Autoreactive CD19⁺CD20⁻ Plasma Cells Contribute to Disease Severity of Experimental Autoimmune Encephalomyelitis Free

Growing evidence suggests that B cells and Abs play an important role in the pathogenesis of multiple sclerosis (MS), an autoimmune disease affecting the CNS (1–7). In fact, >90% of MS patients have the presence of oligoclonal bands in the cerebrospinal fluid, which has been the most prominent immunodiagnostic feature in MS patients, and such bands are produced by Ab-secreting cells (ASCs) (8–11). Autoantibodies can be secreted by short-lived plasmablasts and/or long-lived plasma cells (PCs) and potentially contribute to the pathogenesis of a variety of autoimmune diseases, including MS (10, 12–14). Spleen and lymph nodes are usually the primary sites where activated autoreactive B cells differentiate into proliferating but short-lived autoantibody-secreting plasmablasts (15). Under the influence of essential survival mediators such as IL-6, CXCL12, BAFF, and APRIL, short-lived plasmablasts could migrate to the bone marrow and inflamed tissues where they become long-lived autoreactive PCs and support chronic autoimmunity by persistently secreting pathogenic Abs (16).

These autoantibody-secreting long-lived PCs are extremely refractory to conventional immunosuppressive therapies, including biologics such as B cell depletion regimens (17–20). There is some evidence suggesting that the sparing of long-lived PCs is due to their lack of expression of CD20, which is the target of current B cell depletion therapies such as rituximab (1, 21). Rituximab generates a beneficial net effect in most MS patients even though the circulating total Ab titer, circulating autoantibody titer, and oligoclonal bands in the cerebrospinal fluid are initially unaffected (21). In fact, ∼20% of MS patients treated with rituximab experience relapse of the disease at 48 weeks (22). It is possible that long-lived autoreactive PCs that are not sufficiently eliminated by current therapies could lead to the reactivation of the disease in a subset of MS patients. Thus, long-lived PCs may be a promising new therapeutic target for MS treatment.

Our laboratory recently demonstrated that MEDI551, an anti-human CD19 mAb currently being tested in a phase I clinical trial for relapsing-remitting MS (RRMS) patients (NCT01585766), therapeutically suppresses experimental autoimmune encephalomyelitis (EAE), a widely used mouse model of MS (12). We showed that MEDI551 not only inhibits pathogenic adaptive immune responses, but it also preserves peripheral regulatory mechanisms, including regulatory B cell (Breg) frequencies and Ag-specific regulatory T cell responses. Importantly, MEDI551 depletes autoreactive ASCs in the spleen and also substantially reduces autoreactive ASCs in the bone marrow. As a result, MEDI551-treated animals showed significantly reduced autoantibody levels in both serum and CNS tissues compared with the control Ab-treated mice. Although clinical evidence suggests that myelin-specific Abs might induce tissue damage in MS, the relative contribution of autoantibodies and autoreactive ASCs during the disease course remain unclear (1). Because anti-CD19 mAb targets PCs whereas anti-CD20 mAb largely leaves this population intact, the present study has examined the contribution of autoantibody-producing ASCs to EAE by comparing the efficacy of CD19 and CD20 mAb in an EAE model.

B cell depletion using CD19 mAbs represents a potential new therapy for MS; as such, human CD19 transgenic (hCD19Tg) mice were used in this study to compare the impact of the anti-human CD19 mAb (MEDI551) we used in our previous study (12) with a potent anti-mouse CD20 mAb (MB20-11) (23, 24) on EAE. As previously shown, human CD19 expression in the hCD19Tg mice recapitulates CD19 expression by human pre–, immature, and mature B cells. Additionally, the pattern of human CD19 expression is similar to mouse CD19 in these animals (25–27), for example, on PCs where the mouse CD19 expression is downregulated and human CD19 expression is also downregulated on at least a portion of PCs (23). As a result, anti-human CD19 mAb depletes pre–B cells and mature B cells and also reduces basal serum IgM and IgG levels in hCD19Tg mice (28). MEDI551 was therefore used for these studies so that the results would be more directly translatable into human studies.

Both CD19 and CD20 mAbs significantly suppressed disease severity in a myelin-induced, B cell–dependent EAE model (29). Intriguingly, CD19 mAb surpassed CD20 mAb in disease alleviation, suggesting that autoreactive ASCs targeted by CD19 mAb (but not CD20 mAb) likely impact disease severity. Treatment with either mAb resulted in profound B cell depletion in peripheral blood and secondary lymphoid organs. However, more detailed analysis of cellular components of the adaptive immune compartment revealed that the CD19 mAb more significantly dampened myelin-specific humoral responses. CD20 mAb-treated animals showed various disease severities but the EAE score in individual mice positively correlated with the frequency of the residual autoantibody-secreting PCs in the bone marrow of the treated animals. Analysis of CD19 and CD20 expression on the ASCs revealed that the bone marrow contained a CD19⁺CD20⁻ subpopulation, which was specifically targeted by CD19 mAb but not by CD20 mAb. These data suggest that autoantibody-producing ASCs, especially the long-lived CD19⁺CD20⁻ ones residing in the bone marrow, contribute to disease severity in EAE. We also identified CD19⁺CD20⁻ B cells in the cerebrospinal fluid of MS patients that would not be targeted by CD20 mAb. Collectively, these data suggest that CD19 mAb may represent a new approach for depleting both pathogenic B cells and PCs to concomitantly inhibit the generation of new myelin-specific Abs and reduce preexisting myelin-specific Ab by targeting PCs in MS patients.

hCD19Tg male mice (28), 6–8 wk of age, were used for in vivo studies.

Recombinant human myelin oligodendrocyte glycoprotein (rhMOG)_1–125 was generated as previously described (4). EAE was induced by s.c. immunization at four sites on the back with 100 μg rhMOG emulsified in CFA containing 5 mg/ml mycobacteria (BD Biosciences, San Diego, CA). On days 0 and 2, mice were injected i.p. with 300 ng pertussis toxin (List Biological Laboratories, Campbell, CA). Clinical disease was assessed daily starting at 7 d postimmunization. Scoring was as follows: 0, no disease; 1, loss of tail tone; 2, weakness of hindlimbs; 3, partial hindlimb paralysis; 4, total hindlimb paralysis with or without forelimb paralysis; 5, moribund or death (12, 30).

Immunized mice were tested for grip strength performance using a wire mesh grid connected to a horizontally aligned force meter (San Diego Instruments, San Diego, CA) (30). The grid was secured at a 45° angle, and the top rung of the grid was used for all testing. Mice were held at the base of the tail and supported ventrally while being moved into position to grasp the wire grid. Once successfully grasped, ventral support was released and the tail was gently pulled in a horizontal plane until the animal’s grip was released from the grid. Peak force (in g of force) was captured by the force meter and recorded for later analysis. Both forelimb and hindlimb strength were gauged three times on each of the days as indicated. The mean (in g of force) was calculated. Data are representative of two experiments.

The anti-human CD19 Ab MEDI551 (human IgG1) and a control Ab (16C4-TM, human IgG1) were produced at MedImmune (Gaithersburg, MD) (31). 16C4-TM is a variant Ab of MEDI551, which lacks the ability to elicit B cell depletion via Ab-dependent cellular cytotoxicity due to mutations at its FcRγ binding site (12). The anti-mouse CD20 Ab MB20-11 (mouse IgG2c) (23) was produced at MedImmune, and its isotype control Ab (clone 6.3) was from SouthernBiotech (Birmingham, AL). B cells were depleted with a single dose of 250 μg MEDI551 or MB20-11 via i.p. injection at day 7 after EAE induction. In some cases, mice were given three daily doses at days 7, 8, and 9 of 250 μg MB20-11 as indicated. In the control groups, the same amount of each control Ab or the same volume of PBS instead of B cell–depletion Ab was given by the same method. Depletion was confirmed by staining circulating murine CD19⁺ B cells in peripheral blood taken 5–10 d after Ab administration.

Cells from different tissues and FACS staining were performed as previously described (12). For FACS staining of cells with our 10-color survey panel, the following anti-mouse mAbs were used: anti-CD45 (clone 30-F11, BioLegend, San Diego, CA), anti-CD3e (145-2C11), anti-TCRβ (H57-597), anti-CD4 (RM4-5), anti-B220 (RA3-6B2), anti-CD19 (1D3), anti-CD11b (ICRF44), anti-Gr1 (RB6-8C5), and anti-NK1.1 (PK136). Other fluorescence-labeled mouse Abs used to define B cell subtypes were anti-IgD (11-26C), anti-IgM (R6-60.2), anti-CD93 (AA4.1), anti-CD23 (B3B4), anti-CD138 (281-2), anti-CD1d (1B1), and anti-CD5 (OX-19). The Live/Dead fixable yellow dead cell stain kit (Life Technologies, Grand Island, NY) was used to differentiate viable cells from dead cells. CD45⁺ leukocytes including neutrophils (CD11b⁺Gr1⁺), activated monocytes/microglial cells (CD11b⁺Gr1⁻), and lymphocytes (CD45^hiCD11b⁻) were major populations of immune cells identified in the brain and spinal cords of animals with EAE. A small population of NK cells (NK1.1⁺CD3e⁻) was also identified in both CNS compartments.

For analyzing CD19 and CD20 expression on CD138⁺ PCs, splenocytes and bone marrow cells from EAE mice at the peak of disease were stained with a human CD19 Ab (BV421 mouse anti-human CD19, clone HIB19, BD Biosciences), an unconjugated mouse CD20 Ab (clone MB20-11, isotype mIgG2a, MedImmune) (24, 28), and other Abs to define PCs (IgD⁻CD3⁻CD138⁺B220⁻) and IgD⁺ B cells (IgD⁺mCD19⁺B220⁺). A secondary Ab (FITC anti-mouse IgG2c, Abcam) was used to detect mCD20 Ab. Fluorescence minus one controls were included for proper gating. Cell events were acquired on a FACSAria or FACSCanto (BD Biosciences) and further analyzed using FlowJo software (Tree Star, Ashland, OR).

Serum was obtained from mice at peak of the disease (days 14–16) or at dates as indicated. Brain and spinal cord supernatants were collected while isolating single cells for FACS analysis as described above. For detecting total IgG or MOG-specific IgG, Immulon 2HB plates (Thermo Scientific, Waltham, MA) were coated with unlabeled goat anti-mouse Ig (SouthernBiotech) or rhMOG at 10 μg/ml and then blocked with 1% BSA (Sigma-Aldrich). Diluted serum or supernatants were added to the wells and incubated at room temperature for 2 h. Plate-bound total or MOG-specific IgG and IgM were detected with HRP-conjugated anti-mouse IgG or IgM, respectively (1:5000; BD Biosciences). The MOG-specific IgG titers were quantified using 8-18C5, an anti-MOG monoclonal mouse IgG. The total Ig titers were quantified using commercially available mouse IgG (Thermo Scientific) or mouse IgM (Santa Cruz Biotechnology, Dallas, TX). Signal was developed using TMB substrate (eBioscience) and the reaction was stopped with 1 M HCl. The plates were read at 450 nm wavelength on an Epic plate reader (BioTek, Winooski, VT).

ELISPOT was performed to assess ASCs. In brief, 96-well Immobilon-P MultiScreen plates (Millipore) were coated with 30 μg/ml rhMOG_1–125 to detect MOG-specific ASCs. Plates were coated with goat anti-mouse IgG or IgM (R&D Systems, Minneapolis, MN) to detect total IgG or IgM ASCs. Cell suspensions from spleens and bone marrows were added to individual wells at different dilutions (4 × 10³ to 4 × 10⁵ cells/well). Cells were incubated for 48 h at 37°C in a 5% CO₂ atmosphere. After incubation, plates were washed several times with 0.05% Tween 20 in PBS and incubated with mouse IgG or IgM detection Abs (R&D Systems) overnight at 4°C. The plates were finally developed using ELISPOT blue color module (R&D Systems). ASCs were enumerated under a microscope. The cumulative EAE score is the sum of the daily EAE scores from day 1 through day 18 (which was the day for collecting tissues) for each individual mouse. Correlation analyses were performed between the cumulative EAE score and the frequency of residual autoreactive ASCs in the spleen and bone marrow. A Pearson correlation coefficient and linear regression analysis were used for the comparisons.

This method was used to generate the data presented in Supplemental Fig. 4. Tissue slices were collected in TRIzol (Invitrogen) and homogenized with a tissue homogenizer (Omni International, Kennesaw, GA). RNA isolation was performed according to the TRIzol protocol, and total RNA was quantified in the NanoDrop. One microgram total RNA was reverse transcribed using random primers and SuperScript II. Fifty to one hundred ng cDNA was used per PCR. PCR was performed in an ABI 7500 RT-PCR machine (Applied Biosystems), using custom probes for Baff (forward primer, 5′-GAAGTGTGCCATGTGAGTTATGAG-3′, reverse primer, 5′-TCACCCAAGGCAAAAAGC-3′), April (forward primer, 5′-CGAGTCTGGGACACTGGAATT-3′, reverse primer, 5′-AGATACCACCTGACCCATTGTGA-3′), Il6 (forward primer, 5′-TCGGAGGCTTAATTACACATGTTC-3′, reverse primer, 5′-AAGTGCATCATCGTTGTTCATACA-3′), Cxcl12 (forward primer, 5′-CAAGCATCTGAAAATCCTCAACAC-3′, reverse primer, 5′-CACTTTAATTTCGGGTCAATGCA-3′), and β-actin (forward primer, 5′-CATACGCCTGCAGAGTTAAGCA-3′, reverse primer, 5′-TGGTACGACCAGAGGCATACA-3′), according to the Applied Biosystems protocol (16). To calculate changes in mRNA expression, we first normalized mRNA expression to β-actin mRNA expression, and mRNA expression for each gene in the experimental groups was compared with the level of mRNA expression in PBS-treated EAE mice. Fold changes in mRNA expression were calculated according to the ΔΔCT calculation (change in cycling threshold) recommended by Applied Biosystems.

Patients were recruited to the University of Southwestern Medical Center and gave written consent to participate in accordance with Institutional Review Board protocols. The diagnosis of the patients was based on McDonald criteria (32). Eight patients with RRMS (six females, two males; average age, 41.8 ± 8.4 y), six patients with primary progressive MS (PPMS) (two females, four males; average age, 55.3 ± 6.6 y), and six patients with secondary progressive MS (SPMS) (six females, no males; average age, 55.5 ± 9.4 y) were included in this study. None of the MS patients had an exacerbation at the time of sampling and did not use corticosteroids 60 d prior to sampling. At the time of sampling, 14 of the 20 MS patients were treatment naive to disease modifying immunomodulatory therapies (DMTs), including IFN-β, mAbs, glatiramer acetate (GA), or methotrexate. At the time of sampling, three RRMS patients were on GA, one PPMS patient was on GA, one SPMS patient was on GA, and one SPMS patient was on IFN-β-1a. The average numbers of CD19⁺ B cells/ml cerebrospinal fluid in each cohort were 85 (RRMS), 54 (SPMS), and 25 (PPMS).

Briefly, the cerebrospinal fluid was centrifuged for 10 min at 394 × g at 4°C. The supernatant was removed and the cerebrospinal fluid cell pellet was gently suspended in 200 μl ice-cold FACS buffer (1× PBS with 4% BSA) and enumerated on a hemocytometer. The cerebrospinal fluid pellet was stained with an 11-color panel containing Abs against CD45, CD3, CD4, CD8, CD19, CD20, CD27, CD138, CD16, CD56, and TCRγδ (BD Biosciences, San Jose, CA). Cells were incubated on ice for 20 min in the dark, washed once in FACS buffer (453 × g, 5 min at 4°C), and suspended in 100 μl FACS buffer. Cells were fixed with 2% paraformaldehyde for 20 min on ice and analyzed on a FACSAria within 3 d (BD Biosciences).

Statistical analyses were performed with GraphPad Prism (GraphPad Software, La Jolla, CA). Data were presented as mean ± SEM. The p values were calculated by two-tailed unpaired Student t test. A p value <0.05 was considered statistically significant.

Patients were recruited to the University of Southwestern Medical Center and gave written informed consent to participate in accordance with Institutional Review Board protocols. Animal protocols were approved by the Institutional Animal Care and Research Advisory Committee (University of Texas Southwestern Medical Center).

A B cell–dependent EAE model was used to study the impact of CD19 and CD20 mAb treatment on disease progression (29). hCD19Tg mice were treated with a single 250-μg dose of either an anti-human CD19 (MEDI551) or an anti-mouse CD20 mAb (MB20-11) at day 7 after rhMOG immunization (Fig. 1). Disease severity was evaluated by EAE score and body weight change (Fig. 1A, 1B). All groups of mice showed signs of disease during the course of these experiments, but mice treated with either CD19 mAb or CD20 mAb showed significantly lower disease severity compared with the PBS group (Fig. 1A). Interestingly, the CD19 mAb was more effective than the CD20 mAb in suppressing EAE progression (Fig. 1A, CD19 mAb versus CD20 mAb, p < 0.05, days 15–17, 20–27). In a separate experiment, we found that neither control Abs (16C4-TM, a control for CD19 mAb, and 6.3, a control for CD20 mAb) influenced EAE progression when the mice were treated with a single 250-μg dose at day 7 after rhMOG immunization (Supplemental Fig. 1A). Thus, both CD19 and CD20 mAbs demonstrated efficacy in reducing EAE severity, although the former showed a significantly improved effect.

As expected, body weight change was related to EAE score (Fig. 1B) (33, 34). Both groups treated with B cell depletion Abs gained weight over time, and CD19 mAb–treated mice showed a better recovery from body weight loss than did those treated with CD20 mAb. We also performed a grip test that measures the strength of the forelimbs and hindlimbs separately (Fig. 1C, 1D) (30). In agreement with the disease score, CD19 mAb–treated mice showed accelerated improvement compared with the CD20 mAb–treated mice (p < 0.05 for forelimbs from day 16 to day 24; p < 0.05 for hindlimbs from day 15 to day 28). Specifically, CD19 mAb–treated mice showed significant recovery of strength for both forelimbs (p < 0.05, days 16–26) and hindlimbs (p < 0.05, days 18–28) compared with the PBS group. In contrast, CD20 mAb–treated mice only showed strength improvement at three single time points along the disease course for forelimbs (p < 0.05, days 18, 19, and 23) and one single time point for hindlimbs (p < 0.05, day 19) compared with the PBS group.

Because immune cell infiltration into the CNS is considered a hallmark feature of EAE (35), we investigated the effect of both mAbs on leukocyte dynamics in the CNS (Supplemental Fig. 1B, 1C). As we reported previously (12), CD19 mAb treatment at day 7 led to a significant reduction in the numbers of major CD45⁺ leukocyte populations (neutrophils, monocytes/microglial cells, and lymphocytes) in the spinal cord (Supplemental Fig. 1C, upper panel). Specific to CD45^hiCD11b⁻ lymphocytes in the spinal cord, a significant decrease in CD4, CD8, and γδ T cells and CD19⁺ B cells was observed in CD19 mAb–treated mice (Supplemental Fig. 1C, lower panel). Although CD20 mAb treatment also led to a significant reduction of monocytes/microglia and lymphocytes, including CD4, γδ T cells, and CD19⁺ B cells, the effect of CD20 mAb on other examined cell subsets was not significant (Supplemental Fig. 1C). Neither CD19 mAb nor CD20 mAb had an effect on NK cell frequency. Additionally, CD19 mAb treatment inhibited infiltration of most cell subsets examined to a greater extent than CD20 mAb. However, with the exception of B cells, which were depleted by both Abs, infiltration of most other leukocyte subsets in the brain was not significantly affected by either mAb treatment (Supplemental Fig. 1B). As expected, the control Abs did not affect leukocyte infiltration into the CNS of EAE mice (Supplemental Fig. 1D, 1E).

Previously we showed that Bregs show resistance to B cell depletion mediated by MEDI551, the CD19 mAb used in this study (12). In this study, we tested whether this regulatory population was also modulated by CD20 mAb. Both CD19 mAb and CD20 mAb treatments rapidly remove the vast majority of circulating and tissue B cells, including the major mature B cell subsets (Supplemental Fig. 2). As we reported previously, CD19 mAb spared Bregs because the frequency of Bregs in the residual B cell pool after CD19 mAb treatment was significantly increased compared with the PBS-treated mice (CD19 mAb versus PBS, 17.5 versus 9.8% in the spleen and 18.5 versus 8.4% in the lymph node, both p < 0.01) (Fig. 2). However, the frequency of Bregs in the residual B cell pool after CD20 mAb treatment was similar to the PBS-treated mice in the spleen (CD20 mAb versus PBS, 7.1 versus 9.8%, not statistically significant) and only slightly increased in the lymph node (CD20 mAb versus PBS, 11.4 versus 8.4%, p < 0.05) (Fig. 2). This discordance between CD19 and CD20 mAb treatment on Breg frequency is more readily apparent in the Breg/non-Breg ratio in the residual B cell pool in both lymphoid tissues, which was significantly increased in the CD19 mAb cohort compared with PBS group, whereas the ratio is similar between the CD20 mAb cohort and the PBS group at least in the spleen (Fig. 2, lower panel). Taken together, these results suggest that CD1d^hiCD5⁺ B cells show some resistance to CD19-mediated B cell depletion but are efficiently targeted by CD20-mediated B cell depletion.

A major benefit of CD19 mAb treatment for B cell–dependent diseases is targeting of ASCs, which are often spared by CD20 mAb treatment. Residual autoreactive PCs after anti-CD20 depletion could continuously secrete pathogenic autoantibodies and either contribute to the chronic condition or lead to reactivation of the disease (15, 36–38). To assess the impact of CD19 and CD20 mAbs on humoral immune responses, we measured the serum and CNS Ab levels after a single injection of the mAbs at day 7 after EAE induction (Fig. 3). CD19 mAb significantly reduced total IgGs in the serum, but the CD20 mAb did not (1.8-fold reduction with p < 0.05 for CD19 mAb and 1.1-fold reduction for CD20 mAb compared with the PBS group). Importantly, CD19 mAb reduced MOG-specific IgG in the serum by 7.9-fold compared with the PBS group (p < 0.01) whereas CD20 mAb only reduced MOG-specific IgG in the serum by 1.6-fold compared with the PBS group (Fig. 3A). Furthermore, the effects of the CD19 mAb on MOG-specific IgG were more pronounced than on total serum IgG.

Next, we examined the effect of B cell depletion on Ab levels in the CNS tissues where MOG-specific IgG may mediate tissue damage. In EAE mice, CD19 mAb treatment led to a 2 fold-reduction and a 3.7-fold reduction of total IgGs in the brain and the spinal cord, respectively, whereas CD20 mAb only reduced total IgGs by 1.3-fold in the brain and 2.6-fold in the spinal cord (Fig. 3B, 3C, top panel). CD19 mAb treatment led to 204- and 404-fold reduction of MOG-specific IgG in the brain and spinal cord, respectively. In contrast, CD20 mAb led to 3.8-fold reduction of MOG-specific IgG in the brain and 10.2-fold reduction in the spinal cord (Fig. 3B, 3C, bottom panel). Additionally, control Abs did not affect pre-existing serum IgG or IgM levels in naive hCD19Tg mice. CD19 mAb treatment decreased both IgG and IgM levels, whereas CD20 mAb only decreased pre-existing IgG but did not decrease IgM levels (Supplemental Fig. 3A, 3B).

To further investigate the impact of CD19 and CD20 mAb depletion on ASCs, we quantified the frequency of ASCs in the spleen and the bone marrow (Fig. 4, Supplemental Fig. 3C, 3D). CD19 mAb is more potent than CD20 mAb in reducing total IgG and total IgM-secreting cells in the spleen and total IgG-secreting cells in the bone marrow (Supplemental Fig. 3C, 3D). Short-lived, MOG-reactive plasmablasts in the spleen are also more effectively depleted by CD19 mAb compared with CD20 mAb (Fig. 4A). Long-lived MOG-reactive ASCs in the bone marrow are effectively depleted by CD19 mAb, but not reduced by CD20 mAb (Fig. 4B). We also assessed the mRNA expression for PC survival factors (BAFF, APRIL, IL-6) and chemokine (CXCL12) that participate in the survival and recruitment of PCs to survival niches, primarily in the bone marrow. No significant change of any of these factors was detected in the spleen or the CNS tissues (Supplemental Fig. 4). Thus, these data demonstrated that CD19 mAb potently depletes both short-lived and long-lived autoreactive ASCs in comparison with CD20 mAb.

We observed that individual mice in the CD20 mAb–treated group displayed different degrees of disease severity, which could be due to the incomplete depletion of autoreactive ASCs by the CD20 mAb treatment. Thus, we performed correlation analyses between the cumulative EAE score and the frequency of residual autoreactive ASCs in the spleen (Fig. 4C) and bone marrow (Fig. 4D). We found that following CD20 mAb treatment, the frequency of residual ASCs in the bone marrow significantly correlated with residual disease severity (p = 0.004) such that the higher the residual disease severity, the higher the frequency of residual ASCs in the bone marrow. Residual ASCs in the spleen did not correlate with residual disease severity (p = 0.091). Additionally, no correlation was found between the frequency of Bregs in the residual B cell pool and the cumulative EAE score in CD20 mAb–treated mice (p = 0.215).

Because the residual ASCs in the bone marrow of the CD20 mAb–treated mice correlated with disease, it is possible that the mechanism by which CD19 mAb surpasses CD20 mAb in alleviating disease could be due to targeting of CD19⁺CD20⁻ ASCs in the bone marrow. To test whether long-lived PCs in the bone marrow lack CD20 expression on the surface, we analyzed human CD19 and mouse CD20 expression on PCs in the bone marrow and PC precursors (plasmablasts) in the spleen (Fig. 5). We found that 67.7% of ASCs in the spleen of PBS-treated EAE mice are CD19⁺CD20⁺ and 21.1% of ASCs only express CD19 whereas a small percentage of ASCs are double negative for CD19 and CD20 (9.77% of total ASCs) (Fig. 5A, PBS). In contrast, ASCs in the bone marrow of PBS-treated EAE mice demonstrate a substantial reduction of CD19⁺CD20⁺ ASCs (19.8% in BM compared with 67.7% in SPN) and the frequency of CD19⁺CD20⁻ and CD19⁻CD20⁻ cells are 18.6 and 56.5%, respectively (Fig. 5C, PBS). Next, we investigated the impact of CD19 and CD20 mAb treatment on each of the ASC populations in the spleen (Fig. 5A, 5B) and bone marrow (Fig. 5C, 5D). In the spleen, CD19 mAb treatment removed all of the CD19⁺CD20⁺ ASCs, whereas 12.8% of CD19⁺CD20⁺ ASCs still remained after CD20 mAb treatment (PBS, 67.7%; CD19 mAb, 0%; CD20 mAb, 12.8%). In contrast to CD19 mAb treatment, which removed almost all the CD19⁺CD20⁻ ASCs, CD20 mAb treatment has limited impact on this population because it only slightly reduced the total number of CD19⁺CD20⁻ ASCs without affecting its frequency in the remaining ASC pool (Fig. 5B). In the bone marrow, CD19 mAb surpassed CD20 mAb in removing CD19⁺CD20⁺ ASCs. More importantly, CD19 mAb almost completely removed CD19⁺CD20⁻ ASCs whereas CD20 mAb left this population untouched (Fig. 5C, 5D). The cell number of CD19⁺CD20⁻ ASCs and its ratio to CD19⁻CD20⁻ ASCs in the bone marrow remained largely unchanged in the CD20 mAb treatment group (compare 1:3 versus 1:4, Fig. 5C). These data combined with the correlation data supported our hypothesis that the presence of CD19⁺CD20⁻ ASCs in the bone marrow contributes to the residual disease activity we observed in the mice treated with CD20 mAb.

Interestingly, our analysis of CD19 versus CD20 expression on human B cells revealed that CD19⁺CD20⁻ B cells are present in the cerebrospinal fluid of patients experiencing RRMS (20.18% of total B cells), SPMS (29.58% of total B cells), and PPMS (31.73% of total B cells) (Fig. 6). Note, however, that the absolute number of B cells in the RRMS patients is larger than the progressive MS patient cohorts (85 CD19⁺ B cells/ml in RRMS, 54 CD19⁺ B cells/ml in SPMS, 25 CD19⁺ B cells/ml in PPMS), although the cohorts themselves are small (eight RRMS, six SPMS, six PPMS), and some of them are on DMTs. Nevertheless, this introductory dataset suggests that CD19 mAb and CD20 mAb treatment could have differential depletion effects on human B cells, which could potentially result in different clinical outcomes for MS patients as we have observed in our EAE model.

The goal of this study was to determine whether targeting PCs and/or other potential immune components could advance the current benefit of B cell depletion therapy. To do this, we used a B cell–dependent EAE model and treatment with either a CD19 mAb (which we had previously demonstrated targets PCs) or a CD20 mAb (which spares most PCs). We found that CD19 mAb treatment ameliorates EAE more effectively than does CD20 mAb (Fig. 1, Supplemental Fig. 1). B cell depletion by CD20 and CD19 mAbs in this side-by-side comparison confirmed that both Abs are highly effective to deplete a broad spectrum of B cells in hCD19Tg mice, although CD19 mAb shows better depletion efficiency in some tissues and on certain B cell subtypes likely due to CD19 being expressed both early and late during B cell development (Supplemental Fig. 2).

CD19 mAb treatment more significantly reduced normal and autoreactive anti-MOG serum Ig levels than did CD20 mAb in naive and EAE mice, respectively (Fig. 3A, Supplemental Fig. 3). More importantly, CD19 depletes CNS autoantibodies to a significantly larger extent than did CD20 mAb in the EAE mice (Fig. 3B, 3C). These differences are explained by the finding that in addition to mature B cell depletion by both the CD19 and CD20 mAbs, the CD19 mAb also depletes a significant fraction of plasmablasts and Ab-secreting PCs (Fig. 4). If the Ab pool participates in driving EAE severity, EAE mice that have undergone CD19 mAb treatment should develop EAE when serum from EAE mice induced with rhMOG is passively transferred into them. Others have demonstrated the potent capacity of anti-MOG Abs to confer EAE in other models (4, 5, 15, 17), albeit not CD19 mAb–treated mice.

We further identified that CD19⁺CD20⁻ ASCs were specifically targeted by CD19 mAb but not CD20 mAb, which potentially harbors autoreactivity (Fig. 5). Efforts are underway to use transcriptome analysis to determine the autoreactive potential and proinflammatory signature of CD19⁺CD20⁻ ASCs in comparison with CD19⁻CD20⁻ and CD19⁺CD20⁺ counterparts. However, this outcome would be predicted because CD19⁻CD20⁻ ASCs are not targeted by CD19 mAb treatment and yet there is a profound reduction in EAE severity (Fig. 5). Furthermore, this diminution of EAE severity cannot be attributed to depletion of any other B cell subset as demonstrated by the diminished impact of CD20 mAb on EAE severity (Fig. 1). Finally, the robust correlation between the frequency of residual autoreactive ASCs and EAE severity in the CD20 mAb–treated mice also supports this notion (Fig. 4C, 4D).

One additional possibility, however, is that Bregs may modulate EAE severity. Indeed, our previous study demonstrated that CD19 mAb spares regulatory mechanisms shown to suppress the disease (12). In this study, we demonstrated that Bregs are more resistant to CD19 mAb treatment compared with the other mature B cell subsets. In contrast, we found that CD20 mAb treatment depletes Bregs to a similar extent as other major B cell subsets (Fig. 2). Previous studies identified Bregs as important regulators in controlling EAE largely through the secretion of immunosuppressive cytokines, especially IL-10. Thus, CD20 mAb depletion of regulatory IL-10–competent Bregs could partially contribute to less efficacy of CD20 mAb to suppress EAE than CD19 mAb. It is more likely that the combination of MOG-specific Ab production and B cell Ag presentation and/or costimulation may contribute to EAE pathogenesis (2). Thus, mature B cells, Ab-secreting plasmablasts, and PCs are all important targets for the efficacious treatment of MS.

Interestingly, CD19⁺CD20⁻ B cells are also detected in the cerebrospinal fluid of MS patients (Fig. 6). Further study is required to determine whether the CD19⁺CD20⁻ B cells we detected in the cerebrospinal fluid of MS patients are an important target for therapeutic intervention, as we have demonstrated in EAE. However, the sparing of autoreactive CD19⁺CD20⁻ B cell subtypes by CD20 mAb treatment may provide an explanation for why some patients who are treated with rituximab continue to experience exacerbations (22). Defining the B cell subtypes that do not express CD20 and their autoreactive potential would be exceptionally helpful in such endeavors.

In summary, this study revealed that EAE induced by whole MOG protein immunization readily elicited the formation of a long-standing pool of PCs secreting MOG-specific IgG Abs that may play a fundamental role in EAE pathogenesis. Competent expression of CD19 by a proportion of these PCs, which lack CD20 expression, confers better efficacy of CD19 mAb over CD20 mAb in suppressing the disease. Our study suggests that CD19 represents a very promising therapeutic option for MS patients by more significantly targeting a broader spectrum of B cells including autoreactive CD19⁺CD20⁻ ASCs.

We thank the patients who provided cerebrospinal fluid samples for this study. We also thank Dr. Erik Plautz and Sherry Rovinsky in the Animal Facility of the Department of Neurology and Neurotherapeutics (University of Texas Southwestern Medical Center) for technical assistance. We also thank Dr. Sean Morrison and his team in the Moody Foundation Flow Cytometry Facility at Children’s Research Institute for use of instruments (University of Texas Southwestern Medical Center).

N.L.M. receives grant funding from the National Multiple Sclerosis Society, Diogenix Inc., MedImmune Inc., and TEVA Neuroscience. The other authors have no financial conflicts of interest.