At the start of the experiment (day 0), male C57BL/6J mice aged 6–8 weeks received one intratracheal instillation of BLM at 2 mg/kg, prepared in 50 μL sterile saline. Mice were then monitored for 21 days under SPF housing conditions and euthanized on day 21 for tissue collection and subsequent analyses. After 21 days of BLM treatment, BLM exposure resulted in evident histopathological alterations, as shown by H&E staining, including inflammatory cell infiltration, alveolar enlargement, and septal thickening. Extensive collagen deposition was observed in the tissue sections when examined by Masson’s trichrome staining, primarily localized around bronchioles and within the alveolar interstitium. IHC (Immunohistochemistry) further confirmed elevated expression of Fn1(Fibronectin) (P < 0.001), Vim (P < 0.001), and α-SMA (P < 0.001), indicating enhanced fibrotic activity in BLM-treated lungs (Supplementary Figure 1a). The impact of BLM and normal saline on mouse body weight was monitored over time. In the BLM-treated group, weight gain ceased by day 14, followed by a more pronounced reduction in body weight observed on day 21, indicating progressive systemic effects associated with fibrosis induction (Supplementary Figure 1b). Hydroxyproline assays of lung homogenates demonstrated significantly higher collagen content in the BLM-treated group compared to controls, supporting increased extracellular matrix deposition (Supplementary Figure 1c). Western blot and qPCR analyses showed significant upregulation of fibrosis-related markers, including Fn1, N-cad, α-SMA, and Zeb1 (Zinc Finger E-box Binding Homeobox 1), in lungs from bleomycin-treated mice, relative to controls, at the gene and protein expression levels (P < 0.05) (Supplementary Figure 1d). Our data confirm that BLM-treatment reliably produced a model of pulmonary fibrosis, distinguished by extensive inflammatory infiltration and tissue injury. While ESRP1 is well-known for regulating FGFR2 alternative splicing, which we confirmed via qPCR (Supplementary Figure 1e). This alteration did not fully account for the observed fibrotic phenotype. The primers used for this study are summarized in Table 1. Further mechanistic exploration identified the Epac-Rap1a pathway as the predominant mediator through which ESRP1 exerts its pro-EMT effects in this specific context.

Furthermore, to explore the cellular heterogeneity in the BLM-induced IPF model, we performed sc-RNA seq of lung tissues collected on day 21 post-treatment from both BLM and PBS groups (Figure 1a). After quality control, we profiled 91,418 cells from total lung samples obtained from 3 PBS control mice and 3 BLM treatment mice, identified 10 discrete cell types based on distinct markers, cells were visualized using UMAP (Uniform Manifold Approximation and Projection) projection and grouped into 10 major cell populations based on canonical marker genes (Figure 1b). The identified clusters included: Monocyte_Macrophage, B cells, T/NK cells, Neutrophils, Dendritic Cells (DCs), Fibroblasts, Myofibroblast-like cells, Epithelial cells, Endothelial cells, and Basophils.

Top marker genes for each cell type were identified via differential expression analysis and visualized in violin plots (Figure 1c). For example, Col1a1 (Collagen Type I Alpha 1) and Acta2 (Actin Alpha 2, Smooth Muscle) in myofibroblasts, Epcam in epithelial cells, and Pecam1 in endothelial cells, confirming robust cell-type annotation.

Comparative analysis of cellular composition between BLM and PBS (Phosphate-Buffered Saline) groups revealed noticeable shifts (Figure 1d). BLM-treated lungs exhibited an increase in fibroblasts and myofibroblast-like cells, accompanied by a reduction in epithelial cells and endothelial cells, suggesting fibrotic remodeling and epithelial injury. Quantification of epithelial cell proportions (Figure 1e) showed a marked decrease in the BLM group compared to PBS, though the difference did not reach statistical significance (ns, P > 0.05).

These data demonstrate that bleomycin exposure induces pronounced cellular reorganization in the lung, with expansion of profibrotic mesenchymal populations and concomitant depletion of epithelial compartments, consistent with pathological hallmarks of IPF.

Next, we utilized GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analyses were employed to investigate the functional roles of genes highly expressed in epithelial cells. The GO functions of 30 TOP genes screened by single-cell sequencing showed significant changes in the functions of positive regulation of EMT and response to mchanical stimulus (Figure 2a). Meanwhile, based on functional annotations, it was found that the esrp1 was involved. The KEGG enrichment analysis demonstrated that epithelial cell–associated genes were distributed across the RAP (Ras-related Protein), apelin, and sphingolipid signaling pathways (Figure 2b). These results suggested that epithelial cells are critical population of EMT.

Subsequent genetic screening revealed specific expression of esrp1 in epithelial cells (Figure 2c), and significant elevation in BLM-induced mice compared to control group (Figure 2d). To further elucidate the function of ESRP1, GO and KEGG pathway analysis were conducted. The results showed that ESRP1 regulates RNA splicing and plays a key role in modulating the Rap1 signaling pathway and actin cytoskeleton, processes crucial for fibrosis progression (Figure 2e).

To further clarify the expression of ESRP1, the current study analyzed the expression changes of ESRP1 in the public database GSE218997. The results showed that ESRP1 was highly expressed in BLM-induced lung tissue of mice (Figure 3a). In addition, KEGG enrichment analysis indicated a significant association between the ESRP1 and Rap1a signaling pathways.

Small GTPases serve as multifunctional regulators involved in numerous cellular activities (29). As guanine nucleotide exchange factors (GEFs), Epac proteins mediate the activation of Rap small GTPases. Two isoforms are present in mammals: Epac1, expressed broadly across tissues, and Epac2 (RapGEF4 in humans), which shows tissue-restricted expression in the brain, kidney, and pancreas (30–32). Rap1a activation can facilitate EMT by modulating the expression of cell adhesion molecules and orchestrating cytoskeletal reconfiguration, thereby influencing cell shape and movement (33, 34). Pathological activation of the Epac–Rap1a axis in fibrotic tissues contributes to the amplification of EMT processes, subsequently enhancing extracellular matrix production and deposition, thereby aggravating the extent of fibrosis within lung tissue (35).

Hence, this study utilized western blot analysis together with qPCR to assess the expression patterns of the Epac–Rap1a signaling axis. The results validated the increased expression of ESRP1, Epac, and Rap1a among mice exposed to BLM when evaluated against the control group (Figure 3b). IHC was further conducted to confirm this result. As shown in Figure 3c, the proteins which indicated by brown dots were located within alveolar and bronchial epithelial cells and exhibited higher expression levels in BLM group mice compared to the control group (Figure 3c).

Furthermore, multiplex immunofluorescence revealed a progressive loss of the epithelial marker Epcam in BLM-exposed lungs compared with controls. Concomitantly, both ESRP1 and Rap1a signals were markedly elevated in fibrotic foci on day 21 (Figure 3d). These results demonstrate that ESRP1 contributes to the development in BLM-treated mice, particularly through its impact on epithelial cells.

To validate the expression of ESRP1 in patients with pulmonary fibrosis, lung tissues from both normal and fibrotic samples were collected. H&E staining revealed severe fibrosis in IPF patients compared to healthy controls, characterized by inflammatory cell infiltration and alveolar hemorrhage. Additionally, there was noticeable alveolar enlargement and thickening of the alveolar septa. Masson’s trichrome staining indicated substantial collagen deposition, visible as intense blue staining around the bronchial walls and alveolar septum in IPF patients. Furthermore, IHC analysis demonstrated that protein levels of ESRP1, Epac, and Rap1a were elevated in IPF patients (Figure 4a). IF analysis further corroborated the increased ESRP1 and Rap1a expression in IPF tissues, alongside a reduction in the epithelial cell marker Epcam (Epithelial Cell Adhesion Molecule). Additionally, no significant colocalization was observed between Epacm and ESRP1, or between Epcam and Rap1a (Figure 4b). In healthy lung tissue, alveolar structures remained intact and exhibited only minimal ESRP1 expression, primarily localized to alveolar and bronchial cells. In contrast, lung tissue from IPF displayed disrupted alveolar architecture, proliferation of interstitial fibrous tissue, and elevated Esrp1 expression in bronchial epithelial cells as well as compressed alveolar surfaces. Collectively, these findings indicate that ESRP1 is specifically enriched in epithelial cells residing in the lungs of patients suffering from IPF.

In vitro studies, TGF-β1 serves as a central mediator initiating EMT. Therefore, protein expression analysis was conducted to evaluate EMT markers in 10 ng/mL TGF-β1 stimulated mouse lung epithelial cells (MLE-12). Western blot revealed that TGF-β1 markedly up-regulated ESRP1 protein and simultaneously triggered an EMT signature: E-cad was substantially down-regulated, whereas N-cad expression was significantly elevated (Figure 5a). We also noted enhanced protein expression of Epac and Rap1a in TGF-β1-treated MLE-12 cells compared to the control group (Figure 5a).

Cellular IF (Immunofluorescence) assays further confirmed TGF-β1 treatment increased the expression of Rap1a in the cytoplasm (Figure 5b). KEGG pathway enrichment predicted a potential association between ESRP1 and the Rap1 signaling cascade. To assess the interaction between ESRP1 and Epac, Co-IP was performed on lysates from MLE-12 cells treated with or without TGF-β1. The results showed that anti-ESRP1 antibody co-precipitated Epac from cell lysates. In contrast, normal IgG control antibody failed to precipitate either protein, confirming the specificity of the interaction. Quantitative analyses revealed that TGF-β1 treatment enhanced the interaction between ESRP1 and Epac (36, 37) (mean ± SEM, n = 3, ***P < 0.001; Figure 5c). Thus, TGF-β1 stimulation significantly enhanced the Co-IP enrichment of both Epac and ESRP1, indicating a ligand-induced increase in their mutual association. Moreover, TGF-β1 treatment markedly enhanced this interaction, as quantified by densitometric analysis of precipitated proteins.

Furthermore, MLE-12 cells were transduced with lentiviral vectors to silence or overexpress esrp1, and its expression were quantified by Western Blot and qPCR. Esrp1 knockdown markedly reduced the expression of N-cad, whereas E-cad was significantly increased. Conversely, esrp1 overexpression elevated N-cad transcripts and concurrently suppressed E-cad (Figures 5d,e). These reciprocal changes indicate that ESRP1 transcriptionally regulates EMT.

To elucidate ESRP1 function, lentivirus-mediated gene perturbation was performed in MLE-12 cells. Compared with scrambled controls, esrp1 silencing markedly attenuated the expression of Epac, Rap1a and N-cad, while reciprocally enhancing E-cad abundance. Conversely, overexpression esrp1 elevated Epac, Rap1a and N-cad, accompanied by a concomitant reduction in E-cad (Figure 6a). Collectively, these data demonstrate that ESRP1 positively modulates Epac and Rap1a expression and actively drives EMT.

To clarify the regulatory hierarchy among ESRP1, Epac, and Rap1a, MLE-12 cells were transfected with siRNAs targeting epac and rap1a. Knockdown of epac markedly reduced Rap1a and N-cad levels while increasing E-cad (Figure 6b); conversely, silencing rap1a decreased N-cad and elevated E-cad without affecting Epac abundance (Figure 6c). These results corroborate that the Epac–Rap1a axis promotes EMT. Critically, neither Epac nor Rap1a depletion altered ESRP1 expression, whereas epac knockdown Rap1a expression. Collectively, the data establish that ESRP1 operates upstream of Epac and Rap1a, with Epac positioned upstream of Rap1a in this signaling cascade. These findings indicated that ESRP1 facilitated TGF-β1-induced EMT by activating Epac-Rap1a signaling in vitro.

Upon TGF-β1 stimulation, the protein levels of Rap1a, Epac, ESRP1, and N-cadherin were markedly increased compared with the CON group, whereas E-cadherin was progressively reduced. Notably, ESRP1 overexpression further amplified the TGF-β1–induced changes, as evidenced by higher Rap1a/Epac/ESRP1/N-cadherin expression and a more pronounced decrease in E-cadherin in the oe-Esrp1 + TGF-β1 group. In contrast, in the si-Epac + TGF-β1 group, Rap1a, Epac, ESRP1, and N-cadherin levels were reduced and E-cadherin was elevated relative to the oe-Esrp1 + TGF-β1 group. Importantly, no significant differences were observed in EMT-associated markers or the target proteins between si-Epac + TGF-β1 and si-Epac + oe-Esrp1 + TGF-β1 groups, indicating that Epac knockdown prevented ESRP1 overexpression induce EMT marker alterations under TGF-β1 stimulation (Figure 6d).

Written informed consent was secured from every participant prior to enrollment, and the study protocol was reviewed and approved by the Institutional Review Board at the General Hospital of Ningxia Medical University (Approval No. KYLL-2023-0516) and conducted in accordance with NIH guidelines. Patients diagnosed with lung cancer and lung fibrosis, as identified by CT scans, were recruited from Ningxia, Yinchuan. Samples were collected from areas ≥ 5 cm away from the tumor as healthy controls. The tissue collection occurred between January 2024 and October 2024. IPF diagnosis was based on clinical history, physical exams, pulmonary function tests, HRCT (High-Resolution Computed Tomography), and confirmed by lung biopsy, in alignment with the recommendations issued jointly by the American Thoracic Society (ATS) and the European Respiratory Society (ERS) (3). Healthy Control (HC) samples were obtained from lung cancer patients with histologically normal lung tissue. Additionally, lungs from healthy organ donors and IPF transplant donors were used. All tissue usage was approved by the relevant institutional review boards.

Unless otherwise specified, all reagents were purchased from sigma-Aldrich. BLM was sourced from Solarbio (Cat:IB0871; Beijing, China) for use in the experiments, while recombinant cytokine TGF-β1 was supplied by R&D Systems (Hong Kong, China). The primary antibody against ESRP1 (rabbit polyclonal, Cat No. 21045-1-AP), Epac (rabbit polyclonal, Cat No. 12572-1-AP), Rap1a (mouse monoclonal, Cat No. 68125-1-Ig), N-cadherin (rabbit polyclonal, Cat No. 22018-1-AP) and β-actin (rabbit polyclonal, Cat No. 20536-1-AP) were obtained from Proteintech (Wuhan, China). E-cadherin (rabbit polyclonal, Cat No.#AF0131) was purchased from Affinity (Jiangsu, China). Multiplex immunofluorescence staining was performed according to the manufacturer’s instructions using the Servicebio multiplex immunofluorescence kit (Servicebio, Wuhan, China). The istological and immunohistochemical reagents used in this study were obtained from Zhong-shan-Jin-qiao Biotechnology (Beijing, China). Quantitative real-time PCR kits were sourced from Vazyme (Nanjing, China), and siRNA along with other chemical reagents from Sangon Biotech (Shanghai, China).

The conduct of animal experiments followed the directives and approval of the Ethics Committee for Animal Research (Approval Number: IACUC-NYLAC-2023-252). Male C57BL/6J mice (22 ± 2 g, aged 6–8 weeks) were acquired from the Animal Centre of Ningxia Medical University and used to establish the IPF model. The mice were maintained in a specific pathogen-free (SPF) facility, where they had continuous access to potable water and a nutritionally balanced standard chow diet provided ad libitum. After intraperitoneal anesthesia with 0.1 % pentobarbital sodium (50 mg kg⁻¹), IPF was established through a single intratracheal administration of bleomycin sulfate at a dose of 2 mg/kg dissolved in 50 μL of sterile 0.9% saline. Control mice received an equal volume of vehicle alone. Animals were monitored daily for body weight and general condition. On day 21 post-instillation, mice were euthanized by exsanguination under deep anesthesia, after which lungs were harvested for histopathological examination and biochemical analyses.

Lung tissues from mice in the PBS and BLM groups were dissociated into single-cell suspensions and loaded onto a Chromium Single Cell Controller (10 × Genomics, United States) for single-cell library construction and sequencing. Raw base call files were processed with Cell Ranger (v6.0) to perform demultiplexing, alignment, and UMI counting, and reads were mapped to the mouse reference genome (mm10) to generate the gene–cell count matrix. Downstream analyses were conducted in R using the Seurat workflow. Cells were retained after quality control based on standard metrics (including the number of detected genes/UMIs and the proportion of mitochondrial transcripts), and low-quality cells were removed. The filtered data were normalized and scaled, highly variable genes were identified, and principal component analysis (PCA) was performed. Graph-based clustering was conducted using a shared nearest neighbor (SNN) graph built on the top 10 principal components, and visualization was generated with Uniform Manifold Approximation and Projection (UMAP; RunUMAP). Differentially expressed genes (DEGs) for each cluster (or between groups when applicable) were identified using a Wilcoxon rank-sum test, and multiple testing correction was applied to control the false discovery rate (FDR). Functional enrichment analyses were performed for GO terms and KEGG pathways using curated databases, and significantly enriched terms/pathways were defined based on adjusted P-values.

According to the manufacturer’s guidelines, the MLE-12 were acquired from the Shanghai iCell Bioscience Institute (iCell-m036) and cultured in DMEM/F12 basal medium with added supplements (iCell-0033a). Cells were maintained in a humidified incubator under sterile conditions at 37°C with 5% CO2 until reaching 80–90% confluence. After seeding in 6-well plates overnight, the cells were stimulated with TGF-β₁ (10 ng/mL) for 24 h to construct EMT cell model and then collected for qPCR and western blot.

MLE-12 cells were plated into 6-well plates then infected with the designated lentiviral constructs plus 2 μg/ml polybrene at 70% confluence overnight, followed by 2 μg/mL puromycin selection for 3 days. siRNA duplexes targeting gene-epac/rap1a were transfected at 50 nM using Lipofectamine RNAiMAX according to the manufacturer’s instructions. After 48 h, qPCR and western blot assays were employed to verify knockdown efficiency. The sequence of siRNA used in present study was attached to Table 2. The sequences of the siRNAs (epac and raap1a) used in the present study are listed in Table 3.

Left lungs were fixed overnight in 4% paraformaldehyde, embedded in paraffin, and cut into 4 μm sections. After deparaffinization and rehydration, the slides were stained with hematoxylin and eosin (H&E) to examine tissue architecture and with Masson’s trichrome to evaluate collagen distribution, following the manufacturer’s recommendations. Additional sections were subjected to immunohistochemical analysis.

Cells were first dewaxing and rehydration, followed by antigen retrieval. Endogenous peroxidase activity was removed and incubated with 5% BSA. 1 h later, the sections were stained overnight at 4°C with rabbit anti-Fn (Abcam; Cat: ab2413; Dilution: 1:400), rabbit anti-Vim (Abcam; Cat: ab92547; Dilution: 1:600), rabbit anti-α-SMA (CST; Cat:19245S; Dilution: 1:200), and rabbit anti-ESRP1 (Proteintech;Cat No. 21045-1-AP; Dilution: 1:100), Epac (Proteintech; Cat No. 12572-1-AP; Dilution: 1:200), Rap1a (Proteintech; Cat No. 68125-1-Ig; Dilution: 1:50) and subsequently incubated with the streptavidin-biotin complex (ABC) detection kit (BOSTER, China) for 30 min at room temperature. Each section underwent color development using peroxidase-linked avidin/biotin and 3’-3-diaminobenzidine (DAB) substrate from ZSGB-BIO, Cat# PV-6000D, followed by haematoxylin counterstaining.

Paraffin-embedded sections were dewaxed, rehydrated, and subjected to antigen retrieval using 10 mM sodium citrate buffer (pH 6.0) at 95°C for 15 min. Endogenous peroxidase activity was neutralized with 3% hydrogen peroxide for 10 min, and nonspecific binding was blocked with 5% normal donkey serum for 1 h at room temperature. Slides were then incubated overnight at 4°C with the respective primary antibodies (ESRP1 1:200; Rap1a 1:200) in a humidified chamber. After three 5-min PBS washes, slides were treated with HRP-conjugated secondary antibody for 1 h and developed with iF488-TSA (10 min, RT, dark). A second round of heat-mediated antigen retrieval was performed, followed by incubation with anti-EpCAM (1:100) and the corresponding secondary antibody and iF555-TSA using the same washing regime. Nuclei were counterstained with DAPI (4’,6-Diamidino-2-Phenylindole, Dihydrochloride) for 10 min, autofluorescence quenched for 5 min and sections rinsed under running water for 10 min. Slides were mounted in Prolong Diamond and imaged with a Leica DMi8 confocal microscope.

The concentration of hydroxyproline in the lungs was measured with a commercial ELISA (Enzyme-Linked Immunosorbent Assay) kit (A030-2-1) from Nanjing Jiancheng Bioengineering Institute, following the manufacturer’s guidelines.

Whole proteins were isolated using the Whole Cell Lysis Assay kit (KeyGEN5303, Nanjing, China). Cytoplasmic and nuclear proteins were isolated using the Nuclear and Cytoplasmic Protein Extraction Kit (KGB5302, Nanjing, China) according to the manufacturer’s instructions. Cells were collected and incubated with a cytoplasmic extraction buffer on ice, followed by centrifugation to obtain the cytoplasmic supernatant. The remaining pellets were then resuspended in a nuclear extraction buffer to recover nuclear proteins. Histone-3 and β-actin were employed as internal controls for the nuclear and cytoplasmic fractions, respectively. Equal protein quantities from each sample were separated on a 10% gel, transferred to PVDF membranes, and probed with primary antibodies at 4°C overnight followed by appropriate secondary antibodies (Proteintech, SA00001-2). The protein bands were identified and measured using Superstar ECL (KeyGEN, Nanjing, China) with the Gel Imaging System (Amersham Imagequant800uv, United States). The results were quantitative analysis by using Image J. The antibodies utilized in this research include: rabbit anti-N-cad (1:1,000), rabbit anti-E-cad (1:1,000), mouse anti-Rap1a (1:1,000), rabbit anti-Epac (1:500), and rabbit anti-Esrp1 (1:1,000) and anti-β-actin (1:2,500) polyclonal antibodies were purchased from Proteintech (Wuhan, Hubei, China). The relative expression of target proteins was determined by comparing the gray level ratio of the target bands to β-actin, which served as an internal control.

RNA extraction was carried out with TRIzol reagent (Ambion, United States). A total of 1 μg RNA was then subjected to reverse transcription using a cDNA synthesis kit, following the manufacturer’s instructions (Vazyme Biotech, Nanjing, China). quantitative PCR was performed on a Bio-Rad IQ5 system (Bio-Rad Inc., United States) using HiScript Q RT SuperMix for qPCR (Vazyme, Nanjing, China). Expression levels were normalized against β-actin as the housekeeping gene, and fold changes were computed by applying the 2^–△△CT algorithm. Details of the primer sequences designed for the amplification of mouse genes are presented in Supplementary Table 1.

Cells at 80% confluence were washed twice with chilled PBS and lysed for 30 min on ice using RIPA buffer supplemented with protease inhibitors. Lysates were centrifuged at 12,000 rpm for 15 min at 4°C, and the supernatant was designated as the total protein extract. A minor aliquot of the lysate was retained as the input control. Co-IP was performed with the Beyotime Co-IP kit (P2179S) according to the manufacturer’s instructions.

Briefly, the total protein extract was incubated with primary antibodies against ESRP1 (rabbit polyclonal, ProteinTech, Cat No. 21045-1-AP) or Epac (rabbit polyclonal, ProteinTech, Cat No. 12572-1-AP) at 4°C overnight with gentle rotation. For the negative control, an equal amount of normal rabbit IgG (ProteinTech, Cat No. B900610) was used instead of the specific primary antibody. Subsequently, protein A/G agarose beads (included in the Beyotime Co-IP kit) were added to the mixture and incubated for another 4 h at 4°C. After incubation, the beads were washed five times with ice-cold IP wash buffer (provided in the kit) to remove non-specific bindings, and the immunoprecipitated complexes were eluted with elution buffer (included in the kit) by boiling at 95°C for 10 min. The eluted samples, along with the input control, were subjected to Western blot analysis as described below.

All data in this study were derived from a minimum of three independent experiments for each condition. Statistical analysis was performed using GraphPad Prism 9.5 (GraphPad Software Inc., United States). Statistical analyses included one-way ANOVA for comparisons among multiple groups with a common control and Student’s t-tests for pairwise analyses. Data are expressed as mean ± SD. A probability value of < 0.05 was regarded as statistically significant, with symbols denoting significance as follows: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.