Systems Pharmacology Study of the Anti-Liver Injury Mechanism of Citri Reticulatae Pericarpium

Liver diseases are mostly triggered by oxidative stress and inflammation, leading to extracellular matrix overproduction and prone to develop into liver fibrosis, cirrhosis and hepatocellular carcinoma. Liver injury (LI) refers to various pathogenic factors leading to the destruction of stem cells that then affect the liver’s normal function, causing a series of symptoms and abnormal liver function indicators. Citri Reticulatae Pericarpium (CRP) is one of the most commonly used traditional Chinese medicines; it contains flavonoids including hesperidin, nobiletin, and tangeretin. CRP has antibacterial, antioxidant, and antitumor effects that reduce cholesterol, prevent atherosclerosis and decrease LI. Here we analyzed the components of CRP and their targets of action in LI treatment and assessed the relationships between them using a systems pharmacology approach. Twenty-five active ingredients against LI were selected based on ultra-performance liquid chromatography-quadrupole/time-of-flight mass spectrometry results and databases. The drug targets and disease-related targets were predicted. The 117 common targets were used to construct a protein-protein interaction network. We identified 1719 gene ontology items in LI treatment, including 1,525 biological processes, 55 cellular components, and 139 molecular functions. These correlated with 49 Kyoto Encyclopedia of Genes and Genomes pathways. These findings suggest that CRP may counteract LI by affecting apoptotic, inflammatory, and energy metabolism modules. In vitro experiments suggested that the mechanism may involve hesperidin and naringenin acting on CASP3, BAX, and BCL2 to affect the apoptosis pathway, attenuating liver fibrosis. Naringenin significantly inhibited AKT1 phosphorylation, which in turn mediated activation of the phosphoinositide 3-kinase-Akt signaling pathways against LI. This study provides a reference for systematically exploring the mechanism of CRP’s anti-LI action and is also expands of the application of systems pharmacology in the study of traditional Chinese medicine.


INTRODUCTION
The pathogenetic events leading to liver injury (LI) include oxidative damage, apoptosis, disordered lipid metabolism, and mitochondrial dysfunction. LI refers to one or more results of organ damage, inflammation, bleeding, infection, or other symptoms. Alcohol use is a major predisposing factor for LI (Yang et al., 2020). The annual proportion of alcohol-induced LI has been increasing (Tsutsumi multi-directional pharmacology, bioinformatics, and computer science to construct multi-level "disease-gene-target-drug" networks. This study analyzed and elucidated the links between CRP and LI using a modern systems pharmacology approach and in vitro experiments ( Figure 1).

Compounds Collected From UPLC-Q-TOF-MS/MS Analysis
Identification technology of compounds and metabolites in CRP has developed rapidly, and mass spectrometry identification is a common technical method (Zheng et al., 2013;Zeng et al., 2017;Li et al., 2019a). We added an appropriate amount of methanol to prepare standard solutions of 50-100 μg/ml and accurately weighed CRP powder (1.0 g), placed it in a conical flask containing 80 ml methanol, mixed it well, treated it with an ultrasound machine for 30 min, passed the solution through filter paper, then increased the volume to 100 ml, and finally filtered the solution through a 0.22 μm microporous membrane to obtain the CRP extraction solution. Ultra-Performance Liquid Chromatography-Quadrupole-Time-of-Flight Tandem Mass Spectrometry (UPLC-Q-TOF-MS/MS) conditions were as follows: chromatographic separation was performed at a flow rate of 0.25 ml/min at 40°C, and elution was performed with a linear gradient program according to various periods. The mobile phase system included solvent A (100% acetonitrile, v/v) and solvent B (0.01% formic acid in water, v/v). The process was solvent A (5-12%) for 5 min, (12-21%) for 10 min, (21-35%) for 10 min, (35-45%) for 11 min, (45-48%) for 4 min, (48-60%) for 1 min, (60-85%) for 9 min, (85-95%) for 1 min, (95%) for 2 min, (95-5%) for 1 min, and isocratic elution at 5% for 2.9 min. The ion source gases 1 and 2 were both set to 50 psi, curtain gas was set to 40 psi, ion spray voltage floating was set to 5500 V in the positive mode and 4500 V in the negative mode, ion source temperature was 500°C, collision energy was 40 V, collision energy spread was 15 V, declustering potential was 100 V, and nitrogen was used as a nebulizer and auxiliary gas. Samples were analyzed in both positive and negative ionization modes with a scanning mass-to-charge (m/z) range from 50 to 1,000. Data were collected in information-dependent acquisition mode and were analyzed using the Analyst TF 1.6 data acquisition workstation and PeakView ® 1.2 software. We compared the fragment ion information obtained for various compounds with that of the 22 standards and identified several CRP compounds. Also, we studied and analyzed the literature documenting fragment ion information of CRP components in detail carefully, matched it with our information of multiple compounds, and identified additional compounds.

Compounds Collected From Databases
To collect drug ingredients more thoroughly, we used The Encyclopedia of Traditional Chinese Medicine (http://www. tcmip.cn/ETCM), the Traditional Chinese Medicine Systems Pharmacology database and Analysis Platform (TCMSP, http://lsp.nwu.edu.cn/tcmsp.php), and the TCM database Taiwan (http://tcm.cmu.edu.tw). All three online databases were searched for compounds, using the Chinese name "ChenPi" and the Latin name Citrus reticulata Blanco.

Compound Identification Analysis
We integrated the collected compounds and established a CRP chemical composition. First, the concept of oral bioavailability (OB) was introduced. OB refers to the percentage of unmodified drug entering the circulatory system after oral administration, representing drug utilization efficiency. Higher OB correlates with a higher possibility of clinical usefulness (Lv W. et al., 2020). Drug-likeness (DL) refers to the potential of a compound to become a drug (Huang et al., 2017). Molecules with OB ≥ 30% or DL ≥ 0.18 are considered to have better pharmacological effects . We referred to this principle to select active ingredients for the subsequent step of the analysis. Considering the special biological activities of some compounds, we included them in the final candidate component database.

Identification of the Related Targets and Gene Symbols of CRP Compounds
We used information in the TCMSP (http://lsp.nwu.edu.cn/ tcmsp.php) and Swiss Target prediction (STP, https://www. swisstargetprediction.ch/) to extract protein targets for each CRP component (Mu et al., 2020). After removing redundant information, the targets that could interact with each CRP component were retained.

Acquisition of Liver Injury Gene Targets
The gene targets for LI was obtained from the GeneCards database (https://www.genecards.org/, version 4.9.0) and the Online Mendelian Inheritance in Man database (OMIM, http://www.omim.org/, updated on February 28, 2019) (Mu et al., 2020;Xu et al., 2020). We performed keyword searches using "liver injury" and carefully screened the results. Finally, we obtained genes associated with LI-related diseases.

Drug-Compounds-Genes-Disease Network Construction
To obtain overlapping targets, we first cross-contrasted CRPrelated targets with LI-related targets. We then built a complex information network (D-C-G-D) based on the interactions among drugs (CRP), compounds, gene symbols, and disease (LI) using Cytoscape software (version 3.8.0) and drew a schematic diagram.

Protein-Protein Interaction Network Construction
PPI data were obtained from the STRING database (https:// string-db.org/, Version 11.0, updated on January 19, 2019). The interaction relationships provided by STRING are based on the confidence score and can be used to filter and assess functional genomics data (Tao et al., 2020). The organism species was set to Homo sapiens (Human), and the merged gene symbols were analyzed. We also used the BioGPS database (https://biogps. org) to identify proteins' higher expression in some major organs (Gu et al., 2020). The heat map was generated using Heml software (version 1.0).

Gene Ontology and Kyoto Encyclopedia of Genes and Genomes Pathway Enrichment
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using R software (version 4.0.2). We used the R package "BiocGenerics," a classic and practical tool specially designed for biologists to conduct an in-depth analysis of these targets (Yu et al., 2012). All enriched entries were determined using the Bioconductor database (http:// bioconductor.org/).

Computational Validation of Ingredients-Targets Interactions
Molecular docking is a drug design method that considers receptors' characteristics and the mode of interaction between receptors and drug molecules . We docked a variety of active components and key targets (BAX, BCL2, and caspase-3 [CASP3]) for a total of six component-target interactions. We obtained data from the RCSB Protein Data Bank (PDB, www.rcsb.org); the X-ray crystal structures of the key targets BAX, BCL2, and CASP3 were obtained and confirmed concerning the relevant literature; their PDB IDs were 1F16, 1G5M, and 5IBC, respectively. After processing targets and compounds using PyMOL software (version 1.3), docking work was performed using AUTODOCK VINA software (Version 1.1.2, Scripps Research, San Diego, CA, United States). The required input files for the AutoDock program were prepared using AutoDock tools.

CCK-8 Assay for Cell Viability
Logarithmic growth phase cells were plated in 96-well plates (5 × 10 3 cells per well) and cultured overnight in a 37°C, 5% CO 2 incubator. The following day, various hesperidin and naringenin (0, 25, 50, 75, 100, and 125 μM) were added to fresh medium. After 12 h, the model and administration groups were incubated with 5 ng/ml transforming growth factor (TGF)-β1 for 24 h, and then 10 μL CCK-8 assay solution (Solarbio) was added to each well. A microplate reader (Nanjing Detie Experimental Equipment Co., Ltd. Nanjing, China) was used to measure the optical density (OD) value around 450 nm. Cell survival was calculated as: Ab AC × 100% (Ab and Ac stand for absorbance and absorbance of the control, respectively).

Statistical Analyses
The significance of results was determined based on one-way analysis of variance using Prism 8.0.1 software (Graph Pad Inc. San Diego, CA, United States). All experiments were performed in triplicate, and data are presented as mean ± SD. Differences were considered significant at p < 0.05.

Screening of Active Compounds Collected From UPLC-Q-TOF-MS/MS Analysis
We showed a typical total ion chromatogram of alcohol-soluble components extracted from CRP (Figures 2A,B) and the mass spectrometric cleavage law of hesperidin and nobiletin ( Figures  2C,D). By sampling the primary mass spectrum, we determined each component's relative molecular weight and then obtained fragmentation information based on the secondary mass spectrum. Based on the retention time and mass spectrometric information of the standard material and the compounds reported in the literature, we identified a total of 38 compounds. Twenty-two of these (8, 9, 11, 12, 14, 16, 17, 18, 19, 21, 22, 24, 25, 26, 27, 28, 29, 31, 32, 34, 36, and 37) were identified through careful comparison of information between the samples and standard pure substances. The details are shown in Supplementary Table S1. Through the comparison of the multi-level fragment ion information recorded in the relevant literature and the analysis of the fragmentation law, the other 16 components were identified. Table 1 presents the details of the 38 compounds.

Compounds Collected From Databases
We searched online databases commonly used for TCM ingredients, The Encyclopedia of Traditional Chinese Medicine (http://www.tcmip.cn/ETCM), Traditional Chinese Medicine Systems Pharmacology database and Analysis Platform (http:// lsp.nwu.edu.cn/tcmsp.php), and the TCM database Taiwan (http://tcm.cmu.edu.tw) to produce a more comprehensive compound list Zhu and Hou, 2020). We collected a total of 65 compounds from these databases as candidates for CRP studies (Supplementary Table S2).

Compound Identification Analysis
We collated the results of previous work, combined the components obtained using UPLC/Q-TOF-MS/MS analysis and all CRP components identified in online databases. After removing repeated items, we established a complete CRP internal compound composition library. To screen out the active ingredients of CRP, the classical absorption, distribution, metabolism, and excretion (ADME) parameters of OB and DL were used to screen them. We found five ingredients that met these conditions; however, this result was far below our expectations. CRP contains a large number of flavonoids, including hesperidin and naringin (Dong et al., 2010). Flavonoid glycosides are more easily absorbed after hydrolysis into flavonoid aglycones (Hostetler et al., 2012), and hesperidin is one of the typical constituents (Nectoux et al., 2019). The OB of hesperidin was 13.33%, which was <30%; however, it is an active compound that is significantly effective against LI (Adefegha et al., 2017;Elhelaly et al., 2019). Narirutin has an OB of 8.15% and a DL of 0.75. However, it alleviates LI by regulating protein   phosphorylation in the mitogen-activated protein kinase (MAPK) pathway, which may be related to lipid metabolism (Jianping et al., 2019;Gao et al., 2020). Tangeretin has an OB of 21.38% and a DL of 0.43, but it can ameliorate oxygen-glucose deprivation-induced LI through the JNK signaling pathway . Rosmarinic acid has an OB of 1.38% and a DL of 0.35; it can activate the AMP-activated protein kinase (AMPK) pathway and inhibit inflammation through nuclear factor changes erythroid 2-related factor and nuclear factor-κB signaling and ameliorates obesity and LI (Gyhye et al., 2019). Another mechanism by which compounds counteract LI is antioxidation, and the candidate p-Cymen-8-ol has both antioxidant effects and antibacterial activity (Pires et al., 2006;Santana et al., 2009). Although some components have relatively low kinetic values, they have excellent biological activity; therefore, we still consider them active CRP components. For these reasons, we propose that as long as the candidate components in CRP have good biological activity and intersect with LI's target, they can be considered active CRP compounds for this indication. We selected 25 components as the final active compounds in the candidate compound library ( Table 2).

Identification of Related Targets and Gene Symbols of Compounds in CRP
After we collected the relevant proteins from the TCMSP and STP databases, they were converted in the UniProt database, and redundant terms were removed to obtain 25 components in CRP and the 126 known associated target symbols (Supplementary Table S3).

Acquisition of Known Therapeutic Gene Targets for Liver Injury
A subset of the LI-related targets was obtained from the GeneCards database. We obtained 7372 known LI target symbols. Also, 92 known therapeutic target data for LI were obtained from OMIM. After eliminating redundancies, a total of 7442 known therapeutic targets for LI were collected (Supplementary Table S4).

Drug-Compounds-Genes-Disease Network Construction
We contrasted the obtained drug targets with disease-related genes to obtain their common cross genes. We constructed Venn diagrams ( Figure 3A) with 117 overlaps of 7442 disease gene symbols and 126 drug-gene symbols. To intuitively demonstrate how CRP counteracts LI, we constructed a D-C-G-D network using Cytoscape software ( Figure 3B). D-C-G-D details are in Supplementary Table S5.

Protein-Protein Interaction Network Construction
An action relationship map between genes was obtained after 117 overlapping genes were entered into the STRING online database ( Figure 3C). These genes may be critical targets for LI treatment using CRP. The gene names are listed in Supplementary Table  S6. After removing two free genes, we calculated the topological indices DC, BC, and CC for 115 genes. Degree (DC) indicates how much a gene is linked to other genes, suggesting that the gene may be necessary for the process of CRP counteracting LI. Detailed information about genes and topological indices is provided in Supplementary Table S7. We calculated the mean value of the 115 overlapping gene degrees to be 29.42. All genes with degree values >29.42 were selected, resulting in 25 core targets for interaction networks with other genes. After processing these associated targets using Cytoscape software, the inner loop of Figure 3D shows 25 genes with higher DC values. The deeper red color and large nodes indicate stronger correlations, and these may be critical genes for the CRP treatment of LI. To investigate LI and other major organs' connection, we considered these core targets' expression in various organs. According to the presented results of the heat map ( Figure 3E), the 25 core targets' expression was generally similar in the main organs. For example, AKT1 and MAPK3 were highly expressed in the liver, heart, lung, kidney, small intestine, and colon, while expression levels of PPARG and IL4 were low in these organs. This finding suggests that, in addition to the liver, LI is also closely related to other organs, consistent with the fact that TCM considers the human body as a whole. The specific expression of these core targets in various organs is shown in Supplementary Table S8.

Gene Ontology and Kyoto Encyclopedia of Genes and Genomes Pathway Enrichment
To better summarize their specific functions, we performed GO enrichment analysis of 117 consensus genes and classified and elaborated them according to three modules: biological process (BP), cellular component (CC), and molecular function (MF), including 1525 GO terms for BP terms, 55 GO terms for CC terms, and 139 GO terms for MF terms, for a total of 1719 enriched GO terms. Based on the p. adjust value (p < 0.01), we selected the first ten terms from small to large for the enrichment results of BP, CC, and MF ( Figure 3F). The BP analysis results revealed essential ways in which genes participate in bodily processes. These processes involve responses to nutrient levels (GO: 0031667), aging (GO: 0007568), responses to lipopolysaccharide (GO: 0032496), responses to a molecule of bacterial origin (GO: 0002237), and responses to metal ion (GO: 0010038). CC analysis then identified the sites where these intersection genes function intracellularly and extracellularly, primarily involving membrane raft (GO: 0045121), membrane microdomain (GO: 0098857), membrane region (GO: 0098589), caveola (GO: 0005901), and plasma membrane raft (GO: 0044853). Similarly, from MF analysis, we understand the forms of protein and target binding, primarily involving G protein-coupled amine receptor activity (GO: 0008227), adrenergic receptor activity (GO: 0004935), catecholamine binding (GO: 1901338), heme-binding (GO: 0020037), and tetrapyrrole binding (GO: 0046906). Overall, GO enrichment analysis results suggest that most of the effects of CRP on the liver occur in cell proliferation, regulation of enzyme activity, and energy metabolism. The detailed results are in Supplementary Table S9.
To further identify CRP's potential pathways against LI, we also performed KEGG pathway enrichment analysis on 177 consensus genes, and the screen identified 49 pathways. For multiple pathways obtained by enrichment according to p. adjust (p < 0.01), we selected the top 20 paths for presentation ( Figure 3G). The details of all KEGG pathway enrichment analyses are provided in Supplementary Table S10. Among these pathways, the interleukin (IL)-17 signaling pathway (hsa04657), endocrine resistance (hsa01522), the T cell receptor signaling pathway (hsa04660), Th1 and Th2 cell differentiation (hsa04658), platinum drug resistance (hsa01524), and others are critical pathways by which CRP acts on LI. Some of these are inflammatory pathways, some are apoptotic differentiation pathways, and others are receptor resistance pathways. Activation of inflammatory pathways such as IL-17 signaling (hsa04657) and tumor necrosis factor signaling (hsa04668) can significantly affect hepatocyte function (Yin and Feng, 2017;Doganyigit et al., 2020). Normal hepatocytes accelerate apoptosis after liver damage (Zhang T. et al., 2020). When CRP is administered to humans, the apoptosis pathway (hsa04210) may be affected by proteins such as BAX, BCL2, and CASP3 to maintain normal function. Liver function is also affected by the endocrine system, as endocrine resistance causes toxin-associated steatohepatitis (Clair, et al,2018).

Computational Validation of Ingredients-Targets Interactions
According to the GO and KEGG results, we found that CRP anti-LI is a biological process with multiple targets and pathways. Hesperidin is highly abundant in CRP and is a major active component (Xia et al., 2006). Nobiletin has suitable ADME parameters, suggesting that it may exhibit good drug performance during LI treatment. Based on PPI analysis, we know that some genes linked to naringenin have high metrics. We believe that components such as hesperidin, nobiletin, and naringenin simultaneously affect LI through common targets. Therefore, these compounds were subjected to molecular docking with apoptosis regulators BAX, CASP3, and BCL2 to test our hypothesis. The lower binding energy between molecules represents more potent force binding energy between them, and <0 kcal/mol favors the binding reaction . According to the docking results, the binding energies of hesperidin to BAX and CASP3 were −8.2 and −8.6 kcal/mol, respectively; the binding energies of nobiletin to BAX and BCL2 were −5.9 and −7.7 kcal/mol, respectively; and the binding energies of naringenin to BCL2 and CASP3 were −7.7 and −7.2 kcal/mol, respectively. These results suggest that these compounds and their corresponding proteins have good binding. Figure 4 shows the combinations. Multiple linkages were formed between hesperidin and three amino acid residues in BAX, ARG-37, LEU-45, and ASP-48, including a C-O double bond and strong binding of the amino group ( Figure 4A). Hesperidin interacts with two amino acid residues ASN-208 and ARG-207, in CASP3 ( Figure 4B). These results suggest that a compound simultaneously forms multiple bonds with different amino acids in a protein, allowing the drug to affect the protein. The molecular docking results for nobiletin with BAX and BCL2 are shown in Figures 4C,D, respectively. The C-O double bond on nobiletin binds to the -OH group on SER-55 and the other O atom is connected to the H atom on ASP-519 ( Figure 4C). The two O atoms on nobiletin and the amino acid residue TYR-9 on BCL2 and the H atom on ASP-196 form two hydrogen bonds ( Figure 4D). Figures 4E,F show the docking results of naringenin with BCL2 and CASP3, respectively. The naringenin molecule binds to three residues (ASN-11, TYR-9, and HIS-186) on BCL2 to form multiple hydrogen bonds ( Figure 4E). Naringenin also forms strong hydrophobic binding with residues ARG-207, SER-205, and SER-251 in CASP3 ( Figure 4F). Our analysis showed that hesperidin, nobiletin, and naringenin effectively acts on protein targets closely related to apoptotic effects.

Experimental Validation
Cell Viability CCK-8 assays were performed to determine the effects of different doses of hesperidin and naringenin on LX-2 cell viability ( Figures 5A,D). When hesperidin or naringenin was added at concentrations >75 μmol/L, cell proliferation was not significant, and viability remained high (100-130%). Based on this, three concentrations (75, 100, and 125 μmol/L) were selected for subsequent experiments.

Target Validation
TGF-β1-induced activation of LX-2 cells predisposes to liver fibrosis, resulting from chronic LI developing into cirrhosis that ultimately leads to liver failure (Bestion et al., 2020;Xiang et al., 2020). We chose hesperidin and naringenin to evaluate our prediction results further to validate changes in several relevant proteins in LX-2 cells with or without TGF-β1 (5 ng/ ml) treatment. We performed qRT-PCR to determine the effect of hesperidin on BAX and CASP3 mRNA levels and that of naringenin on CASP3 and BCL2 mRNA levels. Western blot was used to measure naringenin's effect on CASP3, BCL2, ATK1, and pAKT1 protein levels. We found that mRNA levels of BAX and CASP3 were significantly lower after TGF-β1 treatment (p < 0.01, Figure 5B,C,E), BCL2 mRNA was significantly elevated (p < 0.01, Figure 5F), and mRNA levels of CASP3 and BCL2 were consistent with their protein expression results ( Figures 5G,H). Compared with the group that only received TGF-β1 stimulation, The levels of multiple index proteins were adjusted to different extents after treatment with different concentrations of hesperidin or naringenin, which made the results obtained after compound treatment more similar to the normal group. Western blotting revealed that ATK1 was only weakly affected by TGF-β1 and naringenin; protein expression changes in each group were relatively stable ( Figure 5I). Protein levels of pAKT1 increased rapidly after TGF-β1 treatment but decreased significantly after naringenin treatment ( Figure 5J). This suggests that naringenin may have a more pronounced inhibitory effect on AKT1 phosphorylation. These data collectively suggest that CRP reduces LI by mediating cell apoptosis and protein phosphorylation and regulating BAX expression, CASP3, BCL2, and ATK1, all of which are consistent with our molecular docking results. The results support our predicted results based on systems pharmacology.

DISCUSSION
The liver is both a metabolic organ and an important mediator of immune function. LI may cause cancer, and the use of anti-cancer drugs may lead to more severe liver damage (Björnsson et al., 2020). TCM originates from nature and has few side effects, but its application and development have some limitations due to the complexity of the compositions and mechanisms of action. CRP is a pure, natural TCM that can also be used as a food, and a variety of compounds in CRP have been shown to have antiproliferative effects on hepatoma cells (Chu et al., 2017).
In this study, we identified the active components and related targets of CRP. We linked them to LI-related targets to construct a D-C-G-D network that depicts links among 117 consensus genes and diseases. The PPI analysis showed complex interactions between 117 overlapping genes, and topological value analysis further indicated 25 core targets. We also examined these targets' expression relationships in the liver, heart, lung, kidney, small intestine, and colon. The results showed that their expression levels were similar in these organs that belong to the TCM category of "five viscera and six entrails." The theory of TCM posits that the viscera is an organic whole, the center of human life activities, and the organs work together to maintain the living body.
The GO analysis results revealed biological connections between drugs and diseases, and the BP enrichment clarified the response to the stimulation of external substances, such as response to acid chemical (GO:0001101). Hepatocytes are more sensitive to exogenous substance stimulation, and liver lesions develop when individuals regularly work in environments containing chemicals (Zhao et al., 2020). The CC analysis revealed the cellular environment in which the intersection protein or its product occurred, including membrane raft (GO: 0045121). Alcohol alters plasma membranes in vitro; preventing membrane raft oxidation is one approach to inhibit LI caused by excessive alcohol consumption (Chen et al., 2018). MF results suggest how intersection proteins or their products bind or function in vivo after the compound performs its action.
Among them, catecholamine binding (GO: 1901338) is related to the expression of genes, and catecholamine receptors' regulation prevents acute stress-induced LI (Zhu et al., 2014). We also found that a large number of GO terms were related to mitochondria, including regulation of mitochondrial membrane potential (hsa04210), apoptotic mitochondrial changes (GO: 0008637), and mitochondrial outer membrane permeabilization (GO: 0097345). This finding suggests that one of the critical mechanisms by which CRP ameliorates LI is to alter cellular energy metabolism (Xu et al., 2017;Dilberger et al., 2019).
The KEGG pathway enrichment results revealed that inflammatory and apoptotic pathways might be the primary CRP action pathways. CRP mediates the IL-17 signaling pathway by regulating targets such as PTGS2, MAPK3, and CASP3. PTGS2, also known as cyclooxygenase 2, extensively participates in hepatic inflammation induced by xenobiotics . Through the D-C-G-D network, we know that hesperidin, nobiletin, and tangeretin are all involved in regulating PTGS2. The inhibition of PTGS2 mRNA expression is one of the manifestations of LI's acute inflammation resolution . It also suggests that a protein can be affected by several compounds simultaneously. LX-2 cells are prone to proliferate faster, and this can lead to liver fibrosis (Bestion et al., 2020). Our in vitro results suggest that CRP can induce apoptosis in LX-2 cells by regulating targets such as BAX, BCL2, and CASP3, consistent with our finding that the apoptosis pathway (hsa04210) is affected by CRP. The liver has a unique regenerative capacity (Nejak-Bowen and Monga, 2011). Some cancer pathways are associated with uncontrolled cell growth, and their dysregulation can lead to liver cancer. The phosphoinositide 3-kinase (PI3K)-Akt signaling pathway (hsa04151) can be activated by a variety of cell stimulants and toxins and is potentially correlated with liver cancer . Our prediction results suggest that AKT1 has a high degree in this pathway, and the phosphorylated form is a biomarker for cancer and tumor biology (Balasuriya et al., 2020). Our in vitro validation experiments demonstrated that naringenin significantly inhibited AKT1 phosphorylation, which in turn mediated PI3K-Akt signaling pathways to counteract LI. This interesting finding will be the focus of our future study. Nobiletin in CRP inhibits hepatoma metastasis resulting from PI3K-Akt signaling and may become a new compound for liver cancer treatment (Shi et al., 2013). The western blot results indicated that naringenin significantly reduced pAKT1 levels, suggesting that it prevents LI exacerbation. Increased cytochrome P450 expression improves hepatocyte function (Lewis et al., 2020); xenobiotics' metabolism by cytochrome P450 (HSA00980) pathway was analyzed. We found that it resulted from naringenin, tangeretin, narirutin, and other compounds by affecting GSTP1, CYP1A1, and CBR1 targets.
We propose the following mechanism by which CRP counteracts LI. Hesperidin and naringenin in CRP affect the apoptosis pathway by acting on CASP3, BAX, and BCL2, reducing the likelihood of liver fibrosis. Naringenin significantly inhibits AKT1 phosphorylation, in turn mediating PI3K-Akt signaling to counteract LI. Systems pharmacology is a useful strategy to predict therapeutic mechanisms and identify new research directions. We acknowledge that the predictions and actual situations may be biased; however, they complement the in vitro findings, consistent with the molecular docking results that test the system pharmacologybased screening strategy's reliability from a different perspective.

CONCLUSION
Previous drug development models mainly followed the concept of "one drug, one gene, and one disease." In this study, modern technologies such as UPLC-Q-TOF-MS/MS and molecular docking were used to clarify the specific modes of the synergistic effects of multiple CRP components against LI, and provided convincing evidence. System pharmacology of TCM is an emerging multi-field interdisciplinary, but its development still has certain limitations. In addition to being an important reference for CRP in the treatment of liver diseases, our research also has many shortcomings. The specific interactions between all the drug components, proteins and multiple signal pathways involved in it need to be further studied. This approach of Systems pharmacology is a powerful new way to elucidate the mystery of TCM.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
JW and QG conceived and designed the study. SY conceived and designed the in vitro experimental validation. JW, XY, and SY contributed equally to this work and shared first authorship. All authors have read and approved the final manuscript.

FUNDING
This study was supported by the National Traditional Chinese Medicine Standardization Project (ZYBZH-Y-JX-27).