Memory CD4 + T-Cells Expressing HLA-DR Contribute to HIV Persistence During Prolonged Antiretroviral Therapy

Abstract

To date, most assays for measuring the human immunodeficiency virus (HIV-1) reservoir do not include memory CD4+ T-cells expressing the activation marker, human leukocyte antigen-antigen D related (HLA-DR). However, little is known concerning the role these cells play in maintaining persistent HIV-1 during effective antiretroviral therapy (ART). To address this issue, we examined, cellular activation/exhaustion markers (Ki67, CCR5, PD-1, Lag-3 and Tim-3) and viral gag-pol DNA sequences within HLA-DR- and HLA-DR+ memory CD4+ T-cell subsets longitudinally from the peripheral blood of six participants over 3 to ≥15 years of effective therapy. HLA-DR expression was readily detected during the study period in all participants. The average expression levels of CCR5, PD-1 and Tim-3 were higher on the HLA-DR+ T-cell subset whereas the average of LAG-3 expression was higher on their HLA-DR- counterpart. The proportion of HIV-infected cells increased within the HLA-DR+ subset by an average of 18% per year of ART whereas the frequency of infected HLA-DR- T-cells slightly decreased over time (5% per year). We observed that 20-33% of HIV-DNA sequences from the early time points were genetically identical to viral sequences from the last time point within the same cell subset during ART. This indicates that a fraction of proviruses persists within HLA-DR+ and HLA-DR- T-cell subsets during prolonged ART. Our HIV-DNA sequence analyses also revealed that cells transitioned between the HLA-DR+ and HLA-DR- phenotypes. The Ki67 expression, a marker for cellular proliferation, and the combined markers of Ki67/PD-1 averaged 19-fold and 22-fold higher on the HLA-DR+ T-cell subset compared to their HLA-DR- counterpart. Moreover, cellular proliferation, as reflected by the proportion of genetically identical HIV-DNA sequences, increased within both T-cell subsets over the study period; however, this increase was greater within the HLA-DR+ T-cells. Our research revealed that cellular transition and proliferation contribute to the persistence of HIV in HLA-DR+ and HLA-DR- T-cell subsets during prolonged therapy. As such, the HIV reservoir expands during effective ART when both the HLA-DR+ and HLA-DR- cell subsets are included, and therapeutic interventions aimed at reducing the HIV-1 reservoir should target HLA-DR+ and HLA-DR- T-cells.

Keywords: CD4+ T-cells; HIV persistence; HLA-DR; cell activation/exhaustion markers; cellular proliferation; prolonged ART; single-proviral sequencing.

FIGURE 1

HLA-DR expression on CD4+ memory T-cells during ART. Proportions of CD4+ memory T-cells that express HLA-DR during ART were quantified. The linear prediction was performed using mixed effects regression. The slope indicates change of the proportion of HLA-DR+ CD4+ memory T-cells per year on ART and p-value (p) is for the slope.

FIGURE 2

Expression of cellular activation/exhaustion markers on HLA-DR+ and HLA-DR– CD4+ memory T-cells during ART. Proportions of HLA-DR+ and HLA-DR– CD4+ memory T-cells that express CCR5 (A), LAG-3 (B), PD-1 (C), and Tim-3 (D). The colors of data points and lines indicate different participants; 2115 (gray), 2275 (purple), 2013 (orange), 2026 (green), 2046 (maroon), and 2518 (blue). A linear prediction was performed on the log transformed values using mixed-effects linear regression (black line). The changes of expression levels per year on therapy are indicated by exp(slope) (or e^slope). An e^(slope) greater than 1 indicates an increasing rate and less than 1 indicates a decreasing rate of expression. The p-values (p) are derived by comparing the slopes to the null hypothesis of slope = 0, or e^(slope) = 1.

FIGURE 3

Persistence of HIV-infected HLA-DR+ and HLA-DR– CD4+ memory T-cells during ART. The proportion of HIV-infected T-cells (%) was estimated using maximum likelihood model derived from single-proviral sequencing. The change in the frequency of HIV-infected T-cells per year on ART [exp(slope)] was derived from mixed-effects negative binominal regression (black line). The 95% confidence intervals (95% CI) and p-value of the exp(slope) are also shown. The colors of the data points and the lines indicate different participants as shown in the legend.

FIGURE 4

Intact and defective HIV-DNA p6-RT sequences within HLA-DR+ and HLA-DR– CD4+ memory T-cell subsets derived from an acute/early participant. A representative phylogenetic tree of a participant from the AHI group (2275). The inner layer shows the phylogenetic tree constructed by all HIV-DNA p6-RT sequences derived from the HLA-DR+ and HLA-DR– T-cells obtained from participant 2275 during therapy (Visits 1–4; 2.9–15.1 years after ART initiation). The outer layer shows defective genetic features for individual HIV-DNA sequences as shown in the legend.

FIGURE 5

EIS, intact and defective HIV-DNA p6-RT sequences within HLA-DR+ and HLA-DR– CD4+ memory T-cell subsets derived from a chronic participant. A representative phylogenetic tree of a participant from the CHI group (2026). The inner layer shows a maximum likelihood phylogenetic tree that includes all HIV-DNA p6-RT sequences derived from the HLA-DR+ and HLA-DR– T-cells obtained from participant 2026 during therapy (CHI group). The middle layer shows individual HIV-DNA sequences which are part of an EIS within the HLA-DR+ and HLA-DR– T-cells obtained at three earlier time points (Visits 1–3; 5.1–16.0 years after ART initiation) and last time point (Visit 4; 17.7 years after ART initiation), The outer layer shows defective genetic features for individual viral sequences as shown in the legend.

FIGURE 6

HIV-DNA p6-RT sequences persist within HLA-DR+ CD4+ memory T-cell subset during ART. Maximum likelihood phylogenetic trees using HIV-DNA sequences derived from monophyletic clades within the HLA-DR+ T-cell subset (HLA-DR+ to HLA-DR+). The trees were constructed for participants 2026 and 2518. The sampling time points (years after ART initiation) for individual HIV-DNA sequences are indicated under the participant IDs. The viral sequences derived from different sampling time points are shown with different colors that correspond to the visit IDs as shown in the legend. The shades indicate phylogenetic clades that contain identical HIV-DNA sequences derived from the three earlier time points and the last time point.

FIGURE 7

Overall comparison of the HIV-DNA p6-RT sequences that persist within the HLA-DR+ and HLA-DR– CD4+ T-cell subsets during ART. The percent of HIV-DNA sequences that persist within the HLA-DR+ and HLA-DR– T-cells are indicated as “HLA-DR+ to HLA-DR+” and “HLA-DR– to HLA-DR–”, respectively. The percent was estimated by fitting the mixed-effects logistic regression to all participants (blue bar). The 95% CIs of the percent values are shown as red capped lines. The odds ratio (OR) and its 95% CI and p-value compare “HLA-DR+ to HLA-DR+” to “HLA-DR– to HLA-DR–”.

FIGURE 8

HIV-DNA p6-RT sequences from HLA-DR+ T-cells that are identical to sequences from HLA-DR– T-cells during ART. Maximum likelihood phylogenetic trees using HIV-DNA sequences derived from monophyletic clades that show sequences from the first time points of HLA-DR+ T-cells being identical to sequences from HLA-DR– T-cells at the last time point (HLA-DR+ to HLA-DR–). The trees were constructed for participants 2026, 2046, and 2518. The sampling time points (years after ART initiation) for individual HIV-DNA sequences are indicated under the participant IDs. The viral sequences derived from different sampling time points are shown with different colors that correspond to the visit IDs as shown in the legend. The shades indicate phylogenetic clades that contain identical HIV-DNA sequences derived from the HLA-DR+ T-cells obtained from the three earlier time points and the HLA-DR– T-cells from the last time point.

FIGURE 9

HIV-DNA p6-RT sequences from HLA-DR– T-cells that are identical to sequences from the HLA-DR+ T-cells during ART. Maximum likelihood phylogenetic trees using HIV-DNA sequences derived from monophyletic clades that show sequences from the first time points of HLA-DR– T-cells being identical to sequences from HLA-DR+ T-cells at the last time point (HLA-DR– to HLA-DR+). The trees were constructed for all CHI participants. The sampling time points (years after ART initiation) for individual HIV-DNA sequences are indicated under the participant IDs. The viral sequences derived from different sampling time points are shown with different colors that correspond to the visit IDs as shown in the legend. The shades indicate phylogenetic clades that contain identical HIV-DNA sequences derived from the HLA-DR– T-cells obtained from the three earlier time points and the HLA-DR+ T-cells from the last time point.

FIGURE 10

Overall comparison of the HIV-DNA p6-RT sequences that switch between HLA-DR+ and HLA-DR– CD4+ T-cell subsets during ART. The percent of viral sequences derived from the HLA-DR+ T-cell subset from the three early time points which were identical to at least one (HLA-DR–)-derived viral sequence from the last time point is indicated as “HLA-DR+ to HLA-DR–”. The percent of viral sequences derived from the HLA-DR– T-cell subset from the three early time points which were identical to at least one (HLA-DR+)-derived viral sequence from the last time point is indicated as “HLA-DR– to HLA-DR+”. The percent was estimated from the mixed-effects logistic regression fitted to all participants (blue bar). The 95% CIs of the percent values are shown as red capped lines. The odds ratio (OR) and its 95% CI and p-value compare “HLA-DR+ to HLA-DR–” versus “HLA-DR– to HLA-DR+”.

FIGURE 11

HIV-DNA p6-RT sequences derived from EIS within HLA-DR+ and HLA-DR– CD4+ memory T-cells during therapy (CHI participants). (A) Percent of HIV-DNA p6-RT sequences derived from EIS (% EIS) for participants 2013 (orange data points), 2026 (green data points) and 2046 (maroon data points). EIS was characterized using HIV-DNA sequences derived from the three earlier time points combined (“Early time points”) and the viral sequences derived from last time point (“Last time point”). The odds ratio (OR) compares “Last time point” to “Early time points.” The 95% CI (in square brackets) and the p-value (p) of the OR are also shown. The open gray bars indicate mean % EIS and the 95% CI of the mean is shown as the red capped lines. (B) Percent of HIV-DNA p6-RT sequences derived from EIS (% EIS) for participant 2518. The OR compares 17.0 to 15.1 years of ART. The 95% CI (in square brackets) and the p-value (p) are also shown. The filled blue bars indicate % EIS and its 95% CI is shown as the red capped lines. (C) Quantity of HIV-DNA sequences derived from EIS per million cells analyzed within the HLA-DR+ and HLA-DR– T-cell subsets. The gray open bars indicate mean of log transformed EIS per million cells and red capped lines indicate 95% CI of this mean. The fold-difference compares EIS per million cells of “Last time point” to “Early time points.” The 95% CI (in square brackets) and p-value (p) of the fold-difference are also shown. Different colors of the data points represent different participants; 2013 (orange), 2026 (green), 2046 (maroon), and 2518 (blue).