Extracellular vesicle microRNA cargoes from intermittent hypoxia-exposed cardiomyocytes and their effect on endothelium

Intermittent hypoxia (IH), as the crucial pathophysiological feature of obstructive sleep apnea syndrome (OSAS), is an independent risk factor initiating the progression of cardiovascular complications. However, how IH cause cardiovascular injury and initiate inter-organ communication remains unclear. Extracellular vesicle (EV) are reported to be involved in cell-to-cell and organ-to-organ communications through selectively carrying RNA and protein cargos from donor cells and delivering them to recipient cells. By utilizing an unbiased miRNA microarray approach, current study attempted to determine whether IH may alter miRNA proles in cardiomyocyte-derived extracellular vesicles and whether EVs from IH-treated cardiomyocytes could affect endothelial function. from of mouse primary

Obstructive sleep apnea syndrome (OSAS) is an independent risk factor associated with many clinical complications (i.e., hypertension, stroke and coronary heart disease) [1]. Intermittent hypoxia (IH) which commonly seen as the key property of OSAS is considered as the major pathological origin of OSASassociated cardiovascular diseases. Previous ndings suggest that IH resulted in cardiomyocytes apoptosis because of insu cient oxygen supply [2] or through the mediation of miR-146a-5p [3]. Additionally, IH exposure caused earlier and greater in ammatory response in cardiomyocytes as evidenced by increased lactate dehydrogenase and proin ammatory cytokine (IL-1β, IL-6, IL-8, and macrophage migration inhibitory factor) [4]. Recent work of our group proved that myocardial infarction plus IH treatment exacerbates cardiac remodeling via miR-214-3p/CTRP9 pathway [5]. Although previous researches have been devoted to seeking for molecular targets to alleviate IH-related cardiac injury, whereas more novel biomarkers are still needed to be discovered due to the insu cient ndings as yet.
Extracellular vesicles (EVs) are nano-scale bilayer membrane microvesicles produced by most of cells.
The components of EVs closely re ect the physiological and pathological status of their tissue origins and thus are considered as desirable targets for clinical diagnosis and therapeutics. Meanwhile, EVs participate in extensive physiological and pathological processes by transferring complex cargoes including proteins, lipids, and nucleic acids from donor cells to recipient cells [6]. Lots of studies have shown that EV miRNAs are involved in cardiovascular disease [7]. Cardiac broblast EV-miR-423-3p was reported participated in cardioprotective effects through targeting the downstream effector Rap-2c during the acute phase of ischemia-reperfusion injury [8]. While cardiomyocyte-derived EV miR-208a participated in cardiac brosis via promoting broblast proliferation and differentiation into myo broblasts [9]. The decreased EV microRNA-21-5p derived from heart failure patients impaired cardiac regenerative potential by inhibiting angiogenesis and cardiomyocyte survival [10]. And the enriched miR-130b-3p in dysfunctional adipocyte-derived EV was found by our team mediating diabetes-related cardiac injury through suppression of cardioprotective molecules expression in cardiomyocytes [11]. Therefore, systematical and precise dissection of the myocardium-derived EV moleculars under IH conditions is so important as that will help to determine the main mechanism of OSAS or IH-associated cardiovascular diseases, and subsequently aid to develop more effective and precise diagnosis and treatment in clinical practice. However, the disparity of miRNA components in EVs delivered from normoxia-and IH-treated cardiomyocytes is not yet clear.
This study provides the rst evidence that IH causes signi cant alteration in myocardial EV miRNA composition and reveals the adverse effect and the possible mechanism of EVs from IH-treated cardiomyocytes on endothelial function, which will help to understand the mechanistic underpinnings behind the pathology of OSAS and IH and their related cardiovascular diseases.

Cell culture
All animal handling complied with the standard animal welfare regulations of Capital Medical University.
The Animal Subjects Committee of Capital Medical University approved the animal study protocol. C57BL/6 mouse primary cardiomyocytes were isolated as described previously with minor modi cations [12]. Brie y, adult male C57BL/6 mice (8-12 weeks) were subjected to general anesthesia (2% iso uorane). Hearts were harvested and perfused retrogradely via aortic cannulation at 37°C for 5 minutes with a constant ow of Perfusion Buffer (116 mM NaCl, 5.4 mM KCl, 6.7 mM MgCl 2 , 12 mM glucose, 2 mM glutamine, 3.5 mM NaHCO 3 , 1.5 mM KH 2 PO 4 ,1.0 mM NaH 2 PO 4 , 21 mM HEPES) 4.5 mL/min in a Langendorf apparatus (WPI,United states). Then hearts were followed by perfused again in digestion solution supplemented with 0.6mg/ml colleganse II and 15 mM CaCl 2 at 37°C for 10 minutes.
After that, hearts were quickly triturated the ventricular tissue with ne forceps, so that most pieces were smaller than 1 mm 3 . Next the cell suspension was transfered to a 15 mL conical, allowed to settle by gravity for 15 min and carefully removed the supernatant with a transfer pipette to cell culture dishes until only 50 -100 μL of solution remained above the tissue pieces. Then the cell culture dish was moved promptly to a 37 °C incubator with 5% CO 2 and 95% humidity.
HUVECs were purchased from AllCells (Shanghai, China) and cultured with a special medium ECM (ScienCell, San Diego, CA, USA). The 8-10 generations were used for our study and cells were collected for protein detection after treatment with EVs for 24 h or 48h respectively.

Intermittent hypoxia treatment
Intermittent hypoxia (IH) condition was obtained by cycles of 5%-21% O 2 per hour with a BioSpherix-OxyCycler C42 system (BioSpherix, Red eld, NY, USA). Mouse primary cardiomyocytes were subjected to IH or normoxia (Nor) treatment (continuous 21% O 2 as control) for 48h with EV-free medium (SBI, USA) incubation. Then the conditioned medium was collected EV puri cation EVs were isolated from cardiomyocyte culture media according to the method previously described with simple modi cations [13]. Brie y, the culture medium of cardiomyocytes exposed to IH or Nor status was collected and centrifuged progressively at 300 × g for 15 min, 2,000 × g for 15 min, 12,000 × g for 80 min to remove cardiomyocytes and cell debris. Then, the supernatant was passed through a 0.1µm microporous membrane lter (Millipore, MA, USA) and ultracentrifuged at 110,000 × g for 120 min (Beckman Coulter, CA, USA). The EV pellets were washed with PBS to exclude contaminating proteins followed by a second ultracentrifugation at 110,000 x g for 120 min. All the centrifugation steps were performed at 4°C. EVs puri ed from 50mL culture medium of C57BL/6 mouse primary cardiomyocytes (from 5 mice) were collected and frozen at -80°C as one sample for miRNA microarray chip assay.

EVs identi cation
The characterization of EVs was con rmed by measuring the expression of EV-speci c markers, such as CD63, CD81, TSG101, Alix and calnexin via Western blotting analysis. The particle size and number were detected by NanoSight analysis (NS300, Malvern Instruments) and the EV morphology was visualized by transmission electron microscope (TEM, H-600, Hitachi, Western blotting and silver staining Protein was extracted from EVs using a protein extraction kit (BC3710, Solarbio, Beijing, China) and protein concentration was quanti ed using a BCA assay (Thermo Fisher Scienti c, USA) according to the manufacturer's protocol. Protein (5 µg) was separated by 12% SDS-PAGE and transferred onto a nitrocellulose membrane (EMD Millipore, USA). After blocked with 5% nonfat milk, the membranes were incubated with primary antibodies overnight at 4˚C. Then, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1:5,000; K5007; Dako, USA) for 1 h at room temperature. The protein bands were visualized using the SuperSignal chemiluminescent detection module (34080; Pierce) and images were collected using a ChemiDoc TM Touch (Bio-Rad, USA).
Silver staining was performed with the Silver Staining Kit (Invitrogen, USA) following the manufacturer's recommendations. The primary antibody information can be found in Table S2.

Transmission electron microscopy
The morphology of EVs was observed using transmission electron microscopy (TEM). Electron microscope analysis of whole-mounted EVs was carried out according to the reported protocol [13]. Brie y, the EV pellets were xed with 2% PFA and deposited onto EM grids. The grids were separately transferred into 1% glutaraldehyde and methyl cellulose-UA for 10 mins. After air dry, the grid was viewed under FEI Tecnai lectron microscope at 80 kV. miRNA library preparation, microarray chip Assay, and data analysis EV miRNAs were extracted by using the miRNA isolation Kit (catalog number AM1561, Thermo Fisher Scienti c, USA) according to manufacturer's protocol. MicroRNA array analysis was undertaken by Oebiotech Co., Ltd. (Shanghai, China). Brie y, 100ng total miRNA per sample served as an input material for the small RNA library preparation. The libraries were generated using NEBNext® Multiplex Small RNA Library Prep Set for Illumina® (NEB, E7300L). The library quality was assessed on the Agilent Bioanalyzer 2100 system using miRNA High Sensitivity Chips. The chip was sequenced on an Agilent Mouse miRNA 21.0 platform. Known miRNAs were identi ed by comparison to miRBase 21.0 database (http://www.mirbase.org/). The p value < 0.05 and |log 2 (fold change)| > 1 was set as the threshold for signi cantly differential expression by default. Differential miRNAs were exhibited by volcano plot and hierarchical clustering heatmap.

Vascular functional study
Adult male C57BL/6 mice (8-12 weeks) were anaesthetized with iso urane and aortas were dissected out for functional assay. Aortic segments (3 mm in length) were suspended in wire myograph (Danish Myo Technology, Aarhus N, Denmark) to record changes in isometric force. After the activation of phenylephrine (Phe, 1μM) in the organ bath, the vascular tone of REV-treated aorta rings raised from basic 3 mN to 5-7 mN. Then, endothelium-dependent vasodilatation relaxation was assessed by measuring the dilatory responses to cumulative concentrations of acetylcholine (ACh). In addition, endothelium-independent relaxation was assessed by measuring the dilatory responses to cumulative concentrations of sodium nitroprusside (SNP) after aorta rings were contracted by NG-nitro-L-arginine methyl ester (L-NAME, 100μM).

Statistical analysis
Data are shown as mean ± SD for numerical variables. Comparisons were made using the Student's ttest, as appropriate. For all statistical tests, p values less than 0.05 were considered statistically signi cant. All statistical analyses were performed with Graphpad Prism 7.

Characteristics of cardiomyocyte extracellular vesicles
Myocardium-derived EVs were isolated by a combination of ultracentrifugation and ltration. EVs were visualized by transmission electron microscopy after negative staining for morphology and size determination. Results showed that the cardiomyocyte EVs exhibited a near-spherical shape and typical disk-shaped morphology (Fig. 1A). Western blotting analysis indicated that the puri ed EVs contained cardiomyocyte speci city marker α-sarcomeric actin (α-SA) and the typical exosomal markers CD63, CD81, Alix, Tsg101 without endoplasmic reticulum protein Calnexin ( Fig. 1B and Additional le2: Supplemental Table 2). The size distribution of EVs detected by Nanosight NS300 (Malvern, UK) presented that the average diameters of EVs were approximately 121.4±28.9 nm (Nor) and 135.34±17.23nm (IH). The diameters of around 60% EVs were between 40 and 80 nm and more than 80% EVs were less than 200nm (Fig. 1C). Meanwhile, the number of EVs from IH-treated cardiomyocytes was observed signi cantly higher than its counterpart (Fig. 1D).

Microarray chip assay and bioinformatic analysis of EV miRNAs from Nor-or IH-treated cardiomyocytes
To investigate whether IH affects myocardium EV miRNA composition, miRNA microarray analysis was performed. Three biological replicates were performed to obtain reliable results ( Fig. 2A).We quanti ed a total of 1,944 miRNAs from myocardial EVs in the three replicates by comparing to miRBase 21.0 database (http://www.mirbase.org/). To reveal the functional relevance of myocardial EVs under IH status, we mainly focused on the analysis of the differentially expressed miRNAs identi ed in EVs from IH-and Nor-treated cardiomyocytes. The differentially expressed miRNAs were de ned as miRNA with 2fold cutoff and p value less than 0.05. Using these criteria, 63 differentially expressed microRNAs were identi ed in IH-treated myocardial EVs including 32 up-regulated microRNAs and 31 down-regulated miRNAs ( Fig. 2B and Additional le1: supplemental table 1). The target genes of differentially expressed myocardial EV miRNAs were predicted separately by TargetScan, PITA as well as miRNA. org databases (Fig. 2C). 4,364 identical target genes were predicted in three different databases. The putative cellular process or associated functional pathways of these genes were further analyzed using the DAVID bioinformatics tool (KEGG). The top 12 canonical pathways were presented. The signi cant pathways were enriched in PI3K-Akt signal pathway, mTOR signal pathway, FoxO signal pathway and MAPK signal pathway (Fig. 2D). We also categorized these identical target genes based on the GO annotation. The results showed that cytoplasm was the main location of the miRNA-target genes and the top signi cant function was associated with protein binding. In addition, DNA-templated transcription was the most important biological process (Additional le4: Fig. S1).
The effect of EV from IH-exposed cardiomyocytes on endothelial-dependent relaxation Considering the enriched function of differentially expressed EV miRNA was mainly concentrated in Akt pathway, we hoped to explore the effect of IH EV (EV from IH-exposed cardiomyocytes) on endothelial function since Akt/eNOS signaling axis is a well-established way leading to endothelial-dependent relaxation [14]. Consistent with database prediction, total Akt (normalized to GAPDH), phosphorylated Akt (normalized to t-Akt) and phosphorylated eNOS (normalized to t-eNOS) were signi cantly downregulated in IH myocardial EV-treated HUVES compared to its counterpart (Fig. 5A-B). Correspondingly, functional study showed that acetylcholine-caused dose-dependent relaxations in precontracted aortic rings were dramatically impaired in aortas with IH myocardial EV-48 h treatment (Fig. 5C). Whereas, IH myocardial EVs did not affect endothelial independent relaxation induced by NO donor, SNP (sodium nitroprusside) (Fig. 5D).

Discussion
In the present study, we carried out miRNA comparison assay in EVs from IH-or normoxia-treated cardiomyocytes and obtained two novel ndings. First, we identi ed 63 miRNAs in EVs from IH-treated myocarial myocytes with different expression compared to control group. Among them, 11 homologous miRNAs were veri ed with the identical expression alteration as they appeared in the miRNA chip assay. Moreover, we revealed that the expression of miR-181-5p, miR-1a-3p and miR-494-3p has the identical tendency in IH-exposed cardiomyocyte as their expression in EVs from IH-treated cardiomyocyte, whereas the expression of miR-16-2-3p has the opposite tendency. Second, we found IH EVs derived from IHtreated cardiac myocytes effectively inhibited Akt/eNOS sighal and signi cantly impaired endothelialdependent relaxation (Fig. 5E). Extracellular vesicle, as a nanovesicle secreted by various types of cells, encapsulates and transfers a wide range of functional proteins, lipids and nucleic acids to recipient cells to mediate intercellular communication [15]. Growing evidences indicated that EVparticipates in OSAS-or IH-associated cardiovascular diseases through transferring the altered EV components [16,17]. Landscape studies of EV-miRNA changes under OSAS or IH are bound to provide possible insights into the potential molecular mechanisms of IH-related cardiovascular diseases. For instance, intermittent hypoxia altered circulating exosome miRNA cargos that promote the increased permeability and dysfunction of endothelial cells [18]. However, a systematic comparison of the miRNA difference in EVs secreted by cardiomyocytes under intermittent hypoxia and normoxia has not yet been reported. In this study, we identi ed 63 miRNAs with different expression in EVs from IH-treated cardiomyocytes. Among them, 11 homologous miRNAs were veri ed to have the same expression alterations as them appeared in miRNA chip assay. Moreover, 6 miRNAs with up-regulated expression in IH myocardial EVs (miR-125b-5p, miR-1a-3p, miR-22-3p, miR-23a-3p, miR-29a-3p and miR-494-3p) were predicated mainly involved in Akt pathway according to KEGG bioinformatic functional analysis. And further prediction by Targetscan database suggested Akt1/Akt2/Akt3 were the direct target genes of miR-125b, -23a, -22, -1a. Akt/eNOS signaling axis is a wellestablished way leading to endothelial-dependent relaxation [14]. As expected, we proved the adverse effect of EV derived from IH-treated cardiomyocyte on endothelial function as well as the reduced expression of t-Akt, p-Akt and p-eNOS in HUVEC with IH myocardial EV treatment. In addition to the endothelial function, PI3K/Akt signaling pathwaywas also reported mediating cardioprotective effect in cardiomyopathy induced by many diseases [19]. For example, the protective effects of tanshinone IIA on diabetic cardiomyopathy [20], Sulforaphane on hypertrophic cardiomyopathy [21], as well as Darbepoetin alfa on autoimmune cardiomyopathy were all mediated by enhanced Akt activity [22]. Especially, enhanced PI3K/Akt activation reversed the adverse effects of hypoxia-ischemia damage in cardiomyocytes by promoting certain survival-related proteins and anti-apoptotic proteins [23]. PI3K/Akt pathway was also involved in the bene cial effect of resveratrol on chronic intermittent hypoxia-induced cardiac hypertrophy [24]. Moreover, molecular derived from stem cells EVs also participated in cardiomyocyte survival through activating PI3k/Akt pathway [25]. All the previous reports together with our ndings suggests that differentially expressed miRNAs in EVs from IH-treated cardiomyocytes may be involved in IH-triggered cardiovasculopathy via directly regulating Akt activity.
In addition to the predicted functions relevant to PI3K/Akt pathway, the differentially expressed EV miRNAs identi ed in our results have several known pathological functions in cardiovascular system. For example, previous ndings revealed a pivotal role for miR-125b-5p in regulating cardiomyocyte survival during acute myocardial infarction [26] and EVs-derived miR125b-5p under hypoxia-conditioned facilitated ischemic cardiac repair by ameliorating cardiomyocyte apoptosis [27]. Moreover, recent researches indicated miR-22-3p was a key molecule in regulating vascular smooth muscle cell proliferation and migration by targeting HMGB1 [28] and a promising therapeutic target for atherosclerosis treatment [29]. EV miR-23a-3p from mesenchymal stem cells attenuated myocardial injury through suppressing DMT1 expression [30]. And miR-29a-3p played a protective role in TNFα-induced endothelial dysfunction[31]. Therefore, the myocardial EV miRNA components and their variations under IH status will help us to deeply understand the possible mechanism of IH-induced cardiovasculopathy, or help to reveal the function of myocardial EV in IH or OSAS-related cardiovascular diseases or even other relevant complications [32].
Except for the validation of EV miRs with different expression, we also measured the expression of corresponding miRs in IH-treated cardiomyocyte. We found that the expression amount of miR-129b-5p and miR-30c-5p was dramatically higher in cardiomyocytes than that in myocardium EVs, indicating that these miRs mainly exert their function in cardiomyocytes rather than in EVs. On the contrary, the amount of miR-504-3p was signi cantly higher in EVs than that in cardiomyocytes, suggesting miR-504-3p possibly functions mainly by EV mediation. In general, EVs contain multifarious biomolecules with the similar express pro le with their parent cell contents and provide crucial information about an individual's physiological context [33]. Consistent with this view, in our results, the expression of miR-181-5p, miR-1a-3p and miR-494-3p had the identical tendency in IH-exposed cardiomyocyte as their expression in EVs from IH-treated cardiomyocyte. Otherwise, it also has been well recognized that the sorting of EV cargos is a precisely regulated process and some molecules with opposite expression propensity in EVs and their donor cells under speci c status [34]. Our results supported the opinion as well by demonstrating that miR-16-2-3p expression was increased by IH treatment in cardiomyocytes while was inhibited in EVs from IH-treated cardiomyocytes, indicating that most of miR-16-2-3p was kept in cardiomyocyte to play its role under IH status rather than secreted from cardiomyocytes to EVs.
This study found many differentially expressed miRNA in IH myocardial EVs and predicated their function relevant to Akt pathway, whereas, we only restricted our studies to examine the contribution of IH myocardial EVs on HUVECs and endothelial function. Other possible cardiovascular effect of IH myocardial EVs cannot be ignored in explanation of IH-or OSAS-triggered cardiovascular diseases since Akt is a pivotal regulator in a wide range of cellular functions. Moreover, we do have proved that IH myocardial EVs can impair endothelial function and provided the possible miRNAs involved during the process, while we did not precisely illustrate the speci c miRNA of EVs in this regulation mechanism. Therefore, future validation investigation is merited to solve these questions.