Novel Hepatitis C Virus Reporter Replicon Cell Lines Enable Efficient Antiviral Screening against Genotype 1a

Transcription

1 ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Aug. 2010, p Vol. 54, No /10/$12.00 doi: /aac Copyright 2010, American Society for Microbiology. All Rights Reserved. Novel Hepatitis C Virus Reporter Replicon Cell Lines Enable Efficient Antiviral Screening against Genotype 1a Margaret Robinson, Huiling Yang, Siu-Chi Sun, Betty Peng, Yang Tian, Nikos Pagratis, Andrew E. Greenstein, and William E. Delaney IV* Gilead Sciences, 333 Lakeside Drive, Foster City, California Received 26 February 2010/Returned for modification 26 March 2010/Accepted 23 May 2010 The hepatitis C virus (HCV) subgenomic replicon is the primary tool for evaluating the activity of anti-hcv compounds in drug discovery research. Despite the prevalence of HCV genotype 1a ( 70% of U.S. HCV patients), few genotype 1a reporter replicon cell lines have been described; this is presumably due to the low replication capacity of such constructs in available Huh-7 cells. In this report, we describe the selection of highly permissive Huh-7 cell lines that support robust replication of genotype 1a subgenomic replicons harboring luciferase reporter genes. These novel cell lines support the replication of multiple genotype 1a replicons (including the H77 and SF9 strains), are significantly more permissive to genotype 1a HCV replication than parental Huh7-Lunet cells, and maintain stable genotype 1a replication levels suitable for antiviral screening. We found that the sensitivity of genotype 1a luciferase replicons to known antivirals was highly consistent between individual genotype 1a clonal cell lines but could vary significantly between genotypes 1a and 1b. Sequencing of the nonstructural region of 12 stable replicon cell clones suggested that the enhanced permissivity is likely due to cellular component(s) in these new cell lines rather than the evolution of novel adaptive mutations in the replicons. These new reagents will enhance drug discovery efforts targeting genotype 1a and facilitate the profiling of compound activity among different HCV genotypes and subtypes. With 170 million people infected, hepatitis C virus (HCV) represents a significant and immediate global health burden (34). No vaccine is available, and the current standard of care of pegylated alpha interferon (IFN- )/ribavirin combination therapy is poorly tolerated, only partially effective, and contraindicated in many patient populations (10, 22). If untreated, HCV infection leads to an increased risk of liver cancer and/or cirrhosis (28). HCV exhibits considerable genetic diversity and can be divided into seven major genotypes (i.e., genotypes 1 to 7) and several subtypes (e.g., genotypes 1a, 1b, etc.) (27, 29, 30). Genotype 1a is the predominant genotype in North America (representing up to 70% of infections), while genotype 1b is predominant in Europe and Japan (7, 38). The HCV subgenomic replicon, the primary tool for discovering and characterizing inhibitors of HCV replication, is a self-replicating, bicistronic viral RNA (21). The first cistron typically encodes a reporter gene and/or a selectable marker, and expression is driven by the HCV internal ribosome entry site (IRES). The second cistron encodes the HCV nonstructural proteins, and expression is generally driven by the encephalomyocarditis virus (EMCV) IRES. The nonstructural proteins are translated as a polyprotein, proteolytically processed by the viral NS3/4A protease, and form a replication complex in which the NS5B RNA-dependent RNA polymerase replicates the entire replicon RNA (1, 26). By transfecting replicon RNA into a permissive cell line (such as Huh-7), clonal cell lines that stably replicate HCV replicons can be selected. * Corresponding author. Mailing address: 333 Lakeside Drive, Foster City, CA Phone: (650) Fax: (650) E- mail: William.Delaney@gilead.com. Published ahead of print on 1 June With respect to drug discovery, replicon cell lines enable the identification of small-molecule inhibitors, the selection of resistance mutations, and the characterization of an inhibitor s mechanism of action. Historically, replication levels were monitored by measuring HCV RNA (i.e., Northern blotting or reverse transcription [RT]-PCR), but this approach is labor intensive and, consequently, not ideal for high-throughput screening. The incorporation of reporter genes, such as secreted alkaline phosphatase (SEAP), beta-lactamase, or firefly luciferase, facilitated antiviral screening efforts by increasing assay throughput and improving assay statistics (e.g., signal-tonoise ratio and Z factor) (16, 24, 36). The humanized Renilla reniformis luciferase gene (hrluc) is being increasingly utilized in the HCV infectious virus system due to its enhanced signalto-noise ratio and decreased gene length (12, 15, 23). More recently, a humanized Gaussia princeps luciferase gene (hgluc) with even higher substrate turnover rates and smaller gene size than hrluc has been described (32). Despite these advances in reporter gene technology, few stable 1a luciferase replicon cell lines have been described, presumably due to the low replication capacity of such constructs in available Huh-7 cells. Drug discovery efforts involving genotype 1a, including direct potency comparisons of antiviral agents between genotypes 1a and 1b and phenotypic characterization of resistant-mutant 1a replicons, remain low in throughput in the absence of genotype 1a reporter replicon cell lines and cell lines highly permissive for genotype 1a replication. Two distinct strategies have been used to facilitate robust HCV replication in Huh-7 cells. In one approach, combinations of cell culture-adaptive mutations are engineered into the replicon genome prior to transfection (11, 16, 20). The second approach (5, 19, 24) selects an enriched population of Huh-7 cells that are highly permissive for HCV replication through 3099

2 3100 ROBINSON ET AL. ANTIMICROB. AGENTS CHEMOTHER. curing cell lines that stably replicate HCV replicons. For example, Huh7-Lunet and Huh7.5 cells are both derivatives of Huh-7 cells that exhibit markedly enhanced permissivity to replication of genotype 1b replicons (5, 21). Selection of cells based on enhanced permissivity to genotype 1a replication, however, has not been previously described. Here, we report the successful isolation of novel stable cell lines robustly replicating luciferase-encoding genotype 1a HCV replicons. To achieve this, it was necessary to first generate novel cell lines with enhanced permissivity to genotype 1a replication. MATERIALS AND METHODS Cell culture. Huh-luc and Huh7-Lunet cells were obtained from ReBLikon GmbH (Mainz, Germany) (5) All Huh7-Lunet-derived replicon cell lines (including those described below) were propagated in Dulbecco s modified Eagle s medium (DMEM) with GlutaMAX-I (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT), 1 unit/ml penicillin (Invitrogen), 1 g/ml streptomycin (Invitrogen), and 0.1 mm nonessential amino acids (Invitrogen). Replicon cell lines were selected and maintained in 0.5 mg/ml G418 (Geneticin; Invitrogen). Generation of stable replicon cell lines. Amounts of 10 g ofin vitro-transcribed RNA were transfected into Huh7-Lunet, 51C, or 57C cells (described below) by electroporation. Briefly, subconfluent cells were detached by trypsin treatment, collected by centrifugation, washed twice with ice-cold phosphatebuffered saline (PBS), and resuspended at 10 7 cells/ml in Opti-MEM (Invitrogen). Replicon RNA was added to 400 l of cell suspension in a Gene Pulser cuvette (0.4-cm gap). Cells were electroporated at 270 V and 960 F (Bio-Rad Gene Pulser system; Bio-Rad, Hercules, CA). Pulsed cells were incubated at room temperature for 10 min after electroporation and then were resuspended in 30 ml DMEM. Cells were plated into 100-mm-diameter dishes for G418 selection. Cell clones were isolated, expanded, and cryopreserved at early passage levels. To determine the efficiency of G418-resistant colony formation, transfected cells were plated at multiple densities. Twenty-four hours after plating, the medium was replaced with DMEM 10% FBS supplemented with 1.0 mg/ml G418 and refreshed twice per week. Three weeks later, colony plates were used for cell expansion or G418-resistant foci were fixed with 4% formaldehyde and stained with 1% crystal violet and 20% (vol/vol) methanol. Generation of cured cells. The cell lines 1a H77-23, 1a H77-51, 1a H77-57, 1a H77-65, and 1a H77-79 are Huh7-Lunet clones stably transfected with the genotype 1a ph/sg-neo(l I) replicon (Apath, Brooklyn, NY). To cure the cells of HCV RNA, they were cultured in the presence of IFN- (100 IU/ml), BILN (100 nm), and GS-9190 (100 nm) for 1 month. Cells were passaged in medium containing the three drugs once they reached 80% confluence twice a week for a total of 8 passages. Cured cell lines were expanded and cryopreserved at early passage levels, and no more than 4 additional passages occurred before subsequent transfection. Cured cells were designated 23C, 51C, 57C, 65C, and 79C. Two methods were used to confirm that the cured cells lacked detectable HCV replicons. First, cells were subjected to G418 selection (1,000 g/ml) for 3 weeks and determined to be fully sensitive to this cytotoxic agent. Second, the cells were found to lack any NS3 protease activity as determined by a sensitive fluorescent NS3/4A protease assay (31). Plasmids encoding luciferase reporter replicons. Plasmids ph/sg- GlucNeo(L I) and ph/sg-rlucneo (L I) were generated from plasmid ph/ SG-Neo(L I), which encodes a genotype 1a (H77 strain) subgenomic replicon (4) and was obtained from Apath. A novel hgluc-neomycin phosphotransferase fusion (hgluc-neo) gene was designed without the wild-type secretion signal and with a 4-amino-acid (GlyAlaGlyAla) linker between the hgluc and the neomycin genes. The hgluc-neo fusion gene was synthesized by Integrated DNA Technologies (Coralville, IA) and provided in the plasmid pidt-gluc-neo DNA. This plasmid was digested with AscI and AflII, and the excised fragment was ligated with T4 DNA ligase (Promega, Madison, WI) into ph/sg-neo(l I) digested with the same enzymes. The resulting plasmid, ph/sg-glucneo(l I), was sequenced to confirm the correct orientation and sequence of the hgluc-neo gene. The hrluc-neo gene was amplified from pf9 cytomegalovirus (CMV) hrluc-neo Flexi(R) (Promega) by PCR using Accuprime super mix I (Invitrogen) and a primer set of AflII hrluc Fwd and AscI Neo Rev. These two primers have the following sequences and introduce restriction sites for subsequent cloning: AflII hrluc, 5 GTC TTA AGT ACA ACC ATG GCT TCC AAG GTG 3 (AflII site underlined), and AscI Neo Rev, 5 GGC GCG CCT CAG AAG AAC TCG TCA AGA AG 3 (AscI site underlined). The hrluc-neo amplification product was subcloned into pcr2.1-topo (Invitrogen). The resulting plasmid was digested with AscI and AflII, and the excised fragment (hrluc-neo) was ligated with T4 DNA ligase (Promega) into ph/sg-neo(l I) digested with the same enzymes. The resulting vector, ph/sg-hrluc-neo(l I) was sequenced to confirm the correct orientation and sequence of the hrluc-neo fusion gene. Plasmid phcv-sf9-rlucneo(k Y I) was generated from phcv- SF9(K Y I), which encodes a genotype 1a (SF9 strain) subgenomic replicon and was obtained from Robert Lanford (Southwest Foundation for Biomedical Research, San Antonio, TX). The hrluc-neo fusion gene was amplified from pf9 CMV hrluc-neo Flexi(R) (Promega) by PCR using Accuprime super mix I (Invitrogen) and a primer set of BstBI hrluc Fwd and EcoRI Neo Rev. These primers have the following sequences and introduce restriction sites for subsequent cloning: BstBI hrluc Fwd, 5 GAC TTC GAA CAT GGC TTC CAA GGT GTA CGA C 3 (BstBI site underlined), and EcoRI Neo Rev, 5 CTG AAT TCC GGA CGC GTT CAG AAG AAC TCG TC3 (EcoRI site underlined). The hrluc-neo amplification product was subcloned into pcr2.1-topo (Invitrogen). The resulting plasmid was digested with BstBI and EcoRI, and the excised fragment (hrluc-neo) was ligated with T4 DNA ligase (Promega) into phcv-sf9(k Y I) cut with the same enzymes. The resulting vector, phcv- SF9-RlucNeo(K Y I), was sequenced to ensure the correct orientation and sequence of the hrluc-neo fusion gene. Plasmid pfk-rep PI-luc/5.1, which encodes a genotype 1b (Con1 strain) replicon and firefly luciferase reporter driven by the polio IRES, was obtained from ReBLikon GMBH (9). The plasmid pcon1/sg-hrlucneo was generated from the plasmid I389luc-ubi-neo/NS3-3 /ET, which encodes a genotype 1b (Con1 strain) subgenomic replicon and was obtained from ReBLikon GmbH. The hrluc-neo gene was PCR amplified from pf9 CMV hrluc-neo Flexi(R) by PCR using Accuprime super mix I and the primers AscI hrluc Fwd and NotI hrluc Rev. These two primers have the following sequences and carry restriction sites for subsequent cloning: AscI hrluc Fwd, 5 ACT GAC GGC GCG CCA TGG CTT CCA AGG TGT ACG 3 (AscI site underlined), and NotI hrluc Rev, 5 GTC AGT GCG GCC GCT CAG AAG AAC TCG TCA AGA 3 (NotI site underlined). The hrluc-neo amplification product was subcloned into pcr2.1- TOPO. The resulting plasmid was digested with AscI and NotI, and the excised fragment (hrluc-neo) was ligated using T4 DNA ligase into I389luc-ubi-neo/ NS3-3 /ET cut with the same enzymes. The resulting vector, pcon1/sg-hrlucneo, was sequenced to confirm the correct orientation and sequence of the hrluc-neo fusion gene. Plasmid plucneo2a was derived from pjfh1, a plasmid containing the fulllength genotype 2a (JFH-1 strain) genome (Toray, Inc., Japan), as follows: the HCV nonstructural genes, along with the plasmid backbone in pjfh1, were amplified and self-ligated to generate p2asg by PCR using a primer set of HCV2aCoreAfeIrev and HCVMlu2aNS3fw. HCV2aCoreAfeIrev has the sequence 5 TCTAGA AGCGCT tgggcg acggtt ggtgtt tctttt gg 3 (HCV sequence in lowercase) and encodes an AfeI site (underlined). HCVMlu2aNS3fw has the sequence 5 GAGCTT ACGCGT atggct cccatc actgct tatg 3 (HCV sequence in lowercase) and an MluI site (underlined). The AfeI and MluI sites were introduced at the 20th residue of core protein and upstream of NS3, respectively, to allow the insertion of the luciferase reporter and the neomycin phosphotransferase gene. The fragment encoding the reporter, the Neo gene, and the EMCV IRES was amplified by PCR from the pfki389lucubineons3-3 _ET replicon using primers with AfeI and MluI sites at the ends and subsequently cloned into a pta TOPO vector. An XbaI site was knocked out of the luciferase/neo/ires fragment by site-directed mutagenesis so that the final replicon construct would have only one XbaI site. The plasmid plucneo2a was then generated by ligation of the cloned AfeI-MluI fragment containing luciferase/neo/ires into the plasmid p2asg after digestion with AfeI and MluI. The sequence was confirmed by DNA sequencing. RNA transcription. Plasmid DNAs containing genotype 1a subgenomic HCV replicon sequences were linearized with either HpaI (H77 replicons), XbaI (SF9 replicons), or AseI and ScaI (genotype 1b replicons) and purified using a PCR purification kit (Qiagen, Valencia, CA). RNA was synthesized with T7 MEGA- Script reagents (Ambion, Austin, TX) following the manufacturer s suggested protocol, and reactions were stopped by digestion with RNase-free DNase. RNA was purified using an RNeasy kit (Qiagen) in accordance with the manufacturer s protocol. RNA concentrations were determined by measurement of the optical density at 260 nm, and RNA integrity was verified by 0.8% agarose gel electrophoresis and ethidium bromide staining. Extraction, amplification, and genotypic analysis of HCV RNA. HCV RNA isolation, RT-PCR, and sequencing were performed by Tacgen (Hayward, CA). RNA was purified using an RNeasy kit (Qiagen) in accordance with the manufacturer s protocol. RT-PCR was performed using the SuperScript III first-strand

3 VOL. 54, 2010 REPLICATION OF HCV 1a REPLICONS IN NOVEL CELL LINES 3101 FIG. 1. Design of novel 1a reporter replicons. HCV replicons used to generate novel, stable genotype 1a replicon cell lines. (A) A neomycin phosphotransferase (Neo)-encoding H77 1a strain replicon. (B) An H77 1a strain replicon encoding a Gaussia luciferase (Gluc)-Neo fusion reporter. (C) An H77 1a strain replicon encoding a Renilla luciferase (Rluc)-Neo fusion reporter. (D) An SF9 1a strain replicon encoding an Rluc-Neo fusion reporter. A black segment indicates an HCV core sequence. A spotted segment indicates an SF9 sequence. L indicates the P1496L adaptive mutation. I indicates the S2204I adaptive mutation. K indicates the E1202K adaptive mutation. Y indicates the D1431Y adaptive mutation. The 5 and 3 nontranslated regions (NTR) and encephalomyocarditis virus (EMCV) IRES segments are indicated. synthesis system (Invitrogen). DNA samples were sequenced using ABI BigDye version 3.1 chemistry on an ABI PRISM 3700 DNA analyzer. Antiviral compounds. VX-950, BILN-2061, a Bristol-Myers Squibb monomeric NS5A inhibitor (BMS-5Am), and 2 -C-methyl adenosine (2 CMA) were purchased from Acme Bioscience (Belmont, CA). IFN- was purchased from Sigma-Aldrich (St. Louis, MO). The Wyeth HCV NS5B site IV inhibitor HCV- 796 was synthesized by Curragh Chemistries (Cleveland, OH). An Abbott benzothiadiazine NS5B polymerase inhibitor (A ) was synthesized by ChemALong Laboratories (Lemont, IL). Antiviral assays. Replicon cells were seeded in 96-well plates at a density of 5,000 cells per well in 100 l of DMEM culture medium, excluding G418. Compounds were serially diluted in 100% dimethyl sulfoxide (DMSO) and added to cells at a 1:200 dilution, achieving a final concentration of 0.5% DMSO in a total volume of 200 l. In 96-well assays, 3-fold serial drug dilutions with 11 concentrations were used and the starting concentrations were as follows: 50 M (VX-950, 2 CMA, and A ), 5 M (BILN-2061 and HCV-796), and 50 units (IFN- ). Alternately, 2,000 cells/well were seeded in 384-well plates in 90 l of DMEM culture medium, excluding G418. Compounds were added to cells at a 1:225 dilution, achieving a final concentration of 0.44% in a total volume of 90.4 l. Three-fold serial drug dilutions with 10 concentrations were used, and starting concentrations were as follows: 44.4 M (A and BMS-5Am), 22.2 M(2 CMA), 8.88 M (VX-950), 2.2 M (BILN-2061), or 0.44 M (HCV- 796). Cell plates were incubated at 37 C for 3 days, after which culture medium was removed and cells were assayed for luciferase activity as markers for replicon levels. Luciferase expression was quantified using a commercial luciferase assay (Promega). Luciferase levels were converted into percentages relative to the levels in the untreated controls (defined as 100%), and data were fitted to the logistic dose response equation y a/[1 (x/b) c ] using XLFit4 software (IDBS, Emeryville, CA) (y is the amount of normalized luciferase signal, x is the drug concentration, a represents the curve s amplitude, b is the x value at its transition center [EC 50 ], and c is a parameter which defines its transition width). Z determination. To determine the Z factor, 8 wells of a 96-well plate were treated for 72 h with 0.5 M protease inhibitor BILN-2061 (positive control), and 8 wells were treated with 0.5% DMSO (negative control) in triplicate assays. The Z factor was calculated for the antiviral luciferase endpoint using the equation Z 1 [3( p n )/ p n ], where is the standard deviation of the positive (p) or negative (n) control and is the mean of the positive control or negative control (39). stable genotype 1a luciferase reporter replicon cell lines (Fig. 1). Adaptive mutations shown to enhance the replication of subgenomic H77 or SF9 replicons in tissue culture (E1202K, D1431Y, P1496L, and S2204I) were included in the nonstructural genes (37). Selection of cell lines highly permissive to genotype 1a HCV replication. We initially attempted to establish stable genotype 1a luciferase replicon cell lines in Huh7-Lunet cells. However, transfection of these cells with the luciferase replicons described above failed to generate G418 (Geneticin)-resistant colonies in multiple independent experiments. To overcome this, we adopted a strategy used by others to generate cell lines highly permissive to genotype 1b replication. Specifically, we sought to select new cell lines highly permissive to genotype 1a replication by curing cell lines harboring 1a replicons without reporter genes. We established stable genotype 1a replicons without reporter genes by transfecting Huh7-Lunet cells with 1a H77 neo RNA which encodes an H77 1a replicon carrying both the P1496L and S2204I adaptive mutation (25). Individual G418- resistant colonies were isolated and expanded, and replication levels were quantified by measuring NS3 protease activity (Fig. 2.) (35). Five clones exhibiting high levels of NS3 activity, as RESULTS Design and construction of HCV 1a replicon constructs. While a variety of reporter replicon constructs encoding genotype 1b and genotype 2a nonstructural proteins have been described (3, 14, 21), there are few examples of reporter replicons encoding genotype 1a nonstructural genes. We designed and constructed subgenomic replicons with either the hrluc or hgluc reporter gene fused to neomycin in the first cistron and genotype 1a nonstructural genes from the H77 or 1a SF9 strains (4, 8, 17) in the second cistron to enable the selection of FIG. 2. Replication levels in 11 unique clonal cell lines harboring a genotype 1a replicon without a luciferase reporter. After transfection with the 1a H77 neo replicon, individual G418-resistant colonies were isolated, expanded, and tested in an NS3 protease activity assay (as a marker for HCV replication levels). For comparative purposes, the genotype 1b luciferase replicon cell line Huh-Luc was also tested in the assay. RFU, relative fluorescence units.

4 3102 ROBINSON ET AL. ANTIMICROB. AGENTS CHEMOTHER. TABLE 1. Novel cured cell lines are more permissive to genotype 1a replicon than Huh7-Lunet cells Cell line Parental cell line Permissivity to colony formation a 23C 1a H C 1a H C 1a H C 1a H C 1a H77-79 Huh7-Lunet Huh7 a Relative ability to form colonies after genotype 1a replicon transfection and selection. The number of symbols indicates the relative number of colonies formed. well as morphological and growth characteristics similar to those of the original parent cell line (Huh7-Lunet) were further expanded (1a H77-23, 1a H77-51, 1a H77-57, 1a H77-65, and 1a H77-79). These five 1a H77 replicon cell lines were then cured of the replicon by prolonged treatment with a combination of IFN-, the NS3 protease inhibitor BILN-2061, and the novel NS5B polymerase inhibitor GS The resulting cured cell lines, designated 23C, 51C, 57C, 65C, and 79C, were shown to lack detectable HCV replication by two independent methods: first, NS3 protease activity (data not shown) was at or below the level in naïve Huh7-Lunet cells and, second, each clone was fully sensitive to the cytotoxic effects of G418. To assess the ability of the cured cell lines to support genotype 1a replication, we transfected 1a H77 neo RNA into the five cured cell lines and the parental Huh7-Lunet cells. After 3 weeks of G418 selection, the number of G418-resistant colonies observed in the cured cells was significantly higher than that with Huh7-Lunet cells. Among the different 1a cured cell lines, permissivity, as measured by colony formation, ranged from approximately 2-fold (1a H77-23 and 1a H77-79) to 100- fold that observed with Huh7-Lunet cells (Table 1). Two cured replicon cell clones, 51C (Fig. 3A) and 57C, exhibited the highest degree of permissiveness to genotype 1a replication and were used in subsequent studies. To determine if the cured cell lines retained permissivity to replicons encoding other HCV genotypes, we transfected genotype 1a, 1b, and 2a luciferase-encoding replicons into 57C and Huh7-Lunet cells (Fig. 3B). Three days postelectroporation, the luciferase-labeled 1a replicon had replicated at much higher levels in 57C cells ( 30-fold) than in Huh7-Lunet cells. The genotype 1b replicon replicated at similar levels in 57C and Huh7-Lunet cells ( 3-fold greater levels in 57C). 57C cells were less permissive to genotype 2a replicons ( 10-fold) than Huh7-Lunet cells, although this replicon was readily quantifiable in both cell types. Overall, these transient transfection assay results indicate that 57C cells support robust replication of genotypes 1a, 1b, and 2a. However, in comparison to Huh7- Lunet cells, 57C cells were selectively more permissive to genotype 1a rather than globally more permissive to all three genotypes. Generation of cell lines stably replicating luciferase-encoding genotype 1a replicons. To establish stable 1a luciferase replicon cell lines, we electroporated in vitro-transcribed RNA for reporter and parental replicons (Fig. 1) into Huh7-Lunet, 51C, or 57C cells. Remarkably, all 51C and 57C cells transfected with the 1a H77 and 1a SF9 luciferase replicons yielded a significant number of G418-resistant colonies, while Huh7- Lunet cells yielded none (consistent with the results of our initial experiments, data not shown). To assess replication levels in these new genotype 1a stable replicon cell lines, luciferase activity was measured (Fig. 4). All stable 1a replicon luciferase cell lines showed robust luciferase expression, and luciferase signal-to-background ratios in excess of 100 were observed after miniaturization into the 96-well format (Table 2). Z factors, which measure assay quality (39), were calculated for each luciferase reporter replicon using the NS3 inhibitor BILN-2061 as a positive control or an equivalent volume of DMSO as a negative control (Table 2). Over three independent assays, both hrluc and hgluc 1a replicon cell FIG. 3. The cured cell lines are highly permissive to genotype 1a replication but also support replication of genotypes 1b and 2a. (A) RNA encoding the genotype 1a replicon ph/sg-neo(l I) was electroporated into the Huh7-Lunet (top left) or the 51C cured cell line (top right). After 3 weeks of selection with G418, resistant colonies, which harbor the genotype 1a replicon, were visualized with crystal violet. The 51C cell line (top right) gave rise to a significantly greater number of stable replicon clones than the Huh7-Lunet cell line (top left). Confirming that 51C cells are cured of the HCV replicon, electroporation without RNA (Mock) resulted in fully G418-sensitive cells for both cell lines (bottom). (B) Genotype 1a [ph/sg-glucneo(l I)], 1b (pfk-rep PI-luc/5.1), and 2a replicons (plucneo2a) were transfected into either 57C cured (white bars) or Lunet cells (black bars), and luciferase levels at 3 days posttransfection were measured. To control for transfection efficiency, data were normalized to luciferase levels at 4 hours posttransfection. For each genotype, replication in Lunet cells is displayed as a percentage of the replication in 57C cured cells. Error bars show standard deviations.

5 VOL. 54, 2010 REPLICATION OF HCV 1a REPLICONS IN NOVEL CELL LINES 3103 FIG. 4. Robust luciferase expression in 1a luciferase cell clones. Stable cell lines replicating either the 1a H77 Rluc-Neo, 1a H77 Gluc- Neo, or 1a SF9 Rluc-Neo replicon were assayed for luciferase activity. In the nomenclature of the cell lines, 51C and 57C refer to the cured cell line that the cells were derived from, H77 and SF9 refer to the strain of HCV the replicons were constructed with, Gluc and Rluc refer to the reporter gene (Gaussia luciferase or Renilla luciferase), and the terminal number indicates the clone number. Error bars show standard deviations. RLU, relative light units. lines produced Z factors of 0.7, supporting the potential use of these cells in high-throughput screening applications. Drug screening and antiviral activity. To validate these cell lines for antiviral activity assays, we assessed the response of 1a replicons expressing either hgluc or hrluc using a panel of six known HCV replication inhibitors in a 96-well assay format. Compounds were chosen to represent distinct mechanistic classes that target different nonstructural genes, such as NS3 protease inhibitors (BILN-2061 and VX-950) and nucleoside (2 C-methyl adenosine [2 CMA]) or nonnucleoside (HCV-796 and A ) NS5B polymerase inhibitors and host antiviral cytokines (IFN- ) (2, 6, 13, 33). Compounds were also selected to exhibit a broad range of antiviral potency; HCV-796, for example, was approximately 100-fold more potent than 2 CMA in previous replicon assays. Antiviral activity, as measured by 50% effective concentrations (EC 50 s), was consistent for each compound among the various 1a luciferase replicon cell lines (including four 1a H77 luciferase replicon cell lines and three 1a SF9 luciferase replicon cell lines) (Fig. 5). The potencies observed in this experiment for both 1a H77 and 1a SF9 replicons were consistent with EC 50 s we have previously generated in the 1a H77-57 nonreporter replicon cell line (data not shown). These results TABLE 2. Genotype 1a reporter replicons in cured cells produce robust luciferase signal Replicon (cell line) Luciferase signal-to-noise ratio a Z factor b Firefly luciferase 1b (Huh-luc) Renilla luciferase 1a (1a 57C SF9-Rluc3) Gaussia 1a (1a 51C-H77-Gluc1) a Signal-to-noise ratio was calculated by dividing the mean signal from 24 negative-control DMSO-treated wells by the mean signal from 24 positive-control BILN-2061-treated wells. b The Z factor was calculated using 8 positive (BILN-2061 treated) and 8 negative (DMSO treated) control wells from triplicate experiments. FIG. 5. Antiviral response is consistent across multiple genotype 1a replicon cell lines. Novel genotype 1a luciferase-encoding replicon cell lines were seeded in 96-well plates and treated with the indicated HCV inhibitors for three days. After drug treatment, the dose-response for each antiviral was quantified by measuring luciferase activity in duplicate, and EC 50 s were calculated by nonlinear regression. EC 50 s are plotted on the y axis, the antiviral compounds on the x axis, and the cell lines are indicated by the symbols showing the data points (E, 1A 51C-H77-Gluc1;, 1a 51C-H77-Gluc2; ƒ, 1a 51C-H77-Gluc3; F, 1a 57C-H77-RlucP;, 1a 57C-SF9-Rluc1;, 1a 57C-SF9-Rluc2; f, 1a 57C-SF9-Rluc3). Each data point represents the arithmetic mean of the results of two independent experiments. Horizontal lines represent the average EC 50 across all seven cell lines. confirm the phenotypic similarity of 1a replicon cell lines with luciferase reporters (hgluc or hrluc) from different isolates (H77 or SF9) in distinct cured cell lines (51C or 57C). We next compared the potency of compounds between genotype 1a and genotype 1b replicons in a 384-well assay format. In addition to BILN-2061, HCV-796, 2 CMA, and VX- 950, we also tested a monomeric NS5A inhibitor, as this class of compounds has been reported to have significant differences in potency between genotypes 1a and 1b (18). The EC 50 s for each compound against each genotype are summarized in Table 3. 2 CMA, VX-950, and HCV-796 had similar levels of potency against both genotypes ( 2.1-fold differences) (Table 3). The protease inhibitor, BILN-2061, was slightly less potent against the genotype 1a replicon ( 5.7-fold). However, the monomeric NS5A inhibitor was over 1,200-fold less potent against genotype 1a than against genotype 1b. Genotypic analysis of the new stable 1a replicons. In addition to the cellular background, adaptive mutations in the replicon sequence are known to contribute to increased replication efficiency of replicon RNA. To investigate whether adaptive mutations were responsible for the enhanced replication levels of our genotype 1a luciferase replicons, we sequenced the nonstructural regions of replicon RNAs extracted TABLE 3. EC 50 values of direct-acting HCV antivirals against genotype 1b and 1a replicons Compound Genotype 1b EC 50 (nm) against: Genotype 1a Fold difference 2 CMA BILN VX-950 1, , HCV BMS-5Am ,444 N/A 1,201

6 3104 ROBINSON ET AL. ANTIMICROB. AGENTS CHEMOTHER. TABLE 4. Mutations identified in stable genotype 1a replicon cell lines Cell line Parental cell line HCV strain from these cell lines and compared them to baseline replicon sequences. We first sequenced the coding region of the nonluciferase replicon RNA extracted from the five 1a H77 nonreporter cell lines (derived from Huh7-Lunet cells) to determine if any mutations arose during selection (Table 4). For each replicon cell line, we isolated total RNA and sequenced the entire replicon coding regions. Baseline adaptive mutations in the parental 1a H77 Neo replicon included the NS5A mutation, S2204I, and the NS3 mutation, P1496L. In all five of the 1a H77 replicon cell lines, these baseline adaptive mutations were maintained. However, additional HCV mutations were also observed in all five cell lines. One of these cell lines (1a H77-57) contained the previously reported NS4A adaptive mutation K1691R (37). The other clones contained mutations that have not been previously reported (S1289C, A1323S, M1646I, I1694T, and I2273T) but which were not consistently selected across multiple clones. A similar analysis was then performed on the luciferaseencoding replicon cell lines. As with the nonluciferase replicon cell lines, baseline adaptive mutations encoded in the parental H77 replicon were maintained in each clone after selection (Table 4). Similarly, adaptive mutations in the parental SF9 replicon (S2204I in NS5A and E1202K and D1431Y in NS3) were maintained in each clonal cell line after selection. In some but not all of the luciferase-encoding replicon cell lines, additional mutations were identified. The known adaptive mutation K1691R was identified in the 1a H77-57 cell line and also appeared in the 1a 51C-H77-Gluc2 clone which was derived from the H77 1a strain. The NS4B mutation E1726G was observed in two 1a H77 luciferase clones, and the NS5B mutation E2860G was identified in one H77 luciferase clone. However, the 1a H77 Rluc polyclonal pooled cell line did not show any consensus mutations beyond the parental adaptive mutations. Examination of the 1a SF9 luciferase replicons identified two mutations present in one clone each: I1475T in NS3 and L1715S in NS4B. The third SF9 replicon cell line did not contain any mutations. Overall, these analyses did not identify any specific mutations that were required for genotype 1a replication (beyond the parental adaptive mutations) in Huh7-Lunet, 51C, or 57C cell lines. Baseline adaptive mutations DISCUSSION Additional mutation(s) (HCV protein affected) 1a H77-23 Huh7-Lunet H77 P1496L, S2204I S1289C (NS3), I1694T (NS4A) 1a H77-51 Huh7-Lunet H77 P1496L, S2204I A1323S (NS3) 1a H77-57 Huh7-Lunet H77 P1496L, S2204I K1691R (NS4A) a 1a H77-65 Huh7-Lunet H77 P1496L, S2204I I1694T (NS4A) 1a H77-79 Huh7-Lunet H77 P1496L, S2204I M1646I (NS3), I2273T (NS5A) 1a 51C-H77-Gluc1 51C H77 P1496L, S2204I E1726G (NS4B) 1a 51C-H77-Gluc2 51C H77 P1496L, S2204I K1691R (NS4A) a 1a 51C-H77-Gluc9 51C H77 P1496L, S2204I E1726G (NS4B), E2860G (NS5B) 1a 57C-H77-RlucP 57C H77 P1496L, S2204I 1a 57C-SF9-Rluc1 57C SF9 E1202K, D1431Y, S2204I I1475T (NS3) 1a 57C-SF9-Rluc2 57C SF9 E1202K, D1431Y, S2204I 1a 57C-SF9-Rluc3 57C SF9 E1202K, D1431Y, S2204I L1715S (NS4B) a K1691R is a previously reported adaptive mutation (37). While the widely used Huh-7 cell line and, particularly, the derivative cell line cured of the genotype 1b replicon known as Huh7-Lunet support robust replication of genotype 1b reporter replicons, these cell lines do not appear to support the generation of stable genotype 1a reporter replicon cell lines. Indeed, we were unable to generate any genotype 1a stable replicon colonies when using replicon RNAs encoding a luciferase reporter gene. In contrast, but consistent with what others have reported, we were readily able to establish genotype 1a replicon cell lines using replicons without a luciferase reporter. To facilitate the generation of luciferase-encoding genotype 1a replicons, we cured cell lines stably replicating a genotype 1a replicon without a reporter gene, using a combination of HCV inhibitors. The resulting cured Huh7 cell lines were remarkably more permissive to genotype 1a replicons with or without reporter genes; in fact, two out of five cured cell lines had permissivity to nonreporter genotype 1a replicons that was enhanced more than 100-fold compared to the permissivity of Huh7-Lunet cells during colony formation assays. We also tested the ability of the 57C cured cell line to support the replication of genotype 1b and genotype 2a replicons. We found that this cell line supported robust replication of all three genotypes (1a, 1b, and 2a). However, in contrast to genotype 1a, which replicated 30-fold better in 57C than in Huh7-Lunet, genotype 1b showed little enhancement (approximately 3-fold), while genotype 2a replicated better in Huh7- Lunet cells. These results suggest the presence of cellular factors that are able to influence HCV replication in a genotype-specific manner, since permissivity was not globally enhanced for all genotypes. After establishing these 1a reporter replicons, we next validated their utility in antiviral assays. We found that under standard antiviral screening conditions (e.g., 3-day drug treatment of replicon cells in 96-well plates), replicon cell lines encoding either the Renilla or Gaussia luciferase had signalto-noise ratios of 100, with robust Z factors ( 0.7). Furthermore, multiple luciferase cell lines, including those based on the H77 and SF9 1a HCV strains, were demonstrated to have antiviral susceptibilities for multiple drug classes similar to

7 VOL. 54, 2010 REPLICATION OF HCV 1a REPLICONS IN NOVEL CELL LINES 3105 those observed in nonluciferase-containing cell lines. As genotype 1b and 2a luciferase replicons have been previously described (3, 14), these new genotype 1a replicons will allow side-by-side direct comparison of antiviral potency among the three genotypes in a high-throughput reporter format. Furthermore, these 1a replicons encode either Renilla or Gaussia luciferase, which utilize coelenterazine-based substrates, whereas typical genotype 1b or genotype 2b replicons utilize firefly luciferase (which utilizes luciferin-based substrates); this opens the possibility to conveniently multiplex these new 1a replicons with 1b or 2a replicons using commercially available luminescence assay kits designed for this purpose. With the compounds that we tested, we did not note any significant difference in antiviral sensitivity between the H77 and SF9 strains. However, these two 1a strains have a high level of overall homology (99.1% identity at the protein level). The NS3 protease domain and NS5B polymerase have particularly high homology, at 99.4% and 98.9% identity, respectively. NS5A and NS4B, which are emerging targets for drug development, have 98.4% and 97.3% homology, respectively. Overall, having two independent genotype 1a strains to validate inhibitor activity will be valuable for drug discovery efforts, especially given the natural heterogeneity of HCV and the significant impact polymorphisms have already been described to have on some HCV drug classes. We did, in contrast, confirm that a monomeric NS5A inhibitor had dramatically less potency against genotype 1a than against genotype 1b ( 1,200-fold less). Such potency information is valuable for assessing potential variability in the clinic but might also yield valuable insight into inhibitor mechanism of action or aid molecular target identification for new drug classes. Sequencing the 1a replicons stably transfected into Huh7- Lunet or 1a cured cell lines to explore the potential evolution of additional genotype 1a adaptive mutations did not reveal any high-frequency mutations in conjunction with the preexisting adaptive mutations. We observed the previously reported NS4A adaptive mutation K1691R and the nearby I1694T mutation twice each out of 12 sequenced 1a replicon cell lines. I1694, whose role in replication has not yet been characterized, is highly conserved among genotype 1a patient isolates. Positions 1715 (NS4B), 1289 (NS3), and 1646 (NS3) are also highly conserved among genotype 1a patient isolates, and mutations at these residues have an uncertain impact on HCV replication. It should also be noted that some clonal and polyclonal cell lines selected during these studies had no amino acid changes from the parental replicon sequences. Together, these observations suggest that factors specific to the cured cell lines, and not additional adaptive mutations in replicon sequences, are the greatest contributor to the enhanced replication phenotype of genotype 1a replicons in these cells. In summary, the new luciferase replicon cell lines described in this report will enable convenient high-throughput screening and characterization of antivirals against genotype 1a HCV. The approach of creating cell lines with enhanced permissivity was crucial to the generation of the new 1a reporter replicon cell lines and did not rely on identifying adaptive mutations or using preengineered adaptive mutations beyond those already commonly described. Given the high prevalence of chronic genotype 1a infection, we envision that both the cured cell lines and the luciferase-encoding 1a replicons described herein will facilitate efforts to develop new therapies for HCV. Indeed, we have recently taken advantage of the enhanced permissivity of 1a cured cell lines by using them to phenotype chimeric replicons encoding target genes from clinical isolates (data not shown). In this context, these cell lines are particularly valuable, since such chimeric replicons often have significantly reduced replication fitness compared to that of laboratory HCV strains. Finally, these cell lines could be used, along with other permissive cell lines (e.g., Huh7-Lunet and Huh-7.5), to study cellular factors and pathways that govern HCV replication in a pangenotype or genotype-/subtype-specific manner. ACKNOWLEDGMENTS We thank the following members of Gilead Sciences: Matthew Paulson for assistance with luciferase optimization, James Nugteren and Brian Stephens for compound management assistance, Johnny Lee for assistance with replicon assays, Victor Chen and Mark Kenney for informatics support, Hongmei Mo for helpful conversations, and Weidong Zhong for critical review of the manuscript. We also thank Helen Lee and Robert Lanford (Southwest Foundation for Biomedical Research) for helpful discussions. REFERENCES 1. Bartenschlager, R., and V. Lohmann Replication of hepatitis C virus. J. Gen. Virol. 81: Beaulieu, P. L Non-nucleoside inhibitors of the HCV NS5B polymerase: progress in the discovery and development of novel agents for the treatment of HCV infections. Curr. Opin. Invest. Drugs 8: Blight, K. J., A. A. Kolykhalov, and C. M. Rice Efficient initiation of HCV RNA replication in cell culture. Science 290: Blight, K. J., J. A. McKeating, J. Marcotrigiano, and C. M. Rice Efficient replication of hepatitis C virus genotype 1a RNAs in cell culture. J. Virol. 77: Blight, K. J., J. A. McKeating, and C. M. Rice Highly permissive cell lines for subgenomic and genomic hepatitis C virus RNA replication. J. Virol. 76: De Francesco, R., and A. Carfi Advances in the development of new therapeutic agents targeting the NS3-4A serine protease or the NS5B RNAdependent RNA polymerase of the hepatitis C virus. Adv. Drug Deliv. Rev. 59: Di Bisceglie, A. M Hepatitis C. Lancet 351: Feinstone, S. M., H. J. Alter, H. P. Dienes, Y. Shimizu, H. Popper, D. Blackmore, D. Sly, W. T. London, and R. H. Purcell Non-A, non-b hepatitis in chimpanzees and marmosets. J. Infect. Dis. 144: Friebe, P., V. Lohmann, N. Krieger, and R. Bartenschlager Sequences in the 5 nontranslated region of hepatitis C virus required for RNA replication. J. Virol. 75: Fried, M. W., M. L. Shiffman, K. R. Reddy, C. Smith, G. Marinos, F. L. Goncales, Jr., D. Haussinger, M. Diago, G. Carosi, D. Dhumeaux, A. Craxi, A. Lin, J. Hoffman, and J. Yu Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection. N. Engl. J. Med. 347: Guo, J. T., V. V. Bichko, and C. Seeger Effect of alpha interferon on the hepatitis C virus replicon. J. Virol. 75: Hao, W., K. J. Herlihy, N. J. Zhang, S. A. Fuhrman, C. Doan, A. K. Patick, and R. Duggal Development of a novel dicistronic reporter-selectable hepatitis C virus replicon suitable for high-throughput inhibitor screening. Antimicrob. Agents Chemother. 51: Huang, Z., M. G. Murray, and J. A. Secrist III Recent development of therapeutics for chronic HCV infection. Antiviral Res. 71: Kato, T., T. Date, M. Miyamoto, A. Furusaka, K. Tokushige, M. Mizokami, and T. Wakita Efficient replication of the genotype 2a hepatitis C virus subgenomic replicon. Gastroenterology 125: Kim, C. S., J. H. Jung, T. Wakita, S. K. Yoon, and S. K. Jang Monitoring the antiviral effect of alpha interferon on individual cells. J. Virol. 81: Krieger, N., V. Lohmann, and R. Bartenschlager Enhancement of hepatitis C virus RNA replication by cell culture-adaptive mutations. J. Virol. 75: Lanford, R. E., H. Lee, D. Chavez, B. Guerra, and K. M. Brasky Infectious cdna clone of the hepatitis C virus genotype 1 prototype sequence. J. Gen. Virol. 82: Lemm, J. A., D. O Boyle II, M. Liu, P. T. Nower, R. Colonno, M. S. Deshpande, L. B. Snyder, S. W. Martin, D. R. St. Laurent, M. H. Serrano-Wu, J. L. Romine, N. A. Meanwell, and M. Gao Identification of hepatitis C virus NS5A inhibitors. J. Virol. 84:

8 3106 ROBINSON ET AL. ANTIMICROB. AGENTS CHEMOTHER. 19. Lohmann, V., S. Hoffmann, U. Herian, F. Penin, and R. Bartenschlager Viral and cellular determinants of hepatitis C virus RNA replication in cell culture. J. Virol. 77: Lohmann, V., F. Korner, A. Dobierzewska, and R. Bartenschlager Mutations in hepatitis C virus RNAs conferring cell culture adaptation. J. Virol. 75: Lohmann, V., F. Korner, J. Koch, U. Herian, L. Theilmann, and R. Bartenschlager Replication of subgenomic hepatitis C virus RNAs in a hepatoma cell line. Science 285: Manns, M. P., J. G. McHutchison, S. C. Gordon, V. K. Rustgi, M. Shiffman, R. Reindollar, Z. D. Goodman, K. Koury, M. Ling, and J. K. Albrecht Peginterferon alfa-2b plus ribavirin compared with interferon alfa-2b plus ribavirin for initial treatment of chronic hepatitis C: a randomised trial. Lancet 358: Matthews, J. C., K. Hori, and M. J. Cormier Purification and properties of Renilla reniformis luciferase. Biochemistry 16: Murray, E. M., J. A. Grobler, E. J. Markel, M. F. Pagnoni, G. Paonessa, A. J. Simon, and O. A. Flores Persistent replication of hepatitis C virus replicons expressing the beta-lactamase reporter in subpopulations of highly permissive Huh7 cells. J. Virol. 77: Paulson, M. S., H. Yang, I. H. Shih, J. Y. Feng, E. M. Mabery, M. F. Robinson, W. Zhong, and W. E. T. Delaney Comparison of HCV NS3 protease and NS5B polymerase inhibitor activity in 1a, 1b and 2a replicons and 2a infectious virus. Antiviral Res. 83: Reed, K. E., and C. M. Rice Overview of hepatitis C virus genome structure, polyprotein processing, and protein properties. Curr. Top. Microbiol. Immunol. 242: Robertson, B., G. Myers, C. Howard, T. Brettin, J. Bukh, B. Gaschen, T. Gojobori, G. Maertens, M. Mizokami, O. Nainan, S. Netesov, K. Nishioka, T. Shin i, P. Simmonds, D. Smith, L. Stuyver, and A. Weiner Classification, nomenclature, and database development for hepatitis C virus (HCV) and related viruses: proposals for standardization. International Committee on Virus Taxonomy. Arch. Virol. 143: Rodger, A. J., S. Roberts, A. Lanigan, S. Bowden, T. Brown, and N. Crofts Assessment of long-term outcomes of community-acquired hepatitis C infection in a cohort with sera stored from 1971 to Hepatology 32: Simmonds, P Genetic diversity and evolution of hepatitis C virus 15 years on. J. Gen. Virol. 85: Simmonds, P., A. Alberti, H. J. Alter, F. Bonino, D. W. Bradley, C. Brechot, J. T. Brouwer, S. W. Chan, K. Chayama, D. S. Chen, et al A proposed system for the nomenclature of hepatitis C viral genotypes. Hepatology 19: Sudo, K., K. Yamaji, K. Kawamura, T. Nishijima, N. Kojima, K. Aibe, K. Shimotohno, and Y. Shimizu High-throughput screening of low molecular weight NS3-NS4A protease inhibitors using a fluorescence resonance energy transfer substrate. Antivir. Chem. Chemother. 16: Tannous, B. A., D. E. Kim, J. L. Fernandez, R. Weissleder, and X. O. Breakefield Codon-optimized Gaussia luciferase cdna for mammalian gene expression in culture and in vivo. Mol. Ther. 11: Tong, X., Z. Guo, J. Wright-Minogue, E. Xia, A. Prongay, V. Madison, P. Qiu, S. Venkatraman, F. Velazquez, F. G. Njoroge, and B. A. Malcolm Impact of naturally occurring variants of HCV protease on the binding of different classes of protease inhibitors. Biochemistry 45: Wasley, A., and M. J. Alter Epidemiology of hepatitis C: geographic differences and temporal trends. Semin. Liver Dis. 20: Yang, H., and W. E. Delaney IV A novel fluorescence-based protease assay using the endogenous NS3/4A protease activity present in the total cell lysates of HCV replicon cells. J. Clin. Virol. 36:S Yi, M., F. Bodola, and S. M. Lemon Subgenomic hepatitis C virus replicons inducing expression of a secreted enzymatic reporter protein. Virology 304: Yi, M., and S. M. Lemon Adaptive mutations producing efficient replication of genotype 1a hepatitis C virus RNA in normal Huh7 cells. J. Virol. 78: Zein, N. N Clinical significance of hepatitis C virus genotypes. Clin. Microbiol. Rev. 13: Zhang, J. H., T. D. Chung, and K. R. Oldenburg A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen. 4: Downloaded from on September 11, 2018 by guest