Knockdown of circ_0003928 ameliorates high glucose-induced dysfunction of human tubular epithelial cells through the miR-506-3p/HDAC4 pathway in diabetic nephropathy

Serum samples were collected from 41 DN patients (gender: 20 male and 21 female; age: 56.4 ± 6.2) and 29 healthy volunteers (gender: 17 male and 12 female; age: 57.3 ± 5.6) in Hebei General Hospital. The diagnosis of diabetic nephropathy was performed by two experienced renal pathologists. All serum samples were obtained by centrifugation at 3000 rpm for 10 min, and stored at -80℃ for subsequent RNA extraction. The clinical trial obtained approval from the Ethics Committee of Hebei General Hospital (approval number 2020NYH Y-010–04). All participants signed the written informed consent.

Human tubular epithelial cells (HK-2) were purchased from Procell (Wuhan, China), and cultured in Dulbecco’s modified Eagle’s medium (DMEM; Procell) added with 10% fetal bovine serum (FBS; Procell) and 1% penicillin/streptomycin (Procell) at 37˚C in an atmosphere of 5% CO₂. According to the reported method [17, 18], the DN cell model was established by treating HK-2 cells with 30 mmol/L D-glucose (HG; MedChemExpress, Shanghai, China), with 5 mmol/L D-glucose (NG; MedChemExpress) and 30 mmol/L Mannitol (MedChemExpress) acting as controls.

Total RNA from HK-2 cells was extracted using TsingZol (Tsingke, Shanghai, China). RNA quality was measured by detecting the A260/A280 ratio on NanoDrop-1000 apparatus. 2 μg RNA and 100 nm mRNA were reversely transcribed to cDNA using Goldenstar™ RT6 cDNA Synthesis Mix (Tsingke), gDNA remover (Tsingke) and RNasin (Tsingke). 5 μg miRNA was reversely transcribed to cDNA using miRNA synthesis reagents (TaKaRa, Dalian, China), including mRQ Buffer, mRQ Enzyme and RNase I as instructed. Then, qRT-PCR was performed to detect gene expression using T5 Fast qPCR Mix (Tsingke). In brief, 50 ng cDNA was mixed with 10 μL T5 Fast qPCR Mix (SYBR Green), 0.8 μL primers and 0.4 μL ROX Reference Dye, and reacted at 95 °C for 1 min, and 40 cycles at 95 °C for 10 s and 60 °C for 10 s on an Applied Biosystems. The 2^−∆∆Ct method was used to calculate gene expression with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) [19] and U6 as internal controls. The primer sequences are listed in Table 1.

The small interfering RNA against circ_0003928 (si-circ_0003928, 5′-AAGAGGATGGCCCCAGAAGTA-3′), the mimics of miR-506-3p (miR-506-3p, 5′-UAAGGCACCCUUCUGAGUAGA-3′), the inhibitors of miR-506-3p (anti-miR-506-3p, 5′-UCUACUCAGAAGGGUGCCUUA-3′), and respective controls, including siRNA negative control (si-NC), the negative control of miRNA mimics (miR-NC) and the negative control of miRNA inhibitors (anti-miR-NC) were synthesized in Ribobio Co., Ltd. (Guangzhou, China). The open reading frame of HDAC4 was amplified and then inserted into the pcDNA 3.1 vector by GenePharma Co., Ltd. (Shanghai, China) to generate the overexpression plasmid of HDAC4 (HDAC4). Then, cell transfection was performed on HK-2 cells in accordance with the manufacturer’s direction of FuGENE6 (Roche, Basel, Switzerland).

To analyze cell viability, HK-2 cells treated with mannitol, normal glucose (NG), HG, plasmids, small RNA or small interfering RNA were cultured in 96-well plates for 48 h. Then, the cell counting kit-8 (CCK-8) reagent (Beyotime, Shanghai, China) was added to each well of these plates. After 4 h of incubation, these samples were analyzed using a microporous plate spectrophotometer (Bio-Rad, Hercules, CA, USA) with a wavelength of 450 nm.

After various treatments as mentioned above, HK-2 cells were cultured in 6-well plates for 48 h. Afterward, the cells were sub-cultured in the 96-well plates added with EdU-labeled DMEM at 37 °C for 2 h. Subsequent experiments were conducted according to the instruction of the EdU staining kit (Ribobio). Finally, cell proliferation was determined by analyzing EdU-positive cells using a fluorescence microscope (Olympus, Tokyo, Japan).

The assay regarding the detection of reactive oxygen species (ROS), malondialdehyde (MDA) and superoxide dismutase (SOD) levels was performed to evaluate oxidative stress. In brief, 2 × 10⁶ HK-2 cells for each experiment were harvested, and subsequent procedures were carried out following the instruction of the cellular ROS assay kit (Abcam, Cambridge, MA, USA), lipid peroxidation MDA assay kit (Abcam), and SOD activity assay kit (Abcam). Finally, the outputs of these samples were measured by a microporous plate spectrophotometer (Bio-Rad).

HK-2 cells were harvested and lysed using lysis buffer (Beyotime). The cells were centrifuged at 18,000 g for 12 min, and then cell supernatant was collected. These samples were incubated with Ac-DEVD-pNA (Beyotime), and caspase 3 activity was detected using a microplate reader (Bio-Rad).

After various treatments, HK-2 cells were cultured in 12-well plates for 48 h. These cells were harvested by centrifuging and then suspended in Binding buffer (Solarbio, Beijing, China), followed by incubating with Annexin V-FITC (Solarbio) and propidium iodide (Solarbio). At last, the stained cells were analyzed using a flow cytometer (Thermo Fisher, Waltham, MA, USA).

Online database starbase (http://​starbase.​sysu.​edu.​cn/​agoClipRNA.​php?​source=​mRNA) was employed for the prediction of potential regulatory relationships among circ_0003928, miR-506-3p and HDAC4. Then, circ_0003928 and HDAC4 3′UTR sequences harboring the complementary sites of miR-506-3p were amplified by qRT-PCR and introduced into the pmirGLO vector to construct wild-type (WT) reporter plasmids, including circ_0003928-WT and HDAC4-3′UTR-WT. Similarly, the mutant (MUT) plasmids, including circ_0003928-MUT and HDAC4-3′UTR-MUT, were assembled. As instructed [20], cell transfection was performed on HK-2 cells with the use of the reporter plasmids, miRNA mimics, miRNA negative controls and transfection reagents. After 48 h of cell transfection, the cells were harvested for the detection of luciferase activity using a Dual-Lucy assay kit (Solarbio).

The HK-2 cells cultured in 12-well plates were treated with biotin-labeled miR-506-3p mimics (Bio-miR-506-3p) and Bio-miR-NC alone. After 48 h of culture, these cells were lysed using RIP lysis buffer. After that, the supernatant was collected and incubated with streptavidin-coupled beads (Sigma, St. Louis, MO, USA). At last, circ_0003928 level in the co-precipitated RNAs was analyzed by qRT-PCR.

All data were obtained from three independent duplicate tests and analyzed by GraphPad Prism software. Results were shown as means ± standard deviations (SD). Significant differences were compared with Mann–Whitney U test, Student’s t-tests, or analysis of variance. P < 0.05 indicated statistical significance.

Circ_0003928 is located in chr17:27,809,241–27,838,010 and generated by the cyclization of exons 8–15 of the TAOK1 gene (Fig. 1A). Meanwhile, the data from qRT-PCR showed the high expression of circ_0003928 in HG-induced HK-2 cells as compared with normal glucose (NG) or mannitol-induced HK-2 cells (Fig. 1B). The above results suggest that the pathogenesis of DN may be associated with circ_0003928.

HK-2 cells were treated with 30 mmol/L D-glucose to simulate a DN cell model in vitro, regarding 5 mmol/L D-glucose and 30 mmol/L Mannitol as controls. Subsequently, we analyzed the effects of HG treatment on HK-2 cell proliferation, oxidative stress and cell apoptosis using the model. As displayed in Fig. 2A and B, HG treatment inhibited cell viability and cell proliferation in comparison with control groups. Besides, oxidative stress was induced after HG treatment in HK-2 cells. For instance, ROS and MDA levels were increased, and SOD activity was inhibited after HG treatment (Fig. 2C–E). Comparatively, HG treatment increased caspase 3 activity and induced HK-2 cell apoptosis, accompanied by the increases of Bax and cleaved caspase 3 expression and a decrease of Bcl 2 expression (Fig. 2F–K). Collectively, the above evidence suggests the successful establishment of the DN cell model.

We depleted circ_0003928 in HG-induced HK-2 cells to determine the consequential effects on cell proliferation, oxidative stress and cell apoptosis. The data from Fig. 3A first showed that HG-increased circ_0003928 expression was attenuated after circ_0003928 knockdown. Subsequently, HG-induced inhibition of cell viability and cell proliferation were relieved when circ_0003928 expression was decreased (Fig. 3B and C). HG treatment led to increases in OS and MDA levels and a decrease in SOD activity; however, these effects were attenuated after transfection with si-circ_0003928 (Fig. 3D–F). Comparatively, the increased caspase 3 activity and cell apoptotic rate by HG treatment were rescued after circ_0003928 depletion (Fig. 3G and H). Further, knockdown of circ_0003928 remitted HG treatment-triggered dysregulation of Bax, cleaved caspase 3 and Bcl 2 expression (Fig. 3I-L). Therefore, these findings demonstrate that circ_0003928 can regulate HG-induced oxidative stress and apoptosis of HK-2 cells.

Starbase online database was employed to predict the possible target miRNAs of circ_0003928. In this study, we analyzed 4 miRNAs that not only contained circ_0003928-binding sites, but also inhibited DN progression, including miR-142-3p, miR-205-5p, miR-31-5p and miR-506-3p. qRT-PCR first showed that circ_0003928 depletion increased the expression of miR-142-3p and miR-506-3p, especially miR-506-3p, but it had no significant difference in miR-205-5p and miR-31-5p expression (Additional file 1: Figure S1A). Thus, miR-506-3p was chosen as a follow-up study. As presented in Fig. 4A, miR-506-3p contained the binding sites of circ_0003928, suggesting the potential interaction between circ_0003928 and miR-506-3p. Consistently, the data from the qRT-PCR analysis showed that miR-506-3p was significantly downregulated in HG-treated HK-2 cells compared with its expression in the HK-2 cells treated with mannitol or NG (Fig. 4B). Our data confirmed the interaction of circ_0003928 with miR-506-3p. For instance, miR-506-3p overexpression significantly inhibited the luciferase activity of wild-type reporter plasmid of circ_0003928 rather than that of the mutant reporter plasmid, as shown in Fig. 4C. Meanwhile, biotin-labeled miR-506-3p could greatly enrich circ_0003928 in comparison with biotin-labeled miR-NC, as analyzed by RNA pull-down assay (Fig. 4D). Further, circ_0003924 expression was negatively correlated with miR-506-3p in the blood samples of DN patients (Additional file 2: Figure S2A). Taken together, the above evidence demonstrates that circ_0003928 binds to miR-506-3p.

Given the regulatory relationship between circ_0003928 and miR-506-3p, the study continued to analyze whether miR-506-3p participated in the regulation of circ_0003928 toward HG-induced HK-2 cell damage. To this end, we transfected si-circ_0003928, anti-miR-506-3p and matched controls into HG-induced HK-2 cells. Figure 5A first showed that HG-induced inhibition of miR-506-3p expression was enhanced after transfection with anti-miR-506-3p, showing the high efficiency of miR-506-3p knockdown. Then, circ_0003928 knockdown-induced increases of cell viability and cell proliferation were relieved after miR-506-3p knockdown under HG treatment (Fig. 5B and C). Besides, knockdown of circ_0003928 led to decreased ROS and MDA production and increased SOD activity in HG-induced HK-2 cells, but these effects were remitted when miR-506-3p expression was reduced (Fig. 5D–F). Similarly, circ_0003928 depletion decreased caspase 3 activity, cell apoptotic rate and the protein expression of Bax and cleaved caspase 3, and increased Bcl 2 expression; however, these effects were remitted after miR-506-3p depletion (Fig. 5G–L). Collectively, these findings demonstrate that the circ_0003928/miR-506-3p axis is involved in HG-induced HK-2 cell injury.

We further identified miR-506-3p-associated mRNAs. SOX6, CDKN1A, ROCK1, NFAT5, ICAM1 and HDAC4 were chosen for subsequent studies as these mRNAs potentially bound to miR-506-3p and contributed to DN progression. By qRT-PCR analysis, we found that miR-506-3p also affected (downregulated) the expression of ROCK1 and HDAC4, especially HDAC4, among these mRNAs (Additional file 1: Figure S1B). Thus, HDAC4 was employed in the following study. HDAC4 containing the potential binding sites of miR-506-3p (Fig. 6A), as predicted by the starbase online database, was selected for the subsequent study. The study first showed a significantly high expression of HDAC4 in HG-induced HK-2 cells, as compared with control groups (Fig. 6B and C). Subsequently, we found that miR-506-3p mimics significantly reduced the luciferase activity of wild-type reporter plasmid of HDAC4 3′UTR, but that of mutant reporter plasmid had no significant difference after transfection with miR-506-3p mimics (Fig. 6D). Consistently, it was found that miR-506-3p inhibitors dramatically increased the HDAC4 expression (Fig. 6E). As revealed by Spearman correlation analysis, miR-506-3p was negatively correlated with HDAC4 expression in serum samples of DN patients (Additional file 2: Figure S2B). The above data demonstrate that HDAC4 is a target gene of miR-506-3p. Based on the above results, we analyzed the regulatory relationship among circ_0003928, miR-506-3p and HDAC4 using Western blotting. As presented in Fig. 6F, circ_0003928 depletion inhibited HDAC4 production, whereas the effect was relieved after miR-506-3p was downregulated. Taken together, all evidence suggests that circ_0003928 can regulate HDAC4 expression through miR-506-3p.

Considering that circ_0003928 could regulate HDAC4 expression by associating with miR-506-3p, we further analyzed whether the inner mechanism of circ_0003928 in mediating HG-caused HK-2 cell damage involved HDAC4. As shown in Fig. 7A, HG treatment significantly promoted HDAC4 production, which was increased after HDAC4 overexpression. Subsequently, circ_0003928 knockdown promoted the viability and proliferation of HG-induced HK-2 cells, which was attenuated after HDAC4 overexpression (Fig. 7B and C). The decreased ROS and MDA production and increased SOD activity by circ_0003928 knockdown were also relieved by the enforced expression of HDAC4 in HG-treated HK-2 cells (Fig. 7D–F). Comparatively, circ_0003928 depletion inhibited caspase 3 activity, cell apoptotic rate and the protein expression of Bax and cleaved caspase 3, and promoted Bcl 2 production in HG-treated HK-2 cells; however, these effects were remitted when HDAC4 expression was increased (Fig. 7G–L). Thus, these results demonstrate that circ_0003928 regulates HG-induced cell disorders through HDAC4.

The study further detected circ_0003928, miR-506-3p and HDAC4 expression in the serum samples from DN patients and healthy volunteers. As shown in Fig. 8A, circ_0003928 was significantly upregulated in the serum samples of DN patients compared with healthy controls. Subsequently, we observed a lower level of miR-506-3p and a high HDAC4 expression in the serum samples of DN patients compared with control groups (Fig. 8B–D).