NFATc1 affects mouse splenic B cell function by controlling the calcineurin–NFAT signaling network

Stimulation of naive splenic B cells from 6–8-wk-old mice with 10 µg/ml anti-IgM antibody (α-IgM; the F(ab′)₂ fragment of anti–mouse IgM) in vitro leads to strong induction of NFATc1 after ∼24 h. A similar strong induction of NFATc1 was also detected by TPA+ionomycin (T+I) treatment, whereas a much weaker induction was observed after stimulation of B cells with 10 µg/ml LPS or 2 µg/ml α-CD40 (Fig. 1 A and Fig. S1 A). The short NFATc1 protein induced by α-IgM reacted with an antibody raised against the NFATc1 α-peptide and corresponds to NFATc1/αA, whereas the NFATc1 version induced by LPS did not react (Fig. 1 B). The Western blot data on the important role of BCR signals in NFATc1 induction are supported by the strong induction of an Egfp reporter gene controlled by the Nfatc1 locus upon α-IgM stimulation. When splenic B cells from mice bearing a BAC-Nfatc1/Egfp transgene were stimulated by α-IgM, LPS, or α-CD40 for 24 h, a much stronger GFP induction was detected after α-IgM than after LPS or α-CD40 treatment (Fig. 1 C).

To show that BCR-mediated induction of NFATc1 is affected by prototypical signaling molecules of the BCR-signaling pathways, we investigated the effect of α-IgM stimulation on NFATc1 induction in splenic B cells from mice deficient for the protein tyrosine kinase Btk (Kerner et al., 1995; Khan et al., 1995) or the adaptor molecule SLP-65 (Jumaa et al., 1999). As shown in Fig. 1 (D and E), inactivation of either Btk or SLP-65 led to complete loss of NFATc1 induction, indicating that the strong induction of NFATc1/αA is mediated through these signaling molecules, which are known to be key components of BCR signaling.

To investigate the role of NFATc1 in B cell physiology we used mice bearing “floxed” exon 3 alleles of Nfatc1 (Nfatc1^flx/flx). To generate mice in which the murine Nfatc1 gene was inactivated in B cells, Nfatc1^flx/flx mice were crossed with mice that specifically express the cre enzyme in B cells of BM (mb1-cre; Hobeika et al., 2006), spleen (Cd23-cre; Kwon et al., 2008), or GC (Aicda-cre; Kwon et al., 2008). Crossing of Nfatc1^flx/flx mice with the mb1-cre and Cd23-cre lines resulted in the ablation of NFATc1 expression in splenic B cells (Fig. S1, B–D).

Flow cytometric analyses of B cell development in BM from WT and mb1-cre x Nfatc1^flx/flx mice did not show significant differences in the expression of B220, CD43, BP-1, CD24, IgM, and IgD on pro–B cells (Hardy fractions B and C; Hardy and Hayakawa, 2001; CD24⁺BP-1⁻ and CD24⁺BP-1⁺ fractions in B220⁺CD43^hi cells), pre–B cells, and immature and mature B cells (Hardy fractions D–F; Hardy and Hayakawa, 2001; Fig. S2 A). Likewise, the differentiation of LN and splenic NFATc1^−/− B cells appeared to be relatively unaffected (Fig. S2 B), although only 75–80% B cells were counted in preparations of whole splenic B cells from 7 mb1-cre x Nfatc1^flx/flx mice compared with WT mice. Stains of splenic B cells for CD21+CD23 revealed similar ratios of follicular B cells and marginal zone B cells (MZBs) from WT and mb1-cre x Nfatc1^flx/flx mice (Fig. S2 C). Also, the maturation of splenic B cells did not markedly differ between WT and NFATc1^−/− cells (Fig. S2 D). However, in agreement with published data (Berland and Wortis, 2003), in peritoneal washes from mb1-cre x Nfatc1^flx/flx mice a 5–10-fold decrease in the population of B1a cells was detected. In the peritoneum of Cd23-cre x Nfatc1^flx/flx mice, B1a cells that do not express CD23 were found in normal numbers (Fig. S2, E and F).

Investigations of the α-IgM–mediated proliferative capacity of splenic NFATc1^−/− B cells in CFSE dilution assays and [³H]thymidine incorporation experiments showed a marked decrease in their proliferation in vitro, compared with WT B cells. In contrast, no differences in proliferation were observed between NFATc1^−/− and WT B cells upon LPS or α-CD40 stimulation (Fig. 2 A and Fig. S3 A). In addition, NFATc1^−/− splenic B cells showed a distinct increase in α-IgM–mediated apoptosis. This is reflected in the relative drop of living cells (as shown by their SSC-FSC features; Fig. S3 B) and the increase of Annexin V–positive apoptotic NFATc1^−/− cells compared with WT B cells (Fig. 2 B). Similar differences in survival between WT and NFATc1^−/− B cells were detected in cell cycle analysis by PI staining upon α-IgM stimulation for 48 h, showing a pronounced increase in hypoploid sub-G1 cells from ∼20–40% (Fig. S3 C). The increase in α-IgM–mediated AICD is a specific feature of NFATc1^−/− B cells because contrary to those, NFATc2^−/− B cells were resistant to α-IgM–mediated apoptosis (Fig. 2 B), which agrees with the pro-apoptotic activity of NFATc2 in activated T cells, as previously described (Chuvpilo et al., 2002).

For the identification of NFATc1 target genes that control the proliferation and/or AICD of B cells, we investigated gene expression of splenic B cells by using mouse whole-genome expression arrays. Freshly isolated B cells from WT and mb1-cre x Nfatc1^flx/flx mice were either left uninduced or induced by α-IgM for 3, 8, or 16 h. We identified 53 genes whose expressions were enhanced by more than twofold and 37 genes whose expression was suppressed more than twofold by NFATc1 upon α-IgM stimulation for 16 h (Tables S1 and S2). Among the enhanced “top ten” genes are the Spp1, Rcan1, Pdcd1, Tnfsf14, and Fasl genes that encode apoptosis regulators (Fig. 2 C). The Spp1 gene encodes the multifunctional protein osteopontin (Hur et al., 2007); the Rcan1 gene encodes the Cn modulator calcipressin (Vega et al., 2003); the Pdcd1 gene encodes PD-1 (CD279), a member of the Ig supergene family which is expressed on T and B cells (Keir et al., 2008); and the Tnfsf14 gene encodes LIGHT (CD258), a member of the TNF family that is a ligand of the lymphotoxin β receptor (LTβR) and herpes virus entry mediator (HVEM; Scheu et al., 2002). Although the RNA levels of all five genes were almost undetectable in nonstimulated B cells, the expression of Spp1 and Rcan1 genes was rapidly induced by α-IgM treatment for 3 h and decreased (Spp1) or remained quite constant (Rcan1) afterward in NFATc1^−/−, but not WT B cells. In contrast, the expression of Pdcd1, Tnfsf14, and Fasl genes was undetectable upon 3 h induction and increased strongly after 16 h of α-IgM stimulation in WT but not NFATc1^−/− cells (Fig. 2 D).

To show that the Spp1, Rcan1, Pdcd1, Tnfsf14, and Fasl genes are genuine NFATc1 targets in splenic B cells, we performed chromatin immunoprecipitation (ChIP) assays using splenocytes that were either left noninduced or induced with T+I or α-IgM for 16 h. Upon immunoprecipitation with an NFATc1-specific antibody, we detected a marked increase in NFATc1 binding to the promoters of Pdcd1, Tnfsf14, and Fasl genes after both stimulations (Fig. 2 E). In contrast, the Spp1 and Rcan1 promoters already showed NFATc1 binding in nonstimulated cells, but stimulation increased their NFATc1 binding. As expected, for the Il2 promoter an increase in NFATc1 binding was observed upon T+I stimulation but not upon α-IgM stimulation (Fig. 2 E).

To show whether in our assays PD-1 and LIGHT might affect the survival of B cells, we stimulated splenic B cells from PD-1 and LIGHT deficient mice by α-IgM. Compared with WT B cells, both types of B cells showed a strong reduction in AICD (Fig. S3 D). This indicates that three of the most prominent NFATc1 targets in primary B cells, the Pdcd1, Tnfs14 and Fasl genes, code for proapoptotic proteins.

Apart from the Ccnb1/Cyclin B1 gene, among the 53 genes that showed a twofold and stronger increase in expression upon BCR stimulation in WT (compared with NFATc1^−/− B cells) we did not detect genes that directly control cell cycle and proliferation (Table S1). This suggests that NFATc1 enhances the expression of (numerous) genes controlling proliferation only to moderate levels. Such view finds support in gene set enrichment analyses (GSEA) of microarray data in which numerous gene signatures enhancing proliferation were significantly enriched in WT B cells compared with NFATc1^−/− B cells. Among those are eleven signatures that are significantly associated with cell cycle and proliferation (Table I), and numerous cell cycle–associated genes. In addition to the Ccnb1 gene, these include several cyclin genes, which are listed as signature genes in leading edge analyses (Table S3).

Among the 10 genes whose expression was most strongly suppressed by NFATc1 in B cells are the Cd22, Fcer2a/CD23, and S100a4 genes, whose activity modulates (or is modulated by) intracellular Ca²⁺ levels in B cells (Fig. 3 A; Kolb et al., 1990; Nitschke, 2009; Matsuura et al., 2010). These findings suggest that NFATc1, as one of the most prominent NFATc1 targets (along with Rcan1; Figs. 2, C and E), controls BCR-mediated, store-operated Ca²⁺ entry, and thus the Ca²⁺-dependent Cn network, for its own induction in B cells.

Sustained Ca²⁺ entry is necessary for many physiological processes controlling lymphocyte activity, including their proliferation and apoptosis (Hogan et al., 2010). This is illustrated for splenic B cells in Fig. 3 B. Almost all cells proliferated when freshly prepared splenic B cells were cultured in X-vivo medium, a medium rich in Ca²⁺, whereas in a 1:1 mixture of X-vivo and RPMI media (the latter contains about one-third the amount of Ca²⁺ as the X-vivo medium) <50% of B cells were found to proliferate. Adding 10 mM Ca²⁺ to this medium restored the proliferative capacity, whereas inhibiting Cn by cyclosporin A (CsA), either at the start of the experiment or 1 d later, blocked any proliferation of B cells (Fig. 3 B). A similar strong effect has been observed with 10 µM NCI3, a novel Cn inhibitor (Sieber et al., 2007), and 2.5-µM Vivit-11R (Fig. S4, A and C), a specific NFAT inhibitor (Aramburu et al., 1999). The mild proliferative defect of NFATc1^−/− B cells upon α-IgM stimulation (Fig. 2 A and Fig. 3 B) is in line with the observation that NFATc1^−/− B cells exhibit a marked reduction in Ca²⁺ flux (Fig. S4 E), in both the release of Ca²⁺ from intracellular stores and in influx of extracellular Ca²⁺ to ∼60% of level to that in WT cells (Fig. 3 C).

CD22 is an inhibitory co-receptor of BCR that is transiently expressed on splenic B2 cells (Nitschke, 2009). Its surface expression is impaired by BCR stimulation (Lajaunias et al., 2002) and, as flow cytometric assays of WT and NFATc1^−/− B cells imply, NFATc1 contributes to CD22 suppression (Fig. 3 D). This conclusion is supported by the DNA microarray data (Fig. 3 A and Table S2) and semiquantitative RT-PCR assays that showed higher CD22 RNA levels in NFATc1^−/− than WT B cells treated by α-IgM for 16 h (Fig. 3 D). Splenic B cells from CD22^−/− mice show, on the other hand, a strong NFATc1/αA expression in noninduced and LPS-treated B cells, whereas WT B cells express isoforms other than NFATc1/αA under these conditions (Fig. 1 A and Fig. S4 F).

In addition to the Rcan1 gene, the Ppp3cb gene encoding the large CnA β subunit that is predominantly expressed in lymphocytes is a further direct NFAT target. This has been shown previously for cardiomyocytes (Oka et al., 2005), and in ChIP assays we detected the in vivo binding of NFATc1 to the 5′ region of Ppp3cb gene upon α-IgM stimulation (Fig. 3 E, top), whereas no NFATc1 binding was found to the 5′ region of Ppp3ca gene (not depicted). In Western blots using an antibody recognizing CnA α/β proteins, a strong decrease in the α-IgM–mediated induction of one CnA-specific protein was detected in protein extracts from NFATc1^−/− B cells (Fig. 3 E, middle), suggesting an important role of NFATc1 in CnA expression and function. This conclusion is supported by the marked decrease of Cn activity in extracts from freshly prepared NFATc1^−/− B cells, compared with WT B cells (Fig. 3 E, bottom).

The diminished Cn activity suggests that the activation of other NFATc factors might also be impaired in NFATc1^−/− B cells. In immunohistochemical stainings of B cells from WT and mb1-cre x NFATc1^flx/flx mice stimulated for 1 or 2 d with α-IgM and α-CD40 and restimulated for 30 min with T+I, we detected markedly less NFATc2 in nuclei of NFATc1^−/− B cells than WT cells (Fig. 3 F). This suggests an impaired NFATc2 activation in the absence of NFATc1. In RNA levels, no differences were detected for NFATc2 RNA between WT and NFATc1^−/− B cells, whereas a slight decrease in NFATc3 RNA levels was observed for NFATc1^−/− B cells, compared with WT cells (Fig. 3 F). However, because no distinct increase in NFATc3 RNA levels was detected upon α-IgM stimulation in WT B cells (Fig. 3 F), NFATc1 appears to indirectly affect Nfatc3 gene expression.

Collectively, these data indicate that NFATc1 controls Rcan1 and CnA expression, i.e., its immediate upstream control molecules, which, in a feedback loop, affect the transcriptional induction of NFATc1. In addition, NFATc1 seems to also affect upstream BCR signals, probably through CD22 and other targets. A scheme of this network connecting NFATc1 with its upstream activators in B cells is shown in Fig. S5A.

Interruption of BCR signaling by inactivating Btk, SLP-65, PLC-γ2, and further components of the BCR signaling cascade not only leads to a drastic decrease in Ca²⁺ flux, but also to defects in Ig class switching upon immunization with T cell–independent type II (TI-II) antigens (Kerner et al., 1995; Khan et al., 1995; Jumaa et al., 1999; Wang et al., 2000). When we measured Ig levels in sera from nonimmunized WT mice and mice bearing NFATc1^−/− B cells, we did not detect statistically significant differences in the concentration of Ig classes (except IgM in Cd23-cre x Nfatc1^flx/flx mice; Fig. S5 B). However, upon immunization with the TI-II antigen NP-Ficoll, we detected 10-fold lower IgG3 serum levels in Cd23-cre x Nfatc1^flx/flx mice, compared with those from WT mice. Slight, but statistically significant, differences were also detected in IgM levels 2 wk after immunization (Fig. 4 A). Very similar results were obtained upon immunization of Aicda-cre x Nfatc1^flx/flx mice that resulted in tenfold lower IgG3 serum levels compared with WT mice 1, 2, or 3 wk after primary injection. Again, the differences in IgM serum levels were smaller after 2 and 3 wk of immunization, but statistically significant (Fig. 4 A).

These in vivo data are supported by the Ig class switch to IgG3 in cultured B cells. Although in cultures of splenic B cells circular RNA transcripts of IgG3 switch region could be detected in WT B cells after 2 d, no circular transcripts were observed in NFATc1^−/− B cells, whereas circular IgG1 transcripts could be detected in both WT and NFATc1^−/− B cells (Fig. 4 B). Measuring of secreted IgM and IgG3 antibodies revealed a decrease in IgG3 secretion of ∼50% compared with WT cells (Fig. 4 C). After culturing B cells for 5 d, the percentage of CD138⁺ IgG3⁺ plasmablasts generated from NFATc1^−/− cells decreased to ∼30% of IgG3⁺ WT plasmablasts (Fig. 4 D). This might be caused, in part, by a decrease in proliferation and increase in AICD. Indeed, this was found for NFATc1^−/− CD138⁺ plasma cells in vitro (Fig. 4, E and F).

The Ig switch to IgG3 by TI-type II antigen takes place in B cells expressing high levels of NFATc1. This was shown by measuring NFATc1, IgM, and IgG3 levels in preparations of splenocytes from Nfatc1-Egfp reporter mice upon immunization with NP-Ficoll by flow cytometry. A high level of NFATc1 expression was detected for IgM⁺IgG3⁺ B cells, whereas B cells expressing only IgM expressed less NFATc1 (Fig. 4 G). NFATc1/GFP^high cells also expressed high levels of the GC markers PNA and Fas, the plasma cell marker CD138, and the activation marker CD69 (Fig. 4 G), indicating a close link between NFATc1 expression and Ig class switch.

TI-II antigens rapidly stimulate MZBs to produce IgM and IgG3, and defects in Ig class switch upon TI-II antigen immunization are often correlated with a decrease or lack of MZB cells (Martin and Kearney, 2002; Lopes-Carvalho and Kearney, 2004). Surprisingly, we did not notice a reproducible decrease in population of MZB cells in spleens of mice bearing NFATc1^−/− B cells, but defects in their function. When we injected TNP-Ficoll into mice and measured its binding to B cells, we observed a decrease in binding of TNP-Ficoll to NFATc1^−/− MZB cells, compared with MZB cells in WT mice. This is shown in Fig. 5 A for 8–9 mo old mice, whereas in younger 7–8-wk-old mice a 40–60% reduction in TNP-Ficoll binding was observed (unpublished data). 1 h after i.v. injection, TNP-Ficoll formed a ring like structure in the marginal zone of B cell follicles, distal of IgD^high B2 cells and associated with metallophilic macrophages (Fig. 5, B and C). Whereas in spleen from WT mice TNP-Ficoll was found to be evenly distributed and closely associated with IgM^high MZBs, larger aggregates of antigen were detected around, but not associated with CD169⁺ macrophages and IgM^high MZB cells in mice with NFATc1^−/− B cells (Fig. 5, C and D, white arrows). Similar deposits of antigen aggregates were described upon TNP-Ficoll immunization of mice deficient in Pyk-2, complement C3, or complement receptor CR1/2. All of these mice exhibit marked defects in MZB cells (Guinamard et al., 2000).

Contrary to the immunizations with TI-type II antigen, we did not detect remarkable differences in Ig class switch upon immunization with NP-KLH, a T cell–dependent (TD) antigen. Injections of four WT and four Cd23-cre x Nfatc1^flx/flx mice with NP-KLH (in alum) revealed only subtle, albeit statistically significant differences in serum levels of IgG2a one and two wks of immunization, but no differences in IgM, IgG1 and IgG2b levels (Fig. S5 C). On the same hand, no differences were detected in the affinity of IgG1 serum antibodies generated in WT and Cd23-cre x Nfatc1^flx/flx mice (Fig. S5 D). This shows that in contrast to the strong effect on Ig class switch by TI-type II antigens, inactivation of NFATc1 in B cells exerts a mild, if any effect on Ig class switch by TD antigens in the primary response.

B cells control immune reactions by presenting antigens to T cells by releasing cytokines and secreting humoral antibodies. When we tested the effect of NFATc1^−/− C57BL/6 B cells on allogenic T cells from BALB/c mice in MLCs we observed a stronger stimulatory activity of WT B cells than NFATc1^−/− B cells on T cell proliferation and IL-2 synthesis. This indicates NFATc1^−/− B cells as poor activators of T cells (Fig. 6, A and B). This conclusion found support in the weak stimulatory capacity of NFATc1^−/− B cells presenting an OVA peptide (OVA_323-339) to syngenic CD4⁺ T cells from OTII mice that are transgenic (tg) for an OVA-specific TCR (Fig. 6, C and D; Barnden et al., 1998), and NFATc1^−/− c2^−/− double KO B cells displayed even less stimulatory activity (Fig. S6).

Several lines of evidence indicated that B cells can suppress immune reactions by secreting IL-10, and thereby affect the course of autoimmune diseases (Bouaziz et al., 2008; Lampropoulou et al., 2010). Compared with splenic B cells from WT mice, splenic B cells from Cd23-cre x Nfatc1^flx/flx mice harbor twofold more CD19⁺CD5⁺CD1d^high B cells that were described to produce IL-10 (Fig. 7 A; Matsushita et al., 2008). Upon treatment by LPS for 24 h (and, in addition, by T+I for the last 8 h) in vitro, the percentage of IL-10–producing CD5⁺CD1d^high B cells increased twofold (Fig. 7 B; note the lack of any IL-10 signal for splenic B cells from IL-10^−/− mice). A similar increase in the percentage of IL-10–producing CD138⁺ plasmablasts was detected upon differentiation of NFATc1^−/− splenocytes for 3 d in vitro (Fig. 7 C). Incubation with LPS, but not α-IgM, led to a strong increase in IL-10 secretion of both WT and NFATc1^−/− B cells, again with higher IL-10 levels for NFATc1^−/− than for WT B cells (Fig. 7 D). By adding B cells from WT or Cd23-cre x NFATc1^flx/flx mice, which were stimulated overnight with α-IgM or LPS or left unstimulated, to allogenic T cells, NFATc1^−/− B cells suppressed the IFN-γ production of T cells, whereas WT B cells did not or only slightly affected T cell IFN-γ expression. This suppressive effect of NFATc1^−/− B cells was markedly reduced by adding the IL-10–specific antibody JES5-2A5, which relieved IL-10–mediated inhibition of cytokine production of T cells by B cells (Lampropoulou et al., 2008). A very strong suppressive effect on IFN-γ production was exerted by recombinant IL-10 that could be relieved by adding α-IL-10. By adding α-IL-10 to T cell co-cultures, in part the suppressive effect of NFATc1^−/− B cells could be relieved (Fig. 7 E), suggesting that an increase in IL-10 production by NFATc1^−/− splenic B cells contributes to suppression of T cell proliferation in vitro.

These in vitro findings prompted us to investigate whether NFATc1^−/− B cells affect the clinical course of mouse EAE (Hur et al., 2007), a mouse model for human multiple sclerosis, which is strongly affected by IL-10 (Fillatreau et al., 2002; Bouaziz et al., 2008). By inducing adjuvant-driven EAE (Pierau et al., 2009) in mb1-cre x NFATc1^flx/flx and WT mice, we observed that NFATc1 ablation in BM B cells resulted in a much milder EAE than in control mice. Whereas in control mice the onset of EAE was detected 7–8 d after MOG_35-55 peptide injection reaching EAE scores of 2–2.5 at day 20, for mb1-cre x Nfatc1^flx/flx mice first EAE signs were obvious at day 11 upon MOG_35-55 injection, reaching scores of 0.5–1.0 (Fig. 7 F). This shows that the expression of NFATc1 in B cells enhances the severity of EAE.

The results of this study show that ablation of NFATc1 in committed BM pro-B cells or in splenic B cells exerts a relative mild effect on the overall differentiation of peripheral B cells. Peritoneal CD5⁺ B1a cells are the notable exception because their generation is strictly dependent on NFATc1 expression (Fig. S2, E and F; Berland and Wortis, 2002, 2003). However, although NFATc1 activity seems to play a minor role in the development of B cells in BM and periphery, it is of crucial importance for the fate and function of B cells upon BCR triggering. Our data show that NFATc1 controls (a) the proliferation and AICD of splenic B cells (and plasmablasts) upon BCR stimulation, (b) the Ig class switch to IgG3 by the TI-type II antigen NP-Ficoll and the formation of IgG3⁺ plasmablasts in vitro, (c) the IL-10 production of CD1d^highCD5⁺ B cells that suppressed the IFN-γ production by and proliferation of co-cultivated CD4⁺ T cells, and (d) the clinical course of MOG-peptide/adjuvant-driven EAE, a mouse model of human multiple sclerosis. Because NFATc1 ablation led to a marked decrease in the release of Ca²⁺ from endoplasmic stores, Ca²⁺ influx and Cn activity as well as in distinct changes in expression of genes that affect (or are affected by) Ca²⁺ levels, one may conclude that NFATc1 controls Ca²⁺ flux and Cn activity in peripheral B cells.

NFATc1^−/− B cells share several characteristics with B cells from mouse models in that individual components of BCR signaling cascade have been inactivated. Such as with mice deficient for Btk (Kerner et al., 1995; Khan et al., 1995), PLC-γ2 (Hashimoto et al., 2000; Wang et al., 2000), and SLP-65/BLNK (Jumaa et al., 1999; Pappu et al., 1999; Xu et al., 2000), and with Igα^Y204F/Y204F 'knock in' (k.i.) mice (Patterson et al., 2006). Similar to mb1-cre x Nfatc1^flx/flx mice, these mice have a severe block in the generation of B1a cells and in Ig class switch upon immunization with TI-type II, but not TD antigens, and their B cells are characterized by a marked diminished Ca²⁺ flux and BCR-mediated proliferation. However, contrary to B cell development in mb1-cre x Nfatc1^flx/flx mice, SLP-65^−/− mice show a defect at the pre–B cell stage (of CD43⁺BP1⁺IgM⁻ cells) in BM and, therefore, a drastic reduction in number of peripheral B cells (Jumaa et al., 1999). Unlike B cells from mb1-cre x Nfatc1^flx/flx and SLP-65^−/− mice, Igα^Y204F/Y204F k.i. B cells did not show an impaired survival in vitro (Patterson et al., 2006). These observations indicate that NFATc1^−/− B cells share several, but not all properties with those mice defective in individual components of BCR signaling. However, collectively, the phenotype of NFATc1^−/− B cells indicates that the induction of NFATc1 represents an important molecular event of BCR signaling that transmits antigenic signals from the BCR to NFATc1 and, finally, to the numerous NFATc1 target genes that control the proliferation, AICD, cytokine production and TI-type II antigen–mediated Ig switch of splenic B cells.

The diminished release of Ca²⁺ from intracellular stores and the decrease in Ca²⁺ influx in NFATc1^−/− B cells suggest NFATc1 as a key molecule in adjusting the levels of free intracellular Ca²⁺ in B cells. Although it is currently unknown how NFATc1 controls these signaling events, the expression (or activation) of molecules involved in early BCR signaling controlling Ca²⁺ release from endoplasmic stores seem to be under NFATc1 control. One prominent candidate gene is the Cd22 gene, whose expression is enhanced in NFATc1^−/− B cells. Inactivation of the Cd22 gene, on the other hand, resulted in an increase of NFATc1/αA levels in unstimulated B cells and B cells treated with LPS (Fig. S4 F). CD22 affects BCR signals by recruiting the tyrosine phosphatase SHP-1 upon phosphorylation of its multiple motifs, and it is likely that SHP-1 affects the phosphorylation of members of BCR cascade, such as of Btk, SLP-65, or PLC-γ2 that control Ca²⁺ release through activating IP₃ receptors (Nitschke, 2009).

It is currently unknown whether CD22 is the only (or major) target through which NFATc1 affects Ca²⁺ flux. The expression of human SERCA3 (ATP2A3) gene in endothelial cells and mouse IP₃R1 (Itpr1) gene in hippocampal neurons have been described to be under NFAT control (Graef et al., 1999; Hadri et al., 2006), but studies on IP₃R1 expression in T cells (Nagaleekar et al., 2008) did not reveal a positive effect of NFATs on IP₃R1 expression in these cells. Likewise, the expression of PLC-γ2, Btk, and SLP-65 appeared to be quite normal in NFATc1^−/− B cells (unpublished data). However, large proteins such as IP₃R1 and PLC-γ2 are expressed from very complex genes that, like the Rcan1 and Nfatc1 genes, might be under the control of multiple promoters and/or poly A sites and alternately spliced. It remains to be shown whether this is true for other molecules of BCR signaling.

Prominent upstream NFAT targets of Ca²⁺/Cn cascade are Rcan1, the gene encoding the Cn inhibitor calcipression/DSCR1, and the Ppp3cb gene expressing CnA β, the β form of catalytic subunit A of Cn in lymphocytes. For both the Rcan1-4 and Ppp3cb promoters, arrays of NFAT motifs have been described, and the binding of NFAT factors to these promoters has been reported for nonlymphoid cells (Rothermel et al., 2003; Oka et al., 2005). Our data indicate that in B cells Rcan1, CnA β, and NFATc1 form a regulatory network (Fig. S5 A) that helps to fine tune the proper expression of NFATc1, in particular of NFATc1/αA, and the activation of further NFATc factors during immune responses.

NFATc1 expression supports BCR-mediated proliferation and suppresses AICD of primary B cells (Fig. 2), e.g., in the early clonal expansion phase of an immune response. Although there are several studies on a direct NFAT-mediated regulation of cell cycle mediators in T cells (Baksh et al., 2002; Caetano et al., 2002; Carvalho et al., 2007), in our microarray studies among the numerous cyclin, cdk, or cell cycle inhibitor genes we detected only the Ccnb1 gene, encoding Cyclin B1, as a putative NFATc1 target gene in B cells (Table S1). This finding led us to conclude that in B cells, NFATc1 enhances numerous genes controlling cell cycle at moderate levels, and gene set enrichment analyses (Table I) support this view. One exception might be the Bcl2 gene (Table S2) that, in addition to its role in apoptosis, was described to retard cell cycle entry (Vairo et al., 2000). However, the detection of proapoptotic factors FasL, PD-1, and Light molecules as direct NFATc1 target and own experimental data (unpublished data) suggest that at very high concentrations NFATc1/αA could also act in a proapoptotic manner, killing the majority of amplified lymphocytes at the end of an immune response. But further experiments using more appropriate mouse models are necessary to substantiate this hypothesis.

The increase in CD1d^highCD5⁺ B cells producing IL-10 suggests that NFATc1 affects the regulatory capacity of B cells. The Il10ra gene belongs to the 37 genes whose expression was markedly increased in NFATc1^−/− B cells (Table S2), and in 2 out of 3 microarray assays a similar increase was detected for the Il10 gene (unpublished data). This implies that the induction of NFATc1 in B cells suppresses, either directly or indirectly, IL-10 production. This conclusion is supported by the α-IgM–mediated decrease in IL-10 synthesis of CD1d^highCD5⁺ B cells upon BCR triggering and NFATc1 induction, and suggests that the increase in IL-10–producing B cells contributes to the clinical course of EAE in mb1-cre x NFATc1^flx/flx mice. It has previously been shown that B cells play a key role in recovery from EAE by the secretion of IL-10 (Fillatreau et al., 2002). Along with studies on a regulatory role of B cells in the generation of EAE (Yanaba et al., 2008; Pöllinger et al., 2009), our findings lend support to this observation, but also suggest that an impaired antigen-presentation by NFATc1^−/− B cells might ameliorate EAE symptoms (Fig. 6, C and D). In addition to IL-10, other candidates that affect EAE are IFN-γ and osteopontin/Spp1 (Hur et al., 2007), which are both encoded by top ten NFATc1 target genes (Fig. 2 and Table S1).

Collectively, the data of our study show that the induction of NFATc1 plays an important role in the activation and function of splenic B cells upon BCR stimulation. Missing or incomplete NFATc1/αA induction appears to be one reason for the generation of B cell unresponsiveness whereas uncontrolled NFATc1/αA expression could lead to unbalanced immune reactions and autoimmune diseases. Experiments are in progress to elaborate more details of this view.

The B cell stage-specific cre lines mb1 (Cd79a)-cre (Hobeika et al., 2006), Cd23-cre, and Aicda-cre (Kwon et al., 2008) have been described previously. Mice bearing a conditional allele of Nfatc1 (Nfatc1^flx; also designated as Nfat2^flx) were generated by K. Mark Ansel, M.R. Müller, and A. Rao (Harvard Medical School, Boston, MA) by flanking exon 3 of the mouse Nfatc1 gene with loxP sites (to be described in detail elsewhere). All of the following gene-deficient mice were published previously: Btk (Khan et al., 1995); SLP-65 (Jumaa et al., 1999); IL-10 (Kühn et al., 1993); CD22 (Nitschke et al., 1997); Light (Scheu et al., 2002); and PD-1 (Nishimura et al., 1999). Transgenic Nfatc1-Egfp reporter mice were created upon insertion of a Egfp cDNA into exon 3 of mouse Nfatc1 locus cloned in a BAC vector spanning ∼210 kb Nfatc1 DNA. Details of those BAC-tg mice will be published elsewhere. Animal experiments were performed according to project licenses (55.2–2531.01-53/10B), which are approved and controlled by the Regierung von Unterfranken, Würzburg. In most of experiments, 6–8-wk-old mice were used. Naive splenic B cells were isolated using Miltenyi Biotec cell isolation kit to a purity of 95–98% as determined by flow cytometry. 0.5–1 × 10⁶ cells were cultured in 200 µl X-vivo 15 medium (Lonza) and stimulated with 10 µg/ml LPS (Sigma-Aldrich), 1–10 µg/ml α-IgM (F(ab′)₂ fragment goat anti–mouse IgM; Jackson ImmunoResearch Laboratories), 1 or 2 µg/ml α-CD40 (R&D Systems) or TPA and ionomycin as specified in the figure legends. Cell cultures were supplemented with CaCl₂ as specified, and/or with the Cn inhibitors cyclosporin A and FK506, or the NFAT inhibitor 11R-Vivit (EMD). Because of its better solubility and cellular penetration, 11R-Vivit was dissolved in DMSO and used in concentrations between 0.1 and 10 µM. Because DMSO itself interfered with B cell survival, the effect of DMSO on B cells was tested in parallel assays.

Naive CD4⁺ T cells were isolated using a CD4⁺ T cell isolation kit (Miltenyi Biotec), stained with CFSE and incubated with B cells that were preactivated overnight by incubation with 10 µg/ml α-IgM. IL-2 secretion of T cells was measured with a Ready-Set-Go ELISA kit (eBioscience).

Mice were immunized with 200 µg of myelin oligodendrocyte (MOG_35-55) peptide in CFA (Sigma-Aldrich) containing 800 mg of heat-killed Mycobacterium tuberculosis (Difco) as described previously in detail (Pierau et al., 2009). Animal experiments were approved by the state authorities of the Land Sachsen-Anhalt. Mice with a clinical score >2.5 were killed according to the animal statutes. The MOG peptide was provided by D. Rheinhold (Institute of Medical and Clinical Immunology, Magdeburg, Germany).

B cells from WT and mb1-cre x Nfatc1^flx/flx mice were left unstimulated or stimulated with 10 µg/ml α-IgM for 3, 8, and 16 h, respectively. Their RNA was isolated from frozen cells using the TRIzol reagent (Invitrogen). Gene expression analysis was performed in three biological replications using the mouse whole genome 430 2.0 gene expression arrays from Affymetrix. The data were normalized to 500 arbitrary units (MAS5) using the Gene Chip Operating Software (GCOS) from Affymetrix. For statistical analysis absent signals were floored to a common baseline (50 U) and expression values were log₂ transformed.

To compare the gene expression profiles between the WT and Nfatc1^−/− phenotypes for any given time point (3, 8, and 16 h after α-IgM stimulation), data were time zero transformed and sorted for differential expression. Genes that showed at least a twofold difference in expression in all three replicates were determined by overlap analysis using Access. Gene set enrichment analysis (GSEA; http://www.broad.mit.edu/gsea) was performed as described (Subramanian et al., 2005) using 81 lymphoma associated gene signatures published previously (Shaffer et al., 2001) from the signature database of the Staudt laboratory http://lymphochip.nih.gov/signaturedb/ and 1.687 curated gene sets (c2) from the Molecular Signatures Database (MSigDB; Subramanian et al., 2005). Because both phenotypes that were compared in this approach consisted of less than seven samples, “gene set” was chosen as permutation type in the settings of the GSEA software, as recommended by the GSEA team.

The signatures and gene sets were regarded as significantly enriched when the nominal P value and the tail area FDR (tail area false discovery rate) were less than 0.05 and 0.25, respectively. A selective leading edge analysis was performed using significantly enriched gene sets to extract the genes that account for the enrichment score.

For visualization of the data the Cluster and TreeView software packages provided by M. Eisen (http://rana.lbl.gov/EisenSoftware.htm; University of California, Berkeley, Berkeley, CA) were used.

Western blots were performed by separation of whole protein extracts on PAGE-SDS gels followed by immunodetection of NFATc1 using the 7A6 mAb (obtained from BD or Santa Cruz Biotechnology) or polyclonal antibodies (pAB) specific for NFATc1 or NFATc2 (ImmunoGlobe). For detecting NFATc1/αA (Fig. 1 B), a pAb raised against the NFATc1 α-peptide (ImmunoGlobe) was used. As loading control, filters were stained by Ponceau red. Signals were developed by chemiluminescence detection using Super Signal (Thermo Fisher Scientific).

Apoptosis assays were performed by staining with propidium iodide (PI, 125 µg/ml) and Annexin V-FITC (BD), incubated for 15 min at 37°C and analyzed on FACSCalibur using CellQuest software (BD) and FlowJo (Tree Star, Inc.). Electronic compensation of the instrument was done to exclude overlapping emission spectra. In total, 10,000 events were acquired, the cells were gated, and dual parameter dot plots of FL1-H (x-axis; FITC-fluorescence) versus FL2-H (y-axis; PI-fluorescence) were gated in logarithmic fluorescence intensity. For IL-10 detection in specific B cell subset, nonstimulated cells or cells stimulated with α-IgM or LPS were treated by T+I and GolgiPlug for the last 8 h, blocked with mouse Fc receptor specific mAb (BD) and stained with CD5 PE and CD1d FITC mouse-specific antibodies (BD). Cells were fixed and permeabilized using the eBioscience fixing and permeabilizing kit according to manufacturer’s instructions. Next, cells were stained with anti–mouse IL-10 APC antibody (eBioscience). Unstimulated cells and isotype controls were also used in flow cytometric analysis.

For the determination of cell cycle phase distribution of nuclear DNA, cells were first fixed in 100% methanol, permeabilized, and nuclear DNA was labeled with PI (125 µg/ml) after RNase (10 µl from 2 mg/ml stock) treatment. Cell-cycle phase distribution of nuclear DNA was determined on a FACSCalibur using CellQuest software, fluorescence (FL2-A) detector equipped with a 488-nm Argon laser light source and a 623-nm band pass filter (linear scale). In total, 10,000 events were acquired for analysis. CellQuest and FlowJo statistics was used to quantify the data at different phases of the cell cycle. Histogram display of FL2-A (x axis, PI-fluorescence) versus counts (y axis) has been displayed in Fig. S3 C.

NP-Ficoll (10 µg/mouse in PBS) or NP-KLH (100 µg/mouse, Alum precipitated) was i.p. injected into age-matched mice. Levels of NP-specific antibodies were determined by ELISA with 10 µg/ml NP-BSA coated MaxiSorp plates (Nunc). A sample of pooled sera from days 7 and 21 was used as standard and included on every ELISA plate. Total Ig titers from naive mice were determined by ELISA on PolySorp plates (Nunc) coated with isotype-specific antibodies and monoclonal antibody, and isotype antibodies served as standards (SouthernBiotech). Serum samples were applied as serial dilutions, and their concentration was quantified against the standard curve. Isotype-specific antibodies coupled to alkaline phosphatase were used for detection.

10⁶ splenocytes were cultured in 1 ml X-vivo medium without or with LPS (10 µg/ml) or with LPS + IL-4 (10 ng/ml) for 2 d, followed by RNA extraction with TRIzol and cDNA synthesis using the iScript cDNA Synthesis kit (Bio-Rad Laboratories). PCR reactions were performed with the primers for the detection of IgG3 and IgG1 circular transcript as described (Kinoshita et al., 2001). 2.5 × 10⁵ splenocytes were cultured in 1 ml X-vivo medium and stimulated without or with LPS for 6 d. Supernatants were collected and subjected to ELISA. Unlabeled goat anti–mouse IgM and IgG3 were used to coat ELISA plates, mouse IgM and IgG3 (0.1 mg/ml each) were used as standards, and alkaline phosphatase–coupled, isotype-specific antibodies were used for detection. Splenocytes stimulated with LPS for 3 d were stained with antibodies directed against CD19, CD138, and IgG1 and/or IgG3 for the detection of Ig class switch.

4 × 10⁴ cells were incubated in 200 µl X-vivo medium with or without α-IgM or LPS for 72 h, and supernatants were collected. ELISA was performed using a mouse IL-10 ELISA kit (eBioscience) according to the manufacturer’s instructions.

For antigen localization (Fig. 5), 150 µg TNP-Ficoll was injected into the tail vein of mice. 60 min later, B cells were prepared for flow cytometry from one part of the spleen, and the other part was used for histological examination. Cryosections were stained with biotin-conjugated hamster α-TNP-Ficoll IgG (BD), α-CD169/MOMA-1 (Serotec), α-IgD, or α-IgM (SouthernBiotech).

For nuclear localization of NFATc2 (Fig. 3 F), splenic B cells stimulated for 24 h by 10 µg/ml α-IgM and 5 µg/ml α-CD40 and stimulated by T+I for 30 min were harvested by cytospin, dried overnight, fixed in acetone, and stained with antibodies directed against NFATc2 (BD or Santa Cruz Biotechnology). Nuclear localization of NFATc2 in 100 cells from three individual experiments was determined by confocal microscopy.

5 × 10⁶ of freshly prepared murine splenic B cells were incubated in 700 µl X-vivo medium or RPMI 1640 (Invitrogen) containing 5% FCS (PAN), 1 µM Indo-1-AM (Invitrogen), and 0.015% Pluronic F127 (Invitrogen) at 30°C for 25 min. The cell suspension was then diluted with 700 µl medium containing 10% FCS and incubated at 37°C for another 10 min. The cells were washed twice with Krebs-Ringer solution containing 10 mM Hepes, pH 7.0, 140 mM NaCl, 4 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, 10 mM glucose, and 0.5 mM EGTA and diluted to a concentration of 10⁷ cells per ml. For measurement, the cells were diluted 1:10 in Krebs-Ringer solution and a baseline was recorded for 60 s. Ca²⁺ movement was assessed after stimulation of the cells with α-IgM (Jackson ImmunoResearch Laboratories), followed by the addition of Ca²⁺ in denoted concentrations after 2 min of recording. Increases in free intracellular Ca²⁺ were measured in real-time on a LSR II (BD). Raw data files were transferred to FlowJo software and are presented as a median and in comparative overlay analyses.

Cn activity of freshly prepared naive splenic B cells was determined using BIOMOL’s nonradioactive Cn kit according to the manufacturer’s instructions. In brief, 3 × 10⁷ splenic B cells were lysed in 50 µl of lysis buffer (+ protease inhibitors) by pipetting and vortexing, followed by high speed centrifugation for 45 min at 4°C in an Eppendorf centrifuge. For removing free phosphate, supernatants were passed through a desalting spin column, stored on ice, or quick-frozen in liquid nitrogen before storage at −70°C. Aliquots of supernatant corresponding to ∼3 × 10⁶ of B cells were tested in 50-µl assays, containing the RII phosphopeptide Cn substrate. The release of free phosphate was determined in a Malachite green assay. Incubations of parallel assays containing either no substrate, or 10 mM EGTA and EGTA+okadaic acid indicated the specificity of Cn reaction. In immunoblots, a CnA α/β-specific antibody (BD) was used to determine CnA levels (Fig. 3 E).

ChIP assays were performed according to a previously published protocol (Soutoglou and Talianidis, 2002) and slightly modified. In brief, the chromatin of 0.5–1 × 10⁸ splenocytes was cross-linked using 1% formaldehyde, and the reaction was quenched with 125 mM glycine. Nuclei were isolated and treated with DNase I, sonicated, and, after preclearing with protein A–Sepharose that was blocked with salmon sperm DNA, protein–DNA complexes were immunoprecipitated using an NFATc1 mAb (BD), or an unrelated GST antibody as control. After washing, elution, and reversion of cross-links, DNA was isolated by phenol/chloroform and used in PCR reactions that were performed with the following primers amplifying the promoters of the following genes: Il2, forward 5′-CCTGTGTGCAATTAGCTCA-3′ and reverse 5′-CTCTTCTGATGACTCTCTGGA-3′; Fasl, forward 5′-ACTTGCTGAGTTGGACCTCA-3′ and reverse 5′-GTAATTCATGGGCTGCTGCA-3′; Pdcd1, forward 5′-CAGGCACAGAACGTTCAGGAA-3′ and reverse 5′-AACAGGAGTTAGGGATCACA-3′; Rcan1, forward 5′-GCGGCATAGTTCCCACTGGTA-3′ and reverse 5′-GAGTGCTGGGCTTTTCATCCA-3′; Spp1, forward 5′-CACACTCTCCATGCCAGAGCA-3′ and reverse 5′-CACAGAGGCCTCTGCTTGTCA-3′; Tnfsf14, forward 5′-TGTCCGTCTGTCCATCCACCA-3′ and reverse 5′-GGTGCAAGATGCTTGCATGGA-3′; Ppp3ca, forward 5′-TTTTCTTCAGGGTGCCAGTC-3′ and reverse 5′-CAGTCTGTTCCCATCCACCT-3′; Ppp3cb, forward 5′-GCTTAGACTTTTCCAGCTGA-3′ and reverse 5′-GAGGCTACTACAGAAACGGAA-3′.

RNA was purified using TRIzol reagent (Invitrogen) and first strand synthesis was performed with 1 µg of RNA according to manufacturer’s protocol (Fermentas). For PCR reactions, the following primers were used: Cd22, forward 5′-TCCCAGACTTTCCCCTCCCTA-3′ and reverse 5′-CAACCACTTTTACCCACTGA-3′; Nfatc2, forward 5′-GGGTTCGGTGAGTGACAGTT-3′ and reverse 5′-CTCCTTGGCTGTTTGGGATA-3′; Nfatc3, forward 5′-CACGCCGATGACTACTGCAAACTG-3′ and reverse 5′-CCTTGGAGCTGAAATGATGGTGAC-3′.

All statistical analysis was performed using SPSS software. The Mann-Whitney test was used to evaluate the significance of differences in Cn measurements, for other results the unpaired Student’s t test was used. Differences with P values of 0.05 and less are considered as significant. *, P < 0.05; **, P < 0.005, and ***, P < 0.001. If not stated otherwise, mean values from three assays are shown, or one representative experiment is shown from three similar assays.

The GEO accession no. for microarray data is GSE21063.

Fig. S1 shows immunoblots indicating the induction of NFATc1 upon α-CD40 and T+I stimulation and the loss of NFATc1 induction in splenic B cells from mb1-cre and Cd23-cre x Nfatc1^flx/flx mice. Fig. S2 shows flow cytometric data on the role of NFATc1 on B cell development in BM, LN, spleen, and peritoneum. Fig. S3 demonstrates the effect of NFATc1 ablation on the proliferation and survival of splenic B cells. In Fig. S4, data are compiled on the effect of novel Cn inhibitor NCI3 on proliferation of splenic B cells, of the NFAT inhibitor VIVIT on proliferation and apoptosis of splenic B cells, of NFATc1 ablation on Ca²⁺ flux, and on CD22 ablation on NFATc1 induction in splenic B cells. Fig. S5 shows a model on the control of Cn-NFAT network by NFATc1; serum Ig levels of nonimmunized WT, mb1-cre, Cd23-cre, and Aicda-cre x Nfatc1^flx/flx mice; serum Ig levels of WT and Cd23-cre x Nfatc1^flx/flx mice upon immunization with NP-KLH; and the affinity of IgG1 antibody generated upon immunization with anti-NP₄ or anti-NP₁₆ KLH. Fig. S6 shows the effect of NFATc1 and NFATc2 ablation on the capacity of B cells to stimulate T cells to produce IL-2 and to proliferate. In Table S1, genes are compiled that were twofold stronger expressed in WT than NFATc1^−/− splenic B cells. In Table S2, genes are compiled that were twofold stronger expressed in NFATc1^−/− splenic B cells. In Table S3, a leading edge analysis of pathways/gene sets is shown that were significantly enriched in splenic B cells from WT mice compared with those from mb1-cre x Nfatc1^flx/flx mice after 16 h stimulation with α-IgM.

We are very much indebted to A. Dietzel, D. Michel, and I. Pietrowski for excellent technical support. We thank Drs. K. M. Ansel (UCSF) and A. Rao (Harvard Medical School) for generating and providing the Nfatc1^flx/flx mice (funded by National Institutes of Health grants CA42471 and AI40127 to A. Rao). For the kind gift of mouse lines we thank Drs. M. Busslinger (Vienna), G. and P. Matthias (Basel), M. Reth (Freiburg), S. Scheu (Düsseldorf), A. Waisman (Mainz) and H. Wiendl (Würzburg). For reagents and advice we are indebted to Drs. R. Baumgrass and M. Sieber (Berlin), N. Beyersdorf (Würzburg), T. Bopp, and E. Schmitt (Mainz), S. Fillatreau (Berlin), M. Lutz (Würzburg), D. Olive (Marseille), D. Reinhold (Magdeburg), and J. Wienands (Göttingen).

The work was supported by the Deutsche Forschungsgemeinschaft grant TRR52 (to F. Berberich-Siebelt, A. Avots, and E. Serfling), SFB643 and FOR832 Erlangen (to L. Nitschke), and BO 1054-2/2 (to U. Bommhardt); the Land Sachsen-Anhalt (N2/C2 to U. Bommhardt); and the Wilhelm-Sander and Scheel foundations (to E. Serfling).

The authors have not commercial affiliations and declare no competing financial interests.