|Home | About | Journals | Submit | Contact Us | Français|
APOBEC3G (A3G) is an effective cellular host defense factor under experimental conditions in which a functional form of the HIV-encoded protein Vif cannot be expressed. Wild type Vif targets A3G for proteasomal degradation and along with it, any host defense advantage A3G might provide is severely diminished or lost. Recent evidence cast doubt on the potency of A3G in host defense and suggested that it could, under some circumstances, promote the emergence of more virulent HIV strains. In this article, I argue that it is time to recognize that A3G has the potential to act as a double agent. The path forward relies on understanding how cellular and viral regulatory mechanisms enable A3G antiviral function and on developing novel research reagents to explore these pathways.
APOBEC3G (A3G) is a member of a family of cytidine deaminases named after APOBEC11, 2, which was the first enzyme discovered with the capacity for site-specific cytidine to uridine deamination (editing) of apolipoprotein B messenger RNA. Almost nine years ago, Sheehy et al. published a paper showing that the reason why the HIV protein known as Viral infectivity factor (Vif) was required for the virus to infect nonpermissive cells was that these cells expressed A3G. Vif enabled the virus to penetrate host defenses by inducing the destruction of A3G3. At that time the story was simple and exciting: Vif induced the degradation of A3G4, 5 thus preventing A3G from being incorporated into nascent viral particles and thereby neutralizing the ability of A3G to hypermutate HIV single stranded DNA (ssDNA) post-entry during reverse transcription6–8 (Figure 1) (Text Box 1).
In the deaminase dependent mechanism, A3G catalyzes zinc-dependent hydrolytic deamination of deoxycytidine to form deoxyuridine in HIV DNA6, 7. These mutations arise primarily on the HIV minus strand due to A3G having a requirement for a ssDNA substrate8, 42. The frequency and distribution of mutations the HIV proviral DNA is determined both by the 3′ to 5′ processivity of A3G activity on ssDNA43 and limited temporally by the transient availability of ssDNA arising from RNase H-mediated removal of the RNA genome template following reverse transcription34 and prior the formation of dsDNA by second strand synthesis. A3G deaminase activity is thought to be the consequence of protein-protein and protein-ssDNA interactions that drive assembly of larger A3G-ssDNA aggregates presumed to be essential for deaminase activity8, 42, 44.
Vif is a mimic of cellular SOCS box proteins which function as receptors for protein substrates targeted for ubiquitylation through the elongin B/C–culin5–ring box protein 1(RBX2)–E3 ligase complex and degradation via the 26S proteasome45, 46. Vif–A3G binding requires residues within the N-terminus of A3G29, 47–49 and residues in the N-terminus31–33, 50–52 and C-terminus4, 53, 54 of Vif.
In the deaminase-independent mechanism, A3G is predicted to contain an N-terminal and a C-terminal zinc-dependent deaminase or ZDD fold1. Efforts to delineate the contribution of each ZDD to antiviral activity showed that deaminase activity resided exclusively within the C-terminal ZDD15, 55–57. These studies were controversial because they also suggested that deaminase activity might not be required for antiviral activity in experimental systems14, 57, 58. A3G has an intrinsic ability to nonspecifically bind RNA and ssDNA42, 56, 59, 60. This characteristic is undoubtedly essential to the deaminase-independent antiviral activities that have been described in more recent literature wherein A3G binds nucleic acids to inhibit tRNAlys3 priming of first strand synthesis61, strand transfer activity62, reverse transcript elongation63 and inhibition of double stranded proviral DNA integration64, 65.
There is agreement that protein–protein and protein–RNA interactions with the N-terminal half of A3G are required for encapsidation. However, there is disagreement over whether the interactions of A3G with Gag47, 66, 67 and/or viral RNA and/or cellular RNAs59, 60, 68–71 are sufficient to place A3G, along with the viral genome, inside the virion core such that it will be ideally positioned to inhibit reverse transcription post-entry.
In the absence of functional Vif, A3G catalyzes dC to dU mutations primarily in the minus strand reverse transcript, and this templates dG to dA transitions in the protein coding plus strand during viral replication (Text Box 1). Some mutated virions can then integrate into the host cell chromosome, leading to the expression of viral proteins with missense substitutions or nonsense codons9, 10 (Figure 1). These findings are consistent with sequence analysis of HIV isolated from infected patients who had numerous dG to dA polymorphisms; in some instances, inactivating substitutions in the Vif sequence were identified9. The flanking nucleotide sequences of these single nucleotide polymorphisms were the same as those preferred at editing sites for A3G and its homolog A3F9. In a recent study, HIV genomes recovered from infected humanized mice expressing human A3G also contained dG to dA mutations with an A3G flanking sequence preference11. In most experimental systems, the absence of Vif enables sufficient dC to dU mutations leading to the reduction in proviral DNA due to the creation of abasic sites by uracil-DNA glycosylase which is followed by DNA degradation12, 13. In summary, A3G-mediated mutations, which occur in addition to the mutations stemming from low fidelity reverse transcription and recombination of viral genomes, were initially proposed to be solely detrimental to the virus, because of their location in the HIV genome or their abundance.
Soon after the discovery of A3G deaminase-dependent antiviral activity, experiments evaluating the functional requirement of residues in the catalytic domain of A3G through site-directed mutagenesis and deletion analyses revealed that A3G had deaminase-independent antiviral activities14, 15. A3G interactions with viral and host cell proteins and RNAs were shown to inhibit HIV reverse transcription and promote A3G assembly within viral particles (Text Box 1) (Figure 1). Proponents of the deaminase-dependent mechanism argued that deaminase-independent interactions were only apparent because of the supra physiological levels of A3G expressed in transfected cell systems16–18. It is, however, inescapable that A3G subunits have an intrinsic ability to bind proteins, ssDNA and RNA, and therefore deaminase-independent interactions could occur at all levels of expression. It is also important not to overlook the fact that deaminase-independent interactions determine A3G assembly with virions: this is essential according to the deaminase-dependent hypothesis for A3G antiviral activity. What remains unclear is whether the deaminase-independent interactions of A3G that have been proposed to interfere with reverse transcription (Text Box 1) are sufficient to inhibit viral replication within the biological range of A3G expression.
Pursuing the possibility that A3G is a host defense factor, high throughput screening of chemical libraries identified small molecules with antiviral activity that were selected based on their ability to inhibit Vif-dependent A3G degradation19, 20. Similarly, a cell-transducing peptide mimic of the Vif dimerization domain was optimized21 from phage display screening of peptides that disrupted Vif multimers and thereby inhibited viral replication in nonpermissive cells. Both the small molecules and peptide had the anticipated outcome of producing virions that had a higher content of A3G and lower infectivity19–21.
The opinion expressed here is that A3G has the potential of being a double agent enzyme, serving as an antiviral factor or as a facilitator of viral genome diversification which leads to the emergence of drug resistance. Sequence analysis data suggest that varying levels of A3G deaminase-dependent (and therefore deaminase-independent) activity can occur during an infection with wild type virus that expresses functional Vif. Although this theoretically could serve a host defense role, it is uncertain whether the deaminase-dependent mutagenic activity of A3G is sufficient (in activated or resting T cells for example) to exceed a mutagenic threshold for inactivating the HIV genome if the deaminase-independent activity is not sufficient to block viral replication. It should be of concern that discovery of the antiviral mechanisms has relied on overexpressing A3G or A3G mutants, or experimentally ablating A3G or Vif. Logically there is a difference between these engineered systems expressing mRNA and protein from cDNAs and a native setting in which A3G and Vif genes are both expressed and regulated during an inflammatory response. An alternative might the development of target-specific molecular probes that enable evaluating the function of A3G and Vif in native cells following infection with wild type virus. Although there are off-target and toxic artifacts possible within a drug-treatment experimental design, arguably these research reagents hold greater potential than our current approaches to reveal physiological mechanisms that are relevant to those in HIV/AIDS patients.
A3 proteins are not essential for cell survival22, but they could have important functions related to their ability to bind RNA, including the regulation of miRNA functions23 and suppression of endogenous retroviral elements24–27. In fact, the ability of A3 family members to elude inactivation by viral factors such as Vif and A3G/A3F-mediated mutation of viral genomes might have contributed to the rapid evolution of primate retroviral and endogenous retroviral-like elements28. Species-specific and APOBEC3 homolog-specific29 sequence differences might help to determine the present day primate species tropism of HIV/SIV (simian immunodeficiency virus)30–33.
A question that has become increasingly perplexing is why does pre-existing cellular A3G fail to deliver a preemptive antiviral strike on incoming viruses? Why is it necessary to slip a few A3G subunits into the viral particle like the proverbial Greeks in the Trojan Horse? The answer to the question might be that cellular A3G cannot gain access the nucleoprotein core of the virus where reverse transcription takes place unless it is encapsidated or the nascent ssDNA is itself shielded from cellular A3G 34. In addition, cells might regulate A3G for its interaction with cellular RNAs and this might limit the availability of A3G and/or its access to nascent single stranded proviral DNA.
At the heart of this conundrum is the seemingly contradictory concept that the intrinsic ability of A3G to bind RNA is essential for the deaminase-independent antiviral mechanism by which A3G enters the viral particle (Text Box 1) and yet binding to viral RNA in the capsid or binding to cellular RNAs inhibits A3G deaminase-dependent mutagenesis of ssDNA during reverse transcription (Text Box 2). Different cells can regulate the interaction of A3G with cellular RNAs in contrasting ways to generate low molecular mass (LMM) forms of A3G (relatively free of RNA) and high molecular mass (HMM) aggregates of A3G bound to RNA in ribonucleoprotein (RNP) complexes (Text Box 2) (Figure 1). The larger aggregates colocalize with cytoplasmic P-Bodies and Stress Granules that function in RNA and RNP protein degradation and/or serve as temporary storage depots. In HMM aggregates, A3G is believed to have little or no antiviral activity yet Vif mediates A3G degradation in both the LMM and HMM forms 18.
Purified A3G forms homomultimers in a concentration-dependent manner through protein–protein interactions72, 73. Binding of either RNA or ssDNA to A3G will promote higher order oligomerization of A3G subunits as evidenced under defined in vitro conditions by native gel shift analyses42 and atomic force microscopy44. RNP complexes containing A3G referred to as high molecular mass (HMM) complexes colocalize with P-bodies and Stress Granules43, 44, 74–79 where cells degrade or recycle RNAs and RNP proteins. Glycerol gradient sedimentation or size exclusion chromatography of cell extracts have demonstrated that A3G associates with these megaDaltonsized RNPs within minutes of its translation17, 18.
Whereas HMM A3G predominates in cells that are permissive to HIV infection, cell types considered to be refractory to HIV infection (e.g., resting CD4+ T cells, monocytes and mature dendritic cells74, 76, 79) maintain A3G as a component of low molecular mass (LMM) complexes. Through mechanisms that are not understood, cytokines, double stranded RNAs and other growth factors regulate the expression of A3G and the inter-conversion of LMM and HMM complexes. The significance of the inter-conversion of A3G aggregation states to its antiviral activity is currently an underdeveloped issue that needs further investigation36, 37.
All or most of the conversion of LMM to HMM complexes is attributable to RNA-bridged A3G oligomerization and to date no protein components of P-bodies or Stress Granules have been shown to bind directly to A3G. As part of HMM complexes, A3G has low or no deaminase activity, but RNase digestion of isolated HMM complexes restores deaminase activity73, 74.
The higher level of deaminase activity in LMM A3G and its prevalence in nonpermissive cells led to the hypothesis that LMM A3G was necessary and sufficient for cells to be refractory to HIV infection. However, this has been retracted35. Recent experiments testing the hypothesis showed that an aggressive knockdown of LMM A3G in resting CD4+ T cells by RNAi or Vif-mediated degradation of A3G did not render these cells permissive to HIV infection36, 37. The implication of the new data is that LMM A3G pre-existing in cells is not sufficiently available or active enough to inhibit HIV replication.
The new data raise an important question: if A3G in nonpermissive cells is not sufficient to inhibit viral replication, does it have no activity at all, or does it have a low level of mutagenic activity? This is an important question because earlier studies (reviewed in38) warned that rather than providing antiviral host defense, A3G/A3F activities might diversify the viral genome, a tenet predicted by evolutionary biologists studying retroviruses and retroviral elements28. Recent papers have confirmed this possibility under experimental conditions where A3G/A3F promoted mutations that benefited HIV39, 40 and induced a drug-resistant phenotype41. The available data support the possibility that the activity of LMM A3G in infected nonpermissive cells is sufficient to promote mutations in the HIV genome. Moreover, the new data suggest that the mutational activity might not be sufficient to inhibit viral replication either during early or late stages of the viral life cycle (Figure 1).
The important question before us is what level of A3G mutagenic activity might benefit the virus compared to what is necessary to inhibit viral replication? The simplest answer might be that a little DNA deaminase activity benefits the virus and a high level of activity destroys the virus. Unfortunately, we do not know what these levels are, and the location of mutations within the viral genome can be more important than how many mutations there are. To explore this possibility, future research should determine the naturally occurring mechanisms regulating the interconversion of HMM and LMM and how this might determine A3G mutagenic activity. Based on the results, we should gain an understanding of the paradoxical situation wherein A3G host defense mechanisms are sequestered in RNPs and inactivated in permissive cells during an HIV infection. The results will have a significant impact on translational research and therapeutic development using A3G/Vif as a target.
From one point of view, the available data suggest that if naturally expressed A3G activity is too low to provide host defense, inhibiting Vif alone might have limited immediate therapeutic value and could have the unintended long-term consequence of aiding the virus. Conceptually one therapeutic strategy might be to inhibit A3G deaminase activity and allow Vif to destroy A3G. This might hold the potential of reducing the emergence of viral resistance in patients who are also receiving other modalities of antiretroviral therapy. The significant challenge here is that deaminase inhibitors would have to be developed that are selective for A3G/A3F so that they will not be cytotoxic to intermediary metabolism. Skeptics will argue that inhibiting a host defense factor because it might induce beneficial mutations in the virus ignores the greater possibility that A3G mutations might, overall, be deleterious to the virus.
An alternative approach is to activate A3G HMM mutagenic efficiency and enhance its access to viral replication complexes. Could such activators enable cells to ‘fight back’ even though Vif is destroying A3G? In this case, combining A3G activators with Vif inhibitors might address both the need to reduce viral infection and lessen the frequency with which drug resistant strains emerge. Similar to deaminase inhibitors, deaminase activators will need to be A3G/A3F specific.
It is vital that efforts continue to be made to explore cellular and viral factors that regulate the double agent A3G. Undoubtedly, the development of novel probes and research reagents will be crucial to our understanding of mechanisms that determine what role A3G plays in infected cells. Despite progress in these areas, the therapeutic value of A3G inhibitors or activators, with or without Vif antagonists, will not be fully appreciated until such compounds are tested in clinical trials.
The author thanks Jenny M.L. Smith for the illustrations. The author is grateful to Steve Dewhurst, Jason Salter, Jenny Smith and members of the Smith and Wedekind laboratories and OyaGen, Inc for critical reading of the manuscript. This work was supported by Public Health Service grants (NIAID R21 058789, NINDS R21 067671) and the Bill and Melinda Gates foundation grant awarded to HCS, a Public Health Services grant (NIAID R21/R33 076085) awarded to Joseph E. Wedekind and a Public Health Services T32 training grant (NIAID 049815) and Developmental Center for AIDS Research grant (NIAID P30 078498) awarded to Steve Dewhurst.
Conflict of Interest Statement
H.C. Smith is a full time faculty member in the Department of Biochemistry and Biophysics and the Center for RNA Biology at the University of Rochester, School of Medicine and Dentistry, Rochester, NY. He is also founder of OyaGen, Inc and a consultant for the company as its chief scientific officer. OyaGen, Inc, is a therapeutic development company seeking novel therapeutics using APOBEC editing mechanisms as targets (www.oyageninc.com)
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.