Impairment of Human Immunodeficiency Virus Type-1 Integrase SUMOylation correlates with an early replication defect

HIV-1 integrase (IN) orchestrates the integration of the reverse transcribed viral cDNA into the host cell genome and participates also in other steps of HIV-1 replication. Cellular and viral factors assist IN in performing its multiple functions, and post-translational modifications contribute to modulate its activities. Here, we show that HIV-1 IN is modified by SUMO proteins and that phylogenetically conserved SUMOylation consensus motifs represent major SUMO acceptor sites. Viruses harboring SUMOylation site IN mutants displayed a replication defect that was mapped during the early stages of infection, before integration but after reverse transcription. Because SUMOylation-defective IN mutants retained WT catalytic activity, we hypothesize that SUMOylation might regulate the affinity of IN for co-factors, contributing to efficient HIV-1 replication.


HIV-1 integrase (IN) 1 orchestrates the integration of the reverse transcribed viral cDNA into the host cell genome and participates also in other steps of HIV-1 replication. Cellular and viral factors assist IN in performing its multiple functions, and post-translational modifications contribute to modulate its activities.
Here, we show that HIV-1 IN is modified by SUMO proteins and that phylogenetically conserved SUMOylation consensus motifs represent major SUMO-acceptor sites. Viruses harboring SUMOylation-site IN mutants displayed a replication defect that was mapped during the early stages of infection, before integration but after reverse transcription. Since SUMOylation-defective IN mutants retained WT catalytic activity, we hypothesize that SUMOylation might regulate the affinity of IN for co-factors, contributing to efficient HIV-1 replication.
HIV-1 IN is a 288-amino acid protein consisting of three functionally independent domains. The N-terminal domain (NTD) harbors a highly conserved HHCC zinc binding motif that contributes to IN multimerization and enzymatic activities. The central core domain (CCD) contains the catalytic DDE motif that is conserved in all retroviral and retrotransposon INs, and in certain bacterial transposases. The Cterminal domain (CTD) is the least conserved among retroviral IN and binds DNA non specifically (for review see (1)(2)). The best characterized activity of HIV-1 IN is the catalysis of integration, which is crucial for HIV-1 replication (3). This reaction can be reproduced in vitro in the presence of recombinant IN alone and synthetic DNA species mimicking the viral LTR ends and an acceptor substrate (4)(5). However, other components of the preintegration complex (PIC) contribute to the specificity and efficiency of integration in vivo (6)(7)(8). Independently of its enzymatic activity, HIV-1 IN plays additional roles during the viral life cycle that are still ill defined. Indeed, many catalycally-active IN mutants have pleiotropic effects impairing reverse transcription and/or uncoating (9)(10)(11)(12)(13)(14)(15)(16)(17), PIC nuclear import (18)(19) and virion protein composition and/or morphology (16,(20)(21). Post-translational modifications contribute to the regulation of IN activities. HIV-1 IN interacts with and is acetylated by both HAT p300 and GCN5 on C-terminal lysine (K) residues (22)(23). Acetylation increases IN affinity for the viral cDNA, enhances its strand transfer activity in vitro and might regulate the interaction between IN and cellular factors (24). However, the role of this modification during HIV-1 replication is still controversial (25). HIV-1 IN is also ubiquitinated and subsequently degraded by the proteasome (26)(27)(28). In the viral context, IN degradation seems to occur after integration and to be required for correct gap repair (29)(30) and viral genes expression (26). Recently, phosphorylation of HIV-1 IN by cellular JNK has also been proposed to modulate its stability and to be necessary for efficient integration (31). SUMOylation consists of the covalent attachment of small ubiquitin like modifier (SUMO) peptides to a K residue within the consensus motif (ΨKxE/D, where Ψ is a large hydrophobic residue, x any amino acid and E/D glutamic or aspartic acid) of a substrate protein. SUMO-conjugation is mediated by SUMOspecific E1-activating, E2-conjugating and E3ligating enzymes and is reversed by SUMOspecific proteases (reviewed in (32)). In mammals, three major forms of SUMO proteins are expressed. SUMO-1 has approximately 45% amino acid sequence homology to SUMO-2 and SUMO-3, which are 96% identical to each other. SUMO modification is implicated in numerous cellular processes including signal transduction, protein stability and localization, transcriptional regulation, chromatin structure and genome stability (32). It is also well established that viruses interfere with and/or hijack the cellular SUMOylation machinery to replicate (for review, see (33)(34)). Interactions between murine leukemia virus capsid (CA) protein and components of the SUMOylation pathway are required for proper execution of the early steps of replication after reverse transcription, but before integration (35). SUMOylation events have also been implicated in the early phase of HIV-1 infection. Indeed SUMO-2 and RanBP2, a SUMO E3 ligase, were identified in genomewide screens for cell factors that promote HIV-1 reverse transcription and PIC nuclear import, respectively (36)(37). Interaction of HIV-1 or Mason-Pfitzer monkey virus Gag proteins with SUMO-1 and the E2-conjugating enzyme Ubc9 during the late phases of replication have also been reported and are likely involved in the production of fully infectious virions (38)(39)(40)(41). Here, we show that HIV-1 IN is SUMOylated and that three K residues, which are found within conserved consensus motifs, represent the major SUMO-acceptor sites. In the viral context, mutation of SUMO-acceptor residues in IN led to reduced infectivity and slower replication kinetics. Biogenesis, release and reverse transcription steps of mutant HIV-1 particles were not affected. However, cells infected with viruses harboring SUMOylation-defective IN mutants showed a significant decrease in integration events compared to HIV-1 WTinfected cells. Since SUMOylation-site IN mutants retained WT catalytic activity, we inferred that modification by SUMO might participate in the modulation of HIV-1 IN interaction network by regulating its affinity for co-factors, which are required for the efficient execution of early events of HIV-1 replication.
Plasmid construction and Mutagenesis -psPAX2, pWPI and pMD2.G (a gift from D. Trono), pNL4-3EnvFsGFP (a gift from D. Gabuzda), which contains a complete HIV-1 provirus with an env-inactivating mutation and enhanced GFP in the place of Nef (43) and IN WT -Flag, which encodes a codon-optimized IN gene harboring an ATG initiation codon and a C-terminal Flag tag (44) have been already described. The cDNAs encoding WT IN, SUMO-1, -2 and -3, were amplified by PCR from IN-Flag and YFP-SUMO-1, -2 and -3 (45) and were subcloned in frame with an N-terminal 6xHis tag into the pcDNA3.1(-) vector (Invitrogen), yielding His-IN WT , His-SUMO-1, -2 and -3. The cDNA encoding the C-terminal region of LEDGF/p75 (aa 325-530) was amplified by PCR from WT and D366N mutant HA-LEDGF/p75 (46 Virus stock production and infectivity assay -Single-round viruses were produced by cotransfection of 293T cells using a standard calcium phosphate precipitation technique with a plasmid encoding WT or mutant HIV-1packaging DNA (psPAX2) and the genomic transfer vector encoding GFP (pWPI) or the pNL4-3EnvFsGFP vector and an expression vector for the glycoprotein G of vesicular stomatitis virus (VSVg) (pMD2.G). Replicationcompetent viruses were produced by transfecting the pNL4-3 plasmid that encodes a complete HIV-1 infectious provirus. Supernatants were collected 40 h post-transfection, clarified by low speed centrifugation, filtered through 0.45 mpore size filters and treated with 10U/ml Turbo DNAse (Ambion) (1h, RT). Viral particles were concentrated by ultracentrifugation (24 000 rpm, 1h 30 min, 4 °C) using a SW32 rotor (Beckman) on a 20% sucrose cushion. All viral stocks were normalized for the p24 CA antigen content, as determined by ELISA (Zeptometrix) and used to infect target cells (6x10 4 293T or 1x10 6 CEM-GFP cells). After 48h, the percentage of GFPexpressing cells was measured by flow cytometry on a FACSCalibur flow cytometer with CellQuest software (BD Biosciences).
Western blot analysis of viral proteins -Viral proteins associated with virions or with infected cells were analyzed by Western blot with anti-CA and anti-IN antibodies. For quantification of virion-associated CA and IN proteins, secondary antibodies coupled with IRDye near-infrared dyes (IRDye800CW and IRDye680LT, Science Tec) were used. Proteins were visualized on an Odyssey infrared imager and quantified with Odyssey software (LI-COR Biosciences).
Real-time PCR analysis -Total genomic DNA was extracted using the Blood and body fluid kit (QIAGEN) from 293T cells (5x10 5 ) infected with single-round viruses. Full-length reverse transcripts, integrated HIV-1 DNA and 2-LTR circles were quantified using a previously described protocol (49). Parallel infections with heat-inactivated HIV-1 WT viruses were performed to control for residual levels of plasmid DNA that may have resisted DNAse treatment. Viral RNA was extracted with the RNeasy Mini kit (QIAGEN) and amplified with the HIV-1 real time RT PCR Kit (BioEvolution). Real-time PCR and RT-PCR were performed on a Lightcycler 1.0 (Roche Diagnostics).
Vpr-integrase complementation -Viral stocks generated by co-transfecting 293T cells with pNLX.Luc(R-Env-), pRL2P-Vpr-IN WT or mutant, and pNLXE7 were used to infect Jurkat cells (2 × 10 6 /ml, 5 × 10 5 RT cpm), as described (50). Cells were harvested 48 h after infection and lysed in passive lysis buffer (Promega). Frozen and thawed lysates were clarified by centrifugation (18,730×g, 15 min, 4°C), and supernatants were analyzed for luciferase activity in duplicate using the Promega luciferase assay system, an EG&G Berthold Microplate LB 96V luminometer, and a Microlite 1 flat-bottom microtiter plate (Thermo Labsystems). Luciferase activity was normalized to the protein concentration as determined by the Bio-Rad protein assay kit (Bio-Rad) and corrected for background levels from lysates of cells infected with Env-negative controls.
Statistical analyses -Pair-wise comparison between groups was performed using the Student's t test. P<0.05 was set as a threshold for statistical significance.
RESULTS    fig. 3C and 3D). In agreement with previous reports (26)(27)(28), treatment with MG132, an inhibitor of the ubiquitin-proteasome system, prevented the degradation of both WT and mutant IN proteins (fig. 3C) . 4A). Although they can fully recapitulate the early phase of infection, lentiviral vectors lack both the ability to perform the late stages of HIV-1 replication and viral accessory proteins present in their parental counterpart, which are mostly dispensable for replication in vitro but essential for viral pathogenesis in vivo. Thus, we analyzed the effect of the disruption of the SUMOylation sites within IN also on the infectivity of authentic HIV-1 viruses. To this aim, K-to-R substitutions were introduced in a plasmid encoding fulllength HIV-1 proviral DNA with a frame shift in Env and expressing the GFP reporter gene in the place of Nef (NL4-3EnvFsGFP). Mutant virions harboring an IN protein bearing E-to-Q changes at identified SUMOylation consensus motifs were also generated to specifically disrupt SUMOylation without affecting other posttranslational modification that target K residues. Single-round viruses were produced and used in infection experiments as described before. As shown in figure 4B, HIV-1 harboring IN 3KR or IN 3EQ were about 57% and 33% as infectious as HIV-1 WT , respectively. The contribution of non structural viral proteins, absent from the lentivector background, might account for the differential viral infectivity of viruses bearing SUMOylation-deficient IN proteins under these settings. Finally, to study the outcome of the impairment of IN SUMOylation on HIV-1 replication following entry via receptormediated fusion at the plasma membrane, K-to-R changes were introduced into the NL4-3 molecular clone of HIV-1. Human T-lymphoid CEM-GFP cells were infected with equal p24 CA amounts of HIV-1 WT or HIV-1 3KR and the percentage of GFP-positive cells was monitored by flow cytometry over time. Under these experimental conditions, the infectivity of HIV-1 3KR was 32% and 41% that of HIV-1 WT at 48 and 96 h post-infection, respectively ( fig. 4C). Additionally, HIV-1 3KR replicated with a delay of approximately 3 days compared to HIV-1 WT , reaching peak growth at 7 days post-infection. Altogether these findings indicate that SUMOmodified K46, K136 and K244 play a role during HIV-1 replication.

SUMOylation-site mutant HIV-1 is impaired at an early step(s) of replication preceding integration -We observed that simultaneous disruption of IN SUMOylation consensus motifs correlated with an impairment of HIV-1 infectivity.
To establish what stage(s) of HIV-1 replication cycle was affected, we first asked whether the defects observed could be accounted by improper viral particle assembly, composition and/or release. Similar amounts of WT and mutant progeny virions were obtained following transfection of 293T cells with NL4-3EnvFsGFP plasmid encoding IN WT , IN 3KR or IN 3EQ , in independent production experiments, as determined by p24 CA ELISA ( Table 1). The expression levels and patterns of proteasemediated cleavage of the Gag precursor were analyzed and found to be similar for WT and mutant HIV-1-producing cells (data not shown). Analysis of HIV-1 WT , HIV-1 3KR and HIV-1 3EQ protein content by Western blot using nearinfrared dye-conjugated antibodies followed by quantification of the corresponding emission signals on an Odyssey infrared imaging system, showed that WT and mutant viral particles displayed comparable amounts of IN protein relative to CA ( fig. 5A and Table 1). Additionally, two bands reactive to an anti-IN antibody, with the apparent molecular mass of IN monomer and dimer molecules, were detected in all samples by analysis performed under non-reducing conditions (without β-Mercapto-ethanol), as previously described (54 . 5C), the levels of integrated proviruses at 24 and 48 hpi were significantly diminished in cells infected with HIV-1 3KR or HIV-1 3EQ as compared with HIV-1 WT -infected cells ( fig. 5D). We also monitored the formation of 2-LTR circles, which are generally used as a marker of PIC nuclear import (reviewed in (55)), and found that the number of copies of 2-LTR circles formed upon infection with HIV-1 harboring WT or mutant IN proteins was similar (data not shown). We next assessed whether mutation of SUMOylation sites of HIV-1 IN affected its enzymatic activity. For that purpose, we tested the function of SUMOylation site-mutant IN proteins in the viral context using a Vpr-IN complementation assay (56). WT or mutant IN proteins fused to Vpr were expressed together with an HIV-1 proviral vector encoding an active site IN mutant (HIV-1 D64N/D116N ). Vprmediated encapsidation of IN proteins, in which SUMOylated K residues were changed to R either individually or simultaneously, allowed recovery of viral infectivity comparable to that of Vpr-IN WT (figure 5E and data not shown), suggesting that mutation of SUMO-acceptor sites did not significantly affect IN catalytic activity under these infection conditions. Finally, we asked whether impairment of provirus formation upon infection with HIV-1 3KR or HIV-1 3EQ could be accounted for by loss of interaction with LEDGF/p75, an extensively studied IN-interacting protein and an essential chromatin-docking factor for HIV-1 PIC (29,(57)(58). To this aim, 293T cells were cotransfected with an expression vector for Flagtagged IN WT  binding, and correlates to an early replication defect occurring before proviral integration, but after reverse transcription. DISCUSSION HIV-1 IN is the viral enzyme that orchestrates the integration of the viral cDNA into cellular genome, a key event of retroviral replication and the target of novel anti-HIV therapeutic agents. Numerous studies have contributed to quite an extensive understanding of the molecular basis of IN catalytic functions. On the contrary, while it is well established that integrity of IN structure and/or its interaction network are critical for optimal execution of various steps of HIV-1 life cycle other than integration (59), the mechanisms by which IN exerts these additional functions are presently still elusive. To address this issue, IN protein-protein interactions have been widely explored (reviewed in (60)); however, information is scarce on IN posttranslational modifications that represent a common, rapid and generally reversible mechanism for fine tuning of protein activities.
We repeatedly observed that numerous bands were detected in an anti-IN immunoblot following affinity purification of HIV-1 IN under denaturing conditions from either cellular or viral extracts, suggesting that it undergoes a high degree of post-translational modifications (data not shown). It has already been shown that IN is acetylated, ubiquitylated and phosphorylated (22-23, 25-29, 31). Since we had mapped putative sites of SUMO conjugation to K residues at position 46, 136 and 244, located within phylogenetically conserved canonical SUMOylation consensus motifs (32), we asked whether HIV-1 IN is also SUMOylated. We identified these amino acids by comparative analysis of IN sequences, which showed that K46 and K244 are found within SUMOylation consensus motifs common to HIV-1, HIV-2, SIVmac and SIVcpz ( fig. 1A and SI fig. 1). K244 is also conserved among certain nonprimate lentiviruses (61). Notably, analysis of naturally occurring variations within IN sequences of HIV-1 isolates, showed that K46 and K244 are not polymorphic (62)(63). Conversely, K136 is conserved in about one third of HIV-1 strains and displays both interand intra-subtype substitutions to residues that cannot be modified by SUMO with a frequency rate > 0.5% (62)(63). This extent of amino acid conservation suggests a role of these residues in preserved IN functions. Analysis performed in vitro showed that recombinant purified HIV-1 IN is covalently coupled by SUMO proteins. Under these experimental conditions, similar efficiency of modification, but slightly different patterns of conjugation by each paralogue were observed, pointing to the presence of both SUMO-1-and SUMO-2/3-specific attachment sites within IN. We further confirmed modification of IN by SUMO proteins in a cellular context, both in the presence of exogenous or endogenous expression levels of SUMO family members. In both cases SUMO-2/3-conjugated IN species were more abundant than SUMO-1-conjugated forms, likely reflecting the higher availability of SUMO-2/3 compared to SUMO-1 (64). Finally, the formation of numerous highmolecular-weight species is consistent with the addition of multiple SUMO moieties to IN. The attachment of single SUMO molecules to several K residues is supported by the fact that many SUMOylated IN species were detected when an in vitro reaction was conducted in the presence of SUMO-1 alone, which cannot form polySUMO chain. However, the conjugation of polymeric SUMO chains to one K residue of IN under different experimental settings cannot be ruled out. In light of emerging evidence suggesting that SUMO proteins display both common and specific target protein preferences and play both redundant and non-redundant cellular functions, these results might underlie functional differences of SUMO-1-and SUMO-2/3-conjugation to IN. Simultaneous substitution of key residues within the three canonical SUMOylation consensus motifs (either K or E) was required to significantly decrease the occurrence of SUMOylated, but not ubiquitinated or acetylated, IN species. However, IN SUMOylation was not abolished indicating that K46, K136 and K244 represent the major, but not unique, SUMO-acceptor sites. In addition to the ΨKxE/D consensus, the K residue within the reverse E/DxKΨ signature can also be modified by SUMO (65). Concordantly, K residues at position 71 and 258, which are highly conserved among HIV-1 isolates, may represent additional sites of SUMOylation ( fig. S1). Disruption of preferred modified sites might also result in the transfer or the enhancement of SUMO conjugation to other K residues, as reported for other proteins (66)(67) fig. S3A and S3B).
In the viral context, mutation of SUMO acceptor sites correlated with decreased infectivity and slower replication kinetics compared to HIV-1 WT , both in epithelial and T cell lines. Viruses harboring IN 3KR displayed viral particle biogenesis and release similar to the WT counterpart, indicating that mutation of major SUMO-acceptor sites did not affect late stages of HIV-1 replication cycle. We next analyzed viral cDNA synthesis and provirus formation upon infection and found that SUMOylation-site mutant viruses displayed a defect at the integration step, while reverse transcription and PIC nuclear import were mostly unaffected. Altogether, these results show that integrity of identified consensus sites for SUMOylation is required both for optimal SUMOylation levels of IN and efficient proviral integration. Either concomitant or consecutive modification of K residues at position 46, 136 and 244 by SUMO might be required for a favorable outcome of HIV-1 replication. However, residual modification of non canonical SUMOylation sites within IN might compensate for the lack of conjugation to major SUMO-attachment sites and, thus, account for the moderate reduction of viral infectivity. Interestingly, we obtained analogous results when studying HIV-1 3EQ , which bears an IN mutant in which E-to-Q changes were introduced at identified SUMOylation consensus motifs to specifically impair SUMO-conjugation but not other post-translational modifications that occur on K residues. The fact that viruses harboring either IN 3KR or IN 3EQ display an analogous phenotype supports a direct requirement for IN SUMOylation in the optimal execution of an event(s) following reverse transcription and nuclear entry, but before integration. Thus, our results are agreement with data from recent genome-wide studies indicating that components of the SUMOylation pathway promote HIV-1 replication (36)(37) S4). The fact that each HIV-1 virion encapsidates an estimated 250 molecules of IN (70) and that only a small fraction of it might be SUMOylated at steady-state, could at least in part explain these observations. Moreover, SUMOylation is reversible and efficient viral replication might rely on a dynamic process of conjugation-deconjugation of SUMO moieties to IN. We note also that HIV-1 p6 has been shown to be SUMOylated, but the incorporation of the SUMO-conjugated protein into virions has not been detected (38), despite p6 is about 20 times more abundant than IN. Expression of components of the SUMOylation pathway, in particular SUMO-2 and the SUMO E3 ligase RanBP2, in target cells has recently been implicated in the completion of early events of HIV-1 infection (36)(37) (24). Further studies will be required to clarify the molecular mechanisms by which the cellular SUMOylation pathway participates in the control of HIV-1 IN functions during the early steps of viral replication.