Unraveling the cohesin-chromatin interface: identifying protein interactions that modulate chromosome structure and function

Rittenhouse, Natalie L.; Gohil, Riya; Arricastres, June E.; Dowen, Jill M.

doi:10.1186/s13072-025-00596-4

Research
Open access
Published: 02 June 2025

Unraveling the cohesin-chromatin interface: identifying protein interactions that modulate chromosome structure and function

Epigenetics & Chromatin volume 18, Article number: 31 (2025) Cite this article

2160 Accesses
1 Citations
2 Altmetric
Metrics details

Abstract

Background

The evolutionarily conserved cohesin complex is a pleiotropic regulator of chromosome structure and function, participating in sister chromatid cohesion, transcriptional regulation of genes, DNA replication, and DNA repair. Cohesin uses ATP hydrolysis to dynamically extrude DNA loops that bring together cis-regulatory elements and thus regulate gene expression. Some DNA loops are anchored by the binding of CTCF insulator proteins which can stall extruding cohesin complexes, however many DNA loops that connect enhancers and promoters lack CTCF and it is unclear how cohesin is stabilized at these cis-regulatory sites. While cohesin has been found to co-purify with a number of proteins, some of which regulate cohesin function, our current knowledge of cohesin activity is incomplete. Identification of transient or less stable interactions between cohesin and chromatin-associated proteins is crucial for understanding regulation of gene expression and chromosome structure.

Results

Here we utilize a TurboID proximity labeling and mass spectrometry approach for identifying cohesin-interacting proteins. We identify > 400 cohesin-interacting proteins in NIH-3T3 cells, including previously known and potentially novel cohesin interactors. Among the cohesin interactors were chromatin remodeling complexes and histone-modifying complexes. Interactions between seven of these chromatin regulating complexes and cohesin were confirmed with co-immunoprecipitations performed in multiple cell lines. The SWI/SNF complex was found to co-purify with cohesin and SWI/SNF co-occupied enhancers and promoters with cohesin. To investigate the functional relevance of the cohesin-SWI/SNF interaction, we assessed whether the binding of cohesin to the genome is regulated by SWI/SNF or vice versa. Acute small molecule perturbations of SWI/SNF altered the amount of both SWI/SNF and cohesin on chromatin, particularly affecting cohesin binding to CTCF sites.

Conclusions

This work represents the most comprehensive investigation of cohesin-interacting proteins to date. These results identify a physical link between cohesin and a vast number of chromatin-associated proteins inside of cells, including chromatin remodeling complexes and histone-modifying complexes. Furthermore, these results indicate SWI/SNF activity stabilizes cohesin on chromatin particularly at insulator sites. These cohesin interactome data are a resource for future studies aimed at characterizing the functional interactions between cohesin and numerous chromatin-associated proteins in regulating chromosome structure and gene control.

Background

Cohesin is a pleiotropic regulator of chromosome structure with roles in gene regulation, sister chromatid cohesin, DNA replication, and DNA repair. Cohesin is a ring-like complex formed by three core proteins, the structural maintenance of chromosome proteins SMC1 and SMC3, and the kleisin subunit RAD21. RAD21 serves as the interface between the SMC1/SMC3 heterodimer and the STAG subunit of cohesin and it plays key roles in regulating cohesin-mediated chromosome structure and function [1, 2]. RAD21 is phosphorylated during S/G2/M phases of the cell cycle, which promotes cleavage of RAD21 by separase during the metaphase-to-anaphase transition, thus enabling sister chromatids to segregate during mitosis [3,4,5]. RAD21 also contributes to meiotic chromosome cohesion and segregation, DNA replication, and the DNA damage response roles of cohesin [6,7,8,9].

Cohesin hydrolyzes ATP to power the dynamic extrusion of DNA loops that provide structure to chromosomes. The interaction of cohesin with NIPBL/MAU2 is important for both loading of cohesin onto the genome and stimulating ATP hydrolysis during asymmetric DNA loop extrusion [10,11,12]. Cohesin-mediated extrusion of DNA loops is stalled at CTCF bound sites [13]. These cohesin- and CTCF-mediated DNA loops are important for proper regulation of gene expression since they can form insulated neighborhoods that regulate enhancer-promoter communication [14]. Cohesin is removed from the genome during interphase by WAPL. Despite this knowledge, many questions remain unanswered regarding the role of cohesin in maintaining three-dimensional chromosome structure within the nucleus. How does the chromatin environment influence cohesin activity and function? Exploring the cohesin-chromatin axis is crucial for understanding the relationship between cohesin-mediated three-dimensional genome structure, chromatin state and accessibility, and gene expression inside of living cells.

In this study, we use biotin proximity labeling followed by mass spectrometry to identify a comprehensive cohesin interactome. This unbiased approach identifies many previously reported cohesin interactors detected in immunoprecipitation-mass spectrometry (IP-MS) studies and identifies hundreds of novel cohesin-interacting proteins [15,16,17]. Among the cohesin-interacting proteins were numerous chromatin remodeling complexes, such as SWI/SNF, ISWI, and NuRD, and histone-modifying complexes, such as MLL and Polycomb. SWI/SNF was found to co-purify with cohesin and genome-wide binding analysis reveals significant overlap of cohesin binding sites and SWI/SNF binding sites across the genome, specifically at enhancers and promoters. Perturbation of the SWI/SNF complex with small molecules that either inhibit its ATPase activity or target the complex for degradation altered the binding of both SWI/SNF and cohesin to chromatin. Additionally, perturbation of the SWI/SNF complex altered cohesin localization to specific sites in the genome, primarily at enhancers and CTCF binding sites. These data demonstrate that the chromatin regulator SWI/SNF interacts with cohesin and regulates the binding of cohesin to the genome, and thus indicate a relationship between chromatin structure and cohesin-mediated chromosome structure.

Results

Identification of a comprehensive cohesin interactome

We took an unbiased approach to identify proteins that interact with cohesin, including those that interact transiently, inside of cells by performing TurboID-mediated biotin proximity labeling [18]. First, we created doxycycline inducible transgenic NIH-3T3 mouse fibroblast cell lines that express either a 3xHA-TurboID biotin ligase or a fusion of 3xHA-TurboID to the C-terminus of the human cohesin complex subunit RAD21 (RAD21-3xHA-TurboID) (Fig. 1A). Transgenic cells were treated with doxycycline for 36 h to induce expression of either the 3xHA-TurboID or RAD21-3xHA-TurboID transgenes, and a no doxycycline control treatment was performed in cells that contain the RAD21-3xHA-TurboID transgene (No TurboID). Following expression of the transgene, cells were treated with or without biotin, with the No TurboID control cells only being treated with biotin. The use of 3xHA-TurboID and RAD21-3xHA-TurboID expressing cells that were not treated with biotin allows for detection of any background protein biotinylation that occurs within cells. Cells expressing the 3xHA-TurboID and treated with biotin serve as a control for biotinylation of proteins by the free-floating biotin ligase within the nucleus. The No TurboID condition treated with biotin controls for any native biotinylation of proteins performed by endogenous biotin ligases. The RAD21-3xHA-TurboID expressing cells treated with biotin provide an experimental condition in which the biotin ligase is tethered to the cohesin complex, and able to biotinylate proteins that come within a ~ 10 nm radius of cohesin.

Biotinylated proteins from the various control and experimental samples were isolated through cell lysis, nuclear extraction, and enrichment using streptavidin-coated magnetic beads. A quality control check was performed to ensure enrichment of biotinylated proteins by taking 10% of the streptavidin-coated magnetic beads, eluting proteins, and performing a western blot using streptavidin-HRP for visualization of biotinylated proteins (Additional file 1: Figure S1A–C). Samples treated with both doxycycline and biotin had numerous biotinylated proteins in the eluted fraction, while control samples without doxycycline and/or biotin treatment had relatively few biotinylated proteins in all of the fractions. These results indicated strong in vivo biotin labeling of proteins by these constructs and efficient streptavidin enrichment. The remaining 90% of the streptavidin-coated magnetic beads were processed for liquid chromatography tandem mass spectrometry (LC–MS/MS). Biotinylated proteins were identified and the enrichment of biotinylated proteins in the experimental RAD21-3xHA-TurboID + biotin condition relative to two control conditions was determined: (1) The Tier 1 interactome consists of proteins enriched in the RAD21-3xHA-TurboID with biotin condition relative to the 3xHA-TurboID with biotin condition, to identify proteins enriched for cohesin interaction versus the non-targeting active biotin ligase, and (2) The Tier 2 interactome consists of proteins enriched in the RAD21-3xHA-TurboID with biotin condition relative to the No TurboID with biotin condition, to identify proteins enriched for cohesin interaction relative to any native endogenous biotinylation of proteins (Fig. 1A).

Incorporation of the RAD21-3xHA-TurboID fusion protein into cohesin complexes is necessary for identification of proteins that interact with cohesin complexes by TurboID. To test this, cells expressing RAD21-3xHA-TurboID were collected and an immunoprecipitation was performed with antibodies targeting the cohesin complex subunit SMC3, HA, and control IgG (Fig. 1B and Additional file 1: Figure S1D). Subsequent immunoblotting for HA and all three core cohesin subunits (RAD21, SMC3, and SMC1A) revealed that core cohesin subunits co-purify with RAD21-3xHA-TurboID. Approximately 45% of cohesin complex molecules contained the transgenic human RAD21-3xHA-TurboID fusion protein and ~ 55% of cohesin complex molecules contained the endogenous RAD21 protein (Additional file 1: Figure S1E). These results confirm the integration of the RAD21-3xHA-TurboID fusion protein into cohesin complexes.

Our TurboID approach identified known cohesin-interacting proteins and novel cohesin-interacting proteins. Previous systematic analysis of cohesin-interacting proteins utilized IP-MS to identify dozens of cohesin-interacting proteins [15,16,17]. Our Tier 1 and Tier 2 cohesin interactomes identified 412 and 2,047 potential cohesin interactors, respectively (Fig. 1C and Additional file 2: Table S1). The Tier 1 and Tier 2 cohesin interactomes were ~ 2 and ~ 8 times larger, respectively, than the previously identified sets of cohesin interactors detected by immunoprecipitation-mass spectrometry of various cohesin subunits in human HCT116 cells [15]. Our Tier 1 and Tier 2 interactomes included many previously reported cohesin-interacting proteins. Additionally, The Tier 1 and Tier 2 interactomes contained numerous proteins that were not previously reported to interact with cohesin. All proteins identified in the Tier 1 interactome were also present in the Tier 2 interactome. Importantly, known cohesin subunits and regulatory interactors, such as STAG1, STAG2, PDS5A and PDS5B, were significantly enriched in both the Tier 1 and Tier 2 interactomes (Fig. 1D, E). CTCF and YY1 are well established cohesin interactors that were not detected in the Tier 1 interactome or previous IP-MS experiments, yet they were detected in the Tier 2 cohesin interactome. This highlights the fact that the Tier 2 interactome contains transient, or less stable, cohesin-interacting proteins as well as proteins that were enriched over native biotinylation performed by endogenous biotin ligases in the cell (No TurboID condition treated with biotin) but not enriched over biotinylation performed by the free-floating 3xHA-TurboID.

The cohesin interactome captures known and potentially novel biological functions of cohesin

To classify the Tier 1 and Tier 2 cohesin-interacting proteins, Gene Ontology (GO) analysis was performed. Biological processes such as mRNA transport, chromatin and chromosome organization, DNA damage response, and cell cycle, among others, were enriched in both the Tier 1 and Tier 2 interactomes (Fig. 2A and Additional file 3: Table S2). Proteins associated with regulation of chromatin organization, regulation of post-transcriptional gene silencing, and regulation of gene silencing by regulatory ncRNA were highly enriched in the Tier 1 interactome, and also significantly enriched in the Tier 2 interactome. Proteins associated with the well-established cohesin roles of sister chromatid cohesion, cell division, DNA replication, and DNA repair were highly enriched in both the Tier 1 and Tier 2 interactomes (Additional file 1: Figure S2A). Additionally, proteins known to function in chromatin remodeling complexes were highly enriched in the Tier 2 interactome, with many identified in the Tier 1 interactome as well (Fig. 2B). These data both confirm established roles of cohesin and implicate cohesin in new biological functions within cells.

Validation of cohesin interaction with chromatin regulating complexes

The interaction between cohesin and chromatin remodeling complexes is poorly understood and of interest for understanding the role of cohesin in gene control and chromatin state. We identified major chromatin regulating complexes (CRCs) in both the Tier 1 and Tier 2 cohesin interactomes, including chromatin remodeling complexes and histone-modifying complexes (Fig. 2C, D and Additional file 1: Figure S2B). The chromatin remodeling complexes SWI/SNF, ISWI, and NuRD were among the most enriched complexes with at least 50% of their subunits identified in either the Tier 1 or Tier 2 interactomes. The histone-modifying complexes MLL and NuA4, which add methyl and acetyl groups to histone tails at sites of active chromatin, were greatly enriched within the Tier 1 interactome. The repressive Polycomb group complexes (PcG) and mSin3A complex involved in deacetylation of proteins were also enriched within both cohesin interactomes.

To confirm the interaction of cohesin with various CRCs, we performed co-immunoprecipitations of cohesin from NIH-3T3 and mouse embryonic stem cell (mESC) nuclear extracts and immunoblotted for CRC subunits (Fig. 2E, F and Additional file 1: Figure S2C, D). The presence of SWI/SNF complexes was assessed by immunoblotting for SMARCC1, PBAF subcomplex specific subunits PBRM1 and ARID2, and cBAF complex specific subunit ARID1 A. All SWI/SNF subunits were found to co-purify with cohesin, confirming the TurboID results identifying them as cohesin-interacting proteins. The ISWI complex subunit BAZ1B also co-purified with cohesin in both cell types, as did the NuRD complex subunits CHD4 and MTA1. The MLL-associated proteins WDR5 and RBBP5 co-purified with cohesin in both NIH-3T3 and mESCs, and the NuA4 complex subunit EPC1 was also found to strongly interact with the cohesin complex in both cell types. Of the multiple PcG subcomplexes, only the PRC2 subcomplex was assessed due to antibody availability, and the PRC2 subunit SUZ12 was confirmed to co-purify with cohesin. Lastly, the interaction of cohesin with the mSin3A complex was confirmed by immunoprecipitation. These data confirm that chromatin remodelers and histone-modifying complexes interact with the cohesin complex in multiple cell types.

Cohesin co-localizes with SWI/SNF at enhancers, promoters, and CTCF sites

Understanding the functional implications of cohesin interactions with chromatin regulating complexes can provide insight into how cohesin dynamics are regulated. The SWI/SNF chromatin remodeling complex dynamically regulates the accessibility of DNA through ATP-dependent nucleosome shifting and eviction [19]. Active regions of chromatin are continuously being remodeled by SWI/SNF to maintain chromatin accessibility for binding of transcription factors and transcriptional machinery [20]. Interestingly, studies of the SWI/SNF interactome have identified cohesin subunits and known interactors including SMC1A, SMC3, and STAG2 [21, 22]. Additionally, the cohesin complex subunit RAD21 has been found to bind the BRG1 subunit of SWI/SNF in vitro [23].

We investigated whether cohesin and SWI/SNF regulate a set of shared genomic sites by performing and analyzing chromatin immunoprecipitation followed by high throughput sequencing (ChIP-seq). ChIP-seq datasets for the cohesin subunit RAD21, the SWI/SNF subunit BRG1, the histone modification H3K27ac found at active enhancers, the histone modification H3K4me3 found at active promoters, and the transcription factor CTCF were examined. We also performed Micro-C to identify long-range DNA interactions between sites occupied by these factors and PTMs and identified 2,647 topologically associating domains and 4,982 DNA loops in wild-type mESCs. At the Sox2 locus a DNA loop connects the Sox2 gene promoter and a downstream super-enhancer, with many peaks of cohesin, SWI/SNF, CTCF, H3K4me3 and H3K27ac located in the region (Fig. 3A). This analysis identified sites of stable cohesin and SWI/SNF binding across the genome including (1) a class of cohesin binding sites that lack SWI/SNF (green); (2) a class of sites co-occupied by both complexes (purple); and (3) a class of SWI/SNF binding sites that lack cohesin (orange) (Fig. 3A, B). Importantly, a quarter of all cohesin binding sites were shared with SWI/SNF and a tenth of all SWI/SNF binding sites were shared with cohesin, which is significantly more overlap than expected by random chance (Fisher’s exact test, p < 0.00001; ChIPSeeker enrichedPeakOverlap, p < 0.0005) (Fig. 3B). SWI/SNF and cohesin co-bound sites primarily occur at enhancers and promoters that contain CTCF binding sites (Fig. 3C). These data reveal the co-enrichment of cohesin and SWI/SNF at genomic features important for gene regulation. Additionally, the loss or overexpression of SWI/SNF subunits has been shown to correlate with decrease or increase in chromatin interactions, respectively, indicating a potential role for the cohesin-SWI/SNF interaction in linking regulation of chromosome structure and chromatin dynamics [21, 23].

Altered cohesin occupancy at CTCF sites and enhancers upon SWI/SNF perturbation

To determine the functional implications of the cohesin-SWI/SNF interaction and co-localization to regulatory sites, mESCs were treated with one of three small molecules targeting BRG1, the catalytic subunit of SWI/SNF: (1) The ACBI1 PROTAC which recruits an E3 ubiquitin ligase that targets BRG1 for poly-ubiquitination and proteasomal degradation [24]; (2) the bromodomain inhibitor PFI-3 which disrupts the interaction of SWI/SNF with acetylated histone tails and other acetylated proteins [25]; and (3) the ATPase inhibitor BRM014 which inhibits the ATP-dependent nucleosome remodeling activity of SWI/SNF [26], as well as dimethyl sulfoxide (DMSO) as a control (Fig. 4A). Notably, these three small molecules target both BRG1, which is highly expressed in mESCs, and the interchangeable catalytic subunit BRM, which is lowly expressed in mESCs.

Treating mESCs with the ACBI1 PROTAC for 6 h resulted in an ~ 85% decrease in BRG1 protein levels (Additional file 1: Figure S3A). Notably, degradation of BRG1 was accompanied by a ~ 26% decrease in SMARCC1, a core SWI/SNF subunit, and ARID1A, a subunit of the variant complex cBAF. PBRM1, a subunit of the variant complex PBAF exhibited a ~ 65% decrease in protein abundance upon ACBI1 PROTAC treatment, while levels of the ARID2 subunit were unchanged (Additional file 1: Figure S3A, B). The protein levels of cohesin complex subunits SMC1A, SMC3, and RAD21 were largely unaffected by degradation of BRG1 with the ACBI1 PROTAC. These data demonstrate that the ACBI1 PROTAC causes depletion of multiple variant SWI/SNF complex subunits without altering cohesin protein levels in mESCs.

We sought to investigate changes in stable cohesin localization on the genome upon SWI/SNF perturbation. Following the treatments of mESCs with the three small molecules targeting BRG1, the stable binding of cohesin subunit RAD21 was assessed by performing ChIP-seq with two biological replicates. Acute loss of SWI/SNF by ACBI1 PROTAC-mediated degradation has been shown to reduce chromatin accessibility [27,28,29]. Treatment of mESCs with the ACBI1 PROTAC resulted in highly similar RAD21 signal at a union set of RAD21 binding sites detected in either drug treated or DMSO control cells (Additional file 1: Figure S3C). The similar RAD21 signal in treated and control cells indicates that reduced SWI/SNF levels in the nucleus do not cause widespread alterations to stable cohesin binding across the genome. However, differential peak analysis revealed a small number of sites in which RAD21 binding was significantly altered. There were 232 RAD21 peaks with differential signal in ACBI1 PROTAC treated cells, 127 of which have increased RAD21 signal (up RAD21 peaks) and 105 of which have decreased RAD21 signal (down RAD21 peaks) (Fig. 4B, C). Peaks with increased RAD21 signal following ACBI1 treatment often overlap CTCF binding sites that have relatively low signal, while peaks with decreased RAD21 signal overlap with strong CTCF binding sites and active enhancers identified by H3K27ac signal (Fig. 4D and Additional file 1: Figure S3D). Additionally, RAD21 peaks increased upon ACBI1 treatment are most often found in introns and intergenic regions, while decreased RAD21 peaks occur at promoters in addition to introns and intergenic regions (Additional file 1: Figure S3E). Notably, RAD21 peaks with increased RAD21 signal upon ACBI1 treatment more frequently overlap with the anchors of DNA loops compared to non-differential RAD21 peaks of similar signal intensity while RAD21 peaks with decreased signal did not overlap DNA loop anchors more than expected (Fig. 4E). To assess the relationship between SWI/SNF and differential RAD21 signal following ACBI1 treatment, we assessed the distance between each differential RAD21 peak and the nearest BRG1 peak, as well as assessed the distance between the same number of non-differential RAD21 peaks with similar signal intensity and the nearest BRG1 peak (Additional file 1: Figure S3F). Peaks with differential RAD21 signal following ACBI1 treatment were not significantly closer to BRG1 peaks than non-differential RAD21 peaks. These results suggest that acute depletion of SWI/SNF strengthens RAD21 binding to a class of sites that are frequently bound by CTCF and serve as anchors of DNA loops, while also reducing RAD21 binding at a subset of enhancers and CTCF sites.

Treatment of cells with the bromodomain inhibitor PFI-3 has been shown to reduce the binding of SWI/SNF subunit BRG1 to cis-regulatory sites in the genome [30,31,32]. We found that treatment of mESCs with PFI-3 had relatively little effect on RAD21 signal at a set of union RAD21 binding sites identified in either drug treated or DMSO control cells (Additional file 1: Figure S3G). This result suggests that reducing SWI/SNF levels at cis-regulatory sites does not induce widespread changes in stable cohesin binding across the genome. Differential peak analysis identified 23 RAD21 peaks with differential signal, 13 of which have increased RAD21 signal (up RAD21 peaks) and 10 of which have decreased RAD21 signal (down RAD21 peaks) (Fig. 4F). Peaks with differential RAD21 signal following treatment with PFI-3 frequently overlap CTCF binding sites with relatively low signal and not active enhancers (Fig. 4G and Additional file 1: Figure S3H). Additionally, differential RAD21 peaks with PFI-3 treatment are most often found in introns and intergenic regions (Additional file 1: Figure S3I). Notably, RAD21 peaks with differential RAD21 signal upon PFI-3 treatment overlap DNA loop anchors less often than non-differential RAD21 peaks of similar signal intensity (Fig. 4H). We measured the distance between each differential RAD21 peak following PFI-3 treatment and the nearest BRG1 peak, and found no significant difference when compared to non-differential RAD21 peaks (Additional file 1: Figure S3J). These results suggest that inhibition of the BRG1 bromodomain of SWI/SNF causes minimal changes in stable cohesin binding across the genome.

Acute inhibition of SWI/SNF ATPase activity with BRM014 has been shown to reduce chromatin accessibility at enhancers and promoters, and briefly trap BRG1 on the genome before eventual dissociation of inhibited SWI/SNF molecules from chromatin [20, 28, 33]. Treatment of mESCs with BRM014 lead to highly similar RAD21 signal at a union set of RAD21 binding sites identified in either drug treated or DMSO control cells (Additional file 1: Figure S3K). The similar genome-wide RAD21 signal in treated and control cells indicates that loss of SWI/SNF activity does not result in widespread alterations to stable cohesin binding to chromatin. In cells treated with BRM014, there were 88 RAD21 peaks with differential signal, all with increased RAD21 signal (up RAD21 peaks) relative to DMSO control (Fig. 4I, J). Peaks with differential RAD21 signal following treatment with BRM014 frequently overlap CTCF binding sites of relatively low signal, similar to treatment with the ACBI1 PROTAC (Fig. 4K and Additional file 1: Figure S3L). Additionally, differential RAD21 peaks with BRM014 treatment are most often found in introns and intergenic regions, as well as some promoters and transcriptional termination sites (TTS) (Additional file 1: Figure S3M). Notably, RAD21 peaks with increased RAD21 signal upon BRM014 treatment more frequently overlapped with DNA loop anchors than non-differential RAD21 peaks of similar intensity (Fig. 4L). We assessed the relationship between SWI/SNF peaks and peaks of increased RAD21 signal following BRM014 treatment by determining the distance between each differential RAD21 peak and the nearest BRG1 peak. Peaks with increased RAD21 signal following BRM014 treatment did not exhibit a significant difference in distance to the nearest BRG1 peak compared to a control set of non-differential RAD21 peaks of similar signal intensity (Additional file 1: Figure S3N). These results suggest that preventing SWI/SNF-mediated remodeling of chromatin by acute treatment with a small molecule ATPase inhibitor strengthens RAD21 binding to a class of sites that are frequently bound by CTCF and serve as anchors of DNA loops.

Cohesin abundance on chromatin coincides with SWI/SNF abundance

Cohesin and SWI/SNF are thought to be highly processive on chromatin inside of cells, with the majority of molecules actively traveling along DNA and the minority of molecules stably bound to specific sites [34,35,36]. Consistent with this interpretation, ChIP-seq experiments reveal cohesin binding to the anchors of DNA loops, which represent relatively stable binding of cohesin to enhancers, promoters, and CTCF binding sites [14, 37,38,39]. While our prior ChIP-seq experiments identified some changes in these stable cohesin binding events upon acute perturbation of SWI/SNF levels or activity, we sought to determine the extent to which the association of all cohesin molecules (both actively traversing DNA and stably bound to specific sites) with chromatin was altered in an alternative approach. We performed chromatin fractionation following ACBI1, PFI-3, or BRM014 treatment to investigate potential interdependence of cohesin and SWI/SNF in chromatin association (Fig. 5A, B and Additional file 1: Figure S4A–D). In DMSO treated control cells, when chromatin is extracted using a 300 mM NaCl buffer, we found that only ~ 10% of BRG1 and PBRM1 protein was bound to chromatin, while the remaining 90% of these proteins were in the nucleoplasmic fraction. On average, 50% of the SMC1A and SMC3 cohesin proteins were bound to chromatin in 300 mM NaCl buffer (Fig. 5B). Upon treatment with ACBI1, there was both a dramatic decrease in the amount of SWI/SNF in the cells (~ 85% reduction) and the small amount of remaining SWI/SNF showed a ~ 9% median reduction in binding to chromatin (Fig. 5B, Additional file 1: Figure S3A). Notably, cohesin occupancy on chromatin was also reduced by ~ 5% in cells treated with the ACBI1 PROTAC. Cells treated with PFI-3 showed an ~ 7% increase in binding of SWI/SNF to chromatin than DMSO treated cells, and cohesin binding to chromatin was also increased by ~ 12%. Treatment of cells with BRM014 slightly decreased binding of SWI/SNF to chromatin relative to control treated cells by ~ 1% and decreased cohesin binding to chromatin by ~ 3%. These results indicate that small molecule inhibition of the bromodomain of SWI/SNF leads to an increase in both SWI/SNF and cohesin on chromatin, while small molecule degradation of SWI/SNF or ATPase activity inhibition leads to reduced levels of both SWI/SNF and cohesin on chromatin. To test whether the binding of SWI/SNF depends on cohesin, we used siRNAs to deplete cohesin subunit SMC1A and thus reduce cohesin levels on chromatin (Additional file 1: Figure S5A). We performed chromatin fractionation under varying conditions and observed that reduced cohesin levels in the cells did not alter the total amount of SWI/SNF in cells, or the relative abundance of SWI/SNF and the remaining cohesin molecules on chromatin (Additional file 1: Figure S5B, C). Importantly, reducing the levels of cohesin on chromatin did not cause altered SWI/SNF binding to chromatin, indicating that the remodeling activity of SWI/SNF promotes cohesin association with chromatin.

Discussion

In this study, we analyzed proteins that interact with cohesin and assessed the functional relevance of the cohesin-SWI/SNF interaction. By employing TurboID, we discovered hundreds of previously unknown cohesin-interacting proteins, uncovering the largest cohesin interactome dataset to date. These Tier 1 and Tier 2 cohesin interactomes provide a large set of potential cohesin-interacting proteins not detected in previous IP-MS studies, possibly due to their transient or less stable nature. The Tier 1 and Tier 2 cohesin interactomes presented here contain proteins involved in various biological functions, including chromatin remodeling, mRNA transport, regulation of post-transcriptional gene silencing, and regulation of catabolic processes. SWI/SNF and cohesin were found to co-occupy thousands of genomic sites, especially enhancers and promoters. Perturbation of SWI/SNF through BRG1 bromodomain inhibition or targeted degradation of BRG1 reduced stable cohesin binding at a subset of CTCF binding sites and enhancers. Interestingly, both depletion of BRG1 and inhibition of the ATPase-dependent remodeling activity of SWI/SNF increased cohesin enrichment at a subset of CTCF binding sites that serve as DNA loop anchors, indicating that SWI/SNF remodeling activity may be important for maintaining proper three-dimensional genome organization at these sites. Furthermore, SWI/SNF binding to the genome promoted association of cohesin with chromatin. These results not only provide a resource for further study, but also implicate the SWI/SNF chromatin remodeling complex in regulation of cohesin-mediated genome organization.

This study employs acute small molecule treatments targeting SWI/SNF to investigate the effects on cohesin. The ACBI1 PROTAC and the ATPase inhibitor BRM014 have been shown to reduce chromatin accessibility at many sites but also increase chromatin accessibility at a subset of genomic sites [20, 27,28,29, 33]. The genomic sites with increased accessibility after ACBI1 or BRM014 treatment are enriched for the CTCF binding motif [20, 29]. Similarly, the genomic sites with increased accessibility in cells lacking the SWI/SNF subunit SMARCB1 also contain CTCF binding motifs [40]. Consistent with these results we find that sites with increased RAD21 signal following treatment with either ACBI1 or BRM014 are enriched for CTCF binding sites at DNA loop anchors. Furthermore, we show that loss of SWI/SNF binding to the genome decreased the levels of cohesin on chromatin, indicating that SWI/SNF promotes cohesin association with the genome. Loss of SWI/SNF chromatin remodeling activity has also been shown to alter gene expression and differentiation of embryonic stem cells into the neuroectodermal lineage [29, 33]. Together, these results suggest SWI/SNF activity is required for stabilization of cohesin on the genome and proper regulation of cohesin-CTCF mediated genome organization important for maintenance of cell state. The PFI-3 bromodomain inhibitor had been shown to reduce SWI/SNF occupancy at cis-regulatory sites and induce differentiation [30, 41]. We observe a relative increase of SWI/SNF and cohesin on chromatin following PFI-3 treatment but no change in stable cohesin binding at specific sites. These results suggest PFI-3 treatment causes an increase in actively translocating cohesin molecules on the genome, which may alter cohesin-mediated chromosome structure and maintenance of cell state.

The relationship between cohesin and SWI/SNF is likely influenced by several factors. First, SWI/SNF maintains open chromatin environments that promote the loading of cohesin and binding of CTCF [42, 43]. Consistent with this, depletion of SWI/SNF reduced the strength of topologically associating domain (TAD) boundaries and, in neutrophils, SWI/SNF was found to promote recruitment of NIPBL to enhancers [42, 44]. Second, NIPBL can stimulate the activity of chromatin remodelers including human SWI/SNF, thus promoting chromatin accessibility [45]. A yeast study found that NIPBL binds and stimulates RSC-mediated nucleosome remodeling at promoters, creating a nucleosome-free environment ideal for cohesin loading [45]. We observe a relative decrease in cohesin binding to the genome when SWI/SNF is degraded, though whether this is due to a lack of nucleosome-free regions or reduced NIPBL localization is not clear. Third, the bromodomain of BRG1 allows for SWI/SNF interaction with acetylated proteins and cohesin is known to be acetylated, especially at CTCF sites [46]. We observe a relative increase in both SWI/SNF and cohesin binding when the BRG1 bromodomain is inhibited, consistent with a potential bromodomain-mediated interaction of SWI/SNF with acetylated cohesin during chromatin binding. A fourth possibility is that the action of cohesin-mediated DNA loop extrusion and SWI/SNF remodeling of nucleosomes generates DNA twisting and/or torsion that influences the binding and/or properties of the other complex on chromatin [47, 48]. Future studies exploring the role of DNA topology in chromosome structure and dynamics are crucial for understanding the cohesin-SWI/SNF interaction.

Although the findings presented in this study offer valuable insights, several limitations should be considered when interpreting the data and drawing conclusions. First, human RAD21 was expressed in murine NIH3T3 cells, due to reagent availability and the intention to study well synchronized cell populations in the future, and while mouse RAD21 has ~ 96% identity with human RAD21, this may bias or limit the detection of some cohesin interactors [49]. Second, the cohesin interactomes presented here are from asynchronous cells and thus do not interrogate cell cycle-specific roles of cohesin in sister chromatid cohesion or gene regulation [50, 51]. Additionally, cohesin interactions detected by biotin proximity labelling may include transient interactions between cohesin and chromatin-associating proteins as well as direct stable interactions. Further studies are required to determine the direct/indirect nature of these interactions and their functional relevance for the cell. Third, these ChIP-seq studies were performed in biological duplicate, limiting the statistical power to identify differential RAD21 signal in small molecule treated samples compared to DMSO controls, and likely underestimating the number of differential cohesin binding sites [52]. Fourth, while the profiling of cohesin and SWI/SNF binding by ChIP-seq identifies the stable binding sites of these ATP-dependent complexes, CUT&RUN could potentially allow for mapping of dynamically translocating complexes by washing out ATP and thus stalling the cohesin and SWI/SNF complexes at non-stable/transient binding sites across the genome [53]. Lastly, the studies presented here were performed in pluripotent mouse embryonic cells and might not hold true in other cell types with different expression levels of chromatin remodeling complexes. Despite these limitations, the findings presented here improve our understanding of the cohesin-chromatin interaction that shapes chromosome structure and gene regulation.

Conclusions

This work identifies the most complete cohesin interactome to date, including potentially transient interactions not previously known. We find that subunits of chromatin remodeling complexes and histone-modifying complexes are enriched for interaction with cohesin, suggesting a physical and functional link between regulation of nucleosome dynamics and chromosome structure. Perturbation of the chromatin remodeling complex SWI/SNF with small molecules alters the overall association of cohesin with chromatin as well as stable cohesin binding to the genome at specific sites. Future studies can provide additional insights into the mechanisms by which specific chromatin-associated proteins influence cohesin function in a variety of cellular contexts.

Methods

Cell culture

Mouse embryonic stem cells (mESCs; V6.5, male, derived from a C57BL/6(f) × 129/sv(M) cross) were grown under standard conditions [54]. Briefly, cells were grown on gelatinized tissue culture plates with mESCs media containing Knockout DMEM (Thermo Fisher Scientific, 10829–018), 15% fetal bovine serum (VWR, 97068–085), 100 U/ml penicillin, 100 µg/ml streptomycin (Thermo Fisher Scientific, 15140–122), 100 µM beta-mercaptoethanol (Thermo Fisher Scientific, 21–985-023), 1 × non-essential amino acids (Thermo Fisher Scientific, 11140–050), 1 × Glutamax (Thermo Fisher Scientific, 35050–061), and homemade leukemia inducing factor (LIF). NIH-3T3 mouse embryonic fibroblast cells (ATCC, CRL-1658) were grown on tissue culture plates with media containing DMEM (Thermo Fisher Scientific, 11995–065), 10% bovine calf serum (VWR, 10158–358), 100 U/ml penicillin, and 100 µg/ml streptomycin. HEK293T (female human embryonic kidney) cells for spike-in normalization were grown on tissue culture plates with media containing DMEM, 10% bovine calf serum, 1 × Glutamax, 100 U/ml penicillin, and 100 µg/ml streptomycin. TrypLE (Thermo Fisher Scientific, 12–604-039) was used for passaging of all cell lines.

PiggyBac transposon vector construction

To create doxycycline-inducible TurboID vectors, human Rad21 from Addgene plasmid #54248 was cloned into Addgene plasmid #107171 so that 3xHA-TurboID was on the C-terminus of RAD21, followed by Gibson Assembly (NEB, E2611S) to create a plasmid that can express RAD21-3xHA-TurboID. A parent piggyBac transposon vector containing a TRE and an EF1α promoter driving expression of a hygromycin resistance gene was created as previously described (gift from Calabrese Lab) [55]. Both 3xHA-TurboID and RAD21-3xHA-TurboID were cloned behind the TRE of the parent piggyBac transposon vector by digestion with AgeI (NEB, R3552S) and SalI (NEB, R0138S) followed by Gibson Assembly (NEB, E2611S) to generate piggyBac vectors capable of doxycycline-inducible expression of 3xHA-TurboID and RAD21-3xHA-TurboID. The piggyBac transposase (System Biosciences, PB210PA-1) was cloned into pUC19 for plasmid propagation as previously described [56]. Oligonucleotides used for cloning are in Additional file 4: Table S3.

Generation of stable NIH-3T3 TurboID cell lines

NIH-3T3 cells were plated and transfected with pUC19-piggyBac transposase, PB_rtTA_BsmbI (Addgene #126028, gift from Mauro Calabrese), and a piggyBac transposon vector containing either 3xHA-TurboID or RAD21-3xHA-TurboID using Lipofectamine 2000 Reagent (Thermo Fisher Scientific, 11–688-027) according to manufacturer procedure. Two days post transfection, cells were treated with 150 µg/mL Hygromycin (Sigma-Aldrich, 1084355001) and 200 µg/mL G418 (Thermo Fisher Scientific, 10–131-035) for 10 days to select for cells with random insertion of both the PB-rtTA-BsmbI and piggyBac transposon constructs. Cells that survived antibiotic selection were seeded at a low density (1000 cells in a 15 cm cell culture plate). After 4 days of growth, single clonal populations were manually isolated and expanded for ~ 15 days. Clones were treated with 1 µg/mL doxycycline (Sigma-Aldrich, D9891-1G) for 36 h then assessed for expression of 3xHA-TurboID or RAD21-3xHA-TurboID transgenes by western blotting for HA.

Proximity ligation and enrichment of biotinylated proteins

Asynchronous NIH-3T3 cells containing either the 3xHA-TurboID or RAD21-3xHA-TurboID transgenes were grown and treated with 1 µg/mL doxycycline for 36 h, followed by a 2-h 250 µM biotin (Sigma-Aldrich, B4501-1G) treatment. Cells lacking treatment of either doxycycline or biotin were used as controls as shown in Fig. 1A. All treatment conditions were performed in technical triplicate. Following treatment with doxycycline and/or biotin, cells were collected by scraping in PBS, centrifuged, and cell pellets were either frozen for storage at −80 °C or nuclear extracts were prepared. For nuclear protein extraction, cell pellets were resuspended in 200 µL Nuclear Digestion buffer (10 mM HEPES pH 8.0, 10 mM KCl, 0.5% NP-40, 1 × Protease Inhibitor Cocktail (PIC) (Sigma-Aldrich, 11697498001), and 1 × phenylmethylsulfonyl fluoride (PMSF) (Thermo Fisher Scientific, AAJ61473XF) with 1 mM DTT added and incubated at room temperature for 10 min. Cells were then pelleted and washed with 500 µL Nuclear Digestion buffer without DTT, then briefly spun down again for 1 min. Pelleted nuclei were resuspended in 250 µL RIPA + Urea buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1% NP-40, 0.25% deoxycholate, 2 M Urea, 1 × PIC and 1 × PMSF), then supplemented with a final concentration of 0.05U/µL benzonase (Millipore Sigma, E1014) and incubated at 37 °C for 15 min. To stop the benzonase reaction, 250 µL RIPA + Urea buffer with 2 mM EDTA was added to the nuclei and incubated on ice for 5 min. Nuclei were spun down, and the resulting supernatant containing nuclear proteins was quantified by Qubit Protein Quantification (Thermo Fisher Scientific, A50668).

For isolation of biotinylated proteins, 25 µL of streptavidin T1 Dynabeads (Thermo Fisher Scientific, 65601) were prepared by washing three times with 1 mL RIPA + Urea buffer supplemented with 1 mM EDTA, then 500 µg of nuclear protein extract was incubated with the streptavidin T1 Dynabeads overnight at 4 °C. Following overnight incubation of nuclear proteins with streptavidin T1 Dynabeads, samples were placed on a magnet and the supernatant was saved as the flow through fraction. Beads were washed four times with RIPA + Urea buffer supplemented with 1 mM EDTA by rotating for 8 min, then placing on magnet and removing the buffer. Beads were then washed three times with 50 µM Ammonium Bicarbonate pH 7.8 (ABC), rotating 5 min between washes. After the final wash, streptavidin beads were resuspended in 50 µL ABC, 90% of each sample was processed for liquid chromatography with tandem mass spectrometry (LC–MS/MS), while the remaining 10% of biotinylated proteins were eluted off the streptavidin beads by resuspending with Elution Buffer (500 mM Tris pH8.5, 30 mM Biotin, 2% SDS), incubating at room temperature for 30 min with intermittent mixing, adding Laemmli buffer supplemented with DTT (50 mM) and boiling for 10 min prior to western blot analysis.

Liquid chromatography with tandem mass spectrometry (LC–MS/MS)

Biotinylated proteins bound to Streptavidin T1 Dynabeads prepared as previously described were subjected to on-bead trypsin digestion. After the last wash buffer step, 50 µl of 50 mM ammonium bicarbonate (pH 8) containing 1 µg trypsin (Promega) was added to beads overnight at 37 ºC with shaking. The next day, 500 ng of trypsin was added then incubated for an additional 3 h at 37 ºC with shaking. Supernatants from pelleted beads were transferred, then beads were washed twice with 100 µl LC/MS grade water. These rinses were combined with original supernatant, then acidified to 2% Trifluoroacetic acid. Peptides were desalted with peptide desalting spin columns (Thermo) and dried via vacuum centrifugation. Peptide samples were stored at − 80 º C until further analysis. Peptides were reconstituted prior to LC–MS/MS analysis in 25 µl of 2% ACN, 0.1% Formic acid. A pooled sample was created by combining an aliquot of each sample.

LC–MS/MS analysis

Each sample was analyzed by LC–MS/MS using an Easy nLC 1200 coupled to a QExactive HF (Thermo Scientific). The pooled sample was analyzed at the beginning and end of the sequence to assess technical reproducibility. Samples were injected onto an Easy-Spray C18 column (75 μm id × 25 cm, 2 μm particle size) and separated over a 120 min method. The gradient for separation consisted of a step gradient from 5 to 36 to 48% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. The QExactive HF was operated in data-dependent mode where the 15 most intense precursors were selected for subsequent HCD fragmentation. Resolution for the precursor scan (m/z 350–1700) was set to 60,000 with a target value of 3 × 10⁶ ions, 100 ms inject time. MS/MS scans resolution was set to 15,000 with a target value of 1 × 10⁵ ions, 75 ms inject time. The normalized collision energy was set to 27% for HCD, with an isolation window of 1.6 m/z. Peptide match was set to preferred, and precursors with unknown charge or a charge state of 1 and ≥ 8 were excluded.

LC–MS/MS data analysis

Raw data were processed using the MaxQuant software suite (version 1.6.15.0) for peptide/protein identification and label-free quantitation [57]. Data were searched against a Uniprot Reviewed Mouse database (downloaded 07/2021, containing 17,051 sequences) and the MaxQuant contaminants database (245 sequences) using the integrated Andromeda search engine. A maximum of two missed tryptic cleavages were allowed. The variable modifications specified were: N-terminal acetylation and oxidation of Met. Label-free quantitation (LFQ) was enabled. Results were filtered to 1% FDR at the unique peptide level and grouped into proteins within MaxQuant. Match between runs was enabled. Data filtering and statistical analysis was performed in Perseus software (version 1.6.14.0) [58]. The Tier 1 cohesin interactome was made up of proteins significantly enriched in the experimental RAD21-3xHA-TurboID + biotin treatment condition compared to the 3xHA-TurboID + biotin treatment condition with a log2 fold change ≥ 1 and p < 0.05. Additionally, Tier 1 cohesin-interacting proteins were significantly enriched in the experimental RAD21-3xHA-TurboID + biotin treatment compared to both the No TurboID + biotin treatment and the RAD21-3xHA-TurboID without biotin treatment conditions. The Tier 2 interactome was made up of proteins significantly enriched in the experimental RAD21-3xHA-TurboID + biotin treatment condition relative to the No TurboID + biotin treatment condition (Additional file 2: Table S1).

Cohesin-interacting proteins identified in previous IP-MS experiments were overlapped with the Tier 1 and Tier 2 interactomes using DeepVenn [59]. Gene ontology analysis was performed on Tier 1 and Tier 2 cohesin interactomes using the ShinyGO v0.81 software package (FDR < 0.05; GO Biological Processes) [60]. To reduce redundances in GO terms, each term and associated fold enrichment were input into the REVIGO software package for both Tier 1 and Tier 2 associated GO terms [61]. Simplified lists with significance values are in Additional file 3: Table S2. For visualization of select GO terms and associated significance values, bubble plots were created in R using ggplots. To assess which chromatin regulating complexes were enriched in the Tier 1 and Tier 2 cohesin interactomes, subunit compositions of all chromatin regulating complexes were acquired from the EpiFactors database and overlapped with the Tier 1 and Tier 2 cohesin interactomes [62].

Small molecule treatments

For acute degradation of SWI/SNF with the BRG1 PROTAC, ACBI1 (MedChemExpress, HY-128359) was resuspended in dimethyl sulfoxide (DMSO) and mESCs were treated with a final concentration of 1 µM ACBI1 for 6 h. The BRG1 bromodomain inhibitor, PFI-3 (Sigma-Aldrich, SML0939-5MG) was resuspended in DMSO and mESCs were treated with a final concentration of 30 µM PFI-3 for 72 h. The SWI/SNF ATPase inhibitor, BRM014 (MedChemExpress, HY-119374) was resuspended in DMSO and mESCs were treated with a final concentration of 10 µM BRM014 for 3 h. Equal volumes of DMSO were added to cell cultures for corresponding DMSO control samples.

Co-immunoprecipitation

Co-immunoprecipitation assays were performed using the Nuclear Complex Co-IP Kit (Active Motif, 54001) and homemade nuclear digestion buffers for extraction of nuclear proteins. Briefly, cells were collected by scraping in PBS before centrifugation and freezing of cell pellets on dry ice. Cell pellets were stored at − 80 °C until nuclear protein extraction. Nuclei were isolated using the standard kit procedure. Nuclei were lysed using either 300 µL of Nuclear Buffer A (10 mM HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, 350 mM sucrose, 10% glycerol, 1 × PIC) or H250 Nuclear Solubilization Buffer (50 mM HEPES pH 7.9, 250 mM NaCl, 1.5 mM MgCl₂, 25% glycerol, 0.5 mM PMSF, 1 × PIC). Then nuclear extracts were treated with a final concentration of 0.05 U/µL benzonase and incubated at 37 °C for 15 min to degrade nucleic acids. The reaction was stopped by adding 3 µL 0.5 M EDTA and incubated on ice for 5 min prior to spinning down at 5000 × g for 5 min at 4 °C. The supernatant was collected as the nuclear protein fraction and protein levels were quantified using the Qubit Protein Broad Range Assay (Thermo Fisher Scientific, A50668). 200 µg of protein was used per IP following the kit procedure in either 1 × Low Stringency Buffer (75 mM NaCl and 0.1% detergent) or 1 × High Stringency Buffer supplemented with NaCl and detergent (150 mM NaCl and 1% detergent). Protein A Dynabeads (Invitrogen, 10002D) were incubated with antibody 6–8 h prior to overnight incubation with protein extracts at 4 °C. Washing of the beads was performed according to the kit protocol. IP material was eluted off beads by resuspending beads in 50 µL of ChIP Elution Buffer (50 mM Tris pH 8.0, 10 mM EDTA, and 1% SDS) and incubating at 65 °C for 1 h, gently vortexing every 15 min to keep beads in suspension. 20% of eluted proteins and 2% input (4 µg nuclear extract) were assessed by western blot analysis. Antibodies used for IP include: SMC3 (Abcam, ab9263), HA (Cell Signaling Technology, 3724S), and IgG (Bethyl, P120-101).

Western blotting

Cells were collected by trypsinization, counted, and 10 million cells were pelleted before being flash frozen and stored at − 80 °C until protein extraction. Whole cell extracts were generated by resuspending cell pellets in 150 µL Radioimmunoprecipitation Assay (RIPA) buffer containing 50 mM Tris–HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS, supplemented with 1 × PIC, 1 × PMSF, and 250 U/mL Pierce Universal Nuclease (Thermo Fisher Scientific, 88700). Cell suspensions were incubated at room temperature for 15 min, then on ice for 15 min, with brief pipetting to mix every five minutes. Following incubations, cells were centrifuged at max speed for 10 min at 4 °C, and the whole cell protein extract (supernatant) was collected.

Equal volumes of whole cell extract samples were run on 4–20% Tris–Glycine gels (BioRad, 4561094, 4561093, or 4561096) at 225 V for 30 min, then transferred to PVDF membranes (VWR, BSP0161) at 100 V for 75 min. Membranes were blocked for at least 45 min with 5% Blotting-Grade Blocker (BioRad, 1706404) in Tris-buffered Saline with Tween (TBS-T) at room temperature before overnight incubation with primary antibody at 4 °C. Following incubation with primary antibody, membranes were washed 3 times for 10 min with TBS-T prior to incubating with secondary antibody at room temperature for 1 h. Before imaging, membranes were washed 3 times for 10 min with TBS-T, then imaged with Thermo SuperSignal West Pico (Thermo Fisher Scientific, 34577) or Thermo Super Signal West Femto (Thermo Fisher Scientific, 34094) chemiluminescent substrate using an Amersham Imager 600. Protein signal was quantified using ImageStudioLite and box plots were generated using GraphPad Prism.

For Streptavidin-enrichment assay check (Additional file 1: Figure S1A–C), membranes were blocked in 3% BSA in TBS-T for 1 h at room temperature. Membranes were then incubated with Streptavidin-HRP (Cell Signaling Technology, 3999S; 1:1500) for 45 min at 4 °C, followed by washing with TBS-T three times for 10 min. To image the membranes, Thermo SuperSignal West Pico chemiluminescent substrate was used with an Amersham Imager 600.

Primary antibodies used for western blotting include: cohesin subunits SMC1A (Bethyl, A300-080 A), SMC3 (Abcam, ab9263), and RAD21 (Bethyl, A300-080 and Active Motif, 39383); HA (Cell Signaling Technology, 3724S); SWI/SNF subunits SMARCC1 (Cell Signaling Technology 11956S), PBRM1 (Millipore, ABE70), ARID2 (Thermo Fisher Scientific, PA5-35857), ARID1A (Cell Signaling Technology, 12354S), SMARCA4/BRG1 (Abcam, ab110641); ISWI subunit BAZ1B (ABclonal, A9851); NuRD subunits CHD4 (Cell Signaling Technology, 12011S) and MTA1 (Cell Signaling Technology, 5647T); MLL subunits RBBP5 (Bethyl, A300109 A) and WDR5 (Cell Signaling Technology, 13105S); NuA4 subunit EPC1 (ABclonal, A5807); PcG subunit SUZ12 (Cell Signaling Technology, 3737S); mSin3A subunits SAP18 (ABclonal, A43907) and SIN3A (ABclonal, A1577); CTCF (Active Motif, 61312); Actin (Abcam, ab8227); and H3 (Abcam, ab1791). Secondary antibodies used for western blotting include: Goat, anti-Mouse-IgG-HRP (Invitrogen, A16072), Donkey, anti-Rabbit-IgG-HRP (GE Healthcare, NA934-1 mL), Goat, anti-Rabbit-IgG-HRP heavy-chain specific (Invitrogen, A27036), and Mouse, anti-Rabbit-IgG-HRP light-chain specific (Cell Signaling Technology, D4W3E). Representative raw western blots can be found in Additional file 5: Table S4.

Chromatin immunoprecipitation and sequencing (ChIP-seq)

RAD21 ChIP-seq was performed in wildtype mESCs and mESCs treated with SWI/SNF targeting drugs and DMSO controls, while CTCF ChIP-seq was performed in wildtype mESCs as previously described [37, 38]. Briefly, cells (20–30 × 10⁶) were counted and crosslinked with 1% formaldehyde in PBS for 5 min, quenched with 2.5 M glycine, then flash frozen and kept at − 80 °C until chromatin extraction. Crosslinked cells were lysed with 10 mL Lysis 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) by rotating for 10 min at 4 °C. Isolated nuclei were pelleted, then lysed in 5 mL Lysis Buffer 2 (10 mM Tris–HCl pH 8.0, 200 mM NaCl, 1 mM EDTA, and 0.5 mM EGTA) by rotating for 10 min at room temperature. After pelleting, the supernatant was removed and pellets were washed with 5 mL of cold shearing buffer (10 mM Tris pH 7.5, 1 mM EDTA, 0.1% SDS) and spun down again. Chromatin pellets were resuspended in 1 mL shearing buffer with a 5% of HEK293T chromatin spike-in (following the same chromatin extraction procedure) for sonication with a Covaris E220 using the following settings: Duty Factor 5, PIP/W 140, and 200 cycles per burst for 12 min. Chromatin fragments of 200–1000 base pairs were generated. Following sonication, insoluble material was pelleted and removed by spinning for 10 min at 15,000 rpm.

Antibodies were incubated with 30 µL and 50 µL Protein A Dynabeads for 6–8 h and unbound antibody was removed by washing twice with PBS with 1 × PIC prior to addition of chromatin. Chromatin fragments in shearing buffer were supplemented with NaCl and Triton X-100 to be in a ChIP Buffer (15 mM Tris pH7.5, 1.5 mM EDTA, 0.1% SDS, 150 mM NaCl, and 1% Triton X-100) before adding to antibody conjugated beads and rotating overnight at 4 °C. Chromatin from 10 × 10⁶ cells was used per an IP. Following overnight incubation of chromatin with antibody conjugated beads, the beads were washed sequentially with ChIP buffer, Wash Buffer 1 (20 mM Tris–HCl pH 8.0, 500 mM NaCl, 2 mM EDTA, 0.1% SDS, and 1% Triton X-100), Wash Buffer 2 (10 mM Tris–HCl pH 8.0, 250 mM LiCl, 1 mM EDTA, 1% NP-40), and Wash Buffer 3 (10 mM Tris–HCl pH 8.0, 1 mM EDTA, and 50 mM NaCl) by rotating for 5 min at 4 °C and placing on a magnet for each buffer. Chromatin was eluted from beads following the last wash by adding ChIP Elution Buffer (50 mM Tris pH 8.0, 10 mM EDTA, and 1% SDS) and incubating at 65 °C for 1 h, vortexing every 15 min. Eluted chromatin was incubated overnight at 65 °C with 5 µL Proteinase K (New England Biolabs, P8107S) to reverse formaldehyde crosslinks. The next day, DNA was purified using a ChIP DNA Clean and Concentrate kit (Zymo, D5205). Antibodies used for ChIP include RAD21 (Abcam, ab992, targets mouse and human) and CTCF (Active Motif, 61311, targets mouse and human).

Illumina Sequencing libraries were prepared using the Kapa Hyper Prep Kit (Kapa Biosystems, KK8502). Sequencing for wildtype RAD21 and CTCF ChIP-seq was performed on an Illumina NovaSeq 6000 platform and sequencing for BRG1 small molecule ChIP-seq was performed on Illumina NextSeq2000 platform, collecting 2 × 50 bp read pairs per a sample (Additional file 6: Table S5).

ChIP-seq analysis

ChIP-seq analysis was performed using custom scripts that can be found on GitHub (https://github.com/dowenlab). For RAD21 and CTCF ChIP-seq in wildtype mESCs, biological replicates were merged as raw fastq files and reads were aligned to a merged genome containing both the mouse (mm10) and human (hg38) genome assemblies using bowtie (v1.3.1) (parameters -v 2 -p 24 -S -m 1 -best -strata) [63]. The number of multimapping reads that were removed from downstream analysis are indicated in Additional file 6: Table S5. In the merged genome, mouse chromosomes were denoted with a Mchr prefix for distinction from human chromosomes in downstream analysis. Duplicate sequences were removed using samtools (v1.17) markdup (-r -s) [64]. A bam file containing only mouse reads was created using samtools view and converted to bed format with bedtools (v2.25.0) bamtobed, then extended the reads by 200 bp on both sides. To call peaks, extended bed files were analyzed using MACS (v2.1.2) with a false discovery rate of 1% (macs2 callpeak -f BED -g mm -q 0.01) [65]. Peak summits were expanded by 50 bp on either side and peaks that overlap regions prominent with repetitive sequences as defined in the ENCODE mm10 blacklist were removed using bedtools intersect (-v) [66]. In order to remove false positive peaks and establish high confidence peak lists, CTCF samples were filtered for peaks with q-values > 8, while RAD21 peaks were not filtered. The resulting peak files were extended 200 bp on either side and used for downstream analyses. Sequencing statistics and sample information can be found in Additional file 6: Table S5.

Previously published ChIP-seq datasets for BRG1 (GSE198517), H3K27ac (GSE153576), and H3K4me3 (GSE153576) in mESCs were analyzed using custom scripts [33, 39]. H3K27ac and H3K4me3 replicates were analyzed as described above. H3K4me3 peaks were determined using MACS broad peak calling (macs2 callpeak –broad -f BED -g mm -q 0.01). To establish high confidence peak lists, H3K4me3 peaks were filtered for peaks with q-value > 2, while H3K27ac peaks were not filtered. The resulting peak files were extended 200 bp on either side and used for downstream analyses. For BRG1 analysis, replicates were merged and only aligned to the mouse (mm10) genome assembly using bowtie (parameters -v 2 -p 24 -S -m 1 –best –strata –chunkmbs 200). Duplicate sequences were removed, the bam file was converted to a bed file, reads were extended by 200 bp on either side, and peaks were called using MACS as described for RAD21 and CTCF above. The resulting peak file was used for downstream analyses.

For RAD21 ChIP-seq in BRG1 small molecule and DMSO treated mESCs, raw fastq files of biological replicates were aligned to both the mouse (mm10) and human (hg38) genome assemblies using bowtie2 (v2.4.5). Samtools (v1.17) markdup (-r -s) was used to remove duplicate sequences prior to converting the mouse bam file to a bed file using bedtools (v2.25.0) bamtobed and extending the reads by 200 bp on both sides. To call peaks, extended bed files were assessed using MACS (v2.1.2) with a false discovery rate of 1% (macs2 callpeak -f BED -g mm -q 0.01). Peaks in regions known to be rich with sequence repeats as defined in the ENCODE mm10 blacklist was removed using bedtools intersect (-v). To define a high confidence peak set, biological replicate peak files were assessed for reproducibility using IDR (v2.0.3) with a threshold of 0.05. IDR filtered peak files for each drug treatment and respective DMSO were combined using cat and bedtools merge to form a composite peak list for each drug treatment. The resulting peak files were extended 200 bp on either side from the peak summit and used for downstream analyses.

For all datasets, a normalization factor was calculated to account for IP efficiency by first separately counting mouse and human non-duplicate reads with samtools idxstats and awk. Then, a normalization factor (normFactor) was calculated for each dataset using the formula (5/(h/m))*100 where h is the number of human reads in millions and m is the number of mouse reads in millions. The bed file containing mouse reads was converted to a bedgraph file using bedtools genomecov (-bga -scale normFactor) before being converted to a bigwig file with bedGraphToBigWig from ucsctools (v320) [67]. Samples without spike-in normalization (BRG1 and H3K4me3) omitted ‘-scale’ when converting the bed file to a bedgraph file. Z-score normalization of bigwig files was performed using a custom R script from Spencer Nystrom of Dr. Daniel McKay’s lab, before visualization of the data in Figs. 3 and 4.

Signal tracks for ChIP-seq data and Micro-C data were visualized using plotGardener [68]. Peak overlaps were determined using bedtools intersect, and the significance of overlapping cohesin and SWI/SNF peaks compared to all peaks from both datasets was tested with bedtools fisher and the ChIPSeeker enrichedPeakOverlap function (nShuffle = 2000). Heatmpas were generated using z-score normalized bigwig files with deeptools computeMatrix (reference-point) followed by deeptools plotHeatmap (v3.5.4) [69]. UpSet plots were generated in R (v4.1.0) using bedtoolsr and UpSetR using RAD21 peaks (wild type and differential peaks from BRG1 drug treatments), BRG1 peaks, CTCF peaks, H3K27ac peaks, and UCSC transcriptional start sites that overlap H3K4me3 peaks [70, 71]. Peak count matrices were generated using bedtools multicov with bam files for merged replicates (RAD21 from wildtype mESCs, CTCF, H3K27ac, and BRG1) or individual replicates (BRG1 small molecule RAD21 samples) and composite peak files for each BRG1 drug treatment as input. The sum of RAD21 signal for individual replicates from BRG1 drug treated samples were used in generation of XY-plots showing the log2 Drug Treatment signal vs log2 DMSO control signal, which were created in PRISM (v10.3.1). Differential peak analysis was assessed using count matrices for individual replicates of each BRG1 drug treatment and matched DMSO controls with DEseq2 (p < 0.05) [52]. Linked violin plots for differential peaks were made in R with ggplots. Violin plots of BRG1, H3K27ac, and CTCF signal at all RAD21 peaks and differential RAD21 peaks for BRG1 small molecule treatments were made in PRISM using log2 ChIP-seq signal from peak count matrices for BRG1, H3K27ac and CTCF. To test if the distribution of BRG1, H3K27ac, and CTCF signal was significantly different between all RAD21 peaks and differential RAD21 peaks, a non-parametric Mann–Whitney test was performed in PRISM (*< 0.05, ****< 0.0001, ^n.s.not significant). Classification of differential peaks and matched nondifferential peaks at genomic features was assessed with Homer annotatePeaks.pl and visualized using PRISM [72]. Overlap of differential peaks with Micro-C loop anchors was generated with bedtools intersect. The nearest BRG1 peak to a differential RAD21 peak was determined using bedtools closest. Matched non-differential peaks of similar signal to differential peaks were identified using matchRanges in R.

RNAi knockdown

5 × 10⁵ cells wildtype mESCs were plated per well in 6-well plates. 50 nM of siSMC1A reagent (Dharmacon, M-049483–00–0005) or siGLO transfection control reagent (siControl) (Dharmacon, D-001630–01–05) was transfected per well using DharmaFECT1 (Dharmacon, T-2001) following manufacturer’s instructions. 48 h after transfection, cells were collected, counted, and 1 million cells were pelleted before freezing for Micro-C library preparation and nuclear fractionation.

Micro-C and analysis

The Micro-C library was prepared using the Dovetail® Micro-C Kit according to the manufacturer’s protocol. Briefly, the chromatin was fixed with disuccinimidyl glutarate (DSG) and formaldehyde in the nucleus. The cross-linked chromatin was then digested in situ with micrococcal nuclease (MNase). Following digestion, the cells were lysed with SDS to extract the chromatin fragments and the chromatin fragments were bound to Chromatin Capture Beads. Next, the chromatin ends were repaired and ligated to a biotinylated bridge adapter followed by proximity ligation of adapter-containing ends. After proximity ligation, the crosslinks were reversed, the associated proteins were degraded, and the DNA was purified then converted into a sequencing library using Illumina-compatible adaptors. Biotin-containing fragments were isolated using streptavidin beads prior to PCR amplification. Libraries were sequenced on an Illumina NovaSeq platform to generate ~ 1 billion 2 × 50 bp read pairs across two biological replicates (Additional file 6: Table S5).

Initial processing of Micro-C sequencing was performed with the Juicer software package v1.5.6 [73]. Topologically associating domains were identified at 10 kb resolution with the juicer-tools arrowhead program. DNA loops were called using the significant interaction peak caller (SIP) using Knight-Ruiz balanced Micro-C matrices [74]. Overlap of loop anchors with other genomic datasets was assessed using bedtools intersect function [75].

Nuclear fractionation

Following treatment with small molecules targeting BRG1 or DMSO as a control, cells were trypsinized (Gibco, 12604–013) and counted using a Cell Countess. 1 million cells per a condition were used to generate chromatin bound and nucleoplasmic fractions using the Subcellular Protein Fractionation Kit for Cultured Cells (Thermo Fisher Scientific, 7884D). Fractions were extracted following the manufacturer’s instructions for the 50 µL packed cell volume, under varying concentrations of NaCl: 50 mM, 150 mM, and 300 mM. An additional PBS wash was performed between each collection. This process was also performed for siRNA treated mESCs.

SWI/SNF and cohesin levels were assessed by western blotting of the chromatin bound and nucleoplasmic fractions for SWI/SNF subunits BRG1 (Abcam, ab110641) and PBRM1 (Millipore, ABE70), and cohesin subunits SMC1 (Bethyl, A300-080 A) and SMC3 (Abcam, ab9263). Protein signal was quantified using ImageStudioLite (LI-COR, version 5.2.5). The percent chromatin bound protein in the 300 mM NaCl condition was calculated by analyzing the signal in the chromatin bound fraction divided by the total signal in the chromatin bound plus nucleoplasmic fractions (chromatin bound/(chromatin bound + nucleoplasmic) * 100). GraphPad Prism was used to generate box plots and dot plots.

Availability of data and materials

High throughput sequencing datasets created in this article are available at the NCBI GEO data repository under accession number GSE288455 for ChIP-seq data and under accession number GSE288456 for Micro-C data. Code used to analyze sequencing data can be found at https://github.com/dowenlab. Mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD060289 [76].

Abbreviations

LC–MS/MS:: Liquid chromatography tandem mass spectrometry
TRE:: Tetracycline responsive element
GO:: Gene Ontology
mESCs:: Mouse embryonic stem cells
PcG:: Polycomb group
ChIP-seq:: Chromatin immunoprecipitation followed by high throughput sequencing
DMSO:: Dimethyl sulfoxide
CRC:: Chromatin regulating complex
TAD:: Topologically associating domain

References

Nasmyth K, Haering CH. The structure and function of SMC and Kleisin complexes. Annu Rev Biochem. 2005;74(1):595–648.
Article CAS PubMed Google Scholar
Nasmyth K, Haering CH. Cohesin: its roles and mechanisms. Annu Rev Genet. 2009;43(1):525–58.
Article CAS PubMed Google Scholar
Tomonaga T, Nagao K, Kawasaki Y, Furuya K, Murakami A, Morishita J, et al. Characterization of fission yeast cohesin: essential anaphase proteolysis of Rad21 phosphorylated in the S phase. Genes Dev. 2000;14(21):2757–70.
Article CAS PubMed PubMed Central Google Scholar
Hauf S, Waizenegger IC, Peters JM. Cohesin cleavage by separase required for anaphase and cytokinesis in human cells. Science. 2001;293(5533):1320–3.
Article CAS PubMed Google Scholar
Waizenegger IC, Giménez-Abián JF, Wernic D, Peters JM. Regulation of human separase by securin binding and autocleavage. Curr Biol. 2002;12(16):1368–78.
Article CAS PubMed Google Scholar
Xu H, Beasley M, Verschoor S, Inselman A, Handel MA, McKay MJ. A new role for the mitotic RAD21/SCC1 cohesin in meiotic chromosome cohesion and segregation in the mouse. EMBO Rep. 2004;5(4):378–84.
Article CAS PubMed PubMed Central Google Scholar
Watrin E, Peters JM. Cohesin and DNA damage repair. Exp Cell Res. 2006;312(14):2687–93.
Article CAS PubMed Google Scholar
Hou W, Li Y, Zhang J, Xia Y, Wang X, Chen H, et al. Cohesin in DNA damage response and double-strand break repair. Crit Rev Biochem Mol Biol. 2022;57(3):333–50.
Article CAS PubMed Google Scholar
Wu J, Liu Y, Zhangding Z, Liu X, Ai C, Gan T, et al. Cohesin maintains replication timing to suppress DNA damage on cancer genes. Nat Genet. 2023;55(8):1347–58.
Article CAS PubMed Google Scholar
Ciosk R, Shirayama M, Shevchenko A, Tanaka T, Toth A, Shevchenko A, et al. Cohesin’s binding to chromosomes depends on a separate complex consisting of Scc2 and Scc4 proteins. Mol Cell. 2000;5(2):243–54.
Article CAS PubMed Google Scholar
Murayama Y, Uhlmann F. Biochemical reconstitution of topological DNA binding by the cohesin ring. Nature. 2014;505(7483):367–71.
Article CAS PubMed Google Scholar
Barth R, Davidson IF, Van Der Torre J, Taschner M, Gruber S, Peters JM, et al. SMC motor proteins extrude DNA asymmetrically and can switch directions. Cell. 2025;188:749.
Article CAS PubMed Google Scholar
Li Y, Haarhuis JHI, Sedeño Cacciatore Á, Oldenkamp R, Van Ruiten MS, Willems L, et al. The structural basis for cohesin–CTCF-anchored loops. Nature. 2020;578(7795):472–6.
Article CAS PubMed PubMed Central Google Scholar
Dowen JM, Fan ZP, Hnisz D, Ren G, Abraham BJ, Zhang LN, et al. Control of cell identity genes occurs in insulated neighborhoods in mammalian chromosomes. Cell. 2014;159(2):374–87.
Article CAS PubMed PubMed Central Google Scholar
Kim JS, He X, Liu J, Duan Z, Kim T, Gerard J, et al. Systematic proteomics of endogenous human cohesin reveals an interaction with diverse splicing factors and RNA-binding proteins required for mitotic progression. J Biol Chem. 2019;294(22):8760–72.
Article PubMed PubMed Central Google Scholar
Kojic A, Cuadrado A, De Koninck M, Giménez-Llorente D, Rodríguez-Corsino M, Gómez-López G, et al. Distinct roles of cohesin-SA1 and cohesin-SA2 in 3D chromosome organization. Nat Struct Mol Biol. 2018;25(6):496–504.
Article CAS PubMed PubMed Central Google Scholar
Cuadrado A, Giménez-Llorente D, Kojic A, Rodríguez-Corsino M, Cuartero Y, Martín-Serrano G, et al. Specific contributions of cohesin-SA1 and cohesin-SA2 to TADs and polycomb domains in embryonic stem cells. Cell Rep. 2019;27(12):3500-3510.e4.
Article CAS PubMed PubMed Central Google Scholar
Cho KF, Branon TC, Udeshi ND, Myers SA, Carr SA, Ting AY. Proximity labeling in mammalian cells with TurboID and split-TurboID. Nat Protoc. 2020;15(12):3971–99.
Article CAS PubMed Google Scholar
Tyagi M, Imam N, Verma K, Patel AK. Chromatin remodelers: we are the drivers!! Nucleus. 2016;7(4):388–404.
Article CAS PubMed PubMed Central Google Scholar
Iurlaro M, Stadler MB, Masoni F, Jagani Z, Galli GG, Schübeler D. Mammalian SWI/SNF continuously restores local accessibility to chromatin. Nat Genet. 2021;53(3):279–87.
Article CAS PubMed Google Scholar
Wolf BK, Zhao Y, McCray A, Hawk WH, Deary LT, Sugiarto NW, et al. Cooperation of chromatin remodeling SWI/SNF complex and pioneer factor AP-1 shapes 3D enhancer landscapes. Nat Struct Mol Biol. 2023;30(1):10–21.
Article CAS PubMed Google Scholar
Patil A, Strom AR, Paulo JA, Collings CK, Ruff KM, Shinn MK, et al. A disordered region controls cBAF activity via condensation and partner recruitment. Cell. 2023;186(22):4936-4955.e26.
Article CAS PubMed PubMed Central Google Scholar
Shang XY, Shi Y, He DD, Wang L, Luo Q, Deng CH, et al. ARID1A deficiency weakens BRG1-RAD21 interaction that jeopardizes chromatin compactness and drives liver cancer cell metastasis. Cell Death Dis. 2021;12(11):990.
Article CAS PubMed PubMed Central Google Scholar
Farnaby W, Koegl M, Roy MJ, Whitworth C, Diers E, Trainor N, et al. BAF complex vulnerabilities in cancer demonstrated via structure-based PROTAC design. Nat Chem Biol. 2019;15(7):672–80.
Article CAS PubMed PubMed Central Google Scholar
Gerstenberger BS, Trzupek JD, Tallant C, Fedorov O, Filippakopoulos P, Brennan PE, et al. Identification of a chemical probe for family VIII bromodomains through optimization of a fragment hit. J Med Chem. 2016;59(10):4800–11.
Article CAS PubMed PubMed Central Google Scholar
Papillon JPN, Nakajima K, Adair CD, Hempel J, Jouk AO, Karki RG, et al. Discovery of orally active inhibitors of brahma homolog (BRM)/SMARCA2 ATPase activity for the treatment of brahma related gene 1 (BRG1)/SMARCA4-mutant cancers. J Med Chem. 2018;61(22):10155–72.
Article CAS PubMed Google Scholar
Moe KC, Maxwell JN, Wang J, Jones CA, Csaki GT, Florian AC, et al. The SWI/SNF ATPase BRG1 facilitates multiple pro-tumorigenic gene expression programs in SMARCB1-deficient cancer cells. Oncogenesis. 2022;11(1):30.
Article CAS PubMed PubMed Central Google Scholar
Schick S, Grosche S, Kohl KE, Drpic D, Jaeger MG, Marella NC, et al. Acute BAF perturbation causes immediate changes in chromatin accessibility. Nat Genet. 2021;53(3):269–78.
Article CAS PubMed PubMed Central Google Scholar
Hoffman JA, Muse GW, Langer LF, Patterson AI, Gandara I, Ward JM, et al. BRG1 establishes the neuroectodermal chromatin landscape to restrict dorsal cell fates. Sci Adv. 2024;10(9):eadj5107.
Article CAS PubMed PubMed Central Google Scholar
Sharma T, Robinson DCL, Witwicka H, Dilworth FJ, Imbalzano AN. The Bromodomains of the mammalian SWI/SNF (mSWI/SNF) ATPases Brahma (BRM) and Brahma Related Gene 1 (BRG1) promote chromatin interaction and are critical for skeletal muscle differentiation. Nucleic Acids Res. 2021;49(14):8060–77.
Article CAS PubMed PubMed Central Google Scholar
Vangamudi B, Paul TA, Shah PK, Kost-Alimova M, Nottebaum L, Shi X, et al. The SMARCA2/4 ATPase domain surpasses the bromodomain as a drug target in SWI/SNF-mutant cancers: insights from cDNA rescue and PFI-3 Inhibitor studies. Can Res. 2015;75(18):3865–78.
Article CAS Google Scholar
Kim B, Luo Y, Zhan X, Zhang Z, Shi X, Yi J, et al. Neuronal activity-induced BRG1 phosphorylation regulates enhancer activation. Cell Rep. 2021;36(2): 109357.
Article CAS PubMed PubMed Central Google Scholar
Martin BJE, Ablondi EF, Goglia C, Mimoso CA, Espinel-Cabrera PR, Adelman K. Global identification of SWI/SNF targets reveals compensation by EP400. Cell. 2023;186(24):5290-5307.e26.
Article CAS PubMed PubMed Central Google Scholar
Ahmad K, Brahma S, Henikoff S. Epigenetic pioneering by SWI/SNF family remodelers. Mol Cell. 2024;84(2):194–201.
Article CAS PubMed Google Scholar
Klemm SL, Shipony Z, Greenleaf WJ. Chromatin accessibility and the regulatory epigenome. Nat Rev Genet. 2019;20(4):207–20.
Article CAS PubMed Google Scholar
Hansen AS, Pustova I, Cattoglio C, Tjian R, Darzacq X. CTCF and cohesin regulate chromatin loop stability with distinct dynamics. eLife. 2017;3(6): e25776.
Article Google Scholar
Arruda NL, Bryan AF, Dowen JM. PDS5A and PDS5B differentially affect gene expression without altering cohesin localization across the genome. Epigenetics Chromatin. 2022;15(1):30.
Article CAS PubMed PubMed Central Google Scholar
Arruda NL, Carico ZM, Justice M, Liu YF, Zhou J, Stefan HC, et al. Distinct and overlapping roles of STAG1 and STAG2 in cohesin localization and gene expression in embryonic stem cells. Epigenetics Chromatin. 2020;13(1):32.
Article CAS PubMed PubMed Central Google Scholar
Carico ZM, Stefan HC, Justice M, Yimit A, Dowen JM. A cohesin cancer mutation reveals a role for the hinge domain in genome organization and gene expression. PLoS Genet. 2021;17(3): e1009435.
Article CAS PubMed PubMed Central Google Scholar
Sakakura M, Ohta S, Yagi M, Tanaka A, Norihide J, Woltjen K, et al. Smarcb1 maintains the cellular identity and the chromatin landscapes of mouse embryonic stem cells. Biochem Biophys Res Commun. 2019;519(4):705–13.
Article CAS PubMed Google Scholar
Fedorov O, Castex J, Tallant C, Owen DR, Martin S, Aldeghi M, et al. Selective targeting of the BRG/PB1 bromodomains impairs embryonic and trophoblast stem cell maintenance. Sci Adv. 2015;1(10): e1500723.
Article PubMed PubMed Central Google Scholar
Barutcu AR, Lajoie BR, Fritz AJ, McCord RP, Nickerson JA, Van Wijnen AJ, et al. SMARCA4 regulates gene expression and higher-order chromatin structure in proliferating mammary epithelial cells. Genome Res. 2016;26(9):1188–201.
Article PubMed PubMed Central Google Scholar
Valletta M, Russo R, Baglivo I, Russo V, Ragucci S, Sandomenico A, et al. Exploring the interaction between the SWI/SNF chromatin remodeling complex and the zinc finger factor CTCF. Int J Mol Sci. 2020;21(23):8950.
Article CAS PubMed PubMed Central Google Scholar
Zhu Y, Denholtz M, Lu H, Murre C. Calcium signaling instructs NIPBL recruitment at active enhancers and promoters via distinct mechanisms to reconstruct genome compartmentalization. Genes Dev. 2021;35(1–2):65–81.
Article CAS PubMed PubMed Central Google Scholar
Muñoz S, Jones A, Bouchoux C, Gilmore T, Patel H, Uhlmann F. Functional crosstalk between the cohesin loader and chromatin remodelers. Nat Commun. 2022;13(1):7698.
Article PubMed PubMed Central Google Scholar
Van Ruiten MS, Van Gent D, Sedeño Cacciatore Á, Fauster A, Willems L, Hekkelman ML, et al. The cohesin acetylation cycle controls chromatin loop length through a PDS5A brake mechanism. Nat Struct Mol Biol. 2022;29(6):586–91.
Article PubMed PubMed Central Google Scholar
Janissen R, Barth R, Davidson IF, Peters JM, Dekker C. All eukaryotic SMC proteins induce a twist of −0.6 at each DNA loop extrusion step. Sci Adv. 2024;10(50):eadt1832.
Article CAS PubMed PubMed Central Google Scholar
Valdés A, Haering CH. Adding a twist to the loops: the role of DNA superhelicity in the organization of chromosomes by SMC protein complexes. Biochem Soc Trans. 2024;52(6):2487–97.
Article PubMed PubMed Central Google Scholar
McKay MJ, Troelstra C, Van Der P, Kanaar R, Smit B, Hagemeijer A, et al. Sequence conservation of therad21 Schizosaccharomyces pombeDNA Double-strand break repair gene in human and mouse. Genomics. 1996;36(2):305–15.
Article CAS PubMed Google Scholar
Nasmyth K, Peters JM, Uhlmann F. Splitting the chromosome: cutting the ties that bind sister chromatids. Science. 2000;288(5470):1379–84.
Article CAS PubMed Google Scholar
Hnisz D, Day DS, Young RA. Insulated neighborhoods: structural and functional units of mammalian gene control. Cell. 2016;167(5):1188–200.
Article CAS PubMed PubMed Central Google Scholar
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.
Article PubMed PubMed Central Google Scholar
Skene PJ, Henikoff JG, Henikoff S. Targeted in situ genome-wide profiling with high efficiency for low cell numbers. Nat Protoc. 2018;13(5):1006–19.
Article CAS PubMed Google Scholar
Kagey MH, Newman JJ, Bilodeau S, Zhan Y, Orlando DA, Van Berkum NL, et al. Mediator and cohesin connect gene expression and chromatin architecture. Nature. 2010;467(7314):430–5.
Article CAS PubMed PubMed Central Google Scholar
Schertzer MD, Thulson E, Braceros KCA, Lee DM, Hinkle ER, Murphy RM, et al. A piggyBac-based toolkit for inducible genome editing in mammalian cells. RNA. 2019;25(8):1047–58.
Article CAS PubMed PubMed Central Google Scholar
Kirk JM, Kim SO, Inoue K, Smola MJ, Lee DM, Schertzer MD, et al. Functional classification of long non-coding RNAs by k-mer content. Nat Genet. 2018;50(10):1474–82.
Article CAS PubMed PubMed Central Google Scholar
Tyanova S, Temu T, Cox J. The MaxQuant computational platform for mass spectrometry-based shotgun proteomics. Nat Protoc. 2016;11(12):2301–19.
Article CAS PubMed Google Scholar
Tyanova S, Cox J. Perseus: a bioinformatics platform for integrative analysis of proteomics data in cancer research. Cancer Syst Biol. 2018;1711:133–48. https://doi.org/10.1007/978-1-4939-7493-1_7.
Article CAS Google Scholar
Hulsen T. DeepVenn—a web application for the creation of area-proportional Venn diagrams using the deep learning framework Tensorflow.js. arXiv. 2022. https://arxiv.org/abs/2210.04597. Accessed 23 Jan 2025.
Ge SX, Jung D, Yao R. ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics. 2020;36(8):2628–9.
Article CAS PubMed Google Scholar
Supek F, Bošnjak M, Škunca N, Šmuc T. REVIGO Summarizes and Visualizes Long Lists of Gene Ontology Terms. PLoS ONE. 2011;6(7): e21800.
Article CAS PubMed PubMed Central Google Scholar
Medvedeva YA, Lennartsson A, Ehsani R, Kulakovskiy IV, Vorontsov IE, Panahandeh P, et al. EpiFactors: a comprehensive database of human epigenetic factors and complexes. Database. 2015;2015:bav067.
Article PubMed PubMed Central Google Scholar
Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009;10(3):R25.
Article PubMed PubMed Central Google Scholar
Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence alignment/map format and SAMtools. Bioinformatics. 2009;25(16):2078–9.
Article PubMed PubMed Central Google Scholar
Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9(9):R137.
Article PubMed PubMed Central Google Scholar
Amemiya HM, Kundaje A, Boyle AP. The ENCODE blacklist: identification of problematic regions of the genome. Sci Rep. 2019;9(1):9354.
Article PubMed PubMed Central Google Scholar
Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. The human genome browser at UCSC. Genome Res. 2002;12(6):996–1006.
Article CAS PubMed PubMed Central Google Scholar
Kramer NE, Davis ES, Wenger CD, Deoudes EM, Parker SM, Love MI, et al. Plotgardener: cultivating precise multi-panel figures in R. Bioinformatics. 2022;38(7):2042–5.
Article CAS PubMed PubMed Central Google Scholar
Ramírez F, Ryan DP, Grüning B, Bhardwaj V, Kilpert F, Richter AS, et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 2016;44(W1):W160–5.
Article PubMed PubMed Central Google Scholar
Conway JR, Lex A, Gehlenborg N. UpSetR: an R package for the visualization of intersecting sets and their properties. Bioinformatics. 2017;33(18):2938–40.
Article CAS PubMed PubMed Central Google Scholar
Patwardhan M, Wenger C, Davis E, Phanstiel D. Bedtoolsr: an R package for genomic data analysis and manipulation. JOSS. 2019;4(44):1742.
Article PubMed Google Scholar
Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell. 2010;38(4):576–89.
Article CAS PubMed PubMed Central Google Scholar
Durand NC, Shamim MS, Machol I, Rao SSP, Huntley MH, Lander ES, et al. Juicer provides a one-click system for analyzing loop-resolution Hi-C experiments. Cell Syst. 2016;3(1):95–8.
Article CAS PubMed PubMed Central Google Scholar
Rowley MJ, Poulet A, Nichols MH, Bixler BJ, Sanborn AL, Brouhard EA, et al. Analysis of Hi-C data using SIP effectively identifies loops in organisms from C. elegans to mammals. Genome Res. 2020;30(3):447–58.
Article CAS PubMed PubMed Central Google Scholar
Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26(6):841–2.
Article CAS PubMed PubMed Central Google Scholar
Perez-Riverol Y, Bandla C, Kundu DJ, Kamatchinathan S, Bai J, Hewapathirana S, et al. The PRIDE database at 20 years: 2025 update. Nucleic Acids Res. 2025;53(D1):D543–53.
Article PubMed Google Scholar

Download references

Acknowledgements

We thank members of the Dr. Jill Dowen lab, Dr. Daniel McKay lab, and Dr. Doug Phanstiel lab, as well as Dr. Wonho Kim for helpful discussions of data analysis and interpretation. We thank Dr. Mauro Calabrese for providing plasmids and guidance on the creation of the inducible TurboID cell lines. We thank members of the Dr. Brian Strahl lab for assistance with experimental design and quality control for proximity labeling experiments. We thank the labs of Dr. Jesse Raab, Dr. Scott Williams, Dr. Brian Strahl, Dr. Mauro Calabrese, and Dr. Rob McGinty for sharing antibodies used for co-immunoprecipitations. We thank Dr. Michael Guertin for guidance on genomics analysis. We thank the staff of the UNC High Throughput Sequencing Facility and UNC Research Computing for their assistance with sequencing data generation and processing. This research is based in part upon work conducted using the UNC Proteomics Core Facility, which is supported in part by NCI Center Core Support Grant (2P30 CA016086-45) to the UNC Lineberger Comprehensive Cancer Center.

Funding

This work was supported by NIGMS grants R35GM124764 and R35GM152103 to JMD. NLR was supported in part by NIGMS grant T32GM007092 and T32GM135128.

Author information

Authors and Affiliations

Curriculum in Genetics and Molecular Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA
Natalie L. Rittenhouse, Riya Gohil & Jill M. Dowen
Integrative Program for Biological and Genome Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA
Natalie L. Rittenhouse, Riya Gohil & Jill M. Dowen
Department of Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA
June E. Arricastres & Jill M. Dowen
Department of Biochemistry and Biophysics, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA
Jill M. Dowen
Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27599, USA
Jill M. Dowen

Authors

Natalie L. Rittenhouse
View author publications
Search author on:PubMed Google Scholar
Riya Gohil
View author publications
Search author on:PubMed Google Scholar
June E. Arricastres
View author publications
Search author on:PubMed Google Scholar
Jill M. Dowen
View author publications
Search author on:PubMed Google Scholar

Contributions

The study was conceived and the experiments were designed by NLR and JMD. NLR, RG, and JEA conducted the experiments. NLR and RG performed data analysis. The manuscript was written by NLR and JMD, with input from RG. The final manuscript was reviewed and approved by all authors. JMD supervised and acquired funding for the study.

Corresponding author

Correspondence to Jill M. Dowen.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Supplementary figures and corresponding figure legends

13072_2025_596_MOESM2_ESM.xlsx

Additional file 2: Table S1. Cohesin TurboID mass spectrometry Tier 1 and Tier 2 interactomes. RAD21-TurboID mass spectrometry results for Tier 1 and Tier 2 cohesin interactomes

13072_2025_596_MOESM3_ESM.xlsx

Additional file 3: Table S2. GO Biological Processes for Tier 1 and Tier 2 cohesin interactomes. Lists of all GO terms for Tier 1 and Tier 2 cohesin interactomes with enrichment FDR, fold enrichment, number of pathway genes, and number of genes from Tier 1 or Tier 2 found in that pathway

13072_2025_596_MOESM4_ESM.xlsx

Additional file 4: Table S3. Oligonucleotides for piggyBac vector cloning. Oligonucleotides and plasmids used for cloning piggyBac vectors developed for creation of inducible cell lines

13072_2025_596_MOESM5_ESM.xlsx

Additional file 5: Table S4. Representative raw western blots. Representative raw western blot images for all blots in this study

13072_2025_596_MOESM6_ESM.xlsx

Additional file 6: Table S5. Summary and accession numbers of sequencing data used in this study. Summary table of the datasets used in this study including factor accession numbers, background accession numbers, and sequencing statistics

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Rittenhouse, N.L., Gohil, R., Arricastres, J.E. et al. Unraveling the cohesin-chromatin interface: identifying protein interactions that modulate chromosome structure and function. Epigenetics & Chromatin 18, 31 (2025). https://doi.org/10.1186/s13072-025-00596-4

Download citation

Received: 31 January 2025
Accepted: 19 May 2025
Published: 02 June 2025
DOI: https://doi.org/10.1186/s13072-025-00596-4

Unraveling the cohesin-chromatin interface: identifying protein interactions that modulate chromosome structure and function

Abstract

Background

Results

Conclusions

Background

Results

Identification of a comprehensive cohesin interactome

The cohesin interactome captures known and potentially novel biological functions of cohesin

Validation of cohesin interaction with chromatin regulating complexes

Cohesin co-localizes with SWI/SNF at enhancers, promoters, and CTCF sites

Altered cohesin occupancy at CTCF sites and enhancers upon SWI/SNF perturbation

Cohesin abundance on chromatin coincides with SWI/SNF abundance

Discussion

Conclusions

Methods

Cell culture

PiggyBac transposon vector construction

Generation of stable NIH-3T3 TurboID cell lines

Proximity ligation and enrichment of biotinylated proteins

Liquid chromatography with tandem mass spectrometry (LC–MS/MS)

LC–MS/MS analysis

LC–MS/MS data analysis

Small molecule treatments

Co-immunoprecipitation

Western blotting

Chromatin immunoprecipitation and sequencing (ChIP-seq)

ChIP-seq analysis

RNAi knockdown

Micro-C and analysis

Nuclear fractionation

Availability of data and materials

Abbreviations

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Ethics approval and consent to participate

Consent for publication

Competing interests

Additional information

Publisher's Note

Supplementary Information

Additional file 1: Supplementary figures and corresponding figure legends

13072_2025_596_MOESM2_ESM.xlsx

13072_2025_596_MOESM3_ESM.xlsx

13072_2025_596_MOESM4_ESM.xlsx

13072_2025_596_MOESM5_ESM.xlsx

13072_2025_596_MOESM6_ESM.xlsx

Rights and permissions

About this article

Cite this article

Share this article

Keywords

Epigenetics & Chromatin

Contact us