The 18 Chinese medicines of QM (Table 1) were purchased from Beijing General Pharmaceutical Corporation (Beijing, China) and provided by the Department of Pharmacy, Affiliated Hospital of Changchun University of Traditional Chinese Medicine (Changchun, China). The voucher specimens were deposited at the Jilin Ginseng Academy, Changchun University of Chinese Medicine (Changchun, China). According to the standard procedure from Chinese Pharmacopoeia (2020 edition), all of the raw materials (138 g, clinical dosage) were decocted by distilled water twice (1 h each time) at a mass ratio of 1:10 to obtain the water extract of QM. The water extract was filtered and centrifuged to obtain the supernatant, which was dried in vacuum to obtain the powdery extract with a yield of 24.6% (34 g). The QM powdery extract was used for the UHPLC/Q-TOF-MS analysis and further experiments (Zhang et al., 2020).

The following reference compounds were used for the UHPLC/Q-TOF-MS analysis: calycosin 7-O-glucoside (batch number: 5240), complanatuside (batch number: 3037), ononin (batch number: 3811), neoisoliquiritin (batch number: 6971), neoliquiritin (batch number: 7543), formononetin (batch number: 3523), quercetin (batch number: 1116), genistin (batch number: 9435), oleanolic acid (batch number: 1339), oroxylin A (batch number: 6180), isoquercitrin (batch number: 5650), atractylodin A (batch number: 7361), and methylnissolin-3-O-glucoside (batch number: 3816). All were bought from Shanghai Sidande Standard Technical Service Co., Ltd. Shanghai, China. Schaftoside (batch number: DST200524-006), calycosin (batch number: DST190915-012), isoliquiritigenin (batch number: DST190719-016), naringin (batch number: DST191011-099), naringenin (batch number: DST200109-100), peimine (batch number: DST200109-011), peiminine (batch number: DSTDB001201), hesperidin (batch number: DST190716-038), neohesperidin (batch number: DSTDX003901), nobiletin (batch number: DSTDC003701), lobetyolin (batch number: DATDD003501), caffeic acid (batch number: DST190517-013), 5-O-demethylnobiletin (batch number: DSTDQ009401), 3′-demethylnobiletin (batch number: DST210212-120), betulinic acid (batch number: DST190508-026), and isoastragaloside I (batch number: DST180315-022) were purchased from Chengdu Deste Biotechnology Co., Ltd. (Chengdu, China). Liquiritin (batch number: PS012028) was purchased from Chengdu Pusi Biotechnology Co., Ltd. (Chengdu, China). Rapid characterization of the multicomponent from the QM extract was performed using a Waters ACQUITY UPLC I-Class Plus/Xevo G2-XS QTOF system (Waters, Milford, United States). Chromatographic separation was performed on a Waters ACQUITY UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 µm) maintained 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, 2–8% B; 4–7 min, 8–15% B; 7–11 min, 15–21% B; 11–14 min, 21% B; 14–16 min, 21–32% B; 16–19 min, 32% B; 19–24 min, 32–50% B; 24–27 min, 50–80% B; and 27–30 min, 80–95% B. The injection volume was 1 μl. 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 source parameters were as follows: capillary voltage, –2.0 kV in the negative ESI mode and 3.0 kV in the positive ESI mode; source temperature, 100°C; and cone voltage, 40 V. The mass analyzer scanned over the mass range of m/z 50–1,500 at a low collision energy of 6 V and a high collision energy of 40–60 V for the recording of MS¹ and MS² spectra by MS^E, respectively. Processing of the obtained MS^E data was conducted using UNIFI^TM 1.9.3.0 software (Waters, Milford, United States).

Seven male Wistar rats (weighing 250 ± 20 g) were treated by oral gavage with the clinical equal dose of the QM powdery extract (3 g/kg/day, equal to 12.4 g raw materials/kg/day), and the blood samples were collected 10 days after oral administration of QM. Prior to analysis by UHPLC/IM-QTOF-MS, the plasma samples were processed as described below. An aliquot of 200 μl of the plasma sample was added with 600 μl of methanol. The mixed sample was then vortexed for 3 min and centrifuged at 13,000 rpm for 10 min, to precipitate the proteins. The resulting supernatant was evaporated to dryness under a stream of nitrogen at 40°C, and the dried residue was further dissolved in 200 μl of methanol–water (1:1, v/v) solution. After centrifugation at 13,000 rpm for 10 min, the supernatant was used as the test solution and injected for analysis. Characterization of the QM components from the rat plasma was performed using a Waters ACQUITY UPLC I-Class/Vion IMS-QTOF mass spectrometer (Waters, Manchester, United Kingdom) by MS^E in both positive and negative ESI modes, following the same chromatography and MS conditions as described above for characterizing the multiple components from the QM extract.

The PubChem database was used to obtain the molecular structure (SDF format) of each QM compound (Wang et al., 2012). Then, the molecular structural files of identified compounds were uploaded to the PharmaMapper database, the ChemMapper database, and the SwissTarget database to predict their potential targets (Gong et al., 2013; Gfeller et al., 2014; Wang et al., 2017). Meanwhile, PF-related targets were collected using the GeneCards database (Stelzer et al., 2016). The gene and UniProt IDs of these potential targets were obtained from the UniProt database. Finally, the overlapping targets between the potential therapeutic targets of QM and the disease targets in PF were obtained for further network analysis.

Cytoscape 3.8.0 was used to visualize the relationship between components of QM and the potential targets of each compound. To explore the interaction among target proteins, overlapping targets were uploaded to STRING and the information about the protein–protein interaction (PPI) was obtained(Szklarczyk et al., 2015). Importantly, the significant Gene Ontology (GO), KEGG, and Reactome pathways were screened by the DAVID database (Huang et al., 2009).

The molecular structures of the compounds were searched from the PubChem database and reconstructed in the ChemBioDraw. Then, these structures were saved as “mol2” file with the energy minimized and saved as docking ligands as “pdbqt” files in AutoDock version 4.2, respectively. The PDB database (https://www.rcsb.org/) was used to retrieve the 3D structure of TGF-β1 (PDB ID: 1KLA) and TGF-β receptor 1 (TGF-βR1, PDB ID: 1PY5) with the files in the “pdb” format. Discovery Studio Client version 4.5 was used to hydrogenate proteins, remove water, and remove ligand molecules. AutoDock version 4.2 was used to add the nonpolar hydrogen, calculate Gasteiger charges for the protein, save it as “pdbqt” files, and finally, run Vina for molecular docking. The active site of molecular docking was determined by the ligand coordinates in the target protein complex. The ligand was set to be flexible, and the receptor was set to be rigid. Each receptor–ligand interaction generates 50 conformations, and the conformation with the best affinity was used as the final docking conformation. If the binding energy is less than 0, the compound (ligand) and protein (receptor) can bind spontaneously. If binding energy ≤ −1.2 kCal/mol, the ligand–receptor was considered stable (Gomez-Garcia et al., 2018).

Human adenocarcinomic A549 cells with type II alveolar epithelial cell features were bought from the cell bank of the Shanghai Institute of Cell Biology, Chinese Academy, and cultured in RPMI 1640 containing 10% fetal bovine serum (FBS, Clark Bioscience, Claymont, United States), 100 kU/L penicillin, and 100 mg/L streptomycin (Biosharp, Hefei, China) at 37°C in a 5% CO₂ humidified incubator. When the cells were in good condition, TGF-β1 (5 ng/ml) was added to induce A549 cell transformation into mesenchymal-like cells. After 24 h, the A549 cells were observed under the microscope (Olympus CKX53, Japan) to see whether their morphology had changed from hexagonal to a spindle shape after TGF-β1 treatment (Zhang et al., 2019).

A549 cells were seeded into 96-well plates at a density of 3,000, 6,000, or 8,000 cells/well. The effect of QM on cell viability was analyzed after treatment for 24, 48, or 72 h at a concentration of 7.8–1,000 μg/ml. Additionally, A549 cells were divided into the control group (dimethyl sulfoxide, DMSO alone), the model group (5 ng/ml TGF-β1), and the treatment group (5 ng/ml TGF-β1 + 7.8–1,000 μg/ml QM) to examine the effect of QM on the viability of TGF-β1–treated cells. After treatment for 24, 48, or 72 h, MTT (0.5 mg/ml) was added (Solibol, Beijing, China). DMSO (150 μl) was used to dissolve formazan crystals. Absorbance was measured at 490 nm using a microplate reader. The cell survival rate was calculated as the percentage of each group relative to the control group.

Total RNA was extracted from A549 cells using a total RNA extraction kit. Next, 1 μg of total RNA was reverse transcribed into cDNA with the iScript cDNA synthesis kit. A Bio-Rad CFX96 system was used to perform quantitative real-time PCR (qPCR) analysis, and the relative mRNA levels were calculated using the 2^−ΔΔCt method using GAPDH for normalization (Zhang et al., 2020). The primer sequences to amplify the genes encoding N-cadherin, E-cadherin, vimentin, α-SMA, LN, FN, collagen I, TGF-βR1, Smad7, and GAPDH are listed in Supplementary Table S2.

Antibodies against LN (ab11575), FN (ab2413), collagen I (ab34710), α-SMA (ab5694), E-cadherin (ab40772), N-cadherin (ab76011), vimentin (ab8978), TGF-βR1 (ab31013), Smad3 (ab40854), p-Smad3 (S423/S425 and ab52903), Smad7 (ab216428), and GAPDH (ab8245) were purchased from Abcam (Cambridge, MA, United States). After QM treatment, the total protein fraction was extracted from A549 cells by lysis in ice-cold RIPA buffer. Proteins were separated by SDS-PAGE and 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 washing, the membranes were incubated with the HRP-conjugated secondary antibody for 2 h. The protein bands were visualized using a chemiluminescent imaging system (ChemiDoc XRS+, Bio-Rad) and quantified by ImageJ software (Jiang et al., 2021).

The data are expressed as mean ± standard deviation from three independent experiments. All of the data were statistically analyzed using GraphPad prism 9.0. Multiple groups were compared by one-way ANOVA (Tukey’s post hoc test) to determine statistical significance. For all statistical analyses, p < 0.05 was considered to be statistically significant.

To identify the chemical components of QM and analyze the consistency between various batches of QM, UHPLC/Q-TOF-MS analysis was conducted. The extracted ion chromatograms of all the compounds are shown in Figures 1A,B. A total of 56 components were characterized, among which 19 were accurately identified via comparison with the reference compounds. Information of these characterized components, including their assignments to the single drugs, is presented in Supplementary Table S1. These components were mainly from nine botanical drugs, that is, astragali radix (Huang-qi), citri reticulatae pericarpium (Chen-pi), salviae miltiorrhizae radix et rhizoma (Dan-shen), glycyrrhizae radix et rhizoma (Gan-cao), glehniae radix (Bei-sha-shen), atractylodis macrocephalae rhizoma (Bai-zhu), pinelliae rhizoma (Ban-xia), codonopsis radix (Dang-shen), and fritillariae thunbergii bulbus (Zhe-bei-mu). Nineteen compounds were identified by comparison with reference compounds, namely, atractyloside A, complanatuside, schaftoside, calycosin 7-O-glucoside, liquiritin, isoliquiritigenin, naringin, peimine, hesperidin, neohesperidin, peiminine, lobetyolin, neoisoliquiritin, ononin, methylnissolin-3-O-glucoside, calycosin, naringenin, nobiletin, and isoastragaloside I. Moreover, 10 batches of QM samples were analyzed, and similarity values of 0.983–0.999 were obtained by comparing with the reference spectrum, which indicated that the QM formula was reproducible (Figures 1C,D).

Combined with the PubChem database, 43 of the 56 chemical compounds from the UHPLC/Q-TOF-MS analysis had a well-defined molecular structure. Then, according to the “Network Pharmacology Evaluation Method Guidance” (Li, 2021), the 43 chemical compounds were used for the network pharmacological analysis. First, the PubChem database was used to obtain the molecular structure files of the 43 QM components. The potential therapeutic targets of the 43 ingredients were obtained from the PharmaMapper database, the ChemMapper database, the SwissTarget database, and the literature reports (Bunbupha et al., 2020; He et al., 2020a; Wang et al., 2021). These targets were screened according to many parameters, such as fit score ≥3.0 in the PharmaMapper database, score ≥0.4 in the ChemMapper database, and probability ≥0.4 in the SwissTarget database. After screening, we obtained the UniProt ID corresponding to each protein target and deleted reduplicated targets. Thus, a total of 452 potential therapeutic targets of QM were obtained (Supplementary Table S4). Second, the network of 43 active components and 452 potential targets of QM was constructed (Supplementary Figure S1). Furthermore, potential targets in PF were obtained from the GeneCards database, and 642 targets related to PF were obtained with the relevance score >11.63 (depending on the median to screen). Although the relevance score of angiotensin-converting enzyme 2 (ACE2) was lower than 11.63, ACE2 is an important target of COVID-19–related PF (Mourad and Levy, 2020), and QM is a TCM formula for the treatment of COVID-19 convalescent patients with PF. Therefore, ACE2 was also selected as a target of PF in this study, and a total of 643 targets of PF were screened out for further analysis. Based on 452 targets of QM and 643 targets in PF, 70 overlapping targets between QM and PF were obtained (Supplementary Table S3). Then, we established a network between the 43 QM components and 70 overlapping targets (Figure 2A). Through the network, we found that 2-succinyl-6-hydroxycyclohexa-2, 4-diene-1-carboxylic acid, and codonopsine were not connected with the overlapping targets. Then, the 70 overlapping targets were submitted to the STRING database to build the PPI network. As shown in Figure 2B, the PPI network had 70 nodes and 866 edges. Subsequently, topological screening with degree ≥24, betweenness ≥0.01, and closeness ≥0.615 was used to screen key targets (Figure 2B). In addition, the top 20 targets of QM, such as TNF, MMP9, ALB, MAPK1, IGF1, SRC, MAPK8, CASP3, AKT1, MMP2, EGFR, MAPK14, HRAS, MAP2K1, SMAD3, TGFB1, CAT, IL2, ANXA5, and PLG, were selected and ranked with the maximal clique centrality (MCC) score (Figure 2C). Taken together, 70 potential targets of the 43 components of QM were obtained by network pharmacology.

The 70 overlapping targets were submitted to the DAVID database to conduct GO enrichment analysis, from which we obtained 402 biological process (BP) terms (Supplementary Table S5), 36 cellular component (CC) terms (Supplementary Table S6), and 77 molecular function (MF) terms (Supplementary Table S7). The top 20 entries were respectively selected from BP, CC, and MF, in order of -lg p value (Figure 3A). In the BP category, ECM disassembly, mesenchymal cell differentiation, EMT, the collagen catabolic process, and proteolysis were directly related to PF. In the MF category, the targets were mainly involved in protein tyrosine kinase activity, protein kinase activity, TGF-β receptor binding, type II TGF-β receptor binding, protease binding, fibroblast growth factor–activated receptor activity, and fibroblast growth factor binding. The DAVID database was also used to obtain 71 Reactome pathway terms and 99 KEGG pathway terms (Supplementary Tables S8, S9). The top 50 entries of the KEGG database (Figure 3B) and the top 30 entries of the Reactome database (Figure 3C) were selected depending on the -lg p value. The KEGG pathways mainly included FoxO, Rap1, MAPK, VEGF, and TGF-β signaling pathways. The Reactome pathways mainly included activation of matrix metalloproteinases, degradation of the ECM, signaling by SCF-KIT, regulation of IGF transport and uptake by IGFBPs, collagen degradation, and activation of Smads by TGF-β receptor signaling. Collectively, the network pharmacological analysis indicated that QM may inhibit the TGF-β signaling pathway to mediate ECM accumulation, fibroblast activation, and EMT to prevent the progression of PF. The key targets of the TGF-β signaling pathway that participated in PF are shown in Figure 3D.

The cytotoxicity of QM was tested. We found that QM at a concentration of 7.8–1,000 μg/ml was not cytotoxic for A549 cells after treatment for 24, 48, or 72 h (Figure 4A). Moreover, the viability of A549 cells was determined in the presence of TGF-β1 to mimic a pro-fibrotic environment. Compared with the control group, TGF-β1 significantly induced proliferation of A549 cells. After QM treatment at 7.8–1000 μg/ml for 24, 48, or 72 h, the proliferation of TGF-β1–induced A549 cells was decreased (Figure 4B). These data indicate that QM could inhibit the proliferation of TGF-β1–induced A549 cells.

TGF-β1 exerts strong pro-fibrotic effects and can induce EMT (Pezone et al., 2020), which is a phenomenon where inherent epithelial markers such as E-cadherin are lost and mesenchymal characteristics such as N-cadherin, vimentin, and α-SMA are gained (Cruz-Solbes and Youker, 2017). The morphology of A549 cells changes from hexagon to an elongated spindle-like shape when EMT occurs (Zhang et al., 2019). In the present study, after treatment with TGF-β1 (5 ng/ml) for 72 h, a spindle-like shape of A549 cells was observed (Figure 5A). When A549 cells were treated with QM at 125, 250, or 500 μg/ml, the mesenchymal-like morphology was attenuated compared with the TGF-β1–induced group (Figure 5A). To further verify the effects of QM in inhibiting EMT at the molecular level, qPCR and Western blot analyses were used to detect the EMT-related markers. After 72 h of incubation, qPCR results showed that the E-cadherin mRNA expression was decreased, whereas the mRNA levels of N-cadherin, α-SMA, and vimentin were upregulated in TGF-β1–induced A549 cells. Compared with the model group, the expression of E-cadherin was increased, and N-cadherin, α-SMA, and vimentin mRNA levels were downregulated after QM treatment at 125, 250, or 500 μg/ml (Figure 5B). Western blot analysis confirmed the qPCR results (Figures 5C,D). Together, these results demonstrated that QM inhibits EMT in TGF-β1–induced A549 cells.

The above findings revealed that EMT occurred in A549 cells after induction with TGF-β1. Cells with features of mesenchymal cells, such as fibroblasts and myofibroblasts, could secrete ECM proteins, such as collagen I, FN, and LN (Thiery et al., 2009). Excessive deposition of ECM is an important cause of organ fibrosis (Sun et al., 2013). In addition, the results of Reactome pathway enrichment from network pharmacology showed that QM could prevent PF via degrading ECM (Figure 3). To investigate the mechanisms by which QM stimulates ECM degradation, qPCR and Western blot analyses were conducted to investigate the protein and mRNA expression levels of key ECM markers, including LN, FN, and collagen I. The expression of collagen I, FN, and LN was significantly upregulated at the mRNA level in TGF-β1–induced A549 cells. QM treatment for 72 h decreased the level of collagen I, FN, and LN in a dose-dependent manner (Figure 6A). At the protein level, QM significantly reduced the TGF-β1–induced A549 cells which resulted in the increase in LN, FN, and collagen I levels. The inhibitory effects of QM on ECM accumulation were the strongest at a concentration of 500 μg/ml (Figures 6B,C). Together, these results suggested that QM stimulates ECM degradation in TGF-β1–induced A549 cells.

TGF-β/Smads signaling is a classical pathway in the occurrence and progression of organ fibrosis (He X. et al., 2020), giving rise to the occurrence of EMT and ECM deposition (Pardali et al., 2017). Moreover, the results of our network pharmacological analysis also showed that QM may exert its anti-PF effects through the TGF-β signaling pathway (Figures 3B–D). To further verify whether QM could regulate the TGF-β/Smads pathway, TGF-βR1, Smad7, and Smad3 and its phosphorylation, as downstream proteins of TGF-β1, were detected by western blot and/or qPCR. As shown in Figure 7A; Supplementary Figure S2, TGF-βR1 was significantly upregulated, and Smad3 phosphorylation was induced by TGF-β1 in the model group. After QM treatment for 72 h, these indicators were downregulated. Moreover, in TGF-β1–stimulated A549 cells, there was no change in the Smad3 protein level in the QM-treated group, which meant that QM inhibited the TGF-β1–induced increase in the p-Smad3/Smad3 ratio. In addition, as an inhibitor of the TGF-β/Smad3 pathway, TGF-β1 could increase the expression of Smad7 due to feedback action (Wang et al., 2014). But QM treatment had no effect on the increased Smad7 level induced by TGF-β1. Taken together, these results indicated that QM could inhibit the TGF-β1/Smad3 pathway in TGF-β1–induced A549 cells.

To identify the effective components of QM against PF, the QM components that could be absorbed into the bloodstream were characterized by UHPLC/IM-QTOF-MS. By comparing the MS data, the peaks that appeared in the corresponding positions of the QM-administered plasma and QM extract but did not appear in the blank rat plasma were considered as the prototype compounds absorbed into the plasma of rats. Accordingly, 10 prototype compounds were identified in the rat plasma: indole-3-acrylic acid from glehniae radix (Bei-sha-shen), schaftoside from pinelliae rhizoma (Ban-xia), and ononin from astragali radix (Huang-qi), and other seven components were from glycyrrhizae radix et rhizoma (Gan-cao), such as liquiritin, isoliquiritigenin, neoisoliquiritin, licoricesaponin A3, licoricesaponin G2, yunganoside G1, and licoricesaponin K2. The detailed information expressed by # is displayed in Supplementary Table S1. These results demonstrated that glycyrrhizae radix et rhizoma (Gan-cao) might be a key botanical drug of QM for preventing PF. However, the real and potential compounds responsible for the anti-PF effect of QM should be further identified in the mouse model with PF.

We further explored the relationship between i) the nine effective components that can be absorbed into the bloodstream and ii) the TGF-β signaling pathway. Molecular docking was used to analyze the binding energies of the nine compounds with TGF-β1 and TGF-βR1. As shown in Figure 7B, the nine compounds all had good binding affinities with TGF-β1 and TGF-βR1, especially the compounds from glycyrrhizae radix et rhizome (Gan-cao), such as liquiritin, isoliquiritigenin, neoisoliquiritin, licoricesaponin A3, licoricesaponin G2, and licoricesaponin K2. The molecular docking results indicated that the nine compounds from QM could all potentially regulate the TGF-β signaling pathway. However, the pharmacology action of these active components in inhibiting the TGF-β signaling pathway needs to further investigate in the cell or animal models.