Two kinds of novel reassortment H3 subtypes of avian influenza viruses: similar genetic composition, different mammalian transmission capabilities

Abstract

To assess the potential threat of the novel H3 subtype Avian Influenza Virus (AIV) to the poultry industry and human health, whole-genome sequencing and phylogenetic tree and homology analysis were conducted on four H3N3 viruses and one H3N8 virus isolated from three different regions. Representative strains were selected for the study of pathogenicity and transmissibility in Specific Pathogen Free (SPF) chickens, BALB/c mice, and Hartley guinea pigs. Phylogenetic tree and homology analyses indicate that the eight gene segments of the four H3N3 viruses are derived from reassortments involving H3N8 virus (HA gene), H10N3 virus (NA gene), and H9N2 virus (internal genes). Additionally, the gene segments of one H3N8 strain are the result of reassortment between H3N8 virus (HA and NA genes) and H9N2 virus (internal genes). Multiple mammalian-adaptive mutations were detected in the gene fragments, including amino acid substitutions and alterations in glycosylation sites. Animal experimental results indicate that the B166 (H3N3) and K245 (H3N8) isolates were pathogenic to SPF chickens and BALB/c mice, and were also capable of infecting Hartley guinea pigs. Both strains transmitted among chickens, with B166 displaying slightly lower transmissibility than K245. However, only K245 was transmissible via direct contact in Hartley guinea pigs. This highlights the potential public health risks of H3 subtype viruses, posing a threat to the poultry industry and human health, which makes ongoing monitoring and further research crucial.

Introduction

Avian influenza (AI) is a zoonotic disease caused by Avian Influenza Virus (AIV) that primarily affects poultry but can also infect various mammals, including humans(Spackman, 2020). Reassortment of gene segments among various subtypes of AIV can lead to changes in pathogenicity, antigenicity and host range, potentially resulting in immune escape and pandemic(Garten et al., 2009; Kilbourne, 2006).

According to the differences in pathogenicity of AIV infection in chickens, AIV can be categorized into Highly Pathogenic Avian Influenza Virus (HPAIV) and Low Pathogenicity Avian Influenza Virus (LPAIV)(Garten et al., 2009). The H3 subtype of AIV is a low-isolation-rate LPAIV that persists in various bird species, weakening their resistance to opportunistic pathogens and impacting production performance, thus continuing to pose a threat to poultry. When H3 subtype of AIV coexists with other subtypes of AIV, it may lead to virus reassortment, resulting in the emergence of new AIV strains. It is important to note that the new viruses generated by reassortment may have significant impacts on both the poultry industry and public health. In recent years, internal gene reassortment between the H3N8 and H9N2 subtypes of AIV has led to the emergence of a novel H3N8 virus, which has spread widely in poultry populations and subsequently spilled over to humans(Cui et al., 2023). Although human cases remain sporadic, they indicate that the H3N8 subtype of AIV can cross the interspecies barrier and infect mammalian hosts, raising concerns about whether H3 subtype AIV could become a major public health threat in the future.

Through continuous monitoring of AIV, four novel reassortant H3N3 strains and one novel reassortant H3N8 strain were isolated. Based on the isolation results, this study selected representative strains to investigate their pathogenicity in chickens and mice, as well as their transmissibility in chickens and Hartley guinea pigs, aiming to gain an in-depth understanding and theoretically evaluate their potential threat to public health. These findings provide an evidence base to inform future policies for AIV prevention and control.

Materials and methods

Animals and ethics statement

Five-week-old SPF chickens were purchased from Shandong Sparfax Biotechnology Co.Ltd. BALB/c mice aged 4-6 weeks were purchased from Guangdong Zhiyuan Biomedical Science and Technology Co.Ltd. Hartley guinea pigs were purchased from Beijing Viton Lever Biotechnology Co.Ltd. The animals experiment conducted in this study was approved by the Animal Welfare and Ethical Censor Committee of South China Agricultural University (approval number: 2024c038). At the end of the experiment, all surviving animals were euthanized in accordance with ethical requirements.

Sample collection and virus isolation

Samples were collected from active surveillance sites at live poultry markets (LPMs) in Guangdong Province, including cloacal and oropharyngeal swabs from poultry, as well as diagnostic specimens submitted from other provinces veterinary stations and farms when abnormalities were detected.

According to Chapter 6 of the "National Standard for the Diagnosis of Highly Pathogenic Avian Influenza"(GBT18936-2020), virus isolation was performed using 9-11-day-old SPF chicken embryos provided by Guangdong Wens Dahuanong Biotechnology Co., Ltd. The singularity of the samples was initially identified through haemagglutination and haemagglutination inhibition (HA-HI) tests, and samples with mixed infections were purified using the limited dilution method.

Whole genome sequencing of isolates

The purified viral allantoic fluid was processed for RNA extraction following the protocol of the Feijie Total RNA Ultra-Fast Extraction Kit. The complete viral genome was then amplified using two-step RT-PCR and universal primers as reported by Hoffman(Hoffmann et al., 2001). The PCR products of all eight fragments were subjected to agarose gel electrophoresis, and the target fragments were recovered by OMEGA gel recovery kit, and the recovered products were sent to Sangon Biotech (Shanghai) Co.Ltd. Referring to the sequences in NCBI and GISAID Influenza Shared Database preliminary BLAST to determine the subtypes of virulent strains, through the DNA Star Lasergene V7.1 suite, Editseq for view comparison, Seqman for splicing and open reading frame detection, and finally obtain the complete viral gene sequence.

Phylogenetic analysis

Complete coding region sequences of avian influenza viruses (AIVs) from 2000 to 2024, encompassing the HA gene of the H3 subtype and whole genomes of multiple viral subtypes, were retrieved from the GISAID and NCBI databases. Multiple sequence alignment was performed using MAFFT v7.058 in automatic mode. The alignment was manually inspected in MEGA 6.0, and redundant sequences with high similarity were removed using BioAider v1.527, applying a nucleotide sequence identity threshold of 99.9%. The optimal nucleotide substitution model was determined by ModelFinder in PhyloSuite v1.2.2(Zhang et al., 2020), and the maximum likelihood phylogenetic tree was constructed using IQ-TREE. Trees were visualized and annotated with subtype and host information in iTOL (https://itol.embl.de/) and Adobe Illustrator 2018.

Analysis of molecular characterization of viruses

Sequence format conversions and key amino acid changes were analyzed using BioAider V1.527 and MEGA 6.0, and the online website NetNGlyc1.0 (https://services.healthtech.dtu.dk/services/NetNGlyc-1.0/) was used to predict potential glycosylation sites(Pugalenthi et al., 2020). Sequence similarity was analyzed with MegAlign.

Chickens experiment

The virus solution was diluted to 10⁶EID₅₀/0.2 mL for animal experiments(Biacchesi et al., 2005). Each group included 18 five-week-old SPF chickens. Chickens were infected via eye and nasal drops, while the control group was inoculated with 0.2 mL PBS in the same manner. Throat and cloacal swabs were collected at 1, 3, 5, 7, 9, and 11 days post-infection (dpi) from the experimental groups and stored at −80°C in 20% glycerol and double-antibody PBS solution. At 4 dpi, three chickens from the infected and cohabiting groups were randomly dissected, and tissue samples from the heart, liver, spleen, lungs, kidneys, brain, and trachea were collected to assess viral titers. Blood samples were taken at 14 dpi, and HI test was used to detect the presence of viral antibodies in the serum.

Mouse experiment

Thirty mice were randomly assigned to two infection groups or one PBS-treated control group (n = 10). After the viral stock was diluted to 10⁶EID₅₀/50 µL, mice were anesthetized with isoflurane (evidenced by abdominal breathing) and inoculated intranasally with 50 µL of the virus or PBS control.

Over a 14-day observation period, weight and survival were recorded daily. On 4 dpi, three mice per group were dissected to collect seven organ types for viral titer determination. Surviving mice were bled via the retro-orbital route under anesthesia on 14 dpi. The collected serum was RDE-treated and analyzed by HI assay for seroconversion.

Viral transmission in Hartley guinea pigs

Female SPF Hartley guinea pigs weighing 300-350 g were selected for the contact transmission experiment. Twelve guinea pigs were randomly divided into two groups of six: three in the infection group and three in the transmission group. The purified virus stock was diluted to 10⁶EID₅₀/0.3 mL(Cui et al., 2023; Ganti et al., 2022). Three guinea pigs were anesthetized with isoflurane until abdominal breathing was confirmed. A 150 µL viral inoculum was administered into each nostril. The animals were subsequently marked and housed individually in Individual Ventilated Caging (IVC) cages. Twenty-four hours later, three uninfected guinea pigs were introduced into the same IVC cage without physical barriers, allowing direct contact between animals to assess viral transmission under contact conditions. A control group inoculated with PBS was included in the experiment

Nasal lavage samples were collected at 1, 3, 5, 7, and 9 dpi after infection in the challenge group and at 1, 3, 5, 7, and 9 days after exposure in the transmission group, and viral titers were measured. All guinea pigs were observed until 14 dpi, after which blood was collected via cardiac puncture under anesthesia. Serum samples were tested for antibody levels to confirm viral infection.

Quantification and statistical analysis

Statistical analyses were conducted using GraphPad Prism V.6.0. One-way analysis of variance (ANOVA) was used to determine significant differences between groups. Data are presented as mean ± standard deviation, with statistical significance indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Results

Identification results of isolates

Between October 2022 and May 2023, routine surveillance was conducted on poultry farms and in LPMs across multiple regions of China. A total of 819 samples were systematically collected, including visceral tissues from farmed poultry and oropharyngeal/cloacal swabs from birds in LPMs. Through surveillance, four H3N3 and one H3N8 AIVs were isolated, all of chicken origin. Detailed data are provided in Table 1.

Table 1.

Information on related avian influenza strains involved in this study.

Virus names	Subtype	Year	Host	Location	Abbreviation
A/Chicken/Jiangsu/B166/230322/H3N3	H3N3	2023	Chicken	Jiangsu	B166
A/Chicken/Jiangsu/B170/230322/H3N3	H3N3	2023	Chicken	Jiangsu	B170
A/Chicken/Shandong/B210/1/230405/H3N3	H3N3	2023	Chicken	Shandong	B210-1
A/Chicken/Shandong/B210/3/230405/H3N3	H3N3	2023	Chicken	Shandong	B210-3
A/Chicken/Guangdong/K245/230419/H3N8	H3N8	2023	Chicken	Guangdong	K245
A/Jiangsu/428/2021/H10N3*	H10N3	2021	Human	Jiangsu	428
A/Yunnan/0110/2024/H10N3*	H10N3	2024	Human	Yunnan	0110
A/Changsha/1000/2022/H3N8*	H3N8	2022	Human	Hunan	CS1000
A/Henan/4/10CNIC/2022/H3N8*	H3N8	2022	Human	Henan	10CNIC
A/Guangdong/ZS/23SF005/2023/H3N8*	H3N8	2023	Human	Guangdong	ZS-23SF005

Note:.

^⁎refers to the reference strain.

Genetic evolution analysis of the complete genome

We compared the five isolated AIV strains with three previously reported human-origin H3N8 AIV strains—A/Changsha/1000/2022/H3N8 (CS1000), A/Henan/4/10CNIC/2022/H3N8 (10CNIC), and A/Guangdong/ZS/23SF005/2023/H3N8 (ZS-23SF005)—as well as two human-origin H10N3 AIV strains, A/Jiangsu/428/2021/H10N3 (428) and A/Yunnan/0110/2024/H10N3 (0110). Genome-wide nucleotide sequence homology(Clewley and Arnold, 1997) and genetic evolution were analyzed.

The phylogenetic analysis of the HA gene showed that the HA genes of the five H3 subtype isolates were clustered within the Eurasian branch, distinctly separated from the mammalian-origin strains, such as human influenza viruses, and the North American branch. These isolates shared the same branch as three human-origin H3N8 AIVs (Fig. 1). Genetic evolutionary analysis of the six internal genes—PB2, PB1, PA, NP, M, and NS—revealed that the isolates clustered within the same branch as H9N2 subtype AIV and shared a close evolutionary relationship with human-origin H3N8 subtype AIV strains (Fig. 2).Furthermore, the genome-wide phylogenetic analysis revealed that within the Eurasian branch, H3 subtype AIV strains in China have gradually formed a distinct sub-branch, indicating a potential for long-term widespread circulation in the region.

Fig. 1.

Phylogenetic evolutionary tree of HA genes. Three phylogenetic analyses were conducted using the maximum likelihood (ML) method with 5000 bootstrap replicates under the GTR+F + G4 model. The reference sequences were downloaded from the existing databases. The phylogenetic tree of HA was embellished according to branches and subtypes, where the red font represents the isolates of this study and the green font represents the reference human originated strains.

Fig. 2.

Phylogenetic evolutionary tree of genes within H3N3. (A to F) represent the phylogenetic evolutionary trees of PB2, PB1, PA, NP, M and NS genes, respectively. Where the isolates of this study are highlighted in pink and the human originated reference strain is highlighted in blue. The scale bar indicates the branch length based on the number of nucleotide substitutions at each sites. The tree topology was evaluated by 5,000 bootstrap analyses.

Molecular characterization analysis

By referring to the amino acid sequences of 5 human-origin H3N8 subtype AIV strains and comparing the amino acid sites on each gene fragment of the 5 isolates, the analysis shows that the HA protein cleavage patterns of the 5 H3 subtype strains in this study are all PEKQTR↓GIF, and no consecutive basic amino acids were found, which is the molecular characteristic of low pathogenic AIV(Xiong et al., 2014). Both 226Q and 228 G (H3 numbering,the following also applies) are typical avian receptor binding sites that preferentially bind the α−2,3-galactosialate receptor.All five isolates exhibited amino acid changes (G137S, T144A, I155T, and D172Y) compared to the human reference strain(Zou et al., 2020). Additionally, amino acid mutations at sites (34, 188, 229, 259, 261, 344, and 347) matched those in the human-origin strain ZS-23SF005. Analysis of the amino acid sequence of the NA gene revealed that none of the five strains had mutations in H274Y and I312V related to enhanced resistance to oseltamivir and Tamiflu(Liu et al., 2023).

Amino acid sequence analysis revealed that all isolates carried PB2 mutations (I292V, G309D, K318R, I333T, R477G, I495V) linked to increased viral pathogenicity, polymerase activity, and mammalian receptor affinity(Gabriel et al., 2005; Liu et al., 2023) .PB1 mutations (L13P, V171M, R198K, H436Y, L473V) suggest enhanced adaptability to mammals(Brown, 1990; Yang et al., 2021), while I368V promotes transmissibility in ferrets(Chen et al., 2007). In the PA gene, mutations E133K, L295P, and F666L were associated with increased pathogenicity in mice(Chen et al., 2007; de Wit et al., 2010), while N383D enhanced both pathogenicity and transmissibility in ducks(Song et al., 2011).The NP gene mutation M105V, present in all isolates, enhances pathogenicity in chicken(Tada et al., 2011). Mutations N30D and T215A in the M1 gene and L69P in the M2 gene, found in all isolates, are linked to increased pathogenicity in mice(Bosch et al., 1981; Hulse-Post et al., 2007; Tada et al., 2011). Additionally, the D97E mutation in the NS gene, which may enhance human pathogenicity, was consistently present in all four isolates(Seo et al., 2002).

Prediction of potential glycosylation sites on the HA and NA genes of the four H3N3 and one H3N8 subtype AIV strains revealed five conserved N-glycosylation sites on the HA peptide chain in all isolates. These sites were located at positions 38, 54, 181, and 301 in the HA1 coding region and position 499 in the HA2 coding region, consistent with the three human-origin reference strains (Table 2).

Table 2.

Analysis of potential glycosylation sites of HA protein in H3 subtype isolates.

Virus Strains	Potential Glycosylation Sites of HA Proteins
	38	54	181	301	499
	NGT	NAT	NVT	NGS	NGT
A/Chicken/Jiangsu/B166/230322/H3N3	+	+	+	+	+
A/Chicken/Jiangsu/B170/230322/H3N3	+	+	+	+	+
A/Chicken/Shandong/B210/1/230405/H3N3	+	+	+	+	+
A/Chicken/Shandong/B210/3/230405/H3N3	+	+	+	+	+
A/Chicken/Guangdong/K245/230419/H3N8	+	+	+	+	+
A/Changsha/1000/2022/H3N8*	+	+	+	+	+
A/Henan/4/10CNIC/2022/H3N8*	+	+	+	+	+
A/Guangdong/ZS/23SF005/2023/H3N8*	+	+	+	+	+

Note:.

^⁎refers to the reference strain.

Viral replication, pathogenicity, and transmissibility in SPF chickens

To compare the pathogenicity and transmissibility of the novel H3N3 subtype AIV with the novel H3N8 subtype AIV, both containing internal gene replacements from H9N2, one H3N3 isolate (B166) and one H3N8 isolate (K245) were selected for virulence and transmissibility assays in SPF chickens. During the experiment, mild symptoms such as ruffled feathers, warmth-seeking behavior, cold aversion, and diarrhea were observed. Positive swab results confirmed that both novel H3 subtype AIVs shed the virus through the oral cavity and cloaca (Table 3). Based on organ viral load and 14 dpi serum antibody levels (Fig. 3A,B,C), both novel H3 subtype AIVs efficiently replicated in the respiratory system of chickens and transmitted to cohabiting groups through direct contact. Higher viral titers in the trachea compared to the lungs indicate a preference for upper respiratory tract replication. Small amounts of the virus were also detected in extrapulmonary organs, including the heart, liver, spleen, kidneys, and brain, suggesting broad tissue tropism and the potential for systemic infection in chickens (Fig. 3A,B).

Table 3.

Comparison of detoxification of SPF chickens infected with H3N3 subtype isolates with H3N8 subtype.

Experimental Groups and Swab Types			Number of Positive Swabs/Total Number of Swabs Collected
			1dpi			3dpi		5dpi①		7dpi	9dpi	11dpi
B166	Infection group	Throat swab	6/6			6/6		3/3		0/3	0/3	0/3
		Anal swab	4/6			5/6		1/3		1/3	1/3	0/3
	Cohabitation group	Throat swab	1/6			5/6		3/3		3/3	2/3	0/3
		Anal swab	0/6			2/6		1/3		1/3	1/3	0/3
K245	Infection group	Throat swab	6/6			6/6		3/3		0/3	0/3	0/3
		Anal swab	3/6			4/6		2/3		1/3	1/3	0/3
	Cohabitation group	Throat swab	1/6			6/6		3/3		3/3	1/3	0/3
		Anal swab	0/6			2/6		3/3		3/3	0/3	0/3

Experimental Groups and Swab Types				Number of Positive Swabs/Total Number of Swabs Collected
				1dpi	3dpi		5dpi①		7dpi		9dpi	11dpi
PBS control group				Throat swab	-		-		-		-	-
				Anal swab	-		-		-		-	-

Note: “①” Three chickens in the infection and cohabitation group were killed on the fourth day respectively; “-” No positive results were observed.

Fig. 3.

Five-week-old SPF chickens inoculated with 200 µL of virus solution diluted to 10⁶EID₅₀/0.2 mL. (A) The horizontal dashed line indicates the lower limit of detection. Red bars and blue bars charts represent the viral titers of replicating B166 (H3N3) strain in the heart, liver, spleen, lungs, kidneys, brain, and trachea of SPF chickens in the infected and cohabiting groups, respectively. (B) The red bars and blue bars charts represent the viral titers of replicating K245 (H3N8) strain in the heart, liver, spleen, lungs, kidneys, brain, and trachea of SPF chickens in the infected and cohabiting groups, respectively. (C) The red bars and blue bars charts represent antibody levels in chicken serum after infection or exposure to 14 dpi of different strains, respectively.

B166 showed slightly stronger viral replication in the respiratory system than K245, indicating higher invasiveness of the H3N3 subtype AIV in chickens. However, slightly lower serum antibody levels in some cohabiting SPF chickens at 14 dpi suggest that its transmissibility is slightly weaker than that of the H3N8 subtype (Fig. 3C).

Viral replication and pathogenicity in BALB/c mice

To evaluate the potential risk posed by novel reassortant H3 AIV to mammalian and human health, this study assessed the pathogenicity of these strains in BALB/c mice as a mammalian model. During the experimental period, mild clinical signs, including slightly ruffled fur and huddling, were observed in the infected mice. Body weight measurements from 1 to 14 dpi (Fig. 4A) revealed a noticeable reduction among​ the B166 (H3N3) and K245 (H3N8) infection groups during the initial 1-3 dpi. However, this was followed by a gradual recovery, indicating a moderate but non-lethal level of virulence, to which the mice adapted relatively quickly.

Fig. 4.

BALB/c mice were inoculated with 50 µL of virus solution diluted to 10⁶EID₅₀/0.05 mL. (A) Change in body weight of the mice; (B) Red bars represent the antibody levels in the serum of mice infected with different strains for 14 dpi; (C) The horizontal dashed line indicates the lower limit of detection. The individual color bars represent the viral titers of replicating B166 (H3N3) or K245 (H3N8) strains in the heart, liver, spleen, lungs, kidneys, brain, and nasal turbinates of BALB/c mice in the infected groups, respectively.

Assessment of​ viral loads in organ tissues and specific serum antibody levels at 14 dpi (Fig. 4B,C) demonstrated that​ both novel H3 subtypes exhibit pronounced respiratory tropism in BALB/c mice. The B166 strain demonstrated the most efficient replication in the lungs, with infection confined exclusively to the respiratory system. The K245 strain, despite​ its weaker replication in the lungs, nonetheless achieved​ considerable viral titers in the kidneys and spleen. This indicates that K245 has a somewhat broader tissue tropism, albeit with a primary focus on the respiratory tract.

Viral transmissibility in Hartley guinea pigs

To investigate whether the novel H3N3 AIV and H3N8 AIV with similar components pose a potential risk of contact transmission among mammals, we conducted transmission experiments of AIV in guinea pigs. After infection, guinea pigs displayed clinical symptoms such as sneezing and nasal discharge. Viral titers were detected in the nasal lavage fluid of both infection groups, with K245 exhibiting a slightly longer shedding period than B166, as one guinea pig continued shedding until 7 dpi. In the K245 cohabiting group, all three guinea pigs were infected and shed the virus, whereas no viral titers were detected in the B166 cohabiting group. Serum antibody results confirmed infection and viral shedding in both infection groups, but only the K245 cohabiting group showed positive antibodies in all three guinea pigs. No antibodies were detected in the B166 cohabiting group. These findings suggest that while the novel H3N3 AIV isolate can infect guinea pigs, it has not yet developed the ability to transmit between them (Fig. 5A,B,C).

Fig. 5.

Hartley guinea pigs were inoculated with 300 µL of virus solution diluted to 106EID50/0.3 mL. (A) and (B) The horizontal dashed lines indicate the lower limit of detection.The bar chart with three colors shows the viral titers of the B166 (H3N3) or K245 (H3N8) strains in nasal lavage fluid from three guinea pigs in the infection or cohabiting groups at 1, 3, 5, 7, and 9 dpi. (C) The red and blue bars represent antibody levels in Hartley guinea pig serum at 14 dpi following infection or exposure to different strains.

Discussion

AIV exhibit high genomic plasticity, enabling frequent reassortment among co-circulating subtypes and the consequent emergence of novel genotypes. China—home to the world’s most intensive poultry production and vibrant LPMs—provides an ideal ecological arena for these evolutionary events to unfold. Recent H3N8 and H10N3 subtype AIV have been confirmed to originate from reassortment among gene segments of different AIV subtypes(Chen et al., 2023). The recent human infection with the novel H3N8 AIV has attracted widespread attention, prompting intensive investigation of its pathogenicity in poultry and mammals as well as its potential for onward transmission.(Chen et al., 2023; Yang et al., 2023). Although H3 AIV persist in poultry and exhibit a broad host range, they are typically low-pathogenic, causing only mild or sub-clinical infections(Wu et al., 2014). Beyond chickens and ducks, these viruses have repeatedly crossed the species barrier to horses, dogs, pigs and seals(Le et al., 2019). Although human infections remain sporadic, the potential threat of future variants to public health cannot be ignored. Therefore, to reduce the public health risks posed by H3 subtype AIV, it is necessary to strengthen active surveillance of H3 AIV, conduct in-depth research on the molecular characteristics and infection traits of mammalian-adapted variants, and promptly implement appropriate measures to mitigate potential health risks, providing a scientific basis for the prevention and control of possible outbreaks.

In recent years, the novel H10N3 AIV has demonstrated a greater affinity for human receptors to avian receptors. This indicates an increasing risk of H10N3 AIV infecting humans(Le et al., 2019; Liu et al., 2022). All five isolates in this study carried the six internal genes derived from H9N2-subtype AIV. Their HA gene originated from H3N8, while the only difference lay in the NA gene: the H3N3 isolates acquired NA from H10N3, whereas the H3N8 isolates retained the cognate H3N8 NA (Figure S1,S2). Accumulating evidence shows that multiple AIV subtypes now harbour the complete set of internal genes—or critical segments—originating from H9N2, including the human-infecting H3N8 strain identified in 2022 and the highly pathogenic H5N2 recently detected in East China(Cui et al., 2023; Zhang et al., 2023). This indicates that the G57 genotype of H9N2 virus plays a crucial role in the evolutionary reassortment of AIV. Such re-shuffling can abruptly alter virulence, transmissibility and replicative fitness, thereby reshaping epidemic dynamics.

The five AIV isolates in this study exhibited numerous mammalian-adapted mutations, including I292V, G309D, K318R, R477G, and I495V on the PB2 protein; L13P, V171M, R198K, H436Y, and L473V on the PB1 protein; E133K, L295P, and F666L on the PA protein; D97E on the NS1 protein; and N30D, T215A, and L69P on the M2 protein. These mutations in the internal gene-encoded proteins may significantly enhance the virus's replication capacity and virulence in mammals(Leow et al., 2023). Given the high mutation frequency, these amino acid changes in the internal gene-encoded proteins may significantly affect viral infectivity, highlighting the need for ongoing monitoring.

Pathogenicity experiments revealed that both H3N3 and H3N8 subtypes of AIV replicated efficiently in the respiratory system, with a preference for the upper respiratory tract of SPF chickens and mice. In direct-contact transmission experiments, the transmissibility of the two subtypes was compared between chickens and guinea pigs. Whole-genome phylogenetic analysis indicated that the hemagglutinin (HA) genes of H3N8 (K245) and H3N3 (B166) are closely related and belong to the same Eurasian lineage. Their internal gene constellations are also highly similar, with both strains carrying six internal genes derived from H9N2 avian influenza viruses. Despite this genetic resemblance, in vivo transmission studies revealed a marked contrast in phenotype: only the H3N8 subtype exhibited limited transmissibility in guinea pigs, whereas the H3N3 subtype showed no evidence of transmission via direct contact. Based on the data from this study and existing literature, we propose that the differences in transmission capacity between H3N3 and H3N8 viruses among guinea pigs may be partially attributed to functional divergence resulting from differences in the origin of their neuraminidase (NA) genes. There were no obvious abnormal mutations on the NA protein of the isolates in this study, but there were certain differences in potential glycosylation sites from those representing human strains. Changes in certain glycosylation sites of AI viral proteins may have a significant impact on their biological characteristics, including affecting virulence, and alterations in the glycosylation status of antigenic regions may also affect antigenicity, which is worthy of continued attention(Peng et al., 2019). Studies have found that the enzymatic activity of the NA protein enables the virus to cleave sialic acid on the surface of respiratory epithelial cells and on nascent viral particles during infection. This facilitating the release of the viral virions from the host cell, allowing them to infect the next cell(Gaymard et al., 2016). The NA gene of H3N8 originated from the North American clade and is highly homologous to that of human H3N8 strains. Its enzymatic activity may be more adapted to the upper respiratory tract environment of mammals, enabling more efficient cleavage of sialic acid receptors. This facilitates the release of viral particles from cells and their spread within the respiratory tract.

Additionally, structural features of the neuraminidase (NA) protein, such as stalk length, may also influence its function. Shortening of the NA stalk is a typical hallmark of influenza viruses adapting to terrestrial poultry. When crossing into mammalian hosts, influenza viruses typically require readjustment of NA stalk length (usually by elongation or restoration) to adapt to the upper respiratory tract environment of mammals(Durrant et al., 2016). Notably, none of the isolates exhibited alterations in neuraminidase stem length, indicating that the observed differences in transmissibility of the viruses in the guinea pig model were not caused by this structural change.

One possibility is that amino acid substitutions near the catalytic pocket or alterations in surface charge distribution may subtly modulate enzymatic activity or interactions with the respiratory mucus(Shtyrya et al., 2009). More importantly, the functional balance​ between the hemagglutinin (HA) and NA proteins from different sources may vary(Wagner et al., 2002). In this study, the NA of H3N8 may have achieved better cooperation with its cognate HA in the mammalian environment, thereby promoting efficient contact transmission. In contrast, the combination of HA from H3N3 with NA3 may not have evolved or reached such an optimal balance in mammalian hosts, which could hinder effective transmission via respiratory secretions between adjacent individuals. Consequently, NA proteins play an important role in the transmissibility and pathogenicity of influenza viruses(Liu et al., 2023). Thus, understanding how variations in NA protein structure contribute to differences in viral transmissibility remains an important area for further research.

Conclusion

The H3N3 AIV isolated in this study is a low-pathogenic novel influenza virus resulting from genetic reassortment of H3N8, G57-genotype H9N2, and H10N3 viruses. The eight gene segments of H3N8 AIV include HA and NA genes from the H3N8 strain and internal genes reassorted from the H9N2 strain. Both subtype isolates show broad tissue tropism in chickens and mice. In chickens, they primarily shed the virus through the oral cavity and cloaca and can spread via airborne transmission within flocks.The novel H3N8 AIV can transmit between guinea pigs through direct contact, while the H3N3 subtype isolate currently lacks this ability. However, the H3N3 AIV strain theoretically retains the potential to infect mammals, posing a potential threat to public health.

Funding

This work was supported by the Modern Agricultural Research System Innovation Team Project of Guangdong Province (grant no. 2024CXTD15), Science and Technology Program of Guangdong Province (grant no. 2024B1212070013), China Agriculture Research System of Ministry of Finance and Ministry of Agriculture and Rural Affairs (project no. CARS-41) and China National Animal Disease Surveillance and Epidemiological Survey Program (2021–2025) (grant no. 202111).

CRediT authorship contribution statement

Xinyu Han: Writing – original draft, Visualization, Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Muhui Zhong: Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Yujia Yang: Methodology, Investigation, Formal analysis. Shunyin Fang: Methodology, Investigation, Formal analysis. Yuting Shi: Methodology, Investigation, Formal analysis. Yaozhong Lin: Methodology, Investigation, Formal analysis. Xinkui Zhang: Methodology, Investigation, Formal analysis. Wenqi Wu: Methodology, Investigation, Formal analysis. Qinglong Wang: Methodology, Investigation, Formal analysis. Beibei Niu: Methodology, Investigation, Formal analysis. Qiuhong Huang: Methodology, Investigation, Formal analysis. Huifang Yin: Methodology, Investigation, Formal analysis. Ming Liao: Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition, Conceptualization. Weixin Jia: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Disclosures

The authors declare that there is no conflict of interest.