On October 28, 2016, we isolated an H5N6 subtype HPAIV from 1 of 391 fecal samples collected from wild birds in Gokgyo-cheon, South Korea (36°45′12.3″N, 127°07′12.7″E). Gokgyo-cheon is a wild bird habitat for wintering of migratory waterfowl, including mallard (Anas platyrhynchos), spot-billed duck (Anas poecilorhyncha), Mandarin duck (Aix galericulata), and common teal (Anas crecca). The species of the positive fecal sample was identified as Mandarin duck on the basis of DNA barcoding technique as described (9). There were no detectable clinically ill or dead wild birds at the sampling site.

Full-length genome sequencing and phylogenetic analysis were conducted to trace the origin of A/Mandarin_duck/Korea/K16-187-3/2016(H5N6) virus, hereafter referred to as MD/KR/2016. Methods used are detailed in Technical Appendix 1. We entered genome sequences in the GISAID (Global Initiative on Sharing All Influenza Data) EpiFlu database (https://www.gisaid.org) under accession nos. EPI861480–EPI861488. Strains used in analysis are shown in Technical Appendix 2.

The isolate was identified as an HPAIV on the basis of multiple basic amino acids at the HA proteolytic cleavage site (PLRERRRKR/G). GISAID BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) searches indicated that H5 and N6 genes had high nucleotide identity in HA (99.17%) and NA (99.24%) with A/great_egret/Hong_Kong/00032/2016 (H5N6) (Table 1). Internal gene segments, except the polymerase basic 1 (PB1) gene, had high nucleotide identity with other H5N6 subtypes isolated in Guangdong and Jiangxi, China (PB2, 99.09%; polymerase acidic, 98.96%; nucleoprotein, 99.16%; matrix, 98.98%; and nonstructural protein [NS], 98.31%). However, the PB1 gene had high nucleotide identity (97.01%) with H4 low pathogenicity avian influenza viruses (LPAIVs).

In previous phylogenetic analyses, the HA gene of clade 2.3.4.4 viruses was divided into 4 distinct subgroups (Technical Appendix 1 Figure 1) (10). Group intercontinental A (icA) contains H5N8 subtype virus and its reassortant viruses identified in China, South Korea, Japan, Taiwan, Canada, the United States, and countries in Europe during 2013–2016. Group B contains H5N8 subtype viruses identified in China and South Korea during 2013–2014, and in Russia in late 2016. Group C contains H5N1 and H5N6 subtype viruses identified in China, Vietnam, Laos, and Hong Kong, including isolates from humans in Guangdong, Yunnan, and Hunnan Provinces, China. Group D contains H5N6 subtype viruses identified in China and Vietnam, including an isolate from a human in Sichuan Province, China. The HA gene of MD/KR/2016 virus belonged to group C and clustered with H5N6 subtype viruses isolated from humans, cats, and the environment in Guangdong during 2014–2015 and a migratory aquatic bird in Hong Kong during January 2016 (A/great_egret/Hong_Kong/00032/2016 [H5N6]) (Technical Appendix 1 Figure 1).

A previous study reported that A/environment/Guangdong/GZ693/2015 (H5N6), hereafter referred to as GZ693/2015(H5N6), is a 7:1 gene reassortant virus between H5N6 HPAIV and LPAIVs found in southern China (7). MD/KR/2016 clustered with GZ693/2015(H5N6) virus for all 8 genes (Technical Appendix 1 Figure 2). In particular, the HA, NA, PB2, polymerase acidic, nucleoprotein, matrix, and NS protein genes clustered with GZ693/2015(H5N6) and other clade 2.3.4.4 group C H5N6 viruses. The PB1 gene clustered with GZ693/2015(H5N6) (nucleotide identity 92.79%) and LPAIVs, such as H3N2 and H4N2 subtype viruses, from southern China. Phylogenetic analysis and BLAST search collectively suggest that MD/KR/2016 virus had an identical genotype to GZ693/2015(H5N6).

Most of clade 2.3.4.4 group C viruses have leucine or serine at position 129 (H5 numbering) in HA protein. However, MD/KR/2016 had a single amino acid deletion at position 129 (Table 2), as did A/great_egret/Hong_Kong/00032/2016 (H5N6). This deletion at position 129 and phylogenetic network analysis suggested that MD/KR/2016 is closely related to H5N6 subtypes isolated from wild birds in Hong Kong in 2016 (Technical Appendix 1 Figure 3). MD/KR/2016 contained the mutation associated with increased virulence in mammals and mammalian transmissibility (S123P and T156A mutations in the HA gene; P42S and D92E mutation, and elongated C-terminus with PDZ binding motif in NS gene). However, this isolate lacked the Q226L and G228S mutations in HA, which have been associated with increased binding to human-type receptor (α-2,6–linked sialic acid) and lacked Q591K, E627K and D701N mutations in PB2, which have been associated with enhanced pathogenicity and adaptation to mammalian hosts (11). All of the 9 H5N6 subtype human isolates of group C lacked the Q226L and G228S mutations in HA, but 5 viruses contained the E627K mutation in PB2 (Table 2), suggesting that some purported mammalian adaptation amino acid substitutions were not necessary for sporadic virus infection of H5N6 HPAIV in humans.

Wild aquatic birds have been suspected to play a key role in dissemination of HPAIVs to various regions, as seen with clade 2.2 H5N1 HPAIV in 2005, clade 2.3.2.1 H5N1 HPAIV in 2009, and clade 2.3.4.4 H5N8 HPAIV in 2014 (4,12). Some populations of Mandarin ducks are year-round residents in South Korea and Japan; others populations migrate between Russia and eastern Asia (13). In South Korea, HPAIV was detected from Mandarin duck samples in 2010 (H5N1) and 2014 (H5N8) (14,15) and again in 2016 during this study, suggesting that Mandarin ducks are a major host species for clade 2.3.4.4 H5 HPAIV and can disseminate the virus throughout South Korea and into other countries. Detection of the H5N6 HPAIV clade 2.3.4.4 in a migratory bird species in South Korea; reports of H5N6 outbreaks in poultry from China, Laos, and Vietnam; and diagnosis of lethal human cases of highly homologous H5N6 viruses in China raise a concern over the potential for broad geographic dissemination of zoonotic H5N6 HPAIV by wild birds outside eastern Asia.