Endothelial-derived extracellular microRNA-92a promotes arterial stiffness by regulating phenotype changes of vascular smooth muscle cells

Endothelial dysfunction and vascular smooth muscle cell (VSMC) plasticity are critically involved in the pathogenesis of hypertension and arterial stiffness. MicroRNAs can mediate the cellular communication between vascular endothelial cells (ECs) and neighboring cells. Here, we investigated the role of endothelial-derived extracellular microRNA-92a (miR-92a) in promoting arterial stiffness by regulating EC–VSMC communication. Serum miR-92a level was higher in hypertensive patients than controls. Circulating miR-92a level was positively correlated with pulse wave velocity (PWV), systolic blood pressure (SBP), diastolic blood pressure (DBP), and serum endothelin-1 (ET-1) level, but inversely with serum nitric oxide (NO) level. In vitro, angiotensin II (Ang II)-increased miR-92a level in ECs mediated a contractile-to-synthetic phenotype change of co-cultured VSMCs. In Ang II-infused mice, locked nucleic acid-modified antisense miR-92a (LNA-miR-92a) ameliorated PWV, SBP, DBP, and impaired vasodilation induced by Ang II. LNA-miR-92a administration also reversed the increased levels of proliferative genes and decreased levels of contractile genes induced by Ang II in mouse aortas. Circulating serum miR-92a level and PWV were correlated in these mice. These findings indicate that EC miR-92a may be transported to VSMCs via extracellular vesicles to regulate phenotype changes of VSMCs, leading to arterial stiffness.

We previously showed that increased miR-92a level causes EC dysfunction by targeting genes such as Krüppellike factor 2 (KLF2), KLF4, and sirtuin 1 (SIRT1), crucial for a homeostatic endothelium 14 . EC miR-92a can be transported to macrophages via extracellular vesicles (EVs) to downregulate KLF4 levels 11 . Yet, it is not known whether dysregulated miR-92a level in ECs during hypertension onset contributes to arterial stiffness and if so, whether the underlying mechanism involves EC-VSMC communication via miR-92a.
Hence, the objective of this study was to investigate the role of miR-92a in the EC-VSMC crosstalk and its consequential effect in arterial stiffness. Our results reveal an inverse correlation between circulating miR-92a level and PWV in humans and mice. The underlying mechanism involves increased miR-92a level in ECs modulating VSMC plasticity. Locked nucleic acid-modified antisense miR-92a (LNA-miR-92a) ameliorated angiotensin II (Ang II)-induced hypertension in mice, which suggests a therapeutic potential of miR-92a antagonists in ameliorating arterial stiffness.

Results
Increased serum miR-92a level in patients with hypertension. We collected blood from two groups of donors: the patient group (n = 44) with a first diagnosis of hypertension and heathy controls (n = 21) (patient demographics in Table 1). Serum was isolated from the blood and level of miR-92a was measured. As compared with healthy controls, hypertensive patients showed significantly higher circulating miR-92a level (Fig. 1A). Additionally, BMI and levels of blood glucose and triglycerides were higher in hypertensive patients than controls (Table 1). To investigate whether circulating levels of miR-92a were correlated with arterial stiffness, we compared the two groups for PWV, an indicator of arterial stiffness. PWV was higher in hypertensive patients than controls (Fig. 1B). We and others have previously shown that circulating miR-92a level is inversely correlated with endothelial nitric oxide synthase (eNOS)-derived NO bioavailability 14,19 . As anticipated, serum NO level was significantly lower in hypertensive patients than controls (Fig. 1C), but serum ET-1 level was higher (Fig. 1D). Serum miR-92a level was positively correlated with systolic and diastolic blood pressure (SBP and DBP), brachial-ankle PWV (baPWV), and serum ET-1 level (Fig. 1E-H) but inversely correlated with serum NO level (Fig. 1I). Furthermore, baPWV was inversely correlated with serum NO level but positively with serum ET-1 level (Fig. 1J,K). These results suggest that circulating miR-92a level was indicative of endothelial dysfunction and arterial stiffness during hypertension onset.
Ang II induces miR-92a and EC dysfunction. Elevated Ang II level during the hypertension onset has a detrimental role in the cardiovascular system. In vitro, Ang II treatment often causes EC dysfunction. Thus, we examined whether miR-92a in ECs could be induced by Ang II. Ang II treatment increased miR-92a level in ECs dose-and time-dependently ( Fig. 2A,B). Genes crucial for homeostatic ECs, including eNOS, KLF2, and KLF4, are miR-92a targets 14 . Ang II treatment decreased the mRNA and protein levels of eNOS, KLF2, and KLF4 in ECs (Fig. 2C,D), but the expression of pro-inflammatory and -fibrotic ET-1 was increased (Fig. 2C,D). Together, these results indicate that Ang II-induced EC dysfunction might be mediated by elevated miR-92a level. miR-92a regulates phenotype changes of VSMCs. Because the contractile-to-synthetic phenotype change of VSMCs contributes to arterial stiffness 4 , we explored whether increased miR-92a level in ECs could affect the VSMC phenotype. Chang et al. have shown that miR-92a can be secreted from endothelial cells into exosomes to regulate macrophage function 11 . First, the exosomes were isolated from the conditioned media of human umbilical vein ECs (HUVECs) and confirmed by electron microscope and nanoparticle tracking analysis (NTA) with a typical size of approximately 100 nm and a characteristic cup-shaped morphology (Fig. 3A,B). To study whether EC-derived exosomes transfer to VSMCs, we used PKH67 to label the secreted exosomes and cultured VSMCs with the labeled exosomes. The exosomes uptake experiment showed that VSMCs uptake HUVEC-derived exosomes (Fig. 3C,D).
HUVECs were treated with Ang II or PBS for 24 h before co-culture with VSMCs (Fig. 4A). Level of miR-92a was elevated in ECs treated with Ang II and in EVs isolated from conditioned media of Ang II-treated Table 1. Baseline characteristics of healthy controls and patients with hypertension. Data are mean ± SD or number (%). SBP systolic blood pressure, DBP diastolic blood pressure, BMI body mass index, TC total cholesterol, TG triglycerides, LDL low-density lipoprotein. www.nature.com/scientificreports/ ECs (Fig. 4B,C). Furthermore, miR-92a level was elevated in VSMCs co-cultured with Ang II-treated ECs, as compared with phosphate buffered saline (PBS)-treated ECs (Fig. 4D). We then measured mRNA levels of genes related to the contractile and proliferative phenotype. The mRNA levels of contractile markers (α-SMA, smoothelin, and calponin) were decreased, but those of proliferative markers (fibronectin, osteopontin, and thrombospondin) were increased in VSMCs co-cultured with Ang II-treated ECs (Fig. 4E). Given that miR-92a level is significantly increased in carotid arteries after intima injury 13 , we analyzed RNA-seq data (GSE164050) from carotid arteries. Consistent with our results, marker genes of the contractile phenotype of VSMCs were downregulated, and those of the proliferative phenotype were upregulated in carotid arteries (Fig. 4F). We next investigated that EC-derived EVs stimulated by Ang II are responsible for the contractile-to-synthetic phenotype change of VSMCs. HUVECs were incubated with Ang II with 20 μM GW4869 for 24 h to block the formation of EVs or without, prior to ECs co-cultured with VSMCs. As shown in Fig. 4G-I, GW4869 treatment mitigated the contractile-to-synthetic phenotype change of VSMCs. We further treated naïve VSMCs with EVs isolated from conditioned media of Ang II-treated ECs (Fig. 4J). Level of miR-92a was increased in ECs and VSMCs with EVs incubation (Fig. 4K,L). Also, mRNA levels of fibronectin, osteopontin, and thrombospondin were increased, but levels of α-SMA, smoothelin, and calponin mRNA were decreased in VSMCs incubated with EC-derived EVs Data are mean ± SEM. Normally distributed data were analyzed by the two-tailed Student t test (C, D), and the two-tailed Student t test with Welch correction (B) between 2 indicated groups. Non-normally distributed data were analyzed by the Mann-Whitney U test (A) between 2 indicated groups. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.001. r s Spearman correlation coefficient.

LNA-miR-92a reduces hypertension susceptibility.
For translational relevance, we tested the efficacy of exogenously delivered miR-92a in suppressing Ang II-induced hypertension and arterial stiffness. We administered LNA-miR-92a to C57BL/6 mice via tail vein injection (Fig. 5A). LNA-miR-92a administration significantly suppressed the level of miR-92a in circulating CD144 + -EVs and mitigated the Ang II-elevated PWV as compared with control LNA (LNA-Ctrl) ( Fig. 5B,C). Moreover, the level of miR-92a in circulating CD144 + -EVs was positively correlated with PWV in the three groups of mice (Fig. 5D). Consistently, LNA-miR-92a administration reduced the Ang II-elevated SBP and DBP (Fig. 5E,F). With vasorelaxation assays of isolated aortic segments, we assessed the functional outcomes of LNA-miR-92a at the tissue level. The eNOS-dependent vasodilation induced by acetylcholine (ACh) and the endothelium-independent vasodilation induced by sodium nitroprusside (SNP) were impaired in aortas from mice receiving Ang II (Fig. 5G,H). LNA-miR-92a administration reversed both the EC-dependent and -independent vasodilation. To further elucidate the role of miR-92a in VSMC phenotype transition in this mouse model, we assessed transcripts in the mouse aorta that are related to contractile versus proliferative phenotypes. As compared with LNA-Ctrl administration, LNA-miR-92a administration significantly reduced the Ang II-elevated miR-92a levels in aortic ECs and SMCs (Fig. 5I). Consistently, mRNA levels of α-SMA, smoothelin, and calponin were decreased, whereas those of fibronectin, osteopontin, and thrombospondin were increased in aortas of mice receiving Ang II and LNA-miR-92a (Fig. 5J). Thus, the exogenously administered LNA-miR-92a was effective to mitigate arterial stiffness associated with Ang II-induced hypertension in mice. To demonstrate that miR-92a in CD144 + -EVs is crucial for the phenotypic changes of VSMCs in vivo, we also isolated serum CD144 + -EVs from mice treated with Ang II or saline, and then injected these EVs into the wild-type mice. As shown in Supplementary Fig. S3, the serum and VSMC miR-92a levels in mice administered CD144 + -EVs isolated from Ang II-treated mice increased significantly. Consistently, the VSMC contractile markers in mice receiving these CD144 + -EVs significantly decreased, while the proliferative markers increased. These data indicate that the CD144 + -EVs associated miR-92a would contribute, at least in part, to the vascular dysfunction and VSMC phenotype transition in Ang II-administered mice. . Parametric data were analyzed by the one-way ANOVA between multiple groups (A, B) and by the two-tailed Student-t test with Welch correction between 2 indicated groups (C). *p < 0.05, **p < 0.01, ***p < 0.001. The eNOS, KLF4, KLF2, ET-1 and β-actin bands were cropped from full gels ( Supplementary Fig. S1a).

Discussion
Arterial stiffness is considered a complication of hypertension caused by the long-term adverse effects of elevated blood pressure, among other risk factors. Reciprocally, vascular stiffness is a pathological culprit of hypertension 2 . At the molecular level, arterial stiffness is affected by VSMC tone and EC signaling resulting from aging, vascular growth factors, calcification, and activation of innate and adaptive immunity 2,17 . In this study, we found that a cross-talk between dysfunctional ECs and VSMCs can promote arterial stiffness. Specifically, elevated abundance of extracellular miR-92a is transported from impaired ECs to VSMCs, which promotes vascular stiffness (Fig. 6). Highly expressed in ECs, miR-92a induces EC dysfunction and also increases VSMC proliferation, migration, and apoptosis [18][19][20][21][22][23] . MiR-92a can mediate cell-to-cell communication within the vasculature via EVs. Upon vascular damage, activated platelets adhere to the intima, then secrete platelet-derived EVs. These miR-92a-containing EVs, when transported to VSMCs, can induce Col8a1, which augments vascular stiffness 13 . Our results showed that Ang II increased miR-92a level in ECs, which resulted in EC dysfunction. Thus, EC-VSMC communication involving miR-92a might also contributes to the phenotype changes of VSMCs in our experimental conditions. To test this hypothesis, we co-cultured VSMCs with dysfunctional ECs. Results in Figs. 2, 3 and 4 show that impaired ECs with high miR-92a expression upregulated genes promoting proliferation in the co-cultured VSMCs. Thus, the contractile-to-proliferative phenotype change of VSMCs can be induced by miR-92a secreted from ECs. In line with this result, miR-92a level was markedly increased in EVs secreted from Ang II-treated ECs and in co-cultured VSMCs (Fig. 4). Analysis of RNA-seq datasets (GSE164050) revealed that the transcriptomes of injured carotid arteries contained high miR-92a level transported by activated platelets. This in silico analysis showed that the expression of genes related to the contractile phenotype was decreased whereas that related to the proliferative phenotype was increased in the injured carotid artery as compared with controls.
The miR-92a-induced EC dysfunction would be contributed by miR-92a-targeted KLF2, KLF4, and SIRT1 14,18,21 . Whether endothelial KLF2, KLF4 or SIRT1 can transport to VSMCs is unknown. However, VSMCspecific SIRT1 ablation worsens the oxidative and inflammatory responses to Ang II 24 . Thus, besides miR-92a, the proteins targeted by miR-92a may also participate in the EC-VSMC crosstalk. Several studies have reported that Ang II induces the proliferative phenotypic change of VSMCs 25,26 . We also detected the level of miR-92a in VSMCs treated with Ang II, and found that Ang II increased the expression of miR-92a in VSMCs (Supplementary Fig. S2). Thus, we cannot rule out the possibility that Ang II directly regulates contractile-to-synthetic phenotype gene expression changes in VSMCs through changes in SMC miR-92a level.
Macrophages play a critical role in vascular inflammation and remodeling by releasing proinflammatory cytokines and growth factors that act on neighboring ECs and VSMCs [27][28][29] . Of note, EC miR-92a can be also transported to macrophages via EVs. Ample evidence shows that circulating miR-92a level is associated www.nature.com/scientificreports/ with cardiovascular diseases [14][15][16]30 . Reducing miR-92a level in animal models attenuates EC dysfunction and atherosclerosis 14,19 . Circulating miR-92a level was found correlated with PWV in hypertensive patients 16 . In the current study, serum miR-92a level was inversely correlated with PWV in humans. In the Ang II-induced mouse hypertension model, exogenously administered LNA-miR-92a could alleviate Ang II-induced hypertension and arterial stiffness (Fig. 5). LNA-miR-92a may have efficacy in ECs, VSMCs, and macrophages, to collectively ameliorate vascular stiffness. In summary, our study demonstrated that (1) circulating miR-92a level was increased in both hypertensive patients and Ang II-induced hypertensive mice in comparison with healthy controls; (2) circulating miR-92a level was positively correlated with PWV, SBP, and DBP but inversely with NO level in humans; (3) Ang II caused EC dysfunction via EC-derived miR-92a; (4) miR-92a-enriched EVs secreted by dysfunctional ECs played a critical role in modulating the VSMC phenotype transition; and (5) exogenously administered LNA-miR-92a alleviated Ang II-induced hypertension and arterial stiffness in mice. Circulating miR-92a level could be an important biomarker and a potential therapeutic target for hypertension and arterial stiffness. Brachial-ankle PWV (baPWV) measurement. All participants were asked to rest for at least 10 min in the supine position before baPWV measurement. baPWV was measured with an automatic waveform analyzer (VP-1000, Colin, Japan), which automatically records pulse waves of the brachial and posterior tibial arteries with automated oscillometric sensors. After obtaining 5 min of bilateral baPWV values, the average value was used for further analysis.

Human participants and serum samples.
Animal experiments. All animal experiments were approved by the Institutional Animal Ethics Committee of Xi'an Jiaotong University and were carried out in accordance with relevant guidelines and regulations. All mice (C57BL/6J) were kept on a 12-h light/dark cycle and fed a chow diet ad libitum at room temperature. For Ang II-induced hypertension, mice at 9 weeks old were subcutaneous infused with Ang II at 1 μg/kg/min for 28 days by Osmotic minipumps (model 2004, Alzet). One week before Ang II infusion, mice received LNA-Ctrl or LNA-miR-92a at 16 mg/kg body weight by tail-vein injection (Fig. 5A). The second dose of LNA-miR-92a was given 10 days after the minipump implantation. Mice were euthanized by intraperitoneal injection of 200 μL of 2% pentobarbital sodium at the end of day 28 after minipump implantation. LNAs were designed and synthesized by GenePharma Co. Blood pressure was measured by the tail-cuff method (BP-2000, Visitech Systems,  Extracellular vesicle (EV) isolation, analysis, and labeling. CD144-enriched EVs were isolated with procedures described previously 15 with minor modification. Cells, dead cells, and cell debris were removed from the conditioned media by centrifugation at 300g× for 5 min, 2000g× for 15 min and 10,000g× for 30 min, respectively. After centrifugation, the supernatant was ultracentrifuged at 120,000g× at 4 °C for 70 min to pellet the small vesicles (Beckman Optima L-100XP, SW40Ti). The serum was collected and diluted with PBS in a 1:1 ratio, and then ultracentrifuged at 200,000g× at 4 °C for 2 h. The pellet was washed with PBS and filtrated through a 0.22-μm filter. Then, the pellet was immunoblotted with anti-CD144 antibody (Santa Cruz Biotechnology) and incubated with Dynabeads (Invitrogen). Total RNA from CD144-enriched EVs was isolated with TRIzol and with Cel-miR-39 at 2 nM added as a spike-in control. For electron microscopy analysis, the EVs were fixed in 4% paraformaldehyde at 4 °C for 1 h, and then placed on a formvar-coated grid and negatively stained with 2% (w/v) uranyl acetate. Sections were observed using transmission electron microscopy (HITACHI, HT7800). For nanoparticle tracking analysis (NTA) of EVs: Isolated exosome samples were appropriately diluted using 1 × PBS buffer (Biological Industries, Israel) to measure the particle size and concentration. NTA measurement was recorded and analyzed at 11 positions. Particle sizes of EVs were analyzed by nanoparticle tracking analysis using a ZetaView PMX110 (Particle Metrix, Meerbusch, Germany) and the corresponding software ZetaView 8.04.02 SP2. For EVs uptake: A total of 2 × 10 5 HASMCs were incubated with 2 μl PKH63 (Sigma-Aldrich)labeled EC-derived EVs in 24-well plates for 24 h at 37 °C in 95% air and 5% CO 2 . The cells were fixed in 4% paraformaldehyde for 15 min and washed twice with PBS, and then incubated with antibodies against α-SMA, followed by secondary antibody. The samples were counterstained with DAPI, then mounted with fluorescent mounting medium. An Olympus IX81 fluorescence microscope was used to acquire fluorescence images. Normally distributed data (E-H) were analyzed by the two-way ANOVA between multiple groups. Nonnormally distributed data were analyzed by the Kruskal-Wallis test (SMCs in I, α-SMA in J) between multiple groups. *p < 0.05 vs. Saline, # p < 0.05 vs. Ang II + LNA-Ctrl. *p < 0.05, **p < 0.01, ***p < 0.001. r s , Spearman correlation coefficient.    www.nature.com/scientificreports/ of mRNA and miR expression, respectively. The sequences for qPCR primers are in Table 2. The primary antibodies used were anti-KLF4 (Cell Signaling Technology, 12173S), anti-KLF2 (Abcam, ab17008), anti-eNOS (BD Transduction Laboratories, 610296) and anti-ET-1 (Abcam, ab2786). Secondary antibodies were anti-mouse (Jackson, 515-035-003) and anti-rabbit antibodies (Jackson, 111-035045).

Statistical analysis.
Analyses were performed with GraphPad Prism 9. For normally distributed data, the two-tailed Student t test was used to compare two groups and ANOVA with Bonferroni post-hoc test for multiple groups. For non-normally distributed data, the Mann-Whitney U test was used to compare two groups and Kruskal-Wallis test for multiple groups. Correlational analyses involved using Spearman correlation. Data are expressed as mean ± SEM and *p < 0.05 was considered statistically significant.
Consent for publication. The authors agree for publication.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on request.