The rebound-initiating viral reservoir (RIVR) in SIV/SHIV infection is dominated by plasma virus sequences present at the time of ART initiation.

UNLABELLED: Active human immunodeficiency virus (HIV) replication generates long-lived latently infected cells that persist despite effective antiretroviral therapy (ART). Several studies analyzing proviral DNA sequences sampled during ART have shown that these sequences are closely related to plasma virus present around treatment initiation. However, a large proportion of proviral DNA is defective or difficult to reactivate, suggesting that only a subset of latent proviruses contributes to viral rebound in vivo. Here, we used a barcoded simian immunodeficiency virus (SIV) or simian-human immunodeficiency virus (SHIV) model to study virus rebounding after antiretroviral treatment interruption (ATI) and its relationship to plasma virus present before ART initiation. We modeled barcode matches between rebound virus and pre-ART plasma virus to estimate the contribution of different time points to subsequent rebound. We estimate that plasma virus sampled immediately before ART contributes approximately 65%-95% of successful rebound lineages. This remains true after re-treatment and re-interruption of ART, with the circulating virus present just before the second ART contributing most to the second rebound. In contrast, analysis of proviral DNA during ART suggests that plasma virus present at the peak of primary infection contributes more substantially to the DNA reservoir (61%-90%), whereas the virus sampled immediately before ART contributes less. These findings suggest that the observable proviral DNA reservoir is established primarily during peak infection and is supplemented throughout untreated infection. However, post-treatment rebound virus originates from only a small subset of the total proviral DNA reservoir, predominantly formed around the time of ART initiation.

IMPORTANCE

Human immunodeficiency virus infection cannot be eliminated because of long-lived, latently infected cells. Understanding when the latent reservoir forms may help identify the timing of interventions to reduce viral rebound. We analyzed data from macaques infected with a barcoded virus that allowed us to track individual viral lineages. Although many lineages were present in plasma during active infection, only a small number rebounded after antiretroviral treatment (ART) interruption. We compared barcode patterns in rebound virus with plasma virus sampled at different times before treatment to identify when rebounding lineages were observed. We found that the rebound virus was most closely related to the plasma virus present at treatment initiation, suggesting that this may be an important time for the formation of the rebound-initiating reservoir. Rebound virus matched pre-treatment plasma virus more closely than proviral DNA detected during ART, suggesting that on-treatment proviral DNA may poorly predict which viruses will re-emerge after treatment interruption.