Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 May 16.
Published in final edited form as: Mol Cell. 2024 May 16;84(10):1870–1885.e9. doi: 10.1016/j.molcel.2024.04.015

G-quadruplex folding in Xist RNA antagonizes PRC2 activity for step-wise regulation of X-chromosome inactivation

Yong Woo Lee 1, Uri Weissbein 1, Roy Blum 1, Jeannie T Lee 1,*
PMCID: PMC11505738  NIHMSID: NIHMS1990282  PMID: 38759625

SUMMARY

How Polycomb repressive complex 2 (PRC2) is regulated by RNA remains an unsolved problem. While PRC2 binds G-tracts with potential to form RNA G-quadruplexes (rG4), whether rG4’s fold extensively in vivo and whether PRC2 binds folded or unfolded rG4 are unknown. Using the X-inactivation model in mouse embryonic stem cells, here we identify multiple folded rG4’s in Xist RNA and demonstrate that PRC2 preferentially binds folded rG4’s. High-affinity rG4 binding inhibits PRC2’s histone methyltransferase activity and stabilizing rG4 in vivo antagonizes H3K27me3 enrichment on the inactive X-chromosome. Surprisingly, mutagenizing the rG4 does not affect PRC2 recruitment but promotes its release and catalytic activation on chromatin. H3K27me3 marks are misplaced, however, and gene silencing is compromised. Xist-PRC2 complexes become entrapped in the S1 chromosome compartment, precluding the required translocation into the S2 compartment. Thus, Xist rG4 folding controls PRC2 activity, H3K27me3 enrichment, and the step-wise regulation of chromosome-wide gene silencing.

eTOC Blurb

Lee et al. identify RNA G-quadruplexes (rG4) in Xist RNA and demonstrate that they are developmentally specific. rG4 folding controls PRC2 activity, H3K27me3 enrichment, and the step-wise regulation of X-linked gene silencing. When the rG4 sequences are mutated in Xist, PRC2 is hyperactivated and becomes spatially entrapped, preventing the progression of X-inactivation.

Graphical Abstract

graphic file with name nihms-1990282-f0008.jpg

INTRODUCTION

Polycomb repressive complex 2 (PRC2) is an epigenetic complex responsible for trimethylating histone H3 at lysine 27 (H3K27me3) 1. How PRC2 is targeted and how its activity is regulated have been much debated. Multiple studies have implicated RNA in the recruitment of PRC2 26. PRC2 is a robust RNA-binding complex that contacts RNA via several of its subunits, including the catalytic subunit, EZH2, and SUZ12 2,3,7,8. PRC2 has a strong preference for C-rich and G-rich patches in RNA, especially G-rich motifs with the potential to fold into 3D structures known as “RNA G-quadruplexes” (rG4) 912. rG4’s are formed by Hoogsteen base-pairing and organization of guanine tetrads into planar arrays of two, three, or more tetrads 1316. PRC2 interactions with such G-rich sequences regulate RNA POL-II pause-release for precision control of Polycomb target genes 9 and disrupting the PRC2-RNA interactions results in increased transcription elongation and hyperexpression of target genes. Interactions with G-rich RNA have been shown to inhibit the catalytic activity of PRC2 7,8 and/or cause eviction of PRC2 from chromatin 12.

Although its affinity for G4 motifs is known, whether PRC2 is regulated by the folded or unfolded G4 form is not known. The idea that RNA’s binding inhibits PRC2’s activity 812 may also seem contradictory to its role in PRC2 recruitment 2,8. However, early work hinted that PRC2 recruitment, chromatin loading, and H3K27-trimethylation can be parsed: Prior to X-chromosome inactivation (XCI), PRC2 makes contact with Xist RNA in mouse female embryonic stem (ES) cells, but does not contact chromatin or activate its histone methyltransferase (HMT) function until Tsix (Xist’s antisense regulator) is downregulated 2. PRC2’s HMT activity is augmented when RNA binds accessory factors such as JARID2 8. These data indicate that PRC2 recruitment and HMT activation can be uncoupled, though the mechanisms have yet to be elucidated.

Crucially missing is knowledge of structural dynamics at the RNA interface. RNAs are flexible macromolecules with the potential to adopt many conformations, but current models largely hold a static view of RNA-PRC2 relationships. Although current thinking favors PRC2’s affinity for rG4, PRC2 could instead interact with unstructured G-rich structures, especially given uncertainty in whether rG4’s stably fold in vivo 17. In vitro, rG4’s are extremely stable due to extensive hydrogen-bonding and base-stacking. However, one study has proposed that rG4-forming sequences may be globally unfolded in eukaryotes due to their toxicity 17. Nonetheless, in vitro rG4-seq and imaging studies in humans, animals, and plants have suggested existence of thousands of stable rG4’s 1823 and implicated rG4 dynamics in tumorigenesis 24, neurodegeneration 25, viral infection 26,27, RNA splicing 28,29, translation initiation 30, and POL-II pause-release during transcription elongation 9,31. A cryoEM view proposed that RNA induces dimerization and catalytic inhibition of PRC2 32, but left open the question of whether the G-rich RNA is in folded or unfolded form and how the interaction affects gene expression in vivo.

Here we focus on three major questions: (i) Do G-rich motifs fold into rG4 structures in vivo, (ii) does rG4 folding affect PRC2 function in vivo, and (iii) what are the consequences of rG4 loss? We begin by improving existing rG4 detection methods and establishing a pipeline for calling rG4. Using X-chromosome inactivation (XCI) as a model, we identify rG4 structures in Xist RNA and determine their function through mutagenesis in vivo. Surprisingly, mutating the rG4 causes a profound X chromosome-wide H3K27me3 enrichment. PRC2 becomes severely dysregulated and is entrapped in a transitory chromosome compartment, precluding chromosome-wide silencing. Our study thereby establishes the importance of rG4 folding in vivo for the regulation of PRC2 and epigenetic silencing.

RESULTS

rG4-calling pipeline reveals shifting populations of rG4 during ES cell differentiation

We modified existing rG4-seq and RT-Stop protocols 17,3335 based on differential sensitivity of RNA structures to dimethyl sulfate (DMS), reverse transcriptase (RT), and the monovalent cations Li+ versus K+ (Fig 1A). For in vitro analysis (Fig. 1A, top track), we purified chromatin-associated RNAs from ES cells (Fig. S1A) and isolated the polyA+ fraction to enrich for RNA POL-II products. We then fragmented the RNA, ligated adaptors (Fig. S1B, DMS-), denatured the RNA to eliminate secondary structures, and then renatured the RNA in the presence of either the rG4-stabilizing monovalent cation, K+, or the rG4-nonstabilizing cation, Li+. K+ intercalates between G-quartets to offset repulsive negative charges. The RNAs were then reverse-transcribed using Superscript III RT, size-selected, and cloned for sequencing (Fig. S1B, DMS-). rG4’s that reformed in vitro following denaturation would occur only in K+, not Li+. Because RT is stalled by rG4, an rG4 could be inferred by a chain-termination event (“RT-Stop”) 17.

Fig. 1. Identification of in vivo rG4 structures at single-nucleotide resolution.

Fig. 1.

(A) Experimental pipeline for identifying in vitro and in vivo rG4’s in day 7 ES cells. Starting with chromatin-bound RNAs, in vivo experiment involves in-cell DMS treatment, whereas in vitro experiment involves no DMS treatment. RT, reverse transcriptase. Lollipops, DMS-methylation.

(B) DMS-sensitive nucleotides in the context of Watson-Crick, wobble, and Hoogsteen base-pairing. N, nitrogen, highlighted in red for guanine-N7 position; green for adenosine-N1 and cytosine-N3. Red lollipops, guanine-N7 modification. Green lollipops, adenosine-N1 and cytosine-N3.

(C) Summary of rG4 pipeline. rG4’seeker algorithm predicted 894,462 total sites are possible for rG4 formation. Of these, 4,823 sites (from sequences with >6 reads) are statistically significant for RT-stops when folded in vitro under K+ conditions. An additional 615 rG4 in vitro sites are found by analyzing RT-stop frequencies at other G-rich regions. In vivo sites are called from the DMS-treated samples. Among 5,538 in vitro rG4 sites, 54 are empirically observed in vivo. Two biological replicates for each condition were analyzed.

(D) rG4 profile for nucleotides 2415–2480 of representative RNA, Phf1. RT-stop frequency in vitro (DMS-) and in vivo (DMS+). Significant K+-dependent RT-stop positions are indicated by black arrows. Binomial test, P<0.05.

(E) rG4 site density distribution metagene plot. rG4 sites: In silico, n=894,462; in vitro, n=5,538; in vivo, n=54. TSS, transcription start site. TES, transcription end site. EZH2 dCLIP metagene profile 9 is shown as a comparison.

(F) In vivo rG4 folding scores for 5’UTR (n=9 genes), 3’UTR (n=144), and coding sequence (CDS, n=18). *P < 0.05. (Mann-Whitney test).

(G) Number of statistically significantly folded rG4’s in vivo shown in Venn diagrams (FDR-adjusted). Number of distinct and overlapping sites shown in the intersect called from two biological replicates for each differentiation time point. rG4 sites having sufficient sequencing coverage ( > 6) across the time course were analyzed.

For in vivo analysis (Fig. 1A, bottom track), we treated cells with DMS to capture native RNA conformations (Fig. S1B, DMS+), then denatured and renatured the RNA in the presence of K+ versus Li+ prior to RT-primer extension analysis. DMS can methylate 3 of 4 nucleosides in RNA: Adenosine (A) at the N1 position, cytosine (C) at the N3 position, and guanine at the N7 position. However, when RNA is duplexed via Watson-Crick or wobble base-pairing, the N1-A and N3-C bases are protected from DMS, whereas N7-G is not — because N1-A and N3-C participate in hydrogen bonding for base pairing, while the N7 position plays no part (Fig. 1B). By contrast, when RNA is in rG4 conformation, N7-G is protected from DMS, whereas N1-A and N3-C are not. Thus, following denaturation and renaturation, if a G-rich region folded into rG4 in vivo (N7-G protected from DMS), the rG4 would refold in the presence of K+ but not Li+. If the G-rich region did not fold into rG4 in vivo, it would be N7-G-methylated and unable to refold into an rG4, due loss of G-quartet hydrogen-bonds. The in vivo rG4 would be inferred from sequencing of RT primer extension profiles at nucleotide resolution. RT stalls and chain-terminates when there is a folded rG4, a methylated N1-A, or methylated N3-C, but is undeterred by methylated N7-G. Thus, by comparing (1) differential RNA sensitivities to DMS in vivo, (2) differential RT-Stop profiles, and (3) differential renaturation under K+ versus Li+ conditions, our pipeline surveyed possible rG4 sites in vitro and in vivo at nucleotide-resolution (Fig. 1A,B). Henceforth, we dub our rG4-seq pipeline “d-rG4-seq”.

Using two biological replicates, we called significant rG4’s using a binomial test (FDR cutoff 0.1) to compare K+ versus Li+ RT-stop profiles. No obvious stress response was triggered during the DMS treatment in cells (Fig. S1C). Libraries yielded approximately 100–500 million mapped reads (Table S1). In day 7 differentiating ES cells, among 894,462 possible rG4 sites predicted by rG4seeker software 36, only 4,923 rG4’s formed in vitro in a K+-dependent manner (Fig. 1C). An additional 615 rG4’s (not predicted in silico; potential G-quadruplex & G≥40% and potential G-triplex & G≥40%) could also be observed for a total of 5,538 in vitro sites (Fig. 1C, Table S2,S3). Among these, only 54 stably folded in vivo (Fig. 1C, Table S4). K+-dependent RT-stops were enriched (69%) for G’s up to 3 bases upstream of the RT-stop (Fig. S1D,E), consistent with prior studies 17,35. Folded rG4’s frequently occurred within tandem, overlapping rG4 motifs, but only one significant site was usually detected in vivo (e.g., Phf1; Fig. 1D). Overall, in vivo-validated rG4’s were stand-alone structures, whereas 36% of in vitro-rG4’s could occur with one or more rG4’s in the same region (Fig. S1F). Metagene analysis showed that transcription start sites (TSS) and transcription end sites (TES) were enriched for folded rG4 (Fig. 1E). In vivo folding scores were significantly greater for 5’- and 3’-UTR’s than in coding sequences (CDS)(Fig. 1F). Interestingly, the rG4 sites correlated with sites of POL-II pausing 9,12,37.

To ask if rG4 sites changed during ES cell differentiation, we examined ES at differentiation days 0 (pre-XCI), day 3 (XCI initiation), and day 7 (XCI mid-point)(Fig. 1G, Fig. S2AD)38. In day 0 cells, there were 1500–3000 folded rG4’s in vitro, depending on whether FDR-adjusted (more stringent) or P-value-based (less stringent) criteria was applied (Fig. S2A). Very few, if any, were stable enough to be detected in vivo at this timepoint (Fig. 1G, S2B, Table S4S6). In the more stringent analysis, none stably folded on day 0 (Fig. 1G) — consistent with prior assessment in undifferentiated ES cells 17. In the lower stringency analysis, 7 stably folded rG4’s were detectable on day 0 (Fig. S2B; Table S4S6).

Cell differentiation led to a larger number of stable rG4’s. In day 3 cells, approximately 1000 (FDR-adjusted) to 2000 (P-value-based) stable structures could be detected by d-rG4-seq in vitro (Fig. S2A), but only 72 (FDR-adjusted) to 136 (P-value-based) could be observed in vivo (Fig. 1G, S2B; Tables S2S5). In day 7 cells, 5500–6300 rG4’s were detected in vitro, but only 54 (FDR-adjusted) to 196 (P-value-based) sites stably folded in vivo (Fig. 1G, Fig. S2B, Table S4S6). rG4’s were not only more numerous on days 3 and 7, but also exhibited higher scores (Fig. S2C,D; P<0.0001, Wilcoxon test). Intriguingly, the rG4 sites were largely non-overlapping between days 3 versus 7 (Fig. 1G, S2B; S2E; Table S4S6), suggesting that rG4’s fold dynamically in a temporally specific manner. Thus, while the mammalian genome harbors thousands of rG4-forming sequences, only a small subset can fold stably at a given time — with a half-life long enough to be detected by d-rG4-seq — and the subsets represent a traveling population in vivo.

Xist RNA harbors developmentally regulated rG4 structures

Among 96 differentially expressed genes (DEGs) in day 7 cells when compared to day 0 (Fig. S2F), we observed that folded rG4’s occurred mostly in pre-existing transcripts (e.g., Thrap3, Kdelr1 and Srebf2)(Fig. S2F), but genes with large expression changes, such as Xist, Soat1, and Bmf, also showed changes in rG4 folding. Only 4 rG4 genes were in day 7-specific transcripts (H19, Podxl, Arl4c and Flnc; Fig. S2F). rG4-containing transcripts were enriched for RNA regulation, translation, apoptosis, and cytokine/steroid responses between days 3–7 (Fig. S3). To examine rG4 function, we turned to XCI, the mechanism of dosage compensation that silences one of two X-chromosomes in the female 39,40. XCI silencing is initiated by Xist RNA, a 17–20 kb noncoding transcript that is expressed exclusively from the inactivate X (Xi) chromosome and spreads in cis to recruit silencing factors 41. Multiple rG4 motifs occur in Xist RNA (Fig. 2A,S4AB). Three clusters of closely spaced rG4 sites were observed in day 7 female ES cells. In vitro, one significant cluster of 3 rG4’s (K+ dependent RT-stops) occurred between Repeats A and F [AF]. A second cluster of 3 rG4’s occurred in exon 7 [E7], and a third cluster of 3 rG4’s was detected at the 3’ terminus [3’](Fig. 2A,B -arrows). In vivo, only one rG4 at the 3’ site was stable enough to be detected in day 7 cells.

Fig. 2. rG4 motifs regulate Xist RNA and H3K27me3 deposition on the Xi.

Fig. 2.

(A) Schematic presentation of Xist rG4 motif location. RepA RNA, Xist repeat regions and rG4 motifs were indicated on the gene structure diagram. RNA sequence of each rG4 motifs were shown below, with in vitro RT-stop sites in bolded G. In vitro rG4 (AF rG4 and E7rG4), in vivo rG4 (3’rG4) and in silico rG4 (21 motifs) sites are marked as orange, red and gray respectively.

(B) Comparison of rG4 folding in vitro (DMS-) and in vivo (DMS+) for three Xist rG4 regions. Significant RT-stops are indicated with arrows.

(C) RT-qPCR based rG4 induced RT-stop quantification for AF and E7 rG4 region. ****, P < 0.0001 (t-test).

(D) Immuno-RNA-FISH for H3K27me3 and Xist RNA in day7 differentiating EB. Control (TST), AF rG4 mutant (#7, #11), E7 rG4 mutant (#4, #6) and 3’ rG4 mutant (#5, #7) cells were used. Representative confocal microscopy images are shown.

(E) Measured Xist cloud size in Xist rG4 mutant cells. ****, P < 0.0001 (t-test). WT, n=162. AF mut #7, n=74. AF mut #11, n=134. E7 mut #4, n=140. E7 mut #6, n=128. 3’ del #5, n=101. 3’ del #6, n=101.

(F) Quantified H3K27me3 signal intensity co-localized with Xist clouds. ***, P < 0.001. ****, P < 0.0001 (t-test). WT, n=189. AF mut #7, n=136. AF mut #11, n=103. E7 mut #4, n=131. E7 mut #6, n=361. 3’ del #5, n=365. 3’ del #6, n=246.

(G) Number of nuclei showing H3K27me3 signal intensity co-localized with Xist clouds. Ns, not significant. *, P < 0.05. ***, P < 0.001. ****, P < 0.0001 (t-test).

We speculated that other Xist rG4’s may be short-lived, precluding detection using d-rG4-seq. Differential Xist rG4 folding could also occur during XCI. To test these ideas, we performed the more sensitive primer extension-qPCR assay, which relies on site-specific primers designed for the Xist RNA (rather than the whole transcriptome). Indeed, time-course analysis showed that rG4 was detected at the AF site only at day 0 and the E7 rG4 could not fold stably at days 3 and 7 (Fig. 2C). Thus, within Xist, rG4’s also fold dynamically and the AF-rG4 (in particular) may fold stably only in undifferentiated ES cells. These results do not preclude transient rG4 folding below the detection limit of d-rG4-seq.

Loss of Xist rG4 causes a hyper-enrichment of H3K27me3 in vivo

To determine if rG4’s function during XCI, we mutated the three rG4 clusters by shuffling G-nucleotides in AF and E7 motifs and deleting the 3’-rG4 (Fig. S4C,D), specifically targeting the allele located on the future inactive X (Xi) — the X-chromosome of Mus musculus (mus) origin 42. Two independent clones and two biological replicates were analyzed for each region (Fig. S4E,F). All three mutations resulted in loss of rG4 formation in vivo, as evidenced by the loss of K+ dependent RT-stops in a primer extension analysis (Fig. S4G). In AF mutant clones (#7, #11), confocal analysis revealed poor Xist expression, RNA localization (Fig. 2DE), and weakening of H3K27me3 enrichment on the Xi, with ~35% of nuclei exhibiting no enrichment and 47–58% displaying dispersed signals (Fig. 2D,F,G). Quantitation via confocal signals showed 50% loss of Xist and H3K27me3 signals (Fig. 2DG). Because we were not able to separate Xist expression from PRC2 recruitment, we set aside this mutant temporarily.

Because the E7 and 3’ mutants did not have a substantial impact on Xist expression or localization (Fig. 2D,E,S4H), we pursued these mutants to examine the consequences of a pure rG4 loss. Confocal quantitation revealed a surprising hyper-enrichment of H3K27me3 on the Xi of the 3’ mutants (Fig. 2D,F,G). The 3’ mutants exhibited a major increase affecting 75–90% of all nuclei, whereas the E7 mutants demonstrated consistent H3K27me3 levels. As the E7 also rG4 did not demonstrate robust rG4 folding in differentiating ES cells (Fig. 2C), we focused further analysis on the 3’ rG4 mutant. Although Xist localization did not appear to be affected under confocal microscopy (Fig. 2D,E), a higher resolution view may reveal defects. We performed CHART-seq (Capture Hybridization Analysis of RNA Targets with deep sequencing)43 to map Xist-binding sites at nucleotide resolution in 3’ rG4 mutant clones, #5 and #6, in two biological replicates. Intriguingly, while CHART revealed a modest increase in Xist binding across the Xi (Fig. 3A), allele-specific ChIP-seq for H3K27me3 demonstrated super-enrichment of the PRC2 mark across the Xi (Fig. 3B), agreeing with immunostaining results (Fig. 2D). Increased H3K27me3 level was exclusive to the Xi, with no visible H3K27me3 changes on autosomes (e.g., chromosome 13) and the active X (Xa) (Fig. 3C). Zoom-in’s to gene regions confirmed significantly increased levels of H3K27me3 in the 3’ mutant (Fig. 3D,S5A), with a direct correlation between the density of Xist RNA and H3K27me3 (r = 0.75–0.76; Fig. 3E). ChIP-qPCR analysis using allele-specific primers further confirmed the H3K27me3 hyper-enrichment on the Xi (Fig. S5B). Quantitative analysis further showed that the H3K27me3 enrichment was specific to the Xi (Fig. 3F). These data indicate that Xist’s 3’ rG4 antagonizes trimethylation of H3K27 on the Xi.

Fig. 3. Xist 3’rG4 deletion enhances Xist RNA binding and PRC2 mark enrichment on Xi.

Fig. 3.

(A) 3’ rG4 mutant alters S1/S2 compartment Xist binding (CHART-seq) patterns. X chromosome wide Xist enrichments in the indicated cell lines are shown.

(B) Allele-specific ChIP-seq for H3K27me3 in Xist 3’ rG4 mutants. Xi (mus) allele specific H3K27me3 coverage is shown.

(C) Control genomic regions of H3K27me3 ChIP-seq profile in Xist 3’ rG4 mutants. Chromosome 13 (mus) and Xa (cas) allele specific H3K27me3 coverages are shown.

(D) Allele-specific H3K27me3 ChIP-seq profile for representative genes in Xist 3’ rG4 mutants. Xi (mus) and Xa (cas) allele specific H3K27me3 coverages are shown.

(E) Comparison of Xist and H3K27me3 density in Xi genomic regions. Calculated Pearson correlation coefficient of #6 (red) and #5 (blue) is indicated.

(F) Xi(mus), Xa(cas), and autosomal (chromosome 14) allele specific H3K27me3 density plots for genic regions. Ns, not significant. *, P < 0.05. ****, P < 0.0001 (Wilcoxon test).

rG4 folding inhibits the histone methyltransferase activity of PRC2

Although prior studies revealed that RNA binding inhibits the catalytic activity of PRC2 812, it is not known whether specific RNA structures are required for inhibition. To investigate if rG4 folding might underlie PRC2 inhibition, we turned to biochemical analyses. To quantify the strength of PRC2-rG4 interactions, we performed electrophoretic mobility shift assays (EMSA) with P32-labelled Xist rG4 probes AF (46 nt), E7 (43 nt), and 3’ rG4 (61 nt) and E7 rG4 (43 nt) and increasing concentrations of purified recombinant PRC2 (Fig. 4A). We performed EMSA in the presence of K+ versus Li+ to determine the effects of rG4 folding. In the presence of K+, we observed sub-nanomolar binding affinities in all three cases, with Kd 171.6 ± 27 nM for AF, 46 ± 13 nM for E7, and 66.6 ± 16 nM for 3’ rG4 (Fig. 4A). In the presence of Li+, the affinities were significantly reduced (P < 0.0001; 2-way ANOVA) by ~2-fold for AF, 8-fold for E7, and 4-fold for the 3’ motif. Thus, PRC2 clearly favors a folded rG4 structure for binding. Consistent with this, mutating the motifs also substantially reduced affinities (Fig. 4A: AF mut 46 nt, E7 mut 44 nt, and 3’ mut 63 nt).

Fig. 4. rG4 folding enhances PRC2 binding but suppresses its histone methyltransferase activity in vitro.

Fig. 4.

(A) Electromobility shift assays with indicated Xist rG4 and Repeat A RNA (2.5 nM), and purified PRC2 protein. Assays were performed with potassium (K+) or lithium (Li+) containing buffers. PRC2-bound RNA signals were quantified and plotted as a fraction of the total signal within each lane. Graphic summary indicates the hypothetical competition between Xist RNA structures on PRC2.

(B) RNA competition assay using AF rG4 RNA (2.5 nM) and the increasing amount of cold repeat A RNA fragments. PRC2-bound AF rG4 RNA signals were quantified and graphed as a fraction of the total signal within each lane.

(C) PRC2 HMT assay with the increasing amounts of the protein.

(D) Effects of Xist rG4 RNA on PRC2 histone methyltransferase activity. Methyltransferase assays were conducted in the presence of GST-tagged histone H3 peptide fragment (4 uM), S-adenosyl homocysteine (100 uM) and 150 uM PRC2. Assays were performed with potassium (K+) or lithium (Li+)-containing buffers.

We compared the rG4 affinities to that of Repeat A, an Xist element known to bind PRC2 2,44. Intriguingly, despite their smaller sizes, all three Xist rG4’s showed affinities on par with that of the 435-nt Repeat A (Kd 100 ± 11 nM; Fig. 4A, bottom). No binding of PRC2 to the 300-nt negative control MBP RNA was observed (Kd >600 nM). Binding of AF rG4 was competed away by excess Repeat A, but not by a truncated form, Repeat A (I-II)(75 nt)(Fig. 4B), raising the possibility that Repeat A and rG4 in Xist may compete for PRC2, potentially enabling a dynamic regulation of PRC2 recruitment and catalytic activation (Fig. 4A, cartoon).

We then assessed the impact of rG4 folding on PRC2’s HMT activity using chemiluminescence (MTase-gol kit) 45, which typically yields a dose-dependent HMT activity (Fig. 4C). Addition of folded rG4’s (K+ buffer) led to a concentration-dependent reduction of HMT activity for all three Xist rG4’s (Fig. 4D, left panel), with an IC50 of 69 ± 9 (AF), 57.3 ± 2 (E7), and 64.6 ± 6 (3’) nM. This inhibition was proportional to the their PRC2 affinities (Fig. 4C). HMT inhibition by RNA was substantially greater under K+ than Li+ conditions (Fig. 4D), arguing that rG4 folding is required for the effect. Mutating the rG4 motifs similarly blunted HMT inhibition in all three cases (Fig. 4D), further arguing that the folded rG4 structure (not the mere G-richness of the underlying sequence) is crucial.

We sought to test this hypothesis in vivo and turned to pharmacological stabilizers of rG4’s to evaluate consequences for PRC2 activity on chromatin. (There are no known pharmacological inhibitors). The high-affinity compound, phenDC3, harbors a polycyclic aromatic hydrocarbon ring that intercalates into G-stacks to stabilize the rG4 structure 46. We applied 12 uM phenDC3 and analyzed Xist and H3K27me3 patterns at day 7. While Xist expression was not changed, H3K27me3 enrichment was severely reduced by phenDC3 (Fig. 5A,B), supporting the notion that rG4 stabilization inhibits PRC2 activity in vivo. We tested 2 additional rG4 stabilizers, BRACO19 and TMPyP4, and both exhibited dramatic reduction in H3K27me3 accumulation on the Xi (Fig. 5A,B). To confirm the stabilization of rG4 in vivo, we performed RT-stop analysis using the primer extension assay. All three rG4 stabilizers demonstrated significantly increased RT-stops in Xist AF, E7, and 3’ motifs (Fig. 5C,D; S6), though the effects differed in magnitude for the three motifs.

Fig. 5. rG4 stabilizers suppress H3K27me3 deposition on the Xi.

Fig. 5.

(A) Immuno-RNA-FISH for H3K27me3 and Xist RNA in G-quadruplex ligand treated cells. 12 uM of PhenDC3, BRACO19, or TMPyP4 was applied at day 4 and replenished daily with media change. Cells were harvested at day 7 differentiation. Representative confocal microscopy images are shown.

(B) H3K27me3 signals co-localizing with Xist clouds were quantified. ****, P < 0.0001 (Mann-Whitney test).

(C) Effect of phenDC3 on the Xist rG4 folding. Primer extension assay was performed using in vitro transcribed rG4 RNA regions and radiolabeled primers complement to the downstream regions of indicated rG4 motif. Lines and arrowheads mark locations of rG4 motifs and K+-dependent RT-stops. 10 uM phenDC3 was applied to the reaction where indicated.

(D) RT-stop signal plots of the potassium and phenDC3-dependent primer extension assays. RT-stop bands enhanced in the presence of PhenDC3 are indicated with red-colored arrowheads. Primer extension products longer than rG4 motifs were quantified and graphed in the right panels.

Taken together, these data argue that the rG4 structure — not the mere G-richness — is responsible for HMT inhibition. By dynamically folding and unfolding its rG4’s, Xist RNA could regulate recruitment and activation of PRC2’s HMT function (Fig. 5E). We therefore propose that PRC2 is recruited by Xist RNA to the X-chromosome — by either the rG4 itself or other motifs within Xist — but its HMT activity is held in check by binding to folded rG4’s. Subsequent unfolding of the rG4 would reduce affinity for PRC2, enabling activation of PRC2’s HMT activity on chromatin.

To ask if PRC2 remained bound to Xist RNA in the rG4 mutants, we performed RIP assays using antibodies against EZH2 and SUZ12. We first used the fRIP method, which employs light formaldehyde crosslinking and sonication to interrogate RNA-protein association 47. We tested each rG4 mutant separately and performed RT-qPCR in the region flanking the mutated rG4 to determine if PRC2 binding was affected nearby. Intriguingly, none of the rG4 mutations had a negative impact on PRC2-Xist interactions in the flanking region (Fig. 6A, upper panel). In fact, mutating AF and 3’ rG4’s seemed to modestly increase PRC2 binding in the region directly flanking the missing rG4. SUZ12 and EZH2 antibodies yielded similar trends. Because fRIP used formaldehyde crosslinking, PRC2 binding could reflect indirect interactions with Xist. We therefore conducted UV-crosslink RIP (UV-RIP or CLIP) with stringent washes to eliminate non-crosslinked RNA (indirect interactions) 48. SUZ12 UV-RIP also revealed stable Xist-PRC2 interaction without rG4s (Fig. 6A, bottom panel). In contrast, Rps27 And S100a10 — two transcripts that do not interact with PRC2 9 — did not show binding to PRC2 with fRIP or UV-RIP (Fig. S5B). Thus, Xist regions flanking the mutated rG4’s may also play a role in PRC2 binding. Notably, Xist’s Repeat A 2 and Repeat B 49,50 have been shown to contribute to PRC2 recruitment, directly and indirectly. Given that Repeat A competes effectively with the AF rG4 for PRC2 binding (Fig. 4B), we suggest that a dynamic interplay between them may be involved in PRC2 recruitment and regulation of its HMT activity (Fig. 4A, cartoon).

Fig. 6. Xist 3’rG4 deletion traps Xist RNA and PRC2 in the S1 compartment of the Xi.

Fig. 6.

(A) PRC2 binding patterns in the Xist rG4 mutants. Xist regions flanking the indicated rG4 motifs were examined using formaldehyde cross-linking RNA immunoprecipitation (fRIP) or UV-crosslinking RNA immunoprecipitation (UV-RIP) with EZH2 and SUZ12 antibodies. The amount of coprecipitated RNA was measured with qPCR analysis. *, P < 0.05. **, P < 0.01 (t-test).

(B) PRC2 mark enrichment in the 3’ rG4 mutants. SUZ12 and H3K27me3 ChIP-qPCR analysis was performed using allele specific primers targeting Msn gene on the Xi. *, P < 0.05. (t-test).

(C) Xist CHART-seq and allele-specific ChIP-seq profiles for H3K37me3 in Xist 3’ rG4 mutants. Day 7 EB PC1 track taken from 38.

(D) Quantification of Xist and H3K27me3 profiles in 3’ rG4 deletion cell lines. Normalized read density was counted over the S1 and S2 genomic regions. TST versus 3’ rG4 mutants (#5, #6); **, P < 0.01. ****, P < 0.0001 (Wilcoxon test). S1 versus S2 genes; ****, P < 0.0001 (Mann-Whitney test).

(E) Xist CHART density and H3K27me3 coverage plot over S1 and S2 genes in control versus 3’rG4 mutants (#5, #6). Ns, not significant. ****, P < 0.0001 (Wilcoxon test).

(F) H3K27me3 coverage normalized by Xist CHART density. Graphs are plotted over S1 and S2 genes in control versus 3’rG4 mutants (#5, #6). ****, P < 0.0001 (Wilcoxon test).

(G) Schematic of S1 and S2 compartments of the Xi during the process of XCI. The Xist gene is located in the S1 compartment. Xist RNA therefore initially spreads through the S1 compartment (red). Spreading (green arrows) into S2 (blue) is known to require SMCHD1.

3’ rG4 loss causes Xist entrapment and failed XCI silencing

We focused further analysis on the 3’ rG4 site, as this mutation greatly affected PRC2 marks without affecting Xist expression (Fig. 2D,E). Given that Xist localization was overall enriched by CHART-seq (Fig. 3A), we asked if there were underlying heterogeneities. Intriguingly, principal component analysis revealed an anomalous localization in the 1st principal component (PC1). In pre-XCI female ES cells, the Xist locus resides in the active-gene compartment known as the A-compartment. During cell differentiation, Xist RNA is upregulated and spreads throughout the A-compartment, crosses into the B-compartment, and merges A/B compartments into Xi-specific structures known as S1 and S2 compartments 38. The S1 and S2 compartments occupy opposing hemispheres of the Xi 51. In 3’ rG4 mutants, Xist densities were specifically increased within the S1 compartment, as shown by close alignment of Xist signals with positive eigenvectors in PC1 (Fig. 7C – PC1 track marked S1/S2)38. Quantitative analysis confirmed a significant increase in Xist density and H3K27me3 enrichment in the S1 compartment, with relative depletion in the S2 compartment (Fig. 7D). A gene-centric view similarly demonstrated increased localization of Xist and enrichment of H3K27me3 in the S1 compartment (Fig. 7E, S7A). Despite loss of Xist spreading into the S2 compartment, however, genes in the S2 compartment also showed a significant increased H3K27me3 levels (Fig. 6E, S7B).

Fig. 7. Xist 3’rG4 promotes complete Xi gene silencing.

Fig. 7.

(A) Allele-specific H3K27me3, Xist CHART, and RNA-seq profiles of Xist 3’ rG4 mutant. The 10 Mb region around the border of S1 and S2 is shown. The Reference Sequence genes are presented at the bottom track.

(B) Xist 3’rG4 mutant gene expression patterns over S1 and S2 genes. Xa genes were shown as a comparison. Ns, not significant. ****, P < 0.0001 (Wilcoxon test).

(C) S1 and S2 gene expression changes normalized to Xa genes in control (TST) and 3’rG4 mutants (#5, #6). TST versus 3’ rG4 mutants (#5, #6); ***, P < 0.001. ****, P < 0.0001 (Wilcoxon test).

(D) Violin gene expression plot showing Xi gene expression normalized to Xa. ****, P < 0.0001 (Wilcoxon test).

(E) Cumulative gene distribution plot presenting the fraction of Xist 3’ rG4 mutant genes normalized to control (WT). Xa genes were shown as a comparison. p-values are calculated by the Wilcoxon test.

(F) Cartoon depicting the observation that rG4 is required for Xist spreading to the S2 compartment and achieving full gene silencing.

(G) Model of PRC2 recruitment, activation, and eviction by dynamically folding and unfolding of rG4. rG4 folds on nascent transcript, attracting PRC2 to bind but also holding its HMT activity in check. Because RNA competes with DNA for PRC2 binding, PRC2 does not load onto chromatin. Upon rG4 unfolding, PRC2 is released from RNA, loads onto chromatin, where its HMT function is activated (H3K27me3). The reaction is reversible, as the refolding of rG4 attracts PRC2 back onto RNA (which is in spatial proximity), thereby evicting PRC2 from chromatin.

(H) Hit-and-run action along the Xi: In XCI, Xist RNA is present at only 100–200 molecules/cell 53, but its association with catalytic factors such as PRC2 allows for signal amplification. Dynamic rG4 folding/unfolding model enables a “hit-and-run” mechanism 53, whereby Xist RNA is anchored at discrete sites along the Xi and, through 3D loop extrusion, tracks along Xi chromatin to progressively unleash the HMT activity of PRC2 for H3K27me3. rG4 folding attracts PRC2 to Xist RNA and unfolding releases PRC2 onto chromatin to activate its HMT. Because of spatial proximity, rG4 refolding attracts PRC2 back onto Xist where the cycle repeats itself on the next nucleosome. Thus, dynamic RNA conformational changes allow Xist-PRC2 interactions to track along chromatin and overcome the numerosity problem through signal amplification.

Lastly, to determine whether the rG4 mutation affected gene silencing on the Xi, we performed transcriptomic analysis in both 3’ rG4 mutant (#5, #6) with two biological replicates. Despite the hyperactivation of PRC2 in the S1 compartment, transcriptomic analysis showed a loss of gene silencing (Fig. 7AC; S7A). Similarly, S2 genes exhibited a substantial loss of silencing (Fig. 7AC; S7B). Equivalent genes on the Xa showed no significant changes in expression (Fig. 7B). In aggregate across the entire Xi, a dramatic loss of gene silencing was evident (Fig. 7D), with cumulative distribution plots (CDP) showed a substantial right shift indicative of a ~2-fold upregulation of Xi genes (Fig. 7E). Thus, loss of the 3’ rG4 caused a severe dysregulation of gene silencing across the entire X-chromosome, despite the elevated level of the repressive PRC2 mark. These findings indicate that translocation of the Xist-PRC2 complex from S1 to S2 requires the 3’ rG4 (Fig. 7F). We conclude that mutating the 3’ rG4 site leads to (i) entrapment of Xist in the S1 compartment, (ii) a dysregulated hyperactivation of PRC2 in the S1 compartment, and (iii) loss of gene silencing across all of Xi.

DISCUSSION

Here, we have shown that rG4’s fold dynamically and selectively in vivo. Although thousands of sites have the capacity to fold into rG4 in vitro, only a hundred or so can be detected at a given time in ES cells. In undifferentiated ES cells, rG4’s largely do not fold stably enough to be detected in significant numbers using our method. This observation is consistent the inability to detect rG4’s in a prior study 17. In differentiating cells, however, rG4’s are more stable and the number of detectable rG4’s increases. As rG4 site usage is non-overlapping between days 0, 3, and 7, our data argue for dynamic rG4 folding and unfolding in cells on a time scale that is largely too short to be measured using epigenomic methods such as d-rG4-seq52. We propose that many, if not all, of the thousands of in vitro rG4’s may fold in vivo in specific contexts — but only transiently.

Here we have shown that one such context is XCI. We have identified three clusters of rG4’s in Xist RNA. Biochemical analysis demonstrate that PRC2 preferentially binds folded rG4’s in Xist and it is the folded form that inhibits the HMT activity of PRC2. In vivo, pharmacologically stabilizing the rG4 results in a marked depletion of H3K27me3 on the Xi, consistent with the idea that rG4 binding blocks the HMT activity. We also observed that mutating the 3’ rG4 leads to entrapment of Xist in the S1 compartment, a marked hyperactivation of PRC2, and loss of XCI. Our study therefore supports a model in which Xist recruits PRC2 to the X-chromosome, but the rG4 initially holds its HMT activity in check (Fig. 7G). Unfolding of the rG4 would then release PRC2 for its activation on chromatin. While d-rG4-seq cannot visualize dynamic folding and unfolding in real time, the small molecule stabilizers and G-tract mutagenesis in vivo and the K+/Li+ effects in vitro together serve as proxies for folding and unfolding dynamics to provide snapshots of each state.

Our findings are consistent with the eviction model positing that G-tract RNA competes with chromatin for binding and thereby removes PRC2 from chromatin 12. Indeed, eviction could be viewed as the reverse reaction to recruitment. When the G-tract is folded into rG4, PRC2 is recruited by RNA to the locus, but its HMT activity is held in check until the rG4 unfolds — at which point, PRC2 loads onto chromatin and its activity is induced (H3K27me3)(Fig. 7G). When the rG4 refolds, the RNA outcompetes chromatin for PRC2 and PRC2 is evicted from chromatin to rebind RNA 12. CryoEM analysis has suggested that binding of PRC2 to rG4 results in formation of an inactive dimeric complex due to reduced access of histone H3 tail to the catalytic EZH2 subunit, and EZH2 comprises a loop structure that facilitates the release of PRC2 from RNA to DNA upon activation 32. We postulate that rounds of rG4 folding and unfolding would result in a dynamic regulation of PRC2 recruitment, catalytic activity, and eviction from chromatin (Fig. 7G). Currently, the absence of a method for real-time detection of rG4 folding and unfolding events in vivo limits our ability to explore whether such dynamism contributes to PRC2 regulation by Xist rG4s. If rG4’s undergo dynamic cycles of folding and unfolding, it is plausible that an additional 5000–6000 in vitro rG4’s may also participate in PRC2 regulation at a global level. Real-time imaging of rG4’s will be the next challenge toward comprehending the impact of rG4 folding dynamics in epigenetic processes.

In XCI, Xist RNA is present at only 100–200 molecules/cell — and is at a stoichiometric disadvantage relative to the size of the Xi 53. Its engagement of catalytic factors, such as PRC2, amplifies the Xist signal, allowing Xist to act globally. Dynamic rG4 folding and unfolding would facilitate a “hit-and-run” mechanism 53, whereby Xist-PRC2 complexes are anchored at discrete sites along the Xi and, through loop extrusion, would propagate PRC2 and H3K27me3 along the chromatin in cis (Fig. 7H). Repeated cycles of rG4 folding and unfolding would allow PRC2 to track along chromatin and toggle between Xist RNA and Xi chromatin for processive H3K27me3.

These conclusions also have significant implications for the RNA specificity model regarding PRC2 regulation 2,3,5,8,12. It has been suggested that PRC2 may be “promiscuous” or “nonspecific” for RNA due to the occurrence of numerous G-rich binding sites in the transcriptome 10,54,55. Another recent study even suggested that PRC2 is not a true RNA-binding protein and that detected interactions published by other studies 2,3,79,12,32,55 result from spurious in vitro artifacts 56. However, our current study further supports specific PRC2-RNA interactions, as indeed PRC2 has high affinity for folded rG4 structures and the rarity of such folded rG4 structures in vivo accounts for PRC2’s ability to discriminate between RNA species. While our current study implicates a direct physical interaction of Xist rG4 and PRC2, it does not address which region(s) of Xist RNA first recruits PRC2 during XCI. Nor does our study exclude indirect mechanisms. Notably, when one rG4 cluster is mutated, Xist can still recruit PRC2, likely because other domains of Xist RNA might compensate. Previous studies proposed two potential mechanisms — a direct recruitment by the Repeat A motif of Xist 2,8 versus indirect recruitment via Repeat B 50,57, or both 49,58.

Our study reveals an unexpected role of rG4 folding for Xist-PRC2 translocation from the S1 to S2 compartment 38. When the 3’ rG4 is mutated, Xist-PRC2 complexes become trapped within the S1 compartment. Xist and Polycomb proteins are known to depend on each other for proper spreading along the Xi, as PRC1 and PRC2 knockouts show substantial reduction in Xist occupancy 49. Therefore, S2 Xist spreading defects may arise indirectly in part from Polycomb dysregulation caused by loss of Xist rG4 folding. Loss of rG4 also compromises global Xi silencing. Despite repressive H3K27me3 being super-enriched in S1, S1 genes do not correctly silence. Hence, Polycomb marks alone are not sufficient for full gene silencing. Through interactions with PRC2, rG4 folding may also contribute to Xi silencing by controlling POL-II pause-release 9,31. The loss of rG4 may thereby exacerbate XCI dysfunction by impeding Xist transcriptional elongation. As intramolecular G-quadruplexes are highly stable structures in vitro 59, energy may be required to unwind rG4 in vivo. The role of specific RNA helicases will be of major interest in future research.

LIMITATIONS OF THE STUDY

We believe that rG4’s fold in vivo and are not necessarily rare. Fewer than 5% of the 5000–6000 in vitro rG4’s are detected in vivo, but this may be an underestimate. Additional rG4’s may fold too transiently to be detected by a population-based epitranscriptomic assay. It is possible that many, if not all, of the thousands of rG4 motifs may fold in vivo in select conditions. We also note that our rG4-calling method depends on strong RT-Stops in the presence of K+ but not Li+. However, it is known that some robust rG4’s (such as those in TERRA RNA) can fold in the presence of Li+. Thus, our method could under-call the total number of rG4’s in vitro and in vivo. Finally, we note that a recent report suggested that, although the EZH2 antibody used in our study does recognize EZH2, it can cross-react with the RNA-binding protein, SAFB 60. In our study, we validated the PRC2-Xist binding patterns using both EZH2 and SUZ12 antibodies, to alleviate potential concerns regarding antibody specificity.

STAR★Methods

Resource availability

Lead contact

Further information and requests for reagents should be directed to and will be fulfilled by the Lead Contact, Jeannie T. Lee (lee@molbio.mgh.harvard.edu).

Materials availability

Cell lines generated in this study are available upon request to the lead contact (lee@molbio.mgh.harvard.edu).

Data and code availability

Raw high-throughput sequencing data and processed files for DMS-seq, RNA-seq, ChIP-seq, and CHART-seq reported in this paper have been deposited at GEO under accession number: GSE219083. These data are publicly available as of the date of publication. This paper analyzes existing, publicly available data. These accession numbers for the datasets are listed in the key resources table.

Key Resources Table

REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
Mouse monoclonal anti-H3K27me3 Active Motif Cat#AM39535
Rabbit monoclonal anti-H3K27me3 GeneTex Cat#GTX60892
Rabbit monoclonal anti-EZH2 Cell Signaling Cat#5246
Rabbit polyclonal anti-SUZ12 Active Motif Cat#39358
Bacterial and virus strains
BL21(DE3) EMD Millipore Cat# 71401–3
Chemicals, peptides, and recombinant proteins
Recombinant mouse LIF Sigma Cat#ESG1107
Recombinant mouse EZH2 This study N/A
Recombinant mouse SUZ12 This study N/A
Recombinant mouse EED This study N/A
Recombinant mouse RBPA48 This study N/A
Recombinant GST fused histone H3 peptide (16–44 a.a.) This study N/A
SuperScript III reverse transcriptase Invitrogen Cat#18080085
SUPERase•In RNase Inhibitor Invitrogen Cat#AM2694
TURBO DNase Invitrogen Cat#AM2238
RNase H New england biolabs Cat#M0297
T4 Polynucleotide Kinase New england biolabs Cat# M0201
T4 RNA Ligase 2, truncated KQ New england biolabs Cat# M0373
CircLigase ssDNA Ligase Epicentre Cat# CL4111K
RNase A Thermo Scientific Cat# EN0531
Ribonucleoside Vanadyl Complex New england biolabs Cat# S1402S
Dimethyl sulfide Sigma Cat#320293
3X Flag peptide Sigma Cat# F4799
TRIzol Reagent Thermo Scientific Cat# 15596018
Critical commercial assays
MTase-Glo Methyltransferase Assay kit Promega Cat# V7601
Oligotex Direct mRNA kit QIAGEN Cat# 70022
Agencourt AMPure XP Beads Beckman Coulter Cat#A63881
NEBNext Ultra II DNA Library Prep Kit for Illumina New England Biolabs Cat#E7645S
NEBNext Ultra II directional RNA Second Strand Synthesis Module New England Biolabs Cat#E7550S
NEBNext ChIP-Seq Library Prep Master Mix Set for Illumina New England Biolabs Cat#E6240S
NEBNext Poly(A) mRNA Magnetic Isolation Module New England BioLabs Cat#E7490S
Lipofectamine 3000 Transfection Reagent Thermo Fisher Scientific Cat#L3000015
NEBNext Multiplex Oligos for Illumina (Index Primers Set 1) New England BioLabs Cat#E7335S
NEBNext Multiplex Oligos for Illumina (Index Primers Set 2) New England BioLabs Cat#E7500S
Dynabeads Protein G for Immunoprecipitation Invitrogen Cat# 10003D
Deposited data
DMS-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
DMS-seq in WT female mouse differentiating ES cell (day 3) This study GEO:GSE219083
DMS-seq in WT female mouse differentiating ES cell (day 0) This study GEO:GSE219083
Xist CHART-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
Xist CHART-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
Xist CHART-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
In situ Hi-C in WT female mouse differentiating ES cell (day 7) 38 GEO: GSE99991
Experimental Models: Cell Lines
WT Xist female ES cell (TsixTST/+) 42 N/A
AF rG4 mutant Xist female ES cell (TsixTST/+) Clone #7 This study N/A
AF rG4 mutant Xist female ES cell (TsixTST/+) Clone #11 This study N/A
E7 rG4 mutant Xist female ES cell (TsixTST/+) Clone #4 This study N/A
E7 rG4 mutant Xist female ES cell (TsixTST/+) Clone #6 This study N/A
3’ rG4 mutant Xist female ES cell (TsixTST/+) Clone #5 This study N/A
3’ rG4 mutant Xist female ES cell (TsixTST/+) Clone #6 This study N/A
Oligonucleotides
Primers for qPCR,see Table S7 Integrated DNA Technologies N/A
Primers for in vitro trascription,see Table S7 Integrated DNA Technologies N/A
primers for DMS-seq library preparation,see Table S7 Integrated DNA Technologies N/A
Recombinant DNA
pSpCas9(BB)-2A-Puro (PX459) v2.0 61 Addgene Cat#62988
pMB 1609-pRR-EGFP 62 Addgene Cat#65852
pGEX-4T-1-histone H3 peptide (16–44 a.a.) This paper N/A
pJET1.2 Thermo Scientific Cat# K1231
Software and algorithms
SAMtools v1.4.1 63 http://samtools.sourceforge.net/
STAR aligner (v2.7.3) 64 https://github.com/alexdobin/STAR
Cutadapt v1.8.1 65 https://cutadapt.readthedocs.io/en/stable/#
ImageJ v1.53a 66 https://imagej.nih.gov/ij/
bedtools 67 http://bedtools.readthedocs.io/
deepTools 68 https://deeptools.readthedocs.io/en/develop/
Bowtie2 69 http://bowtiebio.sourceforge.net/bowtie2/index.shtml
SAMBAMBA v0.6.6 70 https://github.com/biod/sambamba/releases
Bowtie v1.1.1 71 http://bowtie-bio.sourceforge.net/index.shtml
Python/Biopython v1.70 72 https://biopython.org/
Prinseq-lite v0.20.4 73 https://sourceforge.net/projects/prinseq/files/
Cufflinks v2.2.1 74 http://cole-trapnell-lab.github.io/cufflinks/
Excel Microsoft www.office.com
Prism9 GraphPad www.graphpad.com
Structurefold2 75 https://github.com/StructureFold2/StructureFold2
rG4-seeker 36 https://github.com/TF-Chan-Lab/rG4-seeker

This paper does not use original code.

Additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model and Study Participant Details

Cell growth conditions

Clonal TsixTST/+ mESC (TST) line was cultured as routine procedure. mESC were grown on inactivated feeders in high glucose D-MEM (Thermofisher, 10569044) supplemented with 15% Hyclone FBS (Sigma), 25 mM HEPES pH 7.2–7.5, 1x MEM non-essential amino acids, 1x Pen/Strep, 0.1 mM βME, and 500 U/mL ESGRO recombinant mouse Leukemia Inhibitory Factor (LIF) protein (Sigma, ESG1107) at 37°C with 5% CO2. EB differentiation was performed with growing feeder-depleted cells in the medium without LIF. Cells were kept in suspension until 3 days of differentiation and transferred to gelatinized plates on day 3 for further adherent growth. Where indicated, 12 μM phenDC3 (Sigma, SML2298), BRACO19 (Sigma, SML0560), and TMPyP4 (Calbiochem, 613560) was applied at day 4 and replenished daily with media change. Cells were harvested at day7 differentiation.

Method Details

Generation of Xist rG4 mutant cell lines.

CRISPR-mediated gene base editing was introduced as previously described 76. Xist targeting sgRNA sequence (Table S7) was integrated under the U6 promoter of Cas9/sgRNA ribonucleoprotein expressing plasmid pSpCas9(BB)-2A-Puro (A kind gift from Feng Zhang, Addgene plasmid #62988). Recombination donor plasmid was prepared by introducing indicated Xist mutation to the 0.8 kb long flanking fragment of Xist genomic DNA. Xist sgRNA target sequence was subcloned into the EGFP coding split-site of recombination reporter plasmid pMB1609_pRR-EGFP (A kind gift from Marc Bühler, Addgene plasmid #65852). 2 ug of pSpCas9(BB)-2A-Puro, 1 ug of the doner and 0.4 ug of the reporter plasmids are mixed and co-transfected using lipofectamine 3000 reagent (Thermofisher, L3000075) according to manufacturer’s instructions. 48 hours after transfection, GFP-positive cells were collected by flow cytometry and plated on a feeder-covered plate. After 5 days, 2000 cells are seeded on a 100-mm dish for subsequent clonal cell line isolation. Visible colonies emerged in 3 days, individual colonies were transferred to the feeder-coated 96-well dishes. Clonal cells were initially screened by genotyping PCR using mutant and wild-type gene-specific primers and further verified with sanger sequencing.

DMS treatment

Trypsinized ~ 2 × 107 cells were resuspended in the 5 ml culture medium and quickly transferred to the chemical fume hood. 200 ul (4% v/v) of DMS (Sigma, 32029) is directly applied to the media using a 1 ml syringe. The mixture was incubated at 37°C for 7 min with mild agitation. DMS modification was quenched by adding 1.3 ml βME (Sigma, M3148) after putting the tubes on ice. Cells were recovered by centrifugation at 400 g and 4 °C for 3 min. For complete removal of remaining DMS, cell pellets were repeatedly washed twice with 5ml ice-cold 1x PBS and 25% βME buffer. DMS-treated cells are stored in a –80 °C deep freezer until the RNA purification.

RNA isolation from chromatin-bound fraction

Nuclear fractionation was performed as described 33,34. Frozen cells pellets were placed on dry ice, and resuspended in nuclei isolation buffer (15 mM Tris-HCl pH 7.5, 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 1 mM CaCl2, 250 mM Sucrose, 0.3 % Igepal CA-630, 0.1% βME and 10 mM ribonucleoside vanadyl complex). The solution was repeatedly pipetted until the frozen pellet was completely dissolved. Nuclei were collected at 600 g and 4 °C for 5 min and washed twice with the nuclei isolation buffer without Igepal CA-630. To remove soluble nuclear fraction, nuclei pellet was resuspended in 500 ul NUN1 buffer (20 mM Tri-HCl, pH 8.0, 75mM NaCl, 0.5 mM EDTA, 50% Glycerol and proteinase inhibitor 1xComplete (Roche)) followed by 4.8 ml NUN2 buffer (20 mM HEPES-KOH pH 7.6, 7.5 mM MgCl2, 0.2 mM EDTA, 300 mM NaCl, 1 M Urea, 1% Igepal CA-630). The solution was incubated on ice for 15 min with brief vortexing every 5 mins. An insoluble chromatin pellet was obtained by centrifugation at 10000 g at 4 °C for 15 min centrifugation. 300 μl Trizol (Thermofisher, 15596018) was added to the chromatin pellets, and under freeze-thaw cycles three times. Then the pellets were ground with eppendorf tube pestles (Zymo reaserch, H1001) until completely resuspended in trizol. Additional 700 ml Trizol was applied on the top of the ground mixture and the remaining RNA isolation procedure was performed as following standard Trizol protocol. Poly A+ RNA was collected from 100 – 250 ug chromatin-bound RNA using Oligotex Direct mRNA kit (QIAGEN, 70022).

DMS-seq library preparation

DMS-seq library was prepared according to the previous RNA structure probing protocols with minor modifications 17,77. 1 – 5 ug of poly A+ RNA was chemically fragmented at 95 °C for 8 min in RNA fragmentation buffer (100 mM Tris-HCl pH 8 and 2 mM MgCl2). The fragmented RNA was ethanol precipitated, resuspended in urea gel loading buffer (1x TBE, 3.5M Urea, 0.005% Bromophenol blue, and 0.025% Xylene), and subjected to denaturing urea 8% acrylamide gel fraction. 60– 80 nt RNA fragment was excised and recovered from the gel. RNA samples are dephosphorylated with polynucleotide kinase (PNK) (NEB, M0201) in 10 μl solution comprising 1 μl 10x T4 RNA ligase buffer (NEB, B0216), 1.5 μl T4 PNK and 0.5 μl SUPERase•In RNase Inhibitor (Invitrogen, AM2696) at 37 °C for 5 hours. Then samples are ligated to the 3’ adaptor by directly adding 6 μl 50 % PEG8000, 1 μl 10x T4 RNA ligase buffer, 2 μl 20 μM adenylated 3 DNA adaptor, and 1 μl T4 RNA Ligase 2 truncated K227Q (NEB, M0373) to the dephosphorylated mixture above and incubated at 16 °C for 16 hours. Samples are ethanol precipitated and gel fractionated again, adaptor-ligated products appearing at 85 −105 nt position were gel-purified and resuspended in 13 ul water. For reverse transcription, 13 μl adaptor-ligated RNA was mixed with 4 μl 5x RT buffer (100 mM Tris-HCl pH 7.5, 750 mM KCl (or LiCl), 15 mM MgCl2), 1 μl 10 mM dNTP mix, 2 μl 10 μM phosphorylated iSp18 PE Reverse oligo and 0.1 ul SUPERase•In RNase Inhibitor. The mixture was incubated in PCR machine at 80 °C for 2 min and then 42 °C for 5min, 1 μl 0.1M DTT and 1 μl SuperScript III reverse transcriptase (Invitrogen, 18080085) were added to the mixture while PCR tubes were in the thermocycling block. After incubation at 42 °C for 10 min for cDNA synthesis, samples were moved on ice and then 2 μl 1 M NaOH was added to the reaction mixture. To hydrolyze RNA, samples were treated at 98°C for 15 minutes, and 2 μl 1 M HCl was applied to neutralize the solution. Resulting cDNAs were ethanol precipitated and urea gel fractionated, smearing band appearing at the 78 −105 nt length was purified from the gel. cDNA samples are circularized using 50 U CircLigase (Epicentre, CL4111K) at 60°C for 12 hours. Circularized cDNA was amplified using RT stop amp primer (Table S7) and KAPA HiFi PCR master mix (Roche, KK2601) for 8–11 cycles of PCR. Second size selection was performed with native TBE- 8% acrylamide gel fractionation, PCR bands at 78– 105 bp are purified from the gel. 4 – 5 cycles of additional PCR reactions were run with NEBnext multiplex oligos (NEB, E7335) to introduce Illumina library sequences to the RT-stop profile library. ZR small-RNA PAGE Recovery Kit (Zymo Research, R1070) and SYBR Gold (Invitrogen, S11494) staining were used for gel purification during the preparation process. Sequencing reads were obtained from single-end 150 cycles and 110 G output HiSeq 2000 or novaseq illunima sequencing platforms.

DMS-seq read preprocessing

3’ end adaptor sequences (-a AGATCGGAAGAGC) are trimmed from raw data files using Cutadapt (https://doi.org/10.14806/ej.17.1.200) followed by duplicate removal by Prinseq 73. 6 nt molecular barcodes were further clipped away with -g ^NNNNN argument of Cutadapt. RT-stop reads out of the range of 18– 45 nt length were excluded from the final fastq file using prinseq -max_len 45 -min_len 18 command.

rG4 RT-stop profiling

Our study used K+ dependent RT-Stop frequency rather than fold enrichment in the previous paper 17. The RT-Stop frequency used the number of accumulated reads at the designated nucleotide for normalization while fold enrichment was calculated with the background read density, which was the average number of reads in the same transcript. The local read density could vary at the distant part of the transcript, and that would be further exacerbated in DMS-treated samples due to the different structural profiles. Therefore, RT-stop frequency more accurately reflects the real read depth at the nucleotide potation than the fold enrichment. Furthermore, Li+ control was included in DMS-treated samples as a negative control for better measurement of DMS sensitivity in K+ dependent RT-Stops

For RT-stop profile, processed sequencing reads were aligned with STAR aligner 64 against GRCm38.p6 genome assembly. 1 mismatch was allowed for the read mapping. The subsequent sam files were converted to sorted bam files using samtools 63, RT-stops at the rG4 regions were counted using the rG4’seeker pipeline 36. In vitro rG4 identification was computed from RT-stop counting of two biological replicates. Unknown and G≥40% classes are excluded from the assay due to their absence of rG4 motifs. In vivo rG4 DMS sensitivity was addressed by analyzing K+ dependent RT-stop frequency changes of DMS-treated samples. rG4’s showing binominal probability less than 0.05 in both biological replicates were identified as reference rG4 regions. Among reference rG4’s, FDR-adjusted p-values less than 0.1 in both replicates were indicated as in vivo rG4. Numbers of rG4 sites filtered on each procedure were summarized in Fig. 1C. As for rG4 sites accumulating 0 RT-stop in background Li+ samples, pseudocount +1 was applied for probability calculations. In vivo rG4 folding score was calculated according to the previous study including in vivo Li+ sample as background control.

s=f(DMS+;150mMK)f(DMS+;150mMLi)f(DMS;150mMK)f(DMS;150mMLi)

rG4 folding changes (dFSR) of indicated biological samples were presented as the in vivo rG4 folding score changes normalized by reference rG4 folding scores. Reference rG4’s are the significant in vivo rG4 sites found in indicated samples (day 0, 3, and day 7 for fig. 1G, Table S4S6).

s=rG4foldingscore(invivosamples)rG4foldingscore(control)rG4foldingscore(reference)

We categorized unfolded (dFSR < −0.5), insensitive (−0.5 < dFSR < 0.5) and folding gain (dFSR > 0.5) in vivo rG4 groups which of dFSR values fall into the indicated range. rG4 motif density across the gene body plus 1 kb flanking regions were generated by deeptools 68.

Statistical tests

Lists of in vitro rG4 regions were obtained by introducing non-DMS treated K+ and Li+ RT-stop sequencing files to the rG4 seeker program. The binomial test and FDR correction for in vitro rG4’s were conducted by calculations of rG4 seeker. For the P-value-adjusted in vitro rG4 list, we consolidated the FDR-adjusted in vitro rG4 lists from day 0, 3, and 7 samples. Subsequently, we applied the binomial test again, considering the K+-dependent RT-stop frequency for each time point. In vivo DMS sensitivity of rG4 regions was addressed by binominal test using the BINOM.DIST function of Microsoft Excel. RT-stop frequency in Li+ samples was applied as the probability of success to calculate the statistical significance (p < 0.05 for reference and FDR <0.1 for in vivo rG4 sites) of the RT-stop frequency of K+ samples. As for rG4 sites accumulating 0 RT-stop in background Li+ samples, pseudocount +1 was applied for probability calculations. FDR-adjusted p-values of in vivo rG4’s were obtained by providing raw p-values of binominal test to the R module FDRestimation 78. The Mann-Whitney test and the Wilcoxon test were carried out to calculate the significance of the rG4 folding scores as indicated. GraphPad prism9 was used for calculations. The statistical values of the differential gene expression and gene ontology enrichment were provided from its analysis programs Cufflinks and the Panther gene analysis tool.

Gene ontology (GO) analysis

To identify GO terms enriched in in vivo rG4 genes over background predicted rG4 genes, we utilized the online Panther gene list analysis tool (http://www.pantherdb.org/) under default settings for Mus musculus, using a Fisher statistical overrepresentation test and FDR correction for multiple testing (p < 0.05, FDR corrected). All day3 and day7 in vivo rG4 genes (n=72 and 54) were introduced, whereas the background genes (n=18941; Genes showing predicted rG4 motif with >20 reads) were utilized as a reference set. Per each gene group, we analyzed the percentage of genes in each biological process category and the fold enrichment of each GO term in the rG4 genes over the background.

H3K27me3 IF and Xist RNA FISH

H3K27me3 IF combined with Xist FISH was performed as previously described 38,49. Cells were trypsinized and spun down on the slide glass at 1000 rpm for 3 min. Subsequent samples were submerged in ice-cold CSK buffer for 7 min, followed by fixation with 4% formaldehyde-containing PBS buffer. After blocking the slides in 1% BSA/PBS/0.1% Tween-20, solution 1:500 diluted H3K27me3 antibody (Active Motif, AM39535) was applied on the cytospined sample, washed with PBS/ 0.1% Tween-20, and followed by staining Alexa Flour 488 conjugated secondary antibody. Immunostained samples were fixed in 4% paraformaldehyde for 10 min and dried in the sequential incubation with 70 to 100% increasing ethanol solutions. 40 ng of pSX9–3 plasmid (Xist probe) were labeled with Cy3-dUTP using nick translation kit, added to the in-situ hybridization mix (25% formamide, 2x SSC, 10% dextran sulfate, 0.1 mg/mL mouse Cot-1 DNA and 0.1 mg/mL salmon sperm DNA). Hybridization was performed by incubating samples in a humidified chamber at 37°C for overnight. Samples were washed in 25% formamide/2x SSC at 42°C for 20 min and three times in 2x SSC at 37°C for 5 min each, then mounted with Vectashield mounting media with DAPI (H-1200, Fisher Scientific). Images were acquired on Nikon 90i or Zeiss LSM 800 Airyscan confocal microscope. Image analysis was performed using Image J.

Allele-specific qPCR

qPCR primers were designed based on the 3 nt long nucleotide polymorphism sites on the Msn gene between 129S1/SvJm (mus) and CAST/Eih (cas) genomes (Table S7). Allele specific primers were tested using pure mus and cas DNA before the assays (Fig. S5B).

H3K27me3 ChIP-seq

ChIP-seq was performed as previously described 49. Cells were cross-linked in PBS with 1% formaldehyde at room temp for 20 min with gentle agitation and quenched with 0.125 M glycine for 10 min. Fixed cells were washed twice with PBS, harvested by scraping and stored at −80 °C until use. 10 million cells per ChIP were resuspended with 0.5 ml ice-cold ChIP lysis buffer (50 mM Tris-HCl pH8, 10 mM EDTA pH8, and 1% SDS). Cell extracts were subjected to sonication using the Qsonica apparatus (40% power, 30s on/ 30s off for total sonication time 30 min at 4°C). Sheared chromatin size was examined on the 1% agarose gel electrophoresis and repeated sonication was applied if the fragment is larger than 1 kb marker. 0.1 ml lysate was mixed with 1.2 ml ChIP dilution buffer (20 mM Tris-HCl pH8, 150 mM NaCl, 1% Triton X-100, and 2 mM EDTA pH8), and incubated with 2 μg antibody (H3K27me3, GeneTex GTX60892) for overnight at 4°C. Immunoprotein complex was captured with 20 μL Dynabeads Protein G (Invitrogen, 10003D) for 5 hr incubation. Afterward, beads were washed three times with ChIP wash buffer 1 (20 mM Tris-HCl pH8, 150 mM NaCl, 2mM EDTA pH8, 1% Triton X-100, 0.1% SDS) and once with ChIP wash buffer 2 (20 mM Tris-HCl pH8, 500 mM NaCl, 2mM EDTA pH8, 1% Triton X-100, 0.1% SDS). 100 μL of ChIP elution buffer (1% SDS, 100 mM NaHCO3, 40 ug/ml RNaseA) was added to the washed bead and incubated for 1hr at 37°C to recover bound nucleoproteins, dynabeads were removed by using magnetic stand. Reverse-crosslinking was performed by incubating the supernatant at 65° C overnight. NEBNext Ultra II DNA Kit (NEB, E7645) was used for library preparation by following the product manual. Sequencing reads were obtained from single-end 150 cycles and 110 G output HiSeq 2000 or novaseq illunima sequencing platforms.

H3K27me3 ChIP-seq analysis

Adaptor sequences and PCR duplicates were removed from the raw read files. For Xi (mus) and Xa (cas) allelic analysis, adaptor trimmed total sequencing reads were allele-specifically aligned to 129S1/SvJm (mus) and CAST/Eih (cas) genomes as previously described 79. Allelic alignment generated mapped reads as follows: 16150646 (WT mus rep1), 3377654 (WT cas rep1), 18176384 (3’ cl.5 mus rep1), 3797540 (3’ cl.5 cas rep1), 15294626 (3’ cl.6 mus rep1), 3193004 (3’ cl.6 mus rep1), 29062208 (WT mus rep2), 6081088 (WT cas rep2), 21459680 (3’ cl.5 mus rep2), 4469124 (3’ cl.5 cas rep2), 23165720 (3’ cl.6 mus rep2), and 4824076 (3’ cl.6 cas rep2). FPM and input normalized BigWig track files were generated for visualization and further quantification of H3K27me3 density on the different gene groups. To compare H3K27me3 density over the genes, normalized ChIP-seq signal was computed over index (genecode.vM25) defined gene bodies plus and minus 2 kb area using multiBigwigSummary tool from deepTools.

RNA-seq

Total RNA was extracted using TRizol reagent. Oligo dT magnetic beads (NEB) were used for removing abundant rRNA from WT and Xist 3’ rG4 deletion samples. 50 ng of polyA+ RNA was used for RNA-seq library preparation using NEBNext Ultra Directional RNA Library Prep Kit (NEB). Sequencing reads were obtained from single-end 150 cycles and 110 G output HiSeq 2000 or novaseq illunima sequencing platforms.

RNA-seq analysis

Adaptor sequences were trimmed from the RNA-seq reads as described above. Sequencing reads were allele-specifically aligned to 129S1/SvJm (mus) and CAST/Eih (cas) genomes using STAR aligner. Autosomal gene expression was assessed by aligning sequencing reads to the reference mouse genome. FPKM normalized gene expression level was quantified using the program Cufflinks 74. GraphPad Prism was used for generating cumulative gene fraction and violin plots, and statistical analysis.

Southern blot

Genomic DNA was obtained from the confluent ES cell culture. Feeder depleted cells were collected by centrifugation at 400 g for 5 min, and resulting pellet was resuspended in the 1ml lysis buffer (4M Guanidine thiocyanate, 25mM Sodium citrate, and 0.5% sarcosyl). Total nucleic acids were extracted with phenol:chloroform:isoamyl alcohol (25:24:1 saturated with 10 mM Tris-Cl pH 8.0, 1 mM EDTA), followed by isopropanol precipitation and centrifugation at 14,000 rpm for 20 min at 4 °C. 20 ug of genomic DNA was digested overnight with 100 units of BamH1 and Xba in 200 ul volume. Restriction enzyme digested DNA fragments were fractionated with 1% agarose gel. After electrophoresis, gel was socked with 0.2 N HCl for 10 min, washed in deionized water, neutralized with 0.4 N NaOH, and transferred to Hybond-XL membrane (GE Healthcare). PCR amplified Xist fragment (primer pairs indicated in the Table S7) was labelled with 32P dCTP, hybridization was performed by incubating the membrane with radioactive probe in the Church buffer (7% SDS, 158 mM NaH2PO4, 342 mM Na2HPO4, 1% BSA, and 1 mM EDTA) overnight at 65°C. Membrane was sequentially washed with 2x SSC/0.1% SSC and 1x SSC/0.1% SDS. Radiolabeled signal was detected using Typhoon FLA 9000 imager (GE Healthcare).

qPCR-based rG4 probing

DMS treated RNA samples were purified using Trizol reagent. 3 μg of total RNA was mixed with 4 μl 5x RT buffer (100 mM Tris-HCl pH 7.5, 750 mM KCl (or LiCl), 15 mM MgCl2, 1 μl 10 mM dNTP mix, 2 μl 10 μM PE7 or PE8 Reverse oligo and 0.1 ul SUPERase•In RNase Inhibitor). The mixture was incubated in PCR machine at 80 °C for 2 min and then 42 °C for 5min, 1 μl 0.1M DTT and 1 μl SuperScript III reverse transcriptase (Invitrogen, 18080085) was added to the mixture while PCR tubes are in the thermocycling block. After incubation at 42 °C for 10 min for cDNA synthesis, samples are moved on ice and then 2 μl 1 N NaOH was added to the reaction mixture. To hydrolyze RNA, samples were treated at 98°C for 15 minutes, and 2 μl 1 M HCl was applied to neutralize the solution. The reaction product was recovered by ethanol propitiation. qPCR signal from primer pairs targeting upstream of rG4 motif (AF rG4 F1/PE7 or E7 rG4 F1/PE8) was normalized the signal from rG4 downstream primers (AF rG4 F2/PE7 or E7 rG4 F2/PE8) to quantify rG4 dependent reverse transcriptase stops (Table S7). For mutant rG4 motif probing, qPCR was performed using primer pairs specific for mutant sequences; upstream of rG4 motif (AF mut F/Xist PE7, E7 mut F/Xist PE8 or 3 rG4 F1/3 rG4 PE) and downstream primers (AF rG4 F2/Xist PE7, E7 rG4 F2/Xist PE8 or 3 rG4 F2/3 rG4 PE).

CHART

CHART-seq was performed as previously described 43. 25 million cells were crosslinked in 1X PBS with 1% formaldehyde for 10min at room temperature. The formaldehyde was quenched with the addition of 0.125M glycine and rotation for 5min. Cells were then washed three times with 1x PBS 0.05% Tween-20 and cell pellets were snap freeze in liquid nitrogen. Cell pellets were thawed on ice with 1 ml sucrose buffer (10mM HEPES pH7.5, 0.3M sucrose, 1% Triton X-100, 100mM potassium acetate, 0.1mM EGTA, supplemented with 0.5mM spermidine, 0.15mM spermine, 1mM DTT, 1X Protease Inhibitor Cocktail (11873580001, Sigma), 10U/mL SUPERaseIN RNase Inhibitor (AM2694, Thermo Fisher Scientific)). Resuspended samples were then homogenized with dounce homogenizer (357544, Wheaton) in a volume of 3 ml sucrose buffer. The mixture was layered on a cushion of 7.5ml glycerol buffer (10mM HEPES pH7.5, 25% glycerol, 1mM EDTA, 0.1mM EGTA, 100mM potassium acetate, freshly supplemented with 0.5mM spermidine, 0.15mM spermine, 1mM DTT, 1X PIC, and 5U/mL SUPERaseIN) and centrifugated for 10min in 4C at 1500g. Pellets were resuspended in 3ml 1X PBS and then 3% formaldehyde was added, second fixation was carried out with rotation for 30 min at room temperature. Next, the samples were centrifugated at 1000g for 5min at 4 °C and washed 3 times with ice-cold 1x PBS 0.05% Tween-20. The pellets were resuspended in 1 ml nuclear extraction buffer (50mM HEPES pH 7.5, 250mM NaCl, 0.1mM EGTA, 0.5% N-lauroylsarcosine, 0.1% sodium deoxycholate, 5mM DTT, 10U/mL SUPERaseIN), rotated 10 minutes at 4°C, spinned-down at 400g for 5 minutes at 4C and nuclei were resuspended in 130μl sonication buffer (50mM HEPES pH 7.5, 75mM NaCl, 0.1mM EGTA, 0.5% N-lauroylsarcosine, 0.1% sodium deoxycholate, 0.1% SDS, 5mM DTT, 10U/mL SUPERaseIN). Sonication was performed with Covaris E220 sonicator at power: 140W, duty factor: 10%, cycles per burst: 200 for 5 minutes at 4C. The lysate was pre-cleared with addition of Dynabeads MyOne Streptavidin C1 beads (Invitrogen) in 640μl 2X hybridization buffer (50mM Tris-HCl pH 7.0, 750mM NaCl, 1% SDS, 1mM EDTA, 15% formamide, 1mM DTT, 1mM PMSF, 1X PIC, and 100U/mL SUPERaseIN). Pre-cleared mixtures received 3.6μl from a 10μM mix of Xist probes and hybridization was performed at room temperature in rotation overnight. Next, 120μl of washed beads were added for each reaction and the tubes were incubated in rotation at 37C for 1 hour. Then beads were washed once with 1X hybridization buffer (33% sonication buffer, 67% 2X hybridization buffer) for 10 min, 5 times with 2% SDS buffer (10mM HEPES pH 7.6, 250mM NaCl, 2% SDS, 2mM EDTA, 2mM EGTA, and 1mM DTT) for 5 min and twice with 0.5% NP40 buffer (10mM HEPES pH7.6, 250mM NaCl, 0.5% NP40, 3mM MgCl2, and 10mM DTT) for 5min, at 37C. Beads were resuspended with 180μl of 0.5% NP40 buffer and 20μl RNase H (5U/μl, NEB) for 20 minutes in room temperature to elute the DNA twice. Eluted DNA was incubated with RNaseA for 1h at 37C in rotation, following by Proteinase K (20mg/ml) treatment at 55°C 1hr. To reverse crosslinking the DNA, 300 mM NaCl was added to the samples and incubated at 65°C overnight. DNA was extracted with phenol-chloroform extraction. CHART libraries were than prepared with NEBNext Ultra II DNA Library Prep Kit for Illumina (#E7645S). Libraries were sequenced using ilumina novaseq generating 150bp pair-end reads. Alignment to mouse reference genome was performed using Bowtie2.

Formaldehyde RNA immunoprecipitation (fRIP)

fRIP analysis was performed as previously described 47. Day7 differentiating ES cells were cross-linked in PBS with 0.1% formaldehyde at room temp for 10 min with gentle agitation and quenched with 0.125 M glycine for 10 min. Fixed cells were washed twice with PBS, harvested by scraping and stored at −80 °C until use. Frozen pellets are resuspended in 1 ml of RIPA lysis buffer (50 mM Tris (pH 8), 150 mM KCl, 0.1 % SDS, 1 % Triton-X, 5 mM EDTA, 0.5 % sodium deoxycholate, 0.5 mM DTT, protease inhibitor cocktail (11836153001, Roche) and 100 U/ml RNase Inhibitor (AM2696, Invitrogen)). Lysed cell extract was processed by sonication using the Qsonica apparatus (40% power, 30s on/ 30s off for total sonication time 5 min at 4°C). The remaining cell debris was removed by taking supernatant after spinning down cell lysate 13000 rpm for 10 min at 4°C. We diluted supernatant by adding equal volume of fRIP binding/wash buffer (150 mM KCl, 25 mM Tris (pH 7.5), 5 mM EDTA, 0.5 % NP-40, 0.5 mM DTT, protease inhibitor cocktail and 100 U/ml RNase Inhibitor). Pre-clearing was performed by adding 50 μl of Dynabeads Protein A (10001D, Invitrogen) to the diluted mixture and incubating 4 hours at 4°C. 10% of pre-cleared samples were stored separately for input loading control. 5 μg of rabbit EZH2 antibody (5246S, CELL SIGNALING TECHNOLOGY), SUZ12 antibody (ACTIVE MOTIF, 39057), or rabbit IgG ( )was applied to the pre-cleared mixture and incubated at 4°C overnight. Next day, antibody bound RNA fractions were recovered by incubating the solution with 50 μl of Dynabeads Protein A for 4 hours at 4°C. Beads were washed with fRIP binding/wash buffer 4 times. Antibody bound bead and input samples were incubated in 100 μL of 3× reverse-crosslinking buffer (3× PBS, 6 % N-lauroyl sarcosine, 30 mM EDTA, 15 mM DTT, 20mg/ml Proteinase K) for 1 hour at 42 °C, then another 1 hour at 55 °C. RNA isolation was performed by TriZol extraction and following ethanol precipitation with 2.5 volumes of ethanol and 20 μg of glycogen. To remove DNA contamination, eluted RNA samples were treated with 2 units of TURBO DNase (AM2238, Invitrogen) for 1 hour at 37 °C.

UV crosslinking RNA Immunoprecipitation was performed as described 48. Approximately 1 × 107 cells suspended in PBS were irradiated at 254 nm using a dose of 400 mJ/cm2 of UV light (Stratagene Stratalinker). Post-UV crosslinking, cells were processed similarly to the fRIP procedure, except for the washing steps. Two consecutive bead washes were performed with high salt buffer (1x PBS, 750 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS), low salt buffer (1x PBS, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS), and fRIP binding/wash buffer, each for 5 minutes at room temperature.

qRT-PCR analysis was performed following standard protocol using designated primer pairs (Table S7); AF rG4 (AF rG4 F2-Xist PE7), E7 rG4 (E7 rG4 F2-Xist PE8), 3’ rG4 (3 rG4 F2–3 rG4 PE), Rps27 (RPS27 qRT F-RPS27 qRT R), and S100a10 (S100a10 qRT F-S100a10 qRT R).

Protein purification

Mouse PRC2 proteins were purified from SF9 cells as previously described 8. Purified bacmids containing coding sequences of FLAG-tagged EZH2 and untagged SUZ12, and EED, and RBPA48 were transfected to the SF9 cells using Cellfectin II reagent (Gibco, 10362100) to obtain P1 baculovirus stocks. Baculovirus stocks were further amplified to passage 3 and the virus titer was determined by plaque-forming assay. The resulting P3 virus stocks showed ~10 to 20 X 107 pfu/ml. PRC2 protein expression was performed by infecting SF9 cells, 1 MOI of the virus was added to the 1 × 106/ml SF9 cells in 200 ml culture, incubated for 3 days at 25 °C, and harvested with centrifugation. Each PRC2 component was infected with the individual SF9 culture and combined into a single batch to make the complex during the purification step.

Protein extract was prepared by mixing with 20 ml cell lysis buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 1% NP-40, 0.3% Triton-X, 4 mM EDTA, 1 mM MgCl2, 1 mM DTT, and 20% Glycerol and complete protease inhibitors (Roche)). Cell lysates were incubated with 1 ml M2 anti-FLAG magnetic beads (Sigma, M8823) at 4 °C for 12 hours, beads were washed with 10 ml cell lysis buffer four times. The PRC2 complex was eluted by incubating the beads with the 500 ul elution buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 0.1% Triton-X, 20% Glycerol, and 0.2 μg/ml 3X-FLAG peptide (Sigma, F4799)) at room temperature for 30 min, elution was repeated for three times to maximize the yield. Eluted protein was filtered with Amicon ultra 100 kDa MWCO concentrators (Millipore, UFC8100) to remove free proteins and concentrate the assembled PRC2 complex. Protein concentration was measured using the Bradford protein assay (Bio-Rad) and protein bands in coomassie blue-stained gels were compared with BSA reference samples to confirm the concentration and purity of the proteins.

To obtain the PRC2 HMT assay substrate, histone H3 DNA fragment comprising the amino acid sequence [APRKQLATKAARKSAPSTGGVKKPHRYRP] was cloned into the pGEX-4T1 plasmid. BL21 cells were transformed with the plasmids for the recombinant protein expression. Cells were grown in the LB media at 37 °C until O.D reached to 0.4, and 100 μM IPTG (Sigma-Aldrich) was added for initiating GST protein expression. Cells were incubated for 5 hours at 25 °C to accumulate sufficient amount of GST-H3 recombinant protein in bacterial cells. Cells were collected by centrifugation and lysed by sonication in GST binding buffer (50 mM potassium phosphate buffer pH 7, 5 mM EDTA, 100 mM NaCl, 1% Triton X-100, and 0.15 mM PMSF). After removing cell debris by centrifugation, cell lysate was incubated with Glutathione Sepharose 4B beads (GE Healthcare) for 12 hours at 4 °C. GST protein-bound beads were washed with GST binding buffer 4 times, and elution was performed by incubating the bead with 200 μl of GST elution buffer (50 mM Tris-HCl buffer pH 7, 5 mM EDTA, 100 mM NaCl, 5% glycerol, 25 mM glutathione) for 15 min at room temperature. Protein concentration was quantified by Bradford protein assay (Bio-Rad) and Coomassie blue protein gel staining.

Electromobility shift assay

For rG4 RNA EMSA experiments, single-stranded RNA probes are synthesized using T7 high yield RNA synthesis kit (NEB, E2040S). In vitro transcription templates are prepared by PCR amplifying rG4 motifs with flanking sequence primer pairs including T7 promoter sequence at the 5’ end (Table S7). RNA fragments are radiolabeled with T4 polynucleotide kinase (NEB, M0201S) and γ32P-ATP (PERKINELMER, BLU502Z250UC) followed by purification using the Oligo Clean & Concentrator Kit (Zymo Research, D4060). RNA probes were resuspended in the 10 ul of 2X binding buffer (100 mM Tris-HCl pH 7.5, 5 mM MgCl2, 0.2% NP40, 50 ng yeast tRNA (Ambion Cat# AM7119), and 200 mM KCl (or LiCl)), denatured for 2 min at 80 °C and pre-folded by decreasing temperature to 25 °C at 0.1 °C per second rate. 10 ul of PRC2 was applied as indicated concentrations, assembled binding reaction mixture was incubated at room temperature for 20 min. 4 μl of 6X DNA loading buffer was applied to the samples and RNA protein complex was gel fractionated with 0.8% agarose gel in THEM buffer (66mM HEPES, 34mM Tris, 0.1mM disodium EDTA and 10mM MgCl2). Electrophoresis was performed for 30 min at 120 v volts at 4 °C. THEM buffer was prechilled and gel running was performed on ice to avoid excess heating. Gels were dried and exposed to the phosphoscreen for phosphor imaging.

Histone methyltransferase assay

Histone methyl transferase assay was performed using MTase-Glo Methyltransferase Assay kit (Promega, V7601) following the manufacturer’s instruction. 150 nM of PRC2 histone methyltransferase was incubated with 4 uM of GST fused H3 peptide substrate and 100 uM of S-adenosyl homocysteine (SAM) in HMT buffer (50 mM Tris-HCL pH 8.0, 5 mM MgCl, and 50 mM KCl (or LiCl)) for 3 hours at 25 °C. rG4 RNAs were pre-folded in the HMT buffer by heating at 80 °C for 2 min and gradually cooling down to 25 °C. The indicated amount of rG4 RNA was applied to the HMT reaction before adding SAM. After the incubation, the MTase-Glo reagent was added to the reaction mixture, incubated for 30 min at 25 °C, and then the MTase-Glo detection solution was added to initiate the luciferase reaction. HMT chemiluminescent signal was measured with a microplate reader.

Xist rG4 primer extension assay

Xist RNA templates flanking rG4 sites were in vitro transcribed with PCR amplified Xist fragment derived from pSx9 Xist plasmid. Xist-specific PCR primers contain T7 primer sequences at the 5’ end (Table S7) for supporting RNA synthesis using T7 high-yield RNA synthesis kit (NEB, E2040S). 200 ng of 11 ul templated RNA was mixed with 4 μl 5x RT buffer (100 mM Tris-HCl pH 7.5, 750 mM KCl (or LiCl), and 15 mM MgCl2), 1 μl 10 mM dNTP mix, 0.1 ul SUPERase•In RNase Inhibitor, and 2 μl of γ32P-labeled 1 μM primer extension reverse oligos (Table S7). The RNA templates and reverse primers were incubated in the PCR machine at 80 °C for 2 min and then 42 °C for 5min. 2 ul of 50 uM PhenDC3, BRACO19 or TMPyP4 was applied where indicated and incubated another 42 °C for 5min. For reverse transcription, 0.5 μl 0.1M DTT and 0.5 μl SuperScript III reverse transcriptase (Invitrogen, 18080085) were added to the mixture while reaction tubes were in the thermos block. 20 ul of 2x UREA gel loading buffer was directly added to the reaction mixture and heat treated at 80 °C for 2 min. Reverse transcribed RT products were gel fractionated in denaturing 15% acrylamide TBE-UREA gels, and electrophoresis was performed for 2 hours at 500 V. Radio-labeled decade marker (Ambion, AM7778) was included in the gel electrophoresis for the length comparisons. Gels were exposed to the phosphoscreen for 12 hours for phosphor imaging. Image J was used for the primer extension image analysis.

Supplementary Material

MMC6

Table S5. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 0 EB samples. Related to Figure 1.

MMC8

Table S7. List of primer sequences used for DMS-seq library preparation, in vitro assays and quantitative qPCR. Related to STAR Methods.

MMC7

Table S6. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 3 EB samples. Related to Figure 1.

MMC5

Table S4. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 7 EB samples. Related to Figure 1.

MMC4

Table S3. List of not predicted in vitro rG4 sites identified by K+ dependent RT-stop profiling on non-DMS-treated day 7 EB samples. Related to Figure 1.

MMC2

Table S1. Number of mapped reads in the d-rG4-seq libraries. Related to Figure 1.

MMC3

Table S2. List of in vitro rG4 sites identified by K+ dependent RT-stop profiling on non-DMS-treated day 7 EB samples. Related to Figure 1.

MMC1

HIGHLIGHTS.

  • RNA G-quadruplex (rG4) folding is tightly regulated and developmentally programmed.

  • Folded rG4’s avidly bind PRC2 and inhibit its histone methyltransferase activity.

  • rG4 loss in Xist causes PRC2 hyperactivity on the inactive X chromosome.

  • Xist-PRC2 complexes become spatially entrapped, disrupting X-inactivation.

ACKNOWLEDGEMENTS

We thank all members of the lab for stimulating discussions. This work was funded by a grant from the National Institutes of Health, R01-HD097665, to JTL.

DECLARATION OF INTERESTS

JTL is a cofounder of Fulcrum Therapeutics, a Scientific Advisor to Skyhawk Therapeutics, and a Non-Executive Director of GSK.

Footnotes

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

  • 1.Schwartz YB, and Pirrotta V (2007). Polycomb silencing mechanisms and the management of genomic programmes. Nat Rev Genet 8, 9–22. 10.1038/nrg1981. [DOI] [PubMed] [Google Scholar]
  • 2.Zhao J, Sun BK, Erwin JA, Song JJ, and Lee JT (2008). Polycomb proteins targeted by a short repeat RNA to the mouse X chromosome. Science 322, 750–756. 10.1126/science.1163045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Kanhere A, Viiri K, Araujo CC, Rasaiyaah J, Bouwman RD, Whyte WA, Pereira CF, Brookes E, Walker K, Bell GW, et al. (2010). Short RNAs are transcribed from repressed polycomb target genes and interact with polycomb repressive complex-2. Mol Cell 38, 675–688. 10.1016/j.molcel.2010.03.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Pandey RR, Mondal T, Mohammad F, Enroth S, Redrup L, Komorowski J, Nagano T, Mancini-Dinardo D, and Kanduri C (2008). Kcnq1ot1 antisense noncoding RNA mediates lineage-specific transcriptional silencing through chromatin-level regulation. Mol Cell 32, 232–246. 10.1016/j.molcel.2008.08.022. [DOI] [PubMed] [Google Scholar]
  • 5.Zhao J, Ohsumi TK, Kung JT, Ogawa Y, Grau DJ, Sarma K, Song JJ, Kingston RE, Borowsky M, and Lee JT (2010). Genome-wide identification of polycomb-associated RNAs by RIP-seq. Mol Cell 40, 939–953. 10.1016/j.molcel.2010.12.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Long Y, Hwang T, Gooding AR, Goodrich KJ, Rinn JL, and Cech TR (2020). RNA is essential for PRC2 chromatin occupancy and function in human pluripotent stem cells. Nat Genet 52, 931–938. 10.1038/s41588-020-0662-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zhang Q, McKenzie NJ, Warneford-Thomson R, Gail EH, Flanigan SF, Owen BM, Lauman R, Levina V, Garcia BA, Schittenhelm RB, et al. (2019). RNA exploits an exposed regulatory site to inhibit the enzymatic activity of PRC2. Nat Struct Mol Biol 26, 237–247. 10.1038/s41594-019-0197-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Cifuentes-Rojas C, Hernandez AJ, Sarma K, and Lee JT (2014). Regulatory interactions between RNA and polycomb repressive complex 2. Mol Cell 55, 171–185. 10.1016/j.molcel.2014.05.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Rosenberg M, Blum R, Kesner B, Aeby E, Garant JM, Szanto A, and Lee JT (2021). Motif-driven interactions between RNA and PRC2 are rheostats that regulate transcription elongation. Nat Struct Mol Biol 28, 103–117. 10.1038/s41594-020-00535-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Wang X, Goodrich KJ, Gooding AR, Naeem H, Archer S, Paucek RD, Youmans DT, Cech TR, and Davidovich C (2017). Targeting of Polycomb Repressive Complex 2 to RNA by Short Repeats of Consecutive Guanines. Mol Cell 65, 1056–1067 e1055. 10.1016/j.molcel.2017.02.003. [DOI] [PubMed] [Google Scholar]
  • 11.Kaneko S, Bonasio R, Saldana-Meyer R, Yoshida T, Son J, Nishino K, Umezawa A, and Reinberg D (2014). Interactions between JARID2 and noncoding RNAs regulate PRC2 recruitment to chromatin. Mol Cell 53, 290–300. 10.1016/j.molcel.2013.11.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Beltran M, Tavares M, Justin N, Khandelwal G, Ambrose J, Foster BM, Worlock KB, Tvardovskiy A, Kunzelmann S, Herrero J, et al. (2019). G-tract RNA removes Polycomb repressive complex 2 from genes. Nat Struct Mol Biol 26, 899–909. 10.1038/s41594-019-0293-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Bochman ML, Paeschke K, and Zakian VA (2012). DNA secondary structures: stability and function of G-quadruplex structures. Nat Rev Genet 13, 770–780. 10.1038/nrg3296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Rhodes D, and Lipps HJ (2015). G-quadruplexes and their regulatory roles in biology. Nucleic Acids Res 43, 8627–8637. 10.1093/nar/gkv862. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Dumas L, Herviou P, Dassi E, Cammas A, and Millevoi S (2021). G-Quadruplexes in RNA Biology: Recent Advances and Future Directions. Trends Biochem Sci 46, 270–283. 10.1016/j.tibs.2020.11.001. [DOI] [PubMed] [Google Scholar]
  • 16.Lyu K, Chow EY, Mou X, Chan TF, and Kwok CK (2021). RNA G-quadruplexes (rG4s): genomics and biological functions. Nucleic Acids Res 49, 5426–5450. 10.1093/nar/gkab187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Guo JU, and Bartel DP (2016). RNA G-quadruplexes are globally unfolded in eukaryotic cells and depleted in bacteria. Science 353. 10.1126/science.aaf5371. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Yang X, Cheema J, Zhang Y, Deng H, Duncan S, Umar MI, Zhao J, Liu Q, Cao X, Kwok CK, and Ding Y (2020). RNA G-quadruplex structures exist and function in vivo in plants. Genome Biol 21, 226. 10.1186/s13059-020-02142-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Liu H, Chu Z, and Yang X (2022). A Key Molecular Regulator, RNA G-Quadruplex and Its Function in Plants. Front Plant Sci 13, 926953. 10.3389/fpls.2022.926953. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Cho H, Cho HS, Nam H, Jo H, Yoon J, Park C, Dang TVT, Kim E, Jeong J, Park S, et al. (2018). Translational control of phloem development by RNA G-quadruplex-JULGI determines plant sink strength. Nat Plants 4, 376–390. 10.1038/s41477-018-0157-2. [DOI] [PubMed] [Google Scholar]
  • 21.Yang X, Yu H, Duncan S, Zhang Y, Cheema J, Liu H, Benjamin Miller J, Zhang J, Kwok CK, Zhang H, and Ding Y (2022). RNA G-quadruplex structure contributes to cold adaptation in plants. Nat Commun 13, 6224. 10.1038/s41467-022-34040-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Xu Y, Suzuki Y, Ito K, and Komiyama M (2010). Telomeric repeat-containing RNA structure in living cells. Proc Natl Acad Sci U S A 107, 14579–14584. 10.1073/pnas.1001177107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Laguerre A, Hukezalie K, Winckler P, Katranji F, Chanteloup G, Pirrotta M, Perrier-Cornet JM, Wong JM, and Monchaud D (2015). Visualization of RNA-Quadruplexes in Live Cells. J Am Chem Soc 137, 8521–8525. 10.1021/jacs.5b03413. [DOI] [PubMed] [Google Scholar]
  • 24.Cammas A, and Millevoi S (2017). RNA G-quadruplexes: emerging mechanisms in disease. Nucleic Acids Res 45, 1584–1595. 10.1093/nar/gkw1280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Simone R, Fratta P, Neidle S, Parkinson GN, and Isaacs AM (2015). G-quadruplexes: Emerging roles in neurodegenerative diseases and the non-coding transcriptome. FEBS Lett 589, 1653–1668. 10.1016/j.febslet.2015.05.003. [DOI] [PubMed] [Google Scholar]
  • 26.Meier-Stephenson V, Badmalia MD, Mrozowich T, Lau KCK, Schultz SK, G-emmill DL, Osiowy C, van Marle G, Coffin CS, and Patel TR (2021). Identification and characterization of a G-quadruplex structure in the pre-core promoter region of hepatitis B virus covalently closed circular DNA. J Biol Chem 296, 100589. 10.1016/j.jbc.2021.100589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Zhao C, Qin G, Niu J, Wang Z, Wang C, Ren J, and Qu X (2021). Targeting RNA GQuadruplex in SARS-CoV-2: A Promising Therapeutic Target for COVID-19? Angew Chem Int Ed Engl 60, 432–438. 10.1002/anie.202011419. [DOI] [PubMed] [Google Scholar]
  • 28.Marcel V, Tran PL, Sagne C, Martel-Planche G, Vaslin L, Teulade-Fichou MP, Hall J, Mergny JL, Hainaut P, and Van Dyck E (2011). G-quadruplex structures in TP53 intron 3: role in alternative splicing and in production of p53 mRNA isoforms. Carcinogenesis 32, 271–278. 10.1093/carcin/bgq253. [DOI] [PubMed] [Google Scholar]
  • 29.Weldon C, Dacanay JG, Gokhale V, Boddupally PVL, Behm-Ansmant I, Burley GA, Branlant C, Hurley LH, Dominguez C, and Eperon IC (2018). Specific G-quadruplex ligands modulate the alternative splicing of Bcl-X. Nucleic Acids Res 46, 886–896. 10.1093/nar/gkx1122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kumari S, Bugaut A, Huppert JL, and Balasubramanian S (2007). An RNA G-quadruplex in the 5’ UTR of the NRAS proto-oncogene modulates translation. Nat Chem Biol 3, 218–221. 10.1038/nchembio864. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Ajit K, Alagia A, Burger K, and Gullerova M (2024). Tyrosine 1-phosphorylated RNA polymerase II transcribes PROMPTs to facilitate proximal promoter pausing and induce global transcriptional repression in response to DNA damage. Genome Res 34, 201–216. 10.1101/gr.278644.123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Song J, Gooding AR, Hemphill WO, Love BD, Robertson A, Yao L, Zon LI, North TE, Kasinath V, and Cech TR (2023). Structural basis for inactivation of PRC2 by G-quadruplex RNA. Science 381, 1331–1337. 10.1126/science.adh0059. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Wuarin J, and Schibler U (1994). Physical isolation of nascent RNA chains transcribed by RNA polymerase II: evidence for cotranscriptional splicing. Mol Cell Biol 14, 7219–7225. 10.1128/mcb.14.11.7219-7225.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Pandya-Jones A, Bhatt DM, Lin CH, Tong AJ, Smale ST, and Black DL (2013). Splicing kinetics and transcript release from the chromatin compartment limit the rate of Lipid A-induced gene expression. RNA 19, 811–827. 10.1261/rna.039081.113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Kwok CK, Marsico G, Sahakyan AB, Chambers VS, and Balasubramanian S (2016). rG4-seq reveals widespread formation of G-quadruplex structures in the human transcriptome. Nat Methods 13, 841–844. 10.1038/nmeth.3965. [DOI] [PubMed] [Google Scholar]
  • 36.Chow EY, Lyu K, Kwok CK, and Chan TF (2020). rG4-seeker enables high-confidence identification of novel and non-canonical rG4 motifs from rG4-seq experiments. RNA Biol 17, 903–917. 10.1080/15476286.2020.1740470. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Eddy J, Vallur AC, Varma S, Liu H, Reinhold WC, Pommier Y, and Maizels N (2011). G4 motifs correlate with promoter-proximal transcriptional pausing in human genes. Nucleic Acids Res 39, 4975–4983. 10.1093/nar/gkr079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Wang CY, Jegu T, Chu HP, Oh HJ, and Lee JT (2018). SMCHD1 Merges Chromosome Compartments and Assists Formation of Super-Structures on the Inactive X. Cell 174, 406–421 e425. 10.1016/j.cell.2018.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Jegu T, Aeby E, and Lee JT (2017). The X chromosome in space. Nat Rev Genet 18, 377–389. 10.1038/nrg.2017.17. [DOI] [PubMed] [Google Scholar]
  • 40.Disteche CM (2016). Dosage compensation of the sex chromosomes and autosomes. Semin Cell Dev Biol 56, 9–18. 10.1016/j.semcdb.2016.04.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Brown CJ, Hendrich BD, Rupert JL, Lafreniere RG, Xing Y, Lawrence J, and Willard HF (1992). The human XIST gene: analysis of a 17 kb inactive X-specific RNA that contains conserved repeats and is highly localized within the nucleus. Cell 71, 527–542. 10.1016/0092-8674(92)90520-m. [DOI] [PubMed] [Google Scholar]
  • 42.Ogawa Y, Sun BK, and Lee JT (2008). Intersection of the RNA interference and X-inactivation pathways. Science 320, 1336–1341. 10.1126/science.1157676. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Simon MD, Pinter SF, Fang R, Sarma K, Rutenberg-Schoenberg M, Bowman SK, Kesner BA, Maier VK, Kingston RE, and Lee JT (2013). High-resolution Xist binding maps reveal two-step spreading during X-chromosome inactivation. Nature 504, 465–469. 10.1038/nature12719. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Davidovich C, and Cech TR (2015). The recruitment of chromatin modifiers by long noncoding RNAs: lessons from PRC2. RNA 21, 2007–2022. 10.1261/rna.053918.115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Hsiao K, Zegzouti H, and Goueli SA (2016). Methyltransferase-Glo: a universal, bioluminescent and homogenous assay for monitoring all classes of methyltransferases. Epigenomics 8, 321–339. 10.2217/epi.15.113. [DOI] [PubMed] [Google Scholar]
  • 46.Frasson I, Pirota V, Richter SN, and Doria F (2022). Multimeric G-quadruplexes: A review on their biological roles and targeting. Int J Biol Macromol 204, 89–102. 10.1016/j.ijbiomac.2022.01.197. [DOI] [PubMed] [Google Scholar]
  • 47.D, G.H., Kelley, D.R., Tenen, D., Bernstein, B., and Rinn, J.L. (2016). Widespread RNA binding by chromatin-associated proteins. Genome Biol 17, 28. 10.1186/s13059-016-0878-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Ule J, Jensen K, Mele A, and Darnell RB (2005). CLIP: a method for identifying protein-RNA interaction sites in living cells. Methods 37, 376–386. 10.1016/j.ymeth.2005.07.018. [DOI] [PubMed] [Google Scholar]
  • 49.Colognori D, Sunwoo H, Kriz AJ, Wang CY, and Lee JT (2019). Xist Deletional Analysis Reveals an Interdependency between Xist RNA and Polycomb Complexes for Spreading along the Inactive X. Mol Cell 74, 101–117 e110. 10.1016/j.molcel.2019.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Pintacuda G, Wei G, Roustan C, Kirmizitas BA, Solcan N, Cerase A, Castello A, Mohammed S, Moindrot B, Nesterova TB, and Brockdorff N (2017). hnRNPK Recruits PCGF3/5-PRC1 to the Xist RNA B-Repeat to Establish Polycomb-Mediated Chromosomal Silencing. Mol Cell 68, 955–969 e910. 10.1016/j.molcel.2017.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Lappala A, Wang CY, Kriz A, Michalk H, Tan K, Lee JT, and Sanbonmatsu KY (2021). Four-dimensional chromosome reconstruction elucidates the spatiotemporal reorganization of the mammalian X chromosome. Proc Natl Acad Sci U S A 118. 10.1073/pnas.2107092118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Kharel P, Becker G, Tsvetkov V, and Ivanov P (2020). Properties and biological impact of RNA G-quadruplexes: from order to turmoil and back. Nucleic Acids Res 48, 12534–12555. 10.1093/nar/gkaa1126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Sunwoo H, Wu JY, and Lee JT (2015). The Xist RNA-PRC2 complex at 20-nm resolution reveals a low Xist stoichiometry and suggests a hit-and-run mechanism in mouse cells. Proc Natl Acad Sci U S A 112, E4216–4225. 10.1073/pnas.1503690112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Davidovich C, Zheng L, Goodrich KJ, and Cech TR (2013). Promiscuous RNA binding by Polycomb repressive complex 2. Nat Struct Mol Biol 20, 1250–1257. 10.1038/nsmb.2679. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Kaneko S, Son J, Shen SS, Reinberg D, and Bonasio R (2013). PRC2 binds active promoters and contacts nascent RNAs in embryonic stem cells. Nat Struct Mol Biol 20, 1258–1264. 10.1038/nsmb.2700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Guo JK, Blanco MR, Walkup W.G.t., Bonesteele G, Urbinati CR, Banerjee AK, Chow A, Ettlin O, Strehle M, Peyda P, et al. (2024). Denaturing purifications demonstrate that PRC2 and other widely reported chromatin proteins do not appear to bind directly to RNA in vivo. Mol Cell. 10.1016/j.molcel.2024.01.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Wei G, Almeida M, Bowness JS, Nesterova TB, and Brockdorff N (2021). Xist Repeats B and C, but not Repeat A, mediate de novo recruitment of the Polycomb system in X chromosome inactivation. Dev Cell 56, 1234–1235. 10.1016/j.devcel.2021.04.009. [DOI] [PubMed] [Google Scholar]
  • 58.Colognori D, Sunwoo H, Wang D, Wang CY, and Lee JT (2021). Xist Repeat A contributes to early recruitment of Polycomb complexes during X-chromosome inactivation. Dev Cell 56, 1236–1237. 10.1016/j.devcel.2021.04.007. [DOI] [PubMed] [Google Scholar]
  • 59.Lane AN, Chaires JB, Gray RD, and Trent JO (2008). Stability and kinetics of Gquadruplex structures. Nucleic Acids Res 36, 5482–5515. 10.1093/nar/gkn517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Cherney RE, Mills CA, Herring LE, Braceros AK, and Calabrese JM (2023). A monoclonal antibody raised against human EZH2 cross-reacts with the RNA-binding protein SAFB. Biol Open 12. 10.1242/bio.059955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, and Zhang F (2013). Genome engineering using the CRISPR-Cas9 system. Nat Protoc 8, 2281–2308. 10.1038/nprot.2013.143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Flemr M, and Buhler M (2015). Single-Step Generation of Conditional Knockout Mouse Embryonic Stem Cells. Cell Rep 12, 709–716. 10.1016/j.celrep.2015.06.051. [DOI] [PubMed] [Google Scholar]
  • 63.Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, and Genome Project Data Processing, S. (2009). The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079. 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, and Gingeras TR (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21. 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Marcel M (2011). Cutadapt Removes Adapter Sequences From High-Throughput Sequencing Reads. EMBnet. J. 17, 10–12. [Google Scholar]
  • 66.Schneider CA, Rasband WS, and Eliceiri KW (2012). NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9, 671–675. 10.1038/nmeth.2089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Quinlan AR, and Hall IM (2010). BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842. 10.1093/bioinformatics/btq033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Ramirez F, Ryan DP, Gruning B, Bhardwaj V, Kilpert F, Richter AS, Heyne S, Dundar F, and Manke T (2016). deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res 44, W160–165. 10.1093/nar/gkw257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357–359. 10.1038/nmeth.1923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Tarasov A, Vilella AJ, Cuppen E, Nijman IJ, and Prins P (2015). Sambamba: fast processing of NGS alignment formats. Bioinformatics 31, 2032–2034. 10.1093/bioinformatics/btv098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 71.Langmead B, Trapnell C, Pop M, and Salzberg SL (2009). Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 10, R25. 10.1186/gb-2009-10-3-r25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.Cock PJ, Antao T, Chang JT, Chapman BA, Cox CJ, Dalke A, Friedberg I, Hamelryck T, Kauff F, Wilczynski B, and de Hoon MJ (2009). Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25, 1422–1423. 10.1093/bioinformatics/btp163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Schmieder R, and Edwards R (2011). Quality control and preprocessing of metagenomic datasets. Bioinformatics 27, 863–864. 10.1093/bioinformatics/btr026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 74.Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, Pimentel H, Salzberg SL, Rinn JL, and Pachter L (2012). Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc 7, 562–578. 10.1038/nprot.2012.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Tack DC, Tang Y, Ritchey LE, Assmann SM, and Bevilacqua PC (2018). StructureFold2: Bringing chemical probing data into the computational fold of RNA structural analysis. Methods 143, 12–15. 10.1016/j.ymeth.2018.01.018. [DOI] [PubMed] [Google Scholar]
  • 76.Aeby E, Lee HG, Lee YW, Kriz A, Del Rosario BC, Oh HJ, Boukhali M, Haas W, and Lee JT (2020). Decapping enzyme 1A breaks X-chromosome symmetry by controlling Tsix elongation and RNA turnover. Nat Cell Biol 22, 1116–1129. 10.1038/s41556-020-0558-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 77.Rouskin S, Zubradt M, Washietl S, Kellis M, and Weissman JS (2014). Genome-wide probing of RNA structure reveals active unfolding of mRNA structures in vivo. Nature 505, 701–705. 10.1038/nature12894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 78.Murray MH, and Blume JD (2021). FDRestimation: Flexible False Discovery Rate Computation in R. F1000Res 10, 441. 10.12688/f1000research.52999.2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 79.Pinter SF, Sadreyev RI, Yildirim E, Jeon Y, Ohsumi TK, Borowsky M, and Lee JT (2012). Spreading of X chromosome inactivation via a hierarchy of defined Polycomb stations. Genome Res 22, 1864–1876. 10.1101/gr.133751.111. [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

MMC6

Table S5. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 0 EB samples. Related to Figure 1.

MMC8

Table S7. List of primer sequences used for DMS-seq library preparation, in vitro assays and quantitative qPCR. Related to STAR Methods.

MMC7

Table S6. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 3 EB samples. Related to Figure 1.

MMC5

Table S4. List of in vivo rG4 sites identified by K+ dependent RT-stop profiling on DMS-treated day 7 EB samples. Related to Figure 1.

MMC4

Table S3. List of not predicted in vitro rG4 sites identified by K+ dependent RT-stop profiling on non-DMS-treated day 7 EB samples. Related to Figure 1.

MMC2

Table S1. Number of mapped reads in the d-rG4-seq libraries. Related to Figure 1.

MMC3

Table S2. List of in vitro rG4 sites identified by K+ dependent RT-stop profiling on non-DMS-treated day 7 EB samples. Related to Figure 1.

MMC1

Data Availability Statement

Raw high-throughput sequencing data and processed files for DMS-seq, RNA-seq, ChIP-seq, and CHART-seq reported in this paper have been deposited at GEO under accession number: GSE219083. These data are publicly available as of the date of publication. This paper analyzes existing, publicly available data. These accession numbers for the datasets are listed in the key resources table.

Key Resources Table

REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
Mouse monoclonal anti-H3K27me3 Active Motif Cat#AM39535
Rabbit monoclonal anti-H3K27me3 GeneTex Cat#GTX60892
Rabbit monoclonal anti-EZH2 Cell Signaling Cat#5246
Rabbit polyclonal anti-SUZ12 Active Motif Cat#39358
Bacterial and virus strains
BL21(DE3) EMD Millipore Cat# 71401–3
Chemicals, peptides, and recombinant proteins
Recombinant mouse LIF Sigma Cat#ESG1107
Recombinant mouse EZH2 This study N/A
Recombinant mouse SUZ12 This study N/A
Recombinant mouse EED This study N/A
Recombinant mouse RBPA48 This study N/A
Recombinant GST fused histone H3 peptide (16–44 a.a.) This study N/A
SuperScript III reverse transcriptase Invitrogen Cat#18080085
SUPERase•In RNase Inhibitor Invitrogen Cat#AM2694
TURBO DNase Invitrogen Cat#AM2238
RNase H New england biolabs Cat#M0297
T4 Polynucleotide Kinase New england biolabs Cat# M0201
T4 RNA Ligase 2, truncated KQ New england biolabs Cat# M0373
CircLigase ssDNA Ligase Epicentre Cat# CL4111K
RNase A Thermo Scientific Cat# EN0531
Ribonucleoside Vanadyl Complex New england biolabs Cat# S1402S
Dimethyl sulfide Sigma Cat#320293
3X Flag peptide Sigma Cat# F4799
TRIzol Reagent Thermo Scientific Cat# 15596018
Critical commercial assays
MTase-Glo Methyltransferase Assay kit Promega Cat# V7601
Oligotex Direct mRNA kit QIAGEN Cat# 70022
Agencourt AMPure XP Beads Beckman Coulter Cat#A63881
NEBNext Ultra II DNA Library Prep Kit for Illumina New England Biolabs Cat#E7645S
NEBNext Ultra II directional RNA Second Strand Synthesis Module New England Biolabs Cat#E7550S
NEBNext ChIP-Seq Library Prep Master Mix Set for Illumina New England Biolabs Cat#E6240S
NEBNext Poly(A) mRNA Magnetic Isolation Module New England BioLabs Cat#E7490S
Lipofectamine 3000 Transfection Reagent Thermo Fisher Scientific Cat#L3000015
NEBNext Multiplex Oligos for Illumina (Index Primers Set 1) New England BioLabs Cat#E7335S
NEBNext Multiplex Oligos for Illumina (Index Primers Set 2) New England BioLabs Cat#E7500S
Dynabeads Protein G for Immunoprecipitation Invitrogen Cat# 10003D
Deposited data
DMS-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
DMS-seq in WT female mouse differentiating ES cell (day 3) This study GEO:GSE219083
DMS-seq in WT female mouse differentiating ES cell (day 0) This study GEO:GSE219083
Xist CHART-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
Xist CHART-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
Xist CHART-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
H3K27me3 ChIP-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in WT female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in 3’ rG4 mutant clone 5 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
RNA-seq in 3’ rG4 mutant clone 6 female mouse differentiating ES cell (day 7) This study GEO:GSE219083
In situ Hi-C in WT female mouse differentiating ES cell (day 7) 38 GEO: GSE99991
Experimental Models: Cell Lines
WT Xist female ES cell (TsixTST/+) 42 N/A
AF rG4 mutant Xist female ES cell (TsixTST/+) Clone #7 This study N/A
AF rG4 mutant Xist female ES cell (TsixTST/+) Clone #11 This study N/A
E7 rG4 mutant Xist female ES cell (TsixTST/+) Clone #4 This study N/A
E7 rG4 mutant Xist female ES cell (TsixTST/+) Clone #6 This study N/A
3’ rG4 mutant Xist female ES cell (TsixTST/+) Clone #5 This study N/A
3’ rG4 mutant Xist female ES cell (TsixTST/+) Clone #6 This study N/A
Oligonucleotides
Primers for qPCR,see Table S7 Integrated DNA Technologies N/A
Primers for in vitro trascription,see Table S7 Integrated DNA Technologies N/A
primers for DMS-seq library preparation,see Table S7 Integrated DNA Technologies N/A
Recombinant DNA
pSpCas9(BB)-2A-Puro (PX459) v2.0 61 Addgene Cat#62988
pMB 1609-pRR-EGFP 62 Addgene Cat#65852
pGEX-4T-1-histone H3 peptide (16–44 a.a.) This paper N/A
pJET1.2 Thermo Scientific Cat# K1231
Software and algorithms
SAMtools v1.4.1 63 http://samtools.sourceforge.net/
STAR aligner (v2.7.3) 64 https://github.com/alexdobin/STAR
Cutadapt v1.8.1 65 https://cutadapt.readthedocs.io/en/stable/#
ImageJ v1.53a 66 https://imagej.nih.gov/ij/
bedtools 67 http://bedtools.readthedocs.io/
deepTools 68 https://deeptools.readthedocs.io/en/develop/
Bowtie2 69 http://bowtiebio.sourceforge.net/bowtie2/index.shtml
SAMBAMBA v0.6.6 70 https://github.com/biod/sambamba/releases
Bowtie v1.1.1 71 http://bowtie-bio.sourceforge.net/index.shtml
Python/Biopython v1.70 72 https://biopython.org/
Prinseq-lite v0.20.4 73 https://sourceforge.net/projects/prinseq/files/
Cufflinks v2.2.1 74 http://cole-trapnell-lab.github.io/cufflinks/
Excel Microsoft www.office.com
Prism9 GraphPad www.graphpad.com
Structurefold2 75 https://github.com/StructureFold2/StructureFold2
rG4-seeker 36 https://github.com/TF-Chan-Lab/rG4-seeker

This paper does not use original code.

Additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

RESOURCES