Rtt101‐Mms1‐Mms22 coordinates replication‐coupled sister chromatid cohesion and nucleosome assembly

Abstract

Two sister chromatids must be held together by a cohesion process from their synthesis during S phase to segregation in anaphase. Despite its pivotal role in accurate chromosome segregation, how cohesion is established remains elusive. Here, we demonstrate that yeast Rtt101‐Mms1, Cul4 family E3 ubiquitin ligases are stronger dosage suppressors of loss‐of‐function eco1 mutants than PCNA. The essential cohesion reaction, Eco1‐catalyzed Smc3 acetylation is reduced in the absence of Rtt101‐Mms1. One of the adaptor subunits, Mms22, associates directly with Eco1. Point mutations (L61D/G63D) in Eco1 that abolish the interaction with Mms22 impair Smc3 acetylation. Importantly, an eco1LGpol30A251V double mutant displays additive Smc3ac reduction. Moreover, Smc3 acetylation and cohesion defects also occur in the mutants of other replication‐coupled nucleosome assembly (RCNA) factors upstream or downstream of Rtt101‐Mms1, indicating unanticipated cross talk between histone modifications and cohesin acetylation. These data suggest that fork‐associated Cul4‐Ddb1 E3s, together with PCNA, coordinate chromatin reassembly and cohesion establishment on the newly replicated sister chromatids, which are crucial for maintaining genome and chromosome stability.

Introduction

Faithful chromosome replication and segregation are two cornerstone processes during cell division and proliferation 1, 2, 3. Two identical sister chromatids must be tethered together after their replication in S phase until their segregation in anaphase or meiosis II. This tight physical pairing, termed sister chromatid cohesion (SCC), is not only essential for chromosome segregation, but also participates in homolog pairing, DNA repair, DNA looping, and transcription regulation 4, 5, 6, 7, 8, 9, 10.

SCC is mediated by cohesin, a four‐subunit ring comprising structural maintenance of chromosome proteins Smc1, Smc3, the kleisin subunit Scc1/Rad21, and Scc3/SA, which are conserved from yeast to human 11, 12, 13, 14. Interestingly, cohesin complexes are loaded onto chromatin even before DNA replication 15. However, SCC establishment is obligatorily coupled with the chromosome replication process and the appearance of sister chromatids 16, 17. Among many cohesion factors identified to date, the cohesin acetyltransferase (CoAT), Eco1 (also known as Ctf7), plays an essential role in cohesion establishment. Eco1 targets two highly conserved lysine residues (K112, 113) in the ATPase head domain of yeast Smc3 17, 18, 19, 20, 21. Smc3 acetylation is believed to lock the cohesin exit gate and counteract anti‐cohesion establishment activities mediated by Rad61 and Pds5, because spontaneous mutations in Rad61 and Pds5 are able to bypass the essential requirement of Eco1 19, 22, 23, 24.

Intriguingly, the CoAT activity of Eco1 is restricted to S phase during the normal cell cycle. Activation occurs at least partially through interaction with proliferating cell nuclear antigen (PCNA) and the alternative clamp loader RFC^{(CTF18‐CTF8‐DCC1)} 25, 26, 27, 28. However, abundance of acetylated Smc3 (Smc3ac) is only moderately reduced in the absence of either the Eco1 PIP (PCNA interacting protein) box or the alternative PCNA loader RFC^{(Ctf18‐Ctf8‐Dcc1)}, suggesting additional mechanisms are involved 27, 28, 29, 30. Many replication proteins have been implicated in establishing replication‐coupled cohesion albeit through unknown mechanisms 30, 31.

Rtt101‐Mms1‐Mms22 (homologous to human Cul4‐Ddb1‐Mms22L) is an E3 ubiquitin ligase. Rtt101 is the cullin, Mms1 is a linker subunit, whereas Mms22 is one of the substrate adaptors. This complex is known as a replication fork‐associated E3 required for replicating through natural pausing sites and maintaining replisome integrity in the face of replication stress in both yeast and human 32, 33, 34, 35, 36, 37, 38. Interestingly, Rtt101‐Mms1 E3 has also been demonstrated to function downstream of Rtt109 dependent H3K56 acetylation, in the ubiquitylation of H3‐H4, which is required for deposition of new H3‐H4 on the newly replicated chromatids. This sequential acetylation/ubiquitylation constitutes a recently well‐characterized replication‐coupled nucleosome assembly (RCNA) pathway 39, 40.

In this study, we report robust functional interactions between ECO1 and RTT101‐MMS1‐MMS22. Smc3 acetylation by Eco1 during S phase is significantly reduced in rtt101Δ, mms1Δ, or mms22Δ mutants and this is accompanied by compromised chromatid cohesion. Combination of pol30A251V with rtt101Δ, mms1Δ, or mms22Δ mutants results in an additive loss of Smc3ac, indicating that Rtt101‐Mms1‐Mms22 defines a PCNA‐independent Eco1 regulation pathway. We show that both Rtt109‐dependent H3K56 acetylation and H3K121K122 ubiquitylation by Rtt101‐Mms1‐Mms22 are also required for efficient Smc3 acetylation, illustrating cross talk between RCNA and cohesion pathways. These data reveal a critical role of replication fork‐associated Cul4 E3 ligases in linking two vital replication‐coupled events: SCC establishment and nucleosome assembly.

Results

Functional interactions between RTT101‐MMS1‐MMS22 and ECO1

In order to identify new factors in Eco1‐dependent or independent SCC pathways, we first performed a synthetic genetic screen by crossing eco1 ^ctf7‐203 with the collection of ~1,500 haploid yeast strains carrying single deletion of genes involved in genome stability and cell cycle regulation 39. In addition to previously reported synthetic lethal mutants including chl1Δ and ctf8Δ 41, 42, two new genes, RTT101 and MMS1, were identified to be synthetic lethal with eco1 ^ctf7‐203 in three repeated screens (Figs 1A and EV1A).

Figure 1. Functional interactions between ECO1 and RTT101/MMS1/MMS22 in SCC.

• A
  eco1 ^ctf7‐203 synthetic genetic interactions from a SGA screen. Lines between two genes represent a synthetic lethal interaction.
• B
  Overexpression of RTT101/MMS1/MMS22 suppresses the temperature sensitivity of eco1‐1. Fivefold serial dilutions of the yeast strains expressing the indicated genes under an ADH promoter were grown at the indicated temperatures for 2 days. See Fig EV1E for the eco1 ^ctf7‐203 results.
• C, D
  RTT101, MMS1, or MMS22 overexpression suppresses a SCC defect in the eco1‐1 (C) and eco1 ^ctf7‐203 (D) mutants at different temperatures. Cohesion defects in eco1 mutants in the presence of the indicated plasmid were quantified in the GFP‐marked chromatid loci strains. All cells were synchronized in G1 (α‐factor) at 25°C and then released into fresh medium containing nocodazole to arrest cells prior to anaphase. Over 150 metaphase‐arrested cells were counted for each experiment. Mean ± SD are shown from three independent experiments. Statistical significance was measured using Student's t‐test. **P < 0.01.
• E, F
  RTT101 or MMS1 overexpression partially restores the Smc3ac levels in eco1‐1 mutants at permissive (E) or non‐permissive (F) temperatures. Acetylated Smc3 and Smc3 were immunoblotted in the cell lysates by antibodies specific to Smc3ac and Smc3, respectively. Tubulin was used as a loading control. For the biological sample repeats, see Fig EV1J–K.

Data information: See also Fig EV1.Source data are available online for this figure.

Figure EV1. Functional interaction between Rtt101‐Mms1 and Eco1.

• A
  A representative plate of synthetic genetic arrays. A red circle indicates an inviable double mutant. See also Fig 1A.
• B–D
  The negative genetic interactions between the eco1 ts alleles and rtt101Δ (B), mms1Δ (C), or mms22Δ (D).
• E
  RTT101, MMS1, and MMS22 are potent suppressors of the temperature sensitivity of various eco1 alleles. See also Fig 1C. Fivefold serial dilutions of the yeast strains were grown at the indicated condition for 3 days before being photographed.
• F–I
  RTT101, MMS1, and MMS22 are not general suppressors of the ts alleles of PDS5 (F) and the cohesin subunit encoding genes MCD1 (G), SMC1 (H), and SMC3 (I). The log phase cell cultures of pds5‐3, scc1 ^mcd1‐73, smc1‐259, smc3‐42 harboring high copy ECO1, RTT101, MMS1 or MMS22 were serially diluted fivefold and then incubated at 25, 30 or 34°C for 3 days as described in Fig 1C.
• J, K
  The biological sample repeats of Fig 1F and G.

Source data are available online for this figure.

Since eco1Δ strains are inviable, to validate the genetic interactions, strains carrying three different temperature‐sensitive (ts) alleles of eco1 were generated by plasmid shuffling. Wild‐type (WT) ECO1 was expressed from pRS316/URA3 whereas the eco1 ts allele was introduced in a second plasmid, pRS313/HIS3, in eco1Δ strains carrying WT or deletion alleles of mms1Δ or mms22Δ. pRS316/URA3 expressing ECO1 allows growth of eco1Δ. On 5′ fluoro‐orotic acid (5‐FOA) plates, the ECO1‐URA3 plasmid can be eliminated due to conversion of 5‐FOA to a toxin by the URA3 gene product. Fivefold serial dilutions of log phase cells with indicated genotype were spotted on histidine‐deficient plates in the presence or absence of 5‐FOA. rtt101Δ or mms1Δ mutants grew almost normally in the presence of WT ECO1. However, in combination with any of three eco1 ts alleles, rtt101Δ or mms1Δ strains displayed synthetic lethality. The healthiest eco1 allele, eco1 ^ctf7‐203 , showed the most convincing effect (Fig EV1B and C). Since Rtt101 and Mms1 form Cul4‐family ubiquitin E3 ligases 32, we also identified one of the adaptor subunits, Mms22, displaying similar synthetic interaction with eco1 (Fig EV1D). These results suggest that Rtt101‐Mms1‐Mms22 is required for efficient growth of eco1 hypomorphs.

Novel dosage suppressors of the eco1 mutants

To support this conclusion, we investigated whether overexpression of the individual Rtt101‐Mms1 subunits was able to suppress the conditional lethal phenotype of the eco1 alleles. WT ECO1, RTT101, MMS1, MMS22, and vector alone were transformed into strains carrying the eco1 alleles. Independent isolates for each strain were then plated at permissive and restrictive temperatures. Strikingly, both RTT101 and MMS1 plasmids enabled eco1‐1 (Fig 1B) and eco1 ^ctf7‐203 (Fig EV1E) cells to grow at 37°C to an extent comparable to ECO1 itself and were even more efficient than the well‐known eco1 suppressor, POL30, encoding PCNA 17 (Fig 1B). High copy MMS22 was able to support eco1 cell growth at semi‐permissive temperature (30°C), though more weakly than RTT101 and MMS1. These results allow us to conclude that the temperature sensitivity of eco1 mutants is exquisitely sensitive to Rtt101‐Mms1 levels. RTT101, MMS1, or MMS22 overexpression cannot suppress an eco1‐null allele (data not shown), arguing that they may directly suppress a defect of the mutant Eco1 protein rather than bypassing the requirement for Eco1 function.

Next, we asked whether suppression by RTT101, MMS1, or MMS22 is specific to eco1 mutants or general to mutants defective in SCC. To this end, we repeated high copy suppression experiments on temperature‐sensitive mutants in cohesin complex subunits (Smc1, Smc3, and Mcd1) and Pds5. None of RTT101, MMS1, and MMS22 overexpression suppressed temperature sensitivity of these cohesin component mutants (Fig EV1F–I). Instead, an exacerbation occurred in some cases even at permissive or semi‐permissive temperature. Thus, RTT101‐, MMS1‐, and MMS22‐mediated suppression is likely specific for the eco1 alleles.

The cohesion defect of eco1 can be rescued by RTT101, MMS1, or MMS22 overexpression

Since the essential function of Eco1 lies in cohesion establishment, we next tested whether the physiological basis for dosage suppression is due to restored cohesion in eco1 ts mutants. For this experiment, we analyzed the green fluorescent protein (GFP)‐marked LEU locus (LEU2::lacO256, GFP‐LacI background) in metaphase‐arrested cells 43. Normal paired sister chromatids display a single GFP focus, while loss of cohesion results in two separate foci. Cells were cultured and synchronized in G1 phase at permissive temperature (25°C). After release from α‐factor arrest, each culture was shifted to the indicated temperature in fresh medium supplemented with nocodazole to synchronize cells in M phase. Approximately 65% of metaphase eco1‐1 cells exhibited two GFP foci, an indicator of precocious sister chromatid separation (Fig 1C). Importantly, in the presence of a plasmid expressing RTT101, MMS1, or MMS22, the portion of cohesion‐defective cells was reduced to nearly 32%, which is comparable to complementation by WT ECO1 or POL30 suppression (Fig 1C). Moreover, high copy RTT101 or MMS1 pronouncedly reduced the incidence of separated sister chromatids in eco1 ^ctf7‐203 at either permissive or non‐permissive temperature, whereas MMS22 and POL30 showed a relatively weaker effect (Fig 1D). Taken together, these data reveal a critical role of Rtt101‐Mms1‐Mms22 in compensating for the reduction of Eco1 function in cohesion.

Because the mutant eco1 protein has compromised CoAT activity, which is critical for cohesion establishment during S phase through targeting Smc3, we next examined whether the dosage suppression of eco1 phenotypes correlates with increased Eco1‐dependent Smc3 acetylation. Smc3 acetylation is detected through immunoblotting with an antibody specific to Smc3ac. The Smc3ac levels were extremely low in eco1‐1 cells, but were restored to nearly wild‐type levels by high copy RTT101 and MMS1 at both permissive and non‐permissive temperatures (Figs 1E and F, and EV1J and K), in keeping with their ability to suppress the conditional lethality of eco1 alleles (Fig 1B). Consistent with weaker suppression, high copy MMS22 rescued the Smc3ac levels much less efficiently. These results suggest that Eco1‐mediated acetylation of Smc3 can be strengthened by increased Rtt101‐Mms1 levels.

Efficient Smc3 acetylation by Eco1 requires Rtt101‐Mms1‐Mms22

Next, we tested whether Rtt101‐Mms1‐Mms22 is directly involved in regulation of Eco1 activity. To this end, we compared their null mutants with WT in terms of cell cycle‐regulated Smc3 acetylation. Synchronous cell cultures were collected at different time points after release from α‐factor arrest into the fresh rich media. Similar to previous observations 28, in WT cells, Smc3 acetylation coincided with the time of S phase and peaked at about 60 min following release from G1 (Figs 2A and EV2A). In G2 (90 min), Smc3ac was reduced, but increased again as cells entered the next S phase (120 min). Deletion of RTT101 or MMS1 resulted in a significant impairment in Smc3 acetylation and/or the cell cycle regulation of Smc3 acetylation/deacetylation (Figs 2A and B, and EV2A and B). Correlating with the relatively weaker contribution of MMS22 in ubiquitylation of new histones H3‐H4 40, mms22Δ showed a significant but less severe defect in Smc3 acetylation, implying that there might be redundant adaptor subunits of Rtt101‐Mms1 E3. Given that Smc3 acetylation is a prerequisite of SCC establishment, we then measured the cohesion efficiency in these mutants. Indeed, the reduction in Smc3 acetylation was accompanied by a significant SCC defect at both subtelomeric (Fig 2C) and pericentromeric loci (Fig EV2C). These data indicate a critical role for Rtt101‐Mms1‐Mms22 in promoting efficient Eco1‐dependent Smc3 acetylation and cohesion establishment during normal S phase.

Figure 2. Rtt101‐Mms1‐Mms22 defines a PCNA‐independent regulator of Eco1.

1. Deletion of RTT101, MMS1, and MMS22 leads to reduced Smc3ac during S phase. Cells were grown in rich media and synchronized in G1 by α‐factor before wash and release into fresh media for the indicated times. Lysates were prepared from WT and rtt101Δ/mms1Δ/mms22Δ cells and subjected to immunoblotting by anti‐Smc3 and anti‐Smc3ac. The representative flow cytometry profiles are shown in Fig EV2A.
2. Quantitation of relative Smc3ac during cell cycle in (A). The ratio of acetylated Smc3/Smc3 in each sample was quantified by Quantity One (Bio‐Rad) and plotted as shown in Fig EV2B. The ratio of WT sample in G1 phase was normalized to 1. Three biological repeats were performed, and error bars indicate the standard deviation.
3. rtt101/mms1/mms22 null mutants exhibit precocious sister chromatid separation. Synchronous cells were arrested by nocodazole in metaphase at 30°C and scored for one GFP focus (cohesion proficiency) or two foci (cohesion deficiency) as described in Fig 1D. “n” indicates the total cell number examined for each sample. Mean ± SD are shown from at least three independent experiments. Statistical significance was measured using Student's t‐test. **P < 0.01. See Fig EV2C for centromere cohesion results.
4. Synthetic genetic interactions between rtt101/mms1/mms22 null mutants and pol30A251V. Fivefold serial dilutions (initial OD₆₀₀ = 0.1) of the indicated yeast strains were grown at either 16 or 30°C for 2 days on YPD plates.
5. Combination of RTT101/MMS1/MMS22 deletion with pol30A251V leads to synergistic loss of Smc3 acetylation. Asynchronous cell lysates of each single or double mutant were analyzed for the Smc3ac status as described above.

Data information: See also Fig EV2. Source data are available online for this figure.

Figure EV2. Rtt101‐Mms1‐Mms22 is required for efficient Smc3 acetylation.

1. Representative cell cycle profiles of the samples used for Smc3 acetylation assays. Cells were grown in YPD media and synchronized by addition of 7.5 μg/ml α‐factor. G1‐arrested cells were released by washing twice in fresh medium and continued growth for the indicated time. Cell cycle profiles were analyzed by flow cytometry.
2. Quantitation of Smc3 acetylation. Titration of yeast cell extracts was applied for Western blot against anti‐Smc3ac and anti‐tubulin antibodies, respectively. The intensity of each band was quantified and plotted. Note that the level of Smc3ac is proportional to the tubulin level within the range tested. See Fig 2B.
3. rtt101/mms1/mms22 null mutants also exhibit a sister chromatid cohesion defect in the pericentromeric regions as well. Sister chromatid cohesion was measured as described in Fig 2C, except that chromatid was marked by GFP at TRP1 locus. “n” indicates the total cell number examined for each sample. Mean ± SD are shown from at least three independent experiments. Statistical significance was measured using Student's t‐test. **P < 0.01.
4. RTT101 or MMS22 overexpresssion is able to partially suppress the cold sensitivity of pol30A251V. Fivefold serial dilution assays were conducted as in Fig 1C.

Source data are available online for this figure.

A PCNA‐independent regulator of Eco1‐catalyzed Smc3 acetylation

Given previous observations that PCNA is involved in restricting Eco1 CoAT activity to S phase 27, we next examined the functional relationship between Rtt101‐Mms1‐Mms22 and PCNA. First, deletion of RTT101, MMS1, or MMS22 dramatically exacerbates the growth defect (at 30°C) and cold sensitivity (at 16°C) of pol30A251V, a mutant defective in activating Eco1 (Fig 2D) 27. Second, RTT101 and MMS22 are also high copy suppressors of pol30A251V (Fig EV2D). These results point to a potent functional interplay between Rtt101‐Mms1‐Mms22 and PCNA. Third, although Smc3 acetylation was reduced in rtt101Δ, mms1Δ or mms22Δ, and pol30A251V (Fig 2E), strikingly, in combination with pol30A251V mutation, deletion of individual RTT101, MMS1, or MMS22 resulted in a dramatic, additive loss of Smc3ac (Fig 2E, lanes 4, 6, and 8). Taken together, these data suggest Rtt101‐Mms1‐Mms22 and PCNA precisely control the Eco1‐catalyzed Smc3 acetylation during S phase independently.

Mms22 directly associates with Eco1

Above results indicate potent functional interaction between Rtt101‐Mms1‐Mms22 and the essential cohesion factor Eco1. Since the Rtt101 complex is an E3 ligase, ubiquitylation of Eco1 was a possible mechanism for stimulation of Smc3 acetylation. After many attempts, we have been unable to observe ubiquitylation of Eco1 by Rtt101‐Mms1 (data not shown). Meanwhile, we prepared total soluble yeast proteins to detect Eco1 protein levels by immunoblots. In the absence of Rtt101‐Mms1‐Mms22, the cellular levels of Eco1 were not dramatically changed (Fig 3A). These results indicate that the reduction of Smc3 acetylation in rtt101Δ, mms1Δ, or mms22Δ is not due to regulation of Eco1 expression or protein stability by Rtt101‐Mms1‐Mms22.

Figure 3. Mms22 participates in efficient Eco1 accumulation on chromatin during S phase.

1. The cellular Eco1 levels are not dramatically changed by deletion of RTT101, MMS1, or MMS22. Whole‐cell extracts were prepared as described in Materials and Methods. Eco1‐13MYC was immunoblotted against anti‐MYC antibodies, while tubulin was applied as a loading control.
2. Mms22 interacts with Eco1. MMS22 was cloned into pGADT7 (Table EV2) and introduced to AH109 strain together with an empty pBGKT7 plasmid or carrying the indicated Eco1 mutants. The yeast strains harboring the indicated yeast two‐hybrid plasmids were grown at 30°C on the selective plates.
3. Mms22 binds directly to Eco1 in vitro. Purified recombinant GST‐Eco1 and 6His‐Mms22(200–450) were incubated with glutathione Sepharose in the binding buffer containing 1 μg/μl BSA. The protein bands were revealed by immunoblotting against anti‐His and anti‐GST antibodies, respectively.

Data information: See also Fig EV3. Source data are available online for this figure.

Interestingly, through yeast two‐hybrid assays, we identified physical interaction between Eco1 and the substrate adaptor Mms22 (Fig 3B). No interaction was found between Eco1 and Mms1 or Rtt101. The interaction domain in Mms22 was mapped to a.a. 200–450 (Fig EV3). This Mms22 fragment was able to bind directly to Eco1 as evidenced by in vitro pull‐down assays using purified proteins (Fig 3C, lane 5).

Figure EV3. Mapping interaction between Eco1 and Mms22.

Mms22 truncations were constructed and used in yeast two‐hybrid assays as described in Fig 3B.

Both Mms22‐ and PCNA‐mediated interactions with Eco1 contribute to efficient Smc3 acetylation and cohesion function

To understand the role of Mms22–Eco1 interaction, we sought out separation‐of‐function mutants of Eco1. In Eco1, deletion of a short fragment (a.a.57‐67) abolished its interaction with Mms22 (Fig 4A). Meanwhile, interaction of Eco1 with PCNA was not affected, illustrating that distinct domains of Eco1 are responsible for associating with Mms22 and PCNA, respectively. Notably, when we mutated two highly conserved residues in this Eco1 region to Asp (L61D/G63D, referred to hereafter as eco1LG), the Eco1–Mms22 interaction became barely detectable in either the yeast two‐hybrid (Fig 4A) or GST pull‐down assays (Fig 4B). These data indicate that L61 and G63 residues, located between the N‐terminus and the CoAT domain of Eco1, are necessary for interaction with Mms22. Then, the SCC assays and time course experiments of Smc3 acetylation described in Fig 2 were repeated in eco1LG cells. L61D/G63D mutations caused defects in both cohesion (Fig 4C) and Smc3 acetylation (Figs 4D and EV4A). In addition, overexpression of RTT101 or MMS1 failed to restore the Smc3ac levels in eco1LG mutants (Fig EV4B, compared to Fig 1E and F). The L61D/G63D mutations did not significantly alter CoAT activity on both Scc1 and Eco1 per se as shown by time course in vitro acetylase assays (Fig 4E). Therefore, the Smc3 acetylation and cohesion defects in eco1LG cells are very likely caused by the compromised Eco1‐Mms22 interaction.

Figure 4. Mms22–Eco1 interaction promotes Smc3 acetylation during S phase.

1. Point mutations (L61D/G63D) of Eco1 abolish interaction with Mms22. Yeast two‐hybrid experiments were performed as described in Fig 3B.
2. Direct association between Eco1 and Mms22 becomes compromised in eco1LG mutant. GST pull‐down experiments were performed as in Fig 3C.
3. Precocious sister chromatid dissolution in eco1LG cells. The telomere cohesion assays were performed as described in Fig 1C. “n” indicates the total cell number examined for each sample. Mean ± SD are shown from at least three independent experiments. Statistical significance was measured using Student's t‐test. **P < 0.01.
4. eco1LG cells displayed diminished acetylated Smc3 levels during S phase. To detect Smc3ac, synchronized cell lysates from the indicated time points were immunoblotted. The representative flow cytometry profiles are shown in Fig EV4A.
5. Eco1LG mutant enzyme shows acetyltransferase activity comparable to WT in vitro. The acetyltransferase activity was analyzed by purified enzymes in the presence or absence of acetyl‐CoA (CoA) as an acetyl donor. The acetylated Scc1 or Eco1 was detected by immunoblotting with an anti‐ac‐Lys antibody.
6. eco1LG and pol30A251V mutations cause an additive loss of Smc3 acetylation. Smc3ac in the indicated strains was detected as described in Fig 2E. See Fig EV4C for growth.
7. Combination of eco1LG with a PCNA‐interaction‐defective allele eco1S12AK13A leads to a severe synthetic growth defect. Strains were constructed through plasmid shuffling. A plasmid expressing WT ECO1/URA3 to support the cell growth was marked in gray, which was lost in the presence of 5‐FOA. Fivefold serial dilutions were carried out with the initial OD₆₀₀ is 0.5.

Data information: See also Fig EV4. Source data are available online for this figure.

Figure EV4. A separation‐of‐function mutant eco1LG .

1. Representative cell cycle profiles of the samples used for the time course Smc3 acetylation analysis in Fig 4D. Cells were cultured in SC‐His media and synchronized in G1 by α‐factor and released for the indicated time at 25°C prior to flow cytometry analysis.
2. RTT101 or MMS1 overexpression fails to restore the Smc3ac levels in eco1LG mutant. Smc3 acetylation was analyzed as described in Fig 1E.
3. Synthetic genetic interactions between eco1LG and pol30A251V. Fivefold serial dilutions were conducted as described in Fig 1B at the indicated temperatures. Related to Fig 4G.

Source data are available online for this figure.

To assess the relative contribution of Eco1‐Mms22 and Eco1‐PCNA interactions to regulation of CoAT activity, we compared the overall Smc3ac levels in the corresponding mutants. Single interaction‐defective mutants (pol30A251V or eco1LG) displayed partial loss of Smc3ac, which became barely detectable in the double mutants (Fig 4F, compare lanes 4 with 2 and 3). Moreover, if we combined a PCNA‐interaction‐defective mutant, eco1S12AK13A 27, with eco1LG, severe synthetic sick growth was observed compared to each single mutant (Fig 4G). These results, consistent with synthetic genetic interactions between pol30A251V and deletion of RTT101, MMS1, or MMS22 (Fig 2D), suggest that both Mms22‐ and PCNA‐mediated interactions play crucial roles in targeting Eco1 for Smc3 acetylation during cohesion establishment in S phase.

Cross talk between RCNA and chromatid cohesion

Given that Rtt101‐Mms1 E3 ligase is a critical component of the RCNA pathway 39, 40, we then examined a possible relationship between RCNA and chromatid cohesion using four sets of experiments.

First, we tested whether the role of Rtt101‐Mms1‐Mms22 in activating Eco1 is dependent on H3K56 acetylation by Rtt109, a requirement for RCNA upstream of the Rtt101‐Mms1 E3‐mediated step. Indeed, suppression of eco1 ts alleles by E3 overexpression was abrogated in the absence of either RTT109 (Fig 5A) or RTT106, a histone chaperone also required in the RCNA pathway (Fig EV5A). Moreover, eco1 ts alleles became inviable in the rtt109Δ (Fig 5B) or H3K56R (Fig EV5B) mutants. Interestingly, RTT106 was also a dosage suppressor of eco1‐1, although it was much weaker than another histone chaperone, the CAF‐1 complex (Cac1‐Cac2‐Cac3) (Figs 5C and EV5C). More intriguingly, a single point mutation (F233L) in the Cac1 PIP box (i.e., cac1‐13, a PCNA‐interaction‐defective mutant) 44 completely abolished the CAC1‐mediated suppression (Fig 5C). These results suggest that both upstream Rtt109‐catalyzed H3K56 acetylation and downstream histone chaperones that participate in RCNA contribute to the role of the Rtt101 E3 complex in SCC.

Figure 5. Cross talk between the RCNA and SCC pathways.

1. Suppression of the temperature sensitivity of eco1‐1 by RTT101/MMS22 overexpression is dependent on Rtt109 acetylase. See Fig EV5A for the rtt106Δ results.
2. rtt109Δ and the eco1 ts alleles are synthetic lethal. Fivefold serial dilutions were carried out as above except that the initial OD₆₀₀ is 0.5 instead of 0.1. A plasmid expressing WT ECO1/URA3 to support the cell growth was marked in gray, which was lost in the presence of 5‐FOA.
3. Overexpression of CAF‐1 subunits is able to suppress the temperature sensitivity of eco1‐1 as well. See also Fig EV5B for the 4‐day results.
4. Efficient Smc3 acetylation depends on histone H3K56 acetylation as well. Chromatin‐bound proteins were prepared as described in Materials and Methods. Orc3 was used as an indicator of chromatin‐bound fractions.
5. Ubiquitylation of H3K121K122 is required for Smc3 acetylation. Smc3ac in the indicated strains was detected as described in Fig 2E.
6. RCNA, but not RINA, is required for efficient sister chromatid cohesion. Cohesion assays were performed as described in Fig 1C. “n” indicates the total cell number examined for each sample. Mean ± SD are shown from at least three independent experiments. Statistical significance was measured using Student's t‐test. *P < 0.05, **P < 0.01.

Data information: See also Fig EV5. Source data are available online for this figure.

Figure EV5. Interplay between Eco1 and RCNA components.

1. Suppression effects of RTT101/MMS22 overexpression on the temperature sensitivity of eco1‐1 are abrogated in the absence of histone chaperone Rtt106. See Fig 5A for the rtt109Δ results.
2. H3K56R is synthetic lethal with eco1 alleles. Fivefold serial dilutions were carried out as described in Fig 5B.
3. CAF‐1 subunits are also potent dosage suppressors of eco1‐1. See Fig 5C for the 2‐day results.

Second, we examined the chromatin‐bound fractions of Smc3ac in an RCNA‐defective mutant, non‐acetylatable histone H3K56R. We used chromatin fractionation to separate Triton X‐100‐soluble supernatant and chromatin pellet fractions. The chromatin‐bound Smc3ac levels were reduced about 40% in the H3K56R mutant (Fig 5D). Third, mutating the predominant sites ubiquitylated by the Rtt101 E3, histone H3K121K122, decreased Smc3 acetylation as well (Fig 5E). These suggest that the RCNA pathway is required for efficient Smc3 acetylation.

Fourth, we directly checked the sister chromatid cohesion status in the RCNA mutants. Inactivation of any key component in the RCNA pathway, using rtt109Δ, H3K56R, H3K121K122R, cac1Δ, and H4K5K8K12R mutants, caused a moderate reduction in SCC (Fig 5F). On the contrary, deletion of HIR1 or HIR2, which is involved in replication‐independent nucleosome assembly (RINA) 45, had no apparent effect on SCC (Fig 5F). Taken together, these data suggest that the H3K56ac‐Rtt101‐Mms1 H3 ubiquitylation RCNA pathway is very likely involved in Eco1‐dependent cohesion establishment.

Finally, Rtt101‐Mms1‐Mms22 has been demonstrated to associate with forks through Ctf4 46, 47, a replisomal interaction hub recently also known to link SCC establishment with DNA replication 48, 49. To answer whether Mms22 contributes to cohesion establishment in a same pathway as Ctf4, we constructed their double mutants and compared the cohesion efficiency. As shown in Fig 6A, mms22Δ was epistatic with ctf4Δ regarding cohesion defects, implicating that Mms22 may function through Ctf4 to coordinate the co‐replicational events.

Figure 6. A proposed model for connecting SCC to RCNA by a fork‐associated Cul4 E3 ubiquitin ligase.

1. mms22Δ is epistatic to ctf4Δ in compromising cohesion efficacy. Cohesion assays were performed as described in Fig 1C. “n” indicates the total cell number examined for each sample. Mean ± SD are shown from at least three independent experiments. Statistical significance was measured using Student's t‐test. n.s., not significant; **P < 0.01.
2. The RCNA pathway, constituted by H3K56ac, Rtt101‐Mms1‐Mms22, H3K121K122ub, and PCNA‐dependent CAF‐1 recruitment, deposits new H3‐H4 tetramers onto the newly replicated DNA strands and achieves the local chromatin reassembly. Meanwhile, Mms22, together with PCNA, promotes the CoAT to transiently target Smc3 efficiently, which triggers replication‐coupled SCC.

Discussion

To ensure accurate genetic and epigenetic inheritance, several vital events occur on the newly synthesized DNA strands, including nucleosome assembly, chromatin modifications and SCC which should be temporally coordinated with the course of DNA replication. In this study, we report cross talk between these co‐replicational events through evolutionarily conserved Cul4 E3 ubiquitin ligases.

The cohesive state of preloaded cohesin rings is established mainly through Eco1‐dependent Smc3 acetylation in normal S phase. Therefore, the cohesion decision time point of two newly replicated sister copies should be determined through regulation of Eco1 CoAT activity on Smc3. To achieve this, CoAT activity seems to be controlled in a precise spatiotemporal manner through multilayer mechanisms. First, Eco1 is degraded in late S and G2 in normal cells 20, 28, 50. Second, Eco1 is positioned to act on Smc3 in a tight replication‐coupled manner by multiple factors. In parallel to the RFC^{(Ctf18‐Ctf8‐Dcc1)}–PCNA pathway, here we identified the RCNA (but not RINA) pathway, including both upstream Rtt109‐dependent H3K56ac and downstream H3 ubiquitylation (e.g., K121K122) by Rtt101‐Mms1 E3s.

Based on these findings, we propose a model in which sister chromatid cohesion is established in a chromatin replication (which means DNA replication plus chromatin reassembly)‐coupled manner (Fig 6B). During DNA replication, Eco1 associates with replication fork via interaction with PCNA 25, 26, 27, 28. Rtt101‐Mms1 E3s associating with the forks via Ctf4 hub ubiquitylate new histones H3‐H4 marked with H3K56ac for nucleosome deposition 39, 40. Right upon the completion of a stretch of chromatin DNA, Rtt101‐Mms1‐Mms22 helps Eco1 to target Smc3 within the preloaded cohesin ring on the sites of newly replicated sister chromatids.

This scenario is in agreement with the following facts. First, the CoAT activity of Eco1 is essential, whereas none of the individual Eco1 regulators is indispensable. Second, the inactivation of RFC^{(Ctf18‐Ctf8‐Dcc1)}–PCNA pathway causes only a modest reduction in Smc3 acetylation and a modest (10–20%) defect in cohesion establishment, but this defect is dramatically exacerbated by deletion of the Rtt101‐Mms1‐Mms22 subunits. Third, many replication proteins including ORC, Ctf18, Csm3, Tof1, Mrc1, Ctf4, Fen1 and Chl1, DNA polymerases ε and κ have also been reported to be involved in Smc3 acetylation 25, 31, 41, 42, 51, 52, 53, 54. Intriguingly, PCNA and Rtt101‐Mms1 also directly participate in replication‐coupled histone modifications and chromatin reassembly 45. One of the simplest explanations is that SCC would not be triggered until the completion of chromatin replication, rather than all of these factors acting directly on this process. Currently, we cannot rule out an alternative model in which the RCNA pathway may affect the cohesion establishment through a similar indirect upstream fashion. Fourth, Eco1 appears globally distributed at extraordinarily low levels throughout chromatin 29. Eco1 is very likely presented to target Smc3 in a very transient manner similar to H3K56ac as a marker for signaling the completion of DNA replication and repair 55, 56. Given the potent dosage suppression effects on eco1 ts alleles, it is worthy to point out that Rtt101‐Mms1‐Mms22 could promote cohesion establishment through an anti‐establishment or Eco1‐independent manner as well 57.

Rtt101‐Mms1 (Cul4‐Ddb1) lies at the crossroad between two replication‐coupled events, RCNA and SCC. Both events are important for genome stability maintenance. Particularly, SCC is also required for homolog pairing and recombination between sister chromatids, the favorable partner for cells in S and G2 phases. Whether they also participate in damage‐induced cohesion and thus DNA repair is currently under investigation.

Materials and Methods

Yeast strains and plasmids

Strains and plasmids used in this study (Tables EV1 and EV2) were constructed as described previously 58.

Construction of eco1 alleles

Since ECO1 is an essential gene, the alleles were constructed via plasmid shuffling. Wild‐type ECO1 was cloned and expressed in pRS316/URA3 vector to allow eco1 null mutant grow. eco1 alleles are introduced in pRS313/HIS3 constructs. The pRS316‐ECO1 plasmid was counter‐selected on 5‐FOA plates due to its expression of URA3, which converts 5‐FOA to a cell toxin. The ability to support cell growth was tested for fivefold serial dilution of log phase cells were spotted on SC‐His plates in the presence or absence of 5‐FOA and incubated for 2 days at indicated temperature before photographed.

Protein expression, purification, and preparation of antibodies

Full‐length or truncated forms of pGEX6P‐1‐ECO1, pGEX6P‐1‐eco1LG, pET28a‐mms22(200‐450), pGEX6P‐1‐smc3(522‐681), and pET28a‐SCC1 were expressed in E. coli BL21 (DE3) RIL codon‐plus (Stratagene) and purified by affinity tags and conventional column chromatography. To raise polyclonal antibodies specific to Smc3, purified Smc3(522‐681) was used to immunize rabbit.

Whole‐cell extracts (WCE) and immunoblotting (IB)

WCE of one hundred OD₆₀₀ units of asynchronous or synchronous cells were prepared by glass bead beating as described previously. Protein samples were separated by SDS–polyacrylamide gel electrophoresis (PAGE) and immunoblotted with the antibodies specifically indicated in each figure. Antibodies used in this study are as follows: anti‐ac‐Smc3; mouse anti‐FLAG M2‐specific monoclonal antibody (1:2,000, Sigma); mouse anti‐HA 16B12 (1:1,000, Millipore); polyclonal anti‐GST (glutathione transferase) (1:1,000, ORIGENE); anti‐6His antibodies (1:1,000, ORIGENE); anti‐tubulin (1:10,000, MBL); anti‐ORC3 (1:5,000, gift from Dr. Qing Li); anti‐H4Ac (1:1,000, Millipore); anti‐acetyl lysine (1:1,000, Calbiochem ST1027); protein‐G‐agarose (GE Healthcare). HRP‐conjugated anti‐rabbit or anti‐mouse IgG was used as the secondary antibody (1:10,000, Sigma).

Chromatin fractionation

Chromatin fractionation was basically performed as described previously 59 with some modifications. Briefly, after Triton X‐100 treatment and sucrose cushion of spheroplasts, the chromatin fraction (CHR) is in the pellet, while the supernatant (SN) contains non‐chromatin‐bound proteins.

Synthetic genetic arrays

Temperature‐sensitive (ts) eco1 alleles, kindly provided by Robert Skibbens, were introduced into a query strain for synthetic genetic screens against a non‐essential deletion collection of cell cycle‐related genes at 25°C as described 39.

Time course experiments

Cells were grown, synchronized by 7.5 μg/ml of α‐factor and released into S phase for the indicated time. Whole‐cell extracts or chromatin fractions were prepared and analyzed by Western blotting for each sample. Smc3 and acetylated Smc3 were probed by anti‐Smc3 and anti‐Smc3‐acetyl‐K113 antibodies, respectively, as previously described 28, 60. In parallel, samples were subjected to flow cytometry analysis using a BD Biosciences FACS Verse machine.

GFP dot SCC assays

SCC assays were performed as previously described 43 using the strains (kind gifts provided by Sue Biggins) which contain a tandem of 256 copies of the Lac operator sequence inserted proximal from the telomere (derivatives of SBY180) of ChrIV or the centromere (derivatives of SBY885) of ChrIII.

In vitro acetyltransferase assays

Recombinant WT Eco1 and Eco1LG proteins were purified by conventional affinity chromatograph using glutathione beads (GE Healthcare). Autoacetylation of Eco1 was conducted at 30°C for the indicated time in 20 μl reaction buffer containing 50 mM Tris–HCl (pH 8.0), 0.1 mM EDTA, 50 mM KCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 5% glycerol, 120 μM acetyl‐CoA, and freshly purified Scc1 and GST‐Eco1 or GST‐Eco1LG proteins. Reactions were terminated by the addition of SDS sample buffer at the indicated time. Eco1 acetylation was detected with Western analysis after SDS–PAGE using an anti‐acetyl lysine antibody (Calbiochem ST1027). The amount of proteins loaded was detected by Coomassie brilliant blue staining.

In vitro pull‐down assay

Approximately 10 pM of each protein was mixed with glutathione Sepharose 4B (GE Healthcare Life Sciences) in 100 μl of binding buffer (50 mM HEPES‐NaOH pH 7.6, 150 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM PMSF, 1 μg/μl BSA, and 0.1% Triton X‐100) and incubated for 1 h at 4°C. The beads were washed at least three times prior to Western blotting.

Author contributions

HL, JZ, and QC conceived and designed the overall project; JZ conducted most experiments in yeast with the help from DS; XL constructed some strains and conducted growth assays; LD and XL performed yeast two‐hybrid assays; KS, CL, JT contributed to design and supervise the experiments; HL and JZ wrote the manuscript with input and editing from all of the authors.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

EMBO Reports (2017) 18: 1294–1305