Network Pharmacology-Based Investigation into the Mechanisms of Qibangyishen Formula for the Treatment of Diabetic Kidney Disease

Background Diabetic kidney disease is the main microvascular complication of diabetes, and new treatment strategies are needed. We investigated the multi-component, multi-target, and multi-path mechanism of Qibangyishen formula for the treatment of diabetic kidney disease by network pharmacology methods. Methods We collected the active ingredients and targets of Qibangyishen formula and examined diabetic kidney disease-related genes by searching the ETCM, TCMID, and DisGeNet databases. We constructed and analyzed the genetic network through Cytoscape software and used the STRING platform for pathway enrichment analysis. Results We uncovered 421 active ingredients of Qibangyishen formula, corresponding to 748 targets. A network analysis screen revealed 47 hub-node targets, and 13 of these targets were diabetic kidney disease-related disease genes. Further functional analysis identied 26 targets acting on 11 pathways related to the development of diabetic kidney disease, including PI3K-Akt, thyroid hormone, neurotrophin, Wnt, chemokine, and osteoclast differentiation signaling pathways. Conclusions This study suggests that Qibangyishen formula regulates multiple genes and biological processes related to diabetic kidney disease and provides a foundation for further research and clinical applications.


Background
Diabetic kidney disease (DKD) is a chronic disease caused by diabetes and has a complicated pathogenesis [1][2][3]. The clinical features include continuously increasing excretion of albuminuria and/or glomerular ltration rate (GFR) less than 60 ml/(min·1.73m 2 ) for more than 3 months, which can eventually develop into end-stage renal disease (ESRD). Diabetic kidney disease leads to about 30-50% ESRD cases worldwide [4,5] and is the leading cause of ESRD in middle-aged and elderly people in China [6,7].
There are many clinical treatments for diabetic kidney disease [2,8,9], including life management [10,11], blood sugar control [12,13], blood pressure control [14,15], blood lipid regulation [16,17], uric acid control [18,19], and proteinuria regulation [20]. These treatments have various degrees of e cacy, but many patients with DKD continue to progress to dialysis. Traditional Chinese medicine has also been widely used to treat diabetic kidney disease. Chinese medicine may improve kidney and immune function and elicit therapeutic effects [21][22][23]. Diabetic kidney disease has no precise de nition in ancient Chinese medical literature, but according to etiology and pathogenesis, diabetic kidney disease can be attributed to "edema" and "guange" secondary to Xiaoke. Huang et al. [24] found that QufengTongluo Formula and YiqiHuayu Formula delay renal brosis and reduce urine protein in DKD rats, which may be related to enhanced PTEN expression, inhibition of PI3K/Akt signaling, and reduced transforming growth factor-β (TGF-β) expression. Hong et al. [25] showed that Jiangtang Formula reduces the level of advanced glycosylation products in KK-Ay mice and increases the phosphorylation of PI3K and Akt, which reduces in ammatory responses in DKD rats.
Previously, we found that HUANG QI combined with SAN QI reduces urinary protein excretion, decreases renal pathological changes, and improves apoptosis and shedding of kidney podocytes in DKD rats [33]. NIU BANG ZI extract reduces podocyte damage and proteinuria symptoms of DKD mice by enhancing PP2A activity in podocytes and protects kidney function [34].NIU BANG ZI extract also signi cantly reduces blood sugar in DKD mice and reduces the damage to the kidney [35]. Total glycosides is the main component of the BAI SHAO and reduces serum hs-CRP, TNF-α, and MCP-1 levels in patients with DKD, thereby protecting the kidneys [36]. Peony total glucosides prevent renal tubular interstitial damage in rats with diabetic nephropathy and delay the development of diabetic nephropathy [37]. HEI DOU reduces blood sugar, blood lipids, serum creatinine, and blood urea nitrogen in streptozotocin (STZ)-induced diabetic rats [38], and long-term low-dose HEI DOU protects diabetic renal insu ciency [39]. Although, the effects of individual components of QBYSF have been examined in diabetic murine models and in patients, the mechanism of QBYSF on diabetic kidney disease is not clear yet.
This study uses network pharmacology to screen the main active ingredients, targets, and pathways of QBYSF to analyze its mechanism on the treatment of diabetic kidney disease.

Materials And Methods
Collection of active ingredients and potential targets of QBYSF. ).Based on the pinyin name of each herb, we searched ETCM [40] and TCMID [41] databases to identify the active ingredients of QBYSF. In addition, we searched the ETCM database for the names of traditional Chinese medicines and compounds that are potential targets of the compounds contained in QBYSF.
Collection of targets for diabetic kidney disease.
Using "Diabetic Kidney Disease" as a keyword, we searched DisGeNet [42] to nd genes and targets related to diabetic kidney disease.
Constructing a sub-network affected by diabetic kidney disease.
In November 2019, we downloaded all high-quality binary protein-protein interactions from a high quality proteomics database HINT (High-quality INTeractomes) [43] and used Cytoscape [44] software to construct a genome-wide protein-protein interaction network in humans. This network is named the HINT network, which contains 12933 nodes and 60672 edges. We mapped all disease genes from diabetic kidney disease to the HINT network and extracted all nodes, neighbor nodes, and edges between all nodes related to diabetic kidney disease, but only retained the largest connected components to construct the diabetic kidney disease sub-network.
Constructing an association network between compound herbal medicines and important targets.
We used Cytoscape software to perform topology analysis on the diabetic kidney disease sub-network. We calculated the degree for each node and kept all nodes with a degree greater than twice the average degree for the network's hub nodes. The targets corresponding to the hub nodes were de ned as important targets of QBYSF for the treatment of diabetic kidney disease. Cytoscape software was used to construct an association network between targets and corresponding herbs.

Pathway enrichment analysis
We entered the target gene ID into the STRING [45] platform for pathway enrichment analysis.

Results
Active ingredients and potential targets of QBYSF.
In the ETCM and TCMID databases, the herbs of QBYSF were searched, and 421 active ingredients of QBYSF were obtained (Supplementary Table 1), corresponding to 748 targets (Supplementary Table 2).
The number of compounds and targets of each herbal medicine is shown in Table 1, and the intersection between herbal targets is shown in Fig. 1A. Genes related to diabetic kidney disease.
We searched the DisGeNet database to identify diabetic kidney disease-related target genes. We obtained 559 diabetic kidney disease-related genes (Supplementary Table 3), of which 66 disease genes were potentially targeted by QBYSF (Fig. 1B).
The diabetic kidney disease sub-network.
The diabetic kidney disease sub-network had 3467 nodes and 23661 edges and included genes related to diabetic kidney disease and neighboring genes in the human protein-protein interaction network. We mapped QBYSF targets to this network, identi ed 219 targets, and found 52 related to diabetic kidney disease (Fig. 2).

Identi cation of QBYSF targets.
We calculated the degree for each node in the diabetic kidney disease sub-network, and the average degree for the network was 13.6. The nodes with a degree greater than 27 were labeled hub nodes for the sub-network. There were 444 hub nodes in total, 47 were QBYSF targets, and 13 QBYSF targets were related to diabetic kidney disease.
The correlation network between the 47 QBYSF targets and QBYSF ingredient herbs is shown in Fig. 3A. In general, a natural compound's effect on a target is lower than that of a drug because a drug's structure is designed to increase its binding a nity to the target. Although a single natural compound may have small effect on a target in isolation, targeting the same protein by multiple compounds by multiple compounds may synergistically increase the effect on the target, thereby producing a therapeutic effect. We de ned a protein target score by the number of herb-compound pairs that act on a protein. There were 43 of 47 QBYSF targets with a score equal or greater than two, which suggests that QBYSF may regulate these targets.

Enrichment analysis of QBYSF targets
We examined 219 QBYSF targets including the 47 hub-node targets in the diabetic kidney disease subnetwork using the STRING platform for functional enrichment analysis. We used an FDR < 0.0001 to identify KEGG pathways signi cantly enriched in QBYSF targets and selected the top 20 signal pathways signi cantly regulated by QBYSF (Figure 4).
The signal pathways regulated by the two target sets include PI3K-Akt, thyroid hormone, neurotrophin, and osteoclast differentiation signal pathways ( Figure 4A, B). The intersection of the two sets in Figure   4A and 4B includes 11 signal pathways, which include 26 hub-node targets ( Figure 5).

Discussion
The main characteristics of Chinese medicine are a "holistic concept" and "symptom differentiation and treatment", and network pharmacology research methods are similar to this "holistic concept". Network pharmacology for Chinese medicine is an emerging and rapidly developing cross-disciplinary eld, and includes databases, network servers, and software to study TCM network pharmacology [46,47].
In this study, we used network pharmacology to study the mechanism of Qibangyishen formula for the treatment of diabetic kidney disease. Based on TCM databases, 421 active ingredients in QBYSF were identi ed, corresponding to 748 targets. There were 559 diabetic kidney disease-related genes obtained from the DisGeNet database. Network analysis revealed 47 QBYSF targets, and 13 of these were diabetic kidney disease-related disease genes. Pathway enrichment analysis found that 26 of the 47 targets were regulated in 11 pathways related to endocrine and renal functions, including PI3K-Akt signaling, thyroid hormone signaling, estrogen signaling, insulin signaling, osteoclast differentiation, and Wnt signaling.
The phosphatidylinositol 3'-kinase (PI3K)-protein kinase B (Akt) signaling pathway regulates the cell cycle, transcription, translation, proliferation, growth, and survival in response to extracellular signals such as toxins. Abnormal PI3K/AKT signaling can cause cancer [48][49][50], diabetes [51][52][53], and other diseases. PI3K-Akt signaling is closely related to the pathogenesis of DKD and is a common target in many traditional Chinese medicine compounds. Huang et al. [24] found that enhancing the expression of PTEN, inhibiting the PI3K/Akt signaling pathway, and reducing TGF-β expression can delay renal brosis and reduce urine protein in DKD rats. Hong et al. [25] showed that the Chinese medicine Jiangtang reduces the level of advanced glycosylation products in KK-Ay mice and increases the phosphorylation of PI3K and Akt, thereby reducing the in ammatory response of DKD. Increased activation of PI3K-Akt signaling can also aggravate Tubulointerstitial Fibrosis (TIF) [54].
The Wnt signaling pathway, highly conserved in various species, is a signal transduction pathway that includes proteins that pass signals through cell surface receptors, and is essential in developmental processes. Continuous activation of Wnt signaling induces broblast proliferation, which can lead to renal brosis, which can then be reversed by inhibiting Wnt signaling [55]. Wnt signaling triggers the pathogenesis of podocyte dysfunction and proteinuria [56][57][58]. Targeted inhibition of Wnt /β-catenin signaling transmission can preserve podocyte integrity and is used in animal models [59,60]

Conclusion
In summary, our network pharmacological study suggests that QBYSF protects the kidneys through 26 targets that regulate 11 pathways involved in the progression of diabetic kidney disease, including PI3K-Akt signaling, Wnt signaling, insulin signaling, and chemokine signaling, which delays the progression of diabetic kidney disease.

Authors' Contributions
Zhi Wang, Guihua Jian: conception and design of the study, and generation and collection of data. Wang Zhi: assembly, analysis, and interpretation of data, drafting and revision of the manuscript. All authors approved of the nal version of the manuscript. Figure 1 QBYSF target analysis. (A) The intersection between potential targets of the six herbs in QBYSF. (B) The intersection between potential QBYSF targets and diabetic kidney disease genes.