Xist ribonucleoproteins promote female sex-biased autoimmunity

SUMMARY

Autoimmune diseases disproportionately affect females more than males. The XX sex chromosome complement is strongly associated with susceptibility to autoimmunity. Xist long noncoding RNA (lncRNA) is expressed only in females to randomly inactivate one of the two X chromosomes to achieve gene dosage compensation. Here, we show that the Xist ribonucleoprotein (RNP) complex, comprised of numerous autoantigenic components, is an important driver of sex-biased autoimmunity. Inducible transgenic expression of a non-silencing form of Xist in male mice introduced Xist RNP complexes and sufficed to produce autoantibodies. Male SJL/J mice expressing transgenic Xist developed more severe multiorgan pathology in pristane-induced model of lupus than wild-type males. Xist expression in males reprogrammed T and B cell population and chromatin states to more resemble wild type females. Human patients with autoimmune diseases displayed significant autoantibodies to multiple components of XIST RNP. Thus, a sex-specific lncRNA scaffolds ubiquitous RNP components to drive sex-biased immunity.

In Brief

The Xist RNA protein complex, present only in females, is immunogenic and may underlie female-biased autoimmunity.

Graphical Abstract

INTRODUCTION

Autoimmune diseases are the third-most prevalent disease category, outpaced only by cancer and heart disease¹. Four out of five patients with autoimmune diseases are female. For instance, in systemic lupus erythematosus (SLE), the ratio of patient sex is 9:1 females to males; the ratio in Sjogren’s disease is 19:1 female to male patients^2,3. Although hormones have been extensively studied⁴, the dosage of X chromosome appears to be a major driver of autoimmune risk irrespective of sex or hormonal status in humans and mice^5–8. Patients with Klinefelter syndrome (XXY) are phenotypically males, have male hormonal pattern, but have an elevated risk of autoimmune disease equivalent to females. Specific X-linked genes, such as TLR7, that can escape X inactivation have been nominated as contributors to specific autoimmune diseases^5–8. The genetic risk underlying autoimmune diseases from the second X chromosome in aggregate remain unresolved. In addition, identical twin studies have also shown varying degrees of autoimmune disease penetrance, suggesting a genetic disposition that is also reliant on environmental factors^9,10. Hence, adjuvant triggers¹¹ in addition to genetic predisposition may be initiators of autoimmune disease development.

Mammalian females have a XX genotype and males have a XY genotype. To make the gene expression output roughly equivalent between females and males, every cell in a female’s body epigenetically silences one of two X chromosomes via the action of the long noncoding RNA Xist. Xist is a ~17 kb lncRNA (19 kb in human) that is transcribed only from the inactive X chromosome, and thus not expressed in males. Xist is critical for the establishment of X chromosome inactivation (XCI) spreading from the X-inactivation center and coating the entire inactive X in association with its protein partners. During XCI establishment in mouse embryonic stem cells, Xist associates with 81 unique binding proteins to form an RNP complex, 10 through direct RNA-protein interaction and others through indirect protein-protein interaction^12,13. Xist is widely expressed in adult somatic tissues and associates additional tissue-specific proteins¹⁴. Several Xist binding proteins were previously noted to be autoantigens¹². Studies in SLE patients and mice demonstrated that DNA-autoantibody and RNA-autoantigen immune complexes, such as Sm/RNP and U1A, activate the TLR7, TLR8 and TLR9 pathways of the innate immune system^15–17. The XIST RNP, comprised of a lncRNA, bound RNA binding proteins, and tethered to pieces of genomic DNA, presents qualities resembling nucleic acid-autoantigen immune complexes.

In order to study the impact of the XIST RNP in autoimmune predilection independent of sex chromosome or hormonal background, we utilized an inducible and non-silencing allele of Xist introduced into an autosome in the autoimmune-resistant C57BL/6J and autoimmune-prone SJL/J strain backgrounds. Inducing transgenic Xist RNP formation in male animals allowed the study of this female-specific lncRNA in a male background using a chemically-induced SLE model. Both increased disease severity and elevated autoreactive lymphocyte pathway signatures were observed in the mouse models of pristane-induced SLE. Concurrently, we designed an antigen array to test autoimmune patient seroactivity to XIST-associating proteins and detected significant reactivity towards multiple components of the XIST RNP. Altogether, our data point to a significant role for the Xist RNP as a driver for autoimmunity that may underly the sex-biased female preponderance for developing autoimmune diseases.

RESULTS

Xist RNPs as autoantigens in human disease

A defining feature of many autoimmune diseases is the development of antibodies against self proteins, termed autoantibodies. Many autoantibodies are directed toward nuclear RNA binding proteins, and the nature and titer of such autoantibodies define the type and severity of autoimmune diseases in clinical practice. We and others have identified the constellation of RNA binding proteins associated with Xist RNA in several cell types^12–14. Bibliomic analysis revealed that 30 proteins of Xist RNP constituents have been reported as the targets of autoantibodies (i.e. autoantigens) in one or more human diseases (Figure S1, Table S1). This observation stimulated the hypothesis that Xist RNP may promote female-biased autoimmunity.

Developing the TetOP-ΔRepA-Xist transgenic mouse to model autoimmune diseases

To test Xist ribonucleoprotein as a potential trigger of autoimmunity, we developed a TetOP-ΔRepA-Xist transgenic mouse that enables inducible expression of Xist in male animals. Because Xist expression from an autosome silences the chromosome in cis and is often cell lethal, we chose to use ΔRepA-Xist, a truncation of Xist that removes the A-repeat (RepA) element required for gene silencing activity of Xist¹⁸, but does not ablate chromosome coating or Xist RNP formation. Previous study indicated that 78 of 81 proteins in the Xist RNP associates with ΔRepA-Xist¹². Expression of the ΔRepA mutant Xist is controlled through the Tet-operon promoter (TetOP), and the transgenic cassette is inserted in the Col1A1 locus on chromosome 11 (Figure 1A). Since Xist is expressed on only 1 of the 2 X chromosomes, we used mice heterozygous for TetOP-ΔRepA-Xist (denoted as tgXist onwards) in our studies. After only 2 weeks of doxycycline administration in heterozygous tgXist male mice, expression of tgXist was detectable through Xist qRT-PCR in multiple tissues and as single punctate foci in the nucleus reminiscent of Barr body, as evidenced by RNA fluorescent in situ hybridization (FISH) (Figure 1B,C). We note that in the absence of induction, tgXist has low level of expression that is detectable by qRT-PCR and by FISH; upon induction, tgXist level increases ~100-fold to approximate the level of endogenous Xist in female tissues (Figure 1B–D). tgXist did not reduce chromatin accessibility or RNA expression locally at the locus of transgene insertion or across chromosome 11 (Figure S2A–D), consistent with the notion that ΔRepA-Xist is functionally null for gene silencing¹⁹. Backcrossing of the inducible tgXist transgene into common mouse strains enables the study of Xist in male animals in multiple autoimmune disease models.

Figure 1: tgXist-mice inducibly express ΔrepA-Xist under doxycycline exposure.

(A) Transgene construct of truncated ΔrepA Xist under tet-o-promoter in mice. (B) qRT-PCR and (C) FISH of Xist expression in tgXist-mice after 2 weeks of doxycycline administered in drinking water. (D) qRT-PCR comparison of basal tgXist levels in tgXist male mice without doxycycline exposure and WT male mice. For B and D, number of tgXist Female = 3, tgXist Male (no dox) = 3, tgXist Male + dox = 4, Wt Male = 5. (E) Schematic of tissue collection in pristane-induced SLE model in the C57/BL6J strain. (F) Cohorts in the induced SLE model. See also Figure S2

Effect of genetic background on tgXist and autoimmunity

The pristane model of SLE has a well-documented female bias in disease penetrance and severity in the permissive SJL/J genetic background. In contrast, C57BL/6J mice^20,21, the most widely used genetic background, are autoimmune resistant, and male mice are not expected to develop pristane-induced SLE²⁰. The level at which pristane-induced phenotypes are halted is unclear. We first tested the effect of tgXist expressing on surrogate markers of autoimmunity to understand whether the tgXist transgene changed the requirement for genetic background in this mouse model of autoimmunity. Male and female C57BL/6J mice heterozygous for the ΔRepA-Xist transgene were injected with pristane to chemically induce systemic lupus erythematosus (SLE)^20,22–24 and evaluated for disease (Figure 1E). The pristane-injected transgenic cohorts consisted of a female control untreated for doxycycline (female), male control untreated for doxycycline (tg male), and the test group of male mice continuously treated with doxycycline to induce tgXist expression (male+dox) (Figure 1F). Pristane-treated and untreated wild-type males of the same background were included as additional wild-type male controls. The rationale for each treatment cohort and comparison control is detailed in Table S2.

We found that pristane treated tgXist male mice in C57/BL6 background do not exhibit disease even after 1 year, which is three times the duration for severe disease induction in SJL/J background (16 weeks). While tgXist mice developed anti-nuclear antibodies (ANAb), none of the specific autoantibodies examined were significantly different between tgXist male +pristane vs. WT male+ pristane mice of the same age (Figure S3A–B). A subset of the tgXist male + pristane animals developed autoantibodies to smRNP, Smith proteins, U1-A, and U1–68, but the levels were heterogeneous. These results indicate that in the nonpermissive C57/BL6 background, these animals have a low propensity to develop autoantibody or organ disease after an inflammatory challenge, and tgXist did not bypass this genetic barrier.

Transgenic Xist drives males to female-like changes in T cell profiles

While pristane treatment in C57BL/6J mice did not manifest disease, splenic CD4⁺ T cells (an important cell type for balancing and driving autoimmunity^25–27) from tgXist-expressing male mice showed molecular changes more similar to females than to control (tgXist-non-induced/wild-type) males at several levels. Potential changes in the transcriptional regulation and gene expression of CD4+ splenic T cells were assessed using ATAC-seq and RNA-seq of multiple animals per cohort at 52 weeks post-treatment (Figure 2A, Figure S3C–E). Global comparison of ATAC-seq differential peaks showed several differences between non-tgXist/Xist-expressing animals compared to tgXist-expressing males and wild-type Xist-expressing females, but very few differential peaks between females compared to males expressing tgXist (Figure 2B). PCA comparison of female mice, male mice, and male tgXist-expressing mice (male+dox) showed a separation of mice most highly correlated with sex (Figure 2C). Three of six tgXist-expressing male mice showed female-like skewing (circled in Figure 2C), in keeping with the expected female penetrance percentage.

Figure 2: Bulk ATAC- and RNA-sequencing of splenic CD4+ T-cells from C57BL/6J mice reveal closer correlation between tgXist- and Xist-expressing mice in the pristane-induced SLE model.

(A) Table of treatment cohorts: colors correspond to each treatment cohort and numbers indicate the number of ATAC- and RNA-seq samples. *Due to library quality, 1 of the ATAC-seq libraries was excluded. (B) MA plots comparing differential regions of genomic accessibility in transgenic cohorts. (C) Poly Component Analysis (PCA) plots of ATAC-seq libraries from all mice in the study. Heatmap of (D) ATAC-seq Z-scores and (F) RNA-seq differential gene expression. Differential peak and gene lists generated from comparisons of the Tg Female+ Pristane, WT Male (mock treatment), and Tg Male + Pristane (no dox) groups to WT Male + Pristane cohort. (E) Top 15 differential reactomes associated with cluster 2 genomic regions. (G) CIBERSORT prediction of T-cell subset composition from CD4+ T-cells RNA-seq gene expression libraries. See also Figure S3 and Table S2

ATAC-seq revealed distinct clusters of accessibility corresponding to tgXist/Xist-expressing and non-expressing mice (Figure 2D). tgXist-induced and pristane-injected males displayed similar chromatin accessibility to pristane-injected female positive control mice, with higher accessibility in cluster 2, and were distinct from male negative control groups (tgXist non-induced and wild-type) that displayed higher accessibility in cluster 1 (Figure 2D). Interestingly, the top Reactome categories in Cluster 2, associated with tgXist-expressing males and Xist-expressing females, displayed high TLR pathway signatures (Figure 2E) not present in Cluster 1 (Figure S3D). Tlr9, encoding a pathogen sensor in the innate immune pathway that is highly active in SLE, is significantly more accessible in females and tgXist-induced males (Figure S3E). RNA-seq profiles of gene expression displayed a similar trend and clustering grouped by tgXist/Xist expression with some variability within cohorts (Figure 2F, Figure S3C), suggesting that ATAC-seq provides a more consistent profile of cells in transition. Comparison of dysregulated genes from ATAC- and RNA-seq revealed an overlap of 364 genes (Figure 2G). CIBERSORT²⁸ deconvolution of gene expression signatures also identified a clear segregation of Xist-expressing and non-expressing mice. In particular, tgXist or Xist-expressing males and wild type females, respectively, contained more CD4 memory T-cells while control males expressed greater proportions of naïve T-cells (Figure 2G).

Xist expression in males promotes multi-organ autoimmune pathology

We next tested the effect of tgXist expression in the autoimmune-prone SJL/J mouse background, a widely used strain in multiple autoimmune disease models^22,29,30. Pristane-induced SLE in SJLJ/J mice exhibit many characteristics of human SLE, such as autoantibody development, TLR7 upregulation, multi-organ involvement, and demonstrate a strong female bias^10,22. SJL/J female mice display earlier mortality, more severe nephritis, higher levels of autoantibodies, and are 3.4x more likely to die than male mice ²². This disease model is also more reflective of most human patients than spontaneous SLE models that are restricted to specific genetic mutations^31–34.

We administered a one-time 0.5 mL intraperitoneal injection of pristane to 8–10 week old mice to induce SLE in tgXist and wild-type (WT) mice and continuous ingestion of doxycycline was supplied in drinking water to activate tgXist in selected mice (Figure 3A). The experimental cohorts include the wild-type female pristane-treated positive control, negative controls of mock-injected tgXist- and WT males, the WT male treatment control, and the test group of tgXist-expressing pristane-treated males (Figure 3B, Table S2). Since previous literature determined that WT SJL/J females display mortality as early as 16 weeks post-pristane injection²²,²², we selected the terminal timepoint of 16 weeks post-injection to avoid premature loss of mice.

Figure 3: Increased phenotypic changes in pathophysiology and autoantigen levels in tgXist- and Xist-expressing mice in the SJL/J strain in the pristane-induced SLE model.

(A) Schematic of pristane-induced SLE and strategy for histopathology, serum and sequencing analysis in the SJL/J strain. (B) Treatment cohorts in the SJL/J strain. (C) Table of the mean severity of damage across multiple tissue sites and mice. (D) Representative H&E images of (i) glomeruli (kidney) and (ii) liver sections from each treatment cohort. Arrows demarcate mesangial thickening. (E) Graph of total sum of pathophysiology damage scores. Significance calculated using the Fisher’s Exact Test and FDR<0.05. (F) Median Fluorescence Intensity (MFI) of serum reactivity to autoantigens using a bead-based lupus antigen array after 16 weeks of treatment. Significance calculated using the unpaired Wilcoxon Rank Sum Test and values displayed at FDR < 0.05. Number of mouse serum samples, listed from left to right: wild-type Female + pristane =18, wild-type Male negative control=13, tgXist Male+dox (tgXist control) =11, wild-type Male treatment =12, tgXist Male treatment=10. P-values are indicated; NS indicates not significant or not meeting FDR<0.05. See also Figure S4 and Tables S2 and S7

As SLE is a systemic disease, we used H&E staining to assess the pathology of multiple affected organs at the terminal collection time on a disease scale of 0–5 for each individual organ (Figure 3C,D). In all pristane-treated cohorts, pristane injection caused lipogranulomas in adipose tissues, lymph node hyperplasia and medullary plasmocytosis, as well as varying degrees of extramedullary hematopoiesis and lymphoid expansion in the spleen. However, pristane injection coupled with tgXist expression in males or Xist in WT female resulted in greater incidence and severity of glomerulonephritis (kidney), hepatic lipogranulomas (liver), and pulmonary hemorrhage and lymphohistiocytic alveolitis (lung), which is reflective of disease damage in the kidney, liver, and lungs observed in severe SLE patients³⁵ (Figure 3C). Manifestation of mild signs of early SLE in some pristane-injected male controls was not unexpected because WT SJL/J males are expected to develop milder disease and display later mortality at 24 weeks post-injection²².

To assess whether Xist expression induced female-level autoimmune disease, we summed the disease scores across 6 organs in individual animals, and chose the total pathology score of 10 as the cutoff for severe disease, which clearly distinguished wild type female mice treated with pristane (positive control) vs. wild type male mice treated with pristane (treatment control). Every female mouse had a score of 10 or above; none of the male wild type mice treated with pristane met this cutoff (p =2.01E-6, Fisher’s exact test). In tgXist males treated with pristane, the total pathology increased in a bimodal fashion compared to WT males: 5 of 8 mice achieved female level pathology (score ≥ 10) while 3 animals had much less disease severity (p=0.009, FDR<0.05, Figure 3E). Concordantly, the greatest statistical difference was observed in the comparison between the female positive controls and the pristane-treated WT males (p=2.01E-6), and the significance lessened in the comparison between females and tgXist males treated with pristane (p=0.042, FDR<0.05) (Figure 3E). Thus, a majority of tgXist male mice experienced female level severe pathology in the pristane induced SLE model.

Sera collected at treatment start (day 0, baseline) and terminally (16 weeks post-treatment) were assessed for reactivity to known SLE and SSc antigens using the Luminex bead-based antigen array³⁶. We found 4 known autoantibodies (RIBO P0, RIBO P2, CENPA, CENPB) that are elevated in female WT mice vs. male WT mice treated with pristane. The mean level is increased for all four autoantibodies in tgXist male mice + pristane vs. WT male mice + pristane, but with different degrees of variance. Anti-CENPB is significantly elevated in tgXist male mice treated with pristane vs. WT male mice treated with pristane (p=0.02, FDR<0.05), and the former is no longer significantly different from WT female mice. The other three antibodies showed an intermediate picture. They are elevated in only a subset of tgXist male + pristane animals: they are no longer significantly different from WT female animals with disease, but they are not statistically significant compared to WT male control (Figure 3F). The heterogeneity of known autoantibodies may track with disease severity; all five of animals in the tgXist male + pristane cohort with severe disease had the highest level of anti-RIBO P2 (Figure 3F, Figure S4).

Single-cell ATAC analysis reveals distinct cell type clustering and consistent reformatting of the chromatin landscape in pristane-treated SJL/J animals

To gain insight to the genes and cellular processes underpinning the heightened autoimmunity in tgXist and Xist-expressing mice, we created single cell ATAC+ gene expression libraries of CD45+ (pan-hematopoietic) sorted splenic cells from pristane-injected mice in the SJL/J strain. A key advantage of the single cell multiomics approach is the ability to interrogate all the hematopoietic lineages within the spleen in a single assay. While our bulk ATAC- and RNA-sequencing in the C57BL/6J strain was restricted to CD4+ T-cells, the single cell libraries encompass the whole CD45+ splenic population.

Using the total lowest observed pathology score in female mice (10) as the cutoff (Figure 3B,D), we divided the pristane-induced tgXist-expressing males into disease high (total pathology score ≧ 10) and low (<10) disease groups. Sequencing libraries were made from representative mice selected from the tgXist male high disease (tgM high, n=4), tgXist male low disease (tgM low, n=4), and two pristane-injected control groups: wild-type females (Wt F, n=4) and wild-type males (Wt M, n=3). Due to high mitochondrial/ribosomal RNA content, 2 of the wild-type females were excluded from the gene expression analysis. To maintain consistency, we included only cells with both ATAC and gene expression reads.

The single cell ATAC UMAP formed 19 clusters (Figure S5A) and were assigned cell type identities using imputation of key marker genes (Figure 4A, Figure S5C). The four pristine-treated mouse group were generally evenly distributed across all clusters, although tgM high mice contained a higher fraction of cells within the CD4 Tcell cluster (Figure 4B). TgM high cells also formed a visually separate CD4_Tcell and Bcell_1 group on the UMAP (Figure 4A, 4C), but comparison of distinguishing markers did not show prominent differences from the adjacent CD4_Tcell and Bcell_1 populations. As expected, the tgXist male mice and female mice all showed peaks in the first Xist exon region (the Xist transcription start site is absent in the Xist transgene) that are lacking in the wild-type control males (Figure 4D).

Figure 4: Single cell ATAC comparisons of splenic CD45+ hematopoietic cells from pristane-treated mice in the SJL/J strain.

(A) UMAP of cell type cluster identities and (B) corresponding cell type composition within each pristane-treated mouse group. (C) UMAP clustering and (D) Xist peak tracks from pristane-treated mouse groups (features FDR ≦ 0.1 and Log2FC ≧ 0.5). Pristane-treated mouse groups shown: tgXist male disease high (n=4), tgXist male disease low (n=4), wild-type male (n=3), and wild-type female (n=2). See also Figure S5

Interestingly, Wt F and tgM high cells appear to overlap in the main Bcell_2 cluster (Figure 4A, 4C), which colocalizes Cd19 with the atypical B-cell marker Zeb2 (Figure S5C). However, pairwise comparisons between the mouse groups did not identify significant differences in the chromatin landscape within the main lymphocyte clusters even when comparing the most distinct positive disease control (Wt F) to the males with the lowest disease scores (Wt M and tgM low) (S Figure S5D). Aside from a higher CD4T cell fraction in the tgM high mice and atypical B cell activation markers, there were few noticeable distinguishing features in the single cell ATAC comparisons between the pristane-treated mouse groups, suggestive that all of the pristine-treated mice in the autoimmune-prone SJL/J model may have already undergone epigenetic remodeling preceding physiological disease onset at the time of investigation (16 weeks post-injection).

Single-cell gene expression analysis reveals elevation of atypical B-cells and suppression of T-cell modulators in diseased tgXist animals

We next interrogated the pristane-treated SJL/J groups by single cell gene expression, which may reflect cell states with greater immediacy. The single cell gene expression UMAP formed 12 distinct clusters and were segregated clearly into B-cell, T-cell, NK cell, myeloid and erythroid lineages based on key transcription factors and markers (Figure 5A, Figure S6A). The cellular type identities assigned to cells in the gene expression clusters closely matched the ATAC cell type identities (Figure S5B), further validating the consistency of the assigned cellular identities in both the ATAC and gene expression datasets.

Figure 5: Single cell gene expression comparisons of splenic CD45+ hematopoietic cells from pristane-treated mice in the SJL/J strain.

(A) Cell type cluster identities and (B) pristine-treated mouse groups displayed on the single cell gene expression clustering UMAP. (C) Metrics of differentially expressed B-cell cluster genes of tgXist Male high disease and wild-type female compared to wild-type males. (D) Representative gene expression plots of atypical B cell genes from the Bcell_3 cluster and (E) Metrics of differentially expressed T cell cluster genes of tgXist Male high disease and wild-type female compared to wild-type males. (F) T-cell cluster highlighting significant differentially expressed genes. Pristane-treated mouse groups shown: tgXist male disease high (n=4), tgXist male disease low (n=4), wild-type male (n=3), and wild-type female (n=2). Significance was calculated using the Wilcoxon Rank Sum test. See also Figures S6 and S7

Within the B cell cluster, the tgXist male high disease group overlay with the female wild-type mice, distinct from the wild type male and the relatively unaffected tgXist male low disease animals (Figure 5B, Figure S6B, Figure S7A–D). Since B cells produce the majority of autoantibodies and proinflammatory cytokines that characterize autoimmune pathogenesis^37,38, the high correlation of B-cells, specifically, of the tgXist- diseased males with those from positive disease control females provides another layer of support for the hypothesis that Xist complexes mediate an environment of higher autoimmunity. The tgXist low disease and wild-type males grouped closely together and most of the differences between the tgXist high disease and wild-type female mice lay in the sex chromosomes or ribosomal protein genes. Inspection of the differentially expressed genes comparing the disease-affected tgXist male and female mice with the wild-type males (Figure S6C) revealed that the majority of both downregulated and upregulated genes were shared between the first two groups (Figure 5C). Particularly significant genes included the upregulation of the atypical B-cell marker, Zeb2 (tgM high vs WT M, p = 3.08E-69, WT F vs WT M, p = 1.17E-35) and CD22 (tgM high vs WT M, p = 7.63E-46, WT F vs WT M p = 2.59E-36), a receptor associated with pathogenic B cells³⁹ critical for B-cell proliferation and B cell receptor signaling (Figure S6D). Concurrently, Siglec-g, encoding a receptor associated with promoting B-cell self-tolerance, deficiency of which is associated with increased B-1 cells and multiple autoimmune diseases^40–42, was significantly downregulated (tgM high vs WT M, p = 1.73E-101, WT F vs WT M, p = 1.01E-43) (Figure S6D). Also downregulated were complement receptor 2 (Cr2, tgM high vs WT M, p = 1.30E-153, WT F vs WT M p =3.70E-44) and the paralog to its alternatively spliced form, Cr1l (tgM high vs WT M, p = 4.41E-33, WT F vs WT M, p = 1.35E-33), both of which are important for suppressing autoimmunity^43,44.

Within the B cell clusters, WT F and tgM high uniquely overlapped in a region of Bcell_3 not populated by either WT M or tgM low disease groups (Figure S7A–D). Cluster Bcell_3 contained the highest correlation of gene expression signatures of atypical B-cell markers45 (Figure S7E, highlighted in orange), including upregulation of Cd19, Ms4a1 (gene encoding Cd20), Zeb2, and Fcrl5, a defining marker of atypical memory B-cells in both mice and humans46, as well as downregulation of Cr2 (Cd21), Cd27, and Cxcr5^47,48. Furthermore, Bcell_3 cluster cells from the gene expression data matched solely to the independently clustered multiome ATAC UMAP Bcell_2 cell clusters (Figure S5B), the ATAC clusters imputed to have atypical B-cell characteristics (Figure S5C, Zeb2 imputation). Closer examination of the gene expression Bcell_3 cluster showed significant elevation of Zeb2 (tgM high vs WT M, p = 6.12E-74, WT F vs WT M, p = 8.53E-9) and downregulation of Cr2 (tgM high vs WT M, p = 1.75E-89, WT F vs WT M, p =3.03E-14) and Cxcr5 (tgM high vs WT M, p = 5.65E-23, WT F vs WT M, p = 2.61E-18) in high disease compared to relatively unaffected mouse groups (representative plots, Figure 5D). CD21⁻CD27⁻ double negative are typical distinguishers of atypical B cells while loss of CXCR5 in CD21⁻ effector B cells is a hallmark of atypical B cells in SLE pathogenesis in patients^47,48.

While the UMAP within the T cell clusters was less distinctly demarcated than the B-cell cluster (Figure 5B, Figure S6E), the overall pattern was clear in the shared downregulated programs (Figure 5E). Multiple key T cell regulation and self-tolerance genes were downregulated in the tgXist male high disease and female cohorts compared to the wild-type males. Among the significantly downregulated genes were the glucocorticoid receptor Nr3c1 (tgM high vs WT M, p = 1.37E-33, WT F vs WT M, p = 1.33E-63) involved in Treg modulation of inflammation⁴⁹, Cd37 (tgM high vs WT M, p = 4.60E-12, WT F vs WT M, p = 1.81E-10) which regulates proliferation⁵⁰ and complement-mediated apoptosis of autoreactive T-cells⁵⁰, the CD4 T-cell immune regulator Cd52⁵¹ (tgM high vs WT M, p = 7.24E-13, WT F vs WT M, p = 2.22E-72), and invariant chain Cd74 involved in training antigen immunity in T cells⁵² (tgM high vs WT M, p = 8.15E-37, WT F vs WT M p = 2.27E-96) (representative plots, Figure 5F).

Our multiome sequencing identified clusters suggestive of atypical B cells overlapping closely in both ATAC and gene expression. At the single cell gene expression level, the severity of disease in tgXist-/Xist-expressing animals may be driven by increased atypical B cell activity and decreased immune modulatory programs in both B and T cells. Combined with the heightened pathophysiology scores and increased autoantigen levels, the gene expression data corroborates a role for Xist RNPs in the development of increased and/or more severe autoimmunity in the pristane-induced SLE model.

Autoimmune patients and mice display multiple autoantibodies to the XIST RNP

The imperfect association of known autoantibodies with tgXist-enhanced disease motivated us to consider whether novel autoantibodies to Xist RNP itself exist in human patients. To test whether the XIST RNP is immunogenic in humans, we obtained de-identified sera from patients with dermatomyositis (DM), scleroderma (SSc) and SLE to test for reactivity to XIST complex proteins^12,14. We used protein fragments produced by the Human Protein Atlas for 130 of XIST-associated proteins of interest (selection described in Methods) and 52 for control proteins used as clinical markers for DM, SSc and SLE. Three of the clinical protein antigens overlapped with the XIST ChIRP-MS lists (SSB, SNRPD2, SRP14). When possible, multiple fragments spanning different regions of each protein were used (Figure 6A).

Figure 6: Autoimmune disease individuals experience increased serum reactivity towards antigens of the XIST RNP.

(A) Table of the XIST complex-associated antigen array design showing the number of total unique proteins and corresponding protein fragments counts drawn from XIST ChIRP-MS, clinical disease panel, and control protein lists; and sample numbers of autoimmune patient and general population serum. (B) Volcano plot of serum reactivity of all autoimmune patients and (C) sera grouped by DM, SSc, and SLE patients compared to general population baseline of serum from blood donors. Significant differentially reactive antigens, defined as p_adj < 0.05 and MAD difference > 0, labeled in red. Significance calculated using the Student’s T-test. (D) Metrics of unique antigens from the array with significant elevated serum activity in autoimmune patients compared to the general population. (E) Serum reactivity (MAD) plots of representative antigens significantly reactive in all three autoimmune patient cohorts grouped by disease type and colored by sex (red=female, blue=male). See also Tables S1 and S5

As a general population control, we obtained serum from anonymous donations to the Stanford Blood Center. However, since the mean and median age of these donors were 58 and 61 years at the time of donation, these donors may express some autoreactivity due to advanced age. Nevertheless, autoimmune patients were significantly more reactive to 55 proteins, 16 of which were disease markers, 1 of which overlapped with XIST RNP (SSB), and the remaining 39 antigens from the XIST RNP list (Figure 6B, Table S5). Distinct reactive antigens for each autoimmune disease arose when grouped by disease (Figure 6C) and 9 antigens were shared among all 3 diseases (Figure 6D, Table S5). Of these, TRIM33, or TIF1-γ, is a clinical marker for autoimmune disease (DM)⁵³. The other 8 were XIST RNP components, several of which have recently discovered roles in autoimmunity. HMGB1 was recently identified as an autoantigen in SLE/Sjögren that may be another SS-protein in the SS-A/SS-B family^54,55, HNRNPK is an autoantigen in a subset of Raynaud’s disease^56,57 and aplastic anemia, SAFB is a novel autoantigen in connective tissues detected in interstitial lung disease, and XPO5 is in the SSB/La processing pathway⁵⁸.

In sum, the 3 disease groups were significantly reactive to 79 unique proteins in the array compared to the general population control. Of these 79 proteins, 27 were disease controls, and 53 were associated with the XIST RNP (SSB is also a disease control marker). 37 of the 53 XIST-associated proteins were part of the group of 118 high confidence published XIST RNP complex proteins¹⁴. Of these 37 high confidence XIST complex proteins, 28 have not been described in the current published literature as autoantigens associated with autoimmune disease and are potentially novel biomarkers (Table S5). These results show multiple proteins from the XIST RNPs are novel autoantigens in patients with DM, SSc, and SLE.

Finally, we turned to our mouse model to probe the drivers of autoantibodies to Xist RNP. Using our Xist RNP array, we analyzed sera from the tgXist mice of the SJL/J background to examine the effect of tgXist and pristane induced lupus. We compared sera of each animal longitudinally at 0, 4, 12, and 16 weeks after treatment across 5 mouse cohorts, allowing us to infer a causal relationship between perturbation and autoantibodies to Xist RNP. We compared the autoantibodies in tgXist mice to autoantibodies to the same proteins in human patients with SLE, which grounds the mouse results with human disease relevance (Fig. 7A). First, we observed that female WT mice treated with pristane induced dozens of autoantibodies to Xist RNP by 12 weeks that persist through 16 weeks. These Xist RNP autoantibodies significantly overlap autoantibodies to Xist RNP in human patients with SLE (p= 0.001, Fig. 7B). Second, tgXist expression in males treated with pristane induced many of the same antibodies against Xist RNP, which are at higher levels than in WT male mice treated with pristane at both 12 weeks and 16 weeks (p=6e-11 and p=1e-29 respectively, revised Fig 7B–D). Third, high levels of Xist RNP antibodies are only observed in pristane treated mice, indicating a key role for tissue damage and inflammation. Moreover, wild type female mice had higher Xist RNP antibodies than tgXist male mice (both treated with pristane) at 12 weeks after treatment (p<7e-37); tgXist male mice only reached female level of Xist RNP antibodies at 16 weeks (revised Fig. 7C,D). These results show that tgXist expression in males can promote autoantibodies to Xist RNP in the context of tissue damage, but this occurs with slower kinetics than wild type females.

Figure 7. Model of XIST RNP in autoimmune progression.

(A) Schematic using the Xist antigen array to identify autoantibodies shared in SLE patients and tgXist/Xist-expressing SJL/J mice with/without pristane-induced SLE. (B) Antibody reactivity (MFI normalized to bare bead baseline) against Xist RNP members is shown for the five indicated mouse cohorts across time points. Xist RNP members that are also autoantigens in human SLE patients are indicated on left bar (SLE_antigens). Each row is a Xist RNP member; each column is a mouse serum sample. (C) Quantification of Xist RNP autoantibody reactivity at 12 weeks after treatment, and (D) at 16 weeks after treatment. Significant p-values are indicated (Wilcoxon rank sum test, all with FDR<0.05); non-significant differences are indicated with “NS”. Heat map and box plot visualization plots capped at 1000 MFI to account for outliers. Number of mouse serum samples: wild-type Female + pristane =18, wild-type Male mock treatment =13, tgXist Male PBS+Dox =11, wild-type Male Dox+Pristane =12, tgXist Male Dox+Pristane=10. (E) Autoreactivity to Xist RNPs first causes changes in the chromatin landscape impacting genomic accessibility and changes in lymphocyte gene expression programs prefacing the development of autoantibodies and cascade to prolonged autoreactive activity damaging organs in the final stages of autoimmunity. See also Figure S1 and Table S5

DISCUSSION

Xist lncRNA as a polymeric antigen scaffold in female-biased autoimmunity

Our study nominates Xist ribonucleoprotein complexes as antigenic triggers underlying the greater prevalence of autoimmune diseases in females. Although it is a well-documented fact that females are more prone to autoimmune diseases than males, previous studies primarily examined differences in gene dosage and hormonal background. While prior studies of Xist address altered X-inactivation and the subsequent impact of XCI escape of X-linked genes^14,59,60, this study investigated the immunogenicity of Xist RNP complex itself. We have shown that expression of Xist RNPs in male mice is sufficient to increase disease severity and change the expression and epigenomic profiles of both the B-cell and T-cell effectors of SLE pathogenesis.

Physicians and scientists have long noted that many autoantibodies target large nucleic acid protein complexes, such as chromatin or RNP, in human autoimmune diseases. This feature was exploited by molecular biologists to use patient sera to identify components of the centromere (recognized by autoantibodies in CREST syndrome) or spliceosome (SLE, DM). Immunologists have explained this phenomenon with the idea that large nucleic acid-protein complexes are polymeric, and if exposed in the extracellular space, can cluster and activate immunoreceptors. We propose that the XIST RNP is one such dominant antigenic array that is unique to females. Every cell in a woman’s body has XIST, which is a long polymer (19 kb) and coats the entire inactive X chromosome in the condensed Barr body (an even larger polymer). When a female cell dies due to tissue injury, XIST RNPs will invariably be exposed to the immune system. Our data further suggests a model where XIST contributes to several steps in the progression to autoimmune disease (Figure 7C). In a genetically autoimmune resistant background, low level of XIST, even in the presence of tissue injury, leads to only changes in T cell subsets and chromatin states but no frank organ pathology. These epigenetic changes in accessibility are then subsequently reflected in the gene expression programs upregulating autoreactivity and downregulating immune modulation. Finally, in the context of a permissive genetic background and repeated tissue injury, the presence of XIST RNP exacerbates full blown end organ pathology and activation of multiple immune cell types. Longitudinal studies of sera reactivity and autoimmune disease in humans are consistent with this model ⁶¹.

Opportunities for disease diagnosis and therapy

There are more than 100 known autoimmune diseases that in aggregate afflict ~50 million Americans and comprise one of the top ten leading causes of death for women under the age of 65⁶². Worryingly, cases are increasing yearly on a global scale and recent serologic studies revealed a steep rise of increasing ANA reactivity^63,64. Understanding the risk factors and drivers of autoimmunity has become even more critical in the race to develop effective therapies and sensitive diagnostics specific to each autoimmune disease. However, the high heterogeneity within autoimmune diseases and overlapping traits across diseases have limited our ability to tailor effective therapies and sensitive diagnostics specific to each autoimmune disease⁶⁵. Our discovery of seropositivity towards multiple XIST-associating proteins in autoimmune patients introduces a novel antigen set with clinical potential for enhancing disease detection and monitoring, as autoantibodies are often detected prior to or early in disease onset^61,66. In addition, studies in SSc have also demonstrated the effectiveness of autoantigen analysis in patient stratification and identifying pathogenic pathways^67,68. Profiling the XIST RNP in primary cells, both healthy and diseased, may be useful in advancing our understanding of the aberrant autoreactivity towards proteins within the complex and identify more potential autoantigens.

Currently, there are few targeted therapies for autoimmune diseases available. The most common therapies involve B-cell depletion but are not always effective. There remains a need for more specific pathogenic leukocyte targets. We identified atypical B cells as a population of immune cells that accumulate as a consequence of Xist RNP expression. Atypical B cells (also known as age-associated B cells) are a unique population of B cells that expand with increased TLR7 signaling^69,70 and in female-biased autoimmunity⁷¹. Notably, atypical B cells accumulate in aged female mice but not in age-matched male mice⁶⁹, and atypical B cells are enriched in human or mouse B cells that escape XCI and re-express TLR7^14,72. Thus, atypical B cells appears as the immunological nexus of two potential consequences of mammalian dosage compensation--autoreactivity to Xist RNP and escape from XCI—and suggest that these consequences may synergize to promote female biased immunity. Future studies should address whether and how atypical B cells or other cell types evoked by Xist RNP contribute to autoimmunity.

Limitations of the study

This study employs a truncated Xist missing the A-repeat to model Xist RNP action in male mice. The detection of autoantibodies to Xist RNP in female patients and wild type female mice indicate that tgXist models important aspects of full length Xist in females. Although we have confirmed that immune cell subsets, autoantibodies, and disease penetrance of autoimmunity of tgXist males are at an intermediate level between WT male and WT female, arguing against a neomorphic effect of Xist-ΔA-repeat, it is formally possible that Xist-ΔA-repeat differentially impacts other features of autoimmunity. Fragmentation of cell-free Xist RNP is the inevitable consequence as the complex is released from dying cells and eventually cleared, and this interpretation is fully compatible with our model for female biased immunity. Our transgenic model expressed Xist ubiquitously in male animals, and the role of individual tissues or cell types most responsible the observed phenotypes and the required time window of Xist RNP exposure should be dissected in future studies. Additionally, a known limitation of Tet-regulated expression cassettes is that they can be leaky or silenced over time in vivo^73–76. Thus, there is likely variability or decline of tgXist expression in our transgenic model during the experiment; and the level of Xist required to confer female-level autoimmune risk is unclear. Similarly, this work has not addressed whether fluctuations in Xist level in female individuals may impact autoimmunity. Finally, this study employed a modest number of animals and patient samples, and several results showed large variation between test subjects. Future studies with larger numbers and a detailed focus on exactly which XIST related antigens contribute to female biased immunity will be valuable.

STAR★Methods

Resource Availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Howard Y. Chang (howchang@stanford.edu).

Materials availability

All requests should be directed to the lead contact, except ΔRepA-Xist transgenic mice are available upon request with a completed Materials Transfer Agreement with Anton Wutz (awutz@ethz.ch).

Data and code availability

• Bulk ATAC- and RNA-seq and single cell multiomic data have been deposited in GEO. Accession numbers can be found in the Key Resources Table. The autoantigen array raw data files are included in Supplemental Information. Additional can be found in the Key Resources Table.

• Analysis details are provided in STAR Methods. No original code was generated in this paper.

• Additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
7-AAD	BD Biosciences	Cat # 559925, RRID:AB_2869266
FITC anti-mouse CD4 (clone GK1.5)	BioLegend	Cat # 100406, RRID:AB_312690
TruStain FcX PLUS (anti-mouse CD16/32, clone S17011E)	BioLegend	Cat # 156604, RRID:AB_2783137
APC anti-mouse CD45 (clone 30-F11)	BioLegend	Cat # 103112, RRID:AB_312977
Mouse Xist Stellaris® FISH Probes with Quasar® 570 Dye	LGC Biosearch Technologies	SMF-3011-1, RRID: AB_3076254
R-Phycoerythrin labelled Goat anti-Human IgG Fc	Invitrogen	12-4998-82; AB_465926
Bacterial and virus strains
Biological samples
Human whole blood (healthy control serum)	Stanford Blood Center	https://stanfordbloodcenter.org/researchlabs/research-productsand-services/bloodproducts/
SLE patient sera	This paper; Johns Hopkins	NA
DM patient sera	This paper; Stanford University	NA
SSc patient sera	This paper; Johns Hopkins, Stanford University	NA
Chemicals, peptides, and recombinant proteins
Pristane	Sigma-Aldrich	P9622-10X1ML
PBS 1x	Thermo Fisher Scientific	Cat # 10010049
Doxycyline hyclate	Sigma-Aldrich	D9891-5G
DNAse I	Thermo Fisher Scientific	Cat# 18068015
O.C.T. Compound	Tissue-Tek	Cat# 4583
Fisherbrand™ Superfrost™ Plus Microscope Slides	Fisher Scientific	Cat# 12-550-15
Paraformaldehyde	Fisher Scientific	Cat# 50-980-487
TritonX-100	Acros Organics	AC327371000
Ethanol	Gold Shield	Cat# 412804
Stellaris® RNA FISH Wash Buffer A	LGC Biosearch Technologies	SMF-WA1-60
Stellaris® RNA FISH Hybridization Buffer	LGC Biosearch Technologies	SMF-HB1-10
VECTASHIELD Mounting Medium with DAPI	Vector Laboratories	H-1200
VWR® Micro Cover Glasses	VWR	48366-227
HypoThermosol FRS	BioLifeSolutions	Cat# 101102
CellTrics® Disposable Cell Strainers	Sysmex	04-004-2327
eBioscience™ 1X RBC Lysis Buffer	Thermo Fisher Scientific	Cat# 00-4333-57
BAMBANKER	Wako	Cat# 203-14681
Fetal Bovine Serum	Thermo Fisher Scientific	A3160502
Protector RNase Inhibitor	Sigma-Aldrich	Cat# 3335399001
Critical commercial assays
EasySep Mouse CD4+ T cell Isolation Kit	STEMCELL Technologies	Cat# 19852
Mouse Anti-Nuclear Antibody Kit	MyBioSource.com	MBS731183
RNeasy Plus Mini Kit	Qiagen	Cat# 74136
High-Capacity cDNA Reverse Transcription Kit	Applied Biosystems	Cat# 4368814
TruSeq® Stranded mRNA Library Prep	Illumina	Cat# 20020594
Chromium Single Cell Multiome ATAC + Gene Expression Reagent Bundle	10x Genomics	PN-1000283
Chromium Next GEM Chip J Single Cell Kit	10x Genomics	PN-1000234
Single Index Kit N Set A, 96 rxns	10x Genomics	PN-1000212
Dual Index Kit TT Set A, 96 rxns	10x Genomics	PN-1000215
Deposited data
Raw and processed sequencing data (single cell)	This paper	GEO database: GSE249830
Raw and processed bulk ATAC-seq data	This paper	GEO database: GSE249830
Raw and processed bulk RNA-seq data	This paper	GEO database: GSE249830
Raw and processed SLE antigen array data	This paper	Table S3
Raw and processed XIST antigen array data (patients)	This paper	Table S4
Raw and processed XIST antigen array data (mouse)	This paper	Table S6
Experimental models: Cell lines
Experimental models: Organisms/strains
SJL/J mouse	Jackson Laboratories	Strain# 000686; RRID:IMSR_JAX:000686
C57BL6/J mouse	Jackson Laboratories	Strain# 000664; RRID:IMSR_JAX:000664
ΔRepA-Xist transgenic mouse	This paper	Anton Wutz
Oligonucleotides
Mouse Xist Forward (qRT-PCR) GACAACAATGGGAGCTGGTT	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Mouse Xist Reverse (qRT-PCR) GCAACCCCAGCAATAGTCAT	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Mouse GAPDH Forward (qRT-PCR) TGTGCAGTGCCAGCCTCGTC	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Mouse GAPDH Reverse (qRT-PCR) TGCCACTGCAAATGGCAGCC	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Neomycin Forward (genotyping) AGGATCTCCTGTCATCTCACCTTGCTCCTG	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Neomycin Reverse (genotyping) AAGAACTCGTCAAGAAGGCGATAGAAGGCG	Elim Biopharmaceuti cals, Inc.	Custom Oligos
CMV (genotyping, Forward) GCTGGTTTAGTGAACCGTCAG	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Mouse Xist (genotyping, Reverse) ACAAAGATTGGGCTGTCGAG	Elim Biopharmaceuti cals, Inc.	Custom Oligos
Recombinant DNA
Software and algorithms
Cellranger	10x Genomics	https://support.10xgenomics.com/single-cellgeneexpression/software/overview/welcome
Prism	Graphpad	https://www.graphpad.com/scientific-software/prism/software/prism/
CIBERSORT	Newman et al.28	https://cibersortx.stanford.edu/cshome.php
STAR	Dobin et al.86	https://github.com/alexdobin/STAR
RSEM	Li et al.87	https://github.com/deweylab/RSEM
Bowtie2	Langmead and Salzberg88	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Samtools	Li et al.89	http://www.htslib.org/
MACS2	Zhang et al.90	https://github.com/macs3-project/MACS
DESeq2	Love et al.92	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
g:profiler	Raudvere et al.93	https://biit.cs.ut.ee/gprofiler/gost
Seurat	Hao et al.94	https://satijalab.org/seurat/
ArchR	Granja JM, Corces MR et al.95	https://www.archrproject.com/

Experimental Models and Study Participant Details

Mouse Strains

All mouse work was conducted under Stanford’s approved animal protocol APLAC-14046. Wild-type C57BL/6J (000664) and SJL/J (000686) mice were purchased from Jackson Laboratories. Inducible ΔRepA-Xist transgenic mice: A transgene of tetO linked to Xist cDNA deleted for the A repeat (SacII-XhoI is deleted, appx. 500nt) was targeted to the Col1a1 locus on chromosome 11 in A9 129/Bl6 hybrid ES cells. To insert the ΔRepA-Xist transgene, a homing cassette was first inserted into the 3’-region of Col1a1 using homologous recombination. This homing cassette consisted of a hygromycin (pPGK-hygro-PA) resistance marker, a loxP recombinase site, and a truncated neomycin resistance gene (3’neo-PA). Subsequently, the pPGK-loxP-ΔRepA-Xist cDNA was inserted by Cre mediated recombination followed by G418 selection^18,77. Transgenic mice were generated by injection of the modified ES cells into Bl6 8-cell embryos⁷⁸. Resulting transgenic mice were crossed to a mouse line carrying the R26/N-nlsrtTA doxycycline regulated transactivator^78,79. Mice carrying the tet-O-ΔRepA-Xist and rtTA constructs were backcrossed into the C57BL/6J or SJL/J mice for multiple generations. Tail tips were collected from 5–7 female mice from each generation for speed congenics selection using Charles Rivers’ MAX-BAX 384 SNP panel. The two females with the highest match in the desired background from each generation were selected as breeders. The tgXist-C57BL6/J mice used in this study are > 80% in the C57BL/6J background. The tgXist-SJL/J mice are > 99.99% in the SJL/J strain background. Mice were genotyped for both the Neomycin resistant cassette and from the CMV promoter into the truncated Xist transgene. Studies in the SJL/J background used both mice heterozygous for tgXist and wild-type littermates while studies involving SJL/J mice while wild-type control C57BL/6J mice were obtained from Jackson Laboratories. Genotyping primers are listed in the Key Resources Table.

Clinical Cohorts

All human patient samples were de-identified in this study and obtained under the respective institutions’ IRB-approved protocols.

The biological sex of human serum samples used in this study is indicated in the text and raw data tables in Table S4 in the “SAMPLES” description sheet. All patients were adults. Information regarding age, ancestry, race, ethnicity, and socioeconomic status were not provided in the de-identified clinical patients to the Lead Author in this study and so were not included in the analysis. Due to the small number “n” of male patients, sex was not a separately analyzed variable due to lack of statistical power. Analysis was, instead, separated into disease groups.

General Population

Donated whole blood from a total of 9 male and 8 female donors were obtained from the Stanford Blood Bank following IRB-approved protocols (IRB #13942) with confirmed assent for use in research. Serum was collected as described above in the mouse serum section and frozen in −80°C for long-term storage. During data analysis, 1 of the 9 male samples was filtered out for high IgG background.

Dermatomyositis (Stanford)

Serum from 1 male and 3 female dermatomyositis patients were used in this study. All samples were collected from patients seen at the Stanford outpatient clinics under an IRB-approved protocol (IRB #12047), and all patients provided informed consent to participate. The dermatomyositis cohort has been described previously⁸⁰. All patients met probable or definite DM by 2017 ACR/EULAR IIM Classification Criteria⁸¹. All sera used in the study were also known to contain antibodies against TIF1-γ which was assayed as previously described⁸².

Systemic Sclerosis (Scleroderma, Stanford)

Serum from 24 patients (2 male and 22 female) with systemic sclerosis (SSc) were included in this study from the Stanford cohort. All samples were collected from patients seen at the Stanford outpatient clinics under an IRB-approved protocol (IRB #12047), and all patients provided informed consent to participate. All patients fulfilled 2013 ACR/EULAR classification criteria for SSc. Twelve patients had diffuse and 12 had limited cutaneous SSc. Seven patients had the Scl-70 antibody, 5 had the anti-centromere antibody, 2 patients had RNA polymerase III, 2 had nucleolar ANA, 2 had PM/Scl, and 1 had U1RNP, with the remainder having positive ANA but no known SSc-specific autoantibody.

Systemic Sclerosis (Scleroderma, Johns Hopkins)

Samples were previously collected and stored at −80°C as part of the Hopkins Scleroderma Center cohort protocol (NA_00039566). Serum from 9 males and 31 females were shipped on dry ice to Stanford. Patients were enrolled in the IRB-approved Johns Hopkins Scleroderma Center Research Registry. Scleroderma participants in the registry meet at least one of the following criteria for systemic sclerosis: 1) 2013 ACR/EULAR classification criteria for scleroderma, 2) 1980 ACR classification criteria, 3) having at least 3 of 5 features of the CREST syndrome (calcinosis, Raynaud’s phenomenon, esophageal dysmotility, sclerodactyly, telangiectasia), or 4) having definite Raynaud’s phenomenon, abnormal nailfold capillaries and a scleroderma-specific autoantibody.

Systemic Lupus Erythematosus (Johns Hopkins)

Patients provided written informed consent to participate in the Hopkins Lupus Cohort (IRB study number NA_00039294). Blood was drawn at the time of a clinical blood draw, serum collected and kept in a −80 freezer for long term storage. For this study, 40 serum samples (3 males and 37 females) were randomly selected from patients who were ever: 1) ANA positive with a titer of at least 1:320; and 2) positive to the dsDNA autoantigen. Patients enrolled in the Hopkins Lupus Cohort met either the revised American College of Rheumatology (ACR)⁸³ or Systemic Lupus International Collaborating Clinics (SLICC)⁸⁴ criteria for SLE. Samples were shipped on dry ice to Stanford and stored at –80°C.

Method Details

Induction and evaluation of pristane-induced SLE in mice

Pristane-induction of SLE

Wild-type and tg-Xist mice were injected with a one-time injection of 0.5 mL of pristane (Sigma-Aldrich, P9622–10X1ML) at 8–10 weeks old (SJL/J strain) or 12–14 weeks old (C57BL/6J studies). Control animals were injected with PBS 1x (Thermo Fisher Scientific, 10010049).

Induction of transgene (tgXist)

Simultaneous to the injection date, doxycycline hyclate (Sigma-Aldrich, D9891–5G) was continuously administered at 0.2 g/mL in drinking water to select mice until the terminal timepoint.

For the tgXist expression validation studies, tgXist^+/− and wild-type mice in the C57BL/6J strain were administered doxycycline for 2 weeks and tissues were harvested for qRT-PCR and FISH analysis.

Xist qRT-PCR

Mechanically dissociated thymus, spleen, kidney, and liver cells harvested from tgXist^+/− and wild-type mice in the C57BL/6J were strained through a cell filter, pelleted, and frozen in RLT Buffer with 0.1% BME. RNA was extracted using the RNeasy Mini kit (Qiagen, 74106), genomic DNA was removed using amplification grade DNAse I (Thermo Fisher Scientific, 18068015). RNA concentration was quantified on a Nanodrop. cDNA was made from 1 ug of RNA/sample using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, 4368814). Each Ct value was measured using Lightcycler 480 (Roche) and each mean dCt was averaged from triplicate qRT-PCR reactions. Relative Xist RNA levels was calculated by ddCt method compared to GAPDH controls. Statistical significance was calculated using the Student’s t-test. Primer sequences are listed in the Key Resources Table.

Xist FISH

Mice organs frozen in O.C.T. compound (Tissue-Tek, 4583) were sectioned on a crysostat to 15 μm onto microscope slides (Fisher Scientific, 12–550-15). Sections were washed once with PBS 1x, fixed in 4% paraformaldehyde for 10 min at room temperature, washed twice with PBS 1x for 2–5 min each, then permeabilized on ice with ice cold 0.5% TritonX-100 in PBS for 10 min.

Slides were then rinsed once with PBS and dehydrated sequentially in 70%, 90% and 100% Ethanol (Gold Shield, 412804) for 5 minutes each. Sectioned slides were then allowed to air dry before immersion in freshly made Stellaris^® RNA FISH Wash Buffer A (SMF-WA1–60) for 2–5 minutes at room temperature. Sections were then hybridized overnight for 16–18 hours in the dark at 42C in 250 nM of Mouse Xist Stellaris^® FISH Probes with Quasar^® 570 Dye (LGC Biosearch Technologies, SMF-3011–1) in Stellaris^® RNA FISH Hybridization Buffer (SMF-HB1–10). The next day, slides were incubated twice in Wash Buffer A for 30 minutes at 37C followed by immersion in Stellaris^® RNA FISH Wash Buffer B for 2–5 minutes at 27C, mounted in VECTASHIELD Mounting Medium with DAPI (Vector Laboratories, H-1200) and sealed with cover glass (VWR, 48366–227) and nail polish clear coat. Sectioned organs were imaged on the Zeiss Observer Z.1 using the 63x oil objective, X-Cite Series120 laser, and AxioVision Rel. 4.8 software.

Mouse sera collection

Mice were sedated with isofluorane and retro-orbitally bled immediately prior to the injection and viably bled every 4 weeks until the terminal date. On the terminal date, blood was collected through cardiac puncture. Blood was allowed to clot for 2 hours at room temperature, then spun for 15 minutes at 1500xg at room temperature. Serum was immediately collected, flash-frozen on dry ice and stored at −80C.

ANAb ELISA

Serum ANAb levels were measured using the Mouse Anti-Nuclear Antibody Kit (MyBioSource.com, MBS731183) and assessed on an ELISA plate reader at an optical density of 450 nm. Titration curves and values were calculated using a Four Parameter Logistic Curve.

Tissue collection and preparation of pristane-induced SLE studies

Mice were euthanized by CO₂ asphyxiation and cardiac exsanguination at the terminal timepoints of 16- or 52-weeks post-injection. Terminal cardiac blood was aliquoted into EDTA tubes (400 ml per mouse) and Eppendorf tubes (for serum collection). Total body weights were obtained, and the following organs were weighed individually: thymus, heart, liver, spleen, testes. Gross necropsies were performed on all mice (Table S7). The caudate and papillary liver lobes, left kidney, one half of the thymus and one half of the spleen were dissected and placed on ice in HypoThermosol FRS (BioLifeSolutions, 101102) until all dissections concluded and were either mechanically dissociated using clean scalpels and syringes or frozen in O.C.T. Compound (Tissue-Tek, 4583) at −80C on the same day. The remaining tissues were immersion-fixed in 10% neutral buffered formalin for 72 hours for downstream histology analysis. Dissociated cells were passed through cell filters (Sysmex, 04-004-2327), pelleted at 300xg, incubated for 1 minute at room temperature in eBioscience^™ 1X RBC Lysis Buffer (00-4333-57) to remove red blood cells, and quenched with 10x the volume of PBS 1X. For viable stocks, dissociated cells were frozen in BAMBANKER (Wako, #203-14681) and stored in liquid nitrogen.

CD4+ isolation

CD4+ cells were isolated from freshly dissociated mouse spleen using the EasySep Mouse CD4+ T cell Isolation Kit (STEMCELL Technologies, 19852). CD4+ cells were then immediately used for bulk ATAC-seq, frozen in RLT Buffer Plus, or viably frozen in BAMBANKER. A small aliquot of cells were stained in PBS 1x with the viability marker 7-AAD (BD Biosciences, 559925) and T-cell marker CD4 FITC (clone GK1.5, BioLegend, 100406) to assess purity and viability on a FACS analyzer.

CD45+ isolation

Viable cells frozen in BAMBANKER were thawed at 37C in a bead bath, resuspended in RPMI, and spun at 300xg to pellet and remove the buffer. Cells were incubated with TruStain FcX PLUS (anti-mouse CD16/32, clone S17011E, Biolegend, 156604) to block non-specific binding of immunoglobulin and stained with the pan-hematopoietic marker CD45 APC (clone 30-F11, BioLegend, 103112) and 7-AAD for viability. Viable CD45+ cells were sorted on the BD FACSAria II sorter using a 70 uM nozzle into chilled FBS (Thermo Fisher Scientific, A3160502) with 1% Protector RNAse inhibitor (Sigma-Aldrich, 3335399001) in the Stanford Shared FACS Facility. Sorted cells were then immediately used to prepare sequencing libraries.

Histopathology

Formalin-fixed tissues were processed routinely, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin. All tissues were evaluated blindly by a board-certified veterinary pathologist (KMC). An ordinal histopathologic grading scale (score 0–4) was designed to evaluate glomerulonephritis, hepatic lipogranulomas, pulmonary lymphohistiocytic alveolitis, and splenic extramedullary hematopoiesis. A binary histopathologic grading scale (score 0 or 1) was used to evaluate for intraabdominal lipogranuloma formation and ectopic lymphoid tissue, pulmonary hemorrhage, hemosiderosis, and/or vascular thrombosis, splenic lymphoid hyperplasia and plasmacytosis, and lymph node hyperplasia and medullary plasmacytosis. A total composite score was derived for each mouse. The Fisher’s Exact Test was used to calculate significance between treatment groups. Due to facility and equipment difficulties during the COVID-19 pandemic, some mice are missing scores for kidney or spleen. Only mice with all organs correctly processed and assessed were used to calculate the composite score and included in the final analysis (shown in Figure 3). Complete scores for all SJL/J mice used in the pristane-induced lupus study can be found in the Table S7.

Statistical Analysis of histopathology scores

The Fisher’s Exact Test was used to calculate significance between treatment groups in the histopathology studies. Because the lowest positive female control total body disease score was 10, we set 10 as the cutoff total body disease score to distinguish between low disease and high disease scores.

Sequencing Library Preparation

Bulk sequencing

ATAC-seq libraries of freshly isolated CD4+ cells were prepared using the Omni-ATAC protocol⁸⁵. RNA was extracted using the RNeasy Plus Mini Kit (Qiagen, 74136). TruSeq^® Stranded mRNA Library Prep (Illumina, 20020594) was used to generate polyA-selected RNA-sequencing libraries, cleanup performed on magnets with AMPure XP beads (Beckman Coulter, A63880). The bulk ATAC-seq and RNA-seq libraries were sequenced with paired-end 75 bp read lengths on in the Stanford Functional Genomics Facility an Illumina HiSeq 4000 that was purchased with funds from NIH under award number S10OD018220.

Single cell sequencing

The Chromium Single Cell Multiome ATAC + Gene Expression (10x Genomics) was used to prepare libraries for CD45+ sorted cells with a target of 10,000 cells/sample. Libraries were sent to Novogene for Bioanalyzer trace quality control check and sequencing. Libraries were sequenced on the NovaSeq 6000 at a depth of 20,000 paired reads per cell for gene expression libraries and 25,000 paired reads per cell for ATAC libraries.

Computational Analysis of Sequencing Libraries

Bulk sequencing

The bulk RNA-seq data was aligned to mm10 using STAR⁸⁶. The gene expression read counts was generated using RSEM⁸⁷. The adaptor of paired-end ATAC-seq data was trimmed and aligned to mm10 genome using bowtie2⁸⁸. The mitochondrial reads and reads with low alignment score (<10) were removed. The aligned sam files were converted to bam files and sorted by Samtools⁸⁹. Picard was used to remove duplicate reads and MACS2⁹⁰ was used to call peaks. BEDtools⁹¹ was used to generate read counts from called peaks. Each ATAC-seq peak was annotated by its nearby genes using GREAT under the basal plus extension default setting.

The raw bulk RNA-seq and ATAC-seq read counts were then normalized and analyzed using DESeq2⁹². The differentially expressed genes and ATAC-seq peaks were identified using the negative binomial models. Benjamini hochberg procedure was used to adjust for multiple hypothesis testing. Peaks with FDR < 0.2 and absolute fold change larger than 1.5 were selected as significant. Bioinformatics tool g:Profiler⁹³ was used for pathway enrichment analysis. CIBERSORT²⁸ was used to estimate the abundance of immune cells based on normalized RNA-seq data.

Single cell multiomics sequencing

The single-cell paired RNA and ATAC-seq reads were aligned to the mm10 reference genome using cellranger-arc count (10x Genomics, version 2.0.1).

Gene expression data was filtered to include only barcodes that had nFeature > 500 and pct.mt < 12. Two female samples (F100 and F113) were excluded because, even after stringent filtering, cells from these samples had lower quality metrics than other samples. R v4.1.3 Seurat v4.1.1⁹⁴ and ggplot2 v3.3.6 were used for downstream analysis and visualization. After filtering, clustering and dimensionality reduction was performed using the top 10 principal components (dims=1:10) and a clustering resolution of 0.25. Differential gene expression analysis was performed using Seurat function “FindMarkers” with default settings.

Single cell ATAC data was processed with ArchR v1.0.1⁹⁵. ATAC-seq data was subsetted using the subsetArchRProject function to include only cells with matched gene expression cell IDs and clustered at resolution=0.7 with default settings. Gene imputation features of defining cell type markers were visualized from the GeneScoreMatrix using default settings. Cellular subsets were subsequently subsetted and analyzed with getMarkerFeatures from the PeakMatrix using default settings and maxcells=5000 and k=1000.

Calculation of Chr11 accessibility and gene expression

The aforementioned Chromium Single Cell Multiome ATAC + Gene Expression (10x Genomics) libraries were used in the analysis of Chr11 between the three treatment groups (Wt F+ Pristane, tgXist M + Pristane + Dox, and Wt M + Pristane + Dox).

Chr11 track analysis

The bigwig files for each mouse treatment group were generated using the getGroupBW function in the archR package. Boxplots were visualized in the rtracklayer package in R using bigwig files containing normalized ATAC-seq counts. Genome position-wide scATAC track visualization was obtained using the plotBrowserTrack function in the archR package. The FindMarkers function in the Seurat package was used to identify the differential expressed genes between the pristane+dox treated groups: Wt M versus and tgXist M. A hypergeometric test was then applied to examine whether the genes on chr11 are more likely to be down regulated compared with the genome wide scale.

SLE autoantigen Panel Array

Assay

Mouse serum was assessed using the SLE autoantigen panel and array as previously described³⁶. Serum from the baseline timepoint of 0 weeks (injection start) and terminal timepoints (16 weeks for SJL/J and 52 weeks for C57BL/6J strains) were run in the same assay (sample-matched). Serum was stored long-term at –80°C and prepared as previously described³⁶. Raw MFI values and mouse serum sample information can be found in Table S3.

Analysis

Raw MFI scores were first normalized by subtracting the baseline value of bare bead, with a minimum differential limit set at 0 to account for values below baseline. The difference between terminal and baseline timepoints was then calculated using the normalized values, with a minimum differential limit set at 0 as “depletion” of autoantibodies is not expected to occur over time. Values were plotted in GraphPad Prism and the Wilcoxon Rank Sum Test was used to test for statistical significance. The Benjamini-Hochberg procedure was used to calculate FDR and correct for multiple comparisons.

XIST Autoantigen Array

Autoantigen list

Available recombinant proteins were chosen from a set of XIST-associating proteins of interest from published XIST ChIRP datasets^12,14 at a stringency of log2(EXP/RNAse) > 1 and peptide EXP Sum ≥ 10. Due to the limited availability of recombinant proteins, this criteria is less stringent than that used for the bibliomics analysis in order to include a larger set of XIST-associating proteins of interest. Autoantigens clinically used to screen for DM⁵³, SLE or SSc^96,97 were included as positive controls. 16 exploratory proteins were included from unpublished XIST ChIRP-MS lists. Each protein was represented by one or more protein fragments (20–151 amino acids long) produced within the Human Protein Atlas project (https://www.proteinatlas.org/)^98,99. The full list of autoantigens and their categorization can be found in Tables S4 and S6.

Sample preparation

25 uL of serum per sample were aliquoted in a pre-determined randomized order onto 96-well plates (Thermo Fisher Scientific, AB800150) with mouse and human samples on separate plates. Thermal sealed plates were shipped on dry ice to SciLifeLab in Sweden and stored at –80C upon arrival.

Suspension Bead Array Assay (SciLifeLab)

The antigens were immobilized on color coded magnetic beads (MagPlex, Luminex Corp., Austin, TX) and the assay was run as previously described, with minor alterations¹⁰⁰. The samples were diluted directly before running the assay, prior to adding the secondary antibody the beads incubated for 10 minutes in 0.2% paraformaldehyde to fixate any bound antibodies, and the secondary antibody used was R-Phycoerythrin labelled Goat anti-Human IgG Fc (eBioscience^™; 12–4998-82, Invitrogen).

Data handling and analysis

All data processing from the suspension bead array was performed using R version 4.1.1. Antigen specific background was adjusted for by centering and scaling the 10^th percentile of the mean fluorescence intensity (MFI) value for each antigen to a common value (10^th percentile of the whole dataset). The antigen percentile adjusted MFI values were transformed per sample into the number of “median absolute deviations” (MADs) around the sample, represented as: MADs_ag,sample = (MFIadj_ag,sample − median_sample) / MAD_sample. Raw MFI values can be found in Tables S4 and S6.

For the general population and patient disease comparison analyses, the difference in reactivity was calculated as the difference between the mean of the MAD scores of the comparison groups and significance was calculated using the Student’s T-Test. To remove farlying outliers, MAD scores were first “trimmed” to quantile (0.1, 0.9). Since each protein often contained multiple protein fragments in the assay, the list was next filtered to remove duplicate counts and generate a unique list of reactive proteins. The reactive protein lists were then compared between patient disease groups (Figure 6D,E, Table S5).

Selection of enriched proteins

To find shared reactive proteins in two heterogeneous datasets (pristane-induced SLE mouse and autoimmune disease patients), the difference in MAD scores was calculated in patients as the difference between SLE patient reactivity and the general population (healthy control) and in mice as the difference between 16 weeks and 0 weeks (treatment start baseline) for each group. Protein fragments were considered reactive in patients if MADs_diff > 0 in the quantile “trimmed” set and MADs_diff > 2.5 in the raw datasets. Protein fragments were considered reactive in mice if MADs_diff > 2.5 between 16 and 0 weeks. Since each protein often contained multiple protein fragments in the assay, the list was next filtered to remove duplicate counts and generate a unique list reactive proteins. The enriched protein lists were then compared between the pristane-induced SLE mouse groups and SLE patients (Table S5).

Quantification and Statistical Analyses

All statistical analysis and quantification were conducted in R or GraphPad Prism. Three standardized comparisons are made across mouse cohorts in Figs. 3–5 and 7: (1) WT female mice treated with pristane vs. WT male mice treated with pristane (positive vs. negative control); (2) tgXist male mice +Dox + pristane vs. WT male mice + pristane (test vs. negative control); (3) WT female mice + pristane vs. tgXist male mice + Dox + pristane (positive control vs. test). The p-values are indicated on the figures; multiple hypothesis testing is controlled for using the Benjamini-Hochberg method to estimate false discovery rate (FDR). All indicated significant p-values have a FDR< 0.05 accounting for multiple hypothesis testing. Non-significant comparisons, including those rejected based on FDR, are indicated with “NS” on the figure. Figure legends state the statistical details of the experiments and assays, including exact “n” values, statistical tests, comparisons, and cutoffs specifically used in each figure.

Supplementary Material

1

Figure S1: Established autoantigenic associations of XIST RNPs, Related to Figure 7. Summary of XIST RNP complex proteins with known association as autoantigens in autoimmune disease grouped by disease (Red=positive hit). Full bibliomics information available in Table S1.

2

Figure S2: Chromosome-wide comparison of tgXist transgene insertion in Col1A1 safe harbor locus on Chr11, Related to Figures 1 and 4. (A) Chromatin accessibility at Chr11 assessed using single cell ATAC-seq data (B) ATAC-seq accessibility tracks in the Col1A1 insertion locus (C) Gene expression profile at Chr11 using single cell RNA-seq data (D) Fraction of up- and down-regulated genes on Chr11 and genome-wide comparing tgXist males and Wt male mice. All plot values obtained from single cell multiomic data normalized to sequencing depth from pristane-treated SJL/J mice: number of tgXist M+ Dox + Pristane= 8, Wt M+Dox+Pristane= 3, Wt F + Pristane= 2. Significance was calculated using Fisher’s Exact test.

3

Figure S3: Serum autoantibodies and ATAC and RNA sequencing from pristane-treated mice in the C57BL/6J background, Related to Figure 2. (A) ELISA of Serum ANAb levels, number of tgXist Female = 5, tgXist Male (no dox) = 4, tgXist Male + dox = 6. ANAb of WT male animals were done but did not meet quality control standards and therefore not shown. (B) Median Fluorescence Intensity (MFI) of serum reactivity to representative autoantigens using a bead-based lupus antigen array after 52 weeks of treatment. (A-B) From left to right tgXist Female control to tgXist Male test, number of animals in serum studies: n=5, n=4, n=4, n=6, n=6. (A-B) Statistical significance calculated with Wilcoxon Rank Sum Test, FDR < 0.05. (C) PCA plots of RNA-seq libraries of all five mouse cohorts. (D) Top 15 differential reactomes associated with ATAC-seq cluster 1 genomic regions. (E) Comparison of Fragments Per-Kilobase per-Million mapped fragments (FPKM) in the TLR9 gene region. Significance calculated using the Student’s T-test. * indicated p < 0.05. (C-E) Number of mice used: Wt F+ Pristane = 5, tgXist M+Pristane+Dox=6, tgXist M+Pristane=2, Wt M+Pristane=4, Wt M mock treatment=4.

4

Figure S4: Serum autoantibodies in pristane-treated mice in the SJL/J background, Related to Figure 3. Heatmap of the MFI values at 16 weeks (normalized to 0 weeks) of autoantigens displayed in Figure 3F. Color labels correspond to the mouse treatment group. Each row is an autoantibody; each column is an individual mouse. In the tgXist M+ Dox + pristane group, male mice with female-level total pathology score (score ≥ 10 in Fig 3E) are shown with underlined sample id numbers. However, the autoantibody response are heterogeneous; the four antibodies are a group are not significantly different between tgXist M + Dox + pristane vs Wt M + Dox+ pristane. Number of mouse serum samples, listed from left to right: wild-type Female + pristane =18, wild-type Male negative control=13, tgXist Male+dox (tgXist) control=11, wild-type Male treatment =12, tgXist Male treatment=10.

5

Figure S5, Related to Figure 4: Splenic CD45+ hematopoietic cells single-cell ATAC from pristane-induced SLE mice of the SJL/J strain: (A) Original single-cell ATAC clusters and (B) matched single cell gene expression-determined cell identities displayed on the single cell ATAC UMAP. (C) Localization of defining markers, calculated by imputation from ATAC data, used to determine cell type identity. Imputation scale Log2(NormCounts+1) (D) Pairwise comparison metrics of differential peaks between WT female (positive disease control) and the low disease male control groups (WT Male and tgXist Male low disease) across the four main cellular subsets shown for features FDR ≦ 0.1 and Log2FC ≧ 0.5. Pristane-treated mouse groups shown: tgXist male disease high (n=4), tgXist male disease low (n=4), wild-type male (n=3), and wild-type female (n=2). Significance was calculated using the Wilcoxon Rank Sum test.

6

Figure S6: Splenic CD45+ hematopoietic cells single-cell gene expression from pristine-induced SLE mice of the SJL/J strain, Related to Figure 5: (A) Localization of defining expression markers used to determine immune cell type identities of UMAP clusters. (B) Individual mice displayed on the single cell gene expression clustering UMAP. Individual mouse labels displayed as: transgenic status_sex_total disease damage score_mouse colony ID. (C) Volcano plots of differentially expressed B cell cluster genes comparing tgXist Male high disease and WT Female to WT Male. (D) Representative violin plots of B-cell marker genes from the combined B cell clusters. (E) T cell cluster genes comparing tgXist Male high disease and WT Female to WT Male. Pristane-treated mouse groups shown: tgXist male disease high (n=4), tgXist male disease low (n=4), wild-type male (n=3), and wild-type female (n=2). Significance was calculated using the Wilcoxon Rank Sum test.

7

Figure S7: Distribution of pristane-treated mouse cohorts and atypical B cells, Related to Figure 5. UMAP distribution of cells from (A) Wt female, (B) tgXist Male high disease, (C) Wt Male and (D) tgXist Male low diseases. Blue circles indicate shared overlapping regions in Wt female and tgXist Male high disease pristane-treated mouse groups corresponding to Bcell_3. (E) Dot plot of atypical B-cell marker expression in gene expression clusters. Pristane-treated mouse groups shown: tgXist male disease high (n=4), tgXist male disease low (n=4), wild-type male (n=3), and wild-type female (n=2).

8

Table S1: XIST Bibliomics, Related to Figure 6. Identification of known autoantigens in the XIST RNP and their associated autoimmune diseases and publications. 1) XIST-associated proteins were drawn from published and validated XIST ChIRP-MS datasets. 2) XIST-associated proteins classified as reactive or enriched from the XIST antigen array that were not included in the prior ChIRP-MS lists of high stringency validated XIST RNP proteins.

9

Table S2: Pristane-induced SLE mouse study controls and design, Related to Figures 2 and 3. Study design and appropriate matched control comparisons of each tgXist and wild-type mouse treatment cohort in the 1) C57BL/6J and 2) SJL/J backgrounds.

10

Table S3: Pristane-induced SLE antigen array, Related to Figures 2 and 3. Raw Mean Fluorescence intensity (MFI) values and mouse serum sample information of pristane-induced SLE studies to SLE autoantigens in wild-type and tgXist 1), 2) C57/BL6J and 3), 4) SJL/J mice.

11

Table S4: XIST Antigen Array in patients, Related to Figures 6 and 7. 1) Description and IDs of antigens used in the array to evaluate patient sera reactivity. Rows are labeled with “Gene name_Antigen name”, columns describe association (assay control, clinical disease autoantibody, XIST ChIRP, or Exploratory XIST), gene names, IDs in public databases, control or test status, multiplicity of targets for each fragment (also indicated as a “*” in the row labels as “Gene name*_Antigen name”), and fragment sequence for each antigen. 2) Sample disease and biological sex information. 3) Raw Mean Fluorescence intensity (MFI) values of patient serum reactivity to antigens.

12

Table S5: Reactive Antigens in Autoimmune Disease Patients, Related to Figures 6 and 7. Lists of all reactive antigens in 1) all autoimmune disease patients (All_Patients) and 2) separated by the three autoimmune disease cohorts (By_Disease) evaluated in this study and known status as autoantigens. 3) List of all antigens enriched in SLE patients with outlier-trained MAD difference > 0 and raw MAD difference >2.5-fold MAD scores compared to general population (Described in Methods). XIST RNP complex proteins in blue text. Highlighted antigens correspond to the published high confidence XIST RNP complex proteins¹⁴. “Autoantigen Record” indicates status of the antigen as a known disease control (Control), present in Table S1 XIST Bibliomics publications (Known), or not present in included Bibliomics publications (Unknown). Type of significantly reactive autoimmune disease cohorts indicated in “Array Reactivity”.

13

Table S6: XIST Antigen Array in SJL/J mice, Related to Figure 7. 1) Description and IDs of antigens used in the array to evaluate mouse sera reactivity. Rows are labeled with “Gene name_Antigen name”, columns describe association (assay control, clinical disease autoantibody, XIST ChIRP, or Exploratory Xist), gene names, IDs in public databases, control or test status, multiplicity of targets for each fragment (also indicated as a “*” in the row labels as “Gene name*_Antigen name”), and fragment sequence for each antigen. 2) Sample treatment, timepoint and sex information for each mouse. 3) Raw Mean Fluorescence intensity (MFI) values of mouse sera reactivity to antigens.

14

Table S7: Pristane-induced SLE in SJL/J Mice Histology, Related to Figure 3. Complete full body organ histology scores for all SJL/J mice in the study. Due to pandemic difficulties, a subset of mice were “affected by processing” that rendered the isolated kidney unusable for downstream analysis (indicated as “Y” in the column and “?” under glomerulonephritis) and some mice were missing splenic and lymph node scores. Only mice with all organ scores (in bold text) were included in the final Figure 3 Total Organ Pathology statistical analysis.

HIGHLIGHTS.

• Transgenic mouse models inducibly express Xist in male animals.

• Xist expression in males induce autoantibodies and autoimmune pathology.

• Xist in males reprograms T and B cell populations to female-like patterns.

• Autoantibodies to Xist RNP characterize female-biased autoimmune diseases in patients.