GINS Is a DNA Polymerase ϵ Accessory Factor during Chromosomal DNA Replication in Budding Yeast*

GINS is a protein complex found in eukaryotic cells that is composed of Sld5p, Psf1p, Psf2p, and Psf3p. GINS polypeptides are highly conserved in eukaryotes, and the GINS complex is required for chromosomal DNA replication in yeasts and Xenopus egg. This study reports purification and biochemical characterization of GINS from Saccharomyces cerevisiae. The results presented here demonstrate that GINS forms a 1:1 complex with DNA polymerase ϵ (Pol ϵ) holoenzyme and greatly stimulates its catalytic activity in vitro. In the presence of GINS, Pol ϵ is more processive and dissociates more readily from replicated DNA, while under identical conditions, proliferating cell nuclear antigen slightly stimulates Pol ϵ in vitro. These results strongly suggest that GINS is a Pol ϵ accessory protein during chromosomal DNA replication in budding yeast. Based on these results, we propose a model for molecular dynamics at eukaryotic chromosomal replication fork.

DNA primase initiates DNA replication by synthesizing a short oligoribonucleotides primer, which is immediately elongated by Pol ␣ to form short RNA-DNA fragments on both leading and lagging strand of DNA. Pol ␦ elongates the short RNA-DNA fragment initiated by Pol ␣-primase to make a mature Okazaki fragment (2,10). To carry out processive DNA synthesis in vitro, Pol ␦ requires PCNA and its loader, replication factor C (RF-C) (10). In cooperation with Fen1p (Rad27p), Dna2p and RPA, Pol ␦ also plays a crucial role in processing RNA-linked Okazaki fragments (10).
Although the precise role of Pol ⑀ in vivo is still unclear, it has been implicated in DNA replication, repair, recombination, and mitosis (1,2). Pol ⑀ has at least one essential function in both budding and fission yeasts (11,12), and several lines of evidence suggest that Pol ⑀ plays an essential catalytic role during chromosomal DNA replication. Yeast cells harboring a temperature-sensitive pol2 allele are temperature sensitive for growth and express a thermolabile Pol ⑀ DNA polymerase activity (13). Furthermore, Pol2p is associated with replication forks during S phase (14) and pol2 mutants fail to complete chromosomal DNA replication (13,15). Furthermore, 3Ј-5Ј exonuclease-deficient mutants of POL2 and POL3 (Pol ␦) accumulate strand-specific lesions in chromosomal DNA (16,17,18,19). These observations support models for chromosomal DNA replication in which Pol ⑀ and Pol ␦ play leading strandand lagging strand-specific roles during chromosomal DNA replication, respectively. Pol ⑀ has been proposed as the leading strand DNA polymerase because Pol ⑀ is a highly processive polymerase without PCNA (1,8) and pol3 mutants have defects in maturation of Okazaki fragments (10). Nevertheless, it has been reported that the N-terminal portion of budding yeast Pol ⑀ (Pol2p), which includes motifs required for DNA polymerase and exonuclease activities, is dispensable for DNA replication, DNA repair, and viability (20,21). However, this conclusion is controversial, because other studies suggest that deletion of the N-terminal region of Pol ⑀ confers temperature sensitivity for growth, a defect in DNA elongation, premature senescence, and short telomeres. Furthermore, this pol2p deletion is lethal in combination with temperature-sensitive cdc2 and with exonucleasedeficient Pol ␦ (pol3-01). These results suggest that Pol ⑀ plays a crucial role in maintaining genomic integrity (22,23).
In a previous study, various sld (synthetic lethality with dpb11-1) mutants (24) were identified as yeast mutants that are lethal in combination with temperature-sensitive dpb11-1 (25).
DPB11 encodes Dpb11p, a yeast protein that interacts with Pol ⑀ that is required for initiation of chromosomal DNA replication. Sld1p is identical to Dpb3p, the third subunit of Pol ⑀. Sld2p binds to Dpb11p, and the Sld2p⅐Dpb11p heterodimer facilitates loading of Pol ␣-primase and Pol ⑀ onto replication origins during S phase (26). Sld2p is phosphorylated by S-phase cyclin-dependent protein kinase at the beginning of S phase, and is required for initiation of chromosomal DNA replication (27). Sld3p interacts with Sld4p, which is identical to Cdc45p. Previous studies show that Cdc45p is required for initiation and elongation of chromosomal DNA replication (reviewed in Ref. 2). It has been reported that Sld3 is also important for the progression of DNA replication forks after the initiation step (28), as are Cdc45. In contrast, it does not move with DNA replication forks and only associates with MCM in an unstable manner before initiation. After the establishment of DNA replication forks from early origins, it is no longer essential for the completion of chromosome replication (29). Slp5p is a component of GINS, which also includes Psf1p, Psf2p and Psf3p. GINS is essential for chromosomal DNA replication in yeast (30,31). Finally, Sld6p is identical to Rad53p, which is required for cell cycle checkpoints (32). GINS associates with replication origins during S phase (30,31). It has been suggested that GINS promotes an interaction between the Sld3p⅐Cdc45p complex and the Sld2⅐Dpb11⅐Pol ⑀, and that this could facilitate initiation of DNA replication (27). GINS polypeptides are conserved in eukaryotic cells. Xenopus egg extracts have a complex that resembles GINS, which is also required for the initiation of chromosomal DNA replication (33). Electron microscopy of the Xenopus GINS complex suggests that GINS forms a ring structure that resembles a trimeric PCNA, which is a Pol ␦ clamp. These data are consistent with the idea that GINS is a clamp for Pol ⑀ (33).
This study reports purification and characterization of GINS from yeast cell extracts. GINS does not bind DNA directly, but it forms a 1:1 complex with Pol ⑀ and greatly stimulates its catalytic activity in vitro. In the presence of GINS, Pol ⑀ is more processive and dissociates more readily from replicated DNA. These results suggest that GINS is an accessory protein for Pol ⑀ during chromosomal DNA replication in yeast.
To cell lysate, 0.3 M KCl (final concentration) was added, and incubation was continued at 4°C for 30 min with stirring, followed by centrifugation at 10,000 ϫ g for 30 min at 4°C. The resultant lysate was further supplemented with Triton X-100 and 10% polyethyleneimine solution to give a final concentration of 0.01% (v/v) and 0.4% (v/v), respectively, incubated at 4°C for 30 min, and centrifuged at 10,000 ϫ g for 30 min at 4°C. The resultant pellets were dissolved in buffer A (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 0.01% Triton X-100, and 10 mM 2-mercaptoethanol) containing 1% protease inhibitor mixture and 0.5 M ammonium sulfate, incubated at 4°C for 30 min, and centrifuged at 10,000 ϫ g for 30 min at 4°C. The supernatant, which contained more than 90% of GINS complex, was recovered, and protein was precipitated by addition of ammonium sulfate to a final concentration of 50% saturation. After incubating at 4°C for 30 min, the precipitated protein was collected by centrifugation at 10,000 ϫ g for 30 min at 4°C. The pellet was dissolved in buffer A containing 1% protease inhibitor mixture and 0.1 M NaCl, dialyzed against the same buffer and centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatant was applied to a Mono Q column (HR16/10, GE). After washing the column with buffer A containing 150 mM NaCl, proteins were eluted with a linear gradient from 150 to 600 mM NaCl in buffer A without 2-mercaptoethanol. 6ϫ FLAG-tagged Psf1p was eluted at 350 mM NaCl in Buffer A. Fractions containing 6ϫ FLAG-tagged Psf1p were pooled, supplemented with bovine serum albumin (BSA) to a final concentration of 5 mg/ml and with a mixture of protease inhibitors consisting of 4-(2-aminoethyle)-benzensulfonyl fluoride, aprotinin, benzamidine hydrochloride, leupeptin, and pepstatin A to final concentrations of 1 mM, 2 g/ml, 1 mM, 10 g/ml, and 1 g/ml, respectively, and loaded to a 400 l of anti-FLAG M2 agarose bead column. The column was washed three times with 10 bed volumes of buffer B (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 0.05% (v/v) Tween20, 0.005% (v/v) Nonidet P-40, 2 mM ␤-glycerophosphate, and 2 mM NaF) containing 300 mM NaCl and 0.1 mg/ml BSA, three times with 10 bed volumes of buffer B containing 300 mM NaCl, and once with 4 bed volumes of buffer B containing 150 mM NaCl and 10 g/ml 1ϫ FLAG peptide (Sigma). Bound proteins were eluted in 6 bed volumes of buffer B containing 300 mM NaCl and 100 g/ml 3ϫ FLAG peptide (Sigma). Eluted fractions were combined, applied to a Mono Q column (HR5/5, GE), the column was washed with 5 ml of 100 mM NaCl in Buffer A, and 6ϫ FLAG-tagged Psf1p was eluted with 10 ml of a linear gradient from 100 to 700 mM NaCl in buffer A. Fractions containing 6ϫ FLAG-Psf1p were pooled, dialyzed against buffer G (50% glycerol, 50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA, and 3 mM 2-mercaptoethanol(50 g/total 1 ml) and frozen in liquid nitrogen.
Immunoprecipitation of 6ϫ FLAG-Psf1p from Yeast Cell Extracts-6ϫ FLAG-tagged Psf1p was immunoprecipitated with anti-FLAG M2 agarose beads (Sigma) from the whole cell extracts prepared from asynchronously growing or hydroxyurea (HU)treated yeast YYK61 cells as previously described (24). The precipitates were boiled in SDS-sample buffer at 95°C for 30 min and subjected to SDS-PAGE, followed by immunoblotting.
Glycerol Gradient Sedimentation-Purified GINS and Pol ⑀ (molar ratio was 1:1) were incubated in 100 l of binding mixtures consisting of 50 mM HEPES-KOH pH 7.6, 5 mM MgOAc, 1 mM dithiothreitol, 0.01% Nonidet P-40, 0.1 mg/ml BSA at 4°C for 1 h. For cross-linking experiments, after the incubation, 1 l of 0.2 M dithiobis (succinimidylpropionate) (DSP) was added to a binding mixture, which was further incubated at 24°C for 30 min. The reaction was quenched by addition of 5 l of 1 M Tris-HCl, pH 7.5, and incubation at 24°C for 5 min. The mixtures were layered onto a 5 ml of 15-40% glycerol density gradient in buffer containing 50 mM HEPES-KOH (pH 7.6), 5 mM MgCl 2 , 1 mM dithiothreitol, 0.01% Nonidet P-40, and 50 mM KOAc and centrifuged in a Hitachi P55ST2 rotor at 45,000 rpm for 14 h at 2°C. The sample was fractionated into 23 fractions from top of the gradient and proteins in the fractions were separated by SDS-PAGE and detected by silver staining or immunoblotting.
Immunoblotting-6ϫ FLAG-tagged Psf1p was detected with anti-FLAG (Affinity BioReagents) or anti-Psf1p antibody. 9ϫ Myc-tagged Sld3p was detected with anti-Myc antibody (28). Anti-Psf1p-, -Dpb2p-, -Dpb11p-, -Cdc45p-, and -Sld2p rabbit antibodies were prepared using each protein, which was expressed in Escherichia coli and purified from E. coli. 3 Gel Mobility Shift Assay without Cross-linking-The GINS complex, Pol ⑀ and Cy5-labeled DNA fragments were incubated in reaction mixtures consisting of 20 mM HEPES-KOH (pH 7.6), 0.05% Nonidet P-40, 10% glycerol, and 0.1 mg/ml BSA on ice for 10 min. Ficoll was added to the samples at a final concentration of 15%, and the samples were run on a 4% polyacrylamide gel in TBE. For some experiments, MgOAc was added to the reaction mixture, and TBE at a final concentration of 5 mM. The sequences of the oligonucleotides used were shown in Table 1. Cy5-7 was used as single-stranded linear DNA (sslDNA). Cy5-7 and 71, Cy5-16, and 17, 21, and Cy5-22, Cy5-16 and 18, Cy5-19 and 20 were annealed for preparing a double-stranded linear DNA, 3Ј-overhang DNA, 5Ј-overhang DNA, Y-form DNA and bubble form DNA, respectively. The images were detected using a fluorescent image analyzer FLA-3000 (Fuji film).

TABLE 1 Sequences of the oligonucleotides used for the gel mobility shift assay
Cy5-GGTGAACCTGCAGGTGGGC a All sequences are listed in the 5Ј 3 3Ј direction.

GINS Is Part of the Yeast Replisome-Protein-interacting
partners of GINS were identified by immunoprecipitating 6ϫ FLAG-tagged Psf1p from logarithmically growing yeast cells with anti-FLAG antibody beads. Immunoprecipitates were prepared in the presence or absence of a protein cross-linking agent, and the coprecipitating proteins were analyzed by SDS-PAGE, followed by Western blot. In the absence of cross-linking agent, a small fraction of soluble Dpb2p (the second subunit of Pol ⑀) and Sld3p were detected in the immunoprecipitates (shown by arrows in Fig. 1A), but in the presence of a protein cross-linking reagent, significant amounts of Dpb2p, Dpb11p, phosphorylated Sld2p, Cdc45p (data not shown) and Sld3p coprecipitated with 6ϫ FLAG-Psf1p (Fig. 1A). Other DNA replication factors also coprecipitated with Psf1p including Mcm2p (a member of Mcm2-7p), PCNA, and RPA. The latter two proteins are believed to be involved in lagging strand synthesis catalyzed by Pol ␣-primase and Pol ␦ (2, 10). Yeast cells expressing 3ϫ FLAG-tagged Psf2p were also treated with 0.2 M HU-containing YPD for 2 h, collected, and used to isolate stable S phase-specific protein complexes in the absence of protein cross-linking. Under these conditions, Dpb2p, Mcm10p, Cdc45p, and Mcm2p co-immunoprecipitated with 3ϫ FLAGtagged Psf2p (Fig. 1B). These results suggest that Pol ⑀, Cdc45p, Mcm2p, and Mcm10p form a stable complex with GINS in vivo during S phase. Thus, these results are consistent with the notion that GINS is a part of the yeast replisome.
Purification and Characterization of GINS-The above results suggest that GINS may interact directly with Pol ⑀ and other replication proteins, although these interactions might be weak, transient, and/or salt-sensitive. To investigate this possibility directly, GINS was purified by affinity chromatography from yeast cells expressing 6ϫ FLAG-tagged Psf1p as described under "Experimental Procedures." Purified GINS consisted of four proteins of 35, 30, 25, and 22 kDa, respectively ( Fig. 2A), as previously reported (30). The purity of GINS was estimated to be Ͼ95%, and the protein subunits were approximately equimolar, although the third subunit of GINS (Psf2p) seemed to be a little more than other proteins (actual molar ratio between Sld5p, Psf1p, Psf2p, and Psf3p was 1:1:1.3:1). The identity of each GINS subunit was confirmed by immunoblotting purified GINS with anti-FLAG-, anti-Sld5p-, anti-Psf1p-, anti-Psf2p-, and anti-Psf3p antibodies ( Fig. 2A) and by MALDI-TOF mass spectrometry analysis of gel-purified polypeptides from purified GINS (data not shown). The results confirmed that purified GINS is composed of Sld5p, Psf1p, Psf2p, and Psf3p. Purified GINS did not have any detectable DNA polymerase-, endonuclease-, exonuclease-, DNA helicase-, ATPase-, ssDNA-binding-, or dsDNA binding activity (data not shown).
GINS Specifically Binds to Pol ⑀-The physical interaction between GINS and Pol ⑀ was examined using FLAG-tagged Psf1p and anti-FLAG antibody pull-down assays. When GINS and Pol ⑀ were incubated at 25°C for 30 min, anti-FLAG antibody beads coprecipitated FLAG-tagged Psf1p (GINS) and the Pol ⑀ including Pol2p, Dpb2p, Dpb3p, and Dpb4p (Fig. 3A). GINS appeared to bind to Pol2p⅐Dpb2p as efficiently as to Pol ⑀ holoenzyme (Fig. 3B), but bound much less efficiently to Dpb3p⅐Dpb4p (35) (Fig. 3C) and to the145 kDa-degradation product of the full size Pol2p (265 kDa), which has a DNA  (30) were incubated with (ϩ) or without (Ϫ) formaldehyde and whole cell extracts (WCS) were prepared as described previously (24). 6ϫ FLAG-tagged Psf1p was immunoprecipitated from the whole cell extracts with anti-FLAG beads. Proteins bound to the beads were released by boiling in SDS, and analyzed by SDS-PAGE, followed by immunoblotting with the indicated antibodies. WCS were also analyzed by SDS-PAGE, followed by immunoblotting with the indicated antibodies. In the figure, arrows indicate Sld3p and Dpb2p (a representative subunit of Pol ⑀ holoenzyme) bands that were immunoprecipitated with anti-FLAG beads. B, asynchronously growing cells expressing 3ϫ FLAGtagged Psf2p were treated with 0.2 M HU for 120 min at 30°C, collected by centrifugation, and WCS were prepared without any cross-linking reagent. 3ϫ FLAG-tagged Psf2p was immunoprecipitated with anti-FLAG antibody beads as above, and proteins associated with antibody were analyzed by immunoblotting as A. As a control (Ϫ), untagged yeast cells were grown, treated with 0.2 M HU for 120 min at 30°C, whole cell extracts were prepared and used for immunoprecipitation. Asterisk indicates a protein band that interacts nonspecifically with anti-Dpb11p-or anti-Sld2p antibodies.
polymerization activity as Pol ⑀ (34)(data not shown). However, GINS did not bind to other replicative polymerases, Pol ␦ or Pol ␣-primase complex (Fig. 3D and data not shown for Pol ␣-primase). These results suggest that GINS interacts specifically with the C-terminal-half portion of Pol2p (a catalytic subunit of Pol ⑀) and possibly with Dpb2p, which is the second subunit of S. cerevisiae Pol ⑀. The C-terminal-half portion of Pol2p, where other subunits (Dpb2p, Dpb3p, and Dpb4p) of Pol ⑀ interact and form Pol ⑀ holoenzyme (1), is known to be essential for yeast cell growth (20,21).
The interaction between GINS and Pol ⑀ was also examined by glycerol density gradient sedimentation in the presence or absence of a protein cross-linking agent. GINS and Pol ⑀ co-sedimented through the gradient, when the samples were preincubated with a cross-linking agent, but the complex dissociated spontaneously during sedimentation in the absence of a protein cross-linker (Fig. 4A). These results suggest that the interaction between GINS and Pol ⑀ is relatively weak and/or transient. In contrast, GINS and Pol ⑀ are stable as independent complexes in the absence of protein cross-linking during glycerol density gradient sedimentation (Fig. 4A). The glycerol density gradient data of their apparent molecular weights also suggests that GINS and Pol ⑀ exist as monomers in solution and that they form a 1:1 GINS⅐Pol ⑀ complex (Fig. 4B).
Purified GINS did not exhibit any DNA binding activity in the absence of Mg 2ϩ , where Pol ⑀ readily bound single-stranded-, double-stranded-, 3Ј-tailed doublestranded-, 5Ј-tailed double-stranded, Y-fork like-, and bubble-structure DNA (35) (Fig. 5, A and B), or in the presence of Mg 2ϩ (data not shown).
On the other hand, Pol ⑀⅐GINS significantly bound to 5Ј-tailed double-stranded DNA and caused a further gel shift from that of Pol ⑀ alone in the presence of Mg 2ϩ , although its shift was not enormous, but was very reproducible (Fig. 5C). When anti-FLAG monoclonal antibody was added into the reaction mixture containing GINS and Pol ⑀, a further gel shift was reversed to those of Pol ⑀ alone (Fig. 5C, far right lane). This may be due to disruption of GINS⅐Pol ⑀ complex by antibody binding to GINS. In any case, to our knowledge, this is the first case that binding of eukaryotic DNA polym-  erase and its clamp to a DNA substrate has been demonstrated by gel shift assay. GINS Stimulates DNA Synthesis Catalyzed by Pol ⑀ on a Oligonucleotide Template Primer Substrate-Because GINS specifically interacts with Pol ⑀ in vitro, it seemed possible that GINS has an effect on the catalytic activity of Pol ⑀. This idea was tested by measuring synthesis of DNA catalyzed by Pol ⑀ in the presence or absence of GINS using excess DNA substrate. For this experiment, the DNA substrate was a 65-mer oligonucleotide annealed to a 32 P-labeled 34-mer primer, which was converted to a 65-mer double-stranded DNA product by a DNA synthesis fill-in reaction. In the absence of GINS, although the efficiency was rather low, the fill-in reaction was completed in less than 1 min at 30°C, indicating that the rate of DNA synthesis was Ͼ0.5 n/sec (Fig. 6, A and B). However, the amount of a 65-mer product reached a plateau during the 3ϳ5 min (10 fmol and 35 fmol DNA were synthesized by 25 fmol and 50 fmol Pol ⑀, respectively), suggesting poor release and/or  recycling of Pol ⑀ from the double-stranded reaction product to the unreacted DNA substrate. In contrast, in the presence of GINS, release and recycling of Pol ⑀ appeared to greatly improve, and most of the DNA substrate used was converted to fully double-stranded DNA in 30 min at 30°C (300 fmol and 520 fmol of DNA were synthesized by 25 and 50 fmol of Pol ⑀, respectively) (Fig. 6, A and B). This stimulation result was also confirmed using DNA synthesis assay containing [␣-32 P]dNTPs and a 65-mer template annealed to an unlabeled 34-mer primer (supplemental Fig. S1).
At low Pol ⑀ concentration, GINS stimulated the reaction 15-20-fold, and the extent of the reaction was dependent on the concentration of Pol ⑀ (Fig. 6, A and B). The stimulation was also dependent on the ratio of GINS to Pol ⑀, and increased linearly until a 4:1 ratio of GINS:Pol ⑀ was achieved, after which the amount of DNA synthesis reached a plateau (Fig. 6C). Although data presented above suggest that GINS forms a 1:1 complex with Pol ⑀, the requirement for excess GINS may indicate weak and/or transient interaction between the two proteins under the conditions of in vitro DNA synthesis. Alternatively, excess GINS may spontaneously bind a primer template DNA, although it does not have any binding activity to either ssor dsDNA (Fig. 5). Nevertheless, it should be pointed out that DNA polymerase clamp, PCNA, also stimulated DNA synthesis by Pol ⑀ in the presence of PCNA loader, RF-C (Fig. 6D). The stimulation was only 4 -5-fold under the condition where GINS stimulated the reaction 15-20-fold. Thus, these results suggest that GINS stimulates DNA synthesis by Pol ⑀ and PCNA partially substitutes the role of GINS in DNA synthesis catalyzed by Pol ⑀. Finally, consistent with the results of binding assay (Fig. 3 and data not shown for Pol ␣-primase), the stimulation was very specific for DNA synthesis by Pol ⑀, since neither DNA synthesis by Pol ␦ nor that by Pol ␣-primase was significantly stimulated by addition of GINS (supplemental Figs. S2 and S3).
GINS Stimulates DNA Synthesis Catalyzed by Pol ⑀ on a Singly Primed X174 Single-stranded DNA-Next, the effect of GINS on the rate of DNA synthesis catalyzed by Pol ⑀ was examined using singly primed X174 single-stranded circular DNA (sscDNA) coated with yeast RPA as published (34). After incubation, the products of elongation were analyzed by electrophoresis on an alkaline agarose gel. As shown in Fig. 7A, during the first 3-5 min, DNA synthesis was almost linear in both reactions without and with GINS. Furthermore, total DNA synthesized by Pol ⑀ with GINS was two to three times higher than without GINS and the DNA products produced by Pol ⑀ were longer in the presence of GINS than in the absence of GINS at every time points (0.5, 1, 2, 3, 5, and 10 min). The product DNA at these early time points ran as rather distinct bands, and the size gradually increased to reach the full-length within 10 min incubation with GINS (Fig. 7A). Because the levels of incorporation during the incubation were less than 2% of the total incorporation obtained (only small portion of the input template primers were utilized), these results indicate that Pol ⑀ starts elongation more or less uniformly and that the polymerase has a capacity to elongate DNA all the way around the viral DNA circle without dissociating from it in the presence or absence of GINS. These results suggest that GINS stimulates either the processivity or rate of DNA synthesis catalyzed by Pol ⑀ or it stimulates both. Rates of DNA elongation by Pol ⑀ and Pol ⑀ with GINS, calculated from the maximum size of the elongation products at the first three time points, were about 15 and 30 nucleotides/s, respectively. After 5 min in the time course, the overall rate of DNA synthesis decreased gradually. FIGURE 6. GINS stimulates DNA synthesis by Pol ⑀. A, Pol ⑀ holoenzyme (25 or 50 fmol) was incubated with 600 fmol of DNA substrate (a 32 P-labeled 34-nucleotide primer annealed to a 65-mer oligonucleotide template) with or without GINS (200 fmol) at 30°C. Reaction products were analyzed on a 15% denaturing polyacrylamide gel, followed by autoradiography for 60 min or 18 h. B, results shown in A were quantified by measuring radioactivity of the 65-mer product. C, reactions were carried out and analyzed as in A except that the ratio of GINS to Pol ⑀ was varied as indicated, and reactions were incubated for 5 or 10 min at 30°C. D, PCNA and RF-C also stimulates DNA synthesis catalyzed by Pol ⑀. 25 fmol of Pol ⑀ holoenzyme was incubated with 600 fmol of DNA substrate and 600 fmol of RF-C with or without various amounts of PCNA (0.8, 1.6, and 8 pmol) for 1, 5, and 10 min at 30°C. The products were analyzed as in A, except for autoradiography for 12 h. Reactions were incubated at 30°C with or without GINS (100 fmol) for 0.5, 1, 3, 5, and 10 min. The products were analyzed by 1.2% alkaline agarose gel electrophoresis, followed by autoradiography as published (41). B, reaction mixtures were the same as in A, except for 100 fmol of a 30-mer singly primed X174 sslDNA. C, reaction mixtures were the same as in A, except for with or without GINS (100 fmol), RF-C (250 fmol), and/or PCNA (1.6 pmol) for 20 min. ϩ and Ϫ indicate addition of each protein into the reaction mixture. The bottom numbers in the figure indicate total nucleotides (pmol) synthesized in each reaction.
This may be caused simply by a slowdown of the rate of DNA synthesis. And, these rates were estimated to be about 8 and 20 nucleotides/s, respectively. These are considerably slower than the estimated 35 nucleotides/s rate of replication fork movement in vivo in S. cerevisiae (22).
Although Pol ⑀ holoenzyme itself is able to bind a singly primed X174 sscDNA and GINS does not bind either ssDNA or dsDNA, GINS⅐Pol ⑀ holoenzyme complex would have some difficulty to bind it, due to the predicted structure of GINS (33). Thus, above experiments were repeated using a singly primed X174 sslDNA coated with yeast RPA. As shown in Fig. 7B, GINS also stimulated DNA synthesis by Pol ⑀ as much as with X174 sscDNA. Therefore, it is concluded that GINS⅐Pol ⑀ complex is able to bind sscDNA as efficiently as it binds sslDNA. The in vitro DNA synthesis reactions catalyzed by Pol ⑀ or Pol ⑀⅐GINS were not further stimulated by addition of PCNA and RF-C (Fig. 7C). These results suggest that GINS is an accessory protein for Pol ⑀ and stimulates the Pol ⑀ rate and possibly processivity of DNA synthesis in the absence of DNA polymerase clamp, PCNA, and its loader, RFC, in vitro.

DISCUSSION
This study shows that purified GINS forms a 1:1 complex with Pol ⑀, but not other replicases, (such as Pol ␦ or Pol ␣-primase), and that GINS stimulates Pol ⑀-catalyzed DNA synthesis in vitro. In the presence of GINS, the rate and possibly processivity of DNA synthesis catalyzed by Pol ⑀ increase significantly (Figs. 6 and 7). This stimulation is dependent on the ratio of GINS to Pol ⑀ and also on the ratio of GINS to the DNA substrate (Fig. 6C), unlike other nonspecific DNA-binding proteins. Thus, we concluded that GINS is an accessory protein of Pol ⑀ and Pol ⑀⅐GINS complex catalyzes much more rapid and processive DNA synthesis than Pol ⑀ alone in vitro. Because it is known that Pol ⑀ readily dissociates from a priming site by addition of excess ssDNA, which may be related to the other function of Pol ⑀, S-phase checkpoint regulation (34), we could not use well establishing protocols (such as chasing the reaction with a large excess of template primer or of ssDNA or use of a large excess template primer substrate) for measuring processivity of Pol ⑀ or Pol ⑀⅐GINS complex. However, total DNA synthesized by Pol ⑀ and Pol ⑀⅐GINS during incubation was less than 2% of total DNA template primer, whose molar concentration was 5-fold excess over that of Pol ⑀, and the products were rather uniform in size during the incubation (Fig. 7). Thus, it could be roughly estimated that the processivity of Pol ⑀⅐GINS is Ն5.3 kb/binding event in vitro. Interestingly, Pol ⑀ recycles many times during in vitro DNA synthesis on a linear small template primer in the presence of GINS, but not in its absence ( Fig. 6 and supplementary Fig. S1). These reactions were partially substituted by PCNA and RF-C (Fig. 6D). These characteristics suggest that GINS may increase the suitability of Pol ⑀ for catalyzing leading strand DNA synthesis in vivo. Thus, the data presented here are consistent with the proposal that GINS is a Pol ⑀ accessory factor and that Pol ⑀⅐GINS plays a significant role as a leading strand DNA polymerase.
Previous studies showed that GINS associates with origins of replication and adjacent DNA sequences in yeast DNA (30,31) and that GINS facilitates association of Dpb11p and Cdc45p with chromatin (30). Genetic and physical interaction studies also suggested that GINS interacts with Sld3p and Dpb11p and that Sld3p and Dpb11p facilitate binding of GINS to origins of replication (30). This study provides evidence that GINS is a part of the yeast replisome in vivo (Fig. 1) and that GINS directly interacts with Pol ⑀ in vitro (Figs. 3 and 4). Other groups also showed recently that GINS is one of many replication proteins found at the replication forks (40,41). Particularly, it has been shown that all three Pols, ␣, ␦, and ⑀, Mcm2-7p, Cdc45p, GINS, and Mcm10p are found in the vertebrate replisome. Furthermore, in the presence of the DNA polymerase inhibitor aphidicolin, which causes uncoupling of a highly processive DNA helicase from the stalled replisome, only Cdc45p, GINS, and Mcm2-7p, but not any of Pols, are enriched at the pause site (41). These results are consistent with our finding of a weak and/or transient association of GINS to Pol ⑀ holoenzyme. Therefore, we suggest that Dpb11p, Sld3p, Cdc45p, GINS, and Pol ⑀ assemble in a coordinated manner on replication origins prior to initiation of DNA synthesis in vivo.
One of the functions of PCNA is to recruit Pol ␦ onto a primer site and to stimulate the processivity of Pol ␦ during Okazaki fragments synthesis on lagging strand DNA (reviewed in Ref. 11. Previous studies also suggest that PCNA and RF-C stimulate Pol ⑀-catalyzed DNA synthesis on a singly primed M13 viral sscDNA under certain conditions (42). In this study, we showed that GINS greatly stimulates Pol ⑀-catalyzed DNA synthesis on a 65-mer oligonucleotide template annealed with a 34-mer primer in vitro ( Fig. 6 and supplementary Fig. S1) and PCNA also stimulates DNA synthesis by Pol ⑀ in the presence of RF-C, but its stimulation is much less than with GINS (Fig. 6D). Using singly primed linear X174 (about 5.3 kb) as DNA substrates, Pol ⑀ synthesized a full-length double-stranded DNA product without dissociating from the template and these reactions were stimulated by addition of GINS (Fig. 7, A and B), but not further stimulated by addition of PCNA and RF-C (Fig. 7C). However, these stimulations by addition of GINS were much less than those of a small primer template. This may be because of the unique properties of Pol ⑀. We showed previously that Pol ⑀, which is actively synthesizing DNA, rapidly dissociates from a priming site by addition of a large single-stranded DNA (34). Alternatively, a singly primed X174 ssc or sslDNA used in this study may not be a representative substrate for leading strand synthesis even in the presence of RPA, thus Pol ⑀ may have a difficulty to elongate a primer on ssDNA coated with RPA in the presence of GINS. Consistent with this interpretation, other studies show that PCNA and RF-C greatly stimulate both Pol ␦and Pol ⑀-catalyzed elongation of a short oligonucleotide hybridized on X174 or M13 ssDNA in the presence of high salt (10,42). Thus, we conclude that Pol ⑀ holoenzyme primarily utilizes GINS as an accessory factor for DNA synthesis in vitro and in vivo.
Based on results from our previous studies (30) and the present article, we propose the following model to explain how Pol ⑀⅐GINS complex binds a template-primer DNA during DNA synthesis and how both leading-and lagging strand synthesis are achieved by three different DNA polymerases at eukaryotic chromosomal replication forks. Recently, we observed that Xenopus GINS, which was reconstituted and purified from insect cells (33), stimulates DNA synthesis catalyzed by Xenopus Pol ⑀ as Fig. 6 and supplemental Fig. S1. 4 Thus, it is highly possible that the structure of S. cerevisiae GINS is similar to that of Xenopus GINS and that GINS acts as a clamp for Pol ⑀, although there is no evidence to date that GINS binds ds-or ssDNA in vitro (Fig. 5). In any case, if GINS were a clamp for Pol ⑀, then a few new questions arise as followings; how is GINS⅐Pol ⑀ loaded onto a template primer in vitro and the replication origin in vivo (presumably RNA-DNA primer synthesized by Pol ␣-primase on a leading strand at origin)? Is there a clamp loader for GINS? It is widely believed that each clamp has a specific clamp loader, proteins that usually belong to the AAAϩ protein family (43). Several candidate GINS loaders should be mentioned here, including Mcm2-7p, Pol ⑀, or Pol ⑀⅐Dpb11p⅐Sld2p. Mcm2-7p is the most likely candidate, as it is a AAAϩ protein family, is loaded onto replication origins, is a component of the pre-RC during M and G 1 phase, and is a helicase that is predicted to unwind DNA at the replication fork. However, this is unlikely at present time, as GINS stimulates DNA synthesis catalyzed by Pol ⑀ on a singly primed X174 sscDNA (which has no end) as much as on a singly primed X174 sslDNA (which has ends) without Mcm2-7p (Fig. 7). Thus, as shown in Fig. 8A, we propose that a ringshaped GINS interacts with the C-terminal portion of Pol ⑀ catalytic polypeptide and with the second subunit of Pol ⑀, Dpb2p, and forms a 1:1 complex. Then, the GINS⅐Pol ⑀ subsequently undergoes a conformational transition to an "open structure" that binds to the junction between single-stranded and double-stranded DNA by using a single cleft in the Pol ⑀ catalytic subunit that seems wide enough to accommodate doublestranded DNA (44) and the Dpb2p⅐GINS, which may interact with a single-stranded template DNA. In our model shown in Fig.  8B, GINS can be located on the leading strand of the replication fork between Cdc45p-Mcm2-7p (40,41) and Pol ⑀ during chromosomal DNA replication. Thus, this model differs from a previously proposed model (43). More importantly, the configuration of GINS on the leading strand of the replication fork is different from that of PCNA on the lagging strand, where PCNA is behind Pol ␦ (Fig. 8B). If this model were correct, it is still possible that PCNA is able to bind to the priming site of leading strand as to the priming sites of lagging strand with help of RF-C.
In vivo, Sld2p is phosphorylated by S-Cdk during S phase, Dpb11p⅐Sld2p is formed and subsequently this complex loads Pol ⑀⅐GINS on the replication origin with help of Cdc45p action (30,40). Thus, Dpb11p⅐Sld2p plays a crucial role in recruiting GINS⅐Pol ⑀ to replication origins in yeast. In the scheme of Fig. 8B, GINS is an accessory protein that couples Pol ⑀ to the Cdc45p⅐Mcm2-7p. This is consistent with the finding that Pol ⑀, Cdc45p, and Mcm2-7p co-immunoprecipitate with anti-GINS antibody (Fig. 1B). It is widely believed that the eukaryotic replicative helicase is the heterohexameric Mcm2-7p (43), although this conclusion is not yet firm. Nevertheless, MCM helicase activity is rather weak, somewhat reminiscent of the relatively weak helicase activity of E. coli DnaB. DnaB becomes highly active when coupled to Pol III holoenzyme, and this coupling occurs through the -subunit of the clamp loader (45). If this analogy can be applied to eukaryotic systems, then we propose that Mcm2-7p helicase activity would become highly active when coupled to Pol ⑀⅐GINS by association of Cdc45p. Consistent with this proposal, it has been shown that replication fork movement, but not the initiation timing, is retarded in pol2-16 mutant cells, which express a DNA polymerase domain less Pol ⑀ polypeptide and are presumed that the leading strand synthesis is substituted with Pol ␦ (22). Many other proteins are required for DNA replication in eukaryotes (41,43). Although the functions of these factors are largely unknown, protein interaction studies are consistent with the model proposed in Fig. 8B. Additional biochemical studies of these factors, alone and in various combinations will be required to define and understand their specific roles during eukaryotic chromosomal DNA replication.