Phosphorylation of Mcm2 modulates Mcm2-7 activity and affects the cell's response to DNA damage.

The S-phase kinase, DDK controls DNA replication through phosphorylation of the replicative helicase, Mcm2-7. We show that phosphorylation of Mcm2 at S164 and S170 is not essential for viability. However, the relevance of Mcm2 phosphorylation is demonstrated by the sensitivity of a strain containing alanine at these positions (mcm2(AA)) to methyl methanesulfonate (MMS) and caffeine. Consistent with a role for Mcm2 phosphorylation in response to DNA damage, the mcm2(AA) strain accumulates more RPA foci than wild type. An allele with the phosphomimetic mutations S164E and S170E (mcm2(EE)) suppresses the MMS and caffeine sensitivity caused by deficiencies in DDK function. In vitro, phosphorylation of Mcm2 or Mcm2(EE) reduces the helicase activity of Mcm2-7 while increasing DNA binding. The reduced helicase activity likely results from the increased DNA binding since relaxing DNA binding with salt restores helicase activity. The finding that the ATP site mutant mcm2(K549R) has higher DNA binding and less ATPase than mcm2(EE), but like mcm2(AA) results in drug sensitivity, supports a model whereby a specific range of Mcm2-7 activity is required in response to MMS and caffeine. We propose that phosphorylation of Mcm2 fine-tunes the activity of Mcm2-7, which in turn modulates DNA replication in response to DNA damage.

INTRODUCTION

DNA replication is a tightly regulated process that occurs once and only once per cell division cycle. In addition, the cell must respond to impediments to replication, including DNA damage. One of the targets for regulation is the replicative helicase, comprised of minichromosome maintenance proteins 2 through 7 (Mcm2–7). Unlike other replicative helicases, which contain six identical subunits, Mcm2–7 is comprised of six distinct, but related subunits (1). The requirement for six subunits, with distinct sequences is thought to reflect individual roles for each subunit. Notably, the contribution of subunits to adenosine triphosphate (ATP) hydrolysis by the intact complex varies widely (2–4). For example, mutations that severely affect ATP hydrolysis by Mcm2 do not strongly affect DNA unwinding by Mcm2–7 or viability of Saccharomyces cerevisiae (5,6). We have proposed that ATP hydrolysis by Mcm2 regulates DNA binding by Mcm2–7 since mutations that decrease ATP hydrolysis by Mcm2 increase binding to single-stranded DNA (6).

Mcm2–7 is targeted by several different kinases including CK2, cyclin-dependent kinases (CDK) and Mec1 (7–12). Importantly, Mcm2–7 is the principal target of the essential S-phase kinase, DDK (12,13). DDK, which is comprised of Dbf4 and Cdc7 subunits, activates DNA unwinding by Mcm2–7 in vivo, thereby triggering origin firing throughout S phase (12,14,15). Several of the Mcm subunits are phosphorylated by DDK, including Mcm2 (11,13,16–26). Multiple roles for the phosphorylation of Mcm2 by DDK have been postulated (21,22,24,25). Here, we examined the effects of DDK phosphorylation of Mcm2 on the activity and function of Mcm2–7 in vitro and in vivo. Our results suggest that DDK phosphorylation of Mcm2 is not required for viability, but is important for the cell’s response to DNA damage, possibly through regulating DNA binding and/or DNA unwinding by Mcm2–7.

MATERIALS AND METHODS

Materials

Caffeine, hydroxyurea, methyl methanesulfonate (MMS), calcafluor white and ATP were from Sigma Aldrich (≥99 % purity), molecular biology enzymes from New England BioLabs and primers from Integrated DNA Technologies.

Plasmids

All DNA was amplified with Elongase polymerase (Invitrogen) and the primers mentioned in Supplementary Table S1. Plasmids are described in Supplementary Table S2. Mutated mcm2 fragments were generated using megaprimer polymerase chain reaction (PCR) (27) and inserted into the NdeI/BamHI sites of pET16b, pET24a and/or pMD240. pMD423 and pMD422 were generated by inserting the BsrGI–BamHI fragment of pET24a-mcm2 into the same sites of pMD388 and pMD328, respectively. For pMD367 and pMD371, SphI–BamHI fragments from pET16b-mcm2 and pET16b-mcm2 were inserted into YCplac211 (28). pMD408 is YCplac211 with the XbaI–BamHI fragment of pET16b-MCM2. pMD397 is the BglII–StuI fragment of pET16b-bob1 (a gift from M. O’Donnell, Rockefeller University) in YCplac211. pMD295 and pMD281 are CDC7 and DBF4 inserted into the NdeI/BamHI sites of pCDF-Duet (Novagen) and pET16b (Novagen), respectively. pMD449 contains RFA1, amplified from the yeast genome, fused in-frame to green fluorescent protein (GFP) in YCpDed-GFP (29) using NotI and SacI (blunted).

Yeast strains

Two-step gene replacement (30) was employed to insert mcm2 alleles into BY4743 (MATa/α his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 ura3Δ0/ura3Δ0 MET15/met15Δ0 LYS2/lys2Δ0; (31)). Briefly, pMD367 (mcm2) or pMD371 (mcm2) digested with BsrGI or PstI, respectively, was transformed into BY4743, then Ura+ transformants were streaked on media containing 5-FOA. Colonies from 5-FOA were screened by sequencing, generating MDY104 (MCM2/mcm2) and MDY106 (MCM2/mcm2). MDY139 and MDY159 are spore colonies derived from MDY104 or MDY106 containing mcm2 or mcm2, respectively. The URA3 gene was integrated downstream of mcm2 by transforming YCplac211 containing MCM2 (nucleotides 2023–2607; pMD408) digested with MscI into MDY139 and MDY159 to generate MDY169 and MDY191, respectively. A Ura+ strain wild type for MCM2 (BY4741) was similarly generated (MDY167). Two-step gene replacement was used to generate a cdc7Δ bob1 strain. BY23713 [MATa/α ura3Δ0/ura3Δ0 leu2Δ0/leu2Δ0 his3Δ1/his3Δ1 lys2Δ0/LYS2 met15Δ0/MET15 can1Δ::LEU2+-MFA1pr-HIS3/CAN1 CDC7/cdc7::kanMX; (32)] was transformed with YCplac211-bob1 linearized with MscI. A MATa cdc7Δ bob1 haploid (MDY210) was generated using ‘Magic Marker’ technology (32) and resistance to G418.

Proteins

Mcm2–7 containing PKA-tagged Mcm3 was reconstituted from purified subunits as previously described (3,6). Purified proteins and complexes are shown in Supplementary Figure S1. DDK was purified from cells co-expressing Cdc7 (pMD295) and 6xHis-tagged Dbf4 (pMD281). Lysates, prepared as described (3), were applied to a 20 ml Ni–Sepharose column equilibrated in buffer C (20 mM Tris–HCl pH7.9, 500 mM NaCl, 10% glycerol) with 5 mM imidazole and washed with the same buffer followed by buffer C with 30 mM imidazole. Bound proteins were eluted with a 200 ml, 30 mM–1 M imidazole gradient in buffer C. DDK-containing fractions were dialyzed against 20 mM Tris–HCl pH 7.5, 0.1 mM ethylenediaminetetraacetic acid (EDTA), 10% glycerol and 50 mM NaCl. After dialysis, DTT was added to a final concentration of 2 mM and the dialysate applied to a 1-ml Mono Q column equilibrated in buffer A (20 mM Tris–HCl pH 7.5, 0.1 mM EDTA, 2 mM DTT and 10% glycerol) containing 50 mM NaCl. After washing with the same buffer, retained proteins were eluted with 50–500 mM NaCl in buffer A. Dbf4 and Cdc7 in peak fractions were verified by western blotting with anti-Dbf4 and anti-Cdc7 antibodies (Santa Cruz Biotechnology) and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (London Regional Functional Proteomics Facility).

Fluorescence-activated cell sorting

Cells (7 × 106), fixed in 70% ethanol, were prepared for fluorescence-activated cell sorting (FACS) by incubation with 200 µg RNase A per milliliter for 3 h at 37°C followed by 2 mg Proteinase K per milliliter for 30 min at 30°C. After staining with Sytox Green (Invitrogen), cell sorting was performed on a FACSCalibur (BD Biosciences) by the London Regional Flow Cytometry Facility.

ATP hydrolysis and DNA unwinding assays

ATPase and helicase assays were performed as described (6), except that for DNA unwinding PEG 3350 was added to 5%. For reactions containing DDK, the indicated amounts of DDK were incubated with 200 nM Mcm2–7 and 5 mM ATP for 30 min at 30°C with 50 mM creatine phosphate and 2 µg/ml creatine phosphokinase and started by addition of oligonucleotide substrate. Statistical analyses were performed using analysis of variance (ANOVA) with Tukey’s Multiple Comparison test.

DNA binding

The gel filtration-based assay has been described previously (6). For DNA binding by electrophoretic mobility shift assays (EMSAs), 1 nM of 5′-end 32P-labeled poly-d(T60) was incubated with Mcm2–7 complex for 10 min at 37°C in 6 µl of reaction buffer. Glycerol was added to a final concentration of 5% (v/v) and the reaction then applied to a 5% native (Tris–borate–EDTA) polyacrylamide gel (19:1 acrylamide:Bis–acrylamide; BioShop Canada) containing 5% glycerol, 0.1% NP-40 and 10 mM Mg(CH3COO)2. The gel was resolved at 30 mA for 3 h, dried and exposed to a PhosphorStorage screen and imaged on a Storm 860 scanner (GE Healthcare).

RESULTS

Serines 164 and 170 of Mcm2 are not essential for normal growth

We mapped DDK’s target sites in S. cerevisiae Mcm2 to serines 164 and 170 using in vitro kinase assays on N-terminal deletions and point mutations of Mcm2 (Supplementary Figure S2A–D). S164 and S170 were identified as DDK sites by another group while this study was ongoing (21). We tested whether loss of the potential DDK phosphorylation sites in Mcm2 affected growth using plasmid shuffling. The strain MDY054, containing wild-type MCM2 on a URA3-plasmid, was transformed with a LEU2 plasmid encoding mcm2, mcm2, mcm2 (mcm2) or wild-type MCM2 under control of the MCM6 promoter. The resulting strains were plated on 5-FOA, which is metabolized to a toxic product by Ura3, thus testing whether the mcm2 allele encoded on the LEU2-containing plasmid supports viability. All of the alleles including mcm2 supported viability with growth being comparable to that of cells with wild-type MCM2 at 30°C and 37°C (Figure 1A). A slight reduction in growth was observed with mcm2 at 16°C, whereas strains with the single mutations grew comparably to strains with wild-type MCM2 (Figure 1A). Similar results were observed when mcm2 was integrated at the MCM2 locus (data not shown). We concluded that S164 and S170 are not essential for yeast viability under normal growth conditions. To further test the effects of mutating S164 and S170 on cell growth, we compared MCM2 and mcm2 strains by FACS. Comparison of asynchronously growing cultures indicated the same ratio of cells with 1C and 2C DNA content (Figure 1B). As shown in Figure 1C, the strains also had similar profiles after arrest in G1 with α factor mating pheromone followed by release. Together, these results indicate that there are no gross abnormalities in S phase nor DNA replication initiation when S164 and S170 are substituted with Ala.

Figure 1.

Strains containing mutations in the Mcm2 phosphorylation sites are viable. (A) The growth of haploid mcm2Δ strains bearing plasmid-encoded MCM2 (WT), mcm2, mcm2 or mcm2 was compared. Serial 10-fold dilutions of the strains were spotted on YPD plates and grown at the indicated temperatures. (B) FACS of MCM2 and mcm2 strains was performed on asynchronously growing cultures. (C) MCM2 and mcm2 cells were treated with α-factor mating pheromone for 3 h at 30°C, washed and resuspended in YPD. The DNA content of an aliquot of fixed cells was analyzed by FACS while arrested and at the indicated times after release.

In addition to its essential role in the initiation of DNA replication, DDK participates in the S-phase checkpoint response (12,33–39). Notably, mutations in S. cerevisiae or Schizosaccharomyces pombe Dbf4/Dfp1 result in sensitivity to drugs that induce the S-phase checkpoint, such as hydroxyurea (HU) and MMS (37,38,40,41). To determine whether phosphorylation of Mcm2 by DDK plays a role in the checkpoint response, we subjected cells supported for growth by a plasmid-encoded copy of mcm2 to HU or MMS. Strains containing mcm2 are more sensitive than wild-type MCM2 strains to a short-term exposure of MMS but not HU (Figure 2A). When the strain containing mcm2 was grown on YPD containing caffeine, its growth was also reduced (Figure 2B). Caffeine is a purine analog that inhibits phosphatidylinositol 3 kinase (PI3K)-like kinases, including those involved in the S-phase checkpoint (Tel1 and Mec1) and cell growth [TOR; (42–45)]. The doses of caffeine used here likely cause DNA damage (42). The sensitivity to caffeine was only observed with alanine mutations at both S164 and S170 (mcm2); single mutations did not affect growth (Figure 2B), suggesting that, at least with respect to caffeine sensitivity, S164 and S170 are functionally redundant. Caffeine also affects the cell wall integrity pathway (46); however, the mcm2 strain was not sensitive to calcofluor white (Figure 2C) indicating that phosphorylation at S164 and S170 is not required for the cell wall integrity pathway.

Figure 2.

Sensitivity of mcm2 to genotoxic agents. (A) The survival of MCM2 and mcm2 strains after a 4-h exposure to the indicated concentration of HU or MMS relative to untreated cells was measured as described (37). The assay was performed in triplicate and plotted as mean percent survival ± standard error of the mean (SEM). (B) Serial 10-fold dilutions of strains containing wild-type MCM2, mcm2, mcm2 or mcm2 were spotted on YPD plates containing 10, 15 or 20 mM caffeine and grown at 30°C. (C) Serial 10-fold dilutions of MCM2 and mcm2 were spotted on YPD and YPD containing 10 µg/ml calcofluor white (‘CW’) before incubation at 30°C.

Phosphomimetic mutations at S164 and S170

We asked whether substitution of glutamic acid for serine at positions 164 and/or 170 provided a phosphomimetic effect by examining the drug sensitivity of cells bearing these mutations. Strains with the single (mcm2) or the double mutant (mcm2) alleles on LEU2 plasmids grew to the same extent as cells containing wild-type MCM2 on YPD media at all temperatures tested (Figure 3A). Importantly, strains bearing the Glu substitutions at S164 and/or S170 were not sensitive to caffeine, even at the highest concentration of caffeine supplied (20 mM; Figure 3B). mcm2 cells were also resistant to MMS (Figure 3C). Thus, we concluded that the Glu substitutions at S164 and S170 of Mcm2 were phosphomimetic.

Figure 3.

Mcm2 phosphomimetic mutants. (A) The growth of haploid mcm2Δ strains bearing plasmids encoding MCM2, mcm2, mcm2 or mcm2 was compared. Serial 10-fold dilutions were spotted on YPD plates and grown for 3–5 days at 16, 30 or 37°C. (B) Serial 10-fold dilutions of the indicated strains were spotted on YPD media containing 10, 15 or 20 mM caffeine and grown at 30°C. (C) The survival of MCM2 and mcm2EEstrains after a 4 hour exposure to the indicated concentrations of MMS relative to untreated cells was measured as described (37). (D) Ten-fold serial dilutions of haploid cdc7Δ bob1 strains with wild type MCM2 or mcm2 were spotted on YPD with and without HU (100 mM) or caffeine (10 mM) and incubated at 30°C. (E) A strain deleted at cdc7 and dbf4 and supported for growth with human CDC7 and DBF4 was transformed with plasmids encoding MCM2 or mcm2. Ten-fold serial dilutions of the strains were spotted on YPD with or without 150 mM HU or 15 mM caffeine and incubated at 30°C for 2–6 days.

We next addressed whether the putative phosphorylation of S164 and S170 is the result of DDK activity in vivo. CDC7 is essential, but viability of a cdc7Δ strain can be restored by the bob1 mutation in mcm5 (47). The cdc7Δ bob1 strain is sensitive to HU (47,48). If Mcm2 is the key target of DDK leading to the HU sensitivity of the cdc7Δ bob1 strain, then the introduction of mcm2 should suppress the HU sensitivity. As seen in Figure 3D, mcm2 does suppresses the HU sensitivity as well as the caffeine sensitivity of the cdc7Δ bob1 strain. The difference in growth is dependent on the presence of the HU or caffeine since cdc7Δ bob1 grows similarly on YPD, regardless of whether the mcm2 allele is wild type or phosphomimic (Figure 3D). This result supports the idea that DDK phosphorylates Mcm2 at S164 and S170; however, it is possible that interaction between mcm5-bob1 and mcm2 accounts for the suppression of HU and caffeine sensitivity since MCM2 and MCM5 interact genetically (49). A recent study in our lab (Davey,M.J., Andrighetti,H., Ma,X., and Brandl,C.J., manuscript in preparation) has shown that human DDK supports the essential functions of DDK in yeast, but that yeast with human DDK are sensitive to HU and caffeine (Figure 3E). Of particular note, mcm2 suppresses the sensitivity to HU and caffeine of cells with human DDK (Figure 3E). The simplest explanation for these observations is that, unlike yeast DDK, human DDK fails to efficiently phosphorylate Mcm2 causing sensitivity to genotoxic agents, and that this defect is rescued by the phosphomimetic nature of Mcm2EE. Together, these findings implicate phosphorylation of Mcm2 at S164 and S170 by DDK in response to DNA damage and support the in vitro mapping of DDK sites in Mcm2.

mcm2 results in accumulation of Rpa1 foci

In response to treatment with MMS or other replicative stresses, RPA relocalizes from diffuse staining of the nucleus into discrete foci (50,51). Since RPA is a single-stranded DNA (ssDNA)-binding protein, the RPA foci indicate the accumulation of ssDNA. In normal cells responding to DNA damage, ssDNA is likely generated by repair processes (50,52). Using strains containing Rpa1 tagged with green fluorescent protein (GFP), we compared the RPA foci in mcm2, mcm2 and wild-type strains before and after exposure to MMS. For comparison we also examined a mec2-1 strain that is deficient in the S phase checkpoint (53). In untreated MCM2 cells, GFP-Rpa1 localized to the nucleus with a diffuse signal that transitions into discrete foci after exposure to MMS in 14.9 ± 4.0% of cells (Figure 4). With mcm2, significantly more cells contained foci after MMS treatment (44.3 ± 3.9%), and like the mec2-1 strain foci were seen without MMS treatment. The mcm2 strain had a similar percentage of cells with foci as MCM2 (11.9 ± 1.0%). These results are consistent with an inability of the mcm2 strain to appropriately modulate DNA replication in response to DNA damage.

Figure 4.

Fluorescence microscopy of MCM2, mcm2 and mcm2 cells expressing GFP-tagged Rpa1. (A) MCM2, mcm2 and mec2-1 (53) strains were transformed with a plasmid expressing GFP-Rpa1. Representative bright field (‘BF’) and fluorescent (‘GFP’) images of cells from cultures grown at 30°C in YPD or YPD containing 0.03% MMS for 4 h are shown. The white arrowheads indicate foci. Images were taken on a Nikon Ti microscope at 400X. (B) The mean percent of cells with RPA foci was calculated by counting >100 cells in three replicate experiments. Strains in which the numbers of cells with RPA foci are significantly different (calculated by ANOVA with Tukey’s Multiple Comparison Test) from the MCM2 strain without MMS (*P < 0.05) and with MMS (**P < 0.05) are indicated.

Biochemical activities of the phosphomimetic Mcm2-7 complex

To determine mechanisms by which DDK phosphorylation of Mcm2 exerts its control, we examined the effect of phosphorylation of S164 and S170 on Mcm2 activity in vitro through use of the phosphomimetic mutations in Mcm2. We reconstituted Mcm2–7 with Mcm2 containing glutamic acid residue substitutions at S164 and S170 (Mcm2EE). Note that Mcm proteins are purified from Escherichia coli expression strains and thus not phosphorylated. The final step in the reconstitution is a gel filtration column that separates excess free subunits from the complex. Additionally, the gel filtration column allows us to verify that Mcm2WT–7 and Mcm2EE–7 elute as hexamers (Figure 5A). The peak elution for both complexes occurred around fraction 23, corresponding to a size of ∼610 kDa, close to the predicted size of 608 kDa. Analysis of the peak fractions by quantitative western blotting and densitometry of the Coomassie Blue R250-stained gels indicated equal ratios of Mcm2–7 (within ∼20%).

Figure 5.

Reconstitution of Mcm2WT-7 and Mcm2EE–7 complexes. (A) Mcm2–7 complexes containing wild-type Mcm2 (‘Mcm2WT–7′) or Mcm2EE (‘Mcm2EE–7′) were reconstituted from individual subunits. Shown are Coomassie Blue-stained sodium dodecyl sulfate (SDS) polyacrylamide (6%) gels of the fractions from the final gel filtration step. The elution of molecular size standards from the column is indicated above the gels. The migration of molecular size markers and of Mcms through the gels is indicated on the left and right, respectively. (B) DNA unwinding by Mcm2WT–7 and Mcm2EE–7 on synthetic fork substrates was examined. The peak fraction from each reconstitution was assayed for DNA unwinding. ‘Unwound’ is a control that indicates the extent of reannealing of unwound substrate. The migrations of double-stranded substrate and single-stranded product are indicated on the right. (C) The mean extent of DNA unwinding ± SEM by the indicated amount of each complex was calculated from three replicate experiments. (D) DNA unwinding by Mcm2WT–7 (200 nM) after treatment with DDK is shown. The extent of unwinding is indicated below the gel.

We next measured DNA unwinding by Mcm2WT–7 and Mcm2EE–7 on short oligonucleotide forks whereby DNA unwinding is measured as displacement of a labeled oligonucleotide from its complementary strand. We detected robust unwinding by unphosphorylated Mcm2WT–7 (Figure 5B and C). In contrast, DNA unwinding by Mcm2EE–7 was reduced to about 40–50% of wild-type levels (Figure 5B and C). This suggests that DDK-dependent modification of Mcm2 at S164 and S170 inhibits DNA unwinding by Mcm2–7. Consistent with this idea, treatment of wild-type Mcm2–7 with DDK and ATP under conditions in which Mcm2 is highly phosphorylated and the principal substrate for DDK (Supplementary Figure S2E and F), also decreased DNA unwinding (Figure 5D).

Models for DNA unwinding by hexameric helicases include binding to single- and/or double-stranded DNA followed by translocation (54–61). Thus, we examined DNA binding by Mcm2–7 containing different versions of Mcm2. We first used a gel filtration-based approach (6), in which binding is measured as the amount of Mcm2–7 that co-elutes with ssDNA. Radiolabeled Mcm2–7 or Mcm2EE–7 was incubated with increasing concentrations of M13mp19 circular ssDNA and bound protein separated from free protein via gel filtration. Using this assay we found that Mcm2EE–7 binds DNA with higher affinity than Mcm2–7 (Figure 6A). At the lowest concentration of ssDNA supplied (0.5 nM) there was a >3-fold increase of Mcm2EE–7 bound to DNA compared to Mcm2WT–7. The increased DNA binding with Mcm2EE–7 was similar to Mcm2–7 complexes containing ATP site mutations in Mcm2, including Mcm2K549R in which the conserved P-loop lysine is replaced with an arginine residue [Figure 6A and (6)]. These initial experiments indicated that Mcm2–7 complexes containing Mcm2EE or Mcm2K549R have higher affinity for ssDNA than complexes containing wild-type Mcm2. To confirm that Mcm2EE mimics the effects of phosphorylation, we treated Mcm2WT–7 with DDK and ATP before measuring DNA binding. An increase of ∼2.5–5-fold was observed when Mcm2–7 was phosphorylated by DDK compared to mock-treated (no DDK) Mcm2–7 (Figure 6B).

Figure 6.

ssDNA binding by Mcm2–7. (A) DNA binding was measured using a gel filtration-based assay (6). Binding of Mcm2WT–7 (open triangle), Mcm2EE–7 (filled square) and Mcm2KR–7 (open circle) was determined in triplicate experiments and the mean plotted. Representative elution profiles for each point are shown in Supplementary Figure S3. (B) Mcm2–7 (200 fmol) was incubated in assay buffer with 100 µM ATP with or without DDK for 30 min at 30°C before addition of 50 fmol ssDNA and ATP to 5 mM. After incubation for 10 min at 37°C, the samples were analyzed by gel filtration. The mean amount of Mcm2–7 that co-eluted with DNA in triplicate experiments was determined. (C) A Phosphor screen image of a representative EMSA. The migration of free DNA and protein–DNA complex is indicated on the left. A portion of a gel containing free protein and protein–DNA complex was removed and stained with GelCode Blue (Pierce) and is shown on the right. (D) The mean amount of DNA shifted was calculated and plotted with SEM from three replicate experiments.

DNA binding was also measured by EMSAs, which allow a wider range of concentrations. Increasing amounts of Mcm2WT–7, Mcm2EE–7 or Mcm2K549R–7 were incubated with 1 nM poly-dT60. Binding of Mcm2–7 shifted the DNA to a point near the top of the gel that coincided with the migration of Mcm2–7 as visualized by staining (Figure 6C and D) or western blotting (data not shown). Mcm2WT–7 shifted up to 30 fmol of the ssDNA supplied. This binding is similar to previous reports using this approach (62) and approximates the affinity observed by gel filtration. In contrast, Mcm2EE–7 shifted 8- to 10-fold more poly-dT60 (Figure 6C and D). Mcm2K549R–7 shifted even more ssDNA at levels 15- to 18-fold over Mcm2-7 (Figure 6C and D). Overall, these results demonstrate that the phosphomimetic mutations in Mcm2 enhance ssDNA binding by Mcm2–7.

At first glance, our data seem contradictory. Mcm2–7 containing the Mcm2EE mutant binds DNA better than wild type, but unwinds DNA less well, an activity that requires DNA binding. However, we reasoned that Mcm2EE–7 may bind DNA too tightly, impairing the ability of Mcm2–7 to translocate along DNA. To test this, we examined whether adding NaCl to the DNA-binding assay would reduce the Mcm2EE–7–DNA interaction. Indeed, when NaCl was present at 100 mM, Mcm2EE–7 binding to DNA decreased to a level observed with Mcm2WT–7 (Figure 7A). Moreover, when NaCl was added to Mcm2EE–7 in a DNA-unwinding assay, the amount of unwound substrate increased to a level comparable to Mcm2WT–7 (Figure 7B). These observations suggested that the defect in DNA unwinding by Mcm2EE–7 results from enhanced DNA binding.

Figure 7.

Disruption of DNA binding rescues the DNA unwinding defect of Mcm2EE–7. (A) DNA binding by the indicated concentrations of Mcm2WT–7 (black circle) and Mcm2EE–7 (black square) in the presence of 100 mM NaCl was measured by EMSA and plotted. DNA binding by Mcm2WT–7 (white circle) and Mcm2EE–7 (white square) in the absence of NaCl is shown for comparison. (B) DNA unwinding by the indicated concentrations of Mcm2WT–7 (black circle) and Mcm2EE–7 (black square) in the presence of 100 mM NaCl was measured. DNA unwinding by Mcm2WT–7 (white circle) and Mcm2EE–7 (white square) in the absence of NaCl is shown for comparison.

Since the mcm2 mutation decreases ATP hydrolysis by Mcm2 (6) and like Mcm2EE increases binding to ssDNA by Mcm2–7, we examined whether phosphorylation of Mcm2 affects ATP hydrolysis by Mcm2. The ATP sites of Mcm2–7 are found at subunit interfaces and residues from two subunits are required to form a functional ATP site [Figure 8A; (3,4)]. Accordingly, ATP hydrolysis by Mcm2 is dependent on Mcm6 (3,4). The Mcm2/6 ATP site makes only a small contribution to the total ATP hydrolysis by Mcm2–7 (2) and even a large change in Mcm2 activity has only a modest effect on the intact complex. Thus, we initially measured ATP hydrolysis by the Mcm2/6 pair rather than the intact Mcm2–7 complex. Wild-type, unphosphorylated Mcm2 with Mcm6 hydrolyzed ATP at a rate of 4.9 ± 0.1 min−1 (Figure 8B). When Mcm2EE was mixed with Mcm6, ATP hydrolysis was reduced to ∼75% of wild-type Mcm2/6 (3.6 ± 0.6 min−1). For comparison, ATP hydrolysis by Mcm2K549R was 0.9 ± 0.0 min−1. Mcm2AA had activity similar to wild-type Mcm2 (4.7 ± 0.1 min−1) as would be expected since the proteins are purified from E. coli and hence not phosphorylated (Figure 8B). We also measured the effects of the Mcm2EE mutation on the ATPase activity of the intact complex. There was no significant difference in the rates of Mcm2EE–7 (18.0 ± 0.2 min−1, P > 0.05, n = 3) and Mcm2WT–7 (21.0 ± 1.6 min−1; Figure 8C). For comparison, we also show ATP hydrolysis by Mcm2KR–7 (17.4 ± 0.4 min−1; Figure 8C). The lack of significant difference is not surprising considering the small contribution of Mcm2/6 to the ATPase of intact Mcm2–7.

Figure 8.

Phosphomimetic mutations in Mcm2 reduce ATP hydrolysis. (A) A schematic showing the Mcm2/6 subunit interface. The phosphate binding loop (P-loop) of Mcm2 and the SRF motif of Mcm6, required for ATP hydrolysis by Mcm2 are indicated. K549 is located in the P-loop. (B) ATP hydrolysis by recombinantly expressed Mcm2, Mcm2EE, Mcm2AA and Mcm2K549R in the presence of Mcm6 was measured. The rate of ATP hydrolysis by each pair was calculated and plotted with the standard error of the mean. The rate of hydrolysis by each complex, except between Mcm2/6 and Mcm2AA/6, is significantly different (P < 0.01; n = 3). (C) ATP hydrolysis by Mcm2–7, Mcm2EE–7 and Mcm2KR–7 was measured in triplicate at 1 mM ATP, the rate calculated and plotted with SEM. (D) Haploid strains deleted at mcm2 and containing plasmid-encoded wild-type MCM2, mcm2, mcm2, mcm2 or mcm2 were serially diluted 10-fold and then spotted on YPD with the indicated concentrations of caffeine and grown at 30°C.

Mcm2 phosphorylation affects Mcm2–7 activity in vivo

As shown in Figure 8D, cells bearing mcm2 are sensitive to caffeine similar to mcm2 even though some of the biochemical properties of Mcm2K549R are more similar to Mcm2EE, which does not lead to caffeine sensitivity. To examine the relationship between the changes observed in vitro and the sensitivity to caffeine, we made mcm2 alleles mutated at both the ATP site (K549R) and the DDK phosphorylation sites (EE or AA). Cells containing mcm2 and mcm2 as the sole copy of mcm2 grew similarly to wild-type cells on YPD (Figure 8D). Interestingly, when spotted onto YPD containing caffeine, the growth of the mcm2 strain was markedly reduced relative to both wild-type cells and cells containing the individual mutations (Figure 8D). In contrast, strains with a non-phosphorylatable Mcm2 containing the ATP site mutation (mcm2) were not sensitive to caffeine. These results suggest that a precise range of Mcm2–7 activity is required for cells to respond to DNA damage.

DISCUSSION

Phosphorylation of Mcm2 and cell viability

Here, we show that S. cerevisiae containing Mcm2 that lacks the phosphorylation sites, S164 and/or S170, are viable and progress normally through S phase when Mcm2 is expressed at endogenous levels. Our results are consistent with the recent finding that normal growth is unaffected by deletion of the N-terminal region of Mcm2 (12) or mutation of several phosphorylation sites to alanine (11). Furthermore, it has been demonstrated that phosphorylation of Mcm4 by DDK is sufficient for replication in budding yeast (11,12). These results contrast with the finding that Mcm2S170A expressed from the GAL10 promoter did not support viability (21). We have replicated this latter result, indicating that the cellular effect of Mcm2S170A is concentration dependent (Supplementary Figure S4). It may be that overexpression of Mcm2 causes genomic instability that is dealt with in normally functioning cells, but requires phosphorylation of Mcm2 at S170.

Phosphorylation of Mcm2 in vivo

We propose that phosphorylation of Mcm2 at S164 and S170 is important in response to DNA damage. The sensitivity of the mcm2 strain to caffeine and MMS, the accumulation of RPA foci in this strain and the reversal of these phenotypes by mcm2 strongly suggest a role in the response to DNA damage. MMS is a well-defined base alkylating agent that induces specific cellular responses including S phase checkpoint (63). Caffeine is a purine analog that inhibits kinases of the PI3K-like family, which include the checkpoint kinases, Tel1 and Mec1 (43,45). Inhibition of the checkpoint kinases by caffeine can lead to accumulation of DNA damage (42). Thus, the caffeine sensitive phenotype of cells with mcm2 may not occur in response to inhibition of PI3K-like kinases per se but rather in response to DNA damage generated by high doses of caffeine.

We have mapped S164 and S170 in vitro as potential DDK sites, consistent with a previous study (21). Two lines of evidence support the idea that these are DDK phosphorylation sites in vivo. First, mcm2 suppresses the HU and caffeine sensitivity of a cdc7Δ bob1 strain. Second, the HU and caffeine sensitivity of a strain containing human DDK is also suppressed by mcm2 Furthermore, Mcm2 is already known to be a substrate for DDK in yeast (17) and DDK is known to have roles in responding to DNA damage (26,34,38,40,41,64)

We note that a recent study from the Bell lab indicates that at least a portion of Mcm2 is phosphorylated at S164 and S170 in a CK2-dependent manner in G1 arrested cells (11). Interestingly, DDK and CK2 share similar specificity, targeting acidic sites (65,66). Our work strongly implicates DDK as the kinase that phosphorylates Mcm2 in the response to DNA damage; however, we cannot eliminate the possibility that CK2 has a role. In yeast, the regulatory subunits of CK2 (Ckb1 and Ckb2) are required for recovery from checkpoint and adaptation to promote interaction between a phosphatase (Ptc2) and the checkpoint kinase, Rad53 (67).

Regulation of Mcm2–7 activities by phosphorylation of Mcm2

We propose that phosphorylation of Mcm2 at S164 and/or S170 controls Mcm2-7 in vivo by altering its DNA binding and/or helicase activities. Phosphorylation at S164 and S170 of Mcm2, which increases DNA binding by Mcm2-7 in vitro, may stabilize binding of Mcm2-7 to chromatin when cells respond to DNA damage. Mcm2–7 is more abundant on chromatin than is expected for one complex at each replication fork (1). In mammalian and Xenopus cells, the ‘excess’ Mcm2–7 is not required for normal cell growth but is important when cells are placed under replicative stress (68–70). Alternatively, the decreased DNA unwinding elicited by phosphorylation of Mcm2 may be required in response to DNA damage to slow replication forks and allow time for repair of damaged DNA.

It is possible that the effects on DNA binding and unwinding are mediated through the Mcm2 ATP site since the ATP hydrolysis by the Mcm2/6 site is decreased with the phosphomimetic mutation in Mcm2. Consistent with this idea, the effect of Mcm2EE on DNA binding by Mcm2–7 is similar to what is observed for a mutation that affects ATP hydrolysis by the Mcm2/6 site, Mcm2K549R. Additional studies will be required to determine the exact relationship between ATP hydrolysis by Mcm2/6 and the different phenotypes of MCM2, mcm2, mcm2 and mcm2. In this regard, we cannot exclude that it is the differences in DNA unwinding or an as yet unexamined activity of Mcm2-7 that accounts for the different phenotypes.

Regardless of which activity results in the caffeine sensitivity, relatively small changes detected in vitro result in significant changes in sensitivity to caffeine in vivo. This level of sensitivity to modest changes detected in vitro is not without precedence. An ATP site mutation in the P1 plasmid partition protein, ParA, results in a decrease to about 1/3 of wild-type activity, but has a >20-fold effect on the autorepressor activity of ParA in vivo (71).

SUPPLEMENTARY DATA

Supplementary Data are available at NAR online.

FUNDING

The Canadian Institutes of Health Research (MOP 68926 to M.J.D.); a Natural Science and Engineering Research Council Doctoral Award and Schulich Graduate Awards (to B.E.S.). M.J.D. is a Research Scientist of the Canadian Cancer Society Research Institute. Funding for open access charge: Canadian Institutes of Health Research/University of Western Ontario.

Conflict of interest statement. None declared.