|Home | About | Journals | Submit | Contact Us | Français|
PF-05095808 is a novel biological agent for chronic hepatitis C virus (HCV) therapy. It comprises a recombinant adeno-associated virus (AAV) DNA vector packaged into an AAV serotype 8 capsid. The vector directs expression of three short hairpin RNAs (shRNAs) targeted to conserved regions of the HCV genome. These shRNAs are processed by the host cell into the small interfering RNAs which mediate sequence-specific cleavage of target regions. For small-molecule inhibitors the key screens needed to assess in vitro activity are well defined; we developed new assays to assess this RNA interference agent and so to understand its therapeutic potential. Following administration of PF-05095808 or corresponding synthetic shRNAs, sequence-specific antiviral activity was observed in HCV replicon and infectious virus systems. To quantify the numbers of shRNA molecules required for antiviral activity in vitro and potentially also in vivo, a universal quantitative PCR (qPCR) assay was developed. The number of shRNA molecules needed to drive antiviral activity proved to be independent of the vector delivery system used for PF-05095808 administration. The emergence of resistant variants at the target site of one shRNA was characterized. A novel RNA cleavage assay was developed to confirm the spectrum of activity of PF-05095808 against common HCV clinical isolates. In summary, our data both support antiviral activity consistent with an RNA interference mechanism and demonstrate the potential of PF-05095808 as a therapeutic agent for chronic HCV infection.
Hepatitis C virus (HCV) infection is a leading cause of liver disease worldwide. The World Health Organization estimates that around 170 million people are chronically infected, with 350,000 deaths from hepatitis C-related liver disease per year. Prevalence rates are significant in both developed and developing countries, with a recent report putting the prevalence among Chinese blood donors at 8.7% (11). The standard of care (SOC) for patients with HCV infection has until very recently been the combination of a pegylated interferon with the guanosine analogue ribavirin. Therapy is typically associated with extended dosing of 24 to 72 weeks' duration, low tolerability, and modest efficacy (17). The recent approval of protease inhibitors to supplement the SOC is anticipated to improve response rates and may shorten treatment duration. However, therapy remains an arduous and expensive undertaking, and there is a demonstrable need for improved therapeutics.
An alternative approach to HCV therapy is the use of small interfering RNAs (siRNAs) complementary to the HCV genome. There has been great interest in the therapeutic application of RNA interference (RNAi) since the first report that short RNAs can regulate and suppress complementary RNAs (9). There are ongoing clinical trials using RNAi agents in ophthalmology and oncology and also a recent report of clinical efficacy in respiratory syncytial virus infection (5, 30, 37). There is extensive literature on the activity of siRNAs in in vitro models of infectious disease (3, 13, 21, 33). Challenging viruses such as human immunodeficiency virus and hepatitis C virus with single siRNA agents has been associated with the risk of resistance emergence; a single nucleotide substitution can be sufficient to abrogate RNAi-mediated inhibition (3, 13, 41). However, delivery of multiple siRNAs has been reported to mitigate this risk (13). Consequently the potential of viral vector delivery systems that enable the expression of multiple siRNAs has been explored by many groups (16, 26–28, 32, 44). Because the HCV genome is comprised within a single molecule of RNA, a cleavage event by any one of the multiple, expressed short hairpin RNAs (shRNAs) will render the genome incapable of further translation events or packaging into nascent virions. A further benefit to the use of viral vectors is in clinical practice, where tissue-specific tropism of the delivery vector can be exploited. Adeno-associated virus serotype 8 (AAV-8) vectors have been reported to preferentially transduce liver hepatocytes to completion in animal models (4, 10). PF-05095808 has shown nearly complete transduction of liver hepatocytes in in vivo murine and nonhuman primate studies (D. Suhy et al., submitted for publication).
PF-05095808 (also known as TT-034) comprises a recombinant AAV vector, which directs expression of three shRNAs targeted to the HCV genome (Fig. 1A). All sequences coding for the AAV viral proteins have been removed, and it is therefore replication incompetent and nonintegrating. Since the AAV vector lacks the replication proteins required for genome integration, vector DNA is predominantly stabilized extrachromosomally (29). AAV-8-mediated transgene expression has been shown to persist in the murine livers for a minimum of 120 days (40, 43). It is therefore anticipated that following a single-dose administration of PF-05095808, sustained expression of shRNAs in hepatocytes will be achievable.
Large numbers of small-molecule inhibitors of HCV have emerged from pharmaceutical drug discovery efforts and are progressing through clinical trials. There is a growing understanding of how in vitro activity translates to in vivo efficacy. However, there are substantial challenges in understanding the efficacy profile associated with shRNAs delivered from virus vector systems. Additionally, viral vectors that infect efficiently in the clinical setting may have poor transduction capability in vitro, compounding the difficulties of generating a useful preclinical data package. We report here on building an in-depth understanding of the potency, spectrum, and resistance profile associated with RNA interference mechanisms in HCV model systems. Our studies demonstrate that the three shRNAs of PF-05095808 deliver independent antiviral activity and that both HCV replicon and infectious virus systems are sensitive to transduction by PF-05095808. The impact of the vector delivery vehicle on the observed potency of PF-05095808 is highlighted. The development of a quantitative PCR (qPCR) assay enabling direct measurement of the effector RNAi molecules expressed from PF-05095808 is described. Evidence for an authentic RNA interference mechanism of sequence-specific antiviral activity is obtained from both activity assays and resistance passaging studies. Finally, we demonstrate the spectrum of activity of PF-05095808 against clinical isolates of HCV representative of commonly circulating strains.
The AAV-free helper virus system (Stratagene) was used for construction of the AAV-derived vectors, termed PF-05095808 and PF-05095808 (AAV-2). Vector preparations were generated by transient transfection of HEK293 cells and purified by combined column chromatography and cesium chloride gradient centrifugation as previously described (42), with the modification that Poros 50HQ resin was used for the chromatography step. The purified vector was essentially free of empty capsids. Titers of vector preparations were determined by quantitative real-time PCR, and titers were expressed in terms of vector genomes/milliliter (vg/ml). The primers and probes for the real-time PCR were the following: forward, 5′-AGCTCCACCCTCACTGGTTTTT-3′; reverse, 5′-CAAGCATGATCAGAACGGTTGA-3′; probe, 5′-FAM-TTGTTGAACCGGCCAAGCTGCTG-TAMRA-3′ (where FAM is 6-carboxy-fluorescein and TAMRA is 6-carboxytetramethylrhodamine) (Applied Biosystems). A ViraPower Lentiviral Expression System (Invitrogen) was used for construction of the lentivirus-derived vectors, termed PF-05095808 (LV). Titers of vector preparations were determined using an enzyme-linked immunosorbent assay (ELISA)-based protocol to detect the p24 protein (VPK-107; Cellbiolabs). Recombinant adenoviruses (DE1/E3, adenovirus type 5), containing the PF-05095808 DNA expression cassette, termed PF-05095808 (AdV), were generated at Vector Biolabs, and titers were determined by endpoint dilution assay.
Synthetic ribonucleotides were custom synthesized by Integrated DNA Technologies with the following sequences: shRNA6, CGCGAAAGGCCUUGUGGUACUGAAGCUUGAGUACCACAAGGCCUUUCGCUUUUU; shRNA19, GUCAACUCCUGGCUAGGCAAUUUGUGUAGUUGCCUAGCCAGGAGUUGACUUUUUU; shRNA22, AUUGGAGUGAGUUUAAGCUGAAGCUUGAGCUUAAACUCACUCCAAUUUUUU; siRNA6, GAGUACCACAAGGCCUUUCGC, GAAAGGCCUUGUGGUACUGAA; siRNA19, UAGUUGCCUAGCCAGGAGUUGAC,CAACUCCUGGCUAGGCAAUUUGU; siRNA22, AUUGGAGUGAGUUUAAGCUGA,AGCUUAAACUCACUCCAAUUU.
Huh-7.5 cells (licensed from Apath LLC) were propagated in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, 1× nonessential amino acids, and 1× penicillin-streptomycin. The Huh-7-derived cell line Con1b replicon cells (licensed from RebLikon GmbH) supports replication of subgenomic RNA from the Con1 strain of HCV genotype 1b, which is engineered to encode the firefly luciferase reporter gene and neomycin phosphotransferase gene. Con1b replicon cells were maintained in Huh-7.5 medium supplemented with G418 at 500 μg/ml. The Huh-7-derived cell line H77 1a replicon cells (licensed from Apath LLC) supports replication of subgenomic RNA from the H77 strain of HCV genotype 1a, which is engineered to encode the Renilla luciferase reporter gene and neomycin phosphotransferase gene. The Huh-7-derived cell line JFH 2a replicon cells (licensed from Touray Industries) supports replication of subgenomic RNA from the JFH strain of HCV genotype 2a, which is engineered to encode the Renilla luciferase reporter gene and neomycin phosphotransferase gene.
Luciferase activity was measured using a firefly Britelite assay system (Perkin Elmer) and a Renilla luciferase assay system (Promega). Plates were read on an Envision 2104 multilabel plate reader (Perkin Elmer). Cytotoxicity was measured using a WST-1 proliferation assay (Roche).
The 50% effective concentration (EC50) potencies of synthetic RNAs in the replicon systems were generated by a lipofection reverse transfection protocol using Dharmafect3 (Dharmacon, United Kingdom). To ensure consistent nucleic acid concentration across the dose range, a “stuffer” oligonucleotide (5′-GACCACTTGCCACCCATC-3′) to a final concentration of up to 1 μM was used. In studies using the PF-05095808 vectors, virus inocula were preincubated with target cells for 1 h. For activity assays in HCV replicon lines, cells were incubated for 48 h posttransfection/transduction and then read for luciferase and cytotoxicity activity. For activity assays using the JFH infectious virus (19), cells were incubated for 72 h postinfection and then read for luciferase and cytotoxicity activity. EC50s for inhibitor activity in HCV replication and infection assays were calculated from experimental data using a Pfizer-proprietary statistical analysis plug-in for Microsoft Excel 2007 which fitted the four-parameter Emax model.
Con1b replicon cells were seeded in six-well tissue culture plates and cultured in G418-supplemented medium. Cells were passaged over a period of approximately 6 weeks and subjected to four separate transfections with siRNA or shRNA mixes at concentrations of 7× or 20× EC50. Cells treated with only Dharmafect3 were maintained for equivalent periods of time to serve as passage controls. Following expansion, siRNA/shRNA-resistant cells were subjected to phenotypic (EC50 determination as described above) and genotypic (nucleotide sequencing) analysis. For sequencing analysis, total RNA was extracted from replicon cell pellets (approximately 1 × 106 cells) using an RNeasy Mini Kit (Qiagen). First-strand cDNA synthesis was performed using Superscript III enzyme (Invitrogen) and a replicon-specific primer (5′-GGATGGCCTATTGGCCTGGA-3′). PCR amplification of the HCV nonstructural genes was achieved with AccuPrime Taq DNA Polymerase High Fidelity enzyme (Invitrogen) and gene-specific primers. Nucleotide sequencing of the purified amplicons was performed at Lark Technologies (Cambridge, United Kingdom).
A plasmid containing an HCV Con1b subgenomic replicon cDNA (designated pBB7; licensed from Apath LLC) served as a template for the creation of replicons encoding the C9198G, C9206A, G8546A, and T8903C sequence changes. Site-directed mutagenesis was accomplished using a QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's instructions. The plasmid pBB7 and derivatives thereof were linearized with SpeI and used as templates for production of replicon RNA by in vitro transcription (T7 Megascript Transcription kit; Ambion). Transcribed RNA was purified (Megaclear purification kit; Ambion), and then Huh-7.5 cells (5 × 106) were electroporated with 10 μg of replicon RNA using an Amaxa Nucleofector (Lonza Group, Ltd.) according to the manufacturer's instructions. Following electroporation, 1 × 104 cells were seeded to 96-well culture plates and held overnight at 37°C. Synthetic shRNA oligonucleotides prepared as above were added to electroporated cells and held for a further 72 h of incubation prior to luciferase and cytotoxicity measurements.
MicroRNAs (miRNAs) were purified from replicon cell lysates (miRNeasy Kit, item 74101; Qiagen), and purity was confirmed using an Agilent 2100 Bioanalyzer. A universal cDNA synthesis kit (203300; Exiqon) was used for reverse transcription of microRNAs to cDNA in all the assays. The resultant cDNAs were then taken forward to shRNA-particular qPCR assays. Specific amplification of the shRNA products expressed from PF-05095808 was achieved using a custom miRCURY LNA Universal RT microRNA PCR system (Exiqon) with primer sets particular to the shRNAs: primer shRNA6s2, GAAAGGCCTTGTGGTAC; primer shRNA19s3, GTCAACTCCTGGCTAG; and primer shRNA22as2, GGAGTGAGTTTAAGC. The forward primer was gene specific and designed to a consensus sequence present in the shRNAs, whereas the reverse primer was complementary to the poly(T) adaptor. Measurement of endogenous miR-103 RNA levels (204063; Exiqon) was included in all experiments as an internal control. PCR amplicon detection was by use of a SYBR green dye (203450; Exiqon); dissociation curve analysis was performed on all samples to confirm that only target-specific products were amplified. PCRs were read on an ABI 7900HT machine.
Primers were designed to enable amplification of the two regions of the HCV genome encompassing the shRNA target sites. The target sequence for shRNA6 was contained within a 380-bp PCR amplicon of the 5′ untranslated region (UTR), while the 790-base nonstructural protein 5B (NS5B) amplicon encompassed the target binding sites of both shRNA19 and shRNA22. To generate amplicons from the Con1b and H77 1a replicon cell lines, total cellular RNA was isolated and purified (RNeasy Mini Kit; Qiagen). For clinically derived amplicons, HCV RNA was extracted from 1 ml of HCV-infected plasma (QIAamp UltraSens Virus Kit; Qiagen). The two regions encompassing the shRNA binding sites were amplified from the resulting RNA preparations by first-strand DNA synthesis (SuperScript III First-Strand Synthesis System; Invitrogen), followed by a seminested PCR (AccuPrime Pfx DNA Polymerase; Invitrogen). For the shRNA6 target region derived from the replicon cell lines, primer Tac034 was used for first-strand synthesis, primers Tac023 and Tac033 were used for the first-round PCR, and primers Tac029 and Tac037 were used for the second-round PCR. For the plasma-derived RNA, primers Tac029 and Tac051 were used for the second-round PCR. For the NS5B region containing both the shRNA19 and shRNA22 target sites, an equimolar mixture of primers Tac046 and Tac047 was used for first-strand synthesis, primers Tac049 and Tac050 were used for the first-round PCR, and primers Tac042 and Tac053 were used for the second-round PCR. Primer sequences are as follows: Tac023, TGGGGGCGACACTCCACCAT; Tac029, CTCCACCCTCGAGCACTCCCCTGTGAGGAACTAC; Tac033, CTTTGAGGTTTAGGATTCGTGC; Tac034, GTTGGTGTTACGTTTGG; Tac037, GGTGTTACGTTGCGGCCGCCTTTGAGGTTTAGGATTCGTGC; Tac039, AAGATGTTCATCGAGTCCGACCCT; Tac040, GCCTCCGAATGAGAGTGTTTCGTT; Tac046, AAAAAAAAAAAAAAAAAAAA; Tac047, AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA; Tac051, GGTGTTACGTTGTTTAAACCTTTGAGGTTTAGGATTCGTGC; Tac049, YCTTCACGGAGGCTATGAC; Tac042, GGGACCTCGAGCAACCAGAATACGACTTGGAG; Tac050, TTCAYCGGTTGGGGAGSAGGTAG; Tac053, TGTTTAGTTTAAACTTCAYCGGTTGGGGAGSAGGTAG.
All PCR amplicons were bidirectionally sequenced. The PCR amplicons were ligated into the psiCHECK2 vector (C8021; Promega) using the restriction enzymes XhoI and NotI (replicon-derived 5′ UTR PCR amplicons) or XhoI and PmeI (clinically derived 5′ UTR amplicons and both replicon and clinical NS5B PCR amplicons).
Synthetic complementary oligonucleotides for each of the individual shRNAs encoded by PF-05095808 were annealed and ligated into the pENTR/H1/T0 vector (K4920; Invitrogen) according to the manufacturer's instructions. These were subsequently Gateway cloned into the pLenti-BLOCK-iT-destination vector (V488-20; Invitrogen). These vectors were used to generate lentivirus stocks for transduction of Hek-T-REx-293 cells (Invitrogen). Clonal selection was undertaken to enable identification of single tetracycline-inducible Hek-T-REx-shRNA cell lines. For activity assessment, Hek-T-REx-shRNA cell lines were transiently transfected with psiCHECK2 reporter constructs using Turbofect (R0531; Fermentas) as per the manufacturer's instructions. Plates were incubated for 48 h at 37°C in 5% CO2. Firefly and Renilla luciferase activities from psiCHECK2 vectors were measured for each sample; the firefly luciferase activity served as a transfection control, while Renilla luciferase activity enabled measurement of shRNA-mediated knockdown.
PF-05095808 comprises a recombinant AAV vector which directs expression of three shRNAs targeted to the HCV genome (Fig. 1A). The selection of shRNA was made on the basis of sequence conservation following a Jotun-Hein alignment analysis of approximately 100 full-length HCV sequences and thousands of partial-length sequences accessible from public databases. In the most optimal case, potential target sequences were selected from regions that contained a minimum of 21 nucleotides that would maintain absolute identity. In the absence of complete homology, the stringency was reduced to 80% of all sequences tested. The sensitivities of dozens of potential targetable sites within the HCV genome were tested empirically with individual shRNAs; three of the most potent were placed in PF-05095808. The shRNA6 sequence is complementary to the terminal portion of the 5′ untranslated region (5′ UTR), while shRNA19 and shRNA22 target nonstructural protein 5B encoding the viral polymerase (Fig. 1B).
Each shRNA is independently driven under the control of unique members of the U6 family of small nuclear RNA promoters; these have been weakened to mitigate against the adverse events reported to be associated with high-level shRNA expression (15; Suhy et al., submitted). Attenuated expression was achieved by incorporating the proximal sequence element from a U6-7 promoter into the family of U6 promoters employed for shRNA expression (6). PF-05095808 has therefore been specifically designed to enable dosing at high enough levels to achieve complete transduction of all hepatocytes in the liver by the AAV-8 vector without damagingly high levels of shRNA expression.
To confirm the antiviral activity of the PF-05095808 RNAi agents, synthetically generated shRNAs were delivered by reverse transfection to the HCV Con1b replicon system. The HCV Con1b replicon is a self-replicating subgenomic RNA molecule encoding the HCV NS3-NS5B polyprotein, adapted for efficient replication in Huh-7 cells (24). Levels of HCV RNA replication within these cell lines can be assayed conveniently through luciferase reporter activity, which is under the translational control of the HCV internal ribosome entry site. Antiviral activity against the Con1b replicon was consistently observed, with the following mean EC50s (± standard deviations [SD]): shRNA6, 1 (±1) nM; shRNA19, 54 (±59) nM; and shRNA22, 0.043 (±0.027) nM (Fig. 1C). There was no evidence of cytotoxicity associated with the transfections, as defined by signal of ≤10% in the WST-1 assay, when cells were transfected with up to 1 μM each synthetic shRNA (data not shown).
We next evaluated whether delivery of the PF-05095808 DNA expression cassette via an AAV-8 viral vector, the planned clinical investigational drug, resulted in antiviral activity. Following vector administration, dose-dependent inhibition of the luciferase reporter signal was observed in the Con1b and H77 1a but not the JFH 2a replicon cell lines (Fig. 2A). There was no evidence of cytotoxicity associated with the administration of PF-05095808, as defined by signal of ≤10% in the WST-1 assay, up to an administered dose of 2 × 106 vector genomes (vg)/cell (data not shown). To demonstrate the statistical significance (P test) of the antiviral activity, replicate studies at fixed doses of PF-05095808 were conducted. A dose of 1 × 106 vg/cell of PF-05095808 delivered 71% ± 0.3% inhibition against the Con1b replicon and 13% ± 0.3% inhibition against the JFH 2a replicon cell line. There was a statistically significant difference (P < 0.001) in the level of inhibition observed at doses of 1 × 106 vg/cell and 1 × 105 vg/cell for both the 1b and 2a replicon cell lines. No inhibition was observed in the control treatment group, which received empty AAV-8 capsid, devoid of the recombinant DNA vector genome (Fig. 2C). The reduced inhibition observed against the JFH 2a replicon relative to the Con1b and H77 1a replicons was consistent with the reduced sequence homology at the shRNA target sites in the 2a strain (Fig. 1B). There was no evidence of induction of the interferon response genes following virus vector administration, as measured by MX-1 upregulation (data not shown).
Equivalent studies were performed in the HCV infectious assay system (Fig. 2B) using both the standard JFH 2a strain and a virus designated JFH-808 with modified target sequences to be fully homologous to the shRNA. The initial observations of a dose-dependent response to PF-05095808 treatment were again confirmed by replicate fixed-dose studies (Fig. 2C). Complete inhibition (99% ± 0.6%) of the JFH-808 virus assay signal was observed at the highest dose of 2 × 106 vg/cell PF-05095808, while reduced inhibition (44% ± 7%) was observed against the standard JFH 2a virus. The higher level of inhibition observed against the JFH 2a virus (44% ± 7%) than that against the JFH 2a replicon (13%) may reflect the differences (2-fold) in maximum dose and replication kinetics between systems.
As reported in the literature (7) and observed in our own studies, serotype 8 AAV vectors are known to be highly efficient in vivo but have very poor in vitro transduction properties in target cell lines. Alternative vectors were therefore generated based on AAV serotype 2 (AAV-2), adenovirus (AdV), or lentivirus (LV) systems to improve the efficiency of in vitro delivery of the PF-05095808 expression cassette.
Viral vectors with improved transduction efficiencies significantly enhanced the in vitro efficacy of PF-05095808 (Fig. 3A). Calculated EC50s ranged over 4 logs from 1.19 to 4.97 log10 EC50 (Fig. 3B). Typically, the AAV-based vectors showed a maximum signal inhibition of around 80 to 90%. Inhibition with the LV vectors was also around 80%; higher dosing was precluded by the limiting LV stock titers. With adenovirus, 100% inhibition was observed but was concurrent with limited toxicity at the highest doses as assessed by the WST-1 assay. The rank order of efficacy for the different vector delivery systems was consistent with reported transduction capability (7). We hypothesized that the shifts in efficacy associated with use of alternate viral vectors simply reflected transduction efficiency to target cells.
The experiments with synthetic shRNAs and the PF-05095808 vectors confirmed the antiviral potential of this agent. To quantify the antiviral activity of PF-05095808, direct measurement of the effector RNAi molecules driving HCV genome cleavage is required. Therefore, quantitative PCR (qPCR) protocols specific to the shRNAs generated from PF-05095808 were developed. Host cell Dicer processing of shRNAs produces a family of RNAi molecules with different termini dependent on the cleavage sites in the shRNA stem-loop domains. A qPCR amplification methodology was used which captured all processed shRNAs, regardless of termini (1, 20). Following similar procedures for reverse transcription of microRNAs, the resultant cDNAs were taken forward to qPCR assays specific for each shRNA. The forward PCR primer was designed to the consensus sequence present in the different RNAi species; the reverse primer was complementary to the poly(T) adaptor, thus allowing RNAi species with different 3′ ends to be amplified.
Experiments demonstrated that shRNA expression levels correlated with the observed antiviral activity of PF-05095808 (Fig. 4A). On transduction of the Con1b replicon line with PF-05095808 AAV-2 at doses of 11,000 vg/cell, 1,200 vg/cell and 410 vg/cell, a dose-dependent decrease in replicon activity was observed. Analysis of the associated shRNA levels from these cell lysates showed that at the top dose of PF-05095808, all shRNAs were highly expressed at levels of 600 to 3,300 shRNA copies/cell. A decrease in shRNA expression levels is concomitant with lower doses of PF-05095808. Even though promoters from the same family of U6 polymerase were used, a slightly higher level of the shRNA22 strand was consistently detected relative to strands derived from the shRNAs 6 and 19. No significant difference in the numbers of each specific shRNA/cell was noted between 48 and 72 h postinfection.
The numbers of shRNAs/cell needed for antiviral activity in the different viral vector delivery systems were compared (Fig. 4B). Doses of PF-05095808 vectors predicted to deliver an EC50 were selected. Transduction with PF-05095808 AdV at 29 vg/cell, PF-05095808 AAV-2 at 1,200 vg/cell, and PF-05095808 AAV-8 at 10,000 vg/cell resulted in 47%, 67%, and 36% replicon inhibition, respectively. The levels of expressed shRNA delivering this antiviral effect proved remarkably consistent across the different vector systems, i.e., 800, 600, and 1,400 copies/cell for shRNA22 in the AdV, AAV-2, and AAV-8 systems, respectively. At PF-05095808 AdV doses of 29 vg/cell or higher, there was a ≥2-fold increase in threshold cycle (CT) values of the endogenous miR-103 control (data not shown). This is indicative of a nonspecific impact on host cell microRNA levels from the AdV infection and is consistent with the low level of cytotoxicity signals observed in the WST-1 assay under these conditions. This observation was unique to the PF-05095808 AdV transductions.
The qPCR methodology enabled precise quantitation of the RNAi molecules driving antiviral activity under these experimental conditions. Importantly, the numbers of RNAi molecules needed for efficacy proved to be independent of the transduction efficiency of the viral vector delivery system.
The emergence of resistance in the HCV replicons was screened for following serial passage of the Con1b replicon cell line with synthetic RNAs derived from PF-05095808. The Con1b replicon cells were passaged over a period of approximately 6 weeks and subjected to four separate transfections with siRNA or shRNA mixes at concentrations of 7× or 20× EC50. Replicon lines treated individually with either siRNA22 or shRNA22 showed evidence of a reduction in susceptibility to the challenge agents, as determined by EC50 assessment (data not shown). Replicon lines treated individually with either shRNA6 or shRNA19 showed neither resistance emergence nor replicon clearance. This was attributed to a failure to maintain 100% transduction efficiency and, therefore, consistent selective pressure over the course of the study.
To determine if any treatment-induced changes occurred in shRNA22-treated populations, replicon RNA was sequenced over the entire 5′ UTR and NS5B regions. Within the NS5B domain, nucleotide changes C9198G and C9206A were detected in the target site of shRNA22; these were not present in either passage controls or cell populations exposed to shRNA6 or shRNA19. The C9198G results in an amino acid change of leucine to valine; the C9206A change is silent. The C9198G and C9206A changes were not detected in our analysis of over 1,000 nonredundant genomic HCV sequences from publicly available databases. An additional nucleotide change in the NS5B domain (T8903C) was present across many different treatment groups. The signature nucleotide change G8546A, associated with resistance to the HCV polymerase inhibitor HCV-796, was detected in inhibitor control treatment groups (18). To further analyze sequence changes associated with siRNA22 or shRNA22 treatment, a subset of 50 clones was individually sequenced. The clones were randomly selected from passage control, siRNA22-treated (20× EC50), and shRNA22-treated (20× EC50) lines. This analysis revealed a greater diversity of sequence changes across the target domain, with 10/21 sites varying for shRNA22 and 4/21 for siRNA22 and no changes detected in the passage controls (Fig. 5A).
The C9198G, C9206A, T8903C, and G8546A changes identified from population sequencing were cloned as single or double changes into an HCV Con1b-derived replicon. Replicon lines were then challenged with a fixed dose of shRNA to determine whether there was a sensitivity difference between modified and parental replicons (Fig. 5B). Differences in the mean response from either single or double mutants compared to the control Con1b replicon were investigated, and results are expressed as differences in average response. Individually, the C9198G and C9206A mutations significantly altered the susceptibility of the test replicon cells compared to control HCV Con1b replicon cells to shRNA22 (P < 0.0001) but not to shRNA6 or shRNA19, indicating that these mutants likely arose from selective pressure. Further reduced susceptibility to shRNA22 treatment was observed in the double (C9198G C9206A) and triple (C9198G C9206A T8903C) replicon mutants (P < 0.0001). Furthermore, none of the replicons containing the multiple mutants was resistant to individual treatment with either shRNA6 or shRNA19.
The spectrum of PF-05095808 activity against commonly circulating genotype 1 strains of HCV was assessed using a reporter assay for RNA cleavage. Briefly, HCV target sequences of these common genotype 1 variants were cloned to the psiCHECK2 vector and inserted within the 3′ UTR domain of a Renilla luciferase. Cleavage of the HCV sequence embedded within the Renilla 3′ UTR leads to degradation of the transcript and a concomitant decrease in luciferase reporter signal. The psiCHECK2 vector also independently encodes firefly luciferase, thus enabling normalization for various transfection efficiencies. To enable stable inducible expression of shRNAs 6, 19, or 22, recombinant Hek-T-REx-293 cell lines were engineered by transduction with different lentiviruses encoding the individual shRNAs under the control of a tetracycline-inducible promoter. RNA cleavage activity was calculated by measuring relative Renilla and firefly luciferase signals following transient transfection of the psiCHECK2 vectors to the cognate Hek-T-REx-293-shRNA cell lines.
A clinical isolate panel was generated using HCV sequences with >10% prevalence, as defined by >1,000 nonredundant sequences extracted from the Los Alamos database (22). The panel was constructed by population cloning of the nearly complete 5′ UTR domain or terminal 800 bp of NS5B from 18 HCV genotype 1-infected plasma samples into the psiCHECK2 reporter plasmid. Control sequences were amplified from Con 1b and H77a 1a replicons and cleaved with the following efficiencies for Con1b and H77 1a, respectively: shRNA6 domain, = 74.5 and 73.9%; shRNA19 domain, 63 and 51.4%; shRNA22 domain, 80 and 83%. Reference psiCHECK2 control plasmids containing exact 20-mer target sequence matches were cleaved with high efficiency for all domains (>95%). There was no inhibition of the unmodified psiCHECK2 reporter signal following shRNA induction in the Hek293 cell lines (Fig. 6A).
Population sequencing was performed on all the psiCHECK2 plasmids that comprised the clinical panel. Because all 18 of the HCV 5′ UTR reporter plasmids proved to have entirely homologous shRNA6 target sites, it was not surprising to find that RNA cleavage efficiencies for these plasmids were equivalent to the Con1b control efficiency (Fig. 6B). For the NS5B domain targeted by shRNA22, although 3/18 isolates had single nucleotide changes, only one of these changes impacted potency, as defined by a >50% difference from the Con1b control in the reporter assay (Fig. 6B). There was a significantly higher level of sequence variation associated with the shRNA19 target site, where 15/18 clones showed changes. However, only three of these sequence changes impacted potency, as defined by a >50% difference from the control level (Fig. 6B). For all clinical isolates, the ligation efficiencies of the PCR amplicons to psiCHECK2 vector were confirmed to be >95%, thereby eliminating this as a source of variation in the data set. There was no detectable RNA cleavage when the 5′ UTR isolates were transfected to Hek-T-REx-293 cell lines expressing the NS5B shRNAs and vice versa. The prevalences of HCV sequences that were sensitive or insensitive to shRNA cleavage are reported in Fig. 6C.
Reverse transfection studies with synthetic RNAs demonstrated that each shRNA derived from PF-05095808 delivers an independent antiviral effect against the HCV Con1b replicon. Administration of PF-05095808 AAV-8 to standard HCV replicon and infectious virus systems also revealed dose-dependent antiviral activity, consistent with a sequence-specific effect. Replicon sequences were described on the basis of data from molecular clones. The poor efficiency of AAV-8 vectors in cultured cells continually presented a major challenge to the in vitro characterization of PF-05095808.
There have been numerous reports that RNAi-based agents such as PF-05095808 deliver antiviral activity in HCV replicon systems (14, 31, 35, 45). Our studies were intended to go beyond this proof of principle and to identify methodologies that would enable the linking of in vitro activity with the potential to predict in vivo efficacy. A key step in understanding the potential efficacy of a novel therapeutic is the accurate determination of potency in in vitro activity assays. Our results unequivocally demonstrated that observed potency in vitro was impacted by the transduction efficiency of the vector delivery system, to the extent of a 4-log range in EC50s between the viral variants. There was some concordance in EC50s (~30 vg/cell) against the Con1b replicon when the most efficient AdV and LV vectors for delivery of PF-05095808 were used although it is important that reported virus titers were determined by different methodologies and therefore may not be absolutely comparable. For viral vectors with a good correspondence between in vitro and in vivo transduction efficiencies, one approach to understanding efficacy may be direct measurement of virus genomes within transduced cells. To properly understand the efficacy of PF-05095808, an assay which measured the quantity of RNAi effector molecules required for antiviral activity was required.
A universal qPCR assay for direct quantification of the shRNA was therefore developed. This assay was suitable for both in vitro and in vivo derived sample analysis and so provides a method for linking activity in tissue culture systems with exposure and efficacy in the clinical setting. Using this assay, we were able to reproducibly determine the level of shRNA expression associated with HCV replicon inhibition. The number of shRNAs associated with an approximate EC50 effect across viral vector systems proved remarkably consistent, within a 0.5-log range; this despite a 3-log difference in the virus inoculum needed to deliver antiviral activity. We included the measurement of miR-103 RNA levels as an endogenous control in all experiments to give an independent measure of any nonspecific effects on host cell miRNA levels. In studies with the AdV system at high doses, we did detect indicators of toxicity in both miR-103 RNA and WST-1 assays, consistent with previous literature observations (33). However, our data support the idea that, using AAV-based delivery systems, it is possible to achieve antiviral activity against the HCV replicon by an RNAi-driven mechanism, in the absence of overt cytotoxicity or nonspecific effects on host cell miRNA levels.
We sought to determine the number of shRNA molecules needed for antiviral activity in the replicon system, with a view to building a framework for understanding the required doses for antiviral activity in the clinic. The limiting in vitro transduction efficiency of PF-05095808 in the replicon cell line meant that we were not able to confidently determine shRNA levels associated with a 90 or 99% effective concentration (EC90 or EC99, respectively) effect. The shRNA levels associated with an EC50 effect were in the order of 500 to 2,000 copies/cell under these experimental conditions. Replicon copy numbers are reported to be in the range of 1,000 copies/cell (2). Our data therefore suggest that, at least in this in vitro system, effector RNAi molecule levels do not need to be in great stoichiometric excess over target levels. The work presented in the manuscript uses qPCR methodologies that capture all processed shRNAs, regardless of termini, and thus the quantities reported here are slightly higher than those recorded in parallel PF-05095808 studies (Suhy et al., submitted). There are many challenges, however, in trying to understand the translation between in vitro and in vivo efficacy for agents of this type. Processing of the viral vector prodrug to the active RNAi molecules may vary between cell lines. HCV replication kinetics in the infected patient are poorly understood, and the numbers of genome copies per cell are reported to be significantly lower than in the artificially optimized HCV 1b replicon (23, 34). Given these and other uncertainties, the true potential of this agent to treat chronic HCV infection may fully be assessed only through clinical trials.
To better understand the mode of action of this agent, we screened for resistance emergence following serial passage with short RNAs. Resistance emergence to individual siRNAs and shRNAs was anticipated and has been previously reported (3, 13, 41). Indeed, exposure to siRNA22 and shRNA22 resulted in sequence and sensitivity changes at the target site in the Con1b replicon. Clonal sequencing revealed a greater diversity of changes associated with shRNA treatment than with siRNA treatment (10 versus 4 sites). Literature reports (1, 38) and our own experimental data (unpublished observations) support the finding that Dicer processing of double-stranded RNA results in families of shRNAs with different termini being generated. The diversity of changes associated with shRNA treatment may be reflective of selective pressure from a number of related siRNA molecules.
The predominant target site sequence changes of C9198G and C9206A were associated with a reduction in sensitivity to shRNA22 challenge. The reduction was augmented when the changes were presented as double mutations. The C9206A change lies at the heart of the predicted cleavage domain mediated by the RNA-induced silencing complex (RISC) and as such is consistent with current ideas of the preferred site of RNA cleavage, i.e., nucleotide 9/10 from the 5′ terminus (8). In contrast, the C9198G change is distal to the predicted cleavage site, where it is thought that changes will more likely be tolerated. Sequencing data of 50 further clones revealed nucleotide changes spread over the target domain and brings supporting evidence that changes across the breadth of an shRNA target site can impact cleavage. The nucleotide changes arising from prolonged shRNA22 exposure are consistent with sequence-specific selective pressure and therefore are likely to derive from an RNAi-directed mechanism. Further support for an authentic siRNA mode of action underlying the antiviral activity of PF-05095808 was obtained by 5′ rapid amplification of cDNA ends (RACE) characterization of replicon RNA cleavage products (unpublished observations).
Importantly, individual treatments of shRNA6 or shRNA19 remained fully active and effective against all shRNA22 replicon variants. Not only does this further support a sequence-specific mechanism of action consistent with RNAi activity, it also suggests that activity of PF-05095808 in a clinical setting might be entirely abrogated only in cases where HCV genomes have disruptive, mismatched sequences at all three targeted sites. Experiments testing the activity of PF-05095808 on the shRNA22 variants were not performed because of the difficulty of teasing apart the relative contributions of each of the expressed hairpins toward abrogating replicon activity. Far more sensitive cleavage assays would need to be developed for that type of analysis and is well beyond the scope of the studies presented here. Thus, understanding the barrier to resistance of this novel agent will be very difficult in the context of the current HCV model systems. Short hairpin RNA6 will likely provide the most rigorous barrier to the emergence of resistant strains because conformational restraints require that the primary sequence targeted by shRNA6 is exquisitely conserved. Indeed, to our knowledge there are no reports of resistant variants emerging in the HCV 5′ UTR. There is emerging evidence that with careful selection of direct-acting antiviral agents (DAAs), HCV inhibitors which in isolation suffer from resistance emergence have the potential in dual-therapy regimes to clear virus (25). An HCV therapy which delivers a high barrier to resistance in a single agent would be a big step forward for the field.
The spectrum of activity of specific sequences expressed from PF-05095808 was interrogated using an RNA cleavage reporter assay. This enabled an exclusive focus on the impact of individual sequence changes on RNA cleavage, without the inevitable disturbance of replication capacity of the HCV replicon. Although the clinical isolate data set predicts the relative cleavage efficiencies of different RNA sequences, it does not provide any measure of absolute efficacy. The observation that target sequences were cleaved more efficiently in short RNA domains is likely explained by reduced target accessibility in the extended HCV sequences. This further reinforces the idea that data obtained from such reporter assays must be analyzed in the appropriate context of the known complexity of the secondary structure for the full-length HCV genome.
Across the clinical isolate panel, small but measureable decreases in RNA cleavage efficiency were associated with changes in target sequences, with the exception of sample 18. The result for sample 18 may reflect sequence changes not identified by population sequencing (detection limit of 25%). The majority of single nucleotide changes had no detectable effect on observed cleavage activity (for shRNA19, 7/8 sequence changes; for shRNA22, 2/3 sequence changes). All clinical isolates showed some sensitivity to shRNA cleavage. To provide a measure of the effectiveness of PF-05095808 as an entire system, the three shRNA data sets derived from each individual plasma sample were analyzed to determine the number of target sites inhibited at ≥60%; Con1b control cleavage efficiency was 60 to 80%. For 17 of the 18 plasma HCV RNA samples, ≥60% inhibition of activity was observed with at least two of the shRNAs. The data generated using this novel RNA reporter assay provide supporting evidence that PF-05095808 will be efficacious against commonly circulating clinical isolates of HCV. Additionally the consistent and reproducible data generated support the use of this assay format to characterize any potential resistance emergence following PF-05095808 administration in clinical trials.
HCV therapy is a rapidly evolving field. New inhibitors bring the promise of improved sustained viral response rates but on top of the punishing regime that consists of interferon and ribavirin combination therapy. Not all patients can tolerate or will be suitable for the current SOC-based regime; indeed, there may soon be clearer stratification of those suitable for SOC therapy on the basis of IL28B and associated single nucleotide polymorphisms (12, 36). PF-05095808 represents the prototype of a new approach to HCV therapy, a single-dose treatment that can be administered alone or in combination with other anti-HCV agents. These studies demonstrated that PF-05095808 delivers sequence-specific antiviral activity in the absence of overt cytotoxicity. These studies have also shown the potential of PF-05095808 to target commonly circulating strains of HCV. Existing HCV model systems were used to probe efficacy and resistance emergence; however, clinical trials will be needed to truly understand the therapeutic potential of this compound. For many HCV-infected patients, particularly those living in regions with less developed health care systems, novel therapeutics such as PF-05095808 may provide the best option for combating this insidious infection.
We thank Tanya Parkinson, Paul Targett-Adams, and Mike Westby for useful discussions in preparing the manuscript. We acknowledge Katherine High for scientific discussions and Bernd Hauck and Olga Zelenaia for preparation and quality control testing of vectors.
Published ahead of print 27 December 2011