Members of the APOBEC protein family all contain consensus cytidine-deaminase motifs, some of which have been shown to catalyse the deamination of cytidine to uridine in DNA or RNA. Replication over the mutated uridine (which is recognized as thymidine by the polymerase machinery) then results in a guanosine (G) to adenosine (A) mutation in the pairing strand. APOBEC3G has potent antiretroviral activity, and consistent with its proposed role as a cytidine deaminase, it induces high levels of deleterious G to A mutations in HIV-1 cDNA. Given that APOBEC3G has two consensus cytidine-deaminase motifs, one at each terminus, Sheehy and colleagues set out to find which of these is important for the antiviral effect.

First, they tested the ability of each domain to mutate DNA in an in vitro assay in which mutation of the RNA polymerase B gene of Escherichia coli by APOBEC3G is measured by the frequency of acquired resistance to rifampicin. Mutations in the amino (N)-terminal domain of APOBEC3G did not affect the frequency of rifampicin resistance, whereas mutations in the carboxyl (C)-terminus markedly decreased the frequency of resistant bacterial colonies. Surprisingly, this indicates that only the C-terminal cytidine-deaminase motif of APOBEC3G has DNA-mutating activity.

However, although mutations at the C-terminus inhibited the mutational activity of APOBEC3G, they did not significantly affect antiviral activity against susceptible strains of HIV-1. Comparison of the different APOBEC3G mutants showed that antiviral activity depends on at least one of the domains being intact, but it doesn't matter which one. It therefore seems that there might be an alternative mechanism, independent of DNA deamination, for the antiviral activity of APOBEC3G. Consistent with this, HIV-1 cDNAs recovered from cells infected with APOBEC3G-susceptible virions produced in the presence of C-terminal-mutated APOBEC3G showed no evidence of G to A hypermutation, despite the presence of a strong antiretroviral effect.

The authors experimentally ruled out a dominant-negative effect of C-terminal mutations on any DNA-mutating activity of the N-terminus of APOBEC3G. They also showed that the failure of C-terminal APOBEC3G mutants to edit viral cDNA did not result from a lack of virion packaging. Therefore, these results call into question the previously assumed absolute correlation between cytidine-deaminase activity and antiretroviral function of APOBEC3G, which might have implications for the function of other members of the APOBEC family.