Exploration of the Components and Pharmacological Mechanisms of Keyin Pill-Induced Liver Injury Based on Network Pharmacology

Background Keyin pill (KP), a patented medicine in China, is used to treat psoriasis. However, KP has been reported to have liver toxicity, but its toxic substance basis and underlying mechanisms remain unclear. Therefore, this study aimed to explore the pharmacological mechanisms and components of KP-induced liver injury through animal experiments, UPLC-QTOF/MS combined with network pharmacology. Methods Firstly, based on the immune stress mouse model, liver function parameters and hematoxylin-eosin (H&E) staining were detected to investigate KP-induced liver injury. The UPLC-QTOF/MS method was used to identify the components of KP. CTD database and literature mining were further applied to screen nonliver protective components. Subsequently, the nonliver protective components and their corresponding targets and targets of hepatotoxicity were analyzed by the method of network pharmacology. Finally, key targets from networked pharmacology were examined by the enzyme-linked immunosorbent assay (ELISA) and molecular docking. Results Our results indicated that KP had hepatotoxicity in male Kunming mice, which could favor hepatocyte necrosis and infiltration of inflammatory cells. A total of 70 nonliver protective compounds were identified and screened. The results of network pharmacology illustrated that methoxsalen, obacunone, limonin, and dictamnine might be the main compounds that caused liver damage. The potential hepatotoxicity mechanisms of KP might be through the IL17 and apoptosis pathways to regulate IL6, TNFα, CASP3, and CASP8 targets, thereby causing inflammation, excessive release of factors, and hepatocyte necrosis. The results of the ELISA experiments indicated that KP could increase the release of IL6 and TNFα inflammatory factors in liver tissues. Molecular docking suggested that methoxsalen, obacunone, limonin, and dictamnine had moderate binding ability with CASP3 and CASP8. Conclusion In this study, the material basis and potential pharmacological mechanisms of KP-induced liver injury were preliminarily explored. Our research provides the initial theoretical basis for reducing the toxicity of KP.


Introduction
Keyin pill (KP) is a Chinese patent medicine composed of Smilax glabra Roxb, Dictamnus dasycarpus Turcz, the rhizome of Menispermum dauricum, and Polygonum bistorta L., which is often used in the clinical treatment of blood-heat type psoriasis [1]. Clinical studies showed that KP could alleviate psoriasis symptoms and slow down disease progression [2,3]. However, KP could damage hepatitis and liver function in clinical application [4,5], and the China Food and Drug Administration also issued a risk opinion for liver injury due to KP, which restricts its clinical application.
In recent years, cumulating evidence suggested the toxicity of medicinal materials in KP, including severe or even fatal liver damage, which occurred after taking Dictamnus dasycarpus Turcz and Chinese medicine compounds containing Dictamnus dasycarpus Turcz [6]; Huang et al. revealed that the obacunone, limonin, and dictamnine of Dictamnus dasycarpus Turcz were the main components that cause liver damage [7]; modern pharmacological studies manifested that the water extract of the rhizome of Menispermum dauricum could cause liver damage in mice, and the mechanism of liver damage was related to the induction of lipid peroxidation after causing oxidative stress in the body [8]; dauricine in the rhizome of Menispermum dauricum might be the main component of its hepatotoxicity [9]; the study prompted that Smilax glabra Roxb might have hepatotoxicity [10]. However, due to the complex herb composition of KP, the toxic components and mechanisms of KP leading to liver injury are still unclear. Consequently, it is necessary to carry out extensive research on the material basis and mechanisms of hepatotoxicity from KP.
In addition, KP-induced liver injury has specifc properties. Te incubation period of liver injury caused by KP ranged from 1 to 90 days, and the cumulative dose ranged from 20 to 1800 g. Tere was no diference in the dependence of liver injury on the dose and course of treatment [11]. Terefore, it is necessary to construct an animal model that can refect KP's specifcity of the liver. In the study of the toxicity of traditional Chinese medicine (TCM), the lipopolysaccharide (LPS)-induced immune stress idiosyncratic liver injury model can refect the specifcity of livers to drugs, such as the research of Polygonum multiforum and Xian-Ling-Gu-Bao [12,13].
Network pharmacology is a promising approach to investigating the comprehensive mechanisms of TCM and herbal formulae [14]. Compared with traditional toxicology, network pharmacology has the characteristics of saving resources and high speed in searching for the toxic ingredients and mechanisms of toxic Chinese herbal medicines. Tence, network pharmacology has been widely applied to explore the mechanisms of action of toxic components in TCM [15].
Tis study investigated the dose and pathology of KPinduced hepatotoxicity in male Kunming (KM) mice based on the immune stress model. Te UPLC-QTOF/MS method was used to identify KP compounds and combine the CTD database and literature mining to screen nonliver protective components. Next, KP network analysis was performed to investigate the hepatotoxicity mechanisms and toxic ingredients of KP, including protein-protein interaction (PPI) analysis, functional enrichment analysis, and compoundtarget-pathway (CTP) network construction. Finally, the top-ranked targets obtained by network pharmacology were investigated by applying enzyme-linked immunosorbent assay (ELISA) experiments and molecular docking. Turcz, 30 g of the rhizome of Menispermum dauricum, and 90 g of Polygonum bistorta L. were mixed and decocted three times (2 hours at the frst time, 1 hour at the second time, and 1 hour at the third time). Te decoction was fltered and then transferred to a rotary evaporator for a concentration at 70°C to obtain the whole formula (containing 1 g of herbs per 1 mL). Te whole recipe of KP was then stored in a refrigerator at 4°C. Based on previous studies, it was found that KP caused liver injury with special heterogeneity. Idiopathic drug-related liver injury is the body's idiosyncratic response to drugs, both allergic (immune specifc) and metabolic (metabolic specifc). Drug-induced liver injury induced by the organism's idiosyncratic response to a drug cannot be replicated in commonly used experimental animal models. Terefore, it is necessary to construct an animal model to refect the specifcity of the liver. In the past 30 years, studies have found that the combination of nonliver injury dose LPS and specifc liver injury drugs can increase the liver injury efect of drugs on mice by inducing immune stress [16,17]. In addition, LPS also has the toxic characteristics of targeting the liver. Terefore, in order to avoid the infuence on the liver toxicity induced by KP, LPS 0.1 mg/kg without liver damage dose was selected to construct the immune stress mouse model.

Animal Experiments and
Te male KM mice were randomly divided into 5 groups (n � 8): the high-dose group (14.625 g/kg), the medium-dose group (4.875 g/kg), the low-dose group (1.625 g/kg), the model group, and the control group. KP group and model group mice were injected with LPS 0.1 mg/kg via the tail vein. Te control group was injected with 0.9% NaCl via the tail vein. After 2 hours, 14.625 g/kg (9 times the clinically equivalent dose), 4.875 g/kg (3 times the clinically equivalent dose), 1.625 g/kg (clinical equivalent dose) of KP, and 0.9% NaCl were given by gavage, respectively. After 8 hours, blood from the mice of each group was collected from the retroorbital plexus. Te blood was gathered in a coagulation tube and then centrifuged at 3000 rpm in a high-speed centrifuge for 10 minutes, and then serum was collected and stored in a refrigerator at −80°C. Te livers were removed and fxed in 4% paraformaldehyde and then stored at room temperature until further analysis (Figure 1(a)).

Serum Biochemistry.
Liver function parameters, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (AKP) in the mouse serum were detected by using the "ALT, AST, and AKP detection kits" (Nanjing Jiancheng Institute of Biological Engineering).

Hematoxylin-Eosin (H&E)
Staining. Liver tissues were removed from the fxed liquid. Target tissues were trimmed with a scalpel in the ventilation cupboard. Ten, trimmed tissues and the label were put in the dehydration box. Next, they were dehydrated and dipped in wax. Te modifed tissue sections were mounted on a wax block on the parafn microtome, and the thickness of the sections was 4 μm. Te sections were stained with hematoxylin solution for 3-5 minutes and rinsed with tap water. Te sections were then treated with hematoxylin. Finally, the sections were stained with eosin dye for 5 minutes. Afterward, we observed through microscopic examination, image acquisition, and analysis.

UPLC-QTOF/MS Analysis.
Te processing procedure of 100 μL sample of KP consisted of adding 500 μL pure water, vortex mixing for 30 seconds, ultrasonic treatment with an ice bath for 1 hour, centrifugation at 12000 rpm at 4°C for 10 minutes, and removing the supernatant into a 2 mL sample bottle for machine detection. LC30A UPLC system (Waters Corp, Milford, USA) was conducted for analysis. Te separation was performed on the UPLC BEH C18 column (1.7 μm * 2.1 * 100 mm) with gradient elution of 0.1% formic acid in liquid A water and 0.1% formic acid in liquid B acetonitrile. Te sample injection volume was 5 μL. Primary and secondary MS data were gathered by the AB 5600 Triple TOF TF 1.7 Mass Spectrometer Control Software Analyst, AB Sciex. In each data collection cycle, molecular ions with the strongest strength greater than 100 were selected to collect the corresponding secondary MS data. Bombardment energy: 40 eV, impact energy diference: 20 V, 15 secondary spectra every 50 ms. ESI Ion source parameters were set as follows: atomizing pressure (GS1): 55 Psi, auxiliary pressure: 55 Psi, air curtain pressure: 35 Psi, temperature: 550°C, spray voltage: 5500 V (positive ion mode) or −4000 V (negative ion mode). Te raw mass spectra were imported into Progenesis QI software. Te retention time correction, peak recognition, peak extraction, peak integration, peak alignment, and other work were carried out. Meanwhile, the corresponding Chinese medicine library in the compound was established, and the peaks containing MSMS data were identifed by the self-builttwo-stage mass spectrometry database and the corresponding cracking law matching method.  . Notes: * P value refers to the comparison between the model group and administration groups, and # P value refers to the comparison of each administration group; * P < 0.05, * * P < 0.001, * * * * P < 0.0001, and # P < 0.05.

Target Fishing of Bioactive Components and Hepatic
Injury in the Database. Te smiles format fles for nonliver protective compounds were downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Subsequently, smile format fles were uploaded to SEA (https://sea.bkslab. org/) and Swiss Target Prediction (https://www. swisstargetprediction.ch/) databases to obtain compoundrelated targets. Meanwhile, the targets of nonliver protective ingredients were also screened by HERB (https://herb. ac.cn/) and CTD databases. Targets were only limited to "Homo sapiens" to obtain reliable results. Liver injury was used as the keyword to search for targets in GeneCards (https://www.genecards.org/) and DiGSeE (https://210.107. 182.61/geneSearch/) databases. Liver injury targets with above-average relevance scores were selected from the GeneCards database, and liver injury targets of higher-thanaverage evidence were obtained from the DiGSeE database. Liver injury targets obtained from both databases were deduplicated and uploaded to Uniprot's website (https:// www.uniprot.org/) for standardization.

Potential Targets of Hepatotoxicity Induced by KP.
Te compound targets and hepatotoxicity targets were uploaded to the OmicShare online platform (https://www. omicshare.com/). Te Venn function module was used to screen the overlapping targets. Intersecting targets included 556 targets associated with liver injury in the CTD database. Among the 556 targets, targets with liver injuryrelated scores greater than the average were screened out as direct targets of KP-induced liver injury. Direct targets of liver injury were classifed by pharmacological mechanisms.

PPI Analysis of Direct Targets.
Te direct targets were imported into the STRING database (https://cn.string-db. org/). Te conditions were set as the PPI interaction score >0.4, and the species was restricted to "Homo sapiens." Te results were exported to the TSV format fles. For visual analysis of direct targets, the TSV format fles were imported into Cytoscape 3.7.1 software, and the "Network Analyzer" function was used to analyze the topology structure of the PPI network. Degree > mean value was set as the cut-of threshold, of which met the previously mentioned screening criteria were core targets.

Enrichment Analysis of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG).
In order to clarify the biological functions and related signaling pathways of the targets, the screened direct targets were enriched and analyzed. GO and KEGG functional annotation of direct targets were acquired from the STRING database, with the species limited to "Homo sapiens." Screening criteria were as follows: (1) false discovery rate <0.01; (2) observed gene count was greater than the average; and (3) literature has reported that signaling pathways were relevant to direct or indirect liver injury. Afterwards, the top 10 signaling pathways and GO terms that met the previously mentioned criteria were fltered out and were visualized by using the histogram and bar chart through Bioinformatics online website (https://www.bioinformatics.com.cn/).

Construction and Analysis of the CTP Network.
Te corresponding targets of the top 10 signaling pathways and relevant compounds were imported into Cytoscape 3.7.1 software to construct the CTP network. Te CTP network topological isomerism parameter was analyzed by the network analyzer function. Te targets were regarded as key targets of liver injury induced by KP which met the following two criteria: (1) the CTP network topological isomerism parameter of degree value was greater than the mean value and (2) the topology degree value of the PPI network was greater than the average. Compounds with above-average degrees were considered potentially hepatotoxic compounds. Furthermore, combined with the structure of hepatotoxicity warning [18], the liver injury compounds of KP were screened.

Investigation of Key Targets.
According to the results of network pharmacology, the ELISA experiments were carried out on IL6 and TNFα (abs520004 and abs520010, China) in liver tissues, which were the top 2 infammatory factors.

Molecular Docking.
In order to explore the relationship between liver injury caused by the main components of KP and apoptosis, molecular docking was applied to decipher the interaction between core compounds and hub apoptotic targets. Te three-dimensional structure "SDF" fles of methoxsalen, obacunone, limonin, and dictamnine were downloaded from the PubChem database. PDB format fles of CASP3 (3PD1) and CASP8 (3KJQ) were downloaded from the PDB database (https://www1.rcsb.org/). Molecular docking was carried out on Maestro software (Schrödinger, USA) with the following 4 steps: protein preparation wizard, receptor grid generation, ligand preparation, and ligand docking.
2.9. Statistical Analysis. All data were recorded as the mean ± standard deviation. Comparisons between multiple groups were performed by using one-way ANOVA with Tukey's test using GraphPad Prism software (version 7). A value of P < 0.05 was considered to be statistically signifcant.

Efect of KP on Serum Biochemical Markers.
Liver function parameters of ALT, AST, and AKP were measured to evaluate the efect of KP on the degree of liver injury based on the immune stress model. As shown in Figures 1(b

Histopathological Evaluation of Hepatic Tissue after
Administration of KP. H&E staining was applied to observe the histopathological changes in liver tissues in each group.
In the control group, there was a large amount of hepatocyte steatosis, with tiny round vacuoles in the cytoplasm (black arrow) and no obvious infammatory changes (Figure 2(a)). Compared with the control group, the model group had a large amount of vascular congestion (black arrow), a small amount of hepatic cell steatosis around the portal area, and tiny round vacuoles in the cytoplasm (red arrow), and no obvious infammatory changes (Figure 2(b)). Pathological results from the low-dose group showed focal necrosis of liver lobules, nuclear fragmentation, and a small amount of infammatory cell infltration (black arrow) (Figure 2(c)). Te pathological results of the medium-dose group showed a large number of hepatic cell steatosis and tiny round vacuoles (black arrow) in the cytoplasm and had a lobule with focal necrosis of liver cells and fragmentation of nuclei with a small amount of infammatory cell infltration (red arrow) (Figure 2(d)). In the high-dose group, histopathological analysis of the liver revealed that there was necrosis of liver cells, fragmentation of nuclei, and massive blood vessel congestion (Figure 2(e)). H&E staining indicated that administration of KP could cause cell necrosis and infammatory cell infltration in the liver of mice.

Identifcation of Bioactive Ingredients from KP.
UPLC-QTOF/MS was used to qualitatively analyze the chemical constituents of KP. Te total ion chromatograms (TIC) of KP were scanned in positive and negative ion modes (Figures 3(a) and 3(b)). Based on UPLC-QTOF/MS analysis results, 70 nonliver-protective ingredients were screened by the CTD database, Pharmacopeia (2020), and literature consulting (Table 1).

Acquisition of Component Targets and Hepatotoxicity
Targets. A total of 1618 component targets were screened from SEA, Swiss target prediction, HERB, and CTD databases after removing the duplicated targets. Tere were 2381 targets with an above-average relevance score that were fltered through the GeneCards database, and 106 targets with higher-than-average evidentiary results were achieved in the DiGSeE database. Subsequently, 2357 liver toxicity targets were obtained after normalization by the Uniprot database. Tere were 564 intersection targets that were obtained. Among the intersection targets, 556 targets were associated with liver injury in the CTD database. Tere were 87 targets with higher-than-average inference scores of liver injury evidence in the CTD database. Consequently, the 87 targets were considered to be direct targets of KP-induced liver injury. Direct targets were classifed into 6 categories including the metabolic enzyme family, the transporter family, apoptosis, infammation, nuclear hormone receptors, and oxidative stress according to the pharmacological mechanisms.

PPI Network Analysis.
Tere were 87 targets uploaded to the STRING database to obtain PPI fles, and they constructed a PPI network by using Cytoscape 3.7.1 software. Te PPI network contained 87 nodes and 1221 edges, with an average node degree of 28.06 (Figure 4(a)). A total of 38 targets with a degree value >28.06 were selected as key targets (Figure 4(b)). Among these, the top 10 key targets in terms of degrees of freedom were ALB, IL6, MAPK3, CASP3, TNF, MAPK8, CYCS, CAT, MAPK1, and IL1β, with respective degrees of freedom of 68, 63, 59, 58, 58, 57, 57, 50, 49, and 49.
3.6. Enrichment Analysis. Te GO enrichment analysis was performed to describe the gene's functions of 87 targets. GO terms include 3 categories: biological process (BP), cellular component (CC), and molecular function (MF). Among them, BP was primarily associated with the cellular process, response to chemicals, metabolic process, cell communication, regulation of cell death, and regulation of the Evidence-Based Complementary and Alternative Medicine 5 apoptotic process. CC mainly consisted of intracellular organelle, cytoplasm, intracellular membrane-bounded organelle, and endomembrane system. MF was chiefy involved in protein binding, heterocyclic compound binding, organic cyclic compound binding, and ion binding ( Figure 5(a)). KEGG enrichment analysis showed that apoptosis, hepatitis B, IL17 signaling pathway, TNF signaling pathway, NOD-like receptor signaling pathway, nonalcoholic fatty liver disease, metabolic pathways, the toll-like receptor signaling pathway, bile secretion, and necroptosis were the top 10 signaling pathways ( Figure 5(b)). Te results of the KEGG analysis illustrated that the apoptosis signaling pathway, IL17 signaling pathway, TNF signaling pathway, and NOD-like receptor signaling pathway might be the main signaling pathways of KP-induced liver injury.

CTP Network Analysis.
Tere were 10 pathways combining 47 active components and 68 targets that were used to construct a CTP network. Te CTP network demonstrated 125 nodes (containing 47 chemical compound nodes, 68 target nodes, and 10 pathway nodes) and 386 edges (representing the interaction between chemical compounds, targets, and pathways) ( Figure 6). According to the screening criteria, a total of 21 key targets of liver injury caused by KP were fltered (Table 2), such as IL6, TNF, IL1β, MAPK1, CASP3, CASP8, NFKB1, CXCL8, CCL2, CYP2E1, and CYP3A4. A total of 15 compounds had degree values greater than the mean in the analysis of topological isomerism value. According to previous research on the structure-toxicity relationship of hepatotoxic compounds [18] and literature mining [7,19], the compounds of methoxsalen, obacunone, limonin, and dictamnine might be the material basis of the liver injury induced by KP.     Evidence-Based Complementary and Alternative Medicine group, the IL6 level in the liver was signifcantly increased in the high-dose of the KP group (Figure 7(a)). Compared with control and model groups, the TNFα level in the liver was signifcantly increased in the high-dose of the KP group (Figure 7(b)). Te results revealed that KP caused liver damage by stimulating infammatory factors in the infammatory signal pathway.

Molecular Docking.
Network pharmacology studies predicted that methoxsalen, obacunone, limonin, and dictamnine might be the main substances causing liver injury, and the mechanism may be related to apoptosis. Terefore, the molecular docking approach examined the interactions between these four major compounds and the apoptotic proteins CASP3 and CASP8. Te molecular docking results demonstrated the binding afnity of methoxsalen, obacunone, limonin, and dictamnine to the targets CASP3 and CASP8 (Figures 8(a)-8(d) and 9(a)-9(d)). Te results of docking scores indicated that methoxsalen had stronger binding afnity to CASP3 and CASP8 compared with the other three compounds (Table 3). Methoxsalen could form three hydrogen bonds and one Pi-Pi stacking bond with ARG207, ARG64, GLN1161, and TRP206, respectively (Figure 8(d)). Methoxsalen could form two hydrogen bonding interactions with ARG260 and ARG413 and one Pi-Pi stacking interaction with TYR412 (Figure 9(d)).  Figure 4: PPI analysis results. (a) Pink represents apoptosis-related targets, green represents infammatory targets, red represents oxidative stress-related targets, light purple represents metabolic enzymes, dark purple represents nuclear receptors, and blue represents transportrelated targets. (b) Critical targets with degree value greater than average.  Evidence-Based Complementary and Alternative Medicine 9

Discussion
KP has been used in the treatment of psoriasis for more than 30 years [20]. Previously, more and more concerns over its potential hepatotoxicity have been raised [11]. However, the complex composition of KP makes it difcult to investigate which component produces hepatic toxicity. Terefore, in this study, we combined the animal immune stress model, UPLC-QTOF/MS analysis, and the network pharmacology method to systematically and comprehensively study KPinduced hepatotoxicity. First, the immune stress model was used to investigate the characteristics of KP-induced liver toxicity. Second, the UPLC-QTOF/MS analysis method was applied to identify the components in KP. Tird, networked pharmacologic technology was employed to rapidly screen for key compounds and mechanisms of KP-induced hepatotoxicity. Te characteristics of this research method are that the combination of in vivo experiments and computer screening models can quickly screen hepatotoxic substances and fnd the main mechanisms of hepatotoxicity. Te advantages of this method save the use of experimental animals, experimental costs, and time.
As the main metabolic organ and excretory organ, the liver is the main target organ for toxic injury [21]. Our fndings showed that KP could cause liver cell necrosis, lobular nuclei fragmentation, and infammatory cell infltration in male KM mice. At the same time, there were also changes in serum ALT, AST, and AKP levels. Furthermore, the results of network pharmacology indicated that methoxsalen, obacunone, limonin, and dictamnine might be the main compounds of KP-induced hepatotoxicity. KPinduced hepatotoxicity mechanisms might be primarily through the regulation of IL6, TNFα, CASP3, and CASP8 targets by acting on the IL17 and apoptosis pathways.
Te results of the "toxic component-target-pathway" network showed that methoxsalen, obacunone, limonin, and dictamnine might be the key compounds of KP-induced liver damage. In recent years, compounds with furans, Figure 6: Compound-target-pathway network (yellow nodes represent compounds, green nodes represent signaling pathways, and clustered circles represent targets).  IL6  63  11  MAPK3  59  9  CASP3  58  15  TNF  57  13  MAPK8  57  10  CYCS  50  7  MAPK1  49  9  IL1B  49  9  CXCL8  46  10  CCL2  43  6  CASP8  40  12  MAPK14  40  9  RELA  39  15  FOS  38  10  CASP9  37  8  NFKBIA  33  7  NFKB1  32  13  NOS2  32  18  FASLG  32  6  CYP2E1  31  6  CYP3A4  31  6 10 Evidence-Based Complementary and Alternative Medicine anilines, quinones, hydrazines, thiophenes, arylpropionic acids, and alkynes in their structures may have potential hepatotoxicity [22]. Furan rings generate epoxides or cisaldehyde after activation by metabolic enzymes that can covalently modify the biomolecules to cause cellular damage and toxicities [23]. Dictamnine is metabolized to form epoxides that conjugate with glutathione (GSH) in a very early phase. When hepatic GSH reserves become depleted, the reactive metabolites will bind to functional proteins or other nucleophiles and potentially cause necrosis [24]. Obacunone is metabolized by CYP3A4 into a cis-aldehyde intermediate, which consumes GSH to produce hepatotoxicity [25]. However, obacunone, limonin, and dictamnine in Cortex Dictamni produce hepatotoxicity that may be through additive efects. Dictamnine can not only combine with glutathione in the early stage and consume glutathione but also form adducts with liver proteins to cause liver damage. Obacunone and limonin bind mainly to lysine residues of liver proteins to produce hepatotoxicity [26]. Te research shows that methoxsalen exerts hepatotoxicity by inhibiting the cytochrome CYP450 enzyme in mice [27]. Tese results are in accordance with network pharmacology in our study. Herein, we concluded that the material basis of hepatotoxicity induced by KP was mainly caused by the superimposed efect of furan ring compounds. At present, there are many factors and complex mechanisms of liver injury caused by TCM, including liver enzyme abnormality, immunologic reaction, oxidative stress, lipid metabolism disorder, cholestasis, hepatocyte necrosis caused by mitochondrial dysfunction, and the release of infammatory factors [28]. During hepatocyte necrosis, the release of injury-associated molecules activates toll-like receptors on macrophages and induces proinfammatory signaling pathways leading to cytokine production. Among these proinfammatory cytokines, IL6 and TNFα are considered the key mediators of hepatotoxicity [29]. Our results indicated that the mechanism of KPinduced hepatotoxicity was mainly through the regulation of infammatory targets through the IL17 signaling pathway (Figure 10(a)). ELISA experiments further indicated that KP could promote the release of infammatory factors IL6 and TNFα in the liver of mice. Te production of TNFα is closely related to the activation of Kupfer cells. Elevated TNFα can stimulate hepatocyte apoptosis [30]. Te increased expression of TNFα can initiate the cleavage of pro-CASP8 and/or −CASP10 to activate CASP8/10, and CASP8/10 directly activates downstream CASP3 [31]. Te results of network pharmacology indicated that KP could regulate CASP3 and CASP8 targets by acting on the apoptosis signaling pathway (Figure 10(b)). Te results of molecular docking confrmed that methoxsalen, obacunone, limonin, and dictamnine had binding afnity with apoptotic proteins CASP3 and CASP8. Terefore, we speculated that the mechanism of KP-induced hepatotoxicity could be to promote hepatocyte apoptosis by promoting the release of infammatory factors and the initiation of the apoptotic pathway.
Tis study also has limitations in studying the material basis and mechanisms of liver injury caused by KP. First, although the research methods of network pharmacology have developed rapidly in the past decade, the accuracy of algorithm-based prediction of related targets of compounds needs to be further improved, and the analysis methods of network topological heterogeneity results need to be more

Conclusions
To conclude, we used a research method combining animal experiments, UPLC-QTOF/MS analysis, and network pharmacology to analyze toxic substances and potential mechanisms of KP hepatotoxicity. Te results of animal experiments confrmed that KP was toxic to the liver of mice. Te components of KP were identifed by the UPLC-QTOF/MS analysis method. Trough network pharmacologic analysis, methoxsalen, obacunone, limonin, and dictamnine might be the main substances of liver injury caused by KP. Its potential toxicity mechanisms might be via the IL17 signaling pathway and the apoptosis pathway acting on IL6, TNFα, CASP3, and CASP8 targets to exert hepatotoxic efects. Based on the results of network pharmacology, the ELISA experiments showed that KP might increase the IL6 and TNFα levels in liver tissues. Molecular docking results demonstrated that methoxsalen, obacunone, limonin, and dictamnine could interact with the targets CASP3 and CASP8. Our study lays the theoretical foundation for further study on hepatotoxic compounds and their mechanisms of KP.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.