Blocking heterochromatin spreading constrains cohesin binding at a yeast heterochromatic locus
Paul M. Kraycer, Marc R. Gartenberg
TL;DR
This study shows that limiting the spread of heterochromatin at a yeast locus also limits cohesin binding in that region.
Contribution
The paper demonstrates a direct link between heterochromatin spreading and cohesin domain size at the HMR locus in yeast.
Findings
Truncating heterochromatin at HMR reduced the footprint of Sir3 and Smc3.
Cohesin binding domain size correlates with the size of the heterochromatin domain.
Artificial barriers built from bacterial DNA binding proteins effectively limited heterochromatin spreading.
Abstract
Cohesin mediates central features of chromosome architecture. The protein complex governs sister chromatid cohesion and organizes genomes into loops and domains. In budding yeast, cohesin accumulates at discrete sites on chromosome arms, as well as at domains of heterochromatin. The molecular basis for the distribution at these sites is not yet resolved, although the heterochromatin protein Sir2 has been implicated. If cohesin were to bind a recurring feature of heterochromatin, then size of the cohesin domain would match the size of the heterochromatin domain. To test this hypothesis, the span of heterochromatin at the HMR silent mating-type locus was truncated with artificial barrier elements built from bacterial DNA binding proteins. We found that the most effective barrier reduced the footprint of Sir3 and Smc3, representative components of yeast heterochromatin and cohesin. These…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig 1
Fig 2
Fig 3
Fig 4
Fig 5
Fig 6
Fig 7- —http://dx.doi.org/10.13039/100000002National Institutes of Health
- —New Jersey Commission on Cancer Research
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGenomics and Chromatin Dynamics · Chromosomal and Genetic Variations · Fungal and yeast genetics research
Introduction
Cohesin is a central element of chromosome structure and function, participating in nearly all fundamental chromosomal processes. The protein complex was initially known for holding sister chromatids together during sister chromatid cohesion. Cohesin is now also known to organize individual chromosomes into DNA loops and topologically associated domains (TADs) [1,2]. Mutations in subunits of the complex or its regulators cause developmental diseases known as cohesinopathies [3]. Dysfunction of the complex is also associated with a variety of cancers. The gene encoding the STAG2 subunit of mammalian cohesin, for example, is one of the 12 most frequently mutated genes in human cancers, and is closely associated with glioblastoma, Ewing’s sarcoma and melanoma [4,5].
The four core subunits of cohesin are Smc1, Smc3, Scc3 and Mcd1/Scc1 in yeast (referred to as Mcd1 hereafter). Smc1/3 and Mcd1 form a large protein ring that embraces both sister chromatids to mediate cohesion [6]. Scc3 (SA1/2 or STAG1/2 homologs in other organisms) associates with Mcd1 to mediate interactions with other factors that load, unload and position the complex on chromatin. One factor known as Scc2 (NIPBL in other organisms) stimulates ATPases formed by the head domains of the Smc proteins chromatin loading [7]. The stimulated ATPases also power movement of the cohesin/Scc2 complex at it pumps DNA into loops and TADs [8,9]. Remarkably, this process known as loop extrusion does not seem to require that cohesin embrace DNA topologically [8,10].
In budding yeast, cohesin accumulates in regions surrounding centromeres, as well as between convergently-oriented, transcribed genes [11–14]. The complex also associates with the silenced mating-type loci, HMR and HML, that are packaged in heritable, repressive chromatin structure that functionally resembles heterochromatin in higher eukaryotes [15]. Cohesin binding and cohesion of these sites is heterochromatin-dependent unlike complexes elsewhere in the genome [16–18]. The role of cohesin in yeast heterochromatin biology is unresolved. Whereas some studies found that the complex restricts establishment and expansion of heterochromatin domains others have not [19–22]. In Schizosaccharomyces pombe, where cohesin also associates with heterochromatin, the Pds5 cofactor controls the impact of cohesin on heterochromatin maintenance [23].
Budding yeast heterochromatin assembles from four Sir proteins [15]. At HMR, Sir1 nucleates domains by associating with proteins bound to cis-acting DNA elements known as silencers. Sir1, in turn, recruits Sir4, which acts as a scaffold for association of Sir2 and Sir3. Sir2 is an NAD-dependent, protein deacetylase that acts on histone H4K16Ac, which creates a favored binding site for Sir3. In this way, nucleosomal Sir3 fosters subsequent rounds of Sir2–4 recruitment. Spreading continues until a barrier (i.e., insulator) to further spreading is encountered. The robust stability of heterochromatic silencing (stochastically lost in only one of a 1000 cells [24]) belies the fact that the heterochromatin structure is in flux. Sir proteins exchange on and off nucleosomes readily due to a network of interactions that are weak and exacerbated by limited Sir protein availability [25]. This arrangement likely disfavors promiscuous heterochromatin formation at euchromatic sites that lack silencers, and limits the spread of bona fide heterochromatin beyond insulators.
At HMR, heterochromatin spreading is halted by a fixed barrier defined by a threonine tRNA gene (formally YNCC00114W but referred to here simply as HMR-tDNA) [19,26]. Binding of the RNA polymerase III transcription factor TFIIIC and acetylation of neighboring nucleosomes are key features of the insulating activity, as well as loss of the underlying nucleosome (see [27] and references therein). Indeed, targeted displacement of a histone octamer is sufficient for barrier activity, presumably due to the loss of the nucleosome platform for Sir protein spreading [28].
Although cohesin binding at HMR roughly matches the span of Sir proteins, the molecular feature(s) that dictate this cohesin distribution are not known. Sir2 is a likely mediator but no direct interactions between cohesin and the deacetylase have been reported [17,18]. To investigate the relationship between cohesin and Sir proteins, we reshaped the heterochromatic domain of HMR with an artificial barrier element. We found that limiting the spread of heterochromatin reduced the size of the domain bound by cohesin. These results are consistent with cohesin associating with one or more of the spreading Sir proteins, and binding to the limits of the heterochromatic domain.
Materials and methods
Yeast strains and plasmids
S1 Table lists the strains used and their genotypes. All were derived from strain YLS586 [29], which differs from its W303 progenitor as follows: 1) the B element of the HMR-E silencer was deleted without diminishing silencing, 2) the a1 and a2 mating type genes were replaced with ADE2, and 3) a Ty1 solo δ that resides between HMR-I and HMR-tDNA in W303 but not in the S288C reference map was removed. The parental strain was modified in the following ways. A 356 bp PCR fragment bearing twelve lexA binding sites from strain YCL49 [30] was integrated via Crispr gene editing. A 319 bp EcoRI fragment bearing eight lacI binding sites from pAFS52 [31] was combined with a CgTRP1 fragment in YIplac204 to yield plasmid pPK3. A segment of pPK3 bearing 8xlacO-ScTRP1 was then PCR-amplified and integrated via homologous recombination. In the final engineered loci, the lexA and lacI binding site were inserted at identical positions and were nearly identical in size. LexA was expressed from pAT4, a 2μ-based-plasmid, pRS425 was used as the corresponding empty vector [32]. GFP-lacI and GFP-lacI-4C were expressed at comparable levels from chromosomally integrated, single copy cassettes [33]. A PCR-amplified 3xHA::kanMX tag from pFA6a-3HA-kanMX6 was fused to the 3’-end of genomic SMC3. For silencing assays, an mURA3 reporter gene [34] was integrated downstream of the lexA and lacI binding sites. For transcription activation assays, the mURA3 reporter was replaced with a minimal HIS3 reporter (mHIS3) that contains the ORF and 103 bp of the promoter but not the upstream activating elements [35]. A transcriptional activator control was generated by fusing a Gal4 activation domain fragment (Gal4_AD_, amino acids 768–881) to the C-terminus of GFP-lacI. All genomic modifications were by confirmed by PCR and/or DNA sequencing.
Chromatin immunoprecipitation
ChIP was performed as described previously [33,36]. Strains with lacI barrier elements were grown overnight in YPDA to an OD_600_ between 0.2–0.3 and then arrested in M phase with nocodazole (C_f_ = 10 μg/ml). Strains with lexA barrier elements were grown overnight in 25 mL of SC-leu to an OD_600_ between 0.4–0.6 before the sequential addition of 25 mL of YPDA, and then nocodazole. In both cases, three hours after the addition of nocodazole, cells were fixed by adding 1.4 mL of 37% formaldehyde and rocking for 20 minutes at room temperature (RT). Fixation was quenched by the addition of 2.5 mL of 2.5 M glycine and rocking for three minutes at RT. Cells were pelleted at 2400 rpm for 5 min at 4ºC and washed twice with ice cold TBS (50 mM Tris Cl, pH 7.4, 150 mM NaCl) and then transferred to a 2 mL flat bottom tube. Cells were resuspended in 200 μL fresh ice-cold FALB+ (50 mM HEPES-KOH pH 7.5, 150 mM NaCl, 1 mM EDTA pH 8.0, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 2 mM PMSF, 1 tablet/10mL protease inhibitor cocktail (Roche # 11873580001) and an equal volume of acid-washed glass beads. Cells were ruptured by vortexing at 4ºC for 20 minutes. The soluble fraction was recovered by piercing the flat bottom with a flame-heated 23-gauge needle and spinning the contents at 2000 rpm into a lid-less Eppendorf tube at the bottom of a 15 mL Corning tube. Samples were sonicated in a Bioruptor (Diagenode) for 30 cycles of 30 seconds on and 30 second off at the high setting in the ice water bath to yield bulk chromatin fragments in the range of 200−1000 bp. 730 ul of FALB+ buffer was added to the sonicated samples, which were centrifuged at 15000 rpm at 4ºC for 10 minutes. 100 μL of supernatant was saved as WCE and the remaining supernatant was used for immunoprecipitation with anti-HA for Smc3-3xHA (12CA5, Roche), polyclonal anti-H3 for H3 (Ab1791, Abcam) and for polyclonal anti-Sir3 for Sir3 (rabbit 2934, generously provided by Laura Rusche). 2.5 μL of antibody and 1 μL of sheared salmon sperm DNA were added to each sample before rotating overnight at 4ºC. The next day samples were bound to 40 μL of Protein A magnetic beads (10002D, Invitrogen) that had been prepared by washing with 0.5 mL of 1x PBS-BSA (137 mM NaCl, 2.7 mM KCl, 10 mM Na_2_HPO_4_, 2 mM KH_2_PO_4_, 1%(v/v) BSA). After for four hours at 4ºC on a rotator, the samples were washed with 0.6 mL of four successive buffers at RT, each for 5 min on a rotator: wash 1) FALB (50 mM HEPES-KOH, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS); wash 2) FALB plus 500 mM NaCl; wash 3) ChIP Wash Buffer (10 mM Tris-HCl pH 8.0, 0.25 M LiCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.5% sodium deoxycholate); wash 4) TE (10 mM Tris-HCl pH 8.0, 1 mM EDTA). Immunoprecipitated chromatin was eluted from the beads by adding 100 μL of ChIP Elution Buffer (50 mM Tris-HCl pH 7.5, 10 mM EDTA, 1% SDS) and heating to 65ºC for 10 min. Crosslinks were reversed by adding 100 μL 10% Chelex and heating to 95ºC for 10 minutes. Residual protein was degraded with 1 μL of Proteinase K (10 mg/mL) and heating to 50ºC for 20 mins. For WCE, crosslinks were reversed by similarly adding 100 μL Chelex and heating. RNA was removed by adding 1 μL RNase (10 mg/mL) and heating to at 37ºC for 15 mins. Residual protein was degraded with 1 μL of Proteinase K and heating. Finally, both immunoprecipitated and WCE samples were precipitated with LiCl (Cf = 0.4 M), 1 μL of glycogen (Roche, R0561) and 2.5 volumes of EtOH after first removing the Chelex beads. qPCR was performed with a Rotor-Gene Q. All reported ChIP values were normalized to internal controls: for Smc3-3xHA, a region of high cohesin occupancy on ChrV (549.7 kb); for H3, the same region of ChrV; and for Sir3p, the HML-I silencer. Primers for qPCR are listed in S2 Table. The values reported represent the mean and standard deviation of at least three biological replicates. Statistical significance was calculated using pairwise Student’s t-tests with the Bonferroni correction for multiple testing where appropriate*.*
Silencing and transcription activation assays
Strains were grown overnight to saturation in YPDA or SC media lacking leucine to select for plasmids. Media was exchanged with H_2_O and then cell densities were normalized. For silencing assays, 10-fold serial dilutions were spotted on either SC + 5FOA and SC plates or SC-leu + 5FOA and SC-leu plates. For transcriptional activation assay, the dilutions were spotted on SC, SC-his and SC-his plus 2.5 mm 3-aminotriazole. Growth was recorded every 24 hours using a Bio-Rad ChemiDoc imaging system.
Results
Barrier activity of lexA at HMR
To study the relationship between binding of cohesin and Sir proteins, we aimed to shrink the prototypical heterochromatic domain at HMR with an artificial barrier element. We first used the bacterial lexA protein, which was previously shown to block heterochromatin spreading when targeted to sites at HML [28]. Here, six copies of an identical binding site sequence (the ColE1 operator, which binds two lexA proteins) were positioned roughly 500 bp downstream of the HMR-I silencer (labeled lex ops, Fig 1A). A sensitive reporter gene construct, mURA3, was situated downstream of the lex ops site but before the natural HMR-tDNA barrier [34]. URA3 encodes orotidine 5’-phosphate decarboxylase that converts 5-fluoro-orotic acid (5FOA) into a toxic metabolite. Heterochromatic spreading across mURA3 was scored by viability of serially diluted cultures on plates containing 5FOA. In the absence of lexA, the strain grew well despite the 5FOA challenge, indicating that the reporter gene was repressed at this location (Fig 1B). Deletion of SIR4 eliminated growth, showing that the repression was due to heterochromatic silencing.
LexA reduces the spread of heterochromatic silencing.(A) Schematic of the modified HMR locus. Distances are shown relative to the ORC binding site in HMR-I, the center of the lex operator array and the ATG of mURA3. ADE2 replaces the endogenous mating-type genes but was not used in this study. qPCR measurements in subsequent figures were made at HMR-E, HMR-I, lex ops and dwnX. (B) Silencing of mURA3. Strain MRG7957 (SIR4) with either empty vector (pRS425) or a vector expressing lexA (pAT4) and strain MRG7955 (sir4Δ, pRS425) were spotted in 10-fold dilutions on SC-leu + 5FOA to measure silencing and on SC-leu as a loading control. Omission of leucine selected for plasmids.
Expression of lexA caused a 10-fold loss of viability. This indicates that the protein bound the lex ops site and truncated heterochromatin before it spread to the reporter gene. That colonies still formed in the undiluted culture (the leftmost spot), indicates that silencing persisted in a subset of cells. Together, these results show that lexA limits the spread of heterochromatic silencing at HMR, albeit in a way that is not fully penetrant.
Impact of lexA on Sir3 spreading and cohesin binding
To test whether the artificial barrier blocked the spread of heterochromatin proteins, Sir3 levels were measured with chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR). A heterochromatin free site on chromosome V (534 kb) was used as a negative control. Fig 2A shows that lexA expression did not alter Sir3 occupancy at any site tested, including the HMR silencers upstream of the lex ops site (HMR-E and HMR-I) and at a site immediately downstream (labeled dwnX). The incomplete barrier activity of lexA likely accounts for the detection of Sir3 beyond the artificial barrier. We proffer that the 5FOA silencing assay, which should detect gene expression even if only transient, is more sensitive than the ChIP assay for heterochromatin proteins.
LexA blocks Sir protein spreading inefficiently at HMR.(A) ChIP-qPCR of Sir3. Sir3 occupancy was measured at HMR-E, HMR-I, dwnX and a region of no silencing (ChrV, 534 kb) in strain MRG7957 with either empty vector (pRS425) or a vector expressing lexA (pAT4). All values were normalized to an unlinked heterochromatic site, HML-I. (B) ChIP-qPCR of Smc3-3xHA. Cohesin binding at the same sites was measured relative to a region of low cohesin (ChrV, 534 kb) in the same strain as in panel A. All values were normalized to a cohesin dense site (ChrV 549.7). (C) ChIP-qPCR of Smc3-3xHA in the absence of SIR4. Strains MRG7957 (SIR4) and MRG7955 (sir4Δ) with empty vector were used as in panel B. P values were determined by Student’s t-tests ( = p ≤ 0.05, ** = p ≤ 0.005, *** = p ≤ 0.0005).*
To determine whether lexA altered the distribution of cohesin at HMR, ChIP-qPCR was used to measure the binding of an HA-tagged subunit, Smc3-3xHA. A low cohesin site (ChrV 534 kb) was used as a negative control. As with Sir3, lexA did not change levels of the cohesin subunit upstream or downstream of the lex ops site. Much of the cohesin binding was nevertheless heterochromatin-dependent, as shown by the significant reduction of signal when SIR4 was deleted (Fig 2C). If heterochromatin proteins dictate the binding of cohesin at HMR, the inability of lexA to completely block Sir3 spreading makes it a poor tool to study determinants of cohesin binding at heterochromatin domains.
Barrier activity of lacI chimeras in yeast
Considering the shortcomings of lexA, we turned to another well-characterized bacterial protein, the lac repressor (lacI). Although typically a tetramer, lacI still forms potent DNA binding dimers after removing the C-terminal tetramerization motif [37]. Arrays of dimeric GFP-lac are often used to visualize chromosomal loci in live cells [31]. The following experiments used the prototypical dimeric GFP-lacI, as well as a larger variant that carries four mCherry proteins at the C-terminus, GFP-lacI-4C [33]. The chimeras were originally developed as size-calibrated DNA binding proteins for earlier work in our lab (see Discussion).
The lacI chimeras were expressed in a reporter strain that differed from the one above only by eight lacI operators situated at the lac ops site (Fig 3A). Cells without chimeras (no lacI) grew on 5FOA, establishing that the operator sequences alone are not a barrier (Fig 3B). Expression of GFP-lacI caused a 100-fold loss of viability, indicating that the chimera blocked heterochromatin spreading. The residual growth on 5FOA was eliminated by deletion of SIR4, suggesting that silencing persisted beyond the barrier in some cells. We surmise that GFP-lacI impedes downstream silencing, but as with lexA, the artificial barrier is not fully penetrant. By contrast, expression of GFP-lacI-4C eliminated all growth on 5FOA, phenocopying deletion of SIR4 (Fig 3B). This suggests that the larger lacI chimera generates an effective barrier to the spread of heterochromatin.
GFP-lacI chimeras reduce the spread of heterochromatic silencing.(A) Schematic of modified HMR locus. Distances are shown relative to the center of the eight lacI binding site array, which is labeled lac ops. The qPCR site labeled dwnY was offset 160 bp upstream from dwnX in Fig 1, which was necessary to avoid homologous sequences in the bacterial protein expression vectors. (B) Silencing of mURA3. Strains PMK124 (GFP-lacI), PMK125 (GFP-lacI-4C), PMK126 (no lacI), and MRG7972.0 (GFP-lacI, sir4Δ) were spotted in 10-fold serial dilutions on SC + 5FOA or SC alone.
Heterochromatin spreading and the size of the cohesin bound domain
Sir3 binding was measured to test if the barrier activity of the lacI chimeras on silencing corresponded to truncation of the heterochromatic domain. Without any chimera, Sir3 was detected beyond the lac ops site in concordance with the absence of mURA3 expression (Fig 4A). Like lexA, GFP-lacI did not reduce Sir3 at any of the locations tested, including at site dwnY immediately downstream of the lac operators. By contrast, GFP-lacI-4C reduced Sir3 binding at dwnY greatly (roughly 5-fold). The effect was due to a block of heterochromatin spreading since Sir3 binding at the upstream HMR-I and HMR-E silencers was only modestly affected. Taken together, the silencing and Sir3 mapping assays show that GFP-lacI-4C truncates the HMR heterochromatic domain effectively.
Shrinking the Sir bound domain at HMR constricts the domain bound by cohesin.ChIP-qPCR reactions were performed and normalized as in Fig 2A. (A) ChIP-qPCR of Sir3. HMR-E, HMR-I, dwnY and heterochromatin free site (ChrV, 534 kb) were evaluated in strains PMK126 (no lacI), MRG7615 (GFP-lacI), and MRG7621 (GFP-lacI-4C). (B) ChIP-qPCR of Smc3-3xHA in the same strains as in panel A. (C) Heterochromatin dependence of cohesin binding. ChIP-qPCR was performed on Smc3-3xHA in strains MRG7615 (GFP-lacI, SIR4) and MRG7616 (GFP-lacI, sir4Δ). P values were determined by Student’s t-tests for panel C (as in Fig 2), and with the Bonferroni correction for multiple testing in panels A and B ( = p ≤ 0.0167, ** = p ≤ 0.00167, *** = p ≤ 0.000167).*
ChIP-qPCR of Smc3-3xHA was performed to test whether shrinking the heterochromatic domain also shrinks the domain bound by cohesin. In the absence of any chimera, cohesin was found at the upstream silencers and the downstream dwnY site (Fig 4B). Similar results were found when GFP-lacI was expressed. As with lexA, Smc3 binding was reduced significantly when SIR4 was deleted (Fig 4C). Critically, truncation of the heterochromatin domain by GFP-lacI-4C eliminated cohesin binding beyond the barrier at dwnY whereas binding to the upstream silencers was unchanged. Taken together, these data show that constraining the spread of heterochromatin constrains the size of the cohesin bound domain. The results are consistent with a recurring feature of heterochromatin determining the placement of cohesin within yeast heterochromatin.
Barrier activity correlates with nucleosome occlusion, not transcriptional activation
Potent transcriptional activators act as heterochromatin barriers in yeast (Donze 2001, Ishii 2002). A remote possibility is that GFP-lacI-4C blocks the spread of heterochromatin because it activates transcription. We therefore tested the transcriptional activation potential of the lacI chimeras with a minimal HIS3 reporter gene (mHIS3) situated immediately downstream of the lac binding site array (Fig 5A and Materials and Methods). Expression was measured by growth on media lacking histidine with and without 3-aminotriazole (3AT), a titratable inhibitor of the HIS3 gene product. The approach is similar conceptually to calibrating a two-hybrid assay where one first tests whether bait alone activates transcription [38]. Here, SIR4 was deleted to avoid the confounding effect of heterochromatic transcriptional silencing. In Fig 5B, the strain lacking a chimera (no lacI) did not grow in the absence of histidine. Strains containing GFP-lacI or GFP-lacI-4C grew minimally but this was eliminated with a low level of 3AT. By contrast, a positive control with the Gal4 activation domain (Gal4_AD_) fused to the GFP-lacI C-terminus yielded robust growth. This result shows that the barrier activity of GFP-lacI-4C does not arise from unanticipated transcriptional activity of the chimera.
lacI chimeras are not transcriptional activators.(A) Schematic of the locus used to test transcriptional activation by the lacI chimeras. The mHIS3 reporter lacks upstream activating sequences of HIS3. (B) Strains MRG8007 (GFP-lacI), MRG8003 (GFP-lacI-4C), MRG8005 (no lacI) and MRG7998 (GFP-lacI-GAL4AD) were spotted in 10-fold serial dilution on SC-his and SC-his + 2.5 mM 3AT to measure mHIS3 expression, as well as SC to measure the density of cell loading. GFP-lacI fused to the Gal4 activation domain (GFP-lacI-GAL4AD) served as a positive control for a transcriptional activator.
Nucleosome free regions are barriers to heterochromatin spreading in yeast. ChIP-qPCR of histone H3 was therefore used to test if the artificial barrier proteins block nucleosome formation. H3 occupancy at the similarly positioned lex ops and lac ops sites were compared to a well-positioned nucleosome at HML (labeled HML L3, [39]) and the nucleosome-depleted ACT1 promoter (ACT1p). Fig 6A shows the results for the lex ops-based strains. In the absence of the protein, similar levels of H3 bound the lex ops site and HML L3. Much less histone was found at ACT1p, indicating that the assay can detect differences in nucleosome occupancy. Surprisingly, H3 binding at the lex ops site did not change significantly when lexA was expressed, contrasting previous results with use of the protein as an artificial barrier at HML [28]. The reasons for this discrepancy are not clear. Nevertheless, the limited ability of lexA to displace nucleosomes at the lex ops site of HMR may account for its incomplete barrier activity (Figs 1 and 2). Fig 6B shows the results of lac ops-based strains. Expression of GFP-lacI reduced the H3 signal but not by a statistically significant margin. By contrast, expression of GFP-lacI-4C reduced histone measurably. Significant nucleosome displacement by the larger chimera is consistent with its ability to create an effective barrier to the spread of silencing (Figs 3 and 4).
Activity of the artificial barrier proteins correlates with nucleosome occlusion.ChIP-qPCR of histone H3. (A) The lex ops, HML L3 and ACT1p sites were evaluated in strain MRG7957 with either empty vector (pRS425) or a vector expressing lexA (pAT4). All values were normalized to ChrV 549.7. (B) The lac ops, HML L3 and ACT1p sites were evaluated in strains PMK126 (no lacI), MRG7615 (GFP-lacI), and MRG7621 (GFP-lacI-4C). P values were determined by Student’s t-tests as in Fig 4.
Discussion
Yeast heterochromatin assembly is operationally divided into steps that include 1) nucleation of Sir binding at silencers and 2) spreading of the silencing factors as they polymerize beyond. In this context, the simplest model for cohesin association is that the complex associates with one of the Sir proteins leading to a cohesin footprint that expands with the spread of the heterochromatin domain. In a second model based on nucleation, cohesin loads in a Sir-dependent manner at silencers and then spreads outwardly in a manner independent of Sir protein spreading. Deletion of Sir proteins cannot distinguish between these models because loss of Sirs abolishes both nucleation and spreading. Here, we limited Sir protein spreading but not nucleation with the use of an artificial barrier. We found that the size of the domain bound by cohesin shrank when the heterochromatin domain was restricted. We speculate that artificially extending a heterochromatin domain (e.g., through overexpression of Sir3 [40]) would conversely expand the domain bound by cohesin. Scaling of this kind, with cohesin binding matching the size of a heterochromatin domain, could be due to cohesin associating with a recurring feature of yeast heterochromatin (Fig 7).
Shrinking HMR heterochromatin shrinks the size of the cohesin bound domain.In this model, a featureless ring representing cohesin associates with one of the Sir proteins, thereby coupling distribution of the complex across yeast heterochromatin with Sir protein spreading.
We note that our data cannot rule out a third more complicated possibility: that the artificial barrier GFP-lacI-4C operates by two distinct mechanisms. In the first, the barrier blocks the spread of heterochromatin by nucleosome exclusion (Fig 6). In the second, the barrier independently blocks the spread of mobile cohesin by virtue of its large size. We have previously shown that large obstacles, like GFP-lacI-4C, block cohesin complexes mobilized by transcription [33]. We disfavor this model for reasons described below.
If cohesin associates with a recurring feature of heterochromatin, which heterochromatin factor might be responsible? Sir2 is a favored candidate because tethering fragments of the protein to DNA created an ectopic site of cohesion, even in the absence of Sir3 and Sir4 [17]. Remarkably, only a 63 amino acid carboxy-terminal domain lacking deacetylase activity was required. Moreover, select point mutations within this polypeptide abolished cohesion [18]. Sir2 is also responsible for cohesin binding at other sites of the genome, including the rDNA array, where the deacetylase suppresses RNA polymerase II transcription without the help of other Sirs [34,41,42]. If Sir2 is a recruitment factor, it may not act alone. At the rDNA for example, other Sir proteins bind with Sir2, even if they are not necessary for silencing [43–45]. Additionally, Sir2 point mutations that abolish heterochromatic cohesion do not reduce cohesin binding [18]. If cohesin contacts multiple heterochromatin features, as proposed earlier [16], then binding would persist even if contacts necessary for cohesion were mutated. Taken together, these observations point toward a role for Sir2 in cohesin retention at heterochromatin, but other heterochromatic factors may contribute.
Direct contacts between cohesin and heterochromatin need not be limited to core cohesin subunits. Genome mapping of Scc2 found the protein enriched at HMR [46]. Moreover, a recent search for Scc2 interacting partners found high enrichment of Sir3, not Sir2 [47]. By analogy, the S. pombe heterochromatin factor Swi6 interacts genetically and physically with related cohesin loading factors [48,49]. These observations raise the possibility that Scc2 and Sir3 may also contribute to accumulation of cohesin at yeast heterochromatin. Cohesion of HMR relies on cohesin that is loaded first at the HMR-tDNA barrier by Scc2 [46,50,51]. Deletion of the tDNA reduced but did not eliminate cohesin at HMR, suggesting that loading at other sites contributes. Thus, Scc2 may move with cohesin from multiple sites until their migration is arrested by interactions with a factor(s) within the heterochromatic domain. In this scenario, mobile complexes would bypass HMR if heterochromatin were abolished, consistent with the loss of cohesin at HMR in sir4 mutants (Figs 2 and 4).
In mammals, heterochromatin does not impede loop extrusion [52]. However, cohesin accumulates at pericentric heterochromatin where it confers cohesion necessary for proper chromosome segregation. In this case, heterochromatin HP1 recruits Haspin, which blocks cohesin removal [53,54]. By analogy, cohesin enrichment at yeast heterochromatin might be due similarly to a factor(s) that block disassembly of the complex. Given that transcription can move cohesin complexes [13,14], inhibition of transcription by heterochromatin at HMR could, in effect, block cohesin removal. We disfavor this explanation for cohesin retention at HMR because a) tethering Sir2 fragments restored cohesion of the locus even in the absence of transcriptional silencing, and because b) eliminating transcription of the locus in a heterochromatin deficient strain by deleting the bi-directional promoters did not restore cohesion [16].
The GFP-lacI chimeras were originally developed as a size-calibrated roadblocks to cohesin movement [33]. In that study, constructs with three or more mCherries blocked passage of complexes that were mobilized by transcription. Accordingly, we expected that GFP-lacI-4C would prevent cohesin from arriving at HMR, and cause buildup of the complex at the dwnY site adjacent to the lac operator array. We saw the opposite: GFP-lacI-4C reduced cohesin at the dwnY site. The roadblocking activity we reported previously may be specific for transcription-driven cohesin complexes, and thus not be in play at HMR.
Finally, the artificial barriers described here may also yield additional insight into nucleosome occlusion. Previously, Widom and coworkers described a phenomenon termed collaborative cooperation for unrelated proteins binding the same nucleosome (see Miller [55] and references therein). In brief, the unwinding of nucleosomal DNA by the binding of one protein favors the unwinding and binding of another. Here, GFP-lacI-4C chimeras occlude nucleosomes from binding sites but GFP-lacI does not (Fig 3). The N-terminal DNA binding heads of lacI are identical in these constructs, and far from the mCherries at the lacI C-terminus [56]. Steric clash between these C-terminal additions may transmit strain through the repressor to distort DNA of closely spaced binding sites within nucleosomal DNA. This “bumping of shoulders” may enhance collaborative cooperation in nucleosome displacement by GFP-lacI-4C, as well as native transcriptional regulators.
Supporting information
S1 TableYeast strains.(DOCX)
S2 TableqPCR primers.(DOCX)
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Yatskevich S, Rhodes J, Nasmyth K. Organization of chromosomal DNA by SMC complexes. Annu Rev Genet. 2019;53:445–82. doi: 10.1146/annurev-genet-112618-043633 31577909 · doi ↗ · pubmed ↗
- 2Davidson IF, Peters J-M. Genome folding through loop extrusion by SMC complexes. Nat Rev Mol Cell Biol. 2021;22(7):445–64. doi: 10.1038/s 41580-021-00349-7 33767413 · doi ↗ · pubmed ↗
- 3Banerji R, Skibbens RV, Iovine MK. How many roads lead to cohesinopathies? Dev Dyn. 2017;246(11):881–8.28422453 10.1002/dvdy.24510 · doi ↗ · pubmed ↗
- 4Scott JS, Al Ayadi L, Epeslidou E, van Scheppingen RH, Mukha A, Kaaij LJT, et al. Emerging roles of cohesin-STAG 2 in cancer. Oncogene. 2025;44(5):277–87. doi: 10.1038/s 41388-024-03221-y 39613934 · doi ↗ · pubmed ↗
- 5Dupuy M, Lamoureux F, Mullard M, Postec A, Regnier L, Baud’huin M, et al. Ewing sarcoma from molecular biology to the clinic. Front Cell Dev Biol. 2023;11:1248753. doi: 10.3389/fcell.2023.1248753 37752913 PMC 10518617 · doi ↗ · pubmed ↗
- 6Haering CH, Farcas A-M, Arumugam P, Metson J, Nasmyth K. The cohesin ring concatenates sister DNA molecules. Nature. 2008;454(7202):297–301. doi: 10.1038/nature 07098 18596691 · doi ↗ · pubmed ↗
- 7Bürmann F, Löwe J. Structural biology of SMC complexes across the tree of life. Curr Opin Struct Biol. 2023;80:102598. doi: 10.1016/j.sbi.2023.102598 37104976 PMC 10512200 · doi ↗ · pubmed ↗
- 8Davidson IF, Bauer B, Goetz D, Tang W, Wutz G, Peters J-M. DNA loop extrusion by human cohesin. Science. 2019;366(6471):1338–45. doi: 10.1126/science.aaz 3418 31753851 · doi ↗ · pubmed ↗