Effects of Exonuclease Activity and Nucleotide Selectivity of the Herpes Simplex Virus DNA Polymerase on the Fidelity of DNA Replication In Vivo

Abstract

A mutagenesis system was developed for the in vivo study of the fidelity of DNA replication mediated by wild-type herpes simplex virus type 1 (HSV-1) strain KOS and its polymerase (Pol) mutant derivatives PAA^r5, Y7, and YD12. The pHOS1 shuttle plasmid, which contained the SupF mutagenesis marker gene and the HSV ori_s sequence, was used for analysis of the mutation frequency and the mutation spectrum. All three Pol mutants induced significant increases in the mutation frequencies of the target gene, despite the fact that PAA^r5 was previously shown to have an antimutator phenotype by the thymidine kinase mutagenesis assay (J. D. Hall, D. M. Coen, B. L. Fisher, M. Weisslitz, S. Randall, R. E. Almy, P. Gelep, and P. A. Schaffer, Virology 132:26–37, 1984; C. B. C. Hwang and J.-H. Chen, Gene 152:191–193, 1995). Altered spectra of mutated target genes induced by these three mutants were also observed. The relative frequencies of both deletion and complex mutations found in mutants induced by exonuclease-proficient Pols were significantly higher than those induced by exonuclease-deficient Pols. On the other hand, the exonuclease-deficient Pols induced significant increases in the frequency of base substitutions, which comprised predominantly G · C-to-T · A transversions, as well as mutations at additional hot spots. These results suggest that the HSV-1 DNA Pol can incorporate purine-purine or pyrimidine-pyrimidine mispaired bases which may be preferentially proofread by its intrinsic exonuclease activity. Furthermore, the effects of the sequence context of the target gene and the assay method should also be considered carefully in any analysis of replication fidelity.

DNA polymerase (Pol) is the pivotal enzyme involved in DNA replication. It plays the central role in regulating the fidelity of DNA replication by two different means: selection of the correct nucleotides to be inserted into the growing primer terminus and proofreading or editing of the mispaired nucleotide (24). Studies of the fidelity of DNA replication in vitro have been performed on a variety of DNA Pols; however, in vivo characterization of the fidelity of eukaryotic DNA replication has been difficult and little information is currently available (24). Herpes simplex virus (HSV) DNA replication has proven to be an excellent model for the study of DNA replication, since HSV can be genetically manipulated for in vitro and in vivo characterization. For example, HSV pol mutants with altered drug sensitivities have been isolated and characterized. Studies of these mutants have led to the identification of several conserved regions of the Pol enzyme, among a variety of DNA Pols, which are important for their catalytic activities (7).

The thymidine kinase (tk) gene encoded by HSV type 1 (HSV-1) is not essential for viral replication in cell cultures, yet it is required for the activation of certain antiviral drugs, which are nucleoside analogs, in order to inhibit viral DNA replication. Mutant viruses with tk mutations which fail to activate these drugs are then recognized as drug-resistant mutants. This unique property has also led to the invention of the tk mutagenesis assay (12). We have previously applied the tk mutagenesis assay (15) to examine the spectra of mutations of the tk gene mediated by wild-type strain KOS of HSV-1 and pol mutant PAA^r5 (10). These results indicated that the spectra of mutations of the tk gene are attributable to the phenotype of the pol gene. To have a better understanding of the mechanisms by which the HSV Pol might regulate the fidelity of DNA replication, it is important to examine the effects of other mutant Pols on replication fidelity. The analysis of mutated tk genes, however, is laborious, intense, and tedious. We therefore developed and applied a new system to examine the fidelity of HSV DNA replication mediated by a variety of pol mutants in vivo.

In this system, a shuttle plasmid, pHOS1, which contains a SupF mutagenesis target gene and one of the essential elements required for HSV DNA replication (the ori_s sequence) was constructed. This plasmid was used to examine the mutation frequencies and the spectra of mutations induced by wild-type virus strain KOS; its derivatives, including the PAA^r5 mutant (10); and two exonuclease-deficient (exo⁻) mutants, Y7 and YD12 (18). Results obtained by this mutagenesis assay imply the possible mechanisms by which HSV Pol regulates the fidelity of DNA replication.

MATERIALS AND METHODS

Viruses and cells.

Vero (American Type Culture Collection) and Pol A5 cells were grown and maintained as previously described (18). HSV-1 wild-type strain KOS and its pol mutant derivatives PAA^r5, Y7, YD12, and HP66 were propagated as previously described (18). The PAA^r5 mutant contained an arginine-to-serine mutation at amino acid residue 842 within Pol conserved region III (10). The P5Aph⁺K2 recombinant was a derivative of PAA^r5 in which an altered residue 842 was rescued to the wild-type sequence (8). Y7 and YD12 are exo⁻ mutants due to a mutation(s) in the conserved Exo III motif of the exonuclease domain (18). HP66 is a pol null mutant (21).

Plasmids.

Mutagenesis shuttle plasmid pHOS1 was constructed from mutagenesis shuttle plasmid pZ189 (16, 27) by replacing the simian virus 40 sequences with the ori_s sequence of HSV-1. Briefly, a 230-bp SmaI fragment which contains the HSV-1 ori_s sequence was isolated from pOS822 (30), modified with a BamHI linker, and ligated to the 2,446-bp BamHI fragment of pZ189 to form the pHOS1 plasmid, which also contained the SupF mutagenesis marker gene. Plasmid pSupF1 was constructed by self-ligation of the 2,446-bp BamHI fragment of pZ189.

Southern analysis.

About 5 × 10⁵ Vero cells were transfected with 100 ng of pHOS1 DNA by the DEAE-dextran method (16). Twenty-four hours after transfection, cells were either mock infected or infected with the corresponding virus for 8 h. Total DNA was then isolated, purified, and subjected to digestion with either DpnI or HindIII or a combination of the two enzymes, as indicated. HindIII linearizes pHOS1 DNA, and DpnI cleaves the input plasmid DNA containing the methylated A residue in the GATC sequence into smaller DNA fragments (23). DNA samples were then fractionated on a 0.8% agarose gel, transferred to a hybridization membrane, hybridized with the [α-³²P]dCTP-labeled pSupF1 probe, and exposed to X-ray film for 24 h.

Measurement of mutagenesis frequency.

To measure the mutation frequency of the SupF gene induced by HSV Pol, the mutagenesis assay was performed as follows. Briefly, 2 μg of plasmid pHOS1 was mixed with 5 μl of the Lipofectamine transfection reagent (Life Technologies) and transfected into 5 × 10⁵ Vero cells plated on a 60-mm-diameter dish in accordance with the manufacturer’s protocol. After overnight incubation, transfectants were infected with virus at a multiplicity of 10. Eight hours after infection, samples were washed once with TNE buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA, 150 mM NaCl), and then 1 ml of TNE was added. Cells were then scraped off and collected in an Eppendorf tube. After a brief centrifugation in a microcentrifuge, cells were resuspended in 360 μl of TNE, and then proteinase K and sodium dodecyl sulfate were added to final concentrations of 100 μg/ml and 0.6%, respectively. The sample was incubated at 37°C overnight. Total nucleic acids were then extracted with phenol-chloroform and precipitated with ethanol. Total DNA was then digested with HindIII and fractionated on a 0.8% low-melting-temperature agarose gel by electrophoresis in parallel with HindIII-linearized pHOS1. The DNA sample with a size corresponding to that of pHOS1 was isolated, purified, and recircularized by self-ligation with T4 DNA ligase. The restriction enzyme DpnI was then used to cleave the input plasmid DNA (30). The DNA sample was then purified and electroporated into Escherichia coli MBM7070 host cells. E. coli containing pHOS1 with a mutated SupF gene was identified as white colonies on LB agar plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal), isopropyl-β-d-thiogalactopyranoside (IPTG), and ampicillin (16). The mutation frequency was defined as the ratio of the number of white colonies to the total number of colonies recovered.

Characterization of mutated SupF genes.

Mutant pHOS1 DNA was extracted by the cetyltrimethylammonium bromide DNA precipitation method (25). To classify the type of mutations, each mutant plasmid was directly analyzed by sequencing the SupF gene, using the fmol sequencing kit (Promega, Madison, Wis.) and the EcoRI clockwise primer (New England BioLabs). Mutants containing substituted bases or deletions were identified directly from the sequencing results and classified as either point mutants; deletions with one, two, or three bases deleted; or those with deletions of more than three nucleotides of the SupF sequence. Mutants which contained sequences other than that of the SupF gene were classified as complex mutants; these included rearrangements, insertions, or large deletions of the pHOS1 sequence and were not characterized further. Some mutants which failed to yield any sequencing result were examined by EcoRI digestion, which cleaves immediately after the EcoRI clockwise primer binding site. Loss of the EcoRI restriction site was used to indicate that the mutant might also have lost the primer binding sequence due to deletion or rearrangement, and such mutants were also classified as complex mutants.

Statistics.

The significance (p value) of the differences between the mutation frequencies induced by the viruses was examined by tests of differences between proportions (5). The chi-square values of goodness-of-fit tests (9) were also used to compare the patterns of the types of mutations induced by different viruses.

RESULTS

In this study, we developed a mutagenesis assay to measure the mutation frequency and to characterize the mutational spectra of the SupF gene within shuttle plasmid pHOS1. The pHOS1 plasmid was constructed to contain the HSV-1 ori_s sequence, which could direct the plasmid to replicate only in the presence of factors required for HSV DNA replication (3, 4). pHOS1 also contains the tRNA amber suppressor factor (SupF), the ampicillin gene, and the ColE1 sequence. Therefore, pHOS1 can be propagated in E. coli and be induced to replicate in mammalian cells when these cells are infected with HSV. The SupF gene in pHOS1 is able to suppress the amber mutation in the lacZ gene of the host E. coli MBM7070 and form a blue colony in the presence of X-Gal and IPTG. Whenever the SupF gene is inactivated due to mutation, E. coli MBM7070 can only form a white colony (16, 27). This system allows analysis of the fidelity of DNA replication induced by HSV infection.

Replication of the pHOS1 plasmid is HSV Pol specific.

We first examined whether HSV infection could induce the replication of pHOS1 in Vero cells and whether this replication was Pol dependent. Southern blot analyses were performed as described in Materials and Methods. pHOS1 DNA was recovered efficiently from Vero cells that were either mock infected (Fig. 1A, lanes 1 to 3) or HSV infected (Fig. 1A, lanes 4 to 6). However, only HSV infection could induce plasmid replication and the progeny was resistant to DpnI digestion (Fig. 1A, lane 6), whereas in mock-infected cells, pHOS1 did not replicate and was sensitive to DpnI (Fig. 1A, lane 3). Replication was also dependent on the activity of HSV Pol. This was demonstrated by the observation that the pHOS1 DNA recovered from HP66-infected Vero cells was sensitive to DpnI (Fig. 1B, lane 2), while the pHOS1 progeny recovered from KOS- and PAA^r5-infected cells were not (Fig. 1B, lanes 1 and 3). Furthermore, the production of DpnI-resistant pHOS1 could be rescued by pol constructs with Pol activities but not by the plasmid lacking the pol gene or containing a mutant pol gene defective in complementation of viral replication in HP66-infected Vero cells (17, 18). Therefore, pHOS1 can be induced to replicate in Vero cells upon infection with HSV and HSV Pol is essential for replication of pHOS1.

FIG. 1.

HSV pol-dependent replication of the pHOS1 plasmid. (A) Aliquots of the total DNA recovered from mock (lanes 1 to 3)- or KOS (lanes 4 to 6)-infected cells were either untreated (lanes 1 and 4) or treated with HindIII alone (lanes 2 and 5) or HindIII plus DpnI (lanes 3 and 6) and subjected to Southern blot analysis. (B) DNA samples isolated from KOS (lane 1)-, HP66 (lane 2)-, and PAA^r5 (lane 3)-infected Vero cells were digested with both HindIII and DpnI and subjected to Southern blot analysis. The arrows indicate the positions of the linearized pHOS1.

HSV pol mutants induce increased mutation frequencies of the SupF gene.

The efficacy of this mutagenesis system was tested by examining the mutation frequencies induced by both the wild-type and PAA^r5 viruses, since the PAA^r5 mutant has an antimutator phenotype when the tk gene is used as the target (12, 15). pHOS1 progeny DNA was recovered from Vero cells that had been infected with either KOS or PAA^r5 and was electroporated into the host E. coli MBM7070. Surprisingly, instead of inducing a reduced mutation frequency, PAA^r5 Pol induced a threefold increase in the mutation frequency of the SupF gene compared to that of the wild-type KOS virus (Table 1, experiment 1). Further experiments consistently demonstrated that PAA^r5 induced a 1.6- to 3.25-fold increase in the mutation frequencies of the SupF gene in infected cells, which is significantly different from those induced by the wild-type virus strain KOS Pol (P < 0.01) (Table 1). In one experiment (Table 1, experiment 5), the mutation frequency of the SupF gene induced by marker-rescued recombinant P5Aph⁺K2 (8) was indistinguishable from that induced by the KOS virus. This demonstrated that the increase in the mutation frequency induced by PAA^r5 was due to the altered Pol phenotype. Therefore, PAA^r5 Pol exhibited a modest mutator phenotype with regard to the replication of the SupF gene in shuttle plasmid pHOS1 in virus-infected cells.

TABLE 1.

Frequencies of SupF gene mutations mediated by HSV Pols

Expt	KOS	PAAr5	P5Aph+K2	Y7	YD12
1	15/75,800a (0.02, 1)b	21/32,400 (0.065,c 3.3)	NDd	ND	ND
2	25/68,767 (0.036, 1)	42/41,464 (0.1, 2.8)	ND	ND	ND
3	48/63,165 (0.07, 1)	97/53,168 (0.18, 2.6)	ND	ND	ND
4	43/110,691 (0.038, 1)	71/113,104 (0.062, 1.6)	ND	266/141,722 (0.19,c 5.0)	176/114,421 (0.15,c 3.9)
5	110/170,761 (0.064, 1)	117/103,252 (0.113, 1.8), 87/72,637 (0.12, 1.9)e	65/113,477 (0.057,c 0.9), 117/170,701 (0.069, 1.1)e

^aNumber of mutant colonies/total number of colonies recovered in each experiment. 

^bNumbers in parentheses indicate the relative mutation frequency (percent) followed by the fold increase in the relative mutation frequency induced by each pol gene, with the relative mutation frequency derived from KOS defined as 1. 

^cThe mutation frequencies induced by each mutant pol and KOS pol gene are compared and were evaluated by tests for differences between proportions (5). In all cases, the P value is less than 0.01. In experiment 4, the relative mutation frequencies induced by Y7 and YD12 are also significantly different from that induced by PAA^r5 (P < 0.01). 

^dND, not determined. 

^eTwo independent samples each of PAA^r5- and P5Aph⁺K2-infected Vero cells transfected with pHOS1 DNA were tested in experiment 5. 

In addition to the wild-type and PAA^r5 viruses, the effects of exonuclease activity on the replication fidelity of the SupF genes were also examined. A previous study (18) demonstrated that Y7 and YD12 exhibit 800- and 300-fold increases in mutated tk genes in progeny viruses, respectively. To our surprise, Y7 and YD12 induced only five- and fourfold increases, respectively, in the mutation frequencies of the SupF genes (Table 1, experiment 4). The increased mutation frequencies were significantly different from those induced by the KOS (Table 1, experiment 4; P < 0.01) and PAA^r5 (P < 0.01) viruses. Therefore, the PAA^r5, Y7, and YD12 Pols are less faithful in copying the SupF gene than is wild-type Pol in virus-infected cells under the experimental conditions of this study.

To examine whether the mutations found in the SupF gene were due to the transfection process, pHOS1 DNA was isolated from transfected Vero cells at 24 h after transfection. A fraction of the DNA was then transformed into E. coli without DpnI digestion. The mutation frequency was determined to be 0.003% (12 white colonies in a total of 380,200 colonies). The heterogeneity of the input pHOS1 DNA, which could lead to the formation of white colonies, was also determined to be less than 1 in 10⁵ colonies or a frequency of less than 0.001% (17). Therefore, the observed mutations of the SupF gene represent the results of inaccurate replication of HSV Pol, although a very small portion of these mutations might be due to the experimental procedures.

HSV pol mutants induce altered spectra of mutated SupF genes.

The SupF gene mutations induced by each mutant Pol were characterized and classified as either base substitutions or others as described in Materials and Methods. The results are summarized in Table 2. There was no significant difference between the types of mutations induced by the wild-type and PAA^r5 Pols, which were both exonuclease proficient (13). Two exo⁻ Pols, on the other hand, induced types of mutations significantly different from those induced by exo⁺ Pols. The differences were manifested as increases in the base substitutions and decreases in the deletions and complex mutations induced by exo⁻ Pols. For example, 65 and 73% of the mutations induced by Y7 and YD12, respectively, were point mutations, while KOS and PAA^r5 induced point mutations at 37 and 38%, respectively. Similarly, both exo⁻ Pols induced fewer deletion and complex mutations, with the exception of the complex mutations induced by Y7 Pol (Table 2).

TABLE 2.

Classification of mutant SupF genes induced by HSV-1

Strain (no. of samples)	No. (%) of samples with:				Chi-square (P) valuea
	Base substitutions	Simple insertions and/or deletionsb	Deletions	Complex mutationsc
KOS (181)	68 (37)	12 (8)	41 (23)	60 (33)	0
PAAr5 (198)	76 (38)	10 (5)	39 (20)	73 (37)	2.37 (>0.05)
P5Aph+K2 (55)	21 (38)	1 (2)	13 (24)	20 (36)	2.12 (>0.05)
Y7 (72)	47 (65)	4 (6)	3 (4)	18 (25)	27.14 (<0.01)
YD12 (66)	48 (73)	8 (12)	0	10 (15)	46.12 (<0.001)

^aGoodness-of-fit tests (9) were based on chi-square values obtained by comparison of the observed and expected numbers of each type of mutation induced by each Pol. The expected numbers were predicted from the contribution of each type of mutation induced by the KOS Pol. 

^bA mutated SupF gene containing an insertion or deletion of three or fewer nucleotides was classified as a simple insertion or deletion mutant gene. 

^cA mutated SupF gene associated with an increase in pHOS1 DNA size due to insertion or rearrangement was classified as a complex mutant gene. 

Statistical analysis revealed that the patterns of point mutations induced by two exo⁻ Pols were significantly different from those induced by exo⁺ Pols (Table 3). These differences were predominantly represented by increases in some types of transversions induced by the Y7 and YD12 Pols (Table 3). While about one-third of the base changes induced by exo⁺ Pols were transitions with a majority of G · C-to-A · T changes (Table 3), such alterations were significantly less frequent in mutations mediated by both the Y7 and YD12 Pols. The Y7 and YD12 Pols, on the other hand, predominantly induced G · C-to-T · A transversions (62 and 66%, respectively). Such transversions represented only about 37 and 39% of the mutations induced by the wild-type and PAA^r5 Pols, respectively. Therefore, both the Y7 and YD12 Pols could introduce more G · C-to-T · A transversions during DNA replication, which could contribute significantly to the increased incidences of mutations mediated by these mutant Pols. However, the incidence of G · C-to-C · G transversions induced by the exo⁺ Pols was about 20%, which was higher than that of those induced by Y7 and YD12 Pols (6 and 2%, respectively).

TABLE 3.

Classifications of point mutations

Strain (no. of samples)	No. (%) of samples with:								Chi-square (P) valuea
	Transitions			Transversions
	G · C to A · T	A · T to G · C	Total	G · C to T · A	G · C to C · G	A · T to T · A	A · T to C · G	Total
KOS (84)	28 (33)	1 (1)	29 (34)	31 (37)	17 (20)	5 (6)	2 (2)	55 (65)	0
PAAr5 (93)	26 (28)	4 (4)	30 (32)	36 (39)	18 (19)	5 (5)	4 (4)	63 (68)	9.98 (>0.05)
P5Aph+K2 (29)	6 (21)	3 (10)	9 (31)	12 (41)	6 (21)	2 (7)	0	20 (69)	NDb
Y7 (50)	8 (16)	4 (8)	12 (24)	31 (62)	3 (6)	3 (6)	1 (2)	38 (76)	37.55 (<0.001)
YD12 (51)	8 (16)	2 (4)	10 (20)	34 (66)	1 (2)	3 (6)	3 (6)	41 (80)	31.24 (<0.001)

^aGoodness-of-fit tests were based on chi-square values obtained by comparison of the observed and expected numbers of each type of mutation induced by each Pol. The expected numbers were predicted from the contribution of each type of mutation induced by the KOS Pol. 

^bND, not determined. Statistical analysis was impossible due to the small sample size. 

In addition to the differences in the types of mutations (Table 2) and substituted bases (Table 3) mediated by these Pols, the distributions of the changed bases were also significantly different. A schematic of the point mutations analyzed in this study is shown in Fig. 2, which reveals several interesting observations. For example, nucleotides 109, 123, 133, 156, 159, 168, and 169 were common hot spots of mutations induced by these Pols. While changes were also found at position 115 among the wild-type Pol-induced mutations, it appeared that neither the PAA^r5 Pol nor the two exo⁻ Pols induced any change at this position. Whereas changes at position 129 were relatively common among PAA^r5-induced mutations, fewer or no substituted bases were induced by the wild-type or exo⁻ Pol, respectively. Therefore, it seems that nucleotide 115 was the specific hot spot for the wild-type Pol and that both exo⁻ Pols were not prone to induce a base substitution at position 129, which was the hot spot of G · C-to-C · G transversions. The fact that changes at positions 115 and 129 were not identified among P5Aph⁺K2-induced mutations could be due to the fact that relatively fewer base substitutions were analyzed (Table 3, footnote b). In addition, base substitutions at position 135 were only found among PAA^r5-induced mutations (Fig. 2). Furthermore, it appeared that nucleotide 156 was a hot spot of G · C-to-A · T transitions induced by both exo⁺ Pols, since only one G · C-to-T · A transversion was induced by Y7 Pol and no mutation at this position was induced by YD12 Pol.

FIG. 2.

Schematic of mutations within the SupF gene. The top line shows the coding sequence of the SupF gene. Mutations induced by the KOS, PAA^r5, P5Aph⁺K2, Y7, and YD12 pol genes are shown beneath the sequence for each group. Each letter indicates the changed base found in single mutations, while multiple changes found in the same mutant are underlined. The number in parentheses for one mutant isolated from the PAA^r5 group indicates the change at position 77. Symbols: ▵, nucleotide(s) deleted; +, nucleotide inserted at the corresponding location.

The most dramatic difference regarding the distributions of substituted bases was that 34 and 27% of the Y7- and YD12-induced base substitutions, respectively, were clustered between nucleotides 133 and 139 (Fig. 2). Also, about one-fifth of the point mutations induced by both Y7 and YD12 contained substituted bases at nucleotides 160 and 161. Furthermore, while nucleotide 113 could be a potential hot spot for exo⁻ Pols, a change at this position was rarely found in exo⁺ Pol-induced mutations. Interestingly, mutations at these hot spots specific for the Y7 and YD12 Pols were mainly G · C-to-T · A transversions.

It was also notable that relatively more multiple-base changes were found in a single mutation induced by exo⁺ Pols than exo⁻ Pols. Fifteen and 12 mutations induced by KOS and PAA^r5, respectively, were multiple changes in the mutated SupF gene, while only 2 and 4 mutations induced by Y7 and YD12, respectively, were multiple changes. Although these altered bases could occur in a single DNA replication cycle, it was also possible that these mutations could be the result of the multiple steps of a single change.

Despite the different mutational spectra induced by these Pols, base deletion mutations were commonly found in regions containing reiterative bases, with only a few exceptions (Fig. 2). These mutations also included a deletion of ACC at the end of the SupF gene induced by the wild-type Pol and a deletion of TGG at nucleotides 98 to 103 induced by the PAA^r5 Pol.

DISCUSSION

In vivo mutagenesis system using an amplicon containing HSV-1 ori_s.

In this study, we developed a new method to examine the spectra of mutations induced by the wild-type HSV Pol and three mutant Pols in HSV-infected mammalian cells. Since the HSV Pol is the pivotal protein for HSV DNA replication, the progeny of an ori_s-containing amplicon, pHOS1, isolated from HSV-infected cells can be considered the products of the HSV Pol. Characterization of the mutation frequencies and spectra of mutations of the induced progeny pHOS1 DNA therefore will reflect their contributions to replication fidelity. This study also represents, to our knowledge, the first in vivo study of the sequence changes attributable to loss of the exonuclease proofreading activity of a Pol in mammalian cells.

Mutator versus antimutator.

The PAA^r5 mutant was previously shown to have an antimutator phenotype when the tk gene was the target of mutagenesis (12, 15). In contrast, PAA^r5 revealed a modest mutator phenotype when the SupF gene was the target. Although the Y7 and YD12 Pols were confirmed to have mutator phenotypes by this assay, their mutagenic effects were much weaker than those observed in the tk mutagenesis assay (18). These observed differences suggest that the definition of a pol gene as one that has either a mutator or an antimutator phenotype is dependent on the gene being analyzed and the assay method utilized.

Effects of Pol identity on replication fidelity.

It has been demonstrated that Pol identity plays an important role in misinsertion fidelity (1, 11). For example, Pols lacking intrinsic 3′→5′ exonuclease activity, such as Polβ (20) and the exo⁻ Klenow fragment (2), are prone to form transitional mutations rather than transversions. Unlike Polβ and the Klenow fragment, both HSV exo⁻ Pols had significantly fewer G · C-to-A · T transitions (Table 3). In contrast, these two exo⁻ Pols synthesized predominantly G · C-to-T · A or C · G-to-A · T transversions, which represented two-thirds of the substituted bases analyzed (Table 3). This suggested that misinsertion of dTTP opposite C (template) or dATP opposite G by HSV Pol might occur and that the majority of these misinserted bases might be removed by proofreading activity; an HSV Pol that lacked exonuclease activity might fail to eliminate these mispaired bases. Furthermore, both exo⁺ Pols induced significantly more G · C-to-C · G transversions than did exo⁻ Pols; some of these transversions were exclusive for both exo⁺ Pols, i.e., at nucleotide 129 of the SupF gene (Fig. 2). Therefore, HSV Pol might exhibit specific types of the misinsertion fidelity with exonuclease activity critical for correction of certain types of misinserted nucleotides.

A previous study demonstrated that the PAA^r5 Pol exhibited a higher K_m for deoxynucleoside triphospates (dNTPs) (13), which would allow the discrimination of incorrect nucleotides. Although this altered identity might explain the improved fidelity and altered spectra of the mutated tk genes (13, 15), it could not fully explain our observation that this Pol induced a modest increase in the frequency of the mutated SupF gene. However, it did induce a slightly altered spectrum of mutations, such as the lack of a mutation at position 115 and the gain of a mutation at position 135. Together with the position-specific mutations induced by exo⁻ Pols, these differences suggested that mutations induced by the HSV Pol could also be target gene dependent, which is a feature that could be affected by other factors.

Other factors attributable to fidelity of DNA replication.

If the identity of a Pol were the sole factor determining replication fidelity, one would expect to observe similar outcomes of mutations in both the tk and SupF genes replicated by a particular Pol. The dramatic difference between the mutation frequencies of the tk and SupF genes induced by the PAA^r5 Pol suggests that other factors are also involved in the regulation of replication fidelity. Among these, the effects of the sequence context of the target genes could be critical, which had been demonstrated by in vitro studies of other Pols (1, 11). The fidelity of the PAA^r5 Pol in replicating the SupF gene, therefore, could be dominated by its structure and composition. It is also possible that the tk mutagenesis assay is not sensitive enough to detect all of the mutations. In fact, there are known polymorphisms of the tk gene among different strains of HSV-1 which are equally sensitive to the antiviral drugs used for the selection of tk mutants (12, 15). Similarly, silent mutations of the tk gene, which are sensitive to both acyclovir and ganciclovir, have been frequently found in the laboratory (17). Furthermore, the PAA^r5 Pol may replicate more complex mutations in the viral genomic DNA and escape detection if they become lethal to viral replication. Perhaps the effects of sequence contexts of the target gene could dominate the effects of the Pol’s identity.

The different tk and SupF gene mutation frequencies observed also raised a concern about the positions of the target genes, which might influence replication fidelity; the tk gene is in the context of the viral genome, whereas the SupF gene analyzed in this study is in a replicating plasmid. In other words, the replication mode of the target gene may contribute to the differences in fidelity between these studies. To address this issue, both target genes must be analyzed in the same context and replication mode; i.e., the SupF gene is inserted into and replicates as part of the viral genome. Recently, this type of recombinant virus has been constructed in our laboratory. Preliminary experiments revealed that a recombinant virus derived from the KOS strain had a mutation frequency ranging from 0.03 to 0.1% in the SupF gene, whereas the PAA^r5 derivative had a mutation frequency ranging from 0.04 to 0.15% (17). Although further experiments are required to confirm that each white or light blue colony recovered in these experiments contains a mutated SupF gene, it seems unlikely that the PAA^r5 Pol has the antimutator phenotype with regard to replication of the SupF gene, even when it is placed within the tk locus. Nevertheless, it does seem that the mutation frequency of the SupF gene replicated by the PAA^r5 Pol may be independent of its position or context.

HSV-1 itself has been demonstrated to be mutagenic (6, 16, 28), and its mutagenicity is directly mediated by the structural components of the virion and is independent of the expression of the viral genes (6, 28). This raises the possibility that a small fraction of the mutated SupF genes observed in this study resulted from the mutagenic effects of HSV-1 infection. However, such SupF mutants, if any, should be equally represented in each group, since KOS, PAA^r5, P5Aph⁺K2, and the HP66 mutant all exhibited similar mutagenic effects on the SupF gene (28). Additionally, the presence of the HSV ori_s sequence restricts the replication of pHOS1 to the HSV replication machinery. Therefore, the difference in the mutations induced by these viruses observed in this study should not be considered to be due to the mutagenic effects of HSV infection. It does remain a possibility that a portion of the mutations, especially the complex mutations, resulted from the transfection procedures. In fact, 10 of the 12 mutated SupF genes recovered from the control experiment contained either a deletion or a complex mutation in which the altered sequences could not be sequenced by the EcoRI primer (17). To avoid this possibility in the future, it might be better to use a recombinant virus containing the inserted SupF gene or other reporter genes.

HSV Pol may be unique in regulating replication fidelity.

HSV infection has been demonstrated to induce an imbalance in dNTP pools in infected cells (19). Accompanying this effect, increases in mutations and mutants with altered mutational spectra can be expected. Consistent with this is the observation that there are distinct spectra of mutations between the spontaneous mutations replicated by cellular Pols (14, 16, 22, 26) and those induced by HSV infection (16). However, in an environment of perturbed dNTP pools, the wild-type Pol can replicate the SupF gene with an accuracy similar to that of cellular Pols under normal conditions (this study and references 16 and 28). Thus, the replication machinery of HSV, including the Pol, might have evolved to acquire the unique ability to compensate for a dNTP pool imbalance and favor its own replication.

The HSV Pol might have evolved to have certain types of error-prone replication, such as the incorporation of purine-purine and pyrimidine-pyrimidine mispairs to originate the formation of the transversions that, at most, are about 10-fold less frequent than transitions (29). To accommodate this property, the acquisition of exonuclease activity seems necessary for the correction of some of these misincorporated bases to reduce the rates of transversions. Therefore, the exonuclease activity intrinsic to the HSV-1 Pol may play an important role in maintaining the fidelity of DNA replication.

In conclusion, these experiments demonstrated that the replication fidelity of a Pol can be dramatically influenced by the assay method. The sequence context of the target gene, which might also be affected by the assays used, in addition to the identity of a Pol, should be considered as an important factor in the regulation of replication fidelity. Continuous mutagenesis studies, as well as examination of the kinetic parameters of different mutant Pols, would be very useful in better understanding misinsertion fidelity. Furthermore, such studies are also important for an understanding of how drug-resistant HSV mutants develop, which is an important issue for the successful treatment of HSV infections.