Identification of Hub Genes Correlated with the Initiation and Development in Chronic Kidney Disease via Bioinformatics Analysis

Chronic kidney disease (CKD) is a global public health issue with rising incidence and prevalence rates, that inevitably progresses to end-stage renal disease when faced with excessive loss of nephrons and gradual decline of renal function [1, 2]. The characters of the initiation and progression of CKD are inflammation, oxidative stress and fibrosis [3, 4]. Although prominent events triggered the initiation and progression of CKD have been studied widely, it is still urgent to identify the effective biomarkers or gene targets for suppressing the development of CKD.

Renal inflammation, including pro-inflammatory milieu and immune dysfunction, has been considered an important determinant factor throughout initiation and progression of CKD, regardless of initiating causes [5]. Once stimulated by various pathogenic components, renal parenchymal cells and immune cells secrete various inflammatory mediators, such as chemokines and cytokines, which in turn recruit more immune cells and lead to vicious cycle and fibrosis development [6, 8]. Although numerous studies have demonstrated that chemokines and immune response are tightly associated with CKD, the roles of chemokines and immune cells still need to be further explored.

Chemokines, belonging to chemotactic cytokines, are classified into 4 subfamilies, including CCL, CXCL, CX3CL, and XCL. Chemokines and their receptors are tightly associated with the development of CKD via recruiting multiple inflammatory cells [9, 10]. CCL2 is upregulated in the glomerulus and tubulointerstitium of CKD. The CCL2/CCR2 axis involves in activation and infiltration of macrophages to damage sites and finally induce interstitial renal fibrosis [11]. Deficiency of CCL2 ameliorates renal artery stenosis-induced interstitial inflammation and fibrosis [12]. The enhanced CX3CL1/CX3CR1 axis is observed in the kidney of human and animal CKD models and contributes to infiltration of immune cells and the progression of fibrosis. However, inhibition of CX3CL1/CX3CR1 axis by deficiency or antagonists suppresses the progression of renal fibrosis and impairment in CKD [13]. Increased CCL17 was observed in CKD patients along with decrease of renal function [10]. Although the role of chemokines and their receptors in inciting the development of CKD has been demonstrated, the research on the up- and downstream of chemokines also needs to be further deepened.

This led us to explore the mechanism that regulates renal inflammation in the development of CKD. In the current study, by using multiple bioinformatic analyses of three GEO datasets, hub gene validation of other datasets, and unilateral ureteral obstruction (UUO) operated mice experiment, we found that the expressions of two key genes, Gng2 and Pf4, increased significantly with the development of CKD. These findings indicate that GNG2 and PF4 may be potential molecular biomarkers or targets for the initiation and development of CKD.

The profile datasets GSE38117, GSE87212, GSE125015, and GSE121190 were obtained from the NCBI Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database. The detailed information for these datasets is presented in Table 1. The datasets GSE38117, GSE87212, and GSE125015 were used for analyzing differentially expressed genes (DEGs), function enrichment, protein-protein interactions (PPIs), and hub genes. The dataset GSE121190 was utilized for validating the gene expression levels of hub genes.

The mean value of multiple probes was calculated for a single gene. The DEG identification between Sham control (Con) and UUO groups was performed using the limma (linear models for microarray data) package in R [14]. The |log₂ fold change|> 2 and p values <0.05 were considered the threshold for the DEG selection. The shared DEGs were identified by the venn diagram of DEGs.

DAVID (https://david.ncifcrf.gov/) is an online gene functional annotation database for analyzing the comprehensive sets of the biological meaning of candidate genes [15]. The DEGs were analyzed by DAVID for enriched Gene Ontology (GO) terms (including biological processes (BP), cellular components, and molecular functions) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway related to UUO.

To predict the PPI network of the DEGs, we used the Search Tool for the Retrieval of Interacting Genes (STRING, https://string-db.org/) database, a global online database [16]. The cut-off values were set at combined scores >0.9. Cytoscape (version 3.7.1) was used for PPI network construction [17], and the plug-in APP Molecular Complex Detection (MCODE) was employed to select notable modules within this PPI network [18]. The criteria were set at MCODE scores >7, node cut-off = 0.2, k-core = 2, and maximum depth = 100. Function enrichment of genes in module 1 was analyzed by STRING for enriched GO terms and KEGG pathway. Cytohubba, an APP in Cytoscape, was used to identify hub genes within the PPI network. According to the maximal clique centrality (MCC) method, top ten hub genes were identified and ranked. Additionally, ten hub genes were analyzed by GeneMANIA (http://genemania.org/), which is a gene function prediction web platform to investigate gene-gene interactions and predict the function of genes and gene sets [19].

Age- and weight-matched male C57BL/6 N mice (8- to 12-week-old, 20–29 g) were used in the present experiments. All mice were housed at the Center for Animal Experiments of Wuhan University in specific pathogen-free environments. The experiments were approved by the Ethics Committee of Tongren Hospital, and carried out in keeping with the guidelines of the National Health and Medical Research Council of China. For establishing the UUO model, the left ureter of mice was exposed and tied off with silk suture after anesthetizing with 30% isoflurane; Con mice were suffering the same operation without ligation. Mice were sacrificed for further experiments at 7 and 14 days after operation.

Total RNA was obtained from dissected kidneys by utilizing TRIzol reagents (Sigma-Aldrich, St. Louis, MO, USA). cDNA was obtained by mRNA transcription. Quantitative real-time PCR was carried out using the ABI7900 real-time PCR system (Illumina Eco, USA) with SYBR Premix Ex Taq II (Takara Bio, Otsu, Japan). The sequences of primers used are presented in Table 2.

The Western blotting on tissue lysates was performed as described [20]. In brief, total protein lysates were loaded into SDS-PAGE and transferred onto PVDF membrane (Merch Millipore, Darmstadt, Germany). After incubating with TBS containing 5% skim milk for 1 h at room temperature, the membrane incubated with rabbit polyclonal antibody against α-SMA (14395-1-AP, 1:1,000), mouse monoclonal antibody against fibronectin (FN) (66042-1-Ig, 1:1,000), and mouse monoclonal antibody against GAPDH (sc-365062, 1:2,500) primary antibodies overnight at 4°C and incubated with secondary antibodies for 1 h at room temperature. Band density was detected with the Odyssey CLX Infrared Image System (LI-COR Biosciences, Lincoln, NE, USA), and quantified using NIH ImageJ software.

Paraffin-embedded sections with 4 μm thickness from each isolated kidney were stained with hematoxylin and eosin or Sirius red following the standard procedure, respectively. The tubular damage and tubulointerstitial fibrosis were measured as described [21].

Paraffin-embedded slides with 4 μm thickness were deparaffinized and blocked with 10% normal donkey serum (Vector laboratories, Burlington, Canada) dissolved in PBS. Subsequently, the slides were incubated with α-SMA and FN primary antibodies, then incubated with Alexa Flour 488-conjugate secondary antibody. Finally, slides were stained with DAPI (Life Technologies, Carlsbad, CA, USA). Ten fields (×200 magnification) were randomly selected from 5 individual mice in each group, and the percentage of positive area was semiquantitated utilizing NIH ImageJ software (http://resbweb.nih.gov/ij/).

Statistical analyses were performed with SPSS version 23 (IBM-SPSS Inc., Chicago, IL, USA) using unpaired 2-tailed t test and one-way ANOVA multiple comparisons. All results are expressed as the mean ± standard error. The p value <0.05 was considered significant.

A total of 686, 2,084, and 1,644 DEGs were identified from GSE38117, GSE87212, and GSE125015 datasets, respectively. Volcano plots indicated the upregulated and downregulated DEGs between Con and UUO samples (Fig. 1a). These heatmaps demonstrated the distinct expression levels of DEGs in three datasets (Fig. 1b).

315 shared DEGs were identified in all three datasets (Fig. 2a), including 272 upregulated and 43 downregulated DEGs in UUO compared with Con. To identify further insights into the functional enrichment of shared DEGs and key candidate pathways in the development of UUO-induced obstructive nephropathy, GO terms and KEGG pathway for the DEGs were performed by DAVID. The result showed that the shared DEGs were significantly enriched in processes such as inflammatory response, innate immune response, and immune system process, in BP (Fig. 2b). For the cellular component (CC), the shared DEGs were mainly enriched in the membrane, extracellular region, extracellular exosome, etc. (Fig. 2c). The shared DEGs of molecular function (MF) were most related to protein binding, peptidase activity, etc. (Fig. 2d). Meanwhile, for the enriched KEGG pathways, the shared DEGs were most related to tuberculosis, osteoclast differentiation, chemokine signaling pathway, etc. (Fig. 2e). For the Reactome pathways, the shared DEGs were mainly enriched in platelet degranulation, GPVI-mediated activation cascade generation of second messenger, cell surface interactions at the vascular wall, degradation of the extracellular matrix, etc. (Fig. 2f).

The PPI network of the shared DEGs was built, containing 314 nodes and 438 edges, using the STRING database (Fig. 3a). The top two significant modules were further identified using the Cytoscape APP MCODE (Fig. 3b–c; Table 3). Module 1 contained thirteen genes, Gng2, Gpr18, Gpsm3, Sucnr1, Ccl5, Lpar3, Pf4, Cx3cr1, Anxa1, P2ry12, Ccl9, C3ar1, and Ccl6. Module 1 ranked the highest score, indicating the core function of these genes. The GO functions of the module 1 were related to macrophage activation, migration, and chemotaxis (BPs), chemokine receptor binding and chemokine activity (molecular functions), and extracellular organelles and space (cellular components) (Table 4). KEGG and Reactome pathways implied that these thirteen genes were significantly enriched in chemokine signaling pathway, cytokine-cytokine receptor interaction, hemostasis, and signaling transduction. Chemokine, chemokine interleukin-8-like domain, and G protein-coupled receptor were the features revealed by INTERPRO protein domain analysis (Table 4).

Based on the MCC algorithm, ten hub genes were further identified by Cytoscape APP Cytohubba, including Gng2, Ccl6, Ccl9, Anxa1, Lpar3, Pf4, C3ar1, Sucnr1, Gpsm3, and Gpr18 (Fig. 4a). We further investigated the interaction network of these hub genes and their neighboring genes using GeneMANIA database. As shown in Figure 4b, these interaction networks existed correlations of co-expression (65.99%), predicted (13.23%), shared protein domains (10.40%), and co-localization (10.38%).

The mRNA expression levels of these ten hub genes were validated in the dataset GSE121190. The expression levels of Gng2, Ccl9, and Pf4 were significantly upregulated, while the expression levels of Lpar3 were dramatically downregulated in the UUO group (Fig. 4c).

To figure out the expression changes of the four key genes during renal fibrosis, we established the UUO mouse model for 7 days (7d) and 14 days (14d). We found that damaged tubular number and interstitial fibrotic percentage were significantly increased in both UUO 7d and 14d compared to kidneys from Con, as shown by hematoxylin and eosin and Sirius red staining (Fig. 5a–b). The expression of the fibrotic markers, FN and α-SMA, were dramatically augment in UUO-induced mice compared to those in Con mice as examined by immunofluorescence staining (Fig. 5c). More severe fibrotic lesions were found in UUO 14d compared to those in UUO 7d (Fig. 5b–c). To further confirm above data, we investigated mRNA and protein levels of the fibrotic markers. mRNA and protein expressions of α-SMA and FN were significantly increased in both UUO 7d and UUO 14d compared to those in Con (Fig. 6a–b). We then investigated mRNA levels of the four key genes in our established UUO mice. Consistent with the data of datasets, the mRNA levels of Gng2 and Pf4 were markedly upregulated in both UUO 7d and UUO 14d compared with those in Con. Higher expression levels of Pf4 mRNA were manifested in UUO 14d than UUO 7d. Unexpectedly, the mRNA level of Ccl9 was significantly downregulated in both UUO 7d and 14d. Nevertheless, no significant change of Lpar3 mRNA expression level was observed among these groups (Fig. 6c).

Although there have been numerous explorations of the pathogenesis in CKD, there is still an urgent need to discover effective targets for preventing the development of CKD. In this study, we analyzed and identified the potential novel gene targets or biomarkers in CKD based on bioinformatic analysis of microarray datasets. By integrative analysis of three datasets, GSE38117, GSE87212, and GSE125015, 315 shared DEGs between UUO and control groups were identified. Ten top hub genes, including Gng2, Ccl6, Ccl9, Anxa1, Lpar3, Pf4, C3ar1, Sucnr1, Gpsm3, and Gpr18, ranked the highest MCC score. Subsequently, Gng2, Ccl9, Lpar3, and Pf4 were validated by the other dataset GSE121190 and our UUO-induced mouse model. The mRNA levels of Gng2 and Pf4 were upregulated with the development of CKD; however, the mRNA level of Ccl9 was downregulated in UUO mice. Together, our findings suggest that GNG2 and PF4 may be as potential molecular biomarkers or targets for the development of CKD.

Heterotrimeric guanine nucleotide binding proteins (G proteins) are crucial molecular switches for transmitting signal from G protein-coupled receptor (GPCR) to various downstream effectors [22]. The G proteins, composed of α, β, and γ subunits, are involved in a variety of BPs, including cell proliferation, inflammation, and immune response [23]. The first key gene, G protein subunit gamma 2 (Gng2)-encoded protein GNG2, is a member of the Gγ subunit. It has been reported that GNG2 regulated the inflammatory response during the progression of abdominal aortic aneurysm via interacting with chemokine CXCL1 and its receptor CCR7 [24]. Although the 3’ untranslated region polymorphism of GNG2 is associated with the risk of IgA nephritis in the Chinese Han population [25], there is no direct evidence that showed the role of GNG2 in renal inflammation and fibrosis. Systemic treatment of gallein, an inhibitor of Gβγ subunit-dependent signaling, downregulated renal inflammation and finally protected renal function in AKI and cardiorenal syndrome type 2 models [26, 27]. Therefore, specific inhibition of G proteins, even Gβγ subunits, may be a good alternative strategy for treating multiple diseases [23, 28, 29]. Regarding the role of GNG2 in renal inflammation, further exploration is still needed.

Platelet factor 4 (PF4), also known as C-X-C motif ligand 4 (CXCL4), is a chemokine abundantly expressed in platelet α-granule [30]. It is also expressed in various inflammatory cells, such as monocytes, activated T cells, and plasmacytoid dendritic cells (pDCs) [31, 32]. Accumulated evidence has demonstrated that PF4 plays a key role in a variety of pathophysiological processes, including inflammation, angiogenesis, and cancer [32]. PF4 derived from platelets and other cells participates in the inflammatory response and/or fibrosis in many organs [31, 33]. In CCL4-and thioacetamide-induced chronic liver fibrosis model, knockout PF4 increased expression levels of hepatic stellate cells and platelet-derived PF4 in lesions related to the liver inflammatory response and fibrosis progression [33]. The deficiency of PF4 reduced the infiltration of neutrophils and CD8-positive T cells and ultimately moderated liver fibrosis [33]. In a primary myelofibrosis (a type of myeloproliferative tumor that leads to progressive myelofibrosis) model, specific knockout of PF4 in hematopoietic stem and progenitor cells inhibited the activation of JAK/STAT signaling pathway, and reduced IL6 expression. Then, inhibition of IL6 promoted the myofibroblast phenotype manifested Gli1+ mesenchymal stromal cells, ultimately moderated the severity of myelofibrosis and reduced progressive anemia [32]. Furthermore, the complex of PF4 and DNA activates pDC via binding to TLR9 on pDC, which in turn cause inflammatory response of the system hardened skin [31]. However, administration of exogenous rat-derived PF4 can reduce the production of IL17 in Th17 cells and limit the differentiation of Th17 to alleviate inflammation and renal interstitial fibrosis in rat allogenic kidney transplantation [34]. Although the relationship between PF4 and inflammatory response has been studied for more than 30 years, the role of PF4 in inflammatory response is still contradictory. Therefore, the mechanism of PF4 in CKD and UUO animal models needs to be explored.

C-C motif chemokine ligand 9 (Ccl9) is a murine CC chemokine, and the corresponding human ortholog is Ccl15[35]. Ccl9/Ccl15 encoded protein is a member of the macrophage inflammatory protein-1 family that binds to CCR1 and participates in the regulation of the pathophysiological process in a variety of diseases, including inflammatory response of the asthmatic airway and metastasis of cancer cells [36, 37]. Studies have demonstrated that the increased expression of CCL15 in the circulating blood of patients with CKD is associated with renal damage and impaired renal function [38]. However, the mRNA level of Ccl9 was downregulated in our UUO model. The reasons leading to opposite result may be due to different sample source and expression level. Therefore, the role and biological mechanism of CCL9/CCL15 in the inflammatory response and fibrosis progression of CKD still need to be studied.

While we utilized other datasets and carried out in vivo experiments to confirm the results obtained by bioinformatics analyses, the molecular mechanisms underlying increased Gng2 and Pf4 mRNAs in the kidney of the CKD mouse model were not verified in this study. Expression changes in different cell types, roles, and downstream targets of GNG2 and PF4 need to be further investigated in the kidney of mice with CKD.

We constructed a gene network representing the potential molecular mechanism of CKD in the UUO mouse model. Ten hub genes are promising biomarkers. After validating by using the dataset GSE121190 and our UUO mice, two key genes, Gng2 and Pf4, were seen as potential targets for diagnosing or alleviating the progression of CKD. These findings provide evidence for further studying the progression of CKD.

We appreciate all participants in this study.

This study was performed in line with the guidelines of the National Health and Medical Research Council of China. All animal experiments were approved by the Ethics Committee of Tongren Hospital, approval number [SY2021-020].

The authors declare that there are no conflicts of interests.

This work was supported by grants from the Hubei Key Laboratory Opening Project, Wuhan, Hubei, China (Grant No.: 2021KFY045), National Natural Science Foundation of China (Grant No.: 81770688), and Hubei Leading Talent Program in Medicine, and Application Foundation and Frontier Project of Wuhan (Grant No.: 2020020601012209). The funding organization did not participate in the design of study, analysis of data, and decision of manuscript.

Kai Zhu designed this study, analyzed the data, and discussed and wrote manuscript. Xinxin Li performed experiments and analyzed the data. Likun Gao, Mengyao Ji, Xu huang, Yu Zhao, Wenxiu Diao, Yanqin Fan, and Xinghua Chen analyzed the data. Lei Shen, Lili Li, and Pengcheng Luo discussed and wrote manuscript. All authors have saw and approved the final manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.