Evolution and pathogenicity of H6 avian influenza viruses isolated from Southern China during 2011 to 2017 in mice and chickens

H6 subtype avian influenza viruses spread widely in birds and pose potential threats to poultry and mammals, even to human beings. In this study, the evolution and pathogenicity of H6 AIVs isolated in live poultry markets from 2011 to 2017 were investigated. These H6 isolates were reassortant with other subtypes of influenza virus with increasing genomic diversity. However, no predominant genotype was found during this period. All of the H6N2 and most of the H6N6 isolates replicated efficiently in lungs of inoculated mice without prior adaptation. All of the H6N2 and two H6N6 isolates replicated efficiently in nasal turbinates of inoculated mice, which suggested the H6N2 viruses were more adaptive to the upper respiratory tract of mice than the H6N6 viruses. One of H6N2 virus caused systemic infection in one out of three inoculated mice, which indicated that H6 avian influenza virus, especially the H6N2 viruses posed a potential threat to mammals. Five H6 strains selected from different genotypes caused no clinical signs to inoculated chickens, and their replication were limited in chickens since the viruses have been detected only from a few tissues or swabs at low titers. Our study strongly suggests that the H6 avian influenza virus isolated from live poultry markets pose potential threat to mammals.

Studies with chickens. The pathogenicity of 5 H6 viruses from different genotypes, was evaluated in SFP chickens. Each of six 4-week-old SPF chickens were inoculated i.n. with 10 6.0 EID 50 /bird of each virus in a volume of 100.0 μl, respectively. One day later, three naïve chickens were introduced into each of these isolates to evaluate the transmission of viruses among chickens. Six control birds were inoculated i.n. with the same amount of PBS. Three inoculated chickens in each group were euthanized, and their tracheas, lungs, kidneys, spleens and duodenums were collected for virus isolation at 3 dpi. Cloacal and oropharyngeal swabs were collected from the other 3 chickens at 2, 4 and 6 dpi. All of the tissues were homogenized in PBS (0.1 g per 1.0 ml), and the swabs in 1.0 ml PBS were stirred on vortex, and then all samples were centrifuged at 2500 g for 10 min. Each supernatant was collected, filtered through 0.22 μm filters, and inoculated into allantoic cavities of three 9-day-old embryonated chicken eggs. At 14dpi, blood samples were drawn from the remaining inoculated chickens and three contact chickens for detection of H6 subtype specific antibodies.

Ethics statement and statistical analysis.
All animal studies in this study were conducted in accordance to the guidelines of the Animal Care and Use Committee of Shanghai Veterinary Research Institute, and all animal studies protocols are approved by Shanghai Veterinary Research Institute (Permit number: SHVRI-Po-0120). Good living environment with sufficient food and water were available for all the animals. Comparisons of the weight changes between two groups were determined by nonparametric t-tests using Graph-Pad (Vision 6.0, GraphPad Prism). The differences were considered statistically significant if P values < 0.05(*), P < 0.01(**), P < 0.001(***).

Results
Virus isolation and identification. A total of 22 H6 AIVs were isolated from ducks and geese without any obvious clinical signs in LPMs during 2010-2017. All of viruses were propagated in SPF chicken embryonated eggs and their virus titers ranged from 10 6.3 to 10 9.5 EID 50 /100 μl. Full genomes of 7 H6N2 and 15 H6N6 viruses were sequenced and deposited to the NCBI database, the accession number and abbreviations of viruses used in this study have been listed in Table 1.

Scientific Reports
| (2020) 10:20583 | https://doi.org/10.1038/s41598-020-76541-0 www.nature.com/scientificreports/ Molecular characterization. According to sequence analysis, all of the H6 isolates shared the same amino acids (PQIETR/GLF) with single basic amino acid at the cleavage site between HA1 and HA2, which displayed a low pathogenic feature 32 . The receptor-binding sites in HA protein possessed the amino acid residues Q226 and G228 (H3 HA numbering 33 , which is used throughout the manuscript), suggesting that all H6 isolates would have a higher affinity to α-2,3-linked sialic acid which is predominant in the upper respiratory tract of avian species 34 . Several mutations were found in the receptor-binding area in the HA proteins and listed in Table 2, including the amino acids from 169 to 171 which related to a potential glycosylation site, and the amino acid at 190 which was found to determine the binding capability of H9N2 viruses to lung epithelial cells of mouse and human 35 . The amino acid changes of A138S and P186T/I were appeared in ZJ/B1994/H6N6, FJ/D3480/ H6N6and JS/F336/H6N6. Those 22 viruses contained 119E,274H and 294 N on the NA proteins (N2 numbering 36 ), which suggested that all viruses are sensitive to oseltamivir 37 . There were 11 amino acids deletion in the stalk region of three H6N6 isolates (ZJ/B1994/H6N6, FJ/D3480/H6N6 and JS/F336/H6N6), which might be associated with increased virulence in mammals 38 . It is notable that S31N substitutions in the M2 protein, which are associated with amantadine resistance of influenza virus were observed in GD/E3503/H6N2 and GD/F3891/H6N2 viruses 39 . Amino acid residues were E and D at positions 627 and 701 of the PB2 protein in all those H6 isolates, which have been associated with the pathogenicity of avian influenza viruses in mammals 40 . Previous reports showed www.nature.com/scientificreports/ that the substitution D92E in the NS1 protein associated with reducing its phosphorylation and increasing the virus resistance to interferon 41 , but this substitution was not detected in any of the isolates. In addition, no amino acid changes associated with increased virulence in mammals were detected in the PA or PB1 proteins.
Phylogenetic analysis of HA and NA genes. The nucleotide and amino acid similarity of the HA from 22 isolates in this study were 82.0% to 99.9% and 84.0% to 99.8%, respectively, as shown in Table 3. The phylogenetic tree of HA gene ( Fig. 1a) indicated that all of the H6 isolates were divided into two clades under Eurasian lineage. All the H6N2 isolates were clustered in A/duck/

Phylogenetic analysis of the internal genes.
Phylogenetic analysis of the six internal genes showed that all those 22 H6 AIVs clustered in the Eurasian lineage. The PB1, PB2, PA and NS genes were divided into two clades, and the NP and M genes were fell into 3 clades ( Fig. S1-6). Phylogenetic tree of PB1 genes showed that nine isolates were clustered in the ST339-like clade, in which the viruses were isolated from different birds including chicken, duck, goose and wild bird. The other thirteen isolates were closely related to A/duck/Guangdong/S3180/2010(H6N6) (GD/S3180) (Fig. S1). In the phylogenetic trees of the PB2 and NS genes, only GD/ F1473/H6N2 located in a separated clade ( Fig. S2 and S6). PB2 gene of GD/F1473/H6N2 was related closely to that of the early H5N1 AIV isolate A/goose/Guangdong/1/96 in the BJ/BJ/1/94-likeclade. The NS gene of GD/ F1473/H6N2 was closely related to that of H3N2, H3N8, and H6N2 AIVs isolated from different waterfowls. As for the PA genes, all of the H6N6 viruses and one H6N2 virus GD/F1473/H6N2 were clustered in the ST339like clade (Fig. S3). The other 5H6N2 viruses were located in the Gs/GD196-like clade. The NP gene of the virus JS/E1201/H6N6 was clustered in a separated A/duck/Hunan/573/2002(H6N2)-like (HN/573-like) clade, while other isolates were divided into ST339-likeand A/duck/Mongolia/54/2001(H5N2)-like (Mongolia/54-like) virus groups (Fig. S4). On the NP tree, most of the H6N6 and GD/F1473/H6N2were related to a H5N2 virus A/duck/ Mongolia/54/2001. In the ST339-like clade, the NP genes of 6H6N2 isolates clustered into a A/chicken/Guang-  According to the phylogenetic analysis of the whole genomes of the AIVs isolated from 2011 to 2017, the genotypes of H6 viruses were classified. Based on the HA genes, the viruses were divided into A and B genotypes corresponding to ST/339-like and ST/2853-like clades. H6N2 and H6N6 viruses were divided into A2, B2, A6 and B6 accordingly. Based on the different clades of NA and other 6 internal genes, the genotypes were named according to the chronological order of virus isolation in this study (Fig. 2). The genetic reassortment of H6 AIVs occurred frequently from 2011 to 2017, and resulted in multiple genotypes. The genotype of B602 was detected in 2012, 2015 and 2017, suggesting that the viruses of this genotype have elevated adaptation and replication efficiency in their natural reservoirs.

Pathogenicity in mice.
To evaluate the pathogenicity in mice, groups of white mice were inoculated i.n.
with each H6 isolates (10 6.0 EID 50 /mice). All of the mice used for observation survived till 14 dpi. None of H6N2 viruses, except GD/1268/H6N2, caused significant body weight loss (P < 0.05) in mice compared with the control group (Fig. 3). As for H6N6 viruses, only HN/A729/H6N6 and ZJ/B1994/H6N6 caused significant growth retard in mice, while the clinical signs caused by other viruses were not obvious.
All of the H6H2 isolates were able to replicate efficiently in mouse lung and nasal turbinate without prior adaptation. The mean virus titers were ranged from 10 3.4 to 10 6.0 EID 50 /ml in the lungs and from 10 1.8 to10 4.8 EID 50 /ml in the nasal turbinates (Table 4). Six out of seven H6N2 isolates were detectable in the heart of partial or all the three inoculated mice. Specially, two H6N2 viruses were found in the spleen of one out of three inoculated mice, and one of those viruses GD/E3503/H6N2 caused systemic infection in one inoculated mouse. In addition, the virus was recovered from liver and brain. The GD/1268/H6N2 and GD/1127/H6N2 viruses of A201 genotype were detectable in the lungs, nasal turbinates and hearts consistently, while the 4 isolates of A202 genotype showed diversity in the tissue tropisms ( Table 4).
The H6N6 viruses, except ZJ/B2039/H6N6 and JS/G91/H6N6, replicated in the lungs of inoculated mice. ZJ/B2039/H6N6 has not been detected in any of the other tested tissues, and lower titers of JS/G91/H6N6 were detected only in the nasal turbinate of one out of three mice (Table 4), suggesting the limited replication of those 2 viruses in mice. In addition, the replication of GD/E3798/H6N6 was also limited in mice and lower virus titers were detected in the lungs and nasal turbinates from two out of three mice. Besides those 3 viruses, the other H6N6 isolates replicated well in the lungs and the virus titers ranged from 10 2.1 to 10 6.3 EID 50 /ml. Specially, the average virus titers of 5 isolates reached a high level of over 10 5.0 EID 50 /ml, which suggested those viruses replicated efficiently in the lungs of inoculated mice. In the nasal turbinates, only GD/E3415/H6N6 and FJ/ D3480/H6N6 could be detected in all of the three mice, and the average virus titers were 10 1.1 and 10 2.5 EID 50 /ml, respectively. Except ZJ/B2039/H6N6 and ZJ/B2044/H6N6 which were not recovered from the nasal turbinates  and N6 NA (c) genes were generated by using the neighbor-joining method using MEGA 6.0. The genomic sequences of the viruses listed in black were downloaded from available databases; the viruses listed in red were evaluated in this study. The scale bar represents the distance between sequence pairs, and horizontal distances are proportional to genetic distance.

Pathogenicity and transmission in chickens.
To evaluate the virulence of the H6 isolates in chickens, groups of six 4-week-old SPF chickens were inoculated intranasally with 10 6.0 EID 50 of each H6 isolate. No clinical sign was observed in the inoculated chickens. We further examined the replication of the H6 viruses by detecting the viruses in different tissue and swabs of inoculated chickens. The GD/1127/H6N2 and GD/E3503/ H6N2 viruses were detected in the trachea of one inoculated chicken, respectively. FJ/D3480/H6N6 was detected in the duodenum of one inoculated chicken, while HN/A729/H6N6 and ZJ/B2028/H6N6 were not detected in any tested tissues of 3 inoculated chickens at 3dpi. Only GD/E3503/H6N2 was detected in the oropharyngeal swab from one chicken at 4 dpi. GD/E3503/H6N2 and HN/A729/H6N6 were detected in the cloacal swab from one chicken at 2 dpi and 6 dpi, respectively. None of 3 chickens inoculated with HN/A729/H6N6 showed seropositive to specific H6 antibody, and only some chickens inoculated with the other four viruses were seroconverted. One contact chicken in the GD/E3503/H6N2 group was seroconverted at 14 dpi, while no specific H6 antibody was detected in other contact chickens ( Table 5). The results suggested the replication and transmission of H6 viruses with different genotypes were limited in chickens.

Discussion
Since the late 1990s, H6 AIVs have been circulating in Southern China. The H6 viruses have been become more prevalent over time with an increasing isolation rate of H6N1, H6N2 and H6N6 subtypes 6,15,42 . In this study, we analyzed the molecular evolution and pathogenicity in mice and chickens of H6 AIVs isolated from the LPMs in Southern China from 2011 to 2017, and provided a glimpse of the genetic diversity and pathogenicity of H6 AIVs in mammals and chickens. Multiple H6 virus genotypes emerged from 2011 to 2017 resulted from the internal gene reassortment between H6 and other subtype viruses. It has been noted that most of the H6N1/N2 viruses isolated from Southern China from 2000 to 2005 were clustered with the G1-like, W312-like or H9N2 Ck/Bei-like lineage based on the PB2 and PB1 genes, and the PA and NP genes of most H6 viruses clustered with H5 and H9 viruses, M and NS genes with G1-like or H9N2 Ck/Bei-like lineage viruses 16,22 . Previous studies showed that the H6 isolate from ducks replicated poorly in chicken trachea and that the viruses recovered were W312-like viruses 43,44 . However, in our study, none of the internal genes were related closely to G1-like or W312-like viruses. These results suggested that the H6 subtype viruses in Southern China were genetically diverse, which might increase the potential for H6 viruses to transmit from ducks to other animals. www.nature.com/scientificreports/ Molecular analyses suggested that all of the 22 H6 isolates were low pathogenic AIVs. The receptor-binding sites in the viral HA proteins possess the residues Q226 and G228, similar to those H6 isolates reported previously, which preferentially bind to the α-2,3-linked sialic acid receptors predominant in avian species 45,46 . However, the E190V and N192D substitutions of H6 HA have been associated with interspecies transmission of AIVs from ducks to chickens 47 , but those amino acids were not found in the five H6 AIVs (GD/1127/H6N2, GD/E3503/H6N2, HN/A729/H6N6, ZJ/B2028/H6N6 and FJ/D3480/H6N6) in this study, which would be the most likely reason that the replication of the H6 isolates in chickens was surprisingly limited. The residues 228S, 137N, 186L, A13S, and A193N at HA associated with human receptor-binding preference 45,47 were not found in H6 isolates in this study. Additionally, almost all H6N6 isolates have the substitution of HA at V187D, the binding affinity might be altered to adapt to mammalian receptors 19 .
None of E119V, H275Y, R293K and N295S substitution in the NA were found, which suggested those H6 isolates are sensitive to neuraminidase inhibitors such as oseltamivir. The deletion of 11 amino acids in the stalk region of the NA of FJ/D3480/H6N6, ZJ/B1994/H6N6and JS/F336/H6N6, which was also found in a swine H6N6 virus, might be associated with the infectivity of H6N6 viruses in mammals through affecting NA activity 48 and the balance between HA and NA 49 . It is notable that the S31N substitution in the M2 protein, which is associated with amantadine resistance of influenza virus 50 , was found in GD/E3503/H6N2 and GD/F3891/H6N2. Some of the H6 viruses replicated efficiently on MDCK and A549 cells and in the lungs of mice 17,51 . The direct contact transmission of H6 viruses in guinea pigs was confirmed previously, which suggested the H6 viruses pose a clear threat to mammals 17 . We found that all of the H6H2 isolates were able to replicate efficiently in lung and nasal turbinate without prior adaptation in mice, but the replication ability of H6N6 varies. Even with the same genotype, different H6N6 viruses showed distinct replication abilities in mice, suggesting some amino acid mutations affected the replication of the viruses. Lower titers of some H6 viruses were detected in livers, spleens and brains in a part of inoculated mice, which suggested those viruses posed a potential to adapt and caused a systemic infection in mammals.
Five H6 AIVs selected from different genotypes caused no clinical signs in any of the inoculated chickens. The replication and transmission of H6N2 and H6N6 viruses were limited in SFP chicken in this study, although the pathological changes caused by H6N2 in chickens were reported in California 13 . Previous study showed that H6N2 isolates caused seroconversion in infected chickens, but no virus was recovered from the tissue samples, which suggested that the H6N2 strains replicated poorly and were nonpathogenic to chickens 52 . Our findings showed that the five H6 strains selected from different genotypes caused no clinical signs in any of the inoculated chickens, but the serological test results suggested that chickens in four groups have been infected despite limited recovery of inoculated viruses from the tissue samples. The replication and transmission of H6N2 and H6N6 viruses were limited in SFP chicken in this study.
Overall, our study suggested that the H6 AIVs circulating in South China are genetically diverse and pose potential threat to mammals, and continual surveillance of H6 viruses is necessary in China.
Received: 3 April 2020; Accepted: 8 October 2020 Table 5. Replication of the H6 viruses in chickens. a The number of virus positive samples/total. b The number of antibody positive samples/total, the HI titers ≥ 4 were considered as antibody positive samples.