Impact of Adenosine Triphosphate Synthase Subunit β on TLR Signaling Pathway in Promoting Airway Remodeling and Heterogeneity of Small Airway Epithelial Cells in Chronic Obstructive Pulmonary Disease

Abstract

Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by persistent airway remodeling and epithelial dysfunction, but the molecular mechanisms underlying these pathological changes remain unclear. This study aimed to investigate the role of ATP synthase subunit β (ATP5B) in airway remodeling and epithelial cell heterogeneity in COPD.

Methods

Single-cell RNA sequencing data from small airway epithelial tissues of smokers and nonsmokers were analyzed to identify key cell clusters and hub genes. ATP5B was identified by integrating differentially expressed genes and epithelial cell markers. Functional validation was performed using in vitro experiments with 2% cigarette smoke extract (CSE)-treated BEAS-2B cells and in vivo studies in a CS/LPS-induced COPD mouse model. Molecular assays included RT-qPCR, Western blotting, ELISA, flow cytometry, and histological analysis. Gene set enrichment analysis (GSEA) was applied to explore potential signaling pathways.

Results

Twelve distinct epithelial clusters were identified, and ATP5B emerged as a hub gene significantly upregulated in COPD. Silencing ATP5B reversed CSE-induced apoptosis, reduced pro-inflammatory cytokine release (IL-6, TNF-α), and suppressed epithelial-mesenchymal transition. In vivo, ATP5B knockdown alleviated airway inflammation and remodeling. GSEA and experimental validation confirmed that ATP5B promotes airway remodeling via activation of the Toll-like receptor (TLR) signaling pathway.

Conclusion

These findings suggest that ATP5B may serve as a novel biomarker and therapeutic target for ATP5B plays a crucial role in airway remodeling in COPD by activating the TLR signaling pathway.

Introduction

Chronic obstructive pulmonary disease (COPD) is a severe respiratory disorder caused by smoking, environmental pollution, and genetic factors, making it one of the leading global health concerns [1, 2]. COPD is characterized by airway obstruction, inflammation, and remodeling, a critical feature in disease development and progression [3, 4]. Airway remodeling refers to structural and functional changes in the airways, including epithelial cell proliferation, basement membrane thickening, and smooth muscle contraction [3, 5]. However, the regulatory mechanisms underlying airway remodeling remain poorly understood, posing a significant challenge to developing effective therapeutic strategies.

Adenosine triphosphate (ATP) synthase is a crucial enzyme responsible for the intracellular production of ATP [6]. Previous studies have demonstrated that ATP synthase not only plays a central role in regulating cellular energy metabolism but is also involved in multiple biological processes, including apoptosis and oxidative stress regulation [7–9]. Dysregulation of these functions has been closely linked to the pathological progression of COPD [10]. Adenosine triphosphate synthase subunit β (ATP5B) is a key protein in cellular energy metabolism, forming an essential component of the mitochondrial ATP synthase complex [11]. ATP5B catalyzes the phosphorylation of adenosine diphosphate to generate ATP, providing the necessary energy to maintain cellular function and survival [10]. This process is critical for preserving the structural integrity and functionality of airway epithelial cells, particularly in disease conditions such as COPD, where airway remodeling and epithelial repair mechanisms rely on adequate energy supply [8, 12]. Moreover, ATP5B dysfunction may result in energy metabolism disorders, disrupting cell survival signaling and triggering programmed cell death [11, 13]. In COPD pathogenesis, abnormal apoptotic processes contribute to epithelial cell dysfunction, exacerbating airway inflammation and remodeling [8]. By regulating ATP production, ATP5B indirectly influences oxidative stress response, and its dysfunction may exacerbate oxidative damage, further promoting COPD progression [13, 14]. A study by Zuo et al. [15] reported that ATP5B expression was significantly elevated in asthmatic lung tissue samples compared to normal lung tissues. Additionally, ATP5B was found to promote airway smooth muscle cell proliferation, contributing to airway remodeling. However, research on the specific function and underlying mechanisms of ATP synthase in COPD remains limited [16].

In this study, we aimed to investigate the potential role of ATP5B in COPD and its underlying mechanisms to address the existing research gap regarding ATP synthase in airway remodeling. We utilized the single-cell RNA sequencing (scRNA-seq) dataset GSE123405, obtained from the Gene Expression Omnibus (GEO) database, and performed rigorous data quality control (QC), normalization, and analysis using the Seurat package. Through principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE) analysis, we identified key COPD-associated cell types and used the FindAllMarkers function to determine representative marker genes. The application of these bioinformatics approaches enabled a comprehensive analysis of the dataset, providing a robust foundation for identifying potential hub genes.

By intersecting epithelial cell marker genes with differentially expressed genes (DEGs), ATP5B was identified as a key hub gene. To further validate its role in COPD, we treated human bronchial epithelial cells (BEAS-2B) with cigarette smoke extract (CSE) and assessed the expression of ATP5B and epithelial-mesenchymal transition (EMT) markers using RT-qPCR and Western blot. Additionally, we employed the MTT assay, flow cytometry, and enzyme-linked immunosorbent assay (ELISA) to evaluate the effects of ATP5B on cell viability, apoptosis, and inflammation in BEAS-2B cells.

Our findings from scRNA-seq analysis and in vitro experiments suggest that ATP5B may be crucial in COPD airway remodeling. Further bioinformatics analysis and functional experiments indicate that ATP5B potentially promotes airway remodeling in COPD by activating the Toll-like receptor (TLR) signaling pathway. These results provide new insights into the pathogenesis of COPD and offer potential diagnostic and therapeutic targets, which may contribute to improving patient quality of life and disease management.

Materials and Methods

Data Source

The scRNA-seq dataset GSE123405 [17] was obtained from the GEO database (http://www.ncbi.nlm.nih.gov/geo/). This dataset includes small airway epithelial (SAE) tissue samples from three nonsmokers and three smokers. Additionally, we retrieved the microarray dataset GSE106986 [18], which consists of samples from 14 COPD patients and 5 healthy controls (see online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000547329).

scRNA-seq Analysis

The Seurat package was used to analyze the scRNA-seq data. First, the CreateSeuratObject function (parameters: min.cells = 3, min.features = 200) was applied to process the single-cell dataset and create a Seurat object [19–23]. These parameter settings were based on standard practices to remove low-quality cells while preserving cells with sufficient feature expression for downstream analysis. Retaining cells with more than 200 detected genes ensures that only high-quality cells with adequate gene expression profiles are included, thereby reducing noise introduced by damaged or low-quality cells. Next, the PercentageFeatureSet function calculated the RNA molecule count per cell and the mitochondrial gene percentage (percent.mt). This step is crucial for identifying and filtering out cells with abnormally high mitochondrial gene expression, which may indicate apoptotic or damaged cells that could distort downstream analyses. The FeatureScatter function was then employed to assess the correlation between nFeature and nCount. Following QC, cells were filtered out if they met the following criteria: nFeature <200; nCount <1,000 or >20,000; mitochondrial gene percentage >20%. This filtering step effectively removed non-cell contaminants, cell aggregates, and dead cells from the dataset.

Subsequently, data normalization was performed using the LogNormalize function, and the FindVariableFeatures function was applied to identify the top 1,000 highly variable genes for further analysis. PCA was conducted using the RunPCA function, and key principal components (PCs) were selected based on JackStraw and ScoreJackStraw analyses for subsequent t-SNE clustering. To identify marker genes for each cell cluster, the FindAllMarkers function (parameters: min.pct = 0.25, logfc.threshold = 0.25) was utilized. Annotation of cell types was performed using the CellMarker database. These parameter choices were designed to balance sensitivity and specificity, ensuring the identification of true marker genes that effectively distinguish different cell clusters. Finally, DotPlot and VlnPlot functions were used to visualize the expression patterns of marker genes across different cell types through bubble plots and violin plots. The Seurat package was chosen for this analysis due to its comprehensive functionality, flexibility, and wide acceptance in scRNA-seq data processing, advanced analysis, and visualization. It remains one of the most widely recognized tools for scRNA-seq research.

Differential Expression Analysis

DEGs in the GSE106986 dataset were identified using R’s “limma” package. The selection criteria for DEGs were set as |log₂ (fold change)| >1 and p value <0.05 [24].

Protein-Protein Interaction Analysis

Protein-protein interaction (PPI) analysis was conducted using the STRING database (https://string-db.org/) to explore interactions among the intersection of DEGs and marker genes – the analysis aimed to identify hub genes by determining those with the highest number of adjacent nodes [25, 26].

Receiver Operating Characteristic Analysis

Receiver operating characteristic curves were generated using the pROC package in R to evaluate the accuracy of candidate gene expression in distinguishing COPD status within the GSE106986 dataset.

Functional Enrichment Analysis

Gene set enrichment analysis (GSEA) was performed using GSE106986 data to compare pathway enrichment between high- and low-expression groups of ATP5B based on its median expression value. The “c2.cp.kegg.v7.4.symbols.gmt” gene set from MSigDB was used as a reference to assess ATP5B-regulated pathways. Gene Ontology (GO) functional enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed using an online tool to identify significantly enriched GO terms and KEGG pathways.

Cell Culture and Treatment

Human bronchial epithelial cells (BEAS-2B, ATCC, USA) were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% (volume/volume [v/v]) fetal bovine serum at 37°C in a humidified incubator with 5% CO₂ [27]. CSE was prepared by passing smoke from two unfiltered cigarettes into 10 mL of Dulbecco’s Modified Eagle Medium over 2 min, generating a 100% CSE solution (v/v). The solution was then sterilized using a 0.22-μm filter and stored at −80°C until use. Cells were treated with 2% (v/v) CSE for 24 h. To activate the TLR pathway, cells were stimulated with 10 μg/mL poly(I:C) (SIGMA, Munich, Germany), while phosphate-buffered saline (PBS) treatment served as the control group (online suppl. Fig. S2).

For BEAS-2B cell transfection, BEAS-2B cells were seeded at a density of 2 × 10⁵ cells per well in six-well plates. Lipofectamine 2000 (Invitrogen, USA) was used for plasmid transfection. The cells were divided into two groups: si-NC (negative control for silencing) and si-ATP5B (ATP5B knockdown group). The transfection plasmids were purchased from GenePharma (Shanghai, China).

Establishment and Grouping of the COPD Mouse Model

Male BALB/c mice (6–8 weeks old, weighing 20–22 g) were purchased from Dashuo Laboratory Animal Technology Co., Ltd. (Chengdu, China). The mice were housed under a 12-h light/dark cycle at an 18–22°C temperature and humidity of 50–60%. Before the experiment, the mice were acclimated for 1 week. The mice were then randomly assigned to six groups (n = 6 per group): control group (normal control, exposed to ambient air); COPD group (exposed to cigarette smoke [CS] and intranasally administered lipopolysaccharide [LPS]); sh-NC group (injected with a lentivirus encoding a negative control shRNA and exposed to CS and LPS); sh-ATP5B group (injected with a lentivirus encoding ATP5B shRNA and exposed to CS and LPS); sh-ATP5B + PBS group (injected with a lentivirus encoding ATP5B shRNA, exposed to CS, LPS, and treated with PBS); and sh-ATP5B + poly(I:C) group (injected with a lentivirus encoding ATP5B shRNA and the TLR agonist poly [I:C] [HY-107202, MCE, USA], and exposed to CS and LPS). To establish the ATP5B knockdown mouse model, lentiviral particles encoding either ATP5B shRNA or negative control shRNA (1.0 × 10⁹ TU, 50 μL) were subcutaneously injected into the shaved dorsal skin of the mice [28].

Two weeks after lentiviral injection, on day 1 of COPD induction, the last five groups of mice received an intranasal instillation of 2 μL of 5 mg/mL LPS solution to stimulate infection [28], followed by 1 month of continuous CS exposure. In CSE exposure protocol, CS was generated from Double Happiness cigarettes (China, 12 mg tar). Using a CS generator, mice were exposed to smoke from 10 cigarettes per day (1 cigarette/min, 35 mL puff volume over 2 s, every 60 s) and then placed in a chamber for 30 min. This procedure was performed for 6–8 consecutive weeks. On the final modeling day, an additional intranasal instillation of 2 μL of 5 mg/mL LPS solution was administered to enhance infection. Mice in the sh-ATP5B + PBS group and sh-ATP5B + poly(I:C) group were anesthetized with isoflurane and intranasally administered 100 μg poly(I:C) in 50 μL of sterile PBS [29]. Poly(I:C) (or PBS) administration was performed three times, with a 24-h interval between each dose [29]. Mice were sacrificed on the final day of COPD modeling and lentiviral transfection. Lung edema was assessed using the wet-to-dry weight ratio method (see online suppl. Fig. S3).

Bronchoalveolar Lavage Fluid Collection

A tracheotomy was performed to insert a tracheal cannula, enabling the collection of bronchoalveolar lavage fluid (BALF) using PBS (Wako, 041-20211; total volume: 3 mL). The total cell count in BALF was determined using a hemocytometer, while differential cell counts were assessed by analyzing 300 Diff-Quick-stained cells (Sysmex, 16920).

Hematoxylin and Eosin Staining

Immediately after sacrificing the mice, the right lung was inflated with 10% (v/v) buffered formalin at a pressure of 20 cm H₂O and then fixed. The fixed lung tissue was embedded in paraffin, sectioned into 4 μm slices, and subjected to histological staining. Hematoxylin and eosin (H&E) staining was performed following a standard protocol for histopathological evaluation: staining with hematoxylin (517-28-2, Solarbio) at room temperature for 10 min, followed by rinsing with running water for 30–60 s; differentiation using 1% (v/v) hydrochloric acid ethanol for 30 s, followed by immersion in running water for 5 min; staining with eosin at room temperature for 1 min; dehydration through an alcohol gradient (70%, 80%, 90%, 95%, and 100% [v/v] ethanol, with each step lasting 1 min); clearing with dimethylbenzene (phenol-xylene) for 1 min, followed by two rounds of treatment in xylene I and II for 1 min each; and mounting the stained sections with neutral resin in a fume hood and examining them under a BX50 optical microscope (Olympus). Mean alveolar septal thickness (MAST), mean linear intercept (MLI), and destruction index (DI) were calculated to assess lung structure alterations. Histological changes were scored on a scale from 0 (none) to 4 (severe), evaluating inflammatory cell infiltration and epithelial thickness changes (online suppl. Table S1).

RT-qPCR

Total RNA was extracted from BEAS-2B cells and mouse lung tissues using TRIzol reagent (Invitrogen, Carlsbad, CA, USA), followed by cDNA synthesis using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, USA). RT-qPCR was performed according to the manufacturer’s instructions for SYBR Premix EX Taq (Takara) in a total reaction volume of 20 μL. The PCR program was set as follows: initial denaturation at 95°C for 2 min, followed by 45 cycles of denaturation at 95°C for 15 s and annealing/extension at 60°C for 45 s. Gene expression levels were normalized to β-actin as an internal reference, and primer sequences are listed in online supplementary Table S2. Each experiment was repeated three times.

Western Blot

Total protein was extracted from BEAS-2B cells and mouse lung tissues using RIPA lysis buffer (Shanghai Beyotime Biotechnology, China), and protein concentrations were measured using the BCA Protein Assay Kit (Thermo Scientific, USA). A total of 12% (weight/volume) SDS-PAGE separated proteins and subsequently transferred onto PVDF membranes (Millipore, Germany) using a wet transfer method. Following the transfer, membranes were blocked with nonfat milk and incubated overnight at 4°C with the following primary antibodies: ATP5B (1:1,000, sc-55597, Santa Cruz, USA); E-cadherin (1:1,000, #14472, Cell Signaling); vimentin (1:1,000, #49636, Cell Signaling); N-cadherin (1:1,000, ab76011, Abcam); TGF-β (1:1,000, ab215715, Abcam); matrix metalloproteinase (MMP)-9 (1:1,000, ab76003, Abcam); GAPDH (1:1,000, ab9485, Abcam). After incubation, membranes were washed with TBST and incubated at room temperature for 1 h with HRP-conjugated goat anti-mouse IgG (#91196, Cell Signaling). Protein bands were detected using a chemiluminescent substrate, and ImageJ software (NIH, USA) was used for quantitative analysis. Protein expression levels were normalized to β-actin by calculating the gray intensity ratio. Each experiment was repeated three times.

MTT Assay

Cell proliferation was assessed using an MTT assay kit (Sigma-Aldrich, Zwijndrecht, Netherlands). BEAS-2B cells were seeded at a density of 2 × 10⁴ cells/mL in 24-well plates containing 200 μL of culture medium per well. MTT solution (1 mg/mL) was added to each well, followed by incubation. After removing the MTT solution, dimethyl sulfoxide was added to dissolve the formazan crystals. Absorbance (A value) was measured at 570 nm using a microplate reader (PerkinElmer). Each group included three replicates.

Flow Cytometry

Apoptosis in BEAS-2B cells was quantified using an Annexin V-FITC/PI double-staining assay. Cells were seeded at 2 × 10⁵ cells per well in six-well plates, digested with trypsin, and collected by centrifugation. Apoptotic cells were detected using an Annexin V-FITC apoptosis detection kit (BD Biosciences, Franklin Lakes, NJ, USA), followed by fluorescence analysis of Annexin V/PI staining using a BD FACSCalibur flow cytometer.

Enzyme-Linked Immunosorbent Assay

IL-1β, TNF-α, IL-6, and IFN-γ levels in cell culture supernatants, lung tissues, and BALF were measured using ELISA kits (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions. Each cell experiment was performed in triplicate.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism software. First, normality tests were conducted to assess the distribution characteristics of the data. For normally distributed data, results were presented as mean ± standard deviation, and two groups were compared using the independent sample t test. For non-normally distributed data, results were expressed as the median and interquartile range, and two groups were compared using the Mann-Whitney U test. For multiple-group comparisons, a one-way ANOVA was conducted if the data followed a normal distribution with homogeneous variance. If ANOVA indicated at least one significant difference among the groups, Tukey’s multiple comparison test was applied for pairwise group comparisons. For non-normally distributed data or heterogeneous variance, the Kruskal-Wallis H test was used for multiple-group comparisons, followed by Dunn’s multiple comparison test for post hoc analysis. A p value <0.05 was considered the threshold for statistical significance, indicating that differences were deemed statistically significant when p < 0.05.

Results

Cellular Heterogeneity and Diverse Cell Clusters in SAE of Smokers

Since smoking is one of the most common and significant risk factors for COPD and contributes to airway remodeling in COPD patients [30], we sought to determine the cell type-specific changes in SAE cells induced by smoking. We obtained scRNA-seq data from three nonsmokers and three smokers in the GSE123405 dataset to achieve this. The Seurat package in R was used for data processing. We retained genes detected in at least three cells and filtered out cells with fewer than 200 detected genes. Figure 1a illustrates the nFeature count, nCount per cell, and percent.mt values. Correlation analysis showed a strong correlation between nCount and nFeature (r = 0.97, Fig. 1b), indicating high dataset reliability. To ensure data quality, low-quality cells were filtered using the criteria nFeature >200, 1,000< nCount <20,000, and percent.mt <20%. After filtering, 6,174 high-quality cells were normalized, and variance analysis was performed to identify 1,000 highly variable genes for subsequent analysis (Fig. 1c).

Fig. 1.

QC and variance analysis of single-cell sequencing data from three nonsmokers and three smokers in the SAE. a QC of 13,658 cells from three nonsmokers and three smokers. The three scatter plots display nFeature_RNA (number of detected genes per cell), nCount_RNA (gene expression levels), and percent.mt (mitochondrial gene expression levels). b The correlation between nCount and nFeature, with a correlation coefficient of r = 0.97, indicates the reliability of the dataset. c Scatter plot displaying genes with high-expression variability across different cells, where red dots represent highly variable genes and black dots represent stable genes.

Next, we performed PCA on the 1,000 highly variable genes using the RunPCA function. The results indicated no significant batch effects among the six samples (Fig. 2a). Based on JackStraw and ScoreJackStraw analyses, we selected the top six PCs with p < 0.05 for further analysis (Fig. 2b). Scatter plots and heatmaps of these PCs were generated using the VizDimLoadings and DimHeatmap functions (online suppl. Fig. S4A, B). Using t-SNE dimensionality reduction analysis, we clustered the cells into 12 distinct cell clusters (Fig. 2c), with cells from different samples represented by different colors (Fig. 2d). These findings demonstrate pronounced cellular heterogeneity in the SAE of smokers, confirming the presence of 12 distinct cell clusters within SAE tissues. It should be noted that although this study utilized the same GSE123405 dataset as Zuo et al. [17], scRNA-seq analysis is highly dependent on specific preprocessing steps and parameter settings – such as cell filtering criteria, normalization methods, the number of PCs selected, and clustering resolution. As a result, variations in the presentation of t-SNE clustering plots are expected. These differences are common in scRNA-seq analyses and do not affect the reliability or scientific validity of downstream cell-type annotation or key gene identification in this study.

Fig. 2.

scRNA-seq analysis reveals cell cluster differences in the SAE of nonsmokers and smokers. a PCA clustering analysis of cells. b JackStrawPlot analysis to determine six PCs with p < 0.05 for t-SNE dimensionality reduction analysis. c t-SNE clustering analysis classifying cells into 10 distinct cell clusters. d t-SNE analysis displaying cell distribution differences across sample sources.

Differential Analysis and Marker Gene Annotation Reveal the Molecular Characteristics of 12 Cell Clusters

Differential analysis and marker gene annotation identified 4,989 marker genes that distinguish the 12 cell clusters. Figure 3a presents a heatmap of each cluster’s top five marker genes. Based on marker gene expression patterns, cell-type annotation was performed using the CellMarker database and previously reported gene markers. The 12 clusters were classified into seven known cell types: epithelial cells (clusters 0, 2, 4), regulatory T cells (cluster 1), macrophages (clusters 3, 6), B cells (cluster 5), T cells (clusters 7, 8), airway secretory cells (cluster 9), and neuroendocrine cells (clusters 10, 11) (Fig. 3b). The proportion of each cell type varied significantly across different samples (online suppl. Fig. S5A). Online supplementary Figure S5B displays the top five genes significantly associated with each cell type. We successfully annotated all 12 cell clusters as seven known cell types.

Fig. 3.

Cell annotation. a Heatmap displaying the top five marker genes for each cell cluster, with yellow indicating higher expression levels. b Annotation of 12 cell clusters into seven distinct cell types.

PPI Network Analysis of Differentially Expressed and Marker Genes Identifies Potential Biomarkers in COPD

As EMT is a key feature of airway remodeling in COPD [31], epithelial cells were selected as the target cell population for further investigation. A total of 863 marker genes were identified within this cell group. Additionally, differential expression analysis of the GEO dataset GSE106986 identified 1,103 DEGs in COPD (Fig. 4a). An intersection analysis between epithelial cell marker genes and DEGs yielded 40 overlapping genes (Fig. 4b). A PPI network was then constructed using STRING to analyze these 40 common genes (Fig. 4c), identifying two hub genes, ATP5A1 and ATP5B, which were highly expressed in COPD samples (Fig. 4d). Furthermore, receiver operating characteristic analysis demonstrated that ATP5A1 and ATP5B expression could effectively distinguish COPD patients from healthy controls (Fig. 4e). These findings suggest that ATP5A1 and ATP5B may serve as potential biomarkers and therapeutic targets for COPD airway remodeling.

Fig. 4.

Identification of hub genes potentially promoting COPD airway remodeling. a Volcano plot of DEGs in 14 COPD samples and 5 control samples from GSE106986. b Venn diagram showing the intersection of DEGs and marker genes in GSE106986. c PPI network of marker genes. d Expression levels of hub genes in GSE106986. e ROC curve analysis of hub genes in GSE106986. “*” indicates p < 0.05 compared to the control group. ROC, receiver operating characteristic.

ATP5B Silencing Attenuates CSE-Induced BEAS-2B Cell Injury and EMT

COPD is characterized by irreversible airflow limitation, primarily caused by smoking. The bronchial epithelium is one of CSE’s most affected cell types, making in vitro models an effective representation of long-term CSE exposure [32]. Based on previous studies, 2% CSE treatment of human bronchial epithelial cells (BEAS-2B) is a widely used in vitro model for studying the mechanisms underlying COPD pathogenesis [33, 34]. We assessed the expression levels of ATP5A1 and ATP5B in BEAS-2B cells treated with 2% CSE using RT-qPCR. The results revealed that ATP5B expression was significantly upregulated, whereas ATP5A1 expression remained unchanged (Fig. 5a). Based on these findings, ATP5B was selected as the target gene. Western blot analysis further confirmed that ATP5B protein levels significantly increased following CSE treatment (Fig. 5b).

Fig. 5.

ATP5B mediates CS-induced BEAS-2B cell injury and EMT. a RT-qPCR analysis of ATP5B and ATP5A1 expression in BEAS-2B cells. b Western blot analysis of ATP5B expression in BEAS-2B cells. c RT-qPCR validation of ATP5B knockdown efficiency in each group. d MTT assay measuring BEAS-2B cell viability. e Flow cytometry analysis of apoptosis in BEAS-2B cells. f ELISA quantification of inflammatory cytokines (IL-1β, TNF-α, IL-6, and IFN-γ) in cell supernatants. g RT-qPCR analysis of EMT markers (E-cadherin, vimentin, N-cadherin). h Western blot analysis of EMT markers (E-cadherin, vimentin, N-cadherin). i Western blot analysis of airway remodeling factors TGF-β and MMP-9. Data are expressed as mean ± SD. Comparisons between the two groups were conducted using the independent sample t test. All cell experiments were performed in triplicate. “*” indicates p < 0.05 compared to the control or si-NC group. SD, standard deviation.

To investigate the role of ATP5B in CSE-induced BEAS-2B cell injury and EMT, si-ATP5B plasmids were transfected into CSE-treated BEAS-2B cells. Among the different siRNA sequences tested, si-ATP5B-2 exhibited the most effective silencing and was selected for subsequent experiments (Fig. 5c). MTT assays and flow cytometry showed that, compared to the si-NC group, the si-ATP5B group exhibited increased cell viability and reduced apoptosis (Fig. 5d, e). Furthermore, ELISA analysis of inflammatory cytokines (IL-1β, TNF-α, IL-6, and IFN-γ) in cell supernatants revealed that their expression levels were significantly downregulated in the si-ATP5B group compared to the si-NC group (Fig. 5f).

Furthermore, RT-qPCR and Western blot analyses revealed that, compared to the si-NC group, the si-ATP5B group exhibited increased expression of the epithelial marker E-cadherin, while the expression of mesenchymal markers vimentin and N-cadherin was reduced (Fig. 5g, h). These findings indicate that ATP5B silencing suppresses CSE-induced BEAS-2B cell injury and EMT. Since TGF-β and MMP-9 play essential roles in airway remodeling, epithelial cells produce growth factors, such as TGF-β, in response to epithelial injury, which act on the underlying tissue to regulate repair and remodeling processes. TGF-β is a key mediator of airway remodeling and a well-known inducer of EMT involved in tissue repair [35]. MMPs, particularly MMP-9, are implicated in pulmonary emphysema, contributing to elastin degradation and abnormal alveolar remodeling. CS exposure has been reported to increase MMP-9 levels in mice. Western blot analysis further demonstrated that, compared to the si-NC group, the si-ATP5B group exhibited a significant reduction in TGF-β and MMP-9 expression (Fig. 5i). These results suggest that ATP5B silencing mitigates epithelial cell injury and EMT, potentially influencing airway remodeling in COPD.

ATP5B Promotes Airway Remodeling in COPD by Regulating the TLR Pathway

To further explore the role of ATP5B in COPD airway remodeling, we extracted expression data from 14 COPD patients in the GSE106986 dataset. GO and KEGG analyses were performed to compare DEGs between the ATP5B high-expression and low-expression groups. The results indicated that ATP5B is closely associated with key pathophysiological processes in COPD, including leukocyte migration, apoptosis, and cell adhesion (Fig. 6a–c). Additionally, GSEA was conducted to investigate the potential mechanisms of ATP5B in COPD. Based on the median expression level of ATP5B, COPD samples were divided into high- and low-expression groups for GSEA. The results revealed that the TLR signaling pathway was significantly more active in the ATP5B high-expression group (Fig. 6d), a pathway that has been confirmed to mediate airway remodeling [36]. These findings suggest that ATP5B may promote COPD airway remodeling by regulating the TLR signaling pathway.

Fig. 6.

ATP5B promotes COPD airway remodeling via regulation of the TLR pathway. a Heatmap displaying the top 20 DEGs between the high and low ATP5B expression groups in GSE106986. b GO analysis of ATP5B-associated genes. c KEGG pathway analysis of ATP5B-associated genes. d GSEA analysis shows differences in enriched pathways between the high and low ATP5B expression groups. e RT-qPCR analysis of TLR2, TLR3, and TLR4 expression in BEAS-2B cells across different groups. f RT-qPCR analysis of TLR2, TLR3, and TLR4 expression following ATP5B silencing. g MTT assay measuring BEAS-2B cell viability. h Flow cytometry analysis of apoptosis in BEAS-2B cells. i ELISA quantification of inflammatory cytokines (IL-1β, TNF-α, IL-6, and IFN-γ) in cell supernatants. j RT-qPCR analysis of EMT markers (E-cadherin, vimentin, and N-cadherin). k Western blot analysis of EMT markers (E-cadherin, vimentin, and N-cadherin). l Western blot analysis of airway remodeling markers TGF-β and MMP-9. Data are presented as mean ± SD. Comparisons between the two groups were conducted using the independent sample t test. All cell experiments were performed in triplicate. “*” indicates p < 0.05 compared to the si-NC or si-ATP5B + PBS group. SD, standard deviation.

To validate this hypothesis, we performed RT-qPCR to assess TLR2, TLR3, and TLR4 expression following CSE treatment. The results showed a significant upregulation of TLR2, TLR3, and TLR4 levels in CSE-treated cells (Fig. 6e). Furthermore, TLR2, TLR3, and TLR4 expressions were significantly reduced in the si-ATP5B group compared to the si-NC group (Fig. 6f). To further confirm the role of ATP5B in airway remodeling, poly(I:C), a TLR agonist, was used to stimulate si-ATP5B-transfected cells. The results demonstrated that TLR activation reversed the effects of ATP5B silencing, leading to reduced cell viability, increased apoptosis, elevated inflammation-related cytokine expression, enhanced EMT, and upregulation of airway remodeling markers TGF-β and MMP-9 (Fig. 6g–l). These findings suggest that ATP5B promotes epithelial cell injury and EMT by activating the TLR pathway, thereby contributing to COPD airway remodeling.

Successful Establishment of the COPD Mouse Model

A COPD mouse model was successfully established using CS and LPS exposure. Compared to the control group, the lung wet-to-dry weight ratio was significantly increased in COPD model mice (Fig. 7a). H&E staining was performed to evaluate lung tissue inflammation. In the control group, the airway epithelium remained intact, with no signs of degeneration, necrosis, thickening, or narrowing, and alveolar structures were well preserved. In contrast, COPD model mice exhibited epithelial shedding, extensive inflammatory cell infiltration, reduced alveolar cell count, and significant alveolar fusion. Compared to the control group, MAST was shorter, while MLI, DI, and lung injury scores were significantly elevated in the COPD group (Fig. 7b–e). Next, immune cell infiltration was analyzed. Compared to the control group, COPD model mice exhibited a higher proportion of leukocytes, neutrophils, and macrophages (Fig. 7f). The expression levels of inflammatory factors IL-1β, TNF-α, IL-6, and IFN-γ in the BALF and lung tissue of mice were measured by ELISA. The results showed that, compared with the control group, the expression levels of IL-1β, TNF-α, IL-6, and IFN-γ were significantly upregulated in the COPD group (Fig. 7g).

Fig. 7.

Establishment of the COPD mouse model. a Lung W/D weight ratio in each group. b MAST in lung tissues of each group. c MLI of lung tissues in each group. d DI of lung tissues in each group. e Representative H&E-stained images of lung tissues at ×100 and ×400 magnification, with corresponding lung injury scores; scale bar, 50 μm. f Proportions of immune cells in lung tissues of each group. g Expression levels of IL-6, IL-1β, TNF-α, and IFN-γ in BALF and lung tissues of mice. h Western blot analysis of EMT markers (E-cadherin, vimentin, and N-cadherin) in lung tissues. i RT-qPCR analysis of EMT markers (E-cadherin, vimentin, and N-cadherin) in lung tissues. j Western blot analysis of airway remodeling-related factors TGF-β, MMP-9, and ATP5B in lung tissues. Data are presented as mean ± SD, and two groups were compared using the independent sample t test. n = 6. “*” indicates p < 0.05 compared to the control group. W/D, wet-to-dry; SD, standard deviation.

The expression of EMT markers (E-cadherin, vimentin, N-cadherin) in the lung tissue of mice was assessed by Western blot and RT-qPCR. The results revealed that, compared with the control group, the expression levels of E-cadherin, vimentin, and N-cadherin were upregulated in the lung tissue of the COPD group (Fig. 7h, i). Additionally, Western blot analysis demonstrated that the expression of TGF-β, MMP-9, and ATP5B in the lung tissue of the COPD group was significantly upregulated compared to the control group (Fig. 7j). These findings indicate that in the CS/LPS-induced COPD mouse model, there is a significant inflammatory response, oxidative stress, damage to airway epithelial cells, and EMT.

ATP5B Silencing Suppresses the TLR Pathway, Reducing Inflammation and Oxidative Stress in CS/LPS-Induced COPD Mice

To investigate the role of ATP5B in COPD pathogenesis, we first validated ATP5B expression levels in the lung tissues of different mouse groups. Western blot and RT-qPCR analyses showed that ATP5B expression was significantly reduced in the sh-ATP5B group compared to the sh-NC group. However, ATP5B levels were comparable between the sh-ATP5B + poly(I:C) and the sh-ATP5B + PBS groups (Fig. 8a, b). H&E staining was performed to assess lung inflammation. In the sh-NC group, bronchial epithelial shedding, extensive inflammatory cell infiltration, a reduced alveolar cell count, and significant alveolar fusion were observed. Compared to the sh-NC group, the sh-ATP5B group exhibited reduced inflammatory infiltration, increased MAST, decreased MLI, and significantly lower DI and lung injury scores. However, in the sh-ATP5B + poly(I:C) group, the addition of the TLR agonist poly(I:C) reversed the protective effects of ATP5B silencing, leading to worsened lung pathology (Fig. 8c, d).

Fig. 8.

Effects of ATP5B silencing on COPD mice and the TLR pathway. a Western blot analysis of ATP5B expression in lung tissues of each group. b RT-qPCR analysis of ATP5B knockdown efficiency in lung tissues of each group. c Lung W/D weight ratio, MAST, MLI, and DI in each group. d Representative H&E-stained lung tissue images with lung injury scores at ×100 and ×400 magnification; scale bar, 50 μm. e Proportion of immune cells in lung tissues of each group. f Expression levels of IL-6, IL-1β, TNF-α, and IFN-γ in BALF and lung tissues. g RT-qPCR analysis of TLR2, TLR3, and TLR4 expression in lung tissues of each group. h RT-qPCR analysis of EMT marker mRNA (E-cadherin, vimentin, and N-cadherin) in lung tissues. i Western blot analysis of EMT markers (E-cadherin, vimentin, and N-cadherin) in lung tissues. j Western blot analysis of airway remodeling-related factors TGF-β and MMP-9 in lung tissues. Data are presented as mean ± SD, and two groups were compared using the independent sample t test. n = 6. “*” indicates p < 0.05 between the two groups. W/D, wet-to-dry; SD, standard deviation.

Next, immune cell infiltration was analyzed. The sh-ATP5B group exhibited a lower proportion of leukocytes, neutrophils, and macrophages than the sh-NC group. In contrast, compared to the sh-ATP5B + PBS group, the sh-ATP5B + poly(I:C) group exhibited increased immune cell infiltration, suggesting that TLR activation counteracts the effects of ATP5B silencing (Fig. 8e). ELISA analysis was performed to measure the levels of inflammatory cytokines (IL-1β, TNF-α, IL-6, and IFN-γ) in BALF and lung tissues. Compared to the sh-NC group, the sh-ATP5B group exhibited significantly reduced levels of these inflammatory markers. However, in the sh-ATP5B + poly(I:C) group, IL-1β, TNF-α, IL-6, and IFN-γ levels were significantly increased compared to the sh-ATP5B + PBS group (Fig. 8f).

To further investigate the impact of ATP5B silencing on TLR pathway activation, we analyzed TLR2, TLR3, and TLR4 expression in lung tissues using RT-qPCR. The sh-ATP5B group exhibited significantly lower TLR2, TLR3, and TLR4 expression levels than the sh-NC group. Conversely, TLR2, TLR3, and TLR4 expressions were significantly increased in the sh-ATP5B + poly(I:C) group compared to the sh-ATP5B + PBS group (Fig. 8g).

Additionally, Western blot and RT-qPCR were used to evaluate the expression of EMT markers (E-cadherin, vimentin, and N-cadherin) in lung tissues. Compared to the sh-NC group, the sh-ATP5B group exhibited lower expression levels of these EMT markers. However, E-cadherin, vimentin, and N-cadherin expression was significantly upregulated in the sh-ATP5B + poly(I:C) group compared to the sh-ATP5B + PBS group (Fig. 8h, i). Furthermore, Western blot analysis revealed that TGF-β and MMP-9 expression was significantly downregulated in the sh-ATP5B group compared to the sh-NC group. Conversely, TGF-β and MMP-9 levels were significantly elevated in the sh-ATP5B + poly(I:C) group compared to the sh-ATP5B + PBS group (Fig. 8j). These findings suggest that ATP5B inhibition suppresses TLR pathway activation, reducing inflammation and oxidative stress in CS/LPS-induced COPD mice.

Discussion

This study conducted bioinformatics-based predictions to investigate the mechanisms underlying heterogeneity and airway remodeling in SAE cells of COPD. Unlike previous studies, which primarily focused on the transcriptomic analysis of COPD SAE [37, 38], this study applied scRNA-seq to explore cellular heterogeneity and gene interactions. PPI network analysis was also performed to identify hub genes, providing a more comprehensive understanding of COPD pathogenesis through a novel research strategy.

The GSE123405 dataset was selected for analysis, offering more detailed and accurate cell-type information than previous studies’ datasets. QC and normalization were conducted using the Seurat package, followed by PCA and t-SNE clustering, which identified 12 distinct cell clusters. This study achieved more refined cell-type classification than prior studies, establishing a more robust foundation for downstream analyses.

This study identified COPD-associated DEGs and compared them with findings from previous research. Through PPI analysis, two hub genes, ATP5A1 and ATP5B, were successfully identified, which play a critical role in COPD pathogenesis. Compared to previous studies, the PPI analysis conducted in this study provided a more comprehensive and systematic overview, offering valuable insights into the key genes and pathways involved in COPD airway remodeling.

Furthermore, the expression of the selected hub genes was validated in COPD samples. Consistent with previous findings, ATP5B was found to be upregulated in COPD. GO, KEGG enrichment analysis, and GSEA confirmed that ATP5B regulates airway remodeling in COPD. These results align with prior observations, further supporting the critical role of ATP5B in COPD pathogenesis.

As a key protein in ATP synthesis, ATP5B plays a crucial role in cellular energy metabolism, and its dysfunction may directly affect cellular energy homeostasis, leading to pulmonary cell dysfunction. However, the role of ATP5B in COPD remains unexplored. Recent studies, such as that by Qin et al. [39], have identified glycogen synthase kinase 3β (GSK3β) as a potential regulator of COPD-related inflammation. Like GSK3β, ATP5B is involved in multiple cellular processes, including energy metabolism, neuronal development, and glycogen synthase regulation, which may indirectly affect cellular energy balance and ATP synthesis pathways in COPD. To investigate the role of ATP5B in COPD airway remodeling, this study employed both in vivo and in vitro models. The results demonstrated that ATP5B expression was significantly upregulated in BEAS-2B cells treated with CSE. Further functional experiments revealed that targeting ATP5B suppressed CSE-induced BEAS-2B cell injury and EMT [40, 41]. One of the hallmark features of COPD is structural alterations in the airways, commonly referred to as airway remodeling. This study provides new insights into the role of ATP5B in COPD pathogenesis, suggesting its potential as a therapeutic target for airway remodeling. ATP5B may influence airway smooth muscle cells and fibroblast proliferation and differentiation by modulating cellular energy homeostasis and survival, thereby contributing to airway remodeling. In vivo experiments demonstrated that CS/LPS-induced COPD mice exhibited significant inflammation, oxidative stress, airway epithelial injury, EMT, and elevated expression of TLR pathway-related proteins. Silencing or inhibiting ATP5B in the COPD mouse model resulted in downregulated TLR2, TLR3, and TLR4 expression, reduced inflammatory cytokine levels, and decreased EMT marker expression. In contrast, poly(I:C), a TLR agonist, reversed these effects by increasing TLR expression levels, upregulating inflammatory cytokines, and promoting EMT marker expression. Previous studies have shown that poly(I:C) enhances inflammatory mediator production, leading to airway smooth muscle thickening, goblet cell hyperplasia, and ultimately airway remodeling [42]. Mei et al. [43] also reported that poly(I:C) promotes airway inflammation and remodeling in CS-exposed mice via NF-κB and MAPK signaling pathways. These findings align with previous observations, further confirming the critical role of ATP5B in COPD airway remodeling, potentially mediated through TLR pathway regulation [41].

In summary, scRNA-seq identified 12 distinct cell clusters in the SAE of smokers (online suppl. Fig. S6). Two key genes, ATP5A1 and ATP5B, were found to be potential regulators of COPD airway remodeling, with ATP5B exhibiting the strongest association. Mechanistically, ATP5B promotes COPD airway remodeling by activating the TLR signaling pathway. This study leveraged the scRNA-seq dataset GSE123405 from the GEO database and applied bioinformatics approaches to predict and identify ATP5B as a key gene involved in SAE heterogeneity and airway remodeling in COPD. The findings demonstrate that ATP5B plays a crucial role in COPD airway remodeling, potentially serving as a key driver of disease progression. Furthermore, this study revealed that ATP5B silencing inhibits CSE-induced epithelial cell injury and EMT, suggesting that ATP5B may serve as a potential biomarker and therapeutic target for COPD.

Recent studies have characterized lung cell populations using scRNA-seq, particularly identifying the role of alveolar type 2 cells in COPD pathogenesis and genetic susceptibility. This research identified two alveolar type 2 cell subpopulations crucial in COPD pathobiology and genetic predisposition. The present study’s scientific significance lies in applying bioinformatics approaches to analyze scRNA-seq data, providing the first insights into the potential role of ATP5B in SAE cell heterogeneity and airway remodeling in COPD. Two key genes, ATP5A1 and ATP5B, were identified by comparing DEGs between normal and COPD samples. Further functional enrichment analysis and in vitro experiments demonstrated that ATP5B may promote COPD airway remodeling by activating the TLR signaling pathway. These findings highlight the critical role of ATP5B in COPD airway remodeling, providing new insights into disease pathogenesis and identifying potential biomarkers and therapeutic targets. The inhibition of ATP5B, which reduces CSE-induced epithelial injury and EMT, underscores its potential as a therapeutic candidate for COPD, offering promising implications for diagnosis, disease monitoring, and treatment strategies.

This study has important clinical implications for understanding the pathophysiology of COPD and identifying potential therapeutic strategies. By uncovering the critical role of ATP5B in COPD, this research provides a deeper understanding of the cellular and molecular changes underlying COPD progression. These findings enhance knowledge of COPD pathogenesis and offer new insights for identifying and developing therapeutic targets.

However, this study has several limitations. First, while scRNA-seq technology provides detailed and accurate cell-type information, there are still technical and methodological uncertainties in data analysis. Second, this study relied on a single dataset for prediction and validation, necessitating a larger sample size further to confirm the reliability and reproducibility of the results. Additionally, although in vitro and in vivo experiments provided preliminary evidence, further clinical studies are required to evaluate the potential of ATP5B as a therapeutic target.

Future research should focus on elucidating the precise role of ATP5B in COPD progression and investigating its interactions with other key molecules. Expanding clinical studies will be crucial to validate these findings further and explore ATP5B-related therapeutic strategies and drug development. Ultimately, these advances may contribute to preventing and treating COPD, offering new therapeutic targets and improving patient quality of life.

Statement of Ethics

All animal experiments were approved by the Ethics Committee of Affiliated Hospital of Chengdu University (No. PJ2020-055-01). Ethical approval and consent were not required for the human data used in this study, as all datasets were obtained from publicly available databases.

Conflict of Interest Statement

The author declares no conflict of interest.

Funding Sources

This study was supported by the Foundation of Sichuan Medical and Health Care Promotion Institute (KY2023QN0127), Research Project of Sichuan Medical Association (S2024035), Research Project of Chengdu Medical Association (2024061 and 2021034), Fund Project of Jinniu District Medical Association (JNKY2024-22), the Innovation Team Research Project of Affiliated Hospital of Chengdu University (CDFYCX202209), and Fund Project of Affiliated Hospital of Chengdu University (Y202407).

Author Contributions

Yabo Zhang and H.H. contributed equally to the study design, data acquisition, and manuscript drafting. W.S. and Y. Liu performed the bioinformatics analysis and contributed to data interpretation. Q.Z. and Y. Li conducted in vitro and in vivo experiments. C.L. assisted with histological analysis and data visualization. H.Z. and Yamei Zhang supervised the study, provided critical revisions, and secured funding. All authors read and approved the final manuscript.

Funding Statement

This study was supported by the Foundation of Sichuan Medical and Health Care Promotion Institute (KY2023QN0127), Research Project of Sichuan Medical Association (S2024035), Research Project of Chengdu Medical Association (2024061 and 2021034), Fund Project of Jinniu District Medical Association (JNKY2024-22), the Innovation Team Research Project of Affiliated Hospital of Chengdu University (CDFYCX202209), and Fund Project of Affiliated Hospital of Chengdu University (Y202407).

Data Availability Statement

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.