Inhibition of hepatitis B virus replication by APOBEC3G.

The cytidine deaminase APOBEC3G (A3G) was recently identified as a natural resistance gene that restricts efficient propagation of human immunodeficiency virus and other retroviruses. The enzyme induces massive cytidine to uridine (C3U) deamination of single-stranded retroviral DNA, resulting in DNA degradation or lethal guanine to adenine (G3A) hypermutation (1). Hepadnaviruses, including hepatitis B virus (HBV), replicate by reverse transcription of a pregenomic RNA intermediate inside nucleocapsids, placing them into the family of retroelements (2). These observations, along with an earlier report describing G3A hypermutations in natural HBV variants (3), raise the question of whether HBV represents another potential target for A3G. Turelli et al. showed that this is indeed the case (4). Surprisingly, however, inhibition of viral pregenome packaging rather than induction of G3A hypermutations was identified as the main antiviral mechanism. No significant nucleotide changes were detected in a total of 40 polymerase chain reaction (PCR)–amplified HBV clones derived from cotransfected Huh7 hepatoma cells. Turelli et al. discussed the possibility that A3G-mediated HBV editing may occur in a different cellular context. We investigated the potential antiviral effect of A3G in cotransfected Huh7 cells and another human hepatoma cell line, HepG2 (5). Our results confirm that A3G interferes with proper packaging of viral pregenomic RNA, resulting in a marked suppression of viral DNA synthesis (data not shown). To search for potential A3Gmediated editing of HBV DNA in nucleocapsids that may have escaped the block in RNA packaging, we PCR-amplified newly synthesized HBV DNA from supernatants or cell lysates of cotransfected cells and sequenced individual clones (6). Figure 1 summarizes the results obtained from three experiments in HepG2 cells and two experiments in Huh7 cells. In total, 430 individual clones were sequenced. In Huh7 cells, G3A mutations were rare, irrespective of the presence or absence of A3G, thus confirming the finding of Turelli et al. (4). In A3G-expressing HepG2 cells, the majority of recovered sequences were wildtype as well. However, the number of clones bearing G3A mutations and the overall number of G3A mutations increased significantly (Fisher’s exact test, P 0.034), whereas other nucleotide substitutions were rare (Fig. 1). Further experiments revealed additional G3A mutations in other regions of HBV DNA (Fig. 2), which suggests that they were caused by processive enzymatic activity rather than by global imbalances in the cellular nucleotide pool. Targeted sequence motifs matched well with the hallmarks of A3G action [(7), Fig. 2]. In conclusion, A3G displays a dual antiviral effect: (i) interference with pregenomic RNA packaging and (ii) induction of extensive G3A mutations in a subset of HBV genomes. Interestingly, A3G-mediated editing of HBV DNA appears to be cell line–dependent. The cellular factor(s) accounting for differences in A3G deaminase activity remain to be defined. Conceivably, Huh7 cells either lack a cofactor that is important for deaminase activity or produce a suppressing factor. It is of note that HepG2 cells occasionally yielded some G3A mutations even in the absence of transfected A3G. Because endogenous A3G expression in HepG2 cells was minute (8), this might reflect the activity of another deaminase. Nevertheless,