Adipose mesenchymal stem cell-derived extracellular vesicles containing microRNA-26a-5p target TLR4 and protect against diabetic nephropathy

Diabetic nephropathy (DN) is a complication of diabetes that is increasing in prevalence in China. Extracellular vesicles (EVs) carrying microRNAs (miRs) may represent a useful tool in the development of therapies for DN. Here, we report that EVs released by adipose-derived mesenchymal stem cells (ADSCs) during DN contain a microRNA, miR-26a-5p, that suppresses DN. Using bioinformatic analyses, we identified differentially expressed miRs in EVs from ADSCs and in DN and predicted downstream regulatory target genes. We isolated mesenchymal stem cells (MSCs) from adipose tissues and collected EVs from the ADSCs. We exposed mouse glomerular podocytes and MP5 cells to high glucose (HG), ADSC-derived EVs, miR-26a-5p in-hibitor/antagomir, Toll-like receptor 4 (TLR4) plasmids, or the NF- k B pathway activator (phorbol-12-myristate-13-acetate, or PMA). We used the cell counting kit-8 (CCK-8) assay and flow cytometry to investigate the impact of miR-26a-5p on cell viability and apoptosis and validated the results of these assays with in vivo experiments in nude mice. We found that in DN, miR-26a-5p is expressed at very low levels, whereas TLR4 caused by DN. EVs derived from ADSCs enter mouse glomerular podo- cytes, MP5 cells. MiR-26a-5p contained in the ADSC-derived EVs targets and inhibits TLR4, inactivating the NF- k B/VEGFA pathway in HG-induced MP5 cells, characterized by decreased IKK b , I k B a , and p65 protein expression and a decrease in the extent of the phosphorylation of those proteins. This leads to enhanced viabilityandsuppresses apoptosis ofMP5cells. M., and F. C. writing-review and editing; Q. L. and X. Z. investigation; Y. W., Y. M., F. C., and J. S. resources; Y. W., Y. M., F. C., and J. S. data curation; Y. W., Y. M., F. C., and J. S. soft-ware; Y. W., Y. M., F. C., and J. S. formal analysis; Y. W., Y. M., and F. C. methodology; J. S. funding acquisition.

Diabetic nephropathy (DN), a common microvascular complication of diabetes, is one of the major causes of death among patients with diabetes (1). It is also the leading cause of both chronic kidney and end-stage renal diseases worldwide (2). The incidence of DN is increasing in developing countries, including China (3). DN is characterized by albuminuria and progressive loss of kidney function (4), and risk factors include hyperglycemia and genetic polymorphisms (5). Moreover, the dysfunction of podocytes contributes to the development of DN (6), and DN progression can be slowed by reducing podocyte injury (7). Current treatment methods for DN mainly rely on the management of hyperglycemia and blood pressure (8). Evidence suggests that extracellular vesicles (EVs) slow the development of DN by protecting podocytes from injury (9). Elucidating mechanisms by which EVs regulate podocyte injury may represent a therapeutic target for the treatment of DN.
EVs are nano-sized membrane vesicles that include exosomes and microvesicles, and they are released by many cells types, including mesenchymal stem cells (MSCs) (10,11). MSCs are multipotent stem cells that differentiate into a variety of lineages and exert important functions in bone regeneration and repair (12). A recent study highlighted the role of exosomes derived from adipose-derived mesenchymal stem cells (ADSCs) in protecting against the development of DN (13). EVs are known to contain mRNA, microRNAs (miRs), and long noncoding RNAs, thereby transferring genetic information from one cell to the next (14). Interestingly, miR-26a-5p serves as a candidate biomarker for DN by a meta-analysis of profiling studies (15). Our results also showed that miR-26a-5p is highly expressed in EVs derived from MSCs. Loss of miR-26a led to the development of DN in both cultured podocytes and streptozotocin-induced diabetic mice (16). Additionally, miR-26a-5p was suggested to be one of the therapeutic mediators in DN (15). Based on these findings, we hypothesized that EVderived miR-26a-5p could regulate DN. Toll-like receptor 4 (TLR4), which serves as a signaling receptor for lipopolysaccharides and as a crucial regulator of the innate immunity system, was identified as a downstream target gene of miR-26a-5p (17). One study used M4200 cells to show that TLR4 also participates in the pathogenesis of DN (18). Moreover, downregulation of TLR/NF-kB (NF-kB) using umbelliferone ameliorated renal function in a rat model of DN (19). Notably, the activation of NF-kB is thought to be related to the inflammation and disease development that are associated with DN (20). Upregulation of NF-kB by TLR4 was previously reported in rat pineal glands and human umbilical vein endothelial cells (21,22). In addition, the NF-kB pathway could promote the expression of vascular endothelial growth factor A (VEGFA) (23), which contributes to glomerular endothelial cell dysfunction as well as albuminuria in DN (24). Taking these findings into consideration, we hypothesized that ADSC-derived EVs transferred miR-26a-5p to other cells regulating the progression of DN by targeting TLR4 and modulating NF-kB and VEGFA. Thus, this study may help to identify a novel strategy for the control of DN.

Results
Bioinformatics analyses for the expression of miR-26a-5p and TLR4 in DN Analysis of the EVmiRNA database revealed that the expression of miR-26a-5p in the EVs derived from MSCs was significantly upregulated (Fig. 1A). Furthermore, the miRanda database and Starbase database were used to predict the downstream target genes of differentially expressed miRNAs. These results were then compared with a diabetes-related gene expression data set (GSE21340). Five genes were identified at the intersection of these analyses (SMAD1, CELF2, SMAD6, PYGL, and TLR4) (Fig. 1B). Because the miRanda database predicted a binding site between miR-26a-5p and TLR4 (Fig. 1C), we chose to investigate the expression of TLR4 in the GSE21340 data set. We found that TLR4 expression was significantly higher in DN (Fig. 1D). These results suggest that miR-26a-5p delivered by ADSC-derived EVs protect against DN by regulating TLR4.
Next, EVs isolated from medium used to grow ADSCs were characterized. Dynamic light scattering (DLS) analysis indicated that EVs range in diameter from 30-150 nm. Transmission EM revealed the morphology of EVs to be cup-shaped or spherical. In addition, Western blot analysis of marker proteins . DN is characterized by poor miR-26a-5p expression and high TLR4 expression. A, the expression of miR-26a-5p in EVs. B, the intersection between predicted miR-26a-5p downstream target genes identified by the MiRanda and SatrBase databases and differentially expressed genes from the GSE21340 data set. The number represents the predicted number of target genes. C, a map of the predicted binding site of miR-26a-5p to TLR4 and mutant site. D, the box map for the differential expression of TLR4 gene in diabetes and normal samples. Paired t test was used to compare normal sample and diabetic samples.
of EVs confirmed the expression of CD63 and TSG101, whereas calnexin was absent, indicating that EVs had been successfully isolated (Fig. 2, C-E). The above results demonstrated the successful isolation of EVs from ADSCs.
EVs from ADSCs alleviated the pathological symptoms and suppressed glomerular podocyte apoptosis in a mouse model that spontaneously develops diabetes To evaluate the mouse model of spontaneous diabetes, we compared the levels of protein found in the urine (within 24 h), serum creatinine (Scr), and blood urea nitrogen (BUN) in C57BL/KsJ db/db spontaneous diabetic mice and C57BL/KsJ db/m control mice. The spontaneously diabetic mice had twice the amount of urine protein and 3 times as much Scr and BUN as the control mice (Fig. 3, A-C). The histopathological phenotype was characterized by periodic acid Schiff (PAS) staining of DN tissues. Spontaneously diabetic mice showed an accumulation of extracellular matrix in their kidney tissues, viability of mesangial cells, and a thickening of basement membrane. In addition, there was increased proteinuria in the lumen, hyaline degeneration, and severe vacuolar degeneration of renal tubular epithelial cells in these mice compared with that of control mice, indicating the successful establishment of the DN cell model (Fig. 3D).
To investigate the effect of EVs from ADSCs on spontaneous diabetic mice, we injected ADSC-derived EVs or PBS into mice via the tail vein, and then changes of urine protein, Scr, and BUN levels were subsequently measured, followed by PAS staining to detect histopathological changes. The results illustrated that treatment with EVs derived from ADSCs notably reduced the levels of urine protein, Scr, and BUN in the DN mouse model and alleviated the histopathological changes associated with DN (Fig. 3, A-D).
Next, TUNEL staining was used to detect the apoptosis of glomerular podocytes in mice. Glomerular podocytes from spontaneously diabetic mice (C57BL/KsJ db/db) exhibited a significant increase in apoptosis compared with control mice (C57BL/KsJ db/m mice), and this was reduced by injection of EVs from ADSCs (Fig. 3E). In addition, apoptosis-related proteins (caspase-3, cleaved caspase-3, Bcl-2, Bax) from diabetic mice (C57BL/KsJ db/db) and control (C57BL/KsJ db/m) mice were compared. Bcl-2 protein was significantly reduced in diabetic mice compared with control mice, whereas the protein expression of Bax, cleaved caspase-3, and caspase-3 was increased. Injection of EVs from ADSCs reversed these trends (Fig. 3F). These results demonstrated that ADSC-derived EVs are capable of alleviating the pathological symptoms and inhibiting glomerular podocyte apoptosis in DN mice.

EVs from ADSCs inhibited apoptosis of MP5 cells induced by HG
The ability of mouse glomerular podocytes, MP5 cells, to internalize EVs from ADSCs was assessed using green fluorescent (PKH67)-labeled EVs. The presence of green fluorescence in MP5 cells was observed by fluorescence microscopy, which indicated that MP5 cells were able to internalize ADSC-derived  Flow cytometry was used to detect apoptosis at 48 h after the same treatments described above. The results showed that compared with treatment with either NG or MA, HG induction markedly increased MP5 cell apoptosis, whereas cotreatment of HG and ADSC-derived EVs exhibited reduced cell apoptosis versus HG treatment alone (Fig. 4C). In addition, Western blot analysis was used to detect the expression of apoptosis-related proteins in MP5 cells after 48 h of each treatment. Treatment with HG significantly decreased the protein expression of Bcl-2, whereas increasing that of caspase-3, cleaved caspase-3, and Bax compared with either NG or MA and cotreatment of HG and ADSCderived EVs increased Bcl-2 protein expression and de-creased that of caspase-3, cleaved caspase-3, and Bax (Fig. 4D). These results suggested that EVs from ADSCs are able to suppress HG-mediated injury of podocytes in vitro.   Role of EV-delivered miR-26a-5p in DN EVs from ADSCs delivered miR-26a-5p, which inhibited apoptosis in HG-treated MP5 cells The expression of miR-26a-5p in HG-induced MP5 cells was measured by RT-qPCR after treatment with ADSC-derived EVs or PBS. miR-26a-5p expression was significantly lower after induction with HG than with treatment with NG or MA, but it was notably increased by treatment with ADSC-derived EVs containing miR-26a-5p (Fig. 5A). To confirm that ADSCderived EVs can transfer miR-26a-5p into MP5 cells, Cy3-miR-26a-5p mimic was transfected into ADSCs and then added to MP5 cell media. Fluorescence microscopy revealed that MP5 cells became noticeably red after treatment (Fig. 5B). These results demonstrate that ADSC-derived EVs transfer miR-26a-5p into MP5 cells.
Next, MP5 cell viability was assessed using the CCL-8 assay, and apoptosis was measured by flow cytometry. Under HG conditions, treatment with either miR-26a-5p mimic or ADSCderived EVs markedly improved cell viability and suppressed apoptosis of MP5 cells, whereas ADSC-derived EVs treated with miR-26a-5p inhibitor led to decreased cell viability and increased apoptosis (Fig. 5, D and E).
In addition, Western blot analysis was used to detect apoptosis-related proteins in MP5 cells after each treatment. Treatment with miR-26a-5p mimic or ADSC-derived EVs resulted in increased Bcl-2 protein expression accompanied by a notable decrease in Bax, cleaved caspase-3, and caspase-3 expression. In contrast, ADSC-derived EVs treated with miR-26a-5p inhibitor exhibited markedly decreased Bcl-2 protein expression, whereas the expression of Bax, cleaved caspase-3, and caspase-3 was significantly elevated (Fig. 5F). These results indicate that delivery of miR-26a-5p into MP5 cells by ADSC-derived EVs inhibits HG-induced apoptosis.
RT-qPCR of MP5 cells transfected with miR-26a-5p mimic or inhibitor resulted in a noticeably increased miR-26a-5p expression and a significantly decreased TLR4 expression, whereas opposite effects were detected after treatment with miR-26a-5p inhibitor (Fig. 6, C and D). Western blot analysis revealed that HG treatment of MP5 cells significantly increased TLR4 protein expression compared with treatment with NG or MA treatment, and this could be reversed by treatment with EVs from ADSCs (Fig. 6E). This suggests that miR-26a-5p transferred by ADSC-derived EVs could directly target the TLR4 gene.
To detect the effect of the TLR4 gene regulated by ADSCderived EVs on HG-induced MP5 cells, HG-treated MP5 cells were overexpressed with TLR4 and then treated with ADSCderived EVs. First, Western blot analysis was used to detect the cellular protein expression of TLR4 upon each treatment. It was found that under HG conditions, compared with pcDNA-3.1 treatment, pcDNA-TLR4 significantly upregulated the protein expression of TLR4, which was reversed by treatment of pcDNA-TLR4-and ADSC-derived EVs (Fig. 6F).
Subsequently, CCK-8 assay was performed to detect MP5 cell viability, and flow cytometry was used to detect cell apoptosis. The results displayed that under HG conditions, relative to pcDNA-3.1 treatment, pcDNA-TLR4 significantly inhibited the protein cell viability while promoting cell apoptosis; however, promoted cell viability and inhibited cell apoptosis were detected after treatment with pcDNA-TLR4-and ADSCderived EVs ( Fig. 6G and H).
In addition, Western blot analysis was used to detect apoptosis-related proteins in HG-treated MP5 cells after treatment with each plasmid. pcDNA-TLR4 treatment markedly decreased Bcl-2 protein expression while significantly increasing cleaved caspase-3, caspase-3, and Bax protein expression relative to that of pcDNA-3.1 treatment. The opposite results were found after the treatment of HG-induced MP5 cells transfected with pcDNA-TLR4 and exposed to EVs from ADSCs containing miR-26a-5p (Fig. 6I). Combined with the above results, we concluded that miR-26a-5p delivered by ADSC-derived EVs targets TLR4, inhibiting HG-induced MP5 cell apoptosis.

EVs from ADSCs inhibited HG-induced apoptosis of MP5 cells by downregulating TLR4 and the NF-kB/VEGFA signaling pathway
Subsequently, we explored the role of ADSC-derived EVs in the progression of DN via the regulation of the NF-kB pathway and VEGFA. We first investigated whether miR-26a-5p carrying EVs from ADSCs could inhibit the NF-kB pathway by Role of EV-delivered miR-26a-5p in DN downregulating TLR4 expression. After TLR4 overexpression, MP5 cells induced by HG were further treated with ADSCderived EVs. Immunofluorescence staining revealed that HG treatment of MP5 cells resulted in p65 localization to the nucleus. EVs from ADSCs were able to reduce p65 nuclear localization, keeping more p65 in the cytoplasm. The effects of EVs could be negated by the overexpression of TLR4, which led to the nuclear localization of p65 protein again (Fig. 7A).
Western blot analysis compared the protein expression of NF-kB pathway-related proteins IKKb, IkBa, and p65 and the extent of their phosphorylation in the cells after each treatment. Treatment with EVs from ADSCs and pcDNA-3.1 control plasmid led to significantly decreased protein expression of IKKb, IkBa, and p65 and the extent of their phosphorylation during HG induction, whereas each of these proteins was upregulated by ADSCderived EVs cotreated with pcDNA-3.1-TLR4 (Fig. 7B).
To verify whether EVs from ADSCs can inhibit apoptosis associated with HG treatment by inhibiting the NF-kB pathway, HG-induced MP5 cells were treated with ADSC-derived EVs and the NF-kB pathway activator, phorbol-12-myristate-13-acetate (PMA; 1 mg/ml). Western blot analysis indicated that PMA treatment significantly increased the protein expression of IKKb, IkBa, p65, and VEGFA, as well as the extent of their phosphorylation in HG-induced MP5 cells compared with that of treatment with DMSO. Combined treatment with PMA and ADSC-derived EVs significantly decreased all these indicators compared with PMA treatment alone (Fig. 7C). PMA treatment also notably decreased cell viability, as measured by CCK-8 assay, and increased cell apoptosis, as measured by flow cytometry, relative to treatment with DMSO. This effect was reversed by treatment with PMA and ADSC-derived EVs combined (Fig. 7, D and E).
HG-induced MP5 cells were further treated with ADSCderived EVs after overexpression of VEGFA. VEGFA protein expression was significantly decreased after treatment with ADSC-derived EVs, whereas further overexpression of VEGFA upregulated the VEGFA protein expression (Fig. 7F). HGinduced MP5 cells treated with ADSC-derived EVs plus pcDNA-3.1 displayed increased viability and decreased apoptosis of MP5 cells compared with that of cells treated with PBS. This was reversed by treatment with ADSC-derived EVs and pcDNA-3-VEGFA (Fig. 7G). These data suggest that ADSC-    In panels B-F, *, p , 0.05 versus HG 1 mimic-NC; #, p , 0.05 versus HG 1 miR-26a-5p mimic or HG 1 Exo-inhibitor-NC. Data are expressed as mean 6 standard deviation. Data among multiple groups were compared using one-way ANOVA, followed by Tukey's post hoc test. The experiment was repeated three times.
derived EVs inhibit TLR4 expression, resulting in the downregulation of NF-kB/VEGFA and inhibition of HG-induced MP5 cell injury.
Delivery of miR-26a-5p by ADSC-derived EVs reduced the pathological symptoms and cell apoptosis in spontaneous diabetic mice To explore the effects of miR-26a-5p delivered by ADSCderived EVs in a mouse model of DN, ADSC-derived EVs and/ or miR-26a-5p antagomir were injected into C57BL/KsJ db/db mice via the tail vein. Mice were then observed for changes in urine protein, Scr, and BUN levels, and the histopathological changes of mice were detected using PAS staining. As illustrated, diabetic mice treated with PBS and antagomir NC displayed accumulated extracellular matrix in their kidney tissues, viability of mesangial cells, thickening of basement membrane, and nuclear lamina dispersion, with increased proteinuria in the lumen, hyaline degeneration, and severe vacuolar degeneration of renal tubular epithelial cells. ADSC-derived EVs and antagomir NC treatment led to decreased levels of urine protein, Scr, and BUN in DN modeled mice and alleviated histopathological changes compared with those of PBS and antagomir NC treatment. Treatment with ADSC-derived EVs and miR-26a-5p antagomir treatment markedly increased the levels of urine protein, Scr, and BUN while aggravating the histopath-ological changes in the mice compared with treatment with ADSC-derived EVs and antagomir NC (Fig. 8, A and B). TUNEL staining showed that treatment with both ADSCderived EVs and antagomir NC significantly reduced cell apoptosis compared with that of DN mice treated with PBS and antagomir NC. Apoptosis was notably increased in DN mice treated with both ADSC-derived EVs and miR-26a-5p antagomir compared with treatment with ADSC-derived EVs and antagomir NC (Fig. 8C).
RT-qPCR was used to analyze miR-26a-5p and TLR4 expression in the renal tissues of mice in response to different treatments. Treatment with ADSC-derived EVs and antagomir NC significantly increased miR-26a-5p expression and decreased TLR4 expression compared with PBS plus antagomir NC treatment; however, treatment with ADSC-derived EVs plus miR-26a-5p antagomir decreased miR-26a-5p expression and increased TLR4 expression compared with treatment with ADSC-derived EVs and antagomir NC (Fig. 8D).
Western blot analysis was used to detect the levels of apoptosis-related proteins TLR4 and VEGFA as well as NF-kB pathway-related proteins. Treatment with ADSC-derived EVs and antagomir NC led to marked decreases in Bax, cleaved caspase-3, caspase-3, IKKb, IkBa, p65, and VEGFA protein expression as well as in the extent of IKKb, IkBa, and p65 phosphorylation, accompanied by an increase in the protein expression of Bcl-2 compared with PBS and antagomir NC treatment. These changes were reversed upon treatment with ADSC-derived EVs and miR-26a-5p antagomir (Fig. 8E). Overall, delivery of miR-26a-5p by ADSC-derived EVs significantly reduces the pathological symptoms and cell apoptosis in spontaneously diabetic mice.

Discussion
DN is a fatal complication of diabetes characterized by metabolic disorders and chronic inflammation (25). Increased diabetic podocyte dysfunction aggravates diabetes-induced kidney injury (26). Because EVs have been shown to confer protection against renal diseases (27), we studied the effects of miR-26a-5p delivery by EVs derived from ADSCs on the pathological symptoms of DN. Delivery of miR-26a-5p by ADSC-derived EVs alleviated several pathological symptoms associated with DN.
Initially, we found that miR-26a-5p was highly expressed in EVs obtained from adipose tissue-derived MSCs; however, miR-26a-5p was poorly expressed and TLR4 was highly expressed in DN. As previously reported, miR-26a-5p is highly expressed in exosomes secreted from ADSCs (28). Similarly, a previous study found that miR-26a expression was low in biopsy samples from patients with DN (29). Moreover, metaanalysis revealed that miR-26a-5p expression was downregulated in DN (15), consistent with our results. In addition, high expression of TLR4 was detected in the glomerular basement membrane, proximal convoluted tubule, and the renal interstitial area of the kidney from rats with DN (30). Finally, another study also observed upregulation of TLR4 in db/db mice and SV40-MES-13 cells compared with that of the normal controls (31).
We subsequently demonstrated that EVs derived from ADSCs were able to alleviate the pathological symptoms of DN, as evidenced by a reduction in protein urine, Scr, and BUN, and suppress glomerular podocyte apoptosis in spontaneous diabetic mice. EVs also prevented apoptosis in HG-induced MP5 cells characterized by a reduction in Bax, cleaved caspase-3, and caspase-3 and increased Bcl-2. Transplantation of ADSCs has been highlighted as a potential therapy for DN (32). Consistent with our results, stem cell-derived EVs were capable of suppressing fibrosis in a mouse model of DN (33). In addition,   ADSC-derived EVs alleviated streptozotocin-induced DN in rats (34). Furthermore, this study found that ADSC-derived EVs alleviated signs of DN by delivering miR-26a-5p to nearby cells. In line with our findings, exosomal miR-26a inhibited muscle wasting and renal fibrosis in a mouse model of obstructive kidney disease (35) and suppressed cardiac fibrosis in Role of EV-delivered miR-26a-5p in DN chronic kidney disease (36). Notably, ADSC-derived exosomes transferred miR-181-5p to injured liver cells, preventing liver fibrosis (37). Urinary protein, blood urea nitrogen, and serum creatinine levels were measured to determine the kidney function, and attenuation of DN was associated with reduced levels of Scr, BUN, and urine protein in the streptozotocin-induced DN rat (38,39). At the same time, accelerated apoptosis and renal injury in diabetic rats were associated with decreased Bcl-2 and increased BAX cleaved caspase-3 levels (40,41). These reports support our findings that delivery of miR-26a-5p into MP5 cells by ADSC-derived EVs suppressed the progression of DN. Furthermore, we found that miR-26a-5p delivered by ADSCderived EVs targeted TLR4 in HG-induced MP5 cells, inactivating the NF-kB/VEGFA pathway. Interestingly, upregulated TLR4 increases the susceptibility of diabetic rats to acute kidney injury following myocardial infarction (42). Moreover, reduced expression of TLR4 and NF-kB nuclear translocation caused by catalpol expression suppressed HG-induced podocyte injury (43). DN progression is characterized by the activation of NF-kB in tubular epithelial cells (44), and silencing of VEGA attenuated podocyte injury induced by HG (9). Together, these reports support our findings that TLR4 plays a role in the pathological progression of DN. Additionally, inhibition of TLR4 expression downregulated NF-kB in vitro, thereby alleviating the development of atherosclerosis (45). Therefore, it could be concluded that miR-26a-5p, when delivered by ADSC-derived EVs into MP5 cells, targets TLR4, thereby inactivating the NF-kB/VEGFA signaling pathway and regulating MP5 cell injury.
Our findings demonstrated that delivery of miR-26a-5p by ADSC-derived EVs alleviates the pathological symptoms of DN by targeting TLR4, which subsequently inactivates the NF-kB/ VEGFA signaling pathway (Fig. 9). This suggests a protective role for miR-26a-5p-containing EVs derived from ADSCs during DN, providing a novel therapeutic target for the prevention of DN in the clinical setting. Moreover, this study was the first to discover the effect of the miR-26a-5p/TLR4 axis in DN. Nevertheless, further studies based on human specimens are required to confirm the underlying molecular mechanism.

Ethics statement
This study was approved by the medical ethics committee of Huaihe Hospital, Henan University. All animal experiments were performed according to the provisions of the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health (46).

Bioinformatic analysis
MiRs present in MSC-derived EVs were obtained from the EVmiRNA database (RRID:SCR_018795). The diabetes-related gene expression data set GSE21340 was downloaded from the GEO database (RRID:SCR_005012), which included 10 normal samples and 5 diabetes samples. Differential expression analysis was conducted using the R language "limma" package (47) using normal samples as controls. The threshold for differential expression analysis was set at a log 2 fold change of .1 and p value of ,0.05. Downstream targets genes of different miRs were predicted using the miRanda database (RRID:SCR_017496) and the Starbase database (RRID:SCR_ 016303), and the intersecting genes were overlaid with the differential expression results from the GSE21340 data set. A map for binding sites between miRs and genes was delineated after searching the above miR database. The primers for miR-26a-5p inhibitor, miR-26a-5p mimic, and pCDNA-TLR4 were designed and synthesized by GenePharma (Shanghai, China), whereas the miR-26a-5p antagomir was purchased from Biomics Biotechnologies (Jiangsu, China).

Animal treatment
C57BL/KsJ db/m male mice aged 7 weeks and weighing 18-22 g (control mice, n = 6) and C57BL/KsJ db/db mice (a total of 18 spontaneous diabetic mice assigned into three groups, n = 6) were purchased from Cavens Laboratory Animal (SCX, 2016-0010, Changzhou, China). Three to five mice were placed in each cage, with controlled, alternating 12-h day/night cycles and temperature of 22 6 2°C, with free access to food and water. When reared to the age of 12 weeks, mice were subjected to characterization of DN phenotype by histopathological  , 25 mm). D, the expression of miR-26a-5p and TLR4 in MP5 cells as detected by RT-qPCR. E, the protein expression of apoptosis-related proteins (caspase-3, cleaved caspase-3, Bcl-2, Bax) in renal tissues of mice as detected by Western blot analysis, normalized to GAPDH. Data are expressed as mean 6 standard deviation. Data among multiple groups were compared using one-way ANOVA, followed by Tukey's post hoc test. The experiment was repeated three times. *, p , 0.05 versus PBS 1 antagomir-NC. #, p , 0.05 versus ADSCs-Exo 1 antagomir-NC. examination using periodic acid Schiff (PAS) staining according to the published literature (13). At 13 weeks of age, mice underwent injection with either phosphate-buffered saline (PBS), ADSC-derived EVs, or miR-26a-5p antagomir via tail vein for a 12-week therapy.

Cell isolation and culture
The mice were anesthetized with pentobarbital (200 mg/kg i.p.), and the subcutaneous adipose tissues of the groin were extracted. ADSCs were isolated after collagenase digestion. In short, the fat tissues were cut and washed twice with PBS buffer and centrifuged at 1200 rpm for 10 min. Following the removal of the supernatant, prepared collagenase was added to the precipitates. Detachment was allowed to proceed at 37°C for 40 min and was terminated by the addition of complete medium. Cells were then filtered using a 40-mm cell filter, centrifuged at 1500 rpm for 8 min, resuspended with complete Roswell Park Memorial Institute 1640 medium, and finally cultured in a humidified incubator at 37°C with 5% CO 2 .
In the miR-26a-5p inhibitor experiment, ADSCs were transfected with an miR-26a-5p inhibitor or NC for 48 h, and then ADSC-derived EVs were isolated for further experimentation.

In vitro differentiation of pluripotent ADSCs
Osteoblast differentiation was induced as follows. ADSCs (1 3 10 5 cells/well) were inoculated into a 6-well culture plate, and once they were adherent to the wall, the cells were cultured with osteoblast culture medium (prepared with DMEM supplemented with 0.17 mM vitamin C, 0.5% FBS, 10 mM b-glycerophosphate, 100 nM dexamethasone, and 1% penicillin/streptomycin, all purchased from Sigma-Aldrich Chemical Company [St. Louis, MO, USA]). The cells were cultured at 37°C with 5% CO 2 for 21-28 days, with the medium being replaced every 2 days. After calcified nodules were observed under a light microscope, the cells were washed twice with PBS, fixed with 4% paraformaldehyde for 15 min, washed twice with deionized water, and stained with Alizarin Red S for 30 min. After staining, cells were washed twice with deionized water and observed under an optical microscope.
Adipocyte differentiation was induced as follows. ADSCs (1 3 10 5 cells/well) were inoculated into a 6-well culture plate, and the cells were cultured with adipogenic differentiation culture medium composed of DMEM supplemented with glutamine, 10% FBS, 1 mM rosiglitazone, 1 mM dexamethasone, 0.5 mM 3-isobutyl-1-methylxanthine, 10 mg/ml insulin, 0.2 mM indomethacin, and 1% penicillin/streptomycin. After 3 days,  Role of EV-delivered miR-26a-5p in DN the medium was replaced every other day with low-glucose DMEM containing glutamine, 10% FBS, 1% penicillin/streptomycin, 1 mM rosiglitazone, and 10 mg/ml insulin. The cells were cultured at 37°C with 5% CO 2 for 21-28 days. When the fat droplets were obviously visible under optical microscopy, the cells were washed with deionized water, fixed with 4% paraformaldehyde for 15 min, rinsed with PBS, and then incubated with 65% isopropanol for 4 min. The cells were then stained with new, filtered oil red O for 4 min, washed with deionized water, counterstained with hematoxylin, and observed using an optical microscope.
Chondroblast differentiation was subsequently induced. ADSCs (2 3 10 6 cells/tube) were cultured in a 15-ml centrifuge tube at 37°C with 5% CO 2 for 24 h and then incubated in chondrogenic medium composed of DMEM (4.5 g/liter glucose) supplemented with 100 nM dexamethasone, 0.35 mM proline, 0.17 mM vitamin C, 1 mM sodium pyruvate, 1% insulin-transferrin-selenium, 10 ng/ml transforming growth factor beta 3 (TGFb-3), and 1% penicillin/streptomycin (Sigma-Aldrich Chemical Company, St. Louis, MO, USA) at 37°C with 5% CO 2 for 21-28 days, during which the medium was replaced every 2 days. After the cells were cultured into spheres with a diameter of 1.5-2 mm, the cells were sliced and stained with Alcian blue. Briefly, the cells are washed twice with deionized water, fixed with 4% paraformaldehyde for 15 min, and cut into 10-mm frozen sections. Finally, the sections were stained with 0.5% Alcian blue dye for 30 min and observed under an optical microscope.

Isolation of EVs
At the third passage, ADSCs were inoculated into a 6-cm culture dish and cultured at 37°C with 5% CO 2 . When the cell density reached 70-80%, the cells were cultured in fresh medium containing EVs-free FBS for 24 h, followed by ultracentrifugation at 100,000 3 g for 18 h and filtration using a 0.22-mm filter to obtain EV-free FBS. After 24 h of culture, conditioned medium was collected and EVs were extracted by ultracentrifugation. In short, the cell culture medium was centrifuged at 300 3 g for 10 min, at 2000 3 g for 15 min, and at 12,000 3 g for 30 min, respectively, to remove floating cells and cell fragments. Following filtration using a 0.22-mm filter, the supernatant was collected and ultracentrifuged at 100,000 3 g at 4°C for 2 h, washed with PBS, ultracentrifuged for the second time under the same conditions, and resuspended with PBS. ADSCderived EVs were immediately used in subsequent experimentation or stored at 280°C (48).

Uptake test of EVs
To determine the uptake of ADSC-derived EVs by MP5 cells, EVs were labeled with green fluorescent dye (PKH67, Sigma-Aldrich Chemical Company [St. Louis, MO, USA]) and incubated with MP5 cells at 37°C for 3 h. Following several PBS rinses, the cells were fixed with 4% paraformaldehyde for 15 min, washed with PBS, and then stained with 49,6-diamidino-2-phenylindole (0.5 mg/ml; Invitrogen, Carlsbad, CA, USA). The green fluorescence in MP5 cells was observed under a fluorescence microscope. To detect whether miRs can be transferred from EVs into MP5 cells, MP5 cells treated with ADSC-derived EVs for 3 h were collected for RNA extraction, and miR expression was detected by RT-qPCR.

Determination of urine protein, Scr, and BUN levels
The urine of mice was collected after 12 weeks of treatment and then analyzed using a BCA kit. In short, BCA working solution was mixed with urine, followed by incubation at 37°C for 30 min. Optical density values were detected at 562 nm. After treatment with EVs, the plasma was collected via the abdominal aorta, kept at room temperature for 2 h, and centrifuged at 3500 rpm for 5 min. The levels of Scr and BUN were measured using detection kits (Scr, C011-1, Jiang Cheng Biotechnology, Nanjing, China; BUN, C013-2, Jiang Cheng Biotechnology, Nanjing, China).

Cell transfection
ADSCs were transfected with inhibitor NC, miR-26a-5p inhibitor, pcDNA-3.1, pcDNA-TLR4, and pcDNA-VEGFA. miR-26a-5p inhibitor and inhibitor NC were synthesized by Invitrogen (Carlsbad, CA, USA); pcDNA-3.1/TLR4/VEGFA plasmids were purchased from GenePharma (Shanghai, China). The experiment was carried out according to the instructions of the Lipofectamine 2000 kit (Invitrogen, Carlsbad, CA, USA). In brief, cells were inoculated into a 6-well plate 24 h before transfection. After 6 days of culture, cells were resuspended in conditioned medium and then seeded into a new 6-well plate, followed by transfection according to the above grouping. After transfection, the cells were incubated at 37°C with 5% CO 2 for 6-8 h and further cultured for 24-48 h with complete medium after centrifugation.

Cellular immunofluorescence assay
Cells were fixed with 4% paraformaldehyde for 5 min after they were half-dried. Next, the formaldehyde was removed, and 0.5% Triton X-100 prepared with PBS was used to clear the cells for 15 min. Next, the cells were sealed with 1% BSA at room temperature for 30 min. After removal of the sealing solution, the cells were incubated on a coverslip along with the primary antibody against p65 (no. 8242, 1:200, Cell Signaling Technologies [CST], Beverly, MA, USA) at 4°C overnight and then with the fluorescence-labeled secondary antibody (ab250115, Abcam Inc., Cambridge, MA, USA) at room temperature for 1 h in the dark. Finally, blocking agent (no. MM1402-5ML, MK Bio Science Co., Ltd., Suwon-si, Gyeonggi-do, Korea) containing 49,6-diamidino-2-phenylindole was used to seal the coverslip, and the cells were observed under a fluorescence microscope (Olympus Inc., Tokyo, Japan).

Flow cytometry
Flow cytometry was used to detect the representative surface markers of ADSCs. In short, ADSCs were detached with trypsin to form a single-cell suspension, washed by PBS (without calcium and magnesium), and then blocked with 10% normal goat serum to prevent nonspecific binding. Afterward, a series of monoclonal antibodies to CD14, CD19, CD29, CD34, CD44, CD45, CD73, CD90, HLA-A, -B, and -C, and HLA-DR (1:100, BioLegend, San Diego, CA, USA) labeled with FITC dye were incubated along with the cells for 30 min. A blank control and an FITC single-step staining control were set up, followed by a PBS rinse. Finally, the cells were resuspended with 10% normal goat serum and analyzed with a CyAn ADP analyzer (Beckman Coulter Life Sciences, Brea, CA, USA).
To assess apoptosis, cells were detached with 0.25% EDTAfree trypsin, centrifuged at 1500 rpm for 5 min, resuspended to precipitate, and washed with PBS. After centrifugation at 1500 rpm for 5 min, the cells were resuspended again with 300 ml binding buffer, followed by incubation with 5 ml annexin V-FITC at room temperature in the dark for 5 min. Subsequently, the cells were stained with 10 ml propidium iodine for 10 min, after which the apoptosis was detected using a flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

CCK-8 assay
MP5 cells (3000 cells/well) treated with HG or ADSCderived EVs were inoculated into 96-well plates, followed by incubation for 24, 48, and 72 h. Next, 10 ml CCK-8 reagent was added into each well and incubated for 2 h. The optical density at 450 nm was detected using a microplate reader (Multiskan MK3; Thermo Fisher Scientific, Rockford, IL, USA).

TUNEL staining
In the detection of apoptosis, tissue sections were evaluated according to the reagent specifications (Biobox, BA27). In short, the samples were labeled with DNA fragments in TUNEL solution containing terminal deoxynucleotidyltransferase and streptavidin-FITC. After washing, cells were incubated with paraformaldehyde-conjugated anti-FITC in a dark room at 37°C for 30 min, followed by addition of DAB reaction solution to the slide and 49,6-diamidino-2-phenylindole staining. Following hydrochloric acid alcohol differentiation and ethanol dehydration, the samples were incubated with xylene and sealed with neutral resin. The images were processed and analyzed under a microscope (IX53, Olympus Inc., Tokyo, Japan).

RT-qPCR
Trizol reagent (Invitrogen, Carlsbad, CA, USA) was used to extract total RNA from cultured cells or tissues. Complementary DNA (cDNA) was synthesized from 1 mg total RNA using a Revert Aid first-strand cDNA synthesis kit (Fermentas, Burlington, Ontario, Canada). RT-qPCR analysis then was carried out with the use of the SYBR Premix ExTaq TM II in an ABI PRISM ® 7900HT system (Takara Holdings Inc., Kyoto, Japan). GAPDH was used as an internal reference for determining relative mRNA expression using relative quantification (2 2DDCt method). The PCR primers used in this study are shown in Table 1. MiRs delivered by EVs were isolated using a SeraMir exosome RNA purification kit (System Biosciences, Mountain View, CA, USA). By following the instructions provided by the TaqMan microRNA assist kit (Applied Biosystems, Foster City, CA, USA), miR was converted to cDNA. The FastStart Universal SYBR Green master mix (Roche, Indianapolis, IN, USA) was used for the RT-qPCR reaction, along with the miR-specific forward primers provided by Sangon Biotechnology (Shanghai, China) and general reverse primers provided by the TaqMan microRNA assay kit. The expression of miRs was normalized to U6.

Statistical analysis
Data analysis was performed utilizing SPSS 21.0 (IBM Corp., Armonk, NY, USA). Data are expressed as mean 6 standard deviation. The data between two groups obeying normal distribution and homogeneous variance in paired design were compared using a paired t test and those in unpaired design using unpaired t test. Data among multiple groups were compared using one-way analysis of variance (ANOVA), with a Tukey's test conducted for post hoc analysis. Data among multiple groups at different time points were compared by repeated-measures ANOVA, followed by Bonferroni post hoc test. Pearson's correlation analysis was applied for correlation analysis. The Kaplan-Meier method was used to calculate the survival rate of patients, and log-rank test was employed for single-factor analysis. A p value of ,0.05 indicated a statistically significant difference.

Data Availability
All the data are in the manuscript.