Evolution of dependoparvoviruses across geological timescales-implications for design of AAV-based gene therapy vectors.

ABSTRACT

Endogenous viral elements (EVEs) are genetic remnants of viruses that have integrated into host genomes millions of years ago and retained as heritable elements passed on to offspring until present-day. As a result, EVEs provide an opportunity to analyse the genomes of extinct viruses utilizing these genomic viral fossils to study evolution of viruses over large timescales. Analysis of sequences from near full-length EVEs of dependoparvoviral origin identified within three mammalian taxa, Whippomorpha (whales and hippos), Vespertilionidae (smooth-nosed bats), and Lagomorpha (rabbits, hares, and pikas), indicates that distinct ancestral dependoparvovirus species integrated into these host genomes approximately 77 to 23 million years ago. These ancestral viruses are unique relative to modern adeno-associated viruses (AAVs), and distinct from extant species of genus Dependoparvovirus. These EVE sequences show characteristics previously unseen in modern, mammalian AAVs, but instead appear more similar to the more primitive, autonomously replicating and pathogenic waterfowl dependoparvoviruses. Phylogeny reconstruction suggests that the whippomorph EVE orthologue derives from exogenous ancestors of autonomous and highly pathogenic dependoparvovirus lineages, believed to have uniquely co-evolved with waterfowl birds to present date. In contrast, ancestors of the two other mammalian orthologues (Lagomorpha and Vespertilionidae) likely shared the same lineage as all other known mammalian exogenous AAVs. Comparative in silico analysis of the EVE genomes revealed remarkable overall conservation of AAV rep and cap genes, despite millions of years of integration within the host germline. Modelling these proteins identified unexpected variety, even between orthologues, in previously defined capsid viral protein (VP) variable regions, especially in those related to the three- and fivefold symmetry axes of the capsid. Moreover, the normally well-conserved phospholipase A2 domain of the predicted minor VP1 also exhibited a high degree of sequence variance. These findings may indicate unique biological properties for these virus 'fossils' relative to extant dependoparvoviruses and suggest key regions to explore within capsid sequences that may confer novel properties for engineered gene therapy vectors based on paleovirology data.