A Network Pharmacology-Based Study on Active Ingredients of Corydalis Rhizoma on Coronary Heart Disease

Corydalis Rhizoma(CR) showed a high ecacy for coronary heart disease (CHD). However, the interaction between the active ingredients of CR and the targets of CHD has not been unequivocally explained in previous researches. To study the active components and potential targets of Corydalis Rhizoma and to determine the mechanism underlying the exact effect of Corydalis Rhizoma on coronary heart disease, a method of network pharmacology was used.


Materials And Methods
In this study, we constructed an active components of CR-CHD disease targets network which can directly observed the distribution of active components acting on CHD related targets. Moreover, construction of PPI network enabled us to reveal interactions of relative target proteins, by which we got the core genes with calculation. At last, Gene Ontology(GO) function and Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway enrichment analysis were performed for functional annotation of the relative target genes in CR-CHD network.
2.1.Establishment of active components of CR-relative targets database. Traditional Chinese medicine systems pharmacology database and analysis platform(TCMSP, http://tcmspw.com/index.php, last updated on May 31, 2014) was built for herbal medicines, based on systems pharmacology. It contains all the 499 Chinese herbs registered in the Chinese pharmacopoeia with 29,384 ingredients, 3,311 targets and 837 associated diseases [16]. We performed a research on TCMSP with the keyword of "Corydalis Rhizoma" and the screening conditions were drug similarity (DL) greater than 0.18 and oral bioavailability (OB) greater than 30% [17,18]. The active components of CR-relative targets database was established according to the search result which contained active components of CR and targets corresponding to each component, of which target names were converted into Universal Protein Resource (Uniprot, http://www.uniprot.org/, last updated on November 29, 2019) symbols for subsequent analysis.
2.2. Establishment of disease targets database of CHD. To get a comprehensive collection, the search of disease targets was performed , with the key word of "coronary heart disease", on GeneCards database(https://www.genecards.org/, Version 4.13) which enables researchers to effectively navigate and inter-relate a large number of human genes, diseases, variants, proteins, cells, and biological pathways [19] as well as Online Mendelian Inheritance in Man(OMIM, https://omim.org/, last updated on February 5, 2020) which is the primary repository of comprehensive, curated information on genes and genetic phenotypes and the relationships between them [20]. Database of relative gene targets was built with the obtained targets, converted into Uniprot symbols, from both GeneCards and OMIM database.
2.3. Construction of active components of CR-disease targets of CHD network. Intersection between two gene sets of active components of CR-relative targets database and disease targets database of CHD was employed to draw a Venn diagram [21] showing the distribution of these genes. Cytoscape (version 3.2.1) [22] software was used to construct the active components of CR-disease targets of CHD network, with components and targets are represented as "nodes" while associations between them are encoded by "edges".
2.4. Construction of protein-protein interaction(PPI) network. The String database aims to collect and integrate this information, by consolidating known and predicted protein-protein association data for a large number of organisms [23]. Using the "CR-CHD" common target genes, PPI network [24] was constructed by String database ( https://string-db.org/, Version 11.0), with all the targets were only limited the species as "Homo sapiens", setting combined score 0.4, from which disconnected nodes were hidden. We calculated the number of adjacent nodes of each node in the PPI network, and drew a histogram.
2.5. GO Enrichment and KEGG Pathway Analysis. In order to understand the biological function of "CR-CHD" common genes, we performed the Gene Ontology (GO, http://geneontology.org/, last updated on January 1, 2020) [25] enrichment and Kyoto Encyclopedia of Genes and Genomes(KEGG, https://www.kegg.jp/, last updated on February 18, 2020) [26] pathway analysis using clusterPro ler(version 3.8) and DOSE(version 3.6) software package on R platform. As increasing quantitative data generated from transcriptomics and proteomics requiring integrative strategies for analysis, clusterPro ler automates the process of biologicalterm classi cation and the enrichment analysis of gene clusters [27] while DOSE provides semantic similarity computations among disease ontology(DO) terms, which is important annotation in translating molecular ndings from high-throughput data to clinical relevance, and genes which allows biologists to explore the similarities of diseases and of gene functions in disease perspective [28]. Threshold values were set as P value=0.05 and Q value=0.05. A target gene-molecular function and pathway network was built by using Cytoscape (version 3.2.1) software for a more intuitive visualization, with key modules screened from the network by MCODE plugin of Cytoscape (version 3.2.1) [29,30], setting degree cutoff=2 node score cutoff=0.2 K-score=2 max depth=100.

Results
3.1. Active components of CR-relative targets database. A total of 77 reported ingredients of CR were retrieved by searching the TCMSP database, and 49 active ingredients were screened out using OB and DL from ADME parameters. And 394 related targets of the active components were searched from TCMSP. Details of these components are provided in Table 1. 3.2. Disease targets database of CHD. Removing the same genes, a total of 7173 target genes of CHD were screened from GeneCards database and OMIM database. All of these genes were used for construction of disease targets database of CHD.
3.3. Active components of CR-disease targets of CHD network. As shown in Figure 2, we found out that there are 40 "CR-CHD" common target genes, namely target genes with possible roles in treatment of CHD. The common genes correspond to 48 active components of CR. The common genes and the components corresponded to them were converted into a visualized active components of CR-disease targets of CHD network, shown in Figure 3. The distribution of correspondence between ingredients and target genes can be observed intuitively through the network. According to the analysis of the network, distribution of node degree, namely the number of connections or edges the node has to other nodes, of the network is shown in Figure 4.
There are 3 components with node degree equal or more than 5, including quercetin, C09367 and stigmasterol, while 3 genes with node degree equal or more than 3, which are PTGS1, CHRM3, AR, of which details are listed in Table 2 and Table 3.  3.4. Protein-protein interaction network. As can be seen from Figure 5, the protein-protein interaction of "CR-CHD" common target genes are displayed in the network clearly. Each linkage of target genes represents a meaning, such as protein homology, gene co-expression, inter-genetic adjacency. The more adjacent nodes one node has, the more prominent it is. We can view the key genes with equal or more than 8 adjacent nodes in Figure 6.
3.5. GO enrichment and KEGG pathway analysis of "CR-CHD" common target genes. A total of 70 biological processes and 67 signaling pathways were obtained through GO enrichment analysis and KEGG pathway analysis with the P value 0.05. The rst 20 biological processes were visualized in Figure 7, ranking based on the P values in the order from small to large. Top 5 biological processes which are DNA-binding transcription activator activity, RNA polymerase II-speci c with the highest enrichment score, and the next most critical processes, which were RNA polymerase II transcription factor binding, kinase regulator activity, ubiquitin-like protein ligase binding can be used for the subsequent analysis. A total of 12 related signaling pathways, after removing of the extensive pathway of KEGG pathway enrichment by literature search on PubMed and Embase, are shown in Table 4. Fluid shear stress and atherosclerosis signaling pathway with the most abundant gene enrichment can be the signi cant pathway for further analysis, secondly TNF signaling pathway, apoptosis, MAPK signaling pathway and PI3K-Akt signaling pathway, of which details are visualized in Figure 8. The target gene-molecular function and pathway network was constructed by Cytoscape 3.6.0 software. The network can be seen from Figure 9 that biological processes are represented by small red dots arrayed at the upper part within the circle, while pathways arrayed at the lower part, and dots in the circle indicate genes. Dot size represents a node degree. The bigger the dot is, the more signi cant the gene is. And the genes with node degree equal or more than 10 in target gene-molecular function and pathway network were listed in Table 5. In addition, the network was analyzed by MCODE plugin, and 3 signi cant modules were selected as shown in Figure 10.  3.6. Active ingredients of CR relative to the hub genes. Based on the above analysis, we listed the hub genes, including AR, FOS, CASP3, IL6, MYC, PPAG, NOS3, CASP8, GSTP1, CYP1A1, AHR, HIF1A, ERBB2, IGF2, HSPB1, CRP, CHRM3, DUOX2, PTGS1, which played important roles in each analysis, and active ingredients corresponding to these genes in Table 6. These hub genes and ingredients are the main focus in this study.

Discussion
In this study, we attempted to explore the mechanism of action by which CR treating CHD, with the method of network pharmacology.
In the active components of CR-disease targets of CHD network, quercetin, C09367 and stigmasterol are key ingredients as they are closely related to the most target genes compared to other ingredients of CR. Also, quercetin relates to the majority of hub genes as we concluded in Table 6, while C09367 relates to the rest of hub genes. Quercetin presents signi cant functions as inhibition of LDL oxidation, reduction of in ammatory markers, endothelium-independent vasodilator effects, protection on endothelial function and nitric oxide, and antiplatelet effect, proved by vitro and some animal models, showing the potential to treat cardiovascular diseases [31,32,33]. Also, quercetin showed the analgesic property in various models of in ammation [34]. It is worth mentioning that stigmasterol is related to almost the same genes as C09367, excepting the AR gene. Mouse experiments demonstrated that stigmasterol prevented the HFWD-induced elevation of some di-and triacylglycerol species, decreased serum levels of ceramides, inhibited intestinal absorption of cholesterol and plant sterol, and suppressed hepatic cholesterol [35,36]. Therefore, these key ingredients can achieve a good effect on CHD by lowering cholesterol, inhibiting platelet aggregation, suppressing in ammatory response and relieving the pain.
After analysis of the active components of CR-disease targets of CHD network, we found out that, PTGS1, CHRM3 and AR are the top three most signi cant genes, for a total of 40 "CR-CHD" common target genes, with the most relative ingredients of CR. Prostaglandin G/H synthase 1 protein corresponding to PTGS1 gene, is involved in the generation of thromboxane A2 (TXA2), which promotes platelet activation and aggregation, vasoconstriction and proliferation of vascular smooth muscle cells. Therefore, we expect a modulating effect of CR on the expression of PTGS1 encoding the Prostaglandin G/H synthase 1, proving in future experiments. CHRM3 encodes Muscarinic acetylcholine receptor M3 protein proved to cause an endothelium-independent vasodilatation in a mouse experiment [37]. The muscarinic acetylcholine receptor mediates various cellular responses, including inhibition of adenylate cyclase, breakdown of phosphoinositide and modulation of potassium channels through the action of G proteins [38]. As to AR gene, which encodes androgen receptor protein, is veri ed in some investigations directly that it has atheroprotective effects through both ARdependent and AR-independent mechanisms, causing different incidence of cardiovascular disease between men and women [39,40]. When analyzing the PPI network, genes with the node degree equal or more than 8 were selected as key genes, including CASP3, IL6, MYC, ERBB2, AR, FOS. A member of caspase family, caspase-3, encoded by CASP3 gene, in neuronal cells, has been identi ed as a key mediator of apoptosis [41]. Encoded by IL6, interleukin-6 is increased in a number of cardiovascular diseases, and also associates with a higher incidence of future cardiovascular events, which effects on activity and expression of endothelial nitric oxide synthase and increases vascular superoxide, thus inactivating NO thereby and limiting NO bioavailability [42]. A Swedish cohort indicated that interleukin-6 trans-signaling driven by the IL6 and soluble IL6 receptor binary complex, could be a promising marker of cardiovascular events risk and possibly be used for anti-in ammatory therapy [43]. Expression of ERBB2 relates to mitochondrial function in cardiomyocytes, according to a mouse experiment [44], encoding receptor tyrosine-protein kinase which has a protective effect on cardiomyocytes [45] as well as regulates outgrowth and stabilization of peripheral microtubules. FOS gene could be one of the indexes re ecting myocardial ischemia [46]. By analyzing the target gene-molecular function and pathway network, there were 11 genes, including CASP3, PPARG, NOS3, CASP8, FOS, GSTP1, ERBB3, CYP1A1, AR, AHR, HIF1A, found out to act important roles in the network, as each of them associating with equal or more than 10 molecular functions and pathways. PPARG might increase the risk of CHD in Asian population, as suggested in a meta-analysis [47], as well as PPARGC161T CT/TT was associated with lower levels of blood TC and LDL-C in Han population [48]. NOS3 is signi cantly altered in patients with CHD [49], of which expression reduces in patients with atherosclerosis [50]. Caspase-8, encoded by CASP8 gene controls apoptosis, necroptosis and pyroptosis as a switch [51]. CYP1A1 showed an increased expression in females compared to males under the situation of ischemic heart disease [52]. AHR regulates the expression of members in CYP1 family, including CYP1A1 and CYP1A2. It was demonstrated that the AHR system could induce the reporter gene expression by acute hypoxia, of which induction was transient, in an ischemic hind limb model [53]. Hypoxia inducible factors, including HIF1A, are key oxygen sensors that mediate the ability of the cell to deal with hypoxia [54]. Moreover, 3 modules derived from MCODE analysis, which are module 1 consisting of ERBB2, IGF2, HSPB1, module 2 consisting of CHRM3, CRP, and module 3 consisting of DUOX2, PTGS1. IGF2 may be relevant to the regeneration of the mammalian heart after injury [55], of which expression is induced to increase by hypoxia in rat hearts [56]. An experiment suggested that HSPB1 acts a role that reduced in ammation and healed wound after myocardial infarction(MI), expected to be a target for myocardial repair in MI patients [57]. CRP, which is one of the in ammatory biomarkers, is considered to be an indicator for evaluating severity and prognosis of CHD, as lipoproteins in ammation is considered signi cant in the pathogenesis of CHD [58]. The key genes mentioned above reveal mechanisms underlying the therapeutic effect of CR in the treatment of CHD, participating, to varying degrees, in the process of platelet aggregation, vasoconstriction and vasodilatation, proliferation and repair of vascular endothelial cells and cardiomyocytes, lipid metabolism and in in ammation in the heart. CR may play a full part in the treatment by acting on these key genes. Furthermore, taking these results together, it was concluded that AR, FOS, CASP3 were the 3 most critical genes that played roles in the underlying mechanisms for CR treatment on CHD.
We scanned out the biological process of DNA-binding transcription activator activity, RNA polymerase IIspeci c with the highest enrichment score, and the next most critical processes, which were RNA polymerase II transcription factor binding, kinase regulator activity, ubiquitin-like protein ligase binding after GO enrichment analysis. GO enrichment analysis showed that CR may achieve effects on CHD through the bidirectional regulation of DNA-binding transcription activator activity, RNA polymerase II-speci c, RNA polymerase II transcription factor binding, kinase regulator activity, ubiquitin-like protein ligase binding, though regulation mechanisms still waiting for strong scienti c evidence. After KEGG enrichment analysis and removal of wide range of metabolic pathways, we concluded 5 top pathways with the highest enrichment scores, including uid shear stress and atherosclerosis, TNF signaling pathway, apoptosis, MAPK signaling pathway, PI3K-Akt signaling pathway. Sensing of uid shear stress and atherosclerosis is considered important in processes of vascular development and remodeling [59,60]. Studies demonstrated that TNF antagonists has a potent effect of anti-in ammatory and antioxidant [61,62]. MAPK signaling pathway is activated to regulate apoptosis of cardiac myocytes and angiogenic response of microvascular endothelial cells by related factors, for example, nicotine and protein phosphatase 2A, demonstrating that activation of the MAPK signaling pathway appeared to be a key process in microvascular endothelial cells and cardiac myocytes life-death decisions [63]. The PI3K/Akt signal pathway regulates survival, apoptosis, cell morphology, protein synthesis, and integration of metabolism in cardiomyocytes [64], involved in regulating in ammatory responses, playing a critical role in cardioprotection of preconditioning against ischemia injury [65]. These researches indicate that CR acts on CHD at multiple levels through multiple biological processes and mainstream signaling pathways, mainly regulating vascular development and remodeling, in ammatory process, oxidation, endothelial cell and cardiacmyocyte apoptosis and protein synthesis.
Furthermore, we can understand the advantages of CR in the treatment of CHD and its potential in new drug development by this study.

Conclusions
Traditional Chinese medicine itself with diversi ed structures have multi-targets and multi-pathways effect, thus increasing the di culty to identify mechanisms of active ingredients in CR acting on CHD. Network pharmacology offers a promising new way to solve this problem, basing on the network of "drug-targetdisease-pathway" interactions and bioinformatics analysis. Combined with results of above analyses, it suggests that quercetin plays a dominant role in therapy of CR on CHD, as it relates to the most target genes of CHD. And also, some of these genes corresponding to quercetin are demonstrated to take a critical part in the subsequent PPI network, GO and KEGG enrichment analyses. Among them are FOS and CASP3, which involve in uid shear stress and atherosclerosis, TNF signaling pathway. The next signi cant ingredients are C09367 and stigmasterol, for their common target genes involve in the control of the same biological processes, for example, DUOX2 and PTGS1 participant in heme binding and tetrapyrrole binding, CHRM3 and CRP involve in neurotransmitter binding and ammonium ion binding. Moreover, the AR gene, related to C09367, is the key player in PPI network, GO and KEGG enrichment. In this study, we analyzed the active components of CR as well as the possible mechanisms of these components on CHD, forming the basis of a future research projects. The data used to support the ndings of this study are included within the article and supplementary materials.

Competing Interests
The authors declare that they have no con icts of interest.

Funding and Acknowledgments
This study was supported by the SanMing Project of Shenzhen City of China (grant number: SZSM201612033).

Authors' Contributions
Ying Li and Zhenkun Zhuang searched articles in electronic databases and wrote the manuscript. Mingtai Chen, Haidan Lin and Changjian Yuan analyzed the data. Meihuan Li and Yanhui Wu performed the data extraction. Zhong Zhang designed the study and amended the paper.

Supplementary Materials
Supplementary