Host Restriction of HIV-1 by APOBEC3 and Viral Evasion Through Vif

Abstract

The arms race between virus and host is a constant battle. APOBEC3 proteins are known to be potent innate cellular defenses against both endogenous retroelements and diverse retroviruses. However, retroviruses have developed their own methods to launch counter-strikes. Most primate lentiviruses encode a protein called the viral infectivity factor (Vif). Vif induces targeted destruction of APOBEC3 proteins by hijacking the cellular ubiquitin-proteasome pathway. Here we review the research that led up to the identification of A3G, the mechanisms by which APOBEC3 proteins can inhibit retroelements, and the counter-mechanisms that HIV-1 Vif has developed to evade its antiviral activities.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Great strides have been made in the research of the Human Immunodeficiency Virus (HIV) since it was first identified as the causative agent of Acquired Immunodeficiency Syndrome (AIDS). However, it was not until a few years ago that the function of the viral protein, viral infectivity factor (Vif), was revealed. Vif was found to overcome the host antiviral protein, APOBEC3G (A3G) (Sheehy et al. 2002). The A3G protein belongs to a family of human cytidine deaminases that can inhibit the replication of Vif-deficient viruses. A3G is packaged into newly formed virions of the defective virus, and upon infection of new cells, induces C→U mutations in the minus strand of the newly synthesized viral cDNA (Sheehy et al. 2002; Bieniasz 2004; Rose et al. 2004; Goff 2003; Navarro and Landau 2004; Turelli and Trono 2005; Zhang et al. 2003; Mangeat et al. 2003; Lecossier et al. 2003; Harris et al. 2002, 2003; Mariani et al. 2003; Chiu and Greene 2008; Bishop et al. 2008). In addition, A3G may also have other antiviral activities that have not yet been fully characterized (Newman et al. 2005). Vif, however, counteracts this potent antiviral element of our immune system by hijacking components of the cellular degradation pathway, and targeting A3G for proteasomal degradation (Yu et al. 2003). This prevents A3G from being packaged into budding viruses, and allows them to successfully infect new cells. In this review, we seek to analyze recent data in the field and address areas of research that still need to be tackled.

2 Non-Permissive Cells and the Identification of A3G

HIV-1 Vif was first identified in 1986 as small 192 residue protein, with a molecular weight of 23 kD. It was termed “sor” for short open reading frame protein and categorized as an accessory protein, since Vif deletion viruses were still able to replicate in certain cell lines (Kan et al. 1986; Lee et al. 1986; Sodroski et al. 1986; Strebel et al. 1987). These cell types were called permissive cells. However, it was found that in other cell types (non-permissive cells) virions produced in the absence of Vif were approximately 1,000 times less infectious than those produced from wild-type HIV-1 (Sheehy et al. 2002; Madani and Kabat 1998; Gabuzda et al. 1992; Simon et al. 1998). Permissive cell types included HeLa, HEK 293T, SupT1, and CEM-SS cell lines. On the other hand, primary human T-lymphocytes, macrophages, H9, and CEM cells were shown to be non-permissive, and thus incapable of producing infectious virions (Gabuzda et al. 1992; Simon et al. 1998; von Schwedler et al. 1993). Eventually, cell fusion experiments showed that this non-permissive phenotype was dominant. Permissive 293T cells were fused with the non-permissive HUT78 cells, resulting in virions that were less infectious than those produced from 293T cells alone (Madani and Kabat 1998; Simon et al. 1998). However, the particular restriction factor in these so-called non-permissive cells was not identified until 2002, when Sheehy et al. performed further experiments with two genetically related cell lines, the permissive CEM-SS line, and the parental non-permissive cells CEM T-cell line it was generated from (Sheehy et al. 2002). Comparison of cDNAs generated from these two cell lines consistently produced an approximately 1.5-kb cDNA segment that was expressed in all non-permissive cell lines tested, but not in permissive cell lines. This protein was identified as CEM15 or human cytidine deaminase apolipoprotein B mRNA-editing catalytic polypeptide-like 3G (APOBEC3G, A3G), a member of the APOBEC family of cytidine deaminases. Later studies showed that A3F (Bishop et al. 2004; Liddament et al. 2004; Wiegand et al. 2004; Zheng et al. 2004; Dang et al. 2006), A3B (Bishop et al. 2004; Dang et al. 2006; Yu et al. 2004a; Bogerd et al. 2006a, b) A3DE (Dang et al. 2006), and A3H (Dang et al. 2006; Tan et al. 2008; OhAinle et al. 2008), related deaminase proteins, are also capable of restricting Vif-deficient HIV-1.

3 The APOBEC Family of Cytidine Deaminases

The APOBEC family of cytidine deaminases is a large family of proteins with a conserved zinc-coordinating deaminase motif, H-X-E-(X)_27–28-P-C-X_2–4-C, that is capable of converting cytosine to uracil in RNA or DNA. This family includes APOBEC1, expressed mainly in the small intestine (Teng et al. 1993), and activation-induced cytidine deaminase (AID), a B cell-specific protein, both of which are found on chromosome 12. The family also includes APOBEC2 on chromosome 6, which is highly expressed in muscle tissue (Liao et al. 1999) and APOBEC4 on chromosome 1, reported to be expressed primarily in testes (Rogozin et al. 2005). APOBEC1 is responsible for the C-to-U editing of apolipoprotein B (apo-B) mRNA, and is catalyzed by a multiprotein complex that recognizes an 11-nucleotide sequence downstream of the editing site (Turelli and Trono 2005; Harris and Liddament 2004). AID on the other hand edits single-stranded DNA, and is required for somatic hypermutation (SHM), class switching recombination, and gene conversion of immunoglobulin genes (Turelli and Trono 2005; Harris and Liddament 2004). The cellular functions of the APOBEC2 and APOBEC4 proteins are poorly understood. Much is known on the other hand about the 7 members of the APOBEC3 family: A3A, A3B, A3C, A3DE, A3F, A3G, and A3H. These proteins are found on chromosome 22 (Jarmuz et al. 2002), and are thought to have been generated by multiple, and relatively recent, duplication events. In fact, rodents possess only one APOBEC3 gene which has been shown to be inessential for mouse development, fertility or survival (Mikl et al. 2005). It is thought that this single APOBEC3 gene expanded in primates 40–100 million years ago (Jarmuz et al. 2002). It is significant that several APOBEC3 genes appear to be affected by positive selection pressure, as evidenced by the accumulation of non-synonymous mutations (Sawyer et al. 2004; Zhang and Webb 2004; Conticello et al. 2005). One explanation for this high level of selection pressure is that these proteins evolved as a defense against endogenous retroelements, since the expansion of APOBEC3 proteins in primates appears to coincide with a sharp decline in retrotransposition rates in humans (Waterston et al. 2002). Indeed, it has been shown that several APOBEC3 members are capable of inhibiting a variety of endogenous retroelements as well as various retroviruses.

Although A3G was the first anti-HIV-1 deaminase identified, A3F, A3B, A3DE, and A3H can all inhibit Vif-deficient HIV-1. And although A3C has little effect on HIV-1, it is extremely effective against Vif-deficient simian immunodeficiency virus (SIV) (Yu et al. 2004a). In addition, several APOBEC3 proteins can also inhibit a variety of retroviruses such as mouse mammary tumor virus (MMTV), murine leukemia virus (MLV), human T-cell leukemia virus (HTLV), equine infectious anemia virus (EIAV), Rous sarcoma virus (RSV), Feline leukemia virus (FIV), as well as various foamy viruses (FV). Several APOBEC3 proteins can also inhibit the partial double-stranded DNA virus, Hepatitis B (HBV) ,as well as the single-stranded DNA virus, adeno-associated virus (AAV). And finally a recent study presents evidence that may extend the antiviral repertoire of APOBEC3 proteins to double-stranded DNA viruses, such as human papillomavirus (HPV). Several APOBEC3 proteins can also inhibit various retroelements, including both long terminal repeat (LTR) retrotransposons such as Ty1, IAP, MusD, and HERV-K, as well as non-LTR retroelements such as human LINE-1 and Alu. Table 1 offers a comprehensive list of the APOBEC family of proteins and their known effects on various microorganisms.

Table 1 A comprehensive list of viruses and retroelements that have been found to be affected by various APOBEC family members

4 The Antiviral Mechanism of APOEBC3 Proteins: Deaminase Dependent

APOBEC3 proteins have been reported to have various inhibitory effects on multiple viruses, as mentioned above. For the purpose of simplicity, we will focus mainly on the antiviral effects of A3G on HIV-1, the most well-studied virus. Specifically, A3G has been reported to inhibit: (1) the accumulation of viral DNA (Mangeat et al. 2003; Mariani et al. 2003; Bishop et al. 2008; von Schwedler et al. 1993; Bishop et al. 2006; Holmes et al. 2007; Simon and Malim 1996); (2) the accumulation of two-LTR circle DNA (Luo et al. 2007; Anderson and Hope 2008); and (3) proviral DNA formation (Luo et al. 2007; Miyagi et al. 2007) (Fig. 1).

Fig. 1

A3G in the cell. A3G can be packaged into newly budding virions, which when new infections take place in target cells (right), can edit newly reverse transcribed ssDNA. A3G may also have deaminase-independent functions in producer cells (left), such as inhibition of reverse transcription, strand transfer interactions, and DNA strand elongation

APOBEC3 proteins, like AID, prefer to edit single-stranded DNA. In particular, A3G edits the newly reverse transcribed minus-strand DNA of HIV-1, where it induces C-to-U mutations (Zhang et al. 2003; Mangeat et al. 2003; Lecossier et al. 2003; Harris et al. 2003; Mariani et al. 2003; Yu et al. 2004b; Suspene et al. 2004). Uracil-DNA glycosylases, such as UNG2 or SMUG1, may then generate abasic sites, which in turn may inhibit plus-strand DNA synthesis (Cai et al. 1993; Klarmann et al. 2003; Yang et al. 2007a), or even trigger DNA degradation by cellular endonucleases such as apurinic-apyrimidinic endonucleases (Harris et al. 2003; Yang et al. 2007a). Several laboratories have shown that host UNG-2 can be packaged into HIV-1 virions, via its interaction with other HIV-1 proteins, such as the viral protein R (Vpr) (Mansky et al. 2000) or integrase (IN) (Willetts et al. 1999; Priet et al. 2003). Alternatively, if the minus-strand uracil-containing viral DNA serves as a template for synthesizing the sense-strand DNA, the G-to-A-hypermutated viral DNA may generate premature stop codons or heavily mutated non-functional viral proteins.

Others, however, believe that UNG-2 is dispensable for the antiviral activity of A3G. One group used a bacteriophage protein, uracil DNA glycosylase inhibitor (UGI), in order to demonstrate that the levels of HIV-1 infectivity and A3G restriction were not affected, regardless of the presence or absence of UNG catalytic activity (Kaiser and Emerman 2006; Mbisa et al. 2007). Similar results were obtained with HBV and RSV systems (Nguyen et al. 2007; Langlois and Neuberger 2008). Nevertheless, it cannot be ruled out that Vpr may degrade UNG2 and SMUG1 (Schrofelbauer et al. 2005), thus overcoming the problem of abasic site generation. Alternatively, two other glycosylases, MBD4 and TDG, exist, both capable of acting on dsDNA (Barnes and Lindahl 2004). Neither of these proteins has been studied in the context of A3G antiviral activity. However, deamination may not be the only antiviral mechanism of APOBEC3 proteins.

5 The Antiviral Mechanism of APOEBC3 Proteins: Deaminase Independent

Several groups have shown that A3G and A3F deaminase mutants still retain some anti-HIV-1 activity (Newman et al. 2005; Bishop et al. 2006; Holmes et al. 2007). The first of these deaminase-independent mechanisms is inhibition of reverse transcription (Guo et al. 2006, 2007; Yang et al. 2007b). One group showed that much of the early inhibition of viral DNA production induced by A3G correlated with the inhibition of early minus-sense strong stop DNA, and the inability of tRNA ^Lys₃ to prime reverse transcription (Guo et al. 2006). Later, the same group also used an in vitro system to show that A3G decreased the ability of tRNA ^Lys₃ to bind to viral RNA and initiate reverse transcription (Guo et al. 2007). Similar results were obtained with A3F (Yang et al. 2007b).

The second mechanism described is the inhibition of strand transfer reactions (Mbisa et al. 2007; Li et al. 2007). Li et al. reported that A3G-induced inhibition of both minus- and plus-strand transfer in reverse transcription was responsible for the majority of the reduction in late DNA synthesis. Mbisa et al. showed that a deaminase-defective A3G resulted in defects not only in plus-strand DNA transfer and integration, but also in primer tRNA processing.

Finally, it was also shown that A3G can inhibit reverse transcriptase-catalyzed DNA elongations (Iwatani et al. 2007). Iwatani et al. used purified A3G, NC and RT in an established in vitro system to study reverse transcription. They found that A3G could inhibit all reverse transcriptase (RT)-catalyzed DNA elongation reactions, but did not have any effect on RNase H activity or NC's activities as a nucleic acid chaperone.

Interestingly, although A3G normally exerts its effects by inhibiting new infections after being packaged into newly budded viruses, it may also exert a post-entry block for HIV-1 in resting T cells (Chiu et al. 2005) and dendritic cells (Kreisberg et al. 2006). While activated CD4+ T lymphocytes are highly permissive to wild-type HIV-1 infection, resting PBMCs appear to be resistant to infection (Chiu et al. 2005). Chiu et al. showed that A3G in activated CD4+T cells exists in high-molecular mass ribonucleoprotein complexes (HMM), while A3G in resting CD4+ T cells is predominantly found in an enzymatically active low molecular mass (LMM) form. This LMM form of A3G is thought to cause a post-entry block early in the HIV-1 replication cycle, by impairing reverse transcription (Chiu et al. 2005). These results were reproducible in both naive and memory T cells (Kreisberg et al. 2006), as well as macrophages and dendritic cells (Stopak et al. 2007). Activation of these resting cells with various mitogens and cytokines caused a shift in A3G from LMM to HMM complexes, which also correlated with increased susceptibility to HIV-1 infection (Chiu et al. 2005; Stopak et al. 2007). That this post-entry block is caused by A3G is further supported by data showing that siRNA directed against A3G in unstimulated CD4+ T cells increased cell permissivity to HIV-1 (Chiu et al. 2005).

In addition, sequencing of reverse transcripts from infected cells showed only low levels of G-to-A hypermutation, suggesting that this early restriction is deaminase-independent. Nonetheless, this restriction is not absolute, as newly synthesized viral DNA was detected in resting CD4+ T cells after a 24- to 48-h delay (Chiu et al. 2005). Inactivated T cells was also reported by others to be permissive to HIV-1 infection in some cases (Stevenson et al. 1990; Watson and Wilburn 1992).

In addition, several APOBEC3 proteins appear to inhibit HBV, AAV, IAP, MusD, and LINE-1 in a deaminase-independent manner, which has yet to be described (Bogerd et al. 2006a; Chen et al. 2006; Stenglein and Harris 2006; Muckenfuss et al. 2006), supporting the idea of alternative, and deaminase-independent inhibition mechanisms.

However, there is still no clear consensus on these deaminase-independent activities, with some arguing that these effects may be just a result of protein overexpression to levels far above normal physiological concentrations. One group showed that active site mutants of A3G had no antiviral activity when expressed at levels similar to those observed in primary human T cells (Holmes et al. 2007; Miyagi et al. 2007; Mbisa et al. 2007; Schumacher et al. 2008). Another showed that the catalytic site of A3G and DNA cytosine deamination is important in inhibiting Ty1, MusD, and HIV-1 when expressed at near-physiologic levels (Schumacher et al. 2008). Nevertheless, this does not rule out the deaminase-independent mechanisms described above, as the active site mutants tested may also block important protein functions such as protein–protein or protein–DNA/RNA interactions.

6 A3G Structural Features

To better understand the structural features of APOBEC3 proteins that are important for relationships with viral proteins such as HIV-1 Vif and NC, and in order to inhibit these interactions and thus achieve our ultimate goal of preventing disease, it is essential to have high resolution structures of these proteins.

Three recent reports on the structure of the C-terminal cytidine deaminase domain (CDD) have highlighted several of these structural features. The first structure was resolved by nuclear magnetic resonance (NMR) (Chen et al. 2008). The second more recent structure was resolved by x-ray crystallography (Holden et al. 2008). The second report confirmed many of the structural features and characteristics first reported by Chen et al. but also reported several important differences. While the crystal structure had a five β-sheet core structure surrounded by six α-helices, the NMR structure was reported to have the same core β-structure, but was surrounded by only five α-helices. Furthermore, the A3G crystal structure was reported to have a long and well-defined β2 strand, while the same stretch of amino acids in the NMR structure was short, and interrupted by a 6-residue bulge. In addition, the two active center loops located near the active site were found at different positions. However, the most important difference was in the proposed nucleic acid substrate-binding area. While the crystal structure had a deep, well-defined groove that ran alongside the AC loops and an active site where the cytosine substrate is thought to be deaminated, the NMR structure did not have this groove. Instead, the authors noticed several positively charged residues surrounding the active site region. Several of these amino acids overlap with residues found in the crystal structure groove. These structural differences could be attributed to several mutations made in the A3G construct that was used to obtain the NMR structure. However, these differences may also be a result of the different methodologies used for isolating the protein and/or structural analysis.

Finally, the most recent report (Furukawa et al. 2009), also presents the structure of the wild-type CDD of A3G, resolved in solution by NMR. While this latest structure shares some similarities with the crystal structure (such as the presence of a sixth alpha helix), it also has some features in common with Chen et al.’s NMR structure (such as the second interrupted beta strand). The convergence for the two active center loops near the active site was poor and these elements are not well defined in this structure. In addition, the authors presenting this third structure identified yet another position for the binding of the ssDNA substrate. They propose a model where the ssDNA is positioned along the two alpha helices α1 and α2. Clearly, additional experimental studies, in particular mutational assays will be necessary to determine which, if any, of these three models is correct.

Future studies in this field will most assuredly include attempts to crystallize the N-terminal deaminase domain of A3G, as well as the entire molecule. Furthermore, molecular modeling studies of other deaminase proteins based on this structure may shed light on their different substrate-binding and antiviral specificities.

7 Packaging of APOBEC3 Molecules

In order for APOBEC3 molecules to be effective in inhibiting HIV-1, they need to be effectively packaged into newly budded virions. Studies of A3G packaging requirements revealed that A3G binds to the HIV-1 Gag protein, and in particular to the nucleocapsid (NC) region (Cen et al. 2004; Khan et al. 2005; Schafer et al. 2004; Alce and Popik 2004; Zennou et al. 2004; Luo et al. 2004). Several groups have reported that HIV-1 Gag is necessary and sufficient for A3G packaging; however, these experiments were performed with virus-like particles rather than whole virus, which may have different requirements (Cen et al. 2004; Schafer et al. 2004; Alce and Popik 2004; Svarovskaia et al. 2004; Douaisi et al. 2004). Most groups, however, agree that the efficiency of A3G packaging is significantly enhanced by RNA interactions (Svarovskaia et al. 2004). Although A3G appears to edit only ssDNA, it is nevertheless capable of strongly binding to RNA (Jarmuz et al. 2002; Yu et al. 2004b; Khan et al. 2005; Iwatani et al. 2006; Khan et al. 2007; Wang et al. 2007a; Tian et al. 2007). The NC region of Gag is also important for this interaction with RNA, and, together with A3G, may form a protein:RNA complex (Iwatani et al. 2006; Burnett and Spearman 2007).

In the case of A3G, it is the N terminal cytidine deaminase that is thought to be responsible for the nucleic acid binding properties that contribute to packaging, while the C terminal domain is generally thought to be responsible for deamination (Newman et al. 2005; Iwatani et al. 2006; Navarro et al. 2005; Hache et al. 2005). Several groups have identified specific residues in the N-terminal region of A3G that were shown to affect packaging, in particular, W127 (Huthoff and Malim 2007), which was also shown to be important for RNA-binding.

Although many agree that an RNA interaction is important for efficient packaging, the nature of this RNA is still highly debated. One proposed candidate is the HIV-1 viral genomic RNA, where studies showed that, although A3G could be successfully packaged in the absence of viral RNA, these A3G molecules were not associated with the viral cores of the virions (Khan et al. 2005,2007). A3G molecules were able to be re-associated with the viral cores by the addition of viral RNA in trans. Another RNA candidate is a cellular RNA, and a component of signal recognition particles (SRP), 7SL RNA (Wang et al. 2007a). Indeed, 7SL RNA has been reported to be both abundantly and selectively packaged in HIV-1 virions (Khan et al. 2007; Wang et al. 2007a; Onafuwa-Nuga et al. 2006). Both A3G and NC mutants that were known to have packaging defects, also showed a reduced ability to bind to and package 7SL RNA (Wang et al. 2007a; Bach et al. 2008). Experiments involving overexpression of SRP19, a known 7SL-binding protein, in increasing concentrations in order to reduce the level of free 7SL RNA found in the cell were performed. This overexpression was associated with reduced levels of A3G packaging (Wang et al. 2007a), further supporting the idea that 7SL RNA contributes to the A3G packaging process. Others, however, believe that, although 7SL RNA can bind both A3G and NC, it is not an essential factor for the A3G packaging process (Bach et al. 2008). And although regions in both NC and A3G have been shown to be important for A3G packaging, it is not known whether this interaction is direct or if their interaction could be mediated by other molecules, such as 7SL RNA. This field is still far from a consensus, and more research is needed in order to resolve these discrepancies.

In addition to the mechanism of packaging, the location of packaging is also an important factor for antiviral activity. A3A, another member of the APOBEC3 family, has a potent inhibitory effect on several retroelements (Bogerd et al. 2006b; Chen et al. 2006), but has no effect on HIV-1 (Bishop et al. 2004). However, when A3A was targeted to the viral core, by creating a Vpr-A3A fusion protein, this cytidine deaminase was able to restrict both HIV-1 and SIV in a Vif-independent manner (Aguiar et al. 2008).

However, not all APOBEC3 proteins are packaged by the same mechanism. APOBEC3C, which efficiently restricts SIV but not HIV (Yu et al. 2004a), can nevertheless be packaged into both. A3C also interacts with HIV-1 Gag protein, but unlike A3G, A3C is packaged through a RNA-dependent and NC-independent fashion (Wang et al. 2008). Thus, it appears that individual APOBEC3 proteins have evolved to use different mechanisms for targeting retroviruses, possibly to broaden the range of viruses targeted.

8 Vif Targets APOBEC3 Proteins for Proteasome-Mediated Degradation

Although APOBEC3 proteins may be a formidable weapon in the cell’s antiviral arsenal, several retroviruses have developed their own counter measures that inhibit HIV. The SIV and HIV-1 lentiviruses encode the Vif protein, which can induce the polyubiquitination and degradation of multiple APOBEC3 molecules (Yu et al. 2003; Mehle et al. 2004; Stopak et al. 2003; Marin et al. 2003; Conticello et al. 2003; Sheehy et al. 2003; Liu et al. 2004, 2005). Vif proteins assemble with Cul5, ElonginB, ElonginC, and Rbx1 proteins to form an E3 ubiquitin ligase (Yu et al. 2003; Mehle et al. 2004; Liu et al. 2005; Yu et al. 2004c; Luo et al. 2005; Kobayashi et al. 2005) (Fig. 2).

Fig. 2

The HIV-1 Vif-Cul5 E3 ubiquitin ligase complex; a model of Vif-induced APOBEC3 degradation. Vif binds ElonginC through a viral specific BC-box; ElonginC also interacts with ElonginB. The Vif-ElonginC-ElonginB complex also binds to Cul5, forming the Cul5-EloB/C-Vif E3 ligase. Cul5 also binds Rbx which recruits the E2-ubiquitin conjugating enzyme. Ubiquitin activated by E1 enzyme is transiently transferred to the E2 enzyme, then to the APOBEC3 target molecule. Polyubiquitination of APOBEC3 molecules takes place, followed by degradation by the 26S proteasome

E3 ubiquitin ligases are critical for regulating cellular processes such as mitosis and the cell cycle through targeted protein degradation (Pickart 2004). It is members of the E3 ubiquitin ligase family, such as the cullin-based E3 ligases that mediate protein degradation specificity. These cullins can then form a scaffold on which other E3 ligase protein components assemble and convey the substrate to the E2 ubiquitin-conjugating enzyme. The Vif binding cullin, Cul5, is commonly associated with the ElonginC and ElonginB adaptor molecules. ElonginC recognizes substrate receptor proteins containing a BC-box. The SLQxLA motif in Vif is highly conserved in lentiviral Vifs, and forms a virus-specific BC-box motif that mediates the interaction with ElonginC, which in turn interacts with both ElonginB and Cul5 (Figs. 2 and 3). Vif in turn binds to Cul5 through two other highly conserved sites, the Hx₅Cx_17-18Cx_3-5H motif (Luo et al. 2005; Xiao et al. 2007; Mehle et al. 2006) and a LPx₄L motif downstream (Stanley et al. 2008) (Fig. 3). The first motif is responsible for binding zinc, which stabilizes the molecule, and the second is a highly conserved hydrophobic interface that mediates Cul5 selection.

Fig. 3

APOBEC3 and Vif interaction domains. Various domains for interaction of Vif and APOBEC3 with each other as well as other proteins have been mapped. A3F and A3G have different binding requirements on Vif

In terms of Vif binding to APOBEC3 proteins, several regions have been described in both A3G and A3F, and will be further discussed below. It is also of interest that A3C and A3DE, two APOBEC3 proteins with weaker effects on HIV-1, are also ubiquitinated and degraded by Vif (Dang et al. 2006; Zhang et al. 2008a). And although A3B may also inhibit HIV-1, it is not degraded by Vif, possibly because of its low expression level in T cells.

9 Vif May also Inhibit A3G by Degradation-Independent Mechanisms

Several lines of evidence point to the existence of degradation-independent mechanisms of A3G inhibition in Vif. Recently, a degradation-resistant variant of A3G was identified (Opi et al. 2007). An A3G mutant with a single point mutation at position 97 was found to be defective in multimerization, and at the same time to be impervious to Vif-induced proteasomal degradation, although still capable of binding to Vif (Opi et al. 2007). Surprisingly, although this A3G mutant was not degraded, Vif was still able to prevent its encapsidation into HIV-1 virions, as well as inhibit its antiviral activity (Opi et al. 2007). In fact, one report suggests that Vif’s abilities to establish the production of infectious virus and to degrade APOBEC3 proteins are completely separable functions (Kao et al. 2007). In this study, several different forms of tagged HIV-1Vif were studied. It was found that although Vif expressed from a proviral vector was less efficient at degrading A3G than Vif expressed from codon-optimized vectors, it still was able to efficiently inhibit A3G activity (Kao et al. 2007). In addition, although a YFP-tagged form of Vif was able to efficiently degrade A3G, it was unable to restore viral infectivity (Kao et al. 2007). Further support for this theory comes from Vif mutational studies. Although a point mutation of a serine at position 144 in Vif resulted in reduced levels of viral infectivity, this mutant was found to be able to degrade A36 effectively and efficiently (Mehle et al. 2004). Moreover, HIV-1 is capable of inhibiting the deamination activity of both A3G (Santa-Marta et al. 2004) and AID (Santa-Marta et al. 2007) in bacterial E. coli systems, which do not contain any proteasomal-degradation machinery.

The details and mechanism of this degradation-independent inhibition are not yet known. Two theories that have been proposed are that: (1) Vif competitively binds to a common Vif/A3G binding element that interferes with A3G packaging (Goila-Gaur et al. 2008a); or (2) Vif promotes the shift of A3G from LMM to HMM states (Goila-Gaur et al. 2008a, 2008b). A recent paper also notes that although newly synthesized A3G and stable pre-existing A3G are packaged with the same efficiency into virions, HIV-1 will preferentially degrade newly synthesized A3G (Goila-Gaur et al. 2008b). It is thus possible that Vif may have to use alternative methods to inhibit the action of pre-existing A3G in the cell. Again, this is another area of the Vif:APOBEC field that is ripe for further exploration.

10 Specificity of the Vif:APOBEC Interaction

As discussed above, several important structural elements exist in Vif that allow interactions with various molecules. In addition to the structural elements of Vif necessary for interaction with APOBEC3 proteins, there also exist certain domains on APOBEC3 proteins, important for interactions with Vif. For example, it is known that it is the N-terminal cytidine deaminase domain of A3G which interacts with Vif (Conticello et al. 2003) (Fig. 3). In contrast, it was noticed that adding tags to the C-terminal of A3F induced resistance to degradation by Vif, which was later shown to be due to reduced interaction with Vif (Tian et al. 2006). In fact, the C-terminal of A3F alone is sufficient for interaction with, and degradation by, Vif (Zhang et al. 2008b). Recently published data further specify the region of A3F necessary for interaction with Vif, and map it to an area between amino acids 283–300 (Russell et al. 2008). Furthermore, the authors also show that this stretch of amino acids was sufficient for both interaction with, and degradation by the Vif protein. Therefore, although interaction with Vif maps to the N-terminal of A3G, it appears to be the C-terminal domain that is important for this interaction in A3F (Fig. 3).

Studies to map the specific regions in Vif that are important for A3G and A3F binding revealed that various amino-terminal domains of HIV-1 Vif molecules are involved in distinct substrate APOBEC3 recognitions (Marin et al. 2003; Goila-Gaur et al. 2008b; Tian et al. 2006; Simon et al. 2005; Schrofelbauer et al. 2006; Russell and Pathak 2007; Mehle et al. 2007; He et al. 2008). A region in HIV-1 Vif, spanning a stretch of amino acids from 22 to 44, was found to be important for the suppression of A3G but not A3F (Simon et al. 2005; Russell and Pathak 2007; Mehle et al. 2007). This region is known as the G-box (Fig. 3). On the other hand, several amino acids from 11 to 17 and 74 to 79 of HIV-1 Vif were found to be important for the suppression of A3F but not A3G (Tian et al. 2006; Simon et al. 2005; Schrofelbauer et al. 2006; Russell and Pathak 2007; He et al. 2008) (Fig. 3). Yet another region, from amino acids 52 to 72, was found to be important for both A3G and A3F suppression (He et al. 2008) (Fig. 3). In addition to A3G and A3F, other APOBEC3 family members, such as A3A, A3B, A3C, and A3DE, can also bind to HIV-1 Vif (Dang et al. 2006; Zhang et al. 2008a; Marin et al. 2008). And although A3A and A3B are not thought to be normally degraded by HIV-1 Vif even though they are re-localized by it, certain Vif variants from different HIV-1 strains were capable of degrading them (Marin et al. 2008). However, little is known about how these proteins are recognized by HIV-1 Vifs of different HIV-1 strains, and further exploration is needed.

In addition to the many Vif regions that determine specificity of binding to various APOBEC3 and cellular proteins, various Vifs can also exert species-specific selectivity on the proteins they degrade. For example, HIV-1 Vif can effectively inhibit human A3G, and African green monkey SIV (SIV_Agm) Vif is able to inhibit Agm A3G. However, HIV-1 Vif is unable to recognize and inhibit simian (Agm or rhesus macaque) A3G. Vice versa, SIV_Agm Vif cannot inhibit human or macaque A3G. On the other hand, the Vif protein from macaque-specific SIV (SIV_Mac) can counteract all three A3Gs from humans, Agms, and rhesus macaques (Mariani et al. 2003). It was shown that this species specificity was conferred by a single amino acid, an aspartate at position 128 in A3G. Several independent studies found that this single amino acid residue at position 128 in human A3G was responsible for the species specificity. Changing this residue to its equivalent in Agms (D128K) was enough to reverse the specificity of HIV-1 and SIV_Agm Vifs for the binding and degradation of their respective human and Agm A3Gs (Bogerd et al. 2004; Schrofelbauer et al. 2004; Mangeat et al. 2004; Xu et al. 2004). In contrast, mutations at the equivalent position in A3F, at amino acid 127, had no influence on Vif suppression (Liu et al. 2005).

Finally, it is of interest that phosphorylation of both Vif and A3G can play a role in modulating these interactions. Vif can be phosphorylated at several serine and threonine residues: T96, S144, and T188 (Yang and Gabuzda 1998; Yang et al. 1996). Mutation of these conserved residues does not affect Vif-A3G binding, or A3G degradation (Mehle et al. 2004); however, mutations preventing phosphorylation of S144 in the Vif BC-box significantly decreased Vif function by inhibiting the Vif-ElonginC interaction (Mehle et al. 2004), and thus hindering the process of proteasomal degradation. In addition, a recent report has proposed that A3G is also capable of being phosphorylated (Shirakawa et al. 2008).

Shirakawa et al. showed that protein kinase A (PKA) can bind to and phosphorylate A3G at a threonine at position 32. Phosphorylation of A3G at this position appears to reduce Vif–A3G interaction, thus reducing levels of A3G ubiquitination and degradation. Computer modeling and mutagenesis studies were also used to study an interaction between two amino acids in A3G (T32 and R24) that are proposed to be important for binding to and subsequent degradation by Vif. Thus, phosphorylation may be yet another mechanism whereby protein interaction between Vif, A3G, and proteasomal degradation machinery can be regulated.

11 Vif and Cell Cycle Inhibition

In addition to the antiviral effects Vif exerts through degradation of APOBEC3 proteins, several reports describe a Vif-induced cell cycle delay (Sakai et al. 2006; Wang et al. 2007b; Dehart et al. 2007). Vpr is normally thought to be the HIV-1 protein responsible for the G ₂ cell cycle arrest that has been observed in CD4+ T cells (He et al. 1995; Jowett et al. 1995; Stewart et al. 1997). Unexpectedly, though, studies in HIV-1-infected cells in the absence of Vpr showed an accumulation of cells at the G ₂ phase, and further studies in the presence or absence of Vif confirmed that this phenotype was Vif dependent (Sakai et al. 2006). Additional experiments showed that expression of the Vif protein alone in the absence of HIV-1 infection was enough to increase the ratio of G ₂ :G ₁ cells (Wang et al. 2007b). Moreover, another recent report showed that the G ₂ arrest phenomenon was actually a cell cycle delay, and that Vif uses the same machinery to induce cell cycle delays that it does for APOBEC3 degradation (DeHart et al. 2008). DeHart et al. showed that a Cul5 E3 ligase was required for this phenotype, regardless of the presence or absence of APOBEC3 proteins. However, although the presence of Vif resulted in an accumulation of cells in the G ₂ phase, when followed over a period of several days, it was shown that Vif did not inhibit cell division or reduce the number of dividing cells (DeHart et al. 2008). The degraded substrate responsible for the G2 ₂ disruption has not yet been identified.

12 Other Roles for Vif

In addition to APOBEC3 and proteasomal-degradation machinery proteins, several laboratories have reported that Vif also binds to many other cellular proteins. Ku70, a cellular protein involved in DNA double-strand break repair, and a component of HIV-1 pre-integration complexes (Li et al. 2001), was shown to bind to HIV-1 Vif in a yeast two-hybrid screen of a human lymphocyte cDNA library (Madani et al. 2002). Ku70, together with another protein, Ku80, form a complex (Ku70/80) that can function as a single-stranded DNA-dependent helicase. Therefore, Vif may recruit Ku70 early in the HIV-1 replication cycle to aid in the integration process of the HIV-1 PIC.

In another experiment, a glutathione S-transferase (GST) pull-down assay used to identify Vif-binding partners, showed that Vif bound specifically to the SH3 domain of Hck, a Src family tyrosine kinase (Hassaine et al. 2001). This group further showed that Hck inhibited both the production and the infectivity of HIV-1 viruses in the absence of Vif, and that expression of Hck in Jurkat cells rendered these cells less permissive to infection by Vif deletion viruses. Hck is present in monocytes but not in primary restrictive T cells (Hassaine et al. 2001). Interestingly, Hck also binds to the HIV-1 Nef protein (Saksela et al. 1995), and dominant negative forms of Hck have been shown to block HIV-1 Nef-induced MHC class I downregulation (Chang et al. 2001).

SP140, the nuclear speckle factor, is yet another cellular protein that has been reported to bind to Vif (Madani et al. 2002). SP140 was found to be expressed in non-permissive cells, but absent in permissive cells; however, expressing SP140 in permissive cells did not render them non-permissive. Nonetheless, expression of Vif in cells caused the dispersal of SP140 from nuclear speckles, or its retention in the cytoplasm (Madani et al. 2002). These nuclear speckles are known as PML bodies and have been reported to have functions in transcription, DNA repair, viral defense, cell stress, cell cycle regulation, proteolysis, and apoptosis (Bernardi and Pandolfi 2007).

Other proteins reported to associate with Vif include: SSAT, the spermine/spermidine N1-acetyl-transferase (Lee et al. 1999), a protein involved in polyamine metabolism, the regulation of which could affect viral RNA; cyclophilin A, a member of the peptidyl-prolyl cis-trans isomerase (PPIase) family, which has been shown to interact with many HIV-1 proteins including p55 gag, Vpr, and capsid protein. Cyclophilin A has been shown to be necessary for the formation of infectious HIV virions (Billich et al. 1995); and NVBP, a novel Vif-binding protein (Lee et al. 1999).

Although the interaction of Vif with most of these proteins was reported before Vif was known to degrade A3G, it may be worth re-examining the roles of these proteins in relation to Vif, particularly in light of the recently reported cell cycle disruptions, and the degradation-independent inhibition of A3G induced by Vif.

13 Conclusions

Although the HIV field, and in particular the APOBEC/Vif area, is a rapidly advancing one that aggressively pursues new discoveries, several important questions still remain. It is unclear what (if any) other cellular and antiviral functions APOBEC3 proteins may possess. Similarly, Vif has shown signs of being involved in processes other than APOBEC3 degradation, and these need to be further investigated. The regions and particular amino acids of APOBEC3 and Vif involved in their interaction still need to be fully mapped. In addition, high resolution structures for both Vif and full-length A3G still do not exist. Such structures and further mapping of protein interactions are essential for initiating new drug design studies.