Mms1 and Mms22 stabilize the replisome during replication stress.

Mms1 and Mms22 form a Cul4(Ddb1)-like E3 ubiquitin ligase with the cullin Rtt101. In this complex, Rtt101 is bound to the substrate-specific adaptor Mms22 through a linker protein, Mms1. Although the Rtt101(Mms1/Mms22) ubiquitin ligase is important in promoting replication through damaged templates, how it does so has yet to be determined. Here we show that mms1Δ and mms22Δ cells fail to properly regulate DNA replication fork progression when replication stress is present and are defective in recovery from replication fork stress. Consistent with a role in promoting DNA replication, we find that Mms1 is enriched at sites where replication forks have stalled and that this localization requires the known binding partners of Mms1-Rtt101 and Mms22. Mms1 and Mms22 stabilize the replisome during replication stress, as binding of the fork-pausing complex components Mrc1 and Csm3, and DNA polymerase ε, at stalled replication forks is decreased in mms1Δ and mms22Δ. Taken together, these data indicate that Mms1 and Mms22 are important for maintaining the integrity of the replisome when DNA replication forks are slowed by hydroxyurea and thereby promote efficient recovery from replication stress.

INTRODUCTION

Faithful transmission of the genome from one generation to the next requires the accurate and timely replication of the DNA. Accurate DNA replication requires a complex of proteins that localize to the replication fork, collectively referred to as the replisome ( Aparicio ; Tercero ; Calzada ). Key components of the replisome include Cdc45-MCM-GINS (CMG) which likely compose the replicative DNA helicase ( Aparicio ; Tercero ; Zou and Stillman, 2000; Takayama ; Gambus ) Polα-primase, which primes the leading strand and Okazaki fragments ( Plevani ; Singh and Dumas, 1984), the leading and lagging-strand polymerases (Polε and PolΔ; Pursell ; Nick McElhinny ), Ctf4 and Mcm10, which interact with Polα-primase ( Ricke and Bielinsky, 2004; Gambus ; Tanaka ), and the replication fork–pausing complex (FPC), which comprises Mrc1, Csm3, and Tof1 ( Katou ; Noguchi ; Calzada ; Nedelcheva ; Szyjka ; Bando ).

The timely progression of DNA replication forks can be challenged by a number of conditions, including nucleotide base lesions, abasic sites, interstrand and intrastrand cross-links, DNA secondary structures, sites of strong protein–DNA interactions, transcription complexes, depletion or inhibition of DNA polymerases, and depletion of dNTPs. These challenges are collectively referred to as replication stress. Replication stress is readily induced experimentally by the ribonucleotide reductase inhibitor hydroxyurea ( Krakoff ; Slater, 1973; Alvino ), which results in a depletion of deoxyribonucleotide triphosphates (dNTPs), thereby causing a large decrease in replication fork rate. Hydroxyurea (HU) also activates a cell cycle checkpoint ( Weinert ), which delays cell cycle progression until S phase is complete, stabilizes replication fork proteins at the stalled or slowed forks, and promotes the ability of replication forks to complete DNA synthesis ( Desany ; Alcasabas ; Lopes ; Sogo ; Cobb ; Katou ; Osborn and Elledge, 2003; Tercero ; Lucca ; Bjergbaek ; Feng ; Naylor ; Tittel-Elmer ). Although the mechanisms by which the checkpoint stabilizes replisomes under conditions of replication stress and promotes DNA replication are not entirely known, there are a number of connections between replisome components and the checkpoint. Replisome destabilization has been noted in deletion mutants of several genes with roles in maintaining genome integrity. Deletion of MEC1, RAD53, or SGS1 decreases the association of DNA polymerases with stalled forks ( Cobb ), as does deletion of MRC1 and genes encoding components of the MRX complex ( Lou ; Tittel-Elmer ). In addition, deletion of MRC1, CSM3, TOF1, DIA2, and CTF4 each causes replisome migration to become uncoupled from the nascent DNA chain ( Katou ; Bando ; Mimura ; Tanaka ). These data indicate that a complex network of proteins functions to maintain the integrity of the replisome at the DNA replication fork when cells encounter replication stress.

In budding yeast the genes MMS1 and MMS22 are important for resistance to replication stress, either from dNTP depletion by HU or by DNA damage induced by methyl methanesulfonate (MMS) or camptothecin ( Bennett ; Chang ; Hryciw ; Araki ; Baldwin ; Dovey and Russell, 2007; Dovey ; Yokoyama ; Roberts ). Genetic interaction data suggest that Mms1 and Mms22 function with the cullin Rtt101 ( Pan ; Collins ; Costanzo ; Koh ). Consistent with these data, rtt101Δ shares many phenotypes with mms1 and mms22 deletion mutants. Similar to mms1Δ and mms22Δ strains, rtt101Δ cells are sensitive to agents that perturb replication and induce DNA damage ( Chang ; Luke ; Roberts ). Deletion mutants of rtt101 also have increased rates of spontaneous DNA damage during a normal cell cycle ( Luke ), accumulate at the G2/M transition ( Michel ), and display an abnormal nuclear morphology ( Michel ; Luke ), much like mms1Δ and mms22Δ ( Dovey and Russell, 2007; Dovey ; Yokoyama ; Duro ; Roberts ).

Protein–protein interactions also suggest that Mms1 and Mms22 function in concert with Rtt101 ( Ho ; Pan ; Suter ; Zaidi ; Ben-Aroya ; Mimura ), and sequence conservation further suggests that Rtt101Mms1/Mms22 is similar to the mammalian Cul4Ddb1 complex ( Zaidi ). Cul4Ddb1 is a ubiquitin ligase with a number of roles in genome maintenance (for a review see Jackson and Xiong, 2009), including degradation of the replication factors CDT1 and CHK1 ( Higa ; Leung-Pineda ) and ubiquitination of histones and XPC under conditions of DNA damage ( Sugasawa ; Kapetanaki ; Wang ). As is the case for the Cul4Ddb1 ubiquitin ligase, Rtt101Mms1 forms multiple complexes with several substrate specific adaptors, and these are expected to dictate function ( Zaidi ; Fujii ; Mimura ).

The precise role for this Cul4Ddb1-like complex, Rtt101Mms1/Mms22, in maintaining genome integrity in response to replication stress has yet to be determined. Here we show that Mms1 localizes to regions adjacent to replication origins when replication stress is present, suggesting that Mms1 is recruited to stalled replication forks. We find that deletion of MMS1 or MMS22 reduces the association of replisome proteins with stalled replication forks, and that, in the absence of MMS1 or MMS22 replication forks progress and recover inefficiently in the presence of HU. Together our results indicate that Mms1 and Mms22 are important for the stabilization of protein components at the replication fork during replication stress and thereby function to prevent fork collapse and promote the resumption of DNA synthesis.

RESULTS

Deletion of MMS1 or MMS22 causes an abnormal cell cycle following recovery from replication stress

Deletion of MMS1 or MMS22 renders cells sensitive to DNA-damaging agents ( Bennett ; Chang ; Hryciw ; Araki ; Baldwin ; Dovey and Russell, 2007; Dovey ; Yokoyama ; Roberts ) and to replication stress caused by chronic exposure to HU ( Chang ; Araki ; Dovey and Russell, 2007; Dovey ; Yokoyama ; Roberts ) ( Figure 1A). To delineate the role of MMS1 and MMS22 in HU resistance, we assessed the viability of mms1Δ and mms22Δ over a 24 h period of HU exposure ( Figure 1B). Samples were collected at the indicated times from wild-type, mms1Δ, and mms22Δ cultures treated with 200 mM HU, and viability was measured following plating on media lacking HU. Replication stress began to affect viability in mms1Δ and mms22Δ after 6 h of HU treatment, indicating that even during short-term exposure Mms1 and Mms22 are required for HU resistance. Consistent with previous assays of MMS sensitivity, mms22Δ was more sensitive to HU than was mms1Δ ( Roberts ; Zaidi ; Dovey ).

FIGURE 1:

Mms1 and Mms22 are important for viability in the presence of replication stress. (A) Tenfold serial dilutions of wild-type, mms1Δ, and mms22Δ strains were spotted on YPD containing 0, 50, 100, 150, or 200 mM HU. Plates were incubated at 30°C for 2–3 d. (B) Wild-type, mms1Δ, and mms22Δ cells in early log phase were treated with 200 mM HU. Samples were collected at the indicated time points and plated on YPD to determine the number of colony-forming units. The average of two independent experiments is plotted.

We next examined cell cycle progression in mms1Δ and 
mms22Δ strains during treatment with HU for 90 min, and during a subsequent 3 h recovery period, by flow cytometry and by examination of cell morphology ( Figure 2). Wild-type, mms1Δ, and mms22Δ strains progress similarly during the HU treatment, arresting early in S phase ( Figure 2A). On removal of HU and release into fresh media, the wild-type strain progresses through the cell cycle and completes mitosis within 2 h, as evidenced by the reemergence of cells with 1C DNA contents ( Figure 2A, 120 min) and by the decrease in mitotic cells ( Figure 2, B and C). By contrast, mms1Δ and mms22Δ strains remain with 2C DNA contents and do not complete mitosis within 3 h, suggesting that proper and timely recovery from HU-induced replication fork stalling requires Mms1 and Mms22. In addition, we noted that during recovery from HU the mms1Δ and mms22Δ strains accumulated with a distinct morphology ( Figure 2B). The mutant cells accumulated as large, budded cells with the nuclear DNA spanning the bud neck. This nuclear morphology is similar to that seen during recovery of mms1Δ and mms22Δ cells from MMS damage ( Duro ) and is reminiscent of mutants that attempt mitosis with unreplicated regions ( Torres-Rosell ).

FIGURE 2:

mms1Δ and mms22Δ cells exhibit cell cycle defects during recovery from replication stress. Asynchronously growing cultures (Asy) were arrested in G1 with α-factor (αF), released into 200 mM HU for 90 min, and then washed and released into YPD for 3 h. Samples were collected every 30 min. (A) Flow cytometric analysis of DNA contents. Positions of cells with 1C and 2C DNA contents are indicated. (B) Fluorescence micrographs of fixed cells stained with DAPI to visualize nuclear morphology during HU treatment and subsequent recovery. (C) Large-budded cells with DNA spanning the bud neck were quantified at the indicated times.

Deletion of MMS1 results in abnormal replication fork progression during replication stress

The lengthy mitotic delay in mms1Δ and mms22Δ strains following replication stress suggested that DNA replication in the absence of Mms1 or Mms22 might be defective when HU-induced replication stress is present. We examined replication dynamics in wild-type and mms1Δ strains in the presence of replication stress. G1-arrested cells were released into media containing both bromodeoxyuridine (BrdU), to label newly synthesized DNA, and HU, to stall replication. The wild-type and mms1Δ cultures were sampled at 45 and 90 min, and BrdU-labeled DNA was hybridized to tiling microarrays to identify replicated regions genome wide. The wild-type strain exhibited narrow peaks of BrdU incorporation at early-firing replication origins after 45 min in HU ( Figure 3A, top). These peaks broadened after 90 min ( Figure 3A, bottom), consistent with the observation that HU slows, but does not arrest, DNA synthesis in wild-type cells ( Alvino ). By contrast, DNA synthesis in mms1Δ cells was defective in several ways. First, the BrdU peaks were broader at 45 min when compared with wild-type cells ( Figure 3B, top), indicating that replication forks traveled further in the mutant relative to wild type when replication stress was present. Second, the BrdU peak height at 45 min was reduced in mms1Δ compared with wild type, which might indicate greater variation in the distances traveled by individual replication forks in the population. Third, the BrdU peaks in mms1Δ cells were largely unchanged at 90 min ( Figure 3B, bottom), suggesting the absence of significant DNA replication in the mutant between 45 and 90 min. We conclude that DNA synthesis during replication stress in mms1Δ is initially advanced but then halts as replication forks arrest. We were unable to analyze replication in mms22Δ mutants, as deletion of mms22 in the multicopy thymidine kinase strain that allows incorporation of BrdU resulted in a synthetic sick phenotype (unpublished data ).

FIGURE 3:

mms1Δ cells have an irregular pattern of fork progression in HU. BrdU IP-chip analysis was performed following synchronous release of wild-type (A) and mms1Δ (B) strains into 200 mM HU for 45 min (top) or 90 min (bottom). Enrichment of DNA fragments in the BrdU sample relative to an unreplicated control is shown along chromosome VI. The signal intensity ratio on a log2 scale is shown on the y-axis and the chromosome coordinate is shown on the x-axis. Positive signal represents regions that are replicated, and regions where the positive signal is statistically significant over 300 base pairs are shown in light blue and over 900 base pairs are shown in dark blue. Replication origins (ARSs) are indicated. Early-firing origins are colored red.

Mms1 binds origin-proximal regions when replication 
forks stall

Several studies have suggested that Mms1 and Mms22 function to maintain the stability of the genome ( Bennett ; Chang ; Hryciw ; Araki ; Baldwin ; Dovey and Russell, 2007; Dovey ; Yokoyama ; Duro ; Roberts ; Zaidi ; Ben-Aroya ; Mimura ); however, it remains unclear how they exert this effect. Our data indicated that absence of Mms1 caused defects in DNA replication when replication fork stress was present, suggesting that Mms1 might act directly at stalled replication forks. We examined the association of Mms1 with sites of stalled replication forks using chromatin immunoprecipitation (ChIP) ( Figure 4, A–C). Cells were arrested in G1 phase and released into S phase in the presence of HU. Following cross-linking to preserve protein–DNA interactions, Mms1–DNA complexes were isolated, and the enrichment of an early-firing origin of replication, ARS607, versus that of a late-firing origin, ARS609, was measured using quantitative PCR (qPCR). We observed a twofold enrichment at the early-firing replication origin in the Mms1 immunoprecipitate relative to the late-firing origin, which is dormant in HU. A similar result was found when we compared enrichment of an active origin (ARS1018) with an interorigin region between ARS1009 and ARS1010 that lacks active replication origins ( Figure 4A). Together, these data indicate that Mms1 associates with early-firing origins of replication when replication forks stall. We next looked at Mms1 dynamics by conducting an Mms1 ChIP/qPCR time course with the ARS607/ARS609 probe pair. If Mms1 functions at replication forks, the levels of Mms1 protein detected at these sites may change with time as forks accumulate at the probe region. We detected a modest but significant increase in the amount of early-origin DNA in Mms1 immunoprecipitates between 60 and 120 min ( Figure 4B).

FIGURE 4:

Mms1 binds to origin-proximal regions when replication forks stall and is important for replisome stability at stalled forks. (A–C) ChIP analysis of Mms1-FLAG in the indicated strain backgrounds was performed following synchronous release of cells into S phase in the presence of 200 mM HU for 90 min at 23°C. Enrichment of early-origin DNA fragments relative to late-origin DNA in the Mms1-bound and input fractions was measured by quantitative PCR. The average of at least two experiments is plotted, and the SD is shown. (A) ChIP analysis of Mms1 localization to two early-firing origins of replication. Mms1 association with DNA was analyzed by calculating the ratio of early-origin DNA (ARS607 or ARS1018) to late-origin DNA (ARS609) or DNA associated with an interorigin region (between ARS1009 and ARS1010). (B) Time course of Mms1 localization to origin-proximal regions (ARS607 vs. ARS609). Samples were collected at the indicated time points and processed for ChIP/qPCR. (C) ChIP/qPCR of Mms1 in RTT101 and MMS22 mutants (ARS607 vs. ARS609). (D) ChIP-chip was performed in parallel with BrdU IP-chip following synchronous release of DPB3::flag cells into S phase in the presence of 200 mM HU for 90 min at 23°C. Enrichment of DNA fragments in the BrdU sample relative to an unreplicated control and enrichment of DNA fragments in the Dpb3-bound fraction relative to the unbound fraction are shown along chromosome VI. The signal intensity ratio on a log2 scale is shown on the y-axis and the chromosome coordinate is shown on the x-axis. Positive signal (blue) represents regions that are replicated, and regions where the positive signal is statistically significant over 300 base pairs are shown in light blue and over 900 base pairs are shown in dark blue. Positive signal (yellow/orange) represents occupancy by Dpb3, and regions where the positive signal is statistically significant ( Katou ) over 300 base pairs are shown in yellow and over 900 base pairs are shown in orange. Replication origins (ARSs) are indicated in red for early-firing origins that are active in HU and in black for late-firing origins that are inactive in HU.

Mms1 functions as part of a complex with Rtt101 and Mms22 ( Zaidi ; Mimura ). We examined whether binding of Mms1 at stalled replication forks required either of the Mms1-binding partners. Using ChIP/qPCR, we found that enrichment of the early-origin DNA in Mms1 chromatin immunoprecipitates was abolished in rtt101Δ and in mms22Δ strains ( Figure 4C). Thus the presence of Mms1 at an active replication origin when replication stress is present depends on the known Mms1-binding partners Mms22 and Rtt101.

Because DNA synthesis is defective in mms1Δ strains during replication fork stress and Mms1 associates with replication forks, we performed parallel BrdU immunoprecipitation (IP)-chip and ChIP-chip experiments to examine the localization of fork proteins during DNA replication in mms1 mutants ( Figure 4D). We tagged Dpb3, a subunit of DNA polymerase epsilon (Polε), in wild-type and mms1Δ strains to assess protein–DNA association and DNA synthesis concurrently. As the leading-strand polymerase, Polε localizes to origins of replication and colocalizes with sites of DNA synthesis ( Aparicio ; Hiraga ; Lou ). Strains were blocked in G1 and released into media containing HU to induce replication fork stress and BrdU to label newly synthesized DNA. In the wild-type strain, treatment with HU for 90 min results in overlapping BrdU and Dpb3 signals, with both peaks restricted to regions surrounding early-firing origins of replication. Remarkably, although the profile of BrdU incorporation in the mms1Δ strain shows a broader distribution than in the wild-type strain, Dpb3 localizes to the same chromosome coordinates as in the wild type and does not colocalize with the BrdU signal. This suggests that in mms1Δ, forks that are more distal from the origin no longer have Dpb3 (Polε) associated with them. This is consistent with the data in Figure 3, which indicate little change in the sites of DNA synthesis in mms1Δ between 45 and 90 min. We conclude that the replisome or replisome components have dissociated from these forks and that these forks are no longer synthesizing DNA.

Mms1 and Mms22 stabilize replication proteins at stalled replication forks

Our data indicate that replication forks arrest in mms1Δ and mms22Δ cells in the presence of replication stress and that replication does not resume efficiently during recovery from replication stress. Furthermore, our data suggest that replication fork proteins likely dissociate from forks in the absence of Mms1 or Mms22. We mined high-throughput genetic interaction data sets to identify candidate genes whose function might be influenced by the absence of mms1, mms22, and rtt101 ( Costanzo ; Koh ). The genes encoding members of the Rtt101Mms1/Mms22 complex exhibited positive genetic interactions with genes encoding proteins that reside at the replication fork, including Mrc1, Csm3, and Tof1, and DNA Polε ( Figure 5A). Mrc1-Csm3-Tof1 form a replication fork–pausing complex (FPC) that tethers the MCM helicase to the leading-strand polymerase, Polε, when forks are stalled ( Katou ; Calzada ; Nedelcheva ; Szyjka ; Lou ). Although positive interactions often connect members of the same nonessential complex, analysis of genetic interaction networks showed that the vast majority of positive interactions occur between different protein complexes, which may belong to a common biological pathway ( Baryshnikova ; Costanzo ). Thus these data suggest that the fork-pausing complex or the leading-strand polymerase may be targets of Rtt101Mms1/Mms22 action.

FIGURE 5:

Genetic interactions suggest that Mms1 and Mms22 function with components of the replisome. (A) Genetic interaction map of the Rtt101-Mms1-Mms22 complex. The genetic interactions of MMS1, MMS22, and RTT101 and genes encoding components of the indicated protein complexes were compiled from high-throughput SGA experiments. Genetic interactions were determined both within and between protein complexes. Complexes are displayed as colored nodes: green when interactions between members are positive, red when interactions between members are negative, and gray when the complex includes an essential component. Nodes are connected by green lines when the between-complex interactions are positive and by red lines when the between-complex interactions are negative. (B) Western blot analysis of the indicated strains. Denatured extracts were prepared from asynchronously growing cultures (Asy) and after 90 min in 200 mM HU (90′ HU). Antibodies directed against the FLAG epitope fused to Mrc1, Csm3, and Dpb3 and against phosphoglycerate kinase (loading control) were used to detect each protein.

We assessed whether deletion of MMS1 or MMS22 altered the association of the FPC with forks when replication stress is applied. Steady-state levels of the replication fork proteins Mrc1, Csm3, and Dpb3 were not reduced by deletion of either MMS1 or MMS22, indicating that any observed effects were not due to decreased protein abundance ( Figure 5B). We examined the recruitment of Mrc1 to stalled forks genome wide by ChIP-chip in wild-type, mms1Δ, and mms22Δ strains ( Figure 6, A and B). As reported, Mrc1 localizes to origin proximal regions in HU-treated cells ( Katou ). However in mms1Δ and mms22Δ strains, the enrichment of origin proximal regions was decreased ( Figure 6, A and B, bottom). Although the signal intensity ratio was lower in the mutants, Mrc1 localization occurred at the same chromosome coordinates, suggesting that there is less Mrc1 at these sites and not a spreading of Mrc1 to adjacent regions. This effect was observed genome wide in both mms1Δ and mms22Δ mutants. Consistent with our HU sensitivity data, according to which mms22Δ mutants display a more severe growth defect under conditions of replication stress than mms1Δ mutants, deletion of MMS1 had a more subtle effect on Mrc1 localization than deletion of MMS22. To corroborate this finding, we analyzed the signal intensity peak areas at 12 early-firing origins of replication and compared the distribution in wild-type to that in mms1Δ and mms22Δ strains. The median peak area for Mrc1 was significantly decreased (p < 0.05) in mms1Δ and mms22Δ strains compared with their respective wild-type controls ( Figure 6C). Again, this effect was stronger in the mms22Δ strain. Together, the ChIP-chip data indicate that deletion of MMS1 or MMS22 reduces Mrc1 localization to stalled replication forks under conditions of fork stress. We quantified the reduction in Mrc1 localization to replication origins using qPCR to measure enrichment of the early-firing origin ARS607 compared with the inactive (in HU) origin ARS609 in at least two independent Mrc1 chromatin immunoprecipitates from each of the wild-type, mms1Δ, and mms22Δ strains. The enrichment of the early-firing origin in the Mrc1 immunoprecipitates was reduced to approximately half of the wild-type level in both the mms1Δ and the mms22Δ mutants ( Figure 6D). To confirm that this effect is not unique to Mrc1, we examined the localization of Csm3, a second component of the fork-pausing complex, to these same chromosome coordinates by qPCR. Similar to Mrc1, mms1Δ and mms22Δ strains exhibit a twofold reduction in early origin enrichment in the Csm3 immunoprecipitates ( Figure 6E). These data indicate that Mms1 and Mms22 promote the stable binding of the FPC at replication forks under conditions of replication fork stress.

FIGURE 6:

Deletion of MMS1 or MMS22 reduces the association of the FPC with stalled replication forks. 
(A, B) ChIP-chip analysis was performed following synchronous release of MRC1::flag, mms1Δ MRC1::flag, or mms22Δ MRC1::flag cells into S phase in the presence of 200 mM HU for 90 min at 23°C. After cross-linking and DNA fragmentation, Mrc1 was precipitated. Enrichment of DNA fragments in the Mrc1-bound fraction relative to the unbound fraction is shown along chromosome VI, as in Figure 4D. Replication origins (ARSs) are indicated in red for early-firing origins that are active in HU and in black for late-firing origins that are inactive in HU. (C) The distributions of signal intensity peak areas in the ChIP-chip data for 12 replication origins for the indicated strains is shown as a boxplot, with the median indicated by the horizontal bar. Peak areas were determined by extracting the signal intensity ratio values plotted in A and B at 50-nucleotide (nt) intervals for 5000 nt on either side of each origin and summing those values. Significant differences in distributions between each mutant and its wild-type control as calculated using a Mann–Whitney test are indicated. (D) Enrichment of early-origin DNA fragments relative to late-origin DNA in the Mrc1-bound and input fractions was quantified by quantitative PCR for the indicated strains. The ratio of early-origin DNA (ARS607) to late-origin DNA (ARS609) was quantified for the input and the Mrc1-bound fractions. The average of at least two experiments is plotted, and the SD is shown. (E) Enrichment of DNA fragments in the Csm3-bound fraction from chromatin immunoprecipitations of wild-type, mms1Δ, and mms22Δ strains was quantified by quantitative PCR. The ratio of early-origin DNA (ARS607) to late-origin DNA (ARS609) was quantified for the input and the Csm3-bound fractions. The average of at least two experiments is plotted, and the SD is shown.

Given that Mrc1 links Polε to the replicative helicase when forks stall ( Lou ), and since Polε exhibits positive genetic interactions with the Rtt101-Mms1-Mms22 pathway/complex, we assessed the association of Dpb3, a subunit of DNA Polε, with stalled forks. Deletion of MMS1 or MMS22 resulted in a significant reduction in the amount of Dpb3 at early-firing origins, genome wide, by ChIP-chip ( Figure 7, A–C). As was observed for Mrc1, deletion of MMS22 had a more pronounced effect on Dpb3 localization than MMS1 deletion. We quantified this reduction by qPCR and found that strains carrying deletions of MMS1 or MMS22 had a greater-than-twofold reduction in Dpb3 localization to early-firing origins of replication ( Figure 7D).

FIGURE 7:

Deletion of MMS1 or MMS22 reduces the association of DNA Polε with stalled replication forks. 
(A, B) ChIP-chip analysis was performed following synchronous release of DPB3::flag, mms1Δ DPB3::flag, or mms22Δ DPB3::flag cells into S phase in the presence of 200 mM HU for 90 min at 23°C. After cross-linking and DNA fragmentation, Dpb3 was precipitated. Enrichment of DNA fragments in the Dpb3-bound fraction relative to the unbound fraction is shown along chromosome VI, as in Figure 4D. Replication origins (ARSs) are indicated in red for early-firing origins that are active in HU and in black for late-firing origins that are inactive in HU. (C) The distributions of signal intensity peak areas in the ChIP-chip data for 12 replication origins for the indicated strains is shown as a boxplot, with the median indicated by the horizontal bar. Peak areas were determined by extracting the signal intensity ratio values plotted in A and B at 50-nt intervals for 5000 nt on either side of each origin and summing those values. Significant differences in distributions between each mutant and its wild-type control as calculated using a Mann–Whitney test are indicated. (D) Enrichment of early-origin DNA fragments relative to late-origin DNA from the Dpb3-bound and input fractions was quantified by quantitative PCR for the indicated strains. The ratio of early-origin DNA (ARS607) to late-origin DNA (ARS609) was quantified for the input and the Dpb3-bound fractions. The average of at least two experiments is plotted, and the SD is shown.

Together, these data indicate that Mms1 and Mms22 are important for the stable association of the fork-pausing complex (Mrc1/Tof1/Csm3) and the leading-strand polymerase with the replisome when replication stress is present. The consequence of the partial dissociation of these factors is fork arrest in the presence of replication stress, mitotic defects during recovery from replication stress, and loss of viability.

DISCUSSION

We have characterized the function of MMS1 and MMS22 during replication stress induced by HU. Following short-term exposure to HU, mms1Δ and mms22Δ strains have decreased viability and are unable to complete mitosis efficiently, accumulating with an anaphase-like morphology. DNA replication in the mms1Δ mutant is defective, and replication forks stall in the presence of replication stress. Finally, we find that the association of replisome components with the stressed replication forks decreases by twofold. Together our data suggest a model in which the function of Mms1, and perhaps Mms22, is to promote the stable association of the fork-pausing complex, the leading-strand polymerase, and likely other replisome components with the replication fork when replication stress is present. In the absence of Mms1 (and by extension Mms22) the dissociation of fork components results in replication fork stalling and perhaps collapse, leading to inefficient DNA synthesis during recovery from replication stress ( Figure 8).

FIGURE 8:

Mms1 and Mms22 stabilize the replisome when replication forks stall. In wild-type cells Mms1 and Mms22, either alone or as part of the Rtt101Mms1/Mms22 complex, are important for the stable association of the replisome with replication forks during replication stress. In mms1Δ or mms22Δ strains, the decrease in the association of key replisome components with replication forks results in defects in DNA replication. Under conditions of replication stress, the replisome must be stabilized at stalled forks (1). In mms1Δ/mms22Δ strains, this stabilization is reduced at a subset of replication forks (dotted arrows), causing replisome components to dissociate (2). This effect is confined to a subset of replication origins, leaving some intact forks to resume synthesis once the HU has been removed (3). At many sites, loss of an intact replisome at an origin will be compensated for by a neighboring origin. However, rare regions flanked by collapsed forks will remain unreplicated (4), resulting in mitotic delay, genomic instability, and loss of viability.

DNA synthesis in the presence of replication stress is perturbed in mms1Δ and mms22Δ cells

Our data show that DNA synthesis at replication forks in the presence of HU is defective in mms1Δ mutants. Unexpectedly, mms1Δ displayed broader replication peaks than did the wild-type strain after 45 min in HU. Because the replication peaks were shallower than in wild type 
and there was little difference evident in the flow cytometry profiles, we infer that there is considerable diversity in the amount of DNA synthesis at individual replication forks in individual cells in the population analyzed. This could reflect some degree of HU-resistant DNA synthesis, as we have observed in mutants in elg1 and pol30 (M.B. Davidson, Y. Katou, A. Keszthelyi, J. Ou, T.L. Sing, J.A. Vaisica, A. Chabes, K. Shirahige, and G.W. Brown, unpublished data), or an asynchronous or advanced entry into S phase in mms1Δ. A similar advanced DNA synthesis in HU was seen in ctf18Δ mutants, although in contrast to mms1Δ this was accompanied by checkpoint activation defects ( Crabbe ). Despite this initial burst of DNA synthesis, mms1Δ showed little change in replication peaks between 45 and 90 min, suggesting that forks had completely arrested. This contrasts with wild-type cells, which continue to synthesize DNA in the presence of HU but at lower rates ( Figure 3) ( Alvino ). Furthermore, in mms1Δ strains, we found that Polε does not colocalize with the BrdU signal after 90 min in HU, and in particular origin-distal forks appear to lack Polε ( Figure 4D). Given the replication profile of the mutant, we suggest that during replication fork stress in mms1Δ strains, replication at a subset of origins initially proceeds more rapidly than in wild type, resulting in DNA synthesis at more origin-distal sites. These forks ultimately arrest and lose replisome components, failing to synthesize DNA after 45 min in HU ( Figure 8). Consistent with the arrest and/or collapse of a subset of replication forks in HU in mms1Δ, we found that completion of S phase remained defective in the mutant even after the removal of HU, as evidenced by the persistence of mitotic forms in the mutant. Together our data are consistent with a loss of replisome integrity in mms1Δ.

Mms1 and Mms22 are important for the stabilization 
of the replisome during replication stress

We find that Mms1 and Mms22 are important mediators of Mrc1 localization to sites of stalled replication. Under conditions of replication stress, checkpoint activation is modulated by the phosphorylation of the mediator protein Mrc1 ( Alcasabas ; Naylor ). In addition to its role in checkpoint signaling, Mrc1 is present at replication forks and forms a complex with Csm3 and Tof1 called the FPC ( Katou ; Osborn and Elledge, 2003; Noguchi ; Calzada ; Nedelcheva ; Szyjka ; Lou ; Bando ; Shimmoto ). FPC function is critical not only under conditions of fork stress, but also during a normal, unperturbed cell cycle ( Tourriere ; Lou ). Our data reveal that localization of Mrc1 and Csm3 is decreased twofold in mms1Δ and mms22Δ strains, suggesting that one role of Mms1 and Mms22 is to stabilize the FPC component of the replisome under conditions of replication stress. As a component of the FPC, Mrc1 tethers the leading-strand polymerase, Polε, to the replicative MCM helicase until DNA synthesis can safely resume ( Lou ). We examined the enrichment of Dpb3, a nonessential subunit of Polε, and found that it too was decreased by at least twofold. By both ChIP-chip and ChIP/qPCR, deletion of MMS22 was found to have a more prominent effect on localization of fork proteins than deletion of MMS1 under conditions of replication fork stress. This is consistent with our data demonstrating that mms22Δ mutants are more sensitive to HU than are mms1Δ strains. Likewise, we and others have previously demonstrated that mms22Δ is more sensitive than mms1Δ to the DNA-alkylating agent MMS ( Roberts ; Zaidi ; Dovey ). Although the basis for this increased sensitivity is unknown, mms1Δ mms22Δ double mutants are no more sensitive than mms22Δ in their sensitivity to HU or MMS (unpublished data), suggesting that Mms22 has roles independent of Mms1 under conditions of replication stress and DNA damage.

The effects of MMS1 and MMS22 deletion are likely not restricted to the FPC and Polε. Like the tethering effect of Mrc1 on Polε, Ctf4 binds and regulates Polα ( Gambus ; Tanaka ), and Rtt101-Mms1-Mms22 binds Ctf4 ( Mimura ), indicating that Mms1 and Mms22 could modulate DNA Polα association with the replisome. Furthermore, RTT101, MMS1, and MMS22 have highly correlated genetic interactions with CTF4, suggesting that they may function in concert (C. Boone, unpublished data). The Rtt101Mms1/Mms22 complex could stabilize stalled replication forks by interacting with both the leading and lagging strands of forks via Ctf4 and Mrc1. Consistent with Mms1 and Mms22 promoting replisome stability during replication stress, when we examined association of Cdc45 in mms1Δ and mms22Δ strains by ChIP/qPCR we detected a twofold decrease in Cdc45 at early-firing origins of replication (unpublished data). We propose that Mms1 and Mms22, likely in complex with Rtt101, function to stabilize the replisome ( Figure 8). Dissociation of the replisome from a subset of replication forks results in the replication defects we have observed in mms1Δ and mms22Δ.

It is not clear why the levels of Mrc1, Csm3, and Dpb3 at HU-stalled replication forks decrease by only twofold in mms1Δ and mms22Δ strains. It is possible that some forks are refractory to the absence of Mms1 and Mms22, such that Mrc1, Csm3, and Dpb3 completely dissociate from some forks but are unaffected at others. If this is the case, it is unlikely that these forks represent a specific subset, because the binding signals were uniformly decreased across the genome. Rather, it is more likely to be a stochastic effect, where an effect at a given fork in one cell does not predict or reflect an effect at the same fork in a different cell. An alternative possibility is that the affinity of Mrc1, Csm3, and Dpb3 for the replisome is reduced but not eliminated at all forks. In either case the result would likely be a reduced capacity for resumption of DNA synthesis once the replication stress is removed. This is evident in the mitotic delay and reduced viability in mms1Δ and mms22Δ following recovery from HU.

Is a twofold decrease in fork residence of these proteins likely to have a biological consequence? Short HU exposures have little effect on the viability of cells lacking Mms1 or Mms22, but it is clear that even modest depletion of replisome components can have severe consequences in terms of genome integrity. For example, moderate depletion of DNA Polδ causes genome instability without obvious effects on growth rate ( Lemoine ). Our data indicate that absence of Mms1 or Mms22, and the depletion of Mrc1, Csm3, and Dpb3 at stalled forks that results, causes a defect in DNA replication in the presence of HU and during recovery from HU.

MATERIALS AND METHODS

Yeast strains and media

Yeast strains used in this study are derivatives of BY4741 ( Brachmann ) or W303 and are listed in Table 1. Standard yeast media and growth conditions were used ( Sherman, 1991).

TABLE 1:

Strains used in this study.

Strain	Genotype	Source
ABY8	MATaDPB3-6HIS-10FLAG-KanMX mms22Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
BY4741	MATaleu2Δ0 his3Δ0 ura3Δ0 met15Δ0	Brachmann et al. (1998)
CSM3-3FL	MATaCSM3-6HIS-3FLAG-loxP-KanMX-loxP leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	Katou et al. (2003)
E1670	MATaade2-1 trp1-1 can1-100 his3-11,15 leu2-3112, RAD5+ GAL psi+ ura3::URA3/GPD-TK(7x)	Etienne Schwoba
JOY128	MATaDPB3-6HIS-10FLAG-KanMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY39	MATaMMS1-6HIS-10FLAG-KanMX leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	This study
JVY54	MATaMMS1-6HIS-10FLAG-KanMX rtt101Δ::NatMX leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	This study
JVY56	MATaMMS1-6HIS-10FLAG-KanMX mms22Δ::NatMX leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	This study
JVY65	MATamms1Δ::NatMX ade2-1 trp1-1 can1-100 his3-11, 15 leu2-3112, RAD5+ GAL psi+ ura3::URA3/GDP-TK(7x)	This study
JVY72	MATamms1Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY73	MATamms22Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY76	MATaMRC1-6HIS-3FLAG-loxP-KanMX-loxP mms1Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY77	MATaMRC1-6HIS-3FLAG-loxP-KanMX-loxP mms22Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY80	MATaCSM3-6HIS-3FLAG-loxP-KanMX-loxP mms1Δ::NatMX leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	This study
JVY101	MATaCSM3-6HIS-3FLAG-loxP-KanMX-loxP mms22Δ::NatMX leu2Δ0 his3Δ0 ura3Δ0 met15Δ0	This study
JVY134	MATaDPB3-6HIS-10FLAG-KanMX mms1Δ::NatMX leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	This study
JVY157	MATaDPB3-6HIS-10FLAG-KanMX ade2-1 trp1-1 can1-100 his3-11,15 leu2-3112, RAD5+ GAL psi+ ura3::URA3/GPD-TK(7x)	This study
JVY158	MATamms1Δ::NatMX DPB3-6HIS-10FLAG-KanMX ade2-1 trp1-1 can1-100 his3-11,15 leu2-3112, RAD5+ GAL psi+ ura3::URA3/GDP-TK(7x)	This study
MRC1-3FL	MATaMRC1-6HIS-3FLAG-loxP-KanMX-loxP leu2Δ0 his3Δ1 ura3Δ0 met15Δ0	Katou et al. (2003)

aMontpellier Institute of Molecular Genetics, Montpellier, France.

Hydroxyurea sensitivity assays

To measure sensitivity to continual exposure to HU, cells were grown in 5 ml of YPD overnight at 30°C, serially diluted 10-fold, and spotted onto YPD plates containing 0, 50, 100, 150, and 200 mM HU. Plates were incubated at 30°C for 2–3 d.

To measure sensitivity to acute HU exposure, cells were grown overnight at 30°C in 30 ml of YPD. Cells were then diluted and grown to early-log phase and treated with hydroxyurea at a final concentration of 200 mM. Samples were collected at the indicated time points and plated onto YPD to determine the number of colony-forming units. The average of two independent experiments was plotted.

Flow cytometry and mitotic index

Examination of DNA content in the presence of replication fork stress and during recovery from fork stalling was performed on a Guava flow cytometer (Millipore, Billerica, MA). Data were analyzed using FlowJo Flow Cytometry Analysis Software, version 9.0 (Ashland, OR). Histograms represent the cell cycle distribution of the indicated samples. The y-axis of each graph has been scaled to represent the percentage of the maximum bin contained in that profile. Samples were also processed for parallel analysis of nuclear morphology ( Roberts ). In brief, log-phase cells were blocked in 2 μg/ml α-factor for 3 h at 23°C. Cells were then washed and resuspended in media containing 100 μg/ml pronase (Sigma-Aldrich, St. Louis, MO) and 200 mM HU (Sigma-Aldrich) for 90 min at 23°C with sampling every 30 min. Cells were then washed to remove HU, resuspended in YPD, and incubated for an additional 3 h at 23°C with sampling every 30 min. The 1-ml samples were fixed in 70% ethanol and stored at 4°C. For morphology analysis, 50 μl of cells were harvested, resuspended in 10 μl of VECTASHIELD mounting media with 4,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA), and imaged on a Zeiss Axiovert inverted microscope.

BrdU IP-chip, ChIP-chip, and qPCR

BrdU IP-chip and ChIP-chip experiments were performed essentially as described ( Katou , 2006; Roberts ). Signal intensity peak area was used to assess differences in ChIP-chip profiles between mutants and their respective wild-type controls. For each experiment, the log2 signal intensity values for probes 5000 base pairs on either side of each origin were extracted at 50-nucleotide intervals and the values summed as a measure of peak area. Twelve early-firing origins were selected, including 10 origins that were randomly chosen from the BrdU data (ARS305, ARS306, ARS315, ARS416, ARS606, ARS719, ARS737, ARS920, ARS1015, ARS1211), and ARS607 and ARS1018, which were used for qPCR analysis. The distributions of the signal intensity sums were subjected to the Mann–Whitney test to determine whether the distributions were significantly different (p < 0.05) and are presented as box plots. For quantification of signal by qPCR, input and IP samples of DNA from at least two independent chromatin immunoprecipitations were collected using the same procedure as for ChIP-chip. Following purification of the DNA using a QIAquick PCR Purification Kit (Qiagen, Valencia, CA), samples were quantified using the DyNAmo HS SYBR Green qPCR Kit (Finnzymes, ThermoFisher Scientific, Waltham, MA) in an Applied Biosystems (Foster City, CA) 7500 Real-Time PCR System. Primers for amplification of ARS607 and ARS609 have been previously described ( Lengronne ). Primer sequences were designed for the amplification of DNA proximal to ARS1018 (forward, 5′-TAA CAG AAA ATC CAG ATT TGT ACA GAA AGA-3′; reverse, 5′-ATA TGT AAC CGC AAC AGT AGC CAA-3′) and the interorigin region between ARS1009 and ARS1010 (forward, 5′-TGA ATT AGA TGC TCT TCT GTA TAC TTT CTT-3′; reverse, 5′-GGT ACA TTC ACC TTG GTT TTC AAA TAC GT-3′). Averages of at least two independent experiments were plotted with the SD.

Genetic Interaction Data

Quantitative genetic interactions were identified by synthetic genetic array (SGA) analysis ( Costanzo ) and measured as described elsewhere ( Baryshnikova ; Costanzo ). Protein complexes were assessed for biases in either positive or negative genetic interactions among its members by using a monochromatic purity (MP) score, described elsewhere ( Baryshnikova ). Complexes exhibiting only positive genetic interactions will have an MP score of +1, whereas an MP-score of −1 corresponds to complexes composed exclusively of negative interactions. Complexes whose interactions reflect the background ratio of positive to negative interactions have an MP score equal to 0. The ratio of positive to negative interactions (monochromacity) occurring between pairs of protein complexes was assessed in a similar manner. Protein complex pairs were also assessed for overall enrichment of genetic interactions using a hypergeometric p value.

Immunoblotting

Five optical density units of mid-log-phase cells were fixed with 10% trichloroacetic acid (Sigma-Aldrich), and extracts were prepared as described ( Roberts ). For treatment with hydroxyurea, cells were grown to early log phase at 23°C, arrested in G1 with 2 μg/ml α-factor for 3 h, and synchronously released into 100 μg/ml pronase and 200 mM HU (Sigma-Aldrich) for 90 min before fixation and extract preparation. Proteins were resolved on SDS–PAGE and subjected to immunoblot analysis with monoclonal ANTI-FLAG M2-Peroxidase antibody (Sigma-Aldrich) or anti–phosphoglycerate kinase monoclonal antibody (Molecular Probes, Invitrogen, Carlsbad, CA). Immunoblots were developed using SuperSignal ECL (Pierce Chemical, Rockford, IL).