Hierarchical targeting of subtype C human immunodeficiency virus type 1 proteins by CD8+ T cells: correlation with viral load

Abstract

An understanding of the relationship between the breadth and magnitude of T-cell epitope responses and viral loads is important for the design of effective vaccines. For this study, we screened a cohort of 46 subtype C human immunodeficiency virus type 1 (HIV-1)-infected individuals for T-cell responses against a panel of peptides corresponding to the complete subtype C genome. We used a gamma interferon ELISPOT assay to explore the hypothesis that patterns of T-cell responses across the expressed HIV-1 genome correlate with viral control. The estimated median time from seroconversion to response for the cohort was 13 months, and the order of cumulative T-cell responses against HIV proteins was as follows: Nef > Gag > Pol > Env > Vif > Rev > Vpr > Tat > Vpu. Nef was the most intensely targeted protein, with 97.5% of the epitopes being clustered within 119 amino acids, constituting almost one-third of the responses across the expressed genome. The second most targeted region was p24, comprising 17% of the responses. There was no correlation between viral load and the breadth of responses, but there was a weak positive correlation (r = 0.297; P = 0.034) between viral load and the total magnitude of responses, implying that the magnitude of T-cell recognition did not contribute to viral control. When hierarchical patterns of recognition were correlated with the viral load, preferential targeting of Gag was significantly (r = 0.445; P = 0.0025) associated with viral control. These data suggest that preferential targeting of Gag epitopes, rather than the breadth or magnitude of the response across the genome, may be an important marker of immune efficacy. These data have significance for the design of vaccines and for interpretation of vaccine-induced responses.

FIG. 1.

Distribution of responses from HIV-1-seronegative individuals to each of the peptide pools (Gag to Nef), showing the boundaries of the mean SFU/10⁶ PBMC and 4 standard deviations (dotted lines demarcated by arrow), with the upper threshold being 104 SFU/10⁶ PBMC.

FIG. 2.

Distribution of responses across complete peptide pool sets, showing medians with the 10th and 90th percentiles as well as individual plots. The proportion of individuals responding to each peptide set is shown above the plot, and the significance levels between regions are shown below the plot. Significance was measured by Kruskal-Wallis nonparametric analysis of variance and Dunn's pairwise analysis. n.s., not significant.

FIG. 3.

Responses to peptide pools. (A) Cumulative SFU/10⁶ PBMC responses to each of the peptide pools. (B) Frequencies of individuals recognizing each of the peptide pools, depicted as amino acid numbers, to show more detailed recognition of sites within pools. Consecutive amino acid numbers are shown across all pools, amounting to a total of 3,091 aa, corresponding to the complete expressed genome.

FIG. 4.

Correlations between log₁₀ SFU/10⁶ PBMC values and log₁₀ plasma viral loads showing total responses to all peptide pools (A), Gag pools (B), Pol pools (C), Vif pools (D), the Vpr pool (E), the Tat pool (F), the Rev pool (G), the Vpu pool (H), Env pools (I), and Nef pools (J). Correlations were corrected for multiple tests by Bonferroni's analysis.

FIG. 5.

Correlation between peptide or protein recognition and viral loads. (A) Correlation between the numbers of positive peptide pools recognized per individual (n = 44) and log₁₀ plasma viral loads. Each peptide pool represents a conservative estimate of the number of epitopes recognized. (B) Correlation between the numbers of complete proteins recognized per individual (n = 44) and log₁₀ plasma viral loads. Each region corresponds to one of the nine protein regions, and the numbers recognized measure the breadth of the response across the genome.

FIG. 6.

(A) Cumulative SFU/10⁶ PBMC response to complete sets of peptides (corresponding to whole protein regions) ranked as a hierarchy from highest to lowest recognition. (B) Cumulative response adjusted for protein length ranked as a hierarchy from highest to lowest recognition. (C) Correlation between the hierarchy of responses to Gag and the log₁₀ plasma viral load for both the unadjusted and adjusted ranking for each individual. (D) Correlation for Pol. (E) Correlation for Env. (F) Correlation for Nef. Thick short lines correspond to the unadjusted hierarchical responses, and longer lines correspond to the adjusted hierarchical responses.