Selective Inhibition of Hepatitis C Virus Infection by Hydroxyzine and Benztropine

Abstract

Hepatitis C virus (HCV) infection is a major biomedical problem worldwide as it causes severe liver disease in millions of humans around the world. Despite the recent approval of specific drugs targeting HCV replication to be used in combination with alpha interferon (IFN-α) and ribavirin, there is still an urgent need for pangenotypic, interferon-free therapies to fight this genetically diverse group of viruses. In this study, we used an unbiased screening cell culture assay to interrogate a chemical library of compounds approved for clinical use in humans. This system enables identifying nontoxic antiviral compounds targeting every aspect of the viral life cycle, be the target viral or cellular. The aim of this study was to identify drugs approved for other therapeutic applications in humans that could be effective components of combination therapies against HCV. As a result of this analysis, we identified 12 compounds with antiviral activity in cell culture, some of which had previously been identified as HCV inhibitors with antiviral activity in cell culture and had been shown to be effective in patients. We selected two novel HCV antivirals, hydroxyzine and benztropine, to characterize them by determining their specificity and genotype spectrum as well as by defining the step of the replication cycle targeted by these compounds. We found that both compounds effectively inhibited viral entry at a postbinding step of genotypes 1, 2, 3, and 4 without affecting entry of other viruses.

INTRODUCTION

It is estimated that 150 million humans are chronically infected by hepatitis C virus (HCV), many of whom will suffer from severe liver disease (1). Chronic HCV infection is associated with liver inflammation, fibrosis, cirrhosis, and hepatocellular carcinoma (2). Consequently, HCV infection is one of the leading causes of liver transplantation worldwide. The hepatitis C viruses are genetically diverse and are clustered in 7 major genotypes with a large number of subtypes (3). There are important differences among genotypes regarding geographical distribution, pathogenesis, and response to treatment (4). There is no vaccine against HCV, and the current approved treatments, albeit increasingly effective, are expensive and associated with severe adverse effects (5).

The HCV replication cycle is initiated by the attachment of the viral particles to and entry into the target cells, a multistep process mediated by the viral envelope glycoproteins and host factors that are incorporated into viral particles (6). Virion internalization is triggered by receptor-dependent endocytosis in clathrin-coated pits, and subsequent endosomal acidification triggers fusion of the viral and endosomal membranes, a process mediated by the HCV glycoprotein complex (E1E2) that results in the release of the viral genome to the cytosol (7). Translation of the incoming viral RNAs leads to expression of the viral proteins, viral RNA replication, and assembly of progeny infectious viral particles (reviewed in references 8 and 9).

Until 2011, HCV treatment consisted of the combination of alpha interferon (IFN-α) and ribavirin, which was effective only partially and was associated with severe adverse effects (5). The recent approval of viral protease inhibitors to be used as additives to the IFN-based therapy improved the cure rate significantly, at the expense of additional adverse effects and high cost, factors that limit their implementation (10). Nevertheless, a large number of effective and specific HCV inhibitors (direct acting antivirals; DAA) have been developed and display promising efficacy in clinical trials. However, the genetic diversity of HCV viruses and the stage of liver disease (i.e., cirrhosis) are revealing themselves as obstacles for effective, pangenotypic treatments (5, 11). Thus, the armamentarium against HCV needs to be expanded to combat these diverse viruses, particularly since combination therapy using several drugs is required to circumvent the selection of escape variants during monotherapy (5). In order to contribute to this task, we used an unbiased screening methodology to interrogate a chemical library for HCV inhibitors (12). This library contains 281 clinically approved drugs prescribed for non-HCV applications. Using this strategy, we identified a set of compounds that inhibited HCV infection at nontoxic concentrations. We selected the compounds with the highest therapeutic index (hydroxyzine and benztropine) and characterized the step(s) of the viral replication cycle inhibited by these compounds. All the compounds inhibit HCV entry in various human hepatoma cell lines at a postadsorption step. Interestingly, we observed that genotype 2 envelope glycoproteins are more susceptible to hydroxyzine than those of genotypes 1, 3, and 4, an observation that was not evident for other compounds. Our study identified hydroxyzine and benztropine as effective, selective inhibitors of HCV entry in cell culture. It also revealed important intergenotypic differences in the susceptibility of HCV to the antiviral action of hydroxyzine, indicating that this compound might also be useful for the study of basic aspects of HCV entry.

MATERIALS AND METHODS

Reagents.

An NIH Collection 2 compound library was purchased from Evotec (San Francisco, CA). Hydroxyzine pamoate, fluoxetine, and perphenazine were purchased from Sigma-Aldrich (St. Louis, MO), and benztropine mesylate was purchased from Sellekchem (Houston, TX). 2mAde (2′-c-methyladenosine) was purchased from BOC Sciences (Shirley, NY). All compounds, including those in the library, were dissolved in dimethyl sulfoxide (DMSO) at a final concentration of 10 mM. Except when stated otherwise, the compounds were used at the following concentrations: hydroxyzine and perphenazine at 5 μM, fluoxetine at 3 μM, and benztropine mesylate at 7.5 μM.

Cells.

Human hepatoma Huh-7 cells and derived subclone Huh-7.5.1 clone 2 (here clone 2 cells) as well as human embryonic kidney cells (HEK293T) have been previously described (13–15). These cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 100 mM HEPES, nonessential amino acids (Gibco), and penicillin-streptomycin (Gibco) and 10% fetal bovine serum (FBS). Hepatocarcinoma cell lines BCLC5, BCLC6, and BCLC10 have previously been described (16).

Viruses.

JFH-1 virus was produced from cloned cDNA using a previously described methodology (17). D183 virus (D183v) was previously described (18). Infectious, defective HCV particles produced by transcomplementation (HCVtcp) were generated using the methodology and reagents described in reference 19 and kindly provided by Ralf Bartenschlager.

Library screening.

Compounds were screened at a final 10 μM concentration using a cell-based enzyme-linked immunosorbent assay (ELISA)-colorimetric readout of the infection approach in a 96-well format as described in references 12 and 20. Briefly, this assay allows determining viral spread by measuring the total viral antigen (E2) present in a given well using specific antibodies. Relative infection values are determined using a standard curve generated by serial dilution of the virus inoculum as previously described (12). Once infection efficiency is determined, wells are extensively washed to determine remaining cell biomass in order to evaluate compound cytoxicity. Cell biomass is measured by crystal violet staining and colorimetry at 570 nm as described previously (21). Relative biomass values are expressed as percentages of the values found in the control wells. Since cells are plated at low (approximately 30%) density, the cells proliferate until reaching confluence in the control wells by the end of the experiment 72 h postinfection. Thus, biomass estimation allows identification of false positives in the screening.

Primary hits were identified as those reducing HCV infection by more than 20-fold but maintaining 85% of the cell biomass at 10 μM. Primary hits were subsequently counterscreened at the same concentration for confirmation. Compounds significantly reducing HCV infection but displaying detectable toxicity (biomass < 85%) were counterscreened at 2 μM. Compounds displaying a more than 80% reduction in HCV infection but no significant toxicity (biomass > 85%) were further considered in the study.

Determination of potency (EC₅₀) and toxicity (CC₅₀).

In order to determine potency and cytotoxicity indexes, 10⁴ clone 2 cells were seeded in 96-well plates (approximately 30% confluence). Potency of the compounds (50% effective concentration [EC₅₀], EC₉₀) was calculated by determining relative infection efficiency levels in the presence of serial compound dilutions (ranging from 50 μM to 2.5 nM) at 72 h postinfection, using the colorimetric method described above and as previously described (12, 20). Parallel uninfected cultures, treated with the same compound dilutions, were used to determine 50% cytotoxic concentration (CC₅₀) values by MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide]-formazan cellular viability assays 72 h posttreatment using standard procedures (22). EC₅₀, EC₉₀, and CC₅₀ values were interpolated graphically from the dose-response curves.

Infection experiments. (i) Multiple-cycle infections.

Huh-7.5.1. clone 2 cells (5 × 10⁵ cells; 6-well plate) were inoculated (multiplicity of infection [MOI] of 0.01) with JFH-1 virus (17) in the presence of perphenazine (5 μM), hydroxyzine (5 μM), fluoxetine (3 μM), or benztropine (7.5 μM) and incubated for 6 days at 37°C. Cells were split (1:3) at day 3 postinoculation, at which time the cells were replenished with fresh medium containing the compounds. Intracellular HCV RNA levels in the infected cells were determined at day 6 by real time reverse transcription-quantitative PCR (RT-qPCR) on total cellular RNA extracted using the guanidinium thiocyanate (GTC)-acid phenol extraction method (23), as previously described (17).

(ii) Single-cycle infections.

Huh-7 cells (5 × 10⁴ cells/well; 24-well plates) were seeded in the presence of 5 × 10⁵ focus-forming units (FFU)/well (500 μl) of the D183 virus (18) and perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) Five hours postinfection, cells were washed twice with warm phosphate-buffered saline (PBS) and further incubated for 24 and 48 h in complete medium at 37°C. Samples of the cells and supernatants were collected at 24 and 48 h postinfection for HCV RNA quantification by RT-qPCR and infectivity analyses by titration.

(iii) Infection with HCVtcp.

Cell supernatants containing HCVtcp were mixed 1:1 with the compound dilutions required to obtain the desired final concentrations of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM). Mixtures were used to inoculate Huh-7 cells, which were cultured for 48 h before intracellular luciferase levels were measured using a commercially available kit (luciferase assay system; Promega-Madison, WI).

(iv) Persistent HCV infections.

Establishment of persistent JFH-1 infections was previously described (24). Cells were treated for 48 h (replenishing medium and inhibitors at 24 h) with perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) using DMSO and 2mAde (10 μM) as negative and positive inhibition controls.

(v) Time-of-addition experiments.

In order to discriminate inhibition of particle binding and postbinding events, we incubated prechilled target cells (Huh-7) for 1 h at 4°C in the presence of the perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM). After this adsorption period, cells were washed twice with cold PBS and replenished with warm DMEM–10% FBS and incubated for 72 h at 37°C. Parallel cultures inoculated with virus at 4°C in the absence of inhibitors (1 h) were washed twice with cold PBS and replenished with DMEM–10% FBS containing the compounds. Cultures were incubated in DMEM–10% FBS for an additional 72 h. Relative infection efficiency levels were determined by immunostaining and infection focus counting as previously described (25) using a monoclonal antibody against E2 (AR3A) kindly provided by M. Law (The Scripps Research Institute [TSRI], La Jolla, CA).

HCV pseudotype particle infectivity inhibition assay.

HCV E1/E2-pseudotyped retroviral particles bearing the luciferase reporter gene were generated as described previously (26) using reagents kindly provided by F. L. Cosset (INSERM-Lyon, Lyon, France). As controls, pseudotypes bearing vesicular stomatitis virus (VSV), feline retrovirus (RD114), or influenza virus envelope glycoproteins were produced in parallel. For the inhibition assays, particles were mixed 1:1 with medium containing DMSO or with compound dilutions to achieve the desired final concentration of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM). The mixture was used to inoculate Huh-7 or BCLC-5, -6, or -10 cells at 37°C for another 72-h period, after which infection efficiency was evaluated by measuring reporter gene expression using a commercial kit (luciferase assay system; Promega-Madison, WI). Relative infection values were obtained using DMSO-treated cells as a control (100%). Background levels were established by measuring luciferase expression in cells transduced with HCV pseudoparticles (HCVpp) lacking envelope glycoproteins, as previously described (26).

In vitro transcription and HCV RNA electroporation.

In vitro-transcribed subgenomic HCV replicon RNA was electroporated as described previously (27, 28). At 5 h postelectroporation, compounds were added at the indicated concentrations and left throughout the experiment. Firefly luciferase activities were measured in the sample using a commercial kit (dual-luciferase assay system; Promega-Madison, WI) at different times posttransfection.

Western blot analysis.

Quantification of NS3 protein and human beta actin by Western blot analysis was performed as described previously (28).

RESULTS

Identification of novel compounds inhibiting HCV infection.

In order to identify clinical compounds with potential against HCV infection, we screened a chemical library composed of 281 clinically approved compounds for non-HCV therapeutic purposes. The screening methodology was identical to that described previously (12). Briefly, highly susceptible human hepatoma cells (Huh-7.5.1 clone 2) were inoculated at a low multiplicity of infection (MOI of 0.01) with HCV (D183 [18]). Infected cultures were incubated in the presence of the compounds (10 μM) for 72 h at 37°C, after which time relative infection efficiency and cytotoxicity values were determined as described in Materials and Methods. We identified 12 compounds that reduced HCV infection by more than 1 order of magnitude without significantly reducing cell biomass (Table 1). Figure 1 displays the chemical structure of the primary hits. Interestingly, five of the primary hits were tricyclic antidepressants (Fig. 1; group 1), very similar to those we previously identified as HCV inhibitors using a similar approach (12). Another family of compounds is composed of three synthetic estrogen receptor modulators (SERMs), tamoxifen, clomifene, and raloxifene (group 2; Fig. 1), which have been shown to inhibit several aspects of HCV infection (29), similarly to what we had previously described for another SERM, toremifene citrate (12). Finally, we found hydroxyzine pamoate, a histamine H1 receptor antagonist that is effective in the treatment of chronic dermatological disorders and anxiety; fluoxetine, a highly specific serotonin uptake inhibitor used as an antidepressant; and benztropine mesylate, a muscarinic receptor antagonist used for symptomatic treatment of Parkinson's disease. Since we and others previously characterized compounds similar to those in group 1 described as HCV entry inhibitors (12) and since the antiviral activity of SERMs (group 2) against HCV has been studied previously (12, 29), we focused our attention on the compounds in group 3. In addition, we included in this study perphenazine (from group 1) as a control, since we expected it to block viral entry as we and others have shown previously for compounds with nearly identical structures (12, 30).

TABLE 1.

Primary hits in the compound library screening^a

Group	NCC_Structure_ID	Screening concn (μM)	Name	HCV infection (% of control)	Biomass (%)
1: tricyclic antidepressants	CPD000058254	10	Chlorpromazine	0.14 ± 0.20	90.19 ± 3.54
	CPD000058295	10	Clomipramine	0.22 ± 0.10	93.48 ± 0.76
	CPD000036827	10	Desipramine	0.22 ± 0.17	88.56 ± 5.99
	CPD000058180	10	Perphenazine	0.60 ± 0.52	88.22 ± 10.84
	CPD000058388	10	Imipramine	2.25 ± 1.87	106.96 ± 4.58
2: SERMs	CPD000058508	10	Raloxifene	0.09 ± 0.12	91.82 ± 20.05
	CPD001491671	10	Tamoxifen	1.09 ± 1.53	96.05 ± 4.91
	CPD001317855	2	Clomifene	5.04 ± 0.87	149.75 ± 2.62
3: others	CPD001370751	10	Hydroxyzine	4.59 ± 6.48	87.51 ± 6.8
	CPD000394012	10	Benztropine	0.51 ± 0.50	93.87 ± 3.11
	CPD000058452	2	Fluoxetine	15.41 ± 2.47	159.34 ± 2.00

^aRelative infection efficiency values were determined by colorimetry, using a standard curve with serial virus inoculum dilutions. Relative biomass values were subsequently determined in the same wells, also using a colorimetric readout, setting the biomass measured in the control wells to 100%. Data are shown as averages and standard deviations (SD) of the results of two independent experiments (n = 2).

FIG 1.

Chemical structure of the primary screening hits.

We first determined their potency by evaluating the antiviral activity of serial compound dilutions using the colorimetric readout method described above. The same compound concentrations were tested for cytotoxicity using an MTT-based colorimetric assay (31). The results of this analysis are listed in Table 2 and show that all the selected compounds displayed EC₅₀ values below 1 μM whereas the CC₅₀ values were significantly higher, with indexes (CC₅₀/EC₅₀) higher than 10. We note that hydroxyzine displayed the highest potency and the lowest toxicity, with a therapeutic index of 178.

TABLE 2.

Potency and toxicity of the selected compounds^a

Compound name	EC50 (μM)	EC90 (μM)	CC50 (μM)	S.I.
Hydroxyzine	0.26 ± 0.13	0.55 ± 0.25	46.25 ± 6.25	178
Benztropine	0.67 ± 0.29	2.75 ± 1.08	26.25 ± 1.76	39
Fluoxetine	0.45 ± 0.27	1.50 ± 1.05	10.25 ± 1.76	23
Perphenazine	1.00 ± 0.45	4.50 ± 1.44	10.50 ± 0.70	10

^aPotency (EC₅₀ and EC₉₀) indexes are shown as averages and standard deviations of a minimum of four independent determinations (n = 4). Cytotoxicity (CC₅₀) was calculated in parallel uninfected cell cultures, and values are shown as averages and SD of three independent determinations (n = 3). Selectivity indexes (S.I.) were calculated by determining the ratios of the average CC₅₀ and EC₅₀ values.

We set out to confirm that these compounds are capable of inhibiting HCV spread in cell culture using an alternative readout of infection. Huh-7.5.1. clone 2 cells were inoculated with JFH-1 virus (genotype 2a) at a low multiplicity (MOI of 0.01) in the presence of perphenazine (5 μM), hydroxyzine (5 μM), fluoxetine (3 μM), or benztropine (7.5 μM). Cell lysates were prepared at day 6 postinoculation and processed for Western blot analysis of NS3. Figure 2 shows how JFH-1 spread was inhibited by perphenazine, hydroxyzine, and benztropine by more than 10-fold compared with the virus spread in the presence of the vehicle (DMSO), reinforcing the notion that these compounds are effective HCV inhibitors. Despite its apparent potency, fluoxetine inhibits HCV spread by only less than 1 order of magnitude (Fig. 2), indicating that it is poorly effective at controlling viral spread compared with the other compounds. Therefore, fluoxetine was excluded from further studies.

FIG 2.

Selected compounds inhibit JFH-1 propagation. Huh-7.5.1 clone 2 cells were inoculated at a multiplicity of infection of 0.01 in the presence of perphenazine (5 μM), hydroxyzine (5 μM), benztropine (7.5 μM), or DMSO as a control. (A) Samples of the cells were harvested at day 6 postinoculation and analyzed by Western blotting against NS3 and beta-actin as a loading control. (B) Average values and standard deviations of the normalized NS3 signal are indicated (n = 3).

Selected compounds inhibit early aspects of HCV infection.

In order to determine which aspect(s) of the virus replication cycle is affected by the different compounds, we performed single-cycle infection experiments (MOI of 10) in the presence of the different compounds as described in Materials and Methods. Analysis of extracellular infectivity titers 48 h postinfection confirmed inhibition of the infection, as extracellular infectivity titers were reduced by 1 order of magnitude in the cultures treated with benztropine and 2 orders of magnitude for those treated with perphenazine or hydroxyzine (Fig. 3A). Intracellular HCV RNA levels display a parallel reduction of more than 2 orders of magnitude for the cells treated with perphenazine and hydroxyzine and more than 30-fold for those treated with benztropine (Fig. 3A), indicating that the selected inhibitors interfere with an early step of the infection leading to HCV RNA accumulation.

FIG 3.

Selected compounds inhibit early aspects of HCV infection. (A) Huh7 cells were inoculated at a MOI of 10 in the presence of perphenazine (5 μM), hydroxyzine (5 μM), benztropine (7.5 μM), or DMSO as a control. Samples of cells and supernatants were collected to determine infectivity titers as well as intracellular HCV RNA levels by titration and RT-qPCR, respectively, at the indicated time points. (B) Inhibition of HCVtcp infection. Huh-7 cells were inoculated with HCVtcp in the presence of perphenazine (5 μM), hydroxyzine (5 μM), and benztropine (7.5 μM), and samples of the cells were assayed for luciferase activity 48 h postinoculation. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). (C) Treatment of Huh-7 cells electroporated with a subgenomic HCV RNA replicon bearing a luciferase gene with perphenazine (5 μM), hydroxyzine (5 μM), benztropine (7.5 μM), or DMSO and 2mAde (10 μM) as negative and positive controls, respectively. Data are shown as averages and standard deviations of the results of two independent experiments performed in duplicate (n = 4). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

Similar results were obtained with defective reporter viruses produced by transencapsidation (HCVtcp) that produce a single round of infection (19). Figure 3B shows reductions in infection efficiency of more than 10-fold for the cells treated with perphenazine or hydroxyzine and more than 5-fold for those treated with benztropine determined 48 h postinfection. These results reinforce the notion that all the selected compounds inhibit an early step of the infection leading to viral RNA accumulation.

In order to determine if the selected compounds are targeting initial aspects of HCV RNA replication, Huh-7 cells were electroporated with a subgenomic replicon bearing a luciferase reporter gene to study primary translation and HCV RNA replication independently of viral entry. Treatment of these cells with perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) did not have any impact on HCV RNA replication as shown by accumulations of luciferase 24 h postelectroporation (Fig. 3C) similar to those seen with cells treated with DMSO and in contrast with the marked reduction observed in cells treated with the polymerase inhibitor 2′-c-methyladenosine (2mAde; 10 μM) (32). These results suggest that early HCV RNA replication is not affected by the selected compounds and indicate that these compounds might target a step upstream of primary translation, likely at the level of viral entry.

Analysis of the impact of the selected inhibitors on persistent infections.

Huh-7 cells persistently infected with JFH-1 virus were treated for 48 h with the antiviral concentrations of the compounds or with DMSO as a control. Intracellular HCV RNA and supernatant infectivity titers were measured to determine the impact of the inhibitors on HCV RNA replication and on virus production. Figure 4 shows that, as expected, 2mAde caused a profound reduction in intracellular HCV RNA contents and a comparable reduction in extracellular infectivity titers, indicating that the experimental setup is suitable to identify HCV RNA replication inhibitors. In contrast with 2mAde, none of the selected inhibitors reduced significantly HCV RNA contents in persistently infected cells, indicating that they do not inhibit HCV RNA replication in a persistent infection. Since the analysis of the extracellular infectivity titers could be interfered with by the presence of the compounds in the supernatants, we purified the infectious virions present in the supernatants of the infected cells using microfiltration devices. As shown in Fig. 4, neither hydroxyzine nor perphenazine had any impact on virus production and only a marginal (2-fold) reduction was observed in cells treated with benztropine. Overall, these data suggest that neither persistent HCV RNA replication nor infectious virus production is severely impaired by the selected compounds. Together with the data obtained in single-cycle infection and RNA electroporation experiments (Fig. 3), these results reinforce the notion that the selected inhibitors inhibit an early step of the infection preceding HCV RNA replication.

FIG 4.

Impact of the selected compounds on persistently infected cell cultures. Persistently infected cells were treated for 48 h with perphenazine (5 μM), hydroxyzine (5 μM), benztropine (7.5 μM), or DMSO and 2mAde (10 μM) as negative and positive controls, respectively. Samples of cells and supernatants were collected to determine infectivity titers as well as intracellular HCV RNA levels by titration and RT-qPCR, respectively. Viruses present in the supernatant were partially purified using microfiltration devices to eliminate residual compound present in the supernatant. Infectivity titers were determined in the purified material. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

Selected compounds inhibit entry of several HCV genotypes.

In order to determine if the selected compounds target initial steps of HCV infection, we studied whether they inhibit infection by HCV-pseudotyped retroviral particles (HCVpp). HCVpp infection is a suitable model to study entry inhibitors because it recapitulates early events of the infection mediated by the HCV envelope glycoprotein complexes, such as receptor recognition, particle internalization, and low pH-triggered, E1E2-mediated membrane fusion (26). This system enables measurement of the infectivity of HCVpp-bearing envelopes of different HCV genotypes. Thus, we tested the ability of the selected compounds to inhibit the infection of HCVpp from genotypes 1a, 1b, 2a, 2b, 3a, 4, and 5. Interestingly, we observed that perphenazine displays activity against genotypes 1 through 4 (Fig. 5). The genotype 2a JFH-1 strain, the one used in the original screening, was particularly susceptible to inhibition (Fig. 5). Strikingly, hydroxyzine (5 μM) displayed marked activity only against HCVpp from genotype 2, while the rest of the genotypes appeared to be only marginally susceptible (genotypes 1, 3, and 4) or not susceptible at all (genotype 5). Nevertheless, pronounced inhibition of the HCVpp infection of genotypes 1, 3, and 4, but not genotype 5, was observed with 25 μM hydroxyzine. Finally, benztropine inhibited infection by HCVpp from genotypes 1 through 4, although JFH-1 appeared to be particularly susceptible to the antiviral activity of this compound compared with the other HCV envelopes. On the other hand, genotype 5 was not susceptible to inhibition by any of the compounds. The infectivity of retroviruses pseudotyped with vesicular stomatitis virus (VSV) G protein, with feline retrovirus glycoprotein (RD114), or with influenza virus (FLU) envelope HA and NA glycoproteins was not significantly affected by the presence of the antiviral compounds (Fig. 5), indicating that they specifically inhibited HCV entry of genotypes 1 to 4 at the assayed concentrations.

FIG 5.

Inhibition of HCV entry by perphenazine, hydroxyzine, and benztropine. Huh-7 cells were infected with pseudotyped retroviruses bearing envelope glycoproteins from different HCV genotypes in the presence of perphenazine (5 μM), hydroxyzine (5 μM), benztropine (7.5 μM), or DMSO as a control. In addition, control pseudotypes bearing glycoproteins from VSV, influenza virus (FLU), and feline retrovirus (RD114) were used as a control to determine the specificity of the inhibition. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

In order to verify if inhibition by the selected compounds occurred in non-Huh7 cells, we took advantage of several human hepatocellular carcinoma cell lines, which support efficient infection by HCVpp (M. Coto-Llerena, N. Caro-Pérez, G. Koutsoudakis, L. Boix, J. M. López-Oliva, C. Fernandez-Carrillo, P. Fernandez, J. Bruix, P. Gastaminza, X. Forns, S. Perez-del-Pulgar, unpublished results). Infection of BCLC5, BCLC6, and BCLC10 cell lines with HCVpp (JFH-1) was performed in the presence of the selected compounds or in the presence of vehicle (DMSO). Significant inhibition of HCVpp infection was observed for all three compounds in all cell lines, including the reference Huh-7 cell line (Fig. 6). Although a small reduction in the effectiveness was observed for hydroxyzine in BCLC5 and BCLC6 cells, these results indicate that inhibition of HCV entry is not restricted to Huh-7 cells.

FIG 6.

Inhibition of viral entry by the selected compounds is not restricted to Huh-7 cells. BCL5, BCL6, and BCL10 were infected with HCVpp in the presence of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) or DMSO as a control. Huh-7 cells were infected in parallel as a control. Luciferase activity was measured in cell lysates 48 h postinfection. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

The foregoing data suggest that HCVpp-bearing genotype 2 envelopes are particularly susceptible to the antiviral activity of hydroxyzine. These results were confirmed in multiple-cycle infection experiments performed with chimeric JFH-1 viruses bearing the structural regions from genotype 1a (H77) and genotype 1b (Con1), where hydroxyzine (5 μM) inhibited the spread of genotype 2a (JFH-1) efficiently and inhibited infection by chimeric viruses expressing the structural regions from genotype 1 only slightly (Fig. 7). These differential susceptibility results were not observed with perphenazine and benztropine. These results confirm the differential susceptibilities of genotype 2 to hydroxyzine and reinforce the notion that this compound targets a step of the replication cycle driven by the viral envelope glycoproteins.

FIG 7.

Hydroxyzine displays selective activity against genotype 2 viruses. Huh-7.5.1 clone 2 cells were inoculated at a MOI of 0.01 with JFH-1 and recombinant H77C3 and Con1C3–JFH-1 chimeras in the presence of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) or DMSO as a control. Samples of the cells were collected at day 6 postinoculation, and intracellular HCV RNA levels were determined by RT-qPCR. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

JFH-1 infection is blocked at a postattachment step.

The results described above suggest that all the selected compounds inhibit infection at the level of viral entry, a multistep process that requires attachment to the target cell and internalization of the incoming viral particles. In order to determine whether the compounds target adsorption or internalization of the virion into the target cell, we inoculated naive Huh-7 cells with JFH-1 virus, in the presence of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM), using DMSO as a control. Inoculation was performed at low temperature (4°C) to allow virion adsorption but not internalization (7, 33). After 1 h, the compounds and unbound virus were washed with PBS and the cells were replenished with fresh media without inhibitors (Fig. 8A; adsorption). Parallel cultures were inoculated similarly but in the absence of inhibitors, washed, and incubated in the presence of the selected inhibitors for the rest of the experiment (Fig. 7A; postadsorption). The relative levels of infection efficiency in the different conditions were determined by immunofluorescence microscopy and infection focus counting (25). The results are shown in Fig. 8B, where the relative infectivity levels are shown to have been comparable between the control and the selected compounds, with the exception of perphenazine, when compounds were present during viral adsorption. These results suggest that hydroxyzine and benztropine do not interfere with viral particle stability or with physical attachment of the virion to the target cells. In contrast, the relative infectivity levels were significantly different when the compounds were added postadsorption, indicating that they blocked infection at a step downstream of particle attachment.

FIG 8.

Hydroxyzine and benztropine inhibit viral entry at a postattachment step. (A) Huh7 cells were inoculated with JFH-1 virus dilutions in the presence of perphenazine (5 μM), hydroxyzine (5 μM), or benztropine (7.5 μM) or vehicle (DMSO) as a control at 4°C for 1 h in the presence or absence of the compounds. Cells were washed and replenished with fresh media with or without the compounds, depending on whether the cells had been exposed to the antivirals during the adsorption phase. (B) Relative infectivity titers were determined by immunofluorescence microscopy. Data are shown as averages and standard deviations of the results of two independent experiments performed in triplicate (n = 6). The statistical significance of the differences with the control data set was determined using Student's t test (*, P < 0.05; **, P < 0.01).

DISCUSSION

In this report, we describe the screening of a library of compounds used in clinical practice for non-HCV applications. Using an unbiased screening technology, we have identified 12 nontoxic compounds with activity against HCV in cell culture. The primary hits were divided into three groups (Fig. 1) based on structure and known primary pharmacological utility. The first group is constituted by “tricyclic antidepressants,” both phenothiazines (chlorpromazine and perphenazine) and dibenzazepines (clomipramine, desipramine, and imipramine). Interestingly, in a previous screening, we found that several compounds of this class inhibit HCV entry (12). Mechanistic studies have shown that phenothiazines interfere with E1E2-induced membrane fusion by changing the properties of the endosomal membrane (30). We confirmed that this class of inhibitors interferes with viral entry of different HCV genotypes (Fig. 5) using as a representative member of the group perphenazine, which has also been recently identified as an HCV inhibitor by others (34). We have also reported previously that dibenzazepines inhibit HCV entry (12). The structural similarities between dibenzazepines and phenothiazines; i.e., planar molecules with aromatic rings and a tertiary amine, suggest similar mechanisms of inhibition of HCV entry.

The second group of compounds includes selective estrogen receptor modulators (SERMs). The members of this class of compounds were previously characterized as HCV inhibitors initially in the context of the replicon system, identifying tamoxifen as an inhibitor of HCV RNA replication by interfering with proviral functions of estrogen receptor alpha (ERa) (35). We and others subsequently showed that SERMs could also inhibit viral entry and virion production (12, 29), indicating that SERMs may interfere with multiple steps of the viral life cycle. In this sense, raloxifene, a SERM which displays anti-HCV activity in cell culture (36), ameliorates the cure rates when used in combination with IFN-α and ribavirin in postmenopausal women (37), suggesting that this class of inhibitors may have some therapeutic value in combination with other antivirals. In this study, we focused our attention on hydroxyzine and benztropine, which are benzhydryl-containing compounds with antihistaminic properties. While it is tempting to speculate that these characteristics are responsible for their antiviral activity, further studies will be necessary to determine the structural and pharmacological properties required for antiviral activity.

Study of perphenazine (the putative entry inhibitor), hydroxyzine, and benztropine revealed that they effectively inhibit HCV infection in multiple-cycle (Fig. 2 and 7) and single-cycle (Fig. 3A and B) infection experiments at concentrations in the low-micromolar range. Interestingly, none of these compounds interfere with HCV RNA replication after RNA transfection (Fig. 3C) or in persistently infected cell cultures (Fig. 4), and only benztropine displays a marginal inhibitory effect on virus production. These results indicate that the selected compounds inhibit HCV infection by blocking an early step preceding HCV RNA replication. This hypothesis was confirmed using HCVpp as a surrogate model for HCV entry, as the three compounds inhibited HCVpp infection for genotypes 1, 2, 3, and 4 (Fig. 5). Time-of-addition and low-temperature inoculation experiments showed that all three compounds were most active when added postadsorption (Fig. 8), indicating that they target a step downstream of virion attachment.

Intriguingly, a high concentration (>40-fold the EC₉₀) of hydroxyzine was required for inhibition of entry of genotypes 1, 3, and 4, indicating that genotype 2 strains are particularly susceptible to inhibition by this compound (Fig. 5 and 7). These results may suggest that there are important intergenotypic differences in susceptibilities to similar entry inhibitors. The differential levels of inhibition by hydroxyzine are not unique, as other reported HCV entry inhibitors display important intergenotypic differences. In this regard, two recent reports describe 1,3,5-triazine derivatives as HCV entry inhibitors that selectively block entry of genotype 1a and 1b but not that of other tested genotypes (38, 39). These results may be interpreted as indicating that the inhibitors target directly the envelope glycoproteins, which display very high genetic variability among HCV genotypes (40), and that genetic differences account for the differential levels of susceptibility. Future studies, including studies of selection of resistant mutants, determination of the genetic determinants leading to resistance, and comparison with genotypes displaying reduced or no susceptibility, such as the genotype 5 isolate in our study (Fig. 5), may shed light onto the molecular basis for the differential levels of susceptibility. Alternatively, the different susceptibility levels among the different isolates (Fig. 5) may suggest that, while the routes of entry of different HCV isolates require similar host cell factors (6), there might be differences in the usage of such factors. In fact, it has recently been reported that different HCV isolates may use different members of the claudin family for viral entry as an alternative to claudin-1 (41), which was originally shown to play an essential role in viral entry (42). In a study by Haid et al., when virus infection was neutralized with anti-claudin1 antibodies, some isolates appeared to escape inhibition by using claudin-6 as an alternative receptor (41). These results illustrate the complexity associated with inhibiting entry of different genotypes and reinforce the notion that a large assortment of entry inhibitors might be required for preventing infection by different isolates.

HCV entry inhibitors are not currently used in clinics for the treatment of chronic HCV infection (5). This is in part due to the fact that it is not clear that viral entry is required to maintain chronic HCV infection, as HCV could spread from cell to cell using mechanisms that are different from entry of the incoming virions (43, 44). Nevertheless, it has been recently shown that HCV entry inhibitors contribute to viral suppression in persistently infected cell cultures when coadministered with replication inhibitors (45), underscoring the potential utility of entry inhibitors in combination treatments. While the utility of entry inhibitors in chronic HCV is still under debate, the use of such inhibitors in the transplant setting appears to be a promising approach to prevent liver graft reinfection, which occurs in nearly 100% of the cases. The combination of sofosbuvir and ribavirin has recently been shown to prevent hepatitis C recurrence in around two-thirds of HCV-infected patients treated while awaiting liver transplantation (46). Nevertheless, relapse occurred in patients who had undetectable HCV RNA for less than 30 days before transplantation. It has been proposed that infection of the graft in this situation occurs because residual circulating virions infect the new graft during transplantation (47). Therefore, addition of a drug inhibiting HCV entry might prevent or decrease the incidence of relapse in this liver transplant setting.

While translation of our results to the clinic should be undertaken cautiously, it is noteworthy that there are no reports of liver toxicity induced by either hydroxyzine or benztropine. Moreover, hydroxyzine is indicated for the treatment of pruritus in patients with liver disease, and the peak plasma concentration achieved in these patients falls into the range of concentrations active against HCV in cell culture (Table 2) (48, 49). On the other hand, benztropine displays a narrow therapeutic window and peak plasma concentrations that are lower than those required for antiviral activity in cell culture (50). Despite these potential limitations, we believe that these compounds deserve further investigation as potential anti-HCV antivirals, with hydroxyzine being a better candidate a priori.

Our results show that two drugs approved for clinical applications in humans, hydroxyzine pamoate and benztropine mesylate, are effective against HCV infection in cell culture. We believe that these and other related compounds deserve further investigation for their potential applications for treatment of chronic HCV infection as well as in the transplant setting. In addition to their therapeutic potential, they constitute excellent tools to study basic aspects of HCV entry, including the expected intergenotypic differences in terms of susceptibility to this class of inhibitors. Such studies would certainly contribute to designing better HCV entry inhibitors, which might be important components of future therapeutic strategies in combination with inhibitors targeting other aspects of the infection.