Characterization of the Nuclear Import Pathway for HIV-1 Integrase*

The karyophilic properties of the human immunodeficiency virus, type I (HIV-1) pre-integration complex (PIC) allow the virus to infect non-dividing cells. To better understand the mechanisms responsible for nuclear translocation of the PIC, we investigated nuclear import of HIV-1 integrase (IN), a PIC-associated viral enzyme involved in the integration of the viral genome in the host cell DNA. Accumulation of HIV-1 IN into nuclei of digitonin-permeabilized cells does not result from passive diffusion but rather from an active transport that occurs through the nuclear pore complexes. HIV-1 IN is imported by a saturable mechanism, implying that a limiting cellular factor is responsible for this process. Although IN has been previously proposed to contain classical basic nuclear localization signals, we found that nuclear accumulation of IN does not involve karyopherins α, β1, and β2-mediated pathways. Neither the non-hydrolyzable GTP analog, guanosine 5′-O-(thiotriphosphate), nor the GTP hydrolysis-deficient Ran mutant, RanQ69L, significantly affects nuclear import of IN, which depends instead on ATP hydrolysis. Therefore these results support the idea that IN import is not mediated by members of the karyopherin β family. More generally, in vitro nuclear import of IN does not require addition of cytosolic factors, suggesting that cellular factor(s) involved in this active but atypical pathway process probably remain associated with the nuclear compartment or the nuclear pore complexes from permeabilized cells.

The viral RNA genome undergoes reverse transcription within PICs, leading to double-strand DNA that is ultimately integrated into host DNA by the viral integrase (IN). In contrast to oncoviruses that require mitosis to replicate, HIV is able to infect non-dividing cells, such as macrophages (1)(2)(3). This important characteristic for the physiopathology of HIV infection relies on a specific process that ensures efficient import of the viral DNA through the envelope of the interphase nucleus, prior to integration (4).
Nucleo-cytoplasmic transport processes occur through the nuclear pore complexes (NPC) of the nuclear envelope. NPC have a pore-like, molecular sieve function, whereby molecules smaller than 40 -45 kDa can diffuse into and out of the nucleus. Proteins larger than 40 -45 kDa require a nuclear localization signal (NLS) to be targeted to the nucleus. Different pathways have been described to account for nuclear import of karyophilic proteins, but most of these processes involve nuclear import receptors that belong to a family of related proteins named karyopherins or importins (for review, see Ref. 5). Basic amino acid-rich NLSs are recognized by a heterodimeric receptor composed of kap ␣, an adaptor protein between the NLS and the kap ␤1 protein that promotes interaction of the cargo-containing complex with the NPC (6 -12). The human ribonucleoprotein A1 protein contains another type of NLS, the M9 sequence, rich in glycine and aromatic residues, that directly binds its receptor, transportin or kap ␤2 (13,14). Other members of the kap ␤ family have been identified as import receptors for specific substrates, like SR proteins or ribosomal proteins (15)(16)(17)(18)(19). Beside their ability to interact with both nuclear pore complex and NLS-containing cargo, members of the karyopherin ␤ family also associate with the small Rasrelated GTPase Ran protein in its GTP-bound form (20,21). The nucleotide-bound status of Ran is determined by the chromatin-associated exchange factor RCC1, generating Ran-GTP in the nucleus, and the cytoplasmic GTPase activating protein RanGAP1, depleting Ran-GTP from the cytoplasm. This asymmetric distribution of Ran-GDP and Ran-GTP across the nuclear envelope is probably responsible for the directionality of nuclear transport. Indeed, binding of Ran-GTP to importins promotes the dissociation of the kap-NLS complex and the release of the cargo into the nucleus (22)(23)(24)(25)(26).
Nuclear import of nucleoprotein complexes, such as the HIV-1 PIC, although poorly documented, probably involves multiple NLSs that can be present in both protein and nucleic acid moieties. In particular, it has been reported recently that the central flap of the HIV-1 viral DNA is necessary for nuclear translocation (27). Moreover, three viral proteins, namely IN, the accessory protein viral protein R, and the matrix protein, have been proposed to participate in the nuclear import of HIV-1 PICs, but their respective contribution remains unclear and controversial (for review, see Ref. 28). It has been clearly shown that Vpr and IN not only display karyophilic properties both in vivo and in permeabilized cells but also rapidly accumulate in the nucleus of infected cells (29)(30)(31)(32). In contrast, conflicting data have been reported on the ability of MA to localize in the nucleus (33,34).
The tight association between IN and viral DNA within PIC supports IN as a good candidate for the nuclear import of viral DNA (35,36). HIV-1 IN is a 288-amino acid protein composed of three functionally independent domains: an N-terminal domain, which spans approximately the first 50 amino acids, a catalytic core domain (amino acids 50 -202), and a C-terminal domain (amino acids 202-288) (for review, see Refs. [37][38][39]. The first two domains are relatively well conserved among retroviral integrases. The N-terminal domain contains a zinc finger motif (HHCC) that participates in oligomerization of IN and stimulates catalytic activity. The core domain is responsible for the catalytic activity of the enzyme. Finally, the less conserved C-terminal region displays unspecific DNA binding properties, similar to that of the full-length integrase. Fulllength IN can multimerize to form both dimers and tetramers in solution, and in vitro assays revealed that IN functions as oligomers. It has been well reported that IN exerts pleiotropic effects on the HIV-1 life cycle (38). Whereas some mutations specifically block the integration step into the host DNA, others block replication at a step prior to integration, indicating that IN might be linked to nuclear import or intranuclear routing of the viral DNA. A better understanding of the mechanisms responsible for nuclear import of IN appears therefore to be essential to approach its role in the nuclear translocation of the viral PIC.
In this study, we used an in vitro transport assay based on digitonin-permeabilized cells to analyze the nuclear import pathway of full-length HIV-1 integrase. Although IN has been previously proposed to contain classical basic NLSs, we found that nuclear accumulation of IN does not involve karyopherins ␣, ␤1, and ␤2 and is independent of GTP hydrolysis and Ran. Furthermore, this process appears not to require soluble cytosolic factors but depends on ATP hydrolysis. These results therefore indicate that nuclear accumulation of IN is ensured by an active but atypical mechanism.

EXPERIMENTAL PROCEDURES
Cells and Culture Conditions-Adherent or S3 suspension HeLa cells were maintained in exponential growth in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum.
Preparation of HeLa Cell Cytosol-HeLa cell cytosol was prepared as described by Paschal and Gerace (40). 10 9 exponentially growing HeLa S3 cells were collected by centrifugation at 300 ϫ g for 10 min. The cells were washed twice with phosphate-buffered saline and once with lysis buffer (5 mM Hepes, pH 7.4, 5 mM potassium acetate, pH 7.4, 2 mM magnesium acetate, 1 mM EGTA, 2 mM dithiothreitol, and the following protease inhibitors: 10 g/ml each aprotinin, leupeptin, and pepstatin and 200 g/ml 4-(2-aminoethyl)benzenesulfonyl fluoride from Uptima). The cell pellet was resuspended in one volume of lysis buffer and disrupted in a tight fitting stainless steel homogenizer (as judged by phase contrast microscopy). The homogenate was diluted with 0.1 volume of 10ϫ transport buffer (transport buffer: 20 mM Hepes, pH 7.4, 110 mM potassium acetate, pH 7.4, 2 mM magnesium acetate, 0.5 mM EGTA, 1 mM dithiothreitol, and protease inhibitors) and centrifuged at 40,000 ϫ g for 30 min at 4°C. The supernatant was further centrifuged at 100,000 ϫ g for 1 h. The resulting supernatant (ϳ15 mg/ml as measured with the protein assay kit from Bio-Rad) was aliquoted, frozen in liquid N 2 , and stored at Ϫ80°C.
Recombinant Proteins-Recombinant IN produced in Escherichia coli was a generous gift from S. Escaich (Avantis Pharma, Ivry, France) (32). The expression vector for His-tagged IBB (pKW312) was provided by K. Weis (University of California, San Francisco, CA), and proteins were expressed and purified as described (41). The plasmid for GST-Ran expression was a gift from M. Dasso, and wild-type Ran was purified as described (National Institutes of Health, Bethesda, MD) (42). Expression and purification of RanQ69L was performed essen-tially as described (43) using the expression vector provided by C. Dingwall (SUNY, Stony Brook, New York). The expression plasmid for GST-M9 was a gift from G. Dreyfuss (University of Pennsylvania, Philadelphia, PA), and GST-M9 protein was purified as described previously (14).
Chemical Cross-link between BSA and IN-BSA (300 g) was labeled with 5(6)-carboxyfluorescein-N-hydroxysuccinimide ester (25 g; FLUOS, Roche Molecular Biochemicals) in 100 mM borate, pH 8.5. Resulting BSAfluorescein was coupled to sulfosuccinimidyl-4-(N-maleilidomethyl)cyclohexane-1-carboxylate as a cross-linker (150 g; Sulfo-SMCC, Pierce) in 50 mM borate, pH 7.6, purified on a G50 column equilibrated in 100 mM phosphate, pH 6.0, and concentrated. Fluorescein-BSA-SulfoSMCC was finally cross-linked to recombinant IN (130 g) in 100 mM phosphate, pH 6.0, and fluorescein-BSA-IN was purified on two G75 spin columns to eliminate free integrase and BSA-fluorescein. The resulting fusion protein was analyzed by gel electrophoresis and Western blotting using antiintegrase antibodies. No free IN could be detected in the final fraction that consisted in BSA coupled to at least two IN molecules (data not shown). BSA fused to NLS D or NLS P of IN was obtained by coupling fluorescein-BSA-SulfoSMCC to IN-derived peptides encompassing the putative NLS described in the HIV-1 IN sequence (CGGG 184 NFKRKGGI and CGGG 211 KELQKQITK, respectively) (19).
Nuclear Import Assay-Digitonin-permeabilized HeLa cells were prepared according to Adam et al. (44). Cells grown on coverslips were permeabilized with 55 g/ml digitonin (Sigma) in transport buffer. A standard 50-l nuclear import assay was performed in transport buffer containing an energy-regenerating system (1 mM ATP, 0.5 mM GTP, 10 mM creatine phosphate, and 0.4 units/ml creatine phosphokinase), 15 g/ml BSA-NLS-FITC (45), 25 l of HeLa cell cytosol (ϳ15 mg/ml), and 1.5 g/ml Cy3-coupled recombinant integrase (32). The reaction was allowed to proceed for 30 min at 30°C. Permeabilized cells were subsequently fixed with 2% paraformaldehyde and 0.1% glutaraldehyde, and coverslips were mounted in phosphate-buffered saline containing 50% glycerol. When indicated, cells were pre-incubated with 50 g/ml wheat germ agglutinin (Sigma) for 15 min prior to transport reaction. Competition experiments were performed using a 20-fold excess of non-labeled integrase.
Time-lapse Video Microscopy-Cells grown on 22-mm coverslips were permeabilized as described previously and maintained in transport buffer at 30°C in sealed chambers. The transport reaction was started by adding fluorescent proteins, cytosol, and the energy-regenerating system to the cells. Time-lapse sequences were collected on a Leica DMIRBE epifluorescence microscope equipped with a cooled CCD camera and controlled by Metamorph software (Universal Imaging). Images were recorded every minute.

RESULTS
Nuclear Accumulation of IN Results from a Temperature-dependent Vectorial Process-We recently reported that recombinant HIV-1 integrase coupled to Cy-3 fluorochrome (IN) is imported into nuclei of digitonin-permeabilized HeLa cells in the presence of HeLa cytosol and energy (32) (Fig. 1). This approach was thus chosen to further characterize the mechanisms responsible for the nuclear import of IN. Because the 32-kDa molecular mass of integrase is below the passive diffusion limit of the NPC, we first analyzed whether nuclear localization of IN could result from diffusion. Fluorescein-labeled BSA fused to the NLS of the SV40 large T antigen was used as a positive control for nuclear import in this in vitro system. Nuclear import of both IN and BSA-NLS was completely inhibited upon addition of wheat germ agglutinin (WGA; Fig. 1), a lectin that recognizes N-acetylglucosamine-modified nucleoporins and inhibits many nuclear transport pathways without affecting passive diffusion through the NPC (46)(47)(48). In addition, no nuclear accumulation of IN or BSA-NLS was observed when the nuclear transport reaction was carried out at 4°C (Fig. 1).
To confirm these results, nuclear accumulation of IN was monitored without washing and fixing nuclei using time-lapse video microscopy. Images were acquired every minute, allowing time-lapse nuclear import kinetics to be measured. As shown in Fig. 2, A and B, both IN and BSA-NLS were imported very rapidly into nuclei against a concentration gradient. In contrast, neither carbonic anhydrase, a protein with a molecular mass comparable with IN, nor BSA fused to the reverse NLS accumulated into nuclei under the same experimental conditions ( Fig. 2A). When the nuclear transport reaction was carried out at 4°C in the absence of washing or fixation, IN was excluded from the nucleus, whereas fluorescein-labeled 4-kDa dextran freely diffused to the nucleus (Fig. 2C). Taken together, these results indicate that IN is imported into the nucleus by a temperature-dependent and vectorial process that occurs through the NPC rather than by passive diffusion followed by nuclear trapping.
IN Nuclear Import Is Ensured by a Saturable Process That Does Not Involve Karyopherins ␣, ␤1, and ␤2-As we previously reported (32), IN nuclear import was strongly affected by addition of a 20-fold molar excess of unlabeled IN (Fig. 3A), indicating that IN import corresponds to a saturable process. However, IN did not compete on BSA-NLS nuclear import, suggesting that these proteins utilize distinct import pathways. It has been clearly demonstrated that transport of BSA-NLS is mediated by karyopherins ␣ and ␤1. To ensure that this pathway is not responsible for IN nuclear transport, an import reaction was performed in the presence of either unlabeled IBB, a protein corresponding to the importin ␤ (karyopherin ␤1)-binding site of importin/karyopherin ␣, or a peptide corresponding to the SV40 large T antigen NLS. Addition of IBB or NLS peptide clearly inhibited nuclear import of BSA-NLS (41) but did not affect IN transport, confirming that the classical NLS pathway is not involved in IN nuclear entry (Fig. 3, A and B).
Two different regions of HIV-1 IN have been proposed as putative NLSs ( 186 KRK and 211 KELQKQITK; Refs. 29 and 49) that could be recognized by importin ␣ Rch1. To analyze whether these regions are involved in the IN nuclear transport, an import reaction was performed in the presence of peptides (NLS P and NLS D) encompassing these signals. As shown in Fig. 3B, none of these peptides affected nuclear import of IN or BSA-NLS. In addition, fusion proteins corresponding to fluo-rescein-BSA coupled to NLS P or NLS D did not translocate into the nucleus, whereas BSA coupled to the SV40 large T antigen NLS was transported to this compartment (Fig. 3C). These data indicate that these regions do not function as independent and transferable NLSs recognized by importin ␣. Finally, an unlabeled fusion protein between GST and the nuclear import M9 sequence from human ribonucleoprotein A1 protein (14) did not compete with IN or BSA-NLS import (Fig.  3A), indicating that IN does not use the karyopherin ␤2/transportin-mediated pathway. Together, these results clearly demonstrate that IN nuclear import does not involve karyopherins ␣, ␤1, and ␤2.
The GTPase Ran Is Not Required for IN Nuclear Import-A common feature of karyopherin ␤ proteins involved in nuclear import or importins is their ability to bind their substrate in the cytoplasm in the absence of Ran in its GTP-bound form and to release their cargo in the nucleus upon binding to Ran-GTP (20,21). Consequently, importin-mediated import pathways are in-FIG. 1. Nuclear accumulation of integrase results from a temperature-dependent process that occurs through the nuclear pore complex. Digitonin-permeabilized HeLa cells were incubated for 30 min with Cy3-labeled integrase (left panels) or FITC-labeled BSA-NLS (right panels) in the presence of HeLa cytosol and an ATP-regenerating system at 30°C (upper panels), at 4°C (lower panels), or at 30°C after incubation with 50 g/ml WGA for 15 min prior to the transport reaction (middle panels). After incubation, cells were fixed and analyzed by direct fluorescence.  SLN), and Cy3-labeled carbonic anhydrase (CA) was performed in the presence of HeLa cytosol and an ATP-regenerating system. Nuclear accumulation of these proteins was directly visualized without washing and fixation by time-lapse video microscopy. Images were acquired every minute for 1 h and analyzed using the Metamorph software, and fluorescence intensity measured in the nucleus was expressed as a function of incubation time. B, nuclear accumulation of IN was directly visualized at the various indicated time points by time-lapse video microscopy. C, a nuclear import assay of Cy3-labeled IN or 4-kDa (4kD) FITC-labeled dextran was performed in the presence of HeLa cytosol and an ATP-regenerating system for 30 min at 4°C. Cells were visualized without prior washing and fixation. hibited both by the non-hydrolyzable GTP analog, GTP␥S, which can maintain Ran in a GTP-bound form in the cytoplasm and inhibit the importin/cargo interaction, and also by the dominant negative Ran mutant Q69L, which is unable to hydrolyze GTP. To determine whether nuclear accumulation of IN is ensured by one member of the importin ␤ family, an IN import reaction was performed in the presence of either GTP␥S or RanQ69L. As reported previously (23,50), nuclear import of BSA-NLS was completely inhibited under these conditions, and the fluorescence intensity measured in nuclei incubated with RanQ69L corresponded to 6 Ϯ 0.8% of the control condition. In contrast, IN nuclear accumulation was clearly not affected by GTP␥S and not significantly impaired upon addition of RanQ69L (nuclear fluorescence intensity measured in the presence of RanQ69L corresponded to 89 Ϯ 21% of the control condition; Fig. 4). The Ran GTPase appears therefore not to be essential for IN nuclear import that occurs through a non-classical and probably importin ␤-independent import pathway.
IN Nuclear Import Is Cytosol-independent but ATP-dependent-In agreement with previous studies (44), BSA-NLS was not imported into the nucleus when the transport reaction was carried out in the absence of cell extracts (Fig. 5A). In contrast, IN nuclear accumulation was not affected when HeLa cytosol was omitted from the transport reaction (Fig. 5). This result suggests either that no soluble factor is involved in IN import or that factors responsible for this process are not extracted during the digitonin permeabilization procedure. Nuclear accumulation of IN in the absence of extracts appeared, however, to be completely inhibited by WGA and dependent on addition of energy (Fig. 5B). To confirm that nuclear accumulation of IN in the absence of extracts results from an active transport, nuclear import of a fusion protein consisting in IN chemically fused to BSA (BSA-IN) was analyzed. Although BSA-IN is too large to passively diffuse through NPC, it accumulated in the nucleus in the absence of extracts (Fig. 5C). Upon treatment with WGA or in the absence of energy, BSA-IN was not imported into the nucleus but rather docked at the nuclear envelop (Fig. 5C). These results indicate that IN nuclear import observed in the absence of extracts results from an active transport rather than a diffusion process.
To further characterize the energy requirement of IN transport, import assays were performed in buffer complemented with different nucleotides. Addition of GDP, GTP, or ADP did not promote any significant accumulation of IN into the nucleus ( Fig. 6 and data not shown). In contrast, IN readily entered the nucleus in the presence of ATP. To ensure that endogenous phosphotransferase activity does not form GTP, import assays was performed with ATP and an excess of UDP. A similar import of IN was observed in the presence of ATP alone or supplemented with UDP (data not shown). To test whether ATP hydrolysis was required for this process, an import reaction was carried out with the non-hydrolyzable ATP analog AMP-PNP. As shown in Fig. 6, IN only weakly accumulated in the nucleus under this experimental condition, indicating that ATP hydrolysis is essential for an efficient nuclear import of IN. This result was further confirmed by the increased nuclear import observed when ATP was supplemented with an ATP-regenerating system that maintains a constant ATP concentration. Taken together, these data clearly show that IN nuclear import is ensured by an active non-classical transport pathway that requires ATP hydrolysis but does not involve cytosolic factors or the Ran GTPase.

DISCUSSION
In this study, we used an in vitro system to characterize the mechanisms involved in the nuclear import of full-length HIV-1 integrase. IN rapidly accumulates in the nucleus of digitonin-permeabilized cells in a temperature-and energy-dependent and saturable manner that occurs against a concentration gradient. Nuclear import of IN was completely inhibited by WGA, a lectin that binds specific nucleoporins, thus inhibiting specific transport processes without affecting diffusion (46 -48). Moreover, a fusion protein between IN and BSA that is too large to diffuse through NPC also accumulates in the nucleus. Together these data show that nuclear accumulation of HIV-1 IN does not result from passive diffusion but rather from an active transport that occurs through the NPC.
HIV-1 IN is imported by a saturable mechanism, implying that a limiting cellular factor is responsible for this process. IBB, SV40 T antigen NLS, and GST-M9 were unable to compete on IN nuclear import, indicating that IN does not use karyopherins ␣, ␤1, and ␤2-mediated pathways to enter the nucleus of digitonin-permeabilized cells (14,41). Moreover, neither the non-hydrolyzable GTP analog nor the GTP hydrolysisdeficient Ran mutant, RanQ69L (23, 50), significantly affects and BSA-NLS (right panels) was performed in the presence of HeLa cytosol and an ATP-regenerating system (ϩ extracts) or in buffer containing the energy system alone without cell extracts (Ϫ extracts). B, a nuclear import assay was carried out in the absence of HeLa cytosol with (Control) or without energy (Ϫ energy) or with energy after preincubation with 50 g/ml WGA for 15 min prior to transport reaction (WGA). C, a nuclear import reaction of IN fused to BSA was performed in the absence of HeLa cytosol with (Control) or without energy (Ϫ energy) or with energy after pre-incubation with 50 g/ml WGA for 15 min prior to transport reaction (WGA). with karyopherin ␣ Rch1 through two putative NLSs (29), suggesting that IN would be imported by a karyopherin ␣/␤1mediated pathway. Nevertheless, there is no direct evidence that this interaction is functionally relevant in vivo. Synthetic peptides encompassing these putative NLSs did not affect the nuclear import of either IN and BSA-NLS, and we failed to detect an interaction between IN and Rch1 in a two-hybrid assay (data not shown). Finally, in vitro nuclear import of IN does not require addition of cytosolic factors. Cellular factor(s) involved in this process are therefore not depleted during the permeabilization procedure and remain strongly associated with the nuclear compartment or the NPC. Attempts to prepare nuclei defective for IN import by selective extraction without affecting the integrity of the nuclear envelope have failed so far, and alternative methods should now be considered to identify the cellular partners of IN required for its nuclear import. An intriguing feature of IN nuclear import resides in its requirement for ATP hydrolysis. Although no nuclear import or docking of IN at the NPC was detected in the absence of ATP, a significant association with the nuclear envelop was observed when IN fused to BSA was used as a substrate in an import reaction containing no ATP. These observations suggest that IN may require ATP either to interact with the mediator of its import or to translocate through the NPC.
An increasing number of proteins imported by unconventional mechanisms have been reported recently. In particular, nuclear import of the RNA-binding proteins U1A and U2BЉ, which are components of the uracil-rich small nuclear ribonucleoproteins, displays characteristics closely related to the import of HIV-1 IN. U1A and U2BЉ import occurs independently of members of the karyopherin ␤ family, is not inhibited by non-hydrolyzable GTP analogs or a hydrolysis-deficient Ran mutant, and does not require addition of cytosolic factors in semipermeabilized cells (51). However, extraction of nuclei using a high MgCl 2 concentration led to nuclei deficient for import of U1A and U2BЉ unless extracted nuclear proteins were added in the transport reaction, suggesting that this import pathway involves nucleus-associated factors. Finally, U1A, U2BЉ, and IN are the only proteins described so far to require ATP hydrolysis for their nuclear import. These common features suggest that these proteins may use similar nuclear import pathways. It might thus be interesting to determine whether U1A and U2BЉ are able to cross-compete on the IN import pathway.
Characterization of the domain of IN involved in its nuclear import would certainly provide a useful tool to better understand the molecular mechanisms responsible for this process. Two different regions of HIV-1 IN have been proposed as putative NLSs ( 186 KRK and 211 KELQKQITK; Refs. 29 and 49) that could be recognized by Rch1. We show here that none of these regions is sufficient to target a reporter protein to the nucleus or to interfere with nuclear import of IN or BSA-NLS. Mutations in these sequences were found to affect IN nuclear localization, but contradictory results have been reported recently (52). Moreover, alteration of these motifs also drastically reduces dimerization of the protein and abolishes the reverse transcription and integration processes (31,52). These data indicate that these regions are probably not directly responsible for nuclear import of IN but that mutations within these regions probably affect the overall conformation of the protein, resulting in inhibition of nuclear import. The in vitro reconstitution assay presented here will certainly represent a useful tool to analyze the sequence requirement for IN nuclear import and allow for the independent control of the dimerization, DNA binding, and integration properties of IN.