Evolutionary dynamics of lineage 2 West Nile virus in Europe, 2004-2018: Phylogeny, selection pressure and phylogeography.

West Nile virus (WNV) is an arbovirus causing neuroinvasive disease to humans and equines. Since 2004, lineage 2 WNV strains have been identified in Europe and have been implicated in severe outbreaks, with that of 2018 exceeding the total number from the previous seven years. The aim of this study was to explore the evolutionary process that shapes the genetic diversity of lineage 2 WNV strains (belonging to the Central European/Hungarian subclade) and reconstruct the origin and transmission routes in Europe, and especially in the Balkans. For this purpose, a high number of whole genome sequences (WGSs) were analyzed, along with newly characterized sequences, including strains from the 2018 WNV transmission season in Greece. Maximum likelihood and Bayesian inference methods were used to perform the phylogenetic and phylodynamic analyses and phylogeographic reconstruction. The majority of the Central European/Hungarian lineage 2 strains are grouped in 2 phylogenetic subgroups (Central/South-West European and Balkan) with bush-like topology. Purifying selection shapes their evolution, however, strong evidence of positive selection was revealed in 7 non-structural protein codons of NS1, NS4B and NS5. Thirty-two amino-acid substitutions were fixed in different phylogenetic subgroups, indicating that random genetic drift is responsible for the majority of evolutionary changes. Virus migration, followed by subsequent local evolution is responsible for continuously evolving strains throughout Europe. In total, 10 virus transitions between discrete geographical locations, involving virus spread from Central Europe to other regions, were highly supported. Three novel, independent introductions from Hungary and Bulgaria were responsible for the 2018 re-emergence of WNV in Northern Greece, indicating that Hungary remains an important ecological niche for the virus and has a central role for the dissemination of novel strains in the Balkans. In Northern Greece, tMRCA estimations indicated that a 1-to 2-year period of silent enzootic transmission precedes spread to dead-end hosts. Reconstruction of WNV population dynamics, from WGS data, revealed epidemic patterns characterized by 3- to 5-year oscillations in Europe. Future studies are necessary to determine the possible driving factors for these fluctuations i.e. avian herd immunity and climatic conditions affecting mosquito and bird populations. Maintaining adequate epidemiological surveillance with emphasis on obtaining WGS data, in areas at risk, is crucial for understanding the epidemiology and transmission patterns of WNV. It can further support integrated programs for risk assessment of virus circulation dynamics, aiming to targeted prevention and response measures for veterinary and public health in Europe.