To investigate the basis for the LEDGF/p75 dependence of HIV-1 integrase (IN) nuclear localization and chromatin association, we used cell lines made stably deficient in endogenous LEDGF/p75 by RNAi to analyze determinants of its location in cells and its ability to interact with IN. Deletion of C-terminal LEDGF/p75 residues 340-417 preserved nuclear and chromatin localization but abolished the interaction with IN and the tethering of IN to chromatin. Transfer of this IN-binding domain (IBD) was sufficient to confer HIV-1 IN interaction to GFP. HRP-2, the only other human protein with an identifiable IBD domain, was found to translocate IN to the nucleus of LEDGF/p75(–) cells. However, in contrast to LEDGF/p75, HRP-2 is not chromatin bound and does not tether IN to chromatin. A single classical nuclear localization signal (NLS) in the LEDGF/p75 N-terminal region (146RRGRKRKAEKQ156) was found by deletion mapping and was shown to be transferable to pyruvate kinase. Four central basic residues in the NLS are critical for its activity. Strikingly, however, stable expression studies with NLS(+/–) and IBD(+/–) mutants revealed that the NLS, although responsible for LEDGF/p75 nuclear import, is dispensable for stable, constitutive nuclear association of LEDGF/p75 and IN. Both wild-type LEDGF/p75 and NLS-mutant LEDGF/p75 remain entirely chromatin associated throughout the cell cycle, and each tethers IN to chromatin. Thus, these experiments reveal stable nuclear sequestration of a transcriptional regulator by chromatin during the nuclear-cytosolic mixing of cell division, which additionally enables stable tethering of IN to chromatin. LEDGF/p75 is a multidomain adaptor protein that interacts with the nuclear import apparatus, lentiviral IN proteins and chromatin by means of an NLS, an IBD and additional chromatin-interacting domains.
Lens-epithelium-derived growth factor/p75 (LEDGF/p75) is a member of the hepatoma-derived growth factor (HDGF) family. The other six members are a smaller splicing variant, LEDGF/p52, as well as HDGF and four HDGF-related proteins (HRPs): HRP-1, HRP-2, HRP-3 and HRP-4 (Dietz et al., 2002; Ganapathy et al., 2003; Shinohara et al., 2002). Alternative names for LEDGF/p75 are PC4 and SFRS1 interacting protein 2 (PSIP2) and transcriptional coactivator p75. LEDGF/p75 was identified initially by its copurification with the transcriptional coactivator PC4 (Ge et al., 1998a). Interactions of LEDGF/p75 with components of the general transcription machinery and the transcription activation domain of VP16 indicate participation in transcriptional regulation (Ge et al., 1998a; Ge et al., 1998b). A role in activation of stress response genes has been suggested (Sharma et al., 2000; Shinohara et al., 2002).
LEDGF/p75 is 530 amino acids in length. A 333 amino acid alternatively spliced product of the same gene, LEDGF/p52 (or PSIP1), has the same N-terminal 325 amino acid acids but a different C-terminus that is only 8 amino acids in length (Ge et al., 1998a; Ge et al., 1998b). Both p75 and p52 are nuclear proteins. LEDGF/p75 is less active as a coactivator than LEDGF/p52, but is much more abundant in cells (Llano et al., 2004a; Nishizawa et al., 2001).
Since the discovery that LEDGF/p75 coprecipitates with human immunodeficiency virus type 1 (HIV-1) integrase (IN) (Cherepanov et al., 2003), LEDGF/p75 has drawn increasing interest from researchers studying HIV (Cherepanov et al., 2004; Llano et al., 2004a; Llano et al., 2004b; Maertens et al., 2003; Maertens et al., 2004). IN is a viral enzyme that catalyzes covalent insertion of the reverse-transcribed viral cDNA into a host chromosome. HIV-1 and feline immunodeficiency virus (FIV) IN proteins are known to localize to cell nuclei when expressed in the absence of other viral components, as do some small IN fusions [e.g. to green fluorescence protein (GFP) (Bouyac-Bertoia et al., 2001; Cherepanov et al., 2000; Depienne et al., 2000; Gallay et al., 1997; Llano et al., 2004a; Petit et al., 2000; Pluymers et al., 1999; Woodward et al., 2003)]. This apparent karyophilia has led to the proposal of candidate nuclear localization signals (NLSs) in IN and attempts to implicate them in viral pre-integration complex (PIC) nuclear translocation (reviewed by Goff, 2001). However, some studies were inconsistent with an autonomous NLS in HIV-1 IN. For example, fusions of HIV-1 IN to larger indicator proteins such as pyruvate kinase and β-galactosidase are retained in the cytoplasm (Devroe et al., 2003; Kukolj et al., 1997; Llano et al., 2004a). Semi-permeabilized cell studies suggested that nuclear import of HIV-1 IN is facilitated by a limiting cellular factor that is not cytosolic (Depienne et al., 2001). An additional feature that remained to be explained was the tight association of IN with chromatin within the nucleus (Pluymers et al., 1999).
Insights were provided into these issues by the findings that LEDGF/p75 co-immunoprecipitates and colocalizes with HIV-1 and FIV IN proteins, and determines their nuclear localization (Cherepanov et al., 2003; Llano et al., 2004a; Maertens et al., 2003). LEDGF/p75 was also found to be present in the functional PICS of HIV-1 and FIV (Llano et al., 2004a) and to protect these IN proteins from the ubiquitin-proteasome (Llano et al., 2004b). RNA interference (RNAi) with stably expressed short-hairpin RNAs (shRNA) caused highly effective stable knockdown of LEDGF/p75 (while preserving p52 expression) and produced a definitive phenotype: chromatin association of HIV-1 and FIV IN proteins was disrupted and they relocated permanently and completely to the cytoplasm of knocked down stable cell lines (Llano et al., 2004a). Nuclear/chromatin localization of IN was completely abrogated, consistent with a role for LEDGF/p75 as a lentiviral IN-to-chromatin tethering factor. Transient knockdown of LEDGF/p75 with siRNA also disrupted chromatin association of a GFP-IN fusion protein, in this case shifting it from the nucleus to a diffuse distribution in both nucleus and cytoplasm (Maertens et al., 2003). The failure of HIV-1 IN karyophilia to be transferred by fusion of IN to some NLS reporter proteins (Devroe et al., 2003; Kukolj et al., 1997; Llano et al., 2004a) suggests that these fusion partners disrupt the IN-LEDGF/p75 interaction. So far, only one non-lentiviral IN, from murine leukemia virus (MLV), has been studied for interaction with LEDGF/p75. MLV IN did not interact with LEDGF/p75 and was cytoplasmic in its presence and absence, suggesting the interaction may be lentivirus specific (Llano et al., 2004a).
Unlike LEDGF/p75, LEDGF/p52 does not interact with HIV-1 or FIV IN (Llano et al., 2004a; Maertens et al., 2003). These data suggested that the 205 amino acid LEDGF/p75 C-terminal domain contains the IN-binding domain (IBD), whereas the N-terminal 325 amino acids are likely to determine nuclear location. NLSs mediate nucleopore transit of proteins that exceed the approximately 60 kDa upper limit for passive diffusion (Christophe et al., 2000; Jans et al., 2000; Moroianu, 1999). Several categories have been described (Christophe et al., 2000; Gorlich and Kutay, 1999). The best characterized – `classical' NLSs (Kalderon et al., 1984; Lanford and Butel, 1984) – are generally composed of a few contiguous basic residues, typically 4-6 lysines or arginines (monopartite signal), or of two such stretches separated by a non-conserved spacer of 10-12 residues (bipartite signal). Several other kinds of NLSs also exist, such as the M9 sequence in the hnRNPA1 protein (Siomi and Dreyfuss, 1995); these are quite heterogeneous and generally non-basic in character (Christophe et al., 2000). Moreover, basic amino acid motifs that conform to the classical consensus can be found in non-nuclear proteins and many short basic sequences that are presumptive candidates for classical NLSs turn out not to have this function, making empirical mapping essential (Christophe et al., 2000; Gorlich and Kutay, 1999).
To determine the basis for the regulation of lentiviral IN trafficking by LEDGF/p75, we systematically analyzed this protein for domains mediating two functions: (1) interaction with HIV-1 IN and (2) nuclear localization. To facilitate the analyses, we engineered stable cell lines by RNAi to lack endogenous LEDGF/p75. We then studied mutant protein phenotypes with respect to the nuclear and chromatin association of LEDGF/p75 and IN. LEDGF/p75 NLS-independent nuclear location of LEDGF/p75 and IN that persists through cell division in stable cell lines is shown. To extend our previous comparative analyses, we cloned and sequenced the feline LEDGF/p75 ortholog, and investigated whether LEDGF/p75 interacts with the IN protein of a second oncoretrovirus, avian leukosis virus (ALV).
Materials and Methods
pHIN encodes a C-terminally Myc-epitope-tagged version of HIV-1 IN and has been previously described (Llano et al., 2004a). pPK-LEDGF encodes a fusion of pyruvate kinase (Myc-epitope-tagged) to the amino terminus of LEDGF/p75; it was constructed from the pcDNA3-based version of pMyc-PK (Siomi and Dreyfuss, 1995). LEDGF/p75 fusions were inserted as NheI/NotI fragments into pMyc-PK. Overlap-extension PCR was used for alanine substitution of residues 149-152 in LEDGF/p75 using outer primers 5′-AAATAAGGAAAAGTATGGC-3′ and 5′-AAGCAAGTTCATCCAAGGCC-3′, and inner primers 5′-GGGGGGCAGCGGCAGCGGCAGAAAACAAGTAGAAACTGAGAGG-3′ and 5′-CTGCCGCTGCCGCTGCCCCCCTTCTGGCAGCTTTTGG-3′. To allow expression in the presence of stable RNAi against endogenous LEDGF/p75, this and other mutations used pLEDGF/p75siMut, a LEDGF/p75 cDNA in which seven synonymous nucleotide changes were introduced at the shRNA target site (Llano et al., 2004a). LEDGF/p75Δ340-417 was constructed by overlap-extension PCR using outer primers 5′-CCCGGGTCGACTCTAGAGGTACC-3′ and 5′-TAGGCACCTATTGGTCTTACTGAC-3′ and inner primers 5′-AACATTGTTTTCTTAACTTCTGGC-3′ and 5′-GTTAAGAAAACAATGTTGTATAAC-3′. To enable stable expression, an internal ribosome entry-site-linked puromycin resistance gene (pac) was inserted immediately downstream of LEDGF/p75 genes. A nuclear GFP protein, NLS-GFP, was constructed by inserting the SV40 T antigen NLS (Lanford and Butel, 1984) in frame at the N-terminus of eGFP using oligonucleotide adaptors. LEDGF/p75 segments were fused to the NLS-GFP C-terminus as insertions between BamHI and XbaI of pNLS-GFP. ALV IN was amplified by PCR from RCASBP (Petropoulos and Hughes, 1991) with primers that incorporate a C-terminal Myc epitope tag as previously described for HIV and FIV IN (Llano et al., 2004a). pFlag-HRP-2 was constructed by PCR amplification of HRP-2 from a human cDNA library (ATCC Mammalian Genome Collection), followed by addition of an N-terminal FLAG epitope tag and insertion between PstI and NotI in pCMV (Llano et al., 2004a) and verification by DNA sequencing of identity with the published sequence of Izumoto and colleagues (Izumoto et al., 1997).
Cell culture, transfections and selection of stable cell lines
293T HEK, QT6, Vero and NIH 3T3 cells from the American Type Tissue Culture Collection were grown in Dulbecco's modified Eagle's medium (Gibco BRL) supplemented with 10% fetal calf serum, penicillin and streptomycin. si1340 cells, abbreviated here as L cells, stably express a highly effective anti-LEDGF/p75 short-hairpin RNA (shRNA) (Llano et al., 2004a). siScram cells, abbreviated here as S cells, express a control shRNA. The derivation of this cell line is described elsewhere (Llano et al., 2004a). Transfections were performed by the calcium phosphate coprecipitation method with a total of 2 μg of DNA per well of a six-well plate or 1μg of DNA per chamber in a two-chamber LabTek II glass chamber slide (Nalge Nunc). Briefly, cells were transfected 24 hours after being plated in 2 ml of medium at 0.45 ×106 cells/well or 1 ml of medium at 0.8 ×106 cells/chamber. After 14-16 hours, the transfection mix was replaced with fresh culture medium. Cells were harvested or used for indirect immunofluorescence 40-48 hours after the transfection mix was added. For LEDGF/p75.NLS– and LEDGF/p75Δ340-417 stable cell lines, 3 ×106 L cells were plated in 75 cm2 flasks and cotransfected the next day with 7 μg of DNA that had been linearized at a restriction site in the prokaryotic backbone. After 14-16 hours, the transfection mix was replaced with fresh culture medium. 48 hours after transfection, cells were selected in 3.0 μg/ml puromycin.
Immunoblotting and laser-scanning confocal immunofluorescence microscopy
For Western blotting, cells were lysed in 150 mM NaCl, 0.5% DOC, 0.1% SDS, 1% NP-40, and 150 mM Tris-HCl pH 8.0 plus a protease inhibitor cocktail (Complete-mini; Boehringer). Proteins (30 μg/lane) were resolved in SDS-10% polyacrylamide gels and transferred to Immobilon P membranes (Millipore). Blocked membranes were incubated overnight at 4°C with anti-Myc mAb (clone 9E10; Covance), anti-alpha-tubulin mAb (clone B-5-1-2; Sigma), anti-LEDGF/p75 monoclonal antibody (BD Transduction Laboratories), or anti-Flag mAb (Sigma) in Tris-buffered saline with 5% nonfat milk plus 0.05% Tween 20. After washing, membranes were incubated with the appropriate horseradish peroxidase-tagged secondary antibody. ECL (Amersham Pharmacia Biotech) detected bound antibodies. Indirect immunofluorescence detection of Myc-epitope-tagged proteins was performed by laser-scanning confocal fluorescence microscopy with a monoclonal anti-Myc epitope antibody (Covance; clone 9E10) or a rabbit anti-Myc antibody (Santa Cruz) when costained with anti-LEDGF/p75 mAb. Cells grown in LabTek II chamber slides were fixed with 4% formaldehyde in PBS for 10 minutes at 37°C, washed with PBS, and then permeabilized with ice-cold methanol for two minutes at room temperature. Fixed cells were blocked in 10% fetal calf serum, 20 mM ammonium chloride, and PBS for 30 minutes at room temperature, then incubated with the appropriate antibodies, followed by Alexa-488- or Texas Red-conjugated goat anti-mouse antibody or goat anti-rabbit antibody (Molecular Probes). Nuclear DNA was stained with DAPI (Molecular Probes). Triton X extraction of cells was done as described by Kannouche et al.; briefly, unfixed cells were extracted in 1% Triton X-100 at 37°C prior to fixation in 4% paraformaldehyde and immunofluorescence, with control cells unexposed to the detergent (Kannouche et al., 2004).
48 hours after transfection, cells were lysed for 10 minutes on ice in 350 μl of modified CSK buffer [10 mM Pipes pH 6.8, 10% w/v sucrose, 1 mM DTT, 1 mM MgCl2 plus an EDTA-free protease inhibitors mixture (Roche Molecular Biochemicals)] supplemented with 0.5% NP-40 and 400 mM NaCl. The cell lysates were then centrifuged at 6000 g for 2 minutes. 20 μl supernatant was used for immunoblotting of the total cellular fraction, whereas the rest was incubated with 6 μg of anti-GFP mAb (BD Transduction Laboratories) for 15 minutes, followed by addition of 30 μl sheep anti-mouse IgG magnetic beads (Dynal) and incubated at 4°C for one hour on a rotating platform. Beads were washed three times in CSK buffer containing 0.5% NP-40 and boiled in Laemmli buffer plus β-mercaptoethanol. Samples were separated by SDS-PAGE and analyzed by western blotting with an anti-Myc mAb and anti-GFP mAb as described above.
Feline LEDGF/p75 ortholog cloning
cDNA was prepared from cat peripheral blood mononuclear cell (PBMC) RNA using ProSTAR High fidelity RT-PCR system (Stratagen). The total cDNA was used in a PCR reaction with human LEDGF/p75-specific primers: 5′-AGCTAGCTCTAGAATGACTCGCGATTTCAAACCTGG-3′ and 5′-ATATGCGGCCGCCTAGTTATCTAGTGTAGAATCC-3′. The product was inserted into pcDNA3 as an Nhe/Not fragment and sequenced.
Characterization of an NLS in LEDGF/p75
Interactions of IN with mutant LEDGF/p75 proteins are optimally studied in the absence of the endogenous protein. Therefore, most studies in the present work were carried out in L cells, which stably express a highly effective anti-LEDGF/p75 short-hairpin RNA (shRNA) and lack detectable endogenous LEDGF/p75 (Llano et al., 2004a). A control line (S cells) expresses a control shRNA. Both L and S cells were derived from 293T cells for analyses of subcellular location (Llano et al., 2004a). To override the stable RNAi for experimental purposes in the present work, a LEDGF/p75 cDNA in which seven synonymous nucleotide changes were introduced at the shRNA target site (pLEDGF/p75siMut) was used for all transfected LEDGF/p75 constructs.
First, all 530 amino acids of LEDGF/p75 were fused to the C-terminus of Myc-epitope-tagged pyruvate kinase (PK), yielding PK-LEDGF/1-530. PK is frequently used as an NLS reporter because it is a large (60 kDa), monomeric, exclusively cytoplasmic protein that exceeds the size limit for diffusion through nucleopores (Bouyac-Bertoia et al., 2001; Fouchier et al., 1997; Llano et al., 2004a; Siomi and Dreyfuss, 1995). We used a flexible linker in all fusions (Ser-Gly4-Ser). PK was exclusively cytoplasmic as expected (data not shown), but PKLEDGF/1-530 localized to the nucleus, which is consistent with the existence of a transferable NLS in LEDGF/p75 (Fig. 1A,i). Portions of LEDGF/p75 were then fused to PK. Fusions of the most N-terminal 200 and 325 amino acids to PK resulted in nuclear localization identical to that of LEDGF/p75 (data not shown). By contrast, fusions of the first 100 amino acids (PK-LEDGF/1-100) or of two C-terminal regions (PKLEDGF/306-530 and PK-LEDGF/326-530) led to exclusively cytoplasmic location (data not shown). PK-LEDGF/1-160 and PK-LEDGF/101-530 were exclusively nuclear, whereas PKLEDGF 1-145 was cytoplasmic (Fig. 1A,ii-iv). Expression of single, full-length proteins of predicted molecular mass was verified by immunoblotting for these and all mutant proteins in this study (Fig. 1B).
From these experiments, amino acid residues required for LEDGF/p75 nuclear localization could be clearly deduced to reside within interval 146RRGRKRKAEKQVETE160, which contains the basic, classical NLS-consistent stretch 146RRGRKRKAEKQ156. To confirm this, only amino acids 146-156 were fused to PK (PK-LEDGF/146-156). This fusion protein was exclusively nuclear, identifying this sequence as a transferable NLS (Fig. 2A,i; immunoblotting shown in Fig. 2B). When arginine residues 146 and 147 in PK-LEDGF/146-156 were substituted with alanine (RR146-47AA), the protein was predominantly nuclear in location, but some cytoplasmic immunolabeling was observed (Fig. 2A,ii). When the four central basic residues (149RKRK152) in PK-LEDGF/146-56 were substituted with alanine, the protein was exclusively cytoplasmic (Fig. 2A,iii). Thus, these four basic residues are critical to the NLS, whereas the upstream two basic amino acids 146RR147 appear to be needed for optimal function.
To test this hypothesis further, we alanine-substituted 149RKRK152 in full-length LEDGF/p75, yielding LEDGF/p75.NLS– (Fig. 3A,B). Wild-type LEDGF/p75 was nuclear as expected (Fig. 3A,ii), but LEDGF/p75.NLS– was cytoplasmic in interphase L cells, confirming the results of the PK fusion experiments (Fig. 3A,iii; Fig. 3A,iv is discussed further below).
LEDGF/p75.NLS– interacts with HIV-1 IN and sequesters it in the cytoplasm in interphase yet tethers IN to chromosomes in dividing cells
We then proceeded to use L cells to test whether LEDGF/p75.NLS– colocalizes with IN (Fig. 4A,B). As expected, IN was cytoplasmic in L cells (Fig. 4A,a-d), but was directed to the nucleus by co-expressed wild-type LEDGF/p75 (Fig. 4A,e-h). By contrast, IN and LEDGF/p75.NLS– were excluded from the nuclei of interphase L cells (Fig. 4A,i-l). In both cases, the transcriptional coactivator and HIV-1 IN precisely colocalized (Fig. 4A,g,k). Moreover, over-expression of IN and LEDGF/p75.NLS– in 293T cells (i.e. in the presence of endogenous LEDGF/p75), also resulted in sequestration of IN in the cytoplasm, yielding localizations in these cells opposite to that seen with wild-type LEDGF/p75 and indistinguishable from those in panels i-k of Fig. 4A (data not shown).
However, we then found that, although this NLS mediates nuclear import, it is not required for tethering to chromatin and is thus dispensable for stable residence in nuclei of proliferating cells. We were first intrigued by the observation that intense LEDGF/p75.NLS–-specific immunolabeling of metaphase and anaphase chromosomes could be found in some transiently transfected L cells that had proceeded into mitosis after transfection (see anaphase cell in Fig. 3A,iv, and mitotic cells in Fig. 4A,m-p). This is the same appearance that wild-type LEDGF/p75 has in mitotic cells (Llano et al., 2004a) (data not shown). However, it was difficult to interpret the finding since the great majority of LEDGF/p75.NLS–-transfected cells showed cytoplasmic location (Fig. 3A,iii; Fig. 4A,j). To pursue the issue with greater clarity, we stably expressed LEDGF/p75.NLS– in L cells (Fig. 5A,B). Remarkably, in all cells of this cell line (which we derived twice to verify the result), LEDGF/p75.NLS– has a wild-type, nuclear location (Fig, 5A) – moreover, it mediates precise colocalization and chromatin tethering of IN – regardless of position in the cell cycle (Fig. 5B). Thus, mutation of the classical NLS prevents nuclear import of this protein when the nucleus is intact, but does not abrogate its ability to bind to nuclear DNA. We infer that other domains in the NLS-mutant LEDGF/p75 protein are able to mediate binding to chromatin and nuclear retention when it is exposed to cytoplasmic proteins after nuclear envelope breakdown.
Identification and characterization of the IBD: residues 340-417 are necessary and sufficient to mediate IN interaction
As noted, p52 does not interact with IN, suggesting that the C-terminal domain of LEDGF/p75 (residues 326-530) contains an IBD. We also noted from BLAST and other sequence alignment tool analyses (data not shown) that the region corresponding to residues 338-417 of human LEDGF/p75 is well conserved in different species, including Xenopus (92% conserved, 85% identical). We therefore used a method of fusing segments of the p75-unique C-terminus (residues 306-530 and 330-417) to the C-terminus of a nuclear-targeted GFP test protein (NLS-GFP). We constructed the latter protein by incorporating the SV40 T antigen NLS (Lanford and Butel, 1984) at the GFP N-terminus. The experimental strategy is that the NLS-GFP moiety in the fusions is predicted to confine them to the nucleus (substituting functionally for the N-terminal NLS of LEDGF/p75), allowing potential interactions with untethered, cytoplasmic HIV-1 IN in L cells to be detected by confocal microscopy. Use of L cells both eliminates confounding effects of endogenous LEDGF/p75 and permits cytoplasmically located IN for testing. We initially analyzed their subcellular locations in the absence of IN (Fig. 6A) and verified expression of full-length fusions by immunoblotting (Fig. 6B). As predicted, NLS-GFP and NLS-GFP/306-530 were nuclear (Fig. 6A). NLS-GFP/330-417 was also nuclear, with discrete intra-nuclear foci apparent (Fig. 6A).
We then cotransfected L cells with plasmids encoding HIV-1 IN and the NLS-GFP fusion proteins (Fig. 7). NLS-GFP had no effect on IN, which remained cytoplasmic (Fig. 7A,a-d). By contrast, the GFP protein with the smaller LEDGF/p75 C-terminal fragment (NLS-GFP/330-417) caused a diagnostic relocation and colocalization with IN. IN was re-directed from the cytoplasm to the nucleus (Fig. 7A,i-k). Co-expression of HIV-1 IN with NLS-GFP/306-530 also caused total relocation/colocalization of the fusion protein, but from the nucleus to the cytoplasm (Fig. 7A,e-h). This intriguing opposite polarity is probably due to masking of the NLS of NLS-GFP/306-530 by the interaction with IN or induction of additional protein-protein interactions by the IN–NLS-GFP/306-530 complex. To confirm the interaction between the putative IBD and IN, lysates were also immunoprecipitated. HIV IN co-immunoprecipitated with NLS-GFP/330-417, but not with NLS-GFP (Fig. 7B).
To test the model, we deleted amino acids 340-417 in LEDGF/p75 and tested its interaction with IN in L cells (Fig. 8). LEDGF/p75Δ340-417 was nuclear, with wild-type intranuclear distribution (Fig. 8A, top row) and displayed the predicted reduction in molecular weight (Fig. 8B). This protein had the same appearance in the presence of endogenous LEDGF/p75 (data not shown). However, in contrast to the wild-type protein, Δ340-417 did not interact with HIV-1 IN or sequester it to the nucleus of L cells, in any stage of the cell cycle (Fig. 8A). Taken together, the experiments indicate residues 340-417 in the p75 unique C-terminal domain mediate interaction with HIV-1 IN. All proteins we test that have this IBD colocalize with HIV-1 IN, whether they are LEDGF/p75 proteins or chimeric GFP test proteins. All proteins that lack the IBD do not colocalize with HIV-1 IN.
Comparative virological aspects of the IN-LEDGF/p75 interaction
We have been interested in whether the LEDGF/p75 interaction with IN is lenti-retrovirus specific. FIV IN also displays LEDGF/p75-dependent nuclear localization (Llano et al., 2004a), whereas murine leukemia virus IN does not. In both L and S cells, MLV IN is stably cytoplasmic (Llano et al., 2004a). By contrast, in the present work, we found that ALV IN is stably nuclear in L cells, S cells and Quail (QT6) cells (data not shown), a result that is consistent with the transferable NLS that exists in this protein (Kukolj et al., 1997). FIV IN subcellular localization is the same in both feline and human cells (data not shown). To extend our comparative data with both HIV-1 and FIV IN, we cloned and sequenced the feline LEDGF/p75 ortholog and compared the domains we have mapped. A cDNA was isolated by RT-PCR from Felis catus PBMC mRNA using human-specific primers, and sequenced (GenBank accession no. AY705213). Human and feline LEDGF/p75 show overall 4% non-identity. The NLS is fully conserved. The IBD is also identical in the two proteins except for a conservative valine to isoleucine change at residue 411. Nine of nineteen total variant residues (and six of the nine total non-conservative changes) cluster in the extreme C-terminus between amino acids 485 and 530.
HRP-2 co-expression translocates HIV-1 IN from the cytoplasm to the nucleus in the LEDGF/p75(–) background
BLAST analyses with LEDGF/p75 and LEDGF/p75 fragments identified only one other human protein with a region similar to the IBD. This is hepatoma-derived growth factor (HDGF)-related protein 2 (HRP-2) (Izumoto et al., 1997). The HDGF gene family, of which HRP-2 is a member, is reviewed in Dietz et al. (Dietz et al., 2002). ClustalW alignment showed that residues 463-540 of human HRP-2 possess 54% identity and 83% similarity to the LEDGF IBD (340-417), suggesting a conserved functional domain. We therefore cloned a human HRP-2 cDNA by PCR from a human cDNA library and expressed the protein in L cells (Fig. 9). Although 670 amino acids in length, HRP-2 displayed a molecular mass of approximately 140 kDa (Fig. 9B), which is larger than the predicted mass of 74 kDa. A similarly sized HRP-2 band was observed by Engelman and colleagues (Cherepanov et al., 2004). We speculate that this size shift is due to post-translational modifications (e.g. phosphorylation at serine-acidic clusters) (Meier and Blobel, 1992) that are prominent in HRP-2. HRP-2 was nuclear in the presence and absence of LEDGF/p75 (i.e. in L cells; Fig. 9A, top row) and 293T cells (data not shown). To determine if HRP-2 interacts with HIV-1 IN and can substitute functionally for the nuclear-translocating function displayed by exogenously expressed LEDGF/p75 in L cells, we co-expressed HIV-1 IN with and without HRP-2 in these cells. Similar to the action of LEDGF/p75 (Fig. 4A,e-h), HRP-2 relocated HIV-IN from the cytoplasm into the nucleus where both proteins colocalized (Fig. 9A). Notably however, multiple attempts to co-immunoprecipitate IN with HRP-2 and vice versa from these cells were unsuccessful (data not shown), a result that strongly contrasts with the abundant coprecipitation we observe under the same conditions for LEDGF/p75 and IN (Llano et al., 2004a).
HRP-2 is not a chromatin-tethering protein
Since LEDGF/p75 is tightly associated with chromatin and this forms the basis for its IN-tethering function, we examined whether HRP-2 also displays this property by expressing HRP-2 in L cells, in the presence and absence of exogenous LEDGF/p75 (Fig. 10). The intranuclear distribution of HRP-2 in interphase cells was more homogeneous than that of LEDGF/p75. In addition, the HRP-2 and LEDGF/p75 did not precisely overlap in interphase nuclei (note the red-yellow variegation in the two such nuclei with both proteins in Fig. 10p). However, the striking difference between the proteins is observed in actively dividing cells. HRP-2 was seen to be unassociated with condensed chromatin in cells in various phases of mitosis [e.g. metaphase (Fig. 10p) or telophase (Fig. 10l)]. Note the complete lack of overlap of LEDGF/p75, which is chromosome bound, and HRP-2, which is distributed away from the chromosomes. The distribution of HRP-2 was the same when LEDGF/p75 was absent (Fig. 10a-h). In addition, Triton X-100 extraction of unfixed cells (Kannouche et al., 2004) removed detectable HRP-2 but did not extract LEDGF/p75 from cell nuclei (data not shown). Finally, co-expression of HIV-1 IN and HRP-2 in L cells showed they colocalized precisely in all cell-cycle phases, with the same distribution seen for HRP-2 alone in Fig. 10; however, when all three proteins were co-expressed in L cells, HRP-2 over-expression did not affect the colocalization of IN, LEDGF/p75 and chromatin (data not shown). These results suggest a dominant role for LEDGF/p75 over HRP-2 for IN localization, which is consistent with a weaker interaction of HRP-2 with IN as discussed above.
Here we address two problems that concern LEDGF/p75 domain structure and illuminate the process by which LEDGF/p75 governs the observed subcellular locations of lentiviral IN proteins. First, we determined that a discrete region in the LEDGF/p75 C-terminus is responsible for IN interaction. Second, in the course of identifying and characterizing a single NLS in the N-terminus that determines the nuclear import of the protein, our data revealed novel processes that result in LEDGF/p75 and HIV-1 IN chromatin association. The experiments provide clear evidence for NLS-independent stable nuclear sequestering of a transcriptional regulator by chromatin during the nuclear-cytosolic mixing of cell division. This process in turn enables sequestration of IN through a LEDGF/p75 tether. Overall, LEDGF/p75 can be considered a multi-domain adaptor protein that interacts with the nuclear import apparatus, lentiviral IN proteins, and chromatin by means of an NLS, an IBD and additional chromatin-interacting domains.
Two complementary lines of evidence identify residues 340-417 in LEDGF/p75 as a functional domain necessary and sufficient for interaction with HIV-1 IN. This segment of the p75-unique C terminus is transferable and acts autonomously within the transferred context. In addition, its deletion abrogates IN interaction, in all phases of the cell cycle. There is complete correlation between the presence of the IBD in a protein and IN interaction as assessed by co-immunoprecipitation and confocal immunofluorescence. The location of this IBD in the C-terminus of LEDGF/p75 is consistent with the prior observation that LEDGF/p52 does not interact with IN proteins and lends further support to the chromatin-tethering model.
After this paper was first submitted, Engelman and colleagues published data that identified the NLS and IBD domains using different methods (Cherepanov et al., 2004). Our data are consistent with their identifications of these motifs, although some differences are apparent, and additional findings about subcellular trafficking of LEDGF/p75 and HRP-2 are presented here. They localized the IBD to residues 347-429 using in vitro pull-downs, and this is in good agreement with our mapping to 330-417 as the minimal domain fragment that caused GFP to interact with HIV-1 IN in colocalization and co-immunoprecipitation assays. In addition, we showed that deletion of this region (340-417) from LEDGF/p75 resulted in loss of IN interaction. Thus, considering both studies together, it is reasonable to conclude that residues 347-417 represent the minimal IBD.
The function of HRP-2 is currently unknown. Like other members of the HDGF family, it contains a PWWP domain in its N-terminus (Izumoto, 1997; Diezt et al., 2002). Within this family, only LEDGF/p75 and HRP-2 contain a basic C-terminus, in which the IBD resides. GST pull-downs showed that the homologous IBD region in HRP-2 (amino acids 470-593) interact with IN, and IN could be co-immunoprecipitated from 293T cells with HA-tagged HRP-2 (Cherepanov et al., 2004). We have also shown here that when IN and HRP-2 are co-expressed in L cells, which have been thoroughly documented to exclude lentiviral integrases from the nucleus (Llano et al., 2004a), HIV-1 IN translocates from the cytoplasm to the nucleus. These data suggest that HIV-1 integrase can utilize both LEDGF/p75 and HRP-2 for nuclear localization. However, a key distinction emerged in our experiments: HRP-2 clearly does not associate with chromatin-like LEDGF/p75 and does not tether IN to chromatin (Fig. 10).
In contrast to Cherepanov et al. (Cherepanov et al., 2004), we were unable to co-immunoprecipitate HRP-2 and IN using a variety of extraction and precipitation conditions, although we are able to reciprocally co-immunoprecipitate IN and LEDGF/p75 easily (Llano et al., 2004a; Llano et al., 2004b). In this regard, considerably less HRP-2 than LEDGF/p75 was coprecipitable by IN in the experiments reported (Cherepanov et al., 2004). Thus, the IN–HRP-2 interaction appears to be considerably weaker than the IN-LEDGF/p75 interaction.
In dividing cells, the nuclear envelope breaks down in early prophase with microtubule-induced tearing of the nuclear lamina, allowing mixing of cytoplasmic and nuclear contents, and it reassembles at the end of mitosis after sister chromatid separation (Beaudouin et al., 2002; Foisner, 2003). The only apparent requirement for stable chromatin association of LEDGF/p75.NLS– is that nuclear contents become accessible during cell division. This aspect was clearly established by the LEDGF/p75.NLS– stable cell line, where cytoplasmic LEDGF/p75.NLS– was never visualized in confocal microscopy. In transiently transfected cells, LEDGF/p75.NLS–, which is identical to the wild-type protein except for the four amino acids changed to alanine, was excluded from the nucleus until the advent of cell division, an exclusion that is not seen with wild-type LEDGF/p75. Thus, four basic amino acids in the identified LEDGF/p75 NLS are required for IN to be imported into an intact nucleus and become chromatin associated.
The LEDGF/p75 NLS conforms well to consensus sequences for classical NLSs of the SV40 T antigen-type (Christophe et al., 2000), and functions as a discrete, transferable, linear element. Maertens et al. also analyzed LEDGF/p75 for NLSs, using transiently transfected GFP-LEDGF/p75 and HcRed1-IN fusion proteins (Maertens et al., 2004). Our data agree with Maertens et al. on the identity of the NLS and in the conclusion that NLS-mutant LEDGF/p75 can colocalize with HIV-1 IN in the cytoplasm. Our NLS experiments were different methodologically in that we used immunofluorescence, allowing us to assess directly the location of LEDGF/p75 that is unfused to any protein and is wild type except for the NLS mutation, and to assess its interactions with unfused IN. Because fluorescent proteins (in particular red fluorescent protein) tend to oligomerize and may aggregate (Rizzo and Piston, 2005), this methodological difference might explain the absence of apparent large cytoplasmic co-aggregates or inclusions of colocalized proteins in our study (compare each in Fig. 4). In addition, we extended our analyses to stable expression, which allowed us to follow cells through multiple generations, revealing that this NLS is totally dispensable for stable nuclear localization if cells are dividing. Confocal immunofluorescence shows LEDGF/p75.NLS– to be always and entirely chromatin associated in all cells of the population. These data form an interesting parallel with the results of Devroe et al., who found that the addition of an NES to integrase caused IN to be cytoplasmic after transient transfection, but nuclear after stable expression (Devroe et al., 2003). The finding that ALV IN is nuclear in the presence and absence of LEDGF/p75 shows that stable nuclear residence of a retroviral IN protein does not necessarily imply LEDGF/p75 interaction. This result with a fourth retroviral IN protein also adds support to our hypothesis that the interaction of IN with LEDGF/p75 is lentivirus specific and might therefore be implicated in aspects of replication distinctive to the lentiviridae.
Because the present data involve expression of HIV-1 integrase protein outside the viral context, they do not yet demonstrate a functional role for LEDGF/p75 in HIV-1 replication. The precise role of the IN-LEDGF/p75 interaction in the viral life cycle remains uncertain. A role in determining the predilection of HIV-1 for integration into active genes (Schroder et al., 2002) merits investigation. Our experiments are consistent with a model in which LEDGF/p75 is a multi-domain adaptor protein that can mediate either nuclear import of IN or chromatin sequestration of IN, both of which would be IBD dependent. Retention of both proteins, with LEDGF/p75 tethering IN, is dependent on the binding of LEDGF/p75 to chromatin. No nuclear import is required for stable chromatin association if cells are cycling.
The ability of both LEDGF/p75 and HRP-2 to function in translocating IN to the nucleus in L cells suggests that these proteins mediate nucleopore transit of IN. Alternatively, IN may undergo nuclear import in nondividing cells independently of LEDGF/p75, and then be subjected to nuclear sequestration through interaction with intra-nuclear LEDGF/p75 or HRP-2. However, the latter model would not seem to explain the exclusion of IN from the nuclei of L cells (Fig. 4A,a-c), unless net active export of IN is occurring in these cells. Thus, whereas it is clear that LEDGF/p75 and IN can interact in the cytoplasm, whether IN-LEDGF/p75 complexes are the principal means for the nucleopore transit of IN cannot be stated definitively based on presently available data. Semi-permeabilized cell assays suggest LEDGF/p75 import occurs through the importin alpha/beta pathway (Maertens et al., 2004), which is plausible since this has been the case for all classical NLSs yet studied (Christophe et al., 2000). However, IN nuclear import in the in vitro assay was apparently not enhanced by LEDGF/p75 (Maertens et al., 2004). Furthermore, Depienne et al., using the same digitonin-permeabilization method, reported that HIV-1 IN nuclear import did not involve the importin alpha/beta pathway (Depienne et al., 2001).
Deletion analyses focused on identifying the chromatin-interacting domains in LEDGF/p75 will be useful. Both LEDGF/p75 and HRP-2 contain an N-terminal PWWP domain upstream of the NLS. In other proteins, PWWP domain involvement in protein-protein interactions (Stec et al., 2000) and chromatin association (Ge et al., 2004) has been suggested. Downstream of the NLS, AT-Hook motifs are present in both LEDGF/p75 and HRP-2. Proteins containing AT hooks bind AT-rich DNA and are thought to coregulate transcription by modifying the architecture of DNA to enhance the accessibility of promoters to transcription factors (Aravind and Landsman, 1998; Reeves and Nissen, 1990; Siddiqa et al., 2001). Both of these elements are candidates for mediating chromatin interaction of LEDGF/p75, perhaps cooperatively, and might also influence interactions with the transcription machinery.
We thank G. Dreyfuss (University of Pennsylvania, PA) for pyruvate kinase plasmids, Z. Debyser (Rega Institute, Belgium) for a human LEDGF/p75 cDNA, and G. Bren (Mayo Clinic College of Medicine, MN) for pcDNA3.GFP.