Genetic characterization and pathogenic potential of H10 avian influenza viruses isolated from live poultry markets in Bangladesh

Fatal human cases of avian-origin H10N8 influenza virus infections have raised concern about their potential for human-to-human transmission. H10 subtype avian influenza viruses (AIVs) have been isolated from wild and domestic aquatic birds across Eurasia and North America. We isolated eight H10 AIVs (four H10N7, two H10N9, one H10N1, and one H10N6) from live poultry markets in Bangladesh. Genetic analyses demonstrated that all eight isolates belong to the Eurasian lineage. HA phylogenetic and antigenic analyses indicated that two antigenically distinct groups of H10 AIVs are circulating in Bangladeshi live poultry markets. We evaluated the virulence of four representative H10 AIV strains in DBA/2J mice and found that they replicated efficiently in mice without prior adaptation. Moreover, H10N6 and H10N1 AIVs caused high mortality with systemic dissemination. These results indicate that H10 AIVs pose a potential threat to human health and the mechanisms of their transmissibility should be elucidated.


Results
Isolation of AIVs from Bangladeshi LPMs. We collected 29,305 samples from LPMs through active surveillance in Bangladesh from 2008 to 2017. We screened all samples and confirmed the H5 subtype with quantitative reverse transcription PCR (qRT-PCR). We then subtyped all samples that tested negative for the H5 subtype by HA and NA sequencing. We detected eight different HA subtypes of AIVs, including H1 (5 isolates), H3 (12 isolates), H4 (4 isolates), H5 (182 isolates), H6 (3 isolates), H7 (1 isolate), H9 (1165 isolates), and H10 (8 isolates) 19 . We noted that eight H10 subtype AIVs were isolated from samples collected from three different LPMs in Bangladesh. All of the AIVs were isolated from ducks, with the exception of one H10N7 AIV that was isolated from a chicken (Table 1).

Molecular and phylogenetic analysis of H10 AIVs.
To better understand the genetic relationship of the H10 viruses, we sequenced the complete genomes of eight H10 AIVs isolated from LPMs in Bangladesh. Phylogenetic analysis of the HA sequences revealed that they belonged to the Eurasian lineage. We classified the HA genes into two groups based on branch clustering of the phylogenetic tree (Fig. 1). Group 1 contained H10N7 and H10N9 AIVs in branches that also contained viruses from Europe and Asia. The HA gene of four H10N7 AIVs were most closely related to those of H10N7 AIVs isolated from Jiangxi, China in 2009. Group 2 contained H10N1 and H10N6 AIVs along with Asian AIVs and human H10N8 AIVs (Fig. 1).
The amino acid sequence of the HA protein of all H10 AIVs included a single arginine residue at the cleavage site of HA1 and HA2 (PELMQGR↓GLF), indicating that they were low pathogenic strains. We examined the HA proteins in these AIVs for the presence of N-linked glycosylation motifs by interrogating the proteins for N-(P-[S/T]-P) motifs, which indicate potential N-linked glycosylation sites. We identified four potential glycosylation sites in HA1 at positions 22, 38, 242, and 296, which are highly conserved in all H10 AIVs.
Phylogenetic analysis of the NA genes revealed that all NA subtypes (i.e., N1, N6, N7, and N9) belonged to the Eurasian lineage ( Supplementary Fig. S1). The NA genes of four H10N7 AIVs were most closely related to those of H10N7 AIVs isolated from Jiangxi, China in 2009. The NA genes of the H10N9 and H10N6 AIVs were most closely related to those of AIVs isolated from free-range ducks in the Tanguar haor area in Bangladesh in 2015. This indicates that the NA genes may circulate in Bangladesh between poultry sold in LPMs and free-range ducks. The NA gene of the H10N1 AIV was most closely related to those of H6N1 AIVs that were previously isolated fromChina. In addition, we found that some North American-lineage AIVs from Alaska were related to those from the Eurasian lineage ( Supplementary Fig. S1). Phylogenetic analysis of the polymerase basic 2 (PB2) genes revealed that they formed two distinct groups within the Eurasian lineage, all belonging to the Eurasian lineage ( Supplementary Fig. S2a). The PB2 gene of H10N7 and H10N9 AIVs were located in one group and shared 99% identity with the closest PB2 gene of A/ ruddy shelduck/Mongolia/598C2/2009 (H7N3). The H10N6 AIV PB2 gene was most closely related to AIVs isolated from free-range ducks in the Tanguar haor area in Bangladesh. The PB2 gene of the H10N1 AIV shared 98% nucleotide identity with that of A/tufted duck/Mongolia/1409/2010 (H1N1).  The polymerase basic 1 (PB1) genes of eight H10 AIVs clustered into two groups and belonged to the Eurasian lineage ( Supplementary Fig. S2b). All of our isolates expressed a PB1-F2 protein consisting of 90 amino acids. Furthermore, the H10N1 and H10N9 AIVs had an N66S substitution in PB1-F2 (Supplementary Table S1), which increases virulence and replication efficiency in mammalian hosts 23 . The polymerase acidic (PA) and nucleoprotein (NP) genes of the H10 AIVs clustered into two groups, all of which were of the Eurasian lineage ( Supplementary Fig. S2c,d). The PA and NP genes of the H10N7 and H10N9 AIVs were related to those of AIVs isolated from Europe and Mongolia, whereas the NP gene of the H10N1 and H10N6 AIVs were most closely related to viruses isolated from Eastern China and free-range ducks in the Tanguar haor area in Bangladesh. The matrix (M) and nonstructural (NS) genes of all eight H10 AIVs segregated into two groups ( Supplementary  Fig. S2e,f). The amino acid substitutions at positions 26, 27, 30, 31, and 34 in the M2 protein facilitate resistance to some anti-influenza drugs (i.e., M2 ion channel blockers). We did not observe any of these substitutions in the H10 AIVs we isolated. However, we found that all H10 AIVs contained P42S and V149A substitutions in the NS1 protein, which are associated with virulence and pathogenicity in mammals 24,25 . We did not detect any deletions in the NS1 proteins in any of the H10 AIVs, and all were of allele A and Eurasian lineage.

Antigenic analysis.
To determine the antigenic relations among H10 AIVs isolated from LPMs in Bangladesh, we used a panel of polyclonal and monoclonal antibodies against Eurasian and North American viruses. The H10 AIVs were homogeneous and similar to those of the Eurasian lineage (Table 2). Four H10 AIVs reacted with antisera against A/laughing gull/Delaware Bay/209/2013 (H10N8) and A/glaucous gull/Iceland/ CDX16-4552/2015 (H10N7) from North America, although hemagglutination inhibition (HI) assay titers with the H10N1 and H10N6 AIVs were one-to two-fold higher than were those of the other AIVs (Table 2). Two monoclonal antibodies against A/chicken/Jiangxi/34609/2013 (H10N8) cross-reacted with the H10 AIVs at higher titers and one monoclonal antibody to H10 influenza virus (JX13-3) failed to react with any of the Bangladeshi H10 viruses (Table 2). We found that the H10N7 and H10N9 AIVs from group 1, as classified by their HA gene  phylogenetic trees, exhibited low cross-reactivity with antisera against group 2 AIVs (e.g., H10N1). These results indicate that the H10 AIVs circulating in LPMs in Bangladesh during the surveillance period were undergoing antigenic drift.
We detected the two DK/BD24035/14 (H10N1) and Dk/BD24268/15 (H10N6) AIVs that caused high mortality in mice and were detected in the heart, liver, and brain tissues. In addition, we detected the DK/BD24035/14 (H10N1) AIV in the spleen at 2 and 4 dpi ( Table 3). In contrast, we detected the DK/BD821/09 (H10N7) and DK/ BD8988/10 (H10N9) AIVs in the heart and the DK/BD821/09 (H10N7) AIV in the liver at 2 and 4 dpi ( Table 3). We did not detect any of the AIVs in the intestines of any of the mice.
Histopathologic analysis of the lungs of the mice infected with the DK/BD8988/10 (H10N9) AIV revealed minimal viral antigens in bronchiolar epithelial cells and none in alveolar cells. Infection with the DK/BD821/09 (H10N7) AIV caused intermittent infection of bronchiolar epithelial cells and minimal spread to alveoli. Both the DK/BD24035/14 (H10N1) and Dk/BD24268/15 (H10N6) AIVs caused diffuse infection of bronchiolar epithelial cells and spread to surrounding alveolar epithelial cells (Fig. 4).

Replication of H10 AIVs in A549 cells.
To determine the replication of H10 AIVs in human epithelial cells, we compared the multicycle growth kinetics of four viruses in the human A549 epithelial cell line. All four viruses grew efficiently in A549 cells; however, the viral titers of DK/BD24035/14 H10N1-infected cells were significantly higher than those of DK/BD8988/10 H10N9-infected cells at 12, 24, and 48 h post infection (P < 0.01 at 12 and 48 h post infection and P < 0.05 at 25 h, Fig. 5). The viral titers of DK/BD24035/14 H10N1-infected cells were also significantly higher than those of DK/BD821/09 H10N7-infected cells at 48 h postinfection (P < 0.01, Fig. 5).
Receptor-binding assay. A switch in receptor-binding preference from avian-type receptors to human-type receptors is important for AIV transmission among humans. We tested the receptor-binding preferences of the H10 AIVs and found that all H10 AIVs had higher binding preferences for α2,3 avian-like receptors than for α2,6 human-like receptors (Fig. 6). Expectedly, we found the control A/Aichi/2/1968 (H3N2) human influenza A virus isolate bound to only the human-like receptor, whereas the avian derived virus A/laughing gull/Delaware Bay/42/2006 (H7N3) bound to only the avian-type receptor (Fig. 6).

Antiviral susceptibility.
To determine whether the Bangladeshi H10 AIVs were sensitive to currently approved antiviral drugs, we examined the susceptibility of these H10 AIVs to the NA inhibitors oseltamivir, zanamivir, and peramivir. All four H10 AIVs that we tested were fully susceptible to all three NA inhibitors, with mean 50% inhibitory concentration (IC 50 ) values ranging from 0.12 to 3.06 nM (Table 4).

Discussion
Fatal human cases of avian-origin H10N8 viral infections have caused increased interest in H10 subtype AIVs 5 . Extensive surveillance studies have identified H10 AIVs in waterfowl and LPMs [26][27][28] . Here, we isolated eight H10 AIVs from LPMs in Bangladesh through active surveillance from 2008 to 2017. Our findings demonstrate that these isolates belong to the Eurasian lineage and replicate efficiently in the respiratory system of mice, without adaptation and with some spread beyond the respiratory tract. This enhanced virulence in mice highlights the need for continued monitoring of AIVs in LPMs, and evaluation of the resultant isolates to identify and understand the pathogenic mechanisms of emerging strains with the potential to infect and cause disease in humans. Our phylogenetic analysis of the H10 HA gene revealed that two different groups of Eurasian H10 AIVs were introduced into Bangladesh. The group 2 AIVs H10N1 and H10N6 were similar to H10N8 AIVs that formed a stable lineage in Jiangxi Province, China 26 , suggesting that H10 AIVs may have reassorted with other subtypes in Bangladesh. Our phylogenetic analysis of the NA gene of the H10N1 AIV revealed branching with the NA genes detected in wild and domestic ducks in China. In contrast, the NA genes of H10N6 AIVs branched with those from AIVs recently detected in the Tanguar haor area of Bangladesh and showed similarities to those of Mongolian AIVs. The NA and HA genes of the H10N7 AIVs were similar to those of H10N7 from Jiangxi, China. This suggests that some viruses were introduced to LPMs and others reassorted with viruses circulating in the LPMs or free-range ducks in Bangladesh. The internal genes of the H10 AIVs were more complex and shared high homology with other subtypes of AIVs. Therefore, epidemiologic monitoring of AIVs in migratory birds is essential to better understand their transmission, reassortment and evolution. Antigenic analysis of HA of the H10 AIVs indicated that they share common epitopes. We detected antigenic differences between the A/chicken/ Jiangxi/34609/2013 (H10N8) AIV and the other H10 AIVs that correlated with substitutions in the HA at residues 137 and 83 at antigenic sites A and E, respectively. Overall, we observed limited antigenic drift among the North American and Eurasian lineages.
Migratory birds are important in the evolution, maintenance, and spread of AIVs. Bangladesh is located in the Central Asian flyway and is near the Eastern Asian-Australian and Black Sea-Mediterranean flyways, and is a stopover site for migratory birds. The interaction between wild birds and domestic and/or free-range ducks, which introduces AIVs into poultry systems, was responsible for emerging new strains that infected humans. This was observed in the emergence of novel H7N9 and H10N8 strains in China in 2013 28,29 . We previously showed that the viruses isolated from Tanguar haor area played a role in the development of a new genotype of H5N1 clade 2.3.2.1a through reassortment 22 . Therefore, further study is needed to address the link between AIVs isolated from free-range ducks in the Tanguar haor area and the diversity and evolution of AIVs circulating in LPMs in Bangladesh.
The mouse model has been widely used to evaluate the virulence of AIVs in mammals. We used DBA/2J mice to evaluate the disease potential of H10 AIVs because DBA/2J mice are more sensitive to influenza A infections than are other strains [30][31][32] . We found that all H10 AIVs infected and replicated in mice without prior adaptation, some of the AIVs caused some degree of weight loss, and some were lethal. Furthermore, these AIVs replicated to higher titers in the lungs than in the nasal turbinates, which is consistent with previous findings showing that H10N8 AIVs isolated from LPMs in China replicated well in the respiratory tracts of mice, with higher viral loads in the lungs than in nasal turbinates 33 . Mouse studies can be useful to investigate the potential of low pathogenic AIVs to replicate and cause disease and also can be used to discover the molecular markers which are important for adaptation to a mammalian host 34 . Recent studies demonstrated that many low pathogenic H1 and H7 AIVs isolated from wild birds in North America caused mortality in mice prior to adaptation 31,35 . Recently, we described the replication and pathogenic potential of H3, H7, and H15 viruses from free-range ducks in Bangladesh using mice. We observed that H7N1 and H7N9 viruses caused 100% and 60% mortality respectively, and expanded their tissue tropism beyond the pulmonary tissues to heart, liver, and brain 36 . Our current study showed this for both H10N1 and H10N6 AIVs as well as high mortality in mice.
The PB2 E627K substitution has been shown to confer increased virulence of AIVs in mammals 37,38 . All of the H10 AIVs we isolated did not contain the PB2 E627K substitution. The E627K substitution is present in viruses obtained from human samples infected with H10N8 AIVs but not in poultry isolates 26,27 , suggesting that this substitution is required to improve replication in mammalian hosts. The NS gene of H10N4 AIV has also been shown to contribute to virulence 39 . In 2014, an H10N7 AIV was isolated from seals in northeastern Europe that caused severe mortality 40,41 . An H10N4 AIV was also detected in mink farms in Sweden 42 . We observed that all H10 AIVs had mutations in their NS1 genes that resulted in a P42S substitution, which is associated with virulence and pathogenicity in mammals 24,25 .  AIV mutations conferring resistance to oseltamivir emerged after the antiviral was used to treat patients with H7N9 and H10N8 AIV infections 26,43 . All of the H10 AIVs we tested in this study were susceptible to NA inhibitors. However, monitoring influenza viruses for the presence of drug-resistance markers is important to predict the emergence of strains with such markers.
The H9N2 AIV is endemic in Bangladesh 21,44-46 and has acquired mammalian host-specific mutations in its internal genes, which have been shown to facilitate transmission from avian species to humans 44 . H9N2 AIVs are significant donors of genetic material to emerging zoonotic viruses such as H5Nx, H7N9, and H10N8 AIVs posing an enormous threat to both human health and poultry industry. The wide circulation of H9N2 AIVs in Bangladeshi LPMs affords H9N2 with more opportunities for reassortment with other AIV subtypes, such as H10 AIVs. Transmission of H10 AIVs to humans has resulted in some fatal cases, and serologic evidence indicates that H10 AIVs were previously transmitted among turkey farmers in the United States 47 . This should raise concern about the potential for human-to-human transmission of H10 AIVs. Evaluation of the biologic properties of H10 and other subtypes of AIVs circulating in LPMs is essential for understanding the emergence and evolution of these viruses and to reduce their potential pandemic threat to public health.

Materials and Methods
Ethics statement. All  Deep amplicon sequencing and genetic analysis. Viral RNA was extracted by using an RNeasy kit (QIAGEN); conventional RT-PCR was then performed by using a SuperScript III first-strand synthesis kit (Invitrogen) with the Uni12 influenza primer. Multiplex PCR of all eight gene segments was conducted by using PCR Supermix HiFi (Invitrogen) with the Uni12/13 primers. PCR products were purified on a spin column (QIAGEN). DNA libraries were prepared by using NEXTera XT DNA-Seq library prep kits (Illumina) with 96 dual-index barcodes, according to manufacturer instructions. Pooled libraries were sequenced with an Illumina MiSeq Personal Genome Sequencer and 150 bp paired-end reads. The closest relatives of the viral genes sequenced of the eight isolated viruses were identified in the Global Initiative on Sharing Avian Influenza Data (GISAID) database and the Influenza Sequence Database (ISD) 49,50 . For phylogenetic analysis, multiple-sequence alignments for each data set were conducted using the MAFFT program 51 . MEGA6 was used to construct phylogenetic trees by applying the neighbor-joining method with the Kimura two-parameter distance model and 1,000 bootstrap replicates 52 . Several virus sequences of the North American lineage were also used as an out group in phylogenetic trees.
Antigenic characterization. The HI assay was used to antigenically characterize the viruses. The four H10 AIVs were tested by using reference antisera against H10 AIVs selected from the North American and Eurasian lineages. The antisera against DK/BD24035/14 (H10N1) was generated for this study. Briefly, ferrets were intranasally infected with 1 mL of 10 6 EID 50 /mL viruses and then boosted after 3 weeks by intramuscular injection of virus with adjuvant. Blood was collected 1 week later for serum isolation.