Contrasting Effects of W781V and W780V Mutations in Helix N of Herpes Simplex Virus 1 and Human Cytomegalovirus DNA Polymerases on Antiviral Drug Susceptibility

ABSTRACT

DNA polymerases of the Herpesviridae and bacteriophage RB69 belong to the α-like DNA polymerase family. In spite of similarities in structure and function, the RB69 enzyme is relatively resistant to foscarnet, requiring the mutation V478W in helix N to promote the closed conformation of the enzyme to make it susceptible to the antiviral. Here, we generated recombinant herpes simplex virus 1 (HSV-1) and human cytomegalovirus (HCMV) mutants harboring the revertant in UL30 (W781V) and UL54 (W780V) DNA polymerases, respectively, to further investigate the impact of this tryptophan on antiviral drug susceptibility and viral replicative capacity. The mutation W781V in HSV-1 induced resistance to foscarnet, acyclovir, and ganciclovir (3-, 14-, and 3-fold increases in the 50% effective concentrations [EC₅₀s], respectively). The recombinant HCMV mutant harboring the W780V mutation was slightly resistant to foscarnet (a 1.9-fold increase in the EC₅₀) and susceptible to ganciclovir. Recombinant HSV-1 and HCMV mutants had altered viral replication kinetics. The apparent inhibition constant values of foscarnet against mutant UL30 and UL54 DNA polymerases were 45- and 4.9-fold higher, respectively, than those against their wild-type counterparts. Structural evaluation of the tryptophan position in the UL54 DNA polymerase suggests that the bulkier phenylalanine (fingers domain) and isoleucine (N-terminal domain) could induce a tendency toward the closed conformation greater than that for UL30 and explains the modest effect of the W780V mutation on foscarnet susceptibility. Our results further suggest a role of the tryptophan in helix N in conferring HCMV and especially HSV-1 susceptibility to foscarnet and the possible contribution of other residues localized at the interface between the fingers and N-terminal domains.

IMPORTANCE DNA polymerases of the Herpesviridae and bacteriophage RB69 belong to the α-like DNA polymerase family. However, the RB69 DNA polymerase is relatively resistant to the broad-spectrum antiviral agent foscarnet. The mutation V478W in helix N of the fingers domain caused the enzyme to adopt a closed conformation and to become susceptible to the antiviral. We generated recombinant herpes simplex virus 1 (HSV-1) and human cytomegalovirus (HCMV) mutants harboring the revertant in UL30 (W781V) and UL54 (W780V) DNA polymerases, respectively, to further investigate the impact of this tryptophan on antiviral drug susceptibility. The W781V mutation in HSV-1 induced resistance to foscarnet, whereas the W780V mutation in HCMV slightly decreased drug susceptibility. This study suggests that the different profiles of susceptibility to foscarnet of the HSV-1 and HCMV mutants could be related to subtle conformational changes resulting from the interaction between residues specific to each enzyme that are located at the interface between the fingers and the N-terminal domains.

INTRODUCTION

All antiviral agents currently approved for the treatment of herpes simplex virus (HSV) and human cytomegalovirus (HCMV) infections ultimately target the viral DNA polymerase (1). First-line antiviral agents for the treatment of HSV and HCMV infections include the nucleoside analogues acyclovir and ganciclovir. Both drugs require a first phosphorylation by the virus genome-encoded thymidine kinase (HSV) or UL97 kinase (HCMV) and two subsequent phosphorylations by cellular kinases to be converted into their active forms. The triphosphate forms of acyclovir and ganciclovir are competitive inhibitors of the viral DNA polymerases encoded by the UL30 (HSV) and UL54 (HCMV) genes, respectively. The pyrophosphate analogue foscarnet is a second-line antiviral drug for the treatment of these infections. Foscarnet directly inhibits UL30 and UL54 DNA polymerases by binding to the pyrophosphate binding site and preventing the product release step required for DNA translocation and the promotion of binding of the subsequent deoxynucleoside triphosphate (dNTP), resulting in the cessation of chain elongation.

The Herpesviridae DNA polymerases belong to the family of α-like DNA polymerases (2). Their structure has been likened to a right hand, with the palm, thumb, and fingers domains providing the catalytic site, DNA duplex binding, and nucleotide substrate binding functions, respectively (3–6). The UL30 apoenzyme has been crystallized in the open conformation (7), but crystallographic data for the UL54 DNA polymerase are not available (8). The process of DNA polymerization has been established on the basis of the crystal structure of the bacteriophage RB69 polymerase (gp43) under different conformations (i.e., the apoenzyme state in the absence of DNA binding [6], a binary complex with DNA [9], and the replicating mode of the ternary complex with DNA and nucleotide [5]). It involves the binding of a primer-template duplex DNA to the apoenzyme, which causes the thumb to close down around the DNA while the fingers domain remains in an open conformation. The binding of the incoming nucleotide to this binary complex induces a conformational change in the fingers, which rotate toward the polymerase active site to adopt a closed, catalytically competent conformation. The nucleotide is then transferred onto the 3′ end of the primer strand and the enzyme adopts a pretranslocated state. The pyrophosphate is released, the DNA is translocated, and the enzyme adopts a posttranslocated state. The polymerization cycle can then continue.

Despite a high degree of similarity with the UL30 and UL54 DNA polymerases, the gp43 enzyme is relatively resistant to the broad-spectrum antiviral agent foscarnet (10). It was suggested that the binding site for foscarnet on the UL54 DNA polymerase is located in close proximity to the γ phosphate of the bound nucleotide (11). The triphosphate tail of the incoming nucleotide interacts with the fingers domain of DNA polymerase α, which is formed by two antiparallel α helices called N and P (5, 12). Structure-based alignments of the sequences of helices N and P of UL30 and UL54 DNA polymerases with the sequence of gp43 revealed variable residues in these regions. Chimeric RB69 DNA polymerases were previously engineered to display these critical blocks of amino acids of helices N and P of the HCMV enzyme counterpart (13). The amino acid blocks introduced in both helices and especially in helix N were shown to increase the sensitivity of the chimeric RB69 DNA polymerases to foscarnet compared to that of the wild-type enzyme. In particular, the introduction of the V478W mutation in helix N of the gp43 polymerase appeared to be critical to confer foscarnet susceptibility. Interestingly, the chimeric DNA polymerase with the V478W mutation was shown to crystallize in the closed conformation with the DNA untranslocated, regardless of the addition of foscarnet (13). It was thus suggested that the fingers domain of the mutant gp43 polymerase favors the closed conformation and traps the enzyme in a pretranslocated state, which increases its affinity for foscarnet.

In the study presented here, the V478W mutation that sensitizes the gp43 enzyme to foscarnet was reverted in recombinant HSV-1 (rvHSV-1) and HCMV (rvHCMV) strains by introducing the W781V mutation in the UL30 DNA polymerase and the W780V mutation in the UL54 DNA polymerase, respectively (see the sequence alignment in Fig. 1). Using these recombinant mutant viruses, we further evaluated the impact of the tryptophan located in the fingers domain (helix N) of DNA polymerases on antiviral drug susceptibility, viral replicative capacity, and enzymatic activity in the absence and presence of foscarnet. We also used previously determined X-ray crystal structures of the UL30 apoenzyme along with binary and ternary complexes of gp43 to model key positions in the UL54 DNA polymerase to perform a preliminary analysis of the mechanism of HSV-1 and HCMV susceptibility to foscarnet. Altogether, our results further suggest a role of the tryptophan in helix N of the fingers domain in conferring HCMV and especially HSV-1 susceptibility to foscarnet. Subtle conformational differences resulting from the interactions between residues that are specific to each enzyme and that are located at the interface between the fingers and N-terminal domains might be responsible for the different levels of resistance of HSV-1 and HCMV mutants to foscarnet.

FIG 1.

Sequence alignments of helix N in the DNA polymerases of the RB69 and T4 bacteriophages, HSV-1, and HCMV.

MATERIALS AND METHODS

Cells.

African green monkey kidney cells (Vero; ATCC CCL-81) and human lung fibroblasts (MRC-5; ATCC CCL-171) were obtained from the American Type Culture Collection (Manassas, VA). Vero cells, MRC-5 cells, and human foreskin fibroblasts (HFFs) were grown and maintained in minimal essential medium (MEM; Gibco/Invitrogen, Burlington, Ontario, Canada) supplemented with 10% fetal bovine serum (FBS; PAA Laboratories Inc., Etobicoke, Ontario, Canada).

Generation of recombinant viruses.

A set of five overlapping DNA cosmids (cosmids 24, 32, 48, 51, and 71) containing the entire genome of HSV-1 strain 17 was provided by C. Cunningham (MRC Virology Unit, Glasgow, United Kingdom) (14). The original fragment of cosmid 71 was replaced by three plasmids (pNEB23, pPol6 [containing the UL30 gene], and pNEB10) (15, 16). The W781V mutation was introduced in three independent pPol6 plasmids using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The wild type and three independent recombinant mutant viruses were produced by separate cotransfections of the pool of seven DNA fragments into Vero cells by using Lipofectamine 2000 (Life Technologies, Burlington, Ontario, Canada) in 6-well plates. Reconstituted recombinant viruses were collected when the cytopathic effect was approximately 80 to 90%. The UL30 genes in the mutated pPol6 plasmids and reconstituted viruses were sequenced to confirm that only the desired mutation was introduced.

The HCMV bacterial artificial chromosome (BAC) plasmid derived from the AD169 reference strain (pHB5) (17) was obtained from the laboratory of M. Messerle (Max von Pettenkofer-Institut, Munich, Germany). The Gaussia luciferase (GLuc) gene, driven by the major immediate early (MIE) promoter, was integrated into pHB5 using a two-step bacteriophage lambda Red-mediated recombination protocol (18, 19) to obtain pHB5-GLuc (20). The W780V mutation was introduced into the UL54 gene of two independent pHB5-GLuc mutants using en passant mutagenesis, as previously described (20). MRC-5 cells were transfected by separate electroporation of either the wild type or two independent pHB5-GLuc mutants and the pp71 expression plasmid (pBKCMV82) to enhance the infectivity of the viral DNA (21). Reconstituted recombinant viruses were collected when the cytopathic effect was approximately 80 to 90%. The UL54 genes of the pHB5-GLuc mutants and reconstituted viruses were sequenced to confirm that only the desired mutation was introduced.

Antiviral drug susceptibility assays.

Drug susceptibility testing of recombinant HSV-1 and HCMV was performed in Vero cells and HFFs, respectively, by the plaque reduction assay (PRA) (22, 23). Newly confluent cells seeded in 24-well plates were inoculated with 40 PFU and incubated for 90 min at 37°C in a 5% CO₂ atmosphere. Triplicate wells of infected cells were then incubated with increasing concentrations of foscarnet, acyclovir, or ganciclovir (all from Sigma-Aldrich, St. Louis, MO) in MEM plus 2% FBS containing 0.4% SeaPlaque agarose (Lonza, Rockland, ME) for 3 days (HSV-1) or 7 days (HCMV). Cells were fixed and stained, and the cytopathic effects were counted under an inverted microscope. The concentration of antiviral that reduced the number of plaques by 50% compared to that for control cells without drug (50% effective concentration [EC₅₀]) was determined for each wild-type and mutant recombinant virus.

Viral replicative capacity assays.

The genome copy numbers of the wild-type virus, one representative recombinant HSV-1 mutant, and one representative recombinant HCMV mutant were evaluated by real-time PCR assays at different time points after infection of susceptible cell lines. Briefly, newly confluent Vero cells (96-well plates) and HFFs (48-well plates) were infected with wild-type and mutant recombinant viruses at a multiplicity of infection (MOI) of 0.01 (HSV-1) and 0.05 (HCMV), respectively. Cell lysates were harvested from triplicate wells at 12 h and then daily for 3 days (HSV-1) or daily for 8 days (HCMV) and digested with proteinase K as previously described (24). Viral DNA was extracted from 100 μl of samples with a MagNA Pure LC total nucleic acid isolation kit (Roche Molecular Systems, Laval, Quebec, Canada) and eluted in 100 μl of elution buffer. Real-time PCR assays were performed in duplicate using 5 μl of a 1:100 dilution of extracted DNA, and external standards were run in parallel as previously described (25, 26).

The replicative capacity of the wild type and the three recombinant HSV-1 mutants in Vero cells was determined by the use of a real-time cell analysis (RTCA) system (xCELLigence; Roche Applied Science, Indianapolis, IN) that allows the objective determination of viral cytopathic effects (27). This system measures the impedance using microelectronic sensor arrays integrated in a special cell culture plate (called an E plate [96 wells]). Changes in cellular morphology or growth rate lead to impedance shifts that are measured over time and represented by a dimensionless parameter called the cell index (CI). Newly confluent Vero cells seeded in E plates (96 wells) were inoculated with different recombinant viruses at an MOI of 0.01 and incubated for 90 min at 37°C in a 5% CO₂ atmosphere. The viral suspension was removed and replaced by fresh culture medium (MEM plus 2% FBS), and the CI was recorded every 30 min for 3 days.

The replicative capacity of the wild type and the two recombinant HCMV mutants in HFFs was determined by the GLuc reporter-based assay (20). Briefly, newly confluent HFFs seeded in 24-well plates were inoculated with different recombinant viruses at an MOI of 0.001 and incubated for 90 min at 37°C in a 5% CO₂ atmosphere. The viral suspension was removed and replaced by fresh culture medium (MEM plus 2% FBS). The GLuc activity in cell culture supernatants collected daily for 8 days was measured. The coelenterazine substrate (NanoLight Technology, Pinetop, AZ) was resuspended in acidified methanol at a concentration of 1 mM and then diluted 50-fold in phosphate-buffered saline–5 mM NaCl (28). A volume of 50 μl of coelenterazine substrate was added to 30 μl of cell culture supernatant in a black 96-well plate (Microfluor 2 Black; Thermo, Milford, MA), and the luminescence was measured with a multilabel plate reader (Victor³; PerkinElmer, Waltham, MA) with an acquisition period of 1 s.

Expression of viral DNA polymerase proteins.

The wild-type UL30- and UL54-coding sequences were those of HSV-1 strain 17 (GenBank accession number X14112) and HCMV strain AD169 (GenBank accession number X17403), respectively. The HSV-1 and HCMV DNA polymerase genes were cloned into pCITE4a (EMD BioScience, San Diego, CA) using the NdeI/NcoI and NdeI/SacI restriction sites to generate pCITE-UL30 and pCITE-UL54, respectively. The W781V and W780V mutations were introduced into the UL30 (pCITE-UL30) and UL54 (pCITE-UL54) genes, respectively, using a QuikChange site-directed mutagenesis kit (Stratagene) to produce plasmids pCITE-W781V and pCITE-W780V, respectively. The presence of the desired mutations was confirmed by sequencing the entire UL30 or UL54 gene in the mutated plasmids. Enzymes were expressed by an in vitro transcription-translation system using rabbit reticulocyte lysates (Promega Biosystems, Sunnyvale, CA), as previously reported (29).

Steady-state kinetics of DNA polymerase enzymes and inhibition by foscarnet.

The steady-state kinetics of the UL30 and UL54 DNA polymerases and inhibition by foscarnet were measured by a filter-based assay by using activated calf thymus DNA as the template (11, 30). The reaction was performed in 25 mM Tris-HCl (pH 8.0), 90 mM NaCl, 0.5 mM dithiothreitol, 0.2 mg/ml bovine serum albumin, 5% glycerol, 1 μM [³H]dCTP (specific activity, 20 to 25 Ci/mmol; Moravek, Brea, CA), increasing concentrations (0.3125 to 5 μM) of the other dNTPs (dATP, dGTP, and dTTP), 0.1 mg/ml activated calf thymus DNA (Amersham Biosciences, Piscataway, NJ), and 1 μl of the enzyme-containing transcription-translation mixture in a total volume of 25 μl. The reaction was carried out following the addition of MgCl₂ (final concentration, 10 mM) for 30 min at 37°C (the reaction rates were linear under these experimental conditions). DNA synthesis was stopped by the addition of 600 μl of cold 5% trichloroacetic acid–1% sodium pyrophosphate (NaPP_i), followed by nucleic acid precipitation on ice for 10 min. Samples were then filtered through a Millipore Multiscreen 1.2-μm-pore-size glass fiber C 96-well filtration plate (EMD Millipore, Billerica, MA) and washed 2 times with 5% trichloroacetic acid–1% NaPP_i and once with 99% ethanol. The amount of ³H-labeled DNA retained by each filter membrane was measured by liquid scintillation counting. For inhibition studies, five different concentrations of foscarnet ranging from 0.25 to 2 μM (both wild-type viruses), 1 to 8 μM (UL30 mutant), and 0.5 to 4 μM (UL54 mutant) were added to the reaction mixtures described above. Data obtained in the absence and in the presence of each concentration of foscarnet were fitted to a Michaelis-Menten equation in order to determine the apparent kinetic parameters apparent V_max [V_max(app)] and apparent K_m [K_m(app)] by using the GraphPad Prism software program, version 5.00 (GraphPad Software Inc., San Diego, CA). Data were then represented as a Dixon plot reporting the reciprocal velocity (1/V_max) as a function of the drug concentration. The apparent inhibition constant [K_i(app)] values were estimated from the intercept with the abscissa.

Structural evaluation of UL30 and UL54 DNA polymerases.

The apoenzyme structure of the wild-type UL30 DNA polymerase (PDB accession number 2GV9) (7) was used as the starting model in the absence of a structure of UL54. UL30 was superimposed onto the apoenzyme structure of gp43 (PDB accession number 1WAJ) using least-squares superposition for the palm domain with a root mean square deviation (RMSD) of 1.85 Å over 48 residues from the conserved four beta strands of the fold. The chimeric gp43 carrying the V478W mutation as a ternary complex in the presence of foscarnet (PDB accession number 3KD5) (13) was also superimposed onto the apoenzyme structure of gp43 with an RMSD of 0.6 Å over the 257 C-α residues of the domain. All superpositions were performed using the SuperPose web server (31) within the CCP4 (version 6.3) program suite (32). Structure-based sequence alignments were used to model the residues in the region of interest for the UL54 DNA polymerase, with comparisons and representative mutations made on UL30 using the Coot program (33). All figures were generated using the PyMOL molecular visualization system (Schrödinger, Inc.).

Statistical analyses.

Differences in the replicative capacity of wild-type and mutant recombinant viruses were compared using a two-way analysis of variance (ANOVA). All statistical analyses were carried out using GraphPad Prism software.

RESULTS

Antiviral drug susceptibility testing.

The susceptibility of the wild-type virus and three independent recombinant HSV-1 W781V mutants (rvHSV-1 W781V-1 to rvHSV-1 W781V-3) to foscarnet, acyclovir, and ganciclovir was evaluated in Vero cells by PRA. Table 1 shows that the EC₅₀s of foscarnet, acyclovir, and ganciclovir for the wild-type recombinant HSV-1 strain were 0.06 ± 0.01 mM, 0.05 ± 0.03 μM, and 6.9 ± 1.8 nM, respectively. The foscarnet EC₅₀s for the three recombinant viruses harboring mutation W781V in the UL30 DNA polymerase were 1.8-, 4.3-, and 3-fold higher than the EC₅₀ for the wild type. The acyclovir EC₅₀s for the three independent HSV-1 mutants were 11-, 19-, and 12-fold higher than the EC₅₀ for the wild-type recombinant virus. For ganciclovir, the increases in the EC₅₀s were 2-, 4.6-, and 2.3-fold higher than the EC₅₀ for the wild type. Therefore, the W781V mutation in the UL30 DNA polymerase of HSV-1 confers resistance to all three antiviral drugs.

TABLE 1.

Antiviral drug susceptibility of recombinant HSV-1 wild-type and mutant viruses^a

Recombinant virus	Mean ± SD EC50 (fold change)b
	Foscarnet	Acyclovir	Ganciclovir
rvHSV-1 WT	0.06 ± 0.01 (1)	0.05 ± 0.03 (1)	6.9 ± 1.8 (1)
rvHSV-1 W781V-1	0.11 ± 0.01 (1.8)	0.57 ± 0.14 (11)	14 ± 4 (2)
rvHSV-1 W781V-2	0.26 ± 0.12 (4.3)	0.93 ± 0.42 (19)	32 ± 9.9 (4.6)
rvHSV-1 W781V-3	0.18 ± 0.05 (3)	0.62 ± 0.45 (12)	16 ± 0.8 (2.3)

^aEC₅₀, 50% effective concentration (in millimolar for foscarnet, micromolar for acyclovir, and nanomolar for ganciclovir); WT, wild type.

^bResults are means ± standard deviations from three to seven independent experiments. Fold change, fold change compared to the value for wild-type rvHSV-1 (derived from strain 17). Fold changes in bold correspond to drug resistance, which is typically defined by a 2-fold increase in the EC₅₀ compared to that for the wild type.

The susceptibility of the wild-type virus and two independent recombinant HCMV W780V mutants (rvHCMV W780V-1 and rvHCMV W780V-2) to foscarnet and ganciclovir was evaluated in HFFs by PRA. Table 2 shows that the EC₅₀s of foscarnet and ganciclovir for wild-type recombinant HCMV were 50 ± 7.7 μM and 2.1 ± 0.1 μM, respectively. The foscarnet EC₅₀s for the two recombinant HCMV strains harboring mutation W780V in the UL54 DNA polymerase were 2- and 1.7-fold higher than the EC₅₀ for the wild type. The ganciclovir EC₅₀s for the two recombinant mutants were 1.6- and 1.4-fold higher than the EC₅₀ for the wild type. Therefore, the W780V mutation in the UL54 DNA polymerase of HCMV confers a low level of resistance to foscarnet but does not significantly alter the susceptibility to ganciclovir.

TABLE 2.

Antiviral drug susceptibility of recombinant HCMV wild-type and mutant viruses^a

Recombinant virus	Mean ± SD EC50 (fold change)b
	Foscarnet	Ganciclovir
rvHCMV WT	50 ± 7.7 (1)	2.1 ± 0.1 (1)
rvHCMV W780V-1	99 ± 22 (2)	3.4 ± 0.4 (1.6)
rvHCMV W780V-2	87 ± 24 (1.7)	3 ± 1 (1.4)

^aEC₅₀, 50% effective concentration (in micromolar); WT, wild type.

^bResults are means ± standard deviations from three to six independent experiments. Fold change, fold change compared to the value for wild-type rvHCMV (derived from the AD169 reference strain). The fold change in bold corresponds to drug resistance, which is typically defined by a 2-fold increase in the EC₅₀ compared to that for the wild type.

Viral replication kinetics experiments.

The intracellular viral DNA synthesis of the wild-type virus and representative recombinant HSV-1 (rvHSV-1 W781V-2) and HCMV (rvHCMV W780V-1) mutants in susceptible cell lines was determined by real-time PCR assays. Figure 2A shows that the viral DNA load determined in lysates of Vero cells infected with rvHSV-1 W781V-2 was significantly lower by 70% than that for the wild type on day 3 postinfection (P < 0.001). Figure 2B shows that the genome copy numbers measured in lysates of HFFs infected with rvHCMV W780V-1 were significantly reduced by 72% (P < 0.05), 83%, 88%, 86%, and 79% (P < 0.001 for all other values) on days 4, 5, 6, 7, and 8 postinfection, respectively, compared to the genome copy number for the wild type.

FIG 2.

Kinetics of DNA replication of wild-type (WT) viruses and representative recombinant HSV-1 (A) and HCMV (B) mutants. Vero cells and human foreskin fibroblasts were infected with recombinant viruses at MOIs of 0.01 (HSV-1) and 0.05 (HCMV), respectively. The viral DNA levels in the lysates of infected cells collected at 12 h and daily for 3 days (HSV-1) or daily for 8 days (HCMV) were measured by real-time PCR assays. Results represent the mean ± SEM viral genome copy numbers determined in triplicate wells at each time point. *, P < 0.05 compared to the wild type; ***, P < 0.001 compared to the wild type.

The replicative capacity of the wild-type virus and three independent recombinant HSV-1 W781V mutants in Vero cells was evaluated by RTCA. Cells were infected with the different recombinant viruses at an MOI of 0.01, and the CI was recorded for 3 days. Figure 3A shows that the three clones of HSV-1 harboring mutation W781V in the UL30 DNA polymerase had a defect in replicative capacity compared to their wild-type counterpart. The viral replicative capacity was reduced by 62% (P < 0.05) and 51% (P < 0.001) on days 2 and 3 postinfection, respectively.

FIG 3.

Replicative capacity of wild-type viruses and recombinant HSV-1 (A) and HCMV (B) mutants. (A) Replicative capacity of wild-type virus and three independent recombinant HSV-1 strains harboring mutation W781V in Vero cells determined by real-time cell analysis. Vero cells were infected with the different recombinant viruses at an MOI of 0.01. The cell index (CI) was recorded for 3 days. Data are presented as 1/CI, measured at days 1, 2, and 3, to better represent the decreased replicative capacity of the recombinant mutants compared to that of the wild type. Results are the mean ± SD of sextuplicate determinations and are representative of those of one independent experiment. (B) Replicative capacity of wild-type virus and two independent recombinant HCMV strains harboring mutation W780V in HFFs determined by the GLuc reporter-based assay. GLuc activity was measured in the culture supernatants of HFFs infected with the different recombinant viruses at an MOI of 0.001 collected daily for 8 days. RLU, relative light units. Results are the mean ± SEM from two independent experiments (n = 16 replicates). *, P < 0.05 compared to the wild type; **, P < 0.01 compared to the wild type; ***, P < 0.001 compared to the wild type.

The replicative capacity of the wild type and two independent recombinant HCMV W780V mutants in HFFs was evaluated by the GLuc-based reporter assay. HFFs were infected with the different recombinant viruses at an MOI of 0.001, and the GLuc activity was measured in cell culture supernatants sampled daily for 8 days. Figure 3B shows that the two recombinant HCMV strains harboring the W780V mutation in the UL54 DNA polymerase had altered viral growth compared to that of the wild-type counterpart. The viral replicative capacity decreased by 51% (P < 0.01), 59% (P < 0.001), and 62% (P < 0.001) on days 6, 7, and 8 postinfection, respectively.

DNA polymerase activity and inhibition by foscarnet.

To evaluate the impact of the W781V and W780V mutations on the activity of the UL30 and UL54 DNA polymerases, respectively, we compared the steady-state kinetics of wild-type and recombinant mutant enzymes using a filter-based assay. Table 3 shows that the apparent V_max value of the mutant UL30 DNA polymerase was 6.6-fold lower than that of the wild-type enzyme, whereas its apparent K_m value was 1.6-fold higher. The polymerase activity efficiency (i.e., apparent V_max-to-apparent K_m ratio) of the recombinant mutant enzyme was 11-fold lower than that of the wild-type enzyme. The apparent V_max value of the mutant UL54 DNA polymerase was only slightly reduced (2.3-fold) compared to that of the wild-type enzyme, and the apparent K_m value was increased by 1.8-fold. The apparent V_max-to-apparent K_m ratio of the mutant enzyme was 4.3-fold lower than that of the wild-type UL54 DNA polymerase. By using a similar assay system, we also evaluated the inhibition of the wild-type and mutant UL30 and UL54 DNA polymerase activities in the presence of increasing concentrations of foscarnet. The apparent K_i value for foscarnet against the mutant UL30 DNA polymerase was 45-fold higher than that against the wild-type enzyme. In contrast, the apparent K_i value for the antiviral drug against the recombinant mutant UL54 enzyme increased by only 4.9-fold compared to that against the wild-type counterpart.

TABLE 3.

Steady-state kinetic parameters for recombinant wild-type and mutant UL30 and UL54 DNA polymerases and inhibition of enzyme activity by foscarnet^a

Mutation	Steady-state kinetic parameters				Inhibition by foscarnet
	Vmax(app) (fmol/min)	Km(app) (μM)	Vmax(app)/Km(app)	Fold changeb	Ki(app) (μM)	Fold changeb
UL30
WT	61	0.48	127	1	0.04	1
W781V	9.2	0.77	12	0.09	1.8	45
UL54
WT	8.6	0.55	16	1	0.55	1
W780V	3.7	0.99	3.7	0.23	2.7	4.9

^aResults are representative of those from two independent experiments. WT, wild type.

^bFold change in V_max(app)/K_m(app) values or K_i(app) values compared to the values for the wild-type DNA polymerase.

Structural evaluation of the tryptophan position for UL30 and UL54 DNA polymerases.

The effect of the W781V mutation on HSV-1 DNA polymerase was investigated using the apoenzyme structure of wild-type UL30 (PDB accession number 2GV9) (7) upon least-squares superposition for the palm domain with the gp43 variant carrying the V478W mutation as a ternary complex in the presence of foscarnet (PDB accession number 3KD5) (13). The locations of the fingers residues in relation to the protein domains of gp43 are shown in Fig. 4A, with a focus on key residues involved in likely van der Waals dynamics at the interface between the fingers domain and the N-terminal domain (Fig. 4B). For the variant gp43, the packing interaction between domains involves V478W from fingers helix N, L566 from fingers helix P, and W365 from helix K of the N-terminal domain. The same site is shown in Fig. 4C for the UL30 DNA polymerase, where the equivalent residues are W781 (helix N), V817 (helix P), and V621 (N-terminal domain), respectively. Structure-based sequence alignments were used to model the residues in the same region of the UL54 DNA polymerase, aided by the significant sequence conservation between UL54 and UL30 at the region of interest. This shows that the equivalent residues for the UL54 DNA polymerase are W780 (helix N), F817 (helix P), and I582 (N-terminal domain), respectively (Fig. 4D). The model suggested that the W781V mutation in the UL30 DNA polymerase would likely leave a void in the interdomain packing (due to the smaller V817 and V621 residues) more significant than that for the W780V mutation in the UL54 enzyme due to the bulkier I582 and F817 residues.

FIG 4.

Structural orientation of the fingers domain in gp43, UL30, and UL54 DNA polymerases. (A) Cartoon representation of the complex of RB69 gp43 with a DNA duplex (white surface) showing the conformational change (dashed arrow) of the fingers domain upon nucleotide binding. The fingers domains from the open state (purple) and the closed state (blue) are shown. The remaining domains of gp43 are the N-terminal domain (gray), palm domain (red), thumb domain (green), and exonuclease domain (yellow). The open-state structure is that with PDB accession number 2P5O (45), while the closed state structure is a variant including the V478W mutation (PDB accession number 3KD5) (13) with the active-site magnesium (green balls), 3′ end of the primer strand, and foscarnet. (B) A close-up view of the area in the dashed box in panel A of the interaction between the fingers domain and the N-terminal domain for the RB69 gp43 variant is shown. (C) The apoenzyme structure of HSV-1 UL30 (PDB accession number 2GV9) (7) at the same location described for panel B is shown. (D) The HSV-1 model is used to show the likely positional equivalents to HSV-1 UL30 in HCMV UL54. The boxed arrows above panels B to D suggest the possible relative preference for fingers closure. All figures were made using PyMOL (Schrödinger, Inc.).

DISCUSSION

In the present study, we generated recombinant HSV-1 and HCMV mutants by changing the tryptophan at positions 781 (UL30) and 780 (UL54) in the DNA polymerases to the valine located at an equivalent position in the gp43 enzyme to further evaluate its impact on foscarnet susceptibility. Our results suggest that the tryptophan located in helix N of the fingers domain of DNA polymerases may be involved in the susceptibility of HCMV and especially HSV-1 to foscarnet, as indicated by the increase in the EC₅₀s obtained for revertant viruses (W780V and W781V, respectively) compared to those for their wild-type counterparts. These mutations also resulted in an acyclovir- and ganciclovir-resistant phenotype for the recombinant HSV-1 mutants, whereas the HCMV mutant strains remained susceptible to ganciclovir. In addition, these amino acid changes resulted in an altered replicative capacity for both the recombinant HSV-1 and the recombinant HCMV mutants compared to that of their wild-type counterparts. Of note, neither of these two mutations has yet been detected in clinical samples. Moreover, the tryptophan residue in helix N of the fingers domain is conserved in the DNA polymerases of all human herpesviruses, while a valine is present at this position for the RB69 and T4 phage orthologues.

Contrary to the findings for UL30 and UL54 DNA polymerases, the gp43 enzyme is relatively resistant to foscarnet (10). Like all α-like DNA polymerases, the RB69 enzyme in binary complex with DNA adopts the open conformation for the fingers domain (9). The change of a valine to a tryptophan at position 478 in helix N of the fingers domain sensitizes the RB69 DNA polymerase to the drug (13). The mutant enzyme was cocrystallized in complex with a primer-template terminated with acyclovir in the presence and absence of foscarnet. The V478W mutation likely reduces the population of complexes that exist in the open conformation due to steric hindrance with the tryptophan (W365) located in helix K of the N-terminal domain. The clash between the two tryptophan residues could thus displace the conformational equilibrium of the fingers domain toward the closed conformation, for which foscarnet has a higher binding affinity. The antiviral then inhibits the enzyme activity by trapping the DNA polymerase in a pretranslocated state. The combination of both the V478W and W365V mutations in the gp43 enzyme abrogates the steric clash and restores the resistance to foscarnet. We demonstrated that a mutation opposite the V478W mutation at an equivalent position in the UL30 (i.e., W781V) and UL54 (i.e., W780V) DNA polymerases results in a foscarnet-resistant phenotype of recombinant HSV-1 and, to a lesser extent, HCMV, whereas their wild-type counterparts have a foscarnet-susceptible phenotype, which suggests that this tryptophan is involved in susceptibility to the drug. In the polymerase activity assay, we also showed that the introduction of mutation W781V in the UL30 enzyme resulted in a 45-fold increase in the apparent K_i value compared to that for the wild-type enzyme, which is consistent with a drug-resistant phenotype. In contrast, the inhibition of the mutant UL54 DNA polymerase by the antiviral drug was less significantly (4.9-fold) affected than that of the wild-type enzyme.

Several mutations located in helix N of UL30 (i.e., L774F, L778M, D780N, and L782I) (reviewed in reference 34) and UL54 (i.e., L773V, L776M, V781I, and V787L) (reviewed in reference 35) DNA polymerases were reported to confer resistance to foscarnet, suggesting the role of this helix in the sensitivity of both herpesviruses to this drug. By using a homology model for the UL54 enzyme based on the three-dimensional structure of other α-like DNA polymerases that have already been solved through X-ray crystallography, we have previously examined the role of mutations V781I and V787L in drug resistance (8). Structural evaluation of the wild-type DNA polymerase revealed that the side chain of V781 makes hydrophobic interactions with the nearby C814 located in helix P (and possibly P723 in the palm domain), whereas V787 establishes van der Waals interactions with Q807 from helix P only. The amino acids V781 and V787 may thus help to maintain the local structures of the fingers domain. It was suggested that mutations V781I and V787L introduce bulkier side chains that could result in steric hindrance between the two helices of the fingers domain and could thereby decrease drug susceptibility. It is likely that in the present study the mutant UL30 and UL54 DNA polymerases represent a fine-tuning of the conformational equilibrium of the fingers domain, dictated by several residues at the interface between the fingers domain and N-terminal domain that differ between the two enzymes. It is possible that while the W780/W781 position plays a significant role in the dynamics of the fingers domain for both UL54 and UL30, the presence of the bulkier phenylalanine and isoleucine at this interface for UL54 partially compensates for the tryptophan mutation in that enzyme, while the greater void produced between the two valine residues for the same mutation in UL30 tilts the equilibrium of the fingers domain toward the more open state and reduces the effectiveness of foscarnet. This could also play a role in the significant reduction in the apparent V_max value for the mutant UL30 DNA polymerase compared to that for the wild-type enzyme.

In addition to the resistance to foscarnet, the mutation W781V in recombinant HSV-1 also induces resistance to acyclovir and ganciclovir (14- and 3-fold increases in the EC₅₀s, respectively). Ganciclovir is not an obligate chain terminator like acyclovir (36, 37) but, rather, inhibits the process of DNA synthesis by short-chain termination (38). It is proposed that acyclovir incorporates into the DNA and inhibits the translocation step. In contrast, after incorporation of ganciclovir into the DNA, the translocation step proceeds and the incoming nucleotide is incorporated. The next translocation step may be inhibited because a portion of the sugar that might be required for the efficiency of the reaction is missing in ganciclovir. Thus, the mutation W781V could have different impacts on the susceptibility of HSV-1 to these drugs. It has previously been reported that recombinant HSV-1 harboring mutations at adjacent positions (i.e., D780N and L782I) was resistant to both foscarnet (2.2- and 2.3-fold increases in the EC₅₀s, respectively) and acyclovir (6.1- and 6.4-fold increases in the EC₅₀s, respectively) (15). The mutation W780V in recombinant HCMV slightly decreases the susceptibility to foscarnet but not that to ganciclovir. A recombinant HCMV mutant harboring mutation V781I exhibited a phenotype of resistance to foscarnet (3.8- to 5.2-fold increases in the EC₅₀s) and a phenotype of susceptibility to resistance to ganciclovir (1.0- to 4.5-fold increases in the EC₅₀s) (39, 40). Of interest, V781I is one of the most frequent HCMV mutations, with V715M and L802M being involved in foscarnet resistance in the clinic.

The recombinant HSV-1 and HCMV mutants harboring the respective W781V and W780V mutations in conserved region VI of the UL30 and UL54 DNA polymerases demonstrated viral replicative capacities altered from those of their wild-type counterparts. By using recombinant viruses, we have previously shown that mutations located within conserved regions of the UL30 DNA polymerase (i.e., V715M [region II], L778M [region VI], and N961K [region V]) (15) but not mutation A907V located in a nonconserved region (16) could lead to altered viral replication kinetics. Similarly, UL54 DNA polymerase mutants, such as foscarnet-resistant strains (e.g., with T700A, V715M, and E756D/K mutations [all in region II] and the T821I mutation [in region III]), usually exhibit attenuated or slow growth phenotypes in cell culture compared to the growth phenotypes of their wild-type strains (30, 41–43). We suggest that the subtle changes in the dynamics of the fingers domains of UL30 and UL54 DNA polymerases induced by the respective W781V and W780V mutations may affect critical steps of the DNA polymerization process, such as primer-template binding, nucleotide binding, the efficiency of nucleotide incorporation, and/or the processivity of DNA synthesis, leading to altered viral fitness.

In conclusion, our data suggest that the tryptophan located at positions 781 (UL30) and 780 (UL54) in helix N of the fingers domain may be involved in the susceptibility of HCMV and especially that of HSV-1 to foscarnet. The different profiles of HSV-1 and HCMV mutant susceptibility to foscarnet as well as the altered viral fitness could be related to subtle conformational differences resulting from the interaction between amino acids, specific to each enzyme, that are located at the interface between the fingers domain and the N-terminal domain. These results further demonstrate the usefulness of the bacteriophage RB69 polymerase gp43 as a surrogate enzyme for HSV-1 and HCMV in the discovery and development of new antiherpetic drugs (44).