Network Pharmacology and Molecular Docking Analyses Unveil the Mechanisms of Yiguanjian Decoction against Parkinson's Disease from Inner/Outer Brain Perspective

Objective This study aims to explore the pharmacodynamic mechanism of Yiguanjian (YGJ) decoction against Parkinson's disease (PD) through integrating the central nervous (inner brain) and peripheral system (outer brain) relationship spectrum. Methods The active components of YGJ were achieved from the TCMSP, TCMID, and TCM@Taiwan databases. The blood-brain barrier (BBB) permeability of the active components along with their corresponding targets was evaluated utilizing the existing website, namely, SwissADME and SwissTargetPrediction. The targets of PD were determined through database retrieval. The interaction network was constructed upon the STRING database, followed by the visualization using Cytoscape software. Then, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses on potential targets. Finally, the molecular docking approach was employed to assess the binding affinity between key components and key targets. Results Overall, we identified 79 active components, 128 potential targets of YGJ, and 97 potential targets of YGJ-BBB potentially suitable for the treatment of PD. GO and KEGG analyses showed that the YGJ treatment of PD mainly relied on PI3K-Akt pathway while the YGJ-BBB was mostly involved in endocrine resistance. The molecular docking results displayed high affinity between multiple compounds and targets in accordance with previous observations. Conclusions Our study unveiled the potential mechanisms of YGJ against PD from a systemic perspective: (1) for the YGJ, they have potential exerting effects on the peripheral system and inhibiting neuronal apoptosis through regulating the PI3K-Akt pathway; (2) for the YGJ-BBB, they can directly modulate endocrine resistance of the central nervous and holistically enhance body resistance to PD along with YGJ on PI3K-Akt pathway.


Introduction
Parkinson's disease (PD) is the second most common neurodegenerative disease in the world [1]. Its pathology is characterized by the loss of neurons in the substantia nigra leading to a decrease in dopamine (DA) transmitters in the striatum and the formation of Lewy bodies containing α-synuclein (α-syn) [2]. Clinically, symptoms of PD include typical motor symptoms (bradykinesia, rigidity, resting tremor, and gait disturbances) and non-motor symptoms (sleep disturbances, olfactory disturbances, autonomic dysfunction, and cognitive and psychiatric disturbances) [3]. Currently, levodopa is the standard of care for PD treatment. However, levodopa does not completely cure PD, and long-term treatment is often accompanied by side effects [4]. Therefore, the research of alternative drugs is of great importance for PD treatment.
Although the YGJ has obvious therapeutic effect against PD, there is still no systematic study about this. Network pharmacology, as part of bioinformatics technology, inte-grates systems biology and computational biology [11]. With the rise of technologies such as molecular docking, molecular dynamics simulations, and bioinformatics, in silico strategy has emerged. It can reveal the relationship between molecular monomers with biological pathways or specific diseases by comprehensively studying and expanding the intersection between molecular monomers [12][13][14].    BioMed Research International Sitosterol alpha1

BioMed Research International
Network pharmacology allows the relationships between TCM and disease to be explored as a whole, and the mechanisms between TCM and disease can be systematically revealed and thus provide a systematic approach for the study of TCM treatment of diseases [15]. According to this, we conducted this study to explore the therapeutic mechanisms of YGJ against PD for the first time through the approach of network pharmacology and molecular docking.
In our approach, we assigned all active components to the YGJ group and placed the active components crossing the BBB to the YGJ-BBB group. Through this operation, we can not only explore the effects of all active components of YGJ on peripheral system (outer brain) but also screened out the components that can cross the BBB to specifically explore the therapeutic effects of YGJ on central nervous (inner brain), in order to comprehensively analyze the pharmacodynamic mechanisms of YGJ against PD. Furthermore, a visible graphical abstract for the current flowchart is provided to demonstrate the mechanisms of YGJ against PD concerning central nervous and peripheral systems ( Figure 1).

2.1.
Screening of Active Components. The components of YGJ were collected from multiple databases, including the Traditional Chinese Medicine System Pharmacology Database and Analysis Platform (TCMSP, https://old.tcmsp-e .com/tcmsp.php), Traditional Chinese Medicine Integrated Database (TCMID, http://www.megabionet.org/tcmid/), and Traditional Chinese Medicine Database @ Taiwan (TCM@Taiwan, http://tcm.cmu.edu.tw/index.php). Oral bioavailability (OB) is one of the most important parameters of pharmacokinetics, and the higher the OB value, the better the drug-likeness (DL) of the active component [16]. The Caco-2 screening assay is a valuable tool for testing compounds for intestinal permeability [17]. In this study, the criteria of OB ≥30%, DL ≥0.18, and Caco-2 ≥ -0.4 were used to   [18]. The canonical SMILES of components from TCMID and TCM@Taiwan databases were imported into SwissADME website (http://www.swissadme.ch/) to analyze the bioavailability score. The component with a bioavailability score greater than 0.5 was considered the active component [19,20]. All active components obtained from the above 3 databases were assigned to the YGJ group.

Screening of Active Components
Crossing the BBB. The canonical SMILES of all active components obtained from the above 3 databases were imported into Swiss ADME website to analyze whether they can cross the BBB. The active components crossing the BBB were placed to the YGJ-BBB group.
2.4. Acquisition of Potential Targets against PD. The canonical SMILES of active components were imported into Swiss Target Prediction website (http://www.swisstargetprediction .ch/) to predict targets. The identified PD targets were intersected with the targets of YGJ and YGJ-BBB using the EVenn website tool (http://www.ehbio.com/test/venn/#/) [21]. The intersection targets were considered the potential targets of YGJ and YGJ-BBB against PD.

Construction of Herb-Active Component-Potential
Target Network. The active components and potential targets were imported into Cytoscape 3.9.1 software to construct the "Herb-Active component-Potential target" interactive networks, which can show the interconnection between herbs, their corresponding active components, and potential targets [15]. The Analyze Network tool was used for correlation analyses, which can perform topology analyses on each node in the network, store the calculated values as attributes of the corresponding nodes and edges, and can filter network nodes based on the calculated topology metrics. The key components could be selected after sorting the degree value.

Establishment of PPI Network and Enrichment Analyses.
The potential targets were input into the STRING website (https://cn.string-db.org/) to construct the protein-protein interaction (PPI) networks. The organism criterion was set as Homo sapiens and the minimum required interaction score was fixed at 0.4. The Molecular Complex Detection (MCODE) algorithm, which can detect densely connected regions that are likely to represent molecular complexes in large PPI networks, was based solely on connectivity data [22]. For the MCODE analyses, the resulting TSV files were downloaded from the STRING database and imported into Cytoscape software, and the MCODE plug-in was used to identify the top 3 clusters of each group.

Molecular Docking
Validation. The potential targets from the KEGG pathway with the highest degree of each group were considered to be the key targets [23]. The procedure of molecular docking was as follows: (1) Target file preparation: The three-dimensional (3D) structures of key targets were retrieved from PDB database (https://www.rcsb.org/) and UniProt website (https://www.uniprot.org/) [24]. Water and ligands of the structure were removed using the PyMOL viewer software, hydrogen bonds were added on the structure by AutoDockTools software, and the structure was saved as a PDBQT file [25] (2) Component file preparation: The 3D structures of key components were downloaded from TCMSP and Pubchem database (https://pubchem.ncbi.nlm .nih.gov/). In AutoDockTools, we input the structure of the component as a ligand and set the structure as follows: delete root, show root expansion, and choose torsions. Then, exported the structure to a ligand file in PDBQT format [26] (3) Constructing grid box: Imported the PDBQT structures of target and ligand into AutoDock4 and  9 BioMed Research International defined the grid box of molecular docking. The target structure was used as the grid's center, and the center coordinates (center x/y/z) and box size (size x/y/z) parameters were adjusted to ensure that the grid box completely covers the target structure [27,28] (4) Molecular docking and visualization: Autogrid program was run for the first docking operation. After Genetic Algorithm was used for the calculating, the Autodock program was run for the second docking operation [29]. The binding energy was calculated and successful docking was defined by a binding energy of ≤ -5 kJ/mol [30].  Table 1, and common active components are listed in Table 2. A total of 16 active components that can cross the BBB were identified. To be specific, the number of candidate compounds in BSS, GQZ, and MD was 5, 3, and 8, respectively (Table 3). The results showed that YGJ may synergistically exert the therapeutic effects against PD through multiple potential targets.

Construction and Analyses of Interactive Network.
After the potential targets were obtained, the specific circumstances of the association of each active component with potential targets remain unknown. The "Herb-Active component-Potential target" interactive networks are shown in Figure 3, and 14 active components with degree ≥20 and 12 active components of YGJ-BBB with degree ≥15 were identified to be the key active components [31]; the results are listed in Table 4.

PPI Network
Analyses. There were interactions between potential targets, and each target involved numerous functional pathways; these were all included in the effective mechanisms. We first established the PPI networks to explore the interactions between potential targets. The PPI network of YGJ consists of 128 nodes and 1398 edges, with an average node degree of 21.8 (Figure 4(a)). The PPI network of YGJ-BBB consists of 97 nodes and 861 edges, with an average node degree of 17.8 (Figure 4(b)). The potential targets with high degree in each group are shown in Figures 4(c) and 4(d), and the degree was listed in Supplementary Materials table S1. The top 3 MCODE clusters in each group were screened out (Figures 4(e) and 4(f)), and the details of each cluster are listed in Table 5. According to the results, the clusters of YGJ were related to DA synapse, EGFR tyrosine kinase inhibitor resistance, and calcium signaling pathway; the clusters of YGJ-BBB were related to Rap1 signaling pathway, MAPK signaling pathway, and HIF-1 signaling pathway.

Enrichment
Analyses. GO items with counts greater than 20 of YGJ are shown in Figure 5(a), with a maximum of 20 items included in each analysis. The biological process (BP) results showed that cellular response to nitrogen compound had the highest count; the molecular function (MF) results were mainly focused on phosphotransferase activity, alcohol group as acceptor; the cellular components (CC) results mainly related to dendrite. Top 20 KEGG pathways of YGJ were sorted by P value and shown in Figure 5(b), and the    13 BioMed Research International results were mainly concerned in PI3K-Akt signaling pathway. Given the highly distracting capacity of pathways of cancer, relevant cancer pathways were excluded accordingly.
GO items with counts greater than 20 of YGJ-BBB are shown in Figure 5(c), with a maximum of 20 items included in each analysis. The BP results showed that behavior had the highest count; the MF results were mainly focused on protein kinase activity; the CC results mainly related to dendrite. Top 20 KEGG pathways of YGJ-BBB were sorted by P value and shown in Figure 5(d), and the results were mainly concerned in endocrine resistance. Relevant cancer pathways were also excluded accordingly.    Table 6, and the PDB IDs were listed in Supplementary Materials table S2. The key targets were docked crossly with the key components of each group to predict the effective mechanisms. The successful docking results are listed in Table 7, and 3 representative successful results of

18
BioMed Research International disease progression [32][33][34]. Oxidative stress can not only participate in the formation of amyloid by affecting the structure and self-assembly of α-syn but also be interrelated and interdependent with the inflammatory process [35]. Multiple causes of mitochondrial damage-mediated apoptosis of DA neurons are also associated with the development of PD [36]. Therefore, inhibition of the inflammatory response, oxidative stress, and apoptosis of neuronal cells through systemic effects of the body or interventions targeting the brain has significant therapeutic effects on PD.
Multiple active components screened in this study have been confirmed to be able to exert therapeutic effects on PD through the above-mentioned approaches. Among the active components crossing the BBB, isoimperatorin can inhibit the nuclear factor (NF)-κB pathway, reduce the levels of interleukin (IL)-4, IL-5, IL-6, IL-13, and tumor necrosis factor (TNF)-α, and possess the anti-inflammation effect [37]. Methylophiopogonanone B has antioxidant property, can inhibit the production of malondialdehyde (MDA) and reactive oxygen species (ROS), enhance the activity of superoxide dismutase (SOD), and down-regulate the expressions of Bax/Bcl-2 and caspase-3 to inhibit apoptosis, while can also significantly down-regulate the expression of IL-6 and IL-8 [38]. Ruscogenin can inhibit the activity of the NF-κB pathway to reduce the expression of inflammatory cytokines, including IL-1β and caspase-1, and also reduce the production of ROS, making it both anti-inflammatory and antioxidant properties [39]. Thus, active components crossing the BBB can inhibit inflammatory responses, oxidative stress, and neuronal apoptosis to treat PD.
The MCODE algorithm can detect densely connected regions likely to represent molecular complexes in large PPI networks, based solely on connectivity data [40]. The top 3 clusters of YGJ were related to DA synapse, EGFR tyrosine kinase inhibitor resistance, and calcium signaling pathway. Many disorders with control deficits are associated with abnormal dopamine transmission [24]. Improvement of DA transmission abnormality and elevation of DA level may contribute to rapid reversal of motor complications of PD [41,42]. Epidermal growth factor receptor (EGFR) is a tyrosine kinase receptor involved in cell differentiation and proliferation, and its mutation and amplification are associated with the pathology of neurodegenerative diseases [43,44]. Amyloid can be transported between neuronal cells through EGFR-mediated endocytosis, thereby enabling amyloid transfer throughout the brain [45,46]. Calcium can bind to the C-terminus of αsyn and promote the secretion and aggregation of α-syn [47,48]. Activation of the calcium signaling pathway can promote

20
BioMed Research International the generation of mitochondrial oxidative stress and the apoptosis of substantia nigra DA neurons [49,50]. Therefore, YGJ can improve the transmission abnormality of DA and amyloid and inhibit the secretion and aggregation of α-syn through systemic effects. The top 3 clusters of YGJ-BBB were related to Rap1 signaling pathway, MAPK signaling pathway, and HIF-1 signaling pathway. Activation of the Rasassociation proximate 1 (Rap1) and mitogen-activated protein kinase (MAPK) signaling pathways has been shown to be able to promote inflammatory responses and ROS production [51][52][53][54], while activation of the hypoxia-inducible factor-1 (HIF-1) signaling pathway could inhibit the mitochondriamediated apoptotic process [55]. The MAPK pathway can be activated by its upstream Rap1 pathway, which in turn activates the downstream HIF-1 pathway [56]. Therefore, downregulation of Rap1, MAPK, and HIF-1 pathways can reduce neuronal cell apoptosis by inhibiting inflammatory responses, oxidative stress, and mitochondrial damage in the brain. The efficacy of TCM is represented by the synergy effects of potential targets, but experimental validation would yield massive work and thus hinder our progression when we scrutinized the TCM-targets effects relationship spectrum. KEGG enrichment analysis can find significant signal pathways for the synergy effects of potential targets. Therefore, we need to step-wisely narrow our study objects through KEGG enrichment analysis. According to the results of the KEGG pathway enrichment analyses, YGJ group was mainly focused on PI3K-Akt signaling pathway, while YGJ-BBB group was mainly concerned in endocrine resistance. Activating the PI3K-Akt cell survival pathway can inhibit mitochondrial damage-mediated apoptosis, promote autophagy of amyloid, and play a neuroprotective role [57][58][59]. In the meantime, the activation of the PI3K-Akt pathway can also inhibit inflammation response and oxidative stress [60,61]. EGFR and ERBB2 are upstream ligands [62,63], while CCND1, MTOR, and VEGFA are downstream receptors of PI3K-Akt signaling pathway [64][65][66], and interventions on targets above-mentioned can all have effects on this pathway. The MAPK pathway and the PI3K-Akt pathway have extensive interactions, and the PI3K-Akt pathway can be regulated by acting on the important targets of the MAPK pathway, MAP2K1 and MAPK1 [67,68]. According to the docking results, it is speculated that YGJ may act on CCND1, EGFR, ERBB2, MAP2K1, MAPK1, MTOR, and VEGFA targets to modulate PI3K-Akt signaling pathway, thereby treating PD by affecting the peripheral system.
Studies have shown that the regulation of the endocrine system also plays an important role in the treatment of PD, including sex hormones, insulin, and melatonin [69][70][71][72]. According to the docking results, it is speculated that YGJ-BBB may act on ERBB2, ESR1, ESR2, IGF1R, MAP2K1, and MAPK1 targets to intervene endocrine resistance. Estrogen has been shown to increase the synthesis, release, reuptake, and turnover of DA [73], prevent memory impairment by inhibiting NF-κB activity to modulate neurogenic inflammation [74], and improve mitochondrial damage and restore the activity of antioxidant enzymes [75], thereby reducing the risk of PD and improving symptoms of disorders with control deficits [76,77]. The docking results showed that YGJ-BBB can act on estrogen receptor 1 (ESR1) and estrogen receptor 2 (ESR2), improve the utilization rate of estrogen, and promote the therapeutic effect on PD. There is accumulating evidence that insulin can cross the BBB and influence a multitude of processes in the brain, including modulation of neuronal survival and growth [78], DA transmission [79], maintenance of synapses [80], autophagy of amyloid [81], oxidative stress [82], and neuroinflammation [83]. Meanwhile, a process analogous to insulin resistance exists in the brains of PD patients. Therefore, improving the utilization and restoring the normal function of insulin in the brain could be used as a strategy to slow the progression of PD. The docking results showed that YGJ-BBB can act on insulin-like growth factor 1 receptor (IGF1R) to achieve this. In addition, MAPK signaling pathway is the classical pathway that can be activated by estrogen and insulin. Intervention of the important targets MAP2K1 and MAPK1 in this pathway and the upstream pathway target ERBB2 can affect the function of estrogen and insulin. That is to say, the active components of YGJ that cross the BBB can restore the normal function of estrogen and insulin in the brain by acting on ERBB2, ESR1, ESR2, IGF1R, MAP2K1, and MAPK1, thereby reducing the incidence,

21
BioMed Research International slowing down the progression, and improving related symptoms of PD by affecting the central nervous. It is worth noting that the expression of PI3K-Akt signaling pathway is also closely related to the effects of estrogen and insulin [84][85][86], which indicates that the active components of YGJ can also enhance the therapeutic effects of the active components that cross the BBB on central nervous through peripheral system effects triggered by PI3K-Akt pathway.
Despite the encouraging discoveries, there still exist certain limitations. To begin with, our study is largely constructed upon current databases. However, scientific studies regarding TCM against PD remain insufficient compared to other diseases and thus require further attention. This can be reflected by the KEGG enrichment analyses in which the overlay between disease pathways could induce inadequate inference. Moreover, the comprehensive compounds of herbs in the YGJ are still missing. Those unidentified and unrecorded components were not included and might affect the present result. Although we take a step forward in systemic explanation regarding TCM mechanism from central nervous (inner brain) and peripheral system (outer brain), the determined bioactive component and their corresponding targets cannot fully represent the holistic concept of YGJ. Since the efficacy of YGJ has been clinically testified, we believe that the further advance in computational techniques combined with experimental validation can benefit the exploration of underlying mechanism on YGJ against PD.

Conclusion
In conclusion, 79 active components were screened in this study, of which 16 active components can permeate the BBB. Overall, we identified 128 potential targets of YGJ, 97 potential targets of YGJ-BBB potentially suitable for the treatment of PD. The herb-active component-potential target network, PPI networks, and MCODE networks were constructed through the approach of network pharmacology. It is speculated that orchinol, N-coumaroyltyramine, (S)-p-coumaroyloctopamine, and melianone can bind with the targets related to PI3K-Akt and its upstream and downstream pathways to treat PD by affecting the peripheral system. And bergaptin, ammidin, and ruscogenin, the active components of YGJ that cross the BBB, can bind with targets related to the regulation of multiple hormones to modulate endocrine resistance, such as restoring the normal function of estrogen and insulin to exert therapeutic effects on the central nervous, which can also be enhanced through peripheral system effects of PI3K-Akt pathway. Therefore, not only YGJ can have different therapeutic effects on PD through the effects of peripheral system and central nervous but also the two approaches can work in coordination, thus reflecting the systematic collaboration and singletargetedness of YGJ against PD.

Data Availability
The data used to support the findings of this study are included within the article and supplementary materials.

Ethical Approval
This study did not require ethical approval as no relevant experiments were involved.

Conflicts of Interest
No potential conflict of interest was reported by the authors.

Acknowledgments
Thanks for the assistance of Shandong University of Traditional Chinese Medicine and the research and experimental platform of professor Yiider Tseng. Table S1. Degree of the potential targets of YGJ and YGJ-BBB. Table S2. PDB ID of key targets.