Incorporation of Vpr into Human Immunodeficiency Virus Type 1 Requires a Direct Interaction with the p6 Domain of the p55 Gag Precursor*

The 96-amino acid Vpr protein is the major virion-associated accessory protein of the human immunodeficiency virus type 1 (HIV-1). As Vpr is not part of the p55 Gag polyprotein precursor (Pr55 gag ), its incorporation requires an anchor to associate with the assembling viral particles. Although the molecular mechanism is presently unclear, the C-terminal region of the Pr55 gag corresponding to the p6 domain appears to constitute such an anchor essential for the incorporation of the Vpr protein. In order to clarify the mechanism by which the Vpr accessory protein istrans-incorporated into progeny virion particles, we tested whether HIV-1 Vpr interacted with the Pr55 gag using the yeast two-hybrid system and the maltose-binding protein pull-down assay. The present study provides genetic and biochemical evidence indicating that the Pr55 gag can physically interact with the Vpr protein. Furthermore, point mutations affecting the integrity of the conserved L-X-S-L-F-G motif of p6 gag completely abolish the interaction between Vpr and the Pr55 gag and, as a consequence, prevent Vpr virion incorporation. In contrast to other studies, mutations affecting the integrity of the NCp7 zinc fingers impaired neither Vpr virion incorporation nor the binding between Vpr and the Pr55 gag . Conversely, amino acid substitutions in Vpr demonstrate that an intact N-terminal α-helical structure is essential for the Vpr-Pr55 gag interaction. Vpr and the Pr55 gag demonstrate a strong interaction in vitroas salt concentrations as high as 900 mm could not disrupt the interaction. Finally, the interaction is efficiently competed using anti-Vpr sera. Together, these results strongly suggest that Vprtrans-incorporation into HIV-1 particles requires a direct interaction between its N-terminal region and the C-terminal region of p6 gag . The development of Pr55 gag -Vpr interaction assays may allow the screening of molecules that can prevent the incorporation of the Vpr accessory protein into HIV-1 virions, and thus inhibit its early functions.

Unlike simple retroviruses, HIV-1 1 encodes for regulatory (tat and rev) as well as for accessory (vpr, vpu, vif, and nef) genes, together referred to as the auxiliary genes. The tat and rev regulatory genes have been shown to be absolutely essential for viral replication in vitro (1). Recently, an increasing number of studies demonstrate that mutations affecting the other auxiliary genes, called the accessory proteins, cause significant phenotypic defects in HIV-1 replication, suggesting that these accessory genes may play pivotal roles during in vivo infection and pathogenesis (2).
The Vpr accessory gene product encodes a 14-kDa, 96-amino acid nuclear protein that is expressed late during viral replication in a Rev-dependent manner (3). This protein is highly conserved between HIV-1, HIV-2, and simian immunodeficiency virus (SIV). In addition, HIV-2 and SIVs encode for a protein, Vpx, that has been shown to possess many structural as well as functional similarities with the Vpr protein (4). Functionally, the HIV-1 Vpr protein harbors two main biological activities. First, early during infection of nondividing cells, Vpr is implicated in the nuclear translocation of the preintegration complex (5,6). The mechanism by which Vpr influences the transport of the preintegration complex remains unclear. Although no classical nuclear localization signal have been clearly demonstrated in Vpr, it is likely that Vpr acts through interactions with cellular proteins involved in the nuclear import of macromolecules. In fact, it was recently demonstrated that Vpr could associate with importin-␣ and the nucleoporin Nsp1, and thus possibly play the role of an importin-␤-like protein (7)(8)(9). Consistent with its involvement in the nuclear targeting of the preintegration complex, Vpr was shown to be required for efficient replication in nondividing cells such as monocytes and macrophages (6,10,11). The ability of Vpr to arrest the cell cycle constitutes the second biological activity associated with this protein (12,13). The cytostatic effect of Vpr was shown to result in a specific block in the G 2 phase of the cell cycle, which was correlated with the inactivation of the Cdc2 kinase (14,15). The functional role of Vpr-mediated cell cycle arrest in proliferating and nondividing HIV target cells is still unclear. However, Stewart et al. (16) recently demonstrated that Vpr could also induce apoptosis following cell cycle arrest, suggesting a contribution to CD4 cell depletion during HIV-1 disease. As well, the cell cycle arrest action of Vpr was shown to increase viral expression in dividing T cells as well as in macrophages (11,17,18).
An important feature of the HIV-1 Vpr and the HIV-2/SIV Vpx proteins is that they are selectively incorporated into the virus particles, which indeed suggests an early function for these two proteins during the viral life cycle (19,20). The localization of Vpr and Vpx within virions is still unclear. Immunoelectron microscopic studies suggested that the HIV-1 Vpr protein localized beneath the viral envelope, co-localizing with the Gag p24 core structures (21). However, a more recent analysis by Kewalramani and Emerman (22) placed the Vpx protein within HIV-2 cores. As Vpr and Vpx are not synthesized as part of the Gag or Gag-Pol polyprotein precursors, they must utilize a distinct mechanism in order to be incorporated into virion particles. Lavallée et al. (23) reported that Vpr could be specifically incorporated in trans within virus-like particles originating only from the expression of the Pr55 gag . The Cterminal p6 domain of the Pr55 gag was subsequently demonstrated to be essential for the incorporation of Vpr into virus particles (24,25). Furthermore, the integrity of a very conserved motif in the C-terminal region of the p6 domain, L-X-S-L-F-G, was shown to be critical for Vpr virion incorporation (26,27). Previous studies clearly established that the predicted amphipathic ␣-helical structure located within the N-terminal region of Vpr was important for its packaging into virions (28 -30). Based on the data accumulated so far, the mechanism of Vpr incorporation into virion particles suggests a direct interaction between the p6 domain of the p55 Gag precursor and Vpr. Nonetheless, Vpr was also shown to associate with other domains from the Pr55 gag . The zinc fingers of NCp7 have recently been suggested to be important for the virion incorporation of Vpr (31,32). In addition, evidence suggesting an association between Vpr and the MAp17 has also been obtained (33).
In order to clarify the mechanism by which the HIV-1 Vpr protein is trans-incorporated into virion particles, we used protein-to-protein interaction assays to investigate Vpr-Gag interactions. The present work provides genetic (yeast two-hybrid) and biochemical evidence indicating that the incorporation of Vpr into HIV-1 particles involves a direct association between Vpr and the p55 Gag polyprotein precursor. We demonstrate that the p6 domain is necessary and sufficient for this interaction. The direct binding of Vpr to the p55 precursor may constitute a target for the development of molecules that could prevent Vpr virion incorporation, and thus, Vpr early functions.

EXPERIMENTAL PROCEDURES
Bacterial and Yeast Strains-Manipulations of bacterial strains and of DNAs were performed by standard methods (34) unless otherwise noted. Escherichia coli AG1-competent cells were used for routine DNA manipulations. Yeast strain EGY48 (MATa, trp1, ura3, his3, leu2::plexAop6-leu2) was used as a host strain for all two-hybrid experiments and was obtained from the laboratory of Dr. Roger Brent (Massachusetts General Hospital, Boston, MA).
Construction of Plasmids-To construct the bait plasmids LexA-Vpr, LexA-Vpu, and LexA-Pr55 gag , the vpr and gag genes from the HxBRU provirus plasmid (35) and the vpu gene from the HxBH10 (36) provirus plasmid were amplified by polymerase chain reaction (PCR) using the 5Ј primer 5Ј-GGCCTAAGGACTGGGTACGATCAA-3Ј and the 3Ј primer 5Ј-GACTTTCAGATAACGAATACTA-3Ј for Vpr, the 5Ј primer 5Ј-GAA-GGAGAGGCATCCGTGCGAGAG-3Ј and 3Ј primer 5Ј-GAAGGAGAGG-CATCCGTGCGAGAG-3Ј for Gag, the 5Ј primer 5Ј-GTAGTACATGGG-ATCCAACCTATACA-3Ј and 3Ј primer 5Ј-TCCTTCGGATCCAGTACC-CCATAA-3Ј for Vpu, all containing BamHI sites. The PCR fragments were then cloned in this latter restriction site in translational frame with the codons of the LexA DNA binding domain of the pEG202 vector (37). These latter bait fusion proteins were produced constitutively from pEG202, a 2-m HIS3 ϩ plasmid under the control of the ADH1 promoter and encoding the LexA C-terminal oligomerization region, which contributes to the operator occupancy by LexA derivatives (37). To construct LexA-Vpr mutants, such as LexA-VprE25K, LexA-VprA30F, LexA-VprQ65E, LexA-VprSR79 -80ID, and LexA-VprR80A, the similar Vpr PCR fragments amplified from HxBRU harboring the different Vpr mutants, E25K, A30F, Q65E, SR79 -80ID, and R80A, were fused in frame with the LexA DNA binding domain of the pEG202 vector. The design and construction of these mutants have been described elsewhere (18,29). The Gag BamHI fragment was also cloned in the pET-21c expression vector (Novagen), which was used to produce in vitro labeled Pr55 gag .
The prey plasmids B42-Vpr, B42-Pr55 gag , and B42-Vpu were constructed by digesting the Vpr, Pr55 gag , and Vpu cDNAs from pEG202 with EcoRI and XhoI. These EcoRI-XhoI fragments were placed in pJG4 -5, a 2-m TRP1 plasmid (38), in translational frame with the codons for the simian virus 40 large T nuclear localization signal, the B42 transactivation domain, and the hemagglutinin epitope tag. Because the pJG4 -5 vector is under the control of the GAL1 promoter, the expression of the prey fusion proteins was inducible in yeast grown on minimal medium containing 2% galactose and 1% raffinose (Gal/Raff) but not in yeast grown on 2% glucose (Glc).
To generate Vpr and Vpu expression plasmids, the vpr and vpu genes from the HIV-1 ELI isolate were amplified by PCR using the 5Ј primer 5Ј-AGAGTCGACGAACAAGCCCCAGCAGAC-3Ј and the 3Ј primer 5Ј-GGCCTGCAGTTAGGATCTACTGGATCC-3Ј for Vpr and the 5Ј primer 5Ј-CATGTCGACCAACCTTTAGGGATAATA-3Ј and the 3Ј primer 5Ј-TCTCTGCAGCTACAGGTCATCAATATC-3Ј for Vpu. The PCR products were then directly cloned into the EcoRV restriction site of the pBluescript KSϩ vector (Stratagene). Digestion of pBKSϩ/Vpr and pBKSϩ/Vpu with SalI yielded Vpr and Vpu fragments lacking their initial methionine codon and were placed into the SalI restriction site of the pMAL-c2 vector (New England Biolabs). This placed amino acids 2-96 of Vpr and amino acids 2-81 of Vpu into translational phase with maltose-binding protein (MBP) sequences of the pMAL-c2 vector.
Transformation of Strain with Reporter, Bait, and Prey Plasmids-The selection strain were made by transforming the EGY48 yeast strain with a URA3 lacZ (␤-galactosidase) reporter plasmid and the different HIS3 bait plasmids by the lithium acetate method (34). The yeast selection strain harboring the bait and reporter plasmids were transformed with different prey plasmid DNAs, and tryptophan utilization phenotype was used (in addition to His and Ura markers for bait and LacZ reporter plasmids, respectively) for selection of transformants with the prey plasmids.
Determination of Bait-Prey Interaction-Five independent transformants containing the appropriate bait and prey plasmids were streaked on Glc Ura Ϫ His Ϫ Trp Ϫ medium for amplification. Two days later, transformants were restreaked on plates containing Glc Ura Ϫ His Ϫ Trp Ϫ 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal) medium or Gal/Raff Ura Ϫ His Ϫ Trp Ϫ X-Gal medium to assess transcriptional activation of the lacZ reporter gene.
To investigate Vpr-Gag interaction, we first prepared MBP, MBP-Vpu, and MBP-Vpr-bound amylose resin by incubating equivalent lev-els of MBP, MBP-Vpu, and MBP-Vpr proteins with amylose resin (50% v/v) for 60 min at 4°C. Then, equal amounts of in vitro synthesized [ 35 S]methionine-labeled Pr55 gag (TNT coupled reticulocyte lysates system; Promega) diluted in Column Buffer (20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA) were incubated with MBP, MBP-Vpu, or MBP-Vpr-bound amylose resin for 2 h at 4°C. These complexes were then washed several times with Column Buffer, and bound proteins were eluted with 10 mM maltose, loaded onto a 12.5% SDS-PAGE for autoradiography or Western blot analysis.
Immunoprecipitation and Western Blot-Yeast EGY48 cells expressing B42-Pr55 gag and LexA-Vpr, which was used in yeast two-hybrid assay, were lysed in 500 l of RIPA lysis buffer (41) by beating with glass beads five times for 2 min each. After being removed from beads and cell debris by centrifugation (10,000 ϫ g) at 4°C, the specific proteins in the cell lysates were immunoprecipitated with either a rabbit anti-Vpr serum (23) or a mouse anti-p24 antibody (ATCC HB-9725). Immunoprecipitated proteins were then subjected to SDS-PAGE and subsequently blotted on nitrocellulose (0.45 mm; Schleicher & Schuell). LexA-Vpr and B42-Pr55 gag fusion proteins were identified by the rabbit anti-Vpr serum, or a rabbit anti-p24 antibody (NIH 384), respectively, with highly sensitive ECL chemiluminescence detection system as recommended by the manufacturer (Amersham Pharmacia Biotech).

Genetic Evidence That Vpr and the Pr55 gag Physically
Interact-The experimental system described by Golemis et al. (37) was used to investigate the interaction between Vpr and the Pr55 gag of HIV-1. In order to do so, the Vpr and p55 proteins were fused at the C terminus of the LexA DNA binding domain (bait) and the B42 bacterial transactivator (prey). To monitor the specificity of interactions, the native plasmids containing only the LexA DNA binding domain or the B42 bacterial transactivator domain, or fused with another HIV-1 accessory protein, Vpu, which is unlikely to interact with HIV-1 Gag protein (42), were used as controls of specificity. The host strain (EGY48) contains the LEU2 and the lacZ reporter genes, both carrying LexA operators instead of native upstream regulatory sequences. A EGY48 yeast cell containing a bait (LexA-fusion) plasmid and reporters (LEU2 and lacZ) remains inert for the expression of leucine utilization or ␤-galactosidase activity unless it also contains a prey vector that expresses an interacting protein as a fusion molecule to the B42 acid blob transactivation domain (43). In the above described system (37), the expression of the B42-fusion proteins is conditional on the presence of galactose (Gal/Raff) in the culture medium since the expression is directed by the GAL1 promoter.
Essentially, the yeast strain EGY48 was transformed with different combinations of bait and prey plasmid constructions and selected on Glc Ura Ϫ His Ϫ Trp Ϫ medium. Five independent transformants were then selected and streaked on either 1) Glc Ura Ϫ His Ϫ Trp Ϫ X-Gal medium or 2) Gal/Raff Ura Ϫ His Ϫ Trp Ϫ X-Gal medium. The yeast streaked on Glc Ura Ϫ His Ϫ Trp Ϫ X-Gal medium showed no indication of ␤-galactosidase activity (by the observation of blue colonies), indicating that no protein interaction occurs in the absence of induction of the B42 fusions (data not shown). Fig. 1A shows the results from the different transformation combinations when grown on Gal/Raff Ura Ϫ His Ϫ Trp Ϫ X-Gal medium. As previously shown by other groups (44,45), our result confirmed that the Pr55 gag is capable of homo-oligomerization (lane 3). Because Vpr has been shown to act as a weak transactivator of some cellular promoters as well as the HIV-1 long terminal repeat (46), it was important to verify if Vpr could by itself induce ␤-galactosidase expression when fused to the LexA DNA binding domain. Lane 6 shows that in our yeast two-hybrid system, Vpr does not seem to cause promoter transactivation to levels appreciable for observation. However, when the B42-Pr55 gag fusion was introduced into the FIG. 1. Specificity of interaction between Vpr and the p55 Gag precursor in the yeast two-hybrid system. A, the EGY48 reporter strain containing LexA, LexA-Vpr, or LexA-Pr55 gag was transformed with B42, B42-Vpu, B42-Vpr, or B42-Pr55 gag . The yeast cells (five independent transformants) were then cultured on selective galactose media containing X-Gal and assessed for their ability to interact together. B, a clone shown to be positive for Vpr-Gag interaction was grown in galactose selective media. Yeast cells were lysed and immunoprecipitated with either rabbit anti-Vpr or mouse anti-p24 sera. Samples were then electrophoresed on either a 8.5% (for B42-Pr55 gag ) or 12.5% (for LexA-Vpr) SDS-PAGE, transferred on nitrocellulose, and immunoblotted with either anti-Vpr or anti-p24 antibodies. yeast strain containing the LexA-Vpr fusion, strong ␤-galactosidase expression was detected by the observation of blue colonies (lane 7). This suggests that Vpr and the p55 Gag precursor are directly interacting in vivo in the yeast cell nucleus. Interestingly, the interaction between Vpr and Pr55 gag is observed only in one direction, that is, when Vpr is fused to LexA sequences and the Pr55 gag to the B42 transactivation domain. When the two fusions are switched, the interaction is not observed (lane 4). In addition, evidence of Vpr homo-oligomerization were observed using the yeast two-hybrid system (lane 8), as previously shown by others (44). When B42-Vpu (lanes 2 and 9), LexA (lanes 5 and 10) or B42 (lanes 1 and 6) protein alone were used, growth on X-Gal medium in the presence of galactose did not result in the appearance of blue colonies. As well, controls in which LexA-Vpu was introduced in yeast containing either B42-Vpr or B42-Pr55 gag did not show any sign of ␤-galactosidase expression (data not shown). These interaction experiments were also confirmed using the ONPG colorimetric assay (Fig. 2B). The results presented in Fig. 1A suggest direct and specific binding of Vpr to the Pr55 gag .
Finally, Western blot analysis was performed to confirm that the LexA-Vpr bait and the B42-Pr55 gag prey expressed the expected fusion proteins (Fig. 1B). A 36-kDa LexA-Vpr fusion protein was detected using a rabbit polyclonal anti-Vpr serum, while the B42-Pr55 gag fusion corresponded to a 68-kDa protein using a rabbit anti-p24 serum. Since the B42-Pr55 gag is a large fusion protein, the other bands detected by Western blot analysis are likely to be degradation products generated during yeast protein extraction.
Regions within the Pr55 gag and Vpr Important for the Interaction-Previous reports clearly established that the p6 domain of the Pr55 gag was necessary and sufficient for Vpr virion incorporation (24,25,27). In order to determine the domains of importance for the Vpr-Pr55 gag direct interaction, we first tested different Gag constructs in the two-hybrid system ( Fig.  2A). Kondo et al. (25) showed that the addition of the p6 domain to the C terminus of the MLV p65 Gag precursor was sufficient to mediate Vpr incorporation, while the wild-type MLV Gag precursor was incapable of incorporating Vpr ( Fig. 2A). In order to determine if this specific incorporation of Vpr into MLV/HIV p6 chimeric virus was based on a direct interaction between the chimeric Gag gene product and Vpr, we quantitatively measured ␤-galactosidase activity in the yeast two-hybrid system using the ONPG colorimetric assay. As shown in Fig. 2B, the ␤-galactosidase activity detected for the Pr55 gag -Pr55 gag interaction is roughly 2 times the one detected for the association between Vpr and the Pr55 gag . The ␤-galactosidase activity detected for the Vpr-MLV/HIVp6 interaction is similar to the one observed for the Vpr-Pr55 gag interaction (Fig. 2B). As expected, no interaction was observed with the wild-type p65 MLV Gag precursor. Since the only difference between these two constructs is the addition of the p6 domain, this result suggests that Vpr can specifically associate with the p6 domain of HIV-1, and that p6 gag is necessary and sufficient for the association with Vpr. In order to pinpoint the region of the p6 domain that may act as a potential Vpr binding site, 2 out of the 52 amino acids from the p6 domain were mutated in the Pr55 gag . Leu 44 was changed for a Pro and Phe 45 for a Ser (Fig. 2A). These two substitutions altered the conserved L-X-S-L-F-G motif of the p6 domain shown to be important for Vpr incorporation (26,27,39). As demonstrated in Fig. 2B, this mutation rendered the HIV-1 Pr55 gag completely incapable of associating with Vpr. Interestingly, this mutation in the context of a proviral construct failed to incorporate Vpr (Fig. 5A). In order to confirm that the lack of ␤-galactosidase activity detected with the B42-MLVp65 and the B42-HIV-1 Pr55 gag (p6 L44P/F45S) fusions was not due to stability problems or the absence of protein expression, we performed Western blot analysis to detect these fusion proteins. Western blot clearly demonstrated that the B42-MLVp65, B42-MLV/HIVp6, and the B42-HIV-1 Pr55 gag (p6 L44P/F45S) fusion proteins were expressed to similar levels in the yeast EGY48 as determined by immunoprecipitation followed by Western blot (data not shown). Together, these results suggest that the mechanism by which Vpr is transincorporated into HIV-1 particles involves a direct physical interaction with the p6 domain of the Pr55 gag .
de Rocquigny et al. (31) recently reported that the integrity of the zinc finger structures in the NCp7 was important for Vpr virion incorporation. Consequently, we investigated the effect of NCp7 mutants on the Vpr-Pr55 gag interaction. Fig. 2A demonstrates the structure of the NCp7 zinc fingers. Different mutants affecting the zinc binding domains (H23C is a substitution of His 23 for Cys; C28S/C49S contains substitutions of Cys 28 and Cys 49 for Ser; ⌬K14-T50 is a deletion of both zinc fingers) were fused in translational frame with the B42 transactivator and assayed for ␤-galactosidase activity using the ONPG colorimetric assay in the presence of LexA-Vpr (Fig. 2B). The results demonstrated that neither the HIV-1 Pr55 gag (p7 H23C), the HIV-1 Pr55 gag (p7 C28S/C49S), nor the complete deletion of the NCp7 zinc fingers (HIV-1 Pr55 gag (p7 ⌬K14-T50)) affected the binding between Vpr and the Pr55 gag. This result suggests that the integrity of the NCp7 zinc fingers in the context of the Pr55 gag is not essential for the Vpr-Pr55 gag interaction.
We next wanted to investigate the region of Vpr that is important for its association with the Pr55 gag . In order to address this question, five Vpr point mutants affecting differ-ent structural regions of the protein were fused to LexA sequences (Fig. 3A) and used for interaction experiments using the yeast two-hybrid system. The E25K and A30F mutants affect the predicted N-terminal ␣-helix of Vpr and were shown to be incapable of incorporation into virion particles (29). The Q65E point mutation is located within the second helix of Vpr and affects its intranuclear localization (47). Finally, the SR79 -80ID and the R80A mutants do not lead to cell cycle arrest (11). Yeast containing either wild-type LexA-Vpr or LexA-Vpr mutants were then transformed with the following B42 fusions: the HIV-1 Pr55 gag , the p65 MLV Gag precursor, the MLVGag/HIV p6 fusion, and the HIV-1 Pr55 gag (p6 L44P/ F45S) double mutant, and assayed for interaction using the ONPG colorimetric method. As shown on Fig. 3B, the E25K and the A30F mutants were unable to interact with both the HIV-1 Pr55 gag and the MLVGag/HIV p6 fusion, while the other Vpr mutants were not affected. Similar expression of all LexA-Vpr mutants was confirmed by Western blot analysis (data not shown). Together, this observation suggests that the integrity of the predicted N-terminal ␣-helical structure in Vpr is essential for the physical association with the Pr55 gag . As well, this result confirms the relevance of the binding assay used in this study since the two mutants shown not to interact with the Pr55 gag were shown not to incorporate virion particles (29).
Direct in Vitro Interaction between Vpr and the p55 Gag Precursor-In order to confirm whether the association between Vpr and the Pr55 gag observed in yeast could be reproduced using another approach, we took advantage of an in vitro binding assay using recombinant fusion proteins. First, the vpr gene was introduced into the pMAL-c2 vector in fusion with the maltose-binding protein (MBP). MBP, MBP-Vpu, and MPB-Vpr fusion proteins were then produced in bacteria and purified as described under "Experimental Procedures." Then, the entire Gag open reading frame was used to generate in vitro labeled protein. The in vitro translated Pr55 gag was incubated with MBP, MBP-Vpu, or MBP-Vpr fusion proteins, that were previously immobilized on amylose resin. Following a 2-h incubation, the complexes were washed several times, eluted, and then analyzed on a 12.5% SDS-PAGE. As shown in Fig. 4A, the Pr55 gag was able to specifically interact with MBP-Vpr, and FIG. 3. The Vpr-Pr55 gag interaction requires an intact ␣-helical structure in Vpr. A, the amino acid sequence of the wild-type (WT) Vpr is shown. The position of the predicted amphipathic ␣-helical structures, the leucine/isoleucine-rich (LR) region, and the basic amino acid-rich region in Vpr are also indicated. Design and construction of the E25K, A30F, Q65E, SR79 -80ID, and R80A Vpr mutants were previously described (18,29). B, the interaction strength between B42-Pr55 gag , B42-MLVGag, B42-MLVGag/ HIV-1 p6, and B42-HIV-1 Pr55 gag (p6 L44P/F45S) double mutant and LexA-Vpr, LexA-VprE25K, LexA-VprA30F, LexA-VprQ65E, LexA-VprSR79 -80ID, or LexA-VprR80A was assessed using the liquid ␤-Gal assay. The histogram represents averaged data from at least three different experiments. not with MPB-Vpu nor with MBP alone. To ensure that the 55-kDa band associating with MBP-Vpr was really the p55 Gag precursor, we electrotransferred the SDS gel onto a nitrocellulose filter and performed a Western blot using an anti-p24 antibody. The immunoblot confirmed that the 55-kDa protein was effectively the Pr55 gag (data not shown). An additional 32-kDa band also specifically associated with the MBP-Vpr fusion protein (Fig. 4A, lane 1). This fragment was confirmed to be a Gag-related product by Western blot analysis (data not shown). The smaller proteins in the input loading (lane 4) probably represent nonspecific cleavage products, products from initiation at downstream AUG, or premature translation termination.
We next attempted to analyze the strength of the Vpr-Gag interaction by studying the effect of increasing salt concentration on the recovery of the Pr55 gag by the MBP-Vpr proteins. This technique was previously used to determine the strength of the interaction between the cyclophilins A and B and the Gag protein (48). As demonstrated in Fig. 4B, the Vpr-Pr55 gag interaction could sustain 900 mM NaCl. Coomassie Blue staining of the gel demonstrated that equal amounts of MBP-Vpr were still present after washing with these different salt concentrations (data not shown). Together, these results demonstrate biochemical evidence for a direct binding between the Vpr protein and the Pr55 gag , and confirm the data demonstrated genetically using the yeast two-hybrid system (Fig. 1). Furthermore, we conclude that the association between the Pr55 gag and Vpr is a strong interaction in vitro, sustaining salt concentrations of 900 mM NaCl.
We next wanted to determine if we could interfere with the interaction by specifically blocking the accessibility of the Pr55 gag to the MBP-Vpr fusion protein. In order to address this question, we used two different anti-Vpr antibodies: a polyclonal rabbit antiserum generated against recombinant Vpr protein (23) and a rabbit polyclonal anti-peptide serum that recognizes the first 19 amino acids of Vpr (46). As controls, anti-Vpu and anti-Myc antibodies were used for the experiments. These different antibodies were incubated for 3 h in the presence of immobilized MBP-Vpr, washed several times to remove unbound and nonspecifically attached antibodies, and then presented to equal amounts of [ 35 S]methionine-labeled Pr55 gag . Although the anti-Vpu and anti-Myc antibodies demonstrated slight nonspecific competition with the MBP-Vpr-Pr55 gag interaction, this nonspecific effect never reached the levels observed with the anti-Vpr antibodies in several experiments. Interestingly, the anti-Vpr antibody directed against the first 19 amino acids of Vpr (lane 3) interfered more efficiently with the MBP-Vpr-Pr55 gag interaction than the polyclonal anti-Vpr (lane 2). The result presented in Fig. 4C suggests that the binding between Vpr and p55 can be affected and that the potential use of other molecules could be used to affect the Vpr-Pr55 gag interaction and consequently prevent Vpr virion incorporation.
Mutations That Affect the Binding between Vpr and the Pr55 gag Also Affect Vpr Virion Incorporation-In order to confirm our hypothesis that direct binding between Vpr and the Pr55 gag is required for Vpr virion incorporation, we investigated whether our mutants in the Pr55 gag were still capable of incorporating the Vpr accessory protein into virions. To analyze the HIV-1 Pr55 gag (p6 L44P/F45S) mutant for Vpr virion incorporation, MT-4 cells were transfected with an infectious proviral clone of HIV-1 (pNL4.3) expressing in cis either the wildtype or the Pr55 gag (p6 L44P/F45S) protein. Following transfection, the cells were radiolabeled and the ability of Vpr to be incorporated into virion was monitored by immunoprecipitating both cell and virion-associated viral proteins. We have previously described a ratio system to assess the amount of Vpr found in the virions as a proportion of the total amount found in the cell using other virion proteins as standards (RT) (29). Essentially, the virion-associated Vpr value is calculated based on the level of internal control p66 (RT) present in the virions as well as the proportion of the total Vpr found in the cell as determined by densitometric scanning of autoradiograms. This value is then compared with wild-type virionassociated Vpr value, which is set at 100%. As can be observed in Fig. 5A, substantial amounts of Vpr could be detected when proteins from wild-type virions were immunoprecipitated (lane 8). However, no detectable amount of Vpr protein was observed in virions generated from the proviral construct harboring the p6 L44P/F45S mutation (Fig. 5A, lane 7), even though Vpr was expressed in transfected MT-4 cells (Fig. 5A, lane 3). This observation is consistent with our result (Fig. 2B) that this mutant in the context of the Pr55 gag is incapable of binding Vpr, thus suggesting that a direct interaction between Vpr and the p6 domain of the Pr55 gag is required for Vpr virion incorporation.
We previously demonstrated that proviral constructs harboring the HIV-1 p7 C28S/C49S mutations in the nucleocapsid region of Gag did not extensively affect Vpr virion incorporation (23). This is consistent with our binding analysis (Fig. 2B), suggesting that the integrity of the zinc binding motifs of the NCp7 is not critical for the interaction between Vpr and the Pr55 gag . In order to further confirm this, we decided to analyze the ability of Vpr to incorporate in viruses harboring either the HIV-1 p7 H23C or the HIV-1 p7 ⌬K14-T50 mutations. Because these viruses have been shown to be highly affected in their ability to package HIV-1 RNA (40), these viruses are not replication-competent. Consequently, we decided to use COS-7 cells for viral particle production as described previously (23). Using the same antisera as for the HIV-1 p6 L44P/F45S mutations, viral proteins from lysed cells and sucrose cushionpelleted virions were immunoprecipitated for both the HIV-1 p7 H23C and the HIV-1 p7 ⌬K14-T50 proviral constructs. Fig.  5B shows the virion-associated proteins from one of two independent experiments. As can be seen, both virion particles generated from the HIV-1 p7 ⌬K14-T50 and the HIV-1 p7 H23C proviral constructs were still competent in incorporating Vpr (Fig. 5B, lanes 11 and 12, respectively). Moreover, quantification (Fig. 5C) using densitometric analysis revealed that Vpr is trans-incorporated to levels similar to wild-type virus into the HIV-1 p7 H23C mutant virions. Because the HIV-1 p7 ⌬K14-T50 virion particles are highly affected in their Gag processing (no detectable p66RT band, very low p24/p25 Gag; Fig. 5B, lane 11), we did not quantify Vpr virion incorporation (Fig. 5C). Nonetheless, Fig. 5B (lane 11) shows that Vpr is still incorporated in substantial amounts in the HIV-1 p7 ⌬K14-T50 virion particles. DISCUSSION HIV-1 Vpr is the major virion-associated accessory protein.
As Vpr is not synthesized as part of the Gag polyprotein precursor, it must utilize a distinct mechanism in order to be incorporated into virion particles. It has been clearly demonstrated through Gag deletion analysis that the virion incorporation of the Vpr protein requires the p6 domain from the p55 precursor (24,25,27). However, the association between Vpr virion incorporation and Vpr-Pr55 gag binding was still missing. Several groups (26,27) demonstrated that single amino acid substitutions or deletions of either Leu 44 or Phe 45 in the p6 domain abolished the ability of Vpr to be incorporated in the context of MLV/HIV p6 or Rous sarcoma virus/HIV p6 chimeric viruses. In addition, the inability of our HIV-1 Pr55 gag (p6 L44P/F45S) double mutant to incorporate Vpr was confirmed in the context of an infectious proviral clone of HIV (Fig. 5A). Consequently, the inability of Vpr to interact with the HIV-1 Pr55 gag (p6 L44P/F45S) double mutant (Fig. 2B) brings a direct correlation between the lack of direct binding to the Pr55 gag and the incapacity of Vpr to be incorporated into virion particles.
Evidence of direct interaction between Vpr and other Gag domains has been recently reported. de Rocquigny et al. (31) reported that the zinc fingers of the nucleocapsid protein (NC) were important for Vpr virion association. Moreover, using chemically synthesized peptides, they demonstrated that Vpr could directly interact with NCp7, but not with p6 in vitro. However, our result demonstrated that only mutants in the p6 domain (LF-PS) resulted in the loss of binding between Vpr and the Pr55 gag , while mutations or the complete deletion of the p7 zinc fingers did not affect the interaction (Fig. 2B) and Vpr incorporation (Fig. 5B). It is possible that the zinc fingers of the NCp7 are important for Vpr binding in the context of the mature NC. However, our results indicate that in the context of the Pr55 gag , the zinc finger motifs are less critical for the Vpr-Pr55 gag interaction. Moreover, in contrast to de Rocquigny et al. (31), who demonstrated significant defects in Vpr virion incorporation using NC mutants, in particular with the HIV-1 p7 H23C mutant, our result did not reveal extensive impairment of Vpr virion-incorporating ability (Fig. 5, B and C). It is possible that this discrepancy results from experimental differences. Indeed, the basis for quantification of Vpr encapsidation are different, de Rocquigny's group used Western blot while we used radioimmunoprecipitation. Consequently, the importance of NCp7 in HIV-1 Vpr incorporation still needs to be demonstrated. In fact, Wu et al. (49) deleted the complete NCp8 sequence in HIV-2 and did not affect the ability of Vpx to be incorporated into virions, while removal of p6 sequences resulted in loss of Vpx incorporation.
The structural domains within Vpr important for incorporation have also been studied. The extensive mutagenic analysis of several groups agrees that the amphipathic ␣-helical region located between amino acids 18 and 34 in the N terminus of Vpr is important for the incorporation of this accessory protein (28 -30, 50, 51). Our binding studies using five Vpr mutants, E25K, A30F, Q65E, SR79 -80ID, and R80A, showed that the two mutants that were unable to interact with the Pr55 gag were the mutants that lost their ability to be incorporated (29). This result also correlates the loss of incorporation of Vpr to its incapacity to directly interact with the Pr55 gag . The observation that Vpr could only interact with the Pr55 gag when fused to the C terminus of the LexA DNA binding domain (Fig. 1A, lane  7) but not when fused to the B42 transactivator (Fig. 1A, lane  4) suggests that this interaction requires proper conformation. It is likely that either the B42-Vpr or the LexA-Pr55 gag fusion, or both, are not presented in the proper structure to expose their respective binding domain. Interestingly, we suspect that the p6-binding motif of Vpr requires more than just the predicted N-terminal ␣-helical domain since attempts to compete the in vitro Vpr-Pr55 gag association with a series of Vpr pep-tides were unsuccessful. Peptides harboring Vpr amino acids 1-19, 19 -35, or 23-37 could not compete the interaction between Vpr and the Pr55 gag . Moreover, Yao et al. 2 demonstrated that fusion proteins containing amino acids 1-62, which harbor the predicted N-terminal ␣-helical moiety of Vpr, are incapable of incorporating virion particles, while polypeptides fused to amino acids 1-80 of Vpr are efficiently incorporated. These results suggest that the presence of the leucine/isoleucine-rich domain of Vpr (amino acids 60 -80) might be important for correct folding or exposition of the predicted N-terminal ␣-helical region. Similar results have also been observed by Sato et al. (52). It is noteworthy that our anti-Vpr peptide serum directed against the N-terminal region of Vpr (amino acids [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] was more effective in affecting the interaction with the Pr55 gag than the antibody directed against recombinant Vpr (Fig. 4C), which principally recognizes epitopes lying between amino acids 19 and 72. 3 This suggests that, even though the leucine/ isoleucine-rich region of Vpr might be important, the critical region for the Vpr-Pr55 gag interaction is the predicted N-terminal ␣-helical structure of Vpr.
From our and other groups' results, we present a model for the molecular mechanism by which the Vpr protein is transincorporated into progeny virions. We suggest that HIV-1 Vpr can only associate with p6 in the context of the Pr55 gag . In this context, p6 would have a proper conformation to directly associate with Vpr, and subsequently pull it within forming virions. Then, upon activation of the viral protease, the Pr55 gag would be processed and mature p6 would be release from the Pr55 gag . This release of p6 would result in its dissociation from Vpr. Subsequently, Vpr could associate with other virion proteins such as NC in the viral core in order to fulfill its functional role in the context of the preintegration complex early in infection.
In summary, the results presented here demonstrate direct interaction between Vpr and Gag in the context of the p55 precursor. Furthermore, our results suggest that Vpr transincorporation requires a direct binding to the p6 domain of the Pr55 gag. The development of an assay that demonstrates this critical interaction may allow the screening of molecules that prevent Vpr virion association and thus, Vpr early function, which could ultimately impair HIV-1 infection.