CD22, a B cell–restricted transmembrane glycoprotein, regulates B cell antigen receptor signaling upon interaction with α2,6-linked sialic acid–bearing glycans, which act as ligands and are expressed on B and T cells. In this study, we investigated how the expression of CD22 ligand (CD22L) is modulated following lymphocyte activation or during the course of systemic lupus erythematosus (SLE).
The expression levels of CD22L on B and T cells in nonautoimmune mice were assessed by flow cytometric analysis using a soluble recombinant form of CD22, following stimulation with antigen or mitogen in vitro. In addition, the expression levels of CD22L on circulating lymphocytes were correlated with the progression of SLE in lupus-prone mice.
We observed a constitutive expression of CD22L on mature B cells, but not T cells, in nonautoimmune mice. However, CD22L levels were up-regulated selectively on T cells (but not B cells) stimulated with antigens in vitro, while their expression levels on B cells was up-modulated following polyclonal activation with lipopolysaccharide. Furthermore, expression of CD22L was increased on circulating B cells (and to a lesser extent on T cells) in parallel with progression of SLE in several different lupus-prone mice and in a cohort of (C57BL/6 × [NZB × C57BL/6.Yaa]F1) backcross mice.
The expression of CD22L is differentially regulated in B and T cells, and high expression of CD22L on circulating B cells is a marker for development of severe SLE, suggesting a role for CD22–CD22L interactions in SLE as well as in the regulation of humoral immunity.
CD22 is a B cell–specific transmembrane glycoprotein known to function as a coreceptor for the B cell receptor (BCR). It is variably expressed at different stages during B cell ontogeny, being present at low levels on pre-B cells, at higher levels on mature IgM+,IgD+ B cells, and absent on terminally differentiated plasma cells (1–3). After BCR crosslinking, CD22 is rapidly tyrosine-phosphorylated on its cytoplasmic tail, which results in recruitment and activation of the phosphotyrosine phosphatase, SHP-1 (4). The finding that CD22-deficient B cells exhibit an increased and prolonged Ca2+ signal in response to BCR crosslinking (1, 5–7) indicated that CD22 functions primarily as a negative regulator of BCR signaling. This seemed to be inconsistent with the reduced antibody response to thymus-independent type II antigens observed in CD22−/− mice (1, 7). However, the latter observation is likely to be related to the defective development in CD22−/− mice of marginal zone B cells (8), which play a major role in thymus-independent type II responses (9).
The extracellular domain of CD22 shares sequence similarity with the Siglec (sialic acid–binding Ig-like lectin) subgroup of the Ig superfamily, and recognizes α2,6-linked sialic acid–bearing glycans (10). It is implicated in the adhesion of B cells to various cell types, including lymphocytes, monocytes, erythrocytes, and endothelial cells (11, 12). In fact, it has been demonstrated that CD22 acts as a homing receptor to the bone marrow for mature recirculating B cells (13). CD22 itself and CD45 expressed on B and T cells apparently are major candidates for being CD22 ligands (CD22L) (14–16). More recent studies have demonstrated that the ligand-binding activity of CD22 is critical for its negative regulatory function of BCR signaling (17, 18), suggesting that CD22L expression and its interaction with CD22 play a role in the regulation of B cell activation and antibody response. However, it is still poorly understood how the expression of CD22L is controlled in B and T cells during normal immune responses and in autoantibody-mediated autoimmune disorders, such as systemic lupus erythematosus (SLE).
SLE is an autoimmune disorder characterized by the development of a variety of autoantibodies and lethal glomerulonephritis (19). The dysregulated expression of CD22 and CD22L could have significant consequences, possibly leading to excessive activation of B cells and autoantibody production, which is consistent with the finding that CD22-deficient mice exhibit increased production of IgG anti-DNA autoantibodies (5, 20). In addition, genome-wide mapping analysis for lupus susceptibility loci in lupus-prone NZW mice revealed that an interval containing the Cd22 gene on chromosome 7 is linked with autoantibody production and lupus-like glomerulonephritis (21–23). In the present study, using a soluble recombinant form of murine CD22, we analyzed the modulation of CD22L expression following in vitro activation of B and T cells in nonautoimmune mice and during the course of SLE in different lupus-prone mice.
MATERIALS AND METHODS
D011.10 anti-OVA323–339–I-Ad T cell receptor (TCR) transgenic (24) and Sp6 antidinitrophenyl (anti-DNP) IgM transgenic (25) BALB/c mice were obtained from Dr. A. Rolink (Basel, Switzerland). C57BL/6 (B6), DBA/2, BXSB, NZB, and NZW mice were purchased from Jackson Laboratories (Bar Harbor, ME). BXSB.H2d congenic mice, BXSB.Eα transgenic mice, and B6.Yaa mice have previously been described (26–29). BXSB.ll mice were provided by Dr. F. Dixon (Scripps Research Institute, La Jolla, CA) (30). The F1 hybrid and backcross mice used in this study were obtained by local breeding. Mice were bled from a retroorbital sinus puncture, and resulting sera were stored at −20°C until used.
Recombinant murine CD22–human IgG1 Fc (CD22-Fc) fusion protein.
The complementary DNA (cDNA) encoding the entire extracellular domain of murine CD22 fused to the Fc portion of human IgG1 has been described previously (15). CD22-Fc recombinant protein was produced in COS cells, and purified as described (15).
Flow cytometric analysis.
The expression of CD22L in splenic and peripheral blood cells was analyzed by flow cytometry using a FACSCalibur sorter (Becton Dickinson, Mountain View, CA). Cells were first incubated with CD22-Fc in the presence of anti–Fcγ receptor type II/III (2.4G2) monoclonal antibody (mAb), and then stained with fluorescein isothiocyanate (FITC)–labeled goat anti-human IgG conjugate (Caltag, South San Francisco, CA), followed by staining with phycoerythrin (PE)–labeled anti-B220 (RA3-6B2; PharMingen, San Diego, CA), anti-CD4 (H129.19; PharMingen), or anti-CD8 (53-6.7; PharMingen). The expression of class II major histocompatibility complex (MHC) molecules on B cells was assessed by staining with FITC-labeled anti–I-A (MKD6) and PE-labeled anti-B220.
Spleen cell culture.
B cells were purified from the spleen by adherence of macrophages to plastic culture plates for 1 hour at 37°C and subsequent treatment with anti–Thy-1.2 (AT-83) mAb in the presence of rabbit complement (Cedarlane, Hornby, Ontario, Canada). CD4+ T cells were enriched by treatment with anti-CD8 (H35-17.2) and anti-B220 (RA3-3A1) mAb in the presence of rabbit complement. The purity of B and T cells, as documented by cytofluorometric analysis, was >95%. For stimulation of ovalbumin (OVA)–specific T cells and DNP-specific B cells, 2 × 105 T cells purified from the spleens of BALB/c mice bearing the D011.10 anti-OVA323–339–I-Ad TCR transgene were cultured with or without 2 × 105 B cells purified from the spleens of BALB/c mice expressing the Sp6 IgM anti-DNP transgene, in the presence or absence of DNP-conjugated OVA (50 μg/ml) in a total volume of 200 μl or 1 ml Dulbecco's modified Eagle's medium (DMEM) containing 5% fetal calf serum (FCS).
T cell proliferative responses were assessed by the measurement of 3H-thymidine incorporation during a 24-hour culture, and IgM secretion was determined by enzyme-linked immunosorbent assay (ELISA) in supernatants collected after a 5-day culture. For lipopolysaccharide (LPS) or cytokine treatment experiments, 2 × 106 splenic cells or purified B cells from B6 mice were incubated in 1 ml of DMEM containing 10% FCS, in the presence of LPS (25 μg/ml) or various concentrations of different cytokines. The expression of CD22L on splenic B and T cells, as assessed by the binding of CD22-Fc, was determined by flow cytometric analysis during a 3-day culture.
Serologic and histologic analysis.
IgM concentrations in culture supernatants and serum levels of IgG anti-DNA autoantibodies were determined by ELISA, as previously described (31). Samples of all major organs were obtained at autopsy when mice were moribund, and histologic sections were stained with either the periodic acid–Schiff reagent or hematoxylin and eosin. Glomerulonephritis was scored on a 0–4-point scale based on the intensity and extent of histopathologic changes, as described previously (29). Severe glomerulonephritis (grades 3 and 4) was considered a significant contributor to clinical disease or death.
A cohort of (B6 × [NZB × B6.Yaa]F1) backcross male mice bearing the Yaa mutation (n = 70) were grouped into 2 discrete sets corresponding to the expression level of CD22L on circulating B cells, in order to perform extreme-phenotype analysis (32). Thus, only mice categorized as having either a low or high expression level of CD22L on circulating B cells at 4 months of age were included in the analysis, and the other mice, categorized as having intermediate expression, were excluded from the analysis. The mean (± SD) fluorescence intensity of CD22L on circulating B cells from 4-month-old B6.Yaa male mice (n = 15) was 5.8 ± 0.5. A low CD22L level (CD22Llow) was defined as a mean ± SD value of <7.3 ± 3, and an abnormally high level (CD22Lhigh) was defined as a mean ± SD value of >8.8 ± 6. Statistical analysis for anti-DNA autoantibodies and survival rates was performed by Wilcoxon's 2-sample test. P values less than 0.05 were considered significant.
Spontaneous CD22L expression on B cells, but not T cells, in nonautoimmune mice.
The expression levels of CD22L on B and T cells in splenic and peripheral blood obtained from nonautoimmune B6 mice were first assessed by flow cytometry, using CD22-Fc fusion protein. Peripheral mature B cells exhibited significant binding of CD22-Fc, while binding of CD22-Fc to mature T cells was hardly detectable (Figure 1). Essentially identical results were obtained with 2 other nonautoimmune strains of mice, BALB/c and DBA/2 (results not shown). These findings indicate that CD22L is expressed constitutively on mature peripheral B cells, but not T cells, in nonautoimmune mice.
Increased expression of CD22L on in vitro antigen-stimulated T cells, but not B cells.
To investigate the modulation of CD22L expression following lymphocyte activation by antigens, OVA-specific T cells expressing the D01.10 anti-OVA323–329–I-Ad TCR transgene were isolated by depleting B cells (but not macrophages and dendritic cells) from the spleens of transgenic BALB/c (H2d) mice, and stimulated with DNP-conjugated OVA (50 μg/ml) in vitro. The activation of transgenic T cells in the presence of DNP-OVA was documented by a marked increase in 3H-thymidine uptake 24 hours after incubation (mean ± SD 46,212 ± 3,742 counts per minute), as compared with that in unstimulated transgenic T cells (4,649 ± 924 cpm). Although no detectable CD22-Fc bindings were observed at 24 hours, ∼30% of CD4+ transgenic T cells exhibited a significantly increased binding of CD22-Fc at 72 hours (Figure 2A). Notably, T cells from nontransgenic BALB/c mice cultured with DNP-OVA failed to show any binding of CD22-Fc.
To determine the modulation of CD22L expression following activation of B cells, DNP-specific B cells purified from BALB/c mice expressing the Sp6 anti-DNP IgM transgene were cocultured with OVA-specific transgenic T cells in the presence of DNP-OVA in vitro. In the presence of DNP-specific B cells, a stronger activation of OVA-specific T cells was induced (the mean ± SD 3H-thymidine uptake at 24 hours in the presence of DNP-OVA was 174,161 ± 18,648 cpm and in the absence of DNP-OVA was 5,621 ± 775 cpm). At 48 and 72 hours, respectively, ∼50% and essentially all of the CD4+ transgenic T cells demonstrated up-regulated expression of CD22L (Figure 2B). In contrast, DNP-specific B cells did not exhibit increased binding to CD22-Fc during culture (Figure 2B). Notably, these B cells were efficiently activated, as judged by increased expression of class II MHC (results not shown) and increased secretion of IgM after a 5-day culture (the mean ± SD increase in the presence of DNP-OVA was 920 ± 105 ng/ml and in the absence of DNP-OVA was 48 ± 15 ng/ml). However, the level of IgM in culture supernatants was hardly increased when DNP-specific B cells were incubated with DNP-OVA in the absence of OVA-specific T cells (mean ± SD increase 49 ± 17 ng/ml).
Increased expression of CD22L on B cells stimulated with LPS in vitro.
It was previously shown that B cells stimulated with LPS exhibited enhanced adhesion to BHK cells transfected with murine CD22 cDNA (33). To confirm whether LPS stimulation indeed enhances expression of CD22L, spleen cells obtained from B6 mice were stimulated with LPS. The binding of CD22-Fc was significantly increased on B220+ B cells at 24 hours and peaked 48 hours after LPS stimulation (Figure 3). Enhanced expression of CD22L in response to LPS was similarly observed on purified B cells (results not shown), which provides evidence against a role of cytokines secreted by LPS-activated macrophages. This was further confirmed by the finding that expression levels of CD22L were unchanged in the presence of either interleukin-1 (IL-1), IL-6, or tumor necrosis factor at any dose tested (results not shown).
Increased CD22L expression on B and T cells in lupus-prone mice.
We next evaluated the modulation of CD22L expression on B and T cells during the development of SLE in lupus-prone mice. For this purpose, the CD22L expression levels on peripheral blood lymphocytes, as assessed by the binding of CD22-Fc in flow cytometry, was monitored in lupus-prone BXSB mice. Because only male BXSB mice spontaneously develop severe SLE due to the presence of the Yaa (Y-linked autoimmune acceleration) mutation (34), expression of CD22L in male BXSB mice was compared with that in their female counterparts lacking the Yaa mutation. Significantly increased CD22-Fc binding to circulating B cells was already observed at 2 months of age in 2 of 10 BXSB male mice bearing the Yaa mutation, while levels of CD22-Fc binding in BXSB females were not different from those observed in nonautoimmune B6 mice (Figure 4).
Both the incidence of increased CD22L levels on B cells of BXSB male mice and CD22L expression levels themselves increased with age. At 4 months of age, elevated levels of CD22L on B cells were observed in all BXSB male mice (Figures 4 and 5), 50% of which died of lupus-like glomerulonephritis by the age of 6 months (35). In addition to increased expression on B cells, the expression level of CD22L on circulating CD4+ and CD8+ T cells was also up-regulated in 4-month-old BXSB male mice (Figure 5), although the incidence was lower than that on B cells. In contrast, the levels of CD22L on B and T cells in 6-month-old BXSB female mice remained low, and were essentially identical to those seen in 6-month-old B6 mice. Notably, similar results were obtained with B and T cells from the spleens of BXSB male and female mice (results not shown).
To confirm that the observed age-dependent increase in CD22L expression on circulating lymphocytes from lupus-prone BXSB male mice was indeed due to the development and progression of SLE (and was not uniquely associated with expression of the Yaa gene), we analyzed expression of CD22L on B cells obtained from 3 different substrains of BXSB male mice (BXSB.H2d, BXSB.Eα, and BXSB.ll). These 3 substrains of BXSB mice carry the Yaa mutation but fail to develop severe SLE during the first year of life, because of the presence of the H2d haplotype (26), the transgene encoding an I-E α chain (27), and the ll (long-lived) mutation (30), respectively.
Despite the presence of the Yaa mutation, the expression level of CD22L on peripheral blood B cells from these mice was markedly limited and comparable to that seen in BXSB females (Figure 4). In addition, analysis of 2 other lupus-prone hybrid mice ([NZB × NZW]F1 and [NZB × BXSB]F1) confirmed an age-dependent increase in CD22L expression on B cells in parallel to the development of SLE (Figure 4). Furthermore, comparative analysis of 4-month-old (NZB × BXSB)F1 male and female mice revealed a much higher expression of CD22L on B cells in the male mice that developed an accelerated form of SLE due to the presence of the Yaa mutation (50% mortality at 5.5 months), as compared with the female mice (50% mortality at 9 months) (36).
High CD22L expression on circulating B cells as an active disease marker for development of SLE in (B6 × [NZB × B6.Yaa]F1) backcross male mice.
The age-dependent increase in CD22L expression on peripheral blood lymphocytes in different lupus-prone mice raised the possibility that the increased level of CD22L on circulating B cells could be an active disease marker for the development of SLE. To test this possibility, we investigated whether CD22L expression on B cells from (B6 × [NZB × B6.Yaa]F1) backcross male mice bearing the Yaa mutation correlated with development of a lupus-like autoimmune syndrome.
When a cohort of 70 backcross male mice were examined for CD22L expression on circulating B cells at 4 months of age, levels of CD22L on B cells were found to be highly heterogeneous. From this cohort, 22 CD22Lhigh and 29 CD22Llow mice were selected for further analysis (Figure 6A). At this age, the majority of the CD22Lhigh backcross mice already had significantly elevated titers of IgG anti-DNA autoantibodies (Figure 6B), with the mean values being ∼10-fold higher than those of the CD22Llow backcross mice (P < 0.0001) (Table 1). In contrast, IgG anti-DNA levels in the CD22Llow backcross mice were comparable to those obtained in age-matched B6.Yaa male mice, which fail to develop SLE (29). Notably, IgG anti-DNA autoantibody levels in 7 of 22 CD22Lhigh backcross mice were still within the range of control values obtained in B6.Yaa males at 4 months of age. However, by 8 months of age, all of the CD22Lhigh but only 5 of 29 CD22Llow backcross mice had significantly elevated titers of IgG anti-DNA autoantibodies (Figure 6B). The mean levels of IgG anti-DNA autoantibodies in CD22Lhigh backcross mice were 14 times higher than those in CD22Llow backcross mice (P < 0.0001) (Table 1).
Table 1. IgG anti-DNA autoantibody levels and glomerulonephritis in CD22Lhigh and CD22Llow (B6 × [NZB × B6.Yaa]F1) backcross male mice*
Based on mean fluorescence intensities of CD22 ligand (CD22L) on circulating B cells from 4-month-old B6.Yaa male mice (n = 15), (B6 × [NZB × B6.Yaa]F1) backcross male mice exhibiting a high (>8.8) and low (<7.3) CD22L level were categorized as CD22Lhigh (n = 22) and CD22Llow (n = 29), respectively. Mean (± SD) fluorescence intensities of CD22L for different groups of mice are indicated. Serum levels of IgG anti-DNA autoantibodies in CD22Lhigh and CD22Llow backcross male mice were determined by enzyme-linked immunosorbent assay at 4 and 8 months of age. As a control, serum levels of IgG anti-DNA in 4-month-old B6.Yaa male mice are shown. Results are expressed in titration units (units/ml) in reference to a standard curve obtained from a pool serum of 3–4-month-old MLR-lpr/lpr mice. NT = not tested.
Lethal glomerulonephritis (grades 3 and 4).
16.4 ± 8.1
70 ± 57
139 ± 120
6.5 ± 0.3
8 ± 4
10 ± 9
5.8 ± 0.5
9 ± 4
When the development of a lethal form of lupus-like glomerulonephritis was followed up until mice were 18 months of age, CD22Lhigh backcross mice had an earlier onset and higher incidence of lethal glomerulonephritis than did CD22Llow backcross mice (P < 0.0001) (Figure 7 and Table 1). In fact, 55% (12 of 22) of the CD22Lhigh mice died of glomerulonephritis by 11 months, and 96% (21 of 22) died within 18 months. In contrast, none of the CD22Llow mice died of glomerulonephritis during the first year of life, and only 35% (10 of 29) developed lethal glomerulonephritis within 18 months of age.
In the present study, we assessed the modulation of CD22L on B and T cells following their activation and during the course of lupus-like autoimmune disease. Our results demonstrate that although CD22L was constitutively expressed on B cells, its expression was selectively induced on T cells, but not B cells, after antigenic stimulation in vitro, while the expression level of CD22L on B cells can be further up-regulated following polyclonal activation with LPS. In addition, we observed that expression of CD22L was spontaneously increased on circulating B cells (and to a lesser extent on T cells) in parallel to the progression of SLE in lupus-prone mice. Moreover, the analysis of (B6 × [NZB × B6.Yaa]F1) backcross mice, of which only a fraction spontaneously developed SLE, led to the conclusion that high expression of CD22L on circulating B cells is an active disease marker for SLE. Because CD22 is expressed exclusively on B cells, an induced expression of CD22L on activated T cells suggests a significant role for CD22–CD22L interactions in the regulation of B cell responses and also in the development of SLE and other autoantibody-mediated autoimmune diseases.
Our demonstration of the induction of CD22L expression on antigenically activated T cells may be relevant for the development and regulation of immune responses, because CD22L expression may determine where and to what extent CD22 is ligated, and thus sets signal transduction thresholds for the BCR. Antigen-specific T cells are initially activated by professional antigen-presenting cells, such as dendritic cells, but not by naive B cells. The induced expression of CD22L on activated T cells could facilitate their subsequent interaction with surrounding naive B cells in peripheral lymphoid organs. As proposed previously (4, 37), this interaction could then lead to sequestration of CD22 away from BCR, thereby promoting the activation of antigen-specific B cells through down-modulation of the negative regulatory function of CD22 for the BCR signaling (17, 18). Thus, the induced expression of CD22L on activated T cells may be important in the initial phase of T cell–B cell interaction after antigen challenge.
It is significant that the level of CD22L is progressively increased, in an age-dependent manner, on circulating B and T cells from different lupus-prone mice. The up-regulated expression of CD22L likely reflects the spontaneous activation of B and T cells during the course of autoimmune responses. However, it should be mentioned that the increase in CD22L appears to occur in the entire population of peripheral T cells, independently of CD4 and CD8 phenotypes, despite the fact that the development of SLE is dependent on the activation of CD4+, but not CD8+ T cells (19). This suggests that increased expression of CD22L on the majority of B and T cells in lupus-prone mice may reflect their nonspecific activation in peripheral lymphoid organs. It is noteworthy that polyclonal activation of B and T cells is a common feature of lupus-like autoimmune disease (38–40).
The increased expression of CD22L on B cells in lupus-prone mice may play a more direct role in autoantibody responses. It has been well established that CD22 becomes rapidly tyrosine phosphorylated on its cytoplasmic tail upon crosslinking of BCR, and recruits SHP-1, thereby negatively regulating BCR signaling (4). More recent studies revealed that the ligand-binding activity of CD22 is critical for its negative regulatory function (17, 18). Moreover, only a small fraction (<5%) of the total CD22 associates with BCR (41, 42), and CD22 itself appears to be one of the major CD22 ligands expressed on B cells (15). Thus, highly α2,6-sialylated CD22 on B cells may interact more efficiently as a ligand for CD22, thereby preventing it from associating with BCR on the same B cells. Consequently, BCR-mediated signaling and hence B cell responses to antigen could be potentiated. Therefore, one can speculate that the lack of up-regulation of CD22L on antigen-stimulated B cells may be one of the regulatory mechanisms to prevent excessive activation of B cells, while up-regulation of CD22L on B cells could contribute to overstimulation of B cells, favoring excessive production of autoantibodies in lupus-prone mice.
In addition, a recent study demonstrated that activation of B cells by antigen displayed on the surface of a target cell is depressed if the latter coexpresses CD22L (43). Those authors thus proposed that the CD22–CD22L interaction between autoreactive B cells and target cells could be a mechanism to prevent triggering activation of low-affinity autoreactive B cells upon their contact with autoantigens displayed at high density on a cell surface. In support of this hypothesis, the spontaneous development of lupus-like autoimmune syndrome has been reported in mice deficient in α-mannosidase II, which catalyzes the first step in the biosynthesis of complex asparagine-linked oligosaccharide chains (44). Increased expression of CD22L on autoreactive B cells in lupus-prone mice could compete with expression of CD22L on autologous cells for the interaction of CD22 on B cells, thereby promoting induction of autoimmune responses against membrane antigens, such as anti-erythrocytes, anti-platelets, and anti-lymphocytes, characteristically seen in SLE (19).
The analysis of 3 different lupus-prone mice, BXSB, (NZB × NZW)F1, and (NZB × BXSB)F1, together with (B6 × [NZB × B6.Yaa]F1) backcross mice, has clearly demonstrated that the spontaneously increased expression of CD22L on B cells is closely associated with the development and progression of SLE. Most significantly, the incidence of SLE was much higher in a group of backcross mice exhibiting high levels of CD22L on B cells at 4 months of age than in their littermates exhibiting low levels of CD22L. It should be stressed that a fraction of these backcross mice categorized as CD22Lhigh did not have significant titers of IgG anti-DNA autoantibodies at 4 months of age, when the expression of CD22L on B cells was first analyzed. These mice, however, developed high titers of anti-DNA autoantibodies later, and finally died of lupus-like glomerulonephritis. This suggests that increased expression of CD22L could be used as a predictive marker for the subsequent development of severe SLE.
It would be of interest to determine whether the expression of CD22L on B and T cells is similarly enhanced in other autoimmune diseases mediated not only by autoantibodies but also by CD4 and/or CD8 T cells, such as rheumatoid arthritis, multiple sclerosis, and autoimmune diabetes. Clearly, better understanding of CD22–CD22L interactions in the regulation of autoantibody production would help establish new strategies for the development of therapies in SLE and other autoantibody-mediated autoimmune diseases.
We thank Dr. Edward Clark for critical reading of the manuscript, and Ms Geneviève Leyvraz and Mr. Giuseppe Celetta for their excellent technical help.