How HIV-1 Integrase Associates with Human Mitochondrial Lysyl-tRNA Synthetase

Replication of human immunodeficiency virus type 1 (HIV-1) requires the packaging of tRNALys,3 from the host cell into the new viral particles. The GagPol viral polyprotein precursor associates with mitochondrial lysyl-tRNA synthetase (mLysRS) in a complex with tRNALys, an essential step to initiate reverse transcription in the virions. The C-terminal integrase moiety of GagPol is essential for its association with mLysRS. We show that integrases from HIV-1 and HIV-2 bind mLysRS with the same efficiency. In this work, we have undertaken to probe the three-dimensional (3D) architecture of the complex of integrase with mLysRS. We first established that the C-terminal domain (CTD) of integrase is the major interacting domain with mLysRS. Using the pBpa-photo crosslinking approach, inter-protein cross-links were observed involving amino acid residues located at the surface of the catalytic domain of mLysRS and of the CTD of integrase. In parallel, using molecular docking simulation, a single structural model of complex was found to outscore other alternative conformations. Consistent with crosslinking experiments, this structural model was further probed experimentally. Five compensatory mutations in the two partners were successfully designed which supports the validity of the model. The complex highlights that binding of integrase could stabilize the tRNALys:mLysRS interaction.


Introduction
The human immunodeficiency virus type 1 (HIV-1) is a retrovirus that contains two copies of its genomic single-stranded RNA embedded into the nucleocapsid of mature particles. During the assembly of the virus, some tRNAs from the host cell are selectively packaged into the budding particles [1,2]. This includes tRNA Lys,3 that serves as a primer for initiation of reverse transcription, an essential step of the life cycle of the virus that is believed to take place in the virus shortly after budding, before infection of host cells [3]. In virio analyses reveal that annealing of tRNA Lys, 3 to the primer binding site (PBS) of viral RNA occurs in the viruses [4], consistent with the finding that the level of viral infectivity is correlated with the level of tRNA Lys,3 encapsidation into the virions [5]. Lysyl-tRNA synthetase, an enzyme of the translation machinery that catalyzes aminoacylation of tRNA Lys,3 with
Eluted proteins were either dialyzed against buffer ASU (20 mM Tris-HCl pH 7.5, 100 mM NaCl, 500 mM Urea, 10% glycerol, 1 mM EDTA, 10 mM 2-mercaptoethanol) for IN-H 6 and IN-∆N-H 6 (pI = 8.74 and 9.46, respectively), or against buffer AQU (20 mM Tris-HCl pH 7.5, 500 mM Urea, 10% glycerol, 1 mM EDTA, 10 mM 2-mercaptoethanol) for IN-CCD-H 6 (pI = 7.8), and applied either to a Mono S HR 5/5 column (IN-H 6 and IN-∆N-H 6 ) or to a Mono Q HR 5/5 column (IN-CCD-H 6 ) equilibrated in the same buffers. Proteins were eluted by linear gradients (40 column vol.) of NaCl from 100 to 300 mM (Mono S) or from 0 to 300 mM (Mono Q). Purified proteins were concentrated by ultrafiltration (Vivaspin 6, 10 kDa), dialyzed against storage buffer (20 mM K-phosphate pH 7.5, 100 mM NaCl, 0.02% Triton X-100, 2 mM DTT), and stored at −80 • C. Protein concentrations were determined by using calculated absorption coefficients of 1  6 , and IN-CTD222∆C15-H 6 ) were expressed as described above, except that cells were grown at 37 • C to an A 600 = 0.25, transferred to 28 • C and grown to an A 600 = 0.5, and expression was induced by addition of 1 mM IPTG for 6 h. Cell lysis and Ni-NTA chromatography were conducted as described above. Purification on Mono S HR 5/5 columns was performed as described above, except
The two integrase species from HIV-2_ALL and HIV-2_TRA were expressed in E. coli BL21(DE3) grown at 37 • C to an A 600 = 0.25, transferred to 28 • C and grown to an A 600 = 0.5, and expression was induced by addition of 1 mM IPTG for 6 h. Purification was conducted exactly as described above for IN-H 6 from HIV-1. Protein concentrations were determined by using calculated absorption coefficients of 1.156 and 1.203 A 280 units mg −1 cm 2 , respectively, for IN from HIV-2_ALL and HIV-2_TRA.

HTRF Assay
Homogeneous time-resolved fluorescence (HTRF) assays were performed in black, flat bottom, half-area, non-binding surface, 96-well microplates (Corning #3993). Human mitochondrial LysRS (mLysRS) with a C-terminal HA-tag (mLysRS-HA) was incubated at a dimer concentration of 1.5 nM with the various integrase derivatives carrying a C-terminal His-tag at concentrations indicated in the legends of the figures, in 10 mM Tris-HCl pH 7.5, 50 mM NaCl, 10 mM 2-mercaptoethanol, and BSA at 1 mg/mL, for 1 h on ice. Antibodies (Cisbio) directed to the His-tag and conjugated with Eu 3+ cryptate (Cisbio #61HISKLB), and to the HA-tag conjugated with XL665 (Cisbio #610HAXLB), were added and incubation was continued for 30 min. After addition of 50 mM KF, fluorescence of Eu 3+ cryptate and of XL665 was recorded at 620 nm (I 620 ) and 665 nm (I 665 ), respectively, after excitation of Eu 3+ cryptate at 317 nm, in an Infinite M1000 PRO microplate reader (TECAN). Results are expressed as the ratio of I 665 /I 620 .

Protein Photo-Cross-Linking
The QuickChange Lightning Site-directed Mutagenesis Kit from Agilent Technologies was used to introduce amber (TAG) stop codons at discrete sites within the nucleotide sequence encoding the catalytic domain of mLysRS or the CTD222 domain of integrase from HIV-1 as described previously [24]. Incorporation of p-benzoyl-L-phenylalanine (pBpa) into mutant proteins was conducted in E. coli BL21(DE3) transformed with pEVOL-Bpa that expresses the orthologous supression system [25]. The E. coli strains containing pET28b expressing the various mLysRS or IN-CTD222 mutants carrying TAG stop codons were grown at 37 • C in 1L of LB medium supplemented with kanamycin (50 µg/mL) and chloramphenicol (34 µg/mL), and containing 0.2% arabinose and 1 mM pBpa (Bachem, Bubendorf, Switzerland). When the culture reached an A 600 = 1.0, expression was induced by addition of 1 mM IPTG for 5 h. The mLysRS pBpa variants with pBpa inserted at the 35 distinct positions listed in Table S2 were purified as described previously [24], and the IN-CTD222 pBpa variants with pBpa inserted at the 12 distinct positions listed in Table S3 were purified as described above for wild-type IN-CTD222-H 6 . All the proteins were dialyzed against PBS and stored at −80 • C. During their purification, all the mutants of mLysRS pBpa or IN-CTD222 pBpa were eluted similarly to their wild-type counterparts, Viruses 2020, 12, 1202 5 of 18 mLysRS WT or IN-CTD222 WT , suggesting that their oligomeric structure was not affected by insertion of pBpa, as expected for mutations of residues accessible to the solvent.
Photo-cross-linking was conducted essentially as described in [26]. The different mLysRS pBpa species and IN-CTD222 pBpa species were mixed with IN-CTD222 WT and mLysRS WT , respectively, in a final volume of 80 µL into the wells of a 96-well plate cooled on ice, at protein concentrations indicated in the legends of the figures. Plates were covered with their polystyrene lids and with 3 mm-thick glass plates to filter short-wavelength UV light, and incubated on ice into a CL-1000 Ultraviolet Crosslinker (UVP) equipped with a 365 nm UV lamp. Control samples were withdrawn before starting irradiation, and cross-linked products were analyzed by SDS-PAGE and western blotting after exposure to UV light.

Docking Simulation for the mLysRS and IN Domains
Rigid-body docking was performed between a dimeric structure of lysine-tRNA synthetase and a crystal structure of the C-terminal CTD domain of HIV-1 integrase. The input structure used for lysyl-tRNA synthetase (PDB: 3BJU) was crystallized without a tRNA molecule. The catalytic and tRNA anticodon-binding domains visible in the crystal structure are shared by cytoplasmic and mitochondrial LysRS. We modeled the structure of the bound tRNA using as template the yeast tRNA:aspartyl-tRNA synthetase complex structure (PDB: 1ASY) and superimposed the synthetase chains to deduce the conformation of the bound tRNA. An input structure containing the bound tRNA was preferred so as to prevent docking models to accumulate in unlikely regions where tRNA binds. As for the HIV-1 integrase partner, the input structure was a homodimeric structure containing both the catalytic core and the CTD (PDB: 1EX4). For the docking step, we did not isolate the CTD domain of the HIV-1 integrase, and rather kept all the domains present in the 1EX4 homodimeric structure. In that way, the helical linker connecting the catalytic core and the CTD prevented docking models to accumulate in unlikely regions.
The rigid-body docking step was performed using the InterEvDock2 server [27] that takes into account both the physicochemical nature of protein surfaces and co-evolutionary information and uses three complementary scores Frodock [28], SOAP-PP [29] and InterEvScore [30] to identify the most likely interfaces (http://bioserv.rpbs.univ-paris-diderot.fr/services/InterEvDock2/). In the context of this particular complex, no co-alignments between both partners could be generated since they belong to different species. The docking protocol was performed as described previously following the standard protocol of the server [31,32], using as input the two dimeric structures described above. In the result archive, the 50 best decoys of every three scores (Frodock, SOAP-PP, InterEvScore) used in the consensus selection of the docking models were considered. Among those 150 models, 119 solutions involved the IN-CTD in the interface with LysRS. For the next refinement steps, only the CTD and not the catalytic core domain was considered in the structural models of the complex with LysRS. Models were clustered using fcc [33] with a cutoff threshold of 0.5 and removing similarities between symmetrical structures. Forty-nine non-redundant representative models of complexes were retrieved and were refined using Rosetta [34] through a standard relax protocol under native coordinate constraints and the scoring of the resulting interface energy between IN-CTD and LysRS using the beta_nov15 scoring function. The model with the lowest interface energy reached −45.9 rosetta units, significantly lower than any of the alternative configurations (second best model at −41.3) and was first selected for in-depth structural analysis and design of disruptive compensatory mutants. The coordinates of the refined structural model were deposited on the ModelArchive database (DOI: 10.1016/j.str.2008.12.014) and can be downloaded at (https://modelarchive.org/doi/10.5452/ma-bxirn).

Yeast Two-Hybrid Analysis
The yeast two-hybrid system developed by Brent et al. was used [35]. The mLysRS and IN-CTD222∆C10 coding sequences were inserted into the plasmids pJG4-5 (fused to the B42-activator domain placed under the control of a galactose-inducible promoter) and pEG202 (fused to the LexA DNA binding domain), respectively. Mutations in mLysRS or IN-CTD222∆C10 coding sequences  Table S3 were generated using the QuickChange Lightning Site-directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA). The yeast strain SKY54 (Matα his3 leu2::3LexAop-LEU2 ura3 trp1 lys2::λcI-op-LYS2) [36], which contains a chromosomal LEU2 gene placed under the control of LexA operators was transformed to his + with pEG202-derivatives and to trp + with pJG4-5 derivatives. At least four independent colonies were analyzed for their ability to grow in the absence of leucine (expression of LexAop-LEU2). SKY54 expressing a pair of interactive proteins grew on galactose medium (YNBGal) lacking leucine (expression of B42-fusions that interacted with LexA-fusions) but did not grow on glucose medium (YNB) lacking leucine (no expression of B42-fusions).

Antibodies and Western Blot Analysis
Rabbit anti-IN-CTD antibodies were generated against a synthetic peptide (NFRVYYRDSRDPV WKGPAKLLWKGEGAVVIQDNSDIKVVPRRKAKIIRDYGK) corresponding to residues 222-273 of HIV-1 integrase and affinity purified (GeneCust, Boynes, France). The specificity of these antibodies was controlled by western blotting ( Figure S1). Polyclonal antibodies to LysRS have been described previously [37]. Western blot analyses were conducted with goat anti-rabbit secondary antibodies conjugated to peroxidase (Chemicon) and the SuperSignal West Pico chemiluminescent substrates (Thermo Scientific, Waltham, MA, USA). Chemiluminescence was detected with a LAS-3000 Imaging System (Fuji, Tokyo, Japan).

The C-Terminal Domain Is the Major Region of Integrase Interacting with mLysRS
The packaging of tRNA 3 Lys from the host cell within newly made HIV-1 particles involves the formation of a ternary complex comprising the GagPol viral polyprotein, mitochondrial lysyl-tRNA synthetase and tRNA Lys from the host cell. The integrase domain located at the very C-terminus of the GagPol precursor protein (Figure 1) is the main contributor to the stability of the complex between mLysRS and GagPol [12]. The cryo-EM structural analysis of the HIV-1 STC intasome revealed that integrase is made of three well-defined structural domains ( Figure 1). The α-helical N-terminal domain (NTD) and the β-barrel C-terminal domain (CTD), are connected to the catalytic core domain (CCD) via long spacer polypeptides of 15 and 18 amino acid residues, respectively [22].
To determine which domain of HIV-1 integrase interacts with mLysRS, integrase was expressed in E. coli with a C-terminal His-tag (IN-H 6 ), as well as two derivatives with a deletion of the NTD (IN-∆N-H 6 ) or of the NTD and the CTD (IN-CCD-H 6 ) ( Figure 1A). These three constructs contain the CCD dimerization domain of IN. The apparent dissociation constants of the complexes formed between these three integrase species and mLysRS were determined using a homogeneous time-resolved fluorescence (HTRF) assay described in Khoder-Agha et al. [12]. synthetase and tRNA Lys from the host cell. The integrase domain located at the very C-terminus of the GagPol precursor protein (Figure 1) is the main contributor to the stability of the complex between mLysRS and GagPol [12]. The cryo-EM structural analysis of the HIV-1 STC intasome revealed that integrase is made of three well-defined structural domains (Figure 1). The α-helical N-terminal domain (NTD) and the β-barrel C-terminal domain (CTD), are connected to the catalytic core domain (CCD) via long spacer polypeptides of 15 and 18 amino acid residues, respectively [22].     Figure 1C). These three CTD derivatives interacted with mLysRS with binding affinities of 36 ± 8 nM, 57 ± 15 nM, and 142 ± 46 nM (Figure 2), respectively, showing that the β-barrel domain ( Figure 1A), from residues 222 to 269, is the main region of IN-CTD involved in the interaction with mLysRS.

Integrase from HIV-2 Also Interacts with mLysRS
Integrase from HIV-2 displays only 60% of identical residues as compared to integrase from HIV-1 ( Figure 3 and Table S1). The more variable domain is the NTD, with only 53.3% of identical residues between HIV-1 and HIV-2, the CCDs are more conserved, with 64.1% of identical residues between HIV-1 and HIV-2, and the highest level of conservation is observed for the CTDs, with about 70% of identities between HIV-1 and HIV-2.

Identification of the Amino Acid Residues of mLysRS Involved in Its Interaction with the CTD of Integrase
To identify the surface area of mLysRS involved in its interaction with the CTD of integrase from HIV-1, 35 variants of mLysRS carrying each a single pBpa inserted at the surface of the protein and evenly distributed at the surface of the catalytic domain (Table S2) were isolated as described previously [24]. The wild-type and pBpa-containing mLysRS proteins were incubated on ice at a dimer concentration of 85 nM in the presence of IN-CTD222-H 6 at a monomer concentration of 1.12 µM, in PBS buffer. After 70 min of exposure to UV at 365 nm, samples were analyzed by SDS-PAGE and western blotting using anti-IN-CTD antibodies ( Figure 5). After incubation with seven of the pBpa-containing mLysRS mutants, a high-intensity polypeptide with a molecular mass of 79 kDa was observed, corresponding to the expected size for a cross-linked species containing one monomer of IN-CTD222-H 6 per monomer of mLysRS. Four of the mutants, with pBpa inserted at positions I273, N293, F528, and E576, formed a patch of residues located in the vicinity of each other ( Figure S2). This patch is formed by three residues from one monomer (I273, N293, F528) and one residue from the other monomer (E576). The three other pBpa-containing proteins corresponded to insertion of pBpa at positions H364, Q381, and E418, which are scattered at the surface of mLysRS. Among the other mutants of mLysRS, insertion of pBpa at positions I284 and Y286, residues located within the patch of residues, defined above, also yielded a significant level of cross-linking. The identified residues define one site of interaction located on one side of the dimer of mLysRS, between the acceptor arms of the two tRNA molecules ( Figure S2). Because mLysRS is a symmetrical dimer, the two equivalent patches that are located 30 Å apart could bind the two CTDs of the native dimeric integrase.

Identification of the Amino Acid Residues of the CTD of Integrase Involved in Its Interaction with mLysRS
To search for residues of the CTD of integrase interacting with residues of mLysRS, 12 mutants of IN-CTD222-H 6 containing pBpa exposed at the surface of the protein (Table S3)

Molecular Docking of the CTD of Integrase on mLysRS
In parallel to the identification of the residues involved in the interaction between IN-CTD and mLysRS by experimental methods, we set up an independent docking simulation to explore the most likely interface which could be identified using the InterEvDock2 server [27] and a refinement protocol based on Rosetta software [34] (see Mat & Met). We did not use any of the experimental constraint a priori to guide the docking. Rather, we probed the robustness of the predictive protocol for its ability to retrieve a posteriori solutions consistent with experimental data. A good convergence between experimental and computational approaches could then be considered as a proxy for the reliability of the resulting model. At first, 10,000 decoys were sampled by the InterEvDock2 server and a restricted set of 150 most likely rigid-body docked models could be selected. A clustering step narrowed down the number of solutions down to 49 representative models. Despite the challenging size of the complex and the wide heterogeneity of the solutions, a single solution eventually emerged after refinement of the best 49 candidates (Figure 7). The model with the lowest interface energy as estimated by the Rosetta scoring function reached a value significantly lower than any other alternative solution ( Figure S4). This feature likely accounts for the significant complementarity of the interactions that could be modeled at the interface of the IN-CTD:mLysRS complex. This structural model suggests that electrostatic interactions, for instance between LysRS_R228 and IN_E246, or LysRS_E513/514 and IN_R231, as well as hydrophobic packing, for instance between LysRS_F258 and IN_A248, may have a crucial role at the interface between the two proteins ( Figure S5). These residues are good candidates to test the validity of the model by mutagenesis.

Probing the Putative mLysRS:IN-CTD Interface by Site-Directed Mutagenesis and Y2H
The structural model of mLysRS:IN-CTD interaction (Figure 7) suggested a set of mutations that could alter the interaction. Two types of mutations were introduced in mLysRS or in IN-CTD: mutations that are supposed to create electrostatic repulsion between the two proteins, and mutations that have been suggested to alter hydrophobic packing or to introduce steric clashes (Table S4 and Figure S5). Moreover, in some cases, it was envisioned that a mutation in IN-CTD could be compensated by a mutation in mLysRS (for example interaction between residues IN_R263 and LysRS_E531 might be restored in the double mutant IN_R263E and LysRS_E531R).
The effect of the mutations on the interaction of mLysRS with IN was analyzed in a yeast two-hybrid system previously used to identify the proteins of HIV-1 that interact with mLysRS [8]. Ten mutations were introduced into pEG202 expressing the CTD222∆C10 variant of IN-CTD and eight into pJG4-5 expressing mLysRS (Table S4). Interaction of mLysRS with IN-CTD is reflected by the growth of SKY54 yeast cells on YNBGal medium in the absence of leucine (Figure 8). Mutations R228E, D291K, E513/514R, F528A, E531R, and E576R on mLysRS clearly prevented the growth of SKY54, and mutation F528E had a more moderate effect ( Figure 8A). Mutation LysRS_Y295E had no discernible effect. On the other hand, mutations V250E, R262E, and R263E on IN-CTD222∆C10 reduced the growth of SKY54, and mutation R244D had no visible effect on yeast growth ( Figure 8B). The six other mutations introduced into the CTD of IN had a more moderate effect. In all cases, when mutations had an effect on yeast growth, no massive-growth was seen, as in the case of wild-type mLysRS and IN-CTD222∆C10, but a punctuated growth was observed indicating that only a fraction of yeast cells expressed the LEU2 gene and grew, suggesting that an imperfect pair of interacting proteins can activate expression of LexAop-LEU2, but at a low frequency.
Because mLysRS mutations R228E, E513/514R, F528A, E531R, and E576R prevented growth of SKY54 expressing wild-type IN-CTD222∆C10, we anticipated that introduction of a compensatory mutation into IN-CTD222∆C10 might restore growth of SKY54 expressing double mutations (Table S4). Since SKY54 expressing the mutation IN-K244D grew similarly to wild-type (Figure 8), we did not test the combined effect of mutation LysRS_D291K on yeast growth. In all other cases, growth of SKY54 expressing the two mutant proteins was very similar to growth of SKY54 expressing wild-type proteins ( Figure 9). This result implies that a permutation of the residues in mLysRS and IN-CTD222∆C10 allowed to re-establish a functional interaction between the two proteins. were grown in a galactose-containing medium in the absence of leucine. Yeast cells expressing compensatory mutations were growing much faster than cells expressing a single of these mutations and were able to grow similarly to cells expressing the wild-type proteins.
The structural model of the mLysRS:IN complex recapitulates the residues of the two partners that have been shown to be crucial for the interaction: A248 and V250 of IN form a patch of hydrophobic residues interacting with F528 from mLysRS; R262 and R263 of IN form salt bridges with E576 and E531 from two monomers of mLysRS, respectively ( Figure 10). Insertion of pBpa at positions LysRS_F528, LysRS_E576 and IN_R263 was also found to produce cross-links between the two proteins ( Figures 5 and 6). These seven residues build a hydrophobic and electrostatic motif involved in the assembly of the mLysRS:IN complex. Compensatory mutations of residues LysRS_R228 with IN_E246, and LysRS_E513-514 with IN_R231 also restored the interaction ( Figure   Figure 9. Two-hybrid analysis of compensatory mutations in mLysRS and IN-CTD. Mutants of the CTD of HIV-1 integrase (IN-CTD222∆C10) were co-expressed with mutants of mLysRS in the two-hybrid system, as described in the legend of Figure 8. Cells co-expressing wild-type or mutants of mLysRS with wild-type IN-CTD222∆C10 (first row), wild-type mLysRS with wild-type or mutants of IN-CTD222∆C10 (second row), or a mutant of mLysRS with a mutant of IN-CTD222∆C10 (third row) were grown in a galactose-containing medium in the absence of leucine. Yeast cells expressing compensatory mutations were growing much faster than cells expressing a single of these mutations and were able to grow similarly to cells expressing the wild-type proteins.
The structural model of the mLysRS:IN complex recapitulates the residues of the two partners that have been shown to be crucial for the interaction: A248 and V250 of IN form a patch of hydrophobic residues interacting with F528 from mLysRS; R262 and R263 of IN form salt bridges with E576 and E531 from two monomers of mLysRS, respectively ( Figure 10). Insertion of pBpa at positions LysRS_F528, LysRS_E576 and IN_R263 was also found to produce cross-links between the two proteins ( Figures 5 and 6). These seven residues build a hydrophobic and electrostatic motif involved in the assembly of the mLysRS:IN complex. Compensatory mutations of residues LysRS_R228 with IN_E246, and LysRS_E513-514 with IN_R231 also restored the interaction (Figure 9). These residues, which are located at the periphery of the interface core, stabilize the interaction. A strong cross-link was also observed when pBpa was inserted at IN-R231 ( Figure 6B).

Discussion
In this work, we obtained a structural model of the complex between mLysRS and IN-CTD strongly supported by three sets of experimental constraints, six cross-links with pBpa-containing mLysRS mutants, five cross-links with pBpa-containing IN-CTD proteins, and five pairs of compensatory mutations. This structural model brings several key insights. First, association of IN-CTD requires that mLysRS is a dimer. Indeed, among the residues clearly involved in the interaction, as assessed by compensatory mutant experiments (Figure 9), two belong to one subunit of the dimer (R228 and E576), and four to the other subunit (E513, E514, F528 and E531) ( Figure S5). The two symmetrical binding sites on mLysRS are 30 Å apart, and thus, could bind two IN-CTDs. The polypeptide linkers between IN-CCD and IN-CTD of a dimer of integrase are made of 18 amino acid residues, and were described as non-structured polypeptides in the intasome structure [22] or as long α-helices in the crystal structure of a dimer of IN-CCD-CTD [21]. Therefore, this linker is very flexible and the two CTDs of a dimer of integrase are likely to be able to bind to a dimer of LysRS. This is also consistent with our finding that monomeric IN-CTD binds mLysRS with a K d of about 40 to 140 nM ( Figure 1C), whereas oligomeric IN binds to a dimer of mLysRS with a 20-fold higher affinity (K d of about 3.0 nM, Figure 1A). However, in the context of the GagPol polyprotein, it is not known if a monomeric domain of IN located at the very C-terminus of GagPol interacts with mLysRS to form the GagPol:mLysRS:tRNA Lys,3 packaging complex, or if the IN domain of GagPol is able to form a dimer in the context of the polyprotein precursor. The second noticeable feature of the mLysRS-IN complex is the proximity of the tRNA molecules with the CTD domains of integrase ( Figure 10). The acceptor stem of the tRNA molecule is close to the IN-CTD, which suggests a possible protein-tRNA interaction that could stabilize the association of the tRNA within the mLysRS:IN complex. In particular, Lys266 and Arg269 are directed towards the TΨC stem-loop of the tRNA molecule and could establish salt bridges. This is consistent with our previous report showing that the affinity of tRNA Lys,3 for mLysRS is increased by about two-fold in the presence of a derivative of integrase [12].
The finding that the five compensatory mutants tested in this work ( Figure 9) successfully restored normal growth on yeast cells, strongly argue in favor of the proposed model of mLysRS:IN association. It is noteworthy that mutations introduced in IN-CTD have less pronounced growth phenotypes than those introduced in mLysRS (Figure 8). IN-CTD is a small domain made of 57 amino acid residues that form a β-barrel structure ( Figure S3). Residues of IN-CTD that build the site of interaction with mLysRS are mostly located in loops that join the β-strands of this small structural domain, which suggests that they may be more mobile than the residues of mLysRS engaged in this assembly platform, which are part of a more compact structure, and are generally located in α-helices. Thus, mutations in IN-CTD could be more easily tolerated due to the flexibility of this domain that could accommodate some variations.
Among 20,000 sequences of IN from HIV-1 (from non-redundant GenBank CDS translations+PDB+SwissProt+PIR+PRF), only 16 changes are observed for A248 (11S, 4T and 1V), 17 changes for V250 (15I, 2L and 1G), eight changes for R262 (7K and 1G), and 10 changes for R263 (7K, 2G and 1S). This very high level of conservation is also noted among 425 sequences of IN from HIV-2 (from non-redundant GenBank CDS translations+PDB+SwissProt+PIR+PRF), three changes are observed for R262 (1K and 2S), one change for R263 (1G), and A248 is strictly conserved. Concerning V250, this residue is mainly recovered as Ile (390) or Leu (27), corresponding to conservative changes. The high level of conservation of these residues in HIV-1 and HIV-2 suggests that they are important for the function of integrase. Nevertheless, conservation of functionally important residues is likely to be correlated to the many roles of integrase in the life cycle of the virus.
Integrase is a multifunctional protein involved in many aspects of HIV-1 biology. It catalyzes the strand-transfer reaction during the integration step of viral DNA into host genome, it fulfills an essential role in virus morphogenesis, and, as a component of the GagPol polyprotein precursor, it appears to be necessary for the packaging of tRNA Lys,3 complexed with mLysRS, a crucial step for initiation of reverse transcription. The CTD of IN is involved in these three functional roles. Residues R228, R231, E246, A248, R263, and K266 are predicted to interact with viral DNA and residues R262, R263, and K266 with another monomer of IN, within the strand-transfer complex [22]. Residues K264, K266, R269, and K273 interact with viral RNA, an essential step in virus morphogenesis [17]. Mutation of these residues into Ala generates noninfectious particles that are unable to initiate reverse transcription. In the present study, mutation of residues R231, W243, E246, A248, V250, V259, R262, and R263 of IN-CTD compromised its interaction with mLysRS and are predicted to abolish tRNA Lys,3 packaging into viral particles, and to inhibit viral replication. The conclusions of this work must now be validated by in cellulo approaches. The results obtained in this study offer the opportunity to test several mutants of mLysRS and of the IN domain of GagPol for mLysRS and tRNA Lys,3 packaging, for the initiation of reverse transcription of viral RNA and for HIV-1 infectivity.
Although packaging of tRNA Lys,3 into new virions is absolutely required to generate infectious particles, several of the conserved residues of IN-CTD identified as key residues in the interaction with mLysRS are also involved in electrostatic protein-DNA interactions within the strand transfer complex [22]. Because residues R231, E246, A248, and R263 are predicted to be involved in these two functions, characterization of the effects of their mutation in a cellular system of HIV-1 replication requires a detailed analysis of the stages of the viral life cycle that are affected by these changes.
Two classes of inhibitors have been developed to inhibit IN functions. INSTIs, such as raltegravir or dolutegravir, inhibit the strand-transfer step catalyzed by IN and are used in antiviral therapies [18]. ALLINIs induce abnormal IN multimerization, prevent interaction on IN with viral RNA, generating eccentric non-infectious particles defective in viral replication [17,19]. Because viral resistance to drugs frequently develops, there is a need to develop antiviral drugs with new resistance profiles. The structural model of the complex between IN-CTD and mLysRS reported in this study will provide a support for searching molecules likely to disrupt the interface between the two proteins. The best inhibitor candidates should be able to interfere with both the hydrophobic and electrostatic components of the assembly platform.
Supplementary Materials: The following are available online at http://www.mdpi.com/1999-4915/12/10/1202/s1, Figure S1: Characterization of polyclonal anti-IN-CTD antibodies, Figure S2: Localization of pBpa-cross-linked residues on the 3D structure of LysRS, Figure S3: Localization of pBpa-cross-linked residues on the 3D structure of IN-CTD, Figure S4: Pipeline used for the generation of the structural model of mLysRS:IN-CTD complex, Figure S5: Interaction of suggested key residues at the interface of mLysRS with the CTD of HIV-1 integrase, Table S1: Sequence identities between IN species, Table S2: Position of pBpa insertion into mLysRS, Table S3: Position of pBpa insertion into IN-CTD222∆C10 from HIV-1, Table S4