Downregulation of Erythrocyte miR-210 Induces Endothelial Dysfunction in Type 2 Diabetes

Cardiovascular complications are major clinical problems in type 2 diabetes mellitus (T2DM), and patients with T2DM have poorer clinical outcomes following a cardiovascular event such as myocardial infarction (1). Endothelial dysfunction plays a major role in the etiology of vascular complications in T2DM due to reduced nitric oxide (NO) bioactivity and formation of reactive oxygen species (ROS) promoting proatherosclerotic processes (2,3). However, key events triggering the endothelial dysfunction in T2DM remain to be clearly identified.

Red blood cells (RBC) play a crucial role in cardiovascular homeostasis and contribute to vascular function independently of their function as oxygen carriers (4,5). RBC undergo several functional changes in T2DM including decreased export of NO bioactivity, impaired release of ATP, and increased formation of ROS (5–8). We recently demonstrated that RBC from patients with T2DM and documented endothelial dysfunction in vivo induce endothelial dysfunction in an ex vivo model of isolated aortas from rats and internal mammary arteries from patients without diabetes (9). These observations suggest that the RBC is a trigger and mediator of vascular complications in T2DM, but key mechanisms underlying the interaction of RBC with the vasculature remain largely unknown. miRNAs are 19–25 nucleotides long noncoding RNAs that have emerged as fundamental posttranscriptional regulators of gene expression and function in cardiometabolic diseases including T2DM (10,11). Due to their stability in the circulation and conservation across species, miRNAs can serve as biomarkers as well as mediators of vascular complications in T2DM (12–14). RBC, which constitute ∼45% of the blood volume (hematocrit), are known to contain abundant and diverse miRNAs (15). The expression and function of RBC miRNAs in T2DM have not been explored, however.

Recent studies demonstrated that miR-210 plays a crucial role in maintaining cardiovascular homeostasis. miR-210 acts as a protective factor against vascular (16) and cardiac ischemic injury in mice (17,18). Recent findings showed a significant protective role for miR-210 in enhancing fibrous cap stability in advanced atherosclerotic lesions (19). Of interest, miR-210 levels are decreased in whole blood from T2DM mice (20). It is therefore conceivable that reduced levels of the protective miR-210 are a mechanism contributing to the susceptibility of vascular injury in T2DM. However, the expression of miR-210 in RBC remains unexplored and it is unknown whether dysregulated miR-210 expression in RBC is of importance for endothelial dysfunction in T2DM.

Consequently, we tested the hypothesis that endothelial dysfunction induced by RBC in T2DM is associated with the loss of the protective function of miR-210 in RBC. Using a translational approach, we demonstrate a significant role of downregulated RBC miR-210 in the development of endothelial dysfunction in T2DM through a functional interaction between RBC miR-210 and vascular signaling.

The study design is presented in Supplementary Fig. 1. A total of 36 patients with T2DM were recruited from the Departments of Endocrinology, Karolinska University Hospital and Danderyd Hospital, Stockholm, Sweden, during 2016–2021. T2DM was defined according to the World Health Organization criteria. Thirty-two healthy control subjects free of medication, with no history of cardiovascular disease and with normal fasting glucose, were recruited. The procedure was conducted according to the principles outlined by the Declaration of Helsinki and was approved by the regional ethics review board in Stockholm. All subjects were informed of the purpose and gave their oral and written informed consent.

Animal care and all protocols were approved by the regional ethics committee (17708-2019) and conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publ. no. 85-23, revised 1996). Male Wistar rats and T2DM db/db mice age 15–20 weeks (Charles River Laboratories, Sulzfeld, Germany) were anesthetized with pentobarbital (50 and 100 mg/kg i.p., respectively.) followed by thoracotomy and isolation of thoracic aortic segments. Male T2DM Goto-Kakizaki (GK) rats, miR-210 knockout (KO) mice, and corresponding miR-210 wild-type (WT) littermates were bred at the animal facility of the Karolinska University Hospital and included at 15–30 weeks of age. miR-210 KO mice, characterized previously (21), were provided by Mircea Ivan (Indiana University School of Medicine).

Whole blood was collected in EDTA tubes from a cubital vein of subjects or the thoracic cavity of animals after removal of the heart. RBC were isolated by centrifugation at 4°C and 1,000g (human and rat samples) or 300g (mouse samples) for 10 min followed by three washing cycles with Krebs-Henseleit buffer. Successful removal of white blood cells (>99%) and platelets (≥98%) was verified previously (22). All samples with hemolysis were excluded.

RBC were diluted to a hematocrit of ∼45% with serum-free culture medium (DMEM; Gibco, Waltham, MA) and coincubated with rat and mouse aortas in cell culture incubator at 37°C with 95% O₂ and 5% CO₂ for 18 and 4 h, respectively. Pilot experiments revealed this incubation time to be optimal for rat and mouse RBC. Following the incubation, vessels were mounted in myographs (Danish Myo Technology) in separate organ baths containing 6 mL Krebs-Henseleit buffer. Contractility of vessels was examined with KCl twice (50 and 100 mmol/L for rat aortas; 50 mmol/L for mouse aortas). Thereafter, the vessels were preconstricted with the thromboxane A₂ analog U46619 (30 nmol/L) or phenylephrine (1 μmol/L) before endothelium-dependent (EDR) and endothelium-independent (EIR) relaxations were determined with the administration of cumulatively increasing concentrations of acetylcholine (ACh) and sodium nitroprusside, respectively (9,23). ACh-induced EDR in aortas is completely NOS dependent (24,25).

For investigation of the function of miR-210, miR-210-3p mimic, miR-210-3p inhibitor, or scrambled oligonucleotides (50 nmol/L; Ambion [supplier Thermo Fisher Scientific, Waltham, MA]) were combined with Lipofectamine RNAiMAX, diluted in Opti-MEM, and transfected during the coincubation of human RBC with the rat aorta for 18 h. Following the incubation, segments were rinsed and EDR was determined. The transfection was also applied in the coincubation system where aortas from GK or Wistar rats were incubated with medium only. RBC from miR-210 KO mice and WT littermates were incubated with aortas isolated from WT mice followed by evaluation of EDR and EIR. For investigation of the effect of miR-210 modulation in vivo, WT mice were injected with miR-210 scrambled oligonucleotides and db/db mice were injected with miR-210 mimic or scramble (0.5 mg/kg i.v.; Dharmacon, Lafayette, CO). A phospholipid-oil emulsion formed by combination of the miRNA mimic and the MaxSuppressor In Vivo RNA-LANCEr II agent (Bioo Scientific, Austin, TX) was dispersed in an aqueous solution with use of a lipid extruder. The mice were sacrificed 72 h after the injections, and the RBC were isolated and incubated with aortas from WT mice for EDR and immunohistochemistry. In separate experiments, protein tyrosine phosphatase 1B (PTP1B) inhibitor (10 μmol/L; Calbiochem, Merck, Germany) or the antioxidant N-acetylcysteine (NAC) (10 μmol/L; Sigma-Aldrich, Schnelldorf, Germany) was added in the organ baths after the RBC incubation followed by determination of EDR.

Recipient Wistar rats were anesthetized with pentobarbital 50 mg/kg i.p., tracheotomized, and ventilated via a tracheostomy. RBC (1 mL) from age-matched donor rats (GK and Wistar) were resuspended in 1 mL PBS and injected into the carotid artery of recipients before removal of 2 mL blood from the recipient rat (26). After 4 h, the animal was sacrificed and the aorta was harvested for determination of EDR.

Human carotid arterial endothelial cells (HCATECs) (Lonza, Basel, Switzerland) were cultured in endothelial cell growth medium (Cell Application, San Diego, CA) containing 100 units/mL penicillin-streptomycin in precoated (0.1% gelatin) 75 cm² cell culture flasks. After distribution of cells into six-well culture plates, HCATEC between passages 6 and 8 were treated with RBC (hematocrit 1%) or vehicle when reaching 80% confluence. After 24 h coincubation at 37°C and 5% CO₂, HCATEC were washed more than five times with PBS and harvested for subsequent analyses.

Following RBC incubation, rat or mouse aortic rings were fixed for 24 h in 4% formaldehyde, hydrated in graded ethanol, embedded in paraffin, sectioned, and mounted on coated glass slides (SuperFrost Plus; Thermo Fisher Scientific). For antigen retrieval, slides were subjected to high-pressure boiling in citrate buffer (pH 6.0). After peroxidase inactivation (0.3%) and blockade with goat serum (Abcam, Cambridge, U.K.), aorta cross-sections were incubated overnight (4°C) with the following primary antibodies: mouse monoclonal anti–4-hydroxynonenal (4-HNE) antibody (1:100 dilution, IgG_2b, catalog no. MAB3249; R&D Systems, Minneapolis, MN) and rabbit monoclonal anti-PTP1B antibody (1:100 dilution, IgG, catalog no. ab244207; Abcam). Specific labeling was detected with labeled horseradish peroxidase polymer conjugate as secondary antibody as part of the EnVision+ Dual Link System HRP (Dako [supplier Agilent Technologies, Santa Clara, CA]). Isotype controls were used as negative controls (rabbit IgG or mouse IgG_2b; Abcam). Fields were captured (Leica DM3000 digital microscope; Leica Biosystems, Wetzlar, Germany) and analyzed (ImageJ software 1.53a, Bethesda, MD).

Human carotid artery plaques from patients with T2DM and patients without diabetes obtained from the Munich Vascular Biobank were used for miR-210 expression analysis. The material was fixed for 48 h in 2% zinc-paraformaldehyde, embedded in paraffin, and cut into 5-µm sections per sample. Four sections were stained for hematoxylin-eosin as well as Elastica van Gieson staining. After dehydration, proteinase K treatment was performed followed by hybridization at probe-specific temperatures (miR-210 at 62°C, scrambled negative control, and U6 positive control at 54°C) for 2 h with 50 nmol/L scramble miR-210 and 4 nmol/L U6. Detection was performed by recognition of the DIG-labeled LNA Probes (QIAGEN, Hilden, Germany) with an anti–DIG-AP–coupled antibody (Roche, Basel, Switzerland). Counterstaining was performed with Nuclear Fast Red (Sigma-Aldrich).

Mice were anesthetized with isoflurane and the right carotid artery was isolated, cannulated (PE-10 catheter), and connected to a pressure transducer. After stabilization, heart rate and systolic and diastolic pressure values were registered with PharmLab V5.0 (AstraZeneca R&D, Mölndal, Sweden).

RNA extraction was performed with the QIAGEN miRNeasy Mini (for RBC) and QIAGEN miRNeasy Serum/Plasma Advanced Kit (for plasma). Total RNA concentration was assessed with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific) and diluted in RNAse-free water to a concentration of 2 ng/µL. We used 10 and 100 ng for cDNA synthesis with the TaqMan MicroRNA Reverse Transcription Kit and the TaqMan High-Capacity cDNA Transcription Kit, respectively. PCR amplification was performed in a 7900 HT real-time PCR system (Applied Biosystems, Waltham, MA) using TaqMan Universal PCR Master Mix with UNG (Applied Biosystems) and TaqMan Gene Expression Assays (miR-210 000512, miR-16 000391, PTP1B Hs00942477_m1, U6 001973, RNU48 001006, and RPLPO Hs99999902_m1). Results were normalized to the equal mass of total RNA and the Ct values of miR-16, U6, RNU48, or RPLPO. The relative amount of expression was calculated with the 2^−ΔΔCt method and presented as fold change.

Data are presented as mean ± SD. Differences in concentration-dependent relaxations were analyzed using two-way ANOVA with repeated measures. Multiple comparisons were performed with one-way ANOVA or two-way ANOVA followed by post hoc analysis using the Bonferroni test. Differences between two groups were performed with paired or unpaired two-tailed t test, nonparametric Mann-Whitney U test, or Fisher exact test when appropriate. Normal distribution of data was tested with the D’Agostino-Pearson test. The number of experimental observations (n) refers to the number of animals, cell culture experiments, and RBC donors included. We used one RBC donor for one rat donor. All statistical analysis was calculated with GraphPad Prism 6.05. Statistical significance was accepted when P < 0.05.

The data analyzed during the current study are available from the corresponding author on reasonable request.

Characteristics of subjects included for the collection of RBC and for the measurement of miR-210 expression using in situ hybridization are presented in Table 1 and Supplementary Table 1, respectively.

Incubation with RBC from patients with T2DM impaired EDR but not EIR in rat aortas, indicating the development of endothelial dysfunction (Fig. 1A and B), without affecting smooth muscle function (Fig. 1C). For confirmation that this ex vivo effect of RBC also occurs under in vivo conditions, RBC from GK or healthy Wistar rats were transfused to age-matched Wistar rats in vivo. EDR in aortas isolated from the recipients transfused with RBC from GK rats was significantly impaired in comparison with that of aortas from recipients transfused with Wistar RBC (Fig. 1D).

The expression of miR-210 in RBC from healthy subjects was several hundred–fold higher than in HCATEC (Supplementary Fig. 2A). miR-210 expression was significantly lower in RBC and plasma from patients with T2DM compared with healthy subjects (Fig. 2A and B and Supplementary Fig. 2B). miR-210 expression was lower in RBC from patients with T2DM compared with RBC from healthy subjects also 18 h after transfection with miR-210 scrambled oligonucleotide (Supplementary Fig. 2C). Transfection with miR-210 mimic elevated miR-210 expression in RBC (Fig. 2C) and aortas to a similar extent (Supplementary Fig. 2D). Transfection with miR-210 inhibitor decreased miR-210 expression in RBC from healthy subjects (Fig. 2D) but did not affect miR-210 expression in the coincubated aortas (Supplementary Fig. 2E). In contrast, transfection with miR-210 mimic did not alter miR-210 expression in GK aortas incubated with medium (Supplementary Fig. 2F). This indicates that the transfection more effectively targets RBC than the aortas in the coincubation system and that the change in miR-210 levels in RBC may determine the vascular expression and function. Indeed, the impairment in EDR induced by RBC from patients with T2DM was attenuated when the RBC were transfected with miR-210 but not when exposed to miR-210 scrambled oligonucleotide (Fig. 2E). Conversely, inhibition of miR-210 in RBC from healthy subjects resulted in impairment of EDR (Fig. 2F).

The protective role of RBC miR-210 was further confirmed in miR-210 KO mice. For RBC and aortas from miR-210 KO mice Ct values were 39 ± 1.8 and 36 ± 1.0, respectively, vs. 27 ± 1.0 and 26 ± 0.06 in WT mice (n = 3). Heart rate and systolic blood pressure did not differ between miR-210 KO and WT mice (Supplementary Fig. 3A and B). Both EDR and EIR were markedly impaired in aortas isolated from miR-210 KO mice versus WT mice (Supplementary Fig. 3C and D). Of note, RBC from miR-210 KO mice induced impairment of EDR in aortas from healthy mice in contrast to RBC from WT mice or following incubation with medium only (Fig. 3A). EIR was comparable in mouse aortas from the different incubation groups (Fig. 3B). These observations indicate that the downregulation of miR-210 in RBC signals the development of endothelial dysfunction in T2DM.

We further investigated whether RBC from patients with T2DM affect expression of miR-210 in the endothelium. RBC from patients with T2DM, but not from healthy subjects, reduced the expression of miR-210 in HCATEC (Fig. 3C). To translate this to the clinical setting, we observed that advanced carotid artery plaques from patients with T2DM had significantly lower levels of miR-210 in comparison with plaques from patients without diabetes, with miR-210 primarily localized in the fibrous cap area (Fig. 3D and E and Supplementary Fig. 4).

PTP1B, an enzyme regulating insulin signaling and known to be a target of miR-210 (27), is involved in the development of endothelial dysfunction in diabetes (28,29). The PTP1B inhibitor reversed the impaired EDR in the aorta from GK rats but had no effect on EDR in aortas from Wistar rats (Supplementary Fig. 5A–C), confirming a role of PTP1B in the regulation of endothelial function in diabetes. Interestingly, RBC from patients with T2DM significantly increased mRNA expression of PTP1B in HCATEC compared with RBC from healthy subjects or medium (Fig. 4A) and increased protein expression of PTP1B in rat aortas compared with RBC from healthy control subjects (Fig. 4B and C). This alteration is of functional importance, which is evident from the observation that the impairment in EDR induced by RBC from patients with T2DM was attenuated by inhibition of vascular PTP1B (Fig. 4D). By contrast, PTP1B inhibition had no effects on EDR in rat aortas incubated with RBC from healthy subjects or medium (Fig. 4E and F). Similarly, inhibition of vascular PTP1B attenuated the impairment of EDR in aortas of recipient rats transfused with RBC from GK rats but not RBC from Wistar rats (Fig. 4G and H).

Although it has been shown that PTP1B is a direct target of miR-210 (27), the functional interaction between miR-210 and PTP1B in the regulation of endothelial function has not yet been investigated. We observed that impaired EDR in aortas from miR-210 KO mice was recovered by PTP1B inhibition, while PTP1B inhibition had no effect on EDR in aortas from WT mice (Supplementary Fig. 6A and B). Importantly, knocking down miR-210 in RBC from healthy subjects increased the expression levels of PTP1B and 4-HNE (a marker of oxidative stress) in incubated rat aortas, while overexpressing miR-210 in RBC from T2DM patients decreased the expression levels of these markers (Fig. 5A–D). Impairment of EDR in mouse aortas induced by RBC from miR-210 KO mice was attenuated by inhibition of vascular PTP1B and the antioxidant NAC (Fig. 5E and G), while these inhibitors had no effects on EDR in aortas incubated with RBC from WT mice (Fig. 5F and H). These observations indicate that endothelial dysfunction induced by downregulation of RBC miR-210 is mediated via PTP1B and ROS.

Next, we investigated whether rescue of miR-210 in vivo prevents the RBC-induced endothelial dysfunction in diabetes. Administration of miR-210 mimic to db/db mice elevated miR-210 levels in RBC compared with scramble (Fig. 6A). EDR in aorta from WT mice was impaired and vascular PTP1B and 4-HNE expression were increased following incubation with RBC from db/db mice given scramble in comparison with RBC from WT mice treated with scramble (Fig. 6B–D). This negative effect of RBC from db/db mice on EDR, PTP1B, and 4-HNE expressions was completely prevented when WT mouse aortas were incubated with RBC from db/db mice that were treated with miR-210 mimic in vivo (Fig. 6B–D).

Using a translational approach and multiple experimental models, we demonstrate a novel disease mechanism in T2DM by which downregulation of RBC miR-210 induces endothelial dysfunction via interaction with vascular PTP1B and ROS (Fig. 7). RBC from patients with diabetes had lower miR-210 expression, and RBC from humans and rodents with T2DM induced endothelial dysfunction ex vivo and in vivo. Rescue of RBC miR-210 expression attenuated the endothelial dysfunction induced by RBC in T2DM. RBC from patients with T2DM decreased miR-210 expression in endothelial cells, and miR-210 expression was lower in carotid plaques from patients with T2DM. Moreover, RBC in T2DM upregulated vascular PTP1B and ROS, resulting in endothelial dysfunction. Finally, rescue of miR-210 in vivo prevented the development of endothelial dysfunction induced by RBC from T2DM mice, identifying RBC miR-210 as a potential target for the prevention of endothelial dysfunction in T2DM.

Accumulating evidence suggests that RBC contribute to vascular homeostasis and integrity in addition to their function as gas transporters (4,9,30–32). It is of interest that the function of RBC is altered in several pathophysiological conditions and contributes to disease development (5). In accordance with our recent findings (9,23), altered function of RBC from patients with T2DM induced endothelial dysfunction without affecting smooth muscle function. We previously showed that RBC from T2DM patients with endothelial dysfunction in vivo cause endothelial dysfunction in isolated human internal mammary arteries and rat aortas ex vivo (9) and transfusion of GK RBC into healthy rats induces endothelial dysfunction (26). Impaired release of ATP from RBC in T2DM has also been found to be associated with reduced vasodilation in isolated hamster arterioles (7). These observations confirm that RBC cause endothelial dysfunction in T2DM in various types of vascular beds both ex vivo and in vivo, independently of species.

Circulating miRNAs including miR-21, miR-29b, and miR-126 have been proposed to be diagnostic biomarkers and associated with vascular complications in T2DM (12,13). RBC contain abundant and diverse miRNAs (15), but it is unknown whether RBC miRNAs play a role in the regulation of vascular function in T2DM. Recent studies demonstrated that miR-210 plays a crucial protective role in cardiovascular disease situations. Thus, miR-210 overexpression attenuates vascular and cardiac injury induced by ischemia-reperfusion (16) and hydrogen peroxide (17,18). In patients with unstable carotid artery plaques, miR-210 is the most significantly downregulated of eight deregulated miRNAs in local plasma samples. miR-210 is distinctly localized in the fibrous caps and is suggested to stabilize advanced atherosclerotic lesions (19). In the current study, we found that miR-210 levels were lower in both RBC and plasma from patients with T2DM in comparison with healthy subjects. Together with the observation that miR-210 levels are much higher in RBC than in endothelial cells, and the report that the whole blood miR-210 levels are lower in T2DM mice (20), our findings suggest that RBC are a major source for circulating miR-210. A major finding in our study is that rescue of miR-210 levels in RBC from patients with T2DM attenuated endothelial dysfunction, while inhibition of miR-210 in RBC from healthy subjects led to impairment of endothelial function. The protective role of RBC miR-210 was further confirmed in miR-210 KO mice. Further, in vivo administration of miR-210 mimic to db/db mice, which resulted in elevation of miR-210 levels in db/db RBC, completely prevented the development of endothelial dysfunction induced by RBC from these db/db mice. These observations reveal that downregulation of RBC miR-210 plays a significant role in the development of endothelial dysfunction in T2DM and that RBC miR-210 may serve as a potential therapeutic target for endothelial dysfunction in T2DM.

We further investigated the molecular changes in the endothelium induced by RBC from patients with T2DM. PTP1B, an enzyme regulating insulin signaling, is a direct target of miR-210 (27), and miR-210 negatively regulates PTP1B expression in cardiomyocytes (33). It has been shown to be upregulated in diabetes and to be of importance for the development of endothelial dysfunction via ROS formation in diabetes (29). Several findings in our study suggest that PTP1B and ROS are involved in endothelial dysfunction triggered by the downregulation of miR-210. First, pharmacological inhibition of PTP1B restored endothelial function in aortas isolated from GK rats, indicating a functional role of PTP1B in the regulation of endothelial function in T2DM. Second, RBC from patients with T2DM induced an increase in mRNA PTP1B levels in endothelial cells and protein levels in rat aortas. Third, inhibition of vascular PTP1B prevented the development of endothelial dysfunction induced by RBC from patients with T2DM ex vivo and by RBC transfused from GK rats in vivo. Fourth, inhibition of PTP1B and reduction of oxidative stress attenuated endothelial dysfunction in mouse aortas induced by RBC from miR-210 KO. Fifth, overexpression of miR-210 in RBC from T2DM patients in vitro and in db/db mice in vivo decreased the expression of vascular PTP1B and 4-HNE, a marker of oxidative stress (34), while knocking down miR-210 in RBC from healthy controls in vitro increased these markers.

It was further found that coincubation of RBC from patients with T2DM led to a decrease in miR-210 expression in endothelial cells, indicating that RBC may transmit signaling regulating vascular miR-210 expression. An interesting finding is that atherosclerotic plaques obtained from patients with T2DM had lower expression of miR-210 than those from patients without diabetes. This observation provides an important extension to the clinical setting by demonstrating downregulated miR-210 in patients with T2DM and advanced atherosclerotic cardiovascular disease. The observations that miR-210 expression in endothelial cells following incubation with RBC from healthy subjects was comparable with that observed after incubation with culture medium suggest that a direct transfer of miR-210 from RBC of T2DM patients to endothelial cells seems less likely. However, the mechanism behind the transfer of signaling (including miRNAs) between the RBC and the vasculature is incompletely understood and may involve multiple pathways. Future studies are needed to elucidate such potential signaling pathways.

The current study has certain limitations. First, it is difficult to exclude that medication and comorbidity in the patient group affected RBC function or miR-210 levels. However, our previous data suggest that neither cardiovascular (e.g., statins) nor antidiabetes drugs affect the ability of the RBC to induce endothelial dysfunction (9,23). Moreover, the impairment in endothelial function induced by RBC from patients with T2DM is independent of improvement in glycemic control (23). These observations suggest that comedication does not to any major extent affect the detrimental effect of RBC in patients with T2DM. Furthermore, RBC from GK rats and db/db mice also impair EDR, supporting that RBC induce endothelial dysfunction in T2DM also in the absence of comedication and comorbidities. Second, although the present data clearly indicate an important role of downregulated miR-210 in RBC in endothelial dysfunction, the extent to which this mechanism contributes to the overall vascular dysfunction in T2DM remains unclear. Third, only the rat aorta was used in the current study for evaluation of endothelial function following incubation with human RBC, and ACh-induced relaxation may differ in magnitude between different arterial segments used (35). However, previous studies have shown that the detrimental effect of RBC from patients with T2DM is also present in human internal mammary arteries and hamster arterioles (7,9).

In conclusion, we demonstrate a novel mechanism behind vascular dysfunction in T2DM by which downregulation of miR-210 in RBC induces endothelial dysfunction via interaction with vascular PTP1B and ROS (Fig. 7). Our findings clearly suggest that strategies to increase RBC miR-210 levels have the potential to serve as an effective therapy for the treatment of endothelial dysfunction in patients with T2DM.

Acknowledgments. The patient coordination of David Ersgård and technical assistance by Marita Wallin (Karolinska Institutet) are gratefully acknowledged. The authors thank Dr. Mircea Ivan (Indiana University School of Medicine) for providing the miR-210 KO mice.

Funding. This study was supported by the Swedish Research Council (2020-01372 to J.P. and 013-66-104153-33 to S.-B.C.), the Swedish Heart and Lung Foundation (20190266 to J.P. and 20190341 and 20200326 to Z.Z.), EFSD/Sanofi European Diabetes Research Programme in Macrovascular Complications (to J.P.), the Diabetes Research Wellness Foundation (720-1519-16 and 363-PG to J.P.), the Stockholm County Council ALF (20190031 to J.P.), the Loo and Hans Ostermans Foundation (2018-01213 and 2020-01209 to Z.Z.), Karolinska Institutet grants (2018-01837, 2020-01473, and 2020-02285 to Z.Z.), the Sigurt and Elsa Goljes Memorial Foundation (LA2017-0131 to Z.Z.), the Lars Hiertas Minne Foundation (FO2018-0156 to Z.Z.), the Strategic Research Program in Diabetes (to S.-B.C.) and von Kantzow Foundation (to S.-B.C.) and the Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse (to S.-B.C.).

Duality of Interest. A.C. is supported by a Novo Nordisk postdoctoral fellowship run in partnership with Karolinska Institutet. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. Z.Z. and J.P. designed the study. Z.Z., A.C., C.S., Y.T., A.M., H.W., E.C., T.S, S.N., T.J., and H.J. performed and collected research data. Z.Z., A.C., and A.M. analyzed research data and performed statistical analyses. Z.Z., A.C., A.M., H.J., M.A., X.Z., J.Y., U.H, S.-B.C., L.M., and J.P. contributed to discussion. Z.Z. wrote the manuscript. Z.Z. and J.P. edited the manuscript, and all authors reviewed the final version of the manuscript. Z.Z. and J.P. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in oral form at the European Society of Cardiology Congress, Paris, France, 31 August–4 September 2019.