Shuang Xu,Tonghui Yu,Dan Yuan

Department of Traditional Chinese Medicine,Shenyang Pharmaceutical University,Shenyang 110016,China

AbstractBanxia Xiexin Decoction (BXXXD),a traditional herbal formula,has been used to treat ulcerative colitis (UC) clinically.In this study,chemical compounds and putative targets of BXXXD and UC related therapeutic targets were screened from multiple databases.The protein-protein interaction (PPI) was conducted using String database,and 31 candidate targets were screened from CytoNCA database.The Database for Annotation,Visualization and Integrated Discovery (DAVID) and Metascape database were used for Kyoto Encyclopedia of Genes and Genomes (KEGG) channel and Gene Ontology (GO) enrichment analysis respectively,and the enrichment analysis results were visualized by OmicShare platform.Meanwhile,the interaction network among Chinese herbs,active compounds,candidate targets and pathways was built by Cytoscape 3.7.2 software,and the potential compounds of BXXXD in the treatment of UC were screened.Finally,molecular docking technology was used to verify the putative key compounds.Combined with literature research,5 key compounds for the treatment of UC were identified,which are mainly involved in TNF signaling pathway,cancer signaling pathway,inflammatory bowel disease (IBD),Toll-like receptor signaling pathway,and NF-κB signaling pathway.This study provides a scientific basis for BXXXD as an effective alternative therapeutic agent for UC from a new perspective,and also provides a feasible method for basic chemical research and pharmacological research of BXXXD.

Keywords:Banxia Xiexin Decoction;traditional Chinese medicine;ulcerative colitis;network pharmacology;molecular docking;key compounds

1 Introduction

Ulcerative colitis (UC) is one of inflammatory diseases (IBD),characterized by chronic inflammation of the gastrointestinal tract.Because of the long course of disease,it is difficult to cure and prone to serious complications [1].In the past decades,the incidence of UC has been increasing,which is considered a precancerous lesion of colon cancer.UC has been listed as one of the modern refractory diseases by the World Health Organization and has received increasing attention from the medical community at home and abroad.The clinical manifestations are persistent or recurrent diarrhea,abdominal pain,mucous pus and bloody stools,and tenesmus.However,the cause and pathogenesis of UC are still unclear.It is believed that the occurrence of UC is mainly related to genetic,immune,dietary and other factors.Among them,immune abnormality is considered to be an important factor for the pathogenesis of UC [2,3].

Traditional Chinese medicine (TCM)originated from clinical practices in ancient China and is a comprehensive medical system with potential application value.Different from Western medicine,TCM emphasizes the holistic view and dialectical thinking [4],and aims at a variety of morbidity factors while diagnosing symptoms [5].On the contrary,Western medicine usually uses reductionist methods to adjust the established goal or symptom.The main drugs used in Western medicine to treat UC are aminosalicylic acid preparations led by sulfasalazine,immunomodulators,and antibiotics.However,they cannot effectively treat refractory cases of UC,and the adverse reactions increase during long-term medication.In contrast,TCM shows efficacy and safety in many cases,including the treatment of UC.TCM uses oral decoction,Chinese medicine enema,acupuncture and embedding thread to treat UC,which has the advantages of improving prognosis and reducing side effects [6-8].

TCM formulas are usually composed of many herbs with complex chemical compositions,while UC is related to various factors and mechanisms.Therefore,in order to understand the pathogenesis of UC,network-oriented approach is desirable [9].New research strategies to investigate comprehensive treatment mechanisms based on computational simulation and prediction have flourished in recent years [10].Network pharmacology,a new pattern of drug research,has incomparable advantages over traditional methods in elucidating the comprehensive mechanism.Network models have been used to analyze the relationship among drugs,targets,metabolic pathways,and diseases [11,12],in order to clarify the synergistic effects and potential mechanisms of multi-component and multi-target.It provides other schemes for understanding the interaction between active compounds and candidate targets,such as therapeutic mechanisms elucidation [13],new pharmacological actions prediction [14] and potential biomarkers screening [15].

Banxia Xiexin Decoction (BXXXD) comes from“Treatise on Exogenous Febrile Disease”written by Zhang Zhongjing,a famous medical scientist in Eastern Han Dynasty.The formula consists of Pinelliae Rhizoma,Captidis Rhizoma,Astragali Radix,Zingiberis Rhizoma,Glycyrrhizae Radix,Jujubae Fructus and Ginseng Radix.It has been reported in the literature that BXXXD has a therapeutic effect on UC [16].However,its mechanism of the treatment of UC has not been elucidated.BXXXD contains hundreds of compounds and acts on multiple target proteins,so it is difficult to systematically study the mechanism by conventional methods.Network pharmacology combined with molecular docking technology is very suitable to predict the active compounds in BXXXD and the candidate targets in the treatment of UC,and to clarify the mechanism of action.The results might provide guidance for the improvement of TCM therapy strategies in the future.

2 Materials and methods

The research was conducted in 4 steps:(1)BXXXD and UC related targets were predicted by multiple databases;(2) candidate target acquisition;(3) the related network relationships were constructed using the software of Cytoscape 3.7.2;(4) network analysis and target verification were completed.The flowchart is shown in Fig.1.

Fig.1 Flowchart of the present study

2.1 Chemical compounds database

The chemical compounds of Pinelliae Rhizoma,Captidis Rhizoma,Astragali Radix,Zingiberis Rhizoma,Glycyrrhizae Radix,Jujubae Fructus and Ginseng Radix were retrieved from the Traditional Chinese Medicine Systems Pharmacology database(TCMSP,https://tcmspw.com/tcmsp.php).Oral bioavailability (OB) ≥ 30% and drug-likeness(DL) ≥ 0.18 were used as the screening criteria.Compounds that meet the thresholds of OB and DL will be defined as potential compounds.

2.2 Putative targets of BXXXD and UC related therapeutic targets

TCMSP and Swiss Target Prediction (http://www.swisstarget prediction.ch/) were employed to retrieve compound-related targets according to their computed descriptors or uploading the mol2 structures obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/).

Keyword“Ulcerative Colitis”and species“Homo sapiens”were used to search the therapeutic targets related to UC from seven sources:(1) Online Mendelian Inheritance in Man (OMIM,http://www.omim.org/),(2) Genetic Association Database (GAD,https://geneticassociationdb.nih.gov/),(3) Drugbank database (https://www.drugbank.ca/,version 5.1.3),(4) Comparative Toxicogenomics Database (CTD,http://ctd.mdibl.org),(5) PharmGkB database(https://www.pharmgkb.org/),(6) Therapeutic Target Database (TTD,http://db.idrblab.Net/ttd/),and (7)DisGeNET database (https://www.disgenet.org/).All the protein names were standardized into formal gene symbols via UniProt Knowledgebase (http://www.uniprot.org/).

2.3 Protein-Protein Interaction (PPI) networks analysis

After the putative targets of the compounds in BXXXD and the known therapeutic targets on UC were combined and connected,PPI data were integrated and retrieved from String database (http://string-db.org/cgi/input.pl),with the species set as“Homo sapiens”and the confidence score as 0.

2.4 Network construction

The following networks were constructed:(1) BXXXD chemical compounds-putative targets and UC related therapeutic targets network;(2) medicinal materials-candidate targets network;(3) candidate targets-potential active compounds network;(4) pathways-candidate targets-potential active compounds network.

Cytoscape 3.7.2 software was employed for integration,analysis and visualization of all network data.In the generated network,compounds,targets or diseases were represented by nodes.Edges represented interactions between various nodes.Degree centrality,Betweenness centrality and Closeness centrality of each node,as well as the topological characteristics,were analyzed by CytoNCA to screen candidate targets and potential active compounds of BXXXD.

2.5 Pathway enrichment analysis

To clarify the pathways where the candidate targets were involved,Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes database (KEGG) signaling pathways enrichment analysis were performed using Metascape database(https://metascape.org/) or Database for Annotation,Visualization and Integrated Discovery database(DAVID,https://david.ncifcrf.gov/) respectively.All the results were visualized by OmicShare platform(https://www.omicshare.com/) with“Homo sapiens”as the species and

P

＜ 0.01.

2.6 Molecular docking analysis

AutoDock Vina 1.1.2 was used to study the affinity and action mode of active compounds with target proteins [17].The three-dimensional structure of ligand was downloaded from PubChem and optimized by MMFF94 force field,and the crystal structure of receptor protein was downloaded from RCSB protein database (https://www.rcsb.org/).AutoDockTools 1.5.6 was used to create coordinate files of receptors and ligands in PDBQT format.AutoGrid program was used to calculate the grid point and dock the grid map of each atom in the ligand.Grid point covers the active site of the macromolecule.To increase the accuracy of calculation,the parameter exhaustiveness is set to 20,and default values were used for other parameters.

Molecular docking simulation was completed by AutoDock program.In the present study,semiflexible docking was adopted and the conformation with the lowest binding energy was selected for further interaction analysis.Python Molecule Viewer 1.5.6 was used to visualize the docking complex of proteins and ligands with good affinity,and PyMOL 1.1.7 was used to analyze the binding mode.

3 Results

3.1 BXXXD and UC related therapeutic targets

A total of 211 chemical constituents of 7 Chinese herbs in BXXXD were screened by TCMSP,including 14 compounds in Pinelliae Rhizoma,14 compounds in Captidis Rhizoma,36 compounds in Astragali Radix,5 compounds in Zingiberis Rhizoma,92 compounds in Glycyrrhizae Radix,29 compounds in Jujubae Fructus and 22 compounds in Ginseng Radix.After removing the redundancy,163 chemical compounds and 596 putative targets of BXXXD were obtained.Meanwhile,700 UC related therapeutic targets were retrieved from multiple databases.After merging with the putative compound targets,79 overlapping proteins were obtained,which are recognized as the potential therapeutic targets of BXXXD for treating UC (Fig.2).

Fig.2 79 potential therapeutic targets of BXXXD against UC

3.2 Chemical compounds-Putative targets and UC related therapeutic targets network

The network was constructed by connecting chemical compounds with their potential targets.There were 759 nodes (163 chemical compounds,517 putative targets,and 79 potential therapeutic targets) and 4146 edges.As shown in Fig.3,most of the targets could be regulated by several compounds,while some targets such as IL6,PTGS2,TNF and NFKBIA could be regulated only by one compound.In addition,compound putative targets included some disease targets (potential therapeutic targets),indicating that drugs might indirectly regulate disease-related proteins,and BXXXD might directly act on these proteins or indirectly treat disease by regulating compound putative targets.

Fig.3 BXXXD chemical compounds-putative targets and UC related therapeutic targets network.The network was constructed by linking different herbs of compounds (polychrome triangles) and their putative targets (pink round rectangles and light green V-shapes)

3.3 Candidate targets for BXXXT against UC

79 intersection targets obtained by mapping the chemical composition and the disease were analyzed by PPI network in String database.According to CytoNCA analysis results,nodes with Degree centrality,betweenness centrality and closeness centrality values exceeding the median were identified as candidate targets.Degree centrality,betweenness centrality and closeness centrality of candidates of the chemical profile-based network were shown in Table 1.31 candidate targets were identified,including IL6,ALB,TP53,VEGFA,TNF,CXCL8,STAT3,PTGS2,CASP3,JUN,CCL2,IL1B,MMP9,ICAM1,IL4,IL2,PPARG,CXCL10,IFNG,RELA,VCAM1,CCND1,NOS3,MMP2,HMOX1,CRP,MPO,JAK2,CASP8,NFKBIA and AHR.

Table 1 The detailed Degree centrality,betweenness centrality and closeness centrality of candidate targets of the chemical profile-based network

3.4 Potential active compounds of BXXXD against UC

The network was built to show the relationship between 7 Chinese herbs and their candidate targets.As shown in Fig.4,the network consists of 38 nodes and 93 edges.

Ginseng Radix

(30) has the highest connection with other nodes,followed by Glycyrrhizae Radix (17) and Astragali Radix (17).It was speculated that they might play a major role in the treatment of UC.Besides,the network showed multi-compound and multi-target characteristics of the TCM formula.One medicinal material corresponds to multiple targets and one target corresponds to multiple Chinese herbs.PTGS2 is a good example,which corresponds to 6 Chinese herbs.

Fig.4 Chinese herbs-candidate targets network.The pink round rectangle represents the candidate target,and the orange border hexagon represents 7 Chinese herbs of BXXXD.(Note:RS,Ginseng Radix;BX,Pinelliae Rhizoma;HL,Captidis Rhizoma;GC,Glycyrrhizae Radix;DZ,Jujubae Fructus;GJ,Zingiberis Rhizoma;HQ,Astragali Radix.)

3.5 Potential active compounds of BXXXD against UC

To further analyze the therapeutic effect of BXXXD on UC,candidate targets-interacted compounds network was constructed as shown in Fig.5.153 compounds exerted the therapeutic effect on UC by regulating the candidate targets.Kaempferol,Ginsenoside Rh2,Baicalein,

β

-Carotene,

β

-Sitosterol,Formononetin,Acacetin,Wogonin and Licochalcone A were identified as potential active compounds with Degree Centrality values more than twice the median.Additionally,PTGS2,PPARG,NOS3,CASP3,IL2,RELA,and MMP9 also played an important role in network construction.

Fig.5 Candidate targets-compounds network (Pink octagon represents the candidate targets,green circle represents the related compounds,and green circle with the gray border represents the potential active compounds.Node size indicates the degree of connection between candidate targets and compounds.)

3.6 Pathway enrichment analysis

3.6.1 GO enrichment analysis

GO analysis results of Metascape database(

P

＜ 0.01) were shown in Fig.6.With -log

P

as abscissa and GO term as ordinate,we screened out the top 5 cellular components (CC),top 5 molecular function (MF),and top 15 biological processes (BP)visualization for target protein analysis.The results indicated that these target proteins were significantly correlated with membrane raft,transcription factor complex,outer membrane of plasma,endoplasmic reticulum cavity and secretory particle cavity in CC,cytokine receptor binding,nuclear receptor activity,heme binding,G protein-coupled receptor binding,protein domain-specific binding in MF,and cytokinemediated signaling pathway,negative regulation of cell proliferation,response to antibiotics,apoptosis signaling pathway,nerve enrichment of inflammation,differentiation of leukocytes,and regulation of intercellular adhesion in BP.

Fig.6 GO analysis of candidate targets for BXXXD against UC (Green represents MF,yellow represents CC and orange represents BP.)

3.6.2 KEGG pathway enrichment analysis

KEGG results of David database showed that 58 pathways were related,and the first 20 pathways were visualized by the OmicShare platform.To investigate candidate targets and related compounds involved in the 20 pathways,a network was constructed as shown in Fig.7.The analysis results were consistent with the results obtained by CytoNCA under item 3.5.Therefore,it was speculated that Kaempferol,Ginsenoside Rh2,Baicalein,

β

-Carotene,

β

-Sitosterol,Formononetin,Acacetin,Wogonin,and Licochalcone A were potential active constituents of BXXXD in the treatment of UC.These compounds were from Ginseng Radix,Astragali Radix,Glycyrrhizae Radix,Pinelliae Rhizoma,Zingiberis Rhizoma,and Jujubae Fructus which acted on 8 candidate targets(PTGS2,NOS3,IL2,MMP9,MPO,RELA,TNF,and STAS3).They were larger than the median,and mainly involved in TNF signaling pathway,cancer signaling pathway,inflammatory bowel disease(IBD),Toll-like receptor signaling pathway,NF-κB signaling pathway and T cell signaling pathway.

Fig.7 Active compounds-candidate targets-pathway network (Pink v-shape represents the pathway,yellow border hexagon represents candidate target,and triangle represents the active compounds.Green,pink,orange,and bluepurple triangles represent active compounds with the degree values of 8-9,5-6,3-4,1-2,respectively.)

3.7 Molecular docking

AutoDock Vina 1.1.2 platform was used to calculate the binding free energy of 9 potential active compounds and their related candidate targets(PTGS2,NOS3,IL2,MMP9,MPO,RELA,TNF,and STAS3).The lower the binding free energy of the ligand,the higher the binding affinity for their receptors.

Except for

β

-Carotene,the binding free energy of the other 8 chemical components exhibited was relatively low,ranging from -6.7 to-10.5 kcal/mol.The docking energy value of PTGS2 with Ginsenoside Rh2 was -9.3 kcal/mol,lower than that of reference molecules.The amino acid residues HIS90 and PRO515 in the active region of PTGS2 form hydrogen bonding interaction with Ginsenoside Rh2 molecule and amino acid PHE581 (Fig.8A).The docking energy value of TNF with Kaempferol molecule was -9.1 kcal/mol.At the active site of TNF protein,Kaempferol molecule formed a hydrophobic interaction with ALA51,LEU75 and LEU167 (Fig.8B).In the interaction model between Baicalein and MMP9,Baicalein formed hydrogen bonding interaction with amino acid residue GLN227 of MMP9 and forms hydrophobic interaction with ALA189 and TYR248(Fig.8C).The molecular docking energy value was-10.5 kcal/mol,much lower than the reference cocrystal molecule.Compared with the reference cocrystal molecule,Baicalein had a higher affinity for MMP9 protein.Kaempferol formed a hydrophobic interaction with the amino acid residues PH353,CYS184,and TRP178 in the active region of NOS3(Fig.8D),and its molecular docking energy value was -9.3 kcal/mol.Molecular docking studies showed that the main compounds contained in BXXXD had a significantly binding ability with the selected protein targets,which verified the results of network pharmacology.

Fig.8 Molecular docking results.Autodock Vina was used to calculate the action modes of active compounds with specific proteins:(A) the action mode of Ginsenoside Rh2 and PTGS2;(B) the action mode of Kaempferol and TNF;(C) the action mode of Baicalein and MMP9;(D) the action mode of Kaempferol and NOS3

4 Discussion

Immunomodulatory dysfunction is the internal core mechanism of UC.Therefore,the pathophysiological process of UC might be related to autoantibodies,cellular immunity,cytokines,cyclooxygenase and matrix metalloproteinases,and nitric oxide.UC is characterized by autoimmunity,abnormal expression of cytokines,immune cell infiltration,epithelial cell damage and intestinal barrier dysfunction.Many predictions of signaling pathways involved in the treatment of BXXXD against UC focused on NF-κB signaling pathway.It was reported that NF-κB plays an important role in the occurrence and development of UC.NFκB activated by the upstream molecules of the signaling pathway promoted the expression of a large number of inflammatory mediators,thereby causing intestinal mucosal inflammation and damage.These inflammatory mediators further activated NF-κB,formed a positive feedback loop,and triggered the intestinal inflammatory response.Mucosal damage was then further exacerbated.Tolllike receptors (TLRs),as transmembrane receptors and signal transduction receptors important for innate immunity,activated NF-κB through specific pathways after binding to pathogen-related model molecules,which in turn caused the release of inflammatory mediators [18].Moreover,T cell subsets would differentiate into Th1,Th2.When the ration became unbalanced,regulatory T (iTreg) and Th17 in the intestinal mucosa would be induced,thus producing a large number of pro-inflammatory cytokines,such as tumor necrosis factor (TNF-

α

) and interleukin (IL).These pro-inflammatory cytokines,in turn,activated mucosal proteases,resulting in cell tissue damage and long-term inflammation [19,20].Therefore,inflammation-immune-related pathways,such as Toll-like receptor signaling pathway and T cell signaling pathway,might be one of the potential strategies to treat UC.Molecular docking analysis results showed that Baicalein had the highest binding affinity with MMP9 and MPO.It has been reported that Baicalein can inhibit the expression of MMP9 [21] and downregulate the activity of MPO [22] to achieve the anti-inflammatory purpose.Kaempferol had the highest binding affinity for TNF and NOS3.It has been reported that Kaempferol can inhibit TNF expression and exert anti-inflammatory activity [23].Kaempferol,Baicalein,Wogonin,Acacetin,Licochalcone A,and Formononetin are flavonoids and

β

-Sitosterol is a sterol compound.They were reported to exert anti-inflammatory activity by downregulating the expression of PTGS2 or iNOS [24-32].The results of molecular docking analysis were based on protein co-crystal molecules and were consistent with the results of published reports.Additionally,Kaempferol,Baicalein,Wogonin,Licochalcone A and

β

-Sitosterol had been reported to have therapeutic effects on UC [33-37].However,there was no report about the therapeutic effect of Ginsenoside Rh2,Acacetin and Formononetin on UC.Ginsenoside Rh2,as one of the most important ginsenosides in Ginseng Radix,has antiinflammatory and anti-tumor effects.The results of molecular docking indicated that Ginsenoside Rh2 had a high binding activity with candidate targets.Therefore,Ginsenoside Rh2,Formononetin,Acacetin might be potential active compounds against UC.The finding suggested a new way for the treatment of UC.

PTGS2 (also known as COX-2) produces pro-inflammatory mediators (such as PEG2) and is considered the most important indicator of inflammation.Its expression has been observed to increase in clinical specimens and experimental models [38].The present study screened 8 potential active compounds,which could inhibit PTGS2 activity.More importantly,the 8 potential active compounds had a high binding activity with NOS3,IL2,MMP9,MPO,RELA,CAPS3,TNF,and STAS3.Among them,the levels of TNF,RELA,STAT3,and IL2 reflected the degree of inflammation.Potential active compounds in BXXXD regulated cytokine levels by acting on these targets.Meanwhile,they also had the potential to regulate proteases.They acted on NOS3,MMP9,and MPO,and had the similar therapeutic effects to oral sulfasalazine.Therefore,the potential active compounds screened in this study play a synergistic role in the treatment of UC by regulating multiple targets and multiple pathways.The study provided a scientific basis for BXXXT as an effective alternative therapeutic agent for the treatment of UC.

The research predicted candidate targets of BXXXD against UC and revealed the influence of its chemical constituents on the main pathways through the related candidate targets,thus elucidating the mechanism of BXXXD against UC and providing scientific basis.In addition,we studied the 5 active ingredients that have been reported to treat UC at the molecular level,and the results showed that they all had good binding activity.Therefore,Kaempferol,Baicalein,Wogonin,Licochalcone A and

β

-Sitosterol were proved to be effective ingredients of BXXXT in the treatment of UC.The results provided scientific basis and reference for future clinical medicine.

5 Conclusion

This study aimed to investigate the mechanisms of BXXXD in the treatment of UC.31 candidate targets were screened out from the network pharmacology molecules based on chemical characteristics.Furthermore,we analyzed the interaction among“active ingredients,candidate targets and KEGG pathways”,and used molecular docking to verify the binding activity.Finally,Kaempferol,Baicalein,Wogonin,Licochalcone A and

β

-Sitosterol are selected as the key compounds,which were involved in TNF signaling pathway,cancer signaling pathway,inflammatory bowel disease (IBD),Toll-like receptor signaling pathway,NF-κB signaling pathway,and T cell signaling pathway.