Abstract Body

Background

Regulation of the HIV reservoirs across human tissues is heterogeneous, and the extent to which the HIV transcriptional cascade – from HIV DNA to early transcripts (e.g., unspliced [us] HIV RNAGag) and late transcripts (e.g., multiply spliced [ms] HIV RNATat/Rev) – differs across tissues remains unclear. We examined the influence of tissue environments on HIV transcription under suppressive ART.

Methods

Tissue samples of 24 distinct tissue types (Fig.1) were collected during rapid autopsy from Last Gift participants on long-term suppressive ART. Bulk tissue-associated HIV DNA, us HIV RNAGag and ms HIV RNATat/Rev were quantified by droplet digital PCR. Full-length env single-genome sequencing assessed viral diversity and identified intact and hypermutated clones. Principal Component Analysis of HIV DNA/RNAs (residuals) followed by hierarchical clustering defined tissue clusters (Fig.1). Multivariable mixed-effects models with interactions and Structural Equation Modeling assessed differences across clusters, adjusted for number of clones per sample, and ART regimen and duration.

Results

Twenty-four participants (83.3% male, 65±13 years, mean CD4+ 263±193/µL, ART duration 14±9 years) contributed to 438 samples (n=283 sequenced). Five clusters emerged: C1 (large bowel + abdominal viscera), C2 (small bowel), C3 (spleen, blood), C4 (sanctuaries: brain, testes), C5 (spinal cord sections). Across clusters, the transcriptional cascade differed in 1) Roadblocks: C3-C5 showed a continuous progression from DNA→us→ms (all path coefficients p<0.05), whereas us→ms path was weak in C1 (p=0.15), and DNA→us was weak in C2 (p=0.12). 2) Efficiency: e.g., at comparable HIV DNA levels, C3 showed the steepest increase in us HIV RNAGag (aβ +0.62, p<0.001 vs others), while late transcription was the most efficient in C2 (aβ +0.51, p<0.001). 3) Role of intact clones: higher abundance of intact env clones increased early transcriptional efficiency in C1 (aβ +0.44, p=0.028), while reduced it in C5 (aβ –0.46, p=0.051). Viral diversity was significantly higher in C1-C3 but did not correlate with transcription.

Conclusions

HIV transcriptional cascade differs markedly across tissues, with distinct bottlenecks, efficiencies, and contributions from intact clones. These findings provide the first tissue-level map of HIV reservoir activity, suggesting that tissue environments may imprint specific regulatory signatures on HIV transcription, with direct implications for targeting latency reversal and HIV persistence.

Table or Figure