Aerosol Transmission of Gull-Origin Iceland Subtype H10N7 Influenza A Virus in Ferrets

Subtype H10 avian influenza A viruses (IAVs) have caused sporadic human infections and enzootic outbreaks among seals. In the fall of 2015, H10N7 viruses were recovered from gulls in Iceland, and genomic analyses showed that the viruses were genetically related with IAVs that caused outbreaks among seals in Europe a year earlier. These gull-origin viruses showed high binding affinity to human-like glycan receptors. Transmission studies in ferrets demonstrated that the gull-origin IAV could infect ferrets, and that the virus could be transmitted between ferrets through direct contact and aerosol droplets. This study demonstrated that avian H10 IAV can infect mammals and be transmitted among them without adaptation. Thus, avian H10 IAV is a candidate for influenza pandemic preparedness and should be monitored in wildlife and at the animal-human interface.

ABSTRACT

Subtype H10 influenza A viruses (IAVs) have been recovered from domestic poultry and various aquatic bird species, and sporadic transmission of these IAVs from avian species to mammals (i.e., human, seal, and mink) are well documented. In 2015, we isolated four H10N7 viruses from gulls in Iceland. Genomic analyses showed four gene segments in the viruses were genetically associated with H10 IAVs that caused influenza outbreaks and deaths among European seals in 2014. Antigenic characterization suggested minimal antigenic variation among these H10N7 isolates and other archived H10 viruses recovered from human, seal, mink, and various avian species in Asia, Europe, and North America. Glycan binding preference analyses suggested that, similar to other avian-origin H10 IAVs, these gull-origin H10N7 IAVs bound to both avian-like alpha 2,3-linked sialic acids and human-like alpha 2,6-linked sialic acids. However, when the gull-origin viruses were compared with another Eurasian avian-origin H10N8 IAV, which caused human infections, the gull-origin virus showed significantly higher binding affinity to human-like glycan receptors. Results from a ferret experiment demonstrated that a gull-origin H10N7 IAV replicated well in turbinate, trachea, and lung, but replication was most efficient in turbinate and trachea. This gull-origin H10N7 virus can be transmitted between ferrets through the direct contact and aerosol routes, without prior adaptation. Gulls share their habitat with other birds and mammals and have frequent contact with humans; therefore, gull-origin H10N7 IAVs could pose a risk to public health. Surveillance and monitoring of these IAVs at the wild bird-human interface should be continued.

IMPORTANCE Subtype H10 avian influenza A viruses (IAVs) have caused sporadic human infections and enzootic outbreaks among seals. In the fall of 2015, H10N7 viruses were recovered from gulls in Iceland, and genomic analyses showed that the viruses were genetically related with IAVs that caused outbreaks among seals in Europe a year earlier. These gull-origin viruses showed high binding affinity to human-like glycan receptors. Transmission studies in ferrets demonstrated that the gull-origin IAV could infect ferrets, and that the virus could be transmitted between ferrets through direct contact and aerosol droplets. This study demonstrated that avian H10 IAV can infect mammals and be transmitted among them without adaptation. Thus, avian H10 IAV is a candidate for influenza pandemic preparedness and should be monitored in wildlife and at the animal-human interface.

INTRODUCTION

Influenza A viruses (IAVs) belong to the family Orthomyxoviridae and are classified into different antigenic subtypes based on their surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA). Sixteen HA and nine NA IAV subtypes have been recovered from wild birds (1–3). In addition to infecting humans, IAVs infect a wide range of natural hosts (e.g., avian, swine, canines, and equines), among which migratory waterfowl, shorebirds, gulls, and terns serve as the major IAV reservoirs and play important roles in virus movement, transmission, and genetic reassortment because of their seasonal movements (4, 5).

Subtype H10 IAVs have been recovered from a range of avian and mammalian species. Sporadic cases of H10 avian IAV infection in humans have been reported, but human-to-human transmission has not been established. In 2004, the H10N7 virus caused fever and cough in two infants in Egypt (6); in 2010, two abattoir workers in Australia were found to be H10 virus positive during a low-pathogenic avian influenza outbreak among chickens (7); and in 2013, an H10N8 avian IAV infected three humans in China, resulting in two deaths (8). In addition to humans, H10 IAVs have also been reported in other mammals, including mink (9), seals (10), and pigs (11).

In 2014, dead harbor seals (i.e., Phoca vitulina) in Germany, Sweden, and Denmark were found to be infected with a subtype H10N7 IAV (10, 12, 13). The virus caused respiratory diseases and transmitted efficiently among harbor seals and gray seals (i.e., Halichoerus grypus), leading to multiple outbreaks across multiple regions throughout Europe (14, 15); from March through November 2014, the outbreaks led to 1,400 deaths among a population of approximately 12,000 seals in these regions (13). Serosurveillance suggested that up to 58% of adult harbor seals and 28% of adult gray seals were exposed to this H10N7 virus (14). Laboratory studies in a ferret model showed that this seal H10N7 virus caused respiratory tract inflammation extending from the nasal cavity to bronchi but not lung (16). Similar to the genomic segments of an H3N8 virus isolated from dead North American harbor seals (17), all eight genomic segments of this virus were of avian origin. Prior study suggested that this H3N8 virus could be transmitted among ferrets without adaptation through aerosol droplets (18). Of note, most low-pathogenic avian IAVs replicate strictly in the gastrointestinal tracts of avian hosts and replicate to only a limited extent in mammals.

In 2015, we isolated four H10N7 IAVs from gulls in Iceland; the risk these viruses posed to the sea mammal population and, potentially, to humans was unknown. To assess the risk to mammal populations in the North Atlantic, we characterized the four viruses in terms of their genetic associations with the 2014 European seal virus and their antigenicity, receptor binding preference, pathogenicity, and transmissibility.

RESULTS

Genetic analyses of gull-origin H10N7 IAVs.

We isolated and genetically sequenced IAVs from four gulls, all of which were sampled on the western coast of Iceland during 10 November to 18 November 2015. One of the four isolates [A/glaucous gull/Iceland/4552/2015(H10N7), abbreviated Gg/4552] was from an adult glaucous gull (Larus hyperboreus), and another [A/glaucous gull/Iceland/4270/2015(H10N7), abbreviated Gg/4270] was from a hatch-year glaucous gull. One of the other two isolates [A/Iceland gull/Iceland/4266/2015(H10N7), abbreviated Ig/4266] was from an adult Iceland gull (Larus glaucoides), and the other [A/Iceland gull/Iceland/4402/2015(H10N7), abbreviated Ig/4402] was from a hatch-year Iceland gull. Genetic sequencing showed that the isolates were >99.50% similar to each other across all eight genetic segments. Furthermore, phylogenetic analyses showed that all eight gene segments of these H10N7 IAVs from Iceland were of Eurasian lineage and that four genes (i.e., HA, NA, PB2, and MP) of these Icelandic H10N7 IAVs were genetically similar to an H10N7 IAV isolate [A/harbor seal/Germany/1/2014(H10N7)] recovered from a sick seal on 7 November 2014, whereas four other genes (i.e., PB1, PA, NP, and NS) of these Icelandic H10N7 IAVs belong to at a genetic clade distinct from those of the seal origin isolates (Fig. 1; see Fig. S1 in the supplemental material). Molecular characterization suggested that the PB1, PA, and NP genes of the Iceland gull-origin H10N7 IAVs (e.g., A/glaucous gull/Iceland/4270/2015) and those of seal-origin isolates shared nucleotide sequence identities 93.51%, 93.78%, and 95.49%, respectively, whereas those for all the other 5 genes (HA, NA, PB2, MP, and NS) were >96.00%. For example, the HA and NA genes of the Iceland gull-origin H10N7 IAVs shared 97.69% and 98.68% nucleotide sequence identity, respectively, with HA and NA genes of A/harbor seal/Germany/1/2014(H10N7). Phylogenetic analyses suggested that the NS genes from the seal H10N7 and the gull viruses we recovered do not belong to the same genetic clade, although they shared a nucleotide sequence identity of 96.15%.

FIG 1.

Phylogenetic and antigenic analyses of subtype H10 influenza A viruses (IAVs). (A) Hemagglutinin gene, (B) neuraminidase gene, and (C) polymerase acidic protein. The phylogenetic tree for each gene segment was inferred by using a maximum-likelihood method by running RAxML v8.2.9 and by using a Gamma model of rate heterogeneity and a generalized time-reversible substitution model (72). The bootstrap values were labeled for the selected representative branches with bootstrap values of ≥70. Phylogenetic trees were visualized by using FigTree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). Scale bars represent nucleotide substitutions per site. Red indicates the H10N7 IAV strain isolated from an Iceland gull in 2015; green indicates the subtype H10N7 avian IAVs isolated from European seals in 2014; and the virus marked in blue indicates the IAVs isolated from Iceland from other studies.

Sequence analyses showed that most amino acids at the HA receptor binding sites were conserved among the Iceland gull-origin H10N7 isolates, the seal-origin H10N7 isolates, the H10 isolates from Eurasia, and the H10 isolates from North America (see Table S1 in the supplemental material). The Iceland gull-origin H10N7 isolates and other wild bird-origin H10 isolates had amino acid 222Q (corresponding to 226 at H3), which is identical to those at the HA sequences of two H10 isolates from the 2014 outbreak among European seals (Table S1). However, the other four seal isolates from the 2014 outbreak among European seals have amino acid 222L instead (19) (Table S1). All H10 isolates had amino acid 224G (corresponding to 228 at H3) (Table S1).

Antigenic analyses of gull-origin H10N7 IAVs.

Results from hemagglutination inhibition (HI) assays suggested homologous titers of seven prototype H10 IAVs from Eurasia, North America, and Australia, including A/chicken/Jiangxi/34609/2013 (H10N8), A/laughing gull/DE/209/2013 (H10N8), A/glaucous gull/Iceland/4552/2015 (H10N7), A/seal/Netherlands/P14-221/2014 (H10N7), A/mallard/Netherlands/1/2014 (H10N7), A/duck/Bangladesh/24035/2015 (H10N1), and A/chicken/Germany/N/49 (H10N7), ranged from 1:40 to 1:1,280 (see Table S2 in the supplemental material). The four Iceland gull-origin H10N7 IAVs cross-reacted with ferret sera against these seven H10 prototype viruses, and the HI titers varied from 1:40 to 1:320 (Table S2).

Antigenic cartography analyses using AntigenMap (http://sysbio.cvm.msstate.edu/AntigenMap) (20, 21) showed no clear antigenic clusters among 32 viruses, namely, the 4 Iceland gull-origin H10N7 IAVs, 2 mammalian-origin (seal and mink) H10N7 IAVs, 10 domestic poultry-origin (chicken and duck) IAVs, and 16 wild bird IAVs (Fig. 2). The average antigenic distance between all 32 H10 isolates we tested was 1.3569 (±0.8247 standard deviation [SD]) units, the antigenic distances between 16 wild bird-origin H10 isolates was 1.3228 (±0.9137) units, and the antigenic distance between the four gull H10N7 isolates was 1.2259 (±0.5692) units; each unit corresponds to a 2-fold change in HI titer. The antigenic distance between A/seal/Netherlands/P14-221/2014(H10N7) and four gull H10N7 isolates was 2.1107 (±0.9278) units and that between domestic poultry-origin isolates and four gull H10N7 isolates was 1.2700 (±0.7246) units. Sequence analyses showed that antibody binding sites were conserved among these H10 viruses (data not shown).

FIG 2.

Antigenic map of 32 subtype H10 influenza A viruses (IAVs). The map was constructed using hemagglutination inhibition (HI) data derived from ferret antisera. The open triangles indicate viruses isolated from humans or other mammals; the open circles indicate viruses isolated from domestic poultry; and the black circles indicate viruses isolated from wild birds. IAVs with identical HI titers may have overlapping positions on this antigenic map. The map was constructed using AntigenMap (http://sysbio.cvm.msstate.edu/AntigenMap) (20, 21). Each gridline (horizontal and vertical) in the map represents one antigenic unit distance corresponding to a 2-fold difference in HI titers. Ig/4266, A/Iceland gull/Iceland/4266/2015(H10N7); Gg/4552, A/glaucous gull/Iceland/4552/2015(H10N7); Gg/4270, A/glaucous gull/Iceland/4270/2015(H10N7); Ig/4402, A/Iceland gull/Iceland/4402/2015(H10N7); rg-A/Jiangxi-Donghu/346/2013[R] (6:2) (H10N8), a reassortant with HA and NA genes from A/Jiangxi-Donghu/346/2013 (H10N8) and six other genes from A/PR/8/1934(H1N1).

In summary, the above analyses suggested that the four Icelandic gull H10N7 isolates (Gg/4552, Gg/4270, Ig/4266, and Ig/4402) are antigenically and genetically similar to each other and that these H10N7 isolates do not have significant antigenic diversities from but do have different levels of genetic diversity (e.g., internal gene segments) with the contemporary H10 isolates from avian, mammals, and seals we studied. Based on these results, from these four Icelandic gull H10N7 isolates, Ig/4266 was randomly selected for detailed analyses on its replication, glycan receptor binding, and transmission properties as described in the next sections.

Gull-origin H10N7 IAV had competitive replication in vitro.

To determine the ability of gull-origin H10N7 IAV to replicate in different cells, we infected Madin-Darby canine kidney (MDCK), chicken embryo fibroblast (DF-1), and adenocarcinomic human alveolar basal epithelial (A549) cells with Ig/4266 virus and three other representative H10 isolates [A/seal/Netherlands/P14-221/2014(H10N7) (abbreviated as Seal/221); A/mink/Sweden/E12665/1984(H10N4) (Mink/E12665); and A/chicken/Jiangxi/34609/2013(H10N8) (Ck/34609)] at a multiplicity of infection of 0.01. All four testing viruses replicated efficiently in MDCK, A549, and DF-1 cells, reaching peak virus titers at 48 or 72 hours after inoculation (Fig. 3). In general, these viruses grew better in MDCK cells than in the other cell lines but showed similar growth patterns in MDCK and DF-1 cells. However, in A549 cells, the mink-origin H10N4 virus had a higher virus titer than the other three viruses.

FIG 3.

Replication kinetics of four subtype H10 influenza A viruses isolates in vitro. Growth curves for influenza A/Iceland gull/Iceland/4266/2015(H10N7) virus (Ig/4266), A/chicken/Jiangxi/34609/2013(H10N8) (Ck/34609) virus, A/mink/Sweden/E12665/84(H10N4) virus (Mink/E12665), and A/seal/Netherlands/P14-221/2014(H10N7) virus (Seal/221) in Madin-Darby canine kidney (MDCK), adenocarcinomic human alveolar basal epithelial (A549), and chicken embryo fibroblast (DF-1) cells. Cells were infected with viruses at a multiplicity of infection of 0.01; dashed lines indicate the limit of virus detection, 1.50 log10 (50% tissue culture infective dose/ml [TCID50/ml]). Virus titers are expressed as mean ± standard deviation of log₁₀ TCID₅₀/ml.

Gull-origin H10 IAV had strong binding affinity to alpha 2,6-linked sialic acids and alpha 2,3-linked sialic acids.

To understand the receptor binding properties of these isolates, we tested their receptor binding affinities to two glycan analogs (i.e., Neu5Acα2-3Galβ1-4GlcNAcβ [3ʹSLN, an avian-like receptor] and Neu5Acα2-6Galβ1-4GlcNAcβ [6ʹSLN, a human-like receptor]) for eight representative H10 viruses from various hosts, namely, gull, seal, mink, duck, and chicken, and two prototype influenza viruses, A/California/01/2009(H1N1) and A/Duck/Hunan/795/2002 (HA, NA) × A/PR/8/34(H5N1) (Fig. S1). All of the 8 tested H10 viruses had binding affinities to 3ʹSLN at a sugar-loading concentration of 0.5 μg/ml; however, at the same sugar-loading concentration, isolates from all hosts also showed strong binding responses to 6ʹSLN.

To further determine the binding preference of testing IAV’ (i.e., to 3ʹSLN or 6ʹSLN), we quantified and compared the 50% relative sugar loading (RSL) concentration at half the fractional saturation (f = 0.5) (RSL_0.5) of the tested virus against 3ʹSLN and 6ʹSLN. The higher the RSL_0.5, the smaller the binding affinity. Quantitative analyses showed that Ig/4266 virus had an RSL_0.5 of 0.0835 (±0.0072 standard deviation [SD]) for 3ʹSLN and 0.2917 (±0.0019) for 6ʹSLN, whereas Ck/34609 virus had an RSL_0.5 of 0.0996 (±0.0178) for 3ʹSLN and 0.3398 (±0.0004) for 6ʹSLN.

As expected, our results showed that A/California/01/2009(H1N1) showed binding affinities only to 6ʹSLN (RSL_0.5 of 0.1076 ± 0.0097) but not to 3ʹSLN, whereas A/duck/Hunan/795/2002 (HA, NA) × A/PR/8/34 (H5N1) showed binding affinity to 3ʹSLN (RSL_0.5 of 0.07822 ± 0.0068) but not to 6ʹSLN (Fig. 4).

FIG 4.

Glycan binding specificity of two subtype H10 influenza A viruses to (A) biotinylated α2,3-linked sialic acid (3ʹSLN) and (B) α2,6-linked sialic acid (6ʹSLN) glycan analogs as determined by biolayer interferometry using an Octet RED instrument (Pall FortéBio, Fremont, CA, USA). Streptavidin-coated biosensors were immobilized with biotinylated glycans at different levels. Sugar-loading-dependent binding signals were captured in the association step and normalized to the same background. Binding curves were fitted by using the saturation binding method in GraphPad Prism 7. Horizontal dashed line indicates half of the fractional saturation (f = 0.5); vertical dashed line indicates relative sugar loading (RSL_0.5) at f = 0.5; the higher the RSL_0.5, the smaller the binding affinity. Ig/4266, A/Iceland gull/Iceland/4266/2015(H10N7); Ck/34609, A/chicken/Jiangxi/34609/2013 (H10N8); H1N1, A/California/04/2009(H1N1); and H5N1, A/duck/Hunan/795/2002(HA, NA) × A/PR/8/34 (H5N1).

Gull-origin H10 IAV caused infections in ferrets.

To determine the infectivity and pathogenesis of the gull-origin H10N7 IAV in mammals, we inoculated 6 seronegative ferrets (1 group of 3 ferrets where each ferret was cohoused with an uninoculated ferret [to examine contact transmission] and another group of 3 ferrets where each ferret was cohoused with an uninoculated ferret but not in direct contact [to examine aerosol transmission]; Table 1) with Ig/4266 virus. Three additional ferrets were maintained in a separate room to serve as controls. Two ferrets inoculated with Ig/4266 (from the three contact-transmission groups) and two control ferrets were euthanized and necropsied at 5 days postinoculation (dpi) to determine viral pathogenesis and tissue tropism. Virus shedding levels and seroconversion were determined for the remaining four ferrets.

TABLE 1.

Hemagglutination inhibition titers in serum samples from ferrets inoculated with or exposed to animals inoculated with influenza A/Iceland gull/Iceland/4266/2015(H10N7) virus^a

Ferret ID by group no. and virus exposure route (cage no.)	HI titer by dpi or dpe
	0	10	21
I
Inoculation
E4129 (4)	<10	NA	NA
E4130 (5)	<10	NA	NA
E4132 (6)	<10	80	80
Direct contact
E4133 (4)	<10	<10	40
E4136 (5)	<10	<10	<10
E4137 (6)	<10	160	160
II
Inoculation
E4126 (1)	<10	160	160
E4127 (2)	<10	160	320
E4128 (3)	<10	160	160
Aerosol
E4138 (1)	<10	<10	<10
E4139 (2)	<10	320	160
E4140 (3)	<10	<10	<10
III
Control
E4141 (7)	<10	NA	NA
E4142 (7)	<10	NA	NA
E4143 (8)	<10	<10	<10

^aHI, hemagglutination inhibition; dpe, days postexposure; dpi, days postinoculation; ID, identification; NA, not available because ferrets were euthanized for pathogenicity study. Hemagglutination inhibition assays were performed using turkey erythrocytes.

During the entire experimental period, influenza-like clinical signs (i.e., sneeze) were observed in neither the ferrets inoculated with Ig/4266 nor those exposed to Ig/4266 through direct contact or aerosol droplets. Compared with ferrets in the control group, the results from body temperature monitoring through implant transponder sensor (BioMedic Data Systems, Inc., Seaford, DE) did not show significant consistent temperature elevation in the ferrets inoculated with Ig/4266 or in the ferrets exposed to Ig/4266 through both direct contacts and aerosol droplets (data not shown). The results from body weight monitoring did not show weight loss in all three control ferrets and five out of six ferrets inoculated with Ig/4266 (one ferret [number E4128] inoculated with Ig/4266 had a body weight loss [∼5%] on 3 dpi recovered quickly on 5 dpi) (data not shown). The body weights in the transmission ferrets were not monitored to avoid any potential contaminations.

All four remaining inoculated ferrets seroconverted by 10 dpi; at 21 dpi, HI titers ranged from 1:80 to 1:320 (Table 1). Virus titration in MDCK cells suggested that all six inoculated ferrets shed virus from 3 dpi to 5 dpi and had peak titers of up to 5.00 log₁₀ 50% tissue culture infective dose (TCID₅₀)/ml at 3 dpi (Fig. 5). To determine the tissue tropism of Ig/4266 virus, we quantified viral titers at 5 dpi in 10 tissue sections from each of the two ferrets’ respiratory tracts; tissues sections were from the turbinate, soft palate, upper and distal trachea, and lung (i.e., left cranial and caudal lungs and right cranial, caudal, middle lungs, and right accessory lobes). Except for the left cranial and left caudal lung sections, all tissue samples had H10 IAV-positive titers ranging from 3.75 to 6.25 log₁₀ TCID₅₀/g of tissue. Turbinate sections had an average titer of 5.87 log₁₀ TCID₅₀/g of tissue, whereas accessory sections had an average titer of 3.88 log₁₀ TCID₅₀/g of tissue (Fig. 6). Immunohistochemistry staining of the H10-positive tissue sections showed that IAV-specific antigen was multifocally presented within epithelium cells of the nasal turbinate, trachea, and lung (Fig. 7), a finding consistent with virus titration results (Fig. 6; see Table S3 in the supplemental material). In summary, the average viral loads in turbinate were significantly higher than those in lungs (P < 0.05), and the turbinate has the highest viral loads among all the tested tissues, namely, soft palates (4.63 ± 0.37 SD log₁₀ TCID₅₀/ml), turbinate (5.88 ± 0.38 log₁₀ TCID₅₀/ml), trachea (upper and distal trachea) (4.69 ± 0.31 log₁₀ TCID₅₀/ml), and lung (left cranial and caudal lungs, and right cranial, caudal, middle lungs, and right accessory lobes) (3.25 ± 0.38 log₁₀ TCID₅₀/ml).

FIG 5.

Mean titers of influenza viruses recovered from nasal wash fluids of virus-inoculated and contact ferrets in transmission experiments. Ferrets (n = 3 per group/experiment) were inoculated with 10⁶ 50% tissue culture infectious doses/ml (TCID₅₀/ml) of influenza A/Iceland gull/Iceland/4266/2015(H10N7) virus. Twenty-four hours later, naive ferrets (n = 3) were each randomly paired with an inoculated ferret and either housed in same cages or on a different side of a 1-cm-thick, double-layered, steel partition with 5-mm perforations (A, B, and C represent each direct contact transmission group [white bars, inoculated ferrets; gray bars, naive ferrets]; and D, E, and F represent each aerosol transmission group [white bars, inoculated ferrets; black bars, naive ferrets]). Nasal wash fluids were collected on the indicated days after inoculation or exposure for virus quantification using endpoint titration in Madin-Darby canine kidney cells; ending titers were expressed as log₁₀TCID₅₀/ml. Each panel represents a set of paired ferrets. Dashed lines indicate the limit of detection, log₁₀10^1.5 TCID₅₀/ml. Ferrets represented by the individual panels correspond to those listed in Table 1, namely, (A) ferrets in cage 4, (B) ferrets in cage 5, (C) ferrets in cage 6, (D) ferrets in cage 1, (E) ferrets in cage 2, and (F) ferrets in cage 3.

FIG 6.

Mean titers of influenza virus recovered from respiratory tract tissues of ferrets nasally inoculated with 10⁶ 50% tissue culture infectious doses (TCID₅₀) of influenza A/Iceland gull/Iceland/4266/2015(H10N7) virus. Two ferrets were euthanized 5 days postinoculation (dpi), and virus titers in the respiratory tissues of each ferret were determined by using endpoint titration in Madin-Darby canine kidney cells. The results shown are log10 TCID₅₀/gram. Results for tissues from the control group were negative (data not shown). Abbreviations: SP, soft palate; TR-U, upper trachea; TR-D, distal trachea; LCR, left cranial lung; LCD, left caudal lung; RCR, right cranial lung; RCD, right caudal; RMD, right middle lung; RA, right accessory lobes. Dashed line indicates the limit of detection, 1.50 log₁₀TCID₅₀/ml.

FIG 7.

Pathogenic changes in respiratory tract tissues of ferrets inoculated with influenza A/Iceland gull/Iceland/4266/2015(H10N7) virus at 5 days postinoculation (dpi). Immunohistochemistry staining showed the presence of H10 antigen in respiratory epithelium cells within nasal turbinate (A, brown staining), ciliated epithelial cells within the trachea (B, brown staining), and scattered ciliated epithelial cells in bronchioles tissue were immunoreactive (C, arrows). (D to G) Hematoxylin and eosin-stained ethmoid turbinate sections at 5 dpi in control (D, F) and infected (E, G) ferrets. Ethmoid turbinates from virus-inoculated ferrets had intraluminal aggregates of degenerate neutrophils, macrophages, and cellular debris. In addition, there are scattered neutrophils within the epithelium (G, arrows). Bars = 20 μm/100 μm (in D, E).

To further demonstrate the pathogenesis of Ig/4266 in ferrets, we performed hematoxylin and eosin staining on tissue sections from all locations. Ferrets inoculated with Ig/4266 virus showed moderate inflammatory changes within the ethmoid turbinates with scattered to small clusters of neutrophils within the epithelium and luminal aggregates of degenerate neutrophils, macrophages, cellular debris, and occasional hemorrhage (Fig. 7B and D). One inoculated ferret (E4130) had rare areas of luminal neutrophils and cellular debris within terminal bronchioles (data not shown).

Gull-origin H10 IAV is transmitted by direct contact and by exposure to infectious aerosol droplets.

To determine the transmission ability of Ig/4266 virus through direct contact, we used three sets of ferrets, each comprising a subtype H10 IAV-seronegative contact ferret and an Ig/4266 virus-inoculated ferret (6 ferrets total) (Table 1, group I). The three sets of ferrets were separately housed in partition-free cages. At 21 days post exposure (dpe), two of the three contact ferrets had seroconverted (HI titers of 1:160 and 1:40, respectively) (Table 1). At 5 and 7 dpe, one of the seropositive contacts had a peak virus titer of 3.50 log₁₀ TCID₅₀/ml in nasal wash fluids (Fig. 5C); virus was not detected in nasal wash fluids from the other two contact ferrets (Fig. 5A and B). All three virus-inoculated ferrets had detectable virus titers in their nasal wash fluids (range, 3.00 to 5.00 log₁₀ TCID₅₀/ml). By integrating data from both virological and serological analyses, two out of three direct contact-exposed ferrets became infected with Ig/4266.

We also used three sets of ferrets to determine the transmission ability of Ig/4266 virus through aerosol droplets; each set comprised a subtype H10 IAV-seronegative ferret and an Ig/4266 virus-inoculated ferret (6 ferrets total) (Table 1, group II). The two ferrets in each set were housed in the same cage but separated by a 1-cm-thick, double-layered steel partition with 5-mm perforations. Nonrecirculating air in the cages flowed from the side housing the inoculated ferret through the partition to the side housing the exposure ferret and exhausted to room air through HEPA filtration. All three virus-inoculated ferrets had detectable virus titers in their nasal wash fluids (range, 3.00 to 5.00 log₁₀ TCID₅₀/ml). By 10 dpe, one of the three aerosol-exposed ferrets had seroconverted (HI titer of 1:160) (Table 1), and the same ferret shed virus from 3 to 7 dpe; peak titers (5.50 log₁₀ TCID₅₀/ml) were detected at 5 dpe (Fig. 5E). No virus was detected in nasal wash fluids from the other two aerosol-exposed ferrets (Fig. 5D and F). By integrating data from virological and serological analyses, one out of three aerosol-exposed ferrets became infected with Ig/4266.

A total of three control ferrets (Table 1, group III) were housed in two cages in a separate room from the experimental ferrets. Virus titration of collected tissues, nasal wash fluids, immunohistochemical staining, and serologic assays remained negative for the control ferrets.

DISCUSSION

In this study, we isolated four H10 IAVs from two gull species in Iceland, Iceland gulls and glaucous gulls. Genomic sequence analyses suggested that these isolates were genetically related to Eurasia lineage H10 IAVs, including those that caused the 2014 outbreak among European seals (16, 22, 23). Although genetic connections have been reported among avian IAVs isolated from North America and Iceland (22), none of the gene segments among the four H10 IAVs we isolated for this study were associated with those from North America (Fig. S1). However, in previous studies, both Eurasian and American lineages of AIV have been reported in Iceland, including an H10 subtype of mixed Eurasian and American lineage (22, 23).

Human IAVs preferentially bind to α2,6-linked sialic acids (SAα2,6GA), whereas avian IAVs prefer α2,3-linked sialic acids (SAα2,3GA) (24–26). In this study, biolayer interferometry analyses on the gull-origin H10N7 isolate we recovered (Ig/4266 virus) showed that the isolate bound to both SAα2,3GA and SAα2,6GA, as also shown with three other H10 isolates from chicken, seal, and mink (Fig. S2). These results are consistent with other studies that also used biolayer interferometry analyses to determine binding affinity for avian-origin H10 viruses (27), and they are consistent with results for a study, using turkey erythrocytes expressing both SAα2,3GA and SAα2,6GA or only SAα2,6GA receptors, of seal-origin H10 IAVs that showed significant binding ability to SAα2,3GA and SAα2,6GA (19). However, our results showed that the binding affinities to SAα2,6GA could vary to some extent, depending on the virus strain, but sequence analyses for the reported HA receptor binding sites were mostly conserved among the isolates (Table S1). Of note, a mutation, Q220L (corresponding to 226 in H3 numbering), which was reported to have been acquired in the 220 loop of the HA receptor binding site during the subtype H10N7 IAV outbreaks among European seals, increased replication ability in mammalian cells (A549 cells) and retained replication efficiency in the avian cells (chicken embryo kidney primary cells) (19). This mutation seems to increase the binding affinity of avian-origin H10N7 IAV to SAα2,6GA (19). Mutations Q226L and G228S (in position H3) can increase the ability of avian IAV to infect humans (24, 28–33). Our results suggest that replication of mink-origin H10N4 virus in mammalian cell line A549 and MDCK cells (but not in avian DF-1 cells) was more efficient than that of gull-origin H10 IAVs, although receptor binding properties of these H10 viruses were similar. It is unclear whether the NA or internal genes of mink virus could also contribute to such phenotypic differences in growth analyses (Fig. 3). Nevertheless, the binding ability of the gull-origin H10 IAVs to SAα2,6GA could facilitate viral infection of mammal hosts (e.g., ferrets and humans) expressing SAα2,6GA and facilitate the generation of a more transmissible virus.

Antigenic analyses showed a lack of significant antigenic diversity among 32 H10 IAVs that we tested (Fig. 2); this finding is similar to that from a prior study, in which the antigenic properties of 20 IAVs were compared (34). Sequence analyses showed that the amino acids across the HA antibody binding sites were conserved (data not shown). The limited antigenic diversity for H10 avian IAVs is consistent with the concept of evolutionary stasis in natural reservoirs of IAVs (i.e., aquatic birds, such as migratory waterfowl and shorebirds) (35). These results are consistent with those from other studies of other subtypes of waterfowl-origin IAVs, such as subtype H3 (36) and H7 viruses (37–39).

As one of the key natural reservoirs of IAVs, gulls play an important role in transmission and dissemination of various subtypes of IAVs, including subtypes H5 and H7 (40). Prior studies suggested that gull species seem to be involved in transportation of avian IAVs in the North Atlantic and in the reassortment of these viruses into new genetic combinations (5, 41). The habitat variety of gulls certainly create the possibility of acquiring a variety of IAVs from multiple sources because they are consummate generalists and can be found in practically any habitat, including coastal, marine, brackish, freshwater, agricultural, and even human urban and suburban environments. Glycan receptor characterization showed that gulls (herring gull, laughing gull, and ring-billed gull) had abundant expression of α2,3-linked sialic acid in the respiratory tract (nasal turbinate, trachea, bronchi, and lung) and intestinal tract (duodenum/jejunum and ileum/ceca). In addition, α2,6-linked sialic acid was abundantly expressed in respiratory tracts of gulls; however, little or no expression of α2,6-linked sialic acid was detected in the intestinal tracts of gulls (42). Previous reports indicated that gulls, when infected with highly pathogenic avian IAV, exhibited a variety of clinical and pathological phenotypes; some IAV strains caused severe disease and even death, whereas other strains caused mild lesions and no deaths (43–47). Gulls experimentally infected with low-pathogenic avian IAVs showed no signs of disease, and none died (48–50). Most of these infected gulls shed virus from the oropharynx and cloaca (44–50). Of note, a gull-origin H16N3 IAV was shown to attach to the human respiratory tract, cornea, and conjunctiva, and a human virus was shown to bind to the trachea of gull species tested (51). In summary, prior studies suggest that gulls could be infected with avian or human IAVs and serve as asymptomatic carriers facilitating virus transmission. In this study, the Eurasian lineage H10 viruses, which were associated with outbreaks among seals (Fig. 1), have been identified in at least two gull species in Iceland. Thus, gulls could facilitate spillover of viruses to other hosts within the IAV ecologic environment and to other geographic regions, including North America.

Ferrets have been used as a model of human infection in this study as their physiology and immune system and the influenza-associated glycan receptor distribution in their respiratory tracts are similar to those in humans (52, 53); thus, ferrets have been used widely as a model of influenza infection in humans for influenza risk assessment (e.g., pathogenesis and virus transmission) and to evaluate influenza vaccines (54–57). Human seasonal influenza virus infections in ferrets primarily occur in the upper respiratory tract (58, 59). Instead, avian IAVs (e.g., subtype H5N1) can cause upper respiratory tract and lower respiratory tract manifestations in humans (60). Under laboratory conditions, many avian IAVs, including subtype H5N1 and H7N9, can cause infections in both upper and lower respiratory tracts in ferrets (61, 62). However, most of these laboratory studies in ferrets involve in a high dose (e.g., 10⁵ to 10⁷ PFU, TCID₅₀, or 50% egg infectious doses) usually through intranasal or intratracheal inoculation. Recent studies suggest that the method of virus administration and dosage in the challenges affect the pathogenesis and the tissues where the challenging virus can be recovered (62–65). Human seasonal and pandemic influenza viruses typically can be transmitted efficiently between ferrets through direct contact or aerosol droplets. In our study, gull-origin subtype H10N7 IAV replicated primarily in the upper respiratory tract of ferrets, although some virus was also identified in the lower respiratory tracts of ferrets inoculated with H10N7 virus (Fig. 6; Table S3). In addition, H10N7 virus infection caused moderate histologic changes in ferrets (Fig. 7). In a prior study, pathogenesis was evaluated in ferrets that had been infected with 1 of 20 subtype H10 IAVs from different host species in Eurasia and North America, namely, domestic poultry (chicken, turkey, duck, and quail), wild birds (mallards, ruddy turnstone, northern pintail, American black duck, Eurasian wigeon, and unidentified shorebird species), and mammals (mink, seal, and human) (34). Results showed that the 20 H10 IAVs had robust viral replication in the nasal turbinate and various degrees of replication in the lungs (34); those findings are consistent with our findings for gull-origin H10N7 virus. Furthermore, gull H10N7 virus can be transmitted between ferrets through both direct contact and aerosol droplets, although aerosol transmission is less efficient than direct contact transmission (Table 1). Our findings suggest that the gull-origin H10 IAVs have the potential to infect mammals, including humans, and possibly even result in human-to-human transmission.

Evaluation of the replication and transmission ability of avian-origin IAVs in mammals is a key component of influenza pandemic preparedness. This study supported the preparedness component, showing that, similar to subtype H1, H3, H5, H6, H7, and H9 avian IAVs, subtype H10 avian IAVs can infect mammals, including humans (6–8), minks (9), seals (10), and pigs (11). In addition, our findings suggest that avian H10 IAVs could potentially be transmitted among mammals through direct contact, aerosol droplets, or both, as was confirmed by the outbreaks among European seals (16, 22, 23). Thus, monitoring the enzootic dynamics and evolutionary patterns of H10 IAVs among wildlife, including gulls, would facilitate influenza pandemic preparedness, and continuing surveillance of avian IAVs at the wildlife-domestic animal interface and the animal-human interface is warranted.

MATERIALS AND METHODS

Field sampling.

Gulls were live-captured at Sandgerdi, Iceland (64.041°N, −22.714°W), during 10 November to 18 November 2015, using a cannon-propelled capture net (approximately 18 m long by 12 m wide) as previously described (23). Briefly, captured birds were marked with individual metal bands, identified to species, and ages were determined by plumage characteristics (66). Birds were sampled using a combined oral and cloacal swab where the same swab was first used to swab the oral cavity and then the cloaca. Each swab sample was immediately placed in an individual 2.0-ml cryovial containing 1.25 ml of virus transport medium (23). Vials were held on ice for up to 5 hours prior to being stored in liquid nitrogen or liquid nitrogen vapor. Samples were shipped by private courier from Iceland to Madison, WI, USA, on dry ice in biosecure containers. Once received in the laboratory, samples were stored at −80°C until analyzed.

Detection of IAV-positive samples and virus isolation.

Viral RNA was extracted using the MagMAX-96 avian influenza/Newcastle disease (AI/ND) viral RNA isolation kit (Applied Biosystems Foster City, CA, USA) following the manufacturer’s procedures. Real-time reverse transcriptase PCR (RT-PCR) was performed using previously published procedures, primers, and probe designed to detect the IAV M gene (67). All samples exhibiting a positive cycle threshold from RT-PCR analysis were subjected to virus isolation in 9- to 11-day-old embryonated chicken eggs (68).

Cells.

MDCK cells, A549 cells, and DF-1 cells were obtained from American Type Culture Collection (Manassas, VA, USA). The cells were maintained until use at 37°C under 5% CO₂ in Dulbecco’s modified Eagle medium (Gibco DMEM; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA, USA).

Viruses and virus propagation.

Viral isolates were inoculated in 9-day-old specific pathogen free (SPF) chicken embryonated eggs and incubated at 37°C for 72 hours. Viruses used to generate HI results for antigenic cartography (Fig. 2) include Ck/34609, Seal/221, A/mallard/Northlands/1/2014 (H10N7), A/duck/Bangladesh/24035/2014 (H10N1), A/laughing gull/DE/209/2013 (H10N8), A/duck/Hong Kong/562/1979 (H10N9), Mink/E12665, A/mallard/Netherlands/10240/2002 (H10N7), A/duck/Bangladesh/821/2009 (H10N7), A/duck/Bangladesh/824/2009 (H10N7), A/chicken/Bangladesh/842/2009 (H10N7), A/duck/Bangladesh/8987/2010 (H10N2), A/duck/Bangladesh/8988/2010 (H10N9), A/chicken/Queensland/1/2012 (H10N7), rg-A/Jiangxi-Donghu/346/2013[R] (6 + 1 + 1) inactivated (H10N8), A/ruddy turnstone/King Island/7104CP/2014 (H10N8), A/northern shoveler/CA/27943/2007 (H10N7), A/ruddy turnstone/DE/579/2008 (H10N7), A/shorebird/DE/327/2009 (H10N1), A/ruddy turnstone/DE/6200/2009 (H10N7), A/American coot/MS/09OS615/2009 (H10N3), A/American black duck/New Brunswick/00971/2010 (H10N6), A/northern shoveler/AR/12OS160/2012 (H10N3), A/mallard/MS/12OS443/2012 (H10N1), A/mallard/OH/13OS1979/2013 (H10N7), A/red knot/DE/420/2013 (H10N7), A/spot-billed duck/Alberta/308/2016 (H10NX), and A/ruddy turnstone/DE/409/2016 (H10N5). Viruses used in growth kinetics analyses, glycan receptor binding analyses, and animal studies include Gg/4552, Ig/4402, Gg/4270, Ig/4266, Seal/221, Mink/E12665, A/American black duck/New Brunswick/00971/2010(H10N6), Ck/34609, A/California/01/2009(H1N1), and A/duck/Hunan/795/2002(HA, NA) × A/Puerto Rico/8/34 (H5N1).

Ferret antisera.

Ferret antisera used in HI assays to generate results for antigenic cartography (Fig. 2) include those raised against A/chicken/Jiangxi/34609/2013 (H10N8), A/laughing gull/DE/209/2013 (H10N8), A/glaucous gull/Iceland/4552/2015 (H10N7), A/seal/Netherlands/P14-221/2014 (H10N7), A/mallard/Netherlands/1/2014 (H10N7), A/duck/Bangladesh/24035/2015 (H10N1), and A/chicken/Germany/N/49 (H10N7).

RNA extraction and genomic sequencing.

Viral RNA was extracted by using the KingFisher pure viral NA kit (Thermo Fisher Scientific, Asheville, NC, USA) according to the manufacturer’s instructions. Influenza amplicons were obtained from the 2-step RT-PCR amplification and sequenced using an in-house influenza-targeted sequencing procedure using influenza primers and PCR amplification conditions adapted from previous studies (69, 70). A Nextera XT DNA library preparation kit was utilized to prepare the next-generation sequencing (NGS) libraries prior to sequencing with the MiSeq reagent v3 600 cycles kit with a MiSeq sequencing system (all from Illumina, San Diego, CA, USA) according to the manufacturer’s instructions (37). The quality of paired-end reads obtained from MiSeq sequencing was checked by FastQC (Babraham Bioinformatics, https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and trimmed by Trimmomatic v0.36 (Usadel Lab, http://www.usadellab.org/cms/?page=trimmomatic) using a quality score threshold of 20. After quality trimming was performed, all paired-end reads were aligned to the reference genome of a corresponding virus by using Bowtie 2 (Johns Hopkins University, http://bowtie-bio.sourceforge.net/bowtie2/index.shtml).

In addition to the viruses obtained from viral isolation, the viruses propagated for phenotype analyses were also sequenced to ensure no additional mutations were acquired during virus propagation before they were used in animal experiments and glycan binding.

Sequence alignment and phylogenetic analysis.

Multiple sequence alignments were generated using Muscle v3.8.31 (71). The phylogenetic tree for each gene segment was inferred by using a maximum-likelihood method by running RAxML v8.2.9 and by using a Gamma model of rate heterogeneity and a generalized time-reversible substitution model (72). Phylogenetic trees were visualized by using FigTree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/).

Virus titration.

To quantify virus titer, TCID₅₀ was determined on MDCK cells. Briefly, MDCK cells were maintained as stock cultures in Dulbecco’s modified Eagle medium (DMEM) and replated 1 day before infection in 96-well plates for TCID₅₀ assays. Treated samples and their paired controls were thawed and immediately serially diluted. Cell cultures were then infected for 1 h after three times of phosphate-buffered saline (PBS) (pH 7.4) washes. The medium for dilution and infections was Opti-MEM I reduced serum medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 100 U/ml Gibco penicillin-streptomycin (Thermo Fisher Scientific, Waltham, MA, USA). Postinfection TCID₅₀ cultures were washed and fed with Opti-MEM medium supplemented with 100 U/ml Gibco penicillin-streptomycin and 1 μg/ml tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin from bovine pancreas (Sigma-Aldrich, St. Louis, MO, USA). After 72 hours of incubation postinfection, viral titer of supernatant from 96-well plates was determined by hemagglutination using 0.5% turkey erythrocytes. Results were recorded, and TCID₅₀s were calculated by the Spearman-Karber algorithm (73).

Hemagglutination and HI assays.

Hemagglutination and HI assays were performed using 0.5% turkey erythrocytes as previously described (74).

Virus purification.

The allantoic fluids were harvested and subjected to centrifugation at 1,200 × g for 30 minutes at 4°C. The viruses were then purified by ultracentrifugation at 45,000 × g for 3 hours at 4°C. The pellets were dissolved with 1 ml of phosphate-buffered saline (PBS) (pH 7.4) and subjected to sucrose gradient (30%/60%) centrifugation at 103,000 × g for 1.5 hours at 4°C. The viruses in the intermediate phase were collected and subjected to centrifugation at 103,000 × g for 1 hour at 4°C. The pellets were dissolved in 0.5 ml of PBS and stored at −80°C until use.

Quantification of viral concentration.

The purified viruses were quantified by using the solid-phase enzyme-linked immunosorbent assay (ELISA) method as described elsewhere (75). In brief, ELISA was performed for each testing virus by mouse monoclonal anti-NP antibody (NR-43899; BEI Resources, National Institute of Allergy and Infectious Disease [NIAID], National Institutes of Health [NIH], Bethesda, MD, USA) against goat anti-mouse IgG horseradish peroxidase (HRP) conjugate (Sigma-Aldrich, St. Louis, MO, USA) and 1-Step ultra TMB (3,3',5,5'-tetramethylbenzidine)-ELISA substrate solution (Thermo Fisher, Rockford, IL).

Biolayer interferometry receptor binding assay and data analyses.

Virus receptor affinities were determined by a biolayer interferometry receptor binding assay using an Octet RED instrument (Pall FortéBio, Fremont, CA, USA). To test the potential binding affinity to human- and avian-like receptors, we used two biotinylated glycan analogs, 3ʹSLN and 6ʹSLN (Lectinity Holdings, Moscow, Russia). The glycans were preloaded onto streptavidin-coated biosensors at up to 0.5 μg/ml for 3 minutes in 1× kinetic buffer (Pall FortéBio, Menlo Park, CA, USA). Each test virus was diluted to a final concentration of 100 pM with 1× kinetic buffer containing 10 μM oseltamivir carboxylate (American Radiolabeled Chemicals, St. Louis, MO, USA) and zanamivir (Sigma-Aldrich, St. Louis, MO, USA) to prevent cleavage of the receptor analogs by NA proteins from virus. Association was measured for 30 to 50 minutes at 25°C as described elsewhere (75–78).

Responses were normalized by the highest value obtained during the experiment, and binding curves were fitted by using the binding-saturation method in GraphPad Prism 7. The normalized response curves report the fractional saturation (f) of the sensor surface, as described in a previous study (79). RSL_0.5 was used to quantitate the binding affinity of two selected viruses against two glycan analogs. The higher the RSL_0.5, the weaker the binding affinity.

Virus growth kinetics in vitro.

To determine the replication of the viruses in different cell lines, we infected A549, MDCK, and DF-1 cells with viruses at a multiplicity of infection of 0.01. After absorption for 1 hour at 37°C, the cells were washed with PBS and incubated for 72 hours at 37°C in 5% CO₂ with Opti-MEM I reduced serum medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 1 μg/ml TPCK-treated trypsin from bovine pancreas (Sigma-Aldrich, St. Louis, MO, USA) and with 100 U/ml Gibco penicillin-streptomycin (Thermo Fisher Scientific). Supernatants were collected at 8, 24, 48, and 72 hours after infection and titrated using the TCID₅₀ in MDCK cells.

Animals.

Fifteen 6-month-old ferrets were purchased from Triple F Farms (Sayre, PA, USA). All ferrets were seronegative for IAV, as determined by HI assay.

Pathogenicity of H10N7 IAV from Iceland in ferrets.

We studied the pathogenesis of Iceland H10N7 virus (Ig/4266) in ferrets. Three inoculation-group ferrets were each intranasally administered 0.5 ml of 10⁶ TCID₅₀ of virus (0.25 ml/nose). At the same time, three control ferrets were administered 0.5 ml of PBS. At 5 days postinoculation (dpi), necropsies were performed on two infected and two control ferrets, and respiratory tissues were collected for virus titration in MDCK cells and for immunohistochemistry staining. Tissues were collected from the respiratory system (turbinate, soft palate; upper and distal trachea, left cranial and caudal lung, right cranial and caudal lung, right middle lung, and right accessory lobes).

Transmission study in ferret model.

We used 12 ferrets to evaluate the potential for contact transmission or aerosol transmission of Iceland H10N7 IAV among humans. At 0 dpi, 6 of the 12 ferrets were each intranasally administered 0.5 ml of 10⁶ TCID₅₀ of Ig/4266 virus. At 1 dpi, each of these ferrets was randomly paired with an H10 IAV-seronegative ferret for study of aerosol transmission (group I) and direct contact transmission (group II). Paired group I ferrets (i.e., inoculated and aerosol transmission ferrets) were housed in the same cage but separated by a 1-cm-thick, double-layered steel partition with 5-mm perforations (Allentown, Inc., Allentown, NJ). Paired group II ferrets (i.e., inoculated and contact ferrets) were housed in the same cage without a partition. Nasal wash fluids were collected on 1, 3, 5, 7, and 10 dpi/dpe, and virus titers were determined by TCID₅₀ in MDCK cells. Blood samples were obtained on 0, 10, and 21 dpi/dpe to test for seroconversion.

Immunohistochemical staining.

We performed immunohistochemical staining to confirm the presence of viral antigen in ferret tissues. Tissues were fixed in 10% neutral buffered formalin solution, paraffin embedded, and sectioned before being stained. The sections were deparaffinized by using 100% xylene, 50% ethanol in xylene, 100% ethanol, 95% ethanol, 70% ethanol, 50% ethanol, and distilled water. Antigen retrieval was performed by boiling in 1× antigen retrieval solution (Dako, Carpinteria, CA, USA) in a steamer for 20 minutes. Sections were washed with PBS with 0.5% Tween 20 (PBST) before endogenous peroxidase activity was quenched in 3% H₂O₂. Sections were then blocked with 10% normal goat serum (Invitrogen, Carlsbad, CA, USA) for 1 hour at room temperature before incubation with anti-NP monoclonal antibody (BEI Resources, NIAID, NIH) (80) for 20 hours at 4°C. Sections were treated with a biotinylated goat anti-mouse IgG polyclonal secondary antibody for 30 minutes before ABC reagent (Vectastain, Burlingame, CA, USA) was applied according to the manufacturer’s instructions. Sections were counterstained with hematoxylin, washed, dehydrated, and covered with coverslips.

Histologic tissue inflammation analysis.

Tissues collected from ferrets at 5 dpi were sectioned as previously described and then stained with hematoxylin and eosin. The following tissue sections were analyzed: turbinate, soft palate, upper and distal trachea, left cranial and caudal lung, right cranial and caudal lung, right middle lung, and right accessory lobes.

Statistical analysis.

Student’s t test was used to determine statistical difference between viral titers from different tissues, and an alpha of 0.05 was used as the criteria for significance.

Biosafety and animal handling.

Laboratory and animal experiments were conducted under biosafety level 2 (BSL2) conditions in compliance with protocols approved by the Institutional Animal Care and Use Committee of Mississippi State University or the U.S. Geological Survey National Wildlife Health Center.

Data availability.

Sequencing data of the four H10N7 isolates, Gg/4552, Ig4402, Gg/4270, and Ig/4266, have been deposited in the GenBank database under accession numbers MK100291 to MK100295 and MK100321 to MK100323 (Ig/4266), MK100296 to MK100303 (Ig4402), MK100304 to MK100311 (Gg/4270), and MK100312 to MK100319 (Gg/4552).