Molecular signatures of HIV latency established in different conditions revealed by single-cell RNA sequencing

Because HIV provirus in many cells during latency is not entirely silent, it became possible to use single-cell RNA sequencing (scRNA-seq) to determine gene expression patterns in a subset of latently infected cells. In this study, we aimed to identify molecular signatures, or biomarkers, of latency established in different conditions. We have used two in vitro models of HIV latency. The first model involved infection, activation of CD4+ T cells using anti-CD3/anti-CD28 antibodies and allowing the cells to return to quiescence (the 14-day model). The second model involved direct infection of the resting CD4+ T cells via co-culture with autologous productively infected cells to allow cell-to-cell virus transmission (the 10-day model). Furthermore, we utilized an isogenic pair of CXCR4- and CCR5-tropic viruses for direct infection of resting cells. Finally, CD4+ T cells from people with HIV were also profiled. We show common and unique molecular signatures of latency established in different conditions; a subset of genes identified in vitro was validated using cells from people with HIV. For the in vitro studies, primary CD4+ T cells were isolated from peripheral blood of HIV-seronegative blood donors. For the ex vivo studies, biobanked CD4+ T cell samples from people with HIV were used. Two separate gene expression profiling experiments were conducted using the in vitro models of HIV latency and scRNA-seq. In experiment #1, molecular signatures of latency were compared between the two conditions in which latency was established: following the exposure of cells to T cell receptor stimulus (the 14-day model), or directly in resting cells (the 10-day model). Cells with differential proliferative responses to the T cell receptor stimulus in the 14-day model were tracked using a viable dye carboxyfluorescein succinimidylester (CFSE) to enable comparison of signatures of HIV latency dependent on cell proliferative responses. In experiment #2, gene expression in cells latently infected with viruses of different tropism, CXCR4 and CCR5, were compared. In experiment #3, biobanked samples of CD4+ T cells from people with HIV were used to validate selected biomarkers identified in vitro. These samples were also used to more accurately define a cell as latently infected in vitro, based on HIV read distribution throughout the genome in samples from people with HIV.