Whole‐genome sequencing and antigenic analysis of the first equine influenza virus identified in Turkey

Background In 2013, there was an outbreak of acute respiratory disease in racehorses in Turkey. The clinical signs were consistent with equine influenza (EI). Objective The aim was to confirm the cause of the outbreak and characterise the causal virus. Methods A pan‐reactive influenza type A real‐time RT‐PCR and a rapid antigen detection kit were used for confirmatory diagnosis of equine influenza virus (EIV). Immunological susceptibility to EIV was examined using single radial haemolysis and ELISA. Antigenic characterisation was completed by haemagglutinin inhibition using a panel of specific ferret antisera. Genetic characterisation was achieved by whole‐genome sequencing using segment‐specific primers with M13 tags. Results A H3N8 EIV of the Florida clade 2 sublineage (FC2) was confirmed as the causal agent. The index cases were unvaccinated and immunologically susceptible. Phylogenetic analysis of the HA1 and NA genes demonstrated that A/equine/Ankara/1/2013 clustered with the FC2 strains circulating in Europe. Antigenic characterisation confirmed the FC2 classification and demonstrated the absence of significant drift. Whole‐genome sequencing indicated that A/equine/Ankara/1/2013 is most closely related to the viruses described as the 179 group based on the substitution I179V in HA1, for example A/equine/East Renfrewshire/2/2011, A/equine/Cambremer/1/2012 and A/equine/Saone et Loire/1/2015. The greatest diversity was observed in the NS1 segment and the polymerase complex. Conclusions The first recorded outbreak of EI in Turkey was caused by an FC2 virus closely related to viruses circulating in Europe. Antigenic and genetic characterisation gave no indication that the current OIE recommendations for EI vaccine composition require modification.


| INTRODUCTION
Equine influenza virus (EIV) of the H3N8 subtype is associated with an acute respiratory disease which has immense economic relevance due to its highly contagious nature and potential to disrupt equestrian events. 1,2 Each equine influenza (EI) virion contains 8 segments of negative sense viral RNA named; PB2, PB1, PA, HA, NP, NA, M and NS.
These segments encode at least 10 proteins which include the surface glycoproteins; haemagglutinin (HA) and neuraminidase (NA), the matrix ion channel (M2), the matrix protein (M1), three polymerases (PA, PB1 and PB2), the structural nucleoprotein (NP), the non-structural protein (NS1), the nuclear export protein (NEP), an accessory proapoptotic protein (PB1-F2) 3 and a recently discovered PA-X protein. 4 The HA glycoprotein is particularly important in the evolution of EIV as it stimulates a strong humoral antibody response following infection. 5,6 Mutations in the HA lead to antigenic drift and emergence of antigenically distinct EIV lineages. Subsequent to EI epizootics in 1989, phylogenetic analysis of the HA gene revealed divergence of the circulating H3N8 viruses into the Eurasian and American lineages. 7 In the 1990s, the American lineage diverged into the South America, Kentucky and Florida sublineages. 8 More recently, the Florida sublineage has predominated and evolved into 2 additional distinct clades, Florida clades 1 and 2 (FC1, FC2). 5 FC1 and FC2 viruses are prevalent in the USA and Europe, respectively. 1 However, both have caused outbreaks elsewhere FC1 in South Africa, 9 Japan, 10 Australia 11 and Europe 5,12,13 and FC2 in China 14 Mongolia 15 and India. 16 In endemic populations, EIV persists due to the existence of susceptible partially vaccinated or unvaccinated horses and compromised vaccine effectiveness, which occurs as a result of antigenic mismatch between vaccine and field strain viruses. 17,18 Experimentally infected animals vaccinated with a heterologous vaccine strain shed virus for longer than those that received a homologous vaccine. 19 In countries which are free from EI, the virus has frequently been introduced by the importation of infected vaccinated horses. 20

| Clinical detection
Nasopharyngeal swabs and serum samples were collected from 7 clinically affected horses. The nasopharyngeal swabs were tested by panreactive influenza type A real-time RT-PCR (qRT-PCR) described by Heine et al 2007 22 and the Directigen EZ FluA + B ™ rapid detection kit (BD Diagnostics, Oxford, England). Serum samples were tested by single radial haemolysis (SRH) 23 and the IDScreen influenza A antibody ELISA (IDvet, Grabels, France), a multispecies competition ELISA that detects antibodies against the internal nucleocapsid of the influenza A virus.

| Whole-genome sequencing
A/equine/Ankara/1/2013 was passaged twice in embryonated hens' eggs and diluted to 10 6 EID 50 /mL. RNA was extracted using QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany). For whole-genome sequencing (WGS), one-step PCR was undertaken using SuperScript ™ III One-Step RT-PCR System with Platinum ® Taq High Fidelity (Invitrogen, Carlsbad, California, USA) using overlapping M-13 labelled primers for each of the 8 EIV segments. 25 An additional NA primer (5′-GCCTCACAAAGTGGTTC-3′) was designed to obtain the start of the NA gene.
Genome segments were assembled using Seqman version 14.1.0 (118) 412, DNASTAR, Madison, WI. Multiple nucleotide, and amino acid sequence alignments of each segment were constructed using the ClustalW 26 accessory application in BioEdit sequence alignment editor version 7.2.5. 27 Maximum-likelihood phylogenetic trees were constructed using MEGA7 version 7.0.14 28 with HA1 and NA gene sequences mined from the NCBI GenBank and GISAID databases 29 (see Table S1 for accession codes). The optimum model for each tree was chosen based on the lowest Bayesian information criterion scores. Following phylogenetic classification based on the HA1 gene, the genome sequence of A/equine/Ankara/1/2013 was aligned with the OIE recommended FC2 representative vaccine strain A/equine/ Richmond/1/2007. Amino acid changes observed for A/equine/ Ankara/1/2013 were then compared to other European strains circulating in 2009-2015 to track evolution of EIV (see Table S2 for accession codes).

| Antigenic characterisation
Antigenic characterisation identified A/equine/Ankara/1/2013 as a FC2 strain of EIV (Table 1). A panel of specific antisera raised against FC2 viruses demonstrated similarly high HI titres for A/equine/Ankara/1/2013 as they did against their homologous virus. Additionally, up to an eightfold higher titre was observed with A/equine/Ankara/1/2013 for FC2 than FC1 antisera which confirmed the FC2 classification.

| Phylogenetic classification (HA1 and NA)
A maximum-likelihood phylogenetic tree confirmed that A/equine Ankara/1/2013 is an FC2 strain as it clustered with other isolates in that clade that have the I179V substitution, that is the "179 group"

| Whole-genome sequencing
Whole-genome sequencing was achieved using overlapping M-13 labelled segment-specific primer pairs. The data gener-     Table S1 substitutions that emerged around 2007 (E291D) and 2010 (P103L and V112I). 5,30 In 2011, two subpopulations were identified with an additional change at either position 144 or at 179. 30 Table S1 but the other seven substitutions appear to be unique to A/equine/

| DISCUSSION
The EI genome is subject to a slower rate of evolution than other mammalian influenza viruses; however, mutability as measured by the ratio of non-synonymous to synonymous nucleotide mutations (dN/ dS) was not significantly different for equine, 31 avian 32 or swine influenza viruses. 33 Therefore, the genome of EIV has the potential to experience significant antigenic drift in a similar but decelerated time frame to avian and swine influenza. 31  that the index case of FC2 was transported to Sweden from Spain through the Netherlands. 13 Other studies, based primarily on HA1 analysis, have also shown the ease with which EIV can spread as a result of transport of horses across Europe, 5 specifically within Ireland, 24 the UK 30 and France. 40,41 T A B L E 2 Amino acid differences between A/equine/ which possess a non-inflammatory motif (amino acid L62, H75 and Q79) (Fig. S3.3).
In conclusion, the outbreak of respiratory disease at the racetrack in Ankara was caused by an FC2 virus closely related to those circulating in Europe. There was no indication that the virus was particularly virulent and the majority of horses recovered clinically