Identification and characterization of a functional nuclear localization signal in the HIV-1 integrase interactor LEDGF/p75.

Human lens epithelium-derived growth factor (LEDGF)/p75 protein forms a specific nuclear complex with human immunodeficiency virus type 1 (HIV-1) integrase and is essential for nuclear localization and chromosomal association of the viral protein. We now studied nuclear import of LEDGF/p75 in live and semipermeabilized cells. We show that nuclear import of LEDGF/p75 is GTP-, Ran-, importin-alpha/beta-, and energy-dependent and that the protein competes with the canonical SV40 large T antigen nuclear localization signal (NLS) for nuclear import receptors. We identified the NLS of LEDGF/p75 through deletion analysis and site-directed mutagenesis. The LEDGF/p75 NLS, 148GRKRKAEKQ156, belongs to the canonical SV40-like family. Fusion of this short peptide to the amino terminus of Escherichia coli beta-galactosidase rendered the fusion protein nuclear, confirming that the LEDGF/p75 NLS is transferable. Moreover, a single amino acid change in the NLS was sufficient to exclude the mutant LEDGF/p75 protein from the nucleus and abolish nuclear import of HIV-1 integrase.

has been shown to associate with HIV-1 IN in human cells (30). Using transient knock-down of endogenous LEDGF/p75 via small interfering RNA, LEDGF/p75 was shown to be both necessary and sufficient for accumulation of HIV-1 IN into the nucleus (31). Although the alternative splice variant of LEDGF/p75, p52, is also karyophilic, its nuclear distribution is different from LEDGF/p75 (31,32). p52 did not display affinity for HIV-1 IN in vitro and did not co-localize with fluorescently tagged IN in live cells (31).
The exact cellular function of LEDGF/p75 is unknown. LEDGF/p75 was originally isolated from HeLa cell extracts as an interactor of transcriptional co-activator PC4 and was therefore suggested to play a role in transcriptional regulation (33). Singh and co-workers (34 -36) subsequently reported a role for LEDGF/p75 in cellular stress response and survival. LEDGF/ p75 binds to heat shock and stress-related regulatory DNA elements, and cellular expression of LEDGF/p75 was increased upon heat, oxidative, or chemical stress.
Since LEDGF/p75 was shown to be essential for the nuclear accumulation of HIV-1 IN, we characterized the nuclear import pathway of LEDGF/p75 and identified its NLS. We show that a single point mutation is sufficient to confine LEDGF/p75 and relocate HIV-1 IN to the cytoplasm.
To obtain pGG-P75, the EagI/BamHI fragment from pEGFP-P75 carrying the LEDGF/p75 open reading frame was subcloned between HindIII/BamHI sites of pGG (the EagI and the HindIII termini were treated with mung bean nuclease prior to BamHI digestion).
To make pGG-P75/⌬C and pGG-P75/Ct, the plasmids pEGFP-P75/⌬C and pEGFP-P75/Ct were digested with BglII and treated with T4 DNA polymerase followed by BamHI restriction. The LEDGF/p75 gene fragments were then cloned between the HindIII and BamHI sites of pGG.
LacZ Fusion Constructs-The plasmids pGM-SV40NLS-lacZ and pGM-lacZ were made using the following primers: SV s1 , 5Ј-AAGAGG-AAGGTCCTGCAGATGCTAGATCCCGTCGTTTTACAAC; SV as , 5Ј-CT-ACTCGAGCTATTTTTGACAGGAGACCAACTGG; SV s2 , 5Ј-CTAAAG-CTTGACACCATGGGATCCCCTAAGAAAAAGAGGAAGGTCCTG-CAG; LacZ s , 5Ј-CTAAAGCTTGACACCATGGGCATGCTAGATCCCGT-CGTTTTACAAC. pHRЈCMVlacZ (37) was used as the template to PCR-amplify the Escherichia coli ␤-galactosidase sequences. To build pGM-lacZ, LacZ s and SV as were used as primers. For the construction of pGM-SV40NLS-lacZ, a fragment amplified using SV s1 and SV as primers was used as template in a second round of PCR with primers SV s2 and SV as . Following HindIII and XhoI digestion, the resulting PCR products were subcloned into HindIII and XhoI restriction sites of pcDNA6/V5-HisB (Invitrogen). To construct pGM-P75NLS-LacZ, the SV40 NLS coding sequence was removed from pGM-SV40NLS-lacZ by BamHI and PstI restriction and replaced by a DNA duplex, formed by oligonucleotides 5Ј-GATCCGGGCGTAAGCGGAAGGCTGAAAAACA-ACTGCA and 5Ј-GTTGTTTTTCAGCCTTCCGCTTACGCCCG. All plasmid constructs were verified by sequence analysis to contain no unwanted mutations.
Purification and Fluorescent Labeling of Recombinant LEDGF/p75 and LEDGF/p75 A150 -The bacterial expression vector pCPNat75 for expression of native LEDGF/p75 has been described (31). To produce recombinant LEDGF/p75 A150 , pCPN75-A150 was engineered by QuikChange (Stratagene) using pCPNat75 as template. Wild type LEDGF/p75 and the NLS mutant LEDGF/p75 A150 were expressed in E. coli BL21(DE3) pLysS cells and purified as previously described for the wild type protein (31). The point mutation did not significantly influence binding of the recombinant protein to heparin-Sepharose or cation exchange resins. Both proteins were labeled using Alexa Fluor 633 succinimidyl ester (Molecular Probes, Inc., Eugene, OR) in phosphate buffer according to the manufacturer's instructions. The stoichiometry of label/protein was maintained at Յ1 during labeling.
Cell Culture and Transfections-HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 IU/ml of penicillin, and 100 g/ml streptomycin (Invitrogen) at 37°C in a 5% C0 2 humidified atmosphere. HeLa cells were seeded 1 day prior to transfection into 8-well LabTek chambered cover glass cuvettes. Transfection was performed at ϳ80% confluence with LipofectAMINE 2000 (Invitrogen) using 0.36 g of plasmid DNA.
Western Blot-Cells were lysed 24 h post-transfection in 20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 0.1% SDS, and the total protein concentration was determined using the BCA protein assay (Pierce). Samples containing 15 g of total protein were separated by 11% SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes. Monoclonal mouse anti-LEDGF p75/p52 antibody was purchased from BD Biosciences (San Jose, CA). Horseradish peroxidase-conjugated secondary goat anti-mouse antibody was from Dako (DakoCytomation California Inc., Carpenteria, CA). Detection was carried out using ECLϩ chemiluminescent horseradish peroxidase substrate (Amersham Biosciences).
Semipermeabilized Cells-Rabbit reticulocyte lysate was purchased from Promega (Madison, WI). After ultracentrifugation at 100,000 ϫ g, the lysate was extensively dialyzed against transport buffer (TB, 20 mM Hepes, pH 7.3, 110 mM potassium acetate, 2 mM magnesium acetate, 1 mM EGTA, 2 mM dithiotreitol, 1 g/ml each aprotinin, leupeptin, and pepstatin, and 1 mM phenylmethylsulfonyl fluoride) in a Pierce dialysis cassette with a 10-kDa molecular weight cut-off. The in vitro import assay in semipermeabilized cells was based on methods described in Ref. 38. HeLa cells were seeded on LabTek coverslip cuvettes and used at ϳ50 -60% confluence. Cells were washed twice in ice-cold TB and permeabilized in 40 g/ml digitonin in TB on ice for 5 min. The endogenous cytosol was depleted by washing in ice-cold TB. Import reactions were incubated at 37°C for 30 min and contained 50% rabbit reticulocyte lysate (v/v), an energy-regenerating system (1 mM ATP, 1 mM GTP, 10 mM creatine phosphate, 20 units/ml creatine phosphokinase (Sigma)), and an import substrate in TB. Sulforhodamine B-labeled bovine serum albumin-conjugated NLS (henceforth referred to as SV40-NLS substrate) carrying the SV40 large T antigen NLS was purchased from Sigma. The SV40-NLS substrate was added to the import mixture at a final concentration of 120 g/ml. Alexa 633-labeled LEDGF/p75 and LEDGF/p75 A150 were added at a final concentration of 100 g/ml. To test GTP dependence, GTP was omitted, and 1 mM GTP␥S (Sigma) was added. ATP dependence was tested by replacing ATP with 1 mM AMP-PNP (Sigma). Competition experiments were carried out adding a 5-fold excess of unlabeled substrate. Active nuclear import was inhibited using 50 g/ml wheat germ agglutinin (WGA) (Biomeda, Foster City, CA) in TB. Cells were incubated for 30 min with WGA at 37°C and rinsed once in TB prior to incubation with the import mixture. For reconstitution experiments, a Ran mixture containing 3 M Ran (Jena Bioscience, Jena, Germany) and 0.5 M nuclear transport factor 2 (Sigma) in TB was used, with or without 1 M import receptors imp-␣ (human) and imp-␤ (Drosophila melanogaster) (Jena Biosciences). The import was terminated by washing cells with ice-cold TB followed by fixation in 2% formaldehyde (in phosphate-buffered saline). Slides were mounted in 50% glycerol.
Laser-scanning Microscopy-Confocal microscopy was performed using the Bio-Rad MRC-1024 microscope, interfaced with the Zeiss Axiovert microscope. All single-and two-color images were acquired in 1024 ϫ 1024 resolution mode, using a ϫ 63 water immersion objective. EGFP was excited at 488 nm, rhodamine and HcRed1 at 568 nm, and Alexa 633 at 647 nm.

RESULTS
Computer Analysis-LEDGF/p75 and its smaller splice variant p52 are karyophilic (30 -32). LEDGF/p75 contains about 21% basic residues, which makes it difficult to reliably predict a NLS from the primary sequence. Thus, we turned to computer prediction programs, first using NUCDISC from the PSORT II server (available on the World Wide Web at psort. nibb.ac.jp) (39,40), which looks for canonical NLSs using the patterns shown in Table I. PSORT II uses the k-nearest neighbor algorithm to score putative NLSs (41). Six putative NLSs within LEDGF/p75 were predicted (Table I). We also used the PredictNLS program provided by the Columbia University Bioinformatics Center (available on the World Wide Web at cubic.bioc.columbia.edu/cgi/var/nair/resonline.pl) (42), which searches for possible NLSs by comparing the protein sequence with a data base composed of experimentally determined NLSs expanded with "in silico mutagenesis"-created NLS motifs. Two overlapping NLSs were predicted by PredictNLS, both classified as SV40-like canonical NLSs (Table I).
The NLS of LEDGF/p75 Is Contained within the 325 Nterminal Amino Acid Residues-The LEDGF gene encodes for two alternative splice products, p75 and p52, which contain the same sequence over their N-terminal 325 amino acids (33). Since both LEDGF/p75 and p52 are karyophilic (31,32), we reasoned that the NLS must be confined within the first 325 residues. A pair of deletion mutants was constructed to test our prediction. pEGFP-P75/⌬C and pEGFP-P75/Ct encoded for amino acids 1-325 and 326 -530 of LEDGF/p75, respectively, fused to the C terminus of EGFP. Confocal microscopy of transiently transfected HeLa cells illustrated that full-length LEDGF/p75 fused to EGFP, EGFP-p75, exhibited a typical heterogeneous distribution pattern in the nucleus and was excluded from the nucleoli (Fig. 1A), as was previously reported (31,32). EGFP-p75/⌬C was also concentrated in the nucleus and had a distribution pattern indistinguishable from the fulllength LEDGF/p75 protein (Fig. 1B). EGFP-p75/Ct, however, was dispersed throughout the whole cell with somewhat higher concentration in the nucleus, but the nuclear distribution did not reveal the heterogeneous pattern typical of the full-length protein and EGFP-p75/⌬C (Fig. 1C). Since these EGFP-fused proteins have molecular weights near the exclusion limit of the NPC (64 kDa for EGFP-p75/⌬C and 51 kDa for EGFP-p75/Ct), the deletion mutants and full-length LEDGF/p75 were incorporated into a larger protein, EGFP fused to GST (starting mass 54.6 kDa; abbreviated as GG). As can be seen in Fig. 1E, GG-p75/Ct was excluded from the nucleus, as was GG (data not shown), whereas both the full-length protein (GG-p75; Fig. 2A) and the GG-p75/⌬C (Fig. 1D) were nuclear localized. Therefore, a NLS resides within the amino-terminal 325 residues of LEDGF/p75. The NLS predicted by PSORT II at amino acid position 509 (KKKP; see Table I) therefore does not play a role in the nuclear localization of LEDGF/p75.
Site-directed Mutagenesis of the Putative LEDGF/p75 NLS-On the basis of our deletion analysis and the computer-predicted NLSs, six putative sequences remained to be analyzed. Since one putative NLS recurred in both prediction programs, we first analyzed site-directed mutants centered around Arg 149 in the 146 -156 amino acid stretch of LEDGF/p75 (Table I), in which Lys and/or Arg residues were changed to Ala (Figs. 2 and 3). The mutations were introduced in the pGG-P75 vector. GG-p75, with a predicted molecular mass of ϳ115 kDa, far exceeds the exclusion limit for free diffusion through the NPC. We first constructed pGG-P75 A146 -147 in which Arg 146 and Arg 147 were both changed to Ala. Transient expression in HeLa cells showed that altering these two arginines did not affect the nuclear localization of LEDGF/p75 (Fig. 2B). We then made the following set of mutants: pGG-P75 A150 -152/155 (Fig. 2C), pGG-P75 A150 -152 (Fig. 2D), pGG-P75 A149 -150/155 (Fig. 2E), and pGG-P75 A150/155 (Fig. 2F). All of these LEDGF/p75/NLS mutants were excluded from the nucleus upon transient expression in HeLa cells (Fig. 2, C-F). From this, we deduced that Lys 150 and/or Lys 155 were critical for nuclear localization (Fig. 3A). We therefore introduced single point mutations K150A and K155A into the expression construct. Transient expression in HeLa cells of GG-p75 A150 and GG-p75 A155 revealed that Lys 150 was essential for LEDGF/ p75 nuclear localization under conditions wherein Lys 155 was dispensable (Fig. 3B). All nuclear import-defective LEDGF/p75 mutants showed a diffuse cytoplasmic distribution with varying levels of aggregation (e.g. see Figs. 2F and 3B). Western blot analysis using monoclonal anti-LEDGF/p75 antibody showed that the NLS mutants were all efficiently expressed in HeLa cells and had apparent molecular masses close to the predicted 115 kDa (Fig. 3C).
Co-expression of pGG-p75 A150 Renders HIV-1 IN Cytoplasmic-As previously described (31), fusing EGFP and red fluorescent protein HcRed1 to LEDGF/p75 and HIV-1 IN, respectively, did not perturb their nuclear co-localization upon transient expression in HeLa cells. In addition, the cellular distribution of LEDGF/p75 was not changed when fused to the larger GG double fusion ( Fig. 2A). As expected, upon co-expression with the HcRed1-tagged HIV-1 IN protein (HcRed1-IN), GG-p75 was co-localized with HcRed1-IN in nuclei of transfected cells (Fig. 4A). Strikingly, when HcRed1-IN was coexpressed with the NLS-defective mutant GG-p75 A150 , both proteins became confined to the cytoplasm, forming apparent co-aggregates (Fig. 4B). Similar patterns were observed when other GG-p75 NLS-defective mutants were co-expressed with HcRed1-IN (data not shown).
The LEDGF/p75 NLS Is Transferable-We fused the LEDGF/p75 NLS, 148 GRKRKAEKQ 156 , to the N terminus of the E. coli ␤-galactosidase protein to test whether this sequence is sufficient to target a heterologous protein into the nucleus. Since the NLS of SV40 large T antigen is a canonical example of a transferable NLS (43), we also prepared a control construct that expressed the SV40 NLS fused to ␤-galactosidase. Upon transient transfection of HeLa cells with both constructs and staining with X-gal, ␤-galactosidase activity was clearly detected in the nuclei of transfected cells (Fig. 5, A and  B). As expected, when wild type NLS-less protein was expressed from a control vector, ␤-galactosidase activity was detected throughout the cells (Fig. 5C).
Nuclear Import of LEDGF/p75 Is a Saturable and Temperature-dependent Process and Requires Ran-To characterize the nuclear import pathway of LEDGF/p75 in more detail, we used the semipermeabilized cell system (38). In this assay, the cytoplasmic membrane is first permeabilized with digitonin, and the soluble endogenous cytosolic factors are depleted. Nuclear import of a fluorescently labeled protein can then be studied in the presence of exogenous nuclear import factors. We initially used a rabbit reticulocyte lysate as the complete source for the import factors (38). In the presence of the reticulocyte lysate, ATP, GTP, and an energy regenerating system, both Alexa 633-labeled recombinant LEDGF/p75 and rhodamine-labeled SV40-NLS substrates were readily imported into nuclei (Fig.  6A). Import was not observed when the reticulocyte lysate was omitted from the mixture (Fig. 6A). This result demonstrates that import of LEDGF/p75 is dependent on soluble cellular factors. The lectin WGA inhibits many nuclear transport mechanisms without affecting passive diffusion through the NPC by binding to N-acetylglucosamine-modified nucleoporins (44,45). When we pretreated digitonin-permeabilized HeLa cells with 50 g/ml WGA prior to the addition of LEDGF/p75 or bovine serum albumin-SV40 NLS to the import mixture, no nuclear import was detected (Fig. 6A). In addition, nuclear import was not observed at 4°C or when the ATP-regenerating system was absent and ATP was replaced by nonhydrolyzable AMP-PNP (Fig. 6A). Importantly, an excess of unlabeled recombinant LEDGF/p75 was able to compete with the fluorescent LEDGF/ p75 conjugate for import, which proves that nuclear import of LEDGF/p75 is a saturable process (Fig. 6B).
Nuclear import mediated by cNLSs is dependent on a gradient of RanGTP across the nuclear envelope. imp-␤ can only bind cargo in the cytoplasm if it is released from RanGTP via GTP hydrolysis. At the same time, imp-␤ can only release its cargo in the nucleus by binding to RanGTP. Therefore, imp-␤mediated import pathways are inhibited by RanQ69L, a Ran mutant incapable of hydrolyzing GTP (46). Similarly, nonhydrolyzable analogues of GTP, like GTP␥S, also inhibit cNLSdependent import (47,48). Fig. 6C demonstrates that nuclear accumulation of LEDGF/p75 was dependent on RanGTP, since , and GG-p75/Ct (E). The deletion mutant lacking the C-terminal domain of LEDGF/p75 exhibits a nuclear distribution pattern equal to full-length LEDGF/p75. The C-terminal region however is not able to transfer the GG fusion protein to the nucleus and is thus dispensable for the nuclear localization of LEDGF/p75. import was not detected when RanQ69L was added to the import mixture or when GTP was replaced by GTP␥S.
LEDGF/p75 Is Imported into the Nucleus through the imp-␣/␤ Pathway-Nuclear import of SV40 large T requires both imp-␣ and imp-␤ nuclear transport receptors. The results described above showed that LEDGF/p75 contains a SV40-like NLS and is imported into the nucleus in a temperature-and energy-dependent, saturable manner. Its dependence on Ran and GTP suggested that nuclear import of LEDGF/p75 most likely occurs through the imp-␤ transport receptor. To investigate whether nuclear import of LEDGF/p75 might occur via the same mechanism as SV40 large T, we performed the following competition experiments (Fig. 7B). A 5-fold molar excess of unlabeled LEDGF/p75 was allowed to interact with the import receptors prior to the addition of the rhodamine-conjugated SV40-NLS substrate. Fig. 7B shows that the SV40-NLS substrate was not imported under these conditions. As a control, we analyzed purified NLS mutant LEDGF/p75 A150 protein. As predicted, Alexa 633-labeled mutant protein was exclusively cytoplasmic at conditions that supported the ready transport of wild-type LEDGF/p75 (Fig. 7A). Consistent with this observation, the mutant protein was unable to compete with the SV40-NLS substrate for its nuclear import (Fig. 7B).
Proteins containing basic NLSs can be imported into the nucleus by imp-␤ either with (13) or without (49) the adaptor molecule imp-␣. Another imp-␤-like receptor, imp-7, has been implicated in the nuclear import of the highly positively charged linker histone H1. However, a heterodimer of imp-7 with imp-␤ was required for H1 nuclear import (21). We used a reconstitution assay to determine whether both imp-␣ and imp-␤ were required for LEDGF/p75. Ran GTPase alone or Ran and imp-␤ (Fig. 8) were insufficient to import LEDGF/p75 into the nucleus. It was only upon the addition of both imp-␣ and imp-␤ to the Ran mixture that nuclear import of LEDGF/ FIG. 6. Nuclear import of LEDGF/p75 is a saturable process that depends on cytosolic factors and occurs in a temperature-, Ran-, GTP-, and energy-dependent manner. Upon permeabilization of the plasma membrane and depletion of soluble cytosolic factors, cells were incubated with import mixtures containing 1 mM GTP, an energy-regenerating system, rabbit reticulocyte lysate, and fluorescently labeled import substrate in transport buffer. A, 100 g/ml Alexa 633-labeled LEDGF/p75 (upper panels) or 120 g/ml rhodamine-labeled SV40-NLS substrate (lower panels) (ϩ). Import buffer, the rabbit reticulocyte lysate was omitted. WGA, cells were pretreated with WGA before the addition of the import mixture. 4°C, nuclear import was performed at 4°C. AMP-PNP, ATP and the energy-regenerating system was omitted and replaced by 1 mM AMP-PNP. After incubation, cells were washed once in transport buffer, fixed, and studied by confocal fluorescent microscopy. B, nuclear import of fluorescently labeled LEDGF/p75 was performed in the absence (ϩ) or presence of a 5-fold molar excess of unlabeled LEDGF/p75 (ϩ LEDGF/p75). C, fluorescently labeled LEDGF/p75 was added to the permeabilized HeLa cells in the presence of the rabbit reticulocyte lysate, 1 mM GTP, and an energy-regenerating system (ϩ). RanQ69L, 3 M RanQ69L was included in the reaction mixture. GTP-␥-S, GTP␥S (1 mM) was substituted for GTP.
FIG. 5. The NLS of LEDGF/p75 is transferable. The NLS of LEDGF/p75 (residues 148 -156) or SV40 large T was fused to the N-terminus of ␤-galactosidase, and the fusion proteins were expressed in HeLa cells. Twenty-four h post-transfection, cells were fixed and stained with X-gal. Both the LEDGF/p75 NLS (A) and the SV40 NLS (B) restricted ␤-galactosidase activity to the nuclei of transfected cells. In contrast, cells transfected with a plasmid expressing wild type ␤-galactosidase were stained uniformly (C). p75 was detected, as was the case for the SV40-NLS substrate (Fig. 8).

Identification and Characterization of the LEDGF/p75
NLS-LEDGF/p75 and its splice variant p52 are karyophilic proteins (30 -32). Although both proteins are nuclear, they each have a markedly different spatial and temporal distribution in the nucleus (31,32). LEDGF/p75 and p52 share 325 N-terminal residues (33). p52 has a unique 8-amino acid tail not present in LEDGF/p75, which does not contain an NLS-like pattern. Thus, at least one NLS was expected to reside within the N-terminal 325 residues of LEDGF/p75. Expression of GG-p75/⌬C and GG-p75/Ct deletion mutants in HeLa cells demonstrated that the C-terminal domain of LEDGF/p75 was not karyophilic and therefore does not contain a functional NLS (Fig. 2). Moreover, since the ⌬C mutant of LEDGF/p75 had a nuclear distribution pattern indistinguishable from the fulllength LEDGF/p75, we speculate that it is not the lack of this C-terminal domain in p52 but rather the presence of the extra 8 amino acid residues that determines the specific nuclear distribution of p52. The function of the C-terminal domain of LEDGF/p75 remains to be determined. It has recently been suggested that it might play a role in the interaction of LEDGF/ p75 with HIV-1 IN (31).
Computer analysis of the amino acid sequence predicted several putative NLSs within LEDGF/p75 (Table I). Through site-directed mutagenesis, we identified a single functional NLS in LEDGF/p75, 148 GRKRKAEKQ 156 . Remarkably, a single amino acid change, K150A, was sufficient to block LEDGF/ p75 nuclear localization and, as a consequence, redistribute the mutant protein to the cytoplasm (Fig. 3). We therefore conclude that this single NLS of the eight predicted NLSs functions as a true NLS. We do note, however, that 24 h post-transfection, weak nuclear staining could be detected in about 2-5% of the cells expressing GG-p75 A150 (or any other NLS mutant deficient in targeting GG-p75 to the nucleus) with a pattern similar to the nonmutated LEDGF/p75. We speculate that this low level of nuclear localization was a consequence of mitosis, wherein mutant proteins can bind to mitotic chromosomes following breakdown of the nuclear envelope and subsequently become captured in the nuclei. Compared with 100% nuclear import of wild type LEDGF/p75, this low level of mutant protein import does not affect our conclusions. LaCasse and Lefebvre (50) reported that in 67% of proteins that possess both DNA binding and NLS activities, the NLS overlaps with the DNA binding sequence. Cokol et al. (42) extended the data base search and expanded this number to 90%. Since NLS mutant LEDGF/p75 displayed the wild type distribution in the few cells that supported nuclear import, we speculate that the LEDGF/p75 NLS may not overlap with the protein's DNA binding region(s).
Since transference of the LEDGF/p75 NLS onto the otherwise cytoplasmic ␤-galactosidase conferred nuclear localization to the E. coli protein (Fig. 5), we conclude that the LEDGF/p75 NLS is a linear epitope and is thus not composed of discontinuous epitopes coming together in the tertiary structure of the protein. Indeed, no specific secondary structure was predicted between residues 147 and 170 by various computer algorithms (data not shown).
LEDGF/p75 Makes Use of the imp-␣/␤ Nuclear Import Pathway-Using digitonin-permeabilized HeLa cells, we showed that LEDGF/p75 was transported into the nucleus in an active, energy-dependent fashion. Indeed, nuclear accumulation of LEDGF/p75 was efficiently blocked by WGA and did not occur at 4°C or in the absence of ATP (Fig. 6). Moreover, the NLSdefective K150A mutant failed to be imported in the in vitro system, suggesting that import in semipermeabilized cells and live cells occurred through functionally similar pathways. Since the K150A and wild-type LEDGF/p75 proteins displayed similar purification profiles, our results exclude the possibility that a potential bacterial contaminant affected the observed in vitro nuclear localization profiles.
LEDGF/p75 nuclear import was inhibited when mutant RanQ69L was added to the import mixture or when GTP was replaced by GTP␥S, which implies the function of an imp-␤ like transport receptor (18,19,51). Using a nuclear import reconstitution assay, we demonstrated the dependence of LEDGF/ p75 on both imp-␣ and imp-␤ for its nuclear localization (Fig. 8).
Nuclear Import of HIV-1 IN-Lentiviruses, in contrast to oncoretroviruses, can infect nondividing cells (52)(53)(54)(55). The obvious requirement for productive infection of a nondividing cell is the presence of a nuclear targeting signal within the PIC. It was reported that the PIC is imported into the nucleus in an active, energy-dependent manner (56). The mechanism of lentiviral PIC nuclear import, however, has not been established. Viral proteins Vpr, matrix, and IN as well as the central DNA flap structure have been implicated in HIV nuclear import (24,28,(57)(58)(59)(60)(61)(62). Subsequent research revealed that matrix lacks an NLS and is not strictly required for infection of macrophages (63,64). Moreover, Vpr was shown not to be essential for PIC import and virus replication in nondividing cells (63,65,67). In addition, the central DNA flap appears to be host cell-and viral strain-dependent (68,69).  (59) supported these observations by showing that each part of the putative bipartite NLS did not function in a transferable fashion. It has been suggested that the Y143G mutation in IN might affect the nuclear transport function of the PIC, but no effect on the karyophilic properties of the protein itself was observed (29). Armon-Omer et al. (70) suggested that the 161-173 amino acid stretch in HIV-1 IN comprises a functional NLS, since it displayed some karyophilic properties when fused to bovine serum albumin. However, these data are at odds with previously reported results that this putative NLS was not transferable in live cells (68) and that mutations in this region of IN did not abolish nuclear import of the PIC (68, 71).
Devroe et al. (72) demonstrated that fusing IN to a nuclear export signal was not sufficient to abolish nuclear localization of the constitutively expressed HIV-1 protein. This result suggested that HIV-1 IN was trapped in the nucleus, possibly through association with chromatin or direct binding to DNA. Recently, it was reported that HIV-1 IN nuclear accumulation was dependent on the presence of endogenous LEDGF/p75 protein (31). It was suggested that LEDGF/p75 protein residing in the nucleus might capture HIV-1 IN upon its entrance and then tether the viral protein to chromatin (30,31). Alternatively, active nuclear import of cytoplasmic LEDGF/p75-IN complexes was also hypothesized. In the present study, we confirmed that HIV-1 IN requires LEDGF/p75 for its nuclear localization. We showed that a single amino acid change in the NLS motif of LEDGF/p75 was able to relocate HIV-1 IN to the cytoplasm, corroborating the hypothesis that interaction between HIV-1 IN and LEDGF/p75 can occur in the cytoplasm. According to Görlich and co-workers (73), importins act both as import receptors and as chaperones shielding exposed positively charged domains of cargo proteins. Thus, it was not surprising that disrupting interactions with nuclear import receptors (by mutating the LEDGF/p75 NLS) led to cytoplasmic aggregation of LEDGF/p75-IN complexes.
Several groups have tried to identify the nuclear import pathway used by HIV-1 IN and PICs. The role of the imp-␣/␤ pathway in HIV-1 nuclear import was suggested by several groups (28,60,66,74). Recently, Fassati et al. (74) reported that HIV-1 PICs were imported into the nuclei of HeLa cells and macrophages in the presence of imp-7 and Ran and that import was increased in the presence of imp-␤. They also reported that recombinant HIV-1 IN can be imported in semipermeabilized cells selectively by the imp-␣/␤ pathway; however, import could also be supported by the imp-␤/7 heterodimer (74). On the other hand, Depienne et al. (59) reported that HIV-1 IN was imported into the nuclei of semipermeabilized cells using an uncharacterized ATP-dependent but GTP-and Ran-independent mechanism. Strikingly, the addition of cytosolic factors was not required, suggesting that the necessary import receptors remained in those digitoninpermeabilized cells (59).
In our hands, fluorescently labeled recombinant HIV-1 IN readily accumulated in the nuclei of semipermeabilized cells in a temperature-, GTP-, and ATP-dependent fashion (data not shown). We wondered whether LEDGF/p75 would promote the nuclear import of IN in this in vitro system. When LEDGF/p75 was allowed to form complexes with IN prior to the addition to the import mixture, nuclear accumulation of IN was not enhanced. Although the exact stoechiometry of the IN⅐LEDGF/ p75 complex was not reported, its size was estimated at around 300 kDa, probably containing a pair of IN tetramers (30). If a couple of imp-␣/␤ heterodimers (ϳ150 kDa) were bound to the IN⅐LEDGF/p75 complex, the size of the cargo-importin complex would exceed 500 kDa. We speculate that import of large IN⅐LEDGF/p75 complexes might require specific isoforms of imp-␣ and/or imp-␤ families, which were not functionally present in our in vitro assay. Alternatively, the structure of the NPC might be altered in the digitonin-permeabilized cell system, effectively precluding the efficient nuclear import of large protein complexes.
Multiple elements of the HIV-1 PIC have been proposed as essential for its nuclear localization, which could potentially work in an additive and/or interdependent fashion. It is not uncommon that proteins or complexes make use of several nuclear import pathways (8,10). This would be beneficial for the virus, since one mutation abolishing an interaction with one nuclear import receptor would still allow the virus to gain the nuclear environment through an alternative pathway.
In conclusion, we mapped the nuclear targeting signal in the LEDGF/p75 protein to the short peptide ( 148 GRKRKAEKQ 156 ) and showed that it functioned similarly to the SV40 large T NLS in a RanGTP-and imp-␣/␤-dependent manner. We demonstrated that one single amino acid change, K150A, in LEDGF/p75 was sufficient to abolish the nuclear localization of the mutant protein and co-expressed HIV-1 IN. However, the specific importin-␣/␤ isoforms involved in nuclear import of LEDGF/p75 and/or in its complex with HIV-1 IN are yet to be identified. Experiments are under way to establish the precise role of this cellular factor in retroviral replication.
Acknowledgments-Fluorescence microscopy was performed at the Confocal Microscopy Core Facility of the Brigham and Women's Hospital (Boston, MA). We thank Dr. Philip G. Allen for technical assistance and helpful discussions.