Network Pharmacology Strategy to Investigate the Pharmacological Mechanism of Siwu Decoction on Primary Dysmenorrhea and Molecular Docking Verification

Objective . To study the pharmacological mechanisms of Siwu decoction (SWD) on primary dysmenorrhea (PDM) and verify with molecular docking. Methods . The Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) was utilized to acquire the active compounds and their corresponding target genes. The GeneCards database was utilized in the search for target genes that were associated with PDM. The intersection genes from the active target genes of SWD and those associated with PDM represented the active target genes of SWD that act on PDM. The Gene Ontology (GO) function enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were both carried out by RGUI 3.6.1 and Cytoscape 3.6.0 software. Cytoscape was also utilized for creating a compound-target network, and a protein-protein interaction (PPI) network was created through the STRING database. Molecular docking simulations of the macromolecular protein target receptors and their corresponding compounds were performed using AutoDockTool 1.5.6 and AutoDock Vina software. Results . We identiﬁed 14 active compounds as well as 97 active target genes of SWD by using the TCMSP. We compared the 97 active target genes of SWD to the 299 target genes related to PDM, and 23 active target genes for SWD that act on PDM which correlated with 11 active compounds were detected. The compound-target network as well as the PPI network were created, in addition to selecting the most essential compounds and their targets in order to create a key compound-target network. The most essential compounds were kaempferol, beta-sitosterol, stigmasterol, and myricanone. The key targets were AKT1, PTGS2, ESR1, AHR, CASP3, and PGR. Lastly, molecular docking was used to conﬁrm binding of the target with its corresponding compound. Conclusion . The pharmacological mechanisms of SWD that act on PDM were investigated, and the active compounds in the SWD for treating PDM were further veriﬁed.


Introduction
Primary dysmenorrhea (PDM) is defined as dysmenorrhea resulting from nonpelvic organic lesions [1]. e clinical manifestations of PDM include lower abdominal spasmodic pain during menstrual periods accompanied by headache, nausea, vomiting, and lumbar and leg pain [2].
Oral contraceptives and nonsteroidal anti-inflammatory drugs (NSAIDs) have been prescribed as first choice of treatment in PDM [3,4]. However, these drugs have adverse effects, and their long-term effects are less than optimal. Lately, there have been studies in which Chinese herbal medicine (CHM) was given as PDM therapy, and the results showed that certain curative effect was achieved [5][6][7].
e Siwu decoction (SWD), first used during the Tang Dynasty as the basic prescription and the first prescription in the treatment of gynecological diseases, primarily consists of baishao, chuanxiong, danggui, and shudihuang, with the Latin names Radix Paeoniae Alba (RPA), Rhizoma Chuanxiong (RC), Radix Angelicae Sinensis (RAS), and Radix Rehmanniae Preparata (RRP), respectively [8,9]. SWD is used for tonifying blood, activating blood, regulation of menstruation, and alleviating pain. A metaanalysis and systematic review reported the effects of SWD as treatment for PDM where beneficial and potentially safer in comparison to standard therapy [10].
In the present study, a network pharmacology approach was used to identify the pharmacological mechanism of SWD on PDM; in addition, we performed molecular docking for verification. We first identified both the active compounds of the four herbs as well as the target genes in SWD. en, the active target genes of SWD and the target genes related to PDM were used to detect the active target genes of SWD that act on PDM. Gene Ontology (GO) function enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment were both carried out. Subsequently, we constructed the compound-target network and the protein-protein interaction (PPI) network, and the most essential compounds and targets were screen out. Finally, we used these to create a key compound-target network, followed by performing molecular docking verification. Figure 1 illustrates the flowchart of our study.

Screening the Active Compounds in SWD.
Data on the compounds that can be found in the four herbs (RPA, RC, RAS, and RRP) were acquired from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (http://tcmspw.com/tcmsp.php) [11]. TCMSP is considered as a one-of-a-kind platform of systems pharmacology of CHM that includes details regarding the association between different diseases, medicine, and their targets. e oral bioavailability (OB) was preset at ≥30%, in addition to druglikeness (DL) at ≥0.18, as filter benchmarks during the screening for compounds regarded as active. OB indicates the corresponding quantity and pace at which the body absorbs the drugs into the bloodstream. DL is defined as the equivalence between compounds and previously identified drugs [12].

Screening the Target Genes of Active Compounds.
Data on the genes that are targeted by the active compounds were also retrieved from the TCMSP. In the search format "homo sapiens" was chosen, and we imported the target genes into the Universal Protein Resource (UniProt) knowledge base, an extensive and high-quality online source that contains information regarding annotation data and protein sequences (http://www.uniprot.org/) [13]. en, the human official gene symbols of these target genes were identified, and they were also considered as the proportion of target genes of SWD that are active.

Identification of PDM-Related Target Genes and Acquisition of Active Target Genes of SWD with Effects on PDM.
"Primary dysmenorrhea" was the key word that was used for searching the GeneCards database (https://www.genecards. org/) [14], a database providing comprehensive data on the predicted and annotated type of human genes. e PDMrelated target genes were searched and acquired. en, the intersecting genes derived from the active target genes of SWD and the PDM-related target genes were designated as the active target genes of SWD which have effects on PDM.

GO Function Enrichment and KEGG Pathway Enrichment
Analysis.
e RGUI 3.6.1 as well as org.Hs.eg.db packages were used for retrieval of the entrezIDs of the active target genes. e clusterProfiler package, RGUI, and Cytoscape 3.6.0 software with the plugin apps, CluePedia and ClueGO, were employed to carry out the GO function enrichment analysis as well as analysis of the KEGG pathway enrichment. e GO function enrichment consisted of the analysis of the biological process (BP), molecular function (MF), and cellular component (CC) [15][16][17][18].

Development of the Compound-Target Network and Selection of Key Compounds.
Cytoscape software with the incorporated NetworkAnalyzer tool function was used to develop as well as evaluate a network of compounds and targets [16]. e nodes depict the target genes and compounds, while the edges depict the correlations that exist between them. Based on the degree of the relationship between the target genes and compounds, the key compounds in the SWD that acted on PDM were selected.

Development of PPI Network and Designation of the Key Targets.
A PPI network consisting of the SWD that acts on PDM was constructed after introducing the active target genes into the STRING database, which is used for the discovery of functional interactions within genomewide experimental datasets (https://string-db.org/) [19]. We created a PPI network by setting the type of species being researched as "homo sapiens" and the interaction score at a minimum value of 0.4. en, we used the PPI network to make a topology analysis, and Cytoscape and its Network Analyzer tool were used to determine the exact degree values of every target gene in order to select the key targets of SWD that act on PDM.

Verification of Molecular
Docking. Molecular docking was utilized in order to confirm the binding of target and its corresponding compound. Data on the construction of the small-molecule compounds was acquired via the PubChem Database (https://pubchem.ncbi.nlm.nih.gov/), and the macromolecular protein target receptors were retrieved via the RCSB PDB database (http://www.rcsb.org/). Molecular docking simulations of macromolecular protein target receptors and their corresponding compounds were performed using AutoDockTool 1.5.6 and AutoDock Vina software [20].

Statistical Analysis.
A part of the statistical analysis was simultaneously conducted with the biotechnology add-ons of the software and platforms that have been disclosed in previous sections. In the GO function and KEGG pathway enrichment, an adjusted P (adj. P) value was used and adj. P < 0.05 was perceived as statistically significant.

Screened Active Compounds of SWD.
A total number of 86 compounds were obtained from RPA, 189 compounds from RC, 125 compounds from RAS, and 76 compounds from RRP using the TCMSP (Supplementary Files 1A-1D). By adjusting the filter criteria to OB ≥30% as well as DL ≥0.18, the following number of active compounds was selected: 13 from RPA, 7 from RC, 2 from RAS, and 2 from RRP. Lastly, 20 active compounds of SWD were verified with the exclusion of duplications. Table 1 contains the general data on the active compounds in the SWD.

Attained PDM-Related Target Genes and Identified Active
Target Genes of SWD that Act on PDM. "Primary dysmenorrhea" was the key word that was used to perform a search in the GeneCards database, which resulted in the identification of 299 PDM-related target genes (Supplementary File 3A and 3B).
A comparison was conducted between the previously identified 97 active target genes of SWD and the 299 PDMrelated target genes, leading to the identification of 23 active target genes of SWD that act on PDM ( Figure 2 and Table 2).

e GO Function Enrichment and KEGG Pathway Enrichment Analysis.
e entrez IDs of the active target genes of SWD that act on PDM were retrieved through RGUI as well as org.Hs.eg.db (Table 2). ereafter, we conducted the GO function enrichment and KEGG pathway enrichment analyses with the RGUI and clusterProfiler package. e GO function enrichment analysis of BP indicated the following: active target genes of SWD that act on PDM have been enriched significantly as reaction to toxic substances, the cellular response to xenobiotic stimulus, the metabolic processes of long-chain fatty acids, fatty acids, and unsaturated fatty acids, the response to xenobiotic stimulus, the xenobiotic metabolic process, the maternal process related to female pregnancy, the long-chain fatty acid biosynthetic process, the response to osmotic stress, and other processes (Supplementary File 4A). e 20 most enriched GO BP are visualized in Figure 3(a) and ranked according to their adj. P values.
e GO function enrichment analysis of MF indicated the following: the active target genes of SWD that act on    Figure 3(b) and ranked according to their adj. P values.
e GO function enrichment analysis of CC indicated the following: the active target genes of SWD that act on PDM have been enriched significantly in membrane raft, membrane microdomain, and membrane region (Supplementary File 4C). e three most enriched GO CCs are visualized in Figure 3(c) and ranked according to their adj. P values.
e KEGG pathway enrichment analysis has shown that active target genes of SWD that act on PDM were enriched significantly in platinum drug resistance, xenobiotics metabolism induced by cytochrome P450, endocrine resistance, steroid hormone biosynthesis, arachidonic acid metabolism, metabolism of drugs by cytochrome P450, metabolism of drugs by other enzymes, and epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor resistance (Supplementary File 4D). e eight most enriched KEGG pathways are visualized in Figure 3(d) and ranked according to their adj P values; in addition, the KEGG pathways are shown in Supplementary File 5.
By using ClueGO and CluePedia plugin apps, found in Cytoscape software, the GO function enrichment as well as KEGG pathway enrichment were visualized extra naturally ( Figure 4).

Constructed Compound-Target Network and Selection of Key Compounds.
We utilized Cytoscape software to create a compound-target network and the Network Analyzer tool to perform the analysis. As MOL001918, MOL001924, and MOL000433 had no correspondences to an overlapping target gene, the 23 overlapping active target genes correlated with the following 11 active compounds: 6 from RPA, 5 from RC, 2 from RAS, and 2 from RRP (Supplementary File 6A).
ere were 34 nodes (11 compound nodes as well as 23 target gene nodes) and 62 edges in total identified in the network (Supplementary File 6B, Figure 5). e top four compounds ranked according to their degree within the network were MOL000422 (kaempferol), MOL000358 (beta-sitosterol), MOL000449 (stigmasterol), and MOL002135 (myricanone), which can be regarded as the essential compounds in SWD that act on PDM. General information of the essential compounds can be found in Table 3, and the 2D structure of these compounds acquired from the PubChem Database is shown in Figure 6.

Construction the PPI Network and Selecting Key Targets.
e PPI network was obtained by importing a total of 23 overlapping active target genes into the STRING database. After setting the interaction score to a minimum value of 0.40, 23 target proteins with interactions were identified in   Figure 7). e target genes that were ranked as top six in the network according to degree were selected and regarded as the key targets of SWD that act on PDM (Table 4).

Constructed Key Compound-Target Network. After introducing the key compounds (small-molecule compounds)
and key targets (macromolecular protein target receptors) and their relationships to Cytoscape 3.6.0 software, a key compound-target network could be constructed (Supplementary File 8, Figure 8).

Verification Using Molecular Docking. Data on the 3D
compositions of the small-molecule compounds were acquired through the PubChem Database and that of the macromolecular protein target receptors through the RCSB PDB database. en, AutoDockTool and AutoDock Vina software were used to perform simulations of the molecular docking of potential targets and their designated compounds. Finally, we performed molecular docking to verify the binding of target and its designated component. e simulations of the molecular docking of AKT1-kaempferol are shown in Figure 9, and those of CASP3-beta-sitosterol are shown in Figure 10. e molecular docking simulations of all of the key targets and key compounds are shown in Supplementary File 9.

Discussion
PDM is characterized by cold coagulation, qi stagnation, blood stasis, dampness and heat, and a shortage of blood and qi according to the Traditional Chinese Medicine theory, which considers a blood and qi disorder as the primary etiology of PDM and accentuates that patients should be treated in accordance to these aspects [21,22]. According to modern medicine, the pathogenesis of PDM is primarily related to prostaglandin, vasopressin, oxytocin, estradiol, and endothelin, and it is closely related to immune function and endocrine dysfunction [23,24].
SWD has been considered as the basic prescription and the first prescription in the treatment of gynecological diseases, which consists of RPA, RC, RAS, and RRP. RPA tonifies the blood to collect yin and calms the liver to relieve pain. RC moves qi to relieve pain. RAS tonifies and harmonizes the blood and regulates menstruation to relieve pain. Finally, RRP tonifies the blood and nourishes yin [8,9].
At present, there is a wide application of network pharmacology in research related to CHM. It supports the methodology of researching network-based as well as multicomponent therapies, which is compatible with the multipathway, multitarget, and multicomponent features of CHM [25,26].
In this study, 23 active target genes of SWD that act on PDM were detected. e GO BP function enrichment analysis of SWD as PDM treatment were significantly enriched as reaction to numerous processes as described in the results section and shown in Supplementary File 4A and Figure 3(a). e GO analysis of MF showed that SWD as PDM treatment involved many different activities as described previously in the results section and Supplementary File 4B and Figure 3(b). e results of the GO CC for SWD as PDM treatment are shown in Supplementary File 4C and Figure 3(c). ese results indicated that these enrichments Circles represent active compounds (red represent compounds from RPA, bule represents compounds from RC, green represents compounds from RAS, and yellow represent compounds from RRP), rectangles represent the active target genes, and the edges represent links between the nodes.    Evidence-Based Complementary and Alternative Medicine are closely related to immunity, metabolism, and endocrine activity, which are all strongly associated to the mechanisms of disease in PDM [27][28][29][30]. e results of the KEGG enrichment pathway analysis also demonstrated that numerous pathways were strongly associated to the pathological mechanisms of PDM as shown in Supplementary Files 4D and 5 as well as Figure 3(d). ese results have clearly demonstrated that SWD can act on PDM via various pathways. ese pathways as well as relevant target genes are important and should be further studied.
In the present study, a network of compound-target network was created, followed by the identification of the essential compounds of SWD that act on PDM. e following essential compounds were identified: kaempferol, beta-sitosterol, stigmasterol, and myricanone. Kaempferol can be used to treat numerous acute and chronic inflammation-induced diseases due to its anti-inflammatory properties [31]. Beta-sitosterol has anti-inflammatory, anthelminthic, and antimutagenic activities [32].
Stigmasterol is an organic steroid alcohol that possesses the well-known ability of immune modulation [33]. Myricanone has been reported to have apoptosis-promoting abilities [34]. ese studies were related to the regulation of immunity and endocrine function. e SWD action on PDM could be the result of the interaction of these multiple compounds. However, there are few related studies regarding the single compound action on PDM, which requires further investigation.
e PPI network showed that the actions of SWD on PDM were associated with various targets. Of which, the most essential targets were AKT1, PTGS2, ESR1, AHR, CASP3, and PGR. Previous research indicated that the levels of PTGS2 and ESR1 were abnormal in an estradiol benzoateoxytocin induced dysmenorrhea mice model [9]. It has been shown that the ESR1 gene is statistically significant different in PDM patients and controls [35]. e polymorphisms PGR-CYP17A1-CYP19A1 have been shown to have a genegene interaction when there is a risk of developing  endometriosis [36]. e relationship between these genes and PDM, some of which already have relevant studies, can be further studied, and the possible mechanisms can be explored, which have not previously been investigated.
Molecular docking was also conducted to identify specific interactions among key compounds and their predicted protein targets, which could improve the network's accuracy [37]. e preliminary molecular docking results showed kaempferol, beta-sitosterol, stigmasterol, and myricanone, the key active compounds in RPA, RC, RAS, and RRP, had high binding activities with AKT1, PTGS2, ESR1, AHR, CASP3, and PGR gene target proteins.
ese active compounds may be important foundational materials in SWD for treating PDM through related signaling pathways.
A network pharmacology approach was adopted with the intention of studying the various pharmacological mechanisms of SWD that act on PDM, and the binding of the target and its corresponding compound were verified using molecular docking. However, there are some limitations using these approaches. First, we used the TCMSP database to detect the active compounds of SWD and their corresponding target genes. e criteria for screening the active compounds were fixed, and the target genes associated with PDM were acquired through the GeneCards database. Even though at present these databases are increasingly comprehensive, some compounds and target genes may still have been omitted. Second, the GO function enrichment and KEGG pathway enrichment analyses were undertaken, in addition to the construction of a PPI network to investigate the pathways and target genes of the SWD acting on PDM. ese potential target genes and pathways require further study using experimental analyses.
ird, preliminary molecular docking verification was only conducted in this study, and research regarding the molecular docking of small-molecule compounds and macromolecular protein target receptors should be further studied.

Conclusion
In conclusion, the interactions among the active compounds, pathways, and target genes of SWD acting on PDM, via the effects of various compounds, targets, and pathways, were investigated in the present study. In addition, molecular docking was conducted lastly to identify specific interactions between key compounds and their predicted target proteins, which indicated these compounds may be an important material foundation in the SWD for treating PDM. In sum, the pharmacological mechanisms of SWD that act on PDM were investigated, and the active compounds in the SWD for treating PDM were further verified.

Data Availability
e data used in this study can be found online as supplementary files (http://www.mediafire.com/file/ 87m1y65c596wxbt/Supplementary_Files.zip/file).

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
Dandan Jiang and Xiaoyan Wang created and designed this study and wrote the manuscript. Dandan Jiang and Yufeng Zhang collated and extracted data. Dandan Jiang, Lijun Tian, and Yufeng Zhang conducted analysis of the data. Lijun Tian and Yufeng Zhang supervised and conducted project administration. All the authors have read and authorized the final version of this paper.

Acknowledgments
First, we would like to acknowledge the software and platform that provided the bioinformatic tools. Second, this study was supported in part by the Nantong Science and Technology Plan Project (grant MS12017004-2, recipient Tian) as well as the Research Grant of Jiangyin Hospital of Traditional Chinese Medicine (recipient Zhang). Lastly, we would like to thank LetPub (http://www.letpub.com) for its linguistic guidance while preparing this paper.

Supplementary Materials
Supplementary files can be found online (). Supplementary