Expansion of Autoreactive Unresponsive CD21−/low B Cells in Sjögren's Syndrome–Associated Lymphoproliferation

Authors


Abstract

Objective

Primary Sjögren's syndrome (SS) is an autoimmune disease associated with a high risk of developing non-Hodgkin's lymphoma. This study was undertaken to determine the nature of B cells driving lymphoproliferation in primary SS.

Methods

B cell subsets and function were analyzed in peripheral blood from 66 adult patients with primary SS (including 14 patients with B cell lymphoproliferative disease [LPD]) and 30 healthy donors, using flow cytometry, calcium mobilization, and gene array analysis. The reactivity of recombinant antibodies isolated from single B cells from patients with primary SS and LPD was tested using an enzyme-linked immunosorbent assay.

Results

We observed an expansion of an unusual CD21−/low B cell population that correlated with lymphoproliferation in patients with primary SS. A majority of CD21−/low B cells from patients with primary SS expressed autoreactive antibodies, which recognized nuclear and cytoplasmic structures. These B cells belonged to the memory compartment, since their Ig genes were mutated. They were unable to induce calcium flux, become activated, or proliferate in response to B cell receptor and/or CD40 triggering, suggesting that these autoreactive B cells may be anergic. However, CD21−/low B cells from patients with primary SS remained responsive to Toll-like receptor (TLR) stimulation. Molecules specifically expressed in CD21−/low B cells that are likely to induce their unresponsive stage were detected in gene array analyses.

Conclusion

Patients with primary SS who display high frequencies of autoreactive and unresponsive CD21−/low B cells are susceptible to developing lymphoproliferation. These cells remain in peripheral blood controlled by functional anergy instead of being eliminated, and chronic antigenic stimulation through TLR stimulation may create a favorable environment for breaking tolerance and activating these cells.

Primary Sjögren's syndrome (SS) is a systemic autoimmune disease primarily characterized by chronic inflammation of the exocrine glands, in particular the salivary and lachrymal glands. Extraglandular manifestations occur in many patients and may involve almost any organ. B lymphocyte hyperactivity in primary SS is manifested by the presence of anti-SSA and anti-SSB antibodies, rheumatoid factor (RF), cryoglobulins, and hypergammaglobulinemia. Patients with primary SS have a high risk of developing a lymphoma. Unfortunately, thus far, we have insufficient data to deal with such pertinent concerns. Prolonged B cell survival and excessive B cell activity, probably related to increased production of BAFF (1–3), may lead to lymphomas occurring in 5% of patients with SS (4, 5). Significant predictors of lymphoproliferative disease (LPD) in primary SS include parotid, lymph node, and/or splenic enlargement, monoclonal gammopathy, hypogammaglobulinemia, mixed cryoglobulinemia, palpable purpura, CD4+ T cell lymphopenia, and/or reduced levels of C4 (6–9).

It is proposed that the first event of lymphomagenesis in SS is the chronic stimulation of polyclonal B cells secreting autoreactive antibodies, such as RF. Such autoreactive B cells may become monoclonal, leading to the occurrence of lymphoproliferation. The subsequent step would be a chromosomal abnormality, which would confer to these cells low-grade B cell lymphoma comportment (10). The nonrandom utilization of VH and VL regions by SS-associated lymphoma B cells (11, 12) and the demonstration that these lymphoma B cells may display RF activity (13) support the hypothesis that these lymphomas grow through an autoantigen-driven process.

We report here that an unusual CD21−/low B cell population correlates with the lymphoproliferative status in patients with primary SS. Because CD21 augments B cell receptor (BCR)–mediated signaling as part of the B cell coreceptor complex, its down-regulation may confer a state of anergy to these cells, as has been demonstrated among CD21−/low B cells in patients with rheumatoid arthritis (RA), common variable immunodeficiency, or hepatitis C virus–associated cryoglobulinemia (HCV-MC) (14–16). These CD21−/low B cells are enriched in autoreactive clones that are unresponsive to BCR stimulation, suggesting that these cells are controlled by the tolerizing mechanism of functional anergy. Gene array analyses of CD21−/low B cells revealed molecules that are specifically expressed in these B cells and that are likely to induce their unresponsive stage. These B cells belonged to the memory compartment due to the fact that their Ig genes were mutated, as reported in patients with HCV-MC (15, 16). Taken together, our data suggest that the induction of an unresponsive program in autoreactive B cells may represent a risk of developing B cell lymphoproliferation in primary SS.

MATERIALS AND METHODS

Study subjects.

We recruited 66 patients (59 women and 7 men) with primary SS according to the American–European Consensus Group criteria (17). Their mean ± SD age was 55.9 ± 14.5 years. Fourteen of these patients had B cell LPD. B cell lymphoproliferation was defined as an overt B cell non-Hodgkin's lymphoma (NHL) according to the World Health Organization classification criteria (18) or as the presence of type II mixed cryoglobulinemia (7, 8), which is sometimes associated with lymph node or splenic enlargement. Seven patients had B cell NHL, including 4 marginal zone, 1 lymphocytic, and 2 undetermined lymphomas. Among the 7 remaining patients, who had type II mixed cryoglobulinemia, 2 had lymph node enlargement and 2 had splenomegaly. Blood samples from 30 healthy donors were obtained from Etablissement Français du Sang (Hôpital Pitié-Salpêtrière). The study was performed according to the Declaration of Helsinki. All study subjects provided informed consent, in accordance with the Institutional Review Board.

Phenotypic analysis.

Peripheral blood mononuclear cells (PBMCs) were obtained by density-gradient centrifugation. Phenotypic analyses were performed with anti-human monoclonal antibodies. Anti-human CD10, CD19, CD20, CD21, CD22, CD44, CD69, CD86, HLA–DR, IgM, and annexin V were obtained from Beckman Coulter. Anti-human CD1c, CD25, CD27, CD38, CD84, CD95, BR3, κ light chain, and IgD were obtained from BD Biosciences. Fluorescence-activated cell sorting analyses were performed on a Navios flow cytometer using CXP analysis software (Beckman Coulter). CD19+ B lymphocyte counts (cells per microliter) were established from fresh blood samples using Cyto-Stat tetraCHROME kits with Flowcount fluorescent beads as an internal standard and tetraCXP software with a Navios cytometer according to the recommendations of the manufacturer (Beckman Coulter).

B cell enrichment and cell sorting.

CD27− B cells were enriched from total PBMCs by negative magnetic bead selection using a cocktail of biotinylated antibodies against CD2, CD14, CD16, CD23, CD27, CD36, CD4, and glycophorin A, followed by anti-biotin MicroBeads (naive B cell isolation kit II; Miltenyi Biotec). The purity of CD27− B cells was typically >95%. For proliferation assays, CD27− B cells were separated into CD21+ and CD21−/low fractions using a one-step magnetic bead–based selection process. Cells were fractionated by CD21 with phycoerythrin (PE)–conjugated anti-CD21 (BD Biosciences), followed by anti-PE MicroBeads (Miltenyi Biotec). The purity of CD21+ and CD21−/low fractions was typically >85%.

For gene expression profiling and single-cell sorting, PBMCs were sorted by staining with monoclonal anti-human antibodies against VioBlue-conjugated CD45, PE-conjugated anti-CD27 (Miltenyi Biotec), fluorescein isothiocyanate (FITC)– conjugated CD21, energy-coupled dye (ECD)–conjugated CD19, and allophycocyanin (APC)–conjugated CD10 (Beckman Coulter). For single-cell polymerase chain reaction (PCR), CD10−CD27−CD21−/lowCD19+ B cells from patients with primary SS were sorted on a FACSAria (Becton Dickinson) into 96-well PCR plates.

B cell activation, survival, and proliferation.

Enriched CD27− B cells, which contained CD21+ and CD21−/low B cells from patients with primary SS, were plated at 500,000 cells per well in a 48-well plate in RPMI 10% serum and 20 μg/ml polyclonal F(ab′)2 goat anti-human IgM (Jackson ImmunoResearch), 1 μg/ml anti-human CD40 (Invitrogen), and/or 1 μg/ml CpG (InvivoGen) for 48 hours. To assess B cell activation and survival, cells were stained with the anti-human monoclonal antibodies CD19, CD21, CD25, CD27, CD40, CD69, and CD95. The proportions of apoptotic and dead cells were assessed by flow cytometry to measure binding with annexin V using PE-conjugated annexin V and 7-aminoactinomycin D (7-AAD; BD PharMingen). To assess B cell proliferation, CD21+ and CD21−/low marginal zone B cells were plated at 1 × 105 cells per well of a 96-well flat-bottomed plate with various combinations of the following reagents: 20 μg/ml polyclonal F(ab′)2 goat anti-human IgM; 1 μg/ml anti-CD40; 1 μg/ml CpG; 2 μg/ml Toll-like receptor 3 (TLR-3) agonist (poly[I-C]; InvivoGen), and 1 μg/ml TLR-7 agonist (Gardiquimod; InvivoGen). Cells were incubated for 48 hours and pulsed for 8 hours with tritiated thymidine.

Calcium mobilization.

Cells were resuspended in RPMI 1640 (Gibco Invitrogen) and stained with PE-conjugated anti-CD21, ECD-conjugated anti-CD19 (Beckman Coulter), and APC-conjugated anti-CD27 (BD Biosciences) antibodies for 30 minutes at 4°C. Cells were washed twice in RPMI and resuspended at 1 × 106 cells/ml. Cells were loaded with Fluo-4 AM (Invitrogen) at a final concentration of 5 μM in the presence of 0.2% Pluronic F-127 (Sigma) for 30 minutes at room temperature. Cells were washed twice in RPMI–fetal calf serum 5% and resuspended at 1 × 106 cells/ml. [Ca2+]i was monitored over time by flow cytometry on gated CD19+CD27−CD21+ and CD19+CD27−CD21−/low B cells. Baselines were read for 30 seconds, after which the cells were removed and stimulated with 20 μg/ml of F(ab′)2 anti-IgM (Jackson ImmunoResearch) and then ionomycin at 1 mg/ml (Sigma-Aldrich).

Antibody production, enzyme-linked immunosorbent assays (ELISAs), and immunofluorescence assays.

Cloning strategy, expression vectors, and antibody reactivity against specific antigens have been described previously (19). Highly polyreactive ED38 was used as positive control in HEp-2 reactivity and polyreactivity ELISAs (19). Antibodies were considered polyreactive when they recognized at least 2, and usually all 3, of the antigens analyzed, including double-stranded DNA, insulin, and lipopolysaccharide. All recombinant antibodies were also tested for RF reactivity (anti-IgG) as previously described (20). For indirect immunofluorescence assays, HEp-2 cell–coated slides (Bion Enterprises) were incubated in a moist chamber at room temperature with purified recombinant antibodies at 50–100 μg/ml. FITC-conjugated goat anti-human IgG was used as a detection reagent. Pictures were taken with an Axioskop microscope (Zeiss) using a Plan neofluar objective (40×; 0.75 numerical aperture) and AxioVision 3.1 acquisition software.

Microarray gene expression profile analysis.

RNA was extracted from 105–3 × 105 batch-sorted conventional IgM+CD27−CD21+CD19+ B cells and IgM+CD27−CD21−/lowCD19+ B cells using the NucleoSpin RNA II kit (Macherey-Nagel). Each sample contained 50–200 ng of RNA, and the quality of the purified RNA was assessed with a 2100 Bioanalyzer from Agilent. Using an Illumina TotalPrep RNA Amplification Kit (Applied Biosystems), 50 ng of RNA was amplified and labeled to produce complementary DNA (cDNA). Labeled cDNA was hybridized on whole human genome chips (HumanHT-12 v3 Expression BeadChip kit; Illumina).

PredictSearch, a powerful bioinformatics solution dedicated to identifying relevant correlations between genes and concepts, was used to generate functional networks. This tool is updated daily with the whole NCBI PubMed database and seeks relevant gene–gene or gene–concept correlations within abstracts. It also enables an oriented search for co-citations between genes with action words such as “activation” or “repression” and their semantic variants. PredictSearch software was also implemented with the whole set of transcriptional signatures (>18,000) deposited in the NCBI GEO database and extracted with the DBF-MCL algorithm using the TranscriptomeBrowser tool (21). Furthermore, PredictSearch provides direct access to already published gene pathways.

Statistical analysis.

Data are expressed as the mean ± SD or median. Categorical variables were compared using Fisher's exact test or chi-square test, and continuous variables were compared using the t-test or Mann-Whitney U test when appropriate. All tests were 2-sided with a significance level of 0.05. Graphing and statistical analyses were performed using GraphPad Prism software.

RESULTS

Unusual expansion of peripheral CD27−CD21−/low B cells in patients with primary SS.

We first examined B cell subpopulations in patients with primary SS. We found an increase in the CD27− B cell subset in patients with primary SS and those with primary SS and LPD compared to healthy donors (mean percentage of cells of the CD19+ B cell subset 78.1% in patients with primary SS and 76.8% in patients with primary SS and LPD versus 64.6% in controls; P < 0.001 and P < 0.01, respectively) (Figures 1A and B). In contrast, the absolute numbers of CD27− B cells were decreased in patients with primary SS and patients with primary SS and LPD compared to controls (108.4 cells/μl and 96.4 cells/μl, respectively, versus 129.2 cells/μl in controls; P < 0.0001 for both).

Figure 1.

Expansion of an unusual CD27−CD21−/low B cell population in patients with primary Sjögren's syndrome (SS). A and B, Increased proportion of CD27− B cells in patients with primary SS (pSS) and patients with primary SS and lymphoproliferative disease (LPD) compared with healthy donors (HDs). Values in B are the mean ± SD. C and D, Increased proportion of CD27−CD21−/low B cells in patients with primary SS and LPD compared with patients with primary SS and healthy donors. Results of subanalyses of patients with primary SS and LPD with type II cryoglobulinemia (cryo) and patients with primary SS and LPD with non-Hodgkin's lymphoma (NHL) are shown. In D, circles represent individual patients; horizontal lines show the median. E and F, Decreased expression of IgD and IgM in CD27−CD21−/low B cells (gate II) compared to CD27−CD21+ B cells (gate I). G, CD27−CD21−/low cells display a unique B cell phenotype. Cells were gated on CD27−CD19+ B cells in order to delineate phenotype differences between CD21−/low and CD27−CD21+ B cell populations for B cell markers (CD19, CD20, BAFF receptor [BAFF-R], and κ light chain), maturity markers (CD10, CD38, CD22, and CD40), and activation/proliferation markers (CD69, CD86, CD95, and HLA–DR). ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

Analysis of CD21 expression on B cells showed an increase in both the median percentage and the absolute numbers of CD27−CD21−/low B cells in patients with primary SS and patients with primary SS and LPD compared to healthy donors (median percentage of cells of the CD19+ B cell subset 8.7% in patients with primary SS and 25.9% in patients with primary SS and LPD versus 3.1% in controls [P < 0.0001 for both] and absolute number 12.8 cells/μl in patients with primary SS and 42.7 cells/μl in patients with primary SS and LPD versus 6.2 cells/μl in controls [P < 0.0001 for both]) (Figures 1C and D). Subanalysis of patients with primary SS and LPD (Figure 1D) showed that the patients with type II cryoglobulinemia (n = 7) and those with NHL (n = 7) had a median percentage of CD27−CD21−/low B cells of 18.1% and 64.5%, respectively (P = 0.16). The percentage of CD27−CD21−/low B cells was significantly higher in patients with primary SS and LPD than in patients with primary SS without LPD (P < 0.001).

The expression of IgD and IgM was analyzed on CD27−CD21+ and CD27−CD21−/low B cells from patients with primary SS. CD27−CD21−/low B cells displayed decreased expression of IgD and IgM compared to conventional CD27−CD21+ B cells. While CD27−CD21−/low B cells comprised IgD+ and IgD− cells, 92.5% of the cells were positive for surface IgM (Figure 1E). The mean fluorescence intensity (MFI) of IgD expressed by CD27−CD21−/low B cells was similar to that expressed by CD27−CD21+ B cells. The proportion of IgD+ cells was lower in CD27−CD21−/low cells than in CD27−CD21+ B cells (Figure 1E). In addition, CD27−CD21−/low B cells expressed lower levels of surface IgM than did CD27−CD21+ B cells from the same patients (MFI 27.9 versus 46.6; P < 0.001) (Figure 1F). Four patients with primary SS and LPD had low C4 complement levels, but there was no correlation between CD27−CD21−/low B cell expansion and complement deficiency. We conclude that CD27−CD21−/low B cells are expanded in patients with primary SS, especially those displaying features of lymphoproliferation.

Unique B cell phenotype of CD27−CD21−/low cells.

In order to provide a more precise definition of the phenotype of this expanded population, CD27−CD21−/low B cells were screened by flow cytometry for characteristic B cell markers, maturity markers, and activation/proliferation markers, allowing a comparison with conventional CD27−CD21+ B cells from the same patients. CD27−CD21−/low B cells expressed higher levels of CD19 and CD20 when compared with CD27−CD21+ B cells from the same patients, whereas they expressed lower levels of BAFF receptor and κ light chain (Figure 1G). CD27−CD21−/low B cells did not express markers found on immature B cells, such as the B cell progenitor marker CD10 and, like mature cells, expressed only low levels of the development marker CD38. These markers are commonly used to distinguish new emigrant/immature/transitional B cells (22). Furthermore, the CD27−CD21−/low and conventional CD27−CD21+ B cell populations were both positive for the BCR-associated regulator CD22, which is expressed on mature B cells (23), and for other molecules found only on mature circulating B cells, including CD40 and CD44. Last, analysis of CD27−CD21−/low B cells for signs of recent activation in vivo demonstrated an increased expression of CD69, CD86, and CD95 compared to conventional CD27−CD21+ B cells from the same patients (Figure 1G). Furthermore, CD27−CD21−/low B cells were not proliferating, as assessed by Ki-67 expression. Hence, CD27−CD21−/low B cells show a mildly activated B cell phenotype potentially associated with chronic stimulation.

Expression of highly autoreactive antibodies by CD21−/low B cells from patients with primary SS.

CD27−CD21−/low B cells with a similar phenotype have been reported in several autoimmune diseases (14, 15, 24, 25). CD27−CD21−/low B cells from patients with RA and patients with common variable immunodeficiency express mostly unmutated IgM, whereas those from patients with HCV-MC harbor mutated Ig genes (14–16). To assess whether CD27−CD21−/low B cells from patients with primary SS belonged to the naive or memory B cell fraction, we analyzed the Ig repertoire and reactivity of these B cells using a single B cell cloning approach (19). We found that most CD27−CD21−/low B cells from 4 patients with primary SS expressed diverse IgM antibodies, the majority of which were mutated, suggesting that these B cells belonged to the memory compartment (see Supplementary Tables 1, 2, 3, and 4, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37828/abstract). In addition, somatic hypermutations were mostly located in Ig complementarity-determining regions (CDRs) as in conventional memory B cells, suggesting that CD27−CD21−/low B cells were antigen selected (see Supplementary Figure 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37828/abstract).

Many CD27−CD21−/low B cells expressed HEp-2– reactive and polyreactive antibodies (Figures 2A and B), including some with RF (anti-IgG) reactivity (Figure 2C). CD27−CD21−/low B cells often expressed antibodies recognizing cytoplasmic, and to a lesser extent, nuclear structures with diverse staining patterns (Figures 2D and E). We conclude that CD27−CD21−/low B cells from patients with primary SS are enriched in autoreactive B cell clones that may have been selected by self antigens.

Figure 2.

CD21−/low B cells in primary Sjögren's syndrome (SS) express highly autoreactive antibodies. A–C, Antibodies from CD27−CD21−/low B cells from 4 patients with primary SS and lymphoproliferative disease (patients pSS01, pSS1003, pSS201, and pSS202) were tested by enzyme-linked immunosorbent assay for anti–HEp-2 cell reactivity (A), against double-stranded DNA (dsDNA), insulin, and lipopolysaccharide (LPS) (B), and for rheumatoid factor (anti-IgG) reactivity (C). D and E, CD27−CD21−/low B cells often expressed antibodies recognizing cytoplasmic, and to a lesser extent, nuclear structures. Values in D are the mean ± SD. OD = optical density.

Defects in BCR and CD40-mediated activation in CD27−CD21−/low B cells.

We analyzed the function of CD27−CD21−/low B cells after BCR (using F[ab′]2 anti-IgM), CD40 (using CD40L), and/or TLR-9 (using CpG) triggering by studying the expression of CD25, CD69, CD40, and CD95 on CD27−CD21−/low and conventional CD27−CD21+ B cells from patients with primary SS (Figure 3). BCR, CD40, and/or TLR-3, TLR-7, and TLR-9 triggering induced the expression of CD25, CD69, and CD40 on CD27−CD21+ B cells from patients with primary SS but only resulted in a slight expression of CD95. In contrast, BCR and/or CD40 triggering of CD27−CD21−/low B cells from the same patients failed to properly induce the expression of CD25, CD69, and CD40, while the use of an alternative pathway through TLR-3, TLR-7, and TLR-9 triggering strongly induced the expression of CD25 and CD69 compared with CD27−CD21+ B cells. In addition, CD27−CD21−/low B cells exhibited increased CD95 expression after BCR, CD40, or TLR-3, TLR-7, and TLR-9 stimulation compared with CD27−CD21+ B cells. Thus, CD27−CD21−/low and CD27−CD21+ B cells display distinct B cell responses after stimulation, and CD27−CD21−/low B cells reveal defects in BCR and CD40-mediated activation, but not after TLR triggering (Figure 3).

Figure 3.

CD27−CD21−/low B cells from patients with primary Sjögren's syndrome reveal defects in B cell receptor– and CD40-mediated activation. A–C, CD69 (A), CD25 (B), and CD40 (C) expression levels were not up-regulated on CD27−CD21−/low B cells stimulated with F(ab′)2 anti-IgM or recombinant human CD40L, but CD69 and CD25 were up-regulated on cells stimulated with Toll-like receptor 3 (TLR-3) agonist (poly[I-C]), TLR-7 agonist (Gardiquimod), or TLR-9 agonist (CpG). D, CD27−CD21−/low B cells also exhibited an increased CD95 expression compared with CD27−CD21+ B cells. Values are the mean ± SD from at least 4 independent experiments. ∗ = P < 0.05 versus stimulation with anti-IgM and/or CD40L. MFI = mean fluorescence intensity.

We next analyzed the ability of CD27−CD21−/low B cells to induce intracellular calcium flux upon BCR stimulation, as compared with conventional CD27−CD21+ B cells. In contrast to CD27−CD21+ B cells from patients with primary SS, BCR triggering induced only a weak elevation of intracellular calcium concentration in response to BCR stimulation in CD27−CD21−/low B cells. Hence, CD27−CD21−/low B cells do not exhibit proper calcium flux after BCR stimulation (Figure 4).

Figure 4.

CD21−/low B cells from patients with primary Sjögren's syndrome (pSS) fail to increase [Ca2+]i in response to B cell receptor (BCR) triggering. A, Attenuated calcium responses in CD21−/low B cells compared with CD21+ B cells after BCR crosslinking. B, Fluo-4 intensity in CD21+ B cells from healthy donors (HD) and in CD21+ and CD21−/low B cells from patients with primary SS. C, Slightly higher mean fluorescence of Fluo-4 in the CD21−/low B cell population. D, Higher increase in maximal normalized peak fluorescence of Fluo-4 in the CD21+ B cell population. Values in C and D are the mean ± SD from 5 independent experiments. ∗ = P < 0.05; ∗∗ = P < 0.01.

CD27−CD21 −/low B cells are prone to dying by apoptosis and do not proliferate after BCR and CD40 triggering.

We assessed apoptosis of CD27−CD21−/low B cells by analyzing annexin V and 7-AAD staining. We found that freshly isolated CD27−CD21−/low B cells from patients with primary SS contained a higher frequency of annexin V+7-AAD− early apoptotic cells and annexin V+7-AAD+ late apoptotic and necrotic cells compared with conventional CD27−CD21+ B cells (see Supplementary Figures 2A and B, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37828/abstract). Compared with medium, BCR, CD40, and/or TLR-9 triggering was not able to rescue CD27−CD21−/low B cells from apoptosis. In contrast, stimulated CD27−CD21+ B cells contained lower levels of annexin V–expressing cells (see Supplementary Figures 2A and B). Thus, CD27−CD21−/low B cells are more susceptible to apoptosis than CD27−CD21+ B cells and are poorly rescued by BCR, CD40, and/or TLR-9 triggering.

We next studied the proliferation induced by BCR and TLR triggering using the incorporation of tritiated thymidine. CD27−CD21−/low B cells showed decreased cell proliferation compared with CD27−CD21+ B cells, mainly after BCR triggering and to a lesser extent after TLR-9 stimulation (see Supplementary Figure 2C, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37828/abstract). Thus, CD27−CD21−/low B cells exhibit increased apoptosis and decreased proliferation after stimulation, features that are commonly associated with an anergic phenotype.

Gene expression profiles of CD27−CD21−/low B cells in patients with primary SS reveal the selective up-regulation of an anergic pathway.

In order to define gene signatures associated with the expansion, phenotype, and function of CD27−CD21−/low B cells, we compared gene expression profiles of CD27−CD21−/low B cells with those of conventional CD27−CD21+ B cells isolated from 4 patients with primary SS (Figure 5A). A total of 2,510 transcripts were found to be differentially expressed between the 2 CD27− B cell subpopulations, including 1,934 transcripts that were up-regulated (>1.5-fold change) and 576 transcripts that were down-regulated (<0.7-fold change) in CD27−CD21−/low B cells. Focusing on the transcriptional modulations that discriminate CD27−CD21−/low from CD27−CD21+ B cells in the same patients, we identified transcriptional signatures that are likely to underlie the functional properties of these cells. Hierarchical clustering of differentially expressed genes between CD27−CD21−/low and CD27−CD21+ B cells highlighted an anergy-related transcriptional signature that is specifically up-regulated in CD27−CD21−/low B cells (Figures 5A and B).

Figure 5.

Gene array comparisons of CD21+ and CD27−CD21−/low B cells from patients with primary Sjögren's syndrome (pSS) using the Illumina Human Whole Genome Array. A, Overexpressed transcripts differentially expressed by CD21−/low and CD27−CD21+ B cells from 4 patients with primary SS. Up-regulated and down-regulated transcripts are indicated in red and green, respectively. B, Overexpressed transcripts involved in an anergy and tolerance breakdown cluster. RTKs = receptor tyrosine kinases; BCR = B cell receptor.

The gene content of this transcriptional module was enriched in genes encoding cell surface and intracellular proteins that are involved in signal transduction and in T cell costimulation, such as CD83, CD86/B7-2, and CD72. CBL-B, which is a potent negative regulator of intracellular signaling, was also strongly induced in the CD27−CD21−/low versus CD27−CD21+ B cells. CD72 was shown to function as a regulator of peripheral B cell tolerance, by regulating and being associated with CBL-B, leading to the down-regulation of BCR signaling. Similarly, CD27−CD21−/low B cells up-regulated the transcription of a whole set of immunoreceptor tyrosine–based inhibition motif (ITIM) receptor genes, which are likely to inhibit B cell activation and proliferation, including CD72, SIGLEC, CD22, and FCRL2 genes. We also found a selective up-regulation of FCRL3, which was suggested to play a pivotal role in autoimmunity. Finally, up-regulation of RFNT1/Raftlin, the B cell–specific major raft protein that is necessary for the integrity of lipid rafts and BCR signal transduction, further highlights the related functional involvement of the genes of this transcriptional signature.

Through flow cytometry, we confirmed at the protein level that CD27−CD21−/low B cells up-regulated FCRL2, FCRL3, CD72, CD22, and CD11c, and down-regulated CD1c (Figure 6). Taken together, these findings of transcriptomic analysis of CD27−CD21−/low B cells highlight the selective up-regulation of genes that belong to an inhibitory pathway, which could explain in part the functional anergy of CD27−CD21−/low B cells.

Figure 6.

Fluorescence-activated cell sorting (FACS) analysis of proteins differentially expressed between CD21+ and CD27−CD21−/low B cells from patients with primary Sjögren's syndrome (SS). FACS plot analyses of surface markers overexpressed and down-regulated in CD27−CD21−/low B cells are shown. FCRL2, FCRL3, CD72, CD11c, and CD22 were overexpressed at the protein level in CD27−CD21−/low B cells from patients with primary SS, as found in transcriptomic analysis. In contrast, CD1c marker was down-regulated at the protein level in CD27−CD21−/low B cells from patients with primary SS.

DISCUSSION

In this study, we identified a unique autoreactive B cell population in patients with primary SS that lacks CD21 expression, is refractory to B cell stimulation, and correlates with the lymphoproliferative status. Unresponsive CD21−/low B cells expressing IgM have been identified in patients with RA and patients with common variable immunodeficiency as well as in those with HCV-MC (14–16). The analysis of the Ig repertoire of CD21−/low B cells from patients with primary SS showed that their Ig genes were mutated, indicating that these B cells belonged to the memory compartment. Hence, CD21−/low B cells from patients with primary SS are similar to anergic CD21−/low B cells from patients with HCV-MC, which also expressed mutated IgM (15, 16), whereas most CD21−/low B cells from patients with RA and those with common variable immunodeficiency expressing unmutated Ig likely belong to the naive B cell compartment (14, 24). However, CD21−/low B cells from patients with primary SS differ from their counterparts in patients with HCV-MC, in that they do not express CD27 (15, 16).

Nonetheless, CD21−/low B cells from all patients are refractory to most stimulation, suggesting that autoreactive CD21−/low B cells may use a common anergic program to be tolerized. Indeed, the phenotype of CD21−/low B cells revealed the modulation of the expression of many molecules that may prevent the activation of these B cells. The down-regulation of the complement receptor CR2/CD21 previously reported on mouse and human anergic B cells is likely to result in an increased BCR signaling threshold and compromise the ability of CD21−/low B cells to be activated (14–16, 26, 27). In addition, CD21 is able to break anergy in mouse B cells, further suggesting that its down-regulation contributes to an anergic stage in mice and humans (14–16, 26, 28, 29).

In patients with primary SS, CD21−/low B cells were highly enriched with autoreactive clones. In addition, gene array experiments analyzing CD21−/low B cells from patients with primary SS affected the expression of genes such as CD19, CD22, CD72, and FCRL3, which have been found to be involved in the development of many autoimmune diseases (30–32). These genes encoded molecules belonging to important pathways leading to B cell activation, including those initiated by the BCR. The up-regulated transcription of CD86, CD72, GRB2, CBL-B, and FCRL2 genes is consistent with an anergy-like phenotype. CD86 as well as CD69 and CD95 expression normally induced on activated B cells may reflect the chronic stimulation status of autoreactive CD27−CD21−/low B cells in patients with primary SS. CD72 was shown to function as a regulator of peripheral B cell tolerance, by regulating and being associated with CBL-B, leading to the down-regulation of BCR signaling (33). Interestingly, CBL-B was shown to be a key protein required for the induction of anergy in T cells (34).

The up-regulated transcription of ITIM receptor genes, such as CD72, CD22, FCRL2, and SIGLEC, is also likely to inhibit B cell activation and proliferation. The validation of the up-regulation of many genes by flow cytometry further demonstrates the functional unresponsiveness of CD27−CD21−/low B cells in patients with primary SS. Indeed, CD21−/low B cells from patients with primary SS showed impaired calcium-mediated signaling, did not up-regulate activation markers, and did not proliferate in response to BCR triggering. CD21−/low B cells were also prone to die faster than their CD21+ counterparts, suggesting that they have a shorter half-life. In addition, the increased CD95 expression on CD21−/low B cells could be responsible for their increased susceptibility to cell death. Most of these features are characteristic of anergic mouse B cells (35) and of more recently described anergic human B cells (14–16).

Lastly, we observed that autoreactive CD27−CD21−/low B cells were functionally attenuated after BCR and/or CD40 triggering but not after TLR stimulation as reported in patients with HCV-MC (16). Leadbetter et al (36) showed that activation of autoreactive RF-positive B cells was mediated by immune complexes and required the synergistic engagement of the BCR and members of the myeloid differentiation factor 88–dependent TLR family (TLR-7, TLR-8, and TLR-9); this thus establishes a critical link between the innate and adaptive immune systems in the development of systemic autoimmune disease. Activation in this mode is therefore likely to be a fundamental event in the loss of peripheral B cell tolerance in a wide variety of settings, and these data establish a critical role for endogenous TLR ligands in aberrant activation of the adaptive immune system in autoimmunity. The chronic stimulation of polyclonal B cells secreting autoreactive antibodies, such as RF, is proposed to be the first event of lymphomagenesis in SS (10). Interestingly, patients with primary SS and those with HCV-MC share increased production of BAFF (2, 3) and an increased risk of lymphoma, particularly mucosa-associated lymphoid tissue–type lymphomas (4, 5, 37).

Recalling what has been observed in HCV-MC, the hypothesis of a viral trigger has long been suspected in primary SS. Different studies have detected Epstein-Barr virus (EBV) genome and proteins in the salivary glands of patients with primary SS (38–40). EBV-encoded small RNA, a double-stranded RNA, forms a complex with SSB to induce TLR-3–related secretion of type I interferon and other proinflammatory cytokines (40). Although it is unclear at this point if infectious agents or self-antigens may favor the emergence of CD21−/low B cells in patients with primary SS, their pattern of somatic hypermutations with replacement mutations located in CDRs interacting with antigens highly suggests that these B cells followed an antigen-driven selection as reported for conventional memory B cells. Then, these autoreactive B cells may give rise to the emergence of a preferential clone, leading to lymphoproliferation. The demonstration that these lymphoma B cells may display RF activity (13) supports the hypothesis that SS-associated lymphoma B cells grow through an autoantigen-driven process. However, our sequence analysis of the Ig repertoire of CD21−/low B cells from the 4 patients with primary SS has not identified such monoclonal expansion, which may initially develop in the salivary glands of these patients and may not be identified in the patient's blood until later stages of the disease.

Taken together, our findings indicate that elevated numbers of highly autoreactive unresponsive CD21−/low B cells remain in the blood of patients with primary SS instead of being eliminated. Immune reactions may create a favorable environment to break tolerance and eventually activate these CD21−/low B cells, potentially leading to the development of B cell lymphoproliferation.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Saadoun had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Saadoun, Terrier, Musset, Meffre, Cacoub.

Acquisition of data. Saadoun, Terrier, Bannock, Vazquez, Massad, Kang, Joly, Rosenzwajg, Sene, Benech, Musset, Klatzmann, Meffre, Cacoub.

Analysis and interpretation of data. Saadoun, Terrier, Meffre.

ADDITIONAL DISCLOSURES

Authors Joly and Benech are employees of Prediguard.

Acknowledgements

We thank Marlene Garrido, Nathalie Ferry, Audrey Peres-Lascar, Catherine Blanc, Fabien Pitoiset, and Cesar Prucca for technical assistance. We thank Drs. L. Devine and C. Wang for single-cell sorting.

Ancillary