|Home | About | Journals | Submit | Contact Us | Français|
APOBEC proteins have evolved as innate defenses against retroviral infections. Human immunodeficiency virus (HIV) encodes the Vif protein to evade human APOBEC3G; however, mouse retroviruses do not encode a Vif homologue, and it has not been understood how they evade mouse APOBEC3. We report here a murine leukemia virus (MuLV) that utilizes its glycosylated Gag protein (gGag) to evade APOBEC3. gGag is critical for infection of in vitro cell lines in the presence of APOBEC3. Furthermore, a gGag-deficient virus restricted for replication in wild-type mice replicates efficiently in APOBEC3 knockout mice, implying a novel role of gGag in circumventing the action of APOBEC3 in vivo.
APOBEC3G (hA3G) in humans and its mouse orthologue, APOBEC3 (mA3), act as potent innate defenses against retroviral infection. Both proteins deaminate cytidine in single-stranded DNA, ultimately resulting in hypermutation of newly synthesized proviral DNA (6, 16), although additional deaminase-independent mechanisms of inhibition have been identified (2). Infectious exogenous retroviruses, including human immunodeficiency virus (HIV) and murine leukemia viruses (MuLVs), have evolved mechanisms to circumvent the action of the APOBEC proteins (3, 6). HIV encodes the Vif protein, which facilitates the rapid proteolysis of hA3G, while the mechanism by which exogenous MuLVs evade the action of mA3 is unknown (6).
Exogenous MuLVs, as well as some other gammaretroviruses, encode a glycosylated Gag protein (gGag) originating from an alternate translation start site upstream of the methionine start site of the Gag structural polyproteins (10, 17, 27). gGag is synthesized at similar rates and levels as the structural Gag polyprotein in MuLV-infected cells but is glycosylated and undergoes distinct proteolytic processing (10, 12, 21). A carboxyl fragment of gGag is released from the cell, while an amino fragment is incorporated into the plasma membrane as a type 2 transmembrane protein (12, 25). The functions of gGag remain unclear, but mutations that eliminate its synthesis severely impede in vivo replication of the virus with little, if any, effect on replication in fibroblastic cell lines (7, 19, 26). APOBEC3 proteins are expressed in many tissues in vivo but are poorly expressed in many in vitro cell lines (6), suggesting a possible link between gGag expression and the evasion of mA3 by MuLVs. These studies were undertaken to determine if the expression of the gGag protein facilitated MuLV replication in the presence of mA3 in vitro and in vivo.
Several studies of the effects of mA3 proteins on virus replication examined the infectivity of virions released from cells transfected with cloned proviral DNA in the presence or absence of mA3 (9, 16, 28). Such analyses do not test effects of mA3 present in the cytoplasm of the cells on the infectivity of MuLVs that have not been previously exposed to mA3. To address this issue, we developed an NIH 3T3 cell line (3T3) that expressed a hemagglutinin (HA)-tagged full-length mA3 protein (3T3/mA3) corresponding to the BALB/c allele (4). The infectivity of the two coisogenic MuLVs, CasFrKP (gGag+) and CasFr-3 + 4 (gGag−) (26), in 3T3 and 3T3/mA3 cells was compared using mixtures of the viruses, each carrying a distinct retroviral vector encoding either alkaline phosphatase (LAPSN) or β-galactosidase (G1nβgSvNa), as previously described (14). Experiments were performed using both MuLV-vector combinations. These analyses indicated that cellular mA3 exerted a marked inhibitory effect on the infectivity of gGag− virus but not on the gGag+ virus (Fig. (Fig.1).1). Interestingly, it was recently reported that infection by Moloney MuLV (M-MuLV) (20) and mouse mammary tumor virus (22) was partially inhibited by mA3 and that both virion and cellular mA3 contributed to the inhibition. Furthermore, HIV has also been reported to be inhibited by cytoplasmic hA3G (35).
It is somewhat surprising that cellular mA3 exerts a gGag-dependent effect on infecting MuLVs. The low level of virion-associated gGag may directly influence the action of cellular mA3; however, virion gGag is likely associated with the viral envelope as a type 2 transmembrane protein, and it is difficult to envision how it might interact with cellular mA3. Alternatively, the susceptibility of the gGag− MuLV to cellular mA3 may occur by an indirect mechanism. In this regard, it has been reported that gGag is involved in virion release and that gGag− M-MuLV exhibits an abnormal morphology during virion budding (19). It is conceivable that mature virions may also be altered from an mA3-resistant to an mA3-susceptible phenotype.
A number of studies have reported partial inhibition of ecotropic MuLVs as a result of incorporation of mA3 into progeny virions (16, 20, 28, 33). Indeed, it has been suggested that MuLVs may evade the action of mA3 by exclusion of the protein from virions, although there are conflicting accounts regarding this matter (4, 8, 15, 16, 24, 28). To determine if the presence of gGag influenced the incorporation of mA3, virions were isolated by isopycnic gradient centrifugation (10) from mA3 cells infected with gGag+ or gGag− MuLVs and examined by immunoblot analyses using monoclonal antibodies directly conjugated to horseradish peroxidase. mA3 was readily detected in both virus preparations, with no discernible differences in the levels of virion incorporation in gGag+ or gGag− MuLVs (Fig. (Fig.2A).2A). This result is in agreement with other reports that have shown incorporation of mA3 into virions (4, 15, 24, 28). Furthermore, gGag was also found to be incorporated into virions (Fig. (Fig.2B),2B), consistent with an earlier study (11). Thus, inhibition of mA3-mediated antiviral activity by gGag does not occur simply by preventing incorporation of mA3 into virions.
To determine if virion-incorporated mA3 differentially influenced gGag+ and gGag− MuLVs, we examined the infectivity of viruses released from 3T3/mA3 cells as well as from 3T3 cells lacking mA3. Both cell lines were transduced with the retroviral vector LAPSN, which encodes alkaline phosphatase, to enable the quantification of progeny virus infectivity in single-cycle assays as previously described (14). Cells were infected with gGag+ or gGag− MuLVs, and the infectivity of released viruses was quantified by alkaline phosphatase as well as by focal immunofluorescence assays (31) on mA3−/− Mus dunni cells. Infectivity was normalized to the number of progeny virions released using a colorimetric reverse transcriptase assay (Roche). The retroviral vector assays and the fluorescence assays closely paralleled one another, and their results were combined (Fig. (Fig.2C).2C). These analyses revealed that the specific infectivity of gGag− virus released from 3T3/mA3 cells was markedly decreased compared to that of gGag− virus released from 3T3 cells. In contrast, no decrease in infectivity was observed with the gGag+ virus released from 3T3/mA3 cells compared to 3T3 cells (Fig. (Fig.2C).2C). Experiments using an M. dunni cell line expressing the mA3 protein to assess the effects of cellular as well as virion-incorporated mA3 on the infectivity of gGag+ and gGag− MuLVs yielded similar results (data not shown).
Our analyses indicated that mA3 did not inhibit the gGag+ MuLV; however, a number of studies have reported partial to marked inhibition of other MuLVs (16, 20, 28, 33), all of which encode a gGag protein. A direct comparison of the inhibitory effects of mA3 on M-MuLV and the ecotropic AKV MuLV revealed that AKV was inhibited to a greater extent than M-MuLV (16). Differences in the susceptibility of MuLVs to inhibition by mA3 could reflect differences in the efficacy of their respective gGags to counteract mA3. In this regard, a comparison of the gGag sequences of M-MuLV and AKV reveals extensive amino acid differences in their amino-terminal fragments. Recent studies have also reported that mA3 inhibits the replication of xenotropic murine leukemia-like retrovirus (XMRV) to a much greater extent than M-MuLV (13, 24). All XMRV isolates exhibit a termination codon in the coding sequences of gGag, resulting in a truncation of the protein immediately preceding the transmembrane region (34), suggesting that the sensitivity of XMRV to mA3 may be gGag dependent.
If mA3 restriction is a major factor influencing in vivo replication of MuLVs and its action is sufficiently repressed by gGag, it would be expected that mice lacking mA3 would be permissive to infection by both gGag+ and gGag− MuLVs. To examine this possibility, we determined the level of replication of gGag+ and gGag− MuLVs in mA3 knockout mice (mA3−/−) and their wild-type counterparts (mA3+/+) (29). Comparisons of the replication of gGag+ and gGag− MuLVs revealed a clear influence of 129/Ola mA3 on their replication (Fig. (Fig.3A).3A). In agreement with previous studies on the replication of gGag-deficient mutants (7, 20, 26), the replication of the gGag− MuLV was severely restricted in normal 129/Ola mice. However, in 129/Ola mice lacking mA3, the gGag+ and gGag− MuLVs replicated to equally high levels. These results indicate that the inability of the gGag− MuLV to replicate efficiently in vivo is the result of mA3 expression.
C57BL/6 mice contain the FV-1b allele (32), whereas the MuLVs used in this study are n-tropic. Thus, the levels of replication of the MuLVs in C57BL/6 mice were much lower than those in the 129/Ola mice, which contain the FV-1nr allele (Fig. (Fig.3B)3B) (32) (M. L. Santiago, unpublished results). Nevertheless, replication in these mice was sufficient to observe mA3 inhibition in a gGag-dependent manner. Levels of the gGag− MuLV were restored to the levels observed with the gGag+ MuLV in C57BL/6 knockout mice lacking mA3. It is noteworthy that C57BL/6 mice predominantly express a splice variant mA3 mRNA which lacks exon 5 (1, 23, 29), while 129/Ola mice predominantly express a complete mA3 mRNA. Our results indicate that the MuLV gGag studied here is able to suppress the antiviral effect of both the full-length and exon 5-deleted proteins and further substantiate the role of gGag as an antagonist of the restriction factor.
The studies presented here provide at least partial answers to two difficult questions in retrovirology: those of the function of the gGag of MuLVs and the means by which MuLVs evade the action of APOBEC3. Although gGag of exogenous MuLVs carries out a function similar to that of the Vif protein of HIV, further studies are required to determine similarities and differences in their modes of action. Such studies are particularly relevant in light of recent reports indicating cross-species retroviral infections from mice to humans (18, 30, 34).
This research was supported by the Intramural Research Program of the NIH, NIAID.
Mice were treated in accordance with the regulations and guidelines of the Animal Care and Use Committee of the National Institutes of Health.
We thank Bruce Chesebro, Byron Caughey, Jay Carroll, Lara Myers, and Amanda Duley for helpful discussions and Dan Littman for providing the plasmid encoding mA3.
Published ahead of print on 11 August 2010.