MHC class II–dependent B cell APC function is required for induction of CNS autoimmunity independent of myelin-specific antibodies

Immunization with rhMOG induces EAE in C57BL/6J WT mice (Fig. 1 A and Table 1), but not in B cell–deficient (J_HT) mice (Fig. 1 B and Table 1) or in B cell–depleted mice (not depicted). In contrast, MOG p35–55 and recombinant mouse MOG (rmMOG) can induce EAE in either WT mice (Fig. 1 A) or B cell–deficient mice (Fig. 1 B; Hjelmström et al., 1998; Fillatreau et al., 2002; Oliver et al., 2003). Thus, only rhMOG-induced EAE is considered B cell dependent (Lyons et al., 1999; Oliver et al., 2003; Marta et al., 2005).

When serving as APCs, B cells can produce proinflammatory cytokines that promote development and activation of encephalitogenic T cells (Weber et al., 2010). As IL-6 has a key role in Th17 differentiation (Okuda et al., 1999; Bettelli et al., 2006b), we first addressed whether selective IL-6 production by B cells is required in rhMOG-induced EAE. No difference in disease course was observed between mice containing selective B cell IL-6 deficiency (CD19cre–IL-6^flox/flox) and control (CD19cre–IL-6^wt/wt) mice that were immunized with MOG p35–55 (Fig. 1 C). In contrast, when immunized with rhMOG, mice containing IL-6–deficient B cells developed less severe EAE in comparison with control (CD19cre–IL-6^wt/wt and IL-6^flox/flox) mice (Fig. 1 D) and exhibited reduced IL-17 production in both the periphery and the CNS (Fig. 1 E). These results confirmed the importance of B cell IL-6 production in a model of B cell–dependent CNS autoimmunity.

Although cytokine production by accessory cells can promote proinflammatory T cell polarization, expression of MHC II on APCs is considered an obligate requirement for presentation and activation of Ag-specific CD4⁺ T cells (Schwartz, 1985). Thus, to determine the role of Ag presentation by B cells in EAE, we constructed mice containing MHC II–deficient B cells. Irradiated WT mice were reconstituted with mixed BM cells in a ratio of 80% J_HT and 20% MHC II^−/− (B–MHC II^−/−) or WT (B–MHC II^+/+) BM cells. Thus, in B–MHC II^−/− mice, MHC II deficiency was restricted to B cells, whereas other APCs, including DCs and monocytes, expressed MHC II (Fig. 2 A). The frequencies of B cells and CD4⁺ and CD8⁺ T cells were equivalent in BM chimeras containing either MHC II^−/− or MHC II^+/+ B cells (Fig. 2 A). As shown in Fig. 2 B, BM chimera mice containing MHC II^+/+ B cells developed EAE in response to immunization with rhMOG. In contrast, BM chimera mice containing MHC II–deficient B cells were resistant to clinical or histological EAE induction by rhMOG (Fig. 2 B and Table 1), although, like B cell–deficient J_HT or WT mice, BM chimera mice selectively deficient of MHC II on B cells did develop EAE in response to MOG p35–55 or rmMOG. Thus, EAE induction by rhMOG requires B cell MHC II expression.

Next, we examined how deficiency of MHC II on B cells influenced T cell activation and proinflammatory polarization in EAE induced by MOG p35–55 or rhMOG. Selective B cell MHC II deficiency did not influence proliferative responses to either Ag (Fig. 2 C), presumably because MHC II–competent myeloid cells also can process and present Ags. The contribution of MHC II expression by B cells in the development of pathogenic cells in vivo was also evaluated. No significant differences were observed in peripheral Th1 or Th17 responses in MOG p35–55–immunized BM chimera mice containing either MHC II^+/+ or MHC II^−/− B cells. Also, there were no differences in the frequency of CNS-infiltrating T cells or the number of CNS infiltrates in these mice (Fig. 2 C and Table 1). In contrast, in response to immunization with rhMOG, the frequency of splenic CD4⁺ T cells expressing IFN-γ and IL-17A was significantly decreased in B–MHC II^−/− chimera mice. Furthermore, CNS inflammation was not detected in mice completely lacking MHC II only on B cells. Therefore, MHC II expression on B cells is critical for pathogenic T cell responses in rhMOG-induced EAE.

Antibodies specific for rhMOG are known to be pathogenic (Marta et al., 2005). Although MHC II expression is required for B cell APC function, data indicate that MHC II is also necessary for germinal center formation and subsequent formation of long-lived plasma cells in T cell–dependent responses (Shimoda et al., 2006). Thus, we analyzed production of MOG-specific antibodies in our B–MHC II^−/− mice. Consistent with previous findings, MOG-specific antibodies were produced in control (B–MHC II^+/+) mice in response to rhMOG and rmMOG, but not to MOG p35–55 (Weber et al., 2010). In contrast, anti-MOG IgG was not detected in BM chimera mice containing MHC II–deficient B cells (Fig. 3 A). Interestingly, these data are consistent with the initial work that demonstrated the importance of MHC II in humoral immunity (McDevitt et al., 1972). Absence of MHC II–expressing B cells was also associated with reduction of T follicular helper cells (Tfh cells; Deenick et al., 2010), a specialized T cell subset which provides help for B cell maturation and isotype switching (Fig. 3 B; Vinuesa et al., 2005; Deenick et al., 2010). As B–MHC II^−/− mice did not produce anti-MOG IgG, we tested whether antibodies could restore EAE susceptibility after immunization with rhMOG. Whereas transfer of isotype control OVA-specific mAb had no effect on EAE in B–MHC II^−/− mice immunized with rhMOG (Fig. 3 C and Table 2), administration of pathogenic anti-MOG mAb, 8-18C5, led to development of mild clinical disease, but did not restore EAE severity to the level observed in B–MHC II^+/+ mice. Together, these results emphasize that the humoral contribution of B cells is insufficient to compensate for the lack of MHC II–dependent cellular function in MOG protein–induced EAE.

Besides serving as APCs or a source of autoantibodies, data indicate that through their cell surface–specific BCRs, B cells can transfer their target Ag to other professional APCs, a process which may be MHC II independent (Harvey et al., 2007). Thus, we addressed whether MOG-specific B cells contribute to MOG protein–induced EAE independent of their expression of MHC II molecules. BM chimera mice containing only IgH^MOG-ki B cells that expressed MHC II (IgH^MOG-ki/MHC II^+/+) were susceptible to EAE induced by MOG p35–55, rmMOG, or rhMOG (Fig. 3 D). In contrast, BM chimera mice containing MHC II–deficient IgH^MOG-ki B cells (IgH^MOG-ki/MHC II^−/−) were susceptible to EAE induced by MOG p35–55 or rmMOG but not to EAE induced by rhMOG (Fig. 3 D). Therefore, these results do not support BCR-mediated Ag exchange in rhMOG-induced EAE.

In the experimental system we used thus far, disruption of cellular function (i.e., MHC II deficiency) of B cells was accompanied by a reduction in humoral autoimmunity. It was therefore important to generate a model that would permit examination of the role of MHC II–competent MOG-specific B cells without possible contribution of antibodies. Tg mice containing B cells that express membrane IgH only have been useful for studying APC function in systemic and organ-specific autoimmune diseases (Chan et al., 1999; Wong et al., 2004). Thus, we created Tg mice containing B cells that express membrane MOG-specific Ig (IgH^MOG-mem) but are incapable of secreting Ig. We subcloned the rearranged heavy chain Ig gene from 8-18C5, the same gene used to create the IgH^MOG-ki mice, which are capable of secreting all Ig isotypes, into a construct containing the IgM^a constant region from which the exon encoding for secretion was excised (Fig. 4 A; see Materials and methods; Chan et al., 1999). Mice bearing the IgH^MOG-mem Tg were backcrossed onto the B cell–deficient J_HT strain (IgH^MOG-mem/J_HT; Chen et al., 1993) to prevent the development of B cells expressing endogenous Ig. More than 50% of Tg B cells in the founder line #3088, which was chosen for most of our subsequent experiments, bound MOG protein similarly to B cells in other founder lines and in IgH^MOG-ki mice (Fig. 4 B and not depicted). B cells in those Tg mice also expressed MHC II comparably with IgH^MOG-ki and WT mice (Fig. 4 C). B cells in IgH^MOG-mem/J_HT Tg mice expressed cell surface IgM only (not depicted). Comparable frequencies of CD4⁺, CD8⁺, naive, and effector/memory T cells were detected in IgH^MOG-mem/J_HT, WT, and IgH^MOG-ki mice (not depicted).

The absence of secreted antibody in IgH^MOG-mem/J_HT mice was confirmed in separate analyses of total and MOG-specific IgM, as well as total and MOG-specific IgG in naive, MOG p35–55–immunized, and MOG protein–immunized mice (Fig. 4, D and E). These antibodies were undetectable or observed only in trace amounts in IgH^MOG-mem/J_HT mice and were never significantly more than in B cell–deficient mice. In general, antibodies were 10²–10³ times lower than WT controls. In contrast, naive and immunized IgH^MOG-ki mice secreted much higher concentrations of MOG-specific IgM and IgG.

IgH^MOG-mem/J_HT mice were tested for susceptibility to EAE by immunization with MOG p35–55, rmMOG, and rhMOG. Onset, severity, and disease course in IgH^MOG-mem/J_HT, WT, and IgH^MOG-ki mice immunized with either MOG p35–55 or rmMOG were similar (Fig. 5 A and Table 3). Onset and initial clinical severity were also comparable when these mice were immunized with rhMOG. However, the IgH^MOG-mem/J_HT mice appeared to have a slightly milder disease course after the acute stage, although this difference was not statistically significant. A similar EAE course was observed in another IgH^MOG-mem/J_HT Tg founder line after immunization with rhMOG (not depicted). There were no obvious differences in the extent of meningeal or parenchymal CNS infiltration among any of these mice with EAE (Fig. 5 B and Table 3).

Data from our IgH^MOG-mem/J_HT mice demonstrated that it was the cellular and not the humoral contribution that was required for development of acute disease in B cell–dependent EAE. We therefore examined the role of BCR specificity by comparing EAE induction in our IgH^MOG-mem/J_HT mice with Tg mice containing B cells that express a fixed cell surface IgH specific for nitrophenyl, an irrelevant Ag. Membrane-only nitrophenyl-specific IgH (B1-8; IgH^B1-8-mem/J_HT) mice differ from IgH^MOG-mem/J_HT mice only in the specificity of their rearranged heavy chain gene (Chan et al., 1999). In comparison with B cells from IgH^MOG-mem/J_HT, a much lower frequency of B cells from IgH^B1-8-mem/J_HT were capable of binding rhMOG (Fig. 5 C). As anticipated, B cells from IgH^B1-8-mem/J_HT mice could present MOG p35–55 to MOG-specific 2D2 cells but were inefficient in presentation of MOG protein (Fig. 5 D), which requires processing (Slavin et al., 2001; Bettelli et al., 2003). Like IgH^MOG-mem/J_HT mice, IgH^B1-8-mem/J_HT mice were susceptible to EAE induced by MOG p35–55 or rmMOG, independent of BCR specificity. However, IgH^B1-8-mem/J_HT mice developed a much milder EAE course than IgH^MOG-mem/J_HT mice after immunization with rhMOG (Fig. 5 E), which corresponded to a reduction in MOG-specific proinflammatory Th1 and Th17 cytokine responses in rhMOG-induced EAE (Fig. 5 F). These results clearly support an important role for CNS auto-Ag BCR specificity in B cell APC function in vivo.

It was previously observed when TCR^MOG mice were crossed with IgH^MOG-ki mice, the progeny developed spontaneous paralysis, which was associated with meningeal and parenchymal inflammatory lesions within the optic nerves and spinal cord (OSE; Bettelli et al., 2006a; Krishnamoorthy et al., 2006), a lesion topography reminiscent of neuromyelitis optica. OSE was also associated with elevation in MOG-specific IgG1, and as IgH^MOG-ki B cells can also serve as APCs, it was not clear whether it was their potential role as APCs or production of MOG-specific autoantibodies that promoted development of OSE. Thus, to evaluate the participation of MOG-specific B cells, independent of MOG-specific antibodies, we crossed IgH^MOG-mem/J_HT with TCR^MOG/J_HT mice and evaluated their progeny for spontaneous development of OSE. 55% of IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice and 78% of IgH^MOG-ki × TCR^MOG developed paralysis spontaneously (Table 4). In contrast, spontaneous paralysis was not observed in IgH^B1-8-mem/J_HT × TCR^MOG/J_HT, TCR^MOG/J_HT, or IgH^MOG-ki mice. Onset of paralysis occurred earlier and the clinical severity was greater in IgH^MOG-ki × TCR^MOG mice than in IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice, possibly reflecting participation of MOG-specific antibodies in the manifestation of OSE. Interestingly, the numbers of inflammatory lesions were similar in both IgH^MOG-mem/J_HT × TCR^MOG/J_HT and IgH^MOG-ki × TCR^MOG mice. Prominent meningeal and parenchymal inflammatory foci with demyelination and formation of meningeal lymphoid follicle-like aggregates (Fig. 6, A–E) were observed in IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice, as was described previously in IgH^MOG-ki × TCR^MOG mice with spontaneous OSE (Bettelli et al., 2006a), and resembled similar structures observed in SPMS (Magliozzi et al., 2007). Meningeal foci in IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice were composed predominantly of B cells that infiltrate the parenchyma (Fig. 6 C). Relatively fewer meningeal and parenchymal T cells were detected in these areas (Fig. 6 D). Follicle-like organization was evident using a reticulin stain (Fig. 6 E). Meningeal inflammation and follicle-like structures were not detected in clinically normal IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice (Fig. 6, F–J). Our observations indicate that the APC function of MOG-specific B cells is sufficient to cooperate with MOG-specific T cells in development of OSE and further suggest that in the context of CNS inflammation, anti-MOG antibodies augment phenotypic expression of this spontaneous disease.

Recent MS trials indicate that B cell–depleting agents have a profound beneficial effect, underscoring the importance of B cells in MS pathogenesis (Bar-Or et al., 2008; Hauser et al., 2008; Marriott and O’Connor, 2010; Kappos et al., 2011). Although it was the appreciation of the potential humoral contribution of B cells that inspired testing B cell depletion in MS, the clinical response was not associated with reduction in antibody levels (Bar-Or et al., 2008; Hauser et al., 2008; Marriott and O’Connor, 2010; Kappos et al., 2011). B cells afford multiple functions. Besides serving as a source of antibody-producing plasma cells, B cells can participate as APCs (Lanzavecchia, 1985; Constant et al., 1995a,b; Chan et al., 1999) in Ag transport (Phan et al., 2007) and as regulatory cells (Matsushita et al., 2008). Although MHC II is expressed constitutively on B cells, it is unknown whether B cell Ag presentation may be required for activation of pathogenic CNS auto-Ag–specific T cells in vivo in EAE pathogenesis. In this investigation, we have created models that permitted us to evaluate and distinguish the role of B cells as APCs in vivo from their contribution to humoral CNS autoimmunity. Our results clearly demonstrate how B cell MHC II expression can be pivotal for activation of proinflammatory Th1 and Th17 cells in vivo and susceptibility to CNS autoimmunity. By generation of Tg mice that expressed cell surface myelin-specific BCR, we provided unequivocal evidence that B cell APC function, in the absence of humoral autoimmunity, is sufficient to drive proinflammatory T cell activation in vivo and EAE pathogenesis. Although BCR specificity for MOG protein promoted activation of encephalitogenic proinflammatory T cells, expression of this myelin-specific BCR was insufficient to drive proinflammatory T cell differentiation in the absence of B cell MHC II molecules (Fig. 3 D), providing further evidence for the MHC II-dependent APC participation of B cells. Elimination of B cell APC function can paralyze T cell activation and histological and clinical manifestations of CNS autoimmunity, findings which provide insight regarding B–T communication in MS pathogenesis and the influence of B cell depletion in MS therapy.

With greater appreciation for the role of B cells in MS, it is important to study MS models that require B cells for development of CNS autoimmunity. There are several different murine EAE models, and each one has merits that permit investigation of specific aspects involved in pathogenesis of CNS autoimmunity. With identification of encephalitogenic T cell epitopes within individual myelin Ags, most studies use those corresponding encephalitogenic peptides for EAE induction. In this regard, MOG p35–55 is frequently used in EAE studies. In general, B cells recognize conformational determinants created by longer contiguous or noncontiguous amino acid sequences within native proteins (Roelants, 1972). Whereas MOG p35–55 activates encephalitogenic T cells (Kuchroo et al., 2002), it does not efficiently activate B cells or promote MOG-specific antibody production (Weber et al., 2010). In contrast, MOG protein efficiently leads to both B cell activation and production of anti-MOG antibodies (Weber et al., 2010), and it is recognized that rhMOG, but not rodent MOG isoforms, induces EAE in a B cell–dependent manner (Marta et al., 2005). Interestingly, this model is not DC dependent, as mice depleted of this myeloid APC subset were susceptible to EAE induced by MOG protein (Isaksson et al., 2012). The B cell dependence of rhMOG-induced EAE is attributed to the presence of proline 42 (P42) in its sequence (Oliver et al., 2003; Marta et al., 2005), as substitution of serine, the corresponding residue in rodent MOG, with P42 in rodent MOG protein confers B cell–dependent induction of EAE (Oliver et al., 2003; Marta et al., 2005). Use of rhMOG-induced EAE also permitted us to evaluate the role of B cell cytokine production in expansion of proinflammatory T cells in vivo. Selective deficiency in MHC II–dependent B cell APC function was associated with significant reductions in both Th1 and Th17 cells. As it is recognized that B cells are a source of IL-6, a key Th17-polarizing cytokine (Bettelli et al., 2006b), we investigated how B cell production of IL-6 itself influenced EAE susceptibility. Although selective B cell IL-6 deficiency did not significantly alter the course of MOG p35–55–induced EAE, onset of rhMOG-induced EAE was delayed, severity was attenuated and the frequency of Th17 cells only was significantly reduced in B cell IL-6–deficient mice. Similar results were obtained in mixed BM chimera mice containing IL-6–deficient B cells (unpublished data). These results underscore the importance of IL-6 production by activated B cells in Th17 differentiation in vivo in rhMOG-induced EAE. Interestingly, a recent study evaluated B cell IL-6 production in MOG p35–55–induced EAE and observed that B cell IL-6 production was not required for initiation of EAE, but that the disease course was reduced in mice containing IL-6–deficient B cells (Barr et al., 2012). Thus, B cell IL-6 deficiency may have a variable effect when EAE is induced by MOG p35–55, but it is clear that the impact of B cell IL-6 production is greater in rhMOG-induced EAE. As rhMOG is currently the only known B cell–dependent encephalitogen, it may be advantageous to examine this myelin protein when evaluating the pathogenic role of activated myelin-specific B cells in EAE.

In this study, selective MHC II deficiency in B cells was evaluated using mixed BM chimera mice reconstituted with B cells from either WT mice or from MHC II–deficient mice, ensuring in the latter a complete absence of B cell MHC II expression and MHC II–restricted Ag presentation. As an alternative, we also examined selective B cell MHC II deficiency in CD19-targeted MHC II–deficient (CD19cre–MHC II^flox/flox) mice (Shimoda et al., 2006). It is known that ∼95% of B cells in CD19cre–MHC II^flox/flox mice are MHC II deficient (Shimoda et al., 2006). Although a minority of B cells expressed MHC II molecules, CD19cre–MHC II^flox/flox mice were also resistant to EAE induced with rhMOG (unpublished data). Thus, with both approaches, the results were consistent in demonstrating the in vivo requirement for B cell MHC II expression for EAE induced by this myelin protein. Although selective MHC II deficiency eliminated B cell APC function, it was also associated with a reduction in MOG-specific antibodies. One could contend that the absence of B cell MHC II expression simply eliminated the humoral response, which itself was primarily responsible for EAE pathogenesis. In our study, B cell MHC II deficiency resulted in a reduction in Tfh cells, a subset of CD4⁺ T cells that participate in germinal center formation, generation of memory B cells, and long-lived plasma cells (Crotty, 2011). In this regard, B cell APC function is a prerequisite for humoral immunity as Ag presentation by responding B cells is required for development and maintenance of Tfh, especially when Ag is limiting (Deenick et al., 2010). Nevertheless, it should be recognized that transfer of MOG-specific antibodies alone, or Tg-driven secretion of MOG-specific antibodies alone, does not lead to CNS inflammation and EAE (Linington et al., 1988; Genain et al., 1995; Litzenburger et al., 1998), and, as we have observed, transfer of pathogenic anti-MOG antibodies into B–MHC II^−/− mice that were immunized with rhMOG led to a partial restoration of EAE. Furthermore, despite MHC II expression on myeloid APCs, including DCs, there was a reduction in development of MOG-reactive Th17 cells in mice selectively deficient in B cell MHC II expression, implying direct B–T interaction in vivo.

IgH “membrane-only” BCR Tg mice can distinguish B cell Ag presentation in vivo from humoral immunity (Chan et al., 1999). Although not targeted to a defined self-Ag, IgH^B1-8-mem Tg mice have served to distinguish the contribution of B cells in Ag presentation from their role in the generation of antibodies in murine lupus and arthritis (Chan et al., 1999; O’Neill et al., 2005). To completely dissociate B lymphocyte cellular function from the potential contribution of humoral autoimmunity in EAE pathogenesis, we generated IgH^MOG-mem/J_HT Tg mice. B cells from our IgH^MOG-mem/J_HT mice efficiently presented MOG protein to MOG-specific 2D2 T cells and, despite the inability to secrete MOG-specific IgG or IgM, these mice were completely susceptible to acute clinical and histological EAE induced by rhMOG. Consistent with observations that B cells are more efficient APCs when they recognize the same target Ag as the responding T cells (Lanzavecchia, 1985; Constant et al., 1995a,b), we observed that B cells from IgH^MOG-mem/J_HT mice presented rhMOG to 2D2 T cells and promoted proinflammatory Th1 and Th17 differentiation more efficiently than B cells from IgH^B1-8-mem/J_HT mice, which bound rhMOG less effectively. Furthermore, in contrast to IgH^MOG-mem/J_HT mice, IgH^B1-8-mem/J_HT mice were less susceptible to EAE induction by rhMOG. Collectively, these in vitro and in vivo findings demonstrate how myelin-specific B cells can have a critical role in presentation to myelin-specific T cells in CNS autoimmunity. Although onset of acute EAE was similar in IgH^MOG-mem/J_HT, IgH^MOG-ki, and WT mice, it was of particular interest that the chronic phase appeared less severe in IgH^MOG-mem/J_HT mice than either IgH^MOG-ki or WT mice. Indeed, greater severity could reflect contribution of pathogenic MOG-specific antibodies elicited by immunization of IgH^MOG-ki or WT mice with rhMOG. However, given minor differences observed in the disease course, it may be experimentally challenging to dissect the contribution of anti-MOG antibodies in this model.

With the recent introduction of new agents for relapsing-remitting MS, the unmet need for successful therapy of progressive MS has become more evident (Fox et al., 2012; Brück et al., 2013). Understanding of the genetic and pathogenic mechanisms involved in relapsing-remitting MS may also be advancing more rapidly (Sawcer et al., 2011). Furthermore, it is recognized that there is a paucity of models for progressive MS (Amor et al., 1994; Bettelli et al., 2006a; Krishnamoorthy et al., 2006; Basso et al., 2008). Although models do not entirely reflect spontaneous human disease, they can provide insight regarding specific pathogenic mechanisms. Meningeal B cell follicle–like structures have been observed in SPMS and in IgH^MOG-ki × TCR^MOG mice with OSE. IgH^MOG-mem/J_HT and IgH^MOG-ki mice are complementary; whereas IgH^MOG-mem/J_HT mice cannot produce any antibodies, IgH^MOG-ki mice can secrete all Ig isotypes. Here, by analysis of IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice, we observed it is the cellular participation of MOG-reactive B lymphocytes that is necessary and sufficient for the development of spontaneous OSE and formation of meningeal follicle–like structures. Clinical signs of OSE appeared less severe in IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice than in IgH^MOG-ki × TCR^MOG mice and the onset occurred later, suggesting that MOG-specific antibodies may augment the phenotypic expression. Thus, IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice should provide a foundation to evaluate which individual isotypes of MOG-specific antibodies cooperate with MOG-specific B and T cells in OSE. IgH^MOG-mem mice provide a powerful reagent to directly assess the contribution of select genes that may participate in B–T communication or development of OSE and formation of meningeal follicle–like structures without influence of soluble antibodies or potential antibody–Ag complexes. In summary, our observations in this study underscore the central importance of MHC II–dependent B cell Ag presentation in CNS autoimmunity. Use of IgH^MOG-mem/J_HT mice to study development of spontaneous OSE should provide further insight regarding peripheral and CNS involvement of myelin-specific B cells in MS progression.

WT C57BL/6J CD45.1 mice as well as CD45.2 MHC II^−/− (B6.129S2-H2dlAb1-Ea/J), IL-6^−/− (B6.129S2-Il6^tm1Kopf/J), and CD19cre/cre (homozygous for cre cassette, B6.129P2(C)-Cd19^tm1(cre)Cgn/J) mice were purchased from the Jackson Laboratory. C57BL/6J B cell–deficient J_HT mice (B6.129P2-Igh-Jtm1Cgn/J; Gu et al., 1993) were provided by K. Rajewsky (Harvard University, Boston, MA), and C57BL/6J MOG p35–55–specific TCR Tg (2D2, TCR^MOG) mice (Bettelli et al., 2003) were provided by V.K. Kuchroo (Harvard University). C57BL/6J MOG-BCR knockin (IgH^MOG-ki, also referred to as Th) mice (Litzenburger et al., 1998) were provided by H. Wekerle. C57BL/6 IL-6^flox/flox mice generated by gene targeting technology (Quintana et al., 2013) were provided by J. Hidalgo. For selective deletion of floxed IL-6 gene in B lymphocytes, CD19cre/cre mice were crossed with IL-6^flox/flox mice. Litters heterozygous for the cre cassette (CD19cre–IL-6^flox/flox, CD19cre–IL-6^wt/wt) were used for experiments. IgH^B1-8 (IgH^B1-8-mem) mice (Chan et al., 1999) were backcrossed onto the C57BL/6J background and provided by J.-A. Lyons (University of Wisconsin, Milwaukee, WI) and A.H. Cross (Washington University School of Medicine in St. Louis, St. Louis, MO). IgH^B1-8-mem mice were crossed onto the J_HT background, and IgH^MOG-ki mice were crossed onto the MHC II^−/− background. Animals were housed in a specific pathogen–free barrier facility at the Parnassus Research Center of the University of California, San Francisco (UCSF). All breeding and experiments were reviewed and approved by the UCSF Institutional Animal Care and Use Committee.

To generate IgH^MOG-mem mice, the genomic DNA fragment containing the rearranged heavy chain VDJ of the pathogenic MOG-specific Ig 8-18C5 mAb (Litzenburger et al., 1998) was inserted into the IgM^a constant region of a vector from which the secreted exon and polyadenylation site were excised (IgM^-mem), which permits expression of membrane-bound Ig only (Chan et al., 1999). Specifically, we excised the 3.6-kb fragment containing the rearranged heavy chain enhancer and rearranged VDJ gene segment of the pathogenic MOG-specific Ig by digestion with EcoRI. This fragment was then subcloned into the EcoRI sites in the IgM^-mem only construct. Correct orientation was determined first by PCR and restriction analysis and then confirmed by DNA sequencing of the regions containing the integration sites. The IgH^MOG-mem construct was microinjected into C57BL/6J oocytes, which provided separate founders that were identified by PCR and confirmed by Southern blot. All founders contained B cells that expressed transgene-directed cell surface IgM^a and bound MOG. IgH^MOG-mem mice were crossed to C57BL/6J J_HT mice to generate IgH^MOG-mem/J_HT mice.

TCR^MOG-mem mice were crossed onto the J_HT background to generate TCR^MOG/J_HT mice. IgH^MOG-mem/J_HT and IgH^B1-8-mem/J_HT mice were intercrossed to TCR^MOG/J_HT mice to generate IgH^MOG-mem/J_HT × TCR^MOG/J_HT and IgH^B1-8-mem/J_HT × TCR^MOG/J_HT mice. For comparison of IgH^MOG-mem/J_HT × TCR^MOG/J_HT mice, IgH^MOG-ki mice were crossed with TCR^MOG mice to generate IgH^MOG-ki × TCR^MOG mice (Bettelli et al., 2006a; Krishnamoorthy et al., 2006).

1,200 cGy γ-irradiation (Cs source) was administered to 6-wk-old recipient C57BL/6J mice. On the same day, recipients received 5 × 10⁶ donor BM cells. All BM preparations were depleted of T cells by labeling with a biotinylated anti–Thy-1 (clone T24) before incubation with streptavidin-microbeads (Miltenyi Biotec) for negative selection via magnetic-activated cell sorting (MACS) column (Miltenyi Biotec). To restrict genetic MHC II deficiency to B cells, the BM inocula consisted of 80% BM from J_HT mice supplemented with 20% BM from MHC II–deficient mice. The control group received BM inocula consisting of 80% BM from J_HT mice supplemented with 20% BM from WT mice. In certain experiments, chimeras were generated with 80% BM from J_HT mice and 20% BM from either IgH^MOG-ki/MHC II^−/− or IgH^MOG-ki mice. BM chimeras were housed for at least 8 wk to reconstitute their peripheral lymphoid system before use in EAE experiments. FACS analysis confirmed hematopoietic reconstitution of sublethally irradiated C57BL/6J CD45.1 mice with CD45.2⁺ donor BM for expression of CD45.2 on peripheral blood mononuclear cells. Engraftment of donor CD45.2⁺ cells was ≥ 95%. The BM chimera containing MHC II–deficient B cells showed numbers of CD19⁺ splenocytes equivalent to numbers found in normal WT mice.

Mouse MOG p35–55 (MEVGWYRSPFSRVVHLYRNGK) was synthesized by Auspep. rmMOG was provided C.C.A. Bernard. rhMOG was provided by C.C.A. Bernard and D.E. Jenne. MOG proteins were synthesized, purified, and refolded as previously reported (Clements et al., 2003; Breithaupt et al., 2008).

C57BL/6J mice were injected s.c. with 50 µg MOG p35–55 or 100 µg rmMOG or 100 µg rhMOG emulsified in CFA (DIFCO Laboratories) containing 200 µg heat-killed Mycobacterium tuberculosis (Mtb) H37RA (DIFCO Laboratories) on day 0. Additionally, mice received i.v. 200 ng Bordetella pertussis toxin (List Biological Laboratories) in 0.2 ml PBS on days 0 and 2. Individual animals were observed daily, and clinical scores were assessed with a 0- to 5-point scoring system, as follows: 0 = no clinical disease, 1 = loss of tail tone only, 2 = mild monoparesis or paraparesis, 3 = severe paraparesis, 4 = paraplegia and/or quadriparesis, and 5 = moribund or death. Moribund mice were given disease severity scores of 5 and euthanized.

IgH^MOG-ki mice were immunized with rmMOG (100 µg/mouse). 14–16 d later, mice were bled via cardiac puncture. Blood was allowed to clot, and the serum was removed after centrifugation and frozen at −80°C until use. The presence of high-titer serum antibodies was confirmed by ELISA (described below). The 8-18C5 hybridoma cell line was provided by C. Linington. An IgG1 mouse anti-OVA hybridoma cell line (1B7.11) was purchased from UCSF hybridoma core. Hybridoma cells were cultured in a BD Cell mAb Medium Serum-Free (BD) in large-scale flasks (Corning). mAbs were purified from hybridoma culture supernatants by the UCSF hybridoma core. The concentrations of the proteins were determined by means of the BCA protein assay (Thermo Fisher Scientific), according to the manufacturer’s recommendations.

B–MHC II^+/+ and B–MHC II^−/− mice were immunized with 100 µg rhMOG emulsified in CFA. 200 ng B. pertussis toxin was injected on days 0 and 2. When B–MHC II^+/+ mice showed clinical EAE signs (score ∼2.0), B–MHC II^−/− mice received i.p. injections of 150 µg anti-MOG IgG1 mAb 8-18C5, an equal amount of isotype control OVA-specific mAb 1B7.11, or 250 µl IgH^MOG-ki serum, or non-Tg littermate mice serum three times at 48 h-intervals (Schluesener et al., 1987; Pöllinger et al., 2009). Sera were obtained through cardiac puncture from IgH^MOG-ki or non-Tg littermate mice.

Total serum MOG-specific IgG was quantified using a commercially available anti–mouse MOG (1–125) IgG quantitative ELISA kit (SensoLyte; AnaSpec). Wells were precoated with rmMOG (1–125) protein. Results are expressed as IgG concentrations in µg/ml serum (Figs. 3 A and 4 E) or, to facilitate comparison with other Ig measurements, as OD of serial twofold dilutions (Fig. 4 D). Serum MOG-specific IgM antibodies were measured using a noncommercial ELISA. 96-Maxisorb plates (Costar) were precoated with rmMOG (1–117) protein (10 µg/ml in PBS), blocked with BSA (Sigma-Aldrich), and incubated with sera for 2 h at the indicated dilution. After washing, MOG-specific IgM retained by the plate-bound MOG was detected with horseradish-peroxidase–conjugated anti–mouse IgM (IMMUNO-TEK). Total IgM and IgG were measured using quantitative Ig ELISA kits (IMMUNO-TEK). Total serum IgG or IgM was captured by anti–mouse IgG or IgM antibodies immobilized on micro wells. OVA-coated plates were used as negative controls for nonspecific binding. SOFTmax ELISA plate reader (450-nm wavelength) and software (Molecular Devices) were used for data analysis.

Cells were grown in RPMI supplemented with 10% FCS, β-mercaptoethanol, l-glutamine, and penicillin/streptomycin. For thymidine incorporation, either 2.5 × 10⁵ cells/ml of total splenocytes or 10⁴ cells/ml of purified CD4⁺ T cells plus 2.5 × 10⁵ cells/ml of purified B220⁺ B cells were cultured in 96-well plates in the presence of different concentrations of the rMOG proteins or MOG p35–55 (range 1–50 µg/ml). After 72 h, cells were pulsed with [³H]thymidine and harvested 16 h later. Thymidine incorporation is expressed as cpm. In certain experiments, T cell proliferation was assessed by CFSE dilution assay after gating on CD4⁺ T cells.

Single-cell leukocyte suspensions from spleens and peripheral lymph nodes (paired axillary and inguinal) were generated by gentle dissection. CNS mononuclear cells were isolated from EAE mice at peak of disease after cardiac perfusion with PBS, as described previously (Zeine and Owens, 1992). In brief, minced CNS (spinal cord) tissues were digested with 2.5 mg/ml collagenase D (Roche) at 37°C for 45 min. Mononuclear cells were isolated by passing the tissue through 70-mm cell strainers (BD), followed by discontinuous Percoll gradient (70/30%; Sigma-Aldrich) centrifugation. Lymphocytes were collected from the 30:70% interface and washed. Total cell numbers were determined by counting on a hemocytometer, and viability was assessed by trypan blue exclusion.

MACS (Miltenyi Biotec) was used to purify lymphocyte populations according to the manufacturer’s instructions. CD4⁺ T cells were isolated by depletion of non-CD4⁺ T cells (negative selection, kit #130-090-860). Naive B cells were isolated by depletion from CD43-expressing B cells (activated B cells, plasma cells and CD5⁺ B-1a cells) and non-B cells (kit #130-090-862) according to the manufacturer’s instructions. When necessary, cells were enriched a second time using a fresh MACS column to obtain >95% cell purities, as assessed by flow cytometry.

For two- to six-color immunofluorescence analysis, single-cell suspensions (10⁶ cells) were incubated with anti–mouse FcRIIB/FcRIIIA mAb (2.4G2; BD) to avoid nonspecific staining and were subsequently stained at 4°C using predetermined optimal concentrations of mAb for 30 min, as described previously(Weber et al., 2010). Blood erythrocytes were lysed before staining using FACS lysing solution (BD). Antibodies to the mouse proteins CD45R (B220), FITC/PE/APC/PE-Cy7/PerCP-Cy5.5/APC-Cy7 (RA3-6B2), CD4 FITC/PE (RM4-4), CD4 APC/PE-Cy7/PerCP-Cy5.5/APC-Cy7 (RM4-5), CD11c-APC (N418), CD11b-PE (ICRF44), CD44-PE (IM7), CD62L-APC (MEL-14), IgM-FITC (eB121-15F9), and IgD-FITC (11-26c) were purchased from eBioscience. MHC II I-Ab-FITC (AF6-120.1) was purchased from BD. Isotype- and concentration-matched control antibodies were used to assess nonspecific staining. In certain experiments, cells were fixed in 0.3% paraformaldehyde and kept at 4°C until analysis. Lymphocytes were examined by forward and side light scatter on a FACSCanto II flow cytometer using FACSDiva software (BD).

Lymphocytes were stimulated in vitro with 50 ng/ml PMA (Sigma-Aldrich) and 1 µg/ml ionomycin (Sigma-Aldrich) in the presence of 1 µl/ml brefeldin A (eBioscience) for 6 h before staining. FcRs were blocked before cell surface staining with anti–CD4-PerCP-Cy5.5 (RM4-5; eBioscience). After staining, cells were washed, fixed, and then permeabilized using the Cytofix/Cytoperm Plus Fixation/Permeabilization kit (BD) according to the manufacturer’s instructions. Permeabilized cells were then stained with anti–IL-17–PE-Cy7 (eBio17B7) and anti–IFN-γ–FITC (XMG1.2; all from eBioscience).

For detection of Bcl-6, FcRs were blocked followed by cell surface staining with anti–CD4-PE (GK1.5), anti–PD-1 (CD279)–PE-Cy7 (J43), and anti–ICOS (CD278)-FITC (C398.4A; all from eBioscience) and anti CXCR5-PE (2G8; BD). Bcl-6 staining was subsequently performed using the Foxp3/Transcription Factor Staining Buffer set (eBioscience) and anti–Bcl-6-PE (mGI191E; eBioscience).

Brains, spinal cords, and optic nerves were removed and fixed in 10% neutral-buffered formalin, paraffin-embedded, and sectioned. Representative sections were stained with Luxol fast blue (LFB)–hematoxylin and eosin (H&E; for inflammation and demyelination) and reticulin preparation (for connective tissue) and examined by light microscopy. Meningeal and parenchymal inflammatory foci (>10 clustered inflammatory cells) were counted by a blinded observer (R.A. Sobel). Standard avidin-biotin immunohistochemical staining was performed on the sections with rabbit anti–mouse CD3 (Abcam) and rat anti–mouse CD45R (B220; BD), using reagents from Vector Laboratories. Normal mouse spleen tissue served as positive staining controls; negative controls included omission of the primary antibody. Neuropathologic analysis was performed by R.A. Sobel.

Data are presented as mean ± SEM or SD. For clinical scores, significance between groups was examined using the Mann–Whitney U test. A value of P < 0.05 was considered significant. All other statistical analysis was performed using a one-way multiple-range ANOVA test for multiple comparisons. A value of P < 0.01 was considered significant.

We thank Dr. Patricia Nelson (University of California, San Francisco) for helpful discussion and Kara Pekarek for technical assistance. We thank Drs. Jeri-Anne Lyons and Anne H. Cross for providing us with the IgH^B1-8-mem mice.

N. Molnarfi was supported by an advanced researcher fellowship from the Swiss National Science Foundation (SNF, PA00A-119532) and is a recipient of an advanced researcher exchange 2011 fellowship from the European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS) Foundation. U. Schulze-Topphoff is a fellow of the National Multiple Sclerosis Society (NMSS) and the Deutsche Forschungsgemeinschaft (DFG; SCHU 2587/1-1). M.S. Weber was supported by the NMSS (RG 4450A1/T). T. Prod’homme received fellowship support from the NMSS and Teva Neuroscience. M. Varrin-Doyer is a fellow of the NMSS. J. Hidalgo was supported by SAF2011-23272. Support for this work was provided to S.S. Zamvil by the National Institutes of Health (NIH; R01 AI-073737 and R01 NS-063008), the NMSS (RG 3913 and RG 4768), the Guthy Jackson Charitable Foundation, the Dana Foundation, and the Maisin Foundation and to M.J. Shlomchik by the NIH (R01 AI-043603).

The authors declare no competing financial interests.

Author contributions: N. Molnarfi, U. Schulze-Topphoff, and S.S. Zamvil designed research, analyzed data, and wrote the paper; M.S. Weber, J.C. Patarroyo, M. Varrin-Doyer, T. Prod’homme, and A.J. Slavin analyzed data, gave conceptual advice, and discussed the results at all stages; N. Molnarfi, U. Schulze-Topphoff, J.C. Patarroyo, and T. Prod’homme performed the experiments; M. Varrin-Doyer and A. Shetty assisted in the performance of experiments; R.A. Sobel performed the neuropathologic analysis; C. Linington, J. Hidalgo, D.E. Jenne, H. Wekerle, C.C.A. Bernard, and M.J. Shlomchik contributed new reagents; and S.S. Zamvil supervised the study. All authors read, commented on, and approved the final manuscript.