Prediction of the Potential Mechanism of Triptolide in Improving Diabetic Nephropathy by Utilizing A Network Pharmacology and Molecular Docking Approach

Background: Triptolide (TP) is a major active component of colquhounia root tablet, which has been long been used in China to treat diabetic nephropathy (DN) due to its marked anti‐inflammatory, antiproteinuric, and podocyte‐protective effects. Methods: This study investigated the anti-proteinuria activity and related signaling cascade of TP in DN by utilizing a network pharmacology and molecular docking approach. Results: From the GeneCard, DisGeNET, and National Center for Biotechnology Information Gene databases, 1458 DN targets were obtained and input together with 303 TP targets into Venny2.1.0 for mapping and comparing. In total, 113 common targets of TP and DN were obtained, of which 7 targets were found to play an important role through theoretical inhibitory constant analysis. The common targets were further analyzed by Kyoto Encyclopedia of Genes and Genomes to identify the pathways related to the therapeutic effect of TP on DN. Among them, the seven targets were found to play key roles in six signaling pathways. The molecular docking results also showed TP had good binding ability to the seven targets. Conclusions: Analysis of the common targets and key pathways showed that TP can improve DN via its anti-nephritis, anti-renal fibrosis, antioxidant, and podocyte-protective effects, which might elucidate the mechanism by which TP improves renal function and reduces proteinuria in DN.


Introduction
Diabetic nephropathy (DN), a serious microvascular complication of type 1 and type 2 diabetes, is characterized by proteinuria and persistent renal function injury [1]. Proteinuria, an independent risk factor of disease progression, is the most important clinical characteristic of DN, and is also the leading cause of end-stage renal disease [2,3]. Without early intervention, 50% of patients with microalbuminuria will progress to macroalbuminuria [4]. There are many risk factors for the development of DN including increased inflammatory factors and oxidative stress, changes in fat and protein metabolism, and overexpression of the renin-angiotensin-aldosterone system. These lead to the apoptosis and loss of renal podocytes and decreased filtration capacity of glomeruli, thus aggravating DN [5,6]. Although several recent studies have confirmed that angiotensin-converting enzyme inhibitors/angiotensin receptor blockers can reduce DN proteinuria and play a role in delaying disease progression, they are ineffective in DN patients with normal blood pressure [7]. Due to the limitations of traditional Western medical approaches, some DN patients have turned to alternative treatments such as traditional Chinese medicine (TCM). Triptolide (TP) is a component of the following traditional Chinese herbal medicines: Tripterygium wilfordii Hook. F., Tripterygium hypoglaucum Levl. Hutch, Tripterygium regerii Sprague et Takeda, and Tripterygium forretii Dicls [8]. It is also a major active component of the colquhounia root tablet and tripterygium glycoside tablet, which have long been used in China to treat DN due to their marked anti-inflammatory, anti-proteinuria, and podocyte-protective effects [9][10][11]. Several randomized controlled clinical trials have indicated that TP possibly imparts nephroprotective effects by decreasing proteinuria, serum creatinine levels, and blood urea nitrogen levels [12][13][14]. Although TP is effective for improving renal function and reducing proteinuria in DN, the exact mechanism is still unclear.
Network pharmacology is a research method based on virtual computing technology, high-throughput data, and public database, which combines system computing with experiments, introducing a new field of pharmacology [15,16]. Network pharmacology constructs a multi-level network of disease-phenotype-gene-drug through multi-target interaction [17]. Through analysis of the overall network, we can better predict drug targets to provide help for the re-search and design of new drugs [18]. For TCM, each component in its prescription has its target, and the effect of the drug is often the result of the synergistic effects of multiple component targets [19]. Network pharmacology can reveal the role of various components at the molecular level so that people can make better use of TCM [20].
In this study, we used network pharmacology to identify the potential targets and signaling pathways of the TCM component TP for the treatment of DN and revealed its possible mechanism. Fig. 1 shows a flowchart of the online pharmacological processes of this study.
The PharmMapper database is a platform for targets prediction, which identifies the potential targets of small molecules by a pharmacophore mapping approach. The chemical structure of TP was drawn by ChemDraw (PerkinElmer, Waltham, MA, USA) and uploaded into the PharmMapper database (Fig. 2). In the PharmMapper database, the maximum number of conformation generation was set to 300 and druggable pharmacophore models were selected as the target set. The names of these target genes were converted to official names by UniProt (https://www.uniprot.org/). TCMSP is a systems pharmacology platform used for screening the active ingredients of TCM. After inputting the keyword "triptolide", targets were obtained from the TCMSP database.
GeneCard is a database with relevant information on proteomics, transcriptomics, and genomics [21]. DisGeNET is a comprehensive database of genes related to human disease [22]. NCBI Gene is a database containing information about multiple species [23]. After searching the keywords "diabetic nephropathy" in the above three databases, the targets of DN were obtained.

Construction and Analyses of the PPI Network
The potential targets of TP and the disease targets of DN were mapped and compared with the Venny2.1.0 platform (https://bioinfogp.cnb.csic.es/tools/venny/index.html), in order to obtain the common targets of TP and DN. The STRING database (https://string-db.org/) contains almost all known and predicted information about protein-protein interactions, including direct and indirect interactions. The validity of these interactions was calculated in the form of confidence scores, ranging from 0 to 1 [21]. The medium confidence level was set to greater than 0.4, with the species "Homo sapiens" [22]. The potential targets of TP and the related targets of TP treatment for DN were uploaded to the STRING database. The protein-protein interaction networks of TP and TP-DN were obtained [22,23]. Cytoscape (https://cytoscape.org/) is an open source network software platform, which can be used to visualize the intermolecular interaction network and combine network and gene expression profile data [24]. The TP and TP-DN target networks obtained from the STRING database were imported into Cytoscape software (version 3.8.2, Institute for Systems Biology, Seattle, Washington, USA), and the "Network Analyzer" function was used to analyze the topology parameters of the network [25].

Gene Ontology and Kyoto Encyclopedia of Genes and Genomes Pathway Enrichment Analyses
Metascape database (https://metascape.org/) is a platform for gene annotation analysis, which can analyze the signaling pathways and biological processes of uploaded target genes [26]. For enrichment analysis, the species was set to "Homo sapiens", the p-value cutoff was 0.05, and other parameters were default [27]. Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed on the targets in turn. The results were saved and sorted according to score, and the relevant biological processes and signal pathways were screened [28]. The results of the filters were put input into a bioinformatics online tool (http://www.bioinformati cs.com.cn/), to draw the relevant pictures [29].

Molecular Docking
The crystal structure of candidate protein binding to TP was obtained from the RCSB Protein Data Bank (https:// www.rcsb.org/) and modified using Autodock (version 2.5, Scripps Research, San Diego, California, USA) to remove ligand and water molecules, and add hydrogen and Kollman charge [30,31]. The three-dimensional structure of TP was obtained from DrugBank (https://go.drugbank.com/) and was also modified by Autodock (version 2.5, Scripps Research, San Diego, California, USA) [32]. First, the active site was confirmed by a eutectic small molecule ligand of the proteins. Then the position of the active site with 60 Å outward was taken as the center of the docking box. Second, the Lamarckian genetic algorithm was used to find the best conditions for docking. Finally, the conformation with the lowest energy was selected as the optimal conformation. The docking results were visualized using PyMol (https://pymol.org/2/), where the hydrogen bonds and binding sites were analyzed.

TP-DN Common Targets
A total of 303 targets of TP were obtained through PharmMapper and the TCMSP database. A total of 1458 targets associated with DN were identified in the GeneCard, DisGeNET and NCBI Gene databases. The TP and DN targets were determined using the Venny2.1.0 data platform, which identified common targets (Fig. 3).

TP-DN Target Network
These common targets were imported into the STRING database, and then the protein-protein interaction relationship was transferred to Cytoscape software to generate the TP-DN target network map (Fig. 4). After analyzing the whole network, 113 nodes and 1618 edges were found, with an average node degree of 28.637 and average local clustering coefficient of 0.648. The nodes in the network were sorted according to the degree value (Table 1), which represent the connection among the nodes in the network (the larger the degree value, the more nodes are associated). Among them, the top 10 targets (tumor necrosis factor [TNF], albumin [ALB], AKT1, vascular endothelial growth factor A [VEGFA], Jun proto-oncogene,

Enrichment Analysis of the TP-DN Target Network
The common targets of TP and DN were imported into Metascape for GO and KEGG analyses, and the results were input into a bioinformatics online tool to obtain the enrichment bubble diagram. The results were sorted according to the p-value. In GO analysis, the biological processes were found to be associated with the positive regulation of cell migration, response to hormone, positive regulation of cell motility, positive regulation of cell component movement, and positive regulation of locomotion, among others (Fig. 5). The cell components were associated with vesicle lumen, secretory granule lumen, cytoplasmic vesicle lumen, membrane raft, membrane microdomain, among others (Fig. 6). Molecular functions were correlated with ligand activated transcription factor activity, nuclear receptor activity, phosphatase binding, protein kinase activity, and cytokine receptor binding, among others (Fig. 7). In KEGG analysis (Fig. 8), the top pathways related to DN were selected for further analysis and included advanced glycation end product-receptor for advanced glycation end product (AGE-RAGE), mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase-AKT (PI3K-AKT), relaxin, forkhead box O (FOXO), and TNF signaling pathways.

Molecular Docking
The binding ability of TP to the proteins in the TP-DN target network was evaluated by molecular docking. These  Table 2.

Discussion
DN is a chronic kidney disease and the leading cause of end-stage renal disease in most developed countries [5]. The causes of DN are complex, but inflammation and oxidative stress are known to be involved in its progression [5,33]. As a new approach, network pharmacology can well analyze the overall relationship between drugs and diseases, including how to participate in the therapeutic process [34]. TP markedly attenuates albuminuria and podocyte injury, regulating the T helper cell balance and macrophage infiltration in an animal model of DN [35,36]. To elucidate the possible mechanism and potential targets of TP in the treatment of DN, we constructed and analyzed the targets through network pharmacology and molecular docking.
Among the 113 TP-and DN-related targets, 7 targets were found to play an essential role through network analysis including ALB, AKT1, CASP3, EGFR, STAT3, TNF, and TP53. Individually, ALB functions as an intravascular transporter, which not only binds a variety of ions, hormones, and drugs but also stabilizes osmotic pressure, antiinflammation, and antioxidation [37]. When the concentration of glucose is too high, glycosylation of ALB occurs [38]. After additional events, glycosylated ALB further forms AGEs and stimulates cells to produce oxidative stress, thus damaging cells [39]. CASP3 belongs to the family of cysteine proteases and is an essential factor in regulating apoptosis [40]. High glucose can stimulate mito- chondria, release cytochrome C, and increase the expression of CASP9 and CASP3 [41]. At the same time, activation of CASP12 and CASP3 through endoplasmic reticulum stress can also be independent of mitochondria, resulting in apoptosis [42]. EGFR is an important receptor tyrosine kinase, which is closely related to the development of DN and is widely distributed in glomeruli and renal tubules [43]. EGFR can be activated by high glucose and Src kinase, mediating the phosphorylation of Akt, stimulating a large number of reactive oxygen species (ROS), and inducing the MAPK signal pathway all of which leads to the release of inflammatory factors and reduces insulin secretion in islet cells, resulting in insulin resistance [44]. The Janus kinase/STAT pathway can be activated by high glucose and ROS, and is involved in the pathogenesis of DN [45]. After being phosphorylated, STAT3 enters the nucleus to stimu-late the transcription of target genes, increasing the expression of inflammatory and fibrosis factors [46]. Inhibiting the activity of STAT3 can decrease TNF-α and interleukin beta 1 (IL-b1) levels, ameliorating renal fibrosis [47]. In diabetic patients, the levels of inflammatory factors are significantly elevated [48]. As an inflammatory factor, TNF can greatly promote the development of DN and damage the glomerular filtration barrier [49]. Moreover, it can bind to insulin-like growth factor binding protein-3 to induce the apoptosis of mesangial cells [50]. TP53 is a tumor suppressor, which regulates the apoptosis of podocytes [51]. After phosphorylation, AKT1 activates the MAPK signaling pathway, releasing a large number of inflammatory factors and causing renal fibrosis [52]. High glucose stimulates the EGFR receptor, which activates the PI3K / Akt signaling pathway, which in turn mediates the transcription of genes via the MAPK pathway, leading to TGF-β, Collagen IV, fibronectin as well as TNF-α, IL-1β of the levels rise. The inflammatory factors will trigger ECM and EMT, causing nephritis, renal fibrosis, proteinuria, meanwhile, inflammatory factors will also bind to the corresponding receptors to stimulate cells to release more inflammatory factors. All events ultimately initiate DN.

Conclusions
A total of 113 common targets of TP and DN were identified by using network pharmacology, and the binding ability of TP to these targets was verified by molecular docking experiments. After KEGG enrichment analysis, six pathways were found to play a key role in the therapeutic effect of TP on DN (Fig. 10). The MAPK signaling pathway is a classic inflammatory pathway, which is composed of p38-MAPK, c-Jun N-terminal kinase 1 (JNK1), and extracellular signal-regulated kinase 2 (ERK2) [53]. When stimulated by ROS, p38-MAPK, JNK, and ERK release signaling factors such as ATF1/2 and c-Jun that me-diate the transcription of transforming growth factor beta (TGF-β), IL-1β, fibronectin, and type IV collagen, leading to nephritis, renal fibrosis, podocyte apoptosis, and proteinuria [54]. A high glucose environment can stimulate the glycosylation of serum ALB and gradually transform it into AGEs [38]. AGEs continuously accumulate and activate RAGE, leading to oxidative stress, activation of the MAPK pathway, chronic inflammation, and eventually renal injury [55][56][57]. The PI3K/Akt pathway can activate the nuclear factor kappa B (NF-κB) pathway, increasing the expression of IL-6 and leading to glomerular basement membrane thickening and mesangial expansion [52]. Meanwhile, Akt can phosphorylate FOXO3a in the FOXO signaling pathway, causing extracellular matrix hyperplasia [58]. Relaxin, a member of the insulin family, has vasodilatory and antifibrotic effects. Activation of the relaxin pathway inhibits SMAD2 activation and TGF-β production, reducing synthesis of the extracellular matrix (ECM) [59]. When FOXO3a is phosphorylated by Akt, the expression of bisindolylmaleimide and manganese superoxide dismutase decreases, resulting in ECM accumulation and accelerating the occurrence of DN [60]. Meanwhile, activation of the TNF pathway will elevate the expression of ROS, leading to the altered permeability of the capillary wall and triggering proteinuria [61]. Generally, the results showed that TP could improve DN via its anti-inflammatory, anti-renal fibrosis, anti-oxidant, and podocyte-protective effects.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Author contributions
MY conceived the study, MY and DF designed re-search; XA, DF, YZ, RT, and JZ performed research; DF, ZY performed data analysis; MY and DF prepared all figures and wrote the manuscript; MY edited the final and prepared the revised manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate
Not applicable.