Human nasal epithelial cells were obtained from healthy control subjects (n=9). Subjects had not experienced a symptomatic upper respiratory tract infection in the preceding 6 weeks. Primary ciliated epithelium was obtained by brushing the inferior nasal turbinate and cultured to air-liquid interface as previously described ¹⁰ .

Type-2 alveolar basal epithelial cells (A549) cells (American Type Culture Collection (ATCC), Manassas, VA) were grown in RPMI medium (Gibco) with 10% heat-inactivated foetal calf serum (Sigma), 100 U/ml penicillin, 10 μg/ml streptomycin (pen/strep) and fungizone. Madin-Darby canine kidney (MDCK) (ATCC, Manassas, VA) cells were grown to in Dulbecco's modified eagle medium (DMEM) with 10% foetal calf serum, and pen/strep as above.

Human influenza virus A/Puerto Rico/8/34 (H1N1) (PR8) was grown in fertilized chicken eggs. Eggs were infected at day 9, incubated at 37°C for 2 days and virus harvested in the allantoic fluid at day 11. Allantoic fluid was harvested from uninfected chicken eggs for use as a negative control. Virus stocks were titred by plaque assay using MDCK cells grown to 90% confluence in 96-well dishes. Cells were washed with PBS and infected with serial dilutions of the virus in DMEM for 1 h at 37°C. The inoculum was removed and cells were incubated with 200 μl DMEM (medium containing 1.4% BSA, 2 μg/ml of trypsin and 1x penicillin/streptomycin antibiotics (VX15140122, Fisher)) at 37°C, 5% CO ₂ for 2-3 days. Virus plaques were visualized by staining with 1:200 mouse H36-4.5-2 anti-HA antibodies (in-house monoclonal provided by A. Easton) and an Alexa-594 labelled secondary anti-mouse antibody diluted 1:250 (A-11062, Invitrogen, UK).

Influenza aerosols were generated from a 1 ml suspension containing 10 ¹⁰ plaque-forming units (PFU)/ml using a Pari-Therm nebuliser (PARI Respiratory Equipment Inc, Midlothian, VA, USA). This was connected to a viable impactor (6-stage Microbial sampler, Westech Scientific Instruments, Upper Stondon, Bedfordshire, UK) to collect and fractionate the aerosolised influenza virus particles by aerodynamic size ( Figure 2A) as described previously ¹⁶ . Particles were impacted into separate Petri dishes containing 20 ml of medium (RPMI Media 1640, Life Technologies, Grand Island, NY, USA) with added penicillin (50 μg/ml), streptomycin (50 μg/ml) and Fungizone (1 μg/ml).

Quantification of viable influenza virus in the fractionated aerosolized particles was performed using MDCK cells seeded into 96 well plates. Triplicate wells were exposed to 200 μl of a log dilution series of the impacted air sample at 37°C in 5% CO ₂ for 2 hours. The sample was removed using two washes with DMEM and incubated at 37°C for 48 hours. Influenza virus was detected as above by direct immune-fluorescent staining. The number of immune-fluorescent plaques per well were counted and the total PFU in each sample calculated.

This was performed using the traditional method of liquid immersion cells with fluid containing virus or using our bespoke aerosolisation system ( Figure 1). Frozen aliquots of influenza virus were thawed immediately prior to use and diluted in BEBM basal medium (Lonza, CC3171) to 1x10 ⁵ PFU/ml. Prior to infection, the basolateral culture medium was removed and the apical surfaces of ALI cultures were rinsed with medium. Next, 200 µl of viral suspension (multiplicity of infection (MOI) of ~0.1 (based on an estimate of 5x10 ⁵cells/ALI well)) in BEBM was directly applied to the apical surface for 1 hour and then removed. We also delivered the virus using a bespoke nebulisation system ( Figure 1A), which allows aerosols to be delivered to the surface of ciliated cells in culture under direct vision using an inverted microscope ( Figure 1B). We generated influenza aerosols from suspensions containing doses of 10 ⁵, 10 ⁶ or 10 ⁷ PFU/ml using the Pari-Therm nebuliser. The nebulised virus was then delivered directly to the Transwell inserts via a static-free stainless-steel pipe to negate the effect of static charge on the aerosolised particles generated. Delivery to the cells in culture was via a stainless-steel t-piece at the end of the steel pipe. Both ends were clear to both reduce any pressure effects on the ciliated cultures and to allow real-time observation by high-speed video photography. The entire system was encased in a heated (37°C) Perspex chamber, to mimic body temperature and to prevent the spread of the biological agents. The Pari-Therm was chosen as it warms the aerosols to approximately 32°C thus reducing the possible negative cellular effects of cold aerosol landing on the cells. Cultures were exposed to influenza aerosols for 1 minute. The nebuliser was then switched off and the cultures left for 1 hour. All cultures were maintained at 37°C. Control wells received nebulised allantoic fluid without virus in BEBM. Cells were then washed three times with BEBM to remove unbound virus and the infections were allowed to continue for up to 24 hours. At this time the apical surface of the cells was rinsed and the wash was stored at -80°C for plaque assay and cytokine analyses. Cells were fixed overnight at 4°C with 4% paraformaldehyde in phosphate buffered saline (PBS) for immunostaining.

Cultures were placed in an incubation chamber at 37°C for 30 minutes and were observed via an inverted microscope system (Nikon TU1000, UK). Beating cilia were recorded using a digital high-speed video camera (Lake Image Systems, USA) at a rate of 250 frames per second using an x40 objective as previously described ¹⁷ . Ciliary dyskinesia was defined as ciliated cells that displayed uncoordinated motile cilia or those that beat with a stiff, flickering or twitching motion. The dyskinesia index was calculated as the percentage of dyskinetic ciliated cells relative to the total number of motile ciliated cells.

Chemokines and cytokines were measured using a 96-well multispot assay (Meso Scale Discovery, Maryland, USA) as described previously ¹⁰ . The lower limit of detection was 1 pg/ml.

Cells were fixed with 4% (w/v) paraformaldehyde in phosphate-buffered saline (PBS) overnight at 4°C. Following fixation, cells were washed three times with PBS, blocked with 3% (w/v) BSA in PBS for 10 minutes, and washed again three times with PBS. All subsequent antibody incubations were carried out in PBS containing 1% (w/v) BSA. Reagents used for immunofluorescence in this study were anti-influenza virus HA mouse monoclonal antibody H36-4.5-2 (40 µg/ml) as above and 1:50 anti-β-tubulin III rabbit monoclonal antibody (ab52623, Abcam). Subsequent incubations were performed as described previously ¹⁰ . High resolution optical sections were obtained using a Leica TCS SP5 confocal microscope using a 63x oil objective (numerical aperture, 1.4). Images acquired by confocal microscopy were rendered by Imaris Software v7.2 (Bitplane AG).

Images obtained using a 20x objective were used to quantify the level of anti-β-tubulin III staining by mean fluorescence intensity using NIS Elements Software (Nikon Instruments, Kingston, UK).

Statistical analysis was performed using GraphPad Prism 5 (GraphPad, San Diego, CA, USA). Any difference in the ciliary activity observed for control and influenza virus was determined using paired t-tests. Within-group comparisons of the magnitude of chemokine/cytokine release were conducted using a Wilcoxon signed ranks test. Between-groups comparisons were performed using the Mann–Whitney U-test.

Ethical approval was obtained from the University of Leicester Committee for Research Ethics (Leicester, UK). All adult subjects provided informed written consent.

Work with embryonated hen’s eggs was approved by the University of Warwick Animal Welfare and Ethical Review Body carried out under a licence approved by the UK Home Office as required by the Animal Scientific Procedures Act (1986).

The Pari-therm nebuliser generated infectious influenza virus aerosol particles that ranged from 0.65 to >7 µm in size ( Figure 2B). A substantial proportion (84%) of infectious influenza virus was contained in particles less than 4.7 µm, a particle size where inhalation into the lower respiratory tract is likely. The total number of infectious influenza virus particles produced by the nebuliser was 2x10 ⁹ PFU indicating an approximate 5-fold loss in viral viability or infectivity from the inoculum (10 ¹⁰ PFU). Raw PFU data are available on Figshare ¹⁸

We found that nebulising 1×10 ⁸ PFU for 1 minute resulted in 10 ⁵ PFU landing on the membrane growth area of an empty Transwell (no cells). This aerosol inoculum (MOI of approximately 0.2) was shown to result in approximately 20% of infected epithelial cells being virus-positive after 24 h. In separate wells, cells were infected via liquid immersion using a viral inoculum of MOI of 0.2. This was to confirm that any effects were due to the amount of virus landing on the cells, rather than any difference in the ability of the virus to infect ( Figure 2C). This data showed that the nebuliser was required to contain a virus preparation that was ten-fold higher than that delivered in the immersion system. This was reflected in the amount of virus released by these infected cells into the apical fluid after 24 h infection ( Figure 2D).

Effect of influenza virus infection on ciliary function. We found that influenza virus infection significantly reduced the proportion of epithelial cells with motile cilia ( Table 1). Using the standard liquid immersion method, the number (median (IQR)) of ciliated epithelial cells per field significantly reduced (P<0.05) as early as 18 hours post-influenza virus infection (20 (11-39) cells/4.2 mm ²) compared to the uninfected controls (27 (27-37) cells/4.2 mm ²). After 24 hours infection this reduced to less than half (16 (11-29) cells/4.2 mm ²) that detected in the uninfected controls (33 (27-40) cells/4.2 mm ²). Despite this reduction in the numbers of cells with motile cilia, the ciliary beat frequency of the motile cilia on other ciliated cells in culture was unaffected by influenza virus infection over the 24-hour study period. The median (IQR) ciliary beat frequency of motile cilia on influenza virus infected nasal ciliated cells was indistinguishable after 24 hours (14.7 (9.7-15.2) Hz) from the uninfected controls (14.0 (11.7-15.7) Hz) (P=0.87). The cilia that remained motile following influenza virus infection showed no evidence of ciliary dyskinesia. The median (IQR) dyskinesia index was the same after 24 hours (0% (0%-5%)) as the uninfected controls (0% (0%-0%)) (P=0.87). Raw data on ciliary activity are available on Figshare ¹⁹ .

Infection of ciliated epithelial cells by aerosolised influenza virus results in similar cytopathology as standard method. Human ciliated airway epithelial cells infected with aerosolised influenza A virus resulted in similar levels of cytopathology (cells rounding, lifting off and tight junctions detected) as those immersed in the same concentration of influenza virus. The number (median (IQR)) of motile ciliated epithelial cells per field reduced two-fold to 7.8 (4.0-10.4) cells/4.2 mm ²) compared to the pre-infection level (14.0 (5.2-22.8) cells/4.2 mm ²). The ciliary beat frequency (CBF) of the remaining motile ciliated epithelial cells exposed to aerosolised inoculum of influenza A virus (16.6 (15.4-17.6) Hz) remained unchanged from pre-infection level (15.8 (15.2-16.3) Hz) and the uninfected control of 15.5 (13.1-17.3) Hz. These values do not account for cilia on cells that have become static.

The concentration of inflammatory mediators detected in the apical fluid of cell cultures subjected to the aerosol delivery of the control and influenza virus preparations resulted in significantly lower levels of inflammatory mediators after 24 hours compared to the liquid immersion system ( Table 2). In addition to lower concentrations, the two systems produced a different cytokine response to influenza virus infection. Among the ten cytokines measured, influenza virus infection by liquid immersion caused a significant (P<0.05) downregulation of IL-6 secretion to a median (IQR) of 333 pg/ml (107-1304) from 548 pg/ml (418-1525) in control uninfected cells. However, the aerosol delivery of influenza virus resulted in an upregulation of IL-6 to 1297 pg/ml (1107-1371) from 968 pg/ml (700-1235) in control uninfected cells. The same trend was seen with another anti-inflammatory cytokine IL-13, which showed significant upregulation in the aerosol group from the control 50 pg/ml (48-53) to 65 pg/ml (55-65) following infection, compared to a downregulation in the liquid immersion group from 70 pg/ml (65-76) to 55 pg/ml (22-76), respectively.

Among the nine chemokines measured, ciliated epithelial cells exposed to a liquid immersion inoculum of influenza virus resulted in a significant upregulation of CCL22 from a median (IQR) uninfected control of 2,0545 pg/ml (9986-23181) to 22812 pg/ml (20,342-30,853) ( Table 2), whereas exposure to a aerosolised inoculum and control solution resulted in undetectable levels of this chemokine. MCP-1 and MCP-4 both showed a significant (P<0.05) increase following aerosolised delivery, but this was not significantly altered in the group exposed to the liquid immersion inoculum. However, both delivery systems show similar CXCL8 and CXCL10 responses to influenza virus, where production increased approximately 1.5-fold and 6.5-fold, respectively. In the liquid immersion delivery system, median (IQR) CXCL8 increased from 594 pg/ml (506-989) in the uninfected control cells to 887 pg/ml (692-1190). The aerosol delivery system resulted in much higher concentrations of CXCL8 from 65,523 pg/ml (44,759-86,287) in the uninfected control cells to 109,039 pg/ml (86,701-113,242) compared to the liquid immersion system. CXCL10 increased from 698 pg/ml (433-974) in the uninfected control cells to 4140 pg/ml (823-19884). MIP-1β, Eotaxin-3, TARC and MDC were unable to be detected in the apical fluids of cells exposed to aerosolised medium or virus, but liquid immersion the membrane with medium alone caused levels to exceed 30 pg/ml, 3500 pg/ml, 290 pg/ml and 20,000 pg/ml, respectively. Raw data from measurement of inflammatory mediators are available on Figshare ²⁰ .

The distribution of influenza HA antigen on human ciliated epithelial cells. We have previously shown that RSV preferentially infects ciliated cells and this infection progresses with viral antigen being displayed on the cell surface, leading to viral antigen moving into the ciliary shaft after 24 h ^{1, 2} . To determine whether influenza virus follows a similar pattern of viral antigen spread we used immunofluorescent staining to detect influenza virus haemagglutinin (HA) and ciliary (β-tubulin III) antigens on infected cells. We found a high level of influenza virus HA antigen accumulated on the distal ends of the cilia (ciliary tips) ( Figure 3A, panel A) and the apical cell surface of ciliated cells ( Figure 3A, panel B). We observed the co-localisation of influenza virus and anti-β-tubulin III, indicating virus HA antigen is displayed on the ciliary axoneme ( Figure 3A, panel C). Notably, we found that cells either displayed viral antigens only on the ciliary tips or the full length of the ciliary axoneme. In cells where the apical membrane (cell surface) showed evidence of viral antigen, antigen was always present over the full length of the ciliary axoneme suggesting spread of antigen from the tip to the base of the cilium. Raw mean fluorescence intensity values are available on Figshare

A number of cells were seen with partial and in some cases almost complete loss of cilia ( Figure 3A, panel D). In these cells fragments of β-tubulin III staining was seen the surface of the cell indicating that the cilia may have been lost from infected ciliated cells. To determine whether cilia were lost as a result of influenza infection, we compared the level of anti-β-tubulin III, a marker of cilia, (as measured by fluorescence intensity) in control and infected cultures. This showed that influenza infection resulted in significantly (P<0.01) less anti-β-tubulin III staining compared to uninfected cultures ( Figure 3C).