Sensitivity of hepatitis C virus to cyclosporine A depends on nonstructural proteins NS5A and NS5B†
Article first published online: 28 JUN 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 46, Issue 4, pages 1026–1033, October 2007
How to Cite
Fernandes, F., Poole, D. S., Hoover, S., Middleton, R., Andrei, A.-C., Gerstner, J. and Striker, R. (2007), Sensitivity of hepatitis C virus to cyclosporine A depends on nonstructural proteins NS5A and NS5B. Hepatology, 46: 1026–1033. doi: 10.1002/hep.21809
Potential conflict of interest: Nothing to report.
- Issue published online: 25 SEP 2007
- Article first published online: 28 JUN 2007
- Manuscript Accepted: 7 MAY 2007
- Manuscript Received: 24 MAR 2007
- UW School of Medicine and Public Health from “The Wisconsin Partnership Fund for a Healthy Future”. Grant Number: K08AI055750-03
- National Institutes of Health (NIH)
- American Cancer Society Research Scholar Grant. Grant Number: 07-077-01
HCV reoccurs after liver transplantation and increases mortality. Cyclosporine, but not tacrolimus, has potent antiviral effects against HCV replication in cell culture. To determine the conditions, if any, under which HCV is susceptible to cyclosporine in vivo, we selected for cyclosporine-resistant mutant HCV in vitro. The resulting mutations were mapped to x-ray crystallographic structures and sequence databases. Mutations selected by cyclosporine were clustered in the nonstructural (NS) proteins NS5A and NS5B. Different sets of mutations in NS5A, paired with the same 2 NS5B mutations, conferred different levels of cyclosporine resistance when engineered back into the HCV replicon. Mutations in NS5B are structurally consistent with a proposed model of regulation of RNA binding by cyclophilin B (CyPB). These mutations also highlight a natural polymorphism between different HCV genotypes that correlates with the variation in response to cyclosporine A (CsA) noted in some clinical trials. Replicons engineered to have mutations in only NS5A (P ≤ 0.0001) or only NS5B (P = 0.002) suggest that while both NS5A or NS5B variants alter cyclosporine susceptibility, NS5A has the largest effect. Conclusion: Preexisting sequence variation could alter the effect of cyclosporine on HCV in vivo. (HEPATOLOGY 2007.)
Hepatitis C virus (HCV) chronically infects approximately 170 million people worldwide.1 It is the principal indication for liver transplantation throughout the world. Infection of the new liver graft is universal. The problem is compounded by the fact that the immunosuppressants needed to prevent organ rejection worsen HCV-mediated disease. Consequently, while HCV takes decades to cause significant liver damage in an immunocompetent patient, end-stage liver disease develops in 5 to 7 years in HCV-infected liver transplant patients.2 Initially, recurrent HCV posttransplantation was thought to be fairly benign.3 However, more recent studies have demonstrated that HCV-infected liver transplant patients have significantly higher mortality and morbidity than patients transplanted for cholestatic liver disease.4 This difference may be more pronounced in recent transplants compared to those prior to 1990.5 It remains controversial whether differences between immunosuppressant regimens play a role.
Cyclosporine A (CsA) has recently been demonstrated to have potent anti-HCV activity both in the HCV replicon system and in the recently described JFH-1 2a cell culture system.6–9 CsA binds molecular chaperones called cyclophilins (CyPs) in cells and inhibits their peptidyl-proline isomerase activity.10 The CsA-CyP complex also binds calcineurin and inhibits its activation of T cells.11 FK506 binds FK506-binding proteins and inhibits calcineurin activation independent of CyPs. However, FK506 lacks antiviral activity in the HCV replicon system.7, 12 Certain CsA analogs, such as NIM811 and DEBIO-025, have been shown to have potent anti-HCV activity.9, 13 These CsA analogs retain their ability to bind CyPs, but are no longer recognized by calcineurin. These findings point toward CyP inhibition, not calcineurin inhibition, as being critical for the anti-HCV activity of CsA. Although CsA has been shown to inhibit all the HCV replicons tested thus far, the level of inhibition varies between the different HCV genotypes. The JFH-1 2a replicon has been demonstrated to have lower CsA susceptibility compared to a genotype 1b replicon.8 This, together with clinical data, suggests that different HCV genotypes may have a variable response to immunosuppressants. Thus, treatment of HCV liver transplant patients may need to be tailored based on sequence diversity.
Potentially conflicting models have been proposed to explain the inhibition of the HCV replicon by CsA. Cyclophilin B (CyPB) has been shown to bind the HCV polymerase and regulate the binding of RNA template to the polymerase.14 Another model argues that replication of the HCV replicon depends upon CyPs A, B, and C.6 Both models were based on experiments that used RNA silencing to target specific host CyPs that are known targets of CsA. We undertook a forward genetic approach in which no assumptions were made about the mechanism of CsA inhibition of HCV. We exposed HCV replicons to CsA and selected for HCV replicons with decreased susceptibility to CsA. We identified mutations in the selected replicon pool that mapped to the nonstructural proteins (NS) NS5A and NS5B. These mutations conferred CsA resistance. Mutations in NS5B, the viral RNA-dependent RNA polymerase, fit with a model of CyPB as a functional regulator of the viral polymerase.14 Our study is the first to demonstrate a role of NS5A in the CsA susceptibility of HCV. We show that NS5A has a larger effect than NS5B by separating the NS5A and NS5B mutations. Additionally, our data suggests that naturally occurring HCV polymorphisms may be critical factors in determining whether CsA inhibits HCV in individual patients.
Materials and Methods
Cells and Plasmids.
Huh7 and EN5.3 cells were kindly provided by Dr. Lemon (University of Texas Medical branch, Galveston, TX). Yi et al.15 and Ikeda et al.16 describe the production of EN5.3 cells and the pNtat2ANeo/SI. The plasmids pNtat2ANeo/SI and the pNNeo/3-5b/SI replicon containing the HCV 1bN consensus sequence (GenBank accession no. AF139594) with an adaptive mutation S232I in NS5A were also kindly provided by Dr. Lemon.15, 16 After transfection with replicon RNA, we selected the EN5.3 cells and we maintained them in the growth media containing 2.0 μg/mL blasticidin (Invitrogen) and 0.5 to 1.0 mg/mL G418 (Geneticin, Invitrogen).
CsA Passage Experiments.
We passaged Huh7 cells that were stably transfected with the pNNeo/3-5B/SI replicon in the presence of 1 mg/mL of G418 for 3 weeks both with and without 10 μg/mL of CsA (Sigma). We pooled the replicon cells and split them every 3 to 5 days to maintain a cell confluency of approximately 30% and we replaced the media with new CsA and G418.
Amplification of Replicon RNA Pool by Reverse Transcription–PCR and Cloning.
At the end of 3 weeks, we isolated total RNA using TRIZOL (Invitrogen) from the CsA passaged pool and control replicon cells. We used a SuperScript III One-Step reverse transcription polymerase chain reaction (PCR) mix (Invitrogen) to generate 7 independent PCR amplifications that covered the entire replicon from the HCV internal ribosome entry sequence to the 3′ untranslated region. We cloned the resulting PCR products into TOPO TA vectors (Invitrogen). We sequenced 5 to 16 TOPO TA clones containing PCR products from an individual PCR amplification. We repeated the process for RNA isolated from the control replicon cells. After we used the initial round of sequencing to determine the region with the highest frequency of mutations, 2 independent cDNA reactions were done using SuperScript II Reverse Transcriptase (Invitrogen) followed by PCR amplifications with a high-fidelity polymerase to generate NS5A-NS5B fragments from amino acid 214 of NS5A to amino acid 591 of NS5B. We generated 7 TOPO TA clones from the 2 cDNA amplifications. We then digested these clones with BlpI and ClaI and cloned them into the pNtat2ANeo/SI construct to replace the original sequence and generate mutant replicons 1 to 7. We transfected the original pNtat2ANeo/SI and the mutant replicon RNAs in parallel into EN5.3 cells and passaged them for several weeks in the presence of 0.5 mg/mL G418 and 2 μg/mL blasticidin. A total of 3 of the 7 mutant replicons succeeded in establishing stable replicon cell lines, which we called the CsA-1s, CsA-2s, and CsA-3s (GenBank accession numbers 855173, 860232, and 855179, respectively). For construct CsA-1s 5A, we cut CsA-1s with Blp1 and BstXI and inserted it into pNtat2ANeo/SI. For the CsA-1s 5B construct, we inserted the BstXI-ClaI portion of CsA-1s back into pNtat2ANeo/SI.
RNA Transcription and Transfection.
We linearized replicon DNA with XbaI and transcribed it using a T7 Ampliscribe Flash transcription kit (Epicentre Biotechnology). We transfected RNA into EN5.3 cells with a Transit mRNA transfection kit (Mirus). At 4 to 8 weeks after the establishment of stable replicon cell lines using G418 and blasticidin, we seeded an equal number of cells into 6-well plates with or without CsA, in the absence of G418. On day 6, we removed the media, washed it extensively with phosphate-buffered saline (PBS), and replaced it with new media and CsA. We used the secreted alkaline phosphatase (SEAP) assay to quantitate replicon RNA as described.15
We obtained mature CyPB cDNA by RT-PCR from Huh7 cDNA using primers containing BamHI and EcoRI sites. We generated the glutathione S transferase (GST)-CyPB construct by inserting the CyPB PCR product into pGEX-2T using BamHI and EcoRI restriction sites contained in the primers. We transformed the pGEX-2T and GST-CyPB plasmids into Rosetta bacterial cells (Novagen). We purified the recombinant protein as in Cheng et al.17
In Vitro Translation and Pull-Down Assays.
We generated all NS5B constructs by PCR with forward and reverse primers containing XhoI and PacI, respectively. We inserted PCR products into a modified p-internal ribosome entry sequence vector (Clontech) through the XhoI-PacI restriction sites. We used quick-change mutagenesis to make the P540A mutant of NS5B and to truncate 1bN and 1a NS5B constructs with a stop codon. We generated the NS5B protein using a TNT Quick Coupled Transcription/Translation System (Promega) with the incorporation of [35S]methionine (GE Healthcare) following the manufacturer's protocol. We performed the pull-down assay as described in Cheng et al.17
The SEAP values of the pNtat2ANeo/SI, CsA-1s 5A, and CsA-1s 5B (3 experiments), each measured at 4 different CsA levels, were normalized through division by the corresponding SEAP value at the CsA level of 0 μg/mL within the respective experiment for each replicon. Using SAS 9.1 PROC MIXED, we constructed a fixed-effects linear regression model with repeated measures having the log10-transformed normalized SEAP values as outcomes. Its main effects were the 3 replicons, CsA level (0, 1, 2, or 2.5 μg/mL), and SEAP value replicate number (1, 2, or 3). To model the correlation of the three log10-transformed normalized SEAP values per experimental unit, we chose a compound symmetry covariance structure.
Selection and Cloning of HCV Replicons with Decreased Susceptibility to CsA.
We selected CsA-resistant HCV replicons to uncover more about the antiviral mechanism(s) of CsA. We exposed Huh7 cells bearing the pNNeo/3-5b/SI replicon to 10 μg/mL CsA and 1 mg/mL G418. We used a high concentration of G418 to maintain the replicon despite the presence of CsA. The presence of both G418 and CsA concentration in the media led to cell death after each split, but this lessened over time. We passaged cells continuously in the presence of both drugs. After 3 weeks, we seeded an equal number of CsA-exposed cells and control cells and treated them with different concentrations of CsA to detect resistance via a real-time PCR assay. The CsA-passaged pool demonstrated considerable resistance to CsA compared to control cells (data not shown). We then isolated RNA from stably transfected Huh7/pNNeo/SI cells that were exposed to CsA and control cells. We performed multiple independent reverse-transcriptions and PCR amplifications and we sequenced the products. We obtained several synonymous mutations in both the control replicon and in the CsA-passaged replicon cells. We primarily found nonsynonymous mutations in the CsA-passaged pool. The nonsynonymous mutations that occurred frequently (>80%) were limited to the latter half of NS5A and the carboxy terminus of NS5B (Table 1). In particular, 5 mutations in NS5A and 2 mutations in NS5B showed up repeatedly from independent RT-PCRs. The only other consistent nonsynonymous mutation with a frequency higher than 20% was a glutamic acid to glycine mutation at amino acid 175 of NS3. Since this was genetically distant from the more frequently occurring mutations, we did not further investigate it. We then obtained PCR fragments that covered amino acids 214 of NS5A to 591 of NS5B with fidelity polymerase from cDNA amplified by Superscript II Reverse Transcriptase. We then cloned these PCR fragments into the TOPO TA vector to generate 7 independent amplicons. The same 7 mutations found in the initial round of sequencing occurred repeatedly in almost all 7 amplicons with some individual nonsynonymous mutations (Fig. 1A; Table 1). With the exception of amplicon 4 (Fig. 1A), all other amplicons contained the same 5 NS5A mutations and the 2 NS5B mutations. We cloned the 7 amplicons back into a cell culture–adapted replicon to determine if this locus could alter CsA susceptibility. The recipient replicon, pNtat2ANeo/SI, expressed not only neomycin resistance, but also the HIV trans-activating Tat protein that induces SEAP in EN5.3 cells.15 SEAP levels correlate with viral RNA, which allows replicon RNA levels to be monitored without cell lysis.15 Only 3 of the 7 replicon RNAs replicated efficiently enough to isolate stable cell lines (Fig. 1A). The 3 replicons, referred to as CsA-1s, CsA-2s, and CsA-3s, were then compared to the original pNtat2ANeo/SI replicon-containing cells to measure CsA susceptibility. The IC90 of the replicon-containing mutations selected by CsA exposure increased from 0.5 to 1.3 (1s), 1.2 (2s), and 2.2 (3s) μg/mL CsA (Fig. 1B). All 3 replicons replicated to the same levels as the pNtat2ANeo/SI. We did not observe a difference in kinetics between the pNtat2ANeo/SI and the mutants over a period of 7 days (data not shown). We confirmed by sequencing that the original mutations were still present and that we had found no further nonsynonymous mutations. We performed a second round of isolating stable replicon cell lines for replicon RNAs pNtat2ANeo/SI, CsA-1s, and CsA-3s, to reconfirm the decreased CsA susceptibility. The 2 mutants showed the same response to CsA as the first CsA-1s and CsA-3s replicon cell lines (data not shown).
|Mutations||Frequency of Mutations|
Sequence Variation in NS5A Has a Greater Effect on CsA Susceptibility than NS5B.
Since mutants with different NS5As and the same NS5Bs showed different levels of CsA susceptibility (Fig. 1B), our data implied that NS5A itself altered CsA susceptibility. We made 2 additional constructs, CsA-1s 5A and CsA-1s 5B, to determine if the mutations in NS5A or NS5B alone were sufficient to confer CsA resistance. We obtained stable replicon cell lines for both these constructs, together with the CsA-1s mutant and the pNtat2ANeo/SI replicons. The CsA-1s NS5A mutant conferred CsA resistance to the replicon without concomitant mutations in NS5B (Fig. 2) and, in fact, was not significantly different in CsA susceptibility than CsA-1s. The NS5B mutations alone also conferred a level of CsA resistance (P = 0.002), though the change in CsA susceptibility was relatively small compared to only the NS5A mutations (P ≤ 0.0001) (described in Table 2).
|Effect||Estimate||Standard Error||DF||t value||P > |t||
|0 μg/mL CsA||2.32||0.07||24||32.53||<0.0001|
|1 μg/mL CsA||0.87||0.07||24||12.15||<0.0001|
|2 μg/mL CsA||0.20||0.07||24||2.77||0.0107|
|2.5 μg/mL CsA||Ref||Ref||Ref||Ref||Ref|
Structural Differences in NS5B that Correlate with Genotype and CsA Resistance.
We initially sought to identify what role the NS5B mutations played in conferring resistance to CsA, as our NS5B mutations mapped to a previously identified region of the polymerase that bound CyPB. One proposed mechanism for the antiviral effect of CsA is that it inhibits either a CyPB-NS5B interaction or alters the interaction between NS5B and RNA, which is regulated by CyPB.14 We found 2 mutated NS5B residues in our CsA selection. One of these residues is a proline to threonine change at residue 538. This residue is in close proximity to the proline residue at position 540. A proline to alanine mutation at position 540 had been shown to disrupt the CyPB binding to the polymerase.14 Both the 538 and 540 residues are surface-exposed prolines that likely play important structural roles in the conformation of NS5B (Fig. 3A, dark blue). Our second mutation is a serine to glycine change at position 556. The 556 residue is a naturally occurring glycine in the relatively CsA resistant strain JFH-1 (genotype 2a)8 and in all genotype 2s (Table 3). The wild-type serine residue at position 556 directly interacts with RNA according to the x-ray crystal structure of the polymerase cocrystallized with a RNA oligomer (Fig. 3B, light blue).18 To examine the role of residue 556 in CsA resistance, we compared the 1b NS5B crystal structure (Fig. 3B, green and pink,18 to the YUY1 genotype 2a structure (Fig. 3B, orange),19 which has a naturally occurring glycine at residue 556. The overall root mean squared deviation between the 2 genotype 1b NS5Bs in the NS5B:RNA structure, and between the genotype 1b and 2a NS5Bs is relatively small, as has been reported.19 However, when comparing the carboxyl terminus of both structures, we noticed a large shift in one region of the thumb domain between the genotype 2a structure and two 1b structures. We then calculated the root mean squared deviation20 for the carboxy terminal residues 538-563 (previously linked to CyPB binding), and found it to be much larger between the 2a structure and the two 1b structures (2.494 Å) than between the two 1b structures alone (0.859 Å) suggesting significant structural differences (Fig. 3B). Thus it is possible that the decrease in CsA susceptibility of the CsA-1s NS5B replicon (Fig. 2B) might arise from a change in conformation of the mutant polymerase.
|Subtypes||Sequence||Sequence Variability at Residue 556|
Binding of NS5B to CyPB in the Absence of NS5A.
It was previously demonstrated that residues 521-570 of the polymerase were important for CyPB binding. To determine if changes at position 538 and 556 of the polymerase NS5B alter CyPB binding, we performed a GST-CyPB pull-down assay with the CsA-1s NS5B. We also included a mutated 1bN NS5B with a proline to alanine change at position 540 that was demonstrated to disrupt GST-CyPB binding.14 The 1bN served as a positive control. The H77 1a NS5B was also included in the pull-down because position 540 is an alanine in all 14 sequences from genotype 1a (Table 3) and thus might not be expected to bind CyPB based on previous data. The 1bN, 1a, CsA-1s, and P540A NS5B all interacted with GST-CyPB (Fig. 4A). We observed no interaction of NS5B with GST unless it was fused to CyPB. Further truncations of both 1bN and H771a localized the NS5B region required for the CyPB interaction site to amino acids 398-469 of the NS5B polymerase in the absence of NS5A (Fig. 4B).
Several different HCV replicons (H771a, JFH-1 2a, and 3 different 1bs) are all sensitive to CsA to different degrees in vitro.6, 8, 13 It is unclear if CsA has an antiviral effect in patients. This is an area of high scrutiny because recently, 2 trends have been observed simultaneously in patients. First, there has been a shift from CsA-based regimens to tacrolimus-based regimens.21 Second, while likely multifactorial, recurrent liver disease in HCV-infected patients may be worsening.22 Landmark clinical trials showed no difference between CsA and tacrolimus in liver transplants23, 24 and a recent meta-analysis was also unable to demonstrate differential outcome in CsA-managed compared to tacrolimus-managed HCV-infected liver transplant recipients.25 Studies that have demonstrated a beneficial effect of CsA on HCV have only been able to do so in genotype 1, generally in conjunction with interferon.26–30
We used a forward genetics approach to investigate the mechanism of CsA inhibition of the HCV replicon. CsA selected for mutations in NS5A and NS5B. This is unusual because most antivirals select for relatively uniform, single-point mutations and have a low genetic barrier to resistance.31 Three replicons, each with an NS5A that varied in its carboxy-terminal sequence, and the same 2 mutations in the carboxy-terminus of NS5B, were shown to have reduced CsA susceptibility. Interestingly, CsA-3s was more resistant than the other 2 replicons even though all 3 shared essentially the same NS5B sequence. NS5B mutations fit with a model of CyPB modulating NS5B activity.14 However, at least 1 mutant NS5A was able to confer decreased CsA susceptibility without NS5B mutations. The magnitude of this change in CsA resistance was large compared to that of the NS5B mutations alone. All of the mutant replicons replicated to same levels as the control and showed similar growth kinetics (data not shown). Additionally, all of the mutant replicons contained their original mutations and did not revert back to the wild-type sequence despite the fact that stable cell lines were obtained in the absence of CsA.
A mutant replicon containing NS5B mutations alone conferred a small, but significant amount of resistance compared to wild type. Although our data confirmed a CyPB:NS5B interaction, our attempts with in vitro assays to measure a CyPB enhancement of NS5B:RNA interactions found the background NS5B:RNA interaction to be too high to reliably detect a CyPB enhancement (data not shown). Our mapping of the CyPB:NS5B interaction in the absence of NS5A and RNA implicated a region between 398 and 469 to be important for this interaction. Genotypic and subgenotypic structural differences in the thumb may be important since it is a site for several nonnucleoside allosteric inhibitors in preclinical and clinical trials.19, 32, 33 Our data suggest not only genotype 1b NS5B, but also genotype 1a NS5B maintains the CyPB interaction despite genotype 1a having an alanine at position 540 (Fig. 4; Table 3). Although our mutant NS5B did not abrogate CyPB binding, it remains consistent with a model previously suggested,14 in which NS5B has a significant but not major role in CsA inhibition of the replicon.
The carboxy-terminus of NS5A is polymorphic, thought to be unstructured, and remains functional in the replicon despite insertions and deletions. Nevertheless, chimpanzee experiments have shown this region to be important for in vivo replication and possibly for interferon resistance (Enomoto et al.34 and reviewed in Appel et al.35). Additionally, this region of NS5A binds NS5B.36 It is possible that CsA may be acting by 2 separate mechanisms on NS5A and NS5B or through a single antiviral effect on the NS5A:NS5B replicase complex. Our data does not distinguish between these possibilities, because even though our NS5A and NS5B mutations were selected for and conferred CsA resistance together, they could also do so alone. The NS5B serine/glycine polymorphism at position 556 or NS5A polymorphisms that distinguish genotype 2/3 from genotype 1 could potentially be responsible for the differences in CsA susceptibility.27, 28 Nevertheless, the data presented here points to a polymorphic region of NS5A as having the largest effect on CsA susceptibility.
Nonimmunosuppressive analogs of CsA with even more potent antiviral activity against both HCV and HIV than CsA in cell culture are in early clinical trials.9 These drugs may have an important role to play in therapy, especially in HIV/HCV coinfected patients. Additional studies are needed to determine if the antiviral effect of CsA is clinically relevant, and what resistance mutations, if any, arise in vivo.
We thank Jim Keck for help analyzing crystal structures and Michael Lucey for helpful discussion and comments on this work. We also thank Richard Yang and Dipankar Bhattacharya for technical assistance.
- 27Multicenter randomized trial in HCV-infected patients treated with peginterferon alfa-2A and ribavirin alone after liver transplantation: 18-month report. In: The Liver Meeting 2006; Boston, MA: Schering-Plough; 2006. p. 47., , , , , , et al.