Characteristics of Avian Inuenza H9N2 Virus Isolated from Humans and its Seroprevalence Among Occupationally Exposed Populations in China

Background: The rst human-infected H9N2 inuenza case can be traced back to 1998. Although the H9N2 inuenza virus has low pathogenicity in animals, it donated partial or whole cassettes of internal genes to reassort novel viruses, such as H7N9, H10N8 and H5N6 viruses, that caused human infections with high fatality. Since 2013, sporadic but increasingly frequent human cases caused by H9N2 inuenza virus have been conrmed globally, and most of them were from China. Methods: Information on human infections with H9N2 inuenza virus was collected. Viral molecular determinants were determined by deep sequencing, and phylogenetic analysis was performed using MEGA 6.06. Antigenic analysis was performed by a hemagglutination inhibition (HI) assay. Receptor binding preference analysis was conducted based on a solid-phase binding assay with synthetic sialylglycopolymers. Antiviral susceptibility was determined by a uorescence-based neuraminidase (NA) inhibition assay. Serological study of occupationally exposed populations was performed by HI assay screening and conrmed by microneutralization assay. Results: From 2013 to 2018, 33 human H9N2cases were reported in China, among them 75.7% were children under 10 years old .The 22 viruses were isolated and concentrated in the Y280/G9 lineage of the HA and NA genes. All human H9N2 viruses belonged to the Y280/G9 antigenic lineage, presented a human-like receptor binding preference and remained susceptible to NA inhibitors, but most demonstrated resistance to M2 inhibitors. The seroprevalence of occupationally exposed populations was 2.15%, 3.17%, 2.93% and 1.54% from 2015 to 2018, respectively. A signicant difference in seroprevalence was shown between provinces with human cases (3.66%) and provinces without human cases (2.18%). Conclusions: The continuous antigenic drift and human-like receptor binding preference of the H9N2 virus enable it to have a high risk of causing human infections. The status of the seropositivity in occupationally exposed populations implies a substantial threat to public health. Research on human rst general analysis of all available human-isolated H9N2 inuenza viruses in China till to 2018 and the serological study among occupationally exposed populations nationwide in China. over 10 years before the G57 genotype predominated in vaccinated farm chickens in China. G57 genotype-like viruses nally facilitated the genesis of novel H7N9, H10N8 and H5N6 viruses [23]. Our study showed that recently isolated human H9N2 viruses clustered with G57-like viruses in phylogenetic trees of the HA and NA genes.

Homology analysis of nucleic acids and amino acids was performed on the NCBI website with BLAST. Phylogenetic analysis was performed on MEGA 6.06 using the Maximum Likelihood method. The reliability of the tree topology was assessed by bootstrapping with 1000 replications.

Antigenic analysis
A HI assay was performed to indicate the antigenicity difference. The paired viruses and relevant polyclonal ferret antisera were used as references, including representative viruses of the G1 and Y280/G9 lineage, such as A/quail/HongKong/G1/1997, A/chicken/HongKong/G9/1997, and A/HongKong/308/2014. All sera were pretreated and were diluted with PBS to a nal dilution fold of 1:10. An equal volume of antigen (8 HAUs/50 μl) was added to the sera at a serial 2fold dilution. The HI titers were determined by adding 50 μl/well 1% Turkey RBCs.

Serological study
Routine serological surveillance of occupationally exposed populations for avian in uenza virus has been implemented since 2008 in mainland China. According to the regulatory policy of the national pandemic preparedness plan in China, informed consent was exempted. Serum samples were collected from poultry workers who are working in live poultry markets, large-scale poultry farms, backyard poultry farms, poultry slaughter factories, or wild bird habitats.
Sera were pretreated in 4 volumes of RDE at 37°C for 18 hours followed by 56°C for 30 minutes. The HI assay was performed as described in the WHO manual [9]. An HI titer ≥40 was considered a suspected positive sample, followed by a microneutralization (MN) test for con rmation which was performed as previously described [13]. An MN titer ≥40 was considered seropositive for H9N2 infection.

H9N2 human cases in mainland China
A total of 33 H9N2-infected human cases were documented from 2013 to 2018 in mainland China. Almost all were mild cases except one fatal case, who suffered from an underlying chronic medical condition. Of the 33 cases, the proportion of males and females was 17:16, and the age ranged from 2 months to 84 years. The median age was 4 years old. Most of the individuals were children under 10 years old (25/33, 75.7%). The cases were distributed in 9 provinces and one municipality, which were mostly concentrated in central southern China, and two were located in northern China, including Gansu Province and Beijing municipality.  [5]. Laboratory-con rmed human H9N2 cases were reported by other countries, such as Bangladesh and Egypt, to the WHO; 2 cases were from Bangladesh, and 4 were from Egypt from 2011 to 2016 [5] (Table1).

Genetic characterization
A total of 29 human H9N2 virus sequences are available for analysis, including 24 human viruses (2013-2018), one virus (1999) in mainland China, 3 human viruses from Hong Kong and one from Bangladesh. We also chose 4 animal H9N2 viruses and one environment-related G1-like virus for this analysis. E627K and D701N mutations in the PB2 protein were con rmed for the virulence and transmission of H5N1, H7N9 and H9N2 viruses in mammals [14]. We found that 23/29 viruses contained E627 and that 28/29 contained D701. Only A/Sichuan-Bazhou/1453/2014 presented the change E627K. The other 5 Hunan H9N2 isolates, A/Hunan/34179/2018, A/Hunan/42088/2017, A/Hunan/37286/2017, A/Hunan-Chenzhou/45789/2015 and A/Anhui-Lujiang/39/2018, presented E627V substitutions. The E627V mutation has been studied in the H7N9 virus and con rmed to not compromised tness and transmissibility in both avian and mammalian species [15]. A/Hong Kong/33982/2009 had a D701N substitution.
The amino acid sequence of the HA cleavage site is PSRSSR↓GLF in all tested viruses since 2013, while motifs with more diversity were present in viruses isolated before 2013. All of which indicated molecular markers of low pathogenicity to avian. After 2014, all human H9N2 viruses had acquired Q226L, H183N, A190V and I155T mutations, which were con rmed to be related to human receptor binding preference [16]. Before 2014, human H9N2 viruses contained 2-3 of the 4 abovementioned amino acid site substitutions.
All H9N2 viruses isolated from humans acquired 3 amino acid deletions (positions 63-65) at the NA stalk region after 2013, which were supposed to increase virulence in chickens and mice [17]. NA position 274, 292, 294 and 119 substitutions that supported NA inhibitor resistance were not identi ed in all human The amantadine and rimantadine resistance mutation S31N/G in M2 was present in most H9N2 viruses [19]. Several amino acid changes in internal genes related to increased pathogenicity or virulence in H5N1 avian in uenza virus were not identi ed in H9N2 viruses in this study (Table 2).

Phylogenetic analysis
All human H9N2 in uenza viruses isolated in mainland China in this study belonged to the Y280/G9 lineage. Viruses isolated from 2013 to 2018 were clustered and represented by the viruses A/Hong Kong/308/2014 and A/Anhui-Lujiang/39/2018, candidate vaccine viruses (CVVs). However, A/Guangdong/333/99 was more similar than the CVVs to previous circulating viruses of the Y280/G9 lineage, such as BJ/94. The NA gene phylogenetic tree of H9N2 viruses showed the same model as the HA gene ( Figure 1).

Antigenic analysis
HI titers against the homologous viruses ranged from 1:160-1:5120. A one-way HI assay showed that all the tested human H9N2 in uenza viruses reacted well to at least one of the Y280 lineage reference antisera. A/Hong Kong/308/2014 (abbr. HK308) recommended as a vaccine strain in 2014, reacted well with most of the Y280 lineage viruses (12/17) before 2018. However, none of the 5 viruses isolated in 2018 were well inhibited by ferret antisera raised against the reference viruses before 2018. But all the viruses in 2018 reacted well with A/Anhui-Lujiang/39/2018 (abbr. AHLJ39), which was the representative strain in China and was recommended by the WHO as a new candidate vaccine (Table 3).

Receptor binding preference
All tested viruses were able to bind α2,6 sialylglycopolymers, although two viruses, A/Hong Kong/33982/2009 and A/environment/Guangdong/14883/2016, also bound to α2,3 sialylglycopolymers with moderate a nity. These two viruses presented dual receptor binding features. All human infected H9N2 viruses demonstrated the capacity to bind human receptor-like sialylglycopolymers, but different binding pro les existed. Some viruses (3/18), such as HN37286, JX47249 and GD333, had similar high a nities for 6'-SLN and 6'-SL, and some viruses (2/18), such as BJ58064 and GSJYG1397, had a preference for high a nity for 6'-SLN and lower a nity for 6'-SL. Most viruses (13/18) had a receptor binding preference only for 6'-SLN ( Figure 2).
We further tested the binding capacity of the tested H9N2 viruses with different types of erythrocytes (Supplementary Table 1). All viruses presented no hemagglutination with horse erythrocytes, which express almost exclusively α2,3 Gal linkages. Several studies have shown that human, guinea pig, chicken and turkey erythrocytes express both linkages. In addition, chicken erythrocytes display greater SAα2,3 Gal linkages, while turkey erythrocytes display greater SAα2,6 Gal linkages [20]. This is a possible reason why H9N2 virus titers in turkey RBCs were 0-1024-fold higher than those in chicken RBCs in our study.

Antiviral susceptibility of in uenza H9N2 viruses
All 26 tested viruses were fully susceptible to the two NA inhibitors oseltamivir and zanamivir, with mean IC 50 values ranging from 0.09-0.78 nM. These IC 50 values were similar to those of circulating NA inhibitor-susceptible human H3N2 viruses (0.1 nM for oseltamivir, 0.45 nM for zanamivir ) The fold change of H9N2 viruses IC 50 with NA inhibitor-sensitive reference virus's IC 50 is less than 4 (Supplementary Table 2).
Seroprevalence of H9N2 viruses in occupationally exposed populations Serum samples were collected from occupationally exposed populations from 2015 to 2018, with 15779, 14395, 15863 and 15523 samples, respectively. The seroprevalence rates based on the MN test were 2.15% (340/15779), 3.17% (456/14395), 2.93% (464/15863) and 1.54% (239/15523) each year. We further compared the positive rate of provinces with H9N2 human case occurrences and of those provinces without H9N2 human case occurrences on a yearly basis. The seroprevalence of provinces with human cases was signi cantly higher than that of provinces without human cases (P<0.05) (Table4). We de ned a seropositive rate of 2% as the baseline (average of seropositive rate is 2.4%), and the positive rates of 8 provinces in 2015, 18 provinces in 2016, 15 provinces in 2017 and 8 provinces in 2018 were above the baseline. Discussion H9N2 in uenza virus is usually called a panzootic virus since this subtype of virus has been shown to be a donor of internal genes to generate zoonotic in uenza viruses with pandemic potential. A prior study revealed that H9N2 in uenza virus provided its internal genes to generate a highly pathogenic avian in uenza H5N1 virus, which caused the human H5N1 outbreak in Hong Kong in 1997 [21]. Recent studies also supported the understanding of the emergence of novel in uenza H7N9, H10N8 and H5N6 viruses from reassortment with six internal genes of circulating H9N2 virus [6-8]. All facts indicated that this subtype of in uenza virus had acquired strong abilities of adaptation and spread among the avian population and exhibited high risks to humans after 2013. This situation may have caused human H9N2 in uenza cases to be concentrated after 2013. H9N2 in uenza virus has been one of the predominant subtypes affecting poultry health since the end of the twentieth century. It is stably established in chicken ocks and causes endemic outbreaks in vast areas in China. Moreover, this virus can be easily isolated from wild birds, live poultry markets, backyard ocks and the environment in China, even in air samples in live poultry markets [22]. This prevalence is the reason why humans infected by this virus are frequently reported.
The evolution of the H9N2 in uenza genotype was comprehensive. Multiple H9N2 genotypes were cocirculating over 10 years before the G57 genotype predominated in vaccinated farm chickens in China. G57 genotype-like viruses nally facilitated the genesis of novel H7N9, H10N8 and H5N6 viruses [23]. Our study showed that recently isolated human H9N2 viruses clustered with G57-like viruses in phylogenetic trees of the HA and NA genes.
HA of H9N2 in uenza viruses continues to evolve into two primary lineages (Y280/G9 and G1), which circulate in the poultry of some countries in Asia and the Middle East, such as China, Vietnam, Egypt and Bangladesh. In mainland China, the Y280/G9 virus has a broad range of hosts, such as chickens, ducks, minor poultry species, swine and humans. However, G1-like viruses have been isolated only from minor poultry and humans in southern China [24]. This result was consistent with our surveillance data of poultry-related environmental samples.
Receptor binding preference has important implications for in uenza replication and transmission [25]. In uenza H9N2 viruses have acquired the ability to bind human receptors while circulating in avian species and mammalian hosts. Although this ability was recently demonstrated by most H9N2 viruses, e cient human-to-human transmission was not present. This nding revealed that virus spread among humans needs a more complicated mechanism. Receptor binding preference was also revealed via the hemagglutination phenomenon of H9N2 viruses with different types of erythrocytes.
Clinical signs in chickens with H9N2 virus infection were quite mild. Chickens became more susceptible to secondary infection, especially Escherichia coli infections, with a mortality rate of at least 10%. The virulence of the H9N2 virus was higher in laying hens, and the morbidity of H9N2-infected laying hens was approximately 10% [26]. Human symptoms induced by the H9N2 virus are analogous to seasonal u. The outcomes of patients were quick recovery and no lethality. To date, only one fatal case was found in 2016, which occurred in an individual who suffered from a chronic underlying condition. Most human H9N2 in uenza cases were found through an in uenza-like illness (ILI) surveillance system rather than the unknown pneumonia surveillance system.
Human infection with in uenza H9N2 is often unnoticed since it generally results in mild or asymptomatic illness. Serological surveys are an optimal approach to identify subclinical infections and assess the risk of transmission to humans. A number of serological surveys or studies were carried out in some countries of Asia, Africa, the Middle East and North America [27]. Seroprevalence ranged from 1%-43% by HI and 0.6%-9% by MN, in which the results varied depending upon the infection de nition (cut-offs of antibody titer), viruses (circulating or previously circulated), and methods (HI or MN) used for testing. Selection of low titers as cut-offs can lead to overestimation of in uenza infection. Cross reaction with circulating seasonal in uenza viruses by HI assay usually results in a low sensitivity to in uenza infection determination. Few serological studies of in uenza H9N2 among the human population, especially for occupationally exposed populations, have been performed in some areas, such as Guangzhou, Jiangsu, Shandong, Beijing, and Shanghai, in China or nationwide [28][29][30]. These studies were based on designated populations, small sample sizes or short time durations and can re ect a partial view of the infection status of H9N2 in uenza in China. Seroprevalence ranged from 0.7-15.5%, which was similar to our study, in which 4-year continual surveillance data were analyzed from 2015 to 2018.
At present, the in uenza H9N2 virus has acquired a stronger ability to adapt to avian species and has led to more concern for public health, either directly infecting humans or generating novel in uenza viruses with pandemic potential. Serological surveillance also supported that subclinical infection existence. Efforts should be strengthened to monitor virus changes and infection status in terms of providing prepandemic warnings in a timely manner.

Conclusion
Human in uenza H9N2 cases occurred sporadically in China. There no human-to-human transmission has been identi ed to date. The continuous antigenic drift and human-like receptor binding preference of the H9N2 virus enable it to have a high risk of causing human infections. NA inhibitors are still working as the antiviral against the H9N2 in uenza virus. The seropositivity rate in occupationally exposed populations is variable among different regions, implying a substantial threat to public health. We should strengthen the surveillance of human infection with H9N2 in uenza virus and the research on virological characteristics. Availability of data and materials The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declared that they have no con icts of interest.    Figure 1 Phylogenetic trees of HA and NA genes from in uenza A H9N2 viruses. Phylogenetic tree was constructed using the Maximum Likelihood method in MEGA 6.06. The bootstrap value was tested with 1000 replications. Viruses in black square indicated being selected as the human H9N2 vaccine candidate recommended by WHO.