APOBEC3F properties and hypermutation preferences indicate activity against HIV-1 in vivo

APOBEC3G (CEM15 [1]) deaminates cytosine to uracil in nascent retroviral cDNA [2-5]. The potency of this cellular defense is evidenced by a dramatic reduction in viral infectivity and the occurrence of high frequencies of retroviral genomic-strand G → A transition mutations [2-5]. The overwhelming dinucleotide hypermutation preference of APOBEC3G acting upon a variety of model retroviral substrates is 5′-GG → -AG [2-4, 6-8]. However, a distinct 5′-GA → -AA bias, which is difficult to attribute to APOBEC3G alone [9], prevails in HIV-1 sequences derived from infected individuals (e.g., [10]). Here, we show that APOBEC3F is also a potent retroviral restrictor but that its activity, unlike that of APOBEC3G, is partially resistant to HIV-1 Vif and results in a clear 5′-GA → -AA retroviral hypermutation preference. This bias is also apparent in a bacterial mutation assay, suggesting that it is an intrinsic APOBEC3F property. Moreover, APOBEC3F and APOBEC3G appear to be coordinately expressed in a wide range of human tissues and are independently able to inhibit retroviral infection. Thus, APOBEC3F and APOBEC3G are likely to function alongside one another in the provision of an innate immune defense, with APOBEC3F functioning as the major contributor to HIV-1 hypermutation in vivo.