Molecular basis for the regulation of retroviral nucleosomal integration by chromatin compaction

Abstract

Cellular chromatin is the first non-reversible contact point between incoming infectious agents, such as integrating viruses, and their host genomes. Retroviral integration depends on interactions between the viral intasome and host chromatin, involving interfaces between integrase, target DNA, and nucleosomal histones. While factors like LEDGF/p75 and chromatin compaction influence these interactions, the molecular mechanisms regulating access to nucleosomal interfaces remain unclear. In this study, we dissected how incoming intasomes engage nucleosome surfaces to facilitate integration. Using biochemical assays and molecular docking, we demonstrated that HIV-1 and PFV intasomes specifically bind distinct nucleosomal surfaces. Mapping these interactions onto trinucleosomes and simulating dynamic docking revealed that neighboring nucleosomes can mask functional interfaces, modulating HIV-1’s ability to access its target. Conversely, PFV intasomes are less affected by chromatin compaction due to persistent accessibility of active nucleosomal sites. These findings provide the first structural insights into how chromatin architecture influences retroviral integration. Our results highlight divergent strategies by HIV-1 and PFV to engage chromatin, revealing how nucleosome properties govern the formation of active strand transfer complexes and emphasizing the evolution of virus-specific adaptation to chromatin landscapes.

Graphical Abstract

Graphical Abstract.

Introduction

Retroviral infection requires stable integration of viral DNA into the host chromosomes. In infected cells, integration is carried out by the pre-integration complex (PIC), which contains viral DNA bound to the virally encoded integrase (IN) enzyme, forming the intasome, and other viral/host factors whose nature remains poorly understood. Integration occurs selectively into specific “hot spots” of host chromosomes [1]. Many parameters govern the insertion site preference, including nuclear import processes, cellular targeting pathways, and nucleus architecture [2]. Many studies also indicate that structural features of chromatin, such as targeting epigenetic marks, nucleosomal compaction, and remodeling activities (extensive literature), are important for integration. However, to date, the molecular mechanisms that guide the PIC to specific genomic regions and enable the insertion of PIC-associated viral DNA into specific hot spots of human chromosomes are not fully depicted. Among these factors, the final interaction of the intasome with the nucleosomes, forming the target capture complex (TCC) and finally the strand transfer complex (STC), plays a critical step in the targeting process. The chronology of these events as well as the determinants governing this final chromatin interaction remain to be fully investigated.

The chromatin represents a dynamic complex of DNA and proteins that condense and regulate genetic information in eukaryotic cells. The nucleosome constitutes the core repeating unit of chromatin and consists of 147 bp of DNA wrapped around a histone octamer [3]. Nucleosomes are dynamic structures that can be assembled and disassembled by chaperones or chromatin remodelers and further regulated through posttranslational modifications. The local chromatin environment around the nucleosome imparted by these factors is thus expected to play an important role in establishing a successful integration event. In vitro, integration into nucleosomes can be recapitulated by the assembly of intasomes formed by multimers of IN, the number of which varies from 4 to 16 depending on the virus [4], and viral DNA. The use of these reconstituted intasomes in in vitro integration assays highlighted differences between retroviruses, showing their distinct preferences for chromatin compaction, which was further confirmed in cellulo [5, 6]. Functional engagement with chromatin requires the accessibility of the intasome/nucleosome interfaces, which likely contributes to differential sensitivity of intasomes to chromatin organization. Previous studies have revealed that the retroviral intasomes interact with both DNA and the histone components of the nucleosome, but through distinct binding modes. Indeed, while the PFV intasome binds the H2A/H2B dimer interfaces [7], HIV-1 IN was reported to interact preferentially with histone H4 in vitro, inducing conformational changes within the IN that stimulate integration [8, 9]. In addition, a distinct association with chromatin was also detected in the context of human chromosomes via in vitro chromatin binding assays, highlighting the importance of neighboring chromatin in regulating these functional associations [10].

Taken together, these data suggest that PFV and HIV-1 intasomes engage chromatin through distinct interaction modes. These differences underlie their differential behavior at chromatin contact sites. Integration sensitivity toward the chromatin structure neighboring the insertion site may rely on these different interfaces. In this study, we investigated how the intasome/nucleosome interaction influences their sensitivity to chromatin organization and contributes to the selection of specific regions of chromatin for integration. For this purpose, we compared HIV-1 and PFV retroviral intasomes, which exhibit clearly distinct chromatin preferences, using in vitro integration assays and biochemical approaches with defined chromatin substrates, including isolated nucleosomes and short nucleosome arrays, such as trinucleosomes. These substrates allowed us to assess the impact of local chromatin environment on integration. Complementary molecular docking analyses further revealed that HIV-1 and PFV intasomes differ in their modes of interaction with nucleosomal substrates, resulting in distinct capacities to engage compacted chromatin.

Materials and methods

Protein purifications and intasome assembly

Integrase of PFV has been purified and the intasome assembled as described in [9]. Briefly, PFV IN and its cognate vDNA were mixed, and assembled intasomes were purified by size-exclusion chromatography. Concordantly with published works, the PFV intasome, which is composed of a tetramer of IN, eluted around 11 ml. The HIV-1 IN was expressed in E. coli (Rosetta) and the cells were lysed in buffer containing 50 mM Hepes pH 7.5, 5 mM EDTA, 1 mM DTT, 1 mM PMSF. The lysate was centrifuged, and IN was extracted from the pellet in buffer containing 1 M NaCl, 50 mM Hepes pH 7.5, 1 mM EDTA, 1 mM DTT, 7 mM CHAPS. The protein was then purified on butyl column equilibrated with 50 mM Hepes pH 7.5, 200 mM NaCl, 1 M ammonium sulfate, 100 mM EDTA, 1 mM DTT, 7 mM CHAPS, 10% glycerol. The protein was further purified on heparin column equilibrated with 50 mM Hepes pH 7.5, 200 mM NaCl, 100 mM EDTA, 1 mM DTT, 7 mM CHAPS, 10% glycerol. LEDGF/p75 was expressed in PC2 bacteria, and the cells were lysed in lysis buffer containing 20 mM Tris–HCl pH 7.5, 1 M NaCl, 1 mM PMSF, added lysozyme, and protease inhibitors. The protein was purified by nickel-affinity chromatography and the His-tag was removed with 3C protease at 4°C overnight. After dilution down to 150 mM NaCl, the protein was further purified on SP column equilibrated with 25 mM Tris pH 7.5, 150 mM NaCl (gradient from 150 mM to 1 M NaCl). Then DTT was added to 2 mM final and the protein was concentrated for gel filtration. Gel filtration was performed on a Superdex 200 column (GE Healthcare) equilibrated with 25 mM Tris–HCl, pH 7.5, 500 mM NaCl. Two mM DTT was added to eluted protein that was then concentrated to ~10 mg/ml. The C-terminal domain (CTD, residues 220–270) of HIV-1 integrase was purified as previously described [10]. The CTD was expressed in BL21(DE3) Escherichia coli cells grown in Luria-Bertani (LB) medium supplemented with 50 µg/ml ampicillin and 10% (w/v) sucrose at 37°C with constant shaking at 220 rpm. When the culture reached an OD600 of 0.8, expression was induced by adding 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), and incubation continued overnight at 25°C with shaking at 190 rpm. Cells were harvested by centrifugation at 4000 × g for 20 min at 4°C, resuspended in 10 g/l NaCl, washed, and stored at −20°C before purification. Typically, 4–6 g of wet cell mass was obtained per liter of culture. For purification, the cell pellet was thawed and resuspended in lysis buffer (25 mM HEPES, pH 8.0, 1 M NaCl, 10 mM imidazole, pH 8.0) at a ratio of 10 ml buffer per gram of cells, with the addition of protease inhibitor cocktail (1 tablet per 50 ml). The cells were lysed by ultrasonication using a 13 mm probe, applying 2-s pulses (on/off) at 40% amplitude on ice to prevent overheating. Cell debris was removed by ultracentrifugation at 185 000 × g for 1 h at 4°C using a Ti45 rotor. The clarified lysate was applied to a 5 ml HisTrap FF Crude column (Cytiva) equilibrated with lysis buffer. Unspecifically bound proteins were removed by a stepwise increase of imidazole concentration: 8 column volumes (CV) at 10 mM imidazole, 8 CV at 20 mM, and 8 CV at 40 mM. The protein of interest was then eluted using a linear gradient from 40 to 500 mM imidazole over 12 CV. Fractions containing the purified protein were pooled and concentrated to 0.5 ml using a 15 ml Amicon Ultra-3K centrifugal filter unit (Millipore) before loading onto a Superdex 75 16/60 gel filtration column (Cytiva) pre-equilibrated with storage buffer (25 mM HEPES, pH 8.0, 1 M NaCl). The appropriate fractions were collected and concentrated to ∼5 mg/ml using a 4 ml Amicon Ultra-3K filter unit, aliquoted, and flash-frozen in liquid nitrogen for storage at −80°C. The PFV CTD domain was cloned, produced, and purified as HIV-1 CTD on the basis of the structural definitions from previous intasome studies [7].

Nucleosome assembly

Mononucleosomes and trinucleosomes were assembled as previously described for chromatin assembly [11]. Briefly, 5 µg of biotinylated 147 bp Widom fragment or the corresponding trinucleosomal DNA (Epicypher) was mixed with an excess of 10 µg of human native, H3K36me3-modified, tailless, or mutated recombinant octamers produced in E. coli (purchased from the “Histone Source” Protein Expression and Purification (PEP) facility from the Colorado State University, https://histonesource-colostate.nbsstore.net) in Tris–HCl pH 7.7 and 2 M NaCl in 100 µl final volume. Salt dialysis was then performed to decrease the salt concentration to 0 using Slide-A-Lyzer MINI dialysis device, 7 k MWCO (Fisher Scientific). Assembly was checked by electro-mobile shift assay (EMSA) on 8% native PAGE stained with SYBR safe and sodium dodecyl sulfate–polyacrylamide gel electrophoresis stained with InstantBlue. The acidic patch mutant H2A E92K was purchased at Epicypher (ref. SKU: 16-1030). For reconstituted chromatin fibers, we used an array of 12 repeats of 146 bp 601 positioning sequences separated by 50 bp linkers, previously used and demonstrated its ability to form 30 nm chromatin fibers in vitro [22, 23]. The arrays were assembled via salt dialysis in the presence of core histone octamers and the linker histone H1 to closely replicate the native chromatin environment. The assembly was validated using EMSA after MNase digestion.

Drugs

The CA3 compound was synthetized as previously described [12]. Dolutegravir was purchased at INTERCHIM SA.

Integration assays

Integration assays using purified assembled intasomes were performed as previously [13]. Briefly, 100 ng of DNA assembled on 100 ng of histone octamers (24 nM final concentration) were incubated with 30 nM final concentration of purified intasome in 100 mM NaCl, 20 mM BTP pH 7, 12.5 mM MgSO₄ (for PFV intasome) or in 20 mM HEPES pH 7, 7.5% DMSO, 8% PEG, 10 mM MgCl₂, 20 µM ZnCl₂, 100 mM NaCl, 5 mM DTT final concentration (for HIV-1 intasome) in a final volume of 40 µl. The mix was then incubated at 37°C for 15 min. Then reaction was stopped by the addition of 5.5 µl of a mix containing 5% SDS and 0.25 M EDTA and deproteinized with proteinase K (Promega) for 1 h at 37°C. Nucleic acids were then precipitated with 150 μl of ethanol overnight at −20°C. Samples were then spun at top speed for 1 h at 4°C, and the pellets were dried and then resuspended with DNA loading buffer. Integration products were separated on an 8% native polyacrylamide gel.

For site sequencing, after integration reaction, deproteinization, and DNA precipitation, PCR was performed using the HU5pro (5′ GTGGAAAATCTCTAGCA 3′) and 601rev (5′ CTGGAGAATCCCGGTGC 3′) primers. The PCR products were then purified using the Wizard^® SV gel and PCR cleanup system (Promega, Ref. A9281) and cloned into pGEM^®-T easy vector (Promega, Ref. A1360). Cloned fragments were then sequenced using T7 primer.

Nucleosome bridging assay

The nucleosome bridging assay (NBA) was performed on the basis of typical intasome/nucleosome AlphaLISA previously set up [13]. MN-biotin and MN-DIG (2 nM final concentration) were incubated in buffer NBA (8.1 mM HEPES, pH 7.0, NaCl 100 mM, potassium acetate 13 mM, CaCl₂ 975 µM, ZnCl₂ 3.25 µM, DTT 3.25 mM, 1.25% DMSO). Streptavidin-coated Alpha Donor Beads (Revvity) were added at a final concentration of 5 µg/ml and then incubated at RT with MNs for 5 min before adding the anti-DIG-conjugated AlphaLISA acceptor beads at the same final concentration in the dark. Short spin was performed at 300 rpm and incubation was extended to 10 min at RT in dark. Then 1 µl of freshly assembled intasome was added at 16 µM final concentration of integrase. The plate was incubated at 37°C in a Victor Nivo reader for 15 min before measurement of interaction signals.

Molecular docking

To explore how the HIV-1 and PFV intasomes interact with the human trinucleosome, computational models of these complexes were constructed. The trinucleosome structure was sourced from the 6L4A PDB entry (12.30 Å resolution) [14], while the HIV-1 and PFV intasome models were obtained from the 9C9M (2.01 Å resolution) and 3S3M (2.49 Å resolution) PDB structures, respectively. Protonation of all the amino acid residues for both proteins was predicted using PlayMolecule’s Protein Prepare module at pH 7.4 [15]. Missing residues of the nucleosome were modeled based on PDB structure 1KX5 [16].

To predict the complexes between the trinucleosome and both the HIV-1 and PFV intasomes, the HADDOCK 2.4 web server was used [17]. HADDOCK 2.4 was used for docking, applying a multi-step approach where solutions were grouped based on a 2.0 Å RMSD cutoff and ranked using HADDOCK scores. Active residues were selected based on a previous work [8]. For the HIV-1 intasome, the selected residues were Y227, D229, R231, D232, W235, K236, and D253. For the human nucleosome, the selected residues were R19, K20, V21, L22, and the 51–56 bp region. For the PFV intasome, active residues were selected based on the works of Maskell et al. [7] and Wilson et al. [18]. For the nucleosome, these were K9 and R11 from the N-terminal of H2A and the C-terminal of H2B (T119, S120, A121, and K122) as well as the target area of the DNA (36–41 bp). For the intasome, the residues selected as active were (P135, Q137, K168, A188, P239, and T240 of both chains).

Cells culture, viral transduction, and viral DNA quantifications

K562 cells were plated in 96-multiwell plates at 20 000 cells/well using 100 μl of corresponding media: DMEM or RPMI containing 10% (v/v) fetal calf serum (FCS, Invitrogen) and 50 μg/ml of gentamycin (Invitrogen). Viral transductions were performed using pNL4.3-based pRRLsin-PGK-eGFP-WPRE VSV-G-pseudotyped lentiviruses carrying the wild type or mutated INs using an optimized MOI of 1 in a final volume of 100 µl. MOI was calculated from transducing units (TU) determined by FACS and quantitative PCR-based titration of viral preparations. Cell number was measured using the Countess II (Invitrogen) automatic cell counter. After 5 days, the cells were washed twice with PBS, treated with trypsin for 5 min at 37°C, and centrifuged for 2 min at 2000 rpm. The pellet was washed twice with PBS and resuspended in 200 μl PBS/200 mM EDTA. Fluorescence was quantified using 10 000 cells on a BD FACS CANTO II analyzer and DIVA software (BD Biosciences).

Total DNA (including integrated HIV-1 DNA and episomal HIV-1 DNA) was extracted mostly at 16 h post-transduction when most of the cell lines still showed an active RT process. More precise kinetics were conducted when mentioned. Viral DNA was extracted at 48 h post-transduction for integrated DNA quantifications using the QIAmp blood DNA minikit (Qiagen, Courtaboeuf, France) according to the manufacturer’s protocol. Elution was performed in 30 μl of H₂O. The total HIV-1 DNA was amplified by qPCR using the following primers: WPRE forward 5′-ATG GCT TTC ATT TTC TCC TCC TT-3′, WPRE reverse: 5′-CGG GCC ACA ACT CCT CAT-3′ (intermediate RT), LATE forward 5′-TGT GTG CCC GTC TGT TGT GT-3′, LATE forward 5′-CTG CGT CGA GAG AGC TCT GG-3′ (late RT), EARLY forward 5′-GCC TCA ATAAAG CTT GCC TTG A-3′; early reverse 5′-TGA CTA AAA GGG TCT GAG GGA TCT-3′. GAPDH forward 5′-GGT GAA GGT CGG AGT CAA CGG ATT-3′ and GAPDH reverse 5′-AGG GGT CAT TGA TGG CAA CAA TAT CC-3′ primers were also used for relative quantification of the viral DNA population. Integrated DNA was quantified by Alu qPCR. After initial denaturation step (95°C for 8 min), 100 ng of extracted DNA was amplified using 1 µM LM667, 3 µM Alu1, and 3 µM Alu2 primers for 12 cycles (95°C for 10 s, 60°C for 10 s, and 72°C for 2 min 50 s). After amplification, the product was 1/10 diluted in water, denatured (95°C for 8 min), and amplified in a second PCR using 3 µM lambda and 3 µM early RT R bis primers for 40 cycles (95°C for 10 s, 60°C for 10 s, and 72°C for 9 s). Control quantifications were performed using CD3g OUT and IN primers. LM667 5′-ATGCCACGTAAGCGAAACTCTGGCTAACTAGGGAACCCACTG-3′, Alu1 5′-GCCTCCCAAAGTGCTGGGATTACAG-3′, Lambda 5-ATGCCACGTAAGCGAAACT-3′, Early RT R bis 5′-GGCGCCACTGCTAGAGAT-3′, CD3g OUT Fwd 5′-ACTGACATGGAACAGGGGAA-3′, CD3g OUT Rev 5′-AGCTCTGAAGTAGGGAACATAT-3′, CD3g IN Fwd 5′-GGCTATCATTCTTCTTCAAGGTA-3′, CD3g IN Rev 5′-TTCCTGGCCTATGCCCTTTT-3′.

Results

HIV-1 and PFV intasomes show distinct functional nucleosome interfaces in vitro

Previous structural studies have shown that the PFV intasome targets the nucleosome through interaction with H2A/H2B dimer and nucleosomal DNA [7]. In turn, HIV-1 IN has been presumed to bind other histone components, such as the histone H4 tail [8, 9], suggesting that both intasomes may engage distinct regions of the nucleosome during integration. To further characterize these differences, we analyzed and compared the structure of the HIV-1 and PFV integration products obtained after integration onto mononucleosomes (MNs) assembled with the commonly used Widom 601 sequence, previously shown to provide suitable integration substrates for both models [11].

As shown in Fig. 1A, both PFV and HIV-1 intasomes assembled with their cognate FITC-labeled viral DNA were efficient at integrating into MNs. Analysis of the migration profile of the integration products using polyacrylamide gels revealed distinct insertion patterns for the two intasome types (Fig. 1A). Comparison of the distance of migration of the integration products with the size ladder (Fig. 1B) confirmed that these differences correspond to distinct integration positions. Cloning and sequencing of the integration products (Fig. 1C) confirmed the previously reported major PFV integration sites [7], located at the exposed major groove of the nucleosomal superhelix location (SHL) 3.5 positions, ~36 bp away from the dyad. Moreover, under our experimental conditions, an additional PFV insertion site was observed at ∼41 bp and ∼101 bp (Fig. 1C), consistent with secondary products observed in previous studies [7]. In contrast, for HIV-1, a major insertion site was detected in a different region of the nucleosome, with major insertion sites clustered at 51–56 bp (Fig. 1A–C). Mapping these positions to the nucleosome structure reported in Fig. 1D demonstrated that HIV-1 and PFV intasomes engage distinct regions of the nucleosome to mediate integration in vitro.

Figure 1.

In vitro integration selectivity of HIV-1 and PFV intasomes onto nucleosomes. Integration assays were performed using FITC-labeled intasomes, assembled in the presence of LEDGF/p75 for HIV-1, and Widom 601 nucleosomes. Integration products were analyzed on 8% polyacrylamide gels, visualized by fluorescence scanning or SYBR Safe staining (A). The positions of the full-site and branched half-site products are reported in the figure. Reactions performed without acceptor DNA are also reported as AI (autointegration) conditions. Distances of migration of the integration products were compared with molecular weight markers loaded under similar conditions (B). Integration products obtained in panel (A) were cloned in bacteria and sequenced. The positions of the main HIV-1 and PFV insertion sites were reported onto the 147 bp linear Widom 601 sequence (C). The corresponding insertion sites were subsequently located on the 3D structure of the 601 nucleosome (D).

These differences may be explained by different sensitivities toward nucleosomal structural constraints driven by specific interactions between INs and histone components and/or DNA structures. In particular, HIV-1 integration relies on the functional association between the intasome and histone tails, which may participate in the stabilization of the TCC and STC as well as their activation [8, 9]. To determine whether the selection of the insertion sites at the surface of the nucleosome was dependent on this functional IN/histone association, we compared HIV-1 integration on either native or tailless nucleosomes assembled on the typical Widom 601 positioning sequence and recombinant histone octamers. Deletion of histone tails significantly reduced the integration efficiency (Fig. 2A and B), consistent with previous studies [8], supporting a role of histone tails in intasome stabilization [9]. Nevertheless, the integration sites were located at similar positions in both types of nucleosomes (Fig. 2C). These data indicate that histone tails are not required for the selection of the nucleosomal integration site but instead promote efficient integration by stabilizing the incoming intasome and next STC complex in agreement with the previously reported stimulatory effect of the H4 tail [9].

Figure 2.

Effect of histone tails on HIV-1 insertion site selection at the surface of the nucleosome. Integration assays were performed using radiolabeled HIV-1 intasomes assembled in the presence of LEDGF/p75 and either native or 601 nucleosomes. Integration products were loaded on 8% native polyacrylamide gels (A) and integration was quantified after autoradiography from three independent experiments. Data are reported as means ± SD (B). A t-student statistical analysis was performed to determine the significance of the quantified differences. Insertion site positions along the nucleosome surface were determined by comparison of the distance of migration detected on gel (C).

Together, these findings suggest that the distinct chromatin engagement of HIV-1 and PFV intasomes is unlikely to be driven by differential interactions with histone tail components of the nucleosome. In turn, our results indicate that the binding of the intasome to nucleosomal DNA is a major determinant of integration site selection. Structural differences between retroviral intasomes, especially their quaternary structures, as previously reported [6], may also explain the differences in nucleosomal anchoring caused by differences in steric hindrance. To investigate these points, we further compared the physico-chemical properties of nucleoprotein complexes formed upon integration onto nucleosomes.

In vitro integration of HIV-1 occurs in highly ordered insoluble nucleocomplexes that are not part of aggregates

To further characterize the nucleoprotein complexes formed during retroviral integration into nucleosomes, we first assessed the solubility of the integrated products. For this purpose, integration assays were followed by sequential centrifugation and quantification of the soluble fractions. Centrifugation of the HIV-1 samples after integration followed by deproteinization of the supernatant, as schematized in Fig. 3A, led to a near-complete loss of the detectable integration products in the soluble fraction (Fig. 3B), indicating that these complexes are largely insoluble under experimental conditons. Addition of NaCl to 150 mM final concentration restored the solubility of the strand transfer complex-nucleosome assemblies (STC-nuc), leading to the recovery of most of the integration products, as observed on the polyacrylamide gel (Fig. 3C). In contrast, PFV-derived STC-nuc remained largely soluble in all tested conditions, consistent with their higher solubility (Fig. 3D; see all the quantifications in Fig. 3E).

Figure 3.

Solubility of the HIV-1 and PFV STC in vitro. Concerted integration onto biotinylated 601 mononucleosomes was carried out with either radiolabeled HIV-1 or PFV intasomes and integration products were centrifugated at the reported speed. HIV-1 intasomes were assembled in the presence of LEDGF/p75. Soluble and insoluble fractions were then deproteinized, and the integration products were quantified after capture using streptavidin magnetic beads and radioactivity quantification with scintillator counter [see schematization in (A)]. Results are reported as the % of product measured in the soluble fraction versus the total of products and as means of 3–4 independent experiments ± SD (B). Concerted integration was then performed similarly with addition of 150 mM NaCl final concentration into the reaction products before centrifugation at 6000 rpm for 5 min. Supernatants from HIV-1 (C) and PFV (D) were then deproteinized, precipitated, and loaded on 8% polyacrylamide gels and autoradiographed [quantification in (E)]. Data are reported as means from at least three independent experiments ± SD.

These data suggest that the loss of HIV-1 integration products during centrifugation performed after integration is primarily due to the low solubility of the nucleosomal functional STC-nuc complexes in contrast to PFV STC-nuc, which remain soluble. The formation of poorly soluble HIV-1 nucleoprotein complexes during integration in vitro may be driven by previously reported intasome stacks [19]. Indeed, primate lentiviral intasomes, including HIV-1, tend to assemble in “proto-intasome stacks” of variable length composed of repeating octameric units [20, 21]. These stacks are formed by domain swapping and can be limited by the addition of an excess of C-terminal domain (CTD) of IN [19]. To determine whether the high-order complexes carrying the nucleosomal integration products could be due to these intasome-mediated stacks, we tested the effect of the addition of purified CTD on nucleosomal activity.

As shown in Supplementary Fig. 1A, the addition of the HIV-1 CTD significantly enhanced HIV-1 nucleosome integration. In contrast, and consistent with the absence of stack formation reported for PFV intasomes, the PFV CTD failed to stimulate PFV integration (Supplementary Fig. 1C and D). To rigorously assess the specificity of the HIV-1 CTD effect, we conducted cross-assays. When tested in PFV integration assays, the HIV-1 CTD did not stimulate the reaction. Instead, a dose-dependent inhibition was observed at higher concentrations (Supplementary Fig. 2A), likely due to non-specific binding of the CTD to nucleosomal DNA, which may compete with intasome–DNA interactions. Conversely, the PFV CTD did not enhance HIV-1 integration (Supplementary Fig. 2B) and similarly inhibited the reaction at high concentrations, further supporting a competitive, non-specific interaction with nucleosomal DNA. Collectively, these results demonstrate that the stimulatory effect of the HIV-1 CTD on integration is HIV-1-specific and not a generic phenomenon. They reinforce the hypothesis that the HIV-1 CTD modulates integration by specifically altering the structure or dynamics of HIV-1 intasomes, likely through its previously described role in preventing stack formation—rather than through non-specific mechanisms such as DNA binding.

Taken together, these data indicate that HIV-1 integration occurs in vitro within large functional nucleoprotein complexes that embed the integration products as highly ordered structures that are not driven by intasome stacking. As these large complexes may involve multiple intasome–nucleosome interactions, we next investigated the possibility of a multivalent nucleosomal coordination process using nucleosome bridging assays.

HIV-1 STC bridges multiple nucleosomes in vitro

To better understand how nucleosomes are coordinated within retroviral STC, we set up the AlphaLISA assay schematized in Fig. 4A, using nucleosomes coupled to donor and acceptor beads in the presence of intasomes. As reported in Fig. 4B and experimental setup in Supplementary Fig. 3A, while no interaction signal between beads was detected without the intasome, the addition of the HIV-1 intasome under reaction conditions induced a strong, reproducible, dose-dependent interaction in the 5 to 25 nM intasome range for 4 nM nucleosome. In contrast, the PFV intasome did not induce any significant interaction signal between beads (cf. Fig. 4B). To account for non-functional integrase aggregates, we quantified total and soluble protein fractions after centrifugation, as described previously [9]. Data allowed us to determine that the functional HIV-1 soluble intasome fraction corresponds to ∼40%–60% of the total preparation. Subsequent correction of the intasome concentrations in the dose-dependent bridging experiment reported in Supplementary Fig. 3A led to the determination of an apparent optimal MN/IN bridging stoichiometry for this bridging of ∼0.5–3.75 intasome per nucleosome. This ratio has been further refined by conducting similar bridging experiments using soluble fractions obtained after intasome centrifugation (see Supplementary Fig. 3B), leading to an optimal interaction signal at 6 nM of intasome for 4 nM of nucleosome i.e. ∼1.5 intasomes for one nucleosome. These values are consistent with the model in which one intasome bridges two nucleosomes. Moreover, increasing intasome concentration did not increase the bridging signal (Supplementary Fig. 3B), indicating a specific interaction rather than unspecific IN-nucleosome aggregation.

Figure 4.

In vitro nucleosome bridging by retroviral intasomes (nucleosome bridging assay, NBA). The AlphaLISA-based bridging assay is schematized in (A). Retroviral intasomes, assembled in the presence of LEDGF/p75 for HIV-1 (Intasome with LEDGF/p75), are incubated with a mixture of MN coupled to donor and acceptor beads (i). If only one nucleosome is coordinated within one intasome, no AlphaLISA signal would be detected (ii). Coordination of nucleosomes coupled to the same type of beads will also lead to a lack of signal as reported in the figure. In contrast, the bridging of at least two nucleosomes within either intasome stacks, aggregates or within one intasome forming a TCC will lead to a detectable signal (iii). Addition of HIV-1 CTD can affect the bridging signal mediated by intasome/intasome stacking interactions (iv). Bridging assay has been carried out without or in the presence of HIV-1 or PFV intasome under optimized setup conditions (B). The bridging assay has been performed with the isolated HIV-1 intasome components (intasome assembled with LEDGF/p75, intasome assembled without LEDGF/p75, LEDGF/p75 in the presence of viral DNA and integrase plus LEDGF/p75) with similar concentrations when intasome was used (C). Experiments were conducted in the presence of intasome assembled in the presence of either wild-type LEDGF/p75 or LEDGFΔ141 deleted for the PWWP domain and compared with control conditions without intasome (buffer) (D). Increasing concentrations of 220–270 HIV-1 CTD (E) or CA3 and dolutegravir drugs (F) were added in the MN/HIV-1 intasome bridging assay. Nucleosome bridging signal of intasome assembled with D116A IN catalytic mutation was compared with wild-type intasome and control conditions without intasome (G). Nucleosomes carrying either an acidic patch H2AE92K mutation or deletion in histone tails were used in bridging assay (H). Data are reported as the means of at least three independent experiments ± SD. T-student tests have been performed.

Consistent with previous observation, HIV-1 IN protein alone did not produce any bridging signal (Fig. 4C), also supporting the mechanism of specific interaction rather than nonspecific aggregation due to incorrectly assembled components of the intasome. The HIV-1 IN LEDGF/p75 cofactor used in the intasome assembly was found sufficient to induce a bridging signal alone, suggesting that nucleosome coordination may also be mediated by this factor. Importantly, deletion of the LEDGF/p75 PWWP nucleosome interaction domain strongly decreased the intasome bridging property, confirming that the LEDGF/p75 nucleosome binding function plays a central role in this process (Fig. 4D).

Since previously observed intasome stacking may also lead to bridging signals, we tested the effect of adding increasing amounts of CTD to limit these in vitro nonphysiologically formed stacks. As reported in Fig. 4E, the addition of purified CTD did not affect the bridging signal, confirming that the interaction observed is mostly due to the coordination of multiple nucleosomes to a single intasome or STC rather than several clusters of nucleosomes from multiple bonds to multiple intasomes stacked together.

To further investigate whether nucleosome coordination depends on integration activity or intasome binding to multiple nucleosomes, we performed a nucleosome bridging assay under noncatalytic conditions using the dolutegravir (DTG) integration inhibitor. As reported in Fig. 4F, a similar coordination signal was observed in the presence of DTG, indicating that nucleosome coordination was not dependent on integration catalysis by STC. Similarly, catalytically inactive D116A intasomes retained full bridging activity (Fig. 4G), confirming that the nucleosome bridging was likely dependent on the intasome binding to the nucleosome rather than integration catalysis.

Given that binding of the intasome to the nucleosome may involve interactions among IN, LEDGF/p75, and histone tails [8, 9], we next address the contribution of these tails to this multivalent coordination. To investigate this, we first used CA3 compounds previously shown to bind histones and reported to inhibit the functional association of the intasome with the nucleosome [12, 13]. In contrast to DTG treatment, CA3 inhibited the bridging signal, supporting the importance of histone tails in the bridging process (see Fig. 4F). Furthermore, deletion of H3 or H4 also induced a ∼20% decrease in the bridging signal confirming that these histone regions are partially involved in optimal nucleosome bridging induced by HIV-1 intasome binding (Fig. 4H). In contrast, H2A and H2B tail deletion did not change significantly the intasome bridging property, confirming the specific contribution of H3 and H4.

Additional stacking between adjacent nucleosomes bridged by the intasome may accentuate the interaction signal. Such stacking is known to involve H4 tail interactions with acidic patches [22]. We thus further investigated whether this acidic patch of nucleosome stacking may be responsible for the detected bridging. We used the H2A E92K acidic patch mutant known to prevent H4 binding and nucleosome interactions [23]. As reported in Fig. 4H, the bridging signal was unaffected by the H2A E92K mutation, confirming further that the nucleosome bridging was unlikely due to nucleosome aggregation.

Taken together, these data suggest that a significant proportion of HIV-1 TCCs can bridge at least two nucleosomes under catalytic conditions relevant for retroviral integration. These results highlight the importance of nucleosome coordination, raising the question of the influence of adjacent nucleosomes and high-order chromatin architecture on the integration reaction.

HIV-1 and PFV intasomes are distinctively affected by neighboring nucleosomes

Contact with adjacent nucleosomes may affect access to the target catalytic nucleosome and, thus, integration sensitivity toward chromatin compaction. In particular, the distinct nucleosome interfaces found between PFV and HIV-1 intasomes may account for their previously observed differences in sensitivity toward chromatin architecture, as observed in integration assays using reconstituted polynucleosome templates [6, 8, 24].

To compare HIV-1 and PFV integration in different chromatin architectures, we performed in vitro integration assays using either isolated nucleosome templates or chromatin fibers reconstituted in the presence of the H1 histone linker (Fig. 5A). These fibers were formed of an array of 12 repeats of 146 bp 601 positioning sequences separated by 50 bp linkers, previously shown to form 30 nm chromatin fibers in vitro [25, 26]. Assays revealed that HIV-1 integration is strongly reduced into compacted chromatin, whereas PFV integration is significantly enhanced in compact chromatin conditions (Fig. 5B). These data further confirmed the distinct chromatin compaction preferences of the two types of intasome.

Figure 5.

Impact of adjacent chromatin on HIV-1 and PFV integration. Integration was performed in chromatin fiber containing an array of 12 × 601 sequences separated by 50 bp linkers assembled in the presence of histone H1. Assembly was monitored by MNase protection after 2 min treatment of assembled fiber and corresponding naked DNA (A). The efficiency of HIV-1 and PFV in vitro integration into naked DNA or fiber was compared by bead integration assays. Data are reported as percentage of integration into chromatin fibers; integration into MN is set up at 100% ± SD (three independent experiments) (B). T-student tests have been performed to evaluate the significance of the differences between conditions. MN or the trinucleosomes reported in (C) were then compared by beads integration assay using either PFV or HIV-1 intasomes. Data are reported in (D) as the mean percentage of integration in the trinucleosome structures, integration into MN is set up at 100% ± SD (three independent experiments). T-student statistical tests have been performed to compare each condition. In all these experiments, HIV-1 intasomes were assembled in the presence of LEDGF/p75.

The availability of functional intasome/nucleosome interfaces can be modulated differently by the compaction of chromatin. To better investigate this hypothesis and to more accurately study the impact of neighboring chromatin, we next compared integration onto MN and trinucleosome (Tri-nuc) structures with distinct internucleosome linker sizes (18 bp, 53 bp, and 106 bp), allowing us to directly monitor the regulatory role of neighboring nucleosomes and chromatin compaction, as illustrated in Fig. 5C. As reported in Fig. 5D, quantification of the total amount of HIV-1 integration catalyzed into either the isolated MNs or Tri-nuc showed that the HIV-1 integration efficiency of Tri-nuc was lower than that of MNs. This confirmed that HIV-1 integration is globally inhibited by chromatin compaction.

In addition to overall decrease of integration into compact chromatin, the HIV-1 integration appeared dependent on the level of compaction. Indeed, the integration efficiency of Tri-nuc containing a 53 bp DNA linker (Tri_53) was significantly greater than that of a Tri-nuc with longer (106 bp) or lower (18 bp) linker sizes. This finding suggests that, within compacted structures, specific chromatin configurations alleviate the inhibitory effect of compaction, highlighting the role of neighboring nucleosomes in modulating the formation of the functional nucleosomal STC. In contrast, PFV integration was found to accommodate compacted chromatin more efficiently than HIV-1 and was even stimulated on more compacted structures (Fig. 5B–D). Together, these findings underscore the differential impact of chromatin compaction on retroviral integration.

Neighboring nucleosomes modulate accessibility to functional HIV-1 interfaces

Taken together, these data indicate that both intasome types are affected differently by chromatin compaction in polynucleosome structures and strongly suggest that neighboring nucleosomes modulate access to functional nucleosome interfaces for efficient integration. To further investigate the differences in sensitivity to chromatin compaction, we analyzed the structure of the integration products on native acrylamide gels and quantified the distribution of integration events into either the apical adjacent nucleosomes or the central nucleosome.

The trinucleosome with 53 bp linker, which showed the highest integration efficiency for HIV-1 integration in the previous assays, was used to assess the insertion site preference in compacted chromatin substrates. As reported in Fig. 6A, analysis of the structure of the integration products revealed two major full-site integration products, indicating that the HIV-1 intasome catalyzed the integration predominantly at a single site within the central region of the trinucleosome (see the migration distance analysis in the right panel). The integration sites were found to shift from approximately the size of the 601 nucleosome assembly sequence, confirming the insertion into the central nucleosome region of the acceptor DNA. In contrast, PFV integration into trinucleosome structures led to a similar integration pattern to those observed with MNs but with a 53 bp shift corresponding to the linker length (Fig. 6B). Mapping of these insertion sites further confirmed that the PFV integration occurs across all three nucleosomes within the Tri-Nuc, resulting in multiple integration bands corresponding to each nucleosome site. To validate these data, we performed an integration assay on beads followed by EcoRI cleavage, as reported in Fig. 6C. The data showed that while most of the HIV-1 integration signal remaining in the bead fraction strongly decreased, PFV integration was less affected, leading to a 70% decrease in the signal. This result is consistent with the integration sites mapping, supporting the uniform integration of PFV across nucleosomes, in contrast to the HIV-1 integration restricted to the central nucleosome region (Fig. 6D).

Figure 6.

HIV-1 and PFV integration sites in mono- or trinucleosomes. Integration was carried out using either the isolated MN or the trinucleosome with 53 bp linker and HIV-1 intasomes assembled in the presence of LEDGF/p75 (A) or PFV (B) intasomes before deproteinization and loading into an 8% polyacrylamide gel. Positions of the different integration products on MN are reported on the left side of the gels. The position of the integration product on the trinucleosome surface was estimated from evaluation of the distance of migration using ImageJ software (right panels) and the expected integration products are indicated on the right side of the gels. Beads integration assay was performed, followed by EcorRI cleavage and quantification of the integration remaining on beads (C). Data are reported in (D) as the means of three independent experiments ± SD and t-Student test.

This preferential localization of the HIV-1 integration site at the central nucleosome suggests that one major HIV-1 functional interface is available on this specific trinucleosomal structure, highlighting the requirement for a specific spatial organization of chromatin for productive HIV-1 integration in the Tri-nuc context. Such specific chromatin organization is controlled by the adjacent nucleosome and linker DNA, suggesting their indispensable role for HIV-1 integration. This hypothesis is supported by the fact that both shorter and longer linker lengths reduce HIV-1 integration efficiency (cf. Fig. 5), indicating that an optimal chromatin geometry is required for productive integration.

Taken together, these data suggest that while the central nucleosome may be the preferred substrate for the HIV-1 intasome, additional contacts with neighboring nucleosomes may be required to stabilize STC complex for efficient integration into compacted chromatin, consistent with the nucleosome bridging properties of HIV-1 intasomes (cf. Fig. 4). In contrast, PFV integration sites remain distributed equally across all nucleosomes within the array (see Fig. 6B), indicating that PFV integration is not dependent on chromatin compaction nor neighboring nucleosomes. Alternatively, our results indicate that the functional PFV/nucleosome interfaces remain accessible despite chromatin compaction, unlike those of HIV-1.

H3K36me3 histone modification promotes HIV-1 integration specifically into trinucleosomes

Chromatin structure and dynamics are known to be tightly regulated by histone modifications. Among them, methylation of lysine 36 in histone H3 (H3K36me3) is of particular interest in the context of HIV-1 integration, as it is enriched in actively transcribed regions preferentially targeted by lentivirus (multiple references). In addition, binding of LEDGF/p75 to H3K36me3 was shown to play a direct role in the HIV-1 integration [27]. To assess the role of this modification, we compared integration efficiency in chromatin context using unmodified or H3K36me3-modified MN or trinucleosomes.

As shown in Fig. 7A and B, H3K36me3 did not affect the integration efficiency into mononucleosomes, both for PFV and HIV-1. In contrast, H3K36me3 significantly stimulated HIV-1 into Tri-nuc, while PFV integration remained unchanged (Fig. 7B), consistent with previous structural data indicating no direct interaction with histone H3 [7]. These findings suggest that the impact of histone modification on retroviral integration may depend on the chromatin architecture. Moreover, our data indicate that LEDGF/p75 interaction with H3K36me3 is promoted in the context of polynucleosomes, supporting a model in which multivalent contacts with adjacent nucleosomes provide additional anchoring points that promote efficient HIV-1 integration.

Figure 7.

Effect of H3K36me3 histone modification on HIV-1 (A) and PFV (B) integration on mononucleosomes and trinucleosomes. The efficiency of HIV-1 and PFV in vitro integration into the different nucleosomal DNA was compared by bead integration assays. In these experiments, HIV-1 intasomes were assembled in the presence of LEDGF/p75. Data are reported as percentage of integration into chromatin fibers, with integration into MN being set up at 100% ± SD (three independent experiments) (B). T-student tests have been performed to evaluate the significance of the differences between conditions.

To further explore the mechanistic basis underlying these chromatin-dependent effects, we decided to perform molecular docking simulation of nucleoprotein complexes formed during retroviral integration.

Molecular docking of retroviral intasomes in complex with trinucleosomes

To better understand the interactions between both PFV and HIV-1 incoming intasomes on compacted chromatin structure and investigate further the differences found between these functional interactions, we established a model for the intasomes binding to the Tri-nuc structure. For this purpose, HADDOCK 2.4 was employed to dock both intasomes onto the trinucleosome, specifically at the central single nucleosome and at the adjacent double-nucleosome interaction site. A computational model of the complex was first created. The model for the nucleosome was obtained from the 1KX5 structure (resolution of 1.9 Å) [16], which is available on the Protein Databank. The trinucleosome structure was sourced from the 6L4A PDB entry (12.30 Å resolution) [14], whereas the HIV-1 and PFV intasome models were obtained from the 9C9M (2.01 Å resolution) [19] and 3S3M (2.49 Å resolution) [28] PDB structures, respectively. To compare both types of intasome, we used the catalytic tetramer in both analyses. The representative structures from the best scoring cluster were chosen for this work. This structure was chosen not only because of its position as the best-scoring cluster but also because it most closely resembled the nucleosome-intasome interaction, as described by Maskell et al. [7].

Table 1 summarizes the data obtained with PFV intasome, showing that the HADDOCK docking results indicate a strong binding affinity at both integration sites, with the adjacent double interaction site having a more favorable score (−44.0 ± 13.8) than the single nucleosome interaction site (−38.5 ± 0.8).

Table 1.

Results of the docking of the PFV intasome with the single and double-nucleosome integration sites of the human trinucleosome

PFV Intasome	Central single-nucleosome integration site	Adjacent double-nucleosome integration site
HADDOCK score	−38.5 ± 0.8	−44.0 ± 13.8
RMSD from the lowest-energy structure (Å)	35.6 ± 0.0	1.6 ± 1.7
Van der Waals energy (kcal/mol)	−46.7 ± 1.4	−60.2 ± 7.8
Electrostatic energy (kcal/mol)	−148.3 ± 10.4	−168.3 ± 61.2
Desolvation energy(kcal/mol)	11.1 ± 1.2	15.2 ± 5.9
Restraints violation energy (kcal/mol)	268.0 ± 37.9	346.7 ± 34.7
Buried surface area (Å2)	1761.1 ± 16.5	2048.4 ± 259.3

The lower standard deviation observed for the single nucleosome interaction site suggests that the binding poses were more consistent across docking solutions, whereas the higher variability at the double site implies greater structural flexibility. This finding suggests that interactions at the double nucleosome site may provide additional stabilizing contacts, potentially enhancing integration efficiency. With the help of the protein–ligand interaction profiler [29], the interactions between the intasome and nucleosome were analyzed.

In the case of the central single-nucleosome integration site of PFV intrasome, hydrogen bonds are formed between the hydroxyl group of nucleosome H2A S1 and the amine group of the intasome CTD A328, between the guanidinium group of nucleosome H2A R3 and the amide group of the intasome CTD N360, between the guanidinium group of nucleosome H2A R11 and the guanidinium group of intasome CTD R334, and between the hydroxyl group of nucleosome H2B S120 and the amine group of intasome CCD L230. Notably, as reported in the literature, the intasome interacts with the tail of the H2A histone, particularly with R11, which was reported as a key interacting residue by Wilson et al. [18]. Furthermore, it also interacts with the C-terminus of the H2B histone, which has also been reported in the literature [7, 18]. Figure 8A shows the interaction between the PFV intasome and the human nucleosome at the central single-nucleosome integration site. In this figure, the interaction between the H2A histone tail (orange) and the PFV CTD (yellow) is clearly detected. Furthermore, the C-terminal helix of H2B (light gray) can also interact with the PFV CCD (cyan).

Figure 8.

Docking model of the interaction between the incoming PFV intasome and the single-nucleosome integration site (A) or the double-nucleosome insertion site (B) into trinucleosome. A focus has been made to highlight the hydrogen bonds predicted between the IN and histone components of the nucleosome.

For the double-nucleosome integration site of PFV, most of the nucleosome interfaces were similar to that observed at the single nucleosome site. The main difference is the additional contact with viral DNA (vDNA) alongside interactions between the intasome and histones. Several hydrogen bonds were identified, including interactions between the hydroxyl group of the nucleosome H2A S1 and the hydroxyl group of the intasome CTD S332, between the amino group of the nucleosome H2A K9 and the amine group of the intasome adenosine 17, and between the guanidinium group of the nucleosome H2A R11 and the hydroxyl group of the intasome CCD Y212. As shown in Fig. 8B, the PFV intasome engages the double-nucleosome site interface primarily through interactions with the H2A histone tail (orange). Despite the close proximity of adjacent nucleosomes, the interaction with H2A is thus preserved. In contrast, H2B C-terminal helix does not appear to directly contact the CCD, as observed for the single-nucleosome site, without excluding potential transient interactions due to spatial proximity of H2B. Comparison with the single-nucleosome site indicates that similar nucleosomal regions are involved in intasome binding, which may explain why PFV integration remains efficient at both sites despite the chromatin compaction.

To further confirm this model, we performed additional analyses focusing on the residues previously reported to be involved in the nucleosome interaction. Because docking structures represent static conformations, the dynamic flexibility of the system was considered by examining all nucleosome residues within 6 Å of the intasome residues for potential interactions. The results are presented in Supplementary Fig. 4A. For single nucleosome site, this analysis confirmed interactions between CCD (CCD2 in Supplementary Fig. 4B) of PFV IN and H2A/H2B histone tails, including potential hydrogen bonds involving Q137 with T116 and S120 of H2B and, T240 with S120 of H2B (Supplementary Fig. 4B). In addition, the CCD domain from adjacent IN dimer (CCD1) displayed interactions with the nucleosomal DNA, including a potential hydrogen bond between Q137 and thymine 43, and a possible salt bridge between K159 and the phosphate group of cytosine 42. At the double-nucleosome interface, PFV IN CCD (CCD2) interacted mainly with the H2A histone, including a potential hydrogen bond between T240 and H2A Q6 (Supplementary Fig. 4C), consistent with the static docking model. As observed for the single-nucleosome integration site, the other CCD (CCD1) is mostly interacting with the DNA with possible salt bridges being formed between K159 and cytosine −128 and K168 and thymine −127. Together these additional results support the proposed model underlying the central role of H2A in PFV intasome engagement.

The docking results for the HIV-1 intasome in complex with the trinucleosome are summarized in table 2. The single nucleosome complex has a much more favorable HADDOCK score (−30.8 versus −16.3), indicating that the HIV-1 intasome forms a much stronger and more stable interaction in the single-nucleosome integration site. In contrast, the lower score observed for the double-nucleosome integration site suggests that the additional nucleosome might create steric hindrance that limits HIV-1 intasome binding or make binding less efficient than a single nucleosome interaction. Consistent with biochemical data, these results support a model in which HIV-1 integration is restricted in compacted chromatin substrates and preferentially reoriented toward the central nucleosome, where a more stable interaction with the incoming intasome is allowed.

Table 2.

Results of the docking of the HIV-1 intasome with the single and double-nucleosome integration sites of the human trinucleosome

HIV-1 Intasome	Central single-nucleosome integration site	Adjacent double-nucleosome integration site
HADDOCK score	−30.8 ± 4.7	−16.3 ± 1.6
RMSD from the lowest-energy structure (Å)	19.2 ± 0.5	35.5 ± 0.2
Van der Waals energy (kcal/mol)	−50.0 ± 0.5	−50.1 ± 9.7
Electrostatic energy (kcal/mol)	−122.8 ± 18.5	−115.9 ± 47.3
Desolvation energy (kcal/mol)	18.2 ± 0.4	14.2 ± 4.8
Restraints violation energy	254.8 ± 48.8	428.0 ± 64.5
Buried surface area (Å2)	1761.1 ± 83.4	1767.5 ± 323.2

To elucidate the interactions between the HIV-1 intasome and the nucleosome, we examined the potential contacts within 6 Å of the intasome/nucleosome interfaces via protein–ligand interaction profiler [29]. As shown in Fig. 9A, for the single-nucleosome site, the HIV-1 intasome interacts mostly with the histone H4 tail (red), involving both the CTD (yellow) and the CCD (cyan) domains, in agreement with previous functional studies [8, 9]. Especially, key CTD residues previously implicated in nucleosome binding (Y227, D229, R231, W235, K236, and D253) [8, 9] were also found in our model to interact with histone tails and/or nucleosomal DNA (Fig. 9A and Supplementary Fig. 5). In particular, Y227 and D253 establish hydrogen bonds with H4 R3, consistent with previous biochemical evidence [8].

Figure 9.

Docking model of the interaction between the incoming HIV-1 intasome and the single-nucleosome integration site (A) or the double-nucleosome insertion site (B) into trinucleosome. A focus has been made to highlight the hydrogen bonds predicted between the IN and histone components of the nucleosome.

Furthermore, interactions with the nucleosomal DNA are also identified, such as a hydrogen bond between R231 and cytosine −21, and a salt bridge between K236 and the phosphate group of cytosine −24, as supported by the previously published data [8, 30, 31]. In addition to these previously identified residues, hydrogen bonds are also formed between the guanidinium group of nucleosome H4 R3 and the amide group of the intasome inner CTD N254, and between the N-terminal amine of nucleosome H4 V21 and the backbone carbonyl group of nucleosome CCD G70. Additionally, salt bridges are observed between the amino group of nucleosome H4 K8 and the carboxyl group of the intasome CCD E170, and between the amino group of nucleosome H4 K16 and the carboxyl group of the intasome CTD D232. Together, these results are confirmed by previous functional analyses and further highlight the central role of the H4 histone tail in stabilizing the HIV-1 intasome-chromatin interface via significant interactions with the intasome CTD [8, 9].

In contrast, analysis of the double-nucleosome configuration (Fig. 9B and Supplementary Fig. 5) revealed a marked reduction in interaction compared with the single-nucleosome site. The key CTD residues identified above Y227, D229, R231, W235, K236, and D253 were no longer interacting with the nucleosome. The only observed interaction with H4 was a possible hydrogen bond between the hydroxyl group of nucleosome H4 S1 and the backbone carbonyl group of the intasome CCD E69 for the target central nucleosome. At the double-nucleosome interface, HIV-1 intasome mainly formed contacts with H2A, involving N-terminal tail of H2A (A14 and K15) and HIV-1 IN CTD residues (D256 and S255, Fig. 9B). In addition to the less negative docking score, this result indicates that, in the presence of adjacent nucleosomes, the HIV-1 intasome is unable to establish its preferred H4-dependent interaction interface. This likely prevents the stabilization of the STC, thereby explaining the reduced integration efficiency observed in Tri-nuc substrates.

Importantly, previous functional analysis, including in vitro integration and cellular viral infectivity assays, confirmed the importance of residues R231, D229, and D253 in the functional interaction with the nucleosome supporting our docking model [8, 9]. To further experimentally validate our model, we performed additional functional analysis focusing on additional key CTD residues. To this purpose, D229A, D232A, and K236A IN mutants were generated, and their nucleosomal in vitro integration activities were compared with wild-type and the previously characterized R231A mutant [8, 9]. As reported in Fig. 10A, all the mutants exhibited a significant reduction in integration activity (40%–75%), with K236A showing the strongest effect. Analysis of the structure of integration products on native polyacrylamide gels (Fig. 10B) confirmed the reduced integration activity without effect on the insertion site selection. This indicates that the interaction with the histone tail primary contributes to stabilization of the STC rather than nucleosomal integration site selection, consistent with results obtained using tailless nucleosomes.

Figure 10.

In vitro functional characterization of intasomes mutated in the nucleosome-binding IN CTD domain. Integration activities were either quantified using the nucleosome integration assay on beads (A) or monitored on 8% polyacrylamide gel loading the integration products (B). In these experiments, WT and mutated intasomes were assembled in the presence of LEDGF/p75. Viral infectivity in K562 cells was assessed by GFP-positive cells quantification by flow cytometry five days post-transduction with pRRLsin-PGK-eGFP-WPRE VSV-G-pseudotyped lentiviruses (C). Total DNA synthesis (D) and integration (E) were quantified by qPCR, respectively, 16 h and 48 h post-transduction. Experiments were conducted in the presence of efavirenz 1 µM (EFV) or dolutegravir 1 µM (DTG). In vitro data are reported as the means of three to ten independent experiments ± standard deviations and t-Student tests were performed to assess the significance of the observed differences. In cellulo data are reported as the means of three independent experiments ± standard deviations and t-Student tests were performed to assess the significance of the observed differences. ^****P < .001, ***P < .005, **P< .05, *P < .5.

To further investigate the importance of the identified interface in cellular context, the K236A mutation, showing the highest effect on in vitro nucleosomal integration, was introduced into a pRRLsin-PGK-eGFP-WPRE VSV-G-pseudotyped lentiviral vector, and viral infectivity was compared with wild-type and previously described R231A mutant. As shown in Fig. 10C, the K236A mutant displayed a 70% defect in infectivity, comparable to R231A mutant. Quantification of total DNA synthesis (reverse transcription) and integration by qPCR confirmed a specific defect at the integration step (Fig. 10D and E). Taken together, all these results confirm the importance of these CTD residues in the chromatin integration process, both in vitro and in cellulo, and further support the proposed docking model.

Discussion

Retroviral integration is assumed to occur at the surface of a nucleosome in the host chromosome. Nucleosomes are embedded within different levels of chromatin compaction, prompting the question of whether they engage in long-range interactions or proximal contacts with neighboring chromatin regions. Previous studies have shown that retroviral integration models are differently affected by the degree of chromatin compaction [5, 6]. However, the molecular mechanism underlying this difference in sensitivity remains to be determined. This differential retroviral sensitivity toward chromatin structure could be due to the difference in steric hindrance related to the differences found in the quaternary structures of the intasome [4] as well as possible differences in their anchoring interfaces at the surface of the nucleosome, which may be differentially masked by chromatin compaction.

Exploring these hypotheses by comparing two different intasome types, namely PFV and HIV-1, with different chromatin sensitivity patterns both in cells and in vitro, revealed that both intasome types bind different regions of the nucleosome, leading to different insertion sites (IS) within the corresponding STC. Indeed, while PFV integration occurs mostly on the previously reported IS at exposed major groove of the nucleosomal superhelix location (SHL) 3.5 positions, ~36 bp away from the dyad [7], analysis of the HIV-1 insertion sites revealed that this intasome preferentially binds the 51–56 bp of the nucleosomal DNA. Masking this insertion site in compacted chromatin would lead to a decrease in integration efficiency, as previously observed. Testing this hypothesis by positioning the HIV-1 and PFV nucleosomal insertion sites identified on a tri-nucleosome compacted structure revealed that while the PFV nucleosomal insertion site was accessible on all three nucleosomes, only one HIV-1 IS appeared exposed and suitable for integration and was located on the central nucleosome. In contrast, the corresponding HIV-1 ISs located on adjacent nucleosomes are deeply masked at nucleosome interfaces, preventing integration onto these regions. Molecular docking simulations of intasome binding to these identified sites in the three nucleosomes of a compacted trinucleosome led to interaction models confirming that PFV incoming intasome can efficiently bind all the nucleosomal insertion interfaces. In contrast, HIV-1 binding to adjacent nucleosomes was clearly unfavored compared with that of the central nucleosome, due notably to the masking of both DNA and histone interactions required for efficient integration. More accurate docking analysis of the interaction between HIV-1 intasome and the nucleosome identified key CTD IN residues involved in the association with both nucleosomal DNA and histone H4 tail. Functional analyses of intasomes carrying mutations in the identified positions confirmed the importance of these positions in the chromatin integration process. Taken together, these data confirmed previous studies showing that HIV-1 IN CTD carries a nucleosome interaction domain with an SH3 fold able to interact with histone tail [8, 9]. These results further support a chromatin binding function in retroviral INs.

In vitro integration assays onto trinucleosomes and chromatin fiber structures further support this model, confirming the difference in sensitivity toward chromatin compaction between PFV and HIV-1 and showing that PFV integration accommodates the chromatin compaction more easily than HIV-1 integration does, as suggested by our model. Furthermore, analysis of the trinucleosomal integration products confirmed that the PFV intasome could functionally interact with all three nucleosomes, as expected from our model, and explain its ability to accommodate compaction. In contrast, HIV-1 integration appeared to occur mostly in the central nucleosome region, as expected from the IS masked onto the adjacent nucleosomes, which explains the negative effect of chromatin compaction on this retroviral integration system. These findings provide insights into the molecular mechanisms of retroviral integration. Compared with the HIV-1 intasome, the PFV intasome has a stronger and more stable binding affinity at both central and adjacent nucleosome integration sites. The higher HADDOCK scores and interaction profiles suggest that PFV intasomes can efficiently integrate at both integration sites. In contrast, the HIV-1 intasome has reduced affinity at the adjacent nucleosomal sites, likely due to steric hindrance surrounding the insertion site and weaker interactions with key histone residues. This may explain why the HIV-1 intasome preferentially integrates at the central nucleosome site and is unable to efficiently target each nucleosome within a compacted trinucleosome.

Our analysis also revealed that the ability of reducing the nucleosome linker size to inhibit integration was not linear. Indeed, we unexpectedly found that certain lengths of nucleosome linkers favored integration within the set of polynucleosomes used in our assay. However, the integration efficiency detected in these particular polynucleosomes never reached the level found in mononucleosomes, which may recapitulate the most relaxed chromatin models. Bringing closer to or moving away from the neighboring nucleosomes by modifying the nucleosome linker was indeed found to affect the efficiency of integration. These original results may suggest additional contacts between STC components and adjacent nucleosomes, while integration would be performed mostly onto the central nucleosome. A nucleosome-bridging AlphaLISA assay (NBA, set up in this work) confirmed that, in contrast to PFV, HIV-1 efficiently bridges at least two different nucleosomes. Even if direct contact between nucleosomes cannot be fully ruled out, such nucleosome–nucleosome interactions are not supported by our data obtained with acidic patch mutants. Additionally, this bridging property was probably not due to IN/IN aggregation or intasome/intasome contacts previously shown to induce their oligomerization since the addition of the HIV-1 CTD did not affect the interaction signals. In contrast, omitting the LEDGF/p75 protein largely decreased the association signal as well as the deletion of H3 or H4 tails, in contrast to H2A and H2B. The use of LEDGF/p75 alone confirmed the ability of this factor to bridge several nucleosomes. This property may, thus, participate in the global nucleosome-bridging ability of the HIV-1 intasome. These results may be reminiscent of the previously reported multivalent recognition of nucleosomes by LEDGF/p75 [32], which involves multiple interactions between the PWWP domain of the factor and histone tails and nucleosomal DNA. The strong decrease in the nucleosome-bridging property of intasome formed in the presence of LEDGF/p75 deleted for the PWWP domain confirmed the importance of this region for the nucleosome coordination during integration. In addition to the PWWP interaction with the histone tail, an additional region of interaction was found in the region of the histone octamer acidic patch, and the DNA interface remained dynamic in the context of dinucleosomes. Additionally, specific sites within the intrinsically disordered central region of LEDGF/p75 selectively interact with nucleosomal or extranucleosomal DNA. In concert with our data, these results strongly suggest that the intasome bridging property mostly relies on the LEDGF/p75 cofactor and histone tail interactions, which may be involved in the contact of the HIV-1 intasome with adjacent nucleosomes. Our biochemical data also suggest that IN-histone and intasome–DNA interactions constitute additional secondary determinants of the nucleosome bridging property that may act in concert with LEDGF/p75 for the selection of suitable local chromatin structures optimal for integration, as proposed in the model reported in Fig.11. All together, our data thus revealed an unedited multifactorial nucleosome-bridging function for HIV-1 intasome.

Figure 11.

Model for local chromatin structure scanning by HIV-1 incoming intasome. After a global targeting toward specific chromatin region, the efficiency of retroviral nucleosome binding would rely on the local availability of the related insertion site governed by the level of chromatin compaction. For PFV the functional interface of association with the nucleosome would be accessible in most of the chromatin structures. Compaction of the chromatin may overexpose the optimal intasome-nucleosome interface. This would result in a high capability of PFV intasome to accommodate chromatin structures with a preference for compacted nucleosomes. HIV-1 integration would require increased accessibility of the nucleosome interface in open and specifically structured compacted chromatin structure, allowing the correct association between the targeted and adjacent nucleosomes (as shown in the focused panel).

In this model, the availability of different retroviral ISs modulates the ability of incoming intasomes to efficiently bind nucleosomes in specific chromatin regions. After global intasome targeting toward specific chromatin regions, multivalent interactions between IN, LEDGF/p75, and both targeted and adjacent nucleosomes allow the selection of suitable local structures with nucleosome density and compaction levels suitable for efficient functional docking of the target capture complex and integration catalysis. In the case of PFV, the increased availability of the nucleosomal IS within chromatin would allow the incoming intasome to accommodate broader structures and compaction levels. A higher level of compaction may lead to more exposed PFV ISs and fully mask HIV-1 ISs. This mechanism would constitute an additional parameter in the global intasome targeting toward host chromatin during the final local association process.

In this model, and on the basis of our data, the close chromatin structures as found in heterochromatin regions may prevent HIV-1 insertion, in contrast to PFV, which aligns well with its cellular preference for euchromatin insertions.

Our data additionally suggest that the incoming HIV-1 intasome may form a functional STC on multiple nucleosome structures involving at least two structural nucleosomes neighboring the functional one. In this process, LEDGF/p75 may participate in this functional binding in addition to providing a way to gain access to the dynamic transcribed region of the chromatin where the insertion site is available for integration. This multivalent interaction may be important for stabilizing the STC complex, which may be of particular importance in the highly transcribed regions targeted by HIV-1. Previous remodeling activities, such as SWI/SNF or FACT histone chaperones, which have been shown to improve HIV-1 integration [5], [24], may also help in generating functionally suitable local chromatin structures suitable for integration. SWI/SNF complexes are ATP-dependent chromatin remodelers that disrupt nucleosome positioning, expose linker DNA, and create a more permissive environment for the PIC. Our previous work showed that the presence of SWI/SNF enhances integration efficiency in compacted chromatin substrates, likely by increasing the exposure of linker DNA and destabilizing nucleosome-nucleosome interactions [5]. SWI/SNF could thus partially overcome the inhibitory effect of compact chromatin (e.g. nucleosome fibers) by generating transient gaps or “breathing” regions where the PIC can access its target DNA. FACT (Facilitates Chromatin Transcription) is known to destabilize nucleosomes by removing H2A–H2B dimers, which could directly facilitate PIC access to nucleosomal DNA. In our previous experiments, the addition of FACT significantly improved integration in compacted chromatin structures [24], supporting the hypothesis that FACT-mediated nucleosome destabilization of the surrounding chromatin may be a key step in allowing the PIC to engage with target nucleosomes in more condensed chromatin. FACT may cooperate with the PIC to locally “open” nucleosome arrays, improving the chances of encountering suitable chromatin structures for efficient integration. Both machineries, associated with Pol II and in proximity to the incoming intasome, may cooperate with the integration complex by dynamically remodeling the chromatin landscape, transiently exposing integration sites that would otherwise be inaccessible. We have added further discussion in the manuscript to clarify how FACT and SWI/SNF may facilitate integration in chromatin.

Furthermore, the recently reported histone chaperone activity of LEDDGF/p75 (poster communication in CSH retroviruses meeting 2025 from Joshi et al.) may be directly involved in the local remodeling of the nucleosomal insertion site, directly providing the correct local compaction level for integration during the integration process. This possible chromatin remodeling activity associated with the HIV-1 intasome remains to be fully investigated.

While our in vitro data suggest that specific levels of compaction may be required for efficient HIV-1 integration, direct correlation between the nucleosome repeat length (NRL) used in our biochemical substrates and cellular state of the chromatin remains difficult to carry out. However, we can notice that the theoretical NRL of the 196 bp (146 bp nucleosomal DNA and 50 bp of linker) preferred substrate aligns with the NRL range observed in transcriptionally active regions of mammalian cells (∼180–200 bp) [33]. Most nucleosomes in metazoans are poorly positioned, and it is unknown to what extent bulk genomic nucleosome repeat length reflects the regularity and spacing of nucleosome arrays at individual loci. Modulation of array regularity and NRL depending on functional chromatin states independently of nucleosome positioning has been reported [34]. Literature indicates that nucleosomes are positioned and form a phased array that surrounds the transcription start sites [35–38] of active genes and DNase I hypersensitive sites [39]. More recent work shows that nucleosome arrays downstream of silent promoters are considerably more regular that those downstream of highly expressed ones, despite more extensive nucleosome phasing of the latter [34]. Heterogeneity during cell activation can thus be expected, making it difficult to correlate perfectly the structure of chromatin at the time of cellular integration and from our chromatin array that may constitute a limit of our study. However, the trinucleosome structure revealed by our docking and biochemical analyses may reflect some specific cellular chromatin configurations associated with HIV-1 integration. The trinucleosome model captures the physiological organization of chromatin fibers, in which nucleosomes are arranged into higher-order arrays [3, 40]. Integration into the central nucleosome of such an array could place the provirus within a chromatin context favorable to both strand transfer and subsequent transcriptional activation, while reducing exposure to the repressive environment of dense heterochromatin or the steric constraints imposed by highly compacted nucleosome packing. Indeed, genome-wide studies have established a strong bias of HIV-1 integration for transcriptionally active euchromatin, where nucleosomes are more dynamically positioned and enriched in histone marks linked to elongation, such as H3K36me3 [41, 42]. The central nucleosome in a trinucleosome is therefore more likely to belong to a transcriptionally permissive domain, being less constrained by higher-order folding than nucleosomes at the edges of compacted regions. This structural interpretation is consistent with the function of LEDGF/p75 in tethering the pre-integration complex to H3K36me3-marked nucleosomes within active genes, thereby promoting integration into transcriptionally competent chromatin [43, 44].

The exact role of LEDGF/p75 in this docking process remains to be fully elucidated, and how this factor may influence the structure of the chromatin locally or neighboring it in an integration-coupled manner remains to be answered. Indeed, one other limitation of our docking study is the lack of LEDGF/p75 in the structure used, providing the possibility to detect these interactions. Additional protomers of IN are also missing in the docking study, indicating the detection of potential additional contact with adjacent nucleosomes in these analyses, as predicted from our biochemical experiments. One remaining question in the field is the role of LEDGF/p75 recognition of H3K36me3 mark in the integration process. Indeed, while the association of the H3K36me3 histone mark with cellular integration sites has been broadly reported, the mechanism underlying the integration complexes targeting of these regions and the integration catalysis remains poorly documented. Here, we provide the evidence of an integration modulation by H3K36me3 in a polynucleosome substrate, in contrast to mononucleosomes, further confirming the possible association of the intasome with surrounding chromatin and the possible role of these H3K36me3-enriched regions around the insertion site.

Our model also implies a new notion of the retroviral integration process involving both catalytic and structural nucleosomes. Additional structural approaches, such as cryoelectron microscopy or electron tomography, are needed to gain insight into the atomic interaction within this newly identified integration complex and define all the interactions between the intasome and both structural and catalytic nucleosome components. Such structural insights remain largely limited by the heterogeneity of the HIV-1 STC-Nuc preparations, probably due to the multivalent intasome/nucleosome interactions demonstrated in our work. The data reported here will thus help in defining the best polynucleosomal substrate for future attempts to resolve this structure.

Although additional structural data are needed to fully elucidate the molecular determinants of retroviral intasome interactions with chromatin, our findings reveal that different retroviral species have evolved distinct strategies to engage nucleosomes within specific chromatin regions. These preferred nucleosomal targets vary between viruses, reflecting divergent mechanisms to access suitable regions of host chromatin and optimize host genome invasion.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work has been supported by the French Agence National de Recherche contre le SIDA, ANRS, research agency (ECTZ115893, ECTZ227474) and Foundation pour la Recherche Médicale, FRM, (EQU202303016283).

Data availability

Structural models have been deposited in Zenodo at https://doi.org/10.5281/zenodo.15675341.