Speaker
Description
Background: Hypermutated proviruses, which arise in a single replication cycle when host antiviral APOBEC proteins introduce G-to-A mutations throughout the HIV genome, persist during ART. But, their within-host origins and longevity are poorly understood because standard phylogenetic analyses, which assume that mutations arise over many replication cycles, cannot accommodate them. We describe an approach for within-host phylogenetic analysis and molecular dating of hypermutated proviruses.
Methods: We analyzed longitudinal datasets from 6 women living with HIV, comprising >800 single-genome HIV RNA env sequences collected between seroconversion and ART initiation (median 8 years), along with >600 proviral env sequences sampled over a median 9 years on ART. APOBEC-induced mutations were identified using Hypermut. For each participant we inferred within-host maximum-likelihood phylogenies relating each participant's pre-ART plasma sequences and env-intact on-ART proviruses, with or without hypermutated proviruses. For the latter, APOBEC-mutated sites were either included unaltered, completely stripped from the alignment, or replaced with "R" (mixture of A/G). Proviral ages were inferred using an established root-to-tip regression approach.
Results: Phylogenies featured a median 286 (range:120-421) sequences per participant, including a median 19 (range:5-23) hypermutated proviruses. When directly incorporated into phylogenies, hypermutated proviruses displayed significantly longer branchlengths and higher evolutionary distinctiveness than env-intact proviruses (all p<0.0001), and clustered extensively. All these metrics normalized after stripping hypermutated sites from the alignment, or replacing them with R. The resulting trees produced age estimates of env-intact proviruses that were concordant to those from trees that excluded hypermutated sequences, though coefficients were generally slightly higher using the "R replacement" approach (0.90-0.99). Using the latter trees to molecularly date hypermutated proviruses revealed that these integrated throughout untreated infection, and persisted during ART.
Conclusion: Our approach allows hypermutated sequences to be incorporated into phylogenies for hypothesis testing. Hypermutated proviruses, like other provirus types, are archived throughout untreated infection and persist for years on ART.
