Specific pathogen-free wild type C57BL/6J mice (>8 weeks old, male and female) were housed in an environment with controlled humidity at 24 ± 1°C under a 12 hour light/dark cycle. Water and food were given ad libitum. All animal procedures were approved by the local Animal Care and Use committee of the University of Groningen and performed according to the national guidelines under approval number AVD1050020209205.

The procedure for obtaining human lung tissue was in accordance to the Research Code of the University Medical Center Groningen (UMCG), as stated on https://umcgresearch.org/w/research-code-umcg as well as national ethical and professional guidelines Code of Conduct for Health Research (https://www.coreon.org/wp-content/uploads/2023/06/Code-of-Conduct-for-Health-Research-2022.pdf). The use of left-over lung tissue in this study was not subject to Medical Research Human Subjects Act in the Netherlands, as confirmed by a statement of the Medical Ethical Committee of the University Medical Center Groningen and therefore exempt from consent according to national laws (Dutch laws: Medical Treatment Agreement Act (WGBO) art 458/GDPR art 9/UAVG art 24). All donor material and clinical information were deidentified prior to experimental procedures, blinding any identifiable information to the investigators. Human lung tissue was obtained from excess lung tissue surplus to clinical care requirements derived from lung resection and lung transplant procedures. Histologically normal lung tissue from tumor resection surgery (non-COPD) was obtained from an area as far away as possible from the tumor and all tissues were examined by an experienced pathologist to exclude the presence of abnormalities.

The mouse lung fibroblast cell line CCL206 (CCL-206; ATCC, Wesel, Germany, RRID: CVCL_0437) was cultured in CCL206 medium; DMEM (Lonza, Basel, Switzerland, #42430)/Ham’s F12 (Gibco, #21765) (1:1) supplemented with 10% (v/v) fetal bovine serum (FBS, PAA Laboratories, Pasching, Austria), 2 mM L-glutamine (Life Technologies, Carlsbad, CA, USA, #35050–061), 100 U/ml penicillin/streptomycin (Gibco, Waltham, MA, USA, #15070-063) and 1% amphotericin B (Gibco, #15290-026). Primary human lung fibroblasts were obtained from non-COPD lung material using the crawl out method, as previously described (14). They were maintained in primary fibroblast medium (low glucose DMEM (Biowest, Nuaillé, France) with 10% FBS, 100 U/ml penicillin/streptomycin and 1% glutamax (Gibco, #35050-061)), and used at passage 5 for all experiments. Table 1 summarizes the characteristics of the fibroblast donors of all experiments. All cells were maintained at 37°C with 5% CO₂ and confirmed to be mycoplasma negative before use.

Epcam+ cells were isolated as described previously (15). Briefly, the lungs of adult C57BL/6J mice were instilled with dispase (BD Biosciences, Oxford, UK, #354235) and digested for 45 minutes at 37°C, after which they were further digested with 220 units/ml DNAse1 (Applichem, Darmstadt, Germany #A3778) in DMEM, and filtered through cell strainers (Corning Incorporated, Corning, NY, USA, #431752). The resulting cell suspension was depleted of CD31+ and CD45+ cells using CD31 (Miltenyi, #130-097-418, RRID: AB_2814657) and CD45 (Miltenyi Biotec, Teterow, Germany #130-052-301, RRID: AB_2877061) microbeads in combination with LS columns (Miltenyi #130-091-051). Finally, the suspension was enriched for epithelial cells using CD326 (Epcam) microbeads (Miltenyi #130-105-958).

After digesting the lungs of adult mice and depleting the cell suspension of CD31+ and CD45+ cells (as described above), the cell suspension was divided into three fractions. 1) The first fraction was incubated with CD326 (Epcam) microbeads to obtain epithelial cells as described above (CD31-CD45-Epcam+ fraction). 2) The second fraction was enriched for alveolar type 2 cells (AT2) cells using MHCII-PE antibody (BD Biosciences, #557000) and anti-PE microbeads (Miltenyi, #130-048-801) (CD31-CD45-MHCII+ fraction). 3) The third fraction was also enriched for AT2 cells, with a different approach. This fraction was first incubated with CD326 (Epcam) microbeads and subsequently enriched using MHCII-PE antibody and anti-PE microbeads (CD31-CD45-Epcam+MHCII+ fraction).

The organoid culture was based on published protocols (15). Confluent CCL206 fibroblasts were treated with 10 μg/ml mitomycin C (Sigma-Aldrich, St. Louis, MO, USA, #M4287) for 2 hours in a culture flask to inhibit their proliferation. They were then allowed to recover in normal medium for at least 1 hour, after which they were harvested. Subsequently, 10,000 CCL206 fibroblasts were mixed with 10,000 freshly isolated mouse CD31-CD45-Epcam+ cells, CD31-CD45-MHCII+ cells or CD31-CD45-Epcam+MHCII+ cells. The cell mixtures were seeded into inserts for 24-well plates (Thermo Fisher Scientific, Waltham, USA #10421761) in growth-factor reduced Matrigel (Fisher Scientific, Landsmeer, The Netherlands #11523550) mixed 1:1 with CCL206 medium, and cultured for 14 days with DMEM/Ham’s F12 supplemented with 5% FBS, 100 U/ml penicillin/streptomycin, 1% amphotericin B, 2 mM L-glutamine, insulin-transferrin-selenium (Gibco #15290018), bovine pituitary extract (30 μg/ml, Sigma-Aldrich, #P1476), and recombinant mouse EGF (0.025μg/ml, Sigma-Aldrich, #SRP3196). Y-27632 (10 μM, Tocris Bioscience, Oxford, UK, #1254) was also added for the first 48 hours of culture. Medium was replaced every 2-3 days. Organoid cultures were treated with exacerbation cocktail (see Table 2 ) or the compounds listed in Table 3 during 14 days of culture. Organoid number and size were quantified on day 14, after which organoids were either fixed with acetone/methanol at -20°C for 15 minutes and stored for subsequent immunofluorescent staining or collected for RNA isolation.

50.000 freshly isolated CD31-CD45-Epcam+ cells, CD31-CD45-MHCII+ cells and CD31-CD45-Epcam+MHCII+ cells were spun down onto six 3% (w/v) bovine serum albumin (BSA, Capricorn Scientific, Ebsdorfergrund, Germany) pre-coated microscope slides for 5 min at 550 rpm. Cytospins were air-dried for 30 min and subsequently stored at -20 °C. Samples were permeabilized with 0.1% Triton X in PBS for 10 min, and subsequently blocked for 1 hour with 5% BSA, 2% donkey serum (Jackson Immunoresearch, West Grove, PA, USA, #017-000-121, RRID: AB_2337258), 0.1% Triton X in PBS. Next, samples were incubated overnight at °4 C with rabbit anti-Prosurfactant Protein C (pro-SPC) (MiliporeSigma, Burlington, MA, USA, #AB3786, RRID: AB_91588) diluted 1:1000 in 2% BSA, 2% donkey serum, 0.1% Triton in PBS. They were then washed three times with PBS, and subsequently incubated for 1 hour with the secondary antibody donkey anti-rabbit Alexa fluor 488 (Thermo Fisher Scientific #A21206, RRID : AB_2535792) diluted 1:500. After washing the samples three times with PBS, cytospins were covered with mounting medium containing 4’,6-diamidino-2-phenylindole (DAPI) (Abcam, #104139) and visualized with an Eclipse Ti2-E microscope (Nikon, Tokyo, Japan) at magnification 20x. Representative images were magnified from original 20x images, but were otherwise not altered. Mean Fluorescence Intensity (MFI) was analyzed using ImageJ software (16) by defining the region of interest of each cell and measuring the MFI per cell for pro-SPC staining.

Mouse organoids were stained as previously described (15). Briefly, fixed organoids were blocked inside the insert with 5% BSA in PBS for 2 hours, and then incubated with primary antibodies rabbit anti-pro-SPC (MilliporeSigma, #AB3786) and mouse anti-Acetylated α Tubulin (ACT) (Santa Cruz Biotechnology, Dallas, TX, USA, #sc-23950, RRID: AB_628409) diluted 1:200 in 0.1% BSA 0.1% Triton X in PBS at 4°C overnight. Subsequently, secondary antibodies donkey anti-rabbit Alexa fluor 488 (Thermo Fisher Scientific #A21206, RRID: AB_2535792) and donkey anti-mouse Alexa fluor 568 (Thermo Fisher Scientific #A10037, RRID: AB_2534013) diluted 1:200 in 0.1% BSA 0.1% Triton X in PBS were added for 2 hours. Fixed organoid cultures were then excised from the insert and mounted on glass slides with mounting medium containing DAPI(Abcam). Organoid staining was quantified using a blinded approach on a DM4000B fluorescence microscope (Leica Camera, Wetzlar, Germany).

PCLS were obtained as previously described (17). Briefly, the lungs of adult C57BL/6J mice were inflated with 1.5% low-melting point agarose (Gerbu Biotechnik GmbH, Wieblingen, Germany), separated into lobes, and cut into PCLS of 250 μm thickness using a vibrating microtome (VT1000 S Vibrating blade microtome, Leica). Slices were placed into 12-well plates and treated with vehicle control or EC for 48 hours. They were then stored at -80°C until further processing for RNA isolation, which was performed using Maxwell simplyRNA tissue kit (Promega, Madison, WI, USA, #AS1280) according to manufacturer’s instructions.

300,000 proliferation-inactivated CCL206 fibroblasts were mixed with 300,000 freshly isolated mouse Epcam+ cells and seeded into 6-well plates in 500 μl growth factor reduced Matrigel mixed with 500 μl CCL206 medium. They were cultured for 72 hours with mouse organoid medium supplemented with Y-27632 and in the presence of vehicle control or EC, after which the Matrigel was disrupted by 45 minutes incubation with dispase (1:1). Organoids were brought into a single cell suspension by incubating with trypsin for 5 minutes. Cells were then sorted back into Epcam+ cells and Epcam- cells (CCL206 fibroblasts) using CD326 microbeads and LS columns, and both populations were subjected to RNA isolation using the NucleoSpin^® RNA kit (Macherey-Nagel, Düren, Germany, #740955) according to manufacturer’s instructions. Finally, both populations were bulk RNA-sequenced separately.

RNA concentrations were initially determined on a Nanodrop ND-1000 spectrophotometer, and quantified in detail by a Bioanalyzer and Fragment analyzer system (Agilent). Genomescan (the Netherlands) used the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA, USA #E7760S/L) to process the samples and performed bulk RNA-sequencing using an Illumina NovaSeq 6000 sequencer. Data quality control, adapter trimming, alignment of short reads and feature counting were included in the procedure. Ribosomal (and globin) content was determined to check the library preparation. Samples were checked for possible sample and barcode contamination and various standard quality metrics of the raw data set were determined using FstQC v0.34 and FastQA. The reads were trimmed for adapter sequences using Trimmomatic v0.30 and then aligned with the mouse reference GRCm38 (patch 6). Gene duplicates and genes with an average raw count <1 were removed. Paired-sample analysis using the DESeq2 package 1.34.0 in R 4.2.2 was performed to determine differentially expressed genes (DEGs), which were regarded as significant when padj<0.05. Pathway enrichment analysis was done with all DEGs using the fgsea package 1.20.0 in R. Pathways were considered significant when padj<0.05. Volcano plots were generated using the R package EnhancedVolcano 1.14.0 with cut-offs padj<0.05 and logFold>1. ComplexHeatmap version 2.12.1 was used to create heatmaps with normalized counts obtained from DESeq2 analysis as input. Normalized counts obtained from DESeq2 analysis were also used to report normalized count values. Epithelial signature scores were determined using the marker genes associated with specific epithelial subtypes (from the human lung cell atlas (18)). These gene sets were used as pathway definitions in order to run pathway enrichment analysis with fgsea in R. The complete dataset is available via the Geo Dataset GSE245259.

Primary human fibroblasts of non-COPD donors ( Table 1 ) were seeded into 6-well plates in primary fibroblast medium at a density of 300,000 cells per well. They were then quiesced for 24 hours in low glucose DMEM with 0.1% BSA and antibiotics, and then exposed to EC for 24 hours, after which they were lysed in TRIzol (Invitrogen, Waltham, MA, USA, #15596018). RNA was isolated with TRIzol according to manufacturer’s instructions.

RNA concentrations were quantified with a Nanodrop ND-1000 spectrophotometer. Equal mRNA concentrations were reverse transcribed into cDNA. A 7900HT Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) was used for Real-Time qPCR, with denaturation at 94°C for 30 seconds, annealing at 60°C for 30 seconds and extension at 72°C for 30 seconds for 40 cycles followed by 5 minutes at 72°C, using SYBR green (Roche Applied Science, Mannheim, Germany) as the DNA binding dye. The 2^-ΔΔCT method was used to determine fold changes. Gene expression was normalized to housekeeping genes B2M, SDHA and HMBS for human samples and Rpl13a, B2m and Actb were used as housekeeping genes for mouse samples. The complete lists of human and mouse primer sequences (Biolegio, Nijmegen, the Netherlands) are summarized in Supplementary Tables 1 , 2 ( Supplementary Material ) respectively.

GraphPad Prism 8 was used for statistical testing of all data except RNA sequencing data (see above). The number of biological replicates is described by N. Data are presented with mean and standard deviation unless described otherwise. Normalized data were log transformed before statistical evaluation. All data were checked for normality using the Shapiro-Wilk test. For parametric data, a paired two-tailed t-test, paired one-way ANOVA or two-way ANOVA was used, whereas for nonparametric data, a Wilcoxon or Kruskal-Wallis test was used, where appropriate. In cases where there were missing values an unpaired statistical approach was used. Differences were considered significant when p<0.05.

To investigate the effect of the soluble factor milieu during a COPD exacerbation on epithelial growth and differentiation, we first employed the mouse organoid model, which can be used to assess effects on growth of epithelial progenitors into differentiated epithelial cells over the course of 14 days ( Figure 1A ). A concentration response to EC was examined, in which EC approximated the concentrations of factors found in sputum during an COPD exacerbation (9, 13), with two higher concentrations of the cocktail also being included (ECx3 and ECx10) to account for possible dilution of the measured factor concentrations during sample collection. We found that the number and size of the organoids formed by Epcam+ epithelial cells were significantly increased by all EC concentrations tested ( Figures 1B–D ). Quantitative immunohistochemistry for pro-SPC and acetylated tubulin (ACT) revealed that ECx3 and ECx10 significantly increased the fraction of organoids that were pro-SPC+/ACT+ on day 14 (p=0.0011 and p=0.0057 respectively), which was at the expense of pro-SPC+/ACT- organoids ( Figure 1E ). In addition, using qPCR we found that EC significantly upregulated the expression of alveolar type 2 (AT2) cell marker Lamp3 (p=0.0016) and mucin genes Muc5ac (p=0.0061) and Muc5b (p=0.0343) in mouse organoids on day 14. The expression of Sftpc (marking AT2 cells), as well as markers associated with ciliated cells (Foxj1, Tekt1), club cells (Scgb1a1) and AT1 cells (Hopx, Pdpn, Aqp5) were not significantly affected by the presence of EC, although Sftpc and Hopx expression tended to increase (p=0.1081 and p=0.0680 respectively) ( Figures 1F, G ). As the lowest concentration of EC induced a robust response in mouse organoids, we continued our studies with this concentration of EC only. We then aimed to validate the impact of EC on the growth and differentiation of AT2 cells using the organoid model. To do so, we further enriched the starting population for AT2 cells using major histocompatibility complex class II (MHCII) selection (MHCII+) or Epcam followed by MHCII selection (Epcam+MHCII+), and studied the organoid formation of these cell populations compared to the Epcam+ population ( Figure 1H ). Pro-SPC staining of cytospins from these populations revealed that freshly isolated MHCII+ cells and Epcam+MHCII+ cells had a significantly higher expression of pro-SPC compared to Epcam+ cells, indicating further AT2 cell enrichment ( Figures 1I, J ). EC exposure led to a significant increase in the number of organoids formed by all three cell populations to a similar extent ( Figures 1K, L ), and similarly affected organoid size in all three preparations ( Figure 1M ).

Next, we exposed mouse precision cut lung slices (PCLS) to EC for 48 hours to investigate the effect of EC on already differentiated epithelial cells in their natural context. We found that EC did not affect any of the epithelial markers we tested, including Sftpc, Hopx, Foxj1, Scgb1a1 and Muc5ac ( Supplementary Figures 1A, B ). Collectively this indicates that EC affects the growth and differentiation of epithelial progenitors toward alveolar and airway phenotypes, but has little effect on the already differentiated epithelial cells themselves.

To explore the transcriptional effects of EC on organoids in more detail and to capture the early events associated with EC exposure, we exposed mouse organoid cultures (mouse Epcam+ cells co-cultured with CCL206 fibroblasts) to EC for 72 hours, after which we re-sorted the cells back into epithelial (Epcam+) and CCL206 fibroblast (Epcam-) fractions and subjected them to bulk RNA-sequencing separately ( Figure 2A ). The principal component analysis (PCA) plot indicated that epithelial cells grown in organoid cultures exposed to EC had a different gene signature compared to vehicle control ( Figure 2B ). Indeed, differential expression analysis revealed that EC induced 767 differentially expressed genes (DEGs) in epithelial cells (padj<0.05) ( Figure 2C ). Pathway analysis showed that pathways associated with inflammation (such as inflammatory response, allograft rejection, IL-6/JAK/STAT3 signaling, TNFA signaling via NFKB) and with proliferation (including Myc targets, E2F targets and G2M checkpoint) were significantly enriched in EC-exposed epithelial cells ( Figure 2D ). We note that Epithelial mesenchymal transition was upregulated, potentially related to the decrease in epithelial marker genes, and that pathways associated with lung development and repair including Hedgehog signaling, Wnt/beta-catenin signaling and Notch signaling (11) were reduced ( Figure 2D ). Interestingly, after 72 hour of EC-exposure, epithelial cells already had a reduced expression of various epithelial marker genes, including Sftpc, Abca3, Aqp5, Ager and Scgb1a1 ( Figure 2E ). We also assessed changes in epithelial cell signatures, which were based on the marker genes associated with each epithelial subtype. These epithelial signatures were used to run pathway analysis, which revealed that the AT2 and AT1 cell signatures were significantly reduced, whereas the club cell and bronchial goblet cell signatures were enriched ( Figures 2F, G ). These results indicate that EC affects various cellular processes and pathways associated with repair including proliferation and differentiation already at an early time point.

We also assessed the response of the fibroblasts from mouse organoid cultures exposed to EC for 72 hours. The PCA plot ( Figure 3A ) demonstrated that CCL206 fibroblasts also responded to EC treatment, and differential expression analysis showed that 533 genes were differentially expressed (padj<0.05) ( Figure 3B ). Pathway analysis showed that several pathways associated with proliferation (such as MYC targets, E2F targets and G2M targets) and metabolism (cholesterol homeostasis, glycolysis, oxidative phosphorylation) were upregulated. In addition, processes related to signaling pathways including MTORC1 signaling, KRAS signaling, p53 pathway, IL2/STAT5 signaling and TNF-α signaling via NFKB were enriched, whereas Hedgehog signaling was downregulated ( Figure 3C ). These data indicate that the cocktail not only affected epithelial cells, but also had a large effect on the fibroblasts in the organoid culture system ( Figures 3A–C ). Next, we pre-treated CCL206 fibroblasts with EC for 24 hours, washed the cells, after which we co-cultured them with mouse Epcam+ epithelial cells to form organoids. The co-culture itself was not exposed to EC. The fibroblast pre-treatment with EC was sufficient to induce formation of a significantly higher number of organoids ( Supplementary Figures 2A, B ), though organoid size was not affected ( Supplementary Figure 2C ). These data indicate a role for the fibroblasts in governing, at least part of, the EC effect on the epithelial cells.

In view of these data, we then aimed to identify specific mediators secreted by fibroblasts that were altered by EC and may impact epithelial progenitors. All genes that were significantly altered in the CCL206 fibroblast population re-isolated from EC-exposed organoid cultures were used as the starting point (using padj<0.05 as cut-off value). From these DEGs we extracted the growth factors, which were then selected for being potentially druggable and for having their corresponding receptor expressed on the Epcam+ epithelial cells. Finally, we selected the genes that had a logFold change of >1 or <-1, in order to identify the genes that were most highly changed ( Figure 4A ). Six genes emerged from this filtering process; Areg, Artn, Il33, Pthlh, Ptn and Tgfa ( Figure 4B ). As amphiregulin (Areg), IL-33 (Il33) and TGF-α (Tgfa) are known to influence epithelial cell behavior (19–23), we selected these factors as potential mediators of epithelial cell changes in response to EC. The expression of AREG, IL33 and TGFA was also evaluated in primary human non-COPD fibroblasts in mono-culture after 24 hours of EC-exposure. None of the genes were altered, although IL33 showed a trend toward increased expression (p=0.0857) ( Supplementary Figure 3 ), suggesting that co-culture is required for the EC effect on these genes to be evident.

To assess the role of IL-33, amphiregulin and TGF-α in epithelial cell changes in response to EC, we treated mouse organoids with these factors and studied their response. We also used an antibody against ST2 (receptor for IL-33), and gefitinib, an inhibitor of epidermal growth factor receptor signaling (EGFR, receptor for amphiregulin and TGF-α). The activity of amphiregulin and TGF-α was confirmed by assessing the activation of downstream signaling pathways in CCL206 fibroblasts (data not shown), whereas the activity of IL-33 was previously validated (24). We found that neither IL-33 nor the antibody against the IL-33 receptor affected organoid number or size after 14 days of treatment ( Figures 5A, B ). EC increased the number of organoids formed (p=0.0003), and EC was still able to increase the number of organoids in the presence of the IL-33 receptor antibody (p<0.0001). The expression of epithelial marker genes Aqp5, Hopx, Sftpc, Foxj1, Scgb1a1 and Muc5ac measured in day 14 organoids was not altered by either IL-33 or the antibody against the IL-33 receptor ( Figure 5C and Supplementary Figure 4A ). Whereas EC upregulated the expression of Muc5ac (p=0.0091), the combination of EC and IL-33 receptor antibody did not lead to a significant induction of Muc5ac expression, though there was a trend toward increase (p=0.0782).

After 14 days of exposure to amphiregulin or TGF-α, we observed that organoid formation was not affected ( Figure 5D ). We also found that the increased number of organoids formed in the presence of EC (p<0.0001) could not be inhibited by gefitinib at a concentration of 1 μM. However, gefitinib 10 μM almost completely abolished organoid formation (p<0.0001). The co-stimulation of EC and gefitinib 10 μM also led to a significant reduction in the number of organoids (p<0.0001). A similar trend was observed in organoid size ( Figure 5E ). Amphiregulin and TGF-α both did not have an effect, but gefitinib 1 μM and 10 μM both reduced organoid size (p=0.0199 and p<0.0001 respectively). Organoid size was increased by EC treatment (p=0.0240), which we showed in initial experiments as well. Interestingly, the co-stimulation of EC and gefitinib 10 µM resulted in a significantly decreased organoid size compared to vehicle control (p<0.0001). Finally, we evaluated the effect of amphiregulin and TGF-α on the expression of epithelial marker genes ( Figure 5F and Supplementary Figure 4B ). We note that organoids exposed to gefitinib could not be included in the gene expression studies because of low amounts of RNA isolated (likely due to the low number and small size of organoids formed in the presence of gefitinib). Whereas amphiregulin and TGF-α had no effect on the expression of Hopx, Sftpc, Scgb1a1 and Muc5ac, TGF-α 50 ng/ml significantly downregulated Foxj1 (p=0.0266) and increased the expression of Aqp5 (p=0.0316).