Eco1 is the acetyltransferase that establishes sister‐chromatid cohesion during DNA replication. A budding yeast strain with an eco1 mutation that genocopies Roberts syndrome has reduced ribosomal DNA (rDNA) transcription and a transcriptional signature of starvation. We show that deleting FOB1—a gene that encodes a replication fork‐blocking protein specific for the rDNA region—rescues rRNA production and partially rescues transcription genome‐wide. Further studies show that deletion of FOB1 corrects the genome‐wide replication defects, nucleolar structure, and rDNA segregation that occur in the eco1 mutant. Our study highlights that the presence of cohesin at the rDNA locus has a central role in controlling global DNA replication and gene expression.
Cohesin affects gene expression and DNA replication genome‐wide through the control of processes at the rDNA region in budding yeast. Replication defects at the rDNA in a cohesin acetyltransferase mutant lead to global defects that could contribute to human disease.
A cohesin acetyltransferase mutant in budding yeast has global defects in gene expression and DNA replication.
Recovery of DNA replication at the ribosomal DNA (fob1Δ) rescues local and global defects in replication and transcription.
Global defects in gene expression and DNA replication in Roberts syndrome could stem in part from the ribosomal DNA.
The evolutionarily conserved cohesin complex contributes to chromosome function in many ways. Cohesin contributes to the processes of chromosome segregation, DNA replication, chromosome condensation, and DNA damage repair. Cohesin mutations reduce ribosomal DNA (rDNA) transcription and translation in both budding yeast and human cells . Cohesion also promotes nucleolar structure and function in both budding yeast and human cells , . Roberts syndrome (RBS) is a human disease caused by mutation of ESCO2, a homolog of the yeast cohesin acetyltransferase ECO1 gene . Mutations in cohesin are also associated with Cornelia de Lange syndrome (CdLS) and myeloid neoplasms. These diseases are caused by changes in gene expression, rather than aneuploidy. However, the mechanisms by which the cohesin complex influences the transcriptome are unclear.
Cohesin binds to the approximately 150 highly transcribed tandem repeats that make up the budding yeast rDNA locus . In fact, cohesin binds to the rDNA regions in every eukaryotic genome in which binding has been examined. Replication is a challenge for this highly transcribed region. Fob1 controls rDNA replication in budding yeast, allowing it to occur only in the direction of transcription. The replication fork barrier (RFB) provided by Fob1 ensures that the replication apparatus does not disrupt transcription of the 35S gene , . Human rDNA repeats contain a similar RFB. DNA replication forks move more slowly in human ESCO2 mutant cells . Moreover, the heterochromatic repulsion observed at centromeres and nucleolar organizing centers in RBS cells suggests that these regions might have cohesion defects due to difficulty with replication . The cohesin complex binds adjacent to the RFB in the rDNA  and is important for replication fork restart . These observations indicate an intimate connection between cohesin function and DNA replication, and a special role for cohesin at the rDNA.
In this study, we observed many defects in DNA replication in an eco1 mutant. Defects in replication, rRNA production, and genome‐wide transcription were partially rescued by deleting FOB1. While replication defects have been reported in other cohesin mutants , , , , , it has not been appreciated that replication defects may interfere with transcription of the rDNA region. We propose that replication defects associated with mutations in cohesin greatly influence gene expression.
Results and Discussion
FOB1 deletion partially rescues the genome‐wide expression pattern in an eco1 mutant
We asked how deletion of FOB1 would affect the phenotypes associated with the eco1‐W216G mutation (eco1) that causes decreased acetyltransferase activity in RBS , . Gcn4 is a transcriptional activator that is translated when translational activity is poor . We employed a Gcn4‐lacZ reporter as an indicator for ribosome function. The eco1 strain shows a fourfold increase in β‐galactosidase activity compared to a WT strain . When we deleted the FOB1 gene in the eco1 mutant background, β‐galactosidase levels were reduced (Fig 1A), suggesting that FOB1 deletion rescued the poor translational activity in the eco1 strain. Moreover, when the Fob1 protein was over‐expressed, the β‐galactosidase activity in the eco1 mutant increased further (Supplementary Fig S1), indicating further impaired translational activity.
Ribosome function depends on rRNAs transcribed from the rDNA locus. We speculated that deleting FOB1 rescued ribosome function in the eco1 mutant by rescuing rDNA transcription. We used FISH to detect transcription of a single ribosomal repeat . As previously observed, the rRNA transcript level in the eco1 strain was half that in a WT strain . However, deleting FOB1 in the eco1 strain restored rRNA transcripts to WT levels (Fig 1B). For comparison, we measured rRNA transcripts in an eco1 rad61Δ double mutant strain. RAD61 negatively regulates cohesion establishment and deleting it rescues the temperature sensitivity of the eco1 strain, but not the elevated expression from the Gcn4‐lacZ reporter . While fob1Δ is expected to have an rDNA‐specific effect, rad61Δ should produce a more general effect on cohesin. In contrast to fob1Δ, rad61Δ did not rescue rRNA transcription in the eco1 strain.
Eco1 has other targets in addition to the subunits of the cohesin complex , . To exclude the possibility that fob1Δ might rescue rDNA transcription through a different mechanism, we measured the rRNA level in an smc1‐Q843Δ fob1Δ double mutant. Smc1 is a subunit of the cohesin complex. The mutation is a single amino acid deletion associated with CdLS . The level of rRNA in the smc1‐Q843Δ strain was also rescued by fob1Δ (Fig 1B), suggesting that fob1Δ rescues rDNA transcription through a cohesion‐related mechanism.
To assess the effect of fob1Δ on genome‐wide gene expression in the eco1 strain, we performed microarray analysis of RNA from the following strains: (1) eco1, (2) eco1 fob1Δ, (3) eco1 rad61Δ, (4) fob1Δ, (5) rad61Δ, and (6) WT. Differentially expressed genes were selected based on a fold change between mutant and WT of at least 1.4‐fold and an adjusted P‐value <0.05. The number of differentially expressed genes was less in the eco1 fob1Δ strain (504) than in the eco1 strain (1210) (Supplementary Fig S2). The eco1 fob1Δ strain also had fewer differentially expressed genes than the eco1 rad61Δ strain (843, Fig 1C, Supplementary Fig S2). Since genes containing binding sites for the sequence‐specific transcriptional activators Gcn4 and Tbp1 are differentially expressed in the eco1 strain , we asked whether these targets were less differentially expressed in the double mutant strains. The number of differentially expressed genes with these sites was decreased in the eco1 fob1Δ strain compared to the eco1 and the eco1 rad61Δ strain (Fig 1D). Collectively, these experiments suggest that differential gene expression in the eco1 strain may be due in part to reduced levels of rRNA. Restoration of rRNA levels significantly rescues the transcriptional profile of the mutant.
FOB1 deletion rescues DNA replication defects associated with the eco1 mutation
Given the RFB function of Fob1 at the rDNA, we speculated that fob1Δ would rescue rRNA levels in the eco1 mutant by its effect on DNA replication. To examine DNA replication, we measured cell cycle progression by cytometry analysis. Cells were synchronized in G1 by α‐factor treatment and then released at 33°C to pass through S phase. 33°C is a permissive temperature for growth, but the eco1‐W216G mutation is lethal at 37°C, so we reasoned 33°C might accentuate any phenotype (Supplementary Fig S3). A shift in DNA content was observed at 20 min in the eco1 mutant, indicating earlier progression to S phase than in a WT strain (Fig 2A). However, both WT and eco1 strains complete the shift to 2N at approximately the same time, suggesting that the eco1 strain takes longer to complete replication than WT. To assess the effect of fob1Δ on cell cycle progression in the eco1 strain, we measured cell cycle progression in fob1Δ and eco1 fob1Δ strains. The double mutant did not initiate S phase earlier, suggesting that FOB1 deletion rescued the replication defect (Fig 2A).
We next examined DNA replication in cells synchronized with α‐factor using pulsed field gel electrophoresis (PFGE). In PFGE, chromosomes cannot migrate into the gel while undergoing replication due to replication intermediates. DNA samples were collected at the indicated times following release from G1. Consistent with the cytometry data, less chromosome migration was detectable at 20 min in the eco1 strain compared to a WT strain (Fig 2B). This result confirmed that DNA replication initiated earlier in the eco1 strain, and further demonstrated that all chromosomes were affected. The eco1 fob1Δ strain did not initiate DNA replication early (Fig 2B), suggesting that fob1Δ rescued DNA replication. Therefore, deletion of the rDNA‐specific factor FOB1 appeared to rescue a genome‐wide replication defect in the eco1 mutant.
While Fob1 has fork‐blocking activity, it also regulates recombination and copy number at the rDNA. Eco1 plays a role in DNA damage repair and recombination , , . However, the eco1 mutation does not affect recombination or copy number at the rDNA locus , , nor does it have a synthetic growth phenotype with lower copy number of rDNA (Supplementary Fig S3), suggesting that fob1Δ is unlikely to rescue recombination or copy number issues. In addition, deletion of FOB1 alone does not alter the frequency of origin firing in the rDNA or the fraction of active rDNA genes . Therefore, fob1Δ may rescue the DNA replication defect in the eco1 mutant by allowing bidirectional replication at the rDNA, thereby promoting the completion of rDNA replication. Because rDNA replication and transcription do not occur simultaneously, completion of replication may facilitate efficient transcription of the locus. Deletion of FOB1 has also been shown to relieve replication stress in the smc6‐9 mutant at the rDNA locus , suggesting a shared role for SMC complexes in regulating rDNA replication.
To further address how FOB1 deletion rescues replication of the rDNA locus, we measured replication using BrdU labeling followed by ChIP/qPCR . Cells were arrested in G1 with α‐factor and then released into medium with BrdU. BrdU incorporation was detected using ChIP followed by qPCR. The detection primers were chosen to measure replication at the rARS (primer pairs 3 and 4), or the most distant point from the rARS (primer pairs 1 and 2) when replication is unidirectional. The enrichment for rARS sequences in the eco1 mutant strain was higher than in the WT strain at 20 min, demonstrating that the rDNA begins replication earlier (Fig 2C). However, at the 40‐min time point, the eco1 strain had poor replication of the rARS distal sequences compared to either WT or the eco1 fob1Δ double mutant, strongly suggesting that replication at the rDNA region is incomplete in the single mutant but more complete in the double mutant. A replication fork travels an average of 20 kb in budding yeast, but the average distance is closer to 50 kb at the rDNA, making these replication forks some of the longest in the genome , . Although these ARSs fire early, the replication of the region continues throughout S phase . The observed defects in replication are consistent with the hypothesis that prolonged replication of the rDNA interferes with its transcription in the eco1 mutant strain.
Eco1 regulates origin firing activity
To further address origin firing, we investigated the association of the replication initiation factor Cdc45 with the rARS in WT and eco1 mutant cells using ChIP , . To measure the kinetics of Cdc45 binding, we released yeast from G1 arrest at 16°C to slow down the replication process. The level of Cdc45 binding to the rDNA origin of replication (rARS) in the eco1 mutant peaked at 90 min, earlier than the peak at 105 min observed in WT cells (Fig 3A), further confirming that the rARS fires earlier in the eco1 mutant than in WT.
To study how the eco1 mutation affects replication genome‐wide, we measured DNA content by deep sequencing of genomic DNA in WT and eco1 cells , . Samples of genomic DNA were collected at 0, 20, and 40 min following release from G1 arrest. The origin firing pattern was different between WT and eco1 strains at 20 min (Fig 3B, Supplementary Figs S4 and S5). More early origins fire in the WT strain than in the eco1 mutant strain, but late origins fire about equally well in the two strains at 20 min, indicating that the origin firing sequence is disrupted in the eco1 mutant. Origin firing in the eco1 mutant also occurred at non‐ARS sites as well as mapped ARS sites (Fig 3B, Supplementary Figs S4 and S5), but replication from any single site was generally less pronounced in the eco1 mutant than in the WT. This might be due to the titration of the replication factors by the firing of many additional sites. Replication factors can be limiting for replication progression . Because our previous experiments suggested slow DNA replication in the eco1 mutant, we measured the completeness of DNA replication genome‐wide at late S phase. Replication was less complete in the eco1 mutant at 40 min (Supplementary Fig S6).
To confirm the origin firing defect in the eco1 mutant, we measured origin activity by transforming WT and eco1 mutant strains with plasmids containing (1) no ARS sequence, (2) rARS sequence, or (3) ARS1 sequence . ARS1 is a well‐studied highly efficient early ARS located on chromosome IV. We used these plasmids to assess the ability of these three sequences to promote autonomous plasmid maintenance, likely reflecting the efficiency of firing of the ARS in the genomic context. In the genome, each rDNA repeat contains the rARS sequence. However, in a given cell cycle, approximately 1 in 5 of these rARSs will fire . We observed more transformants for the rARS‐containing plasmid in the eco1 background compared to WT, using the same amount of plasmid DNA (Fig 3C), suggesting more firing of this ARS in the mutant, consistent with the BrdU labeling experiment. An increase in rARS firing could contribute to less transcription of 35S in the context of the genomic locus. The ARS1‐containing plasmid in the eco1 strain had fewer transformants, consistent with the result derived from sequencing that ARS1 fires less efficiently in the eco1 mutant than in WT (Supplementary Fig S5). Interestingly, the no ARS plasmid was replicated with low efficiency in the mutant (Fig 3C), which could reflect the origin fidelity defect observed in genome‐wide sequencing. The above results suggest that Eco1 regulates origin firing.
Cohesin is reported to be enriched at replication origins and to spatially organize replication factories . Cohesin could directly regulate origin firing at ARS sites. Another possibility is that mutations in cohesin alter the dNTP pool . Increases in the nucleotide pool can modulate origin choice and interorigin spacing , . In a genome‐wide proteomic study of the eco1 strain, we found evidence supporting the latter possibility. Many proteins involved in dNTP synthesis were present at higher levels in the eco1 mutant, which could increase the dNTP pool (Supplementary Fig S7). The gene expression profile of the eco1 mutant strain is very similar to starvation , such that the expression of many genes involved in purine, pyrimidine, and amino acid biosynthetic processes is misregulated. However, this signature is not present in the eco1 fob1Δ strain (Supplementary Figs S2 and S7). The misregulation of metabolic processes could cause too many regions to fire, which could subsequently lead to the depletion of nucleotide pools and replication factors such that replication forks cannot proceed with optimal speed . Therefore, cohesin may influence origin usage, firing fidelity, and timing in part through its effect on gene expression.
FOB1 deletion rescues nucleolar morphology and chromosome segregation defects associated with the eco1 mutation
Electron microscopy shows the budding yeast nucleolus, home of the rDNA repeats, as a single dense crescent‐shaped structure abutting the nuclear envelop in a WT strain. However, in the eco1 strain, the nucleolus is irregularly shaped (Fig 4A). To assess the effect of fob1Δ on nucleolar morphology, we analyzed nucleoli in fob1Δ and eco1 fob1Δ strains. FOB1 deletion rescued the irregular nucleolar morphology in the eco1 strain (Fig 4A). In contrast to fob1Δ, rad61Δ had less of a rescue effect for nucleolar structure. The lack of rescue with rad61Δ correlates with the lack of rescue for rDNA transcription and the global transcriptional profile.
Because cohesion establishment is coupled to DNA replication , , , we wondered whether fob1Δ restored nucleolar morphology by improving the levels of acetylated cohesin. We measured the acetylation of K112 and K113 of Smc3, the lysines targeted by Eco1 for replication‐coupled cohesion , . fob1Δ did not rescue acetylation (Fig 4B), suggesting that the recovery of nucleolar morphology in the double mutant is more likely due to the rescue of the replication and transcription of the rDNA locus.
Replication stress could induce chromosome segregation defects and genome instability , . To study rDNA segregation, we used tetR‐YFP to detect tetO repeats inserted in the telomere proximal end of the rDNA . We observed that in the eco1 strain, approximately 50% of spots did not segregate correctly at 80 min after release from G1 (Fig 4C). This is consistent with the finding that cohesin mutation‐induced replication defects lead to segregation defects in mice . In contrast to the delay in separation of the rDNA, we did not observe a delay in centromere segregation (Supplementary Fig S8), suggesting that the rDNA region is specifically delayed in the eco1 mutant.
Next, we addressed whether the rDNA segregation delay in the eco1 strain could be rescued by relieving incomplete replication via fob1Δ. We observed that in the eco1 fob1Δ double mutant strain, the rDNA segregated with normal timing. This suggests that the replication defect induced by the eco1 mutation could cause the rDNA segregation delay.
Figure 4(D) shows a model summarizing the rDNA replication phenotypes for the eco1 and eco1 fob1Δ mutants. Replication stress has been reported to cause sister‐chromatid bridging, especially at fragile loci such as the rDNA . The rDNA locus could play a “sensor” role for cellular functions. Our study suggests that cohesin affects gene expression and DNA replication genome‐wide via control of these same processes at the rDNA region. We speculate that the replication defects associated with cohesin mutations interfere with the transcription of rDNA, leading to transcriptional and translational defects that contribute to human disease.
Materials and Methods
Yeast strains and cell synchronization
Yeast strains and primers used in this study are listed in Supplementary Table S1. Exponentially growing cells were arrested in G1 phase by the addition of α‐factor (1.5 × 10−8 M final) for 2 h. To release cells from α‐factor arrest, cells were spun down and washed twice in media containing 0.1 mg/ml Protease (Sigma, P‐6911).
All deep sequencing and Affymetrix microarray data have been submitted to the NCBI Gene Expression Omnibus (GEO accession number GSE54743). All primary data associated with this manuscript can be found at http://odr.stowers.org/websimr/datasetview/632/0/.
Cytometry and microscopy
Cytometry analysis was performed to analyze cell cycle phase on cells fixed in 70% ethanol. Cells were washed with FACS buffer (50 mM sodium citrate), treated with RNase, stained with propidium iodide (4 μg/ml final), and analyzed by using a MACSQuant analyzer (Miltenyi Biotech). Fluorescence signal was observed using a Zeiss Axiovert 200M microscope (63× objective, NA = 1.40). Image acquisition and analysis was performed with Axiovision (Carl Zeiss).
Pulse‐field gel electrophoresis (PFGE)
PFGE was carried out as previously described .
Yeast cells were grown overnight at 30°C in SD‐ura and then diluted to OD600 = 0.2 in YPD+CSM. Cells were allowed to grow for two generations and were collected. Protein extracts were made by bead beating. β‐galactosidase activity was measured following standardized protocols, using ONPG (o‐nitrophenyl‐β‐D‐galactopyranoside) as the substrate.
Gene expression analysis
Gene expression analysis was carried out using Affymetrix Yeast Genome 2.0 microarrays and following the protocol as previously described .
FISH experiments were carried out following the protocol as previously described .
SL and JLG wrote the paper. GH, LF, CS, and SL conducted data analysis. SL, JLG, KKL, BH, BX, AS, and TB carried out the experiments.
Conflict of interest
The authors declare that they have no conflict of interest.
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Figure S5
Supplementary Figure S6
Supplementary Figure S7
Supplementary Figure S8
Supplementary Table S1
We thank Sue Jaspersen and Jingjing Chen for assistance and helpful suggestions, Ivan Liachko for advice on the analysis of genome‐wide replication data, A. Bedalov and A. Hinnebusch for plasmids, and the Aragon, Pasero, Grunstein, Petes, Kobayashi, and van Oudenaarden laboratories for strains. We thank SIMR for funding.
This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs 4.0 License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
- © 2014 The Authors. Published under the terms of the CC BY NC ND license