Mcm2-7 Is an Active Player in the DNA Replication Checkpoint Signaling Cascade via Proposed Modulation of Its DNA Gate

Abstract

The DNA replication checkpoint (DRC) monitors and responds to stalled replication forks to prevent genomic instability. How core replication factors integrate into this phosphorylation cascade is incompletely understood. Here, through analysis of a unique mcm allele targeting a specific ATPase active site (mcm2DENQ), we show that the Mcm2-7 replicative helicase has a novel DRC function as part of the signal transduction cascade. This allele exhibits normal downstream mediator (Mrc1) phosphorylation, implying DRC sensor kinase activation. However, the mutant also exhibits defective effector kinase (Rad53) activation and classic DRC phenotypes. Our previous in vitro analysis showed that the mcm2DENQ mutation prevents a specific conformational change in the Mcm2-7 hexamer. We infer that this conformational change is required for its DRC role and propose that it allosterically facilitates Rad53 activation to ensure a replication-specific checkpoint response.

INTRODUCTION

The eukaryotic DNA replication checkpoint (DRC) monitors chromosome duplication during S phase and, if irregularities occur, facilitates numerous cellular responses that promote genome stability. In budding yeast, this checkpoint depends upon a phosphorylation cascade initiated by the upstream sensor kinase Mec1/ATR (1, 2), which in turn leads to the phosphorylation and activation of the Mrc1/claspin mediator (3, 4) and ultimately the effector kinase Rad53/Chk1 (5, 6). During deoxynucleoside triphosphate (dNTP) limitation or other forms of replication stress, DRC activation protects genome stability by Rad53-dependent phosphorylation of multiple downstream targets that serve to stabilize nascent replication forks and blocks cell cycle progression, inappropriate recombination (7–9), and the activation of late origins until the stress is alleviated (reviewed in reference 10). In addition to the DRC, a second related pathway that specifically monitors and responds to DNA damage and double-strand breaks also operates during S phase (DNA damage checkpoint [DDC]) (reviewed in reference 11).

How the DRC cascade mechanistically interacts with the core replication machinery is incompletely understood. Current evidence indicates that replication plays a passive role in the process. DNA lesions or stress causes a physical uncoupling between DNA polymerase and the replicative helicase; this in turn results in an aberrantly increased level of single-stranded DNA (ssDNA) production that leads to checkpoint activation (12–14). Correspondingly, normal replication fork formation is a prerequisite for DRC activation (15–18).

However, strong interactions between DRC components and core replication factors, even in the absence of replication stress, suggest that DNA replication in general and the MCM replicative helicase in particular play broader roles in the DRC. The mediator proteins in the cascade (Mrc1/claspin, Tof1/Timeless, and Csm3/Tipin) physically interact with and stabilize both Mcm2-7 and DNA polymerase ε (19–23) and protect fork integrity during replication stress (21, 24). Moreover, these associations are necessary for checkpoint function: loss of the physical interaction between Mrc1 and the Mcm6 subunit (25) causes DNA damage sensitivity, consistent with a DRC defect. Similarly, physical interaction between Mcm7 and Rad17, a component of the checkpoint clamp loader complex (Rad17/Rfc2-5) which, together with the 9-1-1 complex, senses replication stress, is required for normal DRC activity (26).

The present study further explores the possible roles of Mcm2-7 in DRC checkpoint activation and signal transduction. Mcm2-7 is a toroidal AAA⁺ ATPase that comprises the catalytic core of the replicative helicase that unwinds duplex DNA during replication (reviewed in reference 27). The loading and activation of Mcm2-7 are key landmark events that ensure that a single round of DNA replication occurs during each cell cycle (reviewed in reference 28). Interestingly, unlike other hexameric helicases, Mcm2-7 has a unique heterohexameric subunit composition (Mcm2 through -7) that results in 6 distinct ATPase active sites formed at dimer interfaces. This subunit organization allows a division of labor among active sites, with several sites being dedicated to DNA unwinding while other sites appear to form and possibly regulate a structural discontinuity (the Mcm2/5 gate) within the Mcm2-7 ring structure (reviewed in reference 27).

The Mcm gate appears to regulate several aspects of Mcm2-7 function. Biochemical evidence indicates that the gate-open Mcm2-7 conformation lacks helicase activity, while the gate-closed form retains activity (29). In vivo, the Mcm gate has two known functions: (i) to serve as an entry site for DNA loading during origin association (29) and (ii) to regulate DNA unwinding during G₁/S. Interestingly, structural evidence shows that Mcm gate closure and subsequent helicase activation during the G₁/S transition require the loading of the replication factors Cdc45 and GINS to generate the CMG (Cdc45–Mcm2-7–GINS) complex (30, 31). Although the regulation of the Mcm gate conformation is poorly understood, current information indicates that it is modulated by ATPase active sites that flank the gate (i.e., Mcm5/3 and Mcm6/2): mutations in conserved ATPase motifs at these sites biochemically bias the Mcm2-7 complex into a gate-closed form (32). We were therefore interested in the specific possibility that MCM-DNA gate regulation comes into play during the DRC checkpoint response to replication stress.

To address this, we studied a biochemically characterized viable nonnull allele, mcm2DENQ, that surgically inactivates the Walker B ATPase motif of Saccharomyces cerevisiae Mcm2, thereby blocking ATP hydrolysis at the Mcm6/2 active site and biasing the ring into a gate-closed conformation (32, 33). Our interest was piqued, in part, by the fact that the Mcm6 subunit was previously shown to functionally and physically interact with the Mrc1 mediator protein (25).

We characterize the effects of mcm2DENQ on the DRC and DDC responses. These results reveal that Mcm2-7 and specifically the ATPase site inactivated by mcm2DENQ are required at an intermediate step of the DRC signal transduction cascade. We suggest that the involvement of Mcm2-7 at this step helps to ensure the specific discrimination of replication stress from DNA damage stress. We propose specifically that this role is conferred directly because the open-gate conformation of Mcm2-7 allosterically assists the recruitment of Rad53 to Mrc1 to enable effector kinase activation.

MATERIALS AND METHODS

Yeast methods.

S. cerevisiae strains and plasmids are listed in Tables S1 and S2 in the supplemental material. All strains are isogenic derivatives of W303, and construction details are available upon request. For overexpression experiments, plasmids containing either MCM2 (pUP221) or mcm2DENQ (pUP223) expressed under the galactose-inducible GAL1 promoter were integrated into the LEU2 gene of the indicated strains. Cultures were grown at 30°C unless otherwise noted. G₁ cell synchronization was done by using bar1Δ strains, and fluorescence-activated cell sorter (FACS) analysis (34) was performed by using a CyAn ADP analyzer (Beckman Coulter) and the FlowJo software package (Tree Star, Ashland, OR). As both the mcm2DENQ rad9Δ and mcm2DENQ rad9Δ sml1Δ strains quickly develop methyl methanesulfonate (MMS)-resistant variants, these strains were routinely retested for MMS sensitivity prior to experimentation. As no phenotypic differences between these two strain backgrounds were observed, they were used somewhat interchangeably throughout, as detailed in the figure legends.

Immunological methods.

Tubulin immunofluorescence was performed as described previously (35), using a rat antitubulin primary antibody (YOL 1/34; Serotec) and an Alexa Fluor 546–goat anti-rat IgG secondary antibody (catalog number A11081; Invitrogen). In this assay, spindles confined to the mother cell were scored as “short” (premitotic), while those spanning both nascent daughter cells were scored as “long” (postmitotic). Coimmunoprecipitation (co-IP) (36) experiments between Mrc1–triple-hemagglutinin tag (3×HA), Csm3-3×HA, and Mcm2-7 were conducted as previously described (25), except that the medium was supplemented with 2% raffinose during galactose induction. Antibody incubation of the extract was carried out at 4°C with antihemagglutinin (anti-HA) (HA.11; Covance) or one of several pan-Mcm antibodies as indicated (monoclonal AS 1.1 [30] or polyclonal rabbit UM174 [S. P. Bell, MIT]) for 2 h, followed by a 1-h incubation with GammaBind Sepharose-protein G beads (GE Healthcare). FLAG co-IPs were performed as previously described (37), using FLAG-M2 agarose beads (Sigma).

For Mrc1, Rad9, and Rad53 phosphorylation assays and Mcm quantitative Western blot analyses, protein extracts were prepared by the trichloroacetic acid (TCA) precipitation method, and the corresponding blot was visualized by using chemiluminescence (Femto kit; Pierce) and a Fuji LAS-3000 charge-coupled-device (CCD) system in conjunction with Image Gauge software. The following Santa Cruz antibodies were used: anti-Mcm2 (catalog number SC-6680), anti-Mcm5 (catalog number SC-6686), and anti-Rad53 (catalog number SC-6749). Additional antibodies used were as follows: FLAG M2 (catalog number F1804; Sigma) and anti-glucose 6-phosphate dehydrogenase (G6PDH) (catalog number A9521; Sigma). Mcm4 and Mcm6 were visualized by using monoclonal antibodies AS6.1 and AS3.1, respectively (30; A. Schwacha and S. P. Bell, unpublished data).

Mcm2-7 protein stability and limited proteolysis.

Cycloheximide chase experiments were performed as previously described (36). Briefly, cycloheximide was added to exponentially growing cells to a final concentration of 50 μg/ml at time zero. Culture aliquots were withdrawn at the indicated times and subjected to extract preparation and Western blot analysis. Wild-type and Mcm2DENQ Mcm2-7 complexes were purified from baculovirus-infected insect cells (32) and subjected to limited proteolysis (38). Both protein preps were extensively characterized for subunit stoichiometry by quantitative Western blotting following Mcm4 immunoprecipitation (minimal hexamer contents of 81% for the wild-type Mcm2-7 preparation and 59% for the Mcm2-7 preparation containing Mcm2DENQ). To perform limited proteolysis, 2 pmol of purified protein was incubated in S/0.1 buffer (33) containing 2.5 μg/ml trypsin, 10 mM magnesium acetate (MgOAc), and 10 mM ATP (as indicated) in a final volume of 5 μl. At each indicated time point, kill cocktail (1.6× SDS loading dye, 6.67 mM phenylmethylsulfonyl fluoride [PMSF], and 6.67 μg/ml tosyl phenylalanyl chloromethyl ketone [TPCK]) in a total volume of 15 μl in S/0.1 buffer (33) was added to stop the reaction, and the tube was incubated on ice. Proteins were separated on a 7% SDS-PAGE gel and visualized by silver staining.

Sister chromatid cohesion assay.

The sister chromatid cohesion (SCC) assay was performed by using a Lac operator array integrated into chromosome IV at position 932137 (39). Cells were arrested at G₁ with α-factor for 3 h, washed, and released into fresh yeast extract-peptone-dextrose (YPD) medium containing 15 μg/ml of nocodazole for 2 h. The number of cells displaying two separate green fluorescent protein (GFP) dots in close proximity was scored by using a Zeiss Axioskop 40 microscope, and ≥100 cells were scored for each time point.

Genomic methods.

Chromatin immunoprecipitation (ChIP) was performed as described previously (21, 34), using yeast engineered to facilitate bromodeoxyuridine (BrdU) uptake (40). For experiments to examine Mcm localization in G₁, cells were arrested with α-factor at 23°C for 3 h and subsequently processed for ChIP sequencing (ChIP-seq), using either a pan-Mcm2-7 polyclonal antibody (UM174; S. P. Bell, MIT) or anti-Mcm2 (catalog number SC-6680; Santa Cruz), as indicated. Both treatments were highly concordant and exhibited >92% overlap between peaks detected with each antibody from either the wild type or the mcm2DENQ mutant. Cells used for BrdU ChIP-seq were similarly α-factor arrested but released into YPD medium containing 200 mM hydroxyurea (HU) and 400 μg/ml BrdU for 100 to 110 min at 23°C (depending upon the replicate [see the supplemental material]) and subsequently processed for ChIP-seq using an anti-BrdU antibody (catalog number 555627; BD Bioscience). Sequencing libraries were generated from the corresponding immunoprecipitated DNA by using the Illumina sample prep kit, multiplexed, and sequenced on a GAII Illumina sequencer. The resulting data were processed, assembled, and normalized by using standard methods with SCS2.6 software (see the supplemental material). Approximately 5 million reads per experiment were obtained. All genomic experiments were performed in duplicate; for presentation purposes, these replicates were combined following quantile normalization, as indicated. Details of the genomics procedure and analysis may be found in the supplemental material.

ChIP-Seq data accession number.

The primary genomics data from this paper are available from the Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo/) database under accession number GSE38032.

RESULTS

Preliminary considerations and experimental rationale.

The role of Mcm2-7 in the DRC was probed by using the mcm2DENQ allele, a substitution of the two universally conserved acidic residues of the Walker B ATPase motif for their amide counterparts (DE→NQ). Biochemically, the mcm2DENQ mutation abolishes ATP hydrolysis at the Mcm6/2 active site; however, in the context of the complete Mcm2-7 complex, it has little or no effect on in vitro DNA unwinding (32, 33). Interestingly, it is one of the few mcm mutations among many examined that demonstrate defects related to Mcm2/5 gate opening (32). We therefore focused on the effects of mcm2DENQ on DRC checkpoint activation and signal transduction.

As indicated, the mcm2DENQ allele was compared to other yeast mcm alleles with more generic defects. One was mcm4F391, a hypomorphic allele analogous to the mouse mcm4^Chaos3 mutation previously shown to cause mammary adenocarcinomas (referred to here as mcm4^Chaos3) (41, 42); the corresponding yeast mutation also confers genomic instability (43). Two other viable reference MCM alleles that each ablate the Mcm4/6 ATPase site were also examined; the Mcm4 arginine finger mutant allele (mcm4RA) and the Mcm6 Walker B mutant allele (mcm6DENQ) were previously analyzed biochemically, and in contrast to the mcm2DENQ allele, both alleles demonstrate near-normal Mcm2/5 gate conformation (32, 33).

The mcm2DENQ mutant has a classic DRC phenotype.

In wild-type cells, HU generates replication stress and activates the DRC. In contrast, exposure to the radiomimetic chemical MMS generates DNA damage and activates the DDC. In yeast, both signal transduction cascades depend on Mec1 as the sensor and on Rad53 as the terminal effector (reviewed in reference 44) but utilize different mediator proteins as the immediate targets of Mec1 phosphorylation; the DRC is specifically dependent upon Mrc1 (3), while the corresponding DDC mediator is Rad9 (45).

The mcm2DENQ mutant has phenotypes similar to those of mrc1 mutant alleles but different from those conferred by rad9Δ, implying a specific loss of DRC function.

(i) In general, DRC mutants (e.g., mrc1Δ) are specifically sensitive to HU, while DDC mutants (e.g., rad9Δ) are specifically sensitive to MMS (4, 10, 46). In the absence of either chemical, the mcm2DENQ mutant exhibited a nearly normal colony size. However, in the presence of a moderate concentration of HU, the mcm2DENQ mutant grew slowly and formed small colonies (Fig. 1A) but was relatively resistant to MMS (Fig. 1B), consistent with a defect in the DRC but not the DDC. Unlike the mcm2DENQ mutant, the mcm reference mutants produce robust colonies in the presence of HU or MMS, implying that their checkpoint response is normal (Fig. 1B and data not shown).

FIG 1.

The mcm2DENQ mutant has DRC defects. (A) The wild-type (UPY464), mcm2DENQ (UPY499), rad9Δ (UPY630), mrc1Δ (UPY713), and rad9Δ mrc1Δ sml1Δ (UPY715) strains were plated onto YPD lacking (left) or containing (middle) HU. (B) Strains from panel A as well as mcm2DENQ rad9Δ (UPY634), mcm2DENQ rad9Δ sml1Δ (UPY732), mcm2DENQ mrc1AQ (UPY758), mrc1AQ (UPY773), rad9Δ mrc1AQ sml1Δ (UPY745), mcm4^Chaos3 (UPY638), mcm4^Chaos3 rad9Δ (UPY788), mcm2DENQ plus P_GAL1-mcm2DENQ (UPY655), and mcm2DENQ rad9Δ plus P_GAL1-mcm2DENQ (UPY1231) strains were spotted onto the indicated media as 10-fold serial dilutions to assay viability. YPGal, yeast extract-peptone medium plus 2% galactose. (C and D) The indicated strains from panel B were arrested in G₁ and released into YPD plus 0.01% (vol/vol) MMS, and tubulin immunofluorescence was assayed. (C) mcm2DENQ rad9Δ cells undergo inappropriate nuclear segregation and spindle elongation after 4.5 h of exposure to MMS. DAPI, 4′,6-diamidino-2-phenylindole. (D) Time course analyses of spindle elongation following G₁ release. Data represent means and standard deviations from 3 experiments. (E) Viability following MMS exposure. Strains and growth conditions are identical to those described above for panel D, and the cultures were plated at the indicated intervals on YPD without drug to measure viability. Results were normalized to the starting culture viability; means and standard deviations from 3 independent analyses are shown. (F) FACS analysis of the indicated isogenic strains from panel A. Briefly, strains were arrested as described above for panel D and released into fresh YPD with or without 0.033% MMS. Aliquots were taken at the indicated times and processed for FACS analysis. After 45 min (wild type) or 55 min (mcm2DENQ mutant), α-factor was readded to restrict analysis to a single cell cycle. Asterisks denote time points that represent either the apparent start (black) or end (red) of S phase.

(ii) Induction of either a DRC or a DDC response ultimately results in Rad53 phosphorylation and its activation, which triggers a checkpoint response via the subsequent phosphorylation of numerous downstream targets (reviewed in reference 44). However, there is significant cross talk between the two responses (47, 48), and their partial overlap causes diagnostic mutant phenotypes. First, single DRC- or DDC-specific mutants (e.g., mrc1Δ or rad9Δ) still exhibit near-wild-type resistance to the orthologous challenge, i.e., MMS or HU, respectively (Fig. 1A and B). In contrast, the presence of both mutations synergistically results in extreme sensitivity to either chemical (Fig. 1A and B). (It should be noted that although the simultaneous loss of both the DDC and DRC is lethal [e.g., mec1Δ, rad53Δ, and the rad9Δ mrc1Δ double mutant {3, 49, 50}], such lethality can be suppressed by deletion of the SML1 gene [51], which is included in our strains as indicated.)

In analogous double mutant combinations, the mcm2DENQ mutation confers the same effects as canonical DRC-defective alleles (Fig. 1B). First, the mcm2DENQ rad9Δ double mutant was synergistically sensitive to MMS exposure. Second, a combination of mcm2DENQ and mrc1AQ, a nonphosphorylatable DRC allele (4), resulted in the same MMS sensitivity as the corresponding single mutants. (Note that the mcm2DENQ mrc1Δ double mutant is inviable and could not be tested.) Several additional observations suggest that the MMS sensitivity of the mcm2DENQ rad9Δ double mutant stems from a loss of ATP hydrolysis at the Mcm6/2 active site. This phenotype does not arise from a general reduction in mcm2DENQ protein levels, as overexpression of this allele provided no substantial increase in MMS resistance (Fig. 1B, bottom). Moreover, this DRC phenotype was specific to the mcm2DENQ allele, as none of our other reference alleles, either by themselves or in combination with rad9Δ, caused increased MMS sensitivity (Fig. 1B and data not shown). Therefore, we infer that a DRC defect does not simply arise as a secondary consequence of a general mcm defect. As the mcm2DENQ rad9Δ double mutant is considerably more sensitive to MMS than to HU, most of the experiments described below examine the effects of MMS rather than HU unless otherwise specified.

(iii) The DDC and DRC both block cell cycle progression in response to MMS-induced DNA damage. In the DDC-defective rad9Δ mutant, cell cycle progression continues in the presence of DNA damage after a significant lag (52). This lag reflects a concomitant activation of the DRC (see above) and thus is eliminated by DRC-specific mutations. Thus, the simultaneous loss of both checkpoint pathways results in a synergistic increase in aberrant cell cycle progression in the presence of MMS.

By this criterion, the mcm2DENQ mutant exhibits the same phenotype as those of canonical DRC mutants upon MMS exposure. To assay inappropriate mitotic entry, we examined spindle elongation using tubulin immunofluorescence (Fig. 1C). As previously noted (4), wild-type cells grown in the presence of MMS arrest with short spindles that do not extend into the bud, whereas 10 to 20% of cells containing either a rad9Δ or an mrc1Δ mutation attempt cell division and acquire long spindles spanning the bud (Fig. 1D). These two mutations function synergistically, as nearly 50% of rad9Δ mrc1Δ sml1Δ cells acquire long spindles under these conditions (Fig. 1D), indicating abnormal mitotic entry. The mcm2DENQ rad9Δ (sml1Δ) double mutants similarly acquired long spindles, nuclear fragmentation, and inviability (Fig. 1C to E) albeit with somewhat slower kinetics than with a canonical DRC-DDC mutant (e.g., rad9Δ mrc1Δ sml1Δ).

Additionally, FACS analysis demonstrates that, like many checkpoint mutants, the mcm2DENQ mutant grew more slowly than the wild-type strain during unchallenged growth (Fig. 1F), with slightly prolonged S (∼10-min) and G₂ (∼10- to 20-min) phases. Despite the slow growth, the mcm2DENQ mutant by itself has a normal response to DNA damage; in the presence of MMS, the wild-type and mcm2DENQ strains both proceed very slowly through S phase with identical kinetics (Fig. 1F). Thus, the mcm2DENQ mutant has an essentially normal cell cycle response to DNA damage, reflecting the functional redundancy between the DDC and DRC pathways.

However, similar to other DRC alleles, the mcm2DENQ allele demonstrates a synergistic loss of checkpoint control when ablated for the DDC mediator rad9Δ. Although the rad9Δ mutant by itself only partially eliminated the MMS-induced block to S-phase progression, the mcm2DENQ rad9Δ double mutant almost completely eliminated the MMS-induced block to S-phase progression and proceeded through the cell cycle identically in both the presence and absence of MMS (Fig. 1F).

Taken together, the above-described mcm2DENQ phenotypes imply that Mcm2-7 has a specific role in the DRC but no discernible role in the DDC.

The mcm2DENQ mutant demonstrates normal origin loading but defective late origin firing.

To better understand the relationship between the DRC and replication, we assessed the replication ability of the mcm2DENQ mutant. DRC mutations (e.g., in MEC1 or MRC1) inappropriately activate late-firing origins (21, 53) in the presence of replication stress. In contrast, specific loss of the DDC (i.e., rad9Δ) results in little or no change in origin firing under these conditions (53).

We first tested the mcm2DENQ mutant for obvious replication defects. A primary requirement of replication is that Mcm2-7 properly associates with replication origins during G₁ phase (prereplication complex formation) (reviewed in reference 54). Toward this end, we assessed the localization of the Mcm2-7 helicase at origins in G₁ from wild-type and mcm2DENQ strains by ChIP-seq. We identified 377 genomic Mcm association sites in the mcm2DENQ mutant (Fig. 2A). Nearly all such sites were qualitatively similar and concordant with those previously reported for the wild-type strain (oriDB [55]) (Fig. 2A). Thus, the mcm2DENQ mutant appeared to be essentially normal for Mcm2-7 origin loading.

FIG 2.

The mcm2DENQ mutant demonstrates inappropriate late origin firing. Growth conditions are described in Materials and Methods. (A) Mcm ChIP-seq analysis of the wild-type (UPY493) and mcm2DENQ (UPY524) strains during G₁ arrest using the pan-Mcm antibody. A representative region of chromosome XIV is shown, and reads per kilobase per million mapped reads (RPKM) are plotted. ARS, autonomously replicated sequence. (B) BrdU ChIP-seq data for HU (200 mM)-arrested wild-type (UPY493), mcm2DENQ (UPY524), and mrc1Δ (UPY722) strains. The data are plotted similarly to the data in panel A. (C) Box-and-whisker plots showing the medians and quartiles of BrdU enrichment for wild-type (UPY493), mcm2DENQ (UPY524), and mrc1Δ (UPY722) strains at different subsets of origins (“mrc1AQ-specific origins,” those that fire only in the mrc1AQ mutant, and “all late origins minus mrc1AQ origins,” those that fire in the mrc1Δ but not in the mrc1AQ mutant). Significance was determined by a one-tailed Wilcoxon rank sum test to examine if the mutants have differences that are greater than the wild type (WT) (in panel 1, W [sum of ranks] = 18,903 and P = 0.531 for the wild-type and mcm2DENQ strains, and W = 21,619 and P = 0.994 for the wild-type and mrc1Δ strains; in panel 2, W = 26,665 and P < 0.0001 for the wild-type and mcm2DENQ strains, and W = 3,901 and P < 0.0001 for the wild-type and mrc1Δ strains; in panel 3, W = 2,923 and P < 0.0001 for the wild-type and mcm2DENQ strains, and W = 429 and P < 0.0001 for the wild-type and mrc1Δ strains; and in panel 4, W = 10,472 and P = 0.399 for the wild-type and mcm2DENQ strains, and W = 1,065 and P < 0.0001 for the wild-type and mrc1Δ strains). Results for mrc1AQ were reported previously (53).

To examine origin firing in the mcm2DENQ mutant, we modified our ChIP-seq approach to enrich for nascent DNA fragments containing newly incorporated bromodeoxyuridine (BrdU) (56). We observed that during S phase, the sites of BrdU incorporation in the wild-type, mrc1Δ, and mcm2DENQ strains mapped to sites of prereplicative complex (pre-RC) formation at previously identified origins (Fig. 2B) (55). However, the origin subsets activated in the presence of HU in the mcm2DENQ and mrc1Δ strains differed significantly, and similarly, from those observed for the wild type (Fig. 2B).

In comparison to previously reported origin utilization data (53), we found no significant quantitative differences in BrdU incorporation at early origins among the three strains (Fig. 2C1). However, in contrast, we observed, as expected, a marked increase in BrdU incorporation at late origins in the mrc1Δ strain (Fig. 2C2). The mcm2DENQ strain exhibited an intermediate phenotype, with significantly more BrdU incorporation at this subset of late origins than in the wild type (Fig. 2B). Previous studies have shown that different DRC mutants (e.g., mrc1AQ) activate different arrays of late origins (53). Comparison of BrdU incorporation between mrc1AQ-specific origins and all late-activating origins (i.e., those activated in mrc1Δ) revealed a significant increase in BrdU incorporation in the mcm2DENQ mutant (P value of <0.0001) (Fig. 2C3). In contrast, the mcm2DENQ mutant demonstrated no significant enrichment of BrdU in the set of “all late origins minus mrc1AQ origins” that fire in the mrc1Δ strain but not in the mrc1AQ strain (P value of 0.4) (Fig. 2C4). Together, these results underscore the similarities between the mcm2DENQ mutant and established DRC mutants.

FIG 3.

Mcm2-7 is part of the DRC phosphorylation cascade. Representative Western blots are shown from asynchronous cultures grown with or without 200 mM HU or 0.033% MMS for 90 min, as indicated. Induced phosphorylation was calculated as follows: % induced phosphorylation = [(low-mobility species)/(total species)] × 100. Values shown were corrected for any phosphorylation observed in the absence of exogenous DNA damage. (A) Mrc1 phosphorylation. Mrc1-3×HA was probed by using anti-HA antibody to the wild-type (UPY646), mcm2DENQ (UPY647), rad9Δ (UPY659), mcm2DENQ rad9Δ (UPY660), tel1Δ sml1Δ (UPY968), mcm2DENQ tel1Δ sml1Δ (UPY967), and tel1Δ mec1Δ sml1Δ (UPY985) strains. Values shown represent the averages of data from two repeats, and the range between repeats is ≤5%. (B and C) Rad53 phosphorylation. The indicated strains from Fig. 1A as well as the mcm6DENQ (UPY525), mcm6DENQ rad9Δ (UPY919), mcm4RA (UPY529), and mcm4RA rad9Δ (UPY 918) strains were used. Values shown represent the averages of data from ≥3 repeats; in all cases, the standard deviation is <10%. (D) Rad9 phosphorylation. Rad9-3×HA was probed as described above for panel A. Data for the wild-type (UPY648), mcm2DENQ (UPY649), and mcm2DENQ mrc1AQ (UPY758) strains are shown. Values shown represent the averages of data from 2 repeats, and the range between repeats is ≤5%. Note that epitope-tagged Mrc1 and Rad9 are functional and partially alleviate the MMS sensitivity of a rad9Δ mrc1Δ sml1Δ strain (data not shown). *, as previously observed (3), HU causes minimal Rad9 phosphorylation in wild-type and mrc1Δ (and apparently mcm2DENQ) strains.

FIG 4.

(A) Mcm2-7–Mrc1 coimmunoprecipitation. Cell extracts from asynchronous cultures of the wild-type (UPY1044) or mcm2DENQ (UPY1045) strain were immunoprecipitated with anti-HA or anti-MCM antibodies, as indicated. Mrc1 and Mcm6 were visualized by Western blotting using either anti-HA or anti-Mcm6. (B) Similar to panel A except that cells were arrested in S phase by the addition of 200 mM HU, the presence of Mcm proteins was probed with antibody UM174, and blots were visualized by exposure to X-ray film. (C) Similar to panel A, where interactions between Mcm2-7 and Csm3 were tested by reciprocal co-IPs using strains carrying Csm3-3×HA in a wild-type (UPY1057) or mcm2DENQ (UPY1058) background. Mcm2-7 in this experiment was visualized by using antibody UM174. (D) Co-IP experiments analogous to those described above for panel A were conducted on wild-type (UPY1044), tof1Δ (UPY1053), and csm3Δ (UPY1054) strains that all contained the Mrc1-3×HA construct. (E) Wild-type (UPY1101) and mcm2DENQ (UPY1102) strains containing CDC45-3×HA were synchronized and either maintained in α-factor arrest or released into YPD with or without 200 mM HU or 15 μg/ml nocodazole for 2 h to arrest cells in either G₁, S, or G₂ phase, respectively. IP experiments were performed similarly to those described in Materials and Methods using antibody UM174, except that potassium acetate was replaced with potassium glutamate. Samples were analyzed for Mcm2-7 and coprecipitating Cdc45-3×HA via Western blots. (F) Sister chromatid cohesion was assayed in synchronized cells arrested in G₂ for 120 min in nocodazole for the wild-type (UPY613), eco1-1 (K7542), mrc1Δ (UPY744), mrc1AQ (UPY822), mcm2DENQ (UPY606), mcm4^Chaos3 (UPY835), mcm4RA (UPY811), mcm6DENQ (UPY812), and mcm2DENQ/P_GAL1::mcm2DENQ (UPY884) strains. For UPY884, results are shown following growth in glucose (normal Mcm2DENQ levels) and galactose (overexpressed [OE] Mcm2DENQ). Standard deviations of the data are shown.

The mcm2DENQ mutant can sense replication stress but is partially defective in Rad53 phosphorylation.

In contrast to the reference mcm alleles, we demonstrate that the mcm2DENQ mutant is defective in the DRC checkpoint response to HU-induced replication stress. In principle, such a defect could presumably result from a trivial failure to generate the requisite ssDNA signal, due to either a reduced number of replication forks or a partial defect in DNA unwinding (14, 15). However, a more interesting alternative is that the defect arises from disruption of some critical checkpoint function occurring downstream of Mec1 activation that requires a direct involvement of Mcm2-7 in the DRC cascade.

To distinguish between these two possibilities, we investigated the effects of mcm2DENQ on phosphorylation of Mrc1 and Rad53 in response to HU or, as a control, MMS. As described above, either challenge activates both the DRC and the DDC albeit to different relative extents. Members of the cascade are activated through phosphorylation, which can be easily assayed via mobility shifts observed following Western blotting (4, 57, 58).

In wild-type cells under replicative stress, Mrc1 is phosphorylated by Mec1. This, in turn, is required for Rad53 activation and is a diagnostic indicator of initial DRC activation (3, 4). Although we did not examine of the kinetics of Mrc1 phosphorylation following exposure to replication stress, we observed identical steady-state levels of Mrc1 phosphorylation in both the wild-type strain and the mcm2DENQ mutant upon HU treatment. Moreover, identical results were observed for the mcm2DENQ rad9Δ double mutant in HU, where any possible contribution of the DDC is eliminated (Fig. 3A). Previous reports indicate that under certain conditions, Mrc1 can be phosphorylated by either the Mec1 or Tel1 kinase (59). We found that neither the tel1Δ sml1Δ nor the mcm2DENQ tel1Δ sml1Δ mutant blocks Mrc1 phosphorylation, while a mec1Δ tel1Δ sml1Δ triple mutant completely eliminated Mrc1 phosphorylation (Fig. 3A). (Note that the mcm2DENQ mutant is inviable in combination with the mec1Δ sml1Δ mutation.)

Thus, the mcm2DENQ mutant is able to initially recognize and initiate early stages of the DRC cascade in response to replication stress, strongly suggesting that the DRC defect in this mutant is not a trivial consequence of insufficient ssDNA production during replication stress.

The mcm2DENQ mutant must therefore be defective at a later stage of the DRC. To examine this possibility, we next assayed Rad53 phosphorylation in response to either HU or MMS. The wild-type strain, mcm reference alleles, and single mutants defective in either the DRC or DDC all still exhibited robust Rad53 phosphorylation in response to either HU or MMS (40 to 60% of Rad53 in one of several higher-molecular-weight forms), while double mutants that eliminate both pathways (mrc1Δ rad9Δ sml1Δ) completely eliminated phosphorylation (Fig. 3B and C). Furthermore, none of the reference mcm alleles showed any reduction in Rad53 phosphorylation when combined with a rad9Δ mutation in the presence of either MMS or HU (Fig. 3C), indicating proficient DRC activation and implementation (see above). In contrast, the mcm2DENQ mutation conferred the same phenotypes as those of canonical DRC mutations in this test. We observed robust phosphorylation in the mcm2DENQ single mutant and a 2- to 3-fold decrease in Rad53 phosphorylation in the mcm2DENQ rad9Δ double mutant. Moreover, a double mutant containing both mcm2DENQ and a canonical DRC mutation (mrc1AQ) exhibited only a modest reduction in Rad53 phosphorylation, similar to the respective single mutants (Fig. 3B), consistent with Mrc1 and Mcm2-7 functioning in a common pathway. Finally, in the presence of MMS, both the mcm2DENQ mutant and an mcm2DENQ mrc1AQ double mutant demonstrated robust Rad9 phosphorylation (Fig. 3D), confirming normal integrity of the DDC cascade in this strain.

In summary, these results indicate that the DRC phenotype of the mcm2DENQ mutant is not caused by a generic replication defect but rather is caused by a specific defect in the checkpoint cascade, presumably after Mec1 activation and Mrc1 phosphorylation but prior to Rad53 activation. Thus, in addition to its previously described roles (60, 61), Mcm2-7 has an independent specific role in DRC-mediated signal transduction.

The mcm2DENQ mutant maintains associations among Mcm2-7, the DRC mediator proteins, and CDC45.

Although our data indicate that faulty DNA replication is not responsible for the mcm2DENQ DRC phenotype (see above), other indirect causes for this phenotype were also considered.

One such possibility is that within the mcm2DENQ mutant, Mcm2-7 fails to physically interact with the DRC mediator Mrc1, Tof1, or Csm3 (19, 20, 62). To test this possibility, we performed reciprocal coimmunoprecipitation experiments (Fig. 4A). In the mcm2DENQ mutant, we found that the physical association between the Mcm complex and Mrc1 was robust and maintained at nearly wild-type levels both in asynchronous culture (Fig. 4A) as well as during replication stress (Fig. 4B). As the association of Mrc1 with Mcm2-7 has been reported to require Tof1 and Csm3 (19), the robust physical association between Mrc1 and Mcm2-7 in the mcm2DENQ mutant strongly implies that these other two checkpoint mediators properly interact with Mcm2DENQ-containing Mcm2-7 complexes. Our results confirm that the loss of Tof1 greatly reduces the association of Mrc1 with Mcm2-7, implying that the physical association between Mcm2-7 and Tof1 is robust in the mcm2DENQ mutant (Fig. 4D). In contrast to the tof1Δ mutant, loss of Csm3 appears to cause little change in the Csm3–Mcm2-7 association (Fig. 4D). However, we detected an equivalent interaction between Mcm2-7 and Csm3 in both wild-type and mcm2DENQ strains by co-IP experiments (Fig. 4C), suggesting that Csm3 is not needed for the association between Mrc1 and Mcm2-7 (Fig. 4D). Thus, the physical association between Mcm2-7 and both Mrc1 and Csm3 is uncompromised by the mcm2DENQ mutation.

We next examined whether the mcm2DENQ mutation destabilizes the CMG complex during replication stress (Fig. 4E). In the presence of HU, during which the CMG complex would be expected to normally remain intact, Cdc45 remains associated with Mcm2-7 in both the wild-type and mcm2DENQ strains. In contrast, under conditions were the CMG complex is inactive (e.g., during G₁ or G₂), Cdc45 loses the Mcm2-7 association in both the wild-type and mcm2DENQ strains. Thus, the mcm2DENQ mutation appears to cause little or no instability of the CMG complex in the presence of replication stress.

In summary, these results largely rule out the possibility that the mcm2DENQ DRC phenotype is due to a gross inability to recruit either the DRC mediators or Cdc45 to forks or to maintain their physical association during replication stress.

The mcm2DENQ DRC defect is not attributable per se to a loss of sister chromatid cohesion.

Sister chromatid cohesion (SCC) involves the stable physical association between sister chromatids during G₂ (reviewed in reference 63). Most DRC mutants have associated SCC defects (64, 65), raising the possibility that that the DRC defect in the mcm2DENQ mutant is a secondary consequence of faulty cohesion. We examined this using a cytological assay that visualizes a specific test chromosome region marked by a lac operator array bound to a GFP-LacI fusion protein (39).

We first examined SCC during G₂ arrest in the presence of nocodazole (Fig. 4F). In contrast to wild-type cells (∼2% cells with an SCC defect), ∼35% of cells of a cohesin establishment mutant (eco1-1 [66]) contain two GFP foci, indicative of a severe SCC defect. The mcm2DENQ mutant demonstrates an intermediate but significant defect in SCC relative to the wild-type strain (P = 0.004), with a 7-fold increase in defective SCC, with a magnitude similar to that previously observed for an mrc1Δ mutant (P = 0.09 for mcm2DENQ versus mrc1Δ) (65). In common with its DNA damage sensitivity phenotype (Fig. 1B), overexpression of the mcm2DENQ allele is unable to significantly suppress this SCC phenotype, indicating that this defect is not due to a quantitative loss of Mcm2 protein. Furthermore, we find that both the mcm4^Chaos3 (P = 0.07) and mcm6DENQ (P = 0.001) mutants lacking obvious checkpoint defects also demonstrate an SCC defect relative to the wild-type strain, although the difference between mcm4^Chaos3 and the wild type lacks statistical significance (Fig. 4F). In contrast, the mcm4RA mutant has essentially normal SCC relative to the wild-type strain (P = 0.2).

In summary, these data fail to show an obligatory connection between defects in the DRC and those in SCC. In contrast to the DRC defect, the SCC defect appears to be a general phenotype common to diverse MCM alleles. Formally, this finding implies that an SCC defect is insufficient to activate the DRC and thus is not per se the unique cause of the mcm2DENQ DRC defect. It remains to be determined why checkpoint-defective mutants exhibit an SCC defect and, as a separate effect, why SCC is also abrogated by canonical mcm alleles that do not activate the checkpoint.

The mcm2DENQ allele demonstrates few or no off-target defects.

Amino acid substitutions often cause collateral defects in protein expression, folding, or stability (e.g., see reference 67). We examined these issues to determine if they were sufficient to explain the mcm2DENQ phenotypes. Potential collateral defects of this allele in Mcm2 protein expression or stability were examined and ruled out (Fig. 5A and B). Moreover, coimmunoprecipitation experiments indicated that the allele has essential no effect on the assembly or stability of the Mcm2-7 complex in vivo (Fig. 5C), while limited proteolysis experiments using recombinant Mcm2-7 complexes demonstrated that the mcm2DENQ mutation imparts little apparent change to the in vitro conformational stability of the Mcm2-7 complex (Fig. 5D). In contrast, parallel experiments that measure the expression and stability of the mcm6DENQ and mcm4RA proteins demonstrate slight reductions of expression (both alleles) and stability (mcm6DENQ) (Fig. 6). Therefore, we surmise that the DRC defects observed with the mcm2DENQ mutant are unlikely to be caused by a quantitative lack of Mcm2-7.

FIG 5.

The mcm2DENQ mutant lacks significant secondary defects. (A) The wild-type (UPY464) and mcm2DENQ (UPY499) strains were harvested either during asynchronous growth in YPD (Asy) or at the indicated time points following G₁ arrest and release. Cell extracts were analyzed by quantitative Western blotting using antibodies to either Mcm2 or the loading control G6PDH. (B) In vivo Mcm2 stability. Cycloheximide was added to asynchronous cultures of the wild-type and mcm2DENQ strains from panel A and the wild-type strain transformed with an ARS/CEN plasmid encoding carboxypeptidase Y (CPY*) (pUP1106) fused to a 3×HA epitope tag (CPY* is a well-characterized endoplasmic reticulum-associated degradation substrate [36]). Protein extracts were analyzed by Western blotting similarly to panel A, except that anti-HA antibodies were used to probe for CPY*. (Top) representative blots; (bottom) graph of the means and standard deviations from 3 independent assays. (C) Co-IPs of Mcm subunits from an asynchronous culture. UPY1044 (wild type) and UPY1045 (mcm2DENQ mutant) each contain the Mcm4-3×FLAG epitope tag. Western blots were probed with the indicated Mcm subunit-specific antibodies. Rec, recombinant purified Mcm2-7. (D) Limited proteolysis of Mcm2-7 complexes. Shown are silver-stained SDS-PAGE gels of both purified wild-type (top) and mcm2DENQ (bottom) Mcm2-7 hexamers treated with trypsin; numbers represent minutes digested in the presence of active trypsin. Note that the Mcm2, -4, and -7 subunits comigrate in the S. cerevisiae Mcm2-7 complex on SDS-PAGE gels. MW, molecular weight. ATP makes wild-type Mcm2-7 relatively resistant to proteolysis, as shown, and the digests were conducted in both the presence (+) and absence (−) of 10 mM ATP.

FIG 6.

Expression and stability of the Mcm4RA and Mcm6DENQ proteins. (A) In vivo Mcm6DENQ and Mcm4RA protein stability. The wild-type (UPY464), mcm6DENQ (UPY525), and mcm4RA (UPY529) strains were assayed in a manner similar to that described for Fig. 5B. (B) In vivo Mcm4RA and Mcm6DENQ protein expression. The strains from panel A were tested in a manner similar to that described for Fig. 5A; values shown represent the means and standard errors of the means from ≥3 independent experiments.

DISCUSSION

We show that Mcm2-7 is an integral component of the DRC signal transduction cascade, acting essentially at the same step as the Mec1/ATR-mediated phosphorylation of Mrc1. This activity specifically requires the Mcm6/2 ATPase active site that was previously implicated as having a regulatory rather than a DNA-unwinding role in Mcm2-7 (32, 68, 69). Correspondingly, reference mcm alleles that do not affect the Mcm6/2 active site have apparently normal checkpoint signaling (Fig. 1B, 3C, and 6). These results confirm and extend previous findings that implicated Mcm2-7 in the DRC response to MMS-induced damage but did not detect a role in HU-induced replication stress (25).

Our previous studies showed that the Mcm6/2 ATPase site is involved in regulating the opening and closing of the Mcm DNA gate, with the mcm2DENQ mutation biasing the gate into a closed conformation (32). By implication, the role of Mcm2-7 in activating the DRC involves modulation of the Mcm gate, more specifically an open form of the gate, either statically or via cyclic opening and closing. Additional considerations discussed below suggest that the direct involvement of Mcm2-7 in the DRC reflects a broader role for this molecular complex as an important functional connection between the DRC and DNA replication.

As Mcm2-7 is highly conserved among all eukaryotes (reviewed in reference 27), its direct participation in the budding yeast DRC implies a role in the prevention of genomic instabilities linked to DRC defects in cancer and other diseases (70).

Involvement of the Mcm6/2 ATPase active site in the DRC response.

The Mcm6/2 active site biochemically functions to regulate rather than directly participate in DNA unwinding. Early work showed that a specific Mcm subcomplex (Mcm467) was competent to unwind DNA (71), while the addition of the remaining Mcm subunits (i.e., Mcm2, -3, and -5) inhibited this activity (69, 72). In part, this inhibition depends upon conserved ATPase motifs in Mcm2 (i.e., the function of the Mcm6/2 active site) rather than altered oligomerization caused by Mcm2 binding to the Mcm467 subcomplex (69); mutations that reduce but do not eliminate ATP hydrolysis at this site are viable but demonstrate sensitivity to DNA-damaging agents, consistent with a potential checkpoint defect (68).

Moreover, the Mcm6/2 ATPase site is connected functionally and physically to the DRC. Mrc1 binds the C terminus of Mcm6 (25). MCM6 mutants that ablate this interaction are sensitive to DNA-damaging agents and reduce Rad53 phosphorylation in combination with the rad9Δ allele. This phenotype is due specifically to the loss of the Mrc1 association, as the phenotype of this Mcm6 mutant can be suppressed by an Mcm6-Mrc1 fusion protein (25).

DRC mutants are traditionally sensitive to HU (4). Interestingly, both studies described above found that perturbations to the Mcm6/2 site (reduction in ATP hydrolysis or loss of Mrc1 association) cause greater sensitivity to MMS than to HU (25, 68). To a somewhat lesser extent, the mcm2DENQ mutant has similar properties (Fig. 1B). Although the basis of this effect is unknown, it is useful to consider that MMS generates DNA damage ahead of the replication fork, suggesting perhaps that the Mcm6/2 active site is dedicated to dealing with template problems associated with DNA unwinding.

Mcm2-7 as a central coordinator of the DRC.

A checkpoint response has three basic requirements: sensing of the presence of a lesion, transduction of that information into activation of the signaling cascade, and targeting of the effector kinase to elicit the appropriate checkpoint responses. We propose that Mcm2-7 is directly involved as a central coordinator of the DRC.

The need for such an integrative role is emphasized by the fact that in budding yeast, Mec1 is the sensor kinase of both the DRC and the DDC, and similarly, Chk1/Rad53 is the ultimate downstream effector kinase of both pathways but nonetheless mediates appropriate, distinct responses in the two cases. How are appropriate specificities conferred in both the input and output stages? While the two responses are distinguished by alternative mediator proteins, some uncertainty remains about the source of information that actually directs the appropriate response. Mcm2-7 is directly involved in the DNA replication process and, as such, is positioned to mediate DRC function.

Our data in combination with previous results support an integrative role of Mcm2-7 in the DRC. First, Mcm2-7 has a clear mediator function. We show that during an in vivo block to DNA replication, an additional component, Mcm2-7, is required for efficient effector kinase activation. This block is prior to Rad53 activation but after or concurrent with Mrc1 phosphorylation (Fig. 3). Moreover, the mcm2DENQ allele and a phosphorylation-defective mrc1 allele (mrc1AQ) confer essentially the same defect (Fig. 3B), shown by epistasis analysis to occur at the same step in the signal transduction process. Thus, by implication, the missing mediator function is provided by Mcm2-7.

Second, Mcm2-7 likely ensures that Rad53 is activated as part of the DRC to yield a sensible replication-dependent response. Previous results suggest that mediators in addition to Mrc1 (and possibly Csm3 and Tof1) are required in vivo to facilitate appropriate Rad53 activation. Under simplified conditions using either a purified biochemical system (73) or an engineered in vivo system (74), evidence indicates that the only limitation to Rad53 phosphorylation in the DRC is the physical association between Mec1 and Rad53. This physical connection is greatly stimulated by phosphorylated Mrc1, which binds to both kinases. However, in both of these experimental systems, Rad53 activation occurs efficiently even in the absence of DNA replication or damage. As normal DNA replication by its very nature generates replication protein A (RPA)-coated ssDNA known to induce the DRC, additional constraints must be present in vivo to prevent inappropriate DRC activation. Our data confirm this supposition, as we show that, in contrast to the simplified systems, Mcm2-7 is additionally required for full Rad53 activation. As Mcm2-7 is an essential component of active replication forks, this ensures that DRC activation is replication dependent and may additionally help prevent fortuitous DRC or DDC activation.

Models for Mcm2-7 involvement in the DRC signal transduction cascade.

Our previous biochemical evidence indicates that the mcm2DENQ mutation interferes with the ability to open the MCM gate, implying that the mutant protein exists predominantly in a gate-closed conformation (32). We propose that a conformational change involved in gate opening/closing directly regulates the physical interactions between Rad53 and Mrc1.

One specific scenario for the involvement of Mcm2-7 in DRC activation could be as follows. Mcm2-7 functions primarily as a mediator in the DRC (see above). In the absence of replication stress, physical interaction with Mcm2-7 occludes the Rad53-interacting surface of Mrc1 (Fig. 7). Upon DRC activation, a conformational change in Mcm2-7, fueled by ATP hydrolysis at the Mcm6/2 active site (below), relieves Mrc1 inhibition and promotes its Rad53 activation (Fig. 7). As the mcm2DENQ mutant is fundamentally unable to hydrolyze ATP at the Mcm6/2 active site, a reasonable prediction would be that the protein fusion between Mcm6 and Mrc1, previously shown to be competent for Rad53 phosphorylation in the presence of an mrc1Δ mutation (25), would be unable to suppress the mcm2DENQ checkpoint defects. However, the testing of this prediction is complicated by our finding that the Mcm6-Mrc1 fusion construct unexpectedly causes lethality in an mcm2DENQ mutant (F.-L. Tsai, unpublished data). Our hypothesis, which couples a conformational change in a motor protein to a regulatory output, is similar to a recently proposed model that connects defects in topoisomerase II to trigging of a G₂ checkpoint (75).

FIG 7.

Model of Mcm2-7 involvement in the DRC cascade. See the text for details. The conformational change of Mcm2-7 from a closed to an open structure causes a conformational change in Mrc1 that allows its association and subsequent activation (phosphorylation) by Mec1 (not shown). In addition, Mcm gate opening upon DRC activation possibly acts to stop replication elongation. Asterisks denote phosphorylation.

In this scenario, we speculate that upstream activation of the DRC leads to appropriate Mcm ATP hydrolysis and a conformational change in the Mcm ring structure. This outcome then facilitates both subsequent Rad53 phosphorylation as well as helicase inactivation (Fig. 7). This scenario implies that the DRC mediator proteins (particularly Mrc1) should function to regulate Mcm ATP hydrolysis. While such an analysis of Mrc1 has yet to be undertaken, the human proteins that correspond to the budding yeast proteins Tof1 and Csm3 (i.e., Timeless and Tipin) have been shown to block both ATP hydrolysis and DNA unwinding of the human CMG complex in a purified biochemical system (76).