Innate antiviral responses in porcine nasal mucosal explants inoculated with influenza A virus are comparable with responses in respiratory tissues after viral infection

Nasal mucosal explant (NEs) cultured at an air–liquid interface mimics in vivo conditions more accurately than monolayer cultures of respiratory cell lines or primary cells cultured in flat-bottom microtiter wells. NEs might be relevant for studies of host-pathogen interactions and antiviral immune responses after infection with respiratory viruses, including influenza and corona viruses. Pigs are natural hosts for swine influenza A virus (IAV) but are also susceptible to IAV from humans, emphasizing the relevance of porcine NEs in the study of IAV infection. Therefore, we performed fundamental characterization and study of innate antiviral responses in porcine NEs using microfluidic high-throughput quantitative real-time PCR (qPCR) to generate expression profiles of host genes involved in inflammation, apoptosis, and antiviral immune responses in mock inoculated and IAV infected porcine NEs. Handling and culturing of the explants ex vivo had a significant impact on gene expression compared to freshly harvested tissue. Upregulation (2–43 fold) of genes involved in inflammation, including IL1A and IL6, and apoptosis, including FAS and CASP3, and downregulation of genes involved in viral recognition (MDA5 (IFIH1)), interferon response (IFNA), and response to virus (OAS1, IFIT1, MX1) was observed. However, by comparing time-matched mock and virus infected NEs, transcription of viral pattern recognition receptors (RIG-I (DDX58), MDA5 (IFIH1), TLR3) and type I and III interferons (IFNB1, IL28B (IFNL3)) were upregulated 2–16 fold in IAV-infected NEs. Furthermore, several interferon-stimulated genes including MX1, MX2, OAS, OASL, CXCL10, and ISG15 was observed to increase 2–26 fold in response to IAV inoculation. NE expression levels of key genes involved in antiviral responses including IL28B (IFNL3), CXCL10, and OASL was highly comparable to expression levels found in respiratory tissues including nasal mucosa and lung after infection of pigs with the same influenza virus isolate.


Introduction
Influenza A virus (IAV) is a zoonotic respiratory pathogen of global importance in veterinary and human health. Although aquatic birds are the natural reservoir for IAV, other species such as humans and pigs can be hosts. IAV belongs to the Orthomyxoviridae family and is a singlestranded, negative-sense RNA virus with a characteristic segmented genome. The surface glycoproteins hemagglutinin (HA) and neuraminidase (NA) determine the virus subtype (Yoon et al., 2014).
IAV infection in mammals is in general restricted to the respiratory tract. IAV enters the hosts through the nasal cavity, where they encounter the mucosal surface as the first barrier towards infection (Starbaek et al., 2018). Infection of host cells is subsequently mediated through attachment of HA to sialic acid(SA)-coated surface glycoproteins of the respiratory epithelium. The configuration of the SA-linkage is considered a major determinant of IAV host specificity, as avian IAV prefer binding to α2,3-linked SAs, while mammalian IAV generally prefer α2,6-linked SAs (Webster et al., 1992;Byrd-Leotis et al., 2017).
Nasal mucosal explants (NEs) cultured at an air-liquid interface resemble the in vivo situation more accurately than cells grown in 2D flasks or culture plates. NEs maintain tissue complexity and cell-cell interactions including apical tight junctions, intermediate junctions, and desmosomes of the nasal mucosa of healthy individuals (Denney and Ho, 2018). Furthermore, porcine NEs are easily acquired from slaughterhouses and have been shown to remain viable and exhibit minimal changes in morphology, ciliary beating, and number of apoptotic cells for up to 72 h of cultivation at an air-liquid interface (Glorieux et al., 2007;Tulinski et al., 2013). Porcine NEs therefore represent a highly relevant viral infection model for studies of host-pathogen interactions and pathogenesis. Importantly, using NEs as a replacement for live animals is in accordance with the 3R principle, seeking to reduce the number of animals included in a given study (Tannenbaum and Bennett, 2015).
The similarity of the anatomy (e.g. epithelial cell distribution) of the upper respiratory system including the nasal cavity of pigs and humans has recently been reviewed by us and others (Rajao and Vincent, 2015;Iwatsuki-horimoto et al., 2017;Starbaek et al., 2018). Distribution and quantities of mucus-producing goblet cells and ciliated epithelial cells are highly similar, as is the distribution of SA-coated viral receptors in nasal cavities of pigs and humans, thus rendering porcine NEs a promising model also for human respiratory infections (Spicer et al., 1983;Wallace et al., 1994;Shinya et al., 2006;Zhang et al., 2009;Trebbien et al., 2011).
The antiviral immune response to IAV infection is initiated by recognition of the viral pathogen by pattern recognition receptors (PRR) of the host cells in the nasal mucosa and along the respiratory tract. PRRs such as RIG-I (DDX58), TLR3, and MDA5 (IFIH1) will detect viral RNA in the host cell cytoplasm and activate the expression of type I and III interferons (IFNs), which induce the expression of a wide range of antiviral interferon-stimulated genes (ISGs) in infected and neighboring cells (Kato et al., 2006;Pichlmair et al., 2006;Brogaard et al., 2018). The local production of ISGs at the site of infection establishes an antiviral state, where components like MX1, OAS1, OASL, and CXCL10 are important for controlling the IAV replication and production of infectious viral progeny during the first days of disease (Skovgaard et al., 2013;Delgado-Ortega et al., 2014;Kim et al., 2015;Brogaard et al., 2018).
To examine the utility of porcine NEs for the study of host responses to IAV infection, 69 NEs isolated from 11 different pigs were used in this study to analyze the transcriptional response to culturing and to IAV exposure.

Animals
Data presented in this study originates from three independent trials, comprising 69 NEs collected from 11 different pigs (cross-bred Landrace × Yorkshire × Duroc) (Table 1). Animals were acquired from farms of high health status without prior history of respiratory infections. No IAV-specific antibodies were detected in serum samples from the animals prior to the experiment using a commercial ELISA Kit (IDEXX) following the manufacturer's instructions.
Trial 2: 12 NEs were obtained from each of two 6-weeks old piglets at Ghent University (total no. of explants = 24). Animals were obtained, treated, and euthanized as described by Glorieux et al. (2007).
Trial 3: Two NEs from each of six 6-weeks old piglets were included in this study (total no. of explants = 12). These animals originated from the same farm as trial 1. Euthanasia was performed as described in trial 1, but animals were not treated with antibiotics prior to euthanasia.
All work has been carried out in accordance with the EU Directive 2010/63/EU for animal experiments.
Gene expression data from nasal mucosal tissue and lung tissue were included for benchmarking our nasal mucosal explants to respiratory tissue isolated from IAV infected pigs Starbaek et al. in prep). The same A/swine/Denmark/12687/2003 (H1N2) isolate was used for inoculation in all included studies. Lung samples from cross-bred Large White × German Landrace challenged by aerosol exposure were obtained 72 hpi (infected n = 6 and control n = 5) and nasal mucosal tissue from Göttingen Minipigs inoculated by an Table 1 Sampling and inoculation of nasal mucosal explants. Trial 1 was conducted at the National Veterinary Institute, Technical University of Denmark and included three 6-weeks old pigs, from each of which 11 nasal mucosal explants (NEs) were isolated. Trial 2 included two 6-weeks old pigs from each of which 12 NEs were isolated. This trial was performed at Ghent University, Belgium. Trial 3 was conducted at the National Veterinary Institute, Denmark and included six 6-week old pigs from each of which two explants were isolated. Hours post inoculation (hpi). intranasal mucosal atomisation device were obtained 4 days post infection (infected n = 7 and control n = 5). Further details have been described in Starbaek et al. in prep respectively).

Isolation, culture, inoculation, and sampling of nasal mucosal explants
NEs from all trials were isolated and cultured as described by Glorieux et al. (2007) with minor modifications. Briefly, after euthanasia snouts were sawn off the skull in front of the eyes and stripped of skin and muscle tissue. The snouts were split open and mucosa was stripped from the underlying cartilage in the nasal cavity (septum and conchae) and divided into 0.25 cm 2 squares and placed with the epithelium surface upwards on modified cell strainers (VWR) (trial 1 and 3) or stainless-steel supports (trial 2), dimensioned to support the NEs at an air-liquid interface. At this time point (− 20 h post inoculation (hpi)), three NEs (trial 1) were harvested and stored in RNAlater (Qiagen). NEs were supplied with serum-free culture medium from the basal side. The NEs were cultured for approximately 20 h at 37 • C and 5% CO 2 , before three NEs (trial 1) were collected (0 hpi), divided into two pieces, one half was stored in RNAlater at − 20 • C and the other in PBS at − 80 • C. The NEs were inoculated with 0.6 ml virus suspension containing 10 4 (trial 2) and 10 5 (trial 1) TCID 50 of A/swine/Denmark/12687/2003 (H1N2) or virus-free growth medium (mock) (trial 1, 2, and 3) and incubated for 1 h at 37 • C and 5 % CO 2 . After inoculation, the explants were washed three times in pre-heated culture medium (37 • C) and placed at an air-liquid interface in fresh pre-heated culture medium (37 • C) (Fig. 1). The culture medium consisted of 1:1 RPMI (Gibco) and DMEM (Gibco) supplemented by 0.1 μg/ml gentamicin (Gibco) and 100 U/ml penicillin and 0.1 mg/ml streptomycin (Gibco) (trial 1 and 2). In trial 3, the amounts of penicillin and streptomycin was increased to 1000 U/ml and 1 mg/ml respectively, gentamicin was increased to 0.5 mg/ml and finally 5 μg/ml amphotericin B "Fungizone" was added to compensate for the lack of pre-euthanasia antibiotic treatment in these piglets compared to the method described by Glorieux et al. (2007). The virus growth medium varied slightly in trypsin concentration (1 μg/ml trypsin TPCK (Sigma-Aldrich) (trial 1 and 3) vs. 0.4 μl/ml (trial 2)).
Furthermore, 1 % BME vitamins were added to medium in trial 1 and 3.
NEs were harvested at 1 hpi, 24 hpi, and 48 hpi, one piece stored in RNAlater at − 20 • C and the other in PBS at − 80 • C (trial 1 and 3). NEs from trial 2 were preserved in one piece in RNAlater. NEs from trial 3 were not inoculated with IAV and were only used in combination with uninfected NEs from trial 1 to study gene expression as a consequence of the ex vivo conditions after mock inoculation. Number of pigs and NEs, Fig. 1. Schematic representation of the experimental setup. Nasal mucosal isolation (1), air-liquid interface culture procedure (2-3),inoculation with IAV (4), and further cultivation and harvest (5-6). The figure was created with BioRender.com. as well as treatment and time of harvesting in each trial is summarized in Table 1.

Viability of cells in nasal mucosal explants and bacterial contamination
The extend of apoptosis was assessed in NEs of trial 2 by termal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay as described by Glorieux et al. (2007). Matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF MS) was used to check for bacterial contamination of growth medium before and after NE cultivation as previously described (Nonnemann et al., 2019).

Reverse transcription (RT)-qPCR analysis
Total RNA was extracted from NEs stabilized in RNAlater as described previously (Barington et al., 2018). Briefly, explants were homogenized in QIAzol Lysis reagent (Qiagen) using gentleMACS Mtubes (Miltenyi Biotec). RNA was extracted using the miRNeasy mini kit (Qiagen) and treated with RNase-free DNase set (Qiagen) according to the manufacturer's instructions. RNA quantity, purity, and integrity were estimated using the Nanodrop ND-1000 spectrophotometer (Saveen and Werner AB) and the Agilent Bioanalyzer (Agilent Technologies), respectively. 500 ng total RNA was used for the cDNA synthesis (QuantiTect Reverse Transcription Kit (Qiagen)) and an additional DNase treatment was included. Two cDNA replicates were prepared for each RNA sample and -RT controls (reverse transcriptase replaced with water) were included in the reverse transcription. Preamplification was performed using TaqMan PreAmp Master Mix (Applied Biosystems). A pool of 200 nM qPCR primer mix was prepared by combining primer pairs used in the subsequent qPCR. 5 μl TaqMan PreAmp Master Mix, 2.5 μl 200 nM qPCR primer mix and 2.5 μl cDNA was incubated at 95 • C for 10 min followed by cycles of 95 • C for 15 s and 60 • C for 4 min. Residual primers were digested by adding 16 U of Exonuclease I (New England BioLabs) and incubating at 37 • C for 30 min followed by 80 • C for 15 min.
Microfluidic high-throughput qPCR was performed on a BioMark HD real-time instrument (Fluidigm) as previously described . A number of reference genes were included in the panel of genes to allow normalization of the data. Two different dynamic arrays were used in the present study (GE 96.96 and GE 192.24) combining 96 samples with 96 primer assays and 192 samples with 24 primer assays generating 9216 or 4608 parallel qPCR reactions in a single run, respectively. Primer names, sequences, and length of amplicons used in the present study can be seen in Supplementary Table S1.
After qPCR, data was manually curated using Fluidigm Real-Time PCR Analysis software 3.0.2 (Fluidigm), followed by data preprocessing as previously described (Barington et al., 2018) in GenEx5 (MultiD Analyses AB). GeNorm (Vandesompele et al., 2002) and NormFinder (Andersen et al., 2004) were used to identify the most stable reference genes (using the GenEx5 software). All putative reference genes included (GAPDH, HPRT1, RPL13A, PPIA, YWHAZ, and TBP) were validated as appropriate for normalization and used for data normalization. Relative gene expression levels were calculated after transforming normalized Cq values to relative quantities scaled to the sample in the data set which had the lowest expression of the gene in question. P-values to determine statistical significance of differences in gene expression levels were calculated for mock inoculated explants on LOG2 transformed data using ANOVA, after testing for normal distribution of the data and correcting for multiple testing using Benjamini-Hochberg false discovery rate.

Virus replication
Tissue culture infectious dose (TCID 50 ) was determined as described by Van Poucke et al. (2010) of homogenized NEs, however, the incubation was reduced to three days (trial 1 and 3). Briefly, MDCK cells were inoculated with ten-fold serial dilutions of homogenized NEs collected 24 hpi and 48 hpi and incubated at 37 • C with 5 % CO 2 for three days. Virus replication was confirmed by immunocytochemistry of the MDCK cells, fixed in 99% ethanol and stained using an in-house polyclonal rabbit anti-swine IAV antibody, followed by horse radish peroxidase (HRP) conjugated anti-rabbit immunoglobulins and streptavidin HRP conjugate. The staining was developed by precipitated polymerized ethylcarbazole and inspected using an inverted light microscope. TCID 50 was calculated by the Reed and Muench approach. For trial 2, viral replication was confirmed by determining TCID 50 of the NE culture supernatants. MDCK cells seeded in 96-well plates were inoculated in quadruplicate with 10-fold serial dilutions (ranging from 10 0 to 10 -7 ) of NE culture supernatant collected at 24 hpi and 48 hpi, and incubated at 37 • C with 5 % CO 2 for three days. Induction of cytopathic effect (CPE) was recorded and TCID 50 was calculated by the Reed and Muench approach. Influenza RNA levels in NEs were determined as described above by RT-qPCR targeting the influenza matrix protein gene with the following primers (Nagy et al., 2010); Forward Sequence (5 ′ to 3 ′ ) GGCCCCCTCAAAGCCGA and Reverse Sequence (5 ′ to 3 ′ ) CGTCTACGYTGCAGTCC as well as in-house primers targeting the HA gene. The level of initial viral RNA (average of matrix protein assay and HA assay) from the inoculum was measured at 1 hpi, and was scaled to 1 in order to measure any relative increase from this time point.

Gene expression changes during ex vivo explant culture
Changes in gene expression as a consequence of ex vivo culturing were studied for 82 genes involved in inflammation, apoptosis, and antiviral immune responses (Supplementary Table S1) in mock inoculated porcine NEscollected at necropsy (-20 hpi), after 20 h of acclimatization (0 hpi), and at 24 hpi and 48 hpi. Multivariate analysis (principal component analysis (PCA)) of gene expression data, from a total of 24 explants across the four time points, identified three clusters of NE samples having comparable gene expression profiles (Fig. 2, top  panel). NEs collected at necropsy (-20 hpi) form one cluster (green), those collected following 20 h of ex vivo acclimatization (time 0 hpi) form a second cluster (yellow), while the third cluster is constituted of NE samples collected at both 24 and 48 hpi (red and dark red). Thus, despite of differences in trial 1 and 3, 24 hpi and 48 hpi was clearly separated from earlier time points. RNA quality, shown as RNA integrity numbers (RIN), was not affected by the duration of culturing (Fig. 2,  bottom panel).
Quantification of specific mRNA changes resulting from culturing by itself revealed a significant regulation of 24 genes after correction for multiple testing (see Supplementary Table S2). These genes were mainly involved in inflammation or response to inflammation. However, genes involved in apoptosis and response to viral infection were also regulated (see Supplementary Table S2 and Fig. 2, bottom panel). Importantly, several key genes involved in innate antiviral immune response including viral PRRs (MDA5 (IFIH1) and TLR8), Type I IFNs (IFNA and IFNB1), and ISGs (OAS1, MX1 and RNASEL) were significantly downregulated 2-10 fold during culturing. Genes involved in inflammatory response such as IL1A and IL6 were on the contrary upregulated 4 to 44 fold (Fig. 2, bottom panel) after culturing. Furthermore, the acute-phase protein serum amyloid A (SAA) was also found to be significantly upregulated at all three time points following culturing.

Gene expression changes in mucosal explants exposed to influenza A virus during culture
The response to IAV exposure in explants grown at an air liquid interface were analyzed for genes involved in antiviral immune responses and compared to time-matched mock controls. Expression of several pattern recognition receptors including RIG-I (DDX58), MDA5 (IFIH1), TLR3, and the downstream regulatory factor IRF7 was upregulated in virus inoculated NEs compared to the mock inoculated NEs at 24 hpi (Fig. 3).
Likewise, the expression of MX1 and MX2 was upregulated following IAV inoculation at 24 hpi. Type I IFN (IFNB1) and type III IFN (IL28B (IFNL3)), several ISGs (ISG15, IFITM1, IFITM3, OAS1, OASL, EIF2AK2), and the chemokines CXCL10 and CXCL2 were found to be upregulated between 3 and 26 fold in IAV infected NEs at 48 h after viral exposure, compared to mock inoculated NEs harvested at the same time point (Table 2)  , and transcription factor IRF7 as well as the ISGs MX1 and MX2, 24 h after mock inoculation (blue, scaled to 1) or viral inoculation (red) of porcine NEs. Relative expression is presented as mean relative expression ± SEM. Descriptive statistics were used due to small group size. Trial 1 infected n = 6, control n = 3 and trial 2 infected n = 4, control n = 4. their time matched controls (Table 2).

. Increased expression of both IFNs and ISGs from 24 h to 48 h after IAV infection is seen in all investigated antiviral genes compared to
In order to compare changes in expression of antiviral genes in explants of the present study with respiratory tissue from experimentally IAV infected pigs Starbaek et al. in prep.), selected data of gene expression is illustrated in Fig. 4. Comparable changes in expression of key innate antiviral genes including IL28B (IFNL3), OASL, and CXCL10 were seen between cultured NEs 24 h after viral exposure and nasal mucosal tissue and lung tissue isolated from pigs three to four days after infection with the same influenza isolate, A/swine/Denmark/ 12687/2003 (H1N2) (Fig. 4).
Infectious viral titer and viral RNA levels are seen in Table 3. Virus replication could only be detected in trial 2, at 24 and 48 hpi. Viral RNA was detected in infected NEs by qPCR, and increased over time, but at different rates in trial 1 and 2.
Low to moderate levels of apoptotic cells were detected by TUNEL assay. However, no distinct association was found between the extend of apoptotic cells and cultivation time or infection status of the NEs (data not shown). In contrast, several genes associated with apoptosis including FAS, FOS and CASP3 were upregulated 2-4 fold at the three time points following − 20 hpi. (Table S2). No bacterial contamination was found in the growth medium of explants using MALDI-TOF MS.

Discussion
The local innate immune response to infection by IAV is a highly complex process orchestrated by numerous respiratory epithelial cells and epithelium-associated immune cells. The process of viral recognition and activation of signalling pathways leading to type I and III IFN production and subsequent upregulation of antiviral ISGs is paramount for the early control and elimination of any IAV infection (Bowie and Unterholzner, 2008;Starbaek et al., 2018). The epithelium of the nasal mucosa is composed of ciliated cells, secretory cells, and basal cells attached to each other by tight junctions and immune cells, notably resident macrophages and dendritic cells (Knight and Holgate, 2003;Jahnsen et al., 2004;Beule, 2010). This highly multifaceted respiratory environment can be approximated using nasal explants grown at an air-liquid interface for a limited time period (Glorieux et al., 2007;Van Poucke et al., 2010). After isolation, culturing, and inoculation of porcine NEs, we were able to study the molecular antiviral immune response to IAV at the site of infection, in time-matched mock and virus infected explants up to 48 h after viral exposure.
In the present study no decrease in RNA quality or significant correlation of number of apoptotic cells over time was seen in response to ex vivo culturing, which is in accordance with previous results obtained with this model, showing no or little apoptosis and necrosis up to 96 h of ex vivo culturing (Glorieux et al., 2007;Van Poucke et al., 2010). Not surprisingly, the environmental change from in vivo to ex vivo resulted in increased expression of several pro-inflammatory genes, such as IL1A and IL6 and genes associated with apoptosis including FAS and CASP3. In addition, genes involved in viral pathogen recognition, IFN signalling, and several ISGs were found to be regulated over time during cultivation in the absence of virus. Most of these genes were downregulated. These important and opposite effects on the gene expression of proinflammatory as opposed to antiviral immune genes solely brought about by the change in environment, strongly highlight the importance of including and standardizing data to time and handling matched mock controls when studying molecular antiviral immune responses to IAV infection in NEs.
Following IAV inoculation, the expression of several PRRs important Table 2 Differential expression of antiviral genes from mock and virus inoculated explants 24 h and 48 h after inoculation (trial 2). Data is presented as mean relative expression ± SEM in brackets. Differentially expression of more than two folds in the NE are shown in bold. Descriptive statistics were used due to small group size.  Fig. 4. Comparable expression of key innate antiviral genes within cultured explants at 24 h post inoculation compared with in vivo nasal mucosal tissue and lung tissue isolated from pigs infected with the same IAV isolate. Lung data was obtained 3 days post IAV infection (control n = 5 and infected n = 6). Nasal mucosal tissue was collected 4 days post inoculation (control n = 5 and infected n = 7). Relative expression is presented as mean expression ± SEM. Please note the log2 scale on the Y-axis. Descriptive statistics were used due to small group size.

Table 3
Viral replication determined by infectious dose 50% endpoint (TCID50) in MDCK cells, calculated using Reed and Muench method (Reed and Muench, 1938) and RNA expression measured by RT-qPCR. qPCR data is expressed as mean relative expression ± SEM compared to 1 h post inoculation, nvd = no virus detected. for detection of viral RNA was induced in NEs, including RIG-I (DDX58) and MDA5 (IFIH1). We have previously studied the innate immune response in lungs of IAV infected pigs inoculated with the same viral isolate used in the present study, A/swine/Denmark/12687/2003 (H1N2) (Skovgaard et al., 2013;Brogaard et al., 2018). Although the change in gene expression depends on cell composition of the tissue and number of infecting virus particles, gene expression patterns in NEs, nasal mucosa, and lung tissue show comparable patterns of expression for selected antiviral genes. The moderate to high upregulation of IL28B (IFNL3), OASL, and CXCL10 seen in this study was in accordance with previously reported regulation in lung tissue  as well as nasal mucosal tissue isolated from pigs infected with the same IAV isolate (Starbaek et al. in prep.). A strong but transient IFN response is of great importance during viral infection. In the present study, IFNB1 was upregulated in IAV infected NEs 48 h after virus exposure compared to mock inoculated NEs harvested at the same time point. Upregulation of IFNB1 was also found in human nasal epithelial cells 24 h and 48 h after exposure to human H3N2 by Tan et al. (2018). Type III interferon (IFN-λ) has been shown to be among the primary interferons produced by epithelial and dendritic cells in response to IAV infections of the upper respiratory tract (Jewell et al., 2010;Yan et al., 2016;Klinkhammer et al., 2018;Tan et al., 2018). In the present study, IL28B (IFNL3) was found to be upregulated in infected NEs at both 24 hpi and 48 hpi. The results of both type I and III IFNs mirrors previously results obtained from in vivo infected pig lungs and nasal mucosal tissue with the same viral isolate Starbaek et al. in prep) as well as in human nasal epithelial cells in response to H3N2 IAV infection (Yan et al., 2016;Tan et al., 2018).
MX1 and MX2 were upregulated after viral inoculation in both trial 1 and 2. MX1 is a homolog to the mouse Mx1 involved in inhibition of viral transcription in the nucleus (Dreiding et al., 1985;Verhelst et al., 2012), whereas the porcine MX1 is found in the cytoplasm, where it interferes with transport of viral particles to the nucleus, hence inhibiting viral replication (Palm et al., 2010). Upregulation of MX1 has previously been reported in human nasal epithelial cells infected with H1N1 (Kim et al., 2015) and in human blood samples from individuals infected with influenza virus (Andres-Terre et al., 2015). Upregulation of MX2 has previously been reported in IAV infected porcine lung explants, both in case of single-infection with a H1N1 swine IAV and co-infection with swine IAV (H1N1) and porcine reproductive and respiratory syndrome virus (PRRSV) (Dobrescu et al., 2014) and after infection with H3N2 swine IAV . However, the MX2 gene has to our knowledge, not previously been reported to be involved in response to IAV infection of the upper respiratory system in pigs, and its relevance in porcine nasal mucosal tissue remains to be studied in detail.
Other interferon stimulated genes important for influenza viral restriction early after infection, including CXCL10, OAS1, OASL, and ISG15, were found to be highly upregulated in NEs 48 hpi. CXCL10 has previously been found upregulated in human nasal epithelial cells after infection with human H3N2 (Tan et al., 2018) and CXCL10 and OAS1 have been reported to participate in control of infection through suppression of transcription or replication of IAV in human nasal epithelium (Kim et al., 2015). Furthermore, OAS becomes catalytically active in the epithelial cell cytosol upon binding of viral double stranded RNA (dsRNA) during the IAV replication cycle. This activation promotes OAS mediated degradation of viral RNA through RNase L activity and in turn inhibits viral replication (Min and Krug, 2006;Drappier and Michiels, 2015).
Replication of IAV in NEs has previously been reported (Van Poucke et al., 2010). Viral RNA measured by qPCR increased modestly in trial 1 compared to trial 2, and no infectious virus was detected in trial 1. However, reduced viral replication seemed to have only minor effects on the magnitude of antiviral immune responses measured in NEs. Comparing the patterns of gene regulation between trial 1 and 2 for key antiviral genes revealed no systematic differences. Future studies are needed to address the difference in viral replication in order to obtain more reproducible results.
In general, the use of NEs provides easily accessible, cheap samples, and limited space requirements compared to a similar in vivo setup with live pigs. However, it of course comes with some limitations; The NEs are removed from the organism and its natural environment. The tissue lacks the lymph system and influx of immune cells and other immunerelated components. Future studies should hold more biological replicates in order to allow statistical analysis on all data. Furthermore, the experimental settings should be standardized completely between different trials so pigs and NEs are treated under the same conditions and inoculated with the same amount of virus particles. The lack of antibiotic treatment in trial 3 was compensated by using time-matched controls to adjust for a possible difference in the baseline. The presence of other respiratory viruses was not examined, however, nasal explants were isolated from pigs acquired from farms of high health status without prior history of respiratory infections. Therefore, the presence of other respiratory virus cannot be excluded, but it would reflect a natural environment in the nasal mucosa. Virus enters host cells from the apical side of epithelial cells in the airways. In order to mimic nature more closely, virus could be administered on top of the NEs i.e. from the apical side of the cells instead of immersing NEs into virus suspension. Furthermore, it could be speculated, that the incubation temperature should be decreased a few degrees to mimic the nature of a slightly colder environment of the snout/nose. Finally, the TUNEL assay detects DNA fragmentation, which occurs in the late phase of apoptosis. Assays detecting early stage apoptosis, such as caspase activation assay, should be used in future studies.
High similarity of human and pig respiratory epithelium anatomy and antiviral immune proteins (Dawson et al., 2013;Starbaek et al., 2018) suggests that porcine NEs could be a relevant 3R compliant model for the study of early innate host defence against both swine and human adapted IAV. In conclusion, we have shown that NEs are an important valid method for the study of the innate antiviral immune response after IAV infection, though gene expression changes solely as a consequence of ex vivo culturing. However, by careful analysis of time and handling matched mock control samples, transcriptional analysis of the innate antiviral immune response is indeed possible and comparable to responses measured in nasal mucosal tissue and lung tissue isolated from IAV infected pigs.

Funding
The work presented in this article is part of the FluZooMark project supported by Novo Nordisk Foundation (grant NNF19OC0056326).

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.