Discover the key genes for glomerular inflammation in patients with type II diabetic nephropathy based on bioinformatics and network pharmacology

LIU Yi-bu, WEN Min, PENG jin-gang, FAN Ju-di?

1. School of Pharmacy, Guizhou Medical University, Guiyang 550004, China

2. Engineering Technology Research Center of Chemical Synthetic Drug Development and Utilization in Guizhou Province, Guiyang 550004, China

Keywords:

ABSTRACT

1.Introduction

Diabetic Nephropathy (DN) is the most common and serious complication of diabetes mellitus (DM) and the most common cause of end-stage renal disease worldwide[1].Till 2020, the number of patients with diabetes complicated with chronic kidney disease in China reached 24.3 million[1,2].However, the pathogenesis of DN is complex, without clear diagnostic indicator[3-4].Clinical screening for DN is primarily directed at urine albumin, which is not sensitive[3-7].Therefore, the development of DN diagnostic indicators has become the key and difficult problem to guide clinical treatment.At the same time, the pathogenesis of DN is very complex and has not been systematically explained.Understanding the pathogenesis of DN and developing strategies for the treatment of DN and identifying relevant diagnostic indicators can help control its occurrence and progression.Objective:Network pharmacology and bioinformatics were used to analyze the key genes of kidney damage in diabetic nephropathy (DN) patients and explore the effect of expression changes of key genes.Methods:The gene chip dataset GSE96804 of glomerular transcriptome samples and healthy human samples with type II diabetes mellitus was retrieved and obtained from the GEO database.Differentially expressed genes (DEGs) were obtained by differential analysis of standardized genes by R language, and key genes and pathways related to glomerular inflammation in patients with type 2 diabetic nephropathy were obtained through ontological (GO) enrichment and KEGG pathway enrichment analysis through GSEA annotation.The crosstalk between two results was analyzed by protein-protein interaction network (PPI).The human transcript GSE30122 was used for verification, and the crossover differential genes were sorted by Cytoscape to obtain the top ten key genes as Hub genes, and the Nephroseq database was used to explore their impact on patients.Results:Through the analysis of 1235 DGEs based on the limma package of R language, the GSEA-KEGG pathway enrichment found that glomerular inflammation in patients with type 2 diabetic nephropathy mainly affects a variety of biological processes including exogenous stimulation, changes in lipid levels, activation of immune system regulation, cell chemotaxis and differentiation,and kidney development, and DGEs are enriched in transcription factor pathways, tolllike receptor pathways, adipocycyte cytokine signaling pathways, cell adhesion pathways,and cytochemotokine pathways.It is judged to be highly related to diabetic nephropathy inflammation.By aligning with Neqhroseq, 5 key genes with different expression in the glomeruli were obtained.Conclusion:Transcription factor pathway, toll-like receptor pathway,chemokine signaling, cell adhesion pathway, and adipocytokine pathway are the key pathways for inducing inflammation in nephropathy in patients with type 2 diabetes, and CD8A, PTPRC,TCR2, CCL5, and ITGAM may be potential biomarkers for DN diagnosis.

At the same time, DN is accompanied by damage to the glomeruli and renal tubules, as well as inflammatory responses[7].Up to now, it has been reported that three main ways for DN will cause damage to glomerular endothelial cells: targeting the polyol pathway and proteinase C to affect intracellular signaling and metabolism;Glomerular cell dysfunction and macrophage activation induced by synthetic advanced glycosylation end products (AGE) and oxidative stress; Glomerular ultrafiltration and hypertension induced by high glucose and other causes[8-9].Damage to glomerular endothelial cells is accompanied by enrichment and infiltration of inflammatory cells(monocytes, macrophages, lymphocytes), resulting in proteinuria[10].While low-grade glomerular inflammation is thought to be a common pathway for progression of proteinuria glomerulopathy,the use of anti-inflammatory drugs is thought to be beneficial for the control and treatment of the DN course[11].Research and identification of key factors in glomerular inflammation can help discover new targets and develop new strategies for the development and control of diabetic nephropathy inflammation.

Since mid-90s of the 20th century, the completion of the Human Genome Project has marked the beginning of our biologically driven era[12].With the advancement of powerful methods in molecular,structural, and chemical biology, such as genome sequencing, microarray gene expression analysis, RNA interference (RNAi), highthroughput crystallization, etc., methods contain massive amounts of genetic information[13].GEO[14], GeneBank[15], RefSeq[16],UniProt[17] databases, etc.can help us quickly find disease-related sequencing data and protein sequence information, quickly locate the pathway and interaction network of disease occurrence and development, which will accelerate the clinical and pathological research of diseases.

In this paper, based on the mechanism of inflammatory factors affecting the course of DN in previous literature, we focused on the specific clinical feature of induced inflammatory response in patients with diabetic nephropathy.The two sets of glomerular sequencing chip data GSE96804 and GSE30122 in the GEO database were analyzed with differential genes (DGEs) intersected.GO function annotation and GSEA-KEGG pathway enrichment were carried out to identify key pro-inflammatory cytokines and corresponding signaling pathways that would cause glomerular lesions, and intervene in the diagnosis of early diabetic nephropathy.Through the PPI network, the key genes of pathogenesis were locked, and the key diagnostic genes of DN in clinical diagnosis were discussed together with the public database Nephroseq.The aim of our research is to provide guidance and inspiration for precise intervention in early clinical diagnosis and treatment.

2.Materials and methods

2.1 Data Processing and Quality Control

Searching the GEO database (http://www.ncbi.nlm.nih.gov/geo) by the keyword “diabetes kidney glomeruli”, select human source data(Homo sapiens), and select the sequencing method as “Expression profilling by array”.In the obtained data, whether the sequencing site is glomeruli, and the dataset containing only the diseased group or the control group was excluded, and finally the datasets GSE96804 and GSE30122 were screened for analysis.Basic information is shown in the table below (Table 1).

Download the two sets of data mapping matrices and platform files to analyze the quality and difference of the data using the GEO2R platform (https://www.ncbi.nlm.nih.gov/geo/geo2r/).At the same time, based on the platform annotation file, the probe ID in the matrix is marked, which is converted into “ENTREZ_ID”, and then converted into gene name for subsequent analysis.

Tab 1 Dataset information

2.2 Filtering and functional notes of DEGs

The gene expression profiles were analyzed differently by the R-based limma package, and the DEGs were screened with|log2FC|>0.379(FCs>1.3) and P<0.05 as the screening criteria,and R-based ggplot2 and pheatmap were used for visualization,and volcano maps and heat maps were plotted, respectively.At the same time, the obtained differential genes were enriched by gene ontology (GO) and genome encyclopedia (KEGG), andP<0.05 was selected as the screening criteria for screening statistically significant biological process (BP), cellular component (CC) and molecular function (MF).

At the same time, the differential genes that may be filtered by the screening threshold were re-enriched using gene set enrichment analysis (GSEA, software version: 4.3.2), and the KEGG terminology (c2.cp.kegg.v2023.1) and biological processes(c5.go.bp.v2023.1) were annotated using Molecular Signature Datebase (2023) to annotate and analyzeThe gene sets of NES|>1,NOMp<0.05 and FDR<0.25 were considered to be significantly different and enriched.

2.3 Protein-protein interaction (PPI) and screening of core genes

The online website STRING(https://www.cn.string-db.org/) types DEGs to predict the interaction between proteins, imports the network diagram constructed by the intersection of differential genes(up-regulated and down-regulated genes) into Cytoscape, and uses the Cytohuba plug-in to display and screen out the top ten DEGs as Hub genes in the shortest path for the entire network.

2.4 Analysis of DN clinical features of glomerular injury

Based on the public clinical database Nephroseq library (https://www.nephroseq.org), to verify whether the core genes after the intersection of the two sets of datasets have expression differences between the normal group and the model group.At the same time,based on the effects of Nephroseq on the clinical characteristics of Hub gene: glomerular filtration rate (GFR), 24 h proteinuria, and serum creatine, this paper was compared and discussed for screening Hub gene with biomarker potential.

3.Results

3.1 Data quality analysis and identification of DEGs

The database data GSE96804 and GSE30122 meet the quality control requirements of chip sequencing data, and can be well distinguished between the model group and the control group,as shown in Figure 1 below.GSE96804 obtained a total of 1235 DEGs after limma screening, including 417 upregulated genes and 818 downregulated genes.GSE30122 screened 1706 DEGS1706,including 852 up-regulated genes and 854 down-regulated genes.Volcano maps and heat maps plotted by ggplot2 and pheatmap show the clustering of DGES and the differences between different samples, as shown in Figure 2.

Fig 1 GSE96804 differential gene screening

3.2 Analysis of enrichment results

The two groups of differential genes were analyzed by GO enrichment, and the results were shown in Figure 3.GO enrichment results cover the main biological processes of diabetic nephropathy:exogenous stimuli, changes in lipid levels, activation of the regulatory capacity of the immune system, cell chemotaxis and differentiation, and abnormalities in the development and metabolic processes of the kidneys[10-11].Molecules related to inflammatory factors mainly include conversion factors, pro-inflammatory factors,and related adhesion factors, chemokines, toll-like receptors,adipokines, nuclear receptors, etc.[3,18].Further use of GSEA enrichment genes and pathways, the results are shown in Figure 5.The gene enrichment results of six pathways, including transcription factors pathway, toll-like receptor pathway, adipocytokine signaling pathway, adherens pathway, and chemokine pathway, were collected.The two enriched DGEs were intersected, and a total of 174 DGEs associated with inflammatory factors were obtained.Fig.4 GO enrichment analysis results

3.3 Screening and verification of core genes

Using the String database to build a PPI network, 174 DGEs were input into Cytoscape for screening, and the top 10 DGEs were selected as the Hub gene, and the results were shown in the figure below (Fig.5).

Ten Hub genes were entered into the Nephroseq V5 database for retrieval to check whether they were genes with different expression on the clinical sample, and the results were shown in Figure 6.The analysis deleted the geneCD80, which had no expression difference between the normal group and the model group, whileTNF,CD4,CD80, andCD86could not query the sample of type 2 diabetes,and the gene CD40LG could not query the sample of the glomerular part, so it was not considered.Here, a total of 5 Hub genes with expression differences were screened: CD8A, PTPRC, TLR2, CCL5,ITGAM.The difference in the expression of the five genes in the model group (group 2) and the normal group (group 1) in the Neohroseq dataset is shown in Figure 7.All five genes were upregulated in the model group, while the remaining four genes were less expressed in the normal group except ITGAM.

Fig 2 GSE96804 differential gene screening

Fig 4 GSEA-KEGG enrichment analysis

Fig 5 Top 10 key genes associated with inflammatory pathways

3.4 Risk assessment of key genes

By searching the Nephroseq V5 database, the correlation between the Hub genes: CD8A, ITGAM, PTPRC, CCL5, TLR2 and functional differences such as proteinuria, serum creatintine, and glomerular filtration rate was compared.In this regard, the kidney samples of the normal group and the model group were retrieved for differential analysis, and the results are shown in Figure 8～10.Figure 8 shows the correlation between the expression of the four Hub genes and the glomerular filtration rate (ml/min/1.73 m2).High expression of CD8A, PTPRC, ITGAM, TLR2 WILL LEAD TO LOW GFR, CAUSING KIDNEY DAMAGE.Figure 9 shows the correlation between the expression of the three Hub genes and proteinuria (g/24 h).Proteinuria will improve under conditions of high expression of CD8A and ITGAM, while ITGAM gene relies on lower levels of expression.Figure 10 shows the correlation between Hub gene expression and serum creatine (mg/dL).As shown in the figure, in addition to PTPRC, the upregulation of ITGAM,CCL5, and TLR2 will cause an increase in serum sarcosine levels.Therefore, the five Hub genes are indeed related to inflammation in kidney disease and can be used as biomarkers with clinical potential.Fig.10

Fig 6 Analysis of Hub’s gene for differences in expression in the Nephroseq database

Fig 7 Analysis of differences in the expression of Hub gene between groups in the Nephroseq

Fig 8 Comparison of Hubs gene expression and GFR correlation

Fig 9 Comparison of Hubs gene expression with proteinuria

Fig 10 Comparison of Hubs gene expression with serum sarcosine

4.Discussion

The causes of diabetic nephropathy are complex and closely related to kidney dysfunction and changes in some kidney cells.Several studies in diabetic nephropathy had confirmed that inflammationrelated factors and signaling pathways influence the occurrence and progression of diabetic nephropathy[18], and the identification of inflammatory factors help to diagnose and formulate new treatment options.

Based on the bioinformatics method, this study analyzed and compared two human glomerular cell datasets, explored the key genes that cause glomerular inflammation, and compared the glomerular tissue of DN patients’ and healthy people’s glomerular through the intersection analysis of the two biochip datasets from GEO database, and finally intersected 174 inflammation-related DGEs through GO annotation and pathway annotation by GSEA.At the same time, the Nephroseq dataset verified and their performance on key clinical features were proved.The five potential clinical biomarkers: CD8A, ITGAM, PTPRC, CCL5, TLR2 ,were dentified as having potential to provide a new method for the accurate diagnosis of glomerular inflammation caused by nephropathy.The ITGAM pathway is associated with macrophages[19], and the polarization of macrophages and the infiltration of the kidneys will affect the development of DN inflammation[20].CCL5 is a key class of chemokines, and the effects of chemokines and adhesion factors will enrich monocytes and lymphocytes, causing DN[21].Tyrosine protein phosphatase C (PTPYC) is a key gene in regulating intracellular signaling metabolism[22], and is also an important regulator of T cells and B cells.Its differences in expression are related to immunomodulatory and inflammatory factors.TLR2 is a toll-like receptor 2, and its downstream signaling molecules NFκB is an important transcription factor that induces proteinuria and induces the release of inflammatory factors[23].CD8A is an antigen encoded by the CD8 strand, which recognizes the antigen and activates T cells by inducing leukocyte differentiation, the helper T cell receptor (TCR) recognizes the antigen and activates T cells[24-25].Chikako Shimokawa et al.found that CD8 cell activation of T cells in mice will effectively prevent diabetes mellitus in mice induced by streptozotocin[26].At the same time, by comparing with Nephroseq, the performance of DN patients and normal groups on the above five genes was different, and there were significant differences in functions including proteinuria, serum sarcosine,glomerular filtration rate, etc., which had clinical diagnostic potential.

In summary, this study analyzed the differential genes of glomerular inflammation in patients with diabetic nephropathy through bioinformatics, and the results were enriched to 174 significantly different genes, as well as important biological processes including exogenous stimulation, changes in lipid levels, activation of immune system regulation, cell chemotaxis and differentiation,and abnormalities in kidney development and metabolic processes.At the same time, through the screening and comparison of inflammatory pathways, five key genes with biomarker potential were found, from the release of inflammatory factors to the process of immunoregulation and the downstream induction of proteinuria:CD8A, ITGAM, PTPRC, CCL5, TLR2.Since this study had based on different batches of human kidney samples in the public dataset, the course of the disease and the details of the study site of the samples were unknown, and the sample size would be increased to further verify the expression differences of core genes as biomarkers and the survival curve of patients in the later stage of wet experiments through animal models validation and qPCR , to provide a new diagnostic scheme for further optimization of DNinduced nephropathy.

Authors’ Contribution

This paper was supported by the National Science Foundation of Guizhou Provincial Health and Family Planning Commission, and designed to support follow-up research as a background.The first author of the article is in the first place, Peng Jingang is the leader of this project.The corresponding author, Fan Judi, is not a member of the project team, but greatly contributed to the idea guidance and framework design of this article under project’s requirements.All authors have reviewed the contents of the manuscript and there is no conflict of interest.