The Interaction of miR-378i-Skp2 Regulates Cell Senescence in Diabetic Nephropathy

Diabetic nephropathy (DN) is the major cause of end stage renal disease. Proximal tubular epithelial cell (PTEC) injury occurs early in diabetic kidney, and it is correlated with consequent renal failure. Cellular senescence participates in the pathophysiology of DN, but its role remains unclear. We conducted a cross-disciplinary study, including human, in vivo, and in vitro studies, to explore the novel molecular mechanisms of PTEC senescence in DN. We found that HG induced cell senescence in PTECs, supported by enhanced β-galactosidase staining, p53 and p27 expression, and reduced cyclin E levels. Transcriptome analysis of PTECs from a type 2 diabetic patient and a normal individual using next generation sequencing (NGS) and systematic bioinformatics analyses indicated that miR-378i and its downstream target S-phase kinase protein 2 (Skp2) contribute to HG-induced senescence in PTECs. High glucose (HG) elevated miR-378i expression in PTECs, and miR-378i transfection reduced Skp2 expression. Urinary miR-378i levels were elevated in both db/db mice and type 2 diabetic patients, whereas decreased Skp2 levels were shown in proximal tubule of db/db mice and human DN. Moreover, urinary miR-378i levels were positively correlated with urinary senescence-associated secretory phenotype cytokines and renal function in in vivo and human study. This study demonstrates that the interaction between miR-378i and Skp2 regulates PTEC senescence of DN. miR-378i has the potential to predict renal injury in DN. These findings suggest future applications in both therapy and in predicting renal dysfunction of DN.


Introduction
Diabetes mellitus (DM) is an increasingly serious threat to human health and it must be regarded as an important public health issue [1]. Diabetic nephropathy (DN) has been the major cause of 2.7. Ingenuity Pathway Analysis (IPA) The IPA software (IPA, Ingenuity systems, Redwood City, CA, USA) provides "Core Analysis" for genes/proteins. Network analysis and canonical pathways that were obtained from Core Analysis results offer the interaction of genes/proteins. Network graphics with further overlay analysis can be generated by the software [20,23]. The IPA software also offers miRNA-mRNA filter analysis to match miRNAs and their potential targets of mRNAs according to the repression strength.

TargetScan and MiRmap Database
The TargetScan version 7.2 (http://www.targetscan.org) and miRmap (https://mirmap.ezlab.org) software rank potential targets of a specific miRNA based on Context++ score percentiles and miRmap scores, which indicates the repression strength of a miRNA target [20,24]. The TargetScan and miRmap web also can provide miRNA target predictions for different organisms.

RNA Extraction and Reverse Transcription PCR (RT-PCR)
Urine Exosome RNA Isolation Kit (Catalog #47200, Norgen, Thorold, ON, Canada) was used to isolate exosomal RNAs of human and mice urine (5 mL), following the manufacturer's protocol. Oligo (dT) primer and reverse transcriptase (RT; Takara, Shiga, Japan) were utilized to prepare the cDNA after RNA extraction. miRNAs were reverse transcribed using the Mir-X ™ miRNA First Strand Synthesis Kit (Catalog #638313 Takara, Shiga, Japan). We used SYBR Green on the StepOnePlus Real time PCR system (Applied Biosystems, Foster City, CA, USA) to analyze quantitative RNA and miRNA. PCR reaction was carried out with the following temperature profile (95 • C for 10 min, followed by 40 cycles at 95 • C for 15 s and 60 • C for 1 min). Relative expression levels of the mRNA and miRNA in cells were normalized to GAPDH or U6, respectively. miRNAs in exosomes that were isolated from urine were normalized with cel-miR-39 (Exiqon, Vedbaek, Denmark) as a spike-in control, and were then compared with a reference sample. Relative expression was presented using the 2 −∆∆Ct method. The primers used are listed in Table 1.

Experimental Animals
Five-week-old male C57BL/6 mice (n = 3), pathogen-free male db/m mice (non-diabetic animal model) (n = 6), and db/db mice (type 2 DM animal model) (n = 6) were purchased from the National Laboratory Animal Center in Taiwan, and were reared at the experimental animal center of Kaohsiung Medical University, under controlled the temperature at 25 ± 5 • C and the humidity of 55 ± 5% in regular light/dark cycles. At the 12th week, the mean body weight of C57BL/6 mice, db/m mice, and db/db mice was 23.3 ± 0.6 g, 27.8 ± 0.7 g, and 40.4 ± 4.7 g, respectively. Blood and urine samples were collected and kidneys harvested at the 12th week. The kidneys were fixed in 4% paraformaldehyde. All animal experiments in this study were in strict agreement with Kaohsiung Medical University and the Use Committee.

Human Study Participants
One hundred and seven type 2 DM patients with estimated glomerular rate (eGFR) ≥ 30 mL/min/1.73m 2 , and 45 healthy volunteers were enrolled. Study participants were asked to fast for at least 12 h before the collection of urine and blood samples. All urine and blood samples were aliquoted and stored in a −80 • C freezer.
Diabetes was defined as a medical history of diabetes or the use of anti-diabetes agents. Demographic and medical data, including age, gender, and angiotensin-converting enzyme (ACEI)/angiotensin II receptor blocker (ARB) usage, which has the effect of reducing albuminuria, were obtained from medical records and interviews with study participants. Kidney tissue samples were collected from four DN patients scheduled for kidney biopsy and four patients receiving nephrectomy because of upper tract urothelial carcinoma. The study was approved by the Institutional Review Board of the Kaohsiung Medical University Hospital (KMUHIRB-G(I)20160036, KMUHIRB-G(I)20170037, KMUHIRB-20130089).

Laboratory Data Measurement and Quantification of Urinary IL-6, IL-8, MCP-1 and Urinary Albumin/Creatinine Ratio (ACR) in Mice and Humans
Serum creatinine was measured by the compensated Jaffé (kinetic alkaline picrate) method in a Roche/Integra 400 Analyzer (Roche Diagnostics, Mannheim, Germany) using a calibrator traceable to isotope-dilution mass spectrometry [25]. The value of eGFR was calculated using the four-variable equation in the Modification of Diet in Renal Disease (MDRD) study [26]. Serum blood urea nitrogen (BUN) was measured using a glutamate dehydrogenase and urease kinetic method. Glucose was analyzed by six-carbon glucokinase enzyme method.
Levels of IL-6 and MCP-1 in the urine of mice and humans, and IL-8 in the urine of humans, were measured using Magnetic Luminex ® Assay (human Catalog #FCST03 and mice Catalog #LXSAMSM-02 from R&D System, Minneapolis, MS, USA). Levels of urinary albumin of humans and mice were measured using immunoturbidimetric assay with Tina-quant Albumin Gen.2 (ALBT2, Roche Diagnostics, Indianapolis, IN, USA). Concentrations of urine creatinine (Cr) in humans and mice were measured using the enzymatic method (creatinine plus version 2, CREP2, Roche Diagnostics, Indianapolis, IN, USA). Concentrations of IL-6, IL-8, and MCP-1 were corrected by urine Cr before statistical analysis. Normoalbuminuria was defined as urinary ACR <30 mg/g; microalbuminuria was defined as urinary ACR ≥30 mg/g and <300 mg/g; and, macroalbuminuria was defined as urinary ACR ≥300 mg/g. DN was defined as diabetic patients with urinary ACR ≥30 mg/g.

Statistical Analysis
The categorical variables were expressed as percentages. Differences in the distribution of categorical variables were tested using the Chi-square test. The continuous variables were expressed as mean ± S.E.M or median (25th, 75th percentile), as appropriate. The significance of differences in continuous variables between the groups was tested using Student's t-test or one-way analysis of variance (ANOVA), followed by the post hoc test adjusted with a Tukey correction, as appropriate. The association among continuous variables was examined by Spearman correlation.

HG Induces PTEC Senescence in DN
Cell senescence has been reported to participate in the pathophysiologic mechanism of DN [5]. To investigate cell senescence in human PTECs in DN, the cells were treated with NG and HG for seven and nine days, and SAβ-Gal staining was used to examine cell senescence. We found that cell morphologies changed from thin and spindle-shaped to flattened, irregular shapes with increased intracellular debris after HG treatment compared to those after NG treatment. The number of cells with SAβ-Gal staining was also elevated under the HG condition when compared to the NG condition ( Figure 1A). continuous variables between the groups was tested using Student's t-test or one-way analysis of variance (ANOVA), followed by the post hoc test adjusted with a Tukey correction, as appropriate. The association among continuous variables was examined by Spearman correlation.

HG Induces PTEC Senescence in DN
Cell senescence has been reported to participate in the pathophysiologic mechanism of DN [5]. To investigate cell senescence in human PTECs in DN, the cells were treated with NG and HG for seven and nine days, and SAβ-Gal staining was used to examine cell senescence. We found that cell morphologies changed from thin and spindle-shaped to flattened, irregular shapes with increased intracellular debris after HG treatment compared to those after NG treatment. The number of cells with SAβ-Gal staining was also elevated under the HG condition when compared to the NG condition ( Figure 1A).
Western blot was used to evaluate cell senescence markers, such as p53, p27, and cell cyclerelated factor cyclin E in human PTECs under NG or HG conditions. HG elevated p53 and p27 expression and reduced cyclin E1 and cyclin E2 ( Figure 1B-E) expression in PTECs after treatment for three and four days. We further investigated whether cell senescence occurred in mice and humans. IHC results revealed that p53 expression (high positive and positive proportion) in the proximal tubular area was higher in diabetic db/db mice when compared to normal C57B6 mice and non-diabetic db/m mice ( Figure 1F). Also, elevated p53 expression was found at the proximal tubular area of human DN when compared to normal individuals ( Figure 1G). Therefore, we suggest that PTEC senescence may contribute to the development of DN.  The effect of HG on morphological changes and senescence-associated β-galactosidase (SAβ-Gal) staining of human PTECs. PTECs were incubated under normal glucose (NG, 6.2 mM) and HG (30 mM) conditions for seven days and nine days. Cell senescence was assessed using SAβ-Gal staining. HG increased p53 (B) (D3: n = 6, D4: n = 4) and p27 (C) (D3: n = 3, D4: n = 3), and decreased cyclin E1 (D) (D3: n = 3, D4: n = 4) and cyclin E2 (E) (D3: n = 3, D4: n = 4) protein expression in human PTECs after three and four days of treatment. Protein levels were assessed by western blot. The expression of p53 in the proximal tubule of kidneys of mice (F) and humans (G). The kidney sections of C57BL/6 mice, non-diabetic db/m mice, and diabetic db/db mice, and human donors (upper tract urothelial carcinoma, UTUC with normal kidney function and normal glomerulus and proximal tubule) and patients with diabetic nephropathy (DN) were stained with p53 (brown). The images quantification was performed using the IHC Profiler Plugin of ImageJ Software. The bar graph represents the mean ± S.E.M. * p < 0.05, ** p < 0.01, *** p < 0.001 by Student's t test.

Identification of Differentially Expressed Genes Associated with Cell Senescence in PTECs of DM Patients and Normal Individuals
To investigate genes that are potentially associated with PTEC senescence in DN, RNA samples from PTECs of a normal individual and a DM patient were collected and RNAs were profiled by next generation sequence (NGS) and candidate regulators of PTEC senescence were detected by bioinformatics ( Figure 2A). The differentially expressed genes in normal and diabetic PTECs are displayed as a volcano plot in Figure 2B. Differentially expressed protein-coding genes were screened with >2.0-fold-change between normal and diabetic PTECs, with a threshold setting of >0.3 FPKM. After screening, 612 mRNAs with significant <2.0-fold-change of PTECs between diabetic and normal The effect of HG on morphological changes and senescence-associated β-galactosidase (SAβ-Gal) staining of human PTECs. PTECs were incubated under normal glucose (NG, 6.2 mM) and HG (30 mM) conditions for seven days and nine days. Cell senescence was assessed using SAβ-Gal staining. HG increased p53 (B) (D3: n = 6, D4: n = 4) and p27 (C) (D3: n = 3, D4: n = 3), and decreased cyclin E1 (D) (D3: n = 3, D4: n = 4) and cyclin E2 (E) (D3: n = 3, D4: n = 4) protein expression in human PTECs after three and four days of treatment. Protein levels were assessed by western blot. The expression of p53 in the proximal tubule of kidneys of mice (F) and humans (G). The kidney sections of C57BL/6 mice, non-diabetic db/m mice, and diabetic db/db mice, and human donors (upper tract urothelial carcinoma, UTUC with normal kidney function and normal glomerulus and proximal tubule) and patients with diabetic nephropathy (DN) were stained with p53 (brown). The images quantification was performed using the IHC Profiler Plugin of ImageJ Software. The bar graph represents the mean ± S.E.M. * p < 0.05, ** p < 0.01, *** p < 0.001 by Student's t test.
Western blot was used to evaluate cell senescence markers, such as p53, p27, and cell cycle-related factor cyclin E in human PTECs under NG or HG conditions. HG elevated p53 and p27 expression and reduced cyclin E1 and cyclin E2 ( Figure 1B-E) expression in PTECs after treatment for three and four days. We further investigated whether cell senescence occurred in mice and humans. IHC results revealed that p53 expression (high positive and positive proportion) in the proximal tubular area was higher in diabetic db/db mice when compared to normal C57B6 mice and non-diabetic db/m mice ( Figure 1F). Also, elevated p53 expression was found at the proximal tubular area of human DN when compared to normal individuals ( Figure 1G). Therefore, we suggest that PTEC senescence may contribute to the development of DN.

Identification of Differentially Expressed Genes Associated with Cell Senescence in PTECs of DM Patients and Normal Individuals
To investigate genes that are potentially associated with PTEC senescence in DN, RNA samples from PTECs of a normal individual and a DM patient were collected and RNAs were profiled by next generation sequence (NGS) and candidate regulators of PTEC senescence were detected by bioinformatics (Figure 2A). The differentially expressed genes in normal and diabetic PTECs are displayed as a volcano plot in Figure 2B. Differentially expressed protein-coding genes were screened with >2.0-fold-change between normal and diabetic PTECs, with a threshold setting of >0.3 FPKM. After screening, 612 mRNAs with significant <2.0-fold-change of PTECs between diabetic and normal PTECs were identified. Of 612 mRNAs, 280 mRNAs had up-expression and 332 mRNAs had down-expression in diabetic PTECs.
To understand the biological functions of the 612 differentially expressed genes in diabetic PTECs, these genes were first input into the DAVID database for enrichment analysis, using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. The top seven enriched terms of KEGG pathway are shown in Figure 2C. Cell cycle was the most greatly enriched biological processes, and p53 pathway was also one of the top seven of KEGG pathway. These results support the hypothesis that HG induces human PTEC senescence in vitro.
We further utilized the STRING database to identify the potential interaction among these differentially expressed genes associated with cell cycle regulation, based on the results from the DAVID database ( Figure 2D). Of the 14 genes related to cell cycle, Skp2 had the highest correlation with various cell cycle regulators, such as cyclin D2 (CCND2), cyclin B1 (CCNB1), and cyclindependent kinase inhibitor 1C (CDKN1C), as determined by the edge confidence of STRING database. We also used Core Analysis of IPA to analyze 612 differentially expressed genes in diabetic PTECs, with five networks that are related to cell development and cell cycle (Table 2). Skp2 had played a role in the regulation of cell cycle through the upregulation of cyclins ( Figure 2E). Therefore, the loss of Skp2 may cause cell cycle arrest, one factor of the senescence phenomenon in human PTEC.   To understand the biological functions of the 612 differentially expressed genes in diabetic PTECs, these genes were first input into the DAVID database for enrichment analysis, using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. The top seven enriched terms of KEGG pathway are shown in Figure 2C. Cell cycle was the most greatly enriched biological processes, and p53 pathway was also one of the top seven of KEGG pathway. These results support the hypothesis that HG induces human PTEC senescence in vitro.
We further utilized the STRING database to identify the potential interaction among these differentially expressed genes associated with cell cycle regulation, based on the results from the DAVID database ( Figure 2D). Of the 14 genes related to cell cycle, Skp2 had the highest correlation with various cell cycle regulators, such as cyclin D2 (CCND2), cyclin B1 (CCNB1), and cyclin-dependent kinase inhibitor 1C (CDKN1C), as determined by the edge confidence of STRING database. We also used Core Analysis of IPA to analyze 612 differentially expressed genes in diabetic PTECs, with five networks that are related to cell development and cell cycle (Table 2). Skp2 had played a role in the regulation of cell cycle through the upregulation of cyclins ( Figure 2E). Therefore, the loss of Skp2 may cause cell cycle arrest, one factor of the senescence phenomenon in human PTEC.

Decreased Skp2 Expression Contributes to Cell Senescence in DN
According to the results of our bioinformatic analysis, Skp2 participated in the regulation of the cell cycle of diabetic PTECs. We examined whether HG affects Skp2 expression in DN from in vitro, in vivo, to human studies. Decreased Skp2 mRNA was seen in diabetic PTECs compared to normal PTECs ( Figure 3A). Skp2 mRNA levels decreased in human PTECs that were treated with HG for three days ( Figure 3B). Western blot analysis showed that HG reduced Skp2 protein expression in human PTECs on days 3 and 4 ( Figure 3C). Furthermore, the IHC stain was used to investigate Skp2 expression in mice and humans, and it revealed that the Skp2 levels were lower in the proximal tubule area of db/db mice as compared to that of C57B6 mice and db/m mice ( Figure 3D). Consistently decreased levels of Skp2 were observed at the proximal tubular area of human DN patients than that of normal human kidney ( Figure 3E). Since the physiological function of Skp2 assists the cell cycle and HG diminishes Skp2 expression in PTECs, we inferred that decreased levels of Skp2 might cause PTEC senescence in DN.

Identification of Potential miR-378i-Skp2 Interaction in Diabetic PTECs
A large body of evidence shows that miRNAs regulate cellular senescence [19,20]. Based on our findings that Skp2 may be involved in PTEC senescence, we investigated whether loss of Skp2 is via miRNAs in the process of PTEC senescence. The heat map of potentially involved miRNAs in normal and diabetic PTECs is shown in Figure 4A. Of 67 miRNAs differentially expressed, three miRNAs had up-expression and 64 miRNAs had down-expression. Of the three up-expressed miRNAs, the . The kidney sections of C57BL/6 mice, non-diabetic db/m mice, and diabetic db/db mice, and human donors (UTUC with normal kidney function and normal glomerulus and proximal tubule) and patients with DN were stained with Skp2 (brown). The images quantification was performed using the IHC Profiler Plugin of ImageJ Software. The bar graph represents the mean ± S.E.M. * p < 0.05, *** p < 0.001 by Student's t test.

Identification of Potential miR-378i-Skp2 Interaction in Diabetic PTECs
A large body of evidence shows that miRNAs regulate cellular senescence [19,20]. Based on our findings that Skp2 may be involved in PTEC senescence, we investigated whether loss of Skp2 is via miRNAs in the process of PTEC senescence. The heat map of potentially involved miRNAs in normal and diabetic PTECs is shown in Figure 4A. Of 67 miRNAs differentially expressed, three miRNAs had up-expression and 64 miRNAs had down-expression. Of the three up-expressed miRNAs, the pathophysiological function of miR-378i was correlated to cell cycle and G1/S check point regulation, as the top TOX function, as determined by core analysis of IPA database ( Figure 4B). We examined miR-378i expression in PTECs using RT-PCR. miR-378i levels were higher in PTECs of a DM patient than those of a normal individual ( Figure 4C), and HG induced elevated miR-378i levels in human PTECs after treatment for three days ( Figure 4D). pathophysiological function of miR-378i was correlated to cell cycle and G1/S check point regulation, as the top TOX function, as determined by core analysis of IPA database ( Figure 4B). We examined miR-378i expression in PTECs using RT-PCR. miR-378i levels were higher in PTECs of a DM patient than those of a normal individual ( Figure 4C), and HG induced elevated miR-378i levels in human PTECs after treatment for three days ( Figure 4D). We further utilized miRNA target filter of IPA to search the downstream target of miR-378i, which contributes to cell senescence in PTECs, and found that miR-378i targeted Skp2 and regulated Skp2 expression (Table 3). Next, we used TargetScan (7.1 version), which is a web server that predicts the biological targets of miRNAs to search for the presence of sites that matched the seed region of miRNA. The results showed that miR-378i targeted the position 1889-1895 of Skp2 3' UTR ( Figure  4E), with the context score percentile of 91, which meant that the miR-378i-Skp2 link was highly practicable. Transfection of miR-378i mimic decreased the level of Skp2 in HEK2 cells ( Figure 4F). In short, miR-378i modulated Skp2 expression and miR-378i-Skp2 interaction may participate in cell senescence of human PTECs.  We further utilized miRNA target filter of IPA to search the downstream target of miR-378i, which contributes to cell senescence in PTECs, and found that miR-378i targeted Skp2 and regulated Skp2 expression (Table 3). Next, we used TargetScan (7.1 version), which is a web server that predicts the biological targets of miRNAs to search for the presence of sites that matched the seed region of miRNA. The results showed that miR-378i targeted the position 1889-1895 of Skp2 3' UTR ( Figure 4E), with the context score percentile of 91, which meant that the miR-378i-Skp2 link was highly practicable. Transfection of miR-378i mimic decreased the level of Skp2 in HEK2 cells ( Figure 4F). In short, miR-378i modulated Skp2 expression and miR-378i-Skp2 interaction may participate in cell senescence of human PTECs.

Urinary miR-378i Levels are Positively Correlated with Urinary SASP Levels and Albuminuria in Mice and Type 2 DM Patients
Senescent cells have been known to release SASP cytokines, such as IL-6, IL-8, and MCP-1, which can affect signal transduction in neighboring cells and surrounding tissue [8,9]. Since our results demonstrated the interaction between miR-378i and Skp2 in cell senescence of human PTECs, we further investigated the association of miR-378i with SASP and albuminuria in diabetic animal and human studies. Urinary miR-378i levels were measured in db/m mice (n = 6) and db/db mice (n = 6). The metabolic parameters of mice are shown in Table 4. The level of urinary miR-378i was higher in db/db mice than in db/m mice ( Figure 5A). Similarly, SASP cytokines, such as IL-6 and MCP-1, were higher in the urine of db/db mice than db/m mice ( Figure 5B,C). High urinary miR-378i/Cr levels were significantly correlated with urinary IL-6/Cr and MCP-1/Cr levels ( Figure 5D,E). Urinary miR-378i/Cr, IL-6/Cr, and MCP-1/Cr levels were positively correlated with log-formed urinary ACR in mice ( Figure 5F-H). Post-meal blood glucose, mg/dL 204.6 ± 65.9 521.8 ± 44.7 <0.001 Blood urea nitrogen, mg/dL 12.9 ± 1.6 42.7 ± 21.5 0.02 Serum creatinine, mg/dL; 0.0 ± 0.1 0.1 ± 0.1 0.02 Urinary albumin/creatinine ratio, mg/g 26.9 (3.0-30.8) 717.9 (282.1-1737.7) <0.001 Figure 5. Urinary miR-378i is positively correlated with senescence-associated secretory phenotype (SASP) and renal dysfunction in mice. (A) Urinary miR-378i levels were higher in db/db mice (n = 6) compared to db/m mice (n = 6). Urinary SASP levels, including interleukin-6 (IL-6) (B) and monocyte chemoattractant protein-1 (MCP-1) (C) were higher in db/db mice than db/m mice. Urinary-378i was Figure 5. Urinary miR-378i is positively correlated with senescence-associated secretory phenotype (SASP) and renal dysfunction in mice. (A) Urinary miR-378i levels were higher in db/db mice (n = 6) compared to db/m mice (n = 6). Urinary SASP levels, including interleukin-6 (IL-6) (B) and monocyte chemoattractant protein-1 (MCP-1) (C) were higher in db/db mice than db/m mice. Urinary-378i was positively correlated with IL-6 (D) and MCP-1 (E) respectively. High urinary miR-378i (F), IL-6 (G), and MCP-1 (H) were significantly correlated with log-formed urinary albumin-creatinine ratio (ACR) in mice. Exosomal miR-378i in the urine of mice was isolated, then assessed by quantitative real-time PCR normalized to a reference control. Magnetic Luminex ® Assay was used to assess IL-6 and MCP-1 in the urine of mice. Urine albumin was measured using immunoturbidimetric assay, and urine creatinine was determined by an enzymatic method. The bar graph represents the mean ± S.E.M. ** p < 0.01 by Student t test, and p-value of correlation was analyzed by Spearman analysis. The unit of Cr as mg/dL. The unit of IL-6 and MCP-1 as pg/mL. Cr: creatinine.
Forty-five normal individuals and 107 type 2 DM patients were enrolled in our study (Table 5). There was no significant difference of age and sex distribution between the two groups. Thirty-three percentage of type 2 diabetic patients used ACEI/ARB. We found that urinary miR-378i levels were elevated in the type 2 DM patients as compared to normal individuals ( Figure 6A). Type 2 DM patients also had higher urinary IL-6/Cr, IL-8/Cr, and MCP-1/Cr levels than normal individuals ( Figure 6B-D). A positive association of urinary miR-378i/Cr levels with urinary IL-6/Cr, IL-8/Cr, and MCP-1/Cr levels is presented in Figure 6E-G. Urinary miR-378i/Cr, IL-6/Cr, IL-8/Cr, and MCP-1/Cr levels were positively correlated with urinary ACR (Figure 6H-K). Human participants with macroalbuminuria had higher urinary miR-378i levels, IL-6, IL-8, and MCP-1, than those with normoalbuminuria ( Figure 6L-O). After adjusting ACEI/ARB, albuminuria was still significantly and positively associated with urinary miR-378i, IL-6, IL-8, and MCP-1 (Table 6). Besides, urinary miR-378i/Cr, IL-6/Cr, IL-8/Cr, and MCP-1/Cr levels were negatively correlated with eGFR ( Figure 6P-S). Consistent with the results of the in vitro study, miR-378i participates in the mechanism of cell senescence-mediating kidney injury, and urinary miR-378i has the potential for predicting kidney dysfunction.   Exosomal miR-378i in the urine of humans was isolated, then assessed by quantitative real-time PCR normalized to a reference control. Magnetic Luminex ® Assay was used to assess urinary IL-6, IL-8, and MCP-1. Urine albumin was measured using immunoturbidimetric assay and urine creatinine was determined by an enzymatic method. Normoalbuminuria is defined as urinary ACR <30 mg/g; microalbuminuria is defined as urinary ACR ≥30 mg/g and <300 mg/g; macroalbuminuria is defined as urinary ACR ≥300 mg/g. The bar graph represents the mean ± S.E.M. * p < 0.05, ** p < 0.01, *** p < 0.001 by Student t test or ANOVA, followed by the post hoc test adjusted with a Tukey correction, and p-value of correlation was analyzed by Spearman analysis. The unit of Cr as mg/dL. The unit of IL-6 and MCP-1 as pg/mL. Cr: creatinine. The unit of eGFR as mL/min/1.73m 2 .

Discussion
DN develops in one-third of all DM patients, and it accounts as the major cause of ESRD worldwide [28]. Understanding DN mechanisms more clearly, together with early detection of DN onset and progression, greatly assists in avoiding the consequent complications of DN. This cross-disciplinary study demonstrates that HG induces PTEC senescence by enhancing miR-378i expression and reducing Skp2 expression. Furthermore, elevated urinary miR-378i levels are found in diabetic db/db mice and DM patients, and they are positively correlated with SASP, including IL-6, IL-8, and MCP-1, in the urine. The positive association of renal dysfunction with urinary miR-378i is seen in DM patients. This is the first study to provide new perceptions of the unique regulation of miR-378i-Skp2, illustrating the role of PTEC senescence in DN development (Figure 7).

Discussion
DN develops in one-third of all DM patients, and it accounts as the major cause of ESRD worldwide [28]. Understanding DN mechanisms more clearly, together with early detection of DN onset and progression, greatly assists in avoiding the consequent complications of DN. This crossdisciplinary study demonstrates that HG induces PTEC senescence by enhancing miR-378i expression and reducing Skp2 expression. Furthermore, elevated urinary miR-378i levels are found in diabetic db/db mice and DM patients, and they are positively correlated with SASP, including IL-6, IL-8, and MCP-1, in the urine. The positive association of renal dysfunction with urinary miR-378i is seen in DM patients. This is the first study to provide new perceptions of the unique regulation of miR-378i-Skp2, illustrating the role of PTEC senescence in DN development ( Figure. 7). Cell senescence halts PTEC turnover and repair processes and causes progressive inflammation and renal injury [3]. SASP is a possession of senescent cells that coexists with genomic damage and epigenetic abnormality [29]. SASP has played a principal role in tissue microenvironments of renal pathophysiology and it provides biological signal transduction in neighboring cells [30]. SASP promotes the spread of senescence to nearby cells [11] and it further boosts chronic and low-grade inflammation that interrupts homeostasis and results in the development of diseases [29]. Cell senescence participates in pathophysiologic mechanisms of DN [13] and our study shows that HG induces senescence in PTECs.
Based on the bioinformatics results, we found that Skp2 is involved in cell senescence in PTECs in DN. Skp2, a well-characterized F-box protein, is a crucial component of ubiquitin protein ligase complex, called SCF (SKP1-cullin-F-box) [31], which targets phosphorylated cyclin-dependent kinase Cell senescence halts PTEC turnover and repair processes and causes progressive inflammation and renal injury [3]. SASP is a possession of senescent cells that coexists with genomic damage and epigenetic abnormality [29]. SASP has played a principal role in tissue microenvironments of renal pathophysiology and it provides biological signal transduction in neighboring cells [30]. SASP promotes the spread of senescence to nearby cells [11] and it further boosts chronic and low-grade inflammation that interrupts homeostasis and results in the development of diseases [29]. Cell senescence participates in pathophysiologic mechanisms of DN [13] and our study shows that HG induces senescence in PTECs.
Based on the bioinformatics results, we found that Skp2 is involved in cell senescence in PTECs in DN. Skp2, a well-characterized F-box protein, is a crucial component of ubiquitin protein ligase complex, called SCF (SKP1-cullin-F-box) [31], which targets phosphorylated cyclin-dependent kinase inhibitor (p27(Kip1) and p21(Cip1)) for ubiquitin modification and subsequent proteasome degradation, thereby facilitating the activation of cdk2-cyclin E complex for progression into the S-phase during the G1/S transition of the cell cycle [32]. Overexpression of Skp2 leads to cell proliferation and tumorigenesis, and it is associated with the down-regulation of p27 [33]. Skp2 deficiency restricts cell development through the up-regulation of p21, p27, and ATF4 [34]. Down-regulation of Skp2 inhibited mesangial cell proliferation [35,36], and Skp2 accumulation causes podocyte injury [37]. However, the role of Skp2 in PTECs is unknown in DN. In the present study, we have demonstrated that HG induces PTEC senescence by reducing Skp2 expression in PTECs, which stops cell development, leading to cell senescence [34]. Therefore, we have concluded that Skp2 contributes to PTEC senescence, further resulting in DN progression. miRNAs found in biological fluids have been suggested to act in cell-cell communication or as endocrine genetic signals during physiological or pathophysiological processes [38]. The stability of miRNAs in the circulation and in body fluids, their tissue, and disease specificity, and their easy, reliable quantification methods make them feasible as potential biomarkers [39]. miRNAs in the urine have been regarded as potential biomarkers of renal injury [40]. Combination of NGS and IPA analysis confirms that miR-378i has potential activity in the regulation of cell cycle, and Skp2 is the downstream target of miR378i. miR378i has been reported to suppress cell proliferation in hepatocellular carcinoma and colon cancer [41,42]. Our results demonstrate that HG increases the expression of miR-378i in PTECs, and elevated miR-378i is also found in PTECs of type 2 DM patients, supporting the hypothesis that miR-378i contributes to the progression of DN. Computational analysis using IPA shows that miR-378 has a regulatory activity on the cell cycle, and miR-378i mimics decreases Skp2 expression, further corroborating the finding that miR-378i is involved in the cell cycle arrest of senescent PTECs. Moreover, the significantly positive association of urinary miR-378i with urinary SASP, including IL-6, IL-8, and MCP-1, was found in both db/db mice and type 2 DM patients. miR-378i plays a key role in the regulation of PTEC senescence in DN. These results, based on experimental cell studies, animal models, and clinical patients, strongly suggest that miR-378i is one of the critical effectors in PTEC aging enhancing the development of DN in type 2 DM patients.
To examine the predictive activity of miRNAs in renal injury due to DM, we measured urinary miR-378i levels in mice and humans. Our findings show that both diabetic mice and type 2 DM patients have higher miR-378i levels in the urine, and elevated urinary miR-378i is associated with lower eGFR and high albuminuria, a critical renal injury marker. Albuminuria traditionally reflects the glomerular injury. Interestingly, a recent study that was conducted by Wagner et al. provided a new insight that the proximal tubule itself can directly regulate plasma albumin levels through compensating for excessive albumin loss by increasing reclamation [43]. Albuminuria presents as not only glomerular injury, but also the proximal tubular damage [44,45], and it can be considered as the marker of proximal tubule. However, it is difficult to differentiate the origin of albuminuria. Detecting proximal tubular markers (megalin, cubulin) or podocyte indicator (podocin) will help in identifying the source of albuminuria. This is the limitation of this study. Nevertheless, miR-378i is still a potential biomarker that can indicate renal injury in DN and it provides a mechanism for recognizing the early signs of onset or poor progression of DN in clinical practice.

Conclusions
Our study demonstrated that the interaction between miR-378i and Skp2 induces PTEC senescence under the HG condition. miR-378i has predictive activity for renal injury in DN. This study analyzes the new pathophysiological mechanism of cell senescence of DN, and it provides future applications in the prediction of DN progression.