A multi-omics approach to explore the HIV latency through long reads RNA sequencing and whole-genome 5mC and 6mA DNA methylation

The scientific, clinical, and biotechnological fields have experienced remarkable development in the past few years. Clear evidence was the extremely short time used to develop and test SARS-CoV-2 vaccines compared to the median time to discovery and approval of the drugs. However, HIV-1 infection still misses a permanent cure even though forty-one years have passed since its identification. The major obstacle is represented by integrated and transcriptionally silent viral genomes that remain invisible to the immune system due to the absence of a specific marker. An explored way to achieve the eradication of integrated viral DNA was the 'shock and kill' approach. This strategy aims to induce viral transcriptional reactivation using latencyreversing agents (LRA). Thus, infected cells with active proviruses can be killed by cytopathic effects related to HIV or through immune system-based clearance. However, until now no LRA tested has been able to drastically reduce the HIV-1 reservoir. Identifying an effective LRA or a combination of drugs is crucial to unraveling the complex molecular mechanism behind viral latency. In this context, this work aimed to contribute to a better understanding of this complex regulatory network through a multi-omics approach based on the long-read third-generation sequencing of Oxford Nanopore Technologies (ONT). The first part of this project (sub study 1) deals with the evaluation of different in vitro culture conditions to select the optimal experimental setting to study HIV-1 reactivation. The impact of isolating CD4+ T cells before (PREc) or after (POSTc) in vitro culture and the use of autologous plasma (PLSM) or RPMI were evaluated after stimulation with PMA/ionomycin and IL-2. The transcriptome of CD4+ T cells from 3 young adult donors with perinatal HIV in these different experimental settings was characterised using ONT RNA sequencing and a single-subject differential analysis approach. This in-depth analysis showed that isolation timing did not mainly impact gene expression, while RPMI induced stronger T cell activation compared to PLSM. 4 Despite the different magnitude observed, stimulation of T cells cultured in RPMI and plasma showed deregulation of the same pathways and transcription factors. Within the second sub study, the transcriptional dynamics after treatments with two clinically approved LRAs that act with opposite mechanisms, fimepinostat (CUD) and ingenol-3-angelate (PEP), were evaluated within CD4+ T cells of 5 HIV-infected donors and the ACH2 cell line. In cells from a subset of two HIV-infected donors and ACH2 cells, the whole genome DNA methylation levels of two epigenetic markers, N6- methyladenosine (6mA) and 5-methylcytosine (5mC), were explored before and after CUD stimulation. Considering that 6mA has recently been discovered in eukaryotes, including homo sapiens, and that its distribution, functions, and even its effective presence are still debated, in this thesis 6mA was also deeply explored independently by the HIV context. Gene expression and these epigenetic markers were also investigated in the proximity of subject-specific HIV-1 integration sites mapped by the Mathias Lichterfeld team at the Ragon Institute using the MIP-seq assay. Transcriptional analysis confirmed the opposite impact of CUD compared to PEP and PMA/ionomycin and IL-2 (CTR) in T cell activation. Common deregulated genes may play a crucial role in the reactivation of proviruses. Among these, TCF1 was found to be downregulated. This transcription factor belongs to a family of proteins that reduce chromatin accessibility by recruiting histone deacetylases and therefore suppress gene expression. Furthermore, differentially methylated 5mC sites (DMSs) after CUD treatment were found to be enriched in the DNA-binding sequence motif of this protein family. 5mC analysis also revealed that although CUD does not activate the signaling cascade responsible for the main well-known HIV-1 reactivation mechanism, it could contribute by altering 5mC levels within the DNA-binding sites of AP-2 and JUNB instead of modulating its activation. Concerning the whole-genome landscape of 6mA, a number of 100K to 165K 6mA sites (0.006% - 0.009% of 6mA / A) were detected. These sites were found to be concordantly localised among all samples and strongly enriched around the transcription start sites (TSS) of genes, suggesting that 6mA plays specific biological functions. Whole genome 6mA levels were able to cluster ACH2 separately from primary samples, and cells treated with CUD from those left unstimulated. Furthermore, 6mA DMSs were found to be enriched 5 of DNA-binding sites of transcription factors important for HIV-1 latency disruption, including the retinoic acid receptor a (RARa) and thyroid receptors THRa and THRb. Finally, the levels of gene expression, 6mA and 5mC, were integrated and aligned around the subject-specific HIV-1 ISs. This multi-omic analysis revealed a higher level of 5mC methylation in proximity to ISs and enrichment of TSS of genes, 6mA hypermethylated sites, and a higher proportion of hypomethylated CpG sites at -10K bp from IS. These data suggest that silencing mediated by 5mC and transcriptional interference seem to contribute to maintaining the latency of proviruses found in these study participants and that 6mA may influence gene expression.