Background
Equine influenza is one of the major infectious respiratory diseases in equids worldwide. The causative agent, equine influenza virus (EIV), is a highly contagious pathogen endemic in most parts of the world. Like other influenza A viruses, EIV has two surface glycoproteins, haemagglutinin (HA) and neuraminidase (NA). HA plays an essential role in virus entry by attaching to host cell sialic acid receptors and promoting membrane fusion [
1]; it is a major target of neutralizing antibodies and is therefore an important component of commercial vaccines. NA has sialidase activity and is thought to play a role in virus entry as well as exit. Antibodies elicited against influenza NA are also known to contribute to protection and have recently been shown to block both sialidase activity and virus adsorption [
2], however their importance for immunity against EIV remains unclear.
Two influenza subtypes are known to infect horses, H7N7 and H3N8. H7N7 equine influenza was first isolated in Europe in 1956, followed by H3N8 in 1963 [
3,
4]. The two subtypes co-circulated, with reassortment, until the last isolation of H7N7 in the late 1970s [
5,
6]. Following its emergence in South America, most likely from an avian source EIV H3N8 diverged phylogenetically into American and Eurasian lineages [
7,
8], with sufficient antigenic differences to warrant inclusion of both in vaccines. Subsequently three sub-lineages named Kentucky, Florida and South American (Argentinian) emerged within the American lineage [
9], and the Florida sublineage has further diverged into distinct clades 1 and 2 [
10,
11]. Currently, viruses from the Florida sub-lineage are prevalent worldwide and inclusion of representatives of both clade 1 and clade 2 are recommended for vaccines [
12]. Florida clade 1 (FC1) viruses are considered endemic in North America, but have caused major outbreaks in Australia, Japan and South Africa [
13‐
15]. There have also been smaller outbreaks reported in Europe from 2007 to 2009 [
16‐
18]. In Asia and Europe viruses from Florida clade 2 (FC2) predominate, causing significant outbreaks in China, India and Mongolia in recent years and smaller scale outbreaks in multiple countries of Europe [
19‐
23].
Vaccination has been widely used for the control of equine influenza and commercial vaccines have been available for several decades. Despite this, horses immunized with potent vaccines occasionally fail to provide adequate protection. Vaccine breakdown can occur as a consequence of antigenic drift in HA [
24,
25], a well-established phenomenon amongst mammalian influenza A viruses. Cumulative mutations over time result in amino acid substitutions in HA, allowing eventual immune escape of emerging strains from the humoral protection of the host, previously acquired against earlier strains [
26,
27]. Additionally, reassortment among gene segments during mixed infections of influenza viruses can contribute to the appearance of new strains, a process known to have occurred amongst recent EIV strains from different sub-lineages [
10,
11]. To counteract the effects of antigenic drift, surveillance and vaccine strain selection are carried out for EIV. In 2010 the World Organisation for Animal Health (Office International des Epizooties, OIE) updated the recommendations for suitable antigens for commercial vaccines to include a representative strain from both Florida clade 1 and Florida clade 2. However, many vaccine manufacturers still provide products with outdated vaccine strains.
During 2012, an extensive outbreak of equine influenza spread through South America, first reported in Chile in December 2011 and affecting Brazil by February 2012. The outbreak spread rapidly through Chile, Brazil, Uruguay and Argentina [
28,
29] but it is thought likely that other countries in South America were also affected. Horses carrying EIV were transported from Uruguay to Dubai, highlighting the risk to other countries [
23]. Prior to this outbreak, the most recent Brazilian EIV strain to undergo molecular characterisation was A/eq2/Brazil/1987 [
30]. This paper describes the 2012 Brazilian outbreak and the molecular and antigenic characterisation of a representative EIV isolate, A/equine/Rio Grande do Sul/2012.
Discussion
Large-scale epidemics of EIV have been reported since the first transmission of H3N8 in horses in the early 1960s. In some instances these have occurred in immunologically naïve populations, such as the outbreak in South Africa in 2003 and that in Australia in 2007. Epidemics that also affect vaccinated animals suggest that significant antigenic drift away from vaccine strains may have occurred. In 2012, a Florida clade 1 EIV spread through several countries in South America, affecting both vaccinated and unvaccinated animals. In some cases, animals were vaccinated with products that contained outdated strains that did not comply with current OIE recommendations, it is therefore not surprising that these animals showed clinical signs. However, animals vaccinated with a whole inactivated product containing an OIE-compliant Florida clade 1 virus were also affected. Although detailed vaccination records were not available, this is an early indication that the OIE recommendations for this sub-lineage may need updating in the near future.
Antigenic mismatches have been a major variable in well-documented vaccine breakdowns worldwide [
14,
36,
37]. In horses immunized with inactivated vaccines, there is a high correlation between the level of antibodies against HA, determined by single radial haemolysis assay, and the level of protection. However, this applies in cases where the infecting EIV is homologous to the vaccine antigen [
38]. When the infecting virus is antigenically heterologous, the mismatch between the vaccine and the virus can decrease the vaccine efficacy, and consequently higher levels of circulating antibodies are needed in order to neutralize the infection [
39]. In such instances, vaccines may continue to offer protection against clinical signs but not necessarily reduce virus shedding and these animals become a potential source of infection to susceptible horses [
40]. In the absence of serological data we cannot rule out the possibility that vaccinated animals had poor antibody titres, which may have played a part in the outbreak.
Here we describe genetic and antigenic analysis, including determination of the full genome sequence, for a representative isolate of the outbreak in Brazil in 2012. Comparison of HA and NA sequences indicated that there were multiple substitutions between the glycoproteins of these viruses and the OIE-recommended strain South Africa/4/03, however those in HA did not fulfil the much-quoted requirement for ‘four or more changes affecting two or more antigenic sites in HA1’. Nor were there multiple changes in regions previously shown to be important for EIV [
37]. Substitutions of interest included one within antigenic site A of H3 (A138S) and a change within the 220 loop of the receptor-binding region (I223V), which may alter avidity for sialic acid receptors. The reciprocal substitution V223I has been observed in some recent human H3N2 clinical isolates [
41,
42]. None of the amino acid changes observed here were unique to the viruses circulating in South America and were typical of those seen in USA isolates from 2011 onwards, which did not cause large scale epidemics or reports of mass vaccine breakdown.
The results of antigenic analysis using a ferret antisera panel supported the phylogenetic data, with the Brazilian isolates showing good recognition by sera raised against other members of Florida clade 1. The converse also applied, with a ferret serum raised against Rio Grande do Sul/1/12 reacting well against earlier members of the Florida clade 1 sub-lineage. Recognition by sera against Kentucky sublineage viruses was poor, similar to results for earlier FC1 viruses [
10,
23]. These data do not suggest that there has been extensive antigenic drift away from the recommended OIE reference strain South Africa/4/03. Despite this, vaccines containing the recommended strain South Africa/4/03 did not perform as well as expected.
Several of the vaccines available in Brazil during 2012 contained outdated strains, including Kentucky/1/92, Kentucky/94 and Kentucky/97. The antigenic analyses described here indicated that high titres were only obtained with sera from Florida clade 1 strains, not from sera against the Kentucky sub-lineage. Vaccine breakdown with products containing these strains provide field data in support of the decision by the OIE to recommend inclusion of both Florida clade 1 and Florida clade 2 strains. More worryingly, horses that had been vaccinated with South Africa/4/03 were also infected and showed clinical signs. This may have been a consequence of antigenic mismatch with the Rio Grande do Sul/2012 strain, although the HI using ferret antisera had shown an efficient cross-reaction between both strains. However, high levels of cross-reaction are not necessarily correlated to cross-protection [
36,
39]. Alternatively, poor vaccine performance may have been due to a lack of potency or poor adherence to vaccine manufacturer’s recommendations, rather than antigenic drift. Several other risk factors have been associated with influenza infection of vaccinated horses during the last decade [
43]. In this instance, the use of aluminum hydroxide as an adjuvant may be relevant, as seen in the UK outbreak of 2003 ([
39], Table
1). Simple adjuvants such as this may require a closer match between vaccine antigen and outbreak strain in order to be effective than more complex adjuvants that stimulate cell-mediated immunity [
43].
The entry point for equine influenza into Brazil in 2012 is not known, nor has the precise origin of the South American outbreak virus been determined. Based on the reports cited here, after notification by Chile, the outbreak spread quickly across the south of the continent to Uruguay and Southern Brazil then further north. Phylogenetic analysis of HA and NA here and HA1 sequences reported recently for Argentina and Uruguay [
29] showed that South American isolates from 2012 were closely related to the 2011–12 isolates from the USA. These data suggest that these viruses are likely to have originated from the USA, but through possible importation to Chile rather than gradual spread north to south.
Conclusions
The equine influenza epidemic in South America in 2012 was caused by a virus belonging to FC1, similar to those circulating in the USA in the previous year. Phylogenetic analysis indicated that strains from FC1 had acquired several mutations in both HA and NA compared to the OIE recommended strains from 2003, however antigenic assessment using ferret antisera did not identify extensive antigenic drift. In Brazil, the use of vaccine products containing outdated strains may have contributed to the outbreak, since most antigens available on the market fell into this category. However, EIV was also isolated from clinically affected horses vaccinated with an OIE recommended strain. Poor control of horse movements following competition events is likely to have contributed to the spread of EIV.
Consent for publication
Not applicable.
Availability of data and materials
All novel sequence data generated have been deposited with the influenza virus sequence database, GISAID. Accession numbers are provided in Additional file
1: Table S1.
Competing interests
AW, AR and DE declare that they have no competing interests. EAB works as a consultant for and LEdSF is employed by Laboratórios Vencofarma do Brasil. Laboratórios Vencofarma did not provide financial support for these genetic and antigenic studies, or have any influence in the conclusions of this study
Authors’ contributions
This study is a result of collaborative work. DE, EAB, LEdSF, AFA and AAA conceived this study and participated in its design. LEdSF collected field samples from horses and provided data regarding the epidemic in Brazil. EAB, AW and AR carried out the experimental work and sequence analysis. EAR, AW and DE drafted the manuscript. AR participated in the discussion and modification of the manuscript. All authors read and approved the final manuscript.