Long-read sequencing assay allows accurate characterization of the HIV-1 reservoir

Background The advent of near full-length (NFL) HIV-1 proviral genome sequencing greatly expanded our understanding of the quality of the viral reservoir, revealing that only 2-5% of the persistent proviruses in ART-treated individuals can be considered genome-intact. However, current NFL assays are based on labor-intensive and costly principles of repeated PCRs at limiting dilution, restricting their scalability. We developed a long-read sequencing assay to characterize many proviral genomes in parallel from bulk DNA. Methods The sensitivity of the long-read assay was determined on a DNA dilution series of J-Lat in uninfected Jurkat ranging from 80,000 to <8 HIV-1 copies. Next, the assay was performed on 15 chronic ART-suppressed individuals, using a fixed input of 500 ng DNA extracted from peripheral blood CD4 T cells (reservoir sizes ranging from 321 to 6581 total HIV-1 DNA copies/million CD4 T cells). Individual proviruses were tagged with a different unique molecular identifier (UMI) at each end during a single reaction, followed by NFL PCR amplification and long-read sequencing on an Oxford Nanopore MinION. UMI-based demultiplexing allowed for the construction of highly accurate consensus genomes, while excluding aberrant chimeric PCR artefacts. In addition, Full-Length Individual Provirus Sequencing (FLIPS) was performed on 2 individuals. Data from both assays were compared through phylogenetic analyses. Results The lower limit of the long-read assay was found to be