Background
HIV integrase (IN) is a key viral enzyme that catalyzes the integration of viral DNA into the host genome in all retroviruses. HIV-1 IN also functions in other key steps during the viral life cycle, including reverse transcription, nuclear import of the preintegration complex (PIC) and postintegration steps, such as viral protein expression, transcription, packaging and processing [
1‐
4]. Similar to all retroviral INs, HIV-1 IN contains three canonical domains: an N-terminal HH-CC zinc-binding domain, a catalytic core domain (CCD) and a C-terminal DNA-binding domain, each with different individual functions. Full-length IN is a multimeric enzyme that functions as a tetramer [
5]. IN undergoes multiple posttranslational modifications (PTMs) (e.g., ubiquitination, SUMOylation, acetylation and phosphorylation), which have been shown to play versatile roles in the functions of IN and HIV-1 viral replication [
6‐
9]. The SUMOylation of IN has been published earlier [
7]. However, much is still unknown about the physiological mechanisms of this modification.
Small ubiquitin-like modifier (SUMO) proteins are ~ 10 kD in size, and there are four subtypes (SUMO 1–4) in mammals, which are conserved among all eukaryotic cells [
10]. SUMO 1, 2 and 3 are ubiquitous in cells, whereas SUMO4 is only expressed in certain tissues and organs [
10,
11]. While SUMO2 and SUMO3 are 96% identical to each other, SUMO1 only shares 45% amino acid identity with SUMO2/3. SUMOylation is accomplished by a series of enzyme-catalyzed reactions [
12]. Three enzymes are required for all SUMO modification pathways, including SUMO activating enzyme E1 (a heterodimer of Aos1 and Uba2), the unique E2 conjugating enzyme Ubc9 and a number of different E3 ligases, such as PIAS and RanBP2 [
12]. In a previous study, Ubc9 was shown to interact with HIV-1 IN, and SUMO2/3 and Ubc9 negatively regulate the HIV-1 integration step [
13]. Similar to ubiquitination, the substrate proteins can be poly-SUMOylated or mono-SUMOylated at single or multiple Lys targets. The outcomes of SUMO modification vary greatly from protein stability, cytosolic-nuclear translocation, and antagonizing other posttranslational modifications to transcriptional regulation [
10].
During the SUMOylation, SUMOs are covalently conjugated to protein substrates through canonical four-amino-acid SUMO conjugation sites ψ-K-x-D/E (where ψ is a hydrophobic amino acid and x is any amino acid). HIV-1 IN was demonstrated to contain three SUMOylation sites (45LKGE, 135IKQE and 243WKQE) at three Lys residues (K46, K136 and K244), and the impairment of IN SUMOylation correlated with an early replication defect. Even though the mechanism underlying IN SUMOylation is not clear, it was hypothesized that SUMOylation might regulate the proximity between IN and its co-factors, which is indispensable for efficient viral replication [
14]. IN has been shown to interact with multiple host proteins. Interestingly, a number of IN-interacting proteins, such as LEDGF/p75, Ku70, p300 and Rad52, are also SUMOylated [
15‐
18]. LEDGF/p75 is one of the most important cofactors for IN [
19]. Previous studies have shown that LEDGF/p75, as an IN-interacting protein, carried out multiple functions during HIV infection, including tethering IN to transcriptionally active regions of host chromosomes, enhancing the enzymatic activity of IN, stabilizing IN subunit-subunit interactions and promoting IN tetramerization and protecting IN from proteasomal degradation [
20‐
33]. Interestingly, a previous study revealed that SUMOylation–defective IN mutants still retained LEDGF/p75 binding ability [
7], and HIV IN SUMOylation mutations did not affect subcellular localization or viral DNA nuclear import [
7]. Therefore, conclusive biochemical and functional data are still elusive in terms of the impacts of IN SUMOylation during HIV-1 viral replication.
SUMOs can also noncovalently interact with other proteins through specific SUMO-interacting motifs (SIMs). The most well characterized SIM is V/I-x-V/I-V/I or V/I-V/I-x-V/I/L, where x can be any amino acid in a parallel or anti-parallel orientation [
34‐
36]. SUMO-modified proteins can interact with SIM-containing binding partners through noncovalent binding. Notable examples include the SIMs of human TRIM5α, which bind to SUMO-conjugated capsid protein and restrict M-MLV infection [
37], and RanBP2 SIM mediates its binding with the complex of RanGAP1/SUMO1 and Ubc9 [
38]. Overall, the functional consequences of the SUMO-SIM interaction vary considerably in different protein contexts, affecting protein SUMOylation, protein localization, and protein-protein interactions. (see a review [
39]).
Interestingly, IN was shown to interact with SUMO1 and SUMO2 in a yeast two-hybrid system and a coimmunoprecipitation (co-IP) assay [
13]. Thus, the intriguing questions to ask are whether IN bears SIMs and how the IN
SIM-SUMO interaction modulates the multiprotein complex formation among IN and its SUMOylated cellular cofactors to affect different functions of IN. In this study, we examined the amino acid sequence of IN and defined two functional SIMs (200IVDI203 and 257IKVV260) in the catalytic core domain (CCD) and C-terminus. The SIMs of IN were shown to negatively regulate SUMOylation of IN, differentially modulate the IN-LEDGF/p75 and IN-Ku70 interactions, and contribute to the nuclear translocation step of HIV-1 IN. These findings not only uncover two possible SIMs of IN but also provide novel mechanistic insights into the regulation of multiple functions of IN by SIMs.
Discussion
In this study, we report that HIV-1 IN bears two functional SIMs (SIM2 200IVDI203 and SIM3 257IKVV260), that negatively regulate the SUMOylation of IN, as well as the interaction between IN and SUMO E2 conjugation enzyme Ubc9. Also. The results indicate that SIMs in IN are required for its interaction with LEDGF/p75, but not with Ku70. Viruses carrying IN SIM mutants showed impaired viral reverse transcription, nuclear import and integration steps, resulting in defective replication. This report provides the first evidence for the roles of noncovalent SIMs in HIV IN for its functions during viral replication.
Sequence analysis revealed that the three putative SIMs (SIM1 72VILV75, SIM2 200IVDI203 and SIM3 257IKVV260) present in IN conform to this consensus (Fig.
1b). However, our mutational analysis suggests that SIM2 and SIM3 present the major sites for the IN-SUMO interaction. Within these two IN sequences, the hydrophobic residue I200 is implicated in both chromatin and LEDGF/p75 binding [
19]; V260 is involved in the multimerization and structural stabilization of IN [
52,
53]. In this study, the SIM2 mutant I200A/V201A had decreased IN-LEDGF/p75 binding affinity (Fig.
4b), which is consistent with a previous report [
19]. However, it remains an open question as to whether the IN
SIM-SUMO interaction has any impact on its chromatin association or multimerization. Remarkably, V201I, the mutant sequence that still conforms to the consensus SIM, occurs as a natural polymorphism in drug-naïve patients [
54,
55]. This conservative substitution within SIM2 200IVDI203 thus highlights the importance of this sequence in the functions of IN and HIV-1 replication.
Noncovalent SUMO binding or the SIM-SUMO interaction has been shown to facilitate SUMOylation of SIM-containing proteins [
50,
56‐
58]. For example, the interaction between the SIM of Sp100 and SUMO-Ubc9 enhanced the SUMOylation of Sp100 [
56]. This SIM-dependent SUMOylation has also been described in various SUMO targets, such as Daxx, RANBP2/Nup358, HIPK2 and BLM [
50,
57,
58]. Interestingly, our results also showed that the IN SIM mutant 3VI has increased SUMOylation levels compared with INwt, suggesting that IN SIMs negatively regulate SUMOylation (Fig.
3b). Another line of evidence from this study also strongly supports this conclusion. The mutant 3VI had significantly increased Ubc9 binding ability compared with INwt (Fig.
3c). Ubc9 directly binds the SUMO conjugation consensus ψ-K-x-D/E sequence in substrates [
59,
60], and mutating ψ-K-x-D/E abolishes both Ubc9 binding and the SUMOylation of substrate proteins [
60]. Similarly, a previous study reported that Srs2 SUMOylation inhibits its noncovalent SUMO binding, possibly due to the reduced availability of its SIM motif for interactions with SUMOylated proteins in general by Srs2 SUMOylation [
61]. Thus, it appears that noncovalent SUMO binding and covalent SUMO modification or SUMOylation can mutually regulate with each other, either positively or negatively.
The functional outcomes for noncovalent SIM/SUMO binding vary and are largely dependent on SUMOylated proteins and SIM-containing binding partners. SUMO binding through SIM(s) affects protein stability, cytosolic-nuclear translocation, and transcriptional regulation through altered protein-protein or protein-DNA interactions at the molecular level (see a review [
10]). To investigate whether the IN
SIM-SUMO interaction could mediate the binding of IN with its SUMOylated cellular cofactors, we tested the interactions between HIV-1 IN with two SUMOylated proteins, LEDGF/p75 and Ku70. Our data found that while the IN SIM mutant 3VI was severely impaired in IN-LEDGF/p75 binding, it showed a two-fold increase in IN-Ku70 binding affinity (Fig.
4b and c). Additionally, 3VI bound Nup62, a non-SUMOylated cofactor of IN [
62], at the same level as INwt (data not shown). These results together imply that IN SIMs might be involved in the regulation of the interaction between IN and SUMO-conjugated binding partners. Closer examination revealed that the SIM1 mutant M1, which still binds SUMO3, retains the full ability to interact with LEDGF/p75 (Fig.
4b). This observation thus strengthens the notion that SIM2 and SIM3 of IN are required for the IN-LEDGF/p75 interaction. However, we cannot formally exclude the possibility that creating six point mutations in 3VI might have profound impacts on other functions of IN, which may indirectly influence the IN-LEDGF/p75 or IN-Ku70 interaction. For example, V260 has been shown to be critical for the multimerization of IN [
53], and I200 is required for the chromatin binding ability of IN [
63]. The SIM-SUMO interaction has been shown to influence the subcellular localization of SIM-containing proteins. A prominent example is the sequestration of Daxx to PML nuclear bodies, which is mediated through the binding of the SIM located at the C-terminus of Daxx and SUMOylated PML [
50]. Another example is that the SIMs of PML, Sp100 and hDaxx are required for the recruitment of these proteins to herpes simplex virus type 1 (HSV-1)-induced foci, which also recruit SUMO proteins and SUMO E3 ligase PIAS2β [
51]. In the present study, our results from HeLa cells and 293 T cells revealed that the IN SIM mutant 3VI was severely impaired in its nuclear localization (Fig.
5a and b). More specifically, SIM2 200IVDI203 and SIM3 257IKII260 but not SIM1 72VILV75 are involved in the nuclear import of IN proteins (Fig.
5a). As both SIM2 and SIM3 mutants that do not exclusively localize to the nuclei are also impaired for IN-SUMO3 and IN-LEDGF/p75 interactions, and LEDGF/p75 was initially reported to be indispensable for the nuclear import and chromosomal targeting of IN [
31,
33], the cytoplasmic localization pattern of SIM2 and SIM3 mutants might be the reason or result of impaired IN-LEDGF/p75 interaction and/or chromatin association. In accordance with this hypothesis, SIM2 mutant I200A was shown to be defective for chromatin binding in a previous report [
19]. Second, we also consider the possibility that mutations introduced to IN SIMs might block the recruitment of SUMOylated cellular cofactor(s) of IN, which is required for the nuclear translocation of IN proteins. RANBP2 (Nup358) is one of the candidates that is SUMOylated and might be involved in the nuclear import of IN. RANBP2 is a nucleoporin with SUMO E3 ligase activity that harbors both SIMs and SUMO conjugation sites [
35,
57,
63,
64]. As an SUMO E3 ligase, this protein has been shown to promote the SUMOylation of a number of SUMO targets, including Mdm2, HDAC4, topoisomerase II-alpha, PML, or Sp100 [
57,
65‐
67]. Additionally, RANBP2 has been shown to be critical for HIV-1 replication and is involved in the nuclear import of the PIC and integration, both of which are closely linked with the functions of the HIV-1 IN [
68,
69]. We may speculate that SUMOylated proteins, such as RANBP2, might form a complex with IN through binding SIMs and function in the nuclear import of IN. This possibility requires further experimental investigation.
It is known that introducing numbers of IN mutations in the context of virus may cause pleiotropic effects during HIV-1 replication [
1,
70]. In order to limit such pleiotropic effects, we have used a previously described single cycle replicating virus system, in which the viral reverse transcriptase and integrase were complemented
in trans [
43] (Fig.
6a). Through the analyses with this HIV single cycle replication system, it showed that two functional SIMs (SIM2 200IVDI203 and SIM3 257IKVV260) are required for the multiple steps, including reverse transcription, nuclear import, and integration, in the early stage of HIV infection (Fig.
6). These observations are well correlated with the fact of SIMs are important in regulation of its SUMOylation, interacting with LEDGF/p75, and its ability for the nuclear import. Indeed several previous studies shown that the disruption of HIV integrase binding to endogenous LEDGF can significantly interfere with the progeny virus infectiousness [
71,
72]. Another study also reported that a compound, LEDGINs, which specifically inhibit IN interaction with LEDGF/p75, was able to disrupt virus assembly and lead to the large portion of progeny virions display aberrant morphogenesis [
73]. Whether the SIMs in the IN play similar roles during HIV morphogenesis still wait for more detailed studies.