All animal experiments were conducted in ABSL-3 facilities and in accordance with the guidelines of South China Agricultural University’s Institutional Animal Care and Use Committee. All animal protocols were approved by the Committee on the Ethics of Animal Experiments of the ABSL-3 Committee of South China Agricultural University (approval no. L102012017001K).

Two H5N6 viruses—namely, A/oriental magpie-robin/Guangdong/SW8/2014 (SW8) and A/Pallas’s sandgrouse/Guangdong/ZH283/2015 (ZH283)—used in this study were grown and purified three times in Madin–Darby canine kidney cells by standard plaque assay. The stocks of H5N6 viruses were propagated in 9-day-old specific pathogen-free (SPF) chicken eggs at 37°C for 72 h per the procedure (Yuan et al., 2014). Allantoic fluid pooled from multiple eggs was taken for centrifugation for 2 min at 8,000 rpm, from which the supernatant was harvested and subsequently frozen in aliquots at -80°C for further characterization. The 50% egg infectious dose (EID₅₀) titer for egg-grown virus was determined by 10-fold serial dilutions and the titration of each virus in 9-day-old SPF eggs using Reed and Muench’s (1938) method. Six-week-old SPF white leghorn chickens (Guangdong Wens Dahuanong Biotechnology Co., Ltd, Yunfu, China) were held in isolator cages with a feeding space of 117 m³ throughout the duration of each experiment.

In vivo pathogenesis studies of wild bird H5N6 influenza viruses were designed as previously described (Zhang et al., 2008, 2009; Pu et al., 2015). In brief, groups of 12 6-week-old SPF chickens were intranasally inoculated with 0.2 mL of 10⁵ EID₅₀ of SW8 or ZH283, while a control group of 12 chickens was inoculated with 0.2 mL of phosphate buffered saline (PBS) using the same route. Three days later, six inoculated chickens from each group were humanely euthanized to test for viral replication in lung, kidney, spleen, cecal tonsils, bursa of Fabricius, trachea, pancreas, liver, heart, brain, duodenum, ileum, descending colon, and jejunum tissue. The remaining chickens were observed twice daily, at 8:00 and 20:00, for clinical symptoms, morbidity, and mortality for 14 days according to the protocol provided by the World Organisation for Animal Health².

Direct contact virus transmission experiments in chickens were conducted per the procedure of (Yuan et al., 2014). Briefly, the chickens of inoculated groups (n = 6) were intranasally inoculated with 0.2 mL of 10⁵ EID₅₀ of either the SW8 or ZH283 virus in a ABSL-3 laboratory, and after 24 h, additional naïve contact groups (n = 3) were also intranasally inoculated with 0.2 mL of PBS and placed in physical contact in the same cage to share feed and water with chickens inoculated with the virus. At 3 days post-infection (DPI), three inoculated chickens were humanely euthanized, and target tissues (i.e., brain, lung, spleen, and bursa of Fabricius) were harvested to determine viral titers and for RNA extraction. At 3, 5, 7, 9, and 11 DPI, oropharynx and cloacal swabs samples were collected for the detection of viral shedding and suspended in 1 mL of PBS. All tested tissues and swabs samples were harvested for viral detection and titration in SPF chick embryos. All surviving chickens were euthanized at 14 DPI, and the serum was harvested and tested for seroconversion by hemagglutination inhibition testing using 1% turkey erythrocytes (Stephenson et al., 2004).

Total RNA was extracted from the brain, lung, spleen, and bursa of Fabricius of H5N6-inoculated chickens and mock-infected chickens at 3 DPI using the Takara MiniBEST Universal RNA Extraction Kit (Takara Bio Inc., Tokyo, Japan) following the manufacturer’s instructions. Total RNA (1 μg) was reverse-transcribed with the PrimeScript^TM II 1st Strand cDNA Synthesis Kit (Takara Bio Inc.) and stored at -20°C for further study.

Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using a FastStart Universal SYBR Green Master kit (Roche Diagnostics, Shanghai, China). qRT-PCR primers (Table 1) were designed from published target sequences and previously reported (Adams et al., 2009) with Primer Premier 7.0 software (Premier Biosoft, Palo Alto, CA, United States). qRT-PCR was performed on a LightCycler480 (Roche Applied Science, Mannheim, Germany), the products of which were purified by using a DNA gel extraction kit (Takara Bio Inc., Tokyo, Japan). For the purposes of assay validation, purified products were cloned into pMD19-T and sequenced to verify correct target amplification.

The relative expression ratios of target genes in tested tissues vs. those in control tissues were calculated by the 2^-ΔΔCT method using the chicken housekeeping gene glyceraldehyde-3-phosphate-dehydrogenase (NM_204305) as the endogenous reference gene in order to normalize the level of target gene expression (Livak and Schmittgen, 2001). Standard deviations were determined by using the relative expression ratios of three replicates for each gene measured. Differences of virus titers and mRNA expression levels were statistically analyzed with an unpaired non-parametric test and paired Student’s t-test, respectively, using GraphPad Prism version 6.0 (GraphPad Software Inc., La Jolla, CA, United States) software. Compared to the mock-infected control, p < 0.05 and p < 0.01 were considered to indicate a statistically significant difference unless stated otherwise.

In previous research, we characterized the two H5N6 influenza viruses isolated from apparently healthy wild birds in 2014 and 2015 in Guangdong Province, China. On the one hand, SW8 was isolated from an oriental magpie-robin, and its PB2 gene with poultry H5N6 viruses shared the highest nucleotide similarity with that of A/chicken/Dongguan/2690/2013 (H5N6). On the other hand, ZH283 was isolated from a Pallas’s sandgrouse, and its PB2 gene shared the highest nucleotide similarity with that of A/Guangdong/ZQ874/2015 (H5N6) isolated from a 40-year-old woman who reported exposure to domestic poultry³.

To determine the pathogenicity of the viruses in chickens, we intranasally inoculated 6-week-old SPF white leghorn chickens with 10⁵ EID₅₀ of either H5N6 virus (i.e., SW8 or ZH283). All inoculated chickens exhibited clinical signs of illness, including severe depression, cloudy eye, and intermittent head-shaking, and died within 5 DPI, with a mean death time (MDT) of 3.3 to 4.0 d (Figure 1A). SW8 and ZH283 replicated systemically in chickens and at 3 DPI was detectable in all tested organs, including the respiratory tract (i.e., lung and trachea), kidney, lymphoid tissues (i.e., spleen, cecal tonsils, and bursa of Fabricius), pancreas, liver, brain, intestinal tract (i.e., duodenum, ileum, descending colon, and jejunum), and heart. SW8 and ZH283 replicated efficiently in the lower respiratory tract; high viral titers were detected in the lung, with mean titers of 6.33 log₁₀EID₅₀ and 8.58 log₁₀EID₅₀, respectively (Figure 1B). The two novel viruses also replicated in the brain, spleen, and bursa of Fabricius, with mean titers of 4.83–7.17 log₁₀EID₅₀, 5.83–7.33 log₁₀EID₅₀, and 6.08–7.58 log₁₀EID₅₀, respectively (Figure 1B). Overall, the two novel H5N6 influenza viruses of wild bird origin showed high pathogenicity in chickens and could replicate systematically in them.

To evaluate the horizontal intraspecies transmissibility of the two novel H5N6 viruses, three SPF chickens were intranasally inoculated with 0.2 mL PBS and introduced into the same cage as a naïve contact group, which were then housed with chickens inoculated with SW8 or ZH283. Shedding of SW8 could be detected from both oropharyngeal and cloacal swabs within 3 DPI, with viral titers in the ranges of 2.42–3.83 log₁₀EID₅₀ in oropharyngeal swab samples and of 1.52–3.79 log₁₀EID₅₀ in cloacal swab samples. ZH283 could also be detected from oropharyngeal and cloacal swabs within 5 DPI, with viral titers in the range of 4.58–4.75 log₁₀EID₅₀ in oropharyngeal swabs and of 3.50–3.90 log₁₀EID₅₀ in cloacal swabs (Figures 2A,B). Naïve contact chickens co-housed with chickens inoculated with SW8 did not die during the observation time, but all contact group chickens seroconverted and exhibited high titers (9.33 ± 0.58 log2), as shown in Table 2. Viral shedding was observed in both oropharyngeal and cloacal swabs, and viral titers of 1.50–1.83 log₁₀EID₅₀ within 5 DPI were detected in oropharyngeal swabs (Figure 2A); however, viral titers of the cloacal swabs could be detected (1.08 log₁₀EID₅₀) at 3 DPI (Figure 2B). Naïve contact chickens co-housed with chickens inoculated with ZH283 exhibited 100% lethality and mortality, with a MDT of 5.0 days (Table 2), and exhibited clinical signs of illness, including coughing, cloudy eye, and dyspnea. All surviving chickens in the naïve contact group co-housed with ZH283-infected chickens shed virus from the oropharynx and cloaca within 7 DPI, with mean viral titers of 2.75–3.75 log₁₀EID₅₀ in oropharyngeal swabs and of 1.75–4.50 log₁₀EID₅₀ in cloacal swabs (Figures 2A,B). In short, results demonstrate that the two novel H5N6 influenza viruses replicated efficiently in chickens and exhibited efficient transmission via direct contact in the chicken model.

Toll-like receptors are PRRs with a unique and essential physiological function in host immune systems activated by pathogen-associated molecular patterns (Medzhitov, 2001). Expression profiles of two PRRs—TLR3 and TLR7—were examined in the target tissues of H5N6-infected chickens. As shown in Figure 3A, in contrast to mock-infected chickens, their expression level of TLR3 in the brain and lung was significantly elevated when induced by both viruses, with a fold increase of 2.80–78.56 in the brain or lung. In the spleen, the expression level of TLR3 was downregulated in response to SW8 infection, yet upregulated following infection with ZH283. In the bursa of Fabricius, the expression level of TLR3 was markedly downregulated when induced by both viruses. The expression level of TLR7 was upregulated in the lung when induced by SW8 or ZH283, by 1.79- and 19.41-fold, respectively. However, the expression level of TLR7 in the brain, spleen, and bursa of Fabricius showed different expression patterns when induced by the viruses; TLR7 expression level was downregulated when induced by both viruses compared to the control, with a fold change of 0.003–0.78 in all tested tissues except lung tissue. In particular, TLR7 expression remained low and was no longer visible in the bursa of Fabricius when triggered by both viruses. Notably, the expression levels of TLR3 and TLR7 in target tissues induced by ZH283 were generally greater than those induced by SW8 (Figures 3A,B).

As an indispensable regulator of inflammation activation, S1PR1 plays a crucial role in immune cell trafficking and immune response (Rivera et al., 2008). When induced by SW8, S1PR1 expression was upregulated in the brain, lung, and spleen—by 5.51-, 2.29-, and 1.16-fold, respectively—but not in the bursa of Fabricius (0.17-fold). However, the expression level S1PR1 showed different tendencies when infected by ZH283. Unlike the expression level of TLR3 and TLR7, S1PR1 expression in the tested tissues after infection with ZH283 was lower than that in response to infection with SW8 (Figure 3C). Our data indicate that the engagement of PRRs and S1PR1 by the H5N6 influenza virus occurs in a tissue-dependent manner.

The engagement of TLRs by influenza virus in specific target tissues initiated animal immunity via the production of proinflammatory cytokines and chemokines, including IL-1β, IL-6, IL-8, IL-10, TNF-α, and CCL5. As shown in Figures 4A,B,D, the expression level of IL-1β, IL-6, and IL-10 were remarkably unregulated in the lungs of tested chickens when infected by SW8 and ZH283 compared to those of mock-infected chickens. On the contrary, in the brain, spleen, and bursa of Fabricius, the expression levels of IL-1β, IL-6, and IL-10 were downregulated when induced by both viruses. However, the expression levels of IL-8, TNF-α, and CCL5 in the tested tissues of infected chickens showed a different expression patterns. As illustrated in Figure 4F, ZH283 induced an upregulated expression level of CCL5 in all tested tissues, whereas SW8 induced an upregulated expression level of CCL5 in the brain and spleen, but a downregulated one in the lung and bursa of Fabricius. Notably, ZH283-induced expression levels of IL-1β, IL-8, TNF-α, IL-6, and IL-10 were greater than those induced by SW8 in all tested tissues of chickens (Figures 4A–E).

The activation of TLRs also mediated the activation of IFN regulatory factor 3/7, primarily by recruiting MyD88 or TNF receptor-associated factor 6, which ultimately activated I and II IFNs (i.e., IFN-α, IFN-β, and IFN-γ). In the lungs of tested chickens, both ZH283 and SW8 induced significantly upregulated expression levels of IFN-α, IFN-β, and IFN-γ by 7.55- and 75.97-fold, 68.23- and 362.80-fold, 30.11- and 85.31-fold, respectively (p < 0.05) compared to uninoculated chickens (Figures 4G–I). In contrast to the lung, the brain, spleen, and bursa of Fabricius showed different expression patterns in the levels of IFN-α, IFN-β, and IFN-γ in response to ZH283 and SW8 infection. However, ZH283 induced the expression levels of IFN-α, IFN-β, and IFN-γ to a greater extent than SW8 in the tested tissues of infected chickens.

In sum, our data indicate that the mRNA expression profiles of proinflammatory cytokines and chemokines showed different patterns in tested tissues likely associated with the pathogenic difference of both viruses in chickens.

To investigate whether MHC classes I and II molecules were involved in the host innate immune response to H5N6 influenza virus infection, we examined their expression levels in the lung, brain, spleen, and bursa of Fabricius in chickens at 3 DPI. As illustrated in Figures 5A,B, MHC classes I and II molecule expression levels were upregulated in the brain, spleen, and bursa of Fabricius when infected by both viruses. In the lung, in contrast to the mock-infected control, the expression level of the MHC class I molecule was remarkably downregulated (0.063- and 0.20-fold, respectively, p < 0.05); however, that of the MHC class II molecule was significantly upregulated when induced by SW8 and ZH283 (12.83- and 99.08-fold, respectively, p < 0.05). Those results demonstrated that MHC classes I and II molecules could play a significant role in the course of host innate immune response to H5N6 influenza virus infection in chickens.