Regulatory T Cell Dysfunction Acquiesces to BTLA⁺ Regulatory B Cells Subsequent to Oral Intervention in Experimental Autoimmune Encephalomyelitis

B cells have been classically considered a key player in the immune response against microbial infections, mainly through the production of Abs. B cells also have been shown to have a role in the pathogenesis of autoimmune disorders. New clinical interventions have targeted the elimination of B cells for the treatment of rheumatoid arthritis (1), lupus (2), and multiple sclerosis (MS) (3). However, an increasing body of evidence supports the existence of a subpopulation of anti-inflammatory IL-10–producing B cells, which critically affect the outcome of autoimmune diseases both in animal models (4–8) and in human clinical disease (9–12).

Experimental autoimmune encephalomyelitis (EAE) is a T cell–dependent inflammatory disease, principally mediated by IL-17 (13) and GM-CSF (14). EAE is highly reproducible in mice after immunization of susceptible strains with TCR-reactive peptides and mimics many aspects of MS (15). Although EAE is predominantly a T cell–mediated disorder, B cells have been shown to play an important pathogenic role in this disease (reviewed by Gray et al. in Ref. 16). On one hand, autoantibody production may increase demyelination and inflammation, worsening the course of the disease (17, 18). In addition, B cells may function as APCs in early phases of EAE, contributing to the expansion of effector CD4⁺ T cells (19, 20). On the other hand, B cell–deficient mice develop a more aggressive form of EAE (21–24), and IL-10 production by B cells inhibits the lymphocyte response against autoantigens and thus ameliorates EAE (6, 7, 20, 25, 26). As suggested by Matsushita et al. (6), this contradiction may be explained if the existence of regulatory B cells (Bregs) is taken into account. Although pathogenic B cells increase severity of the disease, Bregs can ameliorate symptoms. Therefore, a balance between pathogenic B cells and Bregs can affect the severity of EAE as well as other autoimmune diseases.

The induction of tolerance in humans is problematic, but some success has been achieved for treating allergies (27, 28) and diabetes (29). Oral tolerance in humans is plagued by the lack of effective means to absorb ingested tolerogens from the gastrointestinal tract. Hence, we postulated that targeting tolerogens or seeking alternative sites of treatment, e.g., nasal, offered a means to implement tolerance. We found that reovirus protein σ1 (pσ1) could facilitate uptake of genetically fused tolerogens, when applied mucosally, and stimulate the induction of tolerogen-specific regulatory T cells (Tregs) (30–33). Pσ1 effectively stimulated Ag-specific tolerance to defined Ags: OVA, proteolipid protein (PLP) peptide 139–151, and myelin oligodendrocyte glycoprotein (MOG_29–146) (30–33). PLP_139–151 fused to OVA-pσ1 protected mice against PLP_139–151-induced EAE whereas OVA-pσ1 did not (33). MOG-pσ1 protected against MOG_35–55-induced EAE, but not OVA- pσ1, MOG, or MOG + pσ1 (32), further demonstrating that the tolerogens must be physically coupled to pσ1 to elicit Ag-specific tolerance. In this study, we show that treatment of EAE with MOG-pσ1 increased the percentage and numbers of Bregs and their production of IL-10. This treatment provides the regulatory balance needed to resolve disease.

Herpes virus entry mediator (HVEM) belongs to the TNF receptor family. It regulates the immune response in various pathogen and autoimmune settings through interactions with multiple ligands such as Ig superfamily members, B and T lymphocyte attenuator (BTLA) and CD160, as well as the TNF superfamily ligands LIGHT (lymphotoxin-related inducible ligand that competes for glycoprotein D binding to herpesvirus entry mediator on T cell) and LTα (34, 35). These receptors act bidirectionally having costimulatory or coinhibitory properties. For instance, trans engagement of HVEM with BTLA or LIGHT is costimulatory for the HVEM⁺ cells and promotes cell survival via NF-κB activation (36). In addition, HVEM engagement of LIGHT potently activates T cells which can lead to T cell–mediated intestinal inflammation (37–39). In contrast, HVEM signaling through BTLA results in the inactivation of the BTLA⁺ cells mediated by SH2 domain–containing protein tyrosine phosphatase 1 and 2 (40, 41).

Although large numbers of Tregs are found at sites of chronic inflammation, these fail to curb disease (42–45). To address this paradox, we sought to determine the activation status of Bregs because these cells were induced subsequent to MOG-pσ1 treatment. We found elevated numbers of BTLA⁺ Bregs following MOG-pσ1 treatment. Neutralization of BTLA on Bregs inhibited their capacity to restore suppressive function to EAE Tregs. Thus, we hypothesize that restoration of Treg function is BTLA dependent. Supporting this hypothesis, adoptive transfer of MOG-pσ1–induced IL-10–producing B220⁺(CD19⁺)CD5⁺, but not B220⁺(CD19⁺)CD5^neg cells, significantly increased the percentage and activity of CD25⁺CD4⁺ Tregs, which reduced EAE severity. Furthermore, these Bregs were protective against EAE in BTLA^−/− mice. Such findings underscore the relevance of Bregs and their potential benefit in treating human autoimmune diseases such as MS.

MOG-pσ1 and OVA-pσ1 fusion proteins were prepared as previously described (31, 32) consisting of MOG_29–146 or OVA genetically fused to the N terminus of the whole reovirus pσ1. These recombinant proteins were generated in Pichia pastoris featuring a his-tag to enable their purification (31, 32). Proteins were assessed for purity and quality by Coomassie-stained polyacrylamide gels and by Western blot analysis.

C57BL/6 females (6–8 wk old; Frederick Cancer Research Facility, National Cancer Institute, Frederick, MD) and breeding colonies of Foxp3-mRFP-transgenic, B cell deficient (μMT), IL-10^−/−, and BTLA^−/− mice (The Jackson Laboratory, Bar Harbor, ME) were maintained at Montana State University Animal Resources Center or the University of Florida Animal Center Services. Mice were kept in individual ventilated cages under high-efficiency particulate absorbing-filtered barrier conditions. All animal care and procedures are in compliance with institutional policies for animal health and well-being.

Mice were given a single oral 50-μg dose of MOG-pσ1 or OVA-pσ1 as described previously (30). For EAE induction, mice were challenged s.c. in the flank with 300–350 μg MOG_35–55 peptide (Global Peptide Services, Ft. Collins, CO or Bio-Synthesis, Lewisville, TX) in 100 μl IFA (Sigma-Aldrich, St. Louis, MO) containing 400 μg killed Mycobacterium tuberculosis (Difco Laboratories, Detroit, MI) on day 0 followed by i.p. treatments with 200 ng Bordetella pertussis toxin (List Biological Laboratories, Campbell, CA) on days 0 and 2. Mice were monitored and scored daily for EAE progression (30, 33): 0, normal; 1, a limp tail; 2, hind limb paresis; 3, hind limb paralysis; 4, quadriplegia; and 5, moribund state. In most instances, we observed that the PBS-treated EAE C57BL/6 mice made a slow recovery, and generally about days 40–45, exhibited a clinical score of ≤1, and these are referred to as “naturally recovered” EAE mice.

Prior to dissection, mice were perfused with cold PBS through the left cardiac ventricle. Subsequently, the forebrain and cerebellum were dissected, and the spinal cords (SCs) flushed out with PBS by hydrostatic pressure, minced, and digested with 10 ml HBSS containing 500 U/ml collagenase (Sigma-Aldrich) for 60–90 min at 37°C with shaking to release mononuclear cells. The tissues were then homogenized in cold HBSS buffer (Sigma-Aldrich) as previously described (30) to obtain a single-cell suspension. Centrifuged cells collected from four to five mice were pooled, resuspended in 70% Percoll (Sigma-Aldrich), and underlaid with 30% Percoll. Cells were subjected to gradient density centrifugation, and mononuclear cells were isolated from the 30/70 interphase, then washed, and stained for flow cytometry analysis.

To block nonspecific FcR binding, lymphocytes were preincubated for 15 min at room temperature with anti-CD16/32 FcR block and stained 30 min in FACS buffer at 4°C with fluorochrome-labeled mAbs against B220, CD1d, CD4, CD5, CD19, CD25, and BTLA (clone 8F4; BioLegend). For intracellular expression of cytokines, lymphocytes were assessed stimulated for 5 h with PMA (50 ng/ml) plus ionomycin (500 ng/ml) in the presence of brefeldin A (2 μg/ml), followed by permeabilization with saponin and subsequent staining with mAbs specific IL-10 (clone JES5-16E3) or IL-17 (clone eBio17B7). Nuclear Foxp3 expression by Tregs was measured by FACS analysis, according to the manufacturer’s instructions (Foxp3 Fix/Perm buffer set; BioLegend). Cells were analyzed using a LSRFortessa flow cytometer (BD Biosciences, San Jose, CA) and analyzed with Flow Jo software (Tree Star, Ashland, OR).

Single-cell splenic and lymph node (LN) suspensions were prepared as described previously (29, 30). CD25⁺CD4⁺ Tregs and CD25⁻CD4⁺ T effector (T_eff) cells were purified using a FACSAria (BD Biosciences) and 50,000 T_eff cells were stimulated for 4 d on anti-CD3 mAb-coated wells (5 μg/ml) plus soluble anti-CD28 mAb (5 μg/ml) plus 5 U/ml human IL-2 (28). Bregs were purified as described below. Splenic Tregs and Bregs were added when indicated at a 1:2 ratio. In some instances, Bregs or Tregs were pretreated (10 μg/ml) with anti-BTLA (clone 6F7; eBioscience) or isotype control mAb when indicated. One microcurie per well of [³H]thymidine was added during the last 12 h of culture. The cells were then harvested, and incorporated [³H]thymidine was counted using a Beckman LS 6500 scintillation counter. For some experiments, purified CD4⁺ T cells were restimulated for 4 d with 5 μg/ml MOG_35–55 peptide in the presence of irradiated T cell–depleted splenic feeder cells (30–32).

Supernatants from cultured T cells were analyzed for the production of IFN-γ, TGF-β, IL-4, IL-6, IL-10, and IL-17 by ELISA as described previously (32).

Cell sorting was accomplished using a FACSAria as described previously (30, 32). Briefly, splenic and LN (head and neck LN [HNLN] and mesenteric LN [MLN]) cells from C57BL/6 and IL-10^−/− mice were sorted for Bregs or B cells based on the expression of B220 (or CD19) and CD5. Cell purity exceeded 95%. Sorted cells were kept at 4°C in PBS with 2% FBS until adoptive transfer. Cells were periorbitally transferred into anesthetized mice, and the amount transferred is indicated in each experiment.

The ANOVA followed by posthoc Tukey test was applied to show differences in clinical scores in treated versus PBS mice. The Student t test was used to evaluate the differences between two groups, and ANOVA was used to test differences between multiple groups. The p values < 0.05 were considered statistically significant.

Previous studies have demonstrated the presence of Tregs during EAE and other inflammatory diseases, yet these remained dysfunctional or quiescent and did not reduce inflammation (42–45). Our previous work has shown that the inflammation in EAE mice is rapidly ameliorated with a single low nasal or oral dose of MOG-pσ1 (30, 32). Given this, purified splenic naive Tregs, EAE-derived Tregs from mice at disease onset (day 7 postchallenge [p.ch.], score 1), or Tregs from 15 d after EAE induction (score 3) were added to anti-CD3 plus anti-CD28–stimulated naive T_eff cells to measure inhibition of T_eff cell proliferation. Naive Tregs efficiently inhibited T_eff cell proliferation whereas EAE Tregs did not (Fig. 1A). Tregs obtained from EAE mice 24 h after treatment with MOG-pσ1 (day 16 p.ch.) were also quite effective in suppressing T_eff cell proliferation. Interestingly, Tregs obtained from naturally recovered mice (day 40 p.ch., score 0.5) were as effective as naive Tregs in suppressing T_eff cell proliferation (Fig. 1A), suggesting these Tregs have regained their function.

MOG-pσ1’s therapeutic effect has been shown to be IL-10 dependent (32). To determine the source of IL-10–producing cells, splenic lymphocytes taken from EAE mice previously treated with PBS or MOG-pσ1 were evaluated for their expression of TGF-β and IL-10. In addition to CD4⁺ T cells producing IL-10 (32), a population of IL-10–producing B cells was also found in MOG-pσ1–treated EAE mice (Fig. 1B). B lymphocytes isolated from EAE mice did not produce TGF-β regardless of the treatment (Fig. 1B). MOG-pσ1 did significantly increase the percentage and total numbers of IL-10–producing B cells (Fig. 1B). Further studies set out to assess the impact of MOG-pσ1 on B cells. C57BL/6 and B cell–deficient (μMT) mice were induced with EAE and, 7 d later, orally treated with 50 μg MOG-pσ1 or PBS (Fig. 1C). As expected (24, 46), μMT mice developed a more severe and lingering disease, yet treatment of μMT mice with MOG-pσ1 reduced the disease severity (clinical score) by nearly 50%, showing that MOG-pσ1’s therapeutic effect is partially dependent on B cells.

Both CD5⁺CD1d^Hi (7, 47) and CD5⁺ (7) B cell subsets can serve as Bregs. An analysis was performed to determine which of these Bregs was induced following MOG-pσ1 stimulation. When compared with naive mice, a reduced presence of CD5⁺ B cells was observed in the spleen (Fig. 2A, 2B) and in the HNLNs (Fig. 2C). Interestingly, treatment of EAE mice with MOG-pσ1, but not OVA-pσ1, restored the CD5⁺ B cells to levels similar to those of naive mice (Fig. 2B). In contrast to CD5⁺ B cells that first decreased in diseased EAE mice and then increased either subsequent to MOG-pσ1 treatment or following natural recovery, the B220⁺CD5⁺CD1d^Hi cells could be detected (data not shown) but did not increase as notably as did the CD5⁺ B cells following MOG-pσ1 treatment (Fig. 2A–C). Thus, from this point forward, the described Bregs will be referred to as CD5⁺ (B220⁺ or CD19⁺) B cells. During EAE, the levels of CD5⁺ B cells diminished, and MOG-pσ1 treatment essentially restored to basal levels in the spleens (Fig. 2B) and the total number of CD5⁺ B cells increased in the HNLNs (Fig. 2C). OVA-pσ1 treatment failed to restore CD5⁺ B cells in the spleen (Fig. 2B) but did increase their numbers in the HNLNs (Fig. 2C). These findings support the importance of Bregs in protection conferred by MOG-pσ1 in EAE.

The induced Bregs were further analyzed for their capacity to generate IL-10. EAE mice were evaluated 24 h following MOG-pσ1 or PBS treatment, and spleens, SCs, HNLNs, MLNs, and peripheral LNs (PLNs) were analyzed for intracellular IL-10 production (Fig. 2D). Only SC-infiltrating Bregs were able to produce IL-10 by the PBS-treated group, whereas MOG-pσ1 treatment increased Breg IL-10 production in the SCs, PLNs, and MLNs (Fig. 2E). Of interest, the total number of IL-10–producing Bregs in the SCs was significantly greater in MOG-pσ1–treated EAE than EAE mice (Fig. 2F). Thus, MOG-pσ1 treatment augments these IL-10–producing CD19⁺ (B220⁺) CD5⁺ Bregs during EAE.

To determine whether MOG-pσ1–induced Bregs are protective against EAE, B cells were isolated from C57BL/6 mice induced with EAE and treated 2 wk later with PBS or MOG-pσ1. Twenty-four hours after treatment, combined LN and splenic B cells were cell sorted into B220⁺CD5⁻ (inflammatory) B cells and B220⁺CD5⁺ Bregs, and each was then adoptively transferred into different C57BL/6 mice that were induced with EAE 7 d earlier (Fig. 3). Neither CD5⁻ nor CD5⁺ B cells obtained from PBS-treated donors affected clinical manifestation of EAE (Fig. 3A). Donor B220⁺CD5⁻ B cells from MOG-pσ1–treated mice also failed to ameliorate EAE; however, MOG-pσ1–induced B220⁺CD5⁺ Bregs significantly reduced disease severity and accelerated recovery (p < 0.01) (Fig. 3B). These findings underscore the importance of this IL-10–producing Bregs. The Breg subset, isolated from IL-10^−/− mice, failed to confer protection against EAE when adoptively transferred into C57BL/6 mice with EAE (Fig. 3C). This evidence further confirms the critical role of IL-10⁺ Bregs induced by MOG-pσ1 in mediating protection against EAE. Notably, Bregs from diseased EAE mice failed to confer such protection.

In an effort to understand how Bregs mediate their protection against EAE, both Bregs (B220⁺CD5⁺) and inflammatory B (B220⁺CD5⁻) cells were purified 48h after MOG-pσ1 treatment of EAE C57BL/6 mice. Subsequently, either of these B cell subsets was adoptively transferred into C57BL/6 mice induced with EAE 7 d earlier (Fig. 4A). At the peak of disease (day 14), the activity and quantity of Tregs were determined in recipients (Fig. 4B). Recipients given MOG-pσ1–induced Bregs showed an increase in the percentage of Tregs unlike those recipients given PBS-induced Bregs or MOG-pσ1–induced CD5⁻ B cells. Such evidence suggests that the MOG-pσ1–induced Bregs become activated and stimulate an influx of Tregs or in turn activate quiescent, disease-induced Tregs. The percentage of IL-17–producing T cells among the CD25^HiCD4⁺ and CD25⁻CD4⁺ subsets (48) was measured. Mice adoptively transferred with MOG-pσ1–induced Bregs showed significantly fewer IL-17–producing cells than those found in EAE mice and recipients adoptively transferred with Bregs from PBS-treated EAE mice (p ≤ 0.01) (Fig. 4C).

To determine which cytokines correlated with protection induced upon Breg adoptive transfer, cytokine analyses were performed on purified splenic CD25⁺CD4⁺ T cells isolated 7 d following adoptive transfer. Tregs obtained from mice adoptively transferred with MOG-pσ1–induced Bregs showed significant elevations in IL-4 and IL-10 when compared with control groups or recipients given PBS-induced Bregs (p < 0.01) (Fig. 4D) with concomitant reductions in IFN-γ and IL-17 (p < 0.01). These results demonstrate that MOG-pσ1-induced Bregs are protective by increasing and activating Tregs to produce regulatory cytokines and to simultaneously inhibit proinflammatory cytokine production.

Investigations of how Bregs become activated led to an examination of the role of the BTLA-HVEM-LIGHT pathway, since depending on the ligands’ interactions to their HVEM receptor, this pathway can either be activating or suppressing (34, 35). CD19⁺CD5⁺ B cells, from EAE mice treated with MOG-pσ1 (Fig. 5A), expressed BTLA (Fig. 5B, 5D, 5E), and these BTLA⁺ Bregs from MOG-pσ1–treated mice showed enhanced capacity for producing IL-10 unlike those BTLA⁺CD5⁻ B cells (Fig. 5C, 5E). Although IL-10 production was mostly associated with Bregs in all treatment groups, MOG-pσ1 treatment of EAE mice resulted in a much higher percentage of IL-10–producing BTLA⁺ Bregs than was seen in either naive mice or EAE mice treated with PBS or OVA-pσ1 (Fig. 5B, 5D). Thus, MOG-pσ1 enhances the frequency and the total number of IL-10–producing BTLA⁺ Bregs (Fig. 5B, 5D, 5E).

To determine whether BTLA expression on Tregs was modulated as a consequence of EAE or MOG-pσ1 treatment, these Tregs were also examined (Table I). The percentage of BTLA⁺ Tregs did not change significantly as a consequence of treating EAE mice with MOG-pσ1 or OVA-pσ1 (Table I). Likewise, BTLA expression remained unchanged for CD25⁻CD4⁺ T cells (Table I). To determine whether Bregs enhanced the functionality of EAE-derived Tregs and contributed to EAE recovery, a T cell suppression assay was performed. To assess their function, splenic Tregs were purified from either naive or EAE mice and cocultured with naive CD25⁻CD4⁺ T_eff cells in the presence or absence of MOG-pσ1–induced Bregs. Interestingly, CD5⁺ Bregs, but not CD5⁻ B cells, were able to completely restore the inhibitory activity of EAE-derived Tregs (Fig. 6A). Of note, Bregs required the presence of Tregs because Bregs alone had no inhibitory effect upon T_eff cells, suggesting that Bregs act through Tregs to facilitate their inhibitory activity (Fig. 6A).

To confirm the relevance of BTLA expression by Bregs, studies were performed to assess whether blocking BTLA on MOG-pσ1–induced Bregs impacts the function of EAE-derived Tregs. MOG-pσ1–induced Bregs were pretreated with an anti-BTLA mAb or its isotype control. Isotype control–treated MOG-pσ1–induced Bregs were able to restore Treg function as evidenced by suppression of CD4⁺ T cell proliferation (Fig. 6B). In contrast, MOG-pσ1–induced Bregs treated with an anti-BTLA mAb blocked Bregs from interacting with EAE Tregs and the Tregs remained dysfunctional. BTLA^−/− Bregs were unable to activate EAE-derived Tregs, confirming the importance of BTLA expression on Bregs.

BTLA^−/− mice induced with EAE exhibited significantly more severe clinical disease than did C57BL/6 mice (p = 0.027) (Fig. 7A), similar to what others have shown (49). Moreover, these BTLA^−/− mice were refractive to MOG-pσ1 treatment (Fig. 7A). The observed EAE exacerbation was not due to an intrinsic defect by naive BTLA^−/− Tregs because naive BTLA^−/− Tregs retained the ability to inhibit CD4⁺ T cell proliferation and were able to proliferate after MOG_35–55 stimulation (Supplemental Fig. 1A, 1B). Further analysis of these BTLA^−/− (Foxp3⁺CD25⁺) Tregs from EAE mice showed a failure to produce IL-10 relative to MOG-pσ1–treated BTLA-sufficient mice (Fig. 7B, 7C). Likewise, there were fewer IL-10–producing T_eff cells from BTLA^−/− mice when compared with C57BL/6 mice. Analysis of BTLA^−/− Bregs also showed their inability to upregulate IL-10 following MOG-pσ1 treatment (Fig. 7D, E). Thus, the absence of BTLA prevents the means to costimulate via BTLA, resulting in reduced IL-10 production by both Tregs and Bregs.

We next asked whether MOG-pσ1–induced Bregs could restore Treg function in BTLA^−/− mice. To further address Tregs’ dependency on BTLA⁺ Bregs, C57BL/6-derived BTLA⁺ Bregs were adoptively transferred into BTLA^−/− mice at the time of EAE induction, and recipients were monitored for development of disease. Consistent with the concept of requiring BTLA⁺ Bregs to complete Treg activation, BTLA^−/− recipients exhibited significantly less clinical disease (apparent by both delayed onset and reduced clinical score) than PBS-treated litter mates (**p ≤ 0.01) (Fig. 8A). In fact, very few Bregs were required because as few as 3 × 10⁴ Bregs could effectively dampen EAE.

To determine which cytokines correlated with protection induced after Bregs’ adoptive transfer, cytokine analyses were performed on purified splenic CD4⁺ T cells isolated at the peak of the disease (day 18 p.ch.). Cytokine production was significantly elevated in TGF-β and IL-6 as compared with PBS-treated mice with reduced IFN-γ and IL-17 production (*p ≤ 0.05, **p ≤ 0.01) (Fig. 8B). No differences in IL-10 production were observed. These results show that MOG-pσ1–induced BTLA⁺ Bregs have a critical role in maintaining Treg function, and the adoptive transfer of BTLA⁺ Bregs was sufficient to suppress inflammation and disease as evidenced by reduced clinical scores and less IFN-γ and IL-17 being produced.

MS, the primary cause of paralysis in young adults, is a neurodegenerative disease of the CNS that involves infiltration by activated inflammatory cells, damaging both myelin and axons (13). EAE, a rodent model for MS, recapitulates and provides experimental support for the basic disease mechanism in MS (50). The role of B cells in MS and EAE has generated increasing interest (12, 46, 51–55). Distinct B cell subtypes have recently been associated with EAE disease progression and regulation (6, 7, 16, 53, 56–59), and are at least partially dependent on the choice of Ag (protein or peptide) used to induce the disease (18). A role for B cells has also been suggested in different models of recombinant MOG (rMOG)-induced EAE (18, 60–62). In general, when rMOG was used to induce EAE, pathogenic B cells were induced, and their depletion resulted in less EAE severity (18, 60–62). However, even these studies varied in the type of rMOG used, e.g., human (18) versus mouse (60–62). Whether a greater mobilization of Bregs can be achieved using MOG-pσ1 in rMOG-induced EAE, remains to be determined.

Previous studies have shown that MOG_35-55-induced EAE does stimulate anti-MOG_35-55 Ab responses (23, 32, 63), but EAE can still be elicited in the absence of B cells (24, 50). Abs induced to different MOG peptides as a result of epitope spreading (64) or immunization with MOG protein do contribute to EAE pathogenicity (60–63, 65). Such evidence is consistent with the notion that Abs to the MOG protein are pathogenic to MS patients (66, 67). In contrast, a suppressive role for B cells has been suspected for many years (24, 68). The existence of a regulatory population of B cells has not yet been widely accepted, mainly because of the lack of a unique set of cell surface markers or a defined transcription factor. In mice, at least three different phenotypes have been associated with Bregs: O’Garra et al. (68) first identified the CD5⁺ subpopulation among B cells responsible for IL-10 production; more recently, Bouaziz and colleagues have described CD1d^HiCD5⁺ B cells (generally referred as B10 cells) (69), and Evans et al. attributed IL-10 production to CD21^HiCD23⁺ cells (2B cells) (4). Recently, a role for GITR ligand–expressing (59), IL-35–producing (70), or programmed death ligand 1–expressing B cells (71) controlling autoimmunity independent of IL-10 has also been described, highlighting the complexity of the Breg–Treg interactions, where multiple mechanisms may be involved. Independent of the idea that Bregs constitute a true cell lineage or merely a state of activation by B cells, the production of IL-10 by a subset of B cells has proven critical for the resolution of EAE and other autoimmune disorders both in mice and humans (6, 10, 11, 21, 69). Bregs also have a role in the mechanism of action by Copaxone, a glatiramer acetate copolymer, approved for the treatment of relapsing-remitting MS (5, 12, 25, 52, 56). In a recent clinical trial, Ireland et al. (53) showed that treatment of relapsing-remitting MS patients with glatiramer acetate restored IL-10 production and reduced lymphotoxin-α production by peripheral B cells, thus contributing to the therapeutic effects of glatiramer acetate. This finding further supports our contention that IL-10–producing Bregs contribute to suppression of autoimmunity.

Upon encountering Ag, lymphocyte signaling via T or BCR is modulated by a cognate signal from a costimulatory or coinhibitory molecule (72). One such B cell activation signal, CD40, has recently been found to be important for the generation of Bregs (73). Moreover, the role of HVEM and its ligands BTLA, LIGHT, and CD160 has generated interest (30, 35, 44, 74–77). HVEM is often referred to as a “molecular switch” of the immune system because it facilitates either a costimulatory or coinhibitory signal. In addition, HVEM has bidirectional signaling capacity, transducing a coinhibitory signal by the ligand (BTLA)-positive cells and simultaneously transducing a costimulatory signal for the HVEM (receptor)-positive cells [reviewed by Murphy et al. (35)].

BTLA is highly expressed by anergic cells (78), but for Tregs and T_eff cells, they remain unchanged despite EAE disease or MOG-pσ1 treatment (Table I). In EAE mice, we demonstrate that both Breg and Treg function is absent. Regarding the absence of Breg function in EAE mice, purified Bregs isolated from PBS-treated EAE mice adoptively transferred into EAE mice failed to reduce clinical disease (Fig. 3A). In contrast, purified Bregs from MOG-pσ1–treated EAE mice adoptively transferred into EAE mice reduced disease (Fig. 3B). Such reduction in clinical disease was the result of Breg-induced augmentation Tregs noted by the increased number of Tregs and their activation status to protect against EAE. This combination reduced the presence of IL-17–producing encephalitogenic T_eff cells (Fig. 4). Although BTLA levels on Bregs were not augmented as a consequence of MOG-pσ1treatment, the percentage of and total BTLA⁺ Bregs producing IL-10 were enhanced, further showing that MOG-pσ1 activates these Bregs in an Ag-specific fashion. OVA-pσ1’s inability to replicate MOG-pσ1’s action was evidenced by the failure to elevate the percentage of or total number of activated IL-10–producing Bregs. Increases in IL-10 production by MOG-pσ1–treated mice were not significantly increased for CD5⁻ B cells. Aspects of Breg intervention were also recapitulated using a T suppressor assay in which MOG-pσ1–induced Bregs, but not CD5⁻ B cells, reversed the loss of T cell suppressor activity. Furthermore, neutralization of BTLA interfered with the ability of the MOG-pσ1–induced Bregs to restore EAE Treg suppressor activity (Fig. 6). In addition, IL-10 production by BTLA^−/− mice with EAE was substantially reduced by both Bregs and Tregs. These mice are refractive to MOG-pσ1 treatment of EAE, and clinical disease remained enhanced relative to BTLA-sufficient mice. Only upon adoptive transfer of BTLA-sufficient Bregs from MOG-pσ1–treated C57BL/6 mice did we observe reduction in EAE. Even though these donor cells were IL-10⁺, IL-10 production by BTLA^−/− mice remained undetectable; however, TGF-β was induced by this treatment suggesting alternative mechanisms may have been activated. The universality of BTLA⁺ Bregs’ role to suppress EAE remains to be determined. It is important to bear in mind that EAE is a very heterogeneous disease in terms of its induction, clinical and pathological presentations, and mouse susceptibilities, which have raised some concerns about its relevance as an MS model (15, 79).

The results presented in this paper begin to address the paradox of having large numbers of Tregs at sites of chronic inflammation but still having no apparent effect in subduing this inflammation (42–45). We know that EAE Tregs become activated subsequent MOG-pσ1 treatment because tetramer-specific Foxp3⁺ Tregs were previously found to express CD69 (30). This stimulation reverses the Treg quiescent status, initiating a cascade of events that leads to the restoration of suppressor cell activity via the production of regulatory cytokines that abate clinical disease. Furthermore, the results from this present study show that Tregs purified from EAE mice remain inactive as evidenced by their inability to suppress polyclonally activated CD4⁺ T cell proliferation (Fig. 1A). In contrast, Tregs from EAE mice treated with MOG-pσ1 showed restored suppressor activity prompting us to study their reactivation. Tregs purified from mice naturally recovering from the disease (5 or 6 wk after EAE induction) also regained their capacity to inhibit CD4 T cell proliferation to a level similar to those Tregs derived from naive mice. Future work will be needed to determine whether the naturally recovered Tregs share similar suppressor mechanisms as those with the MOG-pσ1–induced Tregs.

Treg engagement of BTLA⁺ Bregs represents one mechanism for stimulating EAE Tregs. We are currently investigating whether this interaction is HVEM dependent or whether other costimulatory molecules are involved. It has been shown that BTLA interacts in cis with HVEM on cells in which both BTLA and HVEM are expressed (35). The fact that BTLA acts cis with HVEM on Tregs offers one potential pathway by which BTLA⁺ Tregs are rendered quiescent resulting in the inability to control EAE. Tregs possibly become quiescent during EAE as a consequence of persistent coinhibitory molecule signaling via BTLA on Tregs with concomitant downregulation of activation or costimulatory molecules. The introduction of BTLA⁺ cells, much like the Bregs described in this paper, could compete with Tregs’ own BTLA and be rendered active resulting in suppressor function restoration. Defining which specific ligands are induced on Tregs during EAE and how these are regulated by MOG-pσ1 is currently being investigated.

The role of B cells in MS and EAE has generated significant interest, especially since the discovery of IL-10-producing B cells (4, 21, 68, 69). In this study, we show that CD5⁺ Bregs dramatically restore Tregs’ suppressive properties whereas B220⁺CD5⁻, BTLA⁻ B cells, or Bregs from diseased mice do not. In fact, adoptive transfer of disease-induced Bregs into EAE mice failed to stimulate Treg production of regulatory cytokines. In contrast, the EAE recipients given MOG-pσ1–induced Bregs stimulated Tregs to generate regulatory cytokines. Furthermore, BTLA^−/− mice developed a severe form of EAE (49), but adoptive transfer of wild-type (BTLA⁺) Bregs restored protection against EAE (Fig. 8). These BTLA-expressing Bregs interact with Tregs, providing an alternative (or synergistic) pathway for Treg rescue. Of note, the derived Bregs, especially those induced with MOG-pσ1, cannot substitute for Tregs because they are unable to directly suppress T_eff cells. Rather, they are crucial modulators of Tregs’ activity. In contrast, a recent publication (80) reported that innate B regulatory progenitors (c-kit^lowSca-1^lowCD127⁺B220⁺CD19⁺IgM⁻CD1d^intCD43⁺) directly suppress CD4⁺ T cell proliferation. These data suggest a number of regulatory cell subsets, each dampening autoimmune disease.

In summary, these studies show a critical role for BTLA-dependent signaling in the activation of both Bregs and Tregs. The lack or failure of Treg activation has critical consequences for EAE induction, which in turn is important for amelioration of disease by MOG-pσ1. Although previous studies (81, 82) have shown that Bregs are able to selectively induce Foxp3⁺ Treg proliferation, our work provides critical insights to explain how Treg function can be restored following their quiescent status during autoimmune disease. Moreover, our hypothesis proposes a method to restore Treg function, namely with the addition of therapeutically induced Bregs. Because both regulatory subsets are necessary to reduce EAE, utilizing the HVEM-BTLA pathway has promise in the development of effective therapies for inflammatory diseases such as MS.

We thank Jill Bobel for technical assistance.

The authors have no financial conflict of interest.