PTGER3 and MMP-2 play potential roles in diabetic nephropathy via competing endogenous RNA mechanisms

Background Diabetic nephropathy (DN) is a primary complication of diabetes mellitus (DM). The pathology of DN is still vague, and diagnostic accuracy is not enough. This study was performed to identify miRNAs and genes that have possibilities of being used as therapeutic targets for DN in type 2 DM. Methods Human miRNA data GSE51674 and gene data GSE111154 were downloaded from the Gene Expression Omnibus database. Differentially expressed genes (DEGs) and miRNAs (DEmiRNAs) in the kidney between control and DN patients were screened out. The competing endogenous RNA (ceRNA) network was constructed, and key lncRNA-miRNA-mRNA pairs were selected accordingly. Potential drugs targeting DEGs were screened out and validated using PCR analysis. Results Totally, 83 DEmiRNAs and 293 DEGs were identified in GSE51674 and GSE111154, respectively. Thirteen of the top 20 DEmiRNAs (10 up and 10 down) targeted to 47 DEGs. In the ceRNA network, RP11-363E7.4/TTN-AS1/HOTAIRM1-hsa-miR-106b-5p-PTGER3 and LINC00960-hsa-miR-1237-3p-MMP-2 interaction pairs were identified as the key ceRNA network. Interestingly, PTGER3 and hsa-miR-1237-3p were downregulated, and MMP-2 and hsa-miR-106b-5p were upregulated in the kidney of patients with DN compared with normal controls, respectively. PTGER3 and MMP-2 were targeted by drugs including iloprost, treprostinil, or captopril, and the deregulation of the two genes was confirmed in the plasma samples from patients with DN as compared with controls. Conclusions We speculated that the RP11-363E7.4/TTN-AS1/HOTAIRM1-hsa-miR-106b-5p-PTGER3 and LINC00960-hsa-miR-1237-3p-MMP-2 networks were associated with diabetic renal injury.

The pathophysiology leading to DN and resultant renal failure from DM consists of hypertension, altered composition and proliferation, and sclerosis in glomerular [10]. Accordingly, the estimated glomerular filtration rate (eGFR), albuminuria, and hypertension are the clinical manifestations of DN [3]. However, the clinical usage of them could not make an accurate and definite diagnosis of DN and the degree of kidney damage [3]. Herein, DN is often confirmed during postmortem examination because of the lack of definitive clinical findings. Theoretically, the progression and complications of DN could be blocked or slowed down by earlystage interventions [10][11][12]. A major obstacle for that is the lack of accurate biomarkers identifying patients who at high risk of DN. To fulfill this gap, experiments are being carried out to find effective methods for the accurate and preoperative diagnosis of DN [13].
Much evidence has shown that genetic factors are the major predispositions to type 2 DM and DN [1,10]. Many non-coding RNAs and genes like transforming growth factor (TGF) β, nephrin, angiotensin-converting enzyme-2, interleukin-6 (IL-6), and tumor necrosis factor (TNF)-α have been identified to be associated with the development, progression, and prognosis of DN [9,[12][13][14]. These genetic factors modulate extracellular matrix structure as well as the oxidative, fibrotic, and inflammatory responses in the kidney, tubular, and glomerular [12,15,16]. Nevertheless, no single treatment targeting the above factors has been able to reverse or mitigate DN progression, or reduce DN-associated mortality [14]. Therefore, additional factors that have therapeutic targeting possibilities in DN are in urgent need.
This study was performed to identify the genes and miRNAs that were differentially expressed in the kidney of patients with DN. Integrated bioinformatics analysis was performed for screening the miRNAs and genes in DN in type 2 DM. Moreover, the possibility of using them as therapeutic targets would be discussed.

Data processing
The raw data in GSE51674 and GSE111154 datasets were downloaded and processed using the Limma package [17] and Oligo package [18], respectively. Data processing included background correction, normalization, and expression calculation. The probes in GSE51674 and GSE111154 that mapped to human miRNAs and mRNAs, respectively, were retained and used for further analysis. When multiple probes mapped to one gene symbol, the averaged expression level of probes was calculated and regarded as the gene expression level of that gene or miRNA. The probes that did not map to genes or human miRNAs were removed.

Differential expression analysis
The differentially expressed genes (DEGs) and miRNAs (DEmiRNAs) between control and disease samples were identified using the paired t-test in the Limma package [17]. Significant DEGs were identified according to the criteria of p value < 0.05 and |log 2 (fold change, FC)| > 0.585, and significant DEmiRNAs were screened out according to stricter criteria of |log 2 FC| > 2 and adjusted (BH correction) p value < 0.05.

Construction of protein-protein interaction (PPI) network
The protein-protein interaction (PPI) pairs among the products of DEGs were screened in the STRING (Version 10.0, http://www.string-db.org/) [19]. The interaction pairs with a score of higher than 0.4 (medium confidence) were retained and used for the construction of the PPI network. Cytoscape (version 3.2.0, http://www.cytoscape.org/) was employed for the construction of the PPI network. Dominant modules in the PPI network were identified using the MCODE plugin (version 1.4.2, http://apps.cytoscape. org/apps/MCODE) that was provided by Cytoscape. The DEGs in modules were employed for the enrichment analysis.

Functional enrichment analysis
The functional enrichment analysis was performed separately for the DEGs. The clusterprofiler in R was used for enriching the Gene Ontology biological processes and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Significant enrichment was identified with the thresholds of adjusted (BH correction) p value< 0.05 and gene number count ≥2.

Prediction of miRNA-mRNA target
To investigate the correlation between the DEmiRNAs and DEGs in DN, the predictive targets of DEmiRNAs were screened from the DEG list and the miRNA-mRNA regulatory network was constructed accordingly. The targets that were collected by at least 6 of the 7 databases (miRWalk, miRanda, miRDB, miRMap, Pictar2, RNA22, and Targetscan) were regarded as potential targets in our study. Then, the miRNA-mRNA target pairs were obtained and used for the construction of the miRNA-mRNA regulatory network using Cytoscape.

Prediction of lncRNAs that sponge DEmiRNAs
To investigate the competing endogenous RNA (ceRNA) involving lncRNAs, DEGs, and DEmiRNAs in DN, the potential lncRNA-miRNA pairs were identified in the DIANA-LncBase V2 [20]. The lncRNAs that tightly interacted with the DEmiRNAs (score ≥ 0.99) in the kidney were selected. The miRNA-mRNA and lncRNA-miRNA interaction pairs that are linked by the same miRNAs were retained and used for the construction of the lncRNA-miRNA-mRNA (ceRNA) network. Cytoscape was used for the construction of the ceRNA network.

Drug-gene interaction network
Drug-Gene Interaction database (DGIdb) provides information on the interactions between drugs and genes, which might be of great value for settling potential therapeutic strategy that targets the key genes. DGIdb was employed to extract the drugs targeting the DEGs in the above ceRNA network. Drugs that have been admitted by the Food and Drug Administration (FDA) were reserved and used to construct the drug-gene interaction network.

Patients and ethical statement
The peripheral blood samples were collected from three patients with DN and three healthy controls from the Department of Endocrinology, China-Japan Union Hospital of Jilin University, Changchun, China, in Sept 2020. Approval was obtained from the Ethics Committee of China-Japan Union Hospital of Jilin University, and written informed consent was obtained from all participants before the collection of blood samples. All the samples were stored at 4°C before RNA extraction.

Statistical analysis
The expression levels of the MMP-2 and PTGER3 genes were expressed as mean ± standard deviation. The differences in the expression levels of MMP-2 and PTGER3 were analyzed using the t-test in Graphpad Prism 5 (Graphpad Software, San Diego, CA). P < 0.05 was set as the threshold for the significant difference.

Identification of DEmiRNAs and DEGs
After data normalization, a total of 83 DEmiRNAs (|log 2 FC| > 2 and adjusted p < 0.05) and 293 DEGs (p < 0.05 and |log 2 FC| > 0.585) were identified in the GSE51674 dataset and the GSE111154 dataset, respectively ( Fig. 1a and b). The sample clusterings based on these DEmiRNAs and DEGs are presented in Fig. 1c and d, respectively. The heatmaps indicated that expression Fig. 3 The enrichment analysis of the 12-gene module. a, the dominant module that consists of 12 upregulated genes. The node size indicates the coordinated interaction degree. b and c, the biological processes and KEGG pathways that significantly enrich the genes in the module profiles of these genes were different between the DN and control samples.

Analysis of the PPI network
Based on the predicted protein interactions in the STRI NG database, the PPI network that consisted of 228 DEGs and 732 interaction pairs was constructed (Fig. 2).

Construction of the miRNA-mRNA regulatory network
The miRNA-mRNA regulatory network was constructed to investigate the regulatory mechanisms involving the DEmiRNAs and DEGs in DN. Before the prediction of miRNA targets, 10 upregulated and 10 downregulated DEGs with the top absolute values of log 2 FC were selected (Table 1). Then, a total of 47 DEGs (42 up-and 5 downregulated DEGs) were identified to be targeted by 13 DEmiRNAs, including 10 upregulated DEmiRNAs and 3 downregulated DEmiRNAs. Subsequently, the differentially expressed miRNA-mRNA network was constructed ( Fig. 4). In this network, 17, 7, and 6 DEGs were targeted by hsa-miR-106b-5p, hsa-miR-374a-5p, and hsa-miR-1237-3p, respectively.

Identification of potential drug targets
To understand the potential drugs targeting PTGER3 and MMP-2, the drug-gene interactions were predicted and the gene-drug interaction network involving the two genes was constructed (Fig. 6). It was composed of 6 PTGER3 agonists (bimatoprost, dinoprostone, misoprostol, iloprost, dinoprost, and treprostinil) and two MMP-2 inhibitors (captopril and tiludronate; Fig. 6).

Validation of the MMP-2 and PTGER3 expression
The PCR analysis showed that there were significant differences in the expression levels of the MMP-2 and PTGER3 genes in the plasma samples between patients with DN and healthy controls ( Fig. 7a and b). The expression level of the PTGER3 gene in patients with DN was significantly lower than that from healthy controls (p = 0.0360, Fig. 7a), while the MMP-2 gene was significantly upregulated in DN patients as compared with controls (p = 0.0054, Fig. 7b).
The member of the EP receptor family, prostaglandin EP3 receptor gene (PTGER3/EP3) is a G protein-coupled receptor that facilitates a broad range of physiological actions including blunting glucose-stimulated insulin secretion (GSIS) [21]. It has been linked to the dysfunction and loss of β-cells in type 2 DM [21,22]. PTGER3 upregulation in the islets of patients with type 2 DM has been proven previously [23]. Additionally, elevated production of prostaglandin E 2 (PE2) and PTGER3 can be targeted to improve insulin secretion from islets [24]. Pathologically, PTGER3 negatively regulates the production of cyclic AMP (cAMP), a potentiator of GSIS [22,25]. Rankin et al. [21] showed that agonist (sulprostone) and antagonist of EP3r both significantly reduced GSIS and fasting plasma glucagon in non-human primates with noninsulin-dependent diabetes mellitus. Schaid et al. [24] showed that γPTGER3 was constitutively activated in pancreatic β-cells and its action of reducing Fig. 4 The miRNA-mRNA regulatory network. Triangles and circles indicate the miRNAs and genes, respectively. Orange and green color notes the upregulation and downregulation in diabetic samples versus control samples, respectively. The node size indicates the coordinated interaction degree insulin was Gαz dependent. The αPTGER3 activity, however, could be stimulated by sulprostone in Gαz/ cAMP-dependent and -independent signalings. After treatment with sulprostone, Rap1GAP-mediated inhibition of GSIS was potentiated in β-cells [24]. This was in line with the results reported by Rankin et al. [21].
However, the association of PTGER3 expression with nephropathy or other renal diseases is rarely reported [3,26,27]. In renal cell carcinoma (RCC), the lower expression of PTGER3 was correlated with a worse prognosis [26]. In patients with DN, downregulation of PTGER3 in the kidney from donors with early DN versus matched controls has been confirmed by Sircar et al. [3]. What's more, there was a differentially methylated level in the PTGER3 gene between diabetic patients with end-stage renal disease and those without nephropathy [27]. Two or more methylated CpG sites were found in the PTGE R3 gene in diabetic patients with the end-stage renal disease compared with those without nephropathy [27]. An in vivo animal experiment showed that the knockout of PTGER3 did not reduce the levels of plasma and urine glucose in the streptozotocin-induced mouse model of diabetes [28]. However, there was a significant increment in urine osmolality, water reabsorption, and aquaporins expression in mice kidney. Obvious reductions in renal hypertrophy, hyperfiltration, albuminuria, tubular dilation, and nuclear cysts were observed in the kidney in streptozotocin-treated Ep 3 −/− mice compared with wild type mice [28]. These results showed that PTGER3 downregulation was a specific protection for the kidney against metabolic disorders in diabetes. Its downregulation might be used as a biomarker or therapeutic strategy for DN.
Apart from the PTGER3 gene, matrix metalloproteinase-2 (MMP-2) is also differentially methylated in diabetic patients with the end-stage renal disease compared with those Fig. 5 The lncRNA-miRNA-mRNA regulatory network. Diamonds, triangles and circles indicate the lncRNAs, miRNAs and genes, respectively. Orange and green color notes the upregulation and downregulation in diabetic samples versus control samples, respectively. The node size indicates the coordinated interaction degree without nephropathy [27]. MMP-2 is a well-characterized protein responsible for tissue remodeling and extracellular matrix degradation. MMP-2 expression is a marker of newly formed β-cells and is negative in the islets of adult rats [29]. It was positively correlated with periodontitis severity in patients with type 2 DM [30]. In addition, the expression of MMP-2 was higher in patients with type 2 DM compliant with the peripheral arterial disease compared with diabetic patients without arterial diseases and control groups [31,32]. In patients with chronic kidney diseases, creatinine level was positively correlated with MMP-2 (r = 0.39), but in diabetic patients it was negative correlation (r = − 0.72) [31]. Our present study confirmed the upregulation of MMP-2 in DN might be associated with the remodeling of renal and preventing sclerosis in glomerular. The potential of employing if as a biomarker for DN in patients with diabetes should be validated.
Based on the analysis of DEmiRNAs in DN, we identified the potential regulatory networks in DN, including RP11-363E7.4/TTN-AS1/HOTAIRM1-hsa-miR-106b-5p-PTGER3 and LINC00960-hsa-miR-1237-3p-MMP-2. In direct contrary to PTGER3 and MMP-2, hsa-miR-106b-5p and hsa-miR-1237-3p were up-and down-regulated in DN versus control, respectively. Hsa-miR-106b-5p has been proven to contribute to β-cell proliferation following bone marrow transplantation [33]. It was also Fig. 6 The drug and gene interaction network involving two key candidates Fig. 7 The expression of the two genes in the plasma samples from patients with diabetic nephropathy (DN). a and b, the expression levels of the PTGER3 gene and the MMP-2 gene in the plasma samples from patients with DN, respectively. The difference was analyzed using the t-test decreased in diabetic patients with chronic kidney disease versus patients without kidney disease [34], and is significantly linked to kidney function and eGFR [35]. HOTAIRM1 expression is crucial for the differentiation of embryonic stem cells to renal lineage [36]. However, there is less information on the association of these miR-NAs and lncRNAs with DN and the progression of other renal diseases. The tight interactions between these lncRNAs, DEmiRNAs (hsa-miR-106b-5p and hsa-miR-1237-3p), and DEGs (including PTGER3 and MMP-2) indicated the potential and important roles of them in regulating kidney function.
Here in our study, we identified that PTGER3 and MMP-2 were targeted by drugs including iloprost, treprostinil, and captopril. Iloprost and treprostinil are clinically used for severe pulmonary arterial hypertension and chronic thromboembolic pulmonary hypertension [37,38]. Iloprost shows efficacy in decreasing urinary albumin excretion rate in patients with DN and attenuating the progression of DN [39]. Iloprost could downregulate MMP-2 via direct targeting [40]. Since the downregulation of PTGER3 and upregulation of MMP-2 are protective for renal diseases, the agonists of PTGER3 (including iloprost and treprostinil) and inhibitor of MMP-2 (like captopril) might be inapplicable for the treatment of DN. Novel drugs aiming to increase MMP-2 or decrease PTGER3 in the kidney might of great values for preventing DN and renal damages in diabetes.

Conclusions
In summary, we confirmed the crucial roles of PTGER3 downregulation and MMP-2 upregulation in the kidney from patients with DN. They were regulated by lncRNAs (including RP11-363E7.4, TTN-AS1, HOTAIRM1, and LINC00960) and miRNAs (including hsa-miR-106b-5p and hsa-miR-1237-3p) that had been rarely reported in DN. Their dysregulation might be the defense mechanism against renal injury following diabetes. However, more preclinical experiments and clinical trials should be performed to determine the dysregulation of them was a blessing or a curse for DN.