Polycomb-mediated 3D-genome organization controls replication timing

Abstract

Polycomb (PcG) bodies are nuclear foci formed by polycomb protein complexes that contain PcG-bound DNA and are implicated in gene regulation during development and differentiation, although their precise molecular function remains unclear. Using tyramide signal amplification sequencing, we provide a comprehensive view of genomic regions associated with PcG bodies, including specific centromeric and telomeric sites. These regions are enriched for the repressive marks H3K27me3 and H3K9me3, depleted of the active mark H3K4me3, and display low chromatin accessibility. We find a high density of replication origins around PcG bodies, consistent with the replication factor origin recognition complex–associated protein (ORCA)/LRWD1 localizing at these sites. ORCA interacts with the polycomb-repressive complex, stabilizes the H3K27 methyltransferase, and facilitates H3K27me3 deposition at specific chromatin sites. Loss of ORCA affects chromatin organization around PcG bodies, leading to decompaction of the repeat regions, and enhanced initiation from replication origins. Our results suggest that ORCA associates with PcG-bound chromatin to maintain a repressive environment that regulates replication origin firing timing.

Polycomb bodies in the cell nucleus regulate gene expression and play a key role in replication timing control.

INTRODUCTION

Polycomb (PcG) bodies, nuclear structures formed by polycomb group protein complexes, are known to regulate gene expression and required for the maintenance of cellular identity (1, 2). The polycomb group of proteins is responsible for depositing the repressive histone mark, H3K27me3, and this process is critical for regulating gene expression during development and maintaining pluripotency (3). Polycomb proteins comprise two distinct multiprotein complexes, polycomb repressive complex 1 (PRC1) and PRC2, with H2AK119 ubiquitin ligase and H3K27me3 methyltransferase activities, respectively, known to maintain transcriptionally repressed chromatin (4). It is generally believed that the PcG bodies mediate gene repression, playing key roles in dictating when a cell would differentiate and how it would maintain a specialized function. The mechanistic insights into how the PRC1 and PRC2 cooperate to repress gene expression during development or organize the three-dimensional (3D) genome remains to be completely understood (1).

In transformed human cells, polycomb complexes form nuclear foci, the PcG bodies, within which PcG-bound DNA regions are localized and contact each other (2). Normal cells typically lack large PcG bodies and may contain PcG foci of varying size and number, and it is thought that gene expression is well controlled in these cells and they do not require extensive gene silencing and epigenetic memory maintenance that are provided by the PcG bodies (5). Yet, the functional relevance of these structures remains unclear.

We used a chromosomal proximity mapping tool, TSA-seq (tyramide signal amplification sequencing), to identify the genomic regions associated with PcG bodies as it provides precise measurements of chromosome positioning relative to a subnuclear domain (6). Our approach showed a robust association of several centromeric regions, including from chromosome 1, with the PcG body, validating the sequencing approach in light of the cytological data (7). Further, genes associated with the PcG body are enriched for the repressive histone marks H3K27me3 and H3K9me3 and show reduced levels of H3K4me3 and chromatin accessibility. Immediately adjacent to the PcG bodies, where the highest H3K27me3 and H3K9me3 association were identified, we observed repression of genes. In addition, we observed that the late-firing replication origins are organized near PcG body.

Replication origins are bound by prereplicative complex (pre-RC) factors during G1 phase of the cell cycle (8). In eukaryotes, the origin recognition complex (ORC) binds to the replication origin and recruits other factors to establish the pre-RC (9). ORC-associated protein (ORCA/LRWD1) stabilizes the loading of the ORC complex on the chromatin (10, 11). ORC and ORCA interact with modified histone marks on chromatin to regulate processes such as DNA replication initiation, DNA repair, and chromatin organization (12, 13). In postreplicated cells, ORCA preferentially localizes to heterochromatic regions and binds repressive histone marks, including H3K9me3, H4K20me3, and H3K27me3 (14–18). We previously reported that ORCA is selectively enriched with H3K9me3 at centromeres and subtelomeric regions for modulating homologous recombination (13). However, the functional relevance of ORCA’s association with H3K27me3 remains to be understood.

In this study, we report that ORCA plays a role in chromatin compaction through the PRC. Our data show that ORCA interacts with the PRC proteins and is integral to the stability of this complex. Further, ORCA localizes to PcG bodies and affects the 3D-genome organization. The loss of ORCA results in a reduction in the genome-wide distribution of H3K27me3, concomitant with changes in chromatin accessibility and altered transcription of repeats. TSA-seq in cells lacking ORCA, similar to the loss of PRC2 component, EZH2, showed reorganization of the 3D-genome due to the loss of repressive histone marks and decompaction of repeat regions. Last, the replication origin function and efficiency were altered with the loss of ORCA along with changes in genome architecture. We propose that ORCA, by associating with the polycomb complex–bound chromatin, maintains a repressed heterochromatin environment to ensure accurate timing of replication origin firing.

RESULTS

Genome-wide mapping of chromatin organization around the PcG bodies

In mammals, PRC1 and PRC2 complexes consist of core and accessory proteins (19), while the PRC1 core components include RNF2, the ubiquitin ligase and PCGF proteins (polycomb group ring finger, like, PCGF4-BMI1), the PRC2 core components include EED, EZH2, SUZ12, and RbAp 46/48. PcG bodies show distinct nuclear foci, harboring members of the PRC1 and PRC2 complexes, especially of large size in cancerous cells like U2OS but not in diploid fibroblasts like WI38, as can be observed by immunofluorescence localization of RNF2 or BMI1, components of PRC1 (Fig. 1A and fig. S1A). RNF2 staining shows the distinct pattern for PcG bodies, along with other PRC1 members like BMI1 (also known as PCGF4) (20). To map the genomic regions that are around the PcG bodies, we performed a proximity labeling–mediated genome mapping method, TSA-seq. This method has been previously used by us and others to determine the genome organization around other nuclear bodies, such as nuclear speckles and nucleoli (6, 21, 22). TSA uses an antibody-coupled horseradish peroxidase (HRP) to catalyze the formation of diffusible biotin tyramide free radicals and creates a radical concentration gradient at the staining target (Fig. 1B). To convert TSA staining into genomic readout, we performed the RNF2 and BMI1 TSA reaction in U2OS cells with reverse formaldehyde cross-linking, DNA isolation, pull down of biotinylated DNA, and high-throughput sequencing [as described previously (6) and also included within Materials and Methods]. We tested “condition A” (lower tyramide concentration) and condition E (saturated tyramide concentration) (23) and found that condition E produced saturated TSA labeling across the cell and showed maximum DNA pull-down efficiency (fig. S1B). On the basis of these quality control experiments, we used TSA-seq using condition E, consistent with what the Belmont laboratory reported as the ideal condition for nuclear speckle TSA-seq experiments (23). To obtain a comprehensive view of the genome, including centromeres, telomeres, and repeat regions, we aligned TSA-seq and chromatin immunoprecipitation sequencing (ChIP-seq) data to the telomere-to-telomere (T2T) assembly. We generated TSA enrichment genomic maps by plotting the log₂ratio of normalized pulldown read density versus normalized read density of input DNA over a 20-kb sliding window. Twenty-kilobase genomic bins were assigned TSA scores based on distance from the RNF2-stained PcG bodies. The score ranged from 0 to 100, with the region closest to PcG having a score of 100 and the region farthest away a score of 0. The chromosome-wide plot represented 20-kb bins of genome, and regions within a bin (0 to 10, 10 to 20, 20 to 30, 30 to 40…..up to 90 to 100) were assigned a color indicative of their relative position around the PcG bodies (Fig. 1C and fig. S1C). TSA scores that range from 50 to 100 appeared as peaks (for example, a TSA score of 60 would be a peak at +10), and 0 to 50 were represented as valleys (for example, a TSA score of 40 would be a valley at −10) (Fig. 1D). PcG bodies are known to be associated with the periphery of heterochromatin, specifically centromeres and telomeres. Strikingly, the centromere of chromosome 1 was tethered to the PcG bodies [Fig. 1, C (blue region) and D, and fig. S1C], a result consistent with what has been reported earlier (7). TSA-seq of PcG bodies with RNF2 and BMI1 showed a high Pearson coefficient of correlation (r = 0.92), demonstrating the rigor of our approach (Fig. 1E and fig. S1D).

Fig. 1. Genome organization around PcG bodies.

(A) Representative image of immunostaining of RNF2 (green), a component of the PcG bodies in U2OS cells. Scale bar, 5 μm. (B) Schematic of TSA-seq. Primary antibody (RNF2) and HRP-conjugated secondary antibody target the PcG bodies (red circles). HRP catalyzes formation of tyramide-biotin free radicals that diffuses and covalently links to neighboring DNA. Tyramide-biotin–labeled DNA is pulled by streptavidin beads and sonicated. This was followed by DNA library preparation and sequencing. Created in BioRender. Chetlangia, N. (2026) https://BioRender.com/es4iuyi. (C) Chromosome-wide plot with TSA scores for RNF2 TSA-seq. Each chromosome is divided into 20-kb bins; each bin has a TSA score represented by the TSA color scale. Concentric circles show radial staining pattern of TSA labeling; TSA scores 90 to 100 represent the center of the circle, and denote the region closest to the PcG bodies. (D) IGV snapshots of chromosome 1 depicting RNF2, SON (U2OS), and CENPB (K562 cells). TSA-seq score ranges from −50 to +50, and the specific range for each snapshot is noted. Positive values are shown as peaks, and negative values represent valleys. (E) 2D-scatter plot for Pearson coefficient correlation of RNF2 with CENPB (0.25) and SON (0.40), and BMI1 TSA-seq (0.92). Also see fig. S1D. (F) Immunostaining of RNF2 (green) and DNA FISH (red) with pUC1.77 probe (representing Chr1q12 with a TSA score of 100) (left); merged image with DAPI as counterstain (right). Insets show zoomed image of overlap between RNF2 and DNA FISH probe. Scale bar, 5 μm. (G) Quantification of DNA FISH probes from regions with TSA score of 0 to 10, 40 to 50, 70 to 80, 80 to 90, and 90 to 100 (n = 30 for each probe). Box plot denotes the genomic distance distribution with central line indicating the mean value.

To test the potential of TSA to uniquely represent nuclear organization, we compared RNF2 TSA-seq with TSA-seq data of other nuclear compartment markers (21, 23). We aligned TSA-seq of SON [from U2OS cells, (24)], marker of nuclear speckles and CENPB [binds alpha-satellite repeats flanking the centromeres from K562 cells (21) since CENPB binding is highly conserved across cell lines] to T2T and compared it with RNF2 TSA-seq (Fig. 1D). RNF2 TSA-seq showed a positive correlation with SON/nuclear speckles (Fig. 1E). CENPB TSA-seq marking the centromeric positions showed multiple centromeres are enriched in the RNF2 TSA-seq (Fig. 1D and fig. S1F). We found that multiple centromeres and the telomeres of the q-arm of several chromosomes were proximal to the PcG bodies. As can be seen by the chromosome plot, certain centromeres (chr1, 2, and 10) and certain telomeres (chr 2, 4, 15,16, 17, 20, 21, and X) exhibited high PcG TSA scores (Fig. 1C and fig. S1, C and F). A chromosome wide comparison further corroborated that select centromeres are strongly bound to the PcG bodies (fig. S1F).

Next, we were interested in understanding how the chromatin is organized around PcG bodies and the specific chromatin marks associated with these bodies. First, we performed DNA fluorescence in situ hybridization (FISH) to validate that TSA-seq scores were proportional to distance from the PcG bodies. As observed from the chromosome plots (Fig. 1, C and D), the pericentromeric region of chromosome 1 displayed a TSA score of 90 to 100, concordant with a previously reported study (7). DNA FISH using a pUC1.77 probe [specific for the pericentromeric region of chromosome 1, (7)] along with immunofluorescence staining to detect RNF2 showed 100% overlap between them (distance within 150 nm) (Fig. 1F). Next, using five bacterial artificial chromosome (BAC) DNA FISH probes from genomic regions with varying TSA scores (0 to 10, 40 to 50, 70 to 80, 80 to 90, and 90 to 100), we showed that the distance of the DNA FISH foci correlated with the corresponding TSA score, 0 to 10 being the farthest and 90 to 100 completely overlapping with the RNF2-decorated PcG body (Fig. 1G and fig. S1E). On the basis of our combination of DNA sequencing and microscopy, we establish TSA as a powerful method to map the genome that is located in proximity to the PcG bodies.

Genomic and epigenomic features proximal to the PcG bodies

Next, we examined the distribution of epigenomic and genomic features with respect to relative distance from the PcG bodies, as estimated by TSA-seq. We represented the data as histograms denoting the segregation of different features as a function of RNF2 TSA-seq score, with 0 to 10 to 90 to 100 corresponding to the lowest and highest 10% TSA bins, respectively. As evident from the chromosome plots (Fig. 1C and fig. S1C), every TSA score bin covers a different proportion of the genome. We counted the number of genomic base pairs (bp) corresponding to each TSA score bin. Then, we represented the data as percentage of the entire genome covered in each bin. On the basis of such analyses, we observed that most of the genome was located in the 40-to-60 PcG body TSA bins (see the bar plot in Fig. 2A).

Fig. 2. Genomic and epigenomic features relative to proximity to PcG bodies.

(A) Genome is represented as histograms denoting the segregation of different features as a function of RNF2 TSA-seq score, with 0 to 10 to 90 to 100 corresponding to the lowest and highest 10% TSA bins, respectively. The data are shown as percentage of the entire genome covered in each bin. Note that most of the genome is present in the 40-to-60 TSA bins. (B) H3K27me3 (ChIP-seq) enrichment in RNF2 TSA bins. Note an enrichment of H3K27me3 in the 90-to-100 TSA bin. (C) H3K9me3 enrichment in RNF2 TSA bins. TSA bins 80 to 100 show maximum enrichment for H3K9me3. (D) H3K4me3 enrichment in RNF2 TSA bins. Note the reduced H3K4me3 near PcG bodies. (E) Chromatin accessibility by ATAC and its enrichment in TSA bins. (F) Top 10% expressed protein coding genes (by counts per minute) and its correlation with TSA. Note the lack of gene expression immediately next to the PcG bodies. (G) IGV snapshot of chromosome 1 (TSA score 100) with TSA, H3K27me3, H3K4me3, H3K9me3, and ATAC peaks. Note the enrichment of repressive marks and lack of active marks at the TSA100 region. (H) BMI1 (CUT&RUN) enrichment in RNF2 TSA-seq bins. (I) RNF2 (CUT&RUN) enrichment in RNF2 TSA-seq bins. (J) EZH2 (CUT&RUN) enrichment in RNF2 TSA-seq bins. (K and L) IGV snapshots of chromosome 15, and chromosome 1 with TSA and the RNF2 CUT&RUN peaks. ***P < 0.001. ns, not significant.

As an alternate method for analyzing the TSA-seq data, we also divided the genome into an equal number of 20-kb bins based on the TSA percentile (fig. S2A), similar to previous analyses of TSA-seq for speckle-associated regions (6). Using this method, we found that H3K27me3 and H3K9me3 were found in proximity to the top 10% genomic bins (fig. S2, B and C). However, this percentile-based binning method could not distinguish regions highly enriched proximal to PcG body versus those that were close but not directly tethered (fig. S2D), for example, the orange peak at chr1 centromere/pericentromere (using the Min/Max TSA-binning method), to the brown peak that extends several mega base pairs (Mb) upstream and downstream (genome percentile method, in the fig. S2A). Using the Min-Max method (detailed in Fig. 2), we distinguished the genomic regions that fall in the 90-to-100 TSA bin versus those present within the 80-to-90 bin. In contrast, the genome percentile method grouped TSA scores ∼70 to 90 into a single bin, potentially leading to the loss of finer details and misinterpretation of the data (fig. S2D). We, therefore, proceeded with Min-Max method of binning for all the TSA-seq analyses.

PRC2 subunits and H3K27me3 are known to colocalize at PcG bodies, indicating that they are sites of PRC2 enzymatic activity (25, 26). We performed H3K27me3 ChIP-seq in U2OS cells and found that H3K27me3 was most enriched in genomic regions proximal to PcG bodies, as quantified by the fraction of TSA-bin base pairs overlapping the H3K27me3 peaks (Fig. 2B). Despite a small fraction of the genome being covered by the TSA score bins 70 to 100, these bins were highly enriched for H3K27me3. For each TSA bin, we first intersected the genomic regions corresponding to each TSA bin with H3K27me3 peaks using bedtools intersect. We then calculated the total number of base pairs in the resulting intersected regions. This value was divided by the total number of base pairs within the respective TSA bin to obtain the enriched fraction. We observed the highest enriched fraction of H3K27me3 in TSA bin 90 to 100, with the enriched fraction progressively declining with increasing distance from the PcG body.

In mammalian cells, PRC2 and H3K27me3 have been reported to cooperate with H3K9 methylation to maintain heterochromatin organization (27). Further, nucleosomes purified from HeLa cells showed frequent co-occurrence of H3K27me3 with H3K9 methylation (28). We observed that similar to H3K27me3, TSA score bins proximal to PcG bodies (TSA score bins 70 to 100) show higher enrichment of H3K9me3, the highest enrichment being in the 80-to-100 TSA score bins (Fig. 2C). These observations suggest an enrichment of repressive histone marks at the chromatin that are located proximal to the PcG bodies. Consistent with these observations, we found that H3K4me3, a histone mark representing active gene expression, was enriched in genomic regions with TSA score 40 to 80, with low enrichment in the TSA bins 80 to 100 (Fig. 2D). A similar profile was observed for the accessible chromatin (ATAC-seq) (Fig. 2E), implying that active chromatin is located farther away from PcG bodies. As an example, chromosome 1 is shown for the centromere and pericentromeric region, which has a TSA score of 100 (Fig. 2G). In this region, there was enrichment of H3K27me3 and H3K9me3 while H3K4me3 and ATAC peaks were sparse or absent (Fig. 2G). A genomic region on chromosome 17 with a lower TSA score (TSA 60) showed relatively more active marks and open chromatin and fewer repressive marks (fig. S2E).

Furthermore, we have performed a similar analysis for an active nuclear compartment, nuclear speckles. TSA-seq using SON, a core component of nuclear speckles that boost gene transcription and splicing, revealed a profile that is markedly distinct from PcG body, RNF2/BMI1 TSA-seq. Using the same approach described above, we found that H3K4me3, an active chromatin mark, was enriched in regions closest to speckles, whereas H3K27me3 was positioned furthest away (fig. S2, F to H). This is consistent with previously reported findings from the Belmont laboratory (23). These results provided additional rigor to our approach and substantiated the validity of our conclusions.

Last, to address whether there is any correlation between gene expression and their proximity to PcG bodies, we first calculated the gene density in each bin. As expected from the data shown in Fig. 2A, we also found that the maximum gene density was in the 40-to-60 bins. Consistent with this, protein-coding gene expression was maximum from these bins. However, the top 10% expressed genes (coding and noncoding), were distributed in the 40-to-80 TSA score region, with the maximum being in the 70-to-80 TSA bin (Fig. 2F). These observations support the existing notion that gene expression is repressed immediately proximal to the PcG bodies (80 to 100 TSA).

As noted by the Belmont laboratory (6), TSA-seq measures the spatial proximity of chromatin regions bound to specific nuclear domains, independent of whether the protein is located within the cross-linking distance from the chromatin. However, other genomic methods, such as ChIP-seq or CUT&RUN, provide measurements of direct protein-DNA binding genome-wide, representing a molecular proximity tool that offers an occupancy map of proteins likely associated with genomic regions within cross-linking distance of DNA. To rigorously evaluate TSA-seq and to compare it with complementary approaches for identifying PcG target regions, we performed CUT&RUN in U2OS cells using antibodies against RNF2, BMI1, and EZH2. Our results demonstrated that a large number of RNF2 CUT&RUN reads came from 70-to-100 RNF2 TSA-bins, consistent with an enrichment of RNF2 binding to PcG-proximal regions (Fig. 2I). Similar results were obtained when BMI1 or EZH2 CUT&RUN data were compared to RNF2 or BMI1 TSA-seq (Fig. 2, H to J, and fig. S2, I to K). Genomic regions away from the PcG bodies (0 to 50 bins) also showed numerous RNF2/BMI1-bound sites, demonstrating that these approaches are distinct and can clearly differentiate PcG-proximal regions from RNF2/BMI1-bound sites that are located farther from the PcG bodies. For example, the chromosome 1 centromere, which is tethered to PcG bodies shows several prominent CUT&RUN peaks (Fig. 2L and fig. S2M). A prominent TSA-peak on chromosome 15 corresponds to fewer CUT&RUN peaks (Fig. 2K). In contrast, regions within chromosome 5, 8, and 15, identified as distant from PcG bodies, showed distinct RNF2 CUT&RUN peaks (Fig. 2K and fig. S2M), highlighting the limited correlation between TSA-seq and CUT&RUN datasets (Fig. 2L). While TSA-seq and CUT&RUN datasets may show some overlap, the former provides important information on the 3D genome organization, and the latter addresses specific binding sites of a particular protein on the chromatin. We conclude that by using TSA to map the genome near PcG bodies, our method has identified epigenomic features associated with these regions, overcoming the limitations of other genomic methods.

The prereplication complex factor, ORCA, interacts with the PRC complex

We and others have shown that ORCA interacts with repressive histone marks, including the histone mark H3K27me3 written by the PRC2 complex (12, 14, 17). We used an in vivo cell system (CLTon) that contains an array of ∼200 lac operator repeats, stably integrated as a single heterochromatic locus in human osteosarcoma (U2OS), and can be visualized by the yellow fluorescent protein (YFP)–tagged Lac repressor (29). We tethered the triple fusion protein YFP-LacI-EZH2 to the heterochromatic locus and addressed whether ORCA was recruited to the locus (Fig. 3A) (16). We observed that ∼70% of cells showed enrichment of ORCA at the EZH2-tethered loci (Fig. 3B), indicating that ORCA is recruited to the genomic sites containing EZH2 (PRC2 methyltransferase) and/or H3K27me3 mark. Further, coimmunoprecipitation experiments in U2OS cells coexpressing T7-tagged ORCA and c-MYC–tagged EZH2 or YFP-tagged BMI1 confirmed ORCA’s interaction with PRC2 and PRC1, respectively (fig. S3, A and B).

Fig. 3. Interaction of PRC with ORCA.

(A) U2OS 2-6-3 cells expressing YFP-LacI (green) and CFP-ORCA (shown in magenta) or YFP-LacI-EZH2 (green) and CFP-ORCA. Note robust accumulation of ORCA to the YFP-LacI-EZH2 tethered heterochromatic locus. (B) Quantification of the percentage of cells showing overlap of the YFP-LacI or YFP-LacI-EZH2 locus with CFP-ORCA. Error bars showing SD from three biological replicates, ***P ≤ 0.001 by two-tailed Student’s t test. Scale bar, 5 μm. (C) PLA of ORCA and BMI1 (PRC1); cell on the left shows ORCA as a negative control (n = 150). Red foci indicate association of ORCA and BMI1; DAPI (blue). Scale bar, 5 μm. (D) Quantification shown as a bar graph depicting the total number of PLA foci observed in (C). (E) Percentage overlap between ORCA ChIP peaks (1.5, 3, and 5 hours in G1) and H3K27me3 ChIP peaks (base pairs). (F to H) ORCA [ChIP-seq (1.5, 3 and 5 hours in G1)] enrichment in RNF2 TSA-seq bins. As noted in Fig. 2, the data are represented as histograms as a function of RNF2 TSA-seq score, with 0 to 10 to 90 to 100 corresponding to the lowest and highest 10% TSA score, respectively. Note the enrichment of ORCA from 70-to-90 TSA bins. ***P < 0.001. (I) Read profile plot showing H3K27me3, EZH2, RNF2, and BMI1 signal intensity (RPKM) 1 kb upstream and downstream of ORCA peaks. Right, quantification of read counts from each mark at respective ORCA peaks. ***P < 0.001.

To understand the spatial interaction of ORCA with the polycomb complex, we performed a proximity ligation assay (PLA) of ORCA and components of PRC1 (BMI1) and PRC2 (EZH2) (Fig. 3C and fig. S3, C and D). PLA analyses revealed that ORCA interacted with the PRC1 at 10 to 15 foci per cell, similar to the number of PcG bodies present in U2OS cells (Fig. 3D). There were fewer interaction signals when PLA was performed for the association of ORCA with PRC2 (EZH2) (fig. S3D). It is notable that immunolocalization of ORCA, for the most part, showed diffuse nuclear staining and signal discernible at the constitutive heterochromatic sites, including at centromeres and telomeres, upon preextraction procedure before immunostaining (10). Our PLA experiments support the model that ORCA is detected at heterochromatin-enriched PcG bodies although ORCA may not be readily discernible at facultative heterochromatin sites by conventional immunostaining.

Having established that ORCA interacts with the polycomb complex and the fact that it shows spatial proximity to PRC1 and PRC2 components, we next addressed the genome-wide distribution of ORCA to assess whether there is any evidence for enrichment of ORCA at H3K27me3-decorated genomic sites. ORCA is predominantly bound to chromatin in the G1 phase of the cell-cycle (10, 11). As reported previously, we synchronized cells into early, mid, and late G1 by releasing them from mitosis at 1.5, 3, and 5 hours (release from mitosis) representing early, mid, and late G1, and performed ORCA ChIP-seq (30). ORCA-binding sites from early-to-late time points in G1 showed substantial overlap with H3K27me3 peaks (40% overlap with ORCA 1.5 hours ChIP, ∼45% with ORCA 3 hours ChIP, and >50% with ORCA 5 hours ChIP) (Fig. 3E).

We next examined whether ORCA is bound to regions proximal to PcG bodies by evaluating the overlap between ORCA ChIP-seq and the RNF-2 TSA-seq. Most ORCA peaks were present in genomic regions with RNF2 TSA scores of 70 to 90 (Fig. 3, F to H). We overlapped ORCA ChIP-seq with H3K27me3 or RNF2, BMI1, and EZH2 CUT&RUN and presented the results as a profile plot [based on normalized read count, reads per kilobase million (RPKM)], to show overlap between ORCA and PRC1/PRC2 (Fig. 3I and fig. S3E). Note the prominent overlap in ORCA 1.5 and 3 with PRC1 and PRC2, consistent with most ORCA being lost from chromatin by 5 hours (late G1 phase). Across the genome, H3K27me3 levels vary, with both high- and low-signal regions. The lower average signal observed at 5 hours in G1 reflects averaging across a smaller subset of ORCA peaks, many of which are located in regions with relatively low H3K27me3 enrichment. Thus, despite the ∼50% overlap of ORCA peaks overlapping H3K27me3 (Fig. 3E), the H3K27me3 average read density at ORCA-bound sites is reduced at 5 hours. A genome browser view to depict the overlap of ORCA ChIP-seq peaks with that of H3K27me3 and H3K9me3 is shown in fig. S3F. Similarly, the chromosome 1 centromeric region (with a TSA score of 100) showed clear enrichment of ORCA, consistent with ORCA localizing proximal to PcG bodies (fig. S3G). Thus, our results establish the spatial and genome-wide distribution of ORCA near PcG bodies.

PRC2 complex is stabilized by ORCA

To gain functional insights into the role of ORCA in regulating PRC2 function, we investigated how the absence of ORCA affects the localization and enzymatic activity of the PRC2 complex. Immunoblotting experiments demonstrated that the total protein levels of the core PRC2 (EZH2, EED, SUZ12, and RbAp46/48) complex decreased upon the loss of ORCA (fig. S4, A to C). On the other hand, the levels of PRC1 (RNF2 and BMI1) remained largely unchanged (fig. S4, B and C). Because the loss of EZH2 causes destabilization of other core PRC2 complex proteins (31), it was expected to observe that all components of the PRC2 complex were destabilized upon ORCA depletion. Consistent with this, we also observed a decrease in H3K27me3 and me2 protein levels in ORCA-depleted cells (fig. S4, B and C). However, H4K20me3, another repressive histone mark recognized by ORCA, remained mostly unperturbed upon ORCA depletion (fig. S4C). The loss of ORCA also reduced the stability of the H3K9-depositing methyltransferase complex, supporting our model that ORCA functions as a scaffold protein to stabilize HMT complexes (12). We also performed depletion of other ORC components, Orc1 and Orc2, and found that the loss of Orc2 resulted in a reduction in EZH2 (fig. S4A), consistent with our earlier observations that Orc2 stabilizes ORCA (32).

Because ORCA loss causes destabilization of the PRC2 complex, we assessed whether ORCA influences the genome-wide chromatin distribution of H3K27me3. To this end, we performed ChIP-seq of H3K27me3 in control and ORCA-depleted cells. ORCA depletion resulted in a clear reduction in the number of H3K27me3 peaks genome-wide (illustrated in Fig. 4A). In the absence of ORCA, we observed fewer peaks of H3K27me3 at multiple genomic regions (Fig. 4A and fig. S4G). Further, genome-wide, about 42% of H3K27me3 peaks were lost in cells lacking ORCA (Fig. 4B). Notably, ORCA-depleted cells showed changes in PcG body organization, as observed by the localization of BMI1 (Fig. 4C). Control cells showed punctate foci of the PcG bodies, whereas cells lacking ORCA showed elongated “string-like foci,” reminiscent of reorganization of the PcG bodies (Fig. 4, C to E). The spatial redistribution of the PcG bodies was also observed in cells lacking EZH2 (fig. S4, D and E). A large majority of cells lacking ORCA showed this redistribution of the PcG bodies, and these cells showed a decrease in the number of PcG foci per nucleus (Fig. 4E). We observed a clear spatial decompaction of pericentromeric heterochromatin associated with PcG bodies in cells lacking ORCA [using the puC1.77 probe (as noted in Fig. 1)] as observed by DNA FISH (Fig. 4, F and G). These data suggest that reduction in H3K27me3 upon ORCA depletion results in the decompaction of chromatin proximal to PcG bodies.

Fig. 4. Loss of ORCA leads to H3K27me3 loss and PcG body reorganization.

(A) Genome browser view of chromosome 7 with H3K27me3 ChIP-seq peaks in control and ORCA-depleted cells. The features are from the chromosome region depicted by red rectangle. Note the reduction of H3K27me3 in siORCA-treated cells from the region. (B) Venn diagram showing overlap of the H3K27me3 peaks between control and siORCA-treated cells. Note the global loss of H3K27me3 peaks upon the loss of ORCA but emergence of 23,986 new peaks. (C) Immunostaining of BMI1 (green) in control and ORCA-knockdown cells. DAPI is shown in blue. Images were captured using SIM microscopy. Insets (white rectangle) denote zoomed out PcG bodies. Note the reorganization of PcG bodies upon the loss of ORCA. Scale bar, 5 μm. SIM, structured illumination microscopy. (D) Quantification of the number of cells showing changes in PcG body patterning upon the loss of ORCA. Error bars showing SD from three biological replicates, ***P ≤ 0.001 by Welch two sample test. (E) Quantification of the number of PcG foci observed in control and siORCA cells. Note the significant reduction in the number of foci per nucleus upon the loss of ORCA. (F and G) Immunostaining of RNF2 (green) and DNA FISH (red) with pUC1.77 probe (chr1, TSA score 100, as in Fig. 1F) in control and ORCA knockdown cells. DAPI is shown in blue. Scale bar, 5 μm. ***P < 0.001 and *****P < 0.00001.

Further, we also performed CUT&RUN of BMI1 and RNF2 in ORCA-depleted cells and, using diffbind analyses, identified a significant number of regions with reduced binding of PRC1 and PRC2 upon the loss of ORCA (fig. S4F), which is consistent with the fact that the loss of PRC2 integrity results in reduced PRC1 binding at select targets (33). Integrative genomics viewer (IGV) snapshots for chromosomes 1 and 7 illustrate the reduced BMI1 binding in ORCA-depleted cells (fig. S4, G and H). These results demonstrate that although PRC1 protein levels are not greatly altered upon the loss of ORCA, their binding to specific genomic sites is affected in ORCA-depleted cells. The destabilization of the PRC2 complex results in a global reduction in H3K27me3 and also a decrease in PRC1 binding to chromatin sites in ORCA-depleted cells, further supporting the role of ORCA as a scaffold factor.

3D-genome organization is altered upon the loss of ORCA

We next determined the changes in PcG body–associated genome organization in cells lacking ORCA by using RNF2 and BMI1 TSA-seq in control and ORCA-depleted cells. The absence of ORCA correlated with obvious differences in RNF2 TSA scores (Fig. 5A and fig. S5, A, control, and B, ORCA depleted). Subtracting of the control TSA score and the siORCA TSA score, a positive value indicated genomic regions that were moving away from these PcG bodies when ORCA was depleted, and a negative score indicated genomic regions moving closer to the PcG bodies (Fig. 5A, color code denoted in the figure). The violin plot representation corroborated these observations as the bulge of the plot moved downward toward a lower TSA score value upon ORCA depletion (Fig. 5C and fig. S5C), denoting a larger fraction of the genome moving away from the PcG bodies in ORCA-depleted cells. A genome browser view of the chromosome plots of control and siORCA RNF2–TSA-seq, showed large genomic domains where TSA peaks were reduced or converted to valleys upon the loss of ORCA (Fig. 5B and fig. S5K). We validated the movement of the genomic regions by DNA FISH (Fig. 5, D and E). One such representative DNA FISH is shown where a genomic region at the TSA score of 58 (shown as peak value 8) in the control cells is present at the TSA score 38 in ORCA-depleted cells (shown as valley at −12; Fig. 5, D and E). Consistent with previous DNA FISH, we observed that probes from TSA score bins 30 to 40, 60 to 70, and 90 to 100 showed spatial distances from the PcG body in accordance with the TSA score (Fig. 5F). Next, we generated a scatter plot comparing the TSA scores of the corresponding 20-kb genomic bins between control and siORCA conditions. Approximately 28,000 genomic bins had a higher TSA score in control cells, suggesting that they had moved away from the PcG body upon ORCA depletion (Fig. 5G). Also, ∼1200 genomic bins moved closer to the PcG bodies upon loss of ORCA (Fig. 5H), whereas 121,565 genomic bins showed no change in TSA score between siORCA compared to the control (fig. S5D). Genomic region that moved closer to PcG body upon ORCA depletion represented regions that were predominantly located in the TSA 60 to 90 bins in control cells.

Fig. 5. Loss of ORCA results in altered 3D-genome reorganization around PcG bodies.

(A) Chromosome-wide plot of RNF2 TSA-seq (control − siORCA) across 20-kb bins, colored by TSA score [see (A) and (B)]. (B) Top, genome browser view of the TSA-seq of RNF2 control (orange) and ORCA-depleted cells (green) from chromosome 1. y axis is shown from −50 to 50. Bottom, zoomed region. (C) Violin plot of all 20-kb genomic bins showing TSA score distribution in control and siORCA cells. (D) Genome browser view of the TSA-seq of RNF2 control (orange) and ORCA-depleted cells (green) from chromosome 12 [also see (E)]. (E) Immunostaining of RNF2 (green) and DNA FISH (red) in control and siORCA-treated cells [representative image as in (D)]. Insets show zoomed image. Scale bar, 5 μm. (F) Quantification of the distance from the PcG body in nanometer (nm) of DNA FISH in ORCA-depleted cells (probes from TSA score bins 30 to 40, 60 to 70, and 90 to 100). Box plot shows the genomic distance distribution with line indicating the mean value. (G and H) 2D scatter plot showing genomic regions moving away (G) and moving closer (H) from PcG bodies. The x and y axes show RNF2 TSA scores in control and siORCA cells, respectively. Each dot represents a 20-kb bin, colored by control TSA score. Regions moving away: Control-siORCA >13; regions moving closer: Control- siORCA < −10. (I) Stacked bar plot represents the percentage of genome coverage by each TSA score bin in control and siORCA-treated cells. (J and K) Stacked bar plot represents the percentage overlap of H3K27me3 (J) and accessible chromatin (K) with TSA in control and siORCA-treated cells. (L) Stacked bar plot represents top 10% gene expression (in counts per minute) in control and siORCA-treated cells. P values for this figure are included in table S1.

Next, we examined the genomic and epigenomic features in each TSA bin upon the loss of ORCA. Consistent with the data shown in Fig. 5B, a large percentage of the genome present in the TSA bins of 40 to 60 in control cells shifted to bins 20 to 50 in ORCA-depleted cells (Fig. 5I). As shown earlier, fewer H3K27me3 peaks were observed upon the loss of ORCA, yet some marks were enriched proximal to PcG bodies (TSA score 80 to 100) (Fig. 5J). Further, we also observed changes in chromatin accessibility in the absence of ORCA (Fig. 5K and fig. S5, E and F). Last, we performed RNA sequencing (RNA-seq) in control and ORCA-depleted cells and found that the loss of ORCA did not significantly change the expression of the top 10% expressed genes (Fig. 5L and fig. S5G). While the maximum expression in control cells came from the 70-to-80 bin, in ORCA-depleted cells, it was from 70-to-80 and the 40-to-50 bins (Fig. 5L). We conclude that the changes in gene expression upon the loss of ORCA in U2OS were minimal and insignificant (fig. S5, F to I). On the other hand, we observed significant changes in gene expression upon the loss of ORCA in diploid fibroblasts WI38 (fig. S5J). However, these changes may reflect defective cell cycle progression previously observed upon ORCA depletion in diploid cells (10).

Because ORCA loss destabilized EZH2, leading to a concomitant reduction in H3K27me3 and reorganization of PcG bodies, we next assessed whether similar changes were observed upon EZH2 depletion (fig. S4A, showing the efficiency of depletion). As expected, the loss of EZH2, resulted in H3K27me3 reduction (fig. S4B). Further, siEZH2-treated cells showed clustering of PcG bodies (fig. S4, D and E). Furthermore, DNA FISH using chromosome 1 centromere probe confirmed such reorganization (Fig. 6, A and B; note that while in control, only one of the chr1 regions was decompacted, in EZH2si, all three were decompacted). Next, we performed RNF2 TSA-seq in cells lacking EZH2 (Fig. 6C). We observed a large majority of the genome moving away from the PcG bodies (Fig. 6D), similar to what we observe upon the loss of ORCA as can be seen in the violin plot by a downward shift in the distribution toward lower TSA scores. Upon subtraction of the control TSA score and the siEZH2 TSA score, a positive value indicated genomic regions that were moving away from these PcG bodies when EZH2 was depleted, and a negative score indicated genomic regions moving closer to the PcG bodies (Fig. 6E, color code denoted in the figure). IGV snapshots illustrate genomic regions shifting away from the PcG bodies (Fig. 6, F and G). A number of genomic regions that showed this movement were similar in EZH2-depleted and ORCA-depleted cells (fig. S6A).

Fig. 6. Loss of EZH2 results in reorganization of chromatin with respect to the PcG bodies.

(A and B) Immunostaining of RNF2 (green) and DNA FISH (red) with pUC1.77 probe (chr1, TSA score 100, as in Fig. 1F) in control and EZH2 knockdown cells. DAPI is shown in blue. Scale bar, 5 μm. ***P < 0.001. (C) Chromosome-wide plot showing TSA score for RNF2 TSA-seq in EZH2-depleted cells. Each chromosome was divided into 20-kb bins, and each bin was given a TSA score represented by the TSA color scale. (D) Violin plot of all 20-kb genomic bins showing TSA score distribution in control and siEZH2 cells. Note that the bulge of the plot moved downward toward a lower TSA score value upon EZH2 depletion. P values were computed using the Wilcoxon rank-sum test. Statistical analysis using chromosomal location–matched pairwise comparisons revealed a Hodges-Lehmann location shift of ∼10 units and a large effect size (Cliff’s δ = −0.614), indicating that most genomic bins exhibit reduced TSA scores following EZH2 depletion. (E) Chromosome-wide plot showing TSA score for RNF2 TSA-seq of the control minus the TSA score of siEZH2. Each chromosome was divided into 20-kb bins, and each bin was given a TSA score represented by the TSA color scale. (F and G) Representative genome browser view of the TSA-seq of RNF2 control (orange) and EZH2-depleted cells (blue) from chromosome 7 and 11. Note that the valleys become deeper and the peaks become valleys in the absence of EZH2, consistent with the genome moving farther from the PcG bodies.

However, the loss of EZH2 showed more dramatic changes in gene expression (fig. S6B). Just like ORCA-depleted cells, none of the differentially regulated genes in EZH2-depleted cells were polycomb targets as observed by intersecting EZH2 and H3K27me3 genome-wide binding sites (fig. S6C). This supports the conclusion that reduction in H3K27me3 in transformed cells may not be sufficient to cause changes in gene expression. Consistent with previous studies, changes in H3K27me3 alone may not necessarily lead to gene expression changes (34–36). Often, the loss of H3K27me3 may need to be accompanied by an increase in H3K4me3.

One of the important observations we made with the TSA-seq of RNF2 was the tight association of PcG bodies with specific centromeres and telomeres to PcG bodies (Fig. 1C). As previous work from our laboratory has shown that ORCA is enriched at these heterochromatic regions (16), we explored whether ORCA influences the association of these heterochromatic regions with PcG bodies. As the T2T genome assembly enabled mapping of the centromeres and telomeres (37), we compared the representation of these heterochromatic regions within the TSA bins in control and ORCA-depleted cells. Centromere and telomere coordinates were obtained from the T2T/CHM13 genome assembly. Min-Max–normalized TSA-seq signal intensities were averaged across these regions, and profile plot was made using the plotHeatmap function in deepTools.

In contrast to the rest of the genome, most of the centromeres moved closer to PcG bodies [chr 2, 3, 4, 7, 14, 16, 17, 18, 19, 20, and 22), whereas a few centromeres moved away—chr 6, 8, and X—or showed no change (Fig. 7, A to D). In contrast, telomeres, similar to the remainder of the genome, generally moved away (Fig. 7, E to H) in ORCA-depleted cells. Chromosome 3 centromere is shown as a valley in TSA-seq, indicating that it is located away from the PcG bodies (Fig. 7B). Although it is not strongly associated with the PcG body, it shifted closer to the PcG body upon loss of ORCA, as seen by the TSA score that changed from 6 (valley at −44) to 14 (valley at −36) in the absence of ORCA (Fig. 7, C to D). Similarly, the telomere from chromosome 14, which has a TSA score of 56 (peak at 6) in control, shifted to a TSA score of 37 (valley at −13) in ORCA-depleted cells (Fig. 7F), as shown by DNA FISH (Fig. 7, G and H). The relative organization of the PcG bodies to centromeres (fig. S7A) and telomeres (fig. S7B) did not reveal obvious alteration in centromere/telomere structure in the absence of ORCA. Note that PcG bodies themselves exhibited prominent reorganization in the absence of ORCA (fig. S7, A and B). Our results suggest that centromeres seemed to be tethered to the PcG bodies, whereas the rest of the genome including telomeres moved away from these bodies in ORCA-depleted cells, providing clues on how decompaction is accomplished. In cells lacking EZH2, several centromeric regions were flipped (for example, centromeres 3 and 4) and seemed to be localized closer to PcG bodies (fig. S5K). Equally interesting was the observation that a number of acrocentric chromosomes showed large genomic regions moving closer to the PcG bodies in the absence of EZH2 suggesting a role of satellite-repeat containing regions in tethering to PcG bodies (fig. S5K). While these are interesting observations, the mechanistic insights into why centromeric regions seem to tether to PcG will remain the focus of future studies.

Fig. 7. Centromeres and telomeres positioning with respect to the PcG bodies is affected by ORCA.

(A) Profile plot of chromosome-wide centromeric region with TSA score in control and siORCA treated cells. ±2 Mb window was selected spanning the centromeres; y axis shows average TSA score across all centromeres. (B) Genome browser view of chromosome 3 (red rectangle depicting the centromere) with RNF2 TSA-seq and H3K27me3 ChIP-seq in control and siORCA-treated cells. Top two panels depict TSA and bottom two panels depict H3K27me3 peaks. (C) Representative image of RNF2 immunostaining (green) and DNA FISH with a probe from chromosome 3 centromere (red). Note that the TSA score 6 in control is now at 14 in siORCA-treated cells. DAPI is shown blue. Insets depict zoomed foci in white rectangle. Scale bar, 5 μm. (D) Quantification of the DNA FISH data from Fig. 6C. Box plot shows the genomic distance distribution with line indicating the mean value, ***P ≤ 0.001 by Welch two sample test. (E) Profile plot of chromosome-wide telomere with TSA score in control and ORCA knockdown cells. ±2 Mb window was selected spanning telomeres; y axis shows average TSA score across all telomeres. (F) Genome browser view of chromosome 14 (red bar depicting the telomere) with RNF2 TSA and H3K27me3 ChIP-seq in control and siORCA-treated cells. Top two panels depict TSA and bottom two panels depict H3K27me3 peaks. (G) Representative image of RNF2 immunostaining (green) and DNA FISH with a probe from chromosome 14 telomere (red). DAPI is shown blue. Insets depict the foci in white rectangle. Scale bar, 5 μm. (H) Quantification of the DNA FISH data from Fig. 6G. Box plot shows the genomic distance distribution with line indicating the mean value, ***P ≤ 0.001 by Welch two sample test.

In U2OS cells, we observed a reduction in H3K27me3, changes in PcG body patterning, and decompaction of pericentromeric heterochromatin with the loss of ORCA. Also, in diploid WI38 cells, we observed similar decompaction of DNA upon ORCA depletion as noted by DNA FISH with chr 15 satellite III and centromere probe (fig. S7, C and D). Single-molecule RNA FISH with X-inactive specific transcript (XIST) probe also showed decompaction of inactive X chromosome in WI38 cells (fig. S7E), consistent with earlier studies, suggesting the involvement of ORCA/LRWD1 in the maintenance of inactive X chromosome in mouse embryonic stem cells (38). In general, we observed that overall centromeric repeat transcription was altered in the absence of ORCA, although we were unable to assign these transcripts to specific centromeres (fig. S7F). Because accurate assignment of repeat-derived transcripts to specific genomic loci requires high-depth long-read sequencing for accurate resolution, this remains a limitation of our study.

ORCA-mediated chromatin organization around PcG regulates replication timing

ORCA is a part of the preRC, where it acts as a scaffold for the assembly of other ORCs (12). To investigate the potential influence of PcG bodies on replication activity, we performed nascent strand sequencing (NS-seq) (39) in control and ORCA-depleted cells and compared it to our RNF2 TSA-seq dataset. Differentially activated origins were identified using DEseq2 analysis on reads per origin from biological replicates of control and ORCA-depleted cells. Our analysis revealed a substantial fraction of replication origins activated in ORCA-depleted cells compared to control (Fig. 8A). A heatmap of NS-seq data categorized initiated origins as unchanged, down-regulated, and up-regulated in control and ORCA-depleted cells (Fig. 8B). The distribution of mapped origins across TSA score bins in control cells showed that ∼180,000 origins were activated in genomic regions with a TSA score of 40 to 60, coinciding with the maximum genome covered within these bins (Fig. 8C, see the genome coverage in Fig. 5I). Similarly, in ORCA-depleted cells, ∼200,000 origins were activated in genomic regions with a TSA score of 30 to 50, with an additional ∼10,000 origins activated in regions distant from PcG (TSA score of 20 to 30). Further analysis indicated that while genomic regions proximal to PcG (TSA score 60 to 100) activated ∼2000 origins in both control and siORCA cells, siORCA cells exhibited an increased activation of ∼1200 origins within the TSA score range of 80 to 100, compared to ∼750 origins in control cells (Fig. 8D). Also note the obvious increase in the number of fired origins from 80 to 100 in ORCA-depleted cells (fig. S8B). When we used the genome percentile method for TSA binning (fig. S2), most origins were distributed within the TSA bins of 70 to 100 (fig. S8A). These findings suggest a redistribution of replication initiation activity upon ORCA depletion, particularly in genomic regions associated with PcG bodies.

Fig. 8. ORCA controls origin firing proximal to PcG bodies.

(A) Volcano plot depicting differential DNA replication origin activity between control and ORCA-depleted cells. (B) Heatmap of origin stratification in control and ORCA-depleted cells, classified as unchanged, down-regulated, or up-regulated. (C) Stacked bar plot illustrating the stratification of active origins in control and ORCA-depleted cells across TSA score bins. (D) Stratification of origins in the genomic region proximal to PcG (TSA score 70 to 100) in control and ORCA-depleted cells. (E) Stacked bar plot showing TSA bin-specific origin distribution within replication timing domains for control and ORCA-depleted cells. (F) TSA bin–specific origin distribution within replication timing domains in the genomic region near PcG (TSA score 70 to 100). (G) Violin plot of origin activity based on normalized NS signal across TSA bins. Statistical significance was assessed using the Wilcoxon paired test, with a fold change (FC) > 1.5 and P_adj < 0.001 considered biologically significant. *P_adj < 0.001, 1.5 < FC < 2; **P_adj < 0.001, FC > 2; ns, not significant. (H) Line plot showing the averaged median of replicates for origin activity from (G). (I and J) Left: Line plot depicting H3K27me3 (I) and accessible chromatin [ATAC-seq, (J)] signal intensities at replication origins located near PcG bodies in control and ORCA depleted cells, stratified into three genomic domains based on TSA scores (0 to 40, 40 to 70, and 70 to 100). Right: Violin plot showing signal intensities across origins stratified in the indicated TSA genomic domains. Statistical significance was determined by the Wilcoxon paired test, with a FC > 1.5 and P_adj < 0.001 considered biologically relevant. *P_adj < 0.001, 2 < FC < 3; **P_adj < 0.001, 3 < FC < 4; ***P_adj < 0.001, FC > 4; ns, not significant.

Next, we asked whether and how activated origins were distributed within distinct replication timing domains in relation to their proximity to PcG bodies. Most origins in both control and ORCA-depleted cells were activated during the early-to–mid-early replicating genome domain. Further analysis revealed that in wild-type cells, origins from all replication timing domains (early, mid-early, mid-late, and late) were evenly distributed across TSA score bins of 40 to 70 (Fig. 8E). Although the most origins in siORCA cells followed a similar distribution pattern within TSA scores of 30 to 60, ORCA depletion led to the activation of origins more distant from PcG bodies within mid-late replication timing domains, while origins in closer proximity to PcG bodies were enriched in mid-late replicating regions (Fig. 8, E and F, and fig. S8C). A closer analysis revealed that origins activated near PcG bodies in both control and ORCA-depleted cells were predominantly distributed in mid-late replicating domains of the genome (Fig. 8F). In summary, ORCA depletion resulted in significantly higher origin activation at PcG-associated chromatin (RNF2 TSA scores of 80 to 100) in mid-replicating regions compared to control cells (Fig. 8F and fig. S8C).

We then examined whether origin initiation frequency changed in relation to proximity to PcG bodies. To this end, we measured origin frequency as a function of reads per origin within each TSA category and analyzed whether origin strength varied with PcG body proximity in control and ORCA-depleted cells (Fig. 8G). We observed a bimodal distribution of origin activity, with an overall increase in activity in genomic bins with TSA scores of 70 to 100 in ORCA-depleted cells compared to control cells (Fig. 8, G and H). These results suggest that ORCA acts as a repressor of replication origins and their activity in genomic regions proximal to PcG bodies. The number of origins that show no change, up-regulated, or down-regulated and the TSA-score distribution of the differentially activated origins in control versus ORCA-depleted cells are shown in fig. S8D. To assess whether origins were directly regulated by ORCA, we quantified overlap between ORCA ChIP-seq peaks and TSA-defined NS-seq origins (fig. S8E). Permutation testing showed significant enrichment of ORCA at late (TSA 70 to 100) origins (P < 0.0001). Jaccard indices were also highest at this interval (control: 0.0348; siORCA: 0.0289), consistent with preferential ORCA occupancy at late origins relative to early (0 to 40) or mid (40 to 70) intervals. Together with increased median NS-seq signal at TSA 70 to 100 origins upon ORCA depletion, these results support a direct role for ORCA in regulating origins distributed near PcG (RNF2 TSA 70 to 100).

To investigate the epigenetic marks associated with altered replication initiation activity near PcG bodies, we analyzed both active and repressive histone modifications, specifically H3K4me3, H3K27me3, and H3K9me3. In mammalian cells, H3K4me3 is enriched at replication origins (40, 41). Our data confirmed that H3K4me3 signals colocalized (within 1 kb) with replication origins (fig. S8F). Further, we observed that H3K27me3 and H3K9me3 were depleted at origins in control and siORCA cells (fig. S8G). Lower signal intensity of the repressive marks observed in siORCA cells was due to a significant loss of overall repressive histone marks. Last, we categorized origins into three bins based on their distribution with the TSA scores (0 to 40, 40 to 70, and 70 to 100). We observed an overall reduction of H3K27me3 and H3K9me3 across all three bins, consistent with the overall loss of these histone modifications in ORCA-depleted cells (Fig. 8I and fig. S8H). However, the loss was found to be highly significant at origins in TSA score bin 70 to 100. We did not observe significant changes in chromatin accessibility ±1 kb from the origin center across all TSA bins (Fig. 8J). Our results support the model that ORCA interacts with the polycomb complex–bound chromatin to maintain a repressed heterochromatin environment that restricts replication origin firing and ensures accurate replication timing.

DISCUSSION

In the current study, we use a powerful genomic tool, TSA-seq to map the genomic regions that are in proximity to the PcG bodies. We demonstrate the epigenomic features associated with the chromatin that is near versus far from the PcG bodies. Specifically, we show an enrichment of repressive marks, often present at heterochromatic regions, H3K27me3 and H3K9me3, coinciding with gene repression at regions proximal to PcG bodies. Also, we find enrichment of late-firing origins near PcG bodies. Further, the replication initiator factor, ORCA, involved in maintaining a heterochromatic environment associates with the polycomb proteins. The loss of ORCA results in decompaction of heterochromatic sites, prominently next to the PcG bodies, resulting in altered origin firing. These observations suggest that changes in the chromatin environment, proximal to PcG bodies, influence replication origin initiation. PRC1 and PRC2 can establish long-range chromatin interaction; however, the relationship between PcG bodies, 3D-chromatin architecture, and gene repression remains unclear (42–45). We suggest that the association of ORCA to the polycomb proteins enables efficient H3K27me3 deposition and eventual aggregation of these sites to form PcG bodies (46–48). We conclude that chromatin organization is critical not only to regulate gene expression but also for replication origin activity.

Several centromeres and telomeres are positioned in proximity to the PcG bodies, a finding that has not been appreciated. The first report of PcG complex forming the PcG bodies in a series of human cell lines showed that it associated with pericentromeric sequences of chromosome 1, specifically 1q12 (7), an observation we confirm from our TSA-seq experiments. It is known that the chromosome 1 pericentromeric region is enriched for satellite repeats, the majority being satellite 2 and a minor fraction of satellite 3 (49). We note that centromeres of chromosomes 1, 2, 10, and 16 are in proximity to the PcG body. Pericentromeric regions are enriched for satellite DNAs such as satellite II, whereas the centromeric core is composed of alpha-satellite DNA organized into higher-order repeats. Prominently, these centromeres and pericentromeric regions are enriched for H3K27me3 and H3K9me3 domains, supporting the argument that both these repressive marks seem to be needed for tethering these regions to the PcG bodies. Consistent with these observations, we note that the centromeres that are not in proximity to the PcG bodies lack one or the other repressive mark. For example, the pericentromeric region of chromosomes 7, 15, 17, and 22 is also known to harbor satellite 2 but not to the same extent as some other chromosomes. We acknowledge that although centromeric sequences are now included in the T2T genome assembly, these regions are highly repeat-rich and composed of extensive repetitive DNA, which makes accurate mapping of short sequencing reads challenging. Using the GRCh38 (hg38) reference for alignments, we observe strong signal adjacent to centromeric assembly gaps, with enrichment flanking these regions. This pattern indicates that the signal detected at centromeric/pericentromeric regions in the T2T alignments is not random but reflects genuine enrichment, highlighting the advantage of using the T2T assembly for these analyses. Similar observations have been reported in nucleolar TSA-seq studies. Nucleolus-associated chromatin is also highly repetitive and heterochromatic, and using the T2T genome assembly, the strongest signal was observed over the nuclear organizer region (NOR) regions and the second highest over the non-NOR p-arm (21). Telomeric and subtelomeric heterochromatin is also enriched for H3K9me3 and can also contain H3K27me3. We have previously demonstrated that ORCA is enriched at telomeres and plays a key role in establishing heterochromatin (13). PRC2 and H3K27me3 have been implicated in cooperating with the repressive H3K9me3 to establish large heterochromatin domains that are decorated with HP1 (27). Similarly, in Drosophila, telomere-associated sequences can recruit polycomb-group proteins, and this has been suggested to be needed for telomere silencing (50).

Our findings are in line with the recent observations that the loss of polycomb binding mediates long-range effects on H3K27me3 deposition and genome architecture (51). The loss of PRC2 in stem cells and a developmental context have shown their important role in long-range chromatin interaction and controlling gene expression in vivo (44). This work further proposed that developmental gene loci act as nucleation sites from which H3K27me3 spreads across chromatin domains, analogous to the propagation of H3K9me3 domains observed at centromeres and telomeres (52). Loss of H3K27me3 at the polycomb target genes affects the integrity of the PcG bodies, likely due to increased acetylation causing elevated enhancer-promoter looping as observed at the HOXD locus (53). Disruption of PcG bodies can also occur because of the decompaction of polycomb target loci (54), which is primarily because of an imbalance in H3K27me3 and H2AK119ub1 that cause the removal of these marks from the target sites. We find that the ORCA, a heterochromatic-associated protein, is critical for the stabilization of the PRC2. Loss of ORCA leads to reduced H3K27me3 and H3K9me3 and reorganization of the PcG body. We observe similar changes to the PcG body architecture upon the loss of EZH2.

Work from the Mechali laboratory (55) has shown that the efficient origins are associated with open chromatin and are also enriched for polycomb proteins. Further, the Aladjem laboratory showed that late-replicating origins are associated with heterochromatin marks (41). More recently, the Dutta group identified that H3K27me3 is present at origins from multiple datasets (56), and this is consistent with the presence of EZH2 near these origins. The study showed that one class of origins contained >80% RNF2-binding sites, 75% SUZ12-, and 90% EZH2-binding sites, proposing the model that PcG factors and open chromatin domains (marked by the bivalent H3K27me3 and H3K4me3) could play a key role in the choice of some origins in pluripotent cells. All the above point to specific chromatin features that might dictate origin function. The enrichment of SUZ12 at the center of the initiation sites offered evidence that the PcG proteins are likely involved in origin activation rather than preRC establishment.

The presence of ORCA at the PcG bodies, together with the enrichment of replication origins at defined distances from structures, supports a model in which PcG-associated chromatin organization influences origin activation. The loss of PRC2 components has also been shown to increase the number of origins, likely due to the activation of dormant origins (57). These results suggested that the PcG proteins contribute to restricting origin activation within the polycomb domains. The loss of ORCA results in the loss of H3K27me3 and disruption of the PcG bodies, which, in turn, affects replication origin activation. These findings are consistent with observations that the loss of EZH2 leads to H3K27me3 loss, increased open chromatin, and increased replication initiation at a select group of origins, specifically at the “EZH2-bound sites” (58). EZH2-bound gene promoters remained repressed despite the loss of EZH2 but were associated with the up-regulation of origin activity at TSS. The authors proposed a model in which the exposed DNA forms a site of ORC binding at the EZH2-binding site, resulting in increased replication initiation. In light of our current findings, we refine this model to propose that ORCA associates with EZH2 and enables the establishment of H3K27me3 and H3K9me3 domains proximal to the PcG bodies (Fig. 9). This enables heterochromatin organization of centromeres and telomeres and accurate replication initiation of the late-firing origins. We conclude that ORCA association to PcG bodies is a critical regulator of replication timing control.

Fig. 9. Schematic representation of chromatin organization around PcG bodies in control and ORCA-depleted cells.

Note the reorganization of PcG bodies in cells lacking ORCA and consequently firing of a number of origins. The destabilization of the PcG complex is not shown in the image. Created in BioRender. Chetlangia, N. (2026) https://BioRender.com/rl54w33.

MATERIALS AND METHODS

Plasmid constructs

Myc-EZH2 was provided by F. Fuks, Free University Brussels. EZH2 was polymerase chain reaction (PCR) amplified and cloned into pEYFP-LacI vector.

pEGFP-LacI vector was a gift from M. Dundr (59) and used for making pEYFP-LacI vector. T7-ORCA, pEYFP, and cyan fluorescent protein (CFP)–ORCA, pECFP-LacI, and pECFP-LacI-ORCA have been described previously (10).

Cell culture

U2OS cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) containing high glucose, supplemented with 5% fetal bovine serum (FBS). WI38 was also grown in DMEM with 10% FBS.

Depletion of ORCA by small interfering RNA (siRNA) [two independent siRNA oligonucleotides were used (10)]: Cells were grown to 30% confluency. siRNA against control luciferase gene or ORCA or EZH2 was delivered into the cells using Lipofectamine RNAimax at a concentration of 100 nM. siRNA was transfected twice at intervals of 24 hours, and the cells were used for subsequent experiments 24 hours after second round of transfection. Chromatin fractionation was performed as per standard protocol (60).

Immunofluorescence

Fifty to sixty % confluent cells on coverslips were fixed with 2% paraformaldehyde (PFA) in phosphate-buffered saline (PBS, pH 7.4) for 15 min at room temperature (RT), followed by permeabilization with 0.5% Triton X-100 in PBS for 5 min on ice or preextracted before fixing with 0.5% Triton X-100 in cytoskeletal buffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, at 10 mM Pipes at pH 6.8) for 5 min on ice. Cells were blocked with 1% normal goat serum (NGS) for 30 min. Primary antibody incubation was carried out in a humidified chamber for 1 hour, followed by 5-min wash, three times, and followed by secondary antibody incubation for 45 min at RT. After 5-min wash three times, the cells were stained with DAPI (4′,6-diamidino-2-phenylindole) and mounted with VECTASHIELD (Vector Laboratories Inc., Burlingame, CA).

Immunoprecipitation and immunoblots

Cotransfections were done in U2OS cells with cMYC-EZH2 or YFP-BMI1 and T7-ORCA. Cells were lysed 48 hours posttransfection using radioimmunoprecipitation assay (RIPA) lysis buffer with protease and phosphatase inhibitors. Lysates were precleared with Gammabind G Sepharose beads for 1 hour and incubated overnight at 4°C with appropriate antibody. Next day, the lysates were precipitated by Gammabind G Sepharose beads for 2 hours, and beads were washed thrice with RIPA buffer and finally denatured by addition of Laemmli buffer. The complex was analyzed by Western Blotting. For immunoprecipitations and immunoblots, the following antibodies were used: EZH2 (1:500), T7 (1 μg/μl), H4 (1:1000), H3K27me3 (1:1000), H4K20me3 (1:1000), H3K27me2 (1:1000), Suz12, RbAp, EED (1:1000, Cell Signaling Technology, Danvers, MA), BMI1 (1:500) and RNF2 (1: 2000), anti–α-tubulin (1:10,000, Sigma-Aldrich).

Proximity ligation assay

PLA was performed using Sigma-Aldrich Duolink PLA as per the manufacturer’s protocol. Briefly, U2OS cells on coverslips were fixed with 2% PFA for 15 min at RT and permeabilized on ice with PBS containing 0.5% Triton X-100. Coverslips were then blocked at 37°C for 1 hour and incubated with primary antibodies at 37°C for 1 hour. After washing in buffer A, the coverslips were incubated with Duolink PLA probes anti-mouse MINUS and anti-rabbit PLUS for 1 hour at 37°C. The coverslips were then washed in buffer A, and ligation was performed using ligation reaction mixture at 37°C for 30 min. Next, the coverslips were washed in buffer A and incubated in amplification reaction mixture for 100 min at 37°C in the dark. The coverslips were subsequently washed twice in buffer B and once in 0.01× buffer B before DAPI staining.

DNA FISH

U2OS cells on coverslips were fixed with 4% PFA in PBS for 15 min. Cells were quenched with 0.5 M glycine in PBS for three washes of 5 min each. The cells were then permeabilized with 0.5% Triton X-100 for 10 min at RT. Coverslips were blocked with 5% NGS (in PBS) for 30 min followed by primary antibody incubation for 2 hours at RT. After three PBS washes, coverslips were incubated in respective secondary antibodies for 1 hour at RT. After three PBS washes, the cells were postfixed with 4% PFA for 10 min, followed by three washes of 0.5 M glycine. Coverslips were incubated with ribonuclease A (RNase A, 0.1 mg/ml) at 37°C. After three PBS washes, the cells were treated with 0.1 N HCL in 0.5% Triton X-100 on ice for 10 min. After a quick rinse with 2× SSC, the coverslips were incubated in 50% formamide and 2× SSC for 30 min at RT. The recipe for FISH probes is [salmon sperm DNA (10 mg/ml), yeast tRNA (10 mg/ml), human Cot-1 DNA (1 mg/ml), and nick-translated FISH probe]. FISH probes were dried at 40°C using SpeedVac, after which they were dissolved in 10 μl of formamide at 37°C for 10 min. Next, 2 μl of 20× SSC, 4 μl of H₂O, and 4 μl of 50% dextran sulfate were added to each probe (for a 20-μl mix) for each coverslip. The coverslips with FISH probes were heat denatured at 75° to 80°C for 5 min. Hybridization was carried out overnight in a humidified chamber at 37°C. Coverslips were washed three times in 50% formamide +2× SSC at 42°C, followed by three washes with 2× SSC at 42°C and followed by DAPI in 2× SSC. The final wash of 4× SSC was done at RT. The coverslips were mounted using Vectashield. Table 1 provides details of the probes used for DNA FISH used in Figs. 1, 4, 5, and 6.

Table 1. DNA FISH PROBE DETAILS.

Chromosome location	BAC probe	TSA score control	TSA score siORCA
1q12 Satellite III DNA	Vysis CEP 1	100	100
chr8:85,274,962-85,350,486	RP11-1122D5	80–90	80–90
chr1:144,788,251-144,891,229	RP11-466H10	70–80	60–70
chr12:51,918,738-52,057,441	RP11-95A3	40–50	30–40
chr3:94,191,188-94,374,567	RP11-9A9	0–10 (6)	10–20 (14)
chr14:101159600–101161491	RP11-655A19	50–60 (56)	30–40 (37)

smFISH

XIST probes were ordered from Stellaris. smFISH probes were added to Stellaris RNA FISH hybridization buffer (LGC Biosearch Technologies) with 10% formamide at a final concentration of 125 nM. Hybridization was carried out in a humidified chamber in the dark for 6 hours at 37°C. The coverslips were washed with Stellaris RNA FISH Wash Buffer A and mounted with Vectashield (Vector Laboratories).

Chromatin immunoprecipitation

Formaldehyde (Sigma-Aldrich) was added to culture medium to a final concentration of 1%. Cross-linking was allowed to proceed for 10 min at RT and stopped by the addition of glycine at a final concentration of 0.125 M. Fixed cells were washed and harvested. Chromatin was prepared by two subsequent extraction steps (10 min at 4°C) with buffer 1 [50 mM Hepes/KOH (pH 7.5), 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, and 0.25% Triton X-100] and buffer 2 [200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, and 10 mM tris (pH 8)]. Nuclei were then pelleted by centrifugation, resuspended in buffer 3 [50 mM tris (pH 8), 0.1% SDS, 1% NP-40, 0.1% Na-deoxycholate, 10 mM EDTA, and 150 mM NaCl], and subjected to sonication with Bioruptor Power-up (Diagenode, Denville, NJ) yielding genomic DNA fragments with a bulk size of 150 to 300 bp. Chromatin was precleared with Protein A/G ultralink beads (53,133, Pierce, Grand Island, NY) for 2 hours at 4°C and immunoprecipitation with the specific antibodies carried out overnight at 4°C. Immune complexes were recovered by adding preblocked protein A/G ultralink beads and incubated for 2 hours at RT. Beads were washed twice with low salt buffer [0.1% SDS; 1% Triton X-100, 2 mM EDTA, 20 mM tris (pH 8), and 150 mM NaCl], twice with high-salt buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM tris (pH 8), and 500 mM NaCl], once with LiCl wash buffer [10 mM tris (pH 8.0), 1% Na-deoxycholate, 1% NP-40, 250 mM LiCl, and 1 mM EDTA], and twice with TE + 50 mM NaCl. Beads were eluted in TE + 1% SDS at 65°C, and cross-link was reversed O/N at 65°C. The eluted material was phenol/chloroform extracted and ethanol precipitated.

TSA sequencing

Cells were grown in tissue culture flasks and fixed by pouring away growth media, adding freshly made 1.6% PFA in PBS, and incubating at RT for 20 min. The cells were rinsed with PBS and then washed/permeabilized with 0.5% PBST at RT three times for 5 min. Free aldehyde groups were quenched with 20 mM glycine in PBS at RT three times for 5 min. The cells were then washed with PBS and incubated with 1.5% H₂O₂, hydrogen peroxide, in PBS at RT for 1 hour. The cells were rinsed thrice with PBS, blocked with 5% NGS in PBS, in a volume of 1 ml per 25-cm flask surface area at RT for 1 hour, and then incubated with rabbit anti-RNF2 polyclonal antibody, 1:4000 in 0.1% PBST in a volume of 1 ml per 25 cm at 4°C for 10 to 12 hours. The cells were then washed with 0.1% PBST at RT for three times, 5 min each, and incubated with HRP-conjugated goat anti-rabbit polyclonal antibody 1:1000 in 0.1% PBST in a volume of 1 ml per 25 cm at 4°C for 10 to 12 hours. The cells were washed at RT with 0.1% PBST thrice for 5 min each and with PBS for 5 min. The cells were TSA-labeled with a working solution of 50% sucrose/PBS 4 containing 0.0015% H₂O₂ and the condition-specific tyramide-biotin concentration in a volume of 0.8 ml per 10-cm surface area. The cells were incubated at RT for the condition-specific time and then washed at RT with 0.1% PBST 3 × 5 min and with PBS for 5 min. For each staining, a small portion of attached cells were scraped off and loaded onto coverslips for antibiotin and anti-RNF2 and anti-BMI1 immunostaining to visualize the TSA labeling. The remaining cells were washed with high T-E buffer [10 mM tris and 10 mM EDTA (pH 8.0)] at RT for 5 min and lysed with high T-E buffer containing 1% SDS and Proteinase K (0.2 mg/ml). Cell lysates were collected and immediately subjected to genomic DNA extraction. All incubation and washing steps before cell lysing were done by gently shaking the original flasks. Cells on coverslips were stained with streptavidin–Alexa Fluor 594 (Invitrogen) 1:200 and goat anti-rabbit–fluorescein isothiocyanate (Jackson Immuno) 1:500 in GS blocking buffer at RT for 2 hours or at 4°C for 10–12 hours. The cells were then washed with 0.1% PBST at RT for 3 × 5 min. Coverslips were mounted in DAPI containing, antifading medium [DAPI (0.3 μg/ml; Sigma-Aldrich)/10% w/v Mowiol 4-88 (EMD Millipore)/1% w/v DABCO (Sigma-Aldrich)/25% glycerol/0.1 M tris (pH 8.5)]. Genomic DNA was extracted by phenol/chloroform as previously described (6). For samples with low DNA concentration, glycogen (Roche) was added to a final concentration of 0.05 mg/ml to facilitate ethanol precipitation. Isolated DNA was fragmented to 100 to 600 bp using a Bioruptor Pico (Diagenode) machine with a mode of 30-s ON–30-s OFF. Inspection of DNA gels determined the number of required sonication cycles. Biotin labeled DNA fragments were isolated with streptavidin beads as previously described (6). Sequencing libraries were constructed using the TruSeq ChIP Sample Prep Kit (Illumina, IP-202-1012) as per the manufacturer’s protocol. Libraries were amplified by 8 to 12 PCR cycles. Quantitative PCR was used to measure the concentrations of each library, which were then pooled at equimolar concentration for each lane and sequenced (Illumina). Fastq files were generated and demultiplexed with the bcl2fastq Conversion Software (Illumina).

ATAC sequencing

U2OS cells were grown in 10-cm dish and transfected with siRNA oligonucleotides (siGl3 or siORCA). The cells were trypsinized and resuspended in PBS. A total of 50,000 cells were counted and pelleted; after removing the supernatant, the pellet was washed briefly with 100 μl of PBS, and then it was resuspended in 100 μl of ATAC lysis buffer. The resuspended pellet is transferred to a PCR tube on ice. PCR tube was spun at 500g for 10 min, and the supernatant was carefully removed without disturbing the cell pellet. Tagmention mix [25 μl of 2× tagmention buffer, 10× PBS (2 μl), 1.0% digitonin (0.5 μl), 10% Tween 20 (0.5 μl), H₂O (12 μl), and transposomes (10 μl)] was added to the nuclei and incubated at 37°C for 30 min in a thermomixer at 800 rpm. After tagmentation, 250 μl of DNA purification binding buffer and 5 μl of 3 M sodium acetate is added to each tube. Samples are mixed, placed in a column, and centrifuged at 14,000 rpm for 1 min. The flow-through was discarded, and empty wash was performed at 14,000 rpm for 2 min to remove residual wash buffer. The DNA bound to column is eluted by 35 μl of DNA purification elution buffer.

CUT&RUN sample processing and bioinformatic analyses

CUT&RUN was performed using a CUTANA ChIC/CUT&RUN kit (Epicypher, 14-1048) according to the manufacturer’s instructions. A total of 0.5 million cells were light-fixed with 0.1% formaldehyde for 1 min before quenching with 0.125 M glycine. The cells were then permeabilized, bound to concanavalin A magnetic beads, and incubated overnight at 4°C with primary antibodies using a 1:50 dilution. Primary antibodies used were RNF2, BMI1, and EZH2 immunoglobulin G control (Epicypher, 13-0042). After overnight incubation and washing to remove excess antibodies, protein A/G-MNase was added to the antibody-bound beads and activated with calcium at 4°C for 2 hours to cleave target-bound genomic DNA. The samples were then reverse cross-linked under proteinase K treatment overnight at 55°C. CUT&RUN DNA was isolated with column chromatography, and libraries were prepared using an NEBNext Ultra II DNA Library Prep kit for Illumina (New England BioLabs). Libraries were sequenced on an Illumina NovaSeq 6000 with paired-end 150-bp reads.

Trimmed reads were aligned to T2T reference genome using Bowtie2 with options: --local --very-sensitive-local --no-unal --no-mixed --no-discordant --phred33 -I 10. Also, the reads were mapped to the Escherichia coli genome assembly to quantify E. coli spike-in DNA. Low-quality reads (MAPQ <30) were removed using SAMtools. Peak from each replicate was called by magnetic-activated cell sorter (MACS) with stringent parameter. Bedtools intersect (1-bp overlap) was used to find the high confidence overlapping peaks between two replicates. Differential peaks were identified by DiffBind-DEseq2, with an FDR <0.05 and |fold change| >1.5 cutoff.

NS DNA sequencing (NS-seq)

U2OS cells without or with siORCA were harvested, genomic DNA was purified, and NSs were isolated, as described previously (61). Briefly, DNA was denatured by boiling for 10 min, immediately cooled on ice, and fractionated on a neutral sucrose gradient. Fragments of 0.5 to 2 kb (containing NS DNA and broken genomic DNA) were collected and treated with λ exonuclease to remove non–RNA-primed broken genomic DNA. The remaining single-stranded NS DNA were converted to double-stranded DNA using the BioPrime DNA Labeling System (Thermo Fisher Scientific, 18094011). Double-stranded nascent DNA (1 μg) was sequenced using the Illumina genome analyzer II (Solexa). RNAse A pretreated followed by λ exonuclease were used for peak calling.

NS-seq analysis

Sequencing analyses were conducted following established protocols. Raw FASTQ sequencing files underwent initial trimming with Trimmomatic (version 0.36) and Trim Galore (version 0.4.5) to remove low-quality reads. Quality checks were performed using FastQC (version 0.11.5). Trimmed reads were aligned to the T2T genome using the BWA aligner (version 0.7.17) and then normalized to each other using BAMscale (62) “—scale smallest” to generate normalized BigWig files. Peak calling for NS-seq was carried out using the MACS2 (version 2.1.1.20160309) method, and peaks were called against genomic DNA. Replicated NS peaks were identified by intersecting two replicate datasets and extracting common peaks from both control and siORCA samples. Subsequently, peaks from the control and siORCA groups were merged to create a comprehensive origin peak, which was used for quantifying NS-seq signal. NS-signal quantification was performed using deeptools’ “multiBigWigSummary -BED-file” with comprehensive origin peak as BED file using normalized BigWig files to generate count matrix. Violin plots were generated using R with ggplot2.

TSA binning

FastQ files were analyzed using the TSA pipeline developed by Belmont laboratory (https://github.com/lgchang27/TSA-Seq-2020v2) modified by aligning to T2T genome assembly. TSA bigwig files were smoothed to 20-kb window using the code for smoothing in the same pipeline. The smooth files were then used to generate rescaled TSA score BigWig files. Each of the BigWig files was binned into specified coordinates using R. Briefly, the BigWig files were imported into R using the rtracklayer package, and TSA score values were categorized into 10 quantiles. Each imported BigWig file was then scanned to identify TSA scores within 5-kb regions. Adjunct regions with same scores were combined. Transitions between quantiles were recorded as TSA regions specific to each quantile and exported as BED files.

We first intersected the genomic regions corresponding to each TSA bin with H3K27me3 peaks using bedtools intersect. We then calculated the total number of base pairs in the resulting intersected regions. This value was divided by the total number of base pairs within the respective TSA bin. Enriched fraction (y axis in Fig. 2, B to E) = (No. of overlapping base pairs between TSA bin and histone modification peaks / number of base pair in TSA bin).

RNA-seq analysis

Total RNA was extracted using RNeasy Mini Kit and DNA was removed using deoxyribonuclease I (Qiagen, catalog no. 74104). The RNA-seq libraries were prepared with the Kapa Hyper Stranded mRNA library kit (Roche). Read 1 aligns to the ANTISENSE strand and Read 2 aligns to the SENSE strand. RNA-seq FASTQ files were aligned to the T2T star index using STAR aligner version 2.7.10b (63). Aligned reads were subsequently sorted and indexed using samtools (v1.17). Read counts per gene for control and siORCA conditions were computed using the featureCount function from the subread package. Genes with zero expression in both conditions were excluded from further analysis. Differential gene expression was assessed with the DESeq2 R package, and results were visualized using heatmaps generated in R (top 10% counts per minute). For repeat transcript analyses, all sequencing files were aligned to human genome (T2T-CHM13v2.0) using bwa (v0.7.12) and SAMtools (v1.17).

Alignment and quality control of raw sequencing data

FastQ files for TSA-seq, ChIP-seq, ATAC-seq, and RNA-seq were generated from Illumina platform from the Roy J. Carver Sequencing facility at UIUC. The quality of reads (50-bp processed reads, single end) from each sample was assessed before and after trimming with FastQC. Trimming of reads was performed with trimmomatic. All sequencing files were aligned to human genome (T2T-CHM13v2.0) using Bowtie2 (v), and the sam files were converted to bam files using SAMtools (v). For ChIP-seq, peaks were called using MACS (broad peak for histone marks and narrow peak for ORCA ChIP-seq). ATAC-seq peaks were called using the narrow option in MACS.

For comparing TSA-seq with ChIP-seq and ATAC-seq was briefly carried out as TSA bed files and others were intersected using BEDtools to find overlapping peaks, the peaks were counted and normalized by the genomic region (in base pair) of the TSA bin. The normalized number was converted to a percentage. All the bar plots were made in R using ggplot. Heatmaps of centromere and telomere overlap with TSA were made using deeptols. Correlation between TSA-seq was done using plotCorrelation function of deeptools. Chromosome snapshots are taken from IGV on human T2T genome. Chromosome plots were made in R using scatter plot from ggplot package, by averaging the TSA score of the control and siORCA samples.

Quantification and statistical analyses

Statistical analyses were performed by two-tailed Student’s t test unless indicated differently in the figure legends. Quantifications are represented as means ± SD. P value asterisk convention: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Further statistical details of experiments can be found in the figure legends.

Statistical analysis for genomic overlaps

TSA-seq chromatin fractions (control and siORCA conditions) were systematically compared against chromatin marks or replication origins without or with stratification of replication timing using bedtools v2.31.1. For each pairwise comparison, we quantified absolute overlap counts, overlap proportions (fraction of TSA regions intersecting peaks), and Jaccard similarity indices to assess genomic colocalization patterns.

Statistical significance was determined through permutation testing, where the entire set of chromatin peak coordinates were randomly redistributed across the T2T CHM13 reference genome using bedtools shuffle -chrom to preserve chromosome-specific distributions, and overlaps with TSA fractions were recalculated. This randomization and overlap calculation process was repeated multiple times to generate a null distribution of expected overlaps. Permutation sample sizes were adaptively scaled based on TSA fraction size: Datasets with <50 regions used 50 to 100 randomization iterations, 50 to 500 regions used 30 to 75 iterations, and larger datasets used 20 to 50 iterations to balance statistical power with computational efficiency. The actual number of permutation iterations used for each comparison is reported in the results (Permutations_Used column).

Two-tailed P values were calculated as P = 2 × min(P_greater, P_lesser), where P_greater and P_lesser represent the proportion of permutation iterations yielding overlaps greater than or equal to, and less than or equal to, the observed overlap count, respectively. Results were classified using standard significance thresholds: *P < 0.05, **P < 0.01, ***P < 0.001.

For cases where no permutation iterations exceeded the observed overlap (P = 0), we report the minimum detectable P value threshold (1/n_permutations), with detection limits ranging from 5.0 × 10^–2 (20 iterations) to 1.0 × 10^–2 (100 iterations) depending on the adaptive permutation number used. Z scores were calculated as (observed − expected)/standard_deviation from the permutation distribution to quantify effect sizes, with |Z| ≥ 2 indicating medium effect sizes and |Z| ≥ 3 indicating large effect sizes. All analyses used the T2T CHM13 v2.0 assembly coordinates and were implemented using bedtools v2.31.1 and R v4.2 with the GenomicRanges package.

TSA-seq scores were computed for 20-kb genomic bins and compared between control and siORCA, siEZH2 conditions. Distributional differences were assessed using the Wilcoxon rank-sum test. To estimate the magnitude of the shift between conditions, the Hodges–Lehmann estimator was calculated as the median of all pairwise differences between TSA scores from the two groups. Effect size was quantified using Cliff’s delta, a nonparametric statistic that measures the degree of stochastic dominance between two distributions. Cliff’s delta ranges from −1 to 1, where values near 0 indicate substantial overlap between groups and larger absolute values indicate stronger separation. Analyses were performed in R using the effsize package.

Supplementary Materials

The PDF file includes:

Figs. S1 to S8

Legend for table S1

Other Supplementary Material for this manuscript includes the following:

Table S1