Homeostatic expansion of autoreactive immunoglobulin-secreting cells in the Rag2 mouse model of Omenn syndrome

OS patients are characterized by oligoclonal T cells, although they are virtually devoid of circulating B cells (Fig. 1 A and Table S1). However, high IgE and residual IgG and/or IgM serum levels indicate that Ig-producing plasma cells are actively present. Histological analysis of LN biopsies from two OS patients revealed a severe alteration of the microscopic architecture of the organ characterized by expansion of the paracortical area, along with severe depletion of the lymphoid cell population and lack of B cell follicles (Fig. 1 B, top left). Late B cell differentiation stages, including plasma cells and a subset of GC B cells, which are committed to plasma cell differentiation (Angelin-Duclos et al., 2000), are characterized by the expression of both Blimp1 and CD138, or Blimp1 alone, respectively. Accordingly, LN biopsies from healthy donors revealed the presence of a relatively small number of Blimp1⁺CD138⁻ late GC B cells, as well as Blimp1⁺CD138⁺ terminally differentiated plasma cells (Fig. 1 B, top right and inset). We detected Blimp1⁺CD138⁺ double-positive cells in both LN biopsies from OS patients, albeit at different levels (Fig. 1 B, bottom left and right). However, in contrast to LNs from healthy donors, Blimp1⁺CD138⁻ cells were only occasionally detected (unpublished data). Thus, in LNs from OS patients with undetectable circulating B cells, the large majority of Blimp1⁺ cells represent an advanced stage of plasma cell differentiation.

We analyzed BM samples from four OS patients bearing RAG mutations for B cell subset composition. As expected from the recombinase defect, a severe block in B cell differentiation was evident, and mature B cells were barely detectable (Fig. 1 C). To better characterize this developmental arrest, we used the differential expression of marker molecules to define various stages of B cell differentiation (van Zelm et al., 2005). In the BM from age-matched healthy children, ∼18% of the precursor B cell compartment consisted of cytoplasmic (Cy) Igμ⁻ cells, whereas in BM from all analyzed OS patients, ∼92% (82–98%) of the precursor B cell compartment had this phenotype (Fig. 1 D). Interestingly, some CyIgμ⁺ pre–B-II cells and surface membrane (Sm) IgM⁺ immature B cells could be detected in patients OS4 and OS5 (Fig. 1 D). The mature B cell population (CD19⁺ SmIgM⁺ SmIgD⁺) was dramatically reduced in patients (<0.6% vs. 24% in healthy controls; unpublished data). However, ELISpot analysis revealed the presence of ISCs in the BM of all OS patients, producing either IgM or IgG/A (Fig. 1 E).

Given the obvious limitation in performing experiments with tissue samples from OS patients, we addressed the role of B cells in the pathogenesis of Rag2^R229Q mice, which recapitulate the main immune system alterations observed in OS patients (Marrella et al., 2007). Indeed, similarly to the majority of patients, Rag2^R229Q mice showed absent or dramatically reduced numbers of circulating B cells (Fig. 2 A), which is indicative of arrested B cell development. In the BM, Rag2^R229Q mice had a significantly reduced absolute number and percentage of B220⁺ B cells (P < 0.0001; Fig. 2 B and Fig. S1 A). Analysis of early B cell subsets showed a severe block in B cell differentiation at the pro–B cell stage, as indicated by the selective increase of early pro–B cells (B220⁺CD43⁺IgM⁻HSA^lowBP-1⁻; Hardy et al., 1991; P = 0.03; Fig. S1 A and not depicted) and the severely reduced pre–B cells (B220⁺CD43⁻IgM⁻), as well as IgM⁺IgD⁻ immature B cells (P < 0.0001; Fig. 2 B and Fig. S1 A). As a consequence, the absolute number and relative proportion of splenic B220⁺ B cells were profoundly reduced in Rag2^R229Q mice (P < 0.0001; Fig. 2, C and D, and Fig. S1, B and C), with the complete absence of conventional transitional (T1 and T2) and mature B cells (Fig. 2 C and Fig. S1 B). The peripheral B cell pool was mainly composed of IgM⁺IgD⁻ B cells, half of which expressed B220 (43.8 ± 12.3%; Fig. 2 C, bottom), and were negative for the AA4.1 and CD138 markers (not depicted). Notably, the expression of several cell activation markers (CD40, CD69, CD86, and MHC II) was augmented in B220⁺IgM⁺ cells from Rag2^R229Q mice when compared with the same subset from WT littermates (Fig. 2 C). Among IgM⁺ B cells, the mature naive splenic B cell compartment is made up of two populations designed as follicular (FO) B cells (B220⁺CD23^highCD21^low) and marginal zone (MZ) B cells (B220⁺CD23^lowCD21^high). Rag2^R229Q mice had barely detectable IgM⁺ MZ B cells and lacked mature IgD⁺ FO B cells (Fig. 2 D and Fig. S1 C), consistent with the almost complete disruption of both MZ and FO organization (not depicted). In contrast, a CD21⁻CD23⁻B220⁺ lymphocyte population was overrepresented in the mutant mice in comparison with WT littermates (P = 0.0037; Fig. S1 C). The CD21⁻CD23⁻ cells were CD43^−/lowCD5⁻Mac-1⁻ and negative for sIg (not depicted), thus excluding the conventional B-1 lineage. These findings collectively suggest that in Rag2^R229Q mice some B cells may overcome the developmental block and colonize the periphery.

The presence of peripheral B cells in Rag2^R229Q mice prompted us to analyze the IgH repertoire. To this end, we cloned and sequenced the cDNAs encoding the Ig heavy chain (IgH) variable domain of IgM from bulk sorted B220⁺ splenocytes (Table S2). The analysis of these sequences showed one dominant set of clonally related cells (CRCs) in each of the mutant mice, whereas we found only four small CRC clusters in the WT animal (Fig. S2 A). We additionally performed single-cell PCRs on individually sorted B220⁺CD19⁺ splenocytes of WT and Rag2^R229Q mice. Again, although we could not find CRCs in the normal mice, two out of three Rag2^R229Q animals showed CRC clusters (Fig. S2 A). The absence of CRCs from Rag2^R229Q B220⁺CD19⁺ B cells in the third sorting experiment is likely caused by the small number of sequences analyzed. Although we could not definitely demonstrate that Rag2^R229Q mice display a restricted Igh-V repertoire (Fig. S2 B), the results from single-cell PCR show expansion of a few CRCs in Rag2^R229Q mice.

With high variability, the production of serum IgG₁, IgG₂, and IgM was preserved in Rag2^R229Q mice (aged 1–9 mo), whereas IgG₃ and IgA were reduced (P < 0.0001; Fig. 3 A). Consistent with the OS phenotype, IgE levels were significantly higher in mutant mice (P < 0.0001). Thus, Rag2^R229Q mice are able to produce serum Ig of all isotypes, in spite of the apparent lack of mature B cells. To clarify this apparent discrepancy, we enumerated ISCs in cell suspensions derived from BM, spleen, and LNs by ELISpot. Using this analysis, we found a greater proportion of ISCs in secondary lymphoid organs of all Rag2^R229Q mice assayed compared with WT mice (Fig. S3 A). More pronounced differences were scored when ISC counts were normalized to B cell numbers. Indeed, mutant mice displayed 6–165-fold increases in IgM-, IgG-, IgA-, and IgE-secreting cells within the B cell compartment of spleen and LNs, compared with WT littermates (IgM, IgG, IgA, P = 0.008; IgE, P = 0.028; Fig. 3 B). In contrast, in the Rag2^R229Q BM, ISCs were retrieved to a lesser extent and with higher variability (Fig. 3 B). This observation correlated with the reduced expression of CXCL12 in Rag2^R229Q BM, the chemokine required for plasma cells homing and survival in BM niches (Tokoyoda et al., 2004; Fig. S3 B). These data revealed a larger expansion of the ISC population in Rag2^R229Q mice, mainly within the secondary lymphoid organs.

Plasmacytic differentiation is sustained by a program of gene expression, of which Blimp1 is considered to be the primary trigger (Martins and Calame, 2008). By quantitative PCR, we found that expression of Blimp1 is increased by threefold in splenic B220⁺ B cells isolated from mutant mice as compared with controls (P = 0.029; Fig. 3 C). Increased Blimp1 expression also resulted in the induction of Xbp1, which is required for antibody production and secretion (Reimold et al., 2001; P = 0.029; Fig. 3 C). Differentiation to ISC with Blimp1 up-regulation is associated with transcriptional repression of Bcl6 and c-Myc, characteristic of GC B cells (Lin et al., 1997; Shaffer et al., 2000). Accordingly, these transcripts were significantly reduced in Rag2^R229Q B220⁺ cells (P = 0.029; Fig. 3 C). Therefore, although partially limited by the heterogeneity of the sorted B220⁺ cell pool, these findings suggest that splenic B cells from Rag2^R229Q mice might have progressed further through a plasmacytic differentiation pathway than those of WT mice. Accordingly, in mutant mice, Xbp1 and Blimp1 expression were further increased in B220⁻IgM⁺ B cells with respect to B220⁺IgM⁺ B cells, indicative of an advanced stage of plasma cell differentiation (unpublished data). Furthermore, Rag2^R229Q B220⁺ cells retained increased expression of Aid (P = 0.029; Fig. 3 C), which is essential for somatic hypermutation and isotype switching/class switch recombination (CSR).

B cell activation and plasmacytic differentiation might be cooperatively sustained by T cell activation. In Rag2^R229Q mice, CD4⁺ T cells were skewed toward T_H-1 polarization, as indicated by increased production of IFN-γ, but not of IL-4 and IL-10 cytokines upon PMA/ionomycin stimulation (Fig. 4 A). Interestingly, CD4⁺ T cells from Rag2^R229Q mice had a substantially increased percentage of IL-17–producing cells (Fig. 4 A). Accordingly, Rag2^R229Q CD4⁺ T cells displayed significantly increased mRNA levels for the Th1-specific T-bet transcription factor and reduced mRNA levels for GATA-3, which directs Th2 polarization (P = 0.03; Fig. 4 D; Szabo et al., 2002). Significantly higher mRNA levels for the Th17-specific RORγt transcript were also detected. Thus, Th1 and Th17 cells could be implicated in the inflammatory condition characteristic of Rag2^R229Q mice.

Th17 cells secrete large amounts of IL-21, which then acts to amplify the Th17 response in an autocrine fashion (Korn et al., 2007; Zhou et al., 2007). Analysis of serum cytokines by ELISA revealed that mutant mice had elevated levels of IL-21 (P = 0.0113) and IFN-γ (P = 0.0007; Fig. 4 B). Because production of both IL-17 and IL-21 cytokines has been associated to FO helper T cells (T_FH cells; Bauquet et al., 2009; Chtanova et al., 2004), in addition to T_H-17 cells, we analyzed the CD4⁺ T cell subset for the expression of CXCR5 and ICOS, crucially involved in T_FH cells homing and differentiation (Schaerli et al., 2000; Mak et al., 2003). An increased proportion of ICOS⁺ cells was observed in Rag2^R229Q CD4⁺ splenocytes (41.6% vs. 7.7%; n = 6; P = 0.008), and more CD4⁺ ICOS^hi T cells expressed CXCR5 in Rag2^R229Q mice (7.7% vs. 1.5%; n = 6; P = 0.008; Fig. 4 C). Rag2^R229Q CD4⁺ CXCR5⁺ ICOS^hi T cells expressed WT levels of CD40L, but increased levels of programmed death-1 (PD-1; Fig. S4), both of which are co-stimulatory molecules necessary for T_FH cell generation and function (Vinuesa et al., 2005). The mRNA encoding the transcription factor Bcl6, a determinant of the T_FH cell fate (Nurieva et al., 2009), was also significantly increased in Rag2^R229Q CD4⁺ splenocytes (P = 0.03; Fig. 4 D). Overall, these data define subsets of T helper cells with the potential to activate B cells and promote their differentiation into ISC in Rag2^R229Q mice. To better clarify the T cell contribution to B cell phenotype in Rag2^R229Q mice, we treated mutant mice with the anti-CD4 mAb (GK1.5) for 7 wk. This treatment resulted in a complete depletion of circulating, splenic, and BM CD4⁺ T cells (Fig. 5 A). Concomitant to the depletion of CD4⁺ T cells, we observed a significant reduction of the frequency of IgG and IgA but not IgM-secreting cells in the spleen and BM of treated mice (Fig. 5 B), correlating with the serum Ig levels (Fig. 5 C). Anti-CD4 treatment determined a complete normalization of the IgE serum level in Rag2^R229Q mice (Fig. 5 C) and a substantial decrease in the concentrations of CSR-driving cytokines, IFN-γ and IL-21 (Fig. 5 D). Thus, these findings highlight the contribution of CD4⁺ T cells in sustaining B cell activation and plasma cell differentiation in Rag2^R229Q mice.

We next investigated whether Rag2^R229Q mice could mount a T cell–dependent immune response in vivo. Mutant and WT mice (10 wk of age) were immunized into the footpad with OVA in CFA and boosted with recall antigen in immunofluorescence (IFA) on day 21. At regular intervals, sera were analyzed simultaneously for total and OVA-specific antibodies. As expected, in normal mice immunization resulted in a rapid and significant increase of total IgG serum levels (P = 0.03; Fig. 6 A and not depicted), whereas no IgG elevation was observed in Rag2^R229Q mice (Fig. 6 A). Moreover, in WT animals OVA-specific IgG levels were detectable immediately after the boost and sustained after 48 d (Fig. 6 B). In contrast, we could not detect a secondary anti-OVA IgG response in the sera from Rag2^R229Q mice (Fig. 6 B). These results were matched by the analysis of the frequency of antigen-specific ISC as measured by ELISpot. Unimmunized Rag2^R229Q mice made significantly more IgG-secreting cells than WT mice (P = 0.008; Fig. 6 C). However, after boosting, an increased number of IgG plasma cells were detectable exclusively in the spleens of WT mice (P = 0.029) from days 26 to 33, and returned to nearly baseline levels by day 48, which is consistent with the IgG serum levels (Fig. 6 C). A similar trend was observed in the response of WT mice using the ELISpot assay (Fig. 6 D). No OVA-specific IgG-secreting cells could be detected in the Rag2^R229Q mice at any of the time points assayed (Fig. 6 D), indicating that the observed absence in IgG production resulted from an impaired generation of specific antibody-secreting cells.

It is well established that Toll-like receptor (TLR)–mediated activation of mature B cells is required for eliciting humoral immune responses (Pasare and Medzhitov, 2005). Mature murine B cells can be stimulated in vitro by LPS or CpG (the TLR4 and TLR9 ligand, respectively) to proliferate and secrete antibodies (Watson, 1979; Krieg et al., 1995). We evaluated the expression of TLR4 and TLR9 by real-time PCR in purified Rag2^R229Q B220⁺ cells, and found a reduction in comparison to WT cells (Fig. 7 A). Moreover, Rag2^R229Q B220⁺ cells did not cycle in response to LPS or CpG and IL-4 co-stimulation (Fig. 7 B). Because proper interpretation of these results could be hindered by the heterogeneity of the B220⁺ cell subset, we also evaluated the Ig production. Analysis of Ig in the supernatants at day 4 of culture indicated that mutant B cells were able to switch and secrete Ig at a rate comparable to control B cells, and particularly upon CpG stimulation (Fig. 7 C). To confirm in vivo the T cell–independent humoral responsiveness of Rag2^R229Q B cells, we immunized mutant mice i.p. with a mixture of pneumococcal antigens (Pneumovax23-P23) and monitored total and specific IgM serum antibodies at different time points after immunization (from day 10 to 60). The increase in total and specific serum IgM was lower in Rag2^R229Q with respect to WT littermates (Fig. 8, A and B). However, in the Rag2^R229Q B220⁺ population we detected a higher frequency of ISC secreting P23-specific IgM which decayed rapidly (Fig. S5, C and D). The possible homing of specific ISC in the BM was excluded by the failure to detect P23-specific ISC in the BM of mutant mice (unpublished data). The apparent discrepancy between ELISpot results and serum titers could be explained by considering the markedly reduced size of the splenic B cell compartment in mutant mice. Low absolute numbers of P23-specific IgM ISC would be unable to determine WT antibody titers. Moreover, Rag2^R229Q displayed a more rapid decrease in specific serum antibodies reaching the pre-boost level at day 20, indicating a selective impairment in the generation of long-term humoral responses (Fig. 8, A and B). Collectively, our results show that Rag2^R229Q mice are able to mount short-term specific humoral immune responses, which do not require T cell help for their generation.

A proportion of Rag2^R229Q mice spontaneously develop severe alopecia, skin erythrodermia, and wasting syndrome caused by colitis (Marrella et al., 2007). In addition, a minority of them presented with differing degrees of renal damage, resulting in proteinuria. A marked infiltration of oligoclonal activated T lymphocytes was consistently evident in different tissues of mutant mice and correlated with abnormal CXCL10 serum level (P = 0.002; Fig. S6 A). Further characterization of these infiltrates revealed that they were also composed of B220⁺ Ig⁺ B cells (Fig. 9 A; Fig. S6, B and C). The abundance of B220⁺ and IgM⁺ cell infiltrates was lower with respect to CD3⁺ T cells (P = 0.048 and P = 0.0098, respectively; Fig. S6 C), which may reflect the absence of circulating B cells, as well as the significantly lower representation of B lymphocytes in primary and secondary lymphoid organs of Rag2^R229Q mice. Nonetheless, a direct role of B cells in the disease pathogenesis was suggested by the severe glomerulus infiltration of IgM⁺ and IgG⁺ cells (Fig. 9 A, top) in mice with high proteinuria index, whereas no tissue infiltration was observed in nonproteinuric mice (Fig. 9 A, bottom). To further investigate whether autoantibody production might contribute to the autoimmune phenotype, we screened a large cohort of Rag2^R229Q and control animals for the presence of anti–double-stranded DNA (anti-dsDNA) measured by ELISA and IFA with Chionodoxa luciliae, two complementary assays for detection of DNA autoantibodies characterized by low and high specificity, respectively. We observed a significant increase in the frequency of sera positive for both IgG and IgM specific for dsDNA in the group of mutant mice as compared with WT (P < 0.01; Fig. 9 B and not depicted). Notably, high titers of IgG anti-dsDNA were detected in animals with severe kidney infiltrations and urine protein loss (unpublished data). Staining of frozen sections of multiple organs from Rag2/Il2rc double knock-out mice (devoid of endogenous Ig) with sera derived from mutant mice confirmed the presence of autoreactive Ig targeting different tissues (Fig. 9 C).

These findings are indicative of defaults in B cell selection and tolerance. Receptor editing is the mechanism by which B lymphocytes alter their antigen receptors through secondary antibody gene rearrangements, usually involving the κ and λ light chain gene loci. Repeated variable joining rearrangements eventually lead to exhaustion of the recombination potential at the Igk-V gene locus and expression of a λ light chain. A markedly reduced frequency of λ chain⁺ B220⁺ cells was observed in mutant mice (1.5 ± 1% vs. 4.7 ± 1.6%; P = 0.0002; Fig. 9 D), thus suggesting that Rag2^R229Q mutation impairs the process of secondary recombination. In normal conditions, peripheral self-reactive B cells compete with non–self-reactive B cells for survival factors. BAFF is the key regulator of B cell homeostasis beyond the late transitional B cell stage (Meyer-Bahlburg et al., 2008), when expression of BAFF-R is acquired on B cells. Because increased BAFF levels are expected in mice with reduced B cell numbers (Lesley et al., 2004; Lavie et al., 2007), we measured BAFF serum levels in our model. Significantly, higher levels of BAFF were already detected in Rag2^R229Q mice (P < 0.0001; Fig. 9 E) early in life (5-wk-old animals), whereas in WT mice the increase in BAFF serum levels was age-dependent (Fig. S6 D).

As BAFF promotes B cell survival, its overexpression could potentially break B cell tolerance by rescuing self-reactive B cells from deletion. Indeed, mice transgenic for Tnfsf13b (the gene encoding BAFF) develop severe autoimmune symptoms (Mackay et al., 1999; Lesley et al., 2004; Thien et al., 2004). Based on this evidence, we speculated that increased BAFF serum levels in Rag2^R229Q mice could play a crucial role in B cell–mediated autoimmunity by favoring the survival and differentiation of autoreactive ISCs. To directly test this hypothesis, we examined the in vivo effect of selective BAFF-BAFF-R signaling blockade. Mutant mice displaying an overt pathological phenotype (i.e., presence of serum anti-dsDNA IgG and proteinuria) were i.v. injected with three 0.5-mg doses of anti–BAFF-R mAb (9B9) known to prevent BAFF binding (Rauch et al., 2009), and followed for 2 mo. Rag2^R229Q mice treated with anti–BAFF-R mAb showed a significant decrease in the number of splenic B220⁺ B cells compared with PBS-treated (control) animals (P = 0.008; Fig. 10 A). In WT mice, administration of anti–BAFF-R resulted in 90% depletion of FO B cells and 68% depletion of MZ B cells (unpublished data). Interestingly, a marked reduction of both CD4 and CD8 T cells was observed in mutant, but not WT mice, consistent with the role of BAFF signaling on activated T cells (Mackay and Leung, 2006; unpublished data).

The frequency of IgM, IgG, and IgA-ISC in the spleen was not significantly different in treated versus control mutant mice, but because of the smaller splenic B cell counts, the total number of ISC was significantly lower in anti–BAFF-R–treated Rag2^R229Q mice than controls (P < 0.001; Fig. 10 B and not depicted). After euthanization, histological examinations were performed and the degree of cell infiltration was scored in different organs (liver, gut, lung, kidney, and skin). Remarkably, a significantly decreased infiltration index was generally observed in Rag2^R229Q anti–BAFF-R–treated mice compared with controls, and specifically in the extent of infiltrating B220⁺ and IgM⁺ B cells (Fig. 10 C). The potential clinical effect of anti–BAFF-R administration was also evaluated. The Rag2^R229Q mice studied manifested similar proteinuria levels at the beginning of treatment, which significantly worsened over time in mice treated with PBS (P = 0.04; Fig. 10 D). In contrast, stabilization of kidney disease was observed in anti–BAFF-R–treated mice, with 4/7 mice displaying a reduction in the proteinuria index (Fig. 10 D). Consistently, the treated mice had markedly reduced signs of glomerular damage compared with the control mice that showed mesangial cell proliferation and heavy deposition of IgG (Fig. 10 D). Furthermore, 2 mo after treatment, the disappearance of IgG anti-dsDNA antibodies was detected in the majority of the treated mice (Fig. 10 D). Overall, these results support a causative role of B cells in the Rag2^R229Q immunopathology, providing preliminary evidence for a potential therapeutic application of BAFF-BAFF-R signaling blockade.

In this study, we have exploited Rag2^R229Q mice, which reproduce the immunopathology observed in OS patients, to address whether B cells play any role in the pathogenesis of the disease. The impairment of early B cell development in Rag2^R229Q mice determines a dramatic reduction of the peripheral B cell compartment, which is characterized by the presence of Ig-negative precursors and only a few IgM⁺ IgD⁻ B cells, which express several activation markers. FO B cells were almost undetectable, whereas splenic MZ B cells were barely present. In contrast, mutant animals contained an expanded CD21⁻CD23⁻IgM⁻ lymphocyte population. This phenotype could be reminiscent of mice with defective B cell development caused by lack of the surrogate light chain. In fact, these mice displayed an expanded B220⁺CD21⁻CD23⁻IgM^low cell subset and produced ANAs upon in vitro polyclonal stimulation (Harfst et al., 2005; Keenan et al., 2008). Based on this, it can be speculated that in Rag2^R229Q mice most of the peripheral B cells carry autoreactive potential.

The analysis of a large cohort of mutant mice revealed, in some animals, a normal or even higher Ig production compared with WT mice. This variability, resembling the clinical phenotype reported in OS patients (Villa et al., 2001), is actually greater than the one previously reported by Marrella et al. (Marrella et al., 2007). Such a discrepancy could likely reflect the mixed genetic background of our model and the effects of stochastic events occurring during B cell repertoire generation. In mutant mice, the residual serum Ig production is maintained by an enlarged ISC compartment. Excessive commitment to plasma cell fate is sustained by a defined transcription program. The observed increase in Blimp1 likely leads to cessation of cell cycling activity in Rag2^R229Q B cells (Lin et al., 1997), repression of genes required to establish GC reactions, and induction of the Ig secretory pathway (Shaffer et al., 2002; Shaffer et al., 2004). Our findings correlate with previous observations in immunodeficient and/or irradiated hosts, reconstituted with a limited number of peripheral B cells (Agenès and Freitas, 1999). In these animals, B cells quickly expanded, displayed an activated phenotype, and gave rise to an increased proportion of plasma cells. Hence, a compensatory homeostatic regulation might account for the skewed B cell phenotype in our model. During inflammation Aid and Blimp1 expression can be induced in immature/T1 B cells by TLR engagement in a T cell–independent manner (Ueda et al., 2007). Consistently, Aid⁺ immature/T1 B cells were preferentially found outside the conventional GC, at sites exposed to exogenous antigens (Wang et al., 2000; Ueda et al., 2007). The most severe inflammatory and autoimmune-like manifestations in Rag2^R229Q mice affect organs like the gut and the skin, where exposure to microbial antigens could potentially select immature B cell clones for expansion and antibody production. We showed that B cells from Rag2^R229Q mice efficiently responded to TLR ligands and polysaccharides Ag by undergoing rapid plasmacytic differentiation and Ig class switching. The limited variability consequent to impaired recombinase activity and the stochastic nature of B cell repertoire generation might represent factors determining the phenotypic heterogeneity of OS patients.

Several experiments in this study addressed the role of T cells in B cell homeostasis in Rag2^R229Q mice. We found that CD4 T cells were required for B cell activation and differentiation to plasmablasts in Rag2^R229Q mice. We provided evidence that the dominating T cell subsets in mutant mice, namely Th1, Th17, and T_FH cells do have a crucial role in B cell abnormalities. In fact, anti-CD4 treatment caused a significant reduction of ISC compartment and a normalization of serum IgE. These results correlated with the decreased serum concentration of cytokines involved in CSR such as IFN-γ and IL-21. Interestingly, IL-21 has recently been reported to stimulate IgE synthesis in humans, thus highlighting its important role in the occurrence of allergic and atopic disorders (Kobayashi et al., 2009). Furthermore, we demonstrated that T cell help for B cells was not dependent on cognate interaction, as Rag2^R229Q mice were severely impaired in the antibody response to TD antigens. The lack of OVA-specific IgG production might be caused by impaired T-B cell collaboration in the absence of GC-like structures. On the other hand, in this issue, Walter et al. show that adoptive transfer of WT CD4⁺ T lymphocytes was unable to correct defective antibody response to TNP-KLH, indicating B cell intrinsic abnormalities.

In Rag2^R229Q mice, a skewed pattern of cytokines expression and T helper cell bias may favor B cell–mediated autoimmunity. Dysregulated T_FH cells produce large amounts of IL-21, which promotes B cell activation as well as secretion of pathogenic anti-dsDNA autoantibodies and contributes to chronic T cell–mediated autoimmune inflammation by the generation of Th17 cells. Interestingly, it has been shown that IL-17 acts in synergy with BAFF in supporting B cell proliferation and differentiation to plasma cells in SLE patients (Hutloff et al., 2004; Iwai et al., 2003). In our model, homeostatically proliferating T cells are critical for the generation of ISCs, as well as the induction of the hyper-IgE state in mutant mice, and most likely contribute toward the pathogenesis of distinctive OS clinical features.

On the other hand, autoimmunity develops to the same extent in T cell–sufficient and T cell–deficient BAFF-transgenic mice (Groom et al., 2002), indicating that BAFF-mediated rescue of self-reactive B cells may be T cell independent. High-affinity anti-dsDNA serum IgG autoantibodies together with severe kidney damage with proteinuria, which we observed in Rag2^R229Q mice, are pathognomonic signs of B cell–mediated autoimmunity (Sobel et al., 1991; Manson et al., 2009). In the hypomorphic Rag1 murine model described by Walter et al. (2010), only low-affinity IgG autoantibodies were detected and the majority of the mice were devoid of clinical symptoms. This variability in clinical phenotype may reflect the different impact of the genetic defect on the recombination activity and the contribution of environmental factors.

Fewer B cells expressing λ chain in the mutant mice is suggestive of defects in the secondary rearrangements (Vela et al., 2008). Because an impairment in this process would also affect autoreactive Ab-expressing B cells, which could not therefore be silenced by light chain editing, it can be speculated that hypomorphic Rag2 mutation leading to impaired RAGs activity limits the potential to edit autoimmune BCRs. Under physiological conditions, self-reactive B cells that escape deletion and receptor editing, compete with nonautoreactive B cells for peripheral survival signals. Among these signals, BAFF has a crucial role in controlling peripheral B cell homeostasis (Mackay et al., 2003). In Rag2^R229Q mice increased BAFF serum level is likely a consequence of the B cell lymphopenic environment (Lesley et al., 2004; Lavie et al., 2007). It has been shown that increased BAFF levels favor the persistence of self-reactive B cells normally deleted at developmental checkpoints (Lesley et al., 2004; Thien et al., 2004). Indeed, high levels of BAFF have been detected in the serum of patients with various autoimmune conditions (Zhang et al., 2001; Groom et al., 2002; Lindh et al., 2008). In Rag2^R229Q mice high levels of serum BAFF might rescue autoreactive clones and favor their differentiation to ISC. We showed that blocking of BAFF-BAFF-R signaling in Rag2^R229Q mice with an anti-BAFF-R monoclonal antibody led to the disappearance of anti-dsDNA IgG antibodies, a significant amelioration of inflammatory infiltrates and an arrest of renal disease progression. Thus, these data strongly suggest that B cell abnormalities and increased BAFF serum levels play a significant role in autoimmune manifestations in Rag2^R229Q mice. Walter et al. (2010), in the companion paper, also measured elevated serum BAFF levels and autoantibody production in a large proportion of patients with OS and leaky SCID caused by hypomorphic rag mutations. Based on the results presented in this study, we propose that defects in B cell differentiation and function actively contribute to the OS pathogenesis. In support of this hypothesis, we reported the presence of terminally differentiated plasma cells in the lymphoid organs of OS patients devoid of circulating B cells. The similarities between murine and human studies validate the Rag2^R229Q model for further investigation on the contribution of B cell defects to autoimmunity. In conclusion, we have shown that hypomorphic rag mutations do not only result in profound impairment of B cell development but are also associated with the generation of an autoimmune BCR repertoire. We showed that the peripheral cytokines milieu crucially influences the generation of ISC and B cell–mediated immunopathology. The contribution of B cells to autoimmune pathology could provide the basis for a potential use of B cell–directed intervention in OS treatment.

Six unrelated patients with hypomorphic RAG defects whose clinical, immunological, and molecular features were consistent with OS were studied (Table S1). Human samples were collected after signed informed consent in accordance with the protocols of Spedali Civili of Brescia and Erasmus MC Ethical Committees. Mononuclear cells from peripheral blood and BM were purified on Ficoll gradient (Nycomed Pharma A/S).

129Sv/C57BL/6 knock-in Rag2^R229Q mice were previously generated (Marrella et al., 2007), and housed in specific pathogen–free facilities. Immunizations were performed in mice aged 2–3 mo. T cell–dependent antigen response was induced by footpad injection of 100 µl of a 1:1 emulsion of CFA and 100 µg OVA (grade V; Sigma-Aldrich). Secondary response was elicited by boosting with 50 µg OVA in IFA (Sigma-Aldrich). T cell–independent antigen response was induced by i.p. injection of 115 µg Pneumovax-23 (Merck). For CD4⁺ T cell depletion, mice were injected i.v. with two doses of GK1.5 (0.25 mg) 3 wk apart and followed for a total of 7 wk. Anti–BAFF-R (9B9) mAb treatment was performed in mice (3–5 mo age) previously tested for serum autoantibody and urine proteinuria. 3 doses of 0.5 mg were administered i.v. at 3 wk intervals. The mice were followed for 2 mo. In all experiments control mice were injected with PBS. All procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of the San Raffaele Scientific Institute (IACUC318).

Proteinuria was evaluated using Albustix stick (Bayer). Proteinuria index was scored as follows: 0 < 30 mg/dl; 1 = 30 mg/dl; 2 = 100 mg/dl; 3 = 300 mg/dl; and 4 ≥ 2,000 mg/dl.

Mononuclear cells from the peripheral blood and BM of patients were labeled with the following antibodies: anti–human-CD19, CD24, CD38, CD22 (BD); SmIgM, SmIgD, CyIgM (Kallestad/Sanofi-Synthelabo). Cell suspensions from murine BM, spleens, and LNs were stained in PBS 2% FCS-containing antibodies. To reduce nonspecific binding, cells were pretreated with anti-CD16 (BD). The following anti–mouse antibodies were purchased from eBioscience: B220, CD5, CD21, CD23, BP-1, CD40L, and PD-1. Reagents purchased from BD include: IgM, IgD, MHCII, CD40, ICOS, CD69, CD80, CD86, AA4.1, Mac-1, CD4 (RM4-5 and GK1.5), CD138, λ chain, HSA, streptavidin, and appropriate isotype controls. Anti–mouse-CD43 and CXCR5 were obtained from BioLegend. To detect cytokines production by T cells, splenocytes were cultured for 5 h with PMA (50 ng/ml), ionomycin (1 µg/ml), and monensin (1 µg/ml; Sigma Aldrich), and then intracellular IFN-γ, IL-4, IL-10, and IL-17 were stained according to the manufacturer’s instructions (BD). Samples were acquired on a FACSCanto II system (BD) and analyzed with FLOWJO software (version 4.5.4; Tree Star Inc.). Splenic B lymphocytes (B220⁺) used for mRNA analyses were sorted with a FACSAria (BD). The purity was >95%.

Levels of IgG isotypes, IgA, and IgM were measured in culture supernatants and sera by multiplex assay kit (Beadlyte Mouse Immunoglobulin Isotyping kit; Millipore) according to the manufacturer’s instructions and run using a Bio-Plex reader (BioRad Laboratories). Levels of IgE were determined by ELISA assay (BD). Cytokine and chemokine levels in supernatants and sera were measured with specific ELISA kits for mouse BAFF, IL-21, IFN-γ, and CXCL10 (R&D Systems). ISCs were analyzed by ELISpot using the species-specific purified and biotinylated anti-IgG, IgM, IgA, and IgE mAbs (Southern Biotech) and the 3-amino-9-ethylcarbazole (Sigma-Aldrich) as a chromogenic substrate.

For anti-dsDNA, 1:10 diluted serum was applied to fixed C. luciliae slides (Antibodies Inc.), with biotin-conjugated goat anti–mouse IgG (Southern Biotech), and Alexa Fluor 555–conjugated streptavidin (Invitrogen) as detection reagent. A reader blinder to the genotype/treatment of the mice read slides at 1,000×. Anti-dsDNA antibodies were also evaluated by ELISA assay. In brief, polystyrene plates were coated with poly-L-lysine (Sigma-Aldrich) and DNA from calf thymus (Sigma-Aldrich); after post coating with 50 µg/ml of Polyglutamic Acid for 45 min and blocking with PBS 3% BSA, serial dilutions of serum from 1:20 to 1:1,280 were incubated overnight. Bound Abs were detected with alkaline phosphatase-conjugated goat anti–mouse IgG (Southern Biotech). The score of positivity was assigned to sera that were positive for dilution of 1:20 or higher. For anti-cardiolipin antibodies, polystyrene plates were coated with 50 µg/ml of cardiolipin diluted in ethyl alcohol.

Autoantibodies against murine tissues were detected by indirect immunohistochemistry on cryostat tissue sections from adult Rag2/Il2rc^−/− mice to avoid interference from endogenous Ig, and stained with 1/20 dilutions of sera from Rag2^R229Q mice and control littermates. Slides were examined on a Zeiss Axioplan-2 microscope and images were acquired by Olympus DP70 camera using Cell^F imaging software (Soft Imaging System GmbH).

Total splenocytes were labeled with 0.5 µM CFSE (Invitrogen) for 8 min at room temperature. After quenching the labeling reaction by adding FCS, cells were washed twice. Labeled cells (10⁵ cells) were cultured in a 96-well U-bottomed plate with the following stimuli: 2.5 µg/ml CpG-B oligodeoxynucleotide (5′-tccatgacgttcctgacgtt-3′; Alexis Co.), 10 µg/ml LPS (R&D Systems), and 10 ng/ml IL-4 (R&D Systems). The proliferation profile of propidium iodide-negative viable B220⁺ cells was analyzed at day 4 of culture. On the same day, culture supernatants were collected for Ig quantification.

Formalin-fixed paraffin-embedded tissue sections from inguinal LN biopsies of OS patients and patients with unrelated nonimmunological diseases were subjected to routine hematoxylin and eosin (H&E) and double immunostains. In brief, sections were de-waxed, rehydrated, and endogenous peroxidase activity blocked before heat-induced epitope retrieval in 1.0 mM EDTA buffer (pH 8.0) followed by 1 h incubation with primary monoclonal mouse anti–human CD138 (clone NI15, 1:100; Dako) and rat monoclonal anti–Blimp-1 (clone 6D3, 1:50; Santa Cruz Biotechnology Inc.) antibodies. Double immunostainings were obtained by combining a first step using the appropriate biotinylated secondary anti–rat antibody (1:100; Vector Laboratories) followed by streptavidin (SA)-conjugated HRP and diaminobenzidine (Dako) as chromogen, and a second step using a biotinylated secondary anti–mouse antibody (1:100; Vector Laboratories) followed by SA-conjugated alkaline phosphatase (Dako) and Ferangi Blue as chromogen (Biocare Medical). Nuclei were counterstained with Methyl Green.

Formalin fixed paraffin embedded murine samples (liver, kidney, skin, lung, and gut) were subjected to H&E staining to check for basic histopathological changes, and immunohistochemistry was performed as previously described for human samples. Single immunostain reactivity was revealed using Real EnVision Mouse/Rabbit-HRP (Dako) or Super Sensitive IHC Detection System (BioGenex), followed by diaminobenzidine and nuclei counterstained with H&E. The following primary antibodies have been used: monoclonal rat anti-CD3ε (clone CT-CD3; 1:100; Valter Occhiena), rat anti-B220 (clone RA3-6B2; 1:100; Valter Occhiena), rat anti-F4/80 (clone BM8; 1:25; Caltag Laboratories), and polyclonal rabbit anti-Iba-1 (1:400; Wako). Murine plasma cells and Ig deposits have been detected directly by using either an appropriate biotinylated secondary goat anti-murine IgM (Dako) and goat anti–murine IgG (H+L; KPL) or ChemMATE Envision Mouse-HRP detection system (Dako) that bind specifically to the murine Ig. Digital images were then acquired with Olympus DP70 camera mounted on Olympus Bx60 microscope using AnalySIS imaging software (Soft Imaging System GmbH).

The degree of inflammation in the liver, lung, and kidney was double-blinded scored using the following criteria: 0 (none or scarse cell infiltrates); 1 (low to moderate); 2 (moderate to high); 3 (very high). Colitis was graded as follows (Scheiffele and Fuss, 2002): grade 0, no evidence of inflammation; grade 1, low level of leukocyte infiltration around the crypt basis in <10% HPF with no structural changes; grade 2, moderate infiltration in 10–25% HPF involving the lamina propria with minimal epithelial hyperplasia and/or goblet cell depletion; grade 3, high level of infiltration in 25–50% HPF reaching the muscularis mucosa with bowel wall thickening, moderate goblet cell loss, edema, crypt elongation, and epithelial erosions; grade 4: marked degree of transmural infiltration in >50% HPF with massive loss of goblet cell, fibrosis, elongated and distorted crypts with occasional abscesses, ulcerations, and bowel-wall thickening.

For bulk Igh PCRs splenic B220⁺ cells from WT and Rag2^R229Q mice were sorted on a FACS Aria (BD), excluding cell doublets, into 96-well PCR plates (Applied Biosystems; 50 cells/well) containing 20 µl/well of ice-cold 0.5× PBS supplemented with 10 mM DTT, 8 U RNAsin (Promega), and 0.4 U 5′-Prime RNase Inhibitor (Eppendorf). Plates were sealed and immediately frozen on dry ice before storage at −80°C. cDNA was synthesized in a total volume of 40 µl/well in the original 96-well sorting plate. The reverse transcription (RT) reaction was performed by using the High Capacity cDNA Archive kit (Applied Biosystems) following manufacturer’s instructions. cDNA was stored at −20°C. Igh transcripts were amplified by PCR using 3 µl of cDNA as template. All reactions were performed in PCR tubes in a total volume of 50 µl per well containing 100 nM of each primer (msVHE and mCh-mu-inner), 200 µM each dNTP (Invitrogen), and 1.25U HotStar Taq DNA polymerase (QIAGEN). Primers were stored in small aliquots to avoid repeated freezing and thawing and all PCRs were performed with nuclease-free water. PCR was performed for 35 cycles at 95°C for 30 s, 56°C for 30 s, and 72°C for 55 s. For single-cell Igh PCRs the experimental procedures have been described in detail before (Tiller et al., 2009). In brief, single B220⁺CD19⁺ splenocytes were sorted on FACSVantage with DiVa option (BD), using doublet discrimination and single-cell sorting mask settings. Cells were sorted into 96-well plates (Eppendorf) containing 4 µl of 0.5× PBS supplemented with 10 mM DTT (Invitrogen) and 8 U RNAsin (Promega). Synthesis of cDNA was performed in the 96-well plates used for sorting in a total volume of 15 µl containing random hexamer primers (5 µM; Roche), NP-40 (0.3% vol/vol; Sigma-Aldrich), 10 mM DTT, 36 U RNAsin, dNTPs (833 µM of each nucleotide; Invitrogen), and 50 U SuperScript III reverse transcription (Invitrogen). Reverse transcription was performed at 42°C for 5 min, 25°C for 10 min, 50°C for 60 min, and 94°C for 5 min. A semi-nested PCR strategy was used for amplification of Igh transcripts from single cells. First round PCR was performed in a total volume of 40 µl, containing 3.5 µl product of the RT reaction, msVHE and mCh-mu-outer primers (150 nM each), dNTPs (250 µM of each nucleotide), and 1 U HotStar Taq Polymerase. The reaction was performed for 50 cycles at 94°C for 30 s, 56°C for 30 s, and 72°C for 55 s. Second round PCR was performed in a total volume of 40 µl, containing 3.5 µl product of the first round reaction, msVHE and mCh-mu-inner primers (150 nM each), dNTPs (250 µM of each nucleotide), and 1 U HotStar Taq Polymerase. The reaction was performed for 50 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 45 s.

The primers used are as follows: msVHE, 5′-GGGAATTCGA­GGTGCAGCTGC­AGGAGTCTGG-3′; mCh-mu-outer, 5′-AGGGGGCTCTCGCAGGAGACGAGG-3′; mCh-mu-inner, 5′-AGGGGGAAGACATTTGGGAAGGAC-3′.

PCR products were then cloned into a pCR2.1-TOPO TA vector. Ligation was performed in a total volume of 10 µl with 1U T4-Ligate (Invitrogen), 2 µl PCR product, and 50 ng linearized vector. Competent E. coli TOP10F’ bacteria (Takara Bio Inc.) were transformed at 42°C with 10 µl of the ligation product. Colonies were screened by blue–white selection. Sequencing of plasmids and PCR products was performed using vector primer and msVHE, respectively. For analysis of bulk populations, identical nucleotide sequences generated from the same PCR well were excluded from the analysis. The obtained sequences were analyzed using National Center for Biotechnology Information IgBLAST (http://ncbi.nlm.nih.gov/projects/igblast/). Sequences were assigned to the Igh-V segment yielding the highest score, in case of a tie between several segments, the segment using the canonical nomenclature (<Igh-V segment family>.<segment number within family>.<segment number within locus>; Johnston et al., 2006) was assigned. Because of the genetic background of the animals (BX129 or 129XB), no discrimination between segments of the Igh^a (129) and Igh^b (B6) allele were made. Igh-D segments were only assigned if they showed at least six matching bases. Mismatches of the obtained sequences versus the known germline sequence were counted as hypermutations.

Total RNA from sorted B220⁺ cells of pooled mice (n = 5/group) was extracted using a conventional TRIzol-extraction (Invitrogen), treated with 100 U/ml Dnase I (Invitrogen), and reverse transcribed using the High Capacity cDNA Archive kit (Applied Biosystems). Expression of Prmd1/Blimp1 (Mm00476128_m1), Xbp1 (Mm 00457359_m1), Aicda (Mm00507774_m1), Tlr4 (Mm00445274_m1), and Tlr9 (Mm 00446193_m1) was quantified using the Assay on Demand kit (Applied Biosystems) and TaqMan Universal PCR mastermix (Applied Biosystems). qPCR was performed by using an ABI 7900 and analyzed using SDS version 2.3 (Applied Biosystems). cMyc, Bcl-6, T-bet, RORγT, Gata-3, and Cxcl2 expression was quantified by RT-qPCR from 50 ng cDNA in the presence of the TaqMan Universal PCR mastermix, 200 nM gene-specific TaqMan probe (Universal Probes; Roche), and 800 nM of relative primer pairs, using the DNA Engine Opticon 2 System (MJ Research). Thermal cycle conditions were the following: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 60 s. The primers and probes used are as follows: c-Myc, 5′-CCTAGTGCTGCATGAGGAGA-3′ and 5′-TCTTCCTCATCTTCTTGCTCTTC-3′, probe #77; Bcl6, 5′-TTCCGCTACAAGGGCAAC-3′ and 5′-CAGCGATAGGGTTTCTCACC-3′, probe #5 ; T-bet, 5′-TCAACCAGCACCAGACAGAG-3′ and 5′-AAACATCCTGTAATGGCTTGTG-3′, probe #19; RORγT, 5′-CGACTGGAGGACCTTCTACG-3′ and 5′-CCCACATCTCCCACATTGA-3′, probe #52; Gata-3, 5′-TTATCAAGCCCAAGCGAAG-3′ and 5′-TGGTGGTGGTCTGACAGTTC-3′, probe #108; Cxc12, 5′-CTGTGCCCTTCAGATTGTTG-3′ and 5′-TAATTTCGGGTCAATGCACA-3′, probe #41; Cyclophilin, 5′-TTCAAGCTGAAGCACTACGG-3′ and 5′-ATTGGTGTCTTTGCCTGCAT-3′, probe #20. Each sample was analyzed in duplicates, and the relative expression of target genes were defined by calculating the Δcycle threshold, with respect to the housekeeping gene.

Data were compared by a two-tailed Mann-Whitney U test. P values of <0.05 were considered significant. P values for Ig gene repertoire analyses were calculated by 2 × 2 Fisher’s exact test.

Fig. S1 shows the frequencies and absolute counts of different B cell subsets in BM and spleen. Fig. S2 shows the analysis of the Igh repertoire and clonal relationships of splenic B cells of mutant and WT animals. Fig. S3A illustrates that a higher proportion of ISC is present in the lymphoid tissues of Rag2^R229Q mice. Fig. S3B shows defective CXCL12 expression in the BM of Rag2^R229Q mice. Fig. S4 shows the expression of co-stimulatory molecules by CXCR5⁺ ICOS⁺ CD4⁺ T_FH cells. Fig. S5 describes the short-term humoral response to TI antigen. Fig. S6A shows that high serum concentration of the chemokine CXCL10 is present in Rag2^R229Q mice. Fig. S6B shows B cell infiltrates in the liver of mutant mice. Fig. S6C illustrates the degree of cell-specific infiltrates in different organs of mutant mice. Fig. S6D demonstrates that high BAFF serum level is already present in young Rag2^R229Q mice. Table S1 shows the clinical and genetic features of OS patients included in the study. Table S2 indicates that Rag2^R229Q mice have a restricted B cell repertoire.

The technical assistance of Massimiliano Mirolo, Lucia Susani, and Enrica Mira Catò is acknowledged.

This work was supported by grants from the Fondazione Telethon to A. Villa and from the Fondazione Cariplo to A. Villa, P.L. Poliani, and F. Grassi. FIRB grant RBIN04CHXT to P. Vezzoni and ERA/NET “Autoimmune liver disease” to A. Villa partially contributed to the study.

The authors have no financial conflicts of interest.