Pathogenesis and Transmission of Genetically Diverse Swine-Origin H3N2 Variant Influenza A Viruses from Multiple Lineages Isolated in the United States, 2011–2016

Swine-origin influenza viruses of the H3N2 subtype, with the hemagglutinin (HA) and neuraminidase (NA) derived from historic human seasonal influenza viruses, continue to cross species barriers and cause human infections, posing an indelible threat to public health. To help us better understand the potential risk associated with swine-origin H3N2v viruses that emerged in the United States during the 2011-2016 influenza seasons, we use both in vitro and in vivo models to characterize the abilities of these viruses to replicate, cause disease, and transmit in mammalian hosts. The efficient respiratory droplet transmission exhibited by some of the H3N2v viruses in the ferret model combined with the existing evidence of low immunity against such viruses in young children and older adults highlight their pandemic potential. Extensive surveillance and risk assessment of H3N2v viruses should continue to be an essential component of our pandemic preparedness strategy.

ABSTRACT

While several swine-origin influenza A H3N2 variant (H3N2v) viruses isolated from humans prior to 2011 have been previously characterized for their virulence and transmissibility in ferrets, the recent genetic and antigenic divergence of H3N2v viruses warrants an updated assessment of their pandemic potential. Here, four contemporary H3N2v viruses isolated during 2011 to 2016 were evaluated for their replicative ability in both in vitro and in vivo in mammalian models as well as their transmissibility among ferrets. We found that all four H3N2v viruses possessed similar or enhanced replication capacities in a human bronchial epithelium cell line (Calu-3) compared to a human seasonal influenza virus, suggestive of strong fitness in human respiratory tract cells. The majority of H3N2v viruses examined in our study were mildly virulent in mice and capable of replicating in mouse lungs with different degrees of efficiency. In ferrets, all four H3N2v viruses caused moderate morbidity and exhibited comparable titers in the upper respiratory tract, but only 2 of the 4 viruses replicated in the lower respiratory tract in this model. Furthermore, despite efficient transmission among cohoused ferrets, recently isolated H3N2v viruses displayed considerable variance in their ability to transmit by respiratory droplets. The lack of a full understanding of the molecular correlates of virulence and transmission underscores the need for close genotypic and phenotypic monitoring of H3N2v viruses and the importance of continued surveillance to improve pandemic preparedness.

IMPORTANCE Swine-origin influenza viruses of the H3N2 subtype, with the hemagglutinin (HA) and neuraminidase (NA) derived from historic human seasonal influenza viruses, continue to cross species barriers and cause human infections, posing an indelible threat to public health. To help us better understand the potential risk associated with swine-origin H3N2v viruses that emerged in the United States during the 2011-2016 influenza seasons, we use both in vitro and in vivo models to characterize the abilities of these viruses to replicate, cause disease, and transmit in mammalian hosts. The efficient respiratory droplet transmission exhibited by some of the H3N2v viruses in the ferret model combined with the existing evidence of low immunity against such viruses in young children and older adults highlight their pandemic potential. Extensive surveillance and risk assessment of H3N2v viruses should continue to be an essential component of our pandemic preparedness strategy.

INTRODUCTION

The 2009 influenza pandemic was caused by a swine-origin reassortant H1N1 (H1N1pdm09) virus and highlights the pandemic potential of antigenically distinct influenza viruses harbored in pigs (1). Currently, a wide spectrum of genetically diverse influenza viruses are circulating in pigs worldwide; meanwhile, influenza A viruses in swine (IAV-S) continue to evolve rapidly, with the frequent emergence of novel reassortants (2–4). IAV-S can cause acute respiratory illness in pigs, resulting in economic losses in swine industries (5). IAV-S can also transmit directly from swine to humans and cause sporadic human infections (6); swine-origin influenza viruses that are isolated from humans are referred to as variant viruses (7). Currently, three predominant subtypes (H1N1, H1N2, and H3N2) of influenza virus are endemic in swine populations in Asia, Europe, and North America and are responsible for the human infection cases reported to date (3, 8, 9). From 2005 to 2016, more than 400 confirmed human cases of H3N2 variant (H3N2v), H1N2v, and H1N1v virus infection, including one fatal case, have been reported in the United States, among which 372 cases were caused by H3N2v viruses (10–12). The largest H3N2v virus outbreak in the United States was reported during 2011 to 2012, with 321 confirmed cases, the majority of which occurred among children with a history of exposure to pigs at county agricultural fairs (10, 13, 14). Although the number of human infections with H3N2v has decreased since then, multiple cases are reported each year in the United States, emphasizing that swine-origin influenza variant viruses represent a continuous threat to public health (11).

The H3N2 subtype of IAV-S first emerged in the United States in the late 1990s, bearing a triple-reassortant internal gene (TRIG) cassette (15). The triple-reassortant swine (TRS) H3N2 influenza viruses contain human-lineage hemagglutinin (HA), neuraminidase (NA), and PB1 genes; avian-lineage PB2 and PA genes; and swine-lineage NP, M, and NS genes (5). Compared to classical swine H1N1 influenza viruses, H3N2 IAV-S with the TRIG cassette have been shown to be more permissive to reassortment with endemic IAV-S and surface genes from human seasonal influenza viruses. As a result, novel TRS viruses with classical swine-lineage H1 or N1 or human seasonal H1, H3, N1, and N2 emerged following the introduction of the TRIG cassette in pigs (5). The HA genes of TRS H3N2 viruses were introduced from circulating human seasonal influenza viruses from different years and can be clustered into phylogenetically distinct groups (clusters I, II, and III), with cluster III further diversifying into cluster IV in North America (16–18). In 2009, H1N1pdm09 virus containing North American TRS PB2, PA, and PB1 genes; the Eurasian avian-like swine NA and M genes; and the classical swine HA, NP, and NS genes first emerged in Mexico, and the virus soon spread worldwide, with occasional spillovers from humans to animals, including swine, in many countries (1). The introduction of the H1N1pdm09 virus into the swine population coincided with the increasing rates of reassortment and antigenic drift of IAV-S; the diversity of the swine H3N2 viruses was further expanded with the HA gene from cluster IV evolving to sublineages IV(A) to IV(F) after 2009 (4). The vast genetic heterogeneity of influenza viruses circulating in North American pigs not only has posed a challenge to vaccine development and disease control strategies for the swine industry but also has resulted in diversity among the H3N2v viruses that have led to human infections. The majority of H3N2v viruses isolated in the United States during 2011 to 2012 contained the HA of the IV(A) or IV(B) sublineage, the NA of the 2002 human seasonal lineage, and the TRIG cassette, with the exception of the M gene, from H1N1pdm09 virus (11). A novel reassortant H3N2v virus emerged in 2016, with the HA gene being closely related to 2010-2011 human seasonal influenza virus and the rest of the gene constellation being related to the predominant H3N2v viruses from 2011 to 2012 (20). This reassortant H3N2v virus caused 16 human infections during July to August 2016 (11). Due to the continuous evolution and cocirculation of many genetically diverse H3N2 viruses in swine in the United States, it remains challenging to predict which IAV-S may cross species barriers and transmit to humans in the future.

The increasing number of H3N2v infections in humans in recent years has raised concerns regarding the pandemic potential of this virus subtype. Although the HA genes of H3N2 viruses in swine were all derived from human seasonal H3N2 viruses, they have undergone genetic and antigenic evolution in pigs to be antigenically different from both precursor and currently circulating H3N2 viruses (21). Serological evidence has demonstrated that children and older adults have low levels of cross-reactive antibodies to circulating H3N2v viruses, which partially explains the age disparities in H3N2v human infection cases (19). Following outbreaks of H3N2v in 2011, representative viruses were evaluated for their ability to cause disease and transmit in ferrets, which have served as a relevant model of influenza pathogenesis and transmission in humans; those studies found that several H3N2v viruses had the potential to transmit through the air, underscoring their potential to transmit among humans (22). However, rapid evolutionary dynamics, coupled with novel genetic and antigenic changes associated with recent H3N2v viruses, necessitate an updated risk assessment for this group of emerging zoonotic viruses. In the present study, four contemporary H3N2v viruses from the 2011-2016 season were selected, and their ability to replicate was evaluated in both in vitro and in vivo models, as was their ability to cause disease and transmit among ferrets. We found that the genetic diversity displayed by the contemporary H3N2v viruses was also reflected in the distinct pathogenicity and transmissibility phenotypes observed in the ferret model. Our study provides a comprehensive understanding of the pandemic potential of H3N2v viruses that have emerged in recent years.

RESULTS

Genetic analysis of H3N2v viruses.

Similar to the TRS H3N2 viruses harbored in swine hosts, the H3N2v viruses isolated in the United States have exhibited substantial genetic and antigenic diversity (4, 14, 22, 23). Four representative contemporary viruses (A/Iowa/8/2011 [IA/11], A/Ohio/13/2012 [OH/12], A/Michigan/39/2015 [MI/15], and A/Ohio/27/2016 [OH/16]) selected for our study shared an identical internal gene constellation, which contained the TRIG cassette, with the exception of the M gene, derived from H1N1pdm09 virus. The similarity at the protein level was ≥97.6% for PB2, PB1, PA, NP, and M1/M2. The HA genes of the four H3N2v viruses were from different genetic clusters, with IA/11, OH/12, and MI/15 viruses from the IV(A) cluster and OH/16 virus from the 2010-2011 seasonal human-like lineage (Fig. 1). The three H3N2v viruses from the IV(A) cluster shared 98.9 to 99.6% sequence similarity at the protein level but shared only 83.3 to 83.6% similarity with the HA of OH/16 virus from the contemporary human-like lineage. The NA genes of all four H3N2v viruses were derived from the 2002 human seasonal lineage and shared 90.6 to 97.7% similarity at the protein level.

FIG 1.

Genome constellations representative of the recent H3N2v viruses characterized in this study. Gene segments are color coded to indicate ancestral origin: classical swine North American lineage, avian North American lineage, seasonal H3N2 (cluster IV; 1997/1998 lineage), seasonal H3N2 (human like; 2010/2011 lineage), Eurasian swine lineage from H1N1 pdm09 virus, and a gene derived from H1N1pdm09 virus.

Next, we analyzed key residues involved in receptor binding, replication, and virulence in mammals (Table 1). Like previously characterized H3N2v viruses, all of the contemporary H3N2v viruses included in our study had residues 190D, 226I/V, and 228S in the HA, which are indicative of a binding preference for α2,6-linked sialic acids, the receptor for human influenza viruses (24). Like almost all TRS H3N2 viruses, the H3N2v viruses tested had PB2 627E and 701D. However, the viruses contained 271A and S/R at positions 590 and 591, respectively, in PB2, which have been shown to compensate for a lack of the mammalian host adaption markers 627K and 701N (25). We further compared the numbers of glycosylation sites in the HA proteins of H3N2v viruses, which have been shown to play an important role in influenza pathogenesis in mammals (26). Sequence analysis revealed that the four H3N2v viruses evaluated had 6 to 8 potential glycan motifs; in contrast, up to 11 glycan motifs in the HA protein were predicted for the seasonal A/Switzerland/9715293/2013 (SW/13) virus. Furthermore, the four H3N2v viruses analyzed had either a 90- or a 79-amino-acid (aa)-long PB1-F2 protein; both forms have been shown to be functional in mammalian hosts (27). In addition, MI/15 H3N2v had two unique amino acid substitutions, PB2-A661T and PA-V669I (Table 1); these positions have been shown to be involved in modulating polymerase activity in mammals for H1N1pdm09 virus and H5N1 virus, respectively (28, 29). Collectively, we found that genetically diverse H3N2v viruses have emerged in the United States since 2009, and almost all of them possess key mammalian adaptation molecular markers in the HA and PB2 genes.

TABLE 1.

Genetic characteristics of H3N2v and seasonal H3N2 influenza viruses

Virus	Virus abbreviation	Subtype	HA genetic lineage	Receptor binding site (190D, 226I/V, 228S in HA)	No. of glycans in HAa	Residues at PB2 positions (271A, 627E, 701D, 590/591S/R)	Residue at PB2 position 661	PB1-F2 length (aa)	Residue at PA position 669
A/Kansas/13/2009	KS/09	H3N2v	H3-IV(B)	Yes	7	Yes	A	57	V
A/Pennsylvania/14/2010	PA/10	H3N2v	H3-IV(B)	Yes	9	Yes	A	90	V
A/Minnesota/11/2010	MN/10	H3N2v	H3-IV(A)	Yes	8	Yes	A	90	V
A/Indiana/08/2011	IN/11	H3N2v	H3-IV(A)	Yes	8	Yes	A	79	V
A/Iowa/8/2011	IA/11	H3N2v	H3-IV(A)	Yes	8	Yes	A	90	V
A/Ohio/13/2012	OH/12	H3N2v	H3-IV(A)	Yes	8	Yes	A	90	V
A/Michigan/39/2015	MI/15	H3N2v	H3-IV(A)	Yes	8	Yes	T	90	I
A/Ohio/27/2016	OH/16	H3N2v	Seasonal like	Yes	6	Yes	A	79	V
A/Switzerland/9715293/2013	SW/13	H3N2	Seasonal	Yes	11	No	T	90	V

^aThe number of potential glycosylation sites in the HA protein was predicted by the NetNGlyc 1.0 server.

H3N2v viral replication in human airway epithelial cells.

We next compared H3N2v viruses isolated from 2011 to 2016 for their replicative abilities in Calu-3 cells derived from human airway epithelium to characterize virus fitness in vitro and make comparisons with the recent human seasonal SW/13 virus. It was shown previously that the H3N2v viruses from 2009 to 2011 were all able to replicate efficiently in this cell line (22). In this study, we found that the most recent H3N2v isolate, OH/16, exhibited replication kinetics comparable to those of the human seasonal SW/13 virus, reaching peak titers above 10^7.0 PFU/ml at 72 h postinoculation (p.i.) (Fig. 2). MI/15 H3N2v virus showed higher viral titers at 24 h p.i. than did SW/13 virus but reached similar titers at 48 and 72 h p.i. Like previously characterized H3N2v viruses (22), IA/11 and OH/12 viruses possessed an enhanced capacity to replicate in Calu-3 cells, with significantly higher viral titers at 24, 48 h, and 72 h p.i. than those of the SW/13 virus. We conclude that contemporary H3N2v viruses have comparable or enhanced abilities to replicate in Calu-3 cells compared to a human seasonal H3 virus, indicating that H3N2v viruses have strong growth fitness in human airway epithelial cells.

FIG 2.

Viral growth kinetics in Calu-3 cells. Polarized Calu-3 cells grown on 12-well plates with transwell inserts were inoculated at an MOI of 0.01 with each H3N2v virus or the human seasonal SW/13 (H3N2) virus. Supernatants from each transwell were collected at 2, 24, 48, and 72 h p.i. and titrated in MDCK cells by a plaque assay. The mean viral titers from triplicate repeats are plotted as log₁₀ PFU per milliliter ± standard deviations (SD). Statistical analysis between each H3N2v virus and SW/13 virus was performed using two-way analysis of variance (ANOVA) with a Bonferroni posttest with GraphPad Prism. ***, P < 0.001.

H3N2v viral pathogenicity in mice.

Although mice do not exhibit many clinical signs upon influenza virus infection, correlations have been demonstrated between virulence in mice and disease severity in humans for selected influenza A viruses (30, 31). Human seasonal H3N2 influenza viruses, in general, do not cause disease or replicate well in the respiratory tract of mice without prior adaptation (32, 33), but H3N2v viruses from 2009 to 2016 have not been well studied in mice, so we assessed their level of virulence in this model. Groups of 5 mice each were intranasally (i.n.) inoculated with 5 × 10^5.0 PFU of H3N2v virus and observed for morbidity and mortality for up to 14 days p.i. We found that none of the H3N2v viruses caused lethal disease in mice, but viral infection led to various levels of weight loss, with mean maximum weight loss ranging from negligible (OH/12 virus) to 14.7% (A/Kansas/13/2009 [KS/09] virus) (Fig. 3A). Additionally, groups of 3 mice each were inoculated with 5 × 10^5.0 or 1 × 10^3.0 PFU of virus to determine viral titers in the lung and brain tissues on day 3 and day 6 p.i. A/Indiana/08/2011 (IN/11) and MI/15 viruses were sporadically detected at low levels in murine lung tissue in 1 or 2 out of 3 mice, which was comparable to the seasonal A/Panama/2007/1999 (Pan/99) virus (Fig. 3B). In contrast, the remaining six H3N2v viruses exhibited moderate viral replication in the lung tissue of all inoculated mice, with mean viral titers ranging from 10^3.6 to 10^5.7 PFU/ml on day 3 p.i. at the high inoculation dose (5 × 10^5.0 PFU). However, at a lower inoculation dose (1 × 10^3.0 PFU), only KS/09, A/Pennsylvania/14/2010 (PA/10), and OH/16 H3N2v viruses replicated in murine lungs above a mean titer of 10^4.2 PFU/ml. Furthermore, we compared the persistence of viral replication on day 6 p.i. and found that PA/10 virus (following high-dose inoculation) and KS/09 and PA/10 viruses (following low-dose inoculation) persisted through day 6 p.i. in all inoculated mice at titers of ≥10^3.5 PFU/ml (Fig. 3C). Systemic spread to the brain on day 3 or 6 p.i. was not detected for any H3N2 or H3N2v virus tested. Overall, we conclude that the majority of H3N2v viruses tested in our study possessed the ability to replicate in the murine lung, but not in extrapulmonary tissues, causing minimal to moderate levels of weight loss.

FIG 3.

H3N2v viral replication in mouse tissues. Groups of BALB/c mice were i.n. inoculated with 5 × 10^5.0 or 1 × 10^3.0 PFU of each H3N2v or control human seasonal Pan/99 (H3N2) virus. (A) Percent mean maximum weight loss (five mice per group) following i.n. inoculation with 10^5.0 PFU of viruses is shown in a bar graph. (B and C) Viral titers in the murine lung tissues on day 3 (B) or day 6 (C) p.i. were determined by a plaque assay in MDCK cells, and the mean viral titers are plotted as log₁₀ PFU per milliliter and SD (n = 3). The viral detection limit, shown as a dashed line, was 10 PFU/ml. Negative samples were assigned a value of 10 PFU/ml to calculate mean viral titers. The numbers above the bars represent the ratios of positive samples among three inoculated mice.

H3N2v viral pathogenesis and transmission in ferrets.

We next used the ferret model to compare the abilities of H3N2v viruses to cause disease and transmit, as ferrets represent an optimal small-animal model to evaluate influenza virus infections in mammals (34). In this model, human seasonal H3N2 viruses typically cause mild disease, primarily replicate in the ferret upper respiratory tract, and transmit efficiently in both direct contact (DC) and respiratory droplet (RD) transmission models (35–37). In our study, groups of 3 or 6 ferrets each were i.n. inoculated with 10^6.0 PFU of H3N2v viruses and monitored for clinical signs of illness and for viral shedding in nasal wash specimens. Three additional ferrets were inoculated in each virus group for the assessment of the systemic spread of virus on day 3 p.i. We found that all four H3N2v viruses caused moderate weight loss in infected ferrets, with mean maximum weight losses ranging from 12.6 to 17.0% (Table 2) and mild clinical symptoms, including occasional sneezing, anorexia, and inactivity. H3N2v virus inoculation caused a slight increase in body temperature ranging from 1°C to 1.5°C above the baseline in all four groups of ferrets (Table 2). Lethality was not observed in any H3N2v virus-infected animals, except for a single ferret in the IA/11 virus group that required euthanasia on day 8 p.i. due to excessive weight loss. All four H3N2v viruses were able to replicate efficiently in the ferret respiratory tract, as evidenced by high peak viral titers (>10^5.0 PFU/ml) in nasal wash specimens, with virus persisting in nasal wash specimens for 5 to 7 days after inoculation (see Fig. 5A, left, and B, left). High titers in the nasal turbinates were observed 3 days after inoculation with all of the viruses tested, with mean viral titers ranging from 10^4.8 PFU/ml for OH/12 virus to 10^6.0 PFU/ml for IA/11 virus (Fig. 4). The mean viral titers in tracheal tissue ranged from 10^4.6 PFU/g for MI/15 virus to 10^6.0 PFU/g for OH/16 virus (Fig. 4). However, considerable differences were observed in the levels of H3N2v viruses detected in the lower respiratory tract. MI/15 virus was not observed above the limit of detection (<10^1.0 PFU/g) in the lungs of ferrets on day 3 p.i., whereas OH/12 virus replicated to moderate levels (mean viral titer, 10^2.8 PFU/g), and both IA/11 and OH/16 viruses were measured with mean viral titers of 10^4.4 to 10^5.0 PFU/g. The ability of viruses to spread to extrapulmonary tissues was also evaluated at this time point. Virus was either not detected in any systemic tissue (IA/11 virus) or detected sporadically and at low levels in brain and olfactory bulb (OB) from OH/12-, MI/15-, and OH/16-inoculated ferrets and in intestinal tissues from 1/3 OH/12-inoculated ferrets. These findings indicate that the H3N2v viruses lack the ability to consistently replicate outside the ferret respiratory tract and do not possess a tropism for intestinal tissues, which has been reported for some TRS H1N1 and H1N1pdm09 viruses (38, 39).

TABLE 2.

Clinical signs and transmission of H3N2v influenza viruses in ferrets

Virus	Exposure groupa	Mean % max wt loss (days)b	Mean max increase in temp (°C) (days)c	No. of animals with nasal discharge/total no. of animals	No. of animals with sneezing/total no. of animals	Mean max viral titer in NW (log10 PFU/ml)d	No. of animals that seroconverted/total no. of animals (HI titer)	pH threshold for fusiong
IA/11	Inoculated	17.0 (6–8)	1.5 (1–2)	2/3	0/3	6.0	2/2e (1,280)	5.6
	DCT	NTf	NT	NT	NT	NT	NT
	RDT	0	0.9 (2–4)	0/3	0/3	<1	0/3 (<10)
OH/12	Inoculated	12.6 (6–7)	1.3 (2–9)	0/6	5/6	6.4	6/6 (640–1,280)	5.6
	DCT	10.5 (8)	1.2 (3–4)	0/3	3/3	6.3	3/3 (1,280)
	RDT	7.8 (8–9)	1.7 (3–5)	0/3	2/3	6.1	3/3 (1,280)
MI/15	Inoculated	13.3 (6–10)	1.2 (2–6)	0/6	1/6	5.4	6/6 (320–640)	5.6
	DCT	9.6 (4–8)	1.6 (4–7)	2/3	0/3	5.5	3/3 (320–640)
	RDT	4.9 (6–8)	1.0 (4–9)	0/3	0/3	<1	0/3 (<10)
OH/16	Inoculated	16.5 (4–10)	1.0 (1–10)	0/6	2/3	6.1	6/6 (1,280)	5.5
	DCT	10.9 (8)	1.4 (3–4)	0/3	2/3	5.8	3/3 (1,280)
	RDT	8.4 (8–13)	1.0 (3–9)	0/3	1/3	4.9	3/3 (1,280)

^aInoculated indicates inoculation intranasally with 10^6.0 PFU. DCT, direct contract transmission model; RDT, respiratory droplet transmission model.

^bPercent mean maximum weight loss from all infected ferrets. The range of days of observation is shown parenthetically.

^cMean maximum temperature increase over the baseline (ranging from 37.6°C to 39.5°C). The range of days of observation is shown parenthetically.

^dMean maximum viral titers in nasal wash specimens from ferrets (log₁₀ PFU per milliliter).

^eOne inoculated ferret was euthanized due to excessive weight loss; therefore, only the remaining two ferrets were evaluated for seroconversion.

^fNT, not tested.

^gThe highest pH to induce at least 50% syncytia in Vero cells infected with H3N2v influenza viruses. The fusion thresholds for influenza viruses (SW/13 and Pan/99) were 5.4 and 5.6, respectively.

FIG 5.

Transmissibility of H3N2v in direct contact and respiratory droplet transmission ferret models. (A) Three naive contact ferrets were each cohoused with a ferret inoculated with 10^6.0 PFU of OH/12, MI/15, or OH/16 H3N2v virus on day 1 p.i. (B) Three naive contact ferrets were each housed adjacent to a ferret inoculated with 10^6.0 PFU of IA/11, OH/12, MI/15, or OH/16 H3N2v virus in separated cages to allow air exchange but without direct or indirect contact between the animals. Nasal wash specimens from individual inoculated or contact ferrets were collected on days 1, 3, 5, 7, 9, and 11 p.i. or p.c. and then titrated in MDCK cells by a plaque assay. Viral titers are expressed as log₁₀ PFU per milliliter, with a detection limit of 10 PFU/ml.

FIG 4.

H3N2v viral replication in ferret tissues. Three ferrets each were i.n. inoculated with 10^6.0 PFU of IA/11, OH/12, MI/15, or OH/16 H3N2v virus and then humanely euthanized on day 3 p.i. to collect tissues for viral titration in MDCK cells by a plaque assay. The viral titers from individual ferrets are shown as log₁₀ PFU per milliliter for nasal turbinates (Nasal Tur) or log₁₀ PFU per gram for all other tissues. The detection limit, shown by dotted lines, is 10 PFU per ml or g. The intestinal tissue sample represents pooled duodenum, jejunoileal loop, and descending colon, and the brain sample represents pooled anterior and posterior sections.

We next evaluated the ability of H3N2v viruses to transmit using either the DC or RD transmission ferret model. H3N2v viruses in general possess an efficient capacity to transmit among cohoused ferrets, as we have demonstrated previously (22, 40); we therefore selected only OH/12, MI/15, and OH/16 viruses to evaluate their transmission in the DC transmission model. We found that the three H3N2v viruses that we tested transmitted efficiently among cohoused ferrets, as demonstrated by the presence of infectious virus in nasal wash specimens of some contact animals by as early as day 1 postcontact (p.c.) and in all animals by day 3 p.c. (Fig. 5A). The majority of contact ferrets showed clinical signs, and mean maximum virus titers, similar to those observed among inoculated ferrets (Table 2). Successful transmission in the DC transmission model was confirmed by seroconversion; all contact ferrets had hemagglutination inhibition (HI) titers similar to those of the inoculated ferrets on day 21 p.c. (Table 2). H3N2v viruses were then tested for their ability to transmit through the air by respiratory droplets. As shown in Fig. 5B, none of the contact ferrets in the IA/11 and MI/15 groups shed detectable virus in nasal wash specimens or seroconverted at the end of the experiment. In contrast, 3/3 RD contact ferrets exposed to OH/12 and OH/16 virus-infected ferrets exhibited positive viral titers in nasal wash specimens and seroconverted with HI titers against homologous virus comparable to those in inoculated ferrets by day 21 p.c. (Table 2). However, differences in transmission dynamics were observed between these two viruses in this model. Whereas all three RD contact ferrets in the OH/12 group shed virus to high titers on day 3 p.c., only one ferret in the OH/16 group shed detectable virus in nasal wash specimens on day 3 p.c., with the remaining two contact ferrets not becoming virus positive until day 7 or day 11 p.c. The delayed transmission on day 11 associated with OH/16 virus may be due in part to the delayed peak virus shedding of the donor, the inoculated ferret of that ferret pair. Taken together, we conclude that despite genetic and antigenic diversity, all four representative H3N2v viruses caused only moderate disease in ferrets and exhibited generally similar profiles in virus shedding and replication in the upper respiratory tract but possessed differing abilities to replicate in lower respiratory lung tissues. Additionally, contemporary H3N2v viruses transmitted efficiently in the presence of DC but showed various efficiencies of transmission in an RD model.

H3N2v pH threshold for fusion.

An acid-stable HA protein has recently been recognized as an important property for aerosol transmission of influenza viruses (41). To investigate whether the acid stability of H3N2v viruses correlates with virus transmissibility among ferrets, we compared the pH thresholds for fusion for H3N2v viruses using a syncytium formation assay in Vero cells. Human seasonal H3N2 viruses, represented by SW/13 and Pan/99 viruses, have pH thresholds for fusion of 5.4 and 5.6, respectively. The selected H3N2v viruses, IA/11, OH/12, MI/15, and OH/16, exhibited similar pH thresholds for fusion, ranging from 5.5 to 5.6 (Table 2), indicating that H3N2v viruses share acid stability comparable to that of human seasonal H3N2 viruses.

DISCUSSION

The sudden surge in human infections by H3N2v viruses in 2011 to 2012 and again in 2016, coupled with the lack of cross-reactive immunity against H3N2v in young children and older adults, underscores the public health threat of swine-origin influenza viruses (19). Here, we showed that contemporary H3N2v viruses that emerged from genetically diverse IAV-S were well suited to grow in human airway epithelial cells, as demonstrated by their comparable or even enhanced abilities to replicate in Calu-3 cells in comparison with human seasonal H3 viruses. However, H3N2v viruses displayed heterogeneity in their abilities to replicate throughout the respiratory tract in both mice and ferrets and to transmit among ferrets, emphasizing the complexity in assessing the pandemic risk posed by these viruses. Nevertheless, selected H3N2v viruses have exhibited transmission dynamics and efficiency similar to those of seasonal H3N2 viruses, coupled with an enhanced replication capacity in the mammalian lower respiratory tract, indicating that H3N2v viruses have the potential to spread and cause serious disease in humans.

The HA genes of the H3N2v viruses evaluated here are all originally derived from human influenza viruses. Like their human influenza virus precursors, H3N2v viruses possess key residues (190D, 226V/I, and 228S) at the receptor binding site (RBS) responsible for binding to α2,6-linked sialic acids, the preferred receptor for human influenza viruses, suggesting that H3N2v viruses have an inherent property to target the human respiratory epithelium. However, compared to human seasonal viruses, most H3N2v viruses have HA substitutions at positions (e.g., positions 156, 157, 158, 189, and 193) adjacent to the RBS, which have recently been shown to affect binding to selected human glycan analogs (42). Whether these substitutions affect H3N2v virus binding or replication in human respiratory tissues in vivo needs further investigation. In humans, children aged <10 years have shown little or no cross-reactive immunity to H3N2v viruses, compared to a 20 to 30% seropositivity rate among groups aged ≥10 years. Vaccination with the seasonal trivalent influenza virus vaccine did not improve the seroconversion rate among this age group, making this group particularly susceptible to H3N2v virus infection (19, 43). The antigenic divergence between H3N2v and seasonal viruses indicates that the development of H3N2v candidate vaccine viruses that induce a broadly reactive immune response in humans is critical and represents an essential component of pandemic preparedness.

Previous studies have demonstrated that contemporary human seasonal H3N2 viruses are attenuated in mice and do not replicate efficiently, mainly due to the accumulation of N-linked glycosylation sites in the globular head of HA, which results in enhanced neutralization by respiratory tract surfactant protein D (26). Consistent with data from previous studies, we confirmed that the human seasonal virus Pan/99, which bears 11 potential glycosylation sites in the HA, displayed minimal replication in the murine lung. Interestingly, with the exception of IN/11 and MI/15 viruses, the majority of H3N2v viruses included in our study exhibited an ability to replicate moderately in the lungs of mice without prior adaptation. However, the viral tropism for lung tissues apparently was not dependent on the number of glycosylation sites in the HA protein or on the lineages of the HA or M gene. IN/11 and MI/15 viruses had the same number of glycosylation sites in the HA protein and shared the same HA and M gene lineages as some of the replication-competent H3N2v viruses. The influenza virus PB1-F2 protein has previously been identified as a virulence factor in mice, and the expression of full-length PB1-F2 has been associated with the enhanced virulence of several influenza viruses, including highly pathogenic avian influenza (HPAI) H5N1 and the reconstructed 1918 viruses (44, 45). However, the length of PB1-F2 of H3N2v viruses did not appear to modulate viral replication in mice, as the IN/11 and MI/15 viruses had truncated (79-amino-acid) or full-length (90-amino-acid) PB1-F2, respectively, but showed similar lung titers; both protein lengths have been shown to be functional in mammalian hosts compared to the nonfunctional 57-aa truncated protein (27). Moreover, the H3N2v viruses displayed considerable differences in viral persistence in the murine lung, and only KS/09 and PA/10 H3N2v virus-infected mice showed delayed viral clearance at a low viral inoculation dose. Interestingly, these two viruses also caused the most weight loss in mice (approximately 12.1 to 14.7%) among all others tested. Influenza virus persistence in murine lung tissues reflects a balance between viral replication in respiratory tract cells and host immune responses (46, 47). However, whether the H3N2 viruses were able to induce host innate and adaptive immune responses differently, which consequently would contribute to their ability to cause weight loss and delays in viral clearance in mice, needs further investigation. Compared to H3N2v viruses, H1N1pdm09 and TRS H1N1v viruses have exhibited consistent viral replication in murine lungs, with occasional lethality, highlighting the significance of surface HA and NA genes in the pathogenesis of swine-origin influenza viruses in mice (48–50). Although the identification of novel molecular correlates for H3N2v virus virulence in the mouse model is beyond the scope of the present study, a comprehensive pathogenicity comparison of H3N2v viruses spanning several years will have great implications for evaluating the efficacy of novel vaccines and antiviral drugs in this animal model.

In the present study, we further demonstrated that similar to previously characterized H3N2v viruses from 2009 to 2011, the contemporary H3N2v viruses from 2011 to 2016 caused greater weight loss in ferrets than did a seasonal H3N2 virus (51) and replicated efficiently in the ferret upper respiratory tract, shedding high titers of infectious virus on day 1 to day 3 p.i. A previous study from our group showed that ferrets infected by the IN/11 H3N2v virus shed higher viral titers in nasal wash specimens than did ferrets infected by the seasonal Pan/99 virus (52). Despite uniformly efficient replication in the upper respiratory tract, the contemporary H3N2v viruses exhibited substantial differences in their abilities to replicate in ferret lower respiratory tract lung tissues. IA/11 and OH/16 viruses, but not MI/15 or OH/12 viruses, replicated to moderate titers in the lungs of ferrets. The differing capacities to replicate in the ferret lower respiratory tract have also been observed with previously characterized early H3N2v isolates (22). However, similar to the varied pathogenicity among H3N2v viruses in mice, the known molecular markers, including HA receptor binding specificity, mammalian adaptation markers in PB2, and the length of PB2-F1, cannot account for the difference in H3N2v replication in the ferret lower respiratory tract. H3N2v viruses, in general, have exhibited enhanced pathogenicity in ferrets compared to seasonal H3N2 viruses, as demonstrated by higher weight loss and viral shedding as well as better viral replication associated with selected H3N2v viruses; therefore, it is of great importance to closely monitor the disease caused by H3N2v viruses in humans.

So far, although most human infections with H3N2v virus occur only sporadically, without sustained human-to-human transmission, a few small clusters of secondary transmission in humans have been reported (53, 54). In contrast with HPAI H5 viruses, the capacity for efficient transmission in the presence of direct contact shared by all H3N2v viruses that have been characterized to date (22, 40) indicates that swine-origin H3N2v viruses are more adapted to human hosts than are HPAI H5 viruses (55). However, H3N2v viruses have displayed considerable differences in their abilities to transmit in an RD model. Among the eight H3N2v viruses that have been characterized using the RD transmission model here or previously (22), five of them (A/Minnesota/11/2010 [MN/10], IN/11, PA/10, OH/12, and OH/16 viruses) transmitted between 3 of 3 ferret pairs, whereas KS/09 virus transmitted between 2 of 3 ferret pairs, and MI/15 and IA/11 viruses failed to transmit through the air. Our understanding of influenza virus transmission has been advanced by a number of studies involving H5 and H7 avian influenza viruses and the reconstructed 1918 and H1N1pdm09 viruses (30, 56–61). In general, it has been shown that preferred α2,6-linked sialic acid receptor binding and efficient viral replication in the mammalian upper respiratory tract are essential for efficient transmission through the air (62). However, all 8 H3N2v viruses share key residues in HA governing the binding preference for α2,6-linked sialic acid receptors and the S/R residues at positions 590/591 in PB2 to confer efficient viral replication in mammalian hosts, indicating that other unidentified markers govern the heterogeneity of H3N2v RD transmission in the ferret model. Recently, the acquisition of the M segment from the H1N1pdm09 virus has been suggested to enhance RD transmission of H3N2 viruses with an otherwise intact TRIG cassette in the guinea pig model (63). However, both the MI/15 and IA/11 viruses contained the M gene of H1N1pdm09 virus yet cannot readily transmit among ferrets. Additionally, we compared the virus fusion pH, which has been shown to play an important role in virus transmissibility (64). We found that irrespective of their ability to transmit among ferrets, all H3N2v viruses had a pH threshold for fusion similar to those of the seasonal H3N2 viruses that we evaluated, indicating that fusion pH alone is not a direct correlate of the transmissibility of these H3N2v viruses among ferrets. Furthermore, we did not identify any common mutations in each of 8 segments shared by all transmissible H3N2v viruses, although some individual mutations associated with polymerase activity or virulence in mammals have been observed for certain H3N2v viruses, for example, A661T in PB2 and V669I in PA of MI/15 virus (28, 29). We therefore speculate that multiple substitutions across different gene segments, especially in the genetically more diverse HA and NA segments, act synergistically in a strain-specific manner to modulate the transmissibility of H3N2v viruses. Overall, we conclude that the ability for H3N2v virus transmission by RD is a multifactorial trait, and future extensive studies with isogenic recombinant viruses would be necessary in order to identify additional genetic markers of virulence and airborne transmission of swine-origin influenza viruses.

In our study, we demonstrated that several divergent H3N2v viruses exhibit many similarities to the H1N1pdm09 influenza virus, including efficient viral replication in both upper and lower respiratory tract tissues and the capability of transmission by respiratory droplets in a mammalian model (39, 65, 66). This highlights the pandemic potential of these emerging variant virus strains and the necessity for continuous surveillance and close monitoring of cross-reactive immunity in the general population. Furthermore, with the emergence of novel swine H3N2v reassortant viruses with an increased number of gene segments from the H1N1pdm09 virus in the United States (4), future risk assessments will need to be performed to continuously monitor the threat that these emerging zoonotic viruses pose to human health.

MATERIALS AND METHODS

Viruses and cells.

H3N2v influenza viruses IA/11 and OH/12, originally isolated from 1-year-old and 9-year-old males, respectively, were propagated for three passages in Madin-Darby canine kidney (MDCK) cells. H3N2v viruses MI/15 and OH/16, originally isolated from a 43-year-old male and a 2-year-old female, respectively, were passaged in MDCK cells once and then passaged twice in MDCK-SIAT1 cells (stably transfected with the cDNA of 2,6-sialtransferase; ATCC). All patients from whom the four H3N2v viruses were isolated fully recovered. The sequences of stock viruses have been deposited in the GISAID database (accession numbers IA/11, EPI_ISL_99214; OH/12, EPI_ISL_123207; MI/15, EPI_ISL_195645; OH/16, EPI_ISL_232044). The stock virus preparations for H3N2v viruses KS/09, MN/10, PA/10, and IN/11 as well as human seasonal H3N2 viruses SW/13 and Pan/99 were previously described (22). All virus stocks were sequenced and subjected to exclusivity and sterility testing to ensure no contamination from bacteria, fungi, or other subtypes of influenza viruses. The titers of stock viruses were determined in MDCK cells by a standard plaque assay.

Viral infection in Calu-3 cells.

Polarized Calu-3 cells derived from human bronchial epithelium were maintained and cultured in 12-well plates with semipermeable inserts (Costar, Corning, NY) according to previously established methods (67). In brief, Calu-3 cells were inoculated apically with each H3N2v virus or seasonal H3N2 virus at a multiplicity of infection (MOI) of 0.01 for 1 h at 37°C, followed by washing with viral infection medium (0.3% bovine serum albumin in minimal essential medium) to remove the viral inoculum. Fresh viral infection medium was added, and cell culture supernatants were sampled apically at 2, 24, 48, and 72 h p.i. for viral titration in MDCK cells by a plaque assay.

H3N2v viral replication and virulence in mice.

All animal experiments were performed under the guidance of the Centers for Disease Control and Prevention Institutional Animal Care and Use Committee in an Association for Assessment and Accreditation of Laboratory Animal Care International-accredited facility. Female BALB/c mice (Jackson Laboratory), 6 to 8 weeks of age, were inoculated and monitored as previously described (48). Briefly, groups of five mice were i.n. inoculated with 5 × 10^5.0 PFU of each H3N2v virus diluted in 50 μl of phosphate-buffered saline (PBS); the inoculated mice were monitored for morbidity (as measured by weight loss) and mortality up to 14 days p.i. Additional groups of three mice were i.n. inoculated with 5 × 10^5.0 or 1 × 10^3.0 PFU of each H3N2v or seasonal Pan/99 H3N2 virus in 50 μl of PBS, and the inoculated mice were humanely euthanized on day 3 or day 6 p.i. to collect lung and brain tissues. The presence of infectious virus in murine tissues was determined by plaque assays in MDCK cells after the tissues were homogenized in 1 ml of PBS and clarified by centrifugation.

Pathogenesis and transmission of H3N2v virus in ferrets.

Male Fitch ferrets, 4 to 10 months of age, from Triple F Farms (Sayre, PA), were tested to be serologically negative for currently circulating influenza viruses by HI and were housed in a Duo-Flow Bioclean unit (Lab Products Incorporated, Seaford, DE) throughout the study. To assess the ability of H3N2v virus to replicate and cause disease in ferrets, groups of up to nine ferrets each were i.n. inoculated with 10^6.0 PFU of each H3N2v virus diluted in 1 ml of PBS. Three ferrets from each group were humanely euthanized on day 3 p.i. to collect tissues for viral titration, and the remaining three or six inoculated ferrets were monitored daily for clinical signs of illness, weight loss, and lethargy and served as donors in transmission studies as described previously (22, 68). Nasal wash samples were collected on days 1, 3, 5, 7, and 9 p.i. for viral titration by a plaque assay in MDCK cells. The ability of H3N2v virus to transmit among ferrets was evaluated in both DC and RD transmission models as described previously (22). For the DC transmission model, a naive ferret is paired with each inoculated ferret at 24 h p.i. in the same cage. Using this model, transmission may occur between ferret pairs via any contact or airborne route. For RD transmission, a naive ferret is placed in an adjacent cage with a perforated side wall next to each inoculated animal to allow air exchange in the absence of direct or indirect contact. This model of transmission is more restrictive and provides the opportunity for transmission only by an airborne route. Virus transmission was decided based on the detection of infectious viruses in the nasal wash samples collected on days 1, 3, 5, 7, 9, and 11 p.c. and seroconversion to homologous H3N2v virus on day 21 p.c.

Syncytium formation assay in Vero cells.

To compare the pH thresholds for fusion of H3N2v viruses, a syncytium formation assay in Vero cells infected with H3N2v viruses was performed according to previously established methods (69). Briefly, Vero cells were infected in triplicate with each of the H3N2v viruses at an MOI of 1 for 16 h and then treated with 5 μg/ml of tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-trypsin for 15 min. The syncytia were induced by incubating cells with fusion buffer (20 mM HEPES, 2 mM CaCl₂, 150 mM NaCl, 20 mM citric acid monohydrate-sodium citrate tribasic dehydrate) with pH values ranging from 5.2 to 6.0 with 0.1-unit increments for 5 min. The syncytia were visualized by staining cells with anti-NP antibody (clone A1-A3 blend; EMD Millipore) 3 h after fusion induction; the pH threshold for fusion was defined as the highest representative pH value at which at least 50% syncytium formation can be observed among NP-positive cells.