Journal of General Virology (2014), 95, 1969–1982 DOI 10.1099/vir.0.065227-0 Dating the origin of the genus Flavivirus in the light of Beringian biogeography John H.-O. Pettersson and Omar Fiz-Palacios Correspondence Department of Systematic Biology, Evolutionary Biology Centre, Uppsala University, John H.-O. Pettersson Uppsala, Sweden [email protected] or [email protected] The genus Flavivirus includes some of the most important human viral pathogens, and its members are found in all parts of the populated world. The temporal origin of diversification of the genus has long been debated due to the inherent problems with dating deep RNA virus evolution. A generally accepted hypothesis suggests that Flavivirus emerged within the last 10 000 years. However, it has been argued that the tick-borne Powassan flavivirus was introduced into North America some time between the opening and closing of the Beringian land bridge that connected Asia and North America 15 000–11 000 years ago, indicating an even older origin for Flavivirus. To determine the temporal origin of Flavivirus, we performed Bayesian relaxed molecular clock dating on a dataset with high coverage of the presently available Flavivirus diversity by combining tip date calibrations and internal node calibration, based on the Powassan virus and Beringian land bridge biogeographical event. Our analysis suggested that Flavivirus originated ~85 000 (64 000–110 000) or 120 000 (87 000–159 000) years ago, depending on the circumscription of the genus. This is significantly older than estimated previously. In light of our results, we propose that it is likely that modern humans came in contact with several members of the genus Flavivirus much earlier than suggested previously, and that it is possible that the spread of several Received 28 February 2014 flaviviruses coincided with, and was facilitated by, the migration and population expansion of Accepted 4 June 2014 modern humans out of Africa. INTRODUCTION not exclusively, vectored by either Aedes or Culex mosqui- toes. The second group, tick-borne flaviviruses (TBFs), the The genus Flavivirus, within the family Flaviviridae, currently only monophyletic group of Flavivirus, are vectored by consists of .70 virus species distributed all over the globe ixodid ticks, and are further subdivided into a mammal and (Gould et al., 2003; Gubler et al., 2007; Lindenbach et al., a seabird group based on their host specificity (Grard et al., 2007). It includes numerous viruses of major human health 2007; Gritsun et al., 2003). The no-known vector flaviviruses concern, such as Dengue virus (DENV), Japanese enceph- (NKVFs) are associated with either bats or rodents and alitis virus, West Nile virus and the type species for infect vertebrates without an apparent arthropod vector flaviviruses, yellow fever virus (flavus:‘yellow’),givingthe transmitting the viruses (Porterfield, 1980). The majority of family and the genus its name (Gould & Solomon, 2008; Mackenzie et al., 2004). The genome of flaviviruses consists the known flaviviruses are zoonotic, i.e. pathogenic viruses of a positive-sense ssRNA molecule of ~11 kbp. One single that can be transmitted between humans and other animals. ORF encodes three structural proteins (capsid, pre-mem- However, the fourth group, the insect-specific flaviviruses brane and envelope) and seven non-structural proteins (NS1, (ISFs; flaviviruses that are only capable of replicating in NS2A, NS2B, NS3, NS4A, NS4B and NS5) flanked by insect cells) (Kuno, 2007), is most likely an undersampled untranslated regions (Chambers et al., 1990; Lindenbach & and very diverse group (Cook et al., 2012). Rice, 2003). The idea of a molecular clock has been used to address The genus Flavivirus has been divided into four main many hypotheses in the study of emerging viral diseases, groups based on ecological characteristics, molecular phylo- especially for diseases caused by RNA viruses (Bromham & genetic analyses, vector specificity and virus performance in Penny, 2003), e.g. to reject the hypothesis of a potential host cells (Gaunt et al., 2001; Gould et al., 2001; Kuno, spread of human immunodeficiency virus (HIV) in the 2007). Most of the recognized flaviviruses belong to the 1950s through a contaminated polio vaccine (Korber et al., mosquito-borne flaviviruses (MBFs) that are commonly, but 2000). However, dating the origin of viruses is a complex and challenging task. For viruses with high rates of evolu- Three supplementary figures and three supplementary tables are tion, the original phylogenetic signal is difficult to deduce available with the online version of this paper. even with complex evolutionary models because the signal 065227 G 2014 The Authors Printed in Great Britain 1969 J. H.-O Pettersson and O. Fiz-Palacios diminishes with time due to repeated substitutions at the and Russian POWV lineages in North America is consistent same site (Holmes, 2003a). The extremely high substitution with the idea that the colonization of POWV into North rate observed for RNA viruses is mostly due to their error- America happened under a single event by a land route, also prone replication and repair machinery (Bromham & supported by the lack of evidence for continuous movement Penny, 2003; Duffy et al., 2008). Unequal substitution rates of POWV and TBEV between Russia and North America by among lineages of the same genus further adds to the either seabirds or mosquitoes (Heinze et al., 2012). complexity of accurately estimating the time to the most Therefore, it is unlikely that POWV emerged in North recent common ancestor (tMRCA) (Duffy et al., 2008; America before the land bridge became accessible again after Holmes, 2003a; Sanjua´n, 2012). the last glacial maximum. Consequentially, it is improbable that POWV emerged after the Bering Strait was formed. The Reconstruction of divergence times requires a temporal most probable explanation, given the current knowledge and reference to convert branch lengths of a phylogenetic tree present distribution of the POWV lineages, is that during the into time. This temporal reference is usually in the form of course of TBF evolution POWV diverged from other TBF a fossil or biogeographical event (i.e. internal node calibra- lineages in a single introduction event by the land route tion), as in most eukaryote studies, or isolation dates (i.e. during the presence of the Beringian land bridge 15 000– tip date calibration), as with bacteria and virus studies. 11 000 years ago (Heinze et al., 2012). From the biogeographical events, fossils or isolation dates, the ages of internal nodes can be estimated. The match POWV could have been introduced into North America between virus and host phylogenies has led to the sugges- by either humans or other mammals that colonized the tion that some virus lineages originated millions of years Americas during the existence of the Beringian land ago. However, this has been strongly rejected by molecular bridge via one of two routes: the interior route or coastal clock studies indicating a virus origin of only a few thousand route. (i) The interior route became accessible when the years ago (Holmes, 2003a; Worobey et al., 2010). This process of deglaciation created a corridor between the controversy has sparked a debate regarding the validity and Laurentide and Cordilleran glaciers. The corridor did not utility of molecular clock reconstructions using tip dates exist during the last glacial maximum. The corridor versus internal (deep) calibration to study and infer the started to open ~15 000 years ago (Dixon, 2013; Dyke, temporal scale of virus evolution (Sharp & Simmonds, 2004), and became habitable for humans and other 2011). mammals ~13 500 years ago (Dixon, 2013). (ii) Towards the end of the last glacial maximum, the glaciers border- Previous studies based on molecular clocks have suggested ing to the south-west coast of Alaska through western that the Flavivirus clade originated ~10 000 years ago in Canada started to melt. Around 16 000 years ago, the Africa (Gould et al., 2001) from a non-vectored mam- glaciers had receded to the extent that the coastal habitats malian virus ancestor (Gould et al., 2003), followed by the could support human populations (Dixon, 2013 and radiation of TBFs and MBFs during the last 5000 and 3000 references therein). The coastal route is also likely to be years, respectively (Zanotto et al., 1996), and ISFs ~3000 the route by which humans first entered the Americas years ago (Crochu et al., 2004). Other studies have also (Achilli et al., 2013; Bodner et al., 2012; Fagundes et al., dated small clades of individual flaviviruses using tip dates 2008; Goebel et al., 2008; Schurr, 2004). as calibration points. These estimates are broadly congruent with the notion of a Flavivirus origin within the last 10 000 Presently, there are coding nucleotide genomes available from years (e.g. Dunham & Holmes 2007; May et al., 2011; Pan .70 unique strains of the genus Flavivirus, including viruses et al., 2011; Twiddy et al., 2003). However, recent analyses recognized by the International Committee on Taxonomy of on TBFs, based on complete genomes analysed with a Viruses (http://www.ictvonline.org/virusTaxonomy.asp?version= relaxed molecular clock and Bayesian methods, have esti- 2013). Here, we combine this information together with mated the age for the TBF clade to be at least 16 000 years relaxed molecular clock methods implementing a Bayesian (Heinze et al., 2012), indicating that the age for the approach, again to estimate and shed some light on diver- Flavivirus genus as a whole should be significantly older gence times of the genus Flavivirus and groups within. Aided than suggested previously. by biogeographical calibration, we report the first study, to the best of our knowledge, estimating the age and rates of Within the TBFs, a close relationship between the bio- substitution of the genus Flavivirus as a whole, including its geography of the Beringian land bridge, the historical major groups, using complete coding nucleotide genomes.