of June 4, 2018.This information is current as Rheumatoid Arthritis

Proteoglycan-Induced Murine Model of Severe Joint Inflammation in theInnate Immune System Is Required for Expression of CD44 and L-Selectin in the

Finnegan and Katalin MikeczBara Sarraj, Katalin Ludányi, Tibor T. Glant, Alison

http://www.jimmunol.org/content/177/3/1932doi: 10.4049/jimmunol.177.3.1932

2006; 177:1932-1940; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/177/3/1932.full#ref-list-1

, 18 of which you can access for free at: cites 45 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2006 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on June 4, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Expression of CD44 and L-Selectin in the Innate ImmuneSystem Is Required for Severe Joint Inflammation in theProteoglycan-Induced Murine Model of Rheumatoid Arthritis1

Bara Sarraj,* Katalin Ludanyi,† Tibor T. Glant,† Alison Finnegan,*‡ and Katalin Mikecz2†

Proteoglycan (PG)-induced arthritis, a murine model of rheumatoid arthritis, is characterized by autoimmunity against mousecartilage PG and chronic joint inflammation. L-selectin (CD62L) and CD44 are major adhesion molecules on leukocytes thatregulate their homing to lymph nodes and entry into inflamed tissues. In the present study, we studied the requirement for CD44and CD62L expression for mediating lymphocyte homing, thus permitting the development of autoimmunity vs mediating theentry of leukocytes into the joints, thus allowing inflammation in PG-induced arthritis. We immunized wild-type, CD44 knockout(KO), CD62L KO, and double (CD44/CD62L) KO BALB/c mice with PG and monitored the effects of gene deficiencies onPG-specific immunity, arthritis severity, leukocyte trafficking, and the ability of lymphocytes to adoptively transfer disease tosyngeneic SCID mice. Single and double KO mice demonstrated reduced PG-specific spleen cell proliferation, but the productionof Th cytokines and autoantibodies was comparable in KO and wild-type mice. KO leukocytes had reduced ability to adheretightly to the synovial endothelium in arthritic joints. This diminished leukocyte adhesion correlated with the magnitude ofgranulocyte (neutrophil) influx and the severity of inflammation, which were both reduced in the joints of KO mice. However,transfer of spleen cells from mildly arthritic KO donors to SCID hosts resulted in development of severe arthritis. Our resultsindicate that CD44 and CD62L expression in the cells of the innate immune system (granulocytes) is important for their efficientinflux into the joints and also suggest that granulocytes play a crucial role in arthritis progression. The Journal of Immunology,2006, 177: 1932–1940.

R heumatoid arthritis (RA)3 is a human autoimmune dis-ease characterized by chronic inflammation of multiplejoints, ultimately leading to cartilage and bone erosion

and loss of joint function (1, 2). Proteoglycan (PG)-induced ar-thritis (PGIA) in BALB/c mice bears strong resemblance to RA inclinical and histopathological features, immunopathology, as wellas genetic susceptibility and inheritance (3–5). PGIA develops ingenetically susceptible strains (BALB/c or C3H) of mice upon i.p.immunization with human cartilage proteoglycan (hPG) aggrecanin adjuvant (3, 6). A Th1-polarized response to hPG in the immu-nized animals is followed by recognition of self-mouse PG (mPG),and the development of polyarthritis is preceded by, or coincideswith, the appearance of mPG-specific autoantibodies in the circu-lation (4, 5, 7, 8). Locally, PGIA is characterized by influx ofleukocytes into the synovial tissue and joint cavity, synovial liningcell proliferation, and invasion of cartilage and bone by synovialpannus (3, 5).

Sustained migration of leukocytes from the blood into the sy-novial tissue and cavity of multiple joints is a hallmark of thechronic inflammatory process in both PGIA and RA. Cells of boththe adaptive and innate immune system are recruited to the joints;T and B cells are more frequent in the synovial tissue than in thejoint cavity, whereas polymorphonuclear granulocytes (neutro-phils) constitute the major cell population in the synovial fluid (2,3, 5, 9–11). To gain entry into a target tissue, leukocytes imple-ment a sequence of adhesive interactions with the vascular endo-thelium, collectively known as the “multistep paradigm” of ex-travasation (12). Leukocytes first transiently tether to, then roll on,the endothelium under blood flow. Next, chemokine-triggered ac-tivation of leukocyte integrins leads to firm adhesion of these cellsto the endothelium so that they resist detachment from endothelialcells (13, 14). Finally, leukocytes migrate across the vessel wallinto the tissue (12).

L-selectin (CD62L), the lymphocyte homing receptor (15), andCD44, the hyaluronan receptor (16), are two of the major adhesionmolecules expressed by leukocytes of both lymphoid and myeloidorigin (17). In addition to its established role in the homing oflymphocytes to lymph nodes (LNs), CD62L has been shown tomediate leukocyte rolling on inflamed endothelium (18–20).CD44 has been shown to function primarily by supporting leuko-cyte rolling at inflammatory sites (21, 22), but recent observationsin CD44 gene knockout (KO) mice suggest that loss of CD44might facilitate lymphocyte homing to peripheral LNs during in-flammation (23, 24).

In a dermal delayed-type hypersensitivity model, it was shownthat mice deficient in CD62L could not mount a response to epi-cutaneous immunization due to the inability of naive lymphocytesto home to the skin-draining LNs where they could acquire mem-ory and effector functions (18, 25). Expression of CD62L was alsorequired for the prompt initiation of leukocyte migration into the

*Department of Immunology and Microbiology, †Department of Orthopedic Surgery,and ‡Department of Internal Medicine, Section of Rheumatology, Rush UniversityMedical Center, Chicago, IL 60612

Received for publication February 21, 2006. Accepted for publication May 18, 2006.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported in part by National Institutes of Health Grants RA45652and RA051163.2 Address correspondence and reprint requests to Dr. Katalin Mikecz, Department ofOrthopedic Surgery, Rush University Medical Center, Cohn Research Building,Room 712, 1735 West Harrison Street, Chicago, IL 60612. E-mail address:Katalin_Mikecz@rsh.net3 Abbreviations used in this paper: RA, rheumatoid arthritis; CD62L, L-selectin; PG,proteoglycan (aggrecan); hPG, human cartilage PG; IVM, intravital video micros-copy; KO, knockout; LN, lymph node; mPG, mouse PG; PGIA, PG-induced arthritis;PSGL-1, P-selectin glycoprotein ligand-1; WT, wild type.

The Journal of Immunology

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

knee joint in Ag-induced arthritis, a nonautoimmune, monoarticu-lar form of joint inflammation (26). In experimental allergic en-cephalomyelitis, transfer of wild-type (WT) macrophages toCD62L-deficient mice was necessary to cause myelin damage inthe CNS (27). Previous work from our laboratory demonstratedthat DBA/1 mice deficient in CD44 developed collagen-inducedarthritis at reduced incidence and severity despite normal autoim-mune responses to type II collagen (28). In contrast to CD62L KOlymphocytes that migrate poorly to LNs (18, 25), lymphocytesfrom CD44-deficient mice showed an increased propensity tohome to LNs following immunization with type II collagen, bac-terial superantigen (23), or injection with conalbumin (24). Therole of CD62L in autoimmune arthritis has not been investigated.Because of the intricate regulatory pathways involved in the gen-eration of autoimmunity to a joint (cartilage) component, it is ex-pected that induction of a chronic autoimmune form of arthritis,such as PGIA, would be difficult in CD62L KO mice. We hypoth-esized that the LN homing defect (18) could be corrected, at leastin part, by introducing CD44 deficiency into CD62L-null mice(i.e., creation of a double KO). “Restoration” of the adaptive im-mune system could then allow for induction of an autoimmuneresponse to PG in CD44/CD62L double KO mice. Dual adhesionreceptor deficiency is, on the other hand, expected to manifest inimpaired leukocyte recruitment at the periphery (19, 21, 26), thusresulting in reduced inflammation and joint damage.

To dissect the requirement for CD44 and CD62L for the auto-immune vs the inflammatory response in PGIA, we generated dou-ble CD44/CD62L-deficient mice in the PGIA-susceptible BALB/cbackground and compared their Ag-specific T cell responses, au-toantibodies titers, and arthritis development to those in WT,CD44 KO, and CD62L KO mice upon immunization with hPG.Using intravital video microscopy (IVM), we also compared therecruitment of leukocytes to the ankle joints in WT and single anddouble KO mice and examined the relative involvement of leuko-cyte subpopulations in joint inflammation. In addition, we testedthe ability of single and double KO lymphocytes to migrate to thesecondary lymphoid organs of SCID mice and to adoptively trans-fer PGIA. We found that inflammatory, but not autoimmune, re-sponses were compromised by the lack of CD44 and CD62L andthat expression of these molecules in granulocytes was required forthe development of severe arthritis.

Materials and MethodsMice

WT BALB/c mice were purchased from the National Cancer Institute(NCI). Mice deficient in CD44 (in the DBA/1LacJ background) were gen-erated by targeted gene disruption as described previously (28). CD62L-null mice (in the C57BL/6 background) (29) were purchased from TheJackson Laboratory. BALB/c mice lacking CD44 or CD62L were obtainedby backcrossing into the BALB/c background, using a genomic marker-assisted speed congenic method (5, 30), where heterozygous KO maleswith the most complete BALB/c genetic makeup were mated with WTBALB/c females. Using this selection step before each backcross, fullBALB/c background was attained faster (7–9 backcrosses) than with ran-dom cross-breeding (which usually requires �12 backcrosses). FullBALB/c background was confirmed by whole genome screening with 244simple sequence-length polymorphic markers (30). Finally, CD44�/�,CD62L�/�, and CD44�/�/CD62L�/� heterozygous males and females ofpure BALB/c background were intercrossed, and offspring, homozygousfor single and double gene mutations, were identified by genomic PCR(26). The phenotype of adult mice (cell surface CD44 and CD62L expres-sion or lack thereof) was also confirmed by flow cytometry. Female SCIDmice of BALB/c background (NCI/NCrC.B-17-scid/scid), used for celltransfer experiments, were obtained from the NCI and were maintainedunder germfree conditions. All animal procedures were conducted under aprotocol approved by the Institutional Animal Care and Use Committee ofRush University Medical Center.

Induction of primary PGIA and adoptive transfer to SCID mice

hPG was isolated from osteoarthritic cartilage, which was obtained frompatients undergoing joint replacement surgery, and was provided throughthe Orthopedic Tissue, Transplant, and Implant Repository of Rush Uni-versity Medical Center, with the approval of the Institutional ReviewBoard. PG was extracted from pooled cartilage samples as described in ourprevious publications (3–6). WT, single KO, and double KO femaleBALB/c mice at 10–16 wk of age were immunized i.p. with 100 �g of hPG(measured as protein) in dimethyldioctadecyl ammonium bromide adjuvantfour times at 3-wk intervals, according to an established protocol (5, 6).Adoptive transfer of PGIA to SCID mice was performed as described pre-viously (31). In brief, each SCID mouse was injected i.p. with 5 � 107

spleen cells from donors with primary PGIA, along with 200 �g of hPG.Two weeks later, each mouse received 2.5 � 107 donor spleen cells and100 �g of hPG. Animals were inspected for symptoms of arthritis after thethird immunization and the first cell transfer for primary and adoptivePGIA, respectively. According to the presence of redness (erythema) anddegree of swelling, the paws were visually scored on a scale of 0–4 perpaw (0–16 per mouse). The severity of arthritis in each paw was deter-mined as follows: 0, no swelling or redness; 1, mild swelling and erythema;2, moderate swelling; 3, severe swelling; and 4, severe swelling with hard-ening of the periarticular tissue (5, 6). The same investigator inspected andscored the paws twice a week in a blinded manner.

Cell isolation, flow cytometry, and histology

Blood was collected in heparin-containing tubes from anesthetized micethrough retro-orbital puncture. At sacrifice (wk 0, 8, or 11 of immuniza-tion), the spleen and joint-draining (axillary, brachial, inguinal, and pop-liteal) LNs were harvested, and single-cell suspensions were prepared.RBC in the spleen and blood samples were eliminated by hypotonic lysis.Cell counts and viability were determined by trypan blue exclusion. Anklesynovial fluid cells were collected after repeated flushing of the open jointcavity with PBS, followed by gentle scraping of the synovial intima witha pipet tip. Cells in LN, spleen, blood, or joint fluid samples were immu-nostained with fluorescence-conjugated mAbs against adhesion molecules,including CD44, CD62L, and P-selectin glycoprotein ligand 1 (PSGL-1),as well as to leukocyte subpopulation markers such as CD3, CD45R/B220,Gr-1 (granulocytes) (BD Biosciences), and F4/80 (monocytes/macro-phages) (Serotec). Nonspecific binding of the fluorescence-conjugatedmAbs to FcRs was inhibited by preincubation of the cells with unlabeledmAbs to CD16 and CD32 (Fc block; BD Biosciences). Flow cytometrywas performed using a FACSCalibur instrument and CellQuestPro soft-ware (BD Biosciences). The hind limbs of naive or arthritic mice weredissected, fixed in formalin, decalcified, and embedded in paraffin. Thetissue sections were stained with H&E.

Assays of Ag (PG)-specific immune responses

These assays were performed as described before (5–7, 31). In brief, spleenor LN cells were cultured in DMEM (Sigma-Aldrich) supplemented with10% FBS (HyClone) in round-bottom 96-well plates (2 � 105 cells/200�l/well, in triplicate wells) with or without hPG (25 �g protein/ml). Todetermine hPG-specific cell proliferation, the cells were pulsed with[3H]thymidine (1 �Ci/well) on the fourth day of culture and harvested onday 5. Radioisotope incorporation was measured by scintillation countingand expressed in counts per minute. PG-specific IL-2 production by spleencells was determined using a CTLL-2 bioassay, where IL-2 bioactivity wasmeasured in the supernatants after 24 h of culture. To determine the PG-specific production of IFN-� and IL-4, spleen cells were cultured for 5 daysusing the same cell density and Ag dose as indicated for the proliferationassays. The concentration of IFN-� and IL-4 in the culture supernatantswas measured using ELISA (BD Biosciences).

PG-specific Abs in serum were measured by ELISA (6, 31). Briefly,mice were bled at defined time points, and serum samples were stored at�20°C. Maxisorp microplates (Nunc) were coated with hPG or mPG (0.5�g protein/well in sodium carbonate coating buffer (pH 9.5)) overnight at4°C. Fat-free milk was used to block nonspecific protein binding sites.Serially diluted serum samples and internal standard samples (pooled serafrom arthritic mice with known anti-PG Ab titers) were incubated with theimmobilized PGs, and plate-bound hPG- or mPG-specific Abs were de-tected using peroxidase-conjugated rabbit IgG against mouse IgG1 andIgG2a isotypes (Zymed Laboratories), respectively.

Intravital video microscopy

IVM on the mouse ankle synovium was performed as described previously(11). In brief, the mouse was anesthetized by i.m. injection of a xylazine-ketamine mixture. The ankle was immobilized on a custom-built stage, and

1933The Journal of Immunology

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

the Achilles tendon was cut off the calcaneus to expose the ankle synovialtissue to the intravital fluorescence microscope (model Eclipse E600FN;Nikon). The synovium of the open ankle was continuously superfused withwarm PBS through a fine silicon tube attached to the objective lens. Tovisualize leukocytes in the circulation, the mouse was injected i.v. with 10�g of rhodamine 6G (Molecular Probes), a cell-permeable fluorescentDNA-binding dye (11, 26). Leukocyte trafficking was recorded by stream-line acquisition of images with a digital video camera (CoolSnap CF; RSPhotometrics) fitted to the microscope. Blood vessel diameters were mea-sured, and leukocyte adhesion behavior was analyzed offline using Meta-vue software (Universal Imaging). Leukocytes present in postcapillaryvenules with diameters ranging from 20 to 50 �m were selected for anal-ysis. The parameters measured were the number of rolling leukocytes per103 �m (1 mm) vessel perimeter per minute, rolling velocity (micrometerper second), and the number of adherent leukocytes per 105 �m2 of endo-thelial surface per minute (11, 26). For in vivo detection of CD44 andCD62L expression or the presence of granulocytes in the synovial vessels,mice were injected with fluorescence-conjugated mAbs against CD44,CD62L, or Gr-1 (4 �g mAb/mouse) instead of rhodamine 6G.

Lymphocyte trafficking assays

To compare the ability of WT and KO spleen cells to migrate to lymphoidor extra-lymphoid organs, we performed short-term mixed-cell homingassays (23, 28). First, lymphoid tissue (and homeostasis) was reconstitutedin the immunodeficient mice by i.p. injection of splenocytes (5 � 107 cellswith 200 �g of hPG) from immunocompetent WT or KO BALB/c micewith primary PGIA, as described for the adoptive transfer experiments.Two weeks later, the reconstituted SCID mice received a mixture of WTand KO spleen cells from arthritic donor mice after labeling the donor cellswith stable cell-tracing fluorophores of different colors. WT spleen cellswere labeled with either CellTracker Green (catalog no. C7025; 3 �M) orCellTracker Orange (catalog no. C2927; 5 �M) and KO cells with Cell-Tracker Orange, according to the manufacturer’s instructions (MolecularProbes). The green fluorescent (WT) and “orange” (red) fluorescent (WT orKO) spleen cells were mixed at a 1:1 ratio, and a total of 2 � 107 viablecells (with 100 �g of hPG) in a final volume of 150 �l was injected i.v. intoeach SCID recipient. SCID mice previously reconstituted with WT spleno-cytes received a mixture of red and green fluorescence-labeled WT donorcells, and those injected with KO splenocytes received a mixture of red KOand green WT donor cells (23). The hosts were killed 24 h later, andsingle-cell suspensions were prepared from peripheral blood, spleen, jointdraining LNs, and ankle synovial fluid. Labeled donor cells in these sus-pensions, along with those in the original samples of mixed donor cells(before injection), were detected using two-color flow cytometry. The netratios of red:green fluorescent donor cells that had migrated into SCIDtissues were calculated after normalizing the actual ratios of red:green-labeled cells retrieved from the host’s LNs and spleen (RLN and RSpl,respectively) to the ratios of red:green donor cells in the mixtures beforeinjection (RInj) (23). SCID mice injected with a 1:1 mixture of red andgreen fluorescent WT donor cells served as internal controls, where themigration efficiency of differentially labeled WT cells was expected to beidentical.

Statistical analysis

Data were analyzed using one-way ANOVA (SPSS). The post hoc Dun-nett’s t test was used to determine significant differences ( p � 0.05) be-tween the WT and any of the three KO groups of mice. For time courseexperiments conducted on the same sets of mice, repeated measuresANOVA was also performed to determine significant differences ( p �0.05) between the groups in the overall responses.

ResultsThe onset of PGIA is delayed in CD62L KO and CD44/CD62Ldouble KO mice, and disease severity is reduced in CD44 KO,CD62L KO, and double KO mice

To determine whether expression of CD44 and CD62L was re-quired for the development of PGIA, we immunized four geno-types (WT, CD44 KO, CD62L KO, and double KO) of BALB/cmice with hPG. To achieve maximum PGIA incidence and sever-ity in all groups, mice were given four i.p. injections of hPG inadjuvant at wk 0, 3, 6, and 9. They were monitored for the onsetof PGIA and the severity of arthritic symptoms until wk 11. Asshown in Fig. 1A, the incidence of PGIA was reduced in all KOmice as compared with age-matched WT animals. CD62L KO and

double KO mice also showed a delay in arthritis onset, but themajority (70–100%) of the mice developed arthritis by wk 11 re-gardless of genotype (Fig. 1A). The severity of PGIA was reducedin both single and double KO mice relative to WT (Fig. 1B). Thisdifference was significant for the CD62L KO and double KOgroups until the end of the observation period (wk 11) and for theCD44 KO group on wk 7, 9, and 10 (Fig. 1B).

It was very unlikely that genetic “contamination” (microchimer-ism remaining from the original backgrounds) accounted for thereduced incidence and severity of PGIA in KO BALB/c mice, asgenome screening with 244 polymorphic markers (including thosecovering quantitative trait loci that control arthritis susceptibility,severity, and immune responses in PGIA (30)) detected no differ-ence between the WT and any of the KO groups of mice (data notshown).

Ag-specific cell proliferation is reduced, but cytokine and Abresponses are normal in hPG-immunized CD44KO, CD62L KO,and double KO mice

CD62L has been shown to play a crucial role in LN homing andsubsequent activation of lymphocytes upon contact with Ag (18–20). Therefore, the delay in arthritis development and reduced dis-ease severity, particularly in mice lacking CD62L (Fig. 1B), couldbe the consequence of inadequate response to PG immunizationdue to impaired lymphocyte homing. We compared the magnitudeand kinetics of immune responses in the four genotypes of mice by

FIGURE 1. Incidence and severity of primary PGIA in WT, CD44 KO,CD62L KO, and CD44/CD62L double KO BALB/c mice. Mice (n � 20/group) were immunized with hPG four times at 3-wk intervals and mon-itored for the incidence and severity of PGIA. Arrows indicate the third andfourth hPG injections given on wk 6 and 9, respectively. A, Incidence isexpressed as the percentage of mice that developed arthritis. B, Diseaseseverity (cumulative score) is the sum of paw inflammation scores dividedby the number of arthritic mice. The results shown are the means � SEM.The symbols denote significant differences (p � 0.05; Dunnett’s t test)between the WT and CD44 KO (�), WT and CD62L KO (†), and WT anddouble KO (‡) groups at the defined time points. Repeated measuresANOVA indicated that the differences between the WT and KO groupswere significant (p � 0.05) during the entire observation period.

1934 CD44 AND CD62L IN PG-INDUCED ARTHRITIS

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

measuring PG-specific LN and spleen cell proliferation, produc-tion of IL-2, IL-4, and IFN-� by splenocytes in vitro, as well asserum levels of mPG-specific autoantibodies (both Th1- and Th2-polarized isotypes) at wk 0 (before immunization), wk 8 (2 wkafter the third PG injection), and wk 11 (2 wk after the fourth PGimmunization). The hPG-specific proliferation of spleen cells fromall KO mice was significantly weaker than the proliferation of WTsplenocytes at both wk 8 and 11 of immunization (Fig. 2A). LNcells from the immunized mice (including WT) generally gavevery poor proliferative responses to in vitro stimulation with hPG;proliferation was essentially undetectable in all KO LN cultures atwk 8 but was detectable by wk 11 in CD44 KO LN cultures (Fig.2A). IL-2 was produced in approximately equal amounts by hPG-stimulated splenocytes, with the highest levels in double KO cul-tures, but IL-2 could not be detected in the LN cell cultures (datanot shown). PG-specific IFN-� production by spleen cell cultureswas comparable in all genotypes, and less IL-4 was found in theCD44KO and double KO than in the WT cultures at wk 8 but notat wk 11 (Fig. 2B). Serum levels of mPG-specific IgG2a and IgG1

autoantibodies were not reduced in KO mice; in fact, we foundsignificantly elevated levels of the IgG2a isotype of these autoan-tibodies in serum samples of CD44 KO mice at both wk 8 and 11and of the IgG1 isotype in the sera of CD62L mice at wk 11 ofimmunization (Fig. 2C). Taken together, except for a reduced pro-liferative response of spleen cells from all three groups of KO miceto in vitro Ag stimulation (which correlated well with the reducedseverity of PGIA in these groups), adaptive immunity did not seemto be compromised in PG-immunized CD44 KO, CD62L KO, anddouble KO mice.

Leukocytes show altered adhesion behavior in the inflamedsynovial vessels of KO mice

As both CD44 and CD62L have been shown to support leukocyterolling (18, 20, 32–34), the reduced severity of arthritis, observedin mice lacking one or both of these receptors (Fig. 1B), could alsobe attributed to impaired leukocyte rolling and subsequent recruit-ment in the target organs. We performed IVM on mouse anklejoints (11) to determine whether the adhesion behavior of leuko-cytes in the synovial postcapillary vessels of ankle synovium weredifferent in the four genotypes of mice. Both nonarthritic and ar-thritic animals were used for IVM experiments. In the absence ofCD44 or CD62L or both, we expected the number of leukocytesthat rolled on synovial endothelium to decrease, and rolling ve-locity to rise, as a result of loss of one or two major receptors thatcan mediate rolling. However, the number of rolling cells was notreduced in the synovial vessels of arthritic CD44 KO and CD62LKO ankles and was even significantly elevated in double KO, com-pared with WT mice (Fig. 3A). The velocity of rolling cells wassignificantly higher in double KO than WT mice (Fig. 3B). A com-pensatory increase in the circulating pool of leukocytes expressingPSGL-1, another adhesion molecule that mediates rolling via en-dothelial or platelet selectin binding (35), could explain the in-creased propensity of double KO cells to roll, but we found littleevidence for such compensation (the average proportion ofPSGL-1� cells was 31% in WT and 39% in double KO blood).Although the percentage of leukocytes expressing CD62L waslower in CD44 KO than in WT mice (11 and 22%, respectively),the absence of CD44, combined with low CD62L expression inCD44 KO cells, did not result in increased rolling interactions(Fig. 3A) or elevated rolling velocity (Fig. 3B). Importantly, thenumber of firm adherent cells was reduced in the vessels of all KOmice and was significantly lower in CD62L and double KO than inWT mice (Fig. 3C). The difference between WT and KO leuko-cytes, with respect to firm adherence to endothelium, remainedconsistent when WT and KO ankle joints with similar arthritisscores were compared (data not shown).

CD44- and CD62L-expressing cells of the innate immune systemcontribute to the development of severe PGIA

PGIA can be adoptively transferred to syngeneic SCID mice usingspleen cells isolated from BALB/c donors with primary PGIA(31). SCID mice lack functional T and B cells, but they possess anintact innate immune system. Therefore, lymphocyte transfer toSCID mice could provide help in determining whether expressionof CD44 and CD62L in the innate or adaptive arm of the immunesystem is more important for development of a severe form ofadoptive PGIA. If expression of CD44 and/or CD62L in the adap-tive immune system is required for severe disease, then SCID micereconstituted with splenocytes from CD44 KO, CD62L KO, ordouble KO donors with low-severity PGIA are expected to de-velop mild arthritis. However, SCID mice that received spleencells from arthritic KO donors all developed PGIA promptly (Fig.4A) and with similar or even higher severity (Fig. 4B) than their

γ

FIGURE 2. PG-specific immune responses in WT, CD44 KO, CD62LKO, and double KO BALB/c mice. A, hPG-specific proliferation of spleencells (left graph) and joint-draining LN cells (right graph) were assayedbefore (wk 0) and after immunization with hPG (wk 8 and 11) using[3H]thymidine incorporation as described in Materials and Methods (n �3–9 mice/group/time point). B, The concentrations of IFN-� (left graph)and IL-4 (right graph) were measured in hPG-stimulated spleen cell cul-ture supernatants by ELISA (n � 6 mice/group/time point). C, Theamounts of mPG-specific IgG2a (left graph) and IgG1 (right graph) au-toantibodies in serum samples (collected at wk 0, 8, and 11) were deter-mined using ELISA (n � 10–15 mice/group/time point). The error barsrepresent SEM. The symbols are as described in the legend of Fig. 1 anddenote significant differences between the WT and the KO groups.

1935The Journal of Immunology

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

counterparts reconstituted with cells from arthritic WT donors.This observation suggested that the development of adoptive dis-ease was not hampered by the reduced ability of KO donor spleencells to proliferate in the presence of PG (Fig. 2A) and that host(SCID)-derived cells significantly contributed to disease severity.Indeed, in vivo immunostaining identified leukocytes expressingCD44 and CD62L not only in the inflamed synovial venules ofSCID mice reconstituted with WT donor cells (Fig. 4C, left panel),but also in the vessels of SCID hosts that received double KO(CD44/CD62L-negative) splenocytes (Fig. 4C, right panel). Thisprovided in vivo evidence for the recruitment of host (innate im-mune) cells at the site of inflammation in PGIA.

“Arthritogenic” spleen cells home to the LNs in a CD62L-dependent manner and to the spleen in a CD62L-independentmanner but are undetectable in the joint fluid of SCID mice

As spleen (not LN) cells were used for the transfer of PGIA fromarthritic donors to SCID mice, we could not rule out the possibilitythat CD62L-independent occupancy of the SCID LNs by thesplenocytes accounted for the induction of severe disease, partic-ularly in animals injected with cells from CD62L-deficient or dou-ble KO donors. Furthermore, spleens from arthritic donors couldcontain an “effector” population of either lymphoid or nonlym-phoid cells, capable of migrating to the joints directly. To inves-tigate these possibilities, we conducted in vivo trafficking assays

by transferring fluorescence-labeled spleen cells from arthritic WTand KO donors into SCID mice. As described in Materials andMethods, WT donor splenocytes were labeled with a green fluoro-phore and mixed at 1:1 ratio with red fluorescence-labeled cellsfrom each genotype of the KO donors, which allowed for simul-taneous detection of WT and KO cells upon their retrieval from theSCID organs 24 h after transfer. As shown in Fig. 5, A and B,spleen cells from CD44 KO donors migrated significantly betterthan WT cells to the LNs of SCID hosts, whereas LN homing ofCD62L KO cells was impaired severely. Because CD44 KOsplenocytes demonstrated enhanced LN occupancy, it was ex-pected that CD44 deficiency could compensate for this homingdefect of CD62L-null cells. This was not the case, however, assplenocytes lacking both CD62L and CD44 showed very modestimprovement in LN homing compared with spleen cells from

FIGURE 3. Leukocyte adhesion behavior in the synovial postcapillaryvenules of arthritic ankles of WT and KO mice as determined by IVM.IVM was performed on the ankle joints as described previously (11). Bloodleukocytes were visualized by i.v. injection of rhodamine 6G into anes-thetized mice. The adhesion behavior (rolling and firm adhesion) of leu-kocytes in the synovial venules was video recorded and analyzed to de-termine the number of rolling cells per unit length of vessel perimeter (A),rolling velocity (B), and number of adherent cells per unit area of thevascular endothelium (C), as described in Materials and Methods. Similarnumbers of WT and KO ankles were compared at each level of inflam-mation (score 1 through score 4). IVM experiments were scheduled ac-cording to the onset and progression of ankle inflammation in the fourgenotypes of mice and were completed in all groups by wk 16 of immu-nization. Results are expressed as the means � SEM (n � 20–23 mice/group). Asterisks indicate significant differences (p � 0.05; Dunnett’s ttest) between the WT and KO mice.

FIGURE 4. Incidence and severity of adoptively transferred PGIA inSCID mice and participation of host (SCID) cells in the inflammatoryreaction. Incidence (A) and severity (B) of the adoptively transferred dis-ease in syngeneic SCID mice. SCID mice were injected i.p. with spleencells from arthritic WT, CD44 KO, CD62L KO, or double KO BALB/cmice on wk 0 and 2 (arrows) as described in Materials and Methods.Disease incidence and severity were monitored until wk 5 and expressed asshown in Fig. 1. The data in B are the means � SEM (n � 10 recipientSCID mice/group), and the asterisk (wk 1) denotes significant difference(p � 0.05; Dunnett’s t test) between SCID mice injected with cells fromWT and those receiving cells from CD62L KO donors. The overall diseaseseverity was not significantly different among the four groups (p � 0.05;repeated measures ANOVA). C, SCID mice, reconstituted with unlabeledspleen cells from WT (left panel) or double KO (right panel) BALB/cdonors, were injected i.v. with fluorescence-conjugated mAbs againstCD44 (green fluorescence) and CD62L (red fluorescence). Expression ofthese adhesion molecules by leukocytes in the synovial vessels of the hostSCID mice was visualized by IVM. Yellow fluorescence represents leu-kocytes coexpressing (Co-expr.) CD44 and CD62L.

1936 CD44 AND CD62L IN PG-INDUCED ARTHRITIS

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

CD62L single KO mice (Fig. 5, A and B). In contrast to LNs, thespleen was occupied with similar efficiency by WT and KO donorsplenocytes, although relative to WT cells, the ratio of CD44 KOcells was slightly decreased, and the ratios of CD62L and doubleKO cells were increased in the spleens of SCID mice (Fig. 5, C andD). We were unable to detect fluorescent donor cells in the anklesynovial fluid of SCID mice 24 h (or even 72 h) after transfer (datanot shown).

The severity of arthritis is determined by the abundance ofinnate immune cells (granulocytes) in the inflamed joints

As described above, severe PGIA developed promptly in SCIDmice after transfer of KO spleen cells (Fig. 4B), although the LNswere poorly populated in mice reconstituted with cells fromCD62L KO or double KO donors (Fig. 5), and arthritogenic donorsplenocytes did not seem to migrate to the joint cavity. On theother hand, following transfer of unlabeled spleen cells from dou-ble KO donors to SCID mice, leukocytes expressing CD44 andCD62L (obviously of host origin) were detected by IVM in thesynovial vessels of inflamed SCID joints (Fig. 4C). These obser-vations indicated that innate immune cells significantly contributeto the severity of joint inflammation in PGIA. The most likelycandidates are the granulocytes because these cells (mainly neu-trophils) are present in large numbers in the joints of mice withPGIA (3, 5, 11) and other models of immune-mediated arthritis(26, 36), as well as in the inflamed joints of RA patients (37).Therefore, we investigated the relationship between granulocyteinflux and disease severity, first focusing on PGIA in WT mice.IVM following in vivo immunostaining for Gr-1 revealed the pres-ence of very few (mostly rolling) granulocytes in the synovialvenules of noninflamed ankles (score: 0), but the number of cellsthat interacted with the endothelium increased in correlation withthe progression of arthritis (from score 1 to 3) in these joints (Fig.6A). The relative number of granulocytes in the joint fluid followeda similar pattern: very few Gr-1� cells were detected by flow cy-

tometry in the fluid of noninflamed ankles, but the proportion ofgranulocytes “skyrocketed” with arthritis progression (Fig. 6B).This trend was also observed on histological sections of the ankles:as inflammation advanced, more and more polymorphonuclear leu-kocytes accumulated in both the synovial tissue and the joint cav-ity of the affected ankles (Fig. 6C). When comparing the cellularcomposition of synovial fluid cells from WT and KO joints ofmatching inflammation scores, we found that the proportions (%)of Gr-1� cells were somewhat lower in KO than WT joints at alow inflammation score but were similarly high (�90%) at anadvanced degree of inflammation (Fig. 6D). T and B cells andmonocytes/macrophages were minor constituents of the synovialfluid exudate in both WT and KO mice, and, in contrast to gran-ulocytes, these mononuclear cell populations did not show muchchange in size during arthritis progression (data not shown).

DiscussionAs in RA (1, 2, 38, 39), cells of both the adaptive and the innateimmune systems are involved in the initiation and perpetuation ofchronic joint inflammation in PGIA (6). Autoimmunity to mousePG in BALB/c mice develops after an induced response to a het-erologous cartilage Ag (hPG) (3, 5, 6). Recognition of hPG in theadaptive immune system and breakdown of tolerance to a self-cartilage component (mPG), which both occur in the secondarylymphoid organs (LNs and spleen), are followed by the sustainedmigration of effector cells into the target organs (joints). Leukocyteentry to LNs and inflammatory sites has been shown to depend onthe expression of adhesion molecules, including CD62L and CD44(18–23). In this study, we used BALB/c mice lacking CD44,CD62L, or both receptors to elucidate the requirement for theseadhesion molecules for lymphocyte migration to the lymphoid or-gans (and subsequent generation of the immune response), as wellas for leukocyte entry into the joints in PGIA.

As expected, the incidence and severity of arthritis were reducedin the mutant animals as compared with WT mice, and resistance

FIGURE 5. Migration of spleen cells of arthritic WT and KO BALB/c donors to the LNs and spleens of SCID mice. The distribution of fluorescence-labeled WT and KO donor cells are shown in the peripheral LNs (A and B) and the spleens (C and D) of SCID recipients. SCID mice received differentiallylabeled WT or KO donor cells. WT splenocytes were labeled with fluorescent CellTracker Green or Orange and KO cells with CellTracker Orange. Thedifferentially labeled cells were mixed and injected into the SCID hosts as described in Materials and Methods. Green and orange (red) fluorescent donorcells in the SCID LNs and spleen were detected by two-color flow cytometry 24 h later. The net ratio of WT and KO cells in the LNs or spleen wascalculated from the ratio of red:green cells in the retrieved (RLN or RSpl) and injected (RInj) samples. Data shown are means � SEM (n � 6–7 SCIDmice/group). The asterisks indicate significant differences (p � 0.05; Dunnett’s t test) between the net ratios of retrieved WT and KO donor cells. B andD, Representative samples of the flow cytometry profiles of labeled donor cells in the cell suspensions of SCID LNs and spleens, respectively.

1937The Journal of Immunology

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

to severe arthritis was more pronounced in mice lacking CD62L(single or double KO) than in those lacking CD44 only. Theseobservations, which suggest that CD62L might play a greater rolethan CD44 in the development of an inflammatory joint disease,

resonate with our previous study where mice lacking CD62L werefound to be more resistant to a nonautoimmune form of arthritisthan CD44 KO mice (26). It is conceivable that the generation ofthe autoimmune response upon active immunization (as in PGIA)requires a high level of coordination within the adaptive immunesystem by means of directed cell migration to, and acquisition ofimmunological memory in, the lymphoid compartments. Subse-quent development of the inflammatory disease also requires co-operation between adaptive and innate immune system players atthe periphery. The question arises as to which processes and play-ers are affected the most by the loss of CD62L and/or CD44 ex-pression in PGIA.

The crucial role of lymphocyte CD62L in LN homing has beenestablished (15, 18). Although CD44 has not been directly impli-cated in this process, we (23) and others (24) found that CD44-deficient lymphocytes accumulated in LNs with greater efficiencythan WT cells after in vivo activation. Based on these findings, itwas expected that Ag (both hPG and mPG)-specific immune re-sponses would be severely reduced in PG-immunized CD62L-de-ficient mice, enhanced in CD44 KO, and close to normal in CD44/CD62L double KO mice. Besides a surprisingly low magnitude ofPG-specific LN cell proliferation (relative to spleen cells), wefound that the absence of CD62L expression had a negative impacton the Ag-specific response of LN cells in PG-immunized mice.Lymphocytes from the spleens of all KO mice exhibited reducedproliferative responses to hPG, for which there is no clear expla-nation at this time. However, the same cells demonstrated normalcytokine production, suggesting that the signaling function ofsplenocytes was preserved in the KO mice. Compared with WTmice, autoantibodies against mPG of both IgG2a and IgG1 iso-types were produced with equal or greater efficiency in all KOmice, indicating that the lack of CD62L and CD44 expression hadno major effect on autoantibodies production (4, 5) or the Th1/Th2balance associated with PGIA (7).

How could CD62L KO mice mount a globally “normal” (auto-)immune response in the face of a LN homing defect? The answermost likely lies in the systemic (i.p.) route of Ag administrationused to induce PGIA. As confirmed by our cell transfer results inthe present study, CD62L is required for LN homing, but it is notnecessary for lymphocyte entry into the spleen (40). The require-ment for CD62L for acquisition of full immunological memorywas therefore bypassed by the i.p. route of immunization, allowingfor Ag delivery to the spleen, which lymphocytes could enter in aCD62L-independent fashion. The weak LN (vs strong spleen) im-mune responses in the PG-injected mice suggest a greater contri-bution of the spleen than of the LNs to the generation of immu-nological memory, but the best evidence for the prominent role ofthe spleen in PGIA development is the ability of splenocytes fromarthritic BALB/c donors to transfer disease to immunodeficientmice (8, 31). In the present study, spleen cells from WT, CD62LKO, CD44 KO, and double KO BALB/c mice were equallyprompt and effective in inducing severe arthritis upon transfer toSCID mice. This suggested that in PGIA, the spleen was capableof gathering lymphocytes and APCs in a CD62L- and CD44-in-dependent manner, as well as providing a tissue microenvironmentfor the assembly and function of the adaptive (auto)immune sys-tem. Furthermore, the cell transfer studies indicated that spleno-cytes from PG-immunized CD62L KO and double KO donors thatreadily populated the spleen, but migrated poorly to the LNs ofSCID mice, were still potent in “setting the stage” for severe jointinflammation in the SCID host. This raised the possibility that thefailure of these arthritogenic spleen cells to trigger severe diseasein the primary form of PGIA could be due to inefficient recruitmentof KO effector cells to the inflammatory site.

FIGURE 6. Localization and abundance of granulocytes in noninflamedand inflamed ankles of mice with primary PGIA. A, Gr-1� cells were visual-ized in the ankle synovial vessels by IVM following in vivo labeling of bloodleukocytes with fluorescence-conjugated anti-Gr-1 mAb. The IVM snapshots(column A) show granulocytes (Gr-1high cells) interacting with endothelium ina noninflamed ankle (score: 0, top panel), a mildly inflamed (score: �1, middlepanel), or a severely inflamed (score: �3, bottom panel) ankle joint in WTmice. (Note that the video camera captures rolling or adherent granulocytes butnot cells freely flowing in the bloodstream.) B, Granulocytes were also iden-tified in the synovial fluid of WT ankles by in vitro immunolabeling and flowcytometry. The top, middle, and bottom graphs show the percentage of Gr-1�

cells in the joint fluid of noninflamed, mildly inflamed, and severely arthriticankles, respectively. Synovial fluid samples were harvested and pooled from atleast three ankle joints at each level of inflammation. C, Histopathology ofankles from WT mice illustrates the abundance and distribution of inflamma-tory cells in the joint compartments at different clinical scores. Very few leu-kocytes are seen at score 0. Infiltration of both the synovial tissue (St) and jointcavity (Jc) by leukocytes rises with increasing joint swelling (from score �1to �3). High-magnification insets demonstrate that the majority of infiltratingcells in both mildly and severely inflamed joints are neutrophil granulocytes(showing segmented nuclei). H&E staining, Bo: bone (calcaneus). Scale bars,250 �m (50 �m for the insets). D, Percentage of granulocytes in the synovialfluid of ankle joints from WT, CD44 KO, CD62 KO, and double KO mice atdifferent levels of inflammation. Synovial fluid cells were harvested (n � 3ankles/score/group) and pooled according to score and genotype. The pooledcells were labeled with anti-Gr-1 mAb and analyzed by flow cytometry. Thedata shown are from two experiments and indicate the mean percentage ofGr-1� cells in pooled synovial fluid samples of WT and KO ankle joints.

1938 CD44 AND CD62L IN PG-INDUCED ARTHRITIS

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

Using IVM, we found that leukocytes in the synovial venules ofdouble KO mice with primary PGIA exhibited significantly in-creased rolling interactions, elevated rolling speed, and a reducedability to adhere firmly to endothelium as compared with leuko-cytes in WT mice. The simplest explanation for the increased pro-portion of rolling cells in double KO mice is that leukocytes, whichinteracted with the endothelium but failed to adhere, kept rolling.In this respect, the lack of CD62L and CD44 function could havebeen compensated for by PSGL-1 and perhaps also by “partiallyactivated” integrins that are able to support rolling (41). It is likelythat double KO cells were more easily carried away in a fast roll-ing motion by the shear force of the blood flow than their better-equipped WT or single KO counterparts. Fast rolling, in turn,shortened the exposure of leukocytes to chemokines, thus loweringthe rate of integrin-mediated arrest and firm adhesion (14). In thesynovial microcirculation of inflamed joints, all KO leukocytesdemonstrated decreases in firm adhesion compared with WT cellsin the order of double KO � CD62L KO � CD44 KO � WT,which correlated quite well with the severity of primary PGIA inthese groups of mice. As firm adhesion is a committed step that isusually followed by transendothelial migration (12), decreased leu-kocyte adherence could lead to reduced extravasation and less se-vere inflammation. As noted above, spleen cells from these mildlyarthritic KO donors were able to trigger severe arthritis in SCIDmice, suggesting that the bulk of leukocytes migrating into thejoints in primary PGIA did not represent the cells of the adaptiveimmune system.

We showed that leukocytes expressing CD44 and CD62L werebeing recruited in the synovial venules of SCID recipients aftertransfer of splenocytes from double KO donors, thus providing invivo evidence for the participation of host-derived innate immunecells in the local inflammatory process. Furthermore, in primaryPGIA, we demonstrated a positive relationship between the degreeof inflammation and the abundance of granulocytes (neutrophils)that interacted with synovial endothelium and subsequently mi-grated into the synovial tissue and joint cavity. Although T and Bcells, as well as macrophages, could be identified in both nonin-flamed and inflamed joints, these cells were present in low pro-portions, and their numbers did not increase markedly with arthri-tis progression. Collectively, these results suggest that in both theprimary and adoptive forms of PGIA, the severity of inflammationlargely depends on the abundance of innate immune cells (granu-locytes) in the joint. We also found that the ankle joint fluid of KOmice contained fewer granulocytes than the synovial fluid of WTmice at a low degree of inflammation in primary PGIA, but KOgranulocytes “caught up” with WT cells in ankle joints at an ad-vanced stage of inflammation. These results, together with theIVM data, suggest that the lack of CD62L and CD44 expressionlowers the rate of granulocyte influx into the joints, thereby de-laying the onset and slowing down the progression of PGIA. How-ever, as we have shown earlier in WT mice with PGIA (11), at anadvanced stage of inflammation (score �3), the joints become sat-urated with inflammatory cells, leading to a decline in the rate ofleukocyte extravasation. Thus, with slowly continuing recruitment,KO granulocytes could catch up and eventually infiltrate the jointsto the same extent as WT cells did. This was supported by theobservation that the severity of primary PGIA, although reducedinitially in CD44 KO, CD62L KO, and double KO mice, reacheda degree similar to WT by wk 15 of immunization (B. Sarraj andK. Mikecz, unpublished data). Adoptively transferred PGIA de-veloped and progressed in a WT-like fashion in SCID mice recon-stituted with KO spleen cells, suggesting that CD44 and CD62L,expressed in the hosts’ innate immune system, facilitated leuko-cyte recruitment. Therefore, the innate immune cells that contrib-

uted mostly to disease severity must have been neutrophil granu-locytes from the SCID mice. Using IVM after in vivocoadministration of fluorescent mAbs to Gr-1 and CD44, we foundthat the majority of leukocytes interacting with the synovial endo-thelium of SCID mice (which developed arthritis upon transfer ofunlabeled double KO spleen cells) coexpressed Gr-1 and CD44 (B.Sarraj and K. Mikecz, unpublished results), indicating that most ofthe cells recruited in the inflamed joints were indeed host-derivedgranulocytes.

Murine neutrophils lacking CD62L (42) or CD44 (43) havebeen shown to exhibit impaired recruitment in vivo in response to,for example, local stimulation with proinflammatory cytokinessuch as TNF-� or chemokines. The requirement for CD62L (20) orCD44 (22) for T cell migration into inflammatory sites has beenproposed but not firmly established. Because of the paucity of Tcells that extravasate in the joint in PGIA (11), it is difficult toascertain whether expression of these adhesion receptors is neces-sary for the migration of T lymphocytes to this site. Our adoptivetransfer study suggests that, if CD62L and/or CD44 had been crit-ical for the entry of activated (arthritogenic) T cells into the joint,inflammation would have been reduced in SCID mice reconsti-tuted with lymphocytes lacking one or both of these adhesion mol-ecules, but this was not the case.

Taken together, neutrophil granulocytes emerge as the majorinflammatory cell population reliant on the expression of CD62Land, to a lesser extent, of CD44 for efficient influx into the joint inPGIA. In RA, neutrophils emigrate from postcapillary venules lo-cated in the synovial sublining area into the joint cavity in quitelarge numbers, causing severe damage to the cartilage matrix (9,37, 44). Among other factors, elevated serum concentrations ofTNF-�, and the presence of autoantibodies-containing comple-ment-fixing immune complexes in the joint, have been implicatedin neutrophil recruitment in RA (39). As demonstrated by in vivostudies, TNF-�-activated vascular endothelium promotes neutro-phil recruitment (45), and autoantibodies-containing immune com-plexes (e.g., serum from arthritic K/BxN mice) preferentially lo-calize to the joints and initiate neutrophil-dependent arthritis upontransfer to naive mice (36, 46). These finding suggest that TNF-�and autoantibodies, which are mainly (TNF-�) or exclusively (au-toantibodies) produced by adaptive immune cells, can access thejoint and attract granulocytes, thereby effectively connecting theadaptive and innate arms of the immune system in susceptibleindividuals. Therefore, the clinical efficacy of TNF-�- and B cell-targeting therapies in RA can be, in part, attributed to inhibition ofneutrophil recruitment by reducing the interactions between adap-tive and innate immunity (47, 48). Agents that antagonize CD62Land CD44 function, downstream of these interactions, could pro-vide further aid in the therapeutic effort to suspend neutrophil in-flux into the joints in RA.

AcknowledgmentsWe thank Sandor Szanto, Zoltan Szabo, and Vyacheslav Adarichev forhelp with the genotyping and breeding of KO mice. We also thank IstvanGal and Gabor Ludanyi for their assistance in IVM and statistical analysis,respectively.

DisclosuresThe authors have no financial conflict of interest.

References1. Fox, D. A. 2001. Etiology and pathogenesis of rheumatoid arthritis. In Arthritis

and Allied Conditions: A Textbook of Rheumatology. W. J. Koopman, ed. Lip-pincott Williams & Wilkins, Philadelphia, pp. 1085–1102.

2. Goronzy, J. J., and C. M. Weyand. 2005. Rheumatoid arthritis. Immunol. Rev.204: 55–73.

1939The Journal of Immunology

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

3. Glant, T. T., K. Mikecz, A. Arzoumanian, and A. R. Poole. 1987. Proteoglycan-induced arthritis in BALB/c mice: clinical features and histopathology. ArthritisRheum. 30: 201–212.

4. Mikecz, K., T. T. Glant, and A. R. Poole. 1987. Immunity to cartilage proteo-glycans in BALB/c mice with progressive polyarthritis and ankylosing spondy-litis induced by injection of human cartilage proteoglycan. Arthritis Rheum. 30:306–318.

5. Glant, T. T., A. Finnegan, and K. Mikecz. 2003. Proteoglycan-induced arthritis:immune regulation, cellular mechanisms and genetics. Crit. Rev. Immunol. 23:199–250.

6. Glant, T. T., and K. Mikecz. 2004. Proteoglycan aggrecan-induced arthritis: amurine autoimmune model of rheumatoid arthritis. Methods Mol. Med. 102:313–338.

7. Finnegan, A., K. Mikecz, P. Tao, and T. T. Glant. 1999. Proteoglycan (aggrecan)-induced arthritis in BALB/c mice is a Th1-type disease regulated by Th2 cyto-kines. J. Immunol. 163: 5383–5390.

8. O’Neill, S. K., M. J. Shlomchik, T. T. Glant, Y. Cao, P. D. Doodes, andA. Finnegan. 2005. Antigen-specific B cells are required as APCs and autoanti-body-producing cells for induction of severe autoimmune arthritis. J. Immunol.174: 3781–3788.

9. Goetzl, E. J., and I. M. Goldstein. 1985. Cellular components of inflamation:granulocytes. In Textbook of Rheumatology. W. N. Kelley, E. D. Harris, Jr.,S. Ruddy, and C. B. Sledge, eds. Saunders, Philadelphia, pp. 115–144.

10. Zhang, X., T. Nakajima, and C. M. Weyand. 2005. Tissue trafficking patterns ofeffector memory CD4� T cells in rheumatoid arthritis. Arthritis Rheum. 52:3839–3849.

11. Gal, I., E. Bajnok, S. Szanto, B. Sarraj, T. T. Glant, and K. Mikecz. 2005. Vi-sualization and in situ analysis of leukocyte trafficking into the ankle joint in asystemic murine model of rheumatoid arthritis. Arthritis Rheum. 52: 3269–3278.

12. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyteemigration: the multistep paradigm. Cell 76: 301–314.

13. von Andrian, U. H., J. D. Chambers, L. M. McEvoy, R. F. Bargatze, K. E. Arfors,and E. C. Butcher. 1991. Two-step model of leukocyte-endothelial cell interac-tion in inflammation: distinct roles for LECAM-1 and the leukocyte �2 integrinsin vivo. Proc. Natl. Acad. Sci. USA 88: 7538–7542.

14. Constantin, G., M. Majeed, C. Giagulli, L. Piccio, J. Y. Kim, E. C. Butcher, andC. Laudanna. 2000. Chemokines trigger immediate �2 integrin affinity and mo-bility changes: differential regulation and roles in lymphocyte arrest under flow.Immunity 13: 759–769.

15. Gallatin, M., T. P. St. John, M. Siegelman, R. Reichert, E. C. Butcher, andI. L. Weismann. 1986. Lymphocyte homing receptors. Cell 44: 673–680.

16. Aruffo, A., I. Stamenkovic, M. Melnick, C. B. Underhill, and B. Seed. 1990.CD44 is the principal cell surface receptor for hyaluronate. Cell 61: 1303–1313.

17. Yong, K., and A. Khwaja. 1990. Leukocyte cellular adhesion molecules. BloodRev. 4: 211–225.

18. Arbones, M. L., D. C. Ord, K. Ley, H. Ratech, C. Maynard-Curry, G. Otten,D. J. Capon, and T. F. Tedder. 1994. Lymphocyte homing and leukocyte rollingand migration are impaired in L-selectin-deficient mice. Immunity 1: 247–260.

19. Tedder, T. F., D. A. Steeber, and P. Pizcueta. 1995. L-selectin-deficient micehave impaired leukocyte recruitment into inflammatory sites. J. Exp. Med. 181:2259–2264.

20. Ley, K., and G. S. Kansas. 2004. Selectins in T cell recruitment to non-lymphoidtissues and sites of inflammation. Nat. Rev. Immunol. 4: 325–335.

21. DeGrendele, H. C., P. Estess, and M. H. Siegelman. 1997. Requirement for CD44in activated T cell extravasation into an inflammatory site. Science 278: 672–675.

22. Pure, E., and C. A. Cuff. 2001. A crucial role for CD44 in inflammation. TrendsMol. Med. 7: 213–221.

23. Stoop, R., I. Gal, T. T. Glant, J. D. McNeish, and K. Mikecz. 2002. Traffickingof CD44-deficient murine lymphocytes under normal and inflammatory condi-tions. Eur. J. Immunol. 32: 2532–2542.

24. McKallip, R. J., Y. Do, M. T. Fisher, J. L. Robertson, P. S. Nagarkatti, andM. Nagarkatti. 2002. Role of CD44 in activation-induced cell death: CD44-de-ficient mice exhibit enhanced T cell response to conventional and superantigens.Int. Immunol. 14: 1015–1026.

25. Catalina, M. D., M. C. Carroll, H. Arizpe, A. Takashima, P. Estess, andM. H. Siegelman. 1996. The route of antigen entry determines the requirement forL-selectin during immune responses. J. Exp. Med. 184: 2341–2351.

26. Szanto, S., I. Gal, A. Gonda, T. T. Glant, and K. Mikecz. 2004. Expression ofL-selectin, but not CD44, is required for early neutrophil extravasation in antigen-induced arthritis. J. Immunol. 172: 6723–6724.

27. Grewal, I. S., H. G. Foellmer, K. D. Grewal, H. Wang, W. P. Lee, D. Tumas,C. A. Janeway, Jr., and R. A. Flavell. 2001. CD62L is required on effector cellsfor local interactions in the CNS to cause myelin damage in experimental allergicencephalomyelitis. Immunity. 14: 291–302.

28. Stoop, R., H. Kotani, J. D. McNeish, I. G. Otterness, and K. Mikecz. 2001.Increased resistance to collagen-induced arthritis in CD44-deficient DBA/1 mice.Arthritis Rheum. 44: 2922–2931.

29. Robinson, S. T., P. S. Frenette, H. Rayburn, M. Cummiskey, M. Ullman-Cullere,D. D. Wagner, and R. O. Hynes. 1999. Multiple, targeted deficiencies in selectinsreveal a predominant role for P-selectin in leukocyte recruitment. Proc. Natl.Acad. Sci. USA 96: 11452–11457.

30. Adarichev, V. A., J. C. Valdez, T. Bardos, A. Finnegan, K. Mikecz, andT. T. Glant. 2003. Combined autoimmune models of arthritis reveal shared andindependent qualitative (binary) and quantitative trait loci. J. Immunol. 170:2283–2292.

31. Bardos, T., K. Mikecz, A. Finnegan, J. Zhang, and T. T. Glant. 2002. T and B cellrecovery in arthritis adoptively transferred to SCID mice: antigen-specific acti-vation is required for restoration of autopathogenic CD4� Th1 cells in a synge-neic system. J. Immunol. 168: 6013–6021.

32. von Andrian, U. H., J. D. Chambers, E. L. Berg, S. A. Michie, D. A. Brown,D. Karolak, L. Ramezani, E. M. Berger, K. E. Arfors, and E. C. Butcher. 1993.L-selectin mediates neutrophil rolling in inflamed venules through sialyl LewisX-dependent and -independent recognition pathways. Blood 82: 182–191.

33. DeGrendele, H. C., P. Estess, L. J. Picker, and M. H. Siegelman. 1996. CD44 andits ligand hyaluronate mediate rolling under physiologic flow: a novel lympho-cyte-endothelial cell primary adhesion pathway. J. Exp. Med. 183: 1119–1130.

34. Gal, I., J. Lesley, W. Ko, A. Gonda, R. Stoop, R. Hyman, and K. Mikecz. 2003.Role of the extracellular and cytoplasmic domains of CD44 in the rolling inter-action of lymphoid cells with hyaluronan under physiologic flow. J. Biol. Chem.278: 11150–11158.

35. McEver, R. P., and R. D. Cummings. 1997. Role of PSGL-1 binding to selectinsin leukocyte recruitment. J. Clin. Invest.100: S97–S103.

36. Wipke, B. T., and P. M. Allen. 2001. Essential role of neutrophils in the initiationand progression of a murine model of rheumatoid arthritis. J. Immunol. 167:1601–1608.

37. Pillinger, M. H., and S. B. Abramson. 1995. The neutrophil in rheumatoid ar-thritis. Rheum. Dis. Clin. North Am. 21: 691–713.

38. Firestein, G. S., and N. J. Zvaifler. 2002. How important are T cells in chronicrheumatoid synovitis? II. T cell-independent mechanisms from beginning to end.Arthritis Rheum. 46: 298–308.

39. Firestein, G. S. 2003. Evolving concepts of rheumatoid arthritis. Nature 423:356–361.

40. Nolte, M. A., A. Hamann, G. Kraal, and R. E. Mebius. 2002. The strict regulationof lymphocyte migration to splenic white pulp does not involve common homingreceptors. Immunology 106: 299–307.

41. Henderson, R. B., L. H. Lim, P. A. Tessier, F. N. Gavins, M. Mathies, M. Perretti,and N. Hogg. 2001. The use of lymphocyte function-associated antigen (LFA)-1-deficient mice to determine the role of LFA-1. Mac-1, and �4 integrin in theinflammatory response of neutrophils. J. Exp. Med. 194: 219–226.

42. Jung, U., C. L. Ramos, D. C. Bullard, and K. Ley. 1998. Gene-targeted micereveal importance of L-selectin-dependent rolling for neutrophil adhesion.Am. J. Physiol. 274: H1785–H1791.

43. Khan, A. I., S. M. Kerfoot, B. Heit, L. Liu, G. Andonegui, B. Ruffell, P. Johnson,and P. Kubes. 2004. Role of CD44 and hyaluronan in neutrophil recruitment.J. Immunol. 173: 7594–7601.

44. Jasin, H. E. 2001. Mechanisms of tissue damage in rheumatoid arthritis. In Ar-thritis and Allied Conditions: A Textbook of Rheumatology. W. J. Koopman, ed.Lippincott Williams & Wilkins, Philadelphia, pp. 1128–1152.

45. Kunkel, E. J., J. L. Dunne, and K. Ley. 2000. Leukocyte arrest during cytokine-dependent inflammation in vivo. J. Immunol. 164: 3301–3308.

46. Wipke, B. T., Z. Wang, J. Kim, T. J. McCarthy, and P. M. Allen. 2002. Dynamicvisualization of a joint-specific autoimmune response through positron emissiontomography. Nat. Immunol. 3: 366–372.

47. Taylor, P. C., A. M. Peters, E. Paleolog, P. T. Chapman, M. J. Elliott,R. McCloskey, M. Feldmann, and R. N. Maini. 2000. Reduction of chemokinelevels and leukocyte traffic to joints by tumor necrosis factor � blockade inpatients with rheumatoid arthritis. Arthritis Rheum. 43: 38–47.

48. Kotzin, B. L. 2005. The role of B cells in the pathogenesis of rheumatoid arthritis.J. Rheumatol. Suppl. 73: 14–18.

1940 CD44 AND CD62L IN PG-INDUCED ARTHRITIS

by guest on June 4, 2018http://w

ww

.jimm

unol.org/D

ownloaded from