Mechanisms of Quercetin against atrial fibrillation explored by network pharmacology combined with molecular docking and experimental validation

Atrial fibrillation (AF) is a common atrial arrhythmia for which there is no specific therapeutic drug. Quercetin (Que) has been used to treat cardiovascular diseases such as arrhythmias. In this study, we explored the mechanism of action of Que in AF using network pharmacology and molecular docking. The chemical structure of Que was obtained from Pubchem. TCMSP, Swiss Target Prediction, Drugbank, STITCH, Pharmmapper, CTD, GeneCards, DISGENET and TTD were used to obtain drug component targets and AF-related genes, and extract AF and normal tissue by GEO database differentially expressed genes by GEO database. The top targets were IL6, VEGFA, JUN, MMP9 and EGFR, and Que for AF treatment might involve the role of AGE-RAGE signaling pathway in diabetic complications, MAPK signaling pathway and IL-17 signaling pathway. Molecular docking showed that Que binds strongly to key targets and is differentially expressed in AF. In vivo results showed that Que significantly reduced the duration of AF fibrillation and improved atrial remodeling, reduced p-MAPK protein expression, and inhibited the progression of AF. Combining network pharmacology and molecular docking approaches with in vivo studies advance our understanding of the intensive mechanisms of Quercetin, and provide the targeted basis for clinical Atrial fibrillation treatment.


Results
Information on potential targets of Que and AF. The chemical structure of quercetin is shown in Fig. 2a. Que potential targets were obtained by TCMSP, PharmMapper, Swiss Target Prediction, Drugbank, Stitch and Binding database databases, and a total of 449 were obtained by screening de-duplication and integration. A total of 2201 disease targets were screened using CTD, GeneCards, Disgenet and TTD databases, and 3096 differential genes between normal and AF patients were obtained using the GEO database, of which 462 were upregulated and 2634 were downregulated. From the volcano plot (Fig. 2b) and the difference plot (Fig. 2c), it can be seen that there are significant genetic differences between normal human atrial tissues and atrial tissues from AF patients. Combining drug targets, disease targets and GEO targets, 65 core targets, including IL-6, EGFR, MMP9, ESR1 (etc.) were obtained by online mapping of Venn diagram (Fig. 2d). String database, species limited to human, and the protein interactions network was obtained, which was further imported into Cytoscape for topological analysis, and the results suggested that there were 65 nodes and 575 edges in the network, and the key core targets were obtained using cytoHubba plugin (Fig. 2e), ranking the top 10 key targets for treating AF targets IL-6, VEGFA, JUN, MMP9, EGFR, CCND1, CXCL8, PTGS2, ESR1 and MYC, and the detailed information of the top ten targets ranked by the key target Degree value is shown in Table 1.

GO function analysis and classification.
To further clarify the mechanism of action of quercetin in the treatment of AF, we classified 65 core targets into proteins using the PANTHER database (Fig. 3a), found that they were mainly focused on protein modifying enzymes (20.4%), gene-specific transcriptional regulators (18.40%), and metabolite interconverting enzymes (18.40%). 65 core genes were introduced into the R package, and the GO functional results (Fig. 3b), the biological process (BP) was mainly focused on: response to lipopolysaccharide, response to molecule of bacterial origin, regulation of inflammatory response, response to oxidative stress, and epithelial cell proliferation; cell composition (CC) is mainly focused on: vesicle lumen, membrane raft, membrane microdomain, membrane region, focal adhesion; molecular functions (MF) mainly focus on: DNA-binding transcription factor binding, RNA polymerase II-specific DNA-binding transcription factor binding, cytokine receptor binding, monocarboxylic acid binding, and growth factor binding. In addition, we used the MCODE tool to identify highly interconnected clusters in the PPI network, and the MCODE clusters were classified into three types, including regulation of smooth muscle cell proliferation (Fig. 3c), positive regulation of protein import into nucleus (Fig. 3d), negative regulation of intracellular signal transduction (Fig. 3e).
KEGG pathway analysis and core pathway network construction. The pathways of action were clarified by KEGG enrichment analysis of 65 intersecting genes, with 157 enrichment terms, of which the top 30  www.nature.com/scientificreports/ most significant pathways (Fig. 4a), the core targets may affect the lipid and atherosclerosis, AGE-RAGE signaling pathway in diabetic complications, MAPK signaling pathway, and IL-17 signaling pathway, suggesting that Que may be effective in treating AF by modulating the above signaling pathways. The top 10 relevant pathways of Que for AF were screened (Table 2) and imported into Cytoscape to construct the "Que-Major Pathways-AF" network ( Fig. 4b). After screening out the KEGG-enriched pathways that were  www.nature.com/scientificreports/ not related to AF disease, it was found that the important pathway in Que for AF treatment was probably related to MAPK signaling pathway, and the detailed pathway map was obtained by R package analysis (Fig. 5), in which the red area was signal upregulation. Molecular docking and differential analysis of core targets. The five core targets with high Degree values were selected for molecular docking to verify the binding activity of Que active ingredients to potential anti-AF targets. The average docking affinity of the results was − 5.042 kcal/mol, as detailed in Table 3, where the larger the absolute value of the docking affinity, the more stable the binding proved to be, indicating that Que has strong binding to the core targets. The detailed diagram of docking with each key target site is shown in Fig. 6.
Using the difference data of control group and AF group in GEO database, the differential expression of the first five core targets in Que treatment of AF diseases (Fig. 7). VEGFA was significantly upregulated, IL-6, JUN and VEGFA were significantly downregulated compared to controls, while MMP9 targets were not statistically significant, which was considered to be related to the small sample size. The above results suggest that the core targets are closely related to the pathological changes of AF.
Effect of quercetin on atrial fibrillation. In contrast with the visible of p-wave in CON group, in typical AF and Que + AF groups, f-wave was appeared of and of p-wave was disappeared by electrocardiogram meas- www.nature.com/scientificreports/ urement (Fig. 8a). Compared with AF group, the duration of AF was significantly reduced with the increasing concentrations of Que (P < 0.01) (Fig. 8b).
The masson staining results showed that in the atrial myocardial tissue of the CON group, no significant proliferation of blue collagen fibers was happened, and the numbers of collagen fibers in atrial tissues of AF was increased. Compared with the AF group, the numbers of collagen fibers in atrial tissues were decreased significantly with the increasing concentration of Que (P < 0.05) (Fig. 11a).

Effect of Que on protein expression in AF atrial tissue.
Compared with the CON group, p-MAPK protein expression was increased in the AF group (P < 0.01). Compared with the AF group, p-MAPK protein expression was significantly decreased with the increasing concentration Que (P < 0.01). Compared with 75 mg/ kg Que, 150 mg/kg Que in the AF group, p-MAPK protein expression was reduced (P < 0.05). The expression of MAPK protein in 75 mg/kg Que and 150 mg/kg Que groups were no significant difference (P > 0.05) (Fig. 11b).

Discussion
One study found that the increased risk of cardiovascular disease in some populations was associated with low dietary intake of flavonoids, including Que 15 . It has been found that Que played the anti-ischemic and anti-arrhythmic effects in clinical studies [9][10][11] , it leads to great interest and concern of scientists, but the specific mechanisms of Que in the treatment of AF remains unclear. In this study, we investigated the mechanism of Que in the treatment of AF based on network pharmacology analysis. In this study, 449 Que potential targets and 2201 AF targets were obtained from each database respectively, 3096 AF differential genes were obtained from the GEO database. The intersection of the 3 was taken to obtain 65 core genes for Que treatment of AF. The above core genes were screened for the top 5 hub target genes of IL-6, VEGFA, JUN, MMP9 and EGFR by Degree value. GO analysis confirmed that Que had the potential to play an important role in the pathogenesis  www.nature.com/scientificreports/ of AF through multiple biological pathways, and KEGG analysis obtained key pathways of Que for AF, mainly involving lipid and atherosclerosis pathways, the AGE-RAGE signaling pathwayin diabetic complications, MAPK signaling pathway and IL-17 signaling pathway, etc. Through further screening, we found that MAPK pathway plays a key role in the treatment of AF. IL-6 plays a key role in immune, tissue regeneration and metabolism. It was shown that IL-6 in the left atrium was found to promote early atrial fibrosis through the pSTAT3/STAT3 signaling pathway by establishing postoperative AF mice, and similarly higher IL-6 concentrations were found in the pericardial drainage fluid of patients with postoperative AF in a prospective clinical study 16 . Wu reported inhibiting IL-6 inflammatory factor release reduced early atrial fibrosis and its duration in SD rats with postoperative AF 17 . Studies based on clinical followup investigations have found that measurement of IL-6 concentrations could provide an independent prognostic signal as well as a predictor of recurrence for mortality in AF patients 18,19 . Therefore, IL-6 has the potential to be a therapeutic target for AF. The next key gene VEGFA is a key factor in the growth process of blood vessels and endothelial cells. Its main function is to induce endothelial cell proliferation and inhibit apoptosis 20 . It is well known that AF can lead to endothelial dysfunction 21 , and Wang reported that the plasma levels of VEGFA were significantly elevated in AF patients 22 , suggesting that endothelial dysfunction may be one of the etiologies of AF, VEGFA may be a potential therapeutic target for AF.
The next gene is Transcription factor JUN, It has been shown that the transcription factor complex AP-1 plays a major role in cardiac hypertrophy, and knockout of c-jun pure mice develops in a process leading to cardiac malformations 23 , activation of c-jun impaired cell-cell communication between atrial myocytes and ultimately promotes the development of atrial arrhythmias 24,25 . Therefore, JUN has the potential to be a therapeutic target for AF. The next gene MMP9 is a matrix metalloproteinase that plays an important role in local protein hydrolysis, cell migration, apoptosis and signaling in the extracellular matrix. During AF development, overproduction of reactive oxygen species activated MAPK signaling pathway further increasing MMP9 expression 26 . Moe GW reported inhibiting MMP9 expression in canine AF model improved cardiomyocyte hypertrophy and reduced Figure 7. The core action targets of Que on AF were differentially expressed in the GEO data set. Left atrial healthy controls n = 5, AF group n = 5. Values are expressed as mean ± SD. Compared with Control, *P < 0.05 and **P < 0.01. (a) IL-6 differential expression; (b) VEGFA differential expression; (c) JUN differential expression; (d) EGFR differential expression; (e) MMP9 differential expression.  27 . Recent clinical studies have found that the plasma MMP9 level in AF patients were higher than that in controls, which suggested MMP9 could be as a target for the treatment of AF 18,28 . The last of the first five core genes is EGFR, whose main function is to regulate cell growth, proliferation and apoptosis, and is a key regulator of cardiac organogenesis and adult heart 29 . In AF fibrotic atrial tissues, EGFR expression was increased, probably due to the cleavage by pro-fibrotic metalloproteinases during atrial remodeling 30 . The plasma level of EGFR in clinical AF patients was significantly higher than that in the control group 31 , these results suggest that EGFR plays an important role in AF. Overall, the top five core genes were found to be supported by previous studies based on network pharmacology studies. The results of KEGG enrichment analysis suggested that the lipid and atherosclerosis pathway, AGE-RAGE signaling pathway in diabetic complications, MAPK signaling pathway and IL-17 signaling pathway played the major roles in the treatment of AF with Que. One of the main causative factors of atherosclerosis and cardiovascular disease is dyslipidemia, which is considered one of the major risk factors for AF, and several clinical studies had demonstrated that LDL and total cholesterol were negatively correlated with the occurrence of AF [32][33][34] . It is suggested that lipids and atherosclerosis may be involved in the pathogenesis of AF. The AGE-RAGE signaling pathway is an important link in the development of diabetic disease, and this signaling pathway directly mediates diabetic vascular calcification and vasoconstriction-diastolic dysfunction, and a meta-analysis showed that type 2 diabetes increased the risk of AF by 40% 35 . Recent studies had found that the AGE-RAGE signaling pathway was associated with the development of AF 36,37 , and that when AGE bind to RAGE, it produced inflammatory cytokines and cell adhesion molecules that further induced structural and electrical remodeling of the atria. Raposeiras-Roubín found significantly higher AGE and RAGE plasma levels in AF patients than in normal subjects, and several prospective studies have also shown that RAGE predicts future cardiovascular events and death in diabetic patients 38 . This suggesting that this pathway may mediate the progression of AF.
MAPK signaling pathway plays an important role in cell growth, proliferation and fibroblast growth. Li found that atrial structural remodeling in an AF mouse model could be attenuated by decreasing MAPK expression 39 . It has been demonstrated that activation of phosphorylated MAPK significantly increases susceptibility to AF 40 , in detail MAPK pathway (Fig. 5), red upregulated targets get inflammation and apoptosis play an important role in the development of AF disease. Inflammation is one of the important pathogenic mechanisms of AF, and there is an increasing evidence supporting the involvement of inflammation and immune response in the development and progression of AF. Many studies have found that IL-17 signaling pathway stimulates the release of proinflammatory cytokines, which promote myocardial fibrosis and thus induce the development of AF 41,42 . In addition, IL-17 promotes AF development in rat model with aseptic pericarditis by promoting cardiac inflammatory   41,43 . Therefore, the results of the present study suggest that these four signaling pathways may be involved in the mechanism of Que treatment of AF ("Supplementary Information S1"). GO analysis revealed the biological processes mainly focused on responses to lipopolysaccharide, regulation of inflammatory responses, phosphatidylinositol-mediated signaling, responses to oxidative stress and responses to epithelial cell proliferation related to the aforementioned pathway-mediated biological processes, which are also consistent with the aforementioned pathway-mediated biological processes. Molecular docking also suggested strong binding of Que to the above key targets and also validated the therapeutic effect of Que on AF targets at the molecular level.
In order to further validate the results of the above network pharmacology analysis, we performed rat paroxysmal AF model in vivo. Firstly, the paroxysmal AF rat model was successfully constructed by injecting ACh-CaCl 2 mixture solution through tail vein, and we found that AF duration could be obviously reduced by the elevation concentrations of Que. Secondly, HE staining and Masson staining of atrial tissue suggested that with the increasing concentrations of Que, Que could obviously attenuate the morphological and structural damage of atrial tissues induced by paroxysmal AF, as well as reduce atrial fibrosis. Interestingly, there was no difference about the indexes of left ventricular function in each group, however, based on the results of pathological staining results, we considered the current cardiac function was still in a compensatory stage, which was in accordance with previous studies 44 . Finally, combined with the KEGG results, we selected the MAPK pathways that was more www.nature.com/scientificreports/ closely related to the occurrence of AF, and found that with the increasing concentrations of Que intervention, the elevation of atrial p-MAPK protein expression caused by AF of were reduced obviously. Therefore, combining with the network pharmacology, GEO differential expression and in vivo experiments, we hypothesize that Que can medicate some key targets such as IL-6, VEGFA, JUN, MMP9, EGFR, CCND1 and other key signaling pathways such as MAPK pathway, lipid and atherosclerosis, AGE-RAGE pathway in diabetic complications and IL-17 pathway, through which Que can inhibit inflammatory expression, oxidative stress, myocardial fibrosis and alleviate AF-induced atrial remodeling and injury.

Conclusions
In conclusion, this study has identified the core targets and key pathways of quercetin in treatment of atrial fibrillation, which through network pharmacology combined with the GEO database. In paroxysmal atrial fibrillation rat model, quercetin has a protective effect on atrial remodeling, and its mechanism may act through the MAPK signaling pathway. Our study suggests that quercetin can ameliorate the pathophysiological process of atrial fibrillation, which possibly through the synergistic interaction between multiple core targets and pathways, and MAPK is the core target and pathway. These results provide a basis for the clinical application and basic research of quercetin in treatment of atrial fibrillation.
Acquisition of AF related targets. The CTD (http:// ctdba se. org) 50 , GeneCards (https:// www. genec ards. org/) 51 , DISGENET(https:// www. disge net. org/) 52 , Therapeutic Target Database (http:// db. idrbl ab. net/ ttd/) 53 , were used to "Atrial fibrillation" was used as a keyword to search for the relevant genes. The collected targets were combined and eliminated by the median algorithm to obtain disease targets. GSE31821 data were imported into the online analysis tool GEO2R for processing to obtain differentially expressed genes. The gene screening criteria were P < 0.05 and |logFC| ≥ 0.5 to obtain differentially expressed genes between normal tissues and AF, and further volcano maps of differentially expressed genes were drawn 54 .
Cross-tabulation of drugs, diseases and GEO. By using the online mapping tool (http:// www. ehbio. com/ test/ venn/#/), intersections of drug targets, disease targets and differential genes can be obtained.
Construction of protein-protein interaction network (PPI). The intersecting genes were imported into the String database 55 (https:// string-db. org/ cgi/ input. pl) to construct the protein-protein interaction network (PPI), and the species selectors, exported in tsv format. The tsv was imported into Cytoscape 3.7.2 software (http:// www. cytos cape. org/) for adjustment as well as to obtain key targets using the cytoHubba plugin. Simultaneous protein functional classification of the intersecting targets using the Panther database.
GO analysis and KEGG pathway and Que-primary pathway-AF network construction. To further understand the function and role in signaling pathways of the intersecting genes, the gene abbreviations (Symbol) were transformed into gene ID (EntrezID), and the results were analyzed visually by the R language package 6 , as well as the gene ontology (GO) of the above intersecting genes, GO analysis can further explain the role of  56 enrichment analysis was used to investigate the major anti-AF signaling pathways involved in the crossover genes. The core target network of "Que-major pathway-AF" was also constructed. Molecular Docking and differential analysis of core targets. The known key targets were molecularly docked to Que to obtain docking affinities reflecting their stability. The PDB database was used to download the core protein molecular structures and PUBCHEM to download the structures of Que. The protein and Que structures were imported into Auto Dock Tools software for processing and molecular docking, and intermolecular docking affinities were calculated and PyMol 2.4.0 was used for visualization 57,58 . Changes in core targets were analyzed using differential expression between the normal and AF groups from the GEO database. GraphPad Prism software was used for graphical visualization. Values are expressed as mean ± SD.
In vivo experiments. Materials  www.nature.com/scientificreports/ col was approved by the institutional animal care and use Committee of Bengbu Medical College. The rats were randomly divided into 3 groups, control group (CON, n = 6), AF group (AF, n = 6), and quercetin + AF group (Que + AF, n = 12), where the rats in quercetin group 13 were divided into low-dose group (75 mg/kg, n = 6) and high-dose group (150 mg/kg, n = 6), which the rats were intragastric administrated with different concentrations of Que, once/day for 21 days, and the rats in CON and AF were intragastric administrated equal dose of salineion once/day for 21 days. At the end of 21 days, the rats in Que + AF and AF groups were injected with ACh-CaCl 2 mixture (66 ug/ml ACh + 10 mg/ml CaCl 2 ) in tail vein at a dose of 1 ml/kg once a day for 1 week, and equal amount of saline was injected in tail vein in CON group, and atrial tissue was taken after 7 days for next experiment.
ECG data acquisition. Before the experiment, rats were anesthetized using 2% pentobarbital sodium intraperitoneally, limb leads were performed using an electronic ECG machine, ECG was recorded in each group, and non-sinus rhythm was recorded and excluded. When the models of Que + AF and AF had been established, the appearance of f-wave and disappearance of p-wave were the markers for the occurrence of AF, and the disappearance of f-wave and the appearance of p-wave were the markers for the termination of AF 44 , and the time of induction and duration of AF were recorded.
Echocardiography for cardiac function. After ECG recording was completed, echocardiographic testing was performed. Isoflurane anesthesia maintained with small animal ultrasound color Doppler ultrasonography (Vis-ualSonics, Canada), the left ventricular ejection fraction (LVEF), left ventricular fractional shortening (LVFS), left ventricular end-systolic diameter (LVESD), and left ventricular end-diastolic diameter (LVEDD) were measured in each group. The mean of 3 cardiac cycles were used for statistical analysis.
Histopathological changes in the atria. The atrial tissues of each group were partially retained for protein extraction, and the rest were fixed in 4% paraformaldehyde for 24 h. After dehydration and embedding, paraffin sections were made, and the thickness of each section was 4 μm. HE staining and MASSON staining were performed in strict accordance with the instructions of the kit, and the morphology of atrial tissues and collagen deposition in the interstitial space of cardiomyocytes were observed under the light microscope.
Western blotting. The proteins were extracted from the atrial tissue homogenates of each group, electrophoresed and then transferred to the PVDF membrane (cut the length of PVDF membrane according to the molecular weight of target protein), blocking at room temperature for 2 h in 5% non-fat milk powder, where Phospho-MAPK was blocked with 5% BSA for 2 h. GAPDH (1:3000), p44/42 MAPK (1:1000) and Phospho-MAPK (1:1000) antibodies were added sequentially and incubated on a shaker at 4 ℃ for 12 h. On the next day, the PVDF membranes were washed by TBS-T, and then given a secondary antibody incubation at room temperature for 2 h, the image was acquired by chemiluminescence system (VIVILBER machine), and the grayscale values of the strips were analyzed using Image J.
Statistical methods. Data were statistically processed using GraphPad Prism 9.0, and all data were expressed as mean ± standard deviation. One-way ANOVA was used between groups, and P < 0.05 was statistically significant.

Data availability
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.