Overexpression of TLR7 promotes cell-intrinsic expansion and autoantibody production by transitional T1 B cells

To identify which stages of splenic B cell development are affected by TLR7 overexpression, we first compared B cell subset distribution between WT B6 mice and TLR7.1Tg mice, which express 8–16-fold higher TLR7 mRNA compared with WT (Deane et al., 2007). 8–14-wk-old TLR7.1Tg mice had a significant (approximately threefold) increase in the total number of splenic B cells compared with control mice (Fig. 1 A), although the frequency of B220⁺ B cells was decreased due to the expansion of myeloid cells (Deane et al., 2007; Buechler et al., 2013; Sun et al., 2013). Further flow cytometric analyses showed a marked increase in the percentage of CD19⁺B220⁺CD93⁺ (AA4.1) B cells, suggesting an expansion of the immature B cells (Fig.1, B and C). Using a panel of surface markers, we confirmed that the expanded cells were in fact B220⁺CD24^hiCD21^lo T1 B cells (Fig.1 D). The percentage of T1 cells in TLR7.1Tg mice increased nearly threefold, whereas the percentages of T2 (B220⁺CD24^hiCD21^hi/intCD23^hi) and FO (B220⁺CD24^intCD21^int) subsets of total splenic B cells were significantly decreased (Fig.1 E). Actual numbers of T1, T2, and FO B cells increased, with the most significant increases in T1 and FO B cells (Fig.1 F). In accordance with Deane et al. (2007), we also found that the proportion and number of MZ B cells (defined as B220⁺CD24^hiCD23⁻CD21^hi) were significantly reduced in TLR7.1Tg mice (Fig. 1, D–F). Consistent with the flow cytometric data, immunohistological analyses showed significant accumulation of CD93^brightIgD^lowCD138^neg T1-like B cells in the RP areas of the TLR7.1Tg but not the WT mouse spleens (Fig. 1 G).

To determine whether peripheral B cells in TLR7.1Tg mice differed in their Tlr7 expression levels, we performed RT-PCR analyses of sorted B cell T1, T2, FO, and MZ B cell populations. We found that Tlr7 mRNA levels were significantly elevated (9–18-fold) in all B cell subsets of TLR7.1Tg mice, as compared with WT controls, but there was no statistically significant difference between the subsets (Fig. 1 H).

Together, these results showed that Tlr7 is overexpressed in all splenic B cell subsets of TLR7.1Tg mice. T1 and FO B cell subset numbers were the most significantly expanded, whereas MZ B cells decreased significantly. The expansion of T1 cells was associated with marked accumulation of these cells in the splenic RP.

To further understand the basis for the accumulation of T1, T2, and FO B cells in TLR7.1Tg mice, we assessed cell proliferation by a BrdU pulse in vivo, using relatively young (8–12 wk old) TLR7.1Tg and B6 control mice. We found a strikingly higher percentage of BrdU⁺ cells within the T1 B cell subset (≥15%) of TLR7.1Tg mice compared with control B6 mice (≤4.5%; Fig. 2, A and B). A small increase of the percentage of BrdU⁺ cells was also noticeable within the T2 B cell subset; however, little or no BrdU⁺ cells were detectable within the FO or MZ B cell populations (Fig. 2 B). The proliferation of T1 B cells was also supported by histological analyses, which revealed a significant accumulation of BrdU⁺ cells within the splenic RP of TLR7.1Tg mice (Fig. 2 C) where T1 B cells reside (Fig. 1 G; Loder et al., 1999).

To test whether the increased cell proliferation was due to an intrinsic change in T1 B cells overexpressing TLR7, we generated mixed BM chimeric mice by reconstituting lethally irradiated B6 × B6.SJL F1 mice with an equal mix of congenically marked B6.SJL (Ly5.1⁺) and TLR7.1Tg (Ly5.2⁺) donor BM cells. 3 mo after reconstitution, TLR7.1:B6.SJL chimeric mice displayed many of the characteristics of TLR7.1Tg mice, including the development of splenomegaly and myeloid cell expansion (Buechler et al., 2013). Flow cytometric analyses of B cell subsets in mixed BM chimeras revealed a preferential expansion of TLR7.1Tg-derived T1 cells (Fig. 2, D and E). Furthermore, TLR7.1:B6.SJL chimeric mice showed a two- to threefold increase in the percentage of proliferating (BrdU⁺) TLR7.1Tg T1 cells, compared with B6.SJL T1 cells. No significant differences in cell proliferation were observed in either Ly5.1⁺ or Ly5.2⁺ T1 cells in the B6:B6.SJL control chimeras (unpublished data). The effects of TLR7 overexpression in promoting B cell proliferation was selective to the TLR7.1Tg T1 B cells, as we detected no significant differences in proliferation within the T2/MZ or FO B cell subsets from Ly5.2 TLR7.1Tg donors (Fig. 2 F).

To determine whether the proliferation of T1 B cells in TLR7.1Tg was a consequence of TLR7 overexpression by itself or interaction with self-antigens, we crossed TLR7.1Tg mice with hen egg lysozyme (HEL)–specific VDJ9κ5 mice to create TLR7.1Tg mice in which a high proportion of B cells (15–35%) express BCRs specific for HEL (HEL⁺ cells). Analyses of TLR7.1Tg × HEL mice showed lower percentages of T1 cells within the HEL-binding B220⁺ B cell population as compared with B cells that did not bind HEL (HEL⁻ cells; Fig. 2 G). Furthermore, HEL⁺ T1 B cells displayed significantly less BrdU uptake compared with HEL⁻ T1 B cells (Fig. 2 H). Therefore, TLR7 overexpression did not affect the expansion of B cells specific for exogenous antigen. These results suggest that BCR-mediated signaling, presumably via endogenous auto-Ags, is required for T1 B cell expansion.

As TLR signaling has been also implicated in negative selection of autoreactive B cells (Isnardi et al., 2008), we next sought to determine if TLR7 overexpression affected the percentage of cell death in different B cell subsets. Consistent with the notion that the transition from T1 to T2/FO stage represents an important checkpoint for the elimination of potentially autoreactive B cells (Yurasov et al., 2005; Giltiay et al., 2012), we observed an increase in the percentage of Annexin V⁺ T1 cells in both WT and TLR7.1Tg mice compared with other peripheral B cell subsets (Fig. 2 H). The rate of apoptosis was significantly higher in TLR7.1Tg T1 cells compared with the WT T1 cells.

Collectively, these data show that TLR7 overexpression results in a striking increase in both proliferation and cell death of T1 B cells in the RP of the spleen. Because there is an overall expansion of these cells, the rate of proliferation and/or survival must exceed the rate of cell death. This hyper-proliferative phenotype is intrinsic to TLR7Tg T1 B cells but is not present in HEL-specific T1 B cells, strongly supporting the hypothesis that an endogenous antigen is driving their activation.

Because we observed that RP T1 B cells in TLR7.1Tg mice proliferate at high rates in vivo, we asked whether T1 cell subpopulation also contains activated cells with characteristics of plasmablasts. Flow cytometric analysis revealed an increase in the percentage of B220⁺CD138⁺CD44^hi cells, suggesting the presence of Ag-activated (plasmablast) cells. However, these cells constituted a small proportion (∼1%) of the T1 B cells in the TLR7.1Tg mice. Similar percentages of CD138⁺CD44^hi cells were also present within the TLR7.1 FO B cell subset (Fig. 3, A and B).

To further test if TLR7.1Tg T1 and FO cells can proliferate and differentiate into Ab-producing cells, we purified T1 and FO cells from WT or TLR7.1Tg mice (Fig. S1), loaded them with CFSE, and stimulated in vitro with TLR7 agonist (R848), anti-IgM, or SmRNP, a natural RNA-containing ligand. Consistent with the observation that RNA antigens drive the excessive proliferation of T1 cells in vivo, TLR7.1Tg T1 B cells activated by R848 showed a significant increase in the percentage of live cell (scattered gated) and proliferating (CFSE^lo) cells compared with WT T1 cells (Fig. 3 C). We also observed an increased survival and proliferation of TLR7.1Tg T1 cells in response to anti-IgM, whereas less striking differences in survival and proliferation between TLR7.1Tg and WT T1 cell cultures were seen when stimulated by the combination of anti-IgM and R848. Similar to TLR7.1Tg T1 cells, TLR7.1Tg FO B cells were also more responsive to R848 stimulation (Fig. 3 D) but actually responded less to anti-IgM stimulation. Importantly, we found that a portion of TLR7.1Tg T1 cells, but not WT T1 B cells, proliferated in response to stimulation with SmRNP. No responses to SmRNP stimulation were evident in either TLR7.1Tg or WT FO B cell samples.

Because TLR7 and/or BCR stimulation led to cell proliferation rather than death in TLR7.1Tg T1 B cells, we asked whether the in vitro activation of TLR7.1 T1 B cells was also associated with increased immunoglobulin secretion. We observed higher concentrations of both IgM and IgG in nonstimulated cultures of TL7.1Tg T1 B cells and an approximately twofold increase in IgG concentration after activation with R848 or anti-IgM plus R848 (Fig. 4 A). Small but detectable levels of IgG were also found in cultures from TLR7.1Tg T1 cells stimulated with SmRNP. Under the same experimental conditions, sorted FO B cells from TLR7.1Tg mice also produced increased levels of IgG, but at significantly lower levels compared with T1 B cells (Fig. 4 A). Furthermore, although the T1 and FO populations contained similar percentages of CD44^hiCD138⁺ cells (Fig. 3 B), the stimulated T1 cells produced significantly more IgG than the FO cells (Fig. 4 A). In addition, sorted TLR7 Tg T1 cells, from which we removed CD138⁺ cells from the T1 gate, still proliferated at higher rates and produced higher titers of IgG then did the WT T1 cells (unpublished data). Together, these results indicate that the high IgG production by T1 cells was not simply due to CD138⁺ cells.

Further analysis showed that T1 cells predominantly produced IgG2c and IgG2b subclasses (Fig. 4 B). Consistent with these in vitro findings, we detected increased frequencies of IgG2b⁺ and IgG2c⁺ cells within the splenic RP in TLR7.1Tg mice localized in the same area as the T1 cells (Fig. 4 C). Importantly, analysis of IgG produced by TLR7.1Tg T1 cells treated with anti-IgM, R848, or SmRNP also showed reactivity against RNA (Fig. 4 D). ANA test further confirmed the presence of anti-nuclear antibodies in culture supernatants from TLR7Tg T1 cells, stimulated with R848 or anti-IgM plus R848 (Fig. 4 E). The above results indicate that a portion of TLR7.1Tg T1 cells are hyper-responsive to both TLR7 and BCR engagement, resulting in increased survival, proliferation, and secretion of class-switched Abs, including the production of anti-RNA Abs.

In previous studies, T1 B cells were shown to constitutively express AID and respond rapidly to TLR9 and TLR3 stimulation (Han et al., 2007; Ueda et al., 2007). In addition, the transcription factor T-bet is induced in B cells upon activation through TLR7 or TLR9 and mediates IgG2a class-switch recombination (Peng et al., 2002; Berland et al., 2006). Given the surprising predominance of IgG2c and IgG2b Abs produced by purified TLR7.1Tg T1 cells, we tested if AID and/or T-bet might be induced downstream of TLR7. Expression levels of AID encoding gene (Aicda) in freshly isolated TLR7.1Tg T1 was somewhat higher compared with B6 T1 cells, but differences were not statistically significant (Fig. 4 F). Importantly, consistent with Ueda et al. (2007) and Han et al. (2007), T1 cells were the only B cell subset with measurable levels of Aicda mRNA, suggesting that unstimulated T1 cells have the capacity to produce class-switched Abs. Further analysis of TLR7.1Tg and WT T1, T2, FO, and MZ B cells revealed that TLR7.1Tg B cells have a 2.5–5-fold increase in the expression of T-bet encoding gene (Tbx21) compared with WT counterparts (Fig. 4 G). Together, these results demonstrate that additional copies of TLR7 in T1 B cells lead to enhanced expression of AID and T-bet and increased production of IgG2b and IgG2c in response to TLR7 stimulation in vitro (Fig. 4 B).

ssRNA is a ligand for TLR7, yet it is not known where in the spleen B cells encounter RNA that might be free or part of RNPs released from dying cells. To test the role of extracellular RNA as the ligand stimulating TLR7.1Tg B cells, we crossed TLR7.1Tg mice with mice that express bovine RNase A under the human albumin promoter (RNaseTg mice). RNase Tg mice display 5–10-fold higher serum RNase A activity (Sun et al., 2013) but show no alterations in their spleen weight and B cell numbers in the spleen compared with WT mice (Fig. 5 A).

Overexpression of RNase in TLR7.1 × RNase dTg mice led to a reduction in spleen size, which was associated with a modest but not statistically significant reduction in the number of total B cells (Fig. 5 A). However, flow cytometric analyses of individual B cell subsets in TLR7.1 × RNase dTg mice revealed a significant (2–2.5-fold) reduction in the numbers of T1 B cells in the TLR7.1 × RNase dTg mice compared with TLR7.1Tg controls (Fig. 5, B and C). The number of FO B cells was also decreased in TLR7.1 × RNase dTg mice but not as much as the decrease in T1 B cells. In contrast, RNase A overexpression had little effect on MZ B cell numbers, as both TLR7.1 and TLR7.1 × RNase dTg mice had reduced levels of MZ B cells.

The increased expression of RNase A not only reduced TLR7.1Tg T1 B cell numbers but also affected their ability to proliferate in vivo. WT, RNaseTg, TLR7.1Tg, and TLR7.1 × RNase dTg mice were pulsed with BrdU, and BrdU incorporation in different B cell subsets was analyzed as in Fig. 2. The percentage of BrdU⁺ T1 cells in TLR7.1 × RNase dTg mice was significantly reduced compared with the TLR7.1Tg mice (Fig. 5, D and E). No substantial cell proliferation was observed in other B cell subsets. The overexpression of RNase A also significantly limited the accumulation of plasma cells in the TLR7.1Tg mice (Fig. 5, F and G), suggesting that endogenous RNA is required for the generation of Ab-producing cells in TLR7.1Tg mice.

Based on these results, we conclude that the in vivo proliferation and expansion of TLR7.1Tg T1 B cells is driven by endogenous, RNA-containing Ag. In contrast, the reduction of MZ B cells in TLR7.1Tg mice was not significantly affected by the presence of RNase, suggesting that T1 B cell and MZ B cell subsets in TLR7.1Tg mice are regulated by different mechanisms.

The data above revealed that the activation and expansion of T1 B cells is due to cell-intrinsic expression of TLR7 and is driven by endogenous RNA. This activation of T1 cells results in cell proliferation and production of class-switched Abs. Previous studies in mice have postulated the requirement of IFN receptor (IFNαR) for B cell responsiveness to TLR agonists (Green et al., 2009). To test the potential role of type I IFN signaling in the activation of T1 cells in the TLR7.1Tg mice, we crossed the TLR7.1Tg mice to mice deficient in type I IFNαR, thus generating TLR7.1Tg × IFNαR-KO mice. 3–6-mo-old TLR7.1Tg × IFNαR-KO mice had significantly reduced splenomegaly compared with TLR7.1Tg mice; however, the weights of their spleens remained significantly higher than those of WT or IFNαR-KO control mice (Buechler et al., 2013). The reduced splenomegaly in TLR7.1Tg × IFNαR-KO mice correlated with a decrease in the number of myeloid cells (Buechler et al., 2013); however, IFNαR deficiency had no significant effect on the expansion of splenic B cells driven by TLR7 overexpression (Fig. 6 A). Consistent with the lack of reduction in B cell numbers, TLR7.1Tg × IFNαR-KO mice showed only a slight but not statistically significant decrease in the numbers of T1, T2, and FO B cells (Fig. 6 A). There was also no significant decrease in the number of plasma cells in TLR7.1Tg × IFNαR-KO mice compared with TLR7.1Tg mice (unpublished data). Importantly, T1 and FO B cell isolated from TLR7.1Tg × IFNαR-KO mice proliferated at significantly higher rates in response to R848 stimulation, compared with their IFNαR-KO counterparts (Fig. 6 B).

In spite of the lack of an effect on the numbers of T1 and FO B cell subsets, the absence of IFNαR led to a significant increase in the percentage (four- to fivefold) and the number (twofold) of MZ (B220⁺CD21^hiCD1d^hi) B cells in TLR7.1Tg × IFNαR-KO mice, compared with the TLR7.1Tg mice (Fig. 6, C and D). Thus, IFNαR deficiency did not abolish the effects of TLR7 overexpression on the expansion of T1, FO, and plasma cells; intact IFNαR signals, however, were required for TLR7-mediated loss of MZ B cells.

The discovery that TLR7 Tg mice have increased spontaneous proliferation and expansion of T1 B cells in vivo suggested that TLR7 overexpression promoted the activation of a portion of T1 cells (most likely specific for RNA), leading to the production of class-switched Abs. To test whether B cells with specificity for RNA can be stimulated to proliferate and produce Ab at the T1 stage, we used 564Igi^+/− knockin mice, which develop an SLE-like disease through a TLR7-dependent process and in which a high proportion of B cells (5–30%) express BCRs specific for ssRNA (Berland et al., 2006). Flow cytometric analyses of spleen cells from 564Igi^+/− mice revealed a marked increase in the percentage of B220⁺CD24^hiCD21^lo T1 B cells, which predominantly consisted of 564Id⁺ B cells (Fig.7, A–C). Further phenotypic analysis showed that Id⁺ B cells expressed high levels of CD93, low levels of IgD, and low-to-intermediate levels of CD23. As previously reported by Berland et al. (2006), these cells showed relatively low expression of surface IgM (Fig. 7 D). Also consistent with previous reports, we found that H564 Id⁺ cells were excluded from the B cell follicles and MZs and occupy in two major locations: the splenic RP and at the T–B cell borders (Fig. 7 E). Despite the seemingly immature phenotype of Id⁺ cells, 564Igi^+/− mice spontaneously developed class-switched Id⁺ Ab (Fig. 7 F; Berland et al., 2006), suggesting that Id⁺ cells can become activated and differentiate into Ab-producing cells.

To determine whether CD93⁺H564Id⁺ cells can be stimulated to produce Abs in vitro and to compare them with more mature Id⁺ cells, we purified Id⁺CD93^hi and Id⁺CD93^lo cells (Fig. 7 G), stimulated them with a TLR7 agonist (R848), and then measured antibody production in culture supernatants. The Id⁺CD93^hi cells produced significantly higher titer IgG2a and IgG2b Abs compared with Id⁺CD93^lo cells (Fig.7 H). We concluded that the majority of anti-RNA–specific cells in H564Igi mice express a transitional B cell phenotype and are capable of producing class-switched Abs in response to TLR7 stimulation.

Although TLR7 is known to play a pivotal role in autoantibody production in murine lupus (Berland et al., 2006; Christensen et al., 2006; Pisitkun et al., 2006; Santiago-Raber et al., 2010b), where and how TLR7 regulates the development of autoreactive B cells in the periphery is not clear. In this study, we found that increased expression of TLR7 led to a cell-intrinsic expansion of T1 B cells, which was driven by RNA in the RP of the spleen. Furthermore, the in vitro production of class-switched IgG antibodies by T1 B cells suggests that they may be important source of anti-RNA autoantibodies.

The discovery that TLR7.1Tg mice have increased spontaneous proliferation and expansion of T1 B cells in vivo suggested that TLR7 overexpression promoted the activation of a portion of T1 cells, which carry BCRs specific for endogenous Ags, most likely ssRNA or Ags associated with ssRNA. This hypothesis is strongly supported by the fact that the removal of RNA-containing material by RNase overexpression significantly limited both the expansion and proliferation of T1 cells. BM chimera experiments revealed that the increased proliferation of T1 cells was cell-intrinsic and the failure of HEL-specific T1 B cells from TLR7.1Tg × HEL mice to show increased proliferation demonstrates that the results cannot be explained by TLR7 overexpression alone. In addition, even though the TLR7 Tg mice carry a high copy number of TLR7, the fact that Unc93B1 D34A mutant mice phenocopy the TLR7 Tg mice (Fukui et al., 2011) suggest that it is an altered signaling pathway leading to an enhanced response to TLR7 agonists rather than the copy number per se that conveys the immunological abnormalities observed in these two mouse strains.

It was previously shown that the total population of TLR7 Tg splenic B cells was more responsive to TLR7 ligand stimulation in vitro (Deane et al., 2007). We found that the T1 subset in particular proliferated at much higher rates than WT T1 cells, consistent with our in vivo data. Furthermore, stimulation with SmRNP, a natural RNA-containing ligand, also promoted the proliferation of a small but detectable portion of TLR7.1Tg T1 B cells, strongly suggesting that cells with anti-RNA specificity are being activated. Since neither BAFF nor other survival factors were added to the cultures, TLR7 ligation alone promoted a significant expansion of T1 B cells. This suggests that TLR7 overexpression may promote the survival and proliferation of a limited population of cells, most probably those with low to moderate specificity for RNA. Surprisingly, BCR ligation with anti-IgM alone also stimulated greater survival and proliferation of TLR7.1Tg compared with WT T1 B cells. We speculate that the small proportion of responding cells may have already received a TLR7 signal through RNA-reactive BCR in vivo and that anti-IgM acts as a second signal for these cells. Whereas strong BCR ligation induces cell death and is thought to provide a tolerance checkpoint for T1 B cells (King and Monroe, 2000), it is remarkable to note that when activated through TLR7, anti-IgM induces striking proliferation of WT T1 B cells, further supporting a role for TLR7 in imparting an anti-death/survival signal. The mechanisms by which B cells come in contact with and respond to RNP- and deoxyribonucleoprotein (DNP)-containing antigens after exit from the BM are still poorly understood. We found that the expanded T1 cells in the TLR7.1Tg mice were located predominantly in the splenic RP, as were most of the proliferating (BrdU⁺) cells. These findings suggest that the splenic RP represents a primary site for the activation of anti-RNA–specific B cells in the periphery. Despite the expansion of T1 B cells, they were able to undergo apoptosis, and the increased cell death we observed in TLR7.1Tg T1 B cells may have contributed to the antigenic load stimulating other autoreactive B cells. A likely possibility is that B cells directly bind to RNP-containing antigens exposed on the surface of apoptotic cells (Hall et al., 2004; Radic et al., 2004). Alternatively or in addition, B cells may bind RNP or DNP antigens presented on the surface of other cells such as neutrophils, DCs, or macrophages (Qi et al., 2006; Cyster, 2010; Hakkim et al., 2010; Puga et al., 2012). Histological analysis showed a significant accumulation of T1-like (CD93^brightIgD⁻CD138⁻) cells within the splenic RP, a location which would allow interactions with RP macrophages and DCs. The activation of autoreactive T1 B cells in TLR7.1Tg mice may well be driven by these cells, as DCs can directly present Ag to B cells through a recycling pathway following Ag internalization (Bergtold et al., 2005; Qi et al., 2006; Chappell et al., 2012). Recently, Joo et al. (2012) have demonstrated that DCs from SLE patients can efficiently stimulate naive and memory B cells to produce IgG and IgA antibodies, suggesting a direct interplay between myeloid DC and B cells.

Despite their seemingly immature phenotype, recent studies showed that T1 B cells have the potential to become activated and to mount efficient humoral immune responses (Mao et al., 2004; Han et al., 2007; Ueda et al., 2007; Aranburu et al., 2010). Ueda et al. (2007) found that mouse splenic T1 B cells proliferate rapidly in response to TLR ligands, such as CpG DNA (TLR9 ligand) and LPS (TLR4 ligand), and produce IgM and IgG Abs. Notably, the high levels of AID in T1 B cells was associated with class-switch recombination and production of high affinity IgG Abs, which occurred independently of T cell help (Han et al., 2007; Ueda et al., 2007). Human T1-like immature B cells also expand and proliferate in response to bacterial CpG DNA, and these so-called immature B cells acquired a phenotype of IgM⁺ memory B cells, which carry somatic hypermutations; the somatic mutations were mainly in VH1 and VH4/6 genes, which encode for anti-CpG DNA specificity (Aranburu et al., 2010), suggesting possible “cooperation” between the BCR and TLR9 signaling. Recent studies have also reported an increased proliferation of immature (T1 and T2) dual-κ B cells in MRL/lpr mice, which might contribute to the increased frequency of autoreactive B cells in these mice (Fournier et al., 2012).

Here, we found that TLR7 stimulation caused very little production of class-switched IgG by WT T1 B cells. In marked contrast, TLR7 stimulation of TLR7.1Tg T1 B cells led to the production of high concentrations of class-switched IgG Abs. Although AID was only modestly increased in TLR7.1Tg T1 B cells compared with WT, the transcription factor T-bet was expressed at significantly higher levels in TLR7 Tg T1 B cells, suggesting that increased expression of this transcription factor may contribute to increased class switching in T1 B cells in TLR7.1Tg mice. Although our in vivo and in vitro findings support a role for class-switched antibody production by T1 B cells in the splenic RP, we cannot rule out the possibility the IgG2b⁺ and IgG2c⁺ cells found in the RP of TLR7.1Tg mice have developed as a result of GC formation, especially as T cell–dependent generation of GCs and plasmablasts in TLR7 Tg mice has recently been demonstrated (Walsh et al., 2012). Both T1 B cells and FO B cell populations in TLR7.1Tg had increased frequencies of CD44^hiCD138⁺ cells, supporting a model in which both immature and mature B cell populations contribute to the generation of plasma cells in vivo. However, surprisingly, sorted TLR7.1Tg T1 cells produced significantly higher levels of IgG Abs after TLR7 stimulation as compared with FO B cells. This difference might be due to the fact that TLR7.1Tg T1 B cells, unlike TLR7.1 Tg FO B cells, already express AID and high levels of T-bet so that their activation may not require additional T cell help.

Murine B cells lacking type I IFN receptor (IFNαR-KO) respond only weakly to TLR7 ligands, most likely explained by the requirement for type I IFN signaling to up-regulate TLR7 expression in B cells and for B cells to respond to TLR7 agonists (Bekeredjian-Ding et al., 2005; Thibault et al., 2008; Green et al., 2009). Our results indicate that TLR7 overexpression could compensate for the lack of a type I IFN signal in that TLR7.1 × IFNαR-KO mice maintained T1 and FO B cell expansion, T1 B cell activation, and the accumulation of plasma cells. In addition, in TLR7.1 × IFNαR-KO mice, we observed the restoration of MZs and MZ B cell numbers, indicating that the TLR7-mediated loss of MZ B cells requires type I IFN signals. This finding is consistent with previous studies showing that type 1 IFN signals can regulate the MZ B cell population either by promoting MZ B cell egress into the follicles or by influencing their survival (Rubtsov et al., 2008; Wang et al., 2010). Because the overexpression of RNase in TLR7.1 × RNase dTg mice limited the expansion and proliferation of T1 cells and the accumulation of plasma cells but did not restore the MZ B cell population, it appears that the T1 and MZ B cell subsets are regulated differently in TLR7.1Tg mice.

Extra-FO immune responses have been implicated in the activation of autoreactive B cells though TLRs (Christensen and Shlomchik, 2007). The mechanisms that regulate these responses remain unknown. Whereas the T1 stage of B cell maturation may be an important checkpoint for autoreactive B cell deletion (Meffre and Wardemann, 2008), our results demonstrate that when TLR7 responsiveness is increased, RNA-containing Ags can promote the activation (rather than deletion) of T1 B cells within the splenic RP, leading to the production of class-switched Abs. Human and mouse studies have implicated that alteration of TLR7 and/or IRF7 expression can be a key factor in the development of pathogenic anti-RNA Ab in SLE (Subramanian et al., 2006; Deane et al., 2007; Salloum et al., 2010; Shen et al., 2010; Kawasaki et al., 2011). An increased frequency of immature T1/T2-like cells has previously been described in patients with autoimmune diseases, including RA, SLE, and Sjogren’s Syndrome (Daridon et al., 2006; Landolt-Marticorena et al., 2011; Vossenkämper et al., 2012), suggesting that these cells might represent an important source of autoantibodies. Importantly, the immature B cell compartment in both humans and mice has been shown to contain increased frequencies of polyreactive BCR specificities (Yurasov et al., 2005; Andrews et al., 2013), suggesting that if not properly regulated, B cells derived from this compartment have the potential to produce auto-Abs.

In a second model of lupus-like disease associated with anti-RNA autoantibodies, H564Igi^−/+ mice, the majority of H564Id⁺ B cells exhibited a classic immature B cell phenotype, similar to that of T1 and T2 B cells, and, as previously described by Berland et al. (2006), they were largely excluded from the B cell follicles and localize within the splenic RP or at the T-B interface of the follicles. Thus, these Id⁺ cells might be competing with nonautoreactive B cells and may become developmentally arrested at an immature stage, similar to what has been described for anti-dsDNA–specific B cells (Mandik-Nayak et al., 1997). Despite their immature phenotype, Berland at al. (2006) found that H564Igi^−/+ mice spontaneously produced high titers of class-switched Id Abs, indicating that Id⁺ B cells can become activated and produce Abs in vivo. Because very few Id⁺ B cells develop into FO B cells and H564Idi^−/+ mice did not develop Id⁺ GCs, the activation of H564 Id⁺ B cells and generation of autoantibodies most probably occurs outside of the follicles. Consistent with this idea, our in vitro data showed that CD93⁺H564Id⁺ B cells produced higher titers of class-switched Abs upon TLR7 ligand stimulation when compared with CD93⁻H564Id⁺ cells, suggesting again that in a physiological setting, immature B cells might be a major source of Id Ab. It remains possible that some Id⁺ B cells from the FO B cell compartment contribute to auto-Ab production in vivo. Based on our results, it might be worth testing if dysregulated expression of TLR7 can bypass normal requirements for T1 cell differentiation and promote activation and Ab production by B cells arrested at the T1 stage caused by deficiency of BAFF, BAFFR, or SPPL2A (signal peptide peptidase–like 2A; Mackay and Schneider, 2009; Beisner et al., 2013; Bergmann et al., 2013; Schneppenheim et al., 2013).

The overexpression of TLR7 resulted in increased activation of T1 cells in the RP, associated with production of class-switched Abs. Removal of RNA-containing material in TLR7.1 × RNase dTg mice significantly limited the expansion and proliferation of T1 cells, suggesting that endogenous RNA-containing Ags drive their activation.

Enhanced TLR7 responses occur in mice that overexpress TLR7 or have mutations in the TLR chaperone Unc93b (Fukui et al., 2009, 2011), as well as in SLE patients with genetic predisposition (García-Ortiz et al., 2010; Shen et al., 2010; Kawasaki et al., 2011; Lee et al., 2012; Tian et al., 2012) or environmental exposure (e.g., altered response to chronic virus infection; Poole et al., 2009). Our data on T1 B cells that overexpress TLR7 suggest that dysregulation of TLR expression and function in T1 B cells may change their response to Ag, which normally induces T1 B cells to undergo cell death (King and Monroe, 2000; Su et al., 2004). Autoreactive T1 cells with altered TLR7 expression or downstream signaling may respond to both RNA and RNA-associated Ags, resulting in the production of IgG auto-Abs. The IgG auto-Abs produced may not initially be pathogenic but, after T cell help either in the GC (Walsh et al., 2012) or in an extra-FO site, could facilitate the subsequent development of high-affinity pathogenic auto-Abs. Thus, activation of autoreactive T1 B cells may play a key role in the initial development of autoantibodies in SLE.

C57BL/6 mice were purchased from Charles River or The Jackson Laboratory. TLR7.1Tg mice were provided by S. Bolland (National Institutes of Health, Bethesda, MD; Deane et al., 2007); IFNαR-KO mice (Müller et al., 1994) were a gift from D. Stetson (University of Washington, Seattle, WA); TLR7.1Tg × IFNαR-KO mice were generated by crossing TLR7.1Tg mice to IFNαR-KO mice; RNase Tg mice were generated by expressing bovine RNase under the human albumin promoter (Sun et al., 2013); TLR7.1 × RNase dTg mice were generated by crossing TLR7.1Tg mice to RNase Tg mice; VDJ9κ5 IgH and IgL (HEL-Tg) mice were obtained from J. Cyster (University of San Francisco, San Francisco, CA); and TLR7.1 × HEL-Tg mice were generated by TLR7.1Tg mice to HEL-Tg mice. H564Igi mice were described previously (Berland et al., 2006). Mice used in the experiments were between 2 and 6 mo, except for H564Igi mice, which were used at 6–12 mo of age. All mice used were maintained under specific pathogen-free conditions at the University of Washington. All animal procedures were approved by the Institute for Animal Care and Use Committee at the University of Washington.

Mixed BM chimeric mice were generated by lethally irradiating (1,000 rads) F1 progeny of B6 and B6.SJL mice and reconstituting them with a 1:1 ratio of B6 (Ly5.1) and B6.SJL (Ly5.2) or TLR7.1 (Ly5.2) and B6.SJL (Ly5.1) whole BM cells. Mixed BM chimeric mice were reconstituted for 2–4 mo before analysis.

For analysis of splenic B cell, RBC-lysed single cell suspensions were stained with the fluorescently labeled mAbs listed in Table S2. Multicolor flow cytometry was performed by staining the cells with appropriate combinations of mAbs. For BrdU analysis, mice were injected with 1 mg BrdU 1 h before sacrificing. Splenocytes were surface stained with mAbs and BrdU signal was detected using the FITC BrdU Flow kit according to the manufacturer’s instructions (BD). HEL-specific B cells that express VDJ9κ5 IgH and IgL (HEL⁺) B cells were identified using biotinylated HEL. H564Id⁺ cells were identified using Alexa Fluor 647, biotin, or FITC-conjugated anti-H564 Ab. All data were collected using an LSRII or FACScan flow cytometer (BD) and analyzed with FlowJo software (Tree Star).

For RT-PCR analyses, single cell suspensions from individual mice were RBC-lysed and stained with anti-B220, anti-IgM (eBioscience), anti-CD24, anti-CD21 (BD), and anti-CD23 (Invitrogen). A minimum of 200,000 cells was sorted from each subset. Cells were washed with cold PBS and cell pellets were kept at −80°C until use. For in vitro cell culturing experiments, spleens from two to four mice per group were used. Single cell suspensions were prepared under sterile conditions and cells were stained with anti-B220, anti-CD24, and anti-CD21. Post-sort analyses were performed to assess the purity of sorted cells. H564 Id⁺ cells from H564Igi^+/− mice were stained with anti-B220, anti-H564 Id Ab, and anti-CD93 Abs. All cell sorts were performed using a FACSAria II high-speed cell sorter (BD) at 4°C under sterile conditions.

B cell subsets were sorted as described above and loaded with 2.5 µM CFSE in PBS at 37°C for 10 min, quenched with RPMI 1640 containing 10% FBS (Thermo Fisher Scientific), and washed extensively. Cells were resuspended in RPMI 1640 medium containing 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate, 55 mM 2-ME, and 10 mM Hepes, placed in 96-well plates at a concentration of 0.2 × 10⁶ cells/well, and treated with different stimuli. Reagents used for in vitro stimulation were F(ab′)2 fragment goat anti–mouse μ-chain (Jackson ImmunoResearch Laboratories, Inc.), R848 (InvivoGen), and SmRNP (RNP/Sm; Arotec Diagnostics Limited). Cells were cultured at 37°C and 5% CO₂ for 72 h, harvested, and analyzed for CFSE dilution by flow cytometry. Supernatants from each well were frozen and used to assess the production of antibodies.

Spleens from C57BL/6, TLR7.1Tg, and H564Igi^+/− mice were embedded in OTC medium and frozen at −80°C. 6–8 µM sections were fixed to glass slides and kept at −80°C until use. For primary staining, sections were fixed in ice-cold acetone for 5 min, air-dried, rehydrated in PBS for 5 min, and stained at room temperature with combinations of CD93-FITC, IgD-Biotin, IgD-FITC, B220-eFluor450, CD138-PE (all from eBioscience), CD169-FITC (AbD SeroTec), anti–H564-Id-biotin, and IgG2c-biotin or IgG2b-biotin (Jackson ImmunoResearch Laboratories) for 90 min in PBST (PBS containing 0.05% Tween-20). Biotinylated reagents against surface markers were detected by incubation with streptavidin–Alexa Fluor 555 (Invitrogen) or streptavidin–Pacific blue (eBioscience) for 60 min at room temperature in PBST.

For BrdU detection, acetone-fixed slides labeled with IgD-FITC were incubated in Retrievagen A solution (BD) at 89°C for 10 min and allowed to cool at room temperature for 20 min before staining with BrdU-biotin (BD). BrdU-biotin was detected using Cyanine3 TSA amplification (PerkinElmer) according to manufacturer’s instructions. Sections were washed in PBS and mounted in VectaShield HardSet mounting medium (Vector Laboratories). Images were collected on a confocal microscope (LSM 510 META; Carl Zeiss) with LSM 510 (v 4.2) software using 10 or 20× objectives at room temperature. Images were processed using ImageJ (National Institutes of Health) and Photoshop (Adobe) software. The anti-nuclear antibody (ANA) test was performed using HEp-2 antigen substrate slides (MBL Bion). Slides were incubated for 2 h with undiluted culture supernatants, washed, and incubated for 1 h with biotin-conjugated goat anti–mouse IgG (Jackson ImmunoResearch Laboratories), followed by 1 h incubation with streptavidin-FITC (BioLegend). All incubations were performed at room temperature. Slides were washed and mounted in mounting medium. Images were captured on a microscope (Inverted DMIRB; Leica) with MetaMorph software using 20× magnifications.

Total RNA was isolated from sorted cells using the RNeasy mini kit with on-column DNase treatment (QIAGEN). First-strand cDNA was generated using 25 ng of total RNA with the high-capacity cDNA RT-kit using random primers (Applied Biosystems). cDNA was diluted to an equivalent of 0.1 ng/µl total RNA and 8 µl were used per reaction. Primers, as indicated in Table S1, were synthesized (IDT) and diluted to the appropriate concentrations for qPCR using molecular grade water. BLAST results of the primers showed specific sequence homology only to the gene of interest. Reactions in duplicate (20 µl) were run on an ABI StepOne Plus system using a 1:1 mix of template and primer to Absolute Blue SYBR green ROX master mix (Thermo Fisher Scientific), using a two-stage cycle of 95°C for 15 s and 60°C for 1 min repeated for 40 cycles, followed by a dissociation stage. Threshold cycle (Ct) values were determined by setting a constant threshold at 0.2, and fold changes in gene expression were then calculated using the 2^−ΔΔCT method (Livak and Schmittgen, 2001) and relative expression method (Pfaffl et al., 2002).

ELISA plates were precoated with 10 µg/ml goat anti–mouse IgM F(ab′)2 or goat anti–mouse IgG (H+L) mAbs (Jackson ImmunoResearch Laboratories). After washing, plates were blocked with 4% nonfat dry milk in PBS-Tween. Culture supernatant were added in serial dilutions and incubated for 2–4 h at room temperature. Plates were washed and bound Abs were detected using isotype-specific HRP conjugates of anti-IgG1 and anti-IgG2b (ICL); and anti-IgM, anti-IgG2a, and anti-IgG2c (SouthernBiotech). HRP activity was visualized using tetramethylbenzidine peroxidase (Bio-Rad Laboratories). Standard curves were generated using IgM (Jackson ImmunoResearch Laboratories, Inc.), IgG2c, IgG1, IgG2a, and IgG2b (SouthernBiotech) standards and used to calculate the absolute quantities of different antibodies. For detection of anti-RNA Abs, ELISA plates were treated with Poly-l-Lysine (Sigma-Aldrich), coated with 10 µg/ml yeast RNA (Sigma-Aldrich) diluted in PBS and blocked with 2% BSA (Sigma-Aldrich) in PBS. Culture supernatant were added in serial dilutions and incubated overnight at 4°C. Plates were washed, and bound Abs were detected using biotin-conjugated donkey anti–mouse IgG (Jackson ImmunoResearch Laboratories) and streptavidin-AP (Invitrogen). For measurement of 564Igi-Id⁺ antibodies, ELISA plates were coated with 5 µg/well of purified mAb B6-256 anti-Id. Bound serum antibody was detected with alkaline phosphatase (AP)–conjugated isotype-specific goat anti–mouse (SouthernBiotech). AP activity was visualized using pNPP-phosphatase substrate (Sigma-Aldrich).

Statistical significance between groups was determined by two-tailed, unpaired Student’s t test or by one-way ANOVA with Turkey post-test. Graphs and statistical analyses were performed using Prism 5.0 software (GraphPad Software). Values are reported as mean ± SD or ± SEM.

Table S1 shows primer sequences used for qPCR analyses. Table S2 shows antibodies and clones used. Fig. S1 shows gating strategy used for sorting of T1 and FO B populations and post-sort analyses.

We thank Dr. Sylvia Bolland (National Institutes of Health) for kindly providing TLR7.1Tg mice. We thank Nalini Agrawal (Keith Elkon’s Lab) for her help with the study. We thank the University of Washington (UW) Department of Immunology Flow Cytometry Facility and the UW Keck Microscopy Facility for assistance and support with cell sorting and cell imaging.

This study was supported in part by National Institutes of Health grants AR48796 (K.B. Elkon), AI44257 (E.A. Clark), AI52203 (E.A. Clark), and AI081948 (J.A. Hamerman).

The authors have no conflicting financial interests.