Rituximab Efficiently Depletes Increased CD20-Expressing T Cells in Multiple Sclerosis Patients

Ever since the first phase II clinical trials demonstrated rapid and sustained reduction of inflammatory disease activity following a single course of rituximab (RTX) treatment (1, 2), B cell depletion has emerged as a highly promising therapeutic approach in multiple sclerosis (MS). RTX is a chimeric monoclonal anti-CD20 Ab of the IgG1 isotype that triggers rapid complement and NK cell–mediated depletion of CD20-expressing B cells (3). B cell depletion using RTX does not affect the CD19⁺CD20⁻ pro–B cell and CD20⁻CD138⁺ plasma cell populations, and within 6–8 mo following RTX treatment, the CD20⁺ B cell compartment begins to replenish (4), mainly composed of naive B cells (4). B cells of the CD27⁺ memory phenotype remain at significantly lower levels in peripheral blood, often times beyond 12 mo, possibly accounting for a long-lasting beneficial effect of anti-CD20 therapies on MS disease activity that is sustained following repletion of circulating B cells (5).

Low percentages of CD20-expressing T cells in human blood were first described in 1993 (6), but the existence of this rather rare T cell subset has been disputed (7). Others have found that CD20-expressing T cells can exhibit proinflammatory capacity (8, 9). In rheumatoid arthritis (RA), CD20⁺ T cells make up a larger percentage of Th17 cells when compared with healthy individuals (9). However, the overall percentage of CD20⁺ T cells among all T cells does not differ between RA patients and healthy individuals, and the pathological relevance, if any, of CD20⁺ T cells in autoimmune diseases remains entirely unknown. Almost expectedly, during clinical trials in RA, it was noted that CD3⁺ T cells expressing low levels of CD20 are depleted by RTX (4).

In this study, we were interested in unequivocally demonstrating the existence of CD20⁺CD3⁺ cells and determining if these cells indeed belong to a T cell lineage. Furthermore, we sought to evaluate whether CD3⁺CD20⁺ cells were differentially present in the peripheral blood of MS patients compared with healthy donors and to determine their level of depletion in response to RTX treatment in MS patients. To address these questions, we performed extensive flow cytometric phenotypic characterization of B and T lymphocytes and gene expression profiling of CD20⁻ T cells, B cells, and CD20⁺ T cells, from peripheral blood of healthy control subjects (HC), untreated (UNT) MS patients, and MS patients at different time points following RTX treatment.

Peripheral blood was obtained from patients with a confirmed diagnosis of MS who were UNT or had received standard-dose RTX therapy (two infusions of 1 g i.v. each, 2 wk apart) at different time points prior to sample acquisition or from healthy donors (see Table I for sample details). PBMCs were prepared using a Ficoll paque density gradient following standard protocols. These studies were approved by the University of California, San Francisco Committee on Human Research.

Phenotypic analysis of B and T cells was performed using multicolor FACS; see Table I for experiments performed per sample. PBMC were resuspended in PBS/1% BSA, and FcR blocking was performed using mouse serum (The Jackson Laboratory). For B cell analyses, cells were stained with pretitrated volumes of fluorescent-labeled Abs: CD19 (allophycocyanin-Cy7), IgD (PE Cy7), CD27 (Qdot 605), CD24 (PE Alexa 610), CD38 (PerCP Cy5.5), IgM (PE Cy5), IgG (allophycocyanin), CD20 (FITC), CD138 (PE), and CD3 (Pacific Blue). DAPI was added to discriminate dying/dead cells; samples were analyzed on a four-laser FACSAria III (BD Biosciences). CD19⁺ B cells were gated from singlet lymphocytes after exclusion of CD3⁺ T cell and dead cells (DAPI⁺). T cell subsets were separately stained using the following Abs: CD3 (allophycocyanin), CD4 (PerCP Cy5.5), CD8 (allophycocyanin–Alexa Fluor 750), CD20 (FITC), CD27 (Qdot 605), CCR7 (PE), and CD45RA (PE Cy5). CD3⁺ T cells were identified on singlet lymphocytes and further divided into subsets. The CD20 cutoff for negative versus positive was determined by using fluorescence-minus-one control.

PBMC were stained with anti-CD20 FITC (Beckman Coulter), anti-CD19 PE (Beckman Coulter), and anti-CD3 allophycocyanin (Beckman Coulter); dead cells were labeled with DAPI (Invitrogen). Cytometry gating was performed first on lymphocytes, then live cells (DAPI-negative), then on either CD19⁺ or CD3⁺, and lastly on CD20⁺ among CD19⁺ or CD3⁺ cells.

PBMC were stained with Abs against CD3, CD19, and CD20; three populations of lymphocytes were sorted on a MoFlo Astrios (Beckman Coulter) directly into RLT lysis buffer (Qiagen): 1) CD19⁺CD20⁺; 2) CD3⁺CD20⁻; and 3) CD3⁺CD19⁻CD20⁺. Immediately after sorting, cells were frozen in RLT buffer at −80°C. RNA was extracted using the Qiagen micro RNA kit with DNAse treatment (Qiagen). RNA concentrations were measured using a UV spectrophotometer (Nanodrop), and RNA quality was assessed using Pico-RNA chips (Agilent 2100 Bioanalyzer; Agilent Technologies). Microarray hybridizations were performed at the University of California, San Francisco–affiliated Gladstone Institutes’ Genomics Core Facility. The NuGEN Pico V2 kit (NuGEN) was used for cDNA amplification, fragmentation, and biotinylation; biotinylated cDNA was hybridized to Human GeneChipR Gene 2.0 ST microarrays (Affymetrix). The signal intensity fluorescent images were read using the Affymetrix Model 3000 Scanner (Affymetrix) and converted into GeneChip probe results files (CEL). Microarray data were analyzed using the statistical programming language R/Bioconductor. Quality of the analyzed arrays was ascertained using the arrayQualityMetrics package (10). Raw data were processed using the rma function in the "oligo” package (11). Log2-transformed data were filtered for variance using the “genefilter” package (keeping only probes showing a difference between the 10 and 90% quantiles of >1). Hierarchical clustering was performed using the heatmap.2 function of the “gplots” package.

Statistical analyses were performed using GraphPad Prism 6.0 (GraphPad). Statistical methods are indicated in the text and figure legends where applicable.

We studied T cell and B cell subsets in HC and treatment-naive MS patients (Table I). In all subjects (HC and MS), we found CD3⁺ T cells coexpressing low levels of CD20 (Fig. 1A), generally representing <10% of total CD3⁺ cells (Fig. 1B). Given their potentially proinflammatory function (9), we sought to determine if the frequency of CD3⁺CD20^dim T cells in peripheral blood was increased in MS; indeed, the frequency of CD3⁺CD20^dim among all CD3⁺ T cells was modestly higher in MS (7.2 ± 3.6%, mean ± SD; n = 11) compared with 5.4 ± 2.4% (mean ± SD) in HC (n = 18) (p = 0.02, one-tailed t test) (Fig. 1B). To clearly define the lymphocyte lineage of CD3⁺CD20^dim cells, we performed fluorescent flow cytometry imaging (Amnis ImageStream) on HC (n = 3) and treatment-naive MS patients (n = 3), unequivocally revealing CD3 and CD20 expression at the single-cell level (Fig. 1C); >100 CD3⁺CD20^dim T cells were imaged for this experiment (not shown). Lack of expression of CD19 on these cells confirms that they do not belong to a canonical B cell phenotype (Fig. 1C). Transcription profiling and hierarchical clustering of sorted CD3⁺CD20⁻, CD19⁺CD20⁺, and CD3⁺CD20^dim cells from HC (n = 2) further confirmed the T cell lineage of CD3⁺CD20^dim population (Fig. 1D); they express both α/β- and γ/δ-TCRs as indicated by increased expression of TCRα constant chain and TCRγ constant chain (Fig. 1E) compared with B cells. CD20⁺ T cells expressed lower levels of CD20 compared with CD19⁺ B cells (Fig. 1E), hence their designation as CD3⁺CD20^dim T cells.

Extensive flow cytometry immunophenotyping of T cells was performed using PBMC from MS patients (n = 10) and HC (n = 12). The CD3⁺CD20^dim population comprised both CD4⁺ helper and CD8⁺ cytotoxic subsets (Fig. 2A). Interestingly, among CD3⁺CD20^dim T cells, CD8⁺ were modestly increased in MS patients (60.3 ± 14.7%) compared with HC (48 ± 15%) (p = 0.03, one-tailed t test) (Fig. 2B); no significant difference was found between MS and HC in the percentages of CD4⁺ among CD3⁺CD20^dim T cells (42.7 ± 15.7% in HC versus 36 ± 15.6% in MS; p = 0.17, one-tailed t test) (Fig. 2B) or of CD4⁺ or CD8⁺ total CD3⁺ T cells (Supplemental Fig. 1).

CD3⁺CD20^dim T cells also express CD27, CCR7, and CD45RA (Fig. 3A) in different combinations, highlighting the developmental diversity contained within this T cell subset. We evaluated a number of defined subpopulations among CD3⁺CD20^dim T cells; in UNT MS patients the following distribution was found: CD45RA⁺CCR7⁺ naive T cells (28.3 ± 12.3%); CD45RA⁻CCR7⁺ central memory T cells (28.8 ± 13.4%); CD45RA⁻CCR7⁻ effector memory T cells (16.8 ± 8.6%), and CD45RA⁺CCR7⁻ RA⁺ memory T cells (27.5 ± 12.1%) (not shown). There was no significant difference between MS and HC for any of these CD3⁺CD20^dim T cell subsets (not shown). Lastly, surface markers commonly found on B cells (CD19, HLADR, CD24, CD38, CD138, IgD, IgM, and IgG) were not observed on CD3⁺CD20^dim T cells (Fig. 3B).

To study the effect of RTX on CD3⁺CD20^dim T cells and other lymphocyte subsets, we collected PBMC from cross-sectional cohorts of UNT MS patients and compared lymphocyte subsets to their distribution in RTX-treated patients at different time points: 1–12 wk post-RTX treatment (n = 7); 13–24 wk post-RTX (n = 8); 25–36 wk post-RTX (n = 5); and 37–52 wk post-RTX (n = 5). We found near-complete depletion of CD3⁺CD20^dim T cells during weeks 1–12, with mean frequencies of 0.36% (± 0.36) compared with 7.8% (± 3.7) in UNT patients; during weeks 25–36 and weeks 57–52, partial repletion of CD3⁺CD20^dim T cells was observed, with values of 2.7% (± 2.3) and 2.8% (± 1.5), respectively (Fig. 4A). Apart from a slightly increased proportion of overall CD3⁺ T cells during weeks 13–24, no significant effect of RTX on the overall CD3⁺ T cell compartment was present (Fig. 4B), suggesting that following depletion of CD3⁺CD20^dim T cells by RTX, there was rapid homeostatic repopulation with CD20⁻ T cells. A typical representation of the depletion and gradual repopulation of CD3⁺CD20^dim T cells is depicted in Supplemental Fig. 2. As also observed for B cells, very small numbers of CD3⁺CD20^dim T cells remained detectable in PBMC even during RTX treatment (Fig. 4A). Interestingly, the nondepleted CD3⁺CD20^dim T cells that remained in the circulation 1–12 wk following RTX infusion were predominantly comprised of CD4⁺ T cells (Fig. 4C). In contrast, the overall distribution of CD3⁺CD4⁺ and CD3⁺CD8⁺ T cells remained unchanged at all time points following RTX therapy (Fig. 4D).

We were also interested in determining the response of CD19⁺ B cells (Fig. 5A), CD19⁺CD20⁺ B cells (Fig. 5B), and other B cell subpopulations (Fig. 6) to RTX treatment in our patient cohort. Representative flow cytometry plots and gating strategies for B cell subpopulation analysis are shown in Supplemental Fig. 3. In the UNT MS group, the overall CD19⁺ B cell fraction ranged between 4.2 and 11.6% of lymphocytes (8.9 ± 2.8%) (Fig. 5A). As expected (4), RTX induced near-complete depletion of CD19⁺ and CD19⁺CD20⁺ B cells with replenishment during weeks 25–36 (Fig. 5A, 5B); the majority of B cells in replenishing repertoires were composed of Ag-inexperienced IgD⁺CD27⁻ naive and transitional B cell subsets, suggesting influx from CD20⁻ pro–B cells (Fig. 6A). Memory B cell populations and IgD⁻CD27⁻ B cells remained depleted for significantly longer and were low even during weeks 37–52 (Fig. 6A, 6B). We found small percentages of residual B cells at every time point after RTX treatment; interestingly, in the 1–12 wk cohort, the residual B cell compartment was almost entirely composed of IgD⁻CD27⁻ (i.e., double-negative [DN]) B cells (Fig. 6B). We also found that the percentage of CD19⁺CD27^hiCD38^hi plasma cells/plasmablasts (PC) was significantly reduced by RTX during weeks 1–12 and 13–24 (Fig. 6B). Among CD19⁺ cells, PC did not show significant changes in either cohort, but trended toward making up higher numbers among residual CD19⁺ cells (Fig. 6B).

In this report, we unequivocally establish the existence of CD3⁺CD20^dim cells belonging to the T cell lineage in healthy donors and MS patients, thus lending strong support to earlier data suggesting the presence of a CD20-expressing T cell population (6). We confirm that CD20 expression is generally dim on these cells when studied by flow cytometry, further supported by lower levels of CD20 transcripts in these cells. Although we did not perform functional assays, we show that CD3⁺CD20^dim T cells can be either CD4⁺ Th cells or CD8⁺ cytotoxic T cells. Together with previous reports that CD3⁺CD20^dim T cells can assume a proinflammatory Th17 phenotype (9), our data suggest that CD3⁺CD20^dim T cells represent a functionally heterogeneous population that may also potentially be involved in the MS disease process.

A vast body of evidence assigns important roles to T cells in the immune pathology of MS. Partially driven by the strong genetic association of MS susceptibility with HLA-DRB1*1501 and other genes known to be important to Th functions (reviewed in Ref. 12) and by the knowledge that in disease models adoptive transfer of myelin-reactive CD4⁺ T cells elicits MS-like pathology (reviewed in Ref. 13), Th cells are considered necessary contributors to autoimmune demyelination in MS (reviewed in Ref. 14). Accumulating evidence also suggests an important role of CD8⁺ T cells in MS immune pathology. Oligoclonal CD8⁺ T cells are overrepresented in MS lesions (15, 16), and myelin-reactive CD8⁺ T cells are present in peripheral blood of MS patients (17), outnumbering myelin-reactive CD4⁺ T cells (18). In animal models, CD8⁺ T cells reactive against myelin basic protein and myelin oligodendrocyte glycoprotein are encephalitogenic (19), and CD8⁺ T cells reactive against the glial fibrillary acidic protein were recently shown to mediate relapsing-remitting experimental disease (20). In this context, it is interesting that we find increased numbers of CD8⁺ cells among CD3⁺CD20^dim T cells in UNT MS patients compared with HC. In addition, RTX appears to have, overall, a greater impact on the CD8⁺ than the CD4⁺ subset of CD3⁺CD20^dim T cells. It is not known if CD3⁺CD8⁺CD20^dim T cells contribute to autoimmunity in MS and whether depletion of the CD3⁺CD20^dim T cell population contributes to the therapeutic effect of CD20-targeting therapies in MS (21, 22). Based on the preliminary, but highly encouraging clinical results of anti-CD20 therapies in MS, plus the hypothesis that B cell depletion is responsible for this effect, a number of additional B cell–targeting therapeutic approaches are being actively pursued. In particular, a mAb targeting the B cell lineage restricted marker CD19 (23) is currently in phase I clinical evaluation for the treatment of MS (MEDI-551) (24). Compared with anti-CD20 approaches, anti-CD19 therapy targets a broader spectrum of B cells at different developmental stages but is not expected to deplete CD3⁺CD20^dim T cells; accordingly, a difference in therapeutic efficacy between anti-CD19 and anti-CD20 approaches may indirectly identify a pathogenic role, should it exist, of CD3⁺CD20^dim T cells in MS.

A number of studies have also demonstrated effects of anti-CD20 therapy on the T cell compartment. In humans, RTX treatment reduced T cell numbers in the CSF (25, 26) and in the circulation diminished proinflammatory Th1 and Th17 responses of CD4⁺ and CD8⁺ T cells (27). Similarly, in a B cell–dependent model of experimental autoimmune encephalomyelitis induced by immunization of mice with the recombinant extracellular domain of myelin oligodendrocyte glycoprotein, anti-CD20 treatment reduced proinflammatory T cell responses (28, 29). These effects have generally been thought to result from indirect effects of B cell depletion on pathogenic T cells, presumably via blocking B cell–mediated Ag presentation or secretion of proinflammatory B cell cytokines. However, the current data raise the possibility that some of the known effects of RTX on the T cell compartment (25–27), and perhaps also that some of the beneficial effects of RTX in RA and MS, may result from depletion of CD3⁺CD20^dim T cells. Irrespective of the mechanism, in this study, we expand the spectrum of immunological changes induced by CD20-targeting therapy in MS to include near-complete direct depletion of CD3⁺CD20^dim T cells for at least 1 y following RTX infusion. In this regard, it is of great interest that, in preliminary clinical studies of both RTX and ocrelizumab in MS, long-term protection against focal inflammatory disease activity was found even after repletion of circulating B cells occurred (5), consistent with the hypothesis that CD3⁺CD20^dim T cells could have some pathogenic role in MS.

The long-term depletion of CD3⁺CD20^dim T cells also suggests that this population may represent an early developmental stage of T cells that is not rapidly replenished, similar to the prolonged depletion of memory B cells observed after RTX treatment (4). In fact, the diversity of T cell–associated markers expressed by CD20^dim T cells, including helper, cytotoxic, naive, and memory populations detected by flow cytometry and α/β- and γ/δ-TCRs by transcriptomics, all suggest that CD20 expression may be an early thymic event in the development of a diverse CD3⁺CD20^dim T cell population. To our knowledge, no murine equivalent to human CD3⁺CD20^dim T cells has been identified. However, a murine CD20 homolog, MS4aB1, that is expressed on T cells but not B cells was recently described (30). In murine experimental autoimmune encephalomyelitis, treatment with an anti-MS4aB1 Ab was found to ameliorate disease severity and reduce proinflammatory T cell responses (31), theoretically mimicking the therapeutic effect of anti-CD20–mediated CD3⁺CD20^dim T cell depletion, in the absence of B cell depletion, in humans.

CD20 is traditionally considered a B cell lineage-specific marker. CD20 is a trans-membrane protein (32) expressed on the majority of peripheral B cell subsets, but not on early bone marrow pro–B cells or on a subset of terminally differentiated plasma cells primarily residing in secondary lymphoid tissues. There is no known ligand for CD20; functions of CD20 have been proposed to include calcium transport (33) and boosting of T cell–independent B cell responses (34). CD20-targeting therapies are commonly referred to as B cell–depleting therapy; accordingly, we also find the previously described significant impact of RTX on the B cell compartment, with long-term depletion of memory B cells, early replenishment with naive B cell phenotypes, and also reduction of plasmablasts, possibly owing to the higher turnover rates of short-lived plasmablasts (35, 36). We also show for the first time, to our knowledge, that normally rather infrequent CD19⁺CD27⁻IgD⁻ (DN) B cells become significantly depleted by RTX but at the same time comprise the majority of the small B cell population present in peripheral blood of RTX-treated patients; this may suggest relative resistance of DN B cells to CD20-targeted lymphocyte-depleting therapy. Presently, very little is known regarding immunological function of DN B cells, and it has been speculated that this B cell subset may also contribute to autoimmunity, at least in systemic lupus erythematosus (37). The relevance of DN B cells in MS immune pathology is a matter of ongoing investigation.

In summary, we show that in MS increased numbers of a previously neglected T cell subset, CD3⁺CD20^dim T cells, are effectively depleted by RTX. Further studies will be required to understand the developmental origin and evolutionary relevance of CD3⁺ CD20^dim T cells and whether they differ functionally from CD20⁻ T cells. Understanding the pathological relevance of this T cell subset in MS and other autoimmune disorders will likely broaden our understanding of the pathology of human autoimmunity and may reveal novel therapeutic avenues. Current studies in our laboratory are aimed at delineating the functional properties of CD3⁺CD20^dim T cells, defining their potential contribution to MS immune pathology and elucidating the dynamics of the CD3⁺CD20^dim T cell subpopulation through longitudinal follow-up studies of individuals treated with CD20-depleting therapy.

We thank the patients who agreed to participate in this research. We also thank Erica Eggers for excellent technical assistance and the Gladstone Institutes Genomics Core Facility for performing microarray experiments.

D.L. is an employee of F. Hoffmann-La Roche Ltd. H.-C.v.B. has received research funding from F. Hoffmann-La Roche Ltd.