Increased Expression of miR-483-3p Impairs the Vascular Response to Injury in Type 2 Diabetes

Diabetes is among the most prevalent chronic diseases in developed and developing countries. Despite substantial progress in therapy for and prevention of type 2 diabetes (T2D), the excessive cardiovascular risk remains an unmet clinical need (1). Patients with diabetes suffer from a more aggressive course of atherosclerosis and coronary artery disease (CAD), and cardiovascular mortality is disproportionally high in patients with diabetes (2,3). Among the most relevant causes underlying this excessive risk profile is a severely impaired endothelial repair capacity (4). On the other hand, the necessity to replace damaged or detached neighboring cells (e.g., in sections of nonlaminar blood flow) or to populate nonendothelialized sections (e.g., after catheter intervention) arises more frequently in patients with diabetes (5).

Functional impairment of the endothelial cells themselves (6–10), as well as of “accessory” cells with paracrine activity (11–13), is involved in the reduced endothelial repair capacity. Ex vivo cultured “early outgrowth cells” are a heterogeneous myeloid cell population with regenerative potential and have been widely used to model and investigate alterations in patient-derived cells of the hematopoietic lineage. Their exact composition varies between isolation protocols, but most reports agree on a large component of M2-like macrophages (M2MΦs), as well as smaller components of lymphoid and progenitor cells (11,14,15). M2MΦs obtained from patients with high cardiovascular risk, including T2D, exhibit altered secretory activity, thus affecting the healing response of the adjacent endothelial cells (4,16). Mediators regulating the maintenance and restoration of endothelial integrity include peptide messengers; gaseous molecules, such as nitric oxide; radicals; and noncoding RNAs, such as miRNAs. Differential regulation of miRNas in T2D appears to provide an important link between metabolic cues and gene expression, mediating even transgenerational effects (17).

We have previously observed a downregulation of the bradykinin B2 receptor in patients with CAD (11), for which a mature form of miRNA (miR), 483-3p, was predicted to be a potential translational regulator. miR-483-3p has been reported to be modulated in wound healing and cancer progression (18–20). Mechanistic studies point to a role of metabolic dysregulation in the pathological upregulation of miR-483-3p in tumor cells (21). Recent data from in vitro models suggest a role in the apoptosis of “endothelial progenitor cells” and cardiomyocytes in diabetes (22,23). Yet, the in vivo relevance of miR-483-3p for the vascular response to injury, which is delayed in patients with T2D, was not demonstrated before.

We have analyzed the expression and functional relevance of miR-483-3p in human M2MΦs, which are known to crucially modulate the vascular repair process, in HAECs, in two murine models of carotid artery injury, and in aortic specimens from human patients with T2D. miR-483-3p levels were higher in M2MΦs and in the aortic wall of patients with CAD who had T2D than in patients with CAD who did not have T2D. We identified the transcription factor vascular endothelial zinc finger 1 (VEZF1) as a novel target of miR-483-3p. In subsequent functional studies, we could establish that the overexpression of miR-483-3p and consecutive reduction of VEZF1 signaling increases endothelial cell apoptosis, impairs M2MΦ-mediated regenerative function, and ultimately decreases endothelial regenerative capacity in vitro and in vivo.

The study was reviewed and approved by the local ethics committee (Kantonale Ethikkommission Zurich) and complies with International Conference on Harmonisation good clinical practice guidelines. All study participants provided written informed consent before participation or tissue banking. CAD was defined as stenosis of >50% in a coronary artery confirmed by coronary angiography. “Apparently healthy” control subjects were normotensive and normocholesterolemic, and of comparable age as patients without known cardiovascular disease or long-term medication use. Human aortic samples were obtained from patients with CAD, with or without T2D, who were undergoing elective cardiovascular surgery. Patient characteristics are summarized in Tables 1 and 2.

All animal experiments were performed in accordance with the local and national guidelines and were approved by the local Animal Experiments Committee in Zurich and Berlin. B6.BKS(D)-Lepr^db/J and C57BL/6 wild-type mice were purchased from Janvier Labs (Le Genest-Saint-Isle, France). NMRI-Foxn1^ν/Foxn1^ν mice were purchased from Charles River Laboratories (Cologne, Germany).

The 10- to 12-week-old male athymic nude mice (NMRI-Foxn1^ν/Foxn1^ν, for M2MΦ transplantation) and B6.BKS(D)-Lepr^db/J and C57BL/6 wild-type mice (for systemic transduction) were given 200 mg of metamizole as preanesthetic analgesia and subjected to electrical injury of a 4-mm section of the left common carotid artery under general anesthesia with isoflurane as described previously (11). Postoperative analgesia consisted of oral metamizole in the drinking water for 3 postoperative days. Three hours after the surgery, NMRI-Foxn1^ν/Foxn1^ν mice received 3 × 10⁵ M2MΦs i.v. from patients with CAD and T2D or healthy control subjects, who had previously been transfected with power inhibitor of hsa (human)-miR-483-3p (T2D), mimic of hsa-miR-483-3p (healthy), or a scrambled oligonucleotide (both groups) (all from Ambion/Exiqon, Vedbaek, Denmark), and B6.BKS(D)-Lepr^db/J and C57BL/6 mice received anti-miRNA or scrambled control (5 mg/kg body weight, formulated with jetPEI) (Polyplus). Three days after the injury/treatment, mice were injected with Evans blue under anesthesia, and vessels were harvested for quantification of persisting endothelial injury (11) and analysis of miR-483* expression.

Murine cryopreserved carotid arteries were probed with miRCURY LNA miRNA In Situ Hybridization Optimization Kit and probes (sequence: 5′-AAGACGGGAGGAGAGGAGTGA-3′; Exiqon) at Bioneer (Hørsholm, Denmark) according to the protocol of Nielsen et al. (24). Images were acquired using a 20× objective with a Zeiss AxioScan Microscope and analyzed with ImageJ. Human formalin-fixed, paraffin-embedded aortic biopsy specimens were probed with a human miR-483-3p–specific probe (Affymetrix) using the ViewRNA eZ-L Detection Kit–1-Plex (Red; Affymetrix) at Sophistolab (Muttenz, Switzerland). Human CD31 and α-smooth muscle actin were visualized by immunofluorescence staining (both from Abcam) in additional sections to identify smooth muscle/endothelial layers. Images were taken on a Keyence BZ-9000 inverted fluorescence phase contrast microscope at 20× and 40× magnification. The area of miR-483-3p staining was assessed after defining a red color threshold in ImageJ and normalizing the miR-483-3p–positive area to the vessel wall area (px²/px²) visible in each view field.

HAECs were purchased from CellSystems (Troisdorf, Germany) and Lonza (Walkersville, MD) and propagated in fully supplemented EGM-2 with 10% FCS (Lonza) according to the manufacturer recommendations. HAECs were used for assays at passages 6–8. The starvation medium used was EBM-2 (Lonza) supplemented with 100 IU/mL penicillin, 100 μg/mL streptomycin (both Invitrogen, Carlsbad, CA), and 0.5% FCS. Peripheral blood mononuclear cells were isolated from patients, healthy control subjects, or buffy coats, and were used for the generation of M2MΦs by adhesion culture, as described previously (11).

Transfections with mimic or power inhibitor of miR-483-3p (Ambion and Exiqon, respectively) and silencing RNA for VEZF1 (Ambion) and scrambled oligonucleotide (Ambion) were performed by electroporation using the Neon Transfection System and Kit (Thermo Fisher Scientific). Single pulses of 1,400 V for 20 ms (for HAECs) or 1,900 V for 15 ms (for M2MΦs) were used.

Total RNA was isolated (miRNeasy; Qiagen) and retrotranscribed for quantification of either miRNA (miRCURY LNA Universal cDNA Synthesis Kit II; Exiqon) or mRNA (High Capacity cDNA Reverse Transcription Kit; Applied Biosystems). Quantitative PCR was performed in a Viia7 real-time thermal cycler (for miRNA: miRCURY LNA Universal RT miRNA PCR Kit and corresponding primer sets [Exiqon]; for mRNA: SYBR Select Master Mix [Applied Biosystems]; and oligonucleotide primers were synthesized at TIB MOLBIOL [Berlin]).

The capacity of a confluent cell layer of transfected endothelial cells to close a scratched gap within 16 h was assessed as described previously (11). For assessing the effect of miR-483-3p overexpression on M2MΦ paracrine effects, untransfected HAECs were used as described and transfected M2MΦs were added to the freshly scratched HAECs.

HAECs and M2MΦs were cultured and transfected as described above. Prior to assessment of the apoptosis rate, cells were kept in starvation medium containing 10 ng/mL tumor necrosis factor-α (TNF-α) for 16 h. Mitochondrial membrane potential was assessed by JC-1 (Supplementary Figs. 4B and 5B). Binding of annexin V and uptake of propidium iodide (both from BioLegend) were assessed by flow cytometry (Supplementary Figs. 4C and 5C). Overall cell viability was assessed by flow cytometry (Zombie Fixable Viability Kit; Life Technologies).

Protein amount of VEZF1 in HAECs and M2MΦs was assessed by Western blot and immunofluorescence staining in 25 μg of lysate and adherent cells, respectively, using anti-VEZF1 (Sigma-Aldrich) and anti-GAPDH antibodies (Western blot; Cell Signaling Technology), or Hoechst 33342 (immunofluorescence staining). ImageJ was used for densitometric analysis of protein bands and for fluorescence intensity analysis of VEZF1 in adherent cells.

Cells were cotransfected with mimic of miR-483-3p or scrambled control oligonucleotide and empty vector, 3′ untranslated region (UTR) positive control, or VEZF1 3′ UTR of the GoClone luciferase constructs (Switchgear Genomics, La Hulpe, Belgium) according to the manufacturer protocol, and were plated in 96-well white-bottom plates. After 24 h, LightSwitch assay solution was added. Luminescence was read in a Tecan Infinite F200 Pro (Tecan GmbH, Grödig, Austria). Results were plotted as log2 ratios.

M2MΦs were obtained and transfected as described above. Twenty-four hours after transfection, cells were washed with sterile PBS and starvation medium was added (100 μL per 1 × 10⁵ cells). After additional an 24 h, medium was carefully collected and centrifuged to pellet cell debris, and the supernatant was used for analysis. Unconditioned medium was used as a control, and values lower than in unconditioned medium were considered to be zero. Analytes were quantified using proximity extension assay by Olink Proteomics (Uppsala, Sweden).

Depending on the matching of samples and the normality of distribution (D’Agostino-Pearson test), paired or unpaired t test, Wilcoxon signed rank test, or Mann-Whitney U test was used to compare two groups. Three or more groups of unmatched samples were compared by Kruskal-Wallis test or one-way ANOVA, depending on normality. Two-way ANOVA followed by Sidak post-test was used for secretome analysis. A P < 0.05 was considered significant. Box plots show medians with interquartile ranges and ranges.

Expression of hsa-miR-483-3p was higher in M2MΦs obtained from human patients with T2D compared with control subjects without diabetes, both with concomitant CAD (Fig. 1A). hsa-miR-483-3p expression was not different between subjects without diabetes with CAD and subjects without diabetes and without CAD. M2MΦs obtained from patients with diabetes and prediabetes lose their capacity to support vascular healing (4). Increased expression levels of miR-483-3p in this cell type might potentially contribute to their dysfunction. Supporting this hypothesis, transfection with a mimic of miR-483-3p (Fig. 1B and Supplementary Fig. 1) reduced the capacity of transplanted M2MΦs from healthy human individuals to support the reendothelialization of an injury of defined length inflicted upon the common carotid artery in the recipient mice, whereas transfection with a power inhibitor of miR-483-3p partially rescued the capacity of T2D M2MΦs to support in vivo reendothelialization (Fig. 1B–D). In situ hybridization analysis of human aortic sections verified increased miR-483-3p expression in the vascular wall of human T2D patients compared with patients without diabetes with comparable underlying pathology (Fig. 2).

Semiquantitative analysis by in situ hybridization indicated elevated levels of mmu-miR-483* in the endothelial and smooth muscle layers of T2D obese (B6.BKS(D)-Lepr^db/J) mouse aortae (Fig. 3A and B). Intravenous injection of an LNA-based inhibitor of mmu-miR 483* decreased miR-483* expression in the vascular wall of diabetic mice (Fig. 3A and B). An induced injury of the common carotid artery is reendothelialized faster in wild-type mice of the BL/6 background compared with T2D obese mice of the same background (B6.BKS(D)-Lepr^db/J) (Fig. 3C). Systemic injection of inhibitor of the mmu-483* rescued carotid reendothelialization capacity in diabetic obese mice (Fig. 3A and C). No effect on cardiac capillarization or arteriole number was observed in the treated mice (Supplementary Fig. 2).

Exposure of HAECs and M2MΦs to stimuli mimicking aspects of increased inflammatory state (TNF-α) and metabolic dysregulation (hyperinsulinemia, hyperglycemia, and lipotoxicity), all of which are considered to mediate specific mechanisms of vascular dysfunction in T2D, resulted in the upregulation of hsa-miR-483-3p expression in a cell type–specific manner. In endothelial cells, hyperinsulinemia strongly upregulated miR-483-3p, and a trend (P = 0.051) was observed for palmitic acid, mimicking lipotoxicity (Fig. 4A). For M2MΦs, TNF-α was the main stimulus, whereas palmitic acid induced a downregulation of miR-483-3p (Fig. 4B). Hyperglycemia did not result in a significant upregulation of miR-483-3p in our models.

HAECs transfected with a mimic of miR-483-3p were severely impaired in their capacity to effect reendothelialization (Fig. 5A). In separate experiments, closure of a scratch in the endothelial cell layer was less efficient in the presence of M2MΦs from healthy human donors transfected with a mimic of miR-483-3p compared with scrambled control-transfected M2MΦs from the same donors (Fig. 5B). The closure of a scratch injury within an endothelial monolayer relies on several cellular functions, including migration, proliferation, and cell death. Using the fluorescent label CFSE (carboxyfluorescein succinimidyl ester), the proliferation of cells transfected with scrambled oligonucleotide or a mimic of miR-483-3p in fully supplemented growth medium was traced. No significant differences were observed for CFSE fluorescence between scrambled and mimic-transfected HAECs after 48 h (Supplementary Fig. 3). HAECs and M2MΦs were transfected with a mimic of miR-483-3p and subjected to starvation, and apoptosis-associated characteristics were analyzed. The disruption of mitochondrial membrane potential was examined by JC-1 staining, phosphatidylserine exposure on the cell surface by annexin V binding in the absence of propidium iodide uptake, and the disruption of cell membrane integrity by propidium iodide uptake (Supplementary Figs. 4 and 5). In HAECs and M2MΦs, miR-483-3p overexpression led to an increase in those markers of compromised cell viability (Fig. 5C–H). By proximity extension assay, we analyzed the release of 92 proteins with cardiometabolic relevance by M2MΦs transfected with either mimic or power inhibitor of hsa-miR-483-3p or scrambled oligonucleotide (Supplementary Fig. 6). We identified the GAS6 gene to be reduced, and ICAM1, CES1, QPCT, IL-7R, and MET genes to be increased in the supernatant of M2MΦs transfected with a mimic of hsa-miR-483-3p (Supplementary Fig. 6A). When assessing the opposite regulation of protein release by miR-483-3p overexpression versus inhibition, a reduction in the release of lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1) and T-cell Ig and mucin domain containing 4 (TIMD4) with miR-483-3p overexpression and their increased release with miR-483-3p was observed (Fig. 5I and Supplementary Fig. 6B). Vice versa, the release of interleukin (IL)-7R (IL-7R), tenascin-C (TNC), the hepatocyte growth factor receptor MET, the serine protease PRSS2, and ficolin-2 was enhanced with miR-483-3p overexpression and reduced with miR-483-3p inhibition (Fig. 5I and Supplementary Fig. 6B).

In silico analyses have indicated a number of potential targets of miR-483-3p, which might influence the vascular response to injury in M2MΦs and HAECs, including the secreted growth factors IGF-I, platelet-derived growth factor-b, and kit ligand; as well as apoptosis-related factors FAIM (Fas apoptosis inhibitory molecule), BIRC6 (baculoviral IAP repeat-containing protein 6), RNF 32 isoform 2, GUCY1 (guanylate cyclases, soluble), PIP5KA (phosphatidylinositol 4-phosphate-5 kinase A), SAMD9 (SAM domain–containing protein 9), CFLAR (CASP8 and FADD-like apoptosis regulator), NOL3 (nucleolar protein 3), and ANGPTL7 (angiopoietin-related protein 7); and the receptors TrkC (tropomyosin receptor kinase C), B2R (bradykinin receptor B₂), and TGFβ-R (transforming growth factor-β receptor). Of those, only VEZF1, a transcription factor previously associated with endothelial dysfunction in diabetes (25), showed a significant repression in mimic-transfected HAECs and M2MΦs compared with scrambled-transfected controls (Fig. 6A–D and Supplementary Figs. 7–9). miR-483-3p is predicted to bind to a sequence in the 3′ UTR of the VEZF1 transcript conserved among mammals (Supplementary Fig. 10). We could verify direct targeting of the VEZF1 3′ UTR by miR-483-3p by luciferase reporter assay (Fig. 6E). VEZF1 expression was reduced in aortic homogenates of human CAD patients with T2D compared with nondiabetic CAD patients (Supplementary Fig. 11A). Finally, systemic injection of power inhibitor of mmu-miR-483* rescued vascular VEZF1 expression, which was barely detectable in aortae of db/db mice (Fig. 6F and Supplementary Fig. 11B). Silencing of VEZF1 in HAECs increased cell death (Fig. 6G) and reduced in vitro reendothelialization capacity (Fig. 6H).

A dysfunctional vascular response to injury has been described for patients and animal models of T2D and is associated with higher cardiovascular morbidity and mortality in patients with diabetes (26).

In this translational study, we have observed that miR-483-3p , which is extremely low abundant in nondiabetic conditions, is upregulated in individuals with T2D, both murine and human. Consequently, we have investigated the significance of this upregulation in the vasculature and in M2MΦs. Our study thereby addresses two aspects relevant to the maintenance of vascular integrity—the regenerative capacity of the endothelial layer, and the modulation of endothelial survival and function by alternatively activated M2MΦs. Our findings support the hypothesis that elevated levels of miR-483-3p impair endothelial cell survival under stress, thereby limiting vascular repair capacity upon injury. We have identified miR-483-3p–mediated downregulation of the vascular transcription factor VEZF1 as a key mechanism for the loss of endothelial regeneration efficiency in T2D. Vice versa, the inhibition of miR-483-3p restored endothelial repair capacity in vitro and in vivo, either in the endothelium via systemic delivery of power inhibitor of miR-483-3p or by transplanting ex vivo transfected human M2MΦs into a murine model of vascular injury.

miR-483-3p has been studied mainly in the field of cancer and only recently for cardiovascular and metabolic diseases (18,19,23,27,28). Although miR-483-3p is upregulated in certain tumors as well as in diabetes, its role can change from a factor supporting tumor cell proliferation and survival in cancer to inducing or permitting apoptosis in a cardiovascular setting (22,23,27,29,30).

Studies from the field of cancer biology have shown that beyond transcriptional regulation together with its host gene, IGF2, miR-483-3p transcription can be regulated independently of IGF2 transcription via β-catenin/Wnt signaling (31). β-Catenin/Wnt signaling is also enhanced via high glucose in diabetes, thus providing a link for the upregulation of miR-483-3p in both T2D and tumors, albeit with different downstream effects on cell proliferation and survival. Increased glucose levels have also been described to enhance miR-483­-3p transcriptional activation via the O-linked N-acetylglucosamine transferase, which stabilizes the transcriptional complex at the miR-483 promoter (32). Those findings point to a crucial role of glucose levels for the transcription of miR-483­­-3p. Although our data tentatively suggest increased expression of miR-483-3p in HAECs and in M2MΦs after exposure to hyperglycemia, this effect was not significant, in contrast to hyperinsulinemia in HAECs and the inflammatory stimulus TNF-α in M2MΦs. Our data would therefore suggest potential further mechanisms of miR-483-3p transcriptional regulation by insulin-responsive/sensitive elements and/or by inflammatory transcription factors, such as nuclear factor-κB. Together with the cell type–specific differences in target genes and functional effects of miR-483-3p, our data therefore suggest that also transcriptional regulation of miR-483-3p depends on additional factors, which differ between cell types and pathological settings (33).

Among several candidates, we have identified VEZF1 as a target of miR-483-3p. This transcription factor is essential for embryonic angiogenesis (34). Our own present and previous findings indicate a role of VEZF1 not only in embryonic vascular development, but also in the maintenance of the adult endothelium and especially in diabetes-associated endothelial dysfunction (25). VEZF1 has been identified as a coactivator of IL-3 expression (35). However, we could not detect changes in IL-3 protein levels in supernatant of M2MΦs after the inhibition of miR-483-3p or after the silencing of VEZF1 (Supplementary Fig. 12). Aitsebaomo et al. (36) have described mechanistically distinct effects of VEZF1 on the transcriptional regulation of endothelin-1 and IL-3, depending on differential effects of the transcriptional activator Tax. Further potential actions of VEZF1 include the repression of the antiangiogenic factor CITED2 (37), which we could not verify in our setting, and the upregulation of the proangiogenic factor strathmin (38). By extension, those data illustrate that the presence of additional cofactors and the baseline expression levels of target genes modulate the target spectrum of VEZF1 in a cell type–specific manner.

In addition to modulating target gene transcription, VEZF1 has been shown to affect DNA methylation and elongation during transcription, via regulating the expression of the full-length versus the short isoform of DNA methyltransferase Dnmt3b1 (39). In a subsequent study (40), the same group identified VEZF1 to bind to CpG islands, thereby protecting them from de novo DNA methylation. These findings are in line with observations of aberrant DNA methylation being involved in the pathogenesis of type 1 diabetes and T2D (41), which is assumed to play a role in phenomena like the higher frequency of T2D in children born to mothers suffering from malnutrition during pregnancy (42). Interestingly, miR-483-3p expression in adipose tissue of low–birth weight adult humans and in prediabetic adult rats that had been exposed to suboptimal nutrition in early life was elevated, thus suggesting an epigenetic priming by nutrition in early life (17). One might therefore speculate about the existence of an epigenetic feedback mechanism that aggravates both the causes and effects of miR-483-3p overexpression in pre-T2D.

As is typical for miRNAs, miR-483-3p also has multiple targets, with greater or lesser relevance in certain cell types and pathological/physiological situations. Of a number of in silico–predicted and previously described potential targets, only the transcription factor VEZF1 was verified by our analyses in macrovascular endothelial cells and M2-type macrophages. Other targets, such as IGF-I (22), were not confirmed by our analyses (Supplementary Fig. 7 and Supplementary Table 4), potentially because of cell type–specific differences in basal target gene expression and in the presence of transcriptional coregulators.

In order to better understand the paracrine mechanisms affecting endothelial regeneration upon miR-483-3p modulation in M2MΦs, we have used a proximity extension assay to analyze a panel of 92 proteins with cardiometabolic relevance. Upon miR-483-3p overexpression, reduced amounts of the hyaluronan receptor LYVE1, as well as the phosphatidylserine receptor TIMD4 were released, which have both been reported to support proresolving macrophage functions, such as efferocytosis (43,44). Vice versa, both membrane receptors were released in greater amount upon miR-483-3p inhibition. This could, in principle, be explained by increased shedding of microvesicles from the cell surface or disintegration of the cell into apoptotic bodies, as our preparation removed larger cell debris. Because of the opposite direction of change in apoptosis rate in our setting, microvesicle release might be more plausible, although further studies are needed to verify this hypothesis. In addition, the IL-7R, TNC, the hepatocyte growth factor receptor MET, the serine protease PRSS2, and ficolin-2 were released in higher amounts with elevated miR-483-3p levels and released in lower amounts with reduced miR-483-3p levels. For those factors, we assume that their increased release from miR-483-3p–overexpressing M2MΦs and reciprocal regulation by a power inhibitor of miR-483-3p reflect a response to increased activation and/or death and the resulting release of proinflammatory cytokines and damage-associated molecular patterns, which then activate neighboring cells. In support of this hypothesis, IL-7R and MET have been shown to be upregulated by lipopolysaccharides, IL-1β, and TNF-α (45–47). TNC was reported to be upregulated under pathological conditions caused by inflammation and infection (48). Little is known about the role of the serine protease PRSS2 in macrophage function, but it has been observed to be upregulated in Mycobacterium bovis–infected, bovine monocyte–derived macrophages (49). The opsonin ficolin-2 activates the lectin complement pathway and has recently been identified to stimulate macrophage proinflammatory M1 polarization through toll-like receptor four activation (50). Those initial data would tentatively support a role of PRSS2 and ficolin-2 in (secondary) macrophage activation upon miR-483-3p upregulation, as speculated for IL7-R, TNC, and MET. Of note, we do not have information on the molecular mechanisms causing the altered secretion profile upon miR-483-3p manipulation. The proteins identified to be significantly altered in the proteome analysis are probably not direct targets of miR-483-3p, based on in silico prediction (Target Scan 7.1), but may rather be a consequence of several mechanisms acting in parallel, including altered protein transcription/translation (potentially also through VEZF1), as well as altered release mechanisms, including microvesicle shedding and release of apoptotic bodies during apoptosis.

Our in situ hybridization analyses have also revealed an expression and regulation of miR-483-3p in the vascular smooth muscle layer of human aortic biopsy specimens. The overall vascular effects of miR-483-3p therefore likely exceed the mechanisms described here for impairment of endothelial regeneration and might potentially also include the regulation of vascular tone.

Therefore, our observations, together with reports by other working groups (18,22,23), indicate a potential for the therapeutic downregulation of miR-483-3p in patients with T2D. This particular patient population might benefit from improving their response to vascular injury, as is for example evoked by percutaneous catheter intervention, a frequent clinical intervention in patients with T2D because of their higher risk for coronary artery disease (2). Recent advances in RNA therapeutics have raised hopes that these approaches might be translated into a realistic treatment option in the future (51), especially as the endothelium is an organ readily addressable with current transfection approaches.

Acknowledgments. The authors thank Sabine Knüppel (Charité–Universitätsmedizin Berlin) for excellent technical support. The authors also thank iPATH.Berlin, Immunopathology for Experimental Models, core unit of the Charité–Universitätsmedizin Berlin, for support with the histopathological analysis.

Funding. The project was supported by the Novartis Foundation (grant 11A26), the Bank Vontobel Foundation, the German Center for Cardiovascular Research, partner site Berlin (grant FKZ 81Z2100202), and the Zurich Heart House–Foundation for Cardiovascular Research. K.K. was supported by the Deutsche Herzstiftung (Kaltenbach Stipendium) and the EMDO Foundation. T.F.L. is supported by the Swiss National Science Foundation (SNSF) (grant 310030_166576). U.L. is a professor at the Berlin Institute of Health. N.K. was supported by the SNSF (ambizione fellowship PZ00P3_126621).

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

Author Contributions. K.K. designed the experiments, obtained patient material, performed cell culture and animal experiments and data analysis, wrote the article, and reviewed and approved the article. E.T.S. and M.F.M. obtained patient material, performed cell culture and animal experiments, and reviewed and approved the article. T.F.L. and U.L. contributed to and discussed research strategy and data interpretation and reviewed and approved the article. N.K. conceived and supervised the project, designed the experiments, performed data analysis, and wrote the article. K.K. and N.K. 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.