Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Oct 1.
Published in final edited form as: Antiviral Res. 2018 Aug 3;158:45–51. doi: 10.1016/j.antiviral.2018.07.024

Evolutionary pathways to NS5A inhibitor resistance in genotype 1 hepatitis C virus

Shuntai Zhou a,#,#, Sara E Williford a,b,*, David R McGivern a,b, Christina L Burch c, Fengyu Hu a,b,f, Tiffany Benzine a, Paul Ingravallo d, Ernest Asante-Appiah d, Anita YM Howe d,*, Ronald Swanstrom a,e, Stanley M Lemon a,b,#,#
PMCID: PMC6146077  NIHMSID: NIHMS1503051  PMID: 30081054

Abstract

Direct-acting antivirals (DAAs) targeting NS5A are broadly effective against hepatitis C virus (HCV) infections, but sustained virological response rates are generally lower in patients infected with genotype (gt)-1a than gt-1b viruses. The explanation for this remains uncertain. Here, we adopted a highly accurate, ultra-deep primer ID sequencing approach to intensively study serial changes in the NS5A-coding region of HCV in gt-1a- and gt-1b-infected subjects receiving a short course of monotherapy with the NS5A inhibitor, elbasvir. Low or undetectable levels of viremia precluded on-treatment analysis in gt-1b-infected subjects, but variants with the resistance-associated substitution (RAS) Y93H in NS5A dominated rebounding virus populations following cessation of treatment. These variants persisted until the end of the study, two months later. In contrast, while Y93H emerged in multiple lineages and became dominant in subjects with gt-1a virus, these haplotypes rapidly decreased in frequency off therapy. Substitutions at Q30 and L31 emerged in distinctly independent lineages at later time points, ultimately coming to dominate the virus population off therapy. Consistent with this, cell culture studies with gt-1a and gt-1b reporter viruses and replicons demonstrated that Y93H confers a much greater loss of replicative fitness in gt-1a than gt-1b virus, and that L31M/V both compensates for the loss of fitness associated with Q30R (but not Y93H) and also boosts drug resistance. These observations show how differences in the impact of RASs on drug resistance and replicative fitness influence the evolution of gt-1a and gt-1b viruses during monotherapy with an antiviral targeting NS5A.

Keywords: Primer ID sequencing, viral evolution, treatment outcome, resistance-associated substitution, reporter virus

1. INTRODUCTION

New all-oral, combination direct-acting antiviral (DAA) therapies are highly effective against hepatitis C virus (HCV) and result in sustained virological response (SVR) in most (>95%) persons with chronic hepatitis C (Bartenschlager et al., 2013). Most DAA combinations used today include an inhibitor of the nonstructural (NS) 5A protein combined with an NS3/4A protease inhibitor and/or an NS5B RNA-dependent RNA polymerase inhibitor. While treatment failure is uncommon, sequencing of HCV RNA after failure typically suggests the presence of drug-resistant viral variants in patients who do not achieve SVR. NS5A inhibitors are a highly potent class of DAA (Gao et al., 2010; Ivanenkov et al., 2017). NS5A is unique to HCV and closely related viruses with multiple essential functions during the viral life cycle, but no known enzymatic activity (Ross-Thriepland and Harris, 2015). NS5A inhibitors have dual modes of action, both impeding viral RNA synthesis by blocking the formation of replicase complexes, and also preventing the intracellular assembly of infectious virions (Benzine et al., 2017; Berger et al., 2014; Boson et al., 2017; McGivern et al., 2014).

HCV is characterized by a high degree of inter- and intra-host diversity, which likely plays an important role in the response to DAAs and development of resistance (Jabara et al., 2014; Smith et al., 2014). There are 7 HCV genotypes that differ by more than 30% at the nucleotide level, with each further divided into subtypes that differ by 20% (Smith et al., 2014). Genotype 1 (gt-1) is the most common globally and in the U.S. Early phase 1 clinical trials of NS5A inhibitors suggested that the antiviral efficacy of NS5A inhibitors was greater in patients with gt-1b versus gt-1a infection (Lawitz et al., 2012; Liu et al., 2015). Amino acid substitutions in NS5A associated with antiviral resistance (‘resistance-associated substitutions’ or RASs) also differ between gt-1a and gt-1b viruses (Issur and Gotte, 2014; Lontok et al., 2015; Zeuzem et al., 2017). Amino acid substitutions conferring at least a 2.5-fold reduction in susceptibility to an NS5A inhibitor (daclatasvir, ledipasvir, pibrentasvir, or elbasvir) have been identified at residues 24, 28, 30, 31, 32, 38, 58, 92, and 93 within domain I of NS5A (Gottwein et al., 2018; HCV Drug Development Advisory Group, 2012; Zeuzem et al., 2017). However, the diversity of RASs appears to be greater in gt-1a than gt-1b viruses (Zeuzem et al., 2017). More importantly, while pre-existing NS5A RASs may negatively impact the SVR rate in gt-1a patients treated with some NS5A inhibitors, such substitutions have limited impact in gt-1b infections (AASLD-IDSA; Zeuzem et al., 2017).

To better understand the evolutionary pathways leading to NS5A inhibitor resistance in gt-1a and gt-1b infection, we used next generation sequencing (NGS) with tagged cDNA primers (“primer ID sequencing”) (Jabara et al., 2011; Zhou et al., 2015) to intensively characterize HCV diversity in small cohorts of gt-1a and gt-1b infected patients recieving a 5 day course of elbasvir monotherapy. Primer ID sequencing overcomes several limitations of conventional NGS by eliminating multiple types of PCR-related errors, thereby allowing unprecedented sequencing depth and accuracy (overall error rate ~0.01%) (Jabara et al., 2011; Zhou et al., 2015). Each template consensus sequence (TCS) generated by primer ID sequencing represents an original viral RNA template queried at the very first cDNA synthesis step, making primer ID sequencing an excellent tool to study highly diversified viral populations such as HIV-1 and HCV (Barnard et al., 2016; Jabara et al., 2014; Keys et al., 2015; Zhou et al., 2016). Coupling this approach with a unique set of serial serum samples collected before, during and after five days of elbasvir monotherapy, we show here that gt-1a and gt-1b HCV acquire antiviral resistance through distinct selection pathways, and that genotype-specific differences in RAS linkage and frequency can be linked to differences in cell culture assays of viral fitness.

2. Materials and Methods

2.1. Study participants, clinical samples, and primer ID sequencing

Serial plasma samples were collected from subjects participating in a phase 1b clinical trial in which they received either elbasvir (10 mg or 50 mg q.d.) monotherapy or placebo for 5 days (MK-8742 P002 trial, ClinicalTrials.gov Identifier: NCT01532973) (Liu et al., 2015). Seven subjects were infected with gt-1a HCV (4 receiving 50mg of elbasvir, 2 10mg of elbasvir, and 1 placebo) and 6 subjects were infected with gt-1b virus (5 receiving 50mg elbasvir and 1 placebo). RNA encoding domain I of NS5A (amino acids 23–123) was sequenced in an average of 6 serum samples from each subject, collected between 0 (pre-treatment) and 60 days after the initiation of therapy (Table S1). Primer ID libraries were constructed using previously published protocols and primer sequences shown in Table S2 (Zhou et al., 2016; Zhou et al., 2015). Multiplexed libraries were sequenced using Illumina MiSeq (Illumina, San Diego, CA) 300 bp paired-end sequencing.

2.2. Bioinformatics and phylogenetic analysis

The Illumina bcl2fastq pipeline (v 1.8.4) was used for initial de-multiplexing of the data and TCS pipeline (v 1.3.2) to create primer ID template consensus sequences (TCSs) (available at https://github.com/SwanstromLab/PID). In-house scripts were generated to derive abundances of substitutions at each nucleotide and amino acid position in each sample. The primer ID approach allowed identified linkages within each TCS, and neighbor-joining trees were built with MUSCLE (v 3.8.31) (Edgar, 2004a, b).

2.3. In vitro replicative fitness assays

The impact of RASs on replicative fitness and drug resistance were determined using recombinant cell culture-adapted gt-1a H77S.3/GLuc2A (Shimakami et al., 2011) and gt-1b N.2/Gluc2A (Yamane et al., 2014) reporter viruses expressing Gaussia princeps luciferase (GLuc). RASs were introduced into plasmid DNAs using Quikchange XL site-directed mutagenesis (Agilent). Plasmids were linearized with XbaI and transcribed in vitro to produce HCV genomic RNA with the T7 Megascript kit (Ambion, Austin, TX). Huh7 2–3c cells were electroporated with the RNA and replication fitness assessed by monitoring secreted GLuc activity as described previously (Mauger et al., 2015; Shimakami et al., 2011).

3. RESULTS

3.1. primer ID MiSeq sequencing of HCV RNA in serial plasma samples

Approximately 3,000 to 12,000 TCSs were obtained from each clinical sample, providing a sampling depth of 0.1% to 0.03% (defined as the abundance of minor variants with a 95% chance of being detected) (Table S1). However, extremely low or undetectable levels of viremia precluded sequencing of virus in samples collected between 1 to 13 days after starting elbasvir therapy in all gt-1b and some gt-1a-infected subjects.

3.2. Pre-therapy resistance-associated substitutions and intra-host diversity

We considered that differences in the genetic diversity of gt-1a and gt-1b viruses could contribute to differences in the development of NS5A inhibitor resistance and thus treatment outcome. We thus examined the frequency of pre-treatment polymorphisms at the most relevant 16 amino acid positions between codons 23 and 123 of NS5A (Fig. 1A). Significant baseline intra- and inter-host variation existed at codons 28, 30 and 93 (positions of several important RASs), but the intra-host average pairwise diversity (π) of the baseline nucleotide sequences of the viral populations did not differ significantly between gt-1a and gt-1b subjects (π = 0.87 vs. 0.82, p = 0.30). We also calculated estimated genetic distances using the model-based TamuraNei method (TN93 distance) which assesses the distribution of pairwise distances between sequences (Tamura and Nei, 1993). Three of 7 gt-1a subjects (subjects 1, 2, and 5) but only 1 of 6 gt-1b subjects (subject 12) showed a high frequency of TCSs with a TN93 distance greater than 0.04 substitutions/site (p=0.24 by x2 test) (Figs. 1B and 1C). The existence of multiple peaks in the frequency distributions of TN93 distance suggest that that these subjects had several distinct viral lineages before therapy.

Figure 1. Pre-treatment intra-host diversity in HCV RNA encoding domain 1 of NS5A.

Figure 1.

Open in a new tab

(A) Baseline heat map of amino acid substitutions at potential RAS positions in domain I of NS5A in subjects with gt-1a (subjects 1–7) and gt-1b (subjects 8–13) infections. Positions analyzed include those identified by elbasvir-induced changes in Shannon’s entropy (see Figs. S1 and S2). Frequencies of specific substitutions are shown in different colors for each subject at each position (see key at bottom). Highlighted positions are major sites of RASs previously reported. Pos, position; WT, wildtype; aa, amino acid. (B and C) Distributions of TN93 distances of baseline HCV viral populations in gt-1a (B) and gt-1b (C) subjects. The x-axis is the TN93 distance (substitutions per site), the y-axis shows the frequency of TN93 distance. Multiple peaks on the distribution plot suggests that the viral population is comprised of multiple distinct variants.

3.3. Elbasvir-induced changes in sequence encoding NS5A domain I

We calculated Shannon’s entropy as a measure of the variability of amino acid residues at each position within the NS5A protein (Korber et al., 1994). Significant elbasvir-induced changes occurred between baseline and day 28 in gt-1a virus infected subjects at codons 28, 30, 31, 58, and 93 (Fig. S1), and in gt-1b subjects at codons 28, 31, and 93 (Fig. S2). Significant changes in the frequency of amino acid polymorphisms were evident by day 1, but not 4hr after the start of elbasvir therapy. In contrast, no changes were observed in Shannon’s entropy at any position in subjects (7 and 13) receiving placebo.

3.4. Evolutionary pathways to NS5A inhibitor resistance in gt-1a virus

Since primer ID sequencing retains information concerning linkage between substitutions, subsequent analyses focused on the linkage between changes at codons 28, 30, 31, and 93, determining haplotypes defined by these positions at each timepoint. As an example, Subject 2 was infected with gt-1a HCV and recieved 10mg elbasvir (Table S1). At baseline the viral population was homogenous at all four codons, with M28, Q30, L31 and Y93 (“MQLY” haplotype, Fig. 2A, top). By day 2, the viral population had changed dramatically, consistent with a substantial, greater than 3.0 log10 drop in the level of viremia (Fig. 2A, bottom). The original consensus MQLY haplotype had declined to only 14% of the population, and haplotypes with Y93H (MQLH) and L31M (MQMY) dominated with frequencies of 41% and 23%, respectively. Minor haplotypes included those with Q30R (MRLY), Q31H (MHLY) Q30E (MELY), or M28T (TRLY), each at ~5%.

Figure 2. Evolutionary pathways to elbasvir resistance in gt-1a-infected subjects.

Figure 2.

Open in a new tab

(A) gt-1a infected subject 2. (B) gt-1a-infected subject 6 with a dominant pre-existing Q30R substitution. Haplotypes defined by sequence at amino acid positions 28, 30, 31 and 93 (see text) are shown at the bottom of each graph. Dashed lanes indicate when the elbasvir monotherapy ended.

These variants evolved throughout the 5-day treatment period and continued to change in frequency after therapy with the emergence of new haplotypes (Fig. 2A). Variants with Y93H were most abundant at day 4 (79% MQLH), but quickly declined in frequency after cessation of therapy (17% by day 12) and were largely absent by day 60. Y93N (MQLN) represented 4% of the population at day 4, increased to 14% at days 9 and 12, and similarly decreased to ~1% by day 60. In contrast, variants with substitutions at position 30 (MRLY, MELY, and MRMY) began to increase in frequency off drug at day 6, becoming dominant as the viremia rebounded to the baseline level. Variants with substitutions at codon 31 (MQMY and MQVY) persisted at a frequency of 5–10% at the end of the study. Thus, we documented two major pathways to drug resistance in this subject: (1) Y93H/N/S/C, and (2) Q30R/E/H with or without L31M, with less common pathways to resistance including L31M/V, and M28T with or without Q30R. Temporal changes in the relative frequencies of individual haplotypes on and off therapy were likely driven by differences in degrees of elbasvir resistance and replicative fitness. Importantly, we never observed Y93H and Q30R in the same lineage.

Both of these major resistance pathways were observed in 5 of the 6 gt-1a subjects (Figs. 2A and S3). Subject 6, in whom Q30R was dominant at baseline (98% MRLY), was an exception (Fig. 2B). This subject experienced less than a 2 log10 drop in viremia by day 6 despite treatment with 50 mg elbasvir daily. The baseline MRLY haplotype decreased in frequency during and off treatment, dropping to only 6% by the end of the study. L31M/V substitutions began to emerge within the background of Q30R on day 1, with an MRVY haplotype becoming dominant at day 7, and increasing to 92% by day 60 when viremia was greater than 10-fold that at baseline.

3.5. Evolutionary pathways of NS5A inhibitor resistance in gt-1b virus

In gt-1b infected subjects, we observed little or no change in haplotype frequencies 4 hrs after starting elbasvir, but found viremia suppressed so severely by day 1 that changes in the viral population could not be assessed. Major changes were observed in rebounding virus populations, however. For example, viral rebound was sufficient in subject 8 by day 21 to determine that the baseline consensus LRLY haplotype (100%) had been largely replaced by a haplotype with dual Y93H and L31M substitutions (68% LRMH, Fig. 3A). MRLH , LRVH and LRLH haplotypes were also present as minor variants. Unlike what we observed in gt-1a infected subjects, gt-1b haplotypes with Y93H persisted off therapy and remained dominant at the end of the study (Fig. 3A). Resistant variants emerged in a similar fashion in subjects 9, 10 and 11 who were also infected with gt-1b virus (Fig. S3). In each, Y93H was dominant in the rebounding viral population and persisted with or without concomitant L31M/V or L28M substitutions, whereas the baseline consensus haplotype was detectable but remained at a low levels throughout followup.

Figure 3. Evolutionary pathways to elbasvir resistance in gt-1b-infected subjects.

Figure 3.

Open in a new tab

(A) gt-1b-infected subject 8. (B) gt-1b-infected subject 12, with a high-frequency of pre-existing variants with a Y93H substitution. See legend to Fig. 2 for additional details.

Subject 12, who had a high frequency of Y93H at baseline (24% LRLH), was an exception among the gt-1b subjects (Fig. 3B). The baseline consensus haplotype, LRLY, was decreased in frequency by 4 hrs, and present at very low frequency when viremia rebounded at 21 days. Variants with an L31V substitution (LRVY) were dominant in the rebounding virus population (Fig. 3B), but over time this haplotype was partially replaced by an increasing frequency of Y93H variants (LRLH, LRVH, and LRIH haplotypes) which persisted to the end of the study.

3.6. Genetic structure of HCV populations before and after elbasvir treatment

We performed phylogenetic analyses to further examine the linkage between RASs and the genetic structure of viral populations. Fig. 4 compares the phylogenetic trees of HCV in subjects with gt-1a and gt-1b infections at baseline and following elbasvir treatment. In the 1a-infected subject shown (subject 4, Fig. 4A), resistant variants with Y93H/N substitutions can be seen to have arisen by day 7 in multiple independent lineages, some very closely related to baseline lineages. Importantly, there were no lineages with both Y93H/N and Q30R substitutions, consistent with the linkage analysis described above. Similar evolutionary patterns were observed in other gt-1a subjects receiving elbasvir (Fig. S4). In contrast, in gt-1b-infected subjects (for example subject 9, Fig. 4B), the viral population shifted completely post-therapy as Y93H emerged and persisted in novel lineages, some with additional substitutions on top of Y93H, such as L28M, L31V, and L31M. With the exception of subject 12, phylogenetic patterns were similar in the other gt-1b-infected subjects (Fig. S5). In contrast, population structures were stable in placebo recipients (Figs. S6 and S7).

Figure 4. Phylogenetic trees of sequences obtained at 0 hour and first available data point after the 5-day therapy.

Figure 4.

Open in a new tab

(A) gt-1a-infected subject 4, (B) gt-1b-infected subject 9. Sequences in blue were obtained from the 0-hour sample whereas those in red are from the 7-day sample and 21-day sample for subjects 4 and 9, respectively. Haplotyes defined by amino acid residues at positions 28, 30, 31 and 93 are labeled with each major viral linage. Residues shown in black font represent baseline (consensus sequence) whereas those in color are RASs.

3.7. RAS impact on replicative fitness and drug susceptibility of gt-1a and gt-1b virus

Both drug resistance and replicative fitness influence how viral variants emerge and persist with antiviral therapy. To see whether in vitro measurements of these parameters might explain the differences we observed between gt1–1a and gt-1b-infected subjects, we used luciferaseexpressing gt-1a (H77S.3/GLuc2A, MQLY haplotype) and gt-1b (N.2/GLuc2A, LRLY) reporter viruses (Beard et al., 1999; Yamane et al., 2014; Yi et al., 2006) (Fig. 5A). GLuc expression by these viruses provides a measure of their replication. Preliminary studies demonstrated that the EC50 of elbasvir (determined 72 hrs after drug addition) was lower for the gt-1a (2.2 pM) than the gt-1b reporter virus (7.1 pM). The maximal inhibition achieved with concentrations of elbasvir ≥250 pM was also greater for the gt-1a versus gt-1b virus at 48 (p<0.0001 by two-sided t-test) and 72 hrs (p=0.0033) after addition of drug (Fig. 5B). Since NS5A inhibitors impede RNA replication by blocking assembly of the replicase complex, this suggests a significantly shorter half-life for the gt-1a replicase complex (Benzine et al., 2017; McGivern et al., 2014). These findings are consistent with previously published work indicating that elbasvir is at least as potent against gt-1a as gt -1b viruses (Liu et al., 2015).

Figure 5. Impact of RASs on replicative fitness of gt-1a and gt-1b reporter viruses.

Figure 5.

Open in a new tab

(A) Genome organization of gt-1a H77S.3/GLuc2A and gt-1b N.2/GLuc2A reporter viruses that express Gaussia princeps luciferase (GLuc) from sequence inserted upstream of the NS5A-coding region, between p7 and NS2 (Yamane et al., 2014). (B) Inhibition of GLuc secretion by cells infected with (left) gt-1a H77S.3/GLuc2A or (right) N.2/GLuc2A viruses 20, 48 or 72 hrs after the addition of elbasvir at the indicated concentrations. Results represent the mean ± SEM of 3 technical replicates. (C) Huh7 cells were transfected with genomic RNAs representing NS5A haplotypes indicated by the single letter-code for the amino acid residues present at positions 28,30, 31 and 93, with genotype consensus amino acid residues in black and RASs shown in red font. Secreted GLuc activity was measured at 6, 24, 48 and 72hr post transfection, with results normalized to activity at 6h (arbitrarily set to 1.0). Data shown represent the mean ± S.E.M. of 3 independent experiments, each with 3 technical replicates. ΔGDD is replicationincompetent H77S RNA carrying a lethal mutation in the active site of the NS5B RNA-dependent RNA polymerase.

Although replicon studies indicate Y93H/N substitutions cause the highest levels of elbasvir resistance in both gt-1a and gt-1b HCV, reported fold-increases in EC50 were only 17 for gt-1b compared to 220–929 for gt-1a (Liu et al., 2015). The capacity of gt-1a virus to sustain greater elbasvir resistance than gt-1b may thus have contributed to higher levels of viremia during treatment in gt-1a than gt-1b infected subjects. Y93H also caused a much greater loss of replicative fitness in the gt-1a reporter virus: the gt-1a MQLH haplotype replicated at only 3% the level of the wild-type MQLY, whereas the gt-1b LRLH haplotype was 41% as fit as LRLY (Fig. 5C). Replicon studies suggest other gt-1b strains may be more negatively impacted by Y93H substitutions (Liu et al., 2015), but these results are consistent with the persistence of gt-1b and not gt-1a Y93H variants post-treatment (Figs. 2, 3, and S3).

Similar studies help to explain the second-site L31M/V substitutions we observed in gt-1a subjects with Q30R (Figs. 2 and S5). Q30R causes a 16-fold increase in the EC50 of gt-1a replicons (Liu et al., 2015), and this substitution was frequent in the gt-1a subjects we studied (Figs. 2 and S5). However, as noted above, Q30R and Y93H/N substitutions emerged in distinct lineages, and were never found together in a single lineage. Consistent with this, while Q30R (MRLY haplotype) caused a only 4.6-fold loss of fitness in gt-1a virus, combining it with Y93H (MRLH haplotype) was lethal and resulted in no detectable replication (Fig. 5C). In contrast, combining Q30R with L31M (which by itself had no impact on gt-1a replication fitness) completely reversed the loss of fitness resulting from Q30R (Fig. 5C, MRMY haplotype) and also boosted elbasvir resistance from 16-fold to 1489-fold (Table S3). The combination of Q30R and L31V was even more resistant (7143-fold, Table S3). These results explain why Q30R and Y93H were never found in the same lineage, and why Q30R emergence was typically associated with second-site L31M substitutions in gt-1a subjects.

4. DISCUSSION

While NS5A inhibitors are an important component of effective DAA combination therapies for both gt-1a and gt-1b HCV infection, treatment outcome is generally superior in gt-1b-infected persons (Lawitz et al., 2012; Liu et al., 2015). This cannot be attributed to differences in drug potency, as reflected in EC50 values, and the reasons for this difference remain uncertain. We show here that gt-1a and gt-1b viruses differ fundamentally in the way in which resistant variants emerge and persist off therapy, and that some of these differences correlate with genotypespecific differences in the impact of RASs on resistance and replicative fitness.

Our study is unique among others that have examined the response of HCV to NS5A inhibitors. The use of primer ID sequencing eliminated errors commonly introduced by PCR as well as sequencing platform artifacts existing in conventional next generation sequencing strategies (Jabara et al., 2011; Keys et al., 2015; Lee et al., 2017; Zhou et al., 2015). It also allowed us to identify linkage between RASs. The method was sufficiently sensitive to accurately detect minority variants with frequencies as low as 0.03% to 0.1% in the viral population (a function of the number of viral genomes actually sequenced). The number of subjects we studied was relatively small, and of necessity limited by the intensity of the analysis. However, the changes we observed in HCV populations during and after elbasvir therapy are likely representative of those occurring generally in gt-1a and gt-1b infected patients, and provide new insight into how resistance to NS5A inhibitors evolves.

In gt-1a infected subjects, we observed two major independent pathways to resistance (Fig. 6). Y93H/C/N substitutions emerged during treatment, but rapidly diminished in frequency off therapy, becoming minor variants 2 months later. Q30R and L31M/V substitutions were also observed early in treatment, but these RASs, either individually or in combination, increased in frequency following cessation of therapy. The dominance of Y93H during treatment may be explained by its far greater resistance to elbasvir (220-fold increase in EC50) compared to Q30R (16-fold) (Liu et al., 2015) (Table S3). On the other hand, the much greater replicative fitness of Q30R versus Y93H (Fig. 5C) explains the expansion of Q30R genomes off treatment (Fig 6). Moroever, in vitro data show that second-site L31M substitutions fully rescue the fitness of virus with Q30R (Fig. 5C), contributing to Q30R persistence and dramatically boosting elbasvir resistance (Table S3). Interestingly, while we detected Q30R at baseline in all 7 gt-1a infected subjects (Fig. 1A), it was dominant in only one (subject 6, Fig. 2B). A review of the Los Alamos database found Q30R present in only 13 of 3018 (0.4%) gt-1a sequences.

Figure 6. Major evolutionary pathways to NS5A resistance in subjects infected with gt-1a and gt-1b HCV infection.

Figure 6.

Open in a new tab

Changes in the NS5A domain I sequence of gt-1a and gt-1b virus are shown in overlapping or independent Venn diagrams during and after treatment with elbasvir according to the single letter amino acid code for residues at positions 28, 30, 31, and 93. The size of each circle is roughly proportionate to their frequencies in the viral population. The emergence and persistence of variants is driven by positive selective forces related to the magnitude of drug resistance during therapy, and relative replicative fitness subsequently. Note that gt-1a virus has two distinct pathways to resistance, while only one was observed in gt-1b infections.

In contrast to gt-1a infected subjects, we observed only a single pathway to resistance in those with gt-1b infection (Fig. 6). Viremia was extremely low during elbasvir treatment of these subjects. The effective size of the replicating viral population is thus likely to have been much smaller in the gt-1b than gt-1a subjects during therapy, most likely because even the most resistant Y93H and L31F gt-1b variants demonstrate only 15- to 17-fold increases in EC50 (Liu et al., 2015). In contrast to gt-1b infected subjects, where rebounding resistant virus populations demonstrated fundamentally altered structures, RASs emerged from multiple preexisting viral linages in gt-1a subjects (Figs. 4A and S6). This suggests that the effective population size on therapy was sufficiently large in these patients to allow for the deterministic selection of RASs. All possible RASs are likely to have been present before the treatment due to the error-prone nature of HCV RNA synthesis, emerging with drug selection pressure when replication is not fully suppressed. In gt-1b, the limited effective population size during therapy may have allowed only the most resistant variants (Y93H pathway) to be selected.

Unfortunately, available crystallographic models of NS5A structure provide little insight into why the evolution of inhibitor resistance differs in gt-1a and gt-1b viruses. There are several domain I structures reported, but none include the amino-terminal 35 amino acids, and thus we do not know how 3 critical residues (positions 28, 30 and 31) are positioned within the molecule. Moreover, models of the gt-1a (Lambert et al., 2014) and gt-1b (Tellinghuisen et al., 2005) structures differ in that gt-1b is a dimer, while gt-1a is a tetramer, with the relative positioning of the monomer subunits different in each. It is not clear which of these models reflects the native conformation of NS5A, and it is possible that the molecule adopts both folds. A complete molecular explanation for the findings we describe here must thus await a better understanding of the NS5A structure and its association with other components of the HCV replication complex.

Supplementary Material

1

Figure S1. Shannon’s entropy change across the sequenced NS5A amino acid positions from baseline to day 28 in gt-1a subjects. The X-axis represents NS5A amimo acid positions; the Yaxis shows the value of the entropy change between day 28 and baseline. Positions with significant changes in entropy are likely to have sustained amino acid substitutions.

2

Figure S2. Shannon’s entropy change across the sequenced NS5A amino acid positions from baseline to day 28 in gt-1b subjects. See legend to Fig. S2.

3

Figure S3. Emergence of RASs of in subjects with gt-1a and gt-1b HCV during and after elbasvir monotherapy. Haplotypes defined by sequence at amino acid positions 28, 30, 31 and 93 (see text) are shown at the bottom of each graph.

4

Figure S4. Phylogenetic trees of sequences obtained at 0 hour and first available data point after the 5-day therapy in gt-1a subjects (60 days in the placebo recipient). Sequences in blue were obtained from the 0-hour sample whereas those in red are from the 7-day sample and 21-day sample for subjects 4 and 9, respectively. Haplotyes defined by amino acid residues at positions 28, 30, 31 and 93 are labeled with each major viral linage. Residues shown in black font represent baseline (consensus sequence) whereas those in color are RASs.

5

Figure S5. Phylogenetic trees of sequences obtained at 0 hour and first available data point after the 5-day therapy in gt-1b subjects (60 days in the placebo recipient). Sequences in blue were obtained from the 0-hour sample whereas those in red are from the 7-day sample and 21-day sample for subjects 4 and 9, respectively. Haplotyes defined by amino acid residues at positions 28, 30, 31 and 93 are labeled with each major viral linage. Residues shown in black font represent baseline (consensus sequence) whereas those in color are RASs.

6

Table S1 Baseline viremia, elbasvir dose, and numbers of template consensus sequences obtained from serum samples.

7

Table S2. Primers used in primer ID library construction.

8

Table S3. Elbasvir sensitivity of gt-1a and gt-1b replicons and genome-length RNAs with NS5A inhibitor resistance-associated substitutions (RASs). s.

HIGHLIGHTS.

  • Primer ID sequencing identifies NS5A inhibitor RASs with unparalleled accuracy and provides novel insight into RAS linkage.

  • NS5A inhibitor resistance emerges along distinct evolutionary pathways in genotype 1a versus 1b virus infections.

  • Y93H substitutions emerge from within multiple independent lineages present in baseline genotype 1a virus populations.

  • Differences in the in vitro replicative fitness of viruses help to explain the persistence of some RASs off therapy.

Acknowledgements

We thank Drs. Christoph Welsch and Joseph Marcotrigiano for stimulating discussions. This work was funded by a research grant from the Investigator-Initiated Studies Program of Merck, Sharp and Dohme (SML) and grants from the National Institutes of Health [T32-AI007151 (SEW), R01-AI095690 (SML), R21-AI115207 (DRM), R56-AI44667 (RS). This work also received infrastructure support from the UNC Center For AIDS Research [NIH award P30 AI50410] and the UNC Lineberger Comprehensive Cancer Center [NIH award P30 CA16068]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Footnotes

Declaration of Interest

SML has received royalties from AbbVie and grant funding from Merck and Gilead Sciences. DRM has received grant funding from AbbVie and Gilead Sciences. RS is a co-inventor on patents related to primer ID sequencing and has received nominal royalties. EA-A and P.I. are employees of Merck, Sharp and Dohme. All other authors have no competing interests.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

5. REFERENCES

  1. AASLD-IDSA, Recommendations for testing, managing, and treating hepatitis C. http://www.hcvguidelines.org.
  2. Barnard R, Chopra A, James I, Blinco J, Watson MW, Jabara CB, Hazuda D, Lemon SM, Mallal S, Gaudieri S, 2016. Primer ID ultra-deep sequencing reveals dynamics of drug resistance-associated variants in breakthrough hepatitis C viruses: relevance to treatment outcome and resistance screening. Antivir Ther 21, 567–577. [DOI] [PubMed] [Google Scholar]
  3. Bartenschlager R, Lohmann V, Penin F, 2013. The molecular and structural basis of advanced antiviral therapy for hepatitis C virus infection. Nat Rev Microbiol 11, 482–496. [DOI] [PubMed] [Google Scholar]
  4. Beard MR, Abell G, Honda M, Carroll A, Gartland M, Clarke B, Suzuki K, Lanford R, Sangar DV, Lemon SM, 1999. An infectious molecular clone of a Japanese genotype 1b hepatitis C virus. Hepatology 30, 316–324. [DOI] [PubMed] [Google Scholar]
  5. Benzine T, Brandt R, Lovell WC, Yamane D, Neddermann P, De Francesco R, Lemon SM, Perelson AS, Ke R, McGivern DR, 2017. NS5A inhibitors unmask differences in functional replicase complex half-life between different hepatitis C virus strains. PLoS Pathog 13, e1006343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Berger C, Romero-Brey I, Radujkovic D, Terreux R, Zayas M, Paul D, Harak C, Hoppe S, Gao M, Penin F, Lohmann V, Bartenschlager R, 2014. Daclatasvir-like inhibitors of NS5A block early biogenesis of hepatitis C virus-induced membranous replication factories, independent of RNA replication. Gastroenterology 147, 1094–1105 e1025. [DOI] [PubMed] [Google Scholar]
  7. Boson B, Denolly S, Turlure F, Chamot C, Dreux M, Cosset FL, 2017. Daclatasvir Prevents Hepatitis C Virus Infectivity by Blocking Transfer of the Viral Genome to Assembly Sites. Gastroenterology 152, 895–907 e814. [DOI] [PubMed] [Google Scholar]
  8. Edgar RC, 2004a. MUSCLE: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinformatics 5, 113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Edgar RC, 2004b. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32, 1792–1797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gao M, Nettles RE, Belema M, Snyder LB, Nguyen VN, Fridell RA, Serrano-Wu MH, Langley DR, Sun JH, O'Boyle DR 2nd, Lemm JA, Wang C, Knipe JO, Chien C, Colonno RJ, Grasela DM, Meanwell NA, Hamann LG, 2010. Chemical genetics strategy identifies an HCV NS5A inhibitor with a potent clinical effect. Nature 465, 96–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gottwein JM, Pham LV, Mikkelsen LS, Ghanem L, Ramirez S, Scheel TKH, Carlsen THR, Bukh J, 2018. Efficacy of NS5A inhibitors against hepatitis C virus genotypes 1–7 and escape variants. Gastroenterology 154, 1435–1448. [DOI] [PubMed] [Google Scholar]
  12. HCV Drug Development Advisory Group, 2012. Clinically relevant HCV drug resistance mutations. Ann Forum Collab HIV Res 14, 1–10. [Google Scholar]
  13. Issur M, Gotte M, 2014. Resistance patterns associated with HCV NS5A inhibitors provide limited insight into drug binding. Viruses 6, 4227–4241. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ivanenkov YA, Aladinskiy VA, Bushkov NA, Ayginin AA, Majouga AG, Ivachtchenko AV, 2017. Small-molecule inhibitors of hepatitis C virus (HCV) non-structural protein 5A (NS5A): a patent review (2010–2015). Expert Opin Ther Pat 27, 401–414. [DOI] [PubMed] [Google Scholar]
  15. Jabara CB, Hu F, Mollan KR, Williford SE, Menezes P, Yang Y, Eron JJ, Fried MW, Hudgens MG, Jones CD, Swanstrom R, Lemon SM, 2014. Hepatitis C Virus (HCV) NS3 sequence diversity and antiviral resistance-associated variant frequency in HCV/HIV coinfection. Antimicrob Agents Chemother 58, 6079–6092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Jabara CB, Jones CD, Roach J, Anderson JA, Swanstrom R, 2011. Accurate sampling and deep sequencing of the HIV-1 protease gene using a Primer ID. Proc Natl Acad Sci U S A 108, 20166–20171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Keys JR, Zhou S, Anderson JA, Eron JJ Jr., Rackoff LA, Jabara C, Swanstrom R, 2015. Primer ID Informs Next-Generation Sequencing Platforms and Reveals Preexisting Drug Resistance Mutations in the HIV-1 Reverse Transcriptase Coding Domain. AIDS Res Hum Retroviruses 31, 658–668. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Korber BT, Kunstman KJ, Patterson BK, Furtado M, McEvilly MM, Levy R, Wolinsky SM, 1994. Genetic differences between blood- and brain-derived viral sequences from human immunodeficiency virus type 1-infected patients: evidence of conserved elements in the V3 region of the envelope protein of brain-derived sequences. J Virol 68, 7467–7481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lambert SM, Langley DR, Garnett JA, Angell R, Hedgethorne K, Meanwell NA, Matthews SJ, 2014. The crystal structure of NS5A domain 1 from genotype 1a reveals new clues to the mechanism of action for dimeric HCV inhibitors. Protein Sci 23, 723–734. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lawitz EJ, Gruener D, Hill JM, Marbury T, Moorehead L, Mathias A, Cheng G, Link JO, Wong KA, Mo H, McHutchison JG, Brainard DM, 2012. A phase 1, randomized, placebo-controlled, 3-day, dose-ranging study of GS-5885, an NS5A inhibitor, in patients with genotype 1 hepatitis C. J Hepatol 57, 24–31. [DOI] [PubMed] [Google Scholar]
  21. Lee SK, Zhou S, Baldoni PL, Spielvogel E, Archin NM, Hudgens MG, Margolis DM, Swanstrom R, 2017. Quantification of the Latent HIV-1 Reservoir Using Ultra Deep Sequencing and Primer ID in a Viral Outgrowth Assay. J Acquir Immune Defic Syndr 74, 221–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Liu R, Curry S, McMonagle P, Yeh WW, Ludmerer SW, Jumes PA, Marshall WL, Kong S, Ingravallo P, Black S, Pak I, DiNubile MJ, Howe AY, 2015. Susceptibilities of genotype 1a, 1b, and 3 hepatitis C virus variants to the NS5A inhibitor elbasvir. Antimicrob Agents Chemother 59, 6922–6929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lontok E, Harrington P, Howe A, Kieffer T, Lennerstrand J, Lenz O, McPhee F, Mo H, Parkin N, Pilot-Matias T, Miller V, 2015. Hepatitis C virus drug resistance-associated substitutions: State of the art summary. Hepatology 62, 1623–1632. [DOI] [PubMed] [Google Scholar]
  24. Mauger DM, Golden M, Yamane D, Williford S, Lemon SM, Martin DP, Weeks KM, 2015. Functionally conserved architecture of hepatitis C virus RNA genomes. Proc Natl Acad Sci U S A 112, 3692–3697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. McGivern DR, Masaki T, Williford S, Ingravallo P, Feng Z, Lahser F, Asante-Appiah E, Neddermann P, De Francesco R, Howe AY, Lemon SM, 2014. Kinetic analyses reveal potent and early blockade of hepatitis C virus assembly by NS5A inhibitors. Gastroenterology 147, 453–462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Ross-Thriepland D, Harris M, 2015. Hepatitis C virus NS5A: enigmatic but still promiscuous 10 years on! J Gen Virol 96, 727–738. [DOI] [PubMed] [Google Scholar]
  27. Shimakami T, Welsch C, Yamane D, McGivern DR, Yi M, Zeuzem S, Lemon SM, 2011. Protease inhibitor-resistant hepatitis C virus mutants with reduced fitness from impaired production of infectious virus. Gastroenterology 140, 667–675. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Smith DB, Bukh J, Kuiken C, Muerhoff AS, Rice CM, Stapleton JT, Simmonds P, 2014. Expanded classification of hepatitis C virus into 7 genotypes and 67 subtypes: updated criteria and genotype assignment web resource. Hepatology 59, 318–327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Tamura K, Nei M, 1993. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol Biol Evol 10, 512–526. [DOI] [PubMed] [Google Scholar]
  30. Tellinghuisen TL, Marcotrigiano J, Rice CM, 2005. Structure of the zinc-binding domain of an essential component of the hepatitis C virus replicase. Nature 435, 374–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Yamane D, McGivern DR, Wauthier E, Yi M, Madden VJ, Welsch C, Antes I, Wen Y, Chugh PE, McGee CE, Widman DG, Misumi I, Bandyopadhyay S, Kim S, Shimakami T, Oikawa T, Whitmire JK, Heise MT, Dittmer DP, Kao CC, Pitson SM, Merrill AH Jr., Reid LM, Lemon SM, 2014. Regulation of the hepatitis C virus RNA replicase by endogenous lipid peroxidation. Nat Med 20, 927–935. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Yi M, Villanueva RA, Thomas DL, Wakita T, Lemon SM, 2006. Production of infectious genotype 1a hepatitis C virus (Hutchinson strain) in cultured human hepatoma cells. Proc. Natl. Acad. Sci. U.S.A 103, 2310–2315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zeuzem S, Mizokami M, Pianko S, Mangia A, Han KH, Martin R, Svarovskaia E, Dvory-Sobol H, Doehle B, Hedskog C, Yun C, Brainard DM, Knox S, McHutchison JG, Miller MD, Mo H, Chuang WL, Jacobson I, Dore GJ, Sulkowski M, 2017. NS5A resistance-associated substitutions in patients with genotype 1 hepatitis C virus: Prevalence and effect on treatment outcome. J Hepatol 66, 910–918. [DOI] [PubMed] [Google Scholar]
  34. Zhou S, Bednar MM, Sturdevant CB, Hauser BM, Swanstrom R, 2016. Deep Sequencing of the HIV-1 env Gene Reveals Discrete X4 Lineages and Linkage Disequilibrium between X4 and R5 Viruses in the V1/V2 and V3 Variable Regions. J Virol 90, 7142–7158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Zhou S, Jones C, Mieczkowski P, Swanstrom R, 2015. Primer ID Validates Template Sampling Depth and Greatly Reduces the Error Rate of Next-Generation Sequencing of HIV-1 Genomic RNA Populations. J Virol 89, 8540–8555. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

Figure S1. Shannon’s entropy change across the sequenced NS5A amino acid positions from baseline to day 28 in gt-1a subjects. The X-axis represents NS5A amimo acid positions; the Yaxis shows the value of the entropy change between day 28 and baseline. Positions with significant changes in entropy are likely to have sustained amino acid substitutions.

NIHMS1503051-supplement-1.tif (140.5KB, tif)
2

Figure S2. Shannon’s entropy change across the sequenced NS5A amino acid positions from baseline to day 28 in gt-1b subjects. See legend to Fig. S2.

NIHMS1503051-supplement-2.tif (97.2KB, tif)
3

Figure S3. Emergence of RASs of in subjects with gt-1a and gt-1b HCV during and after elbasvir monotherapy. Haplotypes defined by sequence at amino acid positions 28, 30, 31 and 93 (see text) are shown at the bottom of each graph.

NIHMS1503051-supplement-3.tif (2.2MB, tif)
4

Figure S4. Phylogenetic trees of sequences obtained at 0 hour and first available data point after the 5-day therapy in gt-1a subjects (60 days in the placebo recipient). Sequences in blue were obtained from the 0-hour sample whereas those in red are from the 7-day sample and 21-day sample for subjects 4 and 9, respectively. Haplotyes defined by amino acid residues at positions 28, 30, 31 and 93 are labeled with each major viral linage. Residues shown in black font represent baseline (consensus sequence) whereas those in color are RASs.

NIHMS1503051-supplement-4.tif (3.1MB, tif)
5

Figure S5. Phylogenetic trees of sequences obtained at 0 hour and first available data point after the 5-day therapy in gt-1b subjects (60 days in the placebo recipient). Sequences in blue were obtained from the 0-hour sample whereas those in red are from the 7-day sample and 21-day sample for subjects 4 and 9, respectively. Haplotyes defined by amino acid residues at positions 28, 30, 31 and 93 are labeled with each major viral linage. Residues shown in black font represent baseline (consensus sequence) whereas those in color are RASs.

NIHMS1503051-supplement-5.tif (1.8MB, tif)
6

Table S1 Baseline viremia, elbasvir dose, and numbers of template consensus sequences obtained from serum samples.

NIHMS1503051-supplement-6.docx (14.3KB, docx)
7

Table S2. Primers used in primer ID library construction.

NIHMS1503051-supplement-7.docx (60.6KB, docx)
8

Table S3. Elbasvir sensitivity of gt-1a and gt-1b replicons and genome-length RNAs with NS5A inhibitor resistance-associated substitutions (RASs). s.

NIHMS1503051-supplement-8.docx (19.8KB, docx)

RESOURCES