SUMMARY
The long non-coding RNA (lncRNA) XIST establishes X chromosome inactivation (XCI) in female cells in early development and thereafter is thought to be largely dispensable. Here we show XIST is continually required in adult human B cells to silence a subset of X-linked immune genes such as TLR7. XIST-dependent genes lack promoter DNA methylation and require continual XIST-dependent histone deacetylation. XIST RNA-directed proteomics and CRISPRi screen reveal distinctive somatic cell-type-specific XIST complexes, and identify TRIM28 that mediates Pol II pausing at promoters of X-linked genes in B cells. Single-cell transcriptome data of female patients with either systemic lupus erythematosus or COVID-19 infection revealed XIST dysregulation, reflected by escape of XIST-dependent genes, in CD11c+ atypical memory B cells (ABCs). XIST inactivation with TLR7 agonism suffices to promote isotype-switched ABCs. These results indicate cell-type-specific diversification and function for lncRNA-protein complexes and suggest expanded roles for XIST in sex-differences in biology and medicine.
In Brief
XIST restrains human atypical B cell development. A distinct XIST-associated protein complex contributes to XIST-mediated gene silencing in B cells.
Graphical Abstract
INTRODUCTION
Eutherian mammals have evolved a unique strategy to epigenetically silence one of the two X chromosomes (chrX) in females, a process termed X chromosome inactivation (XCI) (Lyon, 1962). XCI is comprised of initiation, establishment and maintenance. During early embryogenesis, XCI is initiated by XIST, a lncRNA that is transcribed and spreads in cis across the entire inactive X (Xi) (Penny et al., 1996). Once XCI is established, the Xi is stably silenced via epigenetic mechanisms such as DNA methylation (Gartler and Riggs, 1983). Nonetheless, over 20% of X-linked genes (termed escapees) escape from XCI (Carrel and Brown, 2017). Some escapees are tissue-specific, highlighting a diversity of XCI maintenance across different tissues (Oliva et al., 2020; Tukiainen et al., 2017). Despite its central role in XCI initiation and establishment, XIST was long thought to be dispensable for XCI maintenance. Genetic deletion of XIST in cultured somatic cells did not reactivate the Xi (Brown and Willard, 1994; Csankovszki et al., 1999). However, recent studies with conditional knockout of mouse Xist in different tissues led to partial Xi reactivation (Adrianse et al., 2018; Yang et al., 2016, 2020). These observations highlight the need to understand when and where XIST is needed for XCI maintenance.
XIST lncRNA comprises several functional modules including A-F repeats (Lu et al., 2017). These repeats can bind and recruit diverse proteins that are responsible for chromatin modification, transcriptional silencing, XIST RNA spreading and coating (Chu et al., 2015; Loda and Heard, 2019). At the onset of XCI, XIST recruits SPEN via A-repeat to facilitate histone deacetylation (Dossin et al., 2020; Lu et al., 2016; Żylicz et al., 2019). Then XIST B and C repeat binding protein HNRNPK recruits Polycomb repressive complex (PRC) 1 and 2, which are responsible for later accumulation of repressive marks H2AK119b and H3K27me3 respectively, further locking in silencing (Almeida et al., 2020). XIST tethering to the Xi relies on C repeat binding protein HNRNPU and E repeat binding protein CIZ1 (Hasegawa et al., 2010; Sunwoo et al., 2017). Although XIST co-factors are well characterized during XCI initiation, the nature and function of the XIST ribonucleoprotein (RNP) complex during XCI maintenance in adult somatic cells is virtually unknown.
Escape from XCI has long been thought to contribute to sex differences. Because chrX harbors a high density of immune genes, incomplete XCI maintenance may lead to elevated X-linked immune gene dosage in females over males, contributing to female-biased autoimmune diseases (Klein and Flanagan, 2016). Indeed, 80% of patients with autoimmune diseases are women and such female-bias can be linked to the dosage of X-linked immune genes (Libert et al., 2010). For example, TLR7, an X-linked gene encoding Toll-like receptor 7, recognizes single-strand (ss) RNA-containing immune complexes that involve in female-biased autoimmunity and ssRNA viral infection (Celhar et al., 2012). The duplication of Tlr7 gene in mice suffices to drive lupus-like symptoms whereas the deletion of Tlr7 ameliorates the disease (Christensen et al., 2006; Subramanian et al., 2006). An important consequence of elevated TLR7 signaling is the formation and activation of CD11c+ atypical memory B cells (ABCs) (Hao et al., 2011; Rubtsov et al., 2011; Rubtsova et al., 2015). ABCs are a unique B cell population that greatly expands in aging, certain infectious diseases (malaria, HIV, and COVID-19 (Woodruff et al., 2020)), and female-biased autoimmunity (systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA)) (Cancro, 2020; Karnell et al., 2017). ABCs accumulate in aged female mice but not in age-matched male mice (Rubtsov et al., 2011). A recent study showed a complete lack of XIST localization on the Xi in CD11c+ ABCs (Pyfrom et al., 2020). Such abnormal XIST localization has also been observed in patients with SLE where CD11c+ atypical B cells are expanded (Wang et al., 2016) (Wang et al., 2016). How and why ABCs are female-biased and potential links to XIST action are largely unknown.
Here, we investigate the role of XIST for XCI maintenance in human B cells. We identify X-linked genes that require ongoing XIST-mediated silencing, characterize the required epigenomic mechanisms, and uncover XIST dysregulation in atypical B cells in multiple human disease states.
RESULTS
XIST is essential for maintaining X-inactivation of immune genes in human B cells
To determine if the XCI is maintained by XIST-mediated silencing or epigenetic memory such as DNA methylation and H3K27me3, we performed CRISPR interference (CRISPRi) of XIST in GM12878 B cells (sgXIST group), a female human B cell line with normal karyotype, leading to 75% XIST knockdown (Figure 1A-B, Figure S1A). We also treated the cells with DNMT and EZH2 inhibitors to block DNA methylation and H3K27me3 (inhibitors group). We performed allelic RNA-seq analysis using heterozygous single nucleotide polymorphisms (SNPs) provided by the phased genome (ENCODE Project Consortium, 2012). For genes with at least 10 reads that can be assigned to alleles, we calculate d-score to measure the allelic bias, which is defined as (readsXi/(readsXi+ readsXa)-0.5). The dscore has a range of −0.5 to +0.5: −0.5 means all the reads are from the active chrX (Xa) and +0.5 means all the reads are from the inactive chrX (Xi); 0 means the reads are biallelic.
Figure 1. XIST is essential for XCI maintenance of a subset of X-linked genes.
A. Schematic view of hypothesis.
B. Schematic view of experimental set up. Control group: B cells perturbed with non-targeting control sgRNA and treated with DMSO for 7 days. Inhibitors group: B cells perturbed with control sgRNA and treated with DNA methylation and Ezh2 inhibitors for 7 days. sgXIST group: B cells perturbed with XIST sgRNA and treated with DMSO for 7 days.
C. Cumulative distribution of d-score of allelic gene expression across the X chromosome in ctrl, inhibitors and sgXIST group. P value was calculated using Kolmogorov–Smirnov test.
D and E. Box plot showing the distribution of d-score of allelic gene expression on chrX (D) and chr1 (E) between control, inhibitors and sgXIST group. P values were from paired t-test.
F. Left heatmap showing the d-score of allelic gene expression on chrX. Right heatmap showing the hierarchical clustering of genes with differential allelic expression between groups. Value depicts the z score of the difference of d-score. Representative genes are highlighted.
G. Bar plot showing the percentage of immune-related gene and genes with other functions between XIST-dependent and -independent genes. P value was from Fisher exact test.
H. Box plot showing the d-score of allelic gene expression on X-linked immune genes. P values were from paired t-test.
I. RNA-seq tracks showing the allelic expression of X-linked immune gene TLR7. Xa, active X chromosome. Xi, inactive X chromosome.
J. Bar plot showing the RNA-seq normalized counts of TLR7 expression. Data are represented as mean +/− SEM
Treatment of epigenetic inhibitors led to 8066 differentially expressed genes (DEG) compared to the control group. In contrast, silencing of XIST led to only 510 DEGs (Figure S1B, Table S1). We noticed that upregulated DEGs after XIST knockdown are more enriched on chrX (24.4%) compared to inhibitors-treated cells (6.8%) (Figure S1D). These results indicated that XIST is the major contributor to maintain the gene dosage on the chrX in somatic B cells, supported by the correlation analysis of the chrX allelic bias across samples (Figure S1C). The cumulative distribution of d-scores across the chrX was significantly increased in sgXIST over control cells (p=0.002, K-S test), without changes of allelic bias on autosomes (Figure 1D-E, Figure S1E). 36 of 130 (27.6%) genes with lower expression in Xi (d-score<0) showed significantly increased d-score in sgXIST cells over control cells (d-scoresgXIST- d-scoreCtrl >0.03, binomial p<0.05), suggesting that they rely on XIST to maintain X-inactivation (Figure 1F). Thus, contrary to expectation, XIST is not fully dispensable for XCI maintenance in adult B cells.
We named these 36 genes as XIST-dependent genes, whereas the remaining 94 genes were named XIST-independent genes because their d-score is not significantly altered after XIST perturbation (Table S2). In control cells, the d-score is significantly higher in XIST-dependent compared to -independent genes, suggesting that formers are not stably silenced (Figure S1F). Indeed, we compared the proportion of variable and constitutive escapees that are defined from Genotype-Tissue Expression (GTEx) study (Tukiainen et al., 2017) and found these escapees are enriched in XIST-dependent compared to -independent genes (Figure S1G). Gene Ontology (GO) analysis of the genes upregulated in sgXIST group showed enrichment in immune response genes including TLR7, GPR174, CXCR3, and CXCL10 (GO enrichment p=7 × 10−5) (Figure S1H). Immune-associated genes are more enriched in XIST-dependent genes compared to -independent genes (p=0.01, Fisher’s exact test) (Figure 1G). We found increased d-score of X-linked immune genes in sgXIST group over control group, supporting that XIST is essential for Xi silencing of these genes (Figure 1H). Specifically, we observed an increased biallelic expression of TLR7 and ∼1.8 fold increase in TLR7 bulk RNA expression after XIST knockdown (Figure 1I-J). Re-expression of XIST by introduction of anti-CRISPR protein in XIST CRISPRi B cells resilenced the Xi (Figure S1I-K), indicating XIST can re-silence a previously active X in B cells.
Analysis of XIST inactivation in human fibroblasts, mouse brain tissue, and GTEx database suggest dependence on XIST and XCI escape occur in multiple lineages with variable magnitude and tissue specificity (Figure S2A-D).
Absence of DNA methylation predicts requirement of XIST for XCI maintenance
We hypothesized that XIST-dependent genes harbor chromatin and/or genetic features that could explain their unstable silencing and continuous dependence on XIST. Indeed, we found significantly lower DNA methylation near their TSSs, and lower H3K27me3 at their gene bodies, compared to XIST-independent genes (Figure 2A-B). Promoters of XIST-dependent genes contain fewer CpG islands (Figure S2E). DNA GC content, proximity to XIST locus, density of LINE, SINE, or LTR element were similar between XIST-dependent and -independent genes (Figure S2E-F). Regression analysis with all epigenomic features showed that low DNA methylation is significantly associated with the XIST-dependent genes and predicted whether the XCI maintenance of genes relies on XIST (Figure 2C-D). We then test if XIST-independent genes may become dependent on XIST for XCI maintenance when epigenetic memory is blocked. Treatment of XIST-depleted cells with epigenetic inhibitors significantly increased the d-score of XIST-independent genes compared to either perturbation alone (Figure 2E). Together, we show that XIST-dependent genes are not stably silenced due to the absence of DNA methylation. Moreover, XIST is functionally redundant with DNA methylation for the maintenance of silencing for a larger set of Xi genes.
Figure 2. Low DNA methylation is associated with genes that continuously rely on Xist for XCI maintenance.
A. Genome tracks showing ATAC-seq, medip-seq, ChIP-seq of H3K27ac, H3K27me3 and H3K9me3 profiles of representative XIST-dependent (TLR7) and -independent (MBNL3) genes. Shaded area highlights the promoter regions.
B. Comparison of DNA methylation, H3K27ac, H3K27me3, chromatin accessibility and H3K9me3 between XIST-dependent and -independent genes. P value was from Mann-Whitney test.
C. Scatter plots showing the linear regression analysis of relationship between XIST dependence (dscoresgXIST-dscorectrl) and epigenetic features.
D. ROC curve of logistic regression analysis showing the prediction of XIST dependence using epigenetic features DNA methylation and H3K27me3.
E. Box plot showing the d-score of XIST-independent genes between inhibitors, sgXIST, and the combination of inhibitors treatment and XIST perturbation group. P values were from paired t-test.
XCI maintenance in adult B cells relies on XIST-mediated histone deacetylation
XIST is critical for XCI initiation via the deacetylation of active mark H3K27ac and the accumulation of repressive mark H3K27me3. We generated XIST A-repeat KO B cells by CRISPR-Cas9 as a XIST KO model. The genomic deletion of XIST A repeat (overlapped with XIST promoter) leads to >99% loss of XIST RNA, consistent with Xist A-repeat KO mouse study (Hoki et al., 2009) (Figure S3A-C). Epigenomic mapping showed a significant increase of H3K27ac deposition at Xi when XIST is ablated, while autosomes showed no change (Figure 3B, Figure S3D), For example, XIST KO led to increased H3K27ac deposition at Xi of the immune gene GPR174 which was reported to impart sex differences in B cell migration (Zhao et al., 2020) (Figure 3A). This re-acetylation of H3K27 after the loss of XIST is preferentially enriched at XIST-dependent genes, and significantly enriched at distal enhancers rather than promoters (Figure 3B-C). This is consistent with the discovery that during XCI initiation, XIST promotes enhancer deacetylation by increasing the activity of histone deacetylase HDAC3, which pre-bound to those enhancers before XIST binding (Żylicz et al., 2019). In contrast, XIST KO did not cause significant change of H3K27me3 occupancy across Xi or at XIST-dependent genes (Figure 3D, Figure S3E). Thus, XIST is continuously required to maintain the silencing status of the Xi especially for XIST-dependent genes in part through H3K27ac deacetylation at enhancers in B cells.
Figure 3. XIST mediates deacetylation of H3K27ac during XCI maintenance in somatic B cell line.
A. H3K27ac ChIP-seq tracks of XIST-dependent gene GPR174.
B. Box plots showing the d-score of chrX and chr1 genes (left), or XIST-dependent and -independent genes (right) between H3K27ac allelic profiles of control and XIST KO group.
C. Box plot showing the H3K27ac allelic bias at enhancer and promoter regions of chrX genes.
D. ox plots showing the d-score of chrX and chr1 genes (left), or XIST-dependent and -independent genes (right) between H3K27me3 allelic profiles of control and XIST KO group.
All P values were from paired t-test.
See also Figure S3
XIST ChIRP-MS uncovers distinct somatic cell XIST RNP complexes for XCI maintenance
XIST is a modular lncRNA that can binds to diverse protein partners during XCI establishment (Chu et al., 2015; McHugh et al., 2015; Minajigi et al., 2015), but the composition of the XIST RNP in somatic cells is not known. We performed XIST ChIRP-MS in the B cell line GM12878 and myeloid cell line K562, which both represent somatic immune cell types with established XCI (Figure 4A, Table S3). We used formaldehyde crosslinking to detect both direct RNA-protein interactions and indirect protein-protein interactions in the XIST complex. RNAse-treated samples served as a negative control to cull non-specific proteins that bind to probes or beads (Figure S4A-B).
Figure 4. Identification of a unique XIST RNP complex in somatic B cell line by XIST CHIRP-MS.
A. Schematic view of ChIRP-MS of XIST bound proteins in ES cell line, GM12878 B cell line, and K562 myeloid cell line.
B. XIST RNA-associated protein network in B cell line revealed by Cytoscape. Proteins with known functions are annotated together. Color and value depicts the fold change of peptide counts of experimental group compared to Rnase treated group.
C-D. Scatter plots showing the comparison of XIST-bound protein enrichment between differentiated ES cell line and B cell line (C) or between myeloid cell line and B cell line (D). Blue: B cell-enriched proteins; green: ES enriched proteins; Orange: myeloid enriched proteins.
Two replicates of XIST ChIRP-MS data in B cells identified 115 XIST binding proteins (Figure S4C). These XIST co-factors can be segregated into groups based on their functions, including chromatin enzymes, nuclear matrix, m6A methylation, and RNA splicing (Figure 4B). The transcriptional repressor SPEN is recruited to Xi by XIST A-repeat and promotes gene silencing via modulating HDAC activity (Dossin et al., 2020; McHugh et al., 2015). We further found distinct XIST binding proteins in B cells such as TRIM28 (also known as KAP1), an epigenetic co-repressor that can bridge multiple histone modifiers for repression (Czerwińska et al., 2017; Iyengar and Farnham, 2011). To determine if the XIST RNP complex is similar between XCI establishment and maintenance, we compared the enrichment of peptide counts for XIST co-factors between B cells and ES cells with inducible expression of XIST. We found 57.8% of XIST co-factors are shared between initiation and maintenance (Figure 4C). This comparison revealed maintenance-specific XIST RNP subunits including RNF40 and TRIM28, suggesting XIST may recruit specific proteins for XCI maintenance. We further compared XIST co-factors between B cells and myeloid cells and found that 71.3% of XIST binding proteins are shared, suggesting that XIST-RNP also has cell-type specific subunits in somatic cells. Notably, TRIM28 is a B cell-specific XIST cofactor, suggesting that TRIM28 might be essential for XIST-mediated XCI maintenance in B cells (Figure 4D). TRIM28 fRIP-qRT-PCR specifically retrieved XIST RNA only in B cells but not in K562 myeloid cells or differentiating ESCs (Figure S4D), further validating that TRIM28 is a B cell-specific XIST binding protein. The XIST interactome in somatic immune cells identified a catalog of cell-type-specific XIST co-factors.
Allelic CRISPRi screen reveals essential XIST co-factors for XCI maintenance in B cells
To functionally validate if the XIST co-facotrs revealed by ChIRP-MS are essential for XCI maintenance, we developed a CRISPRi screen in the dCas9-KRAB expressed B cell line. The readout is the allelic expression of TLR7, an X-linked immune gene whose biallelic expression is associated with female-biased autoimmunity (Souyris et al., 2018). We selected 57 high-confidence XIST co-factors in B cells with a stringent cutoff (peptide counts>25, fold enrichment>10). We designed two sgRNAs targeting TSS adjacent regions per gene and cloned two sgRNAs under human or mouse U6 promoters in the same lentiviral vector. After individual sgRNA transduction and puromycin selection, we enriched for cells undergoing CRISPRi and isolated RNA for TLR7 allelic library preparation (Figure 5A, Table S4).
Figure 5. Allelic CRISPRi screen reveals key XIST cofactors responsible for XCI maintenance of X-linked immune gene TLR7.
A. Schematic view of CRISPRi screen of XIST cofactors on TLR7 allelic gene expression.
B. Bar plot showing the mean d-score of TLR7 from two replicate screens after individual perturbation of XIST cofactors. Blue: negative control (Ctrl); red: positive control (sgXIST). Data are represented as mean +/− SEM.
C. Same as in B. The enriched co-factors compared to negative control are grouped in different functions with color labeled.
D. Scatter plot showing the co-factors with their XIST binding signal shown by fold change of peptide counts and their effect on TLR7 allelic gene expression shown by allelic bias score. Known XIST repeat A, B/C and E binding factors are highlighted.
Gray dashed line in B-D indicates cut-off of the difference of d-score: d-scoreco-factor-dscorectrl=0.1
Depletion of 7 out of 57 XIST cofactors individually increased the d-score of TLR7 compared to non-targeting sgRNA control, suggesting that these co-factors are critical for XCI maintenance of TLR7 (cut-off delta d-score: d-scoreco-factor - d-scoreCtrl >0.1) (Figure 5B). The 7 key XIST co-factors are SPEN, TRIM28, RNF20, RNF40, PTBP1, MATR3, and HNRNPA0, and the first 4 proteins are involved in chromatin modification (Figure 5C). Notably, despite similar XIST binding affinity and perturbation efficiency as XIST A-repeat binding factor SPEN, depletion of XIST repeat B or C binding factors HNRNPK and HNRNPU did not affect XCI maintenance (Figure 5D, Figure S5A). The results indicated that HNRNPK-recruited PRC complexes and subsequent H3K27me3 deposition may be dispensable for XCI maintenance, highlighting SPEN-mediated H3K27ac deacetylation for XCI maintenance (Figure 3E; Figure 5D). To test this, we generated a SPEN mutant B cell line with deletion of SPEN RRM2-4 domain, which directly interacts with XIST for gene silencing (Carter et al., 2020; Dossin et al., 2020; Monfort et al., 2015) (Figure S5B). SPEN mutation led to significant reactivation of Xi specifically at XIST-dependent genes (Figure S5C-D). These data fully support that XIST recruits SPEN to maintain XCI likely via SPEN-HDAC3 mediated H3K27 deacetylation.
PTBP1 and MATR3 have been recently reported to regulate XCI maintenance (Pandya-Jones et al., 2020). PTBP1 and MATR3 are not required for establishment of XCI but only for maintenance in ES cells, which is consistent with our discovery of both factors as required to maintain XCI in adult B cells. Importantly, depletion of B cell-specific XIST cofactor TRIM28 led to Xi reactivation of TLR7, indicating that potential TRIM28-mediated mechanisms participate in XCI maintenance (Figure 5C).
TRIM28 is required for B cell XCI maintenance of a subset of X-linked genes
To further determine if TRIM28 is critical for XCI maintenance, we generated TRIM28 KO B cells, which reduced TRIM28 protein by >95% (Figure S6A). TLR7 was indeed reactivated on Xi in TRIM28-KO cells (Figure 6A). However, d-score analysis across chrX showed a modest effect on Xi reactivation in TRIM28 KO cells (Figure 6B). TRIM28 loss leads to increased d-score in only a subset of genes and 64.7% of them are also XIST-dependent (Figure 6C). The increased d-score of the TRIM28-dependent genes (d-scoreTRIM28 KO-d-scoreCtrl >0.03) is due to re-activation of Xi without significant change on Xa (Figure S6B). ChIP-seq analysis showed TRIM28 occupied ZNF genes on autosomes with concordant deposition of the repressive mark H3K9me3 (Figure S6C), as expected (Iyengar and Farnham, 2011). On chrX, TRIM28 is enriched at the promoters of TRIM28-dependent genes, such as TXLNG compared to TRIM28-independent genes (Figure 6D-E). Allelic analysis of TRIM28 ChIP-seq data revealed that TRIM28 binds to both Xa and Xi (Figure S6D). We compared the allele bias of TRIM28 binding between TRIM28-dependent and independent genes, and found a significantly increased binding at Xi rather than Xa of TRIM28-dependent genes compared to -independent genes (p=0.0008, Wilcoxon rank-sum test) (Figure S6E).
Figure 6. XIST cofactor TRIM28 participates in XCI maintenance of a subset of genes.
A. RNA-seq tracks of TLR7 gene at Xi (blue) and Xa (yellow) between control, TRIM28 KO and sgXIST group.
B. Box plot showing the d-score of chrX genes between control, TRIM28 KO and sgXIST group. P value was from paired t-test.
C. Heatmap showing the hierarchical clustering of genes with differential allelic expression between groups. Value depicts the z score of difference of the d-score.
D. Genome tracks showing TRIM28 ChIP-seq and RNA-seq in control and TRIM28 KO cells for TRIM28-dependent gene TXLNG.
E. Box plot showing TRIM28 binding density at TSS+/−1kb or gene body regions between TRIM28-dependent and -independent genes (left). Average diagram showing the reads per million for TRIM28 ChIP-seq at TRIM28-dependent (red) and -independent genes (blue) at TSS +/−2kb (right).
F. Box plots showing RNA Pol II binding density (log2 counts per kb) (left), promoter pausing ratio (log2(read densityTSS/read densitygene body)) of Ser5 phosphorylated RNA Pol II (middle); and Ser2 phosphorylated RNA Pol II (right) at TRIM28-dependent (red) and -independent genes (blue).
G. Box plot showing GRO-seq analysis of Pol II promoter pausing ratio (log2(read densityTSS/read densitygene body)) at TRIM28-dependent (red) and -independent genes (blue).
H. Bar plot showing the TRIM28 RIP-qPCR on A-F repeats regions of XIST and Actin with no crosslink (blue), formaldehyde crosslink (green), and UV crosslink (orange).
I. A proposed model showing TRIM28-mediated XCI maintenance.
P values in E-G were from Mann-Whitney test.
See also Figure S6
TRIM28 is a transcriptional co-repressor that can evoke several silencing mechanisms. These include recruitment of SETDB1, a histone methyltransferase for H3K9me3, interaction with HDAC-NuRD complex for histone deacetylation, or inhibition of pTEFb to impact RNA polymerase (Pol II) pause release (Ma et al., 2019; Schultz et al., 2001, 2002). Therefore, we examined diagnostic epigenomic features of each of these mechanisms in turn. We observed variable H3K9me3 deposition between TRIM28-dependent and -independent X-linked immune genes (Figure S6F). We also observed a higher H3K27ac at promoters of TRIM28-dependent genes and no significant difference of H3K9me3 and H3K27me3 (Figure S6G), suggesting an alternative regulatory mechanism of TRIM28 beyond erasing H3K27ac or writing H3K9me3. We compared the binding profile of RNA Pol II, Pol II Ser5 phosphorylation (Ser5P, transcription initiation mark), and Pol II Ser2 phosphorylation (Ser2P, transcription elongation mark) between TRIM28-dependent and -independent genes. We found Ser2P enrichment at the promoter relative to the gene body is significantly higher in TRIM28-dependent over -independent genes, whereas their Ser5P occupancy levels are quite similar (Figure 6F). This result indicates that TRIM28-dependent genes may possess higher promoter-proximal pausing of Ser2P Pol II and thus less efficient transcription elongation.
We further measured the Pol II promoter-proximal pausing ratio using GRO-seq data (Core et al., 2014). Indeed, there is a dramatic enrichment of nascent RNA at the promoter of TRIM28-dependent gene TLR7, but not for TRIM28-independent gene SASH3 (Figure S6H). We found a significantly higher promoter-proximal pausing for TRIM28-dependent genes, suggesting that TRIM28 occupancy may inhibit transcription elongation of these genes to maintain XCI (Figure 6G). The RING domain of TRIM28 is responsible for CDK9 sumoylation to inhibit CDK9 kinase activity, that is essential for transcription elongation (Ma et al., 2019) (Ma et al., 2019). The PHD domain of TRIM28 sumoylates adjacent Bromodomain (BR) domain, facilitating recruitment of SETDB1 (Czerwińska et al., 2017). We re-introduced TRIM28 full length protein or TRIM28 mutants with RING domain or PHD-BR (PB) domain deletion into TRIM28 KO cells (Figure S6I). We also created SETDB1 CRISPRi cells which led to 82% knockdown of SETDB1 (Figure S6J). Allelic analysis by pyrosequencing on three TRIM28-dependent immune genes showed that full length TRIM28 protein rescued the TRIM28 KO for gene silencing as expected. Deletion of TRIM28 RING domain led to significant reactivation of all three TRIM28-dependent genes, suggesting RING domain-mediated CDK9 sumoylation and Pol II pausing are important for XCI maintenance of these genes. In contrast, deletion of TRIM28 PB domain or SETDB1 depletion had little effect on the allelic bias, indicating PB domain mediated SETDB1 recruitment and SETDB1-mediated H3K9me3 are dispensable (Figure S6K).
Next, we performed TRIM28 RIP-qRT-PCR to assess the domain of XIST RNA that interacts with TRIM28 (Figure 6H). TRIM28 immunoprecipitation with formaldehyde crosslinking, but not UV crosslinking, specifically retrieved endogenous XIST A-repeat but not other portions, suggesting that TRIM28 indirectly interacts with XIST possibly through interaction of XIST A-repeat binding proteins (Figure 6H-I).
Finally, to test if TRIM28 is required in XCI maintenance in a B cell-specific manner, we analyzed the published shRNA screen data in differentiated ES cells with inducible XIST expression and a GFP reporter (Moindrot et al., 2015). We found that TRIM28 shRNA is not enriched in GFPhi cells (Figure S6L), suggesting that TRIM28 is not required for X-inactivation in differentiated ES cells. We also performed TRIM28 CRISPRi in K562 myeloid cells and achieved ∼70% knockdown of TRIM28 (Figure S6M). The loss of TRIM28 lead to a significant upregulation of four TRIM28-dependent X-linked genes only in B cells but not in myeloid cells (Figure S6N). ChIP-seq analysis showed TRIM28 occupied these gene loci only in B cells (Figure S6N). Thus, both functional and biochemical evidence indicate cell-type specific action of TRIM28 on Xi genes in B cells.
Taken together, these results suggest that B cell-specific XIST cofactor TRIM28 may prevent efficient transcription elongation via RING domain-mediated sumoylation to maintain X-inactivation in B cells (Figure 6I).
Escape of XIST-dependent genes in CD11c+ atypical B cells
Increased expression of X-linked genes along with the aberrant XIST localization was observed in immune cells from SLE patients, suggesting abnormal XCI maintenance may predispose women to autoimmune diseases (Syrett et al., 2019). We performed functional annotation analysis on genes that are overexpressed after XIST depletion in human B cell line. The result showed significant enrichment of genes related to IFN-γ cytokine production, and female-biased autoimmune diseases SLE and RA (Figure 7A), suggesting that XIST-mediated XCI maintenance may be essential to prevent overexpression of X-linked genes that may contribute to autoimmunity. To test this, we used gene set enrichment analysis (GSEA) and found that genes normally silenced by XIST tend to be over-expressed in patients with SLE compared to healthy donors (Figure S7A). Recent studies showed that IgD-CD27-CD11c+ atypical B cells (ABCs) are induced by TLR7 and contribute to pathogenic responses in SLE (Jenks et al., 2018). Our GSEA result showed that XIST-silenced genes tend to be over-expressed in these ABCs of SLE patients (Figure S7A). To further explore the impact of XIST-mediated XCI maintenance in SLE, we leveraged our discovery of “XIST-dependent” genes (X-linked genes whose maintenance are dependent on XIST) and “XIST-independent” genes (control X-linked genes whose maintenance are not). The GSEA analysis showed that XIST-dependent genes are significantly overexpressed in PBMC and IgD-CD27-CD11c+ ABCs of female SLE patients, while the XIST-independent genes were not (Figure 7B, Figure S7B). These results highlight the potential role of XIST dysregulation and subsequent escape of XIST-dependent genes in specific B cell subsets (CD11c+ ABCs) in certain diseases.
Figure 7. Escape of XIST-dependent genes in CD11c+ atypical B cells, which are increased after the loss of XIST.
A. Bar plot showing significantly enriched biological pathway and diseases from genes that are upregulated after XIST perturbation in B cell line.
B. GSEA analysis showing enrichment of XIST-dependent and -independent X-linked genes in PBMC of female SLE patients compared to healthy donors. Normalized Enrichment Score (NES) and FDR q value are shown.
C-D. Uniform manifold approximation and projection (UMAP) plot showing B cell subsets from female COVID-19 patients (C), and XIST escape score (D).
E. Diffusion map showing the pseudotime trajectory using naive cells as root (left). B cell subset projected on diffusion map (right).
F. Violin plot showing the distribution of XIST escape score between CD11c+ atypical memory B cells and conventional memory B cells in female (left) and male (right) COVID-19 patients. P value was from unpaired t-test.
G-H. UMAP showing XIST escape score of CD11c+ atypical memory cells and conventional memory B cells in kidney of female lupus patients (G), and in joint tissue of female RA patients (H). Violin plot showing the distribution of XIST escape score in two B cell subsets (right). P value was from unpaired t-test.
I. Schematic view of XIST KO in human primary B cells upon B cell activation by BCR crosslinking and TLR7 ligand stimulation.
J. Flow cytometry analysis of viable CD19+ human primary B cells between AAVS1 control and XIST KO group(left). Box plot showing the frequency of IgD+CD11c-, IgD-CD11c+, IgDCD11c-B cells among total B cells in two to three replicates from 8 female donors.
K. Flow cytometry analysis of IgG+ B cells gated on IgD-CD11c+ B cell subset. Box plot showing the frequency of IgD-IgG+CD11c+ and IgD-IgG-CD11c+ B cells among total B cells. P value in J-K was from paired t-test.
L. A proposed model of XCI maintenance in human B cells.
We developed a single cell gene expression score, termed “XIST escape score”, to evaluate XIST dysregulation in single cell transcriptome data across diverse human disease states. The XIST escape score is the normalized mean expression value of the XIST-dependent X-linked gene set minus that of the XIST-independent gene set. Thus, the score reflects the coordinate induction of XIST-dependent genes while controlling for overall X-linked gene expression. Applying the XIST escape score to 11 human immune cell scRNA-seq data sets, we discovered specific disease states and B cell subsets where XIST dysregulation occur in vivo (Table S7). Interestingly, we found that escape of XIST-dependent genes occurs during COVID-19 infection, specifically in CD11c+ ABCs (Figure 7C-D). Single cell analysis distinguished CD11c+ ABCs from naive B and conventional memory cells (Figure S7C-D). Cell fate trajectory analysis revealed a bifurcation of the differentiation trajectory from naive B cells to conventional memory B cells vs. CD11c+ ABCs (Figure 7E). Naive B cells with top 10 percentile of XIST escape score project strongly into ABC fate (Figure S7E), suggesting that XIST dysregulation may bias this differentiation choice. XIST dysregulation, as reflected by the gain in XIST escape score, is evident in CD11c+ ABCs from female patients but not male patients (Figure 7F), consistent with this being a female-specific phenomenon related to X chromosome dosage. XIST dyregulation in ABCs during COVID-19 infection was replicated in two additional independent patient cohorts (Table S7).
We discovered a similar XIST dysregulation in kidney infiltrating B cells of lupus nephritis and joint tissue infiltrating B cells of rheumatoid arthritis in female patients. In both cases, CD11c+ ABCs demonstrate significantly higher XIST escape score over conventional memory B cells (Figure 7G-H). Notably, this phenomenon is not observed in B cells from healthy donors, nor female B cells in influenza viral infection, SFTS viral infection, or tumor-infiltrating B cells in breast cancer patients (Table S7). These results suggest that escape of XIST-dependent genes occurs in vivo in several diseases that associated with RNP antigens, and occurs specifically in CD11c+ ABCs that infiltrate the pathologic tissues.
Guided by our single cell trajectory analysis, we hypothesized that XIST dysregulation may promote the differentiation of naive B cells to CD11c+ ABCs. To test this, we deleted XIST in primary human B cells from healthy female donor blood and monitored the B cell differentiation in response to BCR crosslinking and TLR7 stimulation that mimic ssRNA/RNP stimulation. First, we utilized electroporation of Cas9-RNP complex to delete XIST A-repeat or AAVS1 safe harbor locus as control in primary female B cells (Figure 7I; Figure S7F-G). After gating on viable CD19+ B cells (Figure S7H), we found a significant increase of IgD-CD11c+ ABCs in XIST KO group over AAVS1 group (Figure 7J). It is known that TLR7 upregulation can drive the accumulation of CD11c+ atypical B cells (Rubtsov et al., 2011). Indeed, XIST deletion in primary B cells showed ∼2.8 fold increase of TLR7 expression compared to AAVS1 group (Figure S7I). TLR7 escape from XCI has been reported to associate with increased IgG class-switching (Souyris et al., 2018). We observed a significant increase of IgG class-switched CD11c+ ABCs after XIST loss (Figure 7K). Our single cell analysis of CD11c+ ABCs showed unique expression pattern including a mixture of IgD-CD27- and IgD-CD27+ populations and higher expression of T-BET and ZEB2 compared to conventional memory cells (Figure S7D). We found that both IgD-CD27-CD11c+ and IgD-CD27+CD11c+ B cell subsets are indeed significantly increased in XIST KO group (Figure S7J-K), and show increased expression of T-BET and ZEB2, key TFs for CD11c+ ABCs (Cancro, 2020) (Figure S7L).
DISCUSSION
Continual requirement for XIST in XCI maintenance of somatic cells
Our study identified a central role of XIST in maintaining X-inactivation in adult human B cells. Classic studies using somatic cell hybrids that deleted the X inactivation center first suggested that XIST is dispensable for XCI maintenance (Brown and Willard, 1994). More recent studies using conditional Xist knockout mouse models suggest a range of Xi gene reactivation (Adrianse et al., 2018; Bhatnagar et al., 2014; Yang et al., 2016, 2016; Yildirim et al., 2013). These differences may be due to cell type-specific regulatory mechanism of XCI maintenance, and variable dependency of different X-linked genes on XIST for XCI maintenance. With their substantially longer lifespan than mice, human may be a particularly good model to address the epigenetic memory of XCI maintenance.
X-linked genes lacking promoter DNA methylation are particularly sensitive to XIST loss in B cells. After blocking DNA methylation and H3K27me3, genes classified as XIST-independent become reliant on XIST for XCI maintenance. These data demonstrate that the absence of DNA methylation or other epigenetic memory makes XIST RNA-based transcriptional repression necessary. Thus, our data suggest a redundant model for XCI maintenance: epigenetic memory such as DNA methylation and XIST RNA-based transcriptional silencing can compensate each other to maintain a robust and durable X-inactivation (Figure 7L). This is in line with a previous study revealing a synergism of XIST RNA and DNA methylation (Csankovszki et al., 2001).
Cell type-specific XIST RNPs to prevent escape from X-inactivation
Our data revealed TRIM28 as a B-cell specific XIST cofactor that is critical to XCI maintenance of a subset of X-linked genes. Since allelic binding of TRIM28 is not enriched at the Xi and there is no direct interaction between TRIM28 and XIST based on UV RIP, we reason that XIST does not directly recruit TRIM28 to the Xi. It is possible that when XIST recruits proteins to spread across the Xi, the XIST RNP comes into contact with TRIM28 that pre-bound at certain loci, leading to indirect interaction between TRIM28 and XIST RNP on chromatin that we detected by CHIRP-MS and RIP experiments with formaldehyde crosslinking. TRIM28 interacts with the A-repeat of XIST, suggesting functional collaboration with other A repeat binding silencing factors such as SPEN and RBM15. Our data suggest two models: (1) TRIM28 acts at promoters while SPEN and other XIST-cofactors act on both enhancers and promoters (Dossin et al., 2020; Żylicz et al., 2019). (2) XIST-RNP regulates the repressor activity of TRIM28. 35% of TRIM28-dependent genes are not dependent on XIST for silencing, favoring the first model. We showed that TRIM28 promoter occupancy is significantly associated with RNA Pol II promoter-proximal pause to cooperate with XIST-mediated H3K27 deacetylation at enhancers, unveiling a regulatory mechanism for XCI maintenance.
The discovery of unique XIST RNPs in somatic cells suggests that different somatic cells can have overlapping but still different XIST co-factors mediated silencing to prevent XCI escape. Like a transcription factor that collaborates with additional transcription coactivators to regulate different target genes in different cell types, the same lncRNA can also recruit, assemble, and guide different protein complexes to enact distinct consequences in different cell types. Studies of somatic cell XIST RNPs should shed light on XCI escape across mammalian organs and cell types.
XIST dysregulation in pathogenic atypical B cells in human disease
Sex is a major determinant of clinical outcomes in autoimmune diseases and COVID-19 and drives different immune responses in men vs. women (Klein and Flanagan, 2016; Takahashi et al., 2020; Williamson et al., 2020). We developed a computational strategy to track XIST dysregulation in any B cell single-cell gene expression data, documenting the in vivo occurrence of escape of XIST-dependent genes in CD11c+ ABCs in multiple diseases. We further experimentally validated that the loss of XIST can facilitate the naive cell differentiation into CD11c+ atypical B cells, providing a first causal link between a known female-enriched B cell population with XIST dysregulation. ABCs exhibits a strong female bias only in aged mice where the hormone difference is minimized, suggesting that XIST-mediated XCI maintenance contributes to the sex dimorphism of this subset (Rubtsov et al., 2011; Rubtsova et al., 2015). Our functional annotation of XIST-silenced genes showed enrichment of the production of IFN-γ which induces the expression of T-BET, a master TF that promotes the differentiation of CD11c+ ABCs upon TLR7 stimulation (Zumaquero et al., 2019). In addition, CXORF21 is another XIST dependent gene that encodes a key adaptor protein for TLR7 signaling (Heinz et al., 2020). Therefore, XIST is essential to balance the gene dosage of X-linked immune genes that can greatly impact the formation of CD11c+ ABCs, a rare population in healthy donors but aberrantly expanded in certain infectious diseases and female-biased autoimmune diseases (Cancro, 2020). Future work on how XCI maintenance is impaired in ABCs and other cell types will provide a better understanding of sex dimorphism in human diseases.
Limitations of the Study
In this work, we used genetic ablation of XIST to probe the functions of XIST in XCI maintenance. XIST RNA delocalization is the prevalent phenomenon in B cells of female-biased autoimmunity, and the consequences of XIST delocalization may be more subtle or different than XIST inactivation. We relied heavily on an immortalized B cell line for our epigenomic and mechanistic studies, and futures studies are needed to evaluate these mechanisms in normal B cells in vivo. In addition, our identification of Xi genes requiring XIST maintenance is limited by the available SNPs; denser allelic variants may empower discovery of additional genes requiring this mechanism. Finally, our XIST escape score is based on XIST perturbation in B cells and should only be applied to this cell type. Future studies that illuminate cell type-specific requirements of XIST in XCI maintenance will allow a similar approach to be applied additional cell types in the body.
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 ([email protected]).
Materials Availability
The plasmids of TRIM28 mutants with N-terminal 180bp replacement and RING domain or PB domain deletion were available upon request to the Lead Contact.
Data and Code Availability
The accession number for all the sequencing data in this paper is GEO: GSE164596. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD023516. Details of analysis are provided in STAR methods, and further questions should be directed to the Lead Contact.
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Generation of stable cell lines and cell culture
GM and K562 cell lines were cultured in RPMI 1640 (Thermo Fisher Scientific) with 10% FBS (Thermo Fisher Scientific) and 1% Penicillin-Streptomycin (Gibco). HEK 293T cells were cultured in DMEM (Thermo Fisher Scientific) with 10% FBS and 1% Penicillin-Streptomycin. IMR90 cell line were cultured in EMEM (ATCC) with 10% FBS and 1% Penicillin-Streptomycin. Female mESC were cultured on gelatin-coated plates in N2B27 medium for 7 days and then differentiated into NPC using N2B27 medium supplemented with EGF and FGF (both 10ng/mL, Peprotech). To inhibit DNA methylation and H3K27me3, B cell line was treated with 0.3 uM DNMT inhibitor 5-azacytidine and 2 uM EZH2 inhibitor EPZ-6438 for 7 days.
For generation of XIST CRISPRi GM B cell line, we performed lentiviral transduction of sgXIST in dCas9-KRAB expressing GM B cell line which has been previously made in the lab (Rubin et al., 2019). We designed two sgRNAs targeting the XIST promoter and cloned them into one lentiviral vector as previously described (Rubin et al., 2019). To generate the CRISPRi virus, we cultured HEK 293T cells at 4 million per 10cm dish and transfected with 4.5 ug pMP.G, 1.5 ug psPAX2 and 6 ug sgRNA vector using OptiMEM and Lipofectamin 3000 (Cat# L3000015, Thermo Fisher Scientific) at the following day. Two days later, the supernatant was collected and filtered with 0.44 um filter. The virus was concentrated 1:10 using Lenti-X Concentrator (Clontech). dCas9-KRAB+ GM cells were seeded at 300K cells per well of a 6-well plate and 40 uL concentrated virus was added to the media. Two days later, we select the XIST sgRNA expressing cells by adding 2 ug/mL puromycin for seven days and cells expressing mCherry were sorted by BD Aria II FACS sorter. SETDB1 CRISPRi GM B cell line is generated similarly as XIST CRISPRi GM B cell line.
For generation of anti-CRISPR expressing XIST CRISPRi or control CRISPRi B cell line, we performed lentiviral transduction of anti-CRISPR protein in non-targeting control or XIST (sgXIST) CRISPRi GM B cell line which is generated above. To generate the anti-CRISPR virus, we cultured HEK 293T cells at 4 million per 10cm dish and transfected with 4.5 ug pMP.G, 1.5 ug psPAX2 and 6 ug Fuw-AcrIIA4-P2A-GFP vector (Addgene #108247) using OptiMEM and Lipofectamin 3000 at the following day. Two days later, the supernatant was collected and filtered with 0.44 um filter. The virus was concentrated 1:10 using Lenti-X Concentrator. SgNT or sgXIST CRISPRi cells were seeded at 300K cells per well of a 6-well plate and 40 uL concentrated virus was added to the media. A week later, we select the anti-CRISPR and CRISPRi expressing cells by sorting cells expressing both GFP and mCherry using BD Aria II FACS sorter. The anti-CRISPR efficiency is measured by qRT-PCR of XIST.
For generation of XIST CRISPRi IMR90 cell line, we first established dCas9-KRAB expressing IMR90 cell line by lenti-viral transduction of dCas9-BFP-KRAB (Addgene #46911) and sorted BFP+ cells by BD Aria II FACS sorter. Next, we performed transduction of lentivirus expressing sgXIST or sgNT (non-targeting control) in dCas9-KRAB expressing IMR90 cell line and selected by adding 2 ug/mL puromycin for 7 days and sorted cells expressing both BFP and mCherry. The CRISPRi efficiency is measured by qRT-PCR of XIST.
For generation of TRIM28 CRISPRi K562 cell line, we first established dCas9-KRAB expressing K562 cell line by lenti-viral transduction of dCas9-BFP-KRAB and sorted BFP+ cells by BD Aria II FACS sorter. Next, we performed transduction of lentivirus expressing sgTRIM28 or sgNT in dCas9-KRAB expressing K562 cell line and sorted cells expressing both BFP and mCherry. The CRISPRi efficiency is measured by qRT-PCR of TRIM28.
For generation of XIST A repeat KO GM B cell line, we used CRISPR Cas9 RNP editing to cut XIST A repeat. SgRNAs were produced by combining tracr and crRNA according to the Synthego Biosciences protocol. Two sgRNAs (30 uM) targeting both ends of XIST repeat A were complexed with 3 ug Cas9 protein (IDT) for 10 min and then electroporated to GM cells (1700V, 20ms, 1 pulse) using Neon Transfection System (Thermo Fisher Scientific). Transfected cells were sorted into 96-well plate for single cell per well and then single cell clone with repeat A deletion was screened by genomic DNA PCR and RT-qPCR.
GM single cell clones with SPEN RRM2-4 deletion were generated similarly as XIST A repeat KO and screened by genomic DNA PCR and RT-qPCR. GM single cell clones with TRIM28 deletion were generated similarly and screened by western blot. All sgRNA sequences were listed in Table S5.
METHOD DETAILS
CHIRP-MS and proteomics analysis
25 T152 flasks of GM12878 or K562 cells were used for ChIRP-MS (0.5–1 billion cells) as previously described (Chu et al., 2015). Cells were cross-linked in 3% formaldehyde for 30 min followed by 0.125 M glycine quenching for 5 min. Cross-linked cells were lysed in fresh NLB buffer (50 mM Tris pH 7.0, 10 mM EDTA, and 1% SDS) and sonicated in 1 mL Covaris tube for 20 min with the following parameters (Fill level:10; Duty Cycle:15; PIP:140; Cycles/burst:200). The supernatant was pre-cleared with streptavidin beads (Invitrogen) and then treated with or without 30 ug/mL Rnase A (Cat # EN0531, Thermo Fisher Scientific) at 37°C for 45 min. Then the supernatant was hybridized with 8.8 uL XIST probes in Hybridization Buffer (50 mM Tris pH 7.0, 1 mM EDTA, 1% SDS, 750 mM NaCl, and 15% Formamide) at 37°C rotating overnight and added with 880uL of streptavidin beads and rotate for 45 min at 37°C. Beads were washed 5 times in ChIRP Wash Buffer (2X SSC and 0.5% SDS) for 5 min at 37°C. RNA extraction can be performed using a small aliquot of post-CHIRP beads to assess the XIST RNA retrieval efficiency. XIST bound proteins were eluted from beads in Biotin Elution Buffer (12.5 mM biotin (Invitrogen), 7.5 mM HEPES, 75 mM NaCl, 1.5 mM EDTA, 0.15% SDS, 0.075% sarkosyl, 15% Formamide, and 0.02% Na-Deoxycholate) at RT for 20 min then shake at 65°C for 10min, finally precipitated by TCA at 4°C overnight. The proteins were pelleted at 16000 rcf at 4°C for 30 min and the protein pellets were washed once with cold acetone and air-dry for 1 min, then the proteins were solubilized in 1X laemmli sample buffer and boiled at 95C for 30 min. Final proteins were size-selected on 4–12% NuPAGE gels for mass spectrometry. Gel slices were excised and diced to 1 mm cubes prior to proteolytic digestion. Samples were reduced in 5 mM DTT in a 50 mM ammonium bicarbonate buffer at 55°C for 30 mins. After removing the residual solvent, proteins were alkylated with 10 mM acrylamide in the same buffer, rinsed with 50% acetonitrile, and then digested using Trypsin-Lys C (Promega) overnight at 37°C to obtain peptides. Samples were centrifuged to condense particulates so that the solvent including peptides could be collected. A further peptide extraction was performed by the addition of 60% acetonitrile, 39.9% water, 0.1% formic acid and incubation for 10–15 min before collection. Samples were dried by speed vac prior to resuspension in 2% aqueous acetonitrile with 0.1% formic acid for mass spectrometry. Mass spectrometry experiments were performed on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, San Jose, CA) with liquid chromatography using an Acquity M-Class UPLC (Waters Corporation, Milford, MA). For a typical LC/MS experiment, a flow rate of 450 nL/min was used, where mobile phase A was 0.2% formic acid in water and mobile phase B was 0.2% formic acid in acetonitrile. Analytical columns were pulled and packed in-house using fused silica with an I.D. of 100 microns packed with Dr. Maisch 1.8 micron C18 stationary phase to a length of ∼25 cm. Peptides were directly injected onto the analytical column using a gradient (2–45% B, followed by a high-B wash) of 80min. The mass spectrometer was operated in a data-dependent fashion using CID fragmentation for MS/MS spectra generation collected in the ion trap. The collected mass spectra were analyzed using Byonic (Protein Metrics) for peptide identification and protein inference. The search was performed against the Uniprot homo sapiens database, including isoforms. Cysteine modified with propionamide was set as a fixed modification in the search, with other typical modifications, e.g. oxidation of methionine, included as variable modifications. .Data were held to a 12 ppm mass tolerance for precursors and 0.4 Da for MS/MS fragments, allowing up to two missed cleavage sites. Data were validated using the standard reverse-decoy technique at a 1% false discovery rate. The peptide counts in experiment group and Rnase treatment group for each replicate in GM and K562 cells are listed in Table S3. The XIST binding proteins were first identified in each replicate with peptide counts >25 and fold enrichment (experiment group/Rnase treatment group) >1.5. Then the proteins that are overlapped between replicates were identified and further filtered using mean fold enrichment from two replicates>2.5. The highly-confident XIST binding proteins for CRISPRi screen were identified using peptide counts>25 and mean fold enrichment >10.
RNA immunoprecipitation (UV-RIP and formaldehyde-RIP)
The RNA immunoprecipitation protocol was performed as previously described (G Hendrickson et al., 2016) with a few modifications. For UV-RIP, 5 million cells per replicate were crosslinked by UV exposure at 0.3 J/cm2 (254nM UV-C). For formaldehyde-RIP (fRIP), 5 million cells per replicate were fixed in 0.1% formaldehyde for 10 min at RT and quenched by 0.125 M glycine for 10 min at RT. UV-crosslinked or formaldehyde-fixed cells were washed by cold PBS and lysed in lysis buffer (50 mM Tris, 150 mM Kcl, 0,1% SDS, 1% Triton-X, 5 mM EDTA, 0.5% sodium deoxycholate, 0.5 mM DTT, 1XPIC, 100 U/mL RNAseOUT inhibitor) and sonicated using a Covaris Ultrasonicator tube (140W, 10% duty cycle, 200cycles/burst). After 15 min max speed spin at 4°C, the supernatant was pre-cleared with Protein A/G beads for 30 min at 4°C and the beads were disposed. An aliquot of the supernatant was saved as the total input control. The leftover supernatant was incubated with 2 ug mouse IgG (Thermo Fisher) or 2 ug KAP1 (ab10483, Abcam) and rotated overnight at 4°C. 20ul Protein A/G beads were added and rotated for 2 h at 4°C and washed four times with native lysis buffer (25 mM Tris, 150 mM KCl, 5 mM EDTA, 0.5% NP-40, 1X PIC, 100 U/mL RNAseOUT inhibitor, 0.5 mM DTT). The IP beads and input samples were reverse-crosslinked in RCL buffer (2% N-lauroyl Sarcosine, 10 mM EDTA, 5 mM DTT in PBS without Mg and Ca) with Proteinase K and RNAseOUT inhibitor for 1h at 42°C and then another 1h at 50°C, shaking at 1000 rpm. The RNA was purified using Quick-RNA Miniprep Kit (Zymo Research). The RNA quality was checked by BioAnalyzer and the qRT-PCR was performed using Stratagene Brilliant II SYBR Green QRT-PCR Master Mix (Agilent). The primers for RIP qPCR were listed in Table S6.
RNA extraction, RT-qPCR, and pyrosequencing
RNA was extracted in TRIzol using Quick-RNA Miniprep Kit following the manufacturer’s protocol with on-column Dnase digestion (Zymo Research). RT-qPCR was performed using Stratagene Brilliant II SYBR Green QRT-PCR Master Mix (Agilent). To quantify the allelic bias, cDNA was generated by reverse transcription of extracted RNA using SuperScriptIII and the amplified for 5 cycles using targeted qPCR primers. Then the PCR product was purified by Zymo DNA Clean & Concentrator-5 and further amplified using biotinylated primers and sequenced on using Q24 Pyromark (Qiagen). All primers for qRT-PCR and pyrosequencing are listed in Table S6.
RNA-seq
RNA was extracted using Quick-RNA Miniprep Kit with on-column Dnase digestion (Zymo Research). At least 100ng RNA was used to prepare the RNA-seq library using TruSeq® Stranded mRNA Library Prep Kit (Cat# 20020594, Illumina) for each sample following the manufacturer’s instruction. The library was sequenced on an Illumina Hiseq 4000 to generate 2X75 paired-end reads or 2X150 paired-end reads.
Rescue experiment with TRIM28 mutants
Since TRIM28 KO B cells are lethal, we developed a strategy to first express TRIM28 mutants and then knock out the endogenous TRIM28. Lentiviral vectors that carry full length TRIM28 and RING domain or PB (PHD-Bromodomain) domain deletion mutants conjugated to eGFP fused with a 2A peptide were cloned using Gibson Assembly from the PCR product of the plasmid containing TRIM28 cDNA (Addgene #124960). The N-terminal 180 bp sequence of all TRIM28 variants was altered to distinguish with endogenous TRIM28 sequence with silent mutation (producing same amino acids with endogenous TRIM28). Lentivirus expressing TRIM28 mutants were made similarly as above. B cells were transduced with TRIM28 mutants expressing lentivirus for 5 days and GFPhi cells were sorted for endogenous TRIM28 knockout. Two sgRNAs (30 uM) targeting the sequence inside of the N-terminal 180 bp of endogenous TRIM28 were complexed with 3 ug Cas9 protein (IDT) for 10 min and then electroporated to GFPhi B cells (1700V, 20ms, 1 pulse) using Neon Transfection System. The knockout efficiency was measured by indel from sanger sequencing of genomic DNA PCR product of the first 180bp. The percentage of indels was calculated by ICE analysis (Synthego).
CUT&RUN
CUT&RUN for H3K27me3 and H3K9me3 were performed as previously described in (Skene and Henikoff, 2017). Briefly, 0.5 million cells per replicate were bound to 10 uL concanavalin A-coated beads (Cat# BP531, Bangs Laboratories) in Binding Buffer (20 mM HEPES, 10 mM KCl, 1 mM CaCl2 and 1 mM MnCl2). The beads were washed and resuspended in Dig-Wash Buffer (20 mM HEPES, 150 mM NaCl, 0.5 mM Spermidine, 0.05% Digitonin, and 1X PIC). The primary antibody of H3K27me3 (Ab6002, Abcam, 1ug) or H3K9me3 (Ab8898, Abcam, 1ug) was added to the bead slurry and rotated at 4°C overnight. The beads were washed by Dig-Wash Buffer and the secondary antibody (1:50 for rabbit anti-mouse IgG) was added and rotated at 4°C for 1 h, and washed by Dig-Wash buffer. The pA-Mnase fusion protein (a gift from Henikoff lab) was added at 700ng/mL and rotated at 4°C for 1 h. Next, the beads were washed by Dig-Wash buffer twice followed by Low-Salt Rinse Buffer (50 mM HEPES and 0.02% Digitonin). The targeted digestion was performed by incubating at 0°C in Incubation Buffer (3.5 mM HEPES, 10 mM CaCl2, and 0.02% Digitonin) for 30 min and stopped by 1X STOP buffer (20 mM EGTA, 0.02% Digitonin, 25 ug/mL RnaseA, 50 ug/mL Glycogen, and 170mM NaCl). The chromatin was released by incubating at 37°C for 30 min. The DNA extraction was performed by NucleoSpin Gel and PCR Clean-up kit (Cat# 740609.250, Machery-Nagel). The CUT&RUN library was prepared using NEBNext Ultra II DNA library prep kit for Illumina (NEB) following the manufacturer’s protocol, and finally sequenced on an Illumina Hiseq 4000 to generate 2X75 paired-end reads.
ChIP-seq
For H3K27ac ChIP-seq, 1 million cells per replicate were fixed in 1% formaldehyde for 10 min at RT and quenched by 0.125 M glycine for 10 min at RT. For Trim28 ChIP-seq, 1 million cells per replicate were first cross-linked with 2 mM DSG for 40 min at RT and then fixed in 1% formaldehyde for 10 min at RT and finally quenched by 0.125 M glycine for 10 min at RT. Fixed cells were first lysed in membrane lysis buffer (50 mM HEPES-KOH, 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, 0.25% TritonX-100) for 10 min at 4°C. Then pelleted nuclei were lysed in nuclear lysis buffer (10mM Tris, 200mM NaCl, 1mM EDTA, 0.5mM EGTA) for 10min at RT. The chromatin was resuspended in Covaris sonication buffer (10 mM Tris, 1 mM EDTA, 0.1% SDS) and sonicated in a Covaris Ultrasonicator milliTube (140W, 10% duty cycle, 200 cycles/burst) to get 150–700bp length. The sonicated lysate was incubated with 4 ug H3K27ac (ab4729, Abcam) or 5 ug KAP1 (ab10483, Abcam) overnight at 4°C. Antibody-bound chromatin was incubated with Protein-G Dnyabeads (10004D, Invitrogen) for 4 h at 4°C and washed five times with RIPA wash buffer (5 mM HEPES-KOH, 500 mM LiCl, 1 mM EDTA, 1% NP40, 0.7% Sodium Deoxycholate) and eluted in elution buffer (10 mM Tris, 10 mM EDTA, 1% SDS). The eluted sample was incubated at 65°C overnight to reverse the crosslink and treated with RNAse A for 1h at 37°C and Proteinase K for 1h at 55°C. DNA was extracted using ChIP DNA Clean & Concentrator kit (Zymo Research). The ChIP-seq library was prepared using NEBNext Ultra II DNA library prep kit for Illumina (NEB) following the manufacturer’s protocol, and finally sequenced on an Illumina Hiseq 4000 to generate 2X 75 paired-end reads.
CRISPRi screen for allelic expression of TLR7
We designed two sgRNAs per XIST co-factor gene with each targeting a different region in 200 bp after the transcriptional start site using CRISPOR. One sgRNA each was cloned into pMJ117 (human U6 vector) or pMJ179 (mouse U6 vector) which were digested with BstXI and BlpI using NEBuilder HiFi DNA Assembly Master Mix (NEB). Then the corresponding U6 promoter and sgRNA sequences were PCR amplified and cleaned up by ZR-96 DNA Clean & Concentrator-5 (Deep Well). The PCR fragments were assembled into the pU6-sgGFP-NT1 lentiviral vector which was digested with XbaI and XhoI, using NEBuilder HiFi DNA Assembly Master Mix. Finally, the individual colonies with 2X sgRNA plasmid were screened by colony PCR and confirmed by Sanger sequencing. To generate CRISPRi lentivirus, we plated 800K HEK293T cells per well for 6-well pate and for the next day we transfected 0.75 ug pMP.G, 0.25 ug psPAX2, and 1 ug sgRNA vector in 100 uL Opti-MEM using Lipofectamin 3000. After two days, the supernatant was collected and the virus was concentrated 1:10 using Lenti-X Concentrator (Clontech). 200K dCas9-KRAB expressed GM cell line were plated per well in 24 well plates and 40 uL concentrated virus was added the following day. Two days later, we select the Xist-cofactor sgRNA expressing cells by adding 2 ug/mL puromycin for 2–3 weeks. Selection media was refreshed every three days. RNA was extracted using Direct-zol-96 RNA Kit (Zymo Research). TLR7 RT-PCR was performed using Stratagene Brilliant II SYBR Green QRT-PCR Master Mix (50°C for 30 min, 95°C for 10 min, then 35 cycles of 95°C for 30 seconds, 60°C for 1 min, and 72°C for 30 seconds, finally 72°C for 5 min). The second PCR with Ad1_TLR7_F and Ad2_TLR7_R primers was performed using NEBNext Master Mix (2.5uL 1st PCR product with 500 nM primers in program: 98°C for 1 min, 20 cycles of 98°C for 30 seconds, 60°C for 30 seconds, and then 72°C for 40 seconds, finally 72°C for 5 min). The third PCR with indexed primers containing barcodes was performed using NEBNext Master Mix (2uL of 1:10 diluted 2nd PCR product with 1uM indexed primers in program: 98°C for 1 min, 10 cycles of 98°C for 30 seconds, 72°C for 30 seconds, and then 72°C for 40 seconds, finally 72°C for 5 min). Finally, all the samples were pooled and purified using Zymo DNA clean & concentrator. The library was size-selected to 250–300 bp. The gel slices were cut and the DNA was recovered by Zymo Gel DNA recovery kit (Zymo Research). The concentration of the library was quantified by KAPA Library Quantification Kit for Illumina (Roche) and sequenced on an Illumina Miseq 2X75 cycle. Reads were mapped directly to TLR7 reference Xa and Xi allele and counted at each allele for d-score analysis. The co-factors with d-scoreco-factor -d-scoreCtrl >0.1 were labeled as essential XIST co-factors for TLR7 XCI maintenance.
Human primary B cell editing with Cas9 RNP
Buffy coats from female healthy donors were obtained from Stanford Blood Center with consent forms. Peripheral blood mononuclear cells (PBMC) were isolated using Lymphoprep (Cat# 07811, STEMCELL Technologies) density-gradient centrifugation and cryopreserved and stored in −80C. B cells were purified from thawed PBMCs by negative selection using EasySep Human B Cell Enrichment Kit (Cat#19844, STEMCELL Technologies) following the manufacturer’s protocol. Isolated B cells were cultured in IMDM medium supplemented with 10% FBS and 55 mM beta-mercaptoethanol at 1X106 cell/mL and primed with CellXVivo Human B cell expander (1:250 dilution, R&D system), 50 ng/mL IL10 (Cat#200-10-2ug, PeproTech), 10 ng/mL IL15 (Cat#200-15-2ug, PeproTech), 50 ng/mL IL2 (Cat#200-02-10ug, PeproTech) for two days. For B cell electroporation, primed B cells were washed by PBS twice and resuspended in Neon Buffer T (Cat# MPK1025, Thermo Fisher Scientific). Two sgRNAs (30 uM) targeting both ends of XIST repeat A were complexed with 3 ug Cas9 protein (IDT) for 10 min at RT. Cas9 RNP was added to the resuspension to get the final cell density at 3 X10^7 cells/ml and the electroporation was performed (1700V, 20ms, 1pulse) in 10 uL Neon transfection system. Electroporated cells were immediately transferred to pre-warmed B cell expansion medium (RPMI 1640 +10% FBS +1XGlutaMAX (Thermo Fisher Scientific) + 50 ng/mL IL2 + CellXVivo Human B cell expander (1:250 dilution, R&D system)). After two days of culture, an aliquot of cells was saved to test editing efficiency. The left cells were activated in B cell expansion medium supplemented with 1 ug/mL TLR7 agonist R848 (Invivogen) and 5 ug/mL F’2 Fragment Goat Anti-Human IgM (Cat# C840J42, Jackson ImmunoResearch Laboratories) for 5 days and subjected to flow cytometry.
Flow cytometry
B cells were first incubated with Human TruStain FcXTM (Fc Block, BioLegend) in cell staining buffer (Biolegend, Cat# 420201) for 30 min on ice and then stained with viability dye (Thermo Fisher, Cat # R37610), CD27 (clone O323), IgD (clone IA6-2), IgG(clone G18–145), CD19 (clone HIB19), and CD11c (clone Bu15) for 20 min on ice. The stained cells were washed in 1X FACS buffer for twice and are ready for flow cytometry. All flow cytometry antibodies are from Biolegend or BD Biosciences. All flow cytometry was performed using LSRII and FACS data were analyzed by FlowJo.
Computational analysis for RNA-seq
RNA-seq reads were mapped to the human genome (hg19) using STAR with default parameters (--outFilterMultimapNmax 1 --alignEndsType EndToEnd --outSAMattributes NH HI NM MD). Mitochodria reads were removed by samtools and duplicated reads were removed by picard. Quantification of aligned reads at the gene level was performed by HTseq count with default parameters (--stranded=reverse –additional-attr=gene_name). Raw counts were used to identify differentially expressed genes (DEG) using DESeq2 with size factor (total reads) normalization and DEGs were identified if Benjamini & Hochberg adjusted p-value <0.05. For Gene Ontology (GO) term analysis in Figure S1H, PANTHER GO was used. For gene functional annotation in Figure 7A, DAVID gene-annotation enrichment, KEGG pathway, and GAD disease category were used with cut-off (Benjamini & Hochberg adjusted p-value < 0.05 and fold enrichment > 3). In Figure 7, GSEA analysis was performed using Genepattern GSEA module with public available female SLE PBMC and B cell subset RNA-seq data (Jenks et al., 2018; Tokuyama et al., 2018). The gene sets used in GSEA analysis (genes upregulated in sgXIST, XIST-dependent X-linked genes, and XIST-independent X-linked genes) were listed in Table S2.
For allelic-specific RNA-seq analysis, SNP sites for donor NA12878 were extracted from the dbSNP database. SNP sites that are shared between paternal and maternal alleles but different from the human genome were kept as the common NA12878 SNP. SNP sites that are different between paternal and maternal allele were replaced by ‘N’ to build up N-masked reference genome. To maximize the allelic read depth, two biological replicates were pooled for later analysis. RNA-seq reads were aligned to the N-masked reference genome by STAR. After alignment and removal of mitochondria reads, remaining reads were split to maternal-specific (Xa) or paternal-specific (Xi) allele using SNPsplit. Duplicated reads were removed by picard and quantification of allele-specific reads at the gene level was performed via htseq-count similar to the bulk analysis. Genes with allelic reads >10 were retained to calculate the d-score (readsXi/(readsXa+readsXi)-0.5). For genes with d-score <0, genes with d-scoresgXIST-d-scorectrl > 0.03 and binomial P value <0.05 are defined as XIST-dependent genes. The other genes are XIST-independent genes. X-linked immune gene set was curated from review (Libert et al., 2010). The hierarchical clustering was performed using seaborn clustermap with parameters (metric=euclidean, z_score=0). In Figure S1G, the variable escapee, constitutive escapee and the X-inactivated genes were defined from GTEx study (Tukiainen et al., 2017). In Figure S2C-D, the effect size of sex bias for variable escapees in different tissues and the category of variable escapee, constitutive escapee, and the X-inactivated genes were defined from GTEx study (Oliva et al., 2020). In Figure 6, genes with d-scoreTRIM28KO-d-scorectrl > 0.03 and d-scorectrl <0 are defined as TRIM28-dependent genes. The other genes with d-scorectrl <0 are defined as TRIM28-independent genes. Bedgraph files were generated from bam files using bedtools genomcov and then convert to bigwig files for IGV visualization.
Computational analysis for CUT&RUN and ChIP-seq
Reads were mapped to the human genome (hg19) using bowtie2 with default parameters (--very-sensitive). Mitochondrial reads were removed by samtools and duplicated reads were removed by picard. For allelic-specific analysis, to maximize the allelic read depth, two biological replicates were pooled for later analysis. Reads were aligned to the N-masked reference genome by bowtie2. After alignment and removal of mitochondria reads and duplicated reads, remaining reads were split to Xa or Xi allele using SNPsplit. For ChIP-seq of H3K27ac, peaks were called using MACS2 with parameters (--broad -f BAMPE –broad-cutoff 0.01) over input control. The total consensus peak coordinate across samples was obtained using bedtools merge and intersect. The peaks were assigned to the nearest genes using HOMER annotatePeaks. The peaks within TSS +/− 1kb range were defined as promoters and other distal H3K27ac peaks were defined as enhancers. Allelic reads overlapping on peaks were counted using bedtools intersect and peaks with the allelic reads >10 were retained to calculate the d-score (readsXi/(readsXa+readsXi)-0.5). For H3K27me3 and H3K9me3 CUT&RUN and TRIM28 ChIP-seq, reads were aligned and filtered as above. TRIM28 ChIP-seq in K562 cell line is downloaded from public ENCODE dataset ENCFF165QYW. The windows were binned as TSS adjacent regions (TSS +/− 1kb) and gene bodies (TSS+1kb to TES) for each gene. Total and allelic reads overlapping these regions were counted using bedtools and regions with the allelic reads >10 were retained to calculate the d-score. Average plot of TRIM28 bulk signal was generated by ngsplot. Bedgraph files were generated from bam files using bedtools genomcov and then convert to bigwig files for IGV visualization.
Feature comparison
ChIP-seq of H3K27ac (ENCFF180LKW), RNA Pol II (ENCFF368HBX), Ser5P RNA Pol II (ENCFF002UPS), and Ser2P RNA Pol II (ENCFF031RUV), MEDIP-seq (GSE56774)(Sundaram et al., 2014), and GRO-seq (GSE60454)(Core et al., 2014) from public available ENCODE and GEO datasets were used for feature comparison. ATAC-seq of GM12878 cells (Buenrostro et al., 2013) and our CUT&RUN of H3K9me3 and H3K27me3 in control GM cells were used for feature comparison. Reads were aligned to hg19 or N-masked hg19 genome by bowtie2 with mitochondria reads and duplicated reads removal as above. The windows were binned as TSS adjacent regions (TSS +/− 1kb) and gene bodies (TSS+1kb to TES) for each gene. Total and allelic reads overlapping these regions were counted using bedtools and regions with the allelic reads >10 were retained to calculate the d-score. GC content was calculated by bedtools nuc. CpG island, LINE, SINE, and LTR files were downloaded from UCSC Table browser (RepeatMasker) and their density at TSS adjacent regions or gene bodies were counted by bedtools. For linear regression analysis, counts of each feature across X-linked genes were scaled from 0 to 1 individually and analysis was performed by statsmodels linear regression with OLS model. For logistical regression analysis, train and test sets for features were split by sklearn train_test_split with test_size=0.25. Then analysis was performed using sklearn LogisticRegression. GRO-seq analysis was performed using HOMER. The promoter-proximal index was calculated as read densityTSS adjacent regions/read densitygene bodies for each X-linked gene.
Computational analysis for single-cell RNA-seq
Public single-cell RNA-seq data were downloaded from GEO database (GSE149689(Lee et al., 2020), GSE154567(Yao et al., 2021), GSE155673(Arunachalam et al., 2020), GSE142016(Mistry et al., 2019), GSE159117(Zhang et al., 2020), GSE149313, GSE145281(Yuen et al., 2020), GSE135710(Lu et al., 2020)) and ImmPort repository (SDY997(Arazi et al., 2019), SDY998(Zhang et al., 2019)). The single cell data preprocess is done by Scanpy version 1.4.6 (Wolf et al., 2018). The sex of the sample is validated by the expression of male-specific gene DDX3Y, and female-specific gene XIST. The genes that are detected in less than 3 cells were removed using sc.pp.filter_genes(). The cells with fewer than 200 detected genes and higher than 10% mitochondrial genes were removed. The data was normalized to 10000 reads per cell for total-count to correct the library size using sc.pp.normalize_total() with target_sum=1e4. The data was then log transformed using sc.pp.log1p(). The top 2000 highly-variable genes were identified using sc.pp.highly_variable_genes() with min_mean=0.0125, max_mean=3, min_disp=0.5. The total counts per cell and the percentage of mitochondrial genes expressed were regressed out using sc.pp.regress_out(). Each gene is scaled to unit variance using sc.pp.scale() with max_value=10. Principle component analysis was performed using top 2000 highly variable genes using sc.tl.pca(). Then the top 40 principle components were used to compute the neighborhood graph for dimensional reduction using sc.pp.neighbors() with n_neighbors=10, n_pcs=40, random_state=1. The Leiden graph-clustering method was applied to find Leiden clusters using sc.tl.leiden() with different resolution based on datasets. The B cell clusters were subset based on B cell marker genes MS4A1 and CD79A while excluding other immune cell types based on cell-specific marker genes such as CD3E, CD14, PPBP, CD4, CD8A, FCER1A, and CST3 etc. The clusters for B cell subsets were manually annotated as naive, conventional memory, atypical memory B cells based on known cell type markers (IGHD, IGHG1, CD27, ITGAX). The UMAP plots on these marker genes for B cell subsets were generated by sc.tl.umap() and sc.pl.umap(). The compact violin plot for more marker genes of these B cell subsets was generated by sc.pl.stacked_violin(). For XIST escape score analysis, the score is calculated by mean expression of XIST-dependent X-linked genes subtract mean expression of XIST-independent X-linked genes (gene sets from Table S2). The calculation is performed using sc.tl.score_genes() with XIST-independent genes as reference gene pool and XIST-dependent genes as target gene list (ctrl_size=50, n_bins=25). The p value of gene score between subsets was calculated by unpaired t-test using scipy stats ttest_ind() function. For trajectory analysis, a weighted nearest neighbor graph was first constructed t0 build a transition matrix by sc.pp.neighbors() with n_neighbors=20, method=‘gauss’. Then the naive B cell is assigned as a root cell. The diffusion map was computed using n_comps=15 by sc.tl.diffmap(). The Diffusion Pseudotime order was computed by sc.tl.dpt() with n_branchings=1, n_dcs=10 to identify branching point and differentiation endpoint. The diffusion pseudotime plot was generated by sc.pl.diffmap().
QUANTIFICATION AND STATISTICAL ANALYSES
Statistical analyses are performed by Python and R. Information on specific statistical tests are listed in the figure legends and/or method details. Data are represented by mean+/−SEM in bar plots as indicated in the figure legends. P values are listed on figures or figure legends.
Supplementary Material
A. RNA-seq analysis showing the XIST RNA expression across control, sgXIST and inhibitors group.
B. MA plot showing differentially-expressed genes (highlighted in red) between control and inhibitors group (left), or between control and sgXIST group (right). Numbers showing significantly upregulated genes (top) or downregulated genes (down).
C. Heatmap showing Pearson correlation of gene expression allelic bias (d-score) on chromosome X between replicates with the control group (Ctrl), XIST perturbation group (sgXIST) and inhibitors treatment group (inhibitors).
D. Bar plot showing the percentage of genes on chrX and autosomes between genes that are upregulated in inhibitors group and genes that are upregulated in XIST group compared to control group. P value was calculated using Fisher exact test.
E. Box plot showing the log2 counts of expression of genes across the chrX at Xi and Xa in Ctrl, sgXIST and inhibitors group. P value was calculated using paired t-test.
F. Box plot showing the d-score of XIST-dependent and XIST-independent genes in control group using allelic RNA-seq data. P value was calculated using Mann-Whitney test.
G. Bar plot showing the percentage of variable escapee, constitutive escapee and inactive genes between XIST-dependent and -independent genes. P value was calculated using Fisher exact test.
H. Gene Ontology analysis showing top GO terms enriched in genes upregulated in sgXIST group (left) or genes upregulated in inhibitors group (right).
I. qRT-PCR result of XIST RNA expression of B cells that have been transduced with sgXIST or non-targeting control and then overexpressed with or without anti-CRISPRI protein (ACRIIA4).
J. Box plot showing the distribution of d-score of allelic gene expression on the X chromosome in groups as in I. P value was calculated using paired t-test.
K. Cumulative distribution of d-score of allelic gene expression across the X chromosome in groups as in I. P value was calculated using Kolmogorov–Smirnov test.
Data are represented as mean +/− SEM in A and I.
A. Bar plot showing the relative gene expression of XIST from control and sgXIST group in female human B cell line (GM12878) and female human fibroblast cell line (IMR90) from qRT-PCR, or neuronal cells from Xist WT and brain-specific Xist KO mice from RNA-seq data.
B. Cumulative distribution of log2(sgXIST/control) fold between X chromosome and autosomes in three cell types as in A. P value was calculated using Kolmogorov–Smirnov test.
C. Bar plot showing the percentage of XIST-silenced X-linked genes in three cell types. Pie chart showing the number of escapee, variable escapee, X-inactivated genes, and unknown category defined from GTEx study among XIST-silenced X-linked genes in fibroblasts and neuronal cells.
D. Heatmap showing effect size of sex bias in different tissues of variable escapees among XIST-silenced genes in fibroblasts and neuronal cells.
E. Box plots showing the GC percentage, the number of CpG islands, and the distance to XIST between XIST-dependent (blue) and -independent (yellow) genes. P values were calculated via Mann-Whitney test.
F. Box plots showing the density of transposon element LINE, SINE and LTR between XIST-dependent (blue) and -independent (yellow) genes. P values were calculated by Mann-Whitney test.
Data are represented as mean +/− SEM in A and C.
A. Schematic view of XIST repeat A deletion by Cas9-RNP. Genotyping PCR around XIST repeat A showing the deletion of XIST repeat A at one allele in a clonal XIST repeat A deletion cell line (XIST KO, third lane).
B. qRT-PCR results showing the XIST expression in control and XIST KO cells using two primer sets targeting repeat A and exon3 respectively. Data are represented as mean +/− SEM.
C. Genome tracks of H3K4me1, H3K4me3 and H3K27ac of XIST gene in B cell line. Repeat A region is highlighted in orange.
D. Genome tracks of replicates of H3K27ac ChIP-seq and H3K27me3 CUT&RUN in control and XIST KO group at chromosome X.
E. H3K27me3 CUT & RUN track of XIST-dependent gene FIRRE between control and XIST KO group.
A. ChIRP-qPCR result showing the percentage of RNA retrieval of XIST and GAPDH in XIST pulled down portion compared to input RNA in B cell line. RNAse treatment prior to XIST pull-down is showed as a negative control.
B. Coomassie blue stained gel showing the proteins pulled down by XIST probes and negative control with no XIST probes in B cell line.
C. Heatmap showing the z score of peptide counts of XIST-associated proteins with or without Rnase treatment in B cell line.
D. Bar plot showing the TRIM28 formaldehyde RIP-qPCR on A repeat of XIST in B cell line, myeloid K562 cell line, and differentiated ESCs. Western blot showing the pull down efficiency of TRIM28 in three cell types.
Data are represented as mean +/− SEM in A and D.
A. qRT-PCR results showing the relative gene expression of specific XIST co-factors between non-targeting sgRNA perturbed cells (control) and XIST co-factors perturbed cells. Bar plots showing the mean value of RNA fold enrichment from two biological replicates.
B. Genotyping PCR showing clones of B cell line with the deletion of SPEN RRM2-4 deletion (left). The qRT-PCR result showing the relative expression of SPEN RRM2-4 in control group and SPEN RRM2-4 deletion mutant group (right).
C. Box plot showing the distribution of d-score of allelic expression of genes on the chrX between control, sgXIST and SPEN mutant group. P value was calculated by paired t-test.
D. Box plot showing the d-score of allelic expression of XIST-dependent genes and XIST-independent genes in control and SPEN mutant group. P value was calculated by paired t-test.
Data are represented as mean +/− SEM in A and B.
A. Western blot showing the protein level of TRIM28 and GAPDH (internal control) in control and two clonal sgTRIM28 B cell lines (#7 and #25) by Cas9-RNP mediated knocking out. Over 99% TRIM28 proteins are depleted in #25 B cell line and this cell line was used as TRIM28 KO B cells.
B. Box plot showing the log2 counts of expression of TRIM28-dependent genes at Xi and Xa in Ctrl, TRIM28 KO and sgXIST group. P value was calculated using paired t-test.
C. Genome tracks showing the replicates of H3K9me3 CUT&RUN and TRIM28 ChIP-seq at ZNF gene clusters in B cell line.
D. Box plot showing the allelic bias of TRIM28 binding at chrX and autosome chr1. P value was calculated via Mann-Whitney test.
E. Box plot showing the counts of TRIM28 binding at TSS +/− 1kb region of TRIM28-dependent (red) and TRIM28-independent genes (blue) between Xi and Xa. P value was calculated via Mann-Whitney test.
F. Genome tracks showing the H3K9me3 profiles of representative TRIM28-dependent and TRIM28-independent genes in B cell line.
G. Box plots showing the occupancy density of H3K27ac, H3K9me3 and H3K27me3 at TSS +/− 1kb region of TRIM28-dependent (red) and TRIM28-independent genes (blue). P values were calculated via Mann-Whitney test.
H. Genome tracks showing the GRO-seq profile of nascent RNA at representative genes TLR7 (TRIM28-dependent) and SASH3 (TRIM28-independent) in B cell line.
I. Bar plot showing the percentage of indels introduced after endogenous TRIM28 knockout via CRISPR-Cas9 editing in B cells that are overexpressed with TRIM28 full length, RING domain deletion, and PB domain deletion mutants (left). Schematic view of domains in endogenous TRIM28 and exogenous TRIM28 mutants.
J. Bar plot showing the relative expression of SETDB1 in control and sgSETDB1 group using qRT-PCR.
K. Bar plot showing the pyrosequencing result of three TRIM28-dependent genes (TLR7, CXORF21, CXORF38) in control B cells, TRIM28 KO cells, B cells that are first expressed with different TRIM28 mutants and then deplete endogenous TRIM28 with CRISPR-Cas9, B cells with SETDB1 perturbation.
L. Box plot showing the SPEN and TRIM28 shRNA enrichment in GFPhi cells compared to input cells using a published shRNA screen data in differentiated ES cells.
M. Bar plot showing the relative expression of TRIM28 in K562 cells transduced with non-targeting control or sgTRIM28.
N. Genome tracks showing TRIM28 ChIP-seq in GM B cells and K562 myeloid cells on 4 TRIM28-dependent genes (left). Bar plot showing the relative expression fold enrichment of TRIM28-dependent genes in sgTRIM28 group compared to control group in K562 cells (right).
Data are represented as mean +/− SEM in I, J, K, M and N from two biological replicates. P value was calculated using unpaired t-test. * : p<0.05; **: p< 0.001.
A. GSEA analysis showing enrichment of gene set that is upregulated after XIST perturbation in PBMC or IgD-CD27-CD11c+ DN atypical B cells of female SLE patients compared to healthy donors. Normalized Enrichment Score (NES) and FDR q value are shown on the plot.
B. GSEA analysis showing enrichment of XIST-dependent X-linked genes and XIST-independent X-linked genes in IgD-CD27-CD11c+ DN atypical B cells of female SLE patients compared to healthy donors.
C. UMAP plots showing representative markers of naive (IgD+), conventional memory (IgG1+CD27+), and atypical memory (IgG1+CD11c+) B cells.
D. Stacked violin plots showing more markers for these three B cell subsets (naive B cell: CD72, IgD; conventional memory B cell: IgG1, CD27; atypical memory B cell: IgG1, CD11c, T-BET, ZEB2).
E. Diffusion plots showing the pseudotime trajectory started from the naive cells with top10 percentile of XIST escape score. Left: diffusion pseudotime plot. Middle: subset annotated projected on diffusion map. Right: XIST escape score on diffusion map.
F. Gel showing XIST A repeat deletion by genotyping PCR on Cas9-edited human primary B cells from two different female donors. The band highlighted on the gel showed A repeat deletion by sanger sequencing.
G. qRT-PCR result showing XIST RNA expression after XIST repeat A deletion.
H. Gating strategy of viable CD19+ B cells.
I. qRT-PCR result showing TLR7 RNA expression after XIST repeat A deletion. J. Flow cytometry analysis of IgD-CD27-CD11c+ and IgD-CD27+CD11c+ atypical memory B cells in AAVS1 and XIST KO group.
K. Box plot showing the percentage of these B cell subsets among total B cells after XIST repeat A deletion. The data are pooled from 8 healthy donors with 2–3 replicates from each donor. P value was calculated by paired t-test.
L. qRT-PCR result showing Tbx21 and ZEB2 relative RNA expression after the loss of XIST. Data are represented as mean +/− SEM in G, I and L.
DESeq analysis for differentially expressed genes between control, sgXIST, and inhibitors group. Related to Figure 1 and Figure S1
List of guide RNA sequences. Related to STAR Methods
List of primer sequences. Related to STAR Methods
KEY RESOURCES TABLE
| REAGENT or | SOURCE | IDENTIFIER |
|---|---|---|
| Antibodies | ||
| Rabbit polyclonal anti-KAP1 (TRIM28) |
Abcam | Cat# ab10483, RRID:AB_297222 |
| Mouse monoclonal anti-H3K27me3 |
Abcam | Cat# ab6002, RRID:AB_305237 |
| Rabbit polyclonal anti-H3K9me3 |
Abcam | Cat# ab8898, RRID:AB_306848 |
| Rabbit polyclonal anti-H3K27ac |
Abcam | Cat# ab4729, RRID:AB_2118291 |
| Brilliant Violet 605™ anti-human CD27 Antibody | BioLegend | Cat# 302829, RRID:AB_11204431 |
| Brilliant Violet 785™ anti-human IgD Antibody | BioLegend | Cat# 348241, RRID:AB_2629808 |
| APC/Cyanine7 anti-human CD19 Antibody | BioLegend | Cat# 302217, RRID:AB_314247 |
| PE anti-human CD11c Antibody | BioLegend | Cat# 337205, RRID:AB_1236439 |
| PE-Cy™7 Mouse Anti-Human IgG | BD Biosciences | Cat# 561298, RRID:AB_10611712 |
| AffiniPure F(ab’)2 Fragment Goat Anti-Human IgM |
Jackson ImmunoResear ch Labs | Cat# 109–006-129, RRID:AB_2337553 |
| Bacterial and Virus Strains | ||
| 10-beta Competent E.coli (High Efficiency) | NEB | Cat# C3019I |
| Biological Samples |
||
| Human blood buffy coat | Stanford Blood Center | https://stanfordbloodcenter.org/research-labs/research-products-and-services/blood-products/ |
| Chemicals, Peptices, and Recombinant Proteins | ||
| CellXVivo Human B cell expanser | R&D Systems | Cat# CDK005 |
| Recombinant Human IL-10 | PeproTech | Cat# 200-10-2ug |
| Recombinant Human IL-15 | PeproTech | Cat# 200-15-2ug |
| Recombinant Human IL-2 | PeproTech | Cat# 200-02-10ug |
| R848 | Invivogen | Cat# tlrl-r848 |
| 5-azacytidine | Sigma-Aldrich | Cat# A2385-250MG |
| EPZ-6438 | Xcessbio | Cat# M60122-2 |
| Critical Commercial Assays | ||
| Direct-zol RNA Miniprep Kit | Zymo Research | Cat# R2050 |
| NucleoSpin Gel and PCR Clean-up kit | Macherey- Nagel |
Cat# 740609.250 |
| ChIP DNA Clean & Concentrator kit |
Zymo Research | Cat# D5205 |
| Direct-zol-96 RNA Kit | Zymo Research | Cat# R2054 |
| ZR-96 DNA Clean & Concentrator-5 (Deep Well) | Zymo Research | Cat# D4024 |
| Direct-zol DNA/RNA Miniprep Kit | Zymo Research | Cat# R2080 |
| Quick-DNA Miniprep Kit | Zymo Research | Cat# D3025 |
| EasySep Human B Cell Enrichment Kit | stemcell technologies |
Cat# 19554 |
| KAPA Library Quantification Kit | Kapa Biosystems |
Cat# KK4854 |
| TruSeq® Stranded mRNA Library Prep Kit | Illumina | Cat#20020594 |
| NEBNext Ultra II DNA library prep kit for Illumina | NEB | Cat#E7645S |
| NEBuilder® HiFi DNA Assembly Master Mix | NEB | Cat# E2621L |
| Deposited Data | ||
| Raw and analyzed sequencing data | This paper | GEO GSE164596, Table S1 |
| Raw and processed mass spectrometry data | This paper | PRIDE PXD023516, Table S3 |
| Experimental Mocels: Cell Lines | ||
| HEK293T | ATCC | Cat#CRL-3216 |
| GM12878 dCas9-KRAB- BFP-2A-Blast |
This paper | N/A |
| IMR90 dCas9- KRAB-BFP-2A- Blast |
This paper | N/A |
| K562 dCas9- KRAB-BFP-2A- Blast |
This paper | N/A |
| differentiated mESC |
This paper | N/A |
| Oligonucleotides | ||
| sgRNA sequences |
This paper | Table S5 |
| Primers for qPCR, genotyping, pyrosequencing, and CRISPRi screen |
This paper | Table S6 |
| Recombinant DNA | ||
| pMJ117 | Addgene | Cat#85997 |
| pMJ179 | Addgene | Cat#85996 |
| pU6-sgGFP- NT1 |
Addgene | Cat#46914 |
| Fuw-AcrIIA4- P2A-GFP |
Addgene | Cat#108247 |
| Flag-Trim28 | Addgene | Cat#124960 |
| Lenti-Trim28_full length |
This paper | N/A |
| Lenti- Trim28 deltaRI NG |
This paper | N/A |
| Lenti- Trim28 deltaPB |
This paper | N/A |
| Software and Algorithms | ||
| Bowtie2 | (Langmead and Salzberg, 2012) | http://bowtie-bio.sourceforge.net/bowtie2/index.shtml |
| STAR | (Dobin and Gingeras, 2015) | https://github.com/alexdobin/STAR |
| Samtools | (Li et al., 2009) | http://www.htslib.org/ |
| BEDTools | (Quinlan and Hall, 2010) | https://bedtools.readthedocs.io/en/latest/ |
| SNPsplit | (Krueger and Andrews, 2016) | https://www.bioinformatics.babraham.ac.uk/projects/SNPsplit/ |
| Htseq-count | (Anders et al., 2015) | https://htseq.readthedocs.io/en/release_0.111/count.html |
| MACS2 | (Zhang et al., 2008) | https://github.com/macs3-project/MACS |
| DESeq2 | (Love et al., 2014) | http://bioconductor.org/packages/release/bioc/html/DESeq2.html |
| Genepattern GSEA module | (Reich et al., 2006) | https://cloud.genepattern.org/gp/pages/login.jsf |
| Scanpy | (Wolf et al., 2018) | https://scanpy.readthedocs.io/en/stable/ |
| Homer | (Heinz et al., 2010) | http://homer.ucsd.edu/homer/ngs/ |
| FlowJo | TreeStar | https://www.flowjo.com/ |
| ICE | Synthego | https://ice.synthego.com/#/ |
HIGHLIGHTS.
XIST prevents escape of genes with DNA hypomethylated promoters in B cells
XIST maintains X-inactivation through continuous deacetylation of H3K27ac
XIST ChIRP-MS and CRISPRi screen reveal a B cell-specific XIST cofactor TRIM28
XIST loss and TLR7 stimulation promotes CD11c+ atypical B cell formation
ACKNOWLEDGEMENT
We thank CK. Chen, K.L. Huang, N. Pacalin, A. Carter, J. Xu, members of the Chang lab and E. Heard for discussion. We also thank R. Leib at Mass Spectrometry core, PAN facility, FACS facility and SFGF sequencing core at Stanford University. Supported by Scleroderma Research Foundation, Pershing Square Foundation, NIH U54-CA260517, NIH RM1-HG007735 (to H.Y.C.), and a Stanford Dean’s Fellowship (to B.Y.). A.T.S. was supported by the NIH (K08CA230188 and U54260517), a Career Award for Medical Scientists from the Burroughs Wellcome Fund, Fast Grants from Emergent Ventures, and Stanford’s Innovative Medicine Accelerator and Stanford ChEM-H. H.Y.C. is an Investigator of the Howard Hughes Medical Institute.
Footnotes
DECLARATION OF INTERESTS
H.Y.C. is a co-founder of Accent Therapeutics, Boundless Bio, and an advisor to 10x Genomics, Arsenal Biosciences, and Spring Discovery. A.T.S. is a founder of Immunai and Cartography Biosciences and receives research funding from Arsenal Biosciences and 10x Genomics.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- Adrianse RL, Smith K, Gatbonton-Schwager T, Sripathy SP, Lao U, Foss EJ, Boers RG, Boers JB, Gribnau J, and Bedalov A (2018). Perturbed maintenance of transcriptional repression on the inactive X-chromosome in the mouse brain after Xist deletion. Epigenetics Chromatin 11, 50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Almeida M, Bowness JS, and Brockdorff N (2020). The many faces of Polycomb regulation by RNA. Curr. Opin. Genet. Dev. 61, 53–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Anders S, Pyl PT, and Huber W (2015). HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166–169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Arazi A., Rao DA., Berthier CC., Davidson A., Liu Y., Hoover PJ., Chicoin A., Eisenhaur TM., Jonsso AH., L S., et al. (2019). The immune cell landscape in kidneys of patients with lupus nephritis. Nat. Immunol. 20, 902–914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Arunachalam PS, Wimmers F, Mok CKP, Perera RAPM, Scott M, Hagan T, Sigal N, Feng Y, Bristow L, Tak-Yin Tsang O, et al. (2020). Systems biological assessment of immunity to mild versus severe COVID-19 infection in humans. Science 369, 1210–1220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bhatnagar S, Zhu X, Ou J, Lin L, Chamberlain L, Zhu LJ, Wajapeyee N, and Green MR (2014). Genetic and pharmacological reactivation of the mammalian inactive X chromosome. Proc. Natl. Acad. Sci. U. S. A. 111, 12591–12598. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brown CJ, and Willard HF (1994). The human X-inactivation centre is not required for maintenance of X-chromosome inactivation. Nature 368, 154–156. [DOI] [PubMed] [Google Scholar]
- Buenrostro JD, Giresi PG, Zaba LC, Chang HY, and Greenleaf WJ (2013). Transposition of native chromatin for multimodal regulatory analysis and personal epigenomics. Nat. Methods 10, 1213–1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cancro MP (2020). Age-Associated B Cells. Annu. Rev. Immunol. 38, 315–340. [DOI] [PubMed] [Google Scholar]
- Carrel L, and Brown CJ (2017). When the Lyon(ized chromosome) roars: ongoing expression from an inactive X chromosome. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Carter AC, Xu J, Nakamoto MY, Wei Y, Zarnegar BJ, Shi Q, Broughton JP, Ransom RC, Salhotra A, Nagaraja SD, et al. (2020). Spen links RNA-mediated endogenous retrovirus silencing and X chromosome inactivation. ELife 9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Celhar T, Magalhães R, and Fairhurst A-M (2012). TLR7 and TLR9 in SLE: when sensing self goes wrong. Immunol. Res. 53, 58–77. [DOI] [PubMed] [Google Scholar]
- Christensen SR, Shupe J, Nickerson K, Kashgarian M, Flavell RA, and Shlomchik MJ (2006). Toll-like receptor 7 and TLR9 dictate autoantibody specificity and have opposing inflammatory and regulatory roles in a murine model of lupus. Immunity 25, 417–428. [DOI] [PubMed] [Google Scholar]
- Chu C, Zhang QC, da Rocha ST, Flynn RA, Bharadwaj M, Calabrese JM, Magnuson T, Heard E, and Chang HY (2015). Systematic discovery of Xist RNA binding proteins. Cell 161, 404– 416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Core LJ, Martins AL, Danko CG, Waters CT, Siepel A, and Lis JT (2014). Analysis of nascent RNA identifies a unified architecture of initiation regions at mammalian promoters and enhancers. Nat. Genet. 46, 1311–1320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Csankovszki G, Panning B, Bates B, Pehrson JR, and Jaenisch R (1999). Conditional deletion of Xist disrupts histone macroH2A localization but not maintenance of X inactivation. Nat. Genet. 22, 323–324. [DOI] [PubMed] [Google Scholar]
- Csankovszki G, Nagy A, and Jaenisch R (2001). Synergism of Xist RNA, DNA methylation, and histone hypoacetylation in maintaining X chromosome inactivation. J. Cell Biol. 153, 773–784. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Czerwińska P, Mazurek S, and Wiznerowicz M (2017). The complexity of TRIM28 contribution to cancer. J. Biomed. Sci. 24, 63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dobin A, and Gingeras TR (2015). Mapping RNA-seq Reads with STAR. Curr. Protoc. Bioinforma. 51, 11.14.1–11.14.19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dossin F., Pinheiro I., Żylicz JJ., Roensch J., Collombet S., Le Saux A., Chelmicki T., Attia M., Kapoor V., Zhan Y., et al. (2020). SPEN integrates transcriptional and epigenetic control of X-inactivation. Nature 578, 455–460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- ENCODE Project Consortium (2012). An integrated encyclopedia of DNA elements in the human genome. Nature 489, 57–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- G Hendrickson D, Kelley DR, Tenen D, Bernstein B, and Rinn JL (2016). Widespread RNA binding by chromatin-associated proteins. Genome Biol. 17, 28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gartler SM, and Riggs AD (1983). Mammalian X-chromosome inactivation. Annu. Rev. Genet. 17, 155–190. [DOI] [PubMed] [Google Scholar]
- Hao Y, O’Neill P, Naradikian MS, Scholz JL, and Cancro MP (2011). A B-cell subset uniquely responsive to innate stimuli accumulates in aged mice. Blood 118, 1294–1304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hasegawa Y, Brockdorff N, Kawano S, Tsutui K, Tsutui K, and Nakagawa S (2010). The matrix protein hnRNP U is required for chromosomal localization of Xist RNA. Dev. Cell 19, 469–476. [DOI] [PubMed] [Google Scholar]
- Heinz LX, Lee J, Kapoor U, Kartnig F, Sedlyarov V, Papakostas K, César-Razquin A, Essletzbichler P, Goldmann U, Stefanovic A, et al. (2020). TASL is the SLC15A4-associated adaptor for IRF5 activation by TLR7–9. Nature 581, 316–322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, Cheng JX, Murre C, Singh H, and Glass CK (2010). Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hoki Y, Kimura N, Kanbayashi M, Amakawa Y, Ohhata T, Sasaki H, and Sado T (2009). A proximal conserved repeat in the Xist gene is essential as a genomic element for X-inactivation in mouse. Development 136, 139–146. [DOI] [PubMed] [Google Scholar]
- Iyengar S, and Farnham PJ (2011). KAP1 Protein: An Enigmatic Master Regulator of the Genome. J. Biol. Chem. 286, 26267–26276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jenks SA, Cashman KS, Zumaquero E, Marigorta UM, Patel AV, Wang X, Tomar D, Woodruff MC, Simon Z, Bugrovsky R, et al. (2018). Distinct Effector B Cells Induced by Unregulated Toll-like Receptor 7 Contribute to Pathogenic Responses in Systemic Lupus Erythematosus. Immunity 49, 725–739.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Karnell JL, Kumar V, Wang J, Wang S, Voynova E, and Ettinger R (2017). Role of CD11c+ T-bet+ B cells in human health and disease. Cell. Immunol. 321, 40–45. [DOI] [PubMed] [Google Scholar]
- Klein SL, and Flanagan KL (2016). Sex differences in immune responses. Nat. Rev. Immunol. 16, 626–638. [DOI] [PubMed] [Google Scholar]
- Krueger F, and Andrews SR (2016). SNPsplit: Allele-specific splitting of alignments between genomes with known SNP genotypes. F1000Research 5, 1479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee JS., Park S., Jeong HW., Ahn JY., Choi SJ., Lee H., Choi B., Nam SK., Sa M., Kwon J-S., et al. (2020). Immunophenotyping of COVID-19 and influenza highlights the role of type I interferons in development of severe COVID-19. Sci. Immunol. 5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, and 1000 Genome Project Data Processing Subgroup (2009). The Sequence Alignment/Map format and SAMtools. Bioinforma. Oxf. Engl. 25, 2078–2079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Libert C, Dejager L, and Pinheiro I (2010). The X chromosome in immune functions: when a chromosome makes the difference. Nat. Rev. Immunol. 10, 594–604. [DOI] [PubMed] [Google Scholar]
- Loda A, and Heard E (2019). Xist RNA in action: Past, present, and future. PLOS Genet. 15, e1008333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Love MI, Huber W, and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lu Y, Zhao Q, Liao J-Y, Song E, Xia Q, Pan J, Li Y, Li J, Zhou B, Ye Y, et al. (2020). Complement Signals Determine Opposite Effects of B Cells in Chemotherapy-Induced Immunity. Cell 180, 1081–1097.e24. [DOI] [PubMed] [Google Scholar]
- Lu Z, Zhang QC, Lee B, Flynn RA, Smith MA, Robinson JT, Davidovich C, Gooding AR, Goodrich KJ, Mattick JS, et al. (2016). RNA Duplex Map in Living Cells Reveals Higher-Order Transcriptome Structure. Cell 165, 1267–1279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lu Z, Carter AC, and Chang HY (2017). Mechanistic insights in X-chromosome inactivation. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lyon MF (1962). Sex Chromatin and Gene Action in the Mammalian X-Chromosome. Am. J. Hum. Genet. 14, 135–148. [PMC free article] [PubMed] [Google Scholar]
- Ma X, Yang T, Luo Y, Wu L, Jiang Y, Song Z, Pan T, Liu B, Liu G, Liu J, et al. (2019). TRIM28 promotes HIV-1 latency by SUMOylating CDK9 and inhibiting P-TEFb. ELife 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McHugh CA, Chen C-K, Chow A, Surka CF, Tran C, McDonel P, Pandya-Jones A, Blanco M, Burghard C, Moradian A, et al. (2015). The Xist lncRNA interacts directly with SHARP to silence transcription through HDAC3. Nature 521, 232–236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Minajigi A, Froberg J, Wei C, Sunwoo H, Kesner B, Colognori D, Lessing D, Payer B, Boukhali M, Haas W, et al. (2015). A comprehensive Xist interactome reveals cohesin repulsion and an RNA-directed chromosome conformation. Science 349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mistry P, Nakabo S, O’Neil L, Goel RR, Jiang K, Carmona-Rivera C, Gupta S, Chan DW, Carlucci PM, Wang X, et al. (2019). Transcriptomic, epigenetic, and functional analyses implicate neutrophil diversity in the pathogenesis of systemic lupus erythematosus. Proc. Natl. Acad. Sci. U. S. A. 116, 25222–25228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Moindrot B., Cerase A., Coker H., Masui O., Grijzenhout A., Pintacuda G., Schermelleh L., Nesterova TB., and Brockdorff N. (2015). A Pooled shRNA Screen Identifies Rbm15, Spen, and Wtap as Factors Required for Xist RNA-Mediated Silencing. Cell Rep. 12, 562–572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Monfort A, Di Minin G, Postlmayr A, Freimann R, Arieti F, Thore S, and Wutz A (2015). Identification of Spen as a Crucial Factor for Xist Function through Forward Genetic Screening in Haploid Embryonic Stem Cells. Cell Rep. 12, 554–561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Oliva M, Muñoz-Aguirre M, Kim-Hellmuth S, Wucher V, Gewirtz ADH, Cotter DJ, Parsana P, Kasela S, Balliu B, Viñuela A, et al. (2020). The impact of sex on gene expression across human tissues. Science 369. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pandya-Jones A, Markaki Y, Serizay J, Chitiashvilli T, Mancia W, Damianov A, Chronis C, Papp B, Chen C-K, McKee R, et al. (2020). An Xist-dependent protein assembly mediates Xist localization and gene silencing. BioRxiv 2020.03.09.979369. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Penny GD, Kay GF, Sheardown SA, Rastan S, and Brockdorff N (1996). Requirement for Xist in X chromosome inactivation. Nature 379, 131–137. [DOI] [PubMed] [Google Scholar]
- Pyfrom S, Paneru B, Knoxx JJ, Cancro MP, Posso S, Buckner JH, and Anguera MC (2020). The dynamic epigenetic regulation of the inactive X chromosome in healthy human B cells is dysregulated in lupus patients. BioRxiv 2020.11.25.399022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Quinlan AR, and Hall IM (2010). BEDTools: a flexible suite of utilities for comparing genomic features. Bioinforma. Oxf. Engl. 26, 841–842. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reich M, Liefeld T, Gould J, Lerner J, Tamayo P, and Mesirov JP (2006). GenePattern 2.0. Nat. Genet. 38, 500–501. [DOI] [PubMed] [Google Scholar]
- Rubin AJ, Parker KR, Satpathy AT, Qi Y, Wu B, Ong AJ, Mumbach MR, Ji AL, Kim DS, Cho SW, et al. (2019). Coupled Single-Cell CRISPR Screening and Epigenomic Profiling Reveals Causal Gene Regulatory Networks. Cell 176, 361–376.e17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rubtsov AV, Rubtsova K, Fischer A, Meehan RT, Gillis JZ, Kappler JW, and Marrack P (2011). Toll-like receptor 7 (TLR7)–driven accumulation of a novel CD11c+ B-cell population is important for the development of autoimmunity. Blood 118, 1305–1315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rubtsova K, Rubtsov AV, Cancro MP, and Marrack P (2015). Age-Associated B Cells: A T-bet–Dependent Effector with Roles in Protective and Pathogenic Immunity. J. Immunol. 195, 1933–1937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schultz DC, Friedman JR, and Rauscher FJ (2001). Targeting histone deacetylase complexes via KRAB-zinc finger proteins: the PHD and bromodomains of KAP-1 form a cooperative unit that recruits a novel isoform of the Mi-2alpha subunit of NuRD. Genes Dev. 15, 428–443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schultz DC, Ayyanathan K, Negorev D, Maul GG, and Rauscher FJ (2002). SETDB1: a novel KAP-1-associated histone H3, lysine 9-specific methyltransferase that contributes to HP1-mediated silencing of euchromatic genes by KRAB zinc-finger proteins. Genes Dev. 16, 919–932. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Skene PJ, and Henikoff S (2017). An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. ELife 6, e21856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Souyris M, Cenac C, Azar P, Daviaud D, Canivet A, Grunenwald S, Pienkowski C, Chaumeil J, Mejía JE, and Guéry J-C (2018). TLR7 escapes X chromosome inactivation in immune cells. Sci. Immunol. 3. [DOI] [PubMed] [Google Scholar]
- Subramanian S., Tus K., Li Q-Z., Wang A., Tian X-H., Zhou J., Liang C., Bartov G., McDaniel LD., Zhou XJ., et al. (2006). A Tlr7 translocation accelerates systemic autoimmunity in murine lupus. Proc. Natl. Acad. Sci. U. S. A. 103, 9970–9975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sundaram V, Cheng Y, Ma Z, Li D, Xing X, Edge P, Snyder MP, and Wang T (2014). Widespread contribution of transposable elements to the innovation of gene regulatory networks. Genome Res. 24, 1963–1976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sunwoo H, Colognori D, Froberg JE, Jeon Y, and Lee JT (2017). Repeat E anchors Xist RNA to the inactive X chromosomal compartment through CDKN1A-interacting protein (CIZ1). Proc. Natl. Acad. Sci. U. S. A. 114, 10654–10659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Syrett CM, Paneru B, Sandoval-Heglund D, Wang J, Banerjee S, Sindhava V, Behrens EM, Atchison M, and Anguera MC (2019). Altered X-chromosome inactivation in T cells may promote sex-biased autoimmune diseases. JCI Insight 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Takahashi T, Ellingson MK, Wong P, Israelow B, Lucas C, Klein J, Silva J, Mao T, Oh JE, Tokuyama M, et al. (2020). Sex differences in immune responses that underlie COVID-19 disease outcomes. Nature. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tokuyama M, Kong Y, Song E, Jayewickreme T, Kang I, and Iwasaki A (2018). ERVmap analysis reveals genome-wide transcription of human endogenous retroviruses. Proc. Natl. Acad. Sci. U. S. A. 115, 12565–12572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tukiainen T, Villani A-C, Yen A, Rivas MA, Marshall JL, Satija R, Aguirre M, Gauthier L, Fleharty M, Kirby A, et al. (2017). Landscape of X chromosome inactivation across human tissues. Nature 550, 244–248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang J, Syrett CM, Kramer MC, Basu A, Atchison ML, and Anguera MC (2016). Unusual maintenance of X chromosome inactivation predisposes female lymphocytes for increased expression from the inactive X. Proc. Natl. Acad. Sci. U. S. A. 113, E2029–2038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Williamson EJ, Walker AJ, Bhaskaran K, Bacon S, Bates C, Morton CE, Curtis HJ, Mehrkar A, Evans D, Inglesby P, et al. (2020). Factors associated with COVID-19-related death using OpenSAFELY. Nature 584, 430–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wolf FA, Angerer P, and Theis FJ (2018). SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19, 15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Woodruff MC, Ramonell RP, Nguyen DC, Cashman KS, Saini AS, Haddad NS, Ley AM, Kyu S, Howell JC, Ozturk T, et al. (2020). Extrafollicular B cell responses correlate with neutralizing antibodies and morbidity in COVID-19. Nat. Immunol. 21, 1506–1516. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yang L, Kirby JE, Sunwoo H, and Lee JT (2016). Female mice lacking Xist RNA show partial dosage compensation and survive to term. Genes Dev. 30, 1747–1760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yang L, Yildirim E, Kirby JE, Press W, and Lee JT (2020). Widespread organ tolerance to Xist loss and X reactivation except under chronic stress in the gut. Proc. Natl. Acad. Sci. 117, 4262–4272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yao C., Bora SA., Parimon T., Zaman T., Friedman OA., Palatinus JA., Surapaneni NS., Matusov YP., Cerro Chiang G., Kassar AG., et al. (2021). Cell-Type-Specific Immune Dysregulation in Severely Ill COVID-19 Patients. Cell Rep. 34, 108590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yildirim E, Kirby JE, Brown DE, Mercier FE, Sadreyev RI, Scadden DT, and Lee JT (2013). Xist RNA is a potent suppressor of hematologic cancer in mice. Cell 152, 727–742. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yuen KC, Liu L-F, Gupta V, Madireddi S, Keerthivasan S, Li C, Rishipathak D, Williams P, Kadel EE, Koeppen H, et al. (2020). High systemic and tumor-associated IL-8 correlates with reduced clinical benefit of PD-L1 blockade. Nat. Med. 26, 693–698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang B, Zhang Y, Xiong L, Li Y, Zhang Y, Zhao J, Jiang H, Li C, Liu Y, Liu X, et al. (2020). CD127 imprints functional heterogeneity to diversify monocyte responses in human inflammatory diseases. BioRxiv 2020.11.10.376277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang F, Wei K, Slowikowski K, Fonseka CY, Rao DA, Kelly S, Goodman SM, Tabechian D, Hughes LB, Salomon-Escoto K, et al. (2019). Defining inflammatory cell states in rheumatoid arthritis joint synovial tissues by integrating single-cell transcriptomics and mass cytometry. Nat. Immunol. 20, 928–942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, et al. (2008). Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhao R, Chen X, Ma W, Zhang J, Guo J, Zhong X, Yao J, Sun J, Rubinfien J, Zhou X, et al. (2020). A GPR174-CCL21 module imparts sexual dimorphism to humoral immunity. Nature 577, 416–420. [DOI] [PubMed] [Google Scholar]
- Zumaquero E, Stone SL, Scharer CD, Jenks SA, Nellore A, Mousseau B, Rosal-Vela A, Botta D, Bradley JE, Wojciechowski W, et al. (2019). IFNγ induces epigenetic programming of human T-bethi B cells and promotes TLR7/8 and IL-21 induced differentiation. ELife 8, e41641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Żylicz JJ, Bousard A, Žumer K, Dossin F, Mohammad E, da Rocha ST, Schwalb B, Syx L, Dingli F, Loew D, et al. (2019). The Implication of Early Chromatin Changes in X Chromosome Inactivation. Cell 176, 182–197.e23. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
A. RNA-seq analysis showing the XIST RNA expression across control, sgXIST and inhibitors group.
B. MA plot showing differentially-expressed genes (highlighted in red) between control and inhibitors group (left), or between control and sgXIST group (right). Numbers showing significantly upregulated genes (top) or downregulated genes (down).
C. Heatmap showing Pearson correlation of gene expression allelic bias (d-score) on chromosome X between replicates with the control group (Ctrl), XIST perturbation group (sgXIST) and inhibitors treatment group (inhibitors).
D. Bar plot showing the percentage of genes on chrX and autosomes between genes that are upregulated in inhibitors group and genes that are upregulated in XIST group compared to control group. P value was calculated using Fisher exact test.
E. Box plot showing the log2 counts of expression of genes across the chrX at Xi and Xa in Ctrl, sgXIST and inhibitors group. P value was calculated using paired t-test.
F. Box plot showing the d-score of XIST-dependent and XIST-independent genes in control group using allelic RNA-seq data. P value was calculated using Mann-Whitney test.
G. Bar plot showing the percentage of variable escapee, constitutive escapee and inactive genes between XIST-dependent and -independent genes. P value was calculated using Fisher exact test.
H. Gene Ontology analysis showing top GO terms enriched in genes upregulated in sgXIST group (left) or genes upregulated in inhibitors group (right).
I. qRT-PCR result of XIST RNA expression of B cells that have been transduced with sgXIST or non-targeting control and then overexpressed with or without anti-CRISPRI protein (ACRIIA4).
J. Box plot showing the distribution of d-score of allelic gene expression on the X chromosome in groups as in I. P value was calculated using paired t-test.
K. Cumulative distribution of d-score of allelic gene expression across the X chromosome in groups as in I. P value was calculated using Kolmogorov–Smirnov test.
Data are represented as mean +/− SEM in A and I.
A. Bar plot showing the relative gene expression of XIST from control and sgXIST group in female human B cell line (GM12878) and female human fibroblast cell line (IMR90) from qRT-PCR, or neuronal cells from Xist WT and brain-specific Xist KO mice from RNA-seq data.
B. Cumulative distribution of log2(sgXIST/control) fold between X chromosome and autosomes in three cell types as in A. P value was calculated using Kolmogorov–Smirnov test.
C. Bar plot showing the percentage of XIST-silenced X-linked genes in three cell types. Pie chart showing the number of escapee, variable escapee, X-inactivated genes, and unknown category defined from GTEx study among XIST-silenced X-linked genes in fibroblasts and neuronal cells.
D. Heatmap showing effect size of sex bias in different tissues of variable escapees among XIST-silenced genes in fibroblasts and neuronal cells.
E. Box plots showing the GC percentage, the number of CpG islands, and the distance to XIST between XIST-dependent (blue) and -independent (yellow) genes. P values were calculated via Mann-Whitney test.
F. Box plots showing the density of transposon element LINE, SINE and LTR between XIST-dependent (blue) and -independent (yellow) genes. P values were calculated by Mann-Whitney test.
Data are represented as mean +/− SEM in A and C.
A. Schematic view of XIST repeat A deletion by Cas9-RNP. Genotyping PCR around XIST repeat A showing the deletion of XIST repeat A at one allele in a clonal XIST repeat A deletion cell line (XIST KO, third lane).
B. qRT-PCR results showing the XIST expression in control and XIST KO cells using two primer sets targeting repeat A and exon3 respectively. Data are represented as mean +/− SEM.
C. Genome tracks of H3K4me1, H3K4me3 and H3K27ac of XIST gene in B cell line. Repeat A region is highlighted in orange.
D. Genome tracks of replicates of H3K27ac ChIP-seq and H3K27me3 CUT&RUN in control and XIST KO group at chromosome X.
E. H3K27me3 CUT & RUN track of XIST-dependent gene FIRRE between control and XIST KO group.
A. ChIRP-qPCR result showing the percentage of RNA retrieval of XIST and GAPDH in XIST pulled down portion compared to input RNA in B cell line. RNAse treatment prior to XIST pull-down is showed as a negative control.
B. Coomassie blue stained gel showing the proteins pulled down by XIST probes and negative control with no XIST probes in B cell line.
C. Heatmap showing the z score of peptide counts of XIST-associated proteins with or without Rnase treatment in B cell line.
D. Bar plot showing the TRIM28 formaldehyde RIP-qPCR on A repeat of XIST in B cell line, myeloid K562 cell line, and differentiated ESCs. Western blot showing the pull down efficiency of TRIM28 in three cell types.
Data are represented as mean +/− SEM in A and D.
A. qRT-PCR results showing the relative gene expression of specific XIST co-factors between non-targeting sgRNA perturbed cells (control) and XIST co-factors perturbed cells. Bar plots showing the mean value of RNA fold enrichment from two biological replicates.
B. Genotyping PCR showing clones of B cell line with the deletion of SPEN RRM2-4 deletion (left). The qRT-PCR result showing the relative expression of SPEN RRM2-4 in control group and SPEN RRM2-4 deletion mutant group (right).
C. Box plot showing the distribution of d-score of allelic expression of genes on the chrX between control, sgXIST and SPEN mutant group. P value was calculated by paired t-test.
D. Box plot showing the d-score of allelic expression of XIST-dependent genes and XIST-independent genes in control and SPEN mutant group. P value was calculated by paired t-test.
Data are represented as mean +/− SEM in A and B.
A. Western blot showing the protein level of TRIM28 and GAPDH (internal control) in control and two clonal sgTRIM28 B cell lines (#7 and #25) by Cas9-RNP mediated knocking out. Over 99% TRIM28 proteins are depleted in #25 B cell line and this cell line was used as TRIM28 KO B cells.
B. Box plot showing the log2 counts of expression of TRIM28-dependent genes at Xi and Xa in Ctrl, TRIM28 KO and sgXIST group. P value was calculated using paired t-test.
C. Genome tracks showing the replicates of H3K9me3 CUT&RUN and TRIM28 ChIP-seq at ZNF gene clusters in B cell line.
D. Box plot showing the allelic bias of TRIM28 binding at chrX and autosome chr1. P value was calculated via Mann-Whitney test.
E. Box plot showing the counts of TRIM28 binding at TSS +/− 1kb region of TRIM28-dependent (red) and TRIM28-independent genes (blue) between Xi and Xa. P value was calculated via Mann-Whitney test.
F. Genome tracks showing the H3K9me3 profiles of representative TRIM28-dependent and TRIM28-independent genes in B cell line.
G. Box plots showing the occupancy density of H3K27ac, H3K9me3 and H3K27me3 at TSS +/− 1kb region of TRIM28-dependent (red) and TRIM28-independent genes (blue). P values were calculated via Mann-Whitney test.
H. Genome tracks showing the GRO-seq profile of nascent RNA at representative genes TLR7 (TRIM28-dependent) and SASH3 (TRIM28-independent) in B cell line.
I. Bar plot showing the percentage of indels introduced after endogenous TRIM28 knockout via CRISPR-Cas9 editing in B cells that are overexpressed with TRIM28 full length, RING domain deletion, and PB domain deletion mutants (left). Schematic view of domains in endogenous TRIM28 and exogenous TRIM28 mutants.
J. Bar plot showing the relative expression of SETDB1 in control and sgSETDB1 group using qRT-PCR.
K. Bar plot showing the pyrosequencing result of three TRIM28-dependent genes (TLR7, CXORF21, CXORF38) in control B cells, TRIM28 KO cells, B cells that are first expressed with different TRIM28 mutants and then deplete endogenous TRIM28 with CRISPR-Cas9, B cells with SETDB1 perturbation.
L. Box plot showing the SPEN and TRIM28 shRNA enrichment in GFPhi cells compared to input cells using a published shRNA screen data in differentiated ES cells.
M. Bar plot showing the relative expression of TRIM28 in K562 cells transduced with non-targeting control or sgTRIM28.
N. Genome tracks showing TRIM28 ChIP-seq in GM B cells and K562 myeloid cells on 4 TRIM28-dependent genes (left). Bar plot showing the relative expression fold enrichment of TRIM28-dependent genes in sgTRIM28 group compared to control group in K562 cells (right).
Data are represented as mean +/− SEM in I, J, K, M and N from two biological replicates. P value was calculated using unpaired t-test. * : p<0.05; **: p< 0.001.
A. GSEA analysis showing enrichment of gene set that is upregulated after XIST perturbation in PBMC or IgD-CD27-CD11c+ DN atypical B cells of female SLE patients compared to healthy donors. Normalized Enrichment Score (NES) and FDR q value are shown on the plot.
B. GSEA analysis showing enrichment of XIST-dependent X-linked genes and XIST-independent X-linked genes in IgD-CD27-CD11c+ DN atypical B cells of female SLE patients compared to healthy donors.
C. UMAP plots showing representative markers of naive (IgD+), conventional memory (IgG1+CD27+), and atypical memory (IgG1+CD11c+) B cells.
D. Stacked violin plots showing more markers for these three B cell subsets (naive B cell: CD72, IgD; conventional memory B cell: IgG1, CD27; atypical memory B cell: IgG1, CD11c, T-BET, ZEB2).
E. Diffusion plots showing the pseudotime trajectory started from the naive cells with top10 percentile of XIST escape score. Left: diffusion pseudotime plot. Middle: subset annotated projected on diffusion map. Right: XIST escape score on diffusion map.
F. Gel showing XIST A repeat deletion by genotyping PCR on Cas9-edited human primary B cells from two different female donors. The band highlighted on the gel showed A repeat deletion by sanger sequencing.
G. qRT-PCR result showing XIST RNA expression after XIST repeat A deletion.
H. Gating strategy of viable CD19+ B cells.
I. qRT-PCR result showing TLR7 RNA expression after XIST repeat A deletion. J. Flow cytometry analysis of IgD-CD27-CD11c+ and IgD-CD27+CD11c+ atypical memory B cells in AAVS1 and XIST KO group.
K. Box plot showing the percentage of these B cell subsets among total B cells after XIST repeat A deletion. The data are pooled from 8 healthy donors with 2–3 replicates from each donor. P value was calculated by paired t-test.
L. qRT-PCR result showing Tbx21 and ZEB2 relative RNA expression after the loss of XIST. Data are represented as mean +/− SEM in G, I and L.
DESeq analysis for differentially expressed genes between control, sgXIST, and inhibitors group. Related to Figure 1 and Figure S1
List of guide RNA sequences. Related to STAR Methods
List of primer sequences. Related to STAR Methods
Data Availability Statement
The accession number for all the sequencing data in this paper is GEO: GSE164596. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD023516. Details of analysis are provided in STAR methods, and further questions should be directed to the Lead Contact.