Lung tissue specimens were collected from Guangzhou Women and Children's Medical Center, including three lung tissues from type I CAPM patients, three lung tissues from type II CAPM patients, the adjacent tissues as control (N = 3). All participants signed an informed consent form. All experiments were approved by the Medical Ethics Committee of Guangzhou Women and Children's Medical Center, Guangzhou Medical University. The RNA was isolated using the RNeasy Mini Kit (74104, QIAGEN). NanoDrop ND-1000 was used to measure the RNA quantity and quality and an A260/A280 ratio close to 2.0 was considered pure RNA. Standard denaturing agarose gel electrophoresis was used to assess RNA integrity. Sharp and intense bands of 28S and 18S ribosomal RNA represented the qualified RNA integrity, and the intensity ratio of the 28S band/18S bands was ~2-fold.

RNA labeling and array hybridization was performed according to the protocol of Agilent miRNA Microarray System with miRNA Complete Labeling and Hyb Kit (Agilent Technology). Total miRNA from different samples was labeled by Cyanine 3-pCp and the labeled cRNA was redissolved in water after inspissation and desiccation. One micrograms of labeled cRNA was heated at 60°C for 30 min after fragmentation by 11 μL 10× blocking agent and 2.2 μL 25× fragmentation buffer. Finally, the labeled cRNA was diluted in 55 μL of 2× GE hybridization buffer. One hundred microliters of the hybridization solution was dispensed into the gasket slide and assembled to the gene expression microarray slide and then incubated for 17 h at 65°C in an Agilent Hybridization Oven. The Agilent Microarray Scanner (part number G2505C) was used to scan the hybridized arrays after washing and fixing.

Array images were analyzed using Agilent Feature Extraction software (version 11.0.1.1). The GeneSpring GX v14.9 software package (Agilent Technologies) was used to normalize quantile and process subsequent data acquired from the miRNA chip. Processed data were verified and optimized by alignment to human miRNA transcriptome, which represented all known miRNAs. miRNAs that at least three out of nine samples have flags in detected were chosen for subsequently analysis after normalization. Differentially expressed miRNAs between the two samples were filtered using the characteristic log 2 (Fold Change) ≥ 1.0, P ≤ 0.05, and FDR ≤ 1.0. Differentially expressed miRNAs with statistical significance between the two groups were presented through Volcano Plot and Hierarchical Clustering. R scripts were used to draw the hierarchical clustering.

The targets from the GEO database were analyzed by an online program Database for Annotation, Visualization, and Integrated Discovery (DAVID) (https://david.ncifcrf.gov/) (22). The GO chord R package and DAVID database were used to perform GO analysis and KEGG pathway maps with cut-off p < 0.05, respectively.

miRNA's target genes were predicted using targetscan7.1 (http://www.targetscan.org/mmu_71/) and mirdbV6A (http://mirdb.org/miRDB/) databases. In addition, microRNA–target interactions were experimentally validated using the database mirTarbase7.0 (http://mirtarbase.mbc.nctu.edu.tw/php/index.php). The targets of miRNAs were identified by Cytoscape, and modules of key miRNAs were screened by the Molecular Complex Detection (MCODE) with the following default parameters: node score cut-off =0.2, cut-off =2, k-core = 2, and max depth = 100.

Lung tissues were fixed with 4% paraformaldehyde and embedded in paraffin which were sliced into 4 μm sections. For hematoxylin–eosin staining, the section was stained with hematoxylin for 5–8 min after dewaxing and then stained with eosin for 1–3 min. For Masson staining, sections were stained with celestine blue dye solution for 5 min, followed by staining with hematoxylin, ponceau acid fuchsin for 5–8 min, respectively. Then, sections were differentiated with 1% phosphomolybdic acid aqueous solution for 5–10 min and stained with 1% bright green dye for 2–5 min, and mounting. After dehydration, these tissue sections were analyzed under a microscope (Olympus, Japan).

TRIzol reagent (9109, Takara) was used to extract total RNA from lung tissues. Two micrograms of RNA was used for cDNA synthesis using BestarTM qPCR RT kit (2220, DBI) according to the manufacturer's instructions. qRT-PCR was performed using ABI 7500 instrument (ABI7500, ABI, Foster City, CA, USA) in a 20 μL reaction volume, which included 10 μL of Bestar® SybrGreen qPCR master Mix (2043, DBI), 0.5 μL of each primer (10 μM), 1 μL of the cDNA template, and 8 μL of ddH2O. Amplification processes were as follows: 95°C, 2 min; 94°C, 20 s and 58°C, 20 s and 72°C, 20 s for 40 cycles. The relative expression of miRNAs was normalized to U6. Relative expression levels were calculated by 2^−ΔΔCt methods. Primers were synthesized by Sangon Biotech, the details of which are listed in Table 1.

The distribution of miRNA was identified by FISH. Briefly, lung tissue was fixed by 4% paraformaldehyde and cut into 4 μm sections. Then, these sections were pre-hybridized in 200 μL of prehybridization solution for 1 h at 37°C and then hybridized with 250 μL of hybridization solution containing FAM probe (6 ng/μl) overnight at 37°C. Mouse anti-DIG-HRP (G1401, Servicebio) and CY3-TSA (G3016-3, Servicebio) were added to the slides and incubated in a dark condition for 5 min after washing off the hybridization. Cells were then stained with DAPI (1:800) and incubated for 8 min in the dark after washing with wash buffer at 42°C. Finally, different fields were observed under a fluorescence microscope (Olympus, Japan) with FAM at 488 nm. The sequences of the probes are as follows: miR-590-3p: 5′-ACTAGCTTATACATAAAATTA-3′; miR-146a-3p: 5′-CTGAAGAACTGAATTTCAGAGG-3′; miR-140-5p: 5′-CTGTTTGTGGCTGGGCAGACGA-3′; miR-Let-7e-3p: 5′-GGAAAGCTAGGAGGCCGTATAG-3′; miR-21-3p: 5′-CTACCATCGTGACATCTCCATG-3′.

To binding sites of miR-4731-5p for complexin 2 (CPLX2), miR-3150a-3p to vesicle amine transport 1 (VAT1), miR-32-5p to F-box and WD repeat domain containing 7 (FBXW7), and miR-454-3p to SLAIN motif family member 1 (SLAIN1) were predicted using TargetScan. The 3′ UTR of CPLX2, VAT1, FBXW7, and SLAIN1 containing the seed binding site and its mutant sequences were synthesized and then cloned into the pmirGLO vector (Promega), respectively. 293T cells were cotransfected with the pmirGLO vector and miRNA mimics using the Lipo3000 Reagent (Invitrogen). Cell lysates were harvested at 48 h after transfection, and luciferase activities were determined using the Dual-Luciferase Reporter System (Promega) and a microplate reader (Tecan M1000, Switzerland). The sequence of miRNA mimics and miRNA NC were as follows: miR-4731-5p: XX, miR-3150a-3p: XX, miR-32-5p: XX, miR-454-3p: XX, miR-NC: sense: 5′-UUCUCCGAACGUGUCACGUTT-3′, antisense: 5′-ACGUGACACGUUCGGAGAATT-3′.

All statistical analyses were completed using SPSS 21.0 software (IBM, Armonk, NY, USA). The data were presented as mean ± standard deviation. At least three independent experiments were performed. Statistical comparisons were conducted using an unpaired t-test between two groups, and one-way analysis of variance (ANOVA) was used to compare more than two groups. The level of statistical significance was set at p < 0.05.

Type I and type II CPAM patients were collected to perform miRNA chip. As shown in Figure 1A, HE analysis showed that multiple cysts of different sizes were observed in both type I and type II CPAM. Of these, type I CPAM was lined with pseudostratified ciliated columnar epithelium along with mucous cells, whereas type II CPAM was lined with cuboidal to columnar epithelium. Masson staining showed that the collagen fibers were decreased but muscle fibers were increased in both type I and type II CPAM compared with control. Among the 2,549 miRNAs tested using the Agilent 8 × 60 K miRNA-array platform, a total of 2,134 miRNAs were found (Figures 1B,C and Supplementary Table 1). A total of 715 miRNAs were found to be differentially expressed with 394 upregulation and 321 downregulation in the type I CPAM patients (Figure 2A), and 106 miRNAs showed differential expression with 34 upregulation and 72 downregulation in the type II CPAM patients (Figure 2B) compared to healthy individuals (Supplementary Table 2). The results also were presented by heat map (Figures 2C,D). Further analysis showed that 14 upregulated (Figures 3A,C) and 59 downregulated (Figures 3B,C) miRNAs were found in both type I and type II CPAM patients.

To validate the accuracy of the miRNA microarray, the expression of six miRNAs was measured using qRT-PCR in type I and type II CPAM patients, including miR-21-3p, miR-590-3p, miR-146a-3p, and let-7e-3p, which showed downregulation, and miR-4523 and miR-548au-3p, which showed upregulation. As shown in Figure 4A, expression of miR-4523 and miR-548au-3p was significantly increased, whereas expression of miR-21-3p was significantly decreased both in type I and type II CPAM patients. The expression of miR-590-3p and miR-146a-3p was significantly downregulated in type I CPAM and type II CPAM groups compared with the control group, respectively. However, let-7e-3p expression was not changed in both type I and type II CPAM patients compared with the control group. In order to validate the results, the expression and the distribution of these miRNAs were also verified by FISH. The FISH results showed that the green fluorescence of miR-21-3p, miR-140-5p, miR-590-3p, miR-146a-3p, and let-7e-3p was significantly reduced, whereas the green fluorescence of miR-4523 and miR-548au-3p was significantly increased in both type I and type II CPAM patients compared with the control group (Figure 4B). These results were consistent with the results of miRNA microarray, demonstrating the accuracy of the miRNA microarray that was used for further analysis.

It has been demonstrated that miRNA can participate in various development processes by binding the 3′UTR of mRNA. Therefore, we investigated the targets of the common differentially expressed miRNAs by TargetScan7.1 and miRDBV6. For upregulated miRNAs, the results showed that 4,248 and 1,042 targets were predicted for upregulated and downregulated miRNAs both by TargetScan7.1 and miRDBV6, respectively (Supplementary Table 3). According to the miRNA–mRNA network analysis, four miRNAs were the key candidate miRNAs that play important roles in the regulation of CPAM, including miR-4731-5p and miR-3150a-3p, which showed upregulation (Supplementary Figure 1), and miR-32-5p and miR-454-3p, which showed downregulation (Supplementary Figure 2). In addition, targets that showed the best score to bind with these miRNAs were selected, including miR-4731-5p to CPLX2, miR-3150a-3p to VAT1, miR-32-5p to FBXW7, and miR-454-3p to SLAIN1 (Supplementary Figures 1, 2). Dual-luciferase reporter assay showed that the relative luciferase activity was significantly decreased when cotransfection of the 3′ UTR containing wild type binding sequence and miRNA mimics. However, no significant changes were observed when cotransfection the 3′ UTR containing their corresponding mutant binding sequence (Figure 5). These results demonstrated that CPLX2, VAT1, FBXW7, and SLAIN1 were direct targets of miR-4731-5p, miR-3150a-3p, miR-32-5p, and miR-454-3p, respectively.

To determine the function of the miRNAs, GO enrichment analysis was performed using the targets of the common upregulated or downregulated miRNAs. Based on the GOChord plotting function, for biological process (BP), the upregulated targets significantly enriched in regulation of intracellular signal transduction (GO:1902531), regulation of cell communication (GO:0010646), and regulation of signaling (GO:0023051), and the downregulated targets significantly enriched in positive regulation of cellular metabolic process (GO:0031325), positive regulation of nitrogen compound metabolic process (GO:0051173), and positive regulation of cellular process (GO:0048522). For molecular function (MF), upregulated targets were significantly enriched in protein binding (GO:0005515), protein kinase binding (GO:0019901), and kinase binding (GO:0019900), and the downregulated targets were significantly enriched in RNA polymerase II transcription factor activity, sequence-specific DNA binding (GO:0000981), regulatory region nucleic acid binding (GO:0001067), and DNA binding transcription factor activity (GO:0003700). For cellular component (CC), targets with upregulation were significantly enriched in intracellular (GO:0005622), intracellular part (GO:0044424), and cell (GO:0005623), and targets with downregulation were significantly enriched in intracellular part (GO:0044424), intracellular (GO:0005622), and intracellular organelle (GO:0043229) (Figures 6A,B and Supplementary Table 4). These results of GO analysis revealed the function of the targets in CPAM development and progression.

KEGG pathway analysis was performed for further analysis of all targets. The results showed that the targets of common upregulated miRNAs mainly were platelet activation (hsa04611, P = 1.99e−5), Ras signaling pathway (hsa04014, P = 2.56e−5), vascular smooth muscle contraction (hsa04270, P = 4.87e−5), MAPK signaling pathway (hsa04010, P = 5.88e−5), and focal adhesion (hsa04510, P = 9.81e−5) (Figure 7A and Supplementary Table 5). Targets of common downregulated miRNAs mainly were Axon guidance (hsa04360, P = 2.14e−5), proteoglycans in cancer (hsa05205, P = 3.91e−5), pathways in cancer (hsa05200, P = 4.64e−5), FoxO signaling pathway (hsa04068, P = 8.40e−5), and PI3K-Akt signaling pathway (hsa04151, P = 1.94e−5) (Figure 7B and Supplementary Table 5).