Analysis of peripheral B cell subsets in patients with systemic lupus erythematosus (SLE) has provided evidence of specific alterations, such as an expansion of CD27++ plasma cells/blasts and transitional B cells. However, memory B cells in lupus have not been thoroughly investigated, and only recently a CD27− memory B cell subset was identified in the peripheral blood of lupus patients. Focusing on CD27− B cells, this study aimed to identify abnormalities in peripheral B cell subsets in patients with SLE.
Three independent cohorts of lupus patients were used to characterize CD27− memory B cells, using multiparameter flow cytometry and single-cell reverse transcription–polymerase chain reaction of heavy-chain transcripts.
We identified a homogeneous subset of CD27−,IgD−,CD95+ memory B cells with an activated phenotype that was increased in patients with disease flares and that correlated with disease activity and serologic abnormalities. In contrast, the entire subset of CD27−,IgD− B cells was found to be heterogeneous, did not correlate significantly with lupus activity, and was also increased in patients with bacterial infections.
We conclude that CD95 is a useful marker to identify CD27− memory B cells with an activated phenotype, which might serve as a biomarker for lupus activity and as a target of further investigations aiming to elucidate the pathogenic potential of these cells and the mechanisms involved in the generation as well as regulation of this CD27−,IgD−,CD95+ memory B cell subset.
B cell monitoring has been extensively used recently to assess the effect of B cell–depleting or B cell–modulating therapies and the reconstitution of the peripheral blood B cell repertoire after treatment with B cell–depleting drugs. Expression of CD27 has been particularly useful to discriminate B cell subsets. Although CD27 was originally thought to distinguish between memory B cells and plasma cells and between memory B cells and naive B cells (1, 2), more recently the heterogeneity of CD27− B cells has become apparent. The contribution of transitional B cells to this subpopulation has been shown to be high in some patients with systemic lupus erythematosus (SLE) (3). In this regard, after B cell depletion, immature and transitional B cells dominate the peripheral blood for months (4). Conversely, a recent analysis of B cell subsets in patients with Sjögren's syndrome demonstrated a predominance of CD27− B cells, of which ∼10% were B cells with somatically hypermutated immunoglobulin V-region gene rearrangements (5). Moreover, in normal subjects, IgG+,CD27− B cells carrying mutated B cell receptors have been identified both in tonsils and in peripheral blood (6–9). Although in a subset of patients with lupus, the CD27− B cell subpopulation is markedly reduced for several reasons, including long-term immunosuppression, rapid turnover of naive B cells caused by enhanced activation, or disease-associated antibodies to naive B cells (10–13), patients with increased frequencies of CD27− B cells have also been described (14).
Recently, it was reported that CD27−,IgD− B cells represent memory B cells and are associated with increased disease activity and renal disease in lupus patients (15). Similar to transitional B cells, and also typical of memory B cells, these CD27−,IgD− B cells lack expression of the ABCB1 transporter and are therefore unable to extrude rhodamine 123 (9, 15). These studies suggest that B cell homeostasis is disturbed in patients with SLE with elevated percentages of antigen-experienced, activated B cells and transitional B cells and a relative decline in the number of naive B cells. Because CD27 may not be useful for discerning these subpopulations reliably in lupus patients, we evaluated a panel of activation and differentiation markers in 3 independent cohorts of lupus patients, normal healthy subjects, and patients with bacterial infections, using flow cytometry in an attempt to identify and characterize abnormalities in peripheral B cell subset dynamics more completely.
We observed that CD95 is a useful marker for identifying atypical, activated CD27− and largely IgD− memory B cells of pathogenic relevance. CD95 expression by circulating B cells has previously been described to be increased in patients with active SLE (16), but here we report that a specific subset of CD27−,IgD−,CD95+ memory B cells is increased in patients with active disease and comprises a population of larger cells that express activation markers and mutated heavy-chain transcripts. While the results of the current study confirm the previously reported relative expansion of CD27−,IgD− B cells in the peripheral blood of lupus patients, the presence of this subset did not correlate significantly with disease activity. Moreover, this population was found to be expanded in patients with infections as well. CD27−,IgD− B cells are a heterogeneous subpopulation that contains CD10+ transitional B cells in individual lupus patients as well as memory B cells with or without signs of activation (e.g., CD95 expression). The results indicate that CD27−,IgD−,CD95+ B cells fully meet the criteria for memory B cells and apparently result from aberrant B cell activation and/or differentiation in lupus patients with active disease.
PATIENTS AND METHODS
Following informed consent for the protocol approved by the Institutional Review Boards at Charité University Hospitals, Berlin, Germany and Columbia University Medical Center, New York, NY and by the Committee for Clinical Investigations at Albert Einstein College of Medicine, Bronx, NY, samples of whole blood and relevant clinical information were obtained prospectively from 3 cohorts of lupus patients and 4 control populations. SLE patients were recruited sequentially and fulfilled the American College of Rheumatology 1982 revised classification criteria (17). Disease activity was assessed using the SLE Disease Activity Index (SLEDAI) (18). The characteristics of all 3 patient cohorts, including age, sex, disease activity, treatment, major disease manifestations, and serologic parameters as well as B cell counts, are shown in Table 1.
Table 1. Summary of clinical and serologic characteristics of the lupus patients analyzed*
Cohort 1 (n = 29)
Cohort 2 (n = 53)
Cohort 3 (n = 40)
Except where indicated otherwise, values are the number (%) of patients. Cohort 1 included 29 patients with systemic lupus erythematosus (SLE) in whom the surface expression of CD95 by CD27− B cells was analyzed. Cohort 2 included 53 SLE patients in whom the surface expression of IgD by CD27− B cells was analyzed. Cohort 3 included 40 SLE patients in whom the surface expression of IgD and CD95 by CD27− B cells was analyzed. Anti-dsDNA = anti–double-stranded DNA; ELISA = enzyme-linked immunosorbent assay; CNS = central nervous system; SLEDAI = SLE Disease Activity Index.
Not combined with another disease-modifying drug.
Prednisone equivalent at 5–20 mg/day.
Prednisone equivalent at >50 mg/day, administered as an intravenous bolus in most cases.
Cohort 1, which was used to analyze CD95 expression, consisted of patients with active disease and intensive immunosuppressive treatment who were hospitalized at Charité University Hospitals. Cohort 2, which was used to analyze IgD expression, was recruited from the inpatient and outpatient Rheumatology Clinics at Charité University Hospitals and was more heterogeneous than cohort 1 with regard to disease activity. For comparative purposes, 16 and 18 age-, race-, and sex-matched healthy control subjects were analyzed for cohorts 1 and 2, respectively. In addition, 17 patients with nonautoimmune disease who were hospitalized because of bacterial infections were recruited in order to analyze surface expression of CD95 (5 men and 5 women, mean ± SD age 66.6 ± 16.4 years, mean ± SD C-reactive protein [CRP] level 11.7 ± 7.8 mg/dl, mean ± SD leukocyte count 15.9 ± 5.8 cells/nl) or IgD (4 men and 3 women, mean ± SD age 47.7 ± 23.1 years, mean ± SD CRP level 9.1 ± 7.8 mg/dl, mean ± SD leukocyte count 10.2 ± 2.9 cells/nl). In a third cohort comprising 40 ethnically diverse (17 Hispanics, 12 African Americans, 5 Caucasians, 4 African Caribbeans, and 2 Asians) lupus patients recruited from the outpatient Rheumatology Clinics at Charité University Hospitals, Albert Einstein College of Medicine, and Columbia University Medical Center, IgD and CD95 expression were analyzed simultaneously.
Preparation of blood samples and flow cytometric analysis.
Peripheral blood mononuclear cells (PBMCs) were prepared for flow cytometric analysis as previously reported (19). Immunofluorescence labeling was performed by incubating PBMCs with monoclonal antibodies to CD3 (UCHT-1), CD14 (M5E2), CD16 (3G8), IgD (IA6-2), CD95 (DX2), CD19 (SJ25C1), IgM (G20-127), CD5 (UCHT2), CD10 (HI10a), CD44 (G44-26), CD71 (M-A712), IgG (G18-145), HLA–DR (L243), CXCR3 (1C6/CXCR3), or CXCR4 (12G5) (all from BD PharMingen, San Diego, CA) or to CD21 (BU32), CD38 (HIT2), CD86 (BU63), or CD27 (CLB-27/1) (all from Invitrogen, Carlsbad, CA) or to IgA (CBL114F; Chemicon, Temecula, CA) or to CXCR5 (51505.111; R&D Systems, Minneapolis, MN) or to CD19 (HD37; Dako, Glostrup, Denmark). To prevent nonspecific Fc receptor–mediated antibody binding, human immunoglobulin (Beriglobin; Aventis Behring, Marburg, Germany) was added (1:50 dilution) before the incubation step (cohorts 1 and 2) performed in phosphate buffered saline/0.5% bovine serum albumin/5 mM EDTA at 4°C for 10 minutes. Propidium iodide (1 μg/ml; Sigma, Munich, Germany) or 4′,6-diamidino-2-phenylindole (final concentration 200 nM; Invitrogen) was added immediately before cytometric analysis to exclude dead cells. FACSCalibur (cohorts 1 and 2) or LSRII (cohort 3) instruments (Becton Dickinson, San Jose, CA) were used for flow cytometric analysis, and data were analyzed with FlowJo software (Treestar, Ashland, OR).
Fluorescence-activated cell sorting and single-cell reverse transcription–polymerase chain reaction (RT-PCR).
For the analysis of immunoglobulin heavy-chain messenger RNA (mRNA) transcripts of CD95+ and CD95− CD27−,IgD− and CD27+,IgD− B cell subsets, 10 ml of whole blood was obtained from 3 female SLE patients. Two patients (ages 25 years and 32 years, respectively) had disease flares with lupus nephritis, reduced complement levels, elevated levels of anti–double-stranded DNA (anti-dsDNA) antibodies, and circulating immune complexes at the time of analysis. In both patients SLE had been diagnosed 2–3 years previously, and both patients were treated with intravenous cyclophosphamide. The third patient (age 45 years) with longstanding disease had disease flares with Coombs-positive anemia and discoid skin lesions currently treated with azathioprine and prednisone. The patient had anti-Ro and anti-C1q antibodies but was negative for anti-dsDNA antibodies at the time of analysis.
PBMCs were isolated and labeled as previously described. Finally, CD95+ and CD95− CD27−,IgD− and CD27+,IgD− individual B cells (23 cells each) were sorted (MoFlow; DakoCytomation, Hamburg, Germany) into single wells containing modified 1× RT-PCR buffer (5 mM dithiothreitol, 400 ng oligo(dT)18, 0.2 mM dNTP, 1% Triton X-100, 10 units RNasin, 40 units avian myeloblastosis virus reverse transcriptase). First-strand complementary DNA (cDNA) was generated at 42°C for 60 minutes. The immunoglobulin heavy-chain mRNA transcripts were amplified by family- and isotype-specific nested PCR protocols using 5 μl cDNA in the first round (5). Briefly, a set of VH family–specific leader region primers with CH1μ, CH1γ, or CH1α constant region primers was used in the first round. In the second round, 5-μl aliquots of the external PCR mixture were specifically amplified for VH families using the VH family–specific primer complementary to the 5′ framework region 1 in combination with a JH region primer. Following column purification, all PCR products were directly sequenced using the BigDye termination sequencing kit (Perkin-Elmer, Emeryville, CA) and analyzed with the ABI 377 automated sequencer (Perkin-Elmer). V-region gene analysis was performed using Joinsolver (http://joinsolver.niams.nih.gov/index.htm). The maximal PCR error rate of this analysis has been documented to be 1 × 10−4 mutations/basepair (20). Thus, few if any of the nucleotide changes encountered in this study can be ascribed to amplification errors.
Differences between defined patient groups were compared using the nonparametric Mann-Whitney U test. Mean and SD values were determined only if the data showed a Gaussian approximation; otherwise, the median value and the range were used. Correlation was examined by Spearman's rank correlation test. Analysis of mutations was performed using the chi-square test. P values less than 0.05 were considered significant. Data were analyzed using GraphPad Prism 4 software (GraphPad Software, San Diego, CA).
Initially, an analysis of CD95 expression by peripheral B cells was carried out in 29 patients with SLE (cohort 1 in Table 1), 16 healthy subjects, and 10 patients with infections. In patients with SLE, a median 24.9% (range 6.9–65.9) of all peripheral B cells expressed CD95. The frequency of CD95-expressing B cells was significantly higher in patients with SLE than in patients with infections (median 14.0% [range 5.4–41.0]) (P = 0.008) or in normal individuals (median 7.8% [range 5.5–16.1]) (P < 0.0001). The increased expression of CD95 by B cells in patients with SLE and, to some extent, in patients with infections was associated with the presence of CD27++,CD20− plasmablasts or early plasma cells, which were almost uniformly positive for CD95 (in SLE patients, median 9.3% [range 1.8–70.4]; in patients with infections, median 6.7% [range 0.7–32.1]). These cells were, however, almost absent in the peripheral blood of normal healthy subjects. In addition, we found an elevated frequency of CD95-expressing CD27+ memory B cells in patients with SLE (median 11.4% of all B cells [range 1.7–32.1]) compared with patients with infections (median 7.8% [range 2.6–13.7]) (P = 0.04) and normal controls (median 5.1% [range 2.7–13.7]) (P = 0.01).
The upper plot shown in Figure 1A shows a representative example of CD95 and CD27 expression by B cells obtained from a patient with SLE. In the lower plot, results from a normal donor are shown. In addition to the increased frequency of CD27+,CD95+ memory cells and plasmablasts in lupus patients, an additional CD27−,CD95+ B cell subset was also noted. This subset was increased in SLE patients compared with normal subjects (median 2.4% [range 0.7–9.2] versus 1.2% [range 0.4–2.5]; P = 0.0002) (Figure 1B) and patients with infections (median 1.4% [range 0.4–2.3]) (P = 0.003).
A second analysis in another cohort of lupus patients (cohort 2, n = 53) (Table 1) addressed IgD surface expression by B cell subpopulations. As illustrated in Figure 1C, an enlarged CD27−,IgD− B cell subset was observed in patients with SLE but not in normal controls. This subset has been reported in some patients with SLE and has recently been characterized as memory B cells (13–15). We investigated this B cell subset further by comparing lupus patients and healthy controls as well as patients with bacterial infections and observed that the frequency of CD27−,IgD− B cells was significantly increased in SLE patients compared with healthy controls (median 10.2% [range 2.3–46.9] versus 3.4% [range 1.0–10.1]; P < 0.0001) (Figure 1D). Interestingly, this B cell subset was also found to occur at higher frequencies in patients with bacterial infections (median 8.2% [range 5.5–21.4]) than in normal controls (P = 0.009). The frequency of each CD27− B cell subset was examined for a possible relationship to SLE disease activity. While the frequency of CD27−,CD95+ B cells correlated significantly with disease activity as assessed by the SLEDAI (rs = 0.45, P = 0.01) (Figure 1E), no significant correlation was noted between the frequency of CD27−,IgD− B cells and the SLEDAI score in 53 patients (Figure 1F). Of note, however, there was a correlation between the frequencies of CD27−,IgD− B cells and CD27+,IgD− memory B cells (rs = 0.40, P = 0.003).
To validate these findings and further investigate the relationship between CD95+ and IgD− CD27− B cell subsets, analysis of the expression of both markers on CD27− B cells was carried out. This analysis (cohort 3) revealed variable expression of CD95 (median 34.6% [range 12.5–72.4]) (Figure 2A) on CD27−,IgD− B cells. However, most CD27−,CD95+ B cells were IgD− (median 75.7% [range 37.1–91.5]). These results suggest that CD27−,CD95+ B cells largely reside within the CD27−,IgD− B cell subset. Representative examples shown in Figure 2B illustrate that CD27−,IgD−,CD95+ B cells were observed in patients with SLE, especially in those with active disease, while this subset was very uncommon in the peripheral blood of normal subjects. A variable percentage of CD27−,IgD−,CD95+ B cells were also found to be IgM−, indicating that they had undergone immunoglobulin class switching (Table 2 and Figure 2C). This percentage was independent of anti-dsDNA antibody levels or disease activity in 10 patients analyzed.
Table 2. Comparison of VH gene transcripts amplified from CD95+ and CD95− CD27−,IgD− B cells and from CD27+,IgD− B cells from 3 patients with systemic lupus erythematosus*
Except where indicated otherwise, values are the number of transcripts (%) or their mutational frequency (%), respectively. CDR3 = third complementarity-determining region; R = replacement mutation; S = silent mutation; FR = framework region.
Three variable regions (2 from CD95+ B cells) were found to be expressed with 2 different constant regions (μ and γ).
Two variable regions were found to be expressed with 2 different constant regions (γ and α).
Cell size analysis indicated that the CD27−,IgD− B cell subset comprised both small and larger cells, while CD27−,IgD−,CD95+ B cells were uniformly large with increased CD19 expression, in contrast to the CD27−,IgD−,CD95− subset as shown in 2 representative lupus patients (SLE J21 and SLE J44) (Figure 3A). Increased CD19 expression was interpreted as being related to the difference in cell size rather than to an absolute increase in the density of cell surface CD19. As depicted in Figure 3, CD27−,IgD−,CD95+ B cells (and in selected patients [e.g., SLE J21] also their CD95− counterparts) differed from CD27−,IgD+ B cells in the surface expression of various activation markers (CD86, HLA–DR, CD38), chemokine receptors (CXCR3, CXCR4, CXCR5), CR2 (CD21), and CD44. Importantly, CD27−,IgD−,CD95+ B cells displayed these differences more clearly, showing higher levels of CD86, CXCR3, HLA–DR, and CD71 (not shown) expression and lower levels of CXCR4 and CXCR5 expression compared with CD27−,IgD−,CD95− B cells. In contrast, the surface expression of CD21, CD44, or CD38 was either comparable in both subsets, or the difference was more marked in CD27−,IgD−,CD95+ B cells.
As shown in Figure 3, the majority of CD27−, IgD− B cells were CD10− and also CD5− (CD5 not shown). In individual patients, however, a fraction of smaller cells were observed to express CD10 and IgM and were therefore considered to comprise immature or transitional B cells that do not yet express IgD or have it down-regulated (Figures 3B and C). Notably, these cells were exclusively observed in patients with an increased frequency of CD27−,IgD+,CD10+ B cells. Importantly, CD95 and CD10 were never coexpressed. These results suggest that CD27−,IgD−,CD95+ B cells represent a unique subset of pre- and postswitch B cells that are large and that express a specific set of activation and differentiation markers.
To determine whether perturbations in the distribution of any of the CD27− B cell subsets were associated with SLE disease activity, we grouped all patients of cohort 3 according to their SLEDAI scores and assessed the subset distribution. As shown in Figure 4A, patients with disease flares (SLEDAI score >4) had higher frequencies of CD27−,IgD−,CD95+ B cells (P < 0.02), CD27−,CD95+ B cells (P < 0.02), or CD27−,IgD− B cells (P < 0.04) compared with patients with inactive disease (SLEDAI score ≤4). Furthermore, the frequency of circulating CD27−,IgD−,CD95+ B cells (rs = 0.36, P < 0.025) (Figure 4B) or CD27−,CD95+ B cells (rs = 0.36, P < 0.025) (data not shown) correlated significantly with the SLEDAI score in contrast to the frequency of CD27−,IgD− B cells (rs = 0.27, P = 0.09) (data not shown).
An analysis of absolute numbers of CD27− B cell subsets also revealed a significantly increased number of circulating CD27−,IgD−,CD95+ B cells (P < 0.03) or CD27−,CD95+ B cells (P < 0.04) in patients with active disease compared with patients with inactive disease, while a nonsignificant trend toward a difference between the groups was identified when the number of CD27−, IgD− B cells was analyzed (P = 0.09) (Figure 4A). Patients with active disease had a significantly higher percentage of IgD− B cells within the CD27−,CD95+ B cell subset compared with patients with quiescent disease (median 77.5% [range 54.0–91.5] versus 69.6% [range 37.1–90.7]; P < 0.03). In addition, a nonsignificant trend toward a correlation was noted between absolute numbers of CD27−,IgD−,CD95+ B cells (rs = 0.30, P = 0.06), CD27−,CD95+ B cells (rs = 0.30, P = 0.06), or CD27−,IgD− B cells (rs = 0.19, P = 0.20) and the SLEDAI score. Compared with patients with quiescent disease, patients with active disease had an increased percentage of IgD−,CD95+ B cells (median 8.1% [range 2.3–27.6] versus 2.5% [range 0.3–16.6]; P < 0.005) or CD95+ B cells (median 9.9% [range 3.1–30.1] versus 4.4% [range 0.7–18.3]; P < 0.01) in the CD27− B cell subset, and patients with a high percentage of CD95+ B cells in the CD27−,IgD− B cell subset also exhibited an increased percentage of CD95+ B cells in the CD27+,IgD− subset (rs = 0.52, P < 0.001) (data not shown).
When clinical manifestations and serologic findings were analyzed, a significantly increased absolute number but not frequency of peripheral CD27−,IgD−, CD95+ (median 4.6/μl [range 0.6–13.3] versus 2.0/μl [range 0.3–5.3]; P < 0.02) (Figure 4C), CD27−,IgD− (median 11.3/μl [range 1.8–71.9] versus 5.3/μl [range 1.0–31.0]; P < 0.04), or CD27−,CD95+ (median 6.1/μl [range 0.8–17.8] versus 2.9/μl [range 0.6–6.3]; P < 0.007) B cells was noted in patients who showed consumption of complement factors. Despite this link to complement consumption, none of the CD27− B cell subsets was associated significantly with specific clinical manifestations or anti-dsDNA antibody levels (Table 1).
Since mutated and class-switched immunoglobulins are considered to be hallmarks of memory B cells (21, 22), and CD27 surface expression has been shown to coincide with somatic hypermutation (1, 23, 24), we analyzed individual CD95+ and CD95− CD27−,IgD− and CD27+,IgD− B cells using single-cell RT-PCR of immunoglobulin heavy-chain gene transcripts (Table 2). The lupus patients analyzed by flow cytometry seemed to be heterogeneous with regard to the immunoglobulin isotypes expressed by the CD27−,IgD− B cell subset; therefore, 2 patients with a low frequency and 1 patient with a high frequency of IgM-expressing CD27−,IgD− B cells were chosen for further analysis by single-cell RT-PCR.
Although Cα and Cγ transcripts obtained from the CD27−,IgD−,CD95− B cell subset were mutated, Cμ transcripts were in germline configuration (Table 2). Importantly, all transcripts obtained from the CD27−, IgD−,CD95+ B cell subset were mutated even if not immunoglobulin class switched, and the mutational frequency was comparable with the frequency observed in CD27+ memory B cells (7.8% and 7.0%, respectively) (Table 2). The analyses of the pattern of somatic hypermutation revealed no obvious abnormalities comparing CD27+,IgD− and CD27−,IgD− B cells. These findings support the conclusion that both targeting and selection of mutations were similar in both subsets.
Consistent with the results of flow cytometric analysis, patients differed in the immunoglobulin class expressed by CD27− and CD27+ IgD− B cells. Most sequences obtained from patient 1 (6 of 9) were μ transcripts, while in the other 2 patients α and γ transcripts dominated the repertoire.
The current study aimed to investigate the composition of the CD27− B cell subpopulation in patients with SLE. This subset has been described as far more heterogeneous in SLE patients than in normal healthy subjects, with a considerable percentage of transitional B cells (3) or CD27−,IgD− B cells comprising memory B cells (13–15). While the exact circumstances causing these disturbances in peripheral B cell homeostasis are still not fully understood, first observations based on followup analyses of peripheral B cells after treatment with cytotoxic or B cell–depleting drugs suggest that the expansion of transitional B cells in the peripheral blood is relative (3) and might be related to B cell lymphopenia and increased BAFF levels, observed after B cell depletion. In contrast, the expansion of CD27−,IgD− B cells in lupus patients was found to be absolute, and a correlation with disease activity was proposed (15). Although CD27−,IgD− B cells were recently characterized in lupus patients as memory B cells, the results of the current study suggest that this subset is not homogeneous and shows a rather variable composition. In individual patients, CD10+ transitional B cells contribute to this subset. In other patients, small CD27−,IgD− B cells lacking CD10 and also CD95 expression were noted, while in patients with disease flares, activated CD27−,IgD−,CD95+ B cells dominated the subset. The results obtained by immunoglobulin heavy-chain transcript analysis support the conclusion that CD27−,IgD−,CD95+ B cells underwent somatic hypermutation and a variable degree of immunoglobulin class switch, while CD27−,IgD−,CD95− B cells were more heterogeneous in regard to these processes, with only 2 of 5 undergoing class switch and 3 of 5 undergoing somatic hypermutation. In addition, the results of the phenotypic analysis were consistent with the molecular data reflecting the heterogeneity of CD27−,IgD−,CD95− B cells and documenting the activated and homogeneous phenotype of CD27−,IgD−,CD95+ B cells. It should be emphasized that CD27−,IgD−,CD95+ B cells showed an activated phenotype in all patients analyzed, while CD27−,IgD−,CD95− B cells exhibited a variable phenotype when different patients were compared.
In contrast to a previous study by Wei et al (15), we found the entire CD27−,IgD− B cell subset also enhanced in patients with infections. Interestingly, a patient with lobar pneumonia showed a significantly increased frequency and number of CD27−,IgD− B cells, suggesting that these cells might also have a physiologic role in certain immune responses. While enhanced B cell activation is probably one mechanism causing an expansion of CD27−,IgD− B cells in the peripheral blood of patients with SLE and infections, decreased turnover resulting in an accumulation of activated CD27−,IgD−,CD95+ memory B cells might also contribute, especially in lupus patients, since the expression of antiapoptotic factors is increased in these patients and correlates with CD95 expression (25, 26). In line with this hypothesis, an increased frequency and absolute number of CD27−,CD95+ B cells were observed in lupus patients compared with patients with bacterial infections and compared with normal individuals. Moreover, and in contrast to the entire CD27−,IgD− B cell subset, a significant correlation of this subset with disease activity was noted.
Intense interaction of CD40 and CD40 ligand or exposure to high levels of type I interferon can increase CD95 expression in SLE (27, 28). B cells expressing CD95 are usually eliminated by a Fas-dependent mechanism, as also observed in patients with human immunodeficiency virus (29). A strong B cell receptor signal can, however, prevent Fas-induced apoptosis (30, 31). Coligation of the complement receptor CD21 has been shown to provide such a signal (32). In this context, it is interesting to note the association of an increased absolute number of circulating antigen-experienced B cells lacking CD27 expression with complement factor consumption in SLE patients analyzed in the current study.
Since flow cytometry revealed that a median 34.6% (range 12.5–72.4) of all CD27−,IgD− B cells expressed CD95, it appeared reasonable to suggest that CD27−,IgD−,CD95+ B cells are the subset that is actually enhanced in the peripheral blood of patients with disease flares. This was confirmed in an analysis of an ethnically diverse cohort of lupus patients, with results documenting that the CD95+ subset of CD27−,IgD− B cells is associated with active disease, focusing our attention on this unique population of memory B cells.
In contrast to the entire CD27−,IgD− B cell subset, all CD27−,IgD−,CD95+ B cells are uniformly enlarged and express somatically mutated immunoglobulin heavy-chain transcripts with a mutational frequency comparable with that observed in CD27+,IgD− memory B cells. In addition, we observed a significant change in the chemokine receptor expression pattern compared with naive B cells and a loss of CD21, CD38, and CD44 surface expression that was even more pronounced than in CD27+ memory B cells. The phenotype (expression of CXCR3, but lack or decrease of CD27, CD44, CXCR4, and CXCR5 expression) suggests that CD27−,IgD− and especially CD27−,IgD−,CD95+ B cells may represent the product of an aberrant extrafollicular differentiation process of memory B cells. The phenotype of these cells suggests that they are activated but no longer dependent on T cell help and might be designated to migrate to inflamed tissues rather than the bone marrow. In this context, a CD21low B cell subset is generated in mice upon immunization with polysaccharide antigen (33). However, the immunoglobulin genes of CD27−,IgD−,CD95+ B cells are uniformly highly mutated with a pattern comparable with that observed in T cell–dependent germinal center reactions, making it unlikely that these cells underwent differentiation in a T cell–independent manner.
The CD27−,IgD−,CD95+ B cell subset we describe here might significantly overlap with a population of CD21low,CD19high B cells reported to be expanded in the peripheral blood of patients with SLE (34), although a variable percentage of these cells expressed CD27. The molecular mechanisms underlying the lack of CD27 expression remain to be elucidated. Besides down-regulation or shedding of CD27 as has been reported (35), a failure of B cells to up-regulate CD27 before leaving germinal centers has to be considered, since CD27− memory B cells were found and isolated from human tonsils (8). Even though we identified a correlation between the CD27−,IgD−,CD95+ B cell subset and disease activity, it remains to be shown whether this subset is enriched in self-reactive B cells.
Finally, we conclude that CD95 appears to be a useful marker to identify CD27− memory B cells with an activated phenotype that might serve as a biomarker for lupus activity and as a target of further investigations aiming to elucidate the pathogenic potential of these cells and the mechanisms involved in their generation as well as their regulation. Delineation of the mechanisms involved in the generation and regulation of this activated CD27−,IgD−,CD95+ memory B cell subset may broaden our understanding of the immunopathogenesis of SLE and also help to identify new, more specific, therapeutic targets.
Dr. Dörner 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.