Generation of autoimmune-prone mice with a B cell–specific deletion of T-bet. B6.SLE1,2,3 mice (referred to hereafter as SLE mice) were used as a model of spontaneous lupus-like autoimmunity. These mice express intervals of chromosomes 1, 4, and 7 derived from NZM2410 animals on the C57BL/6 background. These genetic intervals have been shown to drive lupus-like disease since SLE, unlike B6, animals contain activated lymphocytes, autoantibodies, and develop glomerulonephritis, with a female bias (21–23). We confirmed that, like other mouse models of SLE, SLE mice accumulate T-bet+ ABCs (Figure 1 and ref. 7).

Figure 1 Generation of SLE × T-betfl/fl × CD19Cre/WT mice. Representative FACS plots and quantification of the frequency of T-bet+ B cells in spleens of 4-month-old C57BL/6, SLE × T-betfl/fl × CD19Cre/WT, and SLE × T-betfl/fl × CD19WT/WT littermate controls. Numbers represent the percentage of cells in the gates. B cells were gated as live, B220+CD19+CD4–CD8–. n = 4 mice per group. Data are presented as mean ± SEM and are representative of more than 5 independent experiments. *P < 0.05 by unpaired 2-tailed Student’s t test.

To study the effect of T-bet expression in B cells in autoimmunity, we intercrossed CD19Cre/WT and T-betfl/fl with SLE mice, generating SLE × T-betfl/fl × CD19Cre/WT mice. This strain contains all of the intervals (from chromosomes 1, 4, and 7) necessary for predisposition to autoimmunity and lacks T-bet expression in B cells. The successful transfer of all intervals specific for SLE mice was checked by PCR (see Methods). B cell–specific T-bet deletion in the SLE × T-betfl/fl × CD19Cre/WT mice was confirmed by both PCR and intracellular staining (Figure 1). We confirmed that the T-bet deletion was B cell specific and that T cells continued to express normal levels of T-bet (data not shown).

Thus, we have successfully generated autoimmune-prone mice with a B cell–specific T-bet deletion.

In these animals Cre is driven by the CD19 promoter with concomitant deletion of CD19; thus, B cells in such mice are heterozygous for expression of CD19. To check that this did not affect disease onset, we confirmed that autoimmunity occurred similarly in SLE × CD19WT/WT and SLE × CD19Cre/WT animals (data not shown), so SLE × T-betfl/fl × CD19WT/WT littermate controls were used for all experiments.

In the absence of T-bet in B cells, SLE mice demonstrate improved kidney function and better survival rates. First we asked whether T-bet expression in B cells was required for the development of kidney pathology, a definitive measure of organ pathology in lupus. First, we measured proteinuria levels in SLE mice with or without T-bet expression in B cells. SLE mice in our facility develop proteinuria by 7 months of age; therefore, we tested 7-month-old mice for the presence of proteinuria. The data indicate that the proteinuria score was significantly reduced in the absence of T-bet–expressing B cells (Figure 2A). The mice were subsequently tested again until they were 12 months old. Only one additional SLE × T-betfl/fl × CD19Cre/WT mouse (out of 9) developed proteinuria during this time frame, indicating that development of proteinuria was prevented rather than delayed in SLE mice with B cell–specific T-bet deletion (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI91250DS1).

Figure 2 B cell–specific T-bet expression is responsible for the development of kidney pathology and rapid mortality in lupus-like autoimmunity. (A) Proteinuria scores of 7-month-old mice (1 = trace, 2 = 30 mg/dl, 3 = 100 mg/dl, 4 ≥ 500 mg/dl). (B–D) Kidney sections from 12-month-old mice were stained with periodic acid–Schiff from SLE × T-betfl/fl × CD19Cre/WT (B), and SLE × T-betfl/fl × CD19WT/WT (C), and C57BL/6 (D) mice. A representative glomerulus is shown in each section. Arrowheads indicate glomeruli. Hypercellularity and endocapillary proliferation were seen in most glomeruli from the SLE × T-betfl/fl × CD19WT/WT mice. Original magnification, ×400. (E) Twenty-five glomeruli for each mouse were evaluated in a blinded analysis in kidney sections. The percentage of affected glomeruli is shown for each individual mouse. (F–H) Five kidneys of 12-month-old mice per genotype were analyzed by immunofluorescence histology (IFH). Representative IFH staining of kidney glomeruli stained with anti-IgG (red) and anti-C3 (green) is shown for: SLE × T-betfl/fl × CD19Cre/WT (F), SLE × T-betfl/fl × CD19WT/WT (G), and C57BL/6 (H) mice. Scale bars: 50 μm. (I) Survival of SLE × T-betfl/fl × CD19Cre/WT, and SLE × T-betfl/fl × CD19WT/WT littermate control mice was followed over the same 12 months (n = 10 mice per group). Comparison of the survival curves and statistical analysis were performed using Prism software. *P < 0.05. Significance determined by Student’s t test (A and E) or Gehan-Berslow-Wilcoxon test (I).

To further evaluate kidney disease in the mice, the kidney glomeruli of SLE × T-betfl/fl × CD19Cre/WT and SLE × T-betfl/fl × CD19WT/WT littermate controls were evaluated by a blinded expert observer. The kidneys were obtained at the time of sacrifice when the mice were 12 months old, or earlier if the mouse exhibited high proteinuria and/or lost 10% of body weight. The results, shown in Figure 2, B–E, indicate that approximately 80% of glomeruli demonstrated hypercellularity and endocapillary proliferation in SLE × T-betfl/fl × CD19WT/WT littermate controls. In contrast, less than 30% of the glomeruli were affected in the SLE × T-betfl/fl × CD19Cre/WT mice.

To confirm that the kidney pathology was associated with immune complex deposition, immunofluorescence histology was performed. Kidney sections were stained with anti-IgG and anti-C3 antibodies to detect immune complex deposition. Our results indicate that, in the absence of T-bet expression in B cells, SLE mice contained less immune complex formation in the kidneys since the size and/or intensity of IgG/C3 stained glomeruli were significantly reduced (Figure 2, F–H).

Together, these data indicate that T-bet expression in B cells plays a critical role for the development of kidney pathology during lupus-like autoimmunity.

Improved survival of SLE mice with B cell–specific T-bet deletion. Since kidney failure is the major cause of death of SLE mice, we asked whether the improved kidney function in SLE × T-betfl/fl × CD19Cre/WT animals also affected their mortality rates. As demonstrated in Figure 2I, the survival of SLE mice was dramatically improved in the absence of T-bet expression in B cells when compared with littermate controls. Only 25% of T-bet–sufficient SLE mice survived until 12 months of age when, in contrast, 90% of SLE mice with B cell–specific T-bet deletion survived until the same age (Figure 2I). This finding demonstrates that T-bet expression in B cells plays a critical role in the development and the consequences of lupus-like autoimmunity.

Overall our data demonstrate that T-bet expression in B cells is essential for the development of kidney pathology and rapid mortality in SLE mice.

B cell–specific T-bet deletion leads to reduced titers of serum IgG2a in SLE mice. Next we asked which changes in the immune system lead to the improved kidney function and reduced mortality in SLE mice that lack B cell–intrinsic T-bet. Multiple groups have previously reported that T-bet expression in B cells is critical for switching B cell isotype expression to IgG2a/c (described herein as IgG2a) (24–26). Therefore, we measured the presence of different IgG subclasses in the sera of SLE × T-betfl/fl × CD19Cre/WT mice compared with SLE × T-betfl/fl × CD19WT/WT littermate controls. Figure 3A demonstrates that the serum concentrations of total IgG were comparable between the experimental and control animals; however, the isotype distribution was altered in the absence of T-bet expression in B cells. In particular, as predicted by previous reports (25–27), in the absence of T-bet+ B cells there was a significant reduction in IgG2a levels, such that it decreased to levels seen in normal C57BL/6 animals. This was accompanied by increased levels of IgG2b and IgG1 in SLE × T-betfl/fl × CD19Cre/WT mice compared with their Cre-negative littermate controls.

Figure 3 B cell–specific T-bet deletion delays the appearance of autoantibodies in SLE mice. (A) Concentrations of total or different subclasses of IgG in the serum of 4-month-old C57BL/6 or SLE × T-betfl/fl × CD19Cre/WT and SLE × T-betfl/fl × CD19WT/WT littermate controls (n = 5). Data are presented as mean ± SEM. *P < 0.05 by 1-way ANOVA followed by Newman-Keuls analysis. (B and C) Titers of anti-chromatin total IgG (B) or IgG2a (C) in the serum of SLE × T-betfl/fl × CD19Cre/WT (red) and SLE × T-betfl/fl × CD19WT/WT (black) littermate controls at different ages as indicated (n = 10–12 mice per group). P values for B and C were calculated using 2-way ANOVA. **P < 0.01, ***P < 0.001. NS, not significant.

B cell–specific T-bet deletion leads to reduction of autoantibodies in SLE mice. The presence of high titers of autoantibodies is essential for the development of kidney failure during the development of lupus-like autoimmunity. Therefore, next we evaluated how T-bet expression in B cells affected the development of autoantibodies. We monitored the appearance, over time, of serum anti-chromatin IgG levels in the SLE mice in the absence or presence of T-bet expression in B cells. The appearance of anti-chromatin total IgG autoantibodies (Figure 3B) and of anti-chromatin IgG2a in particular (Figure 3C) was significantly delayed in the absence of T-bet expression in B cells, when compared with littermate controls. The difference was most marked when the mice were 5 months old. However, later in life, the difference between the SLE × T-betfl/fl × CD19Cre/WT and SLE × T-betfl/fl × CD19WT/WT control mice lessened, although the titers in T-bet–deficient animals were still significantly lower than those in mice containing T-bet+ B cells. This suggests that another transcription factor (or a combination of transcription factors) can compensate for the lack of T-bet expression in B cells, leading to the activation of autoreactive B cells and their production of autoantibodies. One of the obvious candidates for such a transcription factor is EOMES, which has been reported to compensate for lack of T-bet expression in T cells (28). However, we were unable to detect EOMES expression in T-bet–deficient B cells (data not shown), suggesting the involvement of some other yet unknown transcription factor.

B cell–specific T-bet deletion affects both B and T cell compartments during autoimmune responses. To find out why the absence of T-bet expression in B cells led to the delayed appearance of autoantibodies and reduced kidney damage, we compared the B cell compartments of SLE × T-betfl/fl × CD19Cre/WT and SLE × T-betfl/fl × CD19WT/WT littermate control mice. Our data indicate that the absence of T-bet expression in B cells correlated with a reduction in frequencies and numbers of CD11c+ B cells (Figure 4A). Surprisingly, we detected significantly lower frequencies and numbers of GC B cells in mice with a B cell–specific deletion of T-bet (Figure 4B).

Figure 4 Reduced appearance of CD11c+ ABCs, GC B cells, and early plasmablasts in SLE mice with B cell–specific T-bet deletion. Representative FACS plots and quantification of the frequency of (A) CD11c+ B cells, (B) GC B cells, and (C) early plasmablasts in the spleens of 4-month-old C57BL/6 (white bars), SLE × T-betfl/fl × CD19Cre/WT (red bars), and SLE × T-betfl/fl × CD19WT/WT (black bars) mice. B cells were gated as live, B220+CD19+CD4–CD8–. Bar graphs represent mean ± SEM (n = 4 mice per group, representative of 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA followed by Newman-Keuls analysis. NS, not significant.

Our data also demonstrate significant reduction in frequencies and numbers of early plasmablasts in SLE × T-betfl/fl × CD19Cre/WT mice compared with littermate controls (Figure 4C).

Together these data indicate that, in the absence of their T-bet expression, B cells fail to facilitate the formation of GCs during spontaneous autoimmune responses and do not differentiate into plasmablasts, a failure that probably results in decreased autoantibody production.

Antigen (Ag) presentation by B cells plays an important role in spontaneous T cell activation during autoimmune responses (29). Moreover, it has been previously demonstrated that T-bet+ ABCs are potent Ag-presenting cells (14); therefore, the absence of T-bet expression in B cells might affect T cell activation in autoimmunity. To test this hypothesis, we evaluated the T cell compartments in SLE mice in the presence or absence of T-bet expression in B cells. Our data indicate that there was a significant reduction in the percentage and numbers of activated/memory CD4+ T cells but not of CD8+ T cells in the absence of T-bet–expressing B cells (Figure 5, A and B). On the other hand, the percentages and numbers of IFN-γ–producing CD4+ T cells were not affected by the absence of T-bet–expressing B cells, whereas the percentages (but not absolute numbers) of IFN-γ+CD8+ cells did drop significantly (Figure 5, C and D).

Figure 5 Reduced T cell activation in SLE mice with B cell–specific T-bet deletion. Representative FACS plots and quantification of the frequency of (A) activated/memory CD4+ T cells, (B) activated/memory CD8+ T cells, (C) IFN-γ+ CD4+ T cells, and (D) IFN-γ+ CD8+ T cells in the spleens of 4-month-old C57BL/6 (white bars), SLE × T-betfl/fl × CD19Cre/WT (red bars), and SLE × T-betfl/fl × CD19WT/WT (black bars) mice. T cells were gated as live, B220–, CD19–, CD4+ (or CD4–), CD8+ (or CD8–). Bar graphs represent mean ± SEM (n = 4 mice per group, representative of 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA followed by Newman-Keuls analysis. NS, not significant.

GC B cells, preplasmablasts, and activated T cells do appear in aged SLE × T-betfl/fl × CD19Cre/WT mice (>7 months old) (Supplemental Figure 2), which is in line with our observation of the delayed appearance of autoantibodies in these mice (Figure 3), suggesting that the autoimmune response is substantially delayed in SLE mice in the absence of T-bet–expressing B cells.

Overall, these data indicate that T-bet expression in B cells during spontaneous autoimmune responses is critical for B cell activation and generation of GC B cells and early plasmablasts (CD44+CD138+). Moreover, T-bet expression in B cells is also required for the efficient T cell activation during an autoimmune response, perhaps via its effect on B cell Ag-presenting abilities. These changes in B and T cell activation ultimately lead to the reduced titers of autoantibodies and reduction in immune complex formation and deposition in kidneys, which in turn improves the survival of SLE mice.

B cell–intrinsic T-bet expression is critical for the development of spontaneous GCs during autoimmunity. The low frequency of GC B cells in SLE × T-betfl/fl × CD19Cre/WT mice (Figure 4B) suggests that the animals might have impaired formation of spontaneous GCs. To test this, we confirmed by histological analysis of spleens that SLE × T-betfl/fl × CD19Cre/WT mice exhibited a defect in the formation of spontaneous GCs when compared with littermate controls (Figure 6, A–C).

Figure 6 T-bet expression in B cells is required for the formation of spontaneous GCs, but is dispensable for the formation GCs following deliberate immunization. (A–C) Representative immunofluorescent staining of 5 spleen sections obtained from (A) SLE × T-betfl/fl × CD19Cre/WT, (B) SLE × T-betfl/fl × CD19WT/WT, or (C) C57BL/6 mice. (D) Percentage of GC B cells identified by FACS as B220+CD19+CD4–CD8–GL7+CD95+ and (E–G) representative immunofluorescent staining of spleens sections obtained from (E) T-betfl/fl × CD19Cre/WT, (F) T-betfl/fl × CD19WT/WT mice immunized with NP-CGG and alum (day 11) (n = 4 mice per group), or (G) naive C57BL/6 mice. n = 4 mice per group, representative of 3 independent experiments. Scale bars: 100 μm. *P < 0.05, **P < 0.01 by 1-way ANOVA followed by Newman-Keuls analysis. NS, not significant; PNA, peanut agglutinin.

A requirement for T-bet expression in B cells in the formation of GCs was unexpected. Therefore, we wondered whether this requirement was common to all circumstances leading to the creation of GCs, or applied only to their appearance in spontaneous autoimmunity. To answer this question we immunized T-betfl/fl × CD19Cre/WT and T-betfl/fl × CD19WT/WT mice with a conventional protocol involving nitrophenylated chicken γ globulin (NP-CGG) with alum, and analyzed for the presence of GCs in spleens 10 days after immunization. As demonstrated in Figure 6, the absence of T-bet expression in B cells did not affect the appearance of GC B cells (Figure 6D) or GCs (Figure 6, E–G) in response to deliberate immunization with Ag plus an adjuvant.

Together, these data indicate that T-bet expression in B cells is required for the formation of spontaneous GCs during autoimmune responses, but is dispensable for the formation of GCs in response to deliberate immunization, at least with an alum-adjuvanted Ag. These findings are in line with recent reports indicating that IFN-γR expression in B cells is required for the formation of spontaneous GCs during autoimmunity (30, 31) and indicate differential requirements for spontaneous and deliberate GC formation. This finding should be further examined in the future.

T-bet expression in B cells is required for the appearance of autoantibodies in Mer–/– and B6.Nba2 mice. Our data concern the role of T-bet+ B cells in the SLE model of lupus-like disease. However, other mouse models for this malady exist, so we asked whether the effects we observed were confined to SLE mice or were also apparent in other mouse models of SLE. To approach this question we used 2 other mouse strains that are known to develop lupus-like autoantibodies: B6 mice lacking MerTK (Mer–/– mice), which have a defect in the clearance of apoptotic cells and generate autoantibodies with age (32), and B6.Nba2 animals (referred to hereafter as Nba2 mice), which express the autoimmune-predisposing chromosome 1 locus of NZB animals (33). Both Mer–/– and Nba2 mice produce autoantibodies, but do not develop kidney pathology (32, 34). We intercrossed these mice with T-betfl/fl × CD19Cre/WT mice, generating Mer–/– × T-betfl/fl × CD19Cre/WT or Nba2 × T-betfl/fl × CD19Cre/WT mice with B cell–specific T-bet deletions.

As demonstrated in Figure 7A, Mer–/– animals with a B cell–specific T-bet deletion had significantly decreased titers of anti-chromatin antibodies in comparison with their Mer–/– littermate controls that contained T-bet+ B cells. B cell–specific T-bet deletion in Mer–/– mice also resulted in a reduced frequency of GC B cells (Figure 7B). The appearance of spontaneous GCs in such animals was also significantly reduced (data not shown). Similar effects on autoantibody production were detected in Nba2 × T-betfl/fl × CD19Cre/WT mice (Figure 7C).

Figure 7 Mer–/– and Nba2 mice with B cell–specific T-bet deletion exhibit reduced titers of autoantibodies, and reduced frequency of CD11c+ and GC B cells. (A) Titers of anti-chromatin IgG in the serum of Mer–/– and Mer–/– × T-betfl/fl × CD19Cre/WT mice at different ages (as indicated) assessed by ELISA (n = 10 mice per group). (B) Quantification of frequency of T-bet+CD11c+ B cells, GC B cells, and early plasmablasts in 4-month-old Mer–/– × T-betfl/fl × CD19Cre/WT versus Mer–/– mice. Bar graphs represent mean ± SEM (n = 4 mice per group, representative of 3 independent experiments). (C) Titers of anti-chromatin IgG and IgG2a in the serum of Nba2 and Nba2 × T-betfl/fl × CD19Cre/WT mice assessed by ELISA (n = 10 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA followed by Newman-Keuls analysis. NS, not significant.

Since, as mentioned above, Mer–/– and Nba2 mice do not develop kidney pathology, we could not assess the effects of B cell–intrinsic T-bet deletion on the development of these clinical features using these models of SLE. Nevertheless, these data show that our findings about the need for T-bet+ B cells for the appearance of autoimmune symptoms in SLE lupus-like disease is extendable to 2 other models of spontaneous autoantibody production. In combination with the recent discoveries of similar B cells in human patients (16, 17), discussed below, our data suggest that ABCs may be targets of interest in treatment of such diseases.