Specific Inhibition of Human Cytomegalovirus Glycoprotein B-Mediated Fusion by a Novel Thiourea Small Molecule

Abstract

A novel small molecule inhibitor of human cytomegalovirus (HCMV) was identified as the result of screening a chemical library by using a whole-virus infected-cell assay. Synthetic chemistry efforts yielded the analog designated CFI02, a compound whose potency had been increased about 100-fold over an initial inhibitor. The inhibitory concentration of CFI02 in various assays is in the low nanomolar range. CFI02 is a selective and potent inhibitor of HCMV; it has no activity against other CMVs, alphaherpesviruses, or unrelated viruses. Mechanism-of-action studies indicate that CFI02 acts very early in the replication cycle, inhibiting virion envelope fusion with the cell plasma membrane. Mutants resistant to CFI02 have mutations in the abundant virion envelope glycoprotein B that are sufficient to confer resistance. Taken together, the data suggest that CFI02 inhibits glycoprotein B-mediated HCMV virion fusion. Furthermore, CFI02 inhibits the cell-cell spread of HCMV. This is the first study of a potent and selective small molecule inhibitor of CMV fusion and cell-cell spread.

Human cytomegalovirus (HCMV) is a betaherpesvirus with a 50 to 90% prevalence frequency in the adult human population (42). It is an opportunistic pathogen controlled largely by the immune system, especially cell-mediated immunity. Clinically serious disease occurs primarily in three situations involving deficient immune systems: (i) transplant patients suppressed to avoid graft rejection, (ii) AIDS patients, and (iii) congenitally infected infants. In the first two groups, disease can range from a pneumonitis to multiorgan disease and death. Retinitis and blindness, caused by HCMV, are also common in AIDS patients. Disease in infants is typically due to in utero infection and causes neurological sequelae, ranging from hearing loss to mental retardation.

Treatment of HCMV disease is limited to relatively toxic drugs with that target viral DNA synthesis (16, 57). The most commonly used HCMV drug, ganciclovir, is a nucleoside prodrug phosphorylated by the viral UL97 kinase to the nucleotide analog form and is selectively incorporated into progeny DNA by the HCMV UL54-encoded DNA polymerase. Foscarnet is a pyrophosphate analog. Effective concentrations of foscarnet are much higher than that of ganciclovir. Both of these drugs are nephrotoxic; ganciclovir also causes neutropenia and is teratogenic. Cidofovir is a nucleoside phosphonate that is severely renal toxic. In addition to toxicity issues, there are reports of resistance and cross-resistance to the current HCMV drugs (16, 57). Thus, there is a need for other HCMV drugs that are selective, less toxic, and target other aspects of the HCMV replicative cycle.

To identify novel inhibitors of HCMV and herpes simplex virus (HSV), a portion of a proprietary chemical library was screened by using high-throughput HCMV- and HSV-infected cell assays. A screening hit with low-level activity against HSV was identified that was not active against HCMV (3, 56). Initial synthetic chemistry efforts yielded a related bis-aryl thiourea compound with a low micromolar 50% inhibitory concentration (IC₅₀) against HCMV (CFI01) (Fig. 1A); subsequently, a series of about 1,500 related thiourea analogs were made and assayed. Some of these analogs were low-nanomolar inhibitors of HCMV. CFI02 (molecular weight = 491; Fig. 1B) is a representative molecule from this recently described series that specifically inhibits HCMV (compound 38 in Bloom et al. [3]). We show here that a target of CFI02 is inhibition of HCMV glycoprotein B (gB)-mediated fusion of the virion envelope with the cellular plasma membrane.

FIG. 1.

Chemical structure of HCMV inhibitors CFI01 and CFI02.

MATERIALS AND METHODS

Cells and Viruses.

Human foreskin fibroblasts (HFFs) were described previously (25). Mouse embryo fibroblasts were prepared in house from BALB/c mouse embryos. HCMV strain AD169, murine CMV (MCMV) strain Smith, and varicella-zoster virus strain Ellen were obtained from the American Type Culture Collection. Simian cytomegalovirus and rhesus cytomegalovirus (RCMV) were the generous gifts of Wade Gibson (Johns Hopkins) and Scott Wong (Oregon Primate Center), respectively. Herpes simplex virus type 1 (HSV-1) strain Patton was obtained from Richard Hyman (Penn State Uniersity). RV17044, a recombinant HCMV derived from the AD169 parent, expresses the enhanced green fluorescent protein (GFP) from the US9-US10 intergenic region under the control of the simian virus 40 (SV40) promoter. RV17044 was obtained after cotransfection of recombinogenic plasmid, pHXSB-EGFP, and HCMV strain AD169 DNAs into HFF cells as described previously (26, 28). GFP-positive plaques were selected via fluorescence microscopy and purified to homogeneity by repeated passage in HFFs. The proper recombination event was verified by DNA blot hybridization of restriction endonuclease-digested viral DNA (data not shown). RV17044 is an insertional mutant, retaining all genes and transcription units of its AD169 parent. RV699 is an HCMV recombinant mutant virus whose genome contains the prokaryotic β-glucuronidase gene, whose expression is controlled by a viral promoter (26).

Inhibitor-resistant HCMV.

HCMV resistant to CFI02 was derived by passage of parental wild-type strain AD169 in the presence of CFI01, a less-active structural analog of CFI02, at a final concentration of 6 μM. This concentration of CFI01 causes a fourfold reduction in wild-type virus yield. After two serial passages, resistant progeny virus was enriched and plaque purified three times in the presence of the same concentration of CFI01. Cross-resistance to CFI02 was assessed by yield reduction assay. Several independent resistant viruses were obtained in this fashion.

Plasmids.

Plasmids containing viral DNA fragments were cloned into our standard cloning vector, pT7-1 (52). The HindIII-F (bases 64521 to 84684) and XbaI-F (bases 130010 to 146605) DNA fragments from resistant virus RV10325-11 were purified from agarose gels and cloned into the corresponding site of pT7-1 to yield pHindIII-F₁₁, pEcoRI-B₁₁, and pXbaI-F₁₁. The plasmid designated pHB_WT contains the HindIII/BamHI fragment (bases 79913 to 84864) from wild-type HCMV strain AD169 cloned between these sites of pT7-1. Several plasmids are derived from pHB_WT and contain regions from RV10325-11 replacing the corresponding wild-type sequences. pHB₁₁-N contains an NdeI/HpaI DNA fragment (bases 82257 to 83456) from RV10325-11. pHB₁₁-C contains an NsiI/NdeI DNA fragment (bases 81000 to 82257) from RV10325-11. pHB₁₁-N/C contains an NsiI/HpaI DNA fragment (bases 81000 to 83456) from RV10325-11. pHB₂ contains the HindIII/BamHI fragment (bases 79913 to 84864) from resistant strain 10325-2 cloned between these sites of pT7-1. pSK₁₁ contains the SpeI/KpnI DNA fragment (bases 80099 to 84017) from RV10325-11 cloned into the SmaI site of pT7-1.

Recombinogenic plasmid pHXSB-EGFP was used to construct RV17044. It contains a GFP expression cassette (SV40 promoter-driven GFP with SV40 bidirectional polyadenylation signal) flanked by HCMV sequences that directs its recombination within the US10-US9 intergenic region at base 199021 of the HCMV genome (12). Sequentially, the relevant portions of the plasmid are the flanking US11-US10 region (bases 200171 to 199021), the GFP expression cassette, and the flanking US9-US8 region (bases 199022 to 197042).

Automated infected-cell assays. (i) HCMV.

Ninety-six-well plate cultures of HFFs were infected with RV699 (multiplicity of infection of 0.004) in the presence of inhibitor. At 96 h postinfection (hpi), lysis buffer was added (using 50 mM sodium phosphate [pH 7.0] containing 0.1% Triton X-100 and 0.1% Sarkosyl), and the lysates were assayed for β-glucuronidase activity by using the methylumbelliferylglucuronide substrate, which when cleaved yields a fluorescent product. Antiviral activity is indicated by the reduced expression of the HCMV genome resident β-glucuronidase gene, compared to the absence of inhibitor.

(ii) HSV.

Vero cells were infected with HSV-1 strain Patton (multiplicity of infection of 0.006) in the presence or absence of inhibitor. At 24 hpi, the cells are fixed with 50% methanol-50% acetone prior to enzyme-linked immunosorbent assay. The primary antibody is murine anti-HSV gD monoclonal antibody; the secondary antibody is goat anti-mouse immunoglobulin G linked to β-galactosidase. Thus, the extent of viral inhibition is determined by assessing β-galactosidase activity after the addition of the methylumbelliferyl-β-d-galactoside substrate, which is cleaved to yield a fluorescent product.

Virus yield assay.

Plates (10 cm²) of HFFs were infected at the indicated multiplicities of infection in Dulbecco modified Eagle medium (DMEM) containing 2% fetal calf serum and 0.3% dimethyl sulfoxide in the presence or absence of inhibitor. After 2 h of adsorption at 37°C, the inoculum was removed and replaced with fresh medium, with or without inhibitor. After the specified period of time, usually 4 days, the plates were frozen at −70°C. Total virus (extracellular and intracellular) was quantitated by the determination of titers as described previously (26).

Marker transfer.

Transfections were done as described previously (26). Briefly, 1 μg of each of viral genomic DNA and linearized plasmid DNA (i.e., containing the UL55 mutation or control wild-type sequences) were transfected into HFF cells by a coprecipitation technique. After 6 h, the cells were shocked with 20% dimethyl sulfoxide in 1× HeBS buffer (25). Growth medium containing 2 μM CFI02 was used for the selection of resistant virus. The medium was changed every 3 to 4 days for 2 weeks, when the plaques were counted.

PEG reversal experiments.

HFF cells were infected with RV17044 (GFP⁺; multiplicity of infection of 1.0) in DMEM containing 2% fetal calf serum and 2 μM CFI02. After a 2-h adsorption period at 37°C, the medium was removed, and the cells were treated briefly with 50% (wt/wt) polyethylene glycol 8000 (PEG) in DMEM and then washed with 25 and 12.5% PEG in DMEM (49). This step was followed by three washes with PEG-free growth medium. Growth medium was added, and cells were transferred to a 37°C tissue culture incubator. The extent of PEG reversal of inhibitor action was assessed by virus-encoded GFP fluorescence at 48 hpi.

Lipid mixing experiments with R18.

HCMV virions were labeled with R18 (octadecyl rhodamine B chloride [Molecular Probes, Inc.]) as described previously (32). Briefly, 10 μg of R18 was added to 5 × 10⁶ PFU of HCMV virions (previously enriched on a 20 to 60% sucrose gradient, essentially as described previously [6]) in 500 μl of minimal essential medium (MEM) and incubated for 45 min at room temperature. Virions were separated from unincorporated R18 by passage through a Sephadex G-25 spin column and used in mixing experiments the same day. For mixing experiments, 12-well plates of HFF cells were placed on ice, the medium removed, and wells were washed with cold MEM lacking phenol red twice before the addition of 0.5 ml of MEM to each well. Labeled virus (25 to 50 μl; multiplicity of infection [in terms of PFU] of 1 to 3; however, the amount of fusogenic viral particles was much higher) was added, and the plate was incubated for 1 h at 4°C to allow viral adsorption. The inoculum containing unbound virus was removed, and the monolayer washed twice with medium. Warm MEM (0.5 ml) containing inhibitor (at 1 and 10 μg/ml final concentrations) was added to each well; the plates were transferred to 37°C for 1 h to allow viral fusion. The medium was removed, the monolayer was washed with inhibitor-free MEM, and fusion was assessed by fluorescence microscopy.

Protein and DNA analysis.

Radiolabeling, immunoprecipitation, and immunoblotting were done as described previously (27, 29). Antibodies used were as follows: rabbit polyclonal antisera reactive with HCMV gpUS11 (26), rabbit polyclonal antisera reactive with HCMV gpUS2 (20), rabbit polyclonal antisera reactive with HCMV gpUS3 (30), murine monoclonal antibody 13-170 reactive with HCMV IE1/IE2 (Advanced Biotechnologies, Inc.); murine monoclonal antibody TP25.99 reactive with human major histocompatibility complex class I heavy chains (S. Ferrone, New York Medical College), murine monoclonal antibody 13-129 reactive with HCMV pp65 (Advanced Biotechnologies), and murine monoclonal antibody Ber-T9 reactive with human transferrin receptor (Dako). HCMV DNA was analyzed by DNA blot hybridization after isolation by using a modified Hirt procedure (26, 48).

Cytoxicity assay.

An MTS cytotoxicity assay was performed to determine the 50% cytotoxic concentration of various compounds in confluent HFF cell monolayers, as described previously (58).

Immunofluorescence.

Indirect immunofluorescence was done according to a standard protocol (22). Briefly, cells were fixed with 4% paraformaldehyde and then permeabilized with 0.2% Triton X-100, both times in phosphate-buffered saline. After a blocking step with 3% bovine serum albumin in phosphate-buffered saline, primary antibody (rabbit polyclonal anti-pp65) and then donkey anti-rabbit immunoglobulin G-fluorescein isothiocyanate (Pierce) were used. The stained cells were washed and exposed briefly to DAPI (4′,6′-diamidino-2-phenylindole) nuclear fluorophore (Pierce) at 100 ng/ml.

RESULTS

In vitro antiviral and cytotoxicity evaluation.

The anti-HCMV activity of CFI02 was assessed by using automated and manual assays (Table 1). Its IC₅₀ against laboratory adapted HCMV strain AD169 was determined to be 0.04 μM (automated assay), while its IC₉₀ was determined to be 0.12 μM (yield reduction assay), both at a multiplicity of infection of 0.3 (Fig. 2A). CFI02 is a very potent HCMV inhibitor since 1 μM CFI02 reduces yield by >3 logs. When the multiplicity of infection was increased ca. 3- and 10-fold (i.e., multiplicity of infection of 1.0 or 3.0), the IC₉₀ increased only 1.5- and 2.5-fold, respectively. In contrast, the IC₉₀ of the initial analog, CFI01, is a modest 12 μM at multiplicity of infection equal to 0.3 (Fig. 2B), 100-fold less active than CFI02.

TABLE 1.

Antiviral activity and specificity of CFI02

Virusa	Cell type	Concn (μM) as determined by:
		Automated assay (IC50)	Yield reduction assay (IC90)
HCMV (AD169)	HFF	0.04	0.12
HSV-1 (Patton)	HFF	>20	ND
VZV (Ellen)	HFF	>20	ND
RSV (A2)	HFF	>20	ND
MCMV (Smith)	MEFc	NDb	>20
Simian CMV (Colburn)	HFF	ND	>20
RCMV	HFF	ND	>20

^aVZV, varicella-zoster virus; RSV, respiratory syncytial virus.

^bND, not done.

^cMEF, mouse embryo fibroblast.

FIG. 2.

(A) Sensitivity of AD169 to CFI02 at various multiplicities of infection. (B) Comparison of sensitivity of strain AD169 to CFI01 and CFI02. Virus yield assays were used in both.

The specificity of the antiviral effect was evaluated by using other herpesviruses or unrelated viruses in the inhibition assays (Table 1). In each of these assays, the antiviral activity was >20 μM. Even closely related betaherpesviruses (including MCMV, simian CMV, and RCMV) were refractory to inhibition by CFI02. In a cytotoxicity assay, the 50% cytotoxic concentration of CFI02 was >50 μM. Thus, the therapeutic index of this compound is >2,000. We conclude that CFI02 is a potent and specific inhibitor of HCMV.

RV17044.

The HCMV used in many of the mechanism-of-action experiments was a recombinant designated RV17044, which expresses GFP under the control of the SV40 promoter from the US9-US10 intergenic region. This virus is not deleted of any viral genes. It has growth kinetics nearly identical to the HCMV strain AD169 parent (Fig. 3) and is equally sensitive to the inhibitory activity of CFI02: IC₉₀ = 0.14 μM in a yield reduction assay at a multiplicity of infection of 0.3 (data not shown). Thus, the use of RV17044 allowed the visualization of virus-encoded GFP in infected cells throughout infection, as well as the quantitation of virus yield after 4 days.

FIG. 3.

RV17044. (A) Linear representation of the HCMV genome. The location of ORFs US9 and US10 are shown, as is the arrangement of the GFP expression cassette insert within the US9-US10 intergenic region in recombinant strain RV17044. (B) Single cycle growth analysis of AD169 and RV17044 (multiplicity of infection of 1.5).

Time-of-addition experiments.

To assess the portion of the HCMV replication cycle that was inhibited, experiments were performed in which 2 μM CFI02 was added at different times in the replicative cycle: viral fluorescence (48 hpi) and yield (96 hpi) were assessed (Table 2). When inhibitor was present from the time of initial infection (i.e., 0 hpi) through 96 hpi, virus yield was reduced by >4 logs; however, addition of compound beginning 2 h later (i.e., 2 hpi) had no effect on virus yield. Observation of GFP fluorescence was consistent with the viral yield data. These data indicate that the inhibitory effect of CFI02 is exerted within the first 2 h of the replicative cycle. Several important viral genes are expressed during the first few hours of HCMV infection. The effect of CFI02 on expression of these genes, as well as some cellular genes, was assessed by using radiolabeling and immunoprecipitation experiments (Fig. 4). Viral early (gpUS11 and gpUS2 [data not shown]) and immediate-early (IE; IE1, IE2, and gpUS3 [data not shown]) proteins were not expressed under conditions that were determined to be inhibitory in virus yield experiments (e.g., when compound was present at 0 hpi but not when added at 2 hpi). In contrast, cellular transferrin receptor or major histocompatibility complex class I heavy chains (data not shown) were synthesized in equal abundance under all conditions. Consistent with the viral yield experiments and the expression of viral IE and early proteins, HCMV DNA synthesis is inhibited when CFI02 was present from 0 to 2 hpi but proceeded normally when inhibitor was present from 2 to 72 hpi (Fig. 4D).

TABLE 2.

Effect of CFI02 on viral yield

Exposure to 2 μM CFI02 (hpi)a	Viral yield (PFU)a
None	2 × 105
0-96	<1 × 101
2-96	1 × 105
0-2	2 × 101

^aCells were exposed to RV17044 at a multiplicity of infection of 0.1. and were treated with 2 μM CFI02 as indicated.

FIG. 4.

Effect of CFI02 on cellular or viral gene expression and viral DNA synthesis. Uninfected or HCMV-infected HFFs (multiplicity of infection of 3) were exposed to CFI02 (2 μM) as indicated, and proteins radiolabeled from 3 to 6 hpi or DNA was isolated at 72 hpi. Lysates were immunoprecipitated with antibody to IE1/IE2 (A), gpUS11 (B), or human transferrin receptor (C). (D) Infected cell DNAs were cleaved with HindIII and subjected to blot hybridization with probe HX (26) specific for HCMV joint (j) region DNA fragments C and G (12). Two concentrations of DNA were added for each sample, as indicated, for quantitation purposes.

No inactivation of virions.

Since CFI02 appears to act at a very early stage in the HCMV replication cycle, the possibility of this inhibitor directly inactivating virions was examined. In this experiment, RV17044 was incubated in the presence of CFI02 for up to 2 h at either 4 or 37°C and then serially diluted, and the titer was determined (Table 3). Preincubation of CFI02 with virus prior to infection did not result in direct inactivation of virus, as evidenced by no reduction in titer when the inhibitor was subsequently diluted below the effective concentration. Similar data were obtained when the preincubation mix was run over a size-exclusion column (to remove unbound inhibitor) prior to analysis of infectivity (data not shown). These data also suggest that CFI02 does not stably interact with free HCMV virions.

TABLE 3.

Effect of preincubation with CFI02

Incubation condition		Compounda	Virus titer (PFU)
Temp (°C)	Duration (h)
37	0	None	3.79 × 105
37	0	CFI02	3.87 × 105
37	2	None	3.88 × 105
37	2	CFI02	3.78 × 105
4	2	None	3.70 × 105
4	2	CFI02	3.63 × 105

^aCFI02 was used at 2 μM.

Examination of initial events in HCMV replicative cycle.

Since the data have indicated that inhibition by CFI02 is prior to viral IE gene expression, experiments were performed examine initial events in the HCMV infectious cycle, including adsorption, fusion, and cytoplasm-nuclear transport. Detegumentation and nuclear transport of the capsid and (at least) some tegument proteins occurs rapidly after fusion of the viral particle with the cellular plasma membrane (39). An indication that these processes have occurred is the presence of the abundant tegument protein pp65 in the nucleus of infected cells within a few hours after infection. HCMV was exposed to HFF cells for 6 h prior to immunofluoresence analysis for pp65 (Fig. 5). In wells that did not receive CFI02 or that received CFI02 from 2 to 6 hpi, substantial pp65 was detected in the cell nucleus. In contrast, in wells containing CFI02 from 0 to 2 or from 0 to 6 hpi, pp65 was not detected in the nucleus. Thus, detegumentation and cytoplasm-nuclear transport does not occur in the presence of CFI02.

FIG. 5.

Cytoplasm-to-nucleus transport of pp65. HCMV-infected HFFs (multiplicity of infection of 3) were exposed to CFI02 (2 μM) as indicated. At 6 hpi, the cells were fixed and permeabilized for immunofluorescence with anti-pp65 monoclonal antibody and DAPI staining.

Effects on viral adsorption and fusion were assessed in viral yield and immunoblot experiments utilizing a 2-h incubation at 4°C (a temperature that allows binding but not fusion), followed by a temperature shift to 37°C (which allows fusion) (Fig. 6A) (41). When CFI02 was present only during the 2-h adsorption phase and then removed just prior to temperature shift, the 4-day viral yield was reduced by 2 logs (similar to the results obtained if CFI02 is present throughout). When CFI02 was present only after the adsorption phase, the viral yield was reduced ∼1.5 logs. Although the former result may suggest adsorption inhibition, the latter result indicates the inhibition of fusion, since inhibitor was not present during the time of viral adsorption. To further examine a possible effect on adsorption, immunoblot experiments were done. In extracts of cells exposed to virus for 2 h at 4 or 37°C (Fig. 6B), either in the absence or presence of CFI02, the pp65 virion tegument was detected in similar abundance. Again, there was no indication of an effect on adsorption. Thus, inhibition of virus yield in cells exposed to inhibitor only during the adsorption phase is most likely due to the stable association of inhibitor to virus that is bound to the cell surface. PEG chemically mediates the fusion of biological membranes (49). In experiments when PEG was added 2 h after the temperature shift in the presence of CFI02, inhibition of RV17044 was partially reversed, as indicated by the degree of GFP fluorescence at 48 h (Fig. 6C). Taken together, these data strongly support the notion that CFI02 inhibits HCMV fusion with, but not binding to, the plasma membrane.

FIG. 6.

Analysis of the effect of CFI02 on adsorption and fusion. (A) Four-day virus yield of RV17044 (multiplicity of infection of 0.1) after exposure to CFI02 (2 μM) various times postinfection, as indicated. (B) Adsorption assay. HFF cells were left uninfected or were RV17044 infected (multiplicity of infection of 1, 0.3, or 0.1) at either 4 or 37°C and then either not treated or treated with CFI02 (2 μM) for 2 h. After and extensive washing, protein extracts were made for immunoblot analysis with anti-pp65 monoclonal antibody. The location of a background cellular protein (b) that cross-reacts in this assay is indicated. (C) PEG reversal of inhibition. RV17044-infected (multiplicity of infection of 1) HFF cells were untreated (i), treated with 2 μM CFI02 and then PEG-free medium (ii), or treated with CFI02 and PEG in medium (iii) as described in Materials and Methods. Cells were photographed for GFP fluorescence at 48 hpi.

CFI02 does not stably associate with the cell membrane.

The inhibition of HCMV fusion could occur if CFI02 stably interacts with cell membrane protein(s) required for this process. To investigate this possibility, HFF cells were pretreated with CFI02 for 2 h prior to infection by HCMV (i.e., no inhibitor in the medium during infection). Pretreatment of cells resulted in minimal subsequent inhibition of HCMV infection. Pretreatment concentrations of 0.2 μM CFI02 caused no inhibition of HCMV, whereas 0.6 μM CFI02 caused only a twofold inhibition. In contrast, when present from 0 to 2 hpi, 0.2 and 0.6 μM CFI02 caused 10 and >20-fold inhibition. Thus, inhibition of HCMV by CFI02 is not the result of stable association of the compound with cell surface proteins (Fig. 7).

FIG. 7.

Pretreatment of HFF cells with CFI02. HFF cells were exposed to CFI02 for 2 h prior to or just after (i.e., 0 to 2 hpi) exposure to RV17044 (multiplicity of infection of 0.1). For treatment prior to infection, after the 2-h incubation at 37°C, the cells were washed twice and then incubated in inhibitor-free mediuma for 1 h before infection in fresh medium. Virus yield at 4 days postinfection was quantitated.

HCMV resistant to CFI02.

To examine the nature of CFI02 inhibition of HCMV, resistant mutants were generated in the laboratory. Wild-type strain AD169 was passaged and plaque purified in the presence of CFI01 (Fig. 1A), a less-active structural analog from the same chemical series, as described in Materials and Methods. Several independently derived mutants were obtained in this fashion. To identify mutations sufficient to confer resistance, marker transfer experiments of two such mutants (RV10325-2 and RV10325-11) are presented. Since the abundant envelope gB is known to be an essential mediator in the HCMV virion fusion process (4, 41, 53), DNA clones from this region of the mutant viral genome were constructed and tested in these experiments (Fig. 8). The 20-kb HindIII-F region of the RV10325-11 genome, which contains open reading frames UL48-UL56, including gB (UL55), was cloned and subcloned. Marker transfer experiments yielded resistant virus when the mutant clone (pHindIII-F₁₁) but not the analogous wild-type clone (pHindIII-F_WT) was used in conjunction with wild-type AD169 virion DNA (data not shown). As an additional control, another plasmid clone, pXbaI-F₁₁, from the same resistant strain and contains the open reading frames UL87-UL100, the latter of which encodes another abundant virion envelope glycoprotein involved in virion attachment, but not fusion, did not confer resistance in the marker transfer experiments (data not shown). In further marker transfer experiments involving multiple other pHindIII-F₁₁ subclones, those that contained UL55 but not those that did not contain UL55 conferred resistance to CFI02 (data not shown). As a representative example, a pHindIII-F₁₁ subclone designated pSK₁₁, which comprises the UL55 gene with only small amounts of sequence from the neighboring UL54 and UL56 genes, conferred resistance since after transfection with AD169 parental genomic DNA, 99 and 42 PFU were obtained in the absence and presence of CFI02, respectively (Fig. 8B). Sequence analysis of pSK₁₁ revealed two mutations within the UL55 coding region. One mutation was a T→C transition at nucleotide 83281, resulting in an isoleucine-to-threonine mutation at amino acid 171, near the N-terminal portion of UL55 (Fig. 8A). The other was a G→A transition at nucleotide 81296, resulting in a glycine-to-serine mutation at amino acid 733, within a C-terminal UL55 region known as hydrophobic region 1 (Fig. 8A). Each of these mutations was assessed individually for its ability to confer resistance (Fig. 8B). Small regions of the mutant UL55 gene were subcloned in place of the corresponding fragment within an otherwise wild-type plasmid subclone, such that each resulting plasmid had either the amino acid 171 mutation (pHB₁₁-N), the amino acid 733 mutation (pHB₁₁-C), or both (pHB₁₁-N/C). Marker transfer experiments with these plasmids indicated that the C-terminal G733S mutation (13 PFU in the presence of CFI02), but not the I171T mutation (0 PFU in the presence of CFI02), was sufficient to confer resistance (Fig. 8B). The UL55 gB gene-containing region from a second resistant strain (RV10325-2) was also sequenced (the sequencing template was a PCR product generated directly from viral DNA) and cloned. It contains a C→A transversion at nucleotide 81240, resulting in a phenylalanine to leucine mutation at amino acid 751, the first amino acid of hydrophobic domain 2 (51). Marker transfer experiments with pHB₂ verified that this gB mutation does cause resistance, since 36 PFU were obtained in the presence of CFI02, in contrast to 0 PFU obtained when the analogous wild-type clone (pHB_WT) was used (Fig. 8C). Thus, the marker transfer data from resistant strains RV10325-2 and -11 are consistent and demonstrate that individual mutations within the C-terminal gp55 component of gB confer resistance, suggesting that inhibition by CFI02 involves interference with its fusion function(s). RV10325-11 is totally insensitive to CFI02, since in virus yield assays, at concentrations up to 6 μM CFI02 (60 times the parental IC₉₀), there was no effect on RV10325-11 yield (data not shown). Although we have no evidence for a role of genes other than gB in resistance to CFI02, since we did not examine for mutations in other virion glycoprotein genes known to also be involved in fusion (e.g., gH and gL), we cannot assign, nor discount, a possible role of these glycoproteins in CFI02-mediated inhibition.

FIG. 8.

Marker transfer experiments. (A) Schematic representation of the HCMV genome. The approximate locations of the HindIII-F and XbaI-F DNA fragments are indicated. Also shown is an expansion of the UL55 gene region. Filled areas correspond to hydrophobic regions, as follows: signal sequence (ss), hd1, and putative transmembrane hybrophobic region 2 (hd2). Selected restriction sites are indicated, as is the UL55-containing SpeI/KpnI DNA fragment that was cloned to make pSK₁₁. Mutated amino acids are shown, as is the cellular furin cleavage site at UL55 amino acid 459. (B and C) Representative marker transfer experiments. Transfection and/or selection was done in the presence of 2 μM CFI02. wt, wild type.

Inhibition of HCMV fusion in lipid mixing assays.

Membrane fusion between the HCMV lipid envelope and the cell plasma membrane was measured directly by using virus labeled with fluorescence lipid probe R18 (32). Virus was added to HFF cells at 4°C to facilitate binding to the cell surface. After removal of the unbound virus, inhibitor was added, and was fusion induced by shifting the temperature to 37°C. When HCMV fusion occurs, R18 from the viral lipid envelope spreads to cell plasma membrane, resulting in cellular fluorescence. Wild-type and inhibitor-resistant strains of HCMV were used in the presence or absence of CFI02 in the mixing assay (Fig. 9). Lipid mixing was observed with wild-type virus in the absence of inhibitor (Fig. 9A) but was substantially reduced in the presence of 2 and 20 μM CFI02 (Fig. 9B and C). In contrast, CFI02 had no effect on lipid mixing when resistant strain RV10325-11 was used (Fig. 9D to F).

FIG. 9.

Lipid mixing experiments. R18-labeled wild-type strain AD169 and resistant strain RV10325-11 were used (A to C and D to F, respectively). Experiments were performed either in the absence of CFI02 (A and D) or in the presence of 2 μM CFI02 (B and E) or 20 μM CFI02 (C and F).

Effects of CFI02 at late times postinfection.

In addition to being a major component in virion fusion early in the HCMV replication cycle, gB may play a role at late times postinfection in viral release and cell-to-cell spread of virus (41). The possible effect on viral release was assessed by exposing HCMV strain AD169-infected cells (multiplicity of infection of 1.0) to 1.8 μM CFI02 (i.e., ∼15 times the IC₉₀) from 1 to 4 days postinfection. At 4 days postinfection, intracellular and extracellular virus was quantitated by determining the titers (data not shown). As expected, there was no substantial difference in total virus yield between the treated and untreated infected cells. Extracellular and intracellular virus was approximately equal in CFI02-treated cells; in contrast, extracellular virus was five times more abundant than intracellular virus in untreated cells. Thus, a relatively high concentration of CFI02 has a measureable, but minimal, effect on virus release. To determine whether CFI02 prevents cell-to-cell spread of virus, HFF cells were infected at a very low multiplicity with GFP⁺ HCMV (RV17044) in the absence of inhibitor. Inhibitor-containing liquid medium was added after a 2-h adsorption period. After 5 days, viral spread was assessed microscopically, as evidenced by GFP fluorescence (Fig. 10). Two types of controls were done. In plates containing inhibitor-free liquid medium, the majority of plaques contained 10 to 20 GFP⁺ cells (Fig. 10A). This represented both cell-to-cell spread of virus and spread by virus released to into the medium from those cells infected in the first round. In the second control, cells were overlaid with inhibitor-free agarose medium (Fig. 10B). Plaques in this plate usually had 6 to 10 infected cells, representing spread of virus in direct cell-to-cell fashion only. In plates with inhibitor-containing liquid medium, the plaque size was clearly dependent on the concentration of CFI02. In plates containing 0.06 μM CFI02 (i.e., about one-half of the IC₉₀), most plaques were comprised of four to six infected cells; occasionally, singly infected cells were observed (Fig. 10C). At 0.20 μM (about two times the IC₉₀), there were mostly singly infected cells (Fig. 10D). At higher concentrations, only singly infected cells were observed (Fig. 10E and F). At CFI02 concentrations of 0.20 μM and higher singly infected cells predominated (Fig. 10D to F), rather than 6 to 10 infected cell plaques, indicative of cell-cell spread (Fig. 10B); this indicates that CFI02 does indeed prevent this type of viral spread. Thus, CFI02 efficiently inhibits both the reseeding of virus released into the medium and direct cell-to-cell spread of the virus.

FIG. 10.

Inhibition of cell-to-cell spread of HCMV. HFF cells were infected with RV17044 (multiplicity of infection of 0.003) in the absence of inhibitor. CFI02 was added at 2 hpi, as follows: 0.06 μM (C), 0.2 μM (D), 0.6 μM (E), and 2 μM (F). All wells contained liquid medium, except for well B, which contained 0.5% agarose-containing medium. Viral spread, as evidenced by GFP fluorescence, was assessed at 5 days postinfection.

DISCUSSION

A specific and potent small molecule inhibitor of HCMV was identified. CFI02 belongs to a novel series of small molecule HCMV inhibitors for which structure-activity relationship has recently been determined (3). Futhermore, the viral target of the inhibitor is also novel for HCMV: an inhibitor of gB-mediated virion fusion. Evidence for this is as follows: (i) the inhibitor acts within the first 2 h of the HCMV replication cycle; (ii) it remains an effective inhibitor when added after adsorption; (iii) inhibition can be reversed with PEG, a chemical mediator of membrane fusion; (iv) cytoplasm-to-nucleus transport of virions and viral IE-early gene expression were not detected; and (v) mutations sufficient to confer complete resistance are localized in gB.

HCMV gB is a 906-amino-acid major virion envelope glycoprotein (6, 51). Shortly after translation, it forms disulfide-linked homodimers in the endoplasmic reticulum and then is cleaved by cellular furin after amino acid 459 in the trans-Golgi network to yield disulfide-linked amino-terminal gp116 and C-terminal gp55 proteins (2, 7, 9, 10). gB is conserved throughout the herpesvirus family and is known to play an essential and central role in early events (adsorption and fusion) of the HCMV infectious cycle (8, 44). Regarding adsorption, initial interaction of HCMV is with cellular heparin sulfate proteoglycans, mediated by gB and gM (14, 31). This is followed by binding to (unidentified) high-affinity cellular receptors (5, 11). Our data indicate that CFI02 does not interfere with adsorption, since there is no effect on the quantity of virion structural proteins associated with cells in the presence of CFI02 (Fig. 6B). Furthermore, inhibitor added after virus exposure to cells for 2 h at 4°C (conditions under which adsorption does occur) causes maximal inhibition (Fig. 6A).

In addition to adsorption, gB is known to have a role in virion fusion with the plasma membrane (41). HCMV virions fuse with the plasma membrane via a pH-independent mechanism (13). As with adsorption, and by analogy with other herpesviruses, other HCMV virion glycoproteins may contribute to efficient fusion, especially the gH/gL complex (20, 21, 45, 55; E. Kinzler and T. Compton, unpublished data). The exact mechanism by which gB affects or participates in fusion remains to be elucidated. However, when gB was expressed in transfected cells, cellular fusion was observed, and gB antibodies inhibited this fusion (4, 41, 53). Furthermore, C-terminal hydrophobic domain 1 (hd1) and cytoplasmic domains were shown to be required (4, 54). Since mutations that confer resistance to CFI02 have been mapped within the carboxy-terminal half of gB, we propose an interaction between the inhibitor and this glycoprotein. An alternative possibility is that CFI02 stably interacts with cellular protein(s) required for, and thereby blocking, virus fusion. gB is known to interact with at least two classes of cellular molecules (5). CFI02 is unlikely to act by the latter, since preexposure of inhibitor to cells (i.e., prior to infection) results in minimal viral inhibition (Fig. 7). Furthermore, it is improbable that single amino acid mutations in gB would overcome inhibition if CFI02 targeted a cellular protein.

The nature of the interaction between CFI02 and gB is speculative. Its clear that CFI02 does not stably interact with gB on free virions. Preincubation of virions in the presence CFI02, but in the absence of cells, does not inhibit subsequent infection when cells are introduced and the remaining free inhibitor is either diluted below its effective concentration (Table 3) or removed by filtration. Although the mechanism by which gB effects fusion is not known, by analogy with other viral proteins involved in binding and fusion, a conformational change in gB may occur upon virion adsorption. Most notably, conformation changes in influenza virus hemagglutinin and human immunodeficiency virus (HIV) gp120/gp41 occur upon virion binding to cells to facilitate fusion (17, 50). This transient conformationally altered species, necessary to facilitate fusion, may be targeted by inhibitors (18, 19, 24). We hypothesize that an analogous conformation change in gB may expose regions that allow stable interaction with CFI02. gB is a disulfide-linked homodimer and has 11 cysteine residues. In a soluble form of gB, which retains many gB functional properties, 10 of these residues were shown to participate in inter- or intramolecular disulfide bonding (5, 37). One possibility is that a conformation change in gB resulting from binding may involve disulfide-bond shifting to place gB in a conformation required for fusion. Consistent with this hypothesis is the observation that the ability to form free thiol groups is necessary for HCMV virion entry but not for binding (40). CFI02 may stably interact with gB during this transitional period, thereby preventing the fusion conformation. However, disulfide bonding between the thiourea inhibitor and a gB cysteine does not appear to play a major role in its inhibitory activity since the corresponding urea analog has similar inhibitory activity (data not shown). Whatever the nature of the interaction, it does appear to be stable, since removal of inhibitor from the media does not reverse viral inhibition (Table 2), even though inhibited virions can be detected, by immunologic means, to remain bound to cells for greater than 24 h (data not shown).

One very striking feature of CFI02 is the specificity it has for HCMV, while having essentially no inhibitor activity against other viruses, including closely related CMVs that encode a highly homologous gB (Table 1). For example, the gBs of MCMV and RCMV have 45 and 60% overall homology, respectively, to HCMV gB (33, 34, 46). Locally, there are regions of these gBs having much greater identity with the corresponding region of HCMV gB. Mutations that confer resistance to CFI02 are located with in the C-terminal half of UL55, which encodes gB gp55. gB-based fusion assays with either antibodies mapping to gp55, or various gp55 mutants, have indicated that this region plays an essential role in penetration (4, 41, 53, 54). Resistance-conferring mutations G733S and F751L are within regions of high homology with gB of MCMV and RCMV; in both of these betaherpesvirus examples, the wild-type HCMV amino acid is conserved (33, 34, 46). The fact that CFI02 does not inhibit MCMV or RCMV indicates that the gB-CFI02 interaction requires more than just a single contact or interaction domain. Interestingly, G733 is within glycine-rich hd1, a domain proposed to be involved in fusion due to its prediction to form an amphipathic alpha helix (4, 47, 51). Glycine-rich hydrophobic domains are characteristic of some viral fusogenic proteins. It is not known whether the G733S mutation in the resistant mutant causes delayed penetration kinetics. However, G734 and not G733 appears to be the glycine conserved with alphaherpesvirus, influenza virus, Sendai virus, and HIV fusion proteins (47). F751 is the N-terminal amino acid of the hd2 believed to be the transmembrane domain of gB (47, 59). Future studies of additional independent resistant mutants are needed in order to fully understand the nature of the interaction between CFI02 and gB, or perhaps other HCMV glycoproteins.

Although CFI02 is not being developed clinically due to in vivo stability and metabolic issues, it is an example of a specific HCMV inhibitor against a novel target. The viral process inhibited—fusion—is one of the earliest steps in the infective cycle. Clinically, drugs similar to this in function may be advantageous to current anti-HCMV drugs (which target viral DNA replication), since no viral gene products are expressed, thereby avoiding possible cytopathological or immunopathological effect(s) that may be triggered by viral structural, IE, or early proteins. Furthermore, since the target is fusion, cross-resistance with currently approved therapies that target DNA replication (i.e., ganciclovir-, cidofovir-, or foscarnet-resistant isolates) in the clinic would not be an issue (16). In fact, in our studies clinical HCMV isolates resistant to these drugs were sensitive to CFI02 (data not shown). Although, highly active antiretroviral therapy (HAART) has substantially reduced the incidence of CMV disease in HIV-positive individuals, it is still present in HAART failures and transplant patients (15, 35, 43). In fact, recent reports have suggested that the rate of morbidity and mortality in solid organ transplant patients due to ganciclovir-resistant isolates is rising (1, 23, 36, 38). Thus, the need for inhibitors that target other viral functions, such as fusion, are desirable. Our data indicate that specific, highly active small molecular inhibitors against HCMV fusion, such as CFI02, can be obtained, thereby validating this viral target.