Phylogenetic and antigenic characterization of reassortant H9N2 avian influenza viruses isolated from wild waterfowl in the East Dongting Lake wetland in 2011–2012

Background Wild waterfowl are recognized as the natural reservoir for influenza A viruses. Two distinct lineages, the American and Eurasian lineages, have been identified in wild birds. Gene flow between the two lineages is limited. The H9N2 virus has become prevalent in poultry throughout Eurasia, and mainly circulates in wild ducks and shorebirds in North America. Methods In this study, 22 H9N2 avian influenza viruses were isolated from wild waterfowl feces in East Dongting Lake Nature Reserve in November 2011 and March 2012. The phylogenetic, molecular, and antigenic characteristics of these viruses were analyzed based on analyses of the whole genome sequence of each isolate. Results Phylogenetic analyses indicated that these H9N2 viruses were generated by reassortment events. The HA, NA, PA, and NS genes were derived from the American gene pool, and the other four genes were derived from the Eurasian gene pool. Antigenic analyses indicated that these viruses were significantly different from the Eurasian lineage viruses. Conclusions This study presents the isolation of novel intercontinental recombinant H9N2 viruses from wild waterfowl in the East Dongting Lake wetland. The novel genotype H9N2 virus has not been detected in poultry in the region yet, and may be transmitted to naïve birds in poultry farms. Therefore, our results highlight the need for ongoing surveillance of wild birds and poultry in this region.


Background
Influenza A viruses can be divided into different subtypes based upon the two surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA) [1,2]. Wild waterfowl are recognized as the natural reservoir of influenza A viruses, especially low pathogenic avian influenza virus (LPAIV) [3]. To date, all of the HA (H1-H16) and NA (N1-N9) subtype avian influenza viruses (AIVs) have been identified in wild waterfowl, with the exceptions of H17N10 and H18N11, which were isolated from bats [2,4,5]. Previous studies indicated that migratory birds play an important role in the emergence of epidemics in birds, pigs, horses, and humans [1]. Generally, AIVs are nonpathogenic in wild birds, although they sometimes cause significant morbidity and mortality when transmitted to domestic poultry [6][7][8].
The H9N2 subtype influenza virus was first isolated from a turkey in Wisconsin in 1966. It has been most prevalent in wild ducks and shorebirds and has shown no evidence of establishing a stable lineage in land-based poultry in North America [9]. Since the l990s, H9N2 subtype influenza viruses have become prevalent in land-based poultry across East Asia, Middle Asia, Europe and Africa. Epidemiological and phylogenetic studies indicate that three distinct sublineages of H9N2 subtype influenza viruses have been established: Ck/Bj/94-like (A/chicken/Beijing/1/94 or A/duck/Hongkong/Y280/1997), G1-like (A/Quail/Hong Kong/G1/1997) and Korea-like (A/chicken/Y439/1997 or A/chicken/Korea/383490-p96323/1996) [10][11][12].
Poultry infected with H9N2 subtype AIVs have reduced egg production and moderate morbidity when coinfected with other viruses or bacteria [13]. Since 1999, some H9N2 viruses have been identified with potential human virus receptor specificity, and have been occasionally transmitted to human and swine [14][15][16]. Moreover, H9N2 viruses may have contributed internal genes to the highly pathogenic avian influenza (HPAI) H5N1 virus in Hong Kong in 1997 and the novel H7N9 avian influenza virus in mainland China in 2013 [17,18]. Therefore, the H9N2 subtype avian influenza viruses have been classified as candidate viruses with pandemic potential [19,20]. The threat to the poultry industry and public health posed by H9N2 subtype avian influenza viruses should not be ignored.
Hunan East Dongting Lake Nature Reserve is one of the largest wetlands in mainland China and is an important overwintering area and staging site for migratory birds that fly along the East Asia-Australia flyway [21]. Moreover, numerous duck farms are found around the lake [22,23]. Every migrating season, wild birds congregate at the lake where they share a common habitat with domestic ducks, which provides an opportunity for virus reassortment. During active surveillance of AIVs between 2011 and 2012, we isolated H9N2 viruses from wild waterfowl in East Dongting Lake wetlands in November 2011 (7 isolates) and March 2012 (15 isolates). The whole genome sequences of all 22 isolates were obtained, and phylogenetic trees for each gene segment were generated to analyze the relationship of these isolates with other circulating viruses in wild birds or poultry. Furthermore, we performed antigenic analyses to investigate the antigenic characteristics of the isolates.

Virus isolation and sequence analysis
In total, 6621 environmental samples were collected in Hunan East Dongting Lake wetland. H9N2 subtype avian influenza viruses were isolated in November 2011 (7 isolates) and March 2012 (15 isolates) from wild waterfowl feces (Table 1

Phylogenetic analyses
To further characterize the evolution of the 22 H9N2 viruses isolated from wild waterfowl, phylogenetic trees for each gene were constructed. In the HA gene tree, the H9 subtype AIVs clustered into five distinct lineages: G1-like, Y280-like, Korea-like, and two American lineages. The strains that clustered with the G1-like, Y280like, and Korea-like sublineages are known to have been endemic in domestic poultry in Eurasia and Africa [25]. All of the 22 isolates belonged to the American lineage II, which included viruses that are prevalent in wild birds or turkeys in North America. Some genetic exchange was observed between the North American and Eurasian strains, as a strain isolated in Korea clustered with an American lineage, and several H9 subtype AIVs isolated in America clustered with the Korea-like sublineage ( Figure 1a). Similar evolutionary patterns were observed in the N2 gene tree, which also clustered into four lineages: G1-like, Y280like, Korea-like and American lineages. The NA genes of all isolates clustered with the American lineage, and were most closely related to an H5N2 AIV subtype isolated from a snow goose in Montana (Figure 1b). The PA and NS genes of the isolates both clustered with the American lineage and showed high nucleotide similarity with strains isolated from wild birds in North America (Figure 1e and h). By contrast, the PB2, PB1, NP, and M internal genes clustered with the Eurasian gene pool, but were distinct from the H9N2 viruses that circulated in Eurasia and Africa and the corresponding internal genes of the novel H7N9 AIV (Figure 1c, d, f, and g).
Phylogenetic analyses indicated that these isolates were novel recombinant H9N2 viruses. In these viruses, 4 genes (HA, NA, PA, and NS) were derived from the American AIV gene pool and 4 genes (PB2, PB1, NP, and M) were derived from the Eurasian gene pool ( Figure 2).

Molecular analysis
The molecular characteristics of the 22 H9N2 isolates were compared with representative H9N2 virus strains circulating in Eurasia and America. The amino acid sequence of the cleavage sites of all isolates possessed a single basic amino acid (R) within the HA connecting peptide (VPELPKGR↓GLF), which is a typical feature of LPAIVs [1]. The HA receptor-binding pocket included the avian-like motif, Q226 and G228 (H3 numbering, Table 3), suggesting that these viruses could preferentially bind to an avian-like receptor (α2, 3-linked sialic acid) [26]. Analysis of potential HA protein N-X-S/T glycosylation site motifs revealed 8 sites at positions 29, 82, 141, 218, 298, 305, 492, and 551 (Table 4). There were no deletions in the NA stalk region, and no H274Y or N294S substitutions were observed in the NA protein. These properties indicated that these viruses would be sensitive to NA inhibitors, such as oseltamivir [27,28]. They encoded amino acids E and D at positions 627 and 701 of the PB2 protein, respectively, which are characteristics of AIVs [29][30][31]. No amino acid mutations associated with amantadine resistance were observed in the M2 ion channel protein. Additionally, no substitutions associated with increased virulence in mammals were detected in the PB2, PB1, or NS proteins (data not shown).

Antigenic analysis
To assess the antigenic properties of our novel H9N2 isolates, we performed Hemagglutination inhibition (HI) assays with 5 antisera raised against representative H9N2 viruses and two antisera raised against H9N2 isolates from this study (C2148 and PC2539). The HI antibody titers that recognized the C2148 and PC2539 viruses were much higher than the titers against the other H9N2 representative viruses, including the antiserum that was raised against a virus isolated from shorebirds in America (Sb/DE/249/ 06) ( Table 5). These results indicated that the H9N2 viruses isolated in this study were antigenically distinct from the previously identified H9N2 viruses.

Discussion
It is generally accepted that wild waterfowl are the natural reservoir for AIVs, and play important roles in the perpetuation and dissemination of AIVs, especially LPAIV. Previous surveillance studies indicated that AIVs circulate in diverse bird species, including Anseriforms (ducks, geese, and swans) and Charadriiforms (gulls and terns). AIVs preferentially infect cells that line the intestinal tract and numerous viruses can be excreted at high concentrations in wild waterfowl feces. Influenza viruses remain infectious in lake water for up to 4 days at 22°C and for over 30 days at 0°C. Contaminated lake water may result in efficient transmission to naïve birds by the fecal-oral route [32][33][34]. Migrating birds annually travel between breeding and overwintering sites, so AIVs harbored by migrating birds can be distributed along the migration flyway. The Hunan East Dongting Lake Nature Reserve is located along the East Asia-Australia flyway and is a major staging and overwintering site for migratory birds. Moreover, many domestic duck farms are located in this region and employ a free-range style to raise domestic ducks [22]. Because wild waterfowl and domestic ducks may share common habitats, water, and food, genetic exchange between different subtypes of AIVs circulating in wild waterfowl and domestic poultry may be possible in this location [35].
In this study, we obtained 22 H9N2 subtype AIV isolates from wild waterfowl in East Dongting Lake wetland in November 2011 and March 2012. Genetic analyses showed that the 22 H9N2 viruses shared over 99% nucleotide sequence homology with the strain A/Egret/ Hunan/1/2012 (H9N2) in all eight gene segments, which was isolated from an egret in the same region in January 2012. These findings indicated that these viruses had a common origin. Long-term surveillance studies in America and Eurasia demonstrated that the most prevalent AIVs in wild birds could be separated into Eurasian and American lineages. While intercontinental virus exchange existed in migrating birds, its frequency was limited and invasions by whole viruses were not observed [36]. Instances of invasion by the American lineage H9N2 virus into the Eurasian   continent have been reported [24]. Notably, the 22 isolates reported in this study were novel recombinant H9N2 subtype AIVs, which encoded genes derived from both the American and Eurasian gene pools circulating in wild birds. Therefore, genes that originated from the American lineage may have been carried by migratory birds from North America, leading to the generation of multiple AIV reassortants circulating in wild birds in Eurasia. Moreover, this recombinant genotype was detected both by us and Chen and colleagues in different wild birds in three months during 2011-2012 (November 2011, January 2012, and March 2012), suggesting that this virus has been circulating in wild birds in this region [24]. Although H9N2 viruses of this genotype have not been reported in other regions, they may have been distributed to other places by migratory birds along their flyways. H9N2 AIVs have been endemic in domestic poultry across Eurasia and Africa since the late 1990s. The wide circulation of H9N2 viruses provides more opportunities for reassortment with other AIV subtypes in poultry. In the last few decades, H9N2 viruses may have provided internal genes to the HPAI H5N1 and the novel H7N9 avian influenza viruses [17,18]. Additionally, H9N2 viruses have been occasionally transmitted from poultry to mammals (including humans and swine), and some of the H9N2 viruses showed the ability to bind efficiently to α2, 6-linked sialic acid, which can indicate human virus receptor specificity [15,26]. Further studies indicated that H9N2 viruses that encoded Leu226 could replicate in ferrets and be transmitted by direct contact [37]. Therefore, many researchers believe that the H9N2 viruses have the potential to cause a future pandemic. Even if they do not directly cause a pandemic, they could indirectly contribute to an influenza pandemic by contributing internal genes to a reassortant virus [17,18,38]. Our findings demonstrated that all 22 isolates were LPAIVs, and we observed neither G226L substitutions in HA proteins nor any other pathogenicity-associated mutations in other viral proteins. Nevertheless, the pathogenicity of these viruses should be further accessed in poultry and mammalian animal models.
Antigenic analyses suggested that the antigenicity of the isolates were significantly different from the G1-and Y280-like viruses, which have been dominant in domestic poultry in Eurasia and Africa, and the "America" virus, which is closely related to the Korea-like viruses. The antigenic and phylogenetic analyses were generally consistent with each other. Importantly, poultry in mainland China have been widely vaccinated with commercial inactivated vaccines that contain representative prevalent influenza strains circulating in poultry in Eurasia, but not strains of the American lineage. Therefore, the currently used commercial vaccines may not protect poultry or prevent the transmission of the novel H9N2 subtype viruses in China.

Conclusions
In summary, the 22 H9N2 viruses isolated from wild waterfowl in 2011-2012 were novel reassorted H9N2 subtype AIVs with similar genotypes. All isolates encoded genes for proteins with low pathogenic characteristics. Determining whether these H9N2 AIVs can transmit to poultry or other animals or further adapt to new hosts will require continuous monitoring in the future. Our findings extend our knowledge of the ecology of AIVs circulating in wild birds in the Dongting Lake region and highlight the importance of intercontinental AIV gene flow in migratory birds. Therefore, we emphasize the vital need for continued surveillance of AIVs in wild birds and poultry to prepare for and respond to potential influenza pandemics.

Virus detection and isolation
RNA was extracted from 200 μL fecal suspension using the BioRobot Universal system with the QIAamp One-For-All Nucleic Acid Kit (Qiagen, Hilden, Germany) in accordance with the manufacturer's instructions. Influenza A virus was screened using qPCR assay that targeted the influenza Matrix gene. Detection was performed using a Stratagene Mx3005P thermocycler with an AgPath RT-PCR Kit (Applied Biosystems, Foster City, CA, USA) using 5 μL eluate in 25 μL total volume. Each run included 2 negative and 2 positive control samples along with 92 samples. The

Phylogenetic analyses
We performed multiple sequence alignments using the MAFFT software, version 6. Both sequences of the representative H9N2 subtype influenza A virus strains circulating in America and Eurasia and homologous sequences that shared high nucleotide similarities with our H9N2 isolates were included in the phylogenetic analyses. Preliminary phylogenetic trees were constructed to infer the overall topology, using more than 500 sequences for each gene. In order to more clearly define the phylogenetic relationships of the 22 H9N2 virus isolates, representative sequences of each cluster were selected to generate neighbor-joining (NJ) trees. Phylogenetic estimates were calculated by performing 1000 neighbor-joining bootstrap replicates.

Antigenic analyses
Antigenic analyses were performed using 6 polyclonal ferret antisera against A/chicken/Beijing , two representative influenza strains, were also used in our study. HI assays were performed as previously described. Briefly, all sera were treated with receptor destroying enzyme II (RDE) (Denka Seiken, Tokyo, Japan) to remove nonspecific inhibitors of hemagglutination by adding 3 volumes of RDE to tubes with 1 volume of serum. Samples were incubated at 37°C for 16-18 h, and then inactivated at 56°C for 30 min. After RDE inactivation, 6 volumes of phosphate buffered saline (PBS; Thermo Scientific Hyclone) were added. The diluted sera were then serially diluted 2-fold with 25 μL PBS, and equal volumes of antigen (8 HA unit/50 μL) were added to each well. The plates were gently mixed and incubated at room temperature for 20-30 min. Viral titers were determined by adding 50 μL 0.5% turkey red blood cells to each well. The limit of detection for HI titers was ≤ 20.

Nucleotide sequence accession numbers
The nucleotide sequences generated in our study were deposited at GenBank under the accession numbers KF971946 to KF972121.

Competing interests
The findings and conclusions of this report are those of the authors and do not necessarily represent the views of the funding agency. We declare no conflict of interest.
Authors' contributions YLS and SXH designed the research; YZ performed research and drafted the manuscript; TB, YWH, ZHD, HZ, ZYB, MDY, and JFH collected samples and