Processive Replication of Single-stranded DNA Templates by the Herpes Simplex Virus-induced DNA Polymerase*

The DNA polymerase encoded by herpes simplex virus 1 consists of a single polypeptide of M, 136,000 that has both DNA polymerase and 3’+5’ exonuclease activities; it lacks a 5‘+3‘ exonuclease. The herpes polymerase is exceptionally slow in extending a synthetic DNA primer annealed to circular singlestranded DNA (turnover number -0.25 nucleotide). Nevertheless, it is highly processive because of its extremely tight binding to a primer terminus (& e 1 nM). The single-stranded DNA-binding protein from Escherichia coli greatly stimulates the rate (turnover number -4.5 nucleotides) by facilitating the efficient binding to and extension of the DNA primers. Synchronous replication by the polymerase of primed single-stranded DNA circles coated with  the  singlestranded DNA-binding protein proceeds to the last nucleotide of available 5.4-kilobase template without dissociation, despite the 20-30 min required to replicate the circle. Upon completion of synthesis, the polymerase is slow in cycling to other primed single-stranded DNA circles. ATP (or dATP) is not required to initiate or sustain highly processive synthesis. The 3‘+5‘ exonuclease associated with the herpes DNA polymerase binds a 3’ terminus tightly ( K , < 50 nM) and is as sensitive as the polymerase activity to inhibition by phosphonoacetic acid (Ki -4 h ~ ) , suggesting close communication between the polymerase and exonuclease sites.

a single-stranded DNA-binding protein (2). Partial purification of the herpes DNA polymerase has shown it to be an approximately 150-kDa polypeptide (3) in good agreement with the molecular mass of 136 kDa predicted from the nucleotide sequence of the gene (4,5). As a first step in our analysis of HSV-1 DNA replication, we have purified the herpes-induced DNA polymerase to homogeneity and examined the dynamics of its replication of ssDNA templates.
A second viral encoded protein known to be essential for HSV-1 DNA replication, ICP8 (6)(7)(8), binds ssDNA tightly and cooperatively (9) and is therefore analogous to the prokaryotic single-stranded DNA-binding proteins typified by Escherichia coli SSB and T 4 gene 32 protein (1). The interaction of ICP8 with the DNA polymerase in the presence of single-and double-stranded DNA templates is the subject of the accompanying paper (10).

EXPERIMENTAL PROCEDURES
Materials-Unlabeled and labeled nucleotides were purchased from Pharmacia P-L Biochemicals and Amersham Corp., respectively. dAMP-PNP was a gift from Dr. B. Alberts (University of California, San Francisco). $X and Ml3Goril viral DNAs were prepared as described (11); all viral DNA concentrations are expressed as DNA molecules and were calculated using an AZW of 1 as equivalent to 36 pg/ml. (dA)lm and (dT)17 were purchased from Pharmacia P-L Biochemicals. Calf thymus DNA, purchased from Sigma, was activated as described (12). SSB (4 X lo4 units/mg) (13) was a gift from Dr. D. Soltis (this department). DNA polymerase 111 holoenzyme fraction V (7 X lo6 units/mg) was prepared as described (14). E. coli DNA ligase was prepared as described (15). T4 DNA polymerase and T4 polynucleotide kinase were obtained from the United States Biochemical Co. Phosphonoacetic acid was obtained from Sigma. Bio-Gel A-1.5m and protein molecular weight markers were obtained from Bio-Rad. Plastic-backed polyethyleneimine-cellulose sheets (Polygram MN300) were obtained from Brinkmann Instruments; Centricon 10 was from Amicon.
Cells and Viruses"RA305 (16), a thymidine kinase-deficient mutant of HSV-1 [F], was used to infect roller-bottle cultures of Vero cells using a multiplicity of infection of 5-10 plaque-forming units/ cell.
Purification of HSV-1 DNA Polymeruse-The herpes-induced DNA polymerase was purified by a modification of the method of Powell and Purifoy (3). The steps in the purification up to chromatography on phosphocellulose have been described (17). Briefly, nuclei were prepared from 50 roller bottles of infected cells (32 g, wet weight) and extracted with 1.7 M NaC1. The DNA was removed by ultracentrifugation at 100,000 X g, and the supernatant (200 ml) was dialyzed for 6 h against two changes (2 liters each) of buffer B and loaded onto a phosphocellulose column (16-ml bed volume) equilibrated with buffer B. The phosphocellulose column was eluted with a linear gradient from 0.1 to 0.6 M NaCl in a total volume of 200 ml of buffer B. Herpes DNA polymerase activity eluted at approximately 0.3 M One unit is eaual to 1 pmol of DNA synthesis in 10 min at 37 "C.
Relative to the combined fractions Ia and Ib (before dialysis).
NaC1. Polymerase fractions were pooled, dialyzed, and applied to a 5-ml ssDNA-cellulose column (prepared as described in Ref. 18) equilibrated with buffer B containing 30% (w/v) glycerol. The column was eluted with a linear gradient from 0.1 to 1.0 M NaCl in a total volume of 100 ml of buffer B containing 30% (w/v) glycerol. DNA polymerase eluted at approximately 0.3 M NaCl. Active fractions were pooled, dialyzed against buffer C containing 5% (w/v) glycerol, and concentrated by ultrafiltration in a centrifuge (Centricon 10) to 0.75 mg/ml in a total volume of 400 pl. The concentrated DNA polymerase fraction was divided into two equal portions, and each was loaded onto a 10.6-ml linear 10-30% gradient of glycerol in buffer C. The glycerol gradients were centrifuged a t 40,000 rpm in a Beckman SW41.Ti rotor for 40 h at 4 "C. After centrifugation, 0.2-ml fractions were collected from the bottom of the tubes. The purification is summarized in Table I. Fractions at each step of purification were also analyzed by electrophoresis in a 7.5% SDS-polyacrylamide gel, and proteins were visualized by staining with Coomassie Blue (19). Protein concentration was measured by a modification of the method of Bradford (see Ref. 20) using bovine serum albumin as a reference. The concentration of active DNA polymerase molecules was determined from the amount of SSB-coated singly primed circular @X ssDNA that was replicated by a given amount of polymerase in 40 min at 30 "C (assuming one polymerase molecule/primer terminus). The number of circles replicated was calculated from the total amount of DNA synthesis, the specific activity of nucleotides, and the size of the primed ssDNA circle (21) (i.e. moles of nucleotide polymerized per 5386).
Templates-The sequence and synthesis of the deoxyoligonucleotide 15-mer primers are described in a previous report (22). In most of these studies, the @X ssDNA was primed with primer 1; in one experiment, the ssDNA was primed with primer 4 (see text). The M13Goril ssDNA was primed using primer 5. Hybridization of synthetic primers (2.1 p~ 15-mer) to viral ssDNA (160 nM circles) was in 10 mM Tris-C1, 0.3 M NaCl, and 0.03 M sodium citrate (final pH 8.5). The hydridization mixture was heated briefly to 100 "C, allowed to cool to room temperature over 20 min, and incubated a further 1 h at 30 "C.
Primed @X ssDNA labeled at the 5' terminus of the synthetic DNA primer was made by labeling the primer with T4 polynucleotide kinase and [ T -~~P I A T P , and then hybridizing the labeled primer to @X ssDNA, followed by filtration on Bio-Gel A-1.5m in 10 mM Tris-C1 (pH 7.5), 0.5 mM EDTA, 100 mM NaCl to remove excess labeled ATP and primer.
Specific details of DNA synthesis on singly primed @X ssDNA are given in the figure legends. Reactions were incubated at 30 "C, and samples taken at the times indicated were quenched by adding them to an equal volume of 1% SDS, 40 mM EDTA. The samples, after quenching, were usually divided into two parts. One-half was analyzed for DNA synthesis by measuring the total nucleotide incorporated into acid-insoluble material as described (22). (The values for DNA synthesis refer to the entire reaction mixture). The other half was used for the analysis of the replication products by electrophoresis in neutral 0.8% agarose gels as described (22). For autoradiography, dried gels were exposed to Kodak XAR-5 x-ray film.
Measurement of3"Hj' Exonuclease Actiuity-Synthetic DNA hairpin templates containing either paired or unpaired 3"labeled termini were used as substrates for measurement of 3'+5' exonuclease activity. Reaction mixtures (12.5 pl) contained 3.1 pmol of hairpin template, 83 fmol of DNA polymerase, and 60 mM NaCl in buffer A. When present, dNTPs were 60 p~ each. Incubation was at 30 "C. Samples (1 pl) were quenched instantly upon being spotted onto polyethyleneimine-cellulose strips containing ATP and ADP as carrier and developed in 1 M formic acid and 0.5 M LiCl. Positions of unused substrate (origin) and dTMP or dAMP (near the solvent front) were identified by autoradiography and cut out, and their radioactivity was determined by scintillation counting.

HSV-1 DNA Polymerase Is Stimulated by E. coli SSB-
The HSV-1 DNA polymerase was only minimally active with circular 4X ssDNA (5.4 kb) primed with a synthetic 15-mer ( Fig. 1). Activity was, however, stimulated more than 20-fold upon coating the primed ssDNA with E. coli SSB (Fig. 1). The herpes DNA polymerase was also stimulated (1.5-5-fold) by 250 mM NaCl (data not shown), the extent of stimulation depending upon the relative amounts of DNA polymerase and DNA in the assay.
The influence of ionic strength on polymerase activity when the herpes DNA polymerase was in molar excess over singly primed ssDNA circles is shown in Fig. 2. DNA polymerase activity was stimulated %fold by 150 mM NaCl at early times  in the reaction (2-3 min). However, at later times (after 5 min), the polymerase activity was maximal at 60-90 mM NaCl and then decreased progressively as the salt concentration was increased. Analysis of the products of replication by neutral agarose gel electrophoresis showed that most of the primers were extended to a few unique positions on the circular template (data not shown, but see Fig. 3A (Fig. 1). Moreover, the extent of DNA synthesis was proportional to the amount of DNA polymerase added (data not shown). The limited DNA synthesis suggests a highly processive mode of nucleotide polymerization wherein each polymerase molecule completely extends a DNA primer around the ssDNA circle without dissociation and is slow in cycling to another primed template. Analysis of replication products from singly primed SSBcoated ssDNA by native agarose gel electrophoresis supports the highly processive mode of nucleotide polymerization (Fig.  3B). During the time in which full-length products (RF 11) were formed, most of the primed ssDNA remained unchanged (detected by UV-induced ethidium bromide fluorescence).
The 20 min required for the complete replication of a 4X ssDNA molecule (5.4 kb) yields an average turnover number of 4.5 nucleotides/s/polymerase molecule. The lack of significant radioactivity in the region of the gel between the primed ssDNA and RF I1 product after 30 min indicates that cycling of the polymerase to other primed ssDNA molecules is slow. That the remaining primed circles were effective templates was demonstrated by their replication upon further addition of DNA polymerase (not shown).
The herpes DNA polymerase was also highly processive in the absence of SSB. An agarose gel analysis of the replication products formed with singly primed 4X ssDNA showed that most of the DNA templates had not reacted; nevertheless, discrete, partially replicated species and some fully replicated RF I1 molecules were evident (Fig. 3A).
The processivity of the HSV-1 DNA polymerase was demonstrated in a second type of experiment diagramed in Fig.  4A. The DNA polymerase was preincubated with an 18-fold excess of singly primed +X ssDNA coated with SSB so that each polymerase molecule was bound to a primer terminus. dCTP and dGTP (the 3"terminal nucleotides of the primer and the first 4 nucleotides needed for synthesis) were present during the preincubation to protect the primer from removal of the terminal nucleotides by the 3'+5' exonuclease activity of the polymerase (see below). After a short preincubation period, an excess (%fold over 4X ssDNA circles) of challenge DNA, i.e. singly primed, M13Goril ssDNA (8.6 kb), was added. After further preincubation for either 5 s or 3 min, replication was initiated by the addition of dATP and [cY-~~P] dTTP; after 6 min, further incorporation of radioactivity was prevented by addition of excess unlabeled dTTP. Incubation for an additional 40 min ensured complete replication of templates to which a polymerase molecule was bound at the time unlabeled dTTP was added. Analysis of the replication products by agarose gel electrophoresis is shown is Fig. 4B. Essentially all the label was incorporated into the 4X ssDNA template following the 5-s preincubation period. Hence, the herpes DNA polymerase remained bound to the 4X DNA during synthesis. After a 3-min preincubation period, most of the radioactivity was associated with the 4X DNA, showing only minimal transfer of the herpes polymerase to the challenging M13Goril DNA before initiation of synthesis. A control reaction in which the polymerase was added to a mixture of $X and M13Goril ssDNAs showed about twice as much incorporation of labeled nucleotide into the M13Goril ssDNA as into the $X ssDNA (Fig. 4B, third l a n e ) , consistent with the molar excess of M13Goril over 4X ssDNA circles.
The concentrations of primed DNA circles and HSV-1 DNA polymerase in the experiments of Figs. 3 and 4 provide an upper limit of 1 nM for the equilibrium dissociation constant (&) for polymerase binding to a primer-template. Unlike E. coli DNA polymerase I11 holoenzyme, whose tight binding to DNA and highly processive DNA synthesis requires ATP (or dATP) hydrolysis, there was no effect of ATP or dAMP-PNP on processive DNA synthesis by the HSV-1 DNA polymerase (data not shown).

HSV-1 DNA Polymerase Completely Replicates Primed 4X
ssDNA Circles-To determine the extent to which the circular DNA template is replicated by the herpes DNA polymerase, the products of replication of singly primed 4X ssDNA were treated with E. coli DNA ligase, which requires directly apposing 3'-hydroxyl and 5'-phosphoryl termini to form a phosphodiester bond (23). In the experiment shown in Fig. 5, a portion of the reaction mixture was removed after 10 min, and E. coli DNA ligase and NAD+ were added. Samples were removed after an additional 10, 20, and 30 min of incubation and analyzed by electrophoresis in an agarose gel containing ethidium bromide (Fig. 5B, eighth to tenth lanes). Approximately 70% of the RF I1 products of polymerase action were sealed by the DNA ligase within 30 min and as a consequence co-migrated with a closed circular duplex DNA marker. It therefore appears that synthesis by the herpes DNA polymerase proceeds directly to the 5' terminus of the primer, leaving a nick that can be sealed by DNA ligase.

HSV-1 DNA Polymerase Lacks 5 ' 4 '
Exonuclease Activity-The experiment used to test for 5'+3' exonuclease activity associated with the herpes DNA polymerase is diagramed in Fig. 5A. A 2-fold molar excess of polymerase was added to SSB-coated 4X ssDNA primed with a synthetic 15mer labeled with 32P at its 5' terminus. Samples of the reaction were removed at intervals up to 30 min, the time required to replicate the entire circular template. If the polymerase has significant 5'43' exonuclease activity, it should remove the 32P-labeled 5"terminal nucleotide of the primer upon complete replication of the template. As shown in Fig. 5B, the 5'terminal nucleotide persisted throughout the 30-min period of replication. Quantitation of radioactivity of excised gel slices showed that after 30 min, the 32P at the position of the completed RF I1 circles was approximately 75% that of the primer before the beginning of replication (0 min); 19% of the remaining 32P was present in the smear below the RF I1 products.
The reaction of Fig. 5 was initiated by adding DNA polymerase to a solution containing the primed DNA and all four dNTPs; hence, the replication intermediates appeared as a smear. A discrete band of replication intermediates (as in Fig.  3) is observed only when the polymerase is preincubated for a brief period with the primed DNA and synchronous replication is initiated by addition of the dNTPs.
Herpes DNA Polymerase and 3 ' 4 ' Exonuclease Are Present within the Same Polypeptide-The HSV-1 DNA polymerase and its associated 3'45' exonuclease co-sediment perfectly in a 10-30% glycerol gradient at the position of a 158-kDa marker protein (Fig. 6A), consistent with previous reports (24,25). Moreover, as shown in Fig. 6B, the herpes DNA polymerase is nearly homogeneous (>go%) as judged by Coomassie Blue staining of the gradient fractions following SDS-polyacrylamide gel electrophoresis. The DNA polymerase and 3'+5' exonuclease active sites would therefore appear to reside within the same polypeptide chain. The herpes polymerase showed no detectable primase activity (data not shown).

' 4 ' Exonuclease Has Proofreading Activity-An earlier
report demonstrated that the 3'+5' exonuclease associated with the herpes DNA polymerase has proofreading activity (26). We have examined the proofreading capacity of the 3'+ 5' exonuclease using defined synthetic DNA hairpin templates (57-mers) having either a paired or unpaired labeled 3' terminus (diagramed in Fig. 7A). In the presence of the 4 dNTPs, the 3'+5' exonuclease completely removed the unpaired 3'-terminal nucleotide (Fig. 7B). In contrast, the paired 3' terminus was stable to the 3'45' exonuclease in the presence of dNTPs (Fig. 7B), due presumably to the addition of nucleotides to the paired 3' terminus by the polymerase. The 3'+5' exonuclease hydrolyzed an unpaired 3'-terminal nucleotide at just over twice the rate at which the paired 3'terminal nucleotide was hydrolyzed (Fig. 7C). The rates of hydrolysis of paired and unpaired 3'-terminal nucleotides did not change over the range 1 p M to 50 nM template, setting an upper limit of the K , for hydrolysis at 50 nM (Table 11). Since substrate was present at saturating concentrations, apparent turnover numbers for removal of paired and unpaired 3'terminal nucleotides could be calculated from the observed rates of hydrolysis (Table 11). Use of Mn2+ in place of Mg2+ had no effect on the rate of hydrolysis of the paired 3' terminus but stimulated removal of the unpaired 3' terminus 1.5-fold (Table 11). The effect of the herpes-encoded ssDNAbinding protein, ICP8, on the 3'+5' exonuclease activity (Table 11) is considered more fully in the accompanying report (10). During DNA synthesis with the homopolymer template, (dA)looo. (dT),,, a significant amount of dTMP was generated, approximately 10% of the level incorporated into the homopolymer (compare Fig. 7, B and C). However, in the absence of a primer-template, dTMP was not produced (data not shown). The dTMP is most likely formed upon hydrolysis of paired 3' termini during DNA synthesis as observed previously for the herpes polymerase (25) and for other DNA polymerases that contain 3'+5' exonuclease activity (1).
Phosphonoacetic Acid Inhibits DNA Polymerase and 3 ' 4 ' Exonuclease to Equal Extents during DNA Synthesis-Phosphonoacetic acid inhibits both the polymerase and 3'+5' exonuclease activities of the herpes DNA polymerase with a K j value of approximately 4 p M (Fig. 8, A and B ) . In the absence of dNTPs, the Ki for phosphonoacetic acid inhibition of the 3'+5' exonuclease on the hairpin template with either paired or unpaired 3' termini is 50 p~ (Table 11), the same as the K , for hydrolysis of both paired and unpaired 3'-terminal nucleotides (see above).

DISCUSSION
Despite its slow rate of DNA synthesis, the HSV-1-induced DNA polymerase is strikingly processive. Once bound to its primer-template, the herpes polymerase does not dissociate during the approximately 20 min required to fully replicate a 5.4-kb +X ssDNA circle. This highly processive mode of nucleotide polymerization may be of importance in replicating the 150-kb viral chromosome and, even more important, in the synthesis of multiple copies of the genome in rolling circle DNA replication (2). Although it is not known whether Okazaki fragments are intermediates in replication of the herpes chromosome, the high processivity, replication to a nick sealable by DNA ligase, and lack of 5'+3' exonuclease activity are clearly desirable features in the synthesis of discontinuous DNA fragments.
The rate of fork movement during replication of pseudorabies virus, a member of the herpes virus family, is approximately 50 nucleotides/s at 37 "C (26), similar to that of eukaryotic chromosomes (27, 28). One might therefore anticipate a turnover number of at least 30-40 nucleotides/s for the herpes DNA polymerase at 30 "C. The apparent turnover number of 0.25 nucleotide/s with singly primed 4X ssDNa circles is far too low to sustain productive herpes virus infection (assuming 10,000 copies/cell in 10 h). However, the 20fold stimulation of the herpes DNA polymerase upon coating the ssDNA with E. coli SSB approaches the rate in uiuo. HSV-1 may therefore encode a functional analogue of E. coli SSB. Like E. coli SSB, the herpes-induced ICP8 binds tightly and cooperatively to ssDNA (9), is essential for ongoing DNA replication (8), and is present a t stoichiometric levels (29).
However, despite its similarity to the E. coli SSB, binding of ICP8 to ssDNA completely inhibits the replication of singly primed +X ssDNA by herpes DNA polymerase (10). In contrast, ICP8 markedly stimulates synthesis by the polymerase on duplex DNA templates (10). E. coli DNA polymerase I11 holoenzyme hydrolyzes the P,rphosphodiester bond of ATP (or dATP) to initiate highly processive DNA synthesis (22). In contrast, the monomeric herpes DNA polymerase does not require ATP or dATP. Thus, a complex subunit structure and hydrolysis of the terminal phosphate of ATP or dATP are not essential for highly processive DNA synthesis.
The response of the herpes DNA polymerase to ionic strength is complex. At a molar excess of DNA polymerase to    primer termini, the polymerase is stimulated at early times of synthesis by an ionic strength of 100 mM; however, it is inhibited at later times under these conditions. This behavior may be explained by salt stimulation of a rate-limiting step in polymerase catalysis while enzyme is bound to DNA, which is offset by an increase in the stability of DNA sequencespecific pause sites at salt concentrations above 100 mM (e.g. hairpin structures). However, at substoichiometric ratios of polymerase to primer termini, polymerase activity is stimulated up to an ionic strength of 250 mM. Under these conditions, the polymerase can extend a primer to a sequencespecific pause site and then dissociate from the terminus to gain access to an unused primer terminus. Maximum activity of herpes DNA polymerase on SSB-coated singly primed ssDNA circles at moderate ionic strength (60-70 mM) may result from the combined effect of SSB-induced removal of sequence-specific barriers and an intrinsic stimulation of nucleotide polymerization at this ionic strength.
The exceptionally low K,,, value of the 3'41' exonuclease for a primer terminus may underlie the only 2-fold difference in rates of removal of paired and unpaired 3' termini. The 3'+5' exonuclease of the DNA polymerase isolated from HSV-2-infected cells has been reported to remove a 3'-unpaired terminus six times faster than a 3'-paired terminus in the absence of ~N T P s .~ The basis for the discrepancy between that result and the one presented here is unclear, but may reside in the different assay conditions and/or source of enzyme.   It is not surprising that the 3 ' 4 5 ' exonuclease and polymerase activities share the same polypeptide chain. Other polymerases with associated exonuclease activity have both active sites on a single polypeptide (1). However, the roughly equal sensitivity of the herpes polymerase and 3'+5' exonuclease activities to inhibition by phosphonoacetic acid during DNA synthesis was unexpected. In the case of the E. coli DNA polymerase I large fragment (31), crystallographic analysis has demonstrated that the two active sites are physically separated from each other by approximately 25 A (32). Possibly, phosphonoacetic acid binds to one active site and thereby prevents switching of the primer 3' terminus to the other site.
The herpes polymerase differs in many important respects from DNA polymerase (Y (Table 111). These differences may reflect differences in the complexity of replicating host chromosomes organized into a complex nucleosomal structure as compared with the relatively simple 150-kb HSV-1 genome. In both cases, however, the DNA polymerase is very likely associated with accessory replication proteins that could very significantly alter their catalytic properties.