Continuous Reassortment of Clade 2.3.4.4 H5N6 Highly Pathogenetic Avian Influenza Viruses Demonstrating High Risk to Public Health

Since it firstly emerged in China in 2013, clade 2.3.4.4 H5N6 highly pathogenic avian influenza viruses (HPAIVs) has rapidly replaced predominant H5N1 to become the dominant H5 subtype in China, especially in ducks. Not only endemic in China, it also crossed the geographical barrier and emerged in South Korea, Japan, and Europe. Here, we analyzed the genetic properties of the clade 2.3.4.4 H5N6 HPAIVs with full genome sequences available online together with our own isolates. Phylogenetic analysis showed that clade 2.3.4.4 H5N6 HPAIVs continuously reassorted with local H5, H6, and H7N9/H9N2. Species analysis reveals that aquatic poultry and migratory birds became the dominant hosts of H5N6. Adaption to aquatic poultry might help clade 2.3.4.4 H5N6 better adapt to migratory birds, thus enabling it to become endemic in China. Besides, migratory birds might help clade 2.3.4.4 H5N6 transmit all over the world. Clade 2.3.4.4 H5N6 HPAIVs also showed a preference for α2,6-SA receptors when compared to other avian origin influenza viruses. Experiments in vitro and in vivo revealed that clade 2.3.4.4 H5N6 HPAIVs exhibited high replication efficiency in both avian and mammal cells, and it also showed high pathogenicity in both mice and chickens, demonstrating high risk to public health. Considering all the factors together, adaption to aquatic poultry and migratory birds helps clade 2.3.4.4 H5N6 overcome the geographical isolation, and it has potential to be the next influenza pandemic in the world, making it worthy of our attention.


Phylogenetic Analyses
All the H5N6 of full genome sequences until 2019 were downloaded from GISAID (Global Initiative on Sharing All Influenza Data), GenBank (National Center for Biotechnology Information), and IRD (Influenza Research Database), and 974 strains were collected totally. Phylogenetic analysis work was performed on downloaded data, and full-genome sequences of isolated viruses were used. Multiple sequence alignments were performed using Muscle [45]. The maximum likelihood phylogenetic tree was constructed by IQ-TREE software (http://www.iqtree.org/) with the GTRGAMMA model [46]. Ultrafast bootstrap was also implemented, and 1000 replications were run [47].

Receptor Binding Specificity Assay
The receptor binding specificity was analyzed with a solid-phase binding assay. Briefly, plates were coated with serial dilutions of α2,3 glycans (Neu5Aca2-3Galb1-4GlcNAcb-PAA-biotin, 3'SLN, Glycotch) and α2,6 glycans (Neu5Aca2-6Galb1-4GlcNAcb1-PAA-Biotin, 6'SLN, Glycotch) overnight at 4 • C. Then, the glycan solution was removed, and the plates were blocked, washed, and incubated with a solution containing 26 hemagglutination units of influenza virus. The plates were incubated at 4 • C for 12 h. After being washed again, the plates were incubated with mouse monoclonal antibody against NP (Sino Biological). The plates were then washed again and incubated with horseradish peroxidase (HRP)-conjugated goat anti-mouse antibody. Finally, the plates were washed and incubated with 3,3',5,5'-Tetramethylbenzidine (TMB) for 15 min at room temperature. The reaction was stopped with 0.1 M HCl, and the absorbance was read at 450 nM.

Mice and Chickens Challenge Studies
Four-week-old female specific pathogen-free (SPF) BALB/c mice were divided into four groups randomly, 12 mice per group. Mice were inoculated intranasally with 10 6 50% egg infectious doses (EID 50 ) of test viruses or sterile DMEM in a volume of 50 µL. Mice were monitored daily for weight loss over a period of 14 days. Three mice from each group were euthanized at 3 days post-infection (dpi), and lung and brain samples were collected for virus titration. The mice whose weight loss was more than 25% were also humanely euthanized. The mice were euthanized by CO 2 .
A total of 27 six-week-old SPF chickens were divided into three groups, nine chickens per group, and then were inoculated intranasally with 10 6 EID 50 of test viruses in a volume of 200 µL. Non-inoculated hatch-mates were added to each group at 1 day post-infection (dpi, contacts). Six contact chickens for 673 and 674 group and nine contact chickens for 39715 group. Clinical signs were monitored daily until 14 dpi. At 2, 3, and 5 dpi, both cloacal and throat swabs were collected. Chickens were humanely euthanized by cervical dislocation at 2, 3 and 5 dpi. The organs including heart, liver, spleen, kidney, brain, and lung of the chickens succumbing to viral infection were also collected. All the swabs and organs were titrated by embryo eggs. The same actions were also performed on naive contact groups.
The virus titer of mice and chickens were tittered in embryonated eggs. The hemagglutination activity assay was performed by 1% chicken red blood cells.

Phylogenetic Analysis of Surface Genes of H5N6 Viruses
To make a better understand of the establishment and evolution of H5N6, we performed a phylogenetic analysis of all the H5N6 genomes available from GISAID, GenBank, and IRD, together with H5N6 viruses isolated by our lab. The phylogeny analysis result of HA showed that nearly all the H5N6 belonged to clade 2.3.4.4, with only a few exceptions (Figure 1, Figure S1). The HA of clade 2.3.4.4 H5N6 could be divided into three groups: Sichuan-Like (SC-Like), Guangdong-Like (GD-Like), and Worldwide clade 2.3.4.4 H5N2/H5N8-Like (World-H5-Like) ( Figure 1). The SC-Like and GD-Like showed nucleotide similarity to human-origin isolates A/Sichuan/26221/2014(H5N6) and A/Guangzhou/39715/2014(H5N6), respectively. The World-H5-Like showed higher identity with the H5N2 and H5N8 HPAIVs, which spread globally. GD-Like viruses took a major part of all the clade 2.3.4.4 H5N6 HPAIVs.
The phylogeny of NA revealed that the NA of clade 2.3.4.4 H5N6 HPAIVs could be divided into three lineages: Sichuan-Like (SC-Like), Guangdong-Like (GD-Like), and H4N6-Like ( Figure 1, Figure S2). The NA segment of both SC-Like and GD-like viruses exhibited similarity to H6N6 circulating in domestic poultry of China; however, the NA of H4N6-Like viruses is similar to the H4N6 of migratory birds. All the NA genes of SC-Like and GD-Like viruses, according to nucleotide similarity to human-origin isolates A/Sichuan/26221/2014(H5N6) and A/Guangzhou/39715/2014(H5N6), respectively, belonged to ST-192-Like, which was the major lineage of H6N6 circulating among aquatic poultry in China [16]. Furthermore, nearly all the GD-Like NAs had a stalk deletion in the stalk region, with only few exceptions, while the SC-Like and H4N6-Like viruses all harbored a full-length NA without a stalk deletion.  Further analysis of internal genes revealed that the H7N9/H9N2 origin genes showed a district characteristic: gene segments from the same province shared a higher nucleotide identity (Supplementary Figures S3-S8) [8]. This phenomenon means that H5N6 might keep reassorting with local chicken H7N9/H9N2. According to Huang's work on H6 in China, the H6 lineage PB2 belonged to Group I/II of China domestic H6 PB2 [16]. The Japan-Korea lineage PB1 and PA fell into the gene pool of China domestic H6. Furthermore, both Group I/II and gene pool H6 provided genes for H6 circulating in aquatic poultry [16]. These segments might make H5N6 more adapted to aquatic poultry and migratory birds. Additionally, while circulating all over the world, H5N6 also reassorted with the worldwide H5N2/H5N8 carried by migratory birds and local poultry. This was why H5N6 also harbored segments originating from H5N2/H5N8. We also found that the internal genes of H5N6 exhibited a great diversity, and this might help H5N6 cross the inter-species barrier, suggesting its potential threat to world health.

Continuously Reassortment of Clade 2.3.4.4 H5N6 HPAIVs
We found an interesting phenomenon that there were particular HA:NA combinations of A previous study by our group showed that the GD-Like:GD-Like H5N6 exhibited enhanced pathogenicity and transmissibility in chickens as compared to other H5Nx subtypes and H5N6 of other HA-NA combinations [51,52]. It was revealed that the appropriate match of HA and NA promoted the GD-Like:GD-Like combination and became the dominant H5N6 genotype in China. Over time, only the GD-Like:GD-Like combination and World-H5-Like:H4N6-Like combination were left. These combinations might enhance H5N6 replication and transmission. A previous study by our group showed that the GD-Like:GD-Like H5N6 exhibited enhanced pathogenicity and transmissibility in chickens as compared to other H5Nx subtypes and H5N6 of other HA-NA combinations [51,52]. It was revealed that the appropriate match of HA and NA promoted the GD-Like:GD-Like combination and became the dominant H5N6 genotype in China. Over time, only the GD-Like:GD-Like combination and World-H5-Like:H4N6-Like combination were left. These combinations might enhance H5N6 replication and transmission.

Host, Time, and Region Distribution of Clade 2.3.4.4 H5N6 HPAIVs
The conclusion of our summary work shows that the primary hosts of H5N6 were aquatic poultry (ducks and geese) and migratory birds, even though it has a wide host range (Figures 1 and 2). It is possible that the adaption to aquatic poultry and migratory birds of H5N6 was the principal reason why the virus could move over long distances and spread globally. The gene composition of H5N6, especially the internal genes, became more and more complex as time went on ( Figure 2). This revealed the continuous reassortment between H5N6 and other subtype viruses and reassortment between different H5N6 viruses. Additionally, there was also a clear spatio-temporal correlation. First, H5N6 emerged in China and South Asia in 2013. Then it transmitted to South Korea and Japan in 2016. In 2017, H5N6 isolates were reported in Europe. The time clue of evolution also showed that clade 2.3.4.4 H5N6 spread continuously and unrestrictedly ( Figure 2). The hosts of H5N6 also exhibited a diversity. Except domestic poultry and migratory birds, swine, cats, and humans were also found to be infected with H5N6 ( Figure 3). The gene composition of viruses isolated from swine, cats, and humans also showed a multiplicity of sources ( Figure 3). It reminds us that clade 2.3.4.4 H5N6 HPAIVs has a great potential to infect humans.

Host, Time, and Region Distribution of Clade 2.3.4.4 H5N6 HPAIVs
The conclusion of our summary work shows that the primary hosts of H5N6 were aquatic poultry (ducks and geese) and migratory birds, even though it has a wide host range (Figures 1 and  2). It is possible that the adaption to aquatic poultry and migratory birds of H5N6 was the principal reason why the virus could move over long distances and spread globally. The gene composition of H5N6, especially the internal genes, became more and more complex as time went on ( Figure 2). This revealed the continuous reassortment between H5N6 and other subtype viruses and reassortment between different H5N6 viruses. Additionally, there was also a clear spatio-temporal correlation. First, H5N6 emerged in China and South Asia in 2013. Then it transmitted to South Korea and Japan in 2016. In 2017, H5N6 isolates were reported in Europe. The time clue of evolution also showed that clade 2.3.4.4 H5N6 spread continuously and unrestrictedly ( Figure 2). The hosts of H5N6 also exhibited a diversity. Except domestic poultry and migratory birds, swine, cats, and humans were also found to be infected with H5N6 ( Figure 3). The gene composition of viruses isolated from swine, cats, and humans also showed a multiplicity of sources ( Figure 3). It reminds us that clade 2.3.4.4 H5N6 HPAIVs has a great potential to infect humans.     (178) were used as controls. All the H5N6 viruses exhibited a preferential binding to SA2,3Gal receptors ( Figure 4a). However, the human isolate 39715 exhibited a higher binding ability to SA2,6Gal receptors compared to clade 2.3.2 H5N1 virus.
HA of influenza virus was considered to be related with the host range, pathogenicity, and transmissibility in avian and mammalian species.

Replication of Clade 2.3.4.4 H5N6 in Different Cells
To assess the replication ability of clade 2.3.4.4 H5N6 in different cells, we performed a multistep growth curve of several viruses isolated from different hosts in DF-1, ST, and A549 cells. All the viruses replicated efficiently in these three kinds of cells, which could reach a high titer of >7 lg TCID50/mL (Figure 4b

Replication of Clade 2.3.4.4 H5N6 in Different Cells
To assess the replication ability of clade 2.3.4.4 H5N6 in different cells, we performed a multistep growth curve of several viruses isolated from different hosts in DF-1, ST, and A549 cells. All the viruses replicated efficiently in these three kinds of cells, which could reach a high titer of >7 lg TCID50/mL (  All the selected viruses were lethal to mice, and 673 and 39715 could cause a lethality of 100% (Figure 5a,b). All three viruses were capable of replication in the lungs of inoculated mice and could reach as high as 7.25 lg EID 50 /mL (Figure 5c). Additionally, 673 and 39715 could replicate in the brains of inoculated mice; however, 674 was replication-deficient in this organ (Figure 5d). All the selected viruses were lethal to mice, and 673 and 39715 could cause a lethality of 100% (Figure 5a,b). All three viruses were capable of replication in the lungs of inoculated mice and could reach as high as 7.25 lg EID50/mL (Figure 5c). Additionally, 673 and 39715 could replicate in the brains of inoculated mice; however, 674 was replication-deficient in this organ (Figure 5d). This revealed that clade 2.3.4.4 H5N6 can replicate in mammals and exhibits a lethal risk. In fact, 673 could reach a lethality of 100% in mice, while 674 showed a lethality of less than 60%. H5N6 viruses like 673 (SC-HA:SC-NA) cause sudden death to mammalian hosts, diminishing the transmission ability to some extent. However, viruses similar to 674 (GD-HA:GD-NA) are not 100% lethal to mammalian hosts and can also replicate in the lungs of infected animals. Viruses similar to 674 could replicate in the respiratory organs of infected mice but show mild pathogenicity to mice. Additionally, these could achieve some mammalian adaption mutation in the internal genes. The gene composition of 39715 is similar to 674, but 39715 exhibited a high pathogenicity to mice. Mammalian adaption mutations such as E627K or D701N are also found in human isolates. One of reasons why 39715 exhibited a high pathogenicity to mice is that 39715 has the E627K mutation in PB2 (Table S2). This might be one of the reasons why nearly all the human isolates of H5N6 harbor a GD-Like HA and GD-Like NA like 674. This revealed that clade 2.3.4.4 H5N6 can replicate in mammals and exhibits a lethal risk. In fact, 673 could reach a lethality of 100% in mice, while 674 showed a lethality of less than 60%. H5N6 viruses like 673 (SC-HA:SC-NA) cause sudden death to mammalian hosts, diminishing the transmission ability to some extent. However, viruses similar to 674 (GD-HA:GD-NA) are not 100% lethal to mammalian hosts and can also replicate in the lungs of infected animals. Viruses similar to 674 could replicate in the respiratory organs of infected mice but show mild pathogenicity to mice. Additionally, these could achieve some mammalian adaption mutation in the internal genes. The gene composition of 39715 is similar to 674, but 39715 exhibited a high pathogenicity to mice. Mammalian adaption mutations such as E627K or D701N are also found in human isolates. One of reasons why 39715 exhibited a high pathogenicity to mice is that 39715 has the E627K mutation in PB2 (Table S2). This might be one of the reasons why nearly all the human isolates of H5N6 harbor a GD-Like HA and GD-Like NA like 674.

Pathogenicity of Clade 2.3.4.4 H5N6 in Chickens
To investigate the virulence and transmission ability of these viruses in chickens, SPF chickens were inoculated intranasally with 10 6 EID 50 of selected viruses in a volume of 200 µL.
All the selected viruses caused a lethality of 100% to inoculated chickens in three days and were capable of replication in all the organs, including lung, heart, liver, spleen, kidney, and brain ( Table 1). The virus titers in hearts, livers, spleens, kidneys, brains, and lungs were quite high; for example the virus titer in lung could reach > 6 lg EID 50 /mL, and even the virus titer in brain could reach > 5 lg EID 50 /mL. Even though clade 2.3.4.4 H5N6 exhibited a high virulence to chickens, the transmission ability was quite different (Table 2). Contact groups 674 and 39715 shed viruses through throat and cloaca, but 673 contact group only shed viruses at 2 dpi through cloaca. This phenomenon means the transmission ability of 674 and 39715 is stronger than that of 673 in chickens. This should explain why GD-Like HA and GD-Like NA matches became the dominant matches of clade 2.3.4.4 H5N6.

Discussion
In recent years, HPAIVs H5Nx such as H5N2, H5N8, and H5N6 have emerged and replaced predominant H5N1 and became the dominant subtypes. The newly emerged HPAIVs clade 2.3.4.4 H5N6 spread rapidly in South and Central China, replacing H5N1 to become the dominant H5 subtype in China. It is notable that South China and Central China shared the highest aquatic poultry feeding density in China. The human infection cases also took place in these areas. No isolates were reported in other inland provinces with limited waterflow production. Therefore, we hypothesize that aquatic poultry helped H5N6 spread and became endemic in China.
Compared to predominant H5N1 and the worldwide H5N2/H5N8, the newly emerged H5N6 has a more complex and more diverse gene segment composition. Clade 2.3.4.4 HA could be divided into three groups. The H6N6 origin N6 also fell into two groups. During the study of surface genes of H5N6, we found that the match of GD-Like HA with GD-Like NA was significant in China and became the dominant surface gene composition over time. The internal genes of H5N6 were more complex when compared to the surface genes. Further phylogenetic analysis of the internal gene of clade 2.3.4.4 H5N6 viruses in China revealed that the internal genes were similar to H5, H6, and H7N9/H9N2 viruses ( Figures S2-S8), which were the three most common AIVs circulating in the poultry of China. The H5 origin genes all fell into clade 2.3.4, which was a dominant clade in China [53]. Previous research also revealed that clade 2.3.4 AIV could induce severe inflammatory responses in human immune cells compared to clade 2.3.2 and clade 7.2, two other dominant clades in China [54]. Therefore, H5N6 might present a higher risk for public health. H7N9/H9N2 origin internal genes may stem from local poultry because we also discovered an obvious regional signature of these genes. For example, isolates from the same province showed a higher identity (Figures S3-S8); however, H5 origin and H6 origin genes did not have this phenomenon. The H6 origin internal genes were found to be provided by the aquatic poultry of South China. The H5N6 spread to South Korea and Japan all had H6 origin segments. We do believe that H6 origin segments helped H5N6 adapt better to aquatic poultry and migrant birds, allowing it to cross the ocean barrier and spread to South Korea and Japan. As previous findings revealed, HPAIVs H5N2 and H5N8 were spread all over North America, and HPAIV H5N8 circulated in Europe. H5N6 also has the potential to spread globally. Even though the European H5N6 isolates harbored surface genes different from the predominant H5N6 of China, the internal genes shared some identical segments with the H5N6 of China. Compared to the local H5N2/H5N8, the European isolates harbored H7N9/H9N2 lineage origin PB2 and Japan-Korea lineage origin PA. The internal genes of H9N2 were considered as a key factor for the circulation of H7N9 in China [21], and they also played a role in the recent emergence of human H10N8 infections [22]. Up to now, there are no reports about worldwide human infection cases of H5N2/H5N8. However, the European H5N6 harboring H7N9/H9N2 origin segments exhibits a potential threat to human health. The Japan-Korea origin segment might enhance the transmission ability of viruses in migratory birds and aquatic poultry. This highlights the potential of H5N6 to be spread globally and even be a world pandemic. Clade 2.3.4.4 H5N6 HPAIV is capable of replication in avian, swine, and human cells, and it is lethal to both mice and chickens [53,54]. It therefore exhibits a high risk to public health.
Furthermore, aquatic poultry was thought to be a natural reservoir of AIVs. However, not enough attention has been paid to it. Our team insists that the precautions taken regarding AIVs should be moved forward to the surveillance of aquatic poultry. We have invented the H5 vaccines particular for aquatic poultry, D7 and rD8, which have been approved by the Ministry of Agriculture and Rural Affairs of China. More attention should be paid to the AIVs circulating in aquatic poultry.
All the findings combined with our results reveal that clade 2.3.4.4 H5N6 HPAIV shows a preference for migratory birds and aquatic poultry and exhibits great genetic diversity. With the adaption to migratory birds and aquatic poultry, H5N6 has been transported over long distances. During the long-distance transmission of H5N6, it also kept continuously reassorting with local AIVs. Taking all the findings into consideration, H5N6 shows the potential for global spread and poses a great threat to the poultry industry and human health.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2076-0817/9/8/670/s1, Figure S1: Phylogenetic tree of HA, Figure S2: Phylogenetic tree of NA, Figure S3: Phylogenetic tree of PB2, Figure S4: Phylogenetic tree of PB1, Figure S5: Phylogenetic tree of PA, Figure S6: Phylogenetic tree of NP, Figure S7: Phylogenetic tree of M, Figure S8: Phylogenetic tree of NS, Table S1: Basic information of viruses used in this study, Table S2: Key molecular markers of viruses used in this study.