Mechanism of the Switch from NO to H₂O₂ in Endothelium-Dependent Vasodilation in Diabetes

Abstract

Coronary microvascular dysfunction is prevalent among people with diabetes and is correlated with cardiac mortality. Compromised endothelial-dependent dilation (EDD) is an early event in the progression of diabetes, but its mechanisms remain incompletely understood. Nitric oxide (NO) is the major endothelium-dependent vasodilatory metabolite in the healthy coronary circulation, but this switches to hydrogen peroxide (H₂O₂) in coronary artery disease (CAD) patients. Because diabetes is a significant risk factor for CAD, we hypothesized that a similar NO-to-H₂O₂ switch would occur in diabetes.

Methods:

Vasodilation was measured ex vivo in isolated coronary arteries from wild type (WT) and microRNA-21 (miR-21) null mice on a chow or high fat/high sugar diet, and B6.BKS(D)-Lepr^db/J (db/db) mice using myography. Myocardial blood flow (MBF), blood pressure, and heart rate were measured in vivo using contrast echocardiography and a solid-state pressure sensor catheter. RNA from coronary arteries, endothelial cells, and cardiac tissues was analyzed via quantitative real-time PCR for gene expression, and cardiac protein expression was assessed via western blot analyses. Superoxide was detected via electron paramagnetic resonance.

Results:

1) Ex vivo coronary EDD and in vivo MBF were impaired in diabetic mice. 2) Nω-Nitro-L-arginine methyl ester, an NO-synthase inhibitor (L-NAME), inhibited ex vivo coronary EDD and in vivo MBF in WT. In contrast, polyethylene glycol-catalase, an H₂O₂ scavenger (Peg-Cat), inhibited diabetic mouse EDD ex vivo and MBF in vivo. 3) miR-21 was upregulated in diabetic mouse endothelial cells, and the deficiency of miR-21 prevented the NO-to-H₂O₂ switch and ameliorated diabetic mouse vasodilation impairments. 4) Diabetic mice displayed increased serum NO and H₂O₂, upregulated mRNA expression of Sod1, Sod2, iNos, and Cav1, and downregulated Pgc-1α in coronary arteries, but the deficiency of miR-21 reversed these changes. 5) miR-21 deficient mice exhibited increased cardiac PGC-1α, PPARα and eNOS protein and reduced endothelial superoxide. 6) Inhibition of PGC-1α changed the mRNA expression of genes regulated by miR-21, and overexpression of PGC-1α decreased the expression of miR-21 in high (25.5 mM) glucose treated coronary endothelial cells.

Conclusions:

Diabetic mice exhibit a NO-to-H₂O₂ switch in the mediator of coronary EDD, which contributes to microvascular dysfunction and is mediated by miR-21. This study represents the first mouse model recapitulating the NO-to-H₂O₂ switch seen in CAD patients in diabetes.

Introduction

Coronary artery disease (CAD) is a leading cause of mortality and morbidity in the United States. Although the principal focus of CAD research is on obstructive epicardial disease, there is increasing awareness of coronary microvascular dysfunction within the context of ischemic heart disease [1]. Coronary microvascular dysfunction is prevalent in diabetic patients and is independently correlated with cardiac mortality among patients with and without diabetes [27, 45]. Metabolic disorders, especially diabetes mellitus, increase the morbidity and mortality of many microvascular complications such as diabetic retinopathy [3, 37], diabetic nephropathy [8] and diabetic cardiomyopathy. The mortality of patients with CAD and diabetes mellitus increases 2-to 4-fold compared to those nondiabetic patients with CAD [3]. Diabetic patients also have a higher risk of developing CAD and metabolic disorders like hyperglycemia and dyslipidemia that could contribute to endothelial dysfunction by reducing nitrogen oxide (NO)'s bioavailability and increasing the level of reactive oxygen species (ROS) [31, 50].

NO is the major endothelium-derived relaxing factor (EDRF) and contributes to the endothelium-dependent dilation (EDD) in coronary arteries under normal conditions [3, 15]. Previous work revealed that the mediator of flow-induced vasodilation (FMD) in patients with CAD switches from NO-dependent to hydrogen peroxide (H₂O₂)-dependent. In contrast, no such switch occurred in aged patients without CAD [3, 6, 7, 27, 28]. Even though NO and H₂O₂ are both vasodilators, they each have inherent properties [3, 71]. H₂O₂ plays an essential role in pathological conditions such as atherosclerosis and hypertension, compensating for vasorelaxation in large vessels [34]. This NO-to-H₂O₂ transition in EDD might be an early pathologic step in the progression of CAD [38, 43, 56].

Increased ROS levels and reduced NO bioavailability are vital in the induction and progression of microvascular and cardiovascular complications during diabetes. It is well documented that in diabetic animals, EDD in coronary arterioles is impaired [4, 14, 51]. However, the underlying mechanism of this impairment is not entirely understood. The overlap of microvascular dysfunction in CAD and diabetic vascular diseases prompted us to explore whether the NO-to-H₂O₂ switch in mediating vasodilation reported in CAD patients also occurs in diabetes. This study tested this hypothesis in diabetic mouse models, focusing on aortic (macro) and coronary (micro) vasodilation.

To address the mechanism of the NO-to-H₂O₂ switch, we focused on a specific microRNA that we previously found to be upregulated in a diabetic rat model of coronary collateral growth (Supplementary Table 2). It is important to note that diabetes is associated with increased expression of microRNAs. These short non-coding RNA molecules bind to the 3' UTRs of target genes and regulate gene expression post-transcriptionally. MicroRNA-21 (miR-21) plays essential roles in metabolism [9, 27, 58, 75] and is enriched in endothelial cells, where it regulates NO production, cell proliferation, and apoptosis [70, 75], and modulates vascular diseases and remodeling [66, 68]. In this study, we investigated the role of miR-21 in the NO-to-H₂O₂ switch of the mediator of coronary EDD in diabetes by using a genetic knockout model.

Material and Methods

Animals:

All procedures were conducted with the approval of the Institutional Animal Care and Use Committee of Northeast Ohio Medical University and following National Institutes of Health Guidelines for the Care and Use of Laboratory Animals (NIH publication no. 85-23, revised 1996). Mice were housed in a temperature-controlled room with a 12:12-hour (h) light-dark cycle and maintained with access to food and water ad libitum. C57BL/6J wild-type (WT) mice, B6.BKS(D)-Lepr^db/J (db/db) mice and miR-21 null (miR-21^−/−) mice were purchased from the Jackson Laboratories (JAX). The miR-21^−/− mice from JAX were of mixed background with several backcrosses to C57BL/6NJ, and backcrosses were continued for at least 6 more generations onto the C57BL/6NJ background. Littermate WT mice and miR-21^−/− mice on a chow diet (WT and miR-21^−/− + chow) were used as controls. 6-week-old mice were fed a high fat/high sugar (HFHS) diet (Envigo #TD.88137 with 42.7% carbohydrate [mainly sucrose], 42% fat) for at least 5 months [63] to generate diet-induced diabetic mice (WT+ HFHS or miR-21^−/− + HFHS).

Cell culture:

Different lots of WT and db/db mouse coronary endothelial cells (CEC) were purchased from Cell Biologics; healthy human and diabetic patient coronary artery endothelial cells were purchased from Lonza. The cells were cultured in the medium supplied from the companies as instructed. Cells less than passage 6 were used in the experiments. For low and high glucose treatment, cells were cultured in media supplemented with 5.5 mM and 25.5 mM glucose for 72 h. For high lipid treatment, cells were cultured supplemented with 150 μM palmitate, linoleic acid, oleic acid, and 10 μg/ml cholesterol for 72 h with 1% serum.

Isolation of mouse CEC:

Mouse CEC were isolated with a modified protocol based on an online resource (http://vrd.bwh.harvard.edu/) and a publication [39]. Briefly, mouse hearts were dissected and minced into small pieces. After the digestion of the heart using Collagenase I (Worthington), cells were washed and incubated with Dynabeads conjugated with anti-CD31 antibody (Thermo Fisher). The beads with CEC were washed several times and cultured in a mouse endothelial culture medium (Cell Biologics). When confluent, cells were purified with Dynabeads conjugated with anti-Mouse CD102 (ICAM2) antibody. Each animal was used separate isolations, and cells less than passage 6 were used for the experiments.

For loss of function PGC-1α studies, we treated miR-21^+/+ or miR-21^−/− mouse CEC with either vehicle or a PGC-1α inhibitor SR-18292 (SR, MedChemExpress) at 20 μM for 72 h. SR-18292 induces PGC-1α acetylation [59]. Conversely, for the gain of function PGC-1α studies, we overexpressed PGC-1α in WT CEC using adenoviral Ad-PGC-1α and control Ad-GFP (generously provided by Dr. Pere Puigserver from Harvard Medical School) [65]. CEC were transduced with adenovirus at a multiplicity of infection (MOI) of 50 and treated with high glucose (25.5 mM) for 72 h.

Isolated vessel myograph studies:

Thoracic aortas were isolated, and segments were excised, cannulated and incubated in PSS buffer (145 mM NaCl, 4.7 mM KCl, 2.0 mM CaCl₂·2H₂O, 1.17 mM MgSO₄·7H₂O, 3.0 mM MOPS, 1.2 mM NaH₂PO₄-H₂O, 5.0 mM glucose, 2.0 mM pyruvic acid, 0.02 mM EDTA, pH 7.38-7.40) at 37°C in the bath chamber of MultiMyograph System 620M (DMT, Danish Myo Technology A/S). Small coronary arteries (100-200 μm internal diameter) were dissected from the epicardial surface of the left ventricle under a dissecting microscope. After these preparations, two small (20 μm diameter) wires were inserted into the arteries and connected to a force transducer. After tension optimization, arteries were precontracted with the thromboxane mimetic U46619 (1 μM). Once a stable contraction was achieved, pharmacological agents were administered in the bath, and cumulative concentration-response curves were obtained. EDD was measured after administration of acetylcholine (Ach), and endothelium-independent dilation was measured after administration of sodium nitroprusside (SNP). In some studies, pharmacological inhibitors were administered into the bath chamber (100 μM N^G-nitro-L-arginine-methyl ester [L-NAME], 500 units/ml polyethylene glycol-catalase [Peg-Cat], 10 μM Indomethacin) for 30 minutes (min) before recording vasodilation measurements. The area under the Ach relaxation curve was calculated as reported [11, 46, 61].

Nitrite and nitrate measurement:

Serum levels of nitrite and nitrate (NOx) were assessed using a commercial kit (Abcam # ab65328) according to the manufacturer's instructions. Briefly, serum was diluted and ultra-filtered through a 10 kD cutoff filter. The nitrate was converted to nitrite with nitrate reductase based on the Griess reaction. NOx concentrations were determined by measuring optical density (O.D.) at 540 nm with a SpectraMax Plus microplate spectrophotometer (Molecular Devices).

Measurement of H₂O₂:

Serum levels of H₂O₂ were measured using an H₂O₂ assay kit (Abcam #ab10250) according to the manufacturer's instructions. Briefly, the blood was collected via cardiac puncture when the animal was sacrificed. Serum was collected by centrifugation for 15 min at 1000 x g at 4° C, then immediately ultra-filtered through a 10 kD cutoff filter to deproteinize biological fluids. Aliquots were then snap-frozen in liquid nitrogen and transferred to −80° C for storage. Samples were thawed on ice when used for assays. The working solution containing the OxiRed probe and horseradish peroxidase (HRP) was added to serum samples. In the presence of HRP, the OxiRed probe reacts with H₂O₂, and the O.D. of the product was determined at 570 nm by a SpectraMax Plus microplate spectrophotometer (Molecular Devices). A standard curve was generated for each set of samples assayed.

Superoxide measurement by X-band electron paramagnetic resonance (EPR).

EPR spin-trapping measurements of superoxide from mouse CEC were performed with a Bruker EMX Micro spectrometer operating at 9.43 GHz, and the spin trap 5,5 dimethyl-1-pyrroline-N-oxide (DMPO) at a final concentration of 50 mM was used. To prevent any light-induced degradation, care was taken to keep the DMPO-containing solutions covered. The DMPO (ultrahigh purity) was purchased from Dojindo (Rockville, MD). The cell culture containing DMPO (50 mM) was transferred to a 50 μL capillary tube (Drummond Wiretrol, Broomall, PA) and loaded into the EPR resonator (HS cavity, Bruker Instrument, Billerica, MA). EPR spectra were recorded at room temperature with the following parameters: center field 3360 G, sweep width 100 G, power 20 mW, receiver gain 1×10⁵, modulation amplitude 1 G, conversion time 83 ms, time constant 327.68 ms, and numbers of scans 12. Measurements were performed at X-band with 100-kHz modulation frequency with 10-mW microwave power and modulation amplitude of 1.0 Gauss using an EMX HS resonator. The spectral simulations for SOD-sensitive DMPO/^•OH adducts spin quantitation were performed using the WinSim program developed at NIEHS by Duling [22] using the hyperfine coupling constants a^N=a_β^H=14.88 G. The intensity of EPR signal was presented by spin number calculated from double integration of simulated spectrum [30].

RNA isolation and quantitative real-time PCR (qPCR):

RNA was extracted from tissues and cells using Trizol Reagent (Thermo Fisher), miRNeasy or RNeasy Mini Kit (Qiagen), and mRNA levels of gene expression were quantified by qPCR using SYBR Green (GeneCopoeia) on a 7500 Real-Time PCR machine (Applied Biosystems). mRNA levels were normalized to ribosomal protein, large, P0 (36B4), and the relative fold change compared to the controls was calculated. Primers were designed and synthesized by IDT (Integrated DNA Technologies); sequences are listed in Supplementary Table 1. MicroRNAs were extracted from cells and tissues using miRNeasy Mini Kit (Qiagen). Reverse transcription was done with TaqMan MicroRNA Reverse Transcription Kit and quantified using a TaqMan MicroRNA assay kit and TaqMan Universal Master Mix II, no UNG (Thermo Fisher) for real-time-PCR, then normalized to small nuclear RNA U6 (U6), with fold change calculated and compared to the control as described [74].

RNA fluorescent in situ hybridization (FISH) of mouse tissues:

Freshly harvested mouse hearts were quickly perfused with cold 1X phosphate-buffered saline to clear the blood, fixed in 10% neutral buffered formalin at room temperature for 24 h, and washed with 1X phosphate-buffered saline. The sample was dehydrated using a standard ethanol series followed by xylene, then embedded in paraffin using standard procedures. 7-μm sections were prepared for experiments. According to the manufacturer's instruction, RNA FISH was performed with a miRNAscope™ Probe - SR-mmu-miR-21a-5p-S1 and miRNAscope™ Intro Pack HD Reagent Kit (Advanced Cell Diagnostics). Positive and negative controls were used. After RNA FISH, slides were incubated with Isolectin-B4 (1:00 dilution, Cat. # M121411, Thermo Fisher) at 4°C overnight, then counterstained with DAPI (Molecular Probes, Foster City, CA, USA) for nuclei before image acquisition. A confocal microscope captured the fluorescent signal, and the amount of miR-21 expression in endothelium was quantitated by the ratio of colocalization of miR-21 (red) and Isolectin-B4 (green) to the number of Isolectin-B4 (green) positive cells with Image-Pro Premier image analysis software [13].

Western blot:

Protein was extracted from mouse hearts, quantitated, run on a 10% or 12% pre-cast mini gel (BioRad), transferred and blotted with the Peroxisome proliferator-activated receptor (PPAR)-γ coactivator 1-alpha (PGC-1α, ab 54481), Peroxisome proliferator-activated receptor alpha (PPARα, ab 215270), Endothelial nitric oxide synthase (eNOS, sc-376751) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Thermo Fisher, #10941-1-AP) antibodies as described [74]. The protein expression levels were normalized to GAPDH.

Hemodynamics and myocardial blood flow (MBF) measurements and calculation of cardiac work (CW):

Heart rate (HR), mean arterial pressure (MAP), and MBF were measured as previously described [47, 48]. Briefly, mice were anesthetized, the right jugular vein was cannulated for intravenous (i.v.) drug and contrast echocardiography reagent infusions, and the femoral artery was cannulated with a 1.2F pressure catheter (Scisense Inc, Ontario, Canada) to measure MAP and HR. (Fig. 8a). CW is best represented by the triple product of stroke volume (SV) x HR x MAP and directly reflects myocardial oxygen consumption. As previously described, technical limitations regarding the simultaneous measurement of SV and MBF in mice lead us to employ the double product (DP = HR x MAP) index as a surrogate for oxygen consumption to decipher the basis of coronary metabolic dilation. The DP has been demonstrated to be a surrogate for CW, so "CW" notation will be used for the remainder of this manuscript. MBF was measured and calculated using 3-5 different images obtained from the same condition (baseline and treatments with norepinephrine [NE; 2.5 μg/kg/min and 5 μg/kg/min], L-NAME [100 mg/kg] or Peg-Cat [100 units/g] after hexamethonium [5 mg/kg]).

Fig. 8.

Myocardial blood flow (MBF) and cardiac work (CW) at baseline, and during norepinephrine-induced metabolic hyperemia. a. Workflow for NE-induced stress test and MBF measurements. Mice were given norepinephrine (NE) at 2.5 μg/kg/min (NE 2.5) or 5.0 μg/kg/min (NE 5.0) to induce hyperemia. b. MBF in wild-type (WT), db/db and WT+HFHS mice in the absence of inhibitors (n=5-6, *P < 0.05 vs. 3-month-old WT mice (3 mon)). c-e. MBF in WT (c), HFHS-fed WT (d) and db/db mice (e) treated with or without L-NAME or Peg-Cat during NE-induced metabolic hyperemia (n=5-7 mice/group, *P < 0.05 vs. WT 3 mo). f-i. Relationship between CW (double product of heart rate, beats/min and mean arterial pressure, mmHg) and MBF in mice shown in (b-e) (n=5-7). Min, minutes. Mo, month old. HFHS, high fat/high sugar. Two-way ANOVA (b-e) was performed, and linear regression (f-i) analysis was performed (*P < 0.05 vs WT 3 mo, ^# P < 0.05 vs WT+HFHS, ^$P < 0.05 vs db/db 3 mo)

BioRender software was used for Figure 8a with license number IE23AH7TRA.

Statistical Analysis:

Data have been presented as mean ± the standard deviation (SD). Data were analyzed using the PRISM 7.0 statistical software. A one-way or two-way analysis of variance (ANOVA) was made between groups. Comparisons between two groups were made using unpaired Student's t-test. The parametric test was used when the normal distribution test showed that the data distributions did not significantly differ from normal. Otherwise, a nonparametric test was used. Differences were considered statistically significant at a value of P <0.05.

Results

NO-mediated Ach-induced EDD (Ach-EDD) in aortic arteries from diabetic mice.

To determine the impact of diabetes on macrovascular function, we isolated thoracic aortic arteries from 9-month-old chow-fed WT mice, db/db mice, and diet-induced diabetic (WT+HFHS) mice for assessment of Ach-EDD using myography. Plasma lipid and glucose levels in WT mice on the HFHS diet were significantly increased compared to the WT mice on a chow diet (Suppl. Fig. 1a). A glucose tolerance test showed that WT mice on the HFHS diet exhibited a diabetic phenotype (Suppl. Fig. 1b). Compared to WT mice on a chow diet, Ach-EDD was decreased in thoracic aortic arteries from db/db mice and WT on an HFHS diet, and Ach-EDD was inhibited in all groups after treatment with the NO synthase inhibitor L-NAME (Fig. 1a). The endothelium-independent vasodilation response to the NO donor SNP was equivalent in all three groups (Fig. 1b). Together, these data indicate that both the diet-induced and genetic models of diabetes exhibited a diminished macrovascular Ach-EDD response inhibited by L-NAME, suggesting that NO mediates Ach-EDD in aortic arteries in both healthy and diabetic mice.

Fig. 1.

Aortic vasodilation in wild-type and diabetic mice by myography. a. Acetylcholine-induced endothelial dependent dilation (Ach-EDD) in aortic arteries from 9-month-old (9 mo) wild-type (WT) mice, db/db mice, and diet-induced diabetic mice (WT+HFHS) in the presence or absence of L-NAME (NO synthase inhibitor). b. Sodium nitroprusside (SNP)-induced (endothelium-independent) aortic vasodilation for each group. Mo, month old. HFHS, high fat/high sugar. Two-way ANOVA was used for statistical analysis (n=6 mice/group, **P < 0.01 vs. WT 9 mo, ^##P < 0.01 vs. db/db 9 mo, ^$$P < 0.01 vs. WT+HFHS, ^!!P < 0.01 vs. WT 9 mo, and ^&P < 0.05 vs. WT 9 mo)

The mediator of Ach-EDD in diabetic coronary arteries switches from NO to H₂O₂.

Given the differences between the macrovasculature and microvasculature, we sought to characterize the impact of age and diabetes on the mediator of Ach-EDD in coronary arteries. First, we isolated coronary arteries from 3-month, 9-month, and 32-month-old WT mice on a chow diet and assessed Ach-EDD using myography. L-NAME inhibited Ach-EDD in coronary arteries from 3-month (Fig. 2a), 9-month (Fig. 2b), and 32-month-old (Fig. 2c) mice on a chow diet. Second, we compared NO's role to other endothelium-derived relaxing factors (EDRF) on Ach-EDD by administering the H₂O₂ scavenger Peg-Cat and the prostaglandin (PGs) synthesis inhibitor Indomethacin. Neither Peg-Cat nor Indomethacin significantly affected Ach-EDD in the coronary arteries from WT mice on a chow diet of all ages (Fig. 2a-c). Endothelium-independent vasodilation mediated by SNP was not significantly altered in any group (Fig. 2d). The area under the curve calculations and a graphical summary demonstrate the relative contributions of three EDRF (PGs, NO, and H₂O₂) to Ach-EDD in coronary arteries from healthy mice of all ages (Fig. 2e-f) and NO is the major vasodilator. Together, these data demonstrate that L-NAME inhibited Ach-EDD in coronary arteries from healthy WT mice on a chow diet. This reliance on NO as the mediator of vasodilation was not impacted by age.

Fig. 2.

Coronary vasodilation in wild-type mice at different ages by myography. a-c. Acetylcholine-induced endothelial dependent dilation (Ach-EDD) in coronary arteries from 3-month-old (3 mo) (a), 9-month-old (9 mo) (b) and 32-month-old (32 mo) (c) wild-type (WT) mice in the presence or absence of L-NAME, Peg-Cat (H₂O₂ scavenger) or indomethacin (inhibitor of prostaglandin synthesis). d. Sodium nitroprusside (SNP)-induced coronary vasodilation for each group. e. The maximum Ach-EDD induced by vasodilators (prostaglandins (PGs), NO, and H₂O₂) in WT mice at different ages. f. The relative contribution of vasodilators (PGs, NO or H₂O₂) to Ach-EDD was calculated in WT mice at different ages. Mo, month old. Two-way ANOVA was used for all statistical analysis (n=6 mice/group, *P < 0.05, **P < 0.01 vs. WT 3 mo [a], vs. WT 9 mo [b], and vs. WT 32 mo [c])

After observing that L-NAME inhibited Ach-EDD in coronary arteries from all ages of healthy WT mice on a chow diet, we investigated Ach-EDD in diabetic mice and determined the reliance on specific EDRFs. Initially, we found that Ach-EDD was decreased in coronary arteries from 3-month and 9-month-old db/db mice compared to their age-matched WT controls (Fig. 3a-b). Significantly, neither L-NAME nor Indomethacin inhibited Ach-EDD in coronary arteries from db/db mice, but Peg-Cat did (Fig. 3a-b). Diet-induced diabetic mice (WT + HFHS) also exhibited decreased Ach-EDD compared to the control WT mice on a chow diet, further inhibited by Peg-Cat, but not L-NAME or Indomethacin (Fig. 3c). SNP-induced coronary dilation among all three groups was not significantly different (Fig. 3d). The area under the curve calculations and a graphical summary demonstrate the relative contributions of three EDRF (PGs, NO, and H₂O₂) to Ach-EDD in diabetic mouse coronary arteries (Fig. 3e-f) and H₂O₂ is the major vasodilator. These results revealed that both genetic and diet-induced diabetic mice exhibited impaired coronary artery Ach-EDD inhibited by Peg-Cat (H₂O₂-dependent), unlike that of the WT, which was inhibited by L-NAME (NO-dependent), suggesting H₂O₂ plays the dominant role in Ach-EDD in diabetic coronary arteries.

Fig. 3.

Coronary vasodilation in diabetic mice by myography. a-b. Acetylcholine-induced endothelial dependent dilation (Ach-EDD) in coronary arteries from 3-month-old (3 mo) (a) or 9-month-old (9 mo) (b) db/db mice, and age-matched wild-type (WT) controls in the presence or absence of L-NAME, Peg-Cat, or indomethacin. c. Ach-EDD in coronary arteries from 6-8-month-old WT mice fed HFHS diet (WT+HFHS) and age-matched WT controls in the presence or absence of L-NAME, Peg-Cat, or indomethacin. d. Sodium nitroprusside (SNP)-induced coronary vasodilation for each group. e. The maximum Ach-EDD induced by vasodilators (PGs, NO, and H₂O₂) in db/db mice or WT+HFHS mice. f. The relative contribution of individual vasodilators to Ach-EDD. Mo, month old. HFHS, high fat/high sugar. PGs, prostaglandins. Two-way ANOVA was used for all statistical analysis (n=6 mice/group, *P < 0.05, **P < 0.01 vs. db/db 3 mo [a], vs. db/db 9 mo [b], vs. WT+HFHS [c], or ^#P<0.05, ^##P<0.01 vs. age-matched WT control [a-b] or WT+HFHS [c])

miR-21 expression.

Because miR-21 has been implicated in metabolism and cardiovascular diseases, we sought to determine its expression in diabetes. Preliminary RNA-sequencing results showed upregulation of miR-21 in cardiac tissue from diabetic rats on an HFHS diet in a repetitive ischemia coronary collateral growth model (Suppl. Table 2). Therefore, we aimed to characterize miR-21 expression in our mouse and cell models by performing qPCR analysis. First, miR-21 was significantly upregulated in hearts and aortas from db/db mice compared to WT mice on a chow diet (Fig. 4a). Second, miR-21 was significantly upregulated in CEC isolated from db/db mice compared to WT (Fig. 4b). Third, miR-21 was significantly induced by high glucose (25.5 mM) treatment in healthy and diabetic human coronary artery endothelial cells (Fig. 4c). To check the expression of miR-21 in the mouse heart, we performed in situ hybridization, which revealed markedly enhanced expression of miR-21 in endothelial cells in db/db cardiac tissue sections compared to WT (Fig. 4d-e). Together, these data indicate an apparent upregulation of miR-21 in diabetic mouse hearts, endothelial cells (in vitro and in vivo), and cultured human coronary artery endothelial cells subjected to high glucose, suggesting an essential role for miR-21 in diabetes.

Fig. 4.

Induction of miR-21 expression in diabetes. miR-21 expression was determined by qRT-PCR in hearts and aortas from wild-type (WT) and db/db mice (n=5 mice/group) (a), coronary endothelial cells (CEC) from WT and db/db mice (n=3 mice/group) (b), and in healthy or diabetic human CEC treated with high glucose (HG, 25.5 mM) or low glucose (LG, 5.5 mM) (n=3 independent experiments) (c). d-e. miR-21 expression in the endothelium of cardiac tissue from WT or db/db mice was determined by RNAscope in situ hybridization (ISH) (n=4 mice/group). In the representative RNAscope ISH, arrows (white) point to miR-21 (red) expression in endothelial cells (Isolectin-B4, green) (d). The relative expression levels of miR-21 in endothelial cells were quantified by the ratio of miR-21⁺Isolectin-B4⁺ (colocalization of red and green) areas to the number of Isolectin-B4⁺ (green) cells by the Image-Pro Premier image analysis software (n=12 fields, 3 fields/animal). Two-tailed unpaired Student’s t-test was performed for statistical analysis (a, b, and e). Two-way ANOVA was performed for statistical analysis (c). *P < 0.05

The deficiency of miR-21 prevents the Ach-EDD mediator switch from NO to H₂O₂ in diabetic coronary arteries.

Because we observed a markedly increased expression of miR-21 in diabetes (Fig. 4), we sought to investigate the role that miR-21 holds in coronary microvascular function in diabetes. Therefore, we isolated coronary arteries from miR-21^−/− mice on a chow diet (miR-21^−/− + chow) and HFHS diet (miR-21^−/− + HFHS) and assessed Ach-EDD. L-NAME significantly inhibited Ach-EDD in coronary arteries from miR-21^−/− mice on a chow diet, but Peg-Cat and Indomethacin had no significant effect (Fig. 5a), suggesting that NO is the dominant mediator of Ach-EDD in coronary arteries from miR-21^−/− mice on a chow diet. Ach-EDD in coronary arteries from miR-21^−/− mice on an HFHS diet was improved compared to WT mice on an HFHS diet (WT + HFHS) (Fig. 5b). Interestingly, like that of WT and miR-21^−/− on a chow diet, L-NAME significantly inhibited the Ach-EDD in coronary arteries from miR-21^−/− mice fed HFHS diet, but Peg-Cat and Indomethacin did not (Fig. 5b). These data suggest that NO is the dominant mediator of Ach-EDD in coronary arteries from miR-21^−/− mice on an HFHS diet, like that of the WT and miR-21^−/− mice on a chow diet. SNP-induced coronary dilation was not significantly different among the groups (Fig. 5c), suggesting that smooth muscle responsiveness to NO was unchanged. The area under the curve calculations and a graphical summary demonstrate the relative contributions of three EDRF (PGs, NO, and H₂O₂) to Ach-EDD in coronary arteries of the three groups (Fig. 5d-e) and NO is the major vasodilator, suggesting that NO, rather than H₂O₂, plays the dominant role in Ach-EDD in miR-21^−/− mice on an HFHS diet; therefore, the deficiency of miR-21 prevented the NO-to-H₂O₂ switch in coronary arteries from diet-induced diabetic mice.

Fig. 5.

Coronary vasodilation in miR-21 null (miR-21^−/−) mice fed a chow or high fat/high sugar (HFHS) diet by myography. a. Acetylcholine-induced endothelial dependent dilation (Ach-EDD) in coronary arteries from 6–8-month-old chow-fed (a) and HFHS-fed (b) miR-21^−/− mice in the presence or absence of L-NAME, Peg-Cat, or indomethacin. c. Sodium nitroprusside (SNP)-induced coronary vasodilation. d. The maximum Ach-EDD induced by vasodilators (PGs, NO, and H₂O₂) in miR-21^−/− + chow, WT + HFHS, or miR-21^−/− HFHS mice. e. The relative contribution of individual vasodilators to Ach-EDD in miR-21^−/− + chow, WT + HFHS, or miR-21^−/− + HFHS mice. PGs, prostaglandins. Two-way ANOVA was performed for statistical analysis (n=6 mice/group, **P < 0.01 vs. miR-21^−/− + chow [a] and *P< 0.05, **P < 0.01 vs. miR-21^−/− +HFHS [b], or ^#P < 0.05, ^##P < 0.01 vs. WT+HFHS [b])

miR-21 regulates gene expression of vasodilation and ROS signaling pathways in diabetes.

To investigate the mechanism underlying the NO-to-H₂O₂ switch in the mediation of coronary artery vasodilation in diabetic mice, we profiled mRNA expression in isolated coronary arteries and endothelial cells using qPCR. We analyzed essential genes that regulate ROS signaling and endothelial function in diabetes. Our results show that when compared to WT coronary arteries, both db/db mice and WT mice on the HFHS diet expressed a significantly higher level of superoxide dismutase 1 (Sod1) and superoxide dismutase 2 (Sod2) (Fig. 6a). Furthermore, caveolin-1 (Cav1) and inducible NO synthase (iNos) were significantly upregulated in coronary arteries from WT mice on an HFHS diet compared to chow-fed WT mice (Fig. 6a). Catalase (Cat) was significantly upregulated in db/db mouse coronary arteries than chow-fed WT mice (Fig. 6a). We also found that peroxisome proliferator-activated receptor (Ppar)-γ coactivator 1-alpha (Pgc-1α) was downregulated in coronary arteries from both db/db and WT mice on an HFHS diet when compared with WT mice on a chow diet, but endothelial NO synthase (eNos) did not change (Fig. 6a).

Fig. 6.

Gene expression in mouse coronary arteries and coronary endothelial cells by qRT-PCR. a. Relative mRNA levels in coronary arteries from wild-type (WT) mice, WT mice fed an HFHS diet (WT+HFHS) or db/db mice (n=6 mice/group, **P < 0.01 vs. WT). b. Relative mRNA levels in coronary arteries HFHS diet-fed WT or miR-21^−/− mice (n=6 mice/group, ^#P < 0.05 vs. WT+HFHS). c. Relative mRNA levels in coronary endothelial cells (CEC) from WT and db/db mice (n=3 mice/group, **P < 0.01 vs. WT CEC). Pgc-1α, PPARγ coactivator 1 alpha. Sod1, superoxide dismutase 1. Sod2, superoxide dismutase 2. Cat, catalase. Cav1, caveolin 1. eNos, endothelial nitric oxide synthase. iNos, inducible nitric oxide synthase. One-way ANOVA (a) or unpaired Student’s t-test (b and c) was performed for the statistical analysis.

To investigate the effects of miR-21 on these genes, we also compared the gene expression profiles in coronary arteries from miR-21^−/− mice and WT mice on an HFHS diet. Coronary arteries from miR-21^−/− mice on an HFHS diet had significant reductions in the mRNA expression of the upregulated genes seen initially in coronary arteries from WT mice on the HFHS diet (Sod1, Sod2, Cav1, and iNos) and a significant elevation in the expression of the downregulated Pgc-1α (Fig. 6b). Additionally, Cat, a gene seen initially upregulated in coronary arteries from db/db mice was downregulated in coronary arteries from miR-21^−/− mice on an HFHS diet compared to WT mice on an HFHS diet.

To characterize the impact of diabetes on the expression of these genes in endothelial cells, we compared gene expression profiles in CEC from db/db and WT mice on a chow diet. We found that db/db CEC had significantly reduced eNos and elevated iNos, Cav1, and Sod2 compared to WT CEC (Fig. 6c).

Diabetic mice have elevated levels of NO and H₂O₂ in the serum.

To understand the relationship between NO and H₂O₂ bioavailability and their relative contributions during Ach-EDD in coronary arteries of diabetic mice, we assessed the production of NO and H₂O₂. The NO level was determined by measuring NOx (Nitrite/Nitrate), which is highly correlated with NO level in serum [69]. Our results show that serum levels of NO and H₂O₂ were increased in db/db and diet-induced diabetic mice compared to WT mice (Fig. 7a and 7b). Interestingly, the deficiency of miR-21 prevented this change in diet-induced diabetic mice (Fig. 7a-b).

Fig. 7.

Assessment of NO, H₂O₂, and superoxide levels. Serum levels of H₂O₂ (a) as well as nitrite and nitrate (NOx) metabolites (b) were measured in wild-type (WT), WT+HFHS, db/db and miR-21^−/− +HFHS mice (n=6 mice/group, **P < 0.01 vs. WT, ^##P < 0.01 vs. WT+HFHS). c-d. Coronary endothelial cells (CEC) were isolated from WT or miR-21^−/− mice and treated with high lipid (HL, 150 μM palmitate, linoleic acid, oleic acid, and 10 μg/ml cholesterol) or high glucose (HG, 25 mM) for 24 h. Superoxide detection was performed using DMPO (50 mM) spin-trapped and electron paramagnetic resonance (c), and superoxide levels were quantified (d) (n=4-6, **P < 0.01 vs. WT CEC, ^##P < 0.01 vs. WT CEC+HG). HFHS, high fat/high sugar. One-way (a-b) or two-way ANOVA (d) was performed for statistical analysis

miR-21 regulates ROS in mouse coronary endothelial cells.

To study the role of miR-21 on ROS regulation and to signal in endothelial cells under diabetic conditions, we isolated mouse coronary endothelial cells. We treated them with high glucose (25.5 mM) and high lipid (150 μM palmitate, linoleic, oleic acid and 10 μg/ml cholesterol) and superoxide was measured by electron paramagnetic resonance. We demonstrated that WT CEC subjected to high glucose exhibit significantly increased superoxide, and the CEC lacking miR-21 (miR-21^−/− CEC) had normalized superoxide levels under the same conditions (Fig. 7c-d). Mechanistically, this may partially explain why the deletion of miR-21 prevented the NO-to-H₂O₂ switch in diabetic mouse coronary artery Ach-EDD.

Implications of the NO-to-H₂O₂ switch for MBF regulation.

As our previous results (Fig. 2-3) have demonstrated, coronary artery EDD is mediated by NO under normal conditions, but the mediator shifts to H₂O₂ in diabetes. These ex vivo observations led us to hypothesize a functional consequence of this switch that alters MBF regulation in vivo. To determine the impact of the NO-to-H₂O₂ switch on the regulation of myocardial perfusion, we first measured MBF and hemodynamics (HR and MAP) at baseline (after hexamethonium) and during increasing doses of Norepinephrine (NE). Then, after i.v administration of Peg-Cat or L-NAME, we performed another set of MBF measurements (Fig. 8a). NE infusion enhanced myocardial metabolism via CW elevations in all groups. MAP and HR with each dosage are shown in Suppl. Fig. 2a and 2b. Note that MBF increased linearly with CW in all groups, but the hyperemic response (increased blood flow) was reduced in diabetic mice.

We compared MBF in WT mice on a chow diet, WT mice on an HFHS diet, and db/db mice during NE infusion (without Peg-Cat or L-NAME) (Fig. 8b), as well as comparing MBF with and without Peg-Cat or L-NAME for WT mice on a chow diet (Fig. 8c), WT mice on an HFHS diet (Fig. 8d), and db/db mice (Fig. 8e). Since MBF increases with enhanced CW in normal condition, we sought to analyze their relationships. Compared to WT mice on a chow diet, both db/db and WT mice fed HFHS diet exhibited an altered MBF-CW relationship (Fig. 8f). In db/db mice and diet-induced diabetic mice, MBF was significantly lower compared to WT mice during NE-induced metabolic hyperemia (Fig. 8f). Unsurprisingly, L-NAME attenuated NE-induced hyperemia in WT mice on a chow diet (Fig. 8g). Still, Peg-Cat, not L-NAME, attenuated the NE-induced hyperemia in db/db mice and diet-induced diabetic mice (Fig. 8h-i). These in vivo data corroborate our ex vivo data and confirm the diabetic NO-to-H₂O₂ switch in vivo. Altogether, we have shown functional deficits in MBF arising in diabetic mice that may result from a pathological NO-to-H₂O₂ switch in the mediator of vasodilation illustrated in Supplementary Figure 3.

Endothelial miR-21-PGC-1α regulatory axis.

As we have shown, diabetes induces various gene expression changes in coronary arteries and isolated CEC (Fig. 6a and c, respectively). PGC-1α is reported to be a critical regulator of ROS metabolism and mitochondrial homeostasis and has been implicated in the CAD-induced mediator switch [28, 29].

Notably, the deficiency of miR-21 prevented these gene expression changes in diet-induced diabetes (Fig. 6b). Therefore, we sought to investigate further the link between miR-21, PGC-1α, and the regulation of ROS and vasodilation pathway associated genes. First, we isolated CEC from miR-21^−/− and WT mice, subjected them to low glucose (5.5 mM) or high glucose (25.5 mM), and found that Pgc-1α was significantly upregulated in miR-21^−/− CEC under both conditions (Fig. 9a). Next, we subjected WT and miR-21^−/− CEC to high glucose with or without a PGC-1α inhibitor (SR-18292) and characterized their mRNA expression profiles. Initially, we found a significant upregulation of Pparα, reduced Sod2, reduced Cat, and elevated eNos in the miR-21^−/− CEC compared to WT CEC (Fig. 9c). When we treated the miR-21^−/− CEC with SR-18292, we found a significant reduction in Pparα, increased Sod2, increased Cat, and reduced eNos (Fig. 9c), recapitulating the diabetic expression changes we have shown previously (Fig. 6) and linking them to PGC-1α. Lastly, we overexpressed PGC-1α using an adenovirus (Ad-PGC-1α) in WT CEC and treated them with high glucose (25.5 mM) for 72 h. We then profiled miR-21 expression and found that PGC-1α overexpression significantly decreased miR-21 expression, suggesting a regulatory role of PGC-1α on miR-21 expression (Fig. 9b).

Fig. 9.

Gene expression in mouse coronary endothelial cells (CEC). a. Pgc-1α mRNA levels in wild-type (WT) and miR-21^−/− CEC in the presence of low glucose (LG; 5.5 mM) or high glucose (HG; 25.5 mM) (n=4-6 mice/group, *P < 0.05). b. Relative miR-21 expression levels in WT CEC transduced with adenovirus expressing GFP (Ad-GFP; control) or PGC-1α (Ad-PGC-1α) under HG for treatment for 72 h. (n=4-5, **P < 0.01). c. Pparα, Sod2, Cat and eNos expression in WT CEC treated with HG, and miR-21^−/− CEC treated with HG in the presence or absence of a PGC-1α inhibitor SR-18292 (SR; 20 μM) (n=5-11, *P< 0.05). Pgc-1α, PPARγ coactivator 1 alpha. Sod2, superoxide dismutase 2. Cat, catalase. eNos, endothelial nitric oxide synthase. Unpaired Student’s t-test (a, b) or one-way ANOVA (c) was used for statistical analysis

miR-21 regulates PGC-1α, PPARα and eNOS protein expression.

Our gene expression data in coronary arteries and CEC demonstrated that miR-21 regulated Pgc-1α and Pparα mRNA expression. To determine if miR-21 regulated the protein expression of PGC-1α, PPARα and eNOS, we performed western blotting in tissues from miR-21^−/− mice and WT mice. Due to the limited size of coronary arterioles and the low yield of coronary endothelium, we performed the western blotting in whole cardiac tissues. First, we found that eNOS was upregulated in miR-21^−/− mice compared to WT mice on a chow diet (Fig. 10a and c). Next, we found that the deficiency of miR-21 increased PGC-1α and PPARα expression in the mice on an HFHS diet (Fig. 10 b and d), suggesting that miR-21 regulates PGC-1α and PPARα signaling, which could contribute to the prevention of the diabetic NO-to-H₂O₂ switch. Functional protein association networks by STRING (https://string-db.org, Fig. 10e) reveal how PGC-1α and PPARα interact with the ROS-related genes that were shown via qPCR in Fig. 6, which further confirms our data and supports our hypothesis.

Fig. 10.

miR-21 regulates PPARα, eNOS, and PGC-1α expression. a-d. Western blot assays were performed using cardiac tissue lysates. Representative Western blot images are presented (a, b) and protein levels were quantified (c, d) (n=4 mice/group. *P < 0.05. A two-tailed unpaired Student’s t-test was performed for statistical analysis. e. Functional protein association networks via STRING shows the signaling pathway linking PGC-1α and PPARα to the ROS-related genes that were detailed by qPCR in Fig. 6

Discussion

A functional coronary microcirculation is essential for maintaining proper myocardial perfusion; hence impairments in coronary dilation may lead to perfusion insufficiencies and myocardial ischemia. It is well documented that coronary microvascular dysfunction occurs early during diabetic cardiomyopathy [33], but the underlying mechanism is still an enigma. The current study provides new insights into the regulation of diabetic coronary microcirculation. It demonstrates an aberrant shift of EDD mediator in the coronary arteries of db/db and diet-induced diabetic mice. While NO is the predominant mediator of Ach-EDD in most cases, our data suggest a NO-to-H₂O₂ switch in diabetic coronary arteries. To our knowledge, this is the first report that recapitulates clinical observations demonstrating a NO-to-H₂O₂ switch inflow-mediated EDD in CAD in diabetes [23, 43]. Although senescence is commonly linked to the endothelial dysfunction in diabetes [7], our data indicate that the NO-to-H₂O₂ switch in diabetic coronaries was not age-related, clearly showing that NO still mediated Ach-EDD in 32-month-old WT mice, while H₂O₂ mediated Ach-EDD in 3-month-old db/db mice. These results are consistent with clinical reports in which the switch did not occur in elderly healthy patients [7]. Notably, the deficiency of miR-21 prevented the NO-to-H₂O₂ switch in our diet-induced diabetic mice. This suggests that miR-21 is an essential regulator of the endothelial and coronary microvascular function in diabetes. The current study is also the first to verify the NO-to-H₂O₂ mediator switch in vivo. Most studies rely on the isolated microvessels, which we also included in our approach, but we further assessed the effect of the NO-to-H₂O₂ switch on the regulation of MBF in vivo. Our results also suggest enhanced metabolic demand seems unmet by the increased flow in diabetic mice, indicating a higher oxygen extraction rate. Consistent with this, a recent report demonstrated enhanced oxygen extraction, impaired oxygen utilization, and a shift towards anaerobic cardiac metabolism in a "triple hit" (streptozotocin, high-fat diet, and renal embolization) swine model [67]. This could have pathophysiological importance as we also show impaired EDD vasorelaxation and a shift towards H₂O₂-dependent vasodilation.

The balance of NO and H₂O₂ levels is vital for maintaining endothelial homeostasis and function. Under normal conditions, NO is the major EDRF and contributes to the EDD of coronary arteries. As demonstrated, NO is the dominant mediator of Ach-EDD in aortic arteries from WT and diabetic mice, which is consistent with the literature in that NO predominantly regulates the tone of large conduit vessels [41]. Our data also indicate that NO is the primary and dominant mediator of Ach-EDD in healthy coronary arteries but not in diabetic coronary arteries. Previous studies have reported the absence of L-NAME-mediated inhibition of EDD in coronary arteries from db/db mice [4, 14, 51]. However, we are the first to report that the primary mediator of diabetic mouse Ach-EDD is H₂O₂. When NO bioavailability declines in diabetes, H₂O₂ compensates to mediate the EDD. While we did not identify the source of H₂O₂ in this study, it will be an exciting topic for future studies. The adverse outcome of relying on H₂O₂ as a dilator in the coronary circulation might be coronary microvascular dysfunction, an early event for diabetic cardiomyopathy and CAD. In that way, the NO-to-H₂O₂ switch in EDD could be a pathological indicator of coronary microvascular dysfunction in diabetes.

As to the underlying mechanism of the diabetic mouse NO-to-H₂O₂ switch, our study suggests that an imbalance of NO and H₂O₂ might be causal. Compared to WT mice, the serum levels of NO and H₂O₂ are increased in diabetic mice, suggesting a pathological increase in the production of NO and H₂O₂. Levels of NO and H₂O₂ in the coronary arteries are essential for vasodilation but measuring NO in real-time is challenging due to biologic systems' short half-life (ms). NO is rapidly oxidized to nitrite and nitrate inorganic anions, which have been considered markers of NOS activity. This rapid oxidation partly regulates NO bioactivity, where nitrate is the predominant oxidation product in the circulation and diet is the main contributor of exogenous nitrate. In blood and tissues, nitrite can be reduced to NO and other bioactive nitrogen oxides [72]. Taken together, the measurement of NOx is an indirect but helpful assessment of NO. Our gene expression profiling also shows a change in genes' expression that regulates ROS mitigation and NO bioavailability (eNos, iNos, Cav1, Cat, Sod1, and Sod2) in diabetic mice [42, 60, 64].

How does miR-21 regulate the diabetic NO-to-H₂O₂ switch? It is reported that miR-21 is abundantly expressed in endothelial cells, where it governs NO production and ROS-homeostasis by targeting SOD2. miR-21 also plays essential roles in metabolism with implications in diabetic pathology [9, 57, 62, 75]. miR-21 has also been shown to be protective during diabetic cardiomyopathy-induced diastolic dysfunction [18] and confers vascular protection during ischemic preconditioning [73]. However, a miR-21 antagomir also restored the impaired process of coronary collateral growth in the rat model of metabolic syndrome [25, 26], and the deficiency of miR-21 improves the outcome of diabetic retinopathy [10]. This suggests that miR-21 is an essential regulator of vascular endothelial function. In this study, we used a genetic miR-21 knockout mouse model. We showed that the deficiency of miR-21 prevented the NO-to-H₂O₂ switch during Ach-EDD in diet-induced diabetic mouse coronaries, suggesting that miR-21 holds a vital role in regulating coronary endothelial function in diabetes. Moreover, compared to the wild-type mice, miR-21 expression was upregulated both in isolated CEC of db/db mice shown by qPCR and in the endothelial cells in diabetic cardiac tissue shown by in-situ hybridization. In the RNA scope ISH, we noticed that not all the signals of miR-21 co-localized with isolectin-B4, an endothelial cell marker, indicating that the miR-21 is also expressed outside of endothelial cells. However, these signals could have been produced in endothelial cells or other cell types—it is a dynamic process. Extracellular vesicles have been shown to transfer microRNAs and exosomal miR-21 is reported as a major cardioprotective paracrine factor produced by MSCs that could be transferred into the myocardium [40, 54]. miR-21 also contributes to exosome-mediated heart repair by enhancing angiogenesis and cardiomyocyte survival [52]. Moreover, macrophage miR-21 is a key molecule for the profibrotic role of cardiac macrophages [53]. Thus, other potential sources of miR-21 may contribute to the phenotype of the CEC observed in these studies.

The cross-talk of microRNA and transcription factors has been implicated in cardiac hypertrophy[17]. In this study, miR-21 regulates the mRNA and protein expression of PGC-1α and PPARα, two essential regulators of ROS signaling, metabolism, and diabetes. PGC-1α plays an important role in the microcirculation and is involved in the switch of NO-to-H₂O₂ during EDD in CAD patients. It is also reported to have a host of functions in regulating oxidative stress and ROS signaling [62], [12], [19], [20], [16], [35]. Notably, the loss of PGC-1α in non-CAD coronary arterioles produced a diseased (CAD) phenotype characterized by a NO-to-H₂O₂ shift in FMD [29]. Interestingly, our data show that Pgc-1α mRNA expression was significantly reduced in coronary arteries from WT+HFHS and db/db mice, and Pgc-1α is upregulated in miR-21^−/− coronaries and endothelial cells. Our study using CEC from WT and miR-21^−/− mice showed that Pgc-1α expression was upregulated in miR-21^−/− endothelial cells under low glucose (5.5 mM) and high glucose (25.5 mM) compared to the WT. These data suggest PGC-1α signaling to be involved in the NO-to-H₂O₂ switch in EDD in diabetic coronary arteries. Furthermore, our gain and loss of function studies of PGC-1α demonstrate the regulatory role of miR-21 in ROS gene expression is dependent on PGC-1α and overexpression of PGC-1α inhibited miR-21 expression, suggesting that an endothelial miR-21 - PGC-1α regulatory axis is involved in the NO-to-H₂O₂ switch by regulating ROS signaling in diabetic coronary arteries.

We therefore posit that miR-21 may represent a novel therapeutic target for the treatment of coronary microvascular diseases in diabetes. Notably, the use of miR-21 antagomirs will be feasible for clinical studies to influence the pathological NO-to-H₂O₂ switch, thereby improving coronary microvascular function and possibly preventing diabetic cardiomyopathy progression. miR-21 has been proposed as a therapeutic target in hypertension and heart failure [5, 36], and anti-miR-21 oligonucleotides have been used as inhibitors for studies of heart failure and hypertension. RG-012 is a relevant example of a miR-21 inhibitor used in patients with kidney fibrosis [24]. Still, the confounding nature of miR-21 in the context of various diseases requires scrutiny so that off-targeted effects are limited. Systemic delivery lacks tissue-specific targets, so attempts to localize miR-21 inhibition may be of interest, particularly in cases of overlapping diseases where its roles may be opposing.

While we acknowledge the limitations associated with our current study and understand the importance of clinical/translational studies that foster improvements in medicine. Mouse/cell models such as ours provide highly viable and accessible means to study mechanisms and pathways via gain- and loss-of-function genetic manipulations that would be impossible for human studies. Novel disease targets discovered in basic cardiovascular research will eventually need to be confirmed in patients. One aspect of our study deserves particular emphasis: our model of the NO-to-H₂O₂ switch in mediating EDD in diabetes mimics the finding in human patients with CAD [6, 23]. However, there are some aspects of our study that bear further consideration. One pertains to the models of type 2 diabetes used. While various mouse models of diabetes exist, db/db mice are often used as a genetic means to recapitulate diabetes and begin exhibiting a phenotype as early as 6 weeks of age. Although this appears to be wholly unnatural and without parallel in humans, we note that type 2 diabetes is a rising threat in youth [2, 32, 55]. However, the db/db mouse has been criticized as a diabetic model due to abnormal leptin signaling. Therefore, we also utilized a diet-induced diabetic mouse model that better reflects those patients who consume an unhealthy diet (high in fat and sugar) and progress to the development of type 2 diabetes [21, 49, 63]. Like these patients, our mouse model requires a more extended period (5-6 months) to display a diabetic phenotype, especially concerning metabolic derangements and cardiovascular dysfunction.

Regarding our in vitro studies, while the use of human coronary artery endothelial cells purchased from Lonza and more patient samples undoubtedly would have improved the study, acquiring clinical specimens is challenging, especially obtaining live cardiac tissue. Additionally, in the current study we utilized Ach to induce coronary dilation via myography, a method which emphasizes NO, central to endothelial dysfunction, rather than using another method like FMD, which utilizes pressure gradients to produce flow elicited vasodilation [44]. FMD may be more physiological, but difficulties arise when dealing with mouse coronaries, which are highly branched (unlike aortic or mesenteric arteries) and prove challenging to study via FMD. Also, while our dosage of Ach may not be physiological, clinical studies have utilized Ach as an endothelium-dependent probe to measure the coronary flow reserve in patients using a dose ranging from 10⁻⁶ to 10⁻⁴ μM with an infusion rate of 48-120 ml/h for 3 min (equivalent to 0.364 to 108 μg) [14]. Our experiments were done using an Ach concentration ranging from 10⁻⁹ to 10⁻⁵ μM, like a clinical setting. Overall, these limitations do not affect our significant findings in the current study.

In summary, we report a NO-to-H₂O₂ switch in the mediator of Ach-EDD in coronary arteries of db/db mice and diet-induced diabetic mice. The first mouse model recapitulates clinical observation of the NO-to-H₂O₂ switch in EDD in CAD patients in diabetes. More importantly, our study shows that this switch also occurs in vivo, thereby compromising MBF in diabetes. The deficiency of miR-21 could prevent the NO-to-H₂O₂ switch in diet-induced diabetic mice, which suggests that miR-21 might be a new therapeutic target in coronary microvascular diseases and microvascular dysfunction in diabetes.