Licoricesaponin G2 (CAS: 118441-84-2, purity >98%) was sourced from Sichuan Weikeqi Biotechnology Co., Ltd., Sichuan, China. Bleomycin (BLM, HY17565), pirfenidone (HYB0673), TGF-β1 (HY-151427) and Etanercept (HY-108847) were acquired from MedChemExpress, Princeton, United States. RPMI 1640 medium and fetal bovine serum were procured from Gibco, United States, and Clark Bioscience, Claymont, United States, respectively. Minimum Essential Medium was obtained from Procell, Wuhan, China. Paraformaldehyde (4%) and penicillin-streptomycin were supplied by Biosharp, Hefei, China. MTT and Triton X-100 were purchased from Solibol, Beijing, China, while DAPI was sourced from Solibo, Beijing, China. The Magnetic Cell Total RNA Kit and iScript cDNA synthesis kit were procured from Tiangen Biochemical Technology Co., Ltd., Beijing, China. RIPA buffer was obtained from Beyotime Biotechnology, Shanghai, China. PVDF membranes were sourced from Roche, Basel, Switzerland. The Transwell apparatus (8-μm pore size) was purchased from Corning Costar, Cambridge, United States. Antibodies against E-cadherin (ab40772), N-cadherin (ab76011), and GAPDH (ab8245) were acquired from Abcam, Cambridge, United Kingdom. Antibodies against α-SMA (#AF1032) were sourced from Affinity Biosciences, Jiangsu, China. Antibodies against Collagen I (14695-1-AP), TNF-α (CL488-60291), MMP3 (66338-1-Ig), MMP9 (10375-2-AP), β-actin (66009-1-Ig-AP), and goat anti-rabbit/mouse antibodies (SA00001-1, SA00001-2) were obtained from Proteintech, Wuhan, China. p-IKK alpha/beta (BZ16321), NFκB (AP0076), and p-NFκB (BS66162) were purchased from Bioworld, Nanjing, China.

Potential targets for LG2 were identified using several databases, including Drugbank, Swisstarget, Prediction, ZINC, and PharmMapper. After consolidating the data from these sources, duplicate values were eliminated to derive the unique potential targets of LG2.

Using “Pulmonary Fibrosis” as the search keyword, target genes associated with PF were extracted from three reputable databases: GeneCards (https://www.genecards.org/), Online Mendelian Inheritance in Man (OMIM) (https://omim.org/), and PharmGKB (https://www.pharmgkb.org/). These databases are widely recognized as reliable sources for obtaining disease-related targets.

A Venn diagram was utilized to illustrate the overlapping regions among sets of elements. Initially, the principal target genes associated with both PF and LG2 were input into Venny 2.1 (http://bioinfogp.cnb.csic.es/tools/venny/). Subsequently, the intersecting targets of LG2 and PF were identified using the same tool. This analysis ultimately identified the potential target genes of LG2 relevant to the treatment of PF.

A protein-protein interaction (PPI) network was established using the STRING database (https://string-db.org/) (Szklarczyk et al., 2019). Initially, the overlapping targets of LG2 and PF were inputted into the STRING database to create a preliminary PPI network diagram. Subsequently, the species was set to “Homo sapiens,” and isolated proteins were excluded. An enhanced PPI network diagram was then developed using Cytoscape software (version 3.7.2). Within the Cytoscape software, the “Network Analyzer” feature was utilized to conduct an analysis. Key metrics, such as betweenness centrality (BC), closeness centrality (CC), and degree centrality (DC), were selected as screening indices. These degrees were sorted in descending order. Following a comprehensive review of the literature, pivotal targets were identified.

To explore the signaling pathways and potential biological processes involved in LG2 treatment of PF, we entered the intersecting targets of LG2 and PF into the DAVID database (https://david.ncifcrf.gov/) (Dennis et al., 2003). Within the “Functional Annotation” section, we selected CC, molecular functions (MF), and biological processes (BP) under Gene Ontology (GO), along with the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The results were downloaded and saved. Then, these results were filtered using Excel, retaining only those with p-values <0.05 and counts ≥5. Using the bioinformatics platform, the top 30 KEGG pathway enrichment results were presented in a bubble diagram and the top 15 GO enrichment results in a bar chart, based on gene number and p-value. Following a comprehensive literature review, a specific pathway was selected for further investigation.

The Protein Data Bank (PDB) (Protein Data Bank, 2019) is the premier database for the three-dimensional structural data of proteins, nucleic acids, and other biological macromolecules. These structures are primarily determined through experimental techniques such as X-ray crystallography, nuclear magnetic resonance (NMR), and electron microscopy. In our study, we selected two targets as receptors and LG2 for molecular docking validation. Initially, the crystal structure of the target protein was retrieved from the PDB database. Any bound ligand structures were removed, and the ligand binding site or the active site, as identified in the literature, was designated as the binding pocket for the small molecule ligand. Subsequently, after preprocessing the target protein using the PrepWiz module of Schrodinger software, a grid file was generated. Concurrently, details such as the Mol ID number, CAS number, names in both Chinese and English, and the chemical 2D structure of LG2 were obtained from the Traditional Chinese Medicine Systems Pharmacology (TCMSP) database and saved in the mol.2 format. LG2 was processed using the LigPrep module of Schrodinger software. For the final step, the prepared ligand-receptor file underwent molecular docking using the Standard Precision method in Schrodinger’s Glide module. The ionization state was assigned using Epik28 under pH conditions of 7.0 ± 2.0, followed by docking calculations. A default binding score of less than −5.0 kcal/mol indicates favorable binding activity between the ligand and receptor, while a score below −7.0 kcal/mol suggests particularly strong binding affinity (Lipinski et al., 2001).

The experimental protocol involving animals was approved by the Experimental Animal Ethics Committee of Changchun University of Chinese Medicine (Approval No. 2023608). We acquired 8-week-old male C57/BL6 mice from Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China. These mice were maintained under optimal conditions in our animal experiment center at 20°C ± 2°C with 50%–60% humidity. Currently, transtracheal administration is recognized as the more mature method. High-quality research has established typical dosages for BLM-induced PF in mice are 3, 5, and 10 mg/kg (Lee et al., 2020). Based on this, our team explored the pre-experimental conditions using different dosages. We found that although 5 mg/kg and 10 mg/kg could induce PF as well, the modeling mice were in poorer health and higher mortality rates. Therefore, we ultimately chosed 3 mg/kg (Ding et al., 2023) as the optimal dosage to reliably model the onset of PF (the success rate of animal modeling is over 93%). Pirfenidone was selected as the positive control. The mice were randomly assigned to four groups: sham operation, bleomycin (BLM), LG2 (oral gavage: 50 mg/kg), and pirfenidone (PFD, oral gavage: 200 mg/kg), with each group consisting of 10 mice. All mice received an intraperitoneal injection of pentobarbital sodium (30 mg/kg) for anesthesia. The sham operation group was administered an intra-tracheal injection of 0.9% saline (NS), while the other groups received BLM (3 mg/kg). The mice underwent daily oral administration of LG2, with the sham and model groups receiving pure water for 2 weeks, followed by euthanasia and tissue collection for further experimental analysis. The layout and design of the animal experiment are depicted in Figure 5A.

A549 cells (human lung carcinoma) and MRC5 cells (human fetal lung fibroblasts) were obtained from the Cell Bank of Cell Biology at the Chinese Academy (Shanghai, China) and Fu Heng Biology (Shanghai, China), respectively. A549 cells were cultured in RPMI 1640 medium, while MRC5 cells were maintained in Minimum Essential Medium. Both media were supplemented with 10% fetal bovine serum (FBS) and 1% antibiotics (penicillin and streptomycin). The cells were divided into five groups: a control group, a model group treated with TGF-β1 at 5 ng/mL, and three groups exposed to varying concentrations of TGF-β1. To induce EMT, TGF-β1 was added to the cultures (except the control group) upon reaching 60%–70% confluence, and the co-culture was continued for 48 h. All cell lines were incubated at 37°C in a 5% CO₂ atmosphere.

The lung lobes were washed with chilled saline before fixation in 4% paraformaldehyde. The tissues were then embedded in paraffin and sectioned into 3 μm slices, which were subsequently stained with H&E and Masson’s trichrome for detailed histological analysis. The stained sections were meticulously examined using a digital microscope and slide scanner (M8, PreciPoint, Thuringia, Germany). Pathological scoring was conducted based on methodologies described in prior studies (Katoh et al., 1998; Pedroza et al., 2016). Concurrently, to assess the expression levels of proteins related to EMT, ECM, and IHC staining was performed to detect biomarkers including α-SMA, Collagen I, TNF-α, and MMP9. Following dewaxing and antigen retrieval, the tissues were incubated overnight at 4°C with primary antibodies, followed by a 1-h incubation with HRP-labeled secondary antibodies. DAB solution was applied to develop brown staining, and a final counterstain with hematoxylin was performed. The M8 was used to select appropriate fields of view for quantitative analysis of the pathological findings using ImageJ software.

Dewaxed lung tissue sections and treated A549 cells were incubated overnight with primary antibodies, including α-SMA, Collagen I, TNF-α, and MMP9. This was followed by a 1 h incubation with corresponding secondary antibodies at room temperature. Nuclei were stained using DAPI for 15 min. Finally, relevant fields of view were photographed using a fluorescence microscope (Leica DMi8, Germany), and fluorescence intensity was quantified using ImageJ software.

Proteins were extracted from lung tissues and A549 and MRC5 cells using pre-cooled RIPA lysis buffer. The protein concentration was determined using a BCA protein assay. The proteins of interest were then separated by SDS-PAGE and transferred onto PVDF membranes. After blocking for 1 h, the membranes were incubated overnight at 4°C with primary antibodies against E-cadherin, N-cadherin, α-SMA, Collagen I, TNF-α, p-IKK, p-NFκB, NFκB, MMP3, MMP9, β-actin, and GAPDH. Following washing with 1× TBST, the membranes were treated with secondary antibodies for 1 h at room temperature. Protein bands were visualized using a chemiluminescent imaging system (ChemiDoc XRS+, Bio-Rad), and quantification was performed using ImageJ software.

A549 and MRC5 cells were seeded in a 96-well plate at a density of 5,000 cells per well. These cells were treated with various concentrations of LG2, ranging from 0.2 to 50 μmol/L, for 48 h to evaluate the metabolite’s effect on cell viability. Additionally, to assess the impact of LG2 on TGF-β1-treated cells (5 ng/mL), A549 and MRC5 cells were divided into control, model, and treatment groups. After 48 h of incubation, MTT assay was conducted by adding 0.5 mg/mL MTT to each well and incubating for 4 h at 37°C in the dark. Subsequently, 150 μL of DMSO was added to dissolve the formazan crystals. Absorbance was measured at a wavelength of 490 nm using a microplate reader (American Bertten, SynergyMx).

Total RNA was extracted from A549 and MRC5 cells using a total RNA extraction kit. Subsequently, 1 μg of total RNA was reverse transcribed into cDNA utilizing an iScript cDNA synthesis kit. qPCR analysis was conducted on the Bio-Rad CFX96 system (Bio-Rad, Synergy Mx). Data normalization was performed using the 2^−ΔΔCt method with GAPDH as the reference gene (Yang et al., 2021). The primer sequences used are detailed in Table 1.

The migration of A549 cells was assessed using Transwell Boyden chambers equipped with polycarbonate filters containing 8 μm pores. Cells were seeded into the upper chamber pre-coated with Matrigel. These chambers were then placed in 24-well plates and incubated for 24 h in a serum-free medium, treated either with TGF-β1 alone or in combination with LG2. After the incubation, the medium in the upper chamber was removed. The cells were fixed with 4% paraformaldehyde for 1 h. Non-migrated cells on the filter side of the upper chamber were removed using a cotton swab. Following fixation, cells that had migrated to the bottom surface of the membrane were washed three times with PBS and stained with 5% Cresyl violet for 15 min. The migrated cells were then photographed under a microscope for analysis.

The experimental results were expressed as the mean ± standard deviation, based on at least three independent experiments. Data analysis was performed using GraphPad Prism version 9.0. To assess statistical significance among multiple groups, one-way ANOVA followed by Tukey’s post hoc test was utilized. In all statistical analyses, a p-value of less than 0.05 was considered to indicate statistical significance.

After screening, there are 175 potential targets of LG2 were predicted by Durgbank, Swisstarget, Prediction, ZINC and PharmMapper.

The keywords “Pulmonary fibrosis” were used in GeneCards, OMIM, and PharmGKB databases for PF disease targets were searched and integrated to obtain PF-related targets after removing duplicate targets. The three databases of GeneCards, OMIM, and PharmGKB were searched to obtain 6,119 PF disease targets.

Venn analysis of the 6,119 disease targets and 175 drug targets revealed 127 potential targets for LG2 in treating PF, as depicted in Figure 3A.

Utilizing the STRING database’s multiple proteins tool, we uploaded the intersecting targets while specifying H. sapiens as the species. We set the confidence level at greater than 0.9 and excluded disconnected nodes to generate the PPI network. The data, in TSV format, was downloaded and imported into Cytoscape 3.7.2 to visualize a refined PPI network, as illustrated in Figure 3B. This network comprised 118 nodes and 797 edges. Node size and color intensity were determined based on degree values, with larger and darker nodes indicating higher degrees. Edge color intensity was set based on the relationship value score between target points, with darker edges indicating higher scores. Topological analysis was conducted using the Network Analyzer tool in Cytoscape. Median values for BC, CC, and degree centrality (DC) were calculated as 0.00326225, 0.439025935, and 10, respectively. Targets meeting the median for all three metrics were filtered and ranked by degree value in descending order. The top 10 targets, based on a composite ranking of BC, CC, and DC, were identified as VEGFA, SRC, PTGS2, JUN, MMP9, CASP3, STAT3, IL2, FGF2, and PTPRC.

The GO enrichment analysis revealed that among 421 biological process pathways, 111 (with p < 0.05 and count ≥5) were predominantly associated with responses such as “response to drug,” “response to xenobiotic stimulus,” “response to lipopolysaccharide,” as well as the “positive regulation of the ERK1 and ERK2 cascade” and the “positive regulation of the MAPK cascade.” In the category of cellular components, 38 out of 62 pathways (with p < 0.05 and count ≥5) were primarily linked to “membrane raft,” “integral component of plasma membrane,” “plasma membrane,” “caveola,” and “endoplasmic reticulum.” Regarding molecular functions, 42 out of 99 pathways (with p < 0.05 and count ≥5) were chiefly related to “steroid binding,” “heme binding,” and “enzyme binding.” These findings are illustrated in Figure 3C. The KEGG pathway analysis indicated that of the 106 pathways associated with PF, 80 (with p < 0.05 and count ≥5) primarily pertained to “Neuroactive ligand-receptor interaction,” “TNF-α signaling pathway,” “MAPK signaling pathway,” and “EGFR tyrosine kinase inhibitor resistance,” as depicted in Figure 3D. It is well-established that inflammation plays a pivotal role in the progression of PF. Exposure to harmful stimuli can trigger tissue apoptosis and necrosis through inflammatory infiltration. The aggregation of fibroblasts may lead to significant ECM deposition, increased fibrous tissue, and ultimately restricted respiratory function in lung tissues, progressing to respiratory failure. Tumor necrosis factor (TNF-α) receptors, ubiquitously expressed across cells and tissues, have broad and multifaceted roles in inflammatory responses and can either induce or exacerbate tissue fibrosis. Furthermore, TNF-α can diminish the activity of matrix metalloproteinases (MMPs), undermining ECM decomposition by MMPs and exacerbating ECM accumulation. Through network analysis, the TNF-α signaling pathway was identified as shown in Figure 3E. This aligns with the inflammatory perspective, prompting an investigation into whether LG2 could mitigate the onset of EMT/fibroblast myofibroblast transition (FMT), enhance ECM degradation, reduce ECM deposition, and ultimately alleviate PF by suppressing the expression of the TNF-α pathway.

Through network analysis and literature review, we identified two significant targets: TNF-α and MMP9. Specifically, LG2 forms hydrogen bonds with MMP9 at two amino acids, ALA-191 and ASP-235, both located near the active site. Similarly, LG2 establishes hydrogen bonds with TNF-α at three amino acids: GLU-109, SER-74, and ASP-93, all near the active site. These interactions are depicted in Figure 4. Docking studies of these targets with LG2 indicated that both TNF-α and MMP9 demonstrate strong binding affinity for LG2, as detailed in Table 2.

H&E staining revealed significant alterations in the lung tissues of mice induced by BLM alone, as shown in Figure 5B. Compared to the sham-operated group, lung tissues in the BLM group exhibited pronounced pathological changes including extensive connective tissue proliferation (indicated by black arrows), blurred alveolar structures, and widespread infiltration of lymphocytes and neutrophils (yellow arrows). Additionally, notable perivascular infiltration of inflammatory cells forming vascular sleeves was observed (red arrows), alongside cytoplasmic vacuolation in some bronchial epithelial cells (blue arrows). Following treatment with LG2 or pirfenidone (PFD), marked improvements in inflammation and fibrosis were observed in the lung tissues. Masson’s trichrome staining, shown in Figure 5C, further highlighted these changes; the blue-stained areas, indicative of collagen, pointed to the presence of fibrotic tissue. Compared to the sham group, there was a substantial increase in blue collagen fiber deposition in the lung tissues of the BLM group. However, this deposition was notably reduced after treatment, suggesting that LG2 effectively inhibits collagen fiber deposition in BLM-induced lung tissue in mice, as demonstrated in the quantitative analysis of Masson’s trichrome staining (Figure 5D). To further understand the molecular impact of LG2 on PF, we conducted western blot analyses. Compared to the sham group, the BLM group exhibited reduced E-cadherin protein expression and increased levels of N-cadherin, α-SMA, and Collagen I. Post-treatment, there was an upregulation of E-cadherin and a decrease in the other markers (Figures 5E–I). These findings suggest that LG2 treatment significantly alleviates the pathological damage and pulmonary fibrosis in the lung tissues of mice induced by BLM.

Our study investigated the roles of EMT and ECM in the development of PF, specifically examining the regulatory effects of LG2 on BLM -induced PF in mice. We employed IHC and IF to assess the levels of α-SMA and Collagen I proteins in lung tissues. The results demonstrated that in BLM-induced mice, there was an increase in the expression of α-SMA and Collagen I. However, following treatment with LG2 or PFD, the levels of both proteins significantly decreased, as shown in Figures 6A–H.

TGF-β1 exerts a potent fibrotic effect and is known to induce EMT (Phan et al., 2021) and FMT (Zhang et al., 2021). These processes contribute to the formation of cells with mesenchymal characteristics, such as fibroblasts and myofibroblasts, which are central to the secretion of ECM and the progression of fibrosis (Sun et al., 2013). This rationale underpinned our choice of A549 and MRC5 cells for in vitro experiments to further explore the anti-PF mechanism of LG2 in inhibiting TGF-β1-induced transitions. Initially, we assessed the impact of varying concentrations of LG2 (0.2–50 μmol/L) on the viability of these cell types. Notably, compared to the control, A549 cells treated with concentrations of 3.1 μmol/L or higher for 48 h exhibited a significant reduction in viability (Figure 7A). Similarly, MRC5 cell viability showed a marked decrease at concentrations of 25 μmol/L under the same experimental conditions (Figure 7B). Additionally, the viability of A549 and MRC5 cells was assessed under treatment with TGF-β1 to simulate a fibrotic environment (Figures 7C, D). The therapeutic dosages used in subsequent experiments were determined based on these cytotoxicity assessments.

We determined LG2 dosages using the MTT assay. Dosages of 0.8, 1.6, and 3.1 μmol/L were administered to A549 cells and 6.3, 12.5, and 25 μmol/L to MRC5 cells. After an incubation period of 48 h, qPCR analysis showed a decrease in mRNA levels of E-cadherin and an increase in N-cadherin, α-SMA, and collagen I in TGF-β1-stimulated A549 and MRC5 cells (Figures 8A–D, J, K). Subsequent to LG2 treatment, an increase in E-cadherin and a decrease in N-cadherin, α-SMA, and collagen I were observed. Western blot results confirmed these findings (Figures 8E–I, L–N). In summary, LG2 effectively mitigated the initiation of EMT and FMT in TGF-β1-stimulated A549 and MRC5 cells, facilitating ECM degradation. Consequently, we selected A549 cells for further investigation into LG2’s protective mechanisms against PF.

A549 cells undergoing EMT exhibit decreased cell-cell adhesion and enhanced migratory capabilities (Xu et al., 2017). Our findings indicate that TGF-β1 significantly increases the migratory capacity of these cells. However, LG2 treatment notably inhibited this enhanced migration (Figures 9A, B). Additionally, we assessed the expression of α-SMA and collagen I using IF, as both proteins play crucial roles in cell transdifferentiation and migration. The results (Figures 9C–F) demonstrate that LG2 effectively reduced the elevated levels of α-SMA and collagen I induced by TGF-β1.

Having established the significant roles of EMT and ECM in the pathogenesis of PF, we further explored their internal mechanisms of action. Network analysis and molecular docking studies have indicated a close association between the development of PF and the TNF-α signaling pathway. To validate this hypothesis, we initially examined the expression levels of TNF-α, p-IKK, p-NFκB, NFκB, MMP3, and MMP9 using western blot analysis and IHC in vivo and in vitro (Figures 10, 11). As anticipated, compared to the control group, the model group exhibited increased protein levels of TNF-α, p-IKK, p-NFκB, MMP3, and MMP9. Following treatment, the expression levels of above proteins were significantly reduced following treatment with LG2. The TNF-α signaling cascade, known for its critical role in regulating inflammation. In order to further confirm the involvement of the TNF-α pathway in modulating EMT, we introduced the specific TNF-α inhibitor, etanercept, for experimental validation in TGF-β1-induced A549 cells. The cells were categorized into several groups: Ctrl, TGF-β1, TGF-β1+LG2 (1.6 μmol/L), TGF-β1+Etanercept (100 μg/mL) (Gong et al., 2021), and TGF-β1+LG2+Etanercept. Western blot results demonstrated that etanercept considerably reduced the expression of inflammatory proteins (Figures 12A–F). Notably, the simultaneous use of etanercept and LG2 further decreased the inflammatory levels. These results confirm that LG2 exerts anti-fibrosis effects by inhibiting the TNF-α signaling pathway, modulating EMT, and remodeling the ECM.