Abstract
The multifaceted and long-lived nature of HIV-1 reservoirs has proven to be a formidable obstacle in the development of a therapeutic cure of HIV-1. One of the major dimensions of the HIV-1 reservoir is its prevalence and persistence in tissue environments, including in lymph nodes (LN). Within LN, T follicular helper (Tfh) cells are widely considered as the dominant sub-reservoir in viremic people with HIV (PWH). However, whether Tfh cells survive to establish a reservoir in PWH undergoing suppressive antiretroviral therapy (ART) has remained controversial. To address these issues, we deeply phenotyped over 500,000 cells and identified over 2,000 HIV-1 infected cells by employing single-cell multiomics on LN from PWH during viremia and suppressed on ART. While we detected HIV-1 infected Tfh cells, the majority of infected cells during viremia and ART had non-follicular phenotypes and were instead heterogeneously distributed between various CD4+ T-cell subsets. Within-subset comparisons of HIV+ and HIV- cells revealed heightened activation signatures and altered cell cycle states during viremia, but largely similar features during ART. Furthermore, the comparison between viremia and ART in PWH highlighted a noncanonical Tfh subset - defined by high locus accessibility and transcription of PRDM1 - that is selectively depleted during viremia, recovers after ART, and is highly susceptible to infection in vitro. Our work suggests a revised direction for the HIV-1 cure field where a mandate of any comprehensive strategy will be to address the high proportional burden of non-follicular cells within the HIV reservoir.