Mechanism of Magnolia Volatile Oil in the Treatment of Acute Pancreatitis Based on GC-MS, Network Pharmacology, and Molecular Docking

Objective Magnoliae officinalis cortex (MOC) is one of the most frequently used traditional Chinese medicine (TCM) for the treatment of acute pancreatitis (AP). Magnolia volatile oil (MVO) is considered to be one of the main active ingredients in MOC for AP treatment. However, the underlying mechanism of MVO in AP therapy is unknown. Methods An integrated strategy of gas chromatography-mass spectrum (GC-MS), network pharmacology, and molecular docking simulation was employed to predict underlying mechanism of MVO in AP treatment. First, the compounds of MVO were identified by GC-MS, and the targets of the identified characteristic compounds were collected from several databases, as well as AP-related targets. Next, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were carried out to obtain the mechanism. Moreover, the binding activity between core therapeutic targets and their corresponding compounds was evaluated by molecular docking simulation. Results GC-MS results showed a total of 35 compounds that appeared in at least 18 out of 20 chromatograms were considered as characteristic compounds of MVO, and 33 compounds of those were identified. Network analysis demonstrated that 33 compounds regulated 142 AP-related targets. Of those, 8 compounds (α-eudesmol, γ-eudesmol, (−)-terpinen-4-ol, terpineol, hinesol, linalool, borneol, and β-eudesmol) and 8 targets (TNF, IL-1β, PPARγ, PPARα, PTGS2, NCOA1, CNR1, and ESR1) have a close relationship with AP treatment and were recognized as the key active compounds and the core therapeutic targets, respectively. The 142 targets were involved in both inflammation and calcium overload-related biological pathways, such as neuroactive ligand-receptor interaction, estrogen, MAPK, and calcium signaling pathway. Moreover, molecular docking simulation indicated that the 8 core therapeutic targets strongly interacted with their corresponding compounds. Conclusions In summary, the present study elucidated that the efficacy of MVO in AP treatment might be attributed to anti-inflammation and inhibition of calcium overload through multicomponents and multitargets.


Introduction
Acute pancreatitis (AP) is the most frequent digestive system illness with both high morbidity and mortality [1,2]. Te overall incidence rate of AP is 4.8-38 per 100,000 annually [3][4][5][6] and continually increasing [7,8]. Moreover, about 20% of AP might progress into severe acute pancreatitis (SAP) [9] due to associated necrosis of the pancreatic tissue and/or multiorgan failure, resulting in as high as 30% of deaths [10]. Te pathogenesis of AP is complex and still poorly understood, mainly involving trypsinogen activation, pancreatic microcirculation malfunction, endoplasmic reticulum stress, calcium overload, and infammatory [8,11]. Te current clinical therapy of AP is mainly symptomatic improvement, such as pain relief and the correction of fuid, electrolyte, or pH-disorders [12]. However, there are no efective drugs available for treating such a complicated disease. Terefore, the development of more efective new drugs is urgently needed for AP patients. As a key part of complementary and alternative medicine, traditional Chinese medicine (TCM), characterized as multicomponents and multitargets, has been confrmed to have a remarkable efect in AP treatment [11,13,14].
Magnoliae ofcinalis cortex (MOC), obtained from the dried bark of stem, root, and branches of Magnolia ofcinalis Rehd. et Wils or Magnolia ofcinalis Rehd. et Wils. var. biloba Rehd. et Wils, is one of the most commonly used TCM for treating AP [15][16][17]. Magnolia volatile oil (MVO), a complex mixture composed of a large number of terpenoids and their oxygenated derivatives, is the main active ingredient of MOC. Plenty of studies indicate that MVO possesses a wide range of bioactivities including antiinfammatory [18], antibacterial [19,20], antioxidant, and promoting gastric emptying [21] activities, showing a great potential of clinic application of MVO in AP treatment. Moreover, pieces of evidence suggest that the common compounds in MVO, limonene and borneol, have shown potent anti-AP efcacy in several animal models via antioxidative and anti-infammatory [22,23]. However, the key active compounds and underlying mechanism of MVO in AP treatment are not entirely clear, which has driven further investigation.
Network pharmacology is a novel and efective method to systematically and comprehensively describe the law of interactions between components and targets through the integration of chemoinformatics, bioinformatics, network biology, network analysis, and traditional pharmacology [24,25]. Recently, network pharmacology has been successfully employed to investigate the active compounds and the underlying mechanism of TCM with complex chemical composition [26][27][28][29][30]. Molecular docking simulation is an important way of drug virtual screening, which can provide useful information about drug-receptor interactions by calculating the binding energy of small-molecule ligands (drugs) with proteins (receptors) [31][32][33][34][35]. Recently, molecular docking simulation has been widely used to explore the mechanism of TCM [36][37][38][39][40]. However, the active compounds and the underlying mechanism of MVO in AP treatment were yet to be elucidated based on network pharmacology and molecular docking simulation.
As we all know, the chemical composition of TCM is complex and exhibit large diferences in the types and quantities as afected by many factors, such as species, geographical regions, harvesting time, and processing conditions [41][42][43][44][45]. Tis would make it extremely difcult to understand the active compounds and the mechanism of TCM. Tough this problem has been alleviated using network pharmacology-based analysis, the data of compounds were only obtained from few known databases, or from the results of chemical composition analyses of a single sample, the diference in chemical composition of TCM caused by diferences sources was rarely considered in network pharmacology analysis, which would make the result lack generalizability and representativeness because those compounds cannot efectively refect the chemical composition features of TCM from various sources. Terefore, in this study, the volatile compounds in MOC from diferent sources (species and geographical regions) were identifed by gas chromatography-mass spectrum (GC-MS), and the mechanisms of MVO in AP treatment were predicted through network pharmacology and molecular docking simulation. A detailed fowchart of the network pharmacology-based study is presented in Figure 1.

Preparation of Sample Solution.
A total of 20 batches of MOC were collected from Chengdu Lotus Pond Chinese Herbal Medicine Market (Chengdu, China) and authenticated by associate Professor Lu Chen (Chengdu University of TCM). About 100 g of MOC (dried and powdered) and 1,000 mL of water were put into a volatile oil distillation apparatus (recorded in 2015 Edition Chinese Pharmacopoeia) and extracted by steam distillation for 6 h. Te MVO was subsequently collected from the volatile oil extractor and dried by anhydrous sodium sulfate. Twenty milligram of the obtained oil was dissolved in 2 mL of n-hexane and fltered through a 0.22 µm flter, and then the solution was analyzed by GC-MS.

GC-MS Analysis.
Analysis of MVO was performed using Agilent 7890A-5975C GC-MS with an HP-INNOWAX capillary column (30 m × 0.25 mm × 0.25 µm). Te injection volume was 1 μL with 20 : 1 (v/v) ratio split mode. Te carrier gas was helium (99.999% purity, 1 mL/ min). Te temperature of the injector was set as 280°C. Te initial oven temperature was kept at 60°C for 5 min; then, it was gradually raised to 120°C at 10°C/min and to 185°C at 2°C/min and kept for 3 min. Finally, it was raised to 220°C at 8°C/min. Te mass spectrometer was operated at 70 eV in full scan mode. Te compounds in MVO were identifed through both National Institute of Standards and Technology (NIST14) database and literature retrieval. Te relative content of each compound in the chromatogram was calculated by an area normalization method.

Prediction of Putative Targets of MVO.
According to the abovementioned GC-MS results, the compounds that appeared in at least 18 of 20 chromatograms were selected as characteristic compounds of MVO. Te Traditional Chinese Medicine Systems Pharmacology Database (TCMSP, https:// tcmsp-e.com/) and PubChem Database (https://pubchem. ncbi.nlm.nih.gov/) were introduced to collect the information of the chemical candidates. Furthermore, the SwissTargetPrediction Database (https://www. swisstargetprediction.ch/) was applied to predict the targets of the characteristic compounds action. Ten, all the obtained targets were summarized and reimported into the UniProt Database (https://www.uniprot.org/) to convert them into standard target gene names, and nonhuman and unverifed targets were deleted.

Target Prediction of MVO in the Treatment of AP and Construction of the Protein-Protein Interaction (PPI) Network.
In order to obtain more comprehensive information of the AP-related targets, a systematic search was conducted in    [46]. Te PPI network of the compound-AP targets was constructed by importing the overlapping targets into STRING database (https:// string-db.org/) [47]. Te species was set to "Homo sapiens," and the lowest confdence score was set to medium confdence (0.4).

Enrichment Analysis.
To better uncover the potential biological processes and pathways of MVO in AP treatment, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed by introducing the overlapping targets into Metascape database (https://metascape.org/). In our study, the p value was limited to less than 0.05.

Network Construction.
To systematically investigate the pharmacological mechanism of the active ingredients in MVO, a sample-compound-target network was constructed by using Cytoscape software (Version 3.7.2). In addition, a compound-target-pathway network was created by linking the pathways (directly relating to AP treatment mechanism), targets, and compounds using the Cytoscape software.

Molecular Docking
Simulation. Molecular docking simulation was applied to evaluate the interaction relationships between the core therapeutic targets and their corresponding compounds. Te protein crystal structures of each core target and the 3D structures of the compounds were downloaded from the RCSB Protein Data Bank database (PDB, https://www.rcsb.org/) and PubChem database (https://pubchem.ncbi.nlm.nih.gov/), respectively. After pretreatment of ligand (compounds) and receptor (protein crystals) structures, both of them were uploaded to Auto-Dock 4.0 and AutoDock Tools 1.5.6 for molecular docking. Te best docking conformation was picked out following the principles of low energy and reasonable conformation and was visualized using PyMOL.

Chemical Constituent Analysis of MVO.
Te compounds in MVO of 20 samples were identifed with GC-MS analysis, and a total of 23 common compounds were detected in the chromatograms of all MVO samples. Te 23 common compounds could not perfectly refect the compounds characteristics of MVO; hence, the sum of their relative content in nearly half of the sample was less than 80%; particularly in S5 and S19, the values were even lower than 65%, was 62.80% and 59.63%, respectively. In an attempt to get more feature information on the compounds in MVO, 35 main compounds appearing in at least 18 of 20 chromatograms were selected as characteristic compounds of MVO, among them 33 compounds were identifed. Te sum of their relative content was greater than 85% in each sample and consists of 87.51% to 96.41% of the total content. Tose 33 compounds could relatively comprehensively refect the characteristics of the inner main chemical compounds of MVO. Tus, the 33 characteristics compounds were regarded as potential active compounds of MVO in AP treatment and were further used for network pharmacology analysis (Table 1 and Figure 2).

Targets Prediction of MVO in AP Treatment.
In our study, a total of 165 targets of 33 potential active compounds abovementioned were obtained through TCSMP, PubChem, as well as SwissTargetPrediction database, and 9202 APrelated targets were collected from several databases including OMIM, GeneCards, and DrugBank database. Furthermore, Venny 2.1.0 was used for Venn analysis, and 142 overlapping targets of MVO in AP treatment were obtained ( Figure 3). Figure 4(a), according to the topological analysis, the top 8 compounds were identifed as the key active compounds:

Sample-Compound-Target Network. As shown in
, and β-eudesmol (degree � 51). Moreover, all the key active compounds correspond to more than 30 targets, suggesting that these compounds may have multitarget synergistic efects. Figure 4(b), there were 142 nodes and 710 interaction lines in the PPI network diagram with an average node degree of 10 and average aggregation coefcient of 0.458. Te darker the color and the wider the area of the node in the diagram indicate the more signifcant the node. According to the degree value, the top 8 targets (TNF, IL-1β, PPARc, PPARα, PTGS2, NCOA1, CNR1, and ESR1) of the PPI network were identifed as the core therapeutic targets with the degree value greater than or equal to 24, suggesting that these targets may play a crucial role in the network of MVO in AP treatment. Tus, the interaction behaviors between these 8 core therapeutic targets and their corresponding compounds were further evaluated by using molecular docking simulation.

GO and KEGG Pathway Enrichment
Analysis. 142 overlapping targets were introduced into Metascape for enrichment analysis. According to the GO enrichment S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20   Evidence-Based Complementary and Alternative Medicine results, the frst 10 biological processes (BPs), cellular component (CC), and molecular function (MF) were selected for visualization, as shown in Figure 5(a) (p value <0.05). For biological processes, the overlapping targets were mainly enriched in functions associated with response to organic cyclic compound, steroid metabolic process, ion homeostasis, regulation of MAPK signaling pathway, and so on. From the KEGG pathway results, the pathway with the most enriched genes is the neuroactive ligand-receptor interaction pathway (with 28 targets), then followed by pathways in cancer (with 20 targets), PPAR signaling pathway (with 11 targets), T17 cell diferentiation pathway (with 10 targets), cAMP signaling pathway (with 10 targets), calcium signaling pathway (with 9 targets), chemical carcinogenesis pathway (with 8 targets), hepatitis C (with 8 targets), estrogen signaling pathway (with 7 targets), and leishmania infection (with 6 targets) ( Figure 5(b)).

Compound-Target-Pathway Network.
To simplify the analysis, pathways directly relating to AP treatment mechanism, including neuroactive ligand-receptor interaction, estrogen, MAPK, and calcium signaling pathway were constructed into a compound-target-pathway network. Tose 4 pathways were associated with 48 targets (including 6 core therapeutic targets) and 30 potential active components (including 8 key active components) as shown in Figure 6.  Evidence-Based Complementary and Alternative Medicine targets. Docking results are shown in Table 2, a total of 51 pairs of core therapeutic targets (n � 8); their corresponding compounds of MVO (n � 25) complexes were investigated using molecular docking simulation. Most of the binding complexes displayed a relatively strong binding afnity, and the average binding energies of them are −6.16 kcal/mol, which suggested that the potential active compounds of MVO had better binding activity with the core therapeutic targets. Among them, 19 pairs belonged to the core therapeutic targets-key active compounds complexes, and the binding energies of which were from −4.56 kcal/mol to −7.42 kcal/mol. Te top 3 core therapeutic targets-key active compounds docking models are shown in Figure 7, including PTGS2-c-eudesmol (−7.42 kcal/mol), PPARαα-eudesmol (−7.32 kcaJ/mol), and PPARα-hinesol (−7.02 kcaJ/mol).

Discussion
Recently, more and more research studies have focused on TCM-based drug discovery and development to treat disease with complicated pathogenesis such as AP. In addition, the network pharmacology and molecular docking simulation have played a considerable role in the prediction of therapeutic mechanism in this process. In this paper, frst, the compounds in MVO were analyzed by GC-MS. Ten, a range of networks, including sample-compound-target network, PPI network of compound-AP targets, and compound-target-pathway network were constructed to reveal the mechanism of MVO in AP treatment by network pharmacology analyses. Finally, potential intermolecular interactions between the core therapeutic targets and their corresponding compounds based on network results were evaluated by molecular docking simulation.
Te types and amount of compounds in MVO were various with diferent origins, harvesting time, or processing conditions [48][49][50]. Terefore, information of the compound in MVO obtained from few online databases or the GC-MS analysis results of a single sample of MOC would not be representative and universality and might lead to a bias in the result of network pharmacology analysis. To solve this problem, in this paper, 20 batches of MOC were collected from the main production regions, such as Sichuan, Hubei, Chongqing, and Zhejiang, and the compounds in the volatile oil of MOC were analyzed by GC-MS. 33 compounds that appeared in at least 18 of 20 samples were identifed, representing 87.51%-96.41% of the total oil content. Among them, sesquiterpenes and oxygenated sesquiterpenes such as caryophyllene, caryophyllene oxide, calarene, α-, β-, and c-eudesmol are abundantly in MVO, accounting for 0.58-13.43, 2.02-10.14, 1.93-4.63, 14.52-25.99, 10.83-23.07, and 6.50-14.56%, respectively, which was consistent with the previous studies [51,52]. Terefore, those 33 compounds could relatively comprehensively and efectively refect on the characteristic of the inner main compounds of MVO and be used for further network pharmacology analyses.
Based on the network topological analysis, GO and KEGG analysis, there were 8 targets (TNF, IL-1β, PPARc, PPARα, PTGS2, NCOA1, CNR1, and ESR1) and 4 signaling pathways (neuroactive ligand-receptor interaction, estrogen, MAPK, and calcium signaling pathway) which have a very close association with the MVO in AP treatment.
Te acute infammatory response is one of the major pathological features of AP, leading to both the aggravation of local pancreatic damage and multiorgan failure of AP patient [57]. In the present study, there were three pathways (neuroactive ligand-receptor interaction, estrogen, and MAPK signaling pathway) which were associated with antiinfammatory, involving a total of 30 compounds and 47 corresponding targets (including 6 core targets: CNR1, NCOA1, ESR1, PPARc, TNF, and IL-1β).
Te neuroactive ligand-receptor interaction signaling pathway is a collection of all receptors and ligands related to intracellular signaling pathways and located on the plasma membrane [58]. Studies have shown that activation of multiple receptors in this signaling pathway such as α-, β-adrenergic receptor, dopamine receptor, or cholinergic receptors exerts anti-infammatory efects in AP [59]. NCOA1, ESR1, and ESR2 are enriched in the estrogen signaling pathway. Studies have shown that NCOA1 is involved in the acute infammatory process of AP through the regulation of IL-17 [60]. Both ESR1 and ESR2 can mediate anti-infammatory actions. Estradiol may prevent antiinfammatory process in the pancreas tissue via ESR1. Te agonists of ESR1 and ESR2 may reduce oxidative stress, infammatory, and pancreatic damage in the PBDL induced AP model [61].
As key signaling pathways of infammatory diseases such as AP, MAPK signaling pathway regulates the transcription of infammatory factors such as IL-1β, IL-6, TNF-α, and NF-κB [62][63][64]. It was reported that MAPK signaling pathway has a high level of participation the infammatory reaction in the development of AP from local to systemic and is considered as a potential target for AP treatment [64,65].

Evidence-Based Complementary and Alternative Medicine
TNF-α and IL-1β stimulate the production of other infammatory factors such as IL-8 and IL-6 and are regarded as the most prominent "First line" cytokines in AP [66]. TNF-α promotes both neutrophils and macrophages to accumulate in the pancreas, which leads to aggravate the local pancreatic infammation and develop as a systemic infammatory response [67]. Moreover, TNF-α could additionally interact with IL-1β to further induce or aggravate organ injury [66]. It was reported that both the survival state and survival ratio of rats with necrotizing pancreatitis were improved after injection of TNF-α antibody [68].
PTGS2 also known as COX-2, plays a key role in the development and severity of AP [69,70], which causes the release and activation of a large number of digestive enzymes in the pancreas as well as the disorders of pancreatic microcirculation, resulting in severe pathophysiological changes in the pancreas [71]. Mice that lack COX-2 genes have been reported to signifcantly decrease the severity of pancreatitis and pancreatitis-associated lung injury [69,70]. Selective inhibitors of COX-2 were found to reduce pancreatic injury in AP [71].
Sustained elevated levels of Ca 2+ in pancreatic cells are a marker for AP pathogenesis [72]. Tis overload Ca 2+ can lead to disorders of acinar cells secretion, oxidative stress, impaired microcirculation, early activation of zymogens, mitochondrial dysfunction, and cell death, which together mediate the development of AP [73]. Currently, inhibition of intracellular Ca 2+ overload has been considered as one of the most promising therapeutic approaches for AP treatment [74]. In this paper, a total of 9 therapeutic targets were mapped onto the calcium signaling pathway, and those 9 targets corresponded to 20 compounds of MVO, including 6 key compounds: (−)-terpinen-4-ol, terpineol, hinesol, α-, β-, and c-eudesmol.
Trough molecular docking simulation, it was found that the core therapeutic targets displayed good docking afnity for most of their corresponding compounds of MVO with an average binding energy of −6.16 kcal/mol. Notably, three eudesmol isoforms (α-, β, and c-eudesmol) have high binding activity to the core therapeutic targets such as PPARc, PPARα, PTGS2, or/and ESR1. Moreover, those three isoforms are also present at high levels in MVO, and the sum of relative content of them ranges from 31.85 to 63.62%. Consequently, we speculate that α-, β-, and c-eudesmol may play a signifcant role in AP treatment of MVO.
In this paper, we found that the therapeutic mechanism of MVO in AP treatment has the characteristics of multicomponents and multitargets, which mainly involved in 8 key active compounds, 8 core therapeutic targets, and 4 signaling pathway, and associated with anti-infammatory and inhibition of intracellular Ca 2+ overload. Te abovementioned study preliminarily revealed the active compounds and the potential mechanism of action of MVO in AP treatment, which provided enlightening information and basis for further exploration.

Conclusion
In this paper, the integrated strategy of GC-MS, network pharmacology, and molecular docking simulation was used to explore the active compounds as well as therapeutic mechanism of MVO in AP treatment. Te results showed that 8 key active compounds may play pivotal roles in both anti-infammatory and inhibition of intracellular Ca 2+ overload efects in AP treatment by acting primarily on TNF, IL-1β, PPARc, PPARα, PTGS2, NCOA1, CNR1, and ESR1 and involving in neuroactive ligand-receptor interaction, estrogen, MAPK, and calcium signaling pathway. Upon the abovementioned fndings, MVO may be a potential "multicomponents" and "multitargets" agent for AP treatment. Our research will be useful for developing a novel TCM-based drug for clinical AP treatment. However, further extensive experiments are still necessary to verify the therapeutic mechanism of MVO in AP treatment.

Data Availability
Te main data used to support the fndings of this study are included within the article.