HIV is unable to multiply in certain human cells unless it expresses a protein known as virion infectivity factor (Vif). Although Vif was identified in the mid 1980s, it was only last year that a function for this protein was identified when an unequivocal correlation was observed between production of the cellular protein APOBEC3G (also known as CEM15) and the failure of Vif-deficient HIV-1 particles to produce infectious progeny in some host cells. Building on this breakthrough, four papers have just been published that further our understanding of the role of APOBEC3G as an endogenous viral inhibitor and the importance of Vif in overcoming this antiviral strategy.

APOBEC3G is a member of a family of cytidine deaminases that edit RNA, and recent studies have shown that certain members of this family target single-stranded DNA. These observations led several laboratories to hypothesize independently that APOBEC3G could block retroviral infectivity by deaminating cytosine in either virus RNA or DNA, so inducing mutation and/or instability of the virus genome. To address the question of whether APOBEC3G acted on RNA or DNA, Hui Zhang and colleagues sequenced newly transcribed genomic RNA from Vif-defective HIV-1 virions and found no significant mutations, indicating that RNA is not deaminated by APOBEC3G. They went on to investigate whether newly synthesized HIV-1 DNA was a substrate, and found guanine (G) to adenine (A) substitutions in all sequences that were analysed. Denise Lecossier and colleagues carried out a similar analysis on newly reverse transcribed HIV-1 genes Env and U5 from infected cells. They also observed that G to A changes occurred more often in Vif-defective virions than in DNA from wild-type virions. Furthermore, no differences were observed in the frequency of G to A changes in virus DNA that was synthesized by wild-type and Vif-defective viruses produced in permissive cells — that is, cells in which the Vif protein is not required for propagative infection. The study by Lecossier and co-workers also showed that when Vif-defective virions produced in cells that lack APOBEC3G subsequently infected cells expressing this enzyme, their DNA was not edited, indicating that APOBEC3G has to be present in the virus particle to exert its effect.

In an article published simultaneously with the paper of Zhang and co-workers, Bastien Mangeat et al. showed that APOBEC3G was fully active on retrovirus particles that were only distantly related to HIV-1 — APOBEC3G is, therefore, a broadly active anti-retroviral that can both inhibit a wide range of retroviruses and compromise the expression from different pro-viruses. Using a different approach based on mouse leukaemia virus (MLV), Reuben Harris and colleagues also proved that APOBEC3G is a DNA deaminase that is packaged into virions during virus production, culminating in marked deamination of deoxycytidine in the retrovirus minus-strand complementary DNA. Using this system, they showed that MLV produced in cells expressing APOBEC3G had diminished infectivity that correlated with the introduction of a high number of G to A substitutions. These workers also showed that the Vif protein of HIV-1 could protect MLV from APOBEC3G-dependent restriction.

The bottom line from each of these investigations is that when HIV-1 replicates in cells that express APOBEC3G, the Vif protein, through mechanisms that are yet to be defined, can prevent the accumulation of defects in virus proteins that would otherwise result in a non-functional life cycle. The finding that APOBEC3G can act on a broad range of retroviruses, in addition to HIV-1, indicates that hypermutation generated by DNA deamination is a general innate defence mechanism against this important group of pathogens. It can also be argued that the accumulation of APOBEC3G-mediated non-lethal mutations in the replicating virus genome could make a crucial contribution to the variation of virus sequence that is observed in lentivirus populations.