MiR-199b-3p prevent the epithelial-mesenchymal transition (EMT) and brosis of renal tubule by regulating E-cadherin through targeting KDM6A in Diabetic nephropathy (DN)

Bo Tang Department of Nephrology, Qingpu Branch of Zhongshan Hospital A liated to Fudan University Weiliang Li Departmen of Urology, Qingpu Branch of Zhongshan Hospital A liated to Fudan University Tingting Ji Department of Nephrology, Qingpu Branch of Zhongshan Hospital A liated to Fudan University Xiaoying Li Department of Nephrology, Qingpu Branch of Zhongshan Hospital A liated to Fudan University Xiaolei Qu Department of Nephrology, Qingpu Branch of Zhongshan Hospital A liated to Fudan University Shoujun Bai (  baishoujun@126.com ) Qingpu Branch of Zhongshan Hospital A liated to Fudan University https://orcid.org/0000-00023717-5053


Introduction
Diabetic nephropathy (DN) is identi ed as the serious microvascular complication of diabetes which is one of the common causes of end-stage renal disease (1,2). With an increasing number of new diagnoses and a 5-year survival rate of ~ 20%, DN has attracted much attention (3). A growing number of evidence elucidates the injury of proximal tubule critically helps to the renal injury and even the initial stage of DN (4). In recent years, many researchers have indicated that epithelial-mesenchymal transition (EMT), a main pathological process of renal tubular epithelial cell, promotes tubulointerstitial brosis which is the classical feature of end-stage renal disease including DN (5)(6)(7)(8).
Most studies have focused on the role of TGF-β/BMP7/Smad pathways in the progression of DN, while recent studies indicate that chromatin modi cations may also have a signi cant effect on the occurrence of DN (9)(10)(11)(12). KDM6A (also known as Histone demethylase UTX, ubiquitously transcribed tetratricopeptide repeat, X chromosome) was identi ed in 2007 (13), which could speci cally remove diand tri-methyl groups from lysine 27 residue of histone H3 (H3K27) together with KDM6B (also known as JMJD3, jumonji domain containing 3 ) (14,15). In this way, ectopic expression of KDM6A is closely associated with the changed gene expression. Interestingly, several reports have shown that KDM6A play a novel regulatory role in diabetic kidney disease. Chen H et al found KDM6A is in a position to regulate in ammation and DNA damage in db/db mice (16). Lin CL et al found that KDM6A can aggravate diabetic kidney injury through inducing podocytes dysfunction (17). However, little investigation has been done on expounding the role of KDM6A in renal tubulointerstitial brosis of DN.
MiRNAs are endogenous noncoding RNAs with 21-23 nucleotides (nt) which could regulate gene expression via binding to the 3'UTR of target gene mRNA (18). Recent advances in the study of miRNA have shown that miRNA can control the EMT and brosis in renal disease, indicating that miRNA may have a potent effect on DN. For example, miR-21, Let-7d miRNA, miR-302 showed an obvious activity to regulate the process of EMT and brosis in DN (19)(20)(21). We speculate that there is a miRNA participating in the progression of DN by targeting KDM6A mRNA.
In this study, we reported miR-199b-3p could alleviate the process of renal tubule EMT and brosis in DN though binding to KMD6A mRNA. We also explored the deleterious effect of KDM6A on aggravating the renal tubule injury via repressing E-cadherin expression.

Materials And Methods
Cell cultures, reagents and transfections HK-2 cells obtained from China Center for Type Culture Collection (CCTCC) was cultured in MEM (5 mM D-glucose) (Gbico) supplemented with 10% FBS (Gbico), 100ug/ml streptomycin and 100 U/ml penicillin (Hyclone) at 37 °C with 5% CO2. For EMT induction, HK-2 cells were growth in high glucose medium (30 mM D-glucose) or high mannitol medium (5 mM D-glucose and 25 mM mannitol) for 72 h. All primary antibodies were purchased from Abcam and all second antibodies were obtained from Sino Biological. Streptozotocin (STZ) and GSK-J4 were obtained from sigma.
For HK2-E-cadherin stable cell line, HK-2 cells were infected with lentivirus contain E-cadherin encoding sequence or not and selected with puromycin (sigma, 2ug/ml) for 2 weeks. The package of lentivirus was performed in HEK293T co-transfected with psPAX2, pMD2G and pCDH-CMV-Mcs-Ef1-Puro (Addgene) for 72 h. For HK2-miR stable cells, HK-2 cells were infected with lentivirus contain miR-199b-3p or miNC as described (22). The silence RNA for negative control (si-NC), silence RNA for downregulating E-cadherin (si-E-cadherin) were obtained from Invitrogen. The miR-199b-3p inhibitor (inhibitor) and inhibitor NC were obtained from RiboBio. Transfection experiments were performed by using Lipofectamine 2000 reagent according to the manufacturer's instruction (Thermo Fisher Scienti c).
The sequence of siRNA and miRNA was exhibited in Table S2.

Wound Healing Assay
The cells were seed at the density of 5 × 10 5 cells/well in 6-well plates. The cell monolayer was performed by a 200µL pipette tip for forming "wound". Photographs were taken after 0 h and 48 h. The wound areas were measures by ImageJ software and the "relative wound area" was normalized by wound areas at 0 h.

Animals, Diabetic Animal Models And GSK-J4 Treatment
Male C57BL/6 mice (Guangdong Medical Laboratory Animal Center) were intraperitoneally given 190 mg/kg STZ or equivalently normal saline to induce diabetes. The blood glucose levels were equalized by adding 1-2 unit/kg insulin in each diabetic mouse as decreased previously (17) . Mice were seen as diabetes with 200-300 mg/dl of fasting blood glucose. For GSK-J4 treatment, mice were subcutaneously injected with 0.4 mg/kg each day at 1 week after onset of diabetes. MiR-199b-3p knockout mice were obtained from C57BL/6 mice through Nanjing Biomedical Research Institute of Nanjing University.

Mouse Metabolic Measurements
Albuminuria of was quanti ed by albumin and creatinine ratio (ACR) with mouse albumin ELISA kit (Bethyl Laboratories). Creatinine was detected through the Creatinine Assay Kit (BioAssay System). BUN was tested by Urea Assay Kit (BioAssay System). Proteinuria was examined by Proteinuria Elisa kit (Nanjing Anyan Biological Technology). All of those assays were performed at 8 weeks after the onset of diabetes in each mice. Quantitative Real-time PCR Total RNA was extracted using the Trizol reagent (Invitrogen) and cDNAs were gained using PrimeScript RT reagent kit (TaKaRa) except that of miRNA by TaqMan Advanced miRNA cDNA synthesis kit (Waltham). PCR quanti cation was performed as previously decreased (23). To test the expression of miRNA, we used stem-loop qRT-PCR for miR-199a-3p, miR-199b-3p and miR-758-3p following the protocol decreased previously (24,25). Primers was showed in Table S2.

Protein Extraction And Western Blot Analysis
Proteins were extracted from cells and mice kidney (Beyotime Biotechnology) and the protein lysates (20 µg) were quanti ed through bicinochoninic acid assay kit (BOSTER). Then the proteins of each sample were loaded in lane and tested in SDS-PAGE gel at 80V for 30 min and then 100V for 1 h after boiling for 5 min at 100 °C in loading buffer. Next, the proteins were transferred to a nitrocellulose membrane (Boster) at 300A for 1.5 h. The membranes were blocked in Tris-buffered saline contain 0.1% Tween 20 (TBST) and 5% fat-free milk for 1 h. Then the membranes were incubated with rst antibodies overnight at 4 °C. After washing, the membranes were incubated with second antibodies for 2 h at room temperature. The membranes were treated with high-sig electrochemiluminescence kit (Fdbio Science) and detected by an image analysis system (Image-Pro Plus 6.0, Media Cybernetics).

Tissue Histochemical Staining, HE And PAS Staning
Mouse kidneys were obtained in separate treated mice at 8 weeks after the onset of diabetes. Mouse kidneys embedded in para n were cut into 4-mm-thick sections and stained with PAS and HE (Sigma). Stained slides were observed by an experienced pathologist for histopathological examination under light microscopy. For histochemical staining, para n-embedded sections were stained with anti-KDM6A, anti-E-cadherin and anti-TGFβ as described previously (27).

Statistical analysis
Data in the study were analyzed by using Prism (version 5; GraphPad Software) and were compared through the Student's t test. It expressed as the Mean ± SD. Values were considered statistically signi cant if P < 0.05.

Results
KDM6A showed a promotive activity for regulating the EMT in HK-2 cells Bearing in mind the importance of epigenetic modi cation in the development of kidney disease, we rstly investigated the expression and signi cance of KDM6A in the HK-2 cells cultured in high glucose (HG) and mannitol (Mann) medium. First of all, we con rmed the change of HK-2 migration in the HG group, which is closely associated with epithelial-mesenchymal transition (28) Those results illustrated that KDM6A might be able to propel the progression of epithelial-mesenchymal transition in renal tubular epithelial cells.
KDM6A exhibited a pivotal role in kidney injury of DN Basing on the in vitro studies, we next investigated the in vivo role of KDM6A in diabetes kidney of DN using a STZ-induced mouse model (WT-NC groups and WT-STZ groups). We indeed found a higher level of KDM6A was expressed in kidney tissues of the 12-week STZ-induced diabetic mice relative to control groups, both measured by qPCR and Immunohistochemistry (IHC) ( Fig. 2A-C). To further explore the in uence of KDM6A on the progression of diabetes kidney injury, GSK-J4 was applied to treat diabetic mice with negative control (WT-STZ-NC and WT-STZ-G groups). The injury of diabetic kidney was evaluated through the morphology of renal proximal tubule and renal distal tubule which were associated with tubulointerstitial brosis. Obviously, the HE and PAS stain results in kidney of STZ-induced mice showed renal proximal tubular epithelial cells expansion and proximal tubule lumen reduction relative to normal mice, while the degree of proximal tubule injury in diabetic mice was dramatically decreased after treated with GKS-J4 (Fig. 2D, F). Similarly, severe brosis of renal distal tubule in diabetic mice was monitored and the progress of brosis was retarded in the presence of GKS-J4 (Fig. 2E, G). In those GSK-J4-induced or non-GSK-J4-induced diabetic mice, several typical signs of diabetic nephropathy, such as albumin to creatinine ratio (ACR), blood urea nitrogen (BUN), creatinine (Cr), proteinuria, was measured.
All of those sighs were aggravated in STZ-induced mice and restored partly by GSK-J4 treatment (Fig. 2H-K), which were closely consistent with HE and PAS stain results. Those results indicated that the downregulation of KMD6A might alleviate the kidney injury of diabetic nephropathy mice.
Interestingly, the IHC and its quantitative results displayed a suppressed expression of E-cadherin, a maker of EMT and brosis process (14), and an elevated expression of TGF-β, another maker of EMT and brosis process (6), whereas both of them were altered by GSK-J4 (Fig. 2L-O). Treated diabetic mice with GSK-J4, the depressed E-cadherin was recovered and the increased TGF-β was downregulated, totally conforming to the previous results.
Together, the expression of KDM6A was improved in the kidney tissues of STZ-induced diabetic nephropathy mice and the inhibition of KDM6A resulted in a mitigative kidney injury, suggesting that KDM6A could promote the progress of diabetic nephropathy.
KDM6A was the upstream regulator of E-cadherin for repressing the its expression Tons of reports have showed that repressing the epithelial cadherin (E-cadherin) could trigger the EMT and brosis directly (29,30). Depending on the previous work, we assumed that KDM6A might be the upstream effector of E-cadherin and could inhibit the E-cadherin expression, resulting in EMT and even brosis of DN. To testify our assumption, HK-2 cells stably overexpressed E-cadherin (HK2-vehicle and HK2-E-cadherin groups) was constructed. Overexpressing E-cadherin could inhibit the EMT process and reduced the migration of HK-2 cells comparing to HK2-vehicle groups when cells were cultured in HG medium (Fig. 3A). However, as the increasing of E-cadherin in HK2-E-cadherin groups, the expression of KDM6A showed no difference in both transcription and translation level between HK2-vehicle and HK2-Ecadherin groups (Fig. 3B, C). It seemed that KDM6A expression might be not affected by overexpressing E-cadherin.
Furthermore, the silencing RNA, targeting E-cadherin (si-E-cadherin), was applied to knockdown the Ecadherin in HK-2 cells as well as negative control (si-NC) when cells were cultured in mannitol medium. In this condition, the migration of HK-2 was improved dramatically with the E-cadherin knockdown (Fig. 3D).
Notably, the expression of KDM6A was either still in a stable state even if the EMT was aggravated by Ecadherin knockdown (Fig. 3E, F). All those results showed the expression of KDM6A would not be impacted only by variation of E-cadherin expression, suggesting that KDM6A was the upstream regulator of E-cadherin.
To con rm the results, qPCR assay was performed for testing those miRNA expression in HG-induced HK-2 cells and the results showed a similar trend with precious work (Fig. 4A). We also detected the miRNA expression in diabetes kidney using a STZ-induced mouse model. The results was compatible with the consequence in HG-treated HK-2 cell (Fig. 4B).
Considering the greatest change of miR-199b-3p, we assumed that miR-199b-3p could bind to KDM6A mRNA. According to the predicted alignment of miR-199b-3p binding to KDM6A mRNA (Fig. 4C), WT and mutant KDM6A 3'UTR luciferase plasmids as well as miR-199b-3p mimics and mi-NC were designed for dual-luciferase reporter assay in order to con rm the prediction. Comparing with cells co-transfected with WT KDM6A 3'UTR and mi-NC, cells co-transfected with WT KDM6A 3'UTR and miR-199b-3p mimics displayed a conspicuous reduction of luciferase activity. However, when the WT KDM6A 3'UTR was replaced by mutant KDM6A 3'UTR, no luciferase activity change could be observed regardless of cotransfecting with miR-199b-3p mimics or mi-NC (Fig. D).
Those results suggested that miR-199b-3p expression was changed in the EMT of renal tubular epithelial cells and could target to KDM6A, which means miR-199b-3p may play a regulatory role in tubulointerstitial brosis of DN through downregulate KDM6A.

MiR-199b-3p showed a protective effect against epithelial-mesenchymal transition of renal tubular epithelial cells through downregulating KDM6A
For further explore the role of miR-199b-3p in the process of EMT in renal tubular epithelial cells, we transfected the HK-2 cells with lentivirus containing miR-199b-3p mimics or miR-NC and selected the stable cell line (HK2-miRNA and HK2-miNC) and cultured them in HG medium. The expression of miR-199b-3p in both cell lines was tested (Fig. S2A). As the increasing expression of miR-199b-3p in HK2-miRNA, the expression of KDM6A, measured by western blot, was inhibited in HK2-miRNA cells (Fig. 5A) with a slightly lower transcription level of KDM6A detected by the qPCR (Fig. S2A). The expression of TGFβ and E-cadherin was also tested. Decreased expression of TGFβ was observed, whereas the expression of E-cadherin was upregulated relative to HK2-miNC cells (Fig. 5A). The similar tendency was also monitored in transcription level (Fig. S2B). Furthermore, the HK2-miNC cells in HG showed a more serious EMT than HK2-miRNA cells in HG with the deteriorative migration, suggesting that miR-199b-3p could halt HG-induce EMT of HK-2 cells dramatically (Fig. 5B). Those results indicated that miR-199b-3p might be able to downregulate the KDM6A and further protected HK-2 cell from the HG-induced injury.
In order to con rm the in uence of miR-199b-3p on HK-2 cells through downregulating KDM6A, we transfected HK2-miRNA cells with a miR-199b-3p inhibitor (inhibitor) or negative control (inhibitor NC) to determine whether inhibition of miR-199b-3p could exacerbate HG-induced EMT. The qPCR results showed the successful inhibition of miR-199b-3p because of the inhibitor, accompanying with the restored expression of KDM6A (Fig S2. C). Consistently, the maker of EMT, TGFβ and E-cadherin, showed an increased and decreased tendency respectively, no matter in mRNA level or protein level (Fig. 5F, Fig.  S2D). As we supposed, the expression of KDM6A in translation level was certainly restored due to the inhibitor existence (Fig. 5C). In the presence of inhibitor, migration of HK2-miRNA cells were increased, which indicated HK2-miRNA cells were preferred to mesenchymal-like performance compared with inhibitor NC group (Fig. 5D). Those results suggested that the restoration of KDM6A due to repression of miR-199b-3p could aggravate the EMT process.
Together, miR-199b-3p could prevent the degeneration of EMT in renal tubular epithelial cells through suppressing KDM6A expression.
MiR-199b-3p could inhibit the kidney injury caused by diabetic nephropathy by regulating KDM6A expression In order to further con rm the function of miR-199b-3p in vivo, we induced the diabetic nephropathy in miR-199b-3p de cient mice (miR-199b-3p −/− , Fig. S3A) using STZ as well as negative control (miR-199b-3p −/− -NC and miR-199b-3p −/− -STZ). Several typical signs of DN were measured in different groups and all of them showed a similar tendency. The level of ACR, BUN, Cr and proteinuria in miR-199b-3p −/− -STZ groups were signi cantly greater than WT-STZ groups, indicating the protective function was removed due to the miR-199b-3p de ciency (Fig. 6A-D). Furthermore, the pathological features of STZ-induced diabetic kidney in WT and miR-199b-3p −/− mice were observed through HE and PAS stain. Also in the induction of STZ, a more serious renal proximal tubular epithelial cells expansion as well as a more severe proximal tubule lumen reduction was emerged in miR-199b-3p −/− mouse compared to WT mouse (Fig. 6E, G). Consistently, more severe brosis of renal distal tubule was monitored due to the miR-199b-3p de ciency (Fig. 6F, H). Comparing with WT-STZ groups, the IHC and qPCR results in miR-199b-3p −/− -STZ groups showed an increased expression of TGFβ and a decreased expression of E-cadherin, which was consistent with pathological results (Fig. 6I-J, L-M, Fig. S4B). However, the higher level of KDM6A was expressed in diabetic kidney tissues without miR-199b-3p existence, no matter in transcription or translation level (Fig. 6K, N, Fig. S4B).
Summarily, based on those in vivo results, we evidently found the absence of miR-199b-3p was closely associated with the exacerbation of kidney injury in STZ-induced diabetic mice by abolishing the repression of KDM6A, indicating that miR-199b-3p played a protective prole in kidney injury of diabetic nephropathy through downregulating KDM6A.

Discussion
Despite numerous studies, the therapeutic interventions in preventing diabetes nephropathy are still not enough, which means more potential pathways for participating in the progression of diabetes nephropathy have yet to be explored. Here, our study demonstrates that the upregulation of KDM6A in diabetic kidney can inhibit the E-cadherin expression and trigger the epithelial-mesenchymal transition and brosis of renal tubular epithelial cell in diabetic kidney and eventually lead to diabetes nephropathy.
However, we nd a new role of miR-199b-3p in DN, which exhibit the characteristic of protecting the kidney from diabetic-induced injury by repressing the KMD6A expression.
Many studies have suggested that epithelial-mesenchymal transition plays an important role in renal brosis which is a pathological feature of kidney disease containing diabetic nephropathy (5)(6)(7)(8). In the present work, we found that KDM6A could promote the progression of EMT in HK-2 cells induced by HG. Similarly, a positive role of KMD6A in regulating the migration and invasion of hematopoietic stem cells and breast cancer cells, usually caused by EMT, indicated that KMD6A might positively regulate the EMT in those cell types (34,35). Our results from in vivo experiments expounded that the inhibition of KDM6A could decrease the degree of EMT and brosis of diabetic kidney and ease the symptoms of kidney injury. Coincidentally, Lin CL et al also found the suppression of KDM6A in STZ-induced mice showed a protective role in kidney injury. They further demonstrated that KDM6A aggravate DN through disturbing podocyte function by increasing KLF10 which could inhibit nephrin expression (17). Chen H et al showed that overexpression of KDM6A accelerated the progression of diabetic kidney diseases (16). Furthermore, we found that the expression of E-cadherin, a key regulative gene of EMT, would decrease with the increased of KDM6A, whereas KDM6A expression was not affected by the change of E-cadherin expression, indicating that E-cadherin might be the downstream target of KDM6A. However, the role of KDM5A in EMT progression remains controversial. Van den Beucken et al showed that inhibition of KDM6A promote EMT and cancer stemness (36) and Zhou Z et al showed that downregulation of KDM6A repressed the E-cadherin expression (37), which could be attributed to increasing histone H3K27 methylation. Considering multiple performance of KDM6A during EMT in different diseases or cell types, KDM6A may regulate the EMT process in many different pathways. According to that, we hypothesize that other transcription factors might be existed which improve the E-cadherin expression and was repressed by KDM6A. To con rm the hypothesis, more investigation needs to be done in the further.
Many studies showed that miR-199b-3p was expressed in many kinds of tissue and most of them displayed a functional role in different cancer. Graham ME et al showed miR-199b-3p in the serum of patients with papillary thyroid cancer was downregulated dramatically (38). Wang X et al found miR-199b-3p was one of the key miRNA related to hepatocellular carcinoma (39). Koshizuka K et al showed that miR-199b-3p inhibited cancer cell migration and invasion in head and neck cancer by regulating ITGA3 (23). Similarly, Sakaguchi T et al also found miR-199b-3p functioned as a tumour suppressor in bladder cancer by targeting ITGA3 (40). In addition to cancer, miR-199b-3p also play an important role in other disease. The apoptosis of cerebral microvascular endothelial cells was repressed via upregulation of miR-199b-3p in ischemic stroke (25). Wu G et al showed miR-199b-3p was substantially increased in extracellular vesicles from nasal mucus from patients with Allergic Rhinitis (41). Dolz S et al also found miR-199b-3p presented remarkablely higher expression in patients with asymptomatic carotid artery stenosis progression (42). However, the role of miR-199b-3p on diabetic nephropathy is unclear. As described in this study, KDM6A has been reported to be regulated by several miRNAs in kidney tissue. We, for the rst time, de ned KDM6A as a target of miR-199b-3p in diabetic kidney, which may provide new insights into the mechanisms of diabetic nephropathy progression. Although miR-199b-3p was proved to relieve the diabetes-induced kidney injury through inhibiting KDM6A expression here, its upregulation was observed in HG-induced HK-2 cell and kidney tissues of STZ-induced mice, which might be explained by the compensation for the higher expression of KDM6A.

Conclusion
We here provide evidence that KDM6A promotes the epithelial-mesenchymal translation and brosis of renal tubular epithelial cells in kidney of diabetic nephropathy through downregulating the expression of E-cadherin and eventually results in severe kidney injury, while miR-199b-3p rescues the renal damage by repressing KDM6A expression. In the future, miR-199b-3p may be a potential biomarker or target for the diagnosis and treatment for diabetic nephropathy.      Knockout of miR-199b-3p aggravated renal dysfunction in mice. (A-D) Graphs showing the results of ACR, BUN, Cr, proteinuria in different treated mice at 8 weeks after the onset of diabetes. C57BL/6 mice without any treatment (WT-NC), C57BL/6 mice treated with STZ (WT-STZ), miR-199b-3p knockout C57BL/6 mice without any treatment (miR-199b3p-/--NC), miR-199b-3p knockout C57BL/6 mice treated with STZ (miR- immunohistochemical staining results in I-K respectively. n=10. Mean ± standard error of the mean values are presented. *P < 0.05, **P < 0.01, ***P < 0.005, Student's t test.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.