Development of a Novel Dicistronic Reporter-Selectable Hepatitis C Virus Replicon Suitable for High-Throughput Inhibitor Screening

Abstract

Hepatitis C virus (HCV) research and drug discovery have been facilitated by the introduction of cell lines with self-replicating subgenomic HCV replicons. Early attempts to carry out robust, high-throughput screens (HTS) using HCV replicons have met with limited success. Specifically, selectable replicons have required laborious reverse transcription-PCR quantitation, and reporter replicons have generated low signal-to-noise ratios. In this study, we constructed a dicistronic single reporter (DSR)-selectable HCV replicon that contained a humanized Renilla luciferase (hRLuc) gene separated from the selectable Neo^r marker by a short peptide cleavage site. The mutations E1202G, T1280I, and S2197P were introduced to enhance replicative capability. A dicistronic dual-reporter HCV replicon cell line (DDR) was subsequently created by transfection of Huh-7 cells with the DSR replicon to monitor antiviral activity and by the introduction of the firefly luciferase (FLuc) reporter gene into the host cell genome to monitor cytotoxicity. The DDR cell line demonstrated low signal variation within the HTS format, with a calculated Z′ value of 0.8. A pilot HTS consisting of 20 96-well plates with a single concentration (10 μM) of 1,760 different compounds was executed. Hits were defined as compounds that reduced hRLuc and FLuc signals ≥50 and ≤40%, respectively, relative to those in a compound-free control. Good reproducibility was demonstrated, with a calculated confirmation rate of >75%. The development of a robust, high-throughput HCV replicon assay where the effects of inhibitors can be monitored for antiviral activity and cytotoxicity should greatly facilitate HCV drug discovery.

Approximately 170 million people globally test positive for hepatitis C virus (HCV) (9, 11). Infection by HCV results in a high degree of chronic hepatitis. In addition to inducing liver damage, a significant proportion of these infections also result in hepatocellular carcinoma. Although current treatments for hepatitis caused by HCV include interferon in combination with ribavirin (18), approximately 50 to 60% of individuals still are not able to resolve infection (15). Therefore, there is an unmet medical need to develop more effective therapies to treat HCV infection. Until 1999, all cell-based screening efforts for HCV drug discovery relied on surrogate viral systems, such as bovine viral diarrhea virus, and the potential development of assays where activities of specific viral targets could be monitored. In 1999, a significant breakthrough in studying HCV RNA replication occurred when the Bartenschlager laboratory developed the HCV replicon system, a tissue culture system that faithfully mimicked all of the RNA replication events of the HCV life cycle (21). This initiated a phase of intensified research into the mechanisms of HCV RNA translation, replication, and protein processing. It also ushered in a new era for HCV drug discovery, since it was now possible to test the effects of inhibitors of traditional targets, such as NS3 protease, helicase, and NS5B polymerase, in an authentic, in vitro HCV RNA replication system (1).

The original replicon system (21) was constructed by replacement of genes from the HCV genome that are not essential for HCV RNA replication, e.g., the structural genes, p7 and NS2, with a genetic cassette carrying an antibiotic resistance gene and the internal ribosomal entry site (IRES) from encephalomyocarditis virus (EMCV). This resulted in the formation of a dicistronic, selectable, subgenomic HCV replicon (2-4, 21) whose replication requires RNA elements in both nontranslated regions as well as the nonstructural proteins, including NS3 protease, helicase, and polymerase. Therefore, the HCV replicon system can be used for identifying inhibitors against all of these components (1).

Cell-based screening efforts in a high-throughput format to identify novel inhibitors and viral or host targets have recently been described (6, 23, 34). The first generations of selectable and reporter-selectable replicons have not been conducive for carrying out high-throughput screening (HTS) due to the labor-intensive quantitative reverse transcription-PCR methods used in screening and to low signal-to-noise ratios, respectively. To develop a robust screening tool, we developed several reagents derived from the available HCV selectable replicon (7). This work involved the introduction of a short, codon-optimized reporter gene separated from the selectable marker by the sequence for a cleavable peptide and a set of specific tissue culture adaptive mutations. The high signal-to-noise ratio obtained from a replicon with these modifications reduced variability during antiviral testing and enabled testing in an HTS format. The integration of a second reporter gene in the nuclei of the reporter-selectable HCV replicon cells provided a rapid and simple means of measuring cytotoxicity from the same well for which the antiviral activity determination was made. This resulted in the generation of a valuable reagent that can help in the identification of specific anti-HCV inhibitors.

MATERIALS AND METHODS

Inhibitors.

Alpha interferon (IFN-α) and 5,6-dichloro-1-beta-d-ribofuranosyl-benzimidazole (DRB) were obtained from Sigma (St. Louis, MO). Compounds AG-21541 and AG-21033 (22) are HCV NS5B polymerase inhibitors synthesized at Pfizer Global Research & Development (La Jolla, CA).

Cell lines and plasmids.

Huh-7 cell lines were obtained from Apath LLC (St. Louis, MO) and ReBLikon GmbH (Mainz, Germany). An ET cell line carrying a dicistronic firefly (Photinus pyralis) luciferase gene replicon was obtained from ReBLikon GmbH. The plasmid pBB7, which contains a cDNA copy of a dicistronic HCV genotype 1b (Con-1) selectable replicon, was obtained from Apath LLC. The plasmid phRL-CMV (Promega, Madison, WI) contains the gene for humanized Renilla reniformis luciferase (hRLuc).

Huh-7 and ET cells were propagated in Dulbecco's modified Eagle medium (DMEM; Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (HyClone, Logan, UT), 50 IU/ml of penicillin, and 50 μg/ml of streptomycin sulfate (Invitrogen) at 37°C and 5% CO₂. The culture medium for the ET cell line was further supplemented with 250 μg/ml of G418 (Cellgro, Herndon, VA).

Construction of plasmids. (i) pBB7M4hRLuc.

The plasmid pBB7M4 was created by introducing the substitutions I2204S (restoration of a wild-type amino acid codon which is modified in pBB7), E1202G, T1280I, and S2197P into pBB7 by QuikChange site-directed, PCR-based mutagenesis (Stratagene, La Jolla, CA). All mutations were confirmed by sequencing. To ensure that no secondary mutations were introduced during the PCRs, an SspBI-XhoI restriction fragment of the original pBB7 plasmid was replaced with the corresponding SspBI-XhoI fragment from pBB7M4, resulting in the final pBB7M4 plasmid.

To generate a fusion between the hRLuc gene and the sequence of the 2A protease of foot-and-mouth-disease virus (FMDV) (25), two PCRs were employed. In the first PCR, oligonucleotides AscI-hRLuc(+) (5′-CCA GGC GCG CCA TGG CTT CCA AGG TGT ACG ACC CCG AGC-3′) and FMDV2A-hRLuc(−) (5′-GAC TCG ACG TCT CCC GCA AGC TTA AGA AGG TCA AAA TTC AAC AGC TGC TGC TCG TTC TTC AGC ACG CGC TCC ACG-3′) were used to amplify the hRLuc gene from phRL-CMV. The resulting PCR product had an AscI site at the 5′ end and a partial FMDV 2A peptide sequence fused to the hRLuc gene at the 3′ end. In the second PCR, the product from the first PCR was used as the template, and oligonucleotides AscI- hRLuc(+) and AscI-G-FMDV2A(−) (5′-CCA GGC GCG CCC GGG CCC AGG GTT GGA CTC GAC GTC TCC CGC AAG CTT AAG AAG GTC AAA ATT C-3′) were used as primers. The final PCR product encoded an hRLuc-FMDV 2A peptide fusion and was flanked by AscI sites at the 5′ and 3′ ends. This was introduced into a shuttle vector, pcDNA3.1-HCVIRES-Neo^R, which contains the HCV IRES with 36 nucleotides of the core gene fused to the neomycin phosphotransferase gene (Neo^r). The resulting construct is referred to as pcDNA3.1-HCVIRES-hRLuc2A Neo^R (Fig. 1B). The AgeI-PmeI restriction fragment from this construct was used to replace the AgeI-PmeI fragment from pBB7M4, resulting in the construction of pBB7M4hRLuc (Fig. 1C) (10).

FIG. 1.

Construction of the reporter-selectable dicistronic HCV replicon BB7M4hRLuc. (A) The pBB7M4 replicon construct contains nontranslated regions at the 5′ (5′ NTR) and 3′ [poly(U) and U/C rich region (U-U/C) and 98 conserved nucleotides (nt)] ends. The HCV IRES (HI) drives the translation of the amino terminus of the core (36 nt) and Neo^r gene fusion, while the EMCV IRES (EI) drives the translation of the nonstructural (NS) proteins 3 to 5B. Three adaptive changes in NS3 and NS5A (E1202G, T1280I, and S2197P) are indicated by asterisks. (B) Transfer of a reporter-selectable gene cassette from a shuttle vector to the selectable replicon construct pBB7M4. The shuttle construct pcDNA3.1-HCVIRES-hRLuc2A Neo^R contains the entire HCV 5′ nontranslated region (solid line) followed by 36 nt of the core region, the hRLuc reporter gene, the FMDV 2A protease, and the Neo^r gene. This is followed by the vector sequence, as indicated by the dashed line. (C) The product of the transfer is the reporter-selectable HCV replicon construct, pBB7M4hRLuc.

(ii) pcDNA6FLuc.

The plasmid pcDNA6FLuc was constructed by amplifying the firefly luciferase (FLuc) gene from the pGL3 Basic vector (Promega), using a sense primer with a flanking NheI site (5′-GCT AGC ATG GAA GAC GCC AAA AAC ATA AAG AAA GGC CCG GCG CCA TTC TAT CC-3′) and an antisense primer with a flanking PmeI site (5′-GTT TAA ACT CAC ACG GCG ATC TTT CCG CCC TTC TTG GCC TTT ATG AGG-3′). The FLuc PCR product was cloned into the polylinker of pcDNA6 HisA (Invitrogen), using the restriction sites NheI and PmeI, by standard molecular biology techniques. This FLuc gene-containing pcDNA6 version was then used for stable cell line selection using the blasticidin resistance gene while inducing expression of the FLuc gene from the cytomegalovirus promoter in this plasmid.

In vitro RNA transcription and electroporation for generation of replicon cell lines.

BB7M4hRLuc replicon transcripts were prepared in vitro from ScaI-linearized pBB7M4hRLuc by using a T7 Megascript kit (Ambion, Austin, TX) following the manufacturer's protocol. Huh-7 cells were washed twice in phosphate-buffered saline without Ca²⁺ or Mg²⁺ (Invitrogen) and resuspended at 1.5 × 10⁷ cells/ml. Four hundred microliters of cells was mixed with 2 μg of BB7M4hRLuc in vitro transcripts, and electroporation was performed using a Bio-Rad (Hercules, CA) Genepulser unit set to 270 V and 950 μF. Immediately after being pulsed, the cells were collected in 10 ml complete DMEM, and 1, 3, or 5 ml was seeded in a 10-cm dish. After 24 h of incubation, the medium was replaced with 10 ml complete DMEM containing 250 μg/ml G418 and changed weekly until colonies of resistant cells were visible (2 to 3 weeks). Colonies were picked by trypsinization and seeded in 96-well plates with DMEM containing 250 μg/ml G418. Cells were transferred to wells with a greater surface area when confluence was reached, approximately every 2 to 4 days. Once the cells were in six-well plates, 2 × 10⁴ cells/well were seeded in 96-well black-walled plates (Corning, Corning, NY) and incubated for 72 h. Reporter gene expression in these cells was measured with a dual-luciferase reporter kit (Promega) following the manufacturer's instructions, using a 1450 Trilux MicroBeta Jet instrument (Perkin-Elmer, Wellesley, MA). Cell lines with acceptable activity (>1 × 10⁵ relative light units [RLUs] per well) were utilized in antiviral assays. BB7M4hRLuc clone 10 demonstrated the best signal-to-noise ratio and was expanded to generate the dicistronic single-reporter (DSR) replicon cell line (10). The signal-to-noise ratio was calculated by dividing the signal obtained from wells seeded with the replicon cells by the signal obtained from wells with luciferase substrates only.

Generation of DDR replicon cell line.

The pcDNA6FLuc construct (0.5 μg) was electroporated into the DSR replicon cell line as described above. Twenty-four hours after electroporation, the cells were subjected to selection with 250 μg/ml G418 and 6 μg/ml blasticidin (MP Biomedicals, Irvine, CA). After colony formation and expansion, hRLuc and FLuc expression was measured as described above. Clone B6b gave the best expression for both reporter genes and was expanded further to generate the dicistronic dual-reporter (DDR) replicon cell line (8).

Biochemical NS5B polymerase assay.

Recombinant HCV (genotype 1b, strain BK) (27) NS5B polymerase was tested for the ability to perform primer/template-directed transcription in assays that contained 30 mM Tris-HCl (pH 7.2), 10 mM MgCl₂, 20 mM NaCl, 1 mM dithiothreitol, 0.05% Tween 20, 1% glycerol, 5 pmol of biotin-dG₁₂ (primer), 0.5 pmol of poly(rC)₃₀₀ (template), 1 μM GTP, 0.1 to 0.3 μCi of [α-³²P]GTP, and 2.5 pmol (0.15 μg) of HCV polymerase protein in a final volume of 75 μl, as previously described (22). Reactions were initiated by the addition of enzyme and incubated for 30 min at 30°C. Reactions were stopped by the addition of 33 mM EDTA, and polynucleotide products were collected by filtration through DEAE Filtermat papers (Wallac, Wellesley, MA); unincorporated triphosphate was removed by washing the filters with 5% dibasic sodium phosphate. The filters were counted in a Tri-Lux Microbeta scintillation counter (Packard, Wellesley, MA). Compounds to be tested were added at various concentrations from stocks in 10% dimethyl sulfoxide (DMSO)-water (final DMSO concentration, 1% of the reaction mixture). Fifty percent inhibitory concentration (IC₅₀) values were estimated from the primary counts-per-minute (cpm) data (collected in triplicate) by using the formula cpm(I) = cpm(no inhibitor) × {1 − [I]/([I] + IC₅₀)}.

In vitro antiviral and cytotoxicity assays using hRLuc, FLuc, and XTT end points.

Antiviral and cytotoxicity assays were carried out in black-walled, clear-bottomed 96-well plates. Reporter replicon cells were seeded at a density of 2 × 10⁴/well in 100 μl DMEM without selection antibiotics. Eight threefold serial dilutions of compounds were prepared in DMEM and added to the appropriate wells, yielding final concentrations of 320 to 0.1 μM (or 10 to 0.003 IU/ml for IFN) in 200 μl of DMEM. Following 3 days of incubation, 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT; Sigma) and phenazine methosulfate (Sigma) were added to each well at final concentrations of 200 μg/ml and 1 mM, respectively. After 4 hours, the amount of formazan produced was quantified spectrophotometrically at a test reference wavelength of 450 nm and a reference wavelength of 650 nm (24, 31). The medium was removed from all wells, and FLuc and hRLuc activities were measured as described above. The percent inhibition was determined from Luc (FLuc and hRLuc) and XTT values, using the following formulas:

(1)

(2)

The percent inhibition was plotted against the concentration of compound, using XLFit (IDBS, Emeryville, CA), and the inhibitor concentrations giving 50% antiviral inhibition (EC₅₀) and 50% cytotoxicity (CC₅₀) were measured. The selectivity index (SI) was calculated by dividing the CC₅₀ by the EC₅₀.

In vitro antiviral assay using quantitative reverse transcriptase PCR (RT-PCR) (TaqMan).

Twenty thousand cells of the replicon cell lines were incubated with serial half-log dilutions of compounds in black 96-well plates. After incubation for 72 h, the medium was removed, and the cells were washed gently with 200 μl of phosphate-buffered saline. Total RNA was extracted using a QiaAmp RNA extraction kit (QIAGEN, Valencia, CA). Twenty microliters of the RNA was subjected to reverse transcription, using Multiscribe (ABI, Foster City, CA) according to the manufacturer's instructions, by using random hexamers to generate the cDNA. PCRs were carried out for HCV RNA, using the primers 5′-TCC CGG GAG AGC CAT AGT G-3 and 5′-GGC ATT GAG CGG GTT GAT C-3′ and the probe 6-carboxyfluorescein-CCG GAA TTG CCA GGA CGA CCG-BHQ1 according to the manufacturer's directions (ABI). PCRs were also carried out in parallel for an internal control, the mRNA of a housekeeping gene encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH), using the primers 5-GAA GGT GAA GGT CGG AGT C-3′ and 5′-GAA GAT GGT GAT GGG ATT TC-3′ and the probe 6-carboxyfluorescein-CAA GCT TCC CGT TCT CAG CC-6-carboxytetramethylrhodamine. HCV and GAPDH RNA quantifications were based on a standard curve run at the same time that the experimental samples were amplified. The signal for HCV RNA was normalized to that of GAPDH mRNA for each concentration of inhibitor, and percent inhibition was calculated based on comparison with the DMSO-treated wells. EC₅₀ values were calculated from these data, as described above.

To verify that the signal was due to self-replicating RNA and not to integration of nucleic acid into the genome, the TaqMan PCR was done in the presence and absence of reverse transcriptase.

Mock HTS run.

Z′ factor values for both FLuc and hRLuc end points for 10 96-well microtiter plates that were seeded with 2 × 10⁴ cells of the DDR replicon cell line/well were calculated using the following formula (32):

(3)

where “Max” is the maximum signal derived from wells containing DMSO (1%) alone, “Min” is the minimum signal for the antiviral (hRLuc) and cytotoxicity (FLuc) end points derived from wells that contained EC₉₅ and CC₉₅ levels of AG-21541 and DRB, respectively, σ is the standard deviation, and μ is the mean.

Pilot HTS screening using the DDR replicon cell line.

Twenty thousand DDR replicon cells per well were dispensed into 20 96-well plates, using a Multidrop384 system (ThermoLabsystems, Franklin, MA). For all plates, an antiviral control (10 μM of AG-21541) was added to well A12, and a cytotoxicity control (80 μM of DRB) was added to well H12. The remaining wells of column 12 were used as DMSO-only controls. One thousand seven hundred sixty compounds from 20 96-well Pfizer chemical archive plates were then added to plates at a final concentration of 10 μM. After 72 h of incubation, FLuc and hRLuc activities were measured, and percent antiviral inhibition and cytotoxicity were calculated as described above.

RESULTS

Construction of HCV dicistronic reporter-selectable replicon BB7M4hRLuc.

Initially, three adaptive mutations (E1202G, T1280I, and S2197P) (17) were introduced into a modified pBB7 (7) construct that contained the I2204S substitution to generate pBB7M4 (Fig. 1A). A shuttle vector (pcDNA3.1-HCVIRES-hRLuc2A Neo^R) was subsequently constructed that contained an hRLuc gene separated from the neomycin phosphotransferase gene that confers resistance to neomycin (Neo^r) by a cleavable peptide encoding the FMDV 2A protease (Fig. 1B). This cloned reporter gene-selectable marker cassette was used to replace a cassette containing the Neo^r gene only in pBB7M4 to generate pBB7M4hRLuc (10). In transcripts derived from pBB7M4hRLuc, the HCV and EMCV IRESs (HI and EI) are responsible for the translation of the hRLuc-Neo^R cassette and the nonstructural proteins from the NS3-to-NS5B region of the HCV open reading frame, respectively.

Characterization of the DSR replicon cell line.

In vitro transcripts from pBB7M4hRLuc were transfected into Huh-7 cells. Approximately 60 G418-resistant colonies were tested for hRLuc expression, and a clone yielding the highest reporter activity was subsequently expanded to create the DSR replicon cell line (Fig. 2). The hRLuc reporter signal from the DSR cell line was evaluated in parallel with a reporter signal from the ET dicistronic replicon cell line, which contains the FLuc gene. The results indicated a 70-fold higher signal for the DSR cell line than for the ET cell line (Fig. 2A). This translated to a signal-to-noise ratio of 3,500 for the DSR line, compared to 50 for the ET cell line (Fig. 2B). Confirmation of the presence of self-replicating replicon RNA in DSR cells was obtained by real-time RNA quantitation by a TaqMan assay. The signal obtained was approximately 10,000-fold higher when an RT step was included before PCR amplification compared to when no RT step was included (data not shown). This indicated that the hRLuc signal generated from the DSR cell line was due to autonomously replicating replicon RNA and not to integration of replicon cDNA into the cellular genome. Further validation of authentic HCV RNA replication was indicated by the susceptibility of the DSR cell line to IFN, with EC₅₀ values of 0.4 and 0.7 IU/ml (data not shown).

FIG. 2.

Characterization of DSR replicon cell line. Transcripts from pBB7M4hRLuc were transfected into Huh-7 cells to generate the DSR replicon cell line, as described in Materials and Methods. (A) Results are expressed as mean RLUs ± standard deviations for 12 wells of a 96-well plate containing either DSR or ET replicon cells. (B) Signal-to-noise ratios (S:N) for DSR and ET replicon cell lines were derived from the luciferase signals shown in panel A as described in Materials and Methods.

Characterization of the DDR replicon cell line.

To develop a replicon cell line that could allow for antiviral activity and cytotoxicity determinations for inhibitors from the same well, an additional reporter gene, FLuc, was introduced into DSR cells. Expression levels of both hRLuc and FLuc were subsequently measured for several clones (Fig. 3). Clone B6b demonstrated strong and comparable signal strengths for both hRLuc and FLuc that were stable over at least 50 generations (data not shown) and was subsequently selected for further expansion to yield the DDR replicon cell line.

FIG. 3.

Characterization of DDR replicon cell clones. To identify a DDR replicon cell line with optimal hRLuc and FLuc signals, several clones selected with G418 and blasticidin were screened as described in Materials and Methods. Expression levels for both hRLuc and FLuc are shown for seven representative clones and are expressed as mean RLUs ± standard deviations for three wells of a 96-well plate.

To correlate the hRLuc readout with levels of HCV replicon RNA, both the DSR and DDR replicon cell lines were treated with AG-21541, an HCV NS5B polymerase inhibitor (Fig. 4), and levels of hRLuc and HCV RNA were measured. The EC₅₀ values for AG-21541 for the DSR replicon cell line were 3.3 ± 0.43 μM and 2.4 ± 1.3 μM, using hRLuc and RNA end points, respectively. These values are similar to those obtained for AG-21541 with the DDR replicon cell line, where EC₅₀ values of 3.2 ± 1.7 μM and 3.2 ± 1.5 μM were generated from the hRLuc and RNA end points, respectively. The similar EC₅₀ values for the two end points indicate that inhibition at the HCV RNA level was suitably represented by the hRLuc reporter gene level.

FIG. 4.

Chemical structures of small-molecule inhibitors used for validation.

To further evaluate the utility of the hRLuc and FLuc end points to monitor antiviral activity and cytotoxicity, respectively, the DDR cell line was treated with DRB, a cytotoxic nucleoside, with IFN, and with two HCV NS5B polymerase inhibitors (Table 1; Fig. 4) The CC₅₀ values for DRB and AG-21033 using the FLuc end point were comparable to that obtained by the dye reduction methodology using XTT (24, 29, 31) (Table 1). Similarly, no cytotoxicity up to the highest concentration tested was observed for IFN and AG-21541, using either end point. Although potential antiviral activity was also demonstrated for all four compounds (EC₅₀ values ranged from 0.4 to 125 μM), only AG-21541 and IFN demonstrated SIs of >10. The EC₅₀ values determined for the hRLuc end point for this set of treatments were also comparable for both the DDR and DSR replicon cell lines (data not shown).

TABLE 1.

Evaluation of biochemical and antiviral activities of control inhibitors^a

Inhibitor (concn units)	IC50	EC50	XTT CC50	FLuc CC50	SI
DRB (μM)	ND	34.1, 53.7	48.7, 40.4	85.1, 60.7	0.7-2.4
IFN (IU/ml)	ND	0.4, 0.4	>10, >10	>10, >10	>25
AG-21541 (μM)	0.034, 0.014	1.7, 1.5	>320, >320	>320, >320	>188.2->213.3
AG-21033 (μM)	1.4, 0.92	115, 125	245, 242	153, 116	0.9-2.1

^aIC₅₀ values were determined using recombinant HCV NS5B polymerase in a primer/template-directed transcription assay as described in Materials and Methods. EC₅₀s and CC₅₀s were determined with the DDR replicon cell line, using the dual-reporter end points in conjunction with XTT as described in Materials and Methods. Results indicate individual values for two experiments for IC₅₀, EC₅₀, and CC₅₀ values and a range for SIs. ND, not determined.

Characterization of the DDR replicon cell line in an HTS format.

To evaluate the signal variability for both the FLuc and hRLuc end points, six 96-well plates containing DDR replicon cells were plated in the presence of 1% DMSO. The mean signals for both hRLuc and FLuc evaluated 72 h later demonstrated low variability, with a percent coefficient of variance (%CV) of <11 for each plate (Table 2). In a subsequent experiment, 10 96-well plates containing DDR replicon cells were incubated for 72 h, and Z′ values were calculated as described in Materials and Methods. A mean Z′ value of 0.8 (range, 0.7 to 0.8) was calculated for both the FLuc and hRLuc end points, indicating low variation over a large dynamic range (data not shown).

TABLE 2.

Evaluation of the intraplate signal variability of the DDR replicon cell line^a

Plate no.	FLuc signal		hRLuc signal
	Mean RLUs (± SD)	% CV	Mean RLUs (± SD)	% CV
1	325,704 ± 25,052	7.7	479,932 ± 51,541	10.7
2	337,379 ± 22,531	6.7	466,694 ± 46,316	9.9
3	330,355 ± 24,699	7.5	463,802 ± 44,867	9.7
4	325,465 ± 24,637	7.6	451,278 ± 40,791	9
5	340,255 ± 25,474	7.5	485,085 ± 40,700	8.4
6	327,393 ± 28,609	8.7	460,936 ± 48,463	10.5

^aDDR replicon cells were plated in the presence of 1% DMSO. Data are expressed as mean RLUs ± standard deviations for all wells for each 96-well plate.

To test whether the DDR replicon cell line could reproducibly identify HCV inhibitors at different positions in an HTS format, 10 96-well plates containing the DDR replicon cells were incubated with the EC₅₀ concentration (1.7 μM) of AG-21541 in five different well positions (Fig. 5). The results demonstrated comparable antiviral activities (hRLuc inhibition) for all five wells across the 10 plates. In addition, no cytotoxicity (FLuc inhibition) was observed, consistent with previous data (Table 1).

FIG. 5.

Evaluation of the capability of the DDR replicon cell line to identify hits in an HTS format. Ten 96-well plates containing DDR replicon cells were incubated with 1.7 μM (EC₅₀ value) of AG-21541 at five different positions. Data are expressed as mean percentages of inhibition ± standard deviations for both FLuc and hRLuc.

Pilot HTS run with DDR replicon cell line.

To further test the feasibility of using the DDR replicon cell line to identify HCV compounds with specific antiviral activity, we tested 1,760 compounds from the Pfizer compound archive at a concentration of 10 μM in an HTS format using 20 96-well plates. Hit criteria of antiviral inhibition of 50% and 60% were utilized to identify compounds with approximate 50% antiviral inhibition (EC₅₀ values) at or below 10 μM. To differentiate antiviral activity from cytotoxicity while maximizing the number of hits that could be followed up in secondary assays (e.g., dose-response assays, mechanism-of-action studies, etc.), a cytotoxicity threshold of ≤40% was also utilized. Using these criteria, hit rates of 4.8% and 4.0% were obtained when criteria for antiviral activity of ≥50% with cytotoxicity of ≤40% and antiviral activity of ≥60% with cytotoxicity of ≤40% were utilized, respectively. To determine the reproducibility of hit identification, the 20 plates were rescreened. Confirmation rates of 78.8% and 75.4% were measured using criteria of antiviral activity of ≥50% with cytotoxicity of ≤40% and antiviral activity of ≥60% with cytotoxicity of ≤40%, respectively. Although hits from the pilot screen were not subsequently followed up, we utilized the same cutoffs (antiviral activity of ≥60% with cytotoxicity of ≤40%) and initiated an HTS campaign utilizing the full Pfizer chemical file. In this format, we obtained a hit rate of 2.8% and a confirmation rate of 40% (data not shown). To date, we have identified >2,000 antiviral compounds, with EC₅₀ values ranging from <0.01 μM to 9.9 μM and SIs ranging from >10 to >18,900 (data not shown). These data demonstrate the utility of the DDR replicon cell line in an HTS format for identifying compounds with robust antiviral activity.

DISCUSSION

Historically, attempts to propagate infectious HCV in tissue culture have met with limited success. Although recent success has been achieved using a genotype 2 HCV molecular clone (20, 28, 33) and, to some extent, for a genotype 1 clone (30), a highly robust infectious cell culture system for HCV genotype 1 still remains elusive. The subgenomic HCV replicon system still remains the best alternative for studying RNA replication and discovering inhibitors targeting HCV genotype 1, the most prevalent genotype worldwide (12, 26). The utility of this approach was recently exemplified by the design, screening, and development of HCV NS3 protease inhibitors, resulting in the demonstration of efficacy in human clinical trials (19).

While the development of the HCV replicon has been considered a major breakthrough, the replicon system did not become a truly useful and robust tool until cell culture adaptive mutations which allowed high-level replication were identified (7, 13, 17). In this study, we utilized the early-generation selectable replicons to develop a reporter-selectable replicon capable of producing a high signal-to-noise ratio, which is important for rapid screening of HCV inhibitors. The cell line from such a construct was modified further to include a second reporter gene to monitor cytotoxicity in the same well.

The HCV replicon construct described in the current study utilizes a short, codon-optimized reporter gene for human cells, i.e., the hRLuc gene, and a short, 20-amino-acid, self-cleavable peptide of the FMDV 2A protease that results in the cleavage of the hRLuc protein from the Neo^r protein. The signal-to-noise ratio from the DSR replicon cell line was significantly higher than that obtained for the FLuc-containing replicon cell line ET or for other published reporter replicons, including a β-lactamase-containing replicon (34) and a fluorescence resonance energy transfer-based replicon (23). The increase in signal of the DSR replicon may be due to a higher level of replication conferred by the use of a shorter reporter gene (hRLuc gene) as well as the inclusion of three adaptive mutations. Consistent with this proposal, a non-Con-1 genotype 1b replicon that contained a fusion between a nonhumanized RLuc gene of similar size and the Neo^r gene was capable of replication and formation of stable cell lines (16). In contrast, a viable replicon was not obtained when the longer FLuc gene was included in place of the RLuc gene. In addition, Krieger et al. (17) demonstrated that the adaptive mutations E1202G, T1280I, and S2197P constructed in a Con-1 replicon conferred a significant replication advantage compared to the replicon that contained the wild-type sequence. Although we did not evaluate the effect of adaptive mutations in a replicon that contained the longer FLuc gene, it is likely that the combination of a shorter, humanized (codon-optimized) reporter (hRLuc) and highly adaptive mutations provides a higher signal.

The creation of the DDR replicon cell line, which incorporates a reporter gene in both the HCV replicon and the host cell nucleus, allows for the determination of antiviral activity and cytotoxicity in parallel. The combination of two reporter genes in the DDR replicon cell line also takes advantage of Promega's dual-luciferase kit, which enables reading of FLuc and RLuc from the same well using the same machine. The ability to combine both measurements in the same well allows for a more rapid determination of a selectivity window (antiviral activity/cytotoxicity) than that for systems in which compound treatments are separated to obtain antiviral activity and cytotoxicity values. The utility of this system was demonstrated by the antiviral assays performed in this study. In the multidose assay format, although all five compounds tested demonstrated antiviral activity, only IFN and AG-21541 had activities that were clearly differentiated from cytotoxicity (selectivity indices of >25 and >188, respectively). Moreover, using the DDR replicon cell line in an HTS format, we demonstrated that up to 36% of potential antiviral compounds could be removed rapidly through detection of cytotoxicity, in contrast to the case when antiviral activity alone was monitored (data not shown). Furthermore, our system allows for the inclusion of an additional end point to measure cytotoxicity (XTT dye reduction methodology) without affecting the other two reporter end points (data not shown).

The feasibility of an HTS utilizing the DDR replicon cell line was further demonstrated. Typically, an intraplate %CV of 15 or less is considered suitable for most high-throughput runs (23, 34). In our study, the %CV utilizing either the hRLuc or FLuc end point did not exceed 11%. Comparable levels of variability have been obtained for other replicon-based HTSs, e.g., 13%CV for a β-lactamase replicon screen (34) and 12%CV for a fluorescence resonance energy transfer-Alamar blue replicon assay (23). The assay quality of HTSs is commonly measured by the Z′ factor, a value that takes into consideration the dynamic range of the assay as well as variation within the assay (32). Z′ values above 0.5 and that approach 1 indicate a robust assay with low variability (32). In our study, the Z′ value for the DDR replicon cell line was 0.8 utilizing both reporter end points.

As a pilot test, we evaluated 1,760 inhibitors from the Pfizer compound archive in an HTS format utilizing the DDR replicon cell line. Antiviral and cytotoxicity cutoffs were selected to maximize the number of hits with potential antiviral activity that could subsequently be followed up in secondary assays for precise EC₅₀, CC₅₀, and SI determinations as well as various mechanism-of-action studies. Using two different criteria, we obtained hit rates of 4.8% and 4%. Repeat testing of the same 1,760 inhibitors resulted in high confirmation rates, of 79% and 75%, for the two criteria employed. High confirmation rates should increase the confidence in the identification of hits that demonstrate antiviral activity while showing reduced levels of cytotoxicity. Although hits from the pilot screen were not subsequently followed up, we utilized the same cutoffs and initiated an HTS campaign utilizing the full Pfizer chemical file. In this format, we obtained a hit rate of 2.8% and a confirmation rate of 40% (data not shown). Differences in rates between the pilot and full-file screens could be due to the nature of well contents; while plates for the pilot screen were derived from a portion of the Pfizer file where the final compound products are not purified, plates from the full file contained purified well contents. More importantly, subsequent follow-up of hits from the full file screen has led to the identification of >2,000 compounds with robust antiviral activity which is well differentiated from cytotoxicity (SIs range from >10 to >18,900).

The advent of an era of high-throughput cell-based screening has been made possible by several advances in detection methodology, new reagent development, and instrumentation (14). We report here the development of an HCV subgenomic replicon system with a robust signal to monitor HCV replication. We further describe the development of a multiparametric HCV replicon assay where the effects of antiviral agents can be monitored by up to three end points. Evaluating three end points from the same well provides savings in time and reagents as well as consistency and higher quality (5) for both low- and high-throughput formats. Implementation of cytotoxicity and antiviral end points in parallel also allows for more rapid identification of specific antiviral inhibitors and the elimination of cytotoxic compounds at an early stage. Executing an HTS using an HCV replicon should not only provide a wealth of pharmacological information but also be able to broaden the target space to include novel HCV targets.