A Combined Phytochemistry and Network Pharmacology Approach to Reveal the Effective Substances and Mechanisms of Wei-Fu-Chun Tablet in the Treatment of Precancerous Lesions of Gastric Cancer

Wei-Fu-Chun (WFC) tablet is a commercial medicinal product approved by China Food and Drug Administration, which is made of Panax ginseng C.A.Mey., Citrus aurantium L., and Isodon amethystoides (Benth.). WFC has been popularly used for the treatment of precancerous lesions of gastric cancer (PLGC) in clinical practice. In this study, a UHPLC-ESI-Q-TOF/MS method in both positive and negative ion mode was employed to rapidly survey the major constituents of WFC. 178 compounds including diterpenoids, triterpenes, sesquiterpenes, flavonoids, saponins, phenylpropanoids, lignans, coumarins, organic acids, fatty acids, quinones, and sterols, were identified by comparing their retention times, accurate mass within 5 ppm error, and MS fragmentation ions. In addition, 77 absorbed parent molecules and nine metabolites in rat serum were rapidly characterized by UHPLC-ESI-Q-TOF/MS. The network pharmacology method was used to predict the active components, corresponding therapeutic targets, and related pathways of WFC in the treatment of PLGC. Based on the main compounds in WFC and their metabolites in rat plasma and existing databases, 13 active components, 48 therapeutic targets, and 61 pathways were found to treat PLGC. The results of PLGC experiment in rats showed that WFC could improve the weight of PLGC rats and the histopathological changes of gastric mucosa partly by inhibiting Mitogen-activated protein kinase (MAPK) signaling pathway to increase pepsin secretion. This study offers an applicable approach to identify chemical components, absorbed compounds, and metabolic compounds in WFC, and provides a method to explore bioactive ingredients and action mechanisms of WFC.


INTRODUCTION
Traditional Chinese medicine (TCM) is based on the principle of holism, that is, all the body systems are interconnected. Thus, TCM treatments can be multi-target, multi-link, and with minimal sideeffects in the prevention and treatment of precancerous lesions of gastric cancer (PLGC), among other diseases. The Wei-Fu-Chun (WFC) tablet, a well-known Chinese herbal preparation, is composed of three herbs: P. ginseng (HS: 131 g), C. aurantium (ZQ: 250 g), and I. amethystoides (XCC: 2,500 g). Dosage of these herbs was derived from Chinese Pharmacopoeia, 2020 edition. In the first part of this edition, WFC was mentioned to tonify spleen qi, promote blood circulation and detoxification, and to have been employed to treat PLGC in clinic for many years (Jin et al., 2015). PLGC is a group of diseases with pathological features of intestinal metaplasia or dysplasia based on chronic atrophic gastritis (Correa et al., 1975). PLGC is an important stage in the development of superficial gastritis to gastric cancer. According to the Chinese Cancer Statistics released by the Chinese Cancer Registry, the mortality rate of gastric cancer ranked second in China, following lung cancer. . Therefore, clinically halting the progression of PLGC to gastric cancer or reversing it has been the focus of current academic research. Some studies investigated the mechanisms of WFC to treat gastritis in vitro or in vivo (Zhao et al., 2012;Huang et al., 2014). However, no studies on the action mechanism of WFC in the treatment of PLGC have been conducted so far. With the prominence of network pharmacology in system biology, this distinct and novel approach to the study of complex analytical systems is becoming more widely known and more frequently used in the field of drug research. The role of network pharmacology includes uncovering the functions of TCM, providing scientific evidence for TCM, and establishing TCM as a scientifically-proven field . Here, we attempted to explore the action mechanisms of WFC in the treatment of PLGC using network pharmacology and experimental studies.
In this paper, a reliable and rapid ultra-high performance liquid chromatography coupled with electrospray ionization quadrupole time-of-flight mass spectrometry (UHPLC-ESI-Q-TOF/MS) method was established to profile complex compounds in WFC and describe their absorption behavior and metabolites in plasma samples after oral administration of WFC. The network pharmacology method was used to construct the pharmacologynetwork of the "origins-components-targets-pathways" of WFC to systematically predict the active components, targets, and signaling pathways that may treat PLGC. In addition, animal experiment was used to verify the prediction and clarify the mechanisms of action. This study provides a reference for further research and exploration of pharmacodynamics material basis and mechanisms of WFC.

Animal Handling
Animal Model for Drug Metabolism Experiment in Rats Ten specific pathogen-free male Sprague-Dawley rats (200 g) were provided by the Experimental Animal Center of Shanghai University of Traditional Chinese Medicine. Rats were housed in an animal room (24 ± 2°C, 60 ± 5% relative humidity) with the setting of a 12 h dark/12 h light cycle. Before the experiment, rats were given water and fed standard laboratory food for acclimatization for a week. Rats were fasted for 12 h before the sample collection but still had access to water throughout the experiment. WFC (dissolved in purified water) was orally administered to six rats at a dose of 4.0 g/kg (capsule powder/body weight) twice a day consecutively for 7 days to accumulate as many absorbed components as possible. Four rats were assigned to the blank control group. Whole blood samples (0.5 ml) were collected from the sub-orbital vein and placed in heparinized polythene tubes at 15, 30, 60, 90, and 120 min after the last drug administration. Plasma was separated immediately by centrifugation at 12,000 rpm for 10 min at 4°C. All samples were immediately stored at −80°C until further analysis.

Pharmacological Experiment of Precancerous Lesions of Gastric Cancer Model in Rats
Fifty 7-week-old male Wistar rats (171 ± 10 g, SPF Grade) were obtained from the Experimental Animal Center of Shanghai University of Traditional Chinese Medicine. All rats were supported and observed in research facility under alternating light and dark conditions (12 h:12 h). After 3 days of adaptive feeding, all rats were divided into five groups, normal group (N group), model group (M group), high-dose group (WH group), low-dose group (WL group), and control group (VM group), stratified randomly according to the body weight of 10 rats in each group. Based on literature modeling methods (Lin et al., 2019) and with the exception of the N group rats, all rats were given 100 mg/L methyl nitroguanidine aqueous solution drink and 0.3 g/L ranitidine fodder each day, and alternatively gavage fed 150 g/L sodium chloride solution at 56°C and 30% ethanol at 10 ml/kg. All model rats were starved and had satiation disorder. After 4 weeks, drugs were administered by gavage. Drug dosage was as follows: Rats in WH group and WL group were given WFC at 1.44 g/kg and 0.72 g/kg, respectively, and VM group was given vatacoenayme at 0.6 g/kg each day. Rats in N group were given gastric gavage of 10 ml/kg of normal saline. After fasting for 12 h at the end of the 16th week, all rats were sacrificed for sample collection.

Plasma Sample Preparation
All plasma samples from the experimental rats were combined into one sample so as to eliminate the individual variability in each experiment. To get the whole information of absorbed components and metabolites, equivalent plasma was taken from the five time points to form mixture as the analysis plasma. An aliquot of 3 ml of methanol was added to 1 ml of plasma and vortex-mixed for 3 min to precipitate proteins (PPT). After centrifugation at 12,000 rpm for 10 min, the supernatant was transferred into a clean tube and evaporated to dryness under a gentle stream of nitrogen at room temperature. The residue was re-dissolved in 200 μl initial mobile phase through vortex-mixing and ultrasonic processing. The resulting solution was centrifuged at 12,000 rpm for 10 min, and 5 μl of the supernatant was injected into the LC-MS system for analysis. The blank plasma sample was prepared as the drug-containing sample.

Pepsin Detection and Histopathology Observation
Body weight and autonomic activity of rats were observed. Detection of pepsin was conducted in accordance with the kit instructions of Nanjing Jiancheng biological engineering institute. The gastric tissues of rats were subjected to tissue dehydration and paraffin discharge operations after being fully fixed with formaldehyde. The embedded tissues were cut into 0.04 μm blank tissue slices for HE staining and histomorphologic study of gastric mucosa under microscope (100×). Inflammatory cell infiltration, ulcer formation, glandular atrophy, basement membrane thickening, and dysplasia were observed.

Real-Time Polymerase Chain Reaction
According to operating manual, gastric tissues were extracted by "one-step" Trizol method and concentrations were calculated. Obtained RNA templates were reverse-transcribed into c-DNA using TAKARA RNA Reverse Transcription Kit. Next, RT-qPCR was performed following TOYOBO Amplification Kit instructions. Here, each well contained a PCR reaction volume of 10 μl (SYBR Green master mix 5 μl, forward primer 1 μl, reverse primer 1 μl, c-DNA sample 1 μl, and diethylpyrocarbonate 2 μl). ß-Actin was used as an endogenous control gene. PCR reaction consisted of 45 cycles as follows: 50°C for 2 min → 95°C for 1 min → 95°C for 15 s → 60°C for 1 min → 95°C for 15 s → 60°C for 1 min. 2 −ΔCt represents the expression level of the tested genes relative to the normal group of genes.

Mass Spectrometric Conditions
MS detection was performed using Acquity Synapt G2 Q-TOF tandem mass spectrometer (Waters, United Kingdom) connected to the UHPLC system by an ESI interface and controlled by MassLynx version 4.1 (Waters, United Kingdom). The ESI source was operated in both positive (ESI + ) and negative (ESI − ) ionization modes. The optimized conditions to trigger maximum response of metabolites were listed as follows: capillary voltage, −2.5 kV (ESI − ) or +3 kV (ESI + ); sample cone, −25 V (ESI − ) or +30 V (ESI + ); extraction cone, −4.0 V (ESI − ) or +4.0 V (ESI + ); source temperature, 120°C; desolation temperature, 350°C; cone gas (nitrogen) flow, 50 L/h; and desolvation gas (nitrogen) flow, 600 L/h. Argon was used as collision gas. Leucine-enkephalin (2 ng/ml) was used as the lock mass generating a reference ion at m/z of 554.2615 (ESI − ) or 556.2771 (ESI + ) by a lock spray at 5 μl/min to acquire accurate mass during analysis. Data were collected in a centroid mode. MSE approach was conducted with two scan functions. In function 1, the following parameters were set: m/z 50-1,500; scan duration, 0.3 s; interscan delay, 0.024 s; and collision energy ramp, 4 V. In function 2, the following parameters were set: m/z 50-1,500; scan duration, 0.3 s; interscan delay, 0.024 s; and collision energy ramp, 10-30 V. In MSE, MS, and MS/MS data can be acquired almost simultaneously in a single analytical run. Data acquisition and processing were conducted using Waters MassLynx version 4.1.

Construction of In-House Library
By comprehensive document retrieval, information about compounds in the three crude medicinal materials and WFC prescription was collected to form an in-house library of WFC. The in-house library content was described with respect to the known global phytochemical constituents and the metabolites database, including the English name, structure, molecular formula, characteristic fragment ions, all accurate monoisotopic mass values of the related chemical formula, and original source.

MassLynx Processing Approach for Analysis of Wei-Fu-Chun Phytochemical Constituents, Absorbed Compounds and Metabolites
Mass data processing was carried out using MassLynx 4.1, including extracted ion chromatograms (EIC) using a narrow mass window of 0.01 Da, calculation of EIC with mass errors within 5 ppm, and fraction isotope abundance value of 1.0. In particular, EIC were also applied to distinguish mixed peaks eluted almost at the same retention time. Target chemical structures were analyzed and confirmed based on EIC, accurately measured mass value, fragment behavior, and elution order, which were all compared with the in-house library data. Furthermore, WFC components in drugcontaining samples were also determined by comparing the retention time and mass data with blank samples. By comparing postdose rat serum, pre-dose rat serum, and the extracts of WFC by UHPLC-ESI-Q-TOF/MS, absorbed compound profile of rat serum was obtained and analyzed. During the same retention time, peaks appeared in the drugcontaining sample and the extracts sample of WFC, but were absent in the control sample or the peaks in the drug-containing sample were five folds greater than that in the control sample, which was extracted as absorption compound. These candidates were further collated and culled by serious MS spectra analysis to conform the true absorbed compounds.

Active Chemical Compositions of Wei-Fu-Chun and Potential Targets
According to the results of chemical profiling and metabolic study of WFC, we selected absorbed parent molecules and metabolites in rat serum as candidate compounds. The candidate compounds were paired one-to-one with protein targets using Stitch 5.0 database (http://stitch.embl.de/). We set the parameter to "Homo sapiens" with confidence greater than 0. 4, and deleted candidate compounds without corresponding targets. Active components and corresponding potential targets were obtained.

Identifying Precancerous Lesion of Gastric Cancer Related Targets in Wei-Fu-Chun
"Precancerous lesion of gastric cancer" was used as a keyword to search Genecards (https://www.genecards.org/) and OMIM database (http://www.omim.org/) separately and to retrieve therapeutic targets for PLGC. Precancerous lesions of gastric cancer -associated target proteins were collected.

Targets GO-Enrichment and Pathways Enrichment Analysis
The targets of WFC in the treatment of PLGC corresponding to the selected active components were input into DAVID 6.8 database (https://david.ncifcrf.gov/) for GO enrichment analysis, and in KEGG database (http://www.genome.jp/kegg) for pathway enrichment analysis. The parameters were set to p < 0.5 and FDR < 0.05.

Network Construction and Its Features
We input the active component-target pair obtained from String database (http://string-db.org/cgi/input.pl) into Cytoscape 3.1, set the properties of the active components and targets, selected the most suitable node distribution form, and generated the active component-target interaction map. The Frontiers in Pharmacology | www.frontiersin.org November 2020 | Volume 11 | Article 558471 primary pathway, biological processes, cellular components and molecular functions of p < 0.05 were selected. Visualize the network of "origins-components-targets-pathways" with the Merge feature of Cytoscape 3.1.

Statistical Analysis
SPSS 21.0 statistical software was used for data statistics and analysis. Data measurement was expressed as mean ± standard errors (±SEM). One-way ANOVA was used for comparison among groups, and LSD-t test for further two-paired comparison. A p-value <0.05 indicated a significant statistical difference between two groups.

Sample Acquisition and LC-MS Conditions of Developed Method
Nowadays, UHPLC-ESI-Q-TOF/MS capable of high resolution, high sensitivity and high accuracy have been proven to be an excellent technique for quantitative and qualitative analysis of multi-components and metabolites in complex mixture especially in TCM formulas. Here, UHPLC-ESI-Q-TOF/MS has shown superior performance, in terms of high mass resolution and accurate mass measurement, fast scan speed and wide dynamic range of mass analyzer, as well as efficient separation capability and speed of UHPLC. The acquisition of dependable biological samples for UHPLC-ESI-Q-TOF/MS analysis played a crucial part in the comprehensive in vivo screening of WFC compounds. Primarily, all the possible components should be contained in the collected crude samples and have adequate levels to be detected, taken into account that drug administration method and collection time influence sample content (Zhang et al., 2017). LC-MS conditions were optimized to enable samples to obtain the best instrumental performance. Both positive and negative ion modes were employed to screen as many potential compounds as possible, but the negative ion mode provided higher signal intensity and the ability to detect more peak signals. Different mobile phase systems and gradient programs were emphasized and investigated to achieve good ionization and separation behavior. A mixture of 0.1% aqueous formic acid and acetonitrile was nally chosen as the preferred mobile phase. This was because acetonitrile had stronger eluting power and the retention behavior of flavonoids on the reversedphase column was easily affected by pH (increasing pH enhanced the ionization of flavonoids and could reduce the retention in a reversed-phase separation). Thus, small amounts of formic acids were normally included in the solvent to suppress ionization of phenolic or carboxylic groups, hence improving the resolution and reproducibility of each separation. Formic acid buffer have been used as part of the mobile phase for optimizing the analysis time and enhancing separation (Wang et al., 2019). Many components, especially flavonoids and ginsenosides, gave prominent molecular adductive ions when formic acid was used. The results indicated that the proposed method was acceptable and adequate for the comprehensive study of WFC.

Analysis of Wei-Fu-Chun Phytochemistry Components, Absorbed Components, and Metabolites by UHPLC-ESI-Q-TOF/MS
The base peak intensity (BPI) chromatograms of WFC extract at negative and positive ion modes are shown in Figure 1 and Table 2. A total of 178 compounds were identified or tentatively characterized, including 70 terpenes (56 diterpenoids, 12 triterpene, and 2 sesquiterpenes), 51 flavonoids, 33 saponin, six phenylpropanoids, four lignans, three coumarins, three organic acids, two fatty acid, one quinones, one sterol, and four unknown compounds. Structurally related compounds shared analogous MS response and fragment behavior. All compounds were characterized based on their ECs, MS data, and retention behavior with the help of the constructed in-house library and additional literature. The library made the workflow significantly more effective when one was familiar with the detailed information of these extensively investigated molecules and allowed the characterization of these compounds even when reference standards were not available. The calculated correct monoisotopic mass values of quasimolecular ions and rational fragment ions were especially critical to screen the various compounds fast and firmly. There were many isomers inevitably existing in the complex natural compounds and the tiny difference of the mass spectra made them difficult to distinguish. In this case, the electronic effect and their hydrophilicity should be considered (Li et al., 2015). The detailed identified information, including chemical formula, observed mass values of quasi-molecular ions, mass error, and botanical source, is listed in Table 2. A chemical library, including 569 compounds in WFC, was established. By matching against the chemical library, 178 ingredients in WFC were identified. Among them, 93 compounds belonged to XCC, 51 to ZQ, 31 to HS. Three compounds were unknown, meaning that the herb(s) from which they came remained undetermined. In addition, 77 absorbed parent molecules and nine metabolites in rat serum were characterized by UHPLC-ESI-Q-TOF/MS.

Identification of Diterpenoids, Triterpene, and Sesquiterpenes
Diterpenoids are the major bioactive constituents of XCC. About 500 new diterpenoids (mainly ent-kauranoids) with different oxygenations and cleavage patterns have been isolated and characterized from plants of the genus Isodon. In our study, 70 terpenes, including 56 diterpenoids, 12 triterpene, and two sesquiterpenes, were detected and identified in WFC, by means of UHPLC-ESI-Q-TOF/MS in both negative and positive mode according to the literatures (Zhou et al., 2009;Jin et al., 2010). The accurate mass measurements and elemental compositions of molecular ions and main product ion were shown in Table 2. In this study, all terpenes from XCC were ionized as deprotonated molecules [M − H] − in negative mode, however, only part of the terpenes were detected in positive mode.

Identification of Flavonoids
Fifty-one flavonoids and their glycosides in WFC (39 from ZQ, 12 from XCC) were detected according to the literature (Fabre et al., 2001;Shi et al., 2007;Zhang et al., 2012), most of them having a common structure of C6-C3-C6. Their cleavage regularity in extracts has been well described. Simply, for aglycones, the main MS/MS behavior involved RDA fragmentation pathway and losses of small neutral molecules and radicals from Thus, 32 and 38 were deemed narirutin and naringin based on the fragmentations and appearance time. Compounds 42 and 52 were identified as hesperidin and neohesperidin, as both lost a hexose residue (308 Da) at negative and positive conditions (Li et al., 2004). Compound 57 showed identical [M − H] − ion at m/z 463, and fragmentation at m/z 301 and 286, suggesting that compound 57 had a hexose residue (162 Da), and CH 3 (15 Da) in its aglycone. Compound 57 was identified as hesperetin-7-O-glucoside. Other flavonoids were tentatively identified based on the positive and negative parent ion and previous literature report.

Identification of Saponin
Thirty-two saponins (31 from HS) were detected and identified in WFC, by matching the empirical molecular formula and fragment ions with reported MS data of saponins in HS, ZQ, and XCC (Liu et al., 2006;Dan et al., 2009;Zhang et al., 2012). The detailed information was shown in Table 2

Identification of Organic Acids, Fatty Acid, Quinones, Sterol, and Unknown Compounds
Three organic acid, two fatty acid, one quinones, one sterol, and four unknown compounds were identified.

UHPLC-ESI-Q-TOF/MS Analysis of Compounds in Wei-Fu-Chun Absorbed into Blood Circulation
The absorbed compounds were explored based on the hypothesis that the active constituents were those absorbed into tissue (Zhao et al., 2010). By comparing post-dose rat serum, pre-dose rat serum, and the extracts of WFC by UHPLC-ESI-Q-TOF/MS, an absorbed compounds profile of rat serum was obtained and analyzed. The 77 parent molecules in the positive-and negative-ion mode were summarized in Figure 2 and Table 2 and marked with an asterisk (*). Of these 77 parent molecules, there were 24 diterpenoids (23 from XCC, one from ZQ), 20 flavonoids (15 from ZQ, five from XCC), 17 saponins (all are ginsenoside from HS), five triterpenes (three from XCC, two from ZQ), two phenylpropanoids (both are from XCC), two organic acids (both are from XCC), one sesquiterpenes (from ZQ), one sterol (from XCC), one fatty acid (from XCC), and two unknown compounds. Taken together, the most absorbed compounds were terpene from XCC, flavonoids from ZQ, and saponins from HS.

Tentative Identification of the Wei-Fu-Chun Metabolites in Rats
Some drug components could be further metabolized by a variety of metabolic enzymes in the body and form phase I or phase II metabolites, which are highly related to the curative effect and drug elimination. Drug metabolites usually kept the core structure of the parent drug after biotransformation, hence the obtained candidate metabolites were further identified by comparing the change in molecular mass (△M), the retention time, and MS 2 spectral with their parent drugs. Metabolites of all 178 compounds were tentatively predicted in rat plasma according to their molecular weights and literature reports; however, only the metabolites of limonin, ginsenoside Rg1, ginsenoside Rb3, oridonin were found in our study. M1 eluted at 1.07 min was observed at m/z 455.1891, which is 16 Da (−O) lower than limonin. M2 eluted at 1.08 min and the positive ion was observed at m/z 485.1989, which was 14 Da (+CH 2 ) higher than limonin. The peak elution of M4 was at 4.34 min and the positive ion was observed at m/z 489.2314, which was 18 Da (A-ring lactone) higher than that of the parent drug limonin. The two metabolites (M3 and M5) of ginsenoside Rg1 were observed at m/z 859.4590 and 861.4748 in negative mode, which were 14 Da (+O-2H) and 16 Da (+O) higher than limonin . Oridonin + OH (M9) and +2OH (M8) compounds were observed at m/z 379.1540 and 395.1656.

Identification of Active Chemical Compositions, Candidate Targets and Biological Processes for Wei-Fu-Chun Against Precancerous Lesions of Gastric Cancer
The efficacy of TCM is based on the overall regulation of multiple active ingredients acting on different disease-related targets through multiple pathways (Li and Zhang, 2013). Therefore, we collected 86 candidate compounds (Tables 2, 3) and 105 potential targets. 13 active components and 48 protein targets to treat PLGC were then obtained. Detailed information of these therapeutic targets was described in Tables 4, 5 separately.

Enrichment Analysis of Candidate Targets for Wei-Fu-Chun Against Precancerous Lesions of Gastric Cancer
To further explore the possible functions of the 48 therapeutic targets and reveal the relationship between active components and their underlying targets in PLGC, we performed pathway enrichment analysis on PLGC-related protein targets and obtained the main related signaling pathways of target proteins (p-value < 0.01). The 61 anti-PLGC pathways of WFC were listed in Table 6. These therapeutic targets were mainly distributed in PI3K/Akt pathway, MAPK pathway, vascular endothelial growth factor (VEGF) pathway, hypoxiainducible factor-1 (HIF-1) pathway, and tumor necrosis factor (TNF) pathway, and FoxO pathway. In addition, According to the GO enrichment analysis results, we found that the functions of therapeutic targets were involved in multiple biological processes such as peptide chain serine phosphorylation, RNA polymerase II promoter transcription, cellular hypoxia, apoptosis, vascular endothelial cell migration, macrophage differentiation, vascular endothelial growth factor receptor, chemokine biosynthesis, cell proliferation, and nitric oxide biosynthesis (Figure 3).

Construction of Pharmacology-Network for Wei-Fu-Chun Against Precancerous Lesions of Gastric Cancer
Using the Cytoscape 3.1 software, we constructed an "originscomponents-targets-pathways" pharmacology network of WFC ( Figure 4). This network depicted the relationship between 61 pathways, 48 therapeutic targets, 13 active components and corresponding plant materials. The network also consisted of 64 nodes and 75 edges, of which Ginsenoside Rb 1 (C3, degree 7), Naringin (C7, degree 9), Hesperidin (C8, degree 8), Hesperetin (C9, degree 7), and Oridonin (C11, degree 8) had high degrees and were centrally located in the network, suggesting that these active compounds may be the key components of WFC in the treatment of PLGC.

Model Test Results of Precancerous Lesions of Gastric Cancer
The speed of weight gain in the model group was much lower than that in the normal group. At the end of the fourth week, the weight of the high dose group, the low dose group, and the Vitacoenzyme group showed a rising trend after the simultaneous modeling and administration of the high dose group, the low dose group, and the vitamin enzyme group ( Figure 5A). The increase rate of the high dose group was higher than that of the other two groups. The weight of the model group increased and then decreased until the end of the model. Compared with the normal group, there was a significant statistical difference in the activity value of pepsin in the normal group and the model group (p < 0.05). Compared with the model group, the activity value of pepsin in the high-dose group was significantly different (p < 0.05). There was no significant difference in protease activity  between the low dose group and the model group. There was also no significant difference in pepsin activity between the high-dose group and the normal group (p > 0.05) ( Figure 5B). The histopathology of gastric mucosa was demonstrated by hematoxylin eosin staining. Atrophic glands, intestinal metaplasia and lymphocyte infiltration were shown in the model group, and compared with the model group, the above pathological manifestations were alleviated in varying degrees ( Figure 5C). RT-PCR results showed that mRNA expression of VEGF, FOXO4, AKT, TP53, FAS, MAPK8, MAPK11, and MAPK14 in the model group was significantly up-regulated compared with that in the N group (p < 0.05)., while mRNA expression of interleukin-10 (interleukin-10, il-10) was down-regulated (p < 0.01). mRNA expressions of VEGF, FOXO4, AKT, TP53, FAS, MAPK8, MAPK11, and MAPK14 were all significantly down-regulated in the WFC high dose group (p < 0.05) ( Figure 6).

DISCUSSION
WFC, an herbal prescription with three medicines, may contain thousands of compounds, however, naringin is the only maker compound of WFC in 2015 edition of "Chinese Pharmacopoeia." In this study, we hypothesized that unilateral factors and single targets are insufficient to demonstrate the complex mechanisms of WFC. The network pharmacology method used in this study is a novel methodology based on the construction of multilayer networks of disease phenotype-gene-drug to predict drug targets in a holistic manner and promote efficient drug discovery (Liu and Sun, 2012). This method represents a breakthrough in comparison with the traditional herbal medicine research pattern "gene target-disease" and initiates the new pattern of "multiple genes multiple targets-complex diseases" (Hopkins, 2008;Ma et al., 2016). With this method, we proved that XCC was the critical ingredient involved in the treatment of PLGC, which corresponded to the constituents of WFC with 86.8% weight of XCC in WFC. HS was also a major herb that regulated PLGC. Moreover, WFC is a multiple-component complex system. (Zhang et al., 2012) unambiguously identified or tentatively characterized 46 components in WFC tablet, including 26 saponins, 10 flavonoids, and 10 other compounds. However, it is not enough to understand the chemical components of WFC, which makes it rather difficult to define the functions of this herbal medicine from material basis and chemical properties. Therefore, a chemical fingerprint analysis of WFC was necessary. Compared with the material basis of the three herbs in WFC, a chemical fingerprint analysis of WFC was fully carried out by means of UHPLC-ESI-Q-TOF/MS. Finally, a total of 178 compounds were identified in WFC, including 93 compounds (70 terpene) originally from I. amethystoides, 51 compounds from Fructus aurantii, 31 compounds from red ginseng, which basically demonstrated the material basis in WFC. However, the main components' content and dose-effect relationship should be further investigated.
In network pharmacology studies, there were 86 candidate compounds (77 absorbed components and nine metabolites) and 105 potential targets. Thirteen active components and 48 protein targets were selected to explore the effect of WFC against PLGC. The results showed that PI3K/Akt pathway, MAPK pathway, VEGF pathway, HIF-1 pathway, TNF pathway, and FOXO pathway may be involved as the anti-PLGC pathways of WFC. In addition, ginsenoside Rb 1 , naringin, hesperidin, hesperetin, and oridonin may be the key components of WFC in the treatment of PLGC. Ginsenoside Rb1was shown to exert antiinflammatory effects in Modulating TLR-4 dimerization and NF-kB/MAPKs signaling pathways (Gao et al., 2020). Ginsenoside Rb 1 also enhanced the phagocytic capacity of macrophages for  bacteria via activation of the p38/Akt pathway, which may be a useful pharmacological adjuvant for the treatment of bacterial infections in clinically relevant conditions (Xin et al., 2019). Moreover, naringin induced autophagy-mediated growth inhibition by downregulating the PI3K/Akt/mTOR cascade via activation of MAPK pathways in AGS cancer cells (Raha et al., 2015). As for hesperetin, it induced apoptosis in human glioblastoma cells via p38 MAPK activation . Finally, oridonin's anticancer effects on colon cancer were mediated via BMP7/p38 MAPK/p53 signaling (Liu et al., 2018). Overall, this suggested that MAPK pathway, PI3K/Akt pathway, and p38/Akt pathway may be the key pathways of WFC treating PLGC.
To verify the network pharmacology prediction results, in vivo rat experiments were carried out. Pepsin is formed by pepsinogen stimulated by gastric acid. Pepsin is mainly secreted by the main cells. Atrophy of gastric glands and metaplasia of intestinal epithelium reduce the main cells. From superficial gastritis, atrophic gastritis/dysplasia to gastric cancer, pepsinogen in pathological tissues decreases and then pepsin activity decreases. In chronic atrophic gastritis patients with or without intestinal metaplasia, pepsinogen I often decreased (Terasawa et al., 2014). In this experiment, WFC restored pepsin activity in the rat model of PLGC.
MAPK is the main transmitter of intracellular and extracellular signals. P38 and JNK are the important parts of its four major subgroups. JNK pathway and p38 pathway mediate the transmission of cytokines and inflammatory mediators and participate in cell cycle, apoptosis, and migration. This process plays an important role in the development of precancerous lesions to gastric cancer. JNK mainly exists in the cytoplasm and accumulates rapidly and significantly in the nucleus after being activated by the superior kinase. The activation of transcription factors in the nucleus includes TP53, c-Jun and so on (Liu and Lin, 2005), and then produces biological effects. The positive rate of JNK expression in gastric cancer was 75%, and it positively correlated with the size of the tumor and the early and late stage of the tumor (Wang, 2009). Helicobacter pylori can stimulate the invasion of AGS gastric cancer cells through JNK signaling pathway (Díaz-Serrano et al., 2018). Inhibition of JNK pathway can enhance the antitumor effect of trail on MGC803 gastric cancer cells (Liu FIGURE 3 | Biological processes regulated by WFC to treat PLGC. The count of each biological process was shown on the right side of the bar. The p-value of each biological process was less than 0.05. FDR of each biological process was less than 0.05. et al., 2011). P38 has five isomers, p38 α (p38), p38 ß 1, p38 ß 2, p38 γ, p38 δ. 38 α, and p38 ß tissues are widespread. P38 pathway plays an important role in cell cycle regulation and can induce cell cycle. The inhibition of p38 pathway may be one of the mechanisms of synergistic antitumor effect of Adriamycin and monomer PA-2 (Liang et al., 2020). In gastric cancer patients with high expression of lncRNA-aoc4p, inhibiting the expression of lncRNA-aoc4p could reduce the expression level of JNK and p38 protein and inhibit cell proliferation, migration, and invasion (Qu et al., 2019). FOXO4, as a member of the FOXO family of transcription factors, is regulated by microRNA in a variety of cancer cells, and its abnormal expression is closely related to gastrointestinal tumors (Gross et al., 2008;Liu et al., 2020). The current study further proved the therapeutic effect of WFC on PLGC in rats, and preliminarily explored and verified the results of network pharmacology. The detection results of MAPK pathway genes p38 α, p38 β, JNK and its upstream and downstream factors TP53 and VEGF α, as well as FOXO4 and AKT genes showed consistency. We believe that MAPK pathway is involved in the mechanism of action of WFC on PLGC, which needs to be further explored.

CONCLUSION
In conclusion, a simple and reliable UHPLC-ESI-Q-TOF/MS technique was established to profile complex compounds in WFC and describe their absorption behavior in plasma samples after oral administration of WFC. A total of 178 compounds were identified. In addition, 77 absorbed compounds of parent molecules in WFC were detected. Among these compounds, the most absorbed ones were terpenes from XCC, flavonoids from ZQ and saponins from HS. The current study supplemented previous WFC research, and MS data contributed to the identification of allied natural compounds. The screening of in vivo absorbed and metabolic compounds provided constructive material basis for further research on the pharmacology and curative mechanisms of WFC. Moreover, the network pharmacology method was used to predict the active components, corresponding therapeutic targets, and related pathways of WFC in the treatment of PLGC. Finally, based on the major compounds of WFC and their metabolites in rat plasma and existing databases, 13 active components, 48 therapeutic targets, and 61 pathways were found to act against PLGC. The results from rat experiment showed that WFC could improve the FIGURE 4 | Pharmacology-network of the "origins-components-targets-pathways" regulated by WFC. The yellow rectangles represent Chinese herbal medicines, while the purple rhombuses represent active components of WFC. The blue ellipse represents metabolites of ginsenoside Rg1. The red rectangles represent target proteins, and the green ellipses represent pathways in Table 6.
Frontiers in Pharmacology | www.frontiersin.org November 2020 | Volume 11 | Article 558471 weight of PLGC rats and the gastric mucosa histopathological changes partly by inhibiting MAPK signaling pathway to increase pepsin secretion. This study offered an applicable approach for the identification of chemical components, absorbed compounds, and metabolic compounds of WFC, and provided a method to explore bioactive ingredients and action mechanisms of WFC.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

AUTHOR CONTRIBUTIONS
HW, RW, and DX contributed equally. HW conducted the experimental part of the main component analysis of WFC. Network pharmacology was mainly analyzed by RW, DX performed most of the pharmacological experiments, analyzed the data, and participated in the manuscript draft. LD and XL helped complete animal feeding and analysis of the main components in plasma of Weifuchun tablets. YB, XC, and BN polished the manuscript. SW, KL, and WC monitored the drug quality and helped complete the data collation. GY and MS directed the study, and drafted and finalized the manuscript. All authors read and approved the final manuscript.  Frontiers in Pharmacology | www.frontiersin.org November 2020 | Volume 11 | Article 558471