A rare but important host response to HIV is exemplified by individuals (“elite controllers”) who are able to maintain undetectable HIV RNA levels in the absence of therapy for years to decades. Here, we demonstrate that durable control of HIV among elite controllers is strongly associated with the presence of Gag-specific CD4+
T cells and to a lesser extent Gag-specific CD8+
T cells. Our findings show that controllers are enriched for certain class I HLA alleles (i.e., HLA B*13, B*57, B*58, and B*81) and that the presence of these alleles predicts the Gag-specific CD8+
T-cell response. Importantly, we also demonstrate that strong Gag-specific T-cell responses are not present in a subgroup of individuals with excellent viral control and that many individuals control viral replication without “protective” HLA alleles or strong CD8+
T-cell responses. We have also shown that Pol-specific responses, though of a much lower magnitude than Gag-specific responses, are higher in the groups controlling viremia in the absence of therapy. Env-specific responses do not consistently correlate with viral control in this chronically HIV-infected cohort. These latter findings are consistent with findings in the simian immunodeficiency virus model where CD8+
T cells recognizing Gag and Pol epitopes were able to recognize infected cells earlier after infection compared to Env- or Nef-specific CD8+
T cells (51
), as well as in another recent report of elite controllers (46
). Collectively, these data provide strong evidence that HIV-specific T cells are associated with virus control in some, but not all, elite and low-level viremic controllers. It also suggests that HIV-specific T cells may not be necessary for viral control.
Why do some individuals maintain such tight viral control in the absence of therapy, whereas the vast majority of individuals exhibit high levels of viral replication? There are at least several potential mechanisms. One possibility is that elite controllers eradicated replication-competent virus upon first exposure and no longer harbor intact viral genomes. This would have left them seropositive but without active viral replication. In a recent report, Bailey et al. used ultrasensitive reverse transcription-PCR methods to find both plasma viral sequences and CD4+
T-cell-associated proviral DNA in most elite controllers with viral loads <50 copies/ml, arguing against this hypothesis (3
). A second possibility is that elite controllers harbor a replication-impaired virus and that the reduction in viral fitness/pathogenicity allows for a stronger and more effective host response. Such a scenario is reminiscent of our prior studies in treated patients with drug-resistant HIV (“partial controllers on ART”), wherein it appeared that treatment-associated reductions in viral fitness led to activation and preservation of HIV-specific immunity (12
). We have not formally ruled out this possibility, although others have shown that many controllers harbor replication-competent virus (6
). A third possibility is that although there is little viral replication within the peripheral blood compartment, there may be high-level replication in other compartments, such as in the gut or central nervous system, with subsequent maintenance of a potent HIV-specific response. The last possibility, and the one that we address here, is that virus replication is controlled by an effective host immune response.
Defining the role of HIV-specific T-cell responses in humans is challenging. It is difficult to infer whether strong HIV-specific responses result in viral control or whether low-level viral replication results in preservation of HIV-specific T-cell responses (28
). The more likely scenario is that there is a delicate balance between antigenic stimulation and T-cell responses. Low-level viral replication may allow for the maintenance of HIV-specific T cells that are capable of IL-2 production and proliferation. Whether the initial decrease in viral load is due to the presence of these T cells or other features of the immune system cannot be addressed with cross-sectional studies as presented here and elsewhere (14
). However, to address the issue of whether the presence of HIV-specific T cells is simply a consequence of viral control, we analyzed dual cytokine (IFN-γ and IL-2) responses in those who were ART suppressed and compared these individuals to our elite controllers. Despite low levels of viral replication in both groups, elite controllers consistently had higher levels of HIV-specific CD4+
that produce IFN-γ and IL-2 compared to ART-suppressed individuals. Similar trends were observed within the CD8+
T-cell subset. Many have argued that it is in fact viral control itself that allows for the presence of memory cells and thus IL-2 production, but the absence of HIV-specific IL-2 production among ART-suppressed individuals suggests that this is not always the case. It has also been suggested that the absence of T-cell responses in ART-suppressed individuals may be due to immunologic damage during previous high-level viremia (46
). However, this is unlikely to be the case given that, upon ART discontinuation or failure, HIV-specific T-cell responses return (18
One of the main findings of the present study is the fact that, while elite controllers have higher HIV-specific IFN-γ+
T-cell responses as a whole, there is significant heterogeneity within this group. Many subjects exhibit no or only limited HIV-specific T-cell responses. This heterogeneity among controllers is predicted by the class I HLA profile, supporting that HLA mediates its effect via a strong and potent CD8+
T-cell response (i.e., those with “protective” HLA class I alleles have higher frequency CD8+
T-cell responses than those without these alleles). However, approximately a third of the elite controllers in this cohort lacked a “protective” class I allele and had limited Gag-specific CD8+
T-cell responses but still were able to maintain excellent viral control. Furthermore, controllers with protective class I alleles did not have higher CD4+
T counts (data not shown). Thus, the presence of a “protective” HLA allele and a strong adaptive T-cell response does not appear to be necessary for achieving durable viral control. Of note, a recent report by Pereyra et al. failed to detect an association between the presence of certain HLA alleles and the magnitude of HIV-specific responses between elite controllers with protective HLA alleles. This lack of an association may be due to their use of an enzyme-linked immunospot assay (which only allows detection of a single cytokine) and the use of whole peripheral blood mononuclear cells (which does not allow for the delineation of a CD8+
T-cell response from a CD4+
T-cell response) (46
Elite controllers with “protective” class I alleles also had a higher CD8+
T-cell response than noncontrollers with these same alleles. Several possibilities arise to explain this finding. One possibility is that CTL escape in noncontrollers renders the class I-restricted CD8+
T-cell response less robust. Several groups have addressed this possibility among HLA-B*57 expressing individuals and failed to find a strong association between CTL escape mutations and either virus control or T-cell responses (3
). Another possibility is that other factors may be required in addition to a potent T-cell response. For instance, the combination of HLA-Bw4-80I (HLA B alleles, including HLAB*57, that encode molecules that have isoleucine at position 80) and killer cell immunoglobulin-like receptors (KIR) that exert a protective effect against AIDS progression (KIR3DS1 and KIR3DL1) (36
). This and other features of the innate immune response, some linked to class I alleles, may play an integral role in viral control. A third possibility is that there are factors that negate effects conferred by “protective” class I alleles. These may include the absence of a CD4+
T-cell response or altered function of other arms of the immune system. Regardless of the exact reason for this finding, it is clear that the mere presence of a “protective” class I HLA allele does not appear to be sufficient for effective viral control.
Several limitations with regard to our measurements deserve mention. First, the assays used in the present study focused on responses directed against a limited numbers of peptide pools (Gag, Pol, and Env), which were derived from consensus rather than autologous sequences. It is also possible, as recently suggested, that although CD8+
T cells may secrete cytokines in response to cells loaded with synthetic peptides, this may not correlate with the ability of CD8+
T cells to recognize the same epitope in the context of an HIV-infected cell (4
). Second, we measured responses in blood rather than in tissues, where most viral replication occurs. Third, we only measured a focused outcome: the ability to produce IFN-γ and IL-2 after limited exposure to peptides ex vivo. The use of additional cytokines could potentially have increased the antigen-specific response detected. We do not believe that these assay limitations negate our fundamental observation, which is that the levels of Gag- and Pol-specific T cells, as defined by the production of IL-2 and IFN-γ, are consistently higher in elite controllers than in other patient groups.
One of our primary conclusions pertains to the much higher T-cell response in those with undetectable viral loads in absence of therapy compared to those with undetectable viral loads in presence of therapy, which we argue is evidence that the host T-cell response is not simply a consequence of viral control. This argument assumes that the two groups have comparable levels of viral replication, which may not be true. The level of residual viral replication in “elite” controllers is not known and the degree to which virus replication occurs during effective ART remains highly controversial (10
). As has been observed with ART-suppressed patients (35
), most of our elite controllers have low but measurable HIV RNA levels, and there is no apparent association between the level of plasma viremia and the T-cell responses (unpublished data).
In summary, elite and viremic controllers are individuals with chronic HIV infection who are able to maintain viral replication at low levels in the absence of therapy. Here, we show that a robust HIV-specific T-cell response is maintained for many, but not all, elite controllers and that this response is linked to “protective” class I HLA alleles. However, a significant subset of controllers lacks evidence for T-cell control. The clear identification of such individuals using the approach outlined here may allow for more focused studies regarding novel mechanisms of control.