Potent and Selective Inhibition of Human Cytomegalovirus Replication by 1263W94, a Benzimidazole l-Riboside with a Unique Mode of Action^†

Chemicals and reagents.

BDCRB and its l-ribofuranosyl analog (3858W92) were synthesized according to previously published procedures (26). Compound 1263W94 (Fig. 1) and its d-ribofuranosyl and carbocyclic analogs (3322W93 and 2916W93, respectively) were synthesized according to procedures previously described (4). [³H]-1263W94 (14.9 Ci/mmol) was synthesized by Moravek Biochemicals, Inc. (Brea, Calif.). GCV triphosphate and the 1263W94 5′-mono- and triphosphates were prepared by the previously described chemical method (19).

[α-³²P]deoxynucleoside triphosphates ([α-³²P]dNTPs) were obtained from Amersham (Piscataway, N.J.). Cell media and reagents were obtained from Gibco-BRL (Grand Island, N.Y.), except as noted.

Cell and viral culture.

HCMV strain AD169 was obtained from the American Type Culture Collection (Rockville, Md.) and was the laboratory strain used in characterization of the antiviral activity of 1263W94. GCV-resistant recombinants of AD169 were derived from marker rescue experiments with GCV-resistant clinical isolates and wild-type AD169, as reported previously (1, 7, 13, 25). Ten HCMV clinical isolates were obtained from diverse geographical locations within the United States. The HCMV clinical isolates were recovered from immunocompromised patients, including organ transplant patients and human immunodeficiency virus type 1 (HIV-1)-infected patients who had not yet received therapy for HCMV infection. Human T-cell leukemia lines Molt-4, CEM, and CEM CD4⁺ and human B-cell leukemia line IM9 cells were obtained from the American Type Culture Collection. Human diploid fibroblast cells (MRC-5 lung cells, embryonic kidney cells, and human foreskin cells) were from BioWhittaker (Walkersville, Md.). Human umbilical vein endothelial cells (HUVEC) and coronary artery smooth muscle cells (CASMC) were obtained from Clonetics (San Diego, Calif.). Clinical strains were used if they tested negative for the presence of mycoplasma by using the Gen Probe Mycoplasma Rapid Detection System (Fisher Scientific, Pittsburgh, Pa.).

Monolayer cultures of human diploid fibroblast cells were grown at 37°C in Eagle minimal essential medium (MEM) supplemented with either 8 or 2% fetal bovine serum (FBS; HyClone Laboratories, Logan, Utah), 2 mM l-glutamine, 50 U of penicillin G/ml, and 50 μg of streptomycin sulfate/ml (designated MEM 8-1-1 with 8% FBS or MEM 2-1-1 with 2% FBS). Media for HUVEC and CASMC were obtained from Clonetics and cultured according to recommended protocols.

Determination of antiviral activity.

Assessment of antiviral activity and determination of 50% inhibitory concentrations (IC₅₀s) for 1263W94 and other antiviral compounds were made based on three methods of quantitation of virus: (i) measurement of viral plaque formation (plaque reduction assay), (ii) measurement of extracellular infectious virus in a single cycle of viral replication (yield reduction assay), and (iii) measurement of intracellular viral DNA by specific DNA-DNA hybridization after multiple cycles of viral replication. To define the dose-response curves for each antiviral compound assessed, up to eight concentrations of each compound were tested.

Plaque reduction assays were performed with both cell-free virus and cell-associated clinical strains on MRC-5 cells according to standard protocols (24). Plaques were counted by using a binocular microscope at magnifications of ×10 to ×70. IC₅₀s were calculated from the data by using PROBIT, version 79.3 (SAS, Cary, N.C.).

For the virus yield reduction assay, MRC-5 cells were grown to confluence (1.1 × 10⁵ cells/well) in 24-well plates, infected with cell-free virus at a multiplicity of infection (MOI) of ca. 1 and incubated in MEM 2-1-1 at 37°C in the presence or absence of drug. After 72 h, culture supernatants were removed, and limiting dilutions of virus in the culture supernatant were used to infect monolayers of MRC-5 cells. Inoculation media containing residual concentrations of drugs were removed after viral adsorption. Plaques were quantitated as described for the plaque reduction assay. The IC₅₀ values were calculated by fitting viral yield as a function of drug concentration to the Hill equation by weighted linear regression (15, 20).

For assessment of antiviral activity by inhibition of viral DNA synthesis of laboratory strain AD169, 0.7 × 10⁴ MRC-5 cells/well were grown for 3 days to confluence (∼2 × 10⁴ cells/well) in 96-well plates. Medium was removed to leave 20 μl per well, and 150 PFU of AD169 in 25 μl of MEM 2-1-1 was added to each well, for an MOI of ca. 0.008. After a 10-min centrifugation at 1,500 rpm, plates were incubated at 37°C for 90 min to allow viral adsorption to occur, and 180 μl of MEM 2-1-1 containing no drug or containing specified concentrations of antiviral compounds was added to each well. For each antiviral compound investigated, a range of concentrations was chosen to bracket the expected IC₅₀ value. Incubation was continued at 37°C for 6 to 8 days until viral controls without drug showed ∼100% cytopathic effect. Since nearly 99% of the cells were infected by virus produced during the first round of replication, measurement of viral DNA at the end of the second round of replication provides a measure of the antiviral effects of compounds regardless of their mechanism of action. For assessment of antiviral activity of various anti-HCMV compounds against clinical HCMV isolates in a DNA hybridization assay, cell monolayers were infected with cell-associated virus stocks of each clinical isolate corresponding to ca. 100 infectious centers (24).

In both cases, culture medium was then removed by aspiration, and cells were lysed by the addition of 50 μl of lysis buffer (100 mM Tris-HCl, pH 8; 50 mM EDTA; 0.2% sodium dodecyl sulfate; 0.1 mg of proteinase K/ml)/well and incubation for 1 to 2 h at 56°C. The lysate was extracted with phenol, and viral DNA in the resulting aqueous phase was measured by quantitative DNA hybridization as previously described (27). IC₅₀ values were calculated by fitting viral DNA as a function of the drug concentration to the Hill equation as described above.

Selection of drug-resistant virus.

Laboratory strain AD169 was serially passaged in increasing concentrations of 2916W93, beginning with 4 μM 2916W93, approximately twice the IC₅₀, and ending with 80 μM, a concentration higher than the IC₉₀ value for 2916W93. Virus amplification was accomplished after cultures achieved maximum cytopathic effect by passage of either cell-associated or cell-free supernatant virus, depending upon the extent of virus infection and the integrity of the cell monolayer.

Virus that grew in 80 μM 2916W93 at passage 13 was purified by three cycles of plaque purification in the absence of 2916W93 selective pressure. The resulting virus strain, designated 2916^rA, was compared with parent strain AD169 for the ability to replicate in the presence of benzimidazole compounds 1263W94, 2916W93, and BDCRB. GCV was included in the assays for comparison.

Marker rescue experiments.

High-molecular-weight DNA was isolated from virus pellets of strain 2916^rA, and a cosmid library was constructed and screened for HCMV DNA by hybridization against defined fragments of HCMV as described previously (27). Marker rescue experiments were performed as previously described (27) with the cosmids and subfragments from 2916^rA cotransfected with AD169 infectious DNA. After incubation in the absence of drug until maximum cytopathic effect was achieved, 250,000 PFU from each resulting recombinant pool were plated on MRC5 cells and incubated in the presence of 10 μM 1263W94. Incubation was at 34°C, which gave the optimum selective advantage to the resistant mutants. The supernatant virus yield was measured and the cell monolayer was resuspended and used as the inoculum for the second plating.

DNA sequence analysis.

The 2.25-kb DNA fragment encompassing the UL97 open reading frame (ORF) was amplified by PCR from genomic HCMV DNA isolated from wild-type strain AD169, the parent strain 2916^rA, and the 2916^rA-UL97rec virus. The primers used were CPT0 (5′-TCGACGACGCCGTCTA-3′) and CPTZ (5′-CAACCGTCACGTTCCGCG-3′). Sequencing reactions of PCR fragments from the UL97 ORF were performed with primers designed from the published sequence of HCMV AD169 (6) and standard protocols with the ABI 377 fluorescent DNA sequencer (PE Biosystems, Foster City, Calif.). Sequencing data were analyzing by using Sequencher DNA analysis software, version 3.0 (GeneCodes Corporation, Ann Arbor, Mich.).

pUL97 enzyme preparation and assay.

The baculovirus construct BVGSTUL97 (14) encoding the pUL97 protein kinase as a glutathione S-transferase fusion protein was generously provided by D. Coen. The protein was produced and purified exactly as described previously (14). The enzyme was assayed as described, except that sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels from autophosphorylation assays were quantitated after exposure to a phosphorimager screen by using ImageQuant software and, to determine histone phosphorylation, the final reaction mixture was mixed with 150 μl of 0.5% phosphoric acid and applied to a phosphocellulose filter (Millipore MAPHNB010). The filter was washed three times with 300 μl of 0.5% phosphoric acid and air dried. Then, 30 μl of Wallac Optiphase Supermix scintillation fluid was added, and the filter was counted in a Wallac Micro Beta scintillation counter. This filter-binding method reported both autophosphorylation and histone phosphorylation.

The pUL97 protein concentration used was 9 nM in the histone phosphorylation assays and 0.5 μM in the autophosphorylation assays. Reaction times and pUL97 concentrations were such that, in the histone phosphorylation assays, only 5% of the total filter-binding signal was due to autophosphorylation.

Effect of 1263W94 on HCMV DNA synthesis and maturation.

For these studies MRC-5 cells were seeded in 24-well plates at ∼5 × 10⁴ cells/well and grown for 3 days in MEM 8-1-1 to confluence (∼1.1 × 10⁵ cells/well). The cells were infected with AD169 in MEM 2-1-1 at an MOI ranging from 1 to 3 and incubated at 37°C for 90 min to allow viral adsorption. The unadsorbed virus was removed and replaced with 1 ml of MEM 2-1-1. To test the effect of compounds on viral DNA synthesis or maturation, 1263W94, BDCRB, or GCV was added to the medium at the concentrations indicated for each experiment.

For experiments examining DNA synthesis, HCMV-infected cells were treated with drug for 3 to 4 days, and viral DNA was isolated at various times and quantitated by DNA hybridization as described above. For experiments examining the effects of 1263W94 and BDCRB on viral DNA maturation, viral DNA was prepared at various times postinfection and subjected to contour-clamped homogenous electric field gel electrophoresis and DNA hybridization analysis of the proportions of monomeric and concatemeric viral DNA as previously described (27).

A novel method was developed to examine whether 1263W94 inhibited maturation of HCMV DNA, the step in DNA replication and processing affected by BDCRB. The “BDCRB wash-out” was a modification of the experiment described above investigating the effect of 1263W94 or BDCRB on viral DNA maturation. Cells were infected at an MOI of 1.0, and 20 μM BDCRB was added after adsorption. Infected cells were incubated for 4 days to allow maximum accumulation of newly synthesized concatemeric viral DNA. Medium containing BDCRB was then removed, and cultures were washed and refed with medium containing the compound to be tested or with drug-free or BDCRB-containing medium as negative and positive controls, respectively. After 24 h of further incubation, the effects of these inhibitors on the processing and packaging of viral DNA were measured. Since the effects of BDCRB are reversible (26, 27), its removal allows the accumulated concatemeric viral DNA to be rapidly processed to unit length genomes and packaged as infectious virus. Supernatant medium was sampled for measurement of the infectious virus titer. Viral DNA was prepared from the cell monolayers, and the proportions of monomeric and concatemeric viral DNA were measured as described above.

Isolation and assay of HCMV DNA polymerase.

HCMV DNA polymerase was partially purified from lysed HCMV-infected MRC-5 cells by centrifugation and dialysis, followed by anion-exchange chromatography by using a DEAE cellulose column (16). Assay mixtures and methods were the same as previously described (16). The activity of HCMV DNA polymerase was assessed in the presence of 100 μM concentrations of 1263W94, 1263W94-5′-monophosphate (1263W94-MP), or 1263W94-5′-triphosphate (1263W94-TP). Because benzimidazole ribosides could compete with one or more of the natural dNTPs, assays evaluating 1263W94 and the phosphorylated derivatives were performed in a series of four reactions, with each reaction having three dNTPs present at saturating conditions and the fourth dNTP present in a concentration at or near its K_m.

Anabolism of 1263W94 in human fibroblast cells.

Confluent monolayers of MRC-5 cells in 162-cm² flasks, with ca. 10⁷ cells per flask, were mock infected or infected with 2 × 10⁷ PFU of HCMV AD169. High-pressure liquid chromatography-purified [³H]-labeled 1263W94 (diluted to 1 Ci/mmol) in dimethyl sulfoxide was added to the MEM 2-1-1 at a final concentration of 0.5 μM. At the indicated times after infection the cells were washed and extracted by suspension in 11 ml of ice-cold 80% (vol/vol) acetonitrile in water at −70°C overnight (3). The cells were removed by centrifugation, the supernatant liquid was evaporated to dryness, and the extract was dissolved in water. 1263W94 and its potential anabolites were separated on a Partisil 5 SAX RacII anion-exchange column (Whatman, Inc., Clifton, N.J.) at pH 5.0 with a gradient of KCl. Because 1263W94 was not well resolved from the monophosphate form, the samples were also analyzed on a μBondapak C18 column (Waters Corp., Milford, Mass.) at pH 3.5 with an acetonitrile gradient. In both cases fractions were collected, scintillation fluid was added, and the amount of ³H was counted.

Cytotoxicity assays.

Cytotoxicity of 1263W94, BDCRB, and GCV to bone marrow progenitor cells was determined by evaluating growth inhibition (i.e., CFU of granulocytes-macrophages and burst-forming unit-erythroid [see Table 8]), as previously described (11). Cytotoxicity to human leukemia cell lines was determined as previously described (22).

Antiviral activity of benzimidazole compounds.

The antiviral activity of 1263W94 against HCMV strain AD169 grown on MRC-5 lung fibroblast cells is compared to those of other related benzimidazole compounds and of GCV in Table 1. In the benzimidazole series, the most potent compound by DNA hybridization was 1263W94, with a mean IC₅₀ by DNA hybridization ∼4-fold lower than that for GCV. The relative ranking of antiviral potency of the compounds in the series was similar in each of the three antiviral assays employed.

Activity of 1263W94 and GCV against HCMV AD169 was also measured by the yield reduction assay in HUVEC and CASMC. 1263W94 was active at submicromolar concentrations and was ∼5-fold more active than GCV. In experiments performed in triplicate, the mean IC₅₀ values (± the standard error [SE]) for 1263W94 in CASMC and HUVEC were 0.08 ± 0.03 and <0.01 μM (the lowest concentration tested), respectively, compared to the mean IC₅₀ values of 0.52 ± 0.35 and 1.08 ± 0.22 μM for GCV.

To verify that the antiviral activity of 1263W94 was not restricted to the laboratory strain AD169, the antiviral activity of 1263W94 was assessed against 10 clinical isolates of HCMV. All were sensitive to 1263W94, with IC₅₀s obtained in the DNA hybridization assays that ranged from 0.03 to 0.13 μM for 1263W94 and from 0.15 to 1.10 μM for GCV. Similar results were obtained for these 10 clinical strains in the plaque reduction assays, with IC₅₀s ranging from 0.12 to 0.56 μM for 1263W94 and from 0.80 to 7.00 μM for GCV (Table 2).

Activities of 1263W94, GCV, and BDCRB against GCV- or BDCRB-resistant strains were assessed by using the multicycle DNA hybridization assay. The mutations conferring resistance to GCV had previously been mapped to UL97 (1, 7, 13, 25) and to viral DNA polymerase (2), and the BDCRB-resistant mutations to UL89 in the HCMV genome (27). The strains tested included recombinant strains in which the GCV resistance mutations were recombined into an AD169 background. As shown in Table 3, there were no significant differences in IC₅₀s of 1263W94 among seven of the eight strains tested. An eighth strain, 2916^rA, had been selected for resistance to 2916W93 (see below). Notably, these strains that were resistant to BDCRB or to GCV retained sensitivity to 1263W94.

Isolation of drug-resistant virus.

In an effort to confirm that 1263W94 inhibits HCMV through a specific antiviral mechanism that targets a viral gene or gene product, we selected HCMV specifically resistant to 1263W94. Strain AD169 was serially passaged in increasing concentrations of 1263W94 or of the closely related carbocyclic analog 2916W93. The selective pressure resulted in the enrichment of several viral populations capable of variable growth in the presence of the drugs. Plaque purification of the most growth-competent strains selected with 2916W93 yielded strain 2916^rA, which clearly showed a drug-resistant phenotype. Cross-resistance was tested by DNA hybridization assays, which showed strain 2916^rA to be sensitive to both BDCRB and GCV (Table 3). This finding confirms results presented in Gudmundsson et al. (12). Both the BDCRB-resistant strain 1038^r and the GCV-resistant XbaF4 were sensitive to 1263W94.

A cosmid library was constructed from strain 2916^rA, and a fragment identified as S7 was shown to confer resistance to 1263W94 in marker transfer experiments. Analysis of fragment S7 showed that it contained HindIII fragments P, R, S, and T (6), and resistance to 1263W94 was associated only with HindIII fragment S (Table 4). At the same time that these results were obtained, we found that 1263W94 was a potent inhibitor of the HCMV UL97 protein kinase (see below), which is encoded by one of the ORFs on HindIII fragment S. Consequently, the genetic mapping then focused on the UL97 ORF. The 2.25-kb PCR fragment designated PCR-UL97-2916^rA encompassing the UL97 ORF and defined by primers CPTZ and CPTO (see Materials and Methods) was recombined with wild-type strain AD169, and the recombinant progeny virus was found to be resistant to 1263W94 (Table 4). DNA sequencing of the 2.25-kb fragment identified a single nucleotide substitution, T1190G, which would result in a predicted amino acid substitution of Leu to Arg at position 397 of the UL97 polypeptide.

After plaque purification the recombinant virus, now designated 2916^rA-UL97rec, was compared to its parents, the wild-type AD169 and the 2916^rA viruses, in a multicycle DNA hybridization assay for resistance phenotype. The IC₅₀s for 1263W94 were 0.056 ± 0.01 μM for AD169, 0.7 ± 0.3 μM for 2916^rA, and 0.9 ± 0.02 μM for 2916^rA-UL97rec. These data suggest that the mutation in UL97 is sufficient to account for the observed resistance of 2916^rA to 1263W94.

1263W94 inhibits wild-type UL97 protein kinase but not UL97 kinase from the 2916^rA virus.

1263W94 was a potent inhibitor of histone phosphorylation catalyzed by wild-type pUL97 in vitro, with an IC₅₀ of 3 nM, close to one-half of the concentration of active enzyme. In contrast, the IC₅₀ of 1263W94 for histone phosphorylation catalyzed by the UL97 kinase with the Leu397Arg mutation of the 2916^rA virus was increased 20,000-fold to 60 μM. IC₅₀ determinations were not performed for autophosphorylation; however, wild-type pUL97 autophosphorylation was inhibited 99% by 1 μM 1263W94. Autophosphorylation of pUL97 from the 2916^rA virus was inhibited by <5% by 1 μM 1263W94.

1263W94 inhibits HCMV DNA synthesis in infected cells but does not directly inhibit HCMV DNA polymerase.

Viral DNA was measured in HCMV-infected MRC-5 cells during a single-cycle infection in the presence or absence of the antiviral drugs BDCRB, 1263W94, or GCV (Fig. 2). Synthesis of HCMV DNA was unaffected by the presence of 1.5 μM BDCRB, as previously reported (27). However, the continuous presence of 1 μM 1263W94 or GCV, viral DNA synthesis was reduced to <10% of control levels. These measurements were repeated with a range of concentrations of 1263W94 from 0.025 to 0.5 μM and of GCV from 0.5 to 5 μM, permitting the calculation of IC₅₀s for inhibition of DNA synthesis by 1263W94 and GCV at different times postinfection. At 72 h postinfection the IC₅₀ was 0.19 ± 0.07 μM for 1263W94 and 1.9 ± 0.3 μM for GCV; at 92 h postinfection the IC₅₀ was 0.11 ± 0.01 μM for 1263W94 and 0.8 ± 0.13 μM for GCV. The IC₅₀s for inhibition of HCMV DNA synthesis by 1263W94 and GCV in a single cycle of viral replication are similar to those measured for viral replication by any of the standard methods employed. These observations were extended by comparing the inhibition by 1263W94, BDCRB, GCV, and phosphonoformic acid (PFA) (10) of HCMV DNA synthesis and of viral yield at 72 h postinfection (Table 5). For GCV and PFA, the IC₅₀s for DNA synthesis were two- to threefold higher than the IC₅₀s for yield reduction. Similarly, for 1263W94 the IC₅₀s for DNA synthesis were four- to fivefold higher than those for yield reduction; in contrast, for BDCRB the IC₅₀s for DNA synthesis were >750-fold higher than those for yield reduction.

To investigate whether inhibition of viral DNA synthesis by 1263W94 resulted from a direct effect on viral DNA polymerase, HCMV DNA polymerase activity was assessed in the presence of 1263W94, 1263W94-MP, or 1263W94-TP. No significant inhibition of HCMV DNA polymerase was observed in the presence of 100 μM concentrations of any of the compounds tested (Table 6). GCV triphosphate inhibited this HCMV polymerase preparation with a K_i value of 0.43 μM. Thus, 1263W94 inhibits viral DNA synthesis but by a mechanism different from that of GCV.

Phosphorylation of 1263W94 to 1263W94-MP or 1263W94-TP does not occur in uninfected or HCMV-infected human fibroblasts.

To determine whether human fibroblast cells produce phosphorylated anabolites of 1263W94, both uninfected and HCMV-infected MRC-5 cells were cultured in the presence of radiolabeled 1263W94. Anion-exchange and reversed-phase high-pressure liquid chromatography were used to examine cell extracts prepared at 0, 48, and 96 h postinfection. The sensitivities of the methods employed were such that intracellular concentrations of as little as 10 nM would have been detected. No phosphorylated anabolites of 1263W94 were observed in either uninfected or HCMV-infected cells (data not shown).

Effect of 1263W94 on HCMV DNA maturation.

Previous reports demonstrated that BDCRB blocks HCMV replication by inhibiting the maturational cleavage of high-molecular-weight viral DNA concatemers (27), an obligatory process, conserved in all herpesviruses, that occurs in conjunction with viral DNA packaging (28). To determine whether 1263W94 inhibits this process, the accumulation of high-molecular-weight viral DNA in HCMV-infected cells was measured in the presence or absence of 1263W94 (Fig. 3). The total amount of HCMV DNA synthesized decreased with increasing concentrations of 1263W94, but there were only minor changes in the extent of conversion of the precursor HCMV concatemers to mature unit-length viral DNA. In contrast, BDCRB did not inhibit the total amount of HCMV DNA synthesized but clearly inhibited the conversion of HCMV concatemers to monomers in a dose-dependent manner.

The “BDCRB washout” method was also employed to examine more directly the effects of 1263W94 on DNA concatemer cleavage. As shown in Fig. 4 and as quantified in Table 7, the exposure of infected cells to 1263W94 only after removal of the BDCRB-mediated block in viral maturation did not prevent subsequent production of mature HCMV DNA and infectious virus. BDCRB inhibited both measures of viral maturation, with IC₅₀ calculated to be 0.1 μM.

Cytotoxicity results.

Like other benzimidazoles (21), 1263W94 was significantly less toxic to human bone marrow progenitor cells than was GCV (Table 8). Growth inhibition studies with 1263W94 were also conducted in three human T-cell leukemia lines (Molt-4, CEM, and CEM-CD4⁺) and one human B-cell leukemia line (IM9). The IC₅₀s ranged from 35 μM for MOLT-4 cells to 72 μM for CEM-CD4⁺ cells (data not shown).