Footprints of innate immune activity during HIV-1 reservoir cell evolution in early-treated infection

Abstract

Antiretroviral treatment (ART) initiation during the early stages of HIV-1 infection is associated with a higher probability of maintaining drug-free viral control during subsequent treatment interruptions, for reasons that remain unclear. Using samples from a randomized-controlled human clinical trial evaluating therapeutic HIV-1 vaccines, we here show that early ART commencement is frequently associated with accelerated and efficient selection of genome-intact HIV-1 proviruses in repressive chromatin locations during the first year after treatment initiation. This selection process was unaffected by vaccine-induced HIV-1-specific T cell responses. Single-cell proteogenomic profiling demonstrated that cells harboring intact HIV-1 displayed a discrete phenotypic signature of immune selection by innate immune responses, characterized by a slight but significant upregulation of HLA-C, HLA-G, the IL-10 receptor, and other markers involved in innate immune regulation. Together, these results suggest an accelerated immune selection of viral reservoir cells during early-treated HIV-1 infection that seems at least partially driven by innate immune responses.

Figure 1.

Proviral reservoir profile in RIVER study. (A–C) Frequencies of total (A), intact (B), and defective (C) HIV-1 proviruses in PBMCs from the control and treatment groups of the RIVER study at indicated time points. Data from people treated with antiviral therapy for a median of 9 years (ART) and from ECs are shown for comparison. Open circles indicate data at the limit of detection. Horizontal bars indicate the median. FDR-adjusted two-sided Kruskal–Wallis nonparametric tests were used for statistical comparisons relative to EC and ART cohorts. Only significant P values (P < 0.05) are presented. C: RIVER control group; T: RIVER treatment group. (D) Pie charts reflect the composition of HIV-1 DNA sequences from RIVER participants at indicated time points. Data from ART and EC cohorts are shown for reference. FDR-adjusted chi-square tests were used for statistical analysis. Numbers of participants are listed for each study group; numbers of proviral sequences are depicted below pie charts. (E) Proportions of intact HIV-1 DNA copies in total proviral genomes from the control and treatment groups of the RIVER study at randomization, 18 wk, and 1 year time points. Data from ART and EC cohorts are shown for comparison. Horizontal bars indicate the median and n represents the number of study participants. C: RIVER control group; T: RIVER treatment group. (F) The average genetic distance of intact proviruses in each participant from RIVER, ART, and EC cohorts was determined by pair-wise comparisons between all unique intact proviruses within a given study participant. FDR-adjusted two-sided Kruskal–Wallis nonparametric test was used for statistical analysis. Horizontal bars indicate the median, and n represents the number of study participants in whom at least two distinct intact proviruses were detected. (G) Number of HLA class I associated mutations within intact proviral sequences from RIVER, ART, and EC cohorts, determined by an algorithm described by Carlson et al. (2012). Each dot represents one intact provirus. Only clade B sequences were included, and clonal sequences were counted once. FDR-adjusted two-sided Kruskal–Wallis nonparametric test was used. Horizontal bars indicate the median, and n represents the number of intact sequences from each cohort. (H) Proportions of clonal intact HIV-1 proviruses (defined as proviral sequences detected at least two times) within total intact HIV-1 proviruses from RIVER participants at indicated time points, and in ART and EC reference cohorts. Horizontal bars indicate the mean and n represents the number of study participants. All individuals with at least two detectable intact proviral sequences were included. (I and J) Circular maximum likelihood phylogenetic trees of intact HIV-1 proviruses from RIVER control group (I) and RIVER treatment group (J). Participant (P) and clade information were indicated. Each symbol represents one intact provirus. Color coding reflects different participants. n represents the number of intact proviral sequences from each group. Symbols indicate different time points: dot: Randomization time point; square: 18 wk; triangle: 1-year time point; HXB2, HIV-1 reference sequence. (A–H) P values are listed when considered significant (P < 0.05) after adjustment for multiple comparison testing.

Figure S1.

Detailed analysis of HIV-1 DNA sequences from RIVER participants. (A–C) Frequencies of total (A), intact (B), and defective (C) HIV-1 proviruses in PBMCs from all RIVER study participants at indicated time points. Data from people treated with antiviral therapy for a median of 9 years (ART) and from ECs are shown for comparison. Open circles indicate data at the limit of detection. Horizontal bars indicate the median. FDR-adjusted two-sided Kruskal–Wallis nonparametric tests were used for statistical comparisons relative to EC. Wilcoxon matched-pairs signed rank test was used to compare data from RIVER participants at different time points. n represents the number of study participants. (D) Proportions of intact HIV-1 DNA copies in total proviral genomes from all RIVER study participants at randomization, 18 wk, and 1 year time points. Data from ART and EC cohorts are shown for comparison. Horizontal bars indicate the median and n represents the number of study participants. (E) Number of HLA class I–associated mutations within intact proviral sequences from RIVER control and treatment groups, ART and EC cohorts, determined by an algorithm described by Carlson et al. (2012). Each dot represents averaged data from one study participant. Only clade B sequences were included, and clonal sequences were counted once. FDR-adjusted two-sided Kruskal–Wallis nonparametric test was used. Horizontal bars indicate the median and n represents the number of participants from each cohort. (F) Virograms highlighting the proviral reservoir profile in RIVER study participants. Each horizontal line reflects one HIV-1 DNA sequence. Color-coding reflects classification of proviral sequences as intact or defective. Sample time points are indicated on the y-axis.

Figure 2.

Integration site landscape of intact proviruses in RIVER study participants. (A–F) Circos plots reflecting the clonality and chromosomal locations of all intact proviral sequences isolated at indicated time points in each participant (A–C: RIVER control group; D–F: RIVER treatment group). Genomic coordinates are generally indicated in the Hg38 human reference genome nomenclature; in selected cases, coordinates in the T2T reference are indicated if integration sites could not be mapped to the Hg38 genome. Each symbol reflects one intact provirus. Color coding of the symbol and the bracket around the plot reflect indicated time points. Clonal sequences, defined by complete sequence identity and/or identical integration sites, are highlighted. P: participant. Integration sites annotated by the T2T reference genome are indicated. n represents the number of intact proviral sequences from each participant.

Figure 3.

Genomic and epigenetic integration site features of intact proviruses. (A) Pie charts reflect the proportions of intact (upper panel) and defective (lower panel) proviruses within defined genomic regions at indicated time points of the RIVER study, and in ART and EC reference cohorts. Clonal sequences were counted individually, and n represents the number of integration sites. FDR-adjusted chi-square tests were used for statistical analysis. (B) Pie charts reflect the proportions of intact (upper panels) and defective proviruses (lower panels) within defined genomic regions at indicated time points of the control group and treatment group of the RIVER study. Clonal sequences were counted individually, and n represents the number of integration sites. FDR-adjusted chi-square tests were used to compare each pie chart. Comparisons of proviruses in genomic regions with “block and lock” heterochromatin features (group I) and corresponding proportions in genic locations (group II) between individual pie charts were analyzed using FDR-adjusted Fisher’s exact tests. (C) Proportions of intact proviruses located in structural compartments/subcompartments A (transcriptionally-active chromatin) and B (transcriptionally repressed chromatin) are shown, as determined by alignment of integration site coordinates to Hi-C seq data (Rao et al., 2014). Integration sites not covered in the reference dataset were excluded. Clonal sequences with the same integration sites were counted once. FDR-adjusted two-sided Kruskal–Wallis nonparametric test was used. n represents the number of unique integration sites from each cohort.

Figure S2.

Comparative analysis of HIV-1 DNA sequences isolated from RIVER study participants. (A and B) Circos plot reflecting the clonality and chromosomal locations of all defective proviral sequences isolated at indicated time points in participant 3 from RIVER control group (A), and participant 8 from RIVER treatment group. Each symbol reflects one provirus. Color coding of symbol and the bracket around the plot reflect indicated time points. Clonal sequences, defined by complete sequence identity and/or identical integration sites, are highlighted. n represents the number of defective HIV-1 proviral sequences. P: participant. Color-coded arches around the plots indicate types of proviral sequences. (C–E) Chromosomal distance between integration sites of intact proviruses to most proximal host TSSs listed in the genome browser. Clonal sequences with the same integration sites were counted once. FDR-adjusted two-sided Kruskal–Wallis nonparametric test was used. (C and D) Horizontal bars indicate the median, and n represents the number of unique integration sites from each cohort (C) and from each randomization arm (D). (E) Average distance of integration sites of intact proviruses to most proximal TSS in each study participant. Horizontal bars indicate the mean, and n represents the number of participants from each cohort.

Figure 4.

Integration sites of defective HIV-1 proviruses in the RIVER study. (A–D) Circos plots reflecting the clonality and chromosomal locations of all defective proviral sequences isolated at indicated time points in participants 1 (A), 2 (B), 6 (C), and 7 (D). A and B: RIVER control group; C and D: RIVER treatment group. Each symbol represents one defective provirus. Color coding of the symbol and the bracket around the plot reflect indicated time points. Clonal sequences, defined by complete sequence identity and/or identical integration sites, are highlighted. n represents the number of defective HIV-1 proviral sequences from each participant. P: participant. Color-coded arches around the plots indicate types of proviral sequences.

Figure 5.

Phenotypic characteristics of HIV-1 reservoir cells in the RIVER study. (A) 2D UMAP diagrams reflecting the global phenotypic profile of HIV-1 reservoir cells in memory CD4 T cells isolated from peripheral blood of three RIVER study participants. Six computationally defined spherical clusters are indicated. One plot is shown each for category 1, category 2, and category 3 cells. (B) Heatmap representing the normalized phenotypic profile of cells in each spherical cluster, based on 72 surface markers included in this study. (C) Volcano plots reflecting the enrichment ratio of marker-positive cells and corresponding FDR-adjusted (adj.) P values for all 72 surface markers were included in this study. Selected markers were labeled individually. Marker sensitivities, defined as the proportions of marker-positives cells in the indicated categories of cells, were indicated by dot sizes. Comparisons between indicated categories of cells are depicted; bootstrapped data from all three participants are shown. Significance was tested using a two-sided chi-squared test; FDR-adjusted P values are shown. (D) Density plots reflecting the expression of selected surface markers on indicated categories of HIV-1-infected cells from the three RIVER study participants.

Figure S3.

Comprehensive analysis of phenotypic characteristics of HIV-1 reservoir cells. (A) Density plots reflecting the expression of remaining surface markers on indicated categories of cells from three participants. Category 0 cells are shown as a reference. (B) Histograms reflecting expression of HLA-C on carriers of indicated HLA-C alleles. (C) Correlation between enrichment ratios of phenotypic markers on HIV-1 reservoir cells in the RIVER study and in a prior study by Sun et al. (2023). Correlations were analyzed by Spearman correlation coefficients.