Mutation tryptophan to leucine at position 222 of haemagglutinin could facilitate H3N2 influenza A virus infection in dogs

Abstract

An avian-like H3N2 influenza A virus (IAV) has recently caused sporadic canine influenza outbreaks in China and Korea, but the molecular mechanisms involved in the interspecies transmission of H3N2 IAV from avian to canine species are not well understood. Sequence analysis showed that residue 222 in haemagglutinin (HA) is predominantly tryptophan (W) in the closely related avian H3N2 IAV, but was leucine (L) in canine H3N2 IAV. In this study, reassortant viruses rH3N2-222L (canine-like) and rH3N2-222W (avian-like) with HA mutation L222W were generated using reverse genetics to evaluate the significance of the L222W mutation on receptor binding and host tropism of H3N2 IAV. Compared with rH3N2-222W, rH3N2-222L grew more rapidly in MDCK cells and had significantly higher infectivity in primary canine tracheal epithelial cells. Tissue-binding assays demonstrated that rH3N2-222L had a preference for canine tracheal tissues rather avian tracheal tissues, whereas rH3N2-222W favoured slightly avian rather canine tracheal tissues. Glycan microarray analysis suggested both rH3N2-222L and rH3N2-222W bound preferentially to α2,3-linked sialic acids. However, the rH3N2-222W had more than twofold less binding affinity than rH3N2-222L to a set of glycans with Neu5Aca2–3Galb1–4(Fuca-)-like or Neu5Aca2–3Galb1–3(Fuca-)-like structures. These data suggest the W to L mutation at position 222 of the HA could facilitate infection of H3N2 IAV in dogs, possibly by increasing the binding affinities of the HA to specific receptors with Neu5Aca2–3Galb1–4(Fuca-) or Neu5Aca2–3Galb1–3(Fuca-)-like structures that are present in dogs.

Introduction

Influenza A viruses (IAVs) are enveloped, segmented, negative-strand RNA viruses belonging to the family Orthomyxoviridae. In addition to being one of the major causes of respiratory diseases in humans, IAVs naturally infect birds, pigs, horses and sea mammals, and have recently emerged in dogs (Crawford et al., 2005; Li et al., 2010; Song et al., 2008; Yoon et al., 2005). Two subtypes of IAVs, H3N8 and H3N2, have emerged in canine populations in North America and Asia, respectively (Crawford et al., 2005; Li et al., 2010; Song et al., 2008; Yoon et al., 2005). The H3N8 canine influenza virus (CIV) has been shown to be of equine origin, whereas the H3N2 CIV was of avian origin.

The H3N2 avian-origin CIV emerged in Korea and China in 2006 (Li et al., 2010; Song et al., 2008). Under experimental conditions, H3N2 CIV can infect dogs inoculated intranasally or via contact, resulting in respiratory disease (Song et al., 2009). A survey of 829 canine sera (361 farmed and 468 pet dogs) collected between June and December 2007 across Korea showed that the canine populations investigated had an H3N2 seroconversion rate of 19–100 % (Lee et al., 2009). In China, the seroconversion rate for H3N2 IAV was ~6.7 % in the sampled dog population (Li et al., 2010). These surveillance results suggest that H3N2 avian-origin CIV has become endemic in the canine populations in Korea and China.

The haemagglutinin (HA) glycoproteins of IAVs bind and enter host cells depending on the recognition of terminal sialic acid-capped glycosylated molecules (Horimoto & Kawaoka, 2005; Skehel & Wiley, 2000; Webster et al., 1992). Human IAVs preferentially bind to sialyloligosaccharides with N-acetyl sialic acid α2,6-linked galactose (NeuAcα2,6Gal), whereas avian influenza viruses (AIVs) and equine influenza viruses prefer N-acetyl sialic acid α2,3-linked galactose (NeuAcα2,3Gal) (Connor et al., 1994; Rogers & D’Souza, 1989; Rogers et al., 1983). Mutations in HA are frequently detected when AIVs are transmitted from avian species and adapt to mammalian species or are transmitted between mammalian species. For example, Q226L and G228S (for the H3 position) in the 220-loop of the HA receptor binding site can increase the ability of AIV to infect humans (Bateman et al., 2008; Connor et al., 1994; Ha et al., 2003; Liu et al., 2009; Nobusawa & Nakajima, 1988; Vines et al., 1998; Wan & Perez, 2007). Compared with its avian precursor virus, the HA of H3N2 CIV has seven mutations: T10A, D81N, L111I/V, A160T, D172N, W222L and D489N (Li et al., 2010). Of these mutations, only position 222 is located in the receptor binding site (220-loop) of H3N2 HA.

To date, little is known about the factors that resulted in the transmission of H3N2 IAV from avian to canine species. This study aims to characterize the impact of HA W222L, a single mutation detected consistently in H3N2 CIVs, on the receptor binding and host tropism of H3N2 CIVs. Our results suggest the substitution W222L in HA may facilitate H3N2 IAV infections in dogs.

Results

222L is conserved in H3N2 CIVs

A total of 204 and 1300 complete HA1 protein sequences of H3 subtype IAVs from canine and avian species, respectively, were downloaded from public influenza databases. The canine HA1 sequences include 34 from H3N2 and 166 from H3N8 CIVs. The avian HA1 sequences including domestic chicken, turkey, duck, goose, pigeon, and various migratory bird species, such as mallard, teal, coot, widgeon, pintail, northern shoveler, ruddy turnstone, wood duck, red necked stint, snow goose, swan, pelican, and so on. Sequence comparison revealed that leucine (L) is conserved at position 222 of HA in 33 out of 34 (97.06 %) H3N2 and 164 out of 166 H3N8 (98.80 %) CIVs; however, 748 out of 785 (95.29 %) avian HA1 proteins have tryptophan (W) in H3 AIVs, 35 of 88 have L (4.46 %) and two of 88 have arginine (R) (0.25 %), after removing identical sequences.

To better understand the potential roles of position 222 of HA in host tropisms, we analysed the amino acid diversities of this position in the H3 IAVs from other hosts, especially equine (H3N8), swine (H3N2) and human (H3N2). Our results showed that H3N8 equine IAVs have 222W, whereas the equine-origin H3N8 CIVs have 222L (Table 1). The swine IAVs predominantly have 222W with a few 222R; the majority of human IAVs have 222W in the HA genes from those isolates recovered before the year 2003 and 222R after the year 2003.

Table 1. Predominant residues at the receptor binding sites of H3N2 AIVs, CIVs and equine IAVs.

Viruses	Receptor binding sites
	130-Loop							190-Helix								22-Loop
	98	134	135	136	137	138	153	183	188	189	190	191	192	193	194	195	221	222	223	224	225	226	227	228
Avian (H3N2)	Y	G	G	S	G	A	W	H	N	Q	E	Q	T	S	L	Y	P	W	V	R	G	Q	S	G
Canine (H3N2)	Y	G	G	S	G	A	W	H	N	Q	E	Q	T	S	L	Y	P	L	V	R	G	Q	S	G
Canine (H3N8)	Y	G	R	S	G	A	W	H	N	Q/N	E	Q	T	K	L	Y	P	L	V/I	R	G	Q	S	G
Equine (H3N8)	Y	G	R	S	G	A	W	H	N	Q	E	Q	T	K	L	Y	P	W	V	R	G	Q	S	G

As described in Table 1, no other consistent amino acid mutations were observed in or near known receptor binding regions, including the 130-loop, 190-helix and 220-loop, between H3N2 CIVs and their corresponding precursor HAs in H3 AIVs (Li et al., 2010). Based on this observation, the hypothesis was proposed that a W222L mutation in HA could allow H3N2 CIVs to adapt from avian to canine hosts. To determine the role of the substitution W222L in receptor binding and host tropism of H3N2 IAVs, two recombinant viruses rH3N2-222L (canine-like) and rH3N2-222W (avian-like) were generated using reverse genetics based on the A/PR8/1934(H1N1) (PR8) backbone.

W222L increased infectivity and replication efficiency of H3N2 IAV in vitro

To compare infectivity, primary canine tracheal epithelial (CTE) cells were infected with either rH3N2-222L or rH3N2-222W. At 7 h post-infection, 8.77 % (1.45 % sd, n = 1,312) cells were infected with rH3N2-222L, whereas only 2.51 % (0.73 % sd, n = 1,712) cells were infected with rH3N2-222W (P<0.001) (Fig. 1). Texas red-conjugated Maackia amurensis lectin (MAA) staining revealed that the majority of the infected cells were ciliated cells (infection ratio±standard deviation: 0.88±0.089; the number of total cells exacted: n = 3024), suggesting that ciliated cells were the primary target for rH3N2-222L infection.

Fig. 1.

Infectivity assay of rH3N2-222L and rH3N2-222W in primary canine tracheal epithelial (CTE) cells. (a) Micrographs of the infected CTE cells stained with anti-nucleoprotein (NP)–FITC and Texas red–MAA and counterstained with DAPI. (b) Percentage of infected cells. In this assay, the CTE cells were infected with either rH3N2-222L or rH3N2-222W at an m.o.i. of 0.1 p.f.u. for 1 h at 37 °C. To measure the percentile of cells infected with viruses, the infected cells and non-infected cells were quantified from at least eight individual fields.

To evaluate virus replication kinetics, growth curves were determined in MDCK cells. As shown in Fig. 2, rH3N2-222L had a higher peak titre than rH3N2-222W and it yielded four times more viruses than rH3N2-222W at 24 h post-infection. These results indicate the mutation W222L increased the infectivity and replication of H3N2 IAV in vitro.

Fig. 2.

Growth kinetics of recombinant viruses rH3N2-222L and rH3N2-222W in MDCK cells. MDCK cells were infected at a m.o.i. of 0.01 TCID₅₀, respectively. Viruses in the supernatant collected at the indicated time points were titrated by TCID₅₀ in MDCK cells.

L222W affected the haemagglutination assay results

Haemagglutination assays were performed to compare the binding avidity of the two HA forms to turkey, chicken, guinea pig, dog and horse red blood cells (RBCs). In these assays, the virus quantities were normalized to the same HA titre of the WT H3N2 CIV using turkey RBCs. Although the HA titres of both recombinant viruses had similar values in chicken RBCs, the titre of rH3N2-222W virus was at least twofold lower than that of rH3N2-222L in guinea pig, dog and equine RBCs (Table 2). This HA binding profile clearly shows that the W222L mutation affects RBC binding depending on the species of origin, indicating position 222 is involved in virus receptor binding.

Table 2. L222W of HA changed the receptor binding affinity of H3N2 CIVs.

Virus stain	HA titre±sd*
Chicken	Guinea pig	Beagle	Horse
H3N2 WT	64±0	128±0	32±0	53.3±19
rH3N2-222L	64±0	128±0	32±0	64±0
rH3N2-222W	85.3±37	64±0	16±0	12±4

All virus concentrations were normalized to a HA titre of 128 in turkey RBCs.

^*Each assay was repeated three times.

L222W affected virus binding to canine and avian tracheal tissues

FITC-labelled rH3N2-222W and rH3N2-222L were bound to canine and avian tracheal tissues. As shown in Fig. 3, both rH3N2-222W and rH3N2-222L bound with more avidity to the canine tracheal tissue than the avian tracheal tissue. Compared with rH3N2-222W, rH3N2-222L showed slightly stronger binding to canine tracheal tissue. These binding assays indicated W222L substitution in H3N2 CIV conferred changes in the tracheal tissue-binding properties of the reassortant viruses and may increase the binding avidity of H3N2 IAV to canine tracheal tissues during infection.

Fig. 3.

Attachment of recombinant viruses rH3N2-222L and rH3N2-222W to avian and canine tracheal tissues. (a) Micrographs of demonstrating virus binding to avian and canine tracheal tissues. FITC-labelled rH3N2-222L or rH3N2-222W were incubated with ferret tracheal tissues, followed by incubation with rabbit anti-FITC antibody and development with aminoethylcarbazole substrate. The bound viruses are indicated using immunostaining (red-brown) located in the epithelial cilia of these tracheal tissues. (b) Summary of viral binding affinities. ^aMean abundance of cells and strength of signals to which virus attached were scored as follows: –, no attachment detected; +, attachment to rare or few cells; ++, attachment to a moderate number of cells; +++, attachment to many cells with moderate signals; ++++, attachment to many cells with strong signals; +++++, attachment to many cells with strongest signals.

L222W affected binding of H3N2 CIVs to a specific subset of glycans

Glycan microarray screening was used to identify the impact of L222W on the glycan-binding profile of H3N2 IAVs. Our results demonstrated that rH3N2-222L and rH3N2-222W bound preferentially to α2,3-linked glycans (Fig. 4 and Table S1, available in JGV Online), which are prevalent in the canine airway (Oshansky et al., 2011). The L222W substitution resulted in an at least twofold decrease in the mean relative fluorescence units (RFU) to a subset of α2,3-linked glycans with Neu5Aca2–3Galb1–4(Fuca-)- or Neu5Aca2–3Galb1–3(Fuca-)-type structures (Table 3). For example, for Neu5Gca2–3Galb1–4(Fuca1–3)GlcNAcb-Sp0, the RFU value of the rH3N2-222W binding signal was 63.89-fold less than the binding signal of rH3N2L.

Fig. 4.

Comparison of the glycan-binding affinity of H3N2 WT strain, rH3N2-222L and rH3N2-222W to representative glycans on the glycan array listed in Table S1. Different categories of glycans on the array are highlighted as follows: red, α2,3 sulfated sialosides; green, α2,3 di-, tri- and qua-sialosides; yellow, α2,3 linear sialosides; purple, α2,3 fucosylated sialosides; blue, α2,3 internal sialosides; black, α2,3 α2,6 sialosides; grey, α2,6 sialosides in the same order as listed for the α2,3 sialosides (see Table S1 for the name list and structures). RFU, Relative fluorescence units.

Table 3. rH3N2-222L showing higher binding affinity to a subset of glycan than rH3N2-222W.

Structure*	Mean RFU†		Ratio‡
rH3N2-222L	rH3N2-222W
Neu5Gca2–3Galb1–4(Fuca1–3)GlcNAcb-Sp0	6563	103	63.89
Neu5Aca2–3Galb1–4GlcNAcb1–3GalNAc-Sp14	2142	764	2.81
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb-Sp8	6566	617	10.64
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb-Sp0	3744	1131	3.31
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb1–3Galb-Sp8	6235	2514	2.48
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb1–3Galb1–4GlcNAcb-Sp8	10128	1916	5.28
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb1–3Galb1–4(Fuca1–3)GlcNAcb1–3Galb1–4(Fuca1–3)GlcNAcb-Sp0	7615	2576	2.96
Neu5Aca2–3Galb1–4(Fuca1–3)GlcNAcb1–2Mana-Sp0	8680	2642	3.28
Neu5Aca2–3Galb1–4(Fuca1–3)(6S)GlcNAcb-Sp8	9783	3495	2.80
Neu5Aca2–3Galb1–3(Fuca1–4)GlcNAcb-Sp8	6808	1338	5.09

^*Neu5Ac, sialic acid; Gal, galactose; Fuc, fucose; GlcNAc, N-acectyl-d-glucosamine; Glc, glucose; Man, mannose; SP0, -CH₂CH₂NH_2; SP8, -CH₂CH₂CH₂NH_2; Sp9, -CH₂CH₂CH₂CH₂CH₂NH_2; SP14, threonine.

^†RFU values were normalized across slides (see Methods).

^‡rH3N2-222L RFU/rH3N2-222W RFU.

Western blotting showed that canine tracheal tissues, but not avian tracheal tissues, had Neu5Gc glycans

To further test whether Neu5Gc-like glycans are canine tracheal tissue specific, but not chicken tracheal tissue specific, we performed Western blotting assay on protein extracts from both chicken and tracheal tissues using chicken anti-Neu5Gc antibody. Western blotting results showed that protein extracts from avian tracheal tissues had strong reactions with chicken IgY control antibody, but those from canine tracheal tissues did not (Fig. 5a). However, the protein extracts from canine tracheal tissues had strong reactions with chicken anti-Neu5Gc antibody with two specific bands corresponding to ~120 and ~100 kDa, respectively, but those from avian tracheal tissues did not (Fig. 5b).

Fig. 5.

Western blotting results of protein extracts from avian and canine tracheal tissues against chicken IgY control antibody (a) and chicken anti-Neu5Gc antibody (b). The rabbit anti-chicken IgG was used as secondary antibody and the nitrocellulose membrane was visualized in nitro blue tetrazolium/BCIP substrate. Western blotting results showed that protein extracts from avian tracheal tissues have strong reactivity with antibody chicken IgY antibody (control antibody), but those from canine tracheal issues do not (a), and that protein extracts from canine tracheal tissues have with strong reactivity (~120 and ~100 kDa) with chicken anti-Neu5Gc antibody, but those from avian tracheal tissues do not (b).

Discussion

In addition to its presence in avian-like H3N2 CIVs, W222L also appeared in the HA of equine-origin H3N8 CIV, despite the presence of 222W in equine H3N8 IAV (Payungporn et al., 2008). A recent study demonstrated the stability of W222L mutations during the course of experimental infections of H3N8 CIV in naïve and partially immune dogs, and in naturally infected dogs (Hoelzer et al., 2010). These studies support our hypothesis that the consistent mutation W222L at the receptor binding site of HA could contribute to CIV infection. A three-dimensional protein structure model showed that this residue probably participates in virus–receptor binding of H3N8 CIV (von Grotthuss & Rychlewski, 2006). The current study provides experimental evidence supporting the hypothesis that the substitution HA W222L helps in the adaptation of IAVs to canine receptors. In addition to the W222L substitution, there are additional mutations on HA, neuraminidase (NA) and other genomic segments (Li et al., 2010). These mutations may also play a role in CIV host tropism.

This study demonstrated that rH3N2-222L (canine-like) showed greater infectivity than rH3N2-222W (avian-like) in CTE cells in vitro. To our knowledge, this is the first demonstration of a potential selective advantage for rH3N2-222L in CTE cells, supporting the proposition that W222L can increase the infectivity of H3N2 CIVs in canine species. The high infectivity of rH3N2-222L (canine-like) in CTE cells is probably determined by effects of the W222L mutation in receptor binding, which is further confirmed by a stronger binding affinity of rH3N2-222L (canine-like) than rH3N2-222W (avian-like) to canine tracheal tissue.

Glycan microarray technology (Blixt et al., 2004) has been applied widely in identifying the glycan-binding profiles for influenza viruses, such as H5N1 highly pathogenic AIVs (Stevens et al., 2006, 2008), H2N2 and H1N1 pandemic influenza (Bradley et al., 2011a; Chen et al., 2011; Stevens et al., 2008), and H3N2 seasonal influenza viruses (Kumari et al., 2007). In this study, glycan profiling demonstrated that H3N2 CIVs predominantly bound to α2,3-linked sialic acids, which are present on CTE cells (Oshansky et al., 2011). The comparison data with the L222W mutant suggests that H3N2 CIV had a higher binding affinity (at least twofold difference) to the glycans with Neu5Aca2–3GalNAcb1–4GlcNAcb- and Neu5Gca2–3Galb1–4(Fuca1–3)GlcNAcb-like structures. Neu5Gc is a common form of mammalian sialic acid (Varki, 2001) and the results from glycan microarrays are consistent with the results from haemagglutination assays that rH3N2-222L had at least a two-fold higher titre to the mammalian RBCs than rH3N2-222W did (Table 2). Furthermore, Western blotting experiments in this study confirmed that Neu5Gc-like glycans are canine tracheal tissue specific and not chicken tracheal tissue specific (Fig. 5), supporting the idea that these structures account for the phenotypic differences of rH3N2-222L and rH3N2-222W in virus infectivity, haemagglutination and virus-tissue binding assays. The observed structures are conserved fucosylated glycans and further experiments need to be performed to study H3N2 CIV tissue tropisms by comparing fucosylation activities of the tissues across canine respiratory tracks, such as trachea and lung.

IAVs cross species barriers or transmit to other hosts from avian reservoirs mainly through HA, which determines the binding specificity to host sialic acid receptors. For example, avian and equine IAVs bind preferentially to those containing α2,3-N-acetylneuraminic acid-galactose linkages (NeuAcα2,3Gal). Human influenza viruses bind preferentially sialyloligosaccharide receptors with α2,6-acetylneuraminic acid-galactose linkages (NeuAcα2,6Gal) (Connor et al., 1994; Ito, 2000; Ito et al., 1997; Kumari et al., 2007; Rogers & D’Souza, 1989; Rogers & Paulson, 1983; Rogers et al., 1983; Stephenson et al., 2003; Vines et al., 1998; Viswanathan et al., 2010). Swine possess both α2,3- and α2,6-linked receptors, and, therefore, are susceptible to infection with mammalian and avian viruses (Ito et al., 1998). As a result, swine can serve as intermediate hosts for adaptation of AIVs to replication in mammals and as ‘mixing vessels’ to generate genetically novel viruses with pandemic potential (Bochner et al., 2005; Claas et al., 1998; Ma et al., 2008; Myers et al., 2007; Scholtissek et al., 1985; Shinde et al., 2009; Smith et al., 2009; Webster et al., 1992). However, it is still not clear how IAVs are transmitted among those hosts, e.g. avian, canine and equine, all of which have predominantly α2,3-linked glycans in their respiratory tracks. The proposed study suggested that, in addition to the linkages of glycans, e.g. α2,3 or α2,6 linkages, the structural moieties of the glycan receptors affect influenza host tropisms. As described above, H3N2 CIVs have a better binding affinity to Neu5Gc-like glycans, which are canine tracheal tissue specific, but not chicken tracheal tissue specific.

Other studies have also reported that the mutation at residue 222 at HA can affect influenza viral infectivity in mice (Bradley et al., 2011b; Meisner et al., 2008). A virus with a substitution from W to R at residue 222 at HA was isolated from the lungs of mice infected by a Y98F mutant of A/Aichi/2/68(H3N2) and the experiment demonstrated that this mutant increased the viral titres in lung tissues of the infected mice, compared with both the WT strain A/Aichi/2/68(H3N2) and the Y98F mutant of A/Aichi/2/68(H3N2). W222R could introduce a hydrogen bond between the viral HA and host glycan receptor (Gamblin et al., 2004). These results suggested that residue 222 of HA could play an important role in influenza host tropism.

In summary, this study suggests that the W222L substitution in HA could facilitate infection of H3N2 IAV in dogs, possibly by increasing viral binding affinity to canine-specific receptors with Neu5Aca2–3Galb1–4(Fuca-)- or Neu5Aca2–3Galb1–3(Fuca-)-like structures.

Methods

Virus and cells.

A/canine/Guangdong/1/2006(H3N2) was the first isolate identified from pet dogs in south China (Li et al., 2010) and this isolate was used as the WT CIV in this study.

MDCK and human embryonic kidney 293T cells were purchased from the American type Culture Collection. Both cell lines were maintained in Dulbecco's modified Eagle medium (DMEM) (Gibco-BRL), supplemented with 10 % FBS (Atlanta Biologicals), penicillin (100U ml⁻¹)/streptomycin (100 µg ml⁻¹) (Gibco-BRL), at 37 °C with 5 % CO₂. Primary canine tracheal epithelial (CTE) cells were generated in our laboratory using tracheal tissues from a healthy beagle dog according to protocols described previously (Busch et al., 2008; Gray et al., 1996; Quintana et al., 2011; Schroth et al., 1999; Shen et al., 2011; Shibeshi et al., 2008; Sime et al., 1997; Wu et al., 1985). CTE cells were maintained in bronchial epithelial cell growth medium (BEGM) (Lonza Walkersville).

Molecular cloning and mutagenesis.

Viral RNA was isolated from virus A/canine/Guangdong/1/2006(H3N2) using an RNeasy Mini Kit (Qiagen). The full-length HA and NA genes were amplified using SuperScript One-Step RT-PCR with Platinum Taq (Invitrogen) and then cloned into pHW2000 vector (Hoffmann et al., 2000). Mutagenesis of L222W on HA genes was performed using a QuikChange II XL site-directed mutagenesis kit (Agilent) according to the manufacturer's instructions. The primers used for cloning were HA-forward 5′-TATTGGTCTCAGGGAGCAAAAGCAGGGG-3′, HA-reverse 5′-ATGGTCTCCTATTAGTAGAAACAAGGGTGTTT-3′, NA-Forward 5′-TATTCGTCTCAGGGAGCAAAAGCAGGAGT-3′ and NA-reverse ATCGTCTCCTATTAGTAGAAACAAGGAGTTTTTTTGAAC-3′; the primers used for mutagenesis were HA-L222W-forward 5′-CATTGGATCTAGACCCTGGGTAAGGGGCCAATCTG-3′ and HA-L222W-reverse 5′-CAGATTGGCCCCTTACCCAGGGTCTAGATCCAATG-3′. The cloned HA and NA genes and the mutated HA genes were confirmed by Sanger sequencing at the Cornell University Life Sciences Core Laboratories Center.

Generation of recombinant viruses by reverse genetics.

The two plasmids for HA and NA of the H3N2 CIV and six plasmids for the internal genes of A/Puerto Rico/8/34 (H1N1) (PR8) were co-transfected into 293T cells. Cells were incubated at 37 °C for 8 h, followed by replacement of the DNA transfection mixture with 2 ml of Opti-MEM supplemented with penicillin (100 U ml⁻¹)/streptomycin (100 µg ml⁻¹) (Gibco-BRL). After 48 h incubation, the suspension medium was inoculated to 9-day-old embryonated specific pathogen-free chicken eggs and the eggs were maintained at 37 °C for 48 h. The sequences of the HA gene of the rescued viruses were confirmed by Sanger sequencing to be rH3N2 222L and rH3N2-222W.

Infectivity assay in CTE cells.

Confluent CTE cells in BEGM media were washed with PBS and inoculated with virus at a m.o.i. of 0.1 p.f.u. After 1 h incubation, the inocula were removed, and the cell monolayers were washed in PBS three times and then cultured in BEGM media. After 7 h incubation, the infected cells were fixed in 4 % paraformaldehyde, permeabilized using 0.3 % Triton X-100 and blocked using a BSA solution. Immunofluorescence was performed using anti-influenza nucleoprotein (Millipore) as the primary antibody and goat anti-mouse IgG (H+L) FITC-conjugated (Millipore) as the secondary antibody. Texas red-conjugated Maackia amurensis lectin (MAA) was used to determine cell type according to the manufacturer’s instructions (EY Laboratories). Blue DAPI (Sigma-Aldrich) was applied for nuclear staining. The number of immunofluorescent positive cells was manually counted from at least eight individual fields at ×400 magnification. Images were taken using a confocal microscope (LSM510 Meta; Zeiss).

Virus growth curve.

MDCK cells were infected with IAV at a m.o.i. of 0.01 TCID₅₀ per cell. After incubation at 37 °C for 1 h, cells were washed twice with PBS and incubated in Opti-MEM I (GIBCO) containing trypsin-TPCK (1 µg ml⁻¹) at 37 °C, 5 % CO₂. At indicated time points post-inoculation, 200 µl of supernatant was collected, aliquoted and stored at −70 °C until use. Finally, the supernatant from different time points was titrated in MDCK cells to determine the TCID₅₀.

Haemagglutination assay.

The haemagglutination assay was carried out as described elsewhere (Masurel et al., 1981). Briefly, 50 µl of each of the culture supernatants was serially diluted twofold in PBS in round-bottom plates. Subsequently, 50 µl of a 0.5 % suspension of RBCs was added to each well. The plates were incubated at 37 °C for 45 min before recording the HA titres.

Virus-tissue binding.

Concentrated inactivated viruses were labelled with FITC (Sigma) as described previously (van Riel et al., 2006; Xu et al., 2010). Then, 64 haemagglutinating units (HAU) per 50 µl of FITC-labelled viruses were incubated with deparaffinized/rehydrated tissue sections (normal, uninfected canine and avian tracheal tissues) overnight at 4 °C after 3 % H₂O₂ quenching for endogenous peroxidase activity and blocking reagent were applied to the tissue. The slides were washed three times with Tris/HCl sodium buffer (pH 7.2) with 0.05 % Tween-20 to remove unbound viruses. Then the slides were overlaid with a 1 : 50 dilution of polyclonal rabbit anti-FITC/HRP in PBS (DakoCytomation) and incubated at room temperature for 1 h. The signal was amplified with a tyramide signal amplification system following the instructions of the manufacturer (Perkin-Elmer). Tissue sections were developed with aminoethylcarbazole (ENZO Life Sciences) according to the manufacturer's instructions, followed by nuclear counterstaining with haematoxylin (Surgipath SelectTech; Leica). The mean abundance of cells and strength of signals to which virus attached were scored as in Fig. 3(b).

Glycan microarray assay.

Viruses were first purified using a 25 % sucrose cushion as described previously(Bradley et al., 2011a). The purified virus was labelled with desiccated Alexa Fluor 488 succinimidyl ester (Invitrogen) according to the manufacturer’s instructions. After dialysis, the labelled viruses were transferred to clean tubes and stored at 4 °C until used in glycan microarray hybridizations. In glycan hybridization, the version 5.0 glycan slides (Consortium of Functional Glycomics) were used as described (Bradley et al., 2011a). The binding image was read in a Perkin-Elmer ProScanArray scanner and analysed using Imagene version 6.0 image analysis software (BioDiscovery). Relative fluorescence unit (RFU) data were normalized by adjusting the total RFU to the same level across all experiments. A threshold of 2000 RFU was used to floor the samples and only the glycans with at least 2000 RFU were analysed statistically. The Wilcoxon signed rank-sum test was used to compare the glycan-binding patterns among canine H3N2 WT, rH3N2 222L and rH3N2-222W viruses.

Western blot.

Chicken and canine trachea tissues were harvested and frozen at −80 °C. Epithelium was excised from the trachea tissue, manually homogenized and lysed in RIPA buffer in the presence of protease inhibitors. Total protein concentration was determined using the BCA Protein Assay kit (Pierce/ThermoScientific) by following the manufacturer’s protocol. A total of 32 µg protein for each sample was loaded into each lane of the SDS-PAGE gel. Following SDS-PAGE, the separated proteins were transferred to nitrocellulose membranes and blocked in TBS/Tween-20 supplemented with fish gelatin and incubated overnight in either polyclonal anti-Neu5Gc IgY antibody (Sialix) or chicken IgY control antibody (Sialix). The membranes were washed and incubated for 1 h in rabbit anti-chicken IgG secondary antibody (Bethyl), followed by washing and visualization in nitro blue tetrazolium/BCIP substrate.