Mcm proteins are an important family of evolutionarily conserved helicases required for DNA replication in eukaryotes. The eukaryotic Mcm complex consists of six paralogs that form a heterohexameric ring. Because the intact Mcm2-7 hexamer is inactive in vitro, it has been difficult to determine the precise function of the different subunits. The solved atomic structure of an archaeal minichromosome maintenance (MCM) homolog provides insight into the function of eukaryotic Mcm proteins. The N-terminal positively charged central channel in the archaeal molecule consists of β-hairpin domains essential for DNA binding in vitro. Eukaryotic Mcm proteins also have β-hairpin domains, but their function is unknown. With the archaeal atomic structure as a guide, yeast molecular genetics was used to query the function of the β-hairpin domains in vivo. A yeast mcm5 mutant with β-hairpin mutations displays defects in the G1/S transition of the cell cycle, the initiation phase of DNA replication, and in the binding of the entire Mcm2-7 complex to replication origins. A similar mcm4 mutation is synthetically lethal with the mcm5 mutation. Therefore, in addition to its known regulatory role, Mcm5 protein has a positive role in origin binding, which requires coordination by all six Mcm2-7 subunits in the hexamer.
THROUGHOUT evolution, eukaryotic organisms have developed sophisticated mechanisms to tightly regulate the duplication of their chromosomes. Proper duplication of genetic material requires that organisms replicate their DNA only once per cell cycle (Bell and Dutta 2002; Sclafani and Holzen 2007). These events must be coordinated precisely to prevent mutations that may lead to genomic instability, cancer, or cell death. At the heart of this process is the regulation of DNA replication, which can be divided into three basic steps: (i) assembly of the prereplication complex (pre-RC), which accumulates at replication origins; (ii) melting of these origins by helicases required for replication initiation; and (iii) elongation, which occurs during S phase (Bell and Dutta 2002; Sclafani and Holzen 2007).
To initiate DNA replication, several proteins must be recruited to replication origins in a controlled fashion. Orc1-6 proteins [origin recognition complex (ORC)] are bound constitutively to origins in Saccharomyces cerevisiae (Bell and Dutta 2002; Sclafani and Holzen 2007), and together, act as “landing pads” for all other required DNA replication proteins to bind chromatin and activate origins in G1 phase. Specifically, the ORC recruits Cdc6p and Cdt1p, both of which are responsible for loading the Mcm2-7p complex, which most likely acts as the replicative helicase (Tye 1999; Forsburg 2004; Lei 2005). Cdt1p protein binds the Mcm2-7p complex and loads it onto Cdc6p that is already bound to the ORC (Randell et al. 2006; Speck and Stillman 2007). The activities of the Dbf4-dependent kinase (DDK) and the cyclin-dependent kinase (CDK) (Tanaka et al. 2007; Zegerman and Diffley 2007) are then required to activate Cdc45p, which in concert with Sld2p, Sld3p, and the GINS complex loads polymerases onto the pre-RC (Bell and Dutta 2002; Sclafani and Holzen 2007). Once activation of the Mcm2-7p complex has taken place, DNA replication ensues.
The eukaryotic Mcm2-7p complex consists of six essential proteins that form a heterohexameric ring. Subcomplexes of Mcm4/6/7p contain ATPase, helicase, and DNA binding activities (Tye 1999; Forsburg 2004). However, the intact Mcm2-7p complex, or any subcomplexes containing Mcm5p are inactive for DNA binding and helicase activities in vitro (Ishimi et al. 1998; Lee and Hurwitz 2000; Schwacha and Bell 2001), although weak dsDNA binding has been detected recently with recombinant yeast Mcm2-7 complexes in vitro (Bochman and Schwacha 2007). The implication from these in vitro studies is that Mcm4/6/7 complexes are catalytic and Mcm2/3/5 complexes are regulatory. The Mcm protein complex is thought to form a double hexamer, a common architecture for many eukaryotic helicases (Tye 1999; Forsburg 2004). Methanobacterium thermoautotrophicum (MtMcm) represents a simpler system for studying Mcm proteins, in that it has only a single Mcm protein with ATPase and helicase activities (Tye 1999; Forsburg 2004). The N-terminal portion of MtMcm forms a dumbbell-like, double hexamer, which is homohexameric (Fletcher et al. 2003). The MtMcm structure has a long, positively charged channel running through the center of the molecule (Figure 1A; Fletcher et al. 2003). In each monomer, basic residues are found between β-sheets 9 and 10, which forms a β-hairpin with an overall positive charge at the tip in this central channel (Figure 1). Mutation of these basic residues in MtMcm reduces DNA binding activity in vitro (Fletcher et al. 2003). Because yeast Mcm5p is similar to MtMcm in this β-hairpin region (Fletcher et al. 2003) and the role of Mcm5p in DNA binding is unknown as it inhibits DNA binding in vitro (Ishimi et al. 1998; Lee and Hurwitz 2000; Schwacha and Bell 2001), a molecular genetic analysis of the importance of these basic residues in Mcm5p was performed. Strains with mutations in these basic residues have severe defects in the initiation of DNA replication and display a dramatic decrease in the binding of the Mcm complex to replication origins in vivo. These data demonstrate a clear functional conservation of β-hairpin domains in Mcm proteins from different species and suggest that the Mcm2-7 protein complex likely requires coordinated DNA binding from all six members of the heterohexamer.
MATERIALS AND METHODS
Yeast strains, media, and plasmids:
All S. cerevisiae strains used in this study are listed in Table 1. Yeast strains were grown as described previously (Sclafani et al. 1988). To produce mcm5-TP∷HA, we made mutations using the overlap PCR method used previously for mcm5 (Aiyar et al. 1996; Fletcher et al. 2003). Mutations were marked by addition or deletion of a DNA restriction site and all plasmids were verified by PCR followed by restriction digest and DNA sequencing. To produce mcm5-TP, plasmid pRS414-MCM5 (Hardy et al. 1997) was used as a template using the following primers in overlap PCR reactions, where lowercase boldface letters represent the nucleotide change. For K304A: (K304A-fwd) CTATAATTCTgccAATGGgGCCGGA and (pRS414-MCM5-rev) CACTATAGGGCG AATTGGGT, (K304A-rev) TCCGGCgCCATTggcAGAATTATAG and (pRS414-MCM5-fwd) CACTATAGGGCGAATTGGGT (adds BglI site). PCR reactions yielded a 2.0-kb fragment, which was cut with XhoI/PstI and cloned into the XhoI/PstI site of pRS414-MCM5 yielding pRS414-mcm5-K304A. For R311A, pRS414-mcm5-K304A was used as a template with the following primers in overlap PCR reactions: (R311A-fwd) CGGATCaGGAgcGAGCGGGGGTG and (pRS414-MCM5-rev) CACTATAGGG CGAATTGGGT, (R311A-rev) CACCCCCGCTCgcTCCtGATCCG and (pRS414-MCM5-fwd) CACTATAGGGCGAATTGGGT (loses BamHI site). PCR yielded a 2.0-kb fragment, which was cut with XhoI/PstI and cloned into the XhoI/PstI site of pRS414-mcm5-K304A yielding pRS414-mcm5-K304A-R311A. For R324A, pRS414-mcm5-K304A-R311A was used as a template with the following primers in overlap PCR reactions: (R324A-fwd) AGTGGTGTTGCaATTgcAACACCTT AT and (pRS414-MCM5-rev) CACTATAGGGCGAATTGGGT, (R324A-rev) ATAAGGTGTTgcAATtG CAACACCACT and (pRS414-MCM5-fwd) CACTATAGGGCGAAT TGGGT (gains MfeI site). PCR reactions yielded a 2.0-kb fragment, which was cut with XhoI/PstI and cloned into the XhoI/PstI site of pRS414-mcm5-K304A-R311A, yielding pRS414-mcm5-K304A-R311A-R324A, the mcm5 triple mutant plasmid (mcm5-TP). To generate the pRS304 integrating versions of these same plasmids, we cloned the 5.3-kb NotI/XhoI fragment from the pRS414-CDC46 plasmids to the NotI/XhoI site of pRS304.
To generate MCM5∷HA and mcm5-TP∷HA plasmids, we performed a PCR reaction on genomic DNA from strain 908 (MCM5∷HA) generating a 1.6-kb fragment with the following primers: (MCM5-fwd) CACCACTTCCTCCATTTCCACC and (MCM5-rev) CCCCAG ATTTAGTGAATAAGAGCCC. The 1.6-kb genomic fragment was cut with NruI/BclI and cloned into the NruI/BclI site of pRS306-MCM5 yielding pRS306-MCM5∷HA. The pRS306-MCM5∷HA plasmid was then cut with BstAPI/NotI, yielding a 5.8-kb fragment, which was ligated to a 4.1-kb BstAPI/NotI fragment cut from either pRS304-MCM5 or pRS304-mcm5-TP, yielding pRS306-MCM5∷HA and pRS306-mcm5-TP∷HA, respectively.
To generate the MCM4 β-hairpin mutations, the same overlap PCR method was used as for MCM5. MCM4 mutations were marked by the addition or deletion of a DNA restriction site and all plasmids were subsequently verified by PCR followed by restriction digest and DNA sequencing. To produce mcm4-4A, plasmid pRS316-Mcm4 was used as a PCR template for using the following primers in overlap PCR reactions. Lowercase boldface letters represent the nucleotide change. For R445A, (R445A fwd) ATCCCCATTgcAGCgAATTCC and (R445A rev) GGAATTgCGTgcAATGGGGAT (adds an EcoRI site). For K454A, (K454A fwd) CGCGTgCTAgcGTCGTTGTAT and (K454A rev) ATACAACGACgcTAGcACGCG (adds NheI site). For K458A, (K458A fwd) TCGTTGTATgcAACATAtGTC and (K458A rev) GACaTATGTTgcATACA ACGA (adds NheI site). For H465A, (H456A fwd) GATGTGGTggcCGTTAAAAAA and TTTTTTAACGgccACCACATC (adds HaeIII site). The following outside primers were used in all overlap PCR reactions in combination with the above mutagenesis primers: (pRS316-Mcm4 fwd) AGTCAGGGAGAGGGAAACATCAG) and (pRS316-Mcm4 rev) GCAATAGAGCG GGCTAATAAACTG. Final overlap reactions yielded a 1414-bp fragment that was sequentially digested with restriction enzymes NruI and AfeI, and gel purified with the QIAGEN gel purification kit. The final PCR fragment (1232 bp) was ligated into the NruI/AfeI site of pRPL106 (pRS316-MCM4), yielding the final plasmid pRPL107 (pRS316-mcm4-4A). To make the pRS305 LEU2 versions of either MCM4 or mcm4-4A, a 4.7-kb SacI/HindIII fragment from either pRPL106 or pRPL107 was inserted into SacI/HindIII of the vector pRS305, yielding pRAS693 (pMCM4 LEU2) or pRAS691 (pmcm4-4A LEU2) or pRS315, yielding pRAS662 (pMCM4 LEU2) or pRAS685 (pmcm4-4A LEU2).
To produce the mcm4∷hisG-URA3-hisG knockout cassette, the BamHI/BglI fragment from plasmid pNKY51 (Alani et al. 1987) and the hisG-URA3-hisG were inserted into the compatible BclI site of pRAS662, yielding pRAS668. To knock out the MCM4 gene, strain YRL214 was transformed with pRS662 (pMCM4 LEU2), yielding strain RSY1214. RSY1214 was transformed with a SalI/SacI restriction fragment containing the mcm4∷hisG-URA3-hisG disruption from plasmid pRS668. Ura+ transformants were selected on −Ura media, then passed through 5-FOA, yielding strain RSY1220 mcm4∷hisG (pMCM4 LEU2). RSY1220 was then transformed with pRS316-MCM4 URA3 and selected on −Ura media. These URA3 colonies were then screened for the loss of LEU2 on nonselective YEPD medium, yielding strain RSY1225 mcm4∷hisG (pMCM4 URA3).
To make the mcm5-TP mcm4-4A double mutant strains, strain RSY1225 was crossed with RSY1238 to yield strain RSY1240 mcm5Δ KanMX4 trp1∷TRP1 mcm5-TP mcm4∷hisG (pMCM4 URA3) and strain RSY1241 mcm5Δ KanMX4 trp1∷TRP1 MCM5+ mcm4∷hisG (pMCM4 URA3). The presence of the mcm5-TP allele was followed by PCR and restriction digests. Strains RSY1240 and RSY1241 were then transformed with either plasmids pRAS691 (LEU2 mcm4-4A) or pRS693 (LEU2 MCM4), which were digested with HpaI to target integration to the leu2-3, 112 locus. These four strains were subjected to 5-FOA to select for loss of the pMCM4 URA3 plasmid. Only the MCM5 MCM4 (strain RSY1265), MCM5 mcm4-4A (strain RSY1266), and mcm5-TP MCM4 (strain RSY1264) combinations were found as the mcm4-4A mcm5-TP double mutant is inviable (synthetic lethality).
Mcm5 and Mcm4 protein structural alignments and predictions:
Primary sequences were aligned using the CLUSTALW (ver. 1.81) program (http://align.genome.jp/sit-bin/clustalw) as shown in Figure 1C. Although only a portion of the N terminus is shown, full-length sequences were used for the analysis. Protein homology/analogY recognition engine (PHYRE) (http://www.sbg.bio.ic.ac.uk/phyre/index.cgi) (Bennett-Lovsey et al. 2008) was used to produce a 3D-atomic model of ScMcm5 and ScMcm4 proteins shown in Figure 1D. PyMol (version 1.0) was used with the Phyre prediction coordinates to generate the protein fold predictions. This is more accurate than just a BLAST or CLUSTALW alignments as it is a structure-aided alignment. PHYRE produces a 3D-atomic model of the protein by finding a sequence alignment to a known atomic structure in the structural database. In this case, ScMcm5 and ScMcm4 aligned to the solved N-terminal structure of archaeal MtMCM.
Plasmid loss assays:
To generate the strains used for plasmid loss assays, pRS304-MCM5 and pRS304-mcm5-TP were linearized with Bsu36I and transformed into yRL154, integrating at the trp1-289 locus by homologous recombination. Loss of the URA3 MCM5 plasmid was selected using 5-FOA, and Trp+ transformants were selected generating strains yRL214 (MCM5) and yRL220 (mcm5-TP). yRL214 and yRL220 were then transformed with either pDK-243 LEU2 (1x-ARS site) or pDK-368-7 LEU2 (8x-ARS sites), and Leu+ transformants were selected, generating strains yRL230 (MCM5, 1x-ARS), yRL231 (MCM5, 8x-ARS), yRL236 (mcm5-TP 1x-ARS), and yRL237 (mcm5-TP, 8x-ARS). The stability of the ARS plasmids was calculated as described previously (Hogan and Koshland 1992; Loo et al. 1995).
Generation of mcm5ts strains:
To generate the parental mcm5ts strain, plasmid p583 (Dalton and Hopwood 1997) was digested with MluI and ClaI, yielding a 1.8-kb fragment, which was cloned into the MluI and ClaI sites of pRS306-Mcm5 (Pessoa-Brandao and Sclafani 2004) to produce pRAS651 (Table 1). pRAS651 was linearized with NruI and transformed to strain RSY311 (Sclafani et al. 2002), resulting in a targeted duplication event at the MCM5 locus. Ura+ transformants were selected and then purified on 5-FOA media to select for Ura− “pop-outs.” 5-FOAR colonies were screened for the temperature-sensitive phenotype at 37° (frequency 30%). To integrate MCM5 or mcm5-TP into the mcm5ts genome, pRS306-MCM5∷HA and pRS306-mcm5-TP∷HA were both cut with SnaBI to linearize and transformed into the mcm5ts strain, resulting in a targeted duplication event at the MCM5 locus, generating strains yRL251 (mcm5ts∷MCM5∷HA) and yRL253 (mcm5ts∷mcm5-TP∷HA).
Immunoprecipitations and Western blot analysis:
Protein extracts for all immunoprecipitations were prepared as described previously (Pessoa-Brandao and Sclafani 2004). Briefly, 2 mg of protein extract were incubated with 30 μl protein G-sepharose beads (Sigma-Aldrich) blocked in PK lysis buffer (50 mm Tris pH 8.0, 50 mm NaCl, 0.1% Tween, 0.1% Triton-X-100, 1.0 mm EDTA, 0.1% BSA, 0.01% NaAzide) and supplemented with 0.5 mm PMSF, 0.8 μg/ml leupeptin/0.6 μg/ml pepstatin for 3 hr at 4°. Primary antibody was added to each reaction as follows: 50 μg anti-HA antibody (Roche), 2 μg anti-Mcm2p antibody (Santa Cruz), or 2 μg anti-Mcm7p antibody (Santa Cruz). Negative controls were performed in the absence of antibody and containing beads only. All reactions were incubated at 4° with end-over-end rotation for 3 hr. Samples were centrifuged at 1500 rpm at 4°, and supernatant was mixed with 5× boiling sample buffer (1× final), and boiled at 100° for 5 min before loading to SDS–PAGE. Pellets were washed 3 times with 500 μl of lysis buffer supplemented with 0.5 mm PMSF, 0.8 μg/ml leupeptin/0.6 μg/ml pepstatin, and resuspended in 50 μl 2.5× boiling sample buffer. For immunoblot analysis, 10% of each fraction, whole-cell extract, pellet, or supernatant were resolved as described below. Protein extracts for Western blots were prepared as described previously (Pessoa-Brandao and Sclafani 2004). For Western blots, protein extracts were resolved by 7% SDS–PAGE, transferred to nitrocellulose membrane, and probed with anti-mouse Mcm5 monoclonal antibody (Weinreich et al. 1999) at a 1:500 dilution, anti-HA antibody (Roche) at a 1:2500 dilution, anti-Mcm2p antibody (Santa Cruz) at a 1:1000 dilution, anti-Mcm2-7 (Bowers et al. 2004) at a 1:1000 dilution, and anti-Cdt1(UM185) (Bowers et al. 2004) at a 1:15,000 dilution. For Mcm5 and HA blots, a secondary HRP-conjugated anti-mouse antibody (Jackson ImmunoResearch) was used at a 1:3000 dilution. For the Mcm2p blot, a secondary HRP-conjugated anti-goat antibody (Santa Cruz) was used at a 1:3000 dilution. For Cdt1p and Mcm2-7p blots (Bowers et al. 2004), a secondary HRP-conjugated anti-rabbit antibody (Biorad) was used at a 1:3000 dilution. Immunoblots were visualized on film using an ECL chemiluminescence kit (Pierce).
Time course and cell synchrony for ChIP analysis:
The mcm5ts strains were grown in YEPD to a density of 2 × 107/ml at 22° and arrested with nocodazole (final concentration of 10 μg/ml) at 22° for 60 min. Cells were placed at 37° for 120 min to complete the nocodazole arrest and inactivate the endogenous mcm5ts. Cultures were washed twice in 37° YEPD media to remove nocodazole and released to 37° in prewarmed YEPD containing α-factor at 200 nm final concentration. Ten-minute time points were taken for 60 min at 37°.
Chromatin immunoprecipitation (ChIP) analysis:
Chromatin immunoprecipitation (ChIP) analysis was performed as described previously (Meluh and Koshland 1997; Sclafani et al. 2002) with the following modifications: 100 ml yeast cells (2 × 107/ml) were formaldehyde fixed for 2 hr at room temperature. Cells were spheroplasted with Zymolyase 100T (Seikagaku) in 20 mm HEPES pH 7.4, 1.2 m sorbitol supplemented with 0.5 mm PMSF. Spheroplasts were sonicated on ice, 6 pulses (output level 5 with microtip, Fisher Scientific Model 550 sonic dismembrator) for 10-sec intervals. The lysates were centrifuged at 14,000 rpm at 4°, and supernatants were diluted in IP dilution buffer (0.01% SDS, 1.1% Triton X 100, 1.2 mm EDTA, 16.7 mm TRIS pH 8.1, 167 mm NaCl) at 1:9 dilution. Immunoprecipitations were set up with 1.5 ml chromatin solution per IP reaction. For HA immunoprecipitations, chromatin solution was incubated with 25 μg anti-HA antibody (Roche). For Mcm2p immunoprecipitations, chromatin solution was incubated with 2 μg anti-Mcm2p antibody (Santa Cruz). For Mcm7p immunoprecipitations, chromatin solution was incubated with 2 μg anti-Mcm7p antibody (Santa Cruz). All immunoprecipitations were incubated at 4° for 18 hr. BSA blocked protein G-sepharose beads (30 μl, as described above) were added to each reaction, with 4 μg sonicated Lambda DNA (NEB) and incubated at room temperature for 2 hr. ChIP chromatin was subjected to PCR analysis with the following primers: ARS305 fwd-GATTGAGGCCACAGCAAGACCG, ARS305 rev-CTCCGTTTTTAGCCCCCC GTG, ARS1 fwd-GCTGGTGGACTGACGCCAGAAAATGTT, ARS1 rev-GGTGAAA TGGTAAAAGTCAACCCCCTG, ARS501 fwd-CTTTTTTAATGAAGATGACATTG CTCC, ARS5-01 rev-GATGATGATGAGGAGCTCCAATC, ARS305 + 8 kb fwd-TCAT TTCACTGGGTAGTTCGC, and ARS305 + 8 kb rev-CCGACCATACTCACACACAAG. Each PCR product was run on a 1% agarose gel, followed by densitometry analysis using ImageQuant v5.2. For quantitation, the densitometry value for the no antibody controls were subtracted from the densitometry values for both the wild-type (WT) and TP PCR signals. To calculate the percentage of immunoprecipitate (IP), the following formula was used: percentage of IP = densitometry value of IP “sample” divided by 10(a) × densitometry value of the “total”, where (a) is equal to the volume of the IP sample used in the PCR reaction divided by the volume of the total used in the PCR reaction × 100. Typically, 2.5 μl of the IP sample were used per 25 μl PCR reaction, and 1 μl of the total (diluted at 1:20 in H2O) was used per 25 μl PCR reaction. All ChIP experiments were repeated three times.
Fluorescent-activated cell sorter analysis:
Cells were grown in YEPD and processed for fluorescent-activated cell sorter (FACS) analysis (Ostroff and Sclafani 1995).
mcm5-TP shows defects in the initiation phase of DNA replication:
To explore the possibility of a common structure and function of yeast Mcm proteins, the N-terminal MtMcm crystal structure (Fletcher et al. 2003) was used as a model to query the in vivo origin binding properties of the yeast Mcm2-7p complex. Alanines were substituted at positions K304, R311, and R324 in Mcm5, creating a mcm5-triple mutant (mcm5-TP). We focused on the Mcm5 protein because it is believed to have a regulatory function and not be involved in DNA binding from in vitro studies (Ishimi et al. 1998; Lee and Hurwitz 2000; Schwacha and Bell 2001). In addition, because other yeast Mcm proteins have been shown to bind origin chromatin by ChIP (Aparicio et al. 1997; Tanaka et al. 1997), it is assumed that yeast Mcm5 protein also binds, but this has never been demonstrated directly using ChIP. MtMcm and Mcm5 are also alike in that they are of a similar size and both lack an additional N-terminal domain seen in other Mcm proteins such as in Mcm4 (Fletcher et al. 2003) (Figure 1D), which is the site of phosphorylation by DDK and CDK (Sheu and Stillman 2006). Thus, it is important to investigate the cellular role of Mcm5 protein in binding origins of replication by using yeast genetics.
From both the CLUSTALW primary sequence alignment and the structure-aided PHYRE alignment (Figure 1, C and D), these basic residues lie within the potential β-hairpin domain of Mcm5p, between sheets β9 and β10. Mutation of some of these basic residues in two archaeal proteins, MtMcm (R226A K228A) (Fletcher et al. 2003) and in SsoMcm (K246A R247A), reduces DNA binding activity in vitro (McGeoch et al. 2005). In fact, there is conservation of basic residues in between β9 and β10 sheets in all Mcm5 proteins (Figure 1C). The exact alignment is not critical as even in the archaeal proteins, the residues do not exactly line up (residues highlighted in red in Figure 1C).
The archaeal MtMcm lacks the glycine rich sequence in the β-hairpin domain seen in yeast Mcm5, which has nine glycines in the loop region containing the 19 additional amino acids. Glycine residues typically have no secondary structure, and the Phyre program was unable to predict the overall fold of the β-hairpin region due to this glycine-rich region (Figure 1D). In fact, it is not well conserved in other eukaryotic Mcm5 proteins such as from budding yeast Kluyveromyces lactis (KlMcm5) and in the fission yeast Schizosaccharomyces pombe (SpMcm5), both of which have only one glycine in the region (Figure 1C). The Phyre program also had difficulty predicting part of this region in ScMcm4 (Figure 1D). Nonetheless, the prediction showed that both ScMcm5 and ScMcm4 do indeed have a basic residue near the tip (R324 in ScMcm5 and R445 in ScMcm4) (Figure 1D).
Strains yRL214 and yRL220 were constructed that contained either MCM5 or mcm5-TP, respectively, as the sole MCM5 gene integrated at the TRP1 locus, instead of having the MCM5 gene on a ARS CEN plasmid. In this manner, we avoid the possibility that a mutant growth phenotype results from just the single origin (ARS) on the plasmid firing inefficiently. The mcm5-TP strain grows slowly with a 5-hr doubling time as compared to 2 hr for the MCM5 strain in rich YPD medium at 22°. Most of the phenotype is due to the R324A mutation as K304A or R311A single mutants have little phenotype and the R324 mutant has a similar growth defect to the triple mcm5-TP mutant (data not shown).
The mcm5-TP strain also is not temperature sensitive at 37° or cold sensitive at 16°. Western blots were performed on protein extracts from both MCM5 and mcm5-TP strains. Both strains had similar amounts of protein when grown at either 22° or 37° and then probed with an anti-Mcm5 antibody at 22° or 37° (data not shown). Thus, the alanine mutations in the mcm5-TP strain are unlikely to result in unstable Mcm5 protein.
If the β-hairpin domains in Mcm5p were critical for DNA binding in the initiation of DNA replication, a higher rate of loss of minichromosomes would occur in mcm5-TP as compared to MCM5. This is because less efficient DNA binding activity should facilitate a less efficient binding to replication origins (ARS). The stability of minichromosome ARS/CEN plasmids was measured in strains yRL230 and yRL236, which have only an MCM5 or mcm5-TP gene, respectively. This “plasmid-loss” assay has been used effectively to examine the initiation defects in other replication mutants such as cdc6 (Hogan and Koshland 1992) and orc2 (Loo et al. 1995). Because the minichromosome has only one origin, reduced initiation events will result in increased minichromosome loss. By increasing the number of origins, suppression of plasmid loss will occur by mass action. In this case, plasmid pDK368-7, which has eight ARS sites, was used to produce strains yRL231 and yRL237. As expected, the MCM5 strains showed a low rate of plasmid loss (<1%) per generation, regardless of the number of ARS sites on the minichromosome. In contrast, the mcm5-TP strain yRL236 showed a >32% loss rate per generation, which was suppressed to <1% with a minichromosome harboring eight ARS sites in strain yRL237. This phenotype is similar to cdc6 and orc2 mutants defective in the initiation of DNA replication (Hogan and Koshland 1992; Loo et al. 1995), but is different from a cdc17 mutant (DNA polymerase α) defective in the elongation of DNA replication. The cdc17 mutant also displayed an increased level of plasmid loss, but the phenotype was not suppressible by additional ARS origins on the plasmid (Hogan and Koshland 1992). From these data, we conclude that the mcm5-TP strain harbors a defect in the initiation phase of DNA replication, a phenotype consistent with a defect in DNA binding and failure to set up a functional pre-RC in G1 phase.
mcm5-TP is unable to rescue the temperature-sensitive phenotype of a mcm5ts strain:
Mcm5p fusions at either the N or C terminus often damage protein function (Labib et al. 2000; Nguyen et al. 2000). Therefore, an internally HA-tagged MCM5 haploid strain RSY908 was isolated from diploid strain RSY907 by tetrad analysis. Strain RSY907 had been produced by transposon-directed insertions to produce a MCM5∷HAT tag (Ross-MacDonald et al. 1999). We will refer to it as MCM5∷HA from here on. In 10 tetrads analyzed, the MCM5∷HA fusion segregated 2:2. All 10 MCM5∷HA isolates had no phenotype in that they grew at the same rate as an untagged MCM5 strain (1.5 hr doubling time at 30° in rich YPD medium) and were not temperature sensitive at 16° or 37°. The 93-amino-acid HA insertion is located between α-helix 4 and β-sheet 2 on the outside of the protein based on the structure of the archaeal 3D structure (Fletcher et al. 2003). Because this HA tag is on the outside of the protein, it should be useful for immunoprecipitation (see below).
Because the mcm5-TP strain grows poorly and eventually gives rise to “large-colony” suppressors (data not shown), it was difficult to use. Therefore, a strain carrying a conditional mcm5ts mutation [C183Y, which was originally called cdc46-1 (Chen et al. 1992)] was used for the remaining analyses. MCM5 and CDC46 are the same genes and we will use MCM5 only as recommended by the Saccharomyces Genome Database (http://www.yeastgenome.org/). MCM5 gene duplications were produced with the mcm5ts strain and either MCM5∷HA or mcm5-TP∷HA in strains yRL251 and yRL253, respectively. In these strains, both untagged Mcm5ts and tagged Mcm5 wild-type or Mcm5-TP mutant proteins are produced (see below). Thus, the strains grow well at the permissive temperature (22°) and the phenotype of the mcm5-TP mutant will be revealed only at the restrictive temperature (37°). The strains were arrested in G1 phase with α-factor and then released at either the permissive or restrictive temperatures. In the mcm5ts strain, DNA replication is greatly delayed when grown at the restrictive temperature (Figure 2) because the Mcm5ts protein cannot integrate into the Mcm2-7 protein complex and drive DNA replication (Dalton and Hopwood 1997). The mcm5-TP mutant is nearly indistinguishable from the mcm5ts parental strain under restrictive conditions (Figure 2) and therefore, is also defective in DNA replication. Similar to the mcm5ts parental strain, the mcm5ts∷mcm5-TP strain is leaky and forms small colonies after 3 days at 37°, while the MCM5 strain forms large colonies after 1 day (data not shown). Thus, the β-hairpin mutations in mcm5-TP are detrimental to its function and reduce its activity, which explains its inability to efficiently replicate DNA.
Mcm5-TP protein is incorporated into the Mcm2-7p complex and binds the Cdt1 protein:
To determine if the mutant Mcm5-TP protein is able to integrate into the hexameric Mcm2-7 complex, an anti-HA antibody was used to precipitate the HA-tagged Mcm5p from the HA-tagged MCM5 strains grown at the restrictive temperature for 3 hr (37°, Figure 3). Importantly, the untagged mcm5ts protein was shown previously to be defective in associating with the remaining subunits of the Mcm2-7 protein complex (Dalton and Hopwood 1997) and to prevent Mcm7 from binding to origins at the restrictive temperature (Aparicio et al. 1997; Tanaka et al. 1997). The α-HA immunoprecipitates contained comparable Mcm2 protein from both MCM5 and mcm5-TP strains (Figure 3). Furthermore, immunoprecipitates from both MCM5 and mcm5-TP strains also contained a small amount of Cdt1 protein, which loads the Mcm2-7 complex onto chromatin and has been shown to co-immunoprecipitate with Mcm2 (Tanaka and Diffley 2002). Because a similar amount of Mcm2 protein was shown to co-immunoprecipitate with Cdt1 protein in log phase cells as in G1-arrested cells (Tanaka and Diffley 2002), the fact that the mutant mcm5-TP cells become synchronized during this experiment does not affect the result.
Using an antibody which recognizes a common epitope in all six Mcm2-7 proteins (Bowers et al. 2004), both tagged wild-type Mcm5 and mutant Mcm5-TP proteins also co-immunoprecipitate with the entire Mcm2-7 complex (Figure 3). The α-HA antibody quantitatively immunoprecipitates the larger (105 kDa) Mcm5-HA protein, which is immunodepleted from the extract (compare supernatants and pellets, Figure 3), but not the smaller, untagged Mcm5ts protein seen in the Mcm5 blot (Figure 3).
Thus, the β-hairpin mutations in mcm5-TP do not affect the stability of the Mcm5 protein, do not interfere with incorporation of Mcm5p into the Mcm2-7p complex, and do not affect binding to the Cdt1 protein. Therefore, we infer that these mutations do not disrupt the formation of the Mcm2-7p complex and its ability to bind the Cdt1 loader.
We also conclude that the mcm5-TP mutation shows poor complementation of the mcm5ts mutation (Figure 2), because only Mcm2-7 complexes with the Mcm5-TP protein are formed at the restrictive temperature and these complexes have reduced activity. We will demonstrate in Figure 4 below that the Mcm5-TP protein has a reduced ability to bind origin chromatin.
mcm5-TP mutant shows reduced Mcm2p, Mcm5p, and Mcm7p binding to origin chromatin by ChIP:
Chromatin immunoprecipitation (ChIP) (Meluh and Koshland 1997; Sclafani et al. 2002) was used to determine conclusively that the β-hairpin mutations were causing Mcm5-TP protein to bind DNA/chromatin at origins less efficiently. Again, mcm5ts duplication strains with either MCM5∷HA or mcm5-TP∷HA were used. Origin loading was followed over time in synchronized cells from G2/M to G1 phase, during which the Mcm2-7p complex loads onto origins (Aparicio et al. 1997; Tanaka et al. 1997). Cells were arrested at G2/M using nocodazole, then shifted to 37° to inactivate mcm5ts. Under these restrictive conditions the only Mcm5 protein in the Mcm2-7 complex is either tagged wild-type Mcm5 or tagged mutant Mcm5-TP protein. After nocodazole arrest (Figure 4A), nocodazole was washed out, and the cells were released to G1 (37°) using α-factor (αF). Unlike the transition from G1 to S phase (Figure 3), only subtle differences were seen between the MCM5 and mcm5-TP strains during the transition from G2/M to G1 (Figure 4A). At 30 min, there is a slight delay in the transition in the mcm5-TP strain compared to the MCM5 strain. This is consistent with the fact that the Mcm2-7p complex is not required at this stage of the cell cycle (Forsburg 2004).
At the restrictive temperature, only the HA-tagged Mcm5 wild-type or Mcm5-TP mutant protein will be in the Mcm2-7p complex because the Mcm5ts protein is inactive. In the MCM5∷HA strain, a robust PCR signal for ARS1 was detected in Mcm5 (HA), Mcm2, and Mcm7 (Figure 4B) immunoprecipitates, which peaked at ∼30–40 min after nocodazole release. In contrast, the signal was reduced 5- to 10-fold in the mutant mcm5-TP∷HA strain (Figure 4B). When the cells were arrested in mitosis with nocodazole, the signal returned to baseline as expected. No signal was seen when the antibody was omitted as a negative control (data not shown). Similar results were found at origins ARS501 and ARS305 (Figure 4, C and D, respectively), although a higher background was observed at ARS305 in nocodazole (Figure 4D). A non-origin region that is 8 kb upstream of ARS305 (ARS305 + 8 kb) was used as another negative control (Aparicio et al. 1997), and no significant ChIP signal was observed in this non-origin region (Figure 4E).
At the permissive temperature (22°), the situation is different because both Mcm5ts and HA-tagged Mcm5 or Mcm5-TP proteins are expected to be in Mcm2-7 complexes. When the anti-HA antibody is used, only Mcm2-7 complexes containing the HA-tagged protein are immunoprecipitated (Figure 3). In the case of the Mcm5-TP protein complexes, very little would be bound to origins as found using anti-HA antibody (Figure 4). However, when we use anti-Mcm2 or anti-Mcm7 antibodies, Mcm2-7 complexes containing a mixture of both tagged (Mcm5 and Mcm5-TP) and untagged proteins (Mcm5ts) would be immunoprecipitated. In the case of Mcm5ts and Mcm5-TP complexes, we would expect that only 50%, that is, the Mcm5ts complexes, would be bound to the origin as was found (Figure 4).
From these data, we conclude that the β-hairpin domains in Mcm5p are critical for binding of the entire Mcm2-7p complex to chromosomal origins of DNA replication.
β-Hairpin mutations in mcm4 and mcm5 are synthetically lethal:
To begin an analysis of the β-hairpin domains in the Mcm4 protein (Figure 1, C and D), we used a similar strategy as for Mcm5 protein (materials and methods). Four basic residues in the region between β-sheets 9 and 10 were mutated to alanines to produce the mcm4-4A allele (R445A K454A K458A H456A, depicted in red in Figure 1C). Again, even though we have mutated only 4/6 basic residues in this region, our hypothesis is that just reducing the basic charge of the region will have consequences and yield a mutant phenotype. Strains were produced that have null mutations at the endogenous locus (mcm5Δ and mcm4∷hisG) and wild-type or mutant genes integrated at another location (leu2∷LEU2 mcm4 and trp1∷TRP1 mcm5). Strain RSY1240 (mcm5Δ trp1∷TRP1 mcm5-TP mcm4∷hisG (pMCM4 URA3) or strain RSY1241 mcm5Δ trp1∷TRP1 MCM5 mcm4∷hisG (pMCM4 URA3) were transformed with plasmids pRAS691 (LEU2 mcm4-4A) or pRAS693 (LEU2 MCM4) targeted to integrate at the leu2 locus. Loss of the pMCM4 URA3 plasmid will only occur if the integrated mcm4 plasmid can complement the mcm4∷hisG deficiency. Of the four combinations, only the mcm4-4A mcm5-TP double mutant was not recovered on 5-FOA plates at 22°. In addition, the mcm4-4A single mutant (strain RSY1266) was temperature sensitive and did not form colonies at 37°. We conclude that the mcm5-TP and mcm4-4A mutants may have defects in similar cellular processes, most likely, binding to origins.
Using a structure/function approach, we have identified functionally conserved domains in eukaryotic Mcm proteins using the atomic structure of MtMcm as a structural model and yeast as a functional model. Previously, we used this approach to provide a molecular mechanism to explain the bypass of DDK function by the mcm5-bob1 mutation (P83L) (Fletcher et al. 2003). In that case, yeast genetics was used to identify the important residue (P83) and then the MtMCM structure was used to determine the mechanism. In this report, we used the MtMCM structure to identify important residues and then used yeast genetics to determine the effects of their mutation on function in vivo.
Our results clearly indicate that the β-hairpin domain of the yeast Mcm5 protein is important for origin binding and the initiation of DNA replication. Although we have not directly measured DNA binding by Mcm5 protein, we infer from previous in vitro analyzes of archaeal Mcm proteins (Fletcher et al. 2003; McGeoch et al. 2005) and because all eukaryotic Mcm2-7 proteins have similarity to β-hairpin domains (Fletcher et al. 2003) that the DNA binding function of the β-hairpin domains of the six Mcm2-7 eukaryotic proteins has been functionally conserved throughout evolution.
Both archaeal and eukaryotic MCM proteins have two sets of conserved β-hairpins (McGeoch et al. 2005): the N-terminal β-hairpins, which are analyzed in this report, and the smaller C-terminal β-hairpins. In the archaeal MtMCM protein, mutation of the N-terminal β-hairpins reduced binding to both ssDNA and dsDNA in vitro (Fletcher et al. 2005). In the archaeal SsoMCM protein from Sulfolobus solfataricus, mutation of the N-terminal β-hairpins reduced DNA binding by eightfold in vitro (McGeoch et al. 2005), while mutation of the C-terminal β-hairpins reduced DNA binding by only 2.5-fold. Total DNA binding was abolished only when both sets of β-hairpins were mutated. In their model (McGeoch et al. 2005), the N-terminal β-hairpins are more important for binding the helicase onto DNA, while the C-terminal β-hairpins are used to translocate the helicase along the DNA. Our results are consistent with this model in that mutation of the N-terminal β-hairpins in yeast Mcm5p results in a severe defect in Mcm2-7 loading onto origins in G1 phase of the cell cycle (Figure 4) and affects initiation of DNA replication. SF3 viral helicases such as SV40 T antigen and HPV E1 do not contain the N-terminal β-hairpins as they bind DNA in a manner similar to that of a transcription factor (reviewed in Sclafani et al. 2004). However, these viral proteins contain C-terminal hairpins important for translocation (Li et al. 2003; Gai et al. 2004). In this regard, the archaeal helicases are a better model for eukaryotic Mcm2-7 than the viral proteins.
The mutation in the mcm5ts strain used in these studies lies within the zinc-finger domain (C183Y), a domain hypothesized to mediate hexamer–hexamer interactions within the Mcm2-7p complex (Dalton and Hopwood 1997; You et al. 2002; Fletcher et al. 2005). Additionally, interactions between Mcm proteins in the single hexamer have previously been shown to be mediated by the catalytic arginine hairpins and the P-loop domains of the next corresponding Mcm protein (Davey et al. 2003). In the mcm5ts mutant at the restrictive temperature, the Mcm2-7 complex does not form (Dalton and Hopwood 1997), which explains why it cannot bind origins (Aparicio et al. 1997; Tanaka et al. 1997). In contrast, the mcm5-TP mutant protein is part of the Mcm2-7 complex (Figure 3), can bind to the Cdt1 Mcm2-7 protein loader (Figure 3), but is unable to bind origins (Figure 4). These results support the idea that the β-hairpin mutations are not in regions of Mcm5p that are important for Mcm2-7 subunit protein–protein interactions (Davey et al. 2003).
Only subsets of eukaryotic Mcm proteins, specifically the Mcm4/6/7 complex, have been shown to possess DNA binding and/or helicase activities (Ishimi 1997; You et al. 1999; Lee and Hurwitz 2000). Any purified Mcm complexes containing Mcm5, or Mcm2/Mcm3 proteins, are inactive in binding DNA in vitro; thus, Mcm4/6/7 and Mcm2/3/5 complexes may represent catalytic and regulatory trimers, respectively (Lee and Hurwitz 2000; Schwacha and Bell 2001). These observations also make it difficult to measure the DNA binding activity of Mcm5p in vitro. Thus, we exploited a unique HA tag within Mcm5 (Ross-MacDonald et al. 1999) to query the DNA binding abilities of the Mcm complex in vivo using ChIP analysis. While other Mcm proteins have been shown by ChIP to bind origins (Aparicio et al. 1997; Tanaka et al. 1997), Mcm5p has only been shown to bind bulk chromatin (Weinreich et al. 1999). We provide evidence for a specific interaction between origin chromatin and Mcm5p in vivo. We also demonstrate that the conserved β-hairpin domains of Mcm5p are involved in mediating the origin binding activity of the entire Mcm2-7p complex, as has been seen for the archaeal MtMcm (Fletcher et al. 2003) (Figure 4). Previously, we have also used the archaeal structural model for demonstrating the regulatory role of Mcm5p (Hoang et al. 2007). Thus, Mcm5p likely has more than just a simple regulatory function, but actively participates in binding to origins of replication.
Does the Mcm2-7 complex bind dsDNA at the origin? From the channel dimensions of the MtMcm helicase, we calculated that ∼70 bp would be protected by the binding of a Mcm2-7 double hexamer (Fletcher et al. 2003). Additionally, ∼80 bp of origin DNA is protected by bound Mcm2-7 complexes in frog extracts (Edwards et al. 2002). As yeast origins are in intergenic regions of DNA devoid of nucleosomes (Thoma et al. 1984; Venditti et al. 1994) and remain double stranded even after the Mcm2-7 complex has loaded in G1 arrested cells (Geraghty et al. 2000), we can speculate that yeast Mcm2-7 is binding dsDNA at origins in the central channel via the conserved β-hairpins.
It is important to note that the mutations made in MtMcm (Fletcher et al. 2003) and SsoMcm (McGeoch et al. 2005) β-hairpin domains were present at each of the six β-hairpins facing the positively charged central channel. Because the eukaryotic Mcm2-7p complex consists of heterohexamers (unlike MtMcm and SsoMcm), β-hairpin mutations in a single Mcm subunit (e.g., Mcm5p in this report) disrupt origin binding activity in only one of the possible six N-terminal β-hairpins present in the positively charged channel. Perhaps coordinated DNA binding by β-hairpin domains of all six Mcm2-7 proteins is needed for efficient binding. Introduction of a β-hairpin mutation in a second Mcm subunit (Mcm4p in this report) may completely destroy the coordination producing a lethal situation. Similar to our results, recombinant yeast MCM complexes containing a single ATPase mutant subunit are reduced in activity by 6- to 20-fold, which implies that coordination by all six MCM subunits is needed for full activity (Schwacha and Bell 2001). Coordination of both DNA binding and ATPase activities explains why all six Mcm2-7p subunits are required in vivo and in vitro for DNA replication (Forsburg 2004). Coordinated interactions among all six Mcm protein members may be required for full catalytic activity in vivo as proposed (Schwacha and Bell 2001) and as demonstrated here for origin binding activity. These results should be applicable to other eukaryotic systems because of the evolutionary conservation of both structure and function.
We thank Paul Megee for his ChIP expertise, Stephen Dalton for plasmids, and Karen Helm and Mike Ashton in the University of Colorado Cancer Center Flow Cytometry Core Facility. We also thank Xiaojiang Chen and Rui Zhao for help with the structure-aided alignments for yeast and archaeal Mcm proteins. This work was supported by a Public Health Service award from the National Institutes of Health to R.A.S. (GM-35078) and a National Research service award fellowship from the National Institutes of Health to R.P.L. (GM-070403).
↵1 Present address: Department of Laboratory Medicine and Pathology, Medical School, University of Minnesota, Minneapolis, MN 55455.
Communicating editor: N. M. Hollingsworth
- Received February 29, 2008.
- Accepted April 1, 2008.
- Copyright © 2008 by the Genetics Society of America