Origin Single-stranded DNA Releases Sld3 Protein from the Mcm2–7 Complex, Allowing the GINS Tetramer to Bind the Mcm2–7 Complex*

The replication fork helicase in eukaryotic cells is comprised of Cdc45, Mcm2–7, and GINS (CMG complex). In budding yeast, Sld3, Sld2, and Dpb11 are required for the initiation of DNA replication, but Sld3 and Dpb11 do not travel with the replication fork. Sld3 and Cdc45 bind to early replication origins during the G1 phase of the cell cycle, whereas Sld2, GINS, polymerase ϵ, and Dpb11 form a transient preloading complex that associates with origins during S phase. We show here that Sld3 binds tightly to origin single-stranded DNA (ssDNA). CDK-phosphorylated Sld3 binds to origin ssDNA with similar high affinity. Origin ssDNA does not disrupt the interaction between Sld3 and Dpb11, and origin ssDNA does not disrupt the interaction between Sld3 and Cdc45. However, origin ssDNA substantially disrupts the interaction between Sld3 and Mcm2–7. GINS and Sld3 compete with one another for binding to Mcm2–7. However, in a mixture of Sld3, GINS, and Mcm2–7, origin ssDNA inhibits the interaction between Sld3 and Mcm2–7, whereas origin ssDNA promotes the association between GINS and Mcm2–7. We also show that origin single-stranded DNA promotes the formation of the CMG complex. We conclude that origin single-stranded DNA releases Sld3 from Mcm2–7, allowing GINS to bind Mcm2–7.

The initiation of DNA replication involves the complex orchestration of proteins that begin replication fork unwinding (1)(2)(3)(4). Replication fork unwinding is required for the generation of single-stranded DNA templates that are used by the DNA polymerases for copying the genomic information (5). The Mcm2-7 complex is a heterohexameric assembly that binds and hydrolyzes ATP and functions as the motor of the replication fork helicase (6 -8). Sister Mcm2-7 rings load at an origin of replication in late M phase and G 1 (9,10). Recent evidence suggests that the sister Mcm2-7 rings separate upon initiation of DNA unwinding (11).
Sld2, Sld3, and Dpb11 are three proteins in budding yeast that are required for the initiation of DNA replication (17)(18)(19) and also for break-induced replication (20). However, unlike the CMG complex, the Sld3 and Dpb11 proteins do not travel with the replication fork (21,22). Sld3 and Cdc45 bind to early replication origins during the G 1 phase of the cell cycle (18,23). During S phase, the cyclin-dependent kinase (CDK) phosphorylates Sld2 and Sld3, and this event triggers the association of Sld3 with Dpb11 and Sld2 with Dpb11 (24,25). Furthermore, Sld2, Dpb11, GINS, and polymerase ⑀ form a transient preloading complex that is dependent on CDK and important for the recruitment of these factors to origins of replication (23). It has recently been shown that CDK-phosphorylated Sld2 binds tightly to origin single-stranded DNA (ssDNA) in vitro (26).
The Dbf4-dependent kinase (DDK) is also required for the initiation of DNA replication. DDK phosphorylates subunits of the Mcm2-7 complex, and phosphorylation of Mcm4 by DDK is important for cell growth and for the tight association of Cdc45 with the Mcm2-7 complex (27,28). DDK phosphorylation of Mcm4 is also critical to alleviate a replication inhibitory activity of the N terminus of Mcm4 (29).
In this study, we find that Saccharomyces cerevisiae Sld3 binds tightly to single-stranded DNA found at an origin of replication. CDK-phosphorylated Sld3 also binds to origin ssDNA with similar high affinity. Origin ssDNA does not disrupt the interaction between Sld3 and Dpb11 or between Sld3 and Cdc45. However, origin ssDNA disrupts the interaction between Sld3 and Mcm2-7. GINS and Sld3 compete with one another for Mcm2-7 binding. However, in a mixture of Sld3, GINS, and Mcm2-7, origin ssDNA inhibits the interaction between Sld3 and Mcm2-7, whereas origin ssDNA promotes the association between GINS and Mcm2-7. We also show that origin single-stranded DNA promotes the formation of the CMG complex. We conclude that origin single-stranded DNA releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7.

EXPERIMENTAL PROCEDURES
Plasmids-Full-length CDC45 or SLD3 PCR product was cloned into the BamHI/XhoI site of either pET 41a or pET 33b to contain a GST tag or PKA recognition sequence at the N terminus of the protein, respectively. The stop codon was removed at the C terminus to allow for a read-through into the His 6 tag.
Plasmids for the expression of GINS were a generous gift from Mike O'Donnell. These GINS plasmids (pET-Psf3 and Psf5, pLANT-Psf2, pCDF-His-Psf1) were used to simultaneously co-express all four subunits of GINS (30). PKA-GINS has a PKA tag at the N terminus of Psf1. CDK protein expression plasmids and purification methods were a generous gift of Dr. H. Araki (31).
Expression-The plasmids were transformed into Escherichia coli Rosetta 2 (BL21) (EMD Biosciences) and a colony from transformation was used to inoculate 6 liters of ZYM-5052 auto-inducing media (33) containing 100 g/ml kanamycin and 34 g/ml chloramphenicol. For GINS, the media also contained 20 g/ml streptomycin sulfate and 100 g/ml ampicillin. The cells were grown at 37°C until an A 600 of 1.0 was reached, then the temperature was then lowered to 14°C, and the cells were allowed to express protein for 16 h.
Purification of GST Proteins-Harvested cells were lysed with a French press in 200 ml of 50 mM Tris-HCl, pH 8.0, 500 mM NaCl, and 50 mM imidazole. The lysate was then applied to a 20-ml NiSO 4 -charged chelating Sepharose fast flow resin (GE Healthcare). The column was washed with the same buffer. The protein was then eluted in 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10% glycerol, and 250 mM imidazole. Peak fractions were pooled, and 4 ml of glutathione-Sepharose 4B (GE Healthcare) resin were added. The mixture was rotated gently on ice for 1 h and then poured into a column, washed with 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 10% glycerol, and eluted with the same buffer containing 100 mM glutathione. Peak fractions were then pooled, dialyzed against 50 mM Tris-HCl, pH 8.0, 50 mM NaCl, and 10% glycerol, and applied to an 8-ml Source Q column (GE Healthcare). The protein was then eluted with a 20-ml linear gradient from 50 to 500 mM NaCl. Peak fractions were pooled and frozen.
Histidine-tagged proteins were purified using the same procedure as GST proteins, except that the glutathione-Sepharose step was replaced by a size-exclusion chromatography purification step (Superose 6). Coomassie-stained polyacrylamide gels of the final purified products of Sld3, GST-Sld3, Cdc45, GST-Cdc45, and GINS are shown in supplemental Fig. 1, and we estimated the purity of each sample to be ϳ99%.
Analytical Size Exclusion Chromatography-Radiolabeled protein was mixed with unlabeled proteins as described in the figure legends in a final volume of 200 l and incubated for 1 h at 30°C in column buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, and 5% glycerol). The protein mixture was then subjected to 24 ml of Superose 6 (GE Healthcare) size exclusion chromatography, in the same column buffer. Each 250-l fraction was then subjected to SDS-PAGE analysis and quantified using PhosphorImager software.
GST Pulldown Assays-The 100-l GST pulldown reaction contained GST-tagged protein in GST-binding buffer (40 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM EDTA, 10% glycerol, 0.1% Triton X-100, 1 mM DTT, 0.7 g/ml pepstatin, 0.1 mM PMSF, and 0.1 mg/ml BSA), and varying amounts of radiolabeled protein as described in each figure. Reactions were incubated at room temperature for 1 h. Following incubation, reactions were added to 40 l of prepared glutathione-Sepharose and mixed gently. Binding of GST-tagged protein to the beads was performed for 20 min, with gentle mixing every few minutes. Once the binding was complete, the reaction mixture was aspirated, and the beads were washed two times with 0.5 ml GST-binding buffer. After the last wash, 30 l of 5ϫ SDS sample buffer were added to each reaction, and the samples were boiled for 10 min. Samples (20 l) were then analyzed by SDS-PAGE.
Biotin Pulldown Assay-Biotinylated DNA conjugated to streptavidin-agarose magnetic beads (Dynal) were incubated 5 min at 30°C with increasing concentrations of radiolabeled protein in a solution containing 0.1 mM EDTA, 0.2 mM DTT, 10 mM magnesium acetate, 10% glycerol, 40 g/ml BSA, 100 mM NaCl, and 20 mM Tris-HCl, pH 7.5, in a final volume of 25 l. Following the 5-min incubation, the beads were collected at room temperature using a magnet (Dynal). The supernatant was removed, and the beads were washed twice with a solution containing 0.1 mM EDTA, 0.2 mM DTT, 10 mM magnesium acetate, 10% glycerol, 40 g/ml BSA, 100 mM NaCl, and 20 mM Tris-HCl, pH 7.5. The beads were collected with a magnet, the supernatant was removed, and the beads were boiled at 95°C for 10 min in a solution containing 2% SDS, 2 mM DTT, 4% glycerol, 4 mM Tris-HCl, and 0.01% bromphenol blue. The reactions were analyzed by SDS-PAGE. The gel was dried for 1 h at 80°C and exposed to a phosphorimaging screen for 1 h.
Fluorescence Anisotropy-Sld3 binding to DNA was measured by the increase in anisotropy as varying concentrations of Sld3 were incubated with 10 nM fluorescent DNA (5-FAM) for 5 min in a solution containing 0.1 mM EDTA, 0.2 mM DTT, 10 mM magnesium acetate, 10% glycerol, 40 g/ml BSA, 100 mM NaCl, and 20 mM Tris-HCl, pH 7.5, in a final volume of 40 l. Polarized fluorescence intensities were measured at excitation and emission wavelengths of 495 and 538 nm, respectively. The data were plotted as fraction bound versus protein concentration and dissociation constants (K d ) were derived by fitting the data with the equation, y ϭ m 0 /(m 0 ϩ m 1 ).

GST-Sld3
Binds to ARS1 Origin ssDNA-Sld3 binds to early origins in G 1 phase in budding yeast (18,23). The proximity of Sld3 to origin DNA prompted us to investigate whether Sld3 binds directly to origin DNA. A schematic of the ARS1 origin is shown in Fig. 1A. The ARS1 origin contains several elements that are important for the initiation of DNA replication: the A, B1, B2, and B3 elements (34,35). The A element contains the ARS consensus sequence and is a binding site for the origin recognition complex. The B1, B2, and B3 elements encompass a readily unwound region positioned adjacent to the origin recognition complex binding site where the Mcm proteins bind (36 -38). We divided the ARS1 origin into three segments: one encompassing the A and B1 sites, one encompassing the B2 and B3 sites, and one positioned adjacent to the B2 and B3 sites (Fig. 1A).
We purified Sld3 bearing a GST tag to homogeneity and studied its interaction with radiolabeled ssDNA and dsDNA from ARS1 using a GST pulldown assay. We analyzed any bound, radiolabeled DNA by denaturing gel (Fig. 1B). We then quantified the bound DNA with phosphorimaging and plotted the resulting average of similar experiments (Fig. 1C). GST-Sld3 pulls down the majority of ssARS1-2, the sequence of ARS1 encompasses the A and B1 regions. However, GST-Sld3 does not pull down the majority of ssARS1-1, the complementary strand of ssARS1-2, suggesting that GST-Sld3 binding to DNA is strand-specific. Furthermore, GST-Sld3 binds weakly to double-stranded DNA encompassing this same region (dsARS1-12), suggesting that Sld3 binding to DNA is specific for the single-stranded structure. GST-Sld3 also does not pull down the majority of ssARS1-3, ssARS1-4, or dsARS1-34, the single and double DNA strands encompassing the B2 and B3 regions, suggesting that GST-Sld3 binding to ARS1 is region-specific. GST does not pull down any of the input radiolabeled sequences, suggesting that interaction between GST-Sld3 and origin ssDNA depends upon Sld3.
We also studied the interaction of GST-Sld3 with the adjacent complementary single-stranded DNA sequences of ARS1 (ssARS1-5 and ssARS1-6) and found that Sld3 pulls down more than half of the input for one of these sequences (ssARS1-5), FIGURE 1. Sld3 binds to origin ssDNA (ARS1). A, schematic of the origin ARS1. B, 5 pmol of GST-Sld3 or GST was incubated with varying amounts (0.09, 0.28, 0.90, or 2.8 pmol) of radiolabeled ssDNA or dsDNA from ARS1 as described under "Experimental Procedures." The DNA sequences are in Table 1. C, results from experiments shown in B were quantified, averaged, and plotted. The data are mean Ϯ S.E.
whereas it very weakly interacts with the complementary sequence (ssARS1-6). These data suggest that Sld3 binds to two distinct regions of ARS1 (ssARS1-2 and ssARS1-5). GST-Sld3 binds very weakly to the double-stranded DNA encompassing this region (dsARS1-56), consistent with the preference of Sld3 for single-stranded DNA.
CDK-phosphorylated Sld3 with No GST Tag Binds to ARS1 Origin ssDNA-To confirm that Sld3 binds to origin ssDNA, we next utilized a different experimental technique (Fig. 2). We purified Sld3 with no GST tag to examine whether Sld3 with no GST-tag can bind to origin ssDNA. We then radiolabeled Sld3 with either PKA or CDK and pulled down the radiolabeled Sld3 with biotinylated DNA. The bound protein was then analyzed by SDS-PAGE and quantified with phosphorimaging. The PKA phosphorylation is not physiologic; the PKA tag is used to radiolabel the protein only. CDK phosphorylation is physiologic, and CDK phosphorylation of Sld3 during S phase is important for the initiation of DNA replication (24,25). The DNA sequences used were ssARS1-2, ssARS1-4, and dsARS1-12.
At 1 or 3 pmol input of radiolabeled PKA-Sld3, biotinylated ssARS1-2 pulled down more than half of the input Sld3 protein (Fig. 2). These data suggest that Sld3 with no GST tag can bind to origin ssDNA. Furthermore, for 1 or 3 pmol of input CDKradiolabeled Sld3, biotinylated ssARS1-2 also pulled down more than half of the input protein. These data show that the CDK-phosphorylated Sld3 also binds to origin DNA, suggesting that the activity is relevant for S phase, when DNA replication is initiated. However, biotinylated ssARS1-4 or dsARS1-12 did not pull down a high fraction of input PKA or CDK-radiolabeled Sld3, confirming that Sld3 interaction with origin DNA is region-and structure-specific.
Sld3 with No GST Tag Binds to ARS1 Origin ssDNA in Solution with High Affinity (K d ϭ 53 nM ϩ 4 nM)-To determine the solution affinity of Sld3 for origin DNA, we employed the technique of fluorescence anisotropy (Fig. 3). Fluorescently labeled origin DNA was incubated with varying concentrations of Sld3, and the anisotropy change was measured. The Sld3 in these experiments did not contain a GST tag. The data were fit to a single site binding equation, and the apparent K d was calculated to be 53 nM ϩ 4 nM. In contrast, the binding of Sld3 to ssARS1-4 or dsARS1-12 was weak (apparent K d Ͼ 200 nM). These data confirm that binding of Sld3 to ARS1 origin DNA is region-and structure-specific.
Replacement of Origin Thymines with Adenines Completely Abolishes Sld3 Binding-ssARS1-2 is adenine-rich (41% adenine), and we speculated that Sld3 may be binding specifically to the adenines of ssARS1-2 (Fig. 4). We therefore replaced all of the adenines of ssARS1-2 with cytosines (A to C) and measured binding to GST-Sld3 with the GST pulldown assay (Fig. 4A). Surprisingly, the modified DNA (A to C) bound to Sld3 with  wild-type affinity. We therefore tested replacement of thymines with adenines (T to A), guanines with cytosines (G to C), or cytosines with adenines (C to A). Whereas the G-to-C or C-to-A substitutions resulted in no change in affinity of GST-Sld3, the T-to-A substitution completely disrupted binding to GST-Sld3 (Fig. 4A). These data suggest that Sld3 binds specifically to the thymines of ssARS1-2.
There are 21 thymines in the 80-base sequence ssARS1-2. To determine whether Sld3 binds to a specific region of ssARS1-2, we replaced regions of ssARS1-2 with T-to-A substitutions. Replacement of seven or 14 thymines at the 3Ј end of ssARS1-2 had little effect on GST-Sld3 binding (Fig. 4B). However, replacement of seven thymines at the 5Ј end of ssARS1-2 reduced binding of GST-Sld3 to GST background levels (Fig.  4C). These data suggest that Sld3 binds to the seven 5Ј-thymines of ssARS1-2. These seven thymines are positioned in the region between that B1 and B2 elements of ARS1.
GST-Sld3 Binds to ARS305 Origin ssDNA-We next determined whether Sld3 can bind to a second origin of replication, ARS305 (Fig. 5, A and B). Complementary single-stranded DNA and double-stranded DNA were analyzed for binding to GST-Sld3. GST-Sld3 pulled down more than half of input radiolabeled ssARS305-1; however, GST-Sld3 does not pull down a large fraction of either ssARS305-2 or dsARS305-12. These data suggest that Sld3 binds to the single strands of two different origins of replication tested (ARS1 and ARS305). Fur- thermore, the data suggest that Sld3 binding to origin DNA is specific for single-stranded DNA.
We next addressed whether Sld3 contains two different binding sites for ssDNA: one site for ssARS1-2 and a second site for ssARS305-1, or whether ssARS1-2 and ssARS305-1 compete with one another for the same Sld3-binding site. GST-Sld3 was used to pull down radiolabeled ssARS1-2 in the presence of increasing amounts of unlabeled ssARS305-1 (Fig. 5, C and D). As the amount of ssARS305-1 was increased, there was a progressive decrease in the amount of ssARS1-2 bound to GST-Sld3. These data suggest that ssARS1-2 and ssARS305-1 compete for the same Sld3-binding site. Sld3 binds to the seven 5Ј-thymines of ssARS1-2. ssARS305-1 is thymine-rich (64% thymine, see Table 1), and the thymines present in the ssARS305-1 sequence may provide some of the binding specificity for Sld3.
Origin ssDNA (ssARS1-2) Specifically Disrupts Interaction between Sld3 and Mcm2-7-Sld3 binds to Dpb11 in vivo when it is phosphorylated by CDK (24,25). We phosphorylated Sld3 with CDK and tested its interaction with radiolabeled Dpb11 using a GST pulldown assay. We found that approximately onethird of Dpb11 bound to CDK-treated GST-Sld3 (Fig. 6, A and  B). In contrast, very little Dpb11 bound to mock-treated GST-Sld3 (Fig. 6, A and B). We next studied whether ssARS1-2 affects the interaction between Sld3 and GST-Dpb11 (Fig. 6, C  and D). We incubated CDK-treated-GST-Sld3 with radiolabeled Dpb11 and varying concentrations of ssARS1-2. When the ratio of ssARS1-2 to Dpb11 was 3 or 10, there was no effect on radiolabeled Dpb11 interaction with CDK-treated GST-Sld3. These data suggest that origin ssDNA does not influence the interaction between CDK-treated Sld3 and Dpb11.
Sld3 binds to Cdc45 in vivo during different phases of the cell cycle (18). Sld3 also binds to Cdc45 in vitro (Fig. 6, E and F) (39). We next investigated whether ssARS1-2 affects the interaction between Sld3 and Cdc45. GST-Cdc45 was used to pull-down radiolabeled PKA-Sld3 or CDK-Sld3. At molar ratios of DNA to Sld3 of 1, 3, or 10, there was little to no effect on Cdc45 interaction with PKA-Sld3 or CDK-Sld3 (Fig. 6, E and F). These data suggest that ssARS1-2 does not influence the interaction between Sld3 and Cdc45.
Sld3 also binds directly to Mcm2-7 (Fig. 6G) (39). We next investigated whether ssARS1-2 affects the interaction between Sld3 and Mcm2-7. GST-Sld3 was used to pull down radiolabeled Mcm2-7 in the absence of presence of 5 mM ATP or the nonhydrolyzable analog ATP-␥S, and ssARS1-2 was added in varying amounts to the pulldown reaction. ssARS1-2 inhibited the interaction between GST-Sld3 and Mcm2-7 in a concentration-dependent manner (Fig. 6, G and H). The inhibition was similar in the absence or presence of 5 mM ATP or 5 mM ATP-␥S, suggesting that ssARS1-2 disruption of Sld3-Mcm2-7 binding is independent of the nucleotide binding state of Mcm2-7.
Origin Single-stranded DNA (ssARS1-2) Releases Sld3 from Mcm2-7, Allowing GINS to Bind Mcm2-7-We next examined whether ssARS1-2 affects the interaction between GST-Mcm2-7 and GINS (Fig. 7, A and B). GST-Mcm2-7 pulls down the majority of input radiolabeled GINS in the absence or presence of 5 mM ATP. As ssARS1-2 is added to the GST pulldown at ssARS1-2 to Mcm2-7 ratios of 1, 3, 10, or 30, there is little effect on the interaction between GST-Mcm2-7 and radiolabeled GINS. These data suggest that ssARS1-2 does not disrupt the interaction between GINS and Mcm2-7 in the absence or presence of ATP.
We next examined whether ssARS1-2 affects the interaction between GST-Mcm2-7 and radiolabeled GINS in the presence of Sld3. When Sld3 is added to a GST-Mcm2-7 pulldown of GINS, Sld3 inhibits the interaction between GINS and Mcm2-7 (Fig. 7, C and D) (39). When ssARS1-2 is added to the GST-Mcm2-7 pulldown of radiolabeled GINS in the presence of competing Sld3, ssARS1-2 promotes the interaction between GINS and Mcm2-7 in a concentration-dependent manner (Fig.  7, C and D). This effect is similar in the presence or absence of 5 mM ATP. As a control, Sld3 does not bind directly to GINS (Fig.  8, A and B) (39). Cdc45 and Sld3 bind together in vitro (39) and in vivo (18), and thus, we also performed these experiments in the presence of Cdc45. When ssARS1-2 is added to the GST-Mcm2-7 pulldown of radiolabeled GINS in the presence of Cdc45 and Sld3, ssARS1-2 stimulates the interaction between GINS and Mcm2-7 in a concentration-dependent manner (Fig.  7, E and F). These data suggest that Cdc45 does not affect the ability of ssARS1-2 to stimulate GINS binding to Mcm2-7 in the presence of competing Sld3. All of the data discussed thus far indicate that ssARS1-2 releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7.
We next used a solution-based assay, size exclusion chromatography, to determine whether ssARS1-2 promotes the interaction between GINS and Mcm2-7 in the presence of Sld3. First, we conducted a number of control experiments. We studied the interaction between Sld3 and GINS and found no interaction in the absence or presence of ssARS1-2 (Fig. 8, A and B). We also found that ssARS1-2 does not effect the migration of Mcm2-7, as expected for a low molecular weight DNA species (Fig. 8C). We next found that Sld3 can form an interaction with Mcm2-7 with 1:1 stoichiometry in the absence of DNA and that GINS can form an interaction with Mcm2-7 with 1:1 stoichiometry in the absence of DNA (Fig. 8, D-F) (39). However, in the presence of ssARS1-2, the interaction between Sld3 and Mcm2-7 is disrupted (Fig. 8, D and F), consistent with the GST pulldown results (Fig. 6, G and H). In contrast, ssARS1-2 does not directly affect the interaction between GINS and Mcm2-7 in the absence of Sld3 (Fig. 8, E and F), also consistent with the GST pulldown results (Fig. 7, A and B). We next studied a mixture of Sld3, GINS, and Mcm2-7 (Fig.  8, G-I). First, Sld3 was radiolabeled and incubated with unlabeled GINS and Mcm2-7 in the absence of ssARS1-2 (Fig. 8G). In the absence of ssARS1-2, two peaks of roughly equal size were observed, with one peak at an elution of ϳ720 kDa and a second peak at an elution of ϳ68 kDa. Because Sld3 does not bind directly to GINS, the data indicate that roughly half of the Sld3 is forming a complex with Mcm2-7 (high molecular mass peak), whereas the other half of Sld3 is free in solution (low molecular mass peak). We next radiolabeled GINS and incubated it with Sld3 and Mcm2-7 under identical conditions (Fig.  8H). Once again there are two peaks of roughly equal size in the absence of DNA, one of high molecular mass and one of low molecular mass. The high molecular mass peak is consistent FIGURE 6. Origin ssDNA (ssARS1-2) dislodges Sld3 from Mcm2-7. ssARS1-2 does not disrupt the interaction between Sld3 and Dpb11 or between Sld3 and Cdc45. A, 2 pmol of CDK-treated or mock-treated (i.e. no kinase, buffer added only) GST-Sld3 or GST was incubated with varying amounts of PKA-radiolabeled Dpb11 (0.06, 0.2, 0.6, or 2 pmol). Bound radioactive Dpb11 was analyzed by SDS-PAGE. B, results from experiments shown in A were quantified, averaged, and plotted. C, 2 pmol of CDK-treated GST-Sld3 was incubated with 2 pmol of radiolabeled Dpb11 and varying amounts of ssARS1-2 (0, 6, or 20 pmol). Bound radioactive Dpb11 was analyzed by SDS-PAGE. D, results from experiments shown in C were quantified, averaged, and plotted. E, Sld3 containing an N-terminal PKA tag was radiolabeled with either PKA or CDK. PKA-Sld3 and CDK-Sld3 were then matched for radioactive counts and total Sld3 concentration. 2 pmol of PKA-Sld3 or CDK-Sld3 were then incubated with 2 pmol of GST-Cdc45 and varying concentrations of ssARS1-2 (0, 2, 6, or 20 pmol), and the bound radioactive Sld3 was analyzed by SDS-PAGE. F, results from experiments shown in E were quantified, averaged, and plotted. G, 2 pmol of radiolabeled Mcm2-7 in the presence or absence of 5 mM ATP or ATP-␥S was incubated with 2 pmol of GST-Sld3 and varying concentrations of ssARS1-2 (0, 1, 2, 4, or 8 pmol). H, results from experiments shown in G were quantified, averaged, and plotted. The data are mean Ϯ S.E.
with a GINS-Mcm2-7 peak, whereas the low molecular mass peak is consistent with free GINS. Finally, we radiolabeled Mcm2-7 and incubated it with Sld3 and GINS under identical conditions (Fig. 8I). A single, high molecular mass peak is observed in the absence of DNA, which is consistent with a roughly equimolar mixture of Mcm2-7-GINS and Mcm2-7-Sld3 complexes.
We next tested the hypothesis that ssARS1-2 releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7. We radiolabeled Sld3 and incubated it with GINS and Mcm2-7 in the presence of ssARS1-2 (Fig. 8G). In the presence of ssARS1-2, there is one predominant low molecular mass peak, which is consistent with free Sld3. These data suggest that ssARS1-2 disrupts the interaction between Sld3 and Mcm2-7. We next radiolabeled GINS and incubated it with Sld3 and Mcm2-7 in the presence of ssARS1-2 under identical conditions (Fig. 8H).
In the presence of ssARS1-2, there is a predominant high molecular mass peak consistent with a complex of GINS and Mcm2-7. These data suggest that ssARS1-2 promotes the interaction between Mcm2-7 and GINS in the presence of Sld3. Finally, we radiolabeled Mcm2-7 and incubated it with GINS and Sld3 in the presence of ssARS1-2 under identical conditions. In the presence of DNA a single high molecular mass peak is observed, consistent with a GINS-Mcm2-7 complex (Fig. 8I). All of the data discussed thus far indicate that ssARS1-2 releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7.
We previously found that combining equimolar Sld3, Cdc45, Mcm2-7, and GINS results in a roughly equimolar mixture of CMS and CMG complexes (39). We next tested whether the presence of ssARS1-2 will promote the formation of the CMG complex when Sld3, Cdc45, Mcm2-7, and GINS are mixed together (Fig. 9). The data in Fig. 9 suggest that when equimolar Sld3, Cdc45, Mcm2-7, and GINS are mixed in the absence of DNA, there is a roughly equimolar mixture of CMS and CMG complexes, as found previously (Fig. 9, A-D, filled triangles) (39). However, when ssARS1-2 is added to this mixture, the CMS complex is disrupted, whereas formation of the CMG complex is promoted (Fig. 9, A-D, filled circles). These data suggest that origin ssDNA may help promote the formation of the CMG complex in vivo.

DISCUSSION
Sld3 is required for the initiation of DNA replication, but Sld3 does not travel with the replication fork (18,21). We found that Sld3 binds tightly to the single-stranded region of the ARS1 origin encompassing the A and B1 elements (ssARS1-2, K d ϭ 53 nM). Sld3 also binds to ssDNA positioned adjacent to the B2 and B3 elements (ssARS1-5), but Sld3 binds weakly to doublestranded ARS1 DNA. CDK-phosphorylated Sld3 binds to ssARS1-2 with similar high affinity as non-CDK-phosphorylated Sld3, suggesting that Sld3 is active in binding origin ssDNA during S phase. Sld3 also binds tightly to the singlestranded DNA of the ARS305 replication origin, but Sld3 does binds weakly to double-stranded ARS305. Replacement of the seven 5Ј-thymines in ssARS1-2 with adenines reduces Sld3 binding to background levels, suggesting that Sld3 binds to the seven 5Ј thymines of ssARS1-2. These thymines are positioned between the B1 and B2 elements of ARS1. ssARS1-2 specifically inhibits the interaction between Sld3 and Mcm2-7, but ssARS1-2 does not inhibit the interaction between Sld3 and Dpb11 or between Sld3 and Cdc45. In a mixture of Sld3, GINS, and Mcm2-7, ssARS1-2 inhibits the interaction between Sld3 and Mcm2-7, whereas ssARS1-2 promotes the association between GINS and Mcm2-7. We also found that ssARS1-2 promotes the formation of the CMG complex. We conclude that ssARS1-2 releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7.
CDK-phosphorylated Sld2 and Sld3 Bind to Single-stranded Origin DNA with High Affinity-In a previous study, we found that CDK-phosphorylated Sld2 binds to origin single-stranded DNA with high affinity (26). CDK-phosphorylated Sld2 binds to ssARS1-1, the complementary strand to ssARS1-2 (26). These data suggest that Sld2 and Sld3 may bind to complementary single-stranded DNA strands at an origin of replication  (Fig. 10). The region of DNA adjacent to the origin recognition complex binding site is AT-rich and prone to spontaneous melting (37,38). Furthermore, the Mcm proteins initiate unwinding in the neighboring region of DNA (36,40). Moreover, the CMG complex binds to ssDNA, not dsDNA, and the CMG complex unwinds DNA by a steric exclusion mechanism (13).
It is interesting that mutation of seven 5Ј-thymines of ssARS1-2 to adenines abolishes binding of Sld3 to DNA. Similarly, mutation of the thymines of ssARS1-1 to adenines abolishes binding of Sld2 to origin ssDNA (26). Origin sequences are AT-rich (34,36,37), and the preference of binding of Sld2 and Sld3 to thymines may reflect a mechanism to ensure that Sld2 and Sld3 will bind to the AT-rich regions of origin DNA. Previously, we found that Sld2 and Dpb11 stimulate the anneal-ing of ssARS1-1 to ssARS1-2 (26). However, if Sld3 is bound to ssARS1-2, then this annealing activity may be inhibited in vivo.
Role of Sld3 in Initiation of DNA Replication-Sld3 and GINS compete with one another to bind Mcm2-7, and ssARS1-2 influences this competition. When Sld3 and GINS compete for Mcm2-7 binding in the presence of ssARS1-2, ssARS1-2 binds to Sld3. The interaction between Sld3 and ssARS1-2 releases Sld3 from Mcm2-7, allowing GINS to bind Mcm2-7 (Fig. 10). Thus, in a mixture of GINS, Sld3, and Mcm2-7, ssARS1-2 inhibits the interaction between Sld3 and Mcm2-7, whereas ssARS1-2 promotes the interaction between GINS and Mcm2-7. The interaction between GINS and Mcm2-7 is critical for activation of the replication fork helicase because the replication fork helicase in eukaryotes is composed of Cdc45-Mcm2-7-GINS (CMG complex) (12)(13)(14). Thus, one role of The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." The radioactive counts in each fraction were used to calculate the pmol of Sld3 in each fraction. Sld3 (pmol) was then plotted versus the elution of molecular weight standards. B, E, and H, 100 pmol of 32 P-labeled GINS was incubated in the absence or presence of 100 pmol of Sld3, 100 pmol of Mcm2-7, or 200 pmol of ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." C, F, and I, 100 pmol of 32 P-labeled Mcm2-7 was incubated in the absence or presence or 100 pmol of Sld3, 100 pmol of GINS, or 200 of pmol ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." The data are mean Ϯ S.E.
Sld3 is to block the interaction between GINS and Mcm2-7. However, when single-stranded DNA is exposed during S phase, Sld3 releases its grip on Mcm2-7, allowing GINS to bind to Mcm2-7. Treslin and Ticrr have recently been identified as possible Sld3 homologs in higher eukaryotes (41,42). Future work may reveal whether Treslin or Ticrr bind directly to origin ssDNA or whether some other protein fulfills this function in humans.
What Releases Sld3 from Cdc45?-The Sld3-Cdc45 interaction may be disrupted by some mechanism in vivo as Sld3 does not travel with the replication fork, whereas Cdc45 does travel with the replication fork (21). Our data suggest that Cdc45 is more stable in a ternary complex with GINS and Mcm2-7 than in a binary complex with free Sld3 (Fig. 9). Thus, the formation of the CMG complex may itself promote the eventual release of Sld3 from Cdc45. Furthermore, excess GINS (6-fold excess rel-ative to Sld3) will substantially disrupt (Ͼ80%) the interaction between Sld3 and Cdc45 in vitro (39). There are ϳ1430 molecules of Psf1 and ϳ2210 molecules of Psf3 in the cell, in comparison with ϳ125 molecules of Sld3 in the cell (43). The excess of GINS compared with Sld3 in the cell may be sufficient to disrupt the interaction between Sld3 and Cdc45 in vivo. Thus, the formation of the CMG complex may promote the eventual release of Sld3 from Cdc45, or excess cellular GINS may be sufficient to disrupt the interaction between Sld3 and Cdc45 in vivo. On the other hand, there are ϳ1730 molecules of Cdc45 in the cell, compared with ϳ125 molecules of Sld3 (43). Because Cdc45 is in excess compared with Sld3, the disruption of the Sld3-Cdc45 interaction may not be critical for origin firing.
Model for Initiation of DNA Replication (Fig. 10)-The origin recognition complex, Cdc6, and Cdt1 load sister Mcm2-7 hexameric rings around double-stranded DNA at an origin of FIGURE 9. Origin single-stranded DNA (ssARS1-2) promotes the formation of the CMG complex. A, 100 pmol of 32 P-labeled Sld3 was incubated with 100 pmol of Cdc45 and 100 pmol of Mcm2-7 in the absence or presence of 100 pmol of GINS, and/or 200 pmol of ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." The radioactive counts in each fraction were used to calculate the pmol of Sld3 in each fraction. Sld3 (pmol) was then plotted versus the elution of molecular weight standards. B, 100 pmol of 32 P-labeled Cdc45 was incubated with 100 pmol of Sld3 and 100 pmol of Mcm2-7 in the absence or presence of 100 pmol of GINS, and/or 200 pmol of ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." C, 100 pmol of 32 P-labeled Mcm2-7 was incubated with 100 pmol of Sld3 and 100 pmol of Cdc45 in the absence or presence of 100 pmol of GINS, and/or 200 pmol of ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." D, 100 pmol of 32 P-labeled GINS was incubated with 100 pmol of Cdc45 and 100 pmol of Mcm2-7 in the absence or presence of 100 pmol of Sld3, and/or 200 pmol of ssARS1-2. The mixture was then subjected to size-exclusion chromatography as described under "Experimental Procedures." The data are mean Ϯ S.E.