Members of the APOBEC family of deaminases, such as A3G and A3F, have been described as potent retrovirus restriction factors, capable of inhibiting replication of several viruses including HIV-1 in vitro [4, 39–41]. In rodents, it has been clearly demonstrated that APOBEC3 contributes to restriction of Friend MuLV infection in vivo [42, 43]. The role of APOBEC3 in the pathogenesis of AIDS, however, is still under debate [24, 25, 27, 44, 45]. Therefore, we determined A3G and A3F transcription in SIV-infected rhesus macaques and linked it to plasma viral load and disease progression.
Compared with uninfected control animals, we found increased mRNA levels of both A3G and A3F in SIV-infected monkeys during the asymptomatic phase of the disease. So far, several studies have investigated A3G transcription in HIV-infected subjects, however with inconsistent results [25–27, 44, 46, 47]. Whereas some reported increased levels of A3G in HIV-infected subjects [25, 44], others found lower A3G mRNA compared with uninfected individuals [27, 47]. In the few studies on A3F, similarly discrepant results were observed [25, 27]. Some of the inconsistencies might be attributed to methodological differences as both fresh and cryopreserved cells with or without polyclonal stimulation were used. However, the wide variation between infected individuals, also seen in our study, suggests that the most important source of the conflicting data is likely differences in the selection of patients with regard to disease stage and viral load. In the present study, animals that had progressed to AIDS had lower A3G and A3F levels in blood than asymptomatic macaques. Although an association between disease stage in humans and APOBEC3 transcription has not been investigated explicitly, this is corroborated by recent findings of very low A3G mRNA levels in patients with CD4 counts below 200 . On the other extreme, as reported for HIV-patients , we observed the highest mRNA levels among asymptomatic animals in a group of LTNPs, representing animals that had survived for more than three years after infection. This association between APOBEC3 mRNA and disease progression is substantiated by our finding of a negative correlation between APOBEC3 levels and plasma viral load. These results are in agreement with several previous reports, which documented higher A3G [25, 26, 46] and A3F  mRNA levels in PBMC of patients with lower viral load. By contrast, another study did not find a correlation between APOBEC3 mRNA and viral load . As observed previously , APOBEC3 mRNA copies in individuals with high viral load were often below the levels of uninfected controls. These findings may explain the decreased A3G levels in PBMC of infected compared with uninfected individuals reported by some studies, probably due to differences in the composition of the patient groups.
Our study is the first one, which quantified APOBEC3 mRNA in the context of an immunodeficiency virus infection not only in whole PBMC, but also in leukocyte subsets susceptible to infection such as CD4+ T-cells and CD14+ monocytes as well as in the lymph nodes as major virus replication sites. Despite slight variations, the same tendencies were observed. A3G and A3F mRNA levels correlated negatively with viral load and the highest values were observed in LTNPs. Interestingly, the dichotomy in A3G transcription of individuals with higher or with lower A3G levels than uninfected controls was more pronounced in CD4+ T-cells possibly due to differences in the cellular composition. In general, mRNA levels determined in PBMC mirrored the situation in CD4+ T-cells or monocytes. However, this does not reflect the conditions in lymph nodes. In lymph nodes from two different anatomical regions, we also detected high amounts of A3G and A3F mRNA in LTNPs. In contrast to blood however, APOBEC3 levels in lymph nodes were still increased in animals with AIDS. This might be explained by the fact that AIDS-associated immunological alterations occur later in lymphatic tissues than in blood [48, 49]. Since we sacrificed the animals when first signs of AIDS appeared, the situation in lymph nodes may still be more similar to the asymptomatic phase. Also differences in the cellular composition or in the microenvironment between the different compartments  could be responsible as APOBEC3 transcription varies between different cell types and tissues [21, 51, 52] and can be influenced by interferons, cytokines and chemokines [17, 19, 20, 22, 37, 51]. To our knowledge, APOBEC3 mRNA in lymph nodes of HIV-patients has yet not been studied to substantiate our results with findings from the human setting.
Our data clearly demonstrate a negative correlation of A3G and A3F mRNA levels with viral load and an association of high A3G and A3F mRNA levels with prolonged survival, suggesting a possible effect of APOBEC3 on viral replication in vivo favorably influencing disease progression. It has been hypothesized, that the higher A3G levels in LTNPs or patients with low viral load are due to preexisting differences in gene expression . However, our longitudinal studies suggest that A3G transcription is actually up-regulated after infection. This is in line with the results of a recent publication where A3G transcription was compared in the same individuals before and after infection . In addition, we found coordinated transcription of A3G and A3F in all cell populations investigated only in SIV-infected animals, which points to common regulatory mechanisms induced by the infection. Interestingly, increased A3G levels have also been reported in patients with HCV-infection . General to viral infections is a potent induction of type one interferons and several studies have demonstrated stimulation of A3G and A3F expression by interferons in vitro [15, 37, 54, 55]. Therefore, we have determined the transcription of conventionally used prototype interferon-stimulated genes to assess interferon activity in our animals. As reported previously for both HIV-infected humans and SIV-infected macaques [56–58], we observed an early and strong rise of MxA mRNA, which reached its maximum ten days after infection, hence preceding peak viral load. This up-regulation of ISG was paralleled by a more moderate increase in A3G mRNA levels. With resolution of peak viremia both MxA and A3G levels declined but remained slightly above preinfection levels in both PBMC and lymph node cells. These parallel kinetics indicate that APOBEC3 deaminases are downstream effector molecules of a type I interferon response in acute viral infections. In the chronic phase of SIV-infection, however, we found a divergent expression of prototype ISG and A3G/A3F. Whereas MxA and IP-10 transcription increased with higher viral replication, A3G and A3F mRNA levels were negatively correlated with plasma viral load in SIV-infected asymptomatic animals. Thus, other mechanisms must be responsible for the higher APOBEC3 levels found in asymptomatic animals, especially in LTNPs. As acute phase and end stage disease are characterized by a lack of virus specific immunity and LTNPs show a better specific immune response than progressors , mediators associated with a potent immune reaction are likely candidates for this additional APOBEC3 modulation in asymptomatic animals. Cytokines like IL-2 and IL-15 have been shown to induce A3G expression in vitro [19, 22], PBMC from LTNPs produce higher amounts of IL-2  and adjuvanting vaccine vectors with IL-15 improves protection against pathogenic challenge in the SIV/macaque model, partly through an induction of A3G . In this context it would be interesting to explore the effect of HAART on APOBEC3 levels as effective antiretroviral treatment has been shown to restore IL-2 production by antigen specific cells  and IL-15 levels correlated with better outcome of structured treatment interruption . In addition, it has been reported that cross-talk between dendritic cells and T-cells also results in the induction of A3G expression by both contact dependant mechanisms via CD40-CD40L interaction and through soluble mediators such as IL-15 . Dendritic cells are depleted in progressive HIV-infection , but accumulate in lymph nodes of LTNPs exhibiting elevated CD40 expression . Increased signaling through CD40 ligand may thus be responsible for the higher A3G expression in the asymptomatic phase. Consistent with the reported loss of CD40 expressing dendritic cells in AIDS patients , this regulatory pathway attenuates leading to the lower APOBEC3 transcription, which we observed in PBMC of animals with symptoms of AIDS. At this stage of the infection, A3G and A3F mRNA levels once again seem to be governed solely by interferons as they correlated with the transcription of ISGs.
An alternative explanation for the disease stage specific relationship between APOBEC3 and ISG expression could be shifts in the composition of cells. Some groups have reported a stronger induction of A3G and A3F transcription by IFN-alpha for macrophages compared with CD4+ T-cells [21, 51, 55] and for resting compared with activated CD4+ T-cells . However, as we see the same disease stage specific pattern for all cell types investigated, such differences in sensitivity to interferon stimulation probably do not contribute to the discrepancy between APOBEC3 and ISG transcription in asymptomatic animals.
In summary, we think that increased amounts of interferons in SIV-infected macaques influence APOBEC3 mRNA levels throughout the infection. However, as evidenced by the much lower induction of A3G during peak viremia compared with prototype ISG, the effect of interferons on the transcriptional control of APOBEC3 deaminases in vivo is moderate. This interferon-mediated basic regulation is then overlaid by additional modulatory mechanisms, which need time to build up and are strongest in LTNPs, but ultimately vanish during AIDS. Such mechanisms may also potentially become activated in uninfected individuals as various vaccination regimens can induce a long lasting upregulation of A3G in macaques [22, 65]. In addition, several studies have documented increased A3G mRNA levels in HIV-1 exposed but seronegative individuals [26, 44], who are known to mount a subtle HIV-specific cellular immune response . Interestingly, cessation of exposure was associated with both decreased A3G levels  and with a loss of anti-HIV-1 T-cell response . Exploring the mechanisms leading to increased APOBEC3 levels in LTNPs or HIV-1 exposed but seronegative individuals should provide useful information for new therapeutic approaches.
The results from this and previous studies suggest that increased APOBEC3 mRNA levels are associated with lower viral load and slower disease progression. According to our data, A3G and A3F mRNA levels are increased also in CD4+ lymphocytes and CD14+ monocytes, the cellular targets of HIV, and in lymph nodes where the majority of CD4+ T-cells resides and where abundant viral replication takes place. Information on other tissues with high viral load such as mucosa associated lymphoid tissue is sparse. So far, only cervical biopsies have been investigated where A3G levels were comparable between infected and uninfected individuals .
Evidence that APOBEC3 actually exerts its deaminase activity in vivo is derived from studies, which found the typical G to A mutations imprinted on the viral genomes in HIV-infected patients [12, 26, 68]. The extent of these characteristic mutations correlated with A3G transcription , indicating that increased APOBEC3 expression may overcome the restraints imposed by Vif . Moreover, higher hypermutation rates conforming to the A3G and A3F sequence preferences were associated with lower viral load and higher CD4 cell counts [12, 68]. On the other hand, increased A3G/A3F expression may be beneficial even without demonstrable hypermutations as APOBEC3 enzymes can restrict viral replication through mechanism distinct from cytidine deamination [10, 11, 69].