The multidomain HIV-1 Vif protein recruits several cellular partners to achieve neutralization of the antiviral activity of APOBEC3 proteins. Vif neutralizes APOBEC-3G and APOBEC3F predominantly by forming an E3 ubiquitin ligase with Cullin5, ElonginB and ElonginC that targets these proteins for degradation by the ubiquitin-protea-some pathway. Vif associates with the Cullin5-ElonginB-ElonginC complex by binding directly to ElonginC via its SOCS-box motif and to Cullin5 via hydrophobic residues within a zinc-binding region formed by a conserved HCCH motif. The HIV-1 Vif-Cullin5-ElonginBC complex is then able to ubiquitinate the APOBEC3G factor bound to Vif by its N-terminal domain. In this review, we summarize the current knowledge about the structural determinants of Vif that allow it to interact with cellular and viral partners.

To date, little structural data is available on HIV-1 Vif protein and, more importantly no three-dimensional structure exists. This lack of data is first due to the difficulty of expressing high levels of soluble recombinant protein using either prokaryotic or baculovirus expression systems. Vif is a highly basic protein (pI = 10.7) that is conserved in all lentiviruses, except Equine Infectious Anemia Virus [19]. The Vif proteins of closely related HIV-1 strains are highly conserved (e.g. 91% of identity between HIV-1 HXB2 and HIV- 1 MN isolates), while those of slightly more distantly related lentiviruses have diverged significantly (e.g. there is a maximum of 30% of identity between HIV-1 HXB2 and SIV isolates) [20]. Alignments of Vif sequences from HIV-1, HIV-2, and SIV subtypes (Fig. 1) highlight several highly conserved motifs, i.e. an N-terminal tryptophan-rich stretch, a conserved HCCH motif and a SOCS-box motif. The N-terminal tryptophan-rich stretch (residues 1 to 21) is highly conserved (Fig. 1). Interestingly, different tryptophans (Fig. 2a) are required to mediate the Vif recognition and suppression of the target molecules APOBEC3G and APOBEC3F [21]. This point will be discussed in section III. Amino acids 63–70 and 86–89 that are well conserved and predicted to be important for formation of β-strand structures (Fig. 1) are critical for maintaining a normal expression level of Vif and for viral infectivity [22]. In HIV-1, the conserved glutamic acid at position 88 and the tryptophan at position 89 are located within the charged central hydrophilic region ⁸⁸EWRKKR⁹³ that is thought to enhance steady-state expression of Vif in host cells [22]. The HCCH motif (residues 108–139) consists of two conserved His/Cys pairs flanking a predicted α-helix that contains a cluster of hydrophobic conserved residues (Figs. 1, ,2).2). This motif was shown to coordinate one Zn²⁺ ion through the universally conserved residues H108, C114, C133 and H139, and to bind directly to Cullin5 [23,24] (see section IV). The highly conserved ¹⁴⁴SLQYLA¹⁴⁹ motif [20] defines the so-called BC-box motif and is crucial for inactivation of APOBEC3 proteins [13]. This BC-box motif is responsible for the binding to ElonginC, which further targets APOBEC3 antiviral factors to the proteasome. There are high similarities between the conserved SLQ(Y/F)LA ΦΦΦΦ motif (Φ for hydrophobic residue) of Vif and the SOCS-box of SOCS protein (suppressor of cytokine signalling). This subject will be developed in section V and for purpose of clarity, this domain will be called SOCS-box. Vif is a component of cytoplasmic RNPs mediating viral RNA interactions with the Pr55^Gag precursor and cellular factors [25–27] and Vif binds HIV-1 genomic RNA in vivo [25,27] and in vitro [25,28]. In addition, any mutation reducing the affinity of Vif for RNA diminishes viral replication in non-permissive cell lines, suggesting that RNA also plays a central role in Vif function [26,27]. This point will be issued in section VI. Vif is subject to intravirion processing by the HIV-1 protease [29]. Importantly, mutations in Vif that affect Pr-dependent processing also affect the ability of Vif to regulate viral infectivity, suggesting that intravirion processing of Vif is functionally relevant [29]. The processing site is located after L150 dowstream the highly conserved SOCS-box (Fig. 2b). The authors suggest that the Vif processing site is defined by structural constraints rather than by a specific amino acid sequence [30]. Post-translational phosphorylation of viral proteins by cellular kinases regulates HIV-1 infectivity. Many viral proteins are regulated by phosphorylation at different stages of the virus life cycle. Vif is phosphorylated in vitro and in vivo by cellular kinases and Vif phosphorylation is important for HIV-1 replication [31,32]. Four major phosphorylation sites were identified: T96, S144, T155 and T188, the three latter within the C-terminal domain of Vif (Fig. 2a). Importantly, T96 and S144 are highly conserved in all lentiviruses. Mutations of T96 resulted in significant loss of Vif activity and inhibition of HIV-1 replication [32]. Mutation of S144 to alanine resulted in loss of Vif activity, suggesting that phosphorylation at this site plays an important role in regulating HIV-1 replication and infectivity [31]. Synthetic Vif peptides corresponding to the local sequences of the phosphorylation sites were not phosphorylated by MAPK, suggesting that recognition of these sites by MAPK is likely to require structural determinants outside the phosphorylation sites [32]. Like several other proteins encoded by HIV-1, Vif proteins possess a strong tendency toward self-association. In relatively native conditions, Vif proteins form multimers in vitro, including dimers, trimers and tetramers [33]. This propensity for self association, from dimers to tetramers, has been also observed in cross-linking experiments [34]. This behavior was not limited to in vitro studies, as experiments based on co-immunoprecipitation or two-hybrid systems revealed Vif-Vif association within cells [33]. The domain mediating Vif self-association was first identified within the C-terminal domain of the protein in the proline-rich 156–164 region [33]. More precisely, the conserved ¹⁶¹PPLP¹⁶⁴ domain in HIV-1 Vif protein (Fig. 2b) was shown to play a key role in multimerization [35]. Vif multimerization at its C-terminus via this motif is thought to release Pr55^Gag, which binds to this region [35]. In HIV-1, Vif mulimerization is crucial for viral infectivity [35] and for preventing APOBEC3G incorporation into viral particles [36]. However, surprisingly, this multimerization site is not conserved in HIV-2 or SIV (Fig. 1). Some regions of HIV-1 Vif are predicted to be intrinsically disordered (solid lines in Fig. 1). Indeed, large segments of HIV-1 Vif secondary structure are predicted to be random coil (70–84, 130–143, 155–192), with however a more ordered secondary structure predicted for the N-terminus consisting of mostly β-sheets and one α-helix. Very recently, two models for the three-dimensional structure of HIV-1 Vif were presented [37,38]. Briefly, both models are based on predicted secondary structures shown in Fig. 1 and the search in the PDB (Protein Data Bank http://www.rcsb.org/pdb/home/home.do) for protein 3D structures that shares the same secondary structures. The second model [38], which seems to be more complete, incorporated the results of mutagenesis studies [39–41] that indicate that Vif contains at least two functionally crucial domains: an N-terminal domain that is important for binding to APOBEC3G, and a C-terminal region with a conserved SLQ(Y/F)LAΦΦΦΦ motif similar to SOCS-box that is responsible for its interaction with the ElonginB-ElonginC complex [13]. A three-dimensional model of HIV-1 Vif was constructed by homology modelling using two templates: the SOCS-box of VHL (von Hippel-Lindau tumor suppressor protein) as the template for the C-terminal domain, and the N-terminal domain of Narl, whose secondary structure presents strong similarity to the N-terminal domain of Vif [38]. While the two models roughly present the same secondary structure of Vif, the three-dimensional foldings are really divergent. The difficulties lie not only in organizing the secondary elements with respect to each other to access the 3D-fold but also to model the large loops present in Vif. Moreover, none of these models took into account the presence of the HCCH domain that is also crucial for Vif activity and that could be determinant for Vif folding. Each model can however provide specific details about mutation sites, but they should be used with caution, because of lack of structural data such as X-ray diffraction or NMR data.

Vif hijacks the cellular Cullin5-Ring E3 ubiquitin ligase to degrade APOBEC3 proteins (Fig. 3). Vif was shown to recruit Cullin5 in concert with ElonginB and C and Rbx1 to form an E3 ubiquitin ligase complex [13,40,58] similar to ElonginB/C-Cul2-SOCS-box complexes [59,60]. Vif proteins lack a clear Cullin5-box, and so the molecular basis of the recruitment of Cullin5 was first unclear. The first residues shown to interact with Cullin5 but not with ElonginB/C were the highly conserved C114 and C133, upstream the SOCS-box [13] (Fig. 2a). Indeed, mutation of each cysteine residue individually into a serine residue severely impaired the ability of Vif to interact with Cullin5, but not with ElonginB/C. The two cysteine residues were then shown to belong to a completely conserved HCCH motif with the sequence H108-X5-C114-X18-C133-X5-H139 (Fig. 2b). The HCCH sequence and spacing are highly conserved in Vif proteins and are critical for Vif function [61]. All individual residues of this HCCH motif were shown to be necessary for interaction with Cullin5 [62]. Indeed, individual substitutions in each of the His or Cys abolished the interaction with Cullin5 (in SIV-primate) leading the authors to the conclusion that not only the two cysteines but also the two histidines contribute to the interaction [62]. All conserved amino acids of this motif (between H108-H139) except G126 were found to be critical for assembly of the Vif-Cullin5 E3 ubiquitin ligase [61]. This HCCH motif was subsequently shown to form a structural domain that has the potential to bind zinc, making Vif a zinc-binding protein [23,24]. This HCCH motif is reminiscent of a zinc-finger like structure. However, the arrangement of the His and Cys residues and the spacing between them in Vif appear to be unique when compared to other known classes of zinc-binding domains [63]. Here again, it was shown that each individual zinc-coordinating residue was essential for Cullin5 association and APOBEC3G degradation but not for ElonginB/C binding [23,24]. Moreover, a highly conserved stretch of hydrophobic residues F115-X4-I/V120-X2-A123-I/L124 (Figs. 1, ,2)2) within the Vif zinc-binding domain is required for the interaction with Cullin5 [23]. F115 is present in most lentiviral Vif proteins, with W occasionally seen, and F115A, I120S and AL123/124SS mutants lost the ability to interact with Cullin5 [23]. The hydrophobic motif within the zinc-binding domain of Vif forms a critical functional interface with Cullin5. Disruption of this motif results in a defective Vif that is unable to recruit Cullin5 and consequently has lost the ability to degrade APOBEC3G [23]. Interestingly, it has been reported that HIV-1 Vif protein, as well as a minimalist HCCH peptide mimicking the HCCH domain of Vif, changes its conformation upon zinc binding [64]. Moreover zinc binding leads to rapid protein aggregation that is efficiently reversed upon chelating agent treatment like EDTA [64]. Zinc binding generates a conformation capable of forming high order protein assemblies. This self-association property of Vif protein recalls the capacity of Vif to form multimers, but it does not involve the same domain and the multimerization via the PPLP domain lead up to tetramers and not to aggregation. The authors proposed that zinc binding alters the native protein conformation to expose a protein-protein interaction domain. The resulting conformation would display surface side chain residues to the solvent, and could mediate protein-protein interaction, for example with Cullin5 [64]. A recent study demonstrates that the arrangement of the zinc coordination residues, HCCH, and the maintenance of both the conserved residues and the spacing within the HCCH motif are critical for Vif function. Indeed deletion of non conserved residues in each portion of the HCCH motif affected recruitment of Cullin5, APOBEC3G degradation and viral infectivity [61]. It has also been shown that chelation of zinc in cells by a membrane-permeable zinc chelator prevents Vif function in infectivity assays, allowing the virus to become sensitive to the antiviral activity of APOBEC3G [65]. To conclude, the HCCH zinc-binding motif facilitates Vif-Cullin5 binding by playing a structural role in positioning hydrophobic residues for direct contact with Cullin5.

In conclusion, Vif is a multidomain and multifunctional protein that therefore represent a viable target for both therapeutic and preventive interventions. However, there are currently no drugs against this protein in clinical trials. To date, its function and structural features have been fairly well investigated. However, there still exist numerous aspects that need to be resolved. A three-dimensional structure of Vif, alone if Vif is not intrinsically disordered, or in complex with cellular partner(s) would greatly help us to understand its behavior at the molecular level.