In the present study, we examined the cytotoxic capacity of HIV-specific CD8+
T-cells, in response to primary autologous HIV-infected CD4+
T-cell targets, in samples from HIV-negative individuals vaccinated with a replication-incompetent adenovirus serotype 5 HIV vaccine. In prior studies, immunization with this vaccine did not diminish the rate of HIV infection compared to placebo recipients, nor did it lead to reductions in plasma HIV RNA levels among newly infected persons 
. In our analyses, readily detectable HIV-specific CD8+
T-cell cytotoxic responses were observed in vaccinees based on measurements of GrB target cell activity and infected CD4+
T-cell elimination. Significant responses were present a median of 331 days following the last immunization, confirming that long-lived memory cells had been induced with this vaccine strategy. However, the recall cytotoxic capacity of the HIV-specific CD8+
T-cells of vaccinees was modest and overlapped with that of progressors. The magnitude of the cytotoxic responses was not related to the number of vaccinations, nor did it correlate with the percentages of cytokine-secreting T-cells determined by ICS assays. Importantly, we did observe higher cytotoxic responses in vaccine recipients carrying HLA class I alleles that have been associated with immune control over HIV replication, HLA B*27, B*57 or B*58.
Given the lack of an association between traditional measures of HIV-specific T-cell frequencies such as ELISPOT or ICS and immunologic control of HIV after vaccination, there has been increasing interest in measurements of other HIV-specific T-cell functions. Significant activity was observed with viral inhibition assays in recipients of a DNA prime/recombinant Ad5 boost vaccine regimen 
. Spentzou et al. observed relatively low but detectable inhibition by CD8+
T-cells from 7 participants that had received a DNA-recombinant Ad5 prime-boost regimen 
. Freel et al. observed low virus inhibition in 40 participants vaccinated with a similar DNA prime/recombinant Ad5 boost regimen 
. Many of these responses were below the level of detection and the responses of vaccinees were below those of progressors and nonprogressors. Of note, the response of chronic progressors was similar to those of virus controllers 
. Differences between these results and ours may be attributed to differences in vaccine regimen and cohort selection criteria 
. In addition, they may be attributable to the assays used. Although granule exocytosis-mediated killing may be an important contributor in virus inhibition assays, the latter may also measure the effects of CD8+
T-cell proliferation, chemokine or suppressor factor secretion, or cytotoxicity mediated by other mechanisms. Nonetheless, it will be important over the coming years to follow each of these assays for their ability to predict vaccine-induced immunologic control of HIV and for potential clues regarding the mechanism.
Although some HIV-specific CD8+
T-cell cytotoxic capacity was induced by the vaccine approach in the current study, the magnitude necessary for immunologic restriction of HIV remains unclear. Since participants were not infected after vaccination, it was reasonable to suspect their precursor frequencies were most likely lower than those of chronically infected patients. Although this vaccine approach induced pre-challenge levels of Mamu A01-p11CM MHC tetramer+
T-cells in the peripheral blood of rhesus macaques that reached 2% in the Ad5/SIV only arm and levels ranging from 5–25% when preceded by a DNA/CRL1005 prime, the Ad5/HIV vaccine induced only a median response of 0.4–1% HIV-specific CD8+
T-cells based upon ICS in humans in a prior study and 0.5–0.6% in the present study 
. However, it is clear that the modest cytotoxic capacity observed in vaccinees was not simply due to low frequencies of HIV-specific CD8+
T-cells. The frequencies of HIV-specific CD8+
T-cells did not correlate with cytotoxic capacity. Furthermore, some of the individuals with the lowest frequencies of HIV-specific T-cells exhibited the highest CD8+
T-cell cytotoxic responses several years after vaccination.
Another potential factor contributing to the diminished HIV-specific CD8+
T-cell-mediated cytotoxicity observed in vaccinees relative to LTNP might relate to response breadth. That is, there may be additional CD8+
T-cell responses in LTNP targeting proteins outside of the 3 genes included in the vaccine, which might have led to more efficient elimination of HIV-infected CD4+
T-cell targets. This seems unlikely to be the case, however, since the bulk of the cytokine secretory and proliferative responses measured in LTNP have been primarily directed against highly conserved epitopes contained within Nef, Gag and Pol, with only minimal contribution made by responses targeting other gene products 
In order to investigate the cytotoxic responses of the various groups in greater detail, we analyzed them in the context of the true measured E:T ratios following 6-day re-stimulation and found that the per-cell killing capacity for the group of vaccine recipients again remained only slightly higher than that of progressors, but significantly lower than that of LTNP. Although it is not necessarily expected that the CD8+ T-cells of vaccinees would achieve the cytotoxic capacity of chronically infected LTNP, the results of the present study suggest that the human immune response is capable of higher responses. The observation that vaccinee per-cell cytotoxic capacity increased at higher E:T ratios and became more divergent from that of progressors also suggests that induction of higher cytotoxic responses might be attainable with improved vaccine design strategies. These results suggest a threshold level of cytotoxic capacity might have been achieved only in a small subset of patients with protective alleles in this vaccination scheme. Further investigation of the cytotoxic responses directed against other viral infections that have been cleared or are controlled by the host may provide a better context in which to interpret the responses measured in this study. Most importantly, further evaluation of vaccinee cases may reveal whether these measurements of HIV-specific CD8+ T-cell cytotoxic capacity can accurately predict immune control over HIV following vaccination.
In the present study, we did observe a clear effect of protective HLA alleles in priming HIV-specific CD8+
T-cell cytotoxic capacity. However, whether these alleles, in the context of vaccination, are associated with qualitatively different CD8+
T-cell responses or immunologic control has been examined in some prior work with mixed results 
. Preliminary data from the Step study suggested that, after HIV infection, vaccinees with the protective HLA alleles B*57, B*58 and B*27 may have lower HIV RNA levels compared to those with protective alleles that received placebo (Nicole Frahm, personal communication). However, in a subsequent analysis, although those with protective alleles had lower viral loads overall, the difference between placebo and vaccinee cases with protective alleles did not achieve statistical significance (Nicole Frahm, personal communication). Kaslow et al observed that CD8+
T-cells restricted by protective alleles dominate the response to vaccination with an ALVAC-HIV recombinant canarypox 
. In addition, Freel et al. observed some increased HIV inhibition in cells from vaccinees with B*27 or B*57 compared to those who lacked these alleles. However, this was only true of NL4-3 and WEAU viruses 
. Taken together, these data suggest that protective alleles may function to more efficiently prime HIV-specific CD8+
T-cell cytotoxic capacity. The precise mechanism underlying this association, however, is currently unclear. CD8+
T-cell responses restricted by HLA B27 and 57 predominate during acute infection 
and are more likely to maintain the capacity to proliferate after prolonged infection than responses restricted by other alleles 
. Enhanced induction of CD8+
T-cell proliferative responses leading downstream to increased cytotoxic capacity might result from a greater ability of infected cells to stimulate naïve CD8+
T-cells. Alternatively, it might result from an enhanced ability of CD8+
T-cells to respond to infected cells due to interactions related to killer immunoglobulin-like receptor (KIR) ligation, co-stimulatory signal requirements, or a diverse naïve T-cell repertoire. In any event, whether the increased cytotoxic capacity observed in the present study translates into better immunologic control will need to be monitored in future trials.
Although the HIV-specific CD8+
T-cell cytotoxic capacity of vaccinees was less than that observed among LTNP, the cells of some vaccinees rapidly expanded and produced perforin in response to HIV-infected T-cells. In chronic infection, we have observed a very tight correlation between proliferative capacity and perforin expression as we did in the present study 
. However, a range of proliferative responses and perforin expression was observed in vaccinees that, in some cases, overlapped those of LTNP even though these cells did not exhibit cytotoxic responses of similar magnitudes as those of LTNP. It is unlikely that such a large fraction of vaccinees would exhibit immunologic control upon infection. Thus, proliferation or perforin expression, at least under these experimental conditions, are unlikely to be better predictors of immunologic control than cytotoxic capacity in the context of vaccination. The observation of cells with proliferative potential and the ability to make perforin similar to those of LTNP but with only modest cytotoxic capacity may relate to vaccine vector, dose, vaccine administration schedule and timing of post-vaccination PBMC collection for use in the immunologic assays. These might have an impact on the pattern of differentiation or maturation of HIV-specific CD8+
T-cells, which could influence direct measures of killing capacity. Additional studies are currently under way to further characterize the functional capabilities of these cells.
Although many challenges lie ahead, it appears possible that T-cell based immunogens may provide some activity to reduce viral load upon infection. Currently, antibody-based vaccine approaches are receiving increased emphasis with the reported reduction in HIV acquisition observed in the RV144 trial 
. However, an effective T-cell response would potentially complement an effective humoral immune response by reducing viral load in vaccinees that become infected 
. It may also have the effect of reducing the transmission of antibody resistant mutants. Immunologic control in chronically infected LTNP is durable and has lasted more than 25 years in many cases (reviewed in 
). In addition, many patients lack known protective MHC alleles suggesting these are not an absolute requirement for durable immunologic control 
. Evidence of durable immunologic control that extends beyond known protective alleles has also been provided by recent studies in the SIV infection model. Cellular immune responses elicited by infection with recombinant rhesus cytomegalovirus encoding SIV antigens can provide resistance to SIV infection on repeated limiting-dose intrarectal challenge 
. Ongoing studies are examining whether cytotoxic capacity is an important mechanism of immune control in the rhesus macaque-SIV infection model (Daniel Mendoza, unpublished observations). Importantly, our assay will be applied to participants in the Step study who subsequently acquired HIV infection post-vaccination, including a group of individuals who are restricting HIV replication to the limits of detection in currently available viral load assays. These types of analyses will begin to determine the utility of measurements of cytotoxic capacity in predicting vaccine efficacy, and provide further insight into the mechanisms of immunologic control of HIV.