Single-cell analyses identify monocyte gene expression profiles that influence HIV-1 reservoir size in acutely treated cohorts

Abstract

Eliminating latent HIV-1 is a major goal of AIDS research but host factors determining the size of these reservoirs are poorly understood. Here, we investigate the role of host gene expression on HIV-1 reservoir size during suppressive antiretroviral therapy (ART). Peripheral blood cells of fourteen males initiating ART during acute infection and demonstrating effective viral suppression but varying magnitudes of total HIV-1 DNA were characterized by single-cell RNA sequencing. Differential expression analysis demonstrates increased CD14+ monocyte activity in participants having undetectable HIV-1 reservoirs, with IL1B expression inversely associating with reservoir size. This is validated in another cohort of 38 males comprised of different ancestry and HIV-1 subtypes, and with intact proviral DNA assay (IPDA®) measurements. Modeling interactions show monocyte IL1B expression associates inversely with reservoir size at higher frequencies of central memory CD4+ T cells, linking monocyte IL1B expression to cell types known to be reservoirs for persistent HIV-1. Functional analyses reveal that IL1B activates NF-κB, thereby promoting productive HIV-1 infection while simultaneously suppressing viral spread, suggesting a natural latency reversing activity to deplete the reservoir in ART-treated individuals. Altogether, scRNA-seq analyses reveal that monocyte IL1B expression could decrease HIV-1 proviral reservoirs in individuals initiating ART during acute infection.

Fig. 1. Characteristics of study participants and experimental design.

a Distribution of total HIV DNA in Fiebig stage III participants in RV254 at week 48 after ART initiation and their categorization into three groups based on reservoir size. The center of the box plot represents the median (50^th percentile), while the upper and lower bounds of the box represent the lower and upper quartiles (25^th and 75^th percentiles, respectively). The whiskers extend from the box to the data points at most 1.5 times the interquartile range (IQR) from the lower and upper quartiles. b Selected participants from Fiebig stage III with extreme reservoir size phenotypes (undetectable = below LOD; and detectable = high) of cell-associated total HIV DNA in the RV254 Thai discovery cohort (n = 14). Significance was determined by the Mann-Whitney U two-sided test. c Total HIV DNA decay between weeks 0 (AHI) and 48 (after ART initiation). Significance was determined by the Wilcoxon signed-rank two-sided test. d Phenotypes of participants comprising the detectable versus undetectable reservoir size categories. Mean values are shown for each group, NS not significant. Significance was determined by the Mann-Whitney U two-sided test. e Single-cell RNA-seq and multiparameter flow cytometry were performed on all 14 participants. Additional validation by scRNA-seq was performed in an independent AHI cohort from the USA (A5354) (n = 38).

Fig. 2. Differentially expressed genes in monocytes associated with HIV reservoir size during ART.

a scRNA-seq identified 24 unique clusters of immune cell subsets. b CD14+ classical monocytes have the highest number of DEG between the detectable and undetectable reservoir groups. Significance was calculated by the Mann-Whitney U two-sided test with Bonferroni correction. Circle color represents cell subset, while circle size indicates the corresponding number of cells. c Volcano plot shows DEG in all cell types. Genes with P values that are significant after correction are indicated above the horizontal dotted line. Labeled genes have a P < 10e-6 and absolute average log_e fold chang_e ≥ 1 (vertical dotted lines) or P < 10e-100 and absolute average log_e fold change ≥0.5. Significance was calculated by the Mann-Whitney U two-sided test. d The most significant DEG in CD14+ monocytes comparing reservoir groups. Black dots represent the median normalized gene expression values (log_e), and lines represent the interquartile ranges. Teal: undetectable reservoir, red: detectable reservoir. Significance was determined by the Mann-Whitney U two-sided test with Bonferroni correction. e Participant-specific categorical analyses of the most significant DEGs. Normalized gene expression within CD14+ monocytes was averaged per participant and correlation was determined by the Spearman test (n = 14). The association direction is indicated via a monotonic trend line. f, g Table and interaction plots of multiple regression between standardized THBS1 or IL1B expression in monocytes and predicted reservoir size with varying frequency of the CD4+ T_CM population. Nominal P values are indicated for the two-sided interaction analyses.

Fig. 3. Validation of IL1B association with smaller reservoir size from an independent cohort with a different infecting viral subtype across various Fiebig stages.

a HIV DNA levels vary within the A5354 subtype B cohort from the USA (n = 38). The participant samples used in this study are highlighted based on reservoir size. Red: detectable; teal: undetectable; and yellow: middle. Black and White indicate differences in the ancestry of the participants. b Characteristics of participants comprising the detectable, middle, and undetectable reservoir size categories (mean values) and the Mann-Whitney U two-sided test P values comparing the extreme phenotype groups are shown. HIV-1 subtype information was only available for a subset of the participants. NS: not significant c Dimensionality reduction plot of the different immune clusters in this cohort. d CD14+ monocytes have the highest number of normalized DEG associated with reservoir size using a continuous analysis including all 38 participants. e Participant-specific average IL1B expression in CD14+ monocytes categorized by total HIV DNA (n = 38). Spearman correlation P value and rho are shown. The directionality is indicated with a monotonic trend line. Box-whisker plots of the distributions of the three groups are shown below the x-axis. Significance was determined by the Mann-Whitney U two-sided test. f IPDA® measurements from a subset of the participants in this cohort (n = 21). Significance was determined by the Mann-Whitney U two-sided test. g IL1B association with different reservoir type measurements (rows) from the participants with IPDA measurements (n = 21). Significance was assessed using the MAST statistical framework (two-sided with adjustment for multiple corrections). For all box plots, the center of the box plot represents the median (50^th percentile), while the upper and lower bounds of the box represent the lower and upper quartiles (25^th and 75^th percentiles, respectively). The whiskers extend from the box to the data points at most 1.5 times the interquartile range (IQR) from the lower and upper quartiles.

Fig. 4. Pathway analyses identified a distinct signature associating with reservoir size.

a Gene co-expression modules in CD14+ monocytes from the RV254 Thai study. b IL1B is in the M3 WGCNA module which was enriched in cells from RV254 participants with undetectable reservoir based on the top 25 hub genes in the module (detectable=6, undetectable=8). Significance was determined by the Mann-Whitney U two-sided test with Bonferroni correction. c Using the same module hub genes found in RV254, the M3 module was also enriched in cells from the undetectable reservoir participants in the A5354 cohort when HIV DNA levels were grouped categorically (detectable=12, undetectable=11). Significance was determined by the Mann-Whitney U two-sided test with Bonferroni correction. d Average expression of the 25 top hub genes from the M3 module was generally higher in participants with undetectable reservoir in both cohorts. e Predicted protein interaction network of top 25 hub genes using the STRING protein database. Larger nodes have higher degree of connectivity; node color indicates significance in the categorical DEG comparison between the detectable and undetectable groups in RV254 CD14+ monocytes (P value determined using the Mann-Whitney U two-sided test with Bonferroni correction). f Gene ontology analyses of genes enriched in module M3 in CD14+ monocytes. Significance was determined using the Enrichr implementation of Fisher’s exact test. P values were FDR corrected; the ten most significant genesets with adjusted P < 0.05 are shown. The color of y-axis labels indicates the originating database of the gene set: Hallmark: purple, KEGG: brown, GO Biological Process: green, WikiPathways: pink.

Fig. 5. In vitro IL1B activated NF-κB, increased HIV proviral transcription, and inhibited spreading infection.

a Effects of IL1B on NF-κB activity were assessed using A549 NF-κB reporter cells. Cultures were treated with IL1B, TNFα, or LPS and infected with VSV-pseudotyped NL4-3, CH058, or Mock control. After 24 hrs the Alkaline Phosphatase Blue Microwell assay was performed with OD650 values relative to the corresponding no treatment control (NT) reflecting NF-κB expression which is shown on the y-axis. The data presents an average of n = 3 independent experiments ±SEM. Blue: IL1B, teal: TNFα, yellow: LPS. b PBMC from n = 3 participants were treated with IL1B or TNFα and examined for IκBα phosphorylation as described in the methods section. Graphs present the protein expression from these n = 3 participants as mean ± SEM. Corresponding representative uncropped images are provided in the Source Data file. White: no treatment, light blue: IL1B 10 ng/ml, medium blue: IL1B 50 ng/ml, dark blue: TNFα. c Effects of IL1B in vitro when HIV was quantified after a single round of infection. Plots show the relative proportions of pMorpheus-V5 latently (blue) or productively (orange) infected PBMC in cultures treated with IL1B prior to, simultaneously, or after transduction with Env viral particles carrying the indicated Env protein. The background obtained from uninfected cells (Mock) was subtracted from the values obtained for cells infected with the pMorpheus HIV-1 construct pseudo-typed with the indicated envelope proteins. The data represent n = 6 (X4) or n = 3 (R5 and R5X4) individual healthy participants, with error bars representing the average ±SEM. Corresponding gating strategies and representative plots are shown in Supplementary Fig. 7. d Effects of IL1B on spreading HIV-1 infection in cell culture. Using HIV-1 YU-2, bar plots display the relative p24-positive cell fractions after pre-treatment with increasing concentrations of IL1B (from 0.01-10.0 ng/ml, 10-fold increments) across n = 4 different participants. Significance was calculated using the Mann-Whitney U two-sided test. Orange: no treatment, yellow: different concentrations of IL1B treatment. e, f Bar plots display the average infectious virus yields (e) and p24 antigen levels (f) of n = 5 different participants at 4 days post-infection relative to the no IL1B treatment controls normalized to 100% ±SEM. Corresponding replication curves are shown in Supplementary Fig. 9. For all bar graphs shown, bar height and error bars represent mean values +/- SEM. Significant differences in all panels except (d), were determined using two-sided unpaired t-test analyses. Asterisks indicate statistical significance (*P < 0.05, **P < 0.01, *** P < 0.001, ****P < 0.0001). Source data and exact P values are provided in the Source Data file.