PDGF-BB Carried by Endothelial Cell–Derived Extracellular Vesicles Reduces Vascular Smooth Muscle Cell Apoptosis in Diabetes

Cardiovascular complications are a leading cause of morbidity and premature mortality in diabetes (1,2). Structural alterations to vessel walls result in intima-media thickening, which puts individuals at high risk of developing acute cardiovascular events (3,4). Moreover, restenosis is still a major complication in the diabetes setting. A main cause of reocclusion is intimal hyperplasia, which is due to the migration and/or excessive growth of vascular smooth muscle cells (VSMCs). A dysregulated balance between apoptosis and the proliferation of VSMCs seems to play a crucial role in intima-media thickening in individuals with diabetes (5,6). Indeed, in vitro studies have suggested that high glucose (HG) levels induce the expression of bcl-2 family members and inhibit the apoptotic protein inhibitor of apoptosis protein 1 (IAP-1) in VSMCs (7). In addition, Ruiz et al. (8) have demonstrated that VSMCs, recovered from patients with diabetes, show a resistance to apoptosis, which was possibly due to bcl-2 overexpression. Although circulating HG concentration might per se induce VSMC dysfunction, additional events can contribute to this process in vivo.

Several studies have focused on extracellular RNA transporters, indicating that they may be present in biological fluids in the form of vesicles, which have been denoted microvesicles, exosomes, membrane particles, and apoptotic bodies (9,10). Despite the lack of consensus on vesicle classification, the presence of overlapping characteristics and biological activity has evoked the use of the inclusive term “extracellular vesicles” (EVs) (11,12). The paracrine/endocrine effects of EVs have recently gained significant attention (13,14). Indeed, EV biological activity has been linked to the transfer of bioactive molecules, including proteins and microRNA (miRs) (10–14). EVs are widely distributed in human body fluids, while circulating EV cargo usually reflects the cell of origin in its physiological and/or pathological condition (9–15). Indeed, the number and cargo of circulating EVs have been suggested as a means to predict the presence of disease and even the risk of developing disease (16,17).

Increased levels of circulating platelet- and endothelial cell (EC)-derived microparticles have been proposed as “biomarkers” of cell dysfunction (18,19). However, EVs might also deliver specific drivers of disease, as they behave as diffusible vectors of biological activity and participate in exchanging information. This study therefore investigates the role of EC-derived EVs as mediators of VSMC fate in type 2 diabetes (T2D).

Reagents and antibodies are reported in Supplementary Table 1.

A total of 11 individuals with T2D and 6 individuals without diabetes (control subjects), who had undergone carotid endoarteriectomy surgery in our clinic, were included in the study. Clinical characteristics are reported in Supplementary Table 2. All individuals with diabetes were undergoing statin and metformin treatment. Ethics approval was obtained from Azienda Ospedaliero-Universitaria (AOU), Città della Salute e della Scienza di Torino, Torino, Italy. Informed consent was obtained from all individuals in accordance with the Declaration of Helsinki. We had no direct contact with the participants.

Human atherosclerotic plaque specimens were recovered from the above-reported subjects (T2D [D] and nondiabetic [ND] specimens) and processed as previously described (20). Vascular tissue was rinsed three times with PBS, and intima was removed in order to furnish the VSMCs. Tunica media were finely cut into 2- to 3-mm pieces and subjected to enzymatic digestion using collagenase type I (0.1 mg/mL) in DMEM for 1.5 h at 37°C. Digestion media were collected and filtered through nylon mesh cell strainers (100 μm) to remove the undigested explants. The resulting supernatants were centrifuged at 1,200 rpm for 10 min, and cells were plated at 2.5 × 10⁴ cells/cm² and cultured with modified Eagle’s medium supplemented with 20% (v/v) FBS and 1% penicillin-streptomycin. FACS analysis was performed on D VSMCs and ND VSMCs to characterize them, as previously described (20), using antibodies directed to CD31 and α-smooth muscle actin (α-SMA), directly or indirectly conjugated with fluorescein isothiocyanate (FITC) fluorochrome. FITC mouse nonimmune isotypic IgG (BD Bioscience Pharmingen) was used as control.

Primary macrovascular ECs and VSMCs were purchased from Lonza (Basel, Switzerland) and cultured as described by the manufacturer’s instructions. VSMCs and ECs were used at II-III cell-culture passage. For collection of the EVs, ECs were starved under either low glucose (LG) (5 mmol/L) or HG (25 mmol/L) and for 24 h deprived of bovine calf serum. Cell viability was evaluated. Small interfering RNA (siRNA) technology was also used in HG-cultured ECs with siRNA negative control or the platelet-derived growth factor (PDGF)-BB siRNA (Applied Biosystems) (21). EV isolation was obtained from HG-cultured ECs depleted of PDGF-BB. In selected experiments, VSMCs were cultured in LG or HG and then treated in the presence of ND CD31EVs, D CD31EVs, or EC-derived EVs (5 × 10³ EVs/target cell) or stimulated with PDGF-BB. In selected experiments, HG-cultured VSMCs were preincubated with a blocking PDGF receptor β (PDGFRβ) antibody (5 μg/mL). Details are reported in the Supplementary Data. All experiments were performed in accordance with European guidelines and policies and approved by the ethics committee of the University of Turin.

Human serum from all above-described individuals with T2D and control subjects was obtained before surgery and after informed consent. EVs from each participant were obtained by centrifuging serum as previously described (22). The supernatant was subsequently submitted to differential ultracentrifugation at 10,000g and 100,000g for 2 h at 4°C. EV pellets were then resuspended in DMEM and stored at −80°C. FACS analysis of D EVs and ND EVs was performed as previously described (23), using anti-CD31 allophycocyanin (APC), anti-CD14 phycoerythrin (PE), and anti-CD42b FITC antibodies. FITC, PE, or APC mouse nonimmune isotypic IgGs (BD Bioscience Pharmingen) were used as controls. FACS analysis was performed using a Guava easyCyte Flow Cytometer (Merck, Darmstadt, Germany). Fluorochrome-conjugated antibodies were added to a suspension of EVs (2.5 × 10⁶ particles/100 µL) for 15 min at 4°C. Surface marker expression is reported in the representative histograms as the percentage of expression ± SD. The CD31 microbead kit (Miltenyi Biotec, Auburn, CA) was used to isolate CD31EVs from the sera of individuals with T2D (D CD31EVs) and control subjects (ND CD31EVs) (24). Briefly, 0.5 mL freshly thawed plasma was incubated with 100 µL CD31 microbeads for 4 h at 4°C. EVs captured on CD31 antibody–coated magnetic beads were recovered from the magnetic column (MS column) as described in the manufacturer’s instructions. EV-bound beads were submitted to differential ultracentrifugation (Beckman Coulter Optima L-90K ultracentrifuge; Beckman Coulter, Fullerton, CA) for 3 h at 4°C. CD31EVs were either used fresh or were stored at −80°C and then processed for transmission electron microscopy (TEM), biological effects, Western blot, and quantitative RT-PCR (qRT-PCR) analysis. CD63 content in CD31EVs was analyzed by Western blot. Details are reported in the Supplementary Data.

TEM was performed on CD31EVs that had been isolated by ultracentrifugation and resuspended in PBS, placed on 200 mesh nickel formvar carbon-coated grids (Electron Microscopy Science, Hatfield, PA), and left to adhere for 20 min. Grids were then processed as previously described (25) and observed under a Jeol JEM 1010 electron microscope (Jeol, Tokyo, Japan). Details are reported in the Supplementary Data.

ECs were cultured in LG or HG DMEM without bovine calf serum for 24 h to collect the EVs from supernatants as previously described (21,22) and as detailed in the Supplementary Data. EV number and size distribution analysis was performed using a NanoSight LM10 (NanoSight Ltd, Minton Park, U.K.). Results were displayed as number per milliliters and as a frequency size distribution graph, output to a spreadsheet. EC-derived EVs (LG EVs or HG EVs) were processed for biological, Western blot and qRT-PCR analysis (21,22).

Cells and EVs were lysed and protein concentrations were obtained as previously described (25). Protein levels were normalized to α-SMA, β-actin, or CD63 content. Details are reported in the Supplementary Data.

The internalization of EVs was evaluated using confocal microscopy (LSM5-PASCAL; Zeiss, Oberkochen, Germany) as previously described (25). EV pellets were added to HG-cultured VSMCs (2 × 10⁴) pretreated or not with a blocking PDGFRβ antibody. Z-Stack confocal microscopy VSMC images were also obtained (25). Details are reported in the Supplementary Data.

Total RNA was isolated from the VSMCs of atherosclerotic plaque specimens and from human VSMCs, that had been treated as indicated or left untreated, using the TRIzol reagent (Invitrogen) as previously described (26). RNA from cells and EVs was then reverse transcribed using a TaqMan MicroRNA Reverse Transcription Kit, specific for miR-24-3p, miR-221, miR-222, and miR-296-5p, or a SYBR Green miRNA reverse transcription kit specific for miR-21-5p, miR-29a, and miR-145, as indicated. miR expression was normalized to the small nuclear RNA RNU6B. Loss- and gain-of-function experiments were performed in VSMCs that had been transfected with the anti-miR control, the anti–miR-296-5p, pre-miR control, or pre–miR-296-5p oligonucleotides (Applied Biosystems) according to the manufacturer’s instructions (26). Details are reported in the Supplementary Data.

The luciferase reporter assay was performed using a construct generated by subcloning the PCR products amplified from the full-length 3′ untranslated region (UTR) of human BAK1 DNA into the Xba restriction site of the luciferase reporter vector pGL3 (Promega, Madison, WI). The PCR products were obtained using the primers for BAK1 and are reported in detail in the Supplementary Data (26).

Proliferative activity was assayed as previously described (27). For the apoptosis assay, VSMCs were subjected to Muse Annexin V and the cell dead assay (Merck, Darmstadt, Germany) in accordance with the manufacturer’s instructions. Details are reported in the Supplementary Data.

For analysis of the EC-VSMC interaction, 24-well plates were coated with growth factor–reduced Matrigel matrix (BD Biosciences) (28). Briefly, HG-cultured ECs and VSMCs were pretreated, for 24 h, with D CD31EVs, ND CD31EVs, or HG EVs that had been depleted, or not, of PDGF-BB. Red-labeled ECs (PKH26 vital dye) (4.5 × 10⁴) were placed in HG medium on top of the polymerized matrix. Green-labeled VSMCs (PKH67 vital dye) (2 × 10⁴) were then added to ECs. Details are reported in the Supplementary Data.

Scratch assays were performed on HG-cultured VSMCs, treated as indicated, to evaluate cell migration activity. VSMCs were seeded to a final density of 100,000 cells/well for 24 h in order to allow cell adhesion and the formation of a confluent monolayer to occur. Details are reported in the Supplementary Data.

PDGF-BB concentration in D CD31EVs was measured using a commercially available competitive enzyme immunoassay (ELISA) kit (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. For evaluation of mbPDGF-BB, intact or lysate D CD31EVs (2.5 × 10⁸ particles) were compared. The same samples were pretreated with trypsin (0.25%) for 1 h (negative control). Details are reported in the Supplementary Data.

All data are presented as mean ± SEM unless otherwise reported. The D’Agostino-Pearson test was used to test normality. Data on in vitro angiogenesis, cell proliferation, ELISA, apoptosis, and scratch assays; on qRT-PCR miR expression, loss- and gain-of-function experiments, and characterization of recovered EVs; and, lastly, on the densitometric analysis for Western blots were analyzed using the Student t tests for two-group comparison and using one-way ANOVA, followed by the Tukey multiple-comparison test, for three or more groups. All Western blot experiments were performed in triplicate. The minimum sample size was four experiments performed in triplicate, thus ensuring 90% statistical power among experimental groups, and a probability level of 0.05 (two-tailed hypothesis). The cutoff for statistical significance was set at P < 0.05. All statistical analyses were carried out using GraphPad Prism, version 5.04 (GraphPad Software, Inc.).

Circulating EVs are able to modulate cell fate (9,10,13). It was therefore decided to investigate whether and how circulating EVs, derived from ECs of individuals with T2D (D EVs), may impact on VSMC fate. Guava analysis was used to demonstrate the presence of a high percentage of EVs of endothelial origin in the sera of control subjects (ND EVs) and of individuals with T2D (Fig. 1A). A significant reduction of D EVs was detected (Fig. 1A). EC-derived EVs from sera were therefore isolated using CD31-coated magnetic beads (ND CD31EVs and D CD31EVs) and analyzed using TEM (Fig. 1B) and Western blot (Fig. 1C). Functional studies were then performed to evaluate the biological relevance of D CD31EVs in mediating VSMC dysfunction in hyperglycemic condition (HG culture condition). LG-conditioned VSMCs served as control. Consistent with data provided by Ruiz et al. (8), we found that, unlike LG, HG treatment was associated with a significant upregulation of bcl-2 and downregulation of bak/bax (Supplementary Fig. 1A). By contrast, the expression of another member of the bcl-2 protein family, bcl2l2, did not change (Supplementary Fig. 1B). Moreover, HG treatment per se significantly decreased the number of apoptotic cells without affecting VSMC proliferation (Supplementary Fig. 1C and D). Functional studies were then performed on VSMCs cultured in HG conditions and treated with CD31EVs. As shown in Fig. 1D, D CD31EVs, unlike ND CD31EVs, further reduced both the number of apoptotic VSMCs and bak/bax content and increased bcl-2 level (Fig. 1E).

The transfer of proteins and/or genetic information into recipient cells is the main mechanism of EV action (9,10,13,14). CD31EV protein cargo and, in particular, the content of well-known VSMC proliferation/survival factor, PDGF-BB, were therefore analyzed in both CD31EVs and EVs recovered from LG- and HG-treated ECs. Supplementary Fig. 2A shows Nanosight analysis of EVs recovered from LG- or HG-cultured ECs (LG EVs or HG EVs). No differences in EV size or number between LG EVs and HG EVs were detected (data not shown). Conversely, PDGF-BB was found to be enriched only in D CD31EVs (Fig. 2A) and in HG EVs (Supplementary Fig. 2B).

LG- and HG-cultured VSMCs were treated with PDGF-BB to investigate the contribution of PDGF-BB in mediating D CD31EV biological effects. In fact, PDGF-BB further increased bcl-2 content, while reducing the number of apoptotic cells and bak/bax content, in HG-conditioned VSMCs. Conversely, these effects were not detected in VSMCs cultured in LG conditions (Fig. 2B and C). siRNA technology was therefore harnessed to abrogate PDGF-BB expression in EVs (Fig. 2D and E) and validate the role of PDGF-BB in regulating D CD31EV survival signals in HG conditions. As expected, PDGF-BB–depleted EVs no longer increased bcl-2 or decreased bak/bax expression (Fig. 2F) or the number of apoptotic cells (Fig. 2G). For investigation of whether different mechanisms might account for free PDGF-BB and D CD31EV–PDGF-BB–induced effects, HG-cultured VSMCs were pretreated with a blocking PDGFRβ antibody, stimulated with D CD31EVs or free PDGF-BB, and analyzed for bak/bax and bcl-2 expression. As shown in Fig. 3A, PDGFRβ blockade impacts on both free PDGF-BB and D CD31EV–mediated bak/bax and bcl-2 expression. To rule out the possibility that this effect depended on inhibition of PDGFRβ-mediated EV internalization, we performed Z-Stack analysis. Figure 3B shows that PDGFRβ blockade did not hamper EV internalization. Of note, the ELISA assay led to the discovery that PDGF-BB was anchored to the membrane of D CD31EVs (Fig. 3C), indicating that EV membrane-bound (mb)PDGF-BB, by binding to the PDGFRβ, might drive D CD31EV biological effects.

CD31EVs were also evaluated in an EC-VSMC coculture assay. A reduced number of vessels was detected in HG-cultured ECs (data not shown), and D CD31EVs nevertheless promoted VSMC recruitment to neo-formed tubule-like structures (Fig. 4A and B)—unlike ND CD31EVs. A scratch assay also highlighted increased VSMC motility upon D CD31EV treatment (Fig. 4C). As shown in Supplementary Fig. 2C, these effects did not depend on differences in EV vascular endothelial growth factor content. Furthermore, the contribution of PDGF-BB to both processes was validated in experiments using PDGF-BB–depleted EVs; such EVs failed to induce either effects (Fig. 4D–F).

For validation of the above results, VSMCs isolated from either D or ND human atherosclerotic plaque specimens were analyzed for bak/bax and bcl-2 expression. As shown in Fig. 5A, the majority of the recovered cells express VSMC marker. Moreover, as previously reported by Ruiz et al. (8), it was confirmed that D VSMCs expressed high bcl-2 levels. In addition, we found that D VSMCs also expressed low bak/bax content (Fig. 5B and C).

For further insight into the mechanisms regulating VSMC fate, the expression of miRs potentially involved in this process was first analyzed in D VSMCs and compared with ND VSMCs. As shown in Fig. 6A, no significant differences were detected in the expression of miR-21-5p, miR-24-3p, miR-145, miR-29a (not included in the panel: CT >38), miR-221, and miR-222 between D and ND VSMCs (29,30). Moreover, since bak, unlike bax, is one of miR-296-5p putative target genes (TarBase v7.0) (31), its expression was also analyzed. Indeed, high miR-296-5p levels were detected in D VSMCs (Fig. 6A). For validation of the role of miR-296-5p in controlling bak expression, VSMCs cultured in LG or HG conditions were analyzed for miR-296-5p expression. As shown in Fig. 6B, increased miR-296-5p expression was detected upon HG treatment. Moreover, the depletion of miR-296-5p in HG-cultured VSMCs (Supplementary Fig. 3) led to the decrease of bcl-2 levels, the increase of bak/bax content (Fig. 6C), and a consistent increase in the number of apoptotic cells (Fig. 6D). For confirmation of the role of miR-296-5p in bak posttranscriptional regulation, the full-length 3′ UTR BAK1 nucleotide sequence was analyzed for miR-296-5p blasting sequences revealing several base pairings (1,138–1,158) (Fig. 7A). The luciferase assay was used to demonstrate that bak is indeed a direct miR-296-5p target (Fig. 7B). This observation was further validated by gain-of-function experiments (Fig. 7C and D).

For investigation of the contribution of PDGF-BB in mediating miR-296-5p expression, LG- and HG-cultured VSMCs were also treated with PDGF-BB. As shown in Fig. 6E, PDGF-BB further increased miR-296-5p only in HG-cultured VSMCs. These results were validated by siRNA technology (Fig. 6F), thus suggesting that VSMC survival may be under the control of miR-296-5p.

For exclusion of the possibility that D CD31EV–mediated effects could also depend on the delivery of miR-296-5p, the miR-296-5p content was also evaluated in CD31EVs and LG- or HG-cultured EC-derived EVs (LG EVs or HG EVs). Almost undetectable levels of miR-296-5p were found in D CD31EVs (Supplementary Fig. 2D) and in EVs from HG-treated ECs (Supplementary Fig. 2E). Thus, the role of mbPDGF-BB D CD31EVs in mediating miR-296-5p–driven posttranscriptional regulation of bak and its downstream events was further strengthened.

Atherosclerosis, and its associated complications, is a major cause of death worldwide (32). Diabetes accelerates atherosclerosis and restenosis after angioplasty (33,34). Indeed, increased VSMC migration and survival/proliferation are crucial for restenosis, particularly in those with diabetes (35–37). In vitro and ex vivo studies have shown that the upregulation of bcl-2/bcl-xl is crucial for VSMC resistance to apoptosis in diabetes (7,8,38,39). Moreover, Li et al. (7) have reported that while HG enhances the expression of bcl-2 family members in VSMCs, it reduced the expression of IAP-1. We herein demonstrate that VSMCs, exposed to HG concentrations, upregulate bcl-2 and downregulate bak/bax—without affecting VSMC proliferation. Moreover, we discovered that these effects are boosted by D CD31EV treatment. A cell’s survival or death depends on the integrity of the mitochondrial outer membrane (40). In this regard, while the proapoptotic proteins, bak and bax, are involved in the permeabilization of mitochondrial outer membrane, the antiapoptotic bcl-2 family members counteract such proapoptotic signals by preventing cytochrome c efflux (40). Our results therefore indicate that an additional shift in the balance, from apoptotic to antiapoptotic signals, might be the main mechanism behind D CD31EV–induced resistance to apoptosis in HG conditions and suggest that hyperglycemia-mediated cues preferentially translate into VSMC resistance toward apoptosis rather than VSMC proliferation.

The genetic material in EV cargo has sparked considerable interest. The role of miRs as mediators of epigenetic changes has been extensively reported, particularly in diabetes (41,42). The transfer of miRs into recipient cells has been described as a relevant mechanism of EV biological action (9–14). As a matter of fact, Gu et al. (43) recently reported that the transfer of miR-195 from EC-derived EVs to VSMCs regulates VSMC proliferation. Despite the ability of D CD31EVs to induce functional changes in VSMCs, our results demonstrate that D CD31EV–mediated VSMC dysfunction relies on a mechanism independent of the delivery of miRs. Indeed, EVs also transport and deliver proteins that can affect VSMC fate, including PDGF-BB. PDGF-BB is a growth factor known to regulate VSMC outcomes (44,45). In this regard, PDGF-BB derived from platelet and EC is considered a relevant mediator of VSMC dysfunction and restenosis (46). As a matter of fact, we demonstrate that D CD31EVs are enriched in PDGF-BB and that PDGF-BB–enriched D CD31EVs contribute to VSMC dysfunction in the setting of diabetes. Several lines of evidence indicate that downstream signaling events, activated by PDGF-BB, trigger various biological processes, including VSMC migration and recruitment to neo-formed vessels (44,45). It is worth noting that PDGF-BB–depleted EVs failed to induce both VSMC migration and recruitment to neovessels. This suggests that CD31EV–PDGF-BB cargo might play a crucial role in accelerating VSMC dysfunction and restenosis in T2D.

PDGF-BB synthesis and release can increase in response to various stimuli including intima damage (47). As shown herein, in the diabetes setting, CD31EVs constitute a relevant circulating PDGF-BB reservoir and contribute to VSMC fate. We established that PDGF-BB is bound to the membrane of CD31EVs and is required for their biological action but not for their internalization. As a proof of concept, PDGFRβ blockade completely hampers CD31EV-mediated bak/bax expression without impeding their entry into the cell. This indicates that, along with free PDGF-BB, mbPDGF-BB–enriched D CD31EVs contribute to PDGF-BB paracrine effects (48). A cooperative action between CD31EVs and platelet-derived EVs could be postulated in vivo. In fact, EVs released from platelets accumulate in human atherosclerotic plaques and can induce major biological pathways by transferring their PDGF-BB content (49). In line with the results presented herein, it has recently been reported that EC- and platelet-derived EVs are enriched in PDGF-BB in patients with cardiovascular diseases (50).

miRs are key regulators of gene expression, mainly at the posttranscriptional level (51). We herein demonstrate, both ex vivo and in vitro, that the hyperglycemia milieu enhances miR-296-5p expression and modulates bak content in VSMCs and that these effects are strictly dependent on D CD31EV–PDGF-BB cargo rather than EV miR-296-5p delivery. In addition, a shift between antiapoptotic to proapoptotic signals after PDGFRβ blockade (downregulation of bcl-2 expression) was observed. This suggests that PDGF-BB directly, or indirectly by changing the balance of cellular miRs, might transcriptionally or posttranscriptionally regulate its expression. Mutually, these events translate in VSMC resistance to apoptosis.

Emerging evidence suggests that EVs can serve as specific diagnostic/prognostic biomarkers, since they can provide intercellular state information on a given disease condition (52). Increased levels of “small (submicroscopic) membranous particles” of endothelial origin, such as CD31⁺/annexin V⁺ and CD31⁺/CD42⁻, have been detected in circulation in patients with coronary artery disease, suggesting that they may be an additional risk stratification factor (18,19). A significant reduction in CD31EVs (CD31^high/CD42b^low/CD14^low) has been found in individuals with T2D in the current study. The phenotype of the “small membrane particles,” which also includes apoptotic bodies, and the lack of exosome refining (18,19) in coronary artery disease patient studies could explain the discrepancy with our results. However, increasing amounts of evidence indicate that healthy subjects and diseased individuals release EVs with different cargo into circulation (53). Our ex vivo and in vitro results reinforce the notion that D CD31EV cargo, rather than D CD31EV number, is the crucial determinant of their biological activity.

The current study reports that hyperglycemia per se induces epigenetic mechanisms in VSMCs by enriching the circulating CD31EV cargo with mbPDGF-BB, which translates into VSMC resistance to apoptosis (Fig. 8). We are also the first to demonstrate that EV–mbPDGF-BB–mediated miR-296-5p overexpression, and the posttranscriptional regulation of bak, is a relevant mechanism of D CD31EV action. Overall, these results identify D CD31EV–mbPDGF-BB as a novel driver of VSMC dysfunction in the setting of diabetes.

Funding. This work was supported by grants from Ministero dell’Università e della Ricerca Scientifica (ex 60%) (to P.D.) and was also supported by Associazione Italiana per la Ricerca sul Cancro project IG 2015.17630 (to M.F.B.).

Duality of Interest. This work was also supported grant no. 071215 from Unicyte to G.C. and M.F.B. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. G.T. performed ex vivo and in vitro experiments and miR and protein analysis. P.D. performed in vitro angiogenesis assay and FACS analysis. A.R. generated constructs and performed transfections. G.L. performed in vitro experiments and EV isolation. M.G. performed loss- and gain-of-function experiments. S.G. performed Western blot analysis. C.G. performed EV isolation and characterization. A.S. contributed to data interpretation and revised the manuscript. G.C. contributed to the study conception and revised the manuscript. M.F.B. contributed to the study conception, designed the study, and wrote the manuscript. M.F.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.