R2^-/- mice were obtained from the Taconic National Institute of Allergy and Infectious Diseases colony. IRF8^f/f mice have been described previously (21). XCR1-Cre (JAX#035435), CD11c-Cre (JAX#008068) and Itgax-Cre-EGFP (JAX#007567) mice were purchased from Jackson Laboratory. The construct to generate CD11c-Cre mice includes the entire Cd11c gene from BAC clone RP24-361C4 and replaces the first exon for Cre recombinase (28). In the construct to generate Itgax-Cre-EGFP mice, 5.3kb of the Itgax (CD11c) promoter/enhancer directs bicistronic Cre and EGFP protein expression (29). Genotyping and gene excision analysis were performed by Transnetyx. All mice were maintained under specific pathogen-free conditions. Animal studies were conducted according to a protocol approved by the National Institute of Allergy and Infectious Diseases Animal Care and Use Committee.

Spleen, lymph nodes, and bone marrow (BM) single cell suspensions were prepared and stained with antibodies listed in Table 1. Cells were analyzed with a FACSymphony flow cytometer (BD Biosciences) and FlowJo software (Treestar, Version 10.8.1).

For preparing kidney leukocytes, a detailed protocol has been reported recently (31). Briefly, mice were injected with anti-CD45-BV421 antibodies (3 μg in 200 μl) i.v. for 3 min and euthanized immediately. Kidneys were minced with a scissor and digested with the Multi Tissue Dissociation Kit 1 (Miltenyi Biotec) reagents on a GentleMACS Octo Dissociator (Miltenyi Biotec), followed by enrichment for CD45⁺ leukocytes using anti-CD45 microbeads (clone 30F11.1, Miltenyi Biotec) and AutoMACS Pro Separator sorting (Miltenyi Biotec). The eluted cells were then stained and analyzed by flow cytometry.

Approximately 1x10⁷ of mixed BM cells of Irf8^f/fItgax-Cre- (Igh^b), Irf8^f/fItgax-Cre+ (Igh^b), and R2^-/- (Igh^a) mice at a ratio of 1:1 was injected intravenously into lethally irradiated R2^-/- (Igh^b) mice that received a dose of 940 rad 1 day earlier. Three months later, the reconstituted mice were analyzed for autoantibody production and cellular distribution by flow cytometry.

Urinal protein levels were measured with Chemstrip 2GP urine test strips (Roche) according to the manufacturer’s instruction. A Chemstrip was dip into freshly voided urine specimen. A color change from yellow to light green/green occurred within 2 min. Results were obtained by direct visual comparison with the color scale printed on the vial label. Proteinuria was continuously monitored from once daily to once weekly. Protein concentration scores of 0, +1, +2, +3 and +4 correspond to a protein concentration of 0, <30, 30, 100 and 500 mg/dL, respectively. Serum ANA titers were determined by the Hep-2 system described previously (32). Briefly, serum samples were diluted at 1:100, 1:300, 1:900 and 1:2700 with PBS and incubated with Hep-2 substrate slides (MBL, AN-1012) at room temperature for 30 minutes. After washing the slides twice with PBS for 5 minutes, the slides were incubated with a secondary anti-mouse IgG-Alexa488 antibody (minimal x-reactivity) (Biolegend) in the dark at room temperature for 30 minutes. In some experiments, secondary antibody was biotinylated anti-mouse IgG2a/c[a] or IgG2a/c[b] (Table 1), which was revealed by streptavidin-FITC. Afterwards, the slides were washed with PBS three times. Then, they were imaged under a fluorescence microscope and images were taken. ANA titers were scored as follows: 0= negative at 1:100 dilution; 1= positive at 1:100; 2= positive at 1:300; 3= positive at 1:900; 4= positive at 1:2700; 5= positive at 1:9100 dilution.

Kidney tissues were fixed and sectioned by American Histolabs (Gaithersburg, Maryland) for H&E staining. The glomerulonephritis scoring was done by measuring several pathological entities as reported previously (33). The slides were read by a pathologist independently and blindly. Images were taken with an Olympus BX41 microscope (10x and 40x objectives) equipped with an Olympus DP71 camera.

Data were analyzed and figures were made using GraphPad Prism (version 9.0.2). For pairwise comparisons, the appropriate parametric (unpaired Student’s t-test) or non-parametric (Mann-Whitney test) was performed. For multiple Mann-Whitney tests on the same set of data, the Bonferroni’s correction tests were carried out.

R2^-/- mice spontaneously develop ANA and proteinuria starting from 3 months of age, which leads to premature death after 5 months of age (4). Hyperresponsive B cells of R2^-/- mice are responsible for production of autoantibodies and the initiation of lupus-like symptoms. We observed that the number of inflammatory DCs in the kidney positively correlated with the severity of glomerulonephritis (34), raising the possibility that cDCs may contribute to the pathogenesis of autoimmune nephritis. To gain deeper insight of cDC subsets in regulation of autoimmunity, we used CD11c-Cre, a broadly employed transgenic model to target genes in DCs, to delete floxed Irf8, thereby depleting cDC1s (10). As an alternative approach, we used Itgax-Cre-EGFP to delete Irf8 by taking advantage of EGFP tracking of deleted cells. The Itgax gene encodes for CD11c and we therefore expect the same result as in CD11c-Cre lines. By crossing Irf8^f/f with these two Cre lines under the R2^-/- background, we generated IRF8-deficient (designated Irf8^f/fCD11c-Cre+R2^-/- and Irf8^f/fItgax-Cre+R2^-/-, respectively) and -sufficient (designated Irf8^f/fCD11c-Cre-R2^-/- and Irf8^f/fItgax-Cre-R2^-/-, respectively) mice. Littermate mice were used throughout this study.

Compared with Irf8^f/fItgax-Cre-R2^-/-control mice, the mortality of IRF8-targetted Cre+ mice was significantly improved (Figure 1A) and proteinuria were markedly decreased (Figure 1B). This result was consistent with an overall reduction of lethal kidney disease by conditional deletion of Irf8. We confirmed differences in kidney pathology by histological analyses and diagnosis encompassing several parameters, including endocapillary hypercellularity, karyorrhexis, fibrinoid necrosis, hyaline deposits, cellular/fibrocellular crescents, and interstitial inflammation. A combined pathologic score was assigned to each kidney. There was a greater reduction in pathological scores among mice with Cre-targeted Irf8 deletion than in IRF8-sufficient controls (Figures 1C, D). The numbers of kidney-infiltrating immune cells including CD4T, CD8T, monocytes and cDC1s were also markedly reduced in Irf8^f/fItgax-Cre+R2^-/- compared to controls (Figure 1E. Gating schemes were depicted in Supplementary Figure S1). Altogether, we concluded that Itgax-Cre-mediated IRF8 deletion abrogated lethal nephritis in R2^-/- mice.

Since lupus pathology in R2^-/- mice is predated by the presence of autoantibodies in circulation, we determined the titers of serum ANAs. Indeed, Itgax-Cre targeted Irf8 was correlated with almost complete abrogation of the phenotype (Figure 2A). Reduction in autoreactivity in Cre-expressing mice was also correlated with reduced splenomegaly (Figure 2B). Multiple-color flow cytometry analyses of splenocytes revealed several differences in immune cell numbers. Based on the expression of CD26, CD11b and CD24, cDCs (CD11c⁺MHC⁺) were subdivided into cDC1s (CD26⁺CD11b^-CD24⁺) and cDC2s (CD26⁺CD11b⁺CD24^-) (gating strategy was shown in Supplementary Figure S1). As expected, a near complete loss of cDC1s was found in Irf8^f/fItgax-Cre+R2^-/- mice compared to controls (Figure 2C), consistent with the reported essential role for Irf8 in cDC1 development (10). We found no significant differences in other immune cells in the spleen, such as cDC2s, monocytes and CD4Ts (Figures 2D, E; gating strategy shown in Supplementary Figure S1). While the total number of splenic B cells was similar in mice with conditional Irf8 deletion compared to controls (Figure 2E), we decided to estimate the numbers of several activated B cell populations given the stark differences that we had observed in autoantibody titers. Atypical B cells (ABCs) express CD11c and have been associated with autoreactivity in other systems (22). They were almost eliminated by the Itgax-targeted deletion of Irf8 in R2^-/- mice (Figure 2F). In addition, spontaneously activated B cells (germinal center (GC) and plasma cells) were markedly reduced in Irf8^f/fItgax-Cre+R2^-/- mice compared to controls (Gating strategy was shown in Supplementary Figure S1) (Figure 2F). This result suggests that Itgax-Cre expression might result in Irf8 deletion in B cells that normally don’t express CD11c.

Analyses of Irf8^f/fCD11c-Cre+R2^-/- mice revealed very similar results as Irf8^f/fItgax-Cre+R2^-/- mice. Irf8^f/fCD11c-Cre+R2^-/- mice exhibited prolonged survival (Supplementary Figure S2A), significantly reduced proteinuria (Supplementary Figure S2B) and serum ANA levels (Supplementary Figure S2C). The spleen weights of Irf8^f/fCD11c-Cre+R2^-/- mice were significantly reduced compared with controls (Supplementary Figure S2D). Multiple comparisons among immune cells seemed to show a trend in reduction in CD4T, B cells, and monocytes between Irf8^f/fCD11c-Cre+R2^-/- and control mice (Supplementary Figure S2E). However, none of these comparisons reached the threshold for significant differences when the Bonferroni correction was applied. As expected, the frequencies of cDC1s were dramatically decreased, whereas those of cDC2s were slightly increased (Supplementary Figure S2F). The numbers of ABCs were also decreased (Supplementary Figure S2H). In the kidney of Irf8^f/fCD11c-Cre+R2^-/- mice, the numbers of infiltrated leukocytes and lymphocytes were significantly reduced in CD4T, CD8T, monocytes and cDC1s (Supplementary Figure S2G), consisting with the pathological presentations (Supplementary Figure S2I, J). Taken together, the autoimmune manifestations in both Irf8^f/fCD11c-Cre+R2^-/- and Irf8^f/fItgax-Cre+R2^-/- mice were almost completely abrogated, highlighting the critical role of IRF8 in pathogenesis of lupus nephritis.

Analysis of tail genomic DNA revealed mixed genotype results in all Irf8^f/fItgax-Cre+R2^-/- mice but not Irf8^f/fItgax-Cre-R2^-/- controls, suggesting that mosaicism occurred when Itgax-Cre was present (Figure 3A). Flow cytometric analysis of IRF8 protein expression revealed a ~50% reduction of IRF8 in B cells, ABCs, cDC1s, cDC2s and F4/80⁺CD11b^- monocytes (Figure 3B). We confirmed mosaicism in EGFP expression of all immune populations: 100% in DCs, 30-50% in lymphocytes and 10-20% in monocytes (Figure 3C). We tested many B cell subpopulations as the main effect of Irf8 deletion was observed in the production of autoantibodies (Figure 2A). All B cell developmental populations and activated fractions tested showed an average of 50% EGFP positive cells in mice that did not genotype as fully heterozygous (Figures 3D, E). These data suggested that Itgax-Cre was expressed in unidentified hematopoietic precursors that gave rise to progeny immune cells broader than the well-known CD11c⁺ DCs with a result of somatic mosaicism in all immune cells. This result was consistent with a previous report that the CD11c-Cre transgene was found to be expressed in a variety of cell types beyond the expected DCs (35).

To circumvent the “leakiness” issue of Itgax or CD11c-Cre, we crossed Irf8^f/fR2^-/- with Xcr1-Cre+R2^-/- mice with the intention of deleting Irf8 in the cDC1 lineage exclusively, as previously reported (30). However, after three generations, we unexpectedly found complete deletion of Irf8 either at one copy or two copies among progeny mice (Supplementary Figure S3). This prevented production of meaningful mice for experiment.

To determine if IRF8 insufficiency and mosaicism in B cells affected autoantibody production, we generated chimeric mice using a 1:1 mixture of bone marrow cells from Irf8^f/fItgax-Cre+R2^-/- (Igh^b allotype), Irf8^f/fItgax-Cre-R2^-/- (Igh^b allotype), or wild-type R2^-/- (Igh^a allotype) mice (Figure 4A). In this setting, B cell intrinsic effects due to Irf8 deficiency are observed as changes in the ratio between Igh^a and Igh^b allotypes among groups. These two alleles will produce distinct surface IgM allotypes in B cells measurable by flow cytometry. The two alleles also produce distinct released IgG antibodies measured by indirect immunofluorescence with anti-IgG_2a/c^a and anti-IgG_2a/c^b antibodies (lowercase script denotes antibody isotype, while superscript denotes gene allele and, consequently, the donor origin). Three months after reconstitution, splenocytes were analyzed by flow cytometry and serum ANA levels (IgG allotype “a” or “b”) were measured by staining Hep2 cells. The proportion of each donor allotype in various B cell populations was calculated with the gating scheme shown in Figure 4B. The yield of spleen B cells (Figure 4C) and the ratio of donors (IgM^b: IgM^a) in naïve B cells (Figure 4D) were comparable between Group1 and Group 2. However, the frequencies of GC B cells in Group 2 were reduced compared to Group 1 (Figure 4C). Although there was minimum skewing in naïve B cells, ABCs and GC cells were significantly reduced in chimeras that contained 50% cells originated from Itgax-Cre-expressing mice compared to those that contained control Cre-negative populations (Figure 4E). The reduction was exclusively observed in “b” allotypes (red colored data comparing group 1 and group 2) while “a” allotypes (colored blue) were mostly unchanged between group 1 and group 2. Furthermore, B cells from Irf8^f/fItgax-Cre+R2^-/- donors produced significantly reduced levels of serum ANA antibodies (detected as IgG2a/c allotype “b” in Group 2) compared to IRF8-sufficent B cells (detected as IgG2a/c allotype “b” in Group 1) (Figure 4F). We did not observe B cell extrinsic effects of Irf8 deletion in this experiment (i.e. effect of other cells that might express Irf8), because none of the phenotypes in the non-modified (both WT Irf8) allele “a” were reduced in group 2 compared to group 1. Taken together, these data suggest that IRF8 insufficiency in B cells impairs autoantibody production and that partial deletion of Irf8 as mosaic expression is enough to fully abrogate spontaneous germinal centers and measurable autoreactivity.