IRF7 controls spontaneous autoimmune germinal center and plasma cell checkpoints

Systemic lupus erythematosus (SLE) is a debilitating autoimmune disease initiated by dysregulated B cell responses and the emergence of autoreactive B cells (Karrar and Cunninghame Graham, 2018; Nashi et al., 2010; Rawlings et al., 2017). Autoreactive B cells in SLE produce antinuclear antibodies (ANA) that deposit as immune complexes (ICs) in various tissues, including kidneys and vasculature, causing inflammation and end-organ disease (Bagavant and Fu, 2009). Genetic predisposition in concert with TLR activation by self-nucleic acid, permits B cell escape of tolerance mechanisms, leading to the emergence of autoreactive B cells (Celhar and Fairhurst, 2014; Christensen and Shlomchik, 2007; Green et al., 2009). TLR signaling promotes aberrant differentiation of autoreactive B cells into plasma cell (PC) and germinal center (GC) cells to generate autoantibody-producing B cells (Brown et al., 2022; Christensen et al., 2006; Deane et al., 2007; Jackson et al., 2014; Soni et al., 2014); however, the signals downstream of TLR activation and the mechanisms that regulate autoreactive B cells in SLE remain unclear.

We and others previously demonstrated a B cell–intrinsic requirement for TLR7 in promoting autoimmune GC and PC responses and systemic autoimmunity (Brown et al., 2022; Christensen et al., 2006; Cosgrove et al., 2023; Deane et al., 2007; Hwang et al., 2012; Jackson et al., 2014; Soni et al., 2014). Stimulation of endosomal TLRs by nucleic acids, including TLR7, activates NFκB family members and IFN regulatory factors 5 (IRF5) and 7 (IRF7). To date, numerous studies have characterized the involvement of IRF5 and IRF5 risk variants in SLE in mice and humans (Ban et al., 2021; Banga et al., 2020; Cham et al., 2012; Pellerin et al., 2021; Richez et al., 2010). IRF7 was previously shown to promote autoantibody responses but not lupus nephritis in a chemically (pristane)-induced B6 model of SLE (Miyagawa et al., 2016). IRF7 risk alleles are also implicated in the development of human SLE (Salloum et al., 2010; Yasuda et al., 2007). However, IRF7-expressing cell type(s) and mechanisms by which IRF7 drives autoreactive B cell differentiation through PC and GC checkpoints, leading to autoantibody-secreting PCs and autoantibody production in SLE, remain unknown. IRF7 has long been thought to contribute to SLE by regulating type 1 IFN (T1-IFN) production by plasmacytoid dendritic cells (pDCs) (Honda et al., 2005; Yasuda et al., 2007), although IRF7 is also expressed in B cells and monocytes (Ning et al., 2011). Whether and how IRF7 may promote autoimmune GC and PC responses, autoantibody production, and disease in spontaneous SLE models via functioning in cell types other than pDCs is unknown. It is also unclear whether IRF7 transcriptionally controls SLE-promoting gene programs in B cells other than T1-IFN genes.

Here, using FcγRIIB^−/− and FcγRIIB^−/−Yaa spontaneous SLE mouse models, we identified a major role for IRF7 in promoting spontaneous autoimmune PC, GC, and autoantibody responses and SLE-like disease but not in foreign antigen–driven PC, GC, and antibody responses. Single-cell RNA sequencing (scRNAseq) of B cells from FcγRIIB^−/− IRF7^−/− and FcγRIIB^−/− control mice indicated a role for IRF7-driven genetic programs in regulating B cell differentiation through pre-GC to GC and PC populations. Mixed bone marrow (BM) competition chimeras demonstrated a role for IRF7 expression in hematopoietic compartments in driving GC and PC but not myeloid or T cell responses in FcγRIIB^−/− mice. Using B cell–specific mixed BM chimeras, we further demonstrated a B cell–intrinsic requirement of IRF7 in the production of class-switched IgG autoantibodies but not for promoting overall spontaneous autoimmune PC and GC responses. Together, we delineate B cell–intrinsic and –extrinsic functions for IRF7 in regulating B cell transcription, translation, and metabolism, and in breaching tolerance and promoting autoimmune GC and PC responses and systemic autoimmunity development.

To identify the mechanisms by which IRF7 regulates autoimmune GC and PC responses and systemic autoimmunity, we crossed IRF7-deficient (IRF7^−/−) mice with well-characterized SLE-prone FcγRIIB^−/− mice (Bolland et al., 2002; Richez et al., 2010; Shinde et al., 2018). FcγRIIB^−/− mice were originally generated on a 129 background and backcrossed to B6 mice for several generations (Bolland et al., 2002; Richez et al., 2010; Shinde et al., 2018). We further backcrossed FcγRIIB^−/− mice to B6 for at least N12–14 generations prior to intercrossing and then crossing with IRF7-deficient mice. FcγRIIB^−/− mice in our colony are deficient in the Fcgr2b gene and carry 129-derived signaling lymphocyte activation molecule (Slam) family genes that are tightly linked to the Fcgr2b gene (Soni et al., 2015). FcγRIIB^−/− mice had many of the manifestations of human SLE, including splenomegaly with the increased number of splenocytes (Fig. S1, A–C), elevated GC and PC responses, ANAs, and renal disease (Fig. 1 and Fig. S1, A–C) (Bolland et al., 2002; Shinde et al., 2018; Soni et al., 2015). IRF7-deficient FcγRIIB^−/− mice (FcγRIIB^−/−IRF7^−/−) had reduced spleen size and splenocyte number, reduced GC, effector T cell (Teff) and T follicular helper (Tfh) cell, and PC (Fig. 1, A–H and Fig. S1, A–C) responses at 2, 4, and 6 mo of ages. FcγRIIB^−/−IRF7^−/− mice had reduced numbers of autoantibody-producing antibody-forming cells (AFCs) in the spleen and BM (Fig. 1, I and J) and dampened serum ANA reactivity (Fig. 1 K).

In FcγRIIB^−/− mice, autoantibodies started accumulating at 2 mo of age, while at this time point only nucleosome-specific serum IgM, IgG, and IgG2c titers were reduced in IRF7-deficient mice (Fig. S1, D–F). Nucleosome-, double stranded deoxyribonucleic acid (dsDNA), and Smith Ribonucleoprotein (SmRNP)-specific IgG and IgG2c titers were significantly reduced in FcγRIIB^−/−IRF7^−/− mice at 4 and 6 mo of age when substantial amounts of autoantibodies accumulated in FcγRIIB^−/− mice (Fig. 1, L–N). We also found reduced numbers of IgG1- and IgG2c-producing, but not total IgG- and IgG3-producing, splenic and BM AFCs in FcγRIIB^−/−IRF7^−/− mice compared with FcγRIIB^−/− control mice (Fig. S2, A and B). In addition, total serum IgG, IgG1, IgG2c, and IgG3 titers were lower in FcγRIIB^−/−IRF7^−/− mice than in FcγRIIB^−/− control mice (Fig. S2 C). IC deposition and immune cell infiltration in the kidney (Fig. 1, O and P) and kidney pathology (Fig. 1 Q) and scoring for glomerulonephritis, interstitial nephritis, and vessels (Fig. 1 R) were significantly reduced in FcγRIIB^−/−IRF7^−/− mice. FcγRIIB^−/−IRF7^−/− mice also had reduced myeloid cell populations, including conventional DCs and monocyte-derived DCs, although the frequency and number of pDCs were increased (Fig. S3).

Splenomegaly (Fig. 2 A), GC and PC responses (Fig. 2, B–E), and age-associated B cell (ABC) responses, as well as Teff and Tfh cell responses (Fig. 2, G and H), were also reduced in IRF7-deficient FcγRIIB^−/−yaa male mice (Fig. 2 F) that express two copies of TLR7 and present with robust and rapid systemic autoimmune responses and disease (Bolland et al., 2002; Richez et al., 2010). Additionally, FcγRIIB^−/−yaaIRF7^−/− mice showed reduced numbers of autoantibody-producing splenic and BM AFCs (Fig. S4, A and B) and reduced serum titers of IgG and IgG2c autoantibodies to RNA-associated antigen, SmRNP (Fig. 2 I). However, no significant changes were observed in myeloid cell responses except for a consistent increase in the number of pDCs in the absence of IRF7 (Fig. S4, C and D). High-dimensional flow cytometry analysis of immune cell populations within the kidneys of these mice showed significant reductions in the number of CD45⁺ cells, inflammatory DCs (iDCs), macrophages, natural killer (NK) cells, and CD8⁺ Teffs, but not B cells, pDCs, monocyte-derived DCs, conventional DCs, monocytes, neutrophils, and CD4⁺ Teffs in the absence of IRF7 (Fig. 2, J and K). In addition, FcγRIIB^−/−yaaIRF7^−/− mice had reduced serum blood urea nitrogen (BUN) and creatinine levels (Fig. 2, L and M), indicators of improved renal function. Finally, FcγRIIB^−/−yaaIRF7^−/− mice had improved overall survival (Fig. 2 N).

Together, combined data from two SLE mouse models reinforce the role of IRF7 in controlling spontaneous autoimmune GC and PC responses, leading to autoantibody production and the development of SLE-like disease.

We next determined the role of IRF7 in foreign antigen–driven GC, PC, and antibody responses by immunizing B6 control and IRF7^−/− mice with 4-hydroxy-3-nitrophenylacetyl-KLH (NP-KLH) in CFA. We observed no differences in the frequencies of total and NP-specific B cell and GC B cell responses between the strains 14 days after NP-KLH immunization (Fig. 3, A–D). Total and NP-specific CD138⁺TACI⁺ PC responses were similar between the strains except for an increased frequency of immature PCs in IRF7-deficient mice (Fig. 3, E–H). IRF7^−/− mice had a reduced percentage of Teffs, although no difference in Tfh frequency between the strains was observed (Fig. 3, I and J). We also observed no differences in high- (NP₄) and low- (NP₂₈) affinity NP-specific IgG and IgG1 Ab responses (Fig. 3, K and L). We then immunized autoimmune-prone mice sufficient and deficient in IRF7 during ongoing autoimmune responses. Consistent with the data observed in non-autoimmune B6 and IRF7^−/− mice, we found no differences in the frequencies of total and NP-specific B cell and GC B cell responses between FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− mice 14 days after NP-KLH immunization (Fig. 4, A–D). FcγRIIB^−/−IRF7^−/− mice had reduced total PCs but showed increased frequencies of plasmablasts, immature PCs, and NP⁺ PCs (Fig. 4, E–H). FcγRIIB^−/−IRF7^−/− mice further had no deficits in Teff and Tfh (Fig. 4, I and J) and low- and high-affinity NP-specific IgG and IgG1 antibody responses (Fig. 4, K and L). Together, findings from SLE-associated and immunization-induced responses in both non-autoimmune and autoimmune-prone mice highlight a critical role of IRF7 in promoting systemic autoimmune responses but not foreign antigen–induced GC, PC, and antibody responses.

To determine the effects of IRF7 expression in the host microenvironment versus the hematopoietic compartment on autoimmune GC and PC responses in FcγRIIB^−/− mice, we generated competitive mixed BM chimeras. Lethally irradiated wild-type hosts (CD45.1⁺) received a 50:50 mix of BM from FcγRIIB^−/− (CD45.1⁺CD45.2⁺) and FcγRIIB^−/−IRF7^−/− (CD45.2⁺) donors (Fig. 5 A). 8 wk after reconstitution, recipient mice were equally reconstituted by both donor populations (Fig. 5 B). Analysis of B cell populations revealed no significant differences in the development of naïve follicular B cells (Fig. 5, C and D) or ABCs (Fig. 5, E and F). However, we observed a significant reduction in IRF7-deficient B cells within splenic GC (Fig. 5, G and H) and PC responses within the spleen and BM (Fig. 5, I and J). Interestingly, IRF7 deficiency in hematopoietic cells did not affect Teff and Tfh populations (Fig. 5, K and L), and numerous myeloid populations except for iDCs, which remained significantly reduced, and pDCs, which were elevated in the absence of IRF7 (Fig. 5 M). These findings suggest the importance of IRF7 expression in hematopoietic cells in B cell differentiation into GC B cells and PCs during systemic autoimmune responses in FcγRIIB^−/− mice.

Given the effects of IRF7 expression in hematopoietic cells on spontaneous B cell responses, we next examined the role of IRF7 in transcriptional control of B cell responses in SLE by performing scRNAseq on splenic B cells sorted from 3-mo-old FcγRIIB^−/−IRF7^−/− and FcγRIIB^−/− mice. To be able to detect transcriptome in subsets of splenic-activated B cells that represent a small percentage of cells compared with naïve B cells, analysis was performed on B cells composed of 90% IgD⁻ antigen-experienced B cells and 10% IgD⁺ naïve B cells (Fig. 6 A and Fig. S5 A). Unsupervised clustering of 36,842 B cells passing quality control identified 14 B cell clusters as visualized using Uniform Manifold Approximation and Projection (UMAP) (Fig. 6 B; and Fig. S5, B and C). These B cell clusters were annotated based on the expression of canonical markers specific for subpopulations of B cells (Fig. 6 C) and by comparing our new data set with the annotated scRNAseq previously reported by others (Laidlaw et al., 2020). We identified a sharp reduction in the percentage of IRF7^−/− B cells in the pre-GC transition, GC dark and light zones, and PC/blast clusters, but not in the follicular cluster 2, marginal zone (MZ), or switched activated populations (Fig. 6, D and E). Transitional and follicular cluster 1 and naïve-activated B cells were proportionally higher in IRF7^−/− B cells than control cells (Fig. 6 E). In summary, scRNAseq of autoimmune B cells sufficient and deficient in IRF7 identified 14 clusters and revealed significant differences in subpopulations of B cells, pointing to potential regulation of autoimmune GC and PC responses by IRF7-driven transcriptional programs.

To determine the role of IRF7 in B cell differentiation into the GC and PC pathways, we analyzed the top 100 genes differentially expressed between IRF7-sufficient and -deficient B cells across the scRNAseq dataset and identified dynamic transcriptional changes associated with B cell differentiation (Fig. 7 A). To further investigate the involvement of IRF7 in regulating B cell differentiation, we used pseudotime inference to identify transcriptional programs associated with B cell differentiation (Fig. 7 B). The pseudotime trajectory revealed that B cells differentiated in two major pathways: one group differentiated through a pre-GC transitional bridge into GC and PC clusters, and the other group into naïve activated and switched activated B cells and MZ B cells (Fig. 7 B), which can contribute to the generation of autoantibody-producing cells and autoantibodies. We found that IRF7 deficiency in B cells predominantly affected B cell differentiation from pre-GC clusters into GC and PC clusters. We further re-clustered the pre-GC transitional B cell cluster from our scRNAseq dataset to identify underlying subpopulations and their regulation by IRF7 (Fig. 7 C). The re-clustered pre-GC transitional B cells produced five subclusters that allowed further examination. We examined these five subclusters for gene markers associated with B cell differentiation and manually annotated the five subclusters based on cluster-based gene expression. Pre-GC1 expressed numerous genes that distinguished them from naïve follicular B cells, but they retained the most follicular markers in the subcluster. As B cells differentiated through pre-GC stages, they downregulated follicular B cell markers (Cd38, Klf2, and S1pr1) and began to upregulate markers of activation/GC differentiation (Fig. 7 D, S1pr2, Aicda, Mef2b, Mef2c, Pou2f2, Ezh2, Fas, J chain, and Slamf7). Further analysis of scRNAseq data indicated that cells of the pre-GC1 cluster were quiescent, followed by proliferation of B cells at pre-GC2 and pre-GC3 clusters (Fig. 7, E and G). Finally, we determined the percentage of each subcluster that was composed of FcγRIIB^−/− or FcγRIIB^−/−IRF7^−/− B cells and found a stark reduction in FcγRIIB^−/−IRF7^−/− B cells starting at the pre-GC2 cluster (Fig. 7, E and F). These data indicate that IRF7-driven genetic programs regulate B cell differentiation at the pre-GC stages during spontaneous autoimmune B cell responses in SLE-prone FcγRIIB^−/− mice.

To understand the B cell–intrinsic mechanism of IRF7-mediated systemic autoimmunity, we performed pathway analysis on differentially expressed genes identified between FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− mice within the pre-GC clusters of the scRNAseq. The top two pathways identified contained translation (EIF2 signaling) and oxidative phosphorylation (OXPHOS)–related genes, including ribosomal genes that were differentially expressed between FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− pre-GC clusters (Fig. 8, A and B). We also performed IRF7 chromatin immunoprecipitation (ChIP)-seq on antigen-experienced IgD⁻ B cells from FcγRIIB^−/− mice and identified 1229 IRF7 peaks, and ∼32% mapped within 3 kb of a promoter, and the remainder fell largely within intergenic regions or introns (Fig. 8 C). Combined scRNAseq and ChIP-seq analysis identified overlap between 42 upregulated and 29 downregulated genes between FcγRIIB^−/− and FcγRIIB^−/− IRF7^−/− B cells that were IRF7 targets (Fig. 8 D). Surprisingly, we found IRF7 peaks in the promoters of several ribosomal protein genes (Fig. 8 E). To validate the OXPHOS pathway identified in the scRNAseq data set (Fig. 8 A), we performed extracellular flux analysis, which revealed lower OXPHOS and glycolysis in FcγRIIB^−/−IRF7^−/− B cells compared with FcγRIIB^−/− control B cells, including basal respiration and non-mitochondrial O₂ consumption rate (OCR) (Fig. 8, F and G). The differences in the OXPHOS pathway were not a consequence of different proportions of B cell subsets between the two groups. Our analysis of peripheral B cell subsets showed no major differences (data not shown) in the proportions of B cell subsets in the spleens between the two groups except for GC B cells and PCs (Fig. 1). Thus, we do not believe that differences in GC and PC frequency and number, which represent about 5% of total B cells, can alone make up for substantial differences we observed in OCR and extracellular acidification rate (ECAR) values of total B cells between the two groups. To validate the EIF2 signaling pathway (Fig. 8 A) linked to translation, we measured translation and found that FcγRIIB^−/−IRF7^−/− GC B cells and PCs had reduced translation compared with FcγRIIB^−/− control B cell counterparts (Fig. 8 H). FcγRIIB^−/−IRF7^−/− B cells also showed reduced translation compared with FcγRIIB^−/− control B cells after TLR7 stimulation but not IgM and CD40 (Fig. 8 I). These data suggest that IRF7 transcriptionally regulates translation and metabolism in B cells, promoting SLE-associated GC and PC responses.

Given the effects of IRF7 on B cell transcriptome, translation, and metabolism, we measured IRF7 levels in B cell subsets in FcγRIIB^−/− mice. IRF7 expression was elevated in pre-GC and GC B cells relative to non-GC B cells, with the highest expression in GC B cells and much higher expression in FcγRIIB^−/− B cells than in B6 control B cells (Fig. 9, A and B). We also analyzed the published RNAseq dataset (Jenks et al., 2018) and found higher IRF7 expression in SLE B cells than in healthy control B cells (Fig. 9 C). In addition, we analyzed IRF7 expression in transitional 3, resting naïve, activated naïve, double negative (DN2) B cells, switched memory, and plasmablast/plasma cells (ASCs) from healthy and SLE samples. IRF7 expression was increased in activated B cells in SLE patients (Fig. 9 D) and during activation of B cells from healthy peripheral blood mononulcear cells (PBMCs) for 24 h with a TLR7 agonist (Fig. 9 E).

Next, to determine whether IRF7 functioned in a B cell–intrinsic manner to promote spontaneous autoimmune GC and PC responses and autoantibody production, we generated B cell–specific mixed BM chimeras. 20% BM cells from FcγRIIB^−/− or FcγRIIB^−/−IRF7^−/− mice were mixed with 80% BM cells from mu mutant (μMT) mice, which lack B cells, in a 20:80 ratio and transferred into lethally irradiated μMT recipient mice as depicted in Fig. 9 F. This approach created chimeric mice in which B cell compartment was 100% deficient in IRF7 and non-B cell compartment (T cells and myeloid cells) was 80% wild type with 20% IRF7 deficient. Serum ANA positivity and dsDNA- and SmRNP-reactive autoantibody titers in recipient mice with B cells deficient in IRF7 were significantly reduced compared with recipient mice with B cells sufficient in IRF7 (Fig. 9, G–I). However, no major differences in spleen weight (Fig. 9 J) and frequencies of naïve (Fig. 9 K) and activated (Fig. 9 L) B cell populations were observed between the strains except for an increased percentage of total B cells in the absence of IRF7 in B cells. B cell–specific IRF7-deficient mice had reduced frequencies of total CD4 and CD4 Teffs, although no differences in the frequency of Tfh (Fig. 9 M) and myeloid cells (Fig. 9 N) were observed between the strains. These data indicate a B cell–intrinsic requirement of IRF7 in loss of immune tolerance, autoantibody production to DNA- and RNA-associated autoantigens, and CD4⁺ Teff responses, whereas both B cell–intrinsic and –extrinsic contributions of IRF7 are required for driving GC, PC, and Tfh cell responses in SLE-like autoimmunity.

Despite a longstanding association of IRF7 with SLE, the cell-intrinsic role and mechanism by which IRF7 promotes spontaneous SLE-associated GC and PC responses that harbor autoreactive B cells in SLE are not known. Here, using IRF7-deficient mice, competitive mixed BM, and B cell–specific BM chimeras, we identified previously unknown B cell–intrinsic and –extrinsic roles for IRF7 in promoting spontaneous autoimmune GC and PC responses. We found that IRF7 is involved in B cell differentiation into autoantibody-producing cells, steady-state total and ANA-specific class-switched IgG antibody responses, and SLE pathogenesis. IRF7, however, was dispensable for promoting foreign antigen–driven GC, PC, and antibody responses. Through the single-cell transcriptome analysis of SLE-prone B cells, we uncovered the functions for IRF7-driven transcriptional programs in B cell differentiation through pre-GC into GC and PC pathways during systemic autoimmune responses. Analysis of ChIP-seq data identified IRF7 target genes that are differentially expressed in IRF7-deficient SLE-prone B cells. Mechanistically, IRF7 promotes systemic autoimmunity by regulating B cell transcriptome, translation, and metabolism required for B cell activation and differentiation into GC and PC populations during SLE-like autoimmune responses.

Previous studies highlighted the roles of both the extrafollicular (EF) (Jenks et al., 2018; William et al., 2002) and GC pathways (Arkatkar et al., 2017; Cappione et al., 2005; Degn et al., 2017; Jackson et al., 2016; Soni et al., 2014; Tiller et al., 2010; Vinuesa et al., 2009) in the development of autoreactive B cells and pathogenic Ab production in SLE-prone mice and patients. In particular, using the FcγRIIB^−/− SLE mouse model used in the current study, Tiller et al. previously showed the development of autoreactive GC B cells and PCs in which IgG autoantibodies were enriched in GCs and somatically hypermutated (Tiller et al., 2010), highlighting the role of the GC pathway in pathogenic autoantibody production. By blocking the GC pathway, previous studies highlighted a pathogenic role for the EF pathway in systemic autoimmunity (Brown et al., 2022; Voss et al., 2022). We observed dampened GC and PC responses, low numbers of autoantibody-producing splenic and BM AFCs, and reduced serum ANA titers in FcγRIIB^−/− mice in the absence of IRF7. Together, our current findings and the previously described role of the GC pathway in pathogenic Ab production in FcγRIIB^−/− mice (Tiller et al., 2010) strongly suggest the role of IRF7 in the generation of autoreactive PCs and pathogenic Abs through the GC pathway. Several studies have demonstrated that ABCs generated through EF pathways can either differentiate into EF plasmablasts or participate within the GC (Nickerson et al., 2023; Ricker et al., 2021). Deeper understanding of the differential mechanisms by which GC and EF pathways promote high-affinity autoantibody-producing PCs in SLE is warranted to develop targeted therapeutics.

Previous studies have delineated the contributions of IRF5, including a B cell–intrinsic role of IRF5, in promoting systemic autoimmunity and disease (Ban et al., 2021; Banga et al., 2020; Cham et al., 2012; Pellerin et al., 2021; Richez et al., 2010). Here, for the first time, we identified the role of IRF7 in promoting PC and GC responses and autoantibody production. It is thought that there is a significant overlap between IRF5 and IRF7 functions; however, it is not clear whether IRF7 homodimerizes or heterodimerizes with IRF5 or other IRF family members (i.e., IRF3) in B cells to promote GC and PC responses in systemic autoimmunity. The family of IRF proteins is well-known to participate in transcriptional coregulation of gene expression (Andrilenas et al., 2018). Studies focusing on the dynamic interactions and the transcriptional landscape regulated by IRF proteins singularly or cooperatively will help develop a deeper understanding of the mechanisms of IRF7-mediated regulation of B cell responses in SLE.

Through the generation of B cell–specific BM chimeras, we show here a B cell–intrinsic requirement of IRF7 in loss of tolerance and autoantibody production. Our data suggest that B cell–intrinsic IRF7 controls GC and/or PC tolerance checkpoints of autoreactive B cell selection. Our BM chimeric data, OMICS, and mechanistic data together suggest that IRF7-regulated transcriptional programs, translation, and metabolic reprogramming in B cells play critical roles in breaking tolerance and autoreactive B cell selection. B cell–intrinsic IRF7 deficiency, however, was not sufficient to affect overall spontaneous PC, GC, and Tfh responses. These data suggest that development of PC, GC, and Tfh responses in the BM chimeric environment requires both B cell–intrinsic and –extrinsic IRF7. Although correlative, IRF7 is thought to contribute to SLE in a B cell–extrinsic manner by inducing T1-IFNs in pDCs (Honda et al., 2005; Huang et al., 2015; Swiecki and Colonna, 2015). Whether B cell–extrinsic IRF7 contributions to autoimmune PC, GC, and Tfh responses in SLE models are mediated by pDC-derived T1-IFN signaling remain to be delineated, especially given that IRF7 deficiency promoted an increased frequency of pDCs in our competitive BM chimeric mice (Fig. 5 M). Interestingly, in the same chimeric mice, we observed a fivefold reduction in the frequency of iDCs in the absence of IRF7 (Fig. 5 M). We also observed reduced number of iDCs in the kidneys of IRF7-deficient FcγRIIB^−/−Yaa mice. DCs played a major role in priming T cells and in SLE development (Klarquist et al., 2016; Seitz and Matsushima, 2010). Given our data from the BM chimera experiments that indicated both B cell–intrinsic and –extrinsic roles of IRF7, delineation of the mechanism by which pDCs and iDCs contribute to autoimmune PC, GC, Tfh, and autoAb responses in SLE models is warranted. Identifying such mechanisms will require an in vivo system, not yet available, in which IRF7 can be deleted in iDCs and pDCs in SLE-prone mice.

Several risk variants in and around IRF7 are associated with SLE disease activity (Fu et al., 2011; Salloum et al., 2010). A mutation in the IRF7 gene resulting in Q412R is associated with elevated IRF7 expression and an increased T1-IFN signature in SLE patients (Fu et al., 2011). However, it has remained unclear whether the IRF7 risk allele contributed to autoimmune GC and PC responses in SLE in a cell–intrinsic manner independent of T1-IFN production. Generation of mouse models carrying IRF7 risk variants will help address this unanswered question. Here, we have demonstrated that outside of the T1-IFN genes, IRF7 can regulate genes in GC B cells and PCs that are involved in translation and metabolism, which are critical for B cell differentiation through GC and PC pathways in a systemic autoimmune response.

A previous study demonstrated the role of IRF7 in promoting autoantibody responses in a chemically (pristane)-induced B6 model of SLE (Miyagawa et al., 2016). Surprisingly, despite reductions in IC deposition in the kidney of IRF7^−/− mice, kidney pathology and proteinuria were reported in these mice (Miyagawa et al., 2016). The role of IRF7 in autoimmune GC and PC responses and disease in spontaneous SLE models, which may involve differential mechanisms to drive autoimmunity, has remained unclear. We observed reductions in autoantibody production, IC deposition, kidney pathology, and serum BUN/creatinine levels in IRF7-deficient spontaneous SLE-prone mice. We also observed significantly improved survival in TLR7-accelerated SLE-prone mice deficient in IRF7. Our data suggest that IRF7 plays a major role in promoting GC and PC responses and systemic autoimmunity, including tissue inflammation and disease. The discrepancy between our findings and the previous study could be due to differences in the involvement of differential mechanisms between the spontaneous SLE-prone versus chemically induced non-autoimmune B6 mouse models that were used.

We utilized scRNAseq analysis to profile subpopulations of B cells and their differentiation trajectories in SLE-prone mice. Pseudotime trajectory analysis of differentiating B cells revealed that during a systemic autoimmune response, B cells differentiate into two distinct pathways in which one group differentiated through a pre-GC transitional bridge into GC and PC fates, and the other group into naïve activated and switched activated B cells. By comparing single-cell transcriptomic data of FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− B cells, we discovered a significant deficit in IRF7^−/− B cell differentiation into GC and plasmablasts/PC clusters. Through targeted re-clustering of pre-GC transitional cells, we further discovered that B cell differentiation was halted at the pre-GC1 stage in the absence of IRF7, in which the majority of these cells were quiescent compared with B cells in FcγRIIB^−/− mice that progressed to pre-GC2 and pre-GC3 stages and were in the cell cycle. Our OMICs and validation data together suggest that during autoimmune B cell responses in spontaneous SLE-prone mice, IRF7 induces transcription of genes required for protein translation and metabolism in B cells. These IRF7-driven events in B cells, together with B cell–extrinsic factors, likely T1-IFN signaling, are required to meet the demands of activated B cells that can differentiate into GC and PC fates, promoting the generation of pathogenic autoantibody-producing cells and leading to SLE development.

Our identification of critical B cell–intrinsic and –extrinsic roles for IRF7 in regulating their differentiation into the GC and PC pathways does not negate the potential IRF7 role in regulating other B cell populations. Although we focused on the role of IRF7 in B cell differentiation into the GC/PC clusters, we identified differentially expressed genes in the naïve activated, switched activated, and MZ B cell clusters that will be a topic of further investigation at the mechanistic level. Additionally, based on our current findings suggesting both B cell–intrinsic and –extrinsic roles for IRF7 in autoimmune GC and PC responses and in the loss of B cell tolerance in systemic autoimmunity, we primarily focused our investigations into the IRF7 role in B cell responses. We, however, cannot exclude the possibility that IRF7 may contribute to the functionality of myeloid or T cell populations and thereby contribute to the autoimmune B cell response and autoimmunity. Thus, interrogation of IRF7 functions in other cell types in promoting autoimmune GC and PC responses and systemic autoimmunity also will be necessary in the future.

The overall goal of this study was to delineate the cell-intrinsic and -extrinsic mechanisms by which IRF7 promotes autoimmune B cell responses in systemic autoimmunity. This was accomplished by first generating IRF7-deficient, SLE-prone mice and then by generating competitive and B cell–specific mixed BM chimeras. The phenotypic examination of various B and T cell populations was performed by flow cytometry and immunofluorescence (IF) microscopy. Mechanistic studies were performed through single-cell transcriptome and ChIP-seq of B cells from SLE-prone mice sufficient and deficient in IRF7. The number of mice per experiment was determined based on the power calculation of our published data (Chodisetti et al., 2020; Domeier et al., 2018; Fike et al., 2021). Littermate controls were used where possible. Control and experimental mice were age- and sex-matched. All data points generated in various experiments were included in the analysis. All experimental findings were replicated, and the number of replicates is indicated in the figure legends.

C57BL/6 (B6), B6.SJL-Ptprc^aPepc^b/BoyJ (CD45.1), and B6.129S2-Ighm^tm1Cgn/J (B6.μMT) mice were originally purchased from the Jackson laboratories and bred in-house. B6.129P2-Irf7^tm1Ttg/TtgRbRc (IRF7^−/−) mice were purchased from Riken BRC and bred internally. FcγRIIB^−/− mice described previously (Bolland et al., 2002; Shinde et al., 2018) were backcrossed to B6 for several generation before we further backcrossed to B6 for at least N12–14 generation. These mice were intercrossed and then crossed with IRF7-deficient (FcγRIIB^−/−IRF7^−/−) mice. The genome scan of FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− mice from our colony performed by the Jackson Laboratories showed that FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− mice contained >99% and >97% B6 genome, respectively. FcγRIIB^−/−yaa mice were generated by crossing FcγRIIB^−/− mice to B6.yaa mice (from the Jackson Laboratories) internally. All crosses generated for this manuscript were performed in-house. All animal studies were conducted at Pennsylvania State University Hershey Medical Center in accordance with the guidelines approved by our Institutional Animal Care and Use Committee. Animals were housed in a specific pathogen–free barrier facility.

These mice were generated by lethally irradiating 10–12-wk-old female CD45.1 hosts in two doses of 450 rads (X-RAD 320iX Research Irradiator; Precision X-Ray) within a 4-h interval. Within a few hours of irradiation, each recipient mouse received 10 × 10⁶ T cell–depleted BM cells intravenously composed of 5 × 10⁶ cells from FcγRIIB^−/− CD45.1⁺/CD45.2⁺ and 5 × 10⁶ cells from FcγRIIB^−/−IRF7^−/−CD45.2⁺ donors (50:50). Recipient mice were analyzed 8 wk after BM reconstitution.

8–10-wk-old female B6.μMT (B6.129S2-Igh^mtm1Cgn/J) recipient mice were lethally irradiated with 900 rads of x rays (X-RAD 320iX Research Irradiator; Precision X-Ray). Within 4–6 h after irradiation, each recipient intravenously received 10⁷ T cell–depleted BM cells isolated from 8 to 10-wk-old female donor mice, with 80% of cells from B6.μMT mice and 20% from FcγRIIB^−/− or FcγRIIB^−/−IRF7^−/− mice. Recipients were analyzed after 3 mo.

Age- and sex-matched non-autoimmune B6 (8–10 wk old) and autoimmune-prone FcγRIIB^−/− (16–17 wk old) mice were immunized i.p. with 200 µg of NP-KLH (Biosearch Technologies) mixed with CFA (Sigma-Aldrich). Spleens were harvested and processed for various analyses on indicated day after immunization.

Analysis of splenic populations was performed by processing the tissue into a single-cell suspension followed by RBC lysis using Tris-ammonium chloride. Splenocytes were stained using the antibodies listed in the table below (Table 1). For BM analysis, cells were flushed from isolated femurs with RPMI, followed by RBC lysis and staining. For the flow cytometry of the kidney, the kidneys were perfused prior to harvesting. Kidneys were minced into small pieces using a scalpel and incubated with collagenase Type I (Sigma-Aldrich) for 30 min at 37°C prior to RBC lysis and staining. For intracellular staining, cells were fixed using BD Cytofix (BD Biosciences) and permeabilized using BD Phosflow Perm Buffer III according to the manufacturer’s instructions. Staining of unfixed cells was performed at 4°C except for anti-CXCR5 Ab, which was performed at room temperature. Stained cells were analyzed on a BD LSR II or BD FACSymphony A3 (BD Biosciences) using FACSDiva Software. Sorting experiments were performed on a BD ARIA SORP high-performance cell sorter (BD Biosciences). Flow cytometry data were analyzed using FlowJo Software version 10.8.1 (BD) (Table 1).

Kidneys from 6-mo-old mice were fixed in 10% neutral buffered formalin and embedded in paraffin. Kidney sections were cut at 3-μm thickness for periodic acid–Schiff. The slides were scanned on a TissueGnostics TissueFAXS SL imaging system with Kromnigon Spectra Split 7 filters and the TissueFAXS 71.1.138 software. Renal disease was evaluated with a semiquantitative 0–4+ scale as described (Chan et al., 1997). The scale was used for each compartment (glomerular, interstitial, and vascular), with the pathology graded according to specified criteria as absent (0–1), mild (1–2), moderate (2–3), or severe (3–4+). For comparative purposes, all the tissue sections were scored by one observer (R. Caricchio), who was blinded to the origin of the kidney samples.

To detect serum ANAs, ELISA plates (Thermo Fisher Scientific) were first coated with poly-L-lysine, followed by salmon sperm dsDNA (Invitrogen), nucleosome (histone from Sigma-Aldrich on a layer of dsDNA coating), or SmRNP (Arotec Diagnostics). To detect NP-specific Abs, plates were coated with 10 μg/ml NP4-BSA or NP29-BSA. Plates were washed and blocked with 5% FBS in PBS, and serum was added starting at a 1:50 dilution in 5% FBS, followed by 1:2 serial dilution for the remainder of the plate. Antibodies were then detected with one of the following biotinylated antibodies: goat anti-mouse IgM (Jackson Immunoresearch Laboratories), goat anti-mouse IgG (Jackson Immunoresearch Laboratories), or goat anti-mouse IgG2c (Southern Biotech). Plates were then washed and incubated with streptavidin-alkaline phosphatase (Vector Laboratories). P-nitrophenyl phosphate (disodium salt) (Thermo Fisher Scientific) substrate for alkaline phosphatase was used for developing the plates, and plates were read at λ405 nm on Synergy H1 (BioTek Instruments). Serum BUN and creatinine levels were measured via the Urea Nitrogen (BUN) Colorimetric Detection Kit (Arbor Assays) and the Enzymatic Creatinine Test Kit (Diazyme), respectively, according to the manufacturer’s instructions. All ELISA plates were read on Synergy H1 (BioTek Instruments) at the wavelengths indicated by the manufacturer’s instructions.

For detection and quantification of autoantibody producing AFCs, multiscreen 96-well filtration ELISpot plates (Millipore Sigma) were coated with salmon sperm dsDNA (Invitrogen), nucleosome (histone from Sigma-Aldrich on a layer of dsDNA coating), or SmRNP (Arotec Diagnostics). To detect NP-specific Abs, ELISpot plates were coated with 10 μg/ml NP4-BSA or NP29-BSA. Plates were blocked with 5% FBS in PBS. 2 × 10⁶ splenocytes suspended in RPMI 1640 (Corning) containing 10% FBS (Hyclone GE), penicillin-streptomycin (Corning), and Glutamax (Corning) were added to the top row and serially diluted by twofold to the remainder of the rows. Plates were washed, and autoantibody-producing AFCs were detected by incubating with biotinylated anti-κ antibody (Invitrogen) followed by streptavidin-alkaline phosphatase (Vector Laboratories). Plates were then washed and developed via Vector Blue Alkaline Phosphatase Substrate Kit III (Vector Laboratories). ELISpot plates were analyzed using an Immunospot S6 Universal M2 (CTL).

For the detection of ANA seropositivity, serum from mice was diluted 1:50 and incubated on a Hep-2 slide (Antibodies Incorporated). ANAs were then detected using an FITC-rat anti-mouse k antibody (H139-52,1; Southern Biotech). ANA slides were visualized using a Leica DM4000 fluorescence software (Leica Microsystems).

Splenic and kidney IF microscopy was performed by embedding the tissue in OCT compound (Thermo Fisher Scientific) and snap freezing over liquid nitrogen. 5-µM sections were cut and mounted on a ColorFrost Plus Microscope Slide (Thermo Fisher Scientific) and fixed in cold acetone for 20 min. For splenic GC staining, tissue sections were stained with the following antibodies: GL-7 FITC and anti–IgD-APC (11–26c2a; BD Biosciences) and anti–CD4-PE (GK1.2; BioLegend). GC measurements were performed on a random selection of GL-7⁺ staining across the splenic tissue section where total GC area (µM²) was measured using the Leica Application Suite Advanced Fluorescence quantification tool. Kidney sections were stained with anti–CD3-FITC (Immunology Consultant Laboratory) and anti–IgG-PE (Abcam) for the detection of IC depositions. Stained tissue sections were imaged on a Leica DM4000 Fluorescence microscope (Leica Microsystems).

Splenic B cells were isolated from indicated mice via negative selection (StemCell Technologies). Cells were resuspended in XF DMEM pH 7.4 media supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose and plated on poly-L-lysine–coated microchamber wells. OCRs and extracellular acidification rates were measured using a XF HS mini analyzer (Agilent). For the mito stress test, 1.5 μM oligomycin, 2 μM Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), and 0.5 μM antimycin/rotenone were utilized. Data were analyzed using the Agilent Seahorse Wave Software.

Single-cell suspension of splenocytes was prepared as described above. Splenocytes were incubated for 30 min in methionine-free RPMI 1640 medium with 10% of dialyzed FBS at 37°C. Cells were cultured for an additional 2 h for L-Homopropargyl glycine (HPG) incorporation (final concentration 100 μM). Cycloheximide- (100 µg/ml) or puromycin- (10 µg/ml) treated samples were used as negative controls. HPG incorporation in GC B cells and PCs was stained with the Click-iT Plus Alexa Fluor 488 Picolyl Azide Toolkit using the Foxp3 fixation-permeabilization method and was detected by flow cytometry.

PBMCs were isolated from two healthy donor whole blood samples using a Ficoll gradient (BD) under an Institutional Review Board–approved protocol at Penn State College of Medicine. B cells were then isolated from the PBMCs using StemCell EasySep Human B cell Isolation Kit. Isolated B cells were stimulated overnight with 300 ng/ml of R848 in 10% RPMI. RNA was isolated from B cells by the Qiagen RNeasy Kit (Qiagen) according to the manufacturer’s instructions. RNA quality and quantity were assessed via NanoDrop prior to cDNA synthesis (Thermo Fisher Scientific). cDNA was generated using the High-Capacity cDNA Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. TaqMan primers for IRF7 (Hs01014809_g1) were used. Expression levels were normalized by a housekeeping gene (β-actin), and fold change was calculated by the 2(ΔΔCt) method.

Spleens from two 3-mo-old FcγRIIB^−/− and two FcγRIIB^−/−IRF7^−/− mice were processed for cell sorting as described above. CD3^-B220⁺CD19⁺IgD⁻ and CD3⁻B220⁺CD19⁺IgD⁺ cells were sorted from each animal and mixed at a ratio of 9:1 (IgD⁻: IgD⁺) prior to library preparation. Library preparation and sequencing were performed by the Center for Applied Genomics Core at the Children’s Hospital of Philadelphia. Libraries were prepared using the 10x Genomics Chromium Single Cell 3′ Reagent Kit v2 as per the manufacturer’s instructions. Sequencing was performed on an Illumina HiSeq sequencer. Data were then processed through the CellRanger pipeline (10x Genomics). Demultiplexing and alignments were performed against the mm10 transcriptome. Analysis was performed via Seurat package (v. 4.99) within R. Cells expressing <200 or >4,000 unique genes were excluded from further analysis. Cells with >10% mitochondrial gene expression were also excluded. Log normalization and scaling of features were completed before principal component dimensionality reduction, clustering, and visualization via UMAP. Pseudotime analysis was performed using the Monocle3 package in R. A Wilcoxon rank-sum test with a Bonferroni multiple testing correction was used to identify differentially expressed genes between FcγRIIB^−/− and FcγRIIB^−/−IRF7^−/− groups. Gene pathway-based analyses from the “Pre-GC transition” subclusters were performed using Ingenuity Pathway Analysis Software (Qiagen).

Spleens from 10 5–4-mo-old FcγRIIB^−/− mice were pooled in duplicate (five mice in each sample) and processed for cell sorting as described above. CD3⁻B220⁺CD19⁺IgD⁻ B cells were sorted and prepared for ChIP as previously described (Fike et al., 2023). Briefly, cells were fixed using 1.5% formaldehyde for 15 min at 37°C followed by glycine to stop the reaction. Samples were processed further using Cell Signaling ChIP Plus Sonication Chromatin IP Kit (Cell Signaling) according to the manufacturer’s instructions, sonication was performed on a Biorupter PICO (Diagenode). Library prep and sequencing were performed by the Center for Applied Genomics Core at the Children’s Hospital of Philadelphia. Data were processed and analyzed in Bioconductor. Sequences were aligned to reference genome mm9. Peaks were called against input controls using Macs2. Peaks were visualized in IGV.

All data were plotted as the mean of the data ± SEM, unless otherwise indicated. For all datasets, the normality of the data was first assessed based on the number of samples via D’Agostino–Pearson omnibus normality test or the Shapiro–Wilk normality test. Statistical significance between two groups was determined using the Student’s t test when two groups were compared. For datasets containing more than two groups and multiple time points, a two-way ANOVA with a Dunn–Sidak correction for multiple comparisons was performed. For the survival curve, a log-rank (Mantel–Cox) test was performed. The statistical test used is denoted in each figure legend. Statistical analyses were performed using Prism GraphPad software version 6 or 10 (GraphPad Software). P values are shown as NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 for significance.

Fig. S1 describes the reduced spleen size, splenocyte number, and the early stage serum autoantibody titers in the absence of IRF7. Fig. S2 describes the reduced numbers of steady-state IgG1- and IgG2c-producing AFCs and total serum IgG1 and IgG2c titers in the absence of IRF7. Fig. S3 evaluates the effects of IRF7 deficiency on the number of myeloid cells in SLE-prone FcγRIIB^−/− mice. Fig. S4 analyzes the numbers of splenic and BM AFCs and myeloid cell populations. Fig. S5 identifies the B cell clusters through the analysis of single-cell transcriptome of SLE-prone B cells from IRF7-sufficient and -deficient FcγRIIB^−/− mice.

All data are available in the main text or in the supplementary materials. scRNAseq (accession number GSE264033) and ChIP-seq (accession number GSE264034) data have been deposited at the NCBI Gene Expression Omnibus and are publicly available. Accession numbers are also listed in Table 1. Any additional information required to reanalyze the data reported in this paper will be available from the lead contact upon request.

We would like to thank the Pennsylvania State University Hershey Medical Center flow cytometry core facility and Department of Comparative Medicine for help with flow cytometric experiments and animal housing and care, respectively. We thank the support staff within the central facility of the Department of Microbiology and Immunology at Pennsylvania State University College of Medicine. We would like to acknowledge the contributions of Dr. Renata Pellegrino Da Silva, Fernanda Thompson, Jacqueline Smiler, and the other members of the Center for Applied Genomics at CHOP for their help with the scRNAseq and ChIP-seq. We acknowledge the Sanderson Center for Optical Experimentation (SCOPE), Massachusetts Life Science Cener (the SCOPE RRID is SCR_022721) and Leah Whiteman for technical assistance.

This work was primarily supported by the National Institutes of Health grant to Z.S.M. Rahman: RO1 AI162971. This project was partially funded by the following grants: The Judy S. Finkelstein Memorial Research Award at the PennState College of Medicine (A.J. Fike), American Association of Immunology Careers in Immunology Fellowship (A.J. Fike), National Cancer Institute 5T32CA060395-25 (A.J. Fike), Lupus Foundation of America Goldie Simon Preceptorship Award (A.J. Fike), National Heart Lung and Blood Institute grant to M. Trebak: R35 HL150778, and the Pennsylvania Department of Health Tobacco CURE Funds.

Author contributions: A.J. Fike: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, validation, visualization, and writing—original draft, review, and editing. K.N. Bricker: conceptualization, data curation, formal analysis, investigation, methodology, validation, and writing—review and editing. M.V. Gonzalez: data curation, formal analysis, investigation, methodology, software, visualization, and writing—review and editing. A. Maharjan: investigation, validation, and writing—review and editing. T. Bui: resources. K. Nuon: resources. S.M. Emrich: investigation and methodology. J.L. Weber: investigation. S.A. Luckenbill: investigation. N.M. Choi: investigation. R. Sauteraud: formal analysis, software, and visualization. D.J. Liu: data curation, formal analysis, and supervision. N.J. Olsen: resources and writing—original draft, review, and editing. R. Caricchio: formal analysis, investigation, resources, and validation. M. Trebak: conceptualization, funding acquisition, project administration, resources, supervision, validation, and writing—review and editing. S.B. Chodisetti: conceptualization, formal analysis, investigation, methodology, validation, and writing—review and editing. Z.S.M. Rahman: conceptualization, funding acquisition, project administration, supervision, validation, visualization, and writing—review and editing.