Overexpression of Fkbp11, a feature of lupus B cells, leads to B cell tolerance breakdown and initiates plasma cell differentiation

Systemic Lupus Erythematosus (SLE) is a severe systemic autoimmune disease, characterized by multi-organ damages, triggered by an autoantibody-mediated inflammation, and with a complex genetic influence. It is today accepted that adult SLE arises from the building up of many subtle gene variations, each one adding a new brick on the SLE susceptibility and contributing to a phenotypic trait to the disease. One of the ways to find these gene variations consists in comprehensive analysis of gene expression variation in a precise cell type, which can constitute a good complementary strategy to genome wide association studies. Using this strategy, and considering the central role of B cells in SLE, we analyzed the B cell transcriptome of quiescent SLE patients, and identified an overexpression of FKBP11, coding for a cytoplasmic putative peptidyl-prolyl cis/trans isomerase and chaperone enzyme. To understand the consequences of FKBP11 overexpression on B cell function and on autoimmunity's development, we created lentiviral transgenic mice reproducing this gene expression variation. We showed that high expression of Fkbp11 reproduces by itself two phenotypic traits of SLE in mice: breakdown of B cell tolerance against DNA and initiation of plasma cell differentiation by acting upstream of Pax5 master regulator gene.

Introduction immune complex mediated inflammation, which can result in severe glomerolunephritis and vasculitis. The origin of SLE is generally attributed to a combination of a complex genetic influence [1,2], and badly described environmental factors. In line with this theory, the majority of human SLE occurs in adults and is clinically characterized by a succession of flares interspersed with remission phases. Actual treatments consist in non specific drugs such as nonsteroidal antiinflammatory molecules, antimalarial agents, glucocorticoids, and immunosuppressive drugs [3]. A better understanding of the pathology is today necessary to allow for the elaboration of more targeted and less immunosuppressive therapies for SLE, with fewer side effects.
The heterogeneity of SLE is evident from different points of views: the clinical phenotype is different from patient to patient (apart from the almost constant high production of antinuclear autoantibodies), as well as the genetic susceptibility to the disease. Although rare cases of monogenic lupus with a precise phenotype have been described in children (ex. C1q deficiency, TREX1 defects, SAMHD1 defects) [4], we must consider that adult SLE arises from the building up of many subtle gene variations, each one adding a new brick on the SLE susceptibility, and each one contributing to a phenotypic trait to the disease.
Trying to understand the mechanism of the different phenotypic traits of the disease (loss of immune tolerance leading to autoantibody production, defect of apoptotic debris clearance, immune complexes related kidney pathology, diverse skin manifestations, arthritis. . .) is a huge and essential effort. On a strategic point of view, one can guess at least two different roads to identify such molecular mechanisms of the SLE phenotypic expressions. The first one starts from the genomic variants already identified during Genome Wide Association Studies (GWAS). GWAS of SLE patients have identified more than 30 genetic polymorphisms that are associated with SLE, but the combination of these variants differs from patient to patient. These SLE susceptibility genes could affect different steps of SLE development including B cell tolerance breakdown leading to autoantibody production (e.g., PTPN22, BANK1, BLK, LYN), defective clearance of immune complexes and apoptotic debris (e.g., ITGAM, Fc gamma R, Complement), T cell activation (e.g., PTPN22, CTLA4) or organ damage (e.g., Complement, IRF5, TNFAIP3) [1,2].
However, these variations usually do not identify the causal gene, and it is likely that more than one gene variant is necessary to induce a phenotypic trait (because odds ratio compared to healthy individuals are usually weak).
The second road starts from a comprehensive analysis of gene expression variation in a precise cell type. Indeed, transcriptome analysis can constitute a good complementary strategy to GWAS because it could identify some deregulated genes as a resultant of the biological effects of the diverse genetic variants present in a patient.
This work uses this second strategy, and we have chosen to focus on B cells, because of the central role of this cell type in SLE. Indeed, several lines of evidence indicate that B cells are central to the disease process [5]: 1) B cells produce the autoantibodies, some of which are clearly pathogenic by immune complex deposits or by destroying their target; 2) (NZBxNZW)F1 and MRL-Fas lpr/lpr mice (murine models of human SLE) harboring the xid mutation, which inactivates Btk and causes a blockade of B cell development and B cell responses, no longer develop lupus phenotype, including autoantibodies and glomerulonephritis [6,7], as do (NZBxNZW)F1 mice having a very restricted IgM transgenic repertoire [8]; 3) the disease can be transferred in mice by B cells: immunodeficient SCID (severe combined immunodeficiency) mice populated with pre-B cells of (NZBxNZW)F1 mice develop many of the characteristics of (NZBxNZW)F1 mice, suggesting that genetic defects responsible for the development of SLE disease in (NZBxNZW)F1 mice are expressed in their B cells [9].
In order to better understand the role of B cell gene expression abnormalities in SLE immunopathology, we recently analyzed the B-cell transcriptome of SLE patients focusing on the inactive phase of the disease, to avoid gene variation expression linked to B cell activation which accompanies lupus flares [10]. We started to create new mouse models to reproduce the human SLE gene expression variations and have already shown that this functional genomic approach is successful with Carabin gene [11]. One of these genes, FKBP11, is overexpressed in sorted B cells of SLE patients in our analysis. FKBP11 gene encodes the FKBP19 protein, a member of the peptidylprolyl cis/trans isomerase (PPIase) FKBP family. The FKBP19 protein is a FK506 binding protein, containing a N-terminal signal sequence, a PPIase domain, a putative transmembrane domain, and lacking a calcium-binding EF-hand (helix-loop-helix structural domain), which is typical of several FKBP members of the secretory pathway. Notably, it is expressed in lymphoid tissue, in particular during plasma cell differentiation, but its precise biological role in B cells is unknown [12]. Thus, to understand the biological significance of the overexpression of FKBP11 in B cells during human SLE, we created lentiviral transgenic mice reproducing the high level expression of FKBP11. The results show that variation in the expression of this single gene is able to induce two phenotypic traits of human lupus: B cell tolerance breakdown and initiation of plasma cell differentiation. In addition, our results add new important informations on the role of Fkbp11 in B cell physiology.

Overexpression of FKBP11 in a subset of quiescent SLE patients
We recently analyzed a pangenomic transcriptome of purified CD19 þ peripheral B cells in patients with inactive SLE in comparison to B cells from age-and sex-matched controls [10]. FKBP11 was overexpressed in all patients with a strong statistical significance using two different probes in the DNA microarrays (p ¼ 0.006 for the FKBP11 probeset 1, and p ¼ 0.0113 for the FKBP11 probeset 2, see reference [10] and also GEO Series accession number GSE30153 (http://www.ncbi.nlm.nih.gov/geo/query/acc. cgi?acc=GSE30153)). Based on probeset 1, FKBP11 was overexpressed in all patients with a mean increase of 4-fold relative to controls. The overexpression of FKBP11 was much higher (mean of 9-fold over healthy controls) in a subset of five patients (Fig. 1A) displaying a distinct gene expression with many genes implicated in the Unfolded Protein Response (UPR) [10]. FKBP11 overexpression was validated by quantitative real-time RT-PCR in patients versus controls [10]. The role of FKBP11 in B cells and in autoimmunity has never been described. These data led us to investigate in detail the possible role of FKBP11 overexpression in promoting SLE.

Production of lentiviral transgenic mice overexpressing Fkbp11
In order to understand the possible consequences of FKBP11 overexpression on B cell function and autoimmunity, we developed a functional genomic approach consisting in the production and analysis of a new lentigenic mouse model overexpressing Fkbp11, on C57BL/6 background.
The pTRIP lentiviral vector used for the production of Fkbp11 transgenic mice ( Supplementary Fig. S1, A and B) allows for the separated and equal expression of GFP reporter and of murine FKBP19 (protein encoded by Fkbp11 gene), thanks to a T2A peptide derived from picornavirus [13]. The expression of proteins is under the control of the human Ubiquitin C (UbiC) promoter allowing a good expression of the gene of interest in lymphoid organs, notably in B cells [14,15]. We selected a founder with a single proviral integration for further studies, which was named Fkbp11 high (Supplementary Fig S1C). Using the same strategy, we also produced a control transgenic mouse line (GFP þ control line), with a pTRIP lentiviral vector resulting in the expression of GFP alone, from one unique proviral integration site ( Supplementary Fig. S1, D and E).
We first evaluated the expression of Fkbp11 mRNA in lymphoid organs in Fkbp11 high transgenic mice, by quantitative real-time RT-PCR. Fkbp11 is overexpressed in total , and from a subgroup of 5 SLE quiescent patients (black bar) compared to 9 healthy age and sex-matched controls. X-fold represents the 2 exp(Pi-Tmean) value for the patients, where Pi is the value of the FKBP11 probe set signal for a given patient, and T mean , the mean value of signals for the same probe set for the controls. (T mean ¼ 6.096) (Error bars, SEM; Ã P < 0.05, Mann & Whitney test). (B) Quantitative real time RT-PCR analysis of Fkbp11 mRNA expression in total bone marrow cells, splenocytes, purified splenic B cells and total lymph node cells from 3-month-old Fkbp11 high (n ¼ 3) and littermate control mice (n ¼ 3). Each sample was normalized to the endogenous control Hprt1. Each bar represents the level of Fkbp11 mRNA relative to control mice, for each condition ( Ã P < 0.05; ÃÃ P < 0.005; ÃÃÃ P < 0.0005: unpaired t test with Welch correction). (Error bars, SEM). (C) Western blot analysis of FKBP19 expression in splenocytes, purified splenic B cells and total lymph node cells from Fkbp11 high and littermate control mice. Actin was used as loading control.
bone marrow cells, splenocytes, purified splenic B cells and total lymph node cells (Fig. 1B). Western blot analysis confirmed the overexpression of FKBP19 in splenocytes, splenic sorted B cells and lymph node cells in Fkbp11 high mice (Fig. 1C), but FKBP19 expression was barely detectable in bone marrow cells. GFP þ control mice, while expressing GFP in splenocytes, splenic B cells and lymph node cells, did not overexpress Fkbp11 mRNA nor FKBP19 protein (Supplementary Fig. S1F and G) in lymphoid organs, arguing against a transgene-dependent effect for Fkbp11 overexpression in Fkbp11 high mouse line.

Lymphoid hyperplasia and increased T-independent B cell response in Fkbp11 high mice
In order to analyze the consequences of Fkbp11 overexpression in vivo, we studied the immunological phenotype of 8-month-old Fkbp11 high mice. As shown in Table 1, subset proportions were not different between GFP þ control transgenic mice and GFP -(C57BL/6) littermate control mice. Therefore GFP À control mice were used for the rest of the experiments. The control mice used in this study are always littermates for the transgene-positive counterparts.
Interestingly, Fkbp11 high mice developed lymphoid hyperplasia, characterized by an increase of bone marrow, spleen and lymph node total cellularity. This increase occurred in all characterized B and T-cell subpopulations, but was globally more pronounced for B cells (Table 1). Compared to control mice, basal activation of B cells and T cells was not modified in Fkbp11 high mice, as measured by the ex vivo expression of activation markers ( Supplementary  Fig. S2). In addition, at baseline, the production of serum IgG3 was increased in Fkbp11 high mice in a statistically significant manner, compared to control mice, although the levels of serum IgM, IgG, IgG1, and IgG2b was not different (Fig. 2, A and B).
In concordance with the in vivo lymphoid hyperplasia, B cell proliferation was increased after stimulation of splenocytes from Fkbp11 high mice with LPS for 3 days in vitro, compared to splenocytes from littermate control mice (Fig. 2, C-E). Finally, to test B cell responses in vivo, we immunized Fkbp11 high and littermate control mice with T-dependent (ovalbumin, OVA) and T-independent (immunization with 4-hydroxy-3-nitrophenylacetyl (NP)-LPS) antigens. The total IgG, and IgG1 immune response against OVA was unaffected by Fkbp11 overexpression (data not shown). However, compared to control mice, anti NP-LPS IgG immune response was amplified in Fkbp11 high mice compared to control mice (Fig. 2F).
Together, these data show that Fkbp11 overexpression led to lymphoid hyperplasia and increased T-independent B cell responsiveness (higher IgG3 at baseline and more robust anti-NP-LPS IgG after immunization with NP-LPS) in vivo.

Fkbp11 overexpression led to B cell tolerance breakdown
Since we initially observed Fkbp11 overexpression in human SLE B cells, it was of importance to test for the presence of autoantibody production in Fkbp11 high mice. The transgenic mice produced more IgG anti-double stranded (ds) -DNA antibodies than control mice (Fig. 3A). Not only these anti-dsDNA antibodies were produced, but some of the transgenic animals also made more IgM autoantibodies as Rheumatoid Factors (31% of Fkbp11 high mice produced more than the mean value þ standard deviation of controls), anti-thyroglobulin (29%) and anti-actin (57%) (Fig. 3, B-E), indicating a more general breakdown in B cell tolerance in these mice. However, this breakdown of B cell tolerance did not lead to the development of autoimmune symptoms, including glomerulonephritis, as assessed by proteinuria and immunohistological analysis of the kidneys (data not shown, n ¼ 14 mice analyzed in each group).
To gain insight into the breakdown of B cell tolerance in Fkbp11 high mice, we crossed them with the 56R anti-DNA heavy chain transgenic mouse model on C57BL/6 background. The 56R anti-DNA transgenic model (whose B cells express a transgenic heavy chain, associated with a restricted repertoire of endogenous light chains) was shown to produce mainly IgM anti-single-stranded (ss) -DNA [16,17]. Twelvemonth-old 56R/Fkbp11 high mice produced significantly more IgM anti-ssDNA autoantibodies than control 56R animals, confirming that the overexpression of Fkbp11 promoted the loss of B cell tolerance (Fig. 4, A and B).
Considering what is described in the 56R model, this result could be linked to at least three distinct mechanisms: abnormal central autoreactive B cell deletion, defective B cell anergy or defective B cell receptor editing. We did not find any experimental evidence for the first hypothesis since the numbers of IgM aþ (transgenic heavy chain) B cells in the bone marrow ( Supplementary Fig. S3A) and spleen ( Supplementary Fig. S3B), as well as the ratios of IgM aþ and IgM bþ (endogenous heavy chain) B cells in spleen (data not shown), are not modified by the high expression of Fkbp11. To test for the second scenario, we analyzed 56R/ Fkbp11 high mice for the level of surface IgM a expression, and for the B cell response to in vitro activation. We did not observe any downregulation of the basal surface IgM a expression on B lymphocytes ( Supplementary Fig. S3C). In addition, stimulation of B cells from 56R/Fkbp11 high mice with BCR-dependent (anti-IgM antibody, or CG-50, used to specifically stimulate anti-DNA B cells [18,19], or CG-Neg as a negative control [19]), or BCR-independent stimulus  The role of Fkbp11 in B cells and autoimmunity (LPS), did not lead to higher overexpression of activation markers (CD86, CD44, MHCII), to increased proliferation, nor to increased secretion of IgM anti-ssDNA and anti-dsDNA ( Supplementary Fig. S3, D-F), compared to B cells from 56R mice. In conclusion, the breakdown of B cell tolerance in 56R/Fkbp11 high mice could not be explained by a measurable defect in B cell deletion or anergy. Therefore, we wondered if these findings could be explained by a defect of receptor editing. To test this hypothesis, RS rearrangements, which constitute a marker of light chain receptor editing [20], have been analyzed in sorted splenic transitional T1 (B220 þ /IgM aþ /CD21 À /CD23 À ), marginal zone (B220 þ / IgM aþ /CD21 high /CD23 À ), and follicular (B220 þ /IgM aþ / CD21 low /CD23 þ ) [21] autoreactive IgM aþ cells from 56R/ Fkbp11 high mice, compared to 56R mice. We did not observe a significant difference in RS rearrangement levels in each of the IgM aþ autoreactive B cell subpopulations, in 56R/ Fkbp11 high mice compared to 56R mice ( Supplementary  Fig. S3G). Finally, we sequenced the Vk genes in T1 population and found a similar distribution of Vk families between 56R/Fkbp11 high and 56R mice, although with a greater usage of the minor Vk1 and Vk5 families ( Supplementary Fig. 3H). However, we do not know if the usage of these minor families is linked to the anti-DNA activity. Furthermore, we did not find any difference in Jk usage between the two groups of mice (data not shown).
In conclusion, Fkbp11 overexpression facilitates B cell tolerance breakdown by a mechanism which is not known yet. These results may be linked to the variation of expression of Fkbp11, which is tightly regulated during B-cell maturation. Indeed, Fkbp11 expression is increased in bone marrow transitional T1 B cells where central tolerance is known to take place ( Supplementary Fig. S4, A and B).
Fkbp11 overexpression exacerbated some traits of the immunopathology of B6 lpr/lpr mice Considering the breakdown of anti-DNA B cell tolerance on C57BL/6 background, it was interesting to test the effect of Fkbp11 overexpression added to a well-described autoimmune prone genetic defect. The B6 lpr/lpr mice are characterized by the late appearance of anti-DNA antibodies related to the apoptosis defect linked to Fas deficiency [22]. Thus, we crossed the Fkbp11 mice on the lpr/lpr mice on C57BL/6 (B6 lpr/lpr ) background. High expression of Fkbp11 on the B6 lpr/lpr background did significantly increase anti-dsDNA IgM, but not anti-dsDNA IgG production (Fig. 5, A and B). The splenomegaly was increased when Fkbp11 was overexpressed (Fig. 5C). In addition, the number of double negative (DN) T cells in spleen and lymph nodes were increased in Fkbp11 mice on the B6 lpr/lpr background (Fig. 5, D and E). Finally, the IgM and IgG hypergammaglobulinemia was amplified in B6 lpr/lpr mice by the addition of Fkbp11 overexpression (Fig. 5, F and G). Overall, these data showed that some traits of the immunopathology of B6 lpr/lpr mice were exacerbated by Fkbp11 overexpression.
Fkbp11 is implicated in the plasma cell differentiation program, upstream of Pax5 Considering that FKBP11 expression is increased during plasma cell (PC) differentiation [23,24], and that anti-DNA production is increased in Fkbp11 overexpressing mice on C57BL/6 and 56R backgrounds, we decided to study the consequences of Fkbp11 overexpression on B cell differentiation. Known in vitro models of B cell differentiation, and LPS stimulation in particular, result in antibody secreting cells that are mostly B220 low /CD138 þ plasmablasts (PB), and consistently overexpress Prdm1 (coding for Blimp1 protein) [25,26]. Taking advantage of our transgenic construct which directly links GFP and Fkbp11 expression, we followed the fate of GFP-expressing cells after LPS-induced in vitro PB differentiation. As shown in Figure 6A, at day 4, we observed in culture from Fkbp11 high mice a GFP med and a GFP low populations that are almost absent in culture from control mice. In addition, the GFP med population contained the majority of PB, compared to the GFP low population (51% of PB in GFP med B cells, versus 10% in GFP low B cells, P < 0.0001, Mann Whitney test) (Fig. 6B). We have shown by quantitative real-time RT-PCR that Fkbp11 is most importantly (about 5 times, data not shown) overexpressed in sorted GFP med B cells from Fkbp11 high mice, compared to sorted GFP low B cells from the same mice, as expected with the T2A system which permits an equimolar expression of the two proteins. Therefore, it indicates that Fkbp11 overexpression favored PB differentiation. In parallel, using splenic B cells from GFP þ control mice, we confirmed that this increase of GFP expression was not a consequence of ubiquitin promoter activation during LPS induced PB differentiation: indeed, GFP expression was not different in B220 þ CD138 À undifferentiated B cells and PB after 4 days of stimulation of isolated mature B cells with LPS in vitro (Fig. 6C). To better understand the role of Fkbp11 overexpression on B cell differentiation, we quantified the expression of PC differentiation master genes [26][27][28][29][30] by quantitative real-time RT-PCR, in sorted PB and undifferentiated B cells from the transgenic Fkbp11 high mice and littermate control mice, after 4 days of culture of isolated mature B cells with LPS in vitro. As expected, in control mice, the expression of Pax5, Bach2, and Aicda is significantly decreased in sorted PB versus undifferentiated B cells (Fig. 6D), while the expression of Prdm1, Xbp1, and Irf4 is increased (Fig. 6E). Interestingly, at the same time point, sorted PB from Fkbp11 high stimulated B cells expressed significantly lower Pax5, Bach2, and Aicda (Fig. 6D) (which are linked to the PC initiation program), than PB from littermate controls stimulated B cells. However, genes implicated in later stages of PB differentiation (Prdm1 and Xbp1) were not significantly different in PB from Fkbp11 high stimulated B cells compared to PB from littermate controls stimulated B cells (Fig. 6E), and the percentages of PB were not significantly increased after 4 days of culture of B cells from Fkbp11 high mice, compared to control mice (Fig. 6F). The higher downregulation of Pax5 expression in PB from Fkbp11 high stimulated B cells was confirmed at the protein level by Western-Blot analysis (Fig. 6G). In conclusion, considering the PC master gene differentiation program, Fkbp11 overexpression initiates PC differentiation, acting upstream of Pax5 through a higher depression precisely of genes whose downregulation is implicated in the initiation of PC differentiation. The levels of transcription of lately expressed genes, like Xbp1 and Prdm1, are not affected during our in vitro conditions perhaps because Pax5 level is not the unique regulator of Xbp1 expression. For example, the levels of expression of Bcl6 and Irf4, which are also regulators of Prdm1 (and consequently of Xbp1), are not different in PB from the two groups of mice (Fig. 6, D and E). In addition, Xbp1 expression also appears to be influenced by a post-transcriptional mechanism involving the Unfolded Protein Response [31]. Altogether, these results are consistent with the in vivo analysis of splenic B220 À CD138 þ plasma cell population, which was not significantly increased in Fkbp11 high mice, compared to control mice (0.9 AE 0.6% for Fkbp11 high mice versus 1 AE 0.5% for littermate control mice). Finally, to confirm the role of Fkbp11 on the early phase of B cell terminal differentiation, we performed Fkbp11 knockdown experiments in B cells from C57BL/6 mice, using a siRNA that targeting the murine Fkbp11 gene (Supplementary Fig. S5A). This led to the opposite phenotype: expression of Pax5, Bach2, and Aicda was increased after culture of siRNA treated-B cells with LPS in vitro, compared to B cells treated with a control siRNA ( Supplementary  Fig. S5B). Taken together, these results confirmed a positive role of Fkbp11 expression in the initiation of PC differentiation, and this effect is B cell intrinsic.

FKBP11 overexpression in human SLE B cells is linked to plasma cell differentiation
Human peripheral B cells contain a small fraction of PB (CD19 low/þ CD138 þ CD38 þ ). We found a increase of the percentages of circulating PB in the 17 inactive SLE patients (containing the initial subgroup of five patients), compared to controls (1.76% in SLE patients versus 0.84% in controls, Fig. 6H). This minor increase cannot by itself explain the observed variation of FKBP11 expression in total B cells during our transcriptomic analysis (ranging from À2 to 17, with two separate probe sets) [10]. Furthermore, we found a correlation between the percentages of circulating PB and the FKBP11 levels of expression (Spearman correlation test, P ¼ 0.0002; Fig. 6I), suggesting that this overexpression favored PB differentiation, thus reinforcing the results of our in vitro studies with transgenic B cells. Taken together, these results suggest that FKBP11 overexpression initiates PC differentiation, but other unknown molecular events were present in SLE patients to allow terminal PC differentiation.

Discussion
FKBP11 was identified by transcriptome analysis of human SLE B cells during clinically inactive disease. The higher expression of FKBP11 mRNAs in these cells, compared to control B cells, led us to study the effects of FKBP11 overexpression on B cell phenotype and on the development of autoimmunity.
For this purpose, we developed and validated a new model of ubiquitous Fkbp11 overexpression in mice, using a lentigenic transgenesis system. These mice are named Fkbp11 high , and show a 12 times overexpression of Fkbp11 in purified splenic mature B cells.
The phenotype of Fkbp11 high mice included hyperplasia of lymphoid organs (with a more significant increase of B-cell subpopulations), an increase of basal IgG3 production, associated with an increase of B cell responsivity to a T-independent antigen in vivo. This phenotype was not correlated with an increase in basal B cell activation or after stimulation in vitro.
Fkbp11 overexpression was sufficient to break B cell tolerance. This was characterized by a statistically significant increase of IgG anti-dsDNA production, with a sub-group of animals also secreting higher levels of IgM autoantibodies that targeted several autoantigens (including non-nuclear ones). Furthermore, the overexpression of Fkbp11 exacerbated some features of autoimmunity in the B6 lpr/lpr model. Finally, we used the well-documented 56R anti-DNA model on C57BL/6 background to study in more detail the mechanism of tolerance breakdown when Fkbp11 is overexpressed. The 56R model consists in a high affinity heavy chain that binds to ssDNA and a dsDNA and is knocked into the heavy chain locus by gene targeting [16,17]. Initial studies of B cell tolerance in the 3H9 and 56R anti-DNA KI model were conducted on BALB/c background, and have shown that tolerance to nuclear antigens was maintained through a variety of mechanisms, including deletion and receptor editing for high-affinity anti-dsDNA B cells, and anergy of anti-ssDNA B cells, that block autoantibody secretion. The 56R model on C57BL/6 background was shown to be more susceptible to a loss of tolerance [16,32]. Indeed, Liu and colleagues have crossed 3H9 and 56R models (on C57BL/6 background) to the lupus susceptible Sle2z congenic mice and have shown that 56R model was more prone to B cell tolerance breakdown, which was characterized by blockade of receptor editing, a skewing toward marginal zone B cells and preplasmablats, and an increased secretion of autoantibodies [16]. Our results on 56R/Fkbp11 high mice demonstrate that the production of IgM anti-ssDNA autoantibodies (corresponding to the highest proportion of autoantibodies produced by C57BL/ 6-56R mice) is increased. This phenotype was not readily explained by detectable abnormalities in central autoreactive B cell deletion, autoreactive B cell anergy nor autoreactive B cell receptor editing. Thus, if Fkbp11 overexpression by its own was indeed sufficient to induce a breakdown of B cell tolerance, which is a remarkable feature of SLE patients [33], further experiments are needed to explain the precise mechanisms. In addition, we cannot exclude a role of Fkbp11 overexpression in other cell types than B cells in the observed effects.
The increased number of circulating PC is a feature of many patients during SLE, even during inactive phases of the disease [34,35]. Since FKBP11 is overexpressed during PC differentiation [23,24], we asked about the consequences of Fkbp11 overexpression on B cell differentiation when Fkbp11 is overexpressed at a B-cell maturation stage (plasmablast, PB) preceding PC differentiation. It is currently accepted that Pax5 is one of the master proteins controlling PC differentiation [25,26]. However, little is known about the factors that control Pax5 expression [28]. Our data showed that, after 4 days of stimulation of purified splenic mature B cells with LPS in vitro, pro-germinal center B cell genes i.e., Pax5, Bach2, and Aicda (the last two being positively controlled by Pax5) exhibit reduced expression in PB from Fkbp11 high mice, as compared to PB from control mice, and this effect is B cell intrinsic. However, this effect of Fkbp11 overexpression on the PC differentiation genetic program is incomplete since the expression of pro-plasmacytic genes (Prdm1, Xbp1) was not intensified. Thus, the expression of Fkbp11 is not only increased during PC maturation, suggesting a potential role of Fkbp11 in this process, but the overexpression of this gene at a B-cell maturation stage preceding PC differentiation, can promote the initiation of this process. According to our in vitro results, Fkbp11 overexpression is not sufficient to induce a complete PC differentiation genetic program indicating that full PC differentiation requires additional signals that are not present in our in vitro PC differentiation system, but could be present in lupus patients. Importantly, our results localize FKBP19 protein (encoded by Fkbp11 gene) upstream of Pax5 (by direct or indirect effect) during the PC differentiation pathway. FKBP19, as the other members of FKBP family, seems to have a double function with a PPIase activity and a putative chaperone activity [36]. The PPIase activity catalyzes a cis/trans peptidyl-prolyl isomerization of target proteins, which could have an effect on the interactions between two particular partners and on their activation. It is also known that PPIase enzymes interact with a diverse range of intracellular and extracellular targets, however the molecular interactions between PPIase and target proteins are weak. It is tempting to inhibit FKBP19 PPIase activity in order to evaluate its precise role in the development of this phenotype in vivo. However, today there isn't any FKBP19specific inhibitor available [37]. FK506 inhibits the PPIase activity of some FKBPs, particularly FKBP13, therefore it is not specific to FKBP19, and binds FKBP19 200 times less than FKBP13 [12]. Other FKBP's inhibitors exist. They are FK506 structural analogs, but they also lack specificity for one particular member of the family [38]. FKBPs, and potentially FKBP19, are also known to have a chaperone activity, which seems to be independent of the PPIase activity [36]. FKBP19 could therefore play a chaperone role for proteins implicated in the regulation of Pax5, for example, or other targets that remain to be defined in future experiments.
Considering the complexity of SLE immunopathology, our results yield novel insights into the pathogenesis of the disease because they show that overexpression of a single gene, FKBP11, in B cells, is sufficient to induce some biological characteristics of SLE. The dissection of molecular pathways targeted by the FKBP19 protein, and the analysis of FKBP11 expression in larger cohorts of patients, looking for correlation between FKBP11 levels in B cells and clinical phenotypes of SLE patients, may lead to the identification of new molecular targets in SLE.

Patients
Patients with SLE fulfilling at least 4 diagnostic criteria according to the American College of Rheumatology [39] were prospectively included provided that they were in a quiescent phase of the disease (SLEDAI score less than 4) and received minimal treatment [no immunosuppressive drugs; if they needed steroids, the patients were not treated with more than 10 mg of prednisone per day (4 patients)]. 10 patients were treated with hydroxychloroquine [10]. Purified B cells from 17 patients (and 9 age-and sex-matched controls) were subjected to a pangenomic transcriptome analysis (Affymetrix GeneChip human genome U133 plus 2.0). This study was approved by the ethics committee of the Hôpitaux Universitaires de Strasbourg and patients gave their written informed consent.

Virus production and titration
Lentiviral particles were produced by transient transfection of 293T cells as described [40].
Mice In order to produce Fkbp11 lentigenic mice, 10 pL of lentivirus at 5 Â 10 8 IU/ml was injected into the perivitelline space of one-cell stage C57BL/6 mouse embryos (SEAT, Villejuif, France). Embryos were reimplanted into pseudopregnant females. One single-copy Fkbp11 founder (see following paragraph) was selected for further analysis. These Fkbp11 high transgenic mice were then maintained on C57BL/6 background. Progenies were screened by PCR on tail DNA, using following primers: Forward: 5 0 -CACCATGACCCTGCGCCCCTCACT-3 0 and Reverse: 5 0 -CCAGGCACAATC AGCATTGGTAGCT-3 0 . The PCR conditions were: 948C for 5 min; 30 cycles at 948C for 30 s, 558C for 30 s and 728C for 40 s; size of PCR product: 680bp). lpr/lpr Fkbp11 high mice were generated by crossing C57BL/ 6 lpr/lpr mice (B6.MRL-Fas lpr /J; The Jackson Laboratory) with Fkbp11 high mice and by intercrossing between the littermates. 56R mice on C57BL/6 background [16] were kindly provided by M. Weigert (Department of Pathology, University of Chicago, Chicago, IL, USA). 56R/Fkbp11 high mice were generated by crossing 56R mice with Fkbp11 high mice. In each experiment, littermate mice were used as controls. The experiments were conducted on both males and females, except for experiments on lpr/lpr Fkbp11 high mice and B6 lpr/lpr littermate controls, which was conducted exclusively on females. All animal experiments were performed in an animal facility, which is authorized by the ''Direction d epartementale des services v et erinaires'' (Strasbourg, France) and protocols were approved by the ''Comit e d' ethique en mati ere d'Experimentation Animale de Strasbourg'' (CREMEAS, approval number AL/02/15/09/11).

Cell preparation and culture
For activation of splenic mature B cells in vitro, spleen cells were plated at 1.10 6 cells/ml in culture medium composed of RPMI-1640 (Lonza) containing 10% (v/v) FCS (PAN), 50 mM b-Mercaptoethanol (Gibco, Paisley, Scotland), 1% Penicillin/Streptomycin (Gibco), 10mM HEPES (Lonza, Verviers, Belgium), 1 mM Sodium Pyruvate (Lonza). Cells were stimulated with LPS alone (10 mg/ml; Sigma), F(ab') 2 anti-mouse IgM (5 mg/ml; Jackson Immunoresearch), CG-50 (a large DNA fragment containing tandem repeats of hypomethylated CpG which is used to specifically stimulate anti-DNA B cell), or CG-Neg negative control (a large DNA fragment which is devoid of hypomethylated CpG) (0.2 ng/ml), for the indicated time. CG-50 and CG-Neg (HIV-CG À ) plasmids were kindly provided by Prof. Marshak-Rothstein, and were produced as previously described [18,19]. For plasmablast differentiation with LPS in vitro, splenic mature B cells (CD43 À ) were purified using B-cell isolation kit (anti-CD43 (Ly-48) microbeads) according to the supplier's protocol (Miltenyi Biotech, Auburn, CA, USA). B cells were then plated at 1.10 6 cells/mL and were stimulated for 4 days with LPS (10 mg/ml, Sigma) using the same medium as for total splenic cells culture. For knock-down experiments, the efficiency of anti-Fkbp11 siRNA (reference 102718384, Qiagen), compared to a control siRNA (reference 1027280, Qiagen) was tested in L929 murine fibroblasts, using HiPerFect transfection reagent (Qiagen, Hilden, Germany) according to Qiagen protocol specific for fibroblasts, at 24h and 48h after transfection with the siRNAs. For knock-down experiments in B cells, splenic mature B cells were sorted as explained above, and then were transfected with the anti-Fkbp11 siRNA or the control siRNA, using HiPerFect transfection reagent (Qiagen), according to Qiagen protocol specific for suspension cell lines, before adding LPS in the culture medium.

Vk light chain sequencing
Light chain high throughput sequencing was performed using a two-step PCR. In the first step, Vk-Jk1 and Vk-Jk5 rearrangements were amplified using cocktails of primers (Forward: Vk FW1 5 0 -GAY ATT GTG MTS ACM CAR WCT MCA-3 0 ; Reverse: Jk1 5 0 -ACG TTT GAT TTC CAG CTT GG-3 0 , Jk5 5 0 -ACG TTT CAG CTC CAG CTT G-3 0 ). In the second step, band purified amplicons were re-amplified using primers that were modified to include illumina sequencing barcodes and adaptors, using protocols available on the illumina website (http://supportres.illumina.com). Equimolar amounts of each sample library were subsequently sequenced on the Illumina Mi-seq using 2 by 250 base paired end reads. Sequences extending from the Jk end were analyzed using the IMGT high-throughput sequence analysis server and only unequivocally identified Vk genes were included in the analysis.

Flow cytometry analysis and cell sorting
Cell phenotype was performed on lymph nodes, splenic and bone marrow lymphoid populations by four-color fluorescence analysis according to standard protocols. The following antibodies and reagents were used: PerCP, PE, or APC anti-mouse B220, CD3, CD4, CD8, CD19, CD21, CD23, CD44, CD69, CD86, I-A/I-E, CD25, CD138, and IgM (all from BD Biosciences). Propidium iodide was used for live-dead discrimination. For proliferation analysis, cells were permeabilized after extracellular staining and fixed with the cytofix/cytoperm permeabilization kit (BD Biosciences), then stained with the PerCP Cy5.5 anti-Ki67 (BD Biosciences). Cells were analyzed using a FACSCalibur. We then analyzed data with FlowJo software (Treestar). Bone marrow and splenic B-cell subsets ( Supplementary  Fig. S4) on the one hand, and PB and undifferentiated B cells after 4 days of culture with LPS (Fig. 6) on the other hand, were sorted by flow cytometry using a FACS Aria cell sorter (BD Biosciences), after staining with the antibodies indicated in the legend of the figures.

Antibody detection by ELISA
Total IgG, IgG1, IgG2b, IgG3 or IgM levels were measured in serum from 8-month-old mice, and in supernatants after 3 days of stimulation, as previously described [11]. Antithyroglobulin, -dsDNA, -actin, and RF titers, as well as anti-OVA and anti-NP specific antibodies were measured as previously described [11].
Immunization Six-to eight-week-old mice were injected intraperitoneally at days 0, 10, and 20 with 100 mg NP-LPS (Biosearch Technologies). Serum was collected before each injection and at day 30. For immunization experiments with ovalbumin (OVA), six-to eight-week-old mice were injected intraperitoneally with 100 mg OVA (Sigma) in complete Freund's adjuvant (Sigma) at day 0, in incomplete Freund's adjuvant at day 10 and in PBS at day 20. Serum was collected before treatment, and at every time point before the injection.

Statistical analysis
Statistical significance was calculated with a two-tailed Mann & Whitney test or unpaired t test with Welch's correction, Wilcoxon match-pairs test, or Spearman correlation test, using Prism software (GraphPad). plasmids, and M. Duval, D. Lamon, D. Bock, I. Ghazouani and K. Sablon for excellent animal care.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web-site. Figure S1: Creation of Fkbp11 and control GFPþ lentigenic mice, and expression of Fkbp11 in GFPþ control mice. Figure S2: Basal activation of B and T cells is not modified in Fkbp11 high mice compared to control mice. Figure S3: Analysis of deletion and anergy in 56R/Fkbp11 high mice. Figure S4: Fkbp11 expression in B cell subpopulations in wild-type mice.