Positive Impact of HIV-1 gag Cleavage Site Mutations on the Virological Response to Darunavir Boosted with Ritonavir

Abstract

We assessed the roles of baseline gag and gag-pol cleavage site mutations (CSM) on the virological outcome of a darunavir-based regimen in highly antiretroviral-experienced patients. We showed the association, in multivariate analysis, between the A431V gag CSM and the virological response, defined as a reduction in plasma HIV-1 RNA to <50 copies/ml at month 3 (P = 0.028). Our results suggest that a specific gag CSM might have a role on protease inhibitor susceptibility in an inhibitor-specific manner.

HIV resistance to protease inhibitors (PI) results from the selection and accumulation of mutations in protease. However, other genomic regions, regions other than protease, might be involved in PI resistance, especially the protease gag and gag-pol cleavage sites (CS). Several studies have shown that gag CS mutations (CSM) might influence the virological outcome of a PI-based regimen (1, 3, 4, 6, 10, 13, 18, 26). The gag-pol frameshift signal is composed of a slippery sequence (U UUU UUA) and a downstream secondary RNA hairpin structure, allowing the ribosome to pause and leading to a Gag/Gag-Pol production ratio of between 1:10 and 1:20 (8, 12, 25). The aim of this study was to assess the role of baseline substitutions in the C-terminal region of the Gag protein on the virological outcome of treatment-experienced patients receiving darunavir (DRV) in a salvage regimen. The baseline substitutions in the C-terminal regions of Gag and Gag-Pol protein CS p7/p1-transframe protein (p7/p1-TFP) and p1-TFP/p6^gag-p6^pol were studied. Moreover, we analyzed the RNA folding and stability of the baseline gag-pol frameshift signal by measuring the hairpin free energy.

This study included 153 PI-experienced patients, already described in a previous study (5). All had received an antiretroviral-containing regimen that included DRV boosted with ritonavir in association with nucleoside reverse transcriptase inhibitors, and/or nonnucleoside reverse transcriptase inhibitors, and/or enfuvirtide. The median baseline HIV-1 RNA value was 4.7 log₁₀ copies/ml (interquartile range [IQR] of 4.3 to 5.2). The median baseline CD4 cell count was 142 cells/mm³ (IQR of 28 to 264). The patients previously received a median of 4 PI (IQR of 3 to 5) before initiating a DRV-based regimen. The virological response, defined as a reduction in plasma HIV-1 RNA to <50 copies/ml at month 3 (M3), occurred in 55% of patients. In this study, we focused on 130 patients, whose baseline plasma specimens were still available to perform a genotyping assay. Fifty-seven (44%) patients displayed virological failure, and 73 (56%) displayed a virological response at M3. After viral RNA extraction, a Gag-protease region of 323 bp, including the CS p7/p1 and p1/p6^gag in the gag open reading frame and the CS p7/TFP and TFP/p6^pol in the gag-pol open reading frame were amplified and sequenced as already described (16). We focused on specific substitutions documented in the literature: 431, 436, 437, 441, and 449 residues in the gag and gag-pol open reading frames (1, 4, 7, 11, 13, 16, 18, 21, 23, 24, 26, 29) (Fig. 1 A). Differences in the frequency of baseline amino acids in the sequence compared to those in the HXB2 reference strain for these positions and the presence of insertions or deletions in this region were studied according to the virological response at M3 using Fisher exact tests. To estimate whether the mutations, with P < 0.05 in univariate analysis, were an independent predictor of the virological outcome, we used a multivariate logistic regression model that included the following variables: baseline HIV-1 RNA viral load, baseline CD4 cell count, baseline International AIDS Society (IAS) major PI mutations, as well as enfuvirtide coadministration with DRV-ritonavir. RNA hairpin gag-pol frameshift region folding and stability have been determined using Turner's rules, which measure free energy (RNAsoft CombFold [http://www.rnasoft.ca/cgi-bin/RNAsoft/CombFold/combfold.pl]), and were compared between patients considered responders and nonresponders, using the Wilcoxon rank sum test.

FIG. 1.

(A) Gag and Gag-Pol precursors with amino acid sequences of the Gag C-terminal region. INT, integrase; PR, protease; RT, reverse transcriptase; TFP, transframe protein. (B) Number and percentage of patients with the baseline A431V gag cleavage site mutation compared to the HXB2 reference sequence in the gag open reading frame according to the virological response at month 3.

gag sequences were successfully obtained for all plasma samples. The following variables were associated with the virological response at M3 using univariate analysis: baseline HIV-1 RNA viral load, baseline CD4 cell count, number of baseline IAS major PI mutations, and coprescription of enfuvirtide (Table 1). Among these variables, three variables remained independently associated with the virological response at M3 in the multivariate analysis: DRV score, baseline HIV-1 RNA viral load, and coprescription of enfuvirtide. No association between the occurrence of virological failure at M3 and baseline gag and gag-pol CSM has been shown. Similarly, no association between deletions or insertions in the baseline sequences and virological outcome has been evidenced.

TABLE 1.

Factors associated with virological response as assessed by univariate and multivariate regression logistic analyses

Factora	P valueb by:
	Univariate analysis	Multivariate analysis
DRV score for 14R, 20I, 34Q, 47V, 50V, 54M, 55R, 74P, 76V, and 84V		<0.0001
Baseline HIV-1 RNA (log10 copies/ml)	<0.0001	0.0003
Baseline CD4 cell count (no. of cells/mm3)	0.015	0.294
Baseline IAS major PI mutations (≤4 vs >4)	0.009	0.279
Baseline coprescription of T20 in T20-naïve patients	0.029	0.007
No. of active drugs/GSS	0.143
HIV-1 subtype B vs not subtype B	0.300
A431V (p7/p1 Gag cleavage site)	0.036	0.027

^aDRV, darunavir; GSS, genotypic susceptibility score; IAS, International AIDS Society; PI, protease inhibitor; T20, enfuvirtide.

^bThe P values by Fisher's exact test for univariate analysis and the P values by multivariate logistic regression analysis are shown. The significant values in the multivariate analysis are shown in boldface type.

The baseline A431V gag CSM (p7/p1) was present in 46 of the 73 patients (63%) who had a plasma HIV-1 RNA viral load below 50 copies/ml at M3 and in 24 of the 57 (42%) patients with HIV-1 RNA who had a viral load above 50 copies/ml at M3 (Fig. 1B). Baseline A431V gag CSM was associated with a virological response as assessed by univariate analysis (P = 0.036). This substitution remained independently associated with virological response (P = 0.028) in multivariate analysis (Table 1). No association was found between the baseline hairpin free energy and the occurrence of virological failure at M3.

The roles of CSM in p7/p1-TFP and p1-TFP/p6^gag-p6^pol have been well evaluated in vitro. Their presence has been associated with an increase of Gag processing (4, 7, 9, 18, 21, 23), allowing, in some cases, a partial recovery of the viral replicative capacity (4, 18, 23, 26, 29, 30). These substitutions may enhance the affinity of the CS amino acids for protease, either by increasing Van der Waals forces (9, 14, 18) and/or hydrogen bonds (18) or by improving contacts with the protease active site (4, 27). The presence of gag CSM alone (2, 23, 24) or associated with PI mutations (4, 18) may impact on the level of PI phenotypic resistance (20, 22, 23). Mutations in the gag-pol frameshift may contribute to enhanced PI resistance by raising frameshift efficiency, either by increasing slippage (8, 13) or stability of the gag-pol frameshift hairpin (2, 13).

In vivo, several gag and gag-pol CSM were found to be associated with specific PI resistance mutations (1, 3, 10, 13, 14, 16, 18, 19, 30). The prevalence of CSM increases with the number of major PI resistance mutations (23, 30) in treatment-experienced patients compared to antiretroviral-naïve patients (10, 11, 20, 28).

In our study, the presence of the A431V CSM at baseline was significantly and independently associated with a better virological response at M3 in multivariate analysis. This has never been reported for PI-experienced patients. However, we recently reported in antiretroviral-naïve patients receiving a dual-boosted PI regimen (fosamprenavir or atazanavir both associated with saquinavir) as a first-line regimen without any PI mutations in the protease-coding region a positive impact on the virological outcome of the presence, at baseline, of the D437 substitution in the gag-pol open reading frame (17).

In previous studies, the A431V mutation has been associated with virological failure (28), with an increase in Gag precursor cleavage (7, 9, 21), a rise in the viral replicative capacity (20, 26), and also an increase in the phenotypic resistance of almost all the PI tested (4, 23, 24). Furthermore, a study assessing recombinant viruses that harbor the A431V substitution, added to different PI major mutations by site-directed mutagenesis, reported a decrease or increase in the replicative capacity, as well as fluctuations in the level of PI susceptibility (increased or reduced) according to the PI tested (15). Moreover, in another study conducted in highly experienced patients, baseline gag CSM, including A431V, were not associated with virological outcome (23). However, DRV, the last marketed PI, was not tested in any of these latter studies.

Currently, in patients receiving a DRV-containing regimen, only one study has shown an association between the presence of baseline A431V gag CSM and the selection in the protease coding region of the L76V mutation associated with resistance to DRV and lopinavir (16). In our study, the prevalence of the L76V mutation at baseline was too low to enable assessment of such an association. Moreover, genotypic data were not available at the time of failure.

This is the first study to show the positive impact of a gag CSM on the virological outcome of a DRV-containing regimen in highly antiretroviral-experienced patients. DRV displays a potent antiviral activity and has an original chemical structure leading to a specific binding profile with the protease active site which might explain the activity that it maintains on mutant viruses and its genotypic resistance profile. Thus, we could hypothesize that interactions between the enzyme and substrate might impact the nature and position of gag CSM involved in DRV susceptibility. In the literature, A431V has been reported as associated with virological outcome. In contrast to previous studies, our study was carried out in highly antiretroviral-experienced patients that harbor viruses with a median number of 4 major PI resistance mutations and who previously received a median number of 4 PI before initiating the DRV-based regimen (5). Thus, it is difficult to distinguish between whether the A431V mutation had been selected by the previous PI-based regimens or was present in archived PI-resistant variants not detected by bulk sequencing before any PI-based regimen. Further site-directed mutagenesis experiments assessing recombinant viruses harboring the A431V substitution in the presence or absence of DRV resistance-associated mutations would be of interest in order to evaluate the role of this mutation on the virological response. Our results have raised the hypothesis that a specific gag CSM might not have the same impact on virological outcome, according to the PI used as well as the presence or absence of PI-resistance mutations, and could have a direct impact on PI susceptibility in an inhibitor-specific manner.