Antigen-Specific Suppression of Humoral Immunity by Anergic Ars/A1 B Cells

B cells in the preimmune repertoire achieve self-tolerance through at least three distinct mechanisms that include anergy, receptor editing, and clonal deletion (1–6). Although it is not fully understood why one mechanism or another operates in a given autoreactive B cell, high-avidity interactions between the BCR and self-Ag tend to promote receptor editing and clonal deletion, whereas low-avidity interactions favor anergy (3, 7, 8). Anergic B cells may persist with a lifespan of several days (9, 10). In mice, they constitute 2–5% of the B cell repertoire, and in humans they comprise ∼2.5% of peripheral blood B cells (11, 12).

As a consequence of chronic signaling, anergic B cells are generally refractory to acute signals through the BCR, as assessed by Ca²⁺ mobilization, protein tyrosine phosphorylation, proliferation, and Ab secretion (2, 13–17). However, anergy is not necessarily a permanent state. TLR agonists, excess survival signals, withdrawal from the self-Ag, or a very strong BCR signal can reverse various features of anergy and in some cases lead to the secretion of autoantibodies (11, 18–21). This reversible nature of the anergic state has led to the view that anergic B cells are nothing more than a threat to immunological self-tolerance. However, in view of the paradigm that evolution tends to discard structures or processes with no functional value, it is curious that anergic B cells are not promptly eliminated, particularly if they are dangerous. This paradox and the fact that anergic T cells have been demonstrated to possess regulatory functions (22) have given us cause to consider the possibility that anergic B cells may play a role in immune physiology.

To determine the functional capabilities of anergic B cells, we exploited a unique characteristic of the Ars/A1 transgene model of anergy. Ars/A1 B cells express Ig μδ and κ transgenes that together encode a BCR with dual reactivity for the hapten p-azophenylarsonate (Ars) and a self-Ag that includes, but may not be limited to, ssDNA. This self-specificity renders Ars/A1 cells anergic, as assessed by phenotype and function in signaling, proliferation, and Ab production assays (23). In these respects, Ars/A1 B cells are similar to MD4 × ML5 B cells. They differ, however, in that their cognate self-Ags are naturally occurring and ubiquitous in wild-type mice. Consequently, anergy in Ars/A1 B cells is maintained in adoptive recipients by omnipresent self-Ag.

By exploiting these unique features of Ars/A1 B cells, we were able to assess their biological activity in the context of the immune response to Ars. Upon adoptive transfer, Ars/A1 B cells potently suppressed the endogenous recipient IgG immune response to Ars/protein conjugates. This finding suggests that anergic B cells enforce tolerance to self-Ags they engage.

A/J, C57BL/6 (B6), B6xA/J F₁ (B6AF1), B6.129S2-H2^dlAb1-Ea/J (24) (MHC class II [MHC II]^−/−), B6.129S7-Rag1^tm1Mom/J (25) (Rag1^−/−), and B6.129P2-Il10^tm1Cgn/J (26) (IL-10^−/−) mice were purchased from The Jackson Laboratory and housed in the Biological Resource Center at National Jewish Health. Mice carrying Ars-specific canonical Ig μδ and κ transgenes (Ars/A1) or the κ transgene alone (κTg) have been described (23, 27). Ars/A1 and κTg transgenic mice were bred on B6AF1 or C57BL/6 backgrounds; crosses of the transgenes onto a knockout background were done on the C57BL/6 background. All mice were handled and bred with Institutional Animal Care and Use Committee approval in accordance with institutional guidelines.

In standard splenocyte transfers, single-cell suspensions of splenocytes from 8- to 14-wk-old κTg or Ars/A1 donor mice were prepared. Cells were depleted of erythrocytes (RBC lysis buffer; Sigma-Aldrich) and washed once in B cell medium (RPMI 1640 supplemented with Na⁺ pyruvate, l-glutamine, antibiotics, and 10% FCS) and twice in PBS before resuspending to 10⁷ cells/ml in PBS. In most experiments, 10⁶ cells were injected into the lateral tail vein. In some experiments, B cells were purified either by flow cytometry sort or by using a B cell enrichment kit (StemCell Technologies) or depleted using a anti-B220 biotin-coupled Ab and a biotin selection kit (StemCell Technologies) following the manufacturer’s recommendations. Cell purity was determined by flow cytometry to be >95%. Cells were washed twice in PBS and injected i.v. in numbers equivalent to those present in 10⁶ unpurified splenocytes. In some experiments, cells were labeled with 5 μM CFSE (Molecular Probes) as described (11). Cells were washed twice in PBS before i.v. injection into the lateral tail vein.

A single-cell suspension of Ars/A1 splenocytes was prepared. Erythrocytes were removed as mentioned above and cells were divided into four aliquots. Three aliquots were depleted of CD11b⁺ and CD11c⁺ cells using a biotin selection kit (StemCell Technologies) according to the manufacturer’s recommendation. One aliquot was further enriched for B cells and a second aliquot was further enriched for T cells using kits from StemCell Technologies. Populations were analyzed for purity via flow cytometry. Serial dilutions of the various populations (as APCs) were cultured with 10⁵ T hybridoma cells (T17-38). This hybridoma reacts with a peptide from the Vκ region expressed by Ars/A1 B cells (27). At 14 h, IL-2 in the supernatant was quantified using a europium (Eu)³⁺-based fluoroimmunometric assay (anti–IL-2 capture Ab JES6-1A12, anti–IL-2 detection Ab JES6-5H4).

In most experiments, recipients were immunized i.p. immediately after adoptive transfer with 100 μg arsanilated protein (Ars₁₅-keyhole limpet hemocyanin [KLH], Ars₉-OVA, Ars₂-hen egg lysozyme [HEL], or Ars₁₅-chicken γ-globulin) and 100 μg control protein (KLH, HEL, or OVA) in 200 μl IFA/PBS. In initial experiments, injections were done separately on either side of the peritoneum (see Figs. 1, 2, 3C, 5C, 5D). In more recent experiments, proteins were co-emulsified and injected together (see Figs. 3A, 3B, 4C, 6, Supplemental Fig. 2). Booster injections on day 21 were identical to primary immunizations with respect to volume, quantity of immunogen, and adjuvant. Mice were bled from a lateral tail vein on day 21 or 6 d after the booster injection.

All serologic assays were Eu³⁺-based fluoroimmunometric assays as described (27), with one modification: commercially available enhancement solution was substituted with a solution (100 mM sodium acetate, 1 mM thenoyltrifluoroacetone, 750 mM trioctylphosphine oxide [pH 3.2]) made in-house as described (28). For Ag-specific assays, 96-well europium plates (Greiner Bio-One, Frickenhausen, Germany) were coated overnight at 4°C with one of various Ags (10 μg/ml), and Ag-specific serum IgG was quantified with a γ-chain–specific biotin goat anti-mouse IgG (SouthernBiotech, Birmingham, AL). In assays for secreted transgenic Abs, plates were coated overnight at 4°C with 17-63 (a mAb specific for the L chain of the transgene, made in-house). Transgenic Ab was detected with a biotinylated anti-allotypic IgM^a (clone MA-69; BioLegend). End point dilutions and ratios were determined using Excel 2007 for Macintosh; concentrations were extrapolated using Prism 5.0.

Spleens were frozen in Tissue-Tek OCT (Sakura Finetek) and stored at −80°C. Serial sections of 6–8 mm were transferred to microscope slides (ProbeOn Plus; Fisher Scientific), fixed in acetone for 5 min, and stored at −80°C until stained. Sections were blocked with staining buffer (2% FCS in PBS, 0.01% NaN₃) for 10 min at room temperature and incubated for 30 min at room temperature with B220-PE (RA3-6B2) and anti–CD4-allophycocyanin (GK1.5). Slides were analyzed using a Marianas system with Slidebook version 4.0 (Intelligent Imaging Innovations).

Cells were stained following a standard protocol with 30-min incubations at room temperature. Staining with the I-A^b-3K tetramer was performed as described (29). Ca²⁺ mobilization was performed as described (23). Flow cytometric acquisitions or sorts were done on FACScan, LSRII (both BD Biosciences, San Jose, CA), or CyAn and MoFlo XDP (both Beckman Coulter, Fullerton, CA) flow cytometers. Data were analyzed using FlowJo 8.6 (Tree Star, San Carlos, CA).

The following Abs/stains were used: 1) from BioLegend: anti-B220 (RA3-6B2), anti-CD4 (GK1.5), anti-CD44 (IM7), anti-MHC II (M5/114.15.2), anti-CD8 (53-6.7), anti-CD69 (H1.2F3), anti-CD19 (ebio1D3), and anti-ICOS (C398.4A); 2) from eBioscience: anti-CD11c (N418), anti–IL-10 (JES5-16E3), and anti-CD5 (53-7.3); 3) from Becton Dickinson: anti-CD11b (M1/70), anti-CD1d (1B1), and anti-CXCR5 (2G8); 4) from Vector Laboratories: PNA-bio; and 5) made in-house: I-Ab-3K tetramer, E4 F(ab′)₂ (anti-idiotypic, H + L chain-specific for the Ars/A1 BCR) (23), 17-63 (anti-idiotypic, L chain-specific for the Ars/A1 and κTg BCR) (30), and C4H3 (anti–I-A^kHEL46-61) (31). For flow cytometry, Abs were either directly coupled to a fluorochrome or were resolved with a fluorochrome-coupled streptavidin. C4H3 was resolved with an anti-rat IgG coupled to allophycocyanin.

Two micrograms Ars₉-OVA in PBS/5 mM EDTA was maleimide-activated with sulfosuccinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate (Sulfo-SMCC; Pierce) at room temperature according to the manufacturer’s recommendations. Excess Sulfo-SMCC was removed by ultrafiltration (Centricon YM-10; Millipore). The activated Ars₉-OVA was combined with 2 mg lyophilized 3K peptide (FEAQKAKANKAVDGGGC) in 0.5 ml and incubated for 30 min at room temperature. Uncoupled 3K peptide was removed with an Amicon filter (Ultracel 30K; Millipore) (three washes with 4 ml PBS). The quality and purity of Ars₉-OVA-3K was tested in vitro. In a 96-well plate, 10⁶ fresh or paraformaldehyde-fixed Ars/A1, κTg, or C57BL/6 splenocytes were mixed with 10⁵ B3K-F508 hybridoma cells. Cells were incubated for 6 h or overnight with 3K peptide or Ars₉-OVA-3K. Upregulation of CD69 on hybridoma cells was measured by flow cytometry. Batches for immunization were able to upregulate CD69 with fresh APCs but not fixed APCs.

IgG anti–Ars-secreting cells were quantified by ELISPOT as described (30) with the following modifications. Ninety-six–well trays were coated overnight at 4°C with 10 μg/ml Ag (Ars₁₅-BSA) in PBS, and bound Ab was detected with a biotinylated and H chain-specific goat anti-mouse IgG (SouthernBiotech).

Statistical analyses were done using Prism 5.0. Data are shown as the means and SEM. For analyses of curves, the areas under the curves were determined first, after which the data were transformed as y = log(y). Statistical analyses on transformed log data were performed using an unpaired Student t test as specified in the figure legends. Other statistical analyses were performed using one-tailed Mann–Whitney tests. The α level for all tests was set at 0.05.

Ars/A1 mice carry an Ig μδ transgene and a separately integrated Ig κ transgene that encode an Ab directed against the hapten Ars. The Ars/A1 BCR also recognizes ssDNA and possibly other self-Ags, which renders Ars/A1 B cells anergic (23). This dual reactivity of the receptor enabled us to functionally assess anergic B cells in the context of a normal immune response to a foreign Ag. To this end, we transferred 10⁶ splenocytes from Ars/A1 mice or from control mice that carried only the Ars/A1 κTg into wild-type recipient mice and immunized the recipients with an Ars conjugate of KLH (Ars₁₅-KLH) together in IFA. Mice were bled on day 21, and sera were assessed for IgG directed to the hapten, the carrier protein, and the control protein.

Fig. 1A shows that the endogenous IgG immune response of mice that received Ars/A1 cells was strongly suppressed relative to responses by mice that received control κTg cells or no cells. The responses to both the hapten and the carrier were inhibited. To determine whether suppression was Ag specific, we performed a similar experiment in which mice were immunized with Ars₁₅-KLH and separately with OVA. In this case, only the IgG responses to Ars and the carrier protein KLH were inhibited (Fig. 1B). Related experiments produced similar results, in which Ars/A1 recipients often exhibited a >30-fold reduction in IgG anti-Ars titers relative to control recipients. Frequencies of splenic IgG Ab-secreting cells were similarly reduced (Fig. 1F).

Suppression of the response against the carrier protein, although not as strong as that of the anti-hapten response, was consistently observed in all experiments.

In related experiments, we found that Ag-specific immunosuppression by Ars/A1 cells occurred in recipient mice immunized with Ars conjugates of other proteins such as OVA and HEL. Thus, immunosuppression was not an idiosyncrasy of any particular carrier protein. In these experiments, sera were assayed at day 21 of the primary immune response. To determine whether immunosuppression extended into the secondary response, we immunized recipient mice with Ars₉-OVA plus HEL, delivered a booster injection of Ag on day 21, and bled them 6 d later. Sera were analyzed as before for IgG in direct binding assays. As shown in Fig. 1C–E, the IgG immune responses of Ars/A1 recipients remained depressed relative to those of control recipients.

To assess their suppressive potency, we transferred various numbers of Ars/A1 splenocytes to adoptive recipients and immunized them as before. Fig. 2A shows that few Ars/A1 cells were required to mediate immunosuppression: 10⁴ Ars/A1 splenocytes inhibited the primary IgG anti-Ars response by >95% relative to the response of mice that received control κTg cells. In the secondary immune response, the inhibition achieved was not as strong but was still evident and substantial with 10⁵ transferred cells (Fig. 2B). As in other studies, we found that the number of cells that established residence in peripheral lymphoid organs was only a small fraction of the total number transferred. At 16 h after transfer of 10⁵ CFSE-labeled Ars/A1 splenocytes, the spleen contained <2500 Ars/A1 B cells, and with 10⁴ transferred splenocytes this number was reduced to ∼300 (Fig. 2C).

To determine whether B cells in the Ars/A1 splenocytes were responsible for the observed immunosuppression, we depleted Ars/A1 splenocytes of B cells and compared them with whole Ars/A1 splenocytes in the adoptive transfer assay. Numbers of transferred B cell-depleted splenocytes were adjusted to be equivalent to those found in 10⁶, 10⁵, or 10⁴ whole splenocytes. Fig. 3 shows that B cell depletion resulted in loss of all detectable inhibitory activity that could not be accounted for by residual contaminating B cells. In this experiment, 10⁴ whole splenocytes, corresponding to ∼4 × 10³ Ars/A1 B cells, inhibited the primary anti-Ars IgG response by >95%. As an additional test, we performed the converse experiment in which purified Ars/A1 B cells (>98%) were compared with whole Ars/A1 splenocytes. Again, all of the inhibitory activity by transferred splenocytes was accounted for by the Ars/A1 B cells (Fig. 3C). Collectively, these results demonstrate that Ars/A1 B cells are potent and Ag-specific suppressors of humoral immunity.

Several groups have identified immunoregulatory B cells that exert anti-inflammatory function by secreting IL-10 upon activation (32–37). These cells have been variously described as transitional 2, marginal zone precursors (T2-MZPs) or a CD5⁺ subset of marginal zone B cells expressing high levels of CD1d. However, in contrast to T2-MZPs, IgM is downmodulated on Ars/A1 anergic B cells, and Ars/A1 mice lack a distinct population with the characteristic CD5⁺CD1d^hi phenotype of regulatory B cells (23) (Fig. 4A, 4B). To explicitly determine whether Ars/A1 B cells require IL-10 for immunosuppression, we produced IL-10–deficient Ars/A1 mice and tested B cells from these in the adoptive transfer assay. Fig. 4C shows that equivalent inhibition was attained regardless of whether transferred Ars/A1 cells carried a functional IL-10 gene. Therefore, on the basis of their phenotype and mode of function, anergic Ars/A1 B cells appear to be distinct from IL-10–producing regulatory B cells reported by others.

In other models of anergy, autoreactive B cells localize at the interface of T and B cell zones in secondary lymphoid tissues where cognate Ag-specific interactions between T and B lymphocytes are initiated during normal immune responses (10, 38–41). When transferred to adoptive recipients, Ars/A1 B cells localized similarly (Fig. 5A). To determine whether anergic Ars/A1 B cells were presenting self-Ags in MHC II, we tested them for presentation of a peptide derived from the κ-chain V region as a surrogate for self-Ag. Previous studies from our laboratory have shown that Ag-activated B cells self-display peptides from their BCR in MHC II, whereas high-density (ρ > 1.079) resting B cells do not (27). We assayed for display of a defined VκFR1 peptide derived from the Ars/A1 BCR in I-A^k by measuring the IL-2 response of a T cell hybridoma (T17-38) specific for this peptide. T17-38 does not require costimulation for an IL-2 response (27). Splenocytes from Ars/A1 mice were depleted of CD11b⁺/c⁺ cells or enriched for T or B cells and cultured with T17-38. As seen in Supplemental Fig. 1, Ars/A1 B cells effectively displayed the VκFR1 peptide in I-A^k. We take this as an indication that they are capable of processing and presenting in MHC II self-Ags that are engaged by the BCR.

In view of evidence that BCR signaling may be required for Ag presentation in MHC II, it was unclear whether B cells present new Ags acquired through the BCR in MHC II after they have entered the anergic state (42, 43). To test for this, we cultured Ars/A1 B cells with Ars₂-HEL or HEL at various concentrations and assessed MHC II presentation of HEL_46–61-derived peptide using the C4H3 mAb, which binds this peptide in the context of I-A^k (31, 44). In this experiment, Ag presentation in MHC II was more efficient with arsanilated HEL than with HEL, consistent with BCR-specific uptake (Fig. 5B). A capacity to present BCR-acquired Ag de novo is in agreement with prior studies involving MD4 × ML5 anergic B cells (19, 45). Collectively, these observations suggested that anergic B cells might be capable of directly engaging and inhibiting Th cells.

We considered the possibility that Ars/A1 B cells inhibited the humoral immune response by antigenic competition with normal Ag-specific B cells. To test for this, we performed the standard adoptive transfer assay but immunized recipient mice with a 10-fold higher dose of Ag than used in preceding experiments. Fig. 5C shows that even with a dose of 1 mg Ars-OVA, Ars/A1 splenocytes were able to effectively inhibit the immune response to Ars by adoptive recipients. Although this result supported the interpretation that antigenic competition did not account for Ars/A1 suppressive activity, it was not absolutely conclusive because BCR engagement with Ag might be restricted by an Ag-presenting niche in vivo, such as by an APC for B cells (46–50). Therefore, we designed a qualitative test in which Ars/A1 B cells should not be in competition with endogenous Ag-specific B cells. This experiment consisted of the standard adoptive transfer with the exception that the recipients were immunized with Ars₉-OVA together with unconjugated OVA, such that Ars/A1 B cells should not compete with OVA-specific endogenous B cells. The recipients were also injected with HEL as a control Ag. As seen in the center panel of Fig. 5D, the immune response to OVA was again inhibited by Ars/A1 B cells. This result was obtained regardless of whether Ars₉-OVA and OVA were co-emulsified or separately emulsified and injected. To rule out the possibility that Ars/A1 B cells indirectly inhibited the response to OVA by restricting the amount of OVA available to OVA-specific B cells, we tested whether the amount of injected OVA might be limiting in this experiment. To this end, we compared the strength of the immune responses to OVA in mice immunized with OVA alone (100 μg) or with Ars₉-OVA plus OVA (100 μg each) and found that the responses were equivalent (Supplemental Fig. 2). Thus, 100 μg OVA Ag was not limiting. Collectively, these results indicate that antigenic competition is not the major mechanism by which Ars/A1 B cells inhibit the immune response, and they suggest instead that Ars/A1 B cells mediate their effects by acting on carrier protein-specific T cells, in this case, OVA-specific T cells.

The results of the preceding two sections suggested that Ars/A1 B cells directly or indirectly inhibited protein carrier-specific CD4⁺ T cells in an Ag-specific manner. Because Ars/A1 B cells do not express the high levels of CD86 typically seen on activated B cells, we conjectured that they might induce tolerance in T cells by a direct Ag-dependent cognate interaction (21, 23). To test this idea, we bred the Ars/A1 μδ and κ transgenes or the control κ transgene alone into B6 mice that were deficient in MHC II (24). Ars/A1 MHC II^−/− B cells retained their anergic state, as assessed by a BCR-induced Ca²⁺ mobilization assay (Supplemental Fig. 3). When tested in the standard adoptive transfer experiment, the Ars/A1 MHC II^−/− cells still inhibited the endogenous primary and secondary IgG anti-Ars immune responses (Fig. 6A, 6B). However, the inhibition was substantially reduced relative to that obtained with MHC II-sufficient Ars/A1 cells. Two repeats of this experiment produced similar results. This indicates that most of the immunosuppression was due to a direct interaction between CD4 T cells and Ars/A1 B cells. However, redundant suppression mechanisms appear to be at play because the MHC II deficiency in Ars/A1 B cells did not ablate all of their inhibitory activity.

In view of recent studies indicating immunoregulatory properties of IgM (51–54), we determined whether transferred Ars/A1 B cells produced IgM^a (Fig. 6C). Initial results from a competition immunoassay revealed no detectable Ig bearing the Ars/A1 idiotype in sera of adoptive recipients from several experiments (data not shown). However, small amounts of such Ab were detected in some mice with a more sensitive immunoassay using an anti-idiotype capture reagent and an IgM^a-specific detection reagent. No Ars/A1 IgM^a was detected in recipients of MHC II^−/− Ars/A1 cells. This indicated that IgM^a production by Ars/A1 B cells was T cell-dependent, and that the MHC II-independent component of inhibition by Ars/A1 cells was not due to secretion of IgM (Table I). Table I shows the low but variable quantities of Ars/A1 IgM^a measured in adoptive recipients of Ars/A1 splenocytes. Notably, there was no correlation between the presence of IgM^a in a recipient and the degree of inhibition of the endogenous IgG immune response (Fig. 6C). We conclude that immunosuppression by Ars/A1 B cells is not due to their secretion of IgM.

Most of the inhibitory activity of Ars/A1 B cells required their expression of MHC II, indicating a cognate interaction between these cells and CD4 T cells. To seek additional evidence that CD4 T cells were inhibited by anergic B cells, we analyzed the effect of Ars/A1 B cells on the CD4⁺ T cell response to the I-A^b–restricted peptide called 3K (29). We analyzed this response because the 3K peptide elicits proliferation of a natural endogenous population of CD4⁺ T cells. Following adoptive transfer of Ars/A1 or κTg splenocytes, recipients were immunized with a covalent complex of Ars₉-OVA-3K and sacrificed on day 13, when germinal center reactions are normally well developed. In this experiment, the numbers of 3K-specific CD4⁺ T cells were substantially diminished in mice that received Ars/A1 cells relative to those that received control κTg cells. Numbers of 3K-specific T follicular helper (T_FH) cells were similarly reduced in Ars/A1 recipients (Fig. 7, Supplemental Fig. 4). This result indicates that anergic B cells suppress humoral immunity in part by inhibiting the development of Th cells.

In this study, we demonstrate that anergic Ars/A1 B cells are endowed with regulatory capabilities. This was revealed by their Ag-specific suppression of endogenous IgG immune responses in adoptive recipients. Assessing their functional potential was made possible by the fact that the anergic Ars/A1 cells analyzed in this study have a unique dual-reactive BCR that binds ssDNA and the hapten Ars, thus enabling us to evaluate their influence on the specific immune response elicited against arsanilated proteins. We found that suppression of IgG humoral immunity was strong and Ag-specific, and it applied to the carrier protein as well as the hapten, applied to the primary immune response, and extended to the secondary response. Suppression did not require production of IL-10 or IgM by Ars/A1 B cells and operated via redundant mechanisms, one of which involved a cognate interaction between MHC II-restricted T cells and Ars/A1 B cells. Approximately 95% immunosuppression of the anti-Ars response could be achieved with 10⁴ Ars/A1 splenocytes, which contained only ∼4000 Ars/A1 B cells. To our knowledge, this degree of immunosuppression by so few B cells is unprecedented. These results provide the first clear example of Ag-specific immunoregulation by an autoreactive anergic B cell.

Anergic B cells are considered desensitized owing to chronic low-level stimulation by self-Ag. They express reduced levels of CXCR5 and, similar to acutely triggered Ag-reactive B cells, localize to the interface of the B cell follicles and T cell zones in secondary lymphoid organs, where cognate T cell–B cell interactions are normally initiated during primary immune responses (41). Studies with the MD4 × ML5 model of anergy have revealed that anergic B cells can engage in Ag-directed cognate interactions with T cells (55–57). Our experiments similarly showed that Ars/A1 B cells localized to the interface of the T and B cell zones and that they are able to process and present exogenous Ag de novo in MHC II.

Consistent with their APC capabilities and localization at the interface of T and B cell zones, Ars/A1 cells suppressed immunity in a manner that depended partly on a cognate interaction with T cells. This was inferred by the fact that suppression was substantially reduced when Ars/A1 B cells lacked MHC II. Evidence for T cell involvement was also deduced by experiments showing that suppression could not be attributed entirely to antigenic competition at the level of the B cell and by reduced numbers of Th cells, including T_FH cells, in mice that received Ars/A1 B cells.

Although our findings reveal a cognate B cell–T cell interaction in immunosuppression, it is clear that such an interaction did not account for all of the activity of anergic B cells. Another pathway appears to be at play. It is possible that anergic B cells use an indirect pathway of Ag presentation involving other APCs to impart tolerance in T cells. In this regard, a prior report by Townsend and Goodnow (57) demonstrated, in an adoptive transfer model, that Ag (HEL) acquired by donor anergic MD4 × ML5 B cells could stimulate proliferation by HEL-specific T cells that was restricted by host MHC II. Such an indirect pathway of suppression is plausible in our system because we used an adjuvant (IFA) that is not a strong inducer of stimulatory APC activity by dendritic cells. At the same time, other mechanisms could be operating. For example, we have not entirely ruled out any role for antigenic competition or the possibility that Ars/A1 B cells might directly kill or otherwise impair endogenous Ag-specific B cells (58). Regardless, the fact that redundant suppressive mechanisms are at work implicates the physiological importance of anergic B cells in immunoregulation.

Recent studies have identified specific subpopulations of B cells as immunoregulators in models of contact hypersensitivity, allergic airways disease, and autoimmunity (33, 34, 37, 59, 60). In some of these models, regulatory B cells appear to execute their regulatory activities in an Ag-specific manner. In the airways model, anti-inflammatory activity was attributed to regulatory B cell-produced TGF-β, and it was associated with the induction of regulatory T cells (59); however, in the other models, IL-10 appeared to be the principal regulator of inflammation and autoimmunity (33, 34, 37, 60). This stands in contrast to the suppression of humoral immunity we observed, which did not require production of IL-10 by Ars/A1 B cells. Additionally, Ars/A1 B cells were phenotypically distinct from regulatory B cells reported in these other models. These findings, as well as the lack of a requirement for a functional IL-10 gene in Ars/A1 B cells for suppressive activity, indicate that they are a unique regulatory population.

Early studies by several groups implicated resting B cells as inducers of T cell tolerance (61–63). This was initially attributed to an absence of costimulatory molecules, such as CD86, on the B cell surface (61, 64–66). However, subsequent work revealed the importance of limited costimulation for the induction of T cell tolerance (67–69). Additionally, resting B cells are not guided by appropriate chemokines and associated receptors to the anatomical microenvironment where cognate interactions with T cells normally occur. Moreover, monovalent BCR interactions with Ag are generally weak and not likely to lead to efficient uptake and MHC II presentation of Ag. Additionally, it is unclear whether even high-affinity monovalent engagement of Ag by the BCR will lead to Ag presentation in MHC II, as high-density (ρ > 1.079) resting B cells do not display self-peptides from their BCR in MHC II (27). In contrast, when resting B cells are activated, BCR-derived peptides are processed and presented in MHC II. These considerations suggest that B cell activation is required for their APC activity. In principle, activation could occur with a multivalent Ag or a monovalent Ag that is repeatedly displayed on the surface of another cell. In view of the low numbers of anergic Ars/A1 B cells required in this study for immunosuppression and the high numbers of “resting B” cells (∼10⁷) required in preceding studies, we think it is plausible that anergic B cells accounted for the in vivo tolerogenic activity previously ascribed to resting B cells.

Ars/A1 B cells isolated ex vivo, without further manipulation, displayed peptides from their BCR in MHC II (Supplemental Fig. 1). MHC II-associated peptide from the BCR occurs upon BCR aggregation and, as such, may be considered a reporter for presentation of peptides from self-Ag resulting from BCR uptake (27). Anergic MD4 × ML5 B cells also present self-Ag (HEL peptides) in MHC II (19, 45). In view of our findings that anergic Ars/A1 B cells mediate suppression of responses to Ags taken in through the BCR, it is logical to speculate that wild-type An1 cells enforce immunological self-tolerance (11). This is plausible from a quantitative perspective as well. In our model, relatively few Ars/A1 B cells were required for immunosuppression of humoral immunity. With 10⁴ injected Ars/A1 splenocytes, for example, we found that only ∼300 Ars/A1 B cells seeded the spleen. In wild-type nontransgenic mice, An1 cells constitute ∼2–5% of all B cells, which corresponds to ∼1–2.5 × 10⁶ An1 cells per spleen (11). Additionally, ELISPOT assays indicated that ∼10% of An1 B cells react with nuclear Ags (11). This corresponds to >10⁵ nuclear Ag-specific An1 B cells per spleen. At these numbers, there should be sufficient An1 B cells to enforce tolerance with respect to nuclear self-Ags. However, this is dependent on An1 cells possessing suppressive activity, which has not yet been demonstrated. Additional studies will be required to assess wild-type An1 cells for potential regulatory activity. Together with results of prior studies demonstrating that autoreactive anergic T cells are immunosuppressive, our findings suggest a functional regulatory theme that is common to autoreactive anergic T and B lymphocytes, as well as an explanation for their persistence in the immune system (22).

We thank Dr. James Drake for the C4H3 mAb, Dr. Ross Kedl for the FGK4.5 mAb, and Fran Crawford for the I-A^b 3K-tetramer staining reagent. We thank Drs. John Cambier, Philippa Marrack, and Raul Torres for a critical review of the manuscript.

The authors have no financial conflicts of interest.