According to our current understanding of APOBEC3G function, this host restriction factor limits the spread of HIV-1 infection, and other retroviruses [1
], by packaging into the virus during assembly in the producer cell. Following infection of the next target cell, APOBEC3G mediates extensive dC-to-dU deamination of the viral cDNA, incomplete cDNA synthesis, and genome degradation [3
]. Although effective against vif
-deficient HIV-1, the potent antiviral activity of APOBEC3G is successfully neutralized by wild-type HIV-1 through Vif [18
], which functions in concert with an E3 ubiquitin ligase complex to mediate the polyubiquitination and rapid degradation of APOBEC3G through the proteasome [18–20
Despite these significant advances in our understanding of APOBEC3G biology, there remained a considerable lack of detail concerning the subcellular context in which APOBEC3G functions. Cell lines that are permissive (i.e., cells that do not express endogenous APOBEC3G) to vif
-deficient HIV-1 replication can be rendered nonpermissive through either the transient or stable expression of recombinant APOBEC3G. Previously, we investigated the subcellular localization of recombinant APOBEC3G in these cells lines and reported that the protein localized throughout the cytoplasm and also to punctate cytoplasmic foci [26
], which we termed cytoplasmic bodies in reference to the cytoplasmic bodies of another retroviral restriction factor TRIM5α [60
]. The fact that these bodies were present under conditions that rendered cells nonpermissive to vif
-deficient HIV-1 infection suggested to us that these structures could be relevant to the antiviral properties of APOBEC3G.
We began this study by confirming that endogenous APOBEC3G also localizes to cytoplasmic bodies in primary peripheral blood CD4+
T cells, establishing that this was a bona fide property of APOBEC3G in cells that serve as a natural target for HIV-1 infection in vivo and leading us to investigate their identity. Our initial studies revealed that APOBEC3G cytoplasmic bodies were distinct from TRIM5α cytoplasmic bodies and further that these bodies did not overlap with endocytic vesicles including early endosomes, late endosomes, or lysosomes (M. J. Wichroski and T. M. Rana, unpublished data). Using translation inhibitors to monitor the kinetics of cytoplasmic body assembly and disassembly, we observed that they disappeared from the cytoplasm within 60 min of cyclohexamide treatment. This dependence on active translation was a remarkably similar property of proteins that localize to mRNA processing (P) bodies, specialized compartments within the cytoplasm of both yeast and mammalian cells where nontranslating mRNAs accumulate and are subject to degradation or storage [27
]. The localization of APOBEC3G to P-bodies raised the possibility that APOBEC3G could interact with other P-body proteins. Biochemical analysis showed APOBEC3G interactions with P-body proteins that function in cap-dependent translation (eIF4E and eIF4E-T [30
]), translation suppression (RCK/p54 [31
]), RNA interference–mediated post-transcriptional gene silencing (AGO2 [35
]), and decapping of mRNA (DCP2 [27
]). The observation that the interactions between APOBEC3G and these particular P-body proteins were all RNA dependent suggested that APOBEC3G localized to a large multiprotein RNP complex in the cell. Although the significance of the interactions mentioned above toward cellular and/or antiviral APOBEC3G functions remains to be determined, it is of considerable interest that APOBEC3G is associated with cellular machinery that mediates cytoplasmic mRNA processing events. While this manuscript was under review, Beliakova-Bethell et al. [63
] reported an intriguing study linking the assembly of the yeast Ty3 retrotransposon virus-like particles with P-bodies. Therefore, it is quite possible that there is a link between the assembly of human retroviruses/retrotransposons and P-bodies.
Our studies on APOBEC3F and APOBEC3B revealed a possible link between the localization of APOBEC3G and APOBEC3F to P-bodies with potent anti–HIV-1 activity. APOBEC3F, which shares approximately 50% sequence identity with APOBEC3G, exhibits potent anti–HIV-1 activity, and is targeted by Vif for proteasome-mediated degradation [45
], and both proteins are coexpressed in lymphoid cells [55
]. Our results showed that APOBEC3F localizes to P-bodies and hetero-oligomerizes with APOBEC3G through an RNA-dependent interaction. On the contrary, APOBEC3B, which shares approximately 59% sequence identity with APOBEC3G, exhibits only modest anti–HIV-1 activity relative to APOBEC3G and is resistant to Vif due to their inability to interact in the cell [51
]. Interestingly, APOBEC3B rarely localized to P-bodies and was found largely in the nucleus of 293T cells. Furthermore, significant coimmunoprecipitation of APOBEC3B with APOBEC3G was not detected, demonstrating that APOBEC3B does not localize to the same RNP complex shared by APOBEC3G and APOBEC3F. While further studies are necessary to fully assess the role of P-bodies in APOBEC3G antiviral function, these findings provide an interesting link between the potent anti–HIV-1 activities of APOBEC3G and APOBEC3F with their abilities to assemble into an RNP complex and localize to P-bodies. It is also of interest to note that APOBEC3B is not expressed in lymphoid cells [55
] and thus would not encounter HIV-1 in vivo, suggesting a likely explanation as to why this protein has not evolved similarly to APOBEC3G and APOBEC3F with respect to anti–HIV-1 activity.
Considering a primary function of HIV-1 Vif is to restrict the incorporation of APOBEC3G into virions, we also determined whether Vif localized to P-bodies. Although the coexpression of Vif and APOBEC3G leads to a significant reduction in APOBEC3G levels, it is possible to detect cells where APOBEC3G and Vif are visible in the same cell [26
]. In these cases, the remaining APOBEC3G rarely localized to P-bodies; however, in cases where P-body localization could be detected we noticed that Vif localized to these P-bodies as well. This observation suggested that Vif could localize to P-bodies in the presence of APOBEC3G but that the reduction in APOBEC3G levels mediated by Vif also restricted its own localization to P-bodies. This hypothesis was confirmed when it was shown that proteasome inhibition or coexpression of APOBEC3G with the Vif C114S
mutant, a nonfunctional Vif mutant that continues to interact with APOBEC3G [26
], led to complete colocalization of Vif and APOBEC3G at P-bodies. Interestingly, Vif did not localize to P-bodies in the absence of APOBEC3G, showing that the APOBEC3G-Vif interaction was responsible for targeting Vif to P-bodies. The finding that Vif-mediated degradation restricted the localization of both proteins to P-bodies suggests that an important role for Vif could be to selectively remove APOBEC3G from and/or prevent the association of APOBEC3G with P-bodies.
One of the more intriguing aspects of APOBEC3G biology is that it targets a broad range of both exogenous [1
] and endogenous [5
] retroviruses and can also limit the production of infectious hepatitis B virus [7
], a DNA virus the replicates through an RNA intermediate. These findings suggest that APOBEC3G and other members of the APOBEC3 family of proteins may constitute a broad antiviral defense mechanism. This hypothesis is supported by recent findings demonstrating that the expression of APOBEC3G and other APOBEC3 family members is induced by interferon-α in primary monocyte-derived macrophages [68
] and primary hepatocytes [69
], suggesting that these proteins could be up-regulated in vivo as part of the innate immune response to viral infection. The findings that APOBEC3G and APOBEC3F are associated with mRNA processing machinery provide an important clue as to how these proteins may have evolved to exploit a common theme to protect the cell from a broad range of foreign genetic elements. Future studies would reveal the role of P bodies in the assembly of human retroviruses and retrotransposons.