[Frontiers in Bioscience 2, d578-587, December 1, 1997]

	[Frontiers in Bioscience 2, d578-587, December 1, 1997] Reprints PubMed CAVEAT LECTOR
	HIV-1 NUCLEAR IMPORT: IN SEARCH OF A LEADER Michael I. Bukrinsky¹ and Omar K. Haffar² 1The Picower Institute for Medical Research, Manhasset, NY 11030, ²Cytokine Networks Inc., 101 Elliott Ave West, Suite 428, Seattle, WA 98119 Received 11/7/97 Accepted 11/14/97 4. HIV-1 NUCLEAR IMPORT 4.1. Role of active nuclear import in HIV-1 life cycle Oncogenic retroviruses enter the nuclear compartment of target cells during mitosis, when the nuclear envelop is resolved (3,11). Although HIV-1 was shown to infect and replicate in non-dividing cells by utilizing the active nuclear import mechanism (4,10,12), it was hypothesized that in dividing cells, such as activated T lymphocytes, entry of the virus into the nucleus occurred during mitosis, and thus did not require active nuclear importation. This concept was based primarily on the ability of import-deficient mutants to replicate in CD4⁺ T cell lines (4,10,14). Nevertheless, published evidence indicated that active nuclear import mechanism can be functional in HIV-1 infection of immortalized T cells, at least under certain conditions. For instance, while HIV-1 does not replicate in quiescent (G₀) T lymphocytes (52,53), it can productively infect T cells arrested in either G₁-S (54) or G2 (55) phases of the cell cycle, suggesting that cell activation, but not cell division, is necessary for virus replication in T cells. Several lines of evidence support the notion that HIV-1 infection in vivo may occur primarily in activated but non-dividing cells. Firstly, CD4⁺ T lymphocytes replicate very slowly in vivo. For instance, naïve T lymphocytes divide once every 3.5 years, while memory T lymphocytes divide once every 22 weeks (56). Secondly, cell cycle analysis performed on peripheral blood mononuclear cells freshly collected from HIV-1 infected individuals detected approximately 98% of cells in G₀-G₁, 2% in S and G₂ combined, and almost no mitotic cells (57). Furthermore, the combined length of the G₁, S, and G₂ phases covers most of the cycle span of a T cell, while mitosis lasts for only 2-3 hr. Finally, evaluation of HIV-1 replication dynamics in vivo revealed rapid turnover rate for free plasma virus (t_1/2≤6 hr), and rapid loss of virus-producing cells (t_1/2 = 1.6 d) (58). High number of infected cells and a rapid dynamics of virus replication combined with a low number of proliferating T cells argue against HIV-1 replication in dividing cells. The ability of HIV-1 to replicate in interphasic CD4⁺ T lymphocytes suggests that the virus may utilize the same active nuclear importation pathway as used during infection of primary macrophages. The importance of this pathway for the virus is further underscored by the ability of virion-packaged Vpr to arrest infected T cells in the G₂ phase (M. Emerman, personal communication). This effect would make active nuclear import crucial for the survival of the virus in an infected target cell. These considerations clearly indicate that while active nuclear import may not be required for HIV-1 infection of immortalized T cell lines or rapidly dividing mitogen-activated primary T cells in vitro, in the environment of chronic infection in vivo where antigen-driven cell activation is slow and tightly regulated, this cellular pathway may be critical for establishment of HIV-1 infection and propagation of the virus. 4.2. Viral proteins that regulate nuclear import 4.2.1. Matrix antigen (MA) Early studies recognized the karyophilic properties of MA (59-63). The MA protein was the first to be identified as a participant from the viral side in the process of HIV-1 nuclear import (4). Its role turned out to be also the most controversial one. The work by Bukrinsky et al. (4) and by Nadler et al. (47) demonstrated that a basic region in the MA protein encompassing amino acids 25-33, G²⁵KKKYKLKH, functions as an NLS when conjugated to BSA. Compared to the NLS of SV40 large T antigen, this MA NLS was a weak one, requiring the presence of multiple peptides per BSA molecule to achieve partial nuclear localization. Another basic region in the C-terminal part of the MA protein, N¹⁰⁹KSKKKA, was found to be an even weaker NLS, although it was still capable of targeting the BSA-NLS conjugate into the nucleus (47). Given such an incomplete effect of peptides corresponding to the MA NLS when compared to the effect of strong NLSs, such as the SV40 large T antigen NLS, it is not surprising that Fouchier et al. (64) interpreted their results as negative when analyzing the nuclear import function of the MA NLS peptide. These authors further analyzed the intracellular localization of MA fused to pyruvate kinase or maltose binding protein, and detected the proteins only in the cytoplasm. Again, this result is in contrast to results of a similar experiment performed by Gallay et al. (13) who detected nuclear localization of the GST-MA fusion protein. Given the weakness of the MA NLS, these differences may reflect the way that the MA NLS is presented in a particular fusion protein, or simply the size of such protein. In any case, the value of these experiments for assessing the role of the MA protein in HIV-1 nuclear import is at least questionable, since there is no evidence that MA functions in the import process as a fusion protein. A more direct analysis of the MA NLS role in HIV-1 nuclear import came from mutagenesis experiments. These studies clearly demonstrated that mutations introduced into the MA NLS substantially diminished HIV-1 replication in non-proliferating cells (2,4,12,64,65). Surprisingly, the replication defect of the MA NLS mutants was observed to some extent in proliferating cells, such as T cell lines or activated PBLs (64,65). Because viral replication in proliferating cells was considered to be independent of nuclear import, these results were interpreted as an evidence for the lack of MA role in HIV-1 nuclear import (64,65). Recent studies, however, demonstrated significance of the active nuclear import process for effective replication of the virus in activated, proliferating T lymphocytes (see a special section of this review on the role of nuclear import in HIV-1 life cycle). In addition, some of those results (64) were obtained with viruses that carry a functional Vpr gene, thus masking the effect of mutations in the MA NLS (see below). Finally, mutagenesis of the MA NLS was usually limited to substitution of threonines for lysines in positions 26 and 27, while earlier analysis (4) clearly demonstrated that replacement of lysines in positions 26, 27, 30, and 32 was required for complete inactivation of the NLS activity. In addition, the second functional NLS identified in the C-terminal part of MA (47) can partially substitute for the defective N-terminal NLS (M.I.B. and O.K.H., unpublished data). Overall, it appears that although HIV-1 MA carries an NLS(s), it is a rather weak one. How then can it target to the nucleus a large macromolecular complex, such as the HIV-1 pre-integration complex? To some extent the weakness of the MA NLS is compensated by the presence of multiple (1,000) copies of MA in the HIV-1 PIC (66). Presence of multiple NLSs has been shown to improve substantially nuclear import (67). In addition, other proteins within the PIC (e.g. integrase, see below) may contribute their NLSs to the process of HIV nuclear import. However, multiplicity of NLSs on the HIV-1 PIC is not sufficient to make it a strong karyophile without involvement of another viral protein, Vpr. This protein regulates interaction between the viral NLSs and karyopherin alpha, thus effectively enhancing the karyophilic potential of the PIC (see below). 4.2.2. Integrase (IN) A role for integrase in HIV-1 nuclear import has been suggested recently by Gallay and co-workers (13). They demonstrated that IN associates with karyopherin alpha and can target a fusion GST-IN protein into the nucleus of microinjected COS cells. This result contradicts a previously published report (68) in which no karyophilic activity of IN-beta-galactosidase fusion protein was identified. Although the explanation for this disparity suggested by Gallay et al. (68), namely, that the configuration of a particular fusion construct may influence the availability of the NLS, is quite credible, it appears that the inherent weakness of the IN NLS may be another important factor influencing the outcome of these experiments. The contribution of IN to HIV-1 nuclear import is even harder to evaluate on the basis of published results. One of the major problems is incomplete inactivation of the MA NLSs in mutants which are considered to be MA NLS-defective. Indeed, as discussed above, inactivation of Lys²⁶ and Lys²⁷ in the MA NLS is not sufficient to destroy its karyophilic activity, and nuclear import observed after infection with HIV-1 carrying these mutations may be driven by the residual NLS activity of MA. In addition, the multiplicity of infection plays an important role in the outcome of experiments on nuclear import, as demonstrated in a response by Trono and Gallay to a letter by Freed et al. (69). Therefore, experiments with pseudotyped constructs (which carry selected HIV-1 determinants) carrying envelopes of MLV or VSV (70-72) are difficult to interpret given a different route (in case of VSV G protein pseudotyped constructs) and undetermined multiplicity of infection. The only convincing evidence for the role of IN in the import process provided so far is found in the report by Gallay et al. (68) who demonstrated that mutation in the IN gene combined with mutations in Vpr and the MA NLS eliminates nuclear import of HIV-1 PICs in P4 cells, while import of a virus defective in Vpr and the MA NLS is only partially reduced. However, it remains unclear whether Vpr-like cellular proteins (see below) are participating in the nuclear import of HIV-1 in this cell line. 4.2.3. Viral protein R (Vpr) The first glimpse of the Vpr's role in HIV-1 nuclear import came when it was realized that T cell line-adapted HIV-1 strains used in the initial experiments contained a frame-shift mutation in the vpr gene (2). When strains with a functional vpr were used, the effect of inactivating mutations in the MA NLS on nuclear import was greatly diminished (2,14,73). Vpr rescued replication of an MA NLS mutant in macrophages by providing sufficient, although reduced by about 60-80% compared to wild-type virus, nuclear translocation of viral DNA. Studies performed with the cloned vpr gene demonstrated nuclear localization of Vpr after transfection (74). Also consistent with a nuclear import role of this protein is the finding that Vpr is dispensable for HIV-1 replication in dividing cells, such as transformed T cell lines, while being critically required in non-dividing macrophages (75,76). The Vpr protein does not contain a canonical NLS, but does have a cluster of 6 arginine residues at the carboxyl terminus which could be a candidate NLS. This region was initially reported to be both necessary and sufficient to direct Vpr to the nucleus (74). However, later studies provided convincing evidence that the nuclear targeting determinant is likely to reside in the amino-terminal alpha helical half of the protein (77,78). This portion of the protein is also involved in mediating Vpr interactions with cellular protein(s) (79,80). Mutations in the alpha helix domain of Vpr that abolished protein-protein interactions also affected nuclear localization of Vpr (78,81). It appeared therefore that karyophilic properties of Vpr are mediated by a cellular Vpr-interacting protein. This protein was recently identified in our lab (Popov et al., submitted). Not surprisingly, it turned out to be karyopherin alpha. A previous study by Gallay et al. (2869} failed to identify Vpr-karyopherin alphainteraction and concluded that Vpr is imported by a karyopherin alpha-independent mechanism. The reason for this disparity lies in an unusual mode of interaction between Vpr and karyopherin alpha. Whilebinding of MA to karyopherin alpha is mediated by the NLS of MA, binding of Vpr to alpha does not involve an NLS. Therefore, interaction of Vpr and karyopherin alpha could not be competed with an excess of NLS peptide (Gallay et al., 1996). The binding site of Vpr on karyopherin alpha does not appear to overlap with the NLS or karyopherin beta binding sites of alpha; in fact, karyopherin alpha, karyopherin beta,Vpr, and MA can assemble into a tetramer. As a result of Vpr binding to karyopherin alpha the affinity of interaction between the NLS and alphais increased 5-10 fold. This effect explains the enhancing activity of Vpr on HIV-1 nuclear import. It appears that Vpr regulates the nuclear import of HIV-1 preintegration complexes by binding to karyopherin alphaand increasing its affinity for viral NLSs, including the NLS of MA. This binding interaction may allow the PIC to compete efficiently for karyopherin alphabeta heterodimers in the cytosol, and may facilitate docking and movement of the PIC across the nuclear pore complex. Such an activity of Vpr explains why mutation of the MA NLS had only a modest effect on HIV-1 nuclear import (64,65,73). Indeed, other weak NLSs in the HIV-1 PIC can substitute for the MA NLS in the presence of Vpr. The role of Vpr, therefore, is to make the HIV-1 PIC a strong karyophile by enhancing the interaction of its NLSs with karyopherin alpha. These results implicate Vpr as a key regulator of HIV-1 nuclear import. Even though all published reports (75,76,82,83) agree on the role of Vpr in HIV-1 infection of non-dividing cells, the magnitude of this effect clearly differs between experimental systems. Discrepancies may be explained by differences between cell types used or methods of cell cultivation. Nevertheless, the fact that substantial replication of HIV-1 with a mutation in the vpr gene was observed in macrophages (14,65) and growth-arrested HeLa cells (2), while no nuclear import of such a mutant was detected in an in vitro system (Popov et al., submitted), suggests that a cellular protein expressed in those cells can partially substitute for the function of Vpr. The existence of such proteins is also suggested by a conservation of Vpr-binding site on karyopherin alphafrom different species. Indeed, both human and yeast karyopherin alpha bind Vpr (Popov et al, submitted), despite their only 40-50% similarity (84). It appears likely that a high-level expression of such proteins in certain cells (e.g. neurons) makes them susceptible to transduction by Vpr-defective lentiviral vectors (72).

[Frontiers in Bioscience 2, d578-587, December 1, 1997]
Reprints
PubMed
CAVEAT LECTOR

HIV-1 NUCLEAR IMPORT: IN SEARCH OF A LEADER

Michael I. Bukrinsky¹ and Omar K. Haffar²

1The Picower Institute for Medical Research, Manhasset, NY 11030, ²Cytokine Networks Inc., 101 Elliott Ave West, Suite 428, Seattle, WA 98119

Received 11/7/97 Accepted 11/14/97

4. HIV-1 NUCLEAR IMPORT

4.1. Role of active nuclear import in HIV-1 life cycle

Oncogenic retroviruses enter the nuclear compartment of target cells during mitosis, when the nuclear envelop is resolved (3,11). Although HIV-1 was shown to infect and replicate in non-dividing cells by utilizing the active nuclear import mechanism (4,10,12), it was hypothesized that in dividing cells, such as activated T lymphocytes, entry of the virus into the nucleus occurred during mitosis, and thus did not require active nuclear importation. This concept was based primarily on the ability of import-deficient mutants to replicate in CD4⁺ T cell lines (4,10,14). Nevertheless, published evidence indicated that active nuclear import mechanism can be functional in HIV-1 infection of immortalized T cells, at least under certain conditions. For instance, while HIV-1 does not replicate in quiescent (G₀) T lymphocytes (52,53), it can productively infect T cells arrested in either G₁-S (54) or G2 (55) phases of the cell cycle, suggesting that cell activation, but not cell division, is necessary for virus replication in T cells. Several lines of evidence support the notion that HIV-1 infection in vivo may occur primarily in activated but non-dividing cells. Firstly, CD4⁺ T lymphocytes replicate very slowly in vivo. For instance, naïve T lymphocytes divide once every 3.5 years, while memory T lymphocytes divide once every 22 weeks (56). Secondly, cell cycle analysis performed on peripheral blood mononuclear cells freshly collected from HIV-1 infected individuals detected approximately 98% of cells in G₀-G₁, 2% in S and G₂ combined, and almost no mitotic cells (57). Furthermore, the combined length of the G₁, S, and G₂ phases covers most of the cycle span of a T cell, while mitosis lasts for only 2-3 hr. Finally, evaluation of HIV-1 replication dynamics in vivo revealed rapid turnover rate for free plasma virus (t_1/2≤6 hr), and rapid loss of virus-producing cells (t_1/2 = 1.6 d) (58). High number of infected cells and a rapid dynamics of virus replication combined with a low number of proliferating T cells argue against HIV-1 replication in dividing cells.

The ability of HIV-1 to replicate in interphasic CD4⁺ T lymphocytes suggests that the virus may utilize the same active nuclear importation pathway as used during infection of primary macrophages. The importance of this pathway for the virus is further underscored by the ability of virion-packaged Vpr to arrest infected T cells in the G₂ phase (M. Emerman, personal communication). This effect would make active nuclear import crucial for the survival of the virus in an infected target cell.

These considerations clearly indicate that while active nuclear import may not be required for HIV-1 infection of immortalized T cell lines or rapidly dividing mitogen-activated primary T cells in vitro, in the environment of chronic infection in vivo where antigen-driven cell activation is slow and tightly regulated, this cellular pathway may be critical for establishment of HIV-1 infection and propagation of the virus.

4.2. Viral proteins that regulate nuclear import

4.2.1. Matrix antigen (MA)

Early studies recognized the karyophilic properties of MA (59-63). The MA protein was the first to be identified as a participant from the viral side in the process of HIV-1 nuclear import (4). Its role turned out to be also the most controversial one. The work by Bukrinsky et al. (4) and by Nadler et al. (47) demonstrated that a basic region in the MA protein encompassing amino acids 25-33, G²⁵KKKYKLKH, functions as an NLS when conjugated to BSA. Compared to the NLS of SV40 large T antigen, this MA NLS was a weak one, requiring the presence of multiple peptides per BSA molecule to achieve partial nuclear localization. Another basic region in the C-terminal part of the MA protein, N¹⁰⁹KSKKKA, was found to be an even weaker NLS, although it was still capable of targeting the BSA-NLS conjugate into the nucleus (47). Given such an incomplete effect of peptides corresponding to the MA NLS when compared to the effect of strong NLSs, such as the SV40 large T antigen NLS, it is not surprising that Fouchier et al. (64) interpreted their results as negative when analyzing the nuclear import function of the MA NLS peptide. These authors further analyzed the intracellular localization of MA fused to pyruvate kinase or maltose binding protein, and detected the proteins only in the cytoplasm. Again, this result is in contrast to results of a similar experiment performed by Gallay et al. (13) who detected nuclear localization of the GST-MA fusion protein. Given the weakness of the MA NLS, these differences may reflect the way that the MA NLS is presented in a particular fusion protein, or simply the size of such protein. In any case, the value of these experiments for assessing the role of the MA protein in HIV-1 nuclear import is at least questionable, since there is no evidence that MA functions in the import process as a fusion protein.

A more direct analysis of the MA NLS role in HIV-1 nuclear import came from mutagenesis experiments. These studies clearly demonstrated that mutations introduced into the MA NLS substantially diminished HIV-1 replication in non-proliferating cells (2,4,12,64,65). Surprisingly, the replication defect of the MA NLS mutants was observed to some extent in proliferating cells, such as T cell lines or activated PBLs (64,65). Because viral replication in proliferating cells was considered to be independent of nuclear import, these results were interpreted as an evidence for the lack of MA role in HIV-1 nuclear import (64,65). Recent studies, however, demonstrated significance of the active nuclear import process for effective replication of the virus in activated, proliferating T lymphocytes (see a special section of this review on the role of nuclear import in HIV-1 life cycle). In addition, some of those results (64) were obtained with viruses that carry a functional Vpr gene, thus masking the effect of mutations in the MA NLS (see below). Finally, mutagenesis of the MA NLS was usually limited to substitution of threonines for lysines in positions 26 and 27, while earlier analysis (4) clearly demonstrated that replacement of lysines in positions 26, 27, 30, and 32 was required for complete inactivation of the NLS activity. In addition, the second functional NLS identified in the C-terminal part of MA (47) can partially substitute for the defective N-terminal NLS (M.I.B. and O.K.H., unpublished data).

Overall, it appears that although HIV-1 MA carries an NLS(s), it is a rather weak one. How then can it target to the nucleus a large macromolecular complex, such as the HIV-1 pre-integration complex? To some extent the weakness of the MA NLS is compensated by the presence of multiple (1,000) copies of MA in the HIV-1 PIC (66). Presence of multiple NLSs has been shown to improve substantially nuclear import (67). In addition, other proteins within the PIC (e.g. integrase, see below) may contribute their NLSs to the process of HIV nuclear import. However, multiplicity of NLSs on the HIV-1 PIC is not sufficient to make it a strong karyophile without involvement of another viral protein, Vpr. This protein regulates interaction between the viral NLSs and karyopherin alpha, thus effectively enhancing the karyophilic potential of the PIC (see below).

4.2.2. Integrase (IN)

A role for integrase in HIV-1 nuclear import has been suggested recently by Gallay and co-workers (13). They demonstrated that IN associates with karyopherin alpha and can target a fusion GST-IN protein into the nucleus of microinjected COS cells. This result contradicts a previously published report (68) in which no karyophilic activity of IN-beta-galactosidase fusion protein was identified. Although the explanation for this disparity suggested by Gallay et al. (68), namely, that the configuration of a particular fusion construct may influence the availability of the NLS, is quite credible, it appears that the inherent weakness of the IN NLS may be another important factor influencing the outcome of these experiments.

The contribution of IN to HIV-1 nuclear import is even harder to evaluate on the basis of published results. One of the major problems is incomplete inactivation of the MA NLSs in mutants which are considered to be MA NLS-defective. Indeed, as discussed above, inactivation of Lys²⁶ and Lys²⁷ in the MA NLS is not sufficient to destroy its karyophilic activity, and nuclear import observed after infection with HIV-1 carrying these mutations may be driven by the residual NLS activity of MA. In addition, the multiplicity of infection plays an important role in the outcome of experiments on nuclear import, as demonstrated in a response by Trono and Gallay to a letter by Freed et al. (69). Therefore, experiments with pseudotyped constructs (which carry selected HIV-1 determinants) carrying envelopes of MLV or VSV (70-72) are difficult to interpret given a different route (in case of VSV G protein pseudotyped constructs) and undetermined multiplicity of infection. The only convincing evidence for the role of IN in the import process provided so far is found in the report by Gallay et al. (68) who demonstrated that mutation in the IN gene combined with mutations in Vpr and the MA NLS eliminates nuclear import of HIV-1 PICs in P4 cells, while import of a virus defective in Vpr and the MA NLS is only partially reduced. However, it remains unclear whether Vpr-like cellular proteins (see below) are participating in the nuclear import of HIV-1 in this cell line.

4.2.3. Viral protein R (Vpr)

The first glimpse of the Vpr's role in HIV-1 nuclear import came when it was realized that T cell line-adapted HIV-1 strains used in the initial experiments contained a frame-shift mutation in the vpr gene (2). When strains with a functional vpr were used, the effect of inactivating mutations in the MA NLS on nuclear import was greatly diminished (2,14,73). Vpr rescued replication of an MA NLS mutant in macrophages by providing sufficient, although reduced by about 60-80% compared to wild-type virus, nuclear translocation of viral DNA. Studies performed with the cloned vpr gene demonstrated nuclear localization of Vpr after transfection (74). Also consistent with a nuclear import role of this protein is the finding that Vpr is dispensable for HIV-1 replication in dividing cells, such as transformed T cell lines, while being critically required in non-dividing macrophages (75,76).

The Vpr protein does not contain a canonical NLS, but does have a cluster of 6 arginine residues at the carboxyl terminus which could be a candidate NLS. This region was initially reported to be both necessary and sufficient to direct Vpr to the nucleus (74). However, later studies provided convincing evidence that the nuclear targeting determinant is likely to reside in the amino-terminal alpha helical half of the protein (77,78). This portion of the protein is also involved in mediating Vpr interactions with cellular protein(s) (79,80). Mutations in the alpha helix domain of Vpr that abolished protein-protein interactions also affected nuclear localization of Vpr (78,81). It appeared therefore that karyophilic properties of Vpr are mediated by a cellular Vpr-interacting protein.

This protein was recently identified in our lab (Popov et al., submitted). Not surprisingly, it turned out to be karyopherin alpha. A previous study by Gallay et al. (2869} failed to identify Vpr-karyopherin alphainteraction and concluded that Vpr is imported by a karyopherin alpha-independent mechanism. The reason for this disparity lies in an unusual mode of interaction between Vpr and karyopherin alpha. Whilebinding of MA to karyopherin alpha is mediated by the NLS of MA, binding of Vpr to alpha does not involve an NLS. Therefore, interaction of Vpr and karyopherin alpha could not be competed with an excess of NLS peptide (Gallay et al., 1996). The binding site of Vpr on karyopherin alpha does not appear to overlap with the NLS or karyopherin beta binding sites of alpha; in fact, karyopherin alpha, karyopherin beta,Vpr, and MA can assemble into a tetramer.

As a result of Vpr binding to karyopherin alpha the affinity of interaction between the NLS and alphais increased 5-10 fold. This effect explains the enhancing activity of Vpr on HIV-1 nuclear import. It appears that Vpr regulates the nuclear import of HIV-1 preintegration complexes by binding to karyopherin alphaand increasing its affinity for viral NLSs, including the NLS of MA. This binding interaction may allow the PIC to compete efficiently for karyopherin alphabeta heterodimers in the cytosol, and may facilitate docking and movement of the PIC across the nuclear pore complex. Such an activity of Vpr explains why mutation of the MA NLS had only a modest effect on HIV-1 nuclear import (64,65,73). Indeed, other weak NLSs in the HIV-1 PIC can substitute for the MA NLS in the presence of Vpr. The role of Vpr, therefore, is to make the HIV-1 PIC a strong karyophile by enhancing the interaction of its NLSs with karyopherin alpha. These results implicate Vpr as a key regulator of HIV-1 nuclear import.

Even though all published reports (75,76,82,83) agree on the role of Vpr in HIV-1 infection of non-dividing cells, the magnitude of this effect clearly differs between experimental systems. Discrepancies may be explained by differences between cell types used or methods of cell cultivation. Nevertheless, the fact that substantial replication of HIV-1 with a mutation in the vpr gene was observed in macrophages (14,65) and growth-arrested HeLa cells (2), while no nuclear import of such a mutant was detected in an in vitro system (Popov et al., submitted), suggests that a cellular protein expressed in those cells can partially substitute for the function of Vpr. The existence of such proteins is also suggested by a conservation of Vpr-binding site on karyopherin alphafrom different species. Indeed, both human and yeast karyopherin alpha bind Vpr (Popov et al, submitted), despite their only 40-50% similarity (84). It appears likely that a high-level expression of such proteins in certain cells (e.g. neurons) makes them susceptible to transduction by Vpr-defective lentiviral vectors (72).