The Zinc Fingers of HIV Nucleocapsid Protein NCp7 Direct Interactions with the Viral Regulatory Protein Vpr*

The 96-amino acid protein Vpr functions as a regulator of cellular processes involved in human immunodeficiency virus, type 1 (HIV-1) life cycle, in particular by interrupting cells division in the G2 phase. Incorporation of Vpr in the virion was reported to be mediated by the C-terminal domain of the Pr55Gag polyprotein precursor, which includes NCp7, a protein involved in the genomic RNA encapsidation and p6, a protein required for particle budding. To precisely define the Gag and Vpr sequences involved in this protein-protein interaction, NCp7, p6, and Vpr as well as a series of derived peptides were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. Binding assays were carried out by Far Western experiments and by competition studies using (52–96)Vpr immobilized onto agarose beads. The results show that interaction between NCp7 and Vpr occurs in vitroby a recognition mechanism requiring the zinc fingers of NCp7 and the last 16 amino acids of Vpr. Moreover, NCp10, the equivalent of NCp7 in Moloney murine leukemia virus but not polysine inhibits Vpr-NCp7 complexation. Interestingly enough, Vpr was found to interact with Gag, NCp15, and NCp7 but not with mature p6 in vitro.In vivo mutations in NCp7 zinc fingers in an HIV-1 molecular clone led to viruses with important defects in Vpr encapsidation. Together, these results suggest that NCp7 cooperates with p6 to induce Vpr encapsidation in HIV-1 mature particles. The NCp7-Vpr complex could also be important for interaction of Vpr with cellular proteins involved in cell division.

The protein Vpr, which contains 96 amino acids (see Fig. 1A) and could form oligomers, was reported to enhance virus replication, particularly by inducing arrest in cell cycle in the G 2 phase (1-6). Moreover, Vpr was shown to participate with the matrix protein MA in the nuclear transport of the preintegration complex (7). Vpr is present in virions, and its encapsidation has been reported to be dependent on the presence of NCp15, the C-terminal part of the polyprotein Gag (8 -14). In the virions, NCp15 is cleaved by the viral protease to form NCp7 and p6 (15). NCp7 is a small basic protein of 72 amino acids (see Fig. 1A) characterized by the presence of two spatially close zinc fingers of the CX 2 CX 4 HX 4 C type (16,17), NCp7 exhibits nucleic acid binding and annealing activities in vitro and is required for proviral DNA synthesis and virion formation in vivo (reviewed in Ref. 18). The function of p6 remains unclear, although mutations in the sequence of this protein cause alterations in a late step of viral budding (19). Vpr was recently reported to interact with NCp7 in vitro (20) in agreement with experiments in which Vpr was co-precipitated with Gag products from infected cells (11) and with results showing that virus mutants with truncation of the C-terminal domain of Gag were unable to export Vpr from the cell (9,10). Various HIV-1 1 gene manipulations such as selective deletion of p6 or co-transfection of genes encoding Vpr and heterogenous Gag polyproteins supported a critical role for p6 in Vpr incorporation (12). The leucine-rich motif LXSLFG of p6 was suggested to be critical for virion association of Vpr (13,14). However, all the chimeric HIV-1 Gag polyproteins leading to Vpr incorporation contained NCp12 from avian leukosis sarcoma virus or NCp10 from MoMuLV in place of the native NCp7, suggesting that the presence of a nucleocapsid protein preceding the p6 sequence could play a role in incorporation of Vpr in virions. This hypothesis is supported by the observed replacement of a NC protein by another nucleocapsid protein in some steps of the retroviral life cycle (21,22) and by the similarities in the tridimensional structure of NCp7 and NCp10 (16,23). Moreover, the sequence of Vpr necessary for its incorporation into virions remains to be elucitated because the C-terminal domain was suggested to be involved (10), whereas other groups have proposed the involvement of the N-terminal sequence (24,25).
With the aim to characterize more precisely, the interaction of Vpr with the maturation products of NCp15, various in vitro tests were performed using the first chemically synthesized active Vpr. In vitro interactions of this protein with synthetic NCp7, p6, and cellular extracts containing Gag and NCp15 were measured by various methods. In these conditions, Vpr was shown to interact with NCp7 but not with p6. The zinc finger domains of NCp7 and the C-terminal part of Vpr are involved in the complex whose formation is strongly reduced by mutation of the Trp residue on the NCp7 distal zinc finger. NCp10 from MoMuLV, which share common structural features with NCp7 but not with unrelated basic proteins was also found able to recognize Vpr. These results were confirmed and extended by means of site-directed mutagenesis done in the nucleocapsid domain of the molecular clone pNL4.3, suggesting that NCp7 and p6 could cooperate to encapsidate Vpr. Moreover, the NCp7-Vpr complex could play a role in the functions of both proteins during virus replication as recently shown for the activation of the phosphatase protein phosphatase 2A involved in cell division (26).

MATERIALS AND METHODS
Peptide Synthesis-NCp7, p6, Vpr, (1-51)Vpr, and (52-96)Vpr were synthesized on a 433 Automated Peptide Synthesizer (Applied Biosystem) using the procedure already described in detail (32). A biotinyl group was added on the N-terminal residue of peptides when necessary. Amino acids chains were protected by t-butyloxycarbonyl or trityl groups and N␣ by a Fmoc. The hydroxymethylphenoxy-methyl-polystyrene resin was used, and amino acids were incorporated using N,NЈdicyclohexyl carbodiimide/hydroxybenzotriazole as a coupling reagent. At the end of the synthesis, the peptidyl resin was treated for 2 h with trifluoroacetic acid in the presence of scavengers to obtain the crude peptide. Purification was performed by reverse phase HPLC with acetonitril-water gradients. The purity of the peptides already synthesized was checked by electrospray mass spectroscopy. NCp7, Cys 23 -NCp7 and NCp fragments (13-30; 34 -51; 12-53), Acr 37 (12-53)NCp7, and NCp10 were lyophilized with 1.5 equivalents of ZnCl 2 per zinc finger. Details on the synthesis of p6, Vpr, and derivatives will be given elsewhere. 2 Fig. 1B.
Antibodies-A previously characterized NCp7 monoclonal antibody (HH 3 ) was used (27). Monoclonal antibodies against (1-51)Vpr, (52-96)Vpr, and p6 were obtained with the synthetic peptides using the method described in Ref. 27. Their characterization will be reported elsewhere. 2 Far Western Blot Analysis-Vpr, (1-51)Vpr, and (52-96)Vpr were loaded on a 15% SDS-polyacrylamide gel electrophoresis (PAGE) and then transferred onto Hybond-C super membrane (Amersham Corp.). The membrane was treated with the SuperBlock blocking buffer (Pierce) for 2 h at room temperature to discard nonspecific interactions and incubated with 1 M p6 or 1 M NCp7 in 50 mM Tris, pH 7.4, 150 mM NaCl, 0.1% Tween 20 (TBST) overnight at 4°C. Due to the high zinc affinity of NCp7 (28), no ZnCl 2 was added to the buffer. After three successive washes in TBST, bound proteins were detected by incubation firstly with anti-NCp7 (27) or anti-p6 monoclonal antibodies diluted 1:4000 in SuperBlock bloking buffer for 1 h and then with a peroxidase conjugated anti-mouse antibody for 1 h. Complexes were revealed by the ECL method (Amersham Corp.) with peroxidase substrate incubation. This method was also carried out with the Gag polyprotein. Thus, Gag precursor was expressed in recombinant baculovirus-infected cell, and the proteins were loaded on 12% SDS-PAGE and then transferred onto nitrocellulose membranes (29). Blots were incubated with 1 M of Vpr following the procedure described with NCp7, and detection of Vpr was achieved with monoclonal antibodies.
Competition Experiments Using Agarose Beads Immobilized Avidin-25 l of ImmunoPure Immobilized Avidin (Pierce) per experiment were equilibrated in 20 mM Tris, 0.5 M NaCl, 1% Nonidet P-40, 0.1% Tween 20. The beads were then incubated with 1.5 10 Ϫ8 mol of Nterminal biotin (52-96) Vpr in the same buffer. Knowing on one hand by the supplier that the capacity of binding was 2.5 10 Ϫ9 mol of biotin/25 l of beads and on the other hand that the affinity of biotin to Avidine is 10 15 M, we assumed that 2.5 10 Ϫ9 mol of (52-96)Vpr had been captured. After three successive washes with the same buffer, NCp7 (4.5 10 Ϫ7 M) was added and incubated overnight at 4°C. The mixture was centrifugated, and the beads were washed once with TBST, 0.5 M NaCl, 1% Nonidet P-40, 0.1% Tween and once with TBST, 0.15 M NaCl. When NCp7 derivatives were used as competitors, they were preincubated for 1 h with biotin (52-96) Vpr beads at room temperature before the addition of NCp7. When Vpr derivatives were used as competitors, peptides were incubated for 1 h with NCp7 prior to the addition of biotin (52-96)Vpr beads. The same results were obtained after a longer period of incubation, suggesting that equilibrium was reached after 1 h.
NCp7 bound to immobilized biotin (56 -92)Vpr was released by mixing the beads with 1 volume of 2 ϫ loading buffer (50 mM Tris, 10% glycerol, 2% SDS, 0.05% bromphenol blue, 200 mM dithiothreitol) and heated at 80°C for 2 min. To detect NCp7, collected samples were separated by 20% SDS-PAGE and transferred onto Hybond-C super membrane (Amersham Corp.). Blots were revealed with anti-p7 monoclonal antibody (1:4000) diluted in 2% bovine serum albumin TBST followed by horseradish peroxidase-conjugated anti-mouse IgG (Amersham Corp.). Complexes were revealed by ECL. The monoclonal antibody used is selective of NCp7 and was unable to cross-react with NCp7 derivatives or Vpr (data not shown).
Viral Proteins Analysis-Clarified supernatants from DNA transfections (three independent experiments) were ultracentrifuged through a 20% sucrose cushion at 35,000 rpm (Ti 70 rotor) for 90 min at 4°C. Pelleted virions were dissolved in 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA and lysed with loading buffer (63 mM Tris-HCl, pH 6.8, 5% ␤-mercaptoethanol, 2% SDS, 10% glycerol, 0.02% bromphenol blue). Viral proteins were fractionated through a 5-20% gradient SDS-PAGE and electroblotted onto nitrocellulose in Tris-glycine buffer with 30% methanol. Samples were analyzed by immunoblotting with anti-Vpr polyclonal antibodies (Dr. E. Cohen) and anti-CAp24 monoclonal antibodies (Biomérieux) using a chemiluminescence system. Total amounts of CAp24 (under mature form and within its precursor) and Vpr were quantified by using the STORM 840 (Molecular Dynamics) imaging densitometer. Ratios of Vpr encapsidated into viral particles were normalized for the same amount of HIV-1 virions expressed as total Gag proteins, and values are given in percentages of the wild type virus. The values did not differ from Ϯ 15% of the mean.

RESULTS
Peptide Synthesis-In attempt to determine Vpr binding properties, the native peptide containing 96 residues was synthesized using Fmoc chemistry and a procedure already used in the synthesis of NCp7 (32). The first residue Ser 96 was loaded onto 0.13 mmol (150 mg, 0.96 mmol/g) of hydroxymethylphenoxy-methyl-polystyrene resin using N,NЈ-dicyclohexyl carbodiimide/dimethyl amino pyridine as a coupling reagent. Coupling times were increased progressively from 25 to 60 min as a function of the growing size of the peptide. After Arg 77 , Asp 52 , Lys 27 , and Glu 13 , half of the resin was removed both to reduce the peptide resin volume and to increase the amount of Fmoc amino acid added to the reactor vessel. The crude peptide (88 mg) was purified by reverse phase HPLC yielding 10.1 mg of peptide. The product was judged to be Ն95% pure by electrospray mass spectroscopy (Fig. 1B). Based on the secondary structure prediction of native Vpr and on the results already published showing different behavior of the N-and C-terminal parts of Vpr in terms of virion incorporation and nuclear localization (10,24,25), two Vpr sequences were synthesized: (1-51)Vpr and (52-96)Vpr. In this last case, a biotin residue was added to the peptide at the end of the reaction to allow its immobilization onto avidine agarose beads. Finally, to localize the minimal Vpr sequence required for NCp7 binding, three overlapped peptides (52-70)Vpr, (60 -80)Vpr, and (70 -96)Vpr derived from the C terminus of Vpr were synthesized.
The proteins located at the C terminus of the Gag precursor, i.e. the native NCp7 (32) and p6, were synthesized. To investi-gate the role of the NCp7 zinc fingers in the recognition of Vpr, the zinc fingers (13-30)NCp7, (34 -51)NCp7, and the (12-53)NCp7 were prepared. The last peptide encompassing the two CCHC boxes was shown by NMR studies to be the smallest peptide preserving the internally folded structure of NCp7 (16,17). Moreover, the C-(1-13) and N-(51-72) terminal fragments of NCp7 previously synthesized were compared with the zinc finger containing peptides. To study the role of the basic amino acid surrounding the CCHC boxes in Vpr binding, a zinc finger deleted peptide was synthesized by linking 13 VK, 29 RAPRKKG, and 51 TERQANF sequences by means of two Gly-Gly units (Fig. 1A). This peptide devoid of dactyl domains was designated as a-D(13-56)NCp7.
In Vitro Interactions of Vpr with NCp7 but Not p6 and Importance of the C-terminal Domain of Vpr-In vivo genetic analyses suggested that the C-terminal domain of Gag, which is processed in NCp7 and p6 (15), could interact with Vpr. Moreover, both the C and the N termini of Vpr were reported to be necessary for the incorporation of this protein in the virus particles (10,24,25). To understand at the molecular level which sequences are responsible for the interaction between Gag and Vpr, we used a Far Western blot analysis. Therefore, 5 ϫ 10 Ϫ10 mol of Vpr was loaded on a 15% SDS-PAGE gel, transferred onto nitrocellulose membrane, and incubated with pure synthetic NCp7 or p6. The formation of protein-protein complex was revealed using monoclonal antibodies directed against NCp7 or p6. Antibodies did not cross-react with native Vpr or with its C-or N-terminal fragments (data not shown). As depicted in Fig. 2A, a signal corresponding to the interaction between NCp7 and Vpr was detected (left panel). In contrast, even after a long exposure time, no signal could be detected when p6 was used instead of NCp7 (right panel). In addition, when the blot was incubated with NCp7 and p6 and then treated with a mixture of p6 monoclonal antibodies, no signal was observed (data not shown). The same type of experiment was used to investigate the domain of Vpr involved in NCp7 recognition. For this purpose, 1.5, 3, and 4.5 ϫ 10 Ϫ10 mol of (1-51)Vpr or (52-96)Vpr were tested in the same conditions as with Vpr. As depicted in Fig. 2B, only the C terminus of Vpr was found to be able to interact with NCp7. These data were confirmed by competition experiments (Fig. 2C). In this case, the nucleocapsid protein NCp7 was incubated for 1 h in the presence of 100 equivalents of (1-51)Vpr or (52-96)Vpr, and the mixture was incubated with the Vpr containing nitro cellulose membrane. The selective disappearance of the band observed with (52-96)Vpr reinforced the suggestion that the C terminus of Vpr is mainly involved in NCp7 recognition.
Involvement of the Zinc Finger Domain of NCp7 in Vpr Recognition-The domain of NCp7 involved in Vpr recognition was investigated by means of an affinity test based on the immobilization of the (52-96)Vpr sequence substituted on its N-terminal part by a biotin. This residue was linked to the peptide through a caproic acid spacer to avoid steric hindrance between Vpr and avidin. This biotinylated peptide was captured on agarose-avidin beads that were incubated with NCp7 in the absence or in the presence of 1, 5, or 25 equivalents of NCp7 derivatives as competitors. The amount of NCp7 released from the beads was analyzed by successive 20% SDS-PAGE nitrocellulose membrane transfer and Western blot analysis with NCp7 monoclonal antibodies (27). In Fig. 3 (A and B), a comparison between lane 1 and lane 2 confirmed the formation of the NCp7-(52-96)Vpr complex previously characterized by the Far Western method (Fig. 2A). This interaction is not critically dependent on salt concentration because it was observed in the 0.15-1 M NaCl range. The choice of 0.5 M NaCl for competition experiments was driven by the fact that at this concentration, a better ratio between nonspecific (binding of NCp7 to agarose gel) and specific interactions was obtained (data not shown).
To determine if the N-and C terminus of NCp7 were involved in the protein-protein complex, we tested both (1-13)NCp7 and (51-72)NCp7. Fig. 3A (lanes 3 and 4) shows that the N-terminal part of NCp7 did not participate significantly in complex formation because at 5 and 25 equivalents of (1-13)NCp7, no decrease of NCp7 signal was observed. Regarding the C-terminal part of NCp7, the (51-72)NCp7 fragment was unable to inhibit NCp7-Vpr complexation (Fig. 3B, lanes 7 and  8).
The NMR study of NCp7 (16,17) has revealed a spatial proximity of the two zinc fingers partially due to the prolinecontaining conformation of the basic peptide linker 29 RAPRKK. The possibility that this sequence is required for the interaction has been tested by using a peptide in which both zinc fingers were removed and replaced by two Gly-Gly linkers. This peptide, a-D(13-56)NCp7 when tested at 5 and 25 equivalents compared with NCp7 concentration (Fig. 3A, lanes 5 and 6), was unable to displace the NCp7-(52-96)Vpr complex, indicating that the basic amino acid sequences surrounding the zinc finger domains were not directly involved in Vpr recognition. In contrast, five equivalents of (12-53)NCp7, a peptide that encompasses the two zinc fingers, induced a complete disappearance of the NCp7 band (Fig. 3B, lanes 3 and 4). The same experiments carried out with 25 equivalents of zinc finger containing peptides showed that the first finger (13-30)NCp7 alone did not induce a significant reduction in NCp7-Vpr recognition, whereas the distal zinc finger (34 -51)NCp7 totally inhibited complex formation compare Fig. 3A, lanes 7 and 8, with Fig. 3B, lanes 5 and 6). Nevertheless, at 50 equivalents, the proximal zinc finger competes with NCp7 for Vpr recognition.
All these results suggest that both zinc fingers are involved in Vpr recognition and that this interaction is stabilized by the N terminus finger domain. According to the main role of the distal zinc finger, a NCp7 derivative, Aca 37 -NCp7, in which Trp 37 was replaced by an acridinylalanine residue (51), did not displace the NCp7-Vpr complex formation, suggesting a highly specific recognition of the C-terminal zinc finger.
Interaction of NCp10 from MoMuLV with (52-96)Vpr-Owing to the high similarity between nucleocapsid proteins among retroviruses, the occurrence of an interaction between NCp10 from MoMuLV and (52-96)Vpr was suspected. The NCp10 protein, composed of 56 residues and previously synthesized in the laboratory (33,34), was tested as a competitor in the same experiment carried out with NCp7 derivatives. As depicted in Fig. 4, 5 and 25 equivalents of NCp10 (lanes 3 and 4) induced clearly a displacement of the immobilized NCp7-Vpr complex, although to a lesser extent than NCp7. The basic polypeptide polylysine (Fig. 4, lanes 5 and 6) did not inhibit NCp7-(52-96)Vpr binding.
The 80 -96 Amino Acid Sequence of Vpr Is Involved in NCp7 Binding-Competition experiments were carried out with Vpr derivatives to characterize the minimal sequence of Vpr necessary for NCp7 recognition. Increasing concentrations of Vpr derivatives were incubated with native NCp7 prior to the addition of (52-96)Vpr beads. As depicted in protein in Vpr encapsidation, mutations have been introduced in the NC domain of the molecular clone pNL4.3. These mutations affect either the zinc binding domains (H23C is a substitution of Cys 23 for His; ⌬D2 is a deletion of the first zinc finger; H44C is a substitution of Cys 44 for His; ⌬D2 is a deletion of the second zinc finger) or a basic residue in the C-terminal part of NC protein (Leu 59 substituted for Lys). NC mutant and wild type plasmids were transfected into Hela P4 cells. Virions released into the supernatant were pelleted through a 20% sucrose cushion, and viral proteins were analyzed by Western blot. Samples from Hela P4 cells transfections were adjusted for equal amounts of mature CAp24, loaded on a 5-20% gradient SDS-PAGE gel, and analyzed using anti-CAp24 and anti-Vpr antibodies (Fig. 6A).
Mutations H23C, H44C, ⌬D2, and to a lesser extent ⌬D1 and K59L affect Pr55 Gag maturation as judged from an increase in the prominence of the p41 processing intermediate, which is known to contain MAp17 and CAp24 -25 sequences, and an other Gag maturation intermediate shown to contain NC region (data not shown). It should be noted that mature NCp7 was detected in each virion mutant (data not shown).
The relative amounts of CAp24 in mature and immature forms were first measured and compared with CAp24 in the wild type particles (Fig. 6A). The same Western blot allowed a quantitative analysis of the ratio of Vpr incorporated in viral particles (Fig. 6B), compared with the total amount of CAp24 protein (mature and as part of its precursors). This comparison showed pronounced differences, which are expressed as ratios of the relative amount of Vpr in the different mutants versus Vpr in the wild type (Fig. 6B). The percentages given in Fig. 6C were calculated from the amounts of CAp24 (mature and immature forms) versus the amounts of Vpr in wild type (100%) and mutants. The mutants that carried substitutions or deletions affecting the zinc fingers showed a severe defect in Vpr incorporation (10%, compare with the wild type), whereas mutation of the basic residue Lys 59 retained approximately the same amount (80%) of Vpr incorporated into viral particles (Fig. 6C). DISCUSSION The aim of this work was to characterize the domains specifically involved in the binding of Vpr to the C-terminal part of the Pr55 Gag precursor, which was reported to monitor encapsidation of this important regulatory protein. The first interesting result is the direct demonstration that NCp7 binds Vpr and is therefore likely to be involved in the in vivo recognition of the C-terminal part of Gag by Vpr. A reverse Far Western analysis was carried out with the Gag precursor as well as NCp15 (29) blotted onto nitrocellulose membrane. Both proteins were revealed using Vpr or (52-96)Vpr as probes, con-firming the recognition between Vpr and NCp15 (data not shown). Moreover, Far Western blot experiments pointed out that it is the C-terminal part of Vpr that is critically involved in the binding to NCp7 (Fig. 2). The domain of NCp7 involved in the complex with Vpr was analyzed in details by using Nbiotinylated (52-96)Vpr immobilized onto avidin-agarose beads. The results of competition experiments with various NCp7 fragments demonstrate that the distal zinc finger was mainly implicated in Vpr-NCp7 interaction, whereas the flexible N-and the C-terminal parts of NCp7 were not. Moreover, a peptide (13-56)NCp7 containing the basic linker RAPRKKG, but with the two CCHC boxes replaced by two Gly-Gly spacers, was unable to inhibit NCp7-Vpr complex formation, emphasizing the crucial role of the zinc finger domain in the NCp7-Vpr recognition process. The importance of the distal zinc finger of NCp7 for Vpr recognition is supported by the lack of inhibitory properties of NCp7 derivative in which Trp 37 was replaced by an acridinylalanine residue. The NCp7-Vpr interaction seems  6. Ratio of virion-incorporated Vpr. A, Western blot analysis of wild type and NCp7 mutants using anti-p24 antibodies. Virions with or without a mutation in NCp7 released into cell culture supernatants were pelleted, and the proteins were analyzed by Western blot as described under "Materials and Methods" using anti-CAp24 antibodies. Quantification was performed using the STORM 840 (Molecular Dynamics) imaging densitometer, and the values corresponding to the total amount of CAp24 (either as the mature protein or as part of its precursor) are compared with the amount of mature p24 in wild type particles. B, Western blot analysis of Vpr. The same experiment described for A was carried out using anti-Vpr antibodies. A ratio of Vpr found in wild type particles to the amount of Vpr determined for each mutant is given under the blot. C, quantification of incorporated Vpr. Histogram showing the amount of encapsidated Vpr in the various mutants (ratio of CAp24 on Vpr) given as a percentage of Vpr incorporated by the wild type virus. Vpr incorporated into wild type particles is referred to as 100%. WT, wild type. to be rather specific because polylysine failed to hinder NCp7-Vpr recognition (Fig. 4). This former result means that this protein-protein interaction is mainly dependent on hydrophobic contacts and probably hydrogen bond formation. Accordingly, buffers with high ionic strengths (0.5-1 M NaCl) only slightly reduced NCp7-Vpr binding (data not shown).
Implications of zinc finger motifs, different from the retroviral zinc finger, in protein-protein interactions were already reported. Thus, the protein, ZPR1 binds the cytoplasmic tyrosine kinase domain of the epidermal growth factor receptor through its zinc finger domain (CCCC type), and the zinc finger of rabphilin (CCCC type) is also essential for the binding of GTP-rab 3 (35,36). However, to the best of our knowledge, this is the first direct demonstration of a protein-protein interaction involving the zinc finger motifs of NCp7.
The biological relevance of the in vitro results emphasizing the critical role played by the zinc finger domain of NCp7 for Vpr recognition is supported by the results of site-directed mutagenesis experiments (Fig. 6). Thus, the replacement of the proximal and the distal zinc finger by a Gly-Gly unit in ⌬D1 and ⌬D2, respectively, led to a dramatic reduction in Vpr encapsidation. This was also observed when His 23 or His 44 , which behave as zinc ligands, were replaced by a Cys residue. Previous structural and biochemical studies have shown that this mutation did not greatly modify the high affinity of zinc for the NCp7 derivative but induced a change in the conformation of the mutated zinc finger with a loss of the proximity between the two zinc fingers and complete loss of virus infectivity when the mutation was done in the HIV virus (37)(38)(39). In agreement, with in vitro experiments, the replacement of Lys 59 by Leu generates only a slight decrease in encapsidated Vpr levels in the mutant virus.
The involvement of NCp7 in Vpr encapsidation would raise the question of the coexistence of NCp7-Vpr complexes with those resulting from the binding of genomic RNA to the nucleocapsid domain of Gag, a process known to be critical for RNA packaging (15,18). Fluorescence and NMR studies have shown that the zinc finger domains of NC proteins participate in the RNA binding (40,41), a finding supported by the observed reduction in RNA packaging of HIV-1 mutants with defects in the structure of the zinc fingers (37)(38)(39). The packaging of the genomic RNA and the encapsidation of a large number of Vpr molecules could occur by an interaction of Vpr restricted to the accessible domain of the NCp7-RNA complex. Accordingly, recent structural studies of the complex (12-53)NCp7-d(ACGCC) showed that a large part of the CCHC boxes remains free for additional interactions (53).
Competition assays with Vpr demonstrated that the last 16 amino acids of this protein are necessary for NCp7 binding in agreement with mutations showing that the sequence 73-96 is required for Vpr encapsidation (10). The 60 -81 leucine/isoleucine-rich domain and the short 36 -42 sequence of Vpr seem to be involved in dimerization of the protein (1) and accordingly were found in this study not to be involved in the NCp7-Vpr complex. The 80 -96 sequence of Vpr contains a short hydrophobic sequence IGII followed by a sequence enriched in basic amino acids. As shown in Fig. 3A, the positively charged linker of NCp7 did not participate to the recognition of this highly basic C-terminal part of Vpr. We have recently shown by NMR that the (52-96)Vpr domain contains a long ␣-helical 51-80 fragment followed by a less structured (80 -96) C-terminal sequence 3 in agreement with trypsin digestion experiments (1). Moreover, the 1 H NMR parameters corresponding to the (52-96) part of Vpr seem to be not largely modified in the spectrum of (1-96)Vpr, suggesting that the secondary structure of the C-terminal fragment is similar in the native protein and in the truncated domain. At this time, the structure of the complex between the C-terminal sequence of Vpr and the finger domain of NCp7 is still unknown. Nevertheless, one can reasonably hypothesize that, in the complex with NCp7, the largest part of Vpr remains accessible for additional interactions including Vpr oligomerization (1) and binding of other viral and/or host cellular proteins (42)(43)(44)(45).
A striking result of this study is the lack of interaction between p6 and Vpr in our in vitro conditions. These data were confirmed using surface plasmon resonance analysis (Biacore) and enzyme-linked immunosorbent assays showing that Vpr interacted with immobilized NCp7 but not with immobilized p6 (not shown). In contrast, incorporation of Vpr in new virus particles was shown to be critically dependent on the presence of p6, and the interaction between both proteins was proposed to involve the 36 -42 residues of p6 and a predicted ␣-helical domain located in the N-terminal part of Vpr (10 -14). The importance of p6 in Vpr encapsidation was confirmed by mutagenesis experiments showing that the presence of Vpr in virions was dependent on the association of this protein with the C-terminal part of the Gag polyprotein containing either NCp7 or a nucleocapsid protein from MoMuLV or RSV (13,15). Several explanations could be proposed for this apparent contradiction between in vitro and in vivo results. This could come from the construction of the heterologous Gag precursor. As mentioned by Li and co-workers (20), it is worth noting that in all cases the p6 nucleotide sequence was introduced at the 3Ј end of NCp10 or NCp12 sequences in the case of MoMuLV or Rous sarcoma virus, respectively (13,14). These nucleocapsid proteins have strong homologies with NCp7 in terms of primary sequence and in vivo biological properties (15,18,21,22), and the highly folded conformation of the CCHC box of NCp10 is very similar to that of the zinc fingers of NCp7 (17,23,40,46). Interestingly, NCp10 was also able to interact with Vpr in our in vitro conditions (Fig. 4), although its affinity, evaluated by plasmon resonance spectroscopy, appeared significantly lower than that of NCp7 (not shown). Thus, one cannot completely exclude the fact that in the heterologous Gag precursors constructed to study Vpr encapsidation (12)(13)(14), NCp10 (or NCp12) could play the role of NCp7 for Vpr incorporation into budding viruses.
Vpr encapsidation did not occur if the protein p6 was not directly linked to a nucleocapsid protein sequence (13,14), suggesting that in HIV-1 Gag, NCp7 and p6 linked together in NCp15 form could cooperate to direct Vpr encapsidation. Therefore, although p6 was found devoid of well defined conformation in solution (47), the lack of in vitro interaction between Vpr and p6 could be due to the loss in p6 of the conformation, allowing its recognition by Vpr in NCp15. Another possibility to explain the apparent discrepancies between in vitro and in vivo results could be the requirement of an oligomerization of Vpr for its p6-directed encapsidation. Selfassociation of CypA was shown to be critical for its packaging into budding virions through interactions with Gag at the level of a specific matrix sequence (48). The lack of significant Vpr oligomerization, in our in vitro conditions, could inhibit p6 binding. Another likely explanation that could reconcile in vitro and in vivo results is the participation of another (viral or cellular) protein interacting with p6 to form a complex inducing the encapsidation of Vpr bound to the zinc finger domain of NCp7 in the Gag polyprotein. Furthermore, several Gag determinants located at the N terminus of Pr55 Gag could cooperate to ensure an optimal Vpr packaging into newly formed viral particles (49).
In addition to its role in Vpr encapsidation, the Vpr-NCp7 complex could serve to the nuclear transport of the preintegration complex. Likewise, the physical association between MAp17 and Vpr (49) may be related to their putative roles in the nuclear transportation of the HIV-1 preintegration complex to the nucleus, which seems to be supported by the reduction in transportation of the viral preintegration complex in nondividing cells following mutations in both proteins (7). In fact, during the reverse transcription, NCp7 is probably still present in association with the reverse transcripted proviral DNA (50). Due to the lack of a traditional nuclear localization signal in NCp7, the NCp7-DNA complex could shuttle to the nuclear compartment through interaction with Vpr-MA. At this level, a specific contact between the glucorticoid receptor type II and Vpr was recently shown (42). Thus, the translocation of the proviral DNA to the nucleus could occur via a ternary complex involving NCp7 and Vpr or possibly through heterogenous oligomeric including other viral (p24, MA, or IN integrase) and/or cellular proteins. This hypothesis is currently under investigation in the laboratory using a monoclonal antibody against both NCp7 and Vpr. On the other hand, we have very recently shown that the arrest in the G 2 phase of HIV-1 infected cells produced by Vpr involves the formation of a complex between NCp7 and Vpr and that this association inhibits the phosphatase PP2A (26). Owing to the critical role played by Vpr and NCp7 in the retroviral life cycle, the precise molecular interactions between Vpr and NCp7 reported in this paper could facilitate the design of new antiviral agents (52).