Primary bronchial epithelial cells (pBECs) were isolated from lung explant tissue of three CF and three lung emphysema patients undergoing lung transplantation. The tissue was obtained from the Hannover Pathology Tissue collection (Hannover Medical School) and its use was approved by the institutional ethics review board of Hannover Medical School (Reference 2700-2015). Donor characteristics are shown in Supplementary Table S1 .

Cells were isolated using a protocol modified from van Wetering et al. (2000). One to two bronchus rings were used for each isolation. In order to detach cells from the bronchus, the bronchial ring was cut open and incubated for 2 h at 37°C in protease XIV solution (final concentration 1.8 mg/ml, Sigma, St. Louis). It was then transferred to a sterile Petri dish containing PBS supplemented with antibiotics (penicillin/streptomycin, Sigma, St. Louis; MycoZap Plus PR, Lonza, Basel), and cells were detached by gently scraping the inner side of the bronchial ring. Cells were pelleted by centrifugation, resuspended in KSFM complete medium (Gibco, Carlsbad) supplemented with 25 µg/ml bovine pituitary extract (Thermo Fisher Scientific, Waltham), 0.2 ng/ml epidermal growth factor (Thermo Fisher Scientific, Waltham), 1 nM isoproterenol (Sigma, St. Louis), penicillin/streptomycin and MycoZap, and seeded in tissue culture flasks pre-coated with 10 µg/ml fibronectin (VWR, Radnor), 10 µg/ml BSA (Sigma, St. Louis) and 30 µg/ml PureCol collagen (Sigma, St. Louis). Culture medium was replaced every other day until cell layers reached approximately 80% confluence. Cells were then cryopreserved at a controlled freezing rate in KSFM with 0.3 mg/ml BPE and 10% DMSO and stored in liquid nitrogen until further use.

Isolated cells were confirmed to be basal epithelial cells by immunofluorescence. Cells stained positive for cytokeratin 5 (rabbit anti-human cytokeratin 5, Biolegend), p63 (rabbit anti-human p63, Abcam, Cambridge) and negative with the fibroblast-reactive antibody TE-7 (mouse anti-human fibroblast antigen, clone TE7, Merck, Darmstadt).

For each experiment, cells were thawed from cryopreserved stocks and expanded in pre-coated tissue culture flasks using PneumaCult Ex medium including supplements (Stemcell Technologies, Vancouver). Cells were passaged once before seeding into the apical chamber of 12-well plates with transwell inserts at 4x10⁵ cells/insert. Until cell layers were fully confluent, cells were cultured as submerged cultures with PneumaCult Ex medium on both, apical and basal side of the chamber, with medium being changed every other day. Once cell layers reached confluency, the medium on the apical side was removed and cells were left air exposed, while medium on the basal side was replaced with PneumaCult ALI medium including supplements (StemCell Technologies, Vancouver), penicillin/streptomycin (Sigma, St. Louis) and MycoZap Plus PR (Lonza, Basel) in order to initiate differentiation. Medium was refreshed and cells were washed with PBS three times a week until cell layers were fully differentiated after approximately three to four weeks. Cell differentiation was confirmed by an increase in transepithelial electric resistance (TEER), mucus production, and microscopically visible ciliary movement. At least one week prior to infection, cultures were transferred to antibiotic-free medium.

Two P. aeruginosa (PA) isolates, one with non-mucoid phenotype and one with mucoid phenotype, were recovered from one respiratory sample of a CF patient with chronic PA lung infection by using CF-specific cultivation conditions including MacConkey and Pseudomonas CFC selective agar (ThermoFisher, Waltham). Phenotypes were assigned based on colony morphology ( Supplementary Figure 1 ). Species identification was done by Matrix-assisted-laser desorption ionization-time of flight analysis (MALDI-TOF; VITEK MS; bioMérieux, Nürtingen). Subcultures were picked from single colonies and cryopreserved as pure cultures in Microbank vials (Pro-Lab Diagnostics, Round Rock).

For experiments, bacterial suspensions were prepared in PBS from bacteria grown overnight on Columbia Agar plus Sheep Blood ‘Plus’ plates (ThermoFisher Scientific, Waltham). To achieve consistent bacterial inocula, bacterial suspensions were first diluted to an OD₆₀₀ of 0.4 and then further diluted to achieve the desired colony count using a strain-specific dilution factor determined experimentally beforehand.

Human Rhinovirus (HRV) type 16 was originally obtained from ATCC (Manassas) and propagated in HeLa Ohio cells (Sigma Aldrich, St. Louis). For preparation of stocks, sub-confluent cell layers were inoculated with the parent virus pool at a multiplicity of infection of approximately 0.1 in DMEM + 2% fetal bovine serum (FBS). Virus was allowed to attach for 2 h at 33°C under occasional gentle swirling. Unbound virus was then removed by washing cells twice with PBS and fresh infection medium (DMEM+ 2% FBS) was added to the flasks. Approximately three days after infection and when 80-90% of the cell layer showed cytopathic effect, virus pools were harvested. Culture supernatants were collected and cleared of cell debris by centrifugation. Additionally, remaining cells were detached with a cell scraper in a small volume of medium, snap frozen in liquid nitrogen, thawed and centrifuged to remove cell debris. The supernatant of this centrifugation step was added to the cleared culture supernatant. Virus pools were concentrated 10-20x and partially purified by ultrafiltration using Amicon Ultra Centrifugal filter units (cutoff 100 kDa). Virus stocks were quantified by determining the 50% Tissue Culture Infectious Dose (TCID₅₀) on HeLa Ohio cells. TCID₅₀ was calculated using the Spearman-Karber Formula (Hierholzer and Killington, 1996).

To simulate chronic bacterial infection, cells were repeatedly infected with PA for a period of 16 days ( Figure 1A ). For each infection, cell layers were washed thoroughly with PBS and 10³ colony forming units (CFU) of the respective PA strain were added to the apical side of inserts, diluted in 50 µl PBS with 25 µg/ml tobramycin. Cell layers were washed daily, and basal medium was changed every 48 h. On day 16, after apical washing, cells were infected with HRV16 from the apical site at a multiplicity of infection of 1. After incubation at 33°C for 2 h, bacterial infection was carried out as described above, followed by incubation for 48 h at 33°C.

Bacterial infections were carried out in the presence of tobramycin in order to prevent bacterial overgrowth and excessive cell death. Tobramycin was chosen due to its clinical relevance in the treatment of pulmonary PA infections in CF. Both PA isolates were sensitive to tobramycin (minimum inhibitory concentrations [MIC]: 4 µg/ml for non-mucoid isolate, 2 µg/ml for mucoid isolate as determined by E-test). Despite the used concentrations of tobramycin being above the respective MICs of the two isolates determined by E-test, live bacteria could be recovered and cultured from apical washes of the infected cell cultures.

Gene expression and viral RNA were quantified by RT-qPCR. Total RNA was isolated using a peqGOLD Total RNA Kit (VWR, Radnor) and 100 ng RNA was reverse transcribed using iScript gDNA Clear cDNA Synthesis Kit (BioRad, Hercules). For qPCR analysis, cDNA was diluted 1:5. qPCR was then performed using Sso Advanced Universal Probes Mastermix (BioRad, Hercules) on a StepOne Plus Real-Time PCR system (ThermoFisher, Waltham). Oligonucleotide sequences are listed in the Supplementary Information (Table S2) . For gene expression analysis, fold changes relative to uninfected controls were calculated using the 2^-ΔΔCT method (Livak and Schmittgen, 2001) with GAPDH as normalization gene. Intracellular viral RNA copies were quantified using a standard curve derived from synthetic double-stranded DNA.

Protein concentrations of IL-6 and CXCL-8 were quantified by ELISA (Invitrogen, ThermoFisher). Protein concentrations of IL-1β, TNF-α, CXCL-10, and TGF-β were determined by cytometric bead array (BD, Franklin Lakes).

To test for potential degradation of IL-6, 10 µl conditioned culture medium obtained from infection experiments was used either untreated, heat-treated (95°C/5 min), or mixed with 8 µl of different protease inhibitors (or respective controls) and then incubated with 2 µl (0.5 µg/µl) recombinant IL-6 (Peprotech, Rocky Hill) at 37°C. For inhibition of metalloproteases, phosphoramidon (Sigma-Aldrich) was used at a final concentration of 3.4 mM. Nα-Tosyl-L-lysine chloromethyl ketone hydrochloride (TLCK, Sigma-Aldrich), primarily inhibiting serine proteases, was used at a final concentration of 40 mM. Additionally, a proprietary, broad non-metalloprotease inhibitor was used (cOmplete™ Mini EDTA-free Protease Inhibitor Cocktail, Roche, one tab dissolved in 500 µl water).

Proteins were separated by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis using a 15% polyacrylamide gel and transferred onto a nitrocellulose membrane. The membrane was then blocked with 5% skim milk for 1 h at room temperature and subsequently incubated with mouse anti-human IL-6 monoclonal antibody (Clone OTI3G9, Origene, Rockville) diluted 1:50000 in 5% skim milk. After washing with TBST, the membrane was incubated with an HRP-conjugated secondary goat anti-mouse antibody (Thermo Fisher Scientific, Waltham, dilution 1:5000) for 1 h at ambient temperature. After thorough washing of the membrane, signals were detected using Western-Blot Detection Reagents (Amersham/GE Healthcare, Little Chalfont) according to the manufacturer’s protocol. Densitometry was performed for the peak corresponding to full length IL-6 using FIJI/ImageJ2.

Transepithelial electrical resistance (TEER) was measured directly prior to and 48 h after (co-)infection (Millicell ERS-2 instrument, MerckMillipore, Burlington). To measure permeability of the cell layer, 100 µl FITC-labeled dextran (5 mg/ml, average molecular weight 20 kDa, Sigma-Aldrich) were added to the apical side 48 h after (co-)infection. After one hour, fluorescence was measured in medium taken from the basal side (EnVision Multimode Plate Reader, PerkinElmer, Waltham).

Data regarding cytokine levels and gene expression were obtained from six independent experiments, using cells derived from a different donor in each experiment, and with duplicate wells for each condition. The mean value of these duplicate measurements was used for further analysis. Epithelial permeability was assessed in three independent experiments. Data on degradation of recombinant IL-6 and test of inhibitors were generated in three to four independent experiments. Cytokine concentrations and fold changes were logarithmically transformed to approximately conform to normality. For analysis, values below the minimum quantifiable level (MQL) were used as 0.5*MQL value of the respective assay. Statistical significance was tested by analysis of variance (ANOVA) and Tukey’s multiple comparison test for comparison between all experimental groups, or Dunnett’s multiple comparison test for comparison of densitometry data ( Figures 3C, D , comparison to input), all with a significance level of α = 0.05. Figures and text show untransformed values (mean +/- SD) to facilitate interpretation.

We first sought to assess the impact of chronic bacterial infection on the production of inflammatory mediators during secondary viral infection. Therefore, protein concentrations and mRNA levels of key cytokines and chemokines were measured in the basal medium and cell lysates of air-liquid interface cultures following chronic PA infection and acute HRV superinfection.

Concentrations of IL-1β in basolateral media were on average below 10 pg/ml (MQL) for sham-treated cells and for cells exposed to either pathogen alone, but were significantly higher in cells that had been infected with HRV16 following a prior exposure to the non-mucoid PA isolate (28.88 ± 17.69 pg/ml vs. 6.10 ± 2.69 pg/ml, p< 0.001, Figure 1B ). Likewise, pro-IL-1β mRNA levels were significantly higher in cells co-infected with the non-mucoid PA and HRV16 than in cells infected with HRV16 alone (1019 ± 1431 fold change vs. 28 ± 35 fold change relative to uninfected controls, p=0.012, Figure 1C ). PA infection alone had no significant effect on concentrations of IL-6, whereas IL-6 production was strongly induced by infection with HRV16 (2301 ± 1873 pg/ml, Figure 1D ). Strikingly, in cells co-infected with non-mucoid PA and HRV16, virus-induced elevation of IL-6 was almost entirely abrogated (53.17 ± 58.84 pg/ml, p< 0.001 vs. PA-/HRV+), whereas IL-6 concentrations in media of cells co-infected with HRV16 and a mucoid PA isolate were similar to those of cells infected with the virus alone (1876 ± 1067 pg/ml). On mRNA level, compared to HRV infection alone (96 ± 127 fold change), co-infection with the non-mucoid PA isolate significantly increased IL-6 mRNA (4742 ± 8020 fold change, p = 0.04), whereas co-infection with the mucoid isolate did not (79 ± 108, p = 0.999, Figure 1E ).

Other cytokines and chemokines, including CXCL-8 (IL-8), CXCL-10 (IP-10), TGF-β and TNF-α protein, and type I IFN (IFN-β) and type III IFN (IFN-λ1) mRNA were increased to a similar extent during viral infection and co-infections with either of the PA isolates and HRV16. However, TGF-β was significantly increased in co-infection with the non-mucoid PA isolate compared to co-infection with the mucoid PA isolate (298 ± 264 pg/ml vs. 27± 11 pg/ml, p= 0.008, Figure 2 ). Protein and mRNA levels of the molecules listed above were similar in cells isolated from lung explants emphysema patients and those from CF patients.

A significant increase in IL-6 mRNA levels in conjunction with absence of IL-6 protein suggested either degradation or post-transcriptional regulation of IL-6 production in cells co-infected with non-mucoid PA and HRV16. In order to elucidate the divergent effects of PA/HRV co-infection on IL-6 mRNA and protein levels in bronchial epithelial cells, we tested if cell-free conditioned basal media of co-infected cells contain soluble factors that are able to degrade recombinant IL-6 (rIL-6). To this end, media from different infection conditions were mixed with 1 µg rIL-6 and incubated for 4 h at 37°C. Western blot analysis showed that rIL-6 was entirely degraded when mixed with conditioned media obtained from cells co-infected with a non-mucoid PA isolate and HRV16 ( Figure 3A ). In contrast, rIL-6 was not affected by conditioned media of any of the other tested infection conditions, including infection with the individual pathogens, and media from a co-infection with a mucoid PA isolate and HRV16. Similar results were obtained when rIL-6 was mixed with apical wash fluid of the infected cultures instead of culture medium (data not shown).

In line with the results obtained by Western Blot, rIL-6 was no longer detectable by ELISA in samples incubated with media from cells co-infected with non-mucoid PA and HRV16 ( Figure 3B ). If media were heat-inactivated prior to mixing with the recombinant protein, concentrations of IL-6 were comparable to those of controls, suggesting degradation of IL-6 by a heat-labile mediator. A time course series showed two distinct major cleavage products of rIL-6 occurring within minutes after exposure to conditioned media of co-infected cells, with a molecular weight of approximately 16 and 12 kDa respectively (full length IL-6: 21 kDa). Densitometric analysis showed a statistically significant reduction in full length IL-6 already after five minutes. ( Figure 3C , right panel). Addition of a protease inhibitor cocktail (cOmplete EDTA-free Protease Inhibitor Cocktail, Roche) was able to partially inhibit proteolytic degradation of full length rIL-6, while the metalloprotease inhibitor phosphoramidon only inhibited degradation of the 12 kDa fragment. In contrast, addition of the protease inhibitor TLCK, predominantly inhibiting serine proteases, was able to completely block proteolytic degradation of rIL-6 at a final concentration of 40 mM ( Figure 3D ).

Co-infection with the non-mucoid PA isolate and HRV16 appeared to cause more pronounced damage to the cell layer than any of the other infection conditions tested, however, this was not associated with an increase in viral load ( Figure 4A ). As a measure for permeability of the epithelial layer, we measured leakage of FITC-labeled dextran (average molecular weight 20 kDa) from the apical compartment of the transwell system into the basal chamber. Fluorescence intensity in samples taken from the basal compartment of cultures co-infected with the non-mucoid PA isolate and HRV16 was significantly higher than in sham-infected controls or cells infected with HRV16 only ( Figure 4B ). This was not the case for any of the other infection conditions.

As a second method to assess barrier integrity in infected cultures, we determined transepithelial electrical resistance (TEER). Differences between groups failed to reach statistical significance at 48 h post viral infection (NM+/HRV+ vs. PA-/HRV+, p=0.142, Figure 4C ), but showed a similar overall trend as FITC dextran leakage assay.

IL-6 is a key player in regulation of epithelial repair after virus-induced lung injury (Yang et al., 2017). Ciliated cell differentiation of basal cells is dependent on the transcriptional regulator Forkhead box J1 (FOXJ1), which can be activated through the IL-6/STAT3 pathway (Tadokoro et al., 2014). We found FOXJ1 mRNA levels to be decreased in co-infection with the non-mucoid PA isolate, compared to virus-infection alone (0.31 ± 0.17 fold change vs. 1.14 ± 0.49 fold change, p=0.002) ( Figure 5A ). This was accompanied by a statistical significant decrease in the expression of Cilia Apical Structure Protein (SNTN) in cells co-infected with the non-mucoid PA isolate and HRV16 (0.05 ± 0.04 fold change vs. 0.18 ± 0.09 fold change, p=0.002) ( Figure 5B ). On the other hand, the differences in mRNA levels of Secretoglobin (SCGB3A1), a product of secretory cells, did not reach statistical significance between cells co-infected with the non-mucoid PA isolate and single HRV16 infection (0.18 ± 0.11 fold change vs. 0.32 ± 0.12 fold change, p=0.188) ( Figure 5C ).