Collagen I (ab34710), α-SMA (ab5694), E-cadherin (ab40772), N-cadherin (ab76011), TGF-β1 (ab215715), p-Smad3 (S423/S425, ab52903), and goat anti-rabbit/mouse antibodies (ab6721, ab6789) were purchased from Abcam (Cambridge, MA, USA). BLM (HY17565) and pirfenidone (PFD, HYB0673) were obtained from MedChemExpress (Princeton, NJ, USA). Calycosin-7-glucoside (ST08820120-5240), scutellarin (RS03111020-5839), apigenin-7-O-β-D-glucuronic acid (ST08920120-3075), oroxin B (ST09240120-5589), scutellarin methylester (ST15580120-4087), baicalin (RS01861020-5587), salvianolic acid A (ST005570120MG-4881), emodin glucoside (ST10890120-7823), wogonoside (ST08350120-5014), apigenin (RS000411020-5729), scutellarein (ST06340120-3414), baicalein (RS01871020-5699), isoastragaloside IV (ST81100105-6855), wogonin (RS01711020-5325), oroxylin A (ST23660220-6180), emodin (RS01401020-5940), salvianolic acid F (ST22410105mg-5189), salvianolic acid B (ST000500120mg-4065), and luteolin (RS00071020-5864) were purchased from Shanghai Shidande Standard Technical Service Co., Ltd. (Shanghai, China). Astragaloside I (DST190216-016), astragaloside II (DST180315-023), and astragaloside III (DST181118-018) were purchased from Chengdu Durst Biotechnology Co., Ltd. (Chengdu, China).

The 13 Chinese medicines (Table 1) constituting WBT were purchased from Beijing General Pharmaceutical Corporation (Beijing, China) and provided by the Department of Pharmacy, the Affiliated Hospital of Changchun University of Chinese Medicine (Changchun, China). The voucher specimens were deposited at the Jilin Ginseng Academy, Changchun University of Chinese Medicine. According to the preparation processing as traditional use, the standard procedure from Chinese Pharmacopoeia (2020 edition), and previous studies (Huang Q. et al., 2018; Yang et al., 2021), all drugs were blended, mixed, and extracted by 2,000 ml of distilled water (the drug solvent ration of this formula is 9.0) at 100°C for 30 min, which was repeated three times to obtain the aqueous extract of WBT formula. After centrifugation, the supernatant was dried under a vacuum to produce a powdery extract with the drug-extract ratio of 22.47% (obtained 48 g of powdery extract from 214 g of raw materials). The powder was stored at −20°C for further experiments.

As previous described (Zuo et al., 2019; Yang et al., 2021), UHPLC analyses were carried out with a Waters ACQUITY UPLC I-Class Plus/Xevo G2-XS QTOF system (Waters, Milford, USA). Chromatographic separation was performed on a Waters ACQUITY UPLC BEH C18 (4.6 × 150 mm, 3.5 μm) at 30°C. The mobile phase was composed of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) running at 0.3 ml/min, consistent with the following optimal gradient elution program: 0–4 min, 1% B; 4–5 min, 1%–5 B; 5–10 min, 5%–17 B; 10–18 min, 17%–17 B; 18–23 min, 17%–22 B; 23–28 min, 22%–25 B; 28–34 min, 25%–32 B; 34–37 min, 32%–50 B; 37–40 min, 50%–95 B. The injection volume of WBT samples was 5 µl. The high-resolution MS data were recorded on a Xevo G2-XS QTOF mass spectrometer by MS^E in both the positive and negative ESI modes. The parameters of the electrospray ion source were set as follows: capillary voltage, 2.5 kV; cone voltage, 60 V; collision energy, 40–80 eV; scan mass range, 100–1,500 m/z; ion source temperature, 120°C; desolvation gas temperature, 500°C; cone gas flow rate, 50 L/h; desolvation gas flow rate, 800 L/h. The obtained MS^E data were conducted using UNIFI™ 1.9.3.0 software (Waters, Milford, USA).

The PubChem database (https://pubchem.ncbi.nlm.nih.gov/) was used to obtain the molecular structure (SDF format) of each component that identified from WBT powder extract. Then, the molecular structural files were uploaded to the PharmaMapper (http://lilab-ecust.cn/pharmmapper/submitfile.html), ChemMapper (http://lilab-ecust.cn/chemmapper/), and SwissTarget (http://www.swisstargetprediction.ch/index.php) databases to predict the potential targets of each component from WBT (Huang H. et al., 2018; Yang et al., 2020; Mu et al., 2021). The GeneCards database (https://www.genecards.org/) was used to collect PF-related targets (Mu et al., 2021). Finally, the corresponding Gene and UniProt ID of each target were obtained from the UniProt database (https://www.uniprot.org/) (Shawky, 2019). The targets were used for the establishment of a component-target network.

The targets of WBT and PF were combined so that the overlapping targets between PF and WBT could be obtained. Then, the overlapping targets were uploaded to the STRING database, and the information about protein–protein interaction was obtained. Meanwhile, the network of Chinese medicines, chemical ingredients, and potential targets was established and visualized by Cytoscape 3.8.0. The components of WBT and targets of the components were represented by hexagon and (or) rectangle, and the interaction was shown by a connecting line. Finally, depending on the overlapping targets, the significant Gene Ontology (GO) and Reactome Pathway were screened by the Metascape and DAVID databases.

All animal care and procedures were approved by the Experimental Animal Ethics Committee of Changchun University of Chinese Medicine (batch number: 20190134). Seventy of 8-week-old male C57BL/6 mice purchased from Beijing Weitong Lihua Experimental Animal Technology Co., Ltd. [License No. SCXK (Beijing) 2016-0006] were kept at the Animal Experimental Center of Changchun University of Chinese Medicine (Changchun, China) at an ambient temperature (20 ± 2°C) and 50%–60% humidity. In this study, mice were randomly divided into seven groups with 10 mice in each group, including the blank control, sham operation, BLM, BLM + WBT (3, 6, and 12 g/kg, among them, the dose of 6 g/kg/day is equal dosage of clinic, 12 g/kg/day is the higher dosage, and 3 g/kg/day is the lower dosage), and BLM + pirfenidone (PFD, 200 mg/kg) groups, according to the book named Experimental Methodolgy of Pharamacology (2002 version). Except for the control group, mice were intraperitoneally with pentobarbital sodium (10 ml/kg) and intratracheally injected with BLM (5 mg/kg) or 0.9% normal saline (sham operation group), according to the previous report (Shahbaz et al., 2021). WBT and PFD were administered by gavage on the second day after BLM or NS injection for 14 days (once a day). Mice in the blank control, sham, and model groups were given pure water in the same amount of WBT.

Micro-CT imaging was performed with Quantum FX Micro CT (PerkinElmer, Inc., Waltham, MA, USA). After anesthetizing with isoflurane, the lungs were imaged with the help of cardiopulmonary gating techniques. The X-ray system of the scanner used a micro-focusing tube with a focal spot of 5 μm at 4 W and generated X-rays with cone beam geometry. The scanner’s detection system was composed of an amorphous silicon digital X-ray plate, which could acquire projection radiographs at a rate of 30 frames per second. The X-ray tube was set at 90 kV and 80 μA and took the projection radiographs during a 360° gantry rotation, with a total scanning time of 4 min. The reconstructed visual field was 36 mm with a high-resolution scanning mode (Lv et al., 2013; Ruscitti et al., 2017).

H&E and Masson’s trichrome stainings were performed to evaluate the pathological changes of mouse lung tissues from different groups, as previously described (Zhao et al., 2020). For H&E and Masson’s trichrome staining, the exfoliated lung tissue of mice was fixed in 4% paraformaldehyde for 24 h, embedded in paraffin, cut into the 5-µm-thick sections, and stained with H&E and Masson’s trichrome kits. The degrees of inflammatory and fibrotic injuries in the lung tissues were scored according to the Mikawar and Ashcroft methods, respectively (Katoh et al., 1998; Pedroza et al., 2016). Immunohistochemical (IHC) staining was used to analyze the levels of PF-related proteins, such as collagen I, α-SMA, E-cadherin, N-cadherin, and TGF-β1 in the lung tissues, according to the manufacturer’s instructions (Kishi et al., 2018). After dewaxing and antigen repairing, the tissue was incubated with primary antibodies at 4°C overnight, followed by a conjugated secondary antibody and DAB staining. After counterstaining with hematoxylin, the images for antigenic sites were visualized and acquired by a Digital Microscope and Slide Scanner (M8, PreciPoint, Thuringia, Germany). The relative staining score of each protein was determined by calculating the areal density using Image-Pro Plus 6.0 software (Media Cybemetics, Inc., Rockville, MD, USA).

The content of hydroxyproline (HYP) in mouse lung tissue was quantified by the HYP assay kit (Nanjing Jiancheng Institute of Bioengineering), according to a previous study (Lv et al., 2013).

The TC-1 cell line was bought from the iCell Bioscience, Inc. (Shanghai, China), and cultured in RPMI 1640 containing 10% fetal bovine serum (Clark Bioscience, Claymont, USA), penicillin (100 kU/L), and streptomycin (100 mg/L) (Biosharp, Hefei, China) at 37°C in a 5% CO₂ humidified incubator. When the cells were in good condition, in accordance with the previous study (Zeng et al., 2021), TGF-β1 (10 ng/ml) was added to induce TC-1 cells transformation into mesenchymal-like cells.

According to the previous study (Zeng et al., 2021), TC-1 cells were seeded into 96-well plates at a density of 6,000 cells/well, which were used to investigate the effect of WBT on cell viability after treatment for 48 h. To investigate the toxicity of WBT, TC-1 cells were treated with WBT at a concentration of 7.1–1,000 μg/ml for 48 h. In the TGF-β1–induced cell model, TC-1 cells were set as the control [dimethyl sulfoxide (DMSO) alone], model [TGF-β1 (10 ng/ml)], and treatment group [TGF-β1 (10 ng/ml) + WBT (7.1–1,000 μg/ml)] to examine the effect of WBT on the cell viability. After treatment for 48 h, MTT (0.5 mg/ml) was added (Solibol, Beijing, China), and DMSO (150 μl) was used to dissolve formazan crystals, which was measured as the absorbance at 490 nm using a microplate reader. The cell survival rate was calculated as the percentage of each group relative to the control group.

The total RNA of TC-1 cells in different groups was extracted by a total RNA kit. After that, 1 μg of total RNA was reverse-transcribed into cDNA with the iScript cDNA synthesis kit. The Bio-Rad CFX96 system was used to perform quantitative (qPCR) analysis, and the relative mRNA level was calculated using the 2^−ΔΔCt method and normalized by GAPDH (Zhang Z. et al., 2020). The primer sequences of N-Cadherin, E-cadherin, α-SMA, collagen I, and GAPDH are listed in Supplementary Table S1.

The total protein of TC-1 cells in different groups was obtained by an ice-cold RIPA buffer and separated by SDS-PAGE, transferred onto PVDF membranes. After blocking for 1 h at room temperature with a blocking buffer containing 5% BSA, the membranes were incubated with the primary antibody overnight at 4°C. After incubating with the HRP-conjugated secondary antibody for 2 h, the blots were visualized by a chemiluminescent imaging system (ChemiDoc XRS+, Bio-Rad, CA, USA) and quantified by ImageJ software (Jiang et al., 2021).

The data are expressed as mean ± SD. All data were statistically analyzed by GraphPad prism 9.0 software (San Diego, CA, USA). Multiple groups were compared by one-way ANOVA (Tukey’s post hoc) to determine the statistical significance. For all the statistical analyses, p < 0.05 was statistically significant.

After optimization of the chromatographic and mass spectrometric conditions, a total of 42 compounds were identified or deduced from WBT formula, in which 16 components (expressed by * in Supplementary Table S2), including calycosin-7-glucoside, scutellarin/methylester, baicalin, and salvianolic acid A/B, were identified by the reference standards. The 42 compounds in WBT were characterized by comparing the retention time, accurate mass, and fragment ions with those of the standards or data reported by the literature. Furthermore, the attribution to the single herb was confirmed by the base peak intensity (BPI) chromatograms of the blank sample, WBT formula, and individual herbs. The BPI chromatograms of the WBT formula after collection in the positive and negative ion modes are shown in Figures 1A,B. The detailed information for 42 compounds in WBT is shown in Supplementary Table S2. In addition, the fingerprint and the similarity of the WBT formula were further analyzed by UHPLC method. As shown in Figures 1C,D, the percentages of the similarities from all 10 batches of WBT powdery extract were range from 91.0% to 99.3%, which indicates that the WBT powdery extract has good reproducibility. Importantly, we selected six standard compounds in WBT extract to perform quantitative analysis to avoid the risk of artifacts with higher dose in animal study. As shown in Supplementary Figure S1 and Supplementary Table S3, the WBT extract included calycosin-7-glucoside (0.4499 mg/g), baicalin (17.2408 mg/g), salvianolic acid B (3.3389 mg/g), salvianolic acid A (0.96 mg/g), emodin-8-β-D-glucoside (1.0735 mg/g), and wogonoside (6.7344 mg/g), which were the dose range as indicated in the consensus document of clinical evidence-based ethnopharmacology.

According to the PubChem database, the 42 components in the WBT powdery extract were normalized, and those that had not been confirmed potential therapeutic targets were excluded, leaving 36 ingredients for further analysis of the network pharmacology (Table 2). The molecular structure files of the 36 components were obtained from the PubChem database, depending on which the potential targets of 36 components were obtained from several databases, such as PharmaMapper database, ChemMapper database, and SwissTarget database. Moreover, the potential targets were screened according to many parameters (fit score ≥2.5 in the PharmaMapper database, score ≥0.01 in the ChemMapper database, probability ≥0.079 in the SwissTarget database). Afterward, the UniProt ID corresponding to each target was obtained from the UniProt database, and the reduplicative targets were deleted to obtain a total of 687 potential therapeutic targets of 36 components of the WBT powdery extract (Supplementary Table S4). Subsequently, the network of 36 components and their 687 potential targets was constructed (Supplementary Figure S2). In addition, according to the score >11.8, a total of 584 potential targets of PF were selected from the GeneCards database. On the basis of the 687 potential targets of WBT and the 584 potential targets of PF, 93 overlapping targets were obtained, and the network of 36 active components and 93 overlapping targets was established and is shown in Figure 2A and Table 3. Importantly, on the basis of the topological analysis for component-overlapping targets network, the top four components of WBT formula were screened out depending on the degree >48, including baicalein, oroxylin A 7-O-β-D-glucuronide, viscidulin I, and viscidulin III (Supplementary Figure S3). As shown in Figure 2B, the PPI network of potential targets from WBT formula has 93 nodes and 1,272 edges. The 23 main targets, represented by yellow circles, contained ALB, VEGFA, AKT1, TNF, MMP2, MMP9, STAT3, EGFR, MAPK1, FGF2, CASP3, and others after screening with degree ≥27, betweenness ≥0.08, and closeness ≥0.58 (Figure 2C). Importantly, among the 93 overlapping targets, TGF-β and its receptors (TGF-βR1 and TGF-βR2), as classical profibrotic indicators, were also enriched, which are potential targets of 31 components in the WBT formula (Figure 2D). Collectively, 36 components and their 93 potential targets for PF were predicted.

The Metascape database was used to perform GO enrichment for analyzing the potential mechanisms of the WBT formula. For these mechanisms, the 390 biological processes (BPs), 80 cellular components (CCs), and 105 molecular functions (MFs) were obtained and are shown in Supplementary Tables S5–S7. Moreover, the top 20 entries were selected from BP, CC, and MF, in order of -log p value, are shown in Figure 3A. The BPs regulated by WBT included the positive regulation of kinase activity, wound healing, inflammatory response, cytokine production, and the apoptotic signaling pathway. For CCs, CC mainly contained the vesicle lumen, the ECM, and focal adhesion. In the MFs, the targets were mainly involved in peptidase activity, kinase activity, BMP receptor binding, cytokine receptor binding, and other bindings. Furthermore, the DAVID database was used to obtain 91 terms from the Reactome pathway enrichment analysis (Supplementary Table S8). Importantly, the top 20 enrichment pathways significantly involved in the mechanism of WBT formula included the RAF/MAP kinase cascade (KIT, MAPK1, KRAS, TEK, JAK2, HRAS, FGF2, EGFR, IL2, FGFR2, and FGFR1), the degradation of the ECM (MMP1, MMP2, MMP3, MMP7, MMP8, MMP9, MMP12, CASP3, CTSG, PLG, and ELANE), the activation of MMPs (PRSS1, MMP1, MMP2, MMP3, MMP7, MMP8, MMP9, CTSG, PLG, and ELANE), collagen degradation (MMP1, MMP2, MMP3, MMP7, MMP8, MMP9, MMP12, and ELANE), and other pathways (Figure 3B). Remarkably, these potential targets participating in the degradation of ECM were regulated by TGF-β signaling pathway, which is an important regulator in the occurrence and development of fibrosis (Frangogiannis, 2020). Together, the network pharmacology results indicate that WBT may inhibit the TGF-β signaling pathway to regulate ECM production and degradation or other downstream pathways for the blockade of PF progression. However, except for the TGF-β signaling pathway, the mechanisms and targets from the network pharmacology need to be further evaluated.

To further assess the protective effect of WBT against PF, mice were treated with WBT or PFD for 14 days after intratracheal injection of BLM (Figure 4A). After the treatment, the lungs of mice in the different groups (Sham, BLM, BLM + WBT, and BLM + PFD) were observed in three different planes by micro-CT scanning. As shown in Figure 4B, the micro-CT imaging shows that lung tissues from blank and sham groups show clear texture, no abnormal density shadow, and a bronchial vascular bundle that runs normally. The interstitial thickening around bronchial vascular bundles in the axial fiber system and the diffuse interlobular septum thickening in the peripheral fiber system can be markedly observed in the both lungs from BLM-induced model group. Moreover, subpleural linear shadows are seen in many places, indicating that BLM-induced lungs have obvious characteristics of ground glass density and grid-like changes. Compared with model group, the lung texture is clearer, and the bronchial vascular bundles are not significantly thickened in the lungs from WBT-treated group. The original diffuse ground glass density and grid-like change in both lungs were obviously inhibited by WBT treatment (Figure 4B). In addition, BLM-induced decreases of lung volume were slightly improved by WBT treatment, compared with that of the model group (Figure 4C). Moreover, the protective effect of WBT in 12 g/kg was significantly better than that in 6 and 3 g/kg. The improvement for the changes of PF by WBT at 12 g/kg was similar with that of PFD at 200 mg/kg. Therefore, in subsequent experiments, the WBT group of 12 g/kg was used for histopathological and IHC analysis.

H&E and Masson’s trichrome stainings were performed and analyzed to investigate the effect of WBT on histological changes in the lung tissues from BLM-induced mice. The H&E staining results show that many alveolar walls were thickened and accompanied by infiltration of lymphocytes and macrophages in the BLM-induced group, compared with that in the sham group. BLM injection increased the number of inflammatory cells infiltrating around blood vessels and forming vascular sleeves, accompanied by acidic serous substance exudation in the local alveolar cavity (Figure 5A, upper panel). After 14 days of treatment, the extent of tissue destruction was significantly suppressed by WBT or PFD. The bronchial epithelial structure was intact, and the epithelial cells were normal and closely arranged in the WBT group. However, other histological characteristics, including thicker alveolar walls, lymphocyte infiltration, and a small amount of eosinophil and blood in the local alveolar cavity were not obviously improved by WBT (Figure 5A, upper panel). Masson’s trichrome staining confirmed that the blue staining for fibrous collagen deposition as an index of PF in the lung tissues induced by BLM was significantly inhibited by WBT or PFD (Figure 5A, lower panel). The areas of collagen deposition in WBT or PFD-treated lung tissues were markedly lower than that in the BLM-induced group (Figure 5B). Notably, the scores from Mikawar and Ashcroft methods were reduced by WBT on the basis of H&E and Masson’s trichrome stainings, which indicates that WBT can decrease the lung injury induced by BLM (Figure 5C). Together, WBT treatment can improve lung architecture and collagen deposition to block the progression of PF in the BLM-induced mouse model.

As enriched by the network pharmacology, the TGF-β signaling pathway is critical for the protection of WBT against PF. Therefore, the expressions of TGF-β and p-Smad3 in different lung tissues from the sham, model, or WBT/PFD groups were determined by IHC staining. As shown in Figures 6A,B, the increase of TGF-β level induced by BLM is significantly down-regulated by WBT. Furthermore, the phosphorylation of Smad3 is significantly inhibited by the WBT treatment, compared with the BLM-induced model group (Figure 6A, lower panel, and Figure 6C). In addition, WBT and PFD have similar inhibitory effects in TGF-β and p-Smad3 expressions at the protein level. Collectively, WBT blocks the TGF-β/Smad3 pathway in the lung tissues of BLM-induced mice, which confirms the enrichment of TGF-β pathway for WBT by the network pharmacology.

Collagen I is an important component of ECM accumulation, an important characteristic of PF (Long et al., 2018). As shown in Figure 3, the collagen degradation pathway is a potential mechanism of WBT against PF, according to the results of network pharmacological analysis. In the BLM-induced mouse model, the level of collagen I was up-regulated by BLM induction, which was significantly decreased by the treatment of WBT or PFD (Figures 7A,B). Moreover, HYP, a special amino acid extracted from collagen, represents the content of total collagen, which is a theoretically guide for the diagnosis and prognosis of PF (Berisio et al., 2004). The measurement of HYP content in mouse lung tissues showed that the HYP level in BLM-induced model group was higher than that in the sham groups, whereas the content of HYP induced by BLM was significantly decreased by WBT treatment (Figure 7C). In addition, in TGF-β1–induced TC-1 cells (the cytotoxicity test of WBT to TC-1 cells is shown in Supplementary Figure S4), the expression of collagen I was also significantly down-regulated by WBT treatment (Figure 7D). Together, these results indicated that WBT can attenuate collagen accumulation in vivo and in vitro.

EMT is a common developmental process in the pathogenesis of PF, which can be induced by TGF-β (Nieto et al., 2016). Consistent with the previous study (Ding et al., 2021), the treatment of WBT significantly inhibited EMT in vivo and in vitro. In the BLM-induced mouse model, the IHC staining demonstrated that the EMT with the decreased E-cadherin and increased N-cadherin and α-SMA was promoted by BLM, which was blocked by WBT, compared with that of the model group (Figure 8A, middle and lower panels, and Figures 8C,D). In this BLM-induced mouse model, the regulation of WBT on EMT markers was similar to that of PFD (Figure 8). In addition, in TGF-β1–induced TC-1 cells, the changes of α-SMA, E-cadherin, and N-cadherin regulated by the WBT formula at the mRNA and protein levels were similar to those in the BLM-induced mouse model (Figures 8E–G). Together, these data indicate that WBT can inhibit EMT in the BLM-induced mouse model and the TGF-β1–induced cell model.