Novel Reassortant Human-Like H3N2 and H3N1 Influenza A Viruses Detected in Pigs Are Virulent and Antigenically Distinct from Swine Viruses Endemic to the United States

ABSTRACT

Human-like swine H3 influenza A viruses (IAV) were detected by the USDA surveillance system. We characterized two novel swine human-like H3N2 and H3N1 viruses with hemagglutinin (HA) genes similar to those in human seasonal H3 strains and internal genes closely related to those of 2009 H1N1 pandemic viruses. The H3N2 neuraminidase (NA) was of the contemporary human N2 lineage, while the H3N1 NA was of the classical swine N1 lineage. Both viruses were antigenically distant from swine H3 viruses that circulate in the United States and from swine vaccine strains and also showed antigenic drift from human seasonal H3N2 viruses. Their pathogenicity and transmission in pigs were compared to those of a human H3N2 virus with a common HA ancestry. Both swine human-like H3 viruses efficiently infected pigs and were transmitted to indirect contacts, whereas the human H3N2 virus did so much less efficiently. To evaluate the role of genes from the swine isolates in their pathogenesis, reverse genetics-generated reassortants between the swine human-like H3N1 virus and the seasonal human H3N2 virus were tested in pigs. The contribution of the gene segments to virulence was complex, with the swine HA and internal genes showing effects in vivo. The experimental infections indicate that these novel H3 viruses are virulent and can sustain onward transmission in pigs, and the naturally occurring mutations in the HA were associated with antigenic divergence from H3 IAV from humans and swine. Consequently, these viruses could have a significant impact on the swine industry if they were to cause more widespread outbreaks, and the potential risk of these emerging swine IAV to humans should be considered.

IMPORTANCE Pigs are important hosts in the evolution of influenza A viruses (IAV). Human-to-swine transmissions of IAV have resulted in the circulation of reassortant viruses containing human-origin genes in pigs, greatly contributing to the diversity of IAV in swine worldwide. New human-like H3N2 and H3N1 viruses that contain a mix of human and swine gene segments were recently detected by the USDA surveillance system. The human-like viruses efficiently infected pigs and resulted in onward airborne transmission, likely due to the multiple changes identified between human and swine H3 viruses. The human-like swine viruses are distinct from contemporary U.S. H3 swine viruses and from the strains used in swine vaccines, which could have a significant impact on the swine industry due to a lack of population immunity. Additionally, public health experts should consider an appropriate assessment of the risk of these emerging swine H3 viruses for the human population.

INTRODUCTION

Swine have a key role in the ecology of influenza A viruses (IAV) and thus represent a risk for future introductions of swine viruses into the human population. Similar to subtypes that circulate in humans, established swine IAV are of the H1N1, H3N2, and H1N2 subtypes (1), whereas other subtypes are only sporadically detected in swine as a result of interspecies transmission, such as avian-like H3N1 (2) and H2N3 (3) viruses or equine-like H3N8 viruses (4). The porcine respiratory tract contains both human IAV-preferred sialic acid α2,6-galactose (α2,6-Gal)-linked receptors and avian IAV-preferred sialic acid α2,3-Gal-linked receptors (5), providing an underlying biologic basis for swine being intermediary hosts in the evolution of influenza viruses. Unlike the relatively uncommon event of a swine lineage virus becoming established in the human population, human seasonal virus transmission events to swine have repeatedly led to new genetic lineages of novel viruses that became established in various pig populations around the globe (6). Human-origin surface genes have been maintained at a much higher frequency than the internal genes of the seeding virus once it enters a pig population (6), which suggests that barriers to the sustained circulation and efficient adaptation of wholly human viruses exist in swine.

A notable human-to-swine event occurred in the late 1990s when a triple-reassortant internal gene (TRIG) constellation containing swine (M, NP, and NS), avian (PB2 and PA), and human (PB1) influenza virus genes became established among North American swine (7, 8). This constellation of internal genes reassorted with different combinations of surface genes, and as a consequence, the dynamics of influenza virus infection in North American pigs changed drastically. Additionally, more than 49 independent human-to-swine spillover events of the 2009 pandemic H1N1 (H1N1pdm09) strain have occurred globally since it was introduced into the human population (9). These incursions led to multiple reassortment events between H1N1pdm09 and established swine IAV (1, 10), creating unique swine IAV genome configurations and increasing the observed genetic diversity. H1N1pdm09 highlights the pandemic risk of novel viruses generated through the exchange between human and swine virus lineages (11). Furthermore, antigenic drift in viral surface glycoproteins contributes to the evolution of swine IAV (12), resulting in the cocirculation of many antigenically distinct viruses in pigs (1, 13).

Novel H3N2 and H3N1 viruses with contemporary human seasonal H3 genes were identified through the United States Department of Agriculture (USDA) Swine Influenza A Virus Surveillance System. Even though H3N1 viruses have previously been detected in U.S. swine, they are rare (2, 14). The novel H3N1 viruses reported here have a unique combination of surface genes from contemporary human seasonal H3N2 hemagglutinin (HA) and classical swine H1N1 (cH1N1) neuraminidase (NA) and internal genes derived from H1N1pdm09 and hence are distinct from current swine H3 viruses circulating in the United States as well as the human seasonal H3 viruses circulating globally. To assess the impact of these novel H3 viruses, an in vitro genetic and antigenic characterization was conducted along with an in vivo phenotypic characterization. We demonstrate that these novel human-like IAV are virulent in swine and pose a significant threat to the swine population due to an expected lack of population immunity. To further understand the role of gene segments on the striking difference in pathogenesis and transmissibility of these viruses from those of a human seasonal H3N2 virus, we constructed reassortants between the swine human-like H3N1 and the human H3N2 viruses by reverse genetics (rg) and compared the pathogenesis in vivo. Our results suggest that the HA and internal gene constellation are essential for the efficient infection and transmission of the novel human-like H3 viruses.

MATERIALS AND METHODS

Ethics statement.

All animals were housed in biosafety level 2 (BSL2) containment facilities and cared for in compliance with the Animal Care and Use Committee of the National Animal Disease Center.

Viruses and cell lines.

The swine isolates A/Swine/Missouri/A01476459/2012 (H3N2; Sw/MO/12) and A/Swine/Missouri/A01410819/2014 (H3N1; Sw/MO/14) were obtained from the Swine IAV Surveillance System repository held at the USDA National Veterinary Service Laboratories in conjunction with the USDA National Animal Health Laboratory Network (NAHLN). The H3N2 virus was isolated from a breeding herd during the winter of 2012, and the H3N1 virus was isolated from an epidemiologically linked location during the winter of 2013. The HA and NA of the human H3N2 isolate A/Victoria/361/2011 (A/VIC/11; kindly provided by Richard Webby, St. Jude Children's Research Hospital) were genetically similar to the HAs of both swine isolates and the NA of Sw/MO/12, and A/VIC/11 was included as a control. Viruses were propagated in Madin-Darby canine kidney (MDCK) cells.

Reverse-engineered viruses.

The two wild-type viruses with in vivo phenotypes at the opposite ends of the range were chosen to generate reassortants to test the contribution of genes or combinations of genes. Eight viruses were generated by rg using an 8-plasmid system, as previously described (15), in the bidirectional plasmid vector pDP2002, and their genetic constellations are described in Table 1. Gene combinations were verified by full-length sequencing, and viruses were propagated in MDCK cells.

TABLE 1.

Viruses generated by rg using A/VIC/11 and Sw/MO/14 and used as challenge viruses in experiment 2^a

Virus	HA gene origin	NA gene origin	Internal gene origin
Sw/MO/14rg	Sw/MO/14	Sw/MO/14	Sw/MO/14
VIC11-HA/NA	A/VIC/11	A/VIC/11	Sw/MO/14
VIC11-HA	A/VIC/11	Sw/MO/14	Sw/MO/14
VIC11-NA	Sw/MO/14	A/VIC/11	Sw/MO/14
A/VIC/11rg	A/VIC/11	A/VIC/11	A/VIC/11
MO14-HA/NA	Sw/MO/14	Sw/MO/14	A/VIC/11
MO14-HA	Sw/MO/14	A/VIC/11	A/VIC/11
MO14-NA	A/VIC/11	Sw/MO/14	A/VIC/11

^aHA, hemagglutinin; NA, neuraminidase.

Genetic analysis.

Three genes (the HA, NA, and M genes) of the swine isolates were initially sequenced and submitted to GenBank by the submitting NAHLN veterinary diagnostic lab. Following the identification of the human-origin HA gene, 9 swine isolates were subjected to whole-genome next-generation sequencing using an Ion 316 (v2) chip and an Ion PGM 200 (v2) sequencing kit (Life Technologies, Carlsbad, CA) as previously described (16). The HA genes from viruses recovered from pigs directly challenged with virus (referred to throughout as primary pigs) and pigs with indirect contact with primary pigs (referred to throughout as indirect contact pigs) in the in vivo studies were sequenced directly from clinical material by conventional sequencing using a BigDye Terminator (v3.1) cycle sequencing kit (Applied Biosystems, Foster City, CA) per the manufacturer's instruction using previously described primers (17).

Additional representative sequences of North American swine and human viruses were downloaded from GenBank and the Global Initiative on Sharing All Influenza Data (GISAID). Specifically, using BLASTn (18) we identified 15 human isolates from the 2010-2011 influenza season with high HA gene sequence identity and also included randomly selected human isolates from each influenza season from 2008 to 2013. More recent swine human-like H3N1 and H3N2 viruses that were subsequently identified by the USDA surveillance system were also included in the analysis (see Tables S1 and S2 in the supplemental material). The sequences of each of the eight genomic segments were aligned using the default settings in MUSCLE (v3.8.31) software (19), with subsequent manual correction. For each alignment, we inferred the best-known maximum likelihood (ML) tree using the RAxML (v7.4.2) program (20), the rapid bootstrap algorithm, and a general time-reversible (GTR) model of nucleotide substitution with Γ-distributed rate variation among sites. Statistical support for individual branches was estimated by bootstrap analysis, with the number of bootstrap replicates being determined automatically using an extended majority-rule consensus tree criterion (21). The deduced HA1 domain amino acid sequences were aligned and used to identify amino acid differences between the human and the swine viruses.

Animal experiment 1.

Fifty 3-week-old crossbred healthy pigs were obtained from a herd free of IAV and porcine reproductive and respiratory syndrome virus (PRRSV). Prior to the start of the study, the pigs were treated with ceftiofur crystalline free acid and tulathromycin (Zoetis Animal Health, Florham Park, NJ) to reduce bacterial contaminants and were shown to be seronegative for IAV antibodies. Pigs were divided into four groups: nonchallenged (NC) control pigs (n = 5) and pigs challenged with A/VIC/11 H3N2 (n = 10), Sw/MO/12 H3N2 (n = 10), and Sw/MO/14 H3N1 (n = 10).

Challenged pigs were simultaneously inoculated intranasally (1 ml) and intratracheally (2 ml) with 10⁵ 50% tissue culture infective doses (TCID₅₀) per ml of each assigned virus. Inoculation was performed while the pigs were under anesthesia, obtained using an intramuscular injection of a cocktail of ketamine (8 mg/kg of body weight; Zoetis Animal Health, Florham Park, NJ), xylazine (4 mg/kg), and tiletamine-zolazepam (Telazol; 6 mg/kg) (Zoetis Animal Health, Florham Park, NJ). At 2 days postinfection (dpi), five contact pigs were placed in separated raised decks in the same room as each inoculated group to evaluate transmission by indirect contact. Nasal swabs (NS; FLOQSwabs; Copan Diagnostics, Murrieta, CA) were collected at 0, 1, 3, and 5 dpi from primary pigs and at 0 to 5, 7, and 9 days postcontact (dpc) from indirect contact pigs as previously described (22).

Two pigs died from causes unrelated to IAV infection, leaving 8 pigs in the A/VIC/11 group. At 5 dpi, primary pigs were humanely euthanized with a lethal dose of pentobarbital (Fatal Plus; Vortech Pharmaceuticals, Dearborn, MI) and necropsied, when bronchoalveolar lavage fluid (BALF) and tissue samples from the distal trachea and right cardiac or affected lung lobe were collected. Indirect contact pigs were humanely euthanized at 15 dpc for collection of serum to evaluate seroconversion.

Animal experiment 2.

To test the role of the surface genes and internal gene backbones observed in vivo with the wild-type Sw/MO/14 H3N1 virus, the reassortant viruses generated as described above were used in a second pathogenesis study. One hundred twenty-five 3-week-old crossbred healthy pigs obtained from the same herd from which the pigs used in experiment 1 were obtained were used in experiment 2. Groups of 10 pigs were infected with each of the reverse genetics-generated viruses using the same methodology described above, and 5 pigs serving as indirect contacts were introduced at 2 dpi as described above. Nasal swab samples were collected from primary and indirect contact pigs, and necropsies were performed by the same procedures described above for experiment 1.

Virus titers in NS and lungs.

For virus isolation, filtered NS samples were plated onto confluent MDCK cells as previously described (22). Tenfold serial dilutions in serum-free Opti-MEM medium (Gibco, Life Technologies, Carlsbad, CA) supplemented with 1 μg/ml tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-trypsin and antibiotics were prepared for each BALF sample and virus isolation-positive NS sample. Each dilution was plated in triplicate onto phosphate-buffered saline (PBS)-washed confluent MDCK cells in 96-well plates. At 48 h, the plates were fixed with 4% phosphate-buffered formalin and stained using immunocytochemistry as previously described (23). The virus titers (numbers of TCID₅₀ per milliliter) were calculated for each sample according to the method of Reed and Muench (24).

Pathological examination of lungs.

At necropsy, the lungs were removed and evaluated for the percentage of the lung affected by the purple-red consolidation typical of IAV infection. The percentage of the surface of the entire lung affected by pneumonia was calculated on the basis of the weighted proportions of each lobe to the total lung volume (25). Tissue samples from the trachea and lung were fixed in 10% buffered formalin and were routinely processed and stained with hematoxylin and eosin. Microscopic lesions were evaluated by a veterinary pathologist blinded to the treatment groups and scored according to previously described parameters (26). IAV-specific antigen was detected in the trachea and lung tissues using immunohistochemistry (IHC) and scored as previously described (26). Individual scores were summed, and the average group composite scores were used for statistical analysis.

Serology and antigenic cartography.

Two 7-week-old seronegative naive pigs were used for Sw/MO/14 H3N1 antiserum production. Pigs were immunized intramuscularly with 2 doses 2 weeks apart of Sw/MO/14 antigen inactivated by UV irradiation. The antigen was used at 128 hemagglutination units per 50 μl in PBS with a commercial oil-in-water adjuvant (Emulsigen D; MVP Laboratories, Inc., Ralston, NE) at a 1:5 ratio. The pigs were humanely euthanized as described above for blood collection. Prior to hemagglutination inhibition (HI) assays, sera were treated with receptor-destroying enzyme (Sigma-Aldrich, St. Louis, MO), heat inactivated at 56°C for 30 min, and adsorbed with 50% turkey red blood cells (RBCs) to remove nonspecific hemagglutinin inhibitors and natural serum agglutinins. HI assays with samples from the challenged primary and indirect contact pigs were performed with either A/VIC/11, Sw/MO/12, or Sw/MO/14 as the antigen and 0.5% turkey RBCs using standard techniques (27). Reciprocal titers were divided by 10 and log₂ transformed and are reported as the geometric mean.

Two-way HI assays against a reference swine antiserum panel (see Table S3 in the supplemental material) were performed as described above, using a panel of reference swine and human H3N2 viruses, including Sw/MO/12, Sw/MO/14, and A/VIC/11, as HI antigens (28). The reference panel represents H3 viruses historically or currently circulating in pigs in the United States, along with recent and historical representatives of human vaccine strains. The HI assay data and antigenic cartography were used to quantify the antigenic interrelationships between Sw/MO/12, Sw/MO/14, and other H3 isolates, as previously described (12, 29).

Statistical analysis.

The percentage of the lung with macroscopic lesions, microscopic lesion scores, and log₁₀-transformed virus titers in BALF and NS samples were analyzed using analysis of variance, with a P value of ≤0.05 being considered significant (GraphPad Prism [version 6] software; GraphPad Software, La Jolla, CA). Response variables shown to have significant effects by treatment group were subjected to pairwise mean comparisons using the Tukey-Kramer test.

RESULTS

Genetic characterization of the novel H3 viruses.

Phylogenetic analysis of the HA genes of the human-like H3N2 and H3N1 isolates Sw/MO/12 and Sw/MO/14 used in our study and other human-like H3N1 and H3N2 swine viruses identified in GenBank demonstrated that they were most closely related to human seasonal H3N2 strains from 2010-2011 (Fig. 1; see also Fig. S7 in the supplemental material), and they did not cluster with the contemporary circulating swine H3 virus genetic clades (30). The HA genes of the recent swine human-like H3 viruses clustered together in the phylogeny with human seasonal H3 viruses from 2010-2011, suggesting that these swine isolates were of similar ancestry and that the Sw/MO/12 isolate most likely evolved from a human seasonal virus that circulated during the 2010-2011 season. The NA phylogeny indicated that the NA gene of the initially identified human-like H3N2 swine virus (Sw/MO/12) was closely related to human N2 genes that circulated in 2010-2011, similar to the HA phylogeny (Fig. 2A; see also Fig. S8A in the supplemental material). However, the NAs of the more contemporary human-like H3 viruses were closely related to N1 of cH1N1 viruses or N2 of the swine 2002 N2 lineage (Fig. 2B; see also Fig. S8B in the supplemental material). The internal genes of five human-like H3 viruses (the first H3N2 virus and four H3N1 viruses) were all closely related to those of H1N1pdm09 viruses, and more recent human-like swine H3N2 viruses had a combination of internal genes from the TRIG lineage and the M gene of the H1N109pdm lineage (see Fig. S1 to S6 in the supplemental material). The sequences of the viruses recovered from two primary pigs and two indirect contact pigs of each infected group (when recoverable) were sequenced and compared to those of the viruses in the original inoculum to investigate whether amino acid changes occurred after animal passage, and no differences were found.

FIG 1.

Phylogenetic analysis of HA genes of the swine human-like H3 viruses. The maximum likelihood phylogeny of the HAs of 20 human-like H3 swine viruses and 155 H3N2 viruses collected from humans and swine in the United States is shown. The branch color reflects the evolutionary history, and the colors are defined in the key. Numbers above or below the branches indicate the percent bootstrap support; bootstrap support values of ≤50% are not shown. The tree is rooted at the midpoint for clarity, and all branch lengths are drawn to scale. The scale bar indicates the number of nucleotide substitutions per site. A phylogeny with taxon names indicating the viral isolate and the GenBank or GISAID EpiFlu accession number is presented in Fig. S7 in the supplemental material. C-I, C-II, and C-IV, clusters I, II, and IV, respectively.

FIG 2.

Phylogenetic analysis of NA genes of the swine human-like H3 viruses. The maximum likelihood phylogeny of the NAs of 20 human-like H3 viruses and 155 representative viruses collected from humans, swine, and turkeys in the United States is shown. (A) N2 influenza A virus isolates; (B) N1 influenza A virus isolates. Numbers above or below the branches indicate the percent bootstrap support; bootstrap support values of ≤50% are not shown. The trees are rooted at the midpoint for clarity, and all branch lengths are drawn to scale. The scale bar indicates the number of nucleotide substitutions per site. Phylogenies with taxon names indicating the viral isolate and the GenBank or GISAID EpiFlu accession number are presented in Fig. S8 in the supplemental material.

Pathogenesis of swine and human H3 viruses in pigs.

The A/VIC/11 virus did not cause significant macroscopic or microscopic lesions when the lungs of infected pigs were compared to those of noninfected pigs (Table 2). Significantly higher percentages of the lungs of pigs challenged with the swine viruses (Sw/MO/12 and Sw/MO/14) were affected by cranioventral consolidation than were the lungs of pigs challenged with A/VIC/11 (Table 2), with the lungs of Sw/MO/14-infected pigs showing the highest percentage of lesions.

TABLE 2.

Macroscopic pneumonia, lung and trachea microscopic pathology, and lung virus titers obtained in pigs challenged with wild-type viruses and NC controls^a

Challenge virus	Percentage with macroscopic pneumonia	Microscopic pneumonia scoreb	Microscopic tracheitis scorec	Log10 virus titer in BALF
NC control	0.0 ± 0.0A	0.2 ± 0.1A	0.1 ± 0.1A	0.0 ± 0.0A (0/5)d
A/VIC/11	0.0 ± 0.0A	0.8 ± 0.2A	0.8 ± 0.2A	0.6 ± 0.4A (2/8)
Sw/MO/12	4.2 ± 1.0B	4.9 ± 0.5B	2.8 ± 0.5B	3.6 ± 0.2B (10/10)
Sw/MO/14	12.0 ± 0.8C	9.4 ± 0.6C	2.1 ± 0.3B	5.1 ± 0.2C (10/10)

^aResults are shown as means ± standard errors of the means. Different capital letters within the same column indicate significant differences (P ≤ 0.05).

^bThe range of possible scores is 1 to 22.

^cThe range of possible scores is 1 to 8.

^dThe number of virus-positive pigs/total number of pigs tested is indicated in parentheses.

Microscopic lung lesions in the Sw/MO/14 group consisted of moderate to severe, lobular, and patchy to locally extensive interstitial pneumonia and moderately dense peribronchiolar cuffs that extended into the adjacent interstitium. Locally extensive alveolar lumina were expanded by large numbers of neutrophils and macrophages admixed with mild edema. Multifocal bronchi and bronchioles demonstrated moderate to severe epithelial attenuation and necrosis with infiltrates of neutrophils and occasional macrophages in the airway lumen. Pigs challenged with Sw/MO/12 showed less airway impairment than the group challenged with Sw/MO/14, but the pathology of the lungs of pigs challenged with Sw/MO/12 was consistent with that of uncomplicated influenza virus infection. In contrast, the lungs of pigs challenged with A/VIC/11 exhibited minimal, patchy interstitial pneumonia and mild and loosely formed peribronchiolar cuffs. The epithelial attenuation or necrosis in the trachea was mild to moderate in 3 of 10 pigs challenged with Sw/MO/14, although all pigs demonstrated moderate tracheitis, which was also observed in the Sw/MO/12 group. Mild tracheitis was observed in only a few of the A/VIC/11 H3N2-challenged pigs.

The pathology of the lung and trachea observed in pigs challenged with A/VIC/11 generated by rg (A/VIC/11rg) was consistent with that observed in pigs challenged with the wild-type strain in experiment 1; however, the pathology observed in pigs challenged with Sw/MO/14rg was milder than that observed in experiment 1 in pigs challenged with wild-type Sw/MO/14 virus (Tables 1 and 2), although it was still relatively high compared to that observed in pigs challenged with the remaining viruses generated by rg. None of the Sw/MO/14 and A/VIC/11 reassortant viruses generated by rg caused significant macroscopic lung lesions compared to those of noninfected pigs except that the lungs of pigs challenged with VIC11-NA (containing 7 genes of Sw/MO/14) showed a trend for increased macroscopic lung lesions and significant microscopic lung lesion scores.

IAV-specific antigen staining was detected by IHC in the Sw/MO/12- and Sw/MO/14-challenged groups, with the average IHC scores in the lungs being 2.0 ± 0.2 and 5.3 ± 0.4, respectively, and the average scores in the trachea being 2.5 ± 0.4 and 2.6 ± 0.2, respectively. Immunoreactive IAV signals were not observed in any of the A/VIC/11-challenged pigs. In experiment 2 with reassortant viruses, IAV antigen was detected in the lungs and tracheas of pigs challenged with Sw/MO/14rg (IHC scores, 2.15 ± 0.3 and 1.7 ± 0.3, respectively) and VIC11-NA (IHC scores, 2.4 ± 0.4 and 1.7 ± 0.5, respectively) and in the tracheas of pigs challenged with VIC11-HA (IHC score, 0.6 ± 0.4), consistent with the virus titers described below.

Infection and transmission of the human-like swine H3 viruses.

The back-titration of the inocula containing of A/VIC/11, Sw/MO/12, and Sw/MO/14 showed that the titers were 10^4.5, 10^4.5, and 10^4.0 TCID₅₀/ml, respectively. IAV were not isolated from BALF or NS specimens from nonchallenged (NC) control pigs. Virus was detected in the BALF of all pigs challenged with Sw/MO/12 and Sw/MO/14, with the BALF of pigs challenged with Sw/MO/14 showing the highest average virus titers (Table 2). In contrast, the BALF of only two pigs inoculated with A/VIC/11 was virus positive at 5 dpi, and the group mean titer was not significantly different from that for the noninfected group.

The back-titration of the inocula used in experiment 2 showed that the titers ranged from 10^4.25 to 10^5.0 TCID₅₀/ml. Challenge with both rg-generated parental viruses resulted in viral titers in BALF similar to the titers of the wild-type viruses observed in experiment 1 (Table 3). Although the reassortant viruses generated by rg did not result in a significant lung pathology, significant mean viral titers in the lungs were detected in an increased number of pigs in the two groups challenged with viruses containing the HA of Sw/MO/14 on the A/VIC/11 backbone (MO14-HA/NA and MO14-HA; Table 3).

TABLE 3.

Macroscopic pneumonia, lung and trachea microscopic pathology, and lung virus titers in pigs challenged with the various viral constructs and NC controls^a

Challenge virus	Percentage with macroscopic pneumonia	Microscopic pneumonia scoreb	Microscopic tracheitis scorec	Log10 virus titer in BALF
NC control	0.3 ± 0.2A	0.1 ± 0.1A	0.0 ± 0.0A	0.0 ± 0.0A (0/5)d
Sw/MO/14rg	6.3 ± 1.9B	6.4 ± 1.0B	1.1 ± 0.4B	4.1 ± 0.3B (10/10)
VIC11-HA/NA	0.5 ± 0.4A	0.1 ± 0.1A	0.1 ± 0.1A	0.0 ± 0.0A (0/10)
VIC11-HA	0.0 ± 0.0A	0.2 ± 0.1A	0.2 ± 0.1A	0.0 ± 0.0A (0/10)
VIC11-NA	2.7 ± 0.7A	2.3 ± 0.7B	0.2 ± 0.1A	3.6 ± 0.4B,C (9/10)
A/VIC/11rg	0.4 ± 0.2A	0.6 ± 0.2A	0.2 ± 0.2A	0.4 ± 0.3A (2/10)
MO14-HA/NA	1.0 ± 0.3A	0.5 ± 0.2A	0.4 ± 0.2A	1.7 ± 0.4C (7/10)
MO14-HA	1.1 ± 0.5A	0.9 ± 0.2A	0.1 ± 0.1A	2.6 ± 0.4C (9/10)
MO14-NA	1.4 ± 0.6A	0.4 ± 0.1A	0.0 ± 0.0A	0.4 ± 0.2A (2/10)

^aResults are shown as means ± standard errors of the means. Different capital letters within the same column indicate significant differences (P ≤ 0.05).

^bThe range of possible scores is 1 to 22.

^cThe range of possible scores is 1 to 8.

^dThe number of virus-positive pigs/total number of pigs tested are indicated in parentheses.

The magnitude and kinetics of virus shedding in nasal secretions were considerably different between the pigs challenged with the human H3 viruses and the pigs challenged with the swine H3 viruses in experiment 1. Only two primary pigs shed low titers of A/VIC/11 during the study period (Fig. 3). Pigs infected with Sw/MO/12 started shedding at 1 dpi, and all pigs were shedding by 3 dpi. All pigs challenged with Sw/MO/14 shed virus from 1 dpi until the day of necropsy, with the titers at 3 and 5 dpi being similar to those in the pigs challenged with Sw/MO/12 (Fig. 3).

FIG 3.

Nasal viral shedding observed in in vivo experiment 1. The virus titers in nasal swab specimens from primary pigs at 1, 3, and 5 dpi with wild-type Sw/MO/12, Sw/MO/14, or A/VIC/11 (A) and of their respective indirect contact pigs at 4, 5, 7, and 9 dpc (B) are shown. Results are shown as the means and standard errors of the means. The numbers of shedding pigs/total number of pigs are indicated above the bars. Different lowercase letters between groups within the same sampling day indicate significant differences (P ≤ 0.05).

None of the pigs in indirect contact with pigs infected with A/VIC/11 shed virus at any time point. In contrast, pigs in indirect contact with both groups of swine H3 virus-infected pigs shed virus starting at 4 dpc, with the average titers for both groups being similar. One pig in the group in contact with Sw/MO/14-infected pigs was still shedding virus at 9 dpc. By 15 dpc, all pigs in contact with Sw/MO/12- and Sw/MO/14-infected animals had seroconverted to homologous virus (average HI titers, 422.2 ± 13.2 and 2,228.6 ± 12.9, respectively), confirming exposure to the challenge virus. None of the pigs in contact with pigs infected with A/VIC/11 seroconverted.

Pigs infected with both parental rg-generated viruses in experiment 2 showed nasal shedding patterns similar to those of pigs infected with the wild-type viruses in experiment 1 (Fig. 4), consistent with what was observed for viral replication in the lungs. Even though virus was detectable in the lungs of pigs infected with MO14-HA/NA and MO14-HA, a significant loss of nasal viral shedding was detected in pigs infected with all reassortant viruses that contained A/VIC/11 internal genes or NA alone compared to that detected in pigs infected with Sw/MO/14rg (Fig. 4). In contrast, the opposite pattern was observed in pigs infected with virus carrying the HA of A/VIC/11 with the Sw/MO/14 backbone (VIC11-HA/NA and VIC11-HA), with significant virus titers being detected in nasal swabs (Fig. 4), despite limited replication in the lung (Table 3). Apart from the shedding patterns observed in primary infected pigs in experiment 2, only infection with Sw/MO/14rg resulted in airborne transmission to indirect contacts, with the titers being similar to those of wild-type Sw/MO/14 (data not shown).

FIG 4.

Nasal viral shedding observed in in vivo experiment 2 with reassortant viruses. The virus titers in nasal swabs specimens from primary pigs at 1, 3, and 5 dpi with reverse genetics-generated parental virus Sw/MO/14rg (A) or A/VIC/11rg (B) and reassortant viruses with surface genes exchanged on the parental backbones (VIC11-HA/NA, VIC11-HA, and VIC11-NA with the Sw/MO/14rg backbone [A] and MO14-HA/NA, MO14-HA, and MO14-NA with the A/VIC/11rg backbone [B]) are shown. Results are shown as the means and standard errors of the means. The numbers of infected pigs/total number of pigs are indicated above the bars. Different lowercase letters within the same sampling day indicate significant differences (P ≤ 0.05). Levels of lung replication are indicated for comparison, and plus signs illustrate approximate log viral titers. The colored bars indicate genome constellation of rg viruses; colors are defined in the key to Fig. 2.

Antigenic analysis of the novel H3 virus.

The antigenic distances between the human-like H3 viruses (Sw/MO/12 and Sw/MO/14) and human and swine H3N2 reference viruses are shown in Fig. 5, in which the antigens are color coded as described by Lewis et al. (28) (cross-tabulated HI titers are shown in Table S4 in the supplemental material). The human-like swine H3 viruses did not cluster with either of the two major antigenic clusters recently identified for contemporary swine H3 viruses descended from the historical cluster III or with prototypic antigens representing historical swine H3 clusters I and II (Fig. 5A). The novel human-like H3N1 and H3N2 viruses were positioned at least 5 antigenic units away from other contemporary influenza viruses established in swine (Fig. 5B). Sw/MO/12 was located 1.4 antigenic units away from Sw/MO/14. The human seasonal H3 representative A/VIC/11 was located 1.9 and 3.1 antigenic units away from Sw/MO/12 and Sw/MO/14, respectively (Fig. 5B).

FIG 5.

Antigenic relationships between the swine human-like H3 viruses and a panel of reference H3N2 viruses. (A) A three-dimensional antigenic map of swine and human H3 influenza viruses. (B) The antigenic distances between the human-like swine H3 viruses (Sw/MO/12 [a] and Sw/MO/14 [b]) and all viruses represented in the three-dimensional map in panel A. The viruses used in this study, Sw/MO/12 H3N2, Sw/MO/14 H3N1, and A/VIC/11, are represented by green, purple, and gray larger spheres/circles, respectively. Swine and human isolates are colored as described by Lewis et al. (28): light blue, A/Wuhan/359/1995 and the cluster I prototype swine H3N2 virus; light pink, A/Sydney/5/1997, A/Moscow/10/1999, and the cluster II prototype swine H3N2 virus; red and cyan, swine H3 antigenic clusters; multiple colors, outliers; gray, other recent human vaccine strains. The scale bar represents one antigenic unit distance, corresponding to a 2-fold dilution of antiserum in the HI assay.

Human-like H3 genes from swine contained many mutations.

To investigate a possible molecular basis for the antigenic properties and pathogenesis observed with the human-like swine H3 viruses studied here, the deduced HA1 amino acid sequences were compared with those from a panel of reference H3 strains. The sequence of the human-like Sw/MO/14 H3 gene differed at 25 amino acids from that of the human vaccine strain with a similar evolutionary history (A/VIC/11; see Fig. S9 in the supplemental material); 8 of these mutations were located in the previously recognized antigenic sites (sites A to E) (31, 32) (see Fig. S9 in the supplemental material). The sequence of the human-like Sw/MO/12 H3 differed at 18 positions from that of A/VIC/11; 3 of these mutations were located in the previously recognized antigenic sites. The sequence of the Sw/MO/12 H3 gene also differed at 16 positions from the sequence of the Sw/MO/14 H3N1 virus H3 gene (see Fig. S9 in the supplemental material). Positions 140 and 145 differed between Sw/MO/14 and A/VIC/11 and the other swine H3 strains and might be key in determining the relative antigenic map position among these strains. Putative N-linked glycosylation sites were predicted using the Net NGlyc (v1.0) server (http://www.cbs.dtu.dk/services/NetNGlyc/). Substitutions predicted to result in the loss of putative N-linked glycosylation sites were detected at 4 amino acid positions in Sw/MO/14 H3N1 and at 2 positions in Sw/MO/12 H3N2 when their sequences were compared to the sequence of A/VIC/11.

DISCUSSION

Despite a certain level of host specificity, many interspecies influenza A virus transmission events have been documented (33). In that context, pigs are an important natural host for IAV and are closely associated with the ecology and evolution of IAV (33). Notably, human IAV can infect swine and establish new lineages of viruses that become endemic to a particular region (6, 9). The continuous spillover of human viruses into pig populations followed by reassortment and evolution has resulted in the circulation in North America of swine IAV containing human-origin virus segments, such as H3N2 viruses with the TRIG constellation and the human seasonal H1-related viruses known as the delta cluster swine viruses (1, 8, 34). In our study, swine human-like H3 viruses newly identified through the USDA surveillance system caused significant lung pathology in infected pigs, which resulted in airborne transmission. This is consistent with evidence from recent diagnostic investigations that demonstrate that the virus has spread to a location in a second U.S. state without known epidemiologic links to the index case in Missouri. However, submissions to the USDA Swine IAV Surveillance System, including the viruses described in this report, are voluntary and anonymous. Therefore, details regarding the clinical disease on some of the source farms and potential epidemiologic links between the outbreaks are not always available. Both the human-like viruses were antigenically distinct from the swine H3 viruses currently circulating in the United States, and antigenic drift from human seasonal H3N2 vaccine strains was also apparent.

Globally, established strains of IAV in pigs are of three main subtypes: H1N1, H1N2, and H3N2 (1, 33). Nevertheless, H3N1 viruses resulting from the reassortment between swine viruses (14, 35, 36) or from interspecies transmission and reassortment (2, 37, 38) have been detected previously. The HAs of the newly emerging H3N2 and H3N1 viruses that we describe are most genetically similar to those of recent human seasonal H3N2 strains from the 2010-2011 season, suggesting that these viruses evolved from a relatively recent spillover event of a human virus into pigs. These human-like viruses have been detected in multiple reassorted genome constellations containing human H3; either human N2, classical swine N1, or swine 2002-lineage N2; and internal genes from H1N1pdm09 or the TRIG constellation with the H1N1pdm09 M gene. Recently, Nelson et al. (6) showed that human-to-swine transmission has occurred relatively frequently since 1965 in at least 8 countries, often with the replacement of the human IAV internal genes with swine-origin genes, suggesting that reassortment and swine adaptation are important for sustained onward transmission.

The human-like H3N2 virus detected first appears to be a precursor to the H3N1 viruses, differing from the H3N1 virus primarily by mutations in the HA gene and in the subtype of the NA gene. The N1 gene of the H3N1 human-like viruses is of the classical N1 lineage that circulates at a frequency relatively similar to that of N2 in pigs. Two lineages of N2 cocirculate in swine in the United States; one is a human seasonal N2 lineage from approximately 1998, and the other is a more recent human seasonal N2 lineage from approximately 2002 (1). Sw/MO/12-like H3N2 viruses containing human-origin virus NA have not been detected by the USDA surveillance system since 2012, yet the H3N1 virus was repeatedly detected in the 2013-2014 season, suggesting that N1 replaced the human-origin N2 virus, although a direct evolutionary link to an N1 source virus could not be made. However, the most recent evaluation of the surveillance data revealed that human-like H3 viruses with swine N2 of the 2002 lineage are now being detected as third-generation reassortants from a putative human seasonal virus precursor. These findings underscore the likelihood that these novel viruses continue to evolve and adapt to the swine host.

The internal gene constellation also appears to be important in the evolution of these human-like viruses in swine. Reassortants containing surface genes from established viruses and the TRIG constellation with the H1N1pdm09 M gene have become predominant in North American swine IAV (1, 39), and other H1N1pdm09 internal genes are increasingly being detected through the USDA surveillance system (39). The novel field isolates studied here contained all internal genes from H1N1pdm09, leading to the speculation that they may be associated with the fitness of these viruses in the swine host. Indeed, pairing of the A/VIC/11 HA or HA and NA with the H1N1pdm09-lineage internal genes from the Sw/MO/14 virus resulted in significantly higher levels of nasal shedding compared to those of whole human virus. More recent isolates detected in the surveillance system contain the TRIG plus pandemic virus M-gene constellations, but they were detected after the present studies were initiated and will be the subject of future studies.

Our results demonstrate that the human-like viruses efficiently infect pigs, cause moderate to severe pneumonia, and are aerially transmitted to indirect contacts. In contrast, the prototypic human ancestor A/VIC/2011 H3N2 virus did not cause significant pathology and failed to be transmitted to indirect contacts. Unaltered wild-type human IAV were previously shown to cause mild respiratory disease and lung pathology in comparison to the severity of disease and pathology caused by swine-adapted virus (40). Conversely, H1N1pdm09, a swine-origin human seasonal virus, causes typical influenza-like clinical signs and shedding in pigs (41, 42), suggesting that IAV have the potential to be fully adapted to humans and swine. Individual gene segments or mutations within gene segments as well as combinations of genes contribute to viral fitness; for example, an ideal balance between the surface genes HA and NA is necessary to result in effective influenza virus infection (43). Our results suggest that Sw/MO/14 HA alone conferred the ability to replicate in the lungs, regardless of the NA or internal genes of the two backbones paired with it. However, the Sw/MO/14 HA in combination with the other genes (NA and/or internal genes) was critical for the ability to replicate in the nasal epithelium and be transmitted to indirect contacts. While the HA from Sw/MO/14 contributed to replication in the lower respiratory tract, the virus containing the HA of A/VIC/11 replicated in the upper respiratory tract when it was paired with the H1N1pdm09-lineage internal genes of Sw/MO/14. These findings indicate that the Sw/MO/14 HA plays a critical role in the adaptation of these novel viruses to swine, but the combination and balance between viral genes were also essential.

Human influenza viruses have been shown to replicate more efficiently at 33 to 34°C due to amino acid 627K in the PB2 gene (44, 45). In contrast, the baseline body temperature of pigs ranges from 38.5 to 39.5°C and may thus restrict replication like that observed for viruses with the human A/VIC/11 virus backbone in the pig's respiratory tract. However, the H1N1pdm09 virus has been shown to efficiently replicate in both the upper and the lower respiratory tracts of pigs (42), and this internal gene backbone likely contributed to the increased replication of the reassortants with the A/VIC/11 HA in the upper respiratory tract. The ability of the H1N1pdm09 virus to replicate in the lower respiratory tract and thus result in lung pathology has been associated with, among other factors, a lower number of glycosylation sites in the HA and reduced surfactant protein D (SP-D)-mediated clearance (46). The two wild-type human-like swine H3 viruses described here had fewer predicted N-linked glycosylation sites in the HA protein than the putative human IAV ancestor, which might have contributed to their increased pathogenicity in pigs.

Additionally, the presence of carbohydrates on the HA might alter the antigenicity of IAV (47), and the reduction observed in the Sw/MO viruses, in addition to other potential antigenicity-impacting amino acid substitutions, may have impacted the cross-reactivity to the H3 reference antiserum panel. Substitutions in as few as 7 amino acid positions were shown to be largely responsible for the antigenic evolution of H3N2 viruses circulating in humans for 35 years (48). In addition, positions 145 and 159 near the receptor-binding site, among others, are likely responsible for antigenic changes in H3N2 swine virus evolution (28). Amino acid substitutions in these two positions as well as others detected in the human-like swine H3 virus likely contributed to the low cross-reactivity observed here between the human-like Sw/MO viruses and the H3 IAV established in swine. However, the magnitude of the effect of each of these individual substitutions is unclear at the current time. Commercially available swine IAV vaccines in the United States contain swine strains from phylogenetic cluster I and/or IV in their compositions. The human-like H3N2 and H3N1 viruses showed little HI cross-reactivity with current and historical swine H3N2 viruses, and, therefore, the immune responses elicited by the commercial swine vaccines are highly unlikely to result in cross-protection against these novel H3 viruses.

Though new subtypes or genotypes of IAV are sporadically detected in pigs, the properties required for a virus to be efficiently transmitted and become established in pig populations are still largely unknown and likely contextual with the whole genome. The recurring bidirectional exchange between swine and human influenza A viruses has contributed much to the diversity of viruses currently circulating in pigs, and the frequent incursions of human seasonal viruses into swine have greatly influenced the dynamics of IAV evolution in swine. We demonstrated that wild-type field isolates of the human-origin H3N2 and H3N1 swine viruses efficiently infected pigs and resulted in onward transmission. However, the adaptation of human viruses to swine appears to be complex, as the HA gene as well as the internal gene constellation played important but variable roles in infectivity, replication, transmission, and pathogenicity in swine, with viruses showing different phenotypes in the upper respiratory tract versus the lower respiratory tract. Importantly, the novel human-like viruses were antigenically divergent from all U.S. swine viruses included in our contemporary H3N2 serum panel and from the strains used in commercially available swine vaccines; therefore, pigs have likely limited immune protection against these novel human-like viruses. Hence, effective surveillance and close monitoring of the evolution of these human-origin viruses in pigs are critical for vaccine preparedness and to improve preventive measures in the swine industry.