A total of 70 large airway endobronchial biopsy (EBB) samples were collected from participants (12 ACO, 14 asthmatics, 10 COPD ex-smokers (ES), 12 COPD current-smokers (CS), 12 NLFS, and 10 HC) and were obtained from the Tasmanian Respiratory Tissue bank and Newcastle biobanks (Tasmanian Health and Medical Human Research Ethics Committee, ethics IDs: H0013051; the Hunter New England Human Research Ethics Committee reference no: 05/08/10/3.09). The details of tissue collection from the research volunteers were reported in our earlier publications (12–14). Among the participants with ACO, 7 were classified as GOLD stage I and II and 1 as GOLD stage III COPD considering their lung function, and 4 were classified as having severe asthma. The majority of participants with ACO were ex-smokers. Participants with COPD were of mild to moderate COPD (GOLD stage I and II). Fifty percent of the patients with ACO and asthma were on inhaled corticosteroids (800/750 - mcg/day). All participants with asthma and HC were non-smokers. None of the HC subjects had a history of respiratory illness. A summary of the participant demographics is presented in Table 1 .

The formalin-fixed, paraffin-embedded biopsy tissue blocks were sectioned at 3 µm and placed on positively charged glass slides. These sections were dried overnight. Paraffin sections were then dewaxed in xylene, and rehydrated using graded ethanol, followed by a wash in distilled water. Heat-Induced epitope retrieval was conducted in a decloaking chamber (Biocare Medical) using low pH Dako target retrieval solution (Cat#.S1699). The sections were then treated with 3% hydrogen peroxide (H₂O₂) in distilled water (v/v) to block endogenous peroxidase activity. Primary antibodies were applied to the tissue sections as follows: mouse monoclonal E-cadherin (1:200 dilution, ab1416, Abcam, Victoria, Australia), mouse monoclonal N-cadherin (1:150 dilution, ab98952, Abcam, Victoria, Australia), rabbit polyclonal S100A4 (1:1000, A5114, Dako, Victoria, Australia), mouse monoclonal Vimentin (1:200, M7020 Dako, Victoria, Australia), rabbit polyclonal Collagen IV (1:350, ab6586, Abcam, Victoria, Australia), incubated at room temperature. Bound antibodies are then elaborated using polymer enzyme backbone conjugated secondary antibody (Dako EnVision, K5007) and visualised using diaminobenzidine (DAB) chromogen (Dako EnVision, K7005). Harris hematoxylin was used for the nuclear staining. Before performing immunostaining, optimisation of manufacturer-specific immunohistochemical methods was conducted. We have published with these methods before (15–18).

Computer-assisted image analysis was performed with a Leica DM 500 microscope (Leica Microsystems, Germany) and a Leica ICC50W camera. Tissues with visible epithelium, RBM, and lamina propria (LP) were selected for image analysis. The images of the entire tissue area, including epithelium, RBM and LP, were captured at 40X brightfield, avoiding the overlapping area between images. Five images were randomly chosen for percentage stain and cell counts. For E- and N-cadherin, the percent staining expression was measured in the epithelium. For the mesenchymal markers, the marker-positive basal cells (brown) in the epithelial and the marker-positive cells in RBM were counted ( Figures 1E, F ), presented as cells per mm of RBM length. The observer was blinded to patient and diagnosis. The representative tissue micrographs of EMT markers in ACO and HC are provided in Figure 1 .

Length of fragments was measured from Collagen IV-positive tissues which separated the fragments from vessels ( Figure 1I ). Fragmentation of the RBM included pieces apparently hanging off and fully separated from the rest and clefts within the RBM. The total length of splits was added and divided by the length of RBM. Splits that were parallel to the RBM were also included in the measurement. Image analyses were performed using the image analysis software Image-Pro Plus 7.0.

Cigarette smoke extract (CSE) was prepared by bubbling smoke from one filter-less Kentucky research cigarette, 3R4F, which contains 9.5 mg of tar and 0.8 mg of nicotine, through 10 mL of Bronchial Epithelial Basal Medium (BEBM, Lonza) at a rate of one cigarette every 5 minutes (19). This was used immediately in subsequent cell culture experiments by diluting it with BEBM media to achieve a concentration of 1%. We previously determined that this concentration of CSE, assessed via dose-response curves, causes minimal toxicity to the cells while still inducing an immune response. All cells were grown at 37°C with 5% CO2 air.

Primary bronchial epithelial cells (pBECs) were collected from healthy (n=3) and ACO (n=3) subjects via bronchoscopy. pBECs from the healthy control and ACO groups were cultured in separate culture plates as air-liquid interface cultures. In each plate, transwells were spaced out one well apart to minimize cross-contamination. Treatments were added carefully into the basal compartment. A new cell culture pipette tip was used on each treatment for each well on every media change. Primary cells were grown in complete BEGM (Lonza) with growth factor supplements in a submerged monolayer culture. Once confluent, cells were trypsinised using a 1:10 dilution of the standard 0.25% trypsin-EDTA solution (Sigma), and once the cells are dissociated, FBS was used to neutralise and then seeded at 2 × 10⁵ cells in trans-wells in a 12-well plate (Corning) with PneumaCult™-Ex Plus medium containing amphotericin B (final concentration 250 μg/ml), and 2% penicillin-streptomycin until confluent (at least 3 days in both apical and basal compartments) (20–24). Once confluent, apical media was removed and the basal media was replaced with PneumaCult™-ALI medium containing amphotericin B (final concentration 250 μg/ml), and 2% penicillin-streptomycin and maintained at the air-liquid interface (ALI) phase for 21 days until fully differentiated. Fully differentiated ALI wells at day 21 were basolaterally treated with EMT drivers (25) TGF-β (5ng/mL) and interleukin (IL)-13 (10ng/mL) (26) for 10 days or with 1% cigarette smoke extract (CSE) for 2 days. At the end of the treatment, cells were harvested for RNA or protein using appropriate lysis buffers.

Air-liquid interface (ALI) samples were lysed with RIPA buffer containing protease inhibitor (Sigma-Aldrich, Australia) for 20 minutes at 4°C and were assessed for total protein concentrations using pierce™ BCA protein assay kits (Thermofisher Scientific, Australia) (27, 28). A total of 5ug protein was loaded into activated Mini-PROTEAN^® TGX Stain-Free™ precast gels (Bio-Rad laboratories Australia) and SDS-PAGE separation was performed. Standard immunoblot was performed against E-cadherin (14472S, Cell signalling), N-cadherin (ab18203, Abcam), and Vimentin (ab92547, Abcam) as described previously (29). β-actin (ab8227, Abcam) was used as the loading control. Protein bands were detected using SuperSignal West Femto Maximum Sensitivity Substrate reagents (ThermoFisher Scientific Inc.) on the ChemiDoc MP Imaging System (Bio-Rad Laboratories). Protein band density was measured using Image J software and normalized to the loading control (β-actin) level. The relative expression level of protein was calculated and graphed using GraphPad Prism 9.0 software.

Following the data distribution check using D’Agostino & Pearson test, intra- and inter-group variances were analysed using one-way ANOVA or nonparametric with multiple comparisons using uncorrected Dunn’s test. Unless otherwise mentioned, the results are reported as median and range. Univariate Spearman r was used for correlation analysis. Furthermore, the effect of ICS on the EMT markers was explored using a nonparametric test. A P value of <0.05 was considered significant. All analyses were done using GraphPad Prism v9 (San Diego, CA, USA).

The epithelial marker E-cadherin percentage expression in large airway epithelium decreased across all pathological groups ( Figure 2 ). In contrast, the mesenchymal markers N-cadherin percent expression, vimentin, and S100A4 positive cell counts generally increased in all pathological groups except in the patients with asthma. E-cadherin percent expression in patients with ACO was substantially decreased as compared to the HC (P <0.01) and NLFS (P <0.001) groups and tended to be lower than the COPD groups ( Figure 2A ). In addition, a notable decrease in E-cadherin expression was also noted in COPD-ES (P <0.05) and COPD-CS (P <0.05) groups as compared to NLFS; however, compared to HC the decrease in percent expression COPD-ES (P = 0.1021) and CS (P = 0.1178) was not statistically significant.

N-cadherin percent expression in the ACO group appeared to be higher (P = 0.1608) than in the HC group and tended to be lower than in the COPD groups (ES, P = 0.3107; CS, P = 0.0702) and NLFS (P = 0.5487) ( Figure 2B ). Compared to asthma, the percent expression of N-cadherin increased markedly in ACO (P <0.01). Although the percent expression of N-cadherin remained similar in between COPD groups, we further noticed a substantially higher expression of N-cadherin in COPD-ES and CS groups than in the HC (P <001 and <0.05, respectively) and asthma (P <0.001 and <0.0001, respectively). The ratio of E and N cadherin percent expression indicated a dominance of N cadherin, a mesenchymal marker as compared to HC which had near similar proportion of E and N cadherin percent expression ( Figure 3A ).

The number of vimentin-positive basal cells was highest in the patients with ACO followed by the patients with COPD ( Table 2 , Figure 2C ). However, compared to HC, the differences were statistically not significant (ACO, P = 0.6909; COPD-ES, P = 0.7259; CS, P = 0.2686). In asthma, vimentin-positive basal cells were similar to the HC (P = 0.5150).

The number of S100A4-positive basal cells was higher in all pathological groups ( Figure 2D ). Although in the ACO group S100A4 positive basal cells appeared to be higher than the HC (P = 0.3376) and asthma group (P = 0.7417), we noted significantly lower basal cells in ACO than in COPD-ES (P <0.01) and NLFS (P <0.05) groups. S100A4 positive basal cells were tended to be lower in ACO than in COPD-CS group (P = 0.1029). Furthermore, the S100A4-positive basal cells were markedly enhanced in both COPD-ES and CS groups as compared to the HC (P <0.01 and <0.05, respectively) and asthma (P <0.01 and <0.5, respectively). S100A4-positive basal cells in asthma were essentially normal. When we plotted the ratio of vimentin and S100A4 positive basal cells, we noted a higher proportion of S100A4 positive cells across all groups ( Figure 3B ).

The degree of fragmentation was greatest in patients with ACO followed by COPD-CS and ES, and lowest in the asthma group ( Figure 4C ). Enhanced fragmentation in the ACO group was statistically significant than the HC (P <0.01) and asthma (P <0.01). Although the fragmentation in the ACO group appeared to be higher than the COPD-CS (P = 0.3215) and ES (P= 0.2017), the difference was not statistically significant. Furthermore, the degree of fragmentation was substantially higher (P <0.05) in COPD-CS than in the asthma group and tended to be higher than in the HC (P=0.1086). In the COPD-ES group, the fragmentation tended to be higher (P= 0.2039) than the HC and lower than the COPD-CS group (P = 0.7612). The RBM fragmentation in patients with asthma was very similar to HC.

In general, similar to the epithelial basal cells, both vimentin and S100A4 positive cells ( Figures 4A, B ) in the RBM of pathological groups were elevated except in patients with asthma.

The vimentin-positive cells appeared to be elevated in the RBM of ACO as compared to HC although the difference was not statistically (P = 0.1610); however, as compared to asthma the increase in vimentin-positive cells was notable (P <0.05). The number of vimentin-positive cells in patients with ACO was similar to that of COPD-ES (P = 0.7397) and NLFS (P = 0.4331) groups. The COPD-CS group also had high vimentin-positive RBM cells, compared to HC and asthma groups (P <0.05 for both). On the other hand, the vimentin-positive cells in asthma tended to be lower than HC (P = 0.5641). Similar to the epithelial region, the ratio of vimentin and S100A4 RBM cells suggested a higher proportion of S100A4 positive cells across all groups ( Figure 3C ).

In the RBM, a significantly higher number of cells were stained for S100A4 in ACO (P <0.05), COPD-ES (P <0.05), COPD-CS (P <0.01), and NLFS (P <0.001) as compared to HC. The S100A4 stained cells remained comparable among patients with ACO, COPD-ES, COPD-CS, and NLFS. The S100A4 stained RBM cells in asthma remained similar to HC.

We performed an array of correlation analyses using the FEV₁/FVC ratio, smoking history with EMT markers from samples of ACO ( Table 2 ). In addition, we checked correlation ( Table 3 ) between EMT markers from samples of ACO and epithelial macrophage that we reported to be higher in ACO earlier ( Figure 5 ). Notable correlations trend considering spearman r values were (a) a trend of moderate positive correlation between mesenchymal markers (vimentin and S100A4 positive basal cells) in the epithelium and FEV₁/FVC ratio (spearman r, 0.3357 and 0.4825, respectively); (b) a trend of moderate positive correlations between smoking history and mesenchymal markers (Vimentin and S100A4 positive basal cells with the spearman r of 0.3825 and 0.2947, respectively); (c) a trend of weak positive correlation between E-cadherin and epithelial macrophages ( Figure 5A ) (d) significant and moderate positive correlation between epithelial macrophage and epithelial-mesenchymal markers (S100A4 and vimentin positive basal cells with the spearman r of 0.5245 and 0.6294; P = 0.0161 and 0.0420 respectively) ( Figures 5C, D ). We dichotomized the ICS-treated and nontreated patients with ACO to see if there is any effect on the EMT markers ( Figure 6 ) We noted a marked decrease in vimentin (P <0.05) and S100A4 (P <0.001) basal cells in ICS treated patients as compared to patients without ICS treatment. Furthermore, Vimentin and S100A4-positive RBM cells also appeared to be lower in ICS-treated patients as compared to patients without ICS treatment. Interestingly, the degree of fragmentation tended to be higher in ICS treated patients as compared to patients without ICS treatment.

The protein expression of EMT markers; E-cadherin, N-cadherin and Vimentin was measured by performing immunoblot. β-Actin was used as the internal control. The intensity of protein bands was measured, and the fold expression was graphed. Our results show that E-cadherin protein expression is significantly low in the ACO group compared to the healthy group at baseline level ( Figures 7A, B ). IL-13 and TGF-β drastically reduce the expression of E-cadherin in the healthy group but not in ACOS. The expression of N-cadherin protein level was similar between ACO and healthy across all treatments ( Figures 7A, C ). Interestingly, the expression of vimentin protein level was significantly higher in ACOS at baseline (p≤ 0.05) as well as with IL-13 and 1% CSE treatment ( Figures 7A, D ). We also compared relative expression change of N-cadherin and E-cadherin in response to treatment ( Figure 8 ). Although similar in ratios, N-cadherin expression in HC media control was higher than in ACOS (#P<0.05) media control. Within-group variability was more prominent in HC N-cadherin expression across the treatment groups than in ACOS patients. There were higher significant changes in expression (p<0.05, p<0.01, p<0.001, p<0.0001) between N-cadherin and E-cadherin in ACOS than in HC. The ratio change between N-cadherin and E-cadherin was most significant in the ACOS patients induced with TGF-β, double that of HC.