Protective Mechanism of Qi-Dan-Dihuang Decoction in Diabetic Kidney Disease Based on Network pharmacology

Background: Qi-Dan-Dihuang Decoction (QDD) has been used for treating diabetic kidney disease (DKD), but the mechanisms are poorly understood. The aim of this study is to reveal the therapeutic effects and the mechanism of QDD in ameliorating DKD by network pharmacology, in vivo, and in vitro studies. Methods: The effect of QDD on body weight, fast blood glucose, oral glucose tolerance test (OGTT), 24 h urinary protein (24hU-Pro), serum creatinine (Scr), blood urea nitrogen (BUN), and pathological evaluation in kidney were investigated in vivo using C57BLKS/J db/db mice. The main active compounds of QDD, compound-disease interaction targets, and related processes and pathways were discerned by network pharmacology analysis through Chinese Medicine Systems Pharmacology Database (TCMSP) and TCM Database@Taiwan. The protein-protein interaction (PPI) network were established through STRING database. GO and KEGG pathway were used for analysis processes and pathways. Then Western blot was used to veried the predicted results. Finally, cell viability, wound healing and mainly pathway protein expression were detected in vitro using renal tubular epithelial cells HK-2 and NRK-52E cells. Results: Although QDD treatment showed no signicant difference in FBG and AUC of OGTT, but had signicant reduction in Scr level in C57BLKS/J db/db mice. Histopathologic results showed that QDD ameliorated the expansion of mesangial area, thickened membranes of Bowman’s capsules and basement membrane of glomerular capillaries, renal tubular epithelial cells vacuolar degeneration and reversed the glomerular and tubulointerstitial in C57BLKS/J db/db mice. For network pharmacology analysis of QDD, 143 active compounds related to 274 possible targets in QDD obtained and 117 compound-disease interaction targets were screened out combining with Genecards database. 18 key targets was excavated through network topological analysis. GO and KEGG pathway enrichment analysis showed that compound-disease interaction targets were signicantly enriched in processes and pathways that are closely related to DKD. Western blot results showed that QDD signicantly attenuated EMT-related proteins, p-NF-κb, IL-1β, IL-18, p-p38MAPK/p38MAPK, p-AKT/AKT, and p-mTOR/ mTOR protein expressions. Treatment with QDD could alleviate cell viability damaged, EMT process, p-NF-κb, IL-1β, IL-18, p-p38MAPK/p38MAPK, p-AKT/AKT and p-mTOR/ mTOR protein expressions by high D-gulcose. Conclusions: This study provides convincing evidence suggest that QDD protects renal brosis of DKD, by regulating EMT in RTECs and inammatory response through p38MAPK and AKT/mTOR signaling pathways. cell viability damaged, EMT process, p-NF-κb, IL-1β, IL-18, p-p38MAPK/p38MAPK, p-AKT/AKT and p-mTOR/ mTOR protein expressions by high D-gulcose. According to the result, we hypothesize that QDD potentially attenuates brosis by regulating EMT via the p38MAPK and AKT/mTOR signaling pathways.


Introduction
Diabetic kidney disease (DKD) is a common complication of diabetes mellitus, which refers to speci c pathological structural and functional changes in the kidney of diabetes mellitus (DM) patients. It is clinically characterized by proteinuria, hypertension, and progressive reduction in renal functions [1,2]. The pathologic changes in diabetic kidney disease include renal hypertrophy and extracellular matrix accumulation, which contribute to glomerular sclerosis, and gradually lead to renal failure through the

Biochemical Analysis
The body weight was measured weekly. Fast blood glucose and oral glucose tolerance test (OGTT) were assesses every four weeks. After 12 weeks of administration, 24 h urine of mice was collected for 24 h urinary protein (24hU-Pro) detection using Urine protein test kit (Nanjing Jiancheng Bioengineering Institution, China). Serum and kidney were dissected from mice after anesthetized with 1% pentobarbital (40 mg/kg). Serum creatinine (Scr) and blood urea nitrogen (BUN) were measured with serum creatinine and BUN assay kit (Nanjing Jiancheng Bioengineering Institution, China).

Morphologic Evaluation
Kidney tissues xed in 10% formalin in phosphate buffer (pH 7.4) were embedded in para n and 4 µm thick slices and subsequently stained with the hematoxylin-eosin (HE), periodic acid-Schiff (PAS), and Masson methods. Images under bright light microscope were acquired with a digital camera with magni cation ×400 (Nikon, Japan).

Identi cation of candidate compounds in QDD
Compounds of four herbs (HMM, RS, RD, and LC) in QDD were obtained from Traditional Chinese Medicines Systems Pharmacology (TCMSP, http://lsp.nwsuaf.edu.cn/tcmsp.php) database. The compounds of RC were obtained from TCM Database@Taiwan (http://tcm.cmu.edu.tw/zh-tw/index.php) because there is no item for RC in TCMSP database.

Screening for active compounds
The properties of absorption, distribution, metabolism and excretion (ADME) are important indicators for pharmacokinetics of compounds and help to minimize their potential drug-drug interactions in herbs [13].

Identi cation of drug targets and compounds-targets network construction
The targets of active compounds were identi ed from TCMSP database and then transformed into o cial symbols via the UniProt KB search function in the UniProt database (http://www.uniprot.org/) by entering the protein name and de ning the species as "homo sapiens". According to the relationship between herbs, active ingredients and corresponding targets, the "herb-compound-target" network were established by Cytoscape 3.6.1. The nodes in the network represent herbs, active compounds or targets, and the edges indicate the relationships between the herbs and components, or compounds and targets. Then the network analyzer plugin in the software were used to analyze the topological properties of the network, such as degree, betwe/enness centrality and closeness centrality.

Collection of genes related to DKD brosis
The GeneCards database (https://www.genecards.org/) provides comprehensive information on all annotated and predicted human genes, which integrates gene-centric data from about 150 web sources including genomic, transcriptomic, proteomic, genetic, clinical and functional information. The related genes of DKD and brosis together with scores that ranks genes by how closely they are associated with the disease were obtained from GeneCards by the searching strategy "([disorders] (diabetic AND nephropathy) OR [disorders] (diabetic AND kidney AND disease)) AND [disorders] ( brosis)".

PPI network construction and key targets identi cation
The compound-disease interaction targets were obtained by intersecting targets in compounds and disease. The PPI network data was imported from the STRING database (https://string-db.org/) by inputting the protein symbols of interaction targets and choosing "Homo Sapiens" as organism. The Cytoscape was used to draw the PPI network, whose topological features, including degree centrality (DC), betweenness centrality (BC), and eigenvector centrality (EC), Closeness centrality (CC), Network centrality (NC) and Local average connectivity (LAC), were subsequently analyzed by cytoNCA [15,16].
2.10. Gene Ontology and KEGG pathway enrichment analysis for targets DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/) was used to subject GO and KEGG pathway enrichment analysis for compound-disease interaction targets, and p < 0.05 was considered signi cant with hypergeometric test.
ClueGO was utilized to assess the enriched KEGG pathways of interaction targets [17], only terms with at least 5 input genes and 5% of term associated genes were enriched, and terms with p < 0.05 as well as kappa score > 0.3 were shown.

Cell culture and treatment
Human and rat renal tubular epithelial cells HK-2 and NRK-52E cells were purchase from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). NRK-52E cells were maintained in Dulbecco's Modi ed Eagle Medium (DMEM; Gibco, USA), and HK-2 cells were cultured in DMEM/F-12 medium (Gibco, USA), in a humidi ed atmosphere with 5% CO 2 at 37°C. Culture medium for all cells was supplemented with 10% fetal bovine serum (Biological Industries, Israel), 100 U/mL penicillin and 0.1 mg/mL streptomycin (Invitrogen, USA). 30 mmol/L D-glucose in medium for 48 h was used to induce epithelial-mesenchymal transition (EMT) process in the cells.

Cell viability assays
Cells were seeded in 96-well plates in the density of 3 × 10 4 cells/well. After 24 h treatment with different mediums, cells were incubated with MTT for 4 h and 100 µL DMSO was added to each well. The absorbance was recorded at 570 nm by Multiskan FC Microplate Reader (Thermo Scienti c, USA) 2.13. Wound healing assay Cells were plated onto 6-well plates until con uence. Wounds were scratched by 200 µl pipette tip. The culture medium was replaced with fresh serum-free medium. The pictures were taken by an inverted microscope (Nikon, Japan) every 24 h. The wound areas were measured and the relative wound measured were calculated.

Western Blotting
The tissues and cells were lysed in RIPA buffer (Beyotime, China) and protein concentration was measured by BCA Protein Assay Kit (Beyotime, China). A total of 30 µg of total protein was separated by 10% SDS-PAGE and then transferred to a polyvinylidene di uoride lm (Millipore, MA, USA). Immune complexes were detected using enhanced chemiluminescence (ECL kit, A nity, USA). Protein expression levels were normalized to β-actin and quanti ed by Image J (version 1.6)

Statistical analysis
All data are expressed as the means ± SDs and analyzed by SPSS 20.0 (IBM, Armonk, USA). Tukey's test was used for multiple comparison. The values were considered statistically signi cant when p < 0.05.

QDD alleviates renal dysfunction and brosis in db/db mice
Compared to db/m mice, the fast blood glucose, area under the curve (AUC) of OGTT, 24 h U-Pro and Scr were signi cantly increased in db/db mice at 0, 4, 8, 12 weeks, which was reversed by dapagli ozin at 4, 8, 12 weeks. However, QDD treatment showed no signi cant difference in FBG and AUC of OGTT compared to model group. QDD (2% and 4% in diet) and dapagli ozin treatment group mice were signi cantly decreased 24 h U-Pro and Scr levels ( Figure 1A-G). BUN levels shown no signi cant differences in each group ( Figure 1H).
The improvement effects of QDD on pathological changes are shown in Figure 2A-C. H&E staining showed that mesangial area moderately-expanded, membranes of Bowman's capsules and basement membrane of glomerular capillaries thickened, and renal tubular epithelial cell the vacuole denaturalized in kidney of model group. QDD and dapagli ozin could reversed the pathologic changes in different degrees. PAS and MASSON staining results showed that, compared with model group, QDD and dapagli ozin could reverse the kidney glycogen and collagen deposition in glomerular and tubulointerstitial. These results suggested that QDD ameliorates the renal pathological conditions by reducing the damages of kidney in the diabetic db/db mice.

Screening of active compounds in QDD
A total of 657 compounds were retrieved from TCMSP and TCM Database@Taiwan databases, among them, 87 in HMM, 202 in RS, 17 in RR, 71 in RD, and 280 in LC respectively. By ADME screening, 143 candidate compounds of QDD were identi ed after removing 7 duplicate ingredients. More than 50% compounds in HMM, RS, RD and LC, possess high OB value (Figure 3A), demonstrated that QDD exhibited high oral absorption and utilization rate. Furthermore, most of ingredients were exacted from LC, which could explain the reconcile function of LC in QDD.

Targets identi cation of QDD and DKD
Based on TCMSP database, except 9 compounds without any relevant targets, the potential targets of active ingredients in QDD were obtained. In total, 134 of the 143 compounds in QDD were associated with 273 targets after eliminating overlapping proteins. A compound-target network was constructed in Cytoscape ( Figure 3C). In the compound-target relationship, top 10 compounds that connects the most targets were MOL000098 (quercetin), MOL000422 (kaempferol), MOL000006 (luteolin), MOL003896 3.4. Network construction for Compound-Target-Disease 134 compounds and 146 compound-disease interaction targets were used to construct the compoundtarget-disease (CTD) network based on their relationships shown in Cytoscape ( Figure 4A). In the CTD network, compounds or targets with bigger size indicates higher degree, such as quercetin, luteolin, PTGS2, and ESR1, which suggested quercetin and luteolin were probably contribute to the major therapeutic effect through PTGS2 and ESR1. However, some components only had one target, such as denudatin B, doradexanthin, gadelaidic acid, implied that these components may played a less important role or are poorly studied in their therapeutic effects in DKD. The network suggested that the therapeutic effects of QDD for DKD are mediated by multiple compounds via multiple targets.

PPI network of drug-disease interaction targets and network analysis
The PPI information of compound-disease interaction targets obtained from STRING were imported into Cytoscape to established the network ( Figure 4B

GO enrichment analysis
In GO enrichment analysis, there were enriched 300, 53, and 94 signi cantly GO terms in biological processes, cell components and molecular functions and listed and ranked by signi cance in descending order ( Figure 5A). These terms are essential structures and functions for the survival of cells and organisms. Furthermore, there are many GO terms in biological processes corresponding to DKD brosis development processes, such as positive regulation of broblast proliferation, response to insulin, wound healing, in ammatory response, glucose metabolic process, response to leptin, response to oxidative stress, and epithelial to mesenchymal transition ( Figure 5B).

Pathways enrichment analysis for key targets
To examine the genetic pathways that were manipulated by the compounds and in uenced the disease development, the pathway network of interaction targets was constructed by KEGG and visualized by Cytoscape ( Figure 6A). A total of 128 signi cant pathways were identi ed by KEGG enrichment analysis and the top 20 signi cantly enriched pathways were shown ( Figure 6B-C). The potential targets were major distributed in endocrine and metabolic systems, immune and in ammatory systems, and signaling transduction and disease-related pathways. Signaling transduction pathways such as PI3K-Akt signaling pathway, VEGF signaling pathway, MAPK signaling pathway, mTOR signaling pathway, NF-κB signaling pathway, and TGF-beta signaling pathway, are closely related to the pathogenesis of DKD. Our study found, in 123 enriched pathways, 12 and 15 pathways were classi ed to endocrine and immune systems, respectively, which constituted the core pathological mechanisms of DKD. Moreover, we also found that apart from type I and II diabetes mellitus, there were many other diseases highlighted, such as cancer and infectious disease. It indicated that QDD may had a potential therapeutic application for them, which matched the TCM theory of "Homotherapy for Heteropathy". These results suggested that QDD could ameliorated DKD conditions by regulating the endocrine and immune systems through multiple signaling pathways.
3.8. QDD attenuated EMT in the kidney of db/db mice and renal tubular epithelial cells As indicated by the network pharmacology prediction, many candidate targets of QDD were related to the EMT, which is a proven cause of renal brosis. Western blot was used to evaluate the expression levels of EMT-related proteins. As shown in Figure 7A, compared with control group, kidneys in model group displayed higher expression of bronectin, N-cadherin, α-SMA, and Vimentin, and lower expression of ZO-1, E-cadherin. QDD and dapagli ozin treatments could remarkably reverse the changes in EMT-related proteins. These results suggested that QDD attenuated EMT process, which may underline its therapeutic effects on brosis and renal dysfunction in db/db mice.
In HK-2 and NRK-52E cells, high glucose treatment reduced the protein expressions of E-Cadherin, ZO-1, upregulated N-Cadherin, α-SMA, and Vimentin. QDD signi cantly increased the protein levels of E-Cadherin and ZO-1, decreased the protein levels of N-Cadherin, α-SMA, and Vimentin compared to high glucose group ( Figure 7B). Then, we determined the effect of QDD on high glucose cultured tubular epithelial cell migration using wound healing assay. Relative healing areas in the wounded cells were measured at 0, 24, 48, and 72 h after the wound injury. As shown in Figure 7C, the relative healing areas were signi cantly increased by high glucose compared with normal group, and QDD signi cantly reduced healing areas compared with high glucose group, indicated that QDD inhibited high glucose-induced cell migration.
3.9. QDD attenuated in ammatory response in the kidney of db/db mice and renal tubular epithelial cells Compared with control group, the protein expressions of p-NF-κB, IL-1β, and IL-18 in the kidneys were signi cantly increased in model group, and QDD group showed signi cant decreasing in p-NF-κb, IL-1β and IL-18 protein expressions ( Figures 8A). In vitro, QDD suppressed the expressions of p-NF-κb, IL-1β and IL-18 in HK-2 and NRK-52E cells, which were upregulated by high glucose (Figure 8B), indicated that QDD was a promising intervention to resolve in ammation in the kidneys of db/db mice and high glucose-cultured renal tubular epithelial cells.

Effect of QDD on p38MAPK and AKT-mTOR signaling pathways
Network pharmacology predicted that QDD delayed the progression of DKD by regulating MAPK, PI3K-Akt-mTOR signaling pathways, western blot was used to detected the changes in MAPK and PI3K-Akt-mTOR effector proteins. As shown in Figure 9A, mice in model group possessed higher levels of p-p38MAPK/p38MAPK, p-AKT/AKT, and p-mTOR/mTOR than in control group, and QDD treatment signi cantly reduced the protein expressions of p-p38MAPK/p38MAPK, p-AKT/AKT, and p-mTOR/mTOR compared with model group. Moreover, QDD reduced the expressions of p-p38MAPK/p38MAPK, p-AKT/AKT, and p-mTOR/mTOR in HK-2 and NRK-52E cells, which were increased by high glucose ( Figure  9B). These results suggest that QDD alleviated EMT and resolved in ammation in kidneys of db/db mice and high glucose-cultured RTECs probably through the p38MAPK, AKT-mTOR signaling pathways.

Discussion
In this study, we found that QDD treatment could reduced Scr level and ameliorated the histopathological damage in kidney in C57BLKS/J db/db mice. Then, we used network pharmacology to analysis the potential targets of QDD. PPI network analysis revealed that AKT1 and mTOR could be the key targets for renal protective effects of QDD. GO functionally analysis for biological process showed that EMT was signi cantly enriched, and KEGG pathway analysis elucidated that PI3K/AKT signaling pathway, mTOR signaling pathway and MAPK signaling pathway may be closely related to the therapeutic effects of QDD. Western blot of kidney tissues showed that QDD signi cantly attenuated EMT-related proteins, p-NFκb, IL-1β, IL-18, p-p38MAPK/p38MAPK, p-AKT/AKT, and p-mTOR/ mTOR protein expressions. Treatment with QDD could alleviate cell viability damaged, EMT process, p-NF-κb, IL-1β, IL-18, p-p38MAPK/p38MAPK, p-AKT/AKT and p-mTOR/ mTOR protein expressions by high D-gulcose. According to the result, we hypothesize that QDD potentially attenuates renal brosis by regulating EMT via the p38MAPK and AKT/mTOR signaling pathways.
TCM has been used for DKD prevention and treatment for centuries. With the proven therapeutic e cacy and few adverse effects, TCM has attracted increasing attention and wide range of applications in DKD [18][19][20]. However, the complicated ingredients, miscellaneous targets and pathways are still the big challenge for TCM research. Meanwhile, DKD involves variety of physiological and pathological process [21], which makes it more di cult for us to reveal the mechanisms of action of the TCM formula therapeutic effects. Our previous studies demonstrated that QDD attenuated renal brosis in DKD by regulating the renin-angiotensin system and the NF-κB pathway [9,22]. In the present study, we have an in-depth investigation on the therapeutic role of QDD in DKD.
The data in this study suggested that the mechanism of QDD for DKD treatment might have multiple components and multiple targets. Some compounds, which predicted in QDD, have been con rmed to be related to DKD in previous reported studies. For example, quercetin, a compound screened out from HMM and LC, protects renal function in STZ-induced diabetic nephropathy rats, with anti-oxidation, anti-brosis, anti-in ammatory effects, and regulates glucose and lipid metabolism [23][24][25][26]. Luteolin, tanshinone a, formononetin, and isorhamnetin, effective components in QDD, also exhibits renal protective and antidiabetic properties [27][28][29][30].
The relationships between DKD, biological processes, and the enriched pathways have been reported previously, including the regulation of glucose and lipid metabolism, oxidative stress, and signaling transduction pathways [31][32][33][34][35]. Epithelial cells acquire the stromal cell phenotype is the EMT biological process. Under high glucose, renal tubular epithelial cells could transform into mesenchymal-like cells by advancing glycosylation products, chronic in ammation, oxidative stress, and undergoing a series of signaling cascades, to avoid apoptosis [36]. Therefore, EMT plays important role in the development of DKD [37] and is considered to be one of the initiating factors in the occurrence and development of renal TIF [38].
It has been reported that MAPK activation is related to the secretion of TGF-β1 and ECM protein [39]. P38MAPK is positively correlated with the expression of type 3 and type 4 collagen in type 2 diabetic nephropathy rat model [40]. In vivo studies, renal brosis can be improved by blocking MAPK signaling pathway [41]. In vitro studies shown that, under high glucose, p38MAPK activation promotes renal tubular epithelial cell EMT progress [42], and p38 inhibitor SB203580 signi cantly reverses the EMT progress by downregulated the expression of E-Cadherin and type 1 collagen in HK-2 cells [43,44]. mTOR is a serine/threonine kinase, which belongs to the family of phosphatidylinositol kinase-related kinases. MTOR responds to a variety of extracellular stimuli including growth factors, insulin, and nutrients. It regulates various physiological functions such as cell growth and cell cycle mainly through the PI3K/AKT/mTOR pathway [45,46]. Animal experiments and clinical studies have reported the roles of mTOR in diabetic nephropathy. The inhibition on mTORC1 alleviates the morphological changes and dysfunction of diabetic kidneys [47]. In addition, mTOR promotes interstitial brosis by enhancing the proliferation of broblasts, regulation of EMT [48]. In summary, AKT-mTOR plays a role in EMT induced by high glucose and DKD.

Conclusion
The study employs network pharmacology and in vivo and in vitro experiments to demonstrate that QDD delays the progression of DKD by regulating EMT via p38MAPK and AKT/mTOR signaling pathways. The datasets generated for this study are available on request to the corresponding author.