miR-200b Mediates Endothelial-to-Mesenchymal Transition in Diabetic Cardiomyopathy

Focal myocardial fibrosis is a structural hallmark of diabetic cardiomyopathy, manifesting functionally as defective diastolic compliance and cardiac contractility (1–3). Myocardial fibrosis results from increased production and deposition of extracellular matrix (ECM) proteins, such as collagen. Endothelial cells (ECs) are initial targets of hyperglycemic damage and play a major role in the production of ECM proteins in all chronic diabetic complications. At the structural level in the heart, such changes are manifested as thickening of the capillary basement membrane, focal myocardial fibrosis, and so forth (3,4). A result of sustained injury in the fibrotic diseases is that ECs undergo a process of transdifferentiation (i.e., endothelial-to-mesenchymal transition [EndMT]) (5,6). Such changes in the ECs are manifested by a loss of endothelial markers and development of mesenchymal markers (7). Several reports have shown mesenchymal transition of ECs (i.e., EndMT during cardiac fibrosis) (5–7). This EndMT is similar to epithelial-to-mesenchymal transformation in cancer cells. EndMT is a key source of cardiac fibroblasts causing increased production of ECM proteins (5–7) and is characterized by a loss of cell junctions, vascular endothelial (VE)-cadherin, and CD31 expression and a gain of mesenchymal markers, such as vimentin (Vim) and α-smooth muscle actin (SMA) (5–7). Upregulation of p300 has been shown to be a key event in EndMT in the heart (5). The same study also demonstrated the alteration of several microRNAs (miRNAs) in EndMT of nondiabetic etiology (5). We previously demonstrated EndMT in the retina of diabetic animals (8) as well as a pathogenetic role of endothelin 1 (ET-1) in diabetic cardiomyopathy (3,4). Of note, ET-1 has been demonstrated to be of significance in EndMT in the diabetic heart (9). We also have demonstrated that p300 is upregulated in the cardiac tissues of diabetic animals (10). p300-Mediated histone acetylation is a key mechanism in regulating gene transcription. Such epigenetic mechanisms also influence miRNA production and vice versa (10,11).

miRNAs are small (∼20–25-nucleotide) RNA molecules that significantly affect the regulation of gene expression (11,12). The transcription of miRNAs occurs through RNA polymerase II, and their processing to precursor miRNAs (70–100 nucleotides, hairpin-shaped) in the nucleus is mediated by RNAse III Drosha and DGCR8. Following synthesis, they are exported to the cytoplasm by exportin 5. In the cytoplasm, they are processed by Dicer into mature miRNA, the functionally active form. miRNAs bind to specific mRNA targets, causing their degradation or translational inhibition. The importance of miRNAs has been demonstrated in diverse cellular processes (11,12). We have previously demonstrated alterations of multiple miRNAs in ECs exposed to high glucose and in multiple tissues affected in chronic diabetic complications, including the heart. The list includes miR-133a, miR-1, miR-320, miR-146a, and miR-200b. Studies from our laboratory have demonstrated that miR-200b is a key mediator of EndMT in the diabetic retina (8,13–17).

The purpose of the current study was to investigate whether EndMT is of pathogenetic significance in diabetic cardiomyopathy and whether such changes are mediated through alteration of miR-200b and/or other epigenetic mechanisms, such as p300 alterations. We also investigated whether manipulation of miR-200b is a potential therapeutic approach to prevent EndMT. To test miR-200b–mediated EndMT in diabetic cardiomyopathy, we generated endothelial-specific (using a Tie2 promoter) miR-200b–overexpressing mice in which we chemically induced diabetes (8).

We performed all animal experiments according to the Guiding Principles in the Care and Use of Animals. Experimental protocols were reviewed and approved by the Western University Council on Animal Care Committee. The experiments were also designed to conform with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (publication no. 85-23, revised 1996).

As described previously, we used transgenic mice with EC-specific overexpression of miR-200b generated by us (8). We cloned a cDNA fragment containing miR-200b into pg52pSPTg.T2FpAXK (pg52) plasmid containing a Tie2 promoter and enhancer and an SV40 PolyA signal (courtesy of T. Sato, Nara Institute of Science and Technology Graduate School of Biological Sciences, Ikoma, Japan) (18). We excised Tie2-miR-200b transgene by using SalI and injected gel-purified transgene into pronuclei of fertilized eggs from C57BL/6 crossed with CBA/J mice. These were then transferred into pseudopregnant female mice (18). PCR-based assays of the genomic DNA extracted from tail-tip biopsy specimens as a template were used to identify transgenic strains according to the previously described protocol (8). Phenotypic characterization revealed no alteration in the transgenic mice.

We divided male miR-200b transgenic mice and age- and sex-matched wild-type littermates (weighing 23–26 g) into diabetic and control groups. Diabetes was induced by five intraperitoneal injections of streptozotocin (STZ) 50 mg/kg in citrate buffer (pH 4.5) (controls received the same volume of buffer) on consecutive days as previously described (14–16). Hyperglycemia was confirmed by measuring blood glucose levels (at least twice) 3 days after the last STZ injection. Following confirmation of diabetes, the mice were maintained in a hyperglycemic state for 2 months (without exogenous insulin). At the end of the follow-up period, echocardiography was performed in all mice. The mice were then killed, and cardiac tissues and blood were collected. Most of the cardiac tissues were stored frozen (−70°C) for various analysis (see next sections). A small portion of ventricular tissue was fixed in 10% neutral-buffered formalin. These tissues were embedded in paraffin. For histologic studies, 5-μm sections with hematoxylin-eosin, immunofluorescence, and trichrome staining were used.

Lightly anesthetized (1.5% inhaled isoflurane) mice were subjected to echocardiography on a warm handling platform. We used a 40-MHz linear array transducer (MS550D) and Vevo 2100 preclinical ultrasound system (VisualSonics, Toronto, ON, Canada) with nominal in-plane spatial resolution of 40 (axial) × 80 (lateral) μm. M-mode and two-dimensional parasternal short-axis scans (133 frames/s) at the level of the papillary muscles were obtained to assess changes in left ventricular (LV) end-systolic inner diameter (LVIDs) and LV end-diastolic inner diameter (LVIDd). LV volumes at end diastole (LVEDV) and end systole (LVESV) were calculated by Eqs. 1 and 2:

LV fractional shortening (FS) and ejection fraction (EF) were used as indexes of cardiac contractile function and were calculated from the inner diameters by Eqs. 3 and 4:

Color flow–guided, pulsed wave Doppler recordings of the maximal early (E) and late (A) diastolic transmitral flow velocities and Doppler tissue imaging recordings of peak E′ velocity and peak A′ velocity were carried out. Mitral inflow patterns (E/A ratio) and mitral annulus velocities (E′/A′ ratio) were used to assess diastolic dysfunction.

We used isolated primary mouse heart ECs (MHECs) as previously described (19). Briefly, on the day before the experiment, 100 μL sheep anti-rat DynaBeads per litter (5–7 mice) were prepared by washing in 0.1% BSA-PBS and were resuspend 0.1% BSA-PBS. One hundred microliters of beads were mixed with 10 μL of rat anti-mouse platelet EC adhesion molecule 1 monoclonal antibody (Invitrogen Canada, Burlington, ON, Canada) and incubated overnight at 4°C. The beads were then washed and resuspended in DMEM containing 10% FBS. After harvesting, the hearts were cut into small pieces and rinsed in Hanks’ balanced salt solution buffer. They were then digested with collagenase A. Ten percent FBS-DMEM was added. The mixtures were filtered using nylon mesh (70-μm pores), pelleted by centrifugation, washed, and resuspended in the same solution. They were incubated with CD31 beads prepared by the same method. They were washed, and bound cells were cultured in 3.5-cm dishes precoated with 2% gelatin (Sigma, Oakville, ON, Canada) in endothelial basal medium 2 with 10% FBS. The remaining cells were incubated at 37°C for 1 h. The suspended cells (myocytes) and cells attached on the bottom of the flask (fibroblasts) were separated.

Isolated MHECs were plated at a density of 1 × 10⁵ cells/mL. After 24 h of serum-free media incubation, MHECs were incubated with various levels of d-glucose for 48 h. Each experiment was performed with three or more samples.

Isolation of total RNA was performed using TRIzol reagent (Invitrogen Canada) according to the manufacturer’s protocol and as described previously (13–17). The extracted RNA was used for cDNA synthesis by using a High-Capacity cDNA Reverse Transcription Kit (Life Technologies, Grand Island, NY). Quantification of various transcripts was performed by real-time quantitative RT-PCR (LightCycler; Roche Diagnostics, Laval, QC, Canada). Primers for the PCR reactions were designed using Primer 5 software (Supplementary Table 1).

For miRNA analyses, we used a mirVana miRNA Isolation Kit (Ambion, Austin, TX) for miRNA extraction from cardiac tissues collected as described previously (14,16). The cDNA was synthesized with TaqMan microRNA Assay Reverse Transcription Primer and MultiScribe Reverse Transcriptase (Life Technologies) by using the manufacturer’s protocol. Real-time quantitative RT-PCR for miR-200b was performed with the TaqMan microRNA Assay (LightCycler 96) as described previously. The data were normalized to U6 small nuclear RNA to account for differences in reverse transcription efficiencies and the amount of template (13–16). miRNA from serum was extracted using the miRNeasy Serum/Plasma Kit (QIAGEN, Toronto, ON, Canada) according to the manufacturer’s instructions. The isolated serum miRNAs were analyzed as described earlier.

ECs were transfected with mRNA-200b mimic or antagomir (20 nmol/L) using transfection reagent Lipofectamine 2000 (Invitrogen Canada). Scrambled control was used in parallel. miRNA transfection efficiency was determined by real-time RT-PCR by using established techniques (14,15).

The cells and tissues at the end of various treatments were lysed with radioimmunoprecipitation assay buffer. Total protein levels were quantified using the bicinchoninic acid protein assay kit (Pierce, Rockford, IL). Briefly, 30 μg total protein from each sample was loaded to SDS-PAGE and transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA). The membranes were blocked with 5% nonfat milk in PBS containing 0.05% Tween-20. The membranes were incubated with specific primary antibodies (anti-CD31 and anti–fibroblast-specific protein 1 [FSP1] antibodies [Abcam, Cambridge, MA], anti–zinc finger E-box–binding homeobox 1 [ZEB1] antibody [Santa Cruz Biotechnology, Dallas, TX], anti-SMAD/phospho-SMAD2/3 [Cell Signaling, Danvers, MA]). They were then incubated with corresponding horseradish peroxidase–conjugated anti-mouse (Invitrogen Canada) or anti-rabbit (Santa Cruz Biotechnology) secondary antibodies. After washing, the detections were carried out with Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare Life Sciences, Pittsburgh, PA). Blots were stripped and reprobed with β-actin antibody (Abcam) and analyzed by densitometry. ELISA for p300 was performed using commercially available kits (USCN Life Science Inc., San Diego, CA) according to the manufacturer’s instructions.

Cardiac tissue sections were deparaffinized with xylene and alcohol. After washing and incubation with 5% normal serum, the sections were incubated in one of the specific primary antibodies (mouse anti-CD31, rabbit anti-FSP1, rabbit anti-CD34 [Abcam] at 1:100 dilution; mouse anti–α-SMA [Sigma] at 1:400 dilution) followed by incubation in corresponding Alexa Fluor 488–conjugated secondary antibody (1:200 goat anti-mouse IgG or goat anti-rabbit IgG [Invitrogen Canada]). The sections were examined with a fluorescent microscope (Olympus BX51; Olympus America, Melville, NY) and analyzed with ImageJ software.

Data are expressed as mean ± SEM. Significant differences were determined using ANOVA and Student t test, with post hoc analyses as appropriate. Differences were considered significant at P ≤ 0.05.

We first examined miR-200b levels in the hearts of diabetic mice. Quantitative RT-PCR analyses showed that miR-200b levels are reduced in whole hearts of diabetic mice compared with age- and sex-matched nondiabetic controls (Fig. 1A). We then examined which cells are responsible for this reduction, hence, isolated ECs and cardiomyocytes. As expected, there were no significant changes of miR-200b levels in the cardiomyocytes (Fig. 1B), but the ECs showed a significant (∼45%) decrease in miR-200b expression (Fig. 1C). Because cellular distribution had a specific implication in these studies, we also confirmed the cell type that overexpressed miR-200b in the transgenic mice. To this extent, we isolated myocytes, ECs, and fibroblasts from the transgenic and wild-type mice and found no significant differences in miR-200b levels in the fibroblasts and myocytes. The ECs, however, showed an ∼14-fold overexpression of miR-200b in the transgenic mice compared with wild-type controls (Fig. 1D).

We then explored whether cardiac ECs develop glucose-induced miR-200b–dependent EndMT by isolating ECs from wild-type B6 mice and miR-200b transgenic mice, growing them in either 5 or 25 mmol/L glucose and measuring specific markers for EndMT. Glucose exposure of wild-type ECs showed reduced CD31 and increased Vim mRNA expression (Fig. 1E and F). In parallel, mRNA levels of vascular endothelial growth factor (VEGF) and ZEB1 (two additional targets of miR-200b) were increased following 25 mmol/L glucose exposure, but these changes were prevented in the transgenic mouse ECs (Fig. 1G and H).

We examined whether EndMT can produce functional deficits characteristic of diabetic cardiomyopathy and whether such changes are miR-200b dependent. To examine the functional deficits characteristic of diabetic cardiomyopathy, we used STZ-induced diabetic mice. The mice demonstrated hyperglycemia and reduced body weight gain after 2 months of poorly controlled diabetes (Supplementary Table 2) and further showed glucosuria and polyuria (data not shown). Echocardiographic examination showed a slight nonsignificant increase in EF and FS (Fig. 2A and B). However, E/A and E′/A′ ratios were significantly reduced in the wild-type diabetic mice, indicating diastolic dysfunction. These changes were corrected in the transgenic animals (Fig. 2C and D). Additional parameters of echocardiographic data are provided in Supplementary Table 3.

EndMT is characterized by loss of endothelial markers and acquisition of mesenchymal markers (8). To examine the potential role of miR-200b in this process, we examined endothelial and mesenchymal markers in cardiac tissues from miR-200b transgenic mice and littermate controls with and without STZ-induced diabetes. In keeping with our hypothesis, mRNA expression of endothelial markers CD31, VE-cadherin, and CD34 were reduced (∼45%, 35%, and 37%, respectively) in the hearts of wild-type mice with diabetes (Fig. 3A, D, and F) in association with increased mRNA expression of mesenchymal markers FSP1, Vim, α-SMA, and collagen 1α1 (COL1) and 4α1 (COL4) (varying from ∼20–60%), confirming EndMT in animals with diabetes (Figs. 4A, D, and F and 5A and B). The protein levels of endothelial markers, such as CD31 and CD34 (Fig. 3B, C, E, and G), as well as of mesenchymal markers, such as FSP1 and α-SMA (Fig. 4B, C, E, and G), showed similar changes. Additional immunofluorescence stain for CD31, CD34, FSP1, and α-SMA confirmed such alterations in the cardiac microvascular ECs (Figs. 3E and G and 4E and G). Of note, no alterations of collagen 3α1 (COL3) were seen (Fig. 5C).

All such abnormalities were prevented in the miR-200b transgenic mice with diabetes, further establishing a pathogenetic mechanism involving miR-200b in cardiac EndMT in diabetes (Figs. 3–5). However, the level of prevention varied among the transcripts (Figs. 3–5). In parallel, focal myocardial fibrosis was present in the hearts of wild-type diabetic animals (Fig. 5E). Such foci (one to three per animal) were identified in the wild-type diabetic mice and were prevented in the transgenic mice with diabetes. To understand whether circulating miR-200b played a role in the generation or prevention of cardiac EndMT in diabetes, we measured serum miR-200b levels. The results showed no significant differences in the serum miR-200b levels among the various groups, indicating that the observed effects were mediated by autocrine/paracrine effects of miR-200b (Fig. 5D).

All miRNAs, including miR-200b, have multiple targets. Hence, we examined other miR-200b targets modulated by glucose exposure. We have previously demonstrated that miR-200b has additional effects on other molecules of interest in chronic diabetic complications (15). To this extent, we examined VEGF, p300, ZEB1, ZEB2 (all targets of miR-200b), and transforming growth factor-β (TGF-β), molecules that have been demonstrated to play key roles in EndMT in other situations (5–8,15). We measured RNA by RT-PCR and proteins by ELISA or by Western blot. In the hearts of diabetic mice, all these transcripts and the corresponding proteins were elevated, but they were prevented in the miR-200b transgenic animals with diabetes (Fig. 6A–H). We have previously shown that VEGF and p300 are direct targets of miR-200b (8,15). Here, we carried out additional experiments in the ECs by using miR-200b mimics and miR-200b antagomirs, which suggest that these are also directly modulated by miR-200b (Supplementary Fig. 1). To further investigate downstream mediators of TGF-β signaling, we performed Western blot analyses SMAD2/3. These assays showed upregulation of SMAD2/3, an miR-200b target. These changes were also prevented in the miR-200b transgenic mice (Fig. 6I and J).

In the current study, we show that diabetes-induced endothelial phenotypic change (i.e., EndMT) in the heart is responsible for increased ECM protein production. We demonstrated these changes in the glucose-exposed ECs isolated from wild-type and transgenic mice with endothelial-specific miR-200b overexpression and confirmed that miR-200b regulates such processes and that endothelial-specific miR-200b overexpression prevents diabetes-induced cardiac functional and structural abnormalities.

High glucose levels in diabetes is widely recognized as a key initiating factor for multiple chronic diabetic complications (2,8,20,21). ECs are one of the first cell types exposed to hyperglycemia. Elevated intracellular glucose levels may initiate multiple interlinked metabolic abnormalities, such as oxidative stress, protein kinase C activation, and increased nonenzymatic glycation (20,21). These metabolic alterations ultimately converge on the endothelial nucleus. Possibly as a defensive and/or compensatory response to cellular damage, ECs start producing altered amounts of multiple proteins. Such gene transcriptions at several levels are regulated by epigenetic mechanisms, such as acetylation and methylation, at the transcription or posttranscription level through miRNAs (11,12,22). Abnormal protein production may result in cellular phenotypic changes. Although we and others have previously demonstrated some aspects of EndMT (e.g., increased ECM protein production in diabetic cardiomyopathy and in other chronic diabetic complications [8,9,23,24]), the role of specific miRNA has not been demonstrated previously. In the current study, we show glucose-induced reduced expression of EC markers, such as CD31, CD34, and VE-cadherin, and increased expression of multiple mesenchymal markers, such as FSP1, COL4, COL1, SMA, and Vim, confirming EndMT. We further show that miR-200b regulates such processes. Of note, the amount of regulation varied among the transcripts. Overall, 30–40% of EndMT-related transcripts appeared to be influenced by miR-200b, but in the case of some mesenchymal transcripts, miR-200b appeared to have pronounced effects, the exact reasons for which are not clear. In keeping with the current findings, in pressure-induced cardiac damage, EndMT has been shown to contribute significantly to cardiac fibrosis and heart failure (24). However, miR-200b also regulates p300 production (8). As shown in this and our previous reports, a large number of molecules that cause EndMT may further be regulated through p300-dependant histone acetylation (8,10,15,25). Further experiments are needed to decipher the specific contribution of each pathway.

EndMT may happen in several chronic diseases. In the heart, physiologic development of cardiac valves depends on EndMT (5,26,27). In keeping with the current study, EndMT has been demonstrated in various diseases that lead to cardiac fibrosis (24,26,27). Oxidative stress may also be a key factor in such processes (27,28). Increased production of ECM proteins and endothelial dysfunctions are well established features of diabetic cardiomyopathy (10,14,29,30). The data generated from the current study indicate phenotypic change in the ECs linking such processes. We have previously demonstrated the role of miR-200b in augmented VEGF production in diabetes and its role in EndMT in the retina (8,15) and demonstrate similar findings in the current study. VEGF upregulation is known in several chronic diabetic complications (15,21). Furthermore, VEGF alteration has been demonstrated in pressure-induced EndMT (24).

In keeping with our previous finding in diabetic retinopathy, we show that miR-200b also modulates EndMT through modulation of transcription coactivator p300 (8,15). Alterations of p300 have previously been shown in cardiac EndMT (26), and p300 is known to regulate a large number of transcripts (10,25,31). We have also shown that p300 production is regulated by miR-200b (15). Conceptually, through p300, miR-200b may regulate multiple molecules, such as ET-1 and TGF-β. ET-1 has been shown previously to be involved in EndMT in the diabetic heart (9). In keeping with our previous retinal data, this study also showed that TGF-β–mediated SMAD signaling plays a key role in cardiac EndMT (8). We have previously demonstrated that other members of the miR-200 cluster (miR-200a, miR-200c, and miR-429) are not significantly altered by glucose and may not have pathogenetic significance (15). In addition, this study supports the important role ECs play in the pathogenesis of diabetic cardiomyopathy (15,21,30), indicating that alterations in ECs may be a primary initiating factor for the development of diabetic complications and further opening up new targets to block EndMT and its consequences. Such notions need to be further validated by carefully conducted long-term studies.

We also recognize that several other miRNAs may be altered in EndMT (5,14,16,26). In a previous study, miR-125b has been implicated in TGF-β2–mediated cardiac EndMT (5). Whether similar changes or additional miRNA changes occur specifically in diabetic cardiomyopathy would be of interest for future examination.

In summary, the current data indicate an miR-200b–mediated mechanism for the development of EndMT in the context of diabetic cardiomyopathy. Such findings open up the possibility of future RNA-based therapeutic strategies.

Funding. This work was supported in part by grants from the Heart and Stroke Foundation of Ontario and Canadian Diabetes Association and the National Natural Science Foundation of China (grant no. 81200305).

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

Author Contributions. B.F. contributed to the study concept and design, performance of the experiments, data review and analysis, and writing and editing of the manuscript. Y.Ca., S.Che., and X.C. performed the experiments. Y.Ch. contributed to the data review and analysis. S.Cha. contributed to the study concept and design, data review and analysis, and writing and editing of the manuscript. S.Cha. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.