Abstract Body

Background

Antiretroviral therapy (ART) suppresses HIV-1 replication and prevents disease progression in people living with HIV (PLWH). However, the latent reservoir allows viral rebound after analytical treatment interruption (ATI), with wide variation in rebound timing. We hypothesized that specific immune cell states influence rebound kinetics. 

Methods

Peripheral blood mononuclear cells were obtained from 27 PLWH from ATI studies in Thailand(RV254 cohort) without additional intervention. Participants were diagnosed during acute HIV infection (AHI) and were on ART for a median of 3 years. Longitudinal single-cell multiomics profiling, including single-cell transcriptomics, chromatin accessibility, and surface protein expression, were performed at AHI (week 0), 60 weeks post-ART (ART), and before ATI (pre-ATI). Statistical analyses used t-tests, ANOVA and multi-variable regression with cross-validation. 

Results

Participants were classified as early or delayed rebounders based on the median time to viral rebound post ATI (2.7 weeks, IQR 2.0 – 3.7; viral load (VL) >20 copies/mL). Twenty-three immune cell types were identified, with no compositional differences between rebound groups. In contrast, transcriptional profiles strongly associated with rebound timing: monocytes ranked highest at AHI, while naïve CD8+ T cells ranked highest at ART and pre-ATI (MAST, Padj < 0.05). Trajectory analysis of CD8+ T cells showed naïve CD8+ T cells were depleted at AHI but expanded after ART (AHI: 2.8 ± 3.2%; ART: 12.2 ± 6.6%; pre-ATI: 9.4 ± 6.2%; Padj < 0.0001), with the emergence of phenotypically distinct clusters. A novel naïve CD8+ T cells subcluster (C3) that was adjacent to stem-like memory CD8+ T cells (C5) in trajectory space, was significantly enriched in delayed rebounders at pre-ATI (Padj < 0.001). Mechanistically, C3 cells exhibited strong transcriptomic similarity with C5 (r = 0.67, Padj < 0.01), were enriched for TNFα-NF-κB and PI3K–mTOR signaling and displayed increased chromatin accessibility at LEF1, TCF7, and AP-1 binding sites. Further integration with baseline clinical features including VL, identified C3 abundance as the strongest predictor of rebound timing.  

Conclusions

ART induced remodeling of naïve CD8+ T cells, generating a C3 subcluster poised for stem-like memory differentiation. People with higher C3 abundance experience delayed rebound, indicating naïve CD8+ T cell states as key determinants of rebound timing and potential targets for HIV cure.