Biochemical and Pharmacological Analyses of HIV-1 Integrase Flexible Loop Mutants Resistant To Raltegravir

Abstract

Resistance to raltegravir (RAL), the first HIV-1 integrase (IN) inhibitor approved by the FDA, involves three genetic pathways: IN mutations N155H, Q148H/R/K and Y143H/R/C. Those mutations are generally associated with secondary point mutations. The resulting mutant viruses show high degree of resistance against RAL but somehow are affected in their replication capacity. Clinical and virological data indicate the high relevance of the combination G140S+Q148H because of its limited impact on HIV replication and very high resistance to RAL. Here, we report how mutations at the amino acid residues 140 and 148 and 155 affect IN enzymatic activity and RAL resistance. We show that single-mutations at position 140 have limited impact on 3′-processing (3′-P) but severely inactivate strand transfer (ST). On the other hand, single-mutations at position 148 have a more profound effect and inactivate both 3′-P and ST. By examining systematically all the double-mutants at the 140 and 148 positions, we demonstrate that only the combination G140S+Q148H is able to restore the catalytic properties of IN. This rescue only operates in cis when both the 140S and 148H mutations are in the same IN polypeptide flexible loop. Finally, we show that the G140S-Q148H double-mutant exhibits the highest resistance to RAL. It also confers cross-resistance to elvitegravir but less to G-quadraduplex inhibitors such as Zintevir. Our results demonstrate that IN mutations at position 140 and 148 in the IN flexible loop can account for the phenotype of RAL-resistant viruses.

The first molecule approved for the treatment of HIV/AIDS was zidovudine (AZT, GlaxoSmithKline) a chain terminator inhibiting the viral polymerase, reverse transcriptase (RT). AZT was approved by the FDA in March 1987. Over the past 25 years many RT inhibitors and protease (PR) inhibitors have been generated (a total of 22 drugs) (1) to overcome the selection of resistant viruses that appear quickly (6 months) in AZT-treated patient (2). Highly active anti-retroviral therapy (HAART) is generally composed of 3-4 drugs targeting at least 2 viral enzymes at a time. This regimen is very efficient. It reduces viral load and extends the lifetime of HIV-1 infected people. Unfortunately, even with multiple drugs and a very low replication rate, virus diversity (quasi-species) and the poor fidelity of RT still allow the emergence of resistance. In 2003, the first inhibitor of fusion was approved by the FDA followed in 2007 by the first integrase (IN) inhibitor, raltegravir (RAL). Today, the therapeutic armamentarium allows the targeting of 4 different steps of the HIV life cycle including the inhibition of all three viral enzymes (25 drugs in 31 formulations) (3).

IN is required in vivo for the integration of the reverse transcribed viral DNA within genomic DNA. This step of the viral cycle is part of four different processes requiring IN (1). Just after reverse transcription, IN becomes associated with the long terminal repeats (LTR) and processes the viral DNA ends along the motif CAGT. Cleavage of the 3′- extremities of the LTRs (terminal dinucleotide GT 3′ from the conserved CA dinucleotide) is catalyzed by at least a dimer of IN (4). This first activity, 3′-P processing (3′-P), is performed in the cytoplasm within a large nucleo-protein complex composed of viral and cellular co-factors (the pre-integration complex, PIC). The PIC migrates along the microtubule network to the nucleus. Once in the nuclear compartment, the complex interacts with host DNA and the integration of both viral DNA ends occurs 5 bp one from another on opposite strands of the same DNA duplex. This reaction, performed by at least a tetramer of IN (4), is referred to as strand transfer (ST). Inhibitors targeting this activity are called IN strand transfer inhibitors (INSTIs) (1, 5). The last process involved in the completion of integration is the repair of the junctions between viral and cellular DNA. Those reactions are probably done by cellular enzymes and complete the integration of the viral DNA with a 5 bp duplication on each side. Both the 3′-P and ST reactions can be reproduced in biochemical assays using recombinant IN and short oligonucleotides derived from the LTR (6, 7).

IN is a 32 kDa protein issued from the action of PR on the gag-pol precursor. IN can also be produced as recombinant and catalytically active enzyme (6-8). It is composed of 3 domains (9, 10). The N-terminus (amino acids 1-50) contains a zinc-binding motif H₁₂H₁₆C₄₀C₄₃ involved in the oligomerization of IN (11). The core domain (amino acids 50-212) contains the catalytic triad D₆₄D₁₁₆E₁₅₂ consisting in two aspartates and one glutamate residues. This DDE motif is well conserved across the retroviral integrase superfamily, part of the nuclease-transposase superfamilly (including, RNase H, Ruv C, transposases and other retroviral integrases) (12). IN activities require the coordination of 2 divalent metal co-factors with the catalytic DDE triad and most likely together with the viral and host DNA. Although both Mn²⁺ and Mg²⁺ are effective in vitro, it is generally accepted that Mg²⁺ is the physiological metal (10). In the crystal structures, the DDE catalytic site is adjacent to a flexible loop, composed by residues 140 to 149 (see Figure 1A-C) (13), and which is also critical for catalysis. In particular, residue Q148 is implicated in the binding of the viral DNA and is critical for IN activities (14-16). The C-terminus (amino acids 212-288) contains a SH3 like domain and is involved in DNA binding. All 3 domains of IN form homodimers and are implicated in both the viral (donor) and the cellular (target) DNA binding. To date, no 3D structure of the full-length wild-type (WT) enzyme is available. Nevertheless, some structural data from truncated/mutant enzyme provide some insight of the global shape of the protein (13, 17). Unfortunately, the flexible loop (amino acids 140-149) could not be totally resolved and there is also no 3D structure of IN-DNA complexes. This dramatically impairs the rational design of inhibitors.

Figure 1. 3-D structure of HIV-1 IN and mutations investigated in the present study.

A-B. Crystal structure of HIV-1 IN. Amino acids implicated in RAL resistance (N155, G140 and Q148) are highlighted in green. Catalytic triad D64, D116 and E152 are represented in red and the flexible loop residues (140-149) are colored in blue. This structure corresponds to the catalytic core of IN (50-212) with 3 points mutations (G140A, G149A and F185K (pdb file 1bl3) (17). Surface (panel A) and cartoon (panel B) representations were obtained using MacPyMOLX11 hybrid.

C. Alignment of the sequence bordering and including the flexible loop of different IN (Human Immunodeficiency Virus type 1, HIV-1; Simian Immunodeficiency Virus, SIV; Feline Immunodeficiency Virus, SIV; Avian Sarcoma Virus, ASV; Rous Sarcoma Virus, RSV). Amino acids identical to HIV-1 IN sequence have been replaced by a period in other IN sequences. Numbering corresponds to HIV-1 IN.

D. SDS-PAGE of recombinant proteins mutated in their flexible loop and used in this study (10 μg/lane). IN corresponds to WT protein; MW: molecular weight markers (SeeBlue® Plus 2 Pre-tained Standard, Invitrogen™). Abbreviations for the mutant recombinant proteins = S: 140S; A: 140A; H: 148H; R: 148R; K: 148K; SH: 140S-148H; SR: 140S-148R; SK: 140S-148K; AH: 140A-148H; AR: 140A-148R; AK: 140A-148K.

The first IN inhibitor approved by the FDA, raltegravir (RAL, MK-0518, Merck), was originally introduced in regimen of heavily treated patients and is now also used in first line therapy (18-22). Specific mutations within the IN gene have already been identified in RAL-resistant patients (23-25). Three genetic resistance pathways with the primary substitutions Y143R/C, Q148H/R/K and N155H, have emerged in association with secondary mutations at position E92Q/T97A/G163R, G140S/A and E92Q/G, respectively (23, 26, 27). Such mutant viruses show high degree of resistance against RAL but somehow are affected in their replication capacity depending on the mutation (23, 28-30). Elvitegravir (EVG, GS-9137, JTK-303, Gilead) is the next most advanced currently in trials (phase III) (3, 31). Compared to RAL, EVG is more potent both in vitro and ex vivo (15, 32) but also exhibits a higher toxicity in non-infected cells (33). Another limitation of EVG comes from its inactivation by cellular enzymes (34, 35), which can be improved by co-administration with ritonavir (35). Regarding resistance mutations, we recently showed cross-resistance between EVG and RAL for a panel of point mutant IN (15). However, our prior study did not include the mutations that have now emerged from the clinical use of RAL.

In vivo data already suggest that the mutation combination G140S-Q148H is the most relevant one with a very slight impact on virus replication and the highest increase in resistance factor (23, 27, 36). In this particular case, it has been shown that mutation G140S rescued the defective phenotype of mutation Q148H (28). In the present study, we investigated the impact of mutations at position 140 and 148 on the activity of resulting IN and on resistance properties.

Materials and methods

Oligonucleotide Synthesis and Drugs

Oligonucleotides were purchased from Integrated DNA Technologies, Inc. (Coralville, IA). Oligonucleotides 21t (GTGTGGAAAATCTCTAGCAGT), 19t (GTGTGGAAAATCTCTAGCA) and 21b (ACTGCTAGAGATTTTCCACAC) were used to generate the in vitro substrates for IN assays. Single-stranded oligonucleotides 21t and 19t were labeled at the 5′-end using T4 polynucleotide kinase (New England Biolabs, Ipswich, MA) with [γ-³²P] ATP (Perkin-Elmer Life and Analytical Sciences, Boston, MA) according to the manufacturers’ instructions. Unincorporated isotopes were removed using the Mini Quick Spin Oligo Columns (Roche). The DNA duplexes 21t/21b (blunt-ended substrate, 21/21) and 19t/21b (pre-cleaved substrate, 19/21) were annealed by addition of an equal concentration of the complementary strand, heating to 95°C, and slow cooling to room temperature.

Primers used for site-directed mutagenesis G140S (TCAAGCAGGAATTTAGCATTCCCTACAATCC), G140A (TCAAGCAGGAATTTGCCATTCCCTACAATCC), Q148H (CCCTACAATCCCCAAAGTCACGGGGTAATAG), Q148R (CCCTACAATCCCCAAAGTCGCGGGGTAATAG), Q148K (CCCTACAATCCCCAAAGTAAAGGGGTAATAG) and N155H (CTTTAATTCTTTATGCATAGATTCTATTACCCCCTG) correspond to the coding strand. The reverse complementary strand for each primers was also used. Mutated codons are underlined.

Raltegravir (RAL, MK-0518) was purified and elvitegravir (EVG, JTK-303) synthesized as described previously (15). Oligonucleotides 93del (GGGGTGGGAGGAGGGT) and T30923 [(GGGT)₄] were ordered lyophilized from IDT and re-suspended upon arrival with potassium buffer (50 mM TrisHCl, pH 7.5; 10 mM KCl). G-quartets were formed by heating the samples at 98°C for 5 minutes and slow cooling to room temperature before storage at −20°C.

Mutagenesis

IN mutants were generated using the 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 verified by DNA sequencing.

Integrase Purification

Recombinant wild-type (WT) or mutant IN polypeptides were purified from Escherichia coli as described (37). Briefly, the IN gene was cloned into pET15b plasmid (Novagen®, Madison, WI) allowing the expression of N-terminus 6-His tagged protein under IPTG induction (Isopropyl β-D-1-thiogalactopyranoside, Sigma). After mutagenesis, WT and mutants enzymes were expressed in E.coli and purified using a Ni-column (Fast Flow Chelating Sepharose™, GE Healthcare). To allow the purification of multiple enzymes in parallel, we used the Vac-Man® Lab Vacuum Manifold (Promega, Madison, WI) with Poly-Prep® Chromatography columns (Bio-Rad, Hercules, CA). All the enzymes used in this study retained the N-terminal His tag.

Integrase Reactions

IN reactions were carried out by mixing 20 nM DNA with 400 nM IN (unless otherwise indicated) in a buffer containing 50 mM MOPS pH 7.2, 7.5 mM MgCl₂, 14 mM 2-mercaptoethanol, and drugs or 10% DMSO (dimethyl sulfoxide, the drug solvant). Reactions were performed at 37°C for 120 minutes unless otherwise indicated and quenched by addition of an equal volume of loading buffer [formamide containing 1% SDS (sodium dodecyl sulfate), 0.25% bromophenol blue, and xylene cyanol]. Reaction products were separated in 16% poly-acrylamide denaturing sequencing gels. Dried gels were visualized using a Typhoon 8600 (GE Healthcare, Piscataway, NJ). Densitometric analysis was performed using ImageQuant 5.1 software from GE Healthcare.

Results

Mutagenesis and purification of mutant IN polypeptides

To elucidate the role of the flexible loop for IN activity and resistance to INSTIs, we generated a panel of mutations at amino acid positions 140 and 148, commonly mutated in RAL-resistant patients (Figure 1). The glycine residue at position 140 was mutated to serine (G140S) or alanine (G140A) and the glutamine residue at position 148 was mutated to histidine (Q148H), arginine (Q148R) or lysine (Q148K). All combinations of double mutations at these same positions were also engineered (SH, SR, SK, AH, AR and AK). We also mutated the asparagine at position 155 to histidine (N155H) because it has been reported in RAL-resistant patients (38, 39). After sequencing, we confirmed the introduction of the clinically reported mutations in the IN encoding plasmid pET15b. Recombinant enzymes were expressed and purified (Figure 1D).

Biochemical activities of mutant INs

First, we assessed the catalytic properties of the IN mutants using time-course experiments in gel-based assays. Using the full-length substrate corresponding to the viral U5 DNA end (21/21 duplex), we determined simultaneously both the 3′-P and coupled ST activities of the recombinant proteins (Figure 2). The two mutations at position 140 (G to S and G to A) preferentially affected ST activity while having limited impact on 3′-P (Figure 2A). For the mutants at position 148, both ST and 3′-P were severely abolished. When we looked at the double-mutants, only the combination G140S-Q148H (SH) appeared almost fully active for both 3′-P and ST (Figure 2B). The combination SK was the only other one to show some remaining 3′-P activity with 45% of the WT level (Figure 2E).

Figure 2. Biochemical activities of mutant IN enzymes.

A-D. Time-course experiments of 3′-Processing (3′-P; panels A-B) and strand transfer (ST; panels C-D) were performed using a gel-based assay. Reactions were performed with the 21/21 full length (panels A-B) or 19/21 pre-cleaved substrates (panels C-D). Activities of WT and single-mutant enzymes, including G140S/A,Q148H/R/K are represented in panels A and C. Activities of the double-mutants (6 combinations G140S/A-Q148H/R/K) are represented in panels B and D. Reactions were performed at 37°C for 15, 30, 45, 60, and 120 minutes and stopped with the addition of an equal volume of loading buffer.

E. Summary of the 3′-P (using the 21/21 substrate) and ST (using the 19/21 substrate) activities for the 12 enzymes measured after 120 minutes incubation at 37°C. Means and standard deviations (SD) are from 3 to 9 independent determinations. Values are reported on each sides of the graphic (means ± SD).

Because of the defective 3′-P activity of some of the mutants, we directly examined their ST activity using the same gel-based assay but with a pre-cleaved substrate (19/21 duplex) corresponding to the 3′-P product (1, 6, 40, 41). Under these conditions, only the SH mutant was able to catalyze ST close to WT levels (Figure 2C and D). All the other single- and double-mutations had a ST activity below 30% of WT activity. A summary of the biochemical activities of all the mutants at position 140 and 148 is displayed in Figure 2E.

Complementation experiments

We next tested whether the rescue of activity observed in the double-mutant G140S-Q148H required the mutations G140S and Q148H to be in the same molecule (cis) or in two distinct molecules (trans) forming the active dimers or tetramers (4). The activity of the double-mutant (SH) was compared to a mixture of the single-mutants (S+H, Figure 3). Because the results presented in Figure 2 were done with 400 nM enzyme, we first tested the mix of the 140S mutant at 200 nM plus the 148H mutant at 200 nM. The resulting ST activity was not raised above that of each single-mutant alone (Figure 3B). When doubling the amount of enzyme (400 nM of each single-mutant leading to 800 nM of total enzyme in the reaction mix), the ST activity was still not increased to the level of the SH double-mutant (Figure 3A and C). These experiments demonstrate that the G140S and Q148H mutations need to be in the same IN molecule to complement each other.

Figure 3. Cis-complementation for the G140S-Q148H double-mutant.

A. ST activity of WT, G140S (S), Q148H (H), G140S-Q148H (SH) at 400 nM and a mixture of 400 nM G140S plus 400 nM Q148H (800 nM of total enzyme, S+H). Reactions were performed at 37°C for 120 minutes using the 19/21 pre-cleaved substrate.

B. Quantification of the representative experiment shown in panel A (right histogram). The same experiment was performed using 200 nM of IN mutant G140S with 200 nM of Q148H for a total amount of enzyme of 400 nM and quantification of a representative experiment is shown (left histogram).

Effect of the 140-148 flexible loop mutations on RAL resistance

Inhibition of the ST activity of the WT, and the S, H and SH mutants enzymes was examined in the presence of a range of RAL concentrations (Figure 4A). Quantification of ST products shows clear RAL concentration-response for the four enzymes in the range of concentrations used (Figure 4B). The IC₅₀ of RAL for WT was around 70 nM. The Q148H mutant showed resistance to RAL with a 2- to 3-fold increase in the IC₅₀ (≈ 180 nM) while the G140S mutant appears as susceptible to RAL as the WT enzyme. In contrast, the SH double-mutant showed an IC₅₀ shift to 3 μM and a high degree of resistance to RAL (≈ 43-fold).

Figure 4. Activity of RAL on the ST activity of the IN mutants.

A. Representative gel showing the ST activity of WT, G140S (S), Q148H (H) and G140S-Q148H (SH) IN in the presence of RAL (from 111 μM to 1.9 nM in a 3-fold decrement). Reactions were performed using the pre-cleaved 19/21 substrate at 37°C for 120 minutes before loading on 16% PAGE.

B. Inhibition of ST is quantified including data for the IN mutant N155H. Mean and standard deviations derived from at least 3 independents determinations. C. ST activity of WT and mutants IN in the presence of increasing concentrations of RAL. Data were obtained by transformation of panel B data.

Regarding the other mutants (A, R, K, SR, SK, AH, AR and AK), the very low level of activity of several combinations precluded accurate densitometric analysis. However, we visually scored the resistance profile to RAL, and those scores are summarized in Table 1. We also performed parallel experiments using the clinically relevant mutation N155H and found that the resistance of this enzyme was intermediate between the SH and WT enzymes with an IC₅₀ of 600 nM (Figure 4B).

Table 1.

Enzymatic activity and resistance of IN mutants

Enzyme	3′-P1	ST1	Resistance2
WT	++++	++++	−
S	++++	++	−
A	+++	+	+
H	−	+	+
R	−	+	+
K	−	+	+
SH	++++	++++	+++
SR	−	++	+++
SK	+++	+	++
AH	−	+	++
AR	+	+	+++
AK	−	−	+

¹3′-P (using the full-length substrate) and ST (using the pre-cleaved substrate) activities are scored as percent WT activity: − : <10%; + : 10-20%; ++ : 20-40%; +++ : 40-70%; ++++ : 70-100%.

²RAL IC₅₀ increase is scored compared to WT: − : 1-fold; + : 2-3-fold; ++ : 10-25-fold; +++ : up to 50-fold.

To emphasize the selective advantage of each mutations in the presence of RAL, we plotted the ST activity of the mutants in the presence of RAL compared to WT (Figure 4C). The ST activity of the SH double-mutant remained above 50% in the presence of 1 μM of RAL while, under these conditions, the ST activity of the WT enzyme was below 20%. On the other hand, the single-mutants G140S and Q148H are not able to produce more ST than WT at any of the RAL concentrations examined and the N155H mutant shows only a slightly increased ST activity between 30 nM and 2.5 μM of RAL as compared to WT. These results demonstrate the selective advantage of the SH double-mutant in the presence of RAL.

Effect of 140S-148H mutations on 3′-P inhibition by RAL

Although RAL is a potent ST inhibitor, it is also active on 3′-P activity at high concentrations (reported selectivity index around 150) (15). Thus, we investigated the effect of RAL on the 3′-P activity of the SH double-mutant. Using the full-length substrate, we observed an inhibition of the 3′-P and ST activity of WT protein in the presence of RAL (Figure 5A). The IC₅₀ for 3′-P was around 10.5 μM, consistent with a selectivity index of 150 for ST (Figure 5B). Figure 5 also shows that inhibition of 3′-P by RAL the SH double-mutant was observed at higher concentrations than for the WT enzyme (Figure 5A). The IC₅₀ of RAL for the 3′-P activity of the mutant was over 650 μM corresponding to more than 62-fold increase compared to WT (Figure 5B). Similarly to the relative ST activity, we determined the relative 3′-P activity of the SH mutant compared to WT without RAL (Figure 5C). In this case, as for ST, inhibition of the SH double-mutant required a RAL concentration 30 times higher than the concentration required to induce a comparable inhibition of the WT protein.

Figure 5. Resistance of the 3′-P activity of the IN G140S-Q148H mutant to RAL.

A. Representative gel showing the 3′-P activity of WT and G140S-Q148H (SH) IN in the presence of RAL (from 333 μM to 5.6 nM in a 3-fold decrement). Reactions were performed using the blunt-ended 21/21 substrate at 37°C for 120 minutes before loading on a 16% PAGE.

B. Inhibition of 3′-P was quantified for at least 3 independent experiments (means and standard deviations). C. 3′-P activity of WT and mutants IN in the presence of increasing concentrations of RAL compared to WT. Data were obtained by transformation of panel B data.

Cross-resistance of the 140S-148H double-mutant to EVG

We next investigated the effect of EVG on the ST activity of the WT and SH mutant enzymes (Figure 6A). ST activity of WT protein was inhibited with the lowest concentrations of EVG used (in the low nanomolar range). ST products resulting from the SH mutant activity were still observed in the same range of concentrations and the inhibition was observed only for higher concentration (low micromolar inhibition). Quantifications show a shift in the IC₅₀ from 6 nM for the WT protein to 1 μM for the SH double-mutant (Figure 6B). Regarding 3′-P activity, WT enzyme was inhibited by EVG with an IC₅₀ of 8 μM and the SH double-mutant showed a 12- to 13-fold resistance factor with an IC₅₀ of 100 μM (Figure 6B).

Figure 6. Resistance of the G140S-Q148H mutant to EVG.

A. Representative gel showing the ST activity of WT and G140S-Q148H (SH) IN in the presence of EVG (from 1.9 nM to 111 μM in a 3-fold increment). Reactions were performed using the pre-cleaved 19/21 substrate at 37°C for 120 minutes before loading on 16% PAGE.

B. Quantification of 3′-P inhibition (using the blunt-ended 21/21 substrate; PAGE not shown) and ST inhibition. Means and standard deviations were derived from at least 3 independent determinations.

Overcoming the resistance of the SH mutant with other chemotypes

G-quartet-forming oligonucleotides are well established IN inhibitors (42-46). We studied the 3′-P and ST inhibition of both WT and double mutant SH IN by the aptamer 93del (GGGGTGGGAGGAGGGT) and by a zintevir analog, T30923 (GGGTGGGTGGGTGGGT). As already reported (43), the WT enzyme was inhibited by 93del with an IC₅₀ for ST of 100 nM (data not shown). More noticeably, the SH double-mutant was inhibited in the same range of concentration as WT. We next evaluated the 3′-P inhibition of WT and SH double-mutant enzymes by 93del using the full-length DNA substrate (data not shown). The SH double-mutation induced a small increase in IC₅₀ from 0.2 μM for WT to 1.5 μM for the SH double-mutant (8-fold). We also performed similar experiments with an analog of Zintevir, T30923 (GGGTGGGTGGGTGGGT) (44). T30923 inhibited the ST activity of WT enzyme with a lower IC₅₀ than 93del (10 nM, data not shown). With a 3-fold increase in IC₅₀, T30923 showed minimal impact on the ST inhibition of the SH double-mutant (data not shown). When we looked at the 3′-P inhibition (with the full-length substrate, data not shown), the inhibition profile with T30923 was comparable to the profile with 93del. The SH double-mutation induced a small increase in IC₅₀ for 3′-P inhibition by T30923 from 200 nM for WT to 1μM for the SH double-mutant (5-fold).

Discussion

To date, no 3D structure is available for the full-length active IN or for IN bound to DNA. Only, isolated domains have been solved, twice in the presence of a ligand (47, 48). In contrast to the catalytic triad DDE (see Figure 1), which is always defined with metal co-factor, the segment encompassing amino-acid residues 140-149 is consistently not well resolved due to low diffraction. That segment is commonly referred to as the flexible loop (9, 10). The flexibility of the 140-149 segment is probably due at least in part to the presence of 2 glycines (G140 and G149) at each end acting as hinges. Glycine is the amino acid with the smallest side chain, which intrinsically enables the highest degree of rotation of the polypeptide backbone. Mutating residues 140 and 149 to alanine allowed the complete resolution of the loop suggesting that the loop with those mutated hinge residues is less flexible (17).

Here we show that single-mutations at position 140, from glycine to serine or to alanine impair ST activity without inactivating 3′-P (see Figure 2). To date, residue G140 has not been reported to directly interact with DNA. It is generally accepted that IN undergoes a conformational change between 3′-P and ST to accommodate the target (host chromosomal) DNA in the catalytic site of the enzyme while the processed 3′-end of the viral DNA becomes the nucleophile and the target of INSTI (1, 5). The G140S/A mutants could allow an effective interaction with the viral DNA, which would lead to its preserved ability to catalyze 3′-P. This mutant is however not able to catalyze ST. Possibly, this might be due to conformational restriction. Indeed, a recent study on Moloney murine leukemia virus (Mo-MLV) IN proposed a direct correlation between flexibility of the loop and ST activity. Mutations that reduced flexibility specifically impaired ST but not 3′-P or disintegration (49). In the context of the virus, the mutation G140S is known to delay viral replication. This delay was attributed to a lack of integration (50, 51). Our present study suggests this defect is primarily due to impaired ST.

Mutations at position 148 to histidine, arginine or lysine totally inactivated the enzyme for both the 3′-P and ST reactions. In the normal (WT) IN, the glutamine residue at position 148 and the tyrosine 143 of the flexible loop have been shown to interact with the tip of the viral DNA LTR (52). More precisely, chemical cross-linking studies showed a direct interaction between the IN residue at position 148 and the 5′-C on the overhang of the viral DNA lower strand (14, 37). Changing this glutamine residue to histidine, arginine or lysine, which have larger and longer side chains, probably alters viral DNA binding thereby inhibiting both 3′-P and ST. Similarly, mutating Q148 to alanine, asparagine or cysteine was previously shown to block ST activity (14). In vivo, mutations at position 148 markedly decrease the replication capacity of mutant viruses (3, 23). Our data suggest such defects are primarily due to inactivation of both the 3′-P and ST activities of IN.

Simultaneous mutations at both sites (140S and 148H) restored the catalytic activities of the resulting enzyme to almost WT levels and most notably to levels well above each of the single-mutants (see Figures 2 and 3). Our data demonstrate this complementation operates in cis; i.e. both mutations have to be present within the same IN molecule (see Figure 3). Indeed, mixing two single-mutant failed to rescue enzymatic activity. The rescue was only possible with the combination SH (see Figure 2). Any other combination tested (SR, SK, AH, AR or AK) at best only partially affected IN activities (SK mutant, only improving 3′-P). The finding that the flexible loop mutants do not complement each other if they are on different IN molecules is consistent with prior study showing that active site mutants (on the DDE motif) does not complement each other in trans (53). These results demonstrate the interdependency of residues 140 and 148 for IN catalytic activity. Structural studies are warranted to determine whether the SH double-mutant IN will reveal the position of the flexible loop in an active configuration.

Appearance of mutations in patients seems to be dependent on the time of exposure to RAL. The N155 pathway is generally the first one to emerge. Our data show that this mutation confers approximately 10-fold resistance to RAL but also reduces IN’s intrinsic enzymatic activity (see Figure 4). Viruses with the double-mutation G140-Q148 appear as treatment is prolonged (36). Single point mutations in the IN nucleic acid coding sequence are sufficient to produce all the clinically relevant mutants at position 140 and 148 examined here. Mutation G140S was first reported for resistance to L-CA (50, 51) and more recently has been found to also confer minimal resistance to RAL and some diketo acids (28, 54). Here, we show no detectable resistance of the G140S mutant to RAL (see Figure 4) or EVG (data not shown). In contrast, we find all the clinically relevant 148 mutants (Q148H/R/K) resistant to RAL (see Figure 4). However, all those single-mutants present replicative defects (23). Accordingly, we found that those IN mutants are catalytically impaired (see Figures 2 and 4). Moreover, Figure 4C shows that the enzymatic activity of all the single-mutants at positions Q148 is less than that of the WT enzyme in the presence of RAL. This phenotype could explain the tendency of the 148 single-mutants to be quickly replaced by the 140S-148H double-mutants in vivo.

While all the single-mutants impaired IN’s catalytic activity, here we show that the clinically relevant mutant G140S-Q148H, which reestablishes an active site able to carry out both 3′-P and ST, also highly resistant to RAL or EVG. Thus, our experiments demonstrate that the SH double mutation does not restore a proper drug binding site for RAL or EVG. Notably, the SH double-mutant IN was also resistant to 3′-P inhibition by RAL and EVG (see Figures 5 and 6). Thus, in spite of the fact that the 3′-P and ST sites may have different conformations, the SH double mutation alters both sites as revealed by RAL and EVG resistance for both 3′-P and ST. Considering that drug resistance affects not only ST but also 3′-P indicates that RAL and EVG can bind IN in the context of a complex with or without the viral DNA and that the drug binding site in those two conditions involves the flexible loop.

Finally, we show that other kinds of inhibitors such as guanosine quartets oligonucleotides could completely inhibit the SH resistant mutant (data not shown). G-quadraduplexes have been shown to be non-toxic and able to cross the cell membrane, allowing a potential inhibition of intracellular targets (46). Unfortunately, resistant viruses to zintevir presented mutations in the gp120 coding gene, showing that IN was not the primary target of this inhibitor (55). These results show that the SH double-mutant could be directly used to identify new inhibitors to overcome resistance to RAL and EVG. Altogether, our study provide a new insight on the role played by the IN flexible loop during the integration process and drug response. These results may guide future structural studies to better model the IN active site and allow the development of next generation IN inhibitors to overcome RAL resistance.

Abbreviations

3′-P

3′-Processing

AIDS

Acquired Immune Deficiency Syndrome

AZT

Azidothymidine

EVG

Elvitegravir

HAART

Highly Active Anti-Retroviral Therapy

HIV

Human Immunodeficiency Virus

IC₅₀

Inhibitory Concentration 50%

IN

Integrase

INSTI

Integrase Strand Transfer Inhibitor

LTR

Long Terminal Repeat

Mo-MLV

Moloney Murine Leukemia Virus

PIC

Pre-Integration Complex

PR

Protease

RAL

Raltegravir

RT

Reverse Transcriptase

ST

Strand Transfer

WT

Wild-Type