Mapping the evolution of bornaviruses across geological timescales

Abstract

It has always seemed likely that viruses originated early in the history of life. However, until the identification of endogenous viral elements (EVEs), there was little if any direct evidence for most virus groups ever having existed in the distant past (1). EVEs are virusderived DNA sequences found in the germline genomes of metazoan species. Uniquely, they preserve information about the genomes of viruses that circulated tens to hundreds of millions of years ago. Comparative analysis of EVE sequences has now provided robust age calibrations for a diverse range of virus groups, completely transforming perspectives on the longer-term evolutionary interactions between viruses and hosts (2). As progress in mapping the complete genome sequences of species has accelerated, the abundance of “fossilized” viral sequences in eukaryotic genomes has become apparent. Increasingly, the challenge is not finding EVEs, but scaling analytical approaches to tackle the exponentially increasing volume of EVE sequence data (3, 4). In PNAS, Kawasaki et al. (5) utilize sophisticated computational approaches to implement a broad-scale analysis of the viral “fossil record,” focusing on a group of viruses called “bornaviruses.” Bornaviruses (family Bornaviridae) are a poorly understood group of single-stranded negative sense RNA viruses (order Mononegavirales) that infect vertebrates. Until 2015 the family contained only one genus (Orthobornavirus); however, recent progress in sampling viral diversity has led to the establishment of two novel genera (Carbovirus and Curtervirus) (6). The prototype member, Borna disease virus (BDV), infects a variety of mammalian species and is the causative agent of a neurological condition in horses referred to as Borna disease or “sad horse disease.” The name “Borna” derives from an 1895 outbreak that occurred in the vicinity of the town of Borna in Saxony, Germany, and decimated the Prussian cavalry (7). The association of BDV with neurological disorders in mammals, combined with reports [now well established (8)] of zoonotic infections in humans, has driven a decades-long and frequently controversial effort to assess the role of bornavirus infection in human mental health conditions such as schizophrenia and depression (9). A zoonotic transfer event involving a novel bornavirus was recently recorded in Germany following the deaths of three men from progressive encephalitis. Strikingly, all three were also breeders of exotic squirrels, and this epidemiological connection soon led investigators to identify the pathogen responsible— variegated squirrel bornavirus 1 (VSBV-1) (10). Retrospective studies have now established that VSBV-1 is an emerging pathogen that was introduced into the German captive squirrel population via exotic pet trade, probably around 2003, and has caused the deaths of at least four people (11). Information archived in EVE sequences can reveal the long-term evolutionary history of viruses, leading to a more complete understanding of their biology. The distribution and diversity of bornavirus EVEs have been investigated previously, but no previous study approaches the scale of that performed by Kawasaki et al. They searched the genome sequences of 969 eukaryotic species for bornaviral EVEs and identified hundreds of unique loci. Most of these bornavirus “fossils” are highly degraded and fragmentary, making it difficult to do meaningful comparative analyses. However, by artfully combining high-throughput computational approaches with careful human oversight, Kawasaki et al. tease out the evolutionary connections between contemporary bornaviruses and the extinct “paleoviruses” represented by EVE sequences, thereby reconstructing a detailed picture of bornavirus evolution (Fig. 1). Numerous novel subgroups were discovered, some of which may represent novel genera. In addition, EVEs show that the host range of extant bornavirus genera is much broader than expected, encompassing multiple vertebrate classes. Of all the kinds of inferences that can be extracted from EVE sequence data, age calibrations based on the detection of orthologous EVEs in related species are among the most reliable, since they are derived in part from an orthogonal data source (the fossil record of host species) (2). However, identifying EVE insertions