Human immunodeficiency virus type‐1 (HIV‐1) is able to infect non‐dividing cells such as tissue macrophages productively because post‐entry viral nucleoprotein complexes are specifically imported into the nucleus in the absence of mitosis. Although it has been proposed that an amino‐terminal region of the viral matrix (MA, p17Gag) protein harbors a basic‐type nuclear localization sequence (NLS) that contributes to this process, utilization of three distinct nuclear import assays failed to provide any direct supporting evidence. Instead, we found that disruption of this region (26KK→TT) reduces the rate at which the viral Gag polyprotein (p55Gag) is post‐translationally processed by the viral protease. Consistent with the fact that appropriate proteolytic processing is essential for efficient viral growth in all cell types, we also show that the 26KK→TT MA mutation is equivalently deleterious to the replication of a primary macrophage‐tropic viral isolate in cultures of non‐dividing and dividing cells. Taken together, these observations suggest that proteins other than MA supply the NLS(s) that enable HIV‐1 to infect non‐dividing cells.
Human immunodeficiency virus type 1 (HIV‐1) and other lentiviruses replicate efficiently in both non‐dividing (post‐mitotic) and dividing cells (Weinberg et al., 1991; Lewis et al., 1992). Importantly, HIV‐1 infection of non‐dividing cell populations in vivo—and in particular of terminally differentiated tissue macrophages and mucosal dendritic cells—appears to be essential for initiating a pathogenic infection and for establishing viral reservoirs that can persist for extended periods (Gartner et al., 1986; Koenig et al., 1986; Wiley et al., 1986; Schuitemaker et al., 1992; Zho et al., 1993; Connor and Ho, 1994; van‘t Wout et al., 1994; Cornelissen et al., 1995; O'Brien and Pomerantz, 1996). The ability of HIV‐1 to infect non‐dividing cells productively contrasts with the finding that replication of the oncogenic retrovirus murine leukemia virus (MLV) is restricted to dividing (proliferating) cells (Roe et al., 1993; Lewis and Emerman, 1994). This key difference between the life cycles of these two retroviruses has been attributed to the ability of the post‐entry viral nucleoprotein complexes (frequently termed pre‐integration complexes or PICs) of HIV‐1 to be actively imported into the nucleus during interphase (Bukrinsky et al., 1992). In contrast, it has been demonstrated that the PICs of MLV only gain access to the nuclear interior following mitotic nuclear membrane dissolution (Roe et al., 1993).
The trafficking of molecules between the nucleus and the cytoplasm occurs through gated channeled structures known as nuclear pore complexes (NPCs) (Feldherr et al., 1984; Dworetzky and Feldherr, 1988). Because the size limit for diffusion through NPCs is ∼50 kDa, molecules or complexes in excess of this require both a specific targeting signal and energy utilization for transport to take place (Hinshaw et al., 1992; Davis, 1995; Görlich and Mattaj, 1996). Indeed, a number of distinct signals, receptor systems and pathways that mediate nuclear import have already been described (Izaurralde and Mattaj, 1992; Dingwall, 1996; Görlich and Mattaj, 1996). With respect to the import of many proteins, specific peptide motifs that serve as autonomous nuclear localization signals (NLSs) have been defined (Dingwall and Laskey, 1991; Garcia‐Bustos et al., 1991; LaCasse and Lefebvre, 1995). Typically (though not always), these sequences comprise one or two stretches of basic amino acids, the best known example is the NLS of simian virus 40 (SV40) T antigen (Figure 1). Interestingly, the import of multi‐component nucleic acid–protein complexes such as the spliceosomal small nuclear ribonucleoprotein particles (U snRNPs) is more complicated. In particular, the nuclear uptake of U snRNPs is determined by a composite NLS that includes the 5′‐trimethylated cap of the RNA as well as the Sm core proteins (Izaurralde and Mattaj, 1992; Fischer et al., 1993).
By analogy with the mechanism of signal‐mediated nuclear import and given that the size of the HIV‐1 PIC resembles that of the ribosome (Farnet and Haseltine, 1990), one can expect that one or more of the components of the PIC will harbor NLSs that function during infection of non‐dividing cells. To date, two HIV‐1 encoded proteins, matrix (MA or p17Gag) and Vpr, have been implicated as containing NLSs important for such infections (Bukrinsky et al., 1993a; Heinzinger et al., 1994; von Schwedler et al., 1994; Gallay et al., 1995a). In the case of MA, a putative basic‐type NLS at residues 25–33 that bears some sequence similarity to the prototypic basic NLS of T antigen has been recognized (Figure 1). It was further shown that viruses carrying a double lysine to threonine substitution in this region (26KK→TT) displayed levels of infectivity and replication that were diminished in non‐dividing target cells but indistinguishable from corresponding wild‐type viruses in dividing cells (Bukrinsky et al., 1993a; Heinzinger et al., 1994; von Schwedler et al., 1994). In contrast, one group has reported that a 26KK→TT MA mutant virus replicates only marginally less well than the wild‐type virus in non‐dividing, as well as dividing, cell populations (Freed et al., 1995).
Importantly, what has been lacking in the majority of these studies has been the direct assignment of NLS function according to conventional criteria. Specifically, sufficiency in terms of conferring nuclear import on heterologous substrates that have relative molecular masses that are greater than ∼50 kDa, and therefore exceed the limit for diffusion through NPCs, has not generally been demonstrated for components of the HIV‐1 PIC. Accordingly, it is conceivable that the aforementioned mutation in the amino‐terminal basic region of MA could perturb a step (or steps) in the HIV‐1 life cycle that occurs prior to nuclear import but that the ultimate consequence is not manifested until later, namely during nuclear uptake itself. For instance, one could envision that the integrity of the MA basic domain might be important for establishing a PIC conformation that exposes a non‐MA‐encoded NLS to cellular nuclear import factors.
We have therefore been utilizing two complementary experimental strategies to dissect the nuclear import of HIV‐1 PICs. First, we have been assigning and defining NLS function—in terms of sufficiency—for the virally encoded components of PICs by using experimental systems previously developed by various laboratories for the analysis of protein nuclear import. Second, we have been disrupting potential NLSs within infectious molecular clones and then determining the ensuing phenotypes not only in terms of overall replication but also with respect to specific aspects of the life cycle that could be affected by that mutation. Regarding this latter point, rather than restricting our analyses to a lymphocyte‐tropic viral isolate, for example HIV‐1HXB (Shaw et al., 1984), we have been focusing our studies on a naturally macrophage‐tropic isolate, HIV‐1YU‐2 (Li et al., 1992). Indeed, HIV‐1YU‐2 is a more logical choice for these experiments since our long‐term goal is to understand viral infection of cells such as macrophages. For reasons that will become apparent, only data that pertain to the nuclear import capabilities and function of MA are discussed in this manuscript. In summary, we report not only that MA does not appear to possess an NLS, but also that disruption of the previously proposed basic‐type NLS at residues 25–33 both inhibits virus replication equivalently in all cell types and reduces the rate at which the Gag polyprotein is processed.
Evidence that the HIV‐1 MA protein does not harbor an NLS
Here, we have utilized three different experimental approaches to evaluate NLS function in the HIV‐1 MA protein. Importantly, each methodology has previously been used to demonstrate that a given peptide NLS can be sufficient for nuclear import. In each case, we have appended the ‘test’ protein (or peptide) to a heterologous substrate protein such that the final molecule is too large to diffuse (in either direction) through NPCs. According to this strategy, the failure of a given molecule to accumulate in the nucleus following its expression in (or introduction into) the cytoplasm is taken as evidence against that particular test sequence harboring a functional NLS.
For the first set of experiments, transfection‐mediated expression of carboxy‐terminal fusions to the 55 kDa cytoplasmic protein pyruvate kinase (PK) was used to examine the nuclear import of MA in somatic cells (Figure 2). The entire MA coding regions (132 amino acids) of both HIV‐1YU‐2 and HIV‐1HXB were therefore appended to PK, as was the 18 amino acid bipartite basic NLS of the human heterogeneous nuclear ribonucleoprotein particle (hnRNP) K protein which served as a positive control. Following transient transfection of HeLa cell monolayers, the subcellular localizations of the various chimeric proteins were determined by indirect immunofluorescence using a monoclonal antibody that recognizes an introduced epitope tag at the amino‐terminus of PK. As anticipated, PK as a non‐fusion protein was confined exclusively to the cytoplasm (Figure 2, panel A) whereas PK‐NLS was efficiently imported into the nucleus (panel D). In sharp contrast, and unexpectedly, both PK‐MA fusion proteins were localized entirely to the cytoplasm (Figure 2, panels B and C). To investigate the possibility that the amino‐terminal basic domains (residues 25–33) of MA were somehow masked in these fusions, the previously defined K protein NLS (Michael et al., 1995) was inserted between the PK and MA sequences. Both resulting PK–NLS–MA fusions were targeted to the nucleus (Figure 2, panels E and F), thus arguing against NLS occlusion being the reason for the cytoplasmic localization of the PK–MA fusions. To verify that these observations were not a consequence of placing MA at the carboxy‐terminus of PK, we also appended a myristoylation‐defective version of MA (2G→A) to the amino‐terminus of β‐galactosidase (β‐GAL). The inhibition of myristoylation was critical for this experiment as this amino‐terminal acylation mediates the targeting of Gag to the plasma membrane as a prerequisite to viral assembly (Göttlinger et al., 1989; Wang and Barklis, 1993; Spearman et al., 1994; Zhou et al., 1994). The finding that the ∼120 kDa MA–β‐GAL fusion failed to accumulate in the nucleus further suggested that HIV‐1 MA does not harbor an NLS (data not shown).
As a second approach for NLS definition, the nuclear import potential of MA was next evaluated by microinjecting maltose binding protein (MBP) fusions into Xenopus laevis oocytes. Accordingly, the full‐length MA proteins of HIV‐1YU‐2 and HIV‐1HXB, the 26KK→TT mutant derivative of MAYU‐2, residues 25 to 33 of MAHXB and the NLS of K were each translated in vitro as radiolabeled fusions to the carboxy‐terminus of MBP. All five proteins, in addition to MBP as a non‐fusion, were directly injected into the cytoplasms of stage VI Xenopus oocytes and transport to the nucleus assessed 12 h later by manual dissection into nuclear and cytoplasmic fractions followed by analysis on an SDS–polyacrylamide gel (Figure 3). As an internal positive control, all samples were co‐injected with the nucleoplasmin core domain fused to the NLS of the hnRNP A1 protein (termed NPLC‐M9) (Siomi and Dreyfuss, 1995). As predicted, NPLC‐M9 and MBP‐NLS efficiently accumulated in the nucleus (Figure 3, lanes 1, 3, 5, 7, 9 and 11) whereas MBP did not (lane 10). In complete agreement with the data obtained with PK fusions in somatic cells, the MBP fusions carrying full‐length wild‐type MA sequences both remained in the cytoplasm (Figure 3, lanes 1, 2, 5 and 6); not surprisingly, the 26KK→TT mutated derivative was also excluded from the nucleus (lanes 3 and 4). In this experiment, the question of potential masking effects was addressed by using MBP‐MAHXB(25–33). Once again, this MA basic region‐bearing protein failed to display any detectable nuclear import activity (Figure 3, lanes 7 and 8).
As a third and final strategy for the examination of possible NLS function in MA, synthetic peptides that carried introduced amino‐terminal cysteine residues and encompassed residues 25 to 33 of MAYU‐2, residues 25 to 33 of MAHXB, residues 23 to 34 of MAHXB or residues 23 to 34 of the 26KK→TT derivative of MAHXB were generated. Each was covalently coupled to bovine serum albumin (BSA), a protein of 68 kDa, that had previously been fluorescently labeled with Cy‐3. All four BSA–peptide adducts were directly microinjected into the cytoplasms of living HeLa cells together with FITC‐labeled rabbit IgG; the latter serving as a control for cytoplasmic injection since its relative molecular mass of ∼160 kDa precludes transport across the nuclear envelope (Fantozzi et al., 1994; Meyer et al., 1996). The positive and negative control peptides for this experiment were, respectively, the NLS of wild‐type simian virus 40 (SV40) large T antigen and its 128K→T non‐functional derivative (Kalderon et al., 1984). Nuclear import was then determined after 30 min by comparing the localization of the BSA–peptide conjugate with that of IgG using double‐label immunofluorescence (Figure 4). As expected, the wild‐type T antigen NLS mediated efficient import (Figure 4, panel A) whereas the mutated sequence did not (panel B). Once again, all MA‐derived sequences, whether wild‐type or mutated, failed to display any discernible import ability (Figure 4, panels C, D, E and F) and therefore displayed staining patterns that were essentially indistinguishable from the inactive T antigen control (panel B) and the IgG controls (panels I, J, K and L). As an additional control to verify that the BSA–peptide conjugates that failed to accumulate in the nucleus were not, in fact, imported into the nucleus only to be rapidly exported back to the cytoplasm, some microinjections into nuclei were also performed. In these cases, the conjugates (as well as the IgG) remained confined to the nucleus for at least 60 min (data not shown).
In summary, we have exploited three distinct and complementary experimental approaches to investigate bona fide NLS function in the MA protein of HIV‐1. In all cases, we have been unable to obtain any evidence to support this contention.
Importance of the amino‐terminal basic domain of MA for HIV‐1 replication and infection
In light of the above data and previous observations that HIV‐1 harboring the 26KK→TT MA mutation replicates poorly in non‐dividing cells but as well as wild‐type virus in dividing cells (Bukrinsky et al., 1993a; Heinzinger et al., 1994; von Schwedler et al., 1994), it was clearly important to re‐examine the replication and infection phenotypes of a 26KK→TT MA mutant virus. This exact double‐codon substitution was therefore introduced into an infectious proviral clone of HIV‐1YU‐2 and both mutant and wild‐type virus stocks generated by transient transfection of 293T cells. The abilities of these two viruses to establish and maintain spreading infections were determined in monocyte‐derived macrophages (MDMs, non‐dividing cells) and primary blood lymphocytes (PBLs, dividing cells) derived from human peripheral blood as well as in the immortalized human T cell lines C8166‐CCR5 and CEM‐CCR5 (dividing cells) (Figure 5). Of note, these two T cell lines had first been rendered susceptible to HIV‐1YU‐2 infection by stable expression of the CCR5 chemokine receptor, a co‐receptor that is utilized by this particular virus for entry (Choe et al., 1996). Replication was monitored over time as the expression of soluble p24Gag antigen in the culture supernatants. As can be readily visualized from all six viral challenges, the relationship between the replication profiles of the wild‐type and MA mutant viruses was similar regardless of the proliferative status of the cells. Specifically, the MA mutant virus always replicated somewhat less efficiently than wild‐type virus in the context of these spreading infections; this resulted in approximately one‐tenth as much virus accumulation at most time points. These findings demonstrate that the 26KK→TT MA mutation does not selectively inhibit replication of a primary macrophage tropic HIV‐1 isolate in non‐dividing cells. These results therefore stand in contrast to those obtained by some laboratories (Bukrinsky et al., 1993a; Heinzinger et al., 1994; von Schwedler et al., 1994) but not others (Freed et al., 1995).
To verify that the MA mutant HIV‐1 was significantly less infectious than wild‐type virus in single‐cycle challenges, quantitations of infectivity were performed using T cell targets that were either proliferating or had been cell cycle‐arrested (Figure 6). Here, the challenged cells harbored the chloramphenicol acetyl transferase (CAT) gene under the transcriptional control of the HIV‐1 long terminal repeat (LTR) promoter element. As these cells became infected, the viral trans‐activator Tat that was expressed by newly established proviruses not only activated transcription from those proviruses but also induced transcription from the resident HIV‐CAT cassette. As a result, infection was quantitated as the induced accumulation of CAT within 24 h of viral challenge (Simon and Malim, 1996). In six independent infections of proliferating C8166‐CCR5/HIV‐CAT cells, wild‐type HIV‐1YU‐2 was, on average, 2.5‐fold more infectious than its 26KK→TT MA mutant counterpart. Importantly, this differential in infectivity between this pair of viruses was maintained when the cells were rendered non‐dividing by treatment with 4000 rad γ‐irradiation 48 h prior to viral exposure. Thus, as with spreading infections (Figure 5), one‐step measurements of viral infection demonstrated that the 26KK→TT MA mutation is similarly inhibitory in challenges of dividing and non‐dividing cells.
In an additional experiment, HIV‐1 cores were pseudotyped with the envelope glycoprotein of the rhabdovirus vesicular stomatitis virus (VSV G) such that virus binding and entry occurred independent of gp120/41Env, CD4 and CCR5. Interestingly, disruption of the MA basic region still reduced viral infectivity to a significant degree (Figure 6). This observation indicates, therefore, that the 26KK→TT MA mutation likely affects at least one aspect of HIV‐1 nucleoprotein core function.
The amino‐terminal basic domain of MA is important for HIV‐1 Gag processing
In an attempt to identify a possible biochemical explanation for the difference in infectivity between the cores of wild‐type and 26KK→TT MA mutant viruses, the patterns of Gag expression in virus‐producing cells and budded virions were examined. Total cell lysates of cultures expressing wild‐type and 26KK→TT MA mutant HIV‐1YU‐2 and HIV–1LAI proviruses, as well as lysates of viral particles purified from the matched culture supernatants, were prepared and analyzed by Western blotting using monoclonal antibodies specific for capsid (CA, p24Gag) and MA (Figure 7). Strikingly, the profiles of Gag expression were very different between cultures expressing wild‐type or MA mutant viruses. In particular, whereas the levels of total Gag expression were similar, the abundance of certain proteolytic processing intermediates (∼39 kDa and ∼25 kDa) as well as the mature, fully cleaved Gag proteins (most notably MA itself) were substantially reduced in the 26KK→TT samples (Figure 7, compare lanes 2 and 4 with lanes 1 and 3). In contrast, the virion particle Gag proteins were processed to virtual completion in all wild‐type and MA mutant samples (lanes 5–8). These findings are therefore very revealing as they demonstrate that the 26KK→TT MA mutation reduces the rate at which Gag is post‐translationally processed without affecting the ultimate extent (i.e. the endpoint) of cleavage. We speculate that it is the perturbation of normal Gag processing rates by the 26KK→TT MA mutation that results in the generic diminution of infectivity and replication noted for this virus (Figures 5 and 6).
In this report, we have presented several distinct lines of evidence that address the controversial view that the amino‐terminal region of HIV‐1 MA harbors an NLS that directly mediates the nuclear import of viral PICs and is therefore selectively important for infection of non‐dividing cells. First, we have used three well‐established and characterized nuclear import assays to demonstrate that residues 25–33 of HIV‐1 MA are not, by themselves, able to mediate the nuclear import of heterologous marker proteins. This finding was not only true for the full‐length MA protein (Figures 2 and 3) but was also the case for synthetic peptides encompassing residues 23–34 (Figure 4). According to these criteria, we have concluded that the amino‐terminal basic region of MA cannot be regarded as an NLS in the strictest sense of the definition. In fact, the observation that this region does not function as an autonomous NLS is, in retrospect, not so surprising since hydrophobic amino acids such as tyrosine (residue 29) are not usually located within basic‐type NLSs (refer to Figure 1) (LaCasse and Lefebvre, 1995).
Second, we have utilized the same missense MA mutation (26KK→TT) that has been used by others to evaluate the contribution of this region to HIV‐1 infection and replication. In contrast with previous analyses where T cell line‐adapted viruses (HIV‐1HXB) that had been converted to macrophage tropism by env gene substitutions were used, our experiments utilized a primary macrophage‐tropic isolate, HIV‐1YU‐2. We reproducibly found that the replication phenotype of the MA mutant virus was, relative to the wild‐type virus, similar in cultures of non‐dividing and dividing cells. Specifically, replication of the MA mutant virus was approximately one order of magnitude less efficient in MDMs, PBLs and immortalized T lymphoid lines (Figure 5). Importantly, these observations were further underscored by the demonstration that the MA mutant virus was also less infectious in single‐cycle challenges of dividing and non‐dividing T cell populations (Figure 6). Based on these findings, we have concluded that the defect in replication of a virus carrying the 26KK→TT MA mutation appears to be a general one that is unrelated to the proliferative status of the target cells.
Third, we have identified a clear biochemical defect in 26KK→TT MA mutant viruses that manifests itself during the final stages of the viral life cycle. Specifically, both for the HIV‐1YU‐2 isolate examined here and for the HIV–1LAI isolate previously evaluated by others, we have found that missense mutation of the amino‐terminal basic region of MA reduces the rate at which p55Gag polyprotein precursors are proteolytically processed by the viral protease (PR) (Figure 7). Because Gag processing rates can have significant effects on HIV‐1 infectivity (Pettit et al., 1994), our results strongly suggest that at least one aspect of the decreased replication of 26KK→TT MA mutant viruses is attributable to less efficient processing of p55Gag by PR. The notion that the integrity of this region influences cleavage rates is consistent with both biochemical (Spearman et al., 1994; Zhou et al., 1994) and crystallographic (Hill et al., 1996) studies which have indicated that this basic domain is important for the association of p55Gag with the plasma membrane during viral assembly and, by extrapolation, for the activation of PR through Gag dimerization (Katz and Skalka, 1994). Less efficient membrane binding would therefore result in reduced rates of Gag–Gag interaction, PR activation and (ultimately) Gag polyprotein processing.
A critical question that arises from these studies concerns why other groups have found differential replication phenotypes for MA basic domain mutant viruses in non‐dividing versus dividing cells (Bukrinsky et al., 1993a; Heinzinger et al., 1994; von Schwedler et al., 1994). One possible explanation for these discrepancies might be that isolates such as HIV‐1HXB and HIV‐1LAI that have been selected for highly efficient growth in T cell lines may be more tolerant of the 26KK→TT MA mutation in high‐titer infections of dividing T cells than primary isolates such as HIV‐1YU‐2. Notably, experiments carried out with a highly pathogenic isolate of simian immunodeficiency virus (SIVSM Pbj1.9) appear to be generally consistent with our view that MA does not harbor an NLS; here, disruption of the amino‐terminal basic region of MA failed to yield a virus that was selectively impaired for replication in non‐dividing cells (Fletcher et al., 1996). It is also worth emphasizing that our studies have not addressed whether MA is involved in post‐entry events distinct from nuclear import. In particular, it has been suggested that: (i) serine and tyrosine phosphorylation of MA is required for the release of PICs from the membrane following fusion (Bukrinskaya et al., 1996); and (ii) that tyrosine phosphorylation is required for the association of MA with integrase (IN) and, therefore, for MA incorporation into PICs (Gallay et al., 1995a,b).
The conclusion that the HIV‐1 MA protein does not harbor an NLS that directly facilitates infection of non‐dividing cells implies that other components of the PIC—namely, IN, reverse transcriptase (RT), nucleocapsid (NC), Vpr, viral nucleic acids or cellular factors (Bukrinsky et al., 1993b; Gallay et al., 1995b)—must contribute the NLS(s) that target this viral nucleoprotein complex for nuclear import. Indeed, Vpr may be one such protein since vpr‐deficient viruses have been shown to display reduced replicative capacity in non‐dividing cells (Balliet et al., 1994; Heinzinger et al., 1994; Connor et al., 1995; Freed et al., 1995) and Vpr fusion proteins are imported into the nuclei of both transfected HeLa cells and microinjected Xenopus oocytes (data not shown). However, it is also probable that additional components contribute to PIC import as vpr‐deficient viruses are still able to replicate to a significant degree in macrophages. In this regard, we speculate that the combined contributions of Vpr and other PIC constituents to nuclear transport may be somewhat analogous to the composite NLS that determines the import of U snRNPs (Izaurralde and Mattaj, 1992; Fischer et al., 1993). Identifying and defining all the functional NLSs of HIV‐1 PICs will therefore be critical for improving the understanding of how this virus infects cells of the monocyte/macrophage lineage in vivo.
Materials and methods
A pcDNA1 (Clontech) ‐based eukaryotic expression vector that encodes pyruvate kinase (PK) with a c‐Myc epitope tag (Evan et al., 1985) at its amino‐terminus and a polylinker at its carboxy‐terminus has been described (Michael et al., 1995). An analogous vector in which the coding region of maltose binding protein (MBP) was substituted for PK was also constructed. These two vectors served as the substrates for expressing PK–MA and MBP–MA fusion proteins. In all cases, PCR‐amplified fragments corresponding to the entire coding region of MA, or shorter subfragments, were ligated into the parental vectors as XhoI–XbaI fragments. The HIV‐1 templates for PCR amplification were the pIIIB (Hwang et al., 1991) and pYU‐2 (see below) provirus expression vectors; these express HIV‐1HXB and HIV‐1YU‐2, respectively. For the PK–NLS and PK–NLS–MA vectors, an EcoRI–XhoI fragment that contained the 17 amino acid bipartite basic NLS of the hnRNP K protein (Michael et al., 1995) was introduced at the carboxy‐terminus of PK. Similarly, the same hnRNP K NLS‐containing fragment was appended to the carboxy‐terminus of MBP to create the MBP–NLS vector. The vector that expresses the nucleoplasmin core domain fused to the NLS of hnRNP A1 (NPLC‐M9) has been described (Michael et al., 1995).
The wild‐type HIV‐1YU‐2 provirus expression vector employed for these studies, pYU‐2, was a derivative of the original proviral clone (Li et al., 1992) in which the disrupted ATG initiation codon of the vpu gene had been repaired by site‐directed mutagenesis. The 26KK→TT MA mutation was introduced into pYU‐2 by PCR‐mediated site‐directed mutagenesis and the entire amplified region (BssHII–NsiI; positions 707–1249) sequenced on both strands to ensure that no other mutations had been introduced. These DNA manipulations were performed on two independent occasions; the two MA mutant viruses had identical growth phenotypes (data not shown). The wild‐type HIV‐1LAI and 26KK→TT MA mutant vectors have been described (Heinzinger et al., 1994); importantly, the region of MA that spans residues 25–32 is identical in amino acid sequence for HIV‐1LAI and HIV‐1HXB (refer to Figure 1).
The CCR5 cDNA encoding MLV‐based retroviral vector, LP‐M/CCR5, was constructed using a derivative of LN‐M (Simon et al., 1995) in which the neo gene had been replaced with one that confers resistance to puromycin. The VSV G expression vector, pHIT/G, was derived from pHIT123 (Soneoka et al., 1995) by substitution of the MLV env gene with an EcoRI fragment from pJC119 (Rose and Bergmann, 1983) that carried the gene for the glycoprotein of VSV. The MLV gag, gag‐pol expression vector pHIT60 has been described (Soneoka et al., 1995).
Cells and cell lines
The immortalized human cell lines HeLa, 293T, CEM‐SS and C8166, as well as the modified C8166/HIV‐CAT that contains the chloramphenicol acetyl transferase (CAT) gene under the transcriptional control of the HIV‐1 long terminal repeat (LTR) promoter element, have been described (Meyer and Malim, 1994; Simon et al., 1995; Simon and Malim, 1996). Retrovirus‐mediated gene transfer was used to stably express the CCR5 chemokine receptor in CEM‐SS, C8166 and C8166/HIV‐CAT cells such that they were rendered susceptible to HIV‐1YU‐2 infection. To achieve this, a 100 mm diameter culture of 293T cells was transiently transfected with 10 μg each of LP‐M/CCR5, pHIT/G and pHIT60 using calcium phosphate, cell‐free viral supernatant harvested at 24 h and used to challenge target cultures. Stably transduced cell populations were selected and maintained in complete RPMI 1640 medium supplemented with 10% fetal bovine serum and either 200 (CEM‐SS and C8166) or 500 (C8166/HIV‐CAT) ng/ml puromycin.
Localization of pyruvate kinase fusion proteins by indirect immunofluorescence
35 mm diameter sub‐confluent HeLa cell monolayers were transiently transfected with 5 μg of the indicated c‐Myc epitope tagged expression vectors using calcium phosphate. At 24 h, the cells were fixed using paraformaldehyde, permeabilized and the staining pattern for PK determined by hybridization first with the Myc‐specific monoclonal antibody 9E10 (Evan et al., 1985) and then with a Texas red (TXRD)‐conjugated goat anti‐mouse secondary antibody (Meyer and Malim, 1994). The samples were viewed by epifluorescence using a Nikon microphot‐SA microscope at a magnification of ×400.
Nuclear import of maltose binding protein fusions in Xenopus laevis oocytes
The MBP fusion proteins and NPLC‐M9 were synthesized in vitro and labeled with [35S]methionine using a T7‐reticulocyte lysate‐coupled transcription–translation system (Promega Corp.). The reaction products were examined for integrity by SDS–polyacrylamide gel electrophoresis followed by fluorography (data not shown) and each MBP fusion mixed with NPLC‐M9 and injected directly into the cytoplasms of 10–20 stage VI oocytes. At 12 h, the oocytes were manually dissected into nuclear and cytoplasmic fractions and the soluble proteins analyzed by electrophoresis through a 12% SDS–polyacrylamide gel, fluorography and autoradiography (Fischer et al., 1995).
Nuclear import of bovine serum albumin–peptide conjugates in HeLa cells
Fluorescently tagged BSA–peptide adducts were generated by direct labeling of BSA with Cy‐3 (Amersham Corp.) followed by the covalent coupling of synthetic peptides that each carried an introduced amino‐terminal cysteine using the heterobifunctional cross‐linker sulfo‐succinimidyl 4‐(N‐maleimidomethyl) cyclohexane‐1‐carboxylate (Sulfo‐SMCC; Pierce). Efficient coupling of all peptides (at least 10 peptides per BSA monomer) was confirmed by SDS–polyacrylamide gel electrophoresis and Coomassie blue staining (data not shown). These complexes were concentrated to ∼2 mg/ml in phosphate‐buffered saline and injected into HeLa cell cytoplasms or nuclei together with rabbit IgG that had been directly labeled with fluorescein isothiocyanate (FITC; Molecular Probes) (Meyer et al., 1996). After 30–60 min incubation at 37°C, the samples were fixed and visualized by double‐label immunofluorescence at a magnification of ×400. The peptide sequences were: SV40 T antigen wild‐type (residues 125–135), C‐YPKKKRKVEDP; SV40 T antigen 128K→T mutant (Kalderon et al., 1984), C‐YPKTKRKVEDP; MAYU‐2 wild‐type (25–33), C‐GKKQYRLKH; MAHXB wild‐type (25–33), C‐GKKKYKLKH; MAHXB wild‐type (23–34), C‐PGGKKKYKLKHI; MAHXB 26KK→TT mutant (23–34), C‐PGGTTKYKLKHI. All peptides were synthesized using an Advanced Chemtech MPS 396 synthesizer.
Primary cells, HIV‐1 replication and single‐step infections
Monocyte‐derived macrophages (MDMs) and primary blood lymphocytes (PBLs) were derived from the peripheral blood of healthy volunteer donors following venepuncture. MDMs were purified by gelatin‐coated plastic adherence (Collman et al., 1989) and maintained in 24‐well culture dishes at a density of 4×105 cells per well in DMEM supplemented with 20% fetal bovine serum and 100 U/ml recombinant human granulocyte macrophage colony‐stimulating factor (rhGM‐CSF) for 7 days prior to viral challenge. PBLs were purified using Ficoll–Paque, stimulated with 5 μg/ml phytohemagglutinin (PHA) for 72 h and maintained in RPMI 1640 medium containing 20% fetal bovine serum and 20 U/ml recombinant interleukin‐2 (rIl‐2) at a density of ∼1×106 cells/ml (Simon and Malim, 1996).
Stocks of wild‐type and 26KK→TT MA mutant HIV‐1YU‐2 were generated by transient calcium phosphate‐mediated transfection of 100 mm diameter cultures of 293T. At 24 h, the supernatants were harvested and stored in aliquots at −80°C. 4×105 MDMs, 5×106 PBLs, 0.5×106 C8166‐CCR5 cells or 0.5×106 CEM‐CCR5 cells were challenged with wild‐type or mutant stocks corresponding to 10 ng soluble p24Gag (PBLs of donor number 2 were challenged with 100 ng p24Gag), as determined by enzyme‐linked immunosorbent assay (ELISA). Cultures were maintained by replenishing the culture media at 2‐ to 3–day intervals and virus production measured as the expression of soluble p24Gag in the culture supernatants.
For the one‐step infections of C8166‐CCR5/HIV‐CAT or C8166/HIV‐CAT cells, cultures containing 0.5×106 cells were challenged with stocks of wild‐type or 26KK→TT MA mutant HIV‐1YU‐2 corresponding to 10 ng soluble p24Gag, VSV G pseudotyped stocks corresponding to 10 ng soluble p24Gag that had been derived by co‐transfection of the proviral expression vectors with pHIT/G, or medium alone (mock infection). At 24 h post‐infection, whole‐cell lysates were prepared and the levels of CAT activity determined as described (Simon and Malim, 1996). In experiments where the C8166‐CCR5/HIV‐CAT cells were cell cycle‐arrested, cultures were subjected to 4000 rad γ‐irradiation from a 137Cs source and transferred to fresh medium for 48 h prior to viral challenge. For these cells, this treatment typically resulted in a redistribution of cells from ∼40% (G0/G1), ∼34% (S), ∼26% (G2/M) to ∼20% (G0/G1), ∼4% (S), ∼76% (G2/M) as determined by propidium iodide staining and flow cytometry analysis (data not shown).
Analysis of viral Gag proteins
Total cell lysates and viral particles pelleted from culture supernatants by centrifugation at 26 000 g for 60 min at 4°C were derived from 293T cultures transiently transfected with wild‐type and 26KK→TT MA mutant, HIV‐1LAI or HIV‐1YU‐2 provirus expression vectors. The samples were normalized for the amount of p24Gag in the culture supernatants (1.0 μg/ml, wild‐type HIV‐1LAI; 1.4 μg/ml MA mutant HIV‐1LAI; 1.1 μg/ml wild‐type HIV‐1YU‐2; 0.64 μg/ml MA mutant HIV‐1YU‐2), denatured and the proteins resolved on a 12% SDS–polyacrylamide gel before being electrophoretically transferred to nitrocellulose. The filter was hybridized initially with a cocktail of two mouse monoclonal antibodies specific for CA (p24Gag) and MA and subsequently with a horseradish peroxidase‐conjugated goat anti‐mouse antibody. Bound antibody was visualized by enhanced chemiluminescence followed by autoradiography.
We thank Bob Doms, Gideon Dreyfuss, Beatrice Hahn, Alan and Sue Kingsman, Jack Rose, George Shaw and Mario Stevenson for sharing plasmid clones, Paul Bates, Ron Collman and Bob Doms for many helpful discussions, Mike Cancro for assistance with cell irradiation and Laurie Zimmerman for excellent secretarial support. The original HIV‐1YU‐2 proviral clone was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH. This work was supported by the Howard Hughes Medical Institute and US Public Service grant AI38715 (M.H.M.) from the NIAID. U.F. received support from an AIDS fellowship from the Deutsches Krebsforschungszentrum and a grant from the Deutsche Forschungsgemeinschaft (SFB 286).
- Copyright © 1997 European Molecular Biology Organization