The XIST lncRNA is a sex-specific reservoir of TLR7 ligands in SLE

Abstract

Systemic lupus erythematosus (SLE) is a systemic autoimmune disease with a dramatic sex bias, affecting 9 times more women than men. Activation of Toll-like receptor 7 (TLR7) by self-RNA is a central pathogenic process leading to aberrant production of type I interferon (IFN) in SLE, but the specific RNA molecules that serve as TLR7 ligands have not been defined. By leveraging gene expression data and the known sequence specificity of TLR7, we identified the female-specific X-inactive specific transcript (XIST) long noncoding RNA as a uniquely rich source of TLR7 ligands in SLE. XIST RNA stimulated IFN-α production by plasmacytoid DCs in a TLR7-dependent manner, and deletion of XIST diminished the ability of whole cellular RNA to activate TLR7. XIST levels were elevated in blood leukocytes from women with SLE compared with controls, correlated positively with disease activity and the IFN signature, and were enriched in extracellular vesicles released from dying cells in vitro. Importantly, XIST was not IFN inducible, suggesting that XIST is a driver, rather than a consequence, of IFN in SLE. Overall, our work elucidated a role for XIST RNA as a female sex-specific danger signal underlying the sex bias in SLE.

Keywords: Autoimmune diseases; Autoimmunity; Immunology; Innate immunity; Lupus.

Figure 1. XIST is a sex-biased source of putative TLR7 ligands.

(A) Dot plot showing the number of UU dinucleotides and degree of sex bias in expression of all 15,003 transcripts captured in publicly available RNA-sequencing samples from blood (33). (B) Dot plot showing the maximum local (500-nucleotide) UU richness and degree of sex bias of all transcripts in A. (C) Volcano plot showing the sex bias of transcripts from A that also contain the known TLR7 ligand 5′-GUCCUUCAA-3′ (33). (D) Dot plot showing the number of UU dinucleotides and expression level of all transcripts in A. (E) Dot plot showing the maximum local (500-nucleotide) UU richness and expression of all transcripts in A. (F) Dot plot showing each transcript’s rank sum score calculated based on the rank of each transcript’s number of UUs, maximum UU richness, degree of female sex bias, and expression level. (G) Dot plot showing the average normalized rank sums for each transcript in tissues of particular interest in SLE (blood, spleen, and kidney). (H) Dot plot showing the average normalized rank sums for each transcript in all tissues. (I) A line chart indicating the local UU richness surrounding each point in the XIST sequence. The number of UUs in the 500-base section starting with each nucleotide is shown. The A-repeat region and XIST1.1, the region containing the 5′-GUCCUUCAA-3′ motif, are denoted. The dotted line indicates the average UU richness of all transcripts in the human transcriptome. (J) The first 1,000 nucleotides of the XIST sequence, which contains the A-repeat region, are shown. The A-repeats are shown as black text, with the surrounding UU-rich regions highlighted as red text. Nucleotides outside the A-repeat region are shown as gray text. (A–C) Differential expression testing was performed using empirical Bayes estimation with multiple comparisons corrections inherent.

Figure 2. TLR7 ligands in XIST activate human pDCs and HEK-hTLR7 cells.

(A) Sequences of the 4 RNA oligonucleotides used to stimulate pDCs and HEK-hTLR7 cells. (B) ELISA measuring IFN-α production by pDCs from 4 different healthy donors after transfection with 20-mer oligonucleotides. (C) Colorimetric assay showing the production of SEAP by HEK-hTLR7 cells in response to treatment with 20-mer oligonucleotides in 3 technical replicates. All treatments were compared with XIST1.1 by multiple comparisons at each dose. † Indicates XIST1.1 versus RNA9.2s, ^# indicates XIST1.1 versus RNA9.2a, and * indicates XIST1.1 versus polyA. One symbol indicates P < 0.05, 2 symbols indicate P < 0.01, 3 symbols indicate P < 0.001, and 4 symbols indicate P < 0.0001. (D) Fluorescence polarization values of XIST1.1, RNA9.2s, RNA9.2a, and polyA RNA when incubated with increasing doses of human recombinant TLR7. (E) ELISA measuring IFN-α production by pDCs after transfection with A-repeat or control RNA from 4 different healthy donors. IMQ was used as a positive control. (F) Colorimetric assay showing the production of SEAP by HEK-hTLR7 cells after transfection with A-repeat or control RNA in 6 technical replicates. (G) ELISA measuring IFN-α production by pDCs after transfection with A-repeat with or without TLR7 inhibitors HCQ or ODN from 3 different healthy donors. (H) ELISA measuring IFN-α production by pDCs after transfection with varying concentrations of A-repeat RNA or XIST1.1 in 3 technical replicates. A-repeat was compared to XIST1.1 by t test at each dose. (B–H) Error bars indicate 1 standard deviation. * Indicates P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001, and **** indicates P < 0.0001. (B and E–G) All treatments were compared with XIST1.1 or A-repeat by 1-way ANOVA with multiple comparisons correction.

Figure 3. XIST knockdown diminishes TLR7-activating potential of whole cellular RNA.

(A) A431 cells were targeted with CRISPR/Cas9 technology to generate XIST-depleted XIST-A and XIST-B cell populations. Representative RNAScope images showing XIST expression (green) and nuclei (blue) in WT, pCAS, XIST-A, and XIST-B cultures. Scale bar indicates 20 μm. (B) Percentage of XIST-positive cells was determined by masked quantification of 10 randomly chosen RNAScope image fields per cell population. (C) Bar graph showing XIST expression in terms of fragments per kilobase of exon per million mapped fragments (FPKM) in each cell population as measured by RNA sequencing. The dotted line at y = 7 shows the lower limit of detection in our RNA-sequencing experiment. (D) Colorimetric assay measuring SEAP secretion by HEK-hTLR7 cells transfected with fragmented cellular RNA from WT, pCAS, XIST-A, or XIST-B cells. Results shown are pooled from 3 independent experiments. (E) Bar graph showing XIST expression in WT, XIST-KO, and Jurkat cells as measured by qPCR. XIST expression calculated using the ΔΔCt method using GAPDH as an internal control and the WT cell line as the reference. (F) Colorimetric assay measuring SEAP secretion by HEK-hTLR7 cells transfected with fragmented cellular RNA from either WT or XIST-KO A431 cells. (B–F) All conditions compared with WT by multiple comparisons within 1-way ANOVA with multiple-comparison correction. ** indicates P < 0.01, *** indicates P < 0.001, and **** indicates P < 0.0001. (F) Conditions compared by unpaired t test. (B–F) Error bars represent 1 SD.

Figure 4. XIST expression in PBMCs correlates with SLE status and disease activity.

RNA flow cytometry was used to measure XIST and Rpl13A expression levels in PBMCs from women with SLE (n = 11) versus healthy women (n = 12). (A and B) Dot plots showing XIST (A) and RPL13A (B) MFI in SLE and control PBMCs. Healthy controls are also shown in Supplemental Figure 3C. (C–E) XIST MFI in B cells (C), T cells (D), and monocytes (E) from patients with SLE and controls. (F) Bar graph showing XIST MFI in SLE patients with a PGA ≥ 1 (n = 7) and a PGA < 1 (n = 4). (G) Scatterplot of XIST MFI versus SLEDAI. Pearson’s correlation coefficient and P value for linear regression shown. (A–E) * indicates P < 0.05, and ** indicates P < 0.01. Error bars indicate 1 standard deviation. (A–F) Expression of XIST or RPL13A compared between patients with SLE and healthy donors (A–E) or disease groups (F) by t test. (G) Correlation tested by simple linear regression.

Figure 5. XIST activates IFN production in a cell-extrinsic, TLR7-dependent mechanism.

(A) Scatterplot showing average XIST expression in immune cells from 15 patients with lupus nephritis from the AMP data set versus IFN score (37). (B) Bar graphs showing XIST and IRF7 expression in response to IFN-α treatment as measured by qPCR in HEK293 cells, A431 cells, SLE patient PBMCs, and primary human keratinocytes. Upregulation of IRF7 versus XIST or IFI6 versus XIST compared by t test. *** indicates P < 0.001, and **** indicates P < 0.0001. (C and D) Bar graphs showing XIST expression fold-change calculated using the ΔΔCt method relative to GAPDH (C) or the relative quantity released (D) of XIST in EVs, calculated as the fold-change in XIST multiplied by the fold-change in the amount of RNA isolated, as measured by qPCR. Release of XIST RNA compared by t test. **** indicates P < 0.0001. (E) Our model for the role of XIST in SLE pathogenesis. (I) Dying cells release XIST and self-antigens. (II) XIST RNA activates TLR7 in pDCs. (III) Activated pDCs make large amounts of IFN, inducing the IFN signature and an inflammatory environment in the target tissue. (IV) Inflammatory environment and lymphocyte infiltration leads to increased cell death, inducing more XIST release to perpetuate the cycle. (B–D) Error bars represent 1 SD.