Development of a Cell-Based High-Throughput Specificity Screen Using a Hepatitis C Virus-Bovine Viral Diarrhea Virus Dual Replicon Assay

Abstract

The hepatitis C virus (HCV) replicon is a unique system for the development of a high-throughput screen (HTS), since the analysis of inhibitors requires the quantification of a decrease in a steady-state level of HCV RNA. HCV replicon replication is dependent on host cell factors, and any toxic effects may have a significant impact on HCV replicon replication. Therefore, determining the antiviral specificity of compounds presents a challenge for the identification of specific HCV inhibitors. Here we report the development of an HCV/bovine viral diarrhea virus (BVDV) dual replicon assay suitable for HTS to address these issues. The HCV reporter enzyme is the endogenous NS3 protease contained within the HCV genome, while the BVDV reporter enzyme is a luciferase enzyme engineered into the BVDV genome. The HTS uses a mixture of HCV and BVDV replicon cell lines placed in the same well of a 96-well plate and isolated in the same cell backgrounds (Huh-7). The format consists of three separate but compatible assays: the first quantitates the amount of cytotoxicity based upon the conversion of Alamar blue dye via cellular enzymes, while the second indirectly quantitates HCV replicon replication through measurement of the amount of NS3 protease activity present. The final assay measures the amount of luciferase activity present from the BVDV replicon cells, as an indicator of the specificity of the test compounds. This HCV/BVDV dual replicon assay provides a reliable format to determine the potency and specificity of HCV replicon inhibitors.

The hepatitis C virus (HCV) replicon system described by Lohman et al. has provided the first reliable tissue culture based model of HCV replication (36). The subsequent description and isolation of more efficient replicons have improved the reliability even more while providing information on the interplay between virus and cell (2, 6, 27, 36, 44). The replicons were engineered to express neomycin phosphotransferase protein from the native HCV 5′ internal ribosome entry site (IRES) element while the nonstructural proteins NS3-NS5B are translated by the IRES from encephalomyocarditis virus (EMCV) (36). The known viral specific enzymatic activities provided by the replicon include the protease/helicase/ATPase NS3 (11, 16, 29, 52), NS4A (cofactor of NS3 protease) (12, 34), and the RNA-dependent RNA polymerase NS5B (4, 35, 37). The functional roles of NS4B and NS5A are still under investigation, with NS5A being implicated in interferon resistance and NS4B being reported to localize in the endoplasmic reticulum (20, 25, 38, 40, 47, 49, 54). Both proteins are believed to be essential for viral RNA synthesis and contribute to replication efficiency in concert with NS3 and NS5B (6, 21, 30, 44).

The HCV replicon system provides a useful tool for the high-volume screening of compounds effective against HCV replication. Methods used to measure HCV replicon replication include colony selection (35), quantitative RT-PCR (qRT-PCR) for RNA levels (50), immunological methods for proteins such as enzyme-linked immunosorbent assay (32, 47) or Western analysis (42, 44), and expression of reporters such as luciferase and beta-lactamase by replicons (30, 41). Although all these methods can be used for identification of HCV replication inhibitors with the proper controls, issues related to cost, throughput, effects of non-HCV sequences on replication, and ease in distinguishing specific or nonspecific inhibitors need to be addressed (19, 23).

In this report, we describe a cell-based high-throughput screen (HTS) assay which measures three enzymatic functions. The individual assays provide data related to cytotoxicity, HCV inhibition, and specificity in a single well of a 96-well plate. A conventional method, the conversion of Alamar blue dye via cellular enzymes, was used to monitor cytotoxicity (1). A novel fluorescence resonance energy transfer (FRET) assay method was used to determine the potency of inhibitors on HCV replicon replication. This reporter method is based on the assumption that when HCV RNA replication is inhibited, the amount of viral proteins, such as NS3 protease, will decrease. The amount of the NS3 protease can be quantitated by activity and used to quantitate viral RNA levels similar to other enzymatic reporter assays using luciferase, secreted alkaline phosphatase, chloramphenicol transferase, beta-lactamase, or beta-galactosidase (8, 22, 24, 30, 33, 39, 41, 45, 55). We also utilized a bovine viral diarrhea virus (BVDV) replicon cell line containing a luciferase reporter gene as a specificity screen for HCV inhibitors. The use of a mixed-replicon cell format in a 96-well plate provides an economical method for discriminating between the related but distinct viruses HCV and BVDV. Compounds specific for one virus can quickly and easily be discerned, increasing the likelihood of identifying highly selective inhibitors. The use of specific and nonspecific inhibitors is presented to demonstrate the ability of the system to detect BVDV- or HCV-specific inhibitors, increasing the likelihood of identifying compounds targeting viral functions. The HTS assay involves very few manipulations and is specific for the replicon cell lines. The results are also shown to be comparable to those obtained from qRT-PCR, providing a sensitive HTS assay without further purifications or manipulations.

MATERIALS AND METHODS

Generation of HCV and BVDV replicon cell lines.

The HCV replicon cell line was isolated from colonies as described by Lohman et al. (36) and used for all experiments (Fig. 1). HCV replicon cell lines were maintained at 37°C, 5% CO₂, and 100% relative humidity in Dulbecco's modified Eagle medium (11965-084; Life Technologies) with 10% heat inactivated calf serum (Sigma), penicillin-streptomycin (Life Technologies), and Geneticin at 1 mg/ml (Life Technologies).

FIG. 1.

HCV and BVDV replicons. 5′ and 3′ indicate the cis sequence elements at the 5′ and 3′ ends of the RNA molecules necessary for viral replication. Neo is the selectable marker, EMC IRES is an IRES for internal translation of the nonstructural region of the genome, and Luc is the luciferase reporter for BVDV. A ubiquitin sequence (Ub) was placed between the luciferase and neomycin phosphotransferase genes to allow proteolytic processing. NS3 and NS4A encode the protease and activator of protease used in the cell-based FRET reporter assay for HCV, while the luciferase enzymatic activity is the reporter used for BVDV. The sites of polyprotein processing by NS3 are indicated by the arrows.

To generate a BVDV replicon (BVDV-bu), the SphI/BglII fragment from 153E-2 (51) was ligated with the SphI/BssHII fragment from 166A-4 (51), a BssHII/SacII fragment from ubiquitin and a SacII/BglII-digested PCR fragment which was the BVDV NS3 region amplified to incorporate the C terminus of ubiquitin onto the 5′ end of NS3. A firefly luciferase gene was then amplified by standard PCR methods to add BssHII sites at each end and cloned into BVDV-bu at a BssHII site at nucleotide 740 by nondirectional cloning to generate BVDV-Luc. The neomycin gene and EMCV IRES were PCR amplified from the HCV genotype 1b replicon plasmid and ligated into BVDV-Luc to generate the final clone (BVDV-Luc-neo), consisting of the BVDV 5′ untranslated region (UTR) followed by the gene for firefly luciferase, a ubiquitin monomer, the neomycin phosphotransferase gene, the EMCV IRES, BVDV NS3-NS5B, and the BVDV 3′ UTR (Fig. 1). Stable BVDV-Luc-neo cell lines were generated and maintained as described above by selection with G418 at 0.5 mg/ml in Dulbecco's modified Eagle medium. BVDV RNA levels in these cell lines were examined directly by quantitative TaqMan RT-PCR, and BVDV proteins were confirmed by Western blotting (data not shown). In addition, the BVDV luciferase assay was validated in these cell lines by examining luciferase levels in the presence and absence of compound 1453, a specific inhibitor of BVDV replication. As determined by luciferase, the 50 effective concentration (EC₅₀) of compound 1453 was ∼1 μM, comparable to previous results obtained with BVDV infection of MDBK cells (51).

Assays.

qRT-PCR was performed as recommended in the instructions for the Gibco-BRL Platinum quantitative RT-PCR Thermoscript one-step kit on a Perkin-Elmer ABI Prism model 7700 sequence detector. The primers for TaqMan were selected for use following analysis of RNA sequences with Primer Express software from ABI. Primers used for detection of the plus strand RNA were 131F (5′ GGGAGAGCCATAGTGGTCTGC 3′) and 231R (5′ CCCAAATCTCCAGGCATTGA 3′), which amplify the HCV 5′ UTR from nucleotides 131 to 231. The probe used for detection, 5′ FAM-CGGAATTGCCAGGACGACCGG-BHQ1 3′, was obtained from Biosearch Technologies. RNAs were purified from 96 wells by using the RNeasy 96 kit from Qiagen. Experiments were performed in duplicate.

Western analysis was performed according to the instructions for Amersham's chemiluminescence immunology kit (NEL105 Renaissance) using a Molecular Dynamics Storm 860 Phosphorimager and associated software. The primary and secondary antibody dilutions were 1 to 5,000. Antiserum was generated by immunizing rabbits with purified NS3 protease made from an Escherichia coli expression vector encoding the first 181 amino acids of HCV 1a NS3 with subsequent boosters. Blood samples were tested weekly, and boosters continued until a positive signal on a control Western blot was seen relative to a Huh-7 extract. Secondary antibody was a Bio-Rad (170-6515) goat anti-rabbit immunoglobulin G horseradish peroxidase conjugate. The protein samples for Western analysis were from the wells used for the FRET assay and were prepared by the addition of an equal volume of 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) buffer to the FRET assay mixture, followed by heating and loading on a 10% acrylamide gel for SDS-PAGE. Alpha interferon (IFN-α) was obtained from Sigma (I-4276) and stored as recommended.

To perform the screening assay, 96-well cell culture black plates with clear bottoms were used. The FRET peptide was described previously (53). The peptide (Ac-Asp-Glu-Asp [EDANS]-Glu-Glu-Abu-[COO] Ala-Ser-Lys [DABCYL]-NH2) contains a fluorescence donor {EDANS, 5-[(2-aminoethyl)amino]naphthalene-1-sulfonic acid} near one end of the peptide and an acceptor {DABCYL, 4-[(4-dimethylamino)phenyl]azo)benzoic acid} near the other end. The fluorescence of the peptide is quenched by intermolecular resonance energy transfer between the donor and the acceptor, but as the NS3 protease cleaves the peptide, the products are released from resonance energy transfer quenching, and the fluorescence of the donor increases over time as the substrate is cleaved by NS3 protease (Fig. 2). The assay reagent was made from the following: 5× luciferase cell culture lysis reagent from Promega (E153A) diluted to 1× with dH₂O, NaCl added to 150 mM (final), the FRET peptide diluted to 20 μM (final) from a 2 mM dimethyl sulfoxide (DMSO) stock. To prepare the plates, BVDV and HCV cells or HCV cells alone were trypsinized, placed in each well of a 96-well plate, and allowed to attach overnight (10⁴ cells per well [final], equal number of HCV and BVDV cells when mixed). The next day, test compounds were added to wells throughout columns 1 through 10 at a single concentration for HTS; column 11 and column 12 were used as controls and contained a titration of IFN-α or HCV-specific protease inhibitor (Fig. 3) or controls (9, 14, 15, 18, 46). At the highest concentration of IFN-α (1,000 U/ml) or HCV protease inhibitor (250 nM), >95% of HCV replicon replication was inhibited as judged by qRT-PCR (data not shown). All wells had a 0.2-ml final volume and 0.5% DMSO (nontoxic at this level; data not shown). The plates were then placed back in the incubator. Figure 3 shows the layout for HTS of the replicon cell line in a 96-well plate. Seventy-two hours later, the plate was removed and an Alamar blue solution (00-100; Trek Diagnostics) was added to 10% per well for a measure of cellular toxicity. After a reading in a Cytofluor 4000 instrument (PE Biosystems; ∼5 h after Alamar blue addition), plates were rinsed with phosphate-buffered saline and then used for the FRET assay by the addition of 30 μl of the FRET peptide assay reagent (described above) per well. The plate was then placed in the Cytofluor 4000 instrument, which had been set to 340 nm (excitation)/490 nm (emission) and automatic mode, for 20 cycles or less, and the plate was read in a kinetic mode. Typically, the signal-to-noise ratio obtained by using an endpoint analysis after the readings was at least threefold. Following FRET, 40 μl of luciferase substrate (Promega kit for firefly luciferase E4550) was added to each well and the plate was placed in a Top Count (Packard Instruments) programmed for luciferase measurements. In this study, titrations of known inhibitors were used to demonstrate the ability of the triple assay method to discern between HCV and BVDV inhibitors, since random, nonspecific compounds are usually toxic and inhibit HCV as well as BVDV (data not shown).

FIG. 2.

(A) Cell-based FRET assay. When HCV replicon replication is inhibited by inhibitors, the amount of HCV NS3 protease is also decreased. The amount of NS3 protease in the cell lysate then can be quantitated by the amount of the peptide substrate (AcDED[EDANS]EEAbu-y-[COO]ASK[DABCYL]-NH2) being cleaved. When the substrate is cleaved by HCV NS3 protease, it generates fluorescence. Decrease of fluorescence signal in the presence of inhibitors is due to inhibition of replicon replication. (B) Measurement of the increase in fluorescence of the HCV FRET peptide in the HCV replicon cell line and the effect of exposure to various interferon concentrations. The IFN-α concentrations (units per milliliter) used for the different wells are on the right. The assay is linear in this example over a period of 40 min.

FIG. 3.

Diagram of 96-well plate layout for HCV replicon HTS with values from Alamar blue readings and the HCV FRET assay indicated in each well. “Screen” indicates wells with test compounds; “Control” represents wells with DMSO only and is defined as 100% activity (column 11, wells A through H); “Inhibited” contains the greatest amount of a control inhibitor (100% inhibited; wells A12 and B12) and is used to determine background on each plate; “Titration” indicates the titration of interferon or a specific inhibitor and is used as a sensitivity control (wells C12 through H12). The Alamar blue reading from the random compound plate is expressed as a measure of cytotoxicity, with a low number being nontoxic. The compound in F2 shows very little toxicity, while the compound in G5 has substantial toxicity (bold outlines). The percentage of HCV protease activity in each well obtained by the calculations described in the text for the endpoint reading from cycle 21 of the FRET assay is also indicated, with low numbers being indicative of HCV replicon inhibition. The FRET numbers indicate that the compounds in wells F2 and G5 inhibited the HCV replicon ∼74% and 100%, respectively. Comparing the results of the FRET assay with the Alamar assay, it is likely that the inhibition of the HCV replicon for G5 is due to a toxic mechanism while the inhibition due to compound in F2 is not toxic in this assay, suggesting that the compound may be specific for HCV.

Calculations for the three assays were performed to determine BVDV inhibition, HCV inhibition, and percent cytotoxicity. The percent cytotoxicity was determined by Alamar blue conversion to fluorescent product, the percent HCV inhibition was determined by fluorescence increase due to peptide cleavage, and the amount of BVDV inhibition was determined by light units from luciferase activity. The percent BVDV inhibition was quantified relative to a specific BVDV test compound (compound 1453 [51]) in the wells, while wells containing HCV inhibitor only were used as 100% BVDV luciferase activity. The concentration of compound 1453 was chosen so that the highest dilution used inhibited BVDV 100%, was nontoxic to the cells, and did not affect HCV replication (10 μM). The luciferase amounts from 100% inhibited wells were averaged and used as the background luciferase value. This value was subtracted for all wells before percent luciferase activity was calculated. Compound analysis also depended upon the quantification of the relative HCV replicon inhibition and the relative cytotoxicity values. To calculate the HCV replicon inhibition values, an average background FRET signal was obtained from the two wells containing the greatest amount of IFN-α or HCV protease inhibitor at the end of the assay period. These numbers were then subtracted from the average FRET signal obtained from the control wells in row 11, and this number was used as 100% activity. The individual signals in each of the compound test wells were then divided by the averaged control values after background subtraction and multiplied by 100 to determine percent activity. Alternatively, the Cytofluor instrument can be set to calculate rates relative to control wells to determine the EC₅₀s directly from the increase in fluorescence over the linear time of the assay. To calculate cytotoxicity values, the average Alamar blue fluorescence signals from control wells were set as 100% nontoxic. The individual signals in each of the compound test wells were then divided by the average control signal and multiplied by 100% to determine percent viability. EC₅₀s were calculated as the concentration which caused a 50% reduction in HCV RNA, HCV protein amounts, BVDV luciferase values, or FRET activity, while cytotoxic concentrations for 50% reduction were calculated from the Alamar blue values which led to a 50% reduction, using the Cytofluor instrument scaling to control wells. The three numbers generated for the compound plate—percent cytotoxicity, percent HCV activity, and percent BVDV activity—were used to identify compounds of interest for further analysis.

Z′ is calculated as 1 − [(3asds + 3asdb)/(as-ab)], where asds is the standard deviation of the signal, asdb is the standard deviation of the background, as is the average signal, and ab is the average background signal.

RESULTS

The HCV replicon was constructed and isolated as described by Lohman et al. (36). Analysis of the replicon sequence identified a serine deletion at residue 2197, a common adaptive mutation in NS5A (5). A similar BVDV replicon was constructed with a luciferase gene included (Fig. 1). The luciferase gene is fused to a selective marker (neomycin phosphotransferase) via a ubiquitin sequence between the two proteins. The luciferase is released after translation of the fusion protein and proteolytic processing due to ubiquitin. To minimize the difference between host cells, both HCV and BVDV replicons were isolated following RNA transfection into Huh-7 cells.

HCV FRET substrate specificity.

The FRET assay peptide substrate was originally developed for measurement of in vitro NS3 protease activity using purified HCV NS3 protease (53). In the assay, the amounts of NS3 protease and peptide substrate are constant, and the protease activity decreases with increasing concentrations of NS3 protease inhibitors. To develop a cell-based FRET assay, we assumed that when HCV RNA replication is inhibited, the amount of viral proteins such as NS3 protease will decrease. The change in the amount of viral proteins, including NS3 protease, can then be determined by either Western blot analysis or determination of the amount of NS3 protease activity present. Prior to development of a cell-based assay using the FRET peptide, we examined the specificity of the peptide cleavage in crude cell lysates. The peptide substrate was added to individual extracts made from either naive Huh-7 cells or mixtures of Huh-7 plus BVDV replicon cells and was found to yield a substantial increase in fluorescence in extracts only from either cells containing the HCV replicon or cells expressing the HCV NS3 enzyme. This indicated that the assay was specific for the HCV protease and that the HCV FRET peptide was not cleaved by the BVDV protease (data not shown).

Validation of FRET assay.

To validate the HCV FRET assay, we compared the IFN-α EC₅₀ obtained from FRET, Western blotting, and qRT-PCR. The samples for these measurements were from two 96-well plates prepared the same day and treated at the same time with a titration of IFN-α. One plate was used for preparation of RNA for quantitative RT-PCR while the other plate was used for FRET. Samples from the same wells after the FRET assay were used for Western analysis. The FRET assay is illustrated in Fig. 2A, and the results with IFN-α titration following 96 h of incubation are shown as a continuous kinetic graph (Fig. 2B). In the absence of IFN-α, the FRET signal increased with time and was linear for at least 40 min. A decrease in the rate of FRET activity is evident in the graph with increasing IFN-α concentration, indicating replication inhibition with a subsequent decrease in NS3 protease activity. The numbers from the linear range were used for determination of the IFN-α EC₅₀. Instead of kinetic analysis, the endpoint signal could also be used for EC₅₀ determination as long as numbers were taken from the linear portion of the curve. Although no significant EC₅₀ differences were observed between the readout using kinetic analysis and the endpoint signal, we preferred the kinetic analysis because of less interference due to potential compound fluorescence. RNA levels were measured by qRT-PCR (data not shown), while the amount of NS3 protein in each well was quantitated by scanning a Western immunoblot. Similar to the FRET activity, the Western blot analysis showed that the amount of NS3 decreased with increasing concentrations of IFN-α (Fig. 4A). The titration curves comparing qRT-PCR, FRET, and Western blots are shown in Fig. 4B. Our results indicated EC₅₀s (units of IFN-α per milliliter) of 1.9 for the Western blot, 2.9 for the FRET, and 5.3 for qRT-PCR. These values are within threefold of one another and demonstrate equivalency between the assay methods. Our result also demonstrated that the FRET assay, like the Western blot analysis and qRT-PCR, could be used to accurately determine EC₅₀s of inhibitors.

FIG. 4.

(A) Western immunoblot using an anti-NS3 protease serum for the determination of the EC₅₀ of IFN-α. NS3 proteasespecific bands were quantified by phosphorimaging. (B) Comparison of typical values determined by FRET, qRT-PCR, or scanning of a Western blot for titration of interferon in the HCV replicon cell line. Each value represents a well of a 96-well plate at a single interferon concentration relative to a control value. Values obtained were within threefold of each other. Data at the lowest concentration of interferon tended to contain more variation.

Random compound evaluation for FRET and Alamar blue.

To further validate the HTS format for an HCV replicon screen, a plate with fixed concentrations of a diverse collection of compounds was utilized to test both the Alamar blue assay and FRET HCV replicon assay. The layout of a 96-well plate is shown in Fig. 3, and a detailed description is presented in Materials and Methods. A solution of Alamar blue was added to plates after 3 days of incubation with compounds, allowing direct quantification of the level of toxicity in each well by monitoring the cellular conversion of the nontoxic dye. The numbers in the Alamar blue assay are presented as percentage of cytotoxicity with approximately a 12% variation observed in the control cells containing medium only (Fig. 3, wells A11 to H11). After Alamar staining, the same plate was used for the FRET assay. The FRET results are presented as the percent remaining HCV NS3 protease activity (Fig. 3). For example, the HCV NS3 protease activity in wells A12 and B12 was <1%, suggesting that HCV replicon replication was inhibited >99% after 72 h of treatment with IFN at 1,000 U/ml. Modest cytotoxicity (16 to 20%) was observed at this IFN concentration (Fig. 3, wells A12 and B12). Screened compounds are prioritized using the numerical data from both the Alamar (toxicity) and the FRET (HCV replicon) assay. Figure 3 highlights two compounds which showed a noticeable reduction in FRET activity (wells F2 and G5). Inspection of the numbers in both plates indicated that the decreased FRET activity in well G5 is most likely due to cytotoxicity based on the 82% reduction in Alamar staining. In contrast, well F2 is seen to have a noticeable decrease in FRET activity without a corresponding decrease in the Alamar blue measurement, indicating HCV replicon inhibition without measurable toxicity for this compound.

Statistical evaluation of FRET assay.

Minimal signal variation is one of the key factors required for an HTS to lessen the number of false signals. To determine whether the variation in the FRET assay would remain acceptable, 40 additional compound plates were used to quantitate the variation, using statistical analysis to measure the Z′ statistic (56). The Z′ statistic is a measure of the distance between the standard deviations for the signal versus the noise of the assay. If the signal-to-noise scatter in the assay plates is considered acceptable, the Z′ statistic should be 0.5 or greater (56). Forty plates were used to measure the standard deviations and the number distribution between the endpoint signal obtained for the controls and the signal obtained for the background. The data are shown graphically in Fig. 5. Using this calculation, a Z′ of 0.62 was obtained, indicating an acceptable plate-to-plate variation for HTS.

FIG. 5.

Graphical representation of the averaged numbers from 40 separate compound plates used in the Z′ calculation. The numbers at a signal of ∼500 are the readings from the wells containing 1,000 U of IFN and are considered to have 0% FRET activity. The numbers at a signal of ∼1,500 are from wells containing buffer only and are considered to have 100% FRET activity. The Z′ measurement calculates the deviation associated with the measurements, with 2 standard deviations of the means of each measurement being acceptable (see Materials and Methods).

Mixing of replicon cells for specificity.

Potential HCV replication inhibitors, identified from cell based replicon screens, usually go through counterscreens to determine their specificity using different cell lines as well as different viruses. Many viruses, especially the positive-stranded RNA viruses, could serve this purpose (10, 26, 28, 48). BVDV is an HCV-related virus and shares many similarities with HCV regarding genome organization and nonstructural proteins required for viral RNA replication (3). A BVDV replicon cell line was isolated with a luciferase reporter gene in the same Huh-7 cells used to isolate the HCV replicon cell line (Fig. 1). If a compound inhibits replication of both HCV and BVDV replicons with no apparent cytotoxicity, it would not be automatically discarded because of the similarities of both replicons. However, if a compound, especially for nonactive site inhibitors, inhibits the HCV replicon but not the BVDV replicon, it would be assigned a higher priority and be further investigated, since HCV specific inhibitors are being sought. Inhibitors were tested in a mixture of HCV and BVDV replicon cells in the same well to discern specific inhibitors. To develop stringent conditions for specific HCV inhibition, we used the less sensitive FRET assay for the HCV replicon and a more sensitive luciferase assay for the BVDV replicon. Compound 1453, a previously identified specific inhibitor of BVDV RNA replication, was used as a BVDV replicon control (51), and a compound specific for the HCV NS3 protease (7), a peptide mimetic inhibitor, was used as an HCV replicon control (43). The BVDV inhibitor, compound 1453, was identified through a cell-based BVDV screen and is a member of a class of compounds known as cyclic ureas. A resistant virus was isolated, and the mutation was mapped to the BVDV RNA-dependent RNA polymerase. Interestingly, although the inhibitor inhibited replication of the BVDV virus (51) and the replicon (this study) as well as a replication complex, no apparent inhibition was observed in the in vitro polymerase assay (51). HCV and BVDV luciferase replicon cells were mixed and plated together in a 96-well plate. Compound 1453 and an HCV NS3 protease inhibitor were titrated in increasing amounts from rows H to A in five columns. After 72 h of incubation with the inhibitors, the three assays, Alamar blue conversion, HCV FRET peptide cleavage and BVDV luciferase activity, were performed. The plate layout and raw data are shown in Tables 1 to 3. Comparison of the wells containing the titration of the BVDV-specific inhibitor compound 1453 with the Alamar blue numbers indicated no detectable toxicity (Table 1) and no HCV inhibition (Table 2) but titratable BVDV inhibition, with a EC₅₀ of ∼1 μM (Table 3), similar to previously reported EC₅₀s obtained from BVDV viral infection (51). Similarly, the HCV-specific protease inhibitor has no detectable cytotoxicity (Table 1) and no BVDV inhibition (Table 3) but titratable HCV inhibitory activity (Table 2), with an EC₅₀ of 10 nM, which is also similar to reported values (7, 43). DMSO only was added to the cells in columns 11 and 12 as controls for these assays. These results show the ability of the assay to measure cytotoxicity, HCV inhibition, and BVDV inhibition in the same well in a 96-well format. To ensure that the mixture of cells does not interfere with either assay, the HCV NS3 protease inhibitor and compound 1453 were tested in HCV replicon and BVDV replicon cells separately. A change in EC₅₀ was not observed (data not shown).

TABLE 1.

Alamar blue readings for BVDV and HCV replicon cells in a 96-well plate^a

Row	Reading for well containing:
	Compound 1453 (BVDV)					HCV-specific compound					DMSO
	1 (109)	2 (107.9)	3 (111.6)	4 (104.6)	5 (103)	6 (103.9)	7 (98.3)	8 (100.2)	9 (102.5)	10 (99.5)	11 (100.9)	12 (99.1)
A	28,619	35,503	34,325	33,188	33,343	33,562	29,284	33,499	35,040	34,812	32,025	28,673
B	25,647	28,673	26,466	27,080	27,412	25,635	27,437	28,022	28,673	26,157	32,297	27,271
C	33,499	28,821	31,876	24,519	26,329	24,983	24,796	24,784	23,895	25,077	25,635	25,077
D	40,749	39,973	40,920	32,086	30,617	36,756	32,282	29,284	26,739	26,157	27,878	27,852
E	28,686	26,827	29,615	27,969	28,552	25,987	26,243	27,424	25,975	26,011	26,218	27,335
F	28,009	28,526	28,048	28,420	27,412	29,120	25,183	28,009	27,017	26,689	28,048	28,970
G	30,703	30,703	34,357	35,040	31,505	31,936	32,632	28,539	35,453	32,831	30,602	33,327
H	42,782	35,420	37,486	38,411	38,267	37,016	33,831	36,756	38,773	36,894	35,321	35,040

^aCompounds were titrated from top to bottom in five columns each, with the highest concentrations in row A. In columns 11 and 12, DMSO only was added to the wells and represents 100% control. Alamar blue numbers for each well are shown and used to quantify cell viability. Average cell viability values are shown in parentheses for each column, indicating no detectable toxicity for either compound.

TABLE 2.

Relative HCV FRET rates^a

Row	FRET rate in well containing:
	Compound 1453 (BVDV)					HCV-specific compound					DMSO
	1	2	3	4	5	6	7	8	9	10	11	12
A	118.7	119.4	118.7	110.5	108.5	0.8	0.2	−0.8	0.5	−0.7	95.3	84.8
B	111.0	95.2	127.4	115.0	111.6	1.2	2.3	0.0	0.9	0.8	95.9	92.5
C	131.7	118.9	127.5	125.1	118.2	2.1	2.3	2.7	4.0	5.9	109.4	95.7
D	129.4	128.7	121.7	124.4	125.1	2.8	1.9	1.8	1.0	0.6	98.8	99.6
E	115.2	125.0	109.3	121.6	124.4	18.1	18.7	21.1	17.9	16.6	106.8	97.1
F	112.3	116.6	110.4	115.3	108.6	20.8	19.1	18.4	20.6	23.1	97.9	97.8
G	105.8	107.2	81.8	110.7	114.5	87.8	80.1	84.2	88.6	106.2	106.7	103.9
H	109.0	128.6	115.1	116.7	112.3	89.7	73.7	87.2	79.8	84.4	111.0	106.9

^aCompound 1453, the BVDV specific compound (titrated from 10 to 0.3 μM), had no effect on the HCV rates, while the HCV-specific compound (titrated from 0.25 to 0.005 μM) greatly decreased the NS3 protease activity.

TABLE 3.

Relative BVDV luciferase values^a

Row	Luciferase value in well containing:
	Compound 1453 (BVDV)					HCV-specific compound					DMSO
	1	2	3	4	5	6	7	8	9	10	11	12
A	0.2	−0.3	−0.3	−0.3	0.7	112.0	74.3	118.0	100.9	126.2	116.1	115.7
B	−0.1	−0.2	0.1	−0.3	0.5	125.9	104.4	106.0	97.4	78.1	91.7	95.8
C	2.2	1.7	2.1	1.7	1.0	100.3	115.2	103.3	99.6	98.1	89.1	90.9
D	2.7	2.1	2.6	2.1	1.7	104.5	109.8	112.0	96.8	107.1	94.4	95.7
E	29.6	35.2	35.9	45.9	35.4	107.3	106.8	92.4	99.7	89.9	99.1	113.8
F	30.9	32.7	37.5	35.7	36.0	103.8	89.4	96.7	88.7	109.4	105.9	107.8
G	82.3	77.7	87.8	70.1	72.2	84.9	75.6	95.5	101.0	90.9	90.3	111.0
H	91.1	61.9	73.1	65.0	69.6	91.4	89.5	96.3	82.9	78.7	95.8	86.9

^aThe relative BVDV luciferase activity following FRET assay is shown. Compound 1453 greatly decreased the BVDV-specific luciferase activity, while the HCV inhibitor had no effect on the luciferase activity in the BVDV replicon cells.

DISCUSSION

In this report we present an HTS assay method combining an HCV replicon cell line and a BVDV luciferase replicon cell line. We have demonstrated a cell-based HCV-specific FRET protease assay which can quantitate HCV inhibition by using an enzyme required by HCV for replication. We have also incorporated a cytotoxicity assay which uses Alamar blue conversion as a general indication of cytotoxicity. Finally, a BVDV luciferase replicon cell line provides a third assay to measure specificity of the inhibition. The reagents are compatible with each other, requiring only a single wash step followed by addition of reagents and measurements in two machines. The assays allow one to easily discern nontoxic and HCV-specific inhibitory substances.

Development of assays.

The complete assay was developed in two parts, with the first part focused on assaying toxicity along with HCV inhibition. This part examined the use of the HCV FRET assay using the classical viral and HCV inhibitor IFN-α to optimize our methods and signals. The HCV assay 96-well format demonstrated results comparable to those of qRT-PCR over the linear range of the assay and was also equivalent to a reduction seen by Western blotting. As shown in Fig. 3, the assay was also compatible for measuring toxicity of compounds in the same well with the cytotoxicity assay reagent easily removed, with no deleterious effects on the cells. The variation associated with the FRET assay was also shown to be statistically acceptable, as indicated by the variance of the data in Fig. 5 and the Z′ calculation. This was important due to the low signal-to-noise ratio of the assay obtained by fluorescence endpoint analysis and is less of a concern with the high signal-to-noise ratio of the BVDV assay. The assay has a distinct advantage over qRT-PCR in that the assay takes place in situ in a detergent-based crude cell lysate which requires no further preparation prior to the assay. The assay does not involve numerous manipulations to add and subtract reagents after addition of test compounds and relies on a viral protein which is required by the HCV replicon for replication. Occasional variation, when observed, resulted in less HCV protease activity being measured rather than more, demonstrating a tendency toward false positives, which is acceptable for inhibitor screening. The second part of assay development involved mixing of the HCV replicon cells with the BVDV replicon cells. Huh-7 cells were used to select both HCV and BVDV replicon cell lines to provide the same cell background and minimize cell-type-specific effects during testing of compounds. The BVDV luciferase activity measurements provide a demonstration of the relative signal-to-noise ratio and indicate no appreciable interference from the Alamar blue and FRET assays.

HCV substrate.

The FRET protease substrate peptide is a crucial component of the assay which is resistant to cleavage by endogenous Huh-7 cellular proteases over the assay period but is efficiently recognized by the replicon-based HCV NS3 enzyme. An estimate of the enzyme kinetic parameters suggests close agreement for the K_m of the substrate, ∼5 μM, previously determined by in vitro methods (data not shown) (53). Given that the original purpose of the substrate was to monitor in vitro cleavage by using purified rather than crude enzyme, it is also known that the substrate can still be cleaved by the many different genotypes of HCV NS3 protease, providing greater utility for HCV replication assays, including genotype 1a replicons (5, 17) as well as full-length replicons (data not shown) (21). The assay may be amenable to an even greater degree of miniaturization such as a 384-well-based cell culture assay, which could provide greater savings on reagents as well as decrease the time required for screening. Disadvantages of using the FRET assay (and any other reporter assays) include fluorescent compounds which could mask the signal and interfere with measurements. This type of interference can be minimized for the HCV FRET assay by inspection of the curves from each plate in kinetic mode; a fluorescent compound would have a higher fluorescence starting point and, if inhibitory, would not yield an increase of fluorescence over time. The use of the rate of HCV FRET peptide cleavage rather than endpoint signals would also minimize this type of error, and this rate is easily extracted from the Cytofluor software.

Use of NS3 protease as a reporter enzyme.

In addition to the assay mentioned here, the HCV protease and FRET peptide combination might possibly be used as a reporter system. The FRET substrate is relatively resistant to Huh-7, HeLa, and HepG2 cellular proteases (data not shown), suggesting that it is very specific for HCV protease and may be resistant to cellular proteases in other cell types. Placement of the HCV NS3 protease in an expression system (mammalian or bacterial) or in the context of other viruses (13, 31, 33) may allow the FRET assay to provide a sensitive method for using the viral protein in a wider cell repertoire. This type of reporter system would be similar to the commercially available luciferase/beta-galactosidase systems and could measure protein production, promoter strength, cell viability, or other combinations. Adaptation of this method of assay may also be possible with other proteases, provided that a suitable and specific assay substrate can be synthesized. In summary, we describe a convenient three-step HTS assay for the HCV replicon which should enable the rapid identification of HCV-specific inhibitory molecules and maximize the selection of compounds which inhibit HCV through specific mechanisms.