Network Pharmacology-Based Systematic Analysis of Molecular Mechanisms of Dingji Fumai Decoction for Ventricular Arrhythmia

Background Dingji Fumai Decoction (DFD), a traditional herbal mixture, has been widely used to ventricular arrhythmia (VA) in clinical practice in China. However, research on the bioactive components and underlying mechanisms of DFD in VA is still scarce. Methods Components of DFD were collected from TCMSP, ETCM, and literature. The chemical structures of each component were obtained from PubChem. Next, SwissADME and SwissTargetPrediction were applied for compounds screening and targets prediction of DFD; meanwhile, targets of VA were collected from DrugBank and Online Mendelian Inheritance in Man (OMIM). Then, the H-C-T-D network and the protein-protein interaction (PPI) network were constructed based on the data obtained above. CytoNCA was utilized to filter hub genes and VarElect was used to analyze the relationship between genes and diseases. At last, Metascape was employed for systematic analysis on the potential targets of herbals against VA, and AutoDock was applied for molecular docking to verify the results. Results A total of 434 components were collected, 168 of which were qualified, and there were 28 shared targets between DFD and VA. Three function modules of DFD were found from the PPI network. Further systematic analysis of shared genes and function modules explained the potential mechanism of DFD in the treatment of VA; molecular docking has verified the interactions. Conclusions DFD could be employed for VA through mechanisms, including complex interactions between related components and targets, as predicted by network pharmacology and molecular docking. This work confirmed that DFD could apply to the treatment of VA and promoted the explanation of DFD for VA in the molecular mechanisms.


Introduction
In recent years, cardiovascular diseases are the leading cause of death in China [1]. All cardiac conditions, especially ischemic heart disease, can lead to arrhythmias [2]. Among all arrhythmias, ventricular arrhythmia (VA) has the highest mortality. VA is a common but life-threatening disease, mainly including ventricular premature contraction, ventricular tachycardia, ventricular flutter, and ventricular fibrillation, with a clinical presentation that ranges from no symptoms to cardiac arrest [2]. VA is usually generated by all-caused enhanced automaticity or abnormal automaticity, myocardial ischemia, delayed afterdepolarizations, and structural heart disease with reentry [3][4][5]. To prevent adverse events of VA, millions of patients were treated with beta-receptor-blockers, INa antagonists, IKr antagonists, nondihydropyridine calcium antagonists, and other drugs suggested by the 2017 American Heart Association/American College of Cardiology/Heart Rhythm Society guideline [2], but the control of VA is still far from ideal.
Traditional Chinese medicine (TCM) has a clinicalbased development history of over 2000 years [6,7]. Some researchers mentioned that TCM could be used to treat a variety of diseases, including VA [8][9][10]. Besides, related studies suggested that, compared with Western medicine only, patients suffering from various diseases can benefit

Chemical Structures Construction.
e Traditional Chinese Medicine Systems Pharmacology database and analysis platform (TCMSP) [27] and the Encyclopedia of Traditional Chinese Medicine (ETCM) [28] are web-based herb databases, providing comprehensive and standardized information for the commonly used herbs. In this study, the components of each herb in DFD were obtained from TCMSP, ETCM, and published literature. To make the components recognizable for the subsequent analysis work, after removing duplicates, the structure of each component was collected from PubChem [29].

Gastrointestinal Absorption (GA) and Drug-Likeness (DL)
Prediction. Increasingly researchers found that TCM despite their impressive in vitro findings demonstrate less or negligible in vivo activity, resulting in poor absorption and hence poor bioavailability [30]. e absorption, distribution, metabolism, and excretion (ADME) of the drug must be considered by the researcher and developer [31]. Bioabsorption is highly multifactorial but is primarily driven by GA [32]. Besides, DL assesses qualitatively the chance for a molecule to be an oral drug with respect to bioavailability [31]. It was constructed that the estimation of ADME before the drug development studies reduces the possibility of failure [33]. In the mechanism explaining DFD, GA and DL were evaluated using SwissADME, a free tool that could evaluate DL, GA, pharmacokinetics, and medicinal chemistry friendliness of small molecules [31]. After uploading the structure of each compound to SwissADME, if the prediction results of the component suggested high possibility of both GA and DL, it met our inclusion criteria and was adopted for the next screening [31].

Target Prediction and Verification.
In the treatment of diseases, not all absorbable components work; therefore, we filtered out the components with bioactive components from all absorbable components using SwissTargetPrediction, an online tool that can evaluate compounds with a score by fitting a multiple logistic regression on various subsets of known actives to weight structure similarity parameters [34]. Here, we uploaded the structure of each component to SwissTargetPrediction to predict potential targets of DFD, and all possible targets were adopted.
Online Mendelian Inheritance in Man (OMIM) is a knowledge base providing the latest information of human genes [35], and DrugBank is a freely available and comprehensive web resource providing drug-target and drug interaction information [36]. Taking "ventricular arrhythmia" or "arrhythmia of ventricular origin" as a keyword, we obtained VA-related targets from OMIM and DrugBank. Taking the intersection of DFD and VA targets, the common targets between DFD and VA were considered the therapeutic targets of DFD against VA, as described previously [37].
PPI is one of the cores of cellular processing. e analysis of PPI makes the interactions of proteins clear and helps to explain the function of possible protein complexes or functional modules [38]. STRING is a web database providing online analysis of PPI [38]. After uploading the common targets to STRING, we constructed the PPI network. en, the result was imported to Cytoscape (version 3.8.0) for further analysis [39]. CytoNCA plugin in Cytoscape was applied to analyze centrality of certain targets and evaluate protein interaction networks [40].
e VarElect online tool can analyze direct and indirect links between genes and diseases [41]. In this study, the link relationships of potential targets of DFD against VA were analyzed with VarElect; the results helped determine which targets will be included in the next molecular docking.

Biology Functional Analysis.
Since Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) can contribute to the interpretation of system-level data and enable new discoveries [42], in this study, Metascape was employed for GO and KEGG analysis to further explore the complex mechanism of DFD in the treatment of VA. Metascape is a web-based platform providing gene annotation, functional enrichment, and interactome analysis services; monthly database update could keep our analysis results up to date [42]. In our work, GO and KEGG terms with P < 0.01 were considered significantly enrichment analyses.

Molecular
Docking. Molecular docking was used to assess interactions between components and hub targets; the 4 hub targets connected to VA closely were included. e structures of these targets were collected from Protein Data Bank [43]. AutoDock and PyMOL were employed for molecular docking, PyMOL was used to remove the water molecules and isolate proteins of the moleculars [44]. AutoDock was used to add the hydrogen and calculate Gasteiger charges of the moleculars and add the hydrogen for the ligands [45]. At last, molecular docking was conducted using AutoDock to assess the binding energy; the lower the binding energy, the more stable the docking modules [45].  Evidence-Based Complementary and Alternative Medicine prediction results was placed in the Supplementary Materials. All qualified components were adopted for the next screening.

Target Prediction and Verification.
After removal of duplicates, a total of 1096 potential targets of DFD were collected from SwissTargetPrediction. Meanwhile, a total of 260 known therapeutic targets for VA were obtained from OMIM and DrugBank. Taking the intersection of DFD and VA targets, there were 28 shared targets; based on the data obtained above, the Herb-Compound-Targets-Disease (H-C-T-D) network was constructed. e H-C-T-D network was composed of 147 nodes (DFD, VA, 7 herbals, 110 bioactive compounds, and 28 common targets) and 465 edges ( Figure 2). In further analysis, all 28 common targets were uploaded to STRING to construct the PPI network; the results were imported to Cytoscape to calculate the degree value of each gene using CytoNCA plugin and reconstruct the PPI network according to the degree value ( Figure 3). ere were three possible biofunctional modules divided from the PPI network ( Figure 4). After screening by CytoNCA, the top 10 targets were defined as hub targets; the designations and topological parameters of hub targets are shown in Table 1.
We analyzed the 28 common targets using VarElect to investigate the correlation between targets and VA, and the results suggest that 23 targets were related to VA directly, whereas 5 targets were related to VA indirectly ( Table 2); among these targets, Potassium Voltage-Gated Channel Subfamily H Member 2 (KCNH2), Sodium Voltage-Gated Channel Alpha Subunit 5 (SCN5A), Troponin T2-Cardiac Type (TNNT2), and Calmodulin 1 (CALM1) have the highest score of correlation.

Biology Functional Analysis.
e enrichment analysis of GO and KEGG of the 28 common targets was analyzed using Metascape, the results were ranked by −log 10 (P value), and the top 14 of each enrichment item are shown in Figure 5; besides, the functional analysis of the three potential biofunction modules divided from PPI network is shown in Table 3.
In GO and KEGG enrichment analysis, terms with high enrichment scores suggest that the regulation of muscle contraction, regulation of systemic arterial blood pressure by norepinephrine-epinephrine, blood circulation, circulatory system process, adrenergic receptor activity, calcium signaling pathway, adenylate signaling in cardiomyocytes, cGMP-PKG signaling pathway, and neuroactive ligand-receptor interaction could be the most possible mechanisms of DFD in treating VA. For the 3 protein modules, Module 1 can regulate calcium signaling pathway, heart rate, and cAMP signaling pathway. Module 2 can regulate membrane depolarization during an action potential, striated muscle contraction, regulate adrenergic signaling in cardiomyocytes, and cation homeostasis. And Module 3 is more likely to possess the function of regulation of the metabolic process.

Molecular
Docking. Molecular docking was conducted to calculate binding energy between components and hub targets, KCNH2 (PDB ID:1BYW), SCN5A (PDB ID: 4DCK), TNNT2 (PDB ID: 1J1D), and CALM1 (PDB ID: 1CDL); the 4 genes connected closely to VA were included in molecular docking. e docking information of components and hub targets was listed in Table 4, and the results suggested that components in DFD could interact with the hub targets against VA, following the principle that the lower the binding energy, the more stable the docking modules [45]. e lowest 10 binding energy docking modules are shown in Figure 6, the rest docking module could be found in the supplement, and all 10 components interacted with corresponding targets through hydrogen bond mainly.
e results show that Jujubogenin has the highest binding energy connected with HIS70 and ASP67 of KCNH2 ( Figure 6

Discussion
VA is a fatal disease, typical drugs may benefit patients, but its side effects such as respiratory diseases, liver and kidney damage, and bradyarrhythmia can never be ignored. Fortunately, long-time clinical work was told that DFD is an effective herb mixture against antiarrhythmia. Since its excellent clinical efficacy, we conducted a Real-World Trial that included more than 160 patients who suffered premature ventricular contractions to assess the safety and efficacy of DFD for VA and the results demonstrate that DFD combined with metoprolol has better efficacy and safety than placebo combined with metoprolol [24]. Besides, we explored the cellular electrophysiological mechanism of DFD with Chinese hamster ovary cells using whole-cell patch-clamp, and the result suggests that DFD indeed has antiarrhythmic effects based on its antioxidant potential, alleviation of Na + -K + -ATPase and connexin-43, and class I antiarrhythmic properties by suppressing Nav 1.5 dose-dependently with an IC 50 of 24.0 ± 2.4 mg/mL [25]. In this study, the bioactive components and underlying molecular mechanisms of DFD in the treatment of VA were analyzed systematically. rough related information collection and primary screening, we identified 28 potential targets of DFD in the treatment of VA. A PPI network was constructed with STRING and Cytoscape 3.8.0, the top 10°value genes were selected as hub genes, and 3 function modules were divided based on their interactions. All potential genes were analyzed using VarElect, all 10 hub genes are suggested directly related to the treatment of VA, and among these genes, 4 Evidence-Based Complementary and Alternative Medicine KCNH2, TNNT2, and CALM1 as well as SCN5A have the highest scores of correlations; in other words, these 4 genes are the most promising targets for DFD against VA. Recently, KCNH2 could be a hot gene in the study of VA; it could mediate the rapidly activating component of the delayed rectifying potassium current in the heart. Previous research suggested that pathogenic variants in KCNH2 encoding may result in long QT syndrome [46]. Meanwhile, another research based on quantitative analysis of consortium disease cohorts and population controls pointed out that, among patients with long QT syndrome, the mutation probability of the KCNH2 gene is greater than 85% [47]. Besides, another research mentioned the coexpression of CACNA1C and KCNH2 reduces the arrhythmic events [48]. TNNT2 is another hub gene connected to arrhythmias; a genetic analysis suggested that TNNT2 was cosegregated in VAs and sudden death [49]. A study conducted using zebrafish embryos suggested that zebrafish embryos exposed to procymidone are more likely to alter transcription levels of TNNT2 and resulted in arrhythmia as well as increased heart rate finally [50]. Raffaele Coppini conducted a cohort study of patients with hypertrophic cardiomyopathy (HCM), the outcome indicated that, among patients with HCM, most patients have a mutation in TNNT2, and these patients are more likely to suffer from arrhythmias and HCM in the future [51]. SCN5A is pivotal to cardiac electrical conduction and arrhythmic risk; a study provided a new effective therapy to reduce arrhythmia through downregulating the expression of SCN5A [52]. Similarly, there is a study that reported that a combination of quinidine/mexiletine reduces arrhythmia in patients with SCN5A gene mutation [53]. CALM1 is a regulator of voltage-dependent L-type calcium channels; its mutations are related to congenital arrhythmia [54]. Heterozygosity for the CALM1 mutation is causative of an arrhythmia syndrome [55]. Moreover, it can lead to catecholaminergic polymorphic ventricular tachycardia, idiopathic ventricular fibrillation, long QT syndrome, and even sudden death [56].
In the further, the results of GO and KEGG analysis elucidated that the regulation of systemic arterial blood pressure by norepinephrine-epinephrine, muscle contraction, blood circulation, circulatory system process, adrenergic receptor activity, calcium signaling pathway, adenylate signaling in cardiomyocytes, cGMP-PKG signaling pathway, Evidence-Based Complementary and Alternative Medicine and neuroactive ligand-receptor interaction were highly enriched, which revealed the potential mechanisms of DFD in treating VA. Here, the regulation of the calcium signaling pathway is enriched by most of the hub genes. According to a study based on the genomic, transcriptomic, and proteomic data initiated by Dan E Arking, calcium signaling pathway plays an important role in both the depolarization and repolarization of myocardial ischemia; particularly in the repolarization, during the plateau phase of the cardiac action potential, prolonged inward Ca2+ currently leads to delays in ventricular myocyte repolarization [57]. Earlier research also mentioned that Ca 2+ waves can form when the Ca 2+ ion influx into the cell is increased, and the Ca 2+ waves can generate depolarization that triggers arrhythmias [58], it is reasonable to speculate that the regulation of calcium signaling pathways of DFD may be one of the effective methods for VA. Reports also suggested that the adrenergic signaling can increase the transmural difference between Ca 2+ ion transients duration and action potential duration, finally, promoting the formation of delayed afterdepolarizations, the regulation of adenylate cyclase-activating adrenergic receptor signaling pathway and adrenergic receptor signaling pathway of DFD for VA may antiarrhythmia in this way [59]. Adenylate cyclase-modulating G protein-coupled receptor signaling pathway can result in the regulation of G protein-mediated signaling, which is of great importance for the regulation of heart rate and involved in arrhythmias [60]. Besides, as we mentioned above, the potential targets were divided into 3 function modules, as shown in Table 3. e enrichment analysis results indicate Module 1 can regulate calcium signaling pathway, heart rate, and cAMP signaling pathway, which suggest that Module 1 has great potential in the antiarrhythmia. Similarly, the results of enrichment analysis indicate that Module 2 were involved in the regulation of membrane depolarization during an action potential, striated muscle contraction, regulate adrenergic signaling in cardiomyocytes and cation homeostasis, pathways, which has illustrated the antiarrhythmic potential of Module 2. e enrichment analysis of Module 3 may not seem to be ideal, but we found that the 4 genes connected to VA closest are gathered in function Module 3, and it is reasonable to believe that Module 3 has antiarrhythmic effects. Furthermore, as is shown in Figure 5, the multiregulation in different aspects may benefit patients suffering from related diseases such as hypertension, cancer, and other diseases.
Nevertheless, the present study has some limitations. First, although we tried to find out all components, Osdraconis (Fossiliaossiamastodi) and Ostrea Gigas unberg, has only several pieces of research with several components, and were excluded for its poor possibility of GA and DL, but Osdraconis (Fossiliaossiamastodi) and Ostrea Gigas unberg played important roles in DFD; according to the theoretical system of TCM, both Osdraconis (Fossiliaossiamastodi) and Ostrea Gigas unberg can tranquilize the mind; further studies are needed to confirm the sedative mechanism of Osdraconis (Fossiliaossiamastodi) and Ostrea Gigas unberg. Moreover, DFD, as a traditional Chinese decoction, treats VA through multicomponent and multitarget, indicating that the underlying mechanisms are complex as described in our present study. Although we have made some identification on the mechanisms [25],   Figure 4: ree possible biofunctional modules divided from the PPI network. Notes: the score is an indication of the strength of the connection between the gene and the disease.
Evidence-Based Complementary and Alternative Medicine 7 there are still a lot to explore. Here, this study is designed to understand the mechanism profoundly through network pharmacology as well as molecular docking technology. Same as other network pharmacology analyses [37], taking the intersection of targets about DFD and VA in this study not only follows the operating processes of network pharmacology but also makes the result of virtual screening more reliable [61]. Undoubtedly, some targets that involve either DFD or VA might be ignored and missed, which is a common and inevitable issue in network pharmacology. And for this reason, in the collection of components and targets, despite DFD and VA, we tried our best to reduce bias by searching as much more databases as we can. Finally, so many components are boiled together, and more researches

Conclusion
As mentioned above, DFD could be employed for VA through mechanisms, including complex interactions between related components and targets, as predicted by network pharmacology and molecular docking. is work confirmed that DFD could apply to the treatment of VA and promotes the explain of DFD for VA in the molecular mechanisms; similar results can be obtained from previous experiments of cellular electrophysiological mechanisms. e systematic analysis in this work can provide a comprehensive consideration for further studies.

Data Availability
All the data generated or analyzed during this study are included in this published article and its supplementary information files.

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

Authors' Contributions
Yi Liang and Bo Liang contributed equally to this work. Yi Liang, Bo Liang, and Li-Zhi Zhao conceived, designed, and planned the study. Yi Liang and Xin-Rui Wu acquired and analyzed the data. Bo Liang, Wen Chen, and Li-Zhi Zhao interpreted the results. Yi Liang and Bo Liang drafted the manuscript and Bo Liang and Li-Zhi Zhao contributed to the critical revision of the manuscript. All authors read and approved the final manuscript. during the preparation of this manuscript. is work was funded by the General Project of Sichuan Provincial Administration of Traditional Chinese Medicine and Research and Practice Innovation Plan for Postgraduates of Jiangsu, China.

Supplementary Materials
Table S1: comparison of Chinese medicine names and Latin names. Table S2: details of qualified compounds in various herbs. Figure S1: molecular docking modules. (Supplementary Materials)