Extracellular vesicles from human embryonic stem cell-derived cardiovascular progenitor cells promote cardiac infarct healing through reducing cardiomyocyte death and promoting angiogenesis

Human pluripotent stem cells (hPSCs)-derived cardiovascular progenitor cells (CVPCs) are a promising source for myocardial repair, while the mechanisms remain largely unknown. Extracellular vesicles (EVs) are known to mediate cell–cell communication, however, the efficacy and mechanisms of hPSC-CVPC-secreted EVs (hCVPC-EVs) in the infarct healing when given at the acute phase of myocardial infarction (MI) are unknown. Here, we report the cardioprotective effects of the EVs secreted from hESC-CVPCs under normoxic (EV-N) and hypoxic (EV-H) conditions in the infarcted heart and the long noncoding RNA (lncRNA)-related mechanisms. The hCVPC-EVs were confirmed by electron microscopy, nanoparticle tracking, and immunoblotting analysis. Injection of hCVPC-EVs into acutely infracted murine myocardium significantly improved cardiac function and reduced fibrosis at day 28 post MI, accompanied with the improved vascularization and cardiomyocyte survival at border zones. Consistently, hCVPC-EVs enhanced the tube formation and migration of human umbilical vein endothelial cells (HUVECs), improved the cell viability, and attenuated the lactate dehydrogenase release of neonatal rat cardiomyocytes (NRCMs) with oxygen glucose deprivation (OGD) injury. Moreover, the improvement of the EV-H in cardiomyocyte survival and tube formation of HUVECs was significantly better than these in the EV-N. RNA-seq analysis revealed a high abundance of the lncRNA MALAT1 in the EV-H. Its abundance was upregulated in the infarcted myocardium and cardiomyocytes treated with hCVPC-EVs. Overexpression of human MALAT1 improved the cell viability of NRCM with OGD injury, while knockdown of MALAT1 inhibited the hCVPC-EV-promoted tube formation of HUVECs. Furthermore, luciferase activity assay, RNA pull-down, and manipulation of miR-497 levels showed that MALAT1 improved NRCMs survival and HUVEC tube formation through targeting miR-497. These results reveal that hCVPC-EVs promote the infarct healing through improvement of cardiomyocyte survival and angiogenesis. The cardioprotective effects of hCVPC-EVs can be enhanced by hypoxia-conditioning of hCVPCs and are partially contributed by MALAT1 via targeting the miRNA.

Therefore, developing new therapeutic strategies to promote the infarct healing by reducing cardiomyocyte death following acute MI (AMI) and improving the cardiac performance are desirable 11 .
Cell therapy by transplantation of stem/progenitor cells and their derived cardiovascular cells is one of the most notable alternative therapeutic approaches under exploration [12][13][14][15][16][17][18] . Among them, human pluripotent stem cells (hPSCs), including both human embryonic stem cells (hESCs) and human induced pluripotent stem cells (hiPSCs), hold promise for promoting cardiac infarct healing because they can be theoretically produced in unlimited quantities of cells of any lineages, including cardiomyocytes 17,[19][20][21][22][23][24] . hESC-derived cardiovascular progenitor cells (hCVPCs) have been shown to improve cardiac function of rodent infarcted hearts when implanted during the subacute stage of ischemia/reperfusion (I/R) hearts 14 . We found that SSEA1 + hCVPCs promote cardiac infarct healing when given during the early phase of MI in mouse 25 and nonhuman primate 26 models, showing the significant improvement in the recovery of left ventricular (LV) function and the amelioration of cardiomyocyte death. Moreover, the adverse effects of proarrhythmia and tumorigenicity are not observed in the patients with advanced ischemic heart failure (HF) after transplanted SSEA1 + hCVPCs 27,28 . More recently, we found that the beneficial effects of these cells to the infarcted hearts are related to the hCVPCsecreted cytokines via the modulation of macrophage polarization towards a reparative phenotype in the post-MI hearts 25 , while the secretome of hCVPCs and their functions in the cardiac protection remain largely unknown.
The cumulative evidence has revealed that extracellular vesicles (EVs)/exosomes contribute critically to the cardioprotective effects of transplanted stem cells and their derivatives 9,[29][30][31] . It has been shown that administration of the EVs either secreted from hESC-or hiPSC-derived SSEA1 + -CVPCs (hCVPC-EVs) at 3 weeks post MI improves recovery of cardiac function in the mouse model 32 , in part, through the specific microRNA (miRNA) signature contained in the EVs 33,34 . EVs contain proteins, lipids, and nucleic acids (mRNAs, miRNAs, etc.) with important functions 35 . However, it is unclear whether the hCVPC-secreted EVs (hCVPC-EVs) have benefits for promoting cardiac infarct healing when given at the acute phase of MI and, if yes, how they work.
The miRNAs in EVs derived from various types of cells, such as mesenchymal stem cells (MSCs) 36,37 , Sca1 + adult cells 38 , cardiosphere-derived cells 39 , and mouse ESCs 40 , have been shown to protect the heart from MI injury. However, whether the other type of RNAs, such as long noncoding RNAs (lncRNAs), contain in the EVs and what functions they have need more studies, though recent studies showed that the lncRNAs are detected in the EVs derived from human ESC/iPSC-derived cardiomyocytes 41 and LncRNA H19 mediates, at least partially, the cardioprotective role of EVs derived from MSCs in promoting angiogenesis 42 . It has been shown that a lncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) contributes to vascular growth in a mouse retinal angiogenesis model 43 and human umbilical cord MSC-derived EVs prevent aging-induced cardiac dysfunction by releasing lncRNA MALAT1 44 . However, it is unknown whether hCVPC-EVs contain lncRNAs and whether MALAT1 contributes to the cardiac protection of hCVPC-EVs through targeting miRNAs.
In the present study, using a murine permanent MI model involving the administration of SSEA1 + -hCVPCsecreted EVs during AMI, combining in vivo and in vitro analysis, we investigated (i) whether hCVPC-EVs improve infarct healing; (ii) whether hCVPC-EVs contain lncRNA candidates benefiting cardiac protection; and (iii) what the miRNA target of hCVPC-EV-contained lncRNA MALAT1 confers the protection of MALAT1 in cardiomyocyte survival and angiogenesis. Our findings have identified previously unrecognized function of hCVPC-EVs in promoting infarct healing when given at the acute phase of MI and provided new insights into the mechanisms underlying the beneficial effects of hCVPC-EVs in cardiac protection.
Hypoxic conditioning was performed by placing the cells in an oxygen control cabinet (Ruskinn, England) mounted within an incubator and equipped with oxygen controller and sensor for continuous oxygen level monitoring for 48 h in DMEM basal medium. The oxygen concentration in the cabinet was maintained at 1%, with a residual gas mixture composed of 5% CO 2 and balanced N 2 . Normoxic conditioning hCVPCs were incubated under 21% O 2 and 5% CO 2 for 48 h in DMEM basal medium. Both normoxic and hypoxic conditioned medium were collected for EV isolation.

Flow cytometry analysis
Flow cytometry was performed as described previously 21 . Cells to be examined were harvested, dissociated with Accutase (Stem Cell Technologies). The samples were stained for the presence of cardiac progenitor marker PE-conjugated SSEA1 (1:20; eBioscience), and then analyzed and quantified by flow cytometry (FACStar Plus Flow Cytometer, BD bioscience).

EV isolation
The EV-N and EV-H were isolated from normoxic and hypoxic conditioned medium separately as described previously 46 . Briefly, The collected hCVPC conditional medium was centrifugated at 300×g for 30 min followed by 2000×g for 30 min, 4°C to remove cells and dead cells, and then centrifugated at 10,000×g for 30 min, 4°C to remove cell debris, finally centrifugated twice at 100,000×g for 70 min, 4°C with a SW-41 rotor (Beckman Coulter), followed by washing with phosphate-buffered saline (PBS). The final pellet containing EVs was resuspended in PBS and analyzed by NanoSight NS300 (Malvern Panalytical), transmission electron microscope and Western blot, or lysed with QIAzol reagent (#217084, Qiagen) for RNA analysis.

Nanoparticle tracking analysis (NTA)
The NTA was carried out to determine the EV size and concentration by using NanoSight NS300 (Malvern Panalytical) on the isolated EVs as previously reported 38 . The isolated EV pellet as described in the above EV Isolation method was resuspended in PBS, and then 10 μL of it was used for NTA (the sample was diluted to 700 μL with PBS), and 10 μL of it was used for Pierce BCA Protein Assay. During NTA analysis, three 30 s video taken per sample were averaged as one value and five samples were examined in each group. The PBS was subtracted from particle number/mL after quantification. The analysis was performed by using the NTA software (NTA 3.2 Dev Build 3.2.16). Based on the measurement from NTA and Pierce BCA Protein Assay, the 1 μg EV protein had 32.80 ± 8.529 × 10 8 of particles in the EVs secreted from hESC-CVPCs under normoxic cultivation (EV-N) group and 34.60 ± 11.76 × 10 8 of particles in the EVs secreted from hESC-CVPCs under hypoxic cultivation (EV-H) group as shown in Supplementary Fig. S1. Accordingly, the 20 μg EV protein contained about 485-827 × 10 8 particles in the EV-N group, and about 457-927 × 10 8 particles in the EV-H group (p > 0.05, n = 5).

Western blot analysis
The cells and the EVs were lysed with the radioimmunoprecipitation assay lysis buffer. Then, the protein extracts were separated via gel electrophoresis, transferred to a polyvinylidene fluoride membrane, and blocked with 5% bovine serum albumin for 1 h. The membranes were incubated overnight with primary antibody against Alix (1:1000, Cell Signaling Technology), TSG101 (1:1000, Abcam), CD63 (1:200, Santa Cruz Biotechnology), CD81 (1:500, Thermo Fisher Scientific), and CD9 (1:500, Thermo Fisher Scientific). Then, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies at room temperature for 2 h and exposed via enhanced chemiluminescence.

MI and EV treatment
All surgical procedures described were performed in accordance with the Guidelines for Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication, 8th Edition, 2011) and were approved by the Institutional Animal Care and Use Committee of Shanghai Institutes for Biological Sciences. The male C57BL/6 mice aged 10-12 weeks were randomly divided into four groups: Sham + PBS (n = 8), MI + PBS (n = 12), MI + EV-N (n = 12), and MI + EV-H (n = 15) after excluded the LV ejection fraction (LVEF) values greater than 50% (2 in MI + PBS, and 1 in MI + EV-H) at day 2 post MI. The mice were anesthetized via intraperitoneal injection of 50 mg/kg sodium pentobarbital, ventilated with a volumeregulated respirator (SAR830, Cwe Incorporated), and the MI model was induced through the permanent ligation of the left anterior descending (LAD) coronary artery with a 10-0 Prolene suture as previously described 25,40,47 . Of 20 μg EVs, mixed with PBS in a total volume of 20 μL, were directly injected into two sites in the border zones of infarcted myocardium (10 μL each site) after the ligation of LAD coronary artery, and the equal amount of PBS was injected into the sham heart too. Then the muscle and skin of the chest were then sutured with 6-0 and 5-0 Prolene suture, respectively. The body temperature was maintained at 37°C during the surgical procedure.

Echocardiography
Transthoracic echocardiography (Vevo 2100, Visual Sonics) with a 25-MHz imaging transducer was performed on isoflurane anesthetized mice to measure the heart function as previously described 25 . LVEF and LV systolic dimensions (LVDs) were recorded and the average values were collected from each of three consecutive cardiac cycles.

Immunohistochemical staining
Heart tissues were embedded in OCT (SAKURA) compound for histological analysis. were prepared at 400 μm intervals. For fluorescent immunohistochemistry, Transversal frozen sections (5 μm) were fixed with 4% paraformaldehyde, permeabilized in 0.4% Triton X-100 (Sigma), and stained with anti-CD31, α-SMA, and cTnT antibody which were detected by fluorescent conjugated secondary antibodies. Nuclei were stained with DAPI (Sigma). Cell death was evaluated with an In situ Cell Death Detection Kit (Roche Applied Science, Germany) for TUNEL staining as directed by the manufacturer's instructions. The percentage of TUNEL + cardiomyocytes was quantified as the ratio of TUNEL + cTnT + to total cTnT + cells. The immunostaining images were blindly captured using a Zeiss inverted microscope and processed using ZEN software. Three microscopic fields were quantified for each slice. The Image-processing software (Image J) was used as the image quantification software.

Masson's trichrome staining
For the Masson's trichrome-stained images, each heart was made five slices from the point of ligation to the apex of the heart. The morphometric parameters in the five slices of each heart including total LV area and scar area were blindly analyzed, and the scar size was calculated as the total scar area divided by the LV area as described previously 25,47 .

RNA extraction
Total RNA was isolated from hCVPC-EVs using miR-Neasy Micro Kit (Qiagen) according to the manufacturer's instructions. The concentration of RNA fraction was quantified using Nanodrop (Thermo Fisher Scientific) according to the manufacturer's protocol.

RNA-sequencing (RNA-seq) analysis
Five EV samples from each of EV-N and EV-H groups from H9 hESC origin were proceeded for RNA-seq analysis as previously reported 48,49 . Total RNA sequencing was performed with the Truseq Stranded Total RNA Library Prep kit (Illumina, RS-122-2201). This made use of Ribo-Zero to remove abundant cytoplasmic rRNA from total RNA samples. Remaining intact RNA was fragmented using a chemical mix, followed by firstand second-strand cDNA synthesis using random hexamer primers. "End-repaired" fragments are ligated with unique Illumina proprietary adapters. All individually indexed samples were pooled together and multiplexed for sequencing. Libraries were sequenced using the Illumina Hiseq 2500 sequencing system and paired-end 101 bp reads were generated for analysis. Fastq files were aligned against human reference (hg19/hGRC37) using the Tophat2. Duplicate reads were removed using MarkDuplicates from Picard tools, and per gene read counts for Ensembl (v75) gene annotations were computed using htseq-count. Differential gene expression analysis was performed using R-package EdgeR. Principal component analysis was computed using R function prcomp. Hierarchical clustering and heatmaps were generated using R.
Isolation, culture, and oxygen glucose deprivation (OGD) treatment of cardiomyocytes Isolation of neonatal rat cardiomyocytes (NRCMs) were performed as previously reported 50 . Briefly, the NRCMs were prepared by enzymatic digestion of hearts obtained from newborn (1 day old) Sprague-Dawley rat pups and plated on cell culture grade plates (coated with gelatin) at a density of 5 × 10 5 cells/cm 2 in DMEM/F12 medium and maintained at 37°C in humid air with 5% CO 2 . NRCMs were subjected to hypoxia in vitro in an oxygen control cabinet (Ruskinn, England) mounted within an incubator and equipped with oxygen controller and sensor for continuous oxygen level monitoring. A mixture of 85% nitrogen, 10% hydrogen, and 5% CO 2 was utilized to create hypoxia and the O 2 in the chamber was monitored and maintained at a level <0.1%. The NRCMs were treated with OGD injury by cultured in the culture medium without serum and glucose and in hypoxic condition (<0.1% O 2 ) for 6 h as previously reported 51 . The EV-N or EV-H (1 μg/mL), miR-497-mimic (Ribobio, Guangzhou, China) or/and pcDNA3.1(-)-MALAT1 plasmid (Shanghai Integrated Biotech Solutions), MALAT1 GapmeR or control (Guangzhou Epibiotek) with Lipofectamine 3000 (Thermo Fisher Scientific) were added into the culture medium of the NRCMs. Then, the cardiomyocytes were stained by TUNEL, cTnT, and DAPI to evaluate the level of cell death in the NRCMs. The Image-processing software (Image J) was used to quantify the TUNEL + nucleus and total nucleus. The Cell Counting Kit-8 (Sigma-Aldrich) was applied and absorbance of formazan dye produced by living cells was measured in an Infinite M200 Microplate Reader (Tecan, Maennedorf, Switzerland). The culture medium from various groups was collected to measure the activity of lactate dehydrogenase (LDH) release with a LDH Release Assay Kit (Beyotime, China).
Isolation of adult mouse cardiomyocytes (AMCMs) were performed by a langendorff-based approach as previously reported 52 . Briefly, male C57BL/6 mice aged 10-12 weeks were anesthetized with pentobarbital (50 mg/kg body weight, i.p.), then the chest was opened to lift the heart and cut the aorta. The heart was then attached to the perfusion cannula through the aorta and perfused with the digestion buffer (Collagenase II 0.5 mg/ mL (Worthington, USA), Protease XIV 0.05 mg/mL (Sigma-Aldrich, Singapore)) for 10-15 min at the flow rate of 4 ml/min following 3 min of perfusion in the perfusion buffer. Once appeared flaccid, the heart was removed from the cannula and the ventricle was cut into sections followed by dissociated the cells with a plastic pipette in digestion buffer. Further, the cell suspension underwent gravity settling for four rounds within Ca 2+ reintroduction buffers. The cells resuspended with a plating medium were plated onto laminin (5 μg/mL, Thermo Scientific, Singapore) coated culture dish after passed through a 100-μm filter, in a humidified tissue culture incubator (37°C, 5% CO 2 ), and after 1 h, the cells were cultivated in the culture medium 53 . The AMCMs were treated with hCVPC-EVs (1 μg/mL) and PBS (equal volume) for 24 h, and then were washed twice with PBS before RNA extraction and RT-qPCR.
For tube formation assay, 48-well plates were coated with 200 μL of Growth Factor Reduced Matrigel (Corning) per well and incubated at 37°C for 30 min. HUVECs were seeded at a density of 5 × 10 4 cells/cm 2 in the DMEM Basal Medium and maintained at 37°C in humid air with 5% CO 2 for 8 h. The EV-N or EV-H (1 μg/mL), miR-497-mimic (Ribobio, Guangzhou, China) or/and pcDNA3.1(−)-MALAT1 plasmid (Shanghai Integrated Biotech Solutions), MALAT1 GapmeR or control (Guangzhou Epibiotek) with Lipofectamine 3000 (Thermo Fisher Scientific) were added into the culture medium of the HUVECs. Capillary-like structure formation by HUVECs was recorded using a Zeiss inverted microscope. The Image J was used to quantify thetubelength 40 .

Luciferase activity
To verify the direct interactions between MALAT1 and miR-497, the hMALAT1 cDNA full length (8779 bp) was cloned into psiCHECK2 (Promega, Madison, WI) from the pcDNA3.1(−)-MALAT1 plasmid (Shanghai Integrated Biotech Solutions) between the XhoI and NotI restriction endonuclease sites. The map of psiCHECK2-MALAT1 plasmid was shown in Supplementary Fig. S2. The renilla luciferase, hRluc, was used to monitor changes in expression as the result of MALAT1 and miRNA interaction. Synthetic firefly luciferase gene (hluc+) was the control luciferase. SV40 promoter was the promotor of hRluc and MALAT cDNA full length sequence. Lipofectamine 3000 (Thermo Fisher Scientific) was used to cotransfect the plasmids together with synthetic negative control or miR-150-or miR-497-mimic (Ribobio, Guangzhou, China) into HEK293T cells growing in 48well plates. Twenty-four hours later, the cells were lysed in 1× passive lysis buffer, and the activities of Renilla luciferase and firefly luciferase were determined with the Dual Luciferase Reporter Assay System (Promega).

Statistics
Data are expressed as mean ± SEM. Statistical significance was analyzed by using the unpaired Student's t test or one-way analysis of variance (ANOVA) followed with Bonferroni's multiple as appropriate. Two-way ANOVA was applied with Tukey's multiple comparison for analysis of echocardiographic data. Statistical analyses were performed with Graphpad Prism software (version 6.1). A p value <0.05 was considered statistically significant.  Fig. S3b). Transmission electron micrographs of hCVPCs demonstrated the presence of EV-like vesicles within multivesicular bodies (MVBs) in the cytoplasmic area (Fig. 1a). The secreted EVs were isolated from hCVPCs and showed a doublemembrane-bound, cup-shaped typical shape (Fig. 1b). Nanoparticle tracking analysis (NTA) confirmed the mode size of secreted EVs from hCVPCs was around 118 nm in the EV-N and 110 nm in the EV-H (Fig. 1c), with the particle concentrations around 0.82 × 10 8 /mL in the original hCVPC supernatant and 0.95 × 10 8 /mL in the hypoxia-treated hCVPC supernatant (Fig. 1d). The Western blot analysis further confirmed the EV marker proteins ALG-2 interacting protein X (ALIX), tumor susceptibility gene 101 protein (TSG101), CD63, CD9, and CD81 in the hESCs, hCVPCs and the EV preparations derived from these cells (Fig. 1e). These data demonstrate the successful isolation of hCVPC-secreted EVs.

Intramyocardial delivery of hCVPC-EVs improves post-MI cardiac function and reduces scar size
To assess the therapeutic efficacy of hCVPC-EVs when delivered at the acute phase of the murine MI model, the EVs secreted from hCVPCs under normoxic (EV-N) and hypoxic (EV-H) culture conditions were intramyocardially injected in mice with AMI. The functional outcome was assessed at 2, 7, 14, and 28 days post MI as indicated in Fig. 2a. The establishment of MI model by permanent ligation of the LAD coronary artery was confirmed by echocardiography analysis (Fig. 2b). Changes in LVEF and LVDs (Fig. 2b) were similar among the vehicle PBS control, EV-N and EV-H groups at day 2 post MI, indicating a comparable infarcted area among the groups when the MI model was undertaken. However, MI-worsened LVEF and LVDs in the PBS control group were significantly improved in the EV-N and EV-H groups during the observation time up to 28 days post MI, with the EV-H showed a better recovery tendency (Fig. 2b). Consistently, Masson's trichrome-staining analysis showed that the scar area/LV area was significantly smaller in the EV-N and EV-H groups than that in the PBS group at day 28 post MI (Fig. 2c). Micro-PET analysis further showed the uniform uptake of fludeoxyglucose (FDG), demonstrating normal myocardial viability 59 in the heart of Sham mouse, a smaller FDG-uptake area in the MI heart, and partially restored areas in the EV-N and EV-H groups at day 28 post MI (Fig. 2d). Taken together, these results demonstrate that the hCVPC-secreted EVs have cardioprotective effects in augmenting cardiac function and limiting fibrosis formation. Moreover, the EVs secreted from hCVPCs under the hypoxia-culture condition appear to have a better benefit in improving cardiac function of the infarcted hearts.

hCVPC-EVs improve cardiomyocyte survival in vivo and in vitro
To determine whether the inhibition of cardiomyocyte death is one of mechanisms underlying the cardioprotective effects of hCVPC-EVs, we examined the number of TUNEL + -cardiomyocytes in the infarcted hearts. The TUNEL + cardiomyocytes in the border zone of day 3 post MI hearts were significant reduced in the EV-N and EV-H groups and, notably, the number of TUNEL + cardiomyocytes in the EV-H group were lesser than that in the EV-N (Fig. 3a), indicating that the EVs secreted from hypoxia-conditioned hCVPCs have a better effect on the inhibiting of cardiomyocyte death. These findings suggest that hCVPC-EVs delivery can protect cardiomyocytes from MI-induced cell death, and this effect can be enhanced by hypoxia-conditioning of hCVPCs.
To determine the direct effect of hCVPC-EVs on the cardiomyocytes, we examined the role of hCVPC-secreted EVs in the NRCMs with the OGD injury. Consistent with the in vivo data, the OGD treatment significantly reduced the cell viability, while the reduction was significantly attenuated in the EV-N-and the EV-H-treated cells, and the latter showed a better protective effect than that in the EV-N (Fig. 3b). In addition, the OGD-enhanced LDH activity in the culture medium of NRCMs was significantly reduced in the EV-N and EV-H groups (Fig. 3c). These results suggest that both EV-N and EV-H can improve the survival of cardiomyocytes during in vivo MI and in vitro OGD injury.

hCVPC-EVs promote angiogenesis in vivo and in vitro
Angiogenesis can also contribute to the reduction of cell death and fibrosis size in the infarcted hearts. We therefore examined the blood vessel density in the infarcted hearts. Immunohistological chemistry analysis showed that the number of CD31 + vessels (Fig. 4a) and α-SMA + vessels (Fig. 4b) at the border zone of day 28 post MI hearts was significantly increased in the EV-N and EV-H groups compared with these in the PBS control ones. The effect of hCVPC-EVs on endothelial cells was further examined by tube-formation assay using HUVECs. The tube length of HUVECs treated with hCVPC-EV-N or hCVPC-EV-H for 8 h was significantly increased compared with that in the PBS control group, with a better effect in the hCVPC-EV-H than that in the hCVPC-EV-N (Fig. 4c). The HUVEC migration after the inhibition of proliferation by 10 μg/mL mitomycin was promoted in the hCVPC-EV-treated groups (Fig. 4d). These data demonstrate that the beneficial effect of hCVPC-EVs in endothelial cells may contribute to the promoted angiogenesis seen in the hCVPC-EV-treated infarcted hearts.

Identification of RNAs in the hCVPC-EVs
To understand the mechanism of the cardioprotective effect of hCVPCs-Evs, we performed RNA-seq analysis on the EVs isolated from hCVPCs with or without hypoxiaconditioning and then aligned them to a database for RNA annotations (Fig. 5a). Some mRNAs with known functions were detected in both EV-N and EV-H groups, such as insulin like growth factor 2 (IGF2) and superoxide dismutase 2 (Sod2), which were reported to improve cardiomyocyte survival 47,60 . Similarly, ferritin heavy chain 1 (FTH1) and endothelial PAS domain protein 1 (EPAS1) mRNA detected in the EV-N and EV-H have also been shown to promote angiogenesis [61][62][63] . The differential analysis using corrected counts per million (CPM) identified five genes with significant changes between the EV-N and the EV-H groups that were segregated by hierarchical clustering (Fig. 5b). Among them, four mRNAs, FTH1, TPT1, BLOC1S6, MT-ATP6, and one lncRNA MALAT1 were differentially expressed between the two groups (Fig. 5b). The CPM values of the sequencing results showed that MALAT1 abundance in the EV-H was significantly higher than that in the EV-N (Fig. 5c). The differential expression for MALAT1 was further confirmed by RT-qPCR analysis (Fig. 5d). These data suggest that the MALAT1 seems exist in the hCVPC-EVs and might be involved in the beneficial effects of the hCVPC-EVs in the cardioprotection.

LncRNA MALAT1 in the hCVPC-EVs contributes to cardiomyocyte protection and tube formation
To determine whether MALAT1 contributes to cardiomyocyte survival and tube formation of endothelial cells, we performed experiments with knockdown and overexpression of hMALAT1. The hMALAT1 abundance was significantly upregulated in the MI myocardium injected with the hCVPC-EVs compared with these in the Sham and the MI groups (Fig. 6a). Similarly, the hMALAT1 abundance was enhanced in the NRCMs (Fig. 6b) and the AMCMs (Supplementary Fig. S4) treated with the hCVPC-EVs. Overexpression of hMALAT1 was confirmed to improve the viability of NRCMs during OGD injury (Fig. 6c). In addition, the hMALAT1 was knocked down by~90% with MALAT1 GapmeR in the recipient Fig. 3 hCVPC-extracellular vesicles reduce cell death of cardiomyocyte. a Representative and quantification of IHC staining for TUNEL + cells in the border zone of infarcted hearts at day 3 post MI and sham hearts. n = 12 slices from 4 hearts each group. Scale bar, 100 μm. b Cell viability of NRCMs under OGD injury treated with the EV-N and the EV-H, n = 6. Scale bar, 100 μm. c The activity of LDH in the culture medium of NRCMs subjected to a 6-h of OGD injury with and without the treatment of 1 μg EV-N or EV-H. n = 5. *p < 0.05, **p < 0.01, ***p < 0.001. cTnT cardiac troponin T, TUNEL terminal deoxynucleotidyl transferase dUTP nick end labeling, OGD oxygen glucose deprivation.

MALAT1 exerts cardioprotective effects by targeting miR-497
Next, we investigated the potential molecular target of MALAT1 in the cardioprotection. The potential binding of MALAT1 to miR-497 was predicted by using miRcode Website (mircode.org) and DianaTools Website (http:// carolina.imis.athena-innovation.gr/diana_tools/web/index. php?r=lncbasev2%2Findex-predicted). The analysis showed that there were six predicted binding sites of MALAT1 with miR-497 ( Supplementary Fig. S5). The expression of miR-497 was detected in the mouse ventricles of Sham, MI + PBS, and MI + hCVPC-EV groups ( Supplementary  Fig. S6a) as well as in the NRCMs (Supplementary Fig. S6b). To confirm it, we constructed luciferase vectors carrying the hMALAT1 cDNA full length sequence to examine the potential direct bindings between the MALAT1 and miR-497, and miR-150 was used as a negative control. As shown in Fig. 7a, the luciferase activity was significantly decreased in the miR-497-but not the miR-150-mimic transfected cells compared with that of the control-mimic transfected cells. The RNA pull down assay further confirmed that the hMALAT1 was enriched about the 8.2-fold in the Bio-miR-497-5p group (biotin-labeled miR-497-5p) compared with the control group (Bio-NC, biotin labeled scramble RNA) in the HUVEC, indicating that MALAT1 could bind with miR-497-5p (Fig. 7b). Next, we examined the function of MALAT1 and miR-497 on cardiomyocytes and endothelial Fig. 4 Effects of hCVPC-EVs on HUVECs migration, tube formation and NRCMs protection. a, b Representative and quantification of immunohistochemical (IHC) staining for CD31 + endothelial cells (a) and α-SMA + blood vessels (b) in the border zone of infarcted hearts at day 28 post MI. n = 12 slices from 4 hearts each group. c Tube formation of HUVECs treated with EV-N and EV-H (n = 6). d Migration of HUVECs treated with the EV-N and the EV-H after inhibition of proliferation with 10 μg /mL mitomycin for 2 h (n = 5). Scale bar, 100 μm. *p < 0.05, **p < 0.01, ***p < 0.001. cells. In NRCMs, OGD injury-induced reduction of cardiomyocyte viability (Fig. 7c) and increased TUNEL + cardiomyocytes (Fig. 7d) were aggravated by miR-497-mimic and the protective effects of MALAT1 were also reversed by miR-497-mimic. Similarly, the tube formation of HUVECs was impaired by miR-497-mimic and MALAT1 promoted tube formation was also abrogated by miR-497-mimic (Fig. 7e). These results indicate that MALAT1 from hCVPC-EVs may exert its cardioprotective effects by targeting miR-497.

Discussion
In this study, we report that the hESC-CVPC-secreted EVs protect acutely infarcted hearts from progressive worsening of cardiac function and reduce scar formation when delivered at the early phase of MI. Our results point to the protective effects being associated with the improvement in the existing cardiomyocyte survival and vascularization. Further analysis shows that the cardioprotective effects of hCVPC-EVs appear to be partially mediated by the hCVPC-EV-contained lncRNA MALAT1 via targeting miR-497 and the abundance of MALAT1 in the hCVPC-EVs is upregulated in hypoxia-treated hCVPCs (Fig. 8). These findings reveal the previously unrecognized role of the hCVPC-secreted EVs in cardiac protection when given during the acute phase of MI and provide new insights into the RNA content of hCVPC-EVs and how a lncRNA mediator exerts cardioprotective effects.
EVs secreted from stem cells and hPSC-derived cardiac lineage cells hold tremendous promise for cardioprotection, while the secreted EVs are not equal among cells 9,30,40,64,65 . Here, we demonstrate for the first time that the hCVPC-EV delivery administrated at the acute phase of MI improves cardiac function and reduces scar tissue, which recapitulates the beneficial effects of their parent cells in the treatment of AMI in the murine 25 and nonhuman primate model 25,26 . These findings, together with the observations of the cardioprotective effects from various stem cells and their secreted EVs when delivery at the acute phase of MI 39,40,[65][66][67][68] or during early reperfusion 69 , indicate the acute phase of MI is a critical window for promoting infarct healing.
Our findings are consistent with the observation of the improvement of cardiac function when the delivery of EVs either secreted from SSEA1 + -hESC-or hiPSC-derived CVPCs (hCVPC-EVs) into the murine hearts after 2-3 weeks of post-MI chronic HF 32,33 . Moreover, the hCVPC-EVs recapitulate the beneficial effects of their parent cells in the treatment of either AMI 25,26 and chronic post-infarct HF murine model, suggesting that the EVs are active elements of secretome of hPSC-derived cardiac lineage cells, playing major roles in the salvage of infarcted myocardium. Similar observations are obtained from the EVs secreted by hPSC-derived cardiomyocytes 41,65,68,70 . Considering the allogeneic EVs do not induce significant immune responses of the parent cells after repeated dosing 30 , EVs secreted by stem cells and hPSC-derived cardiovascular cells might be used as therapeutic agents, and these vesicles need to be thoroughly dissected to explore the potential use of a cellderived but cell-free therapy for cardioprotection.
Another finding here is that the EVs from hypoxic hCVPCs have better benefits in the improvement of cardiomyocyte survival and the promotion of vascularization than these in the normoxic ones. Similar results showing the enhanced cardioprotective effects from the EVs secreted by hypoxic c-kit + cells 69 and MSCs 13 were observed in a rat model of I/R and a AMI model of nonhuman primates. These observations suggest that the secretome of implanted stem/progenitor cells would alter in the postinfarct tissue and a certain degree of hypoxiapreconditioning of the cells before the implantation would be a way to enhance the therapeutic effect of both cells and secreted vesicles.
It has been shown that the EVs communicate with the target cells directly and transfer the RNAs and proteins representing their parental cells 37,51,71 . Cumulated c The CPM values of the sequencing results showed that MALAT1 abundance in EV-H was significantly higher than that in EV-N. d MALAT1 expression was verified by RT-qPCR. n = 5. *p < 0.05. CPM counts per million, PCA principle component analysis. evidence has shown that the miRNAs in the stem cellsecreted EVs contribute critically to the vesicle-mediated benefits for cardioprotection 40,69,72 . The hCVPC-EVs have been shown containing the specific miRNA signature associated with the tissue-repair pathways 33 , while little is known whether lncRNAs in the EVs from hPSCderived cardiac lineage cells. Here, we identified the existence of lncRNA MALAT1 in the hCVPC-EVs. This is supported by the following observations: (i) the MALAT1 is detected in the RNAseq analysis from the EVs secreted by hCVPC-EV-N and its abundance is higher in the EVs secreted from hypoxic treated hCVPCs than in the hCVPC-EV-N (Fig. 5a-c); (ii) these results are confirmed by RT-qPCR analysis (Fig. 5d); (iii) the high abundance of exogenous hMALAT1 is detected in the cardiac tissue of the infarcted hearts injected with hCVPC-EVs compared with that in the Sham group and the MI group (Fig. 6a); and (iv) the hMALAT1 abundance is significantly upregulated in the cultured NRCMs (Fig. 6b) and the AMCMs (Supplementary Fig. S4) treated with the hCVPC-EVs.
Supportively, the lncRNAs are detected in the EVs secreted from hESC/iPSC-derived cardiomyocytes 41 .
We further confirmed that the MALAT1 protects the cardiomyocytes from OGD injury and promotes tube formation of endothelial cells. These effects may partially contribute to the cardioprotection of hCVPC-EVs as EVs contain many kinds of nucleic acids and proteins. Consistently, MALAT1 has been detected in the EVs from human umbilical cord MSCs and shown to prevent aginginduced cardiac dysfunction 44 . It also contributes the vascular growth in mouse retinal angiogenesis and hindlimb ischemia model 43 . Other kinds of cardioprotective components need to be investigated in further research.
Mechanistically, we identified that MALAT1 exerts its beneficial effects through targeting miR-497. This is supported by the following observations: (i) the luciferase activity assay suggests that miR-497 might be a target of MALAT1 (Fig. 7a); (ii) the miR-497-mimic aggravates the cardiomyocyte injury and canceled the protective effect of MALAT1 in the cardiomyocytes (Fig. 7c, d); (iii) MALAT1-promoted the tube formation of endothelial cells is abrogated by miR-497 (Fig. 7e); and (iv) the RNA pull down assay showed that MALAT1 has an interaction with miR-497 in the HUVECs (Fig. 7b). These data reveal a new target of MALAT1 in the cardioprotection. The deteriorative effects of miR-497 in cardiomyocytes is consistent with the observation showing that the inhibition of miRNA-497 ameliorates anoxia/reoxygenation-induced cardiomyocyte injury by suppressing cell apoptosis and enhancing autophagy 73 . Although our results identify the binding of MALAT1 to miR-497, it is unclear how many MALAT lncRNA need to be delivered to infarcted hearts to sufficiently bind and inhibit miR497. Taken together, the MALAT1 in hCVPC-EVs appears to be one of the mechanisms partially contribute to the cardioprotective effect of hCVPC-EVs via promoting angiogenesis and inhibiting cardiomyocyte death by targeting miR-497. The effects of MALAT1 and miR-497 in the infarcted hearts need to be identified and the mechanisms of hCVPC-EVs in cardioprotection are required to be thoroughly elucidated.
Limitations: The cardiomyocytes we used for the in vitro injury model are NRCMs (Fig. 7), which do not always represent the response in the adult The RNA pull down assay in the HUVECs transfected with biotin-labeled miR-497-5p (Bio-miR-497-5p). The MALAT1 combined with Bio-miR-497-5p was pulled down with Pierce ™ Streptavidin Magnetic Beads. c The cell viability in the NRCMs under OGD injury after transfection of pcDNA3.1-MALAT1 and/or miR-497. d Representative and quantification of IHC staining for TUNEL + NRCMs after OGD injury. e HUVEC tube formation after the transfection of pcDNA3.1-MALAT1 and/or miR-497. TUNEL terminal deoxynucleotidyl transferase dUTP nick end labeling, OGD oxygen glucose deprivation. Scale bar, 100 μm. n = 5 each for (a-e). *p < 0.05, **p < 0.01, ***p < 0.001.
cardiomyocytes. The cardioprotective effects of MALAT1 and the underlying mechanisms need to be verified in the infarcted hearts.
In conclusion, we have shown that the delivery of hCVPC-EVs during early phase of AMI can promote recovery of cardiac function and reduce fibrosis formation. These beneficial effects are enhanced by using the EVs secreted from hypoxia-treated hCVPCs and are associated with the improvement of existing cardiomyocyte survival at the acute phase of infarcted hearts and the promotion of angiogenesis. Moreover, the hCVPC-EVs contain lncRNA MALAT1, which seems partially contribute to the inhibition of cardiomyocyte death and promotion of angiogenesis via targeting miR-497. In addition, the MALAT1 abundance in the hCVPC-EVs can be upregulated by hypoxia-conditioning of hCVPCs.
These findings provide new insights into the cardioprotective mechanisms of hCVPCs and the cell-secreted EVs, and suggest that hCVPC-EVs might be used as a tool to understand the mechanism of infarct healing and promote healing of infarcted hearts. Delivery of hCVPC-EVs during early phase of MI has many beneficial effects of promoting cardiac function, reducing fibrosis, improving cardiomyocyte survival, and promoting angiogenesis. These beneficial effects are enhanced by using the extracellular vesicles secreted from hypoxic hCVPCs. Moreover, the hCVPC-EVs contain lncRNA MALAT1, which appear to contribute to the inhibition of cardiomyocyte death and promoting of angiogenesis via targeting miR-497, and the MALAT1 abundance in the hCVPC-EVs can be up-regulated by hypoxia-conditioning of hCVPCs.