The Human Lagging Strand DNA Polymerase δ Holoenzyme Is Distributive*

Background: Human DNA Pol δ holoenzyme carries out the lagging strand DNA synthesis. Results: Human DNA Pol δ holoenzyme is distributive with nucleotide incorporation rates a hundredfold slower than its prokaryotes/phage counterparts. Conclusion: Human DNA Pol δ holoenzyme dissociates and reassembles during DNA replication. Significance: Probing the characteristics of Pol δ holoenzyme is important for elucidating its mechanistic details and function in vivo. Polymerase δ is widely accepted as the lagging strand replicative DNA polymerase in eukaryotic cells. It forms a replication complex in the presence of replication factor C and proliferating cell nuclear antigen to perform efficient DNA synthesis in vivo. In this study, the human lagging strand holoenzyme was reconstituted in vitro. The rate of DNA synthesis of this holoenzyme, measured with a singly primed ssM13 DNA substrate, is 4.0 ± 0.4 nucleotides. Results from adenosine 5′-(3-thiotriphosphate) tetralithium salt (ATPγS) inhibition experiments revealed the nonprocessive characteristic of the human DNA polymerase (Pol δ) holoenzyme (150 bp for one binding event), consistent with data from chase experiments with catalytically inactive mutant Pol δAA. The ATPase activity of replication factor C was characterized and found to be stimulated ∼10-fold in the presence of both proliferating cell nuclear antigen and DNA, but the activity was not shut down by Pol δ in accord with rapid association/dissociation of the holoenzyme to/from DNA. It is noted that high concentrations of ATP inhibit the holoenzyme DNA synthesis activity, most likely due to its inhibition of the clamp loading process.

Polymerase ␦ is widely accepted as the lagging strand replicative DNA polymerase in eukaryotic cells. It forms a replication complex in the presence of replication factor C and proliferating cell nuclear antigen to perform efficient DNA synthesis in vivo. In this study, the human lagging strand holoenzyme was reconstituted in vitro. The rate of DNA synthesis of this holoenzyme, measured with a singly primed ssM13 DNA substrate, is 4.0 ؎ 0.4 nucleotides. Results from adenosine 5-(3-thiotriphosphate) tetralithium salt (ATP␥S) inhibition experiments revealed the nonprocessive characteristic of the human DNA polymerase (Pol ␦) holoenzyme (150 bp for one binding event), consistent with data from chase experiments with catalytically inactive mutant Pol ␦ AA . The ATPase activity of replication factor C was characterized and found to be stimulated ϳ10-fold in the presence of both proliferating cell nuclear antigen and DNA, but the activity was not shut down by Pol ␦ in accord with rapid association/dissociation of the holoenzyme to/from DNA. It is noted that high concentrations of ATP inhibit the holoenzyme DNA synthesis activity, most likely due to its inhibition of the clamp loading process.
Coordination of the activities of replicative polymerase and accessory proteins is indispensable for the replication of chromosomal DNA in different organisms. This process requires different proteins and enzymes to form a multimeric functional replisome to secure accurate and fast transfer of genetic information from one generation to the next (1,2). DNA replication is semiconservative, and the leading stand synthesis is continuous, whereas the lagging strand is replicated in small fragments (Okazaki fragments) in a discontinuous fashion with intervening short stretches of DNA then sealed by DNA ligase (3). Unlike bacteriophages or prokaryotes, eukaryotic organisms encode two distinctive replicative polymerases to perform leading and lagging strand DNA synthesis (4). Recent investi-gations revealed that polymerase ⑀ (Pol ⑀) 2 and polymerase ␦ (Pol ␦) carry out leading and lagging strand replication, respectively, in Saccharomyces cerevisiae (4). Several lines of evidence suggest that other eukaryotic organisms also encode two replicative DNA polymerases to duplicate their genomic DNA (4 -6). The discrete Okazaki fragment size in eukaryotic cells is 100 -250 bp long, implying that a slow and distributive lagging strand holoenzyme is sufficient to achieve the synthesis of short stretches of DNA. However, recently, Langston and O'Donnell (7) revealed that the S. cerevisiae Pol ␦ holoenzyme is fast and processive and invoked a collision model for the release and recycling of the lagging replicative polymerase. Previous in vitro reconstitution of Pol ␦ holoenzyme with proteins purified from calf thymus provided details about the components and characteristics of the replicative complex in vertebrates (8,9). With singly primed M13, the synthesized products spanned a broad spectrum of primer extension fragments formed with apparent velocities of 1.6 -5.0 nt/s (8,9), much slower than those encountered with phage and bacterial polymerases.
The human Pol ␦ holoenzyme is generated from Pol ␦, PCNA (sliding clamp), and RFC (clamp loader). Human Pol ␦ contains a catalytic subunit p125 and three auxiliary subunits p66, p50, and p12 (10). RFC is a heteropentameric complex, composed of one large subunit and four small subunits, and an ATP-dependent AAA ϩ ATPase that loads PCNA onto DNA (11). PCNA, a homotrimeric clamp, encircles the DNA substrate and is loaded at the primer/template (P/T) junction by RFC, providing a platform for the formation of the holoenzyme. The heterotrimeric RPA (single-stranded DNA (ssDNA)-binding protein) binds to ssDNA to remove secondary structures and prevent endonucleolytic cleavage of the DNA (12). All these components are necessary for optimal DNA synthesis (13). Recently, human lagging strand holoenzyme was reconstituted in vitro, and its processivity was compared with Pol ⑀ (14). Here we report detailed measurements of the kinetics and processivity of this holoenzyme.
In this work, the human Pol ␦ holoenzyme complex was assembled in vitro, and the DNA replication reaction velocity and processivity were examined. The ATPase activities of RFC were measured to deduce the assembly pathway of the holoenzyme. Surprisingly, the dNMP incorporation rate of the reconstituted holoenzyme showed a negative correlation with the ATP concentration used in holoenzyme assembly.

EXPERIMENTAL PROCEDURES
The primers were synthesized by Integrated DNA Technologies. [␣-32 P]ATP, [␥-32 P[ATP, and [␣-32 P]dCTP were purchased from PerkinElmer Life Sciences. DNA markers and T4 polynucleotide kinase were purchased from New England Biolabs. Polyethyleneimine-modified cellulose (PEI cellulose) TLC plates were obtained from EMD Biosciences. ATP␥S was purchased from Calbiochem. dNTPs were from Denville Scientific Inc. Electrophoresis grade agarose was from IBI Scientific. Chromatography paper DE81 was from Whatman International Ltd.
Plasmids and DNA Substrate-The plasmids encoding the wild-type human Pol ␦ were generously provided by Yuji Masuda (13). The plasmid encoding the C-terminal His 6 -tagged wild-type PCNA were generously provided by Ulrich Hübscher (15). The plasmid encoding the subunits of human RPA was a kind gift from Marc Wold (12). The plasmids encoding the subunits of wild-type RFC were a generous gift from Paul Modrich (16). ssM13mp18 DNA was prepared by standard procedures. A catalytically inactive mutant, Pol ␦ AA (D755A/D757A mutant), was constructed by site-directed mutagenesis (forward primer, 5Ј-GTG GTG TAT GGT GCC ACT GCC TCC GTC ATG TG-3Ј, and reverse primer, 5Ј-CAC ATG ACG GAG GCA GTG GCA CCA TAC AC AC-3Ј) and confirmed by sequence analysis.
The DNA substrate for measuring the RFC ATPase activity was prepared by annealing the template (5Ј-CGG ACT GCA CGT GCG TGG GCA TTC GTC GCG CAG GCT CAG CGT CCA CGA GCA CGA-Biotin-3Ј), primer (5Ј-CTG GTC TCG CGA TGG ACG CTG AGC CTG CGC-3Ј), and flap (5Ј-CGT GGT GGT AGG TGA GGG CGG CAC GTG CAG TCC GCG-3Ј). The end was blocked by the addition of NeutrAvidin prior to the measurements.
Proteins-The wild-type Pol ␦ was expressed and purified as described earlier (13). The Pol ␦ AA mutant was expressed and purified similarly to the wild-type enzyme. RPA and PCNA were expressed as reported previously (12,15). Purifications of the wild-type RFC and truncated RFC are described elsewhere. 3 ATP Hydrolysis Assay-Measurement of the intrinsic ATPase activity of RFC was carried out with various concentrations of ATP (0 -1 mM) supplemented with [␣-32 P]ATP in the standard complex reaction buffer (25 mM Tris acetate (pH 7.8), 125 mM KOAc, 10 mM Mg(OAc) 2 , and 1 mM DTT) at 37°C. The ATPase activity of the RFC complex in the presence of PCNA and/or DNA was examined in the presence of 1 mM ATP. When DNA was included in the reaction, the forked DNA described above with an end blocked with a biotin-avidin complex was utilized as a substrate to prevent PCNA from sliding away once loaded. Aliquots of the reactions were drawn at various time points and quenched with 0.5 M EDTA (pH 8.0), and 1 l of the quenched solution was spotted on a PEI cellulose plate and developed in 200 mM potassium P i (pH 7.5) buffer, dried, and exposed to a PhosphorImager screen. The data were analyzed and quantified by ImageQuant software.
ssM13 DNA Primer Extension Assay-Each of the components of ssM13 DNA replication by human Pol ␦ holoenzyme was varied to determine the maximal dNMP incorporation rate. Replication assays were carried out in the standard complex reaction buffer at 37°C. The 38-nucleotide-long primer (5Ј-GGG TTT TCC CAG TCA CGA GGT TGT AAA ACG ACG GCC AG-3Ј, which is complementary to the 6328 -6291 region of ssM13mp18 DNA) was annealed to ssM13 DNA at a ratio of 2:1. In the instances where 32 P-labeled primer was used, the ratio of primer to template was 1.1:1. The primer was radiolabeled with 32 P at the 5Ј-end by T4 polynucleotide kinase using standard procedures (17). A typical reaction contained 10 nM singly primed ssM13 DNA, 100 nM Pol ␦, 100 nM RFC, 200 nM PCNA, 2 M RPA, 50 M dNTPs (10 Ci of [␣-32 P]dCTP was added), and 1 mM ATP in the standard complex reaction buffer that contained 10 mM Mg 2ϩ to provide a standard reaction solution. The reactions were quenched with 2ϫ loading buffer (60 mM NaOH, 2 mM EDTA) at regular intervals, and the products were resolved on a 0.8% alkaline agarose gel, transferred to DE81 paper, exposed to a PhosphorImager screen, and quantified by ImageQuant. The incorporation of radionucleotide into the product of the primer extension ssM13 DNA was calculated from a calibration curve. The ionic strength was normalized to 200 mM salt by the addition of NaCl.
For the ATP␥S addition experiment, a premixed standard reaction mixture was incubated with primed ssM13 DNA for 1 min, and various concentrations of ATP␥S were added to the reaction solution with [␣-32 P]dCTP. The reactions were chased at different times (⌬t), quenched by the quenching buffer, resolved, and quantified as above.
For the polymerase Pol ␦ AA chase experiment, a premixed standard reaction solution was incubated with primed ssM13 DNA for 1 min, and then different concentrations of Pol ␦ AA were added to the reaction solution with [␣-32 P]dCTP. The reactions were chased at different times (⌬t), quenched by the quenching buffer, resolved, and quantified as above.
Polymerase Dilution Experiment-The primer extension reaction contained 10 nM singly primed ssM13 DNA (the primer is labeled with 32 P at the 5Ј-end), 50 nM Pol ␦, 100 nM RFC, 200 nM PCNA, 2 M RPA, 50 M dNTPs, and 1 mM ATP in standard complex reaction buffer. After an incubation at 37°C for 1 min, the reaction was diluted with the primer extension reaction mixture that lacked both the Pol ␦ and the primed ssM13 DNA to obtain 2-, 4-, and 8-fold dilutions of the reactions. The reactions were continued, aliquots were drawn at regular intervals, and the products were resolved on an alkaline agarose gel and quantified as described above.
Monte Carlo Simulation of Dilution Experiment-The Monte Carlo simulation for singly primed ssM13 DNA synthesis was done as described elsewhere. 4 Briefly, the program tests the possibility of the presence of a holoenzyme on a DNA sub-strate by comparing a computer-generated random number with an association/dissociation probability calculated based on the binding kinetics of the holoenzyme. If there is a holoenzyme, in a simulation step, the DNA is elongated by one nucleotide. The simulation step size, ⌬t, was 0.01 s, equal to the reciprocal of the single nucleotide incorporation rate of 100 nt/s (18). The best results were achieved when the association rate constant was set to 1 ϫ 10 6 M Ϫ1 s Ϫ1 and the dissociation rate constant was set to 0.7 s Ϫ1 . A 7-min reaction was performed for each DNA substrate. During the first minute, the concentration of Pol ␦ was 50 nM, and over the next 6 min, the rate constant decreased by the same factor as the dilution in the corresponding experiments. A total of 0.5 million DNA molecules were simulated.

RESULTS
ATPase Activity of RFC and Its Complexes-Human RFC is a member of the AAA ϩ superfamily of ATPases and utilizes ATP to initiate the assembly of the holoenzyme complex by loading PCNA onto the DNA substrate (11). To understand how the complex forms, we examined the ATPase activity of RFC under various conditions, and the results are shown in Table 1. The intrinsic ATPase activity of RFC is low, and the data fit to a Michaelis-Menten equation to yield a k cat of 0.065 Ϯ 0.01s Ϫ1 and a K m(ATP) of 57 Ϯ 6 M. All the other ATPase assays were carried out at a constant ATP concentration of 1 mM. Endblocked forked DNA stimulated the ATPase activity of RFC 4-fold to a k cat of 0.27 Ϯ 0.03 s Ϫ1 and an apparent K m(DNA) of 27 Ϯ 7 nM. PCNA had no effect on ATPase activity of RFC in the absence of DNA. In the presence of 250 nM RFC, 250 nM PCNA, 250 nM DNA, and 1 mM ATP, the ATP hydrolysis increased to an apparent k cat of 0.68 Ϯ 0.06 s Ϫ1 . The addition of Pol ␦ had no significant effect on the ATPase activity of RFC.
DNA Replication Dependence on ATP Concentration-The ATPase activity of RFC showed a constant hydrolysis of ATP under different conditions, suggesting that PCNA was loaded onto the DNA P/T junction continuously even in the presence of human Pol ␦. To further understand whether the loading of PCNA is required after the initiation of DNA replication by Pol ␦ holoenzyme, we carried out an ATP concentration-dependent DNA replication experiment after the optimization of replication assay (supplemental Fig. S1). As shown in Fig. 1, when 0.5 or 1 mM ATP was used, the frontier velocity (length of the longest product divided by reaction time) was 15 nt/s (lanes 2 and 5); however, it dropped slightly to 13 nt/s (lane 8) when 2 mM ATP was used. In the presence of 4 mM ATP (lane 8), the velocity was reduced to 6 nt/s, and it was reduced to 2.5 nt/s when 6 mM ATP was added to the reaction (lane 11).
DNA Replication Dependence on ATP␥S and Rescue by ATP-Unlike the bacteriophage T4 system where the ATPase activity of the clamp loader is dramatically reduced when the replicative polymerase was added to form a holoenzyme complex composed of stoichiometric clamp and polymerase (19,20), Pol ␦ had no effect on the ATP hydrolysis activity of RFC. This phenomenon suggests that RFC might load PCNA onto the P/T  junction of DNA continuously in an ATP-dependent manner. To test this hypothesis, the standard reaction was initiated and carried out for 1 min, chased with different concentrations of ATP␥S with [␣-32 P]dCTP, and then quenched at different time points. The rates of nucleotide incorporation into the extension of primer of ssM13 DNA were measured and shown in Fig. 2. A control experiment verified that ATP␥S cannot support the formation of a catalytically competent holoenzyme (supplemental Fig. S2, lanes 2-6). When no ATP␥S was present, the reaction velocity was 4.0 nt/s, which agreed well with a recent study (14). In the presence of 1, 3, and 5 mM ATP␥S, the nucleotide incorporation rates were 2.0 Ϯ 0.2, 1.  Fig. 3. A control exper-iment in the absence of RFC showed that the DNA synthesis was negligible (supplemental Fig. S3, lane 2). The rescue experiment results showing that the ATP␥S-mediated inhibition of DNA synthesis was relieved by increasing the concentrations of ATP are shown in supplemental Fig. S2, lanes 7-16. DNA Replication Chased with Pol ␦ AA -To examine the processivity with human Pol ␦ holoenzyme and its dissociation from DNA, a catalytically inactive Pol ␦ AA mutant was prepared to chase the standard reaction as was carried out previously in the T4 phage system (21) and is shown in Fig. 3. The rate of dNMP incorporation decreased with increasing amounts of the chase mutant (Fig. 3). When no Pol ␦ AA was added, the DNA synthesis rate was 3.9 Ϯ 0.3 nt/s; however, when an equal amount of Pol ␦ AA was added to chase the reaction, the velocity dropped to 2.0 Ϯ 0.2 nt/s. When a 2-fold excess of the mutant was used, the rate was further reduced to 0.4 Ϯ 0.1 nt/s. No The measured (gray) and calculated (black) rates were plotted against ATP␥S concentration. The experiments were carried out in triplicates to obtain the average and S.E. DNA synthesis was detected when a 4-fold excess of Pol ␦ AA was used to chase the reaction. Theoretically, the DNA synthesis rates were calculated as 4.0, 2.0, 1.3, and 0.5 nt/s when no or 1-, 2-, or 4-fold dead mutant chased the standard reaction, as shown in gray in Fig. 3. The chase experiment with 4-fold excess of Pol ␦ AA was carefully examined for DNA products (supplemental Fig. S5) and revealed a processivity of 150 Ϯ 50 bp.
Human Polymerase ␦ Holoenzyme Is Not Processive-The polymerase chase experiment showed that the human Pol ␦ holoenzyme was not stable and that reassembly of the holoenzyme was indispensable for continued incorporation of nucleotide to extend the primer. The low resolution of the chase experiments precluded the accurate determination of the processivity of human Pol ␦ holoenzyme. To measure the processivity, we conducted the polymerase dilution experiment as well as its simulation by Monte Carlo methods ( Fig. 4 and supplemental Fig. S4). The DNA products became shorter when the reaction was diluted 2-, 4-, and 8-fold. The simulation results indicated that the weak association and fast dissociation rate constants are responsible for the distributive behavior of human Pol ␦ holoenzyme. The calculated processivity is 140 bp (Table S1), which is in excellent agreement with the experimentally determined value of 150 Ϯ 50 bp as noted above. The calculated velocity of the holoenzyme is 6.6 nt/s, which is consistent with the measured rate above (4.0 Ϯ 0.4 nt/s).

DISCUSSION
It is estimated that the length of Okazaki fragments is 100 -250 bp in eukaryotic cells, suggesting that a rapid and highly processive lagging strand replicative complex is not necessary. However, the S. cerevisiae Pol ␦ holoenzyme has a processivity greater than 5 kb (7) and an approximated rate of 60 -150 nt/s (7,22,23). To the contrary, the human Pol ␦ holoenzyme appears to act distributively on an ssM13 template (13, 24 -27). In this study, we measured in vitro these parameters and other properties of DNA synthesis by human Pol ␦ holoenzyme in detail.
PCNA Is Loaded onto P/T Junction Continuously-To understand how the human Pol ␦ holoenzyme forms, ATP hydrolytic activities of RFC with different complexes were examined to elucidate the pathway of Pol ␦ holoenzyme assembly. RFC, by itself, showed very low background ATP hydrolysis activity (0.065 Ϯ 0.01s Ϫ1 ). Upon the addition of a DNA substrate, the ATPase activity was stimulated 4-fold and further stimulated 2.5-fold (0.68 Ϯ 0.06s Ϫ1 ) by adding an equal amount of PCNA (250 nM). In the T4 bacteriophage holoenzyme formation, when the DNA polymerase (gp43) was introduced, the maximal ATPase activity was shut down to a background level, accounted for by a stable clamp-polymerase holoenzyme formation (19,20). However, when human Pol ␦ was added to the PCNA loading system, the ATPase activity remained unchanged, indicating that the Pol ␦-PCNA complex is labile and rapidly exchanges.
ATP Is Necessary to Maintain the Holoenzyme Complex-ATP␥S has been reported to load PCNA in a nonproductive fashion and to prevent the RFC reloading process (11,13,28). To confirm that RFC is required for combined PCNA loading after the preassembly of Pol ␦ holoenzyme, ATP␥S was added to the standard reaction. The DNA synthesis experiments were carried out under optimized conditions of protein concentrations (supplemental Fig. S1). With increasing concentrations of ATP␥S, the DNA synthesis rate decreases and fits the theoretical curve (V ϭ V 0 ϫ [ATP]/([ATP] ϩ [ATP␥S])) very well, suggesting that ATP␥S binds to RFC with similar affinity as ATP and that the binding is reversible. The inhibition by ATP␥S was rescued, when provided with enough ATP (supplemental Fig. S2). Our result is not consistent with an earlier report that 40-fold ATP␥S had no effect on DNA synthesis after the preassembly of calf thymus Pol ␦ holoenzyme (29), but agrees well with the results from an experiment with the human Pol ␦ holoenzyme by Masuda et al. (13) that demonstrated the complete arrest of DNA primer elongation in the presence of ATP␥S. The concentration-dependent inhibition by ATP␥S demonstrated that ATP hydrolysis is necessary for the full activity of human Pol ␦ holoenzyme in DNA synthesis that acts primarily through fueling the activity of RFC to continually load PCNA.
Pol ␦ Readily Dissociates from Holoenzyme-To detect exchange between Pol ␦ polymerase as DNA-bound holoenzyme and Pol ␦ polymerase in solution, Pol ␦ AA was utilized as a trap to chase the primer extension reaction. If the polymerase of bound Pol ␦ holoenzyme freely exchanges with solution polymerase, the rate should fit the equation, Fig. 3. However, if the holoenzyme resists exchange with Pol ␦ AA , the velocity should be higher than that predicted by the free exchange equation. The chase experiment results showed that Pol ␦ AA was a very efficient trap, and the rates are even lower than the prediction for the active exchange model, suggesting that Pol ␦ AA might bind DNA more tightly than that of the wild type. The Pol ␦ AA chase results, combined with RFC ATPase assays, demonstrate that the Pol ␦ holoenzyme is not stable and apt to dissociate and reassemble. The polymerase dilution experiment and the simulation data ( Fig. 4 and supplemental Fig. S4) further support the dissociative behavior of human Pol ␦ holoenzyme. The association and dissociation constants of Pol ␦ at a DNA P/T (Table S1) junction suggest a fast equilibrium between formation and decomposition of the replicative complex, rendering a slow and distributive holoenzyme. Recently, Smith and Whitehouse (30) observed a coupling between lagging strand synthesis and chromatin assembly in S. cerevisiae, which functions as a brake to terminate a processive lagging strand holoenzyme. However, the in vitro experiment on human lagging strand synthesis shows it be distributive, suggesting that the length of Okazaki fragments is an intrinsic property of human Pol ␦ polymerase and does not depend on Flap endonuclease FEN1 and Dna2 helicase/nuclease, enzymes that are necessary for Okazaki fragment maturation (31)(32)(33)(34).
High Concentrations of ATP Inhibit Holoenzyme Assembly and DNA Synthesis Activity-Another interesting result is the inhibition of DNA synthesis by high concentrations of ATP. In the formation of the holoenzyme, ATP is utilized by RFC to load PCNA onto a P/T junction. Generally, with saturating amounts of ATP, high concentrations would have no effect on PCNA loading and subsequent DNA synthesis. However, the frontier velocity of DNA synthesis dropped when the ATP con-centration was increased from 0.5 to 6 mM. This is the first study demonstrating that high concentrations of ATP decreased the DNA incorporation rate by human Pol ␦ holoenzyme. The negative correlation between ATP concentration and holoenzyme activity is most likely due to the inhibition of the RFC loading process, similar to inhibition by ATP of clamp loader activity in the T4 bacteriophage system (35).
In conclusion, the human lagging strand polymerase ␦ in complex with PCNA manifests low intrinsic catalytic activity and processivity in vitro. This property is commensurate with the known size distribution of Okazaki fragments. The accompanying high dissociation rate of the PCNA clamp protein consequently imposes a continuous loading activity on the RFC clamp loader and a high demand for ATP. These measurements serve as a basis for evaluating the effect of additional protein components of the human replisome on DNA replication.