Structure-Analysis of the HIV-1 Integrase Y143C/R Raltegravir Resistance Mutation in Association with the Secondary Mutation T97A^,

Abstract

The HIV-1 integrase (IN) mutations Y143C/R are known as raltegravir (RAL) primary resistance mutations. In a previous study (S. Reigadas et al., PLoS One 5:e10311, 2010), we investigated the genetic pathway and the dynamics of emergence of the Y143C/R mutations in three patients failing RAL-containing regimens. In these patients, the Y143C/R mutation was associated with the T97A mutation. The aim of the present biochemical and molecular studies in vitro was to evaluate whether the secondary mutation, T97A, associated with the Y143C/R mutation could increase the level of resistance to RAL and impact IN activities. Site-directed mutagenesis experiments were performed with expression vectors harboring the region of the pol gene coding for IN. With a 3′-end processing assay, the 50% inhibitory concentrations (IC₅₀) were 1.2 μM, 1.2 μM, 2.4 μM (fold change [FC], 2), and 20 μM (FC, 16.7) for IN wild type (WT), the IN T97A mutation, the IN Y143C/T97A mutation, and the IN Y143R/T97A mutation, respectively. FCs of 18 and 100 were observed with the strand transfer assay for IN Y143C/T97A and Y143R/T97A mutations, with IC₅₀ of 0.625 μM and 2.5 μM, respectively. In the strand transfer assay, the IN Y143C or R mutation combined with the secondary mutation T97A severely impaired susceptibility to RAL compared to results with the IN Y143C or R mutation alone. Assays without RAL suggested that the T97A mutation could rescue the catalytic activity which was impaired by the presence of the Y143C/R mutation. The combination of the T97A mutation with the primary RAL resistance mutations Y143C/R strongly reduces the susceptibility to RAL and rescues the catalytic defect due to the Y143C/R mutation. This result indicates that the emergence of the Y143C/R/T97A double-mutation pattern in patients is a signature of a high resistance level.

INTRODUCTION

Integrase (IN) is required in vivo for the integration of reverse-transcribed viral DNA within genomic DNA. HIV-1 IN is a 32-kDa protein with three different domains (2). The catalytic core contains the catalytic triad DDE (and, more rarely, DDD, DDH, or DED), which is the signature of enzymes belonging to the polynucleotidyl transferase family (27, 39). IN catalyzes two reactions: 3′-end processing (3′-P) and strand transfer (ST). 3′-end processing prepares both ends of the proviral DNA for integration, and final strand transfer joins the viral and nicked chromosomal DNAs (12). As this step is crucial for viral replication, IN represents an important target that has clinical relevance for treating HIV infection and preventing AIDS (30, 35).

Two classes of inhibitors have been described, interfering either with the 3′ processing of the viral DNA long terminal repeats (4, 28) or with the strand transfer of viral DNA into the host genome (17; for a review, see reference 21). Raltegravir (RAL) is an IN strand transfer inhibitor (INSTI) which has been demonstrated to be an effective drug in antiretroviral-naïve (22) and -experienced patients (14, 37) and is to date the only INSTI approved for therapeutic use. Elvitegravir (EVG; GS-9137; Gilead) and S/GSK1349572 (ViiV Healthcare) are the next most advanced antiretroviral drugs (ARVs). However, as reported for other anti-HIV therapeutic compounds, resistance to RAL has been described in vitro and in vivo. The most frequent primary RAL resistance mutations emerging in vivo at virological failure (VF) in the region of the pol gene coding for IN are the Q148H/R/K, N155H and, to a lesser extent, Y143C/H/R mutations (9). Secondary mutations increasing the fitness of the resistant viruses have been identified and include the L74M, E138A/K, and G140A/S mutations for the Q148H/R/K pathway and the L74M, E92A/Q, T97A, Y143H/C, V151I, G163/K/R, and D232N mutations for the N155H pathway (24, 26). The G140S mutation did not confer strong resistance by itself, but it restored the replication capability of the Q148H mutant (10, 23).

In a previous work, we investigated the genetic pathways and the dynamics of emergence of the Y143C/R mutations in the HIV-1 IN gene in three patients failing RAL-containing regimens. The influence of Y143C/R mutations on IN functions and on the sensitivity of IN on RAL was analyzed (31). Enzymatic activities of the IN Y143C/R mutation were severely impaired compared to that of the wild type (WT). However, a high difference in fold change (FC) with RAL was found between in vitro IN mutants carrying the Y143C/R mutation alone or ex vivo with recombinant viruses from patients (31). In vivo, the level of resistance not only was linked to the amino acid found in those positions (R, C, or H) but also was highly modulated by the combination with secondary (compensatory) mutations and by the background IN sequence present in each patient at baseline. Indeed, selection of the Y143R/C mutation was accompanied in all three patients by the T97A mutation (31). Several reported virological data show that the Y143C/H/R mutations were associated with the T97A mutation in patients failing RAL therapy (6, 31). The addition by site-directed mutagenesis of the primary resistant mutation to patient-derived viruses containing the secondary mutation suggests that the secondary mutations (T97A, E92Q, G140S) enhance the resistance to RAL in the presence of the primary resistance mutations (Y143R, N155H, Q148H) (8, 13). Moreover, the T97 position is moderately polymorphic (32). Yet mutation T97A alone influences neither the virological response nor the selection of the Y143R mutation (1), which was shown to be associated with RAL resistance.

To investigate whether, according to virological data previously described, the association of the T97A mutation with the Y143C/R mutation might increase the resistance level to RAL and/or rescue the catalytic defect due to the Y143C/R mutation, we introduced this mutation either alone in the region of the pol gene coding for IN or associated with the Y143C/R mutation. We describe here the role of the T97A mutation combined with the Y143C/R mutation on IN activities and RAL resistance in vitro.

MATERIALS AND METHODS

Mutagenesis.

IN mutants were generated using a Stratagene QuikChange site-directed mutagenesis kit (La Jolla, CA), according to the manufacturer's instructions. The presence of the desired mutations and the integrity of the IN sequence were checked by DNA sequencing. Oligonucleotides (ODN) for mutagenesis were purchased from MWG Biotech (Courtaboeuf, France).

IN purification.

Standard purification of IN was performed in our laboratory essentially as previously described (7, 11, 19). Briefly, IN expression was obtained in Rosetta bacterial strains with 3 h of induction. The soluble fraction containing the HIV-1 IN was loaded on a Hitrap butyl-Sepharose 4B column (1 ml; Pharmacia-LKB), washed with LSC buffer {50 mM HEPES (pH 7.6), 0.2 M NaCl, 0.1 M EDTA, 1 mM dithiothreitol (DTT), 7 mM 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 10% glycerol} and equilibrated with 5 volumes HSC buffer (50 mM HEPES [pH 7.6], 0.2 M NaCl, 1 M ammonium sulfate, 0.1 mM EDTA, 1 mM DTT, 7 mM CHAPS). Proteins were eluted by a decreasing ammonium sulfate concentration (1 to 0 M). Fractions containing IN were pooled and 1/3 diluted with 50 mM HEPES (pH 7.6), 0.1 mM EDTA, 1 mM DTT, 10% glycerol, and 7 mM CHAPS and loaded on a HiTrap heparin Sepharose CL-4B column (1 ml; Pharmacia-LKB), washed with 5 volumes of LSC buffer, and equilibrated with HS buffer (50 mM HEPES [pH 7.6], 1 M NaCl, 0.1 mM EDTA, 1 mM DTT, 10% glycerol, 7 mM CHAPS) and eluted with a NaCl step (0.2 M to 1 M NaCl). Fractions containing IN were pooled and diluted 1:4.3 with 50 mM HEPES (pH 7.6), 0.1 mM EDTA, 1 mM DTT, 10% glycerol, and 7 mM CHAPS and loaded on a HiTrap heparin Sepharose CL-4B column (1 ml; Pharmacia-LKB). The protein was eluted with a linear NaCl gradient (0.2 M to 1 M NaCl).

Elution of all enzymes (mutants and WT) was observed at the same concentration of NaCl throughout the purification process. Fractions containing IN activity were pooled and concentrated, when necessary, by ultrafiltration (Centricon Millipore). Purified IN was kept at −80°C. Proteins were analyzed by electrophoresis by 12% SDS-PAGE. The enzymes have a similar purification rate of 99% homogeneity.

Processing and strand transfer activities.

Standard assays were performed as described previously in 20 mM HEPES (pH 7.5), 10 mM DTT, 7.5 mM MnCl₂, and 0.05% NP-40 in a total volume of 20 μl (7, 11, 19). The reaction mixture was incubated at 37°C for 1 h in the presence of IN and radiolabeled oligonucleotides (1 pmol, 50 nM). To analyze the products, the synthesis was stopped by adding 90 μl of mix urea (777 mM urea, 4 mM EDTA, and 1× TE buffer [7.7 mM Tris-HCl, 0.77 mM EDTA], 300 mM sodium acetate, 444 μg/ml glycogen) followed by phenol-chloroform extraction (1:1 [vol/vol]). DNA was precipitated by 2.5 volumes of absolute ethanol and was dissolved in 95% formamide, 0.5 mM EDTA, 0.025% SDS, 0.025% bromophenol blue, and 0.025% xylene cyanol and then heated for 3 min at 95°C. The reaction products were then analyzed by electrophoresis on 15% polyacrylamide gels with 7 M urea in Tris-borate-EDTA (TBE; pH 7.6) and autoradiographed. The ODN sequences used to perform the processing and strand transfer assays were the following: ODN70, 5′-GTGTGGAAAATCTCTAGCAGT-3′; ODN71, 5′-GTGTGGAAAATCTCTAGCA-3′; ODN72, 5′-ACTGCTAGAGATTTTCCACAC-3′. To perform the 3′ processing assay, the 5′-radiolabeled ODN70 hybridized to ODN72 was used as a substrate, while the 5′-radiolabeled ODN71 hybridized to ODN72 was used as substrate for the strand transfer reaction.

Modeling of the structure of the wild-type and mutated HIV-1 IN. (i.) Sequence comparisons.

Alignment of sequences was performed with ClustalW software using standard parameters. The alignment used for the homology modeling was the one presented in Fig. S1 in reference 29.

(ii) Model building.

The homology building procedure was performed on a Linux PC using the software Modeler version 9v2 (33). The starting template was the published structure of prototype foamy virus (PFV) IN with raltegravir inside (Protein Data Bank [PDB] codes, 3L2T for chain A [resolution 2.85 Å; 15] and 3OYA [resolution 2.65 Å; 16]). Thirty models were generated using standard parameters, and the first three models displaying an energy much lower than the remaining ones were chosen for further studies. Models were checked with MolProbity software (http://molprobity.biochem.duke.edu/).

(iii) Mutations and ligand interaction.

Calculations were performed with an SGI Octane workstation using Insight II and Discover version 2000 (Accelrys, Inc.). The crude model was then submitted to a partial minimization using Discover software with the consistent valence force field process. The backbone was first fixed, and the whole protein was then submitted to 100 steps of the steepest-descent process followed by 1,000 steps of conjugate gradient (CG). The backbone was then unfixed and tethered with a force of 100 kcal/Å, and 1,000 steps of conjugate gradient were applied. At this stage, the protein was superimposed on the X-ray structure of PFV IN (chain A) with raltegravir (and/or DNA) docked inside. The PFV IN was subsequently removed, leaving raltegravir within the structure of the HIV IN. The same procedure was used to dock the DNA strand within the HIV IN. The whole complex with docked RAL was then minimized with tethering on a protein backbone (1,000 steps of CG, 100 kcal/Å), and this constraint was slowly lowered in a stepwise manner to 10 kcal/Å.

Single point mutants were then produced using the homology module of Insight II, and the whole procedure was resumed for each mutant. Finally, we constructed single mutants, with the Y143C or Y143R mutation, and a double mutant, with the T97A and Y143R mutations, and RAL and/or DNA was docked within the structure as described above.

RESULTS

Biochemical activities of mutant IN.

The Y143C/R mutation in patients was found to be accompanied by the T97A mutation, and this double mutation is one of the profiles identified in patients resistant to RAL (5, 31). To elucidate whether the combination of the primary RAL resistance mutations, Y143C or Y143R, with the T97A secondary mutation reduces the susceptibility to RAL and rescues the catalytic defect due to the Y143C/R mutation, we generated mutations at amino acids positions 97 and 143. The threonine residue at position 97 was mutated to alanine (T97A), and the tyrosine residue at position 143 was mutated to cysteine (Y143C) or arginine (Y143R). The combination of double mutations was engineered (Y143C/T97A and Y143R/T97A). Recombinant enzymes were expressed and purified as mentioned in Materials and Methods (see Fig. S1 in the supplemental material).

The 3′-P and ST activities of mutants were compared to those of the wild type. 3′-P activity was moderately impaired with IN 97A compared that of the WT (Fig. 1A and B). With 250 nM IN, the activity of IN 143C/97A was only slightly higher than that of the WT. The 143R/97A mutation also showed slightly higher activity than the WT at the 80 nM IN concentration. Overall, 3′-P activities of IN were not severely impaired by these mutations.

Fig. 1.

Comparison of the 3′-processing activities of WT IN and IN 97A, IN Y143C/97A, and IN Y143R/97A mutants. Different concentrations of wild-type or mutated HIV-1 integrases were incubated for 1 h at 37°C with the 3′-end processing substrate indicated in Materials and Methods. Samples were analyzed by electrophoresis on 16% acrylamide gels/7 M urea in TBE. (A) A typical electrophoresis is shown. (B) Quantification of 3′-end processing activity of IN WT and mutants is shown. The percentage of 3′-P activity of each mutant was normalized versus WT activity at 250 nM taken as 100%. Data are the averages from at least three different experiments. nt, nucleotides.

Some authors in the literature (25) demonstrate that ∼30-bp DNA is better for measuring IN activities than an ∼20-bp substrate. We compared the overall activity of WT IN in the presence of 21 and 38 pb and in the presence of Mg²⁺ and Mn²⁺ (7.5 mM [each]). IN was active in Mg²⁺ and Mn²⁺. However, a better efficiency on a 38-mer substrate than that for the 21-mer was not observed. Therefore, all the following experiments were performed using a 21-mer substrate (data not shown).

On the other hand, the ST activities of the IN 97A and IN143C/97A mutant enzymes were more severely impaired, since only 70% (IN 97A) and 60% (IN143C/97A) of the activity of the WT was measured at an enzyme concentration of 250 nM (Fig. 2A and B). The Y143C/97A combination preferentially affected the ST activity while having limited impact on 3′-P (Fig. 1A and 2A). The Y143R/97A double mutant appeared fully active for both 3′P and ST (Fig. 1A and 2A).

Fig. 2.

Comparison of strand transfer activities of IN WT and mutants. (A and B) Increasing amounts of WT and mutated IN were incubated with the strand transfer substrate for 1 h at 37°C. A typical experiment is shown in panel A. The quantification of activities of IN WT and mutants is shown in panel B. The percentage of ST activity of each mutant was normalized versus WT activity. Data are the averages from at least three different experiments. *, P values of <0.05 are considered statistically significant. (C) Catalytic properties of wild-type and RAL-resistant INs. Strand transfer products are shown. WT and mutated IN (250 nM) were incubated with the strand transfer substrate for 1 h at 37°C. A typical analysis is shown in panel C (overnight exposure). S, substrate.

Secondary mutations increasing the fitness of the resistant viruses have been identified and include the T97A mutation for the Y143C/H/R pathway (5, 8, 26). In vitro, we observed a significant catalytic defect for both 3′-P and ST activities of Y143C/R proteins (31) (Fig. 2C). Interestingly, the Y143C/97A and Y143R/97A double mutants exhibited WT levels of ST activity (Fig. 2). Our in vitro results confirm the in vivo observations: the secondary T97A mutation rescues the catalytic defect due to the Y143C/R primary mutation.

The Y143C/R mutation associated with T97A mutations provides greater resistance than the Y143C/R mutation alone. (i) Comparison of sensitivities of the IN WT and 97A, 143C/97A, and 143R/97A mutants to RAL.

To investigate the possibility that Y143/T97 double mutants have a significant advantage over the single mutants, we compared the 3′-P and ST activities of the IN WT and IN 97A, IN 143C/97A, and IN 143R/97A mutants in the presence of increasing concentrations of RAL. As shown in Fig. 3A, we first tested the mutants for their sensitivity to RAL in the 3′-P assay. At the lowest concentrations of RAL used in the assay (0.41 μM), the 3′-P activity of WT and 97A protein showed partial inhibition (50% inhibitory concentrations [IC₅₀] of 1.2 μM for both INs) whereas IN 143C/97A and 143R/97A mutants remained weakly susceptible to RAL at concentrations up to 11.1 μM, showing a significant resistance to RAL (FC of 2 and 20, respectively) (Fig. 3A, part c). In the same 3′-P assay, ST activity resulting from 3′-P was also observed (Fig. 3A, part b). The ST of the IN WT and 97A mutant was inhibited and more sensitive to RAL than processing, thus showing the specificity of RAL for ST. Furthermore, the IN 143R/97A mutant was more resistant than the IN 143C/97A mutant, since we observed a complete inhibition of ST activity at the concentration 11.1 μM. A concentration of 11 μM was needed to inhibit the ST activity of the IN 143R/97A mutant while ST activity for the Y143C/97A mutant was inhibited by a RAL concentration as low as 1 μM.

Fig. 3.

Comparison of sensitivities to RAL of IN WT and mutants. (A) 3′-P activities in vitro. IN (250 nM) was incubated in the presence of increasing amounts of RAL. Activity was then assayed in the presence of a 3′-P substrate as described in Materials and Methods. A typical analysis is shown in panel A, parts a (2-hour exposure) and b (overnight exposure). The inhibition curve is shown in panel A, part c, and is the result of at least three independent experiments. *, P values of <0.05 are considered statistically significant. (B) ST activities in vitro. IN (250 nM) was incubated in the presence of increasing amounts of RAL. Activity was then assayed in the presence of an ST substrate as described in Materials and Methods. A typical analysis is shown in panel B, part a. The inhibition curve is shown in panel B, part c, and is the result of at least three independent experiments. *, P values of <0.05 are considered statistically significant.

The IC₅₀ for ST activity was also determined in an in vitro dose response assay with a precleaved substrate (19/21 duplex) (Fig. 3B). Under these conditions, the IN 97A mutant was able to catalyze ST in the presence of RAL (Fig. 3B, parts a and b) with an IC₅₀ of 0.002 μM, similar to that of the WT. The Y143C/T97A and Y143R/T97A double mutants conferred a high degree of resistance to RAL (FCs of 18 and 100, respectively).

(ii) Comparison of sensitivities to RAL of IN 143C and 143R single mutants and 143C/97A and 143R/97A double mutants.

To emphasize the selective advantage of the double mutants in the presence of RAL, we compared the 3′-P and ST activities of Y143C/T97A and Y143R/T97A mutants in the presence of RAL compared those of the Y143C and Y143R single mutants (Fig. 4). When using the full-length substrate, there was no significant difference between the 3′-P activities of Y143C and Y143C/T97A proteins in the presence of RAL (Fig. 4A, part a). The difference in IC₅₀ for 3′-P was not statistically significant (7 μM and 11 μM, respectively) (Fig. 4A, part b). We next evaluated the 3′-P inhibition of 143R and Y143R/T97A enzymes. The double mutant induced a strong increase in IC₅₀, from 3.7 μM for Y143R to 60 μM for Y143R/T97A (20-fold) (Fig. 4A, part c). The IC₅₀ for the Y143R mutant was close to that for Y143C (3.7 and 7 μM, respectively).

Fig. 4.

Comparison of sensitivities to RAL of single and double mutants. (A) 3′-P activities in vitro. IN (250 nM) was incubated in the presence of increasing amounts of RAL. Then activity was assayed in the presence of a 3′-P substrate as described in Methods. A typical analysis is shown in panel A, part a. Inhibition curves are shown in panel A, parts b and c, and are the result of at least three independent experiments. *, P values of <0.05 are considered statistically significant. (B) ST activities in vitro. IN (250 nM) was incubated in the presence of increasing amounts of RAL. Activity was then assayed in the presence of a ST substrate as described in Materials and Methods. A typical analysis is shown in panel B, part a. Inhibition curves are shown in panel B, parts b and c, and are the results of at least three independent experiments. *, P values of <0.05 are considered statistically significant.

We then monitored ST activity in the presence of increasing amounts of RAL (Fig. 4B). ST products resulting from IN Y143C and Y143R mutant activities were inhibited at the lowest concentration of RAL used (in the low nanomolar range), with IC₅₀ of 0.007 μM and 0.006 μM, respectively. Inhibition of the Y143C/T97A and Y143R/T97A double mutants required a RAL concentration, respectively, 6 times and >1,000 times higher than the concentration required to induce a comparable inhibition of the Y143C and Y143R IN mutants. There was a shift in the IC₅₀ from 0.007 μM for the 143C protein to 0.625 μM for Y143C/T97A mutant (Fig. 4B, part b) and a shift from 0.006 μM for the Y143R protein to 2.5 μM for Y143R/T97A mutant (Fig. 4B, part c).

Levels of inhibition by RAL of ST activities of the IN WT and single (T97A, Y143C, Y143R) and double mutants were compared in the presence of Mg²⁺ versus Mn²⁺. The IN WT and the IN T97A mutant showed a similar inhibition by RAL, as previously observed in the presence of Mn²⁺ (data not shown). However, IC₅₀ were 10-fold higher in the presence of Mg²⁺. The inhibition abilities of single mutants IN Y143C and IN Y143R was then compared in the presence of Mg²⁺. The IC₅₀ for the Y143C single mutant was higher (0.02 μM) than the IC₅₀ for the Y143R mutant (0.006 μM). In Mn²⁺, the IC₅₀ was similar for these two enzymes (0.006 μM). Finally, the ST activities of the double mutants in the presence of Mg²⁺ were too low to determine an IC₅₀ (data not shown).

Taken together, these data indicate that the primary RAL resistance mutations Y143C or Y143R combined with the T97A secondary mutation strongly reduce the susceptibility to RAL and rescue the catalytic defect due to the Y143C/R mutation.

(iii) Molecular modeling of WT IN and IN mutants.

To propose a mechanism for the resistance induced by the Y143R/C mutation in association with the 97A mutation, we analyzed the structural and molecular effects induced by these RAL resistance mutations. First, we created dynamic models of the wild-type IN and Y143C, Y143R, Y143C/97A, and Y143R/97A drug-resistant mutants of HIV IN starting from the published structure of PFV IN (15, 16, 20). Raltegravir and DNA were then docked to the model (Fig. 5).

Fig. 5.

Molecular modeling of the HIV-1 IN. Modeling of WT IN and IN 143C, IN 143R, Y143R/T97A mutants in complex with DNA and RAL. The protein is in blue, DNA is in green, RAL is shown as the yellow ghost surface and sticks, and residue 143 is in red. IN D64 and D116 are shown in magenta. Note the very close contact between DNA and 143R.

In the WT structure of HIV-1 IN, RAL is located in a pocket near the D116 catalytic amino acid, and the aromatic element of RAL interacts by π-stacking with the Y143 phenol ring. In HIV-1 IN, the OH of T97 residue makes a hydrogen bond with the side chain of residue N120 located in the middle of the very short helix 2 just before the long helix 3, but this is not the case with PFV IN; even if the T97 corresponding amino acid in PFV IN is still threonine in position 166, N120 is replaced by A189 (equivalent to the mutant T97A). Very interestingly, in PFV, this H2 helix is known to have implied contact with the minor groove of the DNA strand (reference 20 and PDB code 3OS0). Near residue N120, the residue N117 on the same segment is very close to raltegravir. When Y143 is mutated to C, residue 143 remains in the same position and the hydrogen bond T97-N120 remains unchanged. However, if Y143 is changed to R, the 143 position is modified: R143 first makes a saline bridge with residue D116, which switches to allow the interaction, and then R143 recruits D64 to form a triad between the three residues. The same situation is observed with the Y143R/T97A double mutant. At the same time, N117 is very close to the guanidinium moiety of R143. With these new positions occupied by residue R143 and N117, the binding pocket for RAL is no longer accessible (Fig. 5). When the DNA strand is docked in all these structures, it appears clearly that the mutation Y143C displays no clash with DNA at the opposite of mutation Y143R, where there is a short distance between 143R and the DNA strand (Fig. 5). But with the T97A mutant, the hydrogen bonds between A97 and N120 no longer exist, a situation which might destabilize or partially unfold the small helix 2 (which is indeed noted after minimization).

DISCUSSION

To date, no three-dimensional structure has been available for the full-length active HIV-1 IN or for HIV-1 IN bound to DNA. Recently, the crystal structure of full-length IN from the prototype foamy virus (PFV) in complex with its cognate DNA was reported (15). The recombinant PFV IN catalyzes efficient concerted insertion of two PFV DNA ends into target DNA and is sensitive to HIV-1 IN INSTIs (38). Raltegravir and elvitegravir inhibit PFV IN in vitro, indicating that the active sites of the FPV and HIV enzymes are quite similar (38). The isopropyl and methyl-oxadiazole groups of raltegravir are involved in hydrophobic and π-stacking interactions with the side chains of Pro214 and Tyr212, respectively (15). Numerous studies have modeled INSTIs at the HIV-1 IN active site (3, 16, 29, 34), with the most recent one focusing on raltegravir and elvitegravir. The binding of raltegravir in the context of these dynamic models of both the WT and the G140S/Q148H drug-resistant enzyme has been studied (29). Recently, the X-ray crystal structure of the PFV intasome together with chemically inhibited structures in the presence of raltegravir or elvitegravir was reported (18). A set of realistic models for the HIV-1 intasome in its active and INSTI-inhibited forms was constructed. In this model, a number of amino acids are in contact with raltegravir in addition to the DDE active-site residues: Gln146, Arg231, Asn117, Tyr143, Asn144, and Pro145.

Here, we show for the first time in vitro that the primary RAL resistance mutations Y143C or Y143R combined with the T97A secondary mutation strongly reduces the susceptibility to RAL and rescue the catalytic defect due to the Y143C/R mutation. The susceptibility to raltegravir and elvitegravir of site-directed mutant viruses carrying primary and secondary resistance IN mutations was recently described. The addition of Y143R primary resistance mutations to patient-derived viruses containing T97A secondary mutations showed dramatically increased levels of resistance to raltegravir and elvitegravir, confirming our results in vitro with raltegravir (8). To understand the origin of the resistance induced by the Y143R/C mutation in association with the 97A mutation and to compare it with results obtained with the Y143R/C mutation alone, we analyzed the structural and molecular effects induced by these RAL resistance mutations. First, we created dynamic models of the WT and Y143C, Y143R, and Y143R/97A drug-resistant mutants of HIV IN in the presence of raltegravir and/or DNA. A recent study demonstrated that raltegravir contacts the three catalytic residues D64, D116, and E152 and interacts with the five residues T66, E92, Y143, Q148, and N155 (22), which are involved in primary resistance (8, 26, 36). Modeling experiments in this work showed that the mutation of Y143 in R closes the lower part of the binding site for RAL and that the inhibitor is no longer able to bind to the enzyme. This could explain the decreased susceptibility of the double mutant to RAL. The addition of the T97A mutation to the single mutant rescues the catalytic activity of IN Y143C/R, which is impaired in the single mutant compared to the WT one. Y143R might immobilize both D64 and D116 of the catalytic site, leading to lower activity. On the other hand, the T97A mutation breaks the interaction between 97A and N120, thus unblocking helix H2. Under these conditions, helix H2 is able to adapt slightly and then allows a small move of the DNA strand to avoid close contact with 143R or 143C. Therefore, in all double mutations, the RAL binding pocket is no longer accessible, which explains the resistance, but helix H2 can adapt and can release the hindrance brought by 143C and particularly by 143R. When T97 is not mutated, the H bond between T97 and N120 maintains helix H2 in position and then prevents the DNA strand from adapting to the hindrance brought by the mutation in position 143.

Altogether, our data suggest that the combination of T97A and Y143R/C mutations leads to an increased resistance to RAL while the IN catalytic activity is maintained.

This observation explains the high frequency of coselection of T97A with Y143R/C in patients failing RAL-containing regimens. The factors leading to the preferential selection of this pattern or of the 155 or 148 mutations patterns need to be elucidated to further understand the dynamics of resistance to RAL in vivo.