This work was based on two study arms. The first study arm included tissue specimens from explanted CLAD allografts at time of retransplantations. The second study arm used prospectively collected serum samples from lung transplant recipients in follow-up at our institution. This study was approved by the Institutional Review Board of the Medical University of Vienna, Austria (EK-No.2106/2017).

For quantification of tissue miR-21, RNA was extracted by 5 × 10 µm tissue sections from FFPE blocks using miRCURY™ RNA Isolation Kit—FFPE samples (Exiqon, Vedbaek, Denmark), according to the manufacturer’s instructions. cDNA was synthesized using High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific, Waltham, Massachusetts, United States) and individual miRNA-specific RT primers (Exiqon, Vedbaek, Denmark). Micro-RNA levels were quantified in duplicates from 4 μl cDNA, with SYBR Green PCR Master Mix and specific primers of the miRCURY LNA™ miRNA PCR assay, using the following settings on an Applied Biosystems™ 7500 Fast Real-Time PCR system (Thermo Fisher Scientific, United States): 2 min, 50°C; 10 s, 95°C; 40 cycles of 10 s, 95°C; 1 min, 60°C.

Serum RNA, including miRNAs, was extracted from 200 μl patient serum, by using the miRNeasy Serum/Plasma Advanced kit (QIAGEN, Germany) according to the manufacturer’s instructions. cDNA was synthesized from 2.5 μl of serum-RNA by using individual miRNA-specific RT primers contained in the 5× miRCURY RT reaction buffer and 10× miRCURY RT enzyme mix (QIAGEN, Germany), by using the following thermal cycler conditions: 60 min, 42°C; 5 min, 95°C. Circulating miRNA levels were quantified in duplicate from 4 μl cDNA, with SYBR Green PCR Master Mix and specific primers of the miRCURY LNA™ miRNA PCR assay, using the amplification condition explained above.

RT-qPCR data were analyzed via the comparative threshold cycle (Ct) method [6]. The concentration of circulating miR-21 was expressed as 2^−∆∆Ct and compared with to control samples.

ISH was performed according to the manufacturer’s protocol (QIAGEN, Hilden, Germany) (10), with some modifications. A double-DIG labeled miRCURY LNA™ microRNA detection probe with the sequences 5′-TCA​ACA​TCA​GTC​TGA​TAA​GCT​A-3′ and a U6 probe (positive control) was used, while a scrambled probe served as negative control. U6 small nuclear RNA (snRNA) is a noncoding RNA transcript used in pre-mRNA splicing expressed in all cells. Therefore, ISH for U6 revealed an intense signal in cell nuclei. Tissue sections (6 μm thick) were deparaffinized in descending ethanol solutions (99, 96, 70%) and digested with Proteinase-K (15 μg/ml) for 30 min at 37° using an Abbott hybridizer System. Then, LNATM probes were denatured and diluted in QIAGEN ISH buffer. Hybridization with 40 nM MiR21 probe, 1 nM U6 probe and 40 nM scramble probe was performed at 50°C for 60 min followed by stringent washes in 5 × SSC, 1 × SSC and 0.2 × SSC at 50°C and blocking. The digoxigenins were recognized by a specific anti-DIG antibody conjugated with Alkaline phosphatase (AP). Samples were stained with freshly prepared NBT/BCIP substrate reagent containing 0.2 mM Levamisole (2 h at 30°C) and slides incubated with KTBT buffer. Nuclear Fast Red was used as counter stain. Sections were analyzed microscopically.

MiR-21 intensity in fibroblast cytoplasm and extracellular matrix was retained for scoring purposes with a minimum cut-off at 10% of cells. Cases were classified as: 0 = negative or faint expression; 1 +: low expression (<25%); 2 +, moderate expression (25–50%); 3 +, strong expression (>50%). Cases with a score of three were regarded as positive in Kaplan-Meier curves.

All stainings were performed on sections of 2–3 µm thickness. IHC was conducted according to a standard protocol using the following antibodies: Vimentin (Clone V9, Biocare medical) at dilutions of 1:300, Notch intracellular domain (NICD) (Clone A-8, Santa Cruz Biotechnology) at dilutions of 1:50, p-Sma and Mad-related protein (SMAD) 2/3 (Clone C-8, Santa Cruz Biotechnology) at dilutions of 1:100, β-catenin (Clone 14, BD Transduction laboratories) at dilutions of 1:100 and E-cadherin (Clone NCH-38, DAKO) at dilutions of 1:2. Staining was either performed with a BenchMark Ultra or a BenchMark XT (Ventana, Tucson, AZ). Negative and positive controls demonstrated appropriate immunolabeling for each staining.

The proportion of epithelial cells or fibroblasts and extracellular matrix that were positive for each marker was classified as: 0 = negative or faint expression; 1 +: low expression (<25%); 2 +, moderate expression (25–50%); 3 +, strong expression (>50%). IHC and ISH were reviewed and scored by two independent researchers (A.B., F.O.), and cases with at least 1 + were regarded as positive.

Categorical variables are reported as percentage, continuous variables as mean (min-max). A X² test, Fisher exact test or one-way ANOVA test was used to test differences. Correlations were quantified with Pearson’s correlation coefficient. The values of serum miR-21 (2^−∆∆Ct) for CLAD patients and stable patients were compared by 2-way repeated measure ANOVA test to evaluate significant trends over time. A p-value of 0.05 or less was considered statistically significant. All the statistical analyses were conducted with IBM SPSS Statistics version 25 (IBM, Chicago, IL) and graphics were designed with GraphPad Prism 8.

Clinical characteristics of the 24 patients included in the histological analysis are shown in Table 1. Seventeen (70%) patients were female, age at transplantation was 26 ± 11 years and at CLAD diagnosis 33 ± 12 years. The most frequent underlying diagnosis was cystic fibrosis (9, 38%), followed by interstitial lung disease (7, 29%) and idiopathic pulmonary arterial hypertension (4, 16%). Thirteen (54%) patients did not receive any induction therapy, 7 (29%) patients received rabbit anti-thymocyte globulins and 4 (17%) alemtuzumab. Fifteen (63%) patients were treated by azithromycin at time of CLAD onset, and twelve (50%) underwent extracorporeal photopheresis.

Clinical characteristics of the 51 patients included in the serum analysis are detailed in Table 1. Gender and age were equally distributed between the CLAD and control group with 10 (40%) vs 14 (53%) females and 49 ± 14 vs 46 ± 13 years. In both groups, the most frequent underlying diagnosis was COPD (16, 64% vs 13, 50%), followed by cystic fibrosis (4, 16% vs 5, 19%). All twelve patients with RAS received azithromycin compared to nine (69%) patients with BOS (p = 0.036). Extracorporeal photopheresis was more often used in BOS (6, 46%) than in RAS (1, 8%) patients (p = 0.035).

MiR-21 was significantly upregulated in CLAD tissue (2^−∆∆Ct: 10.1 ± 9.2 vs 4.3 ± 4.4, p = 0.002) (Figure 1). RT-PCR data were supplemented by in situ hybridization of miR-21, in order to identify distinct expression patterns within the lung parenchyma. Five (30%) BOS samples and 6 (86%) RAS samples showed positive ISH staining for miR-21 (Figure 2; Table 2). In BOS samples, positivity of miR-21 was mainly prevalent in peribronchiolar fibroblasts, in the bronchiolar epithelium as well as in the myofibroblasts of bronchiolar obliterative lesions. In RAS lesions, miR-21 staining was most commonly found in parenchymal fibroblasts and the extracellular matrix (ECM) as well as in the interlobular septa. In both BOS and RAS samples, macrophages showed a positive miR-21 staining (Supplementary Figure S1). Of note, miR-21 was mostly absent in control lung specimens, however slight positivity of miR-21 was observed in bronchiolar epithelium.

Both, BOS (17, 100%) and RAS (6, 86%) specimens, showed strong expression of vimentin in peribronchiolar myofibroblasts, suggesting mesenchymal differentiation (Table 2; Figures 3, 4, Supplementary Figure S2). Notch-intracellular domain was positive in 12 (71%) BOS cases, predominantly in the cytoplasm of peribronchiolar myofibroblasts and bronchiolar epithelium, and in 4 (57%) RAS cases in the cytoplasm of interstitial fibroblasts. Finally, staining of p-SMAD 2/3 was positive in 7 (42%) BOS and 4 (57%) RAS specimens. pSMAD-2/3 expression was prevalent in interstitial fibroblasts as well as in the peribronchiolar myofibroblasts and bronchiolar epithelium in BOS allografts while in RAS allografts it was mainly present in interstitial fibroblasts (Table 2). β-catenin and E-cadherin stainings were negative in specimens of both phenotypes (Supplementary Figure S3). All four markers were negative in control tissue from donor lungs (Supplementary Figure S4). Correlations between IHC stainings of EMT markers and miR-21 ISH were calculated with Pearson correlation coefficients (Table 3). In BOS specimens, a strong positive correlation was found between miR-21 and p-SMAD 2/3 expression in the interstitium (r = 0.649, p = 0.006) as well as between miR-21 and NICD in the bronchiolar epithelium and myofibroblasts in the BO lesions (r = 0.827, p < 0.001). In RAS specimens, a positive correlation was found between miR-21 staining and NICD expression both in the fibroblasts in the interstitium (r = 0.837, p = 0.019) and peribronchial (r = 0.842, p = 0.018).

Figure 5 summarizes analysis of miR-21 expression difference over time between the groups. CLAD patients showed a non-significant trend towards higher miR-21 expression over time compared to stable patients (Panel A, p = 0.110). Panel B shows results of subgroup analysis with RAS patients having a significant increase in serum concentration of miR-21 overtime as compared to stable patients (p = 0.040). No difference was observed between BOS patients and control patients (p = 0.358).