Temporal Dynamics of CD8+ T Cell Effector Responses during Primary HIV Infection

Abstract

The loss of HIV-specific CD8+ T cell cytolytic function is a primary factor underlying progressive HIV infection, but whether HIV-specific CD8+ T cells initially possess cytolytic effector capacity, and when and why this may be lost during infection, is unclear. Here, we assessed CD8+ T cell functional evolution from primary to chronic HIV infection. We observed a profound expansion of perforin+ CD8+ T cells immediately following HIV infection that quickly waned after acute viremia resolution. Selective expression of the effector-associated transcription factors T-bet and eomesodermin in cytokine-producing HIV-specific CD8+ T cells differentiated HIV-specific from bulk memory CD8+ T cell effector expansion. As infection progressed expression of perforin was maintained in HIV-specific CD8+ T cells with high levels of T-bet, but not necessarily in the population of T-betLo HIV-specific CD8+ T cells that expand as infection progresses. Together, these data demonstrate that while HIV-specific CD8+ T cells in acute HIV infection initially possess cytolytic potential, progressive transcriptional dysregulation leads to the reduced CD8+ T cell perforin expression characteristic of chronic HIV infection.

Fig 1. Timing of study participant samples and dynamics of HIV plasma viral loads, CD4⁺ T cell counts and CD8⁺ T cell counts.

(A) Timing of samples relative to estimated time of infection for the three acute/early HIV cohorts: CHAVI (purple), Montreal (blue), and RV217 (red). Plasma HIV RNA copies/ml (B), absolute CD4⁺ T cells counts (C), and CD8⁺ T cell counts for all 32 donors (D). Lowess smoothers were used to represent the mean over time for longitudinal data.

Fig 2. Memory distributions, activation, and proportion of cytotoxic peripheral CD8⁺ T cells for healthy donors and following HIV infection.

(A) Representative flow cytometric plots of CCR7 versus CD45RO from a pre- (day -159) and acute (day 28) infection time point for one donor. (B) Proportion of circulating total memory CD8⁺ T cells for all healthy donors (HD; n = 41), acute HIV time points (23–100 d; n = 27), and chronic HIV time points (>365 d; n = 23). (C) Peak viral load plotted against total memory CD8⁺ T cells at the earliest available time point post-infection (23–41 d) for each RV217 donor. Spearman’s rank correlation test was used to determine significance. (D) Memory subsets as determined by CCR7 and CD45RO staining for all healthy donors (circles), acute HIV time points (squares), and chronic HIV time points (triangles). (E) Proportion of memory CD8⁺ T cells that express HLA-DR over time from infection for four RV217 subjects. Pre-infection time points were set as day 0 for analysis. (F) Representative flow cytometric plots of perforin and HLA-DR expression by memory CD8⁺ T cells from a pre- (day -173), acute (day 31), and chronic (day 420) infection time point for one donor. § Day 420 sample was acquired and analyzed at a later date than earlier samples resulting in a different gating scheme. For consistency, gates were set using naïve (CCR7⁺CD45RO^-) CD8⁺ T cells, which generally do not express perforin or HLA-DR. (G) Proportion of memory CD8⁺ T cells that express perforin for healthy donors (HD), acute HIV time points (23–100 days), and chronic HIV time points (> 365 days). All data represent direct ex vivo assessment with no in vitro stimulation. *** denotes a P value < 0.001. Statistics based on a GEE population-averaged model with Holm adjusted P value. Bars represent approximations of the means generated by the models.

Fig 3. Dynamics of the total CD8+ T cell memory pool and cytotoxic response following infection with HIV, vaccinia virus, yellow fever virus, or influenza.

Proportion of total memory CD8⁺ T cells for longitudinal time points from donors either naturally infected with HIV (n = 11; A), vaccinated with attenuated vaccinia virus (Dryvax, n = 8; B), vaccinated with live yellow fever virus (YFV-17D, n = 10; C), or experimentally infected with influenza (strain H1N1, n = 10; D). Frequency of memory CD8⁺ T cells that express perforin following infection with HIV (E), vaccinia (F), yellow fever (G) or influenza (H). Pre = pre-infection or pre-vaccination time points. Pre-infection time points for HIV, vaccinia, and YFV, were set as day 0 for analysis. All data represent direct ex vivo assessment with no in vitro stimulation. * denotes a P value < 0.05. Statistics based on a GEE population-averaged model with Holm adjusted P value. Bars represent approximations of the means generated by the models. Lowess smoothers were used to represent the mean over time for longitudinal data.

Fig 4. Magnitude and functionality of HIV-specific responses over time.

(A) Frequency of Gag-specific CD8⁺ T cells within the memory CD8⁺ T cell pool over time as determined by measurement of IFN-γ expression or degranulation (CD107a) in response to peptide stimulation (n = 32). (B) Memory distributions for Gag-specific CD8⁺ T cells (red) overlaid on total CD8⁺ T cells (black) for a representative donor. (C) Memory distributions for responding Gag-specific CD8⁺ T cells as determined by CCR7 and CD45RO staining for acute (squares; n = 23), and chronic (triangles; n = 23) HIV time points. (D) Proportion of total responding Gag-specific CD8⁺ T cells that have upregulated IFN-γ or degranulated at acute and chronic time points. (E) Gag-specific CD8⁺ T cells (red) overlaid on total memory CD8⁺ T cells (black) for a representative donor. Percentages represent frequency of responding Gag-specific cells within a quadrant. § Day 420 sample was acquired and analyzed at a later date than earlier samples resulting in a different gating scheme. For consistency gates were set using naïve (CCR7⁺CD45RO^-) CD8⁺ T cells. (F) Proportion of total responding Gag-specific CD8⁺ T cells that upregulated perforin in response to peptide stimulation (n = 28). * denotes a P value < 0.05 and ** denotes a P value < 0.01. Statistics based on a GEE population-averaged model with Holm adjusted P value or random-effects tobit regression. Bars represent approximations of the means generated by the models. Lowess smoothers were used to represent the mean over time for longitudinal data.

Fig 5. T-bet and Eomes expression by total perforin⁺ CD8⁺ T cells for healthy donors and following HIV infection.

(A) Representative flow cytometric plots of T-bet and Eomes expression for total memory (top and middle rows) and perforin⁺ (bottom row) CD8⁺ T cells from the pre- (day -159), acute (day 28), and chronic (day 434) infection time points for one donor. Percentages on top and middle plots are of perforin⁺ cells within the total memory pool. (B) T-bet and Eomes expression by perforin⁺ CD8⁺ T cells for all healthy donors (circles; n = 29), acute HIV time points (squares; n = 21), and chronic HIV time points (triangles; n = 23). Frequency of perforin⁺ cells that are T-bet⁺Eomes⁺ (C) or T-bet^-Eomes^- (D) over time (n = 23). All data represent direct ex vivo assessment with no in vitro stimulation. ** denotes a P value < 0.01 and *** denotes a P value < 0.001. Statistics based on a GEE population-averaged model with Holm adjusted P value. Bars represent approximations of the means generated by the models. Lowess smoothers were used to represent the mean over time for longitudinal data.

Fig 6. T-bet and Eomes expression by responding HIV-specific CD8⁺ T cells over time.

(A) Gag-specific CD8⁺ T cells (red) overlaid on perforin⁺ CD8⁺ T cells (black) from the acute (day 28) and chronic (day 434) infection time points of one donor. Percentages of responding Gag-specific cells within each T-bet and Eomes subset are provided. Frequencies of T-bet⁺Eomes⁺ (B) and T-bet^-Eomes^- (C) responding Gag-specific CD8⁺ T cells over time. (D) Frequencies of T-bet^HiEomes⁺ (blue circles) and T-bet^LoEomes⁺ (red circles) responding Gag-specific CD8⁺ T cells at acute (23–100 days; n = 16) and chronic (>365 days; n = 23) time points. Proportion of T-bet^HiEomes⁺ (E) and T-bet^LoEomes⁺ (F) responding Gag-specific CD8⁺ T cell subsets that express perforin over time. n = 20 for all longitudinal graphs. * denotes a P value < 0.05. Statistics based on a GEE population-averaged model with Holm adjusted P value or random-effects tobit regression. Bars represent approximations of the means generated by the models. Lowess smoothers were used to represent the mean over time for longitudinal data.