Mouse Geminin Inhibits Not Only Cdt1-MCM6 Interactions but Also a Novel Intrinsic Cdt1 DNA Binding Activity*

DNA replication is controlled by the stepwise assembly of a pre-replicative complex and the replication apparatus. Cdt1 is a novel component of the pre-replicative complex and plays a role in loading the minichromosome maintenance (MCM) 2–7 complex onto chromatin. Cdt1 activity is inhibited by geminin, which is essential for the G2/M transition in metazoan cells. To understand the molecular basis of the Cdt1-geminin regulatory mechanism in mammalian cells, we cloned and expressed the mouse Cdt1 homologue cDNA in bacterial cells and purified mouse Cdt1 to near homogeneity. We found by yeast two-hybrid analysis that mouse Cdt1 associates with geminin, MCM6, and origin recognition complex 2. MCM6 interacts with the Cdt1 carboxyl-terminal region (amino acids 407–477), which is conserved among eukaryotes, whereas geminin associates with the Cdt1 central region (amino acids 177–380), which is conserved only in metazoans. In addition, we found that Cdt1 can bind DNA in a sequence-, strand-, and conformation-independent manner. The Cdt1 DNA binding domain overlaps with the geminin binding domain, and the binding of Cdt1 to DNA is inhibited by geminin. Taken together, we have defined structural domains and novel biochemical properties for mouse Cdt1 that suggest that Cdt1 behaves as an intrinsic DNA binding factor in the pre-replicative complex.

Chromosomal DNA replication is subject to strict cell cycle control, which ensures that cells enter S phase once and only once per cell cycle. A considerable body of evidence from both genetic analyses of yeast mutants and biochemical studies using Xenopus egg extracts has shown that the initiation of replication requires the stepwise assembly of protein complexes on chromatin to form a pre-replicative complex (pre-RC) 1 (1)(2)(3)(4)(5)(6)(7)(8).
The pre-RC includes the origin recognition complex (ORC), the minichromosome maintenance protein complex (MCM), and the Cdc6 and Cdt1 proteins. After the activation of S phasepromoting kinases, CDKs, and the Dbf4-dependent kinase, DNA helicase unwinds the two DNA strands, and replication protein A stabilizes single-stranded DNA, thereby allowing an initiation complex to be formed by the loading of DNA polymerases onto the pre-RC. Because most of the components of the pre-RC identified in Saccharomyces cerevisiae and Xenopus have been found in other eukaryotes including humans, it is believed that the mechanisms controlling the initiation of replication are conserved in all eukaryotes. However, the DNA helicase that is associated with the replication fork has not yet been identified, even in S. cerevisiae (1,6,8,9). The best candidate for the replicative DNA helicase is the MCM2-7 complex. The MCM2-7 complex was first identified as a set of genes required for minichromosome maintenance in S. cerevisiae, and it was subsequently identified as a critical component of the replication licensing system in Xenopus egg extracts (3,9,10). MCM2-7 proteins are loaded onto chromatin at late telophase and gradually released as replication forks proceed, and concomitantly, chromatin undergoes a transition to the unlicensed state (3,8,11). The mouse and Schizosaccharomyces pombe MCM4/6/7 complexes possess low but significant DNA helicase activity (12)(13)(14). The presence of single-stranded tails in forked DNA substrates stimulates the processive helicase activity of the S. pombe MCM4/6/7 complex, suggesting that MCM may function as an intrinsic processive DNA helicase whose activity is controlled by as yet undefined modifications or uncharacterized interactions with other factors (15). Thus, to understand the function of the MCM helicase in replication, it is important to determine not only the mechanism by which MCM is loaded onto chromatin but also the interplay between MCM and other replication components of the pre-RC at the onset of replication.
Two factors have been identified as necessary to load MCM proteins onto chromatin: Cdc6 and Cdt1 (1,2,4,8,16). Cdc6 is an essential component of the pre-RC and is conserved among organisms ranging from yeast to humans. Cdc6 is a member of a large superfamily of ATPases known as the AAA ϩ family, and it exhibits significant sequence similarity to subunits of clamploading proteins in eukaryotes (the replication factor C complex) and prokaryotes (the ␥-complex), which load the ringshaped sliding clamp onto DNA (17). Interestingly, S. pombe, mouse, and archaeal MCM complexes have been observed by electron microscopic analysis to have a donut-like shape (18 -20). The loading of MCM onto chromatin is dependent on ORC and Cdc6 in S. cerevisiae, S. pombe, Xenopus, and humans (1,3,9,10). Therefore, Cdc6 has been considered to function as a clamp loader that facilitates the loading of the ring-shaped MCM2-7 complex onto DNA. However, the in vivo subunit composition of the MCM complex remains unclear. In addition, the loading of the MCM2-7 complex onto chromatin has not yet been reconstituted in vitro. Moreover, the stoichiometry of the pre-RC components, including the ORC, the MCM complex, and Cdc6, as well as the local architectural features that are conducive to pre-RC formation and the regulatory mechanisms that govern the conversion of the pre-RC to an initiation complex, remain to be clarified.
Recently, a novel component of the pre-RC, termed Cdt1, has been identified (2,16). Cdt1 was first identified in S. pombe as a gene induced by the CDK-dependent transcription factor cdc10 (21). In both S. pombe and Xenopus, Cdt1 has been shown to be essential for chromosomal replication and the assembly of MCM2-7 on chromatin (22,23). The association of Cdt1 with the pre-RC depends on ORC but is independent of Cdc6 and takes place prior to the association of MCM2-7 with chromatin (22,23). Cdt1 appears to be identical to the essential component of the replication licensing factor RLF-B that has been characterized by Blow and co-workers (24) and Wohlschlegel et al. (25) as the licensing element in Xenopus egg extracts. A Drosophila homologue of Cdt1 has also been isolated (26); the Double Parked gene product is essential for replication, and the protein co-localizes with ORC at sites of chorion gene amplification in ovarian follicle cells. Xenopus, Drosophila, and human Cdt1 are tightly associated with geminin, which is a negative regulator of pre-RC formation that prevents the loading of MCM onto chromatin (1,24,25,27). To date, geminin has been widely used as an inhibitor of pre-RC formation in studies of metazoan DNA replication. In contrast, a geminin counterpart has not been found in S. cerevisiae and S. pombe. Most pre-RC components are conserved among eukaryotes, but geminin is perhaps unique to metazoans. Very recently, an S. cerevisiae homologue of Cdt1 has been identified and shown to be included in the pre-RC and to function in loading the MCM complex onto chromatin (28,29). Surprisingly, S. cerevisiae Cdt1 and Xenopus Cdt1 are only weakly similar (12%), suggesting that amino acid sequence alignments may not be useful in identifying Cdt1 functional domains, regulatory mechanisms, and interactions within the pre-RC. Thus, Cdt1 represents a new target for the regulation of pre-RC formation, and considerable attention has been focused on the molecular function of Cdt1 in DNA synthesis.
To better understand the relevance of the geminin-Cdt1 system to replication in higher eukaryotes, we characterized mouse homologues of Cdt1 (mCdt1). We cloned the mCdt1 cDNA, overexpressed the protein in bacterial cells, and purified it to near homogeneity. Biochemical analysis of purified proteins and yeast two-hybrid analysis allowed us to determine the organizational structure of mCdt1. mCdt1 is divided into three domains: an amino-terminal region, which is poorly conserved among eukaryotes, a central region (amino acids 177-380), which contains a geminin binding site and which is conserved among metazoans, and a carboxyl-terminal region (amino acids 407-477), which binds to MCM6 and is highly conserved among eukaryotes. Unexpectedly, we found that mCdt1 binds to both single-and double-stranded DNA in a sequence-nonspecific manner. The DNA binding domain was found to reside in the amino-terminal region that overlaps with the geminin binding domain. Interestingly, geminin disrupts not only the interaction between mCdt1 and MCM6 but also the mCdt1 DNA binding activity. These results suggest that mCdt1 plays a key role in the establishment of the pre-RC by mediating protein-protein interactions via specific domains and that it possesses an intrinsic DNA binding activity that allows it to anchor the MCM complex onto chromatin.

EXPERIMENTAL PROCEDURES
Cloning of mCdt1 cDNA-The amino acid sequence of the human Cdt1 was used to screen the NCBI EST database for the murine Cdt1 homologue. This search identified three murine expressed sequence tags AI605978, AA139554, and AA671429 (NCBI accession numbers). Sequence-specific primers were used to generate a FLAG tagged fulllength mCdt1 by PCR from a murine embryonic cDNA library (Clontech). PCR primers were as follows: forward primer, 5Ј-CCGCTAGCG-ACTACAAGGACGACGATGACAAGCATATGGCGCAAAGTCGTGTT-ACCGA-3Ј; reverse primer, 5Ј-GGCTCGAGCCCCTCGGCGTGGACGT-G-3Ј. The PCR product was digested with NheI-XhoI and inserted into NheI-and XhoI-digested pET24b (Novagen) to generate pET-FLAG-mCdt1-His 6 . The FLAG tag was incorporated into the forward primer.
Expression and Purification of Recombinant Proteins in E. coli-FLAG-tagged mCdt1 was overproduced in the E. coli strain BL21(DE3) grown in 2ϫ YT medium (1.6% tryptone, 1% yeast extract, and 0.5% NaCl) with 100 g/ml kanamycin. Isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 0.4 mM, and the cells were incubated for an additional 1 h at 25°C. Cells were lysed by sonication in Buffer A containing 50 mM potassium phosphate, pH 7.5, 50 mM KCl, 0.1% Triton X-100, and Complete protease inhibitor mixture (Roche Molecular Biochemicals). The lysate was cleared by centrifugation and incubated at 4°C for 1 h with phosphocellulose resin (Whatman) previously equilibrated with Buffer A. Following incubation, the resin was washed extensively with Buffer A, and the proteins were eluted with Buffer A plus 300 mM KCl. The eluate was mixed with cobalt-chelating TALON-Sepharose (Clontech) equilibrated with Buffer A. The resin was washed with 10 column volumes of Buffer A plus 300 mM NaCl, and the protein was eluted with Buffer A containing 150 mM imidazole and 150 mM KCl. The eluate was mixed with anti-FLAG monoclonal antibody-conjugating Sepharose (Sigma) equilibrated with Buffer A for 2 h at 4°C. Following incubation, the resin was washed extensively with Buffer A, and the proteins were eluted with Buffer A plus 0.1 mg/ml FLAG peptide. Finally, the proteins were applied to a Superdex 200 PC2.3/30 SMART column (Amersham Biosciences) equilibrated with Buffer A containing 150 mM KCl.
Deletion mutants with the GST tag were purified according to the manufacturer's instructions (Amersham Biosciences). T7-and His 6tagged mouse geminin was expressed in bacteria and purified to near homogeneity by cobalt-chelating TALON-Sepharose (Clontech) on a MiniS SMART column followed by a Superdex 200 PC2.3/30 SMART column according to instructions provided by the manufacturer (Amersham Biosciences). Purification of the recombinant mouse MCM4/6/7 complex in insect cells was performed as described previously (13).
Coimmunoprecipitation Analysis-Fifty ng of mCdt1 and 50 ng of MCM4/6/7 complex were mixed in the presence or absence of various amounts of geminin and immunoprecipitated with 1 l of anti-mCdt1 antiserum or preimmune serum that had been pre-absorbed to protein A-Sepharose (Amersham Biosciences) for 4 h at 0°C in 100 l of NET-gel buffer containing 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% Nonidet P-40, and 0.25% gelatin. After washing with NET-gel buffer, precipitates were dissolved with 30 l of 2ϫ Laemmli sample buffer and then subjected to SDS-PAGE and Western blot analysis using anti-MCM6, anti-geminin (Santa Cruz Biotechnology), and anti-mCdt1 antibodies as described previously (31).
GST Pull Down Assay-Fifty ng of GST-tagged mCdt1 mutant proteins and geminin or the MCM4/6/7 complex were mixed with glutathione-Sepharose (Amersham Biosciences) for 4 h at 4°C in 100 l of NET-gel buffer. After washing with NET-gel buffer, GST-tagged proteins were eluted with 30 l of Laemmli sample buffer and then subjected to SDS-PAGE and Western blot analysis using anti-GST (Amersham Biosciences), anti-MCM6 (Santa Cruz Biotechnology), or anti-T7 tag (Novagen) antibodies.
DNA-cellulose Binding Assay-Fifty ng of GST-tagged mCdt1 was incubated with 20 l of single-stranded DNA-cellulose or doublestranded DNA-cellulose (Sigma), which had been equilibrated with Buffer A for 1 h at 4°C. After washing with Buffer A, bound proteins were eluted by increasing concentrations of NaCl. Proteins were subjected to SDS-PAGE followed by silver staining. To identify the mCdt1 domain essential for DNA binding, full-length mCdt1 was mixed simultaneously with five amino-terminally truncated proteins (GST⌬N1-127, GST⌬N1-176, GST⌬N1-293, GST⌬N1-406, and GST⌬N1-478) or three carboxyl-terminally truncated proteins (GST⌬C381-557, GST⌬C180 -557, and GST⌬C131-557) and incubated with DNA-cellulose, and bound proteins were eluted by increasing concentrations of NaCl. Proteins were detected by silver staining. To examine the effect of geminin on the DNA binding activity of mCdt1, increased amounts of T7-tagged geminin were preincubated with 50 ng of GST-tagged mCdt1 for 1 h at 4°C and incubated with double-stranded DNA-cellulose for 1 h 4°C, and unbound proteins were subjected to SDS-PAGE followed by Western blot analysis with an anti-mCdt1 and anti-geminin antibodies.

RESULTS
cDNA Cloning of Mouse Cdt1-To identify the cDNA encoding the mouse Cdt1, we screened the mouse EST database and obtained the clones AI605978, AA139554, and AA671429. The full-length open reading frame was obtained by screening a mouse embryonic fibroblast cDNA library (Clontech). The cloned cDNA encodes a putative protein of 557 amino acid residues with a calculated molecular mass of 61.5 kDa. The predicted protein is 72% identical and 79% similar to human Cdt1, 46% identical to Xenopus Cdt1, and 25% identical to Drosophila Cdt1/DUP. Alignment of the mCdt1 cDNA sequence with the human, Xenopus, Drosophila, S. pombe, and S. cerevisiae sequences reveals extensive conservation in the carboxyl-terminal region of the proteins for a broad range of eukaryotes (Fig. 1). The center region, spanning 200 to 350 residues, is conserved among metazoans but is not found in fungi. Canonical sequence motifs were not detected for the mouse sequence. We also screened genomic sequence databases with the mCdt1 cDNA and found genomic sequences containing mCdt1 open reading frames. The full-length mCdt1 open reading frame consists of 10 exons interrupted by 9 introns and overall spans 4598 bases (data not shown). All exon-intron junction sequences conform to the canonical GT/AG rule. The positions of introns are depicted by arrowheads in Fig. 1B. Interestingly, the central region conserved among metazoans is FIG. 2. mCdt1 interacts with ORC2, geminin, and MCM6 in yeast cells. A, interactions between pairs of fusion proteins containing either an amino-terminal LexA DNA binding domain (BD) or a B42 transactivation domain (AD) were tested by yeast two-hybrid assay with the lacZ reporter gene. Center panel, four independent colonies containing the indicated plasmids were inoculated on selection medium containing the substrate 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) and incubated for 24 h at 37°C. Right panel, ␤-galactosidase reporter activity was measured using the substrate chlorophenol red-␤-D-galactopyranoside (CPRG). Each bar represents three independent measurements. Error bars indicate standard deviation. As a negative control, BD or AD vectors lacking inserts were transformed into yeast. B, yeast two-hybrid analysis using various Cdt1 bait fusions with MCM6 or geminin fusions as a prey to localize the interaction domain of mCdt1 with respect to geminin and MCM6. ␤-Galactosidase activities representing interactions of mCdt1 with geminin or MCM6 are indicated by black and striped bars, respectively. The highly conserved region at the center and carboxyl-terminal region are shown by closed boxes. encoded exactly by exons 4 and 5, whereas the carboxyl-terminal region conserved among eukaryotes is encoded by exons 9 and 10. While our manuscript was in preparation, cDNA encoding mouse Cdt1 homologue was reported (32). The coding sequence of our mCdt1 cDNA was identical to that of the mCdt1 gene published recently (32).
Yeast Two-hybrid Analysis-To analyze mCdt1 protein-protein interactions, we carried out a yeast two-hybrid analysis. Full-length mCdt1 was fused to the LexA DNA binding domain as a bait, and various replication factors, including mouse ORC1-6, MCM2-7, Cdc6, Cdc45, and geminin, were fused to the BA42 transcriptional activation domain as prey, and interactions between bait and prey were monitored in yeast cells. We found that mCdt1 associates tightly with mouse MCM6 and mouse geminin (Fig. 2A). In addition, a weak but significant interaction of mCdt1 with mouse ORC2 was also detected. To further determine which domains are required for interactions with MCM6 and geminin, an mCdt1 deletion series was constructed as baits, as shown in Fig. 2B, and two-hybrid analyses were performed. The amino-terminally truncated mutant ⌬N1-171 and the carboxyl-terminally truncated mutant ⌬C520 -557 strongly interact with both geminin and Mcm6, as efficiently as does full-length mCdt1. The interaction involving the ⌬C520 -557 construct reproducibly yielded higher ␤-galactosidase activity than did interactions involving full-length mCdt1, suggesting that the carboxyl-terminal region (521-557) may inhibit the binding of mCdt1 to geminin or MCM6. In contrast, mutant constructs with larger deletions (⌬N1-458 and ⌬C131-557) are not capable of binding to MCM6 and geminin fusions. Interestingly, the amino-terminally truncated mutant ⌬N1-334 associates with MCM6 but not with geminin, whereas the carboxyl-terminally truncated mutant ⌬C289 -557 associates with geminin but not with MCM6. These results suggest that the central region (residues 172-288) is important for interaction with geminin, whereas the carboxyl terminus (residues 335-519) is required for interaction with MCM6. These results were confirmed by in vitro binding experiments as described below (see Fig. 4).
Expression and Purification of Bacterially Expressed mCdt1-The cDNA encoding mCdt1 was inserted into the prokaryotic expression vector pET24a and flanked with sequences encoding the FLAG peptide at the amino terminus, and the His 6 epitope tag at the carboxyl terminus (Fig. 3A). This construct was introduced into BL21(DE3). Expression of recombinant proteins was induced by isopropyl-␤-D-thiogalactopyranoside, and expressed proteins were purified by sequential column chromatography on phosphocellulose, TALON-chelating Sepharose, and anti-FLAG antibody-conjugating Sepharose. The eluate from the FLAG-Sepharose affinity matrix was analyzed by SDS-PAGE and silver staining as shown in Fig.  3B. FLAG-and His 6 -tagged mCdt1 was purified as a 65-kDa protein to near homogeneity. Several deletion mutants that were designed based on information obtained from yeast twohybrid analyses were also fused with the GST tag (Fig. 3A) and purified as shown in Fig. 3C. Using these purified proteins, we examined the interactions of mCdt1 with geminin, MCM6, and DNA.
Co-immunoprecipitation of mCdt1 with geminin and the MCM4/6/7 Complex-To confirm physical interactions between mCdt1 and geminin or MCM6, as suggested by the yeast two-hybrid analysis, we performed a co-immunoprecipitation analysis using an anti-mCdt1 antibody. Anti-mCdt1 antiserum was raised against a bacterially expressed and purified carboxyl-terminal fragment of mCdt1 that spans residues 425 to 557. As shown in Fig. 4A, we found that MCM6 is co-precipitated with the anti-mCdt1 antibody in the presence of mCdt1. Moreover, in the presence of geminin, the amount of co-precipitated MCM6 is severely reduced in a dose-dependent manner (Fig.  4B, lanes 3-5). Thus, we confirmed that mCdt1 physically interacts with MCM6 and geminin. In addition, geminin interferes with the interaction between mCdt1 and MCM6. We also detected physical interactions by co-immunoprecipitation analysis between hemagglutinin-tagged mCdt1 and T7-tagged ORC2 or T7-tagged MCM6 that were overexpressed in mammalian cultured cells (data not shown).
DNA-cellulose Binding Assay-During the purification of mCdt1 proteins, we noticed that mCdt1 associates with singlestranded DNA-cellulose. Fractionation of mCdt1 was performed using either single-stranded DNA-cellulose or doublestranded DNA-cellulose. In the presence of 50 mM NaCl, mCdt1 associates with both single-and double-stranded DNA-cellulose (Fig. 5A). mCdt1 is eluted from single-stranded DNAcellulose at 200 mM and from double-stranded DNA-cellulose at 250 mM. These results suggest that mCdt1 is capable of associating with either single-stranded or double-stranded DNA as an intrinsic DNA-binding protein. This result was unexpected based on what is known for the S. pombe, S. cerevisiae, Xenopus, Drosophila, and human Cdt1 proteins (22-25, 28, 29, 33, 34). We next examined the effect of geminin on the DNAcellulose binding activity of mCdt1. In the presence of geminin, mCdt1 was found in the flow-through fraction from doublestranded DNA-cellulose, indicating that geminin inhibits the interaction of mCdt1 with double-stranded DNA in a dose-dependent manner (Fig. 5B).
Electrophoretic Mobility Shift Assay of mCdt1-To confirm the ability of mCdt1 to bind to DNA, we performed electrophoretic mobility shift assays. 32 P-End-labeled oligonucleotides with lengths of 17, 37, or 70 bases were incubated with purified mCdt1, and bound proteins were detected by agarose gel electrophoresis followed by autoradiography. Incubation of mCdt1 with the 70-and 37-mer results in the formation of a distinct complex of altered mobility, suggesting that mCdt1 associates tightly with the 37-and the 70-mer in a dose-dependent manner (Fig. 6A). In sharp contrast, mCdt1 causes only a slight increase in the mobility of the 17-mer, indicating that mCdt1 cannot efficiently associate with the 17-mer and that it prefers longer single-stranded oligonucleotides. mCdt1 incubated with DNA does not enter a 4% acrylamide gel, suggesting that the mCdt1⅐DNA complex is a large network (data not shown). It is  6. Electrophoretic mobility shift assay of mCdt1. A, various amounts of mCdt1 were incubated with 5 fmol of 32 P-labeled oligonucleotides (17, 37, or 70 bases in length) for 30 min at 37°C. Protein⅐DNA complexes were analyzed by 1% native agarose gel electrophoresis in 0.5ϫ TBE. Gels were dried, exposed on imaging plates, and analyzed with a BAS2500 Fuji image analyzer. Right and left panels show single-stranded oligonucleotide (ssDNA) and double-stranded oligonucleotide (dsDNA), respectively. B, competition assay of mCdt1 with various DNA substrates. Fifty ng of mCdt1 was incubated with 5 fmol of a 32 P-end-labeled 70-mer in the presence of various amounts of DNA substrates as indicated, and protein⅐DNA complexes were analyzed by agarose gel electrophoresis. AT indicates a 70-base oligonucleotide containing an ARS1 sequence. M13mp18, pUC18(CC), and pUC18(L) indicate single-stranded M13 plasmid DNA, covalently closed circular double-stranded pUC DNA, and SmaI-digested double-stranded linear pUC18 plasmid DNA, respectively. C, full-length mCdt1 (FLAG-mCdt1-His 6 (FLAG-FL; lane 2)) and various noteworthy that the mobility-shifted bands are heterogeneous in size. With lower amounts of mCdt1, a slight shift in mobility is observed with the 37-and the 70-mer. When increased amounts of mCdt1 are added, complexes that migrate even more slowly than the initial specific complex are observed. Thus, a multimeric mCdt1 complex might form in the presence of DNA in a dose-dependent fashion.
Next, we performed competitive electrophoretic mobility shift assays. The 32 P-end-labeled 70-base oligonucleotide was mixed with various amounts of the unlabeled 70-mer, an ATrich 70-mer containing the S. cerevisiae ARS1 sequence, singlestranded M13 plasmid DNA, covalently closed circular doublestranded pUC DNA, or SmaI-digested double-stranded linear pUC18 plasmid DNA (Fig. 6B). In the presence of the unlabeled 17-mer, the association of mCdt1 with the 32 P-end-labeled 70mer is unchanged, consistent with the previous result (Fig. 6A) that association with a 32 P-labeled 17-mer is less efficient. In contrast, other substrates, including the 70-mer, singlestranded DNA, covalently closed circular DNA and linearized double-stranded DNA compete efficiently with the 32 P-labeled 70-mer. Hence, these results indicate that mCdt1 associates with DNA in a sequence-nonspecific, strand-nonspecific, and conformation-nonspecific manner. We also found that poly(dT), salmon sperm DNA, and activated calf thymus DNA compete efficiently with the 32 P-labeled oligonucleotide, whereas yeast transfer RNA or DNA that is treated with high concentrations of ethidium bromide do not (data not shown).
To confirm our conclusion that the amino-terminal region of mCdt1 is a binding domain with DNA, an electrophoretic mobility shift assay with deletion mutant proteins was carried out. The amino-terminally truncated protein (GST⌬N1-293) and the carboxyl-terminally truncated proteins (GST⌬C381-557 and GST⌬C131-557) were mixed with 32 P-end-labeled 70base oligonucleotide, and bound proteins were detected by agarose gel electrophoresis. As shown in Fig. 6C, incubation of the probe with GST⌬C381-557 and GST⌬C131-557 results in the formation of a distinct complex of altered mobility, whereas, GST⌬N1-293 causes only a slight increase in the mobility of 70-mer. These results clearly indicate that the amino-terminal region of mCdt1 (residues 1-293) is responsible for binding to DNA.
Finally, the effect of geminin on the DNA binding activity of mCdt1 was examined. The 32 P-end-labeled 70-mer was mixed with mCdt1 in the presence of various amounts of geminin, and a complex with reduced mobility was detected as shown in Fig.  6D. In the presence of geminin, the intensity of this signal is severely reduced, indicating that geminin inhibits the interaction of mCdt1 with DNA. These results are consistent with our finding (Fig. 5B) that geminin inhibits the interaction of mCdt1 with DNA-cellulose. When excess amounts of geminin were mixed with mCdt1, we observed a distinct shifted complex as shown in Fig. 6D, lanes 4 and 5 (asterisk). This minor signal represents a geminin⅐mCdt1 complex, because the signal is super-shifted by incubation of the complex with either antigeminin or anti-mCdt1 antibodies (data not shown), suggesting that the geminin⅐mCdt1 complex also has a weak affinity for single-stranded DNA. However, most of the probe is observed at the bottom of the gel, indicating that the principal effect of geminin is to inhibit the binding of mCdt1 to DNA. DISCUSSION In this study, we identified domains and novel biochemical properties of the mouse Cdt1 protein. mCdt1 associates with geminin and the MCM4/6/7 complex via the central region, which spans amino acid residues 177-380, and the carboxylterminal region, which spans residues 407-477. In addition, mCdt1 is capable of binding to DNA through its amino-terminal region (residues 1-293). Although mCdt1 prefers longer oligonucleotides such as a 70-mer to the shorter 17-mer, it efficiently binds to DNA in a sequence-nonspecific, strandnonspecific, and conformation-nonspecific manner. Geminin inhibits the ability of mCdt1 to bind both MCM and DNA. These findings will provide new insights for understanding the molecular basis of the initiation of replication.
Mechanism of the Inhibition of Cdt1 Activities by geminin-Our yeast two-hybrid analysis and deletion studies with bacterially expressed and purified proteins revealed that the central region and the carboxyl-terminal region of mCdt1 play roles in binding to geminin and MCM6, respectively. To date, geminin, a specific inhibitor of Cdt1, has been found only in metazoans but not in S. cerevisiae and S. pombe. Interestingly, the carboxyl-terminal region of Cdt1 is conserved among all eukaryotes, including S. cerevisiae and S. pombe, whereas the central region is only conserved among metazoans, including Drosophila, mice, and humans (Fig. 1B). These results suggest that the Cdt1 domain that associates with the MCM4/6/7 complex is conserved among all eukaryotes, whereas the geminin binding domain is conserved only among metazoans.
Although geminin is used widely to inhibit the loading of the MCM complex onto Xenopus chromatin, the mechanism by which geminin inhibits the assembly of the pre-RC is not fully understood (1,8,16). In the presence of geminin, interactions between not only mCdt1 and MCM6 but also between mCdt1 and DNA are disturbed severely (see Figs. 4B and 5B). How does geminin inhibit the binding of Cdt1 to MCM and DNA? The mCdt1 DNA binding region was found to overlap with the geminin binding domain, which is in the central region (177-380). Thus, geminin could mask the DNA binding region by tightly interacting with the mCdt1 central region. However, we have found that the geminin and MCM interaction domains are different. It is tempting to speculate that the association of geminin with the mCdt1 central region causes a conformational change in the overall structure of mCdt1, which concomitantly causes the carboxyl-terminal binding domain to be masked. To clarify the precise mechanism by which geminin regulates Cdt1, structural information will prove useful in elucidating how Cdt1 and geminin interact at the protein level.
The Architecture of the Pre-RC-Previously, we found specific interactions between human MCM10 and human ORC2 or mouse MCM6 by the yeast two-hybrid technique and by coimmunoprecipitation analysis (30). Recently, human MCM10 was found to be expressed at the G 1 /S boundary and to accumulate during S phase (35). In contrast, human Cdt1 accumulates in G 1 phase and rapidly diminishes in S phase (25,33,34). Because the timing of expression of human Cdt1 and human MCM10 are mutually exclusive and because both Cdt1 and MCM10 associate successively with ORC2 and MCM6, it is interesting to speculate that Cdt1 is replaced by MCM10 at the G 1 /S phase boundary and that this change reflects the transiamounts of mCdt1 truncated proteins (GST-⌬N1-293 (lanes 4 -7), GST-⌬C381-557 (lanes 9 -12), and GST-⌬C131-557 (lanes 14 -17)) were incubated with 5 fmol of 32 P-labeled 70-base oligonucleotide for 30 min at 37°C. Protein⅐DNA complexes were analyzed by 1% native agarose gel electrophoresis. Lanes 1, 3, 8, and 13 contained no proteins. D, the effect of geminin on the DNA binding activity of mCdt1. Fifty ng of mCdt1 and various amounts of geminin (25, 50, 100, and 200 ng) were mixed at 4°C for 30 min prior to incubation with 5 fmol of 32 P-labeled oligonucleotides. The asterisk indicates a Cdt1⅐geminin complex that weakly associates with single-stranded oligonucleotide. tion of the pre-RC to an initiation complex, as predicted by Bell and Dutta (1).
Cdc6 has been considered to function as a clamp loader for MCM, because it exhibits similarity to the bacterial clamp loader DnaC (17). After DnaC loads replicative DnaB helicase onto the origin, DnaC plays an additional role in anchoring the replicative DnaB helicase at the origin until the onset of replication, which is mediated by a cryptic single-stranded DNA binding activity (34). In this report, we found that mCdt1 possesses single-stranded DNA binding activity. In addition, it has been reported that Cdt1 assembles onto the pre-RC prior to loading MCM during G 1 phase and rapidly disappears from the nucleus during S phase (25,29,34). Therefore, it is tempting to speculate that the DNA binding activity of Cdt1 may contribute to anchoring the MCM helicase at the origin during G 1 phase, much as DnaC anchors DnaB helicase at the origin until the initiation of replication. To confirm this function for Cdt1 and to characterize its DNA binding activity, it may be necessary to reconstitute an MCM loading and unwinding system in vitro using recombinant MCM2-7, Cdc6, Cdt1, and a physiological template that includes nucleosomes or the nuclear matrix, a feat that has not been achieved to date. The protein-protein interactions that we have identified between Cdt1 and components of the pre-RC will be useful in the in vitro reconstitution of pre-RC formation.
mCdt1 Is a DNA Binding Factor-mCdt1 associates with oligonucleotides longer than 37 bases, double-stranded covalently closed circular DNA, single-stranded circular DNA, poly(dT), and linearized double-stranded DNA. Sonicated salmon sperm DNA and activated calf thymus DNA also compete efficiently with the binding of mCdt1 to a 32 P-labeled oligonucleotide. Hence, mCdt1 associates with DNA in a sequence-, strand-, and conformation-independent manner. However, mCdt1 does not bind to yeast transfer RNA or to DNA that is treated with high concentrations of ethidium bromide (36), suggesting that mCdt1 specifically recognizes DNA (data not shown). This binding profile is reminiscent of that of members of the HMG protein family (37), the RNP-U protein (38), and some initiator proteins such as DnaA, O (39), or yeast ORC (40,41). HMG proteins and the RNP-U protein can associate with a wide range of DNA structures and are considered to modulate chromatin structure (37,38). Although the initiator proteins DnaA, O, and yeast ORC have a high affinity for specific sequences, these proteins also exhibit an intrinsic affinity for a broad range of DNA structures, including singlestranded DNA and double-stranded DNA (39 -41). Sequence alignments show no relationship between mCdt1 and these DNA-binding proteins; therefore, structural information may provide further characterization of the DNA binding property of mCdt1.
Recently, Wilmes and Bell (42) have reported in studies of S. cerevisiae that, in addition to the ARS consensus sequence motif characterized previously, a site called the B2 element requires an unknown DNA binding factor that is not ORC to be unwound. Thus, to understand the molecular mechanism of melting of DNA at origins of replication, the identification of new DNA binding factors is important. Because mCdt1 has a strand-nonspecific DNA binding activity, it may be a good candidate for a putative initiation factor. To test the possibility that multimerized mCdt1 may have a DNA unwinding activity, we assayed its effect on relaxed covalently closed circular DNA as a substrate (43). However, we were unable to detect such an activity using purified mCdt1 protein and covalently closed circular plasmid (data not shown). Future studies will be necessary to define the role of the mCdt1 DNA binding activity at the onset of replication. Localization of mCdt1 in the nucleus, the effect of mCdt1 on the DNA binding activity of ORC, and its effect on MCM helicase activities with respect to physiological substrates should provide new insights into the molecular basis of eukaryotic replication.