A Complex Consisting of Human Replication Factor C p40, p37, and p36 Subunits Is a DNA-dependent ATPase and an Intermediate in the Assembly of the Holoenzyme*

Human replication factor C (hRFC) is a multi-subunit protein complex capable of supporting proliferating cell nuclear antigen (PCNA)-dependent DNA synthesis by DNA polymerases δ and ε. The hRFC complex consists of five different subunits with apparent molecular masses of 140, 40, 38, 37, and 36 kDa. We have previously reported the expression of a three-subunit core complex, consisting of the p40, p37, and p36 subunits following coupledin vitro transcription-translation of the cDNAs encoding these proteins (Uhlmann, F., Cai, J., Flores-Rozas, H., Dean, F. B., Finkelstein, J., O’Donnell, M., and Hurwitz, J. (1996)Proc. Natl. Acad. Sci. U. S. A. 93, 6521–6526). Here we describe the isolation of a stable complex composed of the p40, p37, and p36 subunits of hRFC from baculovirus-infected insect cells. The purified p40·p37·p36 complex, like the five-subunit RFC, contained DNA-dependent ATPase activity that was stimulated by PCNA, preferentially bound to primed DNA templates, interacted with PCNA, and was capable of unloading PCNA from singly-nicked circular DNA. In contrast to the five-subunit RFC, the three-subunit core complex did not load PCNA onto DNA. The p40·p37·p36 complex inhibited the elongation of primed DNA templates catalyzed by the DNA polymerase δ holoenzyme. Incubation of the p40·p37·p36 complex with the hRFC p140 and p38 subunits formed the five-subunit hRFC complex that supported PCNA-dependent DNA synthesis by DNA polymerase δ.

Upon completion of DNA synthesis, the polymerase most likely rapidly dissociates from the tethered complex, leaving the stable PCNA sliding clamp associated with the newly synthesized DNA (12,13). RFC has been shown to efficiently remove PCNA clamps from DNA (9), an activity of particular importance in lagging strand replication which involves the synthesis of a large number of Okazaki DNA fragments. In human cells, the number of Okazaki fragments formed during one round of replication has been estimated to be 100 times greater than the molar amount of cellular PCNA (9). Thus, it is likely that in addition to its role as a clamp loader, RFC also catalyzes the removal of PCNA from DNA to fulfill the requirements for a constant supply of PCNA for further DNA synthesis and other PCNA-dependent reactions.
RFC contains multiple enzymatic activities including the ability to hydrolyze ATP to catalytically load PCNA onto DNA and subsequently recruit pol ␦. Human RFC (hRFC) contains five subunits of 140, 40, 38, 37, and 36 kDa that share conserved regions, referred to as RFC boxes. Two of the conserved regions that contain GXXXXGK(S/T) (box III) and DE(V/A)D sequences (box V) are possible ATP binding sites (14,15), but no single subunit of hRFC has been shown to hydrolyze ATP. The p40 subunit was shown to bind ATP but exhibited no detectable ATP hydrolyzing activity (16,17). The yeast hRFC p36 homologue (Rfc3) exhibited DNA-dependent ATPase activity (18), but the purified hRFC p36 subunit isolated from Escherichia coli or baculovirus overexpression systems failed to show such activity (data not shown). Thus, the identity of the hRFC subunit (or subunits) that catalyzes ATP hydrolysis is not known.
We have previously demonstrated that the hRFC holoenzyme can be reconstituted from its five subunits expressed in either baculovirus-infected insect cells (19) or in a reticulocyte in vitro transcription-translation system (20). Furthermore, it was shown that the hRFC p40, p37, and p36 subunits synthesized in the reticulocyte system formed a stable core complex to which the p140 and p38 subunits bind only when both are present (20).
In this paper, we report that the purified p40⅐p37⅐p36 complex formed in a baculovirus overexpression system contains DNA-dependent ATPase activity that is stimulated by PCNA. This three-subunit complex binds preferentially to primed DNA, interacts with PCNA, and inhibits RFC/PCNA-dependent DNA elongation catalyzed by pol ␦. High levels of the p40⅐p37⅐p36 complex can unload PCNA from DNA but cannot load PCNA onto DNA. We also present evidence that the p40⅐p37⅐p36 complex, when incubated with the p140 and p38 subunits, is converted to the five-subunit RFC active in supporting DNA elongation. This finding indicates that the p40⅐p37⅐p36 complex is an intermediate in assembly of the RFC holoenzyme.
In Vitro Transcription-Translation of RFC p140 and p38 Subunits-Coupled in vitro transcription-translation reactions (12 or 24 l) containing pET16ap140 (37.5 ng/l) and/or pET19bHisp38 DNAs (21 ng/ l) were carried out in the presence or absence of the p40⅐p37⅐p36 complex (83 fmol/l) at 30°C for 90 min as described (20).
Replication Assays-RFC activity was assayed by its ability to support the elongation of singly primed M13 DNA in the presence of pol ␦, PCNA, HSSB, and ATP (20). The standard assay (10 l) contained 40 mM Tris-HCl, pH 7.5, 0.5 mM DTT, 0.01% BSA, 7 mM magnesium acetate, 2 mM ATP, 100 M each of dATP, dGTP, dTTP, 20 M [␣-32 P]dCTP (1-2 ϫ 10 4 cpm/pmol), 4.4 fmol of singly primed circular M13 DNA, 0.25 g HSSB, 50 ng of PCNA (or as indicated), 20 fmol of pol ␦ (or as indicated), and RFC or subunits in amounts as indicated. Reaction mixtures were incubated at 37°C for 30 min, stopped with 10 mM EDTA, and separated by electrophoresis through an alkaline agarose gel (1.5%) followed by autoradiography. For quantitation, an aliquot (1 l) of the reaction mixture was withdrawn, and the amount of acid-insoluble material formed was determined. When in vitro translated products were used, hRFC subunits produced in 24-l reactions were adsorbed to protein A-agarose beads (5 l) containing polyclonal antibodies against the hRFC p37 subunit as described in the above immunoprecipitation procedure. The beads were added to DNA elongation reaction mixtures (15 l) containing 10 fmol of singly primed M13 DNA and other components as described for the standard assay or as indicated. After incubation for 30 min at 37°C with frequent shaking, replication products were analyzed as described above.
ATPase Assay-ATPase activity was measured in reaction mixture (20 l) containing 25 mM Tris-HCl, pH 7.5, 1 mM DTT, 3 mM MgCl 2 , 50 g/ml BSA, 50 M [␥-32 P]ATP (1.5 ϫ 10 4 cpm/pmol), 12.5 M DNA (as nucleotides), and the p40⅐p37⅐p36 complex as indicated. After 60 min at 37°C, aliquots (1 l) were spotted on polyethyleneimine-cellulose thin layer plates which were developed in 1.0 M formic acid, 0.5 M LiCl for 20 min at room temperature. The plates were dried and exposed for autoradiography. One unit of ATPase activity was defined as the formation of 1 nmol of P i under the reaction conditions used as quantitated by phosphorimager (Fuji) analysis.
Protein Interaction Analyses-Protein-protein interactions were examined using surface plasmon resonance. The immobilization of PCNA on sensor chips was carried out using the carbodiimide covalent linkage protocol specified in the manufacturer's instructions (Pharmacia Biosensor). The interaction between immobilized PCNA and bRFC or the p40⅐p37⅐p36 complex in solution was followed by monitoring changes in the surface concentration of proteins on sensor chips using the BIAcore 2000 at room temperature.
PCNA Loading and Unloading Assays-The loading of 32 P-labeled PCNA onto DNA was carried out in reaction mixtures (50 l), containing 0.5 pmol of singly nicked pBluescript DNA, 2.6 pmol of 32 P-labeled PCNA trimers (ϳ1500 cpm/fmol) in 50 l of incubation buffer (20 mM Tris-HCl, pH 7.5, 0.1 M NaCl, 8 mM MgCl 2 , 0.5 mM ATP, 4% glycerol, 5 mM DTT, and 40 g/ml BSA) and 200 fmol of bRFC. Reactions were incubated for 10 min at 37°C, stopped on ice, and then applied at 4°C to a 5-ml gel filtration column (Bio-Gel A15m, Bio-Rad) equilibrated with incubation buffer. Fractions of 170 l were collected, and 32 P was quantitated by Cerenkov counting. The release of 32 P-PCNA complexed to DNA (the unloading reaction) was carried out in reaction mixtures (50 l) containing 83 fmol of PCNA (as trimer) loaded onto singly nicked DNA (isolated by gel filtration), incubation buffer (as described above), and bRFC or the p40⅐p37⅐p36 complex (in amounts as indicated). Mixtures were incubated at 37°C for 10 min, and the reaction was halted by placing tubes on ice. Reactions were then filtered at 4°C through a 5-ml gel filtration column (Bio-Gel, A15m, Bio-Rad) to resolve the 32 P-PCNA (eluting in the included volume) from 32 P-PCNA bound to DNA (eluting in the excluded volume). Fractions of 170 l were collected and the 32 P quantitated by Cerenkov counting.
Preparation of Recombinant Viruses-Recombinant baculoviruses that produced the p40, p37, and p36 subunits of hRFC were as described previously (19).
Large Scale Infection of Insect Cells with hRFC-recombinant Baculoviruses and Preparation of Cell Extracts-Sf9 cells (Invitrogen) were grown at 27°C to a cell density of 2 ϫ 10 6 cells/ml in Grace's medium supplemented with 10% fetal bovine serum. Sf9 cells (2 ϫ 10 8 , 200 ml) were infected simultaneously with recombinant viruses that produced the p40, p37, and p36 subunits of hRFC at a multiplicity of infection of 5 for each virus and were maintained in a 2-liter glass flask at 27°C for 48 h with constant shaking (100 rpm). The cells were then harvested by centrifugation at 300 ϫ g for 15 min. The cell pellet was washed with ice-cold phosphate-buffered saline, resuspended in 2 volumes of hypotonic buffer (50 mM Tris-HCl, pH 8.0, 10 mM KCl, 1.5 mM MgCl 2 , 20 mM sodium phosphate buffer, pH 8.0, 0.5 mM phenylmethanesulfonyl fluoride, 0.2 g/ml aprotinin, 0.2 g/ml leupeptin, and 0.1 g/ml antipain) per volume of packed cells, and lysed with 10 strokes of a Dounce homogenizer. After centrifugation at 2,400 ϫ g for 30 min at 4°C, the supernatant (cytosolic extract, 95 mg, 9 ml) was saved, and the nuclear pellet was resuspended in 2 volumes of extraction buffer (hypotonic buffer without 10 mM KCl) per volume of packed cells and the mixture adjusted to a final concentration of 0.42 M NaCl. After centrifugation at 43,500 ϫ g for 30 min at 4°C, the supernatant (nuclear extract, 35 mg, 5 ml) was combined with the cytosolic extract, and the mixture was centrifuged at 44,000 ϫ g for 30 min at 4°C. The supernatant was used for the purification of the p40⅐p37⅐p36 complex as described below.
Purification of the p40⅐p37⅐p36 Complex-SDS-PAGE (9%) followed by Coomassie staining and Western blot analysis using antibodies specific for the p40, p37, and p36 subunits of RFC were employed to monitor all purification steps which were carried out at 4°C.
Extracts (120 mg of protein, 13 ml), prepared as described above, were adjusted to 0.1 M NaCl and chromatographed on a SP-Sepharose column (1.5 ϫ 2.9 cm) equilibrated with buffer A (25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.01% Nonidet P-40, 1 mM DTT, 0.2 mM phenylmethanesulfonyl fluoride, 0.1 g/ml antipain, 0.2 g/ml leupeptin, 0.2 g/ml aprotinin, and 10% glycerol) plus 0.1 M NaCl. After washing the column with 3-bed volumes of equilibration buffer, bound proteins were eluted using a 50-ml gradient from 0.1 to 0.4 M NaCl in buffer A. Fractions containing the p40, p37, and p36 subunits eluted at 0.25 M NaCl and were pooled (14 mg protein, 15 ml). After adjusting the NaCl concentration to 0.05 M, the pooled fractions were loaded onto a Q-Sepharose column (1.0 ϫ 1.5 cm) that was developed with a gradient (12 ml) of 0.05 to 0.4 M NaCl in buffer A. The p40⅐p37⅐p36 complex, which eluted at 0.3 M NaCl, was identified by SDS-PAGE analysis and the DNA-dependent ATPase activity assay. Prior to this step, the DNA-independent ATPase activity was too high for reliable measurement of DNA-dependent ATPase activity. The pooled fractions containing the p40⅐p37⅐p36 complex (3.7 mg of protein, 4.5 ml) contained 1672 units of DNA-dependent ATPase activity with a specific activity of 452 units/mg. After adjustment to 0.1 M NaCl, these fractions were loaded onto a heparin-Sepharose column (1.0 ϫ 1.3 cm) equilibrated with 0.1 M NaCl in buffer A. A gradient solution (10 ml) of 0.1 to 1 M NaCl in buffer A was used to elute bound proteins. Fractions containing the three RFC subunits eluted at 0.25 M NaCl and were pooled (2.45 mg, 3.5 ml, 1575 units, 643 units/ mg), adjusted to 0.05 M NaCl, and chromatographed through an ATPagarose (Sigma, N-6 attachment, 11 atom spacer) column (1.0 ϫ 1.8 cm) that was developed with a 14-ml 0.05 to 0.5 M NaCl gradient in buffer A. The p40⅐p37⅐p36 complex eluted as a broad peak between 0.1 and 0.5 M NaCl, whereas most of the contaminating proteins eluted between 0.05 and 0.2 M NaCl. Thus, fractions eluting between 0.2 and 0.5 M NaCl were pooled (0.59 mg, 6 ml, 520 units, 880 units/mg) and concentrated using the Ultrafree-15 centrifuge filter membrane (10K, Millipore). An aliquot (200 g, 150 l, 131 units, 657 units/mg) of the concentrated material was then sedimented through a 15-35% glycerol gradient (5 ml) in buffer A plus 250 mM NaCl at 250,000 ϫ g for 24 h at 4°C. Pooled glycerol gradient fractions (100 g, 0.66 ml, 114 units, 1140 units/mg) were stored at Ϫ80°C and showed no loss of activity over a 2-month period with repeated cycles of freezing and thawing. The yield of the p40⅐p37⅐p36 complex obtained from 200 ml of infected Sf9 cells was ϳ0.3 mg, representing a total recovery of 10% of the protein (as estimated by Western blot analysis). Densitometry analysis of Coomassie-stained gels was performed using the Molecular Imager and Imaging Densitometer (Bio-Rad).

Isolation of the p40⅐p37⅐p36 Complex from Baculovirus-in-
fected Insect Cells-The p40⅐p37⅐p36 complex was assembled in vivo by coexpressing the p40, p37, and p36 subunits in baculovirus-infected insect cells. After harvesting the cells, cytosolic and nuclear extracts were prepared as described under "Materials and Methods." The combined cytosolic and nuclear extracts before and after viral infection were analyzed by SDS-PAGE followed by Coomassie staining (Fig. 1A, lanes 1 and 2,  respectively). The overproduced p40, p37, and p36 subunits were not visible on SDS-PAGE after Coomassie staining (lane 2). All three subunits were detected by Western blot analyses using antibodies specific for each subunit, and the amount of p36, p37, and p40 protein formed was estimated to be 0.5-1% total protein (data not shown). The p40⅐p37⅐p36 complex was purified by a number of chromatographic steps and glycerol gradient centrifugation as described under "Materials and Methods." Following glycerol gradient centrifugation, three protein bands were observed that migrated through SDS-polyacrylamide gels at positions corresponding to those of the hRFC p40, p37, and p36 subunits (lane 3). The identity of each subunit was confirmed by Western blot analysis using polyclonal antibodies specific for each subunit (lanes 4 -6). Densitometry analysis of the Coomassie-stained gels indicated that the p40, p37, and p36 subunits were present at a molar ratio of 1.0:1.1: 1.0. The RFC p40, p37, and p36 subunits cosedimented through complex, eluted from the ATP-agarose column (see "Materials and Methods"), was subjected to a 15-35% glycerol gradient centrifugation. Aliquots (10 l) of glycerol gradient fractions were analyzed by SDS-PAGE followed by Coomassie staining. The numbers at the top of the figure represent the glycerol gradient fraction analyzed. C, DNA-dependent ATPase activity of glycerol gradient fractions. Each glycerol gradient fraction (0.5 l) was assayed for its ability to hydrolyze ATP in the presence or absence of X174 ssc DNA as described under "Materials and Methods." the glycerol gradient ( Fig. 2A), peaking in fraction 13, between aldolase (158 kDa, fractions 9 and 10) and BSA (66 kDa, fraction 14) with a sedimentation coefficient of 4.8, indicating that these subunits exist as a stable complex. When individual subunits were subjected to glycerol gradient centrifugation, the p40 and p37 subunits peaked at fraction 16, whereas the p36 subunit was detected in lower fractions of the gradient most likely due to protein aggregation (data not shown). The cosedimentation of the three subunits is consistent with our previous finding that in vitro transcribed-translated hRFC p40, p37, and p36 subunits form a stable complex consisting of equimolar amounts of each subunit (20).
We have also constructed a baculovirus vector that produces the RFC p36 subunit with an additional 6-histidine residue at the N terminus. The p40⅐p37⅐p36 complex was assembled by coexpressing the His-tagged p36 subunit with the p40 and p37 subunits in baculovirus-infected insect cells, and the complex was purified using a Ni 2ϩ affinity column followed by glycerol gradient centrifugation. The p40⅐p37⅐p36 complex that eluted from the Ni 2ϩ column with imidazole contained excess levels of the uncomplexed His-tagged p36 subunit in addition to the p40⅐p37⅐p36 complex. The excess p36 subunit was not totally removed by the glycerol gradient centrifugation. The purified p40⅐p37⅐p36 complex containing either the untagged or Histagged p36 subunit had essentially identical properties (data not shown).
The RFC p40⅐p37⅐p36 Complex Contains DNA-dependent ATPase Activity-ATP hydrolysis is required in order for RFC to load PCNA onto primed DNA templates and recruit pol ␦. However, none of the hRFC subunits isolated from E. coli or baculovirus overexpression systems has been shown to exhibit ATPase activity (17; data not shown) suggesting that multiple subunits may be required for this activity. Therefore, we examined the purified p40⅐p37⅐p36 complex for its ability to hydrolyze ATP, and as shown in Fig. 1C this DNA-dependent activity peaked in fraction 13 of the glycerol gradient, coincidental with the sedimentation of the p40⅐p37⅐p36 protein complex.
The effects of various DNA effectors on the ATPase activity of the complex were examined (Fig. 2A). The p40⅐p37⅐p36 complex possessed weak ATPase activity that was stimulated maximally (34-fold) by X174 ssc DNA, 10-fold by poly(dA) 300 : oligo(dT) 30 , 6-fold by poly(dA) 300 , and 4-fold by oligo(dT) 30 . These properties are similar to those observed for the fivesubunit RFC purified from HeLa cells (3) or from baculovirusinfected insect cells (19). Each molecule of the p40⅐p37⅐p36 complex hydrolyzed 4.5 and 1.4 molecules of ATP/min at 37°C in the presence of X174 ssc DNA and poly(dA) 300 :oligo(dT) 30 , respectively. These values are 2-3 times lower than those observed with the five-subunit hRFC purified from HeLa cells (3) and 4 -7 times lower than that of the bRFC (19). The Histagged p40⅐p37⅐p36 complex contained DNA-dependent ATPase activity that was 2-fold higher than the untagged three-subunit complex. It is possible that the many purification steps required to isolate the untagged p40⅐p37⅐p36 complex leads to some inactivation of this complex.
We also examined the effects of PCNA and HSSB, previously shown to enhance hRFC DNA-dependent ATPase activity (3,19). As shown in Fig. 2B, PCNA stimulated the ATPase activity of the three-subunit complex 2-to 3-fold in the presence of poly(dA) 300 :oligo(dT) 30 but not poly(dA) 300 . This property is identical to that of bRFC and hRFC, suggesting that optimal ATP hydrolyzing activity is achieved when the p40⅐p37⅐p36 complex is associated with PCNA on primed DNA. However, the ATPase activity of the p40⅐p37⅐p36 complex was unaffected by HSSB (data not shown) in contrast to results observed with hRFC and bRFC. It is possible that the stimulatory effects of HSSB depend on its interaction with the p140 and/or p38 subunits.
The RFC p40⅐p37⅐p36 Complex Binds Specifically to Primed DNA-Previous experiments indicated that the interaction between RFC and DNA is mediated by the p140 subunit (16). However, the observation that the p40⅐p37⅐p36 complex contained DNA-dependent ATPase activity suggests that it should also interact with DNA. We examined the DNA-binding properties of this complex using a nitrocellulose filter binding assay. High ionic strength (175 mM NaCl) was required to observe selective binding of bRFC to primed DNA and not to singlestranded DNA (19). However, under these conditions the p40⅐p37⅐p36 complex did not bind to either type of DNA. When the salt concentration was lowered, selective binding of the core complex to primed DNA was observed at 20 mM NaCl. Thus the DNA binding assay with the p40⅐p37⅐p36 complex was carried out at this salt concentration. As shown in Fig. 3, the p40⅐p37⅐p36 complex bound poly(dA) 300 or oligo(dT) 30 inefficiently. However, the complex bound poly(dA) 300 annealed to oligo(dT) 30 , and the binding efficiency was markedly increased as the molar ratio of oligo(dT) 30 to poly(dA) 300 increased, suggesting that the p40⅐p37⅐p36 complex specifically recognized ATPase assays were carried out as described under "Materials and Methods." Reaction mixtures contained the p40⅐p37⅐p36 complex in amounts as indicated, in the absence or presence of 12.5 M (nucleotide concentration) of the following DNAs: oligo(dT) 30 , poly(dA) 300 , poly(dA) 300 :oligo(dT) 30 (1:5 molar ratio), or X174 ssc DNA. B, effect of PCNA on the DNA-dependent ATPase activity of the three-subunit complex. PCNA was added in amounts as indicated to reaction mixtures containing 50 ng of the p40⅐p37⅐p36 complex in the presence of 12.5 M of poly(dA) 300 or poly(dA) 300 :oligo(dT) 30 .
DNA primer ends. The selective binding of the p40⅐p37⅐p36 complex to primed DNA is similar to that of the five-subunit RFC. Selective binding of primed DNA templates under low ionic condition was also observed with each purified subunit of the three-subunit complex. However, the efficiency of binding was lower (2-10-fold). The requirement for low ionic strength to observe DNA binding by the p40⅐p37⅐p36 complex suggests that this three-subunit complex may not be a major factor influencing the binding of RFC to primed DNA.
The RFC p40⅐p37⅐p36 Complex Interacts with PCNA and Inhibits PCNA-dependent DNA Elongation Catalyzed by pol ␦-The p40 and p140 subunits of hRFC have been previously shown to interact with PCNA (17,25). To explore further the interaction between RFC subunits and PCNA, we determined the direct interaction between PCNA and the p40⅐p37⅐p36 complex using the surface plasmon resonance technique as described under "Materials and Methods." As shown in Fig. 4, when solutions of bRFC or the p40⅐p37⅐p36 complex were passed over a sensor surface on which 33 fmol of PCNA (3,000 RU) was immobilized, an increase in mass on the sensor surface was detected. In this experiment, 27 fmol of bRFC (8,104 RU) was retained on the PCNA-coated chip, corresponding to a stoichiometry of ϳ1 molecule of bRFC bound per molecule of PCNA trimer (Fig. 4A). The p40⅐p37⅐p36 complex (4 fmol, 440 RU) also bound a sensor chip to which an equivalent amount of PCNA had been coupled, corresponding to a stoichiometry of ϳ1 molecule of the p40⅐p37⅐p36 complex bound per 8 molecules of PCNA trimer (Fig. 4B). Consistent with this result, about 5 times more PCNA was co-immunoprecipitated with RFC than with the p40⅐p37⅐p36 complex using polyclonal antibodies against the p37 subunit (data not shown).
The p40⅐p37⅐p36 complex did not support the PCNA-dependent DNA elongation reaction catalyzed by pol ␦ (see below, Fig.  7B). However, the three-subunit complex inhibited the elongation of singly primed M13 DNA catalyzed by pol ␦, PCNA, and bRFC. As shown in Fig. 5, the addition of increasing amounts of the p40⅐p37⅐p36 complex to this DNA elongation system resulted in an increased inhibition of DNA synthesis, as evidenced by the accummulation of DNA products shorter than full length (lanes 1-6). The inhibitory effects of the threesubunit complex were pronounced at low levels of PCNA (Fig.  5). As shown in Fig. 5 (lane 7), the addition of high levels of PCNA reversed the inhibition, whereas high levels of pol ␦ or RFC was less effective (data not presented). These observations suggest that the p40⅐p37⅐p36 complex inhibited DNA elongation by competing with the five-subunit RFC for the interaction with PCNA.
Inhibition of the elongation reaction with each of the core FIG. 3. The RFC p40⅐p37⅐p36 complex preferentially binds to primed DNA. A nitrocellulose filter binding assay was used to examine the DNA-binding properties of the p40⅐p37⅐p36 complex as described under "Materials and Methods." The complex, in amounts as indicated, was incubated with 32 P-labeled poly(dA) 300 , oligo(dT) 30 , or poly(dA) 300 : oligo(dT) 30  subunits alone was examined. Only the p40 subunit significantly affected the DNA elongation reaction, and its effect was qualitatively different than that observed with the three-subunit complex. The p40 subunit alone markedly decreased the synthesis of full-length M13 DNA without reduction of nucleotide incorporation, and its effect was evident at high concentrations of PCNA, similar to those used in Fig. 5, lane 7  (data not presented).
The RFC p40⅐p37⅐p36 Complex Unloads PCNA from DNA-The p40⅐p37⅐p36 complex did not load PCNA onto singly nicked circular duplex DNA (data not shown). However, as shown in Fig. 6A, when the three-subunit complex was incubated with 32 P-labeled PCNA complexed with DNA in the absence of ATP, PCNA was displaced from the DNA as shown by the decrease in 32 P-PCNA isolated from the excluded region and the concomitant increase of PCNA detected in the included volume. The amount of PCNA (23, 43, and 55%) displaced from the PCNA⅐DNA complex increased with the addition of increasing amounts of the p40⅐p37⅐p36 complex (6, 12, and 24 pmol, respectively) (Fig. 6, A and B). However, compared with the ATP-dependent bRFC-catalyzed unloading reactions in which 5, 20, and 50 fmol of bRFC protein removed 23, 68, and 88% of the PCNA from DNA, respectively, the efficiency of the threesubunit complex was low (ϳ10 3 -fold less), suggesting that the p38 and/or p140 subunits of hRFC play important roles in the unloading reaction. High levels (11 pmol) of the RFC p40 subunit also unloaded PCNA from DNA (35%) in the absence of ATP but neither the p37 nor p36 subunits unloaded PCNA from DNA under the same conditions (data not shown). These findings suggest that p40 subunit is responsible for the unloading activity of the p40⅐p37⅐p36 complex. It should be emphasized that while RFC unloaded PCNA from DNA in an ATP-dependent manner, the unloading of PCNA by the p40⅐p37⅐p36 complex or the p40 subunit alone did not require ATP, and the efficiency of the reaction was unaffected by ATP (data not presented).
The RFC p40⅐p37⅐p36 Complex Is an Intermediate in the Assembly of the 5-Subunit RFC-Previous studies in which the five cloned human genes of RFC were expressed in an in vitro coupled transcription-translation system showed that these gene products formed a stable five-subunit complex containing approximately equimolar levels of each subunit (20). An examination of the interactions between the RFC subunits indicated that a stable three-subunit core complex containing the p40, p37, and p36 subunits formed that interacted with p38 and p140 only when both subunits were present. These findings suggested a model in which the p38 and p140 subunits bind cooperatively to the core complex in assembling the holoenzyme (20).
The availability of the homogeneous p40⅐p37⅐p36 core complex permitted us to examine whether the isolated complex acted as an RFC holoenzyme assembly intermediate. For this purpose we examined whether RFC activity could be reconstituted upon mixing the isolated baculovirus expressed p40⅐p37⅐p36 complex with the p140 and p38 subunits. As shown in Fig. 7A, when the p40⅐p37⅐p36 complex was incubated with in vitro translated p140 and p38 subunits and immunoprecipitated with antibodies specific for the p37 subunit, a complex containing both 35 S-labeled p140 and p38 subunits was formed (lane 5). Consistent with previous findings, the large subunit did not interact with the p40⅐p37⅐p36 complex in the absence of the p38 subunit (Fig. 7A, lane 3). However, the p38 subunit was co-immunoprecipitated with the p40⅐p37⅐p36 complex in the absence of the p140 subunit (lane 4), although the amount precipitated was about eight times lower than that precipitated when the p140 subunit was also present in the reaction (Fig.  7A, lane 5). This result suggests that hRFC p38 subunit plays an important role in the interaction between the p140 subunit and the p40⅐p37⅐p36 core complex.
The products formed in the reactions described in Fig. 7A were examined for their ability to support the elongation of a singly primed M13 DNA in a replication reaction containing HSSB, pol ␦, and PCNA. As shown in Fig. 7B, the complex containing all five subunits (lane 7) supported DNA synthesis. No replication activity was observed upon omission of RFC (lane 2), p140 and p38 (lane 3), the p40⅐p37⅐p36 complex (lane 4), the p38 (lane 5), or the p140 subunits (lane 6). These results are in keeping with the requirement for all five RFC subunits in the DNA elongation reaction. It should be noted that the adsorption of RFC to protein A beads containing antibodies to the p37 subunit reduced the activity of RFC by approximately 70% (compare lanes 1 and 8). Most of this inhibition was due to the presence of protein A beads which reduced incorporation by approximately 50% (compare lanes 1 and 9). Since the p40⅐p37⅐p36 complex immunoprecipitated on the beads was in molar excess over the reconstituted five-subunit RFC and could potentially inhibit DNA elongation (see Fig. 5), high levels of PCNA (750 fmol) were used to obviate this problem.
It should also be noted that the above experiments were carried out using in vitro translated RFC p140 and p38 subunits. We have also carried out these experiments using purified RFC p140 and p38 subunits isolated from baculovirus-FIG. 6. The p40⅐p37⅐p36 complex unloads PCNA from singly nicked pBS DNA. A, 32 P-PCNA was assembled onto DNA as described under "Materials and Methods" and the product isolated by separation through a Bio-Gel A15m column. The isolated 32 P-PCNA-DNA complex (containing 83 fmol of PCNA) was then incubated in the presence or absence of the p40⅐p37⅐p36 complex (amounts as indicated) as described under "Materials and Methods" with the exception that ATP was omitted prior to gel filtration to resolve free 32 P-PCNA from the 32 P-PCNA⅐DNA complex. Fractions from both the included (free 32 P-PCNA) and excluded regions ( 32 P-PCNA⅐DNA complex) were quantitated by Cerenkov counting. B, the results shown in A are presented graphically with the percentage of PCNA unloaded representing the total amount of PCNA removed from DNA in the presence of the p40⅐p37⅐p36 complex. The unloading of PCNA by bRFC (in the presence of ATP) is presented for comparison. Note that the levels of bRFC were in the fmol range, whereas pmol levels of the p40⅐p37⅐p36 complex were added. infected insect cells. Incubation of the purified three-subunit complex with isolated p140 and p38 subunits did not lead to the assembly of the five-subunit complex or formation of a biologically active product. This failure was traced to a defect in the p38 subunit. Incubation of the in vitro translated p38 subunit with the baculovirus isolated p140 and core complex yielded biologically active five-subunit RFC (data not presented). However, the purified baculovirus expressed p38 subunit could not substitute for the in vitro translated p38 in these experiments (data not shown). These observations suggest that the p38 subunit, when expressed alone in baculovirus-infected insect cells, may be in an aberrant conformation that prevents its interaction with other RFC subunits. The in vitro translated p38, on the other hand, synthesized in the presence of other RFC subunits or alone, can be successfully assembled into the RFC complex. The reasons for this discrepancy are unclear.

DISCUSSION
The mechanism of elongation of primed DNA templates during DNA replication is functionally conserved in prokaryotes and eukaryotes. DNA polymerases use a circular "sliding clamp" and a multi-subunit "clamp loader" as accessory factors to achieve their processivity. The clamp loaders isolated from E. coli, phage T4, and human all consist of five subunits, capable of hydrolyzing ATP in a DNA-dependent manner, binding to primed DNA templates, and interacting with their corresponding clamp proteins and SSBs (26,27). The E. coli and human clamp loaders are also clamp unloaders while the T4 clamp dissociates spontaneously from DNA in the absence of the T4 DNA polymerase (9,28,29) and thus may not require the gene product 44/62 to function as a clamp unloader. Significant progress has been made in all three systems in assigning functions to the individual subunits of the clamp loaders (summarized in Table I).
The intrinsic DNA-dependent ATPase activity of RFC is essential for both the catalytic loading of PCNA onto primed DNA and subsequent recruitment of pol ␦. We report here that a sub-complex consisting of hRFC p40, p37, and p36 subunits hydrolyzes ATP in a DNA-dependent manner with close to 50% of the efficiency observed with the five-subunit hRFC. The contributions of the p140 and p38 subunits to the hRFC ATPase activity remain to be determined. The finding that the hRFC ATPase activity resides in a core p40⅐p37⅐p36 heterotrimer but not with these subunits alone is surprising since the Saccharomyces cerevisiae Rfc3 subunit has been reported to contain DNA-dependent ATPase activity (18). Whether this reflects different functions of these evolutionary related subunits or a unique structural problem with the human p36 subunit is presently unclear.
In E. coli, the ␥ subunit alone binds to ATP but exhibits very FIG. 7. The p40⅐p37⅐p36 complex is an intermediate in the assembly of the five-subunit RFC. A, complex formation after incubation of the p40⅐p37⅐p36 complex and the p140 and p38 subunits. Coupled in vitro transcription-translation reactions (12 l) with plasmid DNAs containing either the coding sequence for hRFC p140 or the p38 subunit were carried out in the presence or absence of the p40⅐p37⅐p36 complex (1 pmol) followed by immunoprecipitation (IP) using a polyclonal antibody against the hRFC p37 subunit. The immunoprecipitated products were analyzed by 9% SDS-PAGE followed by autoradiography to visualize the 35 S-labeled p140 and p38 protein products. Shown in lane 1 is 10% of the reaction mixture containing the p40⅐p37⅐p36 complex and in vitro translated p140 and p38 subunits prior to immunoprecipitation.  little DNA-dependent ATPase activity, whereas the heterodimers ␥␦ and ␥␦Ј and the heterotrimer ␥␦␦Ј all contain significant DNA-dependent ATPase activity (30 -33). These observations are similar to those found with hRFC subunits as reported here. The hRFC p40 subunit can bind ATP (16,17) and thus by analogy with the ␥ subunit in E. coli may be the critical site for the ATPase activity of the clamp loader. It is possible that p40 when expressed alone folds into an aberrant conformation incapable of hydrolyzing ATP but when coexpressed with the p36 and p37 subunits becomes folded into an active conformation capable of hydrolyzing ATP. In support of this, the stable p40⅐p37⅐p36 complex could not be formed in vitro by mixing purified p40, p37, and p36 subunits (data not shown). At present, we cannot rule out the possibility that the p36 or p37 may act as the catalytic subunit of hRFC ATPase, since both contain ATP-binding sequences, although neither purified p36 nor p37 binds ATP (17, data not shown). RFC complexes reconstituted with subunits individually mutated in their putative ATP binding sites should define the subunit(s) essential for ATPase activity.
It was previously reported that RFC bound to DNA through the large subunit (16) and that the p37 subunit had weak DNA-binding activity (17). In this report, we provide evidence that p37, together with the p36 and p40 subunits, possess DNA-binding activity essential for the DNA dependence of the ATPase activity. Based on these observations, we suggest that the p140 subunit mediates the initial RFC DNA binding step followed by a DNA-p40⅐p37⅐p36 interaction required for the stimulation of ATP hydrolysis.
The p40⅐p37⅐p36 complex is not a clamp loader nor does it support PCNA-dependent DNA synthesis by pol ␦, indicating an essential role for the p140 and/or p38 subunit in the loading reaction. The unloading of PCNA from DNA may be less complicated, since both p40 alone and the three-subunit complex were also capable of unloading PCNA. This is consistent with the notion that unloading only requires the PCNA ring to be opened at one or more of the trimer interfaces, while loading requires PCNA opening, positioning on the primed DNA, and subsequent closing of the ring around DNA.
In contrast to the hRFC holoenzyme, unloading of PCNA by the p40⅐p37⅐p36 complex does not require ATP, suggesting that ATP is not essential for opening the PCNA ring structure. Since the p40⅐p37⅐p36 complex can interact with PCNA, it is possible that interactions between the RFC subunits and PCNA are sufficient to open the ring-shaped PCNA, leading to dissociation of PCNA from DNA. Consistent with this, the p40 subunit alone can also unload PCNA in an ATP-independent manner (data not presented). The requirement of ATP in the loading and unloading of PCNA by hRFC may reflect changes in the structure of hRFC induced by ATP binding and hydrolysis that alter the subunit(s) that interact(s) with PCNA contributing to its catalytic activity.
The 10 3 -fold difference in the unloading efficiency between the three-subunit complex and hRFC indicates that the p140 and p38 subunits are essential for efficient unloading. The p140 and p38 subunits may contribute to the unloading reaction in two ways: (i) strong DNA binding of the p140 subunit may bring hRFC into close proximity with PCNA on DNA stabilizing this interaction, and (ii) additional interactions between the p140 or p38 subunits and PCNA may be required for efficient opening of the PCNA ring. Both the p140 and p38 subunits alone interact with PCNA (25, 34, data not shown). Likewise, the loading of PCNA onto DNA may also require multiple interac-tions between the different RFC subunits and PCNA.
Following incubation with the p38 and p140 subunits, the p40⅐p37⅐p36 sub-complex can be converted to the five-subunit complex active in supporting DNA replication. Thus the fivesubunit RFC complex may be assembled in two steps: the formation of the p40⅐p37⅐p36 complex, and the subsequent recruitment of the p140 and p38 subunits. Since p38 interacts with both the p140 subunit and the p40⅐p37⅐p36 complex, it may act as a bridge between the large subunit and the threesubunit core complex. However, the four-subunit complex (p40, p37, p36, and p38) isolated by immunoprecipitation after mixing the p40⅐p37⅐p36 complex with p38 could not be converted to the five-subunit complex following incubation with the p140 subunit (data not shown), suggesting that p38 and p140 subunits bind cooperatively to the core complex. Complexes that contain less than five subunits could not support PCNA-dependent DNA synthesis, indicating that all five subunits are required to constitute functional RFC.