Primary Swine Respiratory Epithelial Cell Lines for the Efficient Isolation and Propagation of Influenza A Viruses

Robust in vitro culture systems for influenza virus are critically needed. MDCK cells, the most widely used cell line for influenza isolation and propagation, do not adequately model the respiratory tract. Therefore, many clinical isolates, both animal and human, are unable to be isolated and characterized, limiting our understanding of currently circulating influenza viruses. We have developed immortalized swine respiratory epithelial cells that retain the ability to differentiate and can support influenza replication and isolation. These cell lines can be used as additional tools to enhance influenza research and vaccine development.

ABSTRACT

Influenza virus isolation from clinical samples is critical for the identification and characterization of circulating and emerging viruses. Yet efficient isolation can be difficult. In these studies, we isolated primary swine nasal and tracheal respiratory epithelial cells and immortalized swine nasal epithelial cells (siNEC) and tracheal epithelial cells (siTEC) that retained the abilities to form tight junctions and cilia and to differentiate at the air-liquid interface like primary cells. Critically, both human and swine influenza viruses replicated in the immortalized cells, which generally yielded higher-titer viral isolates from human and swine nasal swabs, supported the replication of isolates that failed to grow in Madin-Darby canine kidney (MDCK) cells, and resulted in fewer dominating mutations during viral passaging than MDCK cells.

IMPORTANCE Robust in vitro culture systems for influenza virus are critically needed. MDCK cells, the most widely used cell line for influenza isolation and propagation, do not adequately model the respiratory tract. Therefore, many clinical isolates, both animal and human, are unable to be isolated and characterized, limiting our understanding of currently circulating influenza viruses. We have developed immortalized swine respiratory epithelial cells that retain the ability to differentiate and can support influenza replication and isolation. These cell lines can be used as additional tools to enhance influenza research and vaccine development.

INTRODUCTION

Outbreaks of influenza A viruses continue to cause significant morbidity and mortality worldwide (1, 2). The rapid evolution of the virus requires constant surveillance, which is dependent on timely detection, isolation, and characterization of the circulating viruses. Influenza viruses must be amplified in vitro for detailed antigenic and phenotypic analysis (3). Historically, embryonated hen eggs have been used to propagate influenza viruses (4, 5), and most seasonal vaccines continue to be produced in eggs. Mammalian cell lines are a simple alternative for virus isolation from clinical samples. Madin-Darby canine kidney (MDCK) cells are highly susceptible to infection, making them the most widely used cell line for influenza studies (6–9). Several variant MDCK cell lines have been engineered to increase the levels of the α-2,6-linked sialic acids, the primary receptor for human influenza viruses, to support increased isolation of viruses from clinical samples, particularly the recent A/H3N2 viruses (8, 10–13).

The 2009 pandemic and the ongoing introduction of variant H1N1 (H1N1v) and H3N2 (H3N2v) influenza viruses from swine to humans around the world (14–18) highlight the need for surveillance in swine and people at the animal-human interface. The swine respiratory tract possesses both mammalian α2,6-sialic acid and avian α2,3-sialic acid receptors, facilitating gene reassortment between multiple influenza subtypes (19). Indeed, primary respiratory epithelial cells from a variety of species, including swine, support the efficient replication of influenza and other respiratory viruses (20–28). Yet primary cells have several disadvantages, including limited passaging before reaching senescence, creating the need for a steady supply of tissues and/or donors (29, 30). We hypothesized that immortalized swine cell lines could decrease the number of animals required, reduce donor-to-donor variation, and enhance reproducibility.

In these studies, we describe the development and characterization of immortalized swine nasal epithelial cells (siNEC) and tracheal epithelial cells (siTEC) from primary cells. These novel cell lines can be continuously passaged yet retain the ability to differentiate in culture. siNEC and siTEC lines supported the replication of diverse human, swine, and avian influenza virus strains, like primary cultures and, in some cases, more efficiently than MDCK cells. The siNEC and siTEC lines were generally more efficient at generating high-titer viral isolates from clinical samples (swine and human) than MDCK and MDCK-SIAT cells. Overall, we have developed unique cell lines that retain many of the properties of primary cells and support influenza virus replication and isolation, making them an invaluable tool for influenza research and surveillance that may extend to other viruses.

RESULTS

Respiratory cell lines retain properties of primary cells.

Primary epithelial cells were isolated from swine nasal and tracheal tissues (Fig. 1A) and cultured in collagen-coated 100-mm dishes. At 70% confluence, cells were transfected with simian virus 40 (SV40) large T antigen to create immortalized cell lines. Although primary cells did not survive beyond 3 to 6 passages, the immortalized cells were successfully passaged >30 times. Further, the doubling time of the immortalized cell lines was significantly shortened compared to primary cells (Fig. 1B). Since fibroblasts are a common contaminant of primary epithelial cell cultures, siNEC and siTEC were stained with vimentin, a recognized fibroblast marker (31), and assessed by flow cytometry. Compared to swine lung homogenates, siNEC and siTEC had decreased vimentin staining (0.41% and 0.68%, respectively) versus 33.8% in lung homogenates (Fig. 1C and D).

FIG 1.

Characterization of immortalized cells from nasal mucosa and trachea. (A) Swine tissues were collected from the indicated sites. (B) Doubling times of sNEC, siNEC, sTEC, and siTEC were compared over 3 to 6 passages. *, P < 0.05 by unpaired t test; n = 3 to 6. Error bars represent the SEM. (C) Representative histograms showing vimentin expression in swine lung, siNEC, and siTEC. (D) Quantification of fibroblast content from panel C. (E) Transepithelial electrical resistance (TER) measurements of differentiated swine cells cultured at the air-liquid interface. Data represent one independent experiment; n = 3 per group. Error bars represent SD.

Unlike most cell lines (32), siNEC and siTEC retained the ability to differentiate at the air-liquid interface. The cells formed tight junctions, as demonstrated by high transepithelial electrical resistance (TER) (33) (Fig. 1E) and presence of the adherens junction protein β-catenin (Fig. 2). Using β-tubulin as a marker, we observed the presence of ciliated cells in cultures of primary swine nasal epithelial cells (sNEC), primary swine tracheal epithelial cells (sTEC), and siTEC. Interestingly, we were unable to detect ciliated cells in the siNEC cultures, although both siNEC and siTEC secreted mucins (Fig. 3A). These results demonstrate that the swine respiratory cell lines differentiate similar to primary cells and can be grown at the air-liquid interface.

FIG 2.

siNEC and siTEC show characteristics of differentiation. Immortalized and primary NEC and TEC were plated on transwell inserts. After 3 weeks to allow for differentiation, cells were fixed and stained to detect the indicated proteins and images acquired by confocal microscopy. White, β-catenin; red, β-tubulin; blue, DAPI. Scale bar, 20 μm. Data represent two independent experiments; n = 2 to 3 per group.

FIG 3.

Mucin and sialic acid receptor expression on immortalized siNEC and siTEC. siNEC and siTEC were plated on transwell inserts and allowed to differentiate for 3 weeks. (A) Cells were fixed and stained with anti-MUC1 (left) or anti-MUC5AC (right). Magnification, ×63. (B) Cells were stained with Maackia amurensis and Sambucus nigra agglutinin lectins to detect α2,3- and α2,6-sialic acid receptors (green). Blue, DAPI. Magnification, ×60. (C) The percentage of cells positive for α2,3- and α2,6-sialic acid receptors from 8 representative fields were quantified. Primary sNEC and sTEC were included as controls. Data represent two independent experiments; n = 2 to 3/group. Error bars represent SD.

Swine cell lines support viral replication.

Sialic acids are the primary receptors for influenza virus (34). Mammalian viruses preferentially bind the α2,6 conformation, while avian viruses bind the α2,3 conformation (35). The swine respiratory tract expresses both (36, 37). Indeed, siNEC and siTEC lines also expressed both receptors, with α2,6 sialic acid expressed more abundantly than α2,3 (Fig. 3B and C). Because of the sialic acid distribution on the immortalized cells, we hypothesized that they would support influenza viruses from diverse species.

Primary and immortalized cells were infected with a panel of human or swine influenza viruses (Table 1) and viral kinetics assessed. Briefly, differentiated primary and immortalized siNEC and siTEC were infected at a multiplicity of infection (MOI) of 0.1, and apical supernatants were collected from 24 to 72 hours postinfection (hpi). Overall, the replication kinetics and titers of the primary cells and cell lines were similar (Fig. 4). However, the A/California/04/2009 (CA/09) (H1N1) and the A/swine/Iowa/13-1015/2010 (swine/IA H3N2) viruses replicated to ∼2 logs higher in the sNEC than the siNEC. No cytopathic effects (CPE) were observed with CA/09 (H1N1) or A/Switzerland/2013 (H3N2) viruses; however, significant CPE was observed in cells infected with swine viruses. Overall, these data suggest that viral replication dynamics in immortalized and primary swine cells are comparable.

TABLE 1.

Viral strains used

Abbreviation	Virus name	Subtype
CA/09	A/California/04/2009	H1N1
Switzerland	A/Switzerland/9715293/2013	H3N2
Bethesda	A/Bethesda/55/2013	H3N2
RGC	A/red-gartered coot/Chile/C16030/2016	H3N4
Swine/IA	A/swine/Iowa/13-1015/2010	H3N2
Swine/IN	A/swine/Indiana/08/2011	H3N2va
Swine/NC	A/swine/North Carolina/0668/2011	trH3N2b
Swine/TX	A/swine/Texas/4119-2/1998	H3N2

^aVariant H3N2.

^bTriple-reassortant H3N2.

FIG 4.

Viral growth kinetics in primary and immortalized swine cells. (A) sNEC and siNEC were grown to the air-liquid interface and cultured until fully differentiated. Cells were infected with A/California/04/2009 (H1N1), A/Switzerland/9715293/2013 (H3N2), A/swine/Iowa/13-105/2010 (H3N2), or A/swine/North Carolina/0668/2011 (H3N2) at an MOI of 0.1. Apical supernatants were collected at the times indicated and the titers determined by TCID₅₀ assay. *, P < 0.05 by two-way analysis of variance (ANOVA) with repeated measures. Data represent two independent experiments with n = 3/group. Error bars represent SD. (B) sTEC and siTEC were infected as in panel A. Data represent two independent experiments of n = 2/group. Error bars represent SD.

We next compared the replication of a larger panel of human, avian, and swine H3 viruses (Table 1) in siNEC, siTEC, and MDCK cells. In general, the siTEC and MDCK cells had similar kinetics and titers (Fig. 5). Of the viruses tested, only A/Bethesda/55/2013 (H3N2) and A/red-gartered coot/Chile/C16030/2016 (H3N4) viruses replicated more effectively in MDCK cells. In contrast, A/Switzerland/2013 (H3N2) reached titers of 10⁹ 50% tissue culture infective dose (TCID₅₀)/ml by 72 hpi in siTEC but was unable to replicate in MDCK cells, possibly due to reduced α2,6-sialic acid receptor expression compared to respiratory epithelial cells (10, 11). Overall, the siNEC did not support replication as effectively as siTEC or MDCK cells; however, the siNEC continued to support replication of CA/09 (H1N1), and we detected ∼10⁶ TCID₅₀/ml infectious virus particles until at least 13 days postinfection (dpi) (data not shown) with no visible CPE, highlighting that these cells might be persistently infected.

FIG 5.

Viral growth kinetics in fully differentiated immortalized swine cells and MDCK cells. (A) Fully differentiated immortalized swine cells and confluent MDCK cells were infected with A/California/04/2009 (H1N1), A/Switzerland/9715293/2013 (H3N2), A/Bethesda/55/2013 (H3N2), A/red-gartered coot/Chile/C16030/2016 (H3N4) (RGC), A/swine/Iowa/13-105/2010 (H3N2), A/swine/Indiana/08/2011 (H3N2v), A/swine/North Carolina/0668/2011 (H3N2), or A/swine/Texas/4119-2/1998 (H3N2) at an MOI of 0.1. Apical (siNEC and siTEC) and cellular (MDCK) supernatants were collected at the times indicated and titers determined by TCID₅₀ assay. *, P < 0.05 by two-way ANOVA with repeated measures. Error bars represent SD.

Swine cell lines support viral isolation.

An important component of influenza surveillance is the successful isolation of viruses from clinical samples. To determine whether temperature affected viral replication of the primary isolates, monolayers of siNEC, siTEC, and MDCK cells were inoculated with an H1N2 virus isolated from the oral fluid swabs. We also inoculated cocultures of siNEC and siTEC to determine whether viral replication would be more efficient if both nasal and tracheal cells were present. After cells were incubated at 32°C or 37°C for 24 to 72 hpi, supernatants were collected, and viral titers were determined by TCID₅₀ assay on MDCK cells. Overall, viral titers were higher in the swine cell lines than in MDCK cells, especially at 32°C (Fig. 6A and B). Viral titers in the cocultures were comparable to the siNEC and siTEC alone.

FIG 6.

siNEC and siTEC support viral isolation. Monolayers of siNEC, siTEC, or cocultures were inoculated with a swine H1N2 isolate and cultured at 32°C (A) or 37°C (B). Supernatants were collected 3 to 5 days postinoculation and titers of MDCK cells determined by TCID₅₀ assay. Error bars represent SD. (C) siNEC, siTEC, or MDCK were inoculated with universal transport media (UTM) taken from swabs collected from infected swine. Supernatants were collected 3 to 5 days postinoculation and titers of MDCK cells determined by TCID₅₀ assay. Samples represent two independent experiments; n = 2/group. *, P < 0.05 by unpaired t test. Error bars represent SEM. (D) siNEC, siTEC, MDCK, and MDCK-SIAT cells were inoculated with UTM taken from swabs collected from humans known to be infected with swine influenza viruses. Supernatants were collected 3 to 5 days postinoculation and titers of MDCK cells determined by TCID₅₀ assay. Data represent two independent experiments; n = 2/group. *, P < 0.05 by unpaired t test. Error bars represent SEM. Dotted line represents limit of detection.

To determine whether the swine cell lines could be used for surveillance, 25 influenza reverse transcription-quantitative PCR (qRT-PCR)-positive oral fluid or nasal swabs from swine were inoculated onto differentiated swine cell lines or MDCK cells, supernatants were collected at 72 hpi, and viruses were quantified by TCID₅₀ assay on MDCK cells. We successfully isolated viruses from 7 of the 25 samples. We grew 3 isolates from siNEC, siTEC, and MDCK and 3 isolates from swine cells but not MDCK. Only one of the isolates was rescued from MDCK cells alone (Table 2 and Fig. 6C). The viral titers of the isolates harvested from the immortalized swine cells were generally 1 to 2 logs higher than those isolated from MDCK cells.

TABLE 2.

TCID₅₀ of influenza viruses isolated from swine surveillance samples cultured in different cell lines

Sample no.	TCID50/ml of cell linea:
	siNEC	siTEC	MDCK
1	102.5	102.5	102.5
2	104.5	NT	NT
3	103.5	103.5	102.5
4	104.5	103.5	103.5
5	104.5	104.5	NT
6	NT	102.5	NT
7	NT	NT	102.5

^aNT, no titer.

To assess isolation rates from human clinical samples,18 qRT-PCR-positive nasopharyngeal swabs from people with influenza-like illness who were exposed to pigs were inoculated onto differentiated swine cell lines, MDCK and MDCK-SIAT cells, and supernatants collected at 72 hpi. Both qRT-PCR-positive and -negative samples were provided, and laboratory members were blinded to the results. MDCK-SIAT cells isolated the most viruses (14/18 samples), followed by MDCK and siNEC, which isolated 12/18 samples, and siTEC, which recovered virus from 8/18 samples (Table 3). However, the viral titers obtained from MDCK-SIAT were significantly lower than those obtained in differentiated swine cell lines (Table 3 and Fig. 6D).

TABLE 3.

Isolation of influenza virus from human surveillance samples

Sample	Subtype	TCID50/ml of cell line:
		siNECa	siTECa	MDCKa	MDCK-SIATa
A	H1N2	NT	NT	NT	NT
B	H1N2	106.7	NT	NT	NT
C	H1N1	108.5	108.5	103.7	103.3
D	H1N1/N2	NT	NT	103.5	102.6
E	H3N2	108.7	NT	NT	NT
F	H1N1	NT	NT	NT	102.5
G	H1N2	108.5	108.3	106.5	103.3
H	H1N1	107.0	108.0	106.5	105.3
I	H1N2	106.7	106.5	104.7	104.3
J	H1N2	107.0	106.5	108.5	104.5
K	H1N1	107.5	108.5	107.7	104.3
L	H1N2	106.7	106.5	108.3	106.5
M	H1N1	107.7	107.5	108.5	104.8
N	H3N2	105.5	NT	103.7	102.7
O	H3N2	NT	NT	NT	NT
P	H1N2	NT	NT	NT	102.4
Q	Unknown	NT	NT	107.5	102.7
R	H1/H3N2	107.3	NT	107.7	104.5

^aNT, no titer.

Adaptation of human seasonal and zoonotic influenza viruses to cell culture can result in accumulation of mutations (38, 39). Preserving the genomic integrity of influenza virus is crucial in risk assessment and in vaccine production, where cell and egg adaptations can alter antigenic sites, reduce vaccine efficacy, and alter viral growth kinetics. Seasonal influenza viruses propagated through MDCK-SIAT cells show higher genomic stability than those passaged in MDCK cells (40). To determine whether swine-origin influenza viruses show greater genetic stability in siNEC than in MDCK cells, we serially passaged a human H3N2 and a swine H3N1 isolate in duplicate for 6 rounds in siNEC or MDCK cells and performed whole-genome sequencing to quantitate consensus changes and minor variants. Upon propagation in MDCK cells, the human virus acquired dominating mutations in hemagglutinin (HA) G234R and neuraminidase (NA) E433G (Table 4 and Fig. 7). These mutations were not present in the siNEC-passaged virus. No variants detected in siNEC-passaged human H3N2 virus were found at a rate greater than 50% of the population. Propagation of the swine H3N1 virus in MDCK and siNEC cells yielded no variants at a rate greater than 50% of the population (Table 4). Our results suggest that immortalized swine respiratory cells conserve genomic integrity for certain viruses.

TABLE 4.

Combined mutations greater than 10%

Segment	Nucleotide mutation	Amino acid annotation	Frequency (%) in indicated cell line of mutation by:
			Human H3N2a		Swine H3N1b
			MDCK	siNEC	MDCK	siNEC
PB1	G→A	E256K (GAG→AAG)				11.3
	A→G	K388R (AAA→AGA)				15.1
	A→G	Y436C (TAT→TGT)				12.1
	C→T	Q758c (CAA→TAA)		12.5
PB2	A→G	I478V (ATA→GTA)			10.5
PA	A→C	E293D (GAA→GAC)		39.1
HA	G→A	R62K (AGA→AAA)				40.2
	G→A	V178I (GTC→ATC)			37.3
	G→A	G234R (GGG→AGG)	55.3
	C→T	A228V (GCT→GTT)	16.7
	A→C	E319D (GAA→GAC)				40.4
	G→A	G395E (GGG→GAG)		17.4
	G→A	E422K (GAG→AAG)			11.3
	A→G	N461S (AAC→AGC)				44.5
NP	C→A	T130K (ACA→AAA)			10.7d
NA	A→G	I7V (ATA→GTA)			13.2
	A→G	N309S (AAT→AGT)			16.9
	A→G	E433G (GAA→GGA)	83.5d

^aShannon’s entropy for human H3N2 virus passaged in MDCK,1.12, and in siNEC, 1.34.

^bShannon’s entropy for swine H3N1 virus passaged in MDCK, 2.39, and in siNEC, 2.71.

^cStop codon.

^dMutations found in 2 of 2 replicates with average frequency reported. All other mutations found in 1 of 2 replicates.

FIG 7.

Dominating mutations in MDCK-passaged viruses emerge in functional HA regions and NA epitopes. Amino acid variants that emerged at greater than 50% of the viral population in MDCK-passaged human H3N2 and swine H3N1 viruses were mapped onto representative structures by using PyMOL. (A) HA mutation G234R (55.3% relative frequency) mapped onto the hemagglutinin stalk region (PDB: 3TIA; space-filling model). Inset shows ribbon structure detail of wild-type protein (top) and mutated variant (bottom). Amino acid location 243 is highlighted in teal. (B) NA mutation E433G (83.5% relative frequency) mapped onto neuraminidase (PDB: 4WE8; space-filling model). Inset shows ribbon structure detail of wild-type protein (top) and mutated variant (bottom). Amino acid location 433 is highlighted in orange.

DISCUSSION

Swine are considered a ”mixing vessel,” supporting the replication of mammalian and avian strains (41), making primary swine respiratory epithelial cells an excellent model for studying influenza viruses (20, 21, 27, 28). Yet primary cells are difficult to maintain, have a limited passage number, and are often contaminated with fibroblasts. Thus, we set out to immortalize swine respiratory epithelial cells. Unlike primary swine cells, siNEC and siTEC cells could be passaged multiple times, minimizing the use of donor animals and the time-consuming isolation process. Despite immortalization, siNEC and siTEC cells retained the abilities to form tight junctions, secrete mucus, and, in some cases, form cilia (Fig. 2), suggesting a heterogeneous cell population that can recapitulate the respiratory tract comparable to primary cell models (21, 23, 24, 26, 42).

A growing body of literature suggests that the upper and lower respiratory tract may have different responses to infection. For example, interferon lambda (IFN-λ) response in the nasal mucosa was protective against influenza infection, but this was not the case in the lower respiratory tract (43). Similar studies with Sendai virus reported that infection in the nasal cavity elicited protective immunity compared to the lower respiratory tract as the site of infection (44). Differences in receptor distribution between the upper and lower respiratory tract also impact influenza infection (45–47). Thus, immortalized nasal and tracheal/bronchial epithelial cells will be a valuable tool to further study tissue tropism-specific differences.

The currently circulating human seasonal H3N2 influenza viruses are challenging to propagate in cell culture (3). MDCK cells, the gold standard for influenza virus culture, have low expression of α2,6-sialic acid receptors, the preferential binding target for human influenza viruses (3, 27, 48), but the creation of MDCK-SIAT cells has somewhat alleviated this problem (8, 10). The immortalized swine cells expressed both α2,3- and α2,6-sialic acid receptors (Fig. 3) and supported influenza replication without the addition of exogenous trypsin, suggesting the presence of endogenous proteases like primary respiratory cells (49, 50). Human viruses and viral isolation from surveillance samples were able to grow well in siNEC and siTEC, although the siNEC generally had delayed viral kinetics and lower titers compared to siTEC (Fig. 5). One possibility is the lack of ciliated cells present in the siNEC (Fig. 2). Previous studies have found that human H3 viruses preferentially bind ciliated epithelial cells with α2,6 linkages (26, 48). Swine viruses replicated to high titers in both siNEC and siTEC, and we saw significant CPE in both nasal and tracheal cells (Fig. 4 and 6). Interestingly, siNEC shed (CA/09) (H1N1) for >13 days, which could be useful in cell-based vaccine strain production.

Adding the swine cells to surveillance studies could greatly improve the rate of successful isolation of emerging zoonotic viruses and human seasonal viruses. The number of PCR-positive influenza samples that yield negative results in cell culture has increased dramatically since 2010 (51). Using the immortalized swine cells, we were able to isolate viruses from surveillance samples that we were unable to isolate in MDCK and/or MDCK-SIAT cells. Conversely, some viruses could not be isolated in siNEC or siTEC but were rescued from MDCK and/or MDCK-SIAT cells. One caveat is that fully differentiated siNEC and siTEC were used to generate most of our isolates, meaning the immortalized swine cells would need to be cultured for 2 to 3 weeks before use. We had limited success when performing these experiments in monolayers (Fig. 6A and B).

As the virus is passaged through eggs or MDCK cells, selection of minor variants for egg and cell adaptations can emerge, which may alter antigenicity, decrease vaccine efficacy, and lead to inaccurate calculation of zoonotic risk (3, 52–54). Although genomic integrity is somewhat maintained in viruses grown in MDCK cells rather than in eggs (8, 55), we found dominating HA and NA mutations in MDCK-passaged viruses that were not present in siNEC-passaged viruses. While mutations at >10% did emerge in passage through both MDCK cells and siNEC, only MDCK-passaged viruses had detected variants in greater than 50% of sequence reads, including mutations detected at consensus at positions NA E433G and HA G234R (Fig. 7; Table 4).

In this study, we developed immortalized swine respiratory epithelial cells that retained the ability to differentiate and could support robust influenza replication. Using immortalized swine cells not only improved the overall viral isolation rate from clinical samples but, in many cases, generated higher-titer isolates than MDCK or MDCK-SIAT cells. Taken together, our immortalized swine cells have the potential to improve influenza surveillance and advance the field of influenza research and vaccine development.

MATERIALS AND METHODS

Viruses and cells.

A/California/04/2009 (H1N1), A/Switzerland/9715293/2013 (H3N2), A/swine/Iowa/13-1015/2010 (H3N2), and A/red-gartered coot/Chile/C16030/2016 (H3N4) were propagated in 10-day-old embryonated chicken eggs as previously described (5). A/swine/Indiana/08/2011 (H3N2), A/swine/North Carolina/0668/2011 (H3N2), and A/swine/Texas/4119-2/1998 (H3N2) were propagated in MDCK cells (ATCC). A/Bethesda/55/2013 (H3N2) was propagated in MDCK-SIAT cells (a gift from Richard Webby, St. Jude Children’s Research Hospital). Influenza viral titers were determined by TCID₅₀ assay in MDCK cells as previously described (56). MDCK and MDCK-SIAT were cultured in modified Eagle’s medium (MEM) (Corning) containing 200 mM GlutaMax (Gibco) and 10% fetal bovine serum (FBS) (Atlanta Biologicals).

Isolation and culture of immortalized and primary swine cells.

Swine nasal cell brushings and trachea were collected and placed in ice-cold isolation media (Dulbecco’s MEM [DMEM]/F12 [Corning] supplemented with penicillin-streptomycin [100 μg/ml and 100 U/ml, respectively; Gibco] and amphotericin B [0.25 μg/ml]).

(i) Nasal. Cells were collected by swabbing the distal nasal cavity with a FLOQSwab cytology brush (Copan Diagnostics). Brushes were rinsed in phosphate-buffered saline (PBS) to dislodge cells, and the sample was centrifuged. The pellet was resuspended in growth media (DMEM/F12 supplemented with GlutaMax [200 mM], insulin [10 μg/ml; Sigma], transferrin [5 μg/ml; Sigma], cholera toxin [100 ng/ml; Sigma], human epidermal growth factor [25 ng/ml; Thermo Fisher], bovine pituitary extract [30 μg/ml; Corning], 5% FBS, retinoic acid [5 × 10⁻⁷ M; Sigma], penicillin-streptomycin [100 μg/ml; 100 U/ml], and amphotericin B [0.25 μg/ml]). Cells were plated onto rat-tail collagen (Corning)-coated plates at a density of 5 × 10⁵ cells/cm² and incubated at 37°C and 5% CO₂.

(ii) Trachea. Tissues were digested overnight in 0.2% protease XIV (Sigma) in isolation media at 4°C. After the digestion was complete, the interior side of the trachea was lightly scraped with a scalpel to dislodge cells. The supernatant was collected, filtered through a 70-μm cell strainer, and centrifuged. Red blood cells were lysed prior to digestion with DNase (Sigma). Cells were centrifuged and resuspended in isolation media supplemented with 10% FBS and incubated at 37°C and 5% CO₂ in a tissue culture dish for 2 to 4 hours. After the incubation, nonadhered epithelial cells were collected, while contaminating fibroblasts remained adhered to the culture vessel. Cells were centrifuged and resuspended in growth media. Cells were plated onto rat-tail collagen-coated plates at a density of 1 × 10⁵ cells/cm² and incubated at 37°C and 5% CO₂.

(iii) Immortalization of primary swine cells. sNEC passage 13 and sTEC passage 14 were plated in growth media at a density of 5 × 10⁵ in a 6-well plate. Once cells reached 70% confluence, the medium was removed; cells were washed 3 times with PBS and immortalized by using SV40, following the manufacturer’s instructions (abm) and incubated for 8 hours. The infection medium was removed, and cells were reinfected with SV40 overnight. After washing, cells were cultured in growth media until 80% confluent. Cells were passaged >15 times, beyond the limit of primary sNEC and sTEC, and could be successfully passaged up to 30 times.

(iv) Differentiation of immortalized and primary swine cells. Cells were plated onto 24-well transwell inserts coated with rat-tail collagen in growth media. Once cells became 100% confluent and had transwell epithelial resistance (TER) of 1,000 Ω or greater, the apical medium was removed to create an air-liquid interface (ALI). The growth medium was removed and replaced with ALI media (DMEM/F12 supplemented with 200 mM GlutaMax, 2% Nu-Serum [Corning], retinoic acid [5 × 10⁻⁷ M], penicillin-streptomycin [100 μg/ml and 100 U/ml, respectively], and amphotericin B [0.25 μg/ml]) in the basolateral compartment only. Cells were cultured at 37°C and 5% CO₂ for 2 to 3 hours, with the media changed every 48 hours until they became differentiated. For monolayer cultures, cells were plated at a density of 2 × 10⁵ in a 24-well plate, and for cocultures, equivalent numbers (1 × 10⁵) of siNEC and siTEC cells were cultured in a 24-well plate.

Flow cytometry.

Cells were collected by scraping gently and blocked in 5% normal goat serum (Gibco) before staining with anti-vimentin (Abcam; clone EPR3776; 1:100) followed by anti-rabbit Pacific blue-conjugated secondary antibody (Invitrogen; 1:200). Cells were analyzed on the BD LSRFortessa X-20.

TER.

Media were added to the apical compartment of differentiated swine cells. Cells were incubated at 37°C and 5%CO₂ for 30 minutes to equilibrate. TER was measured with an EndOhm chamber and EVOM² epithelial voltohmmeter (World Precision Instruments). After TER measurement, the apical medium was removed.

Immunofluorescent staining and confocal microscopy.

Differentiated cells were fixed in 4% paraformaldehyde (Polysciences Inc.) at room temperature for 15 minutes and permeabilized in 0.1% Triton X-100 (Sigma) for 10 minutes. Membranes were blocked in 2% normal goat serum and 0.5% bovine serum albumin (BSA) (Gibco) before staining with anti-β-catenin (Abcam; clone E247; 1:100), anti-β-tubulin (BioLegend; clone TU27/tubulin; 1:100), anti-MUC5AC (Abcam; clone EPR16904; 1:100), anti-MUC1 (Abcam; clone EP1024Y; 1:100), and either biotinylated Maackia amurensis lectin II (Vector Laboratories; 10 μg/ml) to detect α2,3-linked sialic acid receptors followed by Texas red-conjugated streptavidin (Vector Laboratories; 15 μg/ml), or fluorescein isothiocyanate (FITC)-conjugated Sambucus nigra agglutinin (Vector Laboratories; 10 μg/ml) to detect α2,6-linked sialic acid receptors. After being washed, cells were stained with anti-mouse Alexa Fluor 488 (Invitrogen; 1:500), anti-goat Alexa Fluor 555 (Invitrogen; 1:500), and 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher; 1:1,000) before visualization. Cells were imaged using a Zeiss LSM 780 microscope. To quantify sialic acid expression, α2,6- and α2,3-positive cells were counted from 8 fields chosen at random. Data are presented as a percentage of the total number of cells in each field.

Influenza infection of differentiated cells.

The apical surface was washed twice with PBS to remove mucus. Cells were inoculated with virus (MOI of 0.1) diluted in infection media (MEM supplemented with 200 mM GlutaMax and 0.075% BSA) in the absence of trypsin. After 2 hours, the inoculum was removed, and the apical surface was washed twice with PBS. Cells were incubated at 37°C and 5% CO₂ for the indicated times. The infection medium (100 μl) was added to the apical compartment, and cells were incubated for 30 min at 37°C and 5% CO₂ before the apical supernatant was collected. Virus was quantified by TCID₅₀ assay using MDCK cells and read using hemagglutination assay with 0.5% turkey red blood cells. Titers were calculated by the method of Reed and Muench (56).

Isolation of virus from surveillance samples.

For monolayer cultures, cells were plated in 24-well tissue culture plates as described above and grown to confluence. After washing with PBS, cells were inoculated with 200 μl of DMEM/F12 containing 0.075% BSA, penicillin-streptomycin (100 μg/ml; 100 U/ml) containing 0.5 μg/ml (swine cells) or 1 μg/ml (MDCK and SIAT cells) tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-trypsin (Worthington), and 10 to 50 μl of primary sample, depending on available volume. Cells were incubated for 3 to 5 days while CPE was monitored. Samples were evaluated in duplicate.

For differentiated cultures, cells were plated at 3 × 10⁴ in 24-well transwell inserts and cultured at the air-liquid interface for 2 to 3 weeks to allow differentiation. After the apical surface was washed with PBS, 50 μl of DMEM/F12 containing 0.075% BSA, penicillin-streptomycin (100 μg/ml and 100 U/ml, respectively), and 10 to 50 μl of primary sample (depending on available volume) was added to the apical compartment for 1 h at 37°C and 5% CO₂ in the absence of TPCK-trypsin. The apical supernatants were removed, and cells were incubated for 3 to 5 days, depending on CPE. Supernatants were collected and stored at −70°C until use.

Viral passaging and sequencing.

MDCK cells (2 × 10⁵) or siNEC (2 × 10⁵) were plated in a 24-well tissue culture plate and incubated overnight. Cells were washed twice with PBS and infected with the following influenza viruses (MOI of 0.1): a swine H3N1 isolate and a human H3N2 isolate. After a 72-hour incubation, P1 supernatant was collected and viral titers determined TCID₅₀ on MDCK cells. Freshly plated cells were infected as above with the P1 virus, and passaging was repeated 6 times.

Viral RNA was extracted from 50 μl of pooled cellular supernatants on a Kingfisher Flex magnetic particle processor (Thermo Fisher Scientific) using the Ambion MagMax-96 AI/ND viral RNA isolation kit (Applied Biosystems). The concentration of the extracted viral RNA was determined spectrophotometrically (NanoDrop spectrophotometer) prior to multisegment PCR (MS RT-PCR) using SuperScript III one-step RT-PCR system with platinum Taq high-fidelity DNA polymerases (Thermo Fisher) and an influenza-specific set of universal primers (57) (Opti-F1, 5′-GTTACGCGCCAGCAAAAGCAGG-3′; Opti-F2, 5′-GTTACGCGCCAGCGAAAGCAGG-3′; and Opti,-R1 5′-GTTACGCGCCAGTAGAAACAAGG-3′). The following cycling parameters were then followed: 55°C for 2 min, 42°C for 60 min, 94°C for 2 min, 94°C for 30 s (5 cycles), 44°C for 30 s, 68°C for 3.5 min, 94°C for 30 s (26 cycles), 57°C for 30 s, 68°C for 3.5 min, 68°C for 10 min, and a hold at 4°C. We analyzed 5 μl of the reaction by 0.8% agarose gel electrophoresis to verify the presence of all genomic segments, with PB1 and PB2 migrating together at 2.3 kb and minimal nonspecific amplification below 800 bp.

The MS RT-PCR was purified by using Agencourt AMPure XP kit (Beckman Coulter) according to the manufacturer’s instructions. Briefly, 40 μl of the DNA amplicons was transferred to a 96-well microplate (Bio-Rad) and mixed with Agencourt AMPure XP beads (Beckman Coulter) for magnetic separation of the amplicons. The bound beads were washed twice with 100% ethanol, and the purified amplicons were suspended in 1× Tris buffer (TE). The concentration of the purified DNA was measured spectrophotometrically prior to storage at −20°C (NanoDrop spectrophotometer). DNA amplicons were deep sequenced using Illumina MiSeq technology. Deep sequencing was performed by the St. Jude Children’s Research Hospital Hartwell Center with DNA libraries prepared using Nextera XT DNA-Seq library prep kits (Illumina) with 96 dual-index barcodes and sequenced on an Illumina MiSeq personal genome sequencer to quantify the presence of single-nucleotide variants (SNVs) in passaged and parental viruses.

Reads were mapped to the reference sequence by using CLC Genomics Workbench 9 (Qiagen) (58). Briefly, fastq files were imported and sequences trimmed to remove low-quality bases (<Q30) and reads shorter than 50 bases. Reads were aligned to the reference genome, and the low-frequency variant detection tool was used to identify SNVs at greater than 10% relative frequency. Variants were filtered to remove low-quality variants. Total nucleotide diversity was measured using Shannon’s entropy (H) as follows: $H\left(x\right)=\sum _i^{n}P\left(i\right){\log }_2\hspace{0.17em}P\left(i\right)$ by using the relative frequencies, or P(i), of each SNV (59). Diversity calculations were summed across all eight gene segments with absolute values reported.

Statistical analysis.

Statistical analysis was done using GraphPad Prism v8 as described in figure legends. P < 0.05 was considered significant.

Data availability. Raw sequence reads are available under BioProject number PRJNA666647.