ScienceDirect - Advances in Immunology, Volume 86, Pages 1-319 (2005)

Volume 76pp. 1-368 (2001)Volume 75pp. 1-360 (2000)

Volume 86, Pages 1-319 (2005) ƹ Previous vol/iss Next vol/iss Ʒ

Edited by: Frederick W. AltISBN: 978-0-12-004486-3

Article List Full Abstracts

articles 1 - 10

1. ContributorsPages ix-xPDF (29 K) | View Related Articles

2. Adenosine Deaminase Deficiency: Metabolic Basis of Immune Deficiency and Pulmonary InflammationPages 1-41Michael R. Blackburn and Rodney E. KellemsSummaryPlus | Full Text + Links | PDF (427 K) | View Related Articles

3. Mechanism and Control of V(D)J Recombination versus Class Switch Recombination: Similarities and DifferencesPages 43-112Darryll D. Dudley, Jayanta Chaudhuri, Craig H. Bassing and Frederick W. AltSummaryPlus | Full Text + Links | PDF (882 K) | View Related Articles

4. Isoforms of Terminal Deoxynucleotidyltransferase: Developmental Aspects and FunctionPages 113-136To-Ha Thai and John F. KearneySummaryPlus | Full Text + Links | PDF (683 K) | View Related Articles

5. Innate AutoimmunityPages 137-157Michael C. Carroll and V. Michael HolersSummaryPlus | Full Text + Links | PDF (536 K) | View Related Articles

6. Formation of Bradykinin: A Major Contributor to the Innate Inflammatory ResponsePages 159-208Kusumam Joseph and Allen P. KaplanSummaryPlus | Full Text + Links | PDF (533 K) | View Related Articles

7. Interleukin-2, Interleukin-15, and Their Roles in Human Natural Killer CellsPages 209-239Brian Becknell and Michael A. CaligiuriSummaryPlus | Full Text + Links | PDF (376 K) | View Related Articles

8. Regulation of Antigen Presentation and Cross-Presentation in the Dendritic Cell Network: Facts, Hypothesis, and Immunological ImplicationsPages 241-305Nicholas S. Wilson and Jose A. VilladangosSummaryPlus | Full Text + Links | PDF (922 K) | View Related Articles

9. IndexPages 307-314PDF (56 K) | View Related Articles

10. Contents of Recent VolumesPages 315-319PDF (45 K) | View Related Articles

articles 1 - 10

Home

Browse Search My Settings Alerts Help

file:///D|/1/ScienceDirect/Advances%20in%20Immunology/86/0000.htm (2 of 3)18.02.08 15:54:45

ScienceDirect - Advances in Immunology, Volume 86, Pages 1-319 (2005)

About ScienceDirect | Contact Us | Terms & Conditions | Privacy Policy

Copyright É 2008 Elsevier B.V. All rights reserved. ScienceDirectÈ is a registered trademark of Elsevier B.V.

file:///D|/1/ScienceDirect/Advances%20in%20Immunology/86/0000.htm (3 of 3)18.02.08 15:54:45

Contributors

Numbers in parenthesis indicated the pages on which the authors’ contributions begin.

Frederick W. Alt (43), Howard Hughes Medical Institute, The Children’sHospital Boston, CBR Institute for Biomedical Research, and HarvardMedical School, Boston, Massachusetts 02115

Craig H. Bassing (43), Howard Hughes Medical Institute, The Children’sHospital Boston, CBR Institute for Biomedical Research, and HarvardMedical School, Boston, Massachusetts 02115

Brian Becknell (209), Medical Scientist Program and Integrated BiomedicalGraduate Program, Ohio State University, Columbus, Ohio 43210

Michael R. Blackburn (1), Department of Biochemistry and MolecularBiology, University of Texas Health Science Center at Houston, Houston,Texas 77030

Michael A. Caligiuri (209), Medical Scientist Program, IntegratedBiomedical Graduate Program, Department of Internal Medicine,Division of Hematology/Oncology, and Comprehensive Cancer Center,Ohio State University, Columbus, Ohio 43210

Michael C. Carroll (137), CBR Institute for Biomedical Research, andDepartment of Pediatrics, Harvard Medical School, Boston, Massachusetts02115

Jayanta Chaudhuri (43), Howard Hughes Medical Institute, The Children’sHospital Boston, CBR Institute for Biomedical Research, and HarvardMedical School, Boston, Massachusetts 02115

Darryll D. Dudley (43), Howard Hughes Medical Institute, The Children’sHospital Boston, CBR Institute for Biomedical Research, and HarvardMedical School, Boston, Massachusetts 02115

V. Michael Holers (137), Departments of Medicine and Immunology,University of Colorado Health Sciences Center, Denver, Colorado 80217

Kusumam Joseph (159), Division of Pulmonary/Critical Care Medicine andAllergy/Clinical Immunology, Medical University of South Carolina,Charleston, South Carolina 29425

ix

x contributors

Allen P. Kaplan (159), Division of Pulmonary/Critical Care Medicine andAllergy/Clinical Immunology, Medical University of South Carolina,Charleston, South Carolina 29425

John F. Kearney (113), Division of Developmental and Clinical Immunology,Department of Microbiology, University of Alabama at Birmingham,Birmingham, Alabama 35204

Rodney E. Kellems (1), Department of Biochemistry and Molecular Biology,University of Texas Health Science Center at Houston, Houston, Texas77030

To-Ha Thai (113), Division of Developmental and Clinical Immunology,Department of Microbiology, University of Alabama at Birmingham,Birmingham, Alabama 35204

Jose A. Villadangos (241), Immunology Division and Cooperative ResearchCenter for Vaccine Technology, Walter and Eliza Hall Institute of MedicalResearch, Parkville, Victoria 3050, Australia

Nicholas S. Wilson (241), Immunology Division and The CooperativeResearch Center for Vaccine Technology, The Walter and Eliza HallInstitute of Medical Research, Parkville, Victoria 3050, Australia

Adenosine Deaminase Deficiency: Metabolic Basis ofImmune Deficiency and Pulmonary Inflammation

Michael R. Blackburn and Rodney E. Kellems

Department of Biochemistry and Molecular Biology,University of Texas Health Science Center at Houston,

Houston, Texas 77030

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22. ADA Deficiency in Humans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33. Models of ADA-Deficient Severe Combined Immunodeficiency Disease . . . . . . . . . . . . . . . . . . . . 74. Metabolic Disturbances in ADA Deficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125. Pulmonary Consequences of Elevated Adenosine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216. Additional Physiological Consequences of Elevated Adenosine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Abstract

Genetic deficiencies in the purine catabolic enzyme adenosine deaminase(ADA) in humans results primarily in a severe lymphopenia and immuno-deficiency that can lead to the death of affected individuals early in life. Themetabolic basis of the immunodeficiency is likely related to the sensitivity oflymphocytes to the accumulation of the ADA substrates adenosine and 20-deoxyadenosine. Investigations using ADA-deficient mice have provided com-pelling evidence to support the hypothesis that T and B cells are sensitive toincreased concentrations of 20-deoxyadenosine that kill cells through mechan-isms that involve the accumulation of dATP and the induction of apoptosis. Inaddition to effects on the developing immune system, ADA-deficient humansexhibit phenotypes in other physiological systems including the renal, neural,skeletal, and pulmonary systems. ADA-deficient mice develop similar abnorm-alities that are dependent on the accumulation of adenosine and 20-deoxyade-nosine. Detailed analysis of the pulmonary insufficiency seen in ADA-deficientmice suggests that the accumulation of adenosine in the lung can directly accesscellular signaling pathways that lead to the development and exacerbation ofchronic lung disease. The ability of adenosine to regulate aspects of chroniclung disease is likely mediated by specific interactions with adenosinereceptor subtypes on key regulatory cells. Thus, the examination of ADAdeficiency has identified the importance of purinergic signaling duringlymphoid development and in the regulation of aspects of chronic lung disease.

1advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

2 michael r. blackburn and rodney e. kellems

1. Introduction

Adenosine deaminase (ADA) is an essential enzyme of purine metabolism(Fig. 1) and is highly conserved throughout phylogeny. The initial clue reveal-ing the importance of ADA to mammalian organisms came with the chancediscovery that a form of severe combined immunodeficiency disease (SCID) inhumans was associated with ADA deficiency (Giblett et al., 1972). In thisserendipitous way ADA deficiency was the first of the immunodeficiencydiseases for which the underlying biochemical defect was discovered.Subsequent investigations indicated that ADA deficiency accounts for approx-imately 20% of cases of human SCID and that it is the most severe of theimmunodeficiency diseases, affecting both cell-mediated and humoral immu-nity (Buckley et al., 1997; Hershfield and Mitchell, 2001). Soon after theirdiscovery that defects in ADA were associated with immunodeficiency, Giblettand colleagues examined other immunodeficient individuals for deficiencies in

Figure 1 Catabolism of adenosine and 20-deoxyadenosine. Adenosine and 20-deoxyadenosine aredeaminated to inosine and 20-deoxyinosine by adenosine deaminase (ADA). This is followed bycleavage of the purine base from the ribose or deoxyribose sugar moieties by the enzyme purinenucleoside phosphorylase (PNP) to produce hypoxanthine. Hypoxanthine is salvaged back toinosine monophosphate (IMP) in most tissues by hypoxanthine-guanine phosphoribosyltransferase(HGPRT) or is oxidized first to xanthine and then to uric acid by the enzyme xanthine oxidase(XO). In humans, uric acid is excreted in the urine, whereas in the mouse uric acid can beconverted to allantoin by the enzyme uricase before excretion.

adenosine deaminase deficiency 3

purine catabolic enzymes and found that defects in purine nucleoside phos-phorylase (Fig. 1) also result in immunodeficiency disease (Giblett et al.,1975). These findings demonstrate the importance of purine metabolism indevelopment of the immune system. Deciphering the mechanisms by whichdefects in purine metabolism lead to abnormal lymphopoiesis has proved adifficult task, and although significant progress in this arena has been made,many questions remain. However, efforts to understand the metabolic basis ofthe immunodeficiency associated with ADA deficiency have led to advances inthe treatment of certain leukemias, and the treatment of ADA deficiency inhumans has advanced aspects of enzyme replacement and gene replacementtherapies.

The generation of ADA-deficient mice (Blackburn et al., 1998; Migchielsenet al., 1995; Wakamiya et al., 1995) provided the opportunity to examine thepathways by which disturbances in purine metabolism influence physiologicalsystems in the whole animal. ADA-deficient mice develop a combined immu-nodeficiency similar to that seen in ADA-deficient humans, and experiments inthese mice have provided novel mechanistic information about how the accu-mulation of the ADA substrates adenosine and 20-deoxyadenosine impactcomponents of the immune system. In addition, by removing the majorenzyme that controls adenosine levels in tissues and cells, these mice haveserved as biological screens for physiological processes sensitive to chronicelevations in adenosine (Blackburn, 2003). In particular, ADA-deficient micehave served as a useful model for deciphering the role of adenosine signalingin chronic inflammatory lung diseases such as asthma and chronic obstructivepulmonary disease (COPD). This review discusses the current understandingof how the accumulation of ADA substrates impacts the immune andpulmonary systems by comparing findings in ADA-deficient humans and mice.

2. ADA Deficiency in Humans

ADA deficiency in humans arises from naturally occurring mutations in theADA gene that are inherited in an autosomal recessive manner. Most ADA-deficient humans are diagnosed early in life, when they present with markedlymphopenia; failure to thrive; and opportunistic fungal, viral, and bacterialinfections (Buckley et al., 1997; Hershfield and Mitchell, 2001). These patientshave little to no detectable ADA activity and severe metabolic disturbancesassociated with the loss of ADA activity. The thymus is absent or small anddysplastic in ADA-deficient individuals (Borzy et al., 1979), and they haveseverely reduced numbers of peripheral T, B, and natural killer (NK) cells(Buckley et al., 1997). ADA-deficient SCID is the only immunodeficiencyin which all three cell types are severely reduced in number. Without

4 michael r. blackburn and rodney e. kellems

intervention, ADA-deficient individuals die from overwhelming infectionswithin the first year of life. A smaller population of ADA-deficient patientspresents later in life with a less severe form of immunodeficiency that coin-cides with less severe loss of ADA enzymatic activity and associated metabolicdisturbances (Santisteban et al., 1993). Specific mutations within the ADAgene have been identified for both ‘‘early’’ and ‘‘late’’ onset ADA deficiency(reviewed in Hershfield and Mitchell, 2001), and the severity of the mutations,regarding the loss of ADA enzymatic function, correlates well with the severityof the ensuing disease (Hershfield, 2003).

The most successful treatment for ADA deficiency is histocompatible bonemarrow transplantation from an HLA-matched sibling. Because this treatmentoption is seldom available, alternative treatments have been identified, includ-ing T cell–depleted haploidentical bone marrow transplantation from a parent.However, these approaches have met with limited success. A successful bio-chemical approach for the treatment of ADA deficiency involves the use ofenzyme replacement therapy wherein a polyethylene glycol–modified form ofbovine ADA (PEG–ADA) is provided to patients by twice weekly intramuscu-lar injection (Hershfield et al., 1993). Polyethylene glycol appears to protectthe bovine ADA from proteolytic and immunologic attack, hence increasingthe circulating half-life of this exogenous enzyme. ADA replacement therapy iseffective in reducing the metabolic impact of ADA deficiency and has pro-longed the life of individuals who have in some cases been treated for morethan 8 years (Hershfield, 1995). Relatively few complications have beenreported with respect to allergic reactions or immunogenicity to PEG–ADA,and it appears to be the best option for the prolonged treatment ofADA-deficient patients who lack an HLA-identical marrow donor.

ADA deficiency has received considerable attention as the testing groundfor the development of gene therapy protocols. Several features of ADAdeficiency make it an attractive candidate for gene replacement therapy:bone marrow or cord blood stem cells are relatively accessible cell populations;individuals with as little as 5% normal ADA activity have normal immunefunction, suggesting a high degree of replacement may not be necessary; andevidence exists to suggest that T cells with ADA activity can be selected for andenriched in an ADA-deficient environment. The latter is supported by obser-vations in ADA-deficient individuals where spontaneous clinical remissionsoccurred in association with mosaicism for ADA expression (Hirschhorn et al.,1994, 1996), which in one instance was associated with the reversion of amutation in the ADA allele to normal in lymphoid cells (Hirschhorn et al.,1996). Thus, the hope is that efficient transfer of a recombinant ADA gene intohematopoietic cells will result in the outgrowth of a genetically repairedimmune system. For these and other reasons, ADA gene therapy studies

adenosine deaminase deficiency 5

were the first to use ex vivo approaches to stably introduce new geneticinformation into patients (Blaese et al., 1995; Bordignon et al., 1995).Protocols have been initiated in patients in whom peripheral bloodT lymphocytes (Blaese et al., 1995; Bordignon et al., 1995), bone marrow–derived stem cells (Bordignon et al., 1995; Hoogerbrugge et al., 1996), or cordblood stem cells (Kohn et al., 1998) have been transduced with retroviralvectors carrying recombinant ADA. However, the successes of ADA genetherapy attempts have been limited. Although it appears that transducedcells remain viable in ADA-deficient individuals, the transduction efficiencyand ultimately the expression levels of ADA in transduced cells are not highenough to provide clinical benefit. Patients receiving ADA gene therapy arealso maintained on ADA enzyme replacement therapy. PEG–ADA therapy nodoubt complicates the evaluation of the benefits of ADA gene therapy in thesepatients and may even affect the ability of transduced cells to expand.However, the cessation of PEG–ADA therapy in patients receiving ADAgene therapy results in the return of metabolic disturbances and a decreasein certain lymphoid cell populations (Kohn et al., 1998), suggesting PEG–ADAtherapy provides much better protection than ADA-transduced cells. Thus,until methods for achieving higher levels of transduction and expression areachieved, PEG–ADA therapy will likely remain the treatment of choice for thisdisorder in the absence of a compatible bone marrow donor.

Most immunodeficiency diseases are associated with defects in genes thatencode proteins that play obvious roles as signaling components involved inimmune cell development and function. A well-known example is the X-linkedform of SCID that is associated with defects in the gc chain component ofimportant cytokine receptors such as interleukin (IL-2) (Noguchi et al., 1993).Other examples include defects in adaptor protein kinases involved in T-cellresponses, such as ZAP70 (Elder, 1998) and Lck (Goldman et al., 1998);enzymes involved in the rearrangement of B- and T-cell receptor chaingenes, such as recombination-activating proteins 1 and 2 (RAG-1 and RAG-2) (de Saint-Basile et al., 1991; Schwarz et al., 1996); transporter associatedwith antigen processing 1 or 2 (TAP-1 or TAP-2) (Donato et al., 1995;Furukawa et al., 1999), which are important in processing peptide antigensfor presentation to MHC class I molecules; and transcription factors thatregulate the production of MHC class II molecules (Klein et al., 1993). It isreasonable to expect that defects in these molecules that are important inimmune cell development and function would lead to immunodeficiencydisease. However, it is less clear why defects in a widely distributed enzymeof purine metabolism, such as ADA, would cause such a robust form of SCID.

ADA interacts with the cell surface protein CD26/dipeptidyl-peptidase(Kameoka et al., 1993), and it has been proposed that this interaction may

6 michael r. blackburn and rodney e. kellems

play an important role in immune cell function (Morimoto and Schlossman,1998). The involvement of this interaction in the immunodeficiency seen inADA deficiency seems unlikely (Richard et al., 2000) but needs to be pursuedfurther. ADA is highly expressed in lymphoid cells and in the human thymus(Hirschhorn et al., 1978), suggesting that a special need for the regulation ofadenine nucleosides is required for proper lymphoid development and func-tion. Substantial research has been devoted to investigating the link betweenthe purine metabolic disturbances and the immunologic consequences asso-ciated with ADA deficiency. In this regard, most attention has focused on theimpact of the substrates of the ADA-catalyzed reaction, adenosine and 20-deoxyadenosine (Fig. 1; and see Fig. 4), on aspects of T-cell development. Bothnucleosides are elevated in ADA-deficient patients (Hirschhorn, 1993), andeach possesses distinct cellular activities that can impact lymphocyte develop-ment and function (see Fig. 4). ADA enzyme replacement therapy in humansserves to lower circulating levels of adenosine and 20-deoxyadenosine in asso-ciation with improvement in immune status (Hershfield, 1995; Hershfieldet al., 1993), further adding to the notion that the increased concentrationsof these substrates are responsible for the immunodeficiency seen. Ex vivoexperiments using human cells, as well as investigations in animal models, haveled to the discovery of specific cellular pathways that likely account for thedeletion of lymphocyte populations. These activities are discussed in detail inSection 3.

Immunodeficiency is the most thoroughly studied feature of human ADAdeficiency; however, other abnormalities have been reported (Table 1). A largenumber of ADA-deficient patients have bony abnormalities that include flaredcostochondral junctions in the ribs and short growth plates with few prolifer-ating and some hypertropic and necrotic chondrocytes (Cederbaum et al.,1976). Neurological abnormalities have been reported (Hirschhorn et al.,1980), as have renal defects that include mesangial sclerosis along with unusualcortical adrenal fibrosis (Ratech et al., 1985). Liver abnormalities (Bollingeret al., 1996) and pulmonary insufficiencies of unknown etiology (Hershfieldand Mitchell, 2001) have also been noted. In addition, a higher incidence ofthe inflammatory lung disease asthma, as well as increases in circulatingeosinophils and elevations in serum IgE, have been noted in ADA-deficientindividuals (Hirschhorn, 1999; Kawamura et al., 1998; Levy et al., 1988).Autoimmune features have also been documented (Geffner et al., 1986). Thedegree to which these phenotypes are attributed to the metabolic disturbancesseen in ADA-deficient patients is not known; however, analyses of phenotypesin ADA-deficient mice suggest that chronic elevations in adenosine may beinvolved in some of these features.

Table 1 Phenotypic and Metabolic Disturbances in ADA-Deficient Humans and Mice

ADA-deficient humans ADA-deficient mice

. Lymphopenia: peripheral T, B, andNK cell numbers reduced to lessthan 10% of normal

. Lymphopenia: Splenic T, B, andNK cell numbers reduced to less than10–20% of normal

. Absent or small thymus . Absent or small thymus and spleen(with abnormal germinal centers)

. Costochondral junction abnormalities . Costochondral junction abnormalities

. Renal abnormalities . Renal abnormalities

. Hepatic pathology . Hepatic pathology

. Pulmonary insufficiency(increase in asthma incidence)

. Pulmonary insufficiency (with featuresof asthma)

. Peripheral eosinophilia . Peripheral and lung eosinophilia

. Elevated IgE . Elevated IgE

. Neurological abnormalities . Neurological abnormalities(ventriculomegaly)

. Elevated plasma adenosine . Elevated adenosine levels in plasma,spleen, thymus, liver, lung, kidney,and bone marrow

. Elevated plasma and urine2-deoxyadenosine levels

. Elevated 2-deoxyadenosine levels inplasma, urine,a spleen, thymus, andbone marrow

. Elevated dATP levels inerythrocytes

. Elevated dATP levels in erythrocytes,spleen, thymus, liver, lung, and kidney

. AdoHcy hydrolase inhibition inerythrocytes

. AdoHcy hydrolase inhibition inerythrocytes, spleen, thymus, liver,lung, kidney, and bone marrow

aOur unpublished observations.Abbreviations: ADA, adenosine deaminase; AdoHcy, S-adenosylhomocysteine; NK, natural killer.

adenosine deaminase deficiency 7

3. Models of ADA-Deficient Severe Combined Immunodeficiency Disease

3.1. Generation of ADA-Deficient Mice

In efforts to generate animal models to investigate the mechanisms by whichADA deficiency affects the immune system and to promote the advancementof novel treatment approaches such as ADA enzyme replacement therapy andADA gene therapy, two groups generated ADA-deficient mice. Ada null alleleswere generated by the targeted insertion of the neomycin gene into the mouseAda gene (Migchielsen et al., 1995; Wakamiya et al., 1995). Heterozygoteintercrosses failed to produce ADA-deficient pups, and further analysisrevealed that ADA-deficient fetuses died during the fetal period, just beforeterm or soon after birth. Phenotypic analysis of ADA-deficient fetuses revealedsevere hepatocellular damage and loss of liver function (Migchielsen et al.,

8 michael r. blackburn and rodney e. kellems

1995; Wakamiya et al., 1995). Abnormal lung pathology was also reported(Migchielsen et al., 1995). These defects were associated with marked eleva-tions of adenosine and 20-deoxyadenosine in ADA-deficient fetuses, as well asevidence of dATP accumulation and inhibition of S-adenosylhomocysteine(SAH) hydrolase enzymatic activity, which are downstream metabolic path-ways affected by increases in 20-deoxyadenosine (Migchielsen et al., 1995;Wakamiya et al., 1995). These findings demonstrate that ADA plays an essen-tial role in prenatal development and that the prenatal liver and lung aresensitive to elevations in ADA substrates. However, the prenatal death ofADA-deficient fetuses prevented the detailed analysis of the immune systemand other physiological systems affected in ADA-deficient humans.

ADA is present in virtually all cells, but enzyme concentrations differ as muchas 10,000-fold among cell types (Blackburn and Kellems, 1996). In mice, thehighest levels of ADA enzyme activity are observed in trophoblast cells ofthe placenta (Knudsen et al., 1991), epithelial cells lining the gastrointestinaltract (Chinsky et al., 1990; Witte et al., 1991), thymus (Chinsky et al., 1990), anduterine decidual cells (Knudsen et al., 1991). Within these cells and tissues,ADA gene expression is subject to pronounced developmental control, andtransgenic mouse studies have identified key gene regulatory elements respon-sible for directing expression to specific cell types including the thymus (Aronowet al., 1989), forestomach (Xu et al., 1999), small intestine (Dusing et al., 1997),and trophoblast cells of the placenta (Winston et al., 1992). Transgenic mousestudies led to the identification of an enhancer sequence located between 5and 6 kb upstream of the transcription initiation site of the mouse Ada genethat was capable of directing reporter gene expression in all trophoblast celllineages in a manner that coincides with endogenous Ada expression. Theidentification of the gene regulatory elements from mouse Ada responsible forhigh levels of expression in trophoblast cells (Shi et al., 1997; Winston et al.,1992) provided an approach to rescue ADA-deficient fetuses from prenatallethality.

Trophoblast-specific gene regulatory sequences were used to target theexpression of an ADA-encoding minigene specifically to the trophoblast line-age of transgenic mice (Blackburn et al., 1995). These mice were introducedonto the ADA-deficient background to generate mice that contained the ADA-encoding minigene and were also homozygous for the null Ada allele.Restoring ADA to trophoblast cells of the placenta was sufficient to preventthe liver damage seen in ADA-deficient fetuses and was able to rescue themfrom prenatal lethality. When metabolic disturbances were examined in res-cued ADA-deficient fetuses, it was found that placental ADA was able toprevent the accumulation of 20-deoxyadenosine and dATP in ADA-deficientfetuses while having little effect on lowering the levels of circulating adenosine

adenosine deaminase deficiency 9

(Blackburn et al., 1995). These findings suggest that 20-deoxyadenosinecytotoxicity (see Fig. 4) likely accounts for the liver injury seen in ADA-deficient fetuses. Once born, and with the loss of the placenta, ADA enzymaticactivity was not observed in any of the tissues examined in postnatal ADA-deficient mice (Blackburn et al., 1998). These studies demonstrated the im-portance of the prenatal expression of ADA in the trophoblast cells of mice,but more importantly, they provided a strategy for the generation of postnatalADA-deficient mice in which the mechanistic impact of ADA substrate accu-mulation could be examined.

As with ADA-deficient humans, postnatal ADA-deficient mice develop acombined immunodeficiency (Blackburn et al., 1998). In addition, ADA-defi-cient mice develop a number of nonimmune phenotypes that have also beennoted in ADA-deficient humans (Table 1). These include renal, neurological,and bony abnormalities, as well as a severe pulmonary insufficiency. ADA-deficient mice are relatively normal at birth but fail to thrive and die by 3weeks of age as a result of the severity of the phenotypes mentioned earlier.These animals respond well to ADA enzyme replacement therapy (Blackburnet al., 2000a,b) and have provided the opportunity to examine the metabolicmechanisms underlying both immune and nonimmune phenotypes in thecontext of the whole animal.

3.2. Impact of ADA Deficiency on the Immune System

ADA-deficient mice provided the opportunity to directly assess the effect ofADA deficiency on the thymus and spleen, critical immune organs that are notaccessible in ADA-deficient humans. The status of the immune system pro-gressively deteriorates in ADA-deficient mice until the death of the animals atabout 3 weeks of age (Blackburn et al., 1998). By postnatal day 18, thymocyteand splenocyte numbers are less than 10% of those seen in control littermates(Fig. 2A). This is associated with a decrease in thymus size and a substantialincrease in apoptosis in the cortical and medullary regions of the thymus(Apasov et al., 2001). The severe reduction in thymocyte number was asso-ciated with the almost complete loss of CD4þCD8þ double-positive thymo-cytes. These results indicate that thymocyte development is seriously impairedin ADA-deficient mice in association with a pronounced deficiency in theproduction or stability of double-positive thymocytes. Spleens are also smallerin ADA-deficient mice, and analysis of splenocytes revealed a severe reductionin T, B, and NK cells (Fig. 2B). This combined lymphopenia was also seen inperipheral blood. Thus, as in ADA-deficient humans, ADA-deficient miceexhibit a combined immunodeficiency with a severe reduction in T, B, andNK cells (Table 1).

Figure 2 Severe lymphopenia in ADA-deficient mice. (A) Thymus and spleens were removedfrom 17-day-old control (wild-type or heterozygous) or ADA-deficient mice, and the mean numberof cells in each organ (�SEM, n ¼ 5) was determined. (B) Splenocytes were collected from 17-day-old control and ADA-deficient mice and subjected to flow cytometry. Specific antibodies to cellsurface markers were used to identify T cells (anti-CD3 and anti-TCRb), B cells (anti-CD45R andanti-IgM), and NK cells (anti-DX5). Mean total cell counts �SEM were determined, n ¼ 6. Dataare adapted from Blackburn et al. (2000a).

10 michael r. blackburn and rodney e. kellems

3.3. Impaired Intrathymic T-Cell Development

Decreased numbers of circulating T cells, and a decrease in the number ofCD4þCD8þ double-positive cells, suggest a block at a specific stage in thymo-cyte development. Determining the specific stage at which mouse thymopoi-esis is affected in ADA deficiency came from elegant studies conducted inmouse fetal thymic organ culture (FTOC) (Thompson et al., 2000), an ex vivosystem that allows for the monitoring of thymocyte development in a con-trolled environment. FTOCs performed on thymuses from ADA-deficientmouse fetuses on day 15 of gestation, or from normal fetuses treated withthe potent and specific ADA inhibitor 20-deoxycoformycin, showed a profoundinhibition of thymocyte development past the CD4�CD8�CD44loCD25þ

stage. The inhibition of differentiation in ADA-deficient FTOCs was not due

adenosine deaminase deficiency 11

to a failure of b selection, as T-cell receptor b locus (TCR b) rearrangementsand transcription of the T early a locus occurred normally. Rather, the cellsappeared to be dying from apoptosis, because treatment of ADA-deficientFTOCs with the pan-caspase inhibitor N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone (zVADfmk), overexpression of the antiapoptotic proteinBcl-2, or deletion of apoptotic protease-activating factor 1 (APAF-1) improvedcell yield and the extent of differentiation (Thompson et al., 2000; Van DeWiele et al., 2002). These studies are further supported by findings in geneti-cally modified mice treated with the ADA inhibitor 2 0-deoxycoformycin. It wasshown that the thymocyte apoptosis seen after ADA inhibition in vivo could beprevented by overexpression of Bcl-2 or by the removal of p53 (Benveniste andCohen, 1995). Thus, a strong case can be built that thymocytes in an ADA-deficient environment die by apoptosis as a consequence of failing develop-mental checkpoints during thymopoiesis. As is discussed in Section 4,the majority of available evidence suggests that the depletion of T cells in thethymus is due to the consequences of 20-deoxyadenosine accumulation thatculminates in apoptosis.

3.4. Impaired B-Cell Maturation in the Spleen

Most studies examining the effects of ADA deficiency on lymphocytes havefocused on T-cell development, with relatively little attention being given tothe impact of ADA deficiency on B-cell ontogeny and function. Detailedanalysis of B-cell development and function in ADA-deficient mice providedthe first definitive evidence that there is an intrinsic defect within theB lymphocyte compartment in ADA deficiency (Aldrich et al., 2003).Surprisingly, B-cell development in the bone marrow of ADA-deficient micewas normal; however, spleens were smaller, splenic architecture was abnormal,and splenic B cells showed defects in proliferation and activation. ADA-deficient B cells demonstrated a higher propensity to undergo B cellreceptor-mediated apoptosis, suggesting ADA plays a role in the survival ofB cells during antigen-dependent responses. IgM production by extrafollicularplasmablast cells was higher in B cells isolated from ADA-deficient mice,indicating that activated B cells accumulate extrafollicularly as a result ofpoor or nonexistent germinal center formation. This finding was supportedby the observation that there was a profound loss in germinal center formationin the spleens of ADA-deficient mice. The altered splenic environment andsignaling abnormalities seen may contribute to a block in B-cell antigen-dependent maturation in ADA-deficient spleens. Most evidence suggeststhat defects in B-cell development and function are associated with 20-deox-yadenosine cytotoxicity (see Section 4); however, more work is needed to

12 michael r. blackburn and rodney e. kellems

identify the specific mechanisms involved. Similarly, little is known about thenature of the NK cell lymphopenia seen in ADA-deficient humans or mice,and more research is needed to decipher the mechanisms involved in thedepletion of this cell type.

Thus, specific effects of ADA deficiency on specific lymphocyte populationshave emerged from analysis of cellular populations in ADA-deficient mice andmouse FTOCs. By providing sufficient materials to examine subpopulations ofT and B cells during various stages of their development, ADA deficiency inmice has provided new insight into the profile of ADA-deficient SCID.

4. Metabolic Disturbances in ADA Deficiency

Metabolic disturbances associated with ADA deficiency in humans have beenmonitored in accessible fluids and cellular components such as blood andurine (Hershfield and Mitchell, 2001; Morgan et al., 1987; Simmonds et al.,1978). Elevated levels of adenosine and 20-deoxyadenosine are readilydetected in these samples. ADA-deficient mice provided the opportunity toexamine tissue-specific consequences of ADA deficiency in an animal modelwith features resembling those seen in ADA-deficient humans. There weremarked increases in adenosine and 20-deoxyadenosine concentrations in theserum of ADA-deficient mice (Blackburn et al., 1998). In addition, the abilityto examine metabolic disturbances in a variety of tissues revealed a widespreadaccumulation of adenosine, whereas marked 20-deoxyadenosine accumulationwas predominantly limited to lymphoid tissues such as the bone marrow,thymus, and spleen (Aldrich et al., 2003). Furthermore, treatment of ADA-deficient mice by ADA enzyme therapy was able to prevent the accumulationof 20-deoxyadenosine in the thymus and spleen in association with the preven-tion of lymphopenia (Fig. 3) (Blackburn et al., 2000a). These findings providein vivo precedence for the accumulation of 20-deoxyadenosine in lymphoidorgans being associated with the lymphopenia seen in ADA deficiency.

4.1. Deoxyadenosine Metabolism: Origin and Consequences

20-Deoxyadenosine can be cytotoxic to T cells through a number of pathways(Fig. 4). Increases in 20-deoxyadenosine can lead to the inhibition of S-adeno-sylhomocysteine (SAH) hydrolase, which carries out the hydrolysis of SAH toadenosine and homocysteine (Hershfield et al., 1979; Kredich and Hershfield,1979). Such inhibition can lead to an accumulation of SAH that in turn caninhibit key transmethylation reactions that utilize S-adenosylmethionine(SAM) as a methyl donor. Alternatively, 20-deoxyadenosine can be phosphory-lated to dATP by nucleoside and nucleotide kinases (Ullman et al., 1981), and

Figure 3 Effects of ADA enzyme therapy on the thymus and spleen of ADA-deficient mice.(A) 20-Deoxyadenosine levels were quantified in the thymuses and spleens of 17-day-old controlmice (wild-type or heterozygous) and ADA-deficient mice, and in those of ADA-deficientmice treated with a high dose of PEG –ADA, for up to 7 weeks. Data are presented as mean20-deoxyadenosine levels �SEM, n ¼ 3. nd, not detectable. (B) Splenocytes were collected from17-day-old control and ADA-deficient mice, and from ADA-deficient mice treated with a high doseof PEG –ADA, and subjected to flow cytometry. Specific antibodies to cell surface markers wereused to identify Tcells, B cells, and NK cells. Mean total cell counts �SEM were determined, n ¼ 6.Data are adapted from Blackburn et al. (1998, 2000a).

adenosine deaminase deficiency 13

elevations in dATP can in turn lead to the inhibition of ribonucleotide reduc-tase (Ullman et al., 1979) and in so doing deplete the cell of deoxynucleotidesneeded for DNA synthesis or repair. dATP has also been shown to be impor-tant in the activation of apoptotic protease-activating factor 1 (APAF-1) (Leoniet al., 1998), an important protein in the apoptotic cascade, whereas otherstudies suggest dATP can facilitate the loss of cytochrome c from the mito-chondria (Yang and Cortopassi, 1998), which is a key step in the apoptoticpathway. In nondividing lymphocytes, dATP has been proposed to interfere

Figure 4 Proposed mechanisms by which ADA deficiency can affect cell viability and function. (1)Loss of ADA activity by genetic mutation or pharmacologic inhibition leads to elevations inadenosine (Ado) and 20-deoxyadenosine (dAdo) and loss of inosine (Ino) and 20-deoxyinosine (dIno)production through this pathway. (2) Elevations in Ado can lead to aberrant adenosine receptor(AdoR) signaling that can affect cellular viability or function through mechanisms that are not wellcharacterized. (3) Elevations in dAdo can lead to inhibition of the enzyme S-adenosylhomocysteine(SAH) hydrolase, which is responsible for the hydrolysis of SAH to homocysteine (Hcy) and Ado.SAH hydrolase inhibition can lead to elevations in SAH that can in turn inhibit cellular transmethy-lation reactions (X) that utilize S-adenosylmethionine (SAM) as a methyl group donor. Inhibition oftransmethylation reactions can lead to activation of apoptosis or disruption in key pathways that arenecessary for cell differentiation and function. These pathways are largely uncharacterized as theypertain to ADA deficiency. (4) dAdo can be phosphorylated to dATP through pathways that involvevarious deoxynucleoside and deoxynucleotide kinases. Two that have been shown to be important inthe initial phosphorylation of dAdo to dAMP are adenosine kinase and deoxycytidine kinase.Increases in cellular dATP pools have been associated with the induction of apoptosis throughpathways that include the inhibition of ribonucleotide reductase (RNR), which can decreasedeoxynucleotide pools needed for DNA synthesis and repair; the activation of apoptosis by activationof APAF-1 or other mechanisms that could include activation of poly(ADA-ribose) polymerase;NAD depletion; the induction of double-strand breaks; or the release of cytochrome c from themitochondria. Whereas specific mechanisms are still being worked out, most evidence suggests thatelevations in dAdo leading to increases in dATP and the subsequent induction of apoptosis are likelyresponsible for the lymphoid depletion seen in ADA-deficient humans and mice.

14 michael r. blackburn and rodney e. kellems

with DNA repair by the activation of poly(ADP-ribose) polymerase, which candeplete cellular NAD pools (Seto et al., 1985). Thus, the expansion of dATPpools in lymphoid cells caused by elevations in 20-deoxyadenosine may directlyor indirectly access apoptotic pathways.

adenosine deaminase deficiency 15

The inhibition of SAH hydrolase and the accumulation of dATP are featurescommonly monitored in red blood cells of ADA-deficient patients (Colemanet al., 1978; Hershfield et al., 1979). Similarly, SAH hydrolase inhibition anddATP elevations are seen in tissues of ADA-deficient mice (Aldrich et al., 2003;Blackburn et al., 1998). SAH hydrolase is inhibited in all tissues examined inADA-deficient mice, with the greatest degree of inhibition seen in the thymus,spleen, and bone marrow (Fig. 5A). The accumulation of dATP is also abun-dant, with the highest levels being found in red blood cells, thymus, and spleen

Figure 5 S-Adenosylhomocysteine hydrolase enzymatic activity and dATP levels in immunecompartments of ADA-deficient mice. (A) S-Adenosylhomocysteine (SAH) hydrolase enzymaticactivity was determined in red blood cells (RBC), bone marrow (BM), thymus, and spleen of17-day-old control and ADA-deficient mice. SAH-specific activity is given as nanomoles of adeno-sine converted to SAH per minute per milligram protein �SEM, n ¼ 4. (B) HPLC analysis wasused to quantify dATP levels in RBC, BM, thymus, and spleen of 17-day-old control and ADA-deficient mice. Values are presented as mean nanomoles of dATP per milligram protein �SEM,n ¼ 5. Data are adapted from Aldrich et al. (2003) and Blackburn et al. (1998, 2000a).

16 michael r. blackburn and rodney e. kellems

(Fig. 5B). Interestingly, dATP accumulations are not seen in the bone marrowof ADA-deficient mice (Aldrich et al., 2003). Thus, as with the accumulation of20-deoxyadenosine, the most severe disturbances in SAH metabolism anddATP accumulation are seen in components of the immune system, suggestingthese pathways may be involved in driving the immunodeficiency seen inassociation with ADA deficiency.

4.2. Potential Importance of S-Adenosylhomocysteine HydrolaseInhibition and dATP Accumulation

What then are the relative contributions of SAH hydrolase inhibition anddATP accumulation to the depletion of certain lymphocyte populations inADA deficiency? Treatment of mice with a specific SAH hydrolase inhibitorcaused a block in T-cell development similar to that seen in ADA deficiency(Benveniste et al., 1995). In addition, SAH accumulation has been implicatedin the enhancement of apoptosis (Ratter et al., 1996). These studies suggestthat elevations in 20-deoxyadenosine and the subsequent inhibition of SAHhydrolase can lead to a block in T-cell development either by inhibiting yetunidentified transmethylation reactions that are essential for T-cell develop-ment or by contributing to apoptosis. In contrast, other studies suggest thatinhibition of SAH hydrolase is not involved and implicate the accumulation ofdATP and the subsequent induction of apoptosis in the block in T-cell devel-opment seen in ADA deficiency. Experiments in mouse FTOCs demonstratedthat 20-deoxyadenosine levels are generated as a by-product of apoptosis in thethymus (Thompson et al., 2000). In addition, FTOC studies demonstrated thatthe depletion of T cells in an ADA-deficient environment could be preventedby treatment with an adenosine kinase inhibitor that prevented the accumula-tion of dATP (Van De Wiele et al., 2002). In these experiments, SAH hydrolaseenzymatic activity continued to be inhibited despite rescue of the T-cellphenotype, suggesting that SAH inhibition is not involved and implicatingdATP accumulation in the induction of apoptosis. Additional evidence impli-cating dATP comes from studies in ADA-deficient mice, where it was shownthat there was not a block in B-cell development in the bone marrow of ADA-deficient mice, but defects were noted in the spleen (Aldrich et al., 2003). Inthese studies, SAH hydrolase was significantly inhibited in both the bonemarrow and spleen, whereas dATP accumulation was not found in thebone marrow, suggesting that dATP accumulation contributes to B-cell defectsin ADA-deficient mice. The mechanisms by which dATP is toxic to B cells isnot known, but may involve the activation of apoptotic pathways as appears tobe the case in developing T cells.

adenosine deaminase deficiency 17

The mechanisms behind the lack of dATP accumulation in the bone marrowof ADA-deficient mice are not understood but may relate to the inability of 20-deoxyadenosine to be phosphorylated to dATP in developing B cells of thebone marrow. This addresses a key feature to understanding the sensitivity ofcertain cell types to 20-deoxyadenosine. The relative levels of the enzymes thatare responsible for the phosphorylation of 20-deoxyadenosine to dATP, as wellas the catabolism of dATP, may dictate the vulnerability of certain cell types to20-deoxyadenosine (Carson et al., 1980). Mutations in components that phos-phorylate 20-deoxyadensoine to dATP protect cells from 20-deoxyadenosine(Hershfield et al., 1982; Ullman et al., 1978, 1981). A high incidence of apopto-sis, which can generate 20-deoxyadenosine, occurs in the thymus. This, com-bined with increased levels of deoxynucleoside kinases, could provide therationale for the sensitivity of thymocytes to 20-deoxyadenosine, in that theymay have a greater capacity to accumulate dATP. This hypothesis is far fromconfirmed but provides attractive avenues for continued efforts to understandthe mechanisms underlying the lymphoid toxicity associated with ADA defi-ciency. Additional studies are needed to examine the levels of deoxynucleosidekinases and deoxynucleotide catabolic enzymes, not only in various lymphoidand nonlymphoid organs but also in different lymphoid precursors duringdifferent stages of development. These efforts will help clarify why certaincells, such as developing Tcells, are sensitive to elevations in 20-deoxyadenosine,whereas others, like developing B cells in the bone marrow, are not.

4.3. Adenosine Signaling and Immune Development and Function

Adenosine is a ubiquitous and potent signaling nucleoside. It influences cellfunction by engaging G protein-coupled adenosine receptors that access avariety of intracellular signaling pathways (Fig. 6) (Fredholm et al., 2001).Four adenosine receptors have been identified (A1, A2A, A2B, and A3 adenosinereceptors), and each receptor has a unique affinity for adenosine and a distinctcellular and tissue distribution that can vary among species. A2A and A2B

adenosine receptors are commonly coupled to adenylate cyclase by the stimu-latory G protein (as) and serve to increase intracellular cAMP levels (Londoset al., 1980; Stiles, 1997), whereas A1 and A3 adenosine receptors are coupledto adenylate cyclase by the inhibitory G protein (ai) and hence serve to lowerintracellular levels of cAMP (Londos et al., 1980; Stiles, 1997). In addition,evidence exists to suggest that adenosine receptors can couple to other effectormolecules such as phospholipase C and phosphatidylinositol-3-kinase (PI3kinase) (Olah and Stiles, 1995; Stiles, 1997). Therefore, adenosine receptorsignaling can access pathways that regulate intracellular cAMP and Ca2þ and

Figure 6 Adenosine metabolism and signaling. In response to cellular stress or damage, ATP isreleased into the extracellular space by mechanisms that are not fully understood. ATP is itself apotent signaling molecule via its interaction with P2 purinergic receptors. ATP is rapidly dephos-phorylated by extracellular nucleotidases to form extracellular adenosine (Ado). Ecto-50-nucleotid-ase (ecto50NT) is one such enzyme that plays an important role in regulating local adenosineproduction for receptor signaling. Extracellular Ado can interact with adenosine receptors (AdoR)that are coupled to heterotrimeric G proteins, which in turn couple adenosine receptor activationto various effector molecules that can regulate second-messenger systems to influence cell func-tion. Extracellular Ado can also be deaminated by ADA that can exist extracellularly, or it can betransported into cells via facilitated nucleoside transporters. Intracellular Ado is generated fromthe dephosphorylation of AMP by a cytosolic form of nucleotidase (cyto50NT), or the hydrolysis ofS-adenosylhomocysteine (SAH), in a reaction that also produces homocysteine (Hcy). IntracellularAdo can be transported out of the cell or it can be phosphorylated back to ATP. The first enzymein this step is adenosine kinase (AK). Alternatively, intracellular Ado is deaminated to inosine(Ino) by ADA.

18 michael r. blackburn and rodney e. kellems

can therefore influence cellular physiology through a variety of signalingpathways.

Adenosine signaling plays important roles in regulating homeostasis in anumber of physiological systems including the cardiovascular (Belardinelliet al., 1989), nervous (Fredholm and Dunwiddie, 1988), renal (Churchill,1982), and immune (Huang et al., 1997) systems. In addition, signalingthrough adenosine receptors can affect inflammatory processes that helpregulate the severity of certain diseases or pathological conditions. Perhapsthe best characterized inflammatory aspects of adenosine signaling are theantiinflammatory and protective features of A2A adenosine receptors. Studieshave shown that A2A receptor engagement can promote antiinflammatory

adenosine deaminase deficiency 19

effects that help to regulate the severity of inflammation associated with reper-fusion injury (Linden, 2001; Okusa et al., 1999) or infectious challenges (Ohtaand Sitkovsky, 2001). These protective functions likely involve the engagementof adenosine receptors on neutrophils, lymphocytes, and macrophages(Cronstein and Hasko, 2004). Adenosine has also been shown to protect organssuch as the heart (Lasley et al., 1990) brain (Rudolphi et al., 1992), and liver(Day et al., 2004) from ischemic damage and has been shown to exhibit chemo-protective properties (Fishman et al., 2000). In contrast to these prominentprotective and antiinflammatory aspects, adenosine signaling can also promoteor exacerbate tissue injury (Blackburn, 2003). Such is the case with adenosine-mediated mast cell degranulation, which can influence inflammation, tissueinjury, and physiological alterations seen in chronic lung diseases such as asthma(Forsythe and Ennis, 1999; Fozard and Hannon, 1999). Whether adenosinesignaling serves protective or harmful functions is likely dictated by the celltype-specific expression of adenosine receptors and the effector systems theycouple to, together with the concentration and duration of adenosine producedin the local environment. The ability of adenosine receptor engagement toinfluence so many aspects of disease suggests that they will be laudable targetsfor the development of selective therapeutic compounds. This is no doubt therationale behind the efforts of the many pharmaceutical and biotechnologycompanies with active programs in adenosine-based therapeutics.

T cells express various adenosine receptors (Van De Wiele et al., 2002) andadenosine has been shown to induce apoptosis in T lymphocytes in vitro in areceptor-dependent manner (Kizaki et al., 1990). Thus, the accumulation ofadenosine in the thymus could lead to T-cell apoptosis, which could accountfor the depletion of developing T cells in ADA deficiency. Studies haveaddressed the involvement of specific adenosine receptors during T-cell devel-opment in mouse FTOCs (Van De Wiele et al., 2002). In these studies, fetalthymuses from A2A adenosine receptor- and A3 adenosine receptor-deficientmice were subjected to conditions of ADA deficiency, and T-cell developmentwas monitored. Removal of the A2A and A3 adenosine receptors was notable to prevent the block in T-cell development seen in association withADA inhibition. Similarly, treatment with the nonselective adenosine receptoragonist N-ethylcarboxamidoadenosine (NECA) or the nonselective antagonistXAC (xanthine amine congener) had no effect on T-cell development in thismodel system. These findings provide evidence that signaling through adeno-sine receptors is not responsible for the block in T-cell development seen inADA deficiency. However, it is not clear from these studies whether or notadenosine signaling may play a role in peripheral T-cell function.

Studies of mature cells recovered from the spleens of ADA-deficientanimals revealed that ADA deficiency is accompanied by T-cell receptor

20 michael r. blackburn and rodney e. kellems

activation defects in T cells in vivo (Apasov and Sitkovsky, 1999; Apasov et al.,2001). Ex vivo experiments with ADA-deficient thymocytes and peripheralT cells suggested that elevated adenosine and abnormal adenosine receptorsignaling may be responsible for impaired T-cell receptor signaling. These find-ings suggest that adenosine signaling pathways may regulate thymocyte functionin the periphery. In support of this view, ex vivo ADA-deficient thymocytesdemonstrated inhibited tyrosine phosphorylation of T-cell receptor–associatedsignaling molecules and inhibited T-cell receptor-triggered calcium increases.These studies emphasize the potential importance of adenosine signaling inimpaired T-cell immunobiology. Thus, it is likely that accumulation of 20-deoxyadenosine in the thymus is responsible for the depletion of the majorityof thymocytes through dATP accumulation and apoptosis, and those cells thatsurvive this insult may encounter defects in proper T-cell receptor signalingthat are mediated by widespread accumulations of adenosine. The specificadenosine receptors involved in this process are not clear; however, evidencepoints to the involvement of the A2A adenosine receptor (Apasov et al., 2000).More research into the specific impact of adenosine receptor signaling on T, B,and NK cell function is needed to help clarify the role of these signalingpathways in ADA deficiency, as well as normal immune homeostasis.

4.4. Broader Relevance of ADA Deficiency

Efforts to understand the metabolic mechanisms underlying the lymphopeniaassociated with ADA deficiency have provided novel approaches for thetreatment of certain lymphoid neoplasms (Beutler and Carson, 1993).Initially, it was found that 20-deoxycoformycin, a selective suicide inhibitor ofADA, showed efficacy in the treatment of hairy cell leukemia, a malignancywith a relatively low proliferation rate. Subsequently, researchers found thatthe nonhydrolyzable 20-deoxyadenosine analog, 20-chlorodeoxyadenosine(2CdA; cladribine [Leustatin]), was highly effective in the treatment of hairycell leukemia (Beutler and Carson, 1993). The mechanism of action of 2CdAinvolves its phosphorylation to 2CdATP in lymphoid cells. This feature isattributed to the relatively high levels in lymphoid cells of deoxycytidinekinase, which promotes conversion to 2CdATP, and to the relatively low levelsof nucleotidase, which can break 2CdAMP down. Increases in 2CdATP are inturn thought to induce apoptosis by coactivating two key apoptosis-regulatingfactors: poly(ADP-ribose) polymerase and APAF-1 (Carson and Leoni, 2003;Leoni et al., 1998). The effectiveness of 2CdA in the treatment of hairy cellleukemia suggests that the activation of this pathway may provide an effectivemeans of destroying nonproliferating malignant cells. In addition to the

adenosine deaminase deficiency 21

treatment of malignancies, studies have also shown the potential usefulness of2CdA in the treatment of autoimmune disorders (Beutler and Carson, 1993).Thus, the efforts to understand the metabolic mechanisms of ADA deficiencyhave provided novel and far-reaching approaches for the treatment of morecommon disorders not associated with ADA deficiency.

5. Pulmonary Consequences of Elevated Adenosine

In addition to immunodeficiency, ADA-deficient individuals often developpulmonary insufficiencies (Hirschhorn, 1999; Stephan et al., 1993). It is gen-erally assumed that these pulmonary insufficiencies are associated with oppor-tunistic infections inherent to the compromised immune system of thesepatients. However, the frequent appearance of pulmonary insufficiencies inADA-deficient patients suggests that the metabolic disturbances associatedwith ADA deficiency may have a direct impact on lung development, repair,or function. ADA-deficient mice develop many phenotypes other than immu-nodeficiency, of which the most prominent is pulmonary insufficiency(Blackburn et al., 1998). This has provided the opportunity to examine moreclosely the impact of the ADA-associated metabolic disturbances on the lung.

ADA-deficient mice develop outward signs of respiratory distress during thesecond week of life. This respiratory distress is progressive and is believed tocontribute to the death of these animals by 3 weeks of age (Blackburn et al.,2000b). Analysis of the lungs revealed that the pulmonary insufficiency is notdue to opportunistic infections that may result from the compromised immunesystem in these animals. Instead, it was found that a specific pattern of lunginflammation and damage is induced by the accumulation of adenosine in thelungs (Blackburn et al., 2000b). Accumulation of adenosine has long beennoted in the lungs of individuals suffering from chronic lung disease(Driver et al., 1993); however, the contribution of adenosine to the regulationof lung inflammation and the mechanisms involved are not clear. The seren-dipitous finding of adenosine-dependent lung inflammation and damage inADA-deficient mice has provided an opportunity to examine specific aspects ofadenosine signaling and lung inflammation in the context of the whole animal.

5.1. Adenosine Signaling in Asthma and Chronic ObstructivePulmonary Disease

Asthma and chronic obstructive pulmonary disease (COPD) are lung diseasesthat afflict millions of individuals and result in billions of dollars in annualhealth care costs. Persistent pulmonary inflammation and airway remodeling

22 michael r. blackburn and rodney e. kellems

responses are prominent features of these disorders (Elias et al., 1999). Theinflammation associated with asthma and COPD is driven in part by cytokineand chemokine signaling networks, whereas the chronic remodeling and de-struction of the airways are associated with the activation of growth factorsignaling pathways and disruption of protease/antiprotease balances (Braddinget al., 1997; O’Byrne and Postma, 1999). In contrast to most injury and repairresponses, the inflammation seen in these disorders is chronic and may lastthroughout the life of the affected individual. Although signaling pathwaysassociated with the genesis of inflammation and the control of tissue remodel-ing have been described, little is known about signaling pathways that serve toregulate the chronic nature of these diseases.

Adenosine is rapidly generated as a net result of ATP catabolism that occursin situations of cellular stress or damage (Fig. 6). Therefore, it is not surprisingthat pathologic conditions leading to cellular stress and damage, such as theinflammation and tissue damage seen in chronic lung diseases such as asthma,are associated with increases in adenosine levels (Driver et al., 1993; Huszaret al., 2002). The production of adenosine in the lungs of patients with asthmasuggests that it may play a role in regulating aspects of the disease. There issubstantial clinical and scientific evidence to support this hypothesis.Adenosine can directly influence cellular and physiological processes in thelungs of patients with asthma or COPD (Fozard and Hannon, 1999; Jacobsonand Bai, 1997). Exogenous adenosine can elicit acute bronchoconstriction inpatients with asthma (Cushley et al., 1983) or COPD (Oosterhoff et al., 1993),while having no effect on normal individuals, suggesting a fundamental differ-ence regarding adenosine signaling in these patients. In addition, adenosinesignaling can influence the activity of a number of cell types that play a centralrole in chronic lung disease including mast cells (Marquardt et al., 1978),eosinophils (Walker et al., 1997), macrophages (Hasko et al., 1996), neuronalcells (Bai et al., 1989), epithelial cells (Johnson and McNee, 1985), and smoothmuscle cells (Ali et al., 1994). Adenosine receptor levels are elevated in thelungs of patients with asthma and COPD, which further suggests thatincreased adenosine signaling may be an important feature of these diseases(Walker et al., 1997). Despite these lines of evidence, the significance ofadenosine accumulation in the lung is still not clear nor are the underlyingcellular and molecular mechanisms. Part of the challenge of addressing theseissues is to obtain adequate animal models with which to address the impact ofelevations in endogenous lung adenosine. In this regard, the development ofADA-deficient mice and mice deficient in the various adenosine receptors hasproved useful in uncovering specific cellular pathways in the lung that areactivated in response to chronic elevations in adenosine.

adenosine deaminase deficiency 23

5.2. Pulmonary Inflammation in ADA-Deficient Mice

Adenosine concentrations in the lungs of ADA-deficient mice are estimated toreach 100mM (Blackburn, 2003), which is comparable to those measured influid collected from the lungs of patients with asthma (Driver et al., 1993).Interestingly, 20-deoxyadenosine levels are not markedly elevated in the lungsof ADA-deficient mice (Blackburn et al., 2000b), suggesting any local influ-ences of metabolic disturbances are due to the effects of adenosine. In associ-ation with elevations in lung adenosine, ADA-deficient mice develop severepulmonary inflammation and airway remodeling (Blackburn et al., 2000b).Pulmonary features noted include mast cell degranulation (Zhong et al.,2001), increases in activated alveolar macrophages and eosinophils, mucusmetaplasia, fibrosis, airway enlargement (Blackburn et al., 2000b), and airwayhyperreactivity (Chunn et al., 2001). In addition, there are marked lung-specific elevations of key regulatory cytokines, chemokines, and proteases inADA-deficient mice (Banerjee et al., 2002). Many of these pulmonary featuresare also seen in patients with asthma and/or COPD, suggesting that theseanimals may be useful for examining the role of specific aspects of adenosinesignaling in these disorders. Furthermore, the finding that merely elevatingadenosine levels can access pathways that lead to the promotion of lunginflammation and damage suggests that adenosine signaling is playing an activerole in the exacerbation of chronic lung disease.

5.3. Enzyme Therapy Reversal of Many of the Pulmonary AbnormalitiesAssociated with ADA Deficiency

ADA enzyme therapy was used to determine which of the phenotypes seen inthe lungs of ADA-deficient mice were dependent on elevations in lung adeno-sine. As mentioned earlier, bovine ADA that is covalently linked to polyethyl-ene glycol (PEG–ADA) provides an effective means of lowering systemic andtissue levels of ADA substrates in both ADA-deficient humans (Hershfieldet al., 1993) and mice (Blackburn et al., 2000a). Injecting ADA-deficient micewith PEG–ADA at a stage when lung disease is established rapidly lowers lungadenosine levels and reverses most aspects of lung inflammation and airwayremodeling (Fig. 7). Most notable is the reduction of lung eosinophilia andmucus production in airway epithelium (Blackburn et al., 2000b). ADA en-zyme therapy is also able to prevent mast cell degranulation in the lungs ofADA-deficient mice (Zhong et al., 2001), as well as reverse airway hyperreac-tivity (Chunn et al., 2001). Moreover, lowering lung adenosine levels by ADAenzyme therapy is able to promote the survival of these mice (Blackburn et al.,2000b). ADA-deficient mice treated by ADA enzyme therapy recover rapidly

Figure 7 Effects of ADA enzyme therapy on pulmonary adenosine levels and pulmonary pathol-ogies in ADA-deficient mice. (A) Adenosine and 20-deoxyadenosine levels were quantified in thelungs of 18-day-old control (wild-type or heterozygous) and ADA-deficient mice, and in ADA-deficient mice 72 h after ADA enzyme therapy. Values are presented as mean nanomoles permilligram protein �SEM, n ¼ 4. nd, not detectable. (B) Eosinophil numbers in the bronchialalveolar lavage fluid of 18-day-old control and ADA-deficient mice, and in ADA-deficient mice 72 hafter ADA enzyme therapy, were quantified after staining cytospun cells with Diff-Quik. Data arepresented as total cells �SEM, n ¼ 5. (C) The degree of mucus production in the bronchialairways was determined in 18-day-old control and ADA-deficient mice and in ADA-deficient mice72 h after ADA enzyme therapy. Mucus production was determined by quantifying the degree ofmucus staining in the airways. Data are adapted from Blackburn et al. (2000b) and are presented asmean mucus score �SEM, n ¼ 5.

24 michael r. blackburn and rodney e. kellems

adenosine deaminase deficiency 25

from features of respiratory distress and remain normal as long as they aremaintained on treatment. Removal of treatment results in a rapid rise inadenosine levels and renewed pulmonary complications. These findings dem-onstrate a strong correlation between elevations in lung adenosine concentra-tions in ADA-deficient mice and the activation of cell populations such as mastcells, eosinophils, macrophages, and airway epithelial cells, all of which playcentral roles in chronic lung diseases such as asthma and COPD.

Pulmonary features in ADA-deficient mice, such as mast cell degranulation,eosinophilia, mucus metaplasia, and airway hyperresponsiveness, resemblepulmonary phenotypes seen in allergic asthma. However, there is no indicationof allergen immunization or challenge in ADA-deficient mice. This suggests thatin this model, adenosine may directly access pathways downstream of adaptiveimmune responses typically seen in allergic lung disease. Evidence to supportthis comes from studies in ADA-deficient mice that were mated onto aRag1-deficient background. In these studies, it was shown that the pulmonaryphenotypes seen in ADA-deficient mice persist in the absence of T and Blymphocytes (Chan et al., 2003), suggesting that adenosine is directing itseffects in the lung in a lymphocyte-independent manner. These observationsraise the possibility that adenosine is serving to modulate the aspect of innateimmune responses that promote the development of pulmonary pathologies.Examining the mechanisms through which adenosine directs such responseswould greatly increase our knowledge of adenosine actions in normal anddiseased lungs.

5.4. Signaling Through Adenosine Receptors

Most of the physiological effects of adenosine are attributed to signalingthrough adenosine receptors (Fredholm et al., 2001). Examination of adeno-sine receptor transcript levels in whole lungs of ADA-deficient mice by real-time reverse transcription-polymerase chain reaction (RT-PCR) shows thattranscripts for the A1, A2B, and A3 adenosine receptors are significantly elevat-ed, whereas transcript levels for the A2A adenosine receptor are not altered(Chunn et al., 2001). Increased expression of adenosine receptors in lungtissue, or the influx of inflammatory cells expressing these receptors, likelyrepresents an increased capacity for adenosine signaling in the adenosine-richenvironment of ADA-deficient mice. Evidence to support this has come fromstudies examining the cellular localization and functionality of adenosinereceptors in the lungs of ADA-deficient mice.

26 michael r. blackburn and rodney e. kellems

5.4.1. Mast Cells

One approach to begin to address the functionality of adenosine receptors inADA-deficient mice is to examine the effects of adenosine receptor antago-nism in these environments of elevated endogenous adenosine. This approachhas revealed an important role for adenosine signaling in mast cell degranula-tion (Zhong et al., 2001, 2003). Treatment of ADA-deficient mice with thenonselective adenosine receptor antagonists theophylline and MRS 1220 isable to attenuate mast cell degranulation (Zhong et al., 2001) and airwayhyperresponsiveness (Chunn et al., 2001) caused by elevations in endogenousadenosine. Further analysis using the adenosine receptor antagonist MRS1523, which has relative selectivity for the rodent A3 adenosine receptor(Li and Krilis, 1999), suggests that the A3 adenosine receptor is responsiblefor adenosine-mediated mast cell degranulation in the lungs of ADA-deficientmice (Zhong et al., 2003). The specific adenosine receptors responsible foradenosine-mediated airway hyperreactivity in ADA-deficient mice have notbeen determined. The ability of A3 receptor antagonism to prevent adenosine-mediated mast cell degranulation in ADA-deficient mice is consistent withstudies in A3 receptor-deficient mice (Salvatore et al., 2000; Tilley et al., 2000;Zhong et al., 2003) and other rodent systems (Hannon et al., 1995; Jin et al.,1997; Ramkumar et al., 1993; Reeves et al., 1997) that demonstrate a role forthe A3 receptor in mast cell degranulation. Whether A3 receptor-mediatedmast cell degranulation is the major route of adenosine-mediated mast celldegranulation in humans is not clear. Studies have indicated that the A2B

adenosine receptor is responsible for mast cell degranulation in other speciesincluding humans (Auchampach et al., 1997; Feoktistov et al., 2001). Clarifyingthese species-specific differences will be an important step toward the designof adenosine receptor antagonists for regulating adenosine-mediated mast celldegranulation in asthma.

5.4.2. Eosinophils

In addition to mast cell degranulation, pharmacological and genetic studieshave provided evidence that the A3 adenosine receptor plays a role in regulat-ing eosinophil migration and mucus production in the lungs of ADA-deficientmice (Young et al., 2004). The A3 receptor is expressed in eosinophils andmucus-producing cells in the airways of ADA-deficient mice. Treatment ofADA-deficient mice with MRS 1523, a selective A3 adenosine receptor antag-onist, prevents airway eosinophilia and mucus production. Similar findings areseen in the lungs of ADA/A3 double-knockout mice (Young et al., 2004).Interestingly, these studies demonstrate that although eosinophils are de-creased in the airways of ADA-deficient mice after antagonism or removal of

adenosine deaminase deficiency 27

the A3 receptor, elevations in circulating and lung interstitial eosinophilspersist, suggesting that signaling through the A3 receptor is needed for themigration of eosinophils into the airways. These findings identify an importantrole for the A3 adenosine receptor in regulating lung eosinophilia and mucusproduction in an environment of elevated adenosine.

5.4.3. Overview of Receptor Signaling in the Lungs of ADA-Deficient Mice

The above-described findings demonstrate the utility of using the ADA-defi-cient mouse model to examine the specific roles of the A3 receptor in adeno-sine-dependent phenotypes. Continued efforts using pharmacological andgenetic approaches to examine the function of other adenosine receptors inthis model will provide useful information about the overall picture of adeno-sine signaling and chronic lung disease. Doing so will help guide the processof deciding which adenosine receptors, cell types, and signaling pathwaysto focus on for eventual development of adenosine-based therapeutics.Furthermore, the efficiency of ADA enzyme therapy in reversing aspects oflung inflammation and damage in ADA-deficient mice suggests that there maybe reason to consider the use of exogenous ADA to lower the elevatedadenosine levels in the lungs of patients with asthma or COPD. Researchdesigned to extend the observations seen in ADA-deficient mice toother established models of asthma and COPD will be needed before suchhypotheses can be tested in humans.

5.5. Adenosine in Other Models of Lung Disease

There is substantial evidence that adenosine signaling plays a role in animalmodels of asthma and COPD other than the ADA-deficient model. In anumber of well-controlled studies using allergen-sensitized and challengedBrown Norway rats, Fozard and colleagues examined the mechanisms ofadenosine-induced bronchoconstriction (Ellis et al., 2003; Hannon et al.,2001; Tigani et al., 2002). Their findings suggest that the bronchonstrictiveeffects of adenosine are associated with mast cell degranulation and areadenosine receptor dependent; however, the specific adenosine receptor(s)involved have not been determined. Interestingly, treatment of sensitized andchallenged rats with ADA enzyme therapy, or an ADA inhibitor, did notprevent or enhance, respectively, bronchoconstriction or inflammation (Elliset al., 2003), suggesting that endogenous adenosine accumulation has littleeffect on allergen-induced responses in their model. The reason for this is notclear but may be related to the absolute levels of adenosine that accumulate inthe lung. The degree of lung inflammation and damage in this model may not

28 michael r. blackburn and rodney e. kellems

be severe enough to allow for abundant and persistent adenosine generationlocally in the lung. In addition to the rat, studies in an allergic mouse modelhave demonstrated that adenosine can promote bronchoconstriction (Fan andMustafa, 2002). These findings are consistent with those in the ADA-deficientmouse; however, the specific adenosine receptors involved in adenosine-mediated airway hyperresponsiveness are not known. This physiological pa-rameter might be directly associated with mast cell degranulation, as is seen inhuman asthma patients (Cushley et al., 1984) and rat models (Hannon et al.,2001), or it may represent effects on airway nerves or direct effects on airwaysmooth muscle as has been demonstrated in rabbit models of asthma (Ali et al.,1994; Nyce and Metzger, 1997).

Perhaps the most convincing evidence that adenosine is playing a role inthe exacerbation of lung inflammation and damage has come from studiesin mice overexpressing the helper T cell type 2 (Th2) cytokine IL-13 in thelungs (Blackburn et al., 2003). IL-13–overexpressing mice represent a well-established model of chronic lung inflammation and damage (Zhu et al.,1999, 2002). Investigation of the adenosine signaling pathway in these micedemonstrated that with the progression of lung disease, there is a progressiveincrease in lung adenosine concentrations (Fig. 8A). Interestingly, the levelsof ADA transcripts and activity decrease selectively in the lungs of IL-13–overexpressing mice, suggesting there is an orchestrated trend favoring adeno-sine accumulation in the lungs of these mice. In addition to alterationsin adenosine metabolism, there is augmented expression of the A1, A2B, andA3 adenosine receptors, but not the A2A adenosine receptors, in the lungs ofIL-13–overexpressing mice, a pattern that is also seen in ADA-deficientmice (Chunn et al., 2001). Remarkably, ADA enzyme therapy diminished theIL-13–induced increases in lung adenosine in association with preventing IL-13–induced inflammation (Fig. 8B), chemokine elaboration, tissue fibrosis,and alveolar destruction (Blackburn et al., 2003). These findings suggest thatadenosine signaling is playing a role in IL-13–induced lung injury and addssupport to the notion that ADA enzyme therapy may provide protection fromthe detrimental effects of adenosine in chronic lung disease in general. Theinflammation and damage seen in IL-13–overexpressing mice is remarkablysimilar to that seen in the lungs of ADA-deficient mice (Blackburn et al.,2000b). Investigation into the link between ADA-deficient mice and IL-13–overexpressing mice demonstrated that IL-13 is strongly induced in anadenosine receptor-dependent manner in the lungs of ADA-deficient mice.These findings demonstrate that IL-13 and adenosine stimulate one anotherin an amplification pathway that may contribute to the nature, severity, progres-sion, and/or chronicity of IL-13– and/or Th2-mediated disorders. More re-search into the specific mechanisms by which adenosine influences the

Figure 8 ADA enzyme therapy lowers lung adenosine levels and pulmonary inflammation in IL-13–overexpressing mice. (A) HPLC was used to quantify lung adenosine levels in 3-month-oldcontrol (wild-type) and IL-13–overexpressing mice, and 3-month-old IL-13–overexpressing miceafter 1 month of ADA enzyme therapy. Data are presented as mean nanomoles of adenosine permilligram protein �SEM, n ¼ 4. (B) Total cells were counted in bronchial alveolar lavage fluidcollected from the lungs of 3-month-old control and IL-13–overexpressing mice, and 3-month-old IL-13–overexpressing mice after 1 month of ADA enzyme therapy. Data are adapted fromBlackburn et al. (2003) and are presented as total cells �SEM, n ¼ 4. Tg, transgenic.

adenosine deaminase deficiency 29

production of IL-13 is needed to further validate the role of this amplificationloop in the exacerbation of chronic lung disease (Fig. 9).

5.6. Model for Adenosine Amplification of Lung Disease

Most cells in the body generate adenosine and possess adenosine receptors.A challenge in understanding this ubiquitous signaling system is to determinethe factors that govern its regulation and dictate whether it serves to maintain

Figure 9 Model for adenosine-mediated amplification of lung inflammation and damage. A largevariety of insults, including allergy, infection, injury, and environmental insult, can lead to lunginflammation. Inflammation in the lung can then set up a cascade of events that lead to airwaychanges including mucus metaplasia, fibrosis, other aspects of airway remodeling, and airwayhyperresponsiveness. It is these attributes that ultimately lead to the loss of lung function. Thehypoxia and cellular stress associated with lung inflammation and damage lead to the generation ofhigh concentrations of adenosine through pathways that likely involve enhanced ATP release andecto-50-nucleotidase (CD73) activity. Elevations in adenosine may be enhanced by the localdownregulation of ADA in the lung. Elevations in adenosine may be associated with heightenedor aberrant adenosine receptor signaling. Whereas adenosine signaling may be serving someantiinflammatory roles in the lung, most of the accumulated data suggest that increased adenosinesignaling may serve to amplify existing lung inflammation and airway damage by directlyinfluencing proinflammatory signaling molecules and cell types.

30 michael r. blackburn and rodney e. kellems

homeostasis, protect tissues from injury, or trigger the promotion of inflamma-tion or tissue damage. Experiments in ADA-deficient mice demonstrate thatchronically elevated adenosine can lead to lung inflammation and damage.Furthermore, lowering adenosine levels by ADA enzyme therapy can preventor reverse certain aspects of lung inflammation and damage. These findingssuggest that increases in adenosine levels are detrimental in ADA deficiency;however, studies are needed to determine to what degree adenosine is elevatedin patients with COPD and other chronic lung diseases and whether or notthese elevations activate antiinflammatory or proinflammatory pathways. It is

adenosine deaminase deficiency 31

possible that adenosine may serve both pro- and antiinflammatory roles thatare dictated by the levels of adenosine found in the lung (Cronstein and Hasko,2004; Fig. 9). Lung inflammation and tissue damage likely results in a hypoxicenvironment conducive to the generation of adenosine. At lower levels ofadenosine accumulation, engagement of high-affinity receptors, such as theA1 and A2A receptor, might play important roles in controlling the degree ofinflammation or repair processes in the lung. As the damage becomes moresevere and local adenosine levels increase, low-affinity receptors, such as theA2B adenosine receptor, might serve to access signaling pathways that lead tothe exacerbation of the lung inflammation and damage. In this manner, chronicelevations in lung adenosine levels may serve important roles in promoting thechronic nature of lung diseases such as asthma and COPD (Fig. 9). In thissense, the regulation of mediators such as IL-13 by adenosine may be impor-tant. Examination of the specific adenosine receptors involved in these pro-cesses must be elucidated to help define the mechanisms involved. Doing sowill help identify targets for the development of potential adenosine-basedtherapeutics for the treatment of chronic lung disease.

6. Additional Physiological Consequences of Elevated Adenosine

ADA deficiency in humans is most often associated with SCID. Indeed, it isthe immunodeficiency and consequences thereof that lead to the demise ofADA-deficient humans who are not diagnosed and treated. However, nonim-mune phenotypes have been noted, including neurological (Hirschhorn et al.,1980), renal (Ratech et al., 1985), hepatic (Bollinger et al., 1996), and bony(Cederbaum et al., 1976) abnormalities and pulmonary insufficiencies(Stephan et al., 1993). Studies in ADA-deficient mice suggest that theseabnormalities may not be secondary to the immunodeficiency seen but maybe a direct consequence of the metabolic disturbances associated with ADAdeficiency. As discussed earlier, the pulmonary insufficiencies seen in ADA-deficient mice are mediated by the accumulations of adenosine in the lungs ofthese animals, and they have served as useful models for examining the role ofadenosine signaling in lung inflammation and damage (Blackburn, 2003).Whether the pulmonary insufficiencies in ADA-deficient humans are asso-ciated with aberrant adenosine signaling is not known; however, this featurehas not been examined closely. It has been noted that ADA-deficient patientsmay have a higher incidence of asthma and eosinophilia, and elevations in IgEhave been noted (Hirschhorn, 1999; Kawamura et al., 1998; Levy et al., 1988).More research is needed to determine whether these are primary conse-quences of purine metabolic disturbances in the lungs of ADA-deficientindividuals or whether this represents a species- and model-specific outcome.

32 michael r. blackburn and rodney e. kellems

ADA-deficient mice have served as useful biological screens for adenosinesignaling phenotypes that are associated with ADA deficiency. Ongoing re-search will clarify the pathways involved and their relevance to human disease.To date, ADA-deficient mice have been used to describe the contribution ofadenosine signaling in two physiological processes that have not previouslybeen appreciated in ADA deficiency: defects in alveolar development in thelung (Banerjee et al., 2004) and the reduction in brain matter and secondaryventriculomegaly (Turner et al., 2003).

Alveogenesis and microvascular maturation are the final stages in lungdevelopment in mammals. Alveogenesis in the mouse begins on postnatalday 5, when the process of secondary septation plays a pivotal role in theexpansion of the alveolar sacs and microvascular maturation (Ten Have-Opbroek, 1991). ADA-deficient mice exhibit abnormalities in alveogenesis inassociation with elevated lung adenosine levels (Banerjee et al., 2004). Large-scale gene expression analysis of ADA-deficient lungs, using microarrays,revealed novel relationships between gene expression patterns and elevatedlung adenosine during the stages of alveolar maturation. Genes regulatingapoptosis, proliferation, and vascular development are altered, and decreasedcell proliferation in association with increased alveolar type II cell apoptosis isshown to contribute to abnormal secondary septation in these mice. ADAenzyme therapy allowed for normal patterns of apoptosis, proliferation, andalveolar development in association with prevention of adenosine elevations.These findings were correlated with the presence of adenosine receptors in thedeveloping lung, suggesting the involvement of receptor signaling. Thesestudies provide evidence that elevated lung adenosine can lead to abnormalalveogenesis by disrupting patterns of cell proliferation and apoptosis. Thepresence of alveolar defects has not been documented in ADA-deficienthumans; however, such defects may contribute to the idiopathic pulmonaryinsufficiencies that are occasionally seen.

ADA-deficient mice have also proved useful in the analysis of neurologicaldefects associated with elevations in brain adenosine levels. Periventricularleukomalacia is a neurological disorder characterized by a reduction in brainmatter and secondary ventriculomegaly and is a major cause of developmentaldelay and cerebral palsy in premature infants (Melhem et al., 2000; Volpe,2001). In animal models, features of periventricular leukomalacia can beinduced by hypoxia and by activation of A1 adenosine receptors (Latini andPedata, 2001; Turner et al., 2002). This implies that elevations in brain adeno-sine levels after hypoxia can mediate the development of periventricularleukomalacia through A1 adenosine receptor engagement early in life.However, no direct evidence of endogenous adenosine elevations had been

adenosine deaminase deficiency 33

demonstrated to support this. Studies utilized ADA-deficient mice to test thehypothesis that endogenous elevations of adenosine could directly cause ven-triculomegaly. Analysis of brain pathology in ADA-deficient mice on postnatalday 14 demonstrated that there is a decrease in brain matter, and secondaryventriculomegaly is evident in these mice (Turner et al., 2003). Furthermore,this phenotype is associated with marked elevations in brain adenosine levels.These findings add to the emerging hypothesis that adenosine, via signalingthrough the A1 adenosine receptor, can mediate ventriculomegaly during earlypostnatal development. These studies suggest that the pharmacological block-ade of the A1 adenosine receptor may have clinical utility in the treatment ofadenosine-induced brain injury.

Neurological abnormalities have been documented in ADA-deficient indi-viduals (Hirschhorn et al., 1980); however, they are relatively uncommon, andthe diagnosis of ventriculomegaly or cerebral palsy has not been made.Therefore, the pronounced neurological defects seen in ADA-deficient micemay be unique to this model because of species differences or may be due tothe high levels of adenosine allowed to develop in these animals. Most ADA-deficient individuals are diagnosed early in life and are managed on treatmentprotocols such as ADA enzyme replacement therapy. Therefore, it is less likelythat phenotypes such as ventriculomegaly are allowed to develop in ADA-deficient patients. The same may be true for other phenotypes noticed inADA-deficient mice such as bony and renal abnormalities (Blackburn et al.,1998). These abnormalities have been documented in some of the earliestidentified ADA-deficient patients. However, their prevalence appears to bedecreasing, which may be indicative of improved diagnosis and treatment. Themechanisms underlying the bony and renal defects seen in ADA-deficientmice are still being investigated. Efforts to understand the association ofthese defects with the metabolic disturbances seen will help identify the roleof 20-deoxyadenosine cytotoxicity and/or adenosine signaling in these organsystems.

7. Concluding Remarks

Human genetic disorders can serve as naturally occurring genetic screens thatreveal unexpected genotype–phenotype relationships. In this way, rare geneticdisorders can provide significant insight and medical impact that extend wellbeyond the number of individuals affected with the specific genetic disorder.Such is the case of ADA deficiency (Fig. 10). The relationship between ADAactivity and lymphocyte development prompted the use of ADA inhibitors asimmune suppressants and antileukemic agents (Gan et al., 1987; O’Dwyer

Figure 10 Information acquired from the study of ADA deficiency. Observations in ADA-deficient humans, ex vivo systems, and ADA-deficient mice have provided useful informationabout the mechanisms underlying the immunodeficiency seen in this disorder, as well as novelinformation about pathways that impact a broad range of physiological systems. The immunodefi-ciency seen in association with ADA deficiency is likely due to the accumulation of 20-deoxyade-nosine in primary and secondary immune organs, which, through increases in cellular dATP pools,leads to apoptosis. Efforts to treat the immunodeficiency have led to the advancement of ADAenzyme replacement therapy and gene therapy protocols. In addition, examination of the metabol-ic basis of the immune deficiency has led to the development of novel chemotherapeutic ap-proaches for the treatment of certain leukemias. Work stemming predominantly from observationsin ADA-deficient mice has led to novel hypotheses concerning the role of adenosine signaling inchronic lung disease, development, and neurological function. Further examinations of ADA-deficient mice will yield important information about the role of chronic adenosine elevations inthese and other disorders.

34 michael r. blackburn and rodney e. kellems

et al., 1988). In addition, efforts to understand the metabolic consequences ofADA deficiency led to the development and use of nucleoside analogs andother antimetabolites as pharmacologic agents to treat leukemias and autoim-mune disease (Beutler and Carson, 1993). These latter developments camefrom the realization that 20-deoxyadenosine is cytotoxic to developing T cells,which is likely the major underlying feature that accounts for the immunodefi-ciency seen in ADA-deficient individuals. Thus, efforts to understand ADAdeficiency have provided major inroads to the treatment of deadly diseases,particularly cancer.

adenosine deaminase deficiency 35

Studies conducted in ADA-deficient mice have provided insight into therole of adenosine signaling in development, physiology, and disease(Blackburn, 2003). With the realization that mammalian cells contain fourtypes of adenosine receptors came the interest in deciphering the role ofthese receptors in mammalian cell signaling. Some investigators haveapproached these issues pharmacologically, using receptor-specific agonistsand antagonists (Day et al., 2004; Harada et al., 2000). More recently, inves-tigators have used gene-targeting strategies to create mice lacking specificadenosine receptors (Chen et al., 1999; Salvatore et al., 2000; Sun et al.,2001). The generation of ADA-deficient mice (Blackburn et al., 1998) hasprovided a complementary approach in which the biological consequences ofchronically elevating the ligand adenosine can be observed in vivo. In doing so,chronic elevations in lung adenosine have been shown to be important in theactivation of pathways associated with the exacerbation of chronic lung disease(Blackburn, 2003). Combining pharmacological approaches and geneticapproaches to assess the function of the various adenosine receptors ondifferent aspects of lung disease in ADA-deficient mice has begun to identifynovel pathways that may develop into therapeutic targets for the treatment ofchronic lung diseases such as asthma and COPD. In addition to pulmonaryphenotypes, developmental (Banerjee et al., 2004) and neurological (Turneret al., 2003) abnormalities in ADA-deficient mice have been attributed toabnormal adenosine signaling. Thus, ADA-deficient mice have proved usefulfor the identification of cellular events and physiological processes that aresensitive to elevations in endogenous adenosine.

ADA deficiency has been the testing ground for the development of noveltherapies with considerable potential for many areas of medicine. For exam-ple, ADA deficiency was among the first of the immunodeficiencies for whichbone marrow transplantation, enzyme replacement therapy, and gene therapywere attempted (Hershfield and Mitchell, 2001). Bone marrow transplantationand enzyme therapy have proved successful in treating ADA deficiency andother genetic disorders (Buckley et al., 1997). Gene therapy attempts wereuseful in initial assessments of the safety of this procedure and providedevidence of the feasibility of long-term genetic modification of target cells,using ex vivo approaches (Blaese et al., 1995; Bordignon et al., 1995; Kohnet al., 1998). Inadequately low levels of expression remain a problem in mostcases, as does the complication of assessing the benefit of ADA gene therapywith persistent PEG–ADA treatment. However, these studies did lay thegroundwork for the more successful use of gene therapy to treat X-linkedSCID (Cavazzana-Calvo et al., 2000). In this regard, ADA-deficient mice willserve as useful tools for advancing aspects of ADA gene therapy that willbenefit the broader development of this novel medical therapy.

36 michael r. blackburn and rodney e. kellems

Acknowledgments

We thank Dr. Claudia Andreu-Vieyra and Rebecca Corrigan for helpful comments on thismanuscript. This work was supported by National Institutes of Health Grants AI43572 andHL70953 (to M.R.B.) and DK46207 and HD34130 (to R.E.K.). In addition, M.R.B. is supportedby a Junior Investigator Grant from the Sandler Program for Asthma Research.

References

Aldrich, M. B., Chen, W., Blackburn, M. R., Martinez-Valdez, H., Datta, S. K., and Kellems, R. E.(2003). J. Immunol. 171, 5562–5570.

Ali, S., Mustafa, S. J., and Metzger, W. J. (1994). J. Pharmacol. Exp. Ther. 268, 1328–1334.Apasov, S. G., and Sitkovsky, M. V. (1999). Int. Immunol. 11, 179–189.Apasov, S. G., Chen, J. F., Smith, P. T., Schwarzschild, M. A., Fink, J. S., and Sitkovsky, M. V.

(2000). Br. J. Pharmacol. 131, 43–50.Apasov, S. G., Blackburn, M. R., Kellems, R. E., Smith, P. T., and Sitkovsky, M. V. (2001). J. Clin.

Invest. 108, 131–141.Aronow, B., Lattier, D., Silbiger, R., Dusing, M., Hutton, J., Jones, G., Stock, J., McNeish, J.,

Potter, S., Witte, D., and Wiginton, D. (1989). Genes Dev. 3, 1384–1400.Auchampach, J. A., Jin, X., Wan, T. C., Caughey, G. H., and Linden, J. (1997). Mol. Pharmacol.

52, 846–860.Bai, T. R., Lam, R., and Prasad, F. Y. (1989). Pulm. Pharmacol. 1, 193–199.Banerjee, S. K., Young, H. W., Volmer, J. B., and Blackburn, M. R. (2002). Am. J. Physiol. Lung

Cell. Mol. Physiol. 282, L169–L182.Banerjee, S. K., Young, H. W., Barczak, A., Erle, D. J., and Blackburn, M. R. (2004). Am. J. Respir.

Cell Mol. Biol. 30, 38–50.Belardinelli, L., Linden, J., and Berne, R. M. (1989). Prog. Cardiovasc. Dis. 32, 73–97.Benveniste, P., and Cohen, A. (1995). Proc. Natl. Acad. Sci. USA 92, 8373–8377.Benveniste, P., Zhu, W., and Cohen, A. (1995). J. Immunol. 155, 536–544.Beutler, E., and Carson, D. A. (1993). Blood Cells 19, 559–568.Blackburn, M. R. (2003). Trends Pharmacol. Sci. 24, 66–70.Blackburn, M. R., and Kellems, R. E. (1996). Prog. Nucleic Acid Res. Mol. Biol. 55, 195–226.Blackburn, M. R., Wakamiya, M., Caskey, C. T., and Kellems, R. E. (1995). J. Biol. Chem.

270, 23891–23894.Blackburn, M. R., Datta, S. K., and Kellems, R. E. (1998). J. Biol. Chem. 273, 5093–5100.Blackburn, M. R., Aldrich, M., Volmer, J. B., Chen, W., Zhong, H., Kelly, S., Hershfield, M. S.,

Datta, S. K., and Kellems, R. E. (2000a). J. Biol. Chem. 275, 32114–32121.Blackburn, M. R., Volmer, J. B., Thrasher, J. L., Zhong, H., Crosby, J. R., Lee, J. J., and Kellems, R.

E. (2000b). J. Exp. Med. 192, 159–170.Blackburn, M. R., Chun, G. L., Young, H. W. J., Chunn, J. L., Banerjee, S. K., and Elias, J. A.

(2003). J. Clin. Invest. 112, 332–344.Blaese, R. M., Culver, K. W., Miller, A. D., Carter, C. S., Fleisher, T., Clerici, M., Shearer, G.,

Chang, L., Chiang, Y., Tolstoshev, P., et al. (1995). Science 270, 475–480.Bollinger, M. E., Arredondo-Vega, F. X., Santisteban, I., Schwarz, K., Hershfield, M. S., and

Lederman, H. M. (1996). N. Engl. J. Med. 334, 1367–1371.Bordignon, C., Notarangelo, L. D., Nobili, N., Ferrari, G., Casorati, G., Panina, P., Mazzolari, E.,

Maggioni, D., Rossi, C., Servida, P., et al. (1995). Science 270, 470–475.

adenosine deaminase deficiency 37

Borzy, M. S., Schulte-Wissermann, H., Gilbert, E., Horowitz, S. D., Pellett, J., and Hong, R.(1979). Clin. Immunol. Immunopathol. 12, 31–51.

Bradding, P., Redington, A. E., and Holgate, S. T. (1997). In ‘‘Airway Wall Remodelling in Asthma’’(A. G. Stewart, Ed.), pp. 29–63. CRC Press, Boca Raton, FL.

Buckley, R. H., Schiff, R. I., Schiff, S. E., Markert, M. L., Williams, L. W., Harville, T. O., Roberts,J. L., and Puck, J. M. (1997). J. Pediatr. 130, 378–387.

Carson, D. A., and Leoni, L. M. (2003). Best Pract. Res. Clin. Haematol. 16, 83–89.Carson, D. A., Kaye, J., and Wasson, D. B. (1980). J. Immunol. 124, 8–12.Cavazzana-Calvo, M., Hacein-Bey, S., de Saint Basile, G., Gross, F., Yvon, E., Nusbaum, P., Selz, F.,

Hue, C., Certain, S., Casanova, J. L., Bousso, P., Deist, F. L., and Fisher, A. (2000). Science288, 669–672.

Cederbaum, S. D., Kaitila, I., Rimoin, D. L., and Stiehm, E. R. (1976). J. Pediatr. 89, 737–742.Chan, L. L., Sur, S., Alam, R., Wild, J. S., and Chan, T. (2003). J. Allergy Clin. Immunol. 111, S279.Chen, J. F., Huang, Z., Ma, J., Zhu, J., Moratalla, R., Standaert, D., Moskowitz, M. A., Fink, J. S.,

and Schwarzschild, M. A. (1999). J. Neurosci. 19, 9192–9200.Chinsky, J. M., Ramamurthy, V., Fanslow, W. C., Ingolia, D. E., Blackburn, M. R., Shaffer, K. T.,

Higley, H. R., Trentin, J. J., Rudolph, F. B., Knudsen, T. B., et al. (1990). Differentiation42, 172–183.

Chunn, J. L., Young, H. W., Banerjee, S. K., Colasurdo, G. N., and Blackburn, M. R. (2001). J.Immunol. 167, 4676–4685.

Churchill, P. C. (1982). J. Pharmacol. Exp. Ther. 222, 319–323.Coleman, M. S., Donofrio, J., Hutton, J. J., Hahn, L., Daoud, A., Lampkin, B., and Dyminski, J.

(1978). J. Biol. Chem. 253, 1619–1626.Cronstein, B., and Hasko, G. (2004). Trends Immunol. 25, 33–90.Cushley, M. J., Tattersfield, A. E., and Holgate, S. T. (1983). Br. J. Clin. Pharmacol. 15, 161–165.Cushley, M. J., Tattersfield, A. E., and Holgate, S. T. (1984). Am. Rev. Respir. Dis. 129, 380–384.Day, Y. J., Marshall, M. A., Huang, L., McDuffie, M. J., Okusa, M. D., and Linden, J. (2004). Am.

J. Physiol. Gastrointest. Liver Physiol. 286, G285–G293.de Saint-Basile, G., Le Deist, F., de Villartay, J. P., Cerf-Bensussan, N., Journet, O., Brousse, N.,

Griscelli, C., and Fischer, A. (1991). J. Clin. Invest. 87, 1352–1359.Donato, L., de la Salle, H., Hanau, D., Tongio, M. M., Oswald, M., Vandevenne, A., and Geisert, J.

(1995). J. Pediatr. 127, 895–900.Driver, A. G., Kukoly, C. A., Ali, S., and Mustafa, S. J. (1993). Am. Rev. Respir. Dis. 148, 91–97.Dusing, M. R., Brickner, A. G., Thomas, M. B., and Wiginton, D. A. (1997). J. Biol. Chem.

42, 26634–26642.Elder, M. E. (1998). Semin. Hematol. 35, 310.Elias, J. A., Zhu, Z., Chupp, G., and Homer, R. J. (1999). J. Clin. Invest. 104, 1001–1006.Ellis, K. M., Mazzoni, L., and Fozard, J. R. (2003). Br. J. Pharmacol. 139, 1212–1218.Fan, M., and Mustafa, S. J. (2002). Pulm. Pharmacol. Ther. 15, 147–155.Feoktistov, I., Garland, E. M., Goldstein, A. E., Zeng, D., Belardinelli, L., Wells, J. N., and

Biaggioni, I. (2001). Biochem. Pharmacol. 62, 1163–1173.Fishman, P., Bar-Yehuda, S., Farbstein, T., Barer, F., and Ohana, G. (2000). J. Cell. Physiol.

183, 393–398.Forsythe, P., and Ennis, M. (1999). Inflamm. Res. 48, 301–307.Fozard, J. R., and Hannon, J. P. (1999). Pulm. Pharmacol. Ther. 12, 111–114.Fredholm, B. B., and Dunwiddie, T. V. (1988). Trends Pharmacol. Sci. 9, 130–134.Fredholm, B. B., IJzerman, A. P., Jacobson, K. A., Klotz, K. N., and Linden, J. (2001). Pharmacol.

Rev. 53, 527–552.

38 michael r. blackburn and rodney e. kellems

Furukawa, H., Yabe, T., Watanabe, K., Miyamoto, R., Miki, A., Akaza, T., Tadokoro, K., Tohma, S.,Inoue, T., Yamamoto, K., and Juji, T. (1999). Hum. Immunol. 60, 32–40.

Gan, T. E., Dadonna, P. E., and Mitchell, B. S. (1987). Blood 69, 1376–1380.Geffner, M. E., Stiehm, E. R., Stephure, D., and Cowan, M. J. (1986). Am. J. Dis. Child.

140, 1194–1196.Giblett, E. R., Anderson, J. E., Cohen, F., Pollara, B., and Meuwissen, H. J. (1972). Lancet 2,

1067–1069.Giblett, E. R., Ammann, A. J., Wara, D. W., Sandman, R., and Diamond, L. K. (1975). Lancet 1,

1010–1013.Goldman, F. D., Ballas, Z. K., Schutte, B. C., Kemp, J., Hollenback, C., Noraz, N., and Taylor, N.

(1998). J. Clin. Invest. 102, 421–429.Hannon, J. P., Pfannkuche, H. J., and Fozard, J. R. (1995). Br. J. Pharmacol. 115, 945–952.Hannon, J. P., Tigani, B., Williams, I., Mazzoni, L., and Fozard, J. R. (2001). Br. J. Pharmacol.

132, 1509–1523.Harada, N., Okajima, K., Murakami, K., Usune, S., Sato, C., Ohshima, K., and Katsuragi, T. (2000).

J. Pharmacol. Exp. Ther. 294, 1034–1042.Hasko, G., Szabo, C., Nemeth, Z. H., Kvetan, V., Pastores, S. M., and Vizi, E. S. (1996). J. Immunol.

157, 4634–4640.Hershfield, M. S. (1995). Clin. Immunol. Immunopathol. 76, S228–S232.Hershfield, M. S. (2003). Curr. Opin. Immunol. 15, 571–577.Hershfield, M. S., Kredich, N. M., Ownby, D. R., Ownby, H., and Buckley, R. (1979). J. Clin.

Invest. 63, 807–811.Hershfield, M. S., Fetter, J. E., Small, W. C., Bagnara, A. S., Williams, S. R., Ullman, B., Martin, D.

W., Jr., Wasson, D. B., and Carson, D. A. (1982). J. Biol. Chem. 257, 6380–6386.Hershfield, M. S., Chaffee, S., and Sorensen, R. U. (1993). Pediatr. Res. 33, S42–S47.Hershfield, M. S., and Mitchell, B. S. (2001). In ‘‘The Metabolic and Inherited Basis of Human

Disease’’ (C. R. Scriver, A. L. Beaudet, W. S. Sly, and D. Valle, Eds.), 7th ed., pp. 2585–2625.McGraw-Hill, New York.

Hirschhorn, R. (1993). Pediatr. Res. 33, S35–S41.Hirschhorn, R. (1999). In ‘‘Primary Immunodeficiency Disease: A Molecular and Genetic Ap-

proach’’ (H. D. Ochs, C. I. E. Smith, and J. M. Puck, Eds.), pp. 121–138. Oxford UniversityPress, New York.

Hirschhorn, R., Martiniuk, F., and Rosen, F. S. (1978). Clin. Immunol. Immunopathol. 9, 287–292.Hirschhorn, R., Paageorgiou, P. S., Kesarwala, H. H., and Taft, L. T. (1980). N. Engl. J. Med.

303, 377–380.Hirschhorn, R., Yang, D. R., Israni, A., Huie, M. L., and Ownby, D. R. (1994). Am. J. Hum. Genet.

55, 59–68.Hirschhorn, R., Yang, D. R., Puck, J. M., Huie, M. L., Jiang, C. K., and Kurlandsky, L. E. (1996).

Nat. Genet. 13, 290–295.Hoogerbrugge, P. M., van Beusechem, V. W., Fischer, A., Debree, M., le Deist, F., Perignon, J. L.,

Morgan, G., Gaspar, B., Fairbanks, L. D., Skeoch, C. H., Moseley, A., Harvey, M., Levinsky,R. J., and Valerio, D. (1996). Gene Ther. 3, 179–183.

Huang, S., Apasov, S., Koshiba, M., and Sitkovsky, M. (1997). Blood 90, 1600–1610.Huszar, E., Vass, G., Vizi, E., Csoma, Z., Barat, E., Molnar, V. G., Herjavecz, I., and Horvath, I.

(2002). Eur. Respir. J. 20, 1393–1398.Jacobson, M. A., and Bai, T. R. (1997). In ‘‘Purinergic Approaches in Experimental Therapeutics’’

(K. A. Jacobson and M. F. Jarvis, Eds.), pp. 315–330. Wiley-Liss, Danvers, MA.Jin, X., Shepherd, R. K., Duling, B. R., and Linden, J. (1997). J. Clin. Invest. 100, 2849–2857.Johnson, H. G., and McNee, M. L. (1985). Br. J. Pharmacol. 86, 63–67.

adenosine deaminase deficiency 39

Kameoka, J., Tanaka, T., Najima, Y., Schlossman, S., and Morimoto, C. (1993). Science 261,466–469.

Kawamura, N., Ariga, T., Ohtsu, M., Yamada, M., Tame, A., Furuta, H., Kobayashi, I., Okano, M.,Yanagihara, Y., and Sakiyama, Y. (1998). Immunol. Lett. 64, 49–53.

Kizaki, H., Suzuki, K., Tadakuma, T., and Ishimura, Y. (1990). J. Biol. Chem. 265, 5280–5284.Klein, C., Lisowska-Grospierre, B., LeDeist, F., Fischer, A., and Griscelli, C. (1993). J. Pediatr.

123, 921–928.Knudsen, T. B., Blackburn, M. R., Chinsky, J. M., Airhart, M. J., and Kellems, R. E. (1991). Biol.

Reprod. 44, 171–184.Kohn, D. B., Hershfield, M. S., Carbonaro, D., Shigeoka, A., Brooks, J., Smogorzewska, E. M., Barsky,

L. W., Chan, R., Burotto, F., Annett, G., Nolta, J. A., Crooks, G., Kapoor, N., Elder, M., Wara, D.,Bowen, T., Madsen, E., Snyder, F. F., Bastian, J., Muul, L., Blaese, R. M., Weinberg, K., andParkman, R. (1998). Nat. Med. 4, 775–780.

Kredich, N. M., and Hershfield, M. S. (1979). Proc. Natl. Acad. Sci. USA 76, 2450–2454.Lasley, R. D., Rhee, J. W., Van Wylen, D. G., and Mentzer, R. M., Jr. (1990). J. Mol. Cell. Cardiol.

22, 39–47.Latini, S., and Pedata, F. (2001). J. Neurochem. 79, 463–484.Leoni, L. M., Chao, Q., Cottam, H. B., Genini, D., Rosenbach, M., Carrera, C. J., Budihardjo, I.,

Wang, X., and Carson, D. A. (1998). Proc. Natl. Acad. Sci. USA 95, 9567–9571.Levy, Y., Hershfield, M. S., Fernandez-Mejia, C., Polmar, S. H., Scudiery, D., Berger, M., and

Sorensen, R. U. (1988). J. Pediatr. 113, 312–317.Li, L., and Krilis, S. A. (1999). Allergy 54, 306–312.Linden, J. (2001). Annu. Rev. Pharmacol. Toxicol. 41, 775–787.Londos, C., Cooper, D. M., and Wolff, J. (1980). Proc. Natl. Acad. Sci. USA 77, 2551–2554.Marquardt, D. L., Parker, C. W., and Sullivan, T. J. (1978). J. Immunol. 120, 871–878.Melhem, E. R., Hoon, A. H., Jr., Ferrucci, J. T., Jr., Quinn, C. B., Reinhardt, E. M., Demetrides,

S. W., Freeman, B. M., and Johnston, M. V. (2000). Radiology 214, 199–204.Migchielsen, A. A., Breuer, M. L., van Roon, M. A., te Riele, H., Zurcher, C., Ossendorp, F.,

Toutain, S., Hershfield, M. S., Berns, A., and Valerio, D. (1995). Nat. Genet. 10, 279–287.Morgan, G., Levinsky, R. J., Hugh-Jones, K., Fairbanks, L. D., Morris, G. S., and Simmonds, H. A.

(1987). Clin. Exp. Immunol. 70, 491–499.Morimoto, C., and Schlossman, S. F. (1998). Immunol. Rev. 161, 55–70.Noguchi, M., Yi, H., Rosenblatt, H. M., Filipovich, A. H., Adelstein, S., Modi, W. S., McBride, O.

W., and Leonard, W. J. (1993). Cell 73, 147–157.Nyce, J. W., and Metzger, W. J. (1997). Nature 385, 721–725.O’Byrne, P. M., and Postma, D. S. (1999). Am. J. Respir. Crit. Care Med. 159, S41–S66.O’Dwyer, P. J., Wagner, B., Leyland-Jones, B., Wittes, R. E., Cheson, B. D., and Hoth, D. F.

(1988). Ann. Intern. Med. 108, 733–743.Ohta, A., and Sitkovsky, M. (2001). Nature 414, 916–920.Okusa, M. D., Linden, J., Macdonald, T., and Huang, L. (1999). Am. J. Physiol. 277, F404–F412.Olah, M. E., and Stiles, G. L. (1995). Annu. Rev. Pharmacol. Toxicol. 35, 581–606.Oosterhoff, Y., de Jong, J. W., Jansen, M. A., Koeter, G. H., and Postma, D. S. (1993). Am. Rev.

Respir. Dis. 147, 553–558.Ramkumar, V., Stiles, G. L., Beaven, M. A., and Ali, H. (1993). J. Biol. Chem. 268, 16887–16890.Ratech, H., Greco, M. A., Gallo, G., Rimoin, D. L., Kamino, H., and Hirschhorn, R. (1985). Am.

J. Pathol. 120, 157–169.Ratter, F., Germer, M., Fischbach, T., Schulze-Osthoff, K., Peter, M. E., Droge, W., Krammer,

P. H., and Lehmann, V. (1996). Int. Immunol. 8, 1139–1147.

40 michael r. blackburn and rodney e. kellems

Reeves, J. J., Jones, C. A., Sheehan, M. J., Vardey, C. J., and Whelan, C. J. (1997). Inflamm.Res. 46, 180–184.

Richard, E., Arredondo-Vega, F. X., Santisteban, I., Kelly, S. J., Patel, D. D., and Hershfield, M. S.(2000). J. Exp. Med. 192, 1223–1236.

Rudolphi, K. A., Schubert, P., Parkinson, F. E., and Fredholm, B. B. (1992). Cerebrovasc. BrainMetab. Rev. 4, 346–369.

Salvatore, C. A., Tilley, S. L., Latour, A. M., Fletcher, D. S., Koller, B. H., and Jacobson, M. A.(2000). J. Biol. Chem. 275, 4429–4434.

Santisteban, I., Arredondo-Vega, F. X., Kelly, S., Mary, A., Fischer, A., Hummell, D. S., Lawton, A.,Sorensen, R. U., Stiehm, E. R., Uribe, L., et al. (1993). J. Clin. Invest. 92, 2291–2302.

Schwarz, K., Gauss, G. H., Ludwig, L., Pannicke, U., Li, Z., Lindner, D., Friedrich, W., Seger,R. A., Hansen-Hagge, T. E., Desiderio, S., Lieber, M. R., and Bartram, C. R. (1996).Science 274, 97–99.

Seto, S., Carrera, C. J., Kubota, M., Wasson, D. B., and Carson, D. A. (1985). J. Clin. Invest.75, 377–383.

Shi, D., Winston, J. H., Blackburn, M. R., Datta, S. K., Hanten, G., and Kellems, R. E. (1997).J. Biol. Chem. 272, 2334–2341.

Simmonds, H. A., Sahota, A., Potter, C. F., and Cameron, J. S. (1978). Clin. Sci. Mol. Med. 54,579–584.

Stephan, J. L., Vlekova, V., Le Deist, F., Blanche, S., Donadieu, J., De Saint-Basile, G., Durandy, A.,Griscelli, C., and Fischer, A. (1993). J. Pediatr. 123, 564–572.

Stiles, G. (1997). In ‘‘Purinergic Approaches in Experimental Therapeutics’’ (K. A. Jacobson andM. F. Jarvis, Eds.), pp. 29–38. Wiley-Liss, New York.

Sun, D., Samuelson, L. C., Yang, T., Huang, Y., Paliege, A., Saunders, T., Briggs, J., andSchnermann, J. (2001). Proc. Natl. Acad. Sci. USA 98, 9983–9988.

Ten Have-Opbroek, A. A. (1991). Exp. Lung Res. 17, 111–130.Thompson, L. F., Van de Wiele, C. J., Laurent, A. B., Hooker, S. W., Vaughn, J. G., Jiang, H.,

Khare, K., Kellems, R. E., Blackburn, M. R., Hershfield, M. S., and Resta, R. (2000). J. Clin.Invest. 106, 1149–1157.

Tigani, B., Hannon, J. P., Rondeau, C., Mazzoni, L., and Fozard, J. R. (2002). Br. J. Pharmacol.136, 111–119.

Tilley, S. L., Wagoner, V. A., Salvatore, C. A., Jacobson, M. A., and Koller, B. H. (2000). J. Clin.Invest. 105, 361–367.

Turner, C. P., Yan, H., Schwartz, M., Othman, T., and Rivkees, S. A. (2002). Neuroreport 13,1199–1204.

Turner, C. P., Seli, M., Ment, L., Stewart, W., Yan, H., Johansson, B., Fredholm, B. B.,Blackburn, M., and Rivkees, S. A. (2003). Proc. Natl. Acad. Sci. USA 100, 11718–11722.

Ullman, B., Gudas, L. J., Cohen, A., and Martin, D. W., Jr. (1978). Cell 14, 365–375.Ullman, B., Gudas, L. J., Clift, S. M., and Martin, D. W., Jr. (1979). Proc. Natl. Acad. Sci. USA 76,

1074–1078.Ullman, B., Levinson, B. B., Hershfield, M. S., and Martin, D. W., Jr. (1981). J. Biol. Chem.

256, 848–852.Van De Wiele, C. J., Vaughn, J. G., Blackburn, M. R., Ledent, C. A., Jacobson, M., Jiang, H., and

Thompson, L. F. (2002). J. Clin. Invest. 110, 395–402.Volpe, J. J. (2001). Pediatr. Res. 50, 553–562.Wakamiya, M., Blackburn, M. R., Jurecic, R., McArthur, M. J., Geske, R. S., Cartwright, J., Jr.,

Mitani, K., Vaishnav, S., Belmont, J. W., Kellems, R. E., et al. (1995). Proc. Natl. Acad. Sci. USA92, 3673–3677.

adenosine deaminase deficiency 41

Walker, B. A., Jacobson, M. A., Knight, D. A., Salvatore, C. A., Weir, T., Zhou, D., and Bai, T. R.(1997). Am. J. Respir. Cell Mol. Biol. 16, 531–537.

Winston, J. H., Hanten, G. R., Overbeek, P. A., and Kellems, R. E. (1992). J. Biol. Chem. 267,13472–13479.

Witte, D. P., Wiginton, D. A., Hutton, J. J., and Aronow, B. J. (1991). J. Cell Biol. 115, 179–190.Xu, P. A., Winston, J. H., Datta, S. K., and Kellems, R. E. (1999). J. Biol. Chem. 274, 10316–10323.Yang, J. C., and Cortopassi, G. A. (1998). Biochem. Biophys. Res. Commun. 250, 454–457.Young, H. W., Molina, J. G., Dimina, D., Zhong, H., Jacobson, M., Chan, L. N. L., Chan, T. S., Lee,

J. J., and Blackburn, M. R. (2004). J. Immunol. 173, 1380–1390.Zhong, H., Chunn, J. L., Volmer, J. B., Fozard, J. R., and Blackburn, M. R. (2001). J. Pharmacol.

Exp. Ther. 298, 433–440.Zhong, H., Sergiy, S., Molina, J. G., Sanborn, B. M., Jacobson, M. A., Tilley, S. L., and Blackburn,

M. R. (2003). J. Immunol. 170, 338–345.Zhu, Z., Homer, R. J., Wang, Z., Chen, Q., Geba, G. P., Wang, J., Zhang, Y., and Elias, J. A. (1999).

J. Clin. Invest. 103, 779–788.Zhu, Z., Ma, B., Zheng, T., Homer, R. J., Lee, C. G., Charo, I. F., Noble, P., and Elias, J. A. (2002).

J. Immunol. 168, 2953–2962.

Mechanism and Control of V(D)J Recombination versus ClassSwitch Recombination: Similarities and Differences

Darryll D. Dudley,1 Jayanta Chaudhuri,Craig H. Bassing, and Frederick W. Alt

Howard Hughes Medical Institute, The Children’s Hospital Boston,CBR Institute for Biomedical Research, and Harvard Medical School,

Boston, Massachusetts 02115

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431. Overview: V(D)J and Class Switch Recombination .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442. Antigen Receptor Gene Rearrangement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473. Regulation of V(D)J Recombination .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614. Class Switch Recombination Employs Distinct Mechanisms

for V(D)J Recombination .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695. CSR-Related Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 936. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

Abstract

V(D)J recombination is the process by which the variable region exons encodingthe antigen recognition sites of receptors expressed on B and T lymphocytes aregenerated during early development via somatic assembly of component genesegments. In response to antigen, somatic hypermutation (SHM) and classswitch recombination (CSR) induce further modifications of immunoglobulingenes in B cells. CSR changes the IgH constant region for an alternate set thatconfers distinct antibody effector functions. SHM introduces mutations, at ahigh rate, into variable region exons, ultimately allowing affinity maturation. Allof these genomic alteration processes require tight regulatory control mechan-isms, both to ensure development of a normal immune system and to preventpotentially oncogenic processes, such as translocations, caused by errors in therecombination/mutation processes. In this regard, transcription of substratesequences plays a significant role in target specificity, and transcription ismechanistically coupled to CSR and SHM. However, there are many mechanis-tic differences in these reactions. V(D)J recombination proceeds via precise DNAcleavage initiated by the RAG proteins at short conserved signal sequences,whereas CSR and SHM are initiated over large target regions via activation-induced cytidine deaminase (AID)–mediated DNA deamination of transcribed

1Present address: Immunobiology and Cancer Research Program, Oklahoma Medical ResearchFoundation, Oklahoma City, Oklahoma 73104.

43advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

44 darryll d. dudley ET AL.

target DNA. Yet, new evidence suggests that AID cofactors may help provide anadditional layer of specificity for both SHM and CSR. Whereas repair ofRAG-induced double-strand breaks (DSBs) involves the general nonhomolo-gous end-joining DNA repair pathway, and CSR also depends on at least some ofthese factors, CSR requires induction of certain general DSB response factors,whereas V(D)J recombination does not. In this review, we compare and contrastV(D)J recombination and CSR, with particular emphasis on the role of theinitiating enzymes and DNA repair proteins in these processes.

1. Overview: V(D)J and Class Switch Recombination

The lymphoid arm of the vertebrate immune system has evolved to respondand protect against a diverse set of antigens constantly encountered bythe host. Lymphocytes generate a nearly limitless diversity of antigen recep-tors via processes that direct somatic rearrangements and mutations into thegermline DNA sequences of antigen receptor genes. Variable region exonsof antigen receptors expressed on B and T lymphocytes are generatedvia somatic assembly of component variable (V), diversity (D), and joining (J)gene segments in a process called V(D)J recombination. As the usage of partic-ular gene segments for a given locus is to a certain extent stochastic,this combinatorial joining process generates a highly diverse set of antigenreceptors from a limited number of germline gene segments. B cells are capableof undergoing two additional forms of genetic alteration that enhance the abilityof an antigen-specific B cell to recognize and respond to its cognate antigen.Somatic hypermutation (SHM) introduces a high rate of mutations into thegermline DNA sequences of assembled immunoglobulin heavy (IgH) and light(IgL) chain variable region exons and allows the selection of B cells withreceptors that have increased affinity for a given antigen. IgH class switchrecombination (CSR) adjoins a rearranged variable region exon initially asso-ciated with the Igm constant region (Cm) exons to one of several downstreamsets of CH exons (referred to as CH genes) through the deletion of interveninggermline DNA sequences. This allows expression of an antibody with the sameantigen-binding specificity but with altered CH effector function.

Initiation of the V(D)J recombination reaction requires the products ofrecombination activating genes 1 and 2 (RAGs) (Oettinger et al., 1990;Schatz et al., 1989), which are expressed only in developing lymphocytes(Chun et al., 1991; Mombaerts et al., 1992). RAGs were identified by theirability to confer recombinational activity to a fibroblast cell line harboring adrug-selectable recombination substrate (Oettinger et al., 1990; Schatz et al.,1989). Deficiency in either RAG-1 or RAG-2 leads to a complete block inlymphocyte development at progenitor stages, the first stages at which V(D)J

v(d)j versus class switch recombination 45

recombination normally takes place (Mombaerts et al., 1992; Shinkai et al.,1992). RAGs introduce a DNA double-strand break (DSB) precisely betweena variable region gene-coding segment and an associated recombination signal(RS) sequence (reviewed in Fugmann et al., 2000a; Jung and Alt, 2004). EachRS is made up of conserved heptamer and nonamer sequences and an inter-vening spacer sequence that is either 12 or 23 bp in length. RAGs will mediaterecombination only between antigen receptor gene segments that have RSspacer sequences of 12 and 23 bp, referred to as the 12/23 rule. RAG-inducedDNA breaks are repaired by ubiquitously expressed nonhomologous end-joining (NHEJ) proteins, forming precise signal end joints (SJs) and imprecisecoding end joints (CJs) (reviewed in Bassing et al., 2002b; Jung and Alt, 2004).

Lymphoid-specific expression of RAGs limits V(D)J recombination to B andT lymphocytes (reviewed in Nagaoka et al., 2000). However, to ensure thatT cell receptor (TCR) genes are rearranged to completion only in T cells andthat immunoglobulin genes are rearranged to completion only in B cells, theregulation of V(D)J recombination also involves the lineage-specific accessibil-ity of gene segments (Yancopoulos and Alt, 1985). Such regulated accessibilityof antigen receptor gene segments directs developmental stage-specific rear-rangement. In developing B cells IgH genes are assembled before IgL genes,whereas in developing ab T cells TCRb genes are assembled before TCRagenes (reviewed in Willerford et al., 1996). Regulated accessibility also likelycontributes to the ordered rearrangement of IgH and TCRb genes, whereinD-to-J rearrangements proceed to completion before the onset of V-to-DJrearrangements (Alt et al., 1984; Born et al., 1985; Sleckman et al., 2000).Recombinational accessibility correlates with transcriptional activity of a givenantigen receptor locus, as eliminating transcriptional enhancers often ablatesrearrangement of associated gene segments (reviewed in Bassing et al., 2002b;Sleckman et al., 1996).

CSR and SHM, unlike V(D)J recombination, are dependent on activation-induced cytidine deaminase (AID), a protein expressed only in activatedgerminal center B cells (Muramatsu et al., 2000). Conversely, CSR and SHMdo not require the presence of RAGs, as B cells derived by site-specifictargeting of rearranged IgH and IgL transgenes into the corresponding endog-enous loci of RAG-deficient mice undergo normal levels of CSR (Lansfordet al., 1998) and SHM (Zheng et al., 1998). AID was identified via subtractivecloning of a cell line capable of switching from IgM to IgA on appropriatecellular stimulation (Muramatsu et al., 1999). The absence of AID results inthe loss of CSR and SHM in humans and mice and eliminates gene conversionin chickens, a process related to SHM that allows gene diversification in someanimals (Arakawa et al., 2002; Muramatsu et al., 2000; Revy et al., 2000).In addition, expression of AID in nonlymphoid cell lines induces CSR and

46 darryll d. dudley ET AL.

SHM of transfected substrates, implying that AID is the only lymphoid-specific factor necessary to effect these processes (Okazaki et al., 2002;Yoshikawa et al., 2002). Evidence demonstrates that AID deaminates cyti-dines of single-stranded DNA (ssDNA), thereby introducing DNA lesionsthat effect CSR and SHM (Bransteitter et al., 2003; Chaudhuri et al., 2003;Petersen-Mahrt et al., 2002; Pham et al., 2003; Sohail et al., 2003; Yu et al.,2004). Multiple DNA repair pathways including base excision repair(BER), mismatch repair (MMR), and NHEJ appear to be required for theprocessing and resolution of the AID-initiated DNA lesions during SHM andCSR (Chaudhuri and Alt, 2004; Petersen-Mahrt et al., 2002). The NHEJfactors Ku and DNA-PKcs appear to be required for normal levels of CSR(Casellas et al., 1998; Manis et al., 1998a, 2002a) and may be involved in theresolution of DNA lesions, including DNA DSB intermediates induced byAID (Bross et al., 2000; Chen et al., 2001; Papavasiliou and Schatz, 2000;Wuerffel et al., 1997).

In contrast to the site-specific RSs that target V(D)J recombination, CSR istargeted to large regions (1–12 kb) of repetitive DNA sequences, known asswitch (S) regions, located upstream of all CH genes except Cd (which isregulated at the level of alternate RNA splicing) (Davis et al., 1980; Honjoand Kataoka, 1978; Kataoka et al., 1980). Likewise, SHM mutates noncon-served sequences of rearranged VHDJH and VLJL exons (reviewed in Harriset al., 1999; Jacobs and Bross, 2001). CSR requires the transcription of S regiontarget sequences, as disruption of specific S region transcriptional units elim-inates CSR to the corresponding isotype (reviewed in Manis et al., 2002b),whereas transcription has not been shown to be directly (i.e., mechanistically)involved in V(D)J recombination. In this regard, the transcriptional orientationof an S region is important, as inverted S regions are impaired in their ability tomediate CSR in vivo (Shinkura et al., 2003), in accord with a direct role oftranscription in the process of CSR, as opposed to V(D)J recombination, whichclearly involves a different mechanism. Thus, although enhanced Vk germlinetranscription in vivo enhances Vk rearrangement (Casellas et al., 2002),germline promoter location, rather than transcription through gene seg-ments, may target gene segment accessibility via chromatin remodeling in apolymerase-independent manner (Sikes et al., 1998).

The identification of RAG-1 and -2 was instrumental in elucidating theV(D)J recombination mechanism, which is now understood in some detail(Fugmann et al., 2000a). Likewise, the identification of AID has led to rapidadvances in our understanding of SHM and CSR mechanisms (reviewed inHonjo et al., 2002; Kenter, 2003; Manis et al., 2002b). This review comparesand contrasts the targeting, initiation, and resolution of V(D)J recombinationand CSR.

v(d)j versus class switch recombination 47

2. Antigen Receptor Gene Rearrangement

2.1. Genomic Organization of Murine Antigen Receptor Loci

The antigen receptor expressed on the surface of a B cell normally consists offour polypeptides that are made up of two identical IgH chains and twoidentical IgL chains, with IgL chains being derived from the rearrangementof either Igk or Igl genes (reviewed in Gorman and Alt, 1998). T cells expresssurface receptors made up of either ab or gd heterodimers (reviewed inKisielow and von Boehmer, 1995). The assembly of the variable region exonsof Igk and Igl in developing B cells, as well as the assembly of the variableregion exons of TCRa and TCRg in developing T cells, involves the rearrange-ment of V and J gene segments (reviewed in Bassing et al., 2002b). In contrast,IgH, TCRb, and TCRd variable region exons are assembled from componentV, D, and J gene segments, thus increasing the level of diversification ofrearranged products (reviewed in Bassing et al., 2002b). The variable regionexons of all antigen receptors are then linked to constant region exons via RNAsplicing and subsequently expressed at the cell surface (Fig. 1).

The murine IgH locus consists of some several hundred different V genesegments distributed throughout an approximate 1-Mb region beginning about

Figure 1 Schematic diagram of the murine IgH locus before and after V(D)J recombination. TheVH, DH, and JH gene segments are depicted as rectangles. The 12-bp RS sequences are shown asopen triangles, and the 23-bp RS sequences as solid triangles. The m constant region exons areshown as shaded rectangles, and the switch m region as an oval. The position of the iEm enhancer isindicated by a shaded diamond. The positions of the VH and I exon promoters are shown as solidcircles. Distances between the various elements are not drawn to scale.

Figure 2 Schematic diagram of the murine B cell receptor loci. The V, D, and J gene segments aredepicted as rectangles. The 12-bp RS sequences are shown as open triangles, and the 23-bp RSsequences as solid triangles. Only functional constant region exons are shown, represented bysquares. The positions of various enhancer elements are indicated by circles. The estimatednumber of antigen receptor gene segments for the VH and Vk loci is indicated above each locus.Distances between the various elements are not drawn to scale. Adapted from Hesslein and Schatz(2001).

48 darryll d. dudley ET AL.

100 kb upstream of C m on chromosome 12 (reviewed in Honjo and Matsuda,1995) (Fig. 2). Four J gene segments are positioned in a cluster about 7.5 kbupstream of C m, and 13 known D gene segments are dispersed between theVH and JH gene segments (reviewed in Hesslein and Schatz, 2001). The VH

gene segments are flanked at their 30 ends with RSs containing 23-bpsequences (23-bp RS), as are the JH gene segments at their 50 ends (reviewedin Hesslein and Schatz, 2001). The DH gene segments, on the other hand, areflanked on both sides by RSs with 12-bp spacer sequences (12-bp RS)(reviewed in Hesslein and Schatz, 2001). Thus the 12/23 rule prohibits directVH-to-JH joining and ensures the usage of DH gene segments during normalV(D)J rearrangement, augmenting junctional diversification.

The Igk locus spans approximately 3 Mb of chromosome 6 and containsabout 140 Vk gene segments that can rearrange to 1 of 4 functional Jk genesegments positioned just upstream of a single Ck gene (reviewed in Gormanand Alt, 1998; Schable et al., 1999) (Fig. 2). There is also one nonfunctional Jkgene segment (reviewed in Hesslein and Schatz, 2001). Unlike the IgH andIgl loci, Vk gene segments are found in both transcriptional orientations andthus allow for rearrangement by both deletion and inversion of interveningsequences (reviewed in Gorman and Alt, 1998). Vk gene segments are flankedby 12-bp RSs, and Jk segments by 23-bp RSs (reviewed in Gorman and Alt,1998).

The Igl locus in most mouse strains spans about 200 kb on chromosome 16and has only three functional Vl gene segments, each with a flanking 23-bp RS

v(d)j versus class switch recombination 49

(reviewed in Gorman and Alt, 1998; Selsing and Daitch, 1995; Fig. 2). Thereare three functional and one nonfunctional Cl genes, each of which is asso-ciated with an upstream Jl gene segment flanked by a 12-bp RS (reviewed inGorman and Alt, 1998; Selsing and Daitch, 1995). Two of the Vl genesegments are located upstream of all four Jl–Cl units whereas Vl1 is posi-tioned upstream of only the two 30-most Jl–Cl units and is therefore restrictedin potential rearrangements (reviewed in Gorman and Alt, 1998; Selsing andDaitch, 1995).

The TCRb locus contains two Cb genes, each associated with a single Db

and six functional Jb gene segments positioned upstream (reviewed in Glusmanet al., 2001) (Fig. 3). The entire locus spans nearly 700 kb of mouse chromo-some 6 (reviewed in Glusman et al., 2001; Fig. 3). The Jb gene segments areassociated with 12-bp RSs, whereas the Db segments are flanked on the 50 sideby 12-bp RSs and on the 30 side by 23-bp RSs (reviewed in Hesslein and Schatz,2001). There are about 34 Vb gene segments flanked by 23-bp RSs locatedupstream of the DJb clusters, 14 of which appear to be nonfunctional pseudo-genes (reviewed in Hesslein and Schatz, 2001). There is also one Vb segment,Vb14, found 30 of Cb2 that rearranges by inversion (reviewed in Hesslein andSchatz, 2001). The gene segments and associated RSs of the TCRb locus areorganized in such a way that according to the 12/23 rule, direct Vb-to-Jbrearrangement should be allowed, yet such rearrangements do not normallyoccur (Bassing et al., 2000a; Davis and Bjorkman, 1988; Ferrier et al., 1990).Additional constraints, referred to as beyond 12/23 restriction, ensure that Db

Figure 3 Schematic diagram of the murine T-cell receptor loci. The V, D, and J gene segments aredepicted as rectangles. The 12-bp RS sequences are shown as open triangles, and the 23-bp RSsequences as solid triangles. Only functional constant region exons are shown, represented by solidsquares. The positions of various enhancer elements are indicated by circles. The estimatednumber of antigen receptor gene segments for each locus is indicated above each locus. Distancesbetween the various elements are not drawn to scale. Adapted from Hesslein and Schatz (2001).

50 darryll d. dudley ET AL.

gene segments are utilized during Vb(D)J b rearrangement of the TCR b locusand limit direct Vb-to-J b joining (Bassing et al., 2000a; Jung et al., 2003;Sleckman et al., 2000).

Both TCRa and TCRd gene segments are spread throughout a regionspanning more than 1.3 Mb of mouse chromosome 14 (reviewed in Glusmanet al., 2001; Fig. 3). The single Cd, two Dd, and two Jd gene segments arepositioned between the 30-most Va and 50-most Ja segments, and thus aredeleted after Va-to-Ja rearrangement (reviewed in Hesslein and Schatz, 2001).There are more than 85 Va and 12 Vd gene segments, each adjoined by a 30

23-bp RS, some of which can function as either Va or Vd gene segments,located upstream of the Dd segments (reviewed in Hesslein and Schatz, 2001).There is also one Vd positioned 30 of Cd that has a promoter in the oppositetranscriptional orientation and undergoes inversional recombination (reviewedin Hesslein and Schatz, 2001). Like the Db gene segments, the Dd genesegments have 50 12-bp RSs and 30 23-bp RSs (reviewed in Hesslein andSchatz, 2001).

Furthermore, Jd gene segments have 50 12-bp RSs that according to the12/23 rule might allow for direct Vd-to-Jd joining, although, as with the TCRb

locus, this does not normally occur. At least 60 Ja gene segments are foundupstream of a single Ca gene, each associated with a 50 12-bp RS (reviewed inHesslein and Schatz, 2001).

The TCRg locus is distributed across a region spanning approximately 200 kbof mouse chromosome 13 (reviewed in Glusman et al., 2001; Fig. 3). There areseven Vg gene segments and one Vg pseudogene segment interspersed amongthree functional Jg–Cg units and one nonfunctional Jg–Cg unit (reviewedin Hesslein and Schatz, 2001). All gene segments are positioned in thesame transcriptional orientation, with Vg segments flanked by 23-bp RSs andJg gene segments flanked by 12-bp RSs (reviewed in Hesslein and Schatz,2001).

2.2. Initiation of V(D)J Recombination

2.2.1. Recombinant-Activating Genes 1 and 2

RAGs were identified by transfecting cDNAs into a fibroblast cell line carryingthe stable integration of a V(D)J recombination substrate that can confer drugresistance on successful completion of an RS-directed rearrangement (Schatzand Baltimore, 1988). RAG-1 and RAG-2 are each encoded within a singlecoding exon, and the RAG genes are located within 20 kb of one another in theopposite transcriptional orientation (Oettinger et al., 1990). The close proximi-ty of the two RAG genes, the lack of introns in their coding sequences, and

v(d)j versus class switch recombination 51

their inverted orientation led to the hypothesis that the RAGs were once partof a transposable element that integrated into the vertebrate genome (Agrawalet al., 1998; Lewis and Wu, 1997; Spanopoulou et al., 1996; Thompson, 1995;van Gent et al., 1996a). In support of this theory, RAGs have been shown tocarry out transposition of target sequences in vitro (Agrawal et al., 1998; Hiomet al., 1998) and have been implicated in mediating translocations that occur invivo (Messier et al., 2003; Zhu et al., 2002).

Null mutations in RAGs cause a severe combined immune deficiency(SCID) in humans (Schwarz et al., 1996) and mice (Mombaerts et al., 1992;Shinkai et al., 1992) caused by a complete block in B-and T-cell development.The block in lymphocyte development occurs at the B and T progenitor stages(Mombaerts et al., 1992; Shinkai et al., 1992), the stages at which B cellsnormally rearrange IgH genes and T cells rearrange TCRb, g, and d genes(reviewed in Fehling et al., 1999; Willerford et al., 1996). Furthermore, muta-tions that lead to partial RAG activity in humans cause Omenn syndrome (Villaet al., 1998, 1999; Wada et al., 2000), an SCID disorder characterized byhepatosplenomegaly, lymphadenopathy, eosinophilia, elevated IgE, lack ofcirculating B cells, and a variable number of oligoclonal T cells (reviewed inNotarangelo et al., 1999; Villa et al., 2001).

Aberrant RAG activity has been implicated in translocations betweenimmunoglobulin or TCR and oncogenes such as c-Myc, Bcl-2, and Bcl-6among human T and B lineage lymphomas (reviewed by Mills et al., 2003;Roth, 2003). Some such translocations may involve interchromosomal V(D)Jrecombination involving cryptic RSs in the oncogene loci; whereas others mayinvolve aberrant joining of RAG-initiated DSBs at antigen receptor loci togeneral DSBs on other chromosomes. Clear evidence for the latter process hascome from studies of mouse pro-B lymphomas that arise in an NHEJ- andp53-deficient background (Difilippantonio et al., 2002; Guidos et al., 1996;Zhu et al., 2002). In addition, work has suggested that RAGs initiate transloca-tions by introducing ssDNA nicks, which can be converted to DSBs, at crypticRS or other non-B form DNA structures at various chromosomal locations,such as around the major breakpoint cluster region of human Bcl-2, and,thereby, initiate translocations (Lee et al., 2004; Raghavan and Lieber, 2004;Raghavan et al., 2004).

2.2.2. RAGs Recognize Site-Specific Target Sequences

RAGs recognize and bind to site-specific RSs positioned adjacent to all antigenreceptor gene-coding segments (reviewed in Tonegawa, 1983). Each RS con-sists of a conserved 7-bp sequence (heptamer; consensus, 50-CACAGTG), aconserved 9-bp sequence (nonamer; consensus, 50-ACAAAAACC), and anintervening, relatively nonconserved 12 � 1 or 23 � 1 bp spacer sequence

52 darryll d. dudley ET AL.

(Early et al., 1980; Hesse et al., 1989; Max et al., 1979; Sakano et al., 1979).Although overall highly conserved, there is variation between heptamer andnonamer sequences of individual RSs, with those most closely resemblingthe consensus sequences being the most efficiently rearranged (reviewed inLewis, 1994a). Moreover, not all of the positions within the conserved hepta-mer and nonamer sequences appear to be important for RAG-mediatedcleavage. Whereas changes in the first three nucleotide positions of theheptamer or in the sixth or seventh positions of the nonamer greatly reduceRAG-mediated cleavage of plasmid substrates, changes at other positions arebetter tolerated (Hesse et al., 1989). The spacer sequences also play anessential role in V(D)J recombination, as RAG-mediated cleavage will occuronly when an RS with a 12-bp spacer sequence is paired in complex with an RSwith a 23-bp spacer sequence, a constraint referred to as the 12/23 rule(Eastman et al., 1996; Sakano et al., 1981; van Gent et al., 1996b). The 12/23rule appears to be enforced at the level of binding and assembly of RAGs topaired RSs (Hiom and Gellert, 1998; Mundy et al., 2002) as well as thesubsequent cleavage step (West and Lieber, 1998; Yu and Lieber, 2000).Although much less conservation exists in RS spacer sequences comparedwith heptamer and nonamer sequences, these sequences have also beenshown to influence RAG-mediated cleavage and RS usage (Jung et al., 2003;Nadel et al., 1998). As described above, like gene segments (e.g., allV segments) for any given antigen receptor locus are each associated withRSs with the same length spacer sequences, and thus the 12/23 rule preventsnonproductive V-to-V or J-to-J joining.

The configuration of the heavy chain locus ensures that D gene segmentsflanked with 12-bp RSs will be utilized in all successful VHDJH rearrange-ments, as VH and JH gene segments all have 23-bp RSs (Fig. 2). In contrast, theconfiguration of the TCRb locus, with 23-bp RSs flanking the Vbs and 12-bpRSs flanking the Jbs, should allow direct Vb-to-Jb rearrangement according tothe 12/23 rule, yet this rarely occurs in vivo (Bassing et al., 2000; Sleckmanet al., 2000; Wu et al., 2003) (Fig. 3). Even when Db1 was deleted on bothalleles in mice, Vb-to-Jb1 rearrangements rarely took place and subsequent abT-cell development was severely impaired (Bassing et al., 2000). Severalstudies have shown that this so-called beyond 12/23 restriction is enforced atthe level of specific RSs (Bassing et al., 2000; Jung et al., 2003; Tillman et al.,2003). Indeed, when a Vb 23-bp RS was replaced by the 30 Db1 23-bp RS, the‘‘beyond 12/23 restriction’’ was broken, and direct Vb-to-Jb rearrangement wasdetected (Wu et al., 2003). The strength and efficiency with which the 30 Db123-bp RS mediates rearrangement imply that this RS might contribute toordered rearrangement in which Db1-to-Jb rearrangement takes place beforethe onset of Vb-to-DJb rearrangement.

v(d)j versus class switch recombination 53

In a coupled cleavage reaction involving both 12- and 23-bp RSs, RAGsintroduce DNA DSBs between the heptamers and flanking coding sequences,followed by subsequent ligation of the two blunt RS ends and two modifiedcoding ends. Recombination that takes place between RSs found in theopposite chromosomal orientation will therefore result in the deletion ofintervening DNA sequences in the form of covalently sealed DNA circles(Fujimoto and Yamagishi, 1987; Okazaki et al., 1987; Sakano et al., 1979).Subsequent rounds of cell division result in the permanent loss of thesesequences from the genome (Kabat, 1972; Sakano et al., 1979; Tonegawa et al.,1977). On the other hand, recombination between RSs that are in the samechromosomal orientation leads to an inversion of intervening DNA sequencesand retention of these sequences in the genome (Alt and Baltimore, 1982; Lewiset al., 1982; Malissen et al., 1986; Weichhold et al., 1990; Zachau, 1993). As thepresence of an accessible RS is all that is necessary to render a piece of DNAsusceptible to RAG-mediated cleavage, plasmid substrates have been engi-neered that retain either the RS or coding ends, allowing detailed analysis ofeach type of DNA junction (Hesse et al., 1987; Lewis et al., 1985).

2.2.3. Assembly of Precleavage Complex

In vitro, RAGs are found to cooperatively associate with 12- and 23-bp RSsand their flanking coding gene segments to form a synaptic complex (Bailinet al., 1999; Hiom and Gellert, 1997; Leu and Schatz, 1995). Contacts betweenRAG-1 and nonamer sequences are essential for RS binding, whereas inter-actions with the heptamer or coding sequences appear to help provide speci-ficity to the RAG-binding complex and to promote efficient DNA cleavage(Difilippantonio et al., 1996; Roman and Baltimore, 1996). RAG-1 binding tononamer sequences involves the region between residues 376 and 477 ofRAG-1, with a GGRPR motif (amino acids 389–393 of murine RAG-1) thatis also found in members of the bacterial DNA invertase family forming themain site of interaction (Difilippantonio et al., 1996; Spanopoulou et al., 1996).Independently, RAG-1 binds only weakly to heptamer sequences; however, thepresence of RAG-2 has been shown to help stabilize this interaction (Aidiniset al., 2000; Akamatsu and Oettinger, 1998; Fugmann and Schatz, 2001;Spanopoulou et al., 1996; Swanson and Desiderio, 1999). The region ofRAG-1 that makes contact with the heptamer has been mapped to residues528–760 and also appears to contain the main site of RAG-2 interaction(Arbuckle et al., 2001; Peak et al., 2003). Although the RAG-2 protein doesnot bind DNA independently, RAG-2 does make contact with the RS hepta-mer sequence when in a complex with RAG-1 (Difilippantonio et al., 1996;Spanopoulou et al., 1996; Swanson and Desiderio, 1999).

54 darryll d. dudley ET AL.

Synaptic complex assembly begins in vitro with the binding of RAGs to asingle 12-bp RS referred to as a single complex (SC), followed by integration ofthe companion 23-bp RS target DNA into a paired complex (PC) (Jones andGellert, 2002; Mundy et al., 2002; Swanson, 2002b). The DNA-bending pro-teins HMG1 and HMG2 facilitate the integration of the 23-bp RS and assem-bly of the SC (Rodgers et al., 1999; Swanson, 2002a) and appear to promoteRAG-mediated cleavage (Swanson, 2002a; van Gent et al., 1997). The coordi-nated assembly of the PC and subsequent coupled cleavage requires thepresence of Mg2þ divalent cation, whereas in vitro the presence of Mn2þ

allows cleavage to take place on single RS-containing substrates (van Gentet al., 1996b). By replacing Mn2þ or Mg2þ divalent cations with Ca2þ in vitro,DNA cleavage by RAGs is blocked and the SC consisting of RAGs bound to asingle RS can be isolated as an intermediate of the reaction (Hiom and Gellert,1997). This made it possible to detect two distinct complexes that form on asingle RS, single complex 1 (SC1) and single complex 2 (SC2) (Mundy et al.,2002; Swanson, 2002b). The number of RAG-1 subunits in the SC1, SC2, andPC appears to be the same, although whether there are two (Swanson, 2002b)or more (Mundy et al., 2002) molecules of RAG-1 per complex is still not clear.Studies have consistently found that the slower migrating SC1 contains twosubunits of RAG-2, whereas only a single subunit of RAG-2 exists in SC2(Mundy et al., 2002; Swanson, 2002b). However, crystallization of the complexmay ultimately be required to unequivocally ascertain the stoichiometry.

2.2.4. Biochemistry of the Cleavage Reaction

After assembly of the PC, RAGs introduce a single-strand nick in the DNAbetween the border of the RS heptamer and the gene-coding segment in acoupled cleavage reaction that in vivo requires the presence of both a 12-bpRS and a 23-bp RS (McBlane et al., 1995; van Gent et al., 1996b). This createsa 30-OH on one DNA strand of the gene-coding segment and a 50-phosphategroup on the corresponding RS-containing DNA strand (Fig. 4). The 30-OHthen acts as a nucleophile in attacking the opposite DNA strand in a trans-esterification reaction, forming a covalently sealed hairpin coding end and ablunt, 50-phosphorylated RS end (McBlane et al., 1995; Roth et al., 1992)(Fig. 4). After cleavage, the DNA ends are held together in a postcleavagecomplex that includes the RAGs and all four DNA ends (Agrawal and Schatz,1997; Hiom and Gellert, 1998; Jones and Gellert, 2001; Qiu et al., 2001; Tsaiet al., 2002; Yarnell Schultz et al., 2001).

Mutational studies have identified active catalytic residues in RAG-1 thatwhen mutated result in defects in DNA nicking and hairpin formation,although these residues do not appear to be required for assembly of the PC(Fugmann et al., 2000b; Kim et al., 1999; Landree et al., 1999). The three

Figure 4 Biochemistry of V(D)J recombination. Standard V(D)J recombination results in theformation of precise signal joints and modified coding joints. Products of aberrant V(D)J recombi-nation include hybrid joints, open and shut joints, and transposition events. The rectanglesrepresent V, D, or J gene segments and the solid and open triangles represent 12- and 23-bpRSs, respectively. RAG cleavage and subsequent processing and joining via the NHEJ pathwayleads to the standard V(D)J recombination products shown on the left. Hybrid joints can formwhen the 30-OH of an RAG-liberated RS end attacks the hairpin-coding end of the partner genesegment in the coupled reaction, as shown in the center. RAG-mediated transposition of aliberated 30-OH into an intact piece of double-stranded DNA is depicted on the right.

v(d)j versus class switch recombination 55

identified acidic residues (D600, D708, and E782) are all contained within theactive core RAG-1 protein and likely constitute a DDE motif similar to thatfound in many integrase/transposase family proteins (reviewed in Haren et al.,1999). The DDE triad is thought to function in coordinating two divalentmetal ions (Mg2þ) that facilitate the trans-esterification reaction, one acting asa general base and the other as a general acid (reviewed in Haren et al., 1999).The presence of the DDE motif is consistent with the theory that RAGsstarted out as components of a transposable element that integrated into thevertebrate genome (Agrawal et al., 1998; Spanopoulou et al., 1996; Thompson,1995; van Gent et al., 1996a).

Full-length RAG proteins are relatively insoluble, and therefore elucidationof the biochemistry behind RAG-mediated V(D)J recombination has largelymade use of highly truncated ‘‘core’’ RAG proteins (Kirch et al., 1996;

56 darryll d. dudley ET AL.

McBlane et al., 1995; Sadofsky et al., 1993, 1994; Sawchuk et al., 1997).Although truncated core RAG proteins are capable of mediating completeV(D)J recombination in vitro, the core RAGs carry out the reaction at reducedefficiency both in vitro and in vivo (Akamatsu et al., 2003; Dudley et al., 2003;Kirch et al., 1998; Liang et al., 2002). In the absence of the C-terminal portionof RAG-2, V-to-DJ rearrangements appear more severely affected than D-to-Jrearrangements, suggesting that the noncore region of RAG-2 may play aspecific role during ordered rearrangement of IgH and TCR genes (Kirchet al., 1998; Roman et al., 1997).

In addition to normal CJs and SJs (see below), RAGs can mediate open andshut joints, hybrid joints (HJs), and transpositions both in vitro and in vivo(Fig. 4) (Agrawal et al., 1998; Lewis et al., 1988; Messier et al., 2003;Morzycka-Wroblewska et al., 1988; Sekiguchi et al., 2001). An HJ is definedas the joining of the liberated RS end from one coding segment to the partnerhairpin-coding end participating in the recombination reaction (Fig. 4) (Lewiset al., 1988). A transposition event is similar to a hybrid joint, with insertion ofthe liberated RS end into a double-stranded DNA target sequence instead ofjoining with an RAG-generated coding end (Fig. 4) (Agrawal et al., 1998; Hiomet al., 1998). The truncated core RAGs mediate an increased rate of HJs inNHEJ-deficient cells compared with full-length RAGs, suggesting that thenoncore regions normally function to suppress such aberrant joining events(Sekiguchi et al., 2001). Furthermore, there is an increase in the frequency oftransposition events in core RAG-2–expressing cells compared with controls(Elkin et al., 2003; Tsai and Schatz, 2003), which form by a similar mechanismas that of hybrid joints. Taken together, these studies imply that the noncoreregions of RAGs may have evolved to ensure that RAG-liberated DNA endsare properly joined, thus preventing transposition and other deleterious orineffective recombination reactions (Agrawal and Schatz, 1997; Elkin et al.,2003; Hiom et al., 1998; Messier et al., 2003; Sekiguchi et al., 2001; Steen et al.,1999; Tsai and Schatz, 2003).

2.2.5. Postcleavage Complex

After cleavage, the RAGs remain associated with the four DNA ends in apostcleavage complex, possibly playing a role in the protection of DNA endsfrom degradation, the juxtaposition of ends before rejoining, or recruitmentand activation of end-joining factors for both CJ and SJ formation (Fig. 5)(Agrawal and Schatz, 1997; Hiom and Gellert, 1998; Jones and Gellert, 2001;Qiu et al., 2001; Tsai et al., 2002; Yarnell Schultz et al., 2001). Stability of thepostcleavage complex may also function to inhibit DSB-induced cell cyclearrest and apoptosis, as well as to prevent potentially deleterious transpositionevents (Jones and Gellert, 2001; Perkins et al., 2002). However, studies have

Figure 5 Joining of RAG-mediated DNA double-strand breaks. (A) RAG-1 and RAG-2 cleavageoccurs between RS and coding segments. (B) Ku70 and Ku80 bind to the broken DNA ends.(C) DNA-PKcs and Artemis facilitate the opening and processing (opening) of covalently sealedhairpin coding ends. (D) TdT adds random nucleotides to opened coding ends. XRCC4 and Lig4seal the blunt signal ends and processed coding ends to produce precise signal joints and modifiedcoding joints. In addition, DNA-PKcs functions independently of Artemis to form normal signaljoints.

v(d)j versus class switch recombination 57

shown that signal ends must be deproteinized before rejoining by NHEJfactors in vitro (Leu et al., 1997; Ramsden et al., 1997). In this regard, it wasdemonstrated that the N terminus of RAG-1 has E3 ubiquitin ligase activity(Yurchenko et al., 2003), suggesting a function for RAG-1 in steps beyondrecognition and DNA cleavage. For instance, once the appropriate end-joiningproteins have been recruited or have performed their function, RAG-1–mediated ubiquitination could tag RAG-2 or NHEJ proteins within the com-plex for proteasomal degradation, thus promoting disassembly of the complexand ligation of the DNA ends.

2.2.6. Coding and Signal Joint Formation

RAG-mediated cleavage generates hairpin-coding ends that must be openedand processed before rejoining, whereas the RS ends do not require anyadditional processing and are religated by NHEJ proteins to form precise

58 darryll d. dudley ET AL.

SJs (reviewed in Fugmann et al., 2000a; Fig. 5). Although studies have demon-strated that the RAGs themselves can mediate hairpin opening in vitro(Besmer et al., 1998; Ma et al., 2002; Shockett and Schatz, 1999), evidencehas shown that the NHEJ protein Artemis, in association with DNA-PKcs,likely performs this role in vivo (Ma et al., 2002; Rooney et al., 2002, 2003)(Fig. 5). Hairpin-coding ends are opened via the introduction of a DNA nick,usually within four or five nucleotides 30 of the apex of the hairpin (reviewed inFugmann et al., 2000a; Lieber, 1991; Nadel et al., 1995). Once the hairpins areopened, the 30 overhangs can be filled in via DNA polymerases, thus generat-ing short stretches of palindromic sequences at the junctions of CJs, referredto as P nucleotides (Lafaille et al., 1989; Lewis, 1994b; reviewed in Lewis,1994a; Lieber, 1991). Alternatively, nucleases can chew back the 30 overhangs,resulting in a loss of germline nucleotides at the junction of CJs (reviewed inFugmann et al., 2000a; Lieber, 1991; Nadel et al., 1995). To further diversifyjunctions, the lymphoid-specific protein terminal deoxynucleotidyltransferase(TdT) adds random, nontemplated nucleotides to 30 coding ends and intro-duces so-called N-nucleotide additions, further increasing the diversity ofantigen receptor variable regions (Alt and Baltimore, 1982). Moreover, a splicevariant of TdT appears to function to remove nucleotides from coding junc-tions (Thai et al., 2002). Although TdT is not required for either V(D)Jrecombination or lymphocyte development, it does affect overall repertoirediversification (Gilfillan et al., 1993; Komori et al., 1993). Finally, DNA poly-merase m (polm)–deficient mice have a significant reduction in the length ofVk-to-Jk junctions, suggesting polm plays a role in maintaining CDR3 length ofIgk chains (Bertocci et al., 2003). It is still unclear whether polm regulates theprocessing of coding ends by protecting them from exonucleolytic attack or byfilling in 30 overhangs (Bertocci et al., 2003). Notably, polm shares homologywith TdT.

2.3. Joining of RAG-Mediated DNA Double-Strand Breaks

DNA DSBs can be induced by a variety of agents including ionizing radiation(IR), oxidative stress incurred during normal cellular metabolism, and RAGsduring V(D)J recombination. Mammalian cells have evolved two differentpathways to repair such potentially catastrophic lesions. Homologous recom-bination (HR) is a high-fidelity process that repairs breaks, using a homologouschromosome as a DNA template (reviewed in Hoeijmakers, 2001). NHEJrepairs broken DNA ends in the absence of long stretches of homology,allowing both the loss and addition of nucleotides at the repair junction(reviewed in Khanna and Jackson, 2001). HR takes place predominantly inS and G2 phases of the cell cycle, when homologous templates are both

v(d)j versus class switch recombination 59

available and in close proximity (reviewed in Hoeijmakers, 2001). Conversely,NHEJ appears to be the preferred repair pathway during the G1 phase of thecell cycle, corresponding to the stage at which RAGs are both expressed andactive for recombination (reviewed in Jackson, 2002; Lin and Desiderio, 1995).However, it is clear that NHEJ can function outside of the G1 phase andcomplement the repair activities of HR (Couedel et al., 2004; Mills et al.,2004). Studies involving the transfection of recombination substrates into IR-sensitive cell lines implicated several members of ubiquitously expressedNHEJ proteins as having direct roles in V(D)J recombination (reviewed inBassing et al., 2002b; Taccioli et al., 1993). Members of the NHEJ repairpathway known to be involved in V(D)J recombination include Ku70, Ku80,DNA-PKcs, XRCC4, ligase 4 (Lig4), and Artemis (reviewed in Mills et al.,2003; Rooney et al., 2004). Cells isolated from patients that are unable tocomplete RAG-initiated V(D)J recombination of transiently transfected plas-mid substrates and exhibit IR sensitivity have implicated a seventh potentialmember of the NHEJ group, as genetic analyses have ruled out defects inKu70, Ku80, DNA-PKcs, Artemis, XRCC4, or Lig4 (Dai et al., 2003).

2.3.1. Ku70 and Ku80

Ku70 and Ku80 form a heterodimer (Ku) that directly associates with DNADSBs as well as telomeric regions of chromosomes (reviewed in Critchlow andJackson, 1998). Ku-deficient cell lines are IR sensitive and defective in both CJand SJ formation of transiently transfected recombination substrates, demon-strating that these proteins are essential for normal V(D)J recombination (Guet al., 1997; Nussenzweig et al., 1996; Taccioli et al., 1993, 1994; Zhu et al.,1996). Potential functions for Ku during NHEJ include (1) the protection ofDNA ends generated during V(D)J recombination from unwanted processingor degradation (Boulton and Jackson, 1996; de Vries et al., 1989; Getts andStamato, 1994), (2) the juxtaposition of DNA ends produced by RAG cleavagebefore religation (Boulton and Jackson, 1996; Cary et al., 1997), and (3) therecruitment or activation of DNA repair or DNA damage-sensing proteins(Gottlieb and Jackson, 1993; Lieber et al., 1997; Ramsden and Gellert, 1998;West et al., 1998).

2.3.2. DNA-PKcs and Artemis

DNA-PKcs is a serine/threonine kinase and member of the phosphatidylino-sitol-3-kinase (PI-3 kinase) family that includes the DNA damage responseproteins ataxia telangiectasia mutated (ATM) and ataxia telangiectasia related(ATR) (reviewed in Smith and Jackson, 1999). DNA-PKcs–deficient cell linesdisplay varying degrees of IR sensitivity (Gao et al., 1998; Lees-Miller et al.,

60 darryll d. dudley ET AL.

1995; Taccioli et al., 1998). Furthermore, cell lines derived from SCID mice,which harbor a mutation in DNA-PKcs (Blunt et al., 1995), are severelyimpaired in their ability to form CJs, although SJ formation is relativelyunaffected in these cells (Blackwell et al., 1989; Lieber et al., 1988; Malynnet al., 1988). DNA-PKcs–deficient embryonic stem (ES) cells fail to make CJsbut make fully normal SJs (Gao et al., 1998); however, mouse embryonicfibroblasts from DNA-PKcs–deficient mice, which are also fully deficient forCJ formation, are also somewhat impaired in SJ formation, with many RS joinsin such cells harboring abnormal deletions (Bogue et al., 1998; Errami et al.,1998; Fukumura et al., 1998, 2000; Gao et al., 1998; Priestley et al., 1998).Therefore, although not fully required, DNA-PKcs has some unknown role inSJ formation, and this role is independent of Artemis (see below) and may besubstituted by other factors (e.g., in ES cells). Finally, hairpin-coding endswere shown to accumulate in lymphocytes derived from DNA-PKcs–deficientmice (Roth et al., 1992), thus implicating a function for DNA-PKcs in theprocessing of these V(D)J intermediates.

More recently, DNA-PKcs has been shown to phosphorylate Artemis, an-other member of the NHEJ repair pathway required for the formation of CJsbut not SJs (Ma et al., 2002; Nicolas et al., 1998; Rooney et al., 2002). Thephosphorylated form of Artemis has an endonuclease activity that in vitro iscapable of opening DNA hairpins produced by RAGs (Ma et al., 2002).Sequences of CJ junctions generated from both DNA-PKcs– and Artemis-deficient cells show an increased rate of P-nucleotide additions (Rooneyet al., 2003) consistent with aberrant opening of the hairpin ends (Kienkeret al., 1991; Lewis, 1994b; Rooney et al., 2002; Schuler et al., 1991). In contrastto DNA-PKcs deficiency, SJ formation is normal (both in quantity and quality)in all types of Artemis-deficient cells examined (Noordzij et al., 2003; Rooneyet al., 2003), again supporting a non-Artemis–mediated role for DNA-PKcs inV(D)J recombination and NHEJ.

2.3.3. XRCC4 and DNA Ligase 4

The role of XRCC4 in V(D)J was discovered by expression cloning via com-plementation of an IR-sensitive, V(D)J recombination-defective hamster cellline with a human cDNA library, which was shown to completely complementall IR sensitivity and V(D)J recombination defects of this line (Li et al., 1995).XRCC4 was then shown to associate with DNA Lig4 in vitro (Critchlow et al.,1997; Grawunder et al., 1997). XRCC4- and Lig4-deficient cells were bothshown to exhibit IR sensitivity and an inability to generate either SJs or CJs(Frank et al., 1998; Gao et al., 1998). Thus DNA Lig4 in association withXRCC4 rejoins the four broken DNA ends generated by RAG-mediatedcleavage and likely has a similar role in NHEJ in general.

v(d)j versus class switch recombination 61

3. Regulation of V(D)J Recombination

3.1. RAG-1 and RAG-2 Expression

3.1.1. Lymphoid-Specific Expression of RAGs

The expression of RAG-1 and RAG-2 is predominantly limited to developinglymphocytes, although low levels of RNA transcripts have been detected in themurine central nervous system and in peripheral lymphoid tissues (reviewed inNagaoka et al., 2000). In fact, transcription of the RAG-1 locus is detected inthe earliest lymphocyte progenitors isolated thus far from the bone marrow ofmice (Igarashi et al., 2002). However, the only known defect in RAG-deficientmice is a complete block in lymphocyte development, and therefore the RAGsdo not appear to play a role in the development or function of the centralnervous system (Mombaerts et al., 1992; Shinkai et al., 1992).

The RAG-2 promoter is lymphoid specific and differentially regulated in B andT cells, whereas the basal RAG-1 promoter in both mice and humans does notimpart lymphoid specificity (Lauring and Schlissel, 1999; Monroe et al., 1999a).A more distal element 50 of RAG-2 directs the coordinate and lymphoid-specificexpression of fluorescently tagged RAG-1 and RAG-2 from a bacterial artificialchromosome (BAC) transgene integrated into mice (Yu et al., 1999a). Moreover,RAG transcription appears to be regulated by different cis elements in B andT lymphocytes (Hsu et al., 2003), with both a silencer and antisilencer importantfor proper tissue- and stage-specific expression (Yannoutsos et al., 2004).

Several studies have detected RAG expression in mature B and T cells,leading to the hypothesis that RAGs could function to maintain self-tolerancevia secondary rearrangements of autoreactive receptors (Han et al., 1996;Hikida et al., 1996; Papavasiliou et al., 1997). However, targeted replacementof the RAG-2 gene with sequences encoding a RAG-2:GFP (green fluorescentprotein) fusion protein demonstrated that the low level of RAGs detected inthe periphery likely came from immature lymphocytes that had not completelyshut off RAG expression and yet migrated to the peripheral lymphoid tissues(Kuwata et al., 1999; Monroe et al., 1999b; Yu et al., 1999b). Moreover, RAG-2:GFP expression was not detected when GFP-negative lymphocytes isolatedfrom the periphery of RAG-2:GFP mice were adoptively transferred intoRAG-1–deficient host animals after immunization (Yu et al., 1999b), thusindicating that RAG-2 was not reexpressed in mature lymphocytes.

3.1.2. Allelic Exclusion and Feedback Regulation

Coincident with the onset of RAG expression, IgH rearrangements begin inB220þCD43þckitþCD19þ progenitor B cells (pro-B), and TCRb, TCRg, andTCRd rearrangements take place in CD4�CD8� double-negative (DN) T cells

62 darryll d. dudley ET AL.

(reviewed in Willerford et al., 1996). Because of the inherently imprecisenature of CJs, only one in three rearrangements will be in-frame and capableof expressing a functional protein. Theoretically, lymphocytes could make adifferent receptor chain for each allele and express multiple receptors, each ofa different specificity. However, almost all B cells express the functionalproducts of only one IgH allele and one IgL allele, and in mature ab T cellsonly one TCRb allele is functionally rearranged and expressed, a processreferred to as allelic exclusion (reviewed in Gorman and Alt, 1998; Kisielowand von Boehmer, 1995; Melchers et al., 1999). Thus for the IgH, Igk, Igl, andTCRb loci, only those cells in which the first V(D)J rearrangement is nonpro-ductive go on to rearrange their second allele, preventing the assembly ofmultiple antigen receptors in a single cell (Alt et al., 1984; Yancopoulos and Alt,1985). Both stochastic and regulated models have been proposed to explainallelic exclusion, but the absolute mechanism, which may involve differentmechanistic aspects for different loci, remains enigmatic; although it seemsclear that there must be some form of feedback regulation to prevent openingof the second allele (reviewed by Mostoslavsky et al., 2004). In this context,epigenetic factors, such as asynchronous replication, monoallelic demethyla-tion, and variegated, monoallelic transcriptional activation, which render onlya single allele capable of V-to-(D)J rearrangement initially (Liang et al., 2004;Mostoslavsky et al., 1998), may contribute to initiation of allelic exclusionbefore the feedback signals that maintain allelic exclusion in the face ofcontinued expression of RAG.

The product of a functionally rearranged IgH gene, mIgH, and expression ofa TCRb chain initiate signals that enforce feedback regulation at nonrear-ranged IgH and TCRb alleles, respectively, thus preventing further rearran-gements (reviewed in Gorman and Alt, 1998; Kisielow and von Boehmer,1995; Melchers et al., 1999). In this regard, mIgH associates with the surrogatelight chain proteins, Vpre-B and l5, to form the pre-B cell receptor (pre-BCR)(Melchers et al., 1993). A productive TCRb chain then associates with the pre-Ta protein to form the pre-T cell receptor (pre-TCR) (Saint-Ruf et al., 1994).Expression of the pre-BCR and pre-TCR provides the necessary signals tomediate feedback regulation, cellular expansion, and differentiation to theB220þCD43lo/ckit�CD19þ pre-B and CD4þCD8þ double-positive (DP)T-cell stages, respectively (reviewed in Kisielow and von Boehmer, 1995;Rolink et al., 2001; von Boehmer et al., 1999). Thus, introduction of rearrangedIgH (Spanopoulou et al., 1994; Young et al., 1994) or TCRb (Shinkai et al.,1993) transgene into an RAG-deficient background promotes development tothe pre-B or DP T-cell stages, respectively.

RAG expression is terminated during the phase of cellular proliferation thatoccurs during the transition from pro-B to pre-B in developing B cells, and

v(d)j versus class switch recombination 63

from DN to DP in developing T cells (Grawunder et al., 1995). To facilitatethis, RAG-2 is specifically tagged for degradation by cell cycle–dependentphosphorylation and ubiquitination (Lin and Desiderio, 1993; Mizuta et al.,2002). After the cellular proliferation signaled by expression of the pre-BCR orpre-TCR, RAGs are once again expressed, thus allowing rearrangement of IgLand TCRa genes in pre-B and DP T cells, respectively (reviewed in Nagaokaet al., 2000). During this second wave of RAG expression, further rearrange-ment of the IgH and TCRb loci does not occur, and therefore these loci havesomehow been rendered inaccessible to the RAGs (reviewed in Krangel,2003).

Individual B cells produce immunoglobulin containing either Igk or Igllight chains, but not both, referred to as IgL isotype exclusion (reviewed inMostoslavsky et al., 2004). In pre-B cells, the Igk locus is the first to rearrange,with subsequent rearrangement of Igl genes usually occurring only in cellsthat have failed to generate a productive Igk gene rearrangement (reviewed inGorman and Alt, 1998). The mechanism of sequential Igk versus Igl rear-rangement and whether it is regulated or stochastic also remain unsolvedproblems (reviewed in Mostoslavsky et al., 2004). Once a productive IgLchain is formed, it pairs with the previously rearranged IgH chain to create amature BCR in the form of IgM that is expressed at the surface of thedeveloping B cell. Surface expression of IgM provides further feedbackregulation and allelic exclusion, as well as signaling for the differentiation tothe immature B cell stage and beyond (reviewed in Rolink et al., 2001;Willerford et al., 1996). IgL rearrangements sometimes result in Igk and Iglprotein products that fail to associate with mIgH, and thus are functionallynonproductive (Alt et al., 1980).

Similarly, productive TCRa rearrangement in DP T cells allows for expres-sion of a functional TCRa/b complex that induces progression through posi-tive and negative selection steps that lead to the development of mature CD4þ

or CD8þ single-positive (SP) T cells (reviewed in Kisielow and von Boehmer,1995; Willerford et al., 1996). Unlike other antigen receptor gene loci, theTCRa, g, and d loci do not undergo feedback regulation or allelic exclusionat the level of gene rearrangement (Casanova et al., 1991; Davodeau et al., 1993;Malissen et al., 1988, 1992; Padovan et al., 1993; Sleckman et al., 1998).Preferential pairing of one TCRa chain to the expressed TCRb chain essen-tially prevents the expression of more than one antigen-specific receptor at thesurface of ab T cells carrying two productive TCRa rearrangements (reviewedin Fehling and von Boehmer, 1997; Malissen et al., 1992).

If a newly generated B cell expresses a productive but self-reactive receptor,signaling via the Ig receptor appears to prolong, or possibly reactivate, RAGexpression and allow further rearrangement of IgL chains (reviewed in

64 darryll d. dudley ET AL.

Nemazee and Weigert, 2000; Nussenzweig, 1998). This may lead to thereplacement of a self-reactive receptor with a non-self-reactive one in aprocess termed receptor editing (Gay et al., 1993; Radic et al., 1993; Tiegset al., 1993). There also is some evidence that suggests potential editing ofTCRb genes in peripheral T cells (reviewed by Mostoslavsky and Alt, 2004). Incontrast, both alleles of the TCRa locus appear to rearrange concomitantly inT cells and continue to rearrange throughout the DP T cell stage until positiveselection on self-MHC has successfully occurred (Borgulya et al., 1992;Malissen et al., 1988; Petrie et al., 1995).

3.1.3. Effect of Deregulated RAG Expression

Tight regulatory control over RAG expression is important for normal lympho-cyte development. Overall numbers of B and T cells are dramatically reducedin transgenic mice expressing RAGs continually during all stages of lymphocytedevelopment (Barreto et al., 2001; Wayne et al., 1994a). The reduction in T-cellnumbers reflects a selective reduction in ab T cells, with gd T cell developmentappearing relatively normal (Barreto et al., 2001). As ab T cells undergo a burstof proliferation after productive TCRb rearrangement and expression ofthe pre-TCR, the impairment would be consistent with p53-induced cellcycle arrest or apoptosis caused by an abundance of RAG-generated DNADSBs. Surprisingly, allelic exclusion is maintained even in mice continuallyexpressing RAGs throughout lymphocyte development, indicating that theregulation of antigen receptor gene accessibility is an extremely efficientprocess (Barreto et al., 2001; Wayne et al., 1994b). Finally, transgenic micethat ubiquitously express RAGs die prematurely and are significantly smallerthan control littermates, although it is unclear why (Barreto et al., 2001).

3.2. Regulated Accessibility of Antigen Receptor Gene Segments

3.2.1. Regulated Accessibility and V(D)J Recombination

IgH and TCRb rearrangements take place in an ordered fashion, with DH-to-JH (Alt et al., 1984) and Db-to-Jb (Born et al., 1985) rearrangements proceed-ing to completion on both IgH and TCRb alleles, respectively, before the onsetof subsequent V-to-DJ rearrangements. Lymphoid-restricted RAG expressionlimits V(D)J recombination to developing B and T lymphocytes but cannotaccount for the ordered or stage-specific rearrangement of immunoglobulinand TCR loci (Yancopoulos and Alt, 1985). Regulated gene accessibilityimparts ordered rearrangement of antigen receptor genes such that IgHgenes assemble before IgL genes in developing B cells, and TCRb genesassemble before TCRa genes in developing ab T cells (reviewed by Krangel,

v(d)j versus class switch recombination 65

2003; Mostoslavsky et al., 2003; Yancopoulos and Alt, 1986). Moreover, duringIgH and TCR b gene assembly, D-to-J rearrangements precede V-to-DJ rear-rangements and take place on both alleles before the onset of V-to-DJ rear-rangements (Alt et al., 1984; Born et al., 1985). In this regard, when nucleiisolated from RAG-deficient lymphocytes are incubated with RAGs in vitro,RS cleavage reflects the lineage and stage specificity of the cell from which thenuclei were isolated (Stanhope-Baker et al., 1996). Moreover, the efficiency ofRAG-mediated cleavage of RS-containing extrachromosomal substrates wasfound to be substantially reduced when the substrate was incorporated into anucleosome compared with the same substrate in the form of naked DNA(Golding et al., 1999; Kwon et al., 1998). Thus higher order chromatin struc-ture plays an integral role in the regulated accessibility of antigen receptorgene rearrangement (reviewed in Bassing et al., 2002b; Hesslein and Schatz,2001; Krangel, 2003).

3.2.2. Transcription and V(D)J Recombination

Recombinational activity of integrated immunoglobulin or TCR transgenes isoften largely dependent on the presence and function of associated transcrip-tional enhancer elements (reviewed in Ferrier et al., 1990; Krangel, 2003;Raulet et al., 1985; Sleckman et al., 1996). As the site of integration andnumber of integrated copies can influence the expression of transgenes,gene-targeted ablation of regulatory cis elements at endogenous loci wasused to assess more directly their function in recombination.

There are transcriptional enhancer and promoter elements associated withall antigen receptor loci (reviewed in Hempel et al., 1998; Hesslein and Schatz,2001). Early observations that V(D)J recombination generally correlated withthe appearance of germline transcripts at various antigen receptor loci impli-cated regulatory cis elements as playing a role in recombinational accessibility(Alessandrini and Desiderio, 1991; Fondell and Marcu, 1992; Goldman et al.,1993; Schlissel and Baltimore, 1989; Yancopoulos and Alt, 1985). In the IgHlocus, gene-targeted deletion of the intronic IgH enhancer (iEm) substantiallyreduced VH-to-DJH but not DH-to-JH rearrangement in developingB lymphocytes (Chen et al., 1993; Serwe and Sablitzky, 1993). In the Igklocus, elimination of both the intronic Igk (iEk) and 30 (30Ek) enhancerscompletely abolished Vk-to-Jk rearrangement (Inlay et al., 2002) and sepa-rate deletion of one or the other led to a substantial reduction in Vk-to-Jkrearrangement (Gorman et al., 1996; Xu et al., 1996).

In T cells, deletion of the TCRb enhancer (Eb) substantially reduced thelevel of TCR DJb and VbDJb transcripts, although some Vb-associated germ-line transcripts were still present (Bories et al., 1996; Bouvier et al., 1996;Mathieu et al., 2000). A corresponding decrease in levels of DJb and VbDJb

66 darryll d. dudley ET AL.

rearrangements in Eb�/� lymphocytes was also found, and overall ab T-celldevelopment was severely impaired (Bories et al., 1996; Bouvier et al., 1996).Gene-targeted ablation of the TCRa enhancer (Ea) resulted in a severereduction in germline Ja transcripts and TCRa rearrangement in developingT cells (Sleckman et al., 1997). On the other hand, deletion of Ea did notsubstantially alter the level of TCRd rearrangements in these same cells, eventhough TCRd gene segments are distributed among TCRa gene segments inthis locus (Sleckman et al., 1997). In contrast, mice carrying a deletion of theTCRd enhancer (Ed) had substantial impairment in VdDJd rearrangementsbut were normal for TCRa/b development (Monroe et al., 1999c). WhereasEd was required for TCRd transcripts in DN thymocytes, TCRd transcriptswere unaffected in the few gd T cells that developed in the absence of Ed

(Monroe et al., 1999c). Thus Ed differentially regulates early but not late gd

T-cell processes.Antigen receptor gene segment–associated germline promoters have also

been shown to affect V(D)J recombination. Deletion of the germline Db1promoter substantially reduced germline Db1 transcripts and Db-to-Jb1 rear-rangement levels but did not affect transcription or rearrangement involvingDb2/Jb2 gene segments (Whitehurst et al., 1999). Similarly, deletion of theT early a (TEA) germline promoter upstream of the 50-most Ja gene segmentseliminated germline transcripts associated with upstream Ja gene seg-ments and reduced Va-to-Ja rearrangements corresponding to these sameJa segments (Villey et al., 1996). However, germline Ja transcripts initiatingdownstream of TEA were detectable in thymocytes lacking TEA, and overalllevels of TCRa rearrangements were normal (Villey et al., 1996).

Regarding lineage specificity, targeted replacement of the TCRb enhancerwith iEm promoted transcription and rearrangement of the TCRb locus atlevels substantially higher than in developing T cells with Eb deleted (Borieset al., 1996; Bouvier et al., 1996), demonstrating that iEm can function out-side of the IgH locus to promote accessibility of heterologous sequences.VbDJb rearrangements did not occur at appreciable levels in B lineage cellscarrying iEm in place of Eb, although germline JbCb transcripts were readilydetectable in these same cells (Bories et al., 1996). However, when a largerregion encompassing Cb2 and Eb was replaced with iEm, significant levels ofDb-to-Jb rearrangements were detected in splenic B cells (Eyquem et al.,2002), suggesting that elements within the larger region have the ability tosuppress TCRb accessibility in B lineage cells. When Eb was replaced withthe TCRa enhancer (Ea), there was a significant reduction in Db germ-line transcripts in CD25þCD44�CD4�CD8� (DN3) thymocytes, the stageat which rearrangements of TCRb gene segments normally take place.However, levels of germline Db transcripts were normal in CD4þCD8þ

v(d)j versus class switch recombination 67

thymocytes carrying the Eb-to-Ea replacement, the stage during which TCRais normally active. Thus Ea affected TCRb transcription in a mannercorresponding to its normal functional stage. However, levels of Vb germlinetranscripts in DN3 thymocytes appeared normal, and overall levels of DJb andVbDJb rearrangements were only modestly reduced, demonstrating that Ea

could function to promote VbDJb recombination differently than it wouldnormally at the TCRa locus. Finally, TCRa rearrangements were dramaticallyreduced when Ea was replaced with either the TCRd enhancer (Ed) or iEm

(Bassing et al., 2003b), demonstrating that Ea must carry elements importantfor the regulated rearrangement of TCRa genes that cannot be replaced byother enhancer sequences. Thus, promoter/enhancer interactions clearlyinfluence transcription as well as RAG-mediated recombination at specificsites within antigen receptor loci (reviewed in Bassing et al., 2000; Krangel,2003).

3.2.3. Chromatin Modifications

Chromatin is made up of complexes of protein and DNA that allow thepackaging of approximately 1- to 10-cm lengths of unwound chromosomalDNA into a nucleus with a diameter of only 3–10mm (Alberts et al., 1983).The structure of chromatin begins with 146 bp of DNA wrapped around acomplex of histone proteins that form a nucleosome. Each nucleosome con-sists of eight histone molecules, two copies each of histone family proteins H3,H4, H2A, and H2B (Alberts et al., 1983). The histone protein H1 then linksnucleosomes into the higher ordered structure of 30-nm fibers, which arecondensed even further during interphase of the cell cycle (reviewed inBelmont et al., 1999). The regulation and control of higher order chromatinis an integral component of transcriptional activation and repression of eukary-otic genes (reviewed in Udvardy, 1999), as well as DNA replication (reviewedin Gerbi and Bielinsky, 2002). The epigenetic regulation of chromatin accessi-bility is associated with histone acetylation, phosphorylation, methylation, andubiquitination (reviewed in Berger, 2002), as well as DNA methylation(reviewed in Richards and Elgin, 2002).

Histone modifications have been associated with actively recombining extra-chromsomal substrates and endogenous antigen receptor loci (reviewed inKrangel, 2003; Oettinger, 2004). Treatment with histone deacetylase inhibitorshas been shown to induce RS accessibility and V(D)J recombination within theIgk, TCRg, and TCRb loci of cells otherwise inaccessible because of higherorder chromatin (Agata et al., 2001; Mathieu et al., 2000; McBlane and Boyes,2000). In addition, deletions of cis-regulatory elements necessary for endoge-nous V(D)J recombination have been linked to reduced levels of histoneacetylation of antigen receptor locus–associated sequences (Agata et al.,

68 darryll d. dudley ET AL.

2001; Mathieu et al., 2000). Furthermore, chromatin near the D b1J b1 regionof E b-deficient thymocytes, which have reduced levels of TCRb rearrange-ments (see above), contained alterations in histone acetylation, methylation,and phosphorylation (Spicuglia et al., 2002). These data are consistent with thehypothesis that such combinatorial interactions and modifications lead to anepigenetic regulatory system or a so-called ‘‘histone code’’ that in some mannermay promote recombinational accessibility at antigen receptor loci (reviewedin Bassing et al., 2002b; Jenuwein and Allis, 2001; Oettinger, 2004; Strahl andAllis, 2000). Methylation of specific lysine residues on histone H3 has alsobeen shown to correlate with recombinational activity at the IgH and TCR bloci (Morshead et al., 2003; Ng et al., 2003; Spicuglia et al., 2002; Su et al.,2003). However, targeted recruitment of a histone methyltransferase tochromosomal recombination substrates blocks transcription and recombina-tion of nearby segments (Osipovich et al., 2004), illuminating the complexity ofsuch regulation.

DNA hypomethylation at CpG dinucleotides has been shown to correlatewith transcription in general (Razin and Riggs, 1980), as well as with specificantigen receptor gene segments (Kelley et al., 1988; Mather and Perry, 1981,1983). Regarding recombinational accessibility, methylation at a single CpGsite within the 30 RS of Db1 did not allow cleavage by RAGs (Whitehurst et al.,2000). In addition, Vk-to-Jk rearrangements appear limited to hypomethylatedalleles, and thus DNA methylation may also play a role in allelic exclusion(Mostoslavsky et al., 1998). However, as not all actively recombining antigenreceptor loci display hypomethylated status (Villey et al., 1997), DNA methyl-ation does not always result in elevated recombinational accessibility (Cherryet al., 2000). Therefore, the overall state of recombinationally acces-sible antigen receptor gene segments likely involves a variety of interactingmodifications involving chromatin and DNA (reviewed in Krangel, 2003).

3.2.4. H2AX

Chromatin modifications along antigen receptor loci are also important formonitoring the chromosomal V(D)J recombination reaction to ensure thenormal NHEJ-mediated repair of RAG-generated DSBs. There are threesubfamilies of histone H2A, of which H2AX comprises 10–15% of total H2Aprotein in most mammalian cells (Mannironi et al., 1989). In response to IR-induced DNA DSBs, H2AX is phosphorylated on Ser-139, thus producingg-H2AX. g-H2AX is found in discrete foci at the site of DSBs, and these focioccur at a frequency comparable to the number of induced DSBs (Rogakouet al., 1999). Several DNA repair factors including Rad50, Rad51, andNijmegen breakage syndrome protein (NBS1) have been shown to colocalizewith g-H2AX after the induction of DSBs (Chen et al., 2000; Paull et al., 2000).

v(d)j versus class switch recombination 69

Both DNA-PKcs and ATM are able to phosphorylate H2AX in vitro and/orin vivo (Burma et al., 2001; Rogakou et al., 1998; Stiff et al., 2004).

g-H2AX and NBS1 have been shown to undergo RAG-dependent colocali-zation with the TCR a locus in normal thymocytes (Chen et al., 2000), andH2AX-deficient mice have a slight reduction in overall numbers of B andT lymphocytes (Bassing et al., 2003a; Celeste et al., 2002). Nevertheless, theformation of V(D)J-associated CJs and SJs was found to be normal in H2AX-deficient mice (Bassing et al., 2003a; Celeste et al., 2002) and from recombi-nation substrates transiently transfected into H2AX-deficient ES cells (Bassinget al., 2002a). Clearly, H2AX is not essential for the repair of RAG-inducedDNA DSBs. However, approximately 4% of nontransformed ab T cells fromH2AX-deficient mice contain potential TCR a/ d locus translocations (Celesteet al., 2002) and H2AX-deficient mice may exhibit an increased predispositionto thymic lymphomas with potential TCR a/d locus translocations (Bassinget al., 2003a). Thus, H2AX likely serves a critical role in the suppressionof aberrant V(D)J recombination, possibly through the proposed ‘‘anchoring’’function of H2AX in forming a nucleation site for a number of DNA–protein–protein–DNA interactions that might serve to stabilize synaptic complexes ofchromosomal RAG-cleaved antigen receptor loci (Bassing and Alt, 2004).

4. Class Switch Recombination Employs Distinct Mechanismsfor V(D)J Recombination

4.1. Overview of Class Switch Recombination andSomatic Hypermutation

The consequence of successful V(D)J recombination of IgH and IgL chains indeveloping B cells is the surface expression of IgM and/or IgD. Activation byantigen in the context of certain cytokine stimuli can induce the process ofCSR, whereby the V(D)JH exon initially associated with Cm exons is adjoinedto one of several groups of downstream CH exons (e.g., C g, Ce, and C a,referred to as CH genes) (Fig. 6). Recombination takes place between repeti-tive sequences, termed S regions, which lie just upstream of the various CH

genes (reviewed in Chaudhuri and Alt, 2004). The exchange in CH genes altersthe isotype of expressed antibody from IgM to either IgG, IgE, or IgA, alongwith associated changes in effector function, while maintaining antigen-binding specificity (reviewed in Manis et al., 2002b). DNA sequences locatedbetween the recombining S regions can be detected in the form of circularizedDNA that has been excised from the genome of the effected B cell (Iwasatoet al., 1990). The liberation of circular DNA in CSR is consistent withthe participation of DSB intermediates, analogous to excised SJs in V(D)J

Figure 6 Schematic diagram of the murine IgH locus before and after class switch recombinationbetween Sm and Se. The VH gene segments are depicted as shaded rectangles, the DH segments assolid rectangles, and the JH segments as open rectangles. The Sm regions exons are shown asstriped ovals and constant region exons as solid squares. The position of the iEm and 30 RRenhancers are indicated by diamonds. The positions of the VH and I exon promoters are shownas solid circles. Distances between the various elements are not drawn to scale.

70 darryll d. dudley ET AL.

recombination that require synapsis and repair over appreciable chromo-somal distances. CSR is a mature B-cell–specific process and, unlike V(D)Jrecombination, does not occur in T lineage cells.

In addition to CSR, activation of mature B cells in the context of a germinalcenter reaction can introduce mutations at a high rate (10 �3 to 10�4 per basepair per generation) into assembled IgH and IgL variable region exons via aprocess called SHM. Selection of B cells in which mutated V regions createan antigen receptor of higher affinity than the original results in ‘‘affinitymaturation’’ and the generation of a more effective immune response.

CSR and SHM rely on the activity of the aicda gene, which encodes activa-tion-induced deaminase (AID), which in turn deaminates cytidine residues onDNA and, thus, forms dU/dG mismatched DNA base pairs (Bransteitter et al.,2003; Chaudhuri et al., 2003; Dickerson et al., 2003; Petersen-Mahrt et al., 2002;Pham et al., 2003; Sohail et al., 2003; Yu et al., 2004). CSR and SHM likelyproceed via subsequent excision of the mismatched dU by the base excisionrepair protein uracil DNA glycosylase (UNG). UNG creates an abasic site, anddifferential repair of this lesion apparently leads to either SHM or CSR (Di Noiaand Neuberger, 2002; Petersen-Mahrt et al., 2002; Rada et al., 2002b). Themismatch repair (MMR) proteins Msh2/Msh6 can also bind and process the

v(d)j versus class switch recombination 71

dU:dG mismatch and contribute to the CSR and SHM. In this regard, UNG andMMR deficiencies can impair CSR and alter the spectrum of mutations sus-tained by V genes during SHM. Intriguingly, however, some UNG mutants thathave lost the uracil glycosylase activity retain the ability to mediate CSR in mice,leading to the suggestion that UNG may participate in an as yet unidentifiedmanner in CSR that extends beyond its known enzymatic activity (Begum et al.,2004a).

Transcription of target S region or variable region target sequences isessential for both CSR (Bottaro et al., 1994; Zhang et al., 1993) and SHM(Bachl et al., 2001; Betz et al., 1994; Fukita et al., 1998; Peters and Storb,1996). Each CH gene is organized into a transcriptional unit differentiallyregulated by cytokine-specific transcription factors, thus providing the neces-sary specificity for directing isotype-specific switching (Fig. 6) (reviewed inManis et al., 2002b; Stavnezer, 2000). AID deaminates cytidines of ssDNAin vitro (Bransteitter et al., 2003; Chaudhuri et al., 2003; Dickerson et al.,2003; Pham et al., 2003) but requires transcription of double-stranded dsDNAto generate an appropriate ssDNA substrate (Chaudhuri et al., 2003; Ramiroet al., 2003). Thus, transcription provides an appropriate target substrate forAID to act on (Chaudhuri et al., 2003; Ramiro et al., 2003). The remainder ofthis review focuses on CSR, although contrasts will be made between CSR andSHM where appropriate.

4.2. Organization of Heavy Chain Constant Region Genes

In mice there are eight CH genes located on chromosome 12 and positioneddownstream of the antigen receptor gene segments in a region spanningapproximately 200 kb (Fig. 6). Cm and Cd are the most JH-proximal group ofCH genes, and Cm is the first to associate with a functionally rearrangedV(D)JH exon, and therefore IgM is the first isotype expressed on the surfaceof a B cell. Later in development, expression of IgD occurs via alternative RNAsplicing of the V(D)JH exon to the Cd exons. Thus, IgM and IgD are oftensimultaneously expressed on the surface of the same B cell. Expression of allother immunoglobulin isotypes requires CSR between Sm and downstreamS region sequences (e.g., Sg, Se, or Sa), with subsequent loss of the interven-ing Cm and Cd genes (reviewed in Chaudhuri and Alt, 2004). All CH genes,except Cd, are integrated into transcriptional units consisting of an intervening(I) exon (Lutzker and Alt, 1988), S region, CH exons, and downstream poly-adenylation signal sequences corresponding to the membrane and secretedversions of immunoglobulin. Finally, a region downstream of Ca containingenhancer elements, referred to as the 30 regulatory region (30 RR), is important

72 darryll d. dudley ET AL.

for the production of germline transcripts (GTs) and CSR to the various CH

genes except C g1 (reviewed in Manis et al., 2002b).

4.3. Regulation of Class Switch Recombination

4.3.1. B-C ell Activation and Clas s Switc h Recomb ination

V(D)J recombination and early B-cell development take place in the bonemarrow. On successful expression of surface immunoglobulin in the form ofIgM, immature B cells migrate from the bone marrow to peripheral lymphoidtissues located in the spleen, lymph nodes, and gut-associated lymphoid tissue.Here, at discrete anatomic sites referred to as germinal centers, and often inassociation with T cells, B cells encounter antigen and undergo antigen-drivenclonal expansion that can lead to CSR and/or SHM; although CSR(Macpherson et al., 2001) and SHM (William et al., 2002) can also take placeoutside of germinal centers as well.

CSR is induced in vivo by both T-dependent (TD) and T-independent (TI)antigens. B-cell activation by TD antigens requires interaction of CD40 ligandexpressed on activated T cells and CD40 on the surface of B cells. T-indepen-dent antigens can activate B cells in the absence of direct T- and B-cellinteractions. Type 1 TI antigens, such as lipopolysaccharide (LPS), can act aspolyclonal B-cell activators at high concentrations and are able to activateB cells in the complete absence of T cells. Type 2 TI antigens, on the otherhand, which usually consist of highly repetitious molecules, do not requiredirect B- and T-cell interactions to induce B-cell activation or CSR, althoughB-cell activation and CSR occur inefficiently in the absence of T-cell–derivedcytokines. TD antigen stimulation can be mimicked in vitro by culturing B cellsin the presence of anti-CD40 along with specific cytokines, and TI activationcan be mimicked by treatment with LPS plus or minus the addition of specificcytokines (reviewed in Manis et al., 2002b; Stavnezer, 2000). In concert withantigen-dependent activation, cytokine-induced signaling provides specificityto CSR (reviewed in Manis et al., 2002b; Stavnezer, 2000). For instance, LPSinduces isotype switching to IgG2b and IgG3, whereas LPS plus interleukin 4(IL-4) induces isotype switching to IgG1 and IgE (reviewed in Manis et al.,2002b; Stavnezer, 2000).

4.3.2. Ger mline CH Tr anscrip ts

Isotype switching to a particular CH gene is preceded by transcription of thecorresponding germline sequences (reviewed by Chaudhuri and Alt, 2004;Manis et al., 2002b; Stavnezer, 2000). Transcripts initiate upstream of I exonsfound 50 of each CH gene and terminate at polyadenylation sites located

v(d)j versus class switch recombination 73

downstream of CH genes (Lutzker and Alt, 1988). These transcripts undergoRNA splicing between the I and CH exons to form processed GTs in which theintervening S region sequences are deleted (reviewed in Chaudhuri and Alt,2004). Transcription is regulated by I region promoter sequences that areactivated by CD40-, LPS-, and cytokine-mediated signals (Fig. 6) (reviewedin Stavnezer, 2000).

Deletion of I region promoters and subsequent loss of recombinationinvolving the associated CH gene demonstrated the dependence on S regiontranscription for CSR (Jung et al., 1993; Lorenz et al., 1995; Zhang et al.,1993). Furthermore, targeted replacement of I promoters with heterologous,constitutively active promoters rescues S region transcription and directsisotype switching to the corresponding CH genes (Bottaro et al., 1994;Lorenz et al., 1995; Seidl et al., 1998). GTs do not encode functional proteins,but several studies have implicated RNA splicing and processing as potentiallyplaying a role in the CSR process, as targeted mutations in specific splice sitesstrongly inhibited CSR to the corresponding CH gene (Hein et al., 1998;Lorenz et al., 1995). However, it now seems clear that the major role of GTsis to provide the appropriate ssDNA substrate for AID (see Sections 4.4 and4.5, below) (reviewed in Chaudhuri and Alt, 2004).

In the IgH locus, VH-to-DJH rearrangements are impaired in the absence ofthe intronic enhancer (iE m) that lies in the intronic region between the JH

gene segments and Cm just uptream of Sm; thus it appears that iE m serves anessential role in promoting VH-to-DJH rearrangement in developing B cells(Sakai et al., 1999a; Serwe and Sablitzky, 1993). In this regard, CSR isalso reduced in the absence of iEm, which serves a promoter function forI m-C m–containing GTs (Bottaro et al., 1998; Gu et al., 1993; Sakai et al., 1999a;Su and Kadesch, 1990). Conceivably, CSR that occurs in the absence of iEm

may be mediated by transcription of Im sequences initiated by heterologouspromoters such as those associated with VH or DH gene segments (Gu et al.,1993; Kuzin et al., 2000). Notably, removal of iEm along with a large region ofpotential upstream promoters resulted in reduced Sm recombination, but leftsubstantial recombination within downstream S regions, which manifest asinternal S region deletions (Gu et al., 1993). Thus, iEm does not appear tofunction as an essential enhancer element for promoting CSR to downstreamCH genes, which may be more dependent on the activity of the 30RR (seebelow).

The production of GTs is influenced by enhancer-like elements located inthe region found approximately 15 kb downstream of Ca. The 30RR is an �40-kb region composed of four elements corresponding to hypersensitivity sites(50-HS3a-HS1,2-HS3b-HS4-30), which in various combinations have beenshown to possess locus control region (LCR)–like activity (reviewed in

74 darryll d. dudley ET AL.

Madisen and Groudine, 1994; Manis et al., 2002b). Targeted replacement ofHS1,2 with a pgk-driven neomycin resistance gene cassette (pgk neor) disruptsthe production of GTs and CSR to all CH genes except IgG1 and IgA, some ofwhich are located as far as 120 kb upstream (Cogne et al., 1994). This suggeststhat either I promoter–driven transcription of CH genes is dependent on thissite, or that pgk neor blocked the activity of an unknown downstream element(Cogne et al., 1994; Manis et al., 1998b). In support of the latter possibility,production of GTs and CSR was rescued when the neor cassette was deletedfrom the locus, leaving a loxP site in place of HS1,2 (Manis et al., 1998b).Similar studies demonstrated that HS3a is also dispensable for CSR (Maniset al., 1998b). Furthermore, targeted insertion of the pgk neor cassette atseveral locations throughout the IgH locus inhibited the production of GTsand CSR of CH genes located upstream of the pgk neor insertion, but notdownstream (Seidl et al., 1999). Thus the pgk neor cassette appears to prefer-entially compete with CH promoters for an enhancer activity found in theregion downstream of HS1,2. Clean deletion of HS3b and HS4 together hasthe identical effect as the targeted replacement of HS1,2 or HS3a with the pgkneor cassette, namely a significant reduction in the production of all immuno-globulin isotypes except IgM and IgG1, with a modest reduction in IgA(Pinaud et al., 2001). Finally, insertion of the pgk neor cassette downstreamof HS4 did not reduce CSR or the production of GTs, which, together with theother data, strongly implies that the enhancer activity is contained within theHS3b or HS4 sequences, and possibly both (Manis et al., 2003).

4.3.3. Class Switch Recombination and Somatic HypermutationAre Region-Specific Events

CSR is targeted to S regions, found upstream of each CH gene, that consist of1–12 kb of repetitive sequences (Fig. 6) (Kataoka et al., 1980). Each uniqueS region is made up of tandem repeat units varying in length between 5 bp(Sm) and 80 bp (Sa) (reviewed in Honjo et al., 2002) with some varying degreesof overall homology (reviewed in Stavnezer, 1996). In contrast to the site-specific cleavage mediated by RAGs at RSs in V(D)J recombination, CSRbetween two S regions can occur throughout, and even outside, the corerepeat sequences (Dunnick et al., 1993; Lee et al., 1998; Luby et al., 2001).Sequencing the break points of CSR junctions failed to identify a consensustarget sequence either in relation to the overall S regions or in the context ofthe short pentameric repeats, confirming that CSR is a region-specific ratherthan site-specific event (Dunnick et al., 1993; Lee et al., 1998).

The lack of consensus sequences at CSR junctions suggests that rather thandepending on sequence-specific recognition, the CSR ‘‘recombinase’’ mightinstead be targeted via the formation of S region–specific higher order

v(d)j versus class switch recombination 75

structures. The S regions of most vertebrates are rich in G and C nucleotides,with a bias for purine nucleotides on the nontemplate strand of DNA(Shinkura et al., 2003). Such properties have been shown to promote theformation of RNA–DNA hybrids (Mizuta et al., 2003; Reaban and Griffin,1990; Reaban et al., 1994; Tian and Alt, 2000) that could lead to the generationof structures such as R loops (Tian and Alt, 2000; Shinkura et al., 2003; Yu,2003) and G quartets (Dempsey et al., 1999; Sen and Gilbert, 1988). Incontrast, the S region sequences in Xenopus contain approximately 60%A and T nucleotides, consistent with the overall nucleotide composition ofthe Xenopus genome (Mubmann et al., 1997). However, Xenopus S regions arehighly repetitive and contain palindromic sequences similar to those thattarget SHM, which conceivably could effect CSR in the absence of potentialhigher order structures. (Mubmann et al., 1997; Zarrin et al., 2004).

Deletion of the core tandem repeat sequences of Sm significantly reducedCSR in mice (Luby et al., 2001). The remaining level of CSR detected in thesemice might be due to the retention of a considerable amount of G-rich andshort palindromic sequences left upstream of Cm (Luby et al., 2001), althoughcomplete deletion of all Sm tandem repeats further reduces, but does noteliminate, CSR (Khamlichi et al., 2004). Thus, transcriptional activation fromthe iEm enhancer may by itself be enough to induce low levels of recombina-tion (Khamlichi et al., 2004). On the other hand, complete deletion of Sg1sequences in mice essentially blocks CSR to the deleted allele (Shinkura et al.,2003). Finally, the lack of an associated S region apparently prevents the usageof the cCg gene in humans, as it is located in the correct transcriptionalorientation and lacks mutations in its coding sequence that result in frameshiftor stop codons or would otherwise prevent functional expression (Bensmanaet al., 1988). Therefore, the presence of an S region appears critical for normalCSR. Potential functions of mammalian S regions are discussed in more detailbelow.

SHM is also region specific, as mutations begin in the region just down-stream of an IgH or IgL V promoter, are found throughout the variable regionexons, and are detected as far as 2 kb downstream of V promoters within theintronic region between J and C exons (reviewed in Harris et al., 1999).However, most of the introduced mutations occur within the assembled vari-able regions that form the antigen-binding portion of an antibody molecule orin nearby flanking sequences (reviewed in Harris et al., 1999; Papavasiliou andSchatz, 2002b). Mutations are frequently associated with RGYW sequencemotifs (where R is A or G, Y is C or T, and W is A or T). The most commonchanges are point mutations, with transitions being slightly favoredover transversions, although small deletions and duplications are also detected.Targeting to RGYW sequences likely reflects specific recruitment and

76 darryll d. dudley ET AL.

specificity of the recombinase machinery (Chaudhuri et al., 2004; seebelow), whereas mutations concentrated in sequences making up antigen-binding regions are partly the result of the selection for high-affinity antigenreceptors in response to antigen during affinity maturation (Griffiths et al.,1984).

4.3.4. Indu ced Muta tions in and Around S Regi on Sequenc es

Sequencing of CSR junctions reveals frequent DNA alterations in the form ofsingle-nucleotide mutations and small deletions (Dunnick et al., 1993; Leeet al., 1998). These mutations are AID dependent (Petersen et al., 2001) andcan be found both 5 0 and 30 of the Sm region in wild-type B cells activated for,but that have not undergone, CSR (Dudley et al., 2002; Nagaoka et al., 2002;Petersen et al., 2001). CSR is therefore frequently associated with mutationsthat resemble those induced by SHM, likely reflecting the common AIDdeamination event in the initiation of both CSR and SHM. Thus, the resolu-tion of a common DNA lesion generated by AID by different downstreamrepair pathways could result in either DNA recombination or mutation(reviewed in Chaudhuri and Alt, 2004).

4.4. Activation-Induced Cytidine Deaminase

4.4.1. Discovery and Isolation of Activation-Induced Cytidine Deaminase

The discovery of AID and the subsequent demonstration of its essential role inCSR and SHM have led to rapid advances toward the elucidation of themechanisms that effect CSR and SHM (reviewed in Kenter, 2003; Reynaudet al., 2003). AID was isolated via a subtractive cloning screen from a murineB-cell line (CH12) that on activation switches from IgM to IgA (Muramatsuet al., 1999). Expression of AID is limited to developing germinal center B cells(Muramatsu et al., 1999) and can be induced in vitro by culturing splenicB-cells in the presence of activating stimuli known to induce CSR (Muramatsuet al., 1999). AID deficiency completely abrogates CSR and SHM in bothhumans (Revy et al., 2000) and mice (Muramatsu et al., 2000), and expressionof AID in nonlymphoid cell lines induces at least limited CSR (Okazaki et al.,2002) and SHM (Martin et al., 2002b; Yoshikawa et al., 2002). Furthermore,overexpression of AID in bacteria can lead to mutations in several transcribedgenes (Petersen-Mahrt et al., 2002b; Ramiro et al., 2003). Thus, analogous toRAGs with respect to V(D)J recombination, AID is both necessary and suffi-cient to effect CSR and SHM in the context of proteins expressed in non-lymphoid cells. It is to be noted, however, that the rate of CSR and the

v(d)j versus class switch recombination 77

spectrum of mutations observed in nonlymphoid cells on artificial substratesdo not faithfully recapitulate that observed at endogenous loci in B cells,suggesting that B-cell–specific factors and/or AID modifications (see below)contribute to these processes.

4.4.2. Activation-Induced Cytidine Deaminase Expression

High-level expression of AID via stable integration of an AID transgene caninduce mutations within actively transcribed nonimmunoglobulin genes, in-cluding the AID transgene itself, both in lymphoid and nonlymphoid cell lines(Martin and Scharff, 2002b; Yoshikawa et al., 2002). Constitutive and ubiqui-tous expression of AID in mice via a transgene leads to T-cell lymphomas andadenosarcomas (Okazaki et al., 2003) and deregulated AID expression hasbeen detected in several types of human non-Hodgkin lymphomas (Greeveet al., 2003; Hardianti et al., 2004a,b). Moreover, mature human and mouseB lineage tumors often have translocations that fuse S regions with oncogeneloci (reviewed by Mills et al., 2003). Work has shown, in mouse models, thatsuch translocations are dependent on AID (Ramiro et al., 2004). Thus, tightregulatory control of AID expression is necessary to prevent generalizedgenomic mutations and genomic instability.

AID is expressed in activated B lymphocytes in the context of a germinalcenter reaction, precisely in those cells that undergo SHM and CSR in vivo(Muramatsu et al., 1999). Expression of AID is modulated by inhibitors ofdifferentiation (Id) proteins, as ectopic expression of Id2 or Id3 reduces AIDexpression in activated splenic B cells and inhibits CSR (Gonda et al., 2003;Sayegh et al., 2003). Id proteins are best known as antagonists of the E familyof transcription factors (E proteins), a class of basic helix–loop–helix proteinsthat bind DNA at conserved E box sites as homo- and heterodimers (reviewedin Quong et al., 2002; Sun, 2004). Id proteins form heterodimers withE proteins that are unable to bind DNA, thus negatively regulating transcrip-tional activation by E proteins (Benezra et al., 1990a,b; Christy et al., 1991;Riechmann et al., 1994; Sun et al., 1991). Id proteins have also been shown tointeract with members of Pax and Ets families of transcription factors, likewiseinhibiting their DNA-binding functions (Roberts et al., 2001; Yates et al.,1999). E12, E47, and Pax5 are vital for B-cell development (Bain et al.,1994; Urbanek et al., 1994; Zhuang et al., 1994), and their expression is highlyinduced in mature B cells by CSR-inducing stimuli (Gonda et al., 2003; Quonget al., 1999). E47 and Pax5 have both been shown to bind regulatory elementsupstream of AID in vivo (Gonda et al., 2003; Sayegh et al., 2003), and the Pax5element was shown to be essential for AID gene expression (Gonda et al.,2003).

78 darryll d. dudley ET AL.

4.4.3. Ac tivation- Induced Cytidine Deamin ase Deaminates dCResidue s of Sin gle-Strande d D NA Substrate s: Interac tionwith Replication Prote in A

AID is a ssDNA cytidine deaminase (Bransteitter et al., 2003; Chaudhuri et al.,2003; Dickerson et al., 2003; Pham et al., 2003; Sohail et al., 2003) and suchssDNA substrates are probably revealed during transcription through theS regions (reviewed in Chaudhuri and Alt, 2004). Transcription of S regions,in their physiological orientation, generates ssDNA in the context of R loopsin vitro (Shinkura et al., 2003; Tian and Alt, 2000; Yu et al., 2003) and in vivo(Yu et al., 2003) and such transcribed DNA can serve as targets of AIDdeamination in vitro (Chaudhuri and Alt, 2004; see below).

In addition to R loops, other mechanisms may operate to target AID activityto S regions (see below). In this regard, variable region exons do not haverepetitive sequences or unusual GC content that would lead to R loop forma-tion and do not form R loops when transcribed in vitro (reviewed inChaudhuri and Alt, 2004; Papavasiliou and Schatz, 2002b), yet they aretargeted by AID during SHM. These observations suggested a specific cofactorto target AID during SHM and such a cofactor was identified (Chaudhuri et al.,2004) as replication protein A (RPA), a heterotrimeric ssDNA-binding proteininvolved in replication, recombination, and repair (reviewed in Wold, 1997).RPA stabilizes ssDNA (Wold, 1997) and can bind short stretches of ssDNAbubbles and recruit nucleotide excision and base excision repair proteins(reviewed by Binz et al., 2004; Matsunaga et al., 1996). AID forms a specificcomplex with RPA that facilitates AID-induced DNA deamination of tran-scribed RGYW-containing substrates (Chaudhuri et al., 2004). The efficiencyof substrate binding and deamination by RPA�AID complexes was dependenton the number of RGYW motifs, and deamination was observed at or aroundthese sequences (Chaudhuri et al., 2004). Thus RPA likely functions to targetAID to transcribed SHM hot spots found in V region exons of IgH genes.

Significantly, the AID�RPA complex is B-cell specific, and this specificityappears regulated, at least in part, by the phosphorylation status of AID inB cells (Chaudhuri et al., 2004). Other findings support the notion that theAID�RPA complex may also be involved in CSR, particularly in the context ofXenopus S regions that lack R loop-forming ability but contain regions ofRGYW motifs (Zarrin et al., 2004; see below). Also, it is possible that RPAthat remains bound to the deaminated mammalian S region substrate, asproposed for SHM, also can actively recruit proteins that are downstream ofdeamination, such as UNG and MMR proteins, to the site of initial DNAlesions in CSR.

v(d)j versus class switch recombination 79

4.4.4. Apobec1 and Other Deaminase Family Members

AID shares significant sequence homology with Apobec-1 (34% amino acididentity), a known cytidine deaminase (Muramatsu et al., 1999), and has beenshown to catalyze the deamination of free CTP nucleotides in vitro(Muramatsu et al., 1999; Papavasiliou and Schatz, 2002a). Apobec-1 functionsas an RNA-editing enzyme, inducing a C-to-U conversion at position 6666 ofthe ApoB mRNA transcript, hence changing Gln-2153 into an in-frame stopcodon (reviewed in Chan et al., 1997). The edited transcript encodes ApoB-48,a protein that, although colinear with the N-terminal 2152 residues of full-length ApoB-100, has significantly altered biological function (reviewed inChan, 1992). The shared homology with Apobec-1 led to the proposal thatAID may edit an mRNA transcript of unknown function, thus generating anovel class switch recombinase and/or V region mutator (Muramatsu et al.,2000). This model, which contrasts with most data arguing for a DNA deami-nation activity for AID in CSR and SHM, was supported, albeit quite indirect-ly, by the finding that de novo protein synthesis is required for AID to induceCSR (Begum et al., 2004b; Doi et al., 2003).

The cytidine deaminase ApoBec3G acts as an inhibitor of the humanimmunodeficiency virus type (HIV-1) retrovirus, not by mutating the genomicviral RNA or RNA transcripts, but by introducing dG-to-dA mutations intothe newly synthesized viral DNA (reviewed in Neuberger et al., 2003).Furthermore, although RNA is the physiological substrate for Apobec-1,AID (Bransteitter et al., 2003; Chaudhuri et al., 2003; Dickerson et al.,2003; Ramiro et al., 2003) and other Apobec-1 family members (Harriset al., 2002; Petersen-Mahrt and Neuberger, 2003) can deaminate dC resi-dues of nontranscribed ssDNA or transcribed dsDNA substrates in vitro.Furthermore, overexpression of Apobec-1 in bacteria, as well as otherApobec-1 family members including AID, can lead to dC-to-dG mutations inbacterial DNA (Harris et al., 2002; Petersen-Mahrt et al., 2002). These muta-tions were dependent on the catalytic function of the transfected deaminasevectors, as mutations in Znþ coordination motifs required for deaminaseactivity abolished this mutagenic effect (Harris et al., 2002). Thus Apobec-1family members including AID, but not Apobec-1 itself, function via a DNAdeamination process that is dependent on ssDNA, rather than the previouslyproposed RNA-editing model, further weakening the argument that AID actsvia RNA editing. Notably, Apobec-1 is unable to induce CSR or SHM whenoverexpressed in B cells, as does AID (Eto et al., 2003; Fugmann et al., 2004),although Apobec-1 can function in vitro to deaminate DNA; it is conceivablethat lack of such complementing activity may, in part, reflect inability to recruitcofactors such as RPA. Overall, current evidence suggests the possibility that

80 darryll d. dudley ET AL.

AID may well have evolved from a family of DNA-editing proteins, withApobec-1 being the outlying protein that may have evolved a new RNA-editingfunction.

4.5. Mechanism of Class Switch Recombination andSomatic Hypermutation

4.5.1. S Region Transcription and Activation-Induced CytidineDeaminase Substrate Formation

Mammalian S regions display a strong G-rich nontemplate strand bias(reviewed in Manis et al., 2002b). In this regard, inversion of the Sg1 regionin vivo significantly reduces recombination of the corresponding allele(Shinkura et al., 2003). This strongly suggests that CSR, at least to Cg1, isinfluenced by the transcriptional orientation of S region sequences in vivo.Targeted replacement of Sg1 sequences with randomly generated purine- orpyrimidine-rich sequences supported these findings (Shinkura et al., 2003).Replacement of Sg1 with a 1-kb sequence that when transcribed produces ahighly purine-rich transcript was able to target recombination to Cg1, albeit ata reduced level compared with the endogenous Sg1 sequences (Shinkura et al.,2003). The reduction in CSR efficiency is likely due in significant part to thedifference in overall length of available target sequences (1 kb for the synthe-sized regions compared with 12 kb for the endogenous Sg1 region) (A. Zarrin,and F. Alt, unpublished data). When this 1-kb sequence was inverted, so thatpyrimidine-rich instead of purine-rich transcripts are generated, recombina-tion was reduced to levels comparable to that of an allele completely lackingSg1 (Shinkura et al., 2003).

In vitro, transcription of substrates with pyrimidine-rich sequences on thetemplate strand produces purine-rich transcripts that form stable RNA–DNAheteroduplexes (R loops and collapsed R loops) with the DNA template strandand result in stretches of ssDNA that have been shown to exist in vitro (Fig. 7)(Mizuta et al., 2003; Reaban and Griffin, 1990; Reaban et al., 1994; Tian andAlt, 2000). Furthermore, R loop formation is orientation dependent, as sub-stantial levels of R loops do not form when these same sequences are tran-scribed in the opposite transcriptional orientation (Shinkura et al., 2003; Tianand Alt, 2000). Thus R loop structures form under circumstances that also leadto CSR in vivo and promote stretches of ssDNA that could provide thenecessary substrate for AID deamination (Fig. 7). Notably, such stable R loopstructures were demonstrated to occur within endogenous S regions whenthey were transcribed in vivo, indicating that they may well serve the physio-logical function of providing an AID substrate that was generated from in vitrostudies (Yu et al., 2003).

Figure 7 Transcription of DNA results in the formation of secondary structures that provide thetarget substrate for AID. The indicated secondary structural changes induced into S regions (Sx) bytranscription have all been proposed to play a role in CSR and/or SHM. R loops, which have beendemonstrated to form in vivo, occur when RNA transcripts stably interact with the DNA templatestrand. AID deaminates cytidine residues preferentially on the coding strand, thus leading to DNAlesions that effect CSR and SHM. Adapted from Chaudhuri and Alt (2004).

v(d)j versus class switch recombination 81

Inversion of endogenous Sg1 sequences did not completely abolish recom-bination in vivo (Shinkura et al., 2003). This would imply that either RNA–DNA hybrid structures are still generated at a reduced level when Sg1 istranscribed in the nonphysiological orientation, or there are other means ofproviding the necessary substrates for AID and CSR. In addition, as XenopusS regions are A-T rich instead of G-C rich, the transcription of XenopusS region sequences would not be predicted to form R loops (Mubmann et al.,1997). Transcription of palindrome-containing sequences found in S regions isalso prone to the formation of stem–loop structures (Kataoka et al., 1981;Mubmann et al., 1997; Tashiro et al., 2001). Unlike R loops, stem–loopstructures should form on transcription of palindromic sequences regardlessof transcriptional orientation. However, like R loops, stem loops can promotethe formation of short stretches of ssDNA that could provide appropriatesubstrates for AID. Therefore, there may be several ways in which the unusual

82 darryll d. dudley ET AL.

sequence content of S regions could lead to the generation of structurescontaining ssDNA that would be a substrate for AID activity. Finally, allS regions (from Xenopus to mammals) have a high concentration of theconsensus SHM motif (most notably AGCT), and evidence suggests that thissequence could target AID in CSR in the context of an AID�RPA complex as itdoes in SHM (Zarrin et al., 2004). It has been proposed that S regions mayhave evolved from sequences containing large numbers of SHM motifs (e.g.,AGCT) in lower organisms to the form found in mammals in which high levelsof such motifs remain (to target AID in the context of RPA), but in mammalianS regions, ssDNA generation may have been further augmented via evolutionof ability to form R loops (Zarrin et al., 2004).

4.5.2. AID-Induced Cytidine Deamination

Cytidine deaminases catalyze the conversion of dC to dU residues via thehydrolytic removal of the amino group at the fourth position of the pyrimidinering of cytidine (Betts et al., 1994). The deamination mechanism is likely toresemble that of adenosine deaminases as both involve a zinc atom in theactive site (Betts et al., 1994; Harris et al., 2002). Deamination of dC residuesby AID thereby induces dU/dG mismatches in DNA, the type of DNA lesionsnormally corrected by the base excision repair and MMR pathways (reviewedin Lindahl, 2000).

4.5.3. Base Excision Repair and Uracil DNA Glycosylase

The base excision repair pathway has evolved to provide protection againststructural alterations that can occur in DNA as a result of various endogenousalkylating agents and metabolic reactive oxygen species (reviewed in Lindahl,2000). Such alterations lead to the formation of aberrant nucleotide residuesthat are frequently recognized by DNA glycosylases (reviewed in Krokan et al.,1997). These glycosylases recognize the aberrant nucleotide and remove itfrom the DNA backbone, leaving behind an abasic site (reviewed in Krokanet al., 1997). In the case of cytidine deamination, this function is performed byUNG. The abasic site is then cleaved by an apyrimidic (AP) endonuclease,followed by nucleotide replacement via the action of a polymerase, which isnormally polymerase b (polb) (reviewed in Lindahl, 2000). The final repairstep is ligation, likely involving DNA ligase III (reviewed in Lindahl, 2000).

4.5.4. DNA Deamination Model

The DNA deamination mechanism for CSR and SHM was initially proposedon the basis of the observation that AID expression in bacteria caused muta-tions that somewhat resemble those induced by SHM (Petersen-Mahrt et al.,

Figure 8 The DNA deamination model for CSR and SHM. AID deaminates cytidine residues inS regions, and in the variable region exons of IgH and IgL genes. Steps that lead to SHM areshown on the left and those that effect CSR on the right. UNG, uracil deglycosylase; AID,activation-induced cytidine deaminase. The minor pathway for CSR refers to CSR in the absenceof UNG, likely involving mismatch repair (MMR). Adapted from Rada et al. (2002) and Di Noiaand Neuberger (2002).

v(d)j versus class switch recombination 83

2002). In this model, replication or repair of dU-containing DNA strands leadsto both CSR between S regions and SHM of immunoglobulin VH and VL exons(Fig. 8; Petersen-Mahrt et al., 2002). Mutations could be generated if the dUresidue resulting from AID-induced cytidine deamination is not removed bybase excision repair before DNA replication, wherein the dU is read as a dT,thus resulting in dC-to-dT and dG-to-dA transitions (Fig. 8; Petersen-Mahrtet al., 2002). Alternatively, if the abasic site produced by the function of UNGundergoes replication via a translesional polymerase (reviewed by Chaudhuriand Alt, 2004; Reynaud et al., 2003), subsequent repair of the abasic site wouldlead to both transitions and transversions (Fig. 8; Petersen-Mahrt et al., 2002).Moreover, if an error-prone polymerase were recruited to the abasic siteduring replication, then mutations could be introduced both in and aroundthe initial deaminase-induced lesion, explaining mutations that arise at non-dC

84 darryll d. dudley ET AL.

or non-dG residues in SHM (reviewed in Reynaud et al., 2003). The dU:dGmismatch could also be recognized and processed by components of the MMRmachinery, ultimately leading to the generation of DNA breaks.

Regarding CSR, if multiple cytidine residues are deaminated in close prox-imity and on opposite strands, then base excision repair could lead directlyto staggered DNA breaks (Petersen-Mahrt et al., 2002). Processing of theends of staggered DNA breaks could then result in the formation of DNADSBs that are either blunt or have short overhangs (reviewed in Reynaud et al.,2003). The subsequent repair of DNA DSBs induced in two different S regionsby the NHEJ pathway could thus result in CSR. In addition, there are modelswhereby MMR could also lead to DSBs after AID deamination (reviewed inMartin and Scharff, 2002a).

In support of the DNA deamination model with respect to SHM, thepattern of hypermutations observed in the hypermutating chicken cellline DT40 changes from transversion-dominated mutations to transitionswhen UNG function is inhibited (Di Noia and Neuberger, 2002).Furthermore, there is a dramatic shift in the SHM pattern to transitionsin immunoglobulin V genes of murine UNG-deficient B cells (Rada et al.,2002b). As transversions are primarily dependent on the removal of dU resi-dues by UNG, these data provide further support for the DNA deaminationmodel. With respect to class switching, CSR is substantially reduced in UNG-deficient mice and humans, in accordance with the predictions of the model(Rada et al., 2002b). In addition, as mentioned above, AID has been shown tobe capable of direct deamination of DNA (Bransteitter et al., 2003; Chaudhuriet al., 2003; Dickerson et al., 2003; Petersen-Mahrt et al., 2002; Pham et al.,2003; Ramiro et al., 2003; Sohail et al., 2003; Yu et al., 2004). Finally, theobservation that AID associates with transcribed S regions (Chaudhuri et al.,2004; Nambu et al., 2003) provides strong support for the DNA deaminationmodel.

As mentioned above, work has questioned the precise role of UNG in CSRand SHM, as certain UNG mutants that are catalytically inactive in U removalactivity in vitro are still proficient in mediating CSR and SHM in vivo,suggesting that the role of UNG in CSR and SHM is beyond its DNAglycosylase activity (Begum et al., 2004a). These results were surprisinggiven that human patients with similar mutations have profound defects inCSR (Imai et al., 2003a), leading to the speculation that there may be second-ary mutations in these patients that contribute to the phenotype (Begumet al., 2004a) or that there is some undetected UNG activity in vivo in themouse mutants; overall, these apparently conflicting findings await furtherexperimentation for full resolution.

v(d)j versus class switch recombination 85

4.5.5. Class Switch Recombination versus Somatic HypermutationSpecific Factors

Hyper-IgM syndrome (HIGM) is caused by defects in CSR leading to reducedlevels of IgG, IgA, and IgE (reviewed in Durandy, 2001). Until recently, the knowncauses of HIGM have been thought to affect both CSR and SHM (reviewedin Durandy, 2001). Some patients with HIGM have now been identified with anovel form of HIGM that results in impaired CSR but normal levels of SHM (Imaiet al., 2003b). B cells isolated from the peripheral blood of HIGM4 patientswere activated for CSR in vitro and shown to express substantial levels of AIDand GTs, yet failed to secrete detectable levels of IgE or IgG (Imai et al., 2003b).Furthermore, using a ligation-mediated polymerase chain reaction (LM-PCR)assay, DNA DSBs corresponding to Sm region sequences were readily detec-table in DNA isolated from activated HIGM4 and control B cells (Imai et al.,2003a). Thus HIGM4 may arise from defects in processes downstream of DNAdeamination that are distinct between CSR from SHM.

CSR and SHM share the requirements for AID-induced DNA deamination;however, AID mutants have now been identified that can differentially effectCSR or SHM. Mutations in the C terminus of AID retain SHM activity butare unable to promote CSR in AID�/� B cells (Barreto et al., 2003; Ta et al.,2003). AID C-terminal mutants retained DNA deamination function (Barretoet al., 2003; Ta et al., 2003), and loss of CSR was not due to failure in nucleartransport (Barreto et al., 2003). Furthermore, several mutations in theN terminus of AID had nearly normal CSR activity but were unable to mediateSHM of a retroviral GFP expression construct (Shinkura et al., 2004).Although some of these C-terminal mutants had defects in nuclear import,their ability to effect CSR suggests inefficient nuclear transport is not the causeof defective SHM (Shinkura et al., 2004). AID mutations that uncouple therelated but distinct processes of CSR and SHM suggest specific cofactorsmight exist that interact with these different domains of AID, with RPAbeing one such potential factor. In this regard, the C terminus of AID hasbeen shown to contain a nuclear export sequence that facilitates nuclear exportin a CRM1-dependent pathway (McBride et al., 2004). Although AID must bepresent in the nucleus to effect CSR and SHM, AID is predominantly found inthe cytoplasm (Rada et al., 2002a). Thus AID cofactors such as RPA may play arole in the retention of AID in the nucleus as well as target specificity.

4.5.6. AID-Induced Double-Strand Breaks

Past studies have documented DNA DSBs in S regions of cells stimulated forCSR (Chen et al., 2001; Wuerffel et al., 1997) and in variable regions of B cellsstimulated for SHM (Bross et al., 2000; Papavasiliou and Schatz, 2000). These

86 darryll d. dudley ET AL.

findings led to the proposal that DSBs may be intermediates in these process-es. On identification of AID, early efforts focused on determining whetherAID functioned in the induction or resolution of DNA DSBs (reviewed inChua et al., 2002). The IgH locus has been shown to colocalize with NBS-1and gH2AX foci, which are normally associated with DSBs, following CSRactivation of wild-type but not AID-deficient B cells (Begum et al., 2004b;Petersen et al., 2001). Colocalization was interpreted to reflect the induction ofDNA DSBs in S regions; and, therefore, these studies concluded that AIDdirectly participates in the formation of DNA DSBs during CSR (Petersen et al.,2001). However, H2AX was also found to be associated with V genes (Woo et al.,2003), and, therefore, it is not clear whether the H2AX foci indeed representbreaks at S regions or those at V genes, particularly given that H2AX foci canextend for up to 1 Mb from a DSB (Rogakou et al., 1999). In support ofDSB intermediates in CSR, LM-PCR assays have detected both AID- andUNG-dependent S region breaks (Catalan et al., 2003; Imai et al., 2003a).

In several studies, DNA breaks in V gene segments were found at similarfrequencies in both wild-type and AID-deficient B cells stimulated to undergoSHM (Bross et al., 2002; Papavasiliou and Schatz, 2002a). These results led tothe suggestion that AID was not involved in the induction of DNA DSBsduring SHM, and thus it was speculated that AID might instead be somehowinvolved in the repair of these DSBs (Papavasiliou and Schatz, 2002a).However, it remains unclear whether the DSBs observed in these studieswere actually related to SHM; so the significance of the findings remainsunclear (reviewed in Chua et al., 2002). Overall, it seems likely that CSRworks via a DSB intermediate, whereas SHM does not; this interpretationhas been reinforced by the requirement for factors involved in the DSBresponse (H2AX, DNA-PKcs, Ku, 53BP1, etc.) in CSR but not SHM (seebelow).

4.5.7. Inter nal S Region Deletions Are Anal ogou s to ClassSwitch Recomb ination

A high frequency of internal S region (intra-S) deletions is detected in the S mregion of normal B cells and B cell lines activated for CSR (Alt et al., 1982;Bottaro et al., 1998; Hummel et al., 1987; Winter et al., 1987). Intra-Smdeletions are largely AID dependent, can occur in the absence of an acceptorS region, and are accompanied by mutations in 30 flanking sequences analo-gous to those seen in CSR junctions (Dudley et al., 2002). Thus an intra-Sregion deletion probably reflects the normal CSR mechanism, but in whichrecombination has taken place within homologous sequences of a singleS region rather than between two heterologous S regions. This could result

v(d)j versus class switch recombination 87

via the failure to induce DNA lesions in downstream S regions or in theabsence of S region synapsis.

4.5.8. S Region Mutations

AID mutants with N-terminal alterations that result in selective loss of SHMactivity, but retain the ability to mediate CSR, also generate mutations in Sm(Shinkura et al., 2004). Mutations in AID that selectively lose SHM activitycould be the result of failure to target the AID mutant to V region sequences,for instance because of inability to interact with RPA (Chaudhuri et al., 2004).In this regard, induction of Sm mutations by SHM-defective AID mutantswould simply reflect the differential targeting of AID and not SHM-versusCSR-specific functions of AID. However, loss of AID-targeting activity doesnot preclude an SHM-specific function of AID not directly related to aseparate CSR-specific activity.

Separate studies involving human or murine C-terminal mutants of AIDthat promote SHM and gene conversion but not CSR have given conflictingresults regarding the induction of Sm mutations (Barreto et al., 2003; Shinkuraet al., 2004). CSR-defective mutants of murine AID induced normal levels ofSm mutations, whereas most human CSR-defective AID mutants were unableto promote Sm lesions. Sm mutations induced by CSR-defective murine AIDmutants (Barreto et al., 2003) could be due to the retention of low-level CSRactivity as suggested by an analogous human C-terminal AID mutant, and thusbe independent of SHM activity (Shinkura et al., 2004). This would imply thatAID contains distinct functions for promoting SHM and CSR; or that theCSR-defective mutant forms of AID fail to target S regions. Alternatively,the CSR-defective AID mutants may properly target S regions and effectDNA lesions, as evidenced by the ability to induce mutations in Sm sequencesbut fail to complete actual CSR. In this scenario, CSR-defective mutants mightbe unable to interact with cofactors essential for DNA repair or to facilitateS region synapsis (reviewed by Chaudhuri and Alt, 2004).

4.6. Class Switch Recombination and S Region Synapsis

Recombination between two different S regions takes place over large chro-mosomal distances (up to �175 kb), and these regions must be juxtaposedbefore being joined. Adjoining of S regions could be mediated via associationwith transcriptional promoters, enhancers, chromatin factors, DNA repairproteins, or AID-associated factors, or by interactions involving the S regionsequences themselves.

88 darryll d. dudley ET AL.

4.6.1. Promoter/Enhancer Interactions

CSR is reduced in the absence of iEm, even though the level of steady-statetranscription through Sm appeared unaffected (Bottaro et al., 1998). Promoter/enhancer interactions between iEm and downstream I promoters could effectjuxtaposition of Sm with the S regions of downstream CH genes. However, asSm transcription detected in the absence of iEm could be driven by VH or DH

promoters, normal levels of CSR may be dependent on the specific site oftranscriptional initiation rather than overall levels of transcription (Bottaroet al., 1998). In this regard, replacement of iEm with a pgk promoter returnsthe rate of CSR to approximately normal levels (Bottaro et al., 1998). Similarly,although deletions of HS3b and HS4 in the 30 RR located downstream of theIgH locus result in the reduction of GTs, a role for these sites in the synapsis ofS region sequences cannot be excluded (Pinaud et al., 2001).

4.6.2. H2AX

Effective long-range synapsis of S regions likely relies on chromatin modificationsand associated factors, as indicated by studies of H2AX deficiency (Reina-San-Martin et al., 2003). As noted above, AID-dependent H2AX foci are found at theIgH locus in conjunction with IgH CSR (Petersen et al., 2001). SHM is unaffectedin H2AX-deficient mice, whereas CSR is substantially impaired (Reina-San-Martin et al., 2003). Intra-S region deletions were detected in H2AX-deficientB cells activated for CSR, demonstrating that accessibility of S regions to theCSR machinery and the basic joining mechanism required for CSR is notimpaired by the absence of H2AX (Reina-San-Martin et al., 2003). The re-cruitment and assembly of repair factors at sites of DNA DSBs by g-H2AX hasbeen proposed to facilitate the juxtaposition of broken DNA ends andsubsequent repair by NHEJ proteins (Bassing and Alt, 2004). Thus gH2AXmight similarly promote long-range S region synapsis for the efficient recom-bination between heterologous S regions. In this regard, H2AX-deficient mice,in the absence of the cell cycle checkpoint protein p53, have been shown toundergo translocations involving S region sequences, perhaps indicating thatproper synapsis of S regions during CSR is important for genome stability aswell as CSR (Bassing et al., 2003a). The finding that another protein, 53BP1,proposed to work in the H2AX anchoring mechanism, is also required for CSR(but not SHM) further supports this general model (Manis et al., 2004).

4.6.3. DNA-PKcs

Pro-B cells that lack DNA-PKcs are defective for switching to the IgE isotype(Rolink et al., 1996). However, significant levels of CSR to all immunoglobulinisotypes were detected in a study involving SCID mice reconstituted with

v(d)j versus class switch recombination 89

rearranged IgH and IgL transgenes, which carry a catalytic mutation in DNA-PKcs that abrogates the kinase activity of DNA-PKcs (Bosma et al., 2002; Cooket al., 2003). In contrast, DNA-PKcs–deficient mice had a significant reductionin CSR to all isotypes except IgG1 (Manis et al., 2002). Expression of DNA-PKcs, albeit catalytically inactive, can be detected in cells from SCID mice,leading to the intriguing possibility that serine/threonine kinase activity ofDNA-PKcs is dispensable for CSR, whereas the presence of a noncatalyticDNA-PKcs can provide a necessary function for CSR (Bosma et al., 2002). Inthe context of this model, why there should still be CSR to IgG1 in DNA-PKcs–deficient mice remains a mystery. The fact that DNA-PKcs–deficientB cells switch to IgG1 and not other isotypes implies that recombina-tion between Sm and Sg1 may be mechanistically different than that ofCSR between Sm and other S regions. Alternatively, a general reduction inCSR efficiency in the absence of DNA-PKcs could result in the preferentialdetection of IgG1 simply because it occurs the most efficiently because of itslarge size. Whatever the case, it is notable that DNA-PKcs is able to promotesynapsis of broken DNA ends in vitro (DeFazio et al., 2002), consistent withsuch a function in CSR. In this regard, transformation/transcription domain-associated protein (TRRAP), a distantly related member of the PI-3 kinasefamily found in humans with homologs in both yeast and Caenorhabditiselegans, apparently lacks kinase activity and appears to instead function as ascaffolding protein during chromatin remodeling (McMahon et al., 1998).

4.6.4. Mismatch Repair

Mlh1- and Pms2-deficient mice have a modest reduction in CSR activity, andsequences isolated from S junctions of Mlh1- and Pms2-deficient B cells havean increased rate of microhomologies compared with wild-type B cells(Ehrenstein et al., 2001; Schrader et al., 2002). Yeast homologs of PMS2 andMLH1 can bind two different DNA molecules simultaneously (Hall et al.,2001), leading to the proposal that PMS2 and MLH1 might facilitate S regionsynapsis during CSR (Schrader et al., 2003).

4.6.5. Other Factors

LR-1 is a B-cell–specific heterodimeric protein composed of nucleolin andheterogeneous nuclear ribonucleoprotein D (hnRNP D), in which each sub-unit is capable of low-affinity binding to S region–specific duplex sequences,and with high affinity to sequences in the form of G quartets or G4 DNA(Dempsey et al., 1999; Williams and Maizels, 1991). Consequently, it has beenproposed that LR-1 might bind and capture DNA from two different S regionsand facilitate their synapsis, thus contributing to CSR (Dempsey et al., 1999).

90 darryll d. dudley ET AL.

4.7. Class Switch Recombination and Double-StrandBreak Repair

The detection of closed circles of DNA composed of intervening sequencesbetween two different S regions implied that intermediates of CSR occur inthe form of DNA DSBs (reviewed in Iwasato et al., 1990; Kenter, 2003).Sequences from CSR junctions demonstrate little or no sequence homologybetween donor and acceptor S regions, and CSR junctions frequently containshort deletions or insertions of nucleotides, all of which are consistent with theNHEJ pathway of DNA DSB repair (Dunnick et al., 1993). Lending furthersupport for DSBs as intermediates in CSR, deficiencies in assayed NHEJproteins reduce CSR in mice (Casellas et al., 1998; Manis et al., 1998a,2002a). Finally, deficiency for 53BP1, a DNA damage-sensing protein thatbecomes activated in response to DSBs and is found associated with H2AX,also leads to significantly reduced levels of CSR (Manis et al., 2004; Ward et al.,2004).

4.7.1. Ku

Ku-deficient mice do not develop B or T cells; therefore rearranged IgH andIgL genes must be introduced into these animals to derive mature B cells(Casellas et al., 1998; Manis et al., 1998a). The only detectable IgH isotype inthe serum of these mice is IgM, and splenic B cells isolated from these animalsand stimulated in vitro to undergo specific CSR fail to secrete anything otherthan IgM (Casellas et al., 1998; Manis et al., 1998a). The presence of GTs fromdownstream CH genes and DSBs detected in S g3 sequences suggested thatthe defect in CSR was not due to an inability to initiate the process (Casellaset al., 1998; Manis et al., 1998a). However, as Ku-deficient B cells are alsodefective in proliferation, the lack of CSR could be explained by decreasedsurvival of activated B cells (Manis et al., 1998a). Potentially counteringthis argument, cells that have undergone several rounds of cell division stilldo not undergo CSR (Reina-San-Martin et al., 2003), although it is not clearwhether these cells might represent those that have failed to be completelyactivated.

4.7.2. D NA-PKc s and Artemis

DNA-PKcs–deficient mice have significantly reduced levels of serum isotypes(Manis et al., 2002a), whereas SCID mice that carry DNA-PKcs kinase inactivemutations undergo CSR at nearly normal levels (Bosma et al., 2002; Cook et al.,2003). In this regard, CSR occurs normally in the absence of Artemis (Rooneyet al., submitted), which is activated on phosphorylation by DNA-PKcs (Maet al., 2002), whereas Artemis is essential for opening the hairpin-coding ends

v(d)j versus class switch recombination 91

generated during V(D)J recombination (Rooney et al., 2002). Therefore,DNA-PKcs may provide functions for the repair of DNA damage inducedduring CSR, such as stabilization of the repair complex, synapsis of targetsequence (see above), or recruitment of other essential proteins to the site ofDNA breaks, outside its role in Artemis activation that is required for V(D)J CJformation.

4.7.3. Ataxia Telangiectasia Mutated

Human patients with ataxia telangiectasia mutated (ATM) have normal levelsof SHM in their V region sequences, although an overall reduction in serumimmunoglobulin isotypes and an increase in homology at S region junctionssuggest that ATM does influence CSR (Pan et al., 2002; Pan-Hammarstromet al., 2003; Waldmann et al., 1983; reviewed in Regueiro et al., 2000). ATM isactivated by DNA damage, thereby phosphorylating and activating cell cyclecontrol proteins p53 and Chk2, and thus inducing cell cycle arrest in cellscontaining DSBs (reviewed in Khanna and Jackson, 2001; Shiloh, 2001).However, ATM likely functions beyond sensing DNA damage and cell cycleregulation, as indicated by its ability to phosphorylate the DNA repair proteinNBS1 (Gatei et al., 2000; Lim et al., 2000; Wu et al., 2000; Zhao et al., 2000). Inaddition to the increase in homology at CSR junctions, there are fewer muta-tions and insertions in the sequences around CSR junctions of ATM-deficientB cells than are found in control B cells (Pan et al., 2002). Thus it appears thatATM may function during the repair phase of CSR, although secondary effectscaused by defects in B- and T-cell development and survival could alsocontribute to the observed immunodeficiencies in patients with ATM. In thisregard, ATM-deficient mice initially were not found to have clear-cut defectsin the production of serum IgH isotypes (Barlow et al., 1996; Xu et al., 1996).However, more detailed analyses have now clearly shown a defect in CSR butnormal internal Sm deletions similar to what is seen in H2AX deficiency, whichsupports a role for the DNA DSB response in this process and potentiallysynapsis (Reina-San-Martin et al., 2004; see below).

4.7.4. 53BP1

The role of NHEJ proteins and the likely generation of DNA DSBs duringCSR imply the need to sense and respond to such DNA lesions. 53BP1 wasfound to interact with the DNA damage response and cell cycle checkpointprotein p53 (Xia et al., 2001). 53BP1 was rapidly phoshphorylated in responseto IR (Anderson et al., 2001) and was found in foci that are thought torepresent sites of DNA damage (Anderson et al., 2001; Schultz et al., 2000).Furthermore, 53BP1 colocalized with g-H2AX in nuclear foci that appear after

92 darryll d. dudley ET AL.

DSB induction (Rappold et al., 2001; Rogakou et al., 1999). 53BP1-deficientB cells were dramatically impaired for CSR; although the production of germ-line transcripts and induction of AID expression were normal (Manis et al.,2004; Ward et al., 2004). In contrast, both V(D)J recombination and SHMoccurred normally in 53BP1-deficient mice (Manis et al., 2004; Ward et al.,2004). Thus CSR is highly dependent on DNA damage-sensing proteinsdownstream of AID induction and, thus, likely to influence the DNA repair/S region joining phase. Finally, rare Sm-Sg1 switch junctions amplified from53BP1-deficient B cells are qualitatively similar to wild-type junctions, demon-strating that 53BP1 does not mechanistically affect CSR (Manis et al., 2004). Inthe context of these observations, it has been suggested that 53BP1 may workwith H2AX for S region synapsis via an anchoring mechanism (Bassing and Alt,2004; Manis et al., 2004).

4.7.5. H2AX

Phosphorylation of H2AX on Ser-139 occurs within minutes after treatmentsthat introduce DNA DSBs in yeast and mammalian cells (Downs et al., 2000;Rogakou et al., 1998). g-H2AX appears in discrete nuclear foci that correlate infrequency and nuclear location with induced DSBs (Rogakou et al., 1999). Therapid appearance of g-H2AX foci after the induction of DSBs precedes that ofDNA repair proteins, suggesting that g-H2AX may be involved in the recruit-ment of specific repair factors such as BRCA1, MRE11, RAD50, and NBS1to sites of DNA damage (Paull et al., 2000). Whereas H2AX is required forefficient CSR and AID-dependent foci formation at the IgH locus (see above),it is not required for the process of intra-S region deletions and has beensuggested to be therefore involved in long-range synapsis (Reina-San-Martinet al., 2003), which might occur via an anchoring mechanism as outlined above(Bassing and Alt, 2004; Manis et al., 2004).

4.7.6. NBS1

NBS1 is a DNA repair protein associated with the hMre11/hRad50/NBS1complex that forms nuclear foci in response to DSB-inducing DNA damageand is a target of ATM-mediated phosphorylation (Carney et al., 1998; Maseret al., 1997; Nelms et al., 1998; Wu et al., 2000; Zhao et al., 2000). In yeast,scmre11 and scrad50 mutants have defects in NHEJ and have been linkedgenetically to the same NHEJ pathway as yku70 and lig4, and Mre11, alsoimplicated in microhomology-mediated DNA break repair (reviewed inCritchlow and Jackson, 1998; Paull and Gellert, 2000). Furthermore, NBS1has been detected at nuclear foci that colocalize with the IgH loci in B cellsactivated to undergo CSR, and this colocalization was dependent on the

v(d)j versus class switch recombination 93

presence of AID (Petersen et al., 2001). However, the levels of most serumisotypes in patients with Nijmegan breakage syndrome (NBS) are not substan-tially reduced (reviewed in Shiloh, 1997). B cells of patients with NBS do havean increased frequency of S region junctions with ‘‘imperfect’’ microhomology(four or more nucleotides, with only one mismatch) compared with controls,albeit from a limited number of samples (Pan et al., 2002). Thus NBS1 mayplay a role in the DNA repair phase of CSR or, given its association withg-H2AX, it could be involved in S region synapsis.

4.7.7. Mismatch Repair

During DNA replication, MMR proteins recognize improperly paired nucleo-tide base pairs and mediate the removal and reinsertion of the correct nucleo-tide based on the DNA template strand (Buermeyer et al., 1999). Severalstudies have found an overall decrease in the rate of CSR in the absence ofcertain MMR proteins (Ehrenstein and Neuberger, 1999; Schrader et al.,1999, 2002). CSR junctions were found to occur more frequently in consensusGAGCT and GGGGT sequences, reminiscent of ‘‘hot spot’’ targeting of SHMin the absence of Msh2 (Ehrenstein and Neuberger, 1999; Phung et al., 1998).Moreover, CSR junctions isolated from B cells of Msh2-deficent mice werefound to have slightly decreased lengths of microhomology (Schrader et al.,2002). This would be consistent with the DNA deamination model of CSR, asMMR proteins can extend the region of mutations beyond the original dUresidue induced by the function of AID and UNG (Petersen-Mahrt et al.,2002). These results are consistent with Msh2 playing a more important role inend processing, specifically the removal of 30 nonhomologous overhangs out-side potential regions of microhomology (Schrader et al., 2002). In contrast,there was an increase in microhomology length detected in the CSR junctionsof Mlh1- and Pms2-deficient B cells (Schrader et al., 2002). The increase inmicrohomology at CSR junctions of Pms2- or Mlh1-deficient B cells mightreflect a role for stabilizing CSR intermediates or for S region synapsis, thusrequiring increased sequence homology in their absence for adequate basepair interactions (Schrader et al., 2002, 2003).

5. CSR-Related Diseases

5.1. Hyper-IgM Syndrome Types 1 and 3

Hyper-IgM (HIGM) syndromes are immunodeficiencies caused by geneticdefects that result in abrogation or impairment in CSR (reviewed inDurandy and Honjo, 2001). The first described was X-linked hyper-IgM, or

94 darryll d. dudley ET AL.

HIGM type 1 (HIGM1), caused by mutations in the gene encoding the CD40ligand (CD40L), a membrane glycoprotein expressed on activated T cells(Allen et al., 1993; Aruffo et al., 1993; DiSanto et al., 1993). CD40L interactswith CD40, a member of the tumor necrosis factor receptor family, that isconstitutively expressed on B cells and is variably expressed on T cells, mono-cytes, basophils, dendritic cells, and endothelial cells (reviewed in Grammerand Lipsky, 2000). CD40L binds to CD40 and induces B cell proliferation(Nishioka and Lipsky, 1994; Tohma and Lipsky, 1991), AID induction(Muramatsu et al., 1999) and the production of some immunoglobulin GTs(Fujita et al., 1995; Jumper et al., 1994; Warren and Berton, 1995). Removal ofeither CD40 or CD40L through the use of anti-CD40L antibodies (Foy et al.,1993, 1994) blocks germinal center formation, SHM, and CSR in response toT-dependent antigens. Genetic defects in CD40 lead to an autosomal recessiveform of hyper-IgM, HIGM3, similar to that caused by the absence of CD40L(Ferrari et al., 2001). Thus HIGM1 and HIGM3 are caused by the ablation ofupstream signaling pathways leading to CSR and SHM activation.

5.2. Hyper-IgM Syndrome Type 2

Autosomal recessive hyper-IgM syndrome type 2 is caused by mutationsabrogating the expression or function of AID (Revy et al., 2000). Patientslacking AID have enlarged lymph nodes with correspondingly expanded ger-minal centers (Revy et al., 2000), a characteristic also seen in AID-deficientmice (Muramatsu et al., 2000). These oversized germinal centers likely reflectthe presence of activated B cells that are unable to effect CSR or SHM, andthus accumulate in B-cell follicles of the peripheral lymph tissue.

5.3. Hyper-IgM Syndrome Type 4

Patients with HIGM4 are substantially impaired for CSR, whereas SHM canbe detected in VH regions at levels comparable to that of controls (Imai et al.,2003b). Defects in AID, UNG, or in the expression of GTs were eliminated aspossible causes of HIGM4. Evidence for the existence of a factor differentiallyinvolved in CSR versus SHM is in keeping with the DNA deamination model,in which CSR is effected via DSB intermediates, whereas SHM can beinduced in the absence of DSBs (Petersen-Mahrt et al., 2002). Thus HIGM4is likely caused by defect(s) in factors associated with the targeting of AID toS regions that affect the synapsis of S regions and/or that are involved in anaspect of DNA DSB repair (reviewed in Manis and Alt, 2003).

v(d)j versus class switch recombination 95

5.4. X-Linked Hypohydrotic Ectodermal Dysplasia

Another human disease, X-linked hypohydrotic ectodermal dysplasia (XHM-ED), is characterized by hyper-IgM immunodeficiency caused by missensemutations in the gene encoding NF-kB essential modulator (NEMO)(Doffinger et al., 2001; Jain et al., 2001). NEMO interacts with NF-kB kinasesIKK1 (IkB kinase 1) and IKK2 and is essential for NF-kB activation (Yamaokaet al., 1998). Engagement of CD40 on the surface of B cells with T-cell–expressed CD40L leads to the induction of NF-kB family transcription factors(Berberich et al., 1994; Francis et al., 1995; Lalmanach-Girard et al., 1993).NF-kB family transcription factors mediate the production of Ig1-Cg1 (Linand Stavnezer, 1996; Lin et al., 1998) and Ie-Ce (Iciek et al., 1997) GTs. Not allimmunoglobulin GTs are dependent on NF-kB; thus mutations affectingNF-kB signaling would be predicted to abrogate CSR to some but not allCH genes. However, patients with XHM-ED have undetectable levels of allserum IgGs, and B cells activated in vitro with anti-CD40 fail to effect CSR(Jain et al., 2001). Thus NF-kB signaling in B cells, as with upstream CD40-and CD40L-mediated signaling, is likely involved in overall activation of CSR,perhaps as an activator of AID, although an affect on AID expression in thesepatients has yet to be reported. The developmental aspects of XHM-EDsyndrome can be attributed to defective NF-kB signaling through tumornecrosis factor (TNF) family receptors expressed on embryonic and fetalectoderm-derived tissues (Doffinger et al., 2001). Thus genetic mutationsthat affect CD40, CD40L, or CD40-mediated downstream signalingmolecules all lead to immunodeficiencies with hyper-IgM characteristics.

6. Concluding Remarks

V(D)J recombination and CSR (and the related process of SHM) lead to thedirect alteration of DNA sequences and content in cells of the vertebrateimmune system. V(D)J recombination occurs both in developing B andT lineage cells; whereas CSR and SHM occur only in mature B lineagecells. The potential for deleterious or catastrophic consequences during themanipulation of a cell’s genetic material is obvious; and aberrant V(D)J recom-bination and CSR, and perhaps SHM, have all been implicated in translocationsand other genetic alterations that underlie T lineage [V(D)J recombination] andB lineage [V(D)J recombination, CSR, and SHM] lymphomas. Therefore, allthree of these potentially dangerous genomic alteration processes require tightregulatory control mechanisms. In this context, the proteins that initiate thesegenetic alterations, namely RAGs for V(D)J recombination and AID for CSR

96 darryll d. dudley ET AL.

and SHM, are expressed in tissue- and lineage-specific fashion and are subjectto strict control via posttranslational regulatory processes. Likewise, there iscontrol of the substrate DNA, RS sequences for V(D)J recombination, andS regions for CSR, such that its availability to the initiating enzymes is largelylimited to the appropriate cell types and sequences. Among the major outstand-ing questions is the issue of precisely what aspects of target RS or S region DNAin chromatin make them good substrates for RAG or AID in the appropriate celltypes, lineages, and/or activation stages.

One fundamental difference between CSR and V(D)J recombination is inthe nature of the target sequences recognized by the ‘‘recombinase.’’ RAG-mediated cleavage at the junction of antigen receptor gene-coding segments issite specific and dependent on short, well-defined cis-acting RSs. In contrast,AID deamination of cytidine residues, which appears to initiate CSR (andSHM), is targeted to large S regions that lie upstream of CH genes in the IgHlocus, with recombination occurring throughout the 1- to 12-kb repetitivesequences. Thus, CSR is region specific rather than site specific. Moreover,AID does not appear to recognize specific target sequences with the samedegree of specificity as RAGs, which, in general, recognize specific RSsequences. Instead, AID has been thought to rely on transcription-dependentDNA structures such as R loops that are formed when sequences with certainbase compositions are actively transcribed. Yet, S regions are composed oftandem repeat units with frequent repeats of specific motifs favored by SHM.Thus, in this context, there may still be specific sequences, such as the SHMconsensus, that are preferentially targeted by AID in conjunction with its RPApartner to provide a further degree of specificity in CSR. Although we nowhave some idea about how AID is targeted, there is still much to be learnedabout how AID targeting is so specific for S regions and variable regions andwhy there is not more wide-scale deamination leading to a higher level ofmutation and translocations involving other genes in activated B cells.

In both V(D)J recombination and CSR, the initiating lesion by RAG andAID ultimately appears to lead to a DSB and, subsequently, to employ DSBrepair pathways, most likely NHEJ pathways, for the resolution of the DNAbreaks. Clearly, the classic NHEJ pathway seals both coding and signal jointsin the context of V(D)J recombination. Some evidence suggests this pathway isalso responsible for ligating CSR junctions, although more evidence on thispoint is needed. A significant difference in the joining phase of V(D)J recom-bination and CSR lies in their relative reliance on the DNA DSB response.Thus, V(D)J recombination occurs relatively unimpaired in the absence ofDSB response factors such as H2AX and 53BP1. However, the absence ofthese factors dramatically impairs CSR. One possible explanation is that thefactors are somehow involved in the long-range synapsis of S regions in

v(d)j versus class switch recombination 97

the context of CSR via a general anchoring mechanism proposed to holdDSBs together in chromatin before their joining via NHEJ. In contrast,RAG-generated DSBs appear to be held together in a postcleavage synapticcomplex by the RAGs themselves, which then recruit the NHEJ factors tocomplete the reaction. In both V(D)J recombination and CSR, however, westill know little about the actual process of synapsis and how the involvedproteins contribute to it.

Acknowledgments

We thank Drs. JoAnn Sekiguchi and John Manis for helpful advice and suggestions. We thank Drs.Sean Rooney and Ali Zarrin for communicating unpublished data. C.H.B. is a LymphomaResearch Foundation Fellow. F. W. A. is an investigator with the Howard Hughes MedicalInstitute. This work was supported by an NIH grant to F.W.A. (NIH 5PO1A131541-14).

References

Agata, Y., Katakai, T., Ye, S. K., Sugai, M., Gonda, H., Honjo, T., Ikuta, K., and Shimizu, A. (2001).J. Exp. Med. 193, 873–880.

Agrawal, A., and Schatz, D. G. (1997). Cell 89, 43–53.Agrawal, A., Eastman, Q. M., and Schatz, D. G. (1998). Nature 394, 744–751.Aidinis, V., Dias, D. C., Gomez, C. A., Bhattacharyya, D., Spanopoulou, E., and Santagata, S.

(2000). J. Immunol. 164, 5826–5832.Akamatsu, Y., and Oettinger, M. A. (1998). Mol. Cell. Biol. 18, 4670–4678.Akamatsu, Y., Monroe, R., Dudley, D. D., Elkin, S. K., Gartner, F., Talukder, S. R., Takahama, Y.,

Alt, F. W., Bassing, C. H., and Oettinger, M. A. (2003). Proc. Natl. Acad. Sci. USA 100,1209–1214.

Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., and Watson, J. D. (1983). ‘‘Molecular Biologyof the Cell.’’ Garland Publishing, Inc., New York, NY.

Alessandrini, A., and Desiderio, S. V. (1991). Mol. Cell. Biol. 11, 2096–2107.Allen, R. C., Armitage, R. J., Conley, M. E., Rosenblatt, H., Jenkins, N. A., Copeland, N. G.,

Bedell, M. A., Edelhoff, S., Disteche, C. M., Simoneaux, D. K., Fanslow, W. C., Belmont, J., andSpriggs, M. K. (1993). Science 259, 990–993.

Alt, F., Enea, V., Bothwell, A., and Baltimore, D. (1980). Cell 21, 1–12.Alt, F., Rosenberg, N., Casanova, R., Thomas, E., and Baltimore, D. (1982). Nature 296, 325–331.Alt, F. W., and Baltimore, D. (1982). Proc. Natl. Acad. Sci. USA 79, 4118–4122.Alt, F. W., Yancopoulos, G. D., Blackwell, T. K., Wood, C., Thomas, E., Boss, M., Coffman, R.,

Rosenberg, N., Tonegawa, S., and Baltimore, D. (1984). EMBO J. 3, 1209–1219.Anderson, L., Henderson, C., and Adachi, Y. (2001). Mol. Cell. Biol. 21, 1719–1729.Arakawa, H., Hauschild, J., and Buerstedde, J. M. (2002). Science 295, 1301–1306.Arbuckle, J. L., Fauss, L. A., Simpson, R., Ptaszek, L. M., and Rodgers, K. K. (2001). J. Biol. Chem.

276, 37093–37101.Aruffo, A., Farrington, M., Hollenbaugh, D., Li, X., Milatovich, A., Nonoyama, S., Bajorath, J.,

Grosmaire, L. S., Stenkamp, R., Neubauer, M., Roberts, R. L., Noelle, R. J., Ledbetter, J. A.,Franke, U., and Ochs, H. D. (1993). Cell 72, 291–300.

98 darryll d. dudley ET AL.

Bachl, J., Carlson, C., Gray-Schopfer, V., Dessing, M., and Olsson, C. (2001). J. Immunol. 166,5051–5057.

Bailin, T., Mo, X., and Sadofsky, M. J. (1999). Mol. Cell. Biol. 19, 4664–4671.Bain, G., Maandag, E. C., Izon, D. J., Amsen, D., Kruisbeek, A. M., Weintraub, B. C., Krop, I.,

Schlissel, M. S., Feeney, A. J., van Roon, M., van der Valk, M., te Riele, H. P. J., Berns, A., andMurre, C. (1994). Cell 79, 885–892.

Barlow, C., Hirotsune, S., Paylor, R., Liyanage, M., Eckhaus, M., Collins, F., Shiloh, Y., Crawley,J. N., Ried, T., Tagle, D., and Wynshaw-Boris, A. (1996). Cell 86, 159–171.

Barreto, V., Marques, R., and Demengeot, J. (2001). Eur. J. Immunol. 31, 3763–3772.Barreto, V., Reina-San-Martin, B., Ramiro, A. R., McBride, K. M., and Nussenzweig, M. C. (2003).

Mol. Cell 12, 501–508.Bassing, C. H., and Alt, F. W. (2004). Cell Cycle 3, 149–153.Bassing, C. H., Alt, F. W., Hughes, M. M., D’Auteuil, M., Wehrly, T. D., Woodman, B. B.,

Gartner, F., White, J. M., Davidson, L., and Sleckman, B. P. (2000). Nature 405, 583–586.Bassing, C. H., Chua, K. F., Sekiguchi, J., Suh, H., Whitlow, S. R., Fleming, J. C., Monroe, B. C.,

Ciccone, D. N., Yan, C., Vlasakova, K., Livingston, D. M., Ferguson, D. O., Scully, R., and Alt, F.W. (2002a). Proc. Natl. Acad. Sci. USA 99, 8173–8178.

Bassing, C. H., Swat, W., and Alt, F. W. (2002b). Cell 109, (Suppl. S45–S55).Bassing, C. H., Suh, H., Ferguson, D. O., Chua, K. F., Manis, J., Eckersdorff, M., Gleason, M.,

Bronson, R., Lee, C., and Alt, F. W. (2003a). Cell 114, 359–370.Bassing, C. H., Tillman, R. E., Woodman, B. B., Canty, D., Monroe, R. J., Sleckman, B. P., and Alt,

F. W. (2003b). Proc. Natl. Acad. Sci. USA 100, 2598–2603.Begum, N. A., Kinoshita, K., Kakazu, N., Muramatsu, M., Nagaoka, H., Shinkura, R.,

Biniszkiewicz, D., Boyer, L. A., Jaenisch, R., and Honjo, T. (2004a). Science 305, 1160–1163.Begum, N. A., Kinoshita, K., Muramatsu, M., Nagaoka, H., Shinkura, R., and Honjo, T. (2004b).

Proc. Natl. Acad. Sci. USA 101, 13003–13007.Belmont, A. S., Dietzel, S., Nye, A. C., Strukov, Y. G., and Tumbar, T. (1999). Curr. Opin. Cell Biol.

11, 307–311.Benezra, R., Davis, R. L., Lockshon, D., Turner, D. L., and Weintraub, H. (1990a). Cell 61, 49–59.Benezra, R., Davis, R. L., Lassar, A., Tapscott, S., Thayer, M., Lockshon, D., and Weintraub, H.

(1990b). Ann. N.Y. Acad. Sci. 599, 1–11.Bensmana, M., Huck, S., Lefranc, G., and Lefranc, M. P. (1988). Nucleic Acids Res. 16, 3108.Berberich, I., Shu, G. L., and Clark, E. A. (1994). J. Immunol. 153, 4357–4366.Berger, S. L. (2002). Curr. Opin. Genet. Dev. 12, 142–148.Bertocci, B., De Smet, A., Berek, C., Weill, J. C., and Reynaud, C. A. (2003). Immunity 19,

203–211.Besmer, E., Mansilla-Soto, J., Cassard, S., Sawchuk, D. J., Brown, G., Sadofsky, M., Lewis, S. M.,

Nussenzweig, M. C., and Cortes, P. (1998). Mol. Cell 2, 817–828.Betts, L., Xiang, S., Short, S. A., Wolfenden, R., and Carter, C. W., Jr. (1994). J. Mol. Biol. 235,

635–656.Betz, A. G., Milstein, C., Gonzalez-Fernandez, A., Pannell, R., Larson, T., and Neuberger, M. S.

(1994). Cell 77, 239–248.Binz, S. K., Sheehan, A. M., and Wold, M. S. (2004). DNA Repair (Amst). 3, 1015–1024.Blackwell, K., Malynn, B., Pollock, R., Ferrier, P., Covey, L., Fulop, G., Phillips, R., Yancopoulos,

G., and Alt, F. (1989). EMBO J. 8, 735–742.Blunt, T., Finnie, N. J., Taccioli, G. E., Smith, G. C. M., Demengeot, J., Gottlieb, T. M., Mizuta, R.,

Varghese, A. J., Alt, F. W., Jeggo, P. A., and Jackson, S. P. (1995). Cell 80, 813–823.Bogue, M. A., Jhappan, C., and Roth, D. B. (1998). Proc. Natl. Acad. Sci. USA 95, 15559–15564.Borgulya, P., Kishi, H., Uematsu, Y., and von Boehmer, H. (1992). Cell 69, 529–537.

v(d)j versus class switch recombination 99

Bories, J. C., Demengeot, J., Davidson, L., and Alt, F. W. (1996). Proc. Natl. Acad. Sci. USA 93,7871–7876.

Born, W., Yague, J., Palmer, E., Kappler, J., and Marrack, P. (1985). Proc. Natl. Acad. Sci. USA 82,2925–2929.

Bosma, G. C., Kim, J., Urich, T., Fath, D. M., Cotticelli, M. G., Ruetsch, N. R., Radic, M. Z., andBosma, M. J. (2002). J. Exp. Med. 196, 1483–1495.

Bottaro, A., Lansford, R., Xu, L., Zhang, J., Rothman, P., and Alt, F. W. (1994). EMBO J. 13,665–674.

Bottaro, A., Young, F., Chen, J., Serwe, M., Sablitzky, F., and Alt, F. W. (1998). Int. Immunol. 10,799–806.

Boulton, S. J., and Jackson, S. P. (1996). EMBO J. 15, 5093–5103.Bouvier, G., Watrin, F., Naspetti, M., Verthuy, C., Naquet, P., and Ferrier, P. (1996). Proc. Natl.

Acad. Sci. USA 93, 7877–7881.Bransteitter, R., Pham, P., Scharff, M. D., and Goodman, M. F. (2003). Proc. Natl. Acad. Sci. USA

100, 4102–4107.Bross, L., Fukita, Y., McBlane, F., Demolliere, C., Rajewsky, K., and Jacobs, H. (2000). Immunity

13, 589–597.Bross, L., Muramatsu, M., Kinoshita, K., Honjo, T., and Jacobs, H. (2002). J. Exp. Med. 195,

1187–1192.Buermeyer, A. B., Deschenes, S. M., Baker, S. M., and Liskay, R. M. (1999). Annu. Rev. Genet. 33,

533–564.Burma, S., Chen, B. P., Murphy, M., Kurimasa, A., and Chen, D. J. (2001). J. Biol. Chem. 276,

42462.Carney, J. P., Maser, R. S., Olivares, H., Davis, E. M., Le Beau, M., Yates, J. R., III, Hays, L.,

Morgan, W. F., and Petrini, J. H. (1998). Cell 93, 477–486.Cary, R. B., Peterson, S. R., Wang, J., Bear, D. G., Bradbury, E. M., and Chen, D. J. (1997). Proc.

Natl. Acad. Sci. USA 94, 4267–4272.Casanova, J. L., Romero, P., Widmann, C., Kourilsky, P., and Maryanski, J. L. (1991). J. Exp. Med.

174, 1371–1383.Casellas, R., Nussenzweig, A., Wuerffel, R., Pelanda, R., Reichlin, A., Suh, H., Qin, X. F., Besmer,

E., Kenter, A., Rajewsky, K., and Nussenzweig, M. C. (1998). EMBO J. 17, 2404–2411.Casellas, R., Jankovic, M., Meyer, G., Gazumyan, A., Luo, Y., Roeder, R., and Nussenzweig, M.

(2002). Cell 110, 575–585.Catalan, N., Selz, F., Imai, K., Revy, P., Fischer, A., and Durandy, A. (2003). J. Immunol. 171,

2504–2509.Celeste, A., Petersen, S., Romanienko, P. J., Fernandez-Capetillo, O., Chen, H. T., Sedelnikova,

O. A., Reina-San-Martin, B., Coppola, V., Meffre, E., Difilippantonio, M. J., Redon, C., Pilch, D.R., Olaru, A., Eckhaus, M., Camerini-Otero, R. D., Tessarollo, L., Livak, F., Manova, K., Bonner,W. M., Nussenzweig, M. C., and Nussenzweig, A. (2002). Science 296, 922–927.

Chan, L. (1992). J. Biol. Chem. 267, 25621–25624.Chan, L., Chang, B. H., Nakamuta, M., Li, W. H., and Smith, L. C. (1997). Biochim. Biophys. Acta

1345, 11–26.Chaudhuri, J., and Alt, F. W. (2004). Nat. Rev. Immunol. 4, 541–552.Chaudhuri, J., Tian, M., Khuong, C., Chua, K., Pinaud, E., and Alt, F. W. (2003). Nature 422,

726–730.Chaudhuri, J., Khuong, C., and Alt, F. W. (2004). Nature 430, 992–998.Chen, H. T., Bhandoola, A., Difilippantonio, M. J., Zhu, J., Brown, M. J., Tai, X., Rogakou, E. P.,

Brotz, T. M., Bonner, W. M., Ried, T., and Nussenzweig, A. (2000). Science 290, 1962–1965.

100 darryll d. dudley ET AL.

Chen, J., Young, F., Bottaro, A., Stewart, V., Smith, R. K., and Alt, F. W. (1993). EMBO J. 12,4635–4645.

Chen, X., Kinoshita, K., and Honjo, T. (2001). Proc. Natl. Acad. Sci. USA 98, 13860–13865.Cherry, S. R., Beard, C., Jaenisch, R., and Baltimore, D. (2000). Proc. Natl. Acad. Sci. USA 97,

8467–8472.Christy, B. A., Sanders, L. K., Lau, L. F., Copeland, N. G., Jenkins, N. A., and Nathans, D. (1991).

Proc. Natl. Acad. Sci. USA 88, 1815–1819.Chua, K. F., Alt, F. W., and Manis, J. P. (2002). J. Exp. Med. 195, F37–F41.Chun, J. J., Schatz, D. G., Oettinger, M. A., Jaenisch, R., and Baltimore, D. (1991). Cell 64,

189–200.Cogne, M., Lansford, R., Bottaro, A., Zhang, J., Gorman, J., Young, F., Cheng, H. L., and Alt, F. W.

(1994). Cell 77, 737–747.Cook, A. J., Oganesian, L., Harumal, P., Basten, A., Brink, R., and Jolly, C. J. (2003). J. Immunol.

171, 6556–6564.Couedel, C., Mills, K. D., Barchi, M., Shen, L., Olshen, A., Johnson, R. D., Nussenzweig, A., Essers,

J., Kanaar, R., Li, G. C., Alt, F. W., and Jasin, M. (2004). Genes Dev. 18, 1293–1304.Critchlow, S. E., and Jackson, S. P. (1998). Trends Biochem. Sci. 23, 394–398.Critchlow, S. E., Bowater, R. P., and Jackson, S. P. (1997). Curr. Biol. 7, 588–598.Dai, Y., Kysela, B., Hanakahi, L. A., Manolis, K., Riballo, E., Stumm, M., Harville, T. O., West,

S. C., Oettinger, M. A., and Jeggo, P. A. (2003). Proc. Natl. Acad. Sci. USA 100, 2462–2467.Davis, M. M., and Bjorkman, P. J. (1988). Nature 334, 395–402.Davis, M. M., Calame, K., Early, P. W., Livant, D. L., Joho, R., Weissman, I. L., and Hood, L.

(1980). Nature 283, 733–739.Davodeau, F., Peyrat, M. A., Houde, I., Hallet, M. M., De Libero, G., Vie, H., and Bonneville, M.

(1993). Science 260, 1800–1802.DeFazio, L. G., Stansel, R. M., Griffith, J. D., and Chu, G. (2002). EMBO J. 21, 3192–3200.Dempsey, L. A., Sun, H., Hanakahi, L. A., and Maizels, N. (1999). J. Biol. Chem. 274, 1066–1071.de Vries, E., van Driel, W., Bergsma, W. G., Arnberg, A. C., and van der Vliet, P. C. (1989). J. Mol.

Biol. 208, 65–78.Dickerson, S. K., Market, E., Besmer, E., and Papavasiliou, F. N. (2003). J. Exp. Med. 197, 1291–1296.Difilippantonio, M. J., McMahan, C. J., Eastman, Q. M., Spanopoulou, E., and Schatz, D. G.

(1996). Cell 87, 253–262.Difilippantonio, M. J., Petersen, S., Chen, H. T., Johnson, R., Jasin, M., Kanaar, R., Ried, T., and

Nussenzweig, A. (2002). J. Exp. Med. 196, 469–480.Di Noia, J., and Neuberger, M. S. (2002). Nature 419, 43–48.DiSanto, J. P., Bonnefoy, J. Y., Gauchat, J. F., Fischer, A., and de Saint Basile, G. (1993). Nature

361, 541–543.Doffinger, R., Smahi, A., Bessia, C., Geissmann, F., Feinberg, J., Durandy, A., Bodemer, C.,

Kenwrick, S., Dupuis-Girod, S., Blanche, S., Wood, P., Rabia, S. H., Headon, D. J., Overbeek,P. A., Le Deist, F., Holland, S. M., Belani, K., Kumararatne, D. S., Fischer, A., Shapiro, R.,Conley, M. E., Reimund, E., Kalhoff, H., Abinun, M., Munnich, A., Israel, A., Courtois, G., andCasanova, J. L. (2001). Nat. Genet. 27, 277–285.

Doi, T., Kinoshita, K., Ikegawa, M., Muramatsu, M., and Honjo, T. (2003). Proc. Natl. Acad. Sci.USA 100, 2634–2638.

Downs, J. A., Lowndes, N. F., and Jackson, S. P. (2000). Nature 408, 1001–1004.Dudley, D. D., Manis, J. M., Zarrin, A. A., Kaylor, L., Tian, M., and Alt, F. W. (2002). Proc. Natl.

Acad. Sci. USA 99, 9984–9989.Dudley, D. D., Sekiguchi, J., Zhu, C., Sadofsky, M. J., Whitlow, S., DeVido, J., Monroe, R. J.,

Bassing, C. H., and Alt, F. W. (2003). J. Exp. Med. 198, 1439–1450.

v(d)j versus class switch recombination 101

Dunnick, W., Hertz, G. Z., Scappino, L., and Gritzmacher, C. (1993). Nucleic Acids Res. 21,365–372.

Durandy, A. (2001). Curr. Opin. Allergy Clin. Immunol. 1, 519–524.Durandy, A., and Honjo, T. (2001). Curr. Opin. Immunol. 13, 543–548.Early, P., Huang, H., Davis, M., Calame, K., and Hood, L. (1980). Cell 19, 981–992.Eastman, Q. M., Leu, T. M., and Schatz, D. G. (1996). Nature 380, 85–88.Ehrenstein, M. R., and Neuberger, M. S. (1999). EMBO J. 18, 3484–3490.Ehrenstein, M. R., Rada, C., Jones, A. M., Milstein, C., and Neuberger, M. S. (2001). Proc. Natl.

Acad. Sci. USA 98, 14553–14558.Elkin, S. K., Matthews, A. G., and Oettinger, M. A. (2003). EMBO J. 22, 1931–1938.Errami, A., He, D. M., Friedl, A. A., Overkamp, W. J., Morolli, B., Hendrickson, E. A., Eckardt-

Schupp, F., Oshimura, M., Lohman, P. H., Jackson, S. P., and Zdzienicka, M. Z. (1998). NucleicAcids Res. 26, 3146–3153.

Eto, T., Kinoshita, K., Yoshikawa, K., Muramatsu, M., and Honjo, T. (2003). Proc. Natl. Acad. Sci.USA 100, 12895–12898.

Eyquem, S., Lagresle, C., Fasseu, M., Sigaux, F., and Bories, J. C. (2002). Eur. J. Immunol. 32,3256–3266.

Fehling, H. J., and von Boehmer, H. (1997). Curr. Opin. Immunol. 9, 263–275.Fehling, H. J., Gilfillan, S., and Ceredig, R. (1999). Adv. Immunol. 71, 1–76.Ferrari, S., Giliani, S., Insalaco, A., Al-Ghonaium, A., Soresina, A. R., Loubser, M., Avanzini, M. A.,

Marconi, M., Badolato, R., Ugazio, A. G., Levy, Y., Catalan, N., Durandy, A., Tbakhi, A.,Notarangelo, L. D., and Plebani, A. (2001). Proc. Natl. Acad. Sci. USA 98, 12614–12619.

Ferrier, P., Krippl, B., Blackwell, T. K., Furley, A. J., Suh, H., Winoto, A., Cook, W. D., Hood, L.,Costantini, F., and Alt, F. W. (1990). EMBO J. 9, 117–125.

Fondell, J. D., and Marcu, K. B. (1992). Mol. Cell. Biol. 12, 1480–1489.Foy, T. M., Shepherd, D. M., Durie, F. H., Aruffo, A., Ledbetter, J. A., and Noelle, R. J. (1993).

J. Exp. Med. 178, 1567–1575.Foy, T. M., Laman, J. D., Ledbetter, J. A., Aruffo, A., Claassen, E., and Noelle, R. J. (1994). J. Exp.

Med. 180, 157–163.Francis, D. A., Karras, J. G., Ke, X. Y., Sen, R., and Rothstein, T. L. (1995). Int. Immunol. 7,

151–161.Frank, K. M., Sekiguchi, J. M., Seidl, K. J., Swat, W., Rathbun, G. A., Cheng, H. L., Davidson, L.,

Kangaloo, L., and Alt, F. W. (1998). Nature 396, 173–177.Fugmann, S. D., and Schatz, D. G. (2001). Mol. Cell 8, 899–910.Fugmann, S. D., Lee, A. I., Shockett, P. E., Villey, I. J., and Schatz, D. G. (2000a). Annu. Rev.

Immunol. 18, 495–527.Fugmann, S. D., Villey, I. J., Ptaszek, L. M., and Schatz, D. G. (2000b). Mol. Cell 5, 97–107.Fugmann, S. D., Rush, J. S., and Schatz, D. G. (2004). Eur. J. Immunol. 34, 844–849.Fujimoto, S., and Yamagishi, H. (1987). Nature 327, 242–243.Fujita, K., Jumper, M. D., Meek, K., and Lipsky, P. E. (1995). Int. Immunol. 7, 1529–1533.Fukita, Y., Jacobs, H., and Rajewsky, K. (1998). Immunity 9, 105–114.Fukumura, R., Araki, R., Fujimori, A., Mori, M., Saito, T., Watanabe, F., Sarashi, M., Itsukaichi, H.,

Eguchi-Kasai, K., Sato, K., Tatsumi, K., and Abe, M. (1998). J. Biol. Chem. 273, 13058–13064.Fukumura, R., Araki, R., Fujimori, A., Tsutsumi, Y., Kurimasa, A., Li, G. C., Chen, D. J., Tatsumi, K.,

and Abe, M. (2000). J. Immunol. 165, 3883–3889.Gao, Y., Chaudhuri, J., Zhu, C., Davidson, L., Weaver, D. T., and Alt, F. W. (1998). Immunity 9,

367–376.Gatei, M., Young, D., Cerosaletti, K. M., Desai-Mehta, A., Spring, K., Kozlov, S., Lavin, M. F.,

Gatti, R. A., Concannon, P., and Khanna, K. (2000). Nat. Genet. 25, 115–119.

102 darryll d. dudley ET AL.

Gay, D., Saunders, T., Camper, S., and Weigert, M. (1993). J. Exp. Med. 177, 999–1008.Gerbi, S. A., and Bielinsky, A. K. (2002). Curr. Opin. Genet. Dev. 12, 243–248.Getts, R. C., and Stamato, T. D. (1994). J. Biol. Chem. 269, 15981–15984.Gilfillan, S., Dierich, A., Lemeur, M., Benoist, C., and Mathis, D. (1993). Science 261, 1175–1178.Glusman, G., Rowen, L., Lee, I., Boysen, C., Roach, J. C., Smit, A. F., Wang, K., Koop, B. F., and

Hood, L. (2001). Immunity 15, 337–349.Golding, A., Chandler, S., Ballestar, E., Wolffe, A. P., and Schlissel, M. S. (1999). EMBO J. 18,

3712–3723.Goldman, J. P., Spencer, D. M., and Raulet, D. H. (1993). J. Exp. Med. 177, 729–739.Gonda, H., Sugai, M., Nambu, Y., Katakai, T., Agata, Y., Mori, K. J., Yokota, Y., and Shimizu, A.

(2003). J. Exp. Med. 198, 1427–1437.Gorman, J. R., and Alt, F. W. (1998). Adv. Immunol. 69, 113–181.Gorman, J. R., van der Stoep, N., Monroe, R., Cogne, M., Davidson, L., and Alt, F. W. (1996).

Immunity 5, 241–252.Gottlieb, T. M., and Jackson, S. P. (1993). Cell 72, 131–142.Grammer, A. C., and Lipsky, P. E. (2000). Adv. Immunol. 76, 61–178.Grawunder, U., Leu, T. M. J., Schatz, D. G., Werner, A., Rolink, A. G., Melchers, F., and Winkler,

T. H. (1995). Immunity 3, 601–608.Grawunder, U., Wilm, M., Wu, X., Kulesza, P., Wilson, T. E., Mann, M., and Lieber, M. R. (1997).

Nature 388, 492–495.Greeve, J., Philipsen, A., Krause, K., Klapper, W., Heidorn, K., Castle, B. E., Janda, J., Marcu,

K. B., and Parwaresch, R. (2003). Blood 101, 3574–3580.Griffiths, G. M., Berek, C., Kaartinen, M., and Milstein, C. (1984). Nature 312, 271–275.Gu, H., Zou, Y. R., and Rajewsky, K. (1993). Cell 73, 1155–1164.Gu, Y., Jin, S., Gao, Y., Weaver, D. T., and Alt, F. W. (1997). Proc. Natl. Acad. Sci. USA 94,

8076–8081.Guidos, C. J., Williams, C. J., Grandal, I., Knowles, G., Huang, M. T., and Danska, J. S. (1996).

Genes Dev. 10, 2038–2054.Hall, M. C., Wang, H., Erie, D. A., and Kunkel, T. A. (2001). J. Mol. Biol. 312, 637–647.Han, S., Zheng, B., Schatz, D. G., Spanopoulou, E., and Kelsoe, G. (1996). Science 274,

2094–2097.Hardianti, M. S., Tatsumi, E., Syampurnawati, M., Furuta, K., Saigo, K., Kawano, S., Kumagai, S.,

Nakamura, F., and Matsuo, Y. (2004a). Leuk. Lymphoma. 45, 155–160.Hardianti, M. S., Tatsumi, E., Syampurnawati, M., Furuta, K., Saigo, K., Nakamachi, Y., Kumagai,

S., Ohno, H., Tanabe, S., Uchida, M., and Yasuda, N. (2004b). Leukemia. 18, 826–831.Haren, L., Ton-Hoang, B., and Chandler, M. (1999). Annu. Rev. Microbiol. 53, 245–281.Harris, R. S., Kong, Q., and Maizels, N. (1999). Mutat. Res. 436, 157–178.Harris, R. S., Petersen-Mahrt, S. K., and Neuberger, M. S. (2002). Mol. Cell 10, 1247–1253.Hein, K., Lorenz, M. G., Siebenkotten, G., Petry, K., Christine, R., and Radbruch, A. (1998). J. Exp.

Med. 188, 2369–2374.Hempel, W. M., Leduc, I., Mathieu, N., Tripathi, R. K., and Ferrier, P. (1998). Adv. Immunol. 69,

309–352.Hesse, J. E., Lieber, M. R., Gellert, M., and Mizuuchi, K. (1987). Cell 49, 775–783.Hesse, J. E., Lieber, M. R., Mizuuchi, K., and Gellert, M. (1989). Genes Dev. 3, 1053–1061.Hesslein, D. G., and Schatz, D. G. (2001). Adv. Immunol. 78, 169–232.Hikida, M., Mori, M., Takai, T., Tomochika, K., Hamatani, K., and Ohmori, H. (1996). Science 274,

2092–2094.Hiom, K., and Gellert, M. (1997). Cell 88, 65–72.Hiom, K., and Gellert, M. (1998). Mol. Cell 1, 1011–1019.

v(d)j versus class switch recombination 103

Hiom, K., Melek, M., and Gellert, M. (1998). Cell 94, 463–470.Hoeijmakers, J. H. (2001). Nature 411, 366–374.Honjo, T., and Kataoka, T. (1978). Proc. Natl. Acad. Sci. USA 75, 2140–2144.Honjo, T., and Matsuda, F. (1995). In ‘‘Immunoglobulin Genes’’ (T. Honjo and T. W. Alt, Eds.),

pp. 145–172. Academic Press, San Diego, CA.Honjo, T., Kinoshita, K., and Muramatsu, M. (2002). Annu. Rev. Immunol. 20, 165–196.Hsu, L. Y., Lauring, J., Liang, H. E., Greenbaum, S., Cado, D., Zhuang, Y., and Schlissel, M. S.

(2003). Immunity 19, 105–117.Hummel, M., Berry, J. K., and Dunnick, W. (1987). J. Immunol. 138, 3539–3548.Iciek, L. A., Delphin, S. A., and Stavnezer, J. (1997). J. Immunol. 158, 4769–4779.Igarashi, H., Gregory, S. C., Yokota, T., Sakaguchi, N., and Kincade, P. W. (2002). Immunity 17,

117–130.Imai, K., Slupphaug, G., Lee, W. I., Revy, P., Nonoyama, S., Catalan, N., Yel, L., Forveille, M.,

Kavli, B., Krokan, H. E., Ochs, H. D., Fischer, A., and Durandy, A. (2003a). Nat. Immunol. 4,1023–1028.

Imai, K., Catalan, N., Plebani, A., Marodi, L., Sanal, O., Kumaki, S., Nagendran, V., Wood, P.,Glastre, C., Sarrot-Reynauld, F., Hermine, O., Forveille, M., Revy, P., Fischer, A., and Durandy,A. (2003b). J. Clin. Invest. 112, 136–142.

Inlay, M., Alt, F. W., Baltimore, D., and Xu, Y. (2002). Nat. Immunol. 3, 463–468.Iwasato, T., Shimizu, A., Honjo, T., and Yamagishi, H. (1990). Cell 62, 143–149.Jackson, S. P. (2002). Carcinogenesis 23, 687–696.Jacobs, H., and Bross, L. (2001). Curr. Opin. Immunol. 13, 208–218.Jain, A., Ma, C. A., Liu, S., Brown, M., Cohen, J., and Strober, W. (2001). Nat. Immunol. 2,

223–228.Jenuwein, T., and Allis, C. D. (2001). Science 293, 1074–1080.Jones, J. M., and Gellert, M. (2001). Proc. Natl. Acad. Sci. USA 98, 12926–12931.Jones, J. M., and Gellert, M. (2002). EMBO J. 21, 4162–4171.Jumper, M. D., Splawski, J. B., Lipsky, P. E., and Meek, K. (1994). J. Immunol. 152, 438–445.Jung, D., and Alt, F. W. (2004). Cell 116, 299–311.Jung, D., Bassing, C. H., Fugmann, S. D., Cheng, H. L., Schatz, D. G., and Alt, F. W. (2003).

Immunity 18, 65–74.Jung, S., Rajewsky, K., and Radbruch, A. (1993). Science 259, 984–987.Kabat, D. (1972). Science 175, 134–140.Kataoka, T., Kawakami, T., Takahashi, N., and Honjo, T. (1980). Proc. Natl. Acad. Sci. USA 77,

919–923.Kataoka, T., Miyata, T., and Honjo, T. (1981). Cell 23, 357–368.Kelley, D., Pollok, B., Atchison, M., and Perry, R. (1988). Mol. Cell. Biol. 8, 930–937.Kenter, A. L. (2003). Curr. Opin. Immunol. 15, 190–198.Khamlichi, A. A., Glaudet, F., Oruc, Z., Denis, V., Le Bert, M., and Cogne, M. (2004). Blood 103,

3828–3836.Khanna, K. K., and Jackson, S. P. (2001). Nat. Genet. 27, 247–254.Kienker, L. J., Kuziel, W. A., and Tucker, P. W. (1991). J. Exp. Med. 174, 769–773.Kim, D. R., Dai, Y., Mundy, C. L., Yang, W., and Oettinger, M. A. (1999). Genes Dev. 13,

3070–3080.Kirch, S. A., Sudarsanam, P., and Oettinger, M. A. (1996). Eur. J. Immunol. 26, 886–891.Kirch, S. A., Rathbun, G. A., and Oettinger, M. A. (1998). EMBO J. 17, 4881–4886.Kisielow, P., and von Boehmer, H. (1995). Adv. Immunol. 58, 87–209.Komori, T., Okada, A., Stewart, V., and Alt, F. W. (1993). Science 261, 1171–1175.Krangel, M. S. (2003). Nat. Immunol. 4, 624–630.

104 darryll d. dudley ET AL.

Krokan, H. E., Standal, R., and Slupphaug, G. (1997). Biochem. J. 325, 1–16.Kuwata, N., Igarashi, H., Ohmura, T., Aizawa, S., and Sakaguchi, N. (1999). J. Immunol. 163,

6355–6359.Kuzin, II, Ugine, G. D., Wu, D., Young, F., Chen, J., and Bottaro, A. (2000). J. Immunol. 164,

1451–1457.Kwon, J., Imbalzano, A. N., Matthews, A., and Oettinger, M. A. (1998). Mol. Cell 2, 829–839.Lafaille, J. J., DeCloux, A., Bonneville, M., Takagaki, Y., and Tonegawa, S. (1989). Cell 59,

859–870.Lalmanach-Girard, A. C., Chiles, T. C., Parker, D. C., and Rothstein, T. L. (1993). J. Exp. Med.

177, 1215–1219.Landree, M. A., Wibbenmeyer, J. A., and Roth, D. B. (1999). Genes Dev. 13, 3059–3069.Lansford, R., Manis, J. P., Sonoda, E., Rajewsky, K., and Alt, F. W. (1998). Int. Immunol. 10,

325–332.Lauring, J., and Schlissel, M. S. (1999). Mol. Cell. Biol. 19, 2601–2612.Lee, C. G., Kondo, S., and Honjo, T. (1998). Curr. Biol. 8, 227–230.Lee, G. S., Neiditch, M. B., Salus, S. S., and Roth, D. B. (2004). Cell 117, 171–184.Lees-Miller, S. P., Godbout, R., Chan, D. W., Weinfeld, M., Day, R. S., III, Barron, G. M., and

Allalunis-Turner, J. (1995). Science 267, 1183–1185.Leu, T. M. J., and Schatz, D. G. (1995). Mol. Cell. Biol. 15, 5657–5670.Leu, T. M., Eastman, Q. M., and Schatz, D. G. (1997). Immunity 7, 303–314.Lewis, S., Gifford, A., and Baltimore, D. (1985). Science 228, 677–685.Lewis, S., Rosenberg, N., Alt, F., and Baltimore, D. (1982). Cell 30, 807–816.Lewis, S. M. (1994a). Adv. Immunol. 56, 27–150.Lewis, S. M. (1994b). Proc. Natl. Acad. Sci. USA 91, 1332–1336.Lewis, S. M., and Wu, G. E. (1997). Cell 88, 159–162.Lewis, S. M., Hesse, J. E., Mizuuchi, K., and Gellert, M. (1988). Cell 55, 1099–1107.Li, Z., Otevrel, T., Gao, Y., Cheng, H. L., Seed, B., Stamato, T. D., Taccioli, G. E., and Alt, F. W.

(1995). Cell 83, 1079–1089.Liang, H. E., Hsu, L. Y., Cado, D., Cowell, L. G., Kelsoe, G., and Schlissel, M. S. (2002). Immunity

17, 639–651.Liang, H. E., Hsu, L. Y., Cado, D., and Schlissel, M. S. (2004). Cell 118, 19–29.Lieber, M. R. (1991). FASEB J. 5, 2934–2944.Lieber, M. R., Hesse, J. E., Lewis, S., Bosma, G. C., Rosenberg, N., Mizuuchi, K., Bosma, M. J.,

and Gellert, M. (1988). Cell 55, 7–16.Lieber, M. R., Grawunder, U., Wu, X., and Yaneva, M. (1997). Curr. Opin. Genet. Dev. 7, 99–104.Lim, D. S., Kim, S. T., Xu, B., Maser, R. S., Lin, J., Petrini, J. H., and Kastan, M. B. (2000). Nature

404, 613–617.Lin, S. C., and Stavnezer, J. (1996). Mol. Cell. Biol. 16, 4591–4603.Lin, S. C., Wortis, H. H., and Stavnezer, J. (1998). Mol. Cell. Biol. 18, 5523–5532.Lin, W.-C., and Desiderio, S. (1993). Science 260, 953–959.Lin, W. C., and Desiderio, S. (1995). Immunol. Today 16, 279–289.Lindahl, T. (2000). Mutat. Res. 462, 129–135.Lorenz, M., Jung, S., and Radbruch, A. (1995). Science 267, 1825–1828.Luby, T. M., Schrader, C. E., Stavnezer, J., and Selsing, E. (2001). J. Exp. Med. 193, 159–168.Lutzker, S., and Alt, F. W. (1988). Mol. Cell. Biol. 8, 1849–1852.Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M. R. (2002). Cell 108, 781–794.Macpherson, A. J., Lamarre, A., McCoy, K., Harriman, G. R., Odermatt, B., Dougan, G.,

Hengartner, H., and Zinkernagel, R. M. (2001). Nat. Immunol. 2, 625–631.Madisen, L., and Groudine, M. (1994). Genes Dev. 8, 2212–2226.

v(d)j versus class switch recombination 105

Malissen, M., McCoy, C., Blanc, D., Trucy, J., Devaux, C., Schmitt-Verhulst, A.-M., Fitch, F.,Hood, L., and Malissen, B. (1986). Nature 319, 28–33.

Malissen, M., Trucy, J., Letourneur, F., Rebai, N., Dunn, D. E., Fitch, F. W., Hood, L., andMalissen, B. (1988). Cell 55, 49–59.

Malissen, M., Trucy, J., Jouvin-Marche, E., Cazenave, P. A., Scollay, R., and Malissen, B. (1992).Immunol. Today 13, 315–322.

Malynn, B., Blackwell, T., Fulop, G., Rathbun, G., Furley, A., Ferrier, P., Heinke, L., Phillips, R.,Yancopoulos, G., and Alt, F. (1988). Cell 54, 453–460.

Manis, J. P., and Alt, F. W. (2003). J. Clin. Invest. 112, 19–22.Manis, J. P., Gu, Y., Lansford, R., Sonoda, E., Ferrini, R., Davidson, L., Rajewsky, K., and Alt, F. W.

(1998a). J. Exp. Med. 187, 2081–2089.Manis, J. P., van der Stoep, N., Tian, M., Ferrini, R., Davidson, L., Bottaro, A., and Alt, F. W.

(1998b). J. Exp. Med. 188, 1421–1431.Manis, J. P., Dudley, D., Kaylor, L., and Alt, F. W. (2002a). Immunity 16, 607–617.Manis, J. P., Tian, M., and Alt, F. W. (2002b). Trends Immunol. 23, 31–39.Manis, J. P., Michaelson, J. S., Birshtein, B. K., and Alt, F. W. (2003). Mol. Immunol. 39, 753–760.Manis, J. P., Morales, J. C., Xia, Z., Kutok, J. L., Alt, F. W., and Carpenter, P. B. (2004). Nat.

Immunol. 5, 481–487.Mannironi, C., Bonner, W. M., and Hatch, C. L. (1989). Nucleic Acids Res. 17, 9113–9126.Martin, A., and Scharff, M. D. (2002a). Nat. Rev. Immunol. 2, 605–614.Martin, A., and Scharff, M. D. (2002b). Proc. Natl. Acad. Sci. USA 99, 12304–12308.Maser, R. S., Monsen, K. J., Nelms, B. E., and Petrini, J. H. (1997). Mol. Cell. Biol. 17, 6087–6096.Mather, E., and Perry, R. (1981). Nucleic Acids Res. 9, 6855–6867.Mather, E. L., and Perry, R. P. (1983). Proc. Natl. Acad. Sci. USA 80, 4689–4693.Mathieu, N., Hempel, W. M., Spicuglia, S., Verthuy, C., and Ferrier, P. (2000). J. Exp. Med. 192,

625–636.Matsunaga, T., Park, C. H., Bessho, T., Mu, D., and Sancar, A. (1996). J. Biol. Chem. 271,

11047–11050.Max, E. E., Seidman, J. G., and Leder, P. (1979). Proc. Natl. Acad. Sci. USA 76, 3450–3454.McBlane, F., and Boyes, J. (2000). Curr. Biol. 10, 483–486.McBlane, J. F., van Gent, D. C., Ramsden, D. A., Romeo, C., Cuomo, C. A., Gellert, M., and

Oettinger, M. A. (1995). Cell 83, 387–395.McBride, K. M., Barreto, V., Ramiro, A. R., Stavropoulos, P., and Nussenzweig, M. C. (2004).

J. Exp. Med. 199, 1235–1244.McMahon, S. B., Van Buskirk, H. A., Dugan, K. A., Copeland, T. D., and Cole, M. D. (1998). Cell

94, 363–374.Melchers, F., Karasuyama, H., Haasner, D., Bauer, S., Kudo, A., Sakaguchi, N., Jameson, B., and

Rolink, A. (1993). Immunol. Today 14, 60–68.Melchers, F., ten Boekel, E., Yamagami, T., Andersson, J., and Rolink, A. (1999). Semin. Immunol.

11, 307–317.Messier, T. L., O’Neill, J. P., Hou, S. M., Nicklas, J. A., and Finette, B. A. (2003). EMBO J. 22,

1381–1388.Mills, K. D., Ferguson, D. O., and Alt, F. W. (2003). Immunol. Rev. 194, 77–95.Mills, K. D., Ferguson, D. O., Essers, J., Eckersdorff, M., Kanaar, R., and Alt, F. W. (2004). Genes

Dev. 18, 1283–1292.Mizuta, R., Mizuta, M., Araki, S., and Kitamura, D. (2002). J. Biol. Chem. 277, 41423–41427.Mizuta, R., Iwai, K., Shigeno, M., Mizuta, M., Uemura, T., Ushiki, T., and Kitamura, D. (2003).

J. Biol. Chem. 278, 4431–4434.

106 darryll d. dudley ET AL.

Mombaerts, P., Iacomini, J., Johnson, R. S., Herrup, K., Tonegawa, S., and Papaioannou, V. E.(1992). Cell 68, 869–877.

Monroe, R. J., Chen, F., Ferrini, R., Davidson, L., and Alt, F. W. (1999a). Proc. Natl. Acad. Sci.USA 96, 12713–12718.

Monroe, R. J., Seidl, K. J., Gaertner, F., Han, S., Chen, F., Sekiguchi, J., Wang, J., Ferrini, R.,Davidson, L., Kelsoe, G., and Alt, F. W. (1999b). Immunity 11, 201–212.

Monroe, R. J., Sleckman, B. P., Monroe, B. C., Khor, B., Claypool, S., Ferrini, R., Davidson, L.,and Alt, F. W. (1999c). Immunity 10, 503–513.

Morshead, K. B., Ciccone, D. N., Taverna, S. D., Allis, C. D., and Oettinger, M. A. (2003). Proc.Natl. Acad. Sci. USA 100, 11577–11582.

Morzycka-Wroblewska, E., Lee, F. E. H., and Desiderio, S. V. (1988). Science 242, 261–263.Mostoslavsky, R., and Alt, F. W. (2004). Trends Immunol. 25, 276–279.Mostoslavsky, R., Singh, N., Kirillov, A., Pelanda, R., Cedar, H., Chess, A., and Bergman, Y. (1998).

Genes Dev. 12, 1801–1811.Mostoslavsky, R., Alt, F. W., and Bassing, C. H. (2003). Nat. Immunol. 4, 603–606.Mostoslavsky, R., Alt, F. W., and Rajewsky, K. (2004). Cell 118, 539–544.Mundy, C. L., Patenge, N., Matthews, A. G., and Oettinger, M. A. (2002). Mol. Cell. Biol. 22, 69–77.Muramatsu, M., Sankaranand, V. S., Anant, S., Sugai, M., Kinoshita, K., Davidson, N. O., and

Honjo, T. (1999). J. Biol. Chem. 274, 18470–18476.Muramatsu, M., Kinoshita, K., Fagarasan, S., Yamada, S., Shinkai, Y., and Honjo, T. (2000). Cell

102, 553–563.Mubmann, R., Courtet, M., Schwager, J., and Du Pasquier, L. (1997). Eur. J. Immunol. 27,

2610–2619.Nadel, B., Tehranchi, S., and Feeney, A. J. (1995). J. Immunol. 154, 6430–6436.Nadel, B., Tang, A., Escuro, G., Lugo, G., and Feeney, A. J. (1998). J. Exp. Med. 187, 1495–1503.Nagaoka, H., Yu, W., and Nussenzweig, M. C. (2000). Curr. Opin. Immunol. 12, 187–190.Nagaoka, H., Muramatsu, M., Yamamura, N., Kinoshita, K., and Honjo, T. (2002). J. Exp. Med.

195, 529–534.Nambu, Y., Sugai, M., Gonda, H., Lee, C. G., Katakai, T., Agata, Y., Yokota, Y., and Shimizu, A.

(2003). Science 302, 2137–2140.Nelms, B. E., Maser, R. S., MacKay, J. F., Lagally, M. G., and Petrini, J. H. (1998). Science 280,

590–592.Nemazee, D., and Weigert, M. (2000). J. Exp. Med. 191, 1813–1817.Neuberger, M. S., Harris, R. S., Di Noia, J., and Petersen-Mahrt, S. K. (2003). Trends Biochem.

Sci. 28, 305–312.Ng, H. H., Ciccone, D. N., Morshead, K. B., Oettinger, M. A., and Struhl, K. (2003). Proc. Natl.

Acad. Sci. USA 100, 1820–1825.Nicolas, N., Moshous, D., Cavazzana-Calvo, M., Papadopoulo, D., de Chasseval, R., le Deist, F.,

Fischer, A., and de Villartay, J. P. (1998). J. Exp. Med. 188, 627–634.Nishioka, Y., and Lipsky, P. E. (1994). J. Immunol. 153, 1027–1036.Noordzij, J. G., Verkaik, N. S., van der Burg, M., van Veelen, L. R., de Bruin-Versteeg, S., Wiegant,

W., Vossen, J. M., Weemaes, C. M., de Groot, R., Zdzienicka, M. Z., van Gent, D. C., and vanDongen, J. J. (2003). Blood 101, 1446–1452.

Notarangelo, L. D., Villa, A., and Schwarz, K. (1999). Curr. Opin. Immunol. 11, 435–442.Nussenzweig, A., Chen, C., da Costa Soares, V., Sanchez, M., Sokol, K., Nussenzweig, M. C., and

Li, G. C. (1996). Nature 382, 551–555.Nussenzweig, M. C. (1998). Cell 95, 875–878.Oettinger, M. A. (2004). Immunol. Rev. 200, 165–181.Oettinger, M. A., Schatz, D. G., Gorka, C., and Baltimore, D. (1990). Science 248, 1517–1523.

v(d)j versus class switch recombination 107

Okazaki, I. M., Hiai, H., Kakazu, N., Yamada, S., Muramatsu, M., Kinoshita, K., and Honjo, T.(2003). J. Exp. Med. 197, 1173–1181.

Okazaki, I., Kinoshita, K., Muramatsu, M., Yoshikawa, K., and Honjo, T. (2002). Nature 416,340–345.

Okazaki, K., Davis, D. D., and Sakano, H. (1987). Cell 49, 477–485.Osipovich, O., Milley, R., Meade, A., Tachibana, M., Shinkai, Y., Krangel, M. S., and Oltz, E. M.

(2004). Nat. Immunol. 5, 309–316.Padovan, E., Casorati, G., Dellabona, P., Meyer, S., Brockhaus, M., and Lanzavecchia, A. (1993).

Science 262, 422–424.Pan, Q., Petit-Frere, C., Lahdesmaki, A., Gregorek, H., Chrzanowska, K. H., and Hammarstrom,

L. (2002). Eur. J. Immunol. 32, 1300–1308.Pan-Hammarstrom, Q., Dai, S., Zhao, Y., van Dijk-Hard, I. F., Gatti, R. A., Borresen-Dale, A. L.,

and Hammarstrom, L. (2003). J. Immunol. 170, 3707–3716.Papavasiliou, F., Casellas, R., Suh, H., Qin, X. F., Besmer, E., Pelanda, R., Nemazee, D., Rajewsky,

K., and Nussenzweig, M. C. (1997). Science 278, 298–301.Papavasiliou, F. N., and Schatz, D. G. (2000). Nature 408, 216–221.Papavasiliou, F. N., and Schatz, D. G. (2002a). J. Exp. Med. 195, 1193–1198.Papavasiliou, F. N., and Schatz, D. G. (2002b). Cell 109, (Suppl. S35–S44).Paull, T. T., and Gellert, M. (2000). Proc. Natl. Acad. Sci. USA 97, 6409–6414.Paull, T. T., Rogakou, E. P., Yamazaki, V., Kirchgessner, C. U., Gellert, M., and Bonner, W. M.

(2000). Curr. Biol. 10, 886–895.Peak, M. M., Arbuckle, J. L., and Rodgers, K. K. (2003). J. Biol. Chem. 278, 18235–18240.Perkins, E. J., Nair, A., Cowley, D. O., Van Dyke, T., Chang, Y., and Ramsden, D. A. (2002). Genes

Dev. 16, 159–164.Peters, A., and Storb, U. (1996). Immunity 4, 57–65.Petersen, S., Casellas, R., Reina-San-Martin, B., Chen, H. T., Difilippantonio, M. J., Wilson, P. C.,

Hanitsch, L., Celeste, A., Muramatsu, M., Pilch, D. R., Redon, C., Ried, T., Bonner, W. M.,Honjo, T., Nussenzweig, M. C., and Nussenzweig, A. (2001). Nature 414, 660–665.

Petersen-Mahrt, S. K., and Neuberger, M. S. (2003). J. Biol. Chem. 278, 19583–19586.Petersen-Mahrt, S. K., Harris, R. S., and Neuberger, M. S. (2002). Nature 418, 99–103.Petrie, H. T., Livak, F., Burtrum, D., and Mazel, S. (1995). J. Exp. Med. 182, 121–127.Pham, P., Bransteitter, R., Petruska, J., and Goodman, M. F. (2003). Nature 424, 103–107.Phung, Q. H., Winter, D. B., Cranston, A., Tarone, R. E., Bohr, V. A., Fishel, R., and Gearhart, P. J.

(1998). J. Exp. Med. 187, 1745–1751.Pinaud, E., Khamlichi, A. A., Le Morvan, C., Drouet, M., Nalesso, V., Le Bert, M., and Cogne, M.

(2001). Immunity 15, 187–199.Priestley, A., Beamish, H. J., Gell, D., Amatucci, A. G., Muhlmann-Diaz, M. C., Singleton, B. K.,

Smith, G. C., Blunt, T., Schalkwyk, L. C., Bedford, J. S., Jackson, S. P., Jeggo, P. A., and Taccioli,G. E. (1998). Nucleic Acids Res. 26, 1965–1973.

Qiu, J. X., Kale, S. B., Yarnell Schultz, H., and Roth, D. B. (2001). Mol. Cell 7, 77–87.Quong, M. W., Harris, D. P., Swain, S. L., and Murre, C. (1999). EMBO J. 18, 6307–6318.Quong, M. W., Romanow, W. J., and Murre, C. (2002). Annu. Rev. Immunol. 20, 301–322.Rada, C., Jarvis, J. M., and Milstein, C. (2002a). Proc. Natl. Acad. Sci. USA 99, 7003–7008.Rada, C., Williams, G. T., Nilsen, H., Barnes, D. E., Lindahl, T., and Neuberger, M. S. (2002b).

Curr. Biol. 12, 1748–1755.Radic, M. Z., Erikson, J., Litwin, S., and Weigert, M. (1993). J. Exp. Med. 177, 1165–1173.Raghavan, S. C., and Lieber, M. R. (2004). Cell Cycle. 3, 762–768.Raghavan, S. C., Swanson, P. C., Wu, X., Hsieh, C. L., and Lieber, M. R. (2004). Nature 428,

88–93.

108 darryll d. dudley ET AL.

Ramiro, A. R., Stavropoulos, P., Jankovic, M., and Nussenzweig, M. C. (2003). Nat. Immunol. 4,452–456.

Ramiro, A. R., Jankovic, M., Eisenreich, T., Difilippantonio, S., Chen-Kiang, S., Muramatsu, M.,Honjo, T., Nussenzweig, A., and Nussenzweig, M. C. (2004). Cell 118, 431–438.

Ramsden, D. A., and Gellert, M. (1998). EMBO J. 17, 609–614.Ramsden, D. A., Paull, T. T., and Gellert, M. (1997). Nature 388, 488–491.Rappold, I., Iwabuchi, K., Date, T., and Chen, J. (2001). J. Cell Biol. 153, 613–620.Raulet, D. H., Garman, R. D., Saito, H., and Tonegawa, S. (1985). Nature 314, 103–107.Razin, A., and Riggs, A. D. (1980). Science 210, 604–610.Reaban, M. E., and Griffin, J. A. (1990). Nature 348, 342–344.Reaban, M. E., Lebowitz, J., and Griffin, J. A. (1994). J. Biol. Chem. 269, 21850–21857.Regueiro, J. R., Porras, O., Lavin, M., and Gatti, R. A. (2000). Immunol. Allergy Clin. North Am.

20, 177–205.Reina-San-Martin, B., Difilippantonio, S., Hanitsch, L., Masilamani, R. F., Nussenzweig, A., and

Nussenzweig, M. C. (2003). J. Exp. Med. 197, 1767–1778.Reina-San-Martin, B., Chen, H. T., Nussenzweig, A., and Nussenzweig, M. C. (2004). J. Exp. Med.

200, 1103–1110.Revy, P., Muto, T., Levy, Y., Geissmann, F., Plebani, A., Sanal, O., Catalan, N., Forveille, M.,

Dufourcq-Labelouse, R., Gennery, A., Tezcan, I., Ersoy, F., Kayserili, H., Ugazio, A. G., Brousse,N., Muramatsu, M., Notarangelo, L. D., Kinoshita, K., Honjo, T., Fischer, A., and Durandy, A.(2000). Cell 102, 565–575.

Reynaud, C. A., Aoufouchi, S., Faili, A., and Weill, J. C. (2003). Nat. Immunol. 4, 631–638.Richards, E. J., and Elgin, S. C. (2002). Cell 108, 489–500.Riechmann, V., van Cruchten, I., and Sablitzky, F. (1994). Nucleic Acids Res. 22, 749–755.Roberts, E. C., Deed, R. W., Inoue, T., Norton, J. D., and Sharrocks, A. D. (2001). Mol. Cell. Biol.

21, 524–533.Rodgers, K. K., Villey, I. J., Ptaszek, L., Corbett, E., Schatz, D. G., and Coleman, J. E. (1999).

Nucleic Acids Res. 27, 2938–2946.Rogakou, E. P., Pilch, D. R., Orr, A. H., Ivanova, V. S., and Bonner, W. M. (1998). J. Biol. Chem.

273, 5858–5868.Rogakou, E. P., Boon, C., Redon, C., and Bonner, W. M. (1999). J. Cell Biol. 146, 905–916.Rolink, A., Melchers, F., and Andersson, J. (1996). Immunity 5, 319–330.Rolink, A. G., Schaniel, C., Andersson, J., and Melchers, F. (2001). Curr. Opin. Immunol. 13,

202–207.Roman, C. A., and Baltimore, D. (1996). Proc. Natl. Acad. Sci. USA 93, 2333–2338.Roman, C. A., Cherry, S. R., and Baltimore, D. (1997). Immunity 7, 13–24.Rooney, S., Sekiguchi, J., Zhu, C., Cheng, H. L., Manis, J., Whitlow, S., DeVido, J., Foy, D.,

Chaudhuri, J., Lombard, D., and Alt, F. W. (2002). Mol. Cell 10, 1379–1390.Rooney, S., Alt, F. W., Lombard, D., Whitlow, S., Eckersdorff, M., Fleming, J., Fugmann, S.,

Ferguson, D. O., Schatz, D. G., and Sekiguchi, J. (2003). J. Exp. Med. 197, 553–565.Rooney, S., Chaudhuri, J., and Alt, F. W. (2004). Immunol. Rev. 200, 115–131.Roth, D. B. (2003). Nat. Rev. Immunol. 3, 656–666.Roth, D. B., Menetski, J. P., Nakajima, P. B., Bosma, M. J., and Gellert, M. (1992). Cell 70,

983–991.Sadofsky, M. J., Hesse, J. E., McBlane, J. F., and Gellert, M. (1993). Nucleic Acids Res. 21,

5644–5650.Sadofsky, M. J., Hesse, J. E., and Gellert, M. (1994). Nucleic Acids Res. 22, 1805–1809.Saint-Ruf, C., Ungewiss, K., Groettrup, M., Bruno, L., Fehling, H. J., and von Boehmer, H. (1994).

Science 266, 1208–1212.

v(d)j versus class switch recombination 109

Sakai, E., Bottaro, A., and Alt, F. W. (1999a). Int. Immunol. 11, 1709–1713.Sakai, E., Bottaro, A., Davidson, L., Sleckman, B. P., and Alt, F. W. (1999b). Proc. Natl. Acad. Sci.

USA 96, 1526–1531.Sakano, H., Huppi, K., Heinrich, G., and Tonegawa, S. (1979). Nature 280, 288–294.Sakano, H., Kurosawa, Y., Weigert, M., and Tonegawa, S. (1981). Nature 290, 562–565.Sawchuk, D. J., Weis-Garcia, F., Malik, S., Besmer, E., Bustin, M., Nussenzweig, M. C., and

Cortes, P. (1997). J. Exp. Med. 185, 2025–2032.Sayegh, C. E., Quong, M. W., Agata, Y., and Murre, C. (2003). Nat. Immunol. 4, 586–593.Schable, K. F., Thiebe, R., Bensch, A., Brensing-Kuppers, J., Heim, V., Kirschbaum, T., Lamm, R.,

Ohnrich, M., Pourrajabi, S., Roschenthaler, F., Schwendinger, J., Wichelhaus, D., Zocher, I., andZachau, H. G. (1999). Eur. J. Immunol. 29, 2082–2086.

Schatz, D. G., and Baltimore, D. (1988). Cell 53, 107–115.Schatz, D. G., Oettinger, M. A., and Baltimore, D. (1989). Cell 59, 1035–1048.Schlissel, M. S., and Baltimore, D. (1989). Cell 58, 1001–1007.Schrader, C. E., Edelmann, W., Kucherlapati, R., and Stavnezer, J. (1999). J. Exp. Med. 190,

323–330.Schrader, C. E., Vardo, J., and Stavnezer, J. (2002). J. Exp. Med. 195, 367–373.Schrader, C. E., Vardo, J., and Stavnezer, J. (2003). J. Exp. Med. 197, 1377–1383.Schuler, W., Ruetsch, N. R., Amsler, M., and Bosma, M. J. (1991). Eur. J. Immunol. 21, 589–596.Schultz, L. B., Chehab, N. H., Malikzay, A., and Halazonetis, T. D. (2000). J. Cell Biol. 151,

1381–1390.Schwarz, K., Gauss, G. H., Ludwig, L., Pannicke, U., Li, Z., Lindner, D., Friedrich, W., Seger,

R. A., Hansen-Hagge, T. E., Desiderio, S., Lieber, M. R., and Bartram, C. R. (1996). Science274, 97–99.

Seidl, K. J., Bottaro, A., Vo, A., Zhang, J., Davidson, L., and Alt, F. W. (1998). Int. Immunol. 10,1683–1692.

Seidl, K. J., Manis, J. P., Bottaro, A., Zhang, J., Davidson, L., Kisselgof, A., Oettgen, H., and Alt,F. W. (1999). Proc. Natl. Acad. Sci. USA 96, 3000–3005.

Sekiguchi, J. A., Whitlow, S., and Alt, F. W. (2001). Mol. Cell 8, 1383–1390.Selsing, E., and Daitch, L. E. (1995). In ‘‘Immunoglobulin Genes’’ (T. Honjo and F. Alt, Eds.),

pp. 193–203. Academic Press, San Diego, CA.Sen, D., and Gilbert, W. (1988). Nature 334, 364–366.Serwe, M., and Sablitzky, F. (1993). EMBO J. 12, 2321–2327.Shiloh, Y. (1997). Annu. Rev. Genet. 31, 635–662.Shiloh, Y. (2001). Curr. Opin. Genet. Dev. 11, 71–77.Shinkai, Y., Rathbun, G., Lam, K.-P., Oltz, E. M., Stewart, V., Mendelsohn, M., Charron, J., Datta,

M., Young, F., Stall, A. M., and Alt, F. W. (1992). Cell 68, 855–867.Shinkai, Y., Koyasu, S., Nakayama, K., Murphy, K. M., Loh, D. Y., Reinherz, E. L., and Alt, F. W.

(1993). Science 259, 822–825.Shinkura, R., Tian, M., Smith, M., Chua, K., Fujiwara, Y., and Alt, F. W. (2003). Nat. Immunol. 4,

435–441.Shinkura, R., Ito, S., Begum, N. A., Nagaoka, H., Muramatsu, M., Kinoshita, K., Sakakibara, Y.,

Hijikata, H., and Honjo, T. (2004). Nat. Immunol. 5, 707–712.Shockett, P. E., and Schatz, D. G. (1999). Mol. Cell. Biol. 19, 4159–4166.Sikes, M. L., Gomez, R. J., Song, J., and Oltz, E. M. (1998). J. Immunol. 161, 1399–1405.Sleckman, B. P., Gorman, J. R., and Alt, F. W. (1996). Annu. Rev. Immunol. 14, 459–481.Sleckman, B. P., Bardon, C. G., Ferrini, R., Davidson, L., and Alt, F. W. (1997). Immunity 7,

505–515.Sleckman, B. P., Khor, B., Monroe, R., and Alt, F. W. (1998). J. Exp. Med. 188, 1465–1471.

110 darryll d. dudley ET AL.

Sleckman, B. P., Bassing, C. H., Hughes, M. M., Okada, A., D’Auteuil, M., Wehrly, T. D.,Woodman, B. B., Davidson, L., Chen, J., and Alt, F. W. (2000). Proc. Natl. Acad. Sci. USA 97,7975–7980.

Smith, G. C., and Jackson, S. P. (1999). Genes Dev. 13, 916–934.Sohail, A., Klapacz, J., Samaranayake, M., Ullah, A., and Bhagwat, A. S. (2003). Nucleic Acids Res.

31, 2990–2994.Spanopoulou, E., Roman, C. A., Corcoran, L. M., Schlissel, M. S., Silver, D. P., Nemazee, D.,

Nussenzweig, M. C., Shinton, S. A., Hardy, R. R., and Baltimore, D. (1994). Genes Dev. 8,1030–1042.

Spanopoulou, E., Zaitseva, F., Wang, F. H., Santagata, S., Baltimore, D., and Panayotou, G. (1996).Cell 87, 263–276.

Spicuglia, S., Kumar, S., Yeh, J. H., Vachez, E., Chasson, L., Gorbatch, S., Cautres, J., and Ferrier,P. (2002). Mol. Cell 10, 1479–1487.

Stanhope-Baker, P., Hudson, K. M., Shaffer, A. L., Constantinescu, A., and Schlissel, M. S. (1996).Cell 85, 887–897.

Stavnezer, J. (1996). Adv. Immunol. 61, 79–146.Stavnezer, J. (2000). Curr. Top. Microbiol. Immunol. 245, 127–168.Steen, S. B., Han, J., Mundy, C., Oettinger, M. A., and Roth, D. B. (1999). Mol. Cell. Biol. 19,

3010–3017.Stiff, T., O’Driscoll, M., Rief, N., Iwabuchi, K., Lobrich, M., and Jeggo, P. A. (2004). Cancer Res.

64, 2390–2396.Strahl, B. D., and Allis, C. D. (2000). Nature 403, 41–45.Su, I. H., Basavaraj, A., Krutchinsky, A. N., Hobert, O., Ullrich, A., Chait, B. T., and Tarakhovsky,

A. (2003). Nat. Immunol. 4, 124–131.Su, L.-K., and Kadesch, T. (1990). Mol. Cell. Biol. 10, 2619–2624.Sun, X. H. (2004). Adv. Immunol. 84, 43–77.Sun, X. H., Copeland, N. G., Jenkins, N. A., and Baltimore, D. (1991). Mol. Cell. Biol. 11,

5603–5611.Swanson, P. C. (2002a). Mol. Cell. Biol. 22, 1340–1351.Swanson, P. C. (2002b). Mol. Cell. Biol. 22, 7790–7801.Swanson, P. C., and Desiderio, S. (1999). Mol. Cell. Biol. 19, 3674–3683.Ta, V. T., Nagaoka, H., Catalan, N., Durandy, A., Fischer, A., Imai, K., Nonoyama, S., Tashiro, J.,

Ikegawa, M., Ito, S., Kinoshita, K., Muramatsu, M., and Honjo, T. (2003). Nat. Immunol. 4,843–848.

Taccioli, G. E., Rathbun, G., Oltz, E., Stamato, T., Jeggo, P. A., and Alt, F. W. (1993). Science 260,207–210.

Taccioli, G. E., Gottlieb, T. M., Blunt, T., Priestley, A., Demengeot, J., Mizuta, R., Lehmann, A. R.,Alt, F. W., Jackson, S. P., and Jeggo, P. A. (1994). Science 265, 1442–1445.

Taccioli, G. E., Amatucci, A. G., Beamish, H. J., Gell, D., Xiang, X. H., Torres Arzayus, M. I.,Priestley, A., Jackson, S. P., Marshak Rothstein, A., Jeggo, P. A., and Herrera, V. L. (1998).Immunity 9, 355–366.

Tashiro, J., Kinoshita, K., and Honjo, T. (2001). Int. Immunol. 13, 495–505.Thai, T. H., Purugganan, M. M., Roth, D. B., and Kearney, J. F. (2002). Nat. Immunol. 3, 457–462.Thompson, C. B. (1995). Immunity 3, 531–539.Tian, M., and Alt, F. W. (2000). J. Biol. Chem. 275, 24163–24172.Tiegs, S. L., Russell, D. M., and Nemazee, D. (1993). J. Exp. Med. 177, 1009–1020.Tillman, R. E., Wooley, A. L., Khor, B., Wehrly, T. D., Little, C. A., and Sleckman, B. P. (2003).

J. Immunol. 170, 5–9.Tohma, S., and Lipsky, P. E. (1991). J. Immunol. 146, 2544–2552.

v(d)j versus class switch recombination 111

Tonegawa, S. (1983). Nature 302, 575–581.Tonegawa, S., Hozumi, N., Matthyssens, G., and Schuller, R. (1977). Cold Spring Harb. Symp.

Quant. Biol. 41, 877–889.Tsai, C. L., and Schatz, D. G. (2003). EMBO J. 22, 1922–1930.Tsai, C. L., Drejer, A. H., and Schatz, D. G. (2002). Genes Dev. 16, 1934–1949.Udvardy, A. (1999). EMBO J. 18, 1–8.Urbanek, P., Wang, Z. Q., Fetka, I., Wagner, E. F., and Busslinger, M. (1994). Cell 79, 901–912.van Gent, D. C., Mizuuchi, K., and Gellert, M. (1996a). Science 271, 1592–1594.van Gent, D. C., Ramsden, D. A., and Gellert, M. (1996b). Cell 85, 107–113.van Gent, D. C., Hiom, K., Paull, T. T., and Gellert, M. (1997). EMBO J. 16, 2665–2670.Villa, A., Santagata, S., Bozzi, F., Giliani, S., Frattini, A., Imberti, L., Gatta, L. B., Ochs, H. D.,

Schwarz, K., Notarangelo, L. D., Vezzoni, P., and Spanopoulou, E. (1998). Cell 93, 885–896.Villa, A., Santagata, S., Bozzi, F., Imberti, L., and Notarangelo, L. D. (1999). J. Clin.

Immunol. 19, 87–97.Villa, A., Sobacchi, C., Notarangelo, L. D., Bozzi, F., Abinum, M., Abrahamsen, T. G., Arkwright,

P. D., Baniyash, M., Brooks, E. G., Conley, M. E., Cortes, P., Duse, M., Fasth, A., Filipovich,A. M., Infante, A. J., Jones, A., Mazzolari, E., Muller, S. M., Pasic, S., Rechavi, G., Sacco, M. G.,Santagata, S., Schroeder, M. L., Seger, R., Strina, D., Ugazio, A., Valiaho, J., Vihinen, M., Vogler,L. B., Ochs, H., Vezzoni, P., Friedrich, W., and Schwarz, K. (2001). Blood 97, 81–88.

Villey, I., Caillol, D., Selz, F., Ferrier, P., and de Villartay, J. P. (1996). Immunity 5, 331–342.Villey, I., Quartier, P., Selz, F., and de Villartay, J. P. (1997). Eur. J. Immunol. 27, 1619–1625.von Boehmer, H., Aifantis, I., Feinberg, J., Lechner, O., Saint-Ruf, C., Walter, U., Buer, J., and

Azogui, O. (1999). Curr. Opin. Immunol. 11, 135–142.Wada, T., Takei, K., Kudo, M., Shimura, S., Kasahara, Y., Koizumi, S., Kawa-Ha, K., Ishida, Y.,

Imashuku, S., Seki, H., and Yachie, A. (2000). Clin. Exp. Immunol. 119, 148–155.Waldmann, T. A., Broder, S., Goldman, C. K., Frost, K., Korsmeyer, S. J., and Medici, M. A. (1983).

J. Clin. Invest. 71, 282–295.Ward, I. M., Reina-San-Martin, B., Olaru, A., Minn, K., Tamada, K., Lau, J. S., Cascalho, M.,

Chen, L., Nussenzweig, A., Livak, F., Nussenzweig, M. C., and Chen, J. (2004). J. Cell Biol. 165,459–464.

Warren, W. D., and Berton, M. T. (1995). J. Immunol. 155, 5637–5646.Wayne, J., Suh, H., Misulovin, Z., Sokol, K. A., Inaba, K., and Nussenzweig, M. C. (1994a).

Immunity 1, 95–107.Wayne, J., Suh, H., Sokol, K. A., Petrie, H. T., Witmer-Pack, M., Edelhoff, S., Disteche, C. M., and

Nussenzweig, M. C. (1994b). J. Immunol. 153, 5491–5502.Weichhold, G. M., Klobeck, H.-G., Ohnheiser, R., Combriato, G., and Zachau, H. G. (1990).

Nature 347, 90–92.West, R. B., and Lieber, M. R. (1998). Mol. Cell. Biol. 18, 6408–6415.West, R. B., Yaneva, M., and Lieber, M. R. (1998). Mol. Cell. Biol. 18, 5908–5920.Whitehurst, C. E., Chattopadhyay, S., and Chen, J. (1999). Immunity 10, 313–322.Whitehurst, C. E., Schlissel, M. S., and Chen, J. (2000). Immunity 13, 703–714.Willerford, D. M., Swat, W., and Alt, F. W. (1996). Curr. Opin. Genet. Dev. 6, 603–609.William, J., Euler, C., Christensen, S., and Shlomchik, M. J. (2002). Science 297, 2066–2070.Williams, M., and Maizels, N. (1991). Genes Dev. 5, 2353–2361.Winter, E., Krawinkel, U., and Radbruch, A. (1987). EMBO J. 6, 1663–1671.Wold, M. S. (1997). Annu. Rev. Biochem. 66, 61–92.Woo, C. J., Martin, A., and Scharff, M. D. (2003). Immunity 19, 479–489.Wu, C., Bassing, C. H., Jung, D., Woodman, B. B., Foy, D., and Alt, F. W. (2003). Immunity 18,

75–85.

112 darryll d. dudley ET AL.

Wu, X., Ranganathan, V., Weisman, D. S., Heine, W. F., Ciccone, D. N., O’Neill, T. B., Crick, K. E.,Pierce, K. A., Lane, W. S., Rathbun, G., Livingston, D. M., and Weaver, D. T. (2000). Nature405, 477–482.

Wuerffel, R. A., Du, J., Thompson, R. J., and Kenter, A. L. (1997). J. Immunol. 159, 4139–4144.Xia, Z., Morales, J. C., Dunphy, W. G., and Carpenter, P. B. (2001). J. Biol. Chem. 276, 2708–2718.Xu, Y., Davidson, L., Alt, F. W., and Baltimore, D. (1996). Immunity 4, 377–385.Yamaoka, S., Courtois, G., Bessia, C., Whiteside, S. T., Weil, R., Agou, F., Kirk, H. E., Kay, R. J.,

and Israel, A. (1998). Cell 93, 1231–1240.Yancopoulos, G. D., and Alt, F. W. (1985). Cell 40, 271–281.Yancopoulos, G. D., and Alt, F. W. (1986). Ann. Rev. Immunol. 4, 339–368.Yannoutsos, N., Barreto, V., Misulovin, Z., Gazumyan, A., Yu, W., Rajewsky, N., Peixoto, B. R.,

Eisenreich, T., and Nussenzweig, M. C. (2004). Nat. Immunol. 5, 443–450.Yarnell Schultz, H., Landree, M. A., Qiu, J. X., Kale, S. B., and Roth, D. B. (2001). Mol. Cell 7,

65–75.Yates, P. R., Atherton, G. T., Deed, R. W., Norton, J. D., and Sharrocks, A. D. (1999).

EMBO J. 18, 968–976.Yoshikawa, K., Okazaki, I. M., Eto, T., Kinoshita, K., Muramatsu, M., Nagaoka, H., and Honjo, T.

(2002). Science 296, 2033–2036.Young, F., Ardman, B., Shinkai, Y., Lansford, R., Blackwell, T. K., Mendelsohn, M., Rolink, A.,

Melchers, F., and Alt, F. W. (1994). Genes Dev. 8, 1043–1057.Yu, K., and Lieber, M. R. (2000). Mol. Cell. Biol. 20, 7914–7921.Yu, K., Chedin, F., Hsieh, C. L., Wilson, T. E., and Lieber, M. R. (2003). Nat. Immunol. 4,

442–451.Yu, K., Huang, F. T., and Lieber, M. R. (2004). J. Biol. Chem. 279, 6496–6500.Yu, W., Misulovin, Z., Suh, H., Hardy, R. R., Jankovic, M., Yannoutsos, N., and Nussenzweig, M. C.

(1999a). Science 285, 1080–1084.Yu, W., Nagaoka, H., Jankovic, M., Misulovin, Z., Suh, H., Rolink, A., Melchers, F., Meffre, E., and

Nussenzweig, M. C. (1999b). Nature 400, 682–687.Yurchenko, V., Xue, Z., and Sadofsky, M. (2003). Genes Dev. 17, 581–585.Zachau, H. G. (1993). Gene. 135, 167–173.Zarrin, A. A., Alt, F. W., Chaudhuri, J., Pasquier, L., Stokes, N., and Tian, M. (2004). Nat. Immunol.

5, 1275–1281.Zhang, J., Bottaro, A., Li, S., Stewart, V., and Alt, F. W. (1993). EMBO J. 12, 3529–3537.Zhao, S., Weng, Y. C., Yuan, S. S., Lin, Y. T., Hsu, H. C., Lin, S. C., Gerbino, E., Song, M. H.,

Zdzienicka, M. Z., Gatti, R. A., Shay, J. W., Ziv, Y., Shiloh, Y., and Lee, E. Y. (2000). Nature 405,473–477.

Zheng, B., Han, S., Spanopoulou, E., and Kelsoe, G. (1998). Immunol. Rev. 162, 133–141.Zhu, C., Bogue, M. A., Lim, D. S., Hasty, P., and Roth, D. B. (1996). Cell 86, 379–389.Zhu, C., Mills, K. D., Ferguson, D. O., Lee, C., Manis, J., Fleming, J., Gao, Y., Morton, C. C., and

Alt, F. W. (2002). Cell 109, 811–821.Zhuang, Y., Soriano, P., and Weintraub, H. (1994). Cell 79, 875–884.

Isoforms of Terminal Deoxynucleotidyltransferase:Developmental Aspects and Function

To-Ha Thai1 and John F. Kearney

Division of Developmental and Clinical Immunology, Department of Microbiology,University of Alabama at Birmingham, Birmingham, Alabama 35204

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1131. V(D)J Recombination and Mediating Factors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1142. Junctional Diversity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1143. Origin of TdT.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1154. Transcriptional Regulation of the TdT Gene (Dntt) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1175. TdT and Its Splice Variants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1196. TdT Splice Variants and Junctional Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1237. The TdT Protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1248. TdT-Interacting Proteins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1269. TdT Splice Variants and Repertoire Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

10. Biochemical Properties and Substrate Specificity of TdT .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13011. Expression of Human TdT in Human Leukemias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13112. Possible Aberrant Activity of Human TdT in Leukemias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13213. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

Abstract

The immune system develops in a series of programmed developmental stages.Although recombination-activating gene (RAG) and nonhomologous end-joining(NHEJ) proteins are indispensable in the generation of immunoglobulins andT-cell receptors (TCRs), most CDR3 diversity is contributed by nontemplatedaddition of nucleotides catalyzed by the nuclear enzyme terminal deoxynucleoti-dyltransferase (TdT) and most nucleotide deletion is performed by exonucleases atV(D)J joins. Increasing TdT expression continuing into adult life results inN region addition and diversification of the T and B cell repertoires. In severalspecies including mice and humans, there are multiple isoforms of TdT resultingfrom alternative mRNA splicing. The short form (TdTS) produces N additionsduring TCR and B-cell receptor (BCR) gene rearrangements. Other long isoforms,TdTL1 and TdTL2, have 30 ! 50 exonuclease activity. The two forms of TdTtherefore have distinct and opposite functions in lymphocyte development. Theenzymatic activities of the splice variants of TdT play an essential role in thediversification of lymphocyte repertoires by modifying the composition and lengthof the gene segments involved in the production of antibodies and T-cell receptors.

1Present address: CBR Institute for Biomedical Research, 138 Warren Alpert Building, Boston,Massachusetts 02115.

113advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

114 to-ha thai and john f. kearney

1. V(D)J Recombination and Mediating Factors

Diversity in the antigen-binding or variable region of immunoglobulins andT-cell receptors (TCRs) results from combinatorial rearrangements of thevariable, diversity, and joining gene segments by a process known as V(D)Jrecombination (Hozumi and Tonegawa, 1976). This process normally takesplace at specific stages in developing B and T lymphocytes (Ghia et al., 1996;Haynes and Heinly, 1995; Melchers et al., 1995).

V(D)J recombination is a site-specific reaction that is initiated by lymphoid-specific recombination-activating gene (RAG)-1 and RAG-2 proteins thatsynapse with recombination signal sequences (RSSs) flanking all immunoglob-ulin- and TCR-coding gene segments (Bassing et al., 2002; Fugmann et al.,2000; Grawunder and Harfst, 2001; Lewis, 1994; Oettinger, 1999; Schatz,1997) (Fig. 1). RSSs are composed of a conserved palindromic heptamer andan AT-rich nonamer separated by nonconserved 12- or 23-bp spacers. V(D)Jrecombination occurs predominantly between two coding gene segmentsflanked by RSSs that contain 12- and 23-bp spacers, respectively. This phe-nomenon is known as the 12/23-bp rule. After synapsis of coding gene seg-ments, DNA double-stranded breaks (DSBs) or single-stranded (ss) nicks aregenerated precisely by RAG proteins at the border of RSSs flanking each genesegment. This is followed by a trans-esterification reaction, catalyzed by RAGproteins, in which the 30-OH of the coding strand attacks the opposite DNAstrand (or strand transfer) to form closed hairpin coding ends and blunt50-phosphorylated RSS ends. Hairpins are then resolved by the endonucleaseArtemis (Ma et al., 2002) or other endonucleases generating predominantlycoding ends with 30 or 50 extensions. Subsequently, signal ends or RSS endsare usually precisely joined, whereas coding ends will undergo modificationssuch as nucleotide deletion and addition. The joining phase of the V(D)Jreaction is completed by ubiquitously expressed nonhomologous end-joining(NHEJ) and DSB repair proteins, including Ku70, Ku86, the catalytic subunitof DNA-dependent protein kinase (DNA-PKcs), XRCC4, and ligase IV.

To date, immunoglobulins and TCRs of all vertebrate taxa examined, includ-ing mammals, avians, reptiles, amphibians, teleosts, and cartilaginous fish, aregenerated through V(D)J recombination (Hawke et al., 1996; Kerfourn et al.,1996; Partula et al., 1996; Tjoelker et al., 1990; Turchin and Hsu, 1996).

2. Junctional Diversity

Although RAG and NHEJ proteins are crucial in the generation of immu-noglobulins and TCRs, the majority of junctional diversity is contributed bynontemplated addition (N addition) and deletion of nucleotides at V(D)J joins

isoforms of tdt 115

(Fig. 1). V(D)J joins of terminal deoxynucleotidyltransferase (TdT)-deficientmice are devoid of N nucleotides, strongly demonstrating that TdT is responsi-ble for N addition (Gilfillan et al., 1993; Komori et al., 1993). The identity of theexonucleases involved in nucleotide deletion has proved to be more elusive.

TdT is a nuclear enzyme found predominantly in developing B and T cells inthe bone marrow and the thymus, respectively. However, this strict definitionmay not hold true, because TdT transcripts can also be detected in bonemarrow cells with the phenotypes of myeloid progenitors (see below). TdT isexpressed in fetal human B- and T-cell precursors, and in fetal mice it is barelydetectable. In many vertebrates, the TdT gene (Dntt) is conserved whereV(D)J recombination occurs (Hawke et al., 1996; Kerfourn et al., 1996;Partula et al., 1996; Tjoelker et al., 1990; Turchin and Hsu, 1996).

3. Origin of TdT

TdT belongs to the DNA polymerase (Pol) family X. The amino acid sequenceof DNA polymerase family X members contains the conserved motifGGFRRGKLQGHDVDFLI, for which a function has not yet been deter-mined. However, the underlined D residue is shown to be involved in nucleo-tide binding (see below). It has been suggested that TdTand Polb, implicated inDNA base excision repair (Pelletier et al., 1994; Sawaya et al., 1994; Sugo et al.,2000), share a more recent common ancestral gene. A close examination of thePol X family tree reveals that TdT and the newly identified mammalian Pol m(Aoufouchi et al., 2000; Dominguez et al., 2000) are derived from a more recentcommon ancestor. In addition, an ancient divergence occurs among TdT, Pol m,and Pol b, a close relative of Pol l (Aoufouchi et al., 2000). Pol m has 41% aminoacid identity to TdT and is preferentially expressed in tonsillar germinal centerB cells. These results may indicate that TdT transcripts reportedly detected intonsillar B cells might be those of Pol m. If that is the case, TdT is notreexpressed in mature B cells, as has been suggested (Girschick et al., 2001).Polm has intrinsic TdTactivity; however, unlike TdT, Polm polymerizing activityis enhanced by a template strand independently of cations, and nucleotideinsertion is rather random. These data suggest that the switch to templateindependence evolved more recently and that the common ancestor of TdTand Pol m is probably template dependent. Advances in genomics have allowedus to tentatively identify the putative ancestral gene in the Ciona intestinalisgenome from which TdT and Pol m are derived (Fig. 2). TdT, Pol m, Pol b, andPol l appear to be descendents of the DNA-dependent DNA Pol X familybelonging to the archaebacteria group.

116

Figure 2 Identification of the Dntt putative ancestral gene. The rooted near-neighbor-joiningphylogenetic tree was generated from aligned sequences retrieved from the NCBI, Takifugu, andCiona databases. Note the sequence annotated as Takifugu TdT falls within the Pol m cluster,suggesting that this sequence might represent TdT, not Pol m.

isoforms of tdt 117

4. Transcriptional Regulation of the TdT Gene (Dntt)

Human, bovine, mouse, and rat TdT genes consist of 13 exons, most of whichlocate in the 30 half of the respective gene (Thai et al., 2002; Fig. 3). TdT geneexpression is normally restricted to B- and T-cell precursors. Undoubtedly,mechanisms employed to regulate a gene that is inactive in mature lymphoidcells would be different from those of a gene such as Ig m that is constitutivelyactive. Indeed, mapping studies reveal that the TdT promoter is unusual inmany aspects (Ernst et al., 1996, 1999; Hahm et al., 1994). The minimalpromoter spans a region from positions �111 to þ58, and three controlelements are found within this region. The TdT promoter lacks a TATA boxnormally located 25–30 bp upstream of the transcriptional start site, and

Figure 1 V(D)J recombination. Initiation of V(D)J recombination: RAG-1 and RAG-2 bind andcreate a single-strand nick at the border of the recombination signal sequence (RSS, triangles) andcoding sequence. Nicking: The RAG complex binds stably to the pair of RSS, forming a synapticcomplex that is necessary for the cleavage reaction to occur. Cleavage and hairpin formation: Thesynaptic complex converts the nick into double-strand breaks through a trans-esterification reaction,generating coding end hairpins and blunt signal ends, held together in a postcleavage complex byRAG-1/2. Hairpin resolution: The endonuclease Artemis, complexed with and phosphorylated byDNA-PKcs, resolves the hairpins; the signal joint is formed by precise, head–head ligation of RSS,using the NHEJ machinery. Coding end modification and joining: P addition to, nucleotide deletionof (by TdTL and other, yet to be identified exonucleases) and N addition to (by TdTS) coding endstake place before end joining by the NHEJ machinery.

Figure 3 Genomic organization of human, bovine, mouse, and rat Dntt. The proposed genomicorganization of human, bovine, mouse, and rat TdT genes was based on NCBI and Celera genomicdatabases and data from others (Thai and Kearney, 2004); newly identified exons II, VII, and XIIare represented by slashed boxes, respectively.

118 to-ha thai and john f. kearney

instead contains an initiator (Inr) element (positions �3 to þ5) that overlapsthe transcriptional start site. The TFIID complex can recognize the TATAbox, as well as the Inr element. Mutations of nucleotides at a site where aTATA box might normally reside, at the �30 position, do not affect TdTpromoter function. Likewise, substitution of a genuine TATA box at the �30position only slightly enhances promoter activity relative to the wild-typepromoter; however, promoter activity remains lymphocyte specific. In con-trast, substitutions of two nucleotides within the Inr region completely disruptpromoter activity in lymphoid and nonlymphoid cell lines, despite the pres-ence of a TATA box. These results suggest that, although weak, the Inr isthe core promoter element that functions in concert with other regulators toactivate TdT transcription in a non-tissue-restricted manner. However, despitethe high degree of conservation between the human and the mouse promoterregion, the Inr element is not totally conserved in the human promoter.The human promoter contains several inserts; of interest are the 27-bp insertnear the site where the mouse Inr should be and the 9-bp insert close to the D0

element. The DNA sequence of the 27-bp insert resembles that of the Inrconsensus, suggesting that spacing constraints between the start site andupstream elements are quite loose.

The activity of the Inr is enhanced by downstream basal elements (DBEs)located downstream of the transcriptional start site (positions þ33 to þ58).Mutations in this region abrogate promoter activity in both lymphoidand nonlymphoid cell lines, but factors controlling DBE function have notbeen determined. TdT tissue-specific expression is regulated by a third region

isoforms of tdt 119

called the D0 element and located 60 bp upstream of the start site (positions�70 to �48). Biochemical and genetic studies identify two classes of transcrip-tion factors that bind to this region. The first factor, Elf-1, belongs to the Ets classof proteins. Trans-activation and expression studies show that Elf-1 activates TdTtranscription in immature T and B cells by binding to the second region withinthe D0 element. Mutations that reduce Elf-1 binding but not other Ets proteinswithin this region result in the selective loss of promoter activity in lymphocytes,despite the presence of an inserted TATA box. It is puzzling why Elf-1 wouldconfer stage-specific expression to the TdT gene, because Elf-1 has been impli-cated in the inducible activation of other genes in mature T and B cells. Inaddition, promoter methylation has also been implicated in TdT tissue-specificexpression (Nourrit et al., 1999); however, this study needs to be confirmed. Thesecond factor, Lyf-1, belongs to the Ikaros family of transcription regulators thatintimately associate with pericentromeric heterochromatin and are implicated inheritable gene inactivation (Trinh et al., 2001). Studies demonstrate that Ikarosdimers bind to two regions within the D0 element and can compete with Elf-1activator for binding. Mutations that disrupt Ikaros binding also abrogate TdTdownregulation on lymphocyte differentiation, suggesting that binding of Ikarosdimers within the D0 region represses TdT transcription. Ikaros-dependentrepression of TdT transcription is accompanied by chromatin remodeling, asshown by nuclease sensitivity. Both Ikaros binding and chromatin alterationsprecede pericentromeric repositioning of the TdT gene. Together, these resultslead to the development of the current working model. On lymphocyte differen-tiation, Ikaros dimers compete with Elf-1 for occupancy at the D0 element,thus displacing Elf-1. This event appears to be reversible and is accompanied bychromatin remodeling. Binding of Ikaros dimers to the D0 element facilitatestheir association with multimeric Ikaros complexes positioned at pericentromericfoci, thereby recruiting the TdT gene to these foci to be inactivated.

5. TdT and Its Splice Variants

Three alternative TdT splice variants are found in human, cattle, mouse, andrat (Doyen et al., 1993; Koiwai et al., 1986; Takahara et al., 1994)[Thai and Kearney, 2004; and National Center for Biotechnology Informal(NCBI) rat genomic data base]. In humans, mice, and cattle, the inclusion ofexons XII and VII in the mature transcripts gives rise to TdTL1 and TdTL2,respectively. The exclusion of both exons VII and XII results in the generationof TdTS. In rat, the inclusion of exon II produces TdTL2, whereas TdTL1is derived from mature transcripts containing exon XII. Rat TdTS (rTdTS) isgenerated through the constitutive splicing of both exons II and XII (Fig. 4).Although the presence of these alternatively spliced exons is conserved

Figure 4 The proposed alternative splicing patterns of human, bovine, mouse, and rat Dntt. Redand blue dashed lines represent alternative splicing; green dashed lines indicate constitutivesplicing during normal lymphocyte development.

120 to-ha thai and john f. kearney

evolutionarily, their nucleotide and thus deduced amino acid sequences are not(Fig. 5A and B). These observations suggest that the L1 and L2 inserts may serveas structural rather than protein interaction domains, which allow TdTL1 andTdTL2 to assume conformations different than those of TdTS. The change ofconformation, in turn, may confer different DNA-modifying activities.

In mice, during normal B-cell development in the bone marrow, mTdTSand mTdTL1 transcripts as well as proteins are detected in both cyclingand noncycling pro-B cells where heavy (H) chain gene rearrangementsoccur. In contrast, in cycling and noncycling pre-B cells where light (L) chaingene rearrangements take place, mTdTL1 transcripts and proteins are primarilyseen (Thai et al., 2002). Surprisingly, neither mTdTL2 transcripts nor proteinsare found in either population during normal bone marrow B-cell development;however, the mTdTL2 isoform is readily detected in the transformed myeloidcell line HTX-1. In the bone marrow, the Mac1þGr1þB220þThy1þ andMac1þGr1þB220�Thy� subpopulations express predominantly mTdTS where-as mTdTL1 expression is restricted primarily to the Mac1þGr1þB220�Thy1þ

population; mTdTL2 is not detected in any of these populations (T.-H.Thai, unpublished data). These observations suggest that Dntt transcriptionoccurs in myeloid and T and B precursors in the bone marrow before commit-ment. In the thymus, all populations of T-cell progenitors, excluding the DN1population, (DN2, DN3, ISP, and DP) express mTdTS and mTdTL1 both at thetranscript and protein levels. Again, mTdTL2 is not detected during normalT-cell development in the thymus (T.-H. Thai, unpublished data).

isoforms of tdt 121

Unlike TdT expression patterns in the mouse, during normal human B-celldevelopment, hTdTS and hTdTL2, not hTdTL1, are predominantly expressed inpro-B (CD34þCD19þ sIgM�) and pre-B (CD34�CD19þsIgM�) cells. Asexpected, none of the isoforms are seen in the mature (CD34�CD19þsIgMþ)population. However, this ordered pattern of expression is altered in transformedcell lines because, along with the other isoforms, hTdTL1 is detected in humancell line 697, representative of the pre-B-cell stage, where it is not normallydetected (Thai, unpublished data). To date the expression of hTdT isoformsduring normal human thymocyte development has not been formally assessed.In contrast to adult bone marrow pro-B cells, hTdTL1 is not detected in fetalthymocytes at any age. However, hTdTL1 is detected in transformed T cell lines.Both hTdTS and hTdTL2 are expressed in all stages of thymocyte development[DN (CD4�CD8�), DP (CD4þ CD8þ), and CD4þ and CD8þ thymocytes], aswell as in transformed cell lines. The level of hTdTL2 transcripts is consistentlyhigher than that of hTdTS. On fetal day 91, all three populations studied: DN,DP, and CD4þ and CD8þ thymocytes express less hTdTS than hTdTL2; and thislevel of differential expression persists until fetal day 111, albeit slightly lesspronounced. Moreover, TdT proteins are readily detected in the respectivethymocyte subpopulations by fluorescence-activated cell scrting (FACS) ana-lyses, using TdT-specific monoclonal antibodies (mAbs) (Thai, 2004). In contrastto human fetal thymocytes, both hTdTS and hTdTL2 are expressed at similarlevels in all stages of adult thymocyte development (Thai, 2004). These datasupport previous studies demonstrating that the degree of N addition in TCR-bDJ junctions of human thymocytes increases with age, but the extent of nucleo-tide nibbling remains constant (George and Schroeder, 1992). The persistentexpression of hTdTS and hTdTL2 in all stages of thymocyte development alsoexplains the presence of N addition and nucleotide deletion in human TCR-b, a,g, and d chain genes (Breit and Van Dongen, 1994; Dave et al., 1993; Geneveeet al., 1994; Quiros Roldan et al., 1995; Yoshikai et al., 1986).

Therefore, in normal human B and T cells, exon XII of human Dntt is alwaysexcluded, whereas exon VII can be included (generating hTdTL2) or excluded(generating hTdTS). In contrast, in normal mouse B and T cell progenitors,exon XII of mouse Dntt can be included (generating mTdTL1) or excluded(generating mTdTS), whereas exon VII is always excluded. In the case of cattleand rats, exons XII and VII can be included (producing bTdTL1 and bTdTL2,respectively) or excluded (producing bTdTS) (Takahara et al., 1994, and NCBIrat genomic database, respectively), because transcripts of all three splicevariants can be detected in these two species (Fig. 4). Detailed studies onexpression patterns of bovine and rat TdT splice variants during normal B- andT-cell development should be carried out to confirm these observations. Thus,

122

isoforms of tdt 123

alternative splicing, which appears to be species specific, regulates human,bovine, mouse, and rat TdT isoform expression.

6. TdT Splice Variants and Junctional Diversity

It is evident that mammalian TdT splice variants are alternatively expressedduring normal B- and T-cell development. In addition, we and others haveshown that mTdTS and hTdTS clearly catalyze N addition in V(D)J joins(Benedict et al., 2000; Bentolila et al., 1995; Thai, 2005; Thai et al., 2002).Therefore, it is logical to ask whether the long isoforms possess enzymaticactivity, and if so, do they contribute to junctional diversity?

We have shown both biochemically and genetically that mTdTL1, hTdTL1,and hTdTL2 possess 30 ! 50 exonuclease activity. In a standard recombinationassay, all three long isoforms catalyze the removal of nucleotides from artificialcoding ends but not signal ends (Thai, 2005; Thai et al., 2002). Moreover, theconcomitant expression of TdTS (transferase) and TdTL1 and/or TdTL2 (exo-nuclease) appears to modulate the activity of each other during V(D)J recombi-nation. The activity of mTdTL2 has not been extensively studied, because itsidentity has just been determined in our laboratory. However, preliminary datasuggest that it also possesses 30 ! 50 exonuclease activity. Thus, in mouse andhuman, both TdT long isoforms catalyze nucleotide deletion; however, thereexists a strong evolutionary constraint on alternative splicing to express only oneexonuclease (TdTL1 in mouse and TdTL2 in human) during V(D)J recombina-tion in normal B and T progenitors. One possible explanation for the differentialexpression patterns of mammalian TdTsplice variants is that the coexpression ofall three human TdT isoforms greatly reduces the recombination frequency inthe standard recombination assay (Thai, 2005).

It is noteworthy to point out that coding joins retrieved from cells thatexpress only the long isoform (TdTL1 or TdTL2) contain no P nucleotides,suggesting that TdTL1/2 may be responsible for P nucleotide removal.

Superficial examination of CDR3 joins in bulk peripheral splenic B cells andbone marrow revealed little evidence of a decrease in exonucleolytic activity at V,D, or J coding ends in TdT-deficient mice (Gilfillan et al., 1993; Komori et al.,

Figure 5 Genomic and deduced amino acid sequence analyses of TdT splice variants. (A) Thehuman, bovine, mouse, and rat TdT genomic sequences were retrieved from NCBI and Celeragenome databases, and alignment was done to identify L1 and L2 inserts. (B) The deduced aminoacid sequences of L1 and L2 inserts from all four species were aligned, L1 and L2 inserts are inred and blue, respectively; dashed lines indicate missing residues. In contrast, rat L2 (green) isencoded by exon II, not exon VII, as in human, bovine, and mouse. Moreover, rat exon IV is splicedinto exon V, resulting in the deletion of 11 residues (purple dashed line) from exon V.

124 to-ha thai and john f. kearney

1993). These observations have been construed as evidence that TdTL isoformsare redundant or are not responsible for N deletion during the joining process.However, careful examination of CDR3 joins in FACS-purified B cell progenitorsin the C and C0 fractions revealed that only about 0–7% of wild-type CDR3 V–Dand D–J joins show homology-directed recombination involving one to fivenucleotides, whereas about 27–70% of CDR3 joins in TdT-deficient mice showhomology-directed recombination involving one to five nucleotides (I. Ivalyo andH. Schroeder, personal communications). Therefore when C and C0 B cells areanalyzed before the effects of BCR-mediated selection pressures, these homo-logies are preserved, most likely because of the lack of exonucleolytic activity inthe absence of TdTL. Although this is indirect evidence, these observationsstrongly support a role for TdTL in the removal of these homologous nucleotides.

Undoubtedly there may exist other exonucleases capable of catalyzingnucleotide deletion from coding ends. However, because of the expression ofTdTL isoforms, which is confined primarily to developing B and T cells andtheir presence along with TdTS in cells undergoing V(D)J recombination, wepropose that mammalian TdT long isoform exonuclease activity contributes tothe generation of diversity in V(D)J joins (Fig. 1).

7. The TdT Protein

TdT contains several structural and functional domains (Fig. 6). Thesedomains appear to be conserved among species; therefore, only the humanterminal deoxynucleotidyltransferase (hTdT) structure is described. Like mostnuclear proteins, hTdT contains a conserved nuclear localization signal,PRKKRPR. Three conserved exonuclease (Exo) motifs and three putative

Figure 6 Domain structure of human, bovine, mouse, and rat TdT isoforms. (A) Domainstructure of (top to bottom) human, bovine, and mouse TdT isoforms. (B) Domain structure ofrat TdTL2; note the location of the L2 insert. Rat TdTS (rTdTS) and TdTL1 (rTdTL1) are similarin structure to those of the other three species.

Figure 7 Identification of exonuclease motifs in human, bovine, mouse, and rat TdT. Amino acidsequence alignment of mouse, rat, bovine, and human TdTL isoforms was done to identify exonu-clease core motifs; conserved residues are colored green. The third exonuclease motif is underlinedto show its upstream position proximal to the L1 insert (red).

isoforms of tdt 125

cAMP-dependent phosphorylation consensus sequences are found wherethreonine can be phosphorylated. cAMP site 1 locates just upstream of theExo I motif, whereas site 2 abuts the N-terminal end of the Exo II motif, andall three motifs can be found in the amino acid sequences of human, mouse,bovine, and rat TdT splice variants (Fig. 7; residues in green have been shownto be important for exonuclease activity). Exo motifs have been shown toconfer Exo activity to mTdTL1 (Thai et al., 2002). Although human and bovineTdT can be phosphorylated by a cAMP-dependent kinase in vitro, theiractivity does not appear to be modulated (Chang and Bollum, 1982). On theother hand, phosphorylation has been shown to mediate RAG-2 cell cycle–dependent degradation (Li et al., 1996); perhaps TdT is similarly marked fordegradation during lymphocyte differentiation. Alternatively, phosphorylationmay change the conformation, thereby revealing cryptic DNA-modifyingactivities in TdT, as seen in Artemis (Ma et al., 2002). Whether the proximityof two Exo motifs to two cAMP sites mutually affects their functions has yetto be determined. The DNA-binding domain is located just upstream of thethird cAMP site and the L2 insert. This domain consists of two peptides(DTEGIPCLGSK and GIIEEIIEDGESSEVK) that have been shown tocovalently cross-link to DNA under optimal photolabeling conditions(Farrar et al., 1991). The nucleotide-binding domain is located at the carboxylend of the L2 insert and overlaps the Pol X consensus sequence(GGFRRGKKMGHDVDFLI), where the D residue is found to be crucialfor nucleotide binding (Yang et al., 1994). The Exo III motif lies just upstreamof the human, bovine, mouse, and rat TdT L1 insert, which is located at theN-terminal end of the putative tyrosine phosphorylation site. It has not yetbeen determined whether this site is actually phosphorylated and whether

126 to-ha thai and john f. kearney

phosphorylation modulates TdT activity, but the site appears to be conservedin all TdT proteins identified to date.

After DNA damage induced by radiation, carcinogens, or oxidative freeradicals, cell cycle checkpoints are initiated to allow complete repair of dam-aged DNA, thus curtailing the propagation of permanently altered geneticmaterials. Cell cycle checkpoint and DNA repair proteins, including thetumor suppressor BRCA1 protein, the p53-binding protein (53BP1), theyeast cell cycle checkpoint protein RAD9, the DNA repair protein XRCC1,DNA ligases III and IV, and many others, all contain the globular BRCA1 C-terminal region domain, termed BRCT (Callebaut and Mornon, 1997; Huytonet al., 2000). BRCT modules have been shown to be involved in both BRCT–BRCT and BRCT–non-BRCT-mediated protein interactions and in associa-tions with DNA strand breaks. Among all Pols in Pol X, only TdT and its closerelatives Pol m and Pol l contain the BRCT domain located in the N terminus.The BRCT domain overlaps with Exo I and cAMP sites. A contribution ofthe BRCT domain to TdT function has not yet been determined. Studies havesuggested that the hTdT BRCT domain interacts with Pso4, a human factorinvolves in DNA repair and recombination (Mahajan and Mitchell, 2003).

8. TdT-Interacting Proteins

Because TdT is involved in the diversification of immunoglobulins and TCRsduring V(D)J recombination, it follows that TdT must interact with other factorsinvolved in V(D)J recombination. Indeed, many groups have devoted theirresources to the search for such TdT-interacting proteins. It has been shownin Ku86-deficient mice that the majority of coding joins are devoid of N regionsand that a high proportion of these joins have lost no nucleotides from either endrelative to littermate and C.B-17 SCID controls (Bogue et al., 1997; Puruggananet al., 2001). These observations suggest that Ku proteins interact with TdTS andTdTL1 or TdTL2 (depending on the species) and then recruit them to codingjoins. Subsequently, TdTwas shown to interact with Ku70, Ku86, and the Ku70–Ku86 heterodimer complex (Mahajan et al., 1999). This interaction appears tobe DNA independent. On treatment of cells with a topoisomerase II inhibitoretoposide, TdT and Ku form discrete foci in the nucleus and, with increasingtime, form intracellular complexes dispersed throughout the nucleus.Furthermore, removal of the first 131 amino acids from the TdT N-terminaldomain, which contains the BRCT module, selectively and greatly reduces TdTassociation with Ku70 and the Ku70–Ku86 heterodimer; however, its interac-tion with Ku86 alone remains intact. TdT has also been shown to interactwith DNA-PKcs, another crucial factor required for V(D)J recombination(Mickelsen et al., 1999). These results, although intriguing, fall short in

isoforms of tdt 127

explaining mechanisms and functional consequences of these associations andhave not been confirmed. Through a series of yeast two-hybrid experiments,TdTwas shown to interact with proliferating-cell nuclear antigen (PCNA) and anovel nuclear protein (TdIF1, terminal deoxynucleotidyltransferase-interactingfactor 1) homologous to transcription factor p65, which belongs to the nuclearreceptor superfamily (Ibe et al., 2001; Yamashita et al., 2001). The BRCTdomain does not appear to mediate association with PCNA or TdIF1.Moreover, association with TdIF1 enhances and interaction with PCNA reducesTdT activity. PCNA and TdIF1 do not appear to play any roles in V(D)Jrecombination; therefore, the physiological consequences of these associationson cells are not apparent. However, these data suggest different mechanisms ofTdT regulation and function independent of V(D)J recombination.

9. TdT Splice Variants and Repertoire Development

The most common and direct approach to studying the function of a gene isto create gene-targeted mutant mice and then observe the loss of function and/or change of phenotype. Indeed, TdT-deficient mice were created (Gilfillanet al., 1993; Komori et al., 1993). To date, the TdT deficiency mutation doesnot cause any deleterious diseases or overall phenotypic abnormalities in mice,suggesting that TdT is not required for normal development. However,close examination of the immune system reveals several phenotypic abnorm-alities (Gilfillan et al., 1995b). Studies with TdT-deficient mice incontrovertiblyshow that the majority (�70–80%) of N regions in V(D)J joins are catalyzed byTdTS in vivo. In adult TdT-deficient mice, homology-directed recombinationand canonical joins are enhanced to the levels seen in normal fetal immuno-globulins and TCRs (Fig. 8A and B). CD3hi single-positive selection is moreefficient in the thymuses of TdT-deficient mice, and both CD4þ and CD8þ

lineages are equally affected by the null mutation. The enhanced positiveselection caused by the TdT deficiency mutation is not influenced bythe size of thymuses or the genetic background (i.e., strain and MHC haplo-type) of animals used. Moreover, the increased proportion of CD3hi single-positive cells in TdT-deficient mice appears to result from more thymocytessuccessfully making the transition from the double-positive to the single-positive stage rather than to a more rapid transition between these stages.The more efficient positive selection observed in TdT-deficient mice maybe due to the shortened complementarity-determining region 3 (CDR3) loopof the ab TCR resulting from N-less joining. Although structural evidenceis lacking, this may also account for the cross-reactive neonatal-like T-cellrepertoire in adult TdT-deficient mice, as elegantly demonstrated by Gavinand Bevan (1995), where the TdT-deficient repertoire was also less peptide

Figure 8 Examples of the germline features of mouse T-cell receptor (TCR) and B-cell receptor(BCR) canonical nucleotide sequences. (A) Lymphocytes with the invariant T cell receptor, usingVg3 and Dd2 with the J and D regions, are generated in fetal mouse thymus (in the absence of TdTactivity) and then migrate into the skin, where they remain in the adult as dendritic epidermalT cells (DETCs). (B) A dominant B-1 cell clone (recognized by the idiotype marker T15) expressesthe VHS107/Vk22 HþL chains and the D, JH, and Jk regions shown, is generated in neonatal lifeand dominates the response to bacterial phosphorylcholine in the adult. The V, D, and J assignmentsare shown on top of each rearranged TCR or BCR nucleotide sequence. P nucleotides palindromicto the first nucleotides at the end of germline V, D, and J elements are shown in italics. Shorthomologous regions (underlined) between V, D, and J segments guide the recombination process.There is no TdT-mediated N addition in these joins.

128 to-ha thai and john f. kearney

specific. In addition, the N-less repertoire of B cells is more polyreactive,because TdT-deficient mice exhibit an increase in the phosphorylcholine(PC) response and TEPC-15 idiotypeþ antibody production (Benedict et al.,2001), both of which cannot normally be derived from adult bone marrowprecursors, where constitutive TdT activity prohibits the production of thecanonical germline CDR3 nucleotide sequences. Thus, in the absence of adultTdT activity the fetal/neonatal windows of T and B cell development areextended for the life of the mouse. Although studies are limited, the TdTdeficiency mutation appears to protect against rather than promote autoim-munity, because TdT-deficient mice develop a lower amount of anti-DNA

isoforms of tdt 129

antibodies and rheumatoid factors when immunized with lipopolysaccharide(LPS) (Weller et al., 1997). In addition, when crossed with nonobese diabeticmice or (NZB � NZW)F1 mice, hybrid mice show a reduced incidence ofinsulitis and autoimmune nephritis, respectively (Conde et al., 1998; Gilfillanet al., 1995a). Overall, the immunoglobulin and TCR repertoires of TdT-deficient mice resemble fetal repertoires: homology-directed recombinationis enhanced, canonical joins are overused, CDR3 loops are short and lessdiverse, and response to PC is increased. More significantly, mice havingfetal repertoires throughout life appear to fare normally and robustly againsta variety of immunological insults, including viruses and bacteria. The TdTdeficiency null mutation also affects human immunoglobulin rearrangementby influencing heavy chain variable (VH) and joining (JH) gene segment usage,and high-frequency recombination occurs at sites of short homologies(Tuaillon and Capra, 2000).

It is apparent that TdT is not an essential gene; therefore, to understandthe subtle role that TdT plays in modulating the immune repertoire, the nextobvious question concerns what happens to fetal immunoglobulin and TCRrepertoires if TdT is expressed during ontogeny. Like TdT-deficient mice,TdTS and TdTL1 transgenic mice do not suffer any obvious phenotypic abnorm-alities and thrive in normal housing conditions (Benedict and Kearney, 1999;Benedict et al., 2000; Bentolila et al., 1997; Conde et al., 1998). Moreover, TdTSand TdTL1 mice appear to sustain normal B and T cell development. Asexpected, N regions are present in fetal V(D)J joins and in adult L chains ofTdTS transgenic mice; thus, fetal repertoires resemble those of adults. Theconsequences of this ontogenetically forced N addition are the failure, in adultTdTS transgenic mice, to produce the normally dominant B-cell clones expres-sing the hallmark canonical T15 immunoglobulin receptor and a subsequentfailure to make a T15 anti-PC antibody response on challenges with PC-contain-ing Streptococcus pneumoniae. Administration of sera obtained from adult TdTStransgenic mice immunized with S. pneumoniae failed to protect B-cell–deficient(xid) recipients from challenges with a virulent pneumoccal infection, whereastransfer of sera from normal adult donors immunized with S. pneumoniaeprovided complete protection to xid recipients against a virulent pneumococcalinfection. Expression of TdT in the neonatal spleen reduces reading frame 1(RF1) usage relative to littermate controls exhibiting 80% RF1 usage, andpositive selection of VH81x-encoded H chains during fetal life is almost abol-ished (Marshall et al., 1998). In line with these observations, forced expressionof N nucleotides in L chains that are predominantly N-less decreases VH81xclonal production, thereby negatively affecting positive selection in the spleen,specifically marginal zone B cells (Martin and Kearney, 2000). In contrast,TdTL1 transgenic mice display a normal anti-PC response; however, anti-protein

130 to-ha thai and john f. kearney

antibody responses in these mice have not been determined. However, a reportshows that forced TdT expression in fetal thymus decreases the number of gdT cells and that Vg3Vd1 T cells randomly disseminate in newborn, but not inadult, skin (Aono et al., 2000). Thus, the loss of innocence or the lack of maturity,although not detrimental to immunoglobulin and TCR repertoires, negativelymodulates the immune system.

10. Biochemical Properties and Substrate Specificity of TdT

Although bovine terminal deoxynucleotidyltransferase (bTdT) and hTdTappear to be more related evolutionarily than are mTdT and hTdT, the shortisoform of bovine terminal deoxynucleotidyltransferase (bTdTS) and theshort isoform of mouse terminal deoxynucleotidyltransferase (mTdTS) appearto use Zn2þ preferentially over Mg2þ as divalent cation (Bollum, 1974),whereas hTdTS, purified from leukemic cells, prefers Mg2þ (Deibel andColeman, 1980). The activity of TdTS from all three species is completelyabrogated by EDTA treatment (Bollum, 1974). Because bTdTL1/2 andhTdTL1/2 have not yet been thoroughly characterized, from this point forth,the long isoform of mouse terminal deoxynucleotidyltransferase 1 (mTdTL1)is exclusively described. Unlike mTdTS, mTdTL1 does not require Zn2þ forits Exo activity (Mg2þ appearing to be sufficient), and mTdTL1 is similarlysensitive to EDTA treatment. A pyrophosphorolytic dismutase activity hasbeen ascribed to a calf thymus TdT preparation when Co2þ is used(Anderson et al., 1999). However, we know that all three bTdT isoforms canpotentially be expressed in calf thymus (T.-H. Thai, unpublished observations);thus, it is plausible that the proposed pyrophosphorolytic dismutase activity isactually caused by Exo activity of bTdTL isoforms present in the preparation.bTdTS and mTdTS incorporate purines (A and G) more efficiently thanpyrimidines (C and T) (Basu et al., 1983; Bollum, 1960; Gauss and Lieber,1996). On the other hand, oligodeoxynucleotide primers composed of A orT residues [d(pA)n and d(pT)n, respectively] are better priming substrates forbTdTS and mTdTS than d(pC)n or d(pG)n (Bollum, 1960). mTdTL1 hassimilar substrate requirements. This preferential substrate usage is recapitu-lated using bona fide mouse D elements (DFL16.1 and DQ52), wheremTdTL1 efficiently deletes nucleotides from the 30 end of the forward se-quence, whereas the reverse sequence of the same D element suffers lessnucleotide loss (H. Schroeder, personal communication). The minimum num-ber of residues that a substrate can support priming is three (Bollum, 1974).These data suggest that the overrepresentation of G and C residues inN regions at V(D)J junctions may result from the concomitant activity ofTdTS and TdTL1 (Thai et al., 2002).

isoforms of tdt 131

11. Expression of Human TdT in Human Leukemias

Since the discovery of TdT in 1960, a large body of studies has been publisheddocumenting TdT overexpression in B- and T-cell acute lymphocytic leuke-mias (ALLs) and in acute myelocytic leukemias (AMLs) (Bertazzoni et al.,1982; Greaves et al., 1980; Hoffbrand et al., 1977; Kung et al., 1978; Oiwa et al.,1989). Historically, TdT has been used as a marker in the diagnosis of groups ofhuman leukemias. Its expression in tumors, which often bear markers ofmultiple lineages, indicative of B, T, and myeloid origin, suggests that at leastsome of them are derived from an early multipotential progenitor. Acutelymphocytic leukemia encompasses a group of cancers that represent imma-ture B or T lymphoblasts. Classification of ALL is based on the degree ofdifferentiation of the cells isolated. Approximately 90% of patients with ALLexpress TdT in their blast cells (Hoffbrand et al., 1977; Janossy et al., 1980).The level of TdT expression varies widely, up to 1000-fold, in each cell andbetween each individual as detected by the standard tests now in use. It is notknown whether all hTdT isoforms can be detected in these assays (Huttonet al., 1982), nor is the functional significance of the high TdT expression inthese tumors known. Our preliminary studies show that multiple patterns ofisoforms are expressed in each tumor (R. Schulz, T.-H. Thai, and J. F. Kearney,unpublished observations).

Approximately 18% of (AMLs) express TdT. The highly variable reportedexpression of TdTþ AML cells (0 to >50%) may be due to detection methodsand also may be due to the specific FAB (French–American–British) systemused for classification (Drexler et al., 1993). TdT positivity appears to be higherin ‘‘pure’’ monocytic lines (M0–M1) compared with the monocytic subclasses(M5–M6) (Skoog et al., 1984). Thus, it appears that TdT expression decreasesas the AML cells become more phenotypically monocytic (Paietta et al., 1993).Finally, approximately 30% of patients with blastic transformation in chronicmyelocytic leukemia (CML) exhibit increased TdT activity (Hoffbrand et al.,1977; Hurwitz et al., 1995; Kung et al., 1978). Blast crisis in these patientsportends death, usually within 6 months.

Prognosis and long-term survival studies show that remission rates forpatients with TdT� AML were higher (61%) compared with TdTþ AML(36%). Median survival rates were also higher in those patients who hadTdT-negative AML cells (Del Poeta et al., 1997). Because the signs andsymptoms for ALL and AML are similar, accurate quantification of TdTþ

cells is important for differentiating between ALL and AML, and more precisediagnoses will dictate the appropriate therapy.

Although it is evident that TdT expression has been useful in the detectionand classification of human lymphomas and leukemias, the use of TdT as

132 to-ha thai and john f. kearney

a prognostic marker has met with mixed results (Shurin and Scillian, 1983).ALL remission rates in children exceed 90% with current chemotherapy regi-mens regardless of TdT expression. This is in contrast to AML, which repre-sents less than 20% of all childhood acute leukemias, but represents morethan 30% of childhood deaths from leukemia (Hurwitz et al., 1995). If differ-ences in expression of the isoforms of huTdT can be ascertained more precise-ly, prognostic outcomes can be reevaluated to determine the role of TdT intherapeutic outcomes and survival. Furthermore, the potential role of TdTin transformations can be studied. Transformations herald a poor prognosisas observed in ALL, where the Philadelphia chromosome t(9;22) is seen in only3% of childhood ALL but represents 25% of adult cases, possibly explainingwhy adults fare worse with this disease (Cline, 1994). Further studies on TdTisoforms in these leukemias are now warranted.

12. Possible Aberrant Activity of Human TdT in Leukemias

Although there is extensive literature on the expression patterns of hTdTin leukemia, there are few studies that examine the functional activitiesthat TdT may exert on the development, maintenance, or exacerbation oflymphoid neoplasia. There is abundant evidence for inappropriate rearran-gements of TCRs in B-cell ALL (Dombret et al., 1992), and of both TCRand BCR genes in AML (Foa et al., 1987), although at the time of analysis thesepatterns did not always coincide with TdT expression. There are a fewtantalizing clues that abnormal TdT polymerase activity may be responsiblefor extensive junctional N addition in ALL (Foa et al., 1987; Genevee et al.,1994; Kung et al., 1978; Langlands et al., 1993). In a more limited example of T-ALL, long CDR3 regions of TCRb rearrangements were also found (Yamanakaet al., 1997). Similar findings in AML have been published, in which BCR andTCR rearrangements with long N insertions can be detected together withdeletions of V and J coding regions and multiple Dd segments (Przybylski et al.,1994; Schmidt et al., 1995). It is also of interest that the TCRd locus is often thetarget of unusual rearrangements in AML and ALL (Kimura et al., 1996). It isnot known whether early and/or abnormal expression of TdT isoforms inhematopoietic progenitors affects normal lymphopoiesis and receptor-drivenselection processes in the thymus and the bone marrow, thus giving rise todisordered development. It is assumed that TdT expression is merely a reflec-tion of the stage of development at which a given tumor type arises. However,a reasonable alternative hypothesis is that unchecked or abnormal activityof TdT may have more widespread effects, and it is not known what effectsabnormal expression of TdT isoforms might have on the maintenance andexacerbation of lymphoid malignancies. Indeed, it has been shown that

isoforms of tdt 133

TdT-accessible breaks appear to be scattered over the VH segment obtainedfrom Ramos cells transfected with a TdT-encoding plasmid (Sale andNeuberger, 1998). N insertions, catalyzed by TdT, into the DNA breaks resultin frameshift mutations in the VH region, which in turn generates IgM-lossvariants in culture. These results suggest that accessible DNA breaks within thegenome may serve as substrates for TdT isoforms to modify, either byN addition or N deletion, thus generating mutations.

13. Conclusions

The limited diversity of the early fetal repertoire of B and T cells is guidedby genetic mechanisms involved in the production of the lymphocyte recep-tors that are active during fetal development. The reduced expression andactivity of the mTdT isoforms clearly distinguish lymphocyte developmentin the fetus from that in the adult. Further understanding the function ofthat in relation to RAG and other proteins in receptor gene formation is clearlyof significance to our understanding of the generation of receptor diversity.In the future it will be necessary to determine the expression patterns ofTdT isoforms during T- and B-cell development with respect to ontogeny,the cell cycle, and in relation to other key enzymes involved in recombination.This will involve analysis of in vivo effects of TdTS and TdTL1/2 isoforminteractions with each other and with other DNA-modifying factors. Thereare already clues that substrate specificity preferences of V, D, and J targetsexist for the polymerase and exonuclease activities of TdTS and TdTL1/2 iso-forms. Future studies may reveal that these substrate preferences are thecause of GC nucleotide enrichment at CDR3 joins. TdT isoforms mayalso play a role in the secondary rearrangements involved in k locus inactiva-tion and in receptor revision of autoreactive T and B cells. Further under-standing of the normal structure–function relationship of huTdT isoforms willassist in determining their role in B and T cell development, and theirinvolvement in human lymphoid malignancies. The new information obtainedand the development of new isoform-specific reagents will provide uniquetools to reevaluate these human tumors for TdT expression and assist in thediagnosis and classification of leukemias and may aid in the generation ofbetter antileukemic drugs.

Acknowledgments

We thank Dr. Zeev Pancer for help in generating the phylogenetic tree, Tamer Mahmoud forreading of the manuscript, and Ms. Ann Brookshire for secretarial assistance. This work wassupported in part by NIH grants AI14782, AI14594, AI51533, T32AI07051, and CA13148.

134 to-ha thai and john f. kearney

References

Anderson, R. S., Bollum, F. J., and Beattie, K. L. (1999). Nucleic Acids Res. 27, 3190–3196.Aono, A., Enomoto, H., Yoshida, N., Yoshizaki, K., Kishimoto, T., and Komori, T. (2000). Immu-

nology 99, 489–497.Aoufouchi, S., Flatter, E., Dahan, A., Faili, A., Bertocci, B., Storck, S., Delbos, F., Cocea, L.,

Gupta, N., Weill, J. C., and Reynaud, C. A. (2000). Nucleic Acids Res. 28, 3684–3693.Bassing, C. H., Swat, W., and Alt, F. W. (2002). Cell 109(Suppl.), S45–S55.Basu, M., Hegde, M. V., and Modak, M. J. (1983). Biochem. Biophys. Res. Commun. 111,

1105–1112.Benedict, C. L., and Kearney, J. F. (1999). Immunity. 10, 607–617.Benedict, C. L., Gilfillan, S., Thai, T. H., and Kearney, J. F. (2000). Immunol. Rev. 175, 150–157.Benedict, C. L., Gilfillan, S., and Kearney, J. F. (2001). J. Exp. Med. 193, 89–99.Bentolila, L. A., Fanton d’Andon, M., Nguyen, Q. T., Martinez, O., Rougeon, F., and Doyen, N.

(1995). EMBO J. 14, 4221–4229.Bentolila, L. A., Wu, G. E., Nourrit, F., Fanton d’Andon, M., Rougeon, F., and Doyen, N. (1997).

J. Immunol. 158, 715–723.Bertazzoni, U., Brusamolino, E., Isernia, P., Scovassi, A. I., Torsello, S., Lazzarino, M., and

Bernasconi, C. (1982). Blood 60, 685–692.Bogue, M. A., Wang, C., Zhu, C., and Roth, D. B. (1997). Immunity 7, 37–47.Bollum, F. J. (1960). J. Biol. Chem. 235, PC18–20.Bollum, F. J. (1974). In ‘‘The Enzymes’’ (P. D. Boyer, Ed.), p. 145. Academic Press, New York.Breit, T. M., and Van Dongen, J. J. (1994). Thymus 22, 177–199.Callebaut, I., and Mornon, J. P. (1997). FEBS Lett. 400, 25–30.Chang, L. M., and Bollum, F. J. (1982). J. Biol. Chem. 257, 9588–9592.Cline, M. J. (1994). N. Engl. J. Med. 330, 328–336.Conde, C., Weller, S., Gilfillan, S., Marcellin, L., Martin, T., and Pasquali, J. L. (1998). J. Immunol.

161, 7023–7030.Dave, V. P., Larche, M., Rencher, S. D., Koop, B. F., and Hurwitz, J. L. (1993). Hum. Immunol. 37,

178–184.Deibel, M. R., Jr., and Coleman, M. S. (1980). J. Biol. Chem. 255, 4206–4212.Del Poeta, G., Venditti, A., Aronica, G., Stasi, R., Cox, M. C., Buccisano, F., Bruno, A., Tamburini, A.,

Suppo, G., Simone, M. D., Epiceno, A. M., Del Moro, B., Masi, M., Papa, G., and Amadori, S.(1997). Leuk. Lymphoma 27, 257–274.

Dombret, H., Loiseau, P., Bories, J. C., and Sigaux, F. (1992). Blood 80, 2614–2621.Dominguez, O., Ruiz, J. F., Lain de Lera, T., Garcia-Diaz, M., Gonzalez, M. A., Kirchhoff, T.,

Martinez, A. C., Bernad, A., and Blanco, L. (2000). EMBO J. 19, 1731–1742.Doyen, N., d’Andon, M. F., Bentolila, L. A., Nguyen, Q. T., and Rougeon, F. (1993). Nucleic Acids

Res. 21, 1187–1191.Drexler, H. G., Sperling, C., and Ludwig, W. D. (1993). Leukemia 7, 1142–1150.Ernst, P., Hahm, K., Trinh, L., Davis, J. N., Roussel, M. F., Turck, C. W., and Smale, S. T. (1996).

Mol. Cell. Biol. 16, 6121–6131.Ernst, P., Hahm, K., Cobb, B. S., Brown, K. E., Trinh, L. A., McCarty, A. S., Merkenschlager,

M., Klug, C. A., Fisher, A. G., and Smale, S. T. (1999). Cold Spring Harb. Symp. Quant.Biol. 64, 87–97.

Farrar, Y. J., Evans, R. K., Beach, C. M., and Coleman, M. S. (1991). Biochemistry 30, 3075–3082.Foa, R., Casorati, G., Giubellino, M. C., Basso, G., Schiro, R., Pizzolo, G., Lauria, F., Lefranc,

M. P., Rabbitts, T. H., and Migone, N. (1987). J. Exp. Med. 165, 879–890.

isoforms of tdt 135

Fugmann, S. D., Lee, A. I., Shockett, P. E., Villey, I. J., and Schatz, D. G. (2000). Annu. Rev.Immunol. 18, 495–527.

Gauss, G. H., and Lieber, M. R. (1996). Mol. Cell. Biol. 16, 258–269.Gavin, M. A., and Bevan, M. J. (1995). Immunity 3, 793–800.Genevee, C., Chung, V., Diu, A., Hercend, T., and Triebel, F. (1994). Mol. Immunol. 31, 109–115.George, J. F., Jr., and Schroeder, H. W., Jr. (1992). J. Immunol. 148, 1230–1239.Ghia, P., ten Boekel, E., Sanz, E., de la Hera, A., Rolink, A., and Melchers, F. (1996). J. Exp. Med.

184, 2217–2229.Gilfillan, S., Dierich, A., Lemeur, M., Benoist, C., and Mathis, D. (1993). Science 261, 1175–1178.Gilfillan, S., Bachmann, M., Trembleau, S., Adorini, L., Kalinke, U., Zinkernagel, R., Benoist, C.,

and Mathis, D. (1995a). Eur. J. Immunol. 25, 3115–3122.Gilfillan, S., Benoist, C., and Mathis, D. (1995b). Immunol. Rev. 148, 201–219.Girschick, H. J., Grammer, A. C., Nanki, T., Mayo, M., and Lipsky, P. E. (2001). J. Immunol. 166,

377–386.Grawunder, U., and Harfst, E. (2001). Curr. Opin. Immunol. 13, 186–194.Greaves, M., Paxton, A., Janossy, G., Pain, C., Johnson, S., and Lister, T. A. (1980). Leuk. Res. 4,

1–14.Hahm, K., Ernst, P., Lo, K., Kim, G. S., Turck, C., and Smale, S. T. (1994). Mol. Cell. Biol. 14,

7111–7123.Hawke, N. A., Rast, J. P., and Litman, G. W. (1996). J. Immunol. 156, 2458–2464.Haynes, B. F., and Heinly, C. S. (1995). J. Exp. Med. 181, 1445–1458.Hoffbrand, A. V., Ganeshaguru, K., Janossy, G., Greaves, M. F., Catovsky, D., and Woodruff, R. K.

(1977). Lancet 2, 520–523.Hozumi, N., and Tonegawa, S. (1976). Proc. Natl. Acad. Sci. USA. 73, 3628–3632.Hurwitz, C. A., Mounce, K. G., and Grier, H. E. (1995). J. Pediatr. Hematol. Oncol. 17, 185–197.Hutton, J. J., Coleman, M. S., and Moffitt, S. (1982). Adv. Exp. Med. Biol. 145, 243–257.Huyton, T., Bates, P. A., Zhang, X., Sternberg, M. J., and Freemont, P. S. (2000). Mutat. Res. 460,

319–332.Ibe, S., Fujita, K., Toyomoto, T., Shimazaki, N., Kaneko, R., Tanabe, A., Takebe, I., Kuroda, S.,

Kobayashi, T., Toji, S., Tamai, K., Yamamoto, H., and Koiwai, O. (2001). Genes Cells 6, 815–824.Janossy, G., Hoffbrand, A. V., Greaves, M. F., Ganeshaguru, K., Pain, C., Bradstock, K. F., Prentice,

H. G., Kay, H. E., and Lister, T. A. (1980). Br. J. Haematol. 44, 221–234.Kerfourn, F., Charlemagne, J., and Fellah, J. S. (1996). Immunogenetics 44, 275–285.Kimura, N., Akiyoshi, T., Uchida, T., Okamura, J., Kozuru, M., Nishimura, J., Ohyashiki, J. H.,

Kikuchi, M., Niho, Y., and Okumura, M. (1996). Leukemia 10, 650–657.Koiwai, O., Yokota, T., Kageyama, T., Hirose, T., Yoshida, S., and Arai, K. (1986). Nucleic Acids Res.

14, 5777–5792.Komori, T., Okada, A., Stewart, V., and Alt, F. W. (1993). Science 261, 1171–1175.Kung, P. C., Long, J. C., McCaffrey, R. P., Ratliff, R. L., Harrison, T. A., and Baltimore, D. (1978).

Am. J. Med. 64, 788–794.Langlands, K., Eden, O. B., Micallef-Eynaud, P., Parker, A. C., and Anthony, R. S. (1993).

Br. J. Haematol. 84, 648–655.Lewis, S. M. (1994). Adv. Immunol. 56, 27–150.Li, Z., Dordai, D. I., Lee, J., and Desiderio, S. (1996). Immunity 5, 575–589.Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M. R. (2002). Cell 108, 781–794.Mahajan, K. N., and Mitchell, B. S. (2003). Proc. Natl. Acad. Sci. USA 100, 10746–10751.Mahajan, K. N., Gangi-Peterson, L., Sorscher, D. H., Wang, J., Gathy, K. N., Mahajan, N. P.,

Reeves, W. H., and Mitchell, B. S. (1999). Proc. Natl. Acad. Sci. USA 96, 13926–13931.

136 to-ha thai and john f. kearney

Marshall, A. J., Doyen, N., Bentolila, L. A., Paige, C. J., and Wu, G. E. (1998). J. Immunol. 161,6657–6663.

Martin, F., and Kearney, J. F. (2000). Immunity 12, 39–49.Melchers, F., Rolink, A., Grawunder, U., Winkler, T. H., Karasuyama, H., Ghia, P., and Andersson, J.

(1995). Curr. Opin. Immunol. 7, 214–227.Mickelsen, S., Snyder, C., Trujillo, K., Bogue, M., Roth, D. B., and Meek, K. (1999). J. Immunol.

163, 834–843.Nourrit, F., Coquilleau, I., D’Andon, M. F., Rougeon, F., and Doyen, N. (1999). J. Mol. Biol. 292,

217–227.Oettinger, M. A. (1999). Curr. Opin. Cell Biol. 11, 325–329.Oiwa, K., Koiwai, O., and Kaneda, T. (1989). Jpn. J. Cancer Res. 80, 331–334.Paietta, E., Racevskis, J., Bennett, J. M., and Wiernik, P. H. (1993). Br. J. Haematol. 84, 416–422.Partula, S., de Guerra, A., Fellah, J. S., and Charlemagne, J. (1996). J. Immunol. 157, 207–212.Pelletier, H., Sawaya, M. R., Kumar, A., Wilson, S. H., and Kraut, J. (1994). Science 264,

1891–1903.Przybylski, G., Oettle, H., Ludwig, W. D., Siegert, W., and Schmidt, C. A. (1994). Br. J. Haematol.

87, 301–307.Purugganan, M. M., Shah, S., Kearney, J. F., and Roth, D. B. (2001). Nucleic Acids Res. 29,

1638–1646.Quiros Roldan, E., Sottini, A., Bettinardi, A., Albertini, A., Imberti, L., and Primi, D. (1995).

Immunogenetics 41, 91–100.Sale, J. E., and Neuberger, M. S. (1998). Immunity 9, 859–869.Sawaya, M. R., Pelletier, H., Kumar, A., Wilson, S. H., and Kraut, J. (1994). Science 264,

1930–1935.Schatz, D. G. (1997). Semin. Immunol. 9, 149–159.Schmidt, C. A., Przybylski, G., Seeger, K., and Siegert, W. (1995). Leuk. Lymphoma 20, 45–49.Shurin, S. B., and Scillian, J. J. (1983). Blood 62, 81–84.Skoog, L., Ost, A., Biberfeld, P., Christensson, B., Hast, R., Lagerlof, B., Nordenskjold, B., and

Reizenstein, P. (1984). Br. J. Cancer 50, 443–449.Sugo, N., Aratani, Y., Nagashima, Y., Kubota, Y., and Koyama, H. (2000). EMBO J. 19, 1397–1404.Takahara, K., Hayashi, N., Fujita-Sagawa, K., Morishita, T., Hashimoto, Y., and Noda, A. (1994).

Biosci. Biotechnol. Biochem. 58, 786–787.Thai, T.-H., and Kearney, J. F. (2004). J. Immunol. 173, 4009–4019.Thai, T.-H., Purugganan, M. M., Roth, D. B., and Kearney, J. F. (2002). Nat. Immunol. 3, 457–462.Tjoelker, L. W., Carlson, L. M., Lee, K., Lahti, J., McCormack, W. T., Leiden, J. M., Chen, C. L.,

Cooper, M. D., and Thompson, C. B. (1990). Proc. Natl. Acad. Sci. USA 87, 7856–7860.Trinh, L. A., Ferrini, R., Cobb, B. S., Weinmann, A. S., Hahm, K., Ernst, P., Garraway, I. P.,

Merkenschlager, M., and Smale, S. T. (2001). Genes Dev. 15, 1817–1832.Tuaillon, N., and Capra, J. D. (2000). J. Immunol. 164, 6387–6397.Turchin, A., and Hsu, E. (1996). J. Immunol. 156, 3797–3805.Weller, S., Conde, C., Knapp, A. M., Levallois, H., Gilfillan, S., Pasquali, J. L., and Martin, T.

(1997). J. Immunol. 159, 3890–3898.Yamanaka, K., Kubo, K., Kiyoi, H., Ichihashi, T., Kato, K., Horibe, K., and Naoe, T. (1997). Leuk.

Res. 21, 667–674.Yamashita, N., Shimazaki, N., Ibe, S., Kaneko, R., Tanabe, A., Toyomoto, T., Fujita, K., Hasegawa, T.,

Toji, S., Tamai, K., Yamamoto, H., and Koiwai, O. (2001). Genes Cells 6, 641–652.Yang, B., Gathy, K. N., and Coleman, M. S. (1994). J. Biol. Chem. 269, 11859–11868.Yoshikai, Y., Kimura, N., Toyonaga, B., and Mak, T. W. (1986). J. Exp. Med. 164, 90–103.

Innate Autoimmunity

Michael C. Carroll* and V. Michael Holers {

*CBR Institute for Biomedical Research, and Department of PediatricsHarvard Medical School, Boston, Massachusetts 02115

{Departments of Medicine and Immunology, University of ColoradoHealth Sciences Center, Denver, Colorado 80217

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1371. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1382. Ischemia–Reperfusion Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1383. Reperfusion Injury Mediation by Natural Antibody . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1394. Specificity of Natural IgM-Mediating Reperfusion Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1405. Initiation of Reperfusion Injury by a Single IgM.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1436. Limition of Ischemia-Related Antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1457. Renal Ischemia–Reperfusion.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1478. Fetal Loss Syndromes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

Abstract

The adaptive immune system has evolved highly specific pattern recognitionproteins and receptors that, when triggered, provide a first line of host defenseagainst pathogens. Studies reveal that these innate recognition proteins are alsoself-reactive and can initiate inflammation against self-tissues in a similarmanner as with pathogens. This specific event is referred to as ‘‘innate autoim-munity.’’ In this review, we describe two classes of autoimmune responses, thatis, reperfusion injury and fetal loss syndrome, in which the recognition andinjury are mediated by innate immunity. Both disorders are common and areclinically important. Reperfusion injury (RI) represents an acute inflammatoryresponse after a reversible ischemic event and subsequent restoration of bloodflow. Findings that injury is IgM and complement dependent and that a singlenatural antibody prepared from a panel of B-1 cell hybridomas can restoreinjury in antibody-deficient mice suggest that RI is an autoimmune-typedisorder. Fetal loss syndrome is also an antibody- and complement-dependentdisorder. Although both immune and natural antibodies are likely involved inrecognition of phospholipid self-antigens, inhibition of the complement path-way in rodent models can block fetal loss. As new innate recognition proteinsand receptors are identified, it is likely that innate responses to self representfrequent events and possibly underlie many of the known chronic autoimmunedisorders normally attributable to dysregulation of adaptive immunity.

137advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

138 michael c. carroll and v. michael holers

1. Introduction

The innate immune system is thought of as the host’s first line of defenseagainst microbial infections (Baumgarth et al., 2000; Ochsenbein et al., 1999).Like adaptive immunity, it includes recognition proteins such as collectins(Epstein et al., 1996), serum complement (Reid and Porter, 1981), and naturalantibody (Boes et al., 1998). The latter is produced primarily by a specializedsubset of B cells termed B-1 and is included as a component of innateimmunity (discussed in more detail below) (Hardy et al., 1994). Recognitionof pathogens leads to a cascade of events resulting in initiation of inflammationand infiltration of inflammatory cells. The identification of toll-like receptors(TLR) has greatly expanded our view of the host innate recognition of patho-gens (Janeway and Medzhitov, 1999), and the list of pattern recognitionreceptors (PRRs) and of pathogen-associated molecular patterns (PAMPs),both associated with TLRs (Barton et al., 2003) and other pathways (Aderemand Underhill, 1999), continues to grow. Activation of TLRs results in releaseof cytokines that also induce infiltration and activation of proinflammatory cells(Barton and Medzhitov, 2003). Thus, it is apparent that the innate immunesystem has evolved highly specific recognition molecules that provide a rapidresponse to infectious agents.

Given the highly conserved nature of many of the known PAMPs, it isinevitable that cross-reactivity with host antigens such as heat shock protein70 (HSP-70; Vabulas et al., 2002) occurs, leading to an inflammatory responseagainst self. This type of response is termed innate autoimmunity, as the initialevent is based on innate recognition of self. In this review, we discuss twocommon models of autoimmune response to self-antigens, that is, reperfusioninjury and fetal loss syndrome. Both are examples of innate autoimmunity andrepresent clinically relevant disease in humans. Because the review focuses onearly events in initiation of autoimmune injury, downstream events such asinfiltration of leukocytes, mast cell activation, or injury due to the terminalcomponents of complement are not discussed in depth.

2. Ischemia–Reperfusion Injury

Ischemia–reperfusion injury (RI) represents an acute inflammatory responsefollowing a reversible ischemic event and subsequent restoration of blood flow(Cotran et al., 1994). It is potentially life threatening and is primarily responsi-ble for the tissue injury following reperfusion that occurs in myocardialinfarction, cerebral ischemic events, intestinal ischemia, renal ischemia,and other events such as vascular surgery, trauma, and transplantation. Theresponse of cells to hypoxia is pleiotropic and includes alterations of cytoplasmic

innate autoimmunity 139

architecture, activation of enzymatic pathways, and initiation of gene transcrip-tion (Tatsumi et al., 2003). Mitochondria are particularly sensitive to changesin oxygen concentration (Liu et al., 2003). However, little is known regardingevents outside of the cell that lead to the observed acute inflammatory re-sponse. One proposal is that that surface alterations following hypoxia arerecognized by neutrophils, and this leads to release of reactive oxygen species(ROS) that promote tissue injury (Li and Jackson, 2002). Although neutrophilsare a hallmark of acute inflammation and are observed in RI, their role ininitiation of injury is not clear. In some examples of RI, injury occurs in thepresence of a limited numbers of neutrophils (Briaud et al., 2001; Simpsonet al., 1993).

The first suggestion that the complement system might mediate RI camefrom elegant studies by Fearon and colleagues demonstrating protection ina rat model of coronary ischemia (Weisman et al., 1990). They found thatpretreatment of animals with a soluble inhibitor of activated C3, that is, solubleCR1 (sCR1), dramatically reduced inflammatory injury to the myocardium.Thus, shutting off the complement cascade at the central step of C3 activationwas protective. Subsequent studies in other animal models of RI, includingmodels of cardiopulmonary bypass (CPB) (Chai et al., 2000) and stroke in thepig (Huang et al., 1999) and the mouse skeletal muscle model (Hill et al.,1992), confirmed and extended their observations that blocking of the com-plement system with sCR1 significantly reduced injury. Although these studiessuggested that the complement system was important in RI in various tissues,they did not provide an explanation for how complement activation wasinitiated. That is because each of the three complement activation pathways(classical, alternative, and lectin) converge on C3 and require C3 cleavage tothen generate C5a and the membrane attack complex (MAC).

3. Reperfusion Injury Mediation by Natural Antibody

The early events in RI were first examined using mice bearing specific defi-ciencies in complement or innate immunity. Using a model of skeletal muscleRI, Weiser et al. (1996) demonstrated that mice totally deficient in C3 wereprotected to a similar level as mice pretreated with sCR1. Importantly, micedeficient in C4 were also protected in this model. These results not onlyconfirmed the importance of the complement system in RI but suggestedthat initiation of the cascade was via the classical or lectin pathways. To testa role for antibody, RAG-1–/– mice, which are deficient in mature lymphocytesand thus do not express immunoglobulin, were treated in the RI model.Notably, the RAG-1–/– mice were protected from injury to a similar extent ascomplement-deficient animals. Importantly, reconstitution with normal mouse

Figure 1 Proposed model of reperfusion injury mediated by natural IgM after recognition ofneoepitopes exposed or expressed during ischemia. IgM binding activates the classical pathway ofcomplement, leading to release of proinflammatory peptides C3a and C5a and deposition of themembrane attack complex C5–C9, leading to direct cell lysis.

140 michael c. carroll and v. michael holers

serum restored injury, suggesting that the defect lies in the absence of serumimmunoglobulin. These observations were interpreted to suggest that preex-isting antibody recognized neoepitopes on hypoxic endothelium and initiatedthe classical pathway of complement (Fig. 1). The identification of IgMdeposition on vessels within the reperfused hind limb muscle supported thishypothesis (Weiser et al., 1996). Williams et al. (1999) then extended theseobservations in an intestinal model of RI demonstrating that reconstitution ofRAG-1�/� mice with purified serum IgM restored injury (Fig. 2). These resultsnot only confirmed the importance of serum IgM but suggested a commonmechanism for RI in at least two tissues, that is, skeletal muscle and intestine.

4. Specificity of Natural IgM-Mediating Reperfusion Injury

The first indication of the specificity of natural IgM in RI derived from experi-ments with mice deficient in complement receptors CD21 and CD35 (Cr2�/�).Intriguingly, in independent experiments two groups reported that theCr2�/� mice were protected from injury in an intestinal model of RI(Fleming et al., 2002; Reid et al., 2002). In both reports, reconstitution of thedeficient animals with pooled IgM—prepared from wild-type mice—restoredpathogenic injury to the small bowel as evaluated by histology, myeloperox-idase (MPO) generation, and vascular permeability analysis. By contrast,reconstitution of the deficient animals with IgM prepared from the

Figure 2 Mice deficient in IgM are protected from intestinal reperfusion injury (RI). Reconstitu-tion of mice deficient in recombinase-activating gene (RAG-1null) with pooled IgM from wild-typemice restores injury. RI is based on the permeability index (see Williams et al., 1999).

innate autoimmunity 141

Cr2-deficient mice did not restore injury (Reid et al., 2002). In one report, arole for IgG was suggested on the basis of an increase in local generation ofMPO when the deficient mice were reconstituted with wild-type IgG (Fleminget al., 2002). Because Cr2�/� mice express normal levels of serum IgM, yetwere not injured, it was proposed that ischemia-specific IgM was missing fromthe repertoire. It was concluded that the complement system has an additionalrole in RI and is important in the development or maintenance of the subset ofB cells that secrete the IgM involved in RI. Cr2�/� mice also have an impairedresponse to T-dependent and -independent antigens, as the receptors arerequired for an effective humoral response (for review see Carroll, 1998;Fearon and Cartes, 1995; Hannan et al., 2002).

To identify the population of B cells responsible for secretion of pathogenicIgM, Reid et al. (2002) engrafted Cr2�/� mice with an enriched fraction of B-1cells prepared from Cr2þ/þ wild-type mice. B-1 cells are a major source ofnatural antibody and differ from conventional B cells by several criteria such ascell surface markers, ability to self-replenish, and a limited repertoire that isbiased toward microbial and highly conserved self-antigens (discussed furtherbelow). Characterization of the chimeric mice confirmed the presence ofnormal levels of Cr2þ/þ B-1 cells within the peritoneum within 4–6 weeks ofengraftment (Fig. 3). More importantly, the engrafted mice developed full

Figure 3 Engraftment of Cr2-deficient mice with wild-type peritoneal B-1 cells restores injury.(a) Analysis of peritoneal B-1 cells by FACS after staining with antibody specific for complementreceptors indicates reduction in cell surface level of CD21 and CD35 in Cr2null mice, as expected.By contrast, analysis of peritoneal B cells isolated from Cr2null mice engrafted with wild-type B-1cells 6 weeks previously confirms the presence of CD21þ CD35þ B cells. (b) Cr2null mice areprotected from intestinal RI, but engraftment with wild-type B-1 cells restores injury (see Reidet al., 2002, for details).

142 michael c. carroll and v. michael holers

innate autoimmunity 143

injury in the intestinal RI model. Thus, these results not only confirmed theimportance of IgM in RI but supported a role for B-1 cells as a potential sourceof pathogenic antibody.

5. Initiation of Reperfusion Injury by a Single IgM

The observations taken from the Cr2�/� experiments strongly suggested thatnatural IgM produced by B-1 cells was at least one source—if not the majorsource—of pathogenic antibody in RI. To test further the hypothesis that theantibody was specific and might directly recognize antigens on ischemicendothelium/tissues, a panel of IgM-expressing hybridomas was preparedfrom an enriched fraction of wild-type peritoneal B-1 cells. Supernatant waspooled from a group of 22 hybridomas that produced relatively high levels ofIgM and injected into RAG-1�/� mice before treatment in the intestinal RImodel (Fig. 4; Zhang et al., 2004). Notably, reconstitution with the totalpool restored injury similar to that of wild-type IgM. Thus, the panel of 22hybridomas included a clone or clones that secreted RI-specific IgM. Tofurther identify the clone or clones, pools were prepared from groups of fiveor fewer hybridomas and tested in the RAG-1�/� mice. Using this in vivoapproach, one clone, CM-22, was identified that could restore injury in theimmunoglobulin-deficient mice in the intestinal RI model. Importantly, theother 21 monoclonal antibodies (mAbs) did not restore injury.

Nucleotide sequence analysis of the immunoglobulin heavy and light chaincDNAs expressed by the pathogenic clone CM-22 identified the rearrangedV(D)J as a VH Vm 3.2, DFL 16, and JH1, respectively. The VH–DH and DH–JH junctions bore limited N region addition, and the rearranged VH repre-sented a germline-like sequence with no evidence of somatic mutation. Theimmunoglobul in light chain (Lc) was identified as a member of the Vk 21–12(Jk2) subfamily and is 99% identical to an Lc previously identified as a naturalantibody. Thus, the structure of the immunoglobulin heavy chain (Hc) and Lcsupport the origin of the IgM as a B-1 cell.

Evidence that CM-22 IgM was specific for ischemic intestine came fromhistologic examination of tissues isolated from RAG-1�/� mice treated withIgM isolated from either CM-22 or CM-31 (one of the other mAbs from thepanel). RAG-1�/� mice reconstituted with control IgM failed to develop injuryand appeared similar to the saline controls. By contrast, CM-22-reconstitutedmice developed significant injury to microvilli within the ischemic and reper-fused jejunum (Fig. 5). Pathologic injury correlated with deposition of theCM-22 antibody, which colocalized with complement C4 and C3 within theinjured microvillus (Fig. 6).

Figure 4 Identification of pathogenic IgM that mediates intestinal RI. (A) In vivo screeningapproach used to identify B-1 cell hybridomas that secrete pathogenic IgM. (B) Reconstitution ofRAG-1–/– mice with IgM prepared from clone CM-22 restores injury in the intestinal RI model(see Zhang et al., 2004, for details).

144 michael c. carroll and v. michael holers

The identification of a single mAb that initiates RI in the intestinal modelprovides direct support for the hypothesis that recognition of ischemic tissue isspecific and raises the question of the number of antigens involved andwhether the same or related antigens are also involved in RI among othertissues. Similar experiments were performed in a hind limb model of skeletalmuscle ischemia in which RAG-1�/� mice were reconstituted with IgMisolated from either CM-22 or a control mAb. Notably, mice reconstitutedwith CM-22 IgM developed RI in contrast to those receiving control IgM(W. Austen, M. Zhang, R. Chan, H. Hechtman, M. C. Carroll, and F. D. Moore,

Figure 5 B-1 cell hybridoma clone CM-22 mediates RI in RAG-1–/– mice in the intestinal model.(A) Hematoxylin and eosin–stained cryosections prepared from intestinal tissue of RAG-1–/– micereconstituted with (i) saline, (ii) wild-type IgM, (iii) clone CM-31, or (iv) clone CM-22, 1 h beforetreatment in the intestinal RI model. Arrowheads indicate subepithelial spacing and sloughing ofmicrovilli tips in RI. (B) Injury scores of reconstituted RAG-1–/– mice after treatment in theintestinal RI model. Scores are based on pathology (see Zhang et al., 2004, for details).

innate autoimmunity 145

unpublished results). Thus, in at least two distinct tissues, CM-22 IgM appearsto recognize ischemia-related antigens and initiate injury. It will be importantto extend these studies to other tissues such as myocardium and the CNSstroke model.

6. Limition of Ischemia-Related Antigens

An inherent feature of B-1 cells—the major source of natural antibodies—isthat they are limited in diversity relative to conventional B cells. One explana-tion for the limited repertoire is that B-1 cells develop primarily during the latefetal and neonatal stages. This period of B cell development differs from thatof adults in that terminal deoxynucleotidetransferase (TdT), which is responsi-ble for nucleotide addition to the CDR3 region of the rearranged immuno-globulin heavy chain, is not expressed, and VH usage is biased toward proximalgenes (Hardy et al., 1994; Kantor et al., 1997; Malynn et al., 1990; Seidl et al.,1999). During this period, B-1 cells make up a major fraction of the B cellsproduced. Given the limited repertoire and the bias toward highly conserved

Figure 6 Deposition of IgM and complement within microvilli correlates with induction of injury bypathogenic IgM. RAG-1–/– mice were reconstituted with either CM-31 or CM-22 hybridoma IgM andtreated in the RI model as described in Fig. 5. Red, anti-mouse IgM; green, anti-mouse C4 (e, f, h,and i) or C3 (n, o, q, and r); yellow, colocalization of red and green (see Zhang et al., 2004, for details).

146 michael c. carroll and v. michael holers

innate autoimmunity 147

antigens, it is likely that the antigen/antigens recognized by CM-22 IgM arealso highly conserved and limited in number.

Preliminary evidence in support of a limited number of RI antigens comesfrom two sets of experiments. First, CM-22 appears to precipitate a uniqueprotein/proteins of high molecular weight compared with a control IgM. Inthese experiments IgM containing immune complexes (ICs) were preparedfrom lysates of intestinal tissues of RAG-1�/� mice reconstituted with therespective IgM mAbs before treatment in the intestinal RI model (Zhanget al., 2004). The second line of evidence comes from studies using a peptideinhibitor of CM-22. The peptide was shown to bind specifically to CM-22in vitro and in vivo (M. Zhang, F. D. Moore, and M. C. Carroll, unpublishedresults). Importantly, reconstitution of RAG-1�/� mice with a mixture of CM-22 and the peptide does not restore injury. Moreover, treatment of wild-typemice with the peptide before reperfusion in the intestinal model is alsoprotective. Thus, these two lines of evidence suggest that the number ofantigens involved in RI is limited. It will be important in future experimentsto identify the specific antigen/antigens that are the target for natural IgM andcompare them in various tissues.

In this regard, it will also be relevant to determine whether lack of recognitionof the antigen(s) targeted by CM-22 underlies the lack of RI in Cr2�/� mice.This is important because there are several intestinal antigens with otherphysicochemical properties that are differentially recognized by serum fromCr2þ/þ compared with Cr2�/� mice (L. Kulik and V. M. Holers, unpublishedresults). Thus, whereas results with CM-22 clearly illustrate the key role for thisnatural antibody recognition system in RI in the intestine and skeletal muscle,determination of how CD21 and CD35 themselves influence the evolution ofthe natural antibody repertoire requires additional evaluation.

7. Renal Ischemia–Reperfusion

In addition to playing a major role in the development of natural antibodiesthat recognize antigen(s) and catalyze RI in intestine and skeletal muscle, com-plement as an innate immune mechanism through the alternative pathway canalso function completely independently of natural antibody to identify certainother injured self-tissues. This unique capacity is best exemplified by ischemia–reperfusion injury of the kidney, the pathogenesis of which is reviewed below,in addition to the role of the alternative pathway of complement.

Alternative pathway activation is promoted on surfaces that have neutral orpositive charge characteristics and do not express or bind serum-derivedcomplement inhibitors. This type of activation is directly catalyzed by a processtermed ‘‘tickover’’ of C3, which occurs when C3 spontaneously undergoes a

148 michael c. carroll and v. michael holers

conformation change at a rate of �1%/h. This new conformation of C3, termedC3(H2O), can interact with factor B, allowing factor B to undergo a changethat allows it to then be recognized and cleaved by protease factor D (Muller-Eberhard, 1988). The alternative pathway can, however, also be initiated orgreatly amplified when antibodies block endogenous regulatory mechanisms(Ratnoff et al., 1983), when expression of complement regulatory proteins iseither decreased (Linton and Morgan, 1999; Liszewski et al., 1996; Mizunoet al., 2001; Xu et al., 2000) or is absent such as in the disease paroxysmalnocturnal hemoglobinuria (PNH) (Nishimura et al., 1998), or when dysfunc-tional regulatory proteins are expressed as a result of genetic polymorphisms(Richards et al., 2002, 2003).

RI of the kidney leads to acute renal failure, a condition that may be asso-ciated with a 40–50% mortality rate when it occurs in patients in the hospitalsetting (Levy et al., 1996; Liano et al., 1998; Thadhani et al., 1996). Hemodialysisand other supportive therapies are currently the only treatments for acute renalfailure, although more specific mechanism-based therapies are under activeinvestigation (Sheridan and Bonventre, 2000). Initial studies using animalmodels showed that mice deficient in C3 develop milder renal failure afterischemia–reperfusion, similar to the protective effects of C3 deficiency onintestinal injury. However, in sharp contrast to other organs such as the intes-tine, in the kidney neither C4 nor natural antibody is required to induce injury(Park et al., 2002; Zhou et al., 2000). Indeed, more recent studies have shownthat mice deficient in the alternative pathway complement protein factorB ( f B�/�) are protected from ischemic acute renal failure (Thurman et al.,2003), thus directly implicating this pathway in recognition of ischemic tissue.

The process of RI in the kidney is illustrative of several points relevant tocomplement. First, it is of some interest that there is normally a small amountof C3 that is deposited along Bowman’s capsule of the glomerulus and alongthe basolateral surface of the tubules. This C3 deposition has been suggestedto represent activation of the alternative pathway by ammonia that is producedin the tubules, which can amidate C3 to create a molecule like C3(H2O), whichcan bind factor B and then allow factor B to serve as a substrate for factor D inthe tubules (Nath et al., 1985). Consistent with a role for the alternativepathway in this basal activation process, mice lacking factor B do not exhibitbasal C3 deposition around glomeruli or tubules (Thurman et al., 2003). Afterischemia–reperfusion, C3 is deposited heavily along the basolateral surface ofthe tubules in the outer stripe of the outer medulla at the corticomedullaryjunction, which is also the primary region of histologic injury in the kidney inthis condition. Complement activation must occur via the alternative pathway,as when f B�/� mice are subjected to ischemia–reperfusion, virtually no C3 isseen at this site.

innate autoimmunity 149

Complement activation by the alternative pathway is also causal in thekidney injury and dysfunction that follows ischemia–reperfusion. This is be-cause f B�/� mice demonstrate milder functional impairment and morphologi-cal changes after reperfusion, with serum urea nitrogen (SUN) and histologicinjury scores significantly lower compared with control mice. Of interest, thenumber of neutrophils per high-power field in the tubular region is alsosignificantly lower in f B�/� mice compared with controls, suggesting that thechemotactic effects of C5a and other functionally similar peptides are underalternative pathway control in this setting.

Thus, in the kidney the alternative pathway is capable of specificallyrecognizing ischemic tissue and activating complement, leading to markedlyenhanced tissue injury and neutrophil infiltration in the absence of neoan-tigen recognition by natural antibodies. How this process unfolds specificallyaround tubules at this corticomedullary junction site in the kidney isunder active investigation. One possibility is increased synthesis of alternativepathway components locally that overwhelm inhibitory mechanisms (Fig. 7).Enhanced synthesis of alternative pathway components could come from

Figure 7 Potential mechanisms by which the alternative pathway could be activated around renaltubules during IR injury. (1) Either enhanced levels of complement activators, (2) decreased levelsof complement regulatory proteins, or (3) local production of alternative pathway componentscould promote C3 deposition and tubular injury in an antibody-independent manner.

150 michael c. carroll and v. michael holers

either upregulation in endogenous cells or from newly arriving inflammatorycells at that site. Because the alternative pathway is an enzymatically drivenprocess that is actively inhibited, increasing substrate levels might account forincreased activation. Another possibility is enhanced synthesis of ammonia thatleads to increased amidation of C3, which is capable of interacting with factorB as outlined above. However, our own experiments using NaHCO3 loading( J. M. Thurman and V. M. Holers, unpublished observations), as well as thoseof others using a similar approach to greatly decrease tubular ammonia pro-duction and subsequent amidation of C3 (Sporer et al., 1981), have both failedto protect mice from ischemia–reperfusion injury.

The final possibility, one that we favor, is that alterations of endogenousregulatory proteins occur that lead to diminished inhibition and subsequentalternative pathway activation. In this regard, complement receptor 1–relatedgene/protein y (Crry), a membrane complement regulatory protein that hasbeen extensively characterized (Foley et al., 1993; Kim et al., 1995; Li et al.,1993; Molina et al., 1992), is normally expressed along the basolateral aspect ofrenal tubular cells (Li et al., 1993). Crry is the only membrane complementinhibitor expressed by these cells (Li et al., 1993; Miwa et al., 2001; Qin et al.,2001). Thus, decreased expression of Crry, or alternatively the loss of polarityof Crry expression within the ischemic epithelial cells as the protein movesaway from the basolateral site of initial complement deposition as a conse-quence of metabolic changes during ischemia, could allow alternative pathwayactivation by diminishing local complement inhibitory effects. This issue isunder investigation, using a series of genetic and cell biologic approaches.

One important question with regard to the role of innate immune recogni-tion concerns why the kidney is different from the intestine and other vascu-larized organs. The answer is currently unknown, but there are two likelypossibilities. The first and most straightforward is that the antigen(s) that arerecognized by natural antibodies are not expressed in the kidney. In theabsence of recognition by natural antibody, other innate immune recognitionmechanisms become more apparent. The second is that the vascular supply inthe kidneys to the tubules is complex. Blood first passes through the glomeru-lus and then to the tubules. Vascular tone through autonomic and solublehormonal mechanisms plays an essential role in controlling perfusion of eachregion of the kidney, and thus these effects may alter the ability of innateimmune recognition mechanisms to access the tubules.

Whether additional target organs utilize a similar mechanism to activatecomplement, or whether the ischemic and reperfused kidney is unique in itsability to be directly recognized by this innate immune mechanism, is currentlyunknown. However, it is intriguing that RI in the intestine is also greatlydiminished by experimental blockade of the alternative pathway (Stahl et al.,

innate autoimmunity 151

2003), which might not be expected to happen if the alternative pathwaysimply enhanced complement activation by 2-fold as suggested by serumstudies in factor B–deficient mice (Matsumoto et al., 1997). Thus, it is possiblethat, in a setting in which the classical pathway is engaged by natural antibody,in addition to enhancement of complement activation by the amplification loopthrough C3b, direct recognition of injured tissue by the alternative pathwayand engagement of this pathway using one of the mechanisms discussed abovewith regard to the kidney might also occur.

After activation of the alternative pathway in the kidney, complement-derived mediators can cause renal tubular injury. These factors include C5athrough the C5a receptor (Arumugam et al., 2003; de Vries et al., 2003a,b) aswell as the membrane attack complex (Zhou et al., 2000). Finally, in addition tocomplement, many other proinflammatory pathways are engaged after ische-mia–reperfusion injury (reviewed in Sheridan and Bonventre, 2000). Thus,similarly to intestinal injury, defining what pathways are complement depen-dent and independent should provide important insights into the specific andcausal roles that this system plays relative to other innate immune pathways.

8. Fetal Loss Syndromes

There are several additional conditions in which the alternative pathway as aninnate immune mechanism is likely to be a key component of the recognitionand injury of tissues. One of the most intriguing classes of clinical syndromesin this regard is the fetal loss that occurs in pregnant women as the end resultof many different types of immune recognition mechanisms that target thesemiallogeneic fetus.

The best defined syndrome to date is fetal loss that is caused by anti-phospholipid antibodies. These autoantibodies can develop in patients withsystemic lupus erythematosus as well as independently as the only manifesta-tion of autoimmunity (reviewed in McIntyre et al., 2003). Although the specificpatterns of lipid and lipid:protein recognition vary, what ties these autoanti-bodies together pathophysiologically is their ability to induce thrombosis, orclotting, in vivo. In nonpregnant individuals, the procoagulant effect of theseautoantibodies results in arterial and venous clots with subsequent clinicaleffects such as stroke, deep venous thrombosis, and pulmonary embolism. Inwomen who are pregnant, anti-phospholipid antibodies are additionally asso-ciated with early fetal death as well as growth restriction of surviving fetuses.A predilection for fetal complications appears to be due to the preferentialrecognition of trophoblast cells in the placenta, which are uniquely rich inphosphatidylserine on the external membrane leaflet, by anti-phospholipidautoantibodies.

152 michael c. carroll and v. michael holers

Although IgG anti-phospholipid antibodies are present in patients, anti-phospholipid recognition is common in natural IgM antibodies (McIntyreet al., 2003). In addition, similar to the role of natural antibodies in intestinalischemia–reperfusion injury, single monoclonal anti-phospholipid antibodieswith germline sequences can cause tissue injury fetal loss when transferredinto pregnant mice (Ikematsu et al., 1998). Therefore, anti-phospholipidantibody-mediated injury can also be considered a manifestation of innateimmune recognition.

For several decades the prevailing belief was that alterations of clottingfactors in a prothrombotic direction and/or direct activation of endothelial orinflammatory cells, both mediated solely by the antigen-combining site of anti-phospholipid antibodies, underlay the pathogenesis of this condition andmediated the profound in vivo effects (Espinosa et al., 2003; Rand, 2000).However, the realization that complement activation fragments also demon-strate procoagulant effects in vitro on endothelial cells, platelets, mast cells,macrophages, and polymorphonuclear cells (reviewed in Holers, 2001) led tothe hypothesis that complement activation itself could mediate fetal loss(Holers et al., 2002). Consistent with this novel hypothesis, it was found thatCrry-Ig, a C3 convertase inhibitor active in mice, or subsequently performingexperiments in C3�/� mice, both resulted in complete reversal of the fetalinjury phenotype (Holers et al., 2002). Thus, while anti-phospholipid antibo-dies can demonstrate procoagulant effects in vitro, whether these effects areindeed important to the pathogenesis of disease or whether the antigen-combining site simply targets complement activation to sites of injury remainsan unresolved question.

Relevant to the alternative pathway and innate immunity in this condi-tion, follow-up experiments using additional complement inhibitors andgene-targeted deficient mice revealed that C5a, the classical pathway, thealternative pathway and neutrophils, but not Fc receptors, are key intermedi-aries in anti-phospholipid antibody-mediated loss (Girardi et al., 2003). Inaddition, deposition of C3 in the placenta was correlated with the presenceof neutrophils in the placenta after anti-phospholipid antibody treatment(Girardi et al., 2003). This result strongly suggests that alternative pathwaycomponents carried by these inflammatory cells into the placenta are neces-sary for tissue injury and that the protection afforded by alternative pathwaydeficiency is due to the lack of C3, factor B, and properdin brought in throughthis exogenous source (Girardi et al., 2003). Additional mechanism-basedexperiments are under way to further test this hypothesis, but in support ofthe conclusion that alternative pathway innate immune mechanisms are im-portant in pathogenesis are results demonstrating that a novel inhibitory

innate autoimmunity 153

monoclonal antibody that blocks mouse factor B activity also protects micefrom anti-phospholipid antibody-mediated fetal loss (Thurman et al., 2005).

In addition to this model, studies have shown that the alternative pathway isactivated in the placenta and is pathogenic in causing fetal loss in othercircumstances. One example is the finding that mice with a complete absenceof Crry expression caused by gene targeting are not viable. Specifically, het-erozygous Crry-deficient mice are born in normal numbers and are healthy,but homozygous Crry-deficient mice die in utero (Xu et al., 2000). In thissetting, initial Crry-deficient embryo numbers are normal, but after the depo-sition of C3 through alternative pathway recognition of the placenta and fetus,death ensues and no viable progeny are delivered (Mao et al., 2003).

Finally, whereas results using anti-phospholipid antibodies have been illus-trative of the problem related to those particular antibodies, and Crry-deficientmice illustrate the effects of complete loss of complement regulation, otherstudies have suggested that recurrent cellular immune-mediated fetal loss mayalso be primarily mediated by inappropriate alternative pathway complementactivation (Mellor et al., 2001). Indeed, this area is receiving substantialattention, as it is possible that complement activation is the major pathway offetal loss through multiple immune pathways that are initiated through eitherinnate or adaptive immune recognition mechanisms (Cauchetaux et al., 2003).Thus, as the mechanisms of innate autoimmunity become better understood, itis likely that this pathway of tissue recognition and injury will be increasinglyfound to be central to the pathogenesis of human disease.

References

Aderem, A., and Underhill, D. M. (1999). Mechanism of phagocytosis in macrophages. Annu. Rev.Immunol. 17, 593–623.

Arumugam, T. V., Shiels, I. A., Strachan, A. J., Abbenante, G., Fairlie, D. P., and Taylor, S. M.(2003). A small molecule C5a receptor antagonist protects kidneys from ischemia–reperfusioninjury in rats. Kidney Int. 63, 134–142.

Barton, G. M., and Medzhitov, R. (2003). Toll-like receptor signaling pathways. Science 300,1524–1525.

Baumgarth, N., Herman, O., Jager, G., Brown, L., Herzenberg, L., and Herzenbergh, L. (2000).B1 and B2 cell-derived immunoglobulin M antibodies are non-redundant components of theprotective response to influenza virus infection. J. Exp. Med. 192, 271–280.

Boes, M., Prodeus, A., Schmidt, T., Carroll, M. C., and Chen, J. (1998). A critical role of naturalIgM in immediate response against systemic bacterial infection. J. Exp. Med. 188, 2381–2386.

Briaud, S. A., Ding, Z. M., Michael, L. H., Entman, M. L., Daniel, S., and Ballantyne, C. M.(2001). Leukocyte trafficking and myocardial reperfusion injury in ICAM-1/P-selectin-knockoutmice. Am. J. Physiol. Heart Circ. Physiol. 280, H60–H67.

Carroll, M. C. (1998). The role of complement and complement receptors in induction andregulation of immunity. Annu. Rev. Immunol. 16, 545–568.

154 michael c. carroll and v. michael holers

Cauchetaux, S. M., Kanellopoulos-Langevin, C., and Ojcius, D. M. (2003). At the innate frontiersbetween mother and fetus: Linking abortion with complement activation. Immunity 18,169–172.

Chai, P. J., Nassar, R., Oakeley, A. E., Craig, D. M., Quick, G., Jr., Jaggers, J., Sanders, S. P.,Ungerleider, R. M., and Anderson, P. A. (2000). Soluble complement receptor-1 protects heart,lung, and cardiac myofilament function from cardiopulmonary bypass damage. Circulation 101,541–546.

Cotran, R. S., Kumar, V., and Robbins, S. L. (1994). ‘‘Pathologic Basis of Disease.’’ W. B. Saunders,Philadelphia, PA.

de Vries, B., Kohl, J., Leclerq, W. K. G., Wolfs, T. G. A. M., van Bijnen, A. A. J. H. M., Heeringa, P.,and Buurman, W. A. (2003a). Complement factor C5a mediates renal ischemia–reperfusioninjury independent from neutrophils. J. Immunol. 170, 3883–3889.

de Vries, B., Matthijsen, R. A., Wolfs, T. G. A. M., van Bijnen, A. A. J. H. M., Heeringa, P., andBuurman, W. A. (2003b). Inhibition of complement factor 5 protects against renal ischemia–reperfusion injury: Inhibition of late apoptosis and inflammation. Transplantation 75, 375–382.

Epstein, J., Eichbaum, Q., Sheriff, S., and Ezekowitz, R. (1996). The collectins in innate immunity.Curr. Opin. Immunol. 8, 29–35.

Espinosa, G., Cervera, R., Font, J., and Schoenfeld, Y. (2003). Antiphospholipid syndrome:Pathogenic mechanisms. Autoimmun. Rev. 2, 86–93.

Fearon, D. T., and Carter, R. H. (1995). The CD19/CR2/TAPA-1 Complex of B lymphocytes:Linking natural to acquired immunity. Annu. Rev. Immunol. 13, 127–149.

Fleming, S. D., Shea-Donohue, T., Guthridge, J. M., Kulik, L., Waldschmidt, T. J., Gipson, M. G.,Tsokos, G. C., and Holers, V. M. (2002). Mice deficient in complement receptors 1 and 2 lack atissue injury-inducing subset of the natural antibody repertoire. J. Immunol. 169, 2126–2133.

Foley, S., Li, B., Dehoff, M., Molina, H., and Holers, V. M. (1993). Mouse Crry/p65 is a regulatorof the alternative pathway of complement activation. Eur. J. Immunol. 23, 1381–1384.

Girardi, G., Berman, J., Redecha, P., Spruce, L., Kraus, D. M., Hollman, T. J., Casali, P., Wetsel,R. A., Lambris, J. D., Holers, V. M., and Salmon, J. E. (2003). Complement C5a receptorsand neutrophils mediate fetal injury in the antiphospholipid syndrome. J. Clin. Invest. 112,1644–1654.

Hannan, J., Young, K., Szakonyi, G., Overduin, M. J., Perkins, S. J., Chen, X., and Holers, V. M.(2002). Structure of complement receptor (CR) 2 and CR2–C3d complexes. Biochem. Soc.Trans. 30, 983–989.

Hardy, R. R., Carmack, C. E., Li, Y. S., and Hayakawa, K. (1994). Distinctive developmental originsand specificities of murine CD5þ B cells. Immunol. Rev. 137, 91.

Hill, J., Lindsay, T. F., Ortiz, F., Yeh, C. G., Hechtman, H. B., and Moore, F. D. (1992). Solublecomplement receptor type 1 ameliorates the local and remote organ injury after intestinalischemia–reperfusion in the rat. J. Immunol. 149, 1723–1730.

Holers, V. M. (2001). Complement. In ‘‘Principles and Practices of Clinical Immunology’’ (R. Rich,Ed.), pp. 21.1–21.8. Mosby, St. Louis, MO.

Holers, V. M., Girardi, G., Mo, L., Guthridge, J. M., Molina, H., Pierangelli, S., Espinola, R.,Xiaowei, L. E., Mao, D., Vialpando, C. J., and Salmon, J. E. (2002). Complement C3 activation isrequired for antiphospholipid antibody-induced fetal loss. J. Exp. Med. 195, 211–220.

Huang, J., Kim, L., Mealey, R., Marsh, H., Zhang, Y., Tenner, A., Connolly, E., and Pinsky, D.(1999). Neuronal protection in stroke by an sLex-glycosylated complement inhibitory protein.Science 285, 595–599.

Ikematsu, W., Luan, F. L., LaRosa, L., Beltrami, B., Nicoletti, F., Buyon, J. P., Meroni, P. L., andCasali, P. (1998). Human anticardiolipin monoclonal autoantibodies cause placental necrosis andfetal loss in BALB/c mice. Arth. Rheum. 41, 1026–1039.

innate autoimmunity 155

Janeway, C. A., Jr., and Medzhitov, R. (1999). Lipoproteins take their toll on the host. Curr. Biol. 9,R879–R882.

Kantor, A. B., Merrill, C. E., Herzenberg, L. A., and Hillson, J. L. (1997). An unbiased analysisof V(H)-D-J(H) sequences from B-1a, B-1b, and conventional B cells. J. Immunol. 158,1175–1186.

Kim, Y. U., Kinoshita, T., Molina, H., Hourcade, D., Seya, T., Wagner, L. M., and Holers, V. M.(1995). Mouse complement regulatory protein Crry/p65 utilizes the specific mechanisms of bothdecay-accelerating factor and membrane cofactor protein. J. Exp. Med. 181, 151–159.

Levy, E. M., Viscoli, C. M., and Horwitz, R. I. (1996). The effects of acute renal failure onmortality: A cohort analysis. JAMA 275, 1489–1494.

Li, B., Sallee, C., Dehoff, M., Foley, S., Molina, H., and Holers, V. M. (1993). Mouse Crry/p65:Characterization of monoclonal antibodies and the tissue distribution of a functional homologueof human MCP and DAF. J. Immunol. 151, 4295–4305.

Li, C., and Jackson, R. (2002). Reactive species mechanisms of cellular hypoxia–reoxygenationinjury. Am. J. Physiol. Cell Physiol. 282, C227–C241.

Liano, F., Junco, E., Pasqual, J., Madero, R., and Verde, E. (1998). The spectrum of acute renalfailure in the intensive care unit compared with that seen in other settings: The madrid acuterenal failure study group. Kidney Int. 66, S16–S24.

Linton, S. M., and Morgan, B. P. (1999). Complement activation and inhibition in experimentalmodels of arthritis. Mol. Immunol. 36, 905–914.

Liszewski, M. K., Farries, T. C., Lublin, D. M., Rooney, I. A., and Atkinson, J. P. (1996). Control ofthe complement system. Adv. Immunol. 61, 201–283.

Liu, D., Slevin, J. R., Lu, C., Chan, S. L., Hansson, M., Elmer, E., and Mattson, M. P. (2003).Involvement of mitochondrial Kþ release and cellular efflux in ischemic and apoptotic neuronaldeath. J. Neurochem. 86, 966–979.

Malynn, B., Yancopoulos, G., Barth, J., Bona, C., and Alt, F. (1990). Biased expression of JH-proximal VH genes occurs in the newly generated repertoire of neonatal and adult mice. J. Exp.Med. 171, 843–859.

Mao, D., Wu, X., Deppong, C., Friend, L. D., Dolecki, G. J., Nelson, D. M., and Molina, H.(2003). Negligible role of antibodies and C5 in pregnancy loss associated exclusivelywith C3-dependent mechanisms through complement alternative pathway. Immunity 19, 813–822.

Matsumoto, M., Fukuda, H., Circolo, A., Goellner, J., Strauss-Schoenberger, J., Wang, X., Fujita, S.,Hidvegi, T., Chaplin, D. D., and Colten, H. R. (1997). Abrogation of the alternativecomplement pathway by targeted deletion of murine factor B. Proc. Natl. Acad. Sci. USA 94,8720–8724.

McIntyre, J. A., Wagenknecht, D. R., and Faulk, W. P. (2003). Antiphospholipid antibodies:Discovery, definitions, detection and disease. Prog. Lipid Res. 42, 176–237.

Mellor, A. L., Sivakumar, J., Chandler, P., Smith, K., Molina, H., Mao, D., and Munn, D. H. (2001).Prevention of T cell-driven complement activation and inflammation by tryptophan catabolismduring pregnancy. Nat. Immunol. 2, 64–68.

Miwa, T., Sun, X., Ohta, R., Okada, N., Harris, C. L., Morgan, B. P., and Song, W. C. (2001).Characterization of glycosylphosphatidylinositol-anchored decay accelerating factor (GPI-DAF)and transmembrane DAF gene expression in wild-type and GPI-DAF gene knockout miceusing polyclonal and monoclonal antibodies with dual or single specificity. Immunology 104,207–214.

Mizuno, M., Nishikawa, K., Spiller, O. B., Morgan, B. P., Okada, N., Okada, H., and Matsuo, S.(2001). Membrane complement regulators protect against the development of type II collagen-induced arthritis in rats. Arthritis Rheum. 44, 2425–2434.

156 michael c. carroll and v. michael holers

Molina, H., Wong, W., Kinoshita, T., Brenner, C., Foley, S., and Holers, V. M. (1992). Distinctreceptor and regulatory properties of recombinant mouse complement receptor 1 (CR1) andCrry, the two genetic homologs of human CR1. J. Exp. Med. 175, 121–129.

Muller-Eberhard, H. J. (1988). Molecular organization and function of the complement system.Annu. Rev. Biochem. 57, 321–347.

Nath, K. A., Hostetter, M. K., and Hostetter, T. H. (1985). Pathophysiology of chronic tubulo-interstitial disease in rats: Interactions of dietary acid load, ammonia, and complement compo-nent C3. J. Clin. Invest. 76, 667–675.

Nishimura, J., Smith, C. A., Phillips, K. L., Ware, R. E., and Rosse, W. F. (1998). Paroxysmalnocturnal hemoglobinuria: Molecular pathogenesis and molecular therapeutic approaches.Hematopathol. Mol. Hematol. 36, 119–146.

Ochsenbein, A., Fehr, T., Lutz, C., Suter, M., Brombacher, F., Hengartner, H., and Zinkernagel, R.(1999). Control of early viral and bacterial distribution and disease by natural antibodies. Science286, 2156–2159.

Park, P., Haas, M., Cunningham, P. N., Bao, L., Alexander, J. J., and Quigg, R. J. (2002). Injury inrenal ischemia–reperfusion is independent from immunoglobulins and T lymphocytes.Am. J. Physiol. Renal Physiol. 282, F352–F357.

Qin, X., Miwa, T., Aktas, H., Gao, M., Lee, C., Qian, Y. M., Morton, C. C., Shahsafaei, A., Song,W. C., and Halperin, J. A. (2001). Genomic structure, functional comparison, and tissuedistribution of mouse Cd59a and Cd59b. Mamm. Genome 12, 582–589.

Rand, J. H. (2000). Antiphospholipid antibody-mediated disruption of the annexin-V antithrom-botic shield: A thrombogenic mechanism for the antiphospholipid syndrome. J. Autoimmun. 15,107–111.

Ratnoff, W. D., Fearon, D. T., and Austen, K. F. (1983). The role of antibody in the activation of thealternative complement pathway. Springer Semm. Immunopathol. 6, 361–371.

Reid, K. B. M., and Porter, R. R. (1981). The proteolytic activation systems of complement.Annu. Rev. Biochem. 50, 433–464.

Reid, R. R., Woodcock, S., Shimabukuro-Vornhagen, A., Austen, W. G., Jr., Kobzik, L., Zhang, M.,Hechtman, H. B., Moore, F. D., Jr., and Carroll, M. C. (2002). Functional activity of naturalantibody is altered in Cr2-deficient mice. J. Immunol. 169, 5433–5440.

Richards, A., Goodship, J. A., and Goodship, T. H. (2002). The genetics and pathogenesisof haemolytic uraemic syndrome and thrombotic thrombocytopenic purpura. Curr. Opin. Ne-phrol. Hypertens. 11, 431–435.

Richards, A., Kemp, E. J., Liszewski, M. K., Goodship, J. A., Lampe, A. K., Decorte, R.,Muslumanoglu, M. H., Kavukcu, S., Filler, G., Pirson, Y., Wen, L. S., Atkinson, J. P., andGoodship, T. H. (2003). Mutations in human complement regulator, membrane cofactor protein(CD46), predispose to development of familial hemolytic uremic syndrome. Proc. Natl. Acad.Sci. USA 100, 12966–12971.

Seidl, K., Wilshire, J., MacKenzie, J., Kantor, A., Herzenberg, L., and Herzenberg, L. (1999).Predominant VH genes expressed in innate antibodies are associated with distinctive antigen-binding sites. Proc. Natl. Acad. Sci. USA 96, 2262–2267.

Sheridan, A. M., and Bonventre, J. V. (2000). Pathophysiology of ischemic acute renal failure.Contrib. Nephrol. 132, 7–21.

Simpson, R., Alon, R., Kobzik, L., Valeri, C., Shepro, D., and Hechtman, H. (1993). Neutrophiland nonneutrophil-mediated injury in intestinal ischemia–reperfusion. Ann. Surg. 218,444–453.

Sporer, H., Lang, F., Oberleithner, H., Greger, R., and Deetjen, P. (1981). Inefficacy of bicarbonateinfusions on the course of postischaemic acute renal failure in the rat. Eur. J. Clin. Invest. 11,311–315.

innate autoimmunity 157

Stahl, G. L., Xu, Y., Hao, L., Miller, M., Buras, J. A., Fung, M., and Zhao, H. (2003). Role forthe alternative complement pathway in ischemia/reperfusion injury. Am. J. Pathol. 162,449–455.

Tatsumi, T., Shiraishi, J., Keira, N., Akashi, K., Mano, A., Yamanaka, S., Matoba, S., Fushiki, S.,Fliss, H., and Nakagawa, M. (2003). Intracellular ATP is required for mitochondrial apoptoticpathways in isolated hypoxic rat cardiac myocytes. Cardiovasc. Res. 59, 428–440.

Thadhani, R., Pasqual, M., and Bonventre, J. V. (1996). Acute renal failure. N. Engl. J. Med. 334,1448–1460.

Thurman, J. M., Ljubanovic, C., Edelstein, C. L., Gilkeson, G. S., and Holers, V. M. (2003). Lack ofa functional alternative pathway ameliorates ischemic acute renal failure in mice. J. Immunol.170, 1517–1523.

Thurman, J. M., Kraus, D. M., Girardi, G., Hourcade, D. E., Kang, H.-J., Royer, P. A., Mitchell,L. M., Giclas, P. C., Salmon, J. E., Gilkerson, G. S., and Holers, V. M. (2005). A novel inhibitor ofthe alternative complement pathway prevents antiphospholipid antibody-induced pregnancyloss in mice. Mol. Immunol. 42, 87–97.

Vabulas, R. M., Ahmad-Nejad, P., Ghose, S., Kirschning, C. J., Issels, R. D., and Wagner, H. (2002).HSP70 as endogenous stimulus of the Toll/interleukin-1 receptor signal pathway. J. Biol. Chem.277, 15107–15112.

Weiser, M., Williams, J., Moore, F., Kobzik, L., Ma, M., Hechtman, H., and Carroll, M. (1996).Reperfusion injury of ischemic skeletal muscle is mediated by natural antibody and complement.J. Exp. Med. 183, 2343–2348.

Weisman, H. F., Bartow, T., Leppo, M. K., Marsh, H. C., Carson, G. R., Concino, M. F., Boyle,M. P., Roux, K. H., Weisfeldt, M. L., and Fearon, D. T. (1990). Souble human complementreceptor type I: In vivo inhibitor of complement suppressing post-ischemic myocardial inflam-mation and necrosis. Science 249, 146–151.

Williams, J. P., Moore, F. D., Kobzik, L., Carroll, M. C., and Hechtman, H. B. (1999). Intestinalreperfusion injury is mediated by IgM and complement. J. Appl. Physiol. 86, 938–942.

Xu, C., Mao, D., Holers, V. M., Balanca, B., Cheng, A. M., and Molina, H. (2000). A critical rolefor murine complement regulator Crry in fetomaternal tolerance. Science 99, 498–501.

Zhang, M., Austen, W. G., Jr., Chiu, I., Alicot, E. M., Hung, R., Ma, M., Verna, N., Xu, M.,Hechtman, H. B., Moore, F. D., Jr., and Carroll, M. C. (2004). Identification of a specific self-reactive IgM antibody that initiates intestinal ischemia/reperfusion injury. Proc. Natl. Acad. Sci.USA 101, 3886–3891.

Zhou, W., Farrar, C. A., Abe, K., Pratt, J. R., Marsh, J. E., Wang, Y., Stahl, G. L., and Sacks, S. H.(2000). Predominant role for C5b-9 in renal ischemia/reperfusion injury. J. Clin. Invest. 105,1363–1371.

Formation of Bradykinin: A Major Contributor to the InnateInflammatory Response

Kusumam Joseph and Allen P. Kaplan

Division of Pulmonary/Critical Care Medicine and Allergy/Clinical Immunology,Medical University of South Carolina,

Charleston, South Carolina 29425

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1591. Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1602. Contact Activation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1613. Assembly on Cell Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1664. Activation of the Kinin Cascade: The Role of Endothelial Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1755. Inhibition of Contact Activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1816. Inactivation of Bradykinin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1847. Relations of the Contact Factors to Other Systems .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1858. Considerations in Human Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

Abstract

The plasma kinin-forming cascade can be activated by contact with negativelycharged macromolecules leading to binding and autoactivation of factor XII,activation of prekallikrein to kallikrein by factor XIIa, and cleavage of highmolecular weight kininogen (HK) by kallikrein to release the vasoactive peptidebradykinin. Once kallikrein formation begins, there is rapid cleavage of un-activated factor XII to factor XIIa, and this positive feedback is favored kineti-cally over factor XII autoactivation. Examples of surface initiators that canfunction in this fashion are endotoxin, sulfated mucopolysaccharides, andaggregated Ab protein. Physiological activation appears to occur along thesurface of endothelial cells both by the aforementioned contact-initiated reac-tions as well as bypass pathways that are independent of factor XII. Factor XIIbinds primarily to cell surface u-PAR (urokinase plasminogen activator recep-tor); HK binds to gC1qR via its light chain (domain 5) and to cytokeratin 1 byits heavy chain (domain 3) and, to a lesser degree, by its light chain. Prekal-likrein circulates bound to HK (as does coagulation factor XI), and prekallik-rein is thereby brought to the surface as HK binds. All cell-binding reactionsare dependent on zinc ion. Endothelial cells (HUVECs) have bimolecularcomplexes of u-PAR–cytokeratin 1 and gC1qR–cytokeratin 1 at the cell surfaceplus free gC1qR, which is present in substantial molar excess. Factor XIIappears to interact primarily with the u-PAR–cytokeratin 1 complex, whereasHK binds primarily to the gC1qR–cytokeratin 1 complex and to free gC1qR.

159advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

160 kusumam joseph and allen p. kaplan

Release of endothelial cell heat shock protein 90 (Hsp90) or the enzymeprolylcarboxypeptidase leads to activation of the bradykinin-forming cascadeby activating the prekallikrein–HK complex. In contrast to factor XIIa, neitherwill activate prekallikrein in the absence of HK, both reactions require zinc ion,and the stoichiometry suggests interaction of one molecule of Hsp90 (forexample) with one molecule of prekallikrein–HK complex. The presence offactor XII, however, leads to a marked augmentation in reaction rate via thekallikrein feedback as well as to a change to classic enzyme–substrate kinetics.The circumstances in which activation is initiated by factor XII autoactivationor by these factor XII bypasses are yet to be defined. The pathologic condi-tions in which bradykinin generation appears important include hereditaryand acquired C1 inhibitor deficiency, cough and angioedema due to ACEinhibitors, endotoxin shock, with contributions to conditions as diverse asAlzheimer’s disease, stroke, control of blood pressure, and allergic diseases.

1. Introduction

The plasma pathway by which bradykinin is generated is closely linked to thepathways of coagulation, fibrinolysis, and inflammation and, by analogy with,the alternative complement pathway, the mannose-binding lectin (MBL) com-plement pathway, and Toll-like receptors; it represents part of the ‘‘innate’’inflammatory response rather than a strictly innate ‘‘immune’’ response. Thepathway consists of three requisite plasma proteins, namely, coagulation factorXII (Hageman factor), prekallikrein, and high molecular weight kininogen(HK), which interact as a plasma proteolytic cascade (Kaplan and Silverberg,1987). However the most recent data indicate that there are important inter-actions of those proteins with cell surface ‘‘receptors’’ such that activation ofthe cascade and generation of bradykinin can occur along the cell surface. Keycell surface proteins thus far identified include qC1qR (the receptor for theglobular heads of the C1Q subcomponent of complement) (Joseph et al., 1996),cytokeratin 1 (Hasan et al., 1998; Joseph et al., 1999b), and the urokinaseplasminogen activation receptor (u-PAR) (Colman et al., 1997). These proteinsfulfill binding functions on the surface of microvascular endothelial cells of theskin and lung, as well as human umbilical vein endothelial cells (HUVECs),astrocytes, and possibly smooth muscle cells. HK binding to neutrophils hasbeen shown to be dependent on MAC-1 (CD11b/CD18) (Wachtfogel et al.,1994), whereas HK binding to platelets requires interaction with glycoprotein1b (Bradford et al., 1997; Joseph et al., 1999a). Activation along the cell surfacehas been most extensively studied with HUVECs, and observations havedramatically altered our understanding of the biochemical mechanisms by

formation of bradykinin 161

which these proteins interact. New initiation mechanisms have been describedthat are dependent on cell-derived proteins, such as heat shock protein 90(Hsp 90) (Joseph et al., 2002b) and prolylcarboxypeptidase (Shariat-Madaret al., 2002). In this article we first review the role of each of the plasmaproteins in the generation of bradykinin, and then describe binding andactivation at the plasma–cellular interface with consideration of physiologicaland pathologic consequences of these reactions.

2. Contact Activation

The concept of contact activation was originally developed because itwas found that addition of blood to a glass tube leads to coagulation. Thus‘‘contact’’ with the silicate surface appeared to initiate a proteolytic cascadeculminating in the conversion of fibrinogen of fibrin. At the same time,bradykinin is generated. These reactions that occur during activation in thisfashion are shown in Fig. 1.

Figure 1 Pathway for bradykinin formation indicating the autoactivation of factor XII, the positivefeedback by which kallikrein activates factor XII, cleavage of high molecular weight kininogen(HK) to release bradykinin, formation of factor XII fragment, and enzymatic activation of C1. Thesteps inhibitable by C1 INH are indicated.

162 kusumam joseph and allen p. kaplan

It has been shown that factor XII (Hageman factor) circulates as a b

globulin, of molecular weight 80,000, which autoactivates on binding to sur-faces bearing negative changes (Silverberg et al., 1980a). Because the zymogenhas no detectable enzymatic activity (Silverberg and Kaplan, 1982), it has beenproposed that trace quantities of the active enzyme (factor XIIa) actuallycirculate but that digestion of factor XII by factor XIIa occurs only on bindingof factor XII to the surface. Thus the surface renders factor XII a substrate(Griffin, 1978) for traces of preformed factor XIIa. Prekallikrein (PK) andcoagulation factor XI each circulate as a complex with high molecular weightkininogen (HK) (Mandle et al., 1976; Thompson et al., 1977) with a stoichiom-etry of 1:1 and 1:2, respectively (factor XI is a dimer). The binding sites forprekallikrein and factor XI on HK overlap (Tait and Fujikawa, 1986, 1987) tosuch a degree that HK can bind only one molecule of each, but never both.HK, however, is present in considerable molar excess. Thus we have separatecomplexes of prekallikrein–HK and factor XI–HK and the percentage boundto HK in each case is 85 and 95%, based on equilibrium considerations (Scottand Colman, 1980).

High molecular weight kininogen is a key factor that regulates contactactivation. It is also the link protein that allows assembly of the kinin-formingcascade along the surface of cells, and we therefore consider its structuralfeatures in some detail. Human plasma has two kininogens (Jacobsen, 1966)that are designated high molecular weight kininogen (HK) and low molecularweight kininogen (LK). They are assembled by alternative splicing of theterminal exons (Fig. 2) such that a large portion of their amino acid sequenceis identical (Kitamura et al., 1985). The domain structure of the protein HK isshown in Fig. 3; it consists of six domains. At the N terminus are three domains(encoded by exons 1–9) that are homologous to cystatins and stefans(Kellermann et al., 1986a), including sulfhydryl proteases such as cathepsinB, H, and L, and domains 2 and 3 actually retain cysteine protease inhibitoryactivity (Gounaris et al., 1984; Higashiyama et al., 1986; Ishiguro et al., 1987;Muller-Esterl et al., 1985a). Domain 4 contains the bradykinin sequence plusthe next 12 amino acids. Up to this point LK and HK have identical amino acidsequences. Then exon 10, which includes bradykinin plus domains 5 and 6, isadded for HK (Fig. 2), whereas exon 11 is added for LK with the attachment atthe C terminus of domain 4. When HK is cleaved by plasma kallikrein torelease bradykinin (fast cleavage occurs at a C-terminal Arg–Ser bond, fol-lowed by cleavage at an N-terminal Lys–Arg bond) (Mori and Nagasawa, 1981;Mori et al., 1981), and the kinin-free HK is reduced and alkylated, one canisolate a heavy chain (domains 1–3) and a light chain (the C-terminal 12 aminoacids of domain 4 plus domains 5 and 6) (Thompson et al., 1979). Thus thelight chain of HK and LK are quite different (Kellermann et al., 1986b), and

Figure 2 The gene for HK. Boxes 1–9 represent the exon encoding the heavy chain of both HKand LK. Exon 10 encodes the bradykinin sequence and the light chain of HK, whereas exon 11encodes the light chain of LK. The mature mRNAs are assembled by alternative splicing events inwhich the light chain sequences are attached to the 30 end of the 13-amino acid common sequenceC terminal to bradykinin. HMWK¼HK; LMWK¼LK.

formation of bradykinin 163

this accounts for the difference in molecular weight and many of the functionalproperties of HK that are not shared by LK. It should be noted that plasmakallikrein preferentially cleaves HK (Reddigari and Kaplan, 1988, 1989),whereas tissue kallikrein (encoded by a separate gene from that of plasmakallikrein), cleaves both HK and LK, but with more favorable kinetics if LK isthe substrate (Lottspeich et al., 1984; Muller-Esterl et al., 1985b).

The function of HK in contact activation, as depicted in Fig. 1, are multiple.First, it accelerates the conversion of PK and factor XI to kallikrein and factorXIa, respectively (Griffin and Cochrane, 1976; Meier et al., 1977; Wiggins et al.,1977). This acceleration appears to be due to the ability of PK and factor XI tobind to HK; as a result each of them is in a more favorable conformation foractivation than when they are tested unbound. In addition, HK providesthe attachment to initiating surfaces and brings both PK and factor XI to thesurface as a complex. If PK and factor XI bind to the surface in the absence ofHK, activation by factor XIIa is markedly inhibited, even when comparedwith activation in the fluid phase. Thus the conformational effects of bindingof PK and factor XI to HK are even more evident when activation along the

Figure 3 The structure of HK. The heavy chain region consists of three homologous domains(1–3), of which the latter two are sulfhydryl protease inhibition sites. Domain 4 contains thebradykinin moiety. The light chain region contains the surface binding site (domain 5) andoverlapping binding sites for prekallikrein and factor XI (domain 6).

164 kusumam joseph and allen p. kaplan

surface is compared with fluid-phase activation (since factor XII is activatedalong surfaces, this comparison is made by adding preformed factor XIIa to PKor PK–HK either in solution or bound to a surface).

Figure 1 also depicts a positive feedback in which kallikrein activatessurface-bound factor XII to form factor XIIa (Cochrane et al., 1973; Meieret al., 1977; Silverberg et al., 1980b). In fact, factor XII that is bound under-goes a conformational change that renders it a substrate for factor XIIa (Dunnet al., 1982; Griffin, 1978). Thus autoactivation of factor XII can initiate thecascade once sufficient factor XIIa forms to overcome plasma inhibitors(Silverberg et al., 1980b; Tankersley and Finlayson, 1984), and only a fewpercent conversion to factor XIIa is required. Then the kallikrein formedactivates the remaining surface-bound factor XII at a much more rapid rate.This positive feedback is also accelerated by the presence of HK (Griffin andCochrane, 1976; Meier et al., 1977; Silverberg et al., 1980b). The factor XIIaformed remains attached to the initiating surface; further digestion of factorXIIa by kallikrein (Fig. 1) yields a 32.5-kDa factor XII fragment (factor XIIf)(Dunn and Kaplan, 1982; Kaplan and Austen, 1970, 1971) that retains theactive site of factor XIIa but lacks the binding site to the surface. It is a doubleton sodium dodecyl sulfate (SDS) gels with bands at 30 and 28.5 kDa. Factor

formation of bradykinin 165

XIIf consists of the light chain of factor XIIa, containing the active site,disulfide linked to small C-terminal fragments of the heavy chain of 2000 or500 corresponding to the two SDS bands. Factor XIIf lacks the binding site forthe surface and is released into the fluid phase, where it can continue toactivate prekallikrein until it is ultimately inactivated by plasma proteaseinhibitors. Factor XI is activated by factor XIIa in the presence of HK, butfactor XIIf possesses only 2–4% of the coagulant activity of factor XIIa and HKdoes not augment its reaction rate. Thus factor XIIf can be important forbradykinin formation, but not for intrinsic coagulation.

The mechanism by which HK catalyzes factor XII activation is multifactorialand indirect, since HK does not increase the enzymatic activity of kallikrein,nor does it interact with factor XII to render it a better substrate (Silverberget al., 1980b). Its main effect is to allow dissociation of kallikrein from itscomplex with HK so that it can enzymatically activate factor XII along the cellsurface (Cochrane and Revak, 1980). Kallikrein bound directly to the surface ismuch less effective and cannot disseminate the reaction (Silverberg et al.,1980b). Since HK is required for the formation of kallikrein, that is, activationof PK, the amount of kallikrein is increased when HK is present. Thus theeffective ratio of kallikrein/factor XII in this enzymatic reaction is significantlyaugmented when HK is present.

The percent augmentation of contact activation in the presence of a surfaceplus HK is estimated to be 3000- to 12,000-fold (Rosing et al., 1985; Tankersleyand Finlayson, 1984). If one considers the rate of factor XI activation, HK-deficient plasma is almost as abnormal as factor XII-deficient plasma. HKincreases the rate of formation of kallikrein, it facilitates factor XII conversionto factor XIIa by kallikrein, and it facilitates factor XI activation by factor XIIa.For comparison, it is of interest to consider the rate of factor XI activation inPK-deficient plasma, where the kallikrein feedback activation of factor XII isnot possible, and the only role of HK is in conversion of factor XI to factor XIa.In this case contact activation of coagulation is very slow, but if the time ofincubation of citrated plasma with the surface is increased prior to recalcifica-tion, the clotting time approaches normal (Saito et al., 1974; Weiss et al., 1974;Wuepper, 1973). This is due to gradual conversion of factor XI to factor XIa asa result of factor XII autoactivation on the surface. The next step, conversion offactor IX to factor IXa by factor XIa, is dependent on calcium, and thus aprolonged incubation allows the amount of factor XIa to increase towardnormal. It should be evident from Fig. 1 that plasma that is deficient in eitherfactor XII, PK, or HK cannot generate bradykinin via contact activation. Anybradykinin formed is then dependent on tissue kallikrein activation of LK.

A detailed discussion of the structure of each protein, transcription andtranslational events involved in the synthesis of each protein, and mechanistic

166 kusumam joseph and allen p. kaplan

details regarding activation of factor XII, PK, and factor XI has been presentedin this series previously (Kaplan et al., 1997), and the reader is referred to thatpublication for further details.

3. Assembly on Cell Surfaces

3.1. Binding of High Molecular Weight Kininogen to Human UmbilicalVein Endothelial Cells

The initial studies of the interaction of proteins of the kinin-forming cascadewith cells were performed with platelets (Greengard and Griffin, 1984;Gustafson et al., 1986) and then human umbilical vein endothelial cells(HUVECs) (Schmaier et al., 1988; van Iwaarden et al., 1988). In each instance,HK was shown to bind to each cell type in a zinc-dependent fashion. Thebinding was saturable and reversible; however, binding was found to bedependent on domains 3 and 5 (Hasan et al., 1995; Herwald et al., 1995;Reddigari et al., 1993a), so that both heavy chain and light chain were capableof similar ion-dependent interactions (Reddigari et al., 1993a). Although directprekallikrein binding to HUVECs has been described (Mahdi et al., 2003), it isof doubtful physiological relevance, since prekallikrein is brought to the cellsurface as a bimolecular complex as a result of its interaction with domain 6 ofHK. Factor XII interacts with HUVECs in a similar fashion to HK; theinteraction requires zinc ion and factor XII, and HK can compete for bindingto the cell surface (Reddigari et al., 1993b). The latter observation suggeststhat they bind to similar cell surface proteins with comparable affinity.

We therefore sought to purify and characterize this binding protein; theresults (Joseph et al., 1996) are summarized below and correspond to a p33endothelial cell protein isolated also by Herwald et al. (1996) and identifiedto be gC1qR, the receptor for the globular heads of C1q discovered byGhebrehiwet et al. (1994). A solubilized endothelial cell membrane prepara-tion was passed over an HK affinity column in the presence or absence of zincion and eluted with glycine-HCl (0.1 M, pH 2.5), and the fractions wereneutralized. An aliquot of each eluate fraction (with or without zinc) was spottedonto nitrocellulose membrane and blotted with biotinylated HK, developedwith alkaline phosphatase–streptavidin, followed by reaction with nitrobluetetrazolium/5-bromo-4-chloroindolyl phosphate. A prominent increase in HKbinding was observed after elution in the presence of zinc. These fractions werethen pooled, concentrated, and analyzed by SDS–PAGE. The main feature wasthe appearance of a new prominent band at 33 kDa that was visible withCoomassie stain. In addition, ligand blot experiments demonstrated that bio-tinylated HK bound only to the 33-kDa band. Based on this information, the

formation of bradykinin 167

33-kDa protein was subjected for N-terminal amino acid sequence analysisand the first 13 amino acids were found to be identical to the known NH2

terminus of gC1qR (Ghebrehiwet et al., 1994). We next performed a Westernblot using anti-gC1qR monoclonal antibody 60.11 to further assess the identityof these two proteins. Monoclonal antibody 60.11, which interacts with anepitope at the N terminus of gC1qR, identified the 33-kDa HUVEC-derivedmembrane-binding protein.

We next determined whether factor XII could also bind to gC1qR. HUVECmembrane–purified gC1qR or recombinant gC1qR (rgC1qR) at 1.0 to 2.0 mgwas applied to nitrocellulose membranes and blotted with biotinylated HK orfactor XII in the presence or absence of 50 mM zinc. Sufficient o-phenanthro-line was added to bind the zinc in the purified (but not recombinant) gC1qR toallow zinc-independent binding to be assayed. We found that both HK andfactor XII bind to either purified or rgC1qR in the presence of zinc. Additionalcontrols included biotinylated IgG, which did not bind to gC1qR, and sub-stituting prekallikrein for gC1qR to which biotinylated HK bound, as expected(Mandle et al., 1976). Addition of excess unlabeled HK reversed the ability offactor XII to bind to gC1qR by over 90% as quantitated by scanning, whichsuggests interaction with a common domain within the protein. Factor XIIonly partially (46%) reverses HK binding. This difference may be due to therelative affinity of the two ligands for the gC1qR molecule. Nevertheless, thesedata completely paralleled those observed on binding of factor XII and HK toendothelial cells.

We next attempted to demonstrate that the interaction with gC1qR isindeed responsible for binding to the cell surface, which was addressed byinhibition experiments. HUVECs were incubated for 30 min with HK (8.7 �10�8 M), or anti-gC1qR mAbs 74.5.2 and 60.11. Then, in the presence of50 mM zinc, [125I]HK was added and incubated for 60 min at 37 8C, conditionsknown to saturate the binding sites (Hasan et al., 1995). For each condition,the percentage inhibition of [125I]HK binding was determined. Whereas a 100-molar excess of nonradiolabeled HK inhibited subsequent [125I]HK binding, acomparable concentration of C1q did not, indicating that the binding sites ofgC1qR for C1q and for HK do not overlap. Furthermore, although mAb 60.11did not efficiently inhibit [125I]HK binding to HUVECs, mAb 74.5.2 did. Ithad been shown previously that antibody 60.11 inhibits C1q binding to gC1qR,whereas mAb 74.5.2 does not. Other mAbs recognizing epitopes in differentregions of the molecule were also tested for their inhibitory activity. Of these,only 25.15, which is similar to 74.5.2, was able to inhibit [125I]HK binding toHUVECs. Herwald et al. (1996) also demonstrated that gC1qR, a majorendothelial cell-binding protein for HK, used a domain 5-derived peptidefrom the light chain rather than whole HK as the ligand. In aggregate, these

168 kusumam joseph and allen p. kaplan

data indicate that gC1qR serves as a zinc-dependent binding protein for factorXII as well as for HK, and that binding to HK occurs via the light chain moiety.The specific location within HK for binding to endothelial cells is withindomain 5 (Hasan et al., 1995; Reddigari et al., 1993a), and this also appearsto be the site of interaction with gC1qR. It is also clear that the HK heavychain also binds to endothelial cells. This interaction has been shown torequire domain 3.

The methods employed to identify the cell surface protein that interactswith heavy chain are analogous to those described above for isolation ofgC1qR. Later, a second HK-binding protein was identified in HUVECs byaffinity chromatography employing HK as ligand (Hasan et al., 1998; Shariat-Madar et al., 1999) and was identified as cytokeratin 1. We reasoned that thisprotein might contribute to heavy chain interaction with cells and thereforeprepared an affinity column by covalently coupling peptide LDC27 sequenceto the matrix; this is a 27-amino acid peptide derived from domain 3, which hasbeen identified as an HK-binding site (Herwald et al., 1995). When cellmembranes derived from HUVECs were applied to the column in the pres-ence and absence of 50mM zinc and each was eluted with 0.1 M glycine-HCl,pH 2.5, a band was noted at molecular mass of 68 kDa in the zinc-containingeluate. A ligand blot with HK confirmed binding to this band. When weattempted to sequence it, the N terminus was blocked. We therefore nextdigested the protein with cyanogen bromide and subjected the mixture to massspectrometry. A major peptide at molecular weight 2721.7 was identified, andits sequence was determined and shown to correspond to an internal peptidederived from cytokeratin 1 (Fig. 4) (Joseph et al., 1999b,c). Thus HK bindingto HUVECs appeared to depend on interaction with two proteins, cytokeratin1 and gC1qR, with binding to each by domains 3 and 5 of HK, respectively(i.e., binding of heavy chain to cytokeratin 1 and light chain to gC1qR). Wedemonstrated that gC1qR cannot bind heavy chain at all, whereas cytokeratin1, when tested as a purified protein, can bind both heavy and light chains;however, binding to heavy chain clearly predominates. Factor XII is capableof binding to both proteins. To confirm that these proteins are importantfor binding to endothelial cells, we performed an inhibition experiment inwhich antibody to gC1qR and antibody to cytokeratin 1 were employed. As canbe seen in Fig. 5, antibody to gC1qR inhibited zinc-dependent binding by65%, antibody to cytokeratin 1 inhibited binding by 30%, whereas a combina-tion of antisera inhibited binding by 85%. Since 15% binding corresponds tozinc-independent binding, our data suggest that we account for most, if not all,HK binding to endothelial cells by these two proteins.

A third protein reported to be important for HK binding to HUVECswas identified to be u-PAR (the urokinase plasminogen activator receptor)

Figure 4 Localization of the peptide fragment within the cytokeratin 1 sequence. Completeamino acid sequence of cytokeratin 1 indicates a 24-amino acid internal peptide (boldface, doubleunderlined) corresponding identically to the sequence of the peptide isolated from the cyanogenbromide digest. Fragments were isolated by HPLC and the 2721.7 molecular weight peak wassequenced.

formation of bradykinin 169

by inhibition of HK binding with antisera to domain 2/3 of u-PAR (Colmanet al., 1997). However, we have not been able to isolate u-PAR fromcell membranes (confirmed to contain considerable u-PAR) by HK affinitychromatography. One difference in the experiments is that the studies byColman et al. employed cleaved HK lacking the bradykinin moiety; it ispossible that cleaved HK binds more avidly to u-PAR than does native HKwhile native HK binds more avidly to gC1qR and cytokeratin 1 than it does tou-PAR.

3.2. Binding of Factor XII to HUVEC

Early studies demonstrated that factor XII binds to both gC1qR and cytoker-atin 1 and that it competes for the same binding sites as does HK (Fig. 6). Thefirst study to attempt to identify the binding site on HUVECs (rather thantesting purified proteins shown to bind HK) was a study by Mahdi et al.in which antisera to u-PAR, cytokeratin 1, and gC1qR were employed toinhibit cell binding of factor XII (Mahdi et al., 2002). A surprising result wasthat antibody to u-PAR inhibited best, although the other antisera were

Figure 5 [125I]HK binding to HUVECs and its inhibition by monoclonal antibodies. (A) HUVECswere incubated with [125I]HK (20 nM) in the presence (Z) or absence (m) of 50 mM zinc. After theindicated times, the cells were washed and counted for bound radioactivity. (B) For inhibitionstudies, cells were preincubated with monoclonal antibodies or nonimmune mouse IgG for 30 min.After 30 min, [125I]HK was added and further incubated for 1 hr at room temperature. The linesrepresent HK-binding values after treatment with a monoclonal antibody to gC1qR (s), monoclo-nal antibody to u-PAR (Z), monoclonal antibody to cytokeratin 1 (“), and a combination ofmonoclonal antibodies to gC1qR and cytokeratin 1 (r). Antibody to gp1b (control) showed similarresults to the control mouse IgG. Each point is a mean of three different experiments, performedin triplicate.

170 kusumam joseph and allen p. kaplan

Figure 6 High molecular weight kininogen competes with factor XII for the same binding sites onHUVECs. (A) HUVECs were incubated with [125I]FXII (1mg/ml) in triplicate in the presence ofincremental concentrations of unlabeled factor XII, HK, or normal human IgG for 120 min, andbound ligand was determined. The percentage bound in the presence of a competitor is plottedagainst the concentration of the competitor. (B) HUVECs were incubated with [125I]HK (1mg/ml)in triplicate in the presence of increasing concentration of unlabeled factor XII, and bound ligandwas determined.

formation of bradykinin 171

172 kusumam joseph and allen p. kaplan

contributory. We therefore sought to corroborate the observation by isolationof factor XII–binding proteins directly from HUVEC-derived cell membranepreparations by affinity chromatography employing factor XII as ligand. Themajor zinc-dependent binding protein was clearly u-PAR; gC1qR was alsoisolated as well as small amounts of cytokeratin 1 (our unpublished data). Wecould also demonstrate more avid binding of factor XII to u-PAR than to eithergC1qR or cytokeratin 1 by competitive displacement of factor XII bound toone protein, employing increasing quantities of either of the other two. Thuswe began to develop a model for assembly of the kinin-forming cascade onendothelial cells in which zinc-dependent binding of factor XII is associatedpredominantly with u-PAR, and HK binds to gC1qR as well as cytokeratin 1while prekallikrein is bound to the HK. How that binding occurs, particularlyfor HK, depends on the way these proteins are distributed within the cellmembrane of HUVECs.

3.3. Interaction of gC1qR, Cytokeratin 1, and u-PAR withHUVEC Membranes

One dilemma posed by antibody inhibition studies in which all three antiserawere employed was that the total percent inhibition obtained when the threepercentages were added exceeded 100%. One possible explanation was thatthese proteins might interact in some fashion within the cell membrane anda trimolecular complex containing all three was proposed (Colman andSchmaier, 1997). Thus antisera to one protein might sterically interfere withbinding to the others and falsely influence the percent inhibition observed.A second, different alternative was proposed by another group, who reportedthat HK binding to HUVECs was not due to interaction with proteins at all, butthat proteoglycans such as syndican and glypican were in fact responsible(Renne et al., 2000) and questioned whether gC1qR is truly demonstrablealong the cell membrane surface (Dedio and Muller-Esterl, 1996; Dedio et al.,1998). It should be noted that u-PAR is known to be linked to cell membraneconstituents by a phosphatidylinositol bond, but that gC1qR and cytokeratin 1are not, and the latter two proteins lack typical transmembrane domains. Thusif they are present within cell membranes, the mode of attachment is unknown.Finally, the number of binding sites reported for gC1qR on the cell membranevaried from just under 1 million to 10 million (Motta et al., 1998; Reddigariet al., 1993a, van Iwaarden et al., 1988); questions were raised regarding such ahigh figure, although binding to a cell membrane proteoglycan could achievesuch levels.

We (and others) addressed each of these issues. Employing high-titer,monospecific antisera to gC1qR, the protein was clearly demonstrated to be

formation of bradykinin 173

at the HUVEC surface (Joseph et al., 1999c). Mahdi et al. then demonstratedthe presence of all these proteins within cell membranes by immunoelectronmicroscopy (Mahdi et al., 2001, 2002). Cytokeratin 1 and u-PAR appeared tobe colocalized while gC1qR was present throughout the cell membrane. Weaddressed the question of binding to proteoglycan by employing heparanases,which remove all heparan sulfate–containing structures from cell membranesand demonstrated that HUVECs treated so as to become unreactive to anti-sera to heparan sulfate (the major sulfate source associated with cell surfaceproteoglycans) bound HK normally, and the binding was then inhibited withantisera to gC1qR and cytokeratin 1 (Fernando, 2003a). We next solubilizedpurified cell membranes from HUVECs and demonstrated that gC1qR, cyto-keratin 1, and u-PAR are all present, by immunoblot analysis. There was nosignificant contamination by other cell constituents, particularly mitochondria,which are known to contain large amounts of gC1qR (Dedio et al., 1998). Wethen addressed the interactions of these proteins with each other. First, wecould show that gC1qR binds to cytokeratin 1 but not u-PAR while u-PAR alsobinds to cytokeratin 1, but not to gC1qR. Thus a trimolecular complex is notpossible, but two bimolecular complexes seemed feasible. We then precipi-tated gC1qR and u-PAR from cell membrane preparations and analyzedthe composition of the precipitated materials. Cytokeratin 1 was precipitatedwith both anti-gC1qR and anti-u-PAR; however, the gC1qR–cytokeratin1-containing fraction had no u-PAR while the cytokeratin 1–u-PAR fractioncontained no gC1qR. Our current view of the assembly of the proteins of thekinin-forming cascade on HUVECs envisions factor XII bound to a complex ofu-PAR–cytokeratin 1 while HK binds to a complex of gC1qR–cytokeratin 1. Wedo not know whether HK binds to the complex by both domain 3 and domain 5simultaneously or whether binding to one site affects binding to the other.Complicating this assessment is the fact that the number of gC1qR sites withinthe cell membrane is at least three times that of u-PAR or cytokeratin 1, andthus gC1qR unassociated with either cytokeratin 1 or u-PAR is likely presentand can bind HK or factor XII. Given the relative affinities of factor XII, HKheavy chain, and HK light chain for gC1qR, we would anticipate preferentialbinding of the light chain (domain 5) of HK to gC1qR. Consistent with this isthe prominent inhibition of HK binding to the cell when employing peptideHKH20 derived from domain 5 of the light chain (Nakazawa et al., 2002).

3.4. Binding to Other Cells

The interaction of factor XII and HK with other cell types resembles that seenin HUVECs, although there are differences in the number of binding sites, theaffinity of binding, and the nature of the binding proteins. We reported studies

174 kusumam joseph and allen p. kaplan

of microvascular endothelial cells derived from skin and lung and comparedbinding with that seen in HUVECs, since these cells are more likely to be‘‘physiological’’ (Fernando, 2003b). We also compared binding to HUVECswith that seen with astrocytes, because upregulation of bradykinin B2 recep-tors has been demonstrated within the central nervous system of patientswith Alzheimer’s disease (Jong et al., 2002), and aggregated Ab protein ofAlzheimer’s disease is a potent activator of the kinin cascade (Shibayama et al.,1999). The results are shown in Table 1. It should be noted that a large numberof HK-binding sites was demonstrated for HUVECs by two separate methods,including a fluid phase-based assay in which 850,000 sites per cell weredocumented. It has been suggested that values in the 10 million range(Motta et al., 1998) may have been due to ligand binding to the plates usedrather than to the cells coating the plates, and thus the cell surface representedonly a fraction of the total binding seen (Baird and Walsh, 2002, 2003).However, the inhibition of such binding, employing antisera to the cell surfaceligands, suggests some other interpretation (Mahdi et al., 2003), and our valuesin the fluid phase are in close agreement with binding studies performed withmicrotiter plates (Table 1).

Binding of HK and factor XII to neutrophils and platelets has also beenstudied. HK interacts with neutrophils in a zinc-dependent fashion, analogousto that seen with other cell types; however, the protein with which it interacts isMAC-1 (C3bi receptor; CD11b/CD18) (Wachtfogel et al., 1994). Zinc-depen-dent HK binding to platelets is dependent on interaction with glycoprotein 1b(Bradford et al., 1997; Joseph et al., 1999a). Since both prekallikrein and factor

Table 1 Dissociation Constant and Number of HK-Binding Sites per Cell inHuman Endothelial Cells and Astrocytesa

Cell typeDissociation constant

(Kd, nM)Number of binding sites

per cell (n)

HMVEC-D 1.86 � 0.56 52,833 � 11,121HMVEC-L 4.50 � 1.48 316,306 � 101,031HUVECs 10.35 � 1.02 696,427 � 123,497b

771,666 � 175,000c

Astrocytes 23.73 � 3.61 61,574 � 4887

Abbreviations: HMVEC-D, dermal microvascular endothelial cells; HMVEC-L,lung microvascular endothelial cells; HUVECs, human umbilical vein endothelialcells.

aData represent means � SD, n ¼ 3. The apparent affinity and the number ofbinding sites (n) are calculated by Scatchard analysis of the saturation curve.

bCells attached to 96-well plate.cCells in suspension.

formation of bradykinin 175

XI circulate as complexes with HK, HK–factor XI should bind to HUVECs inan analogous fashion, as does HK–prekallikrein (Shariat-Madar et al., 2001).However, Baird and Walsh (2003) reported that although preformed HK–prekallikrein binds to endothelial cells as a complex, HK–factor XI does not.This is in contrast to studies in which HK is bound, and factor XI is addedseparately. In response, Mahdi et al. agree that HK–prekallikrein predomi-nates, but that HK–factor XI can bind with lower affinity (Mahdi et al., 2003).Binding to platelets differs because factor XI can interact with platelets in theabsence of HK, and there appear to be separate receptors for factor XIand factor XIa (Sinha et al., 1984). Platelets also possess an intrinsic proteinwith factor XI activity that cross-reacts with plasma factor XI immunologically,but differs in molecular weight and isoelectric point (Hsu et al., 1998;Lipscomb and Walsh, 1979). This form of factor XI is present even in patientswho are deficient in plasma factor XI (Tuszynski et al., 1982) and has beenshown to be an alternatively spliced form of factor XI in which one exon ismissing (Hsu et al., 1998). During blood coagulation by the extrinsic (tissuefactor) pathway, these forms of factor XI are more likely activated by thethrombin feedback (Baglia and Walsh, 2000; Baglia et al., 2002; Gailani andBroze, 1991; Naito and Fujikawa, 1991) than by factor XIIa to augment clotformation. This has been shown to occur as a late event within a fibrin matrix(Bouma and Meijers, 2000; Rand et al., 1996). Thus activation on plateletsinvolves factor XII–dependent (Brunnee et al., 1993; Walsh and Griffin, 1981)and independent (thrombin) pathways (von dem Borne et al., 1994), and thefactor XI may be attached to the platelets via HK or by separate receptors. LK,which has a separate light chain from HK, interacts with platelets (as is true ofendothelial cells) and, of necessity, does so solely via domain 3 (Herwald et al.,1995; Jiang et al., 1992a,b).

4. Activation of the Kinin Cascade: The Role of Endothelial Cells

4.1. Activation by Binding to the Cell Surface

We have demonstrated that factor XII can slowly autoactivate when bound toendothelial cells and that addition of kallikrein can digest bound HK to liberatebradykinin at a rate proportional to the kallikrein concentration and with a finalbradykinin level dependent on the amount of bound HK (Nishikawa et al.,1992). Thus, activation of the cascade along the endothelial cell surface islikely; bradykinin is liberated and then interacts with the B2 receptor toincrease vascular permeability. Bradykinin can also stimulate cultured endo-thelial cells to secrete tissue plasminogen activator (Smith et al., 1985), prosta-glandin I2 (prostacyclin), thromboxane A2 (Crutchley et al., 1983; Hong, 1980),

Figure 7 Effect of gC1qR and zinc ion on factor XII–dependent conversion of prekallikrein tokallikrein. Each reaction mixture contained factor XII (1mg/ml), HK (1mg/ml), prekallikrein (1mg/ml), 0.6 mM S2302, and recombinant gC1qR (0–100mg/ml) in HEPES-buffered saline with 50 mMzinc chloride (A) or without zinc chloride (B) at 25 8C. The rate of conversion of prekallikrein tokallikrein was monitored at 405 nm.

176 kusumam joseph and allen p. kaplan

and nitric oxide (Zhao et al., 2001) and can thereby modulate platelet functionand stimulate local fibrinolysis. We next questioned whether factor XII bindingto gC1qR is capable of initiating this cascade. We therefore incubated purifiedfactor XII with a wide concentration of gC1qR (0–100 mg/ml) for a 30-mintime period and prepared replicate samples that were incubated in the ab-sence of zinc ion. As shown in Fig. 7, the rate of prekallikrein conversion tokallikrein increased as the concentration of gC1qR increased (Joseph et al.,2001a,b) and there was no activation if zinc was eliminated from the reactionmixture. Purified cell membrane (native) gC1qR yields a response that isindistinguishable from a recombinant protein, indicating that gC1qR glycosyl-ation does not affect its ‘‘surface’’ properties. If gC1qR is incubated directlywith prekallikrein or with prekallikrein plus HK, there is no conversion ofprekallikrein to kallikrein, again emphasizing the requirement for factor XII.We believe this to be a physiologic phenomenon that is controlled by C1 INHand a2-macroglobulin. This may be one source of the minute quantities offactor XIIa that escape inhibition and that are requisite for contact activation inplasma or during pathologic processes. Other data employing endothelial cellscorroborate the aforementioned effect of gC1qR when endothelial cells areincubated with normal plasma, and the rate of kallikrein formation is com-pared with that seen with plasma deficient in factor XII, prekallikrein, or HK.There was no detectable activation in any plasma except normal plasma (Fig.8A), and the activation was inhibited by antisera to gC1qR and cytokeratin 1(Fig. 8B). When the reaction proceeds beyond 2 hr, the factor XII–deficient

Figure 8 Prekallikrein activation on endothelial cells. (A) Endothelial cells were incubated withnormal, prekallikrein-deficient, factor XII–deficient, or HK-deficient plasmas for 1 hr at 37 8C.After incubation, the cells were washed with HEPES-buffered saline containing 50 mM zincchloride, and prekallikrein activation was monitored by the cleavage of a kallikrein-specificsubstrate, S2302 (0.6 mM), at 405 nm. In (B), endothelial cells were preincubated with antibodiesto cytokeratin 1, gC1qR, or a combination of both for 30 min before addition of normal plasma.

formation of bradykinin 177

plasma activates, but the HK-deficient plasma and prekallikrein-deficientplasma do not, and thus a cell-dependent activation of the prekallikrein inthe presence of HK but the absence of factor FXII appeared possible.

4.2. Factor XII–Independent Activation of thePrekallikrein–HK Complex

Studies have demonstrated that binding of the PK–HK complex to endothelialcells leads to activation in the absence of factor XII (Rojkjaer and Schmaier,1999a,b; Rojkjaer et al., 1998) and that the kallikrein that forms can digest HKto liberate bradykinin and also initiate fibrinolysis (Lin et al., 1997). The latterreaction is dependent on kallikrein activating prourokinase (bound to cellmembrane u-PAR) to urokinase, which in turn converts plasminogen to thefibrinolytic enzyme plasmin. Once such a reaction is set in motion, the additionof factor XII leads to a marked increase in reaction kinetics as a result ofthe conversion of factor XII to factor XIIa by kallikrein. These observationsraise two important questions: (1) What is the nature of the prekallikreinactivator? (2) When factor XII is present (the normal circumstance), isthe cascade initiated by factor XII autoactivation or is the prekallikreinfirst activated by some cell-derived factor and kallikrein then activates thefactor XII?

178 kusumam joseph and allen p. kaplan

We next sought to purify and characterize the cell-derived protein(s) re-sponsible for prekallikrein activation in the absence of factor XII. We firstnoted that this activity was present in both the cell membrane fraction as wellas the cytosol derived from endothelial cells and chose to isolate it from thecytosol. We took advantage of the fact that the prekallikrein-activating moietyappeared to be inhibitable by corn trypsin inhibitor (CTI) (as is factor XIIa);because a CTI affinity column could bind the activity, it was recoverable byeluting the column. A single-step purification, followed by sequence analysis ofsuspect bands seen on SDS gel electrophoresis, ultimately determined thatheat shock protein 90 (Hsp 90) is responsible for the activity seen. Thus whencloned Hsp 90 was incubated with prekallikrein and HK, the prekallikrein wasconverted to kallikrein, and HK was cleaved to liberate bradykinin (Josephet al., 2002a,b; Fig. 9). This is also demonstrable by binding prekallikrein andHK to endothelial cells and assessing the rate of conversion of prekallikrein tokallikrein. Both HK and zinc ion are requisite, and the rate is fast. However,this is in contrast to the slow and factor XII–dependent activation seen whenwhole plasma (Fig. 8) is employed. The reaction is readily demonstrable in thefluid phase as well as by assembly of components along the cell surface;however, it differs strikingly from that seen with factor XIIa. The most criticaldifference is that prekallikrein is not activated unless HK is present. Factor

Figure 9 Prekallikrein activation of Hsp 90. Purified Hsp 90 (2 mg) was incubated withprekallikrein (20 nM), HK (20 nM), zinc (50 mM), and S2302 (0.6 mM), and chromogenic activitywas monitored. Controls were either in the absence of zinc, HK, or both.

formation of bradykinin 179

XIIa readily activates prekallikrein, although the presence of HK does aug-ment the reaction rate. Second, the reaction is stoichiometric, that is, theamount of prekallikrein activated has a 1:1 molar ratio to the amount of Hsp90 present. When we tried to determine the structural features of HK that arerequired, we found that individual heavy and light chains were inactive, andcleaved HK, with bradykinin removed (two-chain HK rather than singlechain), lost about 70% of the activity (Fig. 10). Thus native HK is required.Addition of a peptide that prevents the interaction of prekallikrein withHK also completely inhibits the effect of adding Hsp 90. Hsp 90 is there-fore a stoichiometric activator of the prekallikrein–HK complex and not aprekallikrein activator, as is factor XIIa.

One of the interesting questions we might consider is whether Hsp 90 hasenzymatic activity with the prekallikrein–HK complex as substrate. Hsp 90does have ATPase activity (Richter et al., 2001), but it is not known to be aproteolytic enzyme. The prekallikrein-activating activity can be inhibited bydiisopropylfluorophosphate (DFP), but it has been difficult to characterize theactive site. Although DFP inhibits the reaction, we have been unable toincubate DFP with individual components, dialyze it out, and inhibit thereaction. In fact, if DFP is added to a mixture of Hsp 90, prekallikrein, and

Figure 10 Effect of HK on prekallikrein activation. Cytosol (20 mg) was incubated with 20 nMHK, low molecular weight kininogen (LK), cleaved HK (2C-HK), purified heavy chain of HK (HC-HK), or light chain of HK (LC-HK) in the presence of 20 nM prekallikrein, 50 mM zinc, and0.6 mM S2302. After 2 hr, the chromogenic activity was measured at 405 nm.

180 kusumam joseph and allen p. kaplan

HK, it inhibits conversion of prekallikrein to kallikrein; but if the DFP is thendialyzed out of the mixture, prekallikrein activation then proceeds normally.Thus DFP behaves as a reversible inhibitor instead of as an irreversibleinhibitor and is not phosphorylating an active site serine as is its usual effecton serine proteases. We have considered alternative possibilities, for example,autoactivation of prekallikrein within the prekallikrein–HK complex on addi-tion of Hsp 90, or even that HK becomes an enzyme that converts prekallikreinto kallikrein when Hsp 90 binds.

Other studies have isolated yet another protein with similar functionalcapability. Shariat-Mader et al. isolated a membrane protein that convertsprekallikrein to kallikrein within the prekallikrein–HK complex and identifiedit to be prolylcarboxypeptidase (Motta et al., 2001; Shariat-Madar et al., 2002).This is an exopeptidase that, if its enzymatic capability is relevant, is behavingas an endopeptidase. It is said to be active along the cell surface but not in thefluid phase, which differs from Hsp 90, but its mechanism of action is other-wise strikingly similar. Both require the presence of HK and zinc ion, thereaction in each case is stoichiometric, and each is inhibited by DFP. Althoughthe prolylcarboxypeptidase is assumed to be the enzyme that activates pre-kallikrein within the prekallikrein–HK complex, we suspect that some othermechanism may be operative, perhaps common to both. The prolylcarboxy-peptidase provides an interesting link of the kinin-forming cascade to thebiology of angiotensin, since its function, when originally isolated, was toconvert angiotensin II to angiotensin III, which inactivates it. Thus a moleculethat can generate bradykinin, a vasodilator, inhibits another that is a vasocon-strictor. Hsp 90 is also of particular interest, since this is a protein that isconstitutively present yet upregulated with tissue stress such as hypoxia orduring an inflammatory response.

4.3. Considerations of Activation When Factor XII is Present

Since the endothelial cell participates in the activation of the bradykinin-forming cascade, when all three components are present, factor XII mightbe activated by autoactivation on gC1qR, requiring trace amounts of factorXIIa that is present in plasma, or factor XII might be activated by kallikrein.In the latter scenario, the source of kallikrein would be the stoichiometricinteraction of prekallikrein–HK with Hsp 90 and/or prolylcarboxypeptidase.Formation of factor XIIa then markedly accelerates activation of prekallikrein–HK, since this reaction has typical Michaelis–Menten kinetics. Of course, allthese may be occurring simultaneously, but the evidence thus far suggests thatthe rate of activation of the prekallikrein–HK complex exceeds that of factorXII autoactivation. Thus it is possible that initiation of the cascade on the

formation of bradykinin 181

surface is actually kallikrein, whereas factor XII has the role of an accelerator.Some favor this possibility, employing endothelial cells and purified proteinconstituents (Rojkjaer et al., 1998; Schmaier, 1997, 1998; Schmaier et al.,1999). On a quantitative basis, the cascade remains factor XII dependent.However, the data in Fig. 8 suggest that factor XII may truly initiate whenwhole plasma is studied, that is, minimal dilution in the presence of all theplasma inhibitors.

5. Inhibition of Contact Activation

Regulation of factor XII–dependent pathways occurs by both intrinsic andextrinsic controls. Cleavage of factor XIIa to XIIf (Fig. 1) is one example of anintrinsic control. The factor XIIf produced is not surface bound and is a pooractivator of factor XIa. At the same time, the heavy chain moiety, which has noenzymatic activity, retains the surface-binding site and can compete with factorXII and HK for binding to the surface. Thus, the conversion of factor XIIa tofactor XIIf will reduce the rate of the surface-dependent reactions of coagula-tion, whereas bradykinin generation via fluid-phase activation of prekallikreincontinues. Similarly, digestion of kinin-free HK by factor XIa has beenreported to limit its coagulant activity, (Scott et al., 1985) although, in thiscase, the kinetics appear to be too slow to be of physiologic importance(Reddigari and Kaplan, 1988).

Extrinsic controls are provided by plasma inhibitors for each enzyme.Table 2 indicates the major inhibitors of each active enzyme and, where

Table 2 Plasma Inhibitors of Enzymes of Contact Activation: Relative Percent Contributions toInhibition in Normal Human Plasma

Enzyme

InhibitorFactorXIIa

FactorXIIf Kallikrein

FactorXIaa

C1 inhibitor 91.3 93 52 (84)b 8 (47)Antithrombin IIIc 1.5 4 ND 16 (5)a2-Macroglobulin 4.3 — 35 (16)b —a1-Protease inhibitor — — ND 68 (23.5)a2-Antiplasmin 3.0 3 ND 8c (24.5)

Abbreviation: ND, not determined separately.aData given are from kinetic studies and irreversible complexes formed in plasma are given in

parentheses.bData obtained from generation of kallikrein in situ.cData are for results obtained in the absence of added heparin.

182 kusumam joseph and allen p. kaplan

known, their relative contributions to the total inhibition in plasma. Inhibitionof the contact activation proteases is clearly different from that of the rest ofthe coagulation pathways in that antithrombin III (ATIII) appears to play onlya minor role. Instead, contact activation appears to be limited mainly by C1inhibitor (C1 INH), which is not active against any of the other clotting factorsexcept for inhibition of factor XI. C1 INH is cleaved by the protease it inhibitsand attaches to the active site in a covalent complex. It may remain in a stableform of the acyl enzyme intermediate that characterizes the normal serineprotease mechanism (Travis and Salvesen, 1983). Thus, after a protease hasreacted with C1 INH, it cannot digest protein substrates or hydrolyze smallsynthetic substrates, and the reaction of the active site serine with DFP isabolished.

C1 INH is the only major plasma inhibitor of factor XII and factor XIIf (deAgostini et al., 1984; Forbes et al., 1970; Pixley et al., 1985; Schreiber et al.,1973). Although ATIII can inhibit activated factor XII (Cameron et al., 1989;de Agostini et al., 1984; Stead et al., 1976), its contribution to factor XIIainhibition in plasma is apparently only a few percent of that caused by C1 INH(de Agostini et al., 1984; Pixley et al., 1985). Disagreement exists over theeffect of heparin on the inhibition of activated factor XII by ATIII. Someinvestigators have observed little enhancement of the rate of factor XIIainhibition (Pixley et al., 1991), whereas others have observed a significantincrease (Cameron et al., 1989; Stead et al., 1976). Heparin can act as anactivating surface for contact activation, and factor XII and factor XIIa canbind to it (Hojima et al., 1984; Silverberg and Diehl, 1987). This binding is afactor in the inhibition by ATIII, since inhibition of factor XIIf, which lacks thesurface-binding site, is not augmented in the presence of heparin as much asthat of factor XIIa (Cameron et al., 1989). Curiously, a2-macroglobulin, al-though thought of as a ‘‘universal’’ protease inhibitor (Barrett and Starkey,1973), does not significantly inhibit either form of activated factor XII.

The two major inhibitors of plasma kallikrein are C1 INH and a2-macro-globulin (Gigli et al., 1970; Harpel, 1974; Harpel et al., 1985; McConnell,1972). Together they account for over 90% of the kallikrein inhibitory activityof plasma, with the remainder contributed by ATIII (Schapira et al., 1982c; vander Graaf et al., 1983a). When kallikrein is added to plasma, approximatelyone half is bound to C1 INH and one half to a2-macroglobulin (Harpel et al.,1985). a2-Macroglobulin does not bind to the active site of kallikrein butappears to trap the protease within its structure so as to sterically interferewith its ability to cleave large protein substrates (Barrett and Starkey, 1973).The degree of inhibition is greater than 95%, but the residual activity isdetectable when assayed for lengthy incubation periods. In contrast, digestionof small synthetic substrates is much less affected, and approximately one third

formation of bradykinin 183

of the starting activity is retained. When a surface such as kaolin is added toplasma so that kallikrein is generated in situ, close to 70 or 80% of it is boundto C1 INH (Harpel et al., 1985). The reason for the difference between thepatterns of inhibition of added kallikrein and of endogenously producedkallikrein is unknown. Interestingly, at low temperatures, most of the inhibi-tion of added kallikrein is accounted for by a2-macroglobulin (Harpel et al.,1985); C1 INH appears to be ineffective in the cold (Cameron et al., 1989),and this may underlie the phenomenon of ‘‘cold activation’’ of plasma. Theinhibition of kallikrein by ATIII is also enhanced by heparin (Vennerod et al.,1976) and may therefore become significant in heparinized plasma.

The inhibition profile of factor XI is complicated by the involvement ofseveral factors. In kinetic studies of purified components, a1-antiproteinaseinhibitor (a1-antitrypsin) appears to be the most significant inhibitor of factorXIa (Heck and Kaplan, 1974; Scott et al., 1982b), whereas a1-antitrypsin is nota major inhibitor of other coagulation factors. When the generation of irre-versible enzyme inhibitor complexes was assessed in plasma, however, C1 INHwas found to be the key inhibitor (Wuillemin et al., 1995), with approximatelyequal contributions by a2-antiplasmin and a a1-antiproteinase inhibitor. ATIIIis also an inhibitor of factor XI, with potential for augmentation by heparin,the magnitude of which is unclear (Beeler et al., 1986; Scott et al., 1982a).Physiologic glycosaminoglycans also augment inhibition by C1 INH(Wuillemin et al., 1995, 1996). The combined effects of ATIII and C1 INHmay therefore be most significant at the surfaces, where these substances areplentiful. Finally, platelets secrete protease nexin-2 (a soluble form of amyloidb-protein precursor), which is an efficient but reversible inhibitor in theregulation of factor XIa activity when first generated, although the protectiveeffect of HK against inactivation may also be important (Scandura et al., 1997;Zhang et al., 1997).

The predominant role of C1 INH in the regulation of contact activation inhuman plasma is underscored by the fact that it alone is an efficient inhibitorof activated factor XII, kallikrein, and factor XIa. In plasma from patients withhereditary angioedema (HAE), in which C1 INH is absent, the amount ofdextran sulfate required to produce activation is reduced 10-fold comparedwith normal plasma (Cameron et al., 1989); similar results are obtained in coldplasma. Because some surface was still required for activation under theseconditions, we may surmise that the other inhibitors that are active against thecontact factors do serve to limit their reactions, but that in normal plasma itis inhibition by C1 INH that forms the barrier to the initiation of contactactivation. The plasma concentration of C1 INH is approximately 2 mM, and itis remarkable that its inhibition is ever overcome. That surfaces are able toinduce activation must reflect the protection of the proteases at the surface

184 kusumam joseph and allen p. kaplan

from inhibition. It has also been proposed that kallikrein bound to HK isprotected from inactivation by C1 INH (Schapira et al., 1981, 1982b) anda2-macroglobulin (Schapira et al., 1982b; van der Graaf et al., 1983a,b) andthat factor XIa is similarly protected from a1-antiproteinase inhibitor (Scottet al., 1982b); this mechanism, however, has been ruled out in the case ofkallikrein and C1 INH (Silverberg et al., 1986; van der Graaf et al., 1983a).

6. Inactivation of Bradykinin

Bradykinin is an exceedingly potent vasoactive peptide that can cause venulardilatation, activation of arterial endothelial cells, increased vascular permeabil-ity, hypotension, constriction of uterine and gastrointestinal smooth muscle,constriction of the coronary and pulmonary vasculature, bronchoconstriction,and activation of phospholipase A2 to augment arachidonic acid metabolism.Its regulation is of prime importance, and a variety of enzymes in plasmacontribute to kinin degradation. Carboxypeptidase N (Erdos and Sloane, 1962)removes the C-terminal Arg from bradykinin to leave an octapeptide, des-Arg9

bradykinin (Sheikh and Kaplan, 1986b), which is then digested by angiotensin-converting enzyme (ACE), acting as tripeptidase, to separate the tripeptide,Ser-Pro-Phe, from the pentapeptide Arg-Pro-Pro-Gly-Phe (Sheikh andKaplan, 1986a). Enzymes that have not been characterized rapidly digestSer-Pro-Phe to individual amino acids and more slowly convert the pentapep-tide to Arg-Pro-Pro plus Gly and Phe. The final products of bradykinindegradation are the peptide Arg-Pro-Pro, plus 1 mol each of Gly, Ser, Pro,and Arg, and 2 mol of Phe (Sheikh and Kaplan, 1989a,b). The initial change ofbradykinin to des-Arg9 bradykinin formed by this initial cleavage retains somebut not all the various activities of bradykinin (Marceau and Bachvarov, 1998). Itcan, for example, interact with B1 receptors (Regoli and Barabe, 1980) inducedby inflammation [e.g., interleukin-1 (IL-1) and tumor necrosis factor a (TNF-a)] in the vasculature and cause hypotension, but the activities of bradykinin onthe skin and the contraction of other smooth muscles are abolished. Bradykinininteracts with constitutively expressed B2 receptors to mediate all its functions.Selective B2 and B1 receptor antagonists have been synthesized (Beierwalteset al., 1987; Stewart et al., 1999; Vavrek and Stewart, 1985).

When blood is clotted and serum is studied, all of the reactions for bradyki-nin degradation occur as described, but the rate of initial Arg removal isaccelerated 5-fold compared with plasma (Sheikh and Kaplan, 1989a). Thisis probably due to the action of a plasma carboxypeptidase that is distinct fromcarboxypeptidase N and is expressed (activated) as a result of blood coagula-tion. One such carboxypeptidase is the thrombin-activatable fibrinolysis inhib-itor (TAFI) (Bajzar et al., 1995, 1996). It should also be noted that bradykinin

formation of bradykinin 185

degradation in vivo occurs largely along the pulmonary vasculature and thatendothelial cells there have carboxypeptidase as well as angiotensin-convertingenzyme activities. In the pulmonary circulation, the initial cleavage may occurby angiotensin-converting enzyme acting as a dipeptidase to remove first Phe-Arg and then Ser-Pro (each of which is next cleaved to free amino acids),leaving the pentapeptide Arg-Pro-Pro-Gly-Phe. This is then metabolized fur-ther. The cough, wheeze, and angioedema sometimes associated with use ofACE inhibitors for treatment of hypertension or heart failure is likely due toinhibition of kinin metabolism leading to increased levels of bradykinin(Nussberger et al., 1998). Because bradykinin is a peripheral vasodilator, ithas been considered to be a counterbalance to the vasopressor effects ofangiotensin II. It is clear that the two peptides are also related in terms ofmetabolism, because ACE cleaves His-Leu from the C terminus of angiotensinI, a decapeptide, to leave the octapeptide angiotensin II. Thus, ACE creates avasoconstrictor and inactivates a vasodilator.

7. Relations of the Contact Factors to Other Systems

7.1. Intrinsic Fibrinolytic Cascade

A factor XII–dependent pathway leading to the conversion of plasminogen toplasmin was described in the 1960s and early 1970s (Iatridis and Ferguson,1962; McDonagh and Ferguson, 1970; Ogston et al., 1969), and a defect in thispathway has been observed in plasma deficient in factor XII, prekallikrein, orHK (Colman et al., 1975; Donaldson et al., 1976; Saito et al., 1974; Weiss et al.,1974; Wuepper, 1973; Wuepper et al., 1975). The factor XII–dependentfibrinolytic activity is relatively weak and difficult to demonstrate in wholeplasma, because large quantities of a potent plasminogen activator are notformed. Relatively little plasmin is generated, and this is rapidly inactivated byplasma inhibitors (a2-antiplasmin and a2-macroglobulin). Most studies havetherefore used diluted, acidified plasma (Ogston et al., 1971) or chloroform-treated plasma (Ogston et al., 1969) in which inhibition of contact activationand plasmin is minimized, or have studied a euglobulin preparation thatconcentrates the plasma enzymes and cofactors but limits inhibition (Kluft,1976), or have added organic compounds that destroy a2-antiplasmin and C1INH (Kluft, 1977; Miles et al., 1981, 1983b). Such measures are not needed tostudy blood coagulation or the liberation of bradykinin.

Plasminogen can be activated by kallikrein (Colman, 1969) and factor XIa(Mandle and Kaplan, 1979; Thompson et al., 1977). When purified pre-parations are compared, kallikrein and factor XIa are equipotent as directplasminogen activators (Mandle and Kaplan, 1979). The plasma concentration

186 kusumam joseph and allen p. kaplan

of prekallikrein is approximately 10-fold higher than that of factor XI, however,and, in addition, factor XIIf can readily convert prekallikrein to kallikrein inthe fluid phase (Kaplan and Austen, 1970; Tankersley et al., 1980), whereas ithas minimal activity on factor XI (Kaplan and Austen, 1971). Furthermore,kallikrein can dissociate from surfaces and act in the fluid phase, whereasfactor XIa cannot. For these reasons, kallikrein is more important in thispathway; nevertheless, it is possible to demonstrate a fibrinolytic abnormalityin factor XI–deficient plasma (Saito, 1980).

Activated factor XII (XIIa or XIIf) can also convert plasminogen to plasmin(Goldsmith et al., 1978), but its activity is only 5% that of kallikrein. These areall weak reactions in that the potencies of kallikrein and factor XIa as plasmin-ogen activators are thousands of times lower than that of urokinase (Jorg andBinder, 1985; Mandle and Kaplan, 1979; Miles et al., 1983a). Thus, it can beargued that plasminogen is not a significant substrate for any of them.Although each of these proteins is capable of converting plasminogen toplasmin, the other blood-clotting enzymes—factors IXa, Xa, and VIIa, andthrombin—have no such activity, which argues against this activity being acontingent epiphenomenon.

Later studies of contact-activated fibrinolysis demonstrated that kallikreinactivates the trace quantity of prourokinase in plasma (Hauert et al., 1989;Ichinose et al., 1986; Miles et al., 1981) and that urokinase is the mainplasminogen activator of plasma (Huisveld, 1985). Inhibition by antiurokinaseantisera supports this notion (Miles et al., 1981) as do zymographic gel studiesusing plasma euglobulin preparations (Hauert et al., 1989). Other workershave suggested a role for plasma urokinase in factor XII–independent fibrino-lysis (Kluft et al., 1984). Although urokinase is clearly a much more potentplasminogen activator than any of the enzymes associated with contactactivation, the quantities of urokinase generated are small. If the effects ofa2-antiplasmin and C1 INH are abrogated by addition of flufenamic acidderivatives, contact activation results in the formation of plasmin at approxi-mately 35 ng/ml (Miles et al., 1983b), which represents activation of 0.05%of plasma plasminogen. These observations are shown diagrammatically inFig. 11. Of particular interest are studies in which kinetically favorable prour-okinase activation occurred at the surface of platelets or endothelial cells onaddition of activated factor XII, HK, and prekallikrein (Lenich et al., 1995;Loza et al., 1994). Although the role in vivo of this cascade as a pathway forfibrinolysis is not yet clear, patients with abnormalities of contact activa-tion proteins such as factor XII have died of thrombosis. It is now feltthat a role for contact activation in fibrinolysis is more important physio-logically than any role it has in blood coagulation (hemostasis), sincefactor-deficient patients do not bleed (Kaplan et al., 2002). Furthermore,

Figure 11 Pathways by which plasminogen is converted to the fibrinolytic enzyme plasmin. Themajor activation pathway is dependent on kallikrein conversion of prourokinase to urokinase.However, kallikrein, factor XIa, and factor XIIa are all capable of directly converting plasminogento plasmin.

formation of bradykinin 187

factor XI is considered to be a contributor to the extrinsic (tissue factor)coagulation pathway via the thrombin feedback.

An additional interaction between the kallikrein–kinin system andin vivo fibrinolysis is suggested by the observation that bradykinin is a potentstimulator of the release of tissue plasminogen activator (TPA) from endothelialcells (Smith et al., 1985).

7.2. Interaction with Other Plasma Proteases

Factor XIIf has been reported to activate factor VII (Radcliffe et al., 1977;Seligsohn et al., 1979) and thereby initiate the extrinsic coagulation pathway.This reaction contributes significantly to the kaolin-activated partial thrombo-plastin time (PTT) as usually seen when plasma is exposed to the cold(Gjonnaess, 1972; Laake et al., 1974). This is accentuated in women who useoral contraceptives containing estrogen, apparently owing to an increasedconcentration of factor XII (Gordon et al., 1980, 1983; Jespersen and Kluft,1985; Laake et al., 1974). It is theorized that this pathway might contributeto the increased incidence of thrombosis reported as a complication of oralcontraceptive use.

Factor XIIf (but not factor XIIa) can enzymatically activate the first compo-nent of complement when it is incubated with purified C1 or added to plasma

188 kusumam joseph and allen p. kaplan

(Ghebrehiwet et al., 1981). C1 activation is due to cleavage of the C1rsubcomponent (Ghebrehiwet et al., 1983) by factor XIIf. Little complementactivation is seen when kaolin is incubated with whole plasma, and significantcomplement activation may be seen only under conditions that result insubstantial conversion of factor XIIa to factor XIIf. Once such circumstanceis C1 INH deficiency (i.e., HAE), in which factor XII activation may contributeto complement consumption (Donaldson, 1968; Fields et al., 1983). Kallikreincan also cleave C1 subcomponents, but the net result is destruction rather thanactivation. On the other hand, kallikrein can activate factor B of the alternativecomplement pathway and thereby substitute for factor D (DiScipio, 1982).

7.3. Interaction with Leukocytes

Kallikrein has been reported to interact with human leukocytes in a variety ofways. It is a chemotactic factor for neutrophils (Kaplan et al., 1972) andmonocytes (Gallin and Kaplan, 1974), and it has been shown to cause neutro-phil aggregation (Schapira et al., 1982a) and release of elastase (Klempereret al., 1968). In a rabbit model, kallikrein stimulation of chemotaxis appearedto require cleavage of C5 and release of C5a chemotactic factor (Donaldsonet al., 1977). Therefore C5 bound to the surface of neutrophils can possibly becleaved in the aforementioned reactions. Anti-kallikrein serum was inhibitory,whereas anti-C5 serum had no effect; the authors therefore concluded that theeffect of kallikrein on human neutrophils does not require complement.Furthermore, a degraded form of kallikrein (b-kallikrein), in which theheavy chain is partially digested, is enzymatically active on kininogen to formkinin but possesses a markedly attenuated reactivity with neutrophils (Colmanet al., 1985). Factor XIIa has also been shown to stimulate neutrophils; becausefactor XIIf did not do this, a requirement for a binding site in the heavy chainwas inferred (Wachtfogel et al., 1986). No studies have demonstrated a cellsurface receptor for these enzymes or required cleavage of surface compo-nents. In each instance, the active site of the enzyme is required, and theproenzyme or DFP-treated enzyme is inactive (Kaplan et al., 1972; Schapiraet al., 1982a).

8. Considerations in Human Diseases

8.1. C1 Inhibitor Deficiency

Although C1 INH was defined as an inhibitor of the activated first componentof complement, it is clearly a key control protein of the plasma kinin-formingcascade. The pathogenesis of the swelling in C1 INH deficiency is dependent

formation of bradykinin 189

on the plasma kinin-forming pathway rather than complement; however, it isgermane to review the history of the complement data and to point out how itwas first thought to be the key, and then to present the more recent data thatsuggest otherwise. Intracutaneous injection of C1 into normal individuals wasreported to cause the formation of a small wheal reaction, whereas injectioninto patients with hereditary angioedema yields localized angioedema, that is,an augmented response because of low C1 inhibitor (Klemperer et al., 1968).A kinin-like peptide was isolated from such patients and its formationappeared to be inhibited in C2-deficient plasma. Thus C2 was considered tobe the source of the pathogenic peptide (Donaldson, 1968). However, directdemonstration of such a kinin-like peptide on interaction of activated C1 andC4 and C2 or with C2 alone is lacking. Although it was originally reported thatcleavage of C2b by plasmin generates a kinin (Donaldson et al., 1977),attempts to confirm this experiment have all failed (Colman et al., 1985;Fields et al., 1983). The only identifiable kinin seen in subsequent studieswas bradykinin (Fields et al., 1983). On the other hand, the amino acidsequence of C2b is known, and Strang et al. (1988) synthesized peptides ofvarious lengths and tested each for kinin-like activity. One such peptide wasshown to cause edema when injected intracutaneously, reminiscent of the C2kinin originally described. However, this peptide has not been shown to be acleavage product of C2b, nor has it been shown to be present during attacks ofswelling in patients with hereditary angioedema. Thus, at this point it seemsunlikely that a kinin-like molecule is derived from C2b as a result of enzymaticcleavage. On the other hand, the presence of bradykinin has been documentedas described below, and it is the likely cause of the swelling. In fact, when oneof the proponents of the C2 kinin reexamined kinin formation in the plasma ofpatients with hereditary angioedema, only bradykinin was found (Shoemakeret al., 1994).

It should be noted that 24-hr urine histamine excretion may also beincreased during attacks of angioedema, suggesting that C3a, C4a, or C5ais being generated. Although the plasma levels of C3 and C5 are normalin this disorder, C3 turnover is clearly enhanced (Carpenter et al., 1969).The lesions, however, are not pruritic, and antihistaminics have no effect onthe clinical course of the disease. Thus, complement activation is un-doubtedly occurring, perhaps even during quiescent periods, to lead to a lowlevel of C4, but the vasoactive consequences of augmented complementactivation that occurs during attacks of HAE do not appear to be the causeof the swelling.

C1 inhibitor inhibits all functions of factor XIIa (Gigli et al., 1970; Schreiberet al., 1973) and is one of the two major plasma kallikrein inhibitors, the otherbeing a2-macroblobulin (Harpel et al., 1985), and all functions of kallikrein are

190 kusumam joseph and allen p. kaplan

thereby inhibited including the feedback activation of factor XII, the cleavageof HK, and the activation of plasma prourokinase (Ichinose et al., 1986) to leadto plasmin formation (Fig. 1). C1 inhibitor also inhibits the fibrinolytic enzymeplasmin, although it is a relatively minor inhibitor compared with a2-antiplas-min or a2-macroglobulin. Patients with hereditary angioedema appear to behyperresponsive to cutaneous injections of kallikrein, as they are to C1 (Juhlinand Michaelsson, 1969), and elevated levels of bradykinin and cleaved kinino-gen have been observed during attacks of swelling (Nussberger et al., 1998;Talamo et al., 1969). There is also evidence that C1 activation observed inhereditary angioedema may also be factor XII dependent (Donaldson, 1968).Thus, a factor XII–dependent enzyme may be initiating the classic comple-ment cascade. Plasmin is capable of activating C1s and may represent one suchenzyme (Ratnoff and Naff, 1967). Ghebrehiwet et al. demonstrated thatHageman factor fragment (factor XIIf) can directly activate the classic com-plement cascade by activating C1r and to a lesser degree C1s (Ghebrehiwetet al., 1981, 1983). This may represent a critical link between the intrinsiccoagulation–kinin cascade and complement activation (Fig. 1). The presenceof kallikrein-like activity in induced blisters of patients with hereditary angioe-dema supports this notion (Curd et al., 1980), as does the progressive genera-tion of bradykinin on incubation of hereditary angioedema plasma in plasticnon-contact-activated test tubes (Fields et al., 1983), as well as the low pre-kallikrein HK levels seen during attacks (Schapira et al., 1983). More recentdata support these indirect observations, favoring bradykinin as the criticalpathogenic peptide for hereditary angioedema and—likely—acquired C1 INHdeficiency as well. One unique family has been described in which there is apoint mutation in C1 INH (Ala443 ! Val) leading to inability to inhibit thecomplement cascade but normal inhibition of factor XIIa and kallikrein(Zahedi et al., 1995, 1997). No family member of this type II mutation hashad angioedema. Plasma bradykinin levels have been shown to be elevatedduring attacks of swelling of hereditary and acquired forms of C1 INHdeficiency (Cugno et al., 1996; Nussberger et al., 1998), and local bradykiningeneration has been documented at the site of the swelling (Nussberger et al.,1999).

The role of fibrinolysis also needs to be considered a part of the pathogene-sis of the disease, since antifibrinolytic agents such as e-aminocaproic acid andtranexamic acid appear to be efficacious (Frank et al., 1972; Lundh et al., 1968;Sheffer et al., 1972), and plasmin is generated during active disease (Cugnoet al., 1993). Although kallikrein factor XIa, and even factor XIIa, have someability to activate plasminogen directly, the plasma pathway via the prouroki-nase intermediate appears to be the major factor XII–dependent fibrinolyticmechanism (Fig. 11). Among the functions of plasmin are the activation of

formation of bradykinin 191

C1s, the ability to cleave and activate factor XII just as kallikrein can (Kaplanand Austen, 1971), and digestion of C1 inhibitor (Wallace et al., 1997). Each ofthese would serve to augment bradykinin formation and further deplete thelevels of C1 inhibitor. Thus, the formation of plasmin may in this fashioncontribute to the pathogenesis of the disease.

8.2. Contact Activation in Allergic Diseases

By analogy with observations based on dextran sulfate, naturally occurringglycosaminoglycans or proteoglycans may be able to induce contact activation.We have tested heparin proteoglycan from the Furth murine mastocytoma forits ability to activate a mixture of factor XII and prekallikrein (Brunnee et al.,1993, 1997). There is progressive conversion of prekallikrein to kallikrein asthe concentration of mast cell heparin is increased. The potency of heparinproteoglycan equals that of dextran sulfate, and its activity is inhibited byheparinase I or II, but not by heparitinase or chondroitinase ABC. Of theglycosaminoglycans we have tested, heparin, dermatan sulfate, keratin poly-sulfate, and chondroitin sulfate C are positive in the assay (in that order),whereas heparin sulfate and chondroitin sulfate A are negative. Collagen typesI, III, IV, and V; laminin; fibronectin; and vitronection are also negative.Activation can then occur by release of heparin and/or other mucopolysacchar-ides secreted by mast cells and basophils on exposure to plasma proteins or viainteraction of these proteins with exposed connective tissue proteoglycansduring tissue injury. The proteins of the kinin-forming system have beenshown to be present in interstitial fluid of rabbit skin; thus, the source maynot solely be dependent on exudation and activation of plasma.

Any aspect of inflammation that leads to dilution of plasma constituents orexclusion of inhibitors will augment contact activation, because inhibitoryfunctions are dependent on concentration. Thus, the activatability of plasmacan be shown to be related directly to dilution. Once levels of C1 INH are lessthan 25% of normal (i.e., equivalent to a 1:4 dilution), patients with HAE areprone to attacks of swelling.

Activation of the plasma and tissue kinin-forming systems has been observedin allergic reactions in the nose, lungs, and skin, and include the immediatereaction as well as the late-phase reaction, although the contributions ofthe plasma and tissue kallikrein pathways to each aspect of allergic inflamma-tion are likely quite different. Antigen challenge of the nose followedby nasal lavage revealed an increase in tosyl-l-arginine-O-methyl ester(TAME) esterase activity, which is largely attributable to kallikrein(s) (Proudet al., 1983). The activation was seen during the immediate response as wellas during the late-phase reaction (Creticos et al., 1984). Both LK and HK

192 kusumam joseph and allen p. kaplan

were shown to be present in nasal lavage fluid (Baumgarten et al., 1985),and fractionation of nasal washings demonstrated evidence of both tissuekallikrein (Baumgarten et al., 1986b) and plasma kallikrein (Baumgartenet al., 1986a). Tissue kallikrein can be secreted by glandular tissue as wellas by infiltrating cells, such as neutrophils, and will cleave LK to yield kallidin.Plasma kallikrein will digest HK to yield bradykinin directly. HPLC analysisof kinins in nasal washings revealed both kallidin (lysylbradykinin) and brady-kinin. The latter can be formed from kallidin by aminopeptidase action;however, a portion of the bradykinin is also likely the direct result of plasmakallikrein activity.

Studies of the allergen-induced late-phase reactions in the skin (Atkins et al.,1987, 1992) have demonstrated the presence of kallikrein–C1 INH and acti-vated factor XII–C1 INH complexes in induced blisters observed during an8-hr period. Elevated levels of these complexes were seen between 3 and 6 hr,coincident with the late-phase response and were specific for the antigen towhich the patient was sensitive.

8.3. Regulation of Blood Pressure

The possible relationship of the contact system to blood pressure regulation isan intriguing question. As already seen, ACE creates the hypertensive peptide,angiotensin II, and plays a major role in inactivating the hypotensive productof contact activation, bradykinin. An unfortunate circumstance has dramatizedthe kinin-forming capacity of factor XIIf; trauma patients given plasma proteinfractions as plasma expanders that were contaminated with factor XIIf showedprofound hypotension (Alving et al., 1978, 1980). The mechanism by whichinfused factor XIIf causes hypotension has been demonstrated to be due tobradykinin formation. A more tenuous connection exists between blood pres-sure regulation and the contact system in the factor XII–dependent activationof plasma prorenin. Prorenin is activated to renin by cold treatment of plasmaor acidification to pH 3.3. With acid treatment, most of the renin activity isproduced after reneutralization; this alkaline phase activation is mediated bykallikrein (Derkx et al., 1979; Sealey et al., 1979), as is the cold-inducedactivation (Brown and Osmond, 1984). Kallikrein is able to activate purifiedprorenin (Yokosawa et al., 1979), but when added to plasma, it does not causeprorenin activation in the absence of an acidification step. Although it has beensupposed that the acid treatment serves to destroy kallikrein inhibitors, pro-renin is not activated in plasma deficient in C1 INH or a2-macroglobulin(Purdon et al., 1985). Thus, some other unknown event must occur onacidification. The physiologic significance of this reaction is uncertain.

formation of bradykinin 193

8.4. Kinin in Vascular Disease and Blood Pressure Control

The bradykinin-forming cascade may have a role in cardiovascular diseasesincluding hypertension and diabetes. Numerous interactions exist between theangiotensin cascade, vascular endothelium, and the plasma kinin-formingcascade, as indicated in Fig. 12. Prorenin can be converted to renin by plasmakallikrein in an acidic milieu and is factor XII dependent (Alving et al., 1978,1980). Once there is conversion of angiotensin I to angiotensin II, the interac-tion of angiotensin II with the AT1 receptor on endothelial cells activatesNADPH oxidase and P42/44 MAP kinase (Lu et al., 1998), leading to upregu-lation of the B2 receptor. This may also lead to augmented liberation of Hsp 90(Joseph et al., 2002b) and prolylcarboxypeptidase (Shariat-Madar et al., 2002),each of which interacts with the prekallikrein–HK complex to generate brady-kinin. The interaction of bradykinin with the endothelial cell stimulates theformation of NO (Zhao et al., 2001) and prostacyclin (Crutchley et al., 1983;Hong, 1980). Angiotensin-converting enzyme, expressed on endothelial cells(Caldwell et al., 1976) and present in plasma as kininase II (Yang and Erdos,1967), is responsible for conversion of angiotensin I to angiotensin II and also

Figure 12 Diagrammatic representation of the many interconnections between the renin–angiotensin pathway and the bradykinin-forming pathway.

194 kusumam joseph and allen p. kaplan

degrades bradykinin by first cleaving Phe-Arg from the C terminus, followedby Ser-Pro. Thus angiotensin I, a vasoconstrictor, balances the effects ofbradykinin, NO, and prostacyclin as vasodilators. Of particular interest regard-ing the pathogenesis of diabetic vasculopathy is the ability of glucose toupregulate the B2 receptor (Tan et al., 2004) and to raise the levels of plasmaprekallikrein (Jaffa et al., 2003). The latter effect appears specific, because nosignificant change in factor XII or kininogen levels is noted.

8.5. Other Disorders

Endotoxic shock is associated with depletion of contact activation proteins(Hirsch et al., 1974; Mason et al., 1970; O’Donnell et al., 1976; Robinsonet al., 1975), and serial HK levels have prognostic value because a drop tonear zero usually indicates a fatal outcome, as do lower prekallikrein levels(O’Donnell et al., 1976). A monoclonal antibody to factor XII markedly dimin-ished the mortality by 50% in a baboon model of endotoxic shock (Pixley et al.,1992, 1993), largely due to effects on hypotension and its sequelae. Parametersof disseminated intravascular coagulation (DIC) were unaffected and likelymediated via tissue thromboplastin, although DIC due to endothelial cellinjury and/or endotoxemia is associated with diminished levels of factor XII,prekallikrein, and kallikrein-inhibiting activity.

The synovial fluid of patients with rheumatoid arthritis has been shown tocontain plasma kallikrein, which can activate stromelysin and convert procol-lagenase to collagenase (Nagase et al., 1982). Uric acid and pyrophosphatecrystals can act as surfaces for contact activation (Ginsberg et al., 1980;Kellermeyer and Breckenridge, 1965) and may contribute to the inflammationseen in gout or pseudogout. However, at least one case of gout (Londino andLuparello, 1984) and one of rheumatoid arthritis (Donaldson et al., 1972) havebeen reported in factor XII–deficient subjects.

Pancreatitis, particularly acute hemorrhagic pancreatitis, is associatedwith release of large quantities of tissue kallikrein; thus, kallidin and/or brady-kinin may contribute to the pooling of fluid within the abdominal cavity andhypotension that can result.

The causes of Alzhemer’s disease are not known, although it is associatedwith deposition of b-amyloid protein in the form of plaques as well as � fibrilproteins within neurofibrillary tangles and paired helical filaments. A rarehereditary form of Alzheimer’s disease has been associated with a mutationof the amyloid precursor protein. We have demonstrated that when b-amyloidmonomer aggregates as it does within plaques, it is a potent initiator ofthe plasma kinin-forming cascade. It does so by binding factor XII andHK (Shibayama et al., 1999) and in so doing activates the cascade in a

formation of bradykinin 195

zinc-dependent reaction. Furthermore, the aggregation of b-amyloid has beenshown to be zinc dependent (Bush et al., 1994). In this fashion, b-amyloidresembles the binding we see to cell membranes, and to gC1qR specifically,because the latter protein also activates the cascade. However, activationby aggregated b-amyloid is more rapid than that seen with gC1qR and mayhave features that are reminiscent of both negatively charged surfaces (whichare ion independent) and cell surface initiators. Whether any of the func-tional disturbances of neurons or glial cells seen in Alzheimer’s disease areattributable to the generation of bradykinin remains to be established.

References

Alving, B. M., Hojima, Y., Pisano, J. J., Mason, B. L., Buckingham, R. E., Jr., Mozen, M. M., andFinlayson, J. S. (1978). Hypotension associated with prekallikrein activator (Hageman-factorfragments) in plasma protein fraction. N. Engl. J. Med. 299, 66–70.

Alving, B. M., Tankersley, D. L., Mason, B. L., Rossi, F., Aronson, D. L., and Finlayson, J. S.(1980). Contact-activated factors: Contaminants of immunoglobulins preparations with coagu-lant and vasoactive properties. J. Lab. Clin. Med. 96, 334–346.

Atkins, P. C., Miragliotta, G., Talbot, S. F., Zweiman, B., and Kaplan, A. P. (1987). Activation ofplasma Hageman factor and kallikrein in ongoing allergic reactions in the skin. J. Immunol. 139,2744–2748.

Atkins, P. C., Kaplan, A. P., von Allmen, C., Moskovitz, A., and Zweiman, B. (1992). Activation ofthe coagulation pathway during ongoing allergic cutaneous reactions in humans. J. Allergy Clin.Immunol. 89, 552–559.

Baglia, F. A., and Walsh, P. N. (2000). Thrombin-mediated feedback activation of factor XI on theactivated platelet surface is preferred over contact activation by factor XIIa or factor XIa. J. Biol.Chem. 275, 20514–20519.

Baglia, F. A., Badellino, K. O., Li, C. Q., Lopez, J. A., and Walsh, P. N. (2002). Factor XI binding tothe platelet glycoprotein Ib–IX–V complex promotes factor XI activation by thrombin. J. Biol.Chem. 277, 1662–1668.

Baird, T. R., and Walsh, P. N. (2002). Activated platelets but not endothelial cells participatein the initiation of the consolidation phase of blood coagulation. J. Biol. Chem. 277,28498–28503.

Baird, T. R., and Walsh, P. N. (2003). Factor XI, but not prekallikrein, blocks high molecular weightkininogen binding to human umbilical vein endothelial cells. J. Biol. Chem. 278, 20618–20623.

Bajzar, L., Manuel, R., and Nesheim, M. E. (1995). Purification and characterization of TAFI, athrombin-activable fibrinolysis inhibitor. J. Biol. Chem. 270, 14477–14484.

Bajzar, L., Morser, J., and Nesheim, M. (1996). TAFI, or plasma procarboxypeptidase B, couplesthe coagulation and fibrinolytic cascades through the thrombin–thrombomodulin complex.J. Biol. Chem. 271, 16603–16608.

Barrett, A. J., and Starkey, P. M. (1973). The interaction of a2-macroglobulin with proteinases:Characteristics and specificity of the reaction, and a hypothesis concerning its molecularmechanism. Biochem. J. 133, 709–724.

Baumgarten, C. R., Togias, A. G., Naclerio, R. M., Lichtenstein, L. M., Norman, P. S., and Proud, D.(1985). Influx of kininogens into nasal secretions after antigen challenge of allergic individuals.J. Clin. Invest. 76, 191–197.

196 kusumam joseph and allen p. kaplan

Baumgarten, C. R., Nichols, R. C., Naclerio, R. M., Lichtenstein, L. M., Norman, P. S., andProud, D. (1986a). Plasma kallikrein during experimentally induced allergic rhinitis: Role inkinin formation and contribution to TAME-esterase activity in nasal secretions. J. Immunol.137, 977–982.

Baumgarten, C. R., Nichols, R. C., Naclerio, R. M., and Proud, D. (1986b). Concentrations ofglandular kallikrein in human nasal secretions increase during experimentally induced allergicrhinitis. J. Immunol. 137, 1323–1328.

Beeler, D. L., Marcum, J. A., Schiffman, S., and Rosenberg, R. D. (1986). Interaction of factor XIaand antithrombin in the presence and absence of heparin. Blood 67, 1488–1492.

Beierwaltes, W. H., Carretero, O. A., Scicli, A. G., Vavrek, R. J., and Stewart, J. M. (1987).Competitive analog antagonists of bradykinin in the canine hindlimb. Proc. Soc. Exp. Biol.Med. 186, 79–83.

Bouma, B. N., and Meijers, J. C. (2000). Role of blood coagulation factor XI in downregulation offibrinolysis. Curr. Opin. Hematol. 7, 266–272.

Bradford, H. N., Dela Cadena, R. A., Kunapuli, S. P., Dong, J. F., Lopez, J. A., and Colman, R. W.(1997). Human kininogens regulate thrombin binding to platelets through the glycoproteinIb–IX–V complex. Blood 90, 1508–1515.

Brown, C. F., and Osmond, D. H. (1984). Factor XII and prekallikrein–kallikrein–kinin in thecryoactivation of human plasma prorenin. Clin. Sci. 66, 533–539.

Brunnee, T., La Porta, C., Reddigari, S. R., Salerno, V. M., Kaplan, A. P., and Silverberg, M. (1993).Activation of factor XI in plasma is dependent on factor XII [see comment]. Blood 81, 580–586.

Brunnee, T., Reddigari, S. R., Shibayama, Y., Kaplan, A. P., and Silverberg, M. (1997). Mast cellderived heparin activates the contact system: A link to kinin generation in allergic reactions.Clin. Exp. Allergy 27, 653–663.

Bush, A. I., Pettingell, W. H., Multhaup, G., d Paradis, M., Vonsattel, J. P., Gusella, J. F.,Beyreuther, K., Masters, C. L., and Tanzi, R. E. (1994). Rapid induction of Alzheimer A b

amyloid formation by zinc [see comment]. Science 265, 1464–1467.Caldwell, P. R., Seegal, B. C., Hsu, K. C., Das, M., and Soffer, R. L. (1976). Angiotensin-converting

enzyme: Vascular endothelial localization. Science 191, 1050–1051.Cameron, C. L., Fisslthaler, B., Sherman, A., Reddigari, S., and Silverberg, M. (1989). Studies on

contact activation: Effects of surface and inhibitors. Med. Prog. Technol. 15, 53–62.Carpenter, C. B., Ruddy, S., Shehadeh, I. H., Muller-Eberhard, H. J., Merrill, J. P., and Austen,

K. F. (1969). Complement metabolism in man: Hypercatabolism of the fourth (C4) and third(C3) components in patients with renal allograft rejection and hereditary, angioedema (HAE).J. Clin. Invest. 48, 1495–1505.

Cochrane, C. G., and Revak, S. D. (1980). Dissemination of contact activation in plasma by plasmakallikrein. J. Exp. Med. 152, 608–619.

Cochrane, C. G., Revak, S. D., and Wuepper, K. D. (1973). Activation of Hageman factor in solidand fluid phases: A critical role of kallikrein. J. Exp. Med. 138, 1564–1583.

Colman, R. W. (1969). Activation of plasminogen by human plasma kallikrein. Biochem. Biophys.Res. Commun. 35, 273–279.

Colman, R. W., and Schmaier, A. H. (1997). Contact system: A vascular biology modulator withanticoagulant, profibrinolytic, antiadhesive, and proinflammatory attributes. Blood 90,3819–3843.

Colman, R. W., Bagdasarian, A., Talamo, R. C., Scott, C. F., Seavey, M., Guimaraes, J. A., Pierce,J. V., and Kaplan, A. P. (1975). Williams trait: Human kininogen deficiency with diminishedlevels of plasminogen proactivator and prekallikrein associated with abnormalities of the Hage-man factor-dependent pathways. J. Clin. Invest. 56, 1650–1662.

formation of bradykinin 197

Colman, R. W., Wachtfogel, Y. T., Kucich, U., Weinbaum, G., Hahn, S., Pixley, R. A., Scott, C. F.,de Agostini, A., Burger, D., and Schapira, M. (1985). Effect of cleavage of the heavy chain ofhuman plasma kallikrein on its functional properties. Blood 65, 311–318.

Colman, R. W., Pixley, R. A., Najamunnisa, S., Yan, W., Wang, J., Mazar, A., and McCrae, K. R.(1997). Binding of high molecular weight kininogen to human endothelial cells is mediated via asite within domains 2 and 3 of the urokinase receptor. J. Clin. Invest. 100, 1481–1487.

Creticos, P. S., Peters, S. P., Adkinson, N. F., Jr., Naclerio, R. M., Hayes, E. C., Norman, P. S., andLichtenstein, L. M. (1984). Peptide leukotriene release after antigen challenge in patientssensitive to ragweed. N. Engl. J. Med. 310, 1626–1630.

Crutchley, D. J., Ryan, J. W., Ryan, U. S., and Fisher, G. H. (1983). Bradykinin-induced release ofprostacyclin and thromboxanes from bovine pulmonary artery endothelial cells: Studies withlower homologs and calcium antagonists. Biochim. Biophys. Acta 751, 99–107.

Cugno, M., Hack, C. E., de Boer, J. P., Eerenberg, A. J., Agostoni, A., and Cicardi, M. (1993).Generation of plasmin during acute attacks of hereditary angioedema. J. Lab. Clin. Med. 121,38–43.

Cugno, M., Cicardi, M., Coppola, R., and Agostoni, A. (1996). Activation of factor XII and cleavageof high molecular weight kininogen during acute attacks in hereditary and acquired C1-inhibitordeficiencies. Immunopharmacology 33, 361–364.

Curd, J. G., Prograis, L. J., Jr., and Cochrane, C. G. (1980). Detection of active kallikrein ininduced blister fluids of hereditary angioedema patients. J. Exp. Med. 152, 742–747.

de Agostini, A., Lijnen, H. R., Pixley, R. A., Colman, R. W., and Schapira, M. (1984). Inactivation offactor XII active fragment in normal plasma. Predominant role of C-1-inhibitor. J. Clin. Invest.73, 1542–1549.

Dedio, J., and Muller-Esterl, W. (1996). Kininogen binding protein p33/gC1qR is localized in thevesicular fraction of endothelial cells. FEBS Lett. 399, 255–258.

Dedio, J., Jahnen-Dechent, W., Bachmann, M., and Muller-Esterl, W. (1998). The multiligand-binding protein gC1qR, putative C1q receptor, is a mitochondrial protein. J. Immunol. 160,3534–3542.

Derkx, F. H., Bouma, B. N., Schalekamp, M. P., and Schalekamp, M. A. (1979). An intrinsic factorXII–prekallikrein-dependent pathway activates the human plasma renin–angiotensin system.Nature 280, 315–316.

DiScipio, R. G. (1982). The activation of the alternative pathway C3 convertase by human plasmakallikrein. Immunology 45, 587–595.

Donaldson, V. H. (1968). Mechanisms of activation of C01 esterase in hereditary angioneuroticedema plasma in vitro. J. Exp. Med. 127, 411–429.

Donaldson, V. H., Glueck, H. I., and Fleming, T. (1972). Brief recordings: Rheumatoid arthritis ina patient with Hageman trait. N. Engl. J. Med. 286, 528–530.

Donaldson, V. H., Glueck, H. I., Miller, M. A., Movat, H. Z., and Habal, F. (1976). Kininogendeficiency in Fitzgerald trait: Role of high molecular weight kininogen in clotting and fibrinoly-sis. J. Lab. Clin. Med. 87, 327–337.

Donaldson, V. H., Rosen, F. S., and Bing, D. H. (1977). Role of the second component ofcomplement (C2) and plasmin in kinin release in hereditary angioneurotic edema (H.A.N.E.)plasma. Trans. Assoc. Am. Physicians 90, 174–183.

Dunn, J. T., and Kaplan, A. P. (1982). Formation and structure of human Hageman factorfragments. J. Clin. Invest. 70, 627–631.

Dunn, J. T., Silverberg, M., and Kaplan, A. P. (1982). The cleavage and formation of activatedhuman Hageman factor by autodigestion and by kallikrein. J. Biol. Chem. 257, 1779–1784.

Erdos, E. G., and Sloane, G. M. (1962). An enzyme in human plasma that inactivates bradykininand kallidins. Biochem. Pharmacol. 11, 585.

198 kusumam joseph and allen p. kaplan

Fernando, L. P., Fernando, A. N., Joseph, K., and Kaplan, A. P. (2003a). Assessment of the role ofheparan sulfate in high molecular weight kininogen binding to human umbilical vein endothelialcells. J. Thromb. Haemost. 1, 2444–2449.

Fernando, L. P., Natesan, S., Joseph, K., and Kaplan, A. P. (2003b). High molecular weightkininogen and factor XII binding to endothelial cells and astrocytes. Thromb. Haemost. 90,787–795.

Fields, T., Ghebrehiwet, B., and Kaplan, A. P. (1983). Kinin formation in hereditary angioedemaplasma: Evidence against kinin derivation from C2 and in support of ‘‘spontaneous’’ formation ofbradykinin J. Allergy Clin. Immunol. 72, 54–60.

Forbes, C. D., Pensky, J., and Ratnoff, O. D. (1970). Inhibition of activated Hageman factor andactivated plasma thromboplastin antecedent by purified serum C1 inactivator. J. Lab. Clin. Med.76, 809–815.

Frank, M. M., Sergent, J. S., Kane, M. A., and Alling, D. W. (1972). e-Aminocaproic acid therapy ofhereditary angioneurotic edema: A double-blind study. N. Engl. J. Med. 286, 808–812.

Gailani, D., and Broze, G. J., Jr. (1991). Factor XI activation in a revised model of bloodcoagulation. Science 253, 909–912.

Gallin, J. I., and Kaplan, A. P. (1974). Mononuclear cell chemotactic activity of kallikrein andplasminogen activator and its inhibition by C1 inhibitor and a2-macroglobulin. J. Immunol. 113,1928–1934.

Ghebrehiwet, B., Silverberg, M., and Kaplan, A. P. (1981). Activation of the classical pathway ofcomplement by Hageman factor fragment. J. Exp. Med. 153, 665–676.

Ghebrehiwet, B., Randazzo, B. P., Dunn, J. T., Silverberg, M., and Kaplan, A. P. (1983). Mechan-isms of activation of the classical pathway of complement by Hageman factor fragment. J. Clin.Invest. 71, 1450–1456.

Ghebrehiwet, B., Lim, B. L., Peerschke, E. I., Willis, A. C., and Reid, K. B. (1994). Isolation,cDNA cloning, and overexpression of a 33-kD cell surface glycoprotein that binds to the globular‘‘heads’’ of C1q. J. Exp. Med. 179, 1809–1821.

Gigli, I., Mason, J. W., Colman, R. W., and Austen, K. F. (1970). Interaction of plasma kallikreinwith the C1 inhibitor. J. Immunol. 104, 574–581.

Ginsberg, M. H., Jaques, B., Cochrane, C. G., and Griffin, J. H. (1980). Urate crystal-dependentcleavage of Hageman factor in human plasma and synovial fluid. J. Lab. Clin. Med. 95, 497–506.

Gjonnaess, H. (1972). Cold promoted activation of factor VII. 3. Relation to the kallikrein system.Thromb. Diathesis Haemorrhagica 28, 182–193.

Goldsmith, G., Saito, H., and Ratnoff, O. D. (1978). The activation of plasminogen by Hagemanfactor (factor XII) and Hageman factor fragments. J. Clin. Invest. 12, 54.

Gordon, E. M., Ratnoff, O. D., Saito, H., Donaldson, V. H., Pensky, J., and Jones, P. K. (1980).Rapid fibrinolysis, augmented Hageman factor (factor XII) titers, and decreased C1 esteraseinhibitor titers in women taking oral contraceptives. J. Lab. Clin. Med. 96, 762–769.

Gordon, E. M., Douglas, J., and Ratnoff, O. D. (1983). Influence of augmented Hageman factor(factor XII) titers on the cryoactivation of plasma prorenin in women using oral contraceptiveagents. J. Clin. Invest. 72, 1833–1838.

Gounaris, A. D., Brown, M. A., and Barrett, A. J. (1984). Human plasma a-cysteine proteinaseinhibitor: Purification by affinity chromatography, characterization and isolation of an activefragment. Biochem. J. 221, 445–452.

Greengard, J. S., and Griffin, J. H. (1984). Receptors for high molecular weight kininogen onstimulated washed human platelets. Biochemistry 23, 6863–6869.

Griffin, J. H. (1978). Role of surface in surface-dependent activation of Hageman factor (bloodcoagulation factor XII). Proc. Natl. Acad. Sci. USA 75, 1998–2002.

formation of bradykinin 199

Griffin, J. H., and Cochrane, C. G. (1976). Mechanisms for the involvement of high molecularweight kininogen in surface-dependent reactions of Hageman factor. Proc. Natl. Acad. Sci. USA73, 2554–2558.

Gustafson, E. J., Schutsky, D., Knight, L. C., and Schmaier, A. H. (1986). High molecular weightkininogen binds to unstimulated platelets. J. Clin. Invest. 78, 310–318.

Harpel, P. C. (1973). Studies on human plasma alpha 2-macroglobulin-enzyme interactions.Evidence for proteolytic modification of the subunit chain structure. J. Exp. Med. 138, 508–521.

Harpel, P. C., Lewin, M. F., and Kaplan, A. P. (1985). Distribution of plasma kallikrein betweenC-1 inactivator and a2-macroglobulin in plasma utilizing a new assay for a2-macroglobulin–kallikrein complexes. J. Biol. Chem. 260, 4257–4263.

Hasan, A. A., Zhang, J., Samuel, M., and Schmaier, A. H. (1995). Conformational changes in lowmolecular weight kininogen alters its ability to bind to endothelial cells. Thromb. Haemost. 74,1088–1095.

Hasan, A. A., Zisman, T., and Schmaier, A. H. (1998). Identification of cytokeratin 1 as a bindingprotein and presentation receptor for kininogens on endothelial cells. Proc. Natl. Acad. Sci. USA95, 3615–3620.

Hauert, J., Nicoloso, G., Schleuning, W. D., Bachmann, F., and Schapira, M. (1989). Plasminogenactivators in dextran sulfate-activated euglobulin fractions: A molecular analysis of factor XII-and prekallikrein-dependent fibrinolysis. Blood 73, 994–999.

Heck, L. W., and Kaplan, A. P. (1974). Substrates of Hageman factor. I. Isolation and characteriza-tion of human factor XI (PTA) and inhibition of the activated enzyme by a1-antitrypsin. J. Exp.Med. 140, 1615–1630.

Herwald, H., Hasan, A. A., Godovac-Zimmermann, J., Schmaier, A. H., and Muller-Esterl, W.(1995). Identification of an endothelial cell binding site on kininogen domain D3. J. Biol. Chem.270, 14634–14642.

Herwald, H., Dedio, J., Kellner, R., Loos, M., and Muller-Esterl, W. (1996). Isolation andcharacterization of the kininogen-binding protein p33 from endothelial cells: Identity with thegC1q receptor. J. Biol. Chem. 271, 13040–13047.

Higashiyama, S., Ohkubo, I., Ishiguro, H., Kunimatsu, M., Sawaki, K., and Sasaki, M. (1986).Human high molecular weight kininogen as a thiol proteinase inhibitor: Presence of the entireinhibition capacity in the native form of heavy chain. Biochemistry 25, 1669–1675.

Hirsch, E. F., Nakajima, T., Oshima, G., Erdos, E. G., and Herman, C. M. (1974). Kinin systemresponses in sepsis after trauma in man. J. Surg. Res. 17, 147–153.

Hojima, Y., Cochrane, C. G., Wiggins, R. C., Austen, K. F., and Stevens, R. L. (1984). In vitroactivation of the contact (Hageman factor) system of plasma by heparin and chondroitin sulfateE. Blood 63, 1453–1459.

Hong, S. L. (1980). Effect of bradykinin and thrombin on prostacyclin synthesis in endothelial cellsfrom calf and pig aorta and human umbilical cord vein. Thromb. Res. 18, 787–795.

Hsu, T. C., Shore, S. K., Seshsmma, T., Bagasra, O., and Walsh, P. N. (1998). Molecular cloning ofplatelet factor XI, an alternative splicing product of the plasma factor XI gene. J. Biol. Chem.273, 13787–13793.

Huisveld, I. A., Hospers, J. E. H., van Heeswijk, G. M., Bernink, M. J. E., Erich, W. B. M., andBouma, B. N. (1985). Contribution of contact activation factors to urokinase-related fibrinolyticactivityin whole human plasma. Thromb. Haemost. 54, 102 [abstract].

Iatridis, P. G., and Ferguson, J. H. (1962). Active Hageman factor: A plasma lysokinase of thehuman fibrinolytic system. J. Clin. Invest. 41, 1277.

Ichinose, A., Fujikawa, K., and Suyama, T. (1986). The activation of pro-urokinase by plasmakallikrein and its inactivation by thrombin. J. Biol. Chem. 261, 3486–3489.

200 kusumam joseph and allen p. kaplan

Ishiguro, H., Higashiyama, S., Ohkubo, I., and Sasaki, M. (1987). Mapping of functional domainsof human high molecular weight and low molecular weight kininogens using murine monoclonalantibodies. Biochemistry 26, 7021–7029.

Jacobsen, S. (1966). Separation of two separate substrates for plasma kinin forming enzymes.Nature 210, 98–99.

Jaffa, A. A., Durazo-Arvizu, R., Zheng, D., Lackland, D. T., Srikanth, S., Garvey, W. T., Schmaier,A. H., and Group, D. E. S. (2003). Plasma prekallikrein: A risk marker for hypertension andnephropathy in type 1 diabetes. Diabetes 52, 1215–1221.

Jespersen, J., and Kluft, C. (1985). Increased euglobulin fibrinolytic potential in women on oralcontraceptives low in oestrogen: Levels of extrinsic and intrinsic plasminogen activators, pre-kallikrein, factor XII, and C1-inactivator. Thromb. Haemost. 54, 454–459.

Jiang, Y., Nawarawong, W., Meloni, F. J., and Schmaier, A. H. (1992a). Insights on monoclonalantibodies to kininogens’ heavy chain which influence kininogens’ binding to platelets. Thromb.Haemost. 68, 143–148.

Jiang, Y. P., Muller-Esterl, W., and Schmaier, A. H. (1992b). Domain 3 of kininogens contains acell-binding site and a site that modifies thrombin activation of platelets. J. Biol. Chem. 267,3712–3717.

Jong, Y. J., Dalemar, L. R., Seehra, K., and Baenziger, N. L. (2002). Bradykinin receptor modula-tion in cellular models of aging and Alzheimer’s disease. Int. Immunopharmacol. 2, 1833–1840.

Jorg, M., and Binder, B. R. (1985). Kinetic analysis of plasminogen activation by purified plasmakallikrein. Thromb. Res. 39, 323–331.

Joseph, K., Ghebrehiwet, B., Peerschke, E. I., Reid, K. B., and Kaplan, A. P. (1996). Identificationof the zinc-dependent endothelial cell binding protein for high molecular weight kininogen andfactor XII: Identity with the receptor that binds to the globular ‘‘heads’’ of C1q (gC1q-R). Proc.Natl. Acad. Sci. USA 93, 8552–8557.

Joseph, K., Nakazawa, Y., Bahou, W. F., Ghebrehiwet, B., and Kaplan, A. P. (1999a). Plateletglycoprotein Ib: A zinc-dependent binding protein for the heavy chain of high-molecular-weightkininogen. Mol. Med. 5, 555–563.

Joseph, K., Shibayama, Y., Nakazawa, Y., Peerschke, E. I., Ghebrehiwet, B., and Kaplan, A. P.(1999b). Interaction of factor XII and high molecular weight kininogen with cytokeratin 1 andgC1qR of vascular endothelial cells and with aggregated Ab protein of Alzheimer’s disease.Immunopharmacology 43, 203–210.

Joseph, K., Ghebrehiwet, B., and Kaplan, A. P. (1999c). Cytokeratin 1 and gC1qR mediate highmolecular weight kininogen binding to endothelial cells. Clin. Immunol. 92, 246–255.

Joseph, K., Ghebrehiwet, B., and Kaplan, A. P. (2001a). Activation of the kinin-forming cascade onthe surface of endothelial cells. Biol. Chem. 382, 71–75.

Joseph, K., Shibayama, Y., Ghebrehiwet, B., and Kaplan, A. P. (2001b). Factor XII-dependentcontact activation on endothelial cells and binding proteins gC1qR and cytokeratin 1. Thromb.Haemost. 85, 119–124.

Joseph, K., Tholanikunnel, B. G., and Kaplan, A. P. (2002a). Activation of the bradykinin-forming cascade on endothelial cells: A role for heat shock protein 90. Int. Immunopharmacol.2, 1851–1859.

Joseph, K., Tholanikunnel, B. G., and Kaplan, A. P. (2002b). Heat shock protein 90 catalyzesactivation of the prekallikrein–kininogen complex in the absence of factor XII. Proc. Natl. Acad.Sci. USA 99, 896–900.

Juhlin, L., and Michaelsson, G. (1969). Vascular reactions in hereditary angioneurotic edema. ActaDerm. Venereol. 49, 20–25.

Kaplan, A. P., and Austen, K. F. (1970). A pre-albumin activator of prekallikrein. J. Immunol. 105,802–811.

formation of bradykinin 201

Kaplan, A. P., and Austen, K. F. (1971). A prealbumin activator of prekallikrein. II. Derivation ofactivators of prekallikrein from active Hageman factor by digestion with plasmin. J. Exp. Med.133, 696–712.

Kaplan, A. P., and Silverberg, M. (1987). The coagulation–kinin pathway of human plasma. Blood70, 1–15.

Kaplan, A. P., Kay, A. B., and Austen, K. F. (1972). A prealbumin activator of prekallikrein. 3.Appearance of chemotactic activity for human neutrophils by the conversion of human pre-kallikrein to kallikrein. J. Exp. Med. 135, 81–97.

Kaplan, A. P., Joseph, K., Shibayama, Y., Reddigari, S., Ghebrehiwet, B., and Silverberg, M. (1997).The intrinsic coagulation/kinin-forming cascade: Assembly in plasma and cell surfaces in inflam-mation. Adv. Immunol. 66, 225–272.

Kaplan, A. P., Joseph, K., and Silverberg, M. (2002). Pathways for bradykinin formation andinflammatory disease. J. Allergy Clin. Immunol. 109, 195–209.

Kellermann, J., Lottspeich, F., Henschen, A., and Muller-Esterl, W. (1986a). Amino acidsequence of the light chain of human high molecular mass kininogen. Adv. Exp. Med. Biol.198, 85–89.

Kellermann, J., Lottspeich, F., Henschen, A., and Muller-Esterl, W. (1986b). Completion of theprimary structure of human high-molecular-mass kininogen: The amino acid sequence of theentire heavy chain and evidence for its evolution by gene triplication. Eur. J. Biochem. 154, 471–478.

Kellermeyer, R. W., and Breckenridge, R. T. (1965). The inflammatory process in acute goutyarthritis. 1. Activation of Hageman factor by urate crystals. J. Lab. Clin. Med. 63, 307.

Kitamura, N., Kitagawa, H., Fukushima, D., Takagaki, Y., Miyata, T., and Nakanishi, S. (1985).Structural organization of the human kininogen gene and a model for its evolution. J. Biol.Chem. 260, 8610–8617.

Klemperer, M. R., Donaldson, V. H., and Rosen, F. S. (1968). Effect of C01 esterase on vascularpermeability in man: Studies in normal and complement-deficient individuals and in patientswith hereditary angioneurotic edema. J. Clin. Invest. 47, 604–611.

Kluft, C. (1976). Occurrence of C1 inactivator and other proteinase inhibitors in euglobulinfractions and their influence on fibrinolytic activity. Haemostasis 5, 136–146.

Kluft, C. (1977). Elimination of inhibition in euglobulin fibrinolysis by use of flufenamate:Involvement of C1-inactivator. Haemostasis 6, 351–369.

Kluft, C., Wijngaards, G., and Jie, A. F. (1984). Intrinsic plasma fibrinolysis: Involvement ofurokinase-related activity in the factor XII-independent plasminogen proactivator pathway.J. Lab. Clin. Med. 103, 408–419.

Laake, K., Vennerod, A. M., Haugen, G., and Gjonnaess, H. (1974). Cold-promoted activation offactor VII in human plasma: Studies on the associated acyl-arginine esterase activity. Thromb.Res. 4, 769–785.

Lenich, C., Pannell, R., and Gurewich, V. (1995). Assembly and activation of the intrinsic fibrino-lytic pathway on the surface of human endothelial cells in culture. Thromb. Haemost. 74,698–703.

Lin, Y., Harris, R. B., Yan, W., McCrae, K. R., Zhang, H., and Colman, R. W. (1997). Highmolecular weight kininogen peptides inhibit the formation of kallikrein on endothelial cellsurfaces and subsequent urokinase-dependent plasmin formation. Blood 90, 690–697.

Lipscomb, M. S., and Walsh, P. N. (1979). Human platelets and factor XI. Localization in plateletmembranes of factor XI-like activity and its functional distinction from plasma factor XI. J. Clin.Invest. 63, 1006–1014.

Londino, A. V., Jr., and Luparello, F. J. (1984). Factor XII deficiency in a man with gout andangioimmunoblastic lymphadenopathy. Arch. Internal Med. 144, 1497–1498.

202 kusumam joseph and allen p. kaplan

Lottspeich, F., Kellermann, J., Henschen, A., Rauth, G., and Muller-Esterl, W. (1984). Humanlow-molecular-mass kininogen: Amino-acid sequence of the light chain; homology with otherprotein sequences. Eur. J. Biochem. 142, 227–232.

Loza, J. P., Gurewich, V., Johnstone, M., and Pannell, R. (1994). Platelet-bound prekallikreinpromotes pro-urokinase-induced clot lysis: A mechanism for targeting the factor XII dependentintrinsic pathway of fibrinolysis. Thromb. Haemost. 71, 347–352.

Lu, G., Meier, K. E., Jaffa, A. A., Rosenzweig, S. A., and Egan, B. M. (1998). Oleic acid andangiotensin II induce a synergistic mitogenic response in vascular smooth muscle cells. Hyper-tension 31, 978–985.

Lundh, B., Laurell, A. B., Wetterqvist, H., White, T., and Granerus, G. (1968). A case of hereditaryangioneurotic oedema, successfully treated with epsilon-aminocaproic acid: Studies on C01esterase inhibitor, C01 activation, plasminogen level and histamine metabolism. Clin. Exp.Immunol. 3, 733–745.

Mahdi, F., Shariat-Madar, Z., Todd, R. F., III, Figueroa, C. D., and Schmaier, A. H. (2001).Expression and colocalization of cytokeratin 1 and urokinase plasminogen activator receptoron endothelial cells. Blood 97, 2342–2350.

Mahdi, F., Madar, Z. S., Figueroa, C. D., and Schmaier, A. H. (2002). Factor XII interacts with themultiprotein assembly of urokinase plasminogen activator receptor, gC1qR, and cytokeratin 1 onendothelial cell membranes. Blood 99, 3585–3596.

Mahdi, F., Shariat-Madar, Z., and Schmaier, A. H. (2003). The relative priority of prekallikrein andfactors XI/XIa assembly on cultured endothelial cells. J. Biol. Chem. 278, 43983–43990.

Mandle, R. J., Colman, R. W., and Kaplan, A. P. (1976). Identification of prekallikrein and high-molecular-weight kininogen as a complex in human plasma. Proc. Natl. Acad. Sci. USA 73,4179–4183.

Mandle, R. J., Jr., and Kaplan, A. P. (1979). Hageman-factor-dependent fibrinolysis: Generation offibrinolytic activity by the interaction of human activated factor XI and plasminogen. Blood 54,850–862.

Marceau, F., and Bachvarov, D. R. (1998). Kinin receptors. Clin. Rev. Allergy Immunol. 16,385–401.

Mason, J. W., Kleeberg, U., Dolan, P., and Colman, R. W. (1970). Plasma kallikrein and Hagemanfactor in gram-negative bacteremia. Ann. Internal Med. 73, 545–551.

McConnell, D. J. (1972). Inhibitors of kallikrein in human plasma. J. Clin. Invest. 51, 1611–1623.McDonagh, R. P., Jr., and Ferguson, J. H. (1970). Studies on the participation of Hageman factor in

fibrinolysis. Thromb. Diathesis Haemorrhagica 24, 1–9.Meier, H. L., Pierce, J. V., Colman, R. W., and Kaplan, A. P. (1977). Activation and function of

human Hageman factor: The role of high molecular weight kininogen and prekallikrein. J. Clin.Invest. 60, 18–31.

Miles, L. A., Rothschild, Z., and Griffin, J. H. (1981). Dextran sulphate stimulated fibrinolyticactivity in whole human plasma: Dependence on the contact activation system and on aurokinase-related antigen [abstract]. Thromb. Haemost. 46, 211.

Miles, L. A., Greengard, J. S., and Griffin, J. H. (1983a). A comparison of the abilities of plasmakallikrein, b-factor XIIa, factor XIa and urokinase to activate plasminogen. Thromb. Res. 29,407–417.

Miles, L. A., Rothschild, Z., and Griffin, J. H. (1983b). Dextran sulfate-dependent fibrinolysis inwhole human plasma. J. Lab. Clin. Med. 101, 214–225.

Mori, K., and Nagasawa, S. (1981). Studies on human high molecular weight (HMW) kininogen. II.Structural change of HMW kininogen by the action of human plasma kallikrein. J. Biochem. 89,1465–1473.

formation of bradykinin 203

Mori, K., Sakamoto, W., and Nagasawa, S. (1981). Studies on human high molecular weight(HMW) kininogen. III. Cleavage of HMW kininogen by the action of human salivary kallikrein.J. Biochem. 90, 503–509.

Motta, G., Rojkjaer, R., Hasan, A. A., Cines, D. B., and Schmaier, A. H. (1998). High molecularweight kininogen regulates prekallikrein assembly and activation on endothelial cells: A novelmechanism for contact activation. Blood 91, 516–528.

Motta, G., Shariat-Madar, Z., Mahdi, F., Sampaio, C. A., and Schmaier, A. H. (2001). Assemblyof high molecular weight kininogen and activation of prekallikrein on cell matrix. Thromb.Haemost. 86, 840–847.

Muller-Esterl, W., Fritz, H., Machleidt, W., Ritonja, A., Brzin, J., Kotnik, M., Turk, V.,Kellermann, J., and Lottspeich, F. (1985a). Human plasma kininogens are identical with a-cysteine proteinase inhibitors: Evidence from immunological, enzymological and sequencedata. FEBS Lett. 182, 310–314.

Muller-Esterl, W., Rauth, G., Lottspeich, F., Kellermann, J., and Henschen, A. (1985b). Limitedproteolysis of human low-molecular-mass kininogen by tissue kallikrein: Isolation and character-ization of the heavy and the light chains. Eur. J. Biochem. 149, 15–22.

Nagase, H., Cawston, T. E., De Silva, M., and Barrett, A. J. (1982). Identification of plasmakallikrein as an activator of latent collagenase in rheumatoid synovial fluid. Biochim. Biophys.Acta 702, 133–142.

Naito, K., and Fujikawa, K. (1991). Activation of human blood coagulation factor XI independentof factor XII: Factor XI is activated by thrombin and factor XIa in the presence of negativelycharged surfaces. J. Biol. Chem. 266, 7353–7358.

Nakazawa, Y., Joseph, K., and Kaplan, A. P. (2002). Inhibition of contact activation bya kininogen peptide (HKH20) derived from domain 5. Int. Immunopharmacol. 2,1875–1885.

Nishikawa, K., Shibayama, Y., Kuna, P., Calcaterra, E., Kaplan, A. P., and Reddigari, S. R. (1992).Generation of vasoactive peptide bradykinin from human umbilical vein endothelium-boundhigh molecular weight kininogen by plasma kallikrein. Blood 80, 1980–1988.

Nussberger, J., Cugno, M., Amstutz, C., Cicardi, M., Pellacani, A., and Agostoni, A. (1998). Plasmabradykinin in angio-oedema. Lancet 351, 1693–1697.

Nussberger, J., Cugno, M., Cicardi, M., and Agostoni, A. (1999). Local bradykinin generation inhereditary angioedema. J. Allergy Clin. Immunol. 104, 1321–1322.

O’Donnell, T. F., Jr., Clowes, G. H., Jr., Talamo, R. C., and Colman, R. W. (1976). Kinin activationin the blood of patients with sepsis. Surg. Gynecol. Obstet. 143, 539–545.

Ogston, D., Ogston, C. M., Ratnoff, O. D., and Forbes, C. D. (1969). Studies on a complexmechanism for the activation of plasminogen by kaolin and by chloroform: The participation ofHageman factor and additional cofactors. J. Clin. Invest. 48, 1786–1801.

Ogston, D., Bennett, N. B., Ogston, C. M., and Ratnoff, O. D. (1971). The assay of a plasmacomponent necessary for the generation of a plasminogen activator in the presence of Hagemanfactor (Hageman factor co-factor). Br. J. Haematol. 20, 209–216.

Pixley, R. A., Schapira, M., and Colman, R. W. (1985). The regulation of human factor XIIa byplasma proteinase inhibitors. J. Biol. Chem. 260, 1723–1729.

Pixley, R. A., Cassello, A., De La Cadena, R. A., Kaufman, N., and Colman, R. W. (1991). Effect ofheparin on the activation of factor XII and the contact system in plasma [see comment]. Thromb.Haemost. 66, 540–547.

Pixley, R. A., DeLa Cadena, R. A., Page, J. D., Kaufman, N., Wyshock, E. G., Colman, R. W.,Chang, A., and Taylor, F. B., Jr. (1992). Activation of the contact system in lethal hypotensivebacteremia in a baboon model. Am. J. Pathol. 140, 897–906.

204 kusumam joseph and allen p. kaplan

Pixley, R. A., De La Cadena, R., Page, J. D., Kaufman, N., Wyshock, E. G., Chang, A., Taylor, F. B.,Jr., and Colman, R. W. (1993). The contact system contributes to hypotension but notdisseminated intravascular coagulation in lethal bacteremia: In vivo use of a monoclonal anti-factor XII antibody to block contact activation in baboons. J. Clin. Invest. 91, 61–68.

Proud, D., Togias, A., Naclerio, R. M., Crush, S. A., Norman, P. S., and Lichtenstein, L. M. (1983).Kinins are generated in vivo following nasal airway challenge of allergic individuals withallergen. J. Clin. Invest. 72, 1678–1685.

Purdon, A. D., Schapira, M., De Agostini, A., and Colman, R. W. (1985). Plasma kallikrein andprorenin in patients with hereditary angioedema. J. Lab. Clin. Med. 105, 694–699.

Radcliffe, R., Bagdasarian, A., Colman, R., and Nemerson, Y. (1977). Activation of bovine factorVII by hageman factor fragments. Blood 50, 611–617.

Rand, M. D., Lock, J. B., van’t Veer, C., Gaffney, D. P., and Mann, K. G. (1996). Blood clotting inminimally altered whole blood. Blood 88, 3432–3445.

Ratnoff, O. D., and Naff, G. B. (1967). The conversion of C0IS to C01 esterase by plasmin andtrypsin. J. Exp. Med. 125, 337–358.

Reddigari, S., and Kaplan, A. P. (1988). Cleavage of human high-molecular weight kininogen bypurified kallikreins and upon contact activation of plasma. Blood 71, 1334–1340.

Reddigari, S., and Kaplan, A. P. (1989). Studies of the cleavage of human high molecular weightkininogen by purified plasma and tissue kallikreins, and upon contact activation of plasma. Adv.Exp. Med. Biol. 247B, 317–324.

Reddigari, S. R., Kuna, P., Miragliotta, G., Shibayama, Y., Nishikawa, K., and Kaplan, A. P. (1993a).Human high molecular weight kininogen binds to human umbilical vein endothelial cells via itsheavy and light chains. Blood 81, 1306–1311.

Reddigari, S. R., Shibayama, Y., Brunnee, T., and Kaplan, A. P. (1993b). Human Hageman factor(factor XII) and high molecular weight kininogen compete for the same binding site on humanumbilical vein endothelial cells. J. Biol. Chem. 268, 11982–11987.

Regoli, D., and Barabe, J. (1980). Pharmacology of bradykinin and related kinins. Pharmacol. Rev.32, 1–46.

Renne, T., Dedio, J., David, G., and Muller-Esterl, W. (2000). High molecular weight kininogenutilizes heparan sulfate proteoglycans for accumulation on endothelial cells. J. Biol. Chem. 275,33688–33696.

Richter, K., Muschler, P., Hainzl, O., and Buchner, J. (2001). Coordinated ATP hydrolysis by theHsp90 dimer. J. Biol. Chem. 276, 33689–33696.

Robinson, J. A., Klondnycky, M. L., Loeb, H. S., Racic, M. R., and Gunnar, R. M. (1975).Endotoxin, prekallikrein, complement and systemic vascular resistance: Sequential measure-ments in man. Am. J. Med. 59, 61–67.

Rojkjaer, R., and Schmaier, A. H. (1999a). Activation of the plasma kallikrein/kinin system onendothelial cell membranes. Immunopharmacology 43, 109–114.

Rojkjaer, R., and Schmaier, A. H. (1999b). Activation of the plasma kallikrein/kinin system onendothelial cells. Proc. Assoc. Am. Physicians 111, 220–227.

Rojkjaer, R., Hasan, A. A., Motta, G., Schousboe, I., and Schmaier, A. H. (1998). Factor XII doesnot initiate prekallikrein activation on endothelial cells. Thromb. Haemost. 80, 74–81.

Rosing, J., Tans, G., and Griffin, J. H. (1985). Surface-dependent activation of human factor XII(Hageman factor) by kallikrein and its light chain. Eur. J. Biochem. 151, 531–538.

Saito, H. (1980). The participation of plasma thromboplastin antecedent (factor XI) in contact-activated fibrinolysis. Proc. Soc. Exp. Biol. Med. 164, 153–157.

Saito, H., Ratnoff, O. D., and Donaldson, V. H. (1974). Defective activation of clotting, fibrinolytic,and permeability-enhancing systems in human Fletcher trait plasma. Circ. Res. 34, 641–651.

formation of bradykinin 205

Scandura, J. M., Zhang, Y., Van Nostrand, W. E., and Walsh, P. N. (1997). Progress curve analysis ofthe kinetics with which blood coagulation factor XIa is inhibited by protease nexin-2. Biochem-istry 36, 412–420.

Schapira, M., Scott, C. F., and Colman, R. W. (1981). Protection of human plasma kallikrein frominactivation by C1 inhibitor and other protease inhibitors: The role of high molecular weightkininogen. Biochemistry 20, 2738–2743.

Schapira, M., Despland, E., Scott, C. F., Boxer, L. A., and Colman, R. W. (1982a). Purified humanplasma kallikrein aggregates human blood neutrophils. J. Clin. Invest. 69, 1199–1202.

Schapira, M., Scott, C. F., James, A., Silver, L. D., Kueppers, F., James, H. L., and Colman, R. W.(1982b). High molecular weight kininogen or its light chain protects human plasma kallikreinfrom inactivation by plasma protease inhibitors. Biochemistry 21, 567–572.

Schapira, M., Scott, C. F., and Colman, R. W. (1982c). Contribution of plasma protease inhibitorsto the inactivation of kallikrein in plasma. J. Clin. Invest. 69, 462–468.

Schapira, M., Silver, L. D., Scott, C. F., Schmaier, A. H., Prograis, L. J., Jr., Curd, J. G., andColman, R. W. (1983). Prekallikrein activation and high-molecular-weight kininogen consump-tion in hereditary angioedema. N. Engl. J. Med. 308, 1050–1053.

Schmaier, A. H. (1997). Contact activation: A revision. Thromb. Haemost. 78, 101–107.Schmaier, A. H. (1998). Plasma contact activation: A revised hypothesis. Biol. Res. 31, 251–262.Schmaier, A. H., Kuo, A., Lundberg, D., Murray, S., and Cines, D. B. (1988). The expression of

high molecular weight kininogen on human umbilical vein endothelial cells. J. Biol. Chem. 263,16327–16333.

Schmaier, A. H., Rojkjaer, R., and Shariat-Madar, Z. (1999). Activation of the plasma kallikrein/kinin system on cells: A revised hypothesis. Thromb. Haemost. 82, 226–233.

Schreiber, A. D., Kaplan, A. P., and Austen, K. F. (1973). Inhibition by C1INH of Hagemannfactor fragment activation of coagulation, fibrinolysis, and kinin generation. J. Clin. Invest. 52,1402–1409.

Scott, C. F., and Colman, R. W. (1980). Function and immunochemistry of prekallikrein-highmolecular weight kininogen complex in plasma. J. Clin. Invest. 65, 413–421.

Scott, C. F., Schapira, M., and Colman, R. W. (1982a). Effect of heparin on the inactivation rate ofhuman factor XIa by antithrombin-III. Blood 60, 940–947.

Scott, C. F., Schapira, M., James, H. L., Cohen, A. B., and Colman, R. W. (1982b). Inactivation offactor XIa by plasma protease inhibitors: Predominant role of a1-protease inhibitor and protec-tive effect of high molecular weight kininogen. J. Clin. Invest. 69, 844–852.

Scott, C. F., Silver, L. D., Purdon, A. D., and Colman, R. W. (1985). Cleavage of human highmolecular weight kininogen by factor XIa in vitro: Effect on structure and function. J. Biol.Chem. 260, 10856–10863.

Sealey, J. E., Atlas, S. A., Laragh, J. H., Silverberg, M., and Kaplan, A. P. (1979). Initiation ofplasma prorenin activation by Hageman factor-dependent conversion of plasma prekallikrein tokallikrein. Proc. Natl. Acad. Sci. USA 76, 5914–5918.

Seligsohn, U., Osterud, B., Brown, S. F., Griffin, J. H., and Rapaport, S. I. (1979). Activation ofhuman factor VII in plasma and in purified systems: Roles of activated factor IX, kallikrein, andactivated factor XII. J. Clin. Invest. 64, 1056–1065.

Shariat-Madar, Z., Mahdi, F., and Schmaier, A. H. (1999). Mapping binding domains of kininogenson endothelial cell cytokeratin 1. J. Biol. Chem. 274, 7137–7145.

Shariat-Madar, Z., Mahdi, F., and Schmaier, A. H. (2001). Factor XI assembly and activation onhuman umbilical vein endothelial cells in culture. Thromb. Haemost. 85, 544–551.

Shariat-Madar, Z., Mahdi, F., and Schmaier, A. H. (2002). Identification and characterizationof prolylcarboxypeptidase as an endothelial cell prekallikrein activator. J. Biol. Chem. 277,17962–17969.

206 kusumam joseph and allen p. kaplan

Sheffer, A. L., Austen, K. F., and Rosen, F. S. (1972). Tranexamic acid therapy in hereditaryangioneurotic edema. N. Engl. J. Med. 287, 452–454.

Sheikh, I. A., and Kaplan, A. P. (1986a). Studies of the digestion of bradykinin, Lys-bradykinin, anddes-Arg9-bradykinin by angiotensin converting enzyme. Biochem. Pharmacol. 35, 1951–1956.

Sheikh, I. A., and Kaplan, A. P. (1986b). Studies of the digestion of bradykinin, lysyl bradykinin,and kinin-degradation products by carboxypeptidases A, B, and N. Biochem. Pharmacol. 35,1957–1963.

Sheikh, I. A., and Kaplan, A. P. (1989a). The mechanism of degradation of bradykinin (lysyl-bradykinin) in human serum. Adv. Exp. Med. Biol. 247A, 331–336.

Sheikh, I. A., and Kaplan, A. P. (1989b). Mechanism of digestion of bradykinin and lysylbradykinin(kallidin) in human serum: Role of carboxypeptidase, angiotensin-converting enzyme and deter-mination of final degradation products. Biochem. Pharmacol. 38, 993–1000.

Shibayama, Y., Joseph, K., Nakazawa, Y., Ghebreihiwet, B., Peerschke, E. I., and Kaplan, A. P.(1999). Zinc-dependent activation of the plasma kinin-forming cascade by aggregated b amyloidprotein. Clin. Immunol. 90, 89–99.

Shoemaker, L. R., Schurman, S. J., Donaldson, V. H., and Davis, A. E., III (1994). Hereditaryangioneurotic oedema: Characterization of plasma kinin and vascular permeability-enhancingactivities. Clin. Exp. Immunol. 95, 22–28.

Silverberg, M., and Diehl, S. V. (1987). The autoactivation of factor XII (Hageman factor) inducedby low-Mr heparin and dextran sulphate: The effect of the Mr of the activating polyanion.Biochem. J. 248, 715–720.

Silverberg, M., and Kaplan, A. P. (1982). Enzymatic activities of activated and zymogen forms ofhuman Hageman factor (factor XII). Blood 60, 64–70.

Silverberg, M., Dunn, J. T., Garen, L., and Kaplan, A. P. (1980a). Autoactivation of humanHageman factor: Demonstration utilizing a synthetic substrate. J. Biol. Chem. 255, 7281–7286.

Silverberg, M., Nicoll, J. E., and Kaplan, A. P. (1980b). The mechanism by which the light chain ofcleaved HMW-kininogen augments the activation of prekallikrein, factor XI and Hagemanfactor. Thromb. Res. 20, 173–189.

Silverberg, M., Longo, J., and Kaplan, A. P. (1986). Study of the effect of high molecular weightkininogen upon the fluid-phase inactivation of kallikrein by C1 inhibitor. J. Biol. Chem. 261,14965–14968.

Sinha, D., Seaman, F. S., Koshy, A., Knight, L. C., and Walsh, P. N. (1984). Blood coagulationfactor XIa binds specifically to a site on activated human platelets distinct from that for factor XI.J. Clin. Invest. 73, 1550–1556.

Smith, D., Gilbert, M., and Owen, W. G. (1985). Tissue plasminogen activator release in vivo inresponse to vasoactive agents. Blood 66, 835–839.

Stead, N., Kaplan, A. P., and Rosenberg, R. D. (1976). Inhibition of activated factor XII byantithrombin-heparin cofactor. J. Biol. Chem. 251, 6481–6488.

Stewart, J. M., Gera, L., York, E. J., Chan, D. C., and Bunn, P. (1999). Bradykinin antagonists:Present progress and future prospects. Immunopharmacology 43, 155–161.

Strang, C. J., Cholin, S., Spragg, J., Davis, A. E., III, Schneeberger, E. E., Donaldson, V. H., andRosen, F. S. (1988). Angioedema induced by a peptide derived from complement componentC2. J. Exp. Med. 168, 1685–1698.

Tait, J. F., and Fujikawa, K. (1986). Identification of the binding site for plasma prekallikrein inhuman high molecular weight kininogen: A region from residues 185 to 224 of the kininogenlight chain retains full binding activity. J. Biol. Chem. 261, 15396–15401.

Tait, J. F., and Fujikawa, K. (1987). Primary structure requirements for the binding of human highmolecular weight kininogen to plasma prekallikrein and factor XI. J. Biol. Chem. 262, 11651–11656.

formation of bradykinin 207

Talamo, R. C., Haber, E., and Austen, K. F. (1969). A radioimmunoassay for bradykinin in plasmaand synovial fluid. J. Lab. Clin. Med. 74, 816–827.

Tan, Y., Hutchison, F. N., and Jaffa, A. A. (2004). Mechanism of angiotensin II: Induced expressionof B2 kinin receptors. Am. J. Physiol. Heart Circ. Physiol. 286, H926–H932.

Tankersley, D. L., and Finlayson, J. S. (1984). Kinetics of activation and autoactivation of humanfactor XII. Biochemistry 23, 273–279.

Tankersley, D. L., Fournel, M. A., and Schroeder, D. D. (1980). Kinetics of activation of pre-kallikrein by prekallikrein activator. Biochemistry 19, 3121–3127.

Thompson, R. E., Mandle, R., Jr., and Kaplan, A. P. (1977). Association of factor XI and highmolecular weight kininogen in human plasma. J. Clin. Invest. 60, 1376–1380.

Thompson, R. E., Mandle, R., Jr., and Kaplan, A. P. (1979). Studies of binding of prekallikrein andfactor XI to high molecular weight kininogen and its light chain. Proc. Natl. Acad. Sci. USA 76,4862–4866.

Travis, J., and Salvesen, G. S. (1983). Human plasma proteinase inhibitors. Annu. Rev. Biochem.52, 655–709.

Tuszynski, G. P., Bevacqua, S. J., Schmaier, A. H., Colman, R. W., and Walsh, P. N. (1982). FactorXI antigen and activity in human platelets. Blood 59, 1148–1156.

van der Graaf, F., Koedam, J. A., and Bouma, B. N. (1983a). Inactivation of kallikrein in humanplasma. J. Clin. Invest. 71, 149–158.

van der Graaf, F., Koedam, J. A., Griffin, J. H., and Bouma, B. N. (1983b). Interaction of humanplasma kallikrein and its light chain with C1 inhibitor. Biochemistry 22, 4860–4866.

van Iwaarden, F., de Groot, P. G., and Bouma, B. N. (1988). The binding of high molecular weightkininogen to cultured human endothelial cells. J. Biol. Chem. 263, 4698–4703.

Vavrek, R. J., and Stewart, J. M. (1985). Competitive antagonists of bradykinin. Peptides 6,161–164.

Vennerod, A. M., Laake, K., Solberg, A. K., and Stromland, S. (1976). Inactivation and binding ofhuman plasma kallikrein by antithrombin III and heparin. Thromb. Res. 9, 457–466.

von dem Borne, P. A., Koppelman, S. J., Bouma, B. N., and Meijers, J. C. (1994). Surfaceindependent factor XI activation by thrombin in the presence of high molecular weight kinino-gen. Thromb. Haemost. 72, 397–402.

Wachtfogel, Y. T., Pixley, R. A., Kucich, U., Abrams, W., Weinbaum, G., Schapira, M., and Colman,R. W. (1986). Purified plasma factor XIIa aggregates human neutrophils and causes degranula-tion. Blood 67, 1731–1737.

Wachtfogel, Y. T., DeLa Cadena, R. A., Kunapuli, S. P., Rick, L., Miller, M., Schultze, R. L., Altieri,D. C., Edgington, T. S., and Colman, R. W. (1994). High molecular weight kininogen binds toMac-1 on neutrophils by its heavy chain (domain 3) and its light chain (domain 5). J. Biol. Chem.269, 19307–19312.

Wallace, E. M., Perkins, S. J., Sim, R. B., Willis, A. C., Feighery, C., and Jackson, J. (1997).Degradation of C1-inhibitor by plasmin: Implications for the control of inflammatory processes.Mol. Med. 3, 385–396.

Walsh, P. N., and Griffin, J. H. (1981). Contributions of human platelets to the proteolyticactivation of blood coagulation factors XII and XI. Blood 57, 106–118.

Weiss, A. S., Gallin, J. I., and Kaplan, A. P. (1974). Fletcher factor deficiency: A diminished rate ofHageman factor activation caused by absence of prekallikrein with abnormalities of coagulation,fibrinolysis, chemotactic activity, and kinin generation. J. Clin. Invest. 53, 622–633.

Wiggins, R. C., Bouma, B. N., Cochrane, C. G., and Griffin, J. H. (1977). Role of high-molecular-weight kininogen in surface-binding and activation of coagulation factor XI and prekallikrein.Proc. Natl. Acad. Sci. USA 74, 4636–4640.

Wuepper, K. D. (1973). Prekallikrein deficiency in man. J. Exp. Med. 138, 1345–1355.

208 kusumam joseph and allen p. kaplan

Wuepper, K. D., Miller, D. R., and Lacombe, M. J. (1975). Flaujeac trait: Deficiency of humanplasma kininogen. J. Clin. Invest. 56, 1663–1672.

Wuillemin, W. A., Minnema, M., Meijers, J. C., Roem, D., Eerenberg, A. J., Nuijens, J. H., tenCate, H., and Hack, C. E. (1995). Inactivation of factor XIa in human plasma assessed bymeasuring factor XIa–protease inhibitor complexes: Major role for C1-inhibitor. Blood 85,1517–1526.

Wuillemin, W. A., Eldering, E., Citarella, F., de Ruig, C. P., ten Cate, H., and Hack, C. E. (1996).Modulation of contact system proteases by glycosaminoglycans: Selective enhancement of theinhibition of factor XIa. J. Biol. Chem. 271, 12913–12918.

Yang, H. Y., and Erdos, E. G. (1967). Second kininase in human blood plasma. Nature 215,1402–1403.

Yokosawa, N., Takahashi, N., Inagami, T., and Page, D. L. (1979). Isolation of completely inactiveplasma prorenin and its activation by kallikreins: A possible new link between renin andkallikrein. Biochim. Biophys. Acta. 569, 211–219.

Zahedi, R., Bissler, J. J., Davis, A. E., III, Andreadis, C., and Wisnieski, J. J. (1995). Unique C1inhibitor dysfunction in a kindred without angioedema. II. Identification of an Ala443 ! Valsubstitution and functional analysis of the recombinant mutant protein. J. Clin. Invest. 95, 1299–1305.

Zahedi, R., Wisnieski, J., and Davis, A. E., III. (1997). Role of the P2 residue of complement 1inhibitor (Ala443) in determination of target protease specificity: Inhibition of complement andcontact system proteases. J. Immunol. 159, 983–988.

Zhang, Y., Scandura, J. M., Van Nostrand, W. E., and Walsh, P. N. (1997). The mechanism by whichheparin promotes the inhibition of coagulation factor XIa by protease nexin-2. J. Biol. Chem.272, 26139–26144.

Zhao, Y., Qiu, Q., Mahdi, F., Shariat-Madar, Z., Rojkjaer, R., and Schmaier, A. H. (2001). Assemblyand activation of HK-PK complex on endothelial cells results in bradykinin liberation and NOformation. Am. J. Physiol. Heart Circ. Physiol. 280, H1821–H1829.

Interleukin-2, Interleukin-15, and Their Roles in HumanNatural Killer Cells

Brian Becknell*,{ and Michael A. Caligiuri*, {,{, x, }

*Medical Scientist Program,{Integrated Biomedical Graduate Program,

{Department of Internal Medicine,xDivision of Hematology/Oncology,

}Comprehensive Cancer Center, Ohio State University, Columbus, Ohio 43210

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2091. Introduction.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2092. IL-2 and IL-15 Signaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2113. Role of IL-2 versus IL-15 in Human NK Cell Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2154. Low-Dose IL-2 Therapy Expands the CD56bright NK Subset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2205. IL-2/IL-15 and NK Cell Survival. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2226. IL-2/IL-15 and NK Cell Proliferation .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2247. IL-2/IL-15 and NK Cell Apoptosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2258. IL-2/IL-15 and NK Cell Cytotoxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2269. IL-2/IL-15 and NK Cell Cytokine Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

10. NK Cell–Immune Cell Interactions and Modulation by IL-2/IL-15 .. . . . . . . . . . . . . . . . . . . . . . . . 22811. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

Abstract

Natural killer (NK) cells are CD56þCD3� large granular lymphocytes thatconstitute a key component of the human innate immune response. In additionto their potent cytolytic activity, NK cells elaborate a host of immunoregulatorycytokines and chemokines that play a crucial role in pathogen clearance.Furthermore, interactions between NK and other immune cells are implicatedin triggering the adaptive, or antigen-specific, immune response. Interleukin-2 (IL-2) and IL-15 are two distinct cytokines with partially overlappingproperties that are implicated in the development, homeostasis, and functionof NK cells. This review examines the pervasive effects of IL-2 and IL-15 on NKcell biology, with an emphasis on recent discoveries and lingering challenges inthe field.

1. Introduction

The innate immune system represents the human body’s essential first line ofdefense against infectious disease and malignant transformation. In the im-mune competent host, innate immune effectors act rapidly to restrict the

209advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

210 brian becknell and michael a. caligiuri

dissemination of disease, as well as to trigger the adaptive, or antigen-specific,immune system. Human natural killer (NK) cells are CD56þCD3� large gran-ular lymphocytes that constitute one component of the innate immune system.In addition to their potent cytolytic activity, NK cells elaborate a host ofimmunoregulatory cytokines and chemokines that play a crucial role in patho-gen clearance. In particular, NK cells produce abundant quantities of interfer-on-g (IFN-g), a critical cytokine for the clearance of infectious pathogens as wellas for tumor surveillance (Lieberman and Hunter, 2002; Shankaran et al., 2001).In rodent models, NK cells have been proved essential for the clearance ofcertain tumors, as well as bacterial, fungal, viral, and parasitic infections(Carayannopoulos and Yokoyama, 2004; Kim et al., 2000). Furthermore, inrare cases of human congenital immune deficiencies, the absence of NK cellsproduces a clinical spectrum that parallels classic severe combined immunode-ficiency (SCID) syndromes (Gilmour et al., 2001). The importance of NK cellsis magnified in a host of clinical scenarios in which the adaptive immune systemis compromised. These states include congenital immune disorders, iatrogenicimmune suppression following organ transplantation, and the acquired immunedeficiency syndrome (AIDS). NK cells represent an attractive target fortherapeutic manipulation to fight the rampant opportunistic infections andvirus-induced cancers that arise under these states of adaptive immunoparaly-sis. Indeed, this is the rationale underlying ultralow-dose interleukin-2 therapyto potentiate the antitumor effects of NK cells in AIDS-associated malignancies(Bernstein et al., 1995; Jacobson et al., 1996; Smith, 1988). This approach isfurther substantiated by the advent of NK cell transplantation, in which allo-reactive NK cells have been shown to mediate a potent graft-versus-tumoreffect in patients with acute myeloid leukemia (Ruggeri et al., 2002). Based onthese advances, it is anticipated that a greater mechanistic understanding of NKcells and the innate immune system will provide new means to enhance thefunction of these cells for the benefit of the immunocompromised patient.

The study of interleukin-2 (IL-2) and IL-15 has served as a paradigm forour current understanding of NK cell function and homeostasis. These twocytokines—more than any others—are implicated in the development, surviv-al, proliferation, apoptosis, and effector functions of NK cells. In addition, IL-2 and IL-15 mediate and/or potentiate interactions between NK cells andother immune cells in secondary lymphoid organs and the periphery. Thisreview examines these pervasive effects of IL-2 and IL-15 on NK cell biology.We begin with a discussion of their signal transduction mechanisms. Next, wehighlight the roles of IL-2 and IL-15 in NK cell ontogeny, homeostasis, andfunction. Finally, we conclude by discussing the roles of these two cytokines inmediating interactions between NK cells and other leukocytes in the context ofthe human immune response.

role of il-2 and il-15 in human nk cells 211

2. IL-2 and IL-15 Signaling

2.1. Receptors for IL-2 and IL-15

Interleukin-2 was initially described as a T cell–derived cytokine, or lympho-kine, and is chiefly appreciated for its crucial role in T-cell activation, prolifer-ation, and cell death (Waldmann et al., 2001). Interleukin-15 was firstidentified on the basis of its ability to mimic IL-2–induced T-cell proliferation(Burton et al., 1994; Grabstein et al., 1994). This proliferative effect of IL-15 could be neutralized with antibodies specific to the b and g subunits ofthe previously characterized IL-2 receptor (IL-2R), demonstrating that thesecytokines share these receptor subunits. In contrast, neutralizing anti-bodies to the high-affinity IL-2Ra chain did not inhibit IL-15–induced T-cellproliferation, suggesting that IL-15 possessed its own private high-affinityreceptor (Carson et al., 1994a). Indeed, subsequent cloning and characteriza-tion of the IL-15Ra chain revealed that it possesses extremely high affinity forIL-15 (Kd, �10�11 M)—even in the absence of the IL-2Rbg subunits(Anderson et al., 1995; Giri et al., 1995). In contrast, the IL-2Ra subunitbinds IL-2 with low affinity (Kd, �10�8 M) in the absence of the IL-2Rbg,and a ternary IL-2Rabg complex is requisite for high-affinity binding of IL-2(Kd, �10�11 M; Smith, 1988). In the absence of their private a subunits, bothIL-2 and IL-15 are capable of binding an intermediate-affinity IL-2Rbg

heterodimer and initiating cytoplasmic signal transduction cascades.

2.2. Signal Transduction Cascades Initiated by IL-2/IL-15

In the absence of IL-2/IL-15, two tyrosine kinases associate with the cytoplas-mic tails of the IL-2Rbg heterodimer: Jak1 with b and Jak3 with g. Binding ofIL-2/IL-15 to the bg complex results in tyrosine phosphorylation of STAT3and STAT5 by Jak1 and Jak3, respectively (Miyazaki et al., 1994). Thesephosphorylated STATs then dimerize and translocate to the nucleus, wherethey serve as transcription factors (Leonard, 2001). Consistent with the criticalrole of STAT5 in IL-2/IL-15 signal transduction, knockout mice lackingSTAT5a/b, STAT5b, and Jak3 all have NK cell defects (Imada et al., 1998;Nosaka et al., 1995; Park et al., 1995; Teglund et al., 1998). In additionto Jak/STAT signaling, IL-2/IL-15 binding results in the phosphorylation oftyrosine residues on the cytoplasmic portion of the IL-2Rb chain. Thesephosphotyrosine residues serve as docking sites for SH2, the domain of theadaptor protein Shc, which initiates a Ras/Raf/MAPK cascade that culminatesin AP-1 activation (Zhu et al., 1994). IL-2/IL-15 ligation also results in activa-tion of the phosphatidylinositol-3-kinase (PI-3 kinase)/Akt pathway (Gu et al.,2000), stimulation through src family cytoplasmic tyrosine kinases (Miyazaki

212 brian becknell and michael a. caligiuri

et al., 1995), and activation of the nuclear factor kB (NF-kB; McDonald et al.,1998). Finally, IL-2/IL-15 binding results in induction of BCL-2, a mitochon-drial protein that promotes cell survival (Miyazaki et al., 1995). The impor-tance of these signal transduction pathways in NK cells is highlightedthroughout this review.

2.3. Divergent Functions of IL-2 and IL-15 In Vivo, DespiteSimilarities In Vitro

Despite the shared signal transduction properties of IL-2 and IL-15 in vitro,studies of mice and occasional human patients deficient for these cytokines—or their private a chains—indicate that their in vivo functions are quitedistinct. Indeed, genetic ablation of IL-2 or of the IL-2Ra chain in miceresults in normal NK cell numbers and function (Kundig et al., 1993;Schorle et al., 1991; Willerford et al., 1995). Rather, these mice develop asevere inflammatory disease as a consequence of unabated T-cell proliferation,highlighting the role of IL-2 and its private a chain in activation-induced celldeath of T lymphocytes. Likewise, rare patients have been identified withgenetic deficiencies of IL-2 or IL-2Ra; yet NK cell numbers and functionare preserved in these individuals (DiSanto et al., 1990; Sharfe et al., 1997).Although humans with deficiencies in IL-15 or IL-15Ra have not beenreported, genetic ablation of IL-15 in mice results in a drastic diminution inNK cell number, which can be rescued by exogenous administration of IL-15(Kennedy et al., 2000). Likewise, early experiments revealed a profound NKcell deficiency in IL-15Ra knockout mice (Lodolce et al., 1998); however, aswe discuss below, NK cell expression of IL-15Ra is not required for mostaspects of NK development and mature NK function (Koka et al., 2003).Rather, NK cell–independent IL-15Ra serves to present this cytokine intrans to mature NK cells in the periphery, and this action is critical for NKcell survival (Koka et al., 2003). Thus, in contrast to their shared signaltransduction machinery, IL-2 and IL-15 occupy different physiologic niches.

2.3.1. Resolving the IL-2 versus IL-15 Paradox: Can the IL-15RaChain Signal?

How can we resolve this paradox between shared in vitro signal transductionand disparate in vivo function? Given the unique a subunit of each cytokine’sreceptor, one hypothesis states that the IL-15Ra subunit is capable of eli-citing its own signal transduction cascades. Support for this hypothesis islargely restricted to a series of in vitro studies of IL-15Ra signaling. Forexample, association of IL-15 with its IL-15Ra subunit has been reported toinitiate the recruitment of tumor necrosis factor (TNF) receptor–associated

role of il-2 and il-15 in human nk cells 213

factor 2 (TRAF2) to the short cytoplasmic tail of IL-15Ra in fibroblasts(Bulfone-Pau et al., 1999). This association between TRAF2 and IL-15Ra, asdemonstrated by immunoprecipitation and Western blotting, protects thesecells from TNF-a–induced apoptosis and leads to activation of NF-kB.Similarly, this interaction of IL-15Ra and TRAF2, and IL-15-inducible NF-kB nuclear translocation, have been demonstrated in an erythrocyte cell linelacking the IL-2Rb subunit (Giron-Michel et al., 2003). In addition to TRAF2and NF-kB activation, IL-15 association with IL-15Ra results in activation ofSyk, a cytoplasmic tyrosine kinase, in B and T lymphocytes (Bulanova et al.,2001). This leads to the Syk-dependent phosphorylation of the IL-15Ra

cytoplasmic domain at residue 227 as well as phosphorylation and activation ofphospholipase Cg. This Syk-dependent signal transduction pathway, initiatedby association of IL-15 with the IL-15Ra chain, results in calcium influx as wellas a substantial survival benefit for the cells. This can occur in the completeabsence of the IL-2Rb chain and in the presence of neutralizing antibodies tothe gc chain (Bulanova et al., 2001). Whereas these studies have successfullydocumented IL-15Ra–specific signaling pathways, their relevance in NK cellsand physiologic significance in vivo remain unresolved.

2.3.2. Resolving the IL-2 versus IL-15 Paradox: A Matterof Distribution?

An alternative hypothesis proposes that the distribution of the IL-2Ra andIL-15Ra subunits, as well as the distribution of IL-2 and IL-15 themselves, arethe most likely determinants of the distinct physiologic functions of thesecytokines. Indeed, the expression patterns of each cytokine and its private a

subunit are certainly distinct. Expression of IL-2 mRNA and protein is gener-ally restricted to activated T lymphocytes (Smith, 1988), although some dataindicate that this cytokine can be expressed by dendritic cells in response tolipopolysaccharide (LPS) (Granucci et al., 2001, 2003). In contrast to IL-2 mRNA, the levels of IL-15 transcript are plentiful and ubiquitous acrossnearly all cell types; however, IL-15 protein expression is far less abundant andgenerally restricted to monocytes, dendritic cells, and stromal fibroblasts(Fehniger and Caligiuri, 2001). Competitive binding experiments on humanNK cells, using radiolabeled IL-15 and IL-2, reveal no difference in thepreferential binding of one cytokine over the other, with equal inhibitionof binding in the presence of anti-IL2Rb monoclonal antibody (mAb)(Carson et al., 1994a). This is consistent with the existence of a shared receptorfor these cytokines on NK cells. Scatchard analysis of total NK cells hasidentified high and low-affinity binding sites for both cytokines (Carson et al.,1994a). The few high-affinity binding sites for IL-2 likely represent cells within

214 brian becknell and michael a. caligiuri

the CD56bright NK subset, which exclusively expresses the private high-affinityIL-2Ra (Caligiuri et al., 1990). The distribution of high- and low-affinity IL-15-binding sites among human NK cells is unknown; however, the significanceof IL-15Ra expression on these cells is highlighted by the ability of ultralow-dose IL-15 to promote their survival in serum-free medium (Carson et al.,1994a).

2.4. New Concepts in IL-15 Signaling: IL-15 IsPresented in Trans

Our understanding of IL-15Ra distribution has taken an unexpected twist, asgenetic and biochemical lines of evidence have converged in elegant demon-strations of trans signaling between an IL-15/IL-15Ra high-affinity complexon a ‘‘presenting cell’’ and IL-15bg on a neighboring T or NK cell (Duboiset al., 2002; Koka et al., 2003). Prior to this discovery, our understanding ofIL-15 signaling was based entirely on the model depicted in Fig. 1A.

In this model, IL-15 engages one or more receptor components on a singlecell type, initiating signal transduction cascades within this particular cell. Thisis signaling in cis. However, studies of IL-15Ra knockout animals have defini-tively shown that a complex of IL-15 and IL-15Ra on one cell can presentcytokine and signal through the intermediate-affinity bgc heterodimer on aneighboring cell, i.e., signaling in trans (Fig. 1B) (Koka et al., 2003; Schlunset al., 2004). Moreover, biochemical studies indicate that an IL-15/IL-15Ra

complex on monocytes is capable of multiple rounds of endocytosis followedby cytokine presentation at the cell surface (Dubois et al., 2002). The variousphysiologic roles of cis and trans signaling are only now being delineated. InNK cells, the strongest evidence points toward a role for trans signaling in cellsurvival, which is highlighted later in this review (Koka et al., 2003; Schlunset al., 2004). In CD8þ T cells, trans signaling is required for IL-15–inducedproliferation (Schluns et al., 2004).

The discovery of trans signaling raises the possibility that the regulation ofIL-15 presentation may represent a key point of IL-15/IL-15Ra distributionin vivo. Given the transient nature of IL-15 protein expression, in vivo studieshave failed to identify its physiologic triggers. In vitro, stimulation of humanmonocytes and monocyte-derived cell lines with interferon-g and LPS elicitsIL-15 presentation (Carson et al., 1995; Dubois et al., 2002; Musso et al.,1999). Human and murine monocyte-derived dendritic cells (DCs) produceIL-15 and IL-15Ra in response to treatment with IFN-a (Mattei et al., 2001;Santini et al., 2000). The only negative regulator of IL-15 presentation thus fardescribed is IL-15 itself, which induces a rapid receptor internalization(Dubois et al., 2002).

Figure 1 Cis versus trans signaling by IL-15. (A) Soluble IL-15 is capable of signaling in cisthrough its intermediate-affinity receptor (bgc heterodimer; Kd, �10�8) or the high-affinityreceptor (abgc heterotrimer). Arrows denote the proven/potential ability of the each receptorchain to initiate signaling through its cytoplasmic domain. (B) Alternatively, a complex of IL-15/IL-15Ra on one cell can present cytokine and signal through the intermediate affinity bgc

heterodimer on a neighboring cell, i.e., signaling in trans. What is the precise nature of thephysiologic IL-15–presenting cell? The answer is unknown, but dendritic cells, monocytes, andstromal fibroblasts are the leading candidates.

role of il-2 and il-15 in human nk cells 215

3. Role of IL-2 versus IL-15 in Human NK Cell Development

3.1. Historical Work: Before IL-15

Early studies of human NK cell development focused on the ability of stromalcells to support the differentation of small numbers of NK cells from bonemarrow (BM) progenitors in long-term culture lacking exogenous cytokines

216 brian becknell and michael a. caligiuri

(Pollack et al., 1992). Later, investigators observed that exogenous IL-2 couldpromote NK differentiation of CD34þCD38� hematopoietic progenitors instromal-free cultures, and with higher efficiency in the presence of stroma(Lotzova et al., 1993; Shibuya et al., 1995; Miller et al., 1992). Furthermore,administration of recombinant IL-2 to mice or human patients results in NKcell expansion (Caligiuri et al., 1993; Piguet et al., 1986; Rosenberg et al.,1987). Despite these encouraging findings, their physiologic significance tonormal NK development remained a mystery, given the limited production ofIL-2 by antigen-activated T cells in the periphery, as well as the persistenceof NK cells in mice lacking T cells (as well as in knockout animals lacking theIl-2 gene or the private Il-2Ra subunit; Dorshkind et al., 1985; Kundig et al.,1993; Schorle et al., 1991; Willerford et al., 1995; ). A valuable hint to resolvethis paradox was provided by the observation that mice and human patientslacking either component of the IL-2Rbg complex are severely difficient inNK cells—suggesting a molecule other than IL-2 may be responsible forsignaling through the intermediate-affinity IL-2R during NK development(Cao et al., 1995; DiSanto et al., 1995; Gilmour et al., 2001; Noguchi et al.,1993; Suzuki et al., 1997).

3.2. IL-15 Promotes Human NK Cell Development fromHematopoietic Progenitors

Shortly after its codiscovery by the Grabstein and Waldmann laboratories in1994, numerous studies have implicated IL-15 in the process of human NKcell development. Studies by Mrozek et al. first demonstrated the ability of IL-15 alone to promote differentiation of CD34þ hematopoietic progenitor cells(HPCs) to CD56þ NK cells with large granular lymphocyte (LGL) morpholo-gy, although this occurred without appreciable cell expansion over the 21-dayculture (Mrozek et al., 1996). In seeking to expand the number of NK cellsderived from CD34þ HPCs, two stromal-derived cytokines have been identi-fied: c-Kit ligand (KL) and Flt3 ligand (FL). Culture in either KL or FL resultsin significant expansion in absolute cell number, but these cells failed toacquire CD56 expression or LGL morphology. However, the combination ofIL-15 with KL and/or FL increases the absolute cell number while maintain-ing the cell percentage adopting an NK fate (Mrozek et al., 1996; Yu et al.,1998). Thus, whereas IL-15 alone induces CD56 expression and LGL mor-phology, its coupling with FL or KL increases the number of these NKcells. NK cells derived in combinations of IL-15 with FL or KL are equallycytotoxic as those derived in IL-15 alone. In addition, CD56bright NK cellsderived in the presence of IL-15, or the combination of IL-15 and KL, arecapable of cytokine production [IFN-g, TNF-a, and granulocyte-macrophage

role of il-2 and il-15 in human nk cells 217

colony-stimulating factor (GM-CSF)] and chemokine production (macrophageinflammatory protein-1a, MIP-1a) in response to the monocyte-derived cyto-kines IL-12 and IL-15. In contrast, cells expanded exclusively in the presenceof FL or KL fail to exhibit cytolytic activity or elaborate cytokines, in keepingwith the inability of these cytokines to promote NK differentiation in theabsence of IL-15 (Mrozek et al., 1996; Yu et al., 1998).

3.2.1. Mechanism of Synergy Between IL-15 and Flt3Ligand/c-Kit Ligand

By what mechanism do FL and KL exert their action during human NK celldevelopment? One possible explanation is that these cytokines increase thefrequency of NK precursors among CD34þ HPCs. This hypothesis wasaddressed by limiting dilution analysis (LDA) of NK precursor frequencyamong CD34þ HPCs cultured for 21 days in FL versus KL, compared withfreshly isolated CD34þ HPCs. Indeed, 3 weeks of culture in either FL or KLsignificantly increased the NK precursor frequency when the cells were placedfor 14 days in IL-15 alone. Moreover, FL significantly outpaced KL in thisregard. These data suggest that FL and KL exert some qualitative effect on theNK precursor pool, rendering it more responsive to IL-15. In support ofthis hypothesis, a CD122þCD34bright intermediate cell population is detect-able on 10 days of culture in FL or KL (although to a lesser extent). Isolation ofthis CD122þ population from FL-cultured CD34þ HPCs and LDA in IL-15reveals an NK cell precursor frequency that is 65- to 235-fold higher than thatobserved among freshly isolated CD34þ HPCs (Yu et al., 1998). In support ofthis concept, parallel studies in mouse BM have demonstrated synergisticinteractions between FL or KL and IL-15 in promoting development ofNK cells (Williams et al., 1997). Culture of murine BM progenitors in FL,KL, IL-6, and IL-7 results in the detection of a CD122þ population withheightened NK precursor potential in the presence of IL-15. Furthermore,murine BM contains a CD122þ NK precursor population that differentiatesex vivo into mature NK cells in the presence of IL-15 or high-dose IL-2(Rosmaraki et al., 2001).

3.2.2. Differences between In Vitro–Derived NK Cells andCirculating NK Cells In Vivo

Notably, NK cells derived from in vitro cultures of progenitor cells with IL-15 or high doses of IL-2 are CD56bright with little or no expression of CD2or CD16, in contrast to the majority of NK cells in peripheral blood (Yuet al., 1998). Presumably, additional factors are required for acquisition ofthese additional surface markers of peripheral blood (PB) NK cells. One

218 brian becknell and michael a. caligiuri

soluble factor with an apparently key role in this late maturation of NKis IL-21 (Parrish-Novak et al., 2000). This T-cell–derived cytokine hasbeen shown to promote the acquisition of CD16 and killer immunoglo-bulin receptor (KIR) repertoire by NK cells during their development fromCD34þ cord blood precursors cultured in FL, KL, IL-7, and IL-15 (Sivoriet al., 2003).

3.3. Endogenous Pools of IL-15 for Human NKCell Development

What is the specific cell within bone marrow that is responsible for IL-15production, and in response to what physiologic cues does IL-15 production orsurface presentation occur? The answers to both questions are currentlyunknown, but the leading candidates for the IL-15–producing/presentingcell are stromal fibroblasts, monocytes, and dendritic cells. Analysis of long-term cultures of human BM stromal fibroblasts demonstrates detectable IL-15transcript by reverse transcription-polymerase chain reaction (RT-PCR) andIL-15 protein by enzyme-linked immunosorbent assay (ELISA), whereas IL-2 transcript and protein are consistently undetectable (Cluitmans et al., 1995;Mrozek et al., 1996). In addition, splenic fibroblasts have been shown toexpress surface IL-15, which is necessary and sufficient for cells to supportthe differentiation of autologous CD34þ progenitors to NK cells (Briard et al.,2002). CD14þ peripheral blood monocytes have been shown to producesoluble IL-15 that is detectable by ELISA in response to stimulation withIFN-g and lipopolysaccharide (LPS; Carson et al., 1995). In addition, flowcytometric analysis reveals that stimulation of CD14þ monocytes with IFN-gand LPS induces the surface expression of IL-15 in complex with IL-15Ra

(Dubois et al., 2002; Musso et al., 1999). Although IFN-g and LPS are impro-bable stimuli of IL-15 expression in the context of development, monocytesmaintain a substantial intracellular pool of this cytokine, and it is certainlyconceivable that alternative, currently unknown physiologic cues prompt thesecells to present IL-15 in trans to developing NK cells. Like monocytes,dendritic cells are capable of IL-15 production in response to pathologiccues such as LPS, double-stranded RNA, or type I interferons (Mattei et al.,2001; Santini et al., 2000), and these cells also upregulate expression ofthe IL-15Ra protein in response to these stimuli (Jinushi et al., 2003).Further study is required to implicate one or more of these cell types in IL-15 presentation/secretion during NK cell development and to identify themechanisms that regulate this process in vivo.

role of il-2 and il-15 in human nk cells 219

3.4. How Does IL-15 Promote Human NK Cell Development?

3.4.1. Does IL-15 Initiate an NK Cell Developmental Program?

How does IL-15 contribute to human NK cell development? Does it pro-mote proliferation/survival of an already committed NK precursor, or doesit initiate a developmental program that is required for NK ontogeny? Toaddress this critical question, Yu and colleagues examined the cell cycles status,apoptotic index, and total cell number of CD34þ HPCs cultured first in FL orKL for 3 weeks, followed by the addition of IL-15 for another 14 days (Yu et al.,1998). Whereas the absolute cell number did not change significantly overthe 2 weeks in IL-15, the percentage of CD56þ cells rose from <1 to >80%.However, analyses of cell cycle and apoptosis revealed that a constant fraction ofcells was proliferating and undergoing programmed cell death during this time.Thus, these authors argued, the increase in NK cells can be explained only by theability of IL-15 to promote the differentiation of an NK precursor population tomature NK cells. These data support the hypothesis that IL-15 initiates adevelopmental program that is required for NK cell development.

3.4.2. Does IL-15 Promote Survival/Outgrowth of aCommitted NK Precursor?

However, an alternative hypothesis is that some factor apart from IL-15 isresponsible for commitment to the NK lineage and that IL-15 promotes theselective survival and outgrowth of a small fraction of NK precursor cells. Insupport of this hypothesis, small numbers of NK cells can be found in micelacking IL-15 (Kawamura et al., 2003). In further support, NK cell develop-ment proceeds when NK cell–deficient IL-2/15Rb knockout mice aremodified to express a ubiquitous Bcl2 transgene—suggesting that IL-15 isimportant for survival of the developing NK cell but not lineage commitment(Minagawa et al., 2002). Indeed, as we discuss below, the homeostatic role ofIL-15 as a key survival factor for mature NK is absolutely essential for themaintenance of NK progeny (Cooper et al., 2002; Ranson et al., 2003).However, the same Bcl2 transgene fails to rescue NK cell deficiency in gc

knockout mice (Kondo et al., 1997), implicating either IL-15 itself (signalingthrough agc but not b) or another gc cytokine (IL-2, IL-4, IL-7, IL-9, orIL-21) in lineage commitment.

3.4.3. Can IL-2 Influence NK Cell Development?

Studies of IL-15Ra–/– mice suggest that IL-2 and IL-15 may perform non-redundant roles in guiding the acquisition of the repertoire of activa-ting and inhibitory receptors by developing NK cells (Kawamura et al., 2003).

220 brian becknell and michael a. caligiuri

IL-15Ra expression by developing NK cells is critical for acquisition of allLy49 receptors in the mouse, and mice deficient for IL-15 or IL-15Ra possessan incomplete Ly49 repertoire. Whereas IL-15 is unable to trigger Ly49receptor acquisition in the absence of IL-15Ra, IL-2 is able to rescue ex-pression of certain Ly49 receptors in the IL-15Ra–/– cells. This may occurthrough expression of the unique high-affinity IL-2Ra chain by certainmurine NK precursors, much akin to the selective expression of IL-2Ra byCD56bright NK in humans (Caligiuri et al., 1990). In this regard, it is interes-ting that low-dose IL-2 induces acquisition of killer immunoglobulin-like receptors (KIRs) by CD56bright NK from secondary lymphoid organs(Ferlazzo et al., 2004). Further studies are required to delineate the relativecontributions of IL-2 and IL-15 to NK receptor acquisition by murine andhuman NK cells.

4. Low-Dose IL-2 Therapy Expands the CD56bright NK Subset

4.1. IL-2Ra and c-Kit Are Uniquely Expressed by theCD56bright Human NK Subset

Studies in human peripheral blood have documented the existence of twodistinct NK populations based on cell surface density of the CD56 antigen(neural cell adhesion molecule, N-CAM) (Lanier et al., 1989; Nagler et al.,1989). Indeed, the majority (85–90%) of human peripheral blood NK cells areCD56dim and express high levels of FcgRIII (CD16). The remaining 10–15%of NK cells are CD56brightCD16dim/�. Studies of these subsets at the time oftheir discovery focused exclusively on their cytotoxicity. This work demon-strated that the CD56dim subset displays heightened cytolytic activity whencompared with the CD56bright subset (Lanier et al., 1986). CD56bright NK cellsuniquely express the IL-2Ra chain on their surface, conferring high affinity forIL-2 (Caligiuri et al., 1990). In contrast, CD56dim NK cells express the inter-mediate-affinity IL-2Rbg. This distinction permits CD56bright NK cells toselectively proliferate in response to low-dose (picomolar) quantities of IL-2.In addition, CD56bright cells uniquely express the c-Kit (CD117) receptortyrosine kinase on their surface (Matos et al., 1993). This finding led to studiesdemonstrating a prosurvival effect of c-Kit ligand (KL, stem cell factor) onCD56bright NK cells via increased expression of the antiapoptotic BCL-2 pro-tein (Carson et al., 1994b). These laboratory findings formed the rationale fortherapeutic administration of low-dose IL-2 (alone and more recently incombination with KL) in cancer and AIDS patients with advanced disease topromote the expansion and survival of CD56bright NK cells (Caligiuri et al.,1993; and M. A. Caligiuri, unpublished observations).

role of il-2 and il-15 in human nk cells 221

Why low-dose IL-2 in particular? Although IL-2 therapy at intermediate tohigh doses (nanomolar serum concentrations) has produced significant clinicalresponses in the setting of metastatic renal cell carcinoma and metastaticmelanoma, life-threatening side effects including hypotension and a capillaryleak syndrome have prevented its widespread use (reviewed by Rosenberg,2000). In contrast, regimens of low- and ultralow-dose IL-2 (picomolar serumconcentrations) have yielded expansions of CD56bright NK cells in phase I/IIstudies of patients with cancer and/or human immunodeficiency virus (HIV)infection without significant toxicity (reviewed by Fehniger et al., 2003). Proofof efficacy for this regimen in the clinical setting of immune deficiency and/orcancer must await the development and implementation of randomized phaseIII clinical trials, results of which are pending at this time.

4.2. Why Expand CD56bright NK Cells in Patients withCancer or Immunodeficiency?

The ability of IL-2–expanded CD56bright NK cells to lyse tumor and virus-infected target cells in an MHC-independent manner, together with theircapacity to elaborate cytokines that promote activation of other immunecells, provide a rationale for their expansion in these clinical settings.For example, during HIV infection, depletion of CD4þ T cells results indrastic reductions in IL-2 and IFN-g, two cytokines that are normally re-quired to stimulate CD8þ T cells and initiate an effective cellular immuneresponse. This is one mechanism responsible for increased incidence of op-portunistic infections and malignancy in patients with AIDS (Fauci, 1993).IL-2–expanded CD56bright NK cells produce significant quantities of IFN-g,permitting these cells to serve as surrogates for the CD4þ helper T-cell type 1(Th1) response (Khatri et al., 1998).

4.3. Mechanisms of CD56bright NK Cell Expansion inPatients Receiving Low-Dose IL-2

Administration of low-dose IL-2 to patients results in picomolar concentra-tions of this cytokine that selectively expand CD56bright NK cells within 4–6weeks of therapy, in accordance with the expression of the IL-2 high-affinityreceptor by this population (Bernstein et al., 1995; Caligiuri et al., 1990, 1993).Studies by Fehniger and colleagues focused on the mechanism for the specificoutgrowth of this population, specifically whether IL-2 acts on matureCD56bright NK cell homeostasis or whether it supports the differentiation ofNK cells from progenitor populations in the bone marrow (Fehniger et al.,2000a). Whereas a percentage of CD56bright NK cells proliferates modestly inlow-dose IL-2 in vitro, they exit the cell cycle within 6 days. Moreover, NK

222 brian becknell and michael a. caligiuri

cells from patients treated with low-dose IL-2 fail to transit the cell cycle. Incontrast, low concentrations of IL-2 do promote survival of CD56bright NKcells, in accordance with earlier observations. The presence of a functionalhigh-affinity IL-2 receptor is required for both processes—proliferation andsurvival. Whereas proliferation in response to IL-2 is limited, the survivalbenefit of IL-2 is long lasting and will maintain CD56bright NK cells forweeks at a time. Strikingly, in vitro culture of bone marrow CD34þ progenitorcells in low-dose IL-2 significantly enhances their differentiation to mature NKcells. Thus, the most significant effect of IL-2 on NK cell homeostasis may beits ability to induce the differentiation of NK cells from progenitor popula-tions. The kinetics of IL-2–mediated NK expansion in patients in vivo parallelthose of IL-2–mediated differentiation from bone marrow progenitors in vitro.Significantly, CD56bright cells cultured in low-dose IL-2 show enhanced cyto-kine production (interferon-g) and cytotoxicity compared with cells cultured inmedium alone (Fehniger et al., 2000a). Thus, low-dose IL-2 likely exerts threemajor effects on NK cells in patients receiving this cytokine: (1) increaseddifferentiation from progenitor populations, (2) increased survival, and(3) increased effector function.

5. IL-2/IL-15 and NK Cell Survival

5.1. IL-15 Specifically Promotes NK Cell Survival In Vitroand In Vivo

When cultured under serum-free conditions, NK cells rapidly undergoprogrammed cell death, or apoptosis (Carson et al., 1997a). This findingsuggests that extracellular factors are responsible for sustaining NK cell sur-vival in vivo. Despite the persistence of NK cells in IL-2–deficient mice,treatment with an antibody to IL-2Rb resulted in elimination of murine NKcells (Kundig et al., 1993; Schorle et al., 1991; Tanaka et al., 1993). The discoveryof IL-15 offered a potential explanation for these paradoxical findings. Indeed,addition of low (picomolar) concentrations of IL-15 sustains human NK cellsurvival for over 1 week in serum-free medium (SFM) (Carson et al., 1997a). Asnoted previously, IL-2 also exhibits a prosurvival effect at low doses (Carsonet al., 1997a; Fehniger et al., 2000a). In contrast, other cytokines that share theIL-2Rg chain (i.e., IL-4, IL-7, IL-9, and IL-13) fail to promote NK survival inSFM (Carson et al., 1997a). These results are consistent with the ability of theIL-2Rb subunit to promote cell survival (Miyazaki et al., 1995). Although thesein vitro results suggest a model whereby soluble IL-15 interacts with its high-affinity IL-15R on NK cells, elegant experiments in IL-15Ra–deficient mice

role of il-2 and il-15 in human nk cells 223

suggest that non-NK cells expressing IL-15Ra, in complex with IL-15, areresponsible for presentation of this cytokine in trans to NK cells expressingthe IL-2/15Rb and gc chains in vivo. This model is supported by the ability ofIL-15Ra–/– NK cells to persist on adoptive transfer in wild-type mice, whereaswild-type NK cells fail to survive on transfer to IL-15Ra–/– hosts (Koka et al.,2003).

5.2. IL-15 Promotes Survival Through IncreasedExpression of Bcl-2

How does IL-15 promote cell survival? Culture in SFM and IL-15 maintainslevels of the anti-apoptotic BCL-2 protein in human NK cells, and treatmentof these cells with Bcl-2 antisense oligonucleotide results in downregulation ofBCL-2 protein levels and decreased cell viability (Carson et al., 1997a).Consistent with these observations, adoptive transfer of NK cells from wild-type mice into IL-15–deficient mice results in their rapid disappearancewithin 5 days, whereas transplantation of Bcl-2 transgenic NK cells intoIL-15–deficient recipients results in enhanced survival (Cooper et al., 2002;Ranson et al., 2003). Moreover, when wild-type NK cells are examinedafter transfer into IL-15–deficient mice, their BCL-2 protein expression isdecreased (Ranson et al., 2003). Thus, the prosurvival effects of IL-15 are atleast partially attributed to its ability to enhance BCL-2 levels. The signalingevents responsible for increased BCL-2 in response to IL-15 have not beenelucidated.

5.3. Can Enforced Expression of BCL-2 Serve as a Proxyfor IL-2/IL-15 In Vivo?

With the implication of BCL-2 as one downstream target of IL-15 in NK cells,several investigators have sought to determine whether restoration of BCL-2expression is sufficient to rescue the NK cell deficiences observed in micegenetically deficient in the shared components of the IL-15R/IL-2R. Indeed,enforced expression of BCL-2 rescues the deficiency in NK cell numbersobserved in mice lacking the IL-2Rb chain. Although IL-2Rb–/– Bcl-2tg NKcells are no different from wild-type NK cells in their receptor repertoire andIFN-g production, they completely lack cytotoxic activity (Minagawa et al.,2002). This underlines the pleiotropic roles of IL-2/IL-15 in NK cell ontogeny,homeostasis, and effector function. However, enforced expression of BCL-2 inthe absence of the gc chain does not rescue the NK cell deficiency observed ing�=�

c mice (Kondo et al., 1997).

224 brian becknell and michael a. caligiuri

5.4. IL-2 and IL-15 Transmit BCL-2–IndependentSurvival Signals

Another way in which IL-15 and IL-2 influence NK cell survival appears to bethrough activation of the cytoplasmic tyrosine kinase, Syk (Jiang et al., 2003).Syk-binding sites are found in the IL-2Rb chain and IL-15Ra chain (Bulanovaet al., 2001; Minami et al., 1995; Qin et al., 1994). IL-2–mediated activation ofSyk initiates a PI-3 kinase–dependent transduction cascade that culminates inactivation of Akt/protein kinase B (PKB). Consistent with this pathway, phar-macologic and dominant-negative inhibitors of Syk or PI-3 kinase result inimpaired IL-2–mediated survival and initiation of apoptosis (Jiang et al., 2003).However, mice deficient in Syk maintain normal NK cell numbers (Colucciet al., 1999). These experiments underline the potential differences betweenhuman and murine NK cells, as well as the existence of multiple, functionallyredundant pathways to promote NK cell survival.

6. IL-2/IL-15 and NK Cell Proliferation

6.1. In Vitro Evidence Supports a Role for IL-2/IL-15 inNK Cell Proliferation

The role of IL-2 in proliferation of NK cells was appreciated long before thediscovery of IL-15 through studies of mice transgenic for IL-2 or its high-affinity Tac chain that showed selective proliferation of NK cells but notT cells (Biron et al., 1990; Ishida et al., 1989). In vitro studies of human NKcells revealed that, at low concentrations, IL-2 selectively induces proliferationof the CD25þCD56bright NK subset, because of the selective expression of thehigh-affinity IL-2Rabg ternary complex by this population (Caligiuri et al.,1990; Nagler et al., 1990). Increasing IL-2 concentration to full saturation ofthe high-affinity IL-2R and partial saturation of the intermediate-affinityIL-2Rbgc maximize this proliferation. These effects of IL-2 are diminishedin the presence of neutralizing antibodies to IL-2Ra or IL-2Rb (Carson et al.,1994a). With the discovery of IL-15 and its shared signaling through compo-nents of the IL-2R, it became necessary to compare its influence on NKproliferation with that of IL-2. Not unexpectedly, IL-15 promotes the prolifer-ation of CD56bright NK cells, and this activity is decreased with coincubationwith neutralizing antibodies to IL-2Rb. Consistent with the private usage ofIL-2Ra by its cognate cytokine, neutralizing antibodies to IL-2Ra have noconsequence on IL-15–mediated proliferation of CD56bright NK cells (Carsonet al., 1994a). In contrast to these observations in CD56bright NK cells,CD56dim cells exhibit little to no proliferation in response to a broad rangeof IL-2 or IL-15. (Heterogeneous expression of CD16 within the CD56bright

role of il-2 and il-15 in human nk cells 225

NK subset permits further stratification of its proliferation in response tolow-dose IL-2: CD16dimCD56bright NK cells exhibit greater proliferation inresponse to activation of the high-affinity IL-2R, relative to CD16neg

CD56bright NK cells; Carson et al., 1997b.)

6.2. NK Proliferation In Vivo: Fully Independent of IL-2/IL-15?

Despite the significant impact of IL-2 and IL-15 on CD56bright NK cellproliferation in vitro, cell cycle analysis of CD56bright NK cells obtained frompatients receiving low-dose IL-2 in vivo demonstrates that these NK cells arenot actively dividing (Fehniger et al., 2000a). These observations, together withthe observation that low-dose IL-2 promotes the development of CD56bright

NK cells from CD34þ progenitors, have led to the hypothesis that low-doseIL-2 exerts its actions chiefly through enhanced differentiation of NK cellsfrom a progenitor pool and enhanced NK survival, rather than via in vivoproliferation of mature NK cells (Fehniger et al., 2000a). Thus, the preciseroles of IL-2 and IL-15 in human NK cell proliferation in vivo remain unclearat this time. Indeed, it is conceivable that these cytokines do not play a rolein NK cell proliferation and that they serve critical roles in promoting NKdifferentiation and survival. This hypothesis is supported by studies of NK cellhomeostasis in mice. In lymphocyte-replete mice, bromodeoxyuridine (BrdU)labeling experiments indicate that mature NK cell turnover is virtually nonex-istent and comparable to that of memory T cells (Jamieson et al., 2004).Whereas adoptive transfer to NK cell–replete hosts results in little or noproliferation, NK cells undergo a profound ‘‘homeostatic’’ proliferation ontransplantation to alymphoid hosts (Jamieson et al., 2004; Ranson et al.,2003). Whereas adoptive transfer to IL-15–/– hosts profoundly mitigates NKcell recovery, the degree of their homeostatic proliferation is unaffected by theabsence of IL-15 (Jamieson et al., 2004). These findings argue against a role forIL-15 in homeostatic proliferation but reinforce the essential, nonredundantrole of this cytokine in NK cell survival.

7. IL-2/IL-15 and NK Cell Apoptosis

Whereas IL-15 promotes the survival of resting NK cells in the periphery, thecombination of IL-15 (or IL-2) with IL-12 results in NK cell activation, withconsiderable elaboration of cytokines, such as IFN-g and TNF-a, followedby programmed cell death, or apoptosis (Ross and Caligiuri, 1997). Themechanism of apoptosis in response to IL-12 and IL-15 appears to be viathe production of TNF-a by NK cells, since neutralizing this cytokine or itsassociation with the p80 TNF-a receptor partially abrogates this process (Ross

226 brian becknell and michael a. caligiuri

and Caligiuri, 1997). The in vivo significance of these findings is unknown, butthey likely represent a homeostatic mechanism to downregulate the immuneresponse and thereby prevent the establishment of damaging proinflammatorypositive feedback circuits between NK cells and other innate effectors.

8. IL-2/IL-15 and NK Cell Cytotoxicity

8.1. IL-2 and IL-15 Enhance NK Cell Cytolytic Activity

NK cells are capable of a wide range of cytotoxic mechanisms to lyse targetcells (for a review, see Djeu et al., 2002). These range from the release ofintracellular granules containing cytotoxic proteins (e.g., perforin, granzymes),to the triggering of apoptotic cascades within target cells via the release ofcytotoxic cytokines (e.g., TNF-a) or the direct engagement of death domainreceptors (e.g., Fas) on target cells. Although an in-depth description of thesecytotoxic mechanisms is beyond the scope of this review, it is important to notethe key roles of IL-2 and IL-15 in enhancing NK cell cytolytic activity. Indeed,exposure of NK cells to IL-2 concentrations that saturate the intermediate-affinity IL-2 receptor results in lymphokine-activated killer (LAK) activitytoward tumor cells that otherwise resist NK cell–mediated lysis (Caligiuriet al., 1990). This LAK activity is inhibited in the presence of neutralizingantibodies to the IL-2Rb subunit (Phillips et al., 1989). Preincubation ofCD56dim NK cells with IL-15 triggers cytolysis of NK cell–resistant COLO205 target cells in a dose-dependent manner, which is abrogated in thepresence of anti-IL-2Rb mAb (Carson et al., 1994a). In this study, identicalresults were obtained with IL-2.

8.2. Molecular Basis of IL-2–Enhanced NK Cell Cytotoxicity

Advances in our knowledge of the signal transduction pathways initiated byIL-2 have shed considerable light on the molecular basis of this LAK activity.One mechanism for IL-2–enhanced NK cell cytotoxicity is via upregulation ofthe pore-forming cytolytic effector molecule, perforin. This occurs in part viaSTAT5 association with two enhancer elements in the perforin promoter(Zhang et al., 1999). In addition to STAT5 activation, IL-2 promotes thecytoplasmic stabilization and nuclear translocation of NF-kB p50/p65 com-plexes, which also associate with and upregulate perforin promoter activity(Zhou et al., 2002). The importance of NF-kB for perforin expression isunderscored in patients with hypohidrotic ectodermal dysplasia (HED),who possess mutations in the NEMO/IKK-g gene and display impaired NKcell cytolytic activity, leading to an increased susceptibility to opportunistic

role of il-2 and il-15 in human nk cells 227

infections (Doffinger et al., 2001). The NEMO protein is one component ofthe complex responsible for I-kB phosphorylation and NF-kB activation, andNK cells from HED patients display reduced NF-kB activity. Treatment ofone HED patient with IL-2 partially rescued NF-kB activation and NKcytolytic activity (Orange et al., 2002).

8.3. Role of IL-2 and IL-15 in the Developmental Acquisitionof Cytolytic Function

Studies of human NK cell development point to a vital role for IL-2/IL-15 inthe acquisition of cytolytic activity. Barao and colleagues derived NK cellsin vitro using IL-7, which lack cytotoxicity toward K562 targets. Perforin andgranzyme B levels in these IL-7–derived NK cells were comparable to thosefound in cytotoxic counterparts derived in the presence of IL-2 or IL-15,suggesting that the role of IL-2/IL-15 is not merely to upregulate the expres-sion of these two proteins. In contrast, IL-7–derived NK cells fail to expressthe surface molecule, LFA-1. LFA-1 is a cell adhesion molecule on NK cellsthat engages ICAM-1 on target cells to establish a critical interaction, calledconjugate formation, that must exist prior to the release of cytotoxic granulesby the NK cell. Exposure of IL-7–derived NK cells to IL-2 or IL-15 resultsin their acquisition of LFA-1 surface expression and cytotoxicity that is neu-tralized by antibodies to LFA-1 subunits CD11a and CD18 (Barao et al.,2003). Thus, in addition to enhancing cytotoxicity in mature NK cells, IL-2/IL-15 may serve an essential role in the acquisition of NK cell cytolyticmachinery during NK cell development. This hypothesis is supported bythe findings of Minagawa and colleagues, who sought to rescue the NK celldeficiency observed in IL-2Rb–/– mice through the overexpression of BCL-2.These investigators found that, whereas enforced BCL-2 expression restoredNK cell numbers in IL-2Rb–/– mice, these NK cells completely lacked cytolyticactivity (Minagawa et al., 2002)—an observation that is consistent with anabsolute and nonredundant role for IL-2/IL-15 signaling in the acquisitionand/or maintenance of NK cell cytotoxicity.

9. IL-2/IL-15 and NK Cell Cytokine Production

9.1. IL-2 and IL-15, in Synergy with Monokines, ElicitSignificant Cytokine Production by NK Cells

Numerous studies have documented the roles of IL-2 and IL-15 in elicitingthe production of various immunoregulatory cytokines by human NK cells,including IFN-g, TNF-a, and GM-CSF (for a review, see Cooper et al., 2001).

228 brian becknell and michael a. caligiuri

Whereas IL-2 or IL-15 can elicit little IFN-g production alone, both synergizeequally well with IL-12 to produce substantial quantities of this cytokine(Carson et al., 1994a). In addition to IFN-g production, IL-12 synergizeswith IL-15 or IL-2 in the synthesis of other cytokines, such as TNF-a andGM-CSF. Subtle differences between the ability of IL-15 and IL-2 to elicitcytokine production by NK cells have been noted. For example, Carson andcolleagues observed that GM-CSF production in response to IL-2 alone wasroutinely twice that observed with IL-15, even when both cytokines wereadministered at concentrations that rule out selective use of private high-affinity receptors (Carson et al., 1994a).

9.2. CD56bright Subset is Chiefly Responsible for CytokineProduction by Human NK Cells

Subsequent studies have demonstrated that the CD56bright NK subset pro-duces significantly greater quantities of IFN-g and GM-CSF than CD56dim

NK cells in response to stimulation with IL-2/IL-15 and IL-12 in vitro (Cooperet al., 2001b; Fehniger et al., 1999, 2003). Moreover, because of its uniqueexpression of the HA IL-2R, CD56bright NK cells produce IFN-g in thepresence of low (picomolar) concentrations of IL-2. This raises the possibilitythat this NK subset may participate in cross-talk with T cells in secondarylymphoid organs, where these cells exist in close proximity (Fehniger et al.,2003). Together, these findings are consistent with a model in which CD56bright

NK cells serve to elaborate cytokines that regulate the response of otherimmune cells—such as monocytes, dendritic cells, and other lymphocytes(Cooper et al., 2001a, 2004). IL-2/IL-15 treatment of CD56bright NK cellsresults in the production of different cytokines, depending on the local milieu(Cooper et al., 2001b). For example, the combination of IL-15 and IL-18 resultsin optimal production of GM-CSF but minuscule expression of IL-10; incontrast, treatment with IL-15 and IL-12 results in optimal production of IL-10 but lower production of GM-CSF (Cooper et al., 2001b).

10. NK Cell–Immune Cell Interactions and Modulation by IL-2/IL-15

10.1. CD56bright NK Cells Are Enriched in Secondary LymphoidOrgans, Where T-Cell–Derived IL-2 Can Influence Their Function

CD56bright NK cells express two homing molecules, L-selectin and CCR7, thatdirect the migration of these lymphocytes to peripheral lymph nodes (LNs) viahigh endothelial venules (Campbell et al., 2001; Fehniger et al., 2003; Freyet al., 1998). Accordingly, CD56bright NK cells are enriched in secondary

role of il-2 and il-15 in human nk cells 229

lymphoid organs, particularly LNs and tonsils (Fehniger et al., 2003; Ferlazzoet al., 2004). Immunohistochemical analysis demonstrates that CD56bright

NK cells localize to the parafollicular regions of LNs, where they are inclose proximity to T cells and activated DCs. This observation suggeststhat CD56bright NK cells, which possess the HA IL-2R, may respond toT-cell–derived IL-2 in vivo (Fehniger et al., 2003). In support of this model,coculture of CD56bright NK cells with T cells and APCs indicates that endoge-nous T-cell–derived IL-2 is capable of acting through the HA IL-2R andcostimulates IFN-g production by CD56bright NK cells in the presence ofIL-12. These findings suggest that CD56bright NK cells, through their elabora-tion of IFN-g, may provide an important bridge between innate and adaptiveimmunity (Cooper et al., 2004).

10.2. Low-Dose IL-2 Promotes the Maturation of CD56bright

NK Cells from Secondary Lymphoid Organs

In addition to IFN-g production, low-dose IL-2 may serve to promote thefunctional maturation of LN CD56bright NK cells (Ferlazzo et al., 2004). Thishypothesis is suggested by the finding that administration of low-dose IL-2 topurified LN CD56bright NK cells causes these cells to acquire the surfacephenotype of a CD56dim NK cell: CD16þKIRþNCRþ. In addition, treatmentof LN or tonsillar NK cells with low-dose IL-2 induces these cells to expressperforin and to exhibit cytolytic activity toward NK cell–sensitive targets(Ferlazzo et al., 2004). In contrast, treatment of peripheral blood CD56bright

NK cells with low-dose IL-2 does not induce comparable changes. Thisdifference may reflect the fact that peripheral blood CD56bright NK cellshave yet to enter secondary lymphoid organs. This model of NK cell matura-tion is supported by the finding that circulating CD56bright NK cells expresslymph node–homing molecules (CCR7 and L-selectin), whereas CD56bright

NK cells in secondary lymphoid organs do not express these proteins(Fehniger et al., 2003; Ferlazzo et al., 2004).

Murine bone-marrow–derived dendritic cells (mBMDCs) are capableof IL-2 production on stimulation with LPS (Granucci et al., 2001). DCsfrom IL-2–/– mice exhibit a severe impairment in their ability to stimulateCD4þ and CD8þ T-cell proliferation in a mixed lymphocyte reaction assay,and these IL-2–deficient DCs are reportedly inefficient in activating NK cellresponses (Granucci et al., 2003). Thus, migration of CD56bright NK cells fromperipheral blood to secondary lymphoid organs may influence thedevelopment and function of these cells, under the influence of T-cell– and/or DC-derived IL-2.

230 brian becknell and michael a. caligiuri

10.3. Dendritic Cells and Monocytes Produce IL-15, WhichInfluences NK Cell Cytolytic Function and Cytokine Production

In addition to IL-2, dendritic cells produce IL-15 in response to type Iinterferons (IFN-a and IFN-b), which can increase NK cell activity. Forexample, human and murine monocyte-derived DCs produce IL-15 and IL-15Ra in response to IFN-a treatment (Mattei et al., 2001; Santini et al., 2000).This IL-15/IL-15Ra signals in autocrine fashion to elicit additional productionof IFN-a and upregulation of MICA on the DCs, which associates with theactivating NKG2D receptor on NK cells to promote their cytolytic functionand cytokine production (Jinushi et al., 2003). This mechanism is important foractivation of NK cells by DCs, as underlined by impaired NK activity inpatients with hepatitis C, whose DCs are unable to produce IL-15 in responseto type I interferon treatment (Jinushi et al., 2003).

Flow cytometric analysis of monocytes after stimulation with LPS and IFN-greveals surface expression of IL-15 in complex with IL-15Ra (Dubois et al.,2002; Musso et al., 1999). Coculture experiments of macrophages and auto-logous NK reveal that LPS is sufficient to promote robust IFN-g production—particularly by CD56bright NK cells—which can be inhibited by neutralizingantibodies to IL-12 and IL-15 (Carson et al., 1995; Cooper et al., 2001b). Thus,macrophages produce IL-15 (along with other monokines, such as IL-12, IL-18,and IL-1b) in response to LPS stimulation, which costimulate the production ofIFN-g by NK cells. This IFN-g greatly increases the sensitivity of the macro-phage to LPS, such that subsequent exposure results in cytokine-induced shockand mortality, because of the unhindered release of proinflammatory cytokines.This cycle of sequential priming with IFN-g and LPS exposure is termed thegeneralized Schwartzman reaction (Brozna, 1990). Neutralizing antisera toIL-15 or the IL-2Rb subunit significantly reduce IFN-g levels and subsequentlethality in a murine model of the Schwartzman reaction, suggesting that thiscytokine plays an instrumental role in this process (Fehniger et al., 2000b).Thus, monocyte-derived IL-15 serves a critical role in potentiating IFN-gproduction by NK cells both in vitro and in vivo, and therapeutic strategiesaimed at disrupting this proinflammatory loop may prove efficacious.

11. Conclusions

In this review, we have summarized our current understanding of the rolesof IL-2 and IL-15 in human NK cell development, homeostasis, and function,as well as the potential physiologic roles of these two cytokines in orchestratingthe interactions between NK cells and other immune cells. We have summar-ized these findings in Figs. 2 and 3.

Figure 2 IL-2/IL-15 facilitate NK cell differentiation and contribute to mature NK cellhomeostasis. IL-2 and IL-15 have the ability to promote differentiation of mature NK cells fromNK precursors and also contribute to the homeostasis of mature NK cells, acting variously topromote survival, proliferation, or apoptosis. The physiologic role of these cytokines in prolifera-tion and apoptosis of NK cells is unknown. See text for details.

Figure 3 IL-2/IL-15 contribute to mature NK cell function. IL-2 and IL-15 act at various levels toimpact human NK cell function. They increase the cytolytic activity of CD56bright and CD56dim

NK cells. They greatly enhance cytokine secretion by the CD56bright human NK subset. Specifi-cally in secondary lymphoid tissues (e.g., lymph nodes), IL-2 has been shown to promote theacquisition of KIR and CD16 expression by CD56bright NK cells. See text for details.

role of il-2 and il-15 in human nk cells 231

232 brian becknell and michael a. caligiuri

Although our comprehension of these topics has grown tremendously sincethe discovery of IL-15, several unresolved questions remain to challenge thisfield for years to come. The discovery of IL-15 trans presentation by IL-15Ra,together with the observation that IL-15Ra can signal independently of IL-2Rbg, offer the possibility of reciprocal communication between NK cells andIL-15–presenting cells, such as DCs and monocytes. A more detailed inspec-tion of IL-15Ra signaling and IL-15 presentation (cis versus trans) will un-doubtedly aid our understanding of the specific effects of this cytokine on NKcell ontogeny, function, and homeostasis. Moreover, although IFN-a, IFN-g,and LPS may trigger IL-15 presentation in the context of infectious disease,these are not the likeliest physiologic cues that elicit IL-15 during NK celldevelopment or NK cell survival in the periphery, and defining novel positiveand negative regulators of IL-15 presentation is a major challenge for thefuture of this field. Finally, the discovery of secondary lymphoid organsas havens for CD56bright NK cells, together with the unexpected demonstra-tion that low doses of IL-2 promote their differentiation to CD56dim KIRþ

CD16þ cells, suggest that IL-2 and the microenvironment of these tissuesmay play a qualitative role in the development and maturation of humanNK cells.

Acknowledgments

This work was supported by P01 CA95426 and R01CA68458.

References

Anderson, D. M., Kumaki, S., Ahdieh, M., Bertles, J., Tometsko, M., Loomis, A., Giri, J., Copeland,N. G., Gilbert, D. J., Jenkins, N. A., et al. (1995). Functional characterization of the humaninterleukin-15 receptor a chain and close linkage of IL15RA and IL2RA genes. J. Biol. Chem.270, 29862–29869.

Barao, I., Hudig, D., and Ascensao, J. L. (2003). IL-15-mediated induction of LFA-1 is a late steprequired for cytotoxic differentiation of human NK cells from CD34þ Lin� bone marrow cells. J.Immunol. 171, 683–690.

Bernstein, Z. P., Porter, M. M., Gould, M., Lipman, B., Bluman, E. M., Stewart, C. C., Hewitt,R. G., Fyfe, G., Poiesz, B., and Caligiuri, M. A. (1995). Prolonged administration of low-doseinterleukin-2 in human immunodeficiency virus-associated malignancy results in selective ex-pansion of innate immune effectors without significant clinical toxicity. Blood 86, 3287–3294.

Biron, C. A., Young, H. A., and Kasaian, M. T. (1990). Interleukin 2-induced proliferation ofmurine natural killer cells in vivo. J. Exp. Med. 171, 173–188.

Briard, D., Brouty-Boye, D., Azzarone, B., and Jasmin, C. (2002). Fibroblasts from human spleenregulate NK cell differentiation from blood CD34þ progenitors via cell surface IL-15.J. Immunol. 168, 4326–4332.

Brozna, J. P. (1990). Shwartzman reaction. Semin. Thromb. Hemost. 16, 326–332.

role of il-2 and il-15 in human nk cells 233

Bulanova, E., Budagian, V., Pohl, T., Krause, H., Durkop, H., Paus, R., and Bulfone-Paus, S.(2001). The IL-15R a chain signals through association with Syk in human B cells. J. Immunol.167, 6292–6302.

Bulfone-Pau, S. S., Bulanova, E., Pohl, T., Budagian, V., Durkop, H., Ruckert, R., Kunzendorf, U.,Paus, R., and Krause, H. (1999). Death deflected: IL-15 inhibits TNF-a-mediated apoptosis infibroblasts by TRAF2 recruitment to the IL-15Ra chain. FASEB J. 13, 1575–1585.

Burton, J. D., Bamford, R. N., Peters, C., Grant, A. J., Kurys, G., Goldman, C. K., Brennan, J.,Roessler, E., and Waldmann, T. A. (1994). A lymphokine, provisionally designated interleukinT and produced by a human adult T-cell leukemia line, stimulates T-cell proliferation and theinduction of lymphokine-activated killer cells. Proc. Natl. Acad. Sci. USA 91, 4935–4939.

Caligiuri, M. A., Murray, C., Robertson, M. J., Wang, E., Cochran, K., Cameron, C., Schow, P.,Ross, M. E., Klumpp, T. R., Soiffer, R. J., et al. (1993). Selective modulation of human naturalkiller cells in vivo after prolonged infusion of low dose recombinant interleukin 2. J. Clin. Invest.91, 123–132.

Caligiuri, M. A., Zmuidzinas, A., Manley, T. J., Levine, H., Smith, K. A., and Ritz, J. (1990).Functional consequences of interleukin 2 receptor expression on resting human lymphocytes:Identification of a novel natural killer cell subset with high affinity receptors. J. Exp. Med. 171,1509–1526.

Campbell, J. J., Qin, S., Unutmaz, D., Soler, D., Murphy, K. E., Hodge, M. R., Wu, L., and Butcher,E. C. (2001). Unique subpopulations of CD56þ NK and NK–T peripheral blood lymphocytesidentified by chemokine receptor expression repertoire. J. Immunol. 166, 6477–6482.

Cao, X., Shores, E. W., Hu-Li, J., Anver, M. R., Kelsall, B. L., Russell, S. M., Drago, J., Noguchi,M., Grinberg, A., Bloom, E. T., et al. (1995). Defective lymphoid development in mice lackingexpression of the common cytokine receptor g chain. Immunity 2, 223–238.

Carayannopoulos, L. N., and Yokoyama, W. M. (2004). Recognition of infected cells by naturalkiller cells. Curr. Opin. Immunol. 16, 26–33.

Carson, W. E., Fehniger, T. A., Haldar, S., Eckhert, K., Lindemann, M. J., Lai, C. F., Croce, C. M.,Baumann, H., and Caligiuri, M. A. (1997a). A potential role for interleukin-15 in the regulationof human natural killer cell survival. J. Clin. Invest. 99, 937–943.

Carson, W. E., Fehniger, T. A., and Caligiuri, M. A. (1997b). CD56bright natural killer cell subsets:Characterization of distinct functional responses to interleukin-2 and the c-Kit ligand. Eur.J. Immunol. 27, 354–360.

Carson, W. E., Giri, J. G., Lindemann, M. J., Linett, M. L., Ahdieh, M., Paxton, R., Anderson, D.,Eisenmann, J., Grabstein, K., and Caligiuri, M. A. (1994a). Interleukin (IL) 15 is a novelcytokine that activates human natural killer cells via components of the IL-2 receptor. J. Exp.Med. 180, 1395–1403.

Carson, W. E., Haldar, S., Baiocchi, R. A., Croce, C. M., and Caligiuri, M. A. (1994b). The c-Kitligand suppresses apoptosis of human natural killer cells through the upregulation of bcl-2. Proc.Natl. Acad. Sci. USA 91, 7553–7557.

Carson, W. E., Ross, M. E., Baiocchi, R. A., Marien, M. J., Boiani, N., Grabstein, K., and Caligiuri,M. A. (1995). Endogenous production of interleukin 15 by activated human monocytes is criticalfor optimal production of interferon-g by natural killer cells in vitro. J. Clin. Invest. 96, 2578–2582.

Cluitmans, F. H., Esendam, B. H., Landegent, J. E., Willemze, R., and Falkenburg, J. H. (1995).Constitutive in vivo cytokine and hematopoietic growth factor gene expression in the bonemarrow and peripheral blood of healthy individuals. Blood 85, 2038–2044.

Colucci, F., Turner, M., Schweighoffer, E., Guy-Grand, D., Di Bartolo, V., Salcedo, M., Tybulewicz,V. L., and Di Santo, J. P. (1999). Redundant role of the Syk protein tyrosine kinase in mouse NKcell differentiation. J. Immunol. 163, 1769–1774.

234 brian becknell and michael a. caligiuri

Cooper, M. A., Bush, J. E., Fehniger, T. A., VanDeusen, J. B., Waite, R. E., Liu, Y., Aguila, H. L.,and Caligiuri, M. A. (2002). In vivo evidence for a dependence on interleukin 15 for survival ofnatural killer cells. Blood 100, 3633–3638.

Cooper, M. A., Fehniger, T. A., Fuchs, A., Colonna, M., and Caligiuri, M. A. (2004). NK cell andDC interactions. Trends Immunol. 25, 47–52.

Cooper, M. A., Fehniger, T. A., and Caligiuri, M. A. (2001a). The biology of human natural killer-cell subsets. Trends Immunol. 22, 633–640.

Cooper, M. A., Fehniger, T. A., Turner, S. C., Chen, K. S., Ghaheri, B. A., Ghayur, T., Carson, W. E.,and Caligiuri, M. A. (2001b). Human natural killer cells: A unique innate immunoregulatory rolefor the CD56-bright subset. Blood 15, 3146–3151.

DiSanto, J. P., Keever, C. A., Small, T. N., Nicols, G. L., O’Reilly, R. J., and Flomenberg, N. (1990).Absence of interleukin 2 production in a severe combined immunodeficiency disease syndromewith T cells. J. Exp. Med. 171, 1697–1704.

DiSanto, J. P., Muller, W., Guy-Grand, D., Fischer, A., and Rajewsky, K. (1995). Lymphoiddevelopment in mice with a targeted deletion of the interleukin 2 receptor g chain. Proc. Natl.Acad. Sci. USA 92, 377–381.

Djeu, J. Y., Jiang, K., and Wei, S. (2002). A view to a kill: Signals triggering cytotoxicity. Clin.Cancer Res. 8, 636–640.

Doffinger, R., Smahi, A., Bessia, C., Geissmann, F., Feinberg, J., Durandy, A., Bodemer, C.,Kenwrick, S., Dupuis-Girod, S., Blanche, S., Wood, P., Rabia, S. H., Headon, D. J., Overbeek,P. A., Le Deist, F., Holland, S. M., Belani, K., Kumararatne, D. S., Fischer, A., Shapiro, R.,Conley, M. E., Reimund, E., Kalhoff, H., Abinun, M., Munnich, A., Israel, A., Courtois, G., andCasanova, J. L. (2001). X-linked anhidrotic ectodermal dysplasia with immunodeficiency iscaused by impaired NF-kB signaling. Nat. Genet. 27, 277–285.

Dorshkind, K., Pollack, S. B., Bosma, M. J., and Phillips, R. A. (1985). Natural killer (NK) cellsare present in mice with severe combined immunodeficiency (SCID). J. Immunol. 134,3798–3801.

Dubois, S., Mariner, J., Waldmann, T. A., and Tagaya, Y. (2002). IL-15Ra recycles and presentsIL-15 in trans to neighboring cells. Immunity 17, 537–547.

Fauci, A. S. (1993). Multifactorial nature of human immunodeficiency virus disease: Implicationsfor therapy. Science 262, 1011–1018.

Fehniger, T. A., Bluman, E. M., Porter, M. M., Mrozek, E., Cooper, M. A., VanDeusen, J. B.,Frankel, S. R., Stock, W., and Caligiuri, M. A. (2000a). Potential mechanisms of human naturalkiller cell expansion in vivo during low-dose IL-2 therapy. J. Clin. Invest. 106, 117–124.

Fehniger, T. A., Cooper, M. A., Nuovo, G. J., Cella, M., Facchetti, F., Colonna, M., and Caligiuri,M. A. (2003). CD56bright natural killer cells are present in human lymph nodes and are activatedby T cell-derived IL-2: A potential new link between adaptive and innate immunity. Blood 101,3052–3057.

Fehniger, T. A., Shah, M. H., Turner, M. J., VanDeusen, J. B., Whitman, S. P., Cooper, M. A.,Suzuki, K., Wechser, M., Goodsaid, F., and Caligiuri, M. A. (1999). Differential cytokine andchemokine gene expression by human NK cells following activation with IL-18 or IL-15 incombination with IL-12: Implications for the innate immune response. J. Immunol. 162,4511–4520.

Fehniger, T. A., Yu, H., Cooper, M. A., Suzuki, K., Shah, M. H., and Caligiuri, M. A. (2000b).Cutting edge: IL-15 costimulates the generalized Shwartzman reaction and innate immuneIFN-g production in vivo. J. Immunol. 164, 1643–1647.

Fehniger, T. A., and Caligiuri, M. A. (2001). Interleukin 15: Biology and relevance to humandisease. Blood 97, 14–32.

role of il-2 and il-15 in human nk cells 235

Ferlazzo, G., Thomas, D., Lin, S. L., Goodman, K., Morandi, B., Muller, W. A., Moretta, A., andMunz, C. (2004). The abundant NK cells in human secondary lymphoid tissues require activa-tion to express killer cell Ig-like receptors and become cytolytic. J. Immunol. 172, 1455–1462.

Frey, M., Packianathan, N. B., Fehniger, T. A., Ross, M. E., Wang, W. C., Stewart, C. C., Caligiuri,M. A., and Evans, S. S. (1998). Differential expression and function of L-selectin on CD56bright

and CD56dim natural killer cell subsets. J. Immunol. 161, 400–408.Gilmour, K. C., Fujii, H., Cranston, T., Davies, E. G., Kinnon, C., and Gaspar, H. B. (2001).

Defective expression of the interleukin-2/interleukin-15 receptor b subunit leads to a naturalkiller cell-deficient form of severe combined immunodeficiency. Blood 98, 877–879.

Giri, J. G., Kumaki, S., Ahdieh, M., Friend, D. J., Loomis, A., Shanebeck, K., DuBose, R., Cosman,D., Park, L. S., and Anderson, D. M. (1995). Identification and cloning of a novel IL-15 bindingprotein that is structurally related to the a chain of the IL-2 receptor. EMBO J. 14, 3654–3663.

Giron-Michel, J., Fogli, M., Gaggero, A., Ferrini, S., Caignard, A., Brouty-Boye, D., Baouz, S., LeBousse-Kerdiles, M. C., Peault, B., van Dijk, M., Bulfone-Paus, S., Durali, D., Chouaib, S., andAzzarone, B. (2003). Detection of a functional hybrid receptor gc/GM-CSFRbeta in humanhematopoietic CD34þ cells. J. Exp. Med. 197, 763–775.

Grabstein, K. H., Eisenman, J., Shanebeck, K., Rauch, C., Srinivasan, S., Fung, V., Beers, C.,Richardson, J., Schoenborn, M. A., Ahdieh, M., et al. (1994). Cloning of a T cell growth factorthat interacts with the b chain of the interleukin-2 receptor. Science 264, 965–968.

Granucci, F., Feau, S., Angeli, V., Trottein, F., and Ricciardi-Castagnoli, P. (2003). Early IL-2production by mouse dendritic cells is the result of microbial-induced priming. J. Immunol. 170,5075–5081.

Granucci, F., Vizzardelli, C., Pavelka, N., Feau, S., Persico, M., Virzi, E., Rescigno, M., Moro, G.,and Ricciardi-Castagnoli, P. (2001). Inducible IL-2 production by dendritic cells revealed byglobal gene expression analysis. Nat. Immunol. 2, 882–888.

Gu, H., Maeda, H., Moon, J. J., Lord, J. D., Yoakim, M., Nelson, B. H., and Neel, B. G. (2000).New role for Shc in activation of the phosphatidylinositol 3-kinase/Akt pathway. Mol. Cell. Biol.20, 7109–7120.

Imada, K., Bloom, E. T., Nakajima, H., Horvath-Arcidiacono, J. A., Udy, G. B., Davey, H. W., andLeonard, W. J. (1998). Stat5b is essential for natural killer cell-mediated proliferation andcytolytic activity. J. Exp. Med. 188, 2067–2074.

Ishida, Y., Nishi, M., Taguchi, O., Inaba, K., Hattori, M., Minato, N., Kawaichi, M., and Honjo, T.(1989). Expansion of natural killer cells but not T cells in human interleukin 2/interleukin 2receptor (Tac) transgenic mice. J. Exp. Med. 170, 1103–1115.

Jacobson, E. L., Pilaro, F., and Smith, K. A. (1996). Rational interleukin 2 therapy for HIV positiveindividuals: Daily low doses enhance immune function without toxicity. Proc. Natl. Acad. Sci.USA 93, 10405–10410.

Jamieson, A. M., Isnard, P., Dorfman, J. R., Coles, M. C., and Raulet, D. H. (2004). Turnover andproliferation of NK cells in steady state and lymphopenic conditions. J. Immunol. 172, 864–870.

Jiang, K., Zhong, B., Ritchey, C., Gilvary, D. L., Hong-Geller, E., Wei, S., and Djeu, J. Y. (2003).Regulation of Akt-dependent cell survival by Syk and Rac. Blood 101, 236–244.

Jinushi, M., Takehara, T., Tatsumi, T., Kanto, T., Groh, V., Spies, T., Suzuki, T., Miyagi, T., andHayashi, N. (2003). Autocrine/paracrine IL-15 that is required for type I IFN-mediated den-dritic cell expression of MHC class I-related chain A and B is impaired in hepatitis C virusinfection. J. Immunol. 171, 5423–5429.

Kawamura, T., Koka, R., Ma, A., and Kumar, V. (2003). Differential roles for IL-15R a-chain in NKcell development and Ly-49 induction. J. Immunol. 171, 5085–5090.

Kennedy, M. K., Glaccum, M., Brown, S. N., Butz, E. A., Viney, J. L., Embers, M., Matsuki, N.,Charrier, K., Sedger, L., Willis, C. R., Brasel, K., Morrissey, P. J., Stocking, K., Schuh, J. C.,

236 brian becknell and michael a. caligiuri

Joyce, S., and Peschon, J. J. (2000). Reversible defects in natural killer and memory CD8 T celllineages in interleukin 15-deficient mice. J. Exp. Med. 191, 771–780.

Khatri, V. P., Fehniger, T. A., Baiocchi, R. A., Yu, F., Shah, M. H., Schiller, D. S., Gould, M.,Gazzinelli, R. T., Bernstein, Z. P., and Caligiuri, M. A. (1998). Ultra low dose interleukin-2therapy promotes a type 1 cytokine profile in vivo in patients with AIDS and AIDS-associatedmalignancies. J. Clin. Invest. 101, 1373–1378.

Kim, S., Iizuka, K., Aguila, H. L., Weissman, I. L., and Yokoyama, W. M. (2000). In vivo naturalkiller cell activities revealed by natural killer cell-deficient mice. Proc. Natl. Acad. Sci. USA 97,2731–2736.

Koka, R., Burkett, P. R., Chien, M., Chai, S., Chan, F., Lodolce, J. P., Boone, D. L., and Ma, A.(2003). Interleukin (IL)-15Ra-deficient natural killer cells survive in normal but not IL-15Ra-deficient mice. J. Exp. Med. 197, 977–984.

Kondo, M., Akashi, K., Domen, J., Sugamura, K., and Weissman, I. L. (1997). Bcl-2 rescuesT lymphopoiesis, but not B or NK cell development, in common g chain-deficient mice.Immunity 7, 155–162.

Kundig, T. M., Schorle, H., Bachmann, M. F., Hengartner, H., Zinkernagel, R. M., andHorak, I. (1993). Immune responses in interleukin-2-deficient mice. Science 262,1059–1061.

Lanier, L. L., Le, A. M., Civin, C. I., Loken, M. R., and Phillips, J. H. (1986). The relationship ofCD16 (Leu-11) and Leu-19 (NKH-1) antigen expression on human peripheral blood NK cellsand cytotoxic T lymphocytes. J. Immunol. 136, 4480–4486.

Lanier, L. L., Testi, R., Bindl, J., and Phillips, J. H. (1989). Identity of Leu-19 (CD56) leukocytedifferentiation antigen and neural cell adhesion molecule. J. Exp. Med. 169, 2233–2238.

Leonard, W. J. (2001). Role of Jak kinases and STATs in cytokine signal transduction. Int. J. Hematol.73, 271–277.

Lieberman, L. A., and Hunter, C. A. (2002). Regulatory pathways involved in the infection-inducedproduction of IFN-g by NK cells. Microbes Infect. 4, 1531–1538.

Lodolce, J. P., Boone, D. L., Chai, S., Swain, R. E., Dassopoulos, T., Trettin, S., and Ma, A. (1998).IL-15 receptor maintains lymphoid homeostasis by supporting lymphocyte homing and prolifer-ation. Immunity 9, 669–676.

Lotzova, E., Savary, C. A., and Champlin, R. E. (1993). Genesis of human oncolytic naturalkiller cells from primitive CD34þ CD33� bone marrow progenitors. J. Immunol. 150,5263–5269.

Matos, M. E., Schnier, G. S., Beecher, M. S., Ashman, L. K., William, D. E., and Caligiuri, M. A.(1993). Expression of a functional c-Kit receptor on a subset of natural killer cells. J. Exp. Med.178, 1079–1084.

Mattei, F., Schiavoni, G., Belardelli, F., and Tough, D. F. (2001). IL-15 is expressed by dendriticcells in response to type I IFN, double-stranded RNA, or lipopolysaccharide and promotesdendritic cell activation. J. Immunol. 167, 1179–1187.

McDonald, P. P., Russo, M. P., Ferrini, S., and Cassatella, M. A. (1998). Interleukin-15 (IL-15)induces NF-kB activation and IL-8 production in human neutrophils. Blood 92, 4828–4835.

Miller, J. S., Verfaillie, C., and McGlave, P. (1992). The generation of human natural killer cellsfrom CD34þ/DR� primitive progenitors in long-term bone marrow culture. Blood 80, 2182–2187.

Minagawa, M., Watanabe, H., Miyaji, C., Tomiyama, K., Shimura, H., Ito, A., Ito, M., Domen, J.,Weissman, I. L., and Kawai, K. (2002). Enforced expression of Bcl-2 restores the number of NKcells, but does not rescue the impaired development of NKT cells or intraepithelial lymphocytes,in IL-2/IL-15 receptor b-chain-deficient mice. J. Immunol. 169, 4153–4160.

role of il-2 and il-15 in human nk cells 237

Minami, Y., Nakagawa, Y., Kawahara, A., Miyazaki, T., Sada, K., Yamamura, H., and Taniguchi, T.(1995). Protein tyrosine kinase Syk is associated with and activated by the IL-2 receptor: Possiblelink with the c-Myc induction pathway. Immunity 2, 89–100.

Miyazaki, T., Kawahara, A., Fujii, H., Nakagawa, Y., Minami, Y., Liu, Z. J., Oishi, I., Silvennoinen, O.,Witthuhn, B. A., Ihle, J. N., et al. (1994). Functional activation of Jak1 and Jak3 by selectiveassociation with IL-2 receptor subunits. Science 266, 1045–1047.

Miyazaki, T., Liu, Z. J., Kawahara, A., Minami, Y., Yamada, K., Tsujimoto, Y., Barsoumian,E. L., Permutter, R. M., and Taniguchi, T. (1995). Three distinct IL-2 signaling pathwaysmediated by Bcl-2, c-Myc, and Lck cooperate in hematopoietic cell proliferation. Cell 81,223–231.

Mrozek, E., Anderson, P., and Caligiuri, M. A. (1996). Role of interleukin-15 in the development ofhuman CD56þ natural killer cells from CD34þ hematopoietic progenitor cells. Blood 87, 2632–2640.

Musso, T., Calosso, L., Zucca, M., Millesimo, M., Ravarino, D., Giovarelli, M., Malavasi, F., Ponzi,A. N., Paus, R., and Bulfone-Paus, S. (1999). Human monocytes constitutively express mem-brane-bound, biologically active, and interferon-g-upregulated interleukin-15. Blood 93,3531–3539.

Nagler, A., Lanier, L. L., Cwirla, S., and Phillips, J. H. (1989). Comparative studies of humanFcRIII-positive and negative natural killer cells. J. Immunol. 143, 3183–3191.

Nagler, A., Lanier, L. L., and Phillips, J. H. (1990). Constitutive expression of high affinityinterleukin 2 receptors on human CD16-natural killer cells in vivo. J. Exp. Med. 171,1527–1533.

Noguchi, M., Yi, H., Rosenblatt, H. M., Filipovich, A. H., Adelstein, S., Modi, W. S., McBride,O. W., and Leonard, W. J. (1993). Interleukin-2 receptor g chain mutation results in X-linkedsevere combined immunodeficiency in humans. Cell 73, 147–157.

Nosaka, T., van Deursen, J. M., Tripp, R. A., Thierfelder, W. E., Witthuhn, B. A., McMickle, A. P.,Doherty, P. C., Grosveld, G. C., and Ihle, J. N. (1995). Defective lymphoid development in micelacking Jak3. Science 270, 800–802.

Orange, J. S., Brodeur, S. R., Jain, A., Bonilla, F. A., Schneider, L. C., Kretschmer, R., Nurko, S.,Rasmussen, W. L., Kohler, J. R., Gellis, S. E., Ferguson, B. M., Strominger, J. L., Zonana, J.,Ramesh, N., Ballas, Z. K., and Geha, R. S. (2002). Deficient natural killer cell cytotoxicity inpatients with IKK-g/NEMO mutations. J. Clin. Invest. 109, 1501–1509.

Park, S. Y., Saijo, K., Takahashi, T., Osawa, M., Arase, H., Hirayama, N., Miyake, K., Nakauchi, H.,Shirasawa, T., and Saito, T. (1995). Developmental defects of lymphoid cells in Jak3 kinase-deficient mice. Immunity 3, 771–782.

Parrish-Novak, J., Dillon, S. R., Nelson, A., Hammond, A., Sprecher, C., Gross, J. A., Johnston, J.,Madden, K., Xu, W., West, J., Schrader, S., Burkhead, S., Heipel, M., Brandt, C., Kuijper, J. L.,Kramer, J., Conklin, D., Presnell, S. R., Berry, J., Shiota, F., Bort, S., Hambly, K., Mudri, S.,Clegg, C., Moore, M., Grant, F. J., Lofton-Day, C., Gilbert, T., Rayond, F., Ching, A., Yao, L.,Smith, D., Webster, P., Whitmore, T., Maurer, M., Kaushansky, K., Holly, R. D., and Foster, D.(2000). Interleukin 21 and its receptor are involved in NK cell expansion and regulation oflymphocyte function. Nature 408, 57–63.

Phillips, J. H., Takeshita, T., Sugamura, K., and Lanier, L. L. (1989). Activation of natural killercells via the p75 interleukin 2 receptor. J. Exp. Med. 170, 291–296.

Piguet, P. F., Grau, G., Irle, C., and Vassalli, P. (1986). Administration of recombinant interleukin2 to mice enhances production of hemopoietic and natural killer cells. Eur. J. Immunol. 16,1257–1261.

238 brian becknell and michael a. caligiuri

Pollack, S. B., Tsuji, J., and Rosse, C. (1992). Production and differentiation of NK lineage cells inlong-term bone marrow cultures in the absence of exogenous growth factors. Cell. Immunol.139, 352–362.

Qin, S., Inazu, T., Yang, C., Sada, K., Taniguchi, T., and Yamamura, H. (1994). Interleukin2 mediates p72syk activation in peripheral blood lymphocytes. FEBS Lett. 345, 233–236.

Ranson, T., Vosshenrich, C. A., Corcuff, E., Richard, O., Muller, W., and Di Santo, J. P. (2003).IL-15 is an essential mediator of peripheral NK-cell homeostasis. Blood 101, 4887–4893.

Rosenberg, S. A. (2000). Interleukin-2 and the development of immunotherapy for the treatmentof patients with cancer. Cancer J. Sci. Am. 6(Suppl. 1), S2–S7.

Rosenberg, S. A., Lotze, M. T., Muul, L. M., Chang, A. E., Avis, F. P., Leitman, S., Linehan, W. M.,Robertson, C. N., Lee, R. E., Rubin, J. T., et al. (1987). A progress report on the treatment of157 patients with advanced cancer using lymphokine-activated killer cells and interleukin-2 orhigh-dose interleukin-2 alone. N. Engl. J. Med. 316, 889–897.

Rosmaraki, E. E., Douagi, I., Roth, C., Colucci, F., Cumano, A., and Di Santo, J. P. (2001).Identification of committed NK cell progenitors in adult murine bone marrow. Eur.J. Immunol. 31, 1900–1909.

Ross, M. E., and Caligiuri, M. A. (1997). Cytokine-induced apoptosis of human natural killer cellsidentifies a novel mechanism to regulate the innate immune response. Blood 89, 910–918.

Ruggeri, L., Capanni, M., Urbani, E., Perruccio, K., Shlomchik, W. D., Tosti, A., Posati, S., Rogaia,D., Frassoni, F., Aversa, F., Martelli, M. F., and Velardi, A. (2002). Effectiveness of donor naturalkiller cell alloreactivity in mismatched hematopoietic transplants. Science 295, 2097–2100.

Santini, S. M., Lapenta, C., Logozzi, M., Parlato, S., Spada, M., Di Pucchio, T., and Belardelli, F.(2000). Type I interferon as a powerful adjuvant for monocyte-derived dendritic cell develop-ment and activity in vitro and in Hu-PBL-SCID mice. J. Exp. Med. 191, 1777–1788.

Schluns, K. S., Klonowski, K. D., and Lefrancois, L. (2004). Transregulation of memory CD8 T-cellproliferation by IL-15Raþ bone marrow-derived cells. Blood 103, 988–994.

Schorle, H., Holtschke, T., Hunig, T., Schimpl, A., and Horak, I. (1991). Development andfunction of T cells in mice rendered interleukin-2 deficient by gene targeting. Nature 352,621–624.

Shankaran, V., Ikeda, H., Bruce, A. T., White, J. M., Swanson, P. E., Old, L. J., and Schreiber, R. D.(2001). IFN-g and lymphocytes prevent primary tumour development and shape tumourimmunogenicity. Nature 410, 1107–1111.

Sharfe, N., Dadi, H. K., Shahar, M., and Roifman, C. M. (1997). Human immune disorder arisingfrom mutation of the a chain of the interleukin-2 receptor. Proc. Natl. Acad. Sci. USA 94, 3168–3171.

Shibuya, A., Nagayoshi, K., Nakamura, K., and Nakauchi, H. (1995). Lymphokine requirement forthe generation of natural killer cells from CD34þ hematopoietic progenitor cells. Blood 85,3538–3546.

Sivori, S., Cantoni, C., Parolini, S., Marcenaro, E., Conte, R., Moretta, L., and Moretta, A. (2003).IL-21 induces both rapid maturation of human CD34þ cell precursors towards NK cells andacquisition of surface killer Ig-like receptors. Eur. J. Immunol. 33, 3439–3447.

Smith, K. A. (1988). Interleukin-2: Inception, impact, and implications. Science 240, 1169–1176.Suzuki, H., Duncan, G. S., Takimoto, H., and Mak, T. W. (1997). Abnormal development of

intestinal intraepithelial lymphocytes and peripheral natural killer cells in mice lacking the IL-2 receptor b chain. J. Exp. Med. 185, 499–505.

Tanaka, T., Kitamura, F., Nagasaka, Y., Kuida, K., Suwa, H., and Miyasaka, M. (1993). Selectivelong-term elimination of natural killer cells in vivo by an anti-interleukin 2 receptor b chainmonoclonal antibody in mice. J. Exp. Med. 178, 1103–1107.

role of il-2 and il-15 in human nk cells 239

Teglund, S., McKay, C., Schuetz, E., van Deursen, J. M., Stravopodis, D., Wang, D., Brown, M.,Bodner, S., Grosveld, G., and Ihle, J. N. (1998). Stat5a and Stat5b proteins have essential andnonessential, or redundant, roles in cytokine responses. Cell 93, 841–850.

Waldmann, T. A., Dubois, S., and Tagaya, Y. (2001). Contrasting roles of IL-2 and IL-15 in the lifeand death of lymphocytes: Implications for immunotherapy. Immunity 14, 105–110.

Willerford, D. M., Chen, J., Ferry, J. A., Davidson, L., Ma, A., and Alt, F. W. (1995). Interleukin-2 receptor a chain regulates the size and content of the peripheral lymphoid compartment.Immunity 3, 521–530.

Williams, N. S., Moore, T. A., Schatzle, J. D., Puzanov, I. J., Sivakumar, P. V., Zlotnik, A., Bennett,M., and Kumar, V. (1997). Generation of lytic natural killer 1.1þ, Ly-49� cells from multi-potential murine bone marrow progenitors in a stroma-free culture: Definition of cytokinerequirements and developmental intermediates. J. Exp. Med. 186, 1609–1614.

Yu, H., Fehniger, T. A., Fuchshuber, P., Thiel, K. S., Vivier, E., Carson, W. E., and Caligiuri, M. A.(1998). Flt3 ligand promotes the generation of a distinct CD34þ human natural killer cellprogenitor that responds to interleukin-15. Blood 92, 3647–3657.

Zhang, J., Scordi, I., Smyth, M. J., and Lichtenheld, M. G. (1999). Interleukin 2 receptor signalingregulates the perforin gene through signal transducer and activator of transcription (Stat)5activation of two enhancers. J. Exp. Med. 190, 1297–1308.

Zhou, J., Zhang, J., Lichtenheld, M. G., and Meadows, G. G. (2002). A role for NF-kB activation inperforin expression of NK cells upon IL-2 receptor signaling. J. Immunol. 169, 1319–1325.

Zhu, X., Suen, K. L., Barbacid, M., Bolen, J. B., and Fargnoli, J. (1994). Interleukin-2-inducedtyrosine phosphorylation of Shc proteins correlates with factor-dependent T cell proliferation. J.Biol. Chem. 269, 5518–5522.

Regulation of Antigen Presentation and Cross-Presentation inthe Dendritic Cell Network: Facts, Hypothesis, and

Immunological Implications

Nicholas S. Wilson and Jose A. Villadangos

Immunology Division and The Cooperative Research Center for Vaccine Technology,The Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria 3050, Australia

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2411. Introduction: Dendritic Cells at the Crossroads of Immunity and Tolerance .. . . . . . . . . . . . . . . 2412. Commonalities and Diversity in the DC Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2433. Antigen Presentation Pathways. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2514. Mechanisms and Regulation of Antigen Capture in DCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2525. Antigen Presentation by MHC II Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2576. Control of MHC II Antigen Presentation in DCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2637. DCs and Cross-Presentation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2808. Conclusions and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289

Abstract

Dendritic cells (DCs) are central to the maintenance of immunological toleranceand the initiation and control of immunity. The antigen-presenting properties ofDCs enable them to present a sample of self and foreign proteins, contained withinan organism at any given time, to the T-cell repertoire. DCs achieve this commu-nication with Tcells by displaying antigenic peptides bound to MHC I and MHC IImolecules. Here we review the studies carried out over the past 15 years tocharacterize these antigen presentation mechanisms, emphasizing their signifi-cance in relation to DC function in vivo. The life cycles of different DC populationsfound in vivo are described. Furthermore, we provide a critical assessment ofthe studies that examine the mechanisms controlling DC MHC class II antigenpresentation, which have often reached contradictory conclusions. Finally, wereview findings pertaining to the biological mechanisms that enable DCs to presentexogenous antigens on their MHC class I molecules, a process known as cross-presentation. Throughout, we highlight what we consider to be major knowledgegaps in the field and speculate on possible directions for future research.

1. Introduction: Dendritic Cells at the Crossroads of Immunity and Tolerance

T cells are generated from bone marrow precursors that migrate into thethymus, where they mature into naive T cells. The maturing thymocytesare interrogated and selected based on the quality of the interaction of their

241advances in immunology, vol. 86 � 2005 Elsevier Inc.

0065-2776/05 $35.00 All rights reserved.

242 nicholas s. wilson and jose a. villadangos

T-cell receptors (TCRs) with major histocompatibility complex (MHC)–peptide complexes displayed on the surface of thymic antigen-presentingcells. The thymocytes are selected in a two-pronged process: ‘‘positive’’ selec-tion, which chooses thymocytes that express TCRs capable of some interactionwith the host’s MHC molecule/self-peptide complexes; and ‘‘negative’’ selec-tion, which deletes thymocytes that recognize self-MHC–peptide combina-tions too strongly (autoreactive). Since negative selection is responsible for theelimination of most anti-self T cells, the outcome of this process is referred toas the generation of central tolerance (Starr et al., 2003).

Naive T cells that emerge from the thymus have the potential to interactstrongly with (‘‘recognize’’) foreign peptides bound to host MHC moleculesand become effector T cells in a process referred to as priming. However, it isimportant to note that the repertoire of naive T cells that emerges from thethymus is not ignorant of the peripheral self-MHC–peptide complexes; everyT-cell reacts with self-complexes at a basal strength, a feature that may berequired to maintain the peripheral T-cell compartment (Sprent and Surh,2003). Indeed, the process of central tolerance allows some autoreactive T cellsto escape to the periphery (Bouneaud et al., 2000). These autoreactiveT cells must be eliminated or held in check (tolerized) to avoid autoimmunityin a process that is referred to as peripheral tolerance.

Obviously, a healthy immune system should be efficient at priming anti-foreign T cells while maintaining the self-reactive T cells tolerized. The prob-lem is that the structural basis of the interaction between anti-foreign TCRsand anti-self TCRs with their respective MHC–peptide combinations is thesame. Thus, autoreactive T cells cannot themselves discern whether they arerecognizing self or foreign peptides via their TCRs. It is the context in whichthis interaction takes place that determines whether a given T-cell will beprimed or tolerized. Research conducted on dendritic cells (DCs) has revealedthat this rare and heterogeneous population of leukocytes is in charge ofproviding both the immunogenic and tolerogenic contexts for naive T cells(Banchereau and Steinman, 1998; Steinman, 1991; Steinman et al., 2003).But if DCs have the dual (and paradoxical) role of inducing priming andtolerance, how do they decide between these two opposing outcomes? Ithas been suggested that this decision depends on the developmental stage ofthe DCs (‘‘maturity’’) (Finkelman et al., 1996; Hawiger et al., 2001; Steinmanand Nussenzweig, 2002). Although we are still far from understandingwhat makes a DC tolerogenic or immunogenic, characterizing the antigen-presenting properties of DCs at the different stages of their development hasthus emerged as a primary goal to understand their roles in immunosurveil-lance and tolerance induction. Those properties are the main subject ofthis review.

control of antigen presentation in dendritic cells 243

2. Commonalities and Diversity in the DC Network

2.1. What Is DC Maturation?

Articles on DCs usually begin with a description of a life cycle modeled on thefindings of studies performed first on Langerhans cells (LCs, a type of DCfound in the epidermis) (Romani et al., 1989; Schuler and Steinman, 1985;Steinman, 1991) and later reproduced using DC models grown in vitro (Cellaet al., 1997b; Pierre et al., 1997; Winzler et al., 1997). According to this ‘‘LCparadigm,’’ DCs are present in peripheral tissues in a so-called immature statededicated to sampling their environment (Fig. 1). Immature DCs are highlyendocytic, but express low levels of MHC class II (MHC II) molecules on theirsurface, although these molecules appear abundant in endocytic compart-ments. Another feature of immature DCs is that they are inefficient at activat-ing naive T cells. In the presence of inflammatory compounds expressed byforeign pathogens [lipopolysaccharide (LPS), double-stranded RNA, DNArich in CpG motifs, etc.] or released by damaged tissues (e.g., tumor necrosisfactor a, TNF-a), DCs become ‘‘activated’’ and migrate via the afferentlymphatics to the local draining lymph node (LN) (Larsen et al., 1990;Stoitzner et al., 2003). Simultaneously, DCs acquire a ‘‘mature’’ phenotypecharacterized by a reduced capacity to capture antigens, high expression ofMHC II at the cell surface, and high expression of T-cell costimulatorymolecules such as CD40, CD80, and CD86, which enable mature DCs tostimulate naive T cells (Sharpe and Freeman, 2002). Mature DCs showprolonged presentation of antigens they captured just before or at the timeof receiving their activation signal, but they no longer process and presentnewly encountered antigens. The ability of mature DCs to sustain the presen-tation of the antigens they captured in their immature state has been referredto as antigenic memory.

2.2. DC Subtypes

According to the LC paradigm, DC maturation is linked to tissue localization,with immature cells in the peripheral tissues and mature DCs in the lymphoidorgans. However, the LC represents just one of the multiple types of DCcontained in the body, and it has become apparent that the LC paradigm ofDC life cycle and function does not apply to all DC types (Manickasinghamand Reis e Sousa, 2001; Wilson and Villadangos, 2004). For researchersinterested in antigen presentation but not familiar with the DC field, thesubdivision of DCs into subpopulations can sometimes appear confusing,arbitrary, and irrelevant. However, we consider that a review of the mechan-isms employed by DCs to control antigen presentation would not be complete

Figure 1 Dendritic cell (DC) populations of the spleen and lymph nodes. (A) The spleen (left)receives only blood-derived DCs (red and blue), whereas the lymph nodes (right) receive bothblood-derived and tissue-derived (green) DCs. (B) FACS analysis of the spleen and lymph nodeDC populations (see Table 1). The plots show the expression of CD4 vs CD8 (spleen, left), and of

244 nicholas s. wilson and jose a. villadangos

control of antigen presentation in dendritic cells 245

without a description of the ‘‘life styles’’ for different DC types and how theyagree with, or deviate from, the LC paradigm. Moreover, because our knowl-edge of the antigen-presenting properties of DCs comes from studies usingdifferent types of DC grown in vitro or purified from tissues, it is important tounderstand how the conclusions of the different DC models correlate with thefunctions of DCs in vivo.

Because of the heterogeneity of the DC system, and the changes undergoneby DCs during their development, definition of the DC lineage is usually basedon a set of common phenotypic and functional features rather than a singlesurface marker or capability. Common features of DCs include the expressionof CD11c, the ability to endocytose a large variety of antigens and present themvia MHC II molecules, and a high capacity to stimulate naive T cells. Each DCpopulation can then be further defined based on its expression of a combinationof surface markers. In this section we summarize the most common markersused to discriminate the major DC subtypes known to date. We stress that thisis not by any means an exhaustive description of phenotypic and functionaldifferences among DC subtypes, which have been extensively reviewed else-where (Ardavin, 2003; Hart, 1997; Shortman and Liu, 2002; Wilson andVilladangos, 2004). DC heterogeneity has been characterized best in themouse system, so we will start with murine DC subsets, followed by human DCs.

2.2.1. Tissue-Derived DCs

A major subdivision of DC types can be made according to the paths used toaccess the lymphoid organs: the lymph or the blood (reviewed in Itano andJenkins, 2003; Wilson and Villadangos, 2004). The term ‘‘tissue-derived DCs’’refers to those present in the interstitial spaces of all tissues and in epithelialsurfaces of the gut, the airways, and the skin. These DCs migrate via the

CD205 vs CD8 (lymph node, right) in CD11chigh DCs. The blood-derived DCs comprise threepopulations: CD8+CD4�CD205+ (CD8+ DCs, red), CD4+CD8�CD205� (CD4+ DCs, blue), andCD4�CD8�CD205� DCs (CD4�CD8�, also blue). Note that the CD4+ and CD4�CD8� DCpopulations appear as a single CD8�CD205� group in the lymph node (blue). The tissue-derivedDCs can be distinguished as a CD205+CD8low population (green). These tissue-derived DCsconstitutively migrate from the periphery to the draining lymph nodes (A, top right), where theyhave a mature phenotype, as shown in the immunofluorescence confocal microscopy image shownin (C). In contrast, the blood-derived DCs contained in the spleen and lymph nodes in the steadystate are immature, as illustrated by the accumulation of MHC II molecules in endosomal (Lamp+)compartments (C, top left). The blood-derived DCs can be induced to mature in the lymphoidorgans in response to inflammatory compounds such as LPS or CpG and then acquire a maturephenotype (C, bottom left).

246 nicholas s. wilson and jose a. villadangos

afferent lymphatics into their local LN, where they eventually die (Fig. 1)(Brand et al., 1993; Bujdoso et al., 1989; Geissmann et al., 2002; Howard et al.,1996; Larsen et al., 1990; Lukas et al., 1996). Tissue-derived DCs can bedistinguished from other DC types by their low expression of CD8 and theintermediate to high expression of the C-type lectin CD205 (also known asDEC-205) (Table 1) (Anjuere et al., 1999; Henri et al., 2001; Kamath et al.,2002; Ruedl et al., 2000; Salomon et al., 1998; Stoitzner et al., 2003; Vermaelenet al., 2001; Wilson et al., 2003). LCs belong to this DC group, but of coursethe only LNs where they can be found are the subcutaneous. Thus, thesubcutaneous LNs receives two types of DC via lymph: the ‘‘interstitial DCs’’

Table 1. Mouse Dendritic Cell Populations

Blood derived

Conventional Tissue derived In Blood

pDCs CD4+ CD8+ CD4�CD8� LCs iDCs DCs pDCs

A. LocationSpleen U U U U

Subcutaneous LNs U U U U U U

Visceral LNs U U U U U

Thymus U ? U ?

B. MarkersCD11c þþ þþþ þþþ þþþ þþþ þþþ þþþ þþCD45RA þþþ þþþCD4 var. þþþCD8 var. þþþCD205 þþ þþþ þLangerin þ þþþ þCD11b þþþ þþþ þþþ þþþ þþþ

C. MaturityMaturitya im im im im mat mat im imCostimulatoryb þ þ þ þ þþ þþ þ þAntigen processingc þ þþþ þþþ þþþ þ/– þ/– þþþ þMHC II þ þþ þþ þþ þþþ þþþ þ þ

aim, immature; mat, mature.bT-cell–costimulatory molecules (CD80, CD86, CD40).cThe capacity to endocytose and present new antigens.Abbreviations: pDC, plasmacytoid dendritic cell; LC, Langer hans cell; iDC, immature DC; LN, lymph node.Key: ?, CD4þDCs can be identified, but this could be due to ‘‘pick-up’’ from thymocytes (Vremec et al., 2000);þ, low; þþ, intermediate; þþþ, high; var., heterogeneous (pDCs can be CD4�CD8þ, CD4þCD8�,CD4�CD8þ, or CD4þCD8þ; O’Keeffe et al., 2002a).

control of antigen presentation in dendritic cells 247

of the dermis, which are the skin equivalent of the tissue-derived DCs presentin all other tissues, and LCs from the epidermis (Table 1). These two DC typescan be distinguished from each other by the higher expression of CD205, theintracellular expression of langerin, and the presence of Birbeck granules inthe LCs (Henri et al., 2001; Kamath et al., 2002; Stoitzner et al., 2003;Valladeau et al., 2000; Wilson et al., 2003). Virtually all the tissue-derivedDCs contained in the LNs have a mature phenotype: they express highlevels of MHC II and T-cell costimulatory molecules and can activate naiveT cells, but they are inefficient at processing and presenting newly encoun-tered antigens (Anjuere et al., 1999; Henri et al., 2001; Kamath et al., 2002;Ruedl et al., 2000; Salomon et al., 1998; Stoitzner et al., 2003; Vermaelen et al.,2001; Wilson et al., 2003, 2004). Several studies have demonstrated that tissue-derived DCs contained in the subcutaneous LNs, mesenteric LNs, andmediastinal LNs come from immature DCs that migrated from the skin, thegut, and the airways, respectively (Henri et al., 2001; Huang et al., 2000; Itanoet al., 2003; Macatonia et al., 1987; Vermaelen et al., 2001). So in general thisDC group behaves according to the LC paradigm, but there are two exceptionsto this general rule. First, it has been reported that tissue-derived DCs cantraffic to the LNs while retaining an immature phenotype (Geissmann et al.,2002), although this may only happen in pathological conditions such aschronic skin inflammation. Second, some DCs activated in the peripheraltissue may remain at the site of activation, where they might play an importantrole in driving inflammatory reactions (Huh et al., 2003; Luft et al., 2002;Vermaelen and Pauwels, 2003). It is also important to note that the migrationand maturation of the tissue-derived DCs may not always be triggered byactivatory signals but be part of their normal life cycle even in the absence ofinfections (in the steady state).

2.2.2. Blood–Derived DCs

Another major group of DCs can be found in the LNs and also in the spleen.The spleen can only be accessed via the blood, so probably this is the pathwayused by this DC group to access both the spleen and the LNs (Fig. 1)(reviewed in Itano and Jenkins, 2003; Wilson and Villadangos, 2004). One ofthe DC types belonging to this group of ‘‘blood-derived DCs’’ are the so-calledinterferon-producing cells or plasmacytoid DCs (pDCs), which are character-ized by expression of CD45RA and lower levels of CD11c than other DC types(Table 1; Asselin-Paturel et al., 2001; Bjorck, 2001; Martin et al., 2002; Nakanoet al., 2001; O’Keeffe et al., 2002a). It is unclear whether pDCs play a majorrole in presenting antigens for the initiation of immune responses becausetheir antigen presentation and naive T-cell–stimulatory capacities are much

248 nicholas s. wilson and jose a. villadangos

poorer than those of the other DC types (Krug et al., 2003). Rather, the mainfunction of this DC population appears to be the secretion of interferonsagainst viruses as part of the innate immune response (Colonna et al., 2002).Because this group of DCs is quite unusual and may not be involved inpriming, in this review we use the term ‘‘blood-derived’’ only to refer to the‘‘conventional’’ CD45RA� nonplasmacytoid DCs. These conventional DCscan be subdivided into three populations: one expresses high levels of CD8and CD205 (‘‘CD8þ DCs’’ hereafter), the other expresses CD4 but notCD205 (‘‘CD4þ DCs’’), and the third expresses neither of these three markers(‘‘CD4�CD8� DCs’’) (Table 1) (Anjuere et al., 1999; Henri et al., 2001;Vremec et al., 2000; Wilson et al., 2003). These populations most likely emergefrom three independent developmental streams within the DC lineage (Kamathet al., 2002; Naik et al., 2003; Shortman and Liu, 2002), although some reportssuggest that the CD8þ DCs are derived from one of the CD8� DC populations(Ardavin, 2003). In the steady state, the CD8þ and CD8� DCs are located indifferent areas within the lymphoid organs: the CD8þ DCs are present in theT-cell areas, and the CD8� DCs are found in the marginal zones (De Smedtet al., 1996; Pulendran et al., 1997). The functional significance of this segrega-tion is unclear.

In contrast to the tissue-derived DCs, virtually all the blood-derived DCscontained in the LN and the spleen in the steady state have an immaturephenotype (Fig. 1; Wilson et al., 2003). This statement may appear paradoxical,because the LC paradigm predicts that the DCs reaching the lymphoid organsshould have a mature phenotype. Indeed, the relatively high expression ofMHC II in blood-derived DCs, and their capacity to activate naive T cells inmixed lymphocyte reactions (MLRs) in vitro, led to their consideration asmature DCs (Ruedl et al., 2000; Salomon et al., 1998; Vremec et al., 2000).However, as we discuss later, the level of surface expression of MHC II is amisleading parameter when assessing DC maturity. More importantly, theexperiments that measured the capacity of blood-derived DCs to activatenaive T cells in MLRs had an important caveat: even though freshly isolatedDCs are immature, they become spontaneously activated in culture andquickly acquire a mature phenotype, thus gaining the capacity to activatenaive T cells (De Smedt et al., 1996; El-Sukkari et al., 2003; Inaba et al.,1992; Wilson et al., 2003, 2004). This process of ‘‘spontaneous maturation’’ alsohappens to LCs extracted from the skin and cultured in vitro; indeed, theoriginal definition of DC maturation was based on the phenotypic changesundergone by skin LCs during culture (Schuler and Steinman, 1985). In fact,blood-derived DCs freshly isolated from the spleen or the LN fulfill thedefinition of immaturity: most of their MHC II molecules are contained inendosomal compartments rather than exposed on the cell surface (Fig. 1); they

control of antigen presentation in dendritic cells 249

have the capacity to capture, process, and present newly encountered antigens;they express low levels of T-cell costimulatory molecules; and if they are fixedto prevent their maturation in culture, they are unable to activate naive T cells(Henri et al., 2001; Inaba et al., 1992; Ruedl et al., 2000; Salomon et al., 1998;Vermaelen et al., 2001; Wilson et al., 2003, 2004). Moreover, these DCs can beconverted into mature DCs, similar to the tissue-derived DCs present in theLNs, by injecting stimulatory compounds in vivo or by culture in vitro (Fig. 1)(De Smedt et al., 1996; Inaba et al., 1992; Reis e Sousa and Germain, 1999;Schulz et al., 2000; Wilson et al., 2003, 2004). This implies that in the steadystate virtually all the DCs contained in the spleen, and half of all the DCscontained in the LN, are immature (Wilson et al., 2003).

Does this mean that the blood-derived DC types migrate from the periph-eral tissues without acquiring a mature phenotype? The answer is, probablynot. The three blood-derived DC subsets turn over quickly; for instance, theyare completely replaced in the spleen within 3 to 5 days (Kamath et al., 2000,2002). Yet none of these three subsets has been identified in peripheral tissuesor circulating in the blood (O’Keeffe et al., 2002b, 2003). The blood doescontain plasmacytoid DCs and a different DC population that has somefeatures in common with the CD4þ, CD8þ, and CD4�CD8� DCs, but thisblood DC type is probably not the precursor of the three types contained inthe lymphoid organs (Table 1) (O’Keeffe et al., 2002b, 2003). This suggests thatthe three blood-derived DC populations contained in the spleen and the LNsarise directly in the lymphoid organs from earlier hematopoietic precursors.There, the vast majority spend their entire life span in an immature stateunless they are activated in situ.

Clearly, the life cycle of the blood-derived DC types is different from that ofthe tissue-derived DCs (Fig. 1). Why? We suggest this enables them to play asimilar function to the tissue-derived DCs, and perhaps in addition twodistinct functions. The similar function is immunosurveillance. When onethinks of areas of the body exposed to pathogen infection, the epithelialsurfaces of the skin, the airways, the gut, or the genitourinary tract quicklyspring to mind, but not the blood. However, the blood provides a majorpathway for the entry and spreading of insect-borne pathogens, which includethe causal agents of major diseases: malaria, leishmaniosis, Lyme disease,yellow fever, several forms of encephalitis, etc. The blood-derived DCs areideally located to detect and initiate immune responses against these patho-gens (Henri et al., 2002; Konecny et al., 1999; Perry et al., 2004). In addition,blood-derived DCs may play two additional functions. First, they may provideMHC–peptide ligands for recognition by naive T cells in the steady state,which may be required to maintain T-cell homeostasis (Sprent and Surh,2003). Second, they may be responsible for the induction of peripheral

250 nicholas s. wilson and jose a. villadangos

tolerance by presenting self-antigens in an immature state (Finkelman et al.,1996; Hawiger et al., 2001; Steinman and Nussenzweig, 2002). However, wewish to point out that this does not discard the possibility that the tissue-derived DCs that constantly reach the LNs in the steady state may also play arole in tolerance induction (Albert et al., 2001). Perhaps both (immature)blood-derived DCs and (mature) tissue-derived DCs can induce peripheraltolerance in the steady state, and only when the maturation process has beentriggered by an inflammatory signal associated with infection or tissue damagedo the DCs acquire an immunogenic phenotype. If this were the case, thephenotypic feature distinguishing tolerogenic from immunogenic DCs wouldnot be the maturational status, but rather the ‘‘expression’’ of an additionalsignal induced only by activatory stimuli (Albert et al., 2001; Shortman andHeath, 2001).

2.3. Correlations and Gaps Between the Human and Mouse DC Systems

For obvious reasons, the best-studied DC types in humans are those circulat-ing in the blood and not those found in the lymphoid organs. Human bloodcontains two major DC groups equivalent to those found in mouse blood(Hart, 1997; Shortman and Liu, 2002). The human lymphoid organs alsocontain tissue-derived DCs (in the LNs) (Geissmann et al., 2002; Takahashiet al., 1998) and the two major groups of blood-derived DC: pDCs andconventional DCs (Bendriss-Vermare et al., 2001; McIlroy et al., 2001;Schmitt et al., 2000; Summers et al., 2001; Vandenabeele et al., 2001).However, whether the human conventional DCs comprise the equivalents ofmouse CD4þ, CD8þ, and CD4�CD8� DCs is unknown. As in the mouse, thevast majority of the blood-derived DCs contained in the lymphoid organs ofnoninfected individuals have an immature phenotype (Bendriss-Vermare et al.,2001; McIlroy et al., 2001; Summers et al., 2001), and the number of matureDCs increases dramatically in patients suffering bacterial infections or multi-ple trauma (McIlroy et al., 2001). Therefore it is likely that blood-derived DCsin mice and humans follow a similar life cycle.

2.4. In Vitro-Generated DCs

The development of methods to produce DCs in vitro from precursors foundin mouse bone marrow (BMDCs) (Inaba et al., 1992) or spleen (D1DCs)(Winzler et al., 1997), and from monocyte precursors found in human blood(MoDCs) (Sallusto and Lanzavecchia, 1994), represented a major break-through for studies of DC biology and the development of DC-based im-munotherapies. These methods enabled maintenance of large numbers of DCs

control of antigen presentation in dendritic cells 251

in culture in an immature state and provided the capacity to follow theirdevelopmental changes during maturation triggered by activatory signalsadded to the cultures. Most of the studies on antigen presentation by DCshave been performed on these DC models. However, the in vivo equivalent ofthe in vitro–generated DCs are unknown, so it is important to verify whetherthe findings of studies employing these DC models are applicable to DCsgenerated in vivo.

3. Antigen Presentation Pathways

3.1. Some Antigen Presentation Terminology

To avoid confusion and to enable an informative discussion on the mechanismsthat make DCs highly efficient antigen-presenting cells (APCs), we consideredit would be important to define a number of often incorrectly used terms tocategorize antigens. The term self will be used to refer to antigens that exist ina normal, uninfected individual, such as the proteins encoded by the genomeor the products of the host’s metabolism. Model proteins such as ovalbumin(OVA) or hen egg lysozyme (HEL) expressed in transgenic mice fall into thiscategory. We refer to antigens expressed by an infectious agent or to proteinsinjected in mice as foreign.

We use the term endogenous to refer to components synthesized by theAPCs themselves, as opposed to exogenous components, which are takenchiefly by endocytosis. Thus, viral antigens synthesized by an infected APCare ‘‘endogenous and foreign,’’ whereas a tissue antigen endocytosed by a DCis ‘‘self and exogenous,’’ and an endocytosed bacterium or injected ovalbumin(OVA) is ‘‘foreign and exogenous.’’ The terms cytosolic and endosomal, whichshould only be used to refer to the location where proteins normally performtheir function within the cell, are sometimes used as synonymous with endog-enous and exogenous, respectively, but we think this is misleading. It is truethat most cytosolic proteins are endogenous, but it is incorrect to assume thatall the contents of the endosomes are exogenous. In fact, these contents arepredominantly endogenous, comprising membrane proteins that are deliveredto lysosomal compartments at the end of their life span, or normal componentsof the endocytic route such as proteases, ATPases, etc.

3.2. Two Antigen Presentation Systems

To obtain antigenic peptides for MHC presentation, APCs utilize the twomajor systems employed by eukaryotic cells to dispose of proteins. The firstsystem is found in the endocytic route and comprises a large collection of

252 nicholas s. wilson and jose a. villadangos

proteases with variable substrate specificity and pH requirements (McGrath,1999). Most of the endosomal proteases are collectively known as cathepsins.Endocytosed proteins, whether self or foreign, endogenous or exogenous, aredegraded by these proteases in the endosomal compartments (Honey andRudensky, 2003; Villadangos and Ploegh, 2000). In APCs, the resulting pep-tides can be loaded into the peptide-binding groove of MHC class II (MHC II)molecules and then presented on the plasma membrane for recognition byCD4þ T cells. When DC researchers talk about ‘‘developmental control ofantigen presentation,’’ normally they are refering to the MHC II presentationpathway. This is one of the major themes in the sections that follow.

The second major proteolytic system used by eukaryotic cells is the protea-some, a multimeric complex found in the cytosol that is composed of severalproteolytic and regulatory subunits (Rock et al., 2002). The peptides generatedby the proteasome can be translocated by the TAP transporter into the endo-plasmic reticulum (ER), where they are loaded into the binding groove of newlysynthesized MHC class I (MHC I) molecules. The resulting MHC I–peptidecomplexes then follow the default secretory pathway through the Golgi complexand are displayed on the plasma membrane for inspection by CD8þ T cells.With a few exceptions, all cells are capable of presenting antigens via MHC Imolecules. In most APCs, the polypeptides that access the cytosol are synthe-sized by the APC itself, so most of the antigens presented by MHC I moleculesare endogenous. A notorious exception to this rule is the cross-presentationpathway, which enables DCs and macrophages to present exogenous antigensvia MHC I molecules. The process of antigen degradation by the proteasome(Rock et al., 2002), and the formation of MHC I–peptide complexes in the ER(Cresswell et al., 1999; Purcell, 2000), have been extensively reviewed else-where. It can be stated that DCs do not exhibit specific mechanisms for thecontrol of presentation of endogenous antigens via MHC I molecules, so werefer the reader to those reviews for detailed discussion of the ‘‘classic’’ MHC Ipresentation pathway. In contrast, cross-presentation has special significance inthe DC system and is described in detail after we review the mechanims thatcontrol MHC II antigen presentation.

4. Mechanisms and Regulation of Antigen Capture in DCs

Whether it is for MHC II presentation or for MHC I cross-presentation,sampling of the environment plays a central role in the capacity of DCs tomaintain peripheral tolerance to tissue-specific antigens and to activate im-mune responses against foreign pathogens. In this section we first describe theantigen-capture capabilities of immature DCs, followed by a review of themechanisms suggested to regulate endocytosis during DC maturation.

control of antigen presentation in dendritic cells 253

4.1. Antigen Uptake in DCs

Antigens can be endocytosed by a variety of mechanisms (Fig. 2) (Watts, 1997).Large particulates (bacteria, cells, or artifical beads) are often recognized bymembrane receptors that trigger the formation of large endocytic vesicles(phagosomes), a process known as phagocytosis. Formation of the phagosomerequires recruiting the actin cytoskeleton, which is necessary to mold theplasma membrane around the phagocytosed particle. The process of macro-pinocytosis is mechanistically similar to phagocytosis and also requires actin,but in this case the vesicle (a macropinosome) does not form around a particle;instead it simply engulfs a large portion of extracellular medium. Anothermechanism to engulf extracellular medium is micropinocytosis, which doesnot require actin polymerization for vesicle formation (a pinosome), but in-stead requires the recruitment of the cytosolic protein clathrin to generateclathrin-coated pits. Finally, receptor-mediated endocytosis consists of theinternalization of molecules recognized by specific membrane receptors,which also trigger the formation of clathrin-coated pinosomes. Uptake ofmaterial by macro- and micropinocytosis is often referred to as ‘‘fluid-phase’’endocytosis to indicate that it is nonspecific rather than being triggered byparticular molecular cues intrinsic to the endocytosed material. DCs wereinitially believed to have a low endocytic capacity because, as discussedabove, the protocols used to isolate DCs from tissues stimulated their sponta-neous maturation in vitro, in turn causing the downregulation of endocytosis(reviewed by Steinman and Swanson, 1995). It is now clear that DCs are highlyefficient at all forms of endocytosis in their immature state.

4.1.1. Phagocytosis

DCs can phagocytose bacteria (Rescigno, 2002), yeasts, hyphae (d’Ostianiet al., 2000), protozoans (Konecny et al., 1999), portions of live cells(Harshyne et al., 2001), and whole dying cells. Phagocytosis of apoptotic andnecrotic cells has received particular attention in studies of DC function (for areview, see Heath et al., 2004). Whereas apoptosis is an ordered process usedto eliminate unwanted cells, including those deleted during development andthose at the end of their life span, necrotic death can be associated withpathological conditions such as viral infection. The ability to capture apoptoticand necrotic cells, and distinguish between these two forms of antigen, mayenable DCs to present cell-associated antigens for the maintenance of toler-ance (apoptotic cells) and for induction of immune responses (necrotic cells),respectively (Steinman et al., 2000). It has been postulated that not all DCtypes may possess the capacity to capture cell-associated antigens, suggestinga specialization in DC subset function based on variability in antigen uptake.

Figure 2 Antigen uptake, processing, and presentation in the MHC II presentation pathway.Antigens are endocytosed by actin-dependent (macropinocytosis and phagocytosis) or clathrin-dependent (receptor-mediated endocytosis and pinocytosis) mechanisms. The internalized mate-rial then moves along the endocytic route, which can be subdivided into three major types ofcompartments: early endosomes (EE), late endosomes (LE), and lysosomes (Lys). The physico-chemical conditions in these compartments are progressively more acidic, reducing, and proteo-lytic, leading to gradual degradation of the endocytosed antigens. Antigenic peptides are thusreleased along the entire endocytic route. Newly synthesized MHC II ab dimers are transportedinto the endocytic route in association with the chaperone Ii, which associates with the MHC IImolecules by inserting its CLIP region into the peptide-binding site contained in the ab dimer.The cytosolic tail of Ii carries a sorting motif for delivery of the abIi complexes into EEs directlyfrom the trans-Golgi network or after transient expression on the cell surface. Once the ab–Iicomplexes reach the endocytic route, Ii is degraded in a stepwise fashion. The protease asparaginylendopeptidase (AEP) may be involved in converting full-length Ii into Iip10, which is thendegraded by cathepsin (Cat) S to generate CLIP [one of the major intermediate steps (theab–Iip22 complex) has been removed for simplicity]. The chaperone H-2DM then promotes theexchange of CLIP for antigenic peptides. The conversion of ab–Ii into ab–CLIP has been drawnbetween brackets in all the endosomal compartments to represent that one or more of the stepsinvolved in this process can occur at several stations of the endocytic route. Therefore theab–peptide complexes can be generated in EEs, LEs, or Lys, and be transported from all thesecompartments to the cell surface, although the majority are probably generated in LEs.

254 nicholas s. wilson and jose a. villadangos

Indeed, several groups have reported that only the CD8þ DCs can phagocy-tose and present cell-associated antigens (Iyoda et al., 2002; Schulz and Reise Sousa, 2002; Valdez et al., 2002). However, the studies by Thomson’s group(Morelli et al., 2003) and our own unpublished results indicate that CD8� DCs

control of antigen presentation in dendritic cells 255

are also capable of capturing and presenting cell-associated antigens. Whetherthese discrepancies are due to differences in the cellular antigens analyzed orthe methods of assessment used by the different studies is yet to be estab-lished.

4.1.2. Macropinocytosis

Macropinocytosis is constitutive in immature DCs, whereas in other cells suchas macrophages it must be induced by phorbol esters or growth factors(Swanson and Watts, 1995). It has been estimated that immature DCs caninternalize by macropinocytosis the equivalent of their own volume in 60 min(Sallusto et al., 1995). To maintain a steady uptake of such a large volume ofextracellular fluid, immature DCs rely on regulated expression of several mem-bers of the aquaporin family (de Baey and Lanzavecchia, 2000), membranechannels that facilitate the elimination of excess water across the endosomalmembranes (Engel et al., 2000). Engulfment of large portions of extracellularvolume followed by release of excess water thus results in highly efficientconcentration of solutes in the endosomal compartments of immature DCs.

4.1.3. Receptor-Mediated Endocytosis

Receptor-mediated endocytosis is another strategy of antigen capture fullyexploited by DCs. DCs express a multitude of receptors that they may use totake up different forms of antigens. These include Fc receptors (den Haan andBevan, 2002; Fanger et al., 1996; Ravetch and Bolland, 2001; Sallusto andLanzavecchia, 1994), the immunoglobulin-like molecule ILT3 (Cella et al.,1997a), and members of the C-type lectin family such as the mannose receptor(Sallusto et al., 1995), DEC-205 (Jiang et al., 1995; Mahnke et al., 2000), DC-SIGN (van Kooyk and Geijtenbeek, 2003), langerin (Valladeau et al., 2000),and CIRE (Caminschi et al., 2001). Strikingly, the expression pattern of severalof these putative antigen receptors varies among DC subsets, suggesting thatthe DC populations may be specialized at presenting antigens derived fromdifferent sets of pathogens (Caminschi et al., 2001; Linehan et al., 1999;Mommaas et al., 1999; Valladeau et al., 2000; Vremec and Shortman, 1997).In addition, some of these receptors can deliver activatory or inhibitory signals,so they may have a dual role as both antigen receptors and modulators ofthe immune response (Chang et al., 2002; Chieppa et al., 2003; den Haan andBevan, 2002; Regnault et al., 1999; van Kooyk and Geijtenbeek, 2003).Such receptors could thus be considered ‘‘pathogen recognition receptors’’(Janeway, 1989) akin to the members of the Toll-like receptor (TLR) family(Medzhitov, 2001). Unfortunately, the putative pathogen ligand(s) for ILT3and several of the C-type lectins expressed by DCs are still unknown, so it

256 nicholas s. wilson and jose a. villadangos

remains unclear whether all these molecules have a role in antigen capturein vivo; alternative roles in mediating intercellular interactions or adhesion toextracellular matrix components are also possible (van Kooyk and Geijtenbeek,2003). Characterizing the role of these receptors and their ligands is important,because they may provide suitable targets for the induction of immunogenic ortolerogenic reactions in vivo (Hawiger et al., 2001; Mahnke et al., 2003).

4.2. Control of Endocytosis in DCs

Activation of in vitro-cultured DC leads to a drastic downregulation of phago-cytosis and macropinocytosis (Sallusto et al., 1995; Winzler et al., 1997). Thegroups of Mellman and Watts identified two members of the Rho family ofGTPases, cdc42 and rac1, respectively, as key regulators of this process(Garrett et al., 2000; West et al., 2000), but it is still unknown which are theupstream and downstream factors that control the activity of these twoGTPases during DC maturation. Endocytosis via some of the antigen recep-tors is also downregulated during DC maturation because of decreased ex-pression of the receptors (Sallusto et al., 1995; Stoitzner et al., 2003; Valladeauet al., 2000; Winzler et al., 1997).

These observations have led to the assumption that mature DCs shut downall forms of endocytosis, but this is incorrect. Pinocytosis still occurs in matureDCs at a comparable rate to immature DCs. This point is illustrated by anapproximately equal number of clathrin-coated pinosomes in immature andmature DCs (Garrett et al., 2000; West et al., 2000). In fact, the turnover rateof the bulk of the plasma membrane proteins of immature and mature DCs iscomparable (Wilson et al., 2004); since degradation of membrane proteinsoccurs primarily in the endocytic route, this indicates that the rate of internali-zation and degradation of plasma membrane components does not sub-stantially decrease on maturation. Furthermore, mature DCs still internalizesubstantial amounts of soluble proteins, which are most likely endocytosed bymicropinocytosis or receptor-mediated endocytosis (Pure et al., 1990; Winzleret al., 1997). Supporting the concept that mature DCs are still proficient atcapturing certain antigens, not all the putative antigen receptors are down-regulated during DC maturation; for instance, DEC-205 is upregulated, evenin DC populations that do not express this lectin in their immature state (Inabaet al., 1994; Vremec and Shortman, 1997). These observations imply that theinability of mature DCs to process and present soluble antigens via the MHCII pathway is not due to impaired uptake. As we describe below, this inability isin fact due to downregulation of MHC II synthesis and therefore a lack ofavailable peptide-receptive molecules. Why then are some antigen receptorsstill expressed in mature DCs? One reason might be to provide antigens for

control of antigen presentation in dendritic cells 257

cross-presentation, because mature DCs do not downregulate MHC Isynthesis. On the other hand, perhaps those receptors that are not down-regulated on maturation are not, after all, dedicated to antigen presentationbut rather to adhesion to other cell receptors or extracellular components. Thispoint stresses the need to identify the natural ligands of the putative antigenreceptors expressed by DCs.

Adding further confusion, it is unclear whether the decrease in phagocytosisand macropinocytosis observed in in vitro–generated DCs is reflected in theirin vivo counterparts. Ruedl and Itano have shown that mature skin-derivedDCs are still highly endocytic (Itano et al., 2003; Ruedl et al., 2001), anobservation confirmed by ourselves (unpublished observations). However,LCs isolated from epidermal sheets and cultured in vitro downregulatephagocytosis (Reis e Sousa et al., 1993). One possible explanation for thesediscrepancies might be that phagocytosis is regulated independently of theother phenotypic features that define DC maturation, namely the antigen-presenting and T-cell–stimulatory capacities. If this is the case, DCs maydownregulate phagocytosis in vivo only if they mature as a response to activa-tory signals, but not when they mature in the steady state (Itano and Jenkins,2003; Ruedl et al., 2001).

Finally, it is important to note that measuring endocytosis in vivo withtracers is prone to the influence of DC location within the lymphoid organsand peripheral tissues. For instance, tracers inoculated in the skin are ineffi-ciently captured by blood-derived DCs, but the same tracers injected intrave-nously are preferentially captured by the blood-derived DCs and not thetissue-derived DCs (Itano et al., 2003; Salomon et al., 1998). This probablyreflects different pathways followed by the tracers to enter the lymphoid organ(lymph versus blood), which dictates which DC populations will have access tothe tracer. In addition, based on a highly organized conduit system, thelymphoid organs impose limits to the size of the particles or molecules thatcan access different regions (Gretz et al., 1997, 2000; Kaldjian et al., 2001;Nolte et al., 2003). Since the distribution of DC subsets varies (De Smedt et al.,1996; Pulendran et al., 1997), some DC types may be unable to capture certainantigens, not because they lack the capability, but because they do not haveaccess to the antigens (Itano and Jenkins, 2003).

5. Antigen Presentation by MHC II Molecules

Formation of MHC II–peptide complexes is an intricate process that entailsthe orchestration of multiple cellular processes, including protein-sortingmechanisms, proteolytic activities, the intervention of chaperones, etc. Manyof these mechanisms have been reported to be regulated in DCs in a manner

258 nicholas s. wilson and jose a. villadangos

distinct from other APCs. Although the process of MHC II antigen presenta-tion has been reviewed extensively (Sercarz and Maverakis, 2003; Villadangos,2001; Watts, 1997; Wolf and Ploegh, 1995), in order to discuss the contentiousarea of how this process is regulated in DCs, it is necessary to recapitulate themajor steps involved in this pathway.

5.1. The Endocytic Pathway

Recounting the process of MHC II antigen presentation requires a descriptionof the subcellular structures where formation of the MHC II–peptide com-plexes takes place: the endocytic pathway (Gruenberg, 2001; Mellman, 1996).This pathway is a complex network of tubulovesicular structures, which can bevisualized as a major road spanning from the plasma membrane to the lyso-somes. This road is intersected by secondary incoming and outgoing tracks(Fig. 2). The incoming tracks represent vesicles originating mostly from thecell surface and the distal components of the Golgi complex, and the exit tracksdenote tubules and vesicles that eventually will fuse with the plasma mem-brane. The contents of the endosomal compartments change gradually alongthe major track and therefore it is not possible to identify clear boundariesdividing the endocytic pathway, but for the sake of simplicity, three majorregions are usually distinguished. First, the early endosomes (EEs), whoselimiting membrane and lumen have a similar composition to the plasmamembrane and the extracellular medium, respectively; second, the late endo-somes (LEs), which already contain many components found only in theendocytic route such as proteases, and are more acidic than the EE; third,the lysosomes, which are considered the final station of the endocytic pathway,are acidic and highly proteolytic.

Endocytosed material moves predominantly along the EE–LE–lysosomesaxis toward the lysosomes. As a rule, the number of entry and exit tracks alongthe major axis is most abundant in the EEs and least in the lysosomes, but howany given protein accesses the endocytic route, and how it leaves, is deter-mined by sorting mechanisms that are still poorly understood. For example,the transferrin receptor cycles continuously from the plasma membrane to theEEs and back to the surface (Mellman, 1996); the DC lectin DEC-205transports antigens bound at the plasma membrane to LEs and lysosomes,releases the antigens, and then returns to the cell surface (Mahnke et al.,2000); and the endosomal proteases are delivered to the endocytic routedirectly from the Golgi and are largely retained in endocytic compartments(Chapman et al., 1997). In the case of transmembrane proteins, the ‘‘addresscode’’ that determines how the protein will enter the endocytic route, whetherit will leave and how, can be encoded in its cytoplasmic region (Kirchhausen

control of antigen presentation in dendritic cells 259

et al., 1997). These address codes are probably recognized by cytosolic pro-teins that regulate protein sorting, but little is known about how the informa-tion carried by the transmembrane region is interpreted by the sortingmachinery, or how the machinery works (Gruenberg, 2001).

It is important to point out two features of the endocytic route that arecritical in the context of antigen presentation. First, the sorting mechanismsallow for some degree of ‘‘leakiness,’’ so that, for example, the cathepsins canbe both secreted and delivered to endosomal compartments (Lennon-Dumenil et al., 2001; Wolters and Chapman, 2000). Indeed, the MHC IImolecules are remarkably flexible and follow multiple entry and exit tracks ofthe endocytic pathway (see below). The second feature to consider is that theendocytic route is not always ‘‘linear,’’ as otherwise suggested by our simplifieddescription of the relationships between the different compartments. Rather, itis possible that some compartments ‘‘branch off,’’ becoming distinct reservoirsfor endocytosed material, or parallel subpathways with different dynamics andcomposition than those of the main EE–LE–lysosomal track. The significanceof such branching off is discussed later in the context of MHC II antigenpresentation and MHC I cross-presentation.

5.2. Antigen Degradation

Whatever mechanism is used for their internalization (see above), antigensprogress along the EE–LE–lysosomal axis, where they are exposed to an in-creasingly more acidic, reducing, and proteolytic environment (Fig. 2; Bryantet al., 2002; Honey and Rudensky, 2003; Villadangos and Ploegh, 2000). Theresult is a gradual release of polypeptides along the entire endocytic pathway,a few of which will contain the correct amino acid sequence that permitsthem to lodge into the peptide-binding sites of the particular MHC II allotypesexpressed by the APC (Engelhard, 1994). The objective of the MHC IImolecules is to capture the most diverse array of peptides possible in order todisplay the largest amount of antigenic information to CD4þ T cells(Villadangos, 2001).

5.3. Formation of the MHC II–Peptide Complexes

Newly synthesized MHC II ab dimers are cotranslationally translocated intothe endoplasmic reticulum (ER). However, the peptides that the MHC IImolecules bind are not brought into the ER as is the case for the peptidespresented by MHC I; instead, the MHC II molecules need to be transportedto the endosomal compartments. To enhance their efficiency as peptide bin-ders, the MHC II molecules have evolved a safety mechanism that prevents

260 nicholas s. wilson and jose a. villadangos

their expression as empty dimers on the cell surface: they are unstable and havethe propensity to aggregate if their peptide-binding cavity is empty (Bonnerotet al., 1994; Ceman et al., 1998; Germain and Rinker, 1993; Zhong et al., 1996).In addition, since the peptide-binding site is highly promiscuous, it can prompt-ly associate with other polypeptides contained in the ER (Busch et al., 1996), aprocess that would preclude ab dimers from acquiring peptides in theendosomes. To prevent this early inactivation and to stabilize the conformationof the ab dimer, the MHC II molecules are synthesized in an inactive ‘‘pro-form’’ in which their peptide-binding cleft is occupied by the chaperone invari-ant chain (Ii) (Fig. 2). The Ii also contains in its cytosolic portion a sorting motifthat is recognized in the trans-Golgi network as a signal to haul the ab–Iicomplexes out of the secretory pathway and into the EE/LE regions of theendocytic pathway (Bakke and Dobberstein, 1990; Lotteau et al., 1990; Pieterset al., 1993).

Once the MHC II–Ii complexes reach the endoctytic compartments, the ab

dimers must eliminate Ii to regain their capacity to bind antigenic peptides.This process occurs in several steps. First, several proteases progressivelycleave Ii, producing its major degradation intermediates Iip22, Iip10, andfinally the CLIP fragment, a stretch of amino acids that occupies the MHC IIpeptide-binding cleft (reviewed in Bryant et al., 2002; Honey and Rudensky,2003; Villadangos and Ploegh, 2000). Only one of these proteolytic reactionshas been attributed to a particular enzyme: the cleavage of Iip10, which inDCs is accomplished by cathepsin S. The importance of cathepsin S has beendemonstrated in vivo by the analysis of knockout mice (Nakagawa et al., 1999;Shi et al., 1999), although the impact of the cathepsin S deficiency on MHC IIpeptide binding is strongly dependent on the particular MHC II allelic variant(allotype) expressed by the APC (see More Exceptions than Rules, below)(Nakagawa et al., 1999; Riese et al., 2001; Villadangos et al., 1997, 2001).A study by the Watts group suggests that the protease asparaginyl endopepti-dase (AEP) may be responsible for the initial cleavage of full-length Ii(Manoury et al., 2003), but the role of this enzyme in vivo awaits confirmationin AEP-deficient animals.

The conversion of Iip10 into CLIP has special significance, because itreleases the ab dimers from the cytoplasmic portion of Ii, whose endosomaltargeting motif acts as a retention signal that prevents the ab dimers associatedwith Ii, Iip22, or Iip10 from escaping to the cell surface (Amigorena et al.,1995; Brachet et al., 1997; Driessen et al., 1999; Neefjes and Ploegh, 1992;Pierre and Mellman, 1998). Strikingly, interfering with the kinetics of Iidestruction can have profound effects not only on MHC II trafficking, butalso on the dynamics of endosomal maturation (Gorvel et al., 1995; Gregerset al., 2003; Nordeng et al., 2002; Pieters et al., 1993; Romagnoli and Germain,

control of antigen presentation in dendritic cells 261

1994). Thus, in DCs that cannot convert the ab–Iip10 complex into ab–CLIPbecause of the lack of cathepsin S, the number of endosomal vesicles increasesdramatically (Driessen et al., 1999; and our unpublished observations) andtrafficking of other components of the endocytic pathway is indirectly affected(Riese et al., 2001). This phenomenon may be due to excessive coating ofendosomes expressing Iip10 with the chaperone Hsc70, which associates withthe tail of Ii on the cytosolic face of the endosomes (Lagaudriere-Gesbert et al.,2002). Although it is still unclear how Hsc70 influences endosome biogenesisand maturation, the consequences of Iip10 accumulation in cathepsin S-deficient APCs suggest that the role of Ii may not be solely to control MHCII trafficking and peptide loading but also to act as a master regulator of theendocytic route in APCs. By linking Ii degradation with endosomal trafficking,the APCs may be able to orchestrate the timing of generation of peptide-receptive MHC II dimers with the gradual degradation of endocytosedantigens (Lagaudriere-Gesbert et al., 2002).

Once the MHC II–CLIP complex has been generated, the CLIP peptidemust be substituted for antigenic peptides. This reaction involves the chaper-one H-2DM and (in B cells) its close relative H-2DO, which interact tran-siently with MHC II–CLIP, destabilizing the complex and facilitating therelease of CLIP (reviewed in Alfonso and Karlsson, 2000). This enablesantigenic peptides that contain the correct combination of ‘‘anchor residues’’to associate with the now vacant peptide-binding groove (Engelhard, 1994).The newly formed MHC II–peptide complexes can then be sorted from theendocytic route toward the plasma membrane in transport vesicles (Fig. 2).Since the conversion of Iip10 into CLIP releases the MHC II complex of the Iicytoplasmic tail, this proteolytic step must be quickly followed by the exchangeof CLIP for antigenic peptides. Indeed, in H-2DM–deficient APCs most of thesurface MHC II molecules remain associated with CLIP (Fung-Leung et al.,1996; Martin et al., 1996; Miyazaki et al., 1996). H-2DM not only assists inremoving CLIP, it also acts as a peptide editor, promoting the exchange of low-affinity peptides for high-affinity peptides (Jensen et al., 1999; Katz and Sant,1994; Kropshofer et al., 1996; van Ham et al., 1996). Controlling the abun-dance or location of H-2DM in the endosomal compartments may thus act as arheostat regulating the stability of the MHC II–peptide complexes generatedby the APCs (Kropshofer et al., 1997b).

5.4. Where Does Peptide Binding Occur?

Much of the research performed on MHC II antigen presentation has beendirected toward characterizing the compartments of the endocytic route wherethe MHC II molecules acquire their peptide cargo (Amigorena et al., 1994;

262 nicholas s. wilson and jose a. villadangos

Kropshofer et al., 1997b; Peters et al., 1991; Qiu et al., 1994; Tulp et al., 1994;West et al., 1994; reviewed in Geuze, 1998; Neefjes, 1999). Today, the ac-cepted view is that just as antigens are degraded gradually along the entireEE–LE–lysosomal track, MHC II molecules become receptive to bindingantigenic peptides, or polypeptide precursors, at all the stations of the endo-somal pathway (Castellino and Germain, 1995; Driessen et al., 1999;Villadangos et al., 2000). Likewise, MHC II–peptide complexes can exit fromall compartments to the plasma membrane (Driessen et al., 1999). This abilityenables MHC II molecules to sample peptides from the entire endocyticroute, including peptides that may only be present in EEs because they donot survive the harsher conditions of LEs or lysosomes, and also peptides thatrequire thorough degradation of their polypeptide precursors late in theendocytic route (Sercarz and Maverakis, 2003; Villadangos, 2001; Watts,1997; Wolf and Ploegh, 1995).

Importantly, the spatial organization of the mechanisms employed by MHCII molecules to acquire their peptide cargo implies that all the peptides derivedfrom any endocytosed protein, whether endogenous or exogenous, self orforeign, are included together in a single antigenic peptide pool. Therefore,any mechanism controlling antigen presentation that nonspecifically interfereswith antigen degradation, or with the formation of peptide-receptive MHC IImolecules, should affect the presentation of the entire pool. This concept isimportant when we later evaluate the hypotheses on how MHC II antigenpresentation is regulated in immature and mature DCs.

5.5. More Exceptions than Rules

Although our description of the conversion of MHC II–Ii into MHC II–peptide complexes suggests a rigid sequence of events, this process hasevolved considerable flexibility. Much of the flexibility is imparted by the effectthat MHC polymorphism has on the affinity of the interaction between thepeptide-binding cleft and CLIP. This in turn causes variability in the require-ments of each allotype for accessory molecules. For instance, I-Ab moleculesare highly dependent on association with Ii for conformational stability andeggress out of the ER, but I-Ak can be transported to the cell surface as emptyab dimers in the absence of Ii (Bikoff et al., 1995; Rovere et al., 1998); lack ofcathepsin S impairs conversion of MHC II–Iip10 into MHC II–peptide com-plexes in APCs expressing I-Ab or I-Ad, but Iip10 associates loosely with I-Ak,I-Aq, I-Au, or I-As, enabling these allotypes to acquire peptide cargo even whencathepsin S has been inactivated (Honey et al., 2002; Nakagawa et al., 1999;Riese et al., 2001; Villadangos et al., 1997, 2001); likewise, CLIP can beremoved from some allotypes without the assistance of H-2DM (Avva and

control of antigen presentation in dendritic cells 263

Cresswell, 1994; Bikoff et al., 2001; Brooks et al., 1994; Koonce et al., 2003;Stebbins et al., 1995, 1996; Wolf et al., 1998). Probably these differences alloweach MHC II allelic product to acquire its peptide-binding capacity at differ-ent positions along the endocytic route, in which the concentration of cathep-sin S and H-2DM vary. Since, in normal populations, each individual canexpress several combinations of I-A and I-E (or HLA-DR and HLA-DP)allotypes, this diversity probably contributes to increase the capacity of eachindividual to respond to a variety of antigenic challenges (Villadangos, 2001).

Another source of plasticity in the MHC II presentation pathway is a degreeof ‘‘leakiness’’ in most of the steps described above. Thus, some MHC II–Iicomplexes are not delivered directly to the endocytic route but are transientlyexported to the plasma membrane and then internalized (Fig. 2; Benarochet al., 1995; Koch et al., 1991; Saudrais et al., 1998; Wraight et al., 1990). Inaddition, some peptides can displace the Ii degradation intermediates Iip22or Iip10 and not just CLIP (Denzin et al., 1996; Kropshofer et al., 1997a;Sanderson et al., 1996; Stebbins et al., 1996). Finally, Ii can be eliminated froma fraction of MHC II–Ii complexes in EEs in a reaction independent ofcysteine proteases and H-2DM; a process still poorly understood(Villadangos et al., 2000). All these factors contribute to increase the numberof different endocytic compartments where ab dimers can access the peptidecurrency, thereby enlarging the repertoire of antigenic peptides that APCs canpresent to CD4þ T cells (Villadangos, 2001).

6. Control of MHC II Antigen Presentation in DCs

The studies that have addressed how MHC II antigen presentation is con-trolled in DCs during their maturation have tried to answer three majorquestions. First, why do mature DCs express more MHC II molecules ontheir surface than immature DCs? Second, what causes the striking differencein subcellular distribution of MHC II between immature and mature DCs(Fig. 1C)? Finally, and most importantly, how do the mechanisms that controlantigen presentation in DCs translate into the functional antigen-presentingproperties of immature and mature DCs. Immature DCs efficiently captureantigens, but they are poor at displaying antigenic peptides at the plasmamembrane. In contrast, mature DCs show extended presentation of a ‘‘memo-ry’’ (or a ‘‘snapshot’’) of the antigens they captured at the time of activation, butthey are incapable of presenting subsequently encountered antigens. How arethese changes regulated?

Several mechanisms have been proposed in an attempt to answer the abovequestions, and in summary they can all be broadly assigned to one of twomodels (Fig. 3). The first suggests that immature DCs are inefficient in their

Figure 3 Two models of developmental control of antigen presentation in DCs. In the gain offunction model, immature DCs have impaired formation of ab–peptide complexes as a result ofpoor antigen degradation, inhibition of Cat S, and/or impairment of H-2DM function. Most of theMHC II molecules are thus retained in endosomal compartments and degraded. Activation allowsformation of MHC II–peptide complexes that are then transported to the plasma membrane,where they are long-lived. In the interruption model, immature DCs constitutively generate MHCII–peptide complexes and transport them to the cell surface, but then they are endocytosed anddegraded. Activation downregulates the rate of MHC II–peptide internalization from the plasmamembrane. See Fig. 4 for a more detailed description of the interruption model.

264 nicholas s. wilson and jose a. villadangos

ability to generate MHC II–peptide complexes. On receiving an activationsignal, the now ‘‘maturing’’ DCs rapidly gain the capacity to create and presentMHC II–peptide complexes. We refer to this model as the ‘‘gain of function’’model, for it suggests that DCs gain the capacity to process and presentantigens during maturation. The second model supports the concept thatimmature DCs constitutively generate MHC II–peptide complexes and pres-ent them transiently on their surface, but the complexes are then rapidlyendocytosed and destroyed. According to this model, during maturation,newly generated MHC II–peptide complexes are retained on the plasmamembrane, that is, MHC II–peptide turnover is ‘‘interrupted’’ (Fig. 3). Here

control of antigen presentation in dendritic cells 265

we review the experimental data produced by different groups to analyzeMHC II antigen presentation in DCs, assessing whether the results obtainedsupport the gain of function or the interruption model.

6.1. Control of Endosomal Proteolysis in DCs

Control of antigen degradation has been proposed as one mechanism used byDCs to regulate MHC II antigen presentation. Low proteolytic activity in theendosomal route of immature DCs could limit the availability of peptideligands for MHC II molecules, causing impaired antigen presentation. Thishypothesis is supported by two studies that found that the overall activity ofendosomal cysteine proteases, measured using active site-labeling techniques,increased during maturation in human monocyte–derived DCs (MoDCs) andmouse bone marrow–derived DCs (BMDCs) (Fiebiger et al., 2002; Trombettaet al., 2003). The increase in activity was not simply caused by augmentedprotease expression in mature DCs because the total amount of the enzymesdid not vary significantly on maturation. These studies also showed, using pH-sensitive fluorescent endocytic tracers, that the average pH in the endocyticroute of mature DCs was more acidic than in the immature DCs, consistentwith a low-pH optimum for the majority of endosomal proteases. Therefore,acidification of the endosomal compartments could explain the increase inprotease activity in mature DCs. To further support this, the study byTrombetta and colleagues showed that some components of the vacuolarHþ-ATPase (V-ATPase), the pump responsible for lysosomal acidification,were recruited from the cytosol to the endosomal compartments on DCmaturation, suggesting an active mechanism of control of the endosomal pH(Trombetta et al., 2003). Modulating the endosomal pH can also affect thekinetics of endosomal transport: inhibitors of the V-ATPase, such as concana-mycin A and B, impair trafficking of MHC II and other components of theendocytic pathway from EEs to LEs and from LEs to lysosomes (Benarochet al., 1995; van Weert et al., 1995; Villadangos et al., 2000; Yilla et al., 1993).Together these studies suggest that DCs may control endosomal acidificationas an elegant mechanism to orchestrate antigen internalization and degrada-tion. On the other hand, considering the importance of endosomal proteolysisin cell metabolism, it might be expected that the activity of the endosomalcompartments of immature DCs should still be sufficient to degrade theproteins that are normally eliminated via the endosomal route. This consid-ered, is such ‘‘basal activity’’ sufficient to process antigens in immature DCs?

Several studies have suggested that DCs are indeed inefficient at processingantigens until they receive a maturation stimulus. This hypothesis is supportedby microscopy and biochemical measurements of the rate of degradation of

266 nicholas s. wilson and jose a. villadangos

some exogenous proteins. For instance, hen egg lysozyme (HEL) and horse-radish peroxidase (HRP) were still detectable in endocytic vesicles of im-mature BMDCs up to 2 days after their internalization, and their rates ofdegradation were estimated by Western blot to be less that 25% in 19 h(Trombetta et al., 2003). Activation of the BMDCs led to rapid degradationof the internalized HEL and HRP. These results support the concept that theproteolytic activity in the endosomal compartments of immature BMDCs islow, and this activity is rapidly upregulated on maturation. But are HEL andHRP representative of the bulk of the proteins endocytosed by immatureBMDCs? The BMDCs are cultured in medium supplemented with fetal calfserum (FCS), which typically contains approximately 40 mg of protein permilliliter (yielding a concentration of 4 pg/pl in culture medium supplementedwith 10% FCS). Since the volume of extracellular medium internalized by oneimmature DC has been estimated as 3 pl/h (Sallusto et al., 1995), the amountof protein internalized by immature BMDCs in culture can be estimated to beabout 12 pg/h. This is likely an underestimate, because it does not considerthe amount of protein contained in other cells or cell debris present in theculture, which presumably are also phagocytosed. If the rate of degradation ofthe bulk of the internalized proteins were that observed for HEL or HRP(roughly 25%/day), immature BMDCs would accumulate over 200 pg ofprotein per day. Since the protein content of one whole cell has been estimatedas 200 pg (see, e.g., Princiotta et al., 2003), this rate of accumulation wouldobviously be unsustainable. A similar argument can be made for immatureDCs constitutively sampling their environment in vivo.

Clearly, the half-life of the bulk of the proteins endocytosed by immatureBMDCs must be much shorter than that observed for HEL or HRP. Indeed,bovine serum albumin (BSA), which constitutes approximately 60% of the totalprotein contained in FCS, had a half-life of less than 3 h in immature BMDCs(Trombetta et al., 2003). In fact, immature and mature BMDCs degraded thisprotein with comparable kinetics (Trombetta et al., 2003). Thus, if BSA wereused as a reference model instead of HEL or HRP, it would have to beconcluded that the endosomal proteolytic activities of immature and matureDCs are both high and comparable. In an attempt to obtain a more compre-hensive assessment of such activity, we examined the turnover rate of plasmamembrane proteins, which are mostly degraded in endosomal compartmentsfollowing internalization. We observed that, with the notorious exception ofMHC II molecules (see below), immature and mature DCs degraded most oftheir plasma membrane proteins with comparable kinetics (Wilson et al.,2004). Moreover, the half-life of the bulk of the plasma membrane proteinswas comparable to that of BSA (a few hours) (Wilson et al., 2004). Thereforeit can be concluded that the long half-life of HEL and HRP represents an

control of antigen presentation in dendritic cells 267

exception rather than the rule of protein processing in the endosomal com-partments of immature DCs. An explanation for the exceptional resilience ofHEL and HRP to endosomal degradation might be their resistance to unfold-ing, which in turn may be conferred by the disulfide bonds contained in thesetwo proteins. It would be interesting to test whether immature and matureDCs degrade the denatured forms of HEL and HRP with comparable kinetics.If they did, this would suggest that immature DCs are less capable of unfoldingnative proteins, perhaps due to lower expression of disulfide reductases such asIP30/GILT (Arunachalam et al., 2000; Luster et al., 1988; Maric et al., 1994).

An alternative mechanism that may explain the long half-life of someendocytosed proteins in immature DCs could be the existence of ‘‘storage’’compartments with low proteolytic activity (Lutz et al., 1997). Such compart-ments would represent ‘‘cul-de-sacs’’ branching off the main endocytic track,where a small fraction of the endocytosed material could be stored for laterprocessing, while the remaining protein mass progresses to the lysosomes andis degraded. Some proteins such as HEL or HRP may perhaps be preferen-tially sorted from the major endocytic pathway into these compartments, thusexplaining their unusual long life in immature DCs. Activation signals couldinduce recruitment of proteases into the storage compartments. This hypothe-sis is supported by a study by Driessen and colleagues, which showed that DCactivation induces protease recruitment from lysosomes into late endosomes(Lautwein et al., 2002). However, the existence of these storage compartmentsremains to be verified by other independent studies.

To complicate this picture further, a report by Lennon-Dumenil and colla-borators found that when DCs receive a maturation signal, they downregulatethe proteolytic activity of the endosomal compartments encountered by theendocytosed antigens (Lennon-Dumenil et al., 2002). This conclusion is inapparent contradiction with the studies mentioned above and suggests thatmature DCs degrade endocytosed antigens slower than immature DCs.However, the analysis by Lennon-Dumenil and colleagues used an activeprotease probe linked to latex beads to directly assess the proteolytic activityof the compartments traversed by an endocytic tracer, whereas the formerstudies measured protease activity in cell lysates or disrupted subcellularfractions. In addition, this study measured the changes in endosomal proteol-ysis at early stages of the maturation process. Therefore, it is possible thatalthough the overall proteolytic activity of the endosomal route increases inmature DCs, these gross changes are not uniformly distributed in all theendocytic compartments.

We favor the concept that immature DCs do not accumulate endocytosedproteins, because these would constitute a liability when the DCs encountera pathogen. If immature DCs were constantly accumulating endocytosed

268 nicholas s. wilson and jose a. villadangos

proteins, their endosomal compartments would be ‘‘saturated’’ with self-pro-teins. As a result, a small amount of foreign antigen captured in the presence ofan activatory signal would need to compete with a pool of preendocytosed self-proteins for binding to the available MHC II molecules. A more plausiblescenario is that immature DCs quickly dispose of the proteins they endocytosein the steady state, so that a foreign antigen captured in the context of animmune challenge can achieve a high relative representation on MHC IImolecules, with minimum interference from self-proteins. In this situation, areduction in proteolytic activity in the compartments containing the foreignantigen would provide a more controlled environment for antigenic proces-sing, favoring the preferential loading of newly synthesized MHC II moleculeswith foreign peptides (see below).

In conclusion, while the proteolytic activity of the endocytic route undergoessome changes during DC maturation, which are likely to affect the processingof certain antigens, endosomal proteases contained in immature DCs haveadequate activity to process antigens in the steady state. The question thatremains is whether the resulting antigenic peptides generated in immatureDCs are loaded and presented by MHC II molecules.

6.2. Formation of MHC II–Peptide Complexes in DCs

Indeed, whether immature and mature DCs regulate antigen presentation atthe level of formation of MHC II–peptide complexes has been a contentiousissue. Some reports conclude that immature DCs are inefficient at generatingpeptide-receptive MHC II molecules and therefore cannot present antigens(Inaba et al., 2000; Kleijmeer et al., 2001; Pierre and Mellman, 1998; Turleyet al., 2000), whereas others reach the opposite conclusion (Cella et al., 1997b;Colledge et al., 2002; El-Sukkari et al., 2003; Pierre et al., 1997; Veeraswamyet al., 2003; Villadangos et al., 2001; Wilson et al., 2004). Determining whetherimmature DCs can present antigens is central to understanding their putativerole in maintaining tolerance to self-antigens, a role that would obviouslyrequire them to present those antigens. In this section we describe experi-mental approaches that have been employed to analyze the formation of MHCII–peptide complexes in DCs and the conclusions from such studies.

6.2.1. Is Ii Degradation a Regulatory Mechanism in DCAntigen Presentation?

The first study that suggested the generation of MHC II–peptide complexes inDCs was developmentally regulated reported a difference in degradation ofIip10 between immature and mature mouse BMDCs (Pierre and Mellman,

control of antigen presentation in dendritic cells 269

1998). Immature DCs contained more Iip10 in their endocytic compartmentsthan mature DCs, suggesting that immature DCs were inefficient at degradingIip10. Since Iip10 cleavage is a prerequisite step in MHC II–peptidecomplex formation (see above and Fig. 2), deficient Iip10 processing wouldimpair antigen presentation by immature DCs. Moreover, because theMHC II–Iip10 complexes carry the endosomal retention signal contained inthe cytosolic portion of Ii, the complexes would accumulate in endosomalcompartments, consistent with the observations of microscopy analysis ofMHC II distribution in immature DCs (Fig. 1). This study also reported thepresence of the cysteine protease inhibitor cystatin C (Abrahamson, 1994) inthe endosomal compartments of immature DCs and suggested that cystatinC prevented Iip10 degradation by blocking the activity of cathepsin S (Fig. 3).Cystatin C disappeared from the endosomal route of mature DCs, which led tothe hypothesis that on DC activation cathepsin S would become active anddegrade Iip10, allowing the production of MHC II–peptide complexes (Pierreand Mellman, 1998). These observations supported an attractive gain offunction model to explain how DCs regulate antigen presentation, with cysta-tin C acting as the key switch. However, this study had a potential caveat:measurement of Iip10 content was performed by Western blot analysis ofwhole cells or cell fractions, not from immunoprecipitated MHC II molecules.Therefore, the relative amounts of MHC II molecules that were associatedwith Iip10 in immature and mature DCs were not determined. Thiswas relevant because DCs downregulate MHC II synthesis on maturation(see below; and Cella et al., 1997b; Kampgen et al., 1991; Pure et al., 1990;Rescigno et al., 1998; Villadangos et al., 2001; Wilson et al., 2004). Therefore,immature DCs are at any time synthesizing and processing more MHC IImolecules than mature DCs (a 3-fold difference in BMDCs). This implies thatimmature DCs also contain a larger number of processing intermediates(including MHC II–Iip10 complexes) than mature DCs, even though theyare degrading Iip10 normally. In fact, we have shown by Western blot thatmost of the MHC II molecules contained at any time within immatureBMDCs, D1DCs, or lymphoid organ DCs are associated with peptides, notwith Iip10 (El-Sukkari et al., 2003; Villadangos et al., 2001; Wilson et al., 2004).Therefore, the accumulation of MHC II molecules in the endosomal com-partments of immature DCs (Fig. 1) cannot be attributed to their associationwith Iip10.

The concept that the developmental control of antigen presentation isregulated in DCs independently of the rate of Iip10 degradation was alsosuggested by reports examining ‘‘cathepsin S–independent’’ MHC II allotypes(see above). These allotypes (which include I-Ak, I-Aq, I-Au, and I-As) canconvert MHC II–Iip10 into MHC II–peptide even in cathepsin S–deficient

270 nicholas s. wilson and jose a. villadangos

DCs (Honey et al., 2002; Nakagawa et al., 1999; Riese et al., 2001; Villadangoset al., 1997, 2001). If MHC II antigen presentation were regulated in DCs bycontrolling cathepsin S activity, the cathepsin S–independent allotypes wouldbe oblivious to this control mechanism. However, it was found that theybehave identically to the ‘‘cathepsin S–dependent’’ allotypes (Villadangoset al., 2001). Analysis of DCs from Ii-deficient mice (Rovere et al., 1998) andcathepsin S–deficient mice (Driessen et al., 2001; Nakagawa et al., 1999;Shi et al., 1999) also suggested that the mechanism of control of MHC IIexpression was not based on the kinetics of Ii degradation.

More recently, we were able to directly assess the role of cystatin C incontrolling DC antigen presentation by studying cystatin C knockout mice:we observed no alteration in control of MHC II expression and subcellulardistribution in immature cystatin C-null BMDCs, D1DCs, or spleen DCs (El-Sukkari et al., 2003). Interestingly, we detected a striking difference in thelevels of cystatin C expression among DC populations: CD8þ DCs synthesizemuch larger quantities of cystatin C than do CD8� DCs, and they accumulatecystatin C in MHC IIþ endosomal compartments. This was surprisingbecause cystatin C is considered a ubiquitously expressed protein and itsgene lacks any obvious regulatory elements (Abrahamson, 1994; Abrahamsonet al., 1990; Huh et al., 1995). The differential pattern of expression of cystatinC among closely related DC subsets suggests the existence of hitherto un-known mechanisms of expression of this protease inhibitor. Nevertheless, theexpression of cystatin C in CD8þ DCs did not affect the MHC II presentationof several antigens (El-Sukkari et al., 2003), so it is unclear whether cystatinC plays any specific antigen presentation function in this DC subset.

6.2.2. Is H-2DM a Regulatory Molecule in DC Antigen Presentation?

The chaperone H-2DM has also been suggested as a key regulatory elementcontrolling MHC II antigen presentation in DCs. Analysis of the subcellularlocalization of MHC II and H-2DM in immature D1DCs revealed that MHCII was localized to the internal vesicles of multivesicular bodies, whereas H-2DM localized to the limiting membrane of the bodies (Kleijmeer et al., 2001).The physical separation of H-2DM from MHC II might prevent the conver-sion of MHC II–CLIP into MHC II–peptide and impair antigen presentation.On maturation, the internal vesicles fuse with the outer membrane of themultivesicular bodies, which would enable H-2DM to exert its chaperoneactivity on MHC II–CLIP and allow formation of MHC II–peptide complexes(Kleijmeer et al., 2001). As in the reports that suggested cystatin C representedthe rate-limiting step in the production of peptide-receptive MHC II mole-cules, this study favored the gain of function model of control of antigen

control of antigen presentation in dendritic cells 271

presentation (Fig. 3). However, limitations were also evident: first, it did notexplain how it would operate on MHC II allotypes that do not require H-2DMto convert MHC II–CLIP into MHC II–peptide (Albert et al., 1996; Avva andCresswell, 1994; Bikoff et al., 2001; Koonce et al., 2003; Stebbins et al., 1995;Wolf et al., 1998); second, for the I-Ab allotype, lack of H-2DM activity leads toaccumulation of I-Ab–CLIP complexes, but this study did not observe suchcomplexes in immature DCs; third, in H-2DM-null mice MHC II–CLIPcomplexes are transported normally to the plasma membrane (Fung-Leunget al., 1996; Martin et al., 1996; Miyazaki et al., 1996), so even if the MHC IImolecules of immature DCs failed to eject CLIP, it is unclear how this wouldresult in their poor expression at the plasma membrane. We suggest that thesegregation of H-2DM from MHC II observed in the multivesicular bodies ofimmature DCs may occur after the MHC II molecules have acquired theirpeptide cargo. This segregation, and its control during maturation, may reflectdifferential sorting events that control MHC II trafficking in immature andmature DCs (see below) (van Lith et al., 2001).

6.2.3. Is MHC II–Peptide Complex Formation Regulated in DCs?

Techniques that directly assess the formation of individual MHC II–peptidecombinations have been used by several laboratories to examine how DCscontrol MHC II peptide loading. These include using monoclonal antibodies(mAbs) specific for individual MHC–peptide complexes and antigen presenta-tion assays T-cell hybridomas or naive T cells. In a series of studies, immatureBMDCs were incubated with HEL in the absence or the presence of anactivation stimulus (LPS). Formation of a complex between I-Ak and anHEL-derived peptide was then assessed, employing an mAb or T cells specificfor this complex. Strikingly, the I-Ak–HELpep complexes could not bedetected by microscopy until the BMDCs were activated, suggesting thatimmature BMDCs did not generate peptide-receptive I-Ak molecules (Inabaet al., 2000; Turley et al., 2000). However, the complex could be detectedby Western blot analysis of cell lysates of both immature and mature DCs(Trombetta et al., 2003). Moreover, fluorescence-activated cell sorting (FACS)analysis revealed that the proportion of surface I-Ak molecules that wereassociated with the HEL peptide in immature and mature DCs was compara-ble (Inaba et al., 2000; Veeraswamy et al., 2003). Finally, probably the mostcompelling evidence that I-Ak–HELpep complexes were formed by immatureBMDCs incubated with HEL was that these cells stimulated HEL-specificT cells, demonstrating that they did present the complex (Inaba et al., 2000;Veeraswamy et al., 2003). Therefore, even though immature DCs appear to berelatively inefficient at processing HEL (see above), this deficiency certainly

272 nicholas s. wilson and jose a. villadangos

does not translate to their inability to load and present to T cells I-Ak moleculesloaded with HEL-derived peptides. Indeed, a conclusion of these studies isthat since most antigens are probably more sensitive to proteolysis than HEL(see above), immature DCs will present most antigens even more efficientlythan HEL. Perhaps the failure to detect the I-Ak–HELpep complex by mi-croscopy was due to low sensitivity of the antibody used. In a similar study,Colledge and colleagues tracked formation of a complex between I-Ek and apigeon cytochrome c (PCC)–derived peptide in immature BMDCs, showingthat the complex could be detected by microscopy in endosomal compart-ments just 5 min after incubation of the cells with intact PCC (Colledge et al.,2002). The immature BMDCs then rapidly transported the I-Ek–PCCpepcomplex to the cell surface. This demonstrated that immature BMDCscan eficiently process PCC, load their I-Ek dimers with the resulting anti-genic peptides, and then present these complexes on the plasma membrane.Importantly, all the studies summarized above checked that neither HEL norPCC induced BMDC maturation, discarding the possibility that contaminatingLPS or other inflammatory compounds triggered activation of the immatureBMDCs.

Together these studies support a scenario in which immature BMDCsconstitutively load their MHC II molecules with peptides derived from endo-cytosed antigens and transport the resulting complexes to the cell surface. Isthis also the case for DCs generated in vivo? To answer this question, thegroup of Unanue (Veeraswamy et al., 2003) and ourselves (Wilson et al., 2004)used transgenic mice expressing HEL or OVA as models of self-proteins,respectively, to determine whether the immature lymphoid organ DCs pre-sented antigens in the steady state. Both studies used mAbs and/or antigen-specific T cells to assess formation of I-Ak–HELpep or I-Ab-OVApepcomplexes. The conclusions of these two studies were that in vivo immatureDCs constitutively load their MHC II molecules with peptides derived fromself-proteins and present these complexes to T cells.

Using mAbs and T cells to assess MHC II–peptide loading restricts thescope of studies to the few complexes for which mAbs or T cells have beengenerated. A complementary approach consists of using biochemical methodsto assess the composition of the MHC II molecules. This approach takesadvantage of the unique resistance of the MHC II–peptide complexes todenaturation in SDS at room temperature (Germain and Hendrix, 1991;Springer et al., 1977). Even though the ab–peptide trimers do not containinterchain disulfide bonds, they run in SDS–PAGE as a complex with an Mr ofapproximately 50 kDa. This property is lost if the complexes are first dena-tured at high temperature (‘‘boiling’’). Used in Western blot or combined withmetabolic radiolabeling, this biochemical approach represents a powerful

control of antigen presentation in dendritic cells 273

tool to measure the kinetics, intermediate steps, and efficiency of the conver-sion of MHC II–Ii into MHC II–peptide in DCs and other APCs. Forexample, cathepsin S-deficient DCs show impaired formation of MHC II–peptide SDS-stable complexes and accumulate MHC II–Iip10 complexesinstead; this deficiency can be easily visualized by Western blot or pulse–chase analysis of MHC II immunoprecipitates (Nakagawa et al., 1999; Shiet al., 1999; Villadangos et al., 2000). Likewise, the MHC II–CLIP complexesthat accumulate in H-2DM–deficient APCs can be distinguished from the bulkof normal MHC II–peptide complexes by their distinct mobility in SDS–PAGE (Fung-Leung et al., 1996; Martin et al., 1996; Miyazaki et al., 1996;Villadangos et al., 2000). Studies that have employed this biochemical ap-proach to compare MHC II peptide loading in immature and mature mouseBMDCs (El-Sukkari et al., 2003; Pierre et al., 1997; Veeraswamy et al., 2003),D1DCs (Rescigno et al., 1998; Villadangos et al., 2001), or lymphoid organDCs (El-Sukkari et al., 2003; Wilson et al., 2004), and in human monocyte-derived DCs (Cella et al., 1997a; Saudrais et al., 1998), have reported unim-paired generation of MHC II–peptide complexes in immature DCs. One studyalso indicates that immature DCs express a relatively large number of ‘‘empty’’molecules (ab dimers devoid of antigen peptides), which can be recognizedwith an mAB (Santambrogio et al., 1999). In any case, this population of emptymolecules represented only a fraction of the total (20% approximately,L. Santambrogio, personal communication), with the majority of MHC IImolecules associated with regular peptides. Therefore it can be concludedthat immature DCs from both mouse and human, either generated in vitro orin vivo, constitutively load their MHC II molecules with antigenic peptides.Hence, the mechanisms that control MHC II antigen presentation in DCsmust operate post-MHC II–peptide complex formation.

6.3. Control of MHC II–Peptide Trafficking in DCs

Although the bulk of the MHC II molecules contained at any time in bothimmature and mature DC are loaded with peptides, the fate and localization ofthese complexes clearly differ between the two maturational stages: MHC IImolecules are short-lived and accumulate in endosomal compartments inimmature DCs but are long-lived on the plasma membrane of matureDCs. What mechanism is responsible for these fundamental differences?A straightforward explanation would be that only a few MHC II–peptidecomplexes can escape to the cell surface in immature DCs, the rest beingretained intracellularly and then degraded in lysosomes; on maturation, thesecomplexes would be permitted to traffic to the cell surface. However, analternative possibility is that the MHC II–peptide complexes are in fact

274 nicholas s. wilson and jose a. villadangos

transiently expressed on the surface of the immature DCs and then quicklyinternalized, delivered to lysosomes, and degraded. Quantitation of the num-ber of MHC II–peptide complexes that are transported to the cell surface inimmature DCs, measured by pulse–chase followed by cell surface biotinyla-tion and immunoprecipitation, favor the second mechanism (Baron et al.,2001; Cella et al., 1997a; Villadangos et al., 2001). Therefore, most of theMHC II molecules that accumulate in the endosomal compartments of imma-ture DCs are not really ‘‘retained,’’ but rather are in the process of acquiringpeptide cargo to progress to the cell surface or else are moving along theendocytic route after being internalized from the cell surface (Figs. 3 and 4).An appropriate metaphor would be a playing videotape (see below), where theMHC II molecules are represented by the tape, the portion of the tapeaccessible to the head of the VCR the ‘‘cell surface,’’ and the interior of thecassette the ‘‘endocytic route’’: for an outside observer, most of the tape neverappears to leave the interior of the cassette, but over time the entire length oftape will be transiently exposed on the surface.

What happens when immature DCs receive an activation signal? The disap-pearance of MHC II molecules from internal compartments, and their ac-cumulation on the plasma membrane, suggest a direct transfer of MHC IImolecules from endosomes to the DC surface. However, it has long beenrecognized that the increase in MHC II surface expression that accompaniesLC maturation requires protein synthesis (Pure et al., 1990; Shimada et al.,1987; Witmer-Pack et al., 1988). In accordance with this, we have describedthat brefeldin A (BFA, an inhibitor of protein export out of the ER) andcycloheximide (CHX, a protein synthesis inhibitor) block the increase inMHC II surface expression in maturing D1DCs or spleen DCs withoutimpairing the degradation of the preexisting intracellular MHC II–peptidepool (Villadangos et al., 2001; Wilson et al., 2004). Similar effects have beenobserved with actinomycin D (a transcription inhibitor) (D. Vremec andK. Shortman, personal communication). Therefore, the majority of the MHCII–peptide complexes that accumulate on the surface of maturing DCs origi-nate from de novo synthesis; the preexisting MHC II–peptide complexes thatwere localized in the endosomal compartments when the immature DCsreceived the activation signal contribute little, and most are degraded instead.This conclusion makes sense because removing CLIP from newly synthesizedMHC II–CLIP is kinetically more favorable than displacing peptides frompreloaded MHC II–peptide complexes (Kropshofer et al., 1999). Therefore aDC maturing in response to an endocytosed pathogen will present moreefficiently pathogen-derived peptides by using newly synthesized MHC IImolecules.

Figure 4 DCs regulate MHC II antigen presentation by controlling the rates of MHC II synthesisand MHC II–peptide endocytosis. The MHC II molecules of immature DCs (top left) constitu-tively generate MHC II–peptide complexes, which are transiently expressed on the cell surface,internalized, and then degraded. This causes the apparent accumulation of MHC II moleculesin the endosomal compartments of immature DCs (see Fig. 1C). Therefore, immature DCs canconstantly present self-antigens (open triangles) to T cells. On activation (top right), MHC IIsynthesis is transiently upregulated and turnover of surface MHC II–peptide complexes is down-regulated. This causes an accumulation of newly generated MHC II–peptide complexes on theplasma membrane. Some of these complexes are loaded with peptides derived from antigenscaptured at the site and time of activation (solid triangles). MHC II synthesis is subsequentlydownregulated (bottom left) to prevent the substitution of the MHC II–peptide complexes thatreached the plasma membrane early during maturation. This process enables mature DCs (bottomright) to present a long-lived ‘‘snapshot’’ of the antigenic material captured when they becameactivated, even if the foreign antigens have been exhausted by proteolytic processing. The MHCII–peptide complexes generated on activation may traffic to the cell surface associated tomembrane microdomains.

control of antigen presentation in dendritic cells 275

276 nicholas s. wilson and jose a. villadangos

A fascinating series of studies by the groups led by Kleijmeer (Barois et al.,2002; Kleijmeer et al., 2001), Ploegh (Bertho et al., 2003; Boes et al., 2002,2003), and Mellman (Chow et al., 2002) have shown that maturation signalstrigger dramatic changes in the morphology of the endocytic compartments ofDCs (reviewed in Boes et al., 2004). While the compartments containingMHC II molecules in immature DCs appear vesicular, long tubular structuresare generated on DC maturation, providing exit ‘‘tracks’’ to the cell surface forthe MHC II–peptide complexes. Strikingly, in the absence of T cells thesestructures extend randomly toward the plasma membrane of the activatedDCs, but in their presence the tubules reach out toward the DC–T-cellinterface in an antigen-specific manner (Bertho et al., 2003; Boes et al.,2002, 2003). Although the implication of these observations in vivo are stillunclear, they suggest that T cells signal the maturing DCs to polarize transportof the newly generated MHC II–peptide complexes toward the site of DC–Tcell contact. Moreover, T-cell–costimulatory molecules are delivered to thecell surface in close proximity with the MHC II–peptide complexes (Turleyet al., 2000). These mechanisms have obvious advantages for forming theimmunological synapse (Huppa and Davis, 2003).

Why do the MHC II–peptide complexes accumulate on the surface of thematuring DCs? We support the view that this is due to reduced internalizationof MHC II–peptide complexes during DC maturation (Fig. 4). Indeed, webelieve this is the key step that controls MHC II expression and antigenpresentation in DCs. In immature DCs, surface MHC II–peptide complexesare rapidly internalized and then degraded, but on maturation they areretained on the plasma membrane as a long-lived ‘‘snapshot’’ of antigenicinformation (Baron et al., 2001; Cella et al., 1997b; Villadangos et al., 2001;Wilson et al., 2004). This is not a passive consequence of overall reduction ofmacropinocytosis and phagocytosis in the mature DCs, because the turnoverof most plasma membrane proteins, including MHC I, proceeds at comparablerates in immature and mature DCs (Delamarre et al., 2003; Wilson et al.,2004). This is also the case for the rate of pinocytosis, which shows little changeon DC maturation (see above) (Garrett et al., 2000; West et al., 2000).

Therefore, MHC II–peptide complexes are selectively excluded fromendocytosis and subsequent degradation on DC maturation. This conclusionimplies the existence of a specific mechanism of sorting of MHC II–peptidecomplexes actively regulated during DC maturation. Where and how do thesemechanisms operate? We can speculate on three possibilities. The first mightbe control of internalization of MHC II–peptide complexes from the plasmamembrane, a process that would be downregulated on DC maturation. Thesecond possibility is that the complexes are internalized by default in both theimmature and the mature DCs, but in the mature DCs most of the complexes

control of antigen presentation in dendritic cells 277

are recycled back from EEs to the cell surface instead of proceeding towardLEs and lysosomes, thus following a cycle similar to that of the transferrinreceptor (Mellman, 1996). It is well established that surface MHC II mole-cules can undergo such recycling (Pinet et al., 1995; Ramachandra andHarding, 2000; Reid and Watts, 1990), and indeed the kinetics of internaliza-tion of MHC II in immature DCs suggest that a fraction of the endocytosedMHC II–peptide complexes return to the surface (Cella et al., 1997a;Villadangos et al., 2001). Therefore it is plausible that DC maturation upregu-lates this recycling mechanism, causing the MHC II–peptide complexes ofmature DCs to constantly cycle between the plasma membrane and the EEs(Maxfield and McGraw, 2004). A third mechanism of control could be basedon recruitment of MHC II–peptide complexes to membrane microdomains:Roche’s group described accumulation of MHC II–peptide complexes in lipidrafts on B cells (Anderson et al., 2000), whereas Kropshofer and colleaguesobserved a similar phenomenon in DC, but involving tetraspan rich micro-domains, not lipid rafts (Kropshofer et al., 2002). This mechanism is notincompatible with the previous two; indeed, recruitment to membrane micro-domains might be the factor that modulates internalization and/or recycling ofthe MHC II–peptide complexes in DCs. If this were the case, then the sortingsignals controlling MHC II trafficking may not be contained in the MHC IImolecules themselves but in other, closely associated molecules within themembrane microdomains. In any case, the exact nature of these signals andthe sorting machinery associated with them have yet to be characterized. It isknown that the cytosolic portion of the MHC II b chain contains sorting motifs(for a review see Bakke and Nordeng, 1999), but whether these have a role incontrol of MHC II trafficking in DCs is unknown.

6.4. Control of MHC II Synthesis in DCs

If DC maturation results in ‘‘freezing’’ the MHC II–peptide complexes on theplasma membrane (or their entry in a cycle between the plasma membraneand the EEs), the net addition of newly synthesized MHC II–peptide com-plexes explains the increase in surface expression of MHC II that follows DCactivation. However, this process must stop when the number of MHC IImolecules reaches a certain level. DCs accomplish this by coordinating thereduction in the rate of surface MHC II turnover with a reduction in the rateof MHC II synthesis (Fig. 4). MHC II synthesis increases transiently shortlyafter DC activation, but then decreases gradually (Cella et al., 1997a; Rescignoet al., 1998; Wilson et al., 2004). In DCs generated in vitro, MHC II synthesisis reduced by approximately 70% 24–48 h postactivation (Cella et al., 1997a;Rescigno et al., 1998; Villadangos, 2001). This reduction is much more

278 nicholas s. wilson and jose a. villadangos

dramatic in LCs or splenic DCs activated in vitro, in which MHC II expressionis virtually shut off in 18 h (Kampgen et al., 1991; Pure et al., 1990; Wilson et al.,2004). This downregulation of MHC II synthesis reflects a normal program ofDC maturation in vivo, as it is observed in splenic DCs from mice injected withinflammatory compounds (Villadangos et al., 2001; Wilson et al., 2004), and inthe tissue-derived DCs isolated from LNs, which acquire a mature phenotypein the steady state (see above) (Wilson et al., 2004). In contrast to MHC II, therate of MHC I synthesis increases on DC maturation (Cella et al., 1997a;Delamarre et al., 2003; Rescigno et al., 1998; Villadangos, 2001; Wilson et al.,2004), which is consistent with the continued turnover of surface MHCI observed in mature DCs (Wilson et al., 2004). The surface expression levelof MHC II is thus regulated in DCs by a finely coordinated regulation ofthe rates of MHC II synthesis and MHC II–peptide internalization anddegradation (Fig. 4).

6.5. Corollary: The Cell Biological Basis of ‘‘Antigenic Memory’’

We conclude that the experimental evidence gathered by different laboratoriesin studies of mouse and human DCs obtained from in vitro or in vivo sourcessupport the interruption model of control of MHC II antigen presentation inDCs (Figs. 3 and 4). Using the metaphor described above, MHC II moleculesrepresent the tape of a videocassette. In immature DCs, the MHC II mole-cules are constantly ‘‘unwound’’ (synthesized) from the left ‘‘roll.’’ The MHC IImolecules then bind peptides derived from both endogenous and exogenousself-proteins, which are constantly endocytosed and processed by the imma-ture DCs. The resulting MHC II–peptide complexes are transiently exposedon the plasma membrane for recognition by the ‘‘VCR head’’ (the T cells), butthe complexes are then internalized and degraded (‘‘wound’’ into the videocas-sette roll at the right) (Fig. 4). For immature DCs in uninflamed peripheraltissues, this constant presentation of peptide-loaded MHC II molecules maybe irrelevant, because these DCs are unlikely to encounter naive T cells. Butfor the immature DCs that reside in the lymphoid organs, this process wouldenable DCs to provide T cells with a constant and updated supply of informa-tion about the antigens that reach the lymphoid organs, whether those antigensget there by themselves or are brought by other incoming cells (Belz et al.,2002; Inaba et al., 1998; Pooley et al., 2001; Salomon et al., 1998; Scheineckeret al., 2002; Zhong et al., 1997).

If immature DCs capture even a small amount of foreign antigen in thepresence of maturation signals, ‘‘winding’’ of the ‘‘tape’’ stops and MHC IImolecules start to accumulate on the ‘‘videocassette surface.’’ Many MHC IImolecules will still carry self-peptides, but some will present peptides derived

control of antigen presentation in dendritic cells 279

from the foreign antigen. As the MHC II–peptide complexes accumulate onthe surface of the maturing DCs, MHC II synthesis slows down and eventuallystops (Fig. 4). The net result of these synchronized changes is the characteris-tic increase in the level of surface MHC II that accompanies DC maturation.More importantly, by the time the DCs reach full maturity, even if the antigencaptured when DC activation took place is no longer available and all theendocytosed antigen ‘‘meal’’ has been processed, the mature DCs will haveaccumulated a sufficient number of MHC II molecules loaded with foreignpeptides for scanning by naive T cells. These complexes will be exposed on theDC surface for a long time (‘‘antigenic memory’’), thus increasing the chancesof encountering a pathogen-specific T-cell.

In this scenario, the changes undergone by DCs during the initial period ofmaturation are the most important for efficient antigen presentation. MHC IIsynthesis is initially upregulated (Cella et al., 1997a; Rescigno et al., 1998;Wilson et al., 2004), whereas the proteolytic environment encounteredby recently endocytosed antigen may be partially downregulated (Lennon-Dumenil et al., 2002). Together these two events would facilitate a steadyaccumulation of MHC II molecules loaded with peptides derived from anti-gens captured at the time of activation. The egress of these molecules to thecell surface may initially follow random directions and is accompanied byexpression of T-cell–costimulatory molecules (Turley et al., 2000). But assoon as an antigen-specific T cell detects its cognate MHC II–peptide complex,the T-cell signals the DC to polarize the transport of the MHC II–peptidecomplexes toward the DC–T-cell interface, thus promoting the formation ofthe immunological synapse (Bertho et al., 2003; Boes et al., 2002, 2003).

In conclusion, the major checkpoints that control antigen presentation inDCs and the mechanisms responsible for their antigenic memory are the rateof MHC II synthesis and the rate of turnover of surface MHC II–peptidecomplexes. We think that this interruption model of MHC II antigen presen-tation is consistent with most of the experimental evidence obtained by thedifferent groups engaged in studies of control of MHC II antigen presentationin DCs and can also explain the results that have been interpreted as favoringthe gain of function model.

These conclusions have several implications for studies of DC subtypecharacterization. Surface MHC II levels are often used as a marker of DCmaturity. For this reason, immature blood-derived DCs, which express moreMHC II than, for instance, immature BMDCs, are often referred to as‘‘semimature’’ or ‘‘intermediate’’ DCs. However, to contradict this assumptionthe turnover of surface MHC II is much higher in the immature blood-derivedDCs than in immature BMDCs, D1DCs, or MoDCs (Cella et al., 1997a;Villadangos et al., 2001; Wilson et al., 2004). Likewise, activated lymphoid

280 nicholas s. wilson and jose a. villadangos

organ DCs downregulate MHC II synthesis faster and more profoundly thantheir in vitro–grown counterparts (Cella et al., 1997a; Rescigno et al., 1998;Villadangos et al., 2001; Wilson et al., 2004). Therefore, the expression level ofMHC II is misleading as a marker of the antigen-presenting properties ofdifferent DC types. It is the behavior of the MHC II molecules, and theimmunological manifestation of such behavior (capacity to process and presentnewly encountered antigens and antigenic memory), that distinguish DCmaturity.

Our conclusions also have implications for DC-based immunotherapy.Namely, the success of a DC-based vaccine may depend more on the durationof presentation of MHC II–peptide complexes than on the total amount ofcomplexes generated during preparation of the vaccine (Schuler et al., 2003).Therefore the mechanisms that control MHC II–peptide turnover representpotential targets for drugs that could prolong the half-life of the MHC II–peptide complexes. Conversely, drugs that increased the turnover of self-MHC II–peptide combinations on mature DCs could interfere with autoim-mune responses. Indeed, one of the effects of the immunosuppresive cytokineinterleukin 10 (IL-10) is to counteract the changes induced by DC maturation,promoting MHC II–peptide turnover (Koppelman et al., 1997) and regulatingendosomal proteolysis (Fiebiger et al., 2001).

In addition, some studies suggest that to induce optimal T-cell activation,MHC II–peptide complexes must be generated within the antigen presenta-tion pathway, whereas incubating DCs with synthetic peptides leads to subop-timal recognition by T cells, even if a similar number of MHC II–peptidecomplexes is thus generated on the DC surface (Bertho et al., 2003). Thismight be because only in the first instance do the MHC II–peptide complexeslocalize in membrane microdomains that favor extended MHC II–peptidepresentation and T-cell activation (Anderson et al., 2000; Kropshofer et al.,2002).

7. DCs and Cross-Presentation

7.1. When Cross-Presentation Is Cross-Presentation?

Most cell types in the body express MHC I molecules that are loaded withpeptides generated in the cytosol and then translocated into the ER by theTAP transporter (Fig. 5). These peptides are derived from proteins degradedin the cytosol mostly by the proteasome (Yewdell et al., 2003). Such proteinsinclude any endogenously produced polypeptide, because even noncytosolicproteins such as secreted or membrane proteins, which are cotranslationallytranslocated into the ER, can be processed in the cytosol as defective

Figure 5 Mechanisms of MHC class I ‘‘direct presentation’’ and ‘‘cross-presentation.’’ In thedirect presentation pathway (left), MHC I molecules acquire peptide cargo in the endoplasmicreticulum (ER). The antigenic peptides are derived from polypeptides degraded in the cytosolmostly by the proteasome. The MHC I–peptide complexes shuttle to the cell surface by the defaultsecretory pathway. Cross-presentation entails the formation of a hybrid phagosome–ER hybridcompartment (ergosome), which contains endocytosed exogenous antigens, MHC I molecules, andthe components of the MHC I–peptide loading complex, including the peptide transporter TAP.Exogenous polypeptides are transported to the cytosol, perhaps via the Sec61 complex, and aredegraded in the proximity of the ergosome by the proteasome. The exogenous peptides are thentransported back into the ergosome by TAP for loading onto MHC I molecules. The resultingMHC I–peptide complexes are transported to the cell surface without crossing the Golgi complex.

control of antigen presentation in dendritic cells 281

ribosomal products (DRiPs) (Reits et al., 2000; Schubert et al., 2000; Yewdellet al., 1996, 2003). Exogenous proteins can also be presented by MHC Imolecules, a process that has been termed cross-presentation (Bevan, 1976).This term is sometimes used to refer to presentation of cell-associated anti-gens, even via the MHC II pathway. However, we suggest that the word cross-presentation should be used to refer only to presentation of exogenous anti-gens via MHC I molecules, independently of whether those antigensare soluble, bound to serum Ig as immunocomplexes, cell associated, etc.

282 nicholas s. wilson and jose a. villadangos

Although the existence or practical applications of the cross-presentationpathway are not contested, its significance in maintaining tolerance (cross-tolerance) and inducing immune responses (cross-priming) in vivo is highlycontroversial (Melief, 2003; Zinkernagel, 2002). In this section we focus onadvances in the delineation of the molecular mechanisms involved in cross-presentation. For a more thorough discussion on the role of cross-presentationin the immune system we refer the reader to other reviews (Heath andCarbone, 2001; Heath et al., 2004).

Cross-presentation may occur by different mechanisms. Some proteins cantraverse the plasma membrane and access the cytosol, where they are pro-cessed as endogenous proteins (Jeannin et al., 2000; Kim et al., 1997). Heatshock proteins may be capable of transporting small bound peptides into cells(Srivastava, 2002), although the exact nature of this phenomenon remainscontroversial (Nicchitta, 2003). Endocytosed exogenous proteins, processedin the endocytic route, can generate antigenic peptides suitable for binding toMHC I molecules recycled from the plasma membrane (Svensson et al., 1997).This mechanism is probably independent of the TAP transporter and thecomponents of the MHC I peptide-loading complex and thus is usuallyconsidered a distinct form of cross-presentation, sometimes referred to asthe ‘‘alternative pathway’’ (Campbell et al., 2000; Chen and Jondal, 2004;Gromme et al., 1999; Song and Harding, 1996). Finally, endocytosed antigenscan be actively transferred to the cytosol by a specialized mechanism andaccess the classic MHC I presentation pathway. In this review we focus onthis latter mechanism of cross-presentation because of its particular signifi-cance in DCs. We refer the reader to an excellent previous review by Yewdelland colleagues in this series (Yewdell et al., 1999), which discusses in detail thedifferent mechanisms of cross-presentation mentioned above.

7.2. A Novel Compartment for Cross-Presentation

The first direct evidence to suggest a mechanism of transport from the en-dosomes to the cytosol in cross-presentation was reported by Norbury andcolleagues, first using macrophages (Norbury et al., 1995) and then BMDCs(Norbury et al., 1997). These studies tracked the transport of horseradishperoxidase (HRP) from endosomal compartments into the cytosol. Rodriguezet al., published a similar study tracking the transport of HRP and OVA inD1DCs (Rodriguez et al., 1999). This study also demonstrated that OVA takenup as immunocomplexes was endocytosed and then transferred into the cyto-sol, whereas the Ig portion of the immunocomplex remained in the endosomalcompartments. This indicated that transfer to the cytosol was not simply byphysical disruption of the endocytic vesicle (Reis e Sousa and Germain, 1995)

control of antigen presentation in dendritic cells 283

but rather by a selective mechanism of transport. Indeed, the putative channelthat mediated endosome to cytosol transport could accommodate moleculeswith an Mr of up to 50 to 500 kDa, as indicated by analysis of transfer ofdextrans of variable size (Rodriguez et al., 1999).

A seminal study by the group of Desjardins provided evidence for thegeneration in macrophages of endosomal compartments that contained endo-plasmic reticulum (ER) components (Garin et al., 2001). These compartmentsresult from fusion of ER membranes with phagosomes (Desjardins, 2003).Building on this work, three groups have independently described a similarcompartment in macrophages and DCs, which may provide the physical struc-ture sufficient for cross-presentation (Ackerman et al., 2003; Guermonprezet al., 2003; Houde et al., 2003). We refer to these ER–phagosomal hybridcompartments as ergosomes (Fig. 5). The ergosomes contain endocytosed anti-gens, newly synthesized MHC class I molecules, and the multimolecular ma-chinery required for efficient formation of MHC I–peptide complexes,including the TAP transporter (Purcell, 2000). The study by Cresswell’s groupshowed that macropinosomes can also fuse with ER membranes, at least in DCs(Ackerman et al., 2003). This indicates that the range of antigens that can accessergosomes is probably not restricted to those that are phagocytosed, but alsoincludes soluble antigens taken up by macropinocytosis and perhaps receptor-mediated endocytosis. This suggestion is consistent with the observation thatDCs can cross-present a wide range of antigens including cell-associated,bacterial, immunocomplexed, and soluble proteins (Heath et al., 2004).

It has been hypothesized that the antigens contained in the ergosome aretransferred into the cytosol, where they are processed by the proteasome. Theresulting peptides are then translocated back into the ergosome via TAP togenerate MHC I–peptide complexes, which are then transported to the plasmamembrane (Fig. 5) (Ackerman et al., 2003; Guermonprez et al., 2003; Houdeet al., 2003). Interestingly, a feature of the complexes generated by the cross-presentation mechanism is that they do not appear to traffic through the Golgi,and so their carbohydrates do not undergo the modifications that would makethem resistant to treatment with the enzyme endoglycosidase H (Ackermanet al., 2003). This endoglycosidase H sensitivity may allow the distinctionbetween MHC I–peptide complexes generated via cross-presentation fromthose generated via the classic pathway.

7.3. Where Is the Exit, Please?

Perhaps the major ‘‘black box’’ in this model of cross-presentation is the transferof antigens into the cytosol. It has been suggested that this transport may bemediated by the Sec61 complex, a component of the ER that is also present in

284 nicholas s. wilson and jose a. villadangos

ergosomes (Guermonprez et al., 2003; Houde et al., 2003). The Sec61 complexis a heterotrimer that forms the channel that allows newly synthesized secretoryor membrane proteins to cross from the cytosol into the ER (Rapoport et al.,1996). This channel can also be used to transfer misfolded polypeptides fromthe ER to the cytosol for proteasomal degradation, a process termed ‘‘disloca-tion’’ or ‘‘retrotranslocation’’ (Brodsky and McCracken, 1999; Wiertz et al.,1996). The ergosome-to-cytosol transport of exogenous antigens required forcross-presentation would thus be reminiscent of the process of dislocation(Fig. 5). An involvement of Sec61 in this process is suggested by the observationthat the cholera toxin A1 subunit, a protein known to use Sec61 as a gate to crossfrom within the ER to the cytosol (Schmitz et al., 2000), can be exported fromergosomes (Houde et al., 2003). However, this observation does not prove thatthe cross-presented antigens are in fact translocated via Sec61, only that Sec61maintains its integrity in the ergosome.

The major problem with the hypothesis that Sec61 mediates transport ofexogenous antigens to the cytosol is the size of its channel. The group ofRapoport has provided the closest picture yet of Sec61 by reporting the crystalstructure of SecY, the archaeal homolog of Sec61 (Van den Berg et al., 2004).This study reveals that the diameter of the channel contained in a single SecYcomplex, through which proteins would have to cross the ER membrane, isonly 5–8 A, enough to accommodate only polypeptides in extended conforma-tion or, at most, a disulfide-linked loop. This contrasts with the studies by theWatts and Amigorena groups, which have shown that the ergosome-to-cytosoltransport mechanism can accommodate dextrans with an Mr over 50 kDa andglobular proteins such as enzymatically active HRP (Norbury et al., 1995, 1997;Rodriguez et al., 1999). These observations would suggest that the transport ofantigens from ergosomes to the cytosol is not carried out by Sec61 but by adifferent transporter. On the other hand, the study by Rapoport and co-workersdoes not discard the possibility that several juxtaposed Sec61 complexes couldcreate a much larger pore (Van den Berg et al., 2004). Indeed, electron micros-copy studies of solubilized Sec61 revealed ringlike structures with an open poreof �20 A (Hanein et al., 1996), whereas indirect measurements of the pore size,using fluorescent probes, raise this figure to at least 40 A—sufficient to accom-modate an Fab fragment, for instance (Hamman et al., 1997). Furthermore,previous studies of dislocation of MHC I heavy chains suggested that Sec61 canmediate the transport of glycosylated polypeptides (Wiertz et al., 1996). Basedon the crystal structure reported by Rapoport and colleagues, oligomerization ofSecY to form large pores appears unlikely, but the authors did not discard thispossibility (Van den Berg et al., 2004). Further elucidation of this questionawaits resolution of the three-dimensional structure of the mammalian Sec61complex. In conclusion, Sec61 remains an attractive candidate as the channel

control of antigen presentation in dendritic cells 285

responsible for ergosomal export, but until direct evidence for its involvement isobtained, a verdict for its implication in cross-presentation might be one morecase of ‘‘guilt by association.’’

7.4. Keeping the Team Together

Another intriguing question is whether cross-presentation requires the pres-ence of TAP, MHC I, and the peptide-loading machinery within the ergosome,and a close association between ergosomes and proteasomes. It would bereasonable to assume that the only critical step for cross-presentation wouldbe antigen transport into the cytosol. Once there, the exogenous antigensshould be processed as any endogenous component. However, antigen transferto the cytosol may not necessarily be followed by its presentation via MHC I. Ithas been shown that the products of proteosomal degradation are short lived inthe cytosol, and are quickly eliminated unless they are transported by TAP intothe ER, and then protected by lodging into the MHC I peptide-binding cavity(Reits et al., 2003). Therefore, failure to recruit to the ergosome any of thecomponents of the MHC I antigen presentation machinery could preventcross-presentation from occurring. Indeed, a report from the Mellman grouphas shown that both DCs and CD11c� cells (probably macrophages) growingin bone marrow cultures could transfer soluble OVA to the cytosol, but onlythe BMDCs loaded their MHC I molecules with OVA-derived peptides(Delamarre et al., 2003). Since the CD11c� cells could present endogenousproteins, indicating their MHC I peptide-loading machinery was operative,this report suggested that the failure of the CD11c� cells to cross-present OVAwas due to a deficiency downstream of translocation of OVA to the cytosol.This could be a lack of proteasomes, TAP, MHC I, or other components of thepeptide-loading complex in the ergosomes of CD11c� cells. A report from theJefferies group has described a conserved tyrosine-based sorting motif in thecytosolic portion of the MHC I heavy chain that is required for cross-presentation but not for endogenous antigen presentation (Lizee et al.,2003). This study supports the notion that the recruitment of MHC I mole-cules to ergosomes may depend on an active sorting mechanism that recog-nizes this tyrosine motif; such a mechanism could be operative in BMDCs butnot in CD11c� cells.

7.5. Is Cross-Presentation Regulated During DC Maturation?

Similar to MHC II presentation, MHC I cross-presentation may be develop-mentally regulated in DCs. Signals mediated by Fc receptors (den Haan andBevan, 2002), inflammatory compounds (Datta et al., 2003; Gil-Torregrosa

286 nicholas s. wilson and jose a. villadangos

et al., 2004), CD4þ T cells (Machy et al., 2002), and disruption of intercellularcontacts (Delamarre et al., 2003) have been reported to induce cross-presentation. We stress that we are referring to enhancement of formation ofMHC I–peptide complexes by cross-presentation and not to enhancementof T-cell activation by cross-priming; the second is a direct consequence ofincreased T-cell stimulatory capacity in activated DCs, which may or may notbe accompained by increased efficiency of loading of MHC I molecules withexogenously derived peptides. The reports above suggest that certain compo-nents of the cross-presentation machinery are upregulated on DC maturation,but the mechanisms underlying this regulation are still poorly characterized.

One obvious mechanism that may contribute to enhancement of cross-presentation (and direct presentation) is increased synthesis of MHC I (Cellaet al., 1997a; Rescigno et al., 1998; Villadangos et al., 2001; Wilson et al., 2004)and of the components of the peptide-loading machinery (Gil-Torregrosa et al.,2004). However, since not all activatory signals that enhance MHC I synthesisupregulate cross-presentation (Datta et al., 2003; Delamarre et al., 2003),some more specific mechanisms must exist. Pierre’s group has reported thatactivated DCs concentrate endogenous ubiquitinated proteins and DRiPs incytosolic aggregates (DALIs) (Lelouard et al., 2002). Formation of DALIsmight focus the DC proteasomal activity on exogenous proteins translocatedfrom ergosomes to the cytosol. Another mechanism that might be regulatedis the efficiency of formation of ergosomes or, as mentioned above, therecruitment of MHC I molecules or other components of the peptide-loadingmachinery to these compartments. Evidence for this, however, is still lacking.

A report from the Amigorena group has also shown that mature DCsdownregulate uptake and delivery of antigen to the cross-presentation pathway(Gil-Torregrosa et al., 2004). As described before for the MHC II pathway,transient upregulation of cross-presentation followed by its downregulationmay allow the mature DCs to focus their cross-presenting activity on antigenscaptured at the time of activation. But providing a ‘‘memory’’ of exogenousantigens via cross-presentation would require the DCs to extend the half-life ofthe MHC I–peptide complexes generated by this pathway (Ludewig et al.,2001). If maturing DCs loaded their MHC I molecules with exogenous anti-gens captured in a peripheral tissue but the complexes were rapidly degraded,the cross-presented antigens could be eliminated before or shortly after theDCs reached the lymphoid organs. However, the half-life of surface MHC I hasbeen reported to increase only by 2- to 3-fold (Ackerman and Cresswell, 2003),or not at all (Cella et al., 1997a; Delamarre et al., 2003; Wilson et al., 2004) inmature DCs. This is consistent with the increase in MHC I synthesis observedin mature DCs (Cella et al., 1997a; Rescigno et al., 1998; Villadangos et al.,2001; Wilson et al., 2004). Indeed, it makes sense for mature DCs to keep

control of antigen presentation in dendritic cells 287

presenting endogenous antigens via MHC I molecules to allow detection ofviruses that might infect the DCs themselves. So how can mature DCsmaintain a memory of their cross-presented antigens?

We can provide three hypotheses. The first is that the epitopes known to becross-presented may be particularly stable. It is known that the half-life andimmunogenicity of different MHC I–peptide complexes is determined by theirstability, which in turn depends on the affinity of the MHC I–peptide interac-tion (Chen et al., 1994; Sette et al., 1994). This affinity can vary broadly amongMHC I–peptide combinations. Perhaps only those peptides that interactwith MHC I molecules with high affinity can be cross-presented long enoughto induce cross-priming or cross-tolerance. If this were the case, it would notbe necessary to invoke a specific mechanism of enhancement of the half-life ofthese complexes in mature DCs. The second hypothesis is that a fraction ofthe ergosomes may be relatively stable and not fuse with lysosomes. Asdiscussed for MHC II presentation, DCs may deliver some antigens into‘‘storage’’ compartments with a lower proteolytic activity. Similarly, some ofthe ergosomes may ‘‘branch off’’ the main endocytic track and store some ofthe exogenous antigens for sustained cross-presentation. The third hypothesisis that the MHC I–peptide complexes generated via the cross-presentationpathway may be ‘‘tagged’’ in the ergosomes, so that once they are depositedon the cell surface, their endocytosis and turnover rates are much slowerthan those of their counterparts generated via the classic pathway. Suchtagging might consist of their association with membrane microdomains,posttranslational modifications such as phosphorylation, or the expression ofcarbohydrates not modified by transport through the Golgi complex.

8. Conclusions and Future Directions

Biochemical and cell biological studies have revealed the molecular basis ofsome of the unique antigen presentation capabilities shown by DCs. In general,components of the antigen presentation machinery expressed in DCs are notdifferent from those of other antigen-presenting cells. However, this machineryhas been adapted in DCs to fulfill two distinct functions linked to the matura-tional state of the DCs. In the steady state, immature DCs constitutively expresson their surface MHC class II molecules loaded with self-peptide ligands. Byvirtue of cross-presentation, some DC types can also cross-present on theirMHC I molecules peptides derived from proteins collected from the environ-ment (Heath et al., 2004). This allows immature DCs to provide naive T cellswith MHC–peptide ligands for maintenance of homeostasis and a constantsupply of information about the peripheral self for maintenance of tolerance.In contrast, DCs activated by the presence of pathogens or tissue damage

288 nicholas s. wilson and jose a. villadangos

undergo dramatic changes that result in their presentation of a snapshot of theirantigenic environment at the time and site of activation. This is accomplishedby interrupting the processes of MHC II presentation and MHC I cross-presentation once the activated DC has accumulated a substantial amount ofMHC–peptide complexes on its surface. As has been summarized here, thecheckpoints that control these developmental changes for the MHC II andthe MHC I pathways are different.

The studies carried out on the MHC II pathway indicate that the regulatorymechanisms operate at the level of MHC II synthesis and MHC II–peptideturnover. However, the signals that control the trafficking and turnover of thepeptide-loaded MHC II molecules remain unknown, as does the sortingmachinery involved in interpreting those signals. Is the turnover regulated atthe level of internalization from the plasma membrane or at the level ofrecycling from early endosomes back to the cell surface? Does the associationwith lipid rafts play a direct role in controlling MHC II trafficking? How doesthe formation of tubular tracks for the delivery of MHC II–peptide complexesfit with the dynamics of DC trafficking and maturation and with establishmentof the immunological synapse? The development of sophisticated experimentalsystems that allow direct visualization of the DC–T-cell interaction in thelymphoid organs may provide answers to some of these questions.

In the case of MHC I cross-presentation, the mechanisms of regulation aremuch less defined; these may involve the generation or ergosomes, the recruit-ment of proteasomes to the vicinity of these compartments, or the delivery ofthe components of the MHC I peptide-loading machinery to the ergosomes.The mechanisms involved in formation of the ergosomes themselves, and theminimum requirements that make them antigen presentation–competentcompartments need to be defined. Is Sec61 the channel used by exogenousantigens to access the cytosol? Are the peptides generated by the cross-presentation pathway identical to those generated in the direct pathway?Does the formation of MHC I–peptide complexes in the ergosome necessitateall the components of the peptide-loading machinery involved in the endo-plasmic reticulum? Answering these questions will provide us not only with abetter understanding of the functions played by DCs in tolerance and immu-nity, they may also point to novel targets for the improvement of DC-basedvaccines and other immunotherapeutic strategies.

Acknowledgments

We thank our colleague Bill Heath at the Walter and Eliza Hall Institute for critically reading themanuscript and for suggestion of the term ‘‘ergosome.’’ Nicholas S. Wilson is supported by aMelbourne University Research Scholarship and by a Student Project Grant from the Cooperative

control of antigen presentation in dendritic cells 289

Research Centre for Vaccine Technology. Jose A. Villadangos is funded by a Scholarship of theLeukemia and Lymphoma Society and grants of the Human Frontiers Science ProgramOrganization, the Anti-Cancer Council of Australia, and the Australian National Health andMedical Research Council.

References

Abrahamson, M. (1994). Cystatins. Methods Enzymol. 244, 685–700.Abrahamson, M., Olafsson, I., Palsdottir, A., Ulvsback, M., Lundwall, A., Jensson, O., and

Grubb, A. (1990). Structure and expression of the human cystatin C gene. Biochem. J. 268,287–294.

Ackerman, A. L., and Cresswell, P. (2003). Regulation of MHC class I transport in human dendriticcells and the dendritic-like cell line KG-1. J. Immunol. 170, 4178–4188.

Ackerman, A. L., Kyritsis, C., Tampe, R., and Cresswell, P. (2003). Early phagosomes in dendriticcells form a cellular compartment sufficient for cross presentation of exogenous antigens. Proc.Natl. Acad. Sci. USA 100, 12889–12894.

Albert, L. J., Ghumman, B., and Watts, T. H. (1996). Effect of HLA-DM transfection on hen egglysozyme presentation by T2.Ak cells. J. Immunol. 157, 2247–2255.

Albert, M. L., Jegathesan, M., and Darnell, R. B. (2001). Dendritic cell maturation is required forthe cross-tolerization of CD8þ T cells. Nat. Immunol. 2, 1010–1017.

Alfonso, C., and Karlsson, L. (2000). Nonclassical MHC class II molecules. Annu. Rev. Immunol.18, 113–142.

Amigorena, S., Drake, J. R., Webster, P., and Mellman, I. (1994). Transient accumulation of newclass II MHC molecules in a novel endocytic compartment in B lymphocytes. Nature 369, 113–120.

Amigorena, S., Webster, P., Drake, J., Newcomb, J., Cresswell, P., and Mellman, I. (1995).Invariant chain cleavage and peptide loading in major histocompatibility complex class IIvesicles. J. Exp. Med. 181, 1729–1741.

Anderson, H. A., Hiltbold, E. M., and Roche, P. A. (2000). Concentration of MHC class IImolecules in lipid rafts facilitates antigen presentation. Nat. Immunol. 1, 156–162.

Anjuere, F., Martin, P., Ferrero, I., Fraga, M. L., del Hoyo, G. M., Wright, N., and Ardavin, C.(1999). Definition of dendritic cell subpopulations present in the spleen, Peyer’s patches, lymphnodes, and skin of the mouse. Blood 93, 590–598.

Ardavin, C. (2003). Origin, precursors and differentiation of mouse dendritic cells. Nat. Rev.Immunol. 3, 582–590.

Arunachalam, B., Phan, U. T., Geuze, H. J., and Cresswell, P. (2000). Enzymatic reduction ofdisulfide bonds in lysosomes: Characterization of a g-interferon-inducible lysosomal thiol reduc-tase (GILT). Proc. Natl. Acad. Sci. USA 97, 745–750.

Asselin-Paturel, C., Boonstra, A., Dalod, M., Durand, I., Yessaad, N., Dezutter-Dambuyant, C.,Vicari, A., O’Garra, A., Biron, C., Briere, F., and Trinchieri, G. (2001). Mouse typeI IFN-producing cells are immature APCs with plasmacytoid morphology. Nat. Immunol. 2,1144–1150.

Avva, R. R., and Cresswell, P. (1994). In vivo and in vitro formation and dissociation of HLA-DRcomplexes with invariant chain-derived peptides. Immunity 1, 763–774.

Bakke, O., and Dobberstein, B. (1990). MHC class II-associated invariant chain contains a sortingsignal for endosomal compartments. Cell 63, 707–716.

Bakke, O., and Nordeng, T. W. (1999). Intracellular traffic to compartments for MHC class IIpeptide loading: Signals for endosomal and polarized sorting. Immunol. Rev. 172, 171–187.

290 nicholas s. wilson and jose a. villadangos

Banchereau, J., and Steinman, R. M. (1998). Dendritic cells and the control of immunity. Nature392, 245–252.

Barois, N., de Saint-Vis, B., Lebecque, S., Geuze, H. J., and Kleijmeer, M. J. (2002). MHC class IIcompartments in human dendritic cells undergo profound structural changes upon activation.Traffic 3, 894–905.

Baron, C., Raposo, G., Scholl, S. M., Bausinger, H., Tenza, D., Bohbot, A., Pouillart, P., Goud, B.,Hanau, D., and Salamero, J. (2004). Modulation of MHC class II transport and lysosomedistribution by macrophage-colony stimulating factor in human dendritic cells derived frommonocytes. J. Cell Sci. 114, 999–1010.

Belz, G. T., Behrens, G. M., Smith, C. M., Miller, J. F., Jones, C., Lejon, K., Fathman, C. G.,Mueller, S. N., Shortman, K., Carbone, F. R., and Heath, W. R. (2002). The CD8aþ dendriticcell is responsible for inducing peripheral self-tolerance to tissue-associated antigens. J. Exp.Med. 196, 1099–1104.

Benaroch, P., Yilla, M., Raposo, G., Ito, K., Miwa, K., Geuze, H. J., and Ploegh, H. L. (1995). HowMHC class II molecules reach the endocytic pathway. EMBO J. 14, 37–49.

Bendriss-Vermare, N., Barthelemy, C., Durand, I., Bruand, C., Dezutter-Dambuyant, C., Moulian,N., Berrih-Aknin, S., Caux, C., Trinchieri, G., and Briere, F. (2001). Human thymus containsIFN-a-producing CD11c�, myeloid CD11cþ, and mature interdigitating dendritic cells. J. Clin.Invest. 107, 835–844.

Bertho, N., Cerny, J., Kim, Y. M., Fiebiger, E., Ploegh, H., and Boes, M. (2003). Requirementsfor T cell-polarized tubulation of class IIþ compartments in dendritic cells. J. Immunol. 171,5689–5696.

Bevan, M. J. (1976). Cross-priming for a secondary cytotoxic response to minor H antigenswith H-2 congenic cells which do not cross-react in the cytotoxic assay. J. Exp. Med. 143,1283–1288.

Bikoff, E. K., Germain, R. N., and Robertson, E. J. (1995). Allelic differences affecting invariantchain dependency of MHC class II subunit assembly. Immunity 2, 301–310.

Bikoff, E. K., Wutz, G., Kenty, G. A., Koonce, C. H., and Robertson, E. J. (2001). Relaxed DMrequirements during class II peptide loading and CD4þ T-cell maturation in BALB/c mice.J. Immunol. 166, 5087–5098.

Bjorck, P. (2001). Isolation and characterization of plasmacytoid dendritic cells from Flt3ligand and granulocyte-macrophage colony-stimulating factor-treated mice. Blood 98,3520–3526.

Boes, M., Cerny, J., Massol, R., Op Den Brouw, M., Kirchhausen, T., Chen, J., and Ploegh, H. L.(2002). T-cell engagement of dendritic cells rapidly rearranges MHC class II transport. Nature418, 983–988.

Boes, M., Bertho, N., Cerny, J., Op Den Brouw, M., Kirchhausen, T., and Ploegh, H. (2003). T cellsinduce extended class II MHC compartments in dendritic cells in a Toll-like receptor-dependent manner. J. Immunol. 171, 4081–4088.

Boes, M., Cuvillier, A., and Ploegh, H. (2004). Membrane specializations and endosome matura-tion in dendritic cells and B cells. Trends Cell Biol. 14, 175–183.

Bonnerot, C., Marks, M. S., Cosson, P., Robertson, E. J., Bikoff, E. K., Germain, R. N., andBonifacino, J. S. (1994). Association with BiP and aggregation of class II MHC moleculessynthesized in the absence of invariant chain. EMBO J. 13, 934–944.

Bouneaud, C., Kourilsky, P., and Bousso, P. (2000). Impact of negative selection on the T-cellrepertoire reactive to a self-peptide: A large fraction of T-cell clones escapes clonal deletion.Immunity 13, 829–840.

control of antigen presentation in dendritic cells 291

Brachet, V., Raposo, G., Amigorena, S., and Mellman, I. (1997). Ii chain controls the transport ofmajor histocompatibility complex class II molecules to and from lysosomes. J. Cell Biol. 137,51–65.

Brand, C. U., Gerber, H. A., Hunziker, T., Schaffner, T., Limat, A., and Brathen, L. R. (1993).Phenotype of Langerhans cells in human afferent skin lymph derived from allergic contactdermatitis. Exp. Dermatol. 2, 274–279.

Brodsky, J. L., and McCracken, A. A. (1999). ER protein quality control and proteasome-mediatedprotein degradation. Semin. Cell Dev. Biol. 10, 507–513.

Brooks, A. G., Campbell, P. L., Reynolds, P., Gautam, A. M., and McCluskey, J. (1994). Antigenpresentation and assembly by mouse I-Ak class II molecules in human APC containing deletedor mutated HLA DM genes. J. Immunol. 153, 5382–5392.

Bryant, P. W., Lennon-Dumenil, A. M., Fiebiger, E., Lagaudriere-Gesbert, C., and Ploegh, H. L.(2002). Proteolysis and antigen presentation by MHC class II molecules. Adv. Immunol. 80,71–114.

Bujdoso, R., Hopkins, J., Dutia, B. M., Young, P., and McConnell, I. (1989). Characterizationof sheep afferent lymph dendritic cells and their role in antigen carriage. J. Exp. Med. 170,1285–1301.

Busch, R., Cloutier, I., Sekaly, R. P., and Hammerling, G. J. (1996). Invariant chain protects class IIhistocompatibility antigens from binding intact polypeptides in the endoplasmic reticulum.EMBO J. 15, 418–428.

Caminschi, I., Lucas, K. M., O’Keeffe, M. A., Hochrein, H., Laabi, Y., Brodnicki, T. C., Lew,A. M., Shortman, K., and Wright, M. D. (2001). Molecular cloning of a C-type lectinsuperfamily protein differentially expressed by CD8a� splenic dendritic cells. Mol.Immunol. 38, 365–373.

Campbell, D. J., Serwold, T., and Shastri, N. (2000). Bacterial proteins can be processed bymacrophages in a transporter associated with antigen processing-independent, cysteine pro-tease-dependent manner for presentation by MHC class I molecules. J. Immunol. 164, 168–175.

Castellino, F., and Germain, R. N. (1995). Extensive trafficking of MHC class II-invariant chaincomplexes in the endocytic pathway and appearance of peptide-loaded class II in multiplecompartments. Immunity 2, 73–88.

Cella, M., Dohring, C., Samaridis, J., Dessing, M., Brockhaus, M., Lanzavecchia, A., and Colonna,M. (1997a). A novel inhibitory receptor (ILT3) expressed on monocytes, macrophages, anddendritic cells involved in antigen processing. J. Exp. Med. 185, 1743–1751.

Cella, M., Engering, A., Pinet, V., Pieters, J., and Lanzavecchia, A. (1997b). Inflammatory stimuliinduce accumulation of MHC class II complexes on dendritic cells. Nature 388, 782–787.

Ceman, S., Wu, S., Jardetzky, T. S., and Sant, A. J. (1998). Alteration of a single hydrogen bondbetween class II molecules and peptide results in rapid degradation of class II molecules afterinvariant chain removal. J. Exp. Med. 188, 2139–2149.

Chang, C. C., Ciubotariu, R., Manavalan, J. S., Yuan, J., Colovai, A. I., Piazza, F., Lederman, S.,Colonna, M., Cortesini, R., Dalla-Favera, R., and Suciu-Foca, N. (2002). Tolerization of den-dritic cells by T(S) cells: The crucial role of inhibitory receptors ILT3 and ILT4. Nat. Immunol.3, 237–243.

Chapman, H. A., Riese, R. J., and Shi, G. P. (1997). Emerging roles for cysteine proteases in humanbiology. Annu. Rev. Physiol. 59, 63–88.

Chen, L., and Jondal, M. (2004). Alternative processing for MHC class I presentation by immatureand CpG-activated dendritic cells. Eur. J. Immunol. 34, 952–960.

Chen, W., Khilko, S., Fecondo, J., Margulies, D. H., and McCluskey, J. (1994). Determinantselection of major histocompatibility complex class I-restricted antigenic peptides is explained by

292 nicholas s. wilson and jose a. villadangos

class I-peptide affinity and is strongly influenced by nondominant anchor residues. J. Exp. Med.180, 1471–1483.

Chieppa, M., Bianchi, G., Doni, A., Del Prete, A., Sironi, M., Laskarin, G., Monti, P., Piemonti, L.,Biondi, A., Mantovani, A., et al. (2003). Cross-linking of the mannose receptor onmonocyte-derived dendritic cells activates an anti-inflammatory immunosuppressive program.J. Immunol. 171, 4552–4560.

Chow, A., Toomre, D., Garrett, W., and Mellman, I. (2002). Dendritic cell maturation triggersretrograde MHC class II transport from lysosomes to the plasma membrane. Nature 418, 988–994.

Colledge, L., Bennett, C. L., Reay, P. A., and Blackburn, C. C. (2002). Rapid constitutive gener-ation of a specific peptide–MHC class II complex from intact exogenous protein in immaturedendritic cells. Eur. J. Immunol. 32, 3246–3255.

Colonna, M., Krug, A., and Cella, M. (2002). Interferon-producing cells: On the front line inimmune responses against pathogens. Curr. Opin. Immunol. 14, 373–379.

Cresswell, P., Bangia, N., Dick, T., and Diedrich, G. (1999). The nature of the MHC class I peptideloading complex. Immunol. Rev. 172, 21–28.

Datta, S. K., Redecke, V., Prilliman, K. R., Takabayashi, K., Corr, M., Tallant, T., DiDonato, J.,Dziarski, R., Akira, S., Schoenberger, S. P., and Raz, E. (2003). A subset of Toll-like receptorligands induces cross-presentation by bone marrow-derived dendritic cells. J. Immunol. 170,4102–4110.

de Baey, A., and Lanzavecchia, A. (2000). The role of aquaporins in dendritic cell macropinocy-tosis. J. Exp. Med. 191, 743–748.

Delamarre, L., Holcombe, H., and Mellman, I. (2003). Presentation of exogenous antigens onmajor histocompatibility complex (MHC) class I and MHC class II molecules is differentiallyregulated during dendritic cell maturation. J. Exp. Med. 198, 111–122.

den Haan, J. M., and Bevan, M. J. (2002). Constitutive versus activation-dependent cross-presentation of immune complexes by CD8þ and CD8� dendritic cells in vivo. J. Exp. Med.196, 817–827.

Denzin, L. K., Hammond, C., and Cresswell, P. (1996). HLA-DM interactions with intermediatesin HLA-DR maturation and a role for HLA-DM in stabilizing empty HLA-DR molecules. J. Exp.Med. 184, 2153–2165.

Desjardins, M. (2003). ER-mediated phagocytosis: A new membrane for new functions. Nat. Rev.Immunol. 3, 280–291.

De Smedt, T., Pajak, B., Muraille, E., Lespagnard, L., Heinen, E., De Baetselier, P., Urbain, J.,Leo, O., and Moser, M. (1996). Regulation of dendritic cell numbers and maturation bylipopolysaccharide in vivo. J. Exp. Med. 184, 1413–1424.

d’Ostiani, C. F., Del Sero, G., Bacci, A., Montagnoli, C., Spreca, A., Mencacci, A., Ricciardi-Castagnoli, P., and Romani, L. (2000). Dendritic cells discriminate between yeasts and hyphae ofthe fungus Candida albicans: Implications for initiation of T helper cell immunity in vitro and invivo. J. Exp. Med. 191, 1661–1674.

Driessen, C., Bryant, R. A., Lennon-Dumenil, A. M., Villadangos, J. A., Bryant, P. W., Shi, G. P.,Chapman, H. A., and Ploegh, H. L. (1999). Cathepsin S controls the trafficking and maturationof MHC class II molecules in dendritic cells. J. Cell Biol. 147, 775–790.

Driessen, C., Lennon-Dumenil, A. M., and Ploegh, H. L. (2001). Individual cathepsins degradeimmune complexes internalized by antigen-presenting cells via Fc g receptors. Eur. J. Immunol.31, 1592–1601.

El-Sukkari, D., Wilson, N. S., Hakansson, K., Steptoe, R. J., Grubb, A., Shortman, K., andVilladangos, J. A. (2003). The protease inhibitor cystatin C is differentially expressedamong dendritic cell populations, but does not control antigen presentation. J. Immunol. 171,5003–5011.

control of antigen presentation in dendritic cells 293

Engel, A., Fujiyoshi, Y., and Agre, P. (2000). The importance of aquaporin water channel proteinstructures. EMBO J. 19, 800–806.

Engelhard, V. H. (1994). Structure of peptides associated with class I and class II MHC molecules.Annu. Rev. Immunol. 12, 181–207.

Fanger, N. A., Wardwell, K., Shen, L., Tedder, T. F., and Guyre, P. M. (1996). Type I (CD64) andtype II (CD32) Fc g receptor-mediated phagocytosis by human blood dendritic cells.J. Immunol. 157, 541–548.

Fiebiger, E., Meraner, P., Weber, E., Fang, I. F., Stingl, G., Ploegh, H., and Maurer, D. (2001).Cytokines regulate proteolysis in major histocompatibility complex class II-dependent antigenpresentation by dendritic cells. J. Exp. Med. 193, 881–892.

Fiebiger, E., Maehr, R., Villadangos, J., Weber, E., Erickson, A., Bikoff, E., Ploegh, H. L., andLennon-Dumenil, A. M. (2002). Invariant chain controls the activity of extracellular cathepsin L.J. Exp. Med. 196, 1263–1269.

Finkelman, F. D., Lees, A., Birnbaum, R., Gause, W. C., and Morris, S. C. (1996). Dendriticcells can present antigen in vivo in a tolerogenic or immunogenic fashion. J. Immunol. 157,1406–1414.

Fung-Leung, W. P., Surh, C. D., Liljedahl, M., Pang, J., Leturcq, D., Peterson, P. A., Webb, S. R.,and Karlsson, L. (1996). Antigen presentation and T-cell development in H2-M-deficient mice.Science 271, 1278–1281.

Garin, J., Diez, R., Kieffer, S., Dermine, J. F., Duclos, S., Gagnon, E., Sadoul, R., Rondeau, C., andDesjardins, M. (2001). The phagosome proteome: Insight into phagosome functions. J. Cell Biol.152, 165–180.

Garrett, W. S., Chen, L. M., Kroschewski, R., Ebersold, M., Turley, S., Trombetta, S., Galan, J. E.,and Mellman, I. (2000). Developmental control of endocytosis in dendritic cells by Cdc42. Cell102, 325–334.

Geissmann, F., Dieu-Nosjean, M. C., Dezutter, C., Valladeau, J., Kayal, S., Leborgne, M., Brousse,N., Saeland, S., and Davoust, J. (2002). Accumulation of immature Langerhans cells in humanlymph nodes draining chronically inflamed skin. J. Exp. Med. 196, 417–430.

Germain, R. N., and Hendrix, L. R. (1991). MHC class II structure, occupancy and surfaceexpression determined by post-endoplasmic reticulum antigen binding. Nature 353, 134–139.

Germain, R. N., and Rinker, A. G., Jr. (1993). Peptide binding inhibits protein aggregation ofinvariant-chain free class II dimers and promotes surface expression of occupied molecules.Nature 363, 725–728.

Geuze, H. J. (1998). The role of endosomes and lysosomes in MHC class II functioning. Immunol.Today 19, 282–287.

Gil-Torregrosa, B. C., Lennon-Dumenil, A. M., Kessler, B., Guermonprez, P., Ploegh, H. L., Fruci,D., van Endert, P., and Amigorena, S. (2004). Control of cross-presentation during dendritic cellmaturation. Eur. J. Immunol. 34, 398–407.

Gorvel, J. P., Escola, J. M., Stang, E., and Bakke, O. (1995). Invariant chain induces a delayedtransport from early to late endosomes. J. Biol. Chem. 270, 2741–2746.

Gregers, T. F., Nordeng, T. W., Birkeland, H. C., Sandlie, I., and Bakke, O. (2003). The cytoplasmictail of invariant chain modulates antigen processing and presentation. Eur. J. Immunol. 33, 277–286.

Gretz, J. E., Anderson, A. O., and Shaw, S. (1997). Cords, channels, corridors and conduits: Criticalarchitectural elements facilitating cell interactions in the lymph node cortex. Immunol. Rev. 156,11–24.

Gretz, J. E., Norbury, C. C., Anderson, A. O., Proudfoot, A. E., and Shaw, S. (2000). Lymph-bornechemokines and other low molecular weight molecules reach high endothelial venules via

294 nicholas s. wilson and jose a. villadangos

specialized conduits while a functional barrier limits access to the lymphocyte microenviron-ments in lymph node cortex. J. Exp. Med. 192, 1425–1440.

Gromme, M., Uytdehaag, F. G., Janssen, H., Calafat, J., van Binnendijk, R. S., Kenter, M. J., Tulp,A., Verwoerd, D., and Neefjes, J. (1999). Recycling MHC class I molecules and endosomalpeptide loading. Proc. Natl. Acad. Sci. USA 96, 10326–10331.

Gruenberg, J. (2001). The endocytic pathway: A mosaic of domains. Nat. Rev. Mol. Cell Biol. 2,721–730.

Guermonprez, P., Saveanu, L., Kleijmeer, M., Davoust, J., Van Endert, P., and Amigorena, S.(2003). ER–phagosome fusion defines an MHC class I cross-presentation compartment indendritic cells. Nature 425, 397–402.

Hamman, B. D., Chen, J. C., Johnson, E. E., and Johnson, A. E. (1997). The aqueous pore throughthe translocon has a diameter of 40–60 A during cotranslational protein translocation at the ERmembrane. Cell 89, 535–544.

Hanein, D., Matlack, K. E., Jungnickel, B., Plath, K., Kalies, K. U., Miller, K. R., Rapoport, T. A.,and Akey, C. W. (1996). Oligomeric rings of the Sec61p complex induced by ligands required forprotein translocation. Cell 87, 721–732.

Harshyne, L. A., Watkins, S. C., Gambotto, A., and Barratt-Boyes, S. M. (2001). Dendritic cellsacquire antigens from live cells for cross-presentation to CTL. J. Immunol. 166, 3717–3723.

Hart, D. N. (1997). Dendritic cells: Unique leukocyte populations which control the primaryimmune response. Blood 90, 3245–3287.

Hawiger, D., Inaba, K., Dorsett, Y., Guo, M., Mahnke, K., Rivera, M., Ravetch, J. V., Steinman,R. M., and Nussenzweig, M. C. (2001). Dendritic cells induce peripheral T-cell unresponsive-ness under steady state conditions in vivo. J. Exp. Med. 194, 769–779.

Heath, W. R., and Carbone, F. R. (2001). Cross-presentation, dendritic cells, tolerance andimmunity. Annu. Rev. Immunol. 19, 47–64.

Heath, W. R., Belz, G. T., Behrens, G. M. N., Smith, C. M., Forehan, S. P., Parish, I. A., Davey, G. M.,Wilson, N. S., Carbone, F. R., and Villadangos, J. A. (2004). Cross-presentation, dendritic cellsubsets, and the generation of immunity to cellular antigens. Immunol. Rev. 199, 9–26.

Henri, S., Vremec, D., Kamath, A., Waithman, J., Williams, S., Benoist, C., Burnham, K., Saeland,S., Handman, E., and Shortman, K. (2001). The dendritic cell populations of mouse lymphnodes. J. Immunol. 167, 741–748.

Henri, S., Curtis, J., Hochrein, H., Vremec, D., Shortman, K., and Handman, E. (2002). Hierarchyof susceptibility of dendritic cell subsets to infection by Leishmania major: Inverse relationshipto interleukin-12 production. Infect. Immun. 70, 3874–3880.

Honey, K., and Rudensky, A. Y. (2003). Lysosomal cysteine proteases regulate antigen presentation.Nat. Rev. Immunol. 3, 472–482.

Honey, K., Nakagawa, T., Peters, C., and Rudensky, A. (2002). Cathepsin L regulates CD4þ T-cellselection independently of its effect on invariant chain: A role in the generation of positivelyselecting peptide ligands. J. Exp. Med. 195, 1349–1358.

Houde, M., Bertholet, S., Gagnon, E., Brunet, S., Goyette, G., Laplante, A., Princiotta, M. F.,Thibault, P., Sacks, D., and Desjardins, M. (2003). Phagosomes are competent organelles forantigen cross-presentation. Nature 425, 402–406.

Howard, C. J., Sopp, P., Brownlie, J., Parsons, K. R., Kwong, L. S., and Collins, R. A. (1996).Afferent lymph veiled cells stimulate proliferative responses in allogeneic CD4þ and CD8þ

T cells but not gd TCRþ T cells. Immunology 88, 558–564.Huang, F. P., Platt, N., Wykes, M., Major, J. R., Powell, T. J., Jenkins, C. D., and MacPherson, G.G.

(2000). A discrete subpopulation of dendritic cells transports apoptotic intestinal epithelial cells toT-cell areas of mesenteric lymph nodes. J. Exp. Med. 191, 435–444.

control of antigen presentation in dendritic cells 295

Huh, C., Nagle, J. W., Kozak, C. A., Abrahamson, M., and Karlsson, S. (1995). Structural organ-ization, expression and chromosomal mapping of the mouse cystatin-C-encoding gene (Cst3).Gene 152, 221–226.

Huh, J. C., Strickland, D. H., Jahnsen, F. L., Turner, D. J., Thomas, J. A., Napoli, S., Tobagus, I.,Stumbles, P. A., Sly, P. D., and Holt, P. G. (2003). Bidirectional interactions between antigen-bearing respiratory tract dendritic cells (DCs) and T cells precede the late phase reaction inexperimental asthma: DC activation occurs in the airway mucosa but not in the lungparenchyma. J. Exp. Med. 198, 19–30.

Huppa, J. B., and Davis, M. M. (2003). T-cell-antigen recognition and the immunological synapse.Nat. Rev. Immunol. 3, 973–983.

Inaba, K., Inaba, M., Romani, N., Aya, H., Deguchi, M., Ikehara, S., Muramatsu, S., and Steinman,R. M. (1992). Generation of large numbers of dendritic cells from mouse bone marrowcultures supplemented with granulocyte/macrophage colony-stimulating factor. J. Exp. Med.176, 1693–1702.

Inaba, K., Witmer-Pack, M., Inaba, M., Hathcock, K. S., Sakuta, H., Azuma, M., Yagita, H.,Okumura, K., Linsley, P. S., Ikehara, S., et al. (1994). The tissue distribution of the B7-2costimulator in mice: Abundant expression on dendritic cells in situ and during maturationin vitro. J. Exp. Med. 180, 1849–1860.

Inaba, K., Turley, S., Yamaide, F., Iyoda, T., Mahnke, K., Inaba, M., Pack, M., Subklewe, M.,Sauter, B., Sheff, D., et al. (1998). Efficient presentation of phagocytosed cellular fragments onthe major histocompatibility complex class II products of dendritic cells. J. Exp. Med. 188,2163–2173.

Inaba, K., Turley, S., Iyoda, T., Yamaide, F., Shimoyama, S., Reis e Sousa, C., Germain, R. N.,Mellman, I., and Steinman, R. M. (2000). The formation of immunogenic major histocompati-bility complex class II–peptide ligands in lysosomal compartments of dendritic cells is regulatedby inflammatory stimuli. J. Exp. Med. 191, 927–936.

Itano, A. A., and Jenkins, M. K. (2003). Antigen presentation to naive CD4 T cells in the lymphnode. Nat. Immunol. 4, 733–739.

Itano, A. A., McSorley, S. J., Reinhardt, R. L., Ehst, B. D., Ingulli, E., Rudensky, A. Y., and Jenkins,M. K. (2003). Distinct dendritic cell populations sequentially present antigen to CD4 T cells andstimulate different aspects of cell-mediated immunity. Immunity 19, 47–57.

Iyoda, T., Shimoyama, S., Liu, K., Omatsu, Y., Akiyama, Y., Maeda, Y., Takahara, K., Steinman,R. M., and Inaba, K. (2002). The CD8þ dendritic cell subset selectively endocytoses dying cellsin culture and in vivo. J. Exp. Med. 195, 1289–1302.

Janeway, C. A., Jr. (1989). Approaching the asymptote? Evolution and revolution in immunology.Cold Spring Harb. Symp. Quant. Biol. 54, 1–13.

Jeannin, P., Renno, T., Goetsch, L., Miconnet, I., Aubry, J. P., Delneste, Y., Herbault, N., Baussant,T., Magistrelli, G., Soulas, C., et al. (2000). OmpA targets dendritic cells, induces their matura-tion and delivers antigen into the MHC class I presentation pathway. Nat. Immunol. 1, 502–509.

Jensen, P. E., Weber, D. A., Thayer, W. P., Westerman, L. E., and Dao, C. T. (1999). Peptideexchange in MHC molecules. Immunol. Rev. 172, 229–238.

Jiang, W., Swiggard, W. J., Heufler, C., Peng, M., Mirza, A., Steinman, R. M., and Nussenzweig,M. C. (1995). The receptor DEC-205 expressed by dendritic cells and thymic epithelial cells isinvolved in antigen processing. Nature 375, 151–155.

Kaldjian, E. P., Gretz, J. E., Anderson, A. O., Shi, Y., and Shaw, S. (2001). Spatial and molecularorganization of lymph node T-cell cortex: A labyrinthine cavity bounded by an epithelium-likemonolayer of fibroblastic reticular cells anchored to basement membrane-like extracellularmatrix. Int. Immunol. 13, 1243–1253.

296 nicholas s. wilson and jose a. villadangos

Kamath, A. T., Pooley, J., O’Keeffe, M. A., Vremec, D., Zhan, Y., Lew, A. M., D’Amico, A., Wu, L.,Tough, D. F., and Shortman, K. (2000). The development, maturation, and turnover rate ofmouse spleen dendritic cell populations. J. Immunol. 165, 6762–6770.

Kamath, A. T., Henri, S., Battye, F., Tough, D. F., and Shortman, K. (2002). Develop-mental kinetics and lifespan of dendritic cells in mouse lymphoid organs. Blood 100, 1734–1741.

Kampgen, E., Koch, N., Koch, F., Stoger, P., Heufler, C., Schuler, G., and Romani, N. (1991). ClassII major histocompatibility complex molecules of murine dendritic cells: Synthesis, sialylation ofinvariant chain, and antigen processing capacity are down-regulated upon culture. Proc. Natl.Acad. Sci. USA 88, 3014–3018.

Katz, J. F., and Sant, A. J. (1994). T-cell receptor recognition of MHC class II alloantigens is highlycell type dependent. J. Immunol. 152, 1629–1640.

Kim, D. T., Mitchell, D. J., Brockstedt, D. G., Fong, L., Nolan, G. P., Fathman, C. G., Engleman,E. G., and Rothbard, J. B. (1997). Introduction of soluble proteins into the MHC class I pathwayby conjugation to an HIV Tat peptide. J. Immunol. 159, 1666–1668.

Kirchhausen, T., Bonifacino, J. S., and Riezman, H. (1997). Linking cargo to vesicle formation:Receptor tail interactions with coat proteins. Curr. Opin. Cell Biol. 9, 488–495.

Kleijmeer, M., Ramm, G., Schuurhuis, D., Griffith, J., Rescigno, M., Ricciardi-Castagnoli, P.,Rudensky, A. Y., Ossendorp, F., Melief, C. J., Stoorvogel, W., and Geuze, H. J. (2001). Reorga-nization of multivesicular bodies regulates MHC class II antigen presentation by dendritic cells.J. Cell Biol. 155, 53–63.

Koch, N., Moldenhauer, G., Hofmann, W. J., and Moller, P. (1991). Rapid intracellular pathwaygives rise to cell surface expression of the MHC class II-associated invariant chain (CD74).J. Immunol. 147, 2643–2651.

Konecny, P., Stagg, A. J., Jebbari, H., English, N., Davidson, R. N., and Knight, S. C. (1999).Murine dendritic cells internalize Leishmania major promastigotes, produce IL-12 p40 andstimulate primary T-cell proliferation in vitro. Eur. J. Immunol. 29, 1803–1811.

Koonce, C. H., Wutz, G., Robertson, E. J., Vogt, A. B., Kropshofer, H., and Bikoff, E. K. (2003).DM loss in k haplotype mice reveals isotype-specific chaperone requirements. J. Immunol. 170,3751–3761.

Koppelman, B., Neefjes, J. J., de Vries, J. E., and de Waal Malefyt, R. (1997). Interleukin-10 down-regulates MHC class II ab peptide complexes at the plasma membrane of monocytes byaffecting arrival and recycling. Immunity 7, 861–871.

Kropshofer, H., Vogt, A. B., Moldenhauer, G., Hammer, J., Blum, J. S., and Hammerling,G. J. (1996). Editing of the HLA-DR-peptide repertoire by HLA-DM. EMBO J. 15,6144–6154.

Kropshofer, H., Arndt, S. O., Moldenhauer, G., Hammerling, G. J., and Vogt, A. B. (1997a). HLA-DM acts as a molecular chaperone and rescues empty HLA-DR molecules at lysosomal pH.Immunity 6, 293–302.

Kropshofer, H., Hammerling, G. J., and Vogt, A. B. (1997b). How HLA-DM edits the MHC classII peptide repertoire: Survival of the fittest? Immunol. Today 18, 77–82.

Kropshofer, H., Hammerling, G. J., and Vogt, A. B. (1999). The impact of the non-classical MHCproteins HLA-DM and HLA-DO on loading of MHC class II molecules. Immunol. Rev. 172,267–278.

Kropshofer, H., Spindeldreher, S., Rohn, T. A., Platania, N., Grygar, C., Daniel, N., Wolpl, A.,Langen, H., Horejsi, V., and Vogt, A. B. (2002). Tetraspan microdomains distinct from lipid raftsenrich select peptide–MHC class II complexes. Nat. Immunol. 3, 61–68.

Krug, A., Veeraswamy, R., Pekosz, A., Kanagawa, O., Unanue, E. R., Colonna, M., and Cella, M.(2003). Interferon-producing cells fail to induce proliferation of naive T cells but can promote

control of antigen presentation in dendritic cells 297

expansion and T helper 1 differentiation of antigen-experienced unpolarized T cells. J. Exp. Med.197, 899–906.

Lagaudriere-Gesbert, C., Newmyer, S. L., Gregers, T. F., Bakke, O., and Ploegh, H. L. (2002).Uncoating ATPase Hsc70 is recruited by invariant chain and controls the size of endocyticcompartments. Proc. Natl. Acad. Sci. USA 99, 1515–1520.

Larsen, C. P., Steinman, R. M., Witmer-Pack, M., Hankins, D. F., Morris, P. J., and Austyn, J. M.(1990). Migration and maturation of Langerhans cells in skin transplants and explants. J. Exp.Med. 172, 1483–1493.

Lautwein, A., Burster, T., Lennon-Dumenil, A. M., Overkleeft, H. S., Weber, E., Kalbacher, H.,and Driessen, C. (2002). Inflammatory stimuli recruit cathepsin activity to late endosomalcompartments in human dendritic cells. Eur. J. Immunol. 32, 3348–3357.

Lelouard, H., Gatti, E., Cappello, F., Gresser, O., Camosseto, V., and Pierre, P. (2002).Transient aggregation of ubiquitinated proteins during dendritic cell maturation. Nature417, 177–182.

Lennon-Dumenil, A. M., Roberts, R. A., Valentijn, K., Driessen, C., Overkleeft, H. S., Erickson, A.,Peters, P. J., Bikoff, E., Ploegh, H. L., and Wolf Bryant, P. (2001). The p41 isoform of invariantchain is a chaperone for cathepsin L. EMBO J. 20, 4055–4064.

Lennon-Dumenil, A. M., Bakker, A. H., Maehr, R., Fiebiger, E., Overkleeft, H. S., Rosemblatt, M.,Ploegh, H. L., and Lagaudriere-Gesbert, C. (2002). Analysis of protease activity in live antigen-presenting cells shows regulation of the phagosomal proteolytic contents during dendritic cellactivation. J. Exp. Med. 196, 529–540.

Linehan, S. A., Martinez-Pomares, L., Stahl, P. D., and Gordon, S. (1999). Mannose receptor andits putative ligands in normal murine lymphoid and nonlymphoid organs: In situ expression ofmannose receptor by selected macrophages, endothelial cells, perivascular microglia, andmesangial cells, but not dendritic cells. J. Exp. Med. 189, 1961–1972.

Lizee, G., Basha, G., Tiong, J., Julien, J. P., Tian, M., Biron, K. E., and Jefferies, W. A. (2003).Control of dendritic cell cross-presentation by the major histocompatibility complex class Icytoplasmic domain. Nat. Immunol. 4, 1065–1073.

Lotteau, V., Teyton, L., Peleraux, A., Nilsson, T., Karlsson, L., Schmid, S. L., Quaranta, V., andPeterson, P. A. (1990). Intracellular transport of class II MHC molecules directed by invariantchain. Nature 348, 600–605.

Ludewig, B., McCoy, K., Pericin, M., Ochsenbein, A. F., Dumrese, T., Odermatt, B., Toes, R. E.,Melief, C. J., Hengartner, H., and Zinkernagel, R. M. (2001). Rapid peptide turnover andinefficient presentation of exogenous antigen critically limit the activation of self-reactive CTLby dendritic cells. J. Immunol. 166, 3678–3687.

Luft, T., Jefford, M., Luetjens, P., Toy, T., Hochrein, H., Masterman, K. A., Maliszewski, C.,Shortman, K., Cebon, J., and Maraskovsky, E. (2002). Functionally distinct dendritic cell (DC)populations induced by physiologic stimuli: Prostaglandin E2 regulates the migratory capacity ofspecific DC subsets. Blood 100, 1362–1372.

Lukas, M., Stossel, H., Hefel, L., Imamura, S., Fritsch, P., Sepp, N. T., Schuler, G., and Romani, N.(1996). Human cutaneous dendritic cells migrate through dermal lymphatic vessels in a skinorgan culture model. J. Invest. Dermatol. 106, 1293–1299.

Luster, A. D., Weinshank, R. L., Feinman, R., and Ravetch, J. V. (1988). Molecular and bio-chemical characterization of a novel g-interferon-inducible protein. J. Biol. Chem. 263,12036–12043.

Lutz, M. B., Rovere, P., Kleijmeer, M. J., Rescigno, M., Assmann, C. U., Oorschot, V. M., Geuze,H. J., Trucy, J., Demandolx, D., Davoust, J., and Ricciardi-Castagnoli, P. (1997). Intracellularroutes and selective retention of antigens in mildly acidic cathepsin D/lysosome-associated

298 nicholas s. wilson and jose a. villadangos

membrane protein-1/MHC class II-positive vesicles in immature dendritic cells. J. Immunol.159, 3707–3716.

Macatonia, S. E., Knight, S. C., Edwards, A. J., Griffiths, S., and Fryer, P. (1987). Localization ofantigen on lymph node dendritic cells after exposure to the contact sensitizer fluoresceinisothiocyanate: Functional and morphological studies. J. Exp. Med. 166, 1654–1667.

Machy, P., Serre, K., Baillet, M., and Leserman, L. (2002). Induction of MHC class I presentationof exogenous antigen by dendritic cells is controlled by CD4þ T cells engaging class II moleculesin cholesterol-rich domains. J. Immunol. 168, 1172–1180.

Mahnke, K., Guo, M., Lee, S., Sepulveda, H., Swain, S. L., Nussenzweig, M., and Steinman, R. M.(2000). The dendritic cell receptor for endocytosis, DEC-205, can recycle and enhance antigenpresentation via major histocompatibility complex class II-positive lysosomal compartments.J. Cell Biol. 151, 673–684.

Mahnke, K., Knop, J., and Enk, A. H. (2003). Induction of tolerogenic DCs: ‘‘You are what youeat.’’ Trends Immunol. 24, 646–651.

Manickasingham, S. P., and Reis e Sousa, C. (2001). Mature T-cell seeks antigen for meaningfulrelationship in lymph node. Immunology 102, 381–386.

Manoury, B., Mazzeo, D., Li, D. N., Billson, J., Loak, K., Benaroch, P., and Watts, C. (2003).Asparagine endopeptidase can initiate the removal of the MHC class II invariant chain chaper-one. Immunity 18, 489–498.

Maric, M. A., Taylor, M. D., and Blum, J. S. (1994). Endosomal aspartic proteinases are requiredfor invariant-chain processing. Proc. Natl. Acad. Sci. USA 91, 2171–2175.

Martin, P., Del Hoyo, G. M., Anjuere, F., Arias, C. F., Vargas, H. H., Fernandez, L. A., Parrillas, V.,and Ardavin, C. (2002). Characterization of a new subpopulation of mouse CD8aþ B220þ

dendritic cells endowed with type 1 interferon production capacity and tolerogenic potential.Blood 100, 383–390.

Martin, W. D., Hicks, G. G., Mendiratta, S. K., Leva, H. I., Ruley, H. E., and Van Kaer, L. (1996).H2-M mutant mice are defective in the peptide loading of class II molecules, antigenpresentation, and T-cell repertoire selection. Cell 84, 543–550.

Maxfield, F. R., and McGraw, T. E. (2004). Endocytic recycling. Nat. Rev. Mol. Cell Biol. 5,121–132.

McGrath, M. E. (1999). The lysosomal cysteine proteases. Annu. Rev. Biophys. Biomol. Struct. 28,181–204.

McIlroy, D., Troadec, C., Grassi, F., Samri, A., Barrou, B., Autran, B., Debre, P., Feuillard, J.,and Hosmalin, A. (2001). Investigation of human spleen dendritic cell phenotype and distri-bution reveals evidence of in vivo activation in a subset of organ donors. Blood 97, 3470–3477.

Medzhitov, R. (2001). Toll-like receptors and innate immunity. Nat. Rev. Immunol. 1, 135–145.Melief, C. J. (2003). Mini-review: Regulation of cytotoxic T lymphocyte responses by dendritic

cells: Peaceful coexistence of cross-priming and direct priming? Eur. J. Immunol. 33,2645–2654.

Mellman, I. (1996). Endocytosis and molecular sorting. Annu. Rev. Cell Dev. Biol. 12, 575–625.Miyazaki, T., Wolf, P., Tourne, S., Waltzinger, C., Dierich, A., Barois, N., Ploegh, H., Benoist, C.,

and Mathis, D. (1996). Mice lacking H2–M complexes, enigmatic elements of the MHC class IIpeptide-loading pathway. Cell 84, 531–541.

Mommaas, A. M., Mulder, A. A., Jordens, R., Out, C., Tan, M. C., Cresswell, P., Kluin, P. M., andKoning, F. (1999). Human epidermal Langerhans cells lack functional mannose receptors and afully developed endosomal/lysosomal compartment for loading of HLA class II molecules. Eur.J. Immunol. 29, 571–580.

Morelli, A. E., Larregina, A. T., Shufesky, W. J., Zahorchak, A. F., Logar, A. J., Papworth, G. D.,Wang, Z., Watkins, S. C., Falo, L. D., Jr., and Thomson, A. W. (2003). Internalization

control of antigen presentation in dendritic cells 299

of circulating apoptotic cells by splenic marginal zone dendritic cells: Dependence oncomplement receptors and effect on cytokine production. Blood 101, 611–620.

Naik, S., Vremec, D., Wu, L., O’Keeffe, M., and Shortman, K. (2003). CD8aþ mouse spleendendritic cells do not originate from the CD8a� dendritic cell subset. Blood 102, 601–604.

Nakagawa, T. Y., Brissette, W. H., Lira, P. D., Griffiths, R. J., Petrushova, N., Stock, J., McNeish,J. D., Eastman, S. E., Howard, E. D., Clarke, S. R., et al. (1999). Impaired invariant chaindegradation and antigen presentation and diminished collagen-induced arthritis in cathepsinS null mice. Immunity 10, 207–217.

Nakano, H., Yanagita, M., and Gunn, M. D. (2001). CD11cþ B220þ Gr-1þ cells in mouse lymphnodes and spleen display characteristics of plasmacytoid dendritic cells. J. Exp. Med. 194, 1171–1178.

Neefjes, J. (1999). CIIV, MIIC and other compartments for MHC class II loading. Eur. J. Immunol.29, 1421–1425.

Neefjes, J. J., and Ploegh, H. L. (1992). Inhibition of endosomal proteolytic activity by leupeptinblocks surface expression of MHC class II molecules and their conversion to SDS resistance ab

heterodimers in endosomes. EMBO J. 11, 411–416.Nicchitta, C. V. (2003). Re-evaluating the role of heat-shock protein–peptide interactions in

tumour immunity. Nat. Rev. Immunol. 3, 427–432.Nolte, M. A., Belien, J. A., Schadee-Eestermans, I., Jansen, W., Unger, W. W., van Rooijen, N.,

Kraal, G., and Mebius, R. E. (2003). A conduit system distributes chemokines and small blood-borne molecules through the splenic white pulp. J. Exp. Med. 198, 505–512.

Norbury, C. C., Hewlett, L. J., Prescott, A. R., Shastri, N., and Watts, C. (1995). Class I MHCpresentation of exogenous soluble antigen via macropinocytosis in bone marrow macrophages.Immunity 3, 783–791.

Norbury, C. C., Chambers, B. J., Prescott, A. R., Ljunggren, H. G., and Watts, C. (1997).Constitutive macropinocytosis allows TAP-dependent major histocompatibility complex classI presentation of exogenous soluble antigen by bone marrow-derived dendritic cells. Eur.J. Immunol. 27, 280–288.

Nordeng, T. W., Gregers, T. F., Kongsvik, T. L., Meresse, S., Gorvel, J. P., Jourdan, F., Motta, A.,and Bakke, O. (2002). The cytoplasmic tail of invariant chain regulates endosome fusion andmorphology. Mol. Biol. Cell 13, 1846–1856.

O’Keeffe, M., Hochrein, H., Vremec, D., Caminschi, I., Miller, J. L., Anders, E. M., Wu, L.,Lahoud, M. H., Henri, S., Scott, B., et al. (2002a). Mouse plasmacytoid cells: Long-lived cells,heterogeneous in surface phenotype and function, that differentiate into CD8þ dendritic cellsonly after microbial stimulus. J. Exp. Med. 196, 1307–1319.

O’Keeffe, M., Hochrein, H., Vremec, D., Pooley, J., Evans, R., Woulfe, S., and Shortman, K.(2002b). Effects of administration of progenipoietin 1, Flt-3 ligand, granulocyte colony-stimulating factor, and pegylated granulocyte-macrophage colony-stimulating factor on dendriticcell subsets in mice. Blood 99, 2122–2130.

O’Keeffe, M., Hochrein, H., Vremec, D., Scott, B., Hertzog, P., Tatarczuch, L., and Shortman, K.(2003). Dendritic cell precursor populations of mouse blood: Identification of the murinehomologues of human blood plasmacytoid pre-DC2 and CD11cþ DC1 precursors. Blood 101,1453–1459.

Perry, J. A., Rush, A., Wilson, R. J., Olver, C. S., and Avery, A. C. (2004). Dendritic cells frommalaria-infected mice are fully functional APC. J. Immunol. 172, 475–482.

Peters, P. J., Neefjes, J. J., Oorschot, V., Ploegh, H. L., and Geuze, H. J. (1991). Segregation ofMHC class II molecules from MHC class I molecules in the Golgi complex for transport tolysosomal compartments. Nature 349, 669–676.

300 nicholas s. wilson and jose a. villadangos

Pierre, P., and Mellman, I. (1998). Developmental regulation of invariant chain proteolysis controlsMHC class II trafficking in mouse dendritic cells. Cell 93, 1135–1145.

Pierre, P., Turley, S. J., Gatti, E., Hull, M., Meltzer, J., Mirza, A., Inaba, K., Steinman, R. M., andMellman, I. (1997). Developmental regulation of MHC class II transport in mouse dendriticcells. Nature 388, 787–792.

Pieters, J., Bakke, O., and Dobberstein, B. (1993). The MHC class II-associated invariant chaincontains two endosomal targeting signals within its cytoplasmic tail. J. Cell Sci. 106, 831–846.

Pinet, V., Vergelli, M., Martin, R., Bakke, O., and Long, E. O. (1995). Antigen presentationmediated by recycling of surface HLA-DR molecules. Nature 375, 603–606.

Pooley, J. L., Heath, W. R., and Shortman, K. (2001). Cutting edge: Intravenous soluble antigen ispresented to CD4 T cells by CD8� dendritic cells, but cross-presented to CD8 T cells by CD8þ

dendritic cells. J. Immunol. 166, 5327–5330.Princiotta, M. F., Finzi, D., Qian, S. B., Gibbs, J., Schuchmann, S., Buttgereit, F., Bennink, J. R.,

and Yewdell, J. W. (2003). Quantitating protein synthesis, degradation, and endogenous antigenprocessing. Immunity 18, 343–354.

Pulendran, B., Lingappa, J., Kennedy, M. K., Smith, J., Teepe, M., Rudensky, A., Maliszewski,C. R., and Maraskovsky, E. (1997). Developmental pathways of dendritic cells in vivo: Distinctfunction, phenotype, and localization of dendritic cell subsets in FLT3 ligand-treated mice.J. Immunol. 159, 2222–2231.

Purcell, A. W. (2000). The peptide-loading complex and ligand selection during the assembly ofHLA class I molecules. Mol. Immunol. 37, 483–492.

Pure, E., Inaba, K., Crowley, M. T., Tardelli, L., Witmer-Pack, M. D., Ruberti, G., Fathman, G.,and Steinman, R. M. (1990). Antigen processing by epidermal Langerhans cells correlates withthe level of biosynthesis of major histocompatibility complex class II molecules and expression ofinvariant chain. J. Exp. Med. 172, 1459–1469.

Qiu, Y., Xu, X., Wandinger-Ness, A., Dalke, D. P., and Pierce, S. K. (1994). Separation ofsubcellular compartments containing distinct functional forms of MHC class II. J. Cell Biol.125, 595–605.

Ramachandra, L., and Harding, C. V. (2000). Phagosomes acquire nascent and recycling class IIMHC molecules but primarily use nascent molecules in phagocytic antigen processing.J. Immunol. 164, 5103–5112.

Rapoport, T. A., Jungnickel, B., and Kutay, U. (1996). Protein transport across the eukaryoticendoplasmic reticulum and bacterial inner membranes. Annu. Rev. Biochem. 65, 271–303.

Ravetch, J. V., and Bolland, S. (2001). IgG Fc receptors. Annu. Rev. Immunol. 19, 275–290.Regnault, A., Lankar, D., Lacabanne, V., Rodriguez, A., Thery, C., Rescigno, M., Saito, T., Verbeek,

S., Bonnerot, C., Ricciardi-Castagnoli, P., and Amigorena, S. (1999). Fcg receptor-mediatedinduction of dendritic cell maturation and major histocompatibility complex class I-restrictedantigen presentation after immune complex internalization. J. Exp. Med. 189, 371–380.

Reid, P. A., and Watts, C. (1990). Cycling of cell-surface MHC glycoproteins through primaquine-sensitive intracellular compartments. Nature 346, 655–657.

Reis e Sousa, C., and Germain, R. N. (1995). Major histocompatibility complex class I presentationof peptides derived from soluble exogenous antigen by a subset of cells engaged in phagocytosis.J. Exp. Med. 182, 841–851.

Reis e Sousa, C., and Germain, R. N. (1999). Analysis of adjuvant function by direct visualization ofantigen presentation in vivo: Endotoxin promotes accumulation of antigen-bearing dendriticcells in the T-cell areas of lymphoid tissue. J. Immunol. 162, 6552–6561.

Reis e Sousa, C., Stahl, P. D., and Austyn, J. M. (1993). Phagocytosis of antigens by Langerhanscells in vitro. J. Exp. Med. 178, 509–519.

control of antigen presentation in dendritic cells 301

Reits, E., Griekspoor, A., Neijssen, J., Groothuis, T., Jalink, K., van Veelen, P., Janssen, H., Calafat,J., Drijfhout, J. W., and Neefjes, J. (2003). Peptide diffusion, protection, and degradationin nuclear and cytoplasmic compartments before antigen presentation by MHC class. I.Immunity 18, 97–108.

Reits, E. A., Vos, J. C., Gromme, M., and Neefjes, J. (2000). The major substrates for TAP in vivoare derived from newly synthesized proteins. Nature 404, 774–778.

Rescigno, M. (2002). Dendritic cells and the complexity of microbial infection. Trends Microbiol.10, 425–461.

Rescigno, M., Citterio, S., Thery, C., Rittig, M., Medaglini, D., Pozzi, G., Amigorena, S., andRicciardi-Castagnoli, P. (1998). Bacteria-induced neo-biosynthesis, stabilization, and surfaceexpression of functional class I molecules in mouse dendritic cells. Proc. Natl. Acad. Sci. USA95, 5229–5234.

Riese, R. J., Shi, G. P., Villadangos, J., Stetson, D., Driessen, C., Lennon-Dumenil, A. M., Chu,C. L., Naumov, Y., Behar, S. M., Ploegh, H., et al. (2001). Regulation of CD1 function andNK1.1þ T-cell selection and maturation by cathepsin S. Immunity 15, 909–919.

Rock, K. L., York, I. A., Saric, T., and Goldberg, A. L. (2002). Protein degradation and thegeneration of MHC class I-presented peptides. Adv. Immunol. 80, 1–70.

Rodriguez, A., Regnault, A., Kleijmeer, M., Ricciardi-Castagnoli, P., and Amigorena, S. (1999).Selective transport of internalized antigens to the cytosol for MHC class I presentation indendritic cells. Nat. Cell Biol. 1, 362–368.

Romagnoli, P., and Germain, R. N. (1994). The CLIP region of invariant chain plays a critical rolein regulating major histocompatibility complex class II folding, transport, and peptide occupan-cy. J. Exp. Med. 180, 1107–1113.

Romani, N., Koide, S., Crowley, M., Witmer-Pack, M., Livingstone, A. M., Fathman, C. G., Inaba,K., and Steinman, R. M. (1989). Presentation of exogenous protein antigens by dendritic cells toT-cell clones. Intact protein is presented best by immature, epidermal Langerhans cells. J. Exp.Med. 169, 1169–1178.

Rovere, P., Zimmermann, V. S., Forquet, F., Demandolx, D., Trucy, J., Ricciardi-Castagnoli, P., andDavoust, J. (1998). Dendritic cell maturation and antigen presentation in the absence ofinvariant chain. Proc. Natl. Acad. Sci. USA 95, 1067–1072.

Ruedl, C., Koebel, P., Bachmann, M., Hess, M., and Karjalainen, K. (2000). Anatomical originof dendritic cells determines their life span in peripheral lymph nodes. J. Immunol. 165,4910–4916.

Ruedl, C., Koebel, P., and Karjalainen, K. (2001). In vivo-matured Langerhans cells continue totake up and process native proteins unlike in vitro-matured counterparts. J. Immunol. 166,7178–7182.

Sallusto, F., and Lanzavecchia, A. (1994). Efficient presentation of soluble antigen by culturedhuman dendritic cells is maintained by granulocyte/macrophage colony-stimulating factorplus interleukin 4 and downregulated by tumor necrosis factor a. J. Exp. Med. 179, 1109–1118.

Sallusto, F., Cella, M., Danieli, C., and Lanzavecchia, A. (1995). Dendritic cells use macropinocy-tosis and the mannose receptor to concentrate macromolecules in the major histocompatibilitycomplex class II compartment: Downregulation by cytokines and bacterial products. J. Exp.Med. 182, 389–400.

Salomon, B., Cohen, J. L., Masurier, C., and Klatzmann, D. (1998). Three populations ofmouse lymph node dendritic cells with different origins and dynamics. J. Immunol. 160, 708–717.

Sanderson, F., Thomas, C., Neefjes, J., and Trowsdale, J. (1996). Association between HLA-DMand HLA-DR in vivo. Immunity 4, 87–96.

302 nicholas s. wilson and jose a. villadangos

Santambrogio, L., Sato, A. K., Fischer, F. R., Dorf, M. E., and Stern, L. J. (1999). Abundant emptyclass II MHC molecules on the surface of immature dendritic cells. Proc. Natl. Acad. Sci. USA96, 15050–15055.

Saudrais, C., Spehner, D., de la Salle, H., Bohbot, A., Cazenave, J. P., Goud, B., Hanau, D., andSalamero, J. (1998). Intracellular pathway for the generation of functional MHC class II peptidecomplexes in immature human dendritic cells. J. Immunol. 160, 2597–2607.

Scheinecker, C., McHugh, R., Shevach, E. M., and Germain, R. N. (2002). Constitutive presenta-tion of a natural tissue autoantigen exclusively by dendritic cells in the draining lymph node.J. Exp. Med. 196, 1079–1090.

Schmitt, C., Fohrer, H., Beaudet, S., Palmer, P., Alpha, M. J., Canque, B., Gluckman, J. C., andDalloul, A. H. (2000). Identification of mature and immature human thymic dendriticcells that differentially express HLA-DR and interleukin-3 receptor in vivo. J. Leukoc. Biol.68, 836–844.

Schmitz, A., Herrgen, H., Winkeler, A., and Herzog, V. (2000). Cholera toxin is exported frommicrosomes by the Sec61p complex. J. Cell Biol. 148, 1203–1212.

Schubert, U., Anton, L. C., Gibbs, J., Norbury, C. C., Yewdell, J. W., and Bennink, J. R. (2000).Rapid degradation of a large fraction of newly synthesized proteins by proteasomes. Nature 404,770–774.

Schuler, G., and Steinman, R. M. (1985). Murine epidermal Langerhans cells mature into potentimmunostimulatory dendritic cells in vitro. J. Exp. Med. 161, 526–546.

Schuler, G., Schuler-Thurner, B., and Steinman, R. M. (2003). The use of dendritic cells in cancerimmunotherapy. Curr. Opin. Immunol. 15, 138–147.

Schulz, O., and Reis e Sousa, C. (2002). Cross-presentation of cell-associated antigens by CD8aþ

dendritic cells is attributable to their ability to internalize dead cells. Immunology 107, 183–189.Schulz, O., Edwards, A. D., Schito, M., Aliberti, J., Manickasingham, S., Sher, A., and Reis e Sousa,

C. (2000). CD40 triggering of heterodimeric IL-12 p70 production by dendritic cells in vivorequires a microbial priming signal. Immunity 13, 453–462.

Sercarz, E. E., and Maverakis, E. (2003). Mhc-guided processing: Binding of large antigenfragments. Nat. Rev. Immunol. 3, 621–629.

Sette, A., Vitiello, A., Reherman, B., Fowler, P., Nayersina, R., Kast, W. M., Melief, C. J., Oseroff,C., Yuan, L., Ruppert, J., et al. (1994). The relationship between class I binding affinity andimmunogenicity of potential cytotoxic T-cell epitopes. J. Immunol. 153, 5586–5592.

Sharpe, A. H., and Freeman, G. J. (2002). The B7-CD28 superfamily. Nat. Rev. Immunol. 2,116–126.

Shi, G. P., Villadangos, J. A., Dranoff, G., Small, C., Gu, L., Haley, K. J., Riese, R., Ploegh, H. L.,and Chapman, H. A. (1999). Cathepsin S required for normal MHC class II peptide loading andgerminal center development. Immunity 10, 197–206.

Shimada, S., Caughman, S. W., Sharrow, S. O., Stephany, D., and Katz, S. I. (1987). Enhancedantigen-presenting capacity of cultured Langerhans’ cells is associated with markedly increasedexpression of Ia antigen. J. Immunol. 139, 2551–2555.

Shortman, K., and Heath, W. R. (2001). Immunity or tolerance? That is the question for dendriticcells. Nat. Immunol. 2, 988–989.

Shortman, K., and Liu, Y. J. (2002). Mouse and human dendritic cell subtypes. Nat. Rev. Immunol.2, 151–161.

Song, R., and Harding, C. V. (1996). Roles of proteasomes, transporter for antigen presentation(TAP), and b2-microglobulin in the processing of bacterial or particulate antigens via analternate class I MHC processing pathway. J. Immunol. 156, 4182–4190.

Sprent, J., and Surh, C. D. (2003). Cytokines and T-cell homeostasis. Immunol. Lett. 85, 145–149.

control of antigen presentation in dendritic cells 303

Springer, T. A., Kaufman, J. F., Siddoway, L. A., Mann, D. L., and Strominger, J. L. (1977).Purification of HLA-linked B lymphocyte alloantigens in immunologically active form by pre-parative sodium dodecyl sulfate-gel electrophoresis and studies on their subunit association. J.Biol. Chem. 252, 6201–6207.

Srivastava, P. (2002). Roles of heat-shock proteins in innate and adaptive immunity. Nat. Rev.Immunol. 2, 185–194.

Starr, T. K., Jameson, S. C., and Hogquist, K. A. (2003). Positive and negative selection of T cells.Annu. Rev. Immunol. 21, 139–176.

Stebbins, C. C., Loss, G. E., Jr., Elias, C. G., Chervonsky, A., and Sant, A. J. (1995). Therequirement for DM in class II-restricted antigen presentation and SDS-stable dimer formationis allele and species dependent. J. Exp. Med. 181, 223–234.

Stebbins, C. C., Peterson, M. E., Suh, W. M., and Sant, A. J. (1996). DM-mediated release of anaturally occurring invariant chain degradation intermediate from MHC class II molecules.J. Immunol. 157, 4892–4898.

Steinman, R. M. (1991). The dendritic cell system and its role in immunogenicity. Annu. Rev.Immunol. 9, 271–296.

Steinman, R. M., and Nussenzweig, M. C. (2002). Avoiding horror autotoxicus: the importance ofdendritic cells in peripheral T-cell tolerance. Proc. Natl. Acad. Sci. USA 99, 351–358.

Steinman, R. M., and Swanson, J. (1995). The endocytic activity of dendritic cells. J. Exp. Med.182, 283–288.

Steinman, R. M., Turley, S., Mellman, I., and Inaba, K. (2000). The induction of tolerance bydendritic cells that have captured apoptotic cells. J. Exp. Med. 191, 411–416.

Steinman, R. M., Hawiger, D., and Nussenzweig, M. C. (2003). Tolerogenic dendritic cells. Annu.Rev. Immunol. 21, 685–711.

Stoitzner, P., Holzmann, S., McLellan, A. D., Ivarsson, L., Stossel, H., Kapp, M., Kammerer, U.,Douillard, P., Kampgen, E., Koch, F., et al. (2003). Visualization and characterization ofmigratory Langerhans cells in murine skin and lymph nodes by antibodies against Langerin/CD207. J. Invest. Dermatol. 120, 266–274.

Summers, K. L., Hock, B. D., McKenzie, J. L., and Hart, D. N. (2001). Phenotypic characteriza-tion of five dendritic cell subsets in human tonsils. Am. J. Pathol. 159, 285–295.

Svensson, M., Stockinger, B., and Wick, M. J. (1997). Bone marrow-derived dendritic cells canprocess bacteria for MHC-I and MHC-II presentation to T cells. J. Immunol. 158, 4229–4236.

Swanson, J. A., and Watts, C. (1995). Macropinocytosis. Trends Cell Biol. 5, 424–428.Takahashi, K., Asagoe, K., Zaishun, J., Yanai, H., Yoshino, T., Hayashi, K., and Akagi, T. (1998).

Heterogeneity of dendritic cells in human superficial lymph node: In vitro maturation ofimmature dendritic cells into mature or activated interdigitating reticulum cells. Am. J. Pathol.153, 745–755.

Trombetta, E. S., Ebersold, M., Garrett, W., Pypaert, M., and Mellman, I. (2003). Activation oflysosomal function during dendritic cell maturation. Science 299, 1400–1403.

Tulp, A., Verwoerd, D., Dobberstein, B., Ploegh, H. L., and Pieters, J. (1994). Isolation andcharacterization of the intracellular MHC class II compartment. Nature 369, 120–126.

Turley, S. J., Inaba, K., Garrett, W. S., Ebersold, M., Unternaehrer, J., Steinman, R. M., andMellman, I. (2000). Transport of peptide–MHC class II complexes in developing dendritic cells.Science 288, 522–527.

Valdez, Y., Mah, W., Winslow, M. M., Xu, L., Ling, P., and Townsend, S. E. (2002). Majorhistocompatibility complex class II presentation of cell-associated antigen is mediated byCD8aþ dendritic cells in vivo. J. Exp. Med. 195, 683–694.

Valladeau, J., Ravel, O., Dezutter-Dambuyant, C., Moore, K., Kleijmeer, M., Liu, Y., Duvert-Frances, V., Vincent, C., Schmitt, D., Davoust, J., et al. (2000). Langerin, a novel C-type lectin

304 nicholas s. wilson and jose a. villadangos

specific to Langerhans cells, is an endocytic receptor that induces the formation of Birbeckgranules. Immunity 12, 71–81.

Vandenabeele, S., Hochrein, H., Mavaddat, N., Winkel, K., and Shortman, K. (2001). Humanthymus contains 2 distinct dendritic cell populations. Blood 97, 1733–1741.

Van den Berg, B., Clemons, W. M., Jr., Collinson, I., Modis, Y., Hartmann, E., Harrison, S. C., andRapoport, T. A. (2004). X-ray structure of a protein-conducting channel. Nature 427, 36–44.

van Ham, S. M., Gruneberg, U., Malcherek, G., Broker, I., Melms, A., and Trowsdale, J. (1996).Human histocompatibility leukocyte antigen (HLA)-DM edits peptides presented by HLA-DRaccording to their ligand binding motifs. J. Exp. Med. 184, 2019–2024.

van Kooyk, Y., and Geijtenbeek, T. B. (2003). DC-SIGN: Escape mechanism for pathogens. Nat.Rev. Immunol. 3, 697–709.

van Lith, M., van Ham, M., Griekspoor, A., Tjin, E., Verwoerd, D., Calafat, J., Janssen, H., Reits, E.,Pastoors, L., and Neefjes, J. (2001). Regulation of MHC class II antigen presentation by sorting ofrecycling HLA-DM/DO and class II within the multivesicular body. J. Immunol. 167, 884–892.

van Weert, A. W., Dunn, K. W., Gueze, H. J., Maxfield, F. R., and Stoorvogel, W. (1995). Transportfrom late endosomes to lysosomes, but not sorting of integral membrane proteins in endosomes,depends on the vacuolar proton pump. J. Cell Biol. 130, 821–834.

Veeraswamy, R. K., Cella, M., Colonna, M., and Unanue, E. R. (2003). Dendritic cells process andpresent antigens across a range of maturation states. J. Immunol. 170, 5367–5372.

Vermaelen, K., and Pauwels, R. (2003). Accelerated airway dendritic cell maturation, trafficking,and elimination in a mouse model of asthma. Am. J. Respir. Cell Mol. Biol. 29, 405–409.

Vermaelen, K. Y., Carro-Muino, I., Lambrecht, B. N., and Pauwels, R. A. (2001). Specificmigratory dendritic cells rapidly transport antigen from the airways to the thoracic lymph nodes.J. Exp. Med. 193, 51–60.

Villadangos, J. A. (2001). Presentation of antigens by MHC class II molecules: Getting the most outof them. Mol. Immunol. 38, 329–346.

Villadangos, J. A., and Ploegh, H. L. (2000). Proteolysis in MHC class II antigen presentation:Who’s in charge? Immunity 12, 233–239.

Villadangos, J. A., Riese, R. J., Peters, C., Chapman, H. A., and Ploegh, H. L. (1997). Degradationof mouse invariant chain: Roles of cathepsins S and D and the influence of major histocompati-bility complex polymorphism. J. Exp. Med. 186, 549–560.

Villadangos, J. A., Driessen, C., Shi, G. P., Chapman, H. A., and Ploegh, H. L. (2000). Earlyendosomal maturation of MHC class II molecules independently of cysteine proteases andH-2DM. EMBO J. 19, 882–891.

Villadangos, J. A., Cardoso, M., Steptoe, R. J., van Berkel, D., Pooley, J., Carbone, F. R., andShortman, K. (2001). MHC class II expression is regulated in dendritic cells independently ofinvariant chain degradation. Immunity 14, 739–749.

Vremec, D., and Shortman, K. (1997). Dendritic cell subtypes in mouse lymphoid organs: Cross-correlation of surface markers, changes with incubation, and differences among thymus, spleen,and lymph nodes. J. Immunol. 159, 565–573.

Vremec, D., Pooley, J., Hochrein, H., Wu, L., and Shortman, K. (2000). CD4 and CD8 expressionby dendritic cell subtypes in mouse thymus and spleen. J. Immunol. 164, 2978–2986.

Watts, C. (1997). Capture and processing of exogenous antigens for presentation on MHCmolecules. Annu. Rev. Immunol. 15, 821–850.

West, M. A., Lucocq, J. M., and Watts, C. (1994). Antigen processing and class II MHC peptide-loading compartments in human B-lymphoblastoid cells. Nature 369, 147–151.

West, M. A., Prescott, A. R., Eskelinen, E. L., Ridley, A. J., and Watts, C. (2000). Rac is requiredfor constitutive macropinocytosis by dendritic cells but does not control its downregulation.Curr. Biol. 10, 839–848.

control of antigen presentation in dendritic cells 305

Wiertz, E. J., Tortorella, D., Bogyo, M., Yu, J., Mothes, W., Jones, T. R., Rapoport, T. A., andPloegh, H. L. (1996). Sec61-mediated transfer of a membrane protein from the endoplasmicreticulum to the proteasome for destruction. Nature 384, 432–438.

Wilson, N. S., and Villadangos, J. A. (2004). Lymphoid organ dendritic cells: Beyond theLangerhans cells paradigm. Immunol. Cell Biol. 82, 91–98.

Wilson, N. S., El-Sukkari, D., Belz, G. T., Smith, C. M., Steptoe, R. J., Heath, W. R., Shortman, K.,and Villadangos, J. A. (2003). Most lymphoid organ dendritic cell types are phenotypically andfunctionally immature. Blood 102, 2187–2194.

Wilson, N. S., El-Sukkari, D., and Villadangos, J. A. (2004). Dendritic cells constitutively presentself antigens in their immature state in vivo, and regulate antigen presentation by controlling therates of MHC class II synthesis and endocytosis. Blood 103, 2187–2195.

Winzler, C., Rovere, P., Rescigno, M., Granucci, F., Penna, G., Adorini, L., Zimmermann, V. S.,Davoust, J., and Ricciardi-Castagnoli, P. (1997). Maturation stages of mouse dendritic cells ingrowth factor-dependent long-term cultures. J. Exp. Med. 185, 317–328.

Witmer-Pack, M. D., Valinsky, J., Olivier, W., and Steinman, R. M. (1988). Quantitation of surfaceantigens on cultured murine epidermal Langerhans cells: Rapid and selective increase in thelevel of surface MHC products. J. Invest. Dermatol. 90, 387–394.

Wolf, P. R., and Ploegh, H. L. (1995). How MHC class II molecules acquire peptide cargo:Biosynthesis and trafficking through the endocytic pathway. Annu. Rev. Cell Dev. Biol. 11,267–306.

Wolf, P. R., Tourne, S., Miyazaki, T., Benoist, C., Mathis, D., and Ploegh, H. L. (1998). Thephenotype of H-2M-deficient mice is dependent on the MHC class II molecules expressed. Eur.J. Immunol. 28, 2605–2618.

Wolters, P. J., and Chapman, H. A. (2000). Importance of lysosomal cysteine proteases in lungdisease. Respir. Res. 1, 170–177.

Wraight, C. J., van Endert, P., Moller, P., Lipp, J., Ling, N. R., MacLennan, I. C., Koch, N., andMoldenhauer, G. (1990). Human major histocompatibility complex class II invariant chain isexpressed on the cell surface. J. Biol. Chem. 265, 5787–5792.

Yewdell, J. W., Anton, L. C., and Bennink, J. R. (1996). Defective ribosomal products (DRiPs):A major source of antigenic peptides for MHC class I molecules? J. Immunol. 157, 1823–1826.

Yewdell, J. W., Norbury, C. C., and Bennink, J. R. (1999). Mechanisms of exogenous antigenpresentation by MHC class I molecules in vitro and in vivo: Implications for generatingCD8þ T-cell responses to infectious agents, tumors, transplants, and vaccines. Adv. Immunol.73, 1–77.

Yewdell, J. W., Reits, E., and Neefjes, J. (2003). Making sense of mass destruction: QuantitatingMHC class I antigen presentation. Nat. Rev. Immunol. 3, 952–961.

Yilla, M., Tan, A., Ito, K., Miwa, K., and Ploegh, H. L. (1993). Involvement of the vacuolar Hþ-ATPases in the secretory pathway of HepG2 cells. J. Biol. Chem. 268, 19092–19100.

Zhong, G., Castellino, F., Romagnoli, P., and Germain, R. N. (1996). Evidence that binding siteoccupancy is necessary and sufficient for effective major histocompatibility complex (MHC)class II transport through the secretory pathway redefines the primary function of class II-asso-ciated invariant chain peptides (CLIP). J. Exp. Med. 184, 2061–2066.

Zhong, G., Sousa, C. R., and Germain, R. N. (1997). Antigen-unspecific B cells and lymphoiddendritic cells both show extensive surface expression of processed antigen–major histocompat-ibility complex class II complexes after soluble protein exposure in vivo or in vitro. J. Exp. Med.186, 673–682.

Zinkernagel, R. M. (2002). On cross-priming of MHC class I-specific CTL: Rule or exception?Eur. J. Immunol. 32, 2385–2392.

Index

AACE, see Angiotensin-converting enzymeActivation-induced cytidine deaminase

Apobec-1 homology and function, 79–80discovery and isolation, 76–77expression regulation, 77mutation studies, 85replication protein A interactions, 78single-stranded DNA substrates, 78S region targeting, 78substrate formation, 80–82

ADA deficiency, see Adenosine deaminasedeficiency

Adenosine deaminase deficiencyadenosine signaling and immune

development and function, 17–20CD26–enzyme interactions, 5–620-deoxyadenosine accumulation and toxicity

S-adenosylhomocysteine hydroxylaseinhibition, 12, 16–17

apoptosis derangements, 13–14lymphoid tissues, 12ribonucleotide reductase inhibition, 13

enzyme function, 2gene mutations, 3–4gene therapy prospects, 4–5, 35hairy cell leukemia treatment implications,

20–21knockout mouse models

adenosine receptor expression

eosinophils, 26–27mast cells, 26pulmonary signaling, 27

generation, 7immune system effects

B cell maturation effects, 11–12T cell development, 9–11, 34

307

neurological defects, 32–33phenotypes, 7, 9respiratory effects, 21, 32trophoblast cell essentiality and gene

delivery for fetal rescue, 8–9phenotypic and metabolic disturbances, 6–7pulmonary consequences of adenosine

elevationadenosine amplification model in lung

disease, 29–31enzyme therapy reversal of abnormalities,

23, 25inflammation, 23interleukin-13 mediation of lung injury,

28–29lung disease implications, 21–22, 27–29

severe combined immunodeficiency disease,2–4, 31

treatment, 4–5, 35S-Adenosylhomocysteine hydroxylase,

inhibition in adenosine deaminasedeficiency, 12, 16–17

AID, see Activation-induced cytidinedeaminase

Allergy, contact activation role, 191–192Alzheimer’s disease, contact activation pathway

in pathophysiology, 194–195b-Amyloid, bradykinin formation cascade

component interactions, 194–195Angiotensin-converting enzyme

blood pressure regulation, 192bradykinin inactivation, 184–185

Antigen presentation, see also Dendritic cellsystems, 251–252terminology, 251

Anti-phospholipid antibodies, fetal losssyndrome role, 152–153

308 INDEX

a1-Antitrypsin, contact activationinhibition, 183

Apobec-1, activation-induced cytidinedeaminase homology and function, 79–80

Apoptosis, adenosine deaminase deficiencyderangements, 13–14

Artemisclass switch recombination role, 90–91DNA double-strand break joining mediation

in V(D)J recombination, 60Asthma, adenosine signaling, 21–22, 31ATM, class switch recombination role, 91

BB cell

activation and class switch recombinationregulation, 72

adenosine deaminase deficiency effects onmaturation, 11–12

terminal deoxynucleotidyltransferase splicevariants in development, 120–121,123, 127–129

B cell receptor, V(D)J recombinationgenomic organization in mouse, 47–50overview, 44–46recombinase-activating genes, see

Recombinase-activating genesregulated accessibility of gene segments

chromatin modification, 67–68histone H2AX role, 68–69overview, 64–65transcriptional regulation, 65–67

Bcl-2, cytokine induction and natural killer cellsurvival promotion, 223–224

Bradykinincardiovascular disease and blood pressure

control implications, 193–194contact activation

allergy role, 191–192Alzheimer’s disease implications, 194–195blood cell binding of mediators, 173–175factor XII binding to human umbilical

vein endothelial cells, 169, 172high molecular weight kininogen binding

to human umbilical vein endothelialcells, 166–169

inflammatory disease derangements, 194inhibition, 181–184kinin cascade activation

cell surface binding, 175–177

factor XII-independent activation ofprekallikrein–high molecularweight kininogen complex,177–180

factor XII role, 180–181pathway, 161–166

inactivation, 184–185plasma pathway of generation, 160–161

CC1 inhibitor

contact activation inhibition, 181–183deficiency and disease pathogenesis,

188–191CD26, adenosine deaminase interactions, 5–6Chronic obstructive pulmonary disease,

adenosine signaling, 21–22Class switch recombination

diseases

hyper-IgM syndromes, 93–94X-linked hypohydrotic ectodermal

dysplasia, 95heavy chain constant region gene

organization, 71–72mechanism

activation-induced cytidine deaminaserole, see also Activation-inducedcytidine deaminase

cytidine deamination, 82DNA double-strand break induction,

85–86mutation studies, 85substrate formation, 80–82

Artemis role, 90–91ATM role, 91base excision repair, 82DNA deamination model, 82–84DNA double-strand break repair, 90–92DNA-PKcs role, 88–9153BP1 role, 91–92histone H2AX modifications, 88, 92Ku role, 90mismatch repair, 89, 93NBS1 role, 92–93promoter/enhancer interactions, 88S region

deletions, 86–87mutations, 87recombination and synapsis, 87–89transcription, 80–82

INDEX 309

overview, 44–46, 69–71regulation

B cell activation, 72germline heavy chain constant region

transcripts, 72–74regional specificity in regulation, 74–76S region sequence mutations, 76

V(D)J recombination comparison, 95–97CLIP peptide, major histocompatibility

complex II–peptide complex formation,260–261, 274

Clq receptorhigh molecular weight kininogen and factor

XII binding on human umbilical veinendothelial cells, 166–169, 172

human umbilical vein endothelial cellmembrane interactions, 172–173

ComplementC1 activation, 187–188fetal loss syndrome mediation, 151–153ischemia–reperfusion injury mediation

overview, 139renal ischemia–reperfusion injury,

147–151

COPD, see Chronic obstructive pulmonary

diseaseCrry, fetal loss syndrome role, 152–153CSR, see Class switch recombinationCytidine deaminase, see Activation-induced

cytidine deaminaseCytokeratin 1

high molecular weight kininogen binding onhuman umbilical vein endothelial cells,168–169

human umbilical vein endothelial cellmembrane interactions, 172–173

DDC, see Dendritic cellDendritic cell

antigen capture

macropinocytosis, 255overview, 252–253phagocytosis, 253–255receptor-mediated endocytosis,

255–256regulation, 256–257

antigen presentationdegradation of antigen, 259endocytic pathway, 258–259

major histocompatibility complexII–peptide complex formation,259–261

plasticity, 262–263subcellular localization of peptide binding,

261–262cross-presentation

antigen transfer to cytosol, 283–285compartmentalization, 282–283overview, 280–282peptide-loading machinery, 285regulation during maturation, 285–287

culture from precursors, 250–251immunity and tolerance interactions,

241–242interleukin-15 production and natural killer

cell modulation, 230major histocompatibility complex II antigen

presentation controlantigenic memory biological basis,

278–280endosomal proteolysis regulation, 265–268major histocompatibility complex

II molecule synthesis, 277–278major histocompatibility complex

II–peptide complex formationregulation

H-2DM regulation, 270–271Ii degradation, 268–270techniques for study, 271–273

models, 263–265trafficking control, 273–274, 276–277

maturation, 243prospects for study, 287–288subtypes

blood-derived cells, 247–250mouse versus human systems, 250overview, 243, 245tissue-derived cells, 245–247

DNA ligase 4, DNA double-strand breakjoining mediation in V(D)Jrecombination, 60

DNA mismatch repair, class switchrecombination, 89, 93

DNA-PKcsclass switch recombination role, 88–91DNA double-strand break joining mediation

in V(D)J recombination, 59–60terminal deoxynucleotidyltransferase

interactions, 126

310 INDEX

EElf-1, terminal deoxynucleotidyltransferase

expression regulation, 119Eosinophils, adenosine receptor expression,

26–27

FFactor XII

activation, 164–165, 180b-amyloid interactions, 194–195blood cell binding, 173–175blood pressure regulation, 192bradykinin formation role, 160, 162,

164–165human umbilical vein endothelial cell

binding, 169, 172inflammatory disease derangements, 194inhibition of contact activation, 181intrinsic fibrinolytic cascade, 185–187leukocyte interactions, 188plasma protease interactions, 187–188prekallikrein–high molecular weight

kininogen complex activation, 180–181Fetal loss syndrome, innate autoimmunity,

151–15353BP1, class switch recombination

role, 91–92Flt3 ligand, synergy with interleukin-15 in

natural killer cell development, 217

GGene therapy, adenosine deaminase deficiency,

4–5, 35

HH-2DM, major histocompatibility complex

II–peptide complex formation regulationin dendritic cells, 270–271

Heat shock protein 90, factor XII-independentactivation of prekallikrein–high molecularweight kininogen complex, 178–180

High molecular weight kininogen (HK)b-amyloid interactions, 194–195blood cell binding, 173–175bradykinin formation role, 160, 162–165domains, 162–163factor XII activation, 164–165human umbilical vein endothelial cell

binding, 166–169HK, see High molecular weight kininogen

Histone H2AXclass switch recombination modifications,

88, 92regulation in V(D)J recombination, 68–69

HK, see High molecular weight kininogenHSP90, see Heat shock protein 90Hyper-IgM syndromes

class switch recombination defects, 93–94

IIL-2, see Interleukin-2IL-13, see Interleukin-13IL-15, see Interleukin-15Innate immune system

components, 138autoimmune injury, see Fetal loss syndrome;

Ischemia–reperfusion injuryInterleukin-2

interleukin-15 comparison

cell distribution, 213–214cis versus trans signaling, 214–215signaling, 212–213in vitro versus in vivo similarities, 212

natural killer cell function and homeostasisrole

apoptosis, 225–226CD56bright cell subset

cytokine production, 228expansion with low-dose interleukin-2

therapy mechanisms, 221–222features, 220–221maturation in secondary lymphoid

organs with low-dose interleukin-2therapy, 228–229

rationale for expansion with low-doseinterleukin-2 therapy, 221

cytokine production modulation, 227–228cytolytic activity enhancement, 226–227development studies

cultured cell versus peripheral bloodcell studies, 217–218

historical perspective, 215–216mechanisms, 219–220

overview, 209–210, 231–232proliferation role, 224–225survival promotion mechanisms, 223–224

receptors, 121signal transduction cascades, 211–212

Interleukin-13, lung injury in transgenic mouseand adenosine role, 28–29

INDEX 311

Interleukin-15interleukin-12 comparison

cell distribution, 213–214cis versus trans signaling, 214–215receptor a-chain signaling, 212–213in vitro versus in vivo similarities, 212

natural killer cell function andhomeostasis role

apoptosis, 225–226cytokine production modulation, 227–228cytolytic activity enhancement,

226–227, 230development studies

cultured cell versus peripheral bloodcell studies, 217–218

endogenous cytokine pools, 218hematopoietic precursor differentiation

induction, 216–217mechanisms, 219synergy with c-Kit ligand and Flt3

ligand, 217immune cell sources for modulation, 230overview, 209–210, 231–232proliferation role, 224–225survival promotion

Bcl-2 independent mechanisms,223–224

Bcl-2 induction, 223cultured cell and animal studies,

222–223receptors, 121signal transduction cascades, 211–212

Ischemia–reperfusion injuryacute inflammatory response, 138–139antibody mediation

ischemia-related antigens, 145, 147knockout mouse studies, 140monoclonal immunoglobulin M initiation,

143–145specificity of natural immunoglobulin M,

140–141, 143complement mediation

overview, 139renal ischemia–reperfusion injury,

147–151

KKallikrein

bradykinin formation role,160–162

contact activation inhibition, 181factor XII-independent activation of

prekallikrein–high molecular weightkininogen complex, 177–180

inflammatory disease derangements, 194intrinsic fibrinolytic cascade,

185–187leukocyte interactions, 188prekallikrein complex, 162

Kidney, ischemia–reperfusion injury andcomplement mediation, 147–151

Kininogen, see High molecular weightkininogen

c-Kitligand synergy with interleukin-15 in natural

killer cell development, 217natural killer cell CD56bright subset

expression, 220Ku

class switch recombination role, 90DNA double-strand break joining mediation

in V(D)J recombination, 59terminal deoxynucleotidyltransferase

interactions, 126

LLeukemia

hairy cell leukemia treatment,20–21

terminal deoxynucleotidyltransferaseexpression and activity, 131–133

Ma2-Macroglobulin, contact activation

inhibition, 182–183Major histocompatibility complex class II

antigen presentation, see Dendritic cellMast cells, adenosine receptor

expression, 26

NNatural killer cell

interleukin-2 role in function andhomeostasis

apoptosis, 225–226CD56bright cell subset

cytokine production, 228expansion with low-dose

interleukin-2 therapymechanisms, 221–222

312 INDEX

Natural killer cell (continued)

features, 220–221maturation in secondary lymphoid

organs with low-doseinterleukin-2 therapy, 228–229

rationale for expansion with low-doseinterleukin-2 therapy, 221

cytokine production modulation,227–228

cytolytic activity enhancement, 226–227development studies

cultured cell versus peripheral bloodcell studies, 217–218

historical perspective, 215–216mechanisms, 219–220

overview, 209–210, 231–232proliferation role, 224–225survival promotion mechanisms,

223–224interleukin-15 role in function and

homeostasisapoptosis, 225–226cytokine production modulation,

227–228cytolytic activity enhancement,

226–227, 230development studies

cultured cell versus peripheral bloodcell studies, 217–218

endogenous cytokine pools, 218hematopoietic precursor differentiation

induction, 216–217mechanisms, 219synergy with c-Kit ligand and Flt3

ligand, 217immune cell sources for modulation, 230overview, 209–210, 231–232proliferation role, 224–225survival promotion

Bcl-2-independent mechanisms,223–224

Bcl-2 induction, 223cultured cell and animal studies,

222–223

NBS1, class switch recombination role,

92–93

PPCNA, see Proliferating cell

nuclear antigen

Proliferating cell nuclear antigen, terminaldeoxynucleotidyltransferaseinteractions, 127

RRAGs, see Recombinase-activating genesRecombinase-activating genes

DNA double-strand break joining mediatorsin V(D)J recombination

Artemis, 60DNA ligase 4, 60DNAPKcs, 59–60Ku, 59pathways, 58–59XRCC4, 60

expression regulation in V(D)Jrecombination

allelic exclusion and feedbackregulation, 61–64

downregulation effects, 64lymphoid-specific expression, 61

gene structure, 50–51genetic translocation induction, 51severe combined immunodeficiency

disease defects, 51V(D)J recombination initiation

cleavage reaction biochemistry, 54–56coding and signal joint formation,

57–58postcleavage complex, 56–57precleavage complex assembly,

53–54target sequence recognition, 51–53

Reperfusion injury, seeIschemia–reperfusion injury

Replication protein A, activation-inducedcytidine deaminase interactions, 78

Ribonucleotide reductase, inhibition inadenosine deaminase deficiency, 13

RPA, see Replication protein A

SSAH hydrolase, see S-Adenosylhomocysteine

hydroxylaseSCID, see Severe combined immunodeficiency

diseaseSevere combined immunodeficiency disease

adenosine deaminase deficiency, 2–4, 31recombinase-activating gene deficiency, 51X-linked form, 5

INDEX 313

SHM, see Somatic hypermutationSomatic hypermutation

mechanism

activation-induced cytidine deaminase

role, see also Activation-inducedcytidine deaminase

cytidine deamination, 82DNA double-strand break induction,

85–86mutation studies, 85substrate formation, 80–82

Artemis role, 90–91ATM role, 91base excision repair, 82DNA deamination model, 82–84DNA double-strand break repair,

90–92DNA-PKcs role, 88–9153BP1 role, 91–92histone H2AX modifications, 88, 92Ku role, 90mismatch repair, 89, 93NBS1 role, 92–93promoter/enhancer interactions, 88S region

deletions, 86–87mutations, 76, 87recombination and synapsis, 87–89transcription, 80–82

overview, 69–71regional specificity in regulation, 74–76V(D)J recombination comparison, 95–97

S region, see Class switch recombination;Somatic hypermutation

TT cell

adenosine deaminase deficiency effects ondevelopment, 9–11, 34

adenosine receptors, 19–20terminal deoxynucleotidyltransferase splice

variants in development, 120–121,123, 127–129

T cell receptor, V(D)J recombinationgenomic organization in mouse, 47–50overview, 44–46recombinase-activating genes, see

Recombinase-activating genesregulated accessibility of gene segments

chromatin modification, 67–68

histone H2AX role, 68–69overview, 64–65transcriptional regulation, 65–67

TdIF1, terminal deoxynucleotidyltransferaseinteractions, 127

TdT, see Terminal deoxynucleotidyltransferaseTerminal deoxynucleotidyltransferase

DNA polymerase family X polymerases, 115domains, 124–126leukemia expression and activity,

131–133phylogenetic analysis, 115, 117prospects for study, 133protein–protein interactions, 126–127splice variants

junctional diversity role in V(D)Jrecombination, 123–124

knockout and transgenic mouse studies ofrepertoire development, 127–130

lymphocyte development,120–121, 123

sequences, 122TdTL1, 119TdTL2, 119TdTS, 119

substrate specificity, 130tissue distribution, 115transcriptional regulation of expression,

117, 119

Uu-PAR, see Urokinase plasminogen activator

receptorUrokinase plasminogen activator receptor

high molecular weight kininogen and factorXII binding on human umbilical veinendothelial cells, 168–169, 172

human umbilical vein endothelial cellmembrane interactions, 172–173

VV(D)J recombination

antigen receptor gene rearrangement

genomic organization in mouse, 47–50recombinase-activating genes, see

Recombinase-activating genesregulated accessibility of gene segments

chromatin modification, 67–68histone H2AX role, 68–69overview, 64–65

314 INDEX

V(D)J recombination (continued)

transcriptional regulation, 65–67

class switch recombinationcomparison, 95–97

junctional diversity, 114–115,123–124

overview, 44–46, 114

XX-linked hypohydrotic ectodermal

dysplasia, class switchrecombination defects, 95

XRCC4, DNA double-strand breakjoining mediation in V(D)Jrecombination, 60

Contents of Recent Volumes

Volume 74Biochemical Basis of Antigen-Specific

Suppressor T Cell Factors: Controversiesand Possible AnswersKimishice Isihzaka, Yasuyuki Ishii,Tatsumi Nakano, and Katsuji Sugik

The Role of Complement in B Cell Activationand ToleranceMichael C. Carroll

Receptor Editing in B CellsDavid Nemazee

Chemokines and Their Receptors inLymphocyte Traffic and HIV InfectionPius Loetscher, Bernhard Moser, andMarco Bacciolini

Escape of Human Solid Tumors fromT-Cell Recognition: Molecular Mechanismsand Functional SignificanceFrancesco M. Marincola,Elizabeth M. Jaffee, Daniel J. Hicklin,and Soldano Ferrone

The Host Response to Leishmania InfectionWerner Solbacii and Tamas Laskay

Index

Volume 75Exploiting the Immune System: Toward New

Vaccines against Intracellular Bacteria

315

Jurgen Hess, Ulrich Schaible,Barbel Raupach, andStefan H. E. Kaufmann

The Cytoskeleton in Lymphocyte SignalingA. Bauch, F. W. Alt, G. R. Crabtree, andS. B. Snapper

TGF-� Signaling by Smad ProteinsKohei Miyazono, Peter ten Dijke, andCarl-Henrik Heldin

MHC Class II-Restricted Antigen Processingand PresentationJean Pieters

T-Cell Receptor Crossreactivity andAutoimmune DiseaseHarvey Cantor

Strategies for Immunotherapy of CancerCornelis J. M. Meliey,Rene E. M. Toes, Jan Paul Medema,Sjoerd H. van der Burg,Ferry Ossendorp, andRienk Offringa

Tyrosine Kinase Activation in the Decisionbetween Growth, Differentiation, andDeath Responses Initiated from the B CellAntigen ReceptorRobert C. Hsueh andRichard H. Scheuermann

The 30 IgH Regulatory Region:A Complex Structure in a Search fora Function

316 contents of recent volumes

Ahmed Amine Khamlichi,Eric Pinaud, Catherine Decourt,Christine Chauveau, andMichel Cogne

Index

Volume 76MIC Genes: From Genetics tok Biology

Seiamak Bahram

CD40-Mediated Regulation of ImmuneResponses by TRAF-Dependent andTRAF-Independent Signaling MechanismsAmrif C. Grammer andPeter E. Lipsky

Cell Death Control in LymphocytesKim Newton and Andreas Strassen

Systemic Lupus Erythematosus, ComplementDeficiency, and ApoptosisM. C. Pickering, M. Botto,P. R. Taylor, P. J. Lachmann,and M. J. Walport

Signal Transduction by the High-AffinityImmunoglobulin E Receptor FceRI:Coupling Form to FunctionMonica J. S. Nadler, Sharon A. Matthews,Helen Tuhner, and Jean-Pierre Kinet

Index

Volume 77The Actin Cytoskeleton, Membrane Lipid

Microdomains, and T Cell SignalTransductionS. Celeste Posey Morley andBarbara E. Bierer

Raft Membrane Domains andImmunoreceptor FunctionsThomas Harder

Human Basophils: Mediator Release andCytokine ProductionJohn T. Schroeder,Donald W. MacGlashan, Jr., andLawrence M. Lichtenstein

Btk and BLNK in B Cell DevelopmentSatoshi Tsukada, Yoshihiro Baba, andDai Watanabe

Diversity and Regulatory Functions ofMammalian Secretory Phospholipase A2sMakoto Murakami andIchiro Kudo

The Antiviral Activity of Antibodies in Vitroand in VivoPaul W. H. I. Parren andDennis R. Burton

Mouse Models of Allergic Airway DiseaseClare M. Lloyd,Jose-Angel Gonzalo,Anthony J. Coyle, andJose-Carlos Gutierrez-Ramos

Selected Comparison of Immune andNervous System DevelopmentJerold Chun

Index

Volume 78Toll-like Receptors and Innate Immunity

Shizuo Akira

Chemokines in ImmunityOsamu Yoshie, Toshio Imai, andHisayuki Nomiyama

Attractions and Migrations of LymphoidCells in the Organization of HumoralImmune ResponsesChristoph Schaniel,Antonius G. Rolink,and Fritz Melchers

contents of recent volumes 317

Factors and Forces ControllingV(D)J RecombinationDavid G. T. Hesslein andDavid G. Schatz

T Cell Effector Subsets: Extending theTh1/Th2 ParadigmTatyana Chtanova andCharles R. Mackay

MHC-Restricted T Cell Responses againstPosttranslationally Modified PeptideAntigensIngelise Bjerring Kastrup,Mads Hald Andersen, Tim Elliot, andJohn S. Haurum

Gastrointestinal Eosinophils in Healthand DiseaseMarc E. Rothenberg, Anil Mishra,Eric B. Brandt, and Simon P. Hogan

Index

Volume 79Neutralizing Antiviral Antibody Responses

Rolf M. Zinkernagel, Alain Lamarre,Adrian Ciurea, Lukas Hunziker,Adrian F. Ochsenbein, Kathy D. McCoy,Thomas Fehr, Martin F. Bachmann,Ulrich Kalinke, and Hans Hengartner

Regulation of Interleukin-12 Production inAntigen-Presenting CellsXiaojing Ma and Giorgio Trinchieri

Mechanisms of Signaling by theHematopoietic-Specific Adaptor Proteins,SLP-76 and LAT and Their B CellCounterpart, BLNK/SLP-65Deborah Yablonski andArthur Weiss

XenotransplantationDavid H. Sachs, Megan Sykes,Simon C. Robson, andDavid K. C. Cooper

Regulation of Antibacterial and AntifungalInnate Immunity in Fruitflies and HumansMichael J. Williams

Functional Heavy-Chain Antibodiesin CamelidaeViet Khong Nguyen,Aline Desmyter, andSerge Muyldermans

Uterine Natural Killer Cells in thePregnant UterusChau-Ching Liu andJohn Ding-E Young

Index

Volume 80Protein Degradation and the Generation of

MHC Class I-Presented PeptidesKenneth L. Rock, Ian A. York,Tomo Saric, and Alfred L. Goldberg

Proteoanalysis and Antigen Presentation byMHC Class II MoleculesPaula Wolf Bryant,Ana-Maria Lennon-Dumenil,Edda Fiebiger,Cecile Lagaudriere-Gesbert, andHidde L. Ploegh

Cytokine Memore of T Helper LymphocytesMax Lohning, Anne Richter, andAndreas Radbruch

Ig Gene Hypermutation: A Mechanism is DueJean-Claude Weill, Barbara Bertocci,Ahmad Faili, Said Aoufouchi,Stephane Frey, Annie De Smet,Sebastian Storck, Auriel Dahan,Frederic Delbos, Sandra Weller,Eric Flatter, and Claude-Agnes Reynaud

Generalization of Single ImmunologicalExperiences by Idiotypically MediatedClonal ConnectionsHilmar Lemke and Hans Lange

318 contents of recent volumes

The Aging of the Immune SystemB. Grubeck-Loebenstein and G. Wick

Index

Volume 81Regulation of the Immune Response by

the Interaction of Chemokines andProteasesJo Van Damme and Sofie Struyf

Molecular Mechanisms ofHost-Pathogen Interaction: Entryand Survival of Mycobacteria inMacrophagesJean Pieters and John Gatfield

B Lymphoid Neoplasms of Mice:Characteristics of Naturally Occurring andEngineered Diseasse and Relationships toHuman disordersHerbert Morse et al.

Prions and the Immune System:A Journey Through Gut Spleen,and NervesAdriano Aguzzi

Roles of the Semaphorin Family inImmune RegulationH. Kikutani and A. Kumanogoh

HLA-G Molecules: from Maternal-FetalTolerance to Tissue AcceptanceEdgardo Carosella et al.

The Zebrafish as a Model Organism to StudyDevelopment of the Immune SystemNick Trede et al.

Control of Autoimmunity by Naturally ArisingRegulatory CD4þ T CellsS. Sakaguchi

Index

Volume 82Transcriptional Regulation in Neutrophils:

Teaching Old Cells New TricksPatrick P. McDonald

Tumor VaccinesFreda K. Stevenson, Jason Rice, andDelin Zhu

Immunotherapy of Allergic DiseaseR. Valenta, T. Ball, M. Focke,B. Linhart, N. Mothes,V. Niederberger, S. Spitzauer,I. Swoboda, S.Vrtala, K. Westritschnic, andD. Kraft

Interactions of Immunoglobulins Outside theAntigen-Combining SiteRoald Nezlin and Victor Ghetie

The Roles of Antibodies in Mouse Models ofRheumatoid Arthritis, and Relevance toHuman DiseasePaul A. Monach, Christophe Benoist, andDiane Mathis

MUC1 Immunology: From Discovery toClinical ApplicationsAnda M. Vlad, Jessica C. Kettel,Nehad M. Alajez, Casey A. Carlos, andOlivera J. Finn

Human Models of Inherited ImmunoglobulinClass Switch Recombination and SomaticHypermutation Defects (Hyper-IgMSyndromes)Anne Durandy, Patrick Revy, andAlain Fischer

The Biological Role of the C1 Inhibitorin Regulation of Vascular Permeabilityand Modulation of InflammationAlvin E. Davis, III, Shenghe Cai,and Dongxu Liu

Index

contents of recent volumes 319

Volume 83Lineage Commitment and Developmental

Plasticity in Early LymphoidProgenitor SubsetsDavid Traver and Koichi Akashi

The CD4/CD8 Lineage Choice:New Insights into Epigenetic Regulationduring T Cell DevelopmentIchiro Taniuchi, Wilfried Ellmeier, andDan R. Littman

CD4/CD8 Coreceptors in ThymocyteDevelopment, Selection, and LineageCommitment: Analysis of the CD4/CD8Lineage DecisionAlfred Singer and Remy Bosselut

Development and Function ofT Helper 1 CellsAnne O’Garra andDouglas Robinson

Th2 Cells: Orchestrating Barrier ImmunityDaniel B. Stetson, David Voehringer,Jane L. Grogan, Min Xu, R. Lee Reinhardt,Stefanie Scheu, Ben L. Kelly, andRichard M. Locksley

Generation, Maintenance, and Function ofMemory T CellsPatrick R. Burkett, Rima Koka,Marcia Chien, David L. Boone, andAveril Ma

CD8þ Effector CellsPierre A. Henkart and Marta Catalfamo

An Integrated Model ofImmunoregulation Mediated byRegulatory T Cell SubsetsHong Jiang and Leonard Chess

Index

Volume 84Interactions Between NK Cells and

B LymphocytesDorothy Yuan

Multitasking of Helix-Loop-Helix Proteinsin LymphopoiesisXiao-Hong Sun

Customized Antigens for DesensitizingAllergic PatientsFatima Ferreira, Michael Wallner, andJosef Thalhamer

Immune Response Against DyingTumor CellsLaurence Zitvogel, Noelia Casares,Marie O. Pequignot, Nathalie Chaput,Mathew L. Albert,and Guido Kroemer

HMGB1 in the Immunology of Sepsis(Not Septic Shock) and ArthritisChristopher J. Czura, Huan Yang,Carol Ann Amella, and Kevin J. Tracey

Selection of the T-Cell Repertoire:Receptor-Controlled Checkpoints inT-Cell DevelopmentHarald Von Boehmer

The Pathogenesis of Diabetes in theNOD MouseMichelle Solomon and Nora Sarvetnick

Index

Volume 85Cumulative Subject Index Volumes 66–82