Dual inhibition of HIV-1 replication by integrase-LEDGF allosteric inhibitors is predominant at the post-integration stage

Background LEDGF/p75 (LEDGF) is the main cellular cofactor of HIV-1 integrase (IN). It acts as a tethering factor for IN, and targets the integration of HIV in actively transcribed gene regions of chromatin. A recently developed class of IN allosteric inhibitors can inhibit the LEDGF-IN interaction. Results We describe a new series of IN-LEDGF allosteric inhibitors, the most active of which is Mut101. We determined the crystal structure of Mut101 in complex with IN and showed that the compound binds to the LEDGF-binding pocket, promoting conformational changes of IN which explain at the atomic level the allosteric effect of the IN/LEDGF interaction inhibitor on IN functions. In vitro, Mut101 inhibited both IN-LEDGF interaction and IN strand transfer activity while enhancing IN-IN interaction. Time of addition experiments indicated that Mut101 behaved as an integration inhibitor. Mut101 was fully active on HIV-1 mutants resistant to INSTIs and other classes of anti-HIV drugs, indicative that this compound has a new mode of action. However, we found that Mut101 also displayed a more potent antiretroviral activity at a post-integration step. Infectivity of viral particles produced in presence of Mut101 was severely decreased. This latter effect also required the binding of the compound to the LEDGF-binding pocket. Conclusion Mut101 has dual anti-HIV-1 activity, at integration and post-integration steps of the viral replication cycle, by binding to a unique target on IN (the LEDGF-binding pocket). The post-integration block of HIV-1 replication in virus-producer cells is the mechanism by which Mut101 is most active as an antiretroviral. To explain this difference between Mut101 antiretroviral activity at integration and post-integration stages, we propose the following model: LEDGF is a nuclear, chromatin-bound protein that is absent in the cytoplasm. Therefore, LEDGF can outcompete compound binding to IN in the nucleus of target cells lowering its antiretroviral activity at integration, but not in the cytoplasm where post-integration production of infectious viral particles takes place.


Background
Raltegravir (Merck) and Elvitegravir (Gilead) were introduced in 2007 and 2012 respectively, as the first generation of integrase strand transfer inhibitors (INSTIs) and confirmed integrase (IN) as a clinically validated viral target for antiretroviral (ARV) therapy [1]. The mode of INSTI action was elucidated in complex with a retroviral IN for which the entire 3D structure was defined [2]. However, resistance to INSTIs has emerged in patients [3,4]. A second generation of INSTIs, less sensitive to drugresistance mutations, has been approved (Dolutegravir (DTG) from GSK-Shionogi-ViiV). DTG belongs to the same class of compounds and remains sensitive to the strongest INSTI resistance mutations [5,6]. This highlights the need for integration inhibitors with completely different mechanism of action.
LEDGF/p75 (LEDGF), the main cellular cofactor of IN [7][8][9] is of great interest for the development of a novel generation of integration inhibitors. LEDGF interacts with IN through its C-terminal integrase binding domain (IBD). HIV-1 IN catalytic core (IN-CCD) and N-terminal domains are involved in the interaction with LEDGF [7][8][9][10][11][12]. LEDGF is crucial for integration and replication of HIV [13] although minor residual replication (~10%) was seen in LEDGF-depleted cells [14]. LEDGF functions as a tethering factor for IN, targeting the integration of HIV in actively transcribed gene regions of chromatin [15].
Several inhibitory activities of LEDGINs and tBPQAs have been reported so far. These include the inhibition of IN-LEDGF interaction, the inhibition of IN strand transfer and 3′ processing activities (independent of LEDGF), change in IN oligomerization toward stabilization of IN dimers and inhibition of the formation of the stable IN-viral DNA synaptic complex (SSC) [18,[35][36][37][38]. These compounds are considered as allosteric inhibitors of IN that are able to block HIV integration [18,[35][36][37][38][39][40] and are also referred to as ALLINIs [37,40]. These compounds remain fully active on IN mutants that are resistant to INSTIs and are therefore a promising new class of IN inhibitors. An inhibitory effect of LED-GINs on the infectivity of progeny virions has been reported lately [35,[41][42][43][44][45]. The multiple activities of these compounds raise questions regarding the unicity or multiplicity of their mechanism of action. Here, we explore what mode of action could explain the multiple activities of these inhibitors. We investigate the respective contribution of these different activities to the overall ARV activity of these compounds using a new series of IN-LEDGF inhibitors from the LEDGIN and tBPQA family of compounds.

Development of IN-LEDGF allosteric inhibitors
New IN-LEDGF allosteric inhibitors (INLAIs) of the aryl or heteroaryl-tertbutoxy-acetic acid family were designed. The structure and activities of 7 of these compounds are shown on Table 1. These compounds efficiently inhibited IN-CCD/LEDGF-IBD interaction as well as the interaction between IN and full length LEDGF proteins in homogeneous time-resolved fluorescence (HTRF) assays ( Figure 1A and C). MT4 cells were infected with HxB2 HIV-1 and a subset of 51 compounds showed a good correlation between their ARV activity and their ability to inhibit IN-CCD/LEDGF-IBD or IN-LEDGF interactions ( Figures 1B and D). The most active compound for IN-LEDGF inhibition, Mut101, also had the highest ARV activity (an EC 50 value of 92 nM against HxB2 infection, CC 50 for cytotoxicity was undetectable at over 50 μM (Table 1)). LEDGF was able to compete with these inhibitors, increasing the IC 50 Table S1. Two Mut101 molecules bound to the IN-CCD dimer (Additional file 1: Figure S1A). The ligand was found to be in a pocket surrounded by hydrophobic residues on one side, an acidic region on the other side and basic residues at the bottom of the pocket (Additional file 1: Figure S1B). Three hydrogen bonds link the carboxylic acid group of Mut101 and the protein (Figure 2A), one with the hydroxyl group of the side chain of Thr 174, and two with the amino group of the main chain of His 171 and Glu 170. In addition Mut101 was found to interact with two water molecules (Additional file 1: Figure S1C). The IN-CCD structures with and without Mut101 were superimposed. We found structural differences that appear in two regions ( Figure 2B), which contrasts with previously reported IN-CCD/LED-GIN or tBPQA co-structures where no differences were found [18,36,37].

Effect of IN-LEDGF inhibitors on IN strand transfer and 3' processing activities is independent of LEDGF
We found that these compounds inhibited the IN strand transfer activity as quantitated by ELISA assay ( Figure 3A), in agreement with previously reported data, with IC 50 values in a similar range to those found for inhibition of the IN-LEDGF interaction (Table 1). Activity in the concentration range studied (up to 100 μM) was always partial (reaching a plateau at 56-73% inhibition), which contrasts the full inhibition obtained using Raltegravir.
In contrast with data reported by Christ et al. [35], modification of the order of addition of compounds, before or after DNA in this strand transfer assay, did not result in full inhibition (data not shown). This partial and weaker inhibition than that of INSTIs, was confirmed using a typical assay with radioactive oligonucleotide and gel analysis of the strand transfer products ( Figure 3E-F  Figure 3B). This IN strand transfer inhibition was found regardless of whether or not the donor DNA was preprocessed [36]. Inhibition of IN 3′ processing activity was reported for some INLAIs [37]. We found that increasing concentrations of Mut101 or BI-D lead to a slight decrease in the 3′ processing efficiency (with a maximum of 25-30% inhibition, Figure 3C-D), but their inhibition of the IN strand transfer reaction was more important. ( Figure 3E-F).

IN-LEDGF inhibitors enhance the IN-IN interaction
We   incubation with Mut101 or BI-D results in higher IN oligomerization state (red peaks), that likely corresponds to a partial formation of IN tetramer. Raltegravir had no effect (data not shown). In contrast with some LEDGINs previously described [18]  pocket and mimicking the effect of LEDGF binding to IN [16,46].

Mut101 behaves as an inhibitor of integration in time-of-addition experiments
We performed a time-of-addition experiment (TOA) to identify the HIV-1 replication cycle step that is blocked by Mut101. We used Mut101 at a saturating concentration (25 μM) and single-cycle infection kinetics with VSV-Gpseudotyped Δenv HIV-1 NL4-3 expressing luciferase as a measure of infection. The kinetics of decreased activity after Mut101 addition were very similar to that observed with Raltegravir, but different to those of Nevirapine, suggesting that Mut101 at saturating concentration behaved as an inhibitor of integration ( Figure 5A). This is in full agreement with data reported previously on LEDGINs and tBPQAs [18,36]. The replication cycle analysis by quantitative PCR confirmed that Mut101 inhibited the integration of the proviral DNA ( Figure 5C) but not the production of proviral DNA at reverse transcription ( Figure 5B).

Mut101 remains fully active against HIV-1 mutants that are resistant to INSTIs and other anti-HIV drugs
Mut101 was tested against a panel of virus mutants harboring, in an NL4-3 background, some of the strongest resistant mutations to INSTIs and other classes of ARV drugs used in clinics [47]. These mutants are listed on Table 2. The activity of Mut101 and reference compounds was quantified by the fold change (FC) ratio between EC 50 on resistant virus and EC 50 with the wild type (wt)a measure of compound efficacy on resistant mutant virus. Mut101 had an FC ratio of 1 or lower against all resistant viruses contrasting the results with reference compounds (Table 3). This demonstrates that Mut101, as IN-LEDGF inhibitor, is a candidate for a novel class of drugs that can act on viruses resistant to those currently used in clinics, including INSTIs.
Unlike INSTIs, the Mut101 series of compounds are more potent when assayed with replicative HIV-1 than with non-replicative pseudotyped virus The ARV activity of a drug can be assessed using different assays. Multiple round infection using a replication- competent virus reveals the global ARV activity of a drug, but cannot give an indication as to which step of the viral replication cycle is blocked. All classes of drugs are found fully active in multiple round infection assays. In contrast, in single-round infection, replicationdefective Δenv viruses pseudotyped with an exogenous envelope (VSV-G) can complete viral replication only up to integration. This enables drugs like RT or IN inhibitors (fully active because they act early during the replication cycle, before or at integration) to be distinguished from drugs such as protease inhibitors that act late after integration (inactive in the single cycle assay) (see Table 4). Drugs that act early during reverse transcription (such as AZT and Nevirapine), or at integration (such as Raltegravir) showed ARV activity that is similar or slightly better in single-round (SR) infection assays than in multiple round (MR) infection assays (an EC 50 SR/EC 50 MR ratio of 1 or lower; Table 4). IN-LEDGF inhibitors, as allosteric inhibitors of HIV-1 integrase, were expected to behave similarly to Raltegravir with a SR/MR ratio close to 1. Intriguingly this was not the case. In contrast, Mut101 and the other compounds of this study were much more potent in MR than in SR infection assay with EC 50 SR/EC 50 MR ratios always much higher than 1 and up to 18 for Mut101 (Table 4). Mut101 and the other IN-LEDGF inhibitors also differ from protease inhibitors (PIs) since PIs are active only in MR and completely inactive in SR assays. The Mut101 series of IN-LEDGF inhibitors have an unprecedented mixed profile with moderate ARV activity in SR and more potent activity in MR infection assays. The two doseresponse curves of Mut101 ARV showed that there was no or minimal activity detectable in the SR assay at the concentration resulting in maximum MR activity ( Figure 6A). This suggests that the contribution of integration inhibition (estimated by SR assay) to Mut101 overall ARV activity is minimal at this concentration. This contribution becomes significant only at much higher concentrations, such as those used for TOA experiments. Previous infection experiments studying LEDGINs and tBPQAs ARV activity were performed mostly in MR assay. We analyzed the behavior of a tBPQA, racemic BI-D [48] (structure shown in Additional file 1: Figure S2), to determine if the behavior of the Mut101 compound series is shared by other LEDGINs and tBPQAs. We found a similar discrepancy between high EC 50 in SR (2.4 μM) and much lower EC 50 (0.17 μM) in MR assay.

Mut101 also promotes a post-integration block producing defective HIV-1 progeny virions
The discrepancy between potent ARV activity in MR assays and moderate activity in SR assays, distinguishes Mut101 from INSTIs that specifically block HIV integration. One explanation could be that Mut101 treatment results in a second ARV activity at a late stage of the replication cycle, post-integration. We used the HeLa-LAV system in which the HeLa cell line has been transduced by  HIV-1 LAV virus [49] to test this hypothesis. HIV-1 LAV is constitutively integrated in this cell line and HeLa-LAV cells produce HIV-1 LAV virions that cannot reinfect the cells as they do not express CD4 on their surface. Only drugs that could block virus production at the post-integration step of the HIV-1 replication cycle are expected to be active in this cell line. We treated HeLa-LAV cells with Mut101, Raltegravir, Saquinavir (SQV) or DMSO (as a negative control). The infectivity of viruses produced in the presence of these compounds was tested in TZM indicator cells expressing luciferase and by infection of MT4 cells. The design of this experiment is schematized in Figure 6B. The amount of p24 produced with virus treated by Mut101 was comparable to viruses treated with Raltegravir, DMSO or Mut063 an inactive analogue of Mut101 ( Figure 6C). In contrast, luciferase assay in TZM cells showed that Mut101 and SQV treatments resulted in strong virus infectivity defects; viruses produced in the presence of Raltegravir, DMSO or Mut063 had no infectivity defect ( Figure 6D). These results were confirmed by determining the cytopathic effect of infected MT4 cells using a CellTiter-Glo® assay ( Figure 6E). The infectivity defect was not due to a residual concentration of Mut101 used during virus production since the virus stock was diluted 2000 times, to an inefficient concentration much below its EC 50 . We can also rule out a virucidal effect of Mut101 on virus particles released in the supernatant as Mut101 was unable to inactivate free virus once released in the supernatant of producing cells. Altogether, these results are strongly in favor of a defect provoked at a postintegration step by Mut101 treatment. This defect is additional to the block at integration detected above by the TOA experiment. Western blot using anti-p24 antibody did not detect any perturbation of Gag maturation and CA p24 content in defective virions or in Mut101treated HeLa-LAV cell lysates (data not shown).
A post-integration defect promoted by Mut101 treatment requires Mut101 binding to the LEDGF-binding pocket of IN  Figure 7A). Mut101 had no significant binding to the mutated IN-CCD T174I ( Figure 7B). HIV-1 NL4-3 wt and the NL4-3 IN T174I mutant virus were produced by HEK293T cell transfection in the presence of Mut101, SQV, Raltegravir, Mut063 or DMSO. Virions were harvested and used to infect MT4 cells (as schematized in Figure 7C); their infectivity was tested using a cytopathic CellTiter-Glo® assay. As shown in Figure 7D, NL4-3 wt virus (blue bars) produced in the presence of Mut101 was inactivated and the viability of MT4 cells infected by this virus was preserved. In contrast, the mutant virus T174I (red bars) was insensitive to Mut101 treatment and MT4 cells were fully infected and  their viability abrogated. Both wt and T174I viruses were sensitive to and inactivated by SQV treatment. Raltegravir treatment during virus production had no effect on either virus; these retained full infectivity which was comparable to that observed after DMSO or Mut063 treatment. These results demonstrate that integrase is indeed the unique target of Mut101 for its ARV activity, both at the integration and post-integration steps of the HIV-1 replication cycle. Our co-crystallographic studies with Mut101 bound to IN-CCD allowed us to detect conformational changes resulting from compound binding in the binding site of inhibitors. The structural changes observed when Mut101 is bound to IN confirm and explain the allosteric effect of the IN/LEDGF interaction inhibitor which acts at the post-integration steps. We evidenced a direct correlation between allosteric changes with atomic details and functional effect on IN upon Mut101 binding. Our experiments enabled us to address important questions regarding the unicity or multiplicity of the mechanism of action of these inhibitors, the respective contributions of these inhibitory activities to overall ARV activity, and the specific mode of action of these new ARV agents. We investigated the respective contributions of the two mechanisms to the global ARV activity of these compounds. SR infection assays reflect the activity of an ARV compound during an early step of the HIV replication cycle (up to integration), and MR infection assays reflect global ARV activity. We showed that the post-integration inhibition of the HIV-1 replication cycle is the major mechanism contributing to global Mut101 ARV activity. There was no or minimal ARV activity detectable in SR infection assay at the same Mut101 concentration that achieved 100% inhibition of HIV-1 infection in the MR infection assay. A higher concentration of Mut101 was required to detect ARV activity in the SR assay since its EC 50 in this format (9 μM) was 18 times higher than its EC 50 in MR infection assay (0.49 μM). TOA experiments used a Mut101 concentration (25 μM) that was high enough to permit 100% of ARV activity in the SR infection assay. Our study demonstrates that Mut101 and the other INLAIs of this series are not acting mainly as inhibitors of HIV-1 integration. This is in contrast to early studies reported on LEDGINs, based on MR infection experiments performed at saturating inhibitor concentration, that suggested they act as integration inhibitors [18]. HIV-1 integrase is the unique target of Mut101 for its ARV activity. However, the major action of Mut101 and other related INLAIs is as post-integration inhibitors producing defective infectious HIV-1 virions. Mut101 displays weak activity at early stage integration and potent activity at late stage production of defective virions. We then explored how a compound acting on a unique target (IN) and on a unique binding site (the LEDGF-binding pocket), displays such a difference between its potency on two ARV activities. The  [40]. The EC 50 of racemic BI-D ARV activity was between 2.4 μM and 2.9 μM when tested on wt cells but between 0.16 μM and 0.20 μM (15 to 18 times lower) on LEDGF KO cells, a result not significantly altered by HRP2 disruption. In contrast, the EC 50 of Raltegravir was similar in each cell type. The authors suggest that LEDGF, present in wt cells but not in LEDGF KO cells, can compete with BI-D for binding to the LEDGF-binding pocket of IN. In the presence of a LEDGF competitor in wt cells, the concentrations of BI-D required to achieve similar ARV activity are higher than when LEDGF is absent in KO cells. Strikingly, we found that the EC 50 of BI-D ARV activity on MT4 human cells infected with HIV-1 NL4-3 was 2.4 μM ± 0.5 in SR and 0.17 μM ± 0.03 in MR infection assays. This is very similar to the result found by Wang et al. (Table 5), although they worked with mouse cells and we worked with human cells. The data strongly suggest that a mechanism similar to that observed by Wang et al. (LEDGF competition in SR assay and no competition by LEDGF in MR assay), could explain the difference in ARV activity we found for INLAIs assayed in SR and MR infection assays. These data, and our in vitro data showing that LEDGF can compete with Mut101 for binding to IN, support the model illustrated in Figure  This model suggests that the activity of a protein-protein interaction inhibitor (in this case, concerning the interaction between a viral and a cellular protein) is governed not only by its intrinsic affinity for its target, but also by the cellular compartment in which it is acting. It is the presence or absence of the partner protein of the inhibitor target that could, by competitive binding, negatively affect the level of inhibitor activity.

Discussion
The activity of Mut101 and other INLAIs, at the step of integration, may be explained by impairment of IN-LEDGF interaction and their allosteric inhibitory effect on IN strand transfer catalytic activity. However, we need to understand what molecular mode of action of these compounds explains the post-integration block. Gag maturation and CA composition of defective virions produced in the presence of these compounds was normal [42,43] (E. Le Rouzic unpublished results), suggesting that there is no putative effect on maturation of the Gag precursor. We also know that Mut101 does not inhibit viral protease (D. Bonnard unpublished data). A post-integration stage defect could be related to IN conformational change resulting from compound binding to the LEDGF-binding pocket and IN-IN interaction enhancement ( [42,43] and this study). We showed, for the first time, that INLAIs promoted long range conformational change when they bind to IN-CCD, affecting residues far away from the compound binding site. Such IN conformational change could negatively affect the formation of the stable synaptic complex (SSC) [37], or influence the currently undefined roles of IN during late stages in the HIV-1 replication cycle [51]. Interestingly, it was lately reported [42][43][44][45] that treatment by IN-LEDGF allosteric inhibitors during virus production resulted in a defect in virion morphology with eccentric electron-dense HIV core. Further work is required to answer these questions and defective viruses produced in the presence of Mut101 could be valuable tools for these studies. The LEDGF-binding pocket lies at the dimeric interface of IN, a region crucial for the formation of an active oligomerization state of IN required for its enzymatic activity and specificity [52][53][54]. INLAIs make contacts to both subunits of an IN dimer and promote IN conformational change toward inactive oligomers. These inhibitors should therefore be considered as interfacial inhibitors that bind selectively to macromolecular machine interfaces and often promote allosteric effects [55].
Interestingly, INSTIs that bind at the interface of the IN-DNA-Mg 2+ complex [2] are also considered as archetypal interfacial inhibitors [55].

Conclusion
The dual mode of action of Mut101 compound series, at two different steps of the HIV replication cycle, is unique and unprecedented in all classes of ARV drugs. This could confer a great advantage to this class of ARV compounds from a therapeutic point of view, provided that clinically efficient concentrations can be reached to inhibit also virus replication at integration. The absence of antagonism between Mut101 compounds and INSTIs or the other classes of drugs currently on the market supports their potential for future ARV therapy.
Several acronyms have been proposed for this class of compounds: LEDGIN [18], NCINI [36] and ALLINI [37] have been suggested to underline their mode of action either as LEDGF-IN inhibitors or as Allosteric IN inhibitors. We would like to propose the acronym of INLAI, standing for 'IN-LEDGF Allosteric Inhibitor'. This takes into account both the importance of their interference with LEDGF binding to IN and their powerful allosteric inhibitory activity on IN. Our acronym links both activities in the mode of action and highlights that the binding site of these compounds on IN is the LEDGF-binding pocket.

Virus strains and recombinant HIV-1 molecular clones
HIV-1 NL4-3 and NL4-3Δenv-luc molecular clones were obtained from the NIH AIDS Research and Reference Reagent Program. The SpeI-SalI fragment from pNL4-3 containing the full pol gene was cloned into the pUC18 plasmid. In vitro mutagenesis was performed with the Pfu Turbo (Stratagene) and specific sets of primers to engineer the RT double mutant K103N/Y181C. The mutated fragment was validated by sequencing (Eurofins) and cloned back into pNL4-3 to generate a HIV-1 mutant molecular clone (used as a NNRTI-resistant virus). The molecular clone containing L10R/M46I/L63P/V82T/I84V mutations within the PR-coding region [56] was used as a PR-resistant virus (PI); the clone with M41L/ D67N/T69N/K70R/T215F/K219E within the RT-coding region [57] was used as a NRTI-resistant virus; the clone with M41L/D67N/K103N/M184V/L210W/T215Y within the RT-coding region [57] was used as a NRTI and NNRTI-resistant virus (Multi-drug in this study). PI, NRTIs and Multi-drug resistant clones were obtained through the AIDS Research and Reference Reagent Program. The molecular clone containing G140S/Q148H within the IN-coding region obtained from J-F Mouscadet [58] was used as the INSTI-resistant virus.

Viral stock
293 T (2.2 10 6 cells) were transfected with 6 μg pNL4-3 proviral plasmids (wild-type or drug resistant) using X-tremeGENE 9 reagent (Roche). Cells were washed 24 h later and cell supernatants were collected 48 h post-transfection and stored at −80°C. Single-round viral stocks were produced by co-transfecting pNL4-3Δenv with VSV-G envelope expression vector. Supernatants were collected 2 days after transfection. All viral stocks were quantified for p24 antigen using the Alliance HIV-1 p24 Antigen ELISA (PerkinElmer) and titrated to measure the quantity of infectious particles per mL by infecting TZM-bl indicator cells.

Antiviral assay in MT-4 cells
MT-4 cells growing exponentially at the density of 10 6 / mL were infected with HIV-1 strain NL4-3 at a MOI (multiplicity of infection) of 0.001 for 2 h. The cells were washed with PBS and aliquoted, using 100 μL fresh complete RPMI, into 96-well white plates (Corning) in the presence of different concentrations of compounds. The effective concentration of compound required to inhibit 50% (EC 50 ) of HIV-1 replication was determined after 5 days using the CellTiter-Glo® luminescent reagent (Promega) to quantify cell viability.

Replication-defective-HIV assay
MT-4 cells (growing exponentially at the density of 10 6 /mL) were infected with VSV-G-pseudotyped NL4-3Δenv-luc at a MOI of 0.0001 for 90 minutes. The cells were washed with PBS and aliquoted, using 100 μL fresh complete RPMI, into 96-well white plates (Corning) in the presence of different concentrations of compounds. Luciferase expression was quantified after two days using the One-Glo™ luciferase assay (Promega).

Cytotoxicity assays
Growth inhibition was monitored in a proliferating human T-cell line (MT-4) with different concentrations of compounds. ATP levels were quantified using the CellTiter-Glo® luminescent reagent (Promega) to measure the ability of a compound to inhibit cell growth, an indication of the compound's cytotoxicity. Cytotoxicity was evaluated at either day 2 or day 5.

Time-of-addition experiment
MT-4 cells in a 96-well microtiter plate (10 5 cells per well) were infected with pseudotyped HIV-1 NL4-3 strain at a MOI of 0.001. Compounds were added to single-round infection assays at different time points after infection (0, 2, 3, 4, 5, 6, 8 and 24 h). RAL, NVP and Mut101 were added at 80 nM, 2 μM and 25 μM, respectively. This corresponded to between three and ten times their EC 50 as determined by a drug susceptibility assay (CT-Glo).

Quantification of viral cDNA by real-time PCR
Prior to infection, viral stocks were treated 1 h at 37°C with 100 U per mL of DNAseI (Roche Applied Science). MT4 cells (6x10 6 ) were infected with virus at MOI = 0.001. At 7 h, 24 h and 48 h post-infection, cells were harvested, washed twice in PBS and DNA was extracted using the QIAamp Blood DNA Minikit (Qiagen). Quantifications of viral DNA were performed by real-time PCR using the LightCycler 480 system (Roche Applied Science). Primers, probes, and PCR run conditions were described previously [59]. The copy number of HIV-1 late reverse transcription product (LRT) was determined using standard curves obtained by amplification of cloned DNA containing the matched sequences. The copy number of integrated HIV-1 DNA was determined in reference to a standard curve generated by concomitant two-stage PCR amplification of a serial dilution of the standard HeLa HIVR7-Neo cell DNA [59]. Copy numbers of each viral form were normalized with the number of cells obtained by the quantification by PCR of the β-globin gene according to the manufacturer instructions (Roche Applied Science).

Constructions of epitope-tagged proteins
The His 6 -LEDGF plasmid has been previously described [60]. The plasmid encoding GST-Flag-IBD/LEDGF was constructed by cloning the LEDGF DNA sequence (encoding residues 342 to 507) in fusion with the Flag epitope into pGEX-2 T (GE Healthcare). His 6 -IN plasmid corresponds to pINSD.His and has been previously described [61]. The IN A128T mutant was generated by site-directed mutagenesis from pINSD.His. The full length Flag-tagged integrase sequence from NL4-3 was PCR amplified and cloned between the BamHI and XhoI restriction sites of a pGEX-6P1 vector (GE Healthcare) to generate the expression plasmid GST-Flag-IN. His-CCD and GST-Flag-CCD were obtained by cloning the integrase region (residues 50 to 202, encoding the catalytic core domain) from pINSD.His. Sol [62] into pET15b and pGEX-2 T-Flag, respectively. CCD contains the F185K mutation which greatly improves the solubility of the recombinant protein.
The CCD T174I mutation was introduced into the His-CCD plasmid by site-directed mutagenesis.

Purification of recombinant proteins
Frozen cells pellets from one liter culture were resuspended in 3.5 mL of integrase buffer (50 mM HEPES pH 7.5, 1 M NaCl, 7 mM CHAPS, 5 mM MgCl 2 , 2 mM β-mercaptoethanol, 10% glycerol) (for full length integrase) or the same buffer in a two-fold water dilution (for integrase CCD), containing Complete™ protease inhibitor cocktail (Roche) and benzonase (Sigma). Cells were disrupted using 25 g -30 g, 150-212 μm glass beads (Sigma) and vortexed at 4°C for 10 min. Glass beads were washed three times with 15 mL extraction buffer and whole cell lysate was centrifuged at 109,000 g (R max ) for 1 h at 4°C in a Beckman XL80K ultracentrifuge.
His 6 -tagged IN wt or A128T, or His 6 -tagged IN-CCD lysate was loaded at 3 mL/min on a 5 mL His-Trap FF crude column (GE Healthcare) previously equilibrated with integrase buffer or CCD buffer, respectively, containing 20 mM imidazole. Samples were washed until OD 280nm returned to baseline and bound proteins were then eluted using a 20 to 500 mM imidazole gradient over 20 column volumes. Pooled fractions were concentrated to 2.5 mL using Amicon Ultra 15™ 10 K centrifugal filter devices (Millipore) at 4,000 g and 4°C. Concentrated protein was loaded on a Superdex 200 16/600 PG column (for full length IN) or a Superdex 75 16/600 PG column (for IN-CCD) (GE Healthcare), previously equilibrated with integrase buffer at 4°C. Chromatography was performed at 4°C. The presence of His 6 -Tag IN/CCD in collected fractions was assessed by electrophoresis on NuPAGE Bis-Tris 10% acrylamide gels with MES as the electrophoresis buffer (Invitrogen). Proteins were stained using Imperial Protein Stain TM (Thermo Scientific Pierce). Pooled fractions from Superdex200 or Superdex75 separation were concentrated and stored at −80°C until further use. GST-tagged Flag-CCD and GST-tagged Flag-IBD lysates were loaded at 0.25 mL/min on a 20 mL Glutathione Sepharose 4 Fast Flow (GE Healthcare) column. Bound proteins were eluted using integrase CCD buffer with 20 mM reduced glutathione. Purification was completed as described above. Flag-IN was prepared from a GST-Flag-IN fusion protein using the pGEX-6P expression system (GE Healthcare). After adsorption to the Glutathione Sepharose 4 Fast Flow column, protein corresponding to the 1 liter culture extract was digested by 250 units of PreScission Protease (GE Healthcare) for 16 hours at 4°C. Cleaved protein was eluted by restarting the buffer flow over the column. Purification was carried out by gel filtration on Superdex 200, as described above. rGST was purified on Glutathione Sepharose 4 Fast Flow and Superdex 75 16/600 PG columns as described above but using a PBS buffer.

HTRF®-based CCD-IBD interaction assay
All HTRF® conjugated monoclonal antibodies were purchased from Cisbio Bioassays. IN-CCD/LEDGF-IBD HTRF® assay was performed in 384-well low volume black polystyrene plates (Corning) in CCD-IBD assay buffer (25 mM HEPES pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 0.4 M KF, 0.1% bovine serum albumin, 1 mM DTT). 2 μL of 3-fold serial dilutions of inhibitory compound in 25% DMSO were preincubated for 30 min at room temperature with 8 μL of IN-CCD mixture (75 nM His 6 -IN-CCD, 17 nM XL 665 -conjugated anti-His 6 monoclonal antibody). Then, 10 μL of LEDGF-IBD mixture (20 nM GST-Flag-LEDGF-IBD, 1.8 nM Europium cryptate-labelled anti-GST monoclonal antibody) were added and the plate was incubated for 2.5 h at room temperature before reading the time-resolved fluorescence in a PHERAstar Plus (BMG Labtech) with HTRF module (excitation at 337 nm, dual emission at 620 nm and 667 nm). The HTRF ratio was converted to % inhibition and analyzed by fitting with a sigmoidal dose-response equation with Hill slope to determine the compound IC 50 . Biotin-LTR preprocessed donor DNA, 0.20 μM Digoxigenin (DIG)-Target DNA) was added and the plate was incubated for 2 h at 37°C. The reaction was stopped by addition of 60 μL stop mixture (20 mM Tris-HCl pH 7.6, 0.4 M NaCl, 10 mM Na 2 EDTA, 0.1 mg/mL salmon sperm DNA) and the volume transferred to Reacti-Bind high-binding capacity streptavidin-coated white plates (Thermo Scientific Pierce). After 1 h incubation at room temperature under gentle shaking, integrase and unjoined DNA were removed by three washes with 200 μL wash solution 1 (30 mM NaOH, 0.2 M NaCl, 1 mM Na 2 EDTA). 100 μL of 2000-fold diluted HRP-conjugated anti-DIG Fab (Roche Applied Science) was added and the plate was incubated for 1 h at 37°C. Unbound antibody was removed with wash solution 2 (PBS pH 7.4, 0.05% Tween-20, 0.1% bovine serum albumin), 100 μL of SuperSignal Femto ELISA substrate (Thermo Scientific Pierce) was added and chemiluminescence was immediately read in a PHERAstar Plus with LUM-plus module. The signal, converted to % inhibition, was analyzed by fitting a sigmoidal dose-response curve to determine IC 50 and the inhibition plateau.
IN activity assays -3′-processing, strand transferwere carried out at 37°C with the full-length HIV-1 IN, in a buffer containing 10 mM HEPES (pH 7.2), 1 mM DTT, 7.5 mM MgCl 2 in the presence of 6.25 nM DNA (3′-processing) or 12.5 nM DNA (strand transfer) as described previously [64]. For negative control, 100 mM Na 2 EDTA was added to the reaction before incubation. Products were separated by electrophoresis in denaturing 16% acrylamide/urea gels. Gels were analysed with a Molecular Dynamics STORM phosphoimager and quantified with ImageQuant™ 4.1 software.

Size exclusion chromatography (SEC) experiments with IN liganded with Mut101 and BI-D compounds
SEC was performed with a Superdex 200 10/300 GL column (GE Healthcare) using a flow-rate of 0.4 mL/min in buffer containing 50 mM HEPES, pH 7.5, 1 M NaCl, 7 mM CHAPS, 5 mM MgCl 2 , 10 mM DTT, 10% glycerol at room temperature. His 6 -IN wt (21 μM) or His 6 -IN A128T (40 μM) was incubated for 10 min with 100 μM BI-D or Mut101 before injection on the column. Protein elution was monitored at 280 nm.

Biacore experiments
Experiments were carried out using a Biacore 3000 instrument (GE Healthcare) at 25°C. An anti-GST antibody (GST Capture Kit, GE Healthcare) was immobilized on two flow-cells of a CM5 sensor chip by amine coupling according to the recommendations of the manufacturer. GST-Flag tagged IN CCD proteins (wild type and T174I mutant) at 68 μg/mL in HBS-EP buffer (GE Healthcare) were captured on one flow-cell (8 min injection at 10 μL/ min) while recombinant GST (60 μg/mL in HBS-EP buffer, 8 min injection at 10 μL/min) was injected on the other flow-cell and used as a reference. Kinetics experiments with Mut101 were carried out at 60 μL/min with a 3 min injection of each dilution of the compound in HBS-EP followed by 10 min dissociation. Sensorgrams were evaluated using BiaEvaluation 3.2 software.

Structural studies
Crystallization was performed by the hanging-drop vapordiffusion method at 297 K in 24-well plates. The catalytic domain (CCD) of HIV-1 IN with mutation F185K was expressed and purified as previously described [62]. Prior to any crystallization experiment, the protein was simultaneously dialyzed and concentrated at 277 K with an Amicon Ultra-10 device (Millipore) equipped with a 10 kDa cut-off dialysis membrane. The dialysis solution was 50 mM MES-NaOH pH 5.5, 50 mM NaCl and 5 mM DTT. The protein was concentrated to between 3 mg/mL and 5 mg/mL.
Each hanging-drop consisted of 3 μL protein solution and 3 μL reservoir solution, with 500 μL reservoir solution in the well. Initial screening was carried out using Qiagen kits (Classics & JCSG+) and positive hits were then optimized. The optimized reservoir solution consisted of 1.16-1.36 M ammonium sulfate, 50 mM sodium cacodylate-HCl pH 6.5. The crystals grew to approximate dimensions of 0.2 x 0.2 x 0.4 mm within one week. They were soaked with the Mut101 ligand for 5 days before data collection by adding a 10 mM stock solution of the inhibitor to the drop. The crystals were plunged in oil (FOMBLIN Y LVAC 14/6 from Aldrich) for a few seconds and cryo-cooled in a stream of liquid nitrogen at 100 K. All data were collected at a temperature of 100 K and processed with XDS [65]. All diffraction data were collected using a Pilatus 2 M detector on beamline X06DA (PXIII) at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland. Structure determination was carried out using the CCP4 suite of programs [66]. The structures of the integrase, both in complex with the Mut101 inhibitor or not, were determined by molecular replacement using the program MOLREP [67] and PDB entry 1BHL [68] as the starting model. The models were built manually using the program Coot [69] and refined with the program REFMAC [70]. Arp/Warp [71] was used for the automatic ligand [72] and water molecule fitting. Structures and structure factors have been deposited in the PDB with codes 4LH4 (IN CCD) and 4LH5 (IN CCD with Mut101 inhibitor).
All experiments have been performed under Authorization Number 5606 CA-I, assigned by the French Ministry of Research for work with genetically modified organisms.

Additional files
Additional file 1: Additional methods, figures and tables.
Additional file 2: The movie was generated with PyMOL [73]. The intermediate structures between the initial and final state where generated using the morphing option in Pymol. The two IN monomers are colored in red and gold. The magnesium ion is represented as a green sphere and the coordinating residues of the magnesium and in the Mut101 pocket are represented in sticks. The solvent accessible surface coloring is in red and gold for the carbon atom of the corresponding monomer with the nitrogen in blue, the oxygen in red and sulfur in yellow. Mut101 is represented in cyan.