APOBEC3 Proteins in Viral Immunity

The genomes of mammals and other species contain many genes that restrict infection by viruses. Many antiviral intrinsic restriction factors were discovered through the identification of viral gene products that counteract their action. For example, apolipoprotein B editing complex (APOBEC)3G, a cytidine deaminase (CDA) belonging to the activation-induced CDA (AIDCDA)/APOBEC gene family, was found because the HIV-1–encoded viral infectivity factor (Vif) blocks its activity (1). Many host antiviral genes, including those in the APOBEC3 family, have undergone positive selection by infectious pathogens, resulting in polymorphisms in both regulatory and coding regions (2–5). In addition to the sequence heterogeneity found in polymorphic alleles within a given species, the number of APOBEC3 genes in different species varies as a result of expansion or contraction of the locus, ranging from one gene in mice and rats to seven in primates (APOBEC3A, APOBEC3B, APOBEC3C, APOBECDE, APOBEC3F, APOBEC3G and APOBEC3H) (3, 5–10). Polymorphisms in the mouse Apobec3 gene have been linked to susceptibility to infection by murine retroviruses (11–13), whereas at least one polymorphic APOBEC3H allele (HapII) is believed to confer resistance to HIV-1 infection and disease progression; genome-wide association studies also implicated other human APOBEC3 polymorphisms in HIV-induced disease (11, 14–19). Because HIV is a relatively recent virus in humans, it is likely that other infectious agents contributed to the positive selection of APOBEC3s.

APOBEC3 proteins inhibit replication of HIV-1 lacking vif, which encodes a 23-kDa protein (1, 20–23). In virus-producer cells, Vif binds APOBEC3D, APOBEC3F, APOBEC3G, and APOBEC3H, targeting them for ubiquitinylation and degradation in the proteasome through interactions with a number of cellular factors, including CBF-β, Cul5, and elongins. This prevents APOBEC3 packaging and, thereby, overcomes the antiviral activity (24–30). In contrast, in producer cells infected with vif-deficient HIV or with retroviruses that do not express a Vif protein, APOBEC3 proteins are packaged into virions via interaction with the nucleocapsid protein and viral RNA (31–35).

Once packaged, APOBEC3 proteins inhibit infection by deaminating deoxycytidine residues on minus-strand DNA produced by reverse transcription, introducing G-to-A mutations in newly synthesized HIV-1 coding strand DNA. This leads to both degradation of reversed-transcribed DNA prior to integration and to lethal G-to-A coding strand mutations in the integrated provirus (36). APOBEC3G most often deaminates C residues in CC motifs, whereas the other APOBEC3 proteins preferentially modify those found in CT motifs (20, 37, 38). Because APOBEC3G-induced mutations commonly occur in TGG motifs, deamination often converts the tryptophan codon TGG to a TAG stop codon; in contrast, the other human APOBEC3 proteins, as well as mouse APOBEC3, generate GAA or GA to AAA and AA mutations, which more frequently result in missense mutations (22, 39, 40). APOBEC3-mediated sublethal mutagenesis could lead to both drug-resistant and immune-escape viruses, and lethal mutations may result in the production of truncated viral proteins degraded by the cell’s proteolytic processing pathways, thereby increasing Ag presentation (see “APOBEC3 and the adaptive immune response”) (41).

APOBEC3 proteins also inhibit virus replication by CDA-independent mechanisms (42) (Fig. 1). For example, APOBEC3F lacking CDA activity is as effective as the wild-type protein in inhibiting HIV-1 infection, at least in cultured cells (43). In vitro studies suggested that APOBEC3 proteins inhibit elongation and accumulation of HIV-1, murine leukemia virus, and mouse mammary tumor virus reverse-transcription products (44–46). APOBEC3 proteins also inhibit infection by parvoviruses, as well as retroelement mobilization, without extensively hypermutating their genomes (47–51).

Although virion-packaged proteins appear to restrict infection of T cells, APOBEC3 proteins expressed in human and mouse myeloid and dendritic cells (DCs) also can restrict incoming HIV, mouse mammary tumor virus, and murine leukemia virus by both CDA-dependent and -independent means (52–57). APOBEC3 proteins also were implicated in hypermutating and restricting the hepadnavirus hepatitis B virus and human papillomavirus in hepatocytes and keratinocytes, respectively, as well as HSV-1 and EBV in established cell lines and primary patient samples (58–60); whether this is the result of target cell APOBEC3 activity or packaged viruses in the case of these other virus families is not known.

The different APOBEC3 proteins are expressed at various levels in hematopoietic cell populations, including CD4⁺ and CD8⁺ T cell subsets (e.g., naive and memory), B cells, and myeloid cells (55, 61, 62). CD4⁺ T cell expression of APOBEC3 proteins clearly plays a role in virus restriction. Analysis of the deamination motifs found in HIV-1 cDNA isolated from CD4⁺ T cells indicates that APOBEC3G is the dominant antiviral protein in this cell type (63). Moreover, HIV-1 proviral DNA hypermutation combined with increased APOBEC3G expression levels was linked to lower virus replication and increased CD4⁺ T cells counts in patients (64–67), although individuals carrying the anti–HIV-1 APOBEC3H HapII allele also display lower levels of infection and higher CD4⁺ T cells counts compared with individuals carrying other APOBEC3H alleles (14). More recent work suggests that both cytidine deamination and reverse-transcription inhibition play a role in HIV-1 restriction in primary CD4⁺ T cells (63).

Macrophages are also major targets of infection by HIV and other retroviruses, particularly at mucosal surfaces, serving not only as latent reservoirs but also as APCs (68). However, macrophages are relatively resistant to retrovirus infection due to the expression of APOBEC3 proteins, as well as additional cell host restriction factors, such as SAM domain and HD domain-containing protein 1; the latter is believed to deplete the dNTP pools available to reverse transcriptase for viral cDNA synthesis (69–72). Macrophages express APOBEC3G, APOBEC3F, and APOBEC3DE, and their upregulation is IFN-α dependent, although APOBEC3G restricts HIV-1 more potently than the combined effect of APOBEC3F and APOBEC3DE in this cell type (62, 73). Target cell expression in primary macrophages may also play a role in restricting HIV, as well as other retroviruses (45, 54, 55). APOBEC3A is also expressed at high levels in monocytes, whereas fully differentiated macrophages express low levels (74); however, macrophage expression of APOBEC3A is strongly enhanced by IFN-α treatment (55, 62, 74). This difference in APOBEC3A levels between monocytes and macrophages correlates with susceptibility to HIV-1 infection, with monocytes being more resistant to infection than macrophages; when APOBEC3A was silenced in HIV-1–infected monocytes, virus production was increased (69, 74). APOBEC3G induction by IFN-α in CD14⁺ monocytes is also significantly higher in HIV-1–exposed but seronegative individuals than in HIV-infected individuals or healthy controls (75). Only a small percentage (≤6%) of HIV-1 sequences from infected macrophages contain G-to-A hypermutations; thus, it is possible that APOBEC3 restriction of lentiviruses in macrophages occurs primarily by CDA-independent means (55).

Immature myeloid DCs can be productively infected by HIV-1; these cells may also be used by HIV-1 as a “Trojan horse” to allow for the transmission of virus to the lymph nodes, thus bringing it in the proximity of T cells (76, 77). Transmission of virus from DCs to T cells likely occurs by direct cell–cell transmission via immunological synapses (78, 79). However, myeloid DCs support only a limited amount of HIV replication; this has been attributed, at least in part, to cell host restriction factors, such as APOBEC3 proteins. Activation of myeloid DCs by IFN-α results in the potent upregulation of APOBEC3A, APOBEC3F, and APOBEC3G, without the induction of signals that lead to their maturation (74, 80, 81). Unlike macrophages, HIV-1 DNA in myeloid DCs is hypermutated upon IFN-α treatment; transmission from myeloid DCs to T cells is significantly reduced compared with IFN-naive cells (80).

B cell expression of APOBEC3G is higher than that seen in monocytes (62); nevertheless, because B cells lack the entry receptors required for HIV-1 infection, APOBEC3G is unlikely to play a role in cell-intrinsic restriction. However, B cells are a major source of exosome production upon CD40L and IL-4 stimulation (82). APOBEC3G was shown to be a major exosomal component, and exosomes bearing APOBEC3G can confer anti–HIV-1 activity to Jurkat T cells (83). APOBEC3G levels are upregulated upon treatment of B cells with a combination of agonists, such as CD40L, IL-4, and anti–HLA class II Abs, and autologous CD4⁺ T cells cocultured with B cells activated with these agonists are significantly less infected by HIV-1 (84). This suggests that APOBEC3G derived from exosomes produced by activated B cells confers antiviral activity to CD4⁺ T cells (84). B cell–expressed APOBEC3 proteins may also act on other viruses that infect this cell type, such as EBV (58).

As discussed, APOBEC3 genes are IFN-stimulated genes (ISGs) whose expression is increased in response to various stimuli, including the ligands of TLR3, TLR4, and TLR7 (52, 80, 81, 85–89). Thus, the activation of innate immune responses that occurs upon infection with many viruses could lead to increased APOBEC3 activity. This might be especially important for combating viruses acquired at mucosal sites of infection, where sentinel cell targets, such as macrophages and DCs, are poised to mount such responses.

Expression of IFNs and other cytokines/chemokines is induced by virus infection, the result, in part, of cytosolic nucleic acid sensors that detect single- and double-stranded RNA and DNA (reviewed in Ref. 90). Studies showed that reverse-transcribed retroviral DNA is recognized by several sensors, including cyclic GMP-AMP synthase and members of the absent in melanoma 2 (AIM2)-like receptor family that subsequently signal through the stimulator of IFN genes (STING) pathway (91–93). As described above, APOBEC3 proteins inhibit reverse transcription in addition to introducing CDA-induced mutations. Thus, APOBEC3-mediated inhibition of reverse transcription would seemingly limit cytosolic sensing and the IFN-induced antiviral response. Indeed, we showed recently that, in mice, APOBEC3 in macrophages limits the levels of retroviral reverse transcripts that trigger cytosolic sensing (93). However, reverse transcripts that escaped this blockade activated the STING pathway via cytosolic sensing and induced type 1 IFN expression. This, in turn, increased expression of ISGs like APOBEC3, reducing virus loads in vivo. Because human APOBEC3 proteins also block reverse transcription in sentinel cells and lentivirus reverse transcripts trigger STING via cytosolic sensing, it is likely that similar interactions between the two restriction pathways occur during HIV infection. Interestingly, DNA viruses, like herpesviruses, also induce IFN production via the cytosolic sensor–STING pathway and would be predicted to induce antiviral APOBEC3 expression that could then act on these viruses. This interplay between innate immune sensors and APOBEC3 is likely to afford multiple layers of protection to host against infection by multiple virus families.

APOBEC3G also was linked to increased destruction of HIV-1–infected cells by NK cells (94). Cells infected with HIV lacking or encoding defective Vifs unable to bind APOBEC3G have higher levels of the NKG2D ligands, such as ULBPs and PLAP, resulting in more efficient NK cell–mediated lysis. The increase in lysis was linked to higher uracil content in HIV reverse transcripts caused by APOBEC3G-mediated cytidine deamination. Uracils are removed from DNA via cleavage by uracil–DNA glycosylase, which then activates the base excision–repair pathway. This, in turn, generates gaps or breaks in DNA and subsequently induces the DNA-damage response, a known inducer of NKG2D ligand expression (94) (Fig. 1). Thus, another antiviral function of APOBEC3G may be to recruit lymphocyte effectors, such as NK cells, which, in turn, eliminate infected cells. Whether other human APOBEC3 proteins also affect NK recognition of HIV or other virus-infected cells remains to be determined.

Clearly, APOBEC3 proteins play a role in the intrinsic/innate response to, and the subsequent control of, early virus infection. There is increasing evidence that APOBEC3 proteins also contribute to the adaptive immune response by affecting the generation of CTLs that recognize viral peptides and perhaps B cell production of antiviral Abs. Because APOBEC3 expression is induced by virus infection of APCs, as discussed above, this may afford additional layers of host protection.

It is well-known that viruses causing persistent infection, like HIV-1, generate escape variants. HIV may take advantage of nonlethal mutations induced by APOBEC3 proteins to generate such CTL escape variants (95); APOBEC3G/F hotspots in the HIV genome are enriched in immunogenic CTL epitopes, and deamination at these hotspots diminished CD8⁺ T cell responses to infected cells (39, 40). Thus, it may be beneficial for the virus to preserve APOBEC3G mutation hotspots at locations that influence the presentation and recognition of T cell epitopes and thereby aid in immune escape. In contrast, the host may strengthen its intrinsic ability to control HIV-1 infection by selecting for immune escape variants; there are known CD8 epitopes in Vif itself and Vif immune escape variants would be predicted to more weakly counteract APOBEC3s (96, 97).

CTL responses to viruses may also be increased through APOBEC3G-generated stop codons, leading to premature truncation of proteins and degradation via Ag-processing pathways in HIV-1–infected cells (98, 99) (Fig. 1). Indeed, the virus-specific CTL response to APOBEC3G-expressing, HIV-infected APCs is increased compared with APCs that do not express the CDA (99). Moreover, the activation of HIV-specific CTLs was stronger with vif-deficient HIV-infected APCs. However, a recent report that longitudinally followed pol and env sequences in a large cohort of patients chronically infected with HIV-1 found that the vast majority of the G-to-A mutations fell into the recognition motif for APOBEC3F and the other APOBEC3 proteins and not that of APOBEC3G (39). In this case, missense mutations in viral proteins could also lead to misfolding and protein degradation. Indeed, for many viruses, including HIV-1, there is an inverse correlation between CTL response and viremia and disease progression (100, 101). Thus, the effects of APOBEC3 proteins on the CTL response can have a significant impact on retrovirus infection and evolution.

B cell production of virus-neutralizing Abs may also be altered by APOBEC3 proteins. APOBEC3-knockout mice generate neutralizing Abs against Friend murine leukemia virus (FV) with significantly slower kinetics compared with wild-type mice, although the responses in APOBEC3-knockout mice eventually become as high as those in wild-type animals (13, 102). This effect on the antiviral humoral immune response is likely due to APOBEC3-mediated suppression of virus infection at early times, thereby limiting the number of virus-producing cells in the early stages of infection. This may reduce Ag load and the induction of immune tolerance or prevent FV-mediated damage of critical hematopoietic lineage cells required for generating an immune response (102, 103).

APOBEC3 may also play a role in somatic hypermutation of Abs. AID is the primary CDA involved in somatic hypermutation of Ig genes that results in the diverse repertoire of Abs needed to clear infections (104). One study found that IgH sequences of FV-specific mAbs derived from APOBEC3-knockout mice had significantly lower levels of C-to-T and G-to-A somatic hypermutation and lower virus-binding ability than did those from wild-type mice (105). However, it is not clear that Ig somatic mutations are critical for the control of FV, because AID-knockout mice produce virus-neutralizing IgM Abs and control infection (106). Nevertheless, because the human APOBEC3 proteins APOBEC3A and APOBEC3B have been implicated in mutating genomic DNA, it is possible that they contribute to the development of neutralizing Abs during virus infection (107, 108).

Organisms adapt to infectious agents by developing protective responses; conversely, infectious agents develop adaptive countermeasures to these responses. Host defenses against infectious agents include various mechanisms of innate immunity (e.g., NK cells, TLRs, and IFNs) and adaptive immunity (humoral and cell mediated). APOBEC3 proteins have clear roles in conferring intrinsic host resistance to infection to different viruses as a result of their ability to directly act on viral genomes, thereby generating mutations or blocking viral nucleic acid synthesis. There is also a body of evidence, particularly in the case of retroviruses, that APOBEC3 deaminase activity plays additional antiviral roles in innate and adaptive immunity. Whether these additional mechanisms operate in the antiviral response to other virus families remains to be determined. Moreover, it will be important to demonstrate that these additional roles for APOBEC3 proteins function in the context of virus infection in vivo.

The authors have no financial conflicts of interest.