Identification of Oxidative Stress-Related Biomarkers in Diabetic Kidney Disease

Background Diabetic kidney disease (DKD) is a leading cause of end-stage renal disease throughout the world. In kidney disease, oxidative stress has been linked to both antioxidant depletions and increased reactive oxygen species (ROS) production. Thus, the objective of this study was to identify biomarkers related to oxidative stress in DKD. Methods The gene expression profile of the DKD was extracted from the Gene Expression Omnibus (GEO) database. The identification of the differentially expressed genes (DEGs) was performed using the “limma” R package, and weighted gene coexpression network analysis (WGCNA) was used to find the gene modules that were most related to DKD. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed using “Org.Hs.eg.db” R package. The protein-protein interaction (PPI) network was constructed using the STRING database. The hub genes were identified by the Molecular Complex Detection (MCODE) plug-in of Cytoscape software. The diagnostic capacity of hub genes was verified using the receiver operating characteristic (ROC) curve. Correlations between diagnostic genes were analyzed using the “corrplot” package. In addition, the miRNA gene transcription factor (TF) network was used to explain the regulatory mechanism of hub genes in DKD. Results DEGs analysis and WGCNA-identified 160 key genes were identified in DKD patients. Among them, nine oxidative stress-related genes were identified as candidate hub genes for DKD. Using the PPI network, five hub genes, NR4A2, DUSP1, FOS, JUN, and PTGS2, were subsequently identified. All the hub genes were downregulated in DKD and had a high diagnostic value of DKD. The regulatory mechanism of hub genes was analyzed from the miRNA gene-TF network. Conclusion Our study identified NR4A2, DUSP1, FOS, JUN, and PTGS2 as hub genes of DKD. These genes may serve as potential therapeutic targets for DKD patients.


Introduction
Diabetic kidney disease (DKD), one of the most prevalent and severe chronic microvascular complications of diabetes, is the leading cause of end-stage renal disease [1]. Approximately 40% of diabetes mellitus develop DKD [2]. Although the established control protocols of blood glucose and blood pressure have reduced burden of DKD, the incidence of DKD remained increasing [3]. Terefore, prevention of DKD is the major public health challenge. Te novel strategies and techniques for the early diagnosis and risk prediction are required.
Oxidative stress has long been recognized as a crucial factor in the pathogenesis of DKD [1,2]. Hyperglycaemia increases free radical production leading to oxidative stress which is an imbalance between oxidant and antioxidant levels [4,5]. Oxidative stress stimulates signaling molecules such as nuclear factor-κB (NF-κB) or activator protein-1 (AP-1) [6], ultimately leading to infammation, fbrosis, and apoptotic in renal cells [7,8]. Tus, the key to preventing DKD progression is to eliminate oxidative stress-induced cell damage in the early stages of DKD [9]. Te reliable oxidative stress biomarkers suitable for clinical practice associated with DKD are very important. Tis study aimed to identify potential biomarkers of oxidative stress by bioinformatic approaches to develop new strategies for early prevention and treatment of DKD in clinical research.

Data
Source. GEO is a database containing highthroughput gene expression data, chips, and microarrays. Te two DKD RNA-sequencing datasets, GSE142025 and GSE30528, were downloaded from the GEO database. GSE142025 dataset included 27 kidney biopsy samples from patients with DKD and nine nephrectomy samples from the normal kidney tissue. GSE30528 dataset had nine diabetic kidney disease glomeruli and 13 control glomeruli tissue samples. Te GSE142025 dataset was used for the training, and the GSE30528 dataset was used for validation. Te GO term "response to oxidative stress (GO: 0006979)" was used to identify genes related to oxidative stress.

Identifcation of DEGs.
Te expression matrices were divided into disease and control groups and were screened for DEGs. Identifcation of DEGs between DKD tissue samples and normal tissue samples was performed by using the "limma" R package. Te cutof criteria were as follows: | log 2 fold change (FC)| > 1 and adj. p value <0.05. DEGs meeting the criterion were selected for further analysis.

WGCNA for
Building DKD-Associated Modules. Te R package "WGCNA" [10] was used to identify DKD-related modules and genes. Hierarchical clustering of the gene expression in the samples was performed using the Euclidean distance to detect and eliminate outliers. Genes with similar expression patterns were assigned to a coexpression module. Te pickSoftTreshold function of WGCNA was used to determine the best soft threshold power (β) based on values from 1 to 20.

Gene Set Enrichment Analysis (GSEA) of the Key DKD Genes.
Te key DKD genes were obtained by taking the intersection of DEGs and module genes. GSEA is a computational method for evaluating whether a set of a prioridefned show statistically signifcant and consistent diferences between two biological states. Te GSEA analysis was performed on the key genes of DKD. For GSEA analysis, "h.all.v7.5.1.symbols.gmt" was downloaded from the Molecular Signature Database (MSigDB) as the background gene set.

Functional and Pathway Enrichment Analyses of Candidate Hub Genes.
Te key DKD genes were annotated using the GO term "response to oxidative stress" to get the candidate hub genes. GO term analysis and KEGG pathway analysis were performed using the R package "Org.Hs.eg.db" [11] to identify the functional roles of the candidate hub genes. GO terms or KEGG pathways with adj. p value <0.05 were considered statistically signifcant.

Construction of the PPI Network and Identifcation of Hub
Genes. We constructed a PPI network of candidate hub genes using the STRING database [12] to further screen meaningful hub genes. Te hub genes were identifed using MCODE [13], a Cytoscape plug-in.

Diagnostic Performance Evaluation of Hub Genes by ROC
Analysis. Te ROC curve was plotted, and AUC was calculated using the "pROC" R package [14] to evaluate the capability of selected hub genes and to distinguish DN patients and controls. Spearman correlation coefcients between genes were visualized using the "corrplot" R package [15].

Construction of the Regulatory Network.
MiRNAs can reduce gene expression by binding mRNAs and thereby inducing gene silencing. Transcription factors (TFs), also known as trans-acting factors, are DNA-binding proteins that can activate or inhibit gene transcription by interacting with cis-acting elements of eukaryotic genes. Te miRNet database, a network-based visual analysis tool [16], was used to predict miRNAs and TFs that interacted with the hub genes. Te miRNA gene-TF network was constructed using Cytoscape software.

GEO and Nephroseq v5
Validation. Te expression levels of hub genes associated with DKD were compared with controls using the GEO (GSE30528) and Nephroseq v5 databases in order to verify the robustness of our results in the external dataset.

Identifcation of DEGs between DKD and Control Groups.
We compared DKD and normal tissues in the GSE142025 using the "limma" R package and identifed genes with differential expression in this dataset. Te data were fltered using |log 2 FC| > 1 and adj. p value <0.05. A total of 1,121 DEGs (DKD vs. control) were identifed, including 641 upregulated genes and 480 downregulated genes (Figure 1(a)). Te top 10 upregulated genes and top 10 downregulated genes (sorted by adj. p value) are shown in Figure 1(b).

Discovery of the DKD-Related Modules and Genes.
To identify genes associated with DKD, we performed WGCNA. First, the hierarchical clustering of the samples showed that there were no outliers (Figures 2(a) and 2(b)). Ten, β � 6 was selected as the appropriate soft-thresholding power to ensure a scale-free analysis (Figure 2(c)). Highly related genes were divided into the same module, and a total of 20 modules were identifed (Figure 2(d)). A heatmap was generated to show the global outline of the relationship between the modules and DKD (Figure 2(e)). According to the correlation results, a strong positive correlation was observed between the light cyan module (Cor � 0.75, p value � 1E − 06), tan module (Cor � 0.69, p value � 2E − 05), yellow module (Cor � 0.62, p value � 2E − 04), and DKD, and a strong negative correlation between purple module (Cor � −0.85, p value � 1E − 09), red (Cor � −0.79, p value � 1E − 07) module, and DKD. In addition, 3,238 genes within these coexpression modules were associated with DKD.

Screening and Analysis of Key Genes and Candidate Hub
Genes of DKD. Te selected DEGs and module genes were intersected, and a total of 160 overlapping genes were screened out as the key genes for DKD (Figure 3(a)). Ten, the GSEA analysis was performed on these 160 key genes using the hallmark gene sets in the MSigDB database ( Figure 3(b)). Te results showed that two pathways were signifcantly enriched, namely "HALLMARK_HYPOXIA" and "HALL-MARK_TNFA_SIGNALING_VIA_NFKB." Ten, a total of 9 out of 160 genes were annotated in the GO term "response to oxidative stress (GO: 0006979)," which were NR4A2, DUSP1, NR4A3, FOS, JUN, PTGS2, GSTP1, NOL3, and RPS3. Tese genes were identifed as candidate hub genes for DKD. To explore the potential function of these candidate hub genes, we performed GO and KEGG enrichment analysis using the R package "Org.Hs.eg.db." Figures 3(c)-3(e) show the candidate hub genes were mainly involved in "response to oxidative stress," "cellular response to chemical stress," and "cellular response to oxidative stress" for the aspects of biological process (BP), "transcription regulator complex" for the aspects of cellular component (CC), and "DNA-binding transcription activator activity," "DNA-binding transcription activator activity, RNA polymerase II-specifc," and "DNAbinding transcription factor binding" for the aspects of molecular function (MF). KEGG pathway analysis indicated the candidate hub genes were mainly enriched in "fuid shear stress and atherosclerosis," "leishmaniasis," and "IL-17 signaling pathway" (Figure 3(f)).

Identifcation of Hub Genes by the PPI Network.
To better understand the interaction among the identifed nine candidate hub genes, we used the STRING online server to construct a PPI network (Figure 4(a)). Te entire PPI network was analyzed using MCODE, and the genes were selected as hub genes in the core module ( Figure 4(b)). Te fve genes, NR4A2, DUSP1, FOS, JUN, and PTGS2, were in the core module and were selected as hub genes. Te expression of all hub genes was downregulated in DKD patients ( Figure 4(c)).

Evaluation of the Diagnostic Efect of the Hub Gene on DKD.
In this study, the ROC curve was plotted to evaluate the diagnostic ability of DKD for fve hub genes and calculated the corresponding AUC values ( Figure 5(a)). Te results show that NR4A2 (AUC � 1), DUSP1 (AUC � 0.996), FOS (AUC � 0.971), JUN (AUC � 0.971), and PTGS2 (AUC � 0.959) can be utilized as distinguishing feature parameter for DKD. Te correlations between fve hub genes were then analyzed ( Figure 5(b)). NR4A2, DUSP1, FOS, JUN, and PTGS2 were positively correlated with each other. Te correlation between DUSP1 and FOS was the highest at 0.96.

Expression Validation of Hub
Genes. Te expressions of the hub genes were further verifed in the validation GSE30528 dataset (Figure 7(a)). Te results showed that the expression Up Down Evidence-Based Complementary and Alternative Medicine 3 pattern of the fve hub genes in the validation set was comparable to the training set. In the validation set, however, only a signifcant diference in the DUSP1 expression was observed between DKD and control samples. We further examined the expressions of the hub genes in DKD and control samples using the online tool Nephroseq V5 (Figure 7(b)). Te result showed that hub genes NR4A2, DUSP1, FOS, and JUN had the same expression patterns as the training set.
Sample clustering to detect outliers  A26A  A24A  A19A  A20A  N13B  N6B  N10B  N5B  N7B  N2A  N3A  N8B  N12B  A29A  B6B  B8B  B3B  B5B  B4B  B9B  A33A  A17A  A30A  A13A  A21A  A25A  A23A  A32A  A11A   A15A  A28A  A12A  A14A  A22A  60   80    regulation of vascular associated smooth muscle cell proliferation vascular associated smooth muscle cell proliferation response to cAMP response to organophosphorus response to glucocorticoid response to purine−containing compound response to corticosteroid response to hydrogen peroxide cellular response to reactive oxygen species cellular response to extracellular stimulus regulation of neuron death cellular response to external stimulus response to steroid hormone neuron death regulation of apoptotic signaling pathway response to extracellular stimulus response to reactive oxygen species cellular response to oxidative stress cellular response to chemical stress response to oxidative stress

Discussion
Oxidative stress, as an important factor in the development and progression of DKD, has been a research hotspot. Te aim of this study was to investigate disease evaluation and diagnosis markers.
Te nine screened candidate genes for oxidative stress were subjected to GO and KEGG enrichment analyses based on data acquisition from online databases. On the basis of response to oxidative stress, cellular response to chemical stress, cellular response to oxidative stress, and various other physiological processes, these genes could form a transcription regulator complex with other proteins to regulate the gene expression. Subsequently, this complex will help in regulating molecular functions such as DNA-binding transcription activator activity, RNA polymerase IIspecifc, and DNA-binding transcription factor binding. In this way, these genes would further regulate fuid shear stress and atherosclerosis signaling pathway, leishmaniosis signaling pathway, IL-17 signaling pathway, TNF signaling pathway, and many others. Consequently, these genes were involved in oxidative stress, infammatory response, and cell apoptosis. Among them, the IL-17 signaling pathway and TNF signaling pathway are closely associated with the occurrence and development of diabetic nephropathy. IL-17 is a key cytokine secreted by CD4 + T cells [17]. In the high glucose state, the accumulation of AGEs increases the expression of IL-17 [18,19], and after binding to its receptor, it    Evidence-Based Complementary and Alternative Medicine promotes IL-6, monocyte chemotactic protein (MCP-1), and regulates activation of monocytes. Te production of the transforming growth factor TGF-β1 and the activated B cell nuclear factor kappa light chain enhancer (NF-κB) mediate the infammatory response associated with the tissue damage in DKD and glomerulosclerosis. However, it is important to note that other studies have come to the opposite conclusion. In a clinical study, IL-17 expression decreased in DKD patients with disease progression [20]. In addition, in the STZ-treated mouse model of IL-17 knockout, it was found that IL-17A knockout mice aggravated renal infammation and fbrosis by inhibiting autophagy [21]. Although the specifc role of IL-17 in diabetic nephropathy is still unknown, it is clear that IL-17 has a signifcant role in the onset of diabetic nephropathy. Te TNF family is involved in diferent states of DKD by regulating immune function. In animal studies and clinical trials, a high glucose environment can induce oxidative stress, resulting in an increase in the expression of TNF-α, which can aggravate DKD injury [22][23][24]. Its mechanism is related to the induction of podocyte retinoic acid receptor responder protein 1 (RARRES1) expression to activate podocyte apoptosis [25], inhibition of NRF2/KEAP1/ARE pathway to reduce antioxidant capacity [26], to promote NF-κB release [27] and other infammatory factors release, and drive Egr-1 activation [28] to regulate cell diferentiation and growth. Osteoprotegerin (OPG) may be among the most studied TNF family. OPG expression was found to be upregulated in the kidneys of diabetic patients and correlated with the severity of DKD [29]. Tese results show that the hub gene screened by our study is closely associated with the occurrence of DKD and also suggests a potential pathway, providing a new theoretical foundation for researchers into the mechanism of DKD. For further exploration of the key genes afecting DKD, fve hub genes (NR4A2, DUSP1, FOS, JUN, and PTGS2) were screened by the PPI network, all of which were underexpressed in DKD patients. Gene expression of NR4A2, DUSP1, FOS, JUN, and PTGS2 was negatively correlated with DKD. In addition, all these fve hub genes showed strong diagnostic ability based on their diagnostic performance by the ROC curve.
NR4A2 is a member of the superfamily of steroidthyroid hormone-retinoid receptors. Under conditions of cellular response to external stimuli, the genes of this family would rapidly produce corresponding proteins, which are transcription factors to regulate gene expression [30]. Cumulative evidence has revealed the association of NR4A2 with oxidative stress. For instance, Popichak KA reported that NR4A2 could inhibit the infammatory response through NF-κB [31,32]. Kaoru found that NR4A2 could suppress LPS-stimulated monocytes/macrophages to induce TNF mRNA expression in a non-NF-κB manner, which could protect islet cells by alleviating endoplasmic reticulum stress in islet β cells [33]. Additional in vivo experiments showed that NR4A2 was involved in stress response [34].
In addition, DUSP1 could be considered an inhibitor of MAPK activity because it dephosphorylates MAPK at both tyrosine and threonine residues intracellularly and negatively regulates the production of proinfammatory factors  Evidence-Based Complementary and Alternative Medicine [35]. Ikuyo discovered in his research that the transcription of the DUSP1 gene was inhibited under oxidative stress [36]. Leblanc et al. [34] previously reported an increased number of activated P38MAPK in the glomeruli of DKD, which can be explained by the downregulation of DUSP1 and is consistent with our results. In this regard, inducing the expression of DUSP1 could serve as a potential treatment for DKD.
AP-1 complex, composed of FOS and JUN, is an important transcription factor in the nucleus [37]. It has a regulatory role in biological processes such as cell proliferation, diferentiation, and apoptosis [38]. Existing studies support its involvement in the occurrence and development of various diseases, such as colorectal cancer [39] and pancreatic cancer [40]. Under the high glucose environment, the production of ROS can induce increased FOS and JUN protein expression in the renal tissue, leading to the activation of AP-1 to initiate the transcription of its downstream genes (TGF-1, FN, and Laminin B). It may also lead to the accumulation of extracellular matrix and the thickening of the glomerular basement membrane in the renal tissue, thereby promoting the occurrence and development of DKD [37,41,42]. At the same time, there may be a change in the expression of FOS. Duan found an increased level of FOS in the proximal renal tubules in the event of acute renal injury, which was, however, highly unstable and degraded rapidly [43]. Te existence of TRE inhibits its transcription on the promoter of the FOS gene; however, high glucose may cause the increase of FOS protein and promote the binding of AP-1 to TRE (specifc TPA-responsive element), resulting in feedback inhibition of its own expression.
PTGS2 plays an essential role in the process of oxidative stress [44], and COX-2, encoded by PTGS2, is a key regulator of renal hemodynamics and promoter of infammation, which also has a role in podocyte injury. It has been documented that the expression of PTGS2 in podocytes increased signifcantly under high glucose conditions [45]. Luo et al. reported an increase in the expression of COX-2 in insulin-dependent diabetes mellitus. In addition, Jia et al. revealed that the expression of COX-2 was increased in the renal tissue when there was renal injury [46]. While these fndings are inconsistent with the decrease of PTGS2 gene expression to some extent in DKD patients observed in our study, however, further experimental validation is required.
In conclusion, our study suggests a close relationship between oxidative stress and DKD based on function and expression analysis of the screened oxidative stress-related genes in DKD. Oxidative stress-related genes may be considered novel biomarkers for the diagnosis and prognosis of DKD progression. Our study's fndings may serve as a basis for future experimental confrmation, as well as research into disease pathogenesis and clinical treatment. However, our research also has some limitations. For example, we need more samples from multiple centers to verify the role of these hub genes in DKD. In addition, we need to further fnd the regulatory pathways of these hub genes in DKD through sequencing and other means, which is of great signifcance for a deeper understanding of the mechanism of hub genes in DKD.

Data Availability
Te datasets for this study are available in the GEO database. GEO belongs to public databases. Users can download relevant data for free for research and publish relevant articles.

Ethical Approval
Te patients involved in the database have obtained ethical approval. Tis study is based on open-source data, so there are no ethical issues.

Consent
Not applicable.

Conflicts of Interest
Te authors declare that they have no conficts of interest.