We show for the first time that HIV-1-positive elite suppressors are characterized by a PBMC miRNA profile that in general resembles that of uninfected individuals. However, we also find that ES, on the basis of miRNA expression, are not an entirely homogenous group, and that different mechanisms, shaped or marked by different miRNA expression patterns, underlie elite control in different individuals. We demonstrate further that miRNA expression is closely tied to CD4+ T-cell count in both ES and viremic patients, but that CD4 + T-cell counts do not fully explain the observed differences in miRNA profile. We describe significant differential expression of several miRNAs in control, ES, and viremic patients that have not previously been reported in connection with HIV-1 infection, e.g., miR-31 and its star partner, miR-31*. Confidence in our findings is bolstered by rigorous cross-validation by two profiling platforms, multiple-replicate testing by individual qPCR assays, and the use of multiple data normalization and analysis methods to ensure method-independence of results.
Despite the general proximity to controls, elite suppressors as a group display a pattern of expression for several specific miRNAs (e.g., miR-125b and miR-150) that closely resembles that of viremic patients. This finding gives rise to several hypotheses. The similar levels of some miRNAs in all infected samples may imply that immunologic responses to the presence of even low levels of virus and/or an immunologic imprint in the early phase of infection are sufficient to effect reduced expression of specific miRNAs. Alternatively, it is possible that elite suppressors, even prior to infection, maintain lower baseline levels of these miRNAs in comparison with the general population, and that this is protective or indicative of mechanisms of protection. These hypotheses should be tested in longitudinal studies.
A second major implication of our results for miRs-125b and -150 (as well as -382, although it does not reach significance) has to do with the reported role of these miRNAs in latency and control of viral replication . We find no negative correlation of these miRNAs with viral load of viremic patients. In fact, there is a positive trend for miR-125b and viral load. Although ES may have slightly higher levels of these miRNAs than viremic patients, the difference does not seem to be of sufficient magnitude to explain the profound gap in viral loads. It is of course possible that elite suppressors experience a significant decline of these miRNAs in the global PBMC population, but that a small number of specific cells--perhaps infected cells--experience an increase that keeps viral replication in check.
Conversely, the observed alterations in PBMC miRNA expression cannot be attributed solely to infected cells. For miRNAs that are less abundant in PBMC from viremic patients, downregulation must be due almost entirely to bystander effects. Only a small percentage of PBMC are infected in HIV-positive individuals. Thus, even if specific PBMC miRNAs were absent from infected cells, this difference would be undetectable in the overall PBMC population. For overexpressed miRNAs, as well, bystander effects would have to be invoked unless infected cells upregulated these miRNAs by orders of magnitude. Although possible, miRNA expression differences of hundreds- or thousands-fold are rarely observed in vivo or in vitro. We conclude that most of the expression differences we observe are the result of host responses to infection, indirect effects of viral products, or both.
In addition to miRNAs with known roles in general HIV-1 infection, we describe several previously unreported associations of miRNAs with HIV-1 infection and progression. For three reasons, we suggest that at least one of these miRNAs, miR-31, has widespread implications for the interplay of HIV and T cell identity in latency and disease progression. First, miR-31 expression differences distinguish cells that influence HIV disease: previous publications consistently report that T-regulatory cells contain lower levels of miR-31 than do other T-cells [38, 39]. Second, miR-31 is induced by CCAAT/enhancer binding protein beta (C/EBP beta) , specific isoforms of which can either inhibit or stimulate retroviral replication [41, 42]. Third, the validated target of miR-31, the special AT-rich binding protein 2 (SATB2) , is part of a gene regulatory family needed for differentiation within the T-cell lineage. These observations, along with the consistent and profound downmodulation of miR-31 in PBMC of viremic patients and, to a lesser extent, in cells of some elite suppressors, demand experimentation on how the presence and absence of this miRNA affect T-cell differentiation and activation in the context of HIV infection and latency.
Our data suggest that changes in the levels of miR-31, miR-150, and others in PBMC during infection are the result of regulation and are not entirely (if at all) due to CD4+ T-cell decline. This conclusion is strongly supported by a recent study of miRNA profiles in specific blood cell populations . With the possible exception of miR-125b, miRNAs that are positively correlated with CD4+ T-cell count in our work are not found exclusively in CD4+ T-cells or naïve T-cells and thus could not be expected to decline to the observed extent even with complete CD4+ T-cell depletion. Indeed, some miRNAs that have been found at high levels across PBMC subsets (miR-16) or are even enriched in CD4+ T-cell subsets (miR-181 family members, miR-130b) --and would therefore be expected to decline along with CD4+ T-cells--are, to the contrary, negatively correlated with CD4+ T-cell count. Further work is needed to characterize the effects of HIV and immune responses on miRNA expression in PBMC and specific cell types both in vivo and, as far as possible, in primary culture ex vivo.
Several pioneering studies have already initiated the process of answering these questions, foremost among them the work of Houzet, et al.  and Huang, et al. . Building on earlier investigations of HIV and miRNA using HIV-1-transfected or -infected cell lines [16, 17], Houzet, et al. reported differential regulation of miRNAs in PBMC of HIV-infected, viremic individuals. The direction of this regulation was primarily downward, consistent with our results. Perhaps most importantly, our groups both observed downregulation of miR-150 and miR-29 family members, which have been characterized as miRNAs with HIV regulatory roles. Although Houzet, et al., observed a greater number of significant differences between controls and infected individuals than we report here, we do not see this difference as a discrepancy, due to the sizes of the respective studies, different study design, and use of different profiling platforms. Instead, the similarities of our finding--especially regarding latency-associated miRNAs--are telling and important, and should prompt additional follow-up studies.
Regarding the Huang, et al. study, we find significant and consistent downregulation of two of five miRNAs (miRs -125b and -150) that these authors found to be associated with control of HIV-1 transcripts in CD4+ T-cells and downregulated in CD4+ T-cells of HIV-1 patients . We also observed that another of these five previously reported miRNAs, miR-382, was downregulated approximately two-fold, but not significantly, in PBMC of both ES and viremic patients (not shown). Interestingly, although the five miRNAs from the Huang, et al. CD4+ T-cell study were reported to be downregulated during differentiation of monocytes to macrophages in a follow-up study , another group observed downmodulation of only one of these five, miR-223 . In the future, multiple validation techniques and analysis of larger numbers of samples will help to address the important issue of miRNAs involved in HIV control and latency in specific target cell types.
Of potential clinical importance, this study suggests that miRNA profiles may serve as biomarkers for the identification of ES who could benefit from closer monitoring and/or antiretroviral treatment (ART). Among the ES examined here, three had distinctly lower CD4+ T-cell counts. ES with low CD4+ T-cell counts have been described previously [44–46]; one elite suppressor with a low CD4+ T-cell count developed Kaposi's Sarcoma despite undetectable viral load (46). For some ES with declining CD4+ T-cell counts, ART has been initiated, in some cases with positive CD4 response [44, 47]. On the basis of the individual qPCR results presented here, two of the three elite suppressors with low CD4+ T-cell counts are unambiguously clustered with the viremic patients. It will be important to expand our investigations to discover additional correlates of miRNA expression and to determine whether miRNA profiles predict clinical decline, facilitating the decision of when treatment should be initiated in ES with declining CD4+ T-cell counts.