The human immunodeficiency virus type 1 integrase protein has karyophilic properties; that is, it localizes to the cell nucleus according to a range of assays. As an essential component of the preintegration complex, it has been suggested that the karyophilic properties of integrase might facilitate transport of the preintegration complex through the nuclear pore complexes of nondividing cells. However, no experiments have satisfactorily identified a nuclear localization signal within integrase. In this work, we investigated the karyophilic properties of integrase in intact cells with hopes of identifying a genuine transferable nuclear localization signal. Our results confirm that integrase tightly binds chromosomal DNA in vivo. However, our analysis determined that large integrase fusion proteins are unable to access the nucleus, indicating that integrase might lack a transferable nuclear localization signal. In addition, we present several lines of evidence to indicate that DNA binding might facilitate integrase nuclear accumulation. Furthermore, our data indicate integrase is degraded in the cytoplasm by a proteasome-dependent process, an event that probably contributes to the apparent nuclear accumulation of integrase. These results provide new insight into human immunodeficiency virus type 1 integrase intracellular dynamics.
In contrast to simpler retroviruses, lentiviruses such as human immunodeficiency virus type 1 (HIV-1) have evolved the unique ability to infect nondividing cells. Thus, the HIV preintegration complex (PIC) probably contains viral proteins with nuclear localization signals (NLSs) that promote transport of PICs through the nuclear pores of nondividing target cells. Of the known PIC constituents, studies have implicated matrix, viral protein R (Vpr) and, more recently, integrase (IN) as factors facilitating PIC nuclear import (Fouchier and Malim, 1999). IN is a logical candidate because it is the essential DNA recombinase of PICs and has karyophilic properties.
Several groups have sought to characterize IN import. Dargemont and colleagues examined IN nuclear import in an in vitro transport assay using digitonin-permeabilized HeLa cells (Depienne et al., 2001). This assay suggested that IN was imported into the nucleus by a rapid, ATP- and temperature-dependent, saturable mechanism. IN import did not require cytosolic factors, was independent of Ran/GTP and was not competed by excess NLS peptides (Depienne et al., 2001). Malim and colleagues reported the identification of a novel NLS within IN that played an essential role in HIV-1 infectivity in both dividing and nondividing cells (Bouyac-Bertoia et al., 2001). However, we subsequently determined that the IN `NLS' mutants failed to mislocalize in transiently transfected cells and that the mutant viruses were primarily replication defective owing to defective IN catalysis (Limón et al., 2002).
Malim and colleagues likewise confirmed that the IN `NLS' mutant viruses were not deficient in PIC nuclear import (Dvorin et al., 2002). Another report determined that mutation of Cys130 to Gly abolished viral infectivity, reduced IN oligomerization and stability, and resulted in diffuse nuclear and cytoplasmic staining by transient transfection (Petit et al., 1999). Although C130G apparently mislocalized IN, Mammano and colleagues attributed this mislocalization to structural perturbations, not the disruption of an NLS (Petit et al., 1999).
Thus, despite several attempts to describe the pathway, it is currently unclear how HIV-1 IN is imported into the nucleus. Several reports have even acknowledged that IN localization could result from passive diffusion and DNA binding (Kukolj et al., 1997; Petit et al., 1999). In this work, we investigated the karyophilic properties of HIV-1 IN in intact cells with the hopes of identifying a genuine transferable NLS.
Materials and Methods
An epitope-tagged version of HIV-1NL4-3 IN was constructed by PCR amplifying codon-optimized IN from pHDM.Hgpm2 (kind gift of R. Mulligan, Harvard Medical School) and inserting it in-frame with the N-terminal FLAG and hemagglutinin (HA) tags of pOZ-N (kind gift of Y. Nakatani, Dana-Farber Cancer Institute). This FLAG/HA-tagged IN open reading frame (hereafter referred to as IN) was subcloned into pMSCV-Puro (Clontech Laboratories, Palo Alto, CA) to generate a vector useful for both transient transfections and packaging murine-based retroviruses to generate stably transduced cell lines. To generate IN-NES, the coding sequence for the nuclear export signal (NES) from cAMP-dependent protein kinase inhibitor (PKI) [L38ALKLAGLDI47 (numbers indicate position within human PKIα)] was attached just upstream of the IN stop codon by PCR. Site-directed mutagenesis was performed with the QuickChange Mutagenesis Kit (Statagene, La Jolla, CA). To generate pGFP-PK-IN, full-length HIV-1 IN was amplified by PCR and inserted in frame with PK from pGFP-PK (Sherman et al., 2001).
Cell culture and immunofluorescence microscopy
For transient transfections, HeLa cells were plated at 20,000 cells per well in Nunc Lab-Tek II chamber slides (Nalge Nunc International, Rochester, NY) and transfected the following day using the Fugene6 Transfection Reagent (Roche Molecular Biochemicals, Indianapolis, IN). 18-24 hours following transfection, IN localization was monitored by indirect immunofluorescence microscopy using a DeltaVision platform, as described (Limón et al., 2002), except that the 12CA5 anti-HA antibody was used as the primary antibody. DNA was visualized with 4′,6-diamidino-2-phenylindole (DAPI), which was present in the Vectashield mounting medium (Vector Laboratories, Burlingame, CA). Retrovirus packaging was performed as previously described (Devroe and Silver, 2002). HeLa cells were infected overnight with the various IN-expressing retrovirus vectors. 24 hours following selection, the cells were replated in media containing 1 μg ml–1 puromycin. Following expansion of the various IN-expressing cell populations, IN localization was monitored as described above. For proteasome-inhibition experiments, cells were pulsed with 5 μM MG-132 (Calbiochem-Novabiochem, San Diego, CA), reconstituted in dimethyl sulfoxide (DMSO), for 5 hours. Cells treated with DMSO were used as a control. For quantitative data analysis, images were analysed with MetaMorph version 4.6r5 software (Universal Imaging Corporation, Downingtown, PA) using the integrated Measure Colocalization function.
SDS-PAGE and western blotting
Whole cell extracts were prepared by lysing in 50 mM Tris pH 7.5, 300 mM NaCl, 0.5% Triton X-100, containing Complete Protease Inhibitor Cocktail Tablets (Roche Molecular Biochemicals). Protein concentration was determined by BioRad DC assay (BioRad Laboratories, Hercules, CA). An equal quantity of protein was analysed by western blotting using the 12CA5 anti-HA antibody (to detect IN and IN-NES) or an anti-green-fluorescent-protein (GFP) antibody (to detect GFP fusions). In some cases, the anti-FLAG M2 antibody (Sigma, St Louis, MO) was used to detect IN. To prepare nuclear/cytoplasmic extracts, cells were washed twice in PBS, resuspended in two packed cell volumes (PCV) of 10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA pH 8, 1 mM MgCl2, 2 mM DTT, containing Complete Protease Inhibitors Cocktail. After a 15 minute incubation on ice, NP-40 was added to a final concentration of 0.1%, samples were incubated on ice for 2-3 minutes and centrifuged at 8000 g for 10 minutes. The supernatant (`cytoplasm') was saved. The pellet was resuspended in 2 PCV of 20 mM HEPES pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM DTT, plus Complete Protease Inhibitors Cocktail. Nuclear proteins were extracted for 15 minutes on ice with intermittent vortexing, followed by centrifugation at 19,000 g for 20 minutes. The supernatant (`nuclear extract') and cytoplasm were subsequently analysed by western blot analysis as described above. For immunoprecipitation studies, whole cell extracts were prepared in 50 mM HEPES pH 7.5, 500 mM NaCl, 1% Triton X-100, 5 μM MG-132, plus Complete Protease Inhibitors Cocktail. Lysates were precleared with Protein-A/Sepharose (Amersham Biosciences, Uppsula, Sweden) prior to a 1 hour incubation at 4°C with anti-FLAG M2 affinity matrix (Sigma) or rabbit polyclonal anti-ubiquitin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) prebound to Protein-A/Sepharose. Immunopecipitates were washed three times in the immunoprecipitation buffer, followed by two washes in PBS. Bound proteins were eluted from the beads by boiling in 2× SDS-PAGE sample buffer containing 5% β-mercaptoethanol and 50 mM DTT.
HIV-1 integrase accumulates in the nucleus and binds DNA
To examine steady-state localization of IN, we performed indirect immunofluorescence microscopy on HeLa cells transiently or stably expressing an epitope-tagged version of IN. We determined that IN localization (Fig. 1A) closely mirrored DNA staining (Fig. 1B). Furthermore, IN coated the condensed chromosomes of mitotic cells (Fig. 1C-F), as has previously been reported with untagged IN (Cherepanov et al., 2000). Recent experiments indicate that IN is tightly associated with an insoluble nuclear pellet, but can be partially solubilized with buffer containing relatively high concentrations of salt (Cherepanov et al., 2003) (E.D. et al., unpublished) or by DNase I digestion followed by extraction under hypotonic conditions (Cherepanov et al., 2003). These results highlight that HIV-1 IN tightly binds chromosomal DNA in vivo, a phenomenon that might be mediated by the nonspecific DNA-binding properties of the central core (Engelman et al., 1994; Heuer and Brown, 1997; Katzman and Sudol, 1995; Shibagaki and Chow, 1997) and/or C-terminal domains (Engelman et al., 1994; Heuer and Brown, 1997; Lutzke et al., 1994; Vink et al., 1993; Woerner and Marcus-Sekura, 1993).
Several reports have apparently identified mutations within IN that prevent its nuclear accumulation following transient transfection and instead mislocalize the mutant IN to the cytoplasm (Bouyac-Bertoia et al., 2001; Gallay et al., 1997; Petit et al., 2000). However, in our hands, the V165A (Bouyac-Bertoia et al., 2001), K186Q and Q214L/Q216L (Gallay et al., 1997; Petit et al., 2000) mutant proteins behaved essentially identically to wild-type (WT) IN in transient transfections (Limón et al., 2002) (E.D. et al., unpublished). Of note, Malim and colleagues confirmed that IN Val165 does not play a specific role in PIC import (Dvorin et al., 2002). Additionally, Tsurutani et al. (Tsurutani et al., 2000) reported that the K186Q change did not affect the nuclear accumulation of a GFP/HIV-1-IN (GFP-IN) fusion protein. The conflicting reports on the K186Q change suggest that the nature of the IN localization assay might in large part affect the outcome of the experiment. Because PIC nuclear localization is probably essential for HIV-1 replication in nondividing cells and several groups have failed to reproduce results of studies on individual HIV-1 karyophiles (Dvorin et al., 2002; Limón et al., 2002; Tsurutani et al., 2000) (summarized in Fouchier and Malim, 1999), we expanded our analysis of HIV-1 IN nuclear localization and intracellular trafficking.
HIV-1 IN fails to localize a large fusion protein to the nucleus
Although there have been many reports that HIV-1 IN accumulates in the nucleus, it is currently unclear how HIV-1 IN is imported into the nucleus. Because the monomeric form of IN is only ∼32 kDa, IN could theoretically enter the nucleus by passive diffusion. Two different GFP-IN fusion proteins (each ∼60 kDa) also accumulated within the nucleus of transfected cells (Limón et al., 2002; Pluymers et al., 1999; Tsurutani et al., 2000), a finding that nearly rules out passive diffusion because the nuclear pore complex typically excludes proteins larger than 40-60 kDa (reviewed in Mattaj and Englmeier, 1998). However, Skalka and colleagues demonstrated that an IN/β-galactosidase fusion protein (∼148 kDa) was exclusively cytosolic, under conditions where the analogous fusion to avian sarcoma virus IN was nuclear (Kukolj et al., 1997). Although this result suggested that HIV-1 IN might lack an NLS, it is possible that the C-terminal β-galactosidase fusion partner somehow masked a viral NLS. As mentioned above, GFP fused to the N-terminus of IN localized to the nucleus, suggesting that the domain organization of GFP-IN might be important for nuclear import. As a result, we revisited this line of experimentation by fusing HIV-1 IN to the C-terminus of a large (∼82 kDa) cytosolic GFP/pyruvate-kinase fusion protein (GFP-PK) that had previously been used to study the NLS within Vpr (Sherman et al., 2001). The NES-like C-terminal 86 amino acids of PK were removed in this construct (data not shown).
In agreement with previous findings, GFP-PK was exclusively cytosolic (Fig. 2Aa) under conditions where GFP-PK fused to the Vpr NLS (GFP-PK-Vpr73-96, ∼84 kDa) accumulated within the nucleus (Fig. 2Ab) (pGFP-PK and pGFP-PK-Vpr73-96 kindly provided by W. Greene, University of California, San Francisco). We determined that GFP-PK-IN (∼115 kDa) was almost invariably cytosolic (Fig. 2Ac-e), although rare cells also contained some degree of nuclear staining (Fig. 2Ae,f). The inclusion of two different flexible linkers between the PK and IN domains in GFP-PK-IN yielded identical results (data not shown). Results of western blotting revealed that all three GFP-PK-IN fusions were intact in transfected cells (Fig. 2B and data not shown). Although this observation ruled out the possibility that GFP-PK-IN was cytosolic because of proteolytic removal of a potential NLS within the IN domain, it is possible that proteolytic separation of GFP from PK-IN contributed to the observed nuclear accumulation in a small proportion of cells (Fig. 2Ae,f). Alternatively, infrequent nuclear accumulation might have occurred via transient exposure of intact GFP-PK-IN to cellular DNA during mitosis. In an attempt to test this, live cell imaging was performed on GFP-PK-IN-expressing HeLa cells using a Nikon TE2000E microscope equipped with a heated, humidified microscope stage (equipment and advice kindly provided by R. King, Harvard Medical School). The results of this analysis were inconclusive, because only a few transiently transfected cells entered mitosis and those that did invariably died by apoptosis during the course of the experiment (data not shown).
Green and colleagues discovered that GFP-PK fused to full-length Vpr (GFP-PK-Vpr) was cytoplasmic but shuttling, because incubation with leptomycin B (a specific inhibitor of nuclear export) resulted in nuclear accumulation of GFP-PK-Vpr (Sherman et al., 2001). However, incubation with leptomycin B failed to cause any nuclear accumulation of GFP-PK-IN, ruling out the possibility that GFP-PK-IN was shuttling in this system (data not shown). Thus, although GFP-IN can accumulate within the nucleus (Limón et al., 2002; Pluymers et al., 1999; Tsurutani et al., 2000), we fail to find any evidence for a transferable IN NLS using the more rigorous assay of 115-kDa fusion proteins.
Localization of IN fused to an NES
Our above analysis highlights that IN accumulates in the cell nucleus and binds chromosomal DNA (Fig. 3A). Because a specific mutation within IN that universally inhibits nuclear accumulation has yet to be described, we turned to a different strategy and attached the NES from the cAMP-dependent protein kinase inhibitor (LALKLAGLDI) (Wen et al., 1995) to the C-terminus of IN (Fig. 3B). We reasoned that, following nuclear entry, IN-NES should be rapidly exported back into the cytoplasm, thus providing a novel system to examine IN transport in intact cells.
Results of indirect immunofluorescence microscopy demonstrated that IN-NES was cytosolic in most transiently transfected cells (Fig. 3Cd,e), with a small proportion of cells exhibiting some degree of nuclear accumulation (Fig. 3Ce). Thus, under these assay conditions, an NES overrode the inherent karyophilic properties of IN. Interestingly, when stable cell populations were generated by transduction, IN-NES now localized to the nucleus (Fig. 3Cf) in a pattern indistinguishable from WT IN (Fig. 3Cc). Western-blot analysis confirmed that the NES peptide was retained in stably expressing cells (Fig. 3D). It is worth emphasizing that the apparent increased amount of IN (and IN-NES) expressed in our stably transduced population was not indicative of higher IN expression on a per-cell basis. Becuase the stably transduced cells were selected in the presence of puromycin, all cells in the population expressed the IN (or IN-NES) protein. By contrast, only a relatively small proportion of cells (∼10-15%) expressed IN after transient transfection (data not shown). Because IN-NES localization differed depending on assay conditions, it was important to consider the main differences between transient transfection and stable transduction (Fig. 3E). For our transient transfection, HeLa cells were processed 18-24 hours after transfection. Thus, of the cells successfully transfected with IN-NES, only a small proportion would have undergone mitosis prior to immunofluorescence microscopy. By contrast, owing to the selection of stable cell lines, many rounds of cell division occurred prior to immunofluorescence microscopy. These experiments suggest that over time, and conceivably as a result of mitoses, the karyophilic properties of IN eventually overpower a strong NES.
Cytoplasmic IN is subject to proteasome-dependent turnover
The dramatic relocalization of IN-NES over time prompted us to examine the dynamics of cytoplasmic IN. The results of this analysis demonstrated that a proteasome-dependent process degrades cytoplasmic IN. After a 5-hour pulse with MG-132, we observed a stabilized population of IN predominantly localized to the cytoplasm (Fig. 4A). We consider it unlikely that the new IN population was initially nuclear but became stabilized upon export. Rather, we propose that the stabilized population represented newly translated IN that had escaped the cellular degradation machinery.
To characterize the mechanism regulating cytoplasmic IN instability, IN was immunoprecipitated under stringent conditions (500 mM NaCl, 1% Triton X-100) from transiently transfected cell lysates following a 5 hour pulse with MG-132. This analysis revealed an accumulation of high-molecular-weight ubiquitin conjugates specific to IN-expressing cells (Fig. 4B, left). Based on this, we conclude that either a population of IN is multiply ubiquitinated or IN associates tightly with other multiply ubiquitinated proteins. To address these two alternatives, the reverse experiment was performed using anti-ubiquitin antibodies for the immunoprecipitation. The recovered IN migrated with a mobility consistent with unmodified protein (Fig. 4B, right). Based on this, it seems likely that IN associates tightly with multiply ubiquitinated cellular protein(s), although the alternative model that low levels of ubiquitinated IN associate with unmodified IN cannot be excluded.
It has previously been shown that HIV-1 IN overexpressed in cells is degraded by the N-end rule (Mulder and Muesing, 2000). As opposed to HIV-1-processed IN, which has a Phe at the N-terminus (Lightfoote et al., 1986), our system was engineered to express IN with an N-terminal Met. Because Met is a `stabilizing' residue (Varshavsky, 1996), an N-end-rule-independent mechanism might be responsible for the turnover of FLAG-HA-IN. However, we cannot rule out the possibility that the N-terminal Met was removed following translation, rendering the protein subject to N-end-rule instability. Regardless of the mechanism, our results highlight the fact that cytoplasmic IN is subject to proteasome-dependent degradation.
Altered subcellular distribution of an IN DNA-binding mutant
Since HIV-1 IN binds chromosomal DNA in vivo (Fig. 1), we considered whether or not DNA binding might mediate IN nuclear accumulation by serving as an IN `sink.' If DNA binding helps to mediate IN nuclear accumulation, then a DNA-binding mutant should affect the pattern of IN staining. Therefore, we examined the localization of K156E/K159E IN, a mutant that has previously been classified as defective for specific (viral) and nonspecific (chromosomal) DNA-binding activity (Jenkins et al., 1997). At steady state, a large proportion of K156E/K159E IN-expressing cells exhibited a significant degree of cytosolic staining (Fig. 5Ad), in contrast to WT IN (Fig. 5Aa). Following incubation with MG-132, the stabilized WT IN was readily detected in the cytoplasm (Fig. 5Ab,c; quantified in Fig. 5B), a finding consistent with our biochemical fractionation (Fig. 4A). Of note, the localization of DMSO-treated K156E/K159E IN mirrored that of WT IN stabilized with MG-132 (Fig. 5A, compare b,c to d; Fig. 5B). Incubation with MG-132 further increased the cytoplasmic signal of K156E/K159E IN (Fig. 5Ae,f,B).
Although K156E/K159E IN was significantly defective for DNA binding in in vitro assays, this function was not completely inactivated (Jenkins et al., 1997). In line with this observation, it is worth noting that the C-terminal domain of IN, which has been implicated in both nonspecific (Engelman et al., 1994; Heuer and Brown, 1997; Lutzke et al., 1994; Vink et al., 1993; Woerner and Marcus-Sekura, 1993) and specific (Gao et al., 2001) DNA binding, is intact in the K156E/K159E mutant. Consistent with this notion, a considerable degree of nuclear accumulation was evident in transiently transfected cells at steady-state (Fig. 5Ad), and some residual nuclear accumulation was observed following MG-132 treatment (Fig. 5Af, arrow). Stably transduced cells expressing K156E/K159E IN exhibited more cytoplasmic staining than WT IN, although significant nuclear staining was also observed (Fig. 5C). These results suggest that residual DNA binding by K156E/K159E IN in vivo, which is perhaps mediated by the intact C-terminal domain, might help to mediate significant nuclear accumulation.
The current work sheds light on several aspects of IN intracellular transport. First, our analysis failed to reveal a transferable NLS within IN, although we cannot formally rule out the possibility that the different GFP-PK-IN fusions studied here somehow masked an NLS within IN. Second, our experiments with IN fused to an NES suggest DNA binding might be sufficient to localize IN to the cell nucleus over time. Third, our experiments with the proteasome inhibitor MG-132 suggest that IN is selectively degraded in the cytoplasm. Finally, we determined that a DNA-binding IN mutant exhibits less nuclear accumulation than IN. Based on these data, we propose the nuclear compartment might act as a `sink' for IN in transiently transfected or stably transduced cells. In summary, IN can access the nuclear compartment, where it binds tightly to chromosomal DNA and is refractory to proteasome-dependent degradation. These findings readily explain the apparent karyophilic properties of IN in the absence of a transferable NLS. Our results emphasize that, if HIV-1 IN does contain a bona fide NLS, the inherent karyophilic properties of IN will certainly confound its identification.
In light of our current data, it is interesting to revisit previously published work concerning IN import. Dargemont and colleagues' in vitro transport assay indicated that import was saturable, did not require cytosol or Ran/GTP and could not be competed by excess NLS peptides (Depienne et al., 2001). In retrospect, this unusual pathway could be partially explained by accumulation within the nucleus because of DNA binding. However, DNA binding alone does not fully explain why the import appeared rapid, or ATP- and temperature-dependent in this assay (Depienne et al., 2001). Aside from the permeabilized cell assay, other groups have reportedly identified mutations that inhibit IN nuclear import in transiently transfected cells (Bouyac-Bertoia et al., 2001; Gallay et al., 1997; Petit et al., 2000), yet these results have been difficult to reproduce (Limón et al., 2002; Tsurutani et al., 2000) (E.D. et al., unpublished). Another report determined that the C130G mutation abolished viral infectivity, reduced IN oligomerization and stability, and resulted in diffuse nuclear and cytoplasmic staining by transient transfection (Petit et al., 1999). In our hands, we observed increased cytoplasmic staining for C130G IN, similar to that of K156E/K159E (data not shown). Because multimerization is important for IN's DNA-binding activity (Lutzke and Plasterk, 1998) and the C130G change reduced oligomerization (Petit et al., 1999), we speculate that altered DNA binding might have impaired the karyophilic properties of C130G. In summary, we are unaware of any mutation(s) within IN that categorically prevents its nuclear accumulation and instead hypothesize that the K156E/K159E and C130G IN mutants partially mislocalize because of altered DNA-binding activities.
It is interesting to notice the somewhat paradoxical differences observed between transient transfections and stable infections. Our transient transfection assays examine localization soon after IN expression in cells that, in general, have not undergone cell division. Transient transfections typically result in high-level gene expression, owing partly to high gene dosage and high rates of transcription. Thus, expression levels are apparently sufficient to visualize even `unstable' protein populations that are subject to proteasomal degradation. We believe that transient transfections are useful for examining `early' events/localization following gene expression. In transiently transfected cells, a significant proportion of IN-NES is cytoplasmic under conditions where WT IN is almost exclusively nuclear. These results highlight the fact that, in the short-term, nuclear accumulation of IN-NES is reduced. In contrast to transient transfections, stable populations are useful for examining sustained, long-term effects of gene expression. Because the infected cell population is stably selected, multiple rounds of cell division occur prior to replating for immunofluorescence microscopy. During this process, IN (or any of the IN derivatives described here) can readily access and bind chromosomal DNA following mitotic nuclear envelope breakdown. In addition, stable cells typically have a much lower gene dosage. These attributes favor the detection of degradation-resistant protein populations. Taken together, it is not surprising that IN-NES is predominantly localized to the cell nucleus in stably infected cell populations. Our data imply that, eventually, the karyophilic properties of IN will overpower the actions of a strong NES.
It is possible to portray our results as evidence that IN plays no role in facilitating PIC nuclear import. However, various mutations within IN reduce the levels of circular forms of HIV-1 DNA in acute infections of dividing target cells (Bouyac-Bertoia, 2001; Engelman et al., 1997; Limón et al., 2002; Tsurutani et al., 2000; Wiskerchen and Muesing, 1995) (A. Limón and A.E., unpublished). Because the circular forms of HIV-1 DNA are produced in the nucleus, these results suggest that many mutations within IN can reduce the nuclear import of incoming viral PICs. Thus, although the isolated IN protein might lack a transferable NLS, IN probably plays a key structural role in stabilizing or orienting the incoming PIC and/or exposing a novel NLS that might be composed of multiple PIC constituents. Our findings suggest it might be extremely difficult (if not impossible) to dissect the process of PIC nuclear import by studying individual PIC components. Indeed, the `real' karyophilic determinant might be composed of multiple PIC components, or perhaps the entire PIC, a concept that should significantly affect future attempts to uncover the mechanism(s) by which HIV-1 traverses the nuclear envelope to infect nondividing cells.
During the peer review of this manuscript, a paper was published (Maertens et al., 2003) indicating that p75/LEDGF, an HIV-1 IN-interacting protein (Cherepanov et al., 2003), is important for nuclear/chromosomal targeting of HIV-1 integrase in human cells. In p75/LEDGF knock-down cells, HIV-1 IN lost its prominent chromosomal staining pattern and exhibited an essentially diffuse nuclear/cytoplasmic distribution. These results are consistent with our data: they agree with our conclusions that HIV-IN might not have a transferable NLS, that cytoplasmic IN is highly unstable and prone to degradation, and that the karyophilic properties can be attributed to the interaction of IN with a cellular component in the nucleus. Although Maertens et al. (Maertens et al., 2003) were unable to determine whether cytoplasmic p75/LEDGF facilitates IN nuclear import or p75/LEDGF retains intranuclear IN, our observations that large GFP-PK-IN fusions were predominantly cytoplasmic (Fig. 2A) indicate that LEDGF probably interacts with IN after nuclear entry.
Taken together, our findings highlight the fact that the nucleus acts as a sink for HIV-IN, where it tightly associates with chromatin and p75/LEDGF, and is protected from cytoplasmic degradation. It will be important to determine whether p75/LEDGF helps to stabilize integrase and the PIC, acts as an integration cofactor, or directs integrase/PICs to an appropriate chromosomal site for integration.
This research was supported by an HHMI Predoctoral Fellowship (E.D.), NIH grants AI39394 and AI52014 (A.E.), and grants from the NIH and the Claudia Adams Barr Fund (P.S.). We thank W. Green, Y. Nakatani and R. Mulligan for providing plasmids, R. King for microscope use and advice, G. Nolan for providing Phoenix-Ampho cells, and T. Stoyanova for generating plasmids containing the K186Q, Q214L/Q216L and K156E/K159E IN mutations.