of February 3, 2019.This information is current as

Surface Isotypes and Differentiation AntigensCell Subsets Defined by the Expression of Identification of Murine Germinal Center B

Randolph J. Noelle and Thomas J. WaldschmidtStephen M. Shinall, Mercedes Gonzalez-Fernandez,

http://www.jimmunol.org/content/164/11/5729doi: 10.4049/jimmunol.164.11.5729

2000; 164:5729-5738; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/164/11/5729.full#ref-list-1

, 32 of which you can access for free at: cites 71 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 © 2000 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 February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on February 3, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

Identification of Murine Germinal Center B Cell SubsetsDefined by the Expression of Surface Isotypes andDifferentiation Antigens1

Stephen M. Shinall,* Mercedes Gonzalez-Fernandez,† Randolph J. Noelle,† andThomas J. Waldschmidt2*‡

Germinal centers (GCs) are inducible lymphoid microenvironments that support the generation of memory B cells, affinitymaturation, and isotype switching. Previously, phenotypic transitions following in vivo B cell activation have been exploited todiscriminate GC from non-GC B cells in the mouse and to delineate as many as seven distinct human peripheral B cell subsets.To better understand the differentiative processes occurring within murine GCs, we sought to identify subpopulations of GC Bcells corresponding to discrete stages of GC B cell ontogeny. We performed multiparameter flow-cytometric analyses of GC B cellsat consecutive time points following immunization of BALB/c mice with SRBC. We resolved the murine GC compartment intosubsets based on the differential expression of activation markers, surface Ig isotypes, and differentiation Ags. Class-switched andnonswitched GC B cells emerged contemporaneously, and their relative frequencies remained nearly constant throughout the GCreaction, perhaps reflecting the establishment of a steady state. A significant percentage of the nonswitched B cells with a GCphenotype exhibited surface markers associated with naive B cells, including CD23, surface IgD, and high levels of CD38 consistentwith either prolonged recruitment into the GC reaction or protracted expression of these markers during differentiation withinthe GC. Expression of the activation marker BLA-1 was dynamic over time, with all GC B cells being positive early afterimmunization, followed by progressive loss as the GC reaction matured into the second and third week. Implications of theseresults concerning GC evolution are discussed.The Journal of Immunology,2000, 164: 5729–5738.

T hree to 4 days following immunization with a thymus-dependent Ag, an oligoclonal cohort of B cells activatedwithin or along the borders of the T cell areas of second-

ary lymphoid tissue colonizes adjacent B cell follicles, formingstructures termed germinal centers (GCs).3 GCs are inducible lym-phoid microenvironments comprised primarily of Ag-specific Bcells, Ag-specific CD41 Th cells, follicular dendritic cells, andmacrophages, and are the principal anatomic sites associated with:1) expansion of Ag-specific B cell clones; 2) affinity maturation viasomatic hypermutation and subsequent Ag-driven selection; 3) ap-optosis of B cells failing this stringent selection process; 4) gen-eration of memory B cells (1–8).

The GC reaction is critically dependent upon cognate interac-tions between Ag-specific B and T cells, and represents one of twodevelopmental pathways to which activated B cells commit duringT-dependent responses, the other being the differentiation to Ab-secreting plasma cells. Despite the accumulating data regardingmolecular requirements for GC formation, such as the critical role

for CD40-CD40L interactions (9) and the developmental require-ments for TNF-a, TNFR1, LT-a, LT-b, and LT-bR (10–14),much remains unclear regarding the processes associated withGCs. Little is known regarding the mechanisms that regulate com-mitment of B cells to the GC, somatic hypermutation, and differ-entiation to the memory cell or plasma cell stage. Although currentinvestigation is beginning to reveal candidate molecules that par-ticipate in murine GC regulation and terminal differentiation, suchas OX40-OX40L and the transcription factor Blimp-1 (15–18), amore detailed understanding of the various stages of GC matura-tion is needed. An initial step toward this goal is the phenotypicand functional characterization of GC B cell subsets.

During primary B lymphopoiesis, committed precursors progressthrough successive stages of maturation and selection, a process ac-companied by the acquisition of unique cell surface phenotypes de-terminable by multiparameter flow cytometry (19–22). Indeed, thephenotypic characterization of distinct B cell precursor populationshas greatly facilitated subsequent investigations of genes required foror regulating B lineage commitment and differentiation. Exploitingchanges in cell surface phenotype following peripheral B cell activa-tion, Pascual et al. (23) similarly utilized flow cytometry to identifyphenotypes potentially corresponding to sequential stages of periph-eral B cell differentiation in the human. Subsequent isolation and mo-lecular and functional analysis have suggested developmental rela-tionships between these subsets and permitted the construction of alinear model of Ag-driven B cell differentiation (4). In addition, recentin vitro studies have partially recapitulated Ag-induced developmentalstages associated with the GC (24).

In the mouse, several groups have employed flow cytometryand/or immunohistology to describe phenotypic characteristicsdistinguishing GC B cells from other B cell subsets (25–28). Initialstudies examining GC B cell phenotypes elicited during primary or

*Interdisciplinary Graduate Program in Immunology and‡Department of Pathology,University of Iowa College of Medicine, Iowa City, IA 52242; and†Department ofMicrobiology, Dartmouth Medical School, Lebanon, NH 03756

Received for publication June 21, 1999. Accepted for publication March 14, 2000.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisementin accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by National Institutes of Health Grant AI 31265.2 Address correspondence and reprint requests to Dr. Thomas J. Waldschmidt, De-partment of Pathology, University of Iowa College of Medicine, Iowa City, IA 52242.E-mail address: thomas-waldschmidt@uiowa.edu3 Abbreviations used in this paper: GC, germinal center; AFC, Ab-forming cell; BSS,balanced salt solution; HSA, heat-stable Ag; LT, lymphotoxin; NP, (4-hydroxy-3-nitrophenyl)acetyl; PALS, periarteriolar lymphatic sheaths; PNA, peanut agglutinin;SB, staining buffer; sIg, surface Ig.

Copyright © 2000 by The American Association of Immunologists 0022-1767/00/$02.00

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

secondary humoral response utilized heterologous erythrocytes(SRBC) as Ag (25). More recently, the phenotype of GC B cellsevoked during the genetically restricted response of C57BL/6 miceto the hapten (4-hydroxy-3-nitrophenyl)acetyl (NP) has been par-tially characterized (27, 28). However, no report explicitly inves-tigating the ontogeny and distribution of selected activation Agsand surface Ig (sIg) isotypes within the GC B cell compartment hasbeen published. Such a study would have the potential to identifydiscrete GC B cell subsets that possibly correspond to distinctstages of GC B cell ontogeny. The SRBC system, previously ex-ploited by Nieuwenhuis and colleagues (8) to determine the oli-goclonality of the GC reaction in rats and Kraal et al. (25) topartially examine murine GC B cell phenotypes, was selected forstudy due to its potential to induce robust polyclonal GC responsesin an adjuvant-independent manner.

This work presents multiparameter flow-cytometric data resolv-ing GC B cells (defined as B2201PNAhigh splenic mononuclearcells from immunized mice) into discrete subsets based on thedifferential expression of several cell surface markers, includingBLA-1, sIgM, sIgD, sIgG1, CD23, and CD38. Examination of sIgheavy and light expression by GC B cells revealed that isotypeswitching preceded or was concomitant with GC formation, estab-lished IgM as the predominant isotype expressed by GC B cells,and revealed an apparent steady state between switched and non-switched GC B cells. B2201PNAhigh B cells defined by sIgD,CD38, and CD23 expression provided evidence for protracted re-cruitment into GCs throughout the first 2–3 wk of the primaryresponse, revealing populations potentially corresponding to earlystages following commitment to, or differentiation within, the GCreaction. Expression of the activation Ag BLA-1 was dynamicduring the GC response, and GC B cell subsets defined by BLA-1expression or absence correlated with the inductive, established,and dissociative phases of the GC response. Further phenotypicheterogeneity within the GC B cell compartment was revealed byfour-color flow cytometry, wherein B2201PNAhigh cells were si-multaneously examined for the expression of combinations of theaforementioned surface determinants. Implications of these resultson the further study and characterization of GC B cells arediscussed.

Materials and MethodsMice

Eight- to 16-wk-old female BALB/c mice were obtained from CharlesRiver Laboratories (Wilmington, MA) and housed in a specific pathogen-free facility. CD40L2/2 mice were bred in our colony from breeding pairskindly supplied by Dr. Jacques Peschon (Immunex, Seattle, WA).TNFR12/2 mice were provided by Dr. Charles Lutz (University of Iowa,Iowa City, IA). Following immunization, animals were housed in a cleanhousing facility until sacrificed on specific days postimmunization.

Abs for flow cytometry

The following reagents were used during the flow-cytometric studies:RA3-6B2, a rat IgG anti-mouse B220; 1D3, a rat IgG anti-mouse CD19;clone 90, a rat IgG anti-mouse CD38; Jo2, a hamster IgG anti-mouse CD95(PharMingen, San Diego, CA); 53-10.1, a rat IgG anti-mouse BLA-1;M1/69, a rat IgG anti-mouse CD24(heat-stable Ag (HSA)); 1C10, ratIgG anti-mouse CD40; CY34.1.2, mouse IgG anti-mouse CD22; 7E9,rat IgG anti-mouse CD21/CD35;PS/2, rat IgG anti-mouse VLA-4; F441,rat IgG anti-mouse LFA-1; YN1, rat IgG anti-mouse ICAM-1; b-7-6, ratIgG anti-mouse IgM; 11-26, rat IgG anti-mouse IgD; B3B4, a rat IgGanti-mouse CD23; GL-1, rat IgG anti-mouse CD86; GL-7, a rat IgMdirected against an undefined Ag; S7, a rat IgG anti-mouse CD43; M5114,a rat IgG anti-mouse MHC II; 281-2, a rat anti-mouse Syndecan 1 (Phar-Mingen); polyclonal goat anti-mousek light chain; polyclonal goat anti-mouse IgG1; PNA, specific for terminal galactosyl (b-1,3) N-acetylgalac-toseamine residues. Biotin-conjugated polyclonal goat anti-mouse IgG1,PE-conjugated polyclonal goat anti-mousek light chain, and PE-strepta-vidin were purchased from Southern Biotechnology Associates (Birming-

ham, AL). Chromatographically purified rat and goat IgG (Jackson Immu-noResearch, West Grove, PA) were used as isotype controls. FITC-PNAand FITC-streptavidin were purchased from Vector Laboratories (Burlin-game, CA). Texas Red-streptavidin was purchased from Leinco Technol-ogies (Ballwin, MO) and The Jackson Laboratory (Bar Harbor, ME). mAbswere semipurified from HB101 serum-free supernatants by 50% ammo-nium sulfate precipitation and conjugated to biotin, PE, Texas Red, orCyanine 5.18 using standard procedures. Cyanine 5.18-streptavidin wasprepared in our laboratory using standard procedures.

Immunizations

Mice were injected once i.p. with 0.2 ml of 10% v/v SRBC (ColoradoSerum Company, Denver, CO) in balanced salt solution (BSS; equivalentto 1–53 108 SRBC) or 25mg of TNP-Ficoll (Solid Phase Sciences, SanRafael, CA). Splenocytes were examined at 4, 6, 8, 10, 12, 14, 16, 18, 20,22, 24, and 26 days postimmunization.

Cell preparation

On the selected days postimmunization, animals were anesthetized withMetofane (methoxyfluorane) and animals sacrificed by cervical dislocation.Spleens were removed and cell suspensions prepared by pressing the ex-cised spleens between the frosted ends of two glass slides, followed bysuspension in BSS. Cells were washed by centrifugation at 1500 rpm for 7min at 4°C and resuspended in staining buffer (SB; BSS supplemented with5% bovine calf serum and 0.1% sodium azide). Mononuclear cells wereisolated by density centrifugation over FicoLite-LM (Atlanta Biologicals,Norcross, GA) for 20 min at 2500 rpm at room temperature. Cells at theinterface were collected and resuspended in BSS and again washed bycentrifugation at 1500 rpm for 7 min at 4°C. Viable cells were enumeratedby hemocytometer and resuspended at a final concentration of 503 106

cells/ml in SB.

Flow-cytometric analysis

A total of 5 3 105 splenocytes in SB was added to the appropriate mixtureof biotinylated, FITC-, PE-, cyanine 5.18-, or Texas Red-conjugated Abs,and fluorochrome-conjugated PNA in the presence of 25mg 2.4G2 (anti-FcgR) and 10ml normal rat serum to minimize Fc receptor-dependentbackground staining. Cells were incubated on ice for 20 min, then washedtwice with SB. Biotinylated reagents were visualized by the addition ofappropriately conjugated streptavidin. After 20 min on ice, cells were againwashed twice with SB, suspended in fixative (1% formaldehyde in 1.253PBS), and run on a FACSVantage flow cytometer (Becton Dickinson,Mountain View, CA) featuring a primary argon ion laser and a rhodamine6G CR559 dye head laser (Coherent, Palo Alto, CA) pumped by a secondargon ion laser. A minimum of 30,000 events was collected per sample.Low angle and orthogonal light scatter were used to exclude dead cells anddebris, and electronic compensation was utilized to correct for spectraloverlap between FITC/PE and Texas Red/cyanine 5.18. Data were storedon a VAX station 3200 computer equipped with DESK FACS analysissoftware (kindly provided by Wayne Moore, Stanford University, Stanford,CA). Two-color contours are represented as 2% probability plots. Two-color contours from four-color flow-cytometric studies are represented as5% probability plots. In those experiments in which goat anti-k light chainor goat anti-IgG1 Abs were utilized, it was necessary to first incubate cellswith these reagents in the absence of rat or mouse Abs and blocking agents.Cells were washed after treatment with the goat Abs, followed by additionof blocking agents, and stained as described above. To facilitate analysisand comparison of populations defined by the expression or absence ofparticular surface determinants, data for each time point presented in Figs.3A, 4A, and 5Awere derived from the same mouse.

Confocal microscopy and image analysis of fresh splenicsections

Confocal microscopic analysis was performed as described previously(29). Briefly, spleens from unimmunized or SRBC-immunized animalswere harvested and embedded in 4% agarose in HBSS and sectioned in avibrating microtome (Microcut H12; Energy Beam Sciences, Agawam,MA), and 500-mm sections were incubated overnight at 4°C with FITC-conjugated anti-IgD (11–26) and cyanine 5.18-conjugated PNA (both usedat a concentration of 10–20mg/ml in PBS/1% BSA/0.02% sodium azide).Mouse IgG was included to minimize Fc-dependent background staining.After washing, the samples were mounted in slides with PBS/azide/10%glycerol and analyzed on the confocal microscope (Bio-Rad 1024; Bio-Rad, Hercules, CA).

5730 MURINE GERMINAL CENTER B CELL SUBSETS

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

ResultsKinetics of GC B cell induction

Using multiparameter flow cytometry, we identified and enumer-ated GC B cells 4–26 days following primary immunization ofBALB/c mice with SRBC. B2201PNAhigh cells, previously de-fined by flow cytometry and immunofluorescence as GC B cells(25, 26), were apparent by day 4 postimmunization, reached a peakfrequency by days 6–8, and then decreased to one-half this max-imum frequency between days 12 and 14 postimmunization (Fig.1). B2201PNAhigh cells could still be identified into the third andfourth week postimmunization, but by day 42, only a residualB2201PNAhigh population was detected (Fig. 1, and data notshown). This kinetic profile is consistent with previous immuno-histologic, stathmokinetic, and flow-cytometric analyses of GCevolution (28, 30–32). These results also demonstrate the robustnature of the SRBC response and the potential to analyze GC Bcells in detail.

To confirm the B2201PNAhigh cells as an exclusive GC B cellpopulation, we further defined the conditions under which cells ofthis phenotype could be elicited. GCs are Ag inducible, T celldependent, and are absent in mice rendered deficient in CD40L(33) or TNFR1 (34). Correspondingly, the B2201PNAhigh cellswere observed in mice immunized 6 days previously with SRBC inBSS, but not with BSS alone nor with TNP-Ficoll, a TI-2 Aginducing few or no GCs (35), confirming both the Ag and T celldependence of the B2201PNAhigh population (Fig. 1, and data notshown). Likewise, B2201PNAhigh cells were undetectable follow-ing SRBC immunization of mice lacking CD40L or TNFR1 (datanot shown).

To further substantiate the identity of the B2201PNAhigh cells,we more thoroughly characterized the phenotype of this popula-tion. The B2201PNAhigh cells were determined to consist of ma-ture B cells because they uniformly expressed the B cell lineage

markers CD19, CD21/35, CD22, and CD40, excluding contami-nation by either non-B cells or immature B cells (Fig. 2 and TableI). In addition, and consistent with earlier findings (25, 28), theB2201PNAhigh population was not enriched for Ab-forming cells(AFCs), as demonstrated by the virtual absence of AFC-associatedmarkers such as CD43 and Syndecan-1 (CD138) (Table I). Fur-thermore, as shown in Fig. 2, the majority of B2201PNAhigh cellsexpressed heightened levels of CD95 (Fas) and the Ag recognizedby the mAb GL-7, as has been previously reported for GC B cells(5, 36). Although GL-7 expression roughly overlapped with highavidity for PNA, GL-7 expression was heterogeneous within theB2201PNAhigh population, and a population of B2201PNAlow

cells expressing GL-7 was detected (Fig. 2). The significance ofGL-7 heterogeneity within the GC compartment, also apparent inimmunohistochemical analyses of GCs (37), is unclear at this time.Collectively, these data demonstrate that B2201PNAhigh cells are

FIGURE 2. B2201PNAhigh cells are B cells that exhibit the phenotypicattributes of GC B cells. At day 8 postimmunization, splenic mononuclearcells were stained with cyanine 5.18 anti-B220, FITC-PNA, and biotin-conjugated Abs specific for the indicated surface determinants, followed byPE-streptavidin. The contour plots are derived from B2201-gated spleno-cytes and depict the correlated expression of CD19, CD95, and GL-7 withPNA. Identical results were obtained on days 4–14 (data not shown). Thenumbers within the GL-7 contour plot are the percentages of GL-71 andGL-72 cells within the B2201PNAhigh population, and are representativeof three independent experiments. The range of percentage ofB2201PNAhigh cells that were GL-71 at that time point was 64–71%.

Table I. Differential marker expression on germinal B cell subsetsa

Follicular Early GCb Late GCb

IgM Intermediate Bimodalc BimodalKappa 1 1 1IgG1 2 Bimodal BimodalBLA-1 2/low High BimodalIgD High Bimodal BimodalCD23 1 Bimodal BimodalCD38 High Bimodal BimodalGL-7 2 1 1B7.2 (CD86) 2 1 1/bimodalCD19 1 1 1CD21/35 1 1 1CD22 1 1 1CD40 1 1 1Fas (CD95) 2 1 1HSA (CD24) Intermediate Bright BrightLFA-1 (CD11a) Intermediate Intermediate IntermediateVLA-4 (CD49d/CD29) Intermediate Low LowICAM-1 (CD54) Intermediate Intermediate IntermediateSyndecan-1 (CD138) 2 2 2CD43 2 2 2MHC II Intermediate High High

a GC B cells were defined as B2201PNAhigh splenic mononuclear cells.b Early GCs are those present 4–6 days postimmunization; late GCs are those

present 8–14 days postimmunization.c Indicates that populations both expressing and lacking the given surface deter-

minant were observed.

FIGURE 1. Kinetics of the B2201PNAhigh population elicited by SRBCimmunization. On the indicated days postimmunization, splenocytes fromanimals challenged with SRBC were spun through FicoLite-LM, andmononuclear cells stained with cyanine 5.18-anti-B220 and FITC-PNA.Numbers within the contour plots are the percentages of viable mononu-clear cells with a B2201PNAhigh phenotype observed at that given timepoint and are representative of three to four independent experiments. Therange of frequencies of B2201PNAhigh for each time point is as follows:day 4 (4.2–5.8%), day 6 (5.8–9.5%), day 8 (6.3–9.6%), day 10 (3.5–6.7%),day 12 (3.5–5.9%), day 14 (2.9–4.6%), day 16 (1.9–2.9%), day 18 (2.1–4.0%), day 20 (1.8–3.7%), day 22 (1.7–3.6%), day 24 (1–2%).

5731The Journal of Immunology

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

mature B cells present only when GC formation occurs and whichpossess the phenotypic attributes of GC B cells, but not AFCs.

Delineation of GC B cell subsets

Having established a reproducible system for induction of GC re-sponses, we further analyzed the phenotype of murine GC B cellsby investigating the expression and distribution of selected cellsurface markers within the GC B cell compartment at consecutivetime points following primary immunization. Cohorts of micewere challenged with SRBC, and 4–18 days following immuni-zation, splenic mononuclear cells were subjected to multiparam-eter flow-cytometric analysis in which mAbs specific for one ortwo of the surface markers listed in Table I were used in conjunc-tion with anti-B220 mAb and fluorochrome-conjugated PNA todelineate GC B cells. As shown in Fig. 2 and summarized in TableI, GC B cells prominently expressed CD19, CD21/35, CD22,CD40, and CD95. In addition, GC B cells also were found toexpress high levels of HSA (CD24) and class II MHC molecules,intermediate levels of CD11a and CD54, and low levels of VLA-4(Table I). However, these proteins were uniformly expressed byGC B cells on all days examined (data not shown), and discrete GCB cell subpopulations defined by the presence or absence of thesesurface markers were therefore not evident. In contrast, GC B cell

subsets defined by the expression of sIg isotypes and activationmarkers were readily detected.

GC B cell subsets delineated by IgM or IgG1 expression

Because GCs are associated with isotype switching (reviewed inRef. 38; see also Refs. 30, 32, and 39), we evaluated the distribu-tion and kinetics of sIgM, sIgG1, and sIgG2a expression followingprimary antigenic challenge. In Fig. 3A, sIgM1 GC B cells weredetected at day 4 and represented approximately two-thirds of theGC B cell compartment, with one-third being sIgM2. Correspond-ingly, approximately one-fourth to one-third of the GC B cell pop-ulation was sIgG11.

Surprisingly, despite the expansion and subsequent diminutionof the GC B cell compartment over time, the percentage of totalGC B cells expressing or lacking either sIgM or sIgG1 remainednearly constant throughout the primary GC reaction, even 3–4 wkpostimmunization (Fig. 3A). In contrast, and characteristic of theBALB/c response, sIgG2a1 GC B cells were infrequent upon

FIGURE 3. The relative frequencies of sIgM1 and sIgG11 GC B cellsremain constant throughout the primary GC response.A, Splenic mononu-clear cells were isolated 4–18 days postimmunization and stained withcyanine 5.18 anti-B220, FITC-PNA, and biotin-conjugated anti-IgM orgoat anti-mouse IgG1, followed by PE-streptavidin. The contour plotsshow the correlated expression of sIgM or sIgG1 and PNA by B2201-gatedcells. B, Cells were stained with Texas Red anti-B220, FITC-PNA, PE-conjugated anti-IgM, and biotin anti-mouse IgG1, followed by cyanine5.18-streptavidin. The contour plots show the correlated expression ofsIgM and sIgG1 by B2201PNAhigh-gated cells. The numbers within thecontour plots inA represent the percentages of total B2201PNAhigh cellsdetermined to be positive or negative for sIgM or sIgG1, and are repre-sentative of three to four independent experiments. Inflections betweenpopulations revealed by anti-IgM or anti-mouse IgG1 staining served as thecutoff for designation of cells as either positive or negative for the respec-tive marker. Ranges of percent positives for sIgM were as follows: day 4(55–82%), day 8 (55–82%), day 12 (56–72%), day 18 (58–74%). Rangeof percent positives for sIgG1 were as follows: day 4 (20–29%), day 8(25–34%), day 12 (22–26%), day 18 (12–22%).

FIGURE 4. BLA-1 expression is dynamic throughout the primary GCresponse and with sIgM resolves the GC compartment into four subsets.A,Splenic mononuclear cells were isolated 4–14 days postimmunization andstained with cyanine 5.18 anti-B220, FITC-PNA, and biotin anti-BLA-1,followed by PE-streptavidin. The contour plots show the correlated ex-pression of BLA-1 and PNA by B2201-gated cells. Numbers within thecontour plots represent the average percentages of total B2201PNAhigh

cells determined to be BLA-11 or BLA-12 (n 5 4). The crosshairs rep-resent the cutoff for the determination of BLA-11 vs BLA-12 based onstaining with isotype controls. The range of percent positives for each timepoint is as follows: day 4 (88–94%), day 6 (84–87%), day 8 (58–85%),day 10 (57–73%), day 12 (26–60%), day 14 (42–62%).B, Splenic mono-nuclear cells were stained with Texas Red anti-B220, FITC-PNA, PE-BLA-1, and biotin anti-IgM, followed by cyanine 5.18-streptavidin. Con-tour plots indicate correlated BLA-1 vs sIgM expression within theB2201PNAhigh-gated population at days 8, 12, and 18. Numbers in thecontour plots are the average percentages of the respective GC B cell sub-sets within the total B2201PNAhigh population (n5 4): (clockwise fromthe upper right) BLA-11sIgM1, BLA-12sIgM1, BLA-12sIgM2,BLA-11sIgM2.

5732 MURINE GERMINAL CENTER B CELL SUBSETS

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

SRBC immunization, and not always detected (data not shown).Using four-color flow cytometry to simultaneously examine ex-pression of sIgM and sIgG1, a mutually exclusive pattern of sIgMand sIgG staining was observed on days 4, 8, 12, and 18 (Fig. 3B).This demonstrates the sIgG11 cells to be bona fide class-switchedGC B cells, and not cells with Fc receptor-bound immune com-plexes. In addition, the relative frequencies of sIgM1 and sIgG11

cells did not change by treating cells with a low pH buffer beforestaining (data not shown). Although IgG1 is the major target ofisotype switching upon SRBC challenge in BALB/c mice, the re-sults in Fig. 3Bindicate switching to other isotypes. Because vir-tually all of the GC B cells at all time points werek light chainpositive (data not shown) (26, 40), the sIgM, sIgG1 negative cellsare likely to represent infrequent sIgG21, sIgA1, and sIgE1 cells.

The contemporaneous emergence of sIgM1 and sIgG11 GC Bcells, as evidenced by our flow-cytometric analyses, suggests thatthe initiation of isotype switching occurs very early after, concom-itant with, or even before the induction of the GC reaction follow-ing SRBC immunization. Intriguingly, there was no progressiveconversion, as measured by frequency, from sIgM1 to class-switched GC B cells during the primary response. Furthermore, thedata presented in this work corroborate earlier immunohistochem-ical and flow-cytometric studies indicating that class switchingprecedes somatic hypermutation within the murine GC, a processthat is not discernible by current methodologies until 6–8 daysfollowing immunization (30, 32). This is in apparent contrast to thehuman GC reaction, as Liu and colleagues have recently providedevidence that a subset of human GC B cells undergoes classswitching, and that this occurs subsequent to the initiation of so-matic hypermutation (41, 42).

BLA-1 expression is dynamic and correlates with distinct phasesof the GC reaction

The uncharacterized 53-kDa cell surface determinant BLA-1 waspreviously reported to be expressed by GC B cells (27, 43). Inter-estingly, BLA-11 cells were capable of adoptively transferringmemory responses 6 days, but not 6 wk following immunization(43), and hapten-specific, sIgG11 B cells were found to expressBLA-1 at day 6, but not at day 24 postimmunization (27). To-gether, these data indicate that BLA-1 expression is modulatedduring the evolution of the GC reaction. We therefore studied thekinetics and distribution of BLA-1 expression on GC B cells. InFig. 4A, the vast majority of GC B cells were BLA-11 at 4 dayspostimmunization, consistent with previous phenotypic analyses.By day 8, however, BLA-1 expression became heterogeneous; ap-proximately two-thirds of GC B cells were BLA-11, but one-thirdwas now BLA-12. The frequency of BLA-11 GC B cells contin-ued to decrease until by day 12 approximately one-half of the GCB cells were BLA-11 and the other half BLA-12, a pattern alsoevident during the third week postimmunization (Fig. 4B). By24–26 days following challenge, little or no surface BLA-1 waspresent on GC B cells (data not shown).Therefore, the pattern ofBLA-1 expression closely correlated with the distinct phases of GCevolution, with uniform expression typical of the inductive phase,heterogeneous expression characteristic of the established phase, andrelative absence representative of the dissociative phase.

To further examine the distribution of BLA-1 within the GC B cellpopulation, we evaluated the expression of BLA-1 with respect to sIgMamong GC B cells at 8, 12, and 18 days postimmunization. In Fig. 4B,sIgM and BLA-1 expression by B2201PNAhigh-gated cells defined foursubsets at day 8 postimmunization. The predominant GC B cell subset atthis time expressed both sIgM and BLA-1 and represented approx-imately one-half of the total GC B cell population and three-quar-ters of sIgM1 GC B cells. A prominent sIgM2 BLA-11 popula-

tion was also evident at day 8, accounting for nearly one-third oftotal GC B cells and greater than one-half of sIgM2 GC B cells.Two minor subpopulations, namely sIgM1 BLA-12 and sIgM2

BLA-12, each accounted for less than one-fifth of total GC B cellsat day 8. The frequency of BLA-11 GC B cells decreased betweendays 8 and 12 postimmunization, and this reduction was observedin both the sIgM1 and sIgM2 GC B cell populations. The relativefrequencies of sIgM1 and sIgM2 GC B cells expressing BLA-1observed at day 12 were maintained as late as day 18 (Fig. 4B).

FIGURE 5. Subsets of B2201PNAhigh cells defined by the expressionof CD38, CD23, and sIgD may correspond to early intermediates of GC Bcell ontogeny.A, Splenic mononuclear cells on days 4, 8, 12, and 18postimmunization were stained with cyanine 5.18-B220, FITC-PNA, bi-otin anti-CD38, biotin anti-CD23, or biotin anti-IgD, followed by PE-streptavidin. The contour plots show the correlated expression of thesemarkers and PNA by B2201-gated cells. Inflections between the popula-tions revealed by CD38 and sIgD staining served as the cutoff for deter-mination of CD38high and CD38low or sIgD1 and sIgD2, respectively. Thecrosshairs within the CD23 contour plots represent the cutoff for determin-ing CD231 and CD232 based on staining with isotype controls.B andC,Splenic mononuclear cells at 4, 8, 12, and 18 days postimmunization werestained with Texas Red anti-B220, FITC-PNA, and either cyanine 5.18anti-IgD and biotin anti-CD23 followed by PE-streptavidin (B) or PE anti-CD38 and biotin anti-CD23 followed by cyanine 5.18-streptavidin (C).Contour plots represent the correlated expression of sIgD and CD23 (B) orCD23 and CD38 (C) by B2201PNAhigh-gated cells. Data presented inAwere derived from the same experiment and are representative of at leastfour experiments at each time point. Ranges of percent CD38high cells are:day 8 (39–66%), day 12 (33–34%), day 18 (27–54%). Ranges of percentpositives for CD23 are: day 4 (61–72%), day 8 (58–69%), day 12 (62–70%), day 18 (42–65%). Ranges of percent positives for sIgD expressionare: day 4 (17–27%), day 8 (14–41%), day 12 (08–35%), day 18(14–19%).

5733The Journal of Immunology

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

These data indicate that BLA-1 expression is not restricted tosIgM1 GC B cells and that BLA-1 expression decreases as a func-tion of time among both sIgM1 and sIgM2 GC B cell subsets.

Subsets of B2201PNAhigh cells revealed by the expression ofsIgD, CD23, and CD38

It is largely accepted that most GC B cells are derived from maturenaive B cells expressing sIgM and sIgD (5). The vast majority ofGC B cells, however, have been reported to be sIgD2, includingthose bearing sIgM (8, 25, 26). sIgD1 GC B cells from humantonsils have been previously described (44, 45) and more recentlyisolated and partially characterized (46–49). GC B cells express-ing both sIgM and sIgD most likely represent a transitional stagebetween follicular mantle and GC B cells and may correspond toputative GC founder cells (48). We sought to identify such GCfounder cells by examining within the GC B cell compartment theexpression and distribution of molecules associated with naive Bcells, namely sIgD, CD23, and CD38. CD23 is expressed late in Bcell ontogeny, but is absent on isotype-switched B cells (50–52).Expression of CD23 on GC B cells, as tested by immunohisto-chemical means, has been difficult to determine due to robust ex-pression of this molecule on follicular dendritic cells (53, 54).CD38 is expressed by naive B cells, down-regulated by GC Bcells, and subsequently up-regulated by memory B cells (55–58).

As seen in Fig. 5A, subsets of B2201PNAhigh cells bothexpressing and lacking CD38, CD23, and sIgD were observedfollowing SRBC immunization. At day 4, B2201PNAhigh cellsuniformly expressed CD38 (albeit at lower levels than onB2201PNAlow cells), more than one-half were CD231, and nearlyone-fifth were sIgD1. By day 8, a B2201PNAhigh population ex-pressing high levels of CD38 could be identified, and the emer-gence of this prominent subset coincided with an increase in fre-quency of sIgD1 cells, suggesting that the CD38high cells alsoexpressed sIgD. By days 12 and 18 of the primary response, theCD38high subset was less numerous, representing approximatelyone-third of the B2201PNAhigh population. These data are thusconsistent with previous reports demonstrating diminished CD38expression on GC B cells following immunization (57, 58). Al-though the sIgD1 subset had also decreased in relative frequencyat these later time points, approximately one-half or more of theB2201PNAhigh cells remained CD231. Because B2201PNAhigh

cells expressing CD23 were less numerous than those bearingsIgM, but more numerous than cells displaying sIgD or CD38 (seeFigs. 3Aand 5A), we speculated that the loss of sIgD and CD38precedes that of CD23 within the sIgM1 GC B cell population. Wefurther speculated that simultaneous evaluation of CD23 and sIgDor CD38 would identify discrete or overlapping populations pos-sibly representing early intermediates in GC B cell ontogeny. Thiswas confirmed in Fig. 5B, in which sIgD1B2201PNAhigh cellswere found to express high levels of CD23. A CD231sIgD2 pop-ulation was also observed, in accordance with the data in Fig. 5A,and suggesting that the loss of sIgD precedes that of CD23. ThesIgD1B2201PNAhigh cells were also found to be CD38high andsIgM1 (data not shown). In Fig. 5C, simultaneous examination ofCD23 and CD38 expression revealed a CD38high population thatsimilarly expressed high levels of CD23 at most time points. Al-though CD38 expression within the B2201PNAhigh compartmentwas not bimodal at day 4, a subset of CD23high cells expressingelevated levels of CD38 was detected. Fig. 5C also demonstrates asmall CD231CD38low population, again indicating transition ofGC B cells to a CD38low phenotype precedes loss of CD23. To-gether, these data indicate that the acquisition of PNA receptorscan occur before the loss of sIgD, CD23, and CD38, and the re-sulting sIgM1sIgD1CD231CD38highB2201PNAhigh cells may

represent the earliest determinable stage of commitment to the GCreaction. This early stage is most likely followed by sequential lossof IgD and CD38, leaving a sIgM1sIgD2CD231CD38low inter-mediate stage GC B cell. It is also of interest that both thesIgM1sIgD1CD231CD38high and sIgM1sIgD2CD231CD38low

GC subsets are present at all time points examined, suggesting thatfounder cells are continually recruited into the GC reaction ormaintained with a founder phenotype once the GC has beenestablished.

Confocal microscopic analysis of B2201PNAhigh cellsexpressing IgD

As the IgD1B2201PNAhigh cells were an unexpected finding, es-pecially at the later time points, we further explored the location ofthese cells relative to GCs. Specifically, we used two-color con-focal microscopy of fresh splenic sections to determine whetherthese cells were within or outside of established GCs. This tech-nology was chosen because it leaves sensitive epitopes intact andgenerally allows for greater sensitivity compared with other tech-niques. Spleens were harvested from animals immunized 8 dayspreviously with SRBC (corresponding to the peak of the primaryGC reaction and the time at which the IgD1B2201PNAhigh cellswere most numerous in the previous flow-cytometric experiments).Splenic sections were then stained with FITC-conjugated anti-IgD(green) and cyanine 5.18-conjugated PNA (blue) to delineate bothIgD1 cells and GCs. In spleens from unimmunized control mice,IgD1 cells were limited to the follicles, and PNA staining exhib-ited a background reticular pattern consistent with quiescent spleen(data not shown). In a day 8 immunized spleen, however, PNA1

GCs were readily apparent. As illustrated in Fig. 6, IgD1PNA1

double staining cells were also evident (appearing as turquoisecells), and surprisingly, could be found in several follicular sites.In addition to a small number residing within the GC, clusters ofIgD1PNA1 cells were observed in the mantle both at the GCmantle interface and in areas more distant from the GC. This wasa consistent finding in three separate immunized mice (data notshown). Thus, while IgD1PNA1 cells were indeed part of the GC,a significant fraction of this population was found outside of the

FIGURE 6. Two-color confocal microscopic analysis of IgD1PNAhigh

cells. Spleens from mice immunized 8 days earlier with SRBC were sec-tioned and stained with FITC anti-IgD and cyanine 5.18-PNA. Dark blue(PNA1) areas represent the GC, and green (IgD1) areas represent the fol-licular mantle. IgD1PNA1 double staining cells are seen as turquoise cellsor clusters. Results are representative of three immunized spleensexamined.

5734 MURINE GERMINAL CENTER B CELL SUBSETS

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

GC. Although the meaning of this result is presently unclear, thefinding of non-GC IgD1PNA1 cells may be consistent with IgD1

cells being continually recruited into the GC reaction (foundercells) or IgD1 cells exiting the GC as post-GC or memory B cells.

Kinetics and phenotype of GC B cells 3 and 4 wk followingantigenic challenge

We subsequently extended our phenotypic analysis of GC B cellsto include the relatively uncharacterized late phase of the GC re-sponse, primarily to determine whether the GC B cell subsets thatwere established early following immunization persisted through-out the response. As shown in Fig. 1, B2201PNAhigh cells werestill consistently detected 16–26 days postimmunization. Interest-ingly, and despite the reduction in the overall frequency of GC Bcells, the subsets defined by sIgM, CD23, and CD38 expressionwere still present and in the relative frequencies observed upontheir establishment during the inductive phase (data not shown).sIgD expression was less evident at these times with between 10and 15% of total B2201PNAhigh cells expressing sIgD (data notshown). In addition, fewer GC B cells expressed BLA-1 at theselater time points, although 20–30% of GC B cells were BLA-11

as late as days 22–24 (data not shown).

DiscussionSimilar to primary B lymphopoiesis in the bone marrow, Ag-driven peripheral B cell maturation in the mouse is accompaniedby profound changes in cell surface phenotype. In peripheral lym-phoid organs, B cells committed to differentiate into plasma cellsdown-regulate certain B cell-specific markers such as sIg andB220, but acquire cell surface determinants such as syndecan-1and PC-1 (27, 32, 59–61). Likewise, B cells participating in theGC reaction are distinguishable from other B cell subsets due totheir high avidity for PNA (25, 26, 62). Previous phenotypic stud-ies have provided significant information regarding the moleculesexpressed by murine GC B cells and have enabled investigators tounambiguously identify this B cell population (25, 26); however, adetailed evaluation of GC B cell phenotypes at numerous timepoints following immunization has not been described. Such anexamination would have the potential to delineate populations cor-responding to distinct stages of GC B cell ontogeny, from GCfounder cell to memory or plasma cell precursor, wherein the var-ious processes associated with the GC (e.g., somatic hypermuta-tion, isotype switching, and differentiation to memory B cells orplasma cells) are thought to occur.

The multiparameter flow-cytometric data presented in this studyprovide a comprehensive phenotypic analysis of the murine GC Bcell compartment at consecutive intervals following primary im-munization of BALB/c mice with SRBC. We defined GC B cellsas B2201PNAhigh splenic mononuclear cells, and the validity ofthis combination of surface determinants was verified in severalways. In addition to demonstrating this population’s Ag, T cell,and CD40 dependence, B2201PNAhigh cells were found to becomprised exclusively of mature B lineage cells that expressedmarkers previously ascribed to GC B cells, but were devoid ofmarkers associated with AFCs (Figs. 1 and 2, Table I, and data notshown). We subsequently resolved the GC B cell population intodiscrete subsets based on the expression and distribution of variousIg isotypes (sIgM, sIgD, and sIgG1), the activation marker BLA-1,and differentiation Ags (CD23, CD38).

sIgM was found to be expressed by more than one-half of GC Bcells isolated 4–26 days following SRBC immunization (Fig. 3A,data not shown). sIgG11 cells were also evident at day 4 postim-munization, and represented between one-quarter to one-third of

GC B cells at that and all subsequent time points (Fig. 3A). ThesIgG11 cells accounted for the majority of sIgM2 B cells, and fewbona fide sIg2 cells were observed at any time postimmunization(Fig. 3, data not shown). Strikingly, although the frequency of thetotal GC B cell population waxed and waned, the relative frequen-cies of sIgM1 and sIgG11 GC B cells remained nearly constant,possibly reflecting the establishment of a dynamic steady state be-tween class-switched and nonswitched GC B cells (Fig. 3).

The kinetics and distribution of Ig isotypes observed during theprimary anti-SRBC response are consistent with earlier studiesdemonstrating the rapidity of isotype switching in vivo (27, 30, 32,59), but raise questions regarding the extent, kinetics, and durationof the isotype switch reaction with respect to GC formation andevolution. In immunohistochemical analyses of C57BL/6 mice re-sponding to a thymus-dependent form of the hapten NP, isotype-switched cells were first detected within extrafollicular PALS foci(from which both AFCs and GC founder cells originate), and themajority of GCs observed (first seen in this report at day 8) werecomprised almost entirely of isotype-switched cells, indicating arapid and near total conversion from sIgM to sIgG in both PALSfoci and GCs (30). Although subsequent studies by Kelsoe andcolleagues reported GC formation as early as day 4 postimmuni-zation (63), the heavy chain isotype expressed within these earlyGCs was not examined. Using multiparameter flow cytometry,other studies of the primary anti-NP response revealed that Ag-specific, isotype-switched GC and non-GC B cells displayed sim-ilar kinetics with respect to appearance and exponential expansion,but the relative frequencies of nonswitched vs switched NP-bind-ing GC B cells were not evaluated (27, 32, 59).

In the primary anti-SRBC response of BALB/c mice describedin this study, sIgG11 and sIgM1 GC B cells emerged at the sametime as GC B cells first become discernible, and the relative fre-quencies of these two populations did not change appreciably overthe ensuing 2–3 wk of the response. Surprisingly, no gradual (orsudden) diminution in the sIgM1 population coincident with anequivalent increase in the sIgG11 population was observed (Fig.3). These findings could be interpreted to mean that the isotypeswitch reaction not only begins to occur before or concomitantwith the induction of GCs, but that class switching is restricted tothis early phase of B cell activation. Thus, the relative frequenciesof the switched and nonswitched populations established duringthe early stage of B cell activation would remain essentially un-changed after this window of opportunity for class switching hadpassed. Alternatively, the present data could imply that isotypeswitching involvingm and/or other heavy chain genes may occurcontinuously throughout the GC response, with regulatory mech-anisms maintaining a nearly constant frequency of both switchedand sIgM1 B cells.

Although it is unclear which mechanism accounts for the almostfixed frequency of switched and nonswitched cells in response toSRBC challenge, evidence for both exists in other systems. Spe-cifically, studies have provided support for isotype switching atextrafollicular sites early in the response as well as in the GC laterafter Ag challenge. As discussed above, immunohistochemicalstudies by Kelsoe and colleagues have provided evidence forswitching in both PALS foci and GC (30). Toellner et al. (39) andPeakman and Maizels (64), combining RT-PCR or in situ hybrid-ization with immunohistochemistry, respectively, examined the lo-calization of switch transcript production in situ and reported thepredominant site of isotype switching during anti-NP responses tobe in the extrafollicular T cell regions of immunized spleen. Peak-man and Maizels (64) documented that peak transcript productionpreceded GC formation, although switch transcripts were detectedwithin GCs subsequent to their appearance, suggesting GCs also

5735The Journal of Immunology

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

have the potential to serve as sites of isotype switching. Consistentwith this, Zinkernagel and colleagues (65), performing immuno-histochemical studies following primary infection with vesicularstomatitis virus, reported both switched and nonswitched GC Bcell populations at day 12 following infection. This is of interestbecause PALS foci are not observed in response to vesicular sto-matitis virus immunization. Finally, evidence for isotype switchingwithin the GC also comes from human studies in which switchtranscripts and DNA excision loops, intermediates in deletionalisotype switch recombinations, were detected in sort-purified GCB cell subsets (41, 42). It is thus possible that the present resultscould be explained by either switching before GC formation,within the GC, or both. Experiments examining sIgM1 andsIgG11 GC B cells at various times after SRBC immunization forintermediates of the isotype switching reaction are in progress todirectly address this issue.

The persistence of sIgM1 cells throughout all stages of the pri-mary response was particularly striking, and could also be ex-plained by a protracted recruitment of mature sIgM1 cells into theGC reaction, although some of the sIgM1 cells may well representthe precursors of sIgM1 memory cells, direct evidence for whichhas been reported in the human (66–69). However, most sIgM1

GC B cells also expressed CD23 (data not shown), a determinantexpressed late in B cell ontogeny, but absent from isotype-switched or sIgM1 memory B cells (50, 51, 67). Moreover, and asdiscussed below, a significant proportion of sIgM1 GC B cellssimultaneously expressed sIgD, CD23, and levels of CD38 com-parable with follicular B cells. Collectively, these phenotypic dataare consistent with many of the sIgM1 cells corresponding to GCfounder cells or other early stages of GC B cell differentiation.

Recently, human GC founder cells have been identified and par-tially characterized (46–49). This population exhibited a pheno-type reminiscent of both follicular and GC B cells, including theexpression of sIgD, sIgM, CD23, and CD38. In our studies, mAbsdirected against murine sIgD, CD23, and CD38 revealed subpopu-lations of B2201PNAhigh cells expressing these respective markers(Fig. 5A). This phenotype is virtually identical to that described forhuman GC founder cells, allowing for the possibility that murinesIgD1CD231CD38highB2201PNAhigh cells likewise represent thefirst stage of GC commitment. Surprisingly, B2201PNAhigh cellsexpressing sIgD, CD38, or CD23 were detected after day 10, whenthe peak of the GC response (as measured by the frequency ofB2201PNAhigh cells) had diminished (Fig. 5A). Although the sig-nificance of this observation is currently unclear, it could suggestthat recruitment from the follicular B cell pool may occur for anextended period of time following SRBC challenge.

Further phenotypic analysis revealed that CD23, CD38, andsIgD demarcated discrete but overlapping populations potentiallycorresponding to early stages of GC B cell ontogeny (Fig. 5,B andC, and data not shown). From the kinetic analyses and the four-color flow-cytometric data, it would appear that during the processof commitment to the GC reaction, sIgM1sIgD1CD231CD38high

follicular B cells acquired receptors for PNA before the loss ofsIgD. Based on the comparative frequencies of sIgM, sIgD, CD23,and CD38-expressing GC B cells, it is likely that both sIgD andCD38 were subsequently down-regulated by the sIgM1 GC Bcells. In contrast, CD23 expression was maintained until thesIgM1 GC B cells either underwent isotype switching or pro-ceeded further along the GC developmental trajectory toward non-switched memory B cells, as few sIgM1CD232 GC B cells weredetected until about day 8 (data not shown).

Although sIgD1B2201PNAhigh cells have previously been de-tected 12 days following immunization of C57BL/6 mice with NPkeyhole limpet hemocyanin (70) and IgD1 cells can be found

within presumptive GCs 9 days after immunization with pigeoncytochromec (71), IgD expression by murine GC B cells has notbeen well described. To more thoroughly investigate theIgD1B2201PNAhigh population observed by flow cytometry, weemployed two-color confocal microscopy of splenic sections toexamine cellular localization. IgD1PNAhigh cells detected 8 daysfollowing immunization were found within the follicular mantle ofsecondary follicles, along the border of the follicular mantle andthe subjacent GC, and within the GC (Fig. 6). The presence ofnon-GC IgD1PNAhigh cells is of particular interest, and can beinterpreted several ways. One could suggest these cells to be at anearly stage of activation and commitment to the GC, and hencefounder cells, as discussed above. Alternatively, these cells mightbe IgD1 GC B cells emigrating from the GC, and hence representa population of memory cells, especially at later time points. It isalso possible that both processes are going on simultaneously.

That murine IgD1PNAhigh cells might represent post-GC mem-ory cells is all the more intriguing with the recent revelation of apopulation of human memory B cells expressing not only sIgM,but also sIgD (68, 69). In the human studies, CD27 expression wasfound to distinguish human memory B cells (both sIgM1 andsIgM2) from naive B cells (68, 69), raising the possibility that thismarker may prove useful in murine studies as well. Inpreliminaryexperiments, however, few if any murine GC B cells were found toexpress detectable levels of CD27 at days 4, 8, 12, and 18 postim-munization (Shinall and Waldschmidt, unpublished observations).

Unlike the situation described above for isotype distribution,expression of the activation Ag BLA-1 within the GC B cell com-partment was found to be dynamic over time. BLA-1 expressionwas initially uniform, but became heterogeneous by the end of thefirst week. By day 12 and thereafter, approximately one-half of theGC B cell population expressed BLA-1, and by days 24–26, mostGC B cells were devoid of this marker (Fig. 4A, and data notshown). Intriguingly, simultaneous evaluation of sIgM and BLA-1expression at different times following immunization revealed fourGC B cell subsets with the predominant GC B cell population,expressing both sIgM and BLA-1 (Fig. 4B). BLA-1 expressiondecreased among both sIgM1 and sIgM2 cells as a function oftime. These data confirm and extend earlier studies that docu-mented expression of BLA-1 in GCs at early but not late timepoints following immunization (27, 43).

The identity, function, and regulation of expression of BLA-1remain unclear, although polyclonal activators such as LPS caninduce BLA-1 expression on cultured B cells and such BLA-11

cells have the forward light scatter characteristics of B cell blasts(43). Given the distribution and kinetics of BLA-1 expressionwithin the GC compartment and the restriction of BLA-1 to LPSblasts in vitro, it is conceivable that BLA-1 expression may coin-cide with proliferating GC B cells. Alternatively, expression of thisAg may be induced early during GC B cell ontogeny, and subse-quently become down-regulated as a function of time, cell divi-sion, or availability of stimuli (e.g., Ag, T cell-derived costimula-tion), and thus may serve to distinguish recently recruited fromresident GC B cells. We are currently investigating the relationshipbetween BLA-1 expression and proliferative and differentiationalstatus within the murine GC.

In summary, the present work identifies GC B cell subsets po-tentially corresponding to distinct stages of GC B cell ontogenydefined by the differential expression of isotypes (IgM, IgD, IgG1),the activation marker BLA-1, and differentiation Ags (CD38,CD23) as well as the kinetics of their induction and persistencefollowing primary immunization of BALB/c mice with SRBC. Im-portantly, in comparative studies, the subsets identified in thisstudy were conserved in other strains of mice immunized with

5736 MURINE GERMINAL CENTER B CELL SUBSETS

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

SRBC and in BALB/c mice immunized with other thymus-depen-dent Ags, suggesting that these subsets are general rather thanunique to the SRBC response (S. Shinall and T. Waldschmidt,manuscript in preparation). The generality of the findings maytherefore allow for further work in the mouse describing subsets orstages of GC B cells in which distinct genetic programs are turnedon, leading to affinity maturation and memory cell formation.

AcknowledgmentsWe thank Dr. Lee Herzenberg for providing 53-10.1 mAb (anti-BLA-1),Dr. Douglas Fearon for providing the 1D3 mAb (anti-CD19), DNAX Re-search Institute (Palo Alto, CA) for providing the 1C10 mAb (anti-CD40),Dr. John Kearney for providing clone 90 mAb (anti-CD38), Teresa Dulingfor expert operation of the flow cytometer, Lorraine Tygrett for experttechnical assistance, and Carla Hartl for assistance in the preparation of thismanuscript.

References1. Kosco-Vilbois, M. H., J.-Y. Bonnefoy, and Y. Chvatchko. 1997. The physiology

of murine germinal center reactions.Immunol. Rev. 156:127.2. Liu, Y.-J., and C. Arpin. 1997. Germinal center development.Immunol. Rev.

156:111.3. Maclennan, I. C. M., A. Gulbranson-Judge, K.-M. Toellner,

M. Casamayor-Palleja, E. Chan, D. M.-Y. Sze, S. A. Luther, and H. A. Orbea.1997. The changing preference of T and B cells for partners as T-dependentantibody responses develop.Immunol. Rev. 156:53.

4. Liu, Y.-J., and J. Banchereau. 1996. The paths and molecular controls of periph-eral B-cell development.Immunologist 4:55.

5. Tsiagbe, V. K., G. Inghirami, and G. J. Thorbecke. 1996. The physiology ofgerminal centers.Crit. Rev. Immunol. 16:381.

6. Kelsoe, G. 1995. In situ studies of the germinal center reaction.Adv. Immunol.60:267.

7. MacLennan, I. C. M. 1994. Germinal centers.Annu. Rev. Immunol. 12:117.8. Kroese, F. G. M., W. Timens, and P. Nieuwenhuis. 1990. Germinal center reac-

tion and B lymphocytes: morphology and function.Curr. Top. Pathol. 84:103.9. Foy, T. M., A. Aruffo, J. Bajorath, J. E. Buhlmann, and R. J. Noelle. 1996.

Immune regulation by CD40 and its ligand GP39.Annu. Rev. Immunol. 14:591.10. Matsumoto, M., Y.-X. Fu, H. Molina, and D. D. Chaplin. 1997. Lymphotoxin-

a-deficient and TNF receptor-I-deficient mice define developmental and func-tional characteristics of germinal centers.Immunol. Rev. 156:137.

11. Liu, Y.-J., and J. Banchereau. 1996. Mutant mice without B lymphocyte follicles.J. Exp. Med. 184:1207.

12. Alimzhanov, M. B., D. V. Kuprash, M. H. Kosco-Vilbois, A. Luz,R. L. Turetskaya, A. Tarakhovsky, K. Rajewsky, S. A. Nedospasov, andK. Pfeffer. 1997. Abnormal development of secondary lymphoid tissues in lym-photoxinb-deficient mice.Proc. Natl. Acad. Sci. USA 94:9302.

13. Futterer, A., K. Mink, A. Luz, M. H. Kosco-Vilbois, and K. Pfeffer. 1998. Thelymphotoxinb receptor controls organogenesis and affinity maturation in periph-eral lymphoid tissues.Immunity 9:59.

14. Rennert, P. D., D. James, F. Mackay, J. L. Browning, and P. S. Hochman. 1998.Lymph node genesis is induced by signaling through the lymphotoxinb receptor.Immunity 9:71.

15. Stuber, E., and W. Strober. 1996. The T cell-B cell interaction via OX40-OX40Lis necessary for the T cell-dependent humoral immune response.J. Exp. Med.183:979.

16. Stuber, E., M. Neurath, D. Calderhead, H. P. Fell, and W. Strober. 1995. Cross-linking of Ox40 ligand, a member of the TNF/NGF family, induces proliferationand differentiation in murine splenic B cells.Immunity 2:507.

17. Turner, C. A. Jr., D. H. Mack, and M. M. Davis. 1994. Blimp-1, a novel zinc-finger-containing protein that can drive the maturation of B lymphocytes intoimmunoglobulin-secreting cells.Cell 77:297.

18. Lin, Y., K. K. Wong, and K. Calame. 1997. Repression of c-myctranscription byBlimp-1, an inducer of terminal B cell differentiation.Science 276:596.

19. Rajewsky, K. 1992. Early and late B-cell development in the mouse.Curr. Opin.Immunol. 4:171.

20. Hardy, R. R., C. E. Carmack, S. A. Shinton, J. D. Kemp, and K. Hayakawa. 1991.Resolution and characterization of pro-B and pre-pro-B cell stages in normalmouse bone marrow.J. Exp. Med. 173:1213.

21. Li, Y.-S., K. Hayakawa, and R. R. Hardy. 1993. The regulated expression of Blineage associated genes during B cell differentiation in bone marrow and fetalliver. J. Exp. Med. 178:951.

22. Li, Y-S., R. Wasserman, K. Hayakawa, and R. R. Hardy. 1996. Identification ofthe earliest B lineage stage in mouse bone marrow.Immunity 5:527.

23. Pascual, V., Y.-J. Liu, A. Magalski, O. de Bouteiller, J. Banchereau, andJ. D. Capra. 1994. Analysis of somatic mutation in five B cell subsets of humantonsil. J. Exp. Med. 180:329.

24. Arpin, C., J. Dechanet, C. Van Kooten, P. Merville, G. Grouard, F. Briere,J. Banchereau, and Y.-J. Liu. 1995. Generation of memory B cells and plasmacells in vitro.Science 268:720.

25. Kraal, G., I. L. Weissman, and E. C. Butcher. 1982. Germinal centre B cells:antigen specificity and changes in heavy chain class expression.Nature 298:377.

26. Butcher, E. C., R. V. Rouse, R. L. Coffman, C. N. Nottenburg, R. R. Hardy, andI. L. Weissman. 1982. Surface phenotype of Peyer’s patch germinal center cells:implications for the role of germinal centers in B cell differentiation.J. Immunol.129:2698.

27. Lalor, P. A., G. J. V. Nossal, R. D. Sanderson, and M. G. McHeyzer-Williams.1992. Functional and molecular characterization of single, (4-hydroxy-3-nitro-phenyl)acetyl (NP)-specific, IgG1

1 B cells from antibody-secreting and memoryB cell pathways in the C57BL/6 immune response to NP.Eur. J. Immunol. 22:3001.

28. Kimoto, H., H. Nagaoka, Y. Adachi, T. Mizuochi, T. Azuma, T. Yagi, T. Sata,S. Yonehara, Y. Tsunetsugu-Yokota, M. Taniguchi, and T. Takemori. 1997. Ac-cumulation of somatic hypermutation and antigen-driven selection in rapidly cy-cling surface Ig1 germinal center (GC) B cells which occupy GC at a highfrequency during the primary anti-hapten response in mice.Eur. J. Immunol.27:268.

29. Gonzalez, M., F. Mackay, J. L. Browning, M. H. Kosco-Vilbois, and R. J. Noelle.1998. The sequential role of lymphotoxin and B cells in the development ofsplenic follicles. J. Exp. Med. 187:997.

30. Jacob, J., R. Kassir, and G. Kelsoe. 1991. In situ studies of the primary immuneresponse to (4-hydrxoy-3-nitrophenyl)acetyl. I. The architecture and dynamics ofresponding cell populations.J. Exp. Med. 173:1165.

31. Hollowood, K., and J. Macartney. 1992. Cell kinetics of the germinal centerreaction: a stathmokinetic study.Eur. J. Immunol. 22:261.

32. McHeyzer-Williams, M. G., M. J. McLean, P. A. Lalor, and G. J. V. Nossal.1993. Antigen-driven B cell differentiation in vivo.J. Exp. Med. 178:295.

33. Xu, J., T. M. Foy, J. D. Laman, E. A. Elliott, J. J. Dunn, T. J. Waldschmidt,J. Elsemore, R. J. Noelle, and R. A. Flavell. 1994. Mice deficient for the CD40ligand. Immunity 1:423.

34. Matsumoto, M., S. Mariathasan, M. H. Nahm, F. Baranyay, J. J. Peschon, andD. D. Chaplin. 1996. Role of lymphotoxin and the type I TNF receptor in theformation of germinal centers.Science 271:1289.

35. Weissman, I. L., G. A. Gutman, S. H. Friedberg, and L. Jerabek. 1976. Lymphoidtissue architecture. III. Germinal centers, T cells, and thymus-dependent vs thy-mus-independent antigens.Adv. Exp. Med. Biol. 66:229.

36. Han, S., B. Zheng, Y. Takahashi, and G. Kelsoe. 1997. Distinctive characteristicsof germinal center B cells.Semin. Immunol. 9:255.

37. Miller, C., G. Kelsoe, and S. Han. 1994. Lack of B7-2 expression in the germinalcenters of aged mice.Aging Immunol. Infect. Dis. 5:249.

38. Stavnezer, J. 1996. Antibody class switching.Adv. Immunol. 61:79.39. Toellner, J.-M., A. Gulbranson-Judge, D. R. Taylor, D. M.-Y. Sze, and

I. C. M. MacLennan. 1996. Immunoglobulin switch transcript production in vivo relatedto the site and time of antigen-specific B cell activation.J. Exp. Med. 183:2303.

40. Hayakawa, K., R. Ishii, K. Yamasaki, T. Kishimoto, and R. R. Hardy. 1987.Isolation of high-affinity memory B cells: phycoerythrin as a probe for antigen-binding cells.Proc. Natl. Acad. Sci. USA 84:1379.

41. Martinez-Valdez, H., F. Malisan, O. de Bouteiller, C. Guret, J. Banchereau, andY. J. Liu. 1995. Molecular evidence that in vivo isotype switching occurs withinthe germinal centers.Ann. NY Acad. Sci. 764:151.

42. Liu, Y.-J., F. Malisan, O. de Bouteiller, C. Guret, S. Lebecque, J. Banchereau,F. C. Mills, E. E. Max, and H. Marinez-Valdez. 1996. Within germinal centers,isotype switching of immunoglobulin genes occurs after the onset of somaticmutation.Immunity 4:241.

43. Hardy, R. R., K. Hayakawa, D. R. Parks, L. A. Herzenberg, andL. A. Herzenberg. 1984. Murine B cell differentiation lineages.J. Exp. Med.159:1169.

44. Hsu, S.-M., and E. S. Jaffe. 1984. Phenotypic expression of B-lymphocytes. II.Immunoglobulin expression of germinal center cells.Am. J. Pathol. 114:396.

45. Nahm, M. H., P. A. Takes, M. B. Bowen, and K. A. Macke. 1989. Subpopulationsof B lymphocytes in germinal centers. II. A germinal center B cell subpopulationexpresses sIgD and CD23.Immunol. Lett. 21:201.

46. Billian, G., C. Bella, P. Mondiere, and T. Defrance. 1996. Identification of atonsil IgD1 B cell subset with phenotypical and functional characteristics ofgerminal center B cells.Eur. J. Immunol. 26:1712.

47. Liu, Y.-J., O. de Bouteiller, C. Arpin, F. Briere, L. Galibert, S. Ho,H. Martinez-Valdez, J. Banchereau, and S. Lebecque. 1996. Normal humanIgD1IgM2 germinal center B cells can express up to 80 mutations in the variableregion of their IgD transcripts.Immunity 4:603.

48. Lebecque, S., O. de Bouteiller, C. Arpin, J. Banchereau, and Y.-J. Liu. 1997.Germinal center founder cells display propensity for apoptosis before onset ofsomatic mutation.J. Exp. Med. 185:563.

49. Arpin, C., O. de Bouteiller, D. Razanajaona, I. Fugier-Vivier, F. Briere,J. Banchereau, S. Lebecque, and Y.-J. Liu. 1998. The normal counterpart of IgDmyeloma cells in germinal center displays extensively mutated IgVH gene,Cm-Cd switch, andl light chain expression.J. Exp. Med. 187:1169.

50. Kikutani, H., M. Suemura, H. Owaki, H. Nakamura, R. Sato, K. Yamasaki,E. L. Barsumain, R. R. Hardy, and T. Kishimoto. 1986. Fce receptor, a specificdifferentiation marker transiently expressed on mature B cells before isotypeswitching.J. Exp. Med. 164:1455.

51. Waldschmidt, T. J., D. H. Conrad, and R. G. Lynch. 1988. The expression of Bcell surface receptors. I. The ontogeny and distribution of the murine B cell IgEFc receptor.J. Immunol. 140:2148.

52. Best, C. G., J. D. Kemp, and T. J. Waldschmidt. 1995. Murine B-cell subsetsdefined by CD23.Methods Enzymol. 8:3.

53. Maeda, K., G. F. Burton, D. A. Padgett, D. H. Conrad, T. F. Huff, A. Masuda,A. K. Szakal, and J. G. Tew. 1992. Murine follicular dendritic cells and lowaffinity Fc receptors for IgE (FceRII). J. Immunol. 148:2340.

5737The Journal of Immunology

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from

54. Maeda, K., M. Matsuda, and Y. Imai. 1995. Follicular dendritic cells: structure asrelated to function.Curr. Top. Microbiol. Immunol. 201:119.

55. Lund, F., N. Solvason, J. C. Grimaldi, R. M. E. Parkhouse, and M. Howard. 1995.Murine CD38: an immunoregulatory ectoenzyme.Immunol. Today 16:469.

56. Lund, F. E., D. A. Cockayne, T. D. Randall, N. Solvason, F. Schuber, andM. C. Howard. 1998. CD38: a new paradigm in lymphocyte activation and signaltransduction.Immunol. Rev. 161:79.

57. Oliver, A. M., F. Martin, and J. F. Kearney. 1997. Mouse CD38 is down-regu-lated on germinal center B cells and mature plasma cells.J. Immunol. 158:1108.

58. Ridderstad, A., and D. M. Tarlinton. 1998. Kinetics of establishing the memoryB cell population as revealed by CD38 expression.J. Immunol. 160:4688.

59. Smith, K. G., T. D. Hewitson, G. J. V. Nossal, and D. M. Tarlinton. 1996. Thephenotype and fate of the antibody-forming cells of the splenic foci.Eur. J. Im-munol. 26:444.

60. Takahashi, T., L. J. Old, and E. A. Boyse. 1970. Surface alloantigens of plasmacells.J. Exp. Med. 131:1325.

61. Stearne, P.A., I. R. van Driel, B. Grego, R. J. Simpson, and J. W. Goding. 1985.The murine plasma cell antigen PC-1: purification and partial amino acid se-quence.J. Immunol. 134:443.

62. Rose, M. L., M. S. C. Birbeck, V. J. Wallis, J. A. Forrester, and A. J. S. Davies.1980. Peanut lectin binding properties of germinal centers of mouse lymphoidtissue.Science 284:364.

63. Jacob, J., and G. Kelsoe. 1992. In situ studies of the primary immune response to(4-hydroxy-3-nitrophenyl)acetyl. II. A common clonal origin for periarteriolarlymphoid sheath associated foci and germinal centers.J. Exp. Med. 176:679.

64. Peakman, M. C., and N. Maizels. 1998. Localization of splenic B cells activatedfor switch recombination by in situ hybridization with Ig1 switch transcript andRad51 probes.J. Immunol. 161:4008.

65. Bachmann, M. F., B. Odermatt, H. Hengartner, and R. M. Zinkernagel. 1996.Induction of long-lived germinal centers associated with persisting antigen afterviral infection.J. Exp. Med. 183:2259.

66. Van Ets, J. H., F. H. J. Gmelig Meyling, and T. Lotenberg. 1992. High frequencyof somatically mutated IgM molecules in the human adult blood B cell repertoire.Eur. J. Immunol. 22:2761.

67. Klein, U., R. Kuppers, and K. Rajewsky. 1997. Evidence for a large compartmentof IgM-expressing memory B cells in humans.Blood 89:1288.

68. Klein, U., K. Rajewsky, and R. Kuppers. 1998. Human immunoglobulin(Ig)M1IgD1 peripheral blood B cells expressing the CD27 cell surface antigencarry somatically mutated variable region genes: CD27 as a general marker forsomatically mutated (memory) B cells.J. Exp. Med. 188:1679.

69. Tangye, S. G., Y.-J. Liu, G. Aversa, J. H. Phillips, and J. E. de Vries. 1998.Identification of functional human splenic memory B cells by expression ofCD148 and CD27.J. Exp. Med. 188:1691.

70. Ford, J. E., M. G. McHeyzer-Williams, and M. R. Lieber. 1994. Analysis ofindividual immunoglobulinl light chain genes amplified from single cells isinconsistent with variable region gene conversion in germinal-center B cell so-matic mutation.Eur. J. Immunol. 24:1816.

71. Mikszta, J. A., L. J. McHeyzer-Williams, and M. G. McHeyzer-Williams. 1999.Antigen-driven selection of TCR in vivo: related TCRa-chains pair with diverseTCR b-chains.J. Immunol. 163:5978.

5738 MURINE GERMINAL CENTER B CELL SUBSETS

by guest on February 3, 2019http://w

ww

.jimm

unol.org/D

ownloaded from