|Home | About | Journals | Submit | Contact Us | Français|
Hepatitis C virus (HCV) infection leads to acute and chronic liver diseases, and new classes of anti-HCV therapeutics are needed. Cyclosporine A (CsA) inhibits HCV replication, and CsA derivatives that lack the immunosuppressive function are currently in clinical trials as candidate anti-HCV drugs. Here, we characterize several independently derived HCV replicons with varying levels of CsA resistance due to mutations in nonstructural protein 5B (NS5B), the HCV-encoded polymerase. Mutant HCV replicons engineered with these mutations showed resistance to CsA. The mutations reside in two distinct patches in the polymerase: the template channel and one face of a concave surface behind the template channel. Mutant NS5B made by cells expressing the HCV replicon had increased ability to bind to RNA in the presence of CsA. Purified recombinant NS5B proteins containing the mutations were better at de novo initiated RNA synthesis than the wild-type control. Furthermore, the mutant proteins were able to bind RNA with approximately eightfold higher affinity. Last, mutation near the template channel alleviated the lethal phenotype of a mutation in the concave patch, P540A. This intramolecular compensation for the HCV replicase function by amino-acid changes in different domains was further confirmed in an infectious cell culture-derived virus system.
Increased level of CsA resistance is associated with distinct mutations in the NS5B gene that increase RNA binding in the presence of CsA, and the intramolecular communications between residues of the thumb and the C-terminal domains are important for HCV replicase function.
Cyclosporine A (CsA), a commonly used immunosuppressant for transplant patients, has recently emerged as a potential new anti-HCV therapeutic. CsA and its derivatives potently inhibit HCV replication both in cell-culture systems and in mice with transplanted human liver (1-3), although no consensus has emerged on the in vivo benefits of using CsA over Tacrolimus (FK506), a compound that lacks anti-HCV effect in vitro, for HCV-infected liver-transplant patients (4-9). More recent clinical trials with a CsA derivative, DEBIO-025, yield promising results in HIV and HCV coinfected patients, but drug resistance in vivo has not been studied (10).
Studies of the antiviral effect of CsA on HCV replicons has led to the identification of Cyclophilin (CyP) as an essential cofactor for HCV replication (11-13); point mutations in nonstructural protein 5B (NS5B) and NS5A that are associated with CsA resistance in vitro (14, 15). In addition, NS5B interacts with both CyPA and CyPB both in vitro and in vivo (12, 13). These data establish NS5B as an indirect viral target for the CsA-mediated inhibition of HCV replication. The CyPA-NS5B interaction has also been shown to be the principal mediator of cyclosporine resistance in vitro (12).
How CyPA affects NS5B function is unclear. The crystal structure of HCV NS5B lacking the C-terminal transmembrane domain has been solved. It adopts a typical right-handed shape that contains finger, palm, and thumb domains surrounding the active site (16-18). Characterization of allosteric, nonnucleoside inhibitors revealed extensive communication between the different domains (19). Here we identify a series of mutations in NS5B that conferred resistance to CsA and show that the mutations can affect RNA binding and synthesis by the HCV polymerase.
Cyclosporine A was purchased from Alexis Corporation (San Diego, CA). GS5- and CsA-resistant replicons have been described previously (14, 20). All replicon cells are routinely cultured in DMEM with 10% FBS and 500 μg/ml of G418.
These experiments were performed using published procedures (14, 20). Detailed information on these methods can be found in the supplemental methods. Primer sequences used in the cloning and sequencing are shown in the supplemental tables 1 and 2.
Transient replication was measured by isolation of total cellular RNAs from cells 7 h, 4 days, and 10 days after electroporation of the Rep1b or JFH-1 RNA. We performed real-time RT-PCR with a SYBR® Green PCR kit (Applied Biosystems) to detect HCV IRES and GAPDH expression. Sequences of the qRT-PCR primers are shown in the supplemental table 3. In vitro transcription, electroporation, colony formation, and Poly-U RNA binding assays were performed as previously described (12, 14).
Infection of Huh-7.5 cells by cell culture-derived virus was performed as previously described (21). Immunostaining of infected Huh-7.5 cells was carried out according to standard methods with a mouse monoclonal antibody against the core protein (Affinity Bioreagents, Golden, CO) and a FITC-conjugated goat-anti-mouse secondary antibody for detection (Sigma, St. Louis, MO).
Wild-type and mutant versions of the NS5B Δ21 were expressed in E. coli as C-terminal 6xHis tagged proteins and purified first by a metal ion chromatography and then with polyuridylate resins. Standard RdRp assays were performed as previously described (22). Protein concentrations were determined by comparison with known concentrations of BSA in Coomassie blue-stained SDS-PAGE.
Differential scanning fluorimetry was performed in a Stratagene MX3005P machine with an excitation of 492 nm and emission of 610 nm. Each sample was prepared in a total volume of 50 μl containing solutions of protein at 2.5 μM final concentration, SYPRO orange (Molecular probes) at 2.5 × final concentration. The samples were heated at a rate of 0.5°C/min, from 25 to 95°C, and the Tm values were calculated from the maximum of the first derivative with KaleidaGraph software (Synergy Software, Reading, PA).
Measurements were made with an LS55 spectrometer (Perkin-Elmer) and cuvettes with an optical path length of 0.4 cm at 22°C. The fluorescein-labeled RNA SL-UC was at 0.2 μM in 50 mM Tris-Cl (pH 7.5) and 25 mM NaCl. The final volume of protein added did not exceed 5% of the initial sample volume. Excitation wavelength was 495 nm, and each emission was scanned from 510 to 560 nm. The average of 10 scans was taken per data point and the data was analyzed by nonlinear least-square fitting using KaleidaGraph.
We have recently isolated CsA-resistant replicons using a combination of antibiotic selection and cell sorting (14). When single-cell clones of the resistant population were analyzed, various levels of CsA resistance were observed (Fig. 1A). The three single-cell clones with the highest levels of resistance were RS1-2, RS1-8, and RS1-9 (Fig. 1A); other clones had lower levels of resistance, similar to that of RS1-10 (supplemental Fig. 1A). To confirm that the CsA resistance of these single-cell clones was indeed conferred by the replicon RNA, we introduced total RNAs isolated from these cells, which contain the replicon RNAs, into naïve Huh-7.5 cells. The new replicons are designated RS1-2.2, RS1-8.2, RS1-9.2, and RS1-10.2, respectively. The replicon RNAs from all these clones phenocopied the resistant phenotype to the new replicon cells (supplemental Fig. 1B). Importantly, the relative level of resistance was maintained in the progeny replicons, demonstrating that the replicon RNAs in these clones contain distinct mutations that are responsible for the resistant phenotype.
HCV NS5B gene, which encodes the viral RdRp, has been implicated in the CsA-mediated inhibition of HCV and CsA resistance in vitro (12-15). We sequenced the NS5B region of the single cell clones and compared them to the wild-type sequence represented by GS5 replicon. Despite having a similar level of resistance, RS1-8, RS1-9 and RS1-2 contained different mutations in the NS5B coding region. A single common I432V mutation was found in RS1-2 NS5B while combinations of two mutations I11V/I454V and Q438R/E440G were found in RS1-8 and RS1-9, respectively. To determine whether these mutations could confer CsA resistance, we engineered them into a wild-type genotype 1b replicon, Rep1b, and assayed the resultant constructs for CsA resistance. Expression of the NS5A protein of Rep1b was below the detectable level when the replicon was treated with 0.375 μg/ml of CsA. In contrast, NS5A could be detected when the mutant replicons were treated with the same concentration of CsA (Fig. 1B). This result was confirmed with a colony-formation assay. In the absence of CsA treatment, electroporated replicon RNAs resulted in formation of G418-resistant colonies as a result of the expression of the neomycin phosphotransferase gene carried by these replicons. Inclusion of 0.375 μg/ml of CsA abolished the ability of the wild-type replicon (Rep1b) to form colonies, whereas the three mutant replicon RNAs retained colony formation (Fig. 1C). No significant differences were observed between the mutant and wildtype replicons’ ability to form colonies in the absence of CsA. These results demonstrate that these NS5B mutations are sufficient to confer increased CsA resistance without enhancing replication. However, the full level of CsA resistance of the original resistant isolates may require mutations in other genes (compare the CsA dosages in Fig. 1A and Fig. 1B and see supplemental table 4).
All of the mutations lie within two regions of the NS5B structure (Fig. 2A). First, mutations I432V, I11V and V405L are within the template channel or the Δ1 loop that extends from the finger subdomain to the thumb subdomain to form a latch over the template channel. This locking mechanism is primarily due to hydrophobic interactions between the Δ1 loop and the residues on the outer surface of the template channel (22). The second group of mutations, Q438R, E440G, I454V, and W550R, lie along one side of a concave surface at the back of the template channel, along a groove. Mutation R556V could not be mapped because it was not present in the crystal structure, but the trajectory of the C-terminal tail is projected to be near the template channel. To extend this analysis, we also examined the locations of the two mutations in NS5B identified by Fernandes et al. (15), S556G and P538T. S556G is located within the template channel and P538T is within the same side of the concave surface of NS5B, lining the aforementioned groove. A compilation of all nine mutations in the NS5B structure is shown in Fig. 2B; including a cut-away view that better exposes the locations of the mutations within the template channel.
On the basis of their locations in the NS5B structure, we hypothesize that the mutations could affect RNA binding by NS5B. To test this hypothesis, we assayed whether NS5B produced in cells could bind poly-U (pU) RNA in vitro in a CsA-sensitive manner (13). Cell lysates were incubated with pU resin, then washed and subjected to western blot analysis with antiserum to NS5B. Binding to protein G Sepharose served as a negative control. All NS5B variants were able to bind pU RNA in the absence of CsA (Fig. 3A). The interaction between the wild-type NS5B and pU RNA was reduced by more than 90% upon CsA treatment. In contrast, all three of the NS5B proteins produced by mutant replicons retained pU RNA binding when treated with CsA (Fig. 3A). The reductions in RNA binding by mutant proteins containing I432V, Q438R/E440G, and I11V/I454V were 0%, 5-10%, and 40-50%, respectively (Fig 3B). Note that these cells were treated with CsA for only 22 hrs, a time at which total NS5B protein level has not yet decreased. Longer treatment with CsA (>24 hours) would have resulted in a reduction in total NS5B.
To further determine if the mutations in NS5B that confer resistance to CsA affect NS5B activity, we expressed several of the mutants in E. coli in a form that lacked the C-terminal transmembrane domain (NS5BΔ21). The mutant proteins include I432V, I11V/I454V, Q438R/E440G, and a mutant that contained all five mutations (I11V, I432V, I454V, Q438R, and E440G), named Mut5 (Fig. 4A).
The HCV RdRp can initiate RNA synthesis de novo from the 3′ end of the viral genome (23, 24) or by extension from a primed template (25). Both activities could be measured in the same reaction with RNA LE19 that exists in equilibrium between a monomer and a dimer. The monomer forms an intramolecular hairpin that directs de novo initiation to produce a 19-nucleotide (nt) product. A dimer can serve as the template-primer complex for an extension reaction 32 nt in length (22) (Fig. 4B). In addition, template-switch products that are multimers of the 19-nt RNA are also produced from some of the de novo initiated 19-mers will (26). All mutant proteins retained the ability to synthesize the de novo transcript, the primer-extended products and the template switch products (Fig. 4C), indicating that the mutations that conferred resistance to CsA did not affect residues of the polymerase that are essential for catalytic activity.
The mutant proteins had more subtle differences with regard to RNA synthesis. First, RNA synthesis by I432V was increased several folds (Fig. 4C). Higher template concentrations were previously reported to inhibit RNA synthesis by Δ21 protein (21). In contrast, I432V was not inhibited in RNA synthesis at higher ligand concentrations (supplemental Fig. 2). Second, mutants Q438R/E440G, I11V/I454V, and Mut5 produced up to 4.6-fold more of the 19-nt de novo initiated RNA than of the 32-nt primer-extension product (Fig. 4D). Notably, I432V produced a ratio of the de novo-initiated and the primer-extended products similar to that of Δ21, suggesting that I432V is functionally distinct from the other mutants. Because Mut5 also contained the I432V mutation, the effects of the other mutations on the ratio of RNA synthesis appear to dominate that of I432V.
The conformation of the HCV RdRp is linked to the mode of RNA synthesis. To examine the effects of the mutations on NS5B conformation, we selected mutant proteins I432V (near the template channel) and Q438R/E440G (in the concave surface) as representing the two classes of mutant proteins. Differential scanning fluorimetry (27) was used to measure proteins’ unfolding as a function of temperature. The assay uses SYPRO orange, which fluoresces when it binds to hydrophobic regions of proteins that are exposed upon thermal denaturation. Both I432V and Q438R/E440G differed distinctly in thermal denaturation profiles from Δ21 (Fig. 5A) and from each other, in agreement with the observed differences in RNA synthesis (Fig. 4C). The functional and spectroscopic analyses support the model that the CsA-resistant mutations affect the conformation of the NS5B protein.
A fluorescence anisotropy assay was used to determine the affinity of RdRp-RNA binding. The RNA used was chemically synthesized to contain 5′ fluorescein (Fig. 5B) and a stable hairpin needed for efficient de novo initiation of RNA synthesis (28, 29). The binding isotherms for the three proteins were complex to fit, perhaps reflecting that the proteins may exist in equilibrium of two or more conformers, but when fitted to the same hyperbola equation, the affinities for SL-UC binding were 16 μM for Δ21 and less than 2 μM for I432V and Q438R/E440G (Fig. 5B). The apparent Kd values were reproducible within 10% in two independent experiments. These results show that mutations that conferred CsA resistance can increase RNA binding by NS5B in the absence of other proteins. Strikingly, both I432V (near the template channel) and Q438R/E440G (the concave surface of the RdRp), conferred increased RNA binding by the HCV RdRp.
We sought additional evidence for the functional relationship between the template channel and the concave surface of the HCV RdRp. A point mutation, P540A, that resides in the concave surface of the HCV RdRp abolished the replication of a genotype 1b replicon (13) and I432V rescued replication when engineered into the P540A background (14), demonstrating communication between the mutations at the two patches. We tested all the CsA-resistance mutations for their ability to alleviate the lethal effect of P540A in a transient replication assay. Replicon RNAs carrying P540A or P540A plus the CsA-resistance mutations were delivered into Huh-7.5 cells by electroporation. Total RNAs were extracted and analyzed by real-time RT-PCR. The input RNAs were at comparable levels in all the replicon derivatives at 7 h. After 4 days in culture, however, the P540A RNA had decreased to that of the negative control whereas the replicon derivatives had levels close to that of the wild type (Fig. 6). At day 10, the RNA levels of P540A/Q438R and P540A/E440G were lower than that of the wild type but three orders higher than that of P540A (Fig. 6A). The colony-formation assay results confirmed these observations (Fig. 6B). These results provide additional evidence that the two patches in the NS5B protein affected by CsA are functionally related.
The NS5B gene of GT2a carries a valine at position 454 instead of an isoleucine in the wild-type sequence. Importantly, the JFH-1 isolate of the GT2a genome can efficiently producing infectious particles in cell culture (30). To examine if Val454 could counteract the lethal phenotype of P540A in this genetic background, we engineered the P540A mutation into the JFH-1 genome and examined its effect on viral replication and infection of this isolate. In the transient replication assay, the P540A mutant was impaired in its ability to replicate but accumulated to a level higher than a non-viable control virus (GDD changed to GND in the catalytic core of NS5B) (Fig. 7A). In contrast, the double mutant of V454I and P540A (PAVI) was not viable whereas the V454I single mutant replicated close to wild-type level (Fig. 7A). There is a significant decrease in RNA level from 7 hour to day 4 for all the samples, likely a result of the higher amount of input RNA (10 μg as opposed to 1 μg used for the replicons in Fig. 6A) needed to produce infectious viruses. Nevertheless, for the wild-type, V454I, and P540A, the RNA level stabilized from day 4 through day 10, indicating sustained replication. Furthermore, when the supernatants from these cells were tested for their abilities to infect naïve Huh-7.5 cells, similar results were obtained (Fig. 7B). These results strongly suggest that the presence of a naturally occurring Valine at position 454 could prevent the lethal effect of P540A in an experimental setting that recapitulates the full cycle of HCV infection.
Here, we report that mutations in the HCV polymerase that confer CsA-resistant replication increased RNA binding by NS5B, which is an indirect viral target of CsA’s inhibitory effect on HCV replication. Together with our previous report that CyPA interacts with NS5B and regulates HCV replication (12), these results suggest a model in which CyPA assists NS5B in binding RNA for HCV RNA synthesis. When mutations arise within NS5B that improve RNA binding, CyPA is not as necessary for NS5B function. Structural analysis revealed that the mutations associated with CsA resistance and a lethal mutation for which they can compensate are located in two distinct patches in NS5B, one near the template channel and one on the back of the template channel. The two patches both affected the same activity, perhaps RNA binding with or without CyPA. Although the template channel certainly contacts the RNA, the patch behind the template channel may serve to stabilize RNA binding either directly or with assistance from CyPA.
NS5B exists in a closed conformation that is correlated with de novo initiation which, on the basis of the ternary structure of the phage ϕ6 RdRp, is likely to be the form that binds single-stranded RNA for de novo initiation (31). The template channel is too narrow, however, to accommodate the double-stranded RNA intermediate that must form during RNA synthesis; the RdRp must assume a more open conformation during elongative RNA synthesis (22). The open and closed conformations probably exist in equilibrium and we propose that CyP and/or mutations within the RdRp could affect the equilibrium. With the CsA mutations, the increased de novo initiation versus primer extension by several of the mutations led us to propose that at least some of the mutations affect the conformation of the RdRp, especially mutations I432V and I11A, which could contact the Δ1 loop.
The NS5B mutations did not completely restore CsA resistance when engineered into a Con1 replicon background. Mutations elsewhere in the genome, particularly in the NS5A region, can probably augment the CsA resistance in combination with the NS5B mutations (14, 15). NS5A has been proposed to serve as a cofactor for NS5B, and interactions between the two have been reported (32, 33). NS5A-NS5B interaction may therefore modulate the conformation of NS5B and NS5A mutations may synergize with the NS5B mutations to increase CyP or RNA-binding. Interestingly, although the CsA-resistant replicons contained mutations in NS5A and NS3 (supplemental table 4), these mutations were also found in GS5, which is a CsA-sensitive replicon (12, 14). It is possible that mutations in other NS proteins, although unable to confer resistance by themselves, can augment CsA-resistance when combined the NS5B mutations.
Polymorphism of the HCV NS genes may affect HCV replication fitness and drug sensitivity in vivo. Although the P540A is clearly lethal to genotype 1b replicon, an alanine at position 540 exists in many naturally occurring isolates. Interestingly, all these isolates also contain at least one of the rescuing mutations that we identified as natural variations (14). Taking advantage of the naturally occurring I454V variation in the JFH-1 isolate, we confirmed the intramolecular compensation using a full-length infection system. These results demonstrate that the mutations we found to confer CsA resistance in cultured cells could be found in natural HCV isolates. Also of note, the concentrations of CsA that we used here to select resistant replicons are in the same range of the blood concentrations of CsA and DEBIO-025 in liver transplant and HCV infected patients, respectively. These data suggest that CsA-based therapies may have varying effects based on the patient population.
We thank Drs. Takaji Wakita and Charlie Rice for reagents, Ruth Didier for assistance with flow cytometry, and Dr, Anne B. Thistle for proofreading.
Funding sources: This work was supported by the James and Esther King Biomedical Research Program (HT) and by NIH/NIAID AI073335 (CK).