B Cell Antigen Presentation Is Sufficient To Drive Neuroinflammation in an Animal Model of Multiple Sclerosis

Multiple sclerosis (MS), a chronic demyelinating disease of the CNS affecting close to 2.3 million people worldwide, is a leading cause of disability in young adults (1, 2). Moderately efficacious immune-modulating therapies for MS have been developed, in part with the aid of the CD4 T cell–dependent animal model experimental autoimmune encephalomyelitis (EAE). By generating and presenting target autoantigens, APCs play an essential role in coordinating the behavior of CD4 T cells and inflammatory destruction of myelin during EAE (3, 4). Combined expression of MHC class II (MHCII), costimulatory molecules, and cytokines by APCs regulates CD4 T cell functional traits in both peripheral and CNS compartments and ultimately directs the inflammatory cascade of events resulting in myelin and nerve damage (4, 5). The identity and characteristics of APCs involved in initiating and propagating inflammation within the CNS have been under intense scrutiny (3, 5). Although dendritic cells (DCs) were suggested to serve all required APC roles in EAE and MS, they are not sufficient to generate maximal disease in recombinant myelin oligodendrocyte glycoprotein (rMOG)-immunization models of EAE or for the development of spontaneous optic neuritis (6). Thus, additional APCs must participate in the generation and propagation of myelin-reactive CD4 T cells in autoimmune neuroinflammation. Extensive studies examined the contribution of other APCs, such as monocytes, macrophages, and microglia, in EAE, and results suggested that they work in concert with DCs to promote disease (3).

Several studies identified the contributions of another professional APC, B cells, in the pathogenesis of CNS inflammatory demyelination, offsetting the earlier viewpoint that B cells are not required for EAE, which was suggested by work in mice genetically deficient in B cells (7). For example, myelin oligodendrocyte glycoprotein (MOG)-specific Ig increases disease severity of EAE (8–10), and greater numbers of MOG-specific B cells combined with T cells recognizing cognate Ag results in spontaneous inflammatory demyelination in the CNS (11, 12). Further, B cell depletion after the onset of EAE can ameliorate inflammation and clinical disease (13, 14). Moreover, subsets of B cells identified by their production of IL-10, IL-6, or IL-35 were shown to modulate the severity of EAE (15–17). Alternatively, B cells have suppressive traits during EAE, because depletion of B cells before peptide immunization can exacerbate disease (13). In sum, B cells are clearly implicated in the pathogenesis of EAE, via multiple mechanisms, including cytokine and Ig production, as well as regulation of CD4 T cell function.

The importance of B cells in MS is underscored by the demonstration that B cell–depletion therapy can be highly efficacious for certain patients (18). However, the mechanisms by which removal of B cells from MS patients results in clinical benefit remain unclear. Although plasma cells and Ig are typical features of the MS plaque (2, 19), and localized intrathecal production of Ig is detected in most patients with MS (20), the efficacy of B cell depletion in MS appears to be independent of any effects on plasma cells or Ig (21–23). Furthermore, follow-up studies on MS patients undergoing B cell depletion revealed alterations in proliferation and proinflammatory cytokine production by CD4 T cells (21). These studies raise questions regarding the degree to which B cell Ag presentation, rather than Ig production, drives neuroinflammation during MS. B cells were recognized to function as APCs in neuroinflammation, particularly after the induction of EAE via immunization with rMOG (14). Subsequent work suggested that B cell Ag presentation is required to initiate disease induced by human rMOG immunization in a B cell–dependent form of EAE (24). However, whether B cells are capable of independently driving CD4 T cell autoreactivity to myelin targets during EAE has not been determined.

Hence, we sought to determine the sufficiency of B cells for APC function during EAE. We began our studies by generating a murine system for the conditional expression of MHCII to restrict expression of MHCII to B cells. We originally observed that B cell Ag presentation is not sufficient for the initiation or propagation of EAE (25). However, because of the clinical evidence for B cell Ag presentation during EAE and MS, we chose to further explore the role of Ag-specific B cells. In this study, we found that maximal disease in protein-induced active EAE models is dependent upon B cell Ag presentation. Further, narrowing the repertoire of B cell specificity to MOG results in enhanced Ag processing and presentation, facilitating EAE development. To our knowledge, this is the first demonstration in vivo that B cells can serve as the primary APC and coordinate CD4 T cell autoreactivity to myelin Ags during EAE.

C57BL/6 (B6) mice were purchased from the Jackson Laboratory. B^MHCII mice were generated as previously reported (25). DC^MHCII mice (26) were kindly provided by Dr. Terri Laufer (University of Pennsylvania, Philadelphia, PA), and IgH^MOG mice (10) were kindly provided by Dr. Hartmut Wekerle (Max Planck Institute of Neurobiology, Munich, Germany). IAβ^f/f mice were described previously (27) and were bred to CD19^Cre mice obtained commercially from The Jackson Laboratory. Thymic grafting was performed as described using thymi from B6 mice (6). Animals were housed in a specific pathogen–free barrier facility at the Washington University School of Medicine. All breeding and experimental protocols were performed in accordance with protocols reviewed and approved by the Washington University Animal Studies Committee.

Active and passive EAE were induced, as described (25, 28). Briefly, active EAE was induced in mice by s.c. injection of 200 μg MOG_35–55 emulsified in CFA (Sigma-Aldrich) containing 500 μg H37RA (Difco, Detroit, MI) and i.p. injection with 200 ng pertussis toxin (Enzo Life Sciences) in 0.2 ml PBS on day 0. Additionally, mice received 200 ng pertussis toxin i.p. 48 h later. Alternatively, mice were injected with 150 μg rMOG protein [residues 1–125 of human MOG protein (6)] in 500 μg CFA. Passive EAE was established by transfer of 1 × 10⁷ MOG-specific, Thy1.1⁺ encephalitogenic cells, as reported (25, 29). Mice were observed daily, and clinical score was assessed using a five-point scoring system, as follows: 0 = no disease; 1 = limp tail; 2 = mild hindlimb paresis; 3 = severe hindlimb paresis; 4 = complete hindlimb plegia or quadriplegia; and 5 = moribund or dead.

Prior to perfusion, spleens were harvested from experimental mice. Single-cell suspensions from spleens were treated with ACK erythrocyte lysis buffer. Mice were perfused with 25 ml ice-cold PBS, and brains and spinal cords were collected from perfused mice and processed with a dounce homogenizer to obtain single-cell suspensions. CNS cells were purified by centrifugation for 30 min in a 30% Percoll (GE Healthcare Life Sciences) solution, as previously reported (25). Cells were incubated with the anti-FcR Ab 2.4G2 prior to the addition of Abs. The following Abs were purchased from BD Biosciences: CD45-FITC, CD8α–allophycocyanin–H7, CD19-allophycocyanin-H7, CD19-BV510, B220–PE–Texas Red, B220–PE-CF594, CD11b–Alexa Fluor 700, and MHC-v450. MHCII–Pacific Blue and CD11c-PECy7 Abs were purchased from eBioscience. The following Abs were purchased from BioLegend (San Diego, CA): CD138-PE, MHCII (I-A/I-E)–Pacific Blue, Thy1.1-PerCP, and CD4-allophycocyanin. Cells were acquired on a Gallios flow cytometer (Beckman Coulter) and analyzed with FlowJo software (TreeStar), with doublets being excluded.

Mice were sacrificed and perfused with 25 ml cold PBS, followed by 20 ml 4% paraformaldehyde (Sigma-Aldrich). Brains and spinal cords were removed and fixed in 4% paraformaldehyde overnight or longer. Tissues were embedded in paraffin and sectioned (5 μm) by the Developmental Biology Histology and Microscopy Core at Washington University. Representative sections were stained for myelination with solochrome cyanine, as previously reported (30), and H&E for inflammation. Slides were examined by light microscopy using a Nikon 90i motorized upright digital microscope with camera and MetaMorph software (Molecular Devices).

Animals were sacrificed by cardiac puncture, and blood samples were allowed to clot at room temperature. Serum was obtained by centrifugation of clotted blood at 15,000 × g for 30 min and frozen at −20°C until use. Total serum MOG-specific IgG was quantified by plating serial serum dilutions on 96-well plates precoated with 10 μg/ml rMOG in coating buffer (sodium bicarbonate). Plates were blocked with 1% BSA in PBS and incubated with sera overnight at 4°C. After washing, MOG-specific IgG retained by the plate-bound rMOG was detected with alkaline phosphatase–conjugated goat anti-mouse IgG F(ab′)₂ (Jackson ImmunoResearch) and SIGMAFAST p-Nitrophenyl phosphate tablets with SIGMAFAST Tris Buffer tablets in dH2O (both from Sigma-Aldrich). Absorbance was measured at 450 nm with a μQuant plate reader, and KC Junior software (both from Bio-Tek) was used for data analysis.

8.18C5-producing hybridoma cells were grown in CELLine Flasks (BD Biosciences). Supernatant was collected, and rAbs were purified with a Hi-Trap Protein G Column HP (GE Healthcare Life Sciences). Abs were eluted with 0.1 M glycine into microcentrifuge tubes that contained 1 M Tris and were dialyzed against PBS. Ab concentrations were determined by absorbance at 280 nm. 8.18C5 (200 μg i.v.) was injected on days 7 and 9 posttransfer of encephalitogenic T cells.

Generation of the MOG-specific T cell hybridoma 204.62 was performed according to standard protocol (31). Briefly, B6 mice were immunized via footpad injection of 10 nM MOG_35–55 emulsified in CFA (Difco). Seven days following immunization, the draining popliteal lymph nodes were dispersed and fused according to standard protocols (31). Hybridomas were used in Ag-presentation assays, as described previously (25). Spleens were processed for B cell enrichment by AutoMACS (Miltenyi Biotec, Auburn, CA). Single-cell suspensions were plated, washed, and combined with Ag. A total of 5 × 10⁵ APCs was cultured with 5 × 10⁴ hybridoma cells with Ag overnight at 37°C. Proliferation of CTLL-2 was measured after the addition of supernatant by [³H]thymidine incorporation, as described (31). EC₅₀ was calculated by constructing a dose-response curve for each population of APCs, using the concentration of rMOG providing half-maximal IL-2 response by the hybridoma 204.62.

The Mann–Whitney U test was used for comparisons of EAE severity. Student t tests were used for comparisons of leukocytes and Ig values.

We showed previously that Ag presentation by DCs alone is capable of mediating full disease in active EAE induced by immunization with MOG_35–55 but not rMOG protein or for the development of spontaneous optic neuritis (6). Based on these results, along with the results of B cell–depletion studies in MS and EAE, we hypothesized that B cell Ag presentation cooperates with DC Ag presentation to maximize the initial response to rMOG protein during active EAE induction. To explore the possible cooperation between DC and B cell Ag presentation during the initiation of EAE, we bred mice in which MHCII expression is restricted to DCs [DC^MHCII (26)] to mice in which MHCII expression is restricted to B cells [B^MHCII (25)], generating mice in which DCs and B cells are the only MHCII-bearing cells (DC^MHCII×B^MHCII). After thymic grafting from B6 neonates, mice were immunized with rMOG, according to standard protocols (6). Although B^MHCII and DC^MHCII mice were highly resistant to rMOG-induced EAE as reported, the combination of DC and B cell MHCII expression significantly increased the susceptibility to EAE (Fig. 1A), both in terms of incidence and severity (Table I). These results suggest that B cell Ag presentation is required to enhance the encephalitogenic response by CD4 T cells during EAE.

Based on our findings in DC^MHCII×B^MHCII mice, we hypothesized that B cell Ag presentation is necessary for EAE development. Thus, we examined the requirement for B cell Ag presentation during EAE using a genetic model of conditional MHCII deletion (27). We crossed IAβ^f/f mice (27) with CD19^Cre mice to eliminate MHCII expression by B cells (IAβ^f/f×CD19^Cre mice). Using rMOG protein as immunogen, Zamvil and colleagues (24) reported that IAβ^f/f×CD19^Cre mice are resistant to EAE. We sought to determine the influence of B cell MHCII expression during the secondary phases of EAE. EAE was induced in B6 and IAβ^f/f×CD19^Cre mice by transfer of encephalitogenic CD4 T cells. We found that IAβ^f/f×CD19^Cre mice exhibited similar EAE severity as did wild-type (WT) controls (Table I). The degree of MHCII elimination in IAβ^f/f×CD19^Cre mice was reported to be very high but with a variable percentage of MHCII⁺ cells remaining (27). Thus, we performed flow cytometric assessment of MHCII expression in IAβ^f/f×CD19^Cre mice during EAE, which revealed a sizeable residual expression of MHCII by B cells in blood and spleen (Supplemental Fig. 1). These findings demonstrate incomplete elimination of B cell MHCII expression in IAβ^f/f×CD19^Cre mice with EAE, leaving open the possibility that the severity of disease in mice entirely lacking MHCII expression by B cells is reduced. Nonetheless, our results demonstrate a requirement for B cell expression of MHCII during the initiation of EAE, but they call into question the role for B cell Ag presentation during the propagation of responses by encephalitogenic CD4 T cells during EAE.

These results, along with the reports of several other investigators (21, 24), suggest that B cell Ag presentation is critical to the development of CD4 T cell–mediated inflammatory CNS demyelination. However, B cell MHCII expression alone is not sufficient to support active or passive EAE (25). Thus, we sought to determine the basis of resistance to EAE in B^MHCII mice, in which B cells exclusively express MHCII (25). Because cognate B and T cell interactions were implicated during pathogenic CD4 T cell–dependent responses, including autoimmune CNS demyelination (14, 21), we reasoned that increasing the efficiency with which APCs acquire and present Ag to CD4 T cells would facilitate disease. To test this hypothesis, we increased the precursor frequency of MOG-specific B cells by crossing B^MHCII mice to IgH^MOG mice, which express a receptor highly specific for MOG (B^MHCII×IgH^MOG mice). Active immunization of B^MHCII×IgH^MOG mice with rMOG did not elicit signs of EAE (Table I). Following the receipt of MOG-specific encephalitogenic CD4 T cells, WT mice developed clinical evidence of EAE beginning, on average, at day seven (Fig. 1B, Table I). In contrast, B^MHCII mice were entirely resistant to EAE after transfer of encephalitogenic CD4 T cells, as previously reported (25). However, B^MHCII×IgH^MOG mice with greater B cell specificity for MOG developed typical signs of EAE, beginning, on average, 17 d following the transfer of donor T cells (Fig. 1B, Table I). Of note, a statistically significant and reproducible delay in onset was observed (Fig. 1C), and a mild, but statistically significant, reduction in disease severity was also seen consistently compared with WT mice with passive EAE (Table I).

The spinal cord is the primary site of inflammatory damage in this murine system of EAE in mice on the B6 background (32). Hence, we examined spinal cords from WT, B^MHCII, and B^MHCII×IgH^MOG mice for pathologic changes. Histologic assessment revealed inflammatory infiltrates and foci of demyelination in WT spinal cords after passive EAE induction. Similarly, regions of inflammatory demyelination were observed in spinal cord samples from B^MHCII×IgH^MOG mice that developed passive EAE (Fig. 2). The lack of clinical disease and the absence of myelin loss or inflammatory infiltrates within the spinal cord in B^MHCII mice were consistent with previous findings (25) (Fig. 2). No differences in the anatomic regions of immune cell infiltration or myelin loss were observed between WT and B^MHCII×IgH^MOG mice.

At both the peak of disease and during chronic stages of passively induced EAE, flow cytometric analysis revealed a typical inflammatory composition of activated microglia with infiltrating macrophages and lymphocytes in spinal cord tissue from WT and B^MHCII×IgH^MOG mice (Fig. 3A, 3B). When mice reached the peak of disease, the fractions of CD45^intCD11b^int microglia and CD11b⁻CD45^hi lymphocytes were elevated in B^MHCII×IgH^MOG mice compared with WT mice. However, the absolute numbers of these cells were comparable between the groups (Fig. 3C); the discrepancy between the percentage and the absolute number of cells likely reflects, in part, the difference in the overall composition of mononuclear cells between groups, considering the lack of endogenous CD4⁺ T cells in B^MHCII×IgH^MOG mice. During the late stages of passive EAE, no differences in the frequency or absolute number of microglia or lymphocytes were found between WT and B^MHCII×IgH^MOG mice (Fig. 3D). Notably, both the fraction and absolute number of MHCII-bearing cells within the spinal cord were drastically reduced in B^MHCII×IgH^MOG mice compared with WT mice during the chronic phases of disease (Fig. 3C, 3D). Because of the requirement for Ag presentation within the CNS (4, 33), we hypothesized that an accumulation of B cells within the spinal cord during EAE in B^MHCII×IgH^MOG mice would be evident; however, similar to WT mice, B cells were observed in low proportion and number at both early and late time points (Fig. 3C, 3D), consistent with the previous description of minimal B cell CNS infiltration during standard models of EAE in B6 mice (34, 35). These results demonstrate that increased numbers of B cells specific for cognate Ag can function alone to propagate autoreactive CD4 T cell inflammatory responses in the CNS, resulting in typical clinicopathologic features of EAE.

Approximately 30% of B cells from IgH^MOG mice express BCR specific for MOG (10). Although soluble Ig can facilitate CNS demyelination (36–38), increasing the BCR specificity for MOG is also likely to enhance recognition of target Ag along with increasing the efficiency of Ag processing and presentation by B cells (39, 40). We sought to distinguish the role of soluble MOG-specific Ig in B cell–mediated EAE. Thus, we quantified serum Ig recognizing MOG protein and its conformational epitopes in WT, B^MHCII, and B^MHCII×IgH^MOG mice. As expected, low levels of serum MOG-specific IgG were detected in WT mice with passive EAE (Fig. 4A). These low levels of anti-MOG IgG were similar to the levels found in WT or B^MHCII mice without disease (Fig. 4A). In contrast, serum from naive B^MHCII×IgH^MOG mice contained approximately twice the amount of MOG-specific IgG in comparison with WT and B^MHCII mice (p < 0.0001, Fig. 4A). The level of MOG-specific IgG increased after the development of passive EAE in B^MHCII×IgH^MOG mice, but this difference was not statistically significant (p = 0.07, Fig. 4A). Serial dilutions were performed to quantify the range of IgG in each group (Fig. 4B). These results suggested that the presence of serum MOG-specific Ig may act to facilitate EAE in B^MHCII×IgH^MOG mice. Transfer of the MOG-specific mAb 8.18C5 into animals exhibiting clinical signs of EAE is known to increase disease severity and speed the course of disease (9, 38) or to initiate disease after immunization in B cell–deficient mice (36). Thus, we examined whether soluble Ig specific for MOG can trigger disease induction in B^MHCII mice. Transfer of 8.18C5 7 d after induction of EAE resulted in significantly elevated levels of serum MOG-specific IgG (p = 0.0006, Fig. 4A). As expected, transfer of 8.18C5 resulted in enhanced disease severity in WT mice (Fig. 5), demonstrating the efficacy of 8.18C5 in the positive-control arm of the experiment. In contrast, administration of 8.18C5 to B^MHCII mice did not alter resistance to the development of EAE (Fig. 5). Identical EAE resistance was observed when serum from protein-immunized WT mice with EAE was administered to B^MHCII mice (data not shown).

Because soluble MOG-specific IgG was not sufficient to induce disease in our model system, we hypothesized that B^MHCII×IgH^MOG mice are susceptible to disease as a result of enhanced detection, acquisition, and presentation of self-Ags by B cells due to the higher frequency of MOG-specific B cells. To test this, we cocultured a highly sensitive MOG-specific T cell hybridoma with rMOG and B cells from WT, B^MHCII, or B^MHCII×IgH^MOG spleens. Whole WT spleens were capable of stimulating MOG-specific CD4 T cells. In comparison, MOG-specific CD4 T cell responses were slightly reduced with populations of B cells enriched from WT spleens (Fig. 6A). A substantial increase in T cell responses elicited from B^MHCII×IgH^MOG mice splenic B cells was observed compared with WT B cells and even more so in comparison with B^MHCII B cells (Fig. 6A). Specifically, half-maximal stimulation was achieved at 381 nM of rMOG using B cells enriched from WT spleens in comparison with 70.81 nM of rMOG using B cells from B^MHCII×IgH^MOG mice. Of note, processing and presentation of rMOG by DCs in DC^MHCII mice is not deficient (Fig. 6B, 6C). These results demonstrate the exquisite capacity for B cells from B^MHCII×IgH^MOG mice to capture and present Ag to CD4 T cells, which reflects passive EAE induction when B cells are the only functional APC.

Based on our in vivo system of conditional MHCII expression, we originally observed that B cell Ag presentation alone was insufficient to initiate or propagate EAE (25). In our present study, we find that B cells are capable of independently performing APC functions during EAE, but only when enough B cells recognize cognate Ag. To our knowledge, this is the first demonstration that B cells can serve all Ag-presentation functions during passive EAE. The importance of B cell Ag presentation during the initiation of autoimmune CNS demyelination is also highlighted by our finding that B cells cooperate with DCs to greatly augment disease severity in an active immunization EAE model. These results reveal the capacity of B cells to directly regulate cognate CD4 T cell responses during autoimmunity and highlight the selective ability for B cell Ag presentation to propagate CD4 T cell autoreactivity targeting the CNS.

Molnarfi et al. (24) recently developed a system for studying B cell Ag presentation that is complementary to the model used in the current study. Using a T cell– and B cell–dependent active immunization model of EAE (7), they demonstrated the necessity of B cell expression of MHCII, rather than Ab production, for initiating a pathogenic T cell response during neuroinflammation. In contrast, our work demonstrates the functional cooperation between DC and B cell Ag presentation during the initiation of active EAE, as well as the sufficiency of B cell Ag presentation for disease progression in passive EAE. Certainly, other APCs are capable of complementing DCs in the activation of autoreactive CD4 T cell responses. Macrophage and monocyte APCs were implicated in the initiation of EAE (41), and microglia have long been suspected of facilitating CD4 T cell autoreactivity in EAE (42). Nonetheless, our results implicate a role for B cell Ag presentation in EAE and suggest potential value in targeting cognate B cell–T cell interactions in MS.

The requirement for MOG-specific BCRs to facilitate EAE in our system suggests that MOG-specific Ig may be critical for disease induction. However, we were unable to elicit disease in B^MHCII mice by addition of either soluble Ig specific for MOG or serum from immunized WT mice. These findings are consistent with the hypothesis that membrane-bound BCR specific for MOG is critical to the propagation of CD4 T cell autoreactivity, likely for efficient capture of target Ag. Indeed, recently published work by Zamvil and colleagues (24) demonstrates a critical role for B cell APC function independent of Ig secretion. Immunization with human rMOG is a well-characterized B cell–dependent model of active EAE (7, 36, 43). However, there are critical differences in amino acid sequences of the main encephalitogenic peptide (MOG_35–55) processed from human and rodent MOG protein that could affect the immune response through altered Ag detection, processing, and presentation. Using rodent rMOG allowed us to study a relevant autoimmune reaction in B^MHCII×IgH^MOG mice, because the 8.18C5 H chain of IgH^MOG mice was developed specifically in response to rodent MOG.

To confirm the Ig-independent nature of our EAE model, we treated B^MHCII mice that had received encephalitogenic T cells with either 8.18C5 or serum from rMOG immunized WT mice. Although previous research indicates that serum from WT mice is not pathogenic, we used this to control for the possible immune-activating effects of serum components like cytokines and chemokines. Additionally, our ex vivo Ag presentation results support the hypothesis that MOG-specific BCR enhances the ability of B cells to present cognate Ag, with B cells from B^MHCII×IgH^MOG mice eliciting an ∼5-fold greater stimulation of MOG-specific CD4 T cells compared with WT B cells. These data do not detract from the pathogenic role of soluble MOG-specific Ig in vivo, but rather demonstrate that a distinct pathway dependent upon membrane-bound MOG-specific Ig is critical at later stages during EAE as a mechanism to capture and process target Ag.

We believe that our work may have implications regarding the pathogenesis of MS. The presence of intrathecally generated oligoclonal Ig has been used as a diagnostic feature of MS (2). Although the targets of these Abs remain unclear (44, 45), the presence of oligoclonal Ig in the cerebrospinal fluid space in most patients with MS raises the important question about our model of where B cell cognate interactions with T cells occur during disease. Clonal expansion and Ig class switching and secretion are direct results of B cell Ag presentation to cognate T cells. Indeed, Lambracht-Washington et al. (46) reported that clonally expanded B cells from the cerebrospinal fluid of MS patients produce Ig with reactivities to myelin basic protein and some polyreactivity to GFAP and CNPase. In our model, up to 30% of B cells recognize a CNS-specific Ag, a situation that might reflect the prevalence or expansion of B cells in MS patients. Whether B cells acquire and present Ag in the CNS compartment is a critical question that is amenable to address using our unique system.

Therefore, we propose a model in which myelin autoreactivity initiates when B cells cooperate with DCs to prime CD4 T cells, followed by germinal center reactions and B cell clonal expansion that results in B cells assuming a primary role in Ag presentation to propagate CD4 T cell responses. Thus, the number of B cells expressing BCRs specific for cognate autoantigen is highly relevant in promoting the likelihood of B and T cell interactions to perpetuate disease. Although the upper limit of Ag-specific B cells required for disease in our model was defined as ∼30% by previous characterizations of the IgH^MOG mouse (10), our attempts to define a threshold for autoreactive B cells in EAE initially involved administration of rMOG and adjuvant several days prior to the initiation of passive EAE in B^MHCII mice. B^MHCII mice treated with rMOG and CpG or with rMOG and CpG and anti-CD40 Ab were resistant to passive EAE (data not shown). Thus, activating MOG-specific B cells via preconditioning with Ag and adjuvant does not overcome the barrier to EAE when B cells function as the sole APC. Nonetheless, a logical extension of our findings is identifying the clinically relevant threshold for B cell Ag specificity. Theoretically, enhanced efficacy of B cell–depletion therapy could be achieved by specifically targeting autoreactive B cells that have expanded to deleterious levels, a concept that is recognized [e.g., (46)] but impeded by the confusion over actual CNS targets of lymphocytes. Perhaps Ag-specific targeting of B cells is more practical in neuromyelitis optica, an inflammatory disease similar to MS in which B and T cell autoreactivity toward a specific protein target—aquaporin 4—is defined (47).

What remains to be demonstrated is the requirement for DCs at later stages of EAE and MS. In fact, this is highly controversial given the recent observation that DC depletion may exacerbate EAE (48). However, in contrast to DCs that are short-lived, B cells can mature into long-lived memory phenotype with the capacity to indefinitely extend autoimmune responses. Additionally, depletion of B cells may unleash potent proinflammatory monocyte-derived APCs in the setting of MS (49), and the complexity of B cell–targeted therapy in MS was highlighted recently in a clinical trial using the B cell–depleting agent, atacicept, which led to exacerbation of MS rather than improvement (50).

Our data demonstrate that minimal MHCII expression within the CNS is required during the propagation of EAE. In light of the requirement for the generation of target Ags from CNS tissue (51, 52), these data suggest that myelin Ags derived from the CNS can be generated in sufficient quantity by a relatively minimal number of APCs. The small numbers of B cells observed in the CNS of B^MHCII×IgH^MOG mice that develop EAE may be a result of B cells being more facile in capturing soluble protein Ags. Alternatively, protein Ags may be directly targeted from the periphery by B cells after an initial breach of the CNS compartment, particularly by highly activated T cells. This would serve as an early step in the genesis of cognate Ags, resulting in the delay in disease observed in B^MHCII×IgH^MOG mice compared with WT mice. In specific circumstances, accumulation of B cells in the CNS during MS and EAE occurs in the form of ectopic lymphoid follicles (34, 53). Whether this process requires MHCII has yet to be explored.

We thank Gurumoorthy Krishnamoorthy and Hartmut Wekerle for generously providing IgH^MOG mice and advice on experimental approaches. We thank Drs. Anne Cross, Robyn Klein, Raj Apte, Emil Unanue, and Laura Piccio for critical review of the manuscript, as well as Soomin Shin and Gretchen McGee for other technical support. We acknowledge the technical assistance of Bryan Bollman for completion of the histochemical experiments.

The authors have no financial conflicts of interest.