APOBEC3 Cytidine Deaminases: Distinct Antiviral Actions along the Retroviral Life Cycle*

The field of human immunodeficiency virus (HIV) biology has been galvanized by the discovery of innate APOBEC3 cytidine deaminases, which pose powerful barriers to the replication of HIV and other retroviruses. Rapid progress has been made in defining their action, intriguing regulation within cells, expanded range of retroviral targets, and counterstrikes utilized by retroviruses against them. Although scientifically fascinating, advances in APOBEC3 biology may lead to new antiviral drugs and improved lentiviral vectors for gene therapy.


Mutagenesis by Virion-incorporated APOBEC3G
A3G packaging into HIV-1 virions involves assembly with the nucleocapsid region of the Gag viral protein and possibly binding of RNA (19 -26). A3G appears to selectively target single-stranded DNA (27,28) and yields dC to dU mutations in the viral minus-strand DNA formed during reverse transcription (RT) (5, 29 -32). These uracil-containing DNA molecules are either degraded, probably by DNA repair enzymes such as uracil DNA glycosylase (33) and apurinic-apyrimidinic endonucleases, or serve as templates for the synthesis of plus-strand DNA, yielding dG to dA hypermutated proviruses encoding altered proteins. These events potently halt viral spread. The single-stranded specificity also accounts for the increase in mutation frequency in a 5Ј to 3Ј direction (27). The 3Ј regions in the minus-strand remain single-stranded for the longest time as they await plus-strand synthesis. This antiviral activity of human A3G also extends to other lentiviruses including simian immunodeficiency virus (SIV), equine infectious anemia virus (EIAV), and even distantly related retroviruses such as murine leukemia virus (MLV) and foamy viruses (6,29,31,34,35) Other Antiviral APOBEC3 Proteins Intriguingly, other APOBEC3 family members also exert antiviral activity. A3F induces dG to dA transitions in viral plus-strand DNAs, and its action is blocked by HIV-1 Vif (18, 36 -38). However, the two enzymes differ in their consensus sequence for mutation: A3G favors 5Ј-CC dinucleotides (underline denotes the site of deamination) (27,29,31,32), whereas A3F prefers 5Ј-TC (18,36,37). Mutations in the A/T-rich proviruses present in AIDS patients bear the footprints of both A3G and A3F action. The hypermutation specificity of A3G and A3F is governed by the C-terminal deaminase domain (39). Human A3B also has moderate activity against HIV-1 and an A3F-like target site consensus. However, A3B is resistant to Vif (18,40). A3B is not expressed in many of the natural cellular targets of HIV-1 and may not normally influence HIV-1 replication in vivo. Neither A3F nor A3B is active against MLV whereas A3G is (18,41). Thus, the spectrum of viruses impaired by A3F and A3B may well be more restricted.
Vif appears to act by accelerating degradation of A3G by the 26 S proteasome and partially impairing translation of A3G mRNA (7). Vif induces polyubiquitylation of A3G and targets it for destruction by the proteasome (45,46). Vif contains two key domains, one that binds to A3G and a SLQ(Y/F)LA motif for A3G degradation via the proteasome (45). The latter is similar to a conserved sequence in the BC-box of the suppressors of cytokine signaling proteins. Vif recruits an E3 ligase complex of elongins B and C, Cul5, and Rbx1 (47). Binding of Vif to elongin C depends on the SLQ(Y/F)LA motif in Vif and two conserved cysteines (48) and is negatively regulated by serine phosphorylation in the BC-box motif (49). Dominant-negative mutants of Cul5, which block modification by Nedd8 or assembly with Rbx1, prevent the degradation of A3G and restore its antiviral activity (47). Vif also induces proteasomal degradation of A3F (36 -38, 50).
Unlike A3G, the viral protection conferred by Vif appears highly species- specific (6), For example, although Vif from the SIV of African green monkey (SIVagm) triggers the degradation of its own host's A3G, it fails to neutralize human or chimpanzee A3G. Similarly, HIV-1 Vif does not impair the action of African green monkey or rhesus macaque A3G. These species-specific effects are likely key to the species tropisms of retroviruses and may thwart many cross-species transmissions. Remarkably, this apparent specificity is regulated by amino acid 128 in A3G (aspartic acid in humans and lysine in African green monkey) (51)(52)(53)(54). Interestingly, Vif of SIV from rhesus macaque (SIVmac), and possibly sooty mangabey (SIVsm), neutralizes A3G orthologs in most of the other species, including humans (6, 54), a compelling correlation with the fact that SIVsm was zoonotically transmitted to humans and gave rise to HIV-2.
Although available data argue that Vif-induced degradation of A3G is key, Vif might also physically exclude A3G from virions by sequestering it from sites of viral budding (6,42). Moreover, the intracellular depletion of A3G and rescue of viral infectivity might be functionally separable activities of Vif (49,55). For example, infectious HIV viruses can be produced by cells where Vif only moderately alters the steady-state A3G levels (55). The S144A mutation in Vif, which prevents phosphorylation at this site, compromises infectivity of progeny virions but induces A3G degradation (49). Thus, Vif phosphorylation may regulate another Vif function that confers full viral infectivity. Such ancillary actions of Vif deserve continued investigation.

Counterstrikes by Other Susceptible Viruses
Other retroviruses have evolved different escape mechanisms. Several APO-BEC3 proteins potently inhibit the infectivity of viral vectors based on primate foamy viruses (PFV). The packaging of APOBEC3 into PFV virions induces dG 3 dA hypermutations in PFV reverse transcripts. PFV Bet appears to function as the foamy virus analogue of HIV-1 Vif preventing the incorporation of the human A3G/A3F proteins into progeny virions. However, Bet does not deplete human A3G/A3F in virion producer cells (35). Similar findings have emerged in studies of a distantly related feline foamy virus (34). Another example is provided by MLV, which replicates effectively in murine APO-BEC3-positive cells yet does not encode a vif-or a bet-like gene. The MLV Gag protein has evolved to lack binding, and hence virion packaging, of the cognate murine APOBEC3. However, MLV infectivity is still restricted by heterologous APOBEC3 proteins that retain the ability to interact with MLV Gag (41,56). The lentivirus EIAV, also devoid of a Vif but sensitive to human A3G, may have evolved another mechanism for evading A3G orthologs in its natural hosts. It may replicate in cells lacking the enzyme, but an occult Vif-like activity or a Gag protein invisible to the deaminase cannot be excluded. EIAV, similar to feline immunodeficiency virus, also encodes a dUTPase that may provide a separate pathway to protect the virus from the effects of deamination.

APOBEC3G Restricts HIV-1 Post-entry in Resting CD4 T Cells
How can ⌬Vif virions from permissive cells infect nonpermissive cells laden with cytoplasmic A3G but escape its potential antiviral effects? The answer emerged in studies of the cellular A3G protein. A3G exists in two different sized forms, an enzymatically active low molecular mass (LMM) and an inactive high molecular mass (HMM) ribonucleoprotein complex (Fig. 1). In cells susceptible to HIV infection, such as H9 T cells, A3G is sequestered in an inactive HMM complex that is unable to block replication of incoming virions. Vif targets this HMM A3G complex for polyubiquitylation by recruitment of key E3 ligase components, including Cul5 and elongin B (57).
Strikingly, A3G exists only in the LMM form in peripheral blood-derived resting CD4 T cells and monocytes (57), both of which are refractory to HIV-1 infection (58 -62). LMM A3G is converted to a HMM complex when resting CD4 T cells are activated or when monocytes are induced to differentiate into macrophages. This change correlates with increased susceptibility of these cells to HIV-1 infection. When expression of the LMM form in resting CD4 T cells was "knocked down" by RNA interference, HIV-1 infection occurred (57). Thus, LMM A3G is a potent post-entry restriction factor for HIV-1 in peripheral blood-derived resting CD4 T cells and likely in freshly isolated monocytes.
RT intermediates of incoming virions may be substrates for LMM A3G. Alternatively, LMM A3G (8) may target genomic RNA-primer tRNA complexes and impair the RT reaction, leading to a kinetic delay in the accumulation of late RT products, which can be rescued by knockdown of LMM A3G (57). Only low levels of dG to dA hypermutations occur in the reverse transcripts of resting CD4 T cells (57). Thus, A3G-restricting activity may not depend strictly on deamination.
Anti-HIV-1 activity also persists for select A3G mutants lacking the C-terminal active enzymatic site (63). Their retained active N-terminal sites might mediate RNA binding. Several questions remain. How does the LMM A3G complex block HIV-1 infection in resting CD4 T cells? Is it analogous to the way that A3G mutants lacking cytidine deaminase activity impair HIV infection? Can the restricting activity of LMM A3G be recapitulated in cells already harboring HMM A3G by expression of A3G variants that cannot be recruited into the HMM complex? Because the RT block in resting CD4 T cells is not completely relieved by small interfering RNA-mediated knockdown, and A3G and A3F signature hypermutations are detectable in a minor portion of the viral cDNA sequences in resting CD4 T cells (57), are other APOBEC3 members also regulated in the same manner as A3G and can they similarly participate in the post-entry restriction block encountered in resting CD4 T-cells?
How is A3G incorporated into virions? Assembly of A3G with viral Gag (19 -26) or genomic HIV-1 RNA (20, 22) appears important. Do these factors promote release of A3G from the cellular HMM complex and lead to its incorporation as a fully active LMM enzyme complex? Alternatively, A3G might be incorporated as an inactive HMM complex perhaps bound to and inactivated by HIV-1 genomic RNA. In this case, some mechanism for intravirion activation of A3G must be operational.

Other Exogenous APOBEC3 Targets
The action of APOBEC3 on reverse transcripts suggested that these enzymes might also attack other retroelements. Hepatitis B virus (HBV) is the prototypic member of the hepadnavirus family. Like the retroviruses, hepadnaviral replication involves RT of a pregenomic RNA intermediate within nucleocapsids in the cytoplasm of virus producer cells. Both human A3G and A3F interfere with the HBV life cycle in a cotransfected Huh7 hepatoma cell line (64,65). A3G/A3F appears to block the incorporation of HBV RNA into the subgenomic particles, destabilize the viral RT complex, and preclude HBV DNA accumulation in virions. Cytidine deaminase activity seems dispensable (64). The same phenomenon prevails in the HepG2 hepatoma cell line, but APOBEC3-mediated editing occurs in some HBV DNAs (66 -69). Although the precise antiviral mechanism requires further investigation, it seems to parallel the post-entry restricting function of A3G against HIV, which may not be strictly dependent on cytidine deamination.
The deaminase-independent antiviral activity of A3G also targets HTLV-1 (70). A3G packaged into HTLV-1 particles can reduce infectivity of HTLV-1 particles (70) although the effects were more modest when compared with HIV-1 (71). Signature mutations, however, do not accumulate at significant levels (70). HTLV-1 preferentially infects activated CD4 T cells that express key receptors for viral entry and thus may bypass the post-entry restricting activity of LMM A3G in resting CD4 T cells.

Roles beyond Blocking Exogenous Retroelements
A3G might also act on endogenous retroelements. These elements, which correspond to long interspersed nuclear elements (LINEs), short interspersed nuclear elements, and long terminal repeat (LTR) retrotransposons, are present at high copy number in ancestral genomes and have contributed as a major driving force to genome evolution (72). Most retroelements have lost replication competence, but some, including murine intracisternal A-particle and MusD sequences, remain mobile through retrotransposition, an intracellular process involving RT intermediates. Both human and murine A3Gs markedly inhibit retrotransposition of intracisternal A-particle and MusD elements (73) and even the yeast Ty1 LTR retrotransposon (74,75) in vitro. Cytidine deaminase activity appears critical for this inhibition (73,75). Indeed, the mouse genome contains many retrotransposon proviruses that harbor mutations consistent with APOBEC3-mediated editing (73). In addition, APOBEC genes have been subject to strong positive selection throughout the history of primate evolution (76,77). This selection appears older than the modern lentiviruses, suggesting that these genes expanded to prevent genome instability caused by endogenous retroviruses (77). Surprisingly, LINE-1 (L1) non-LTR retrotransposons do not seem to be affected by A3G (73,78). This could reflect the nuclear localization of its RT intermediates. Nonetheless, the effect of APOBEC3 proteins on non-LTR retroelements requires careful evaluation. Their replication cycle involves cytoplasmic RNA intermediates, and the APOBEC3 proteins are competent for RNA binding (8). Furthermore, the relatively narrow tissue distribution of the APOBEC3 proteins in immune cells and germ line cells (8) raises the intriguing possibility of a natural editing substrate essential for the function of these cells or the presence of harmful elements in these cells that must be nullified. Significant effort should be aimed at identifying the "natural" cellular targets of APOBEC3 proteins.

APOBEC3 as New Therapeutic Targets
Insights into the mechanism by which Vif counteracts A3G have revealed promising new targets for anti-HIV-1 drug development. The optimal strategy would be to identify small molecule inhibitors that selectively disrupt the association of Vif with A3G/A3F proteins. Others might include blocking recruitment of the E3 ligase by Vif. By suppressing the activity of Vif, these drugs would free the endogenous A3G/A3F proteins to act against HIV-1.
Information on factors that regulate APOBEC3 expression might also provide a means to counteract the Vif-mediated degradation. A3G is at low levels in resting T cells but strongly induced when the cells are activated (7). PMA stimulation of H9 T cells activates A3G gene transcription involving the mitogen-activated protein kinase signaling pathway (79). Identifying the signaling pathways involved in induction of APOBEC3 expression might be therapeutically valuable if APOBEC3 levels could be safely boosted beyond the capacity of Vif.
The discovery that cellular LMM A3G is a post-entry restriction factor in resting CD4 T cells (57) could prompt evaluation of approaches to produce this restricting activity or prevent its natural inactivation in cells that are normally highly permissive for HIV-1 infection. The conversion of LMM to HMM A3G after T cell activation merits careful assessment regarding the range of signals and the molecular pathways capable of inducing this response. Cytokines may confer permissivity for HIV-1 infection in resting T cells (80,81). Thus, cytokine stimulation could promote the assembly of A3G into the latent HMM complexes as they create a permissive state for HIV-1 in resting T cells in the absence of full-fledged cellular activation and proliferation. It will be important to determine whether cytokine treatment of T cells can also up-regulate expression of A3G and whether the increasing A3G levels are required for the assembly of HMM complexes.
Another important task is to define the protein and RNA cofactors in the HMM A3G complex (57). It remains unclear whether these HMM complexes correspond to the A3G-containing cytoplasmic bodies that lack vimentin cages (82). Small molecule inhibitors of HMM A3G complex formation might be valuable, as this host protein-protein interface would not be expected to undergo rapid sequence variation like Vif. A putative mutant of A3G constitutively residing in the LMM complex, even upon T cell stimulation, would also provide early resistance against incoming viruses. However, an effect of such disassembled forms of A3G on host genomic DNA must be excluded in these activated and often proliferating cellular hosts. It will be of great value to determine if LMM but catalytically inactive A3G could mediate the post-entry block. For lentivirus vector-mediated gene therapy in T cells, stimuli promoting the HMM A3G complex formation without potentially promoting cell activation or differentiation could be a crucial step in developing therapeutic strategies for a variety of diseases including cancers.
The role of APOBEC3 members in HBV infection is unclear. APOBEC3 expression has not been demonstrated in human hepatocytes, the primary target for HBV. Interferons inhibit HBV replication by mediating noncytopathic HBV clearance in vivo and in liver cell lines (83). A specific APOBEC family member might be up-regulated in hepatocytes by interferons and in turn facilitate HBV clearance. A better understanding of the potential participation of APOBEC3 in this response could facilitate the identification of new therapeutic strategies for HBV.

Concluding Remarks
HIV-1 RT is vulnerable to lethal editing by virion-incorporated host A3G cytidine deaminase. Unfortunately, the HIV-1 Vif protein circumvents this potent antiviral defense. HIV-1 also encounters a powerful post-entry restriction block mediated by the LMM form of A3G present in resting CD4 T cells. these cells by creating a potent post-entry block to viral replication. During cellular activation and differentiation, these proteins are recruited into an enzymatically inactive HMM complex. In this form, A3G fails to attack the incoming virions as shown in B and C. B, lethal editing mediated by virion-incorporated cytidine deaminases. In the absence of Vif, A3G is effectively incorporated into HIV virions budding from virus producer cells and massively mutates nascent HIV DNA during reverse transcription in the next round of infection. These events halt HIV growth. C, neutralization of APOBEC3 proteins by Vif. Expression of Vif from the integrated provirus impairs the translation of A3G mRNA and targets the A3G protein for proteasomal degradation. These two activities of Vif effectively eliminate A3G from the cell and lead to production of fully infectious virions lacking A3G. Ub, ubiquitin; E2, ubiquitin carrier protein. D, anti-HBV activity. Cellular A3G blocks incorporation of HBV RNA into the subgenomic particles, thereby destabilizing the viral reverse transcription complex and precluding HBV DNA accumulation.
This new restricting activity of A3G is not strictly dependent on editing and is not antagonized by Vif. These findings highlight the multifunctional properties of A3G as an antiviral factor acting at distinct steps along the retroviral life cycle, utilizing its inherent deaminase activity or a deaminase-independent antiviral mechanism. Acting in concert with other family members, A3G proteins also likely function as important guardians of the integrity of the human genome through their effects on endogenous retroviruses and possibly other retroelements.