Inhibition of miR-200c Restores Endothelial Function in Diabetic Mice Through Suppression of COX-2

Diabetes affects 9.5% of the adult population worldwide (1). Most patients with diabetes die of cardiovascular complications, including coronary heart disease, stroke, and nephropathy (2,3). Endothelial dysfunction associated with reduced nitric oxide bioavailability is one of the important initiators of vascular pathogenesis leading to the development of diabetic cardiovascular events (4). Hyperglycemia decreases the bioavailability of endothelium-derived relaxing factors such as nitric oxide but increases the production of endothelium-derived contracting factors, thereby contributing to endothelial dysfunction (5,6). However, the molecular mechanisms for the hyperglycemia-mediated disrupted balance between endothelium-derived relaxing factors and endothelium-derived contracting factors remain largely unclear.

microRNAs (miRNAs) are the small noncoding RNAs that bind to sequence-specific mRNA and consequently inhibit translation (7). Growing evidence has pinpointed miRNAs as important modulators of cardiovascular health and disease. For instance, miRNAs participate in vascular inflammation (8), arterial remodeling (9), smooth muscle plasticity (10), atherosclerosis (11), and endothelial cell apoptosis (12). Nevertheless, the role of miRNAs in the development of diabetic endothelial dysfunction is sparsely studied.

The miR-200 family, which comprises miR-200c, -200a, -200b, -141, and -429, plays a role in the development of diabetes complications. Downregulation of miR-200b increases vascular endothelial growth factor expression, promotes angiogenesis, and ameliorates diabetic retinopathy (13), whereas elevated miR-200b induces inflammation by promoting the expression of cyclooxygenase-2 (COX-2) and MCP-1 in vascular smooth muscle cells (14). In diabetic mouse glomeruli and renal mesangial cells, an increased miR-200 family (miR-200b and miR-200c) may promote the expression of fibrotic genes in diabetic nephropathy (15). Our initial screening of miRNAs showed a marked increase of miR-200c expression in db/db mouse aortas. However, the exact involvement of miR-200c in diabetic endothelial dysfunction has never been investigated to our knowledge. The current study, therefore, hypothesized that miRNA-200c plays a critical role in diabetes-associated endothelial dysfunction and that inhibition of miR-200c is effective as a novel intervention against diabetic vasculopathy.

C57BL/6 mice, db/db, and db/m⁺ mice were purchased from the Laboratory Animal Center of The Chinese University of Hong Kong (CUHK). COX-2^−/− mice were supplied by The University of Hong Kong. All animal experiments were approved by the CUHK Animal Experimentation Ethics Committee and performed in compliance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication, eighth edition, updated 2011).

All collection and treatment of human samples were conducted under the guidelines established by the Joint CUHK-New Territories East Cluster Clinical Research Ethics Committee. Human renal artery specimens were obtained from nephrectomy patients (with a poorly functioning kidney [e.g., hydronephrosis, renal calculus]) with hyperglycemia (diabetes) or normal blood glucose (nondiabetes) after obtaining informed consent. The mean age of patients was 58 (range 44–72) years.

Primary mouse endothelial cells (MAECs) were cultured as previously reported (16). The H5V mouse endothelial cell line was purchased from American Type Culture Collection and cultured in DMEM (Gibco, Gaithersburg, MD).

Mouse thoracic aortas were dissected in Krebs-Henseleit solution containing (in mmol/L): 119 NaCl, 4.7 KCl, 2.5 CaCl₂, 1 MgCl₂, 25 NaHCO₃, 1.2 KH₂PO₄, and 11 d-glucose (17). Some aortas were treated with high glucose (HG) or transduced with adenovirus-mediated miR-200c for 24 h in DMEM. Changes in isometric force were then recorded by myograph (Danish Myo Technology, Aarhus, Denmark). Acetylcholine (ACh)-induced endothelium-dependent relaxations (EDRs) were examined. All drugs and chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Resistance mesenteric arteries were dissected in sterilized PBS and transduced with the indicated adenovirus for 24 h. Flow-mediated dilatation (FMD) was recorded by a Zeiss Axiovert 40 microscope, model 110P, with video camera and monitored with MyoVIEW software (Danish Myo Technology) as previously described (18).

Aortic segments (2 mm in length) were incubated in 5 μmol/L dihydroethidium (DHE) (Molecular Probes, Eugene, OR), cut open, and observed under an FV1000 confocal microscope (Olympus, Tokyo, Japan) (excitation 515 nm; emission 585 nm). Reactive oxygen species (ROS) released from MAECs was also measured by an EMX electron paramagnetic resonance (EPR) spectrometer (Bruker Daltonics, Bremen, Germany) with 100 μmol/L 1-hydroxy-2,2,6,6-tetramethyl-4-oxo-piperidine hydrochloride (TEMPONE-H; Alexis) and 5,5-dimethyl-1-pyrroline-N-oxide (Alexis) as spin trap agents. HX-XO (100 μmol/L hypoxanthine + 0.01 units/mL xanthine oxidase) was used as positive control.

A fragment containing mature miR-200c flanked by 150 base pairs upstream and 150 base pairs downstream of genomic sequence was amplified by PCR from mouse and human genomic DNA and cloned into pAdTrack-U6 (which also contains a green fluorescent protein [GFP] cassette) (19,20). Oligonucleotides against the mature sequence of miR-200c gene, CCGGTCCATCATTACCCGGCAGTATTATTTTTC (forward) and TCGAGAAAAATAATACTGCCGGGTAATGATGGA (reverse), were annealed and cloned into pAdTrack-U6 and named as anti–miR-200c. Mutant miR-200c sequence was modified from TAATACTGCCGGGTAATGATGGA to ATAAAGTCCCGGGTAATGATGGA. The cDNA of mouse ZEB1 was subcloned from pcDNA-ZEB1-HIS (a gift from G. Goodall, The University of Adelaide, Adelaide, SA, Australia) into pAdTrack-CMV. All vectors were constructed to adenoviral plasmid through recombining with pAdEasy-1 in BJ5183 Escherichia coli. After PacI (New England Biolabs, Ipswich, MA) digestion, adenoviral plasmids were transfected into HEK293 cells to generate infectious adenovirus particles. pAdTrack-CMV-GFP (Ad-GFP) and pAdTrack-U6 (named as miR-Ctrl) adenoviruses were used as control in corresponding experiments.

Proteins prepared from MAECs or mouse aortas were dissolved in 5× SDS loading buffer, denatured at 95°C for 5 min, developed with SDS-PAGE, and transferred to a polyvinylidene fluoride membrane (Millipore, Bedford, MA) for Western blotting using antibodies detected by an ECL chemiluminescence system (GE Healthcare, Pittsburgh, PA). The primary antibodies used were anti-COX-2 (1:1,000; Abcam, Cambridge, U.K.), anti-GAPDH (1:5,000; Ambion, Austin, TX), and anti-ZEB1 (1:1,000; Santa Cruz, Dallas, TX).

miRNAs from endothelial cells or aortas were extracted using a mirVana miRNA Isolation Kit (Ambion). miRNA screening was done using an NCode SYBR GreenER miRNA qRT-PCR Kit (Invitrogen) (Thermo Scientific). Primers were mmu-miR-223: tgtcagtttgtcaaatacccca; mmu-miR-200c: taatactgccgggtaatgatgga; mmu-miR-24: tggctcagttcagcaggaacag; mmu-miR-320: aaaagctgggttgagagggcga; mmu-miR-503: tagcagcgggaacagtactgcag; mmu-miR-200b: taatactgcctggtaatgatga; mmu-let-7b: tgaggtagtaggttgtgtggtt; mmu-miR-20b: caaagtgctcatagtgcaggtag; mmu-miR-21: tagcttatcagactgatgttga; mmu-miR-146a: tgagaactgaattccatgggtt; mmu-miR-221: agctacattgtctgctgggtttc; mmu-miR-155: ttaatgctaattgtgataggggt; and mmu-miR-23b: atcacattgccagggattacc. miR-200a, -b, and -c expression was determined by Applied Biosystems TaqMan MicroRNA Assay (Life Technologies) using the ViiA 7 System (Applied Biosystems) (8). Primer identification catalog numbers were 000502 for mmu-miR-200a-3p/hsa-miR-200a-3p, 002251 for mmu-miR-200b-3p/hsa-miR-200b-3p, 000505 for mmu-miR-200c-3p/hsa-miR-200c-3p, and 001973 for snRU6.

Total RNA from MAECs or mouse aortas was extracted using TRIzol reagent (Invitrogen). Reverse transcription was carried out in 2 μg total RNA using an iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). cDNA was amplified using SYBR Green Real-Time PCR Master Mix (Life Technologies). GAPDH was used as endogenous control. Primers for mouse ZEB1 were sense: 5′ CAAACACCACCTGAAAGAGCAC 3′, antisense: 5′ AAGAGATGGCGAGGAACACTG 3′; for mouse GAPDH, sense: 5′ AGGTCGGTGTGAACGGATTTG 3′, antisense: 5′ TGTAGACCATGTAGTTGAGGTCA 3′; for mouse Cbr1, sense: 5′ CGCAGGCATCGCCTTCAA 3′, antisense: 5′ CCTCTGTGATGGTCTCGCTT 3′; for mouse Fam213b, sense: 5′ GGAGCATCCTGGACCAACAC 3′, antisense: 5′ GGCAGCTGGTAGGATGCTTA 3′; for mouse prostaglandin E₂ (PGE₂) synthase, sense: 5′ CATCAAGATGTACGCGGTGG 3′, antisense: 5′ CTCCACATCTGGGTCACTCC 3′; for mouse prostacyclin (PGI₂) synthase, sense: 5′ GGTGGCGGTGACTTGTTGC 3′, antisense: 5′ TCCAACGGAGGCTCACCAG 3′; and for mouse thromboxane A₂ (TXA₂) synthase, sense: 5′ ACCTACTTCTTTCTCCACCACCT 3′, antisense: 5′ TGATGCCCAACTTCTCCAGTC 3′.

The plasmid pcDNA-ZEB1-HIS contains coding sequence for mouse ZEB1. The expression plasmids (pcDNA-c-Fos, c-Jun, p65, and p50) of c-Fos, c-Jun (2 units AP-1), p65, and p50 (2 units nuclear factor-κB [NF-κB]) were provided by D.P. Liu (Peking Union Medical College, Beijing, China). A 1,301-base pair human COX-2 promoter fragment (−1,267 to +34) was cloned into the KpnI/HindIII site of pGL3-basic to construct pGL-COX-2. The purified luciferase plasmids were transfected into H5V cells cultured in a 24-well plate using Lipofectamine 2000 (Invitrogen) pRL-TK reporter (Promega, Madison, WI) as internal control. Luciferase activity was assessed by using the Dual-Luciferase Reporter Assay System (Promega).

Prostaglandin F_2α, PGE₂, 6-keto PGF_1α (the stable product of PGI₂), and thromboxane B₂ (the stable product of TXA₂) were measured using the high-performance liquid chromatography–coupled mass spectrometry method (21). Briefly, after treatment with miR-200c overexpressing virus for 24 h, C57BL/6 mouse aortas were transferred to Krebs-Henseleit solution and exposed to ACh 3 μmol/L for 3 min. The levels of prostaglandins in the aortas and medium were measured by Agilent 1100 series high-performance liquid chromatography (Agilent Technologies, Santa Clara, CA) and Bruker Daltonics micrOTOF-Q mass spectrometer.

Mice were fasted for 8 h and then loaded orally with glucose 1.2 g/kg body weight. Blood glucose was measured at times 0, 15, 30, 60, 90, and 120 min with a commercial glucometer (Ascensia ELITE; Bayer, Mishawaka, IN). For the insulin tolerance test, mice were fasted for 2 h and then injected with insulin 0.75 units/kg body weight. Blood glucose was measured as aforementioned.

Results are presented as mean ± SEM of the number of separate experiments. Concentration-response curves were analyzed using GraphPad Prism version 4.0 software. Statistical significance was determined by Student t test (two-tailed) or one-way ANOVA followed by the Bonferroni post hoc test when more than two treatments were compared. P < 0.05 indicates statistical difference between groups.

We first performed quantitative PCR (qPCR) to detect the relative expression of a number of miRNAs, which were reportedly present in endothelial cells, and found that the miR-200c level was markedly higher in aortas of diabetic db/db mice than in nondiabetic mice (Supplementary Fig. 1A). Furthermore, the TaqMan probe-based qPCR analysis showed that miR-200c was the most abundant isoform among the miR-200 family members, >100-fold more than miR-200a and miR-200b in nondiabetic db/m⁺ mouse aortas (Fig. 1A) and ∼30-fold more than miR-200a and miR-200b in renal arteries of human subjects without diabetes (Fig. 1B). We thus selected miR-200c as the target molecule for subsequent studies. The expression of miR-200c in db/db mouse aortas and in renal arteries from patients with diabetes was approximately twofold greater than that of respective arteries from nondiabetic mice (Fig. 1C) or human subjects (Fig. 1D). The miR-200c expression was also elevated in HG (30 mmol/L, 24 h)-treated MAECs, and this effect was reversed by cotreatment with ROS scavenger tempol (4-hydroxy-2,2,6,6-tetramethylpiperidine-N-oxyl) (100 μmol/L) (Fig. 1E). Furthermore, 24-h treatment with 100 μmol/L HX-XO to release ROS augmented miR-200c expression in MAECs, which was abolished by tempol (Fig. 1F). However, miR-200c overexpression by adenoviral (ad-miR-200c) transduction did not affect ROS production in MAECs, as detected by EPR spectroscopy (Fig. 1G and Supplementary Fig. 1B), and in the endothelium of db/m⁺ mouse aortas, as indicated by en face DHE staining (Fig. 1H and Supplementary Fig. 1C). These results indicate that hyperglycemia upregulates miR-200c expression most likely through ROS elevation.

Ex vivo transduction of ad-miR-200c markedly impaired ACh-induced EDRs and increased the miR-200c level in db/m⁺ mouse aortas, whereas miR-Ctrl overexpressing virus (ad-miR-Ctrl) and mut-miR-200c overexpressing virus (ad-mut-miR-200c) had no effect (Fig. 2A and B and Supplementary Fig. 2A and B). By contrast, endothelium-independent relaxations to sodium nitroprusside were unaffected by ad-miR-200c (Supplementary Fig. 2C), suggesting that miR-200c–reduced EDRs were likely caused by dysfunction of endothelial cells but not vascular smooth muscle cells. Of note, miR-200c overexpression also attenuated EDRs in human renal arteries (Fig. 2C). Cotreatment with ad-anti–miR-200c reversed miR-200c–induced impairment of EDRs in both mouse aortas (Fig. 2A) and human renal arteries (Fig. 2C).

To explore the pathological role of elevated miR-200c in diabetic endothelial dysfunction, db/db mouse aortas were cultured and transfected with ad-anti–miR-200c for 24 h. Ad-anti–miR-200c dramatically reduced miR-200c expression (Supplementary Fig. 3A) and improved EDRs (Fig. 2D and E) in db/db mouse aortas in a dose-dependent manner. Inhibition of miR-200c by anti–miR-200c also improved FMD of second-order mesenteric resistance arteries (Fig. 2F and Supplementary Fig. 3B). In addition, 24-h incubation with ad-anti–miR-200c protected against HG (30 mmol/L, 48 h)-induced attenuation of EDRs in db/m⁺ mouse aortas (Fig. 2G). These results suggest that miR-200c elevation is most likely to mediate endothelial dysfunction in diabetic mice.

Transduction (24 h) with ad-miR-200c suppressed the expression of ZEB1 at mRNA and protein levels in MAECs, and both effects were reversed by ad-anti–miR-200c (Fig. 3A and B), suggesting the involvement of ZEB1 inhibition in miR-200c–mediated endothelial dysfunction. Indeed, transduction with a ZEB1-overexpressing virus (ad-ZEB1, 24 h) (Fig. 3C) abolished ad-miR-200c–induced impairment of EDRs in db/m⁺ mouse aortas (Fig. 3D).

ZEB1 expression was significantly decreased both in db/db mouse aortas (Fig. 4A) and in renal arteries from patients with diabetes (Fig. 4B) compared with that in respective control subjects without diabetes. Inhibition of miR-200c by anti–miR-200c increased ZEB1 expression in db/db mouse aortas (Fig. 4C). Likewise, ad-anti–miR-200c also prevented HG (30 mmol/L, 24 h)-induced ZEB1 downregulation in MAECs (Fig. 4D). These results indicate that decreased ZEB1 expression due to miR-200c upregulation may account for diabetic endothelial dysfunction. Indeed, overexpression of ZEB1 by transduction with adenovirus (ad-ZEB1) (Supplementary Fig. 4A) reversed endothelial dysfunction in db/db mouse aortas and mesenteric arteries (Fig. 4E and F and Supplementary Fig. 4B and C). Consistent with a previous report in other tissues (22), overexpression of ZEB1 dose-dependently inhibited miR-200c expression in db/db mouse aortas and in MAECs (Fig. 4G and H).

To further confirm the significant role of miR-200c elevation in the maintenance of diabetic endothelial dysfunction, ad-anti–miR-200c or ad-miR-Ctrl was administered to db/db mice through tail intravenous injection (10⁷ plaque-forming units [pfu]). After 1 week of treatment, EDRs in ad-anti–miR-200c–treated mice were markedly augmented in aortas, accompanied by reduced miR-200c expression but elevated ZEB1 expression (Fig. 5A–C). Similarly, in vivo transduction of ad-ZEB1 in db/db mice also improved EDRs, increased ZEB1 expression, and decreased miR-200c expression (Fig. 5D–F).

We and others demonstrated before that COX-2 is an important mediator of endothelial dysfunction (23–25), and ZEB1 is reported to suppress COX-2 expression (14). We next tested the hypothesis that COX-2 contributes to miR-200c–induced endothelial dysfunction on the basis of our observation that miR-200c suppressed ZEB1 expression in endothelial cells. As expected, miR-200c failed to attenuate EDRs in aortas of COX-2^−/− mice (Fig. 6A and Supplementary Fig. 5A). In support of the functional results, miR-200c elevated COX-2 protein expression in MAECs (Fig. 6B), which was reversed by overexpression of ZEB1 or anti–miR-200c (Fig. 6B and C). However, COX-1 protein expression was not changed by miR-200c (Fig. 6B and C). Moreover, the luciferase reporter gene assay demonstrated that miR-200c increased promoter activity of COX-2, which was inhibited by overexpression of ZEB1 but not by overexpression of c-Fos/c-Jun (AP-1) or p65/p50 (NF-κB) (Fig. 6D), suggesting that ZEB1 directly inhibits miR-200c–stimulated COX-2 upregulation at the transcriptional level in mouse endothelial cells. In vitro evidence also showed that ZEB1 overexpression downregulated COX-2 expression in db/db mouse aortas (Fig. 6E). Further experiments showed that miR-200c (24 h) did not change the mRNA levels of prostaglandin F_2α synthases (Cbr1 and Fam213b), PGE₂ synthase, PGI₂ synthase, and TXA₂ synthase (Supplementary Fig. 5B), but it increased PGE₂ levels by approximately fivefold in the medium (3–4 vs. 15–20 ng/mL) and by approximately twofold in the aortas (0.15–0.2 vs. 0.4–0.5 ng/mg). PGI₂ (assayed in the form of 6-keto PGF_1α) level was also increased by approximately twofold in the medium (10–13 vs. 17–25 ng/mL) but not in the aortas (Supplementary Fig. 5C and D). We next tested whether these two elevated prostaglandins affected EDRs in mouse aortas. PGE₂ at a concentration of 100 nmol/L (35.2 ng/mL, approximately twofold of the level in the medium) attenuated ACh-induced relaxation (Supplementary Fig. 5E). However, 100 nmol/L PGI₂ (37.4 ng/mL, approximately twofold of the level in the medium) had no effect, indicating that PGE₂ is likely involved in miR-200c–induced endothelial dysfunction in mouse aortas.

The key findings of the current study are that 1) miR-200c expression is elevated in arteries from diabetic mice and patients with diabetes, and hyperglycemia stimulates miR-200c expression through a ROS-dependent mechanism; 2) miR-200c overexpression impairs EDRs in nondiabetic mouse aortas, and suppression of miR-200c by ad-anti–miR-200c improves EDRs in db/db mouse aortas and FMD of resistance mesenteric arteries; 3) miR-200c suppresses ZEB1, leading to COX-2 upregulation, which in turn impairs endothelial function in diabetic mice; and 4) ZEB1 overexpression restores endothelial function in both conduit and resistance arteries in diabetes. The results highlight the prospect of an miR-200c-ZEB1-COX-2 axis as a potentially novel therapeutic target to restore endothelial function in diabetes (Fig. 6F).

Approximately 2,000 miRNAs have so far been identified in human, which are believed to control the expression of at least 30% of genes and participate in nearly every aspect of biological processes, including cell proliferation, differentiation, apoptosis, and development (26). Most studies have focused on the role of miRNAs in the development and growth of cancer, so the involvement of miRNAs in the development of diabetes and its vascular complications is known to a much lesser degree. Although our initial screening showed a markedly elevated expression of miR-200c in db/db mouse aortas, the pathological significance of miR-200c upregulation in diabetic vasculopathy is basically unexplored.

ROS is critically involved in hyperglycemia-induced endothelial dysfunction because elimination of ROS by tempol improves endothelial function in db/db mouse aortas (27). The current study shows that HG is able to upregulate miR-200c expression, which is mediated by oxidative stress because HG-stimulated miR-200c upregulation in MAECs is reversed by tempol. Furthermore, HX-XO, a ROS generator, increases miR-200c expression. Likewise, another study showed that H₂O₂ raises miR-200c expression in human endothelial cells (28). By contrast, miR-200c overexpression does not affect ROS production in endothelial cells, as reflected by both in situ DHE staining and EPR spectroscopy. These results suggest that hyperglycemia-associated oxidative stress is most likely to account for miR-200c upregulation in endothelial cells under diabetic conditions. These initial observations drove us to elucidate the pathophysiological role of miR-200c in diabetic endothelial dysfunction aided by the use of viral constructs in a series of in vitro, ex vivo, and in vivo studies. We first demonstrate that overexpression of miR-200c impairs endothelial function in aortas from nondiabetic mice and that such impairment can be reversed by inhibition of miR-200c in both conduit and resistance arteries in db/db mice. Taken together, we established that elevated expression of miR-200c in diabetic conditions is likely a key mediator of diabetic endothelial dysfunction.

We next studied the detailed molecular mechanisms underlying miR-200c–induced endothelial dysfunction. ZEB1 is a zinc finger transcriptional repressor and is expressed in various types of cells. miR-200 family is known to inhibit epithelial-to-mesenchymal transition during cancer development by repression of ZEB1 (29–31). One previous study showed that miR-200c suppresses ZEB1 expression and subsequently leads to endothelial cell senescence and apoptosis (28). We also found that miR-200c inhibits ZEB1 expression in nondiabetic mouse aortas, whereas overexpression of ZEB1 attenuates miR-200c–induced impairment of endothelial function in these aortas, indicating that ZEB1 downregulation is likely involved in miR-200c–induced endothelial dysfunction in normal mice. However, whether ZEB1 also plays a significant role in diabetes-related endothelial dysfunction is unknown. The current study shows that ZEB1 expression is decreased in db/db mouse aortas and that suppression of miR-200c restores the diminished ZEB1 expression. More importantly, transduction of ZEB1 overexpressing adenovirus profoundly improves endothelial function in db/db mice without affecting insulin sensitivity (Supplementary Fig. 6), and this effect is almost identical to that of miR-200c inhibition. Therefore, hyperglycemia in diabetes probably upregulates miR-200c expression, which subsequently downregulates ZEB1 expression, leading to impaired endothelial function. This notion is supported by the observation that ZEB1 overexpression is accompanied simultaneously by miR-200c downregulation in aortas from not only nondiabetic mice but also diabetic db/db mice. Previous studies on epithelial cells showed that miR-200 and ZEB are mutually negative regulators of their expression and control the typical epithelial-to-mesenchymal transition–associated properties in cancer development (22). Here, we also demonstrate that this reciprocal negative regulatory circuit plays an important role in the maintenance of diabetic endothelial dysfunction.

Previous studies revealed that upregulated COX-2 expression and activity are crucial contributors to diabetes- or hyperglycemia-associated endothelial dysfunction (32,33). However, the exact molecular mechanism mediating hyperglycemia-stimulated COX-2 upregulation remains unclear. Few earlier studies suggested that the phosphatidylinositol 3-kinase/Akt signaling and transcriptional factors NF-κB and CREB are involved (34,35). Here, we report that miR-200c stimulation elevates COX-2 expression in nondiabetic mouse aortas, which is reversed by anti–miR-200c or ZEB1 overexpression. In addition, ZEB1 overexpression downregulates COX-2 expression in db/db mouse aortas. These results suggest that the miR-200c/ZEB1 negative feedback loop mediates a substantial part of hyperglycemia-stimulated COX-2 upregulation in endothelial cells under diabetic conditions. The essential role of COX-2 in miR-200c–induced endothelial dysfunction is finally confirmed by the inability of miR-200c to attenuate EDRs in aortas from COX-2^−/− mice. The current study shows that miR-200c–induced COX-2 upregulation results in a fivefold increase of PGE₂ in the medium and a twofold increase in mouse aortas. PGI₂ (assayed in the form of 6-keto PGF_1α) was also increased by twofold only in the medium. PGE₂ induces vasoconstriction by binding to PGE₂ receptors 1 and 3 (36,37). PGI₂ is reported to induce contraction in aortas of spontaneously hypertensive rats and aged Wistar-Kyoto rats, probably by interacting with TXA₂ receptors (38). Therefore, either PGE₂ or PGI₂ is a possible candidate to mediate miR-200c–induced endothelial dysfunction. Further experiments showed that pretreatment with PGE₂ but not PGI₂ can actually attenuate ACh-induced relaxations in C57BL/6 mouse aortas, thus suggesting that PGE₂ is likely to contribute to COX-2 upregulation–associated endothelial dysfunction in mice.

The clinical use of several COX-2 inhibitors was found to increase thrombotic cardiovascular events (39,40). This adverse effect is likely related to the disrupted prothrombotic/antithrombotic balance. Human platelets contain only COX-1, and the major prostaglandin is the potent proaggregatory and vasoconstrictive TXA₂ (41), whereas endothelial cells normally express COX-2, and the major COX-2–derived arachidonic acid product is PGI₂ in endothelial cells (42,43). PGI₂ can interact with platelet prostacyclin receptor to inhibit platelet aggregation. However, selective inhibition of COX-2 reduces the production of PGI₂ in endothelial cells, while leaving platelet production of TXA₂ intact, tilting the prothrombotic/antithrombotic balance toward thrombosis (39,40), which is probably the main cause of cardiovascular events associated with the use of COX-2 inhibitors in some patients. On the other hand, the current experimental results show that COX-2 upregulation leads to elevated production of PGE₂, the latter of which can reduce EDR, suggesting that the type of prostaglandins derived from COX-2 upregulation induced in various situations may determine its impact (harmful or beneficial) on vascular function.

Like diabetic mouse arteries, renal arteries from patients with diabetes also exhibit an increased miR-200c expression and decreased ZEB1 expression. Moreover, overexpression of miR-200c impairs endothelial function in renal arteries from subjects without diabetes, which is again reversed by cotreatment with ad-anti–miR-200c. Although the renal vascular reactivity can be influenced by the varied disease backgrounds of the patients with diabetes from whom the arteries were obtained, the current study provides useful evidence for the first time to our knowledge that abnormal miR-200c/ZEB1 expression may involve endothelial dysfunction in patients with diabetes.

In conclusion, we demonstrate upregulated miR-200c in arteries from both diabetic mice and patients with diabetes. Inhibition of miR-200c by anti–miR-200c and ZEB1 overexpression effectively restores the impaired endothelial function in both conduit and resistance arteries of diabetic mice through inhibiting the expression and activity of COX-2, an important mediator of vascular dysfunction probably through elevating prostaglandins such as PGE₂. The current study not only deepens our understanding about the role of miRNAs in vascular pathophysiology associated with diabetes but also adds candidates to the existing list of therapeutic anti-miRs should their safety be approved. Taken together, miR-200c could serve as a promising target for ameliorating diabetic vasculopathy.

Acknowledgments. The authors thank Gregory Goodall (The University of Adelaide, Adelaide, SA, Australia) and De-Pei Liu (Peking Union Medical College, Beijing, China) for providing the plasmids.

Funding. This study was supported by the Research Grants Council of Hong Kong (CUHK2/CRF/12G, T12-402/13N), National Natural Science Foundation of China (81270932, 81471082, and 91339117), Beijing Natural Science Foundation (5122028), Hong Kong Scholarship Program, and CUHK High Promise Initiatives.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.Z. and J.L. contributed to the study design, conduct of the experiments, data analysis, and preparation of the manuscript. D.Q., L.W., and J.-Y.L. contributed to the conduct of the experiments and data analysis. C.W.L. and Z.G. contributed to the functional studies. P.L. provided the plasmids. G.L.T. provided the knockout mice. H.K.L. performed the bioassays. C.F.N. provided the human samples. R.C.W.M. and X.Y. contributed to the discussion. Y.H. contributed to the study design and preparation of the manuscript. Y.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.