Comprehensive evaluation of hepatitis B virus reverse transcriptase substitutions associated with entecavir resistance

Authors


  • Potential conflict of interest: At the time these studies were conducted, all authors were employees and stockholders of Bristol-Myers Squibb Co.

Abstract

Virologic resistance emerging during entecavir (ETV) therapy for hepatitis B virus (HBV) requires three substitutions in the viral reverse transcriptase (RT), signifying a high barrier to resistance. Two of these substitutions are associated with lamivudine resistance (LVDr) in the tyrosine-methionine-aspartate-aspartate (YMDD) motif (rtM204V and rtL180M), whereas the other occurs at one or more positions specifically associated with ETV resistance (ETVr): rtT184, rtS202, or rtM250. Although a variety of substitutions at these primary ETVr positions arise during ETV therapy, only a subset give rise to clinical virologic breakthrough. To determine the phenotypic impact of observed clinical and potential new ETVr substitutions, a comprehensive panel of clones containing every possible amino acid at the three primary ETVr positions in LVDr HBV was constructed and analyzed in vitro. A range of replication capacities was observed for the panel, but none of the mutations rescued replication of the LVDr mutant to the wild-type level. More clones with residue rtS202 substitutions were severely impaired than those at rtT184 or rtM250. A wide variety of ETV susceptibilities was observed, ranging from approximately eight-fold (no increase over the LVDr parent) to greater than 400-fold over the wild-type. A correlation was identified between clinically observed substitutions and those displaying higher in vitro replication and resistance, especially those from virologic breakthrough patients. Conclusion: The high number of tolerated and resistant ETVr substitutions is consistent with models predicting that the mechanism for ETVr is through enhancement of LVDr changes in the RT deoxyribonucleotide triphosphate (dNTP)-binding pocket. (HEPATOLOGY 2008.)

The genetic changes that impart resistance to the four approved nucleoside analogs used for the treatment of chronic hepatitis B virus (HBV) have been described (Table 1). Lamivudine (LVD) resistance (LVDr) arises from substitutions of the methionine (rtM204) to valine (rtM204V), isoleucine (rtM204I), or rarely serine (rtM204S) in the tyrosine-methionine-aspartate-aspartate (YMDD) motif of the HBV reverse transcriptase (RT). The rtM204V change is always accompanied by an rtL180M growth adaptive substitution,1 whereas the rtM204I substitution is found with or without rtL180M. Additionally, the rtV173L substitution has been found in some LVDr isolates and increases HBV replication capacity in vitro.2 Resistance to telbivudine also arises from changes of the YMDD motif rtM204.3 In contrast, resistance to adefovir (ADV) arises through rtN236T or rtA181V changes, whereas recent studies indicate that the rtA181T change often associated with ADV is more likely related to LVD therapy.4, 5

Table 1. HBV Antiviral Resistance Substitutions
AntiviralHBV RTHBsAg*Reference
  • *

    ETVr HBsAg changes listed are in addition to those associated with the LVDr changes.

  • Only found in a single patient, in combination with rtT184G.17

 rtM204V + rtL180MsI195M1, 16
Lamivudine, TelbivudinertM204I (ATT codon)sW196L 
rtM204I (ATA codon)sW196stop 
 rtV173LsE164D8
AdefovirrtA181VsL173F7
rtN236TNo change 
Entecavir rtT184Anone 
  rtT184CsL175F, sL176V 
  rtT184FsL175F 
  rtT184GsL176V 
  rtT184Inone 
  rtT184LsL175F 
 rtM204V+rtL180M plus:rtT184MsL176stop19
  rtT184SsL175F, sL176V 
  rtS202Cnone 
  rtS202Gnone 
  rtS202IsV194F 
  rtM250Vnone 
  rtM250Lnone 
  rtT184Inone
 rtM204I plus:rtT184SsL175F, sL176V19
 rtM250Inone
  rtM250Lnone 

Entecavir (ETV), a novel deoxyguanosine analog, has demonstrated efficacy for treating HBV patients infected with wild-type or LVDr HBV.6–9 Comprehensive resistance monitoring during clinical studies has revealed less than 1% viral resistance in nucleoside-treatment-naive (nucleoside-naive) patients through four years of therapy, reflecting a high genetic barrier to resistance that likely stems from potency in suppressing HBV DNA replication as well as a need for three HBV RT substitutions for high-level ETV resistance (ETVr).10 These three substitutions include primary LVDr substitutions, rtM204V and rtL180M, plus an additional change at rtT184, rtS202, or rtM250.11 This is in contrast to resistance to LVD and ADV, where changes in only one or two residues at primary resistance positions result in virologic breakthrough. However, in ETV-treated patients infected with LVDr HBV, the frequency of resistance is increased because ETV is somewhat less potent (approximately eight-fold relative to wild-type) in this population and because only one additional RT substitution is required for breakthrough. In addition, ETVr substitutions have been found to exist in a small proportion of LVDr HBV before ETV therapy, suggesting that LVD therapy can select for primary ETVr substitutions.11

In patients with LVDr HBV, a number of ETVr substitutions can emerge at positions rtT184, rtS202, and rtM250, although not all confer high-level resistance12 or result in subsequent virologic breakthrough11 (Table 1). In addition, rtI169 substitutions have occasionally been associated with ETVr, apparently functioning as growth adaptive or compensatory changes associated with the primary resistance changes noted.13 Furthermore, only viruses with ETVr changes in the LVDr HBV background rtM204V and rtL180M, and not rtM204I, display significant reductions in ETV susceptibility.12 Some of the ETVr changes at rtT184 and rtS202G also result in substitutions in the overlapping HBV surface antigen gene (Table 1). Those for ADVr rtN236T and the ETVr rtM250 position substitutions, however, occur after the hepatitis B surface antigen (HBsAg) open reading frame and do not result in HBsAg changes.

Not all patients with ETVr substitutions show virologic breakthrough or primary nonresponsiveness, and some show virologic breakthrough only after further changes at ETVr positions. Moreover, patient isolates with different ETVr substitutions show varied levels of ETV susceptibility. To explore the basis for the substitutions found, especially those in breakthrough isolates, and to define potential new ETVr substitutions that might arise, we tested a panel of LVDr HBV (rtM204V and rtL180M) containing every possible amino acid residue at ETVr positions rtT184, rtS202, or rtM250. The laboratory clones associated with greater replication capacity and ETVr correlated with isolates derived from patients participating in ETV clinical studies.

Abbreviations

ADV, adefovir; EC50, concentration of drug required to reduce viral replication by 50%; ETV, entecavir; ETVr, entecavir resistant, entecavir resistance; HBV, hepatitis B virus; LVD, lamivudine; LVDr, lamivudine resistant, lamivudine resistance; RT, reverse transcriptase; YMDD, tyrosine-methionine-aspartate-aspartate;

Materials and Methods

HBV Expression Constructs.

Site-directed mutagenesis (Stratagene Quick change kit) was used to generate the various HBVs by modifying a plasmid13 encoding wild-type genotype D, ayw serotype HBV,14 (NCBI accession number U95551), similar to and derived from pCMV-HBV.15

HBV Phenotypic Susceptibility Assay.

Cell culture susceptibility assays were performed by transfection of HepG2 hepatoma cells with HBV expression plasmids in the presence of a titration of antiviral agents, followed by the quantification of replicated, immunocaptured HBV using DNA dot-blot hybridization, as described.13 HBV nucleocapsids were captured from detergent-treated culture media for released virus, or from intracellular lysates for intracellular replicated HBV. Intracellular lysates were prepared by incubating cells in 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.8% wt/vol Nonidet-P40 for 1 hour before capturing with anti-HBV core rabbit polyclonal antibody. ETV was prepared at Bristol-Myers Squibb. ETV dose–response curves were analyzed and average EC50 calculations from duplicates performed using Xfit for Excel (ID Business Solutions Ltd.).

HBV Core Protein Enzyme-Linked Immunosorbent Assay.

An enzyme-linked immunosorbent assay for intracellular levels of HBV core protein was used to determine the consistency of the transfection efficiencies of the HBV mutant plasmids. HepG2 cells, transfected with HBV expression plasmids as described, were incubated with Tris-buffered saline (50 mM Tris [pH 8.0], 150 mM NaCl) containing 0.8% Nonidet-P40 for 1 hour at room temperature. Lysates were clarified by centrifugation at 183 ×g for 5 minutes and used immediately or stored at −20°C. Immulon 4 HBX enzyme-linked immunosorbent assay plates (Thermo Electron) were coated with 500 ng/well anti-core monoclonal antibody (Virogen) in 50 mM NaHCO3 (pH 9.6) at 4°C overnight. Plates were washed 3 times with phosphate-buffered saline (PBS) containing 0.05% (vol/vol) Tween 20, then blocked with Superblock in PBS (Pierce) for 1 hour at room temperature. Plates were washed with PBS-Tween 20, and clarified cell lysates prepared as noted were added, followed by a 2-hour incubation at room temperature. Plates were then washed three times with PBS-Tween 20 and incubated with anti-core rabbit polyclonal antibody (Dako) diluted 1:1,000 in Superblock containing 0.05% (vol/vol) Tween 20 for 1 hour at room temperature. After three washes with PBS-Tween 20, horseradish peroxidase–coupled anti-rabbit polyclonal antibody diluted 1:3,000 in Superblock-Tween 20 was added and incubated a further 1 hour at room temperature. The plates were then washed six times with PBS-Tween 20. Then, 3,3′,5,5′ tetramethylbenzidine substrate (Pierce) was added and color development stopped after 5 to 10 minutes by addition of 1N HCl. Quantification was performed using a Spectra Max 250 microplate reader (Molecular Devices) at 450 nm. Purifed HBV core antigen (Virogen) was serial diluted in Tris-buffered saline–0.8% Nonidet-P40 (range, 0.5-10,000 pg/well final concentration) and used to create concentration standards for final quantification of core protein in cell lysates.

HBV Clinical Isolates.

The ETVr substitutions identified in clinical isolates were the result of population sequencing of amplified HBV RT from patient serum samples, as described,11 and updated with the results after 4 years of ETV therapy.10 Isolates with mixtures in the population sequence that obscured the identification of residues at ETVr positions rtT184, rtS202, or rtM250 were cloned and subsequently identified by sequencing of individual clones.

Results

Panel of ETVr Substitutions.

To evaluate the ETVr substitutions that could potentially emerge in patients, a panel of mutants containing all of the possible amino acid substitutions at the three primary ETVr residues was constructed by site-directed mutagenesis of a genotype D HBV clone containing LVDr substitutions rtM204V and rtL180M. Because poor replication capacity could preclude the measurement of susceptibility levels in culture, all clones included the rtV173L substitution reported by Delaney et al.2 to enhance HBV replication. In agreement with those studies, we found that viruses encoding the rtV173L along with LVDr substitutions replicated at a level of 66% of wild-type HBV, whereas the HBV with only the LVDr substitutions replicated at levels that were only 31% of wild-type (data not shown). Several different nucleotide changes at ETVr codons were required to create every possible residue change, including those that require one, two, and three base changes from the wild-type. The sV175F, sL176V, sL176stop, and sV194F HBsAg changes found in clinical isolates are represented within the engineered ETVr panel (Supplementary Table 1). In addition, several other HBsAg changes resulted from ETVr substitutions that have not been observed clinically. The rtM250 residue does not overlap the HBsAg gene, and consequently these codon changes did not result in changes to the HBsAg.

Phenotypic Analysis of ETVr HBV Panel.

Several phenotypic assessments of the ETVr mutant panel were performed in culture. First, the intracellular expression of HBV core protein was assessed to verify the transfection efficiency and expression of each clone. Second, the level of replicated intracellular HBV DNA found within HBV nucleocapsids was determined as a measure of the replication capacity of the clones. Third, the extracellular replicated and released HBV DNA found in secreted virions was assessed. Lastly, the virion and nucleocapsid extracellular HBV DNA in the presence of a titration of ETV was used to assess the ETV susceptibility of the various clones, which was expressed as the effective concentration that resulted in 50% inhibition of HBV DNA production (ETV EC50). Extracellular HBV could not be consistently, quantitatively immunocaptured with anti-HBsAg antibodies, presumably because of an overabundance of subviral particles released from cells. Therefore, anti-HBV core antibodies were used to quantitatively capture detergent-treated nucleocapsids.

Replication Capacity and Secretion of HBV Mutants.

The level of replicated HBV DNA associated with each clone in the panel was assessed by quantification of nucleocapsid-associated HBV DNA. Both intracellular and extracellular assessments were made because changes in the HBV polymerase residues could affect the sequence of the overlapping HBsAg that is involved in virion secretion. In addition, the levels of HBV core antigen produced within cells was determined to ensure that the effects seen were not the result of altered expression of the various plasmids or inconsistencies in transfection efficiency. The results of the panel of rtT184, rtS202, and rtM250 substitutions in LVDr HBV are shown in Fig. 1. HBV core protein expression was relatively consistent for the various clones; however, the replication capacities of the various HBV mutants varied widely. For the rtT184 panel, most residues yielded replication-competent HBV, although HBV DNA levels were still below that of the wild-type, which displayed 187% and 205% of the LVDr intracellular and extracellular controls, respectively (Fig. 1A). Although many replicated as well as the LVDr virus control, reduced replication was observed for charged residues rtT184R, K, H, and D, and to a lesser degree for E. Aromatic residue substitutions rtT184W and Y, as well as P, also replicated poorly. For all but one mutant, the levels of extracellular and intracellular HBV DNA did not vary appreciably; suggesting that changes in the overlapping HBsAg did not affect secretion. The exception was the rtT184M substitution, which causes a stop codon in the overlapping HBsAg gene and resulted in an increased level of extracellular HBV DNA relative to intracellular. It has been noted that HBsAg deletions can cause increased secretion of naked HBV core particles.16 Because cell culture systems involve secretion of both enveloped and unenveloped nucleocapsids, we were unable to differentiate between the two forms associated with the rtT184M substitution.

Figure 1.

The cell culture replication capacity of LVDr HBV containing various ETVr substitutions. Intracellular and extracellular nucleocapsid-associated HBV DNA levels are shown as a fraction of the LVDr clone analyzed in parallel (mean ± the standard deviation of at least three independent experiments). The intracellular and extracellular HBV DNA levels of the wild-type were 187% and 205% that of the LVDr control, respectively. Intracellular HBV core levels are shown in percent wild-type. (A) rtT184 substitutions. (B) rtS202 substitutions. (C) rtM250 substitutions.

Similar results were found with the panel of rtS202 substituted LVDr viruses with respect to core protein expression and levels of intracellular and extracellular HBV nucleocapsid DNA (Fig. 1B). Notably, most substitutions in the rtS202 panel resulted in substantially lower replication, often resulting in levels that were unable to be accurately measured for ETV susceptibility (see below). The clones that yielded replication-competent virus contained both large (for example, F, Y) and small (for example, G, A, C) residues, whereas charged residues (D, E, K, R) severely restricted replication capacity. In some cases, very similar residues such as I and V and L displayed markedly different properties. These differences did not appear to be related to the overlapping HBsAg changes because both the I and L substitutions caused the same sV194Y HBsAg change.

The replication capacities of LVDr HBV with various rtM250 substitutions also displayed widely variant replication levels; however, most did replicate to detectable levels. Residues both larger and smaller than the wild-type methionine replicated reasonably, whereas residues that were basic (rtM250 K and R) or rtM250P, which can induce a structural turn, displayed very poor replication. Because the rtM250 substitutions do not overlap the HBsAg coding region, it was not surprising that intracellular and extracellular HBV DNA levels were similar in all cases.

No rtT184, rtS202, or rtM250 substitutions in LVDr HBV exhibited substantially increased replication above the LVDr virus control, and all were impaired relative to the wild type. However, to ensure that ETVr did not increase replication at lower, physiological ETV levels, replication was assessed at 15 nM ETV (1/2 maximum concentration [Cmax] levels17) to provide selective pressure at a level achieved in patients. Replication was not increased by the presence of ETV (Table 2), suggesting that the mechanism for ETVr was not through increased fitness. These results agree with more definitive in vitro enzyme studies proving that the mechanism for ETVr is through decreased binding of ETV-triphosphate to the polymerase active site.18

Table 2. Replication of Mutants in the Presence of ETV
 HBV DNA Replication (relative to LVDr)*
Amino Acid ChangeNo ETV15 nM ETVRatio
  • *

    Extracellular nucleocapsid-associated HBV DNA levels normalized to the LVDr control analyzed in parallel (mean of at least three independent experiments). Experiments were performed in the presence or absence of 15 nM ETV.

  • Ratio of replication level in the presence of ETV compared to the no-ETV control.

rtT184A1.01.11.1
 C0.81.01.3
 F0.70.81.1
 G1.21.31.1
 I0.70.81.2
 L1.11.31.2
 S1.71.81.0
rtS202C1.01.11.1
 G1.01.41.3
rtM250L1.01.31.3
 V0.91.21.3

Phenotypic Analysis.

ETV susceptibility (EC50) was determined for all clones (Fig. 2). We were not able to accurately measure the EC50 of clones that replicated and released very low levels of extracellular HBV DNA (generally <10% of LVDr control), and these were therefore excluded from the data analysis. Additionally, ETV susceptibilities for recombinant clones containing RTs from patient isolates with ETVr substitutions were compared in parallel (Fig. 2). The average EC50 values of the LVDr parents containing the rtV173L mutation and the wild-type were 40.3 ± 18 nM and 6.8 ± 4.0 nM, respectively, resulting in a decrease in EC50 compared with wild-type, similar to that of clones lacking the rtV173L change.

Figure 2.

ETV susceptibility of LVDr HBV with various ETVr substitutions. ETV EC50 values from at least three independent experiments are shown for the laboratory site-directed mutants (black bars) or patient isolates (gray bars) LVDr HBVs with rtM204V + rtL180M substitutions. The EC50 of the LVDr laboratory clone without ETVr substitutions is indicated by the horizontal line and was 40.3 ± 17.8 nM (n = 21), whereas that of the wild-type in parallel was 6.8 ± 4.0 nM (n = 20). Error bars indicate standard deviation for the site-directed clones or the range for patient isolates (isolates without error bars were from a single patient only). The substitutions that resulted in extraordinarily low levels of extracellular HBV DNA did not yield reasonable dose–response curves and are shown without EC50 values. Substitutions labeled with an asterisk characteristically lead to virological breakthrough.

Substitutions at rtT184.

ETV susceptibility varied widely for the panel of LVDr clones with rtT184 substitutions (Fig. 2A), ranging from greater than 1,000 nM to levels similar to those of the LVDr parent. Various amino acids could be substituted at position rtT184, including large, small, acidic, and basic residues, and there was no correlation of resistance levels with replication capacities. As discussed, viruses with substitutions at rtT184P, R, and K were impaired to a degree that prevented reliable susceptibility levels assessment. Similar to the laboratory-constructed mutants, patient isolates with rtT184 substitutions also showed a wide ETV susceptibility range, but the subset of residues that emerge during ETV therapy was more limited. Significantly, there was a close correlation between the resistance levels of the laboratory mutants and the clinical isolates. In general, isolates that displayed the lowest levels of ETV resistance (rtT184S, I, C, and G) did not result in virological breakthrough without being combined with other ETVr residues.11 Patient isolates were also found with rtT184 substitutions in the LVDr rtM204I background, but these were limited to rtT184 S or I substitutions and yielded average ETV EC50 values of 39 and 43 nM, respectively. However, these combinations did not occur in high numbers and were not found in patients experiencing virological breakthrough.

Substitutions at rtS202.

Of the rtS202 substituted clones that replicated to detectable levels, a range of ETV susceptibilities were found, ranging from greater than 3000 nM EC50 to the levels seen with the wild-type rtS202 in LVDr HBV (Fig. 2B). Unlike the ETVr rtT184 substitutions, the single rtS202 substitutions found in clinical HBV isolates to date are restricted to rtS202C and G. Although rtS202G changes were predominant, a single isolate contained the rtS202C change. One other patient isolate included a virus with an rtS202I change, but it was only found in combination with an additional rtT184G substitution.12 Recombinant HBV studies showed that the isolate with the rtS202I change in LVDr HBV alone had severely impaired replication, which was partially restored by rtT184G.13 Of the rtS202G and C substitutions found outside of combinations, only rtS202G resulted in breakthrough.11

Substitutions at rtM250.

ETV susceptibilities for the rtM250 substituted clones were also generally decreased relative to the LVDr control (Fig. 2C), the exceptions being large residues rtM250F, Y and W, which had little effect on ETV susceptibility. Very few rtM250 changes were found in clinical isolates. ETVr rtM250V and rtM250L changes were found in clinical isolates with the rtM204V and rtL180M LVDr backbone, yielding high levels of resistance. Whereas the rtM250V substitution in LVDr HBV led to virological breakthrough, the single patient involved was treated with the lower 0.5-mg ETV dose rather than the approved 1.0-mg dose.12 Other patients have exhibited virological breakthrough with rtM250 V or L changes, but only when found with other isolates with more prevalent rtT184 or rtS202 substitutions. Substitutions rtM250L and I have also been found in patients with the rtM204I LVDr substitution, but the levels of resistance for these clones is relatively low (average ETV EC50 of 121 and 48 nM, respectively).

Correlation Between In Vitro and Clinical Profiles.

We surmised that a correlation may exist between the phenotypic aspects identified in vitro and the repertoire and clinical outcome of the substitutions found in patients. Therefore, both the intracellular replication capacity and ETV susceptibility for each laboratory clone that was phenotyped were compared (Fig. 3). The analysis revealed that HBV with substitutions that exhibited a combination of high replication capacity and reduced ETV susceptibility correlated with substitutions that emerge during ETV therapy, with a few exceptions. Replication capacities of all clones with ETVr substitutions were impaired relative to the wild-type with those found clinically generally replicating at levels that were similar to the LVDr virus assayed in parallel. A wide range of ETV susceptibilities were observed; however, those that were found clinically had ETV EC50 values of 100 nM or higher. Those that had the highest ETV EC50 levels (highest resistance) were also found among patients with virological breakthrough. The few exceptions to these trends included the rtT184Q substitution that was not found clinically but had replication levels 61% of the wild-type and an ETV EC50 of 341 nM. The rtT184Q caused an sL176K substitution in the overlapping HBsAg that has not been found in any other clinical isolate. Also, rtS202A was an exception in that it had relatively high replication (71% of wild-type) and ETV resistance (EC50 = 263 nM). Whereas the rtS202G and rtS202C isolates did not change the overlapping HBsAg, the rtS202A caused a sV194H change. The rtM250 substitutions that had high replication and resistance but were not found clinically were the rtM250I and rtM250C changes. None of the rtM250 changes resulted in HBsAg substitutions. The rtM250I substitution has been observed, but only in isolates with the rtM204I and not the rtM204V + rtL180M LVDr backbone, whereas the rtM250C change has not been seen clinically. Additionally, as discussed, the rtM250L change in the rtM204V + rtL180M LVDr backbone has been observed clinically but was not found in the absence of viruses with other ETVr substitutions; therefore, it cannot be unequivocally associated with virological breakthrough. Altogether, the results of this combined analysis suggest that those substitutions found clinically generally have higher levels of replication and resistance, with those resulting in virological breakthrough having the highest levels of resistance.

Figure 3.

Correlation between phenotyping and clinical results. Results of individual clones in Figs. 1 and 2 are plotted according to replication capacity (intracellular HBV DNA relative to rtM204V and rtL180M control) and resistance (ETV EC50 in log nM). Letters within symbols indicate substituted residues for rtT184 (squares), rtS202 (diamonds), and rtM250 (circles). Filled symbols indicate substitutions that were found in patient isolates. Underlined residues indicate substitutions that have characteristically been found among patient isolates at the time of virological breakthrough. The results for the control wild-type HBV with rtV173L and the LVDr parent clone (rtM204V, rtL180M, and rtV173L) are shown as WT and LVDr, respectively. The dashed gray lines indicate the approximate cutoffs of replication capacity and resistance.

The number of nucleotide changes required for particular substitutions could also be a potential basis for a restricted repertoire of clinically observed substitutions. The rtT184 codon seen in patient isolates is always ACT. Substitutions at this position which are the result of a single nucleotide change in the codons are A, I, and S; substitutions C, F, G, L, and M require two nucleotide changes. The rtT184Q substitution can be achieved using one of two codons, CAA or CAG, both of which would result from 3 nucleotide changes from the wild-type. This could explain why rtT184Q is not seen clinically, despite high levels of replication and resistance. Two codons are seen in the clinic encoding rtS202; most patient isolates have AGT as the rtS202 codon, but AGC is also observed. Only rtS202I causes an HBsAg substitution (sV194F); rtS202C or G do not result in any HBsAg substitution. All substitutions observed in patient isolates at rtS202 are the result of a single nucleotide substitution. Based on the two clinically observed codons at this position, rtS202A must be generated by mutating the first two nucleotides of the codon; thus, using the codon GCT would result in an sV194L substitution, and rtS202A using GCC would result in an sV194P substitution.

All substitutions found clinically require codons with either one or two nucleotide changes from the wild-type. No substitutions requiring three changes were found in clinical isolates. However, there were many amino acid substitutions requiring only one or two nucleotide changes from the wild-type that were not found clinically. Therefore, the number of nucleotide changes was not a significant factor in determining whether a substitution emerged in patients.

Discussion

In contrast to the other approved HBV antiviral agents (LVD, ADV, and telbivudine), multiple substitutions (rtM204V and rtL180M together with a substitution at rtT184, rtS202, or rtM250) are required for ETV resistance and resistance-related virological breakthrough to occur. However, ETV retains in vitro and in vivo efficacy against LVDr HBV and, molecular modeling has predicted that ETV retains access to a unique hydrophobic pocket in the HBV RT nucleotide binding site present in the LVDr RT.18 Molecular models of ETVr further suggest that additional changes at rtT184, rtS202, or rtM250 decrease access to the binding pocket in LVDr HBV, resulting in phenotypic resistance.20

The rtT184 mutant panel yielded more viruses that replicated (>10% of LVDr HBV) as compared with the rtS202 substitutions. Molecular modeling suggests that these residues adjoin each other, and that hydrogen bonding between them stabilizes the position of the rtS202-containing YMDD loop.20 Thus, large or small substitutions that sterically affect the YMDD positioning or disrupt the hydrogen bonding should cause similar affects. We hypothesize that an additional hydrogen bond between the rtS202 backbone residue and the valine at rtM204 maintains the local configuration or curvature of the tip of the YMDD loop, which may be important for replication.18 This would explain why the bulk of rtS202 substitutions have substantial negative impacts on viral replication. Nevertheless, there are far more clones with rtS202 substitutions that display reduced ETV susceptibility than have been found in HBV from patients on ETV therapy (Fig. 2). Perhaps the lack of rtS020 ETVr changes found in clinical isolates is attributable to changes to the overlapping HBsAg gene, resulting in better serological recognition or by impairing progeny HBV assembly, secretion, or infection of cells. However, HBsAg could not be the sole reason why some rtS202 substitutions are found clinically and others are not because some share the same HBsAg substitutions. Furthermore, although many rtM250 substitutions replicated and displayed substantially reduced ETV susceptibility, very few appeared in clinical isolates. Mechanistic studies and molecular modeling suggest that rtM250, in the primer grip region of HBV polymerase.1 may reposition the portion of the nucleotide-binding pocket that comprises the primer template. Therefore, they may be more sensitive to change than our current understanding of the other ETVr residues. In contrast to the rtS202 changes, rtM250 substitutions do not overlap with or cause changes in the HBsAg. Therefore, factors aside from replication, ETV susceptibility, and overlapping HBsAg changes must play a role in restricting those substitutions that are observed clinically. For example, the ability of the variants to infect cells and spread within the liver cannot be measured using current cell culture assays.

Some patients have developed primary ETVr substitutions while receiving LVD therapy, before treatment with ETV.11, 13 These have been restricted, however, to substitutions that do not have a great impact on ETV susceptibility, either the rtT184S or I change, the rtS202C change, or the rtM250I or L change in an rtM204I LVDr background. After ETV therapy, however, there has been a different and more varied repertoire of ETVr substitutions found clinically, with varying levels of replication and resistance. This finding supports the model in which ETVr substitutions act through further augmentation of LVDr changes in HBV RT with respect to the nucleotide-binding pocket, rather than rtT184, rtS202, or rtM250 substitutions acting directly on the pocket. This is consistent with the finding that clinical ETVr substitutions recombined into wild-type HBV (without LVDr changes) do not substantially affect HBV growth or ETV susceptibility.13 This also explains why ETVr substitutions are not found in HBV in the absence of LVDr changes.

Our results using enzyme kinetics18 and replication capacity, as well as those of Tenney et al.,13 show that phenotypic resistance to ETV results from reduced ETV-triphosphate binding to ETVr HBV RT and contradicts the hypothesis that it is the result of overcoming a replication deficit.21 The findings here suggest that ETVr HBV were still growth deficient relative to wild-type HBV, thus explaining why they often constitute only a portion of a patient's quasispecies isolates. Additional changes in HBV RT have been found to arise subsequent to or simultaneous with the primary ETVr residues, most notably various substitutions at residue rtI169, as well as a handful of other novel substitutions13 (and unpublished results). These changes are in addition to the primary ETVr changes and likely the result of selection for higher replication capacities and may be somewhat unique to ETV, similar to the adaptive substitutions at positions rtV1732 and rtL8022 that arise in some patient isolates in response to LVD therapy. Our ongoing analysis of cloned isolates from patient quasispecies has revealed complex mixtures of isolates and various ETVr substitutions,12 similar to the findings with LVD.23

In summary, recombinant HBV containing RT substitutions rtT184A, F, I, L, M, S, rtS202A, C, G, or rtM250C, I, L,V in a background with LVDr (rtM204V and rtL180M) exhibited a higher relative level of phenotypic resistance to ETV in cell culture. In addition, although replication of these mutants was impaired relative to the wild-type, the HBV DNA levels were similar to that of LVDr virus. Together, these characteristics generally correlated with those of isolates found in patients on ETV therapy, and more specifically, those experiencing virological breakthrough. These results support the role of these substitutions as important clinical markers of ETV resistance.

Acknowledgements

The authors acknowledge our Bristol-Myers Squibb colleagues Patricia Poundstone for excellent nucleotide sequencing, Mary Jane Plym for some plasmids, and Bruce Kreter for comments on the manuscript. We also thank Dr. Steven Goff and Dr. George Acs for plasmids pCMV-HBV and pTHBV-1, respectively.

Ancillary