Adult cardiomyocytes‐derived EVs for the treatment of cardiac fibrosis

Abstract

Cardiac fibrosis is a common pathological feature of cardiovascular diseases that arises from the hyperactivation of fibroblasts and excessive extracellular matrix (ECM) deposition, leading to impaired cardiac function and potentially heart failure or arrhythmia. Extracellular vesicles (EVs) released by cardiomyocytes (CMs) regulate various physiological functions essential for myocardial homeostasis, which are disrupted in cardiac disease. Therefore, healthy CM‐derived EVs represent a promising cell‐free therapy for the treatment of cardiac fibrosis.

To this end, we optimized the culture conditions of human adult CMs to obtain a large yield of EVs without compromising cellular integrity by using a defined combination of small molecules. EVs were isolated by ultracentrifugation, and their characteristics were analysed. Finally, their effect on fibrosis was tested.

Treatment of TGFβ‐activated human cardiac fibroblasts with EVs derived from CMs using our culture system resulted in a decrease in fibroblast activation markers and ECM accumulation. The rescued phenotype was associated with specific EV cargo, including multiple myocyte‐specific and antifibrotic microRNAs, although their effect individually was not as effective as the EV treatment. Notably, pathway analysis showed that EV treatment reverted the transcription of activated fibroblasts and decreased several signalling pathways, including MAPK, mTOR, JAK/STAT, TGFβ, and PI3K/Akt, all of which are involved in fibrosis development. Intracardiac injection of CM‐derived EVs in an animal model of cardiac fibrosis reduced fibrotic area and increased angiogenesis, which correlated with improved cardiac function.

These findings suggest that EVs derived from human adult CMs may offer a targeted and effective treatment for cardiac fibrosis, owing to their antifibrotic properties and the specificity of cargo.

1. INTRODUCTION

Cardiovascular diseases are the leading cause of death worldwide (McDonagh et al., 2021). Although there are multiple types of cardiovascular diseases, nearly all of them involve cardiac fibrosis (Frangogiannis, 2021; Tian et al., 2017). In adults, tissue repair after an injury is generally facilitated by fibrosis (Nagpal et al., 2016). Specifically, in the heart, due to its limited regenerative capacity with only 1% of annual turnover (Bergmann et al., 2009), the fibrotic scar acts to preserve myocardial structure under tissue loss and mechanical stress. However, an erroneous superactivation of fibroblasts may occur, causing excessive cell division and extracellular matrix (ECM) accumulation. This phenomenon increases the stiffness and disrupts the electrical properties of the heart muscle, decreasing cardiac function and eventually causing death (Furtado et al., 2016; Nagpal et al., 2016; Weber et al., 2013). Despite the improvements, there is no current treatment capable of curing cardiac dysfunction and effectively decrease cardiac fibrosis, only drugs to ameliorate the consequences, such as β‐blockers (McDonagh et al., 2021). Thus, the treatment of excessive fibrosis is a major therapeutic goal that must be archived.

Initial studies of myocardial regeneration focused on the use of embryonic stem cells (ESs) and mesenchymal stem cells (MSCs), which were directly implanted into the heart (Hong et al., 2014; Sharma et al., 2017; Stamm et al., 2003). However, transplanted cell retention was lower than 5% after a few days (Hong et al., 2014; Tang et al., 2005), showing only moderate regenerative results. Nevertheless, the conducted experiments revealed that the clinical benefits exerted by the implanted cells predominantly arise from their secretome, including extracellular vesicles (EVs) (Chong et al., 2014; Hong et al., 2014; Khan et al., 2015).

EVs, including exosomes and microvesicles, are lipid bilayer structures (40–1000 nm) released by all cells in the body that act as a communication tool among cells, both nearby or entering the bloodstream and traveling to distant places (Witwer & Théry, 2019). EVs contain multiple molecules, including RNA, DNA and proteins. This cargo is responsible for the altered behaviour of the recipient cell (Ratajczak et al., 2006; Skog et al., 2008). EV communication has been found both in physiological and pathological conditions, and the fate depends on the cell of origin state; erratic malignant cells will contain malicious cargo with the objective of promoting tumour growth (Kosaka et al., 2016), while healthy cells will secrete EVs beneficial for the correct functioning of the tissue homeostasis (Buzas, 2022). Hence, healthy cells‐derived EVs may be a novel therapeutic agent. Indeed, the use of EVs as a cell‐free therapy is very promising for multiple reasons, including reduced ethical concerns, high stability of the product, and low immunogenicity allowing a nonpersonalized treatment (Rogers et al., 2020). Pioneering studies in EV research concentrated on undifferentiated cells such as induced pluripotent stem cells (iPSCs) or MSCs, which showed a partial shrinkage in myocardial infarct (MI) size by transmitting cytoprotective microRNAs (Arslan et al., 2013; Charles et al., 2020; Wang et al., 2015). Per contra, we are planning to use EVs derived from cells resident in the adult human heart (cardiac cells; CC) and focus on their effect on human cardiac fibroblasts (CFs). Since EVs contain molecules characteristic of their cell of origin (Gao et al., 2020; Jenjaroenpun et al., 2013), we believe that the EV cargo derived from adult CCs will contain multiple molecules to regulate cardiac‐specific pathways that may be beneficial for cardiac fibrosis and heart remodelling. Specifically, we focused on microRNAs contained in EVs since microRNA are easily capable of altering the transcriptome of recipient cells (Simons & Raposo, 2009). However, in order to collect a substantial quantity of EVs, a significant number of cells is required. Consequently, one of the primary limitations in our experimental design is the terminal differentiation of adult CCs and their limited lifespan in vitro. Thus, prior to EV collection, we aimed to improve the primary culture of CCs. For that, the combination of the small molecules Y‐27632, a Rock inhibitor (Y); A‐83‐01, a TGF‐β inhibitor (A); CHIR9902, a GSK3 inhibitor (C); and PD0325901, a MEK inhibitor (P), was used in accordance with previous studies where it was acquired as a long‐term culture while maintaining the original cell function (Katsuda et al., 2017, 2019; Kawamata & Ochiya, 2010).

Here, we show an extended adult human CCs, especially cardiomyocyte (CM) cell culture by the use of small molecules, which maximizes the extensive collection of EVs without inducing any malignant phenotype. CM‐derived EV cargo was confirmed to contain multiple antifibrotic and cardiac‐specific microRNAs. Due to this beneficial cargo, EVs reverted the phenotype of activated fibroblasts in vitro, and in vivo reduced fibrosis and overall, improved cardiac function.

2. METHODS

2.1. Cell culture

PC‐100‐021 (lot. 81030171), coronary artery smooth muscle cells, and human cardiac myocytes (HCM, lot. #436z024.6, 33‐year‐old female) cardiomyocyte primary cell lines were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) and Promocell (Heidelberg, Germany), respectively, and tested to verify the absence of mycoplasma contamination. PC‐100‐021 cells were cultured in Vascular Cell Basal Medium containing basic fibroblast growth factor (FGF) (2 ng/mL), insulin (5 μg/mL), ascorbic acid (50 μg/mL), L‐glutamine (10 mM), epithelial growth factor (EGF) (0.5 ng/mL) and 5% foetal bovine serum (FBS), which are contained in the culture medium kit (ATCC). HCM cells were cultured in Myocyte Growth Medium Kit with Basic FGF (2 ng/mL), Insulin (5 μg/mL), EGF (0.5 ng/mL) and 5% FBS, which are contained in the medium kit (Promocell). PC‐100‐021 and HCMs were cultured in a collagen dish (AGC, Tokyo, Japan).

In addition to the preestablished medium, multiple combinations of small molecules were added at the following final concentrations: Y‐27632 (10 μM) (Wako, Japan), A‐83‐01 (1 μM) (Wako), CHIR99021 (3 μM) (Axon Medchem, Groningen, Netherlands) and PD0325901 (1 μM) (Wako). The combinations were Y, A, C, P, YA, YC, YP, AC, AP, CP, YAC, YAP, ACP, YCP and YACP. In the case of more than one small molecule combination, the concentration was additive. For instance, YAC, Y‐27632 (10 μM), A‐83‐01 (1 μM) and CHIR99021 (3 μM).

The cardiac fibroblast primary cell line was purchased from Promocell and maintained in Dulbecco's modified Eagle's medium (DMEM) (Sigma, CA, USA) with 10% heat‐inactivated FBS (Thermo Fisher Scientific, LA), 1% L‐glutamine (Gibco, LA, USA) and 1× antibiotics and antimitotic (Gibco). MSC cells were cultured in MesenPRO (Thermo Fisher Scientific) for expansion. For EV collection, the medium was changed to StemPro (Thermo Fisher Scientific). All the cells were grown at 37°C in a 5% CO₂ incubator, the medium was changed every 3 days, and the cells were passaged when they reached 70%–80% confluence.

For cardiac fibroblast activation, 10 ng of Recombinant Human TGFβ1 (Preprotech, Cranbury, NJ, USA) was added for 24 h prior to any EV treatment. During the EV treatment, TGFβ1 was also present in the media.

2.2. EV collection

Once the cell culture was close to confluence, the cells were washed with phosphate buffered saline (PBS), and the culture medium was replaced with advanced DMEM (Thermo Fisher Scientific). Cells were cultured for 2 more days, collected, and centrifuged at 2000 × g for 10 min at 4°C. For removal of cellular debris, the supernatant was filtered through a 0.22 μm filter (Millipore, Burlington, MA, USA). The conditioned medium was then used for EV isolation. Samples were ultracentrifuged at 210,000 × g for 70 min at 4°C using an SW41Ti rotor (Beckman Coulter, CA, USA) and resuspended in PBS. For animal experiments, the resuspended EVs were further concentrated using a pressure evaporator (Labconco Refrigerated CentriVap Concentrator, Labconco, KS, USA). The protein concentration of the EV fraction was determined using a Quant‐iT Protein Assay with a Qubit 2.0 Fluorometer (Invitrogen).

To remove possible protein contaminants, size exclusion chromatography (SEC) columns were used to collect EVs. Conditioned medium was first centrifuged at 2000 × g for 10 min and filtered to remove cell debris. A total of 10 mL of conditioned medium was loaded into to a single column and fractionated according to manufactures protocol (#ICS35‐1693, Izon Science, New Zealand). The eluate was collected and fractions from 4 to 8, which contain mostly EVs, were collected and mixed. Then Amicon Ultra‐15 (#UFC910024, Merck Millipore) were used to concentrate the samples.

2.3. Nanoparticle tracking analysis

For quantification of the particle number in EV samples, nanoparticle tracking analysis (NTA) was carried out using the NanoSight LM10‐HS system (NanoSight, United Kingdom). The results are presented as the average ± SD of three independent experiments of the 60‐s video with camera level 13.

2.4. Cryo‐electronic microscopy

Cryo‐EM was used for direct visualization of vesicles. Two microliters of EV pellets, isolated as described above, were applied to a Quantifoil holey carbon grid (Mo, R1.2/1.3, Quantifoil Micro Tools GmbH, Germany) and vitrified by rapid plunging in precooled liquid ethane with Vitrobot Mark IV (Thermo Fisher Scientific). Until imaging, the vitrified specimens were stored under liquid nitrogen. The EM grids were examined at liquid nitrogen temperature with a cryo‐electron microscope (JEM‐2200FS, JEOL, Japan) equipped with a field emission gun and an omega‐type energy filter at an accelerating voltage of 200 kV. A slit width of 15 eV was used to obtain a zero‐energy loss electron beam. Images were recorded with a DE‐20 direct electron detection device (Direct Electron LP, CA, USA).

2.5. ExoScreen

The detailed principle and analytical methods were presented in a previous report (Yoshioka et al., 2014). Prior to the experiment, AlphaLISA unconjugated acceptor beads (PerkinElmer, Inc., Waltham, MA, USA) were conjugated with the typical EV markers CD9 (SHI‐EXO‐M01, CosmoBio, Tokyo, Japan) and CD63 (SHI‐EXO‐M02, CosmoBio).

A 96‐well half‐area white plate was filled with 5 μL of sample and mixed with 5 nM biotinylated antibodies (PerkinElmer Inc., Waltham, MA, USA) and 50 μg/mL acceptor bead‐conjugated antibodies against human CD9 and CD63. After 3 h of incubation, AlphaScreen streptavidin‐coated donor beads (PerkinElmer Inc.) were added and incubated for 30 more minutes in the dark. The plate was read in an EnSpire Alpha 2300 Multilabel Plate Reader (PerkinElmer Inc.) using an excitation wavelength of 680 nm and emission detection set at 615 nm.

2.6. EV internalization

Purified EVs derived from CM or MSCs were labelled with a PKH26 red fluorescence labelling kit (Sigma–Aldrich) and washed five times using a 100‐kDa filter (Microcon YM‐100, Millipore) to remove excess dye. PKH67‐ or PKH26‐labelled EVs were added to several types of human fibroblasts, mouse foetal fibroblasts, or cardiomyocytes. Sixteen hours later, cells were fixed with 4% paraformaldehyde (Wako) and stained with ActinGreen 488 ReadyProbes Reagent (Thermo Fisher Scientific) for 30 min and Hoechst33342 (Dojindo, Kumamoto, Japan) for 15 min. The amount of internalized EV in HUVECs was captured by confocal microscopy (Olympus, Tokyo, Japan) and quantified.

2.7. Metabolome analysis

For metabolite extraction, 50 μL of the EV sample was mixed with 200 μL of methanol containing internal standards (H3304‐1002, Human Metabolome Technologies, Yamagata, Japan) at 0°C to inhibit enzymatic activity. Subsequently, 150 μL of Milli‐Q water was added, and the mixture was subjected to centrifugal filtration through a Millipore 5‐kDa cut‐off filter (Ultrafree MC PLHCC, Merck Millipore) at 9100 × g, 4°C for 120 min to eliminate macromolecules. The resulting filtrate was then evaporated under vacuum and reconstituted in 25 μL of Milli‐Q water for subsequent metabolome analysis.

Metabolome analysis was performed following HMT's ω Scan package protocol, utilizing the capillary electrophoresis Fourier transform mass spectrometry (CE‐FTMS) as previously described (McDonagh et al., 2021). CE‐FTMS analysis was conducted using an Agilent 7100 CE capillary electrophoresis system, with the spectrometer scanning from m/z 60 to 900 in positive mode, and from m/z 70 to 1050 in negative mode, respectively (Sasaki et al., 2019). Peaks were extracted using MasterHands, an automatic integration software developed by Keio University (Japan), to obtain peak information including m/z, peak area and migration time (MT) (Sugimoto et al., 2010). The areas of annotated peaks were then normalized to internal standards and sample volume to derive relative levels of each metabolite. Y‐27632 was quantified based on a single‐point calibration curve using standard compounds. The minimum detectable concentration of Y‐27632 is between 0.03 and 0.01 μmol/g.

2.8. Coculture and EV treatment of fibroblasts

For evaluation of the effects of CM‐derived EVs on fibrosis activation, two systems were used. First, 10 μg/mL TGFβ was added to fibroblasts cultured in 24‐well plates. Twenty‐four hours later, CM was seeded on cell culture inserts (0.4 μm, Corning, NY, USA) and cocultured with and without small molecules in FBS (‐) medium for 48 h. Then, fibroblast RNA or protein was extracted. For EV treatment, fibroblasts cultured in 24‐well plates were treated with TGFβ for 24 h in FBS (‐) medium prior to EV treatment. The medium was changed, 5 μg/mL EVs were added, and the cells were cultured for 2 more days.

2.9. RNA extraction, RT‒qPCR and next generation sequencing (NGS)

Total RNA from cultured cells and exosomal RNA was extracted using QIAzol and the miRNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The extracted RNA concentration was measured by Nanodrop (Thermo Fisher Scientific).

For mRNA expression analysis, 1 μg of total RNA was reverse transcribed with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) and then quantified by TaqMan Gene expression analysis on a StepOne Real‐Time PCR System (Applied Biosciences). The expression levels were normalized to those of ACTB (TaqMan assay ID: Hs01060665), and the relative fold changes in mRNA expression levels were calculated using the formula 2^−ΔΔCt. The TaqMan probes used in this study were actin acta 2 (ACTA2) (Hs00426835_g1), GATA binding protein 4 (GATA4) (Hs00171403_m1), vascular cell adhesion molecule 1 (VCAM1) (Hs01003372_m1), myosin light chain 2 (MYL2) (Hs00166405_m1), troponin I3 (TNNI3) (Hs00165957_m1), cyclin dependent kinase inhibitor 1A (CDKN1A) (Hs00355782_m1) and cyclin dependent kinase inhibitor 2A (CDKN2A) (Hs00923894_m1). Prior to NGS analysis, the RNA quality of the samples was assessed by using an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA) with the Agilent RNA6000 pico kit and Agilent small RNA kit (Agilent Technologies).

For microRNA sequencing, libraries were constructed using the QIAseq miRNA Library Kit (#331505, Qiagen) according to the manufacturer's protocols. The pooled libraries were sequenced using NextSeq 500 (Illumina, Inc., San Diego, CA) in 76‐base‐pair (bp) single‐end reads. Then, the original FASTQ files generated by NextSeq were uploaded to the Qiagen GeneGlobe Data Analysis Center (https://geneglobe.qiagen.com) and aligned to the miRBase v21 (http://www.mirbase.org) databank. The unique molecular index (UMI) counts were quantified using the trimmed mean of the M‐value (TMM) method (Tian et al., 2017).

For RNA sequencing, library construction was performed with a TruSeq Stranded mRNA Library Prep Kit (Illumina) according to the manufacturer's protocols. The libraries of the samples were sequenced using NovaSeq 6000 (Illumina) in 100‐base‐pair (bp) paired‐end reads. The sequence reads were aligned to the mouse reference genome (mm10) using STAR 2.7.5c (Dobin et al., 2013). The aligned reads were subjected to downstream analyses using StrandNGS 3.4 software (Agilent Technologies, Santa Clara, CA). The read counts allocated for each gene and transcript (Ensembl Genes 2016.12.01) were quantified using the transcripts per million (TPM) method (Parrish et al., 2014; Wagner et al., 2012).

2.10. Immunoblotting

Cells or EVs were lysed to extract protein using Mammalian Protein Extract Reagent (Thermo Fisher Scientific), and the protein concentration was measured using Qubit (Invitrogen). For cell lysate, 10 μg of each total protein sample, or 1 μg of EV protein, was separated using SDS‒PAGE on a 4%–15% gradient gel (MiniPROTEAN TGX Gel, GE Healthcare, IL, USA). Following electrophoresis, the proteins were transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore). Then, the membranes were blocked with Blocking One (Nacalai Tesque, Japan) for 30 min at room temperature (RT) and incubated overnight at 4°C with primary antibodies against α‐SMA (#14243, Cell Signaling Technologies, MA, USA), actin (#mab1501, Millipore), Vinculin (#13901S, CST), Fibronectin (#ab2413), CD9 (SHI‐EXO‐M01, CosmoBio), CD63 (SHI‐EXO‐M03, CosmoBio), CD81 (SHI‐EXO‐M03, CosmoBio), Calnexin (#2679, CST). Then, the membranes were incubated for 1 h with the appropriate secondary antibody linked to horseradish peroxidase anti‐mouse IgG (#NA931, GE Healthcare, WI, USA) or anti‐rabbit IgG (#NA934, GE Healthcare). The signal was developed using ImmunoStar LD (Wako), and imaging and subsequent quantification were performed using Fusion Solo software with a Fusion Solo S imaging system (Vilber, France).

For microRNA mimic effect on the signalling pathway analysis, the ProteinSimple kit was used. (#SM‐W004A, Biotechne, MN, USA). Five micrograms of total protein were used to measure the protein expression of mothers against decapentaplegic homolog 2 (Smad2) (#5339, CST), mTOR (#2972, CST), FOXO1 (#2880, CST), PTEN (#9559, CST), Ras (#3965, CST) and Akt (#9272, CST), following the manufacture's protocol. The secondary antibody used was the one recommended by the assay´s protocol (#DM‐001; #DM‐002, Biotechne).

2.11. Animal experiments

Animal experiments in this study were performed in compliance with the Institutional Animal Care and Use Committee and approved by the institutional official of LSI Medicinie Safety Institute Corporation (approval No. 2021‐0376) according to the Guidelines for Animal Studies (LSIM Safety Institute Corporation). Eight‐week‐old male C57BL/6J mice were used for animal experiments. Initially, an osmotic pump (Mini osmotic pump model 2006, Muromachi Kikai Co., Ltd., Japan) containing angiotensin II (Sigma) or PBS (Wako), for the sham group, was injected intracutaneously through a 1‐cm incision made in the skin between the scapula of the mice. The osmotic pump had a flow rate of 0.15 μL/h with a concentration of 36 μg/body/day. On Day 28, intramyocardial administration of 15 μg of EVs was performed. Under anaesthesia, EV or PBS (in the groups PBS and Sham) was administered into the left ventricular muscle using a catheter with a 30 G needle in two different sites. Animals were sacrificed 2 weeks after the treatment. Hearts were harvested and a transverse cut twice, one proximal to the injection site and another on the lower third of the heart (Figure S15). The portions were then fixed in 4% paraformaldehyde.

2.12. Echocardiography

Echocardiography was performed once a week from the week following the implantation of the osmotic pump. Mice were placed under inhalation anaesthesia with isoflurane, and the fur was removed. Then, the probe was applied to the chest, and cardiac function was measured in M‐mode using an ultrasonic diagnostic imaging system (XARIO SSA‐660A, Toshiba Medical System Corporation). Left ventricular ejection fraction [LVEF = (LVIDd3‐LVIDs3)/LVIDd3] and left ventricle diameter shortening rate [LVFS = (LVIDd‐LVIDs) × 100/LVIDd] were analysed.

2.13. Immunocytochemistry for ECM quantification

CFs treated with several EVs in 24‐well plates were fixed with 4% paraformaldehyde. After 10 min of fixation, the cells were washed with PBS, permeabilized with 0.1% Triton X‐100 and blocked in 5% BSA (Sigma) for 30 min. Primary antibodies against α‐SMA (#14245, Cell Signaling Technology, MA, USA), fibronectin (#ab2413, Abcam) and collagen type I (#ab745, Abcam) were applied to the cells and incubated overnight at 4°C. Subsequently, the cells were incubated with secondary antibodies (Alexa Fluor 594 goat anti‐rabbit A‐11012; Alexa Fluor 594 goat anti‐mouse A‐21125, Invitrogen) for 60 min at RT. Finally, VECTASHIELD mounting medium containing 4',6‐diamidino‐2‐phenylindole (DAPI)Optimal (H‐1000, Vector Laboratories, Burlingame, CA, USA) was used to seal the slides. The stained cells were observed using a BZ‐X700 microscope (Keyence, Japan). The stained area percentage was calculated using the image analysis application software for BZ‐X700 (Keyence) and normalized according to the cell nuclei number.

2.14. microRNA mimic transfection

For identification of the functional role of microRNAs, CFs were cultured in 24‐well plates until confluence reached 70%–80%. Cells were transfected with 20 nM of each microRNA mimic according to the transfection protocol of Lipofectamine RNAiMAX (Invitrogen). After 24 h, the medium was changed, and at 48 h, proteins were collected for further experiments.

The microRNA mimics (Table S1) were used along with the negative control. In addition, a mixture of all microRNAs was transfected in two manners: The first one with an equal amount of all microRNA mimics, and the second one relative to the abundance found in Y‐derived EVs.

2.15. Blood pressure

Blood pressure was measured once a week from the week following the implantation of an osmotic pump after echo measurement. Blood pressure was measured using a non‐invasive warming sphygmomanometer (BP‐98A‐L, Softlon Co., Ltd.).

2.16. Histology‐fibrosis staining

For calculation of infarct size, an improved version of Sirius red staining (Sirius red/fast green) was used (Segnani et al., 2015). Hearts previously fixed in 4% paraformaldehyde were rehydrated and stained according to the manufacturer's protocol (#9046, Chondrex, WA, USA). The stained sections were analysed using BZ‐H3M software (Keyence) and normalized to the total area. All cardiac areas from two sections per heart were calculated.

2.17. Immunohistochemistry

Paraffin‐embedded hearts from mouse experiments were rehydrated. Next, antigen unmasking was performed by heating the samples in Immunosaver (NEM, Japan) for 3 min at 120°C. After endogenous peroxidase inactivation, permeabilization, and blocking using Protein Block Serum‐Free (Dako, Dinamarca), primary antibodies diluted in REAL antibody diluent (Dako) were incubated overnight at 4°C. The primary antibodies used were anti‐CD31 (#ab28364, Abcam, England), anti‐Vimentin (1:100, #5741S, CST), anti‐collagen type I (#ab745, Abcam), anti‐collagen type III (#ab747, Abcam), and anti‐αSMA (#14245, CST). Next, samples were incubated for 1 h at RT with ImPRESS IgG‐peroxidase kits (Vector Labs, CA, USA), and colour development was performed using an ImmPACT DAB substrate kit (Vector Laboratories) under light microscopy, followed by counterstaining with Mayer's hematoxylin (Applied Biosystems) for 3 min. The stained cells were observed by BZX‐700 (Keyence). The stained area was calculated with BZ‐H3M software (Keyence).

2.18. Immunofluorescence

Optimal cutting temperature compound (OCT)‐embedded hearts were fixed with paraformaldehyde (PFA) for 10 min, washed with PBS, and permeabilized with 0.1% Triton‐X (Sigma‒Aldrich). Because the anti‐CD9 primary antibody's host was a mouse and the stained tissue origin was also a mouse, the Mouse on Mouse Immunodetection Kit (M.O. M, Vector Laboratories) was used according to the manufacturer's instructions. After incubation with biotinylated anti‐mouse IgG reagent, other cardiac structure‐specific primary antibodies (CD31, αSMA, Col I and Vim) were applied for overnight. Finally, the secondary antibodies Alexa Fluor 594 anti‐streptavidin (#532356, Invitrogen) and Alexa Fluor 488 goat anti‐rabbit IgG (Thermo Fisher Scientific) were incubated at RT for 1 h. Sections were mounted using DAPI/Fluoromount (#0100‐20, CosmoBio). Fluorescence images were taken using an LSM 700 confocal microscope (Zeiss, Oberkochen, Germany).

2.19. Statistical analysis

All studies utilized at least three biological replicates, including mouse experiments. All the data are expressed as the mean ± standard deviation. A two‐tailed Student's t‐test or the Bonferroni method was used to determine the significance of the data. Unless stated, the results were considered statistically significant when the p‐value was < 0.05. In significant values, * denotes p‐value < 0.05; **denotes p‐value < 0.01.

3. RESULTS

3.1. Extended cell culture

A large amount of EVs is needed to use EVs as therapeutic agents. Therefore, an equally large number of cells are needed. However, adult CCs, similar to most primary cells, have a very short in vitro lifespan. Based on previous studies in our laboratory (Katsuda et al., 2017; Kawamata & Ochiya, 2010), several combinations of small molecules were used to expand the primary cell culture. Two different cell lines, PC‐100‐021 and HCM, were selected as representative of two types extensively present in the cardiac were selected: coronary artery smooth muscle cells and cardiomyocytes, respectively. Notably, HCM cells expressed the typical CM marker Troponin T and PC‐100‐021 the smooth muscle cells marker αSMA (Figure S1). Both cell lines were treated with the 16 possible combinations of the four molecules: Y (Y‐27632; Rock inhibitor), A (A‐83‐01; TGF‐β inhibitor), C (CHIR9902; GSK3 inhibitor), and P (PD0325901; mitogen‐activated protein kinase, MEK inhibitor). The molecule Y resulted in the best treatment in both cell lines, which allowed the culture extension of HCM cells for more than 90 days, while nontreated CMs (NT), stopped dividing around Day 60, and PC‐100‐021 lasted a maximum of 60 days, while NT lasted less than 20 days (Figure 1a, Figure S2a). In contrast, P‐containing combinations rapidly induced senescence in cells, visibly changing their morphology at Day 30 (Figure S2b). Other molecules, such as A, C, or YA resulted efficient in only one of the cell lines, being A the only molecule that could expand HCM culture for over 120 days.

FIGURE 1.

Establishment of extended adult‐CM culture. (a) Representative image at Day 0 in the upper part and 60 in HCM cells and at Day 30 in PC‐100‐021 cells under different cell culture treatments in the lower part. Scale: 200 nm (b) CMs treated with Y and A presented higher gene expression of early differentiation markers (GATA4, VCAM) and lower expression of terminal differentiation markers (MYL2, TNNI3) in comparison to NT cells. Senescence marker expression (CDKN1A, CDKN2A) was not induced in Y‐ and A‐treated CMs at the collection point. Representative experiment n = 3. (c) NGS of treated HCM cells followed by pathway signalling analysis showed inhibition of signalling pathways related to cardiovascular diseases. (d) Y‐treated HCMs did not present activated malignancy‐related pathways. (e) Y‐treated HCMs had cardiovascular system development and functional pathways activated. A, A‐treated CM; NT, non‐treated CM; Y, Y‐treated CM.

These initial results pointed out that culturing CMs with Y or A alone resulted in the most suitable cell type due to the extended lifespan. Following CM early differentiation (GATA4, VCAM), terminal differentiation (MYL2, TNNI3) and senescence markers (CDKN1A, CDKN2A) were analysed by qPCR. Both molecules, but especially Y, maintained cell growth without inducing the expression of terminal differentiation or senescence markers (Figure 1b).

Changes in the cell culture, could also induce malignant alterations. Hence, RNA from NT and Y‐treated HCM cells was sequenced, and their differential expression was compared using Ingenuity Pathway Analysis (IPA) software. As shown by NGS and further signalling pathway analysis, cardiovascular disease‐related pathways and malignant induction‐related pathways were inhibited in Y‐treated HCM cells (Figure 1c,d). Furthermore, cardiovascular system development and function‐related pathways were activated (Figure 1e). Similar results were observed in A‐treated HCM cells, as well (Figure S3).

3.2. EVs from CCs

Next, the possible changes in EV secretion and characteristics between the different treatments on CCs (NT, Y, A) after 1–2 months of culture were assessed. Cryo‐transmission electron microscopy (TEM) images showed the typical EV morphology, a bilipid rounded structure, with a size of approximately 100–200 nm (Figures 2a and S4), as further confirmed by NTA (Figure 2b). The typical EV markers CD9, CD63 and CD81 were detected by immunoblotting (Figure 2c) and quantified by ExoScreen, a method that provides more accurate quantitative expression than immunoblotting (Figure 2d) by detecting a double protein, and only when the distance is lower than 200 nm with a luminesce reaction between a donor and an acceptor beads conjugated to two antibodies (Kawamata & Ochiya, 2010). In general, all the CCs secreted EVs with similar size that were taken up by CFs (Figure 2e).

FIGURE 2.

EV characterization; all CMs presented similar EVs. (a) Representative cryo‐EM images showing a standard structure of all types of CM‐derived EVs. Scale: 200 nm. (b) EV median number and size, analysed by NTA. n = 3 (c) Representative immunoblotting image of the common EV markers, CD9, CD63, and CD81, which were enriched in the EV fraction in comparison to whole cell lysate protein samples. n = 3 (d) EV markers shown by ExoScreen, which presents a more accurate quantification. All samples were normalized to the total protein amount. (e) HCM‐derived EV internalization assay by cardiac fibroblasts. Scale: 50 μm. A, A‐treated CM‐derived EVs; NT, non‐treated CM‐derived EVs; Y, Y‐treated CM‐derived EVs.

3.3. CM‐derived EVs effectively reduced fibroblast activation in vitro

Fibrosis is the result of an erroneous activation of fibroblasts (Kong et al., 2014). Thus, for evaluation of the antifibrotic effect of CC‐derived EVs, CFs were activated with TGFβ, the most commonly used cytokine (Ruiz‐Villalba et al., 2020). According to preliminary studies, 24 h with TGFβ culture was sufficient to induce the fibroblast activation marker ACTA2, and the ECM major component, fibronectin (Figure S5). Importantly, since our objective was to reduce the already established fibrosis in the myocardium, for all our experiments, first, fibroblast activation was established, and later on, CC coculture or EV treatment was assessed in the continued presence of TGFβ (Figure S6a).

A significant increase in the αSMA marker after TGFβ addition, followed by a decrease under CC coculture, especially Y‐CMs, was observed. The decrease in fibroblast activation markers in Y‐CCs was of 20%–30% in both cell lines (Figure S6b,c). Notably, the main effector in the coculture system was confirmed to be EVs since almost exact results were found when, instead of coculture, activated fibroblasts were treated with 5 μg/mL EVs (Figure 3a). Both the mRNA and protein levels of αSMA were reduced 30% in comparison with TGFβ‐activated fibroblasts (Figure 3b,c). In addition, not only αSMA but also fibronectin protein levels were quantified. Although fibronectin expression was reduced in both cell lines, significant differences between activated fibroblasts and Y‐EV‐treated fibroblasts were observed only in PC‐100‐021 cells (Figure S6d).

FIGURE 3.

EV reduces fibroblast activation in vitro. (a) Experimental design. When fibroblast confluency reached 70%, the medium was changed, and 10 μg/mL TGFβ was added. Twenty‐four hours later, EVs were added. Two days later, mRNA or protein was collected for further analyses. (b) ACTA2 mRNA levels relative to TGFβ (+) fibroblasts. Y‐EVs reduced the expression of ACTA2 by 20.9% ± 8 in HCM cells and 20.1% ± 11 in PC‐100‐021 cells. Non‐treated CM derived EV treatment only reduced significantly a 12.5% ± 3 in HCM cells and 12.9% ± 11 in PC‐100‐021 cells. HCM: TGFβ (+) vs. NT p‐value = 0.021; TGFβ (+) vs. Y p‐value = 0.042; TGFβ (+) vs. A p‐value = 0.044. PC‐100‐021: TGFβ (‐) vs. TGFβ (+) p‐value = 0.014; TGFβ (+) vs. Y p‐value = 0.048. n = 3 (c) αSMA protein levels relative to TGFβ (+) fibroblasts. Y‐EV treatment reduced the expression of fibroblast activation markers by 25.6% ± 2 in HCM cells and 35% ± 12 in PC‐100‐021 cells. HCM: TGFβ (‐) vs. TGFβ (+) p‐value < 0.001; TGFβ (‐) vs. NT p‐value = 0.031; TGFβ (+) vs. Y p‐value = 0.001. PC‐100‐021: TGFβ (‐) vs. TGFβ (+) p‐value = 0.008; TGFβ (+) vs. NT p‐value = 0.037; TGFβ (+) vs. Y p‐value = 0.004. n = 3. (d and e) Extracellular matrix quantification by the expression of fibronectin and collagen type I. EV treatment reduced ECM similar levels to αSMA expression in HCM cells. Y‐EV reduced 29.3% ± 12 of fibronectin and 31.1% ± 6 of collagen type I. Scale: 200 μm. αSMA: TGFβ (‐) vs. TGFβ (+) p‐value < 0.001; TGFβ (‐) vs. A p‐value < 0.001; TGFβ (+) vs. Y p‐value = 0.018; Y vs. A p‐value = 0.028. Fibronectin: TGFβ (‐) vs. TGFβ (+) p‐value < 0.001; TGFβ (‐) vs. Y p‐value = 0.022; TGFβ (+) vs. Y p‐value < 0.001; TGFβ (+) vs. A p‐value < 0.001. Collagen I: TGFβ (‐) vs. TGFβ (+) p‐value = 0.036; TGFβ (+) vs. Y p‐value = 0.038. n = 3. A, A‐treated CM; NT, non‐treated CM; Y, Y‐treated CM.

During EV purification, the cells were washed with PBS. However, since CCs were initially cultured in the presence of the small molecules Y and A, there is still a small possibility that the medium contains these molecules and that the observable effect is due to these molecules. Thus, the inhibitory effect was tested in a coculture system of CMs with and without Y/A (Figure S7a,b). The coculture system with the addition of Y and A presented much more potent protein inhibition, as expected since Y and A are inhibitors of molecules within the TGFβ pathway (Rock and TGFβ inhibitors, respectively). To further exclude the potential direct influence of Y within the HCM‐EV sample, two additional experiments were conducted. First, the EV sample was subjected to mass spectrometry and further metabolomic analysis, revealing no detection of Y (Table S2). In parallel, EVs were purified using size exclusion chromatography (SEC) columns to eliminate any potential smaller contaminants. The inhibitory effect of EVs on αSMA expression was assessed by comparing EVs isolated via ultracentrifugation (Y‐EV) and those obtained through SEC (s.Y‐EV, signifying SEC‐purified Y‐EV). Samples were normalized, with a total of 2.5 × 10¹²particles/mL. This particle count is equivalent to 5 μg/mL of total EV protein isolated by ultracentrifugation. Notably, no significant distinctions were observed between the two samples (Figure S7c,d). Thus, these findings corroborate the absence of Y and underscore the direct impact of EVs.

Fibroblast activation is a hallmark of fibrosis, but the most damaging phenotype of fibrosis is the result of fibroblast activation and the secretion and deposition of ECM, since it stiffens the heart and reduces its function (Silva et al., 2021). ECM deposition was quantified by immunofluorescence of the two main components in the heart: fibronectin and collagen type I. Figure 3d is a representative result of the effect of Y‐HMC‐derived EV treatment on activated fibroblasts (Figure 3d). The results are again in concordance with αSMA expression. The secretion of ECM was reduced by approximately one third compared to that of TGFβ‐activated fibroblasts, and the levels of collagen type I were almost reduced to basal expression (Figure 3e). Similar results were observed in the cell line PC‐100‐021 as well (Figure S8).

Unfortunately, EVs derived from A‐treated CM had a high variance in the fibroblast activation markers αSMA and fibronectin (Figures 3b,c and S6) and ECM deposition (Figure 3d,e). Additionally, as mentioned before, the small molecule A had an effect on only one of the two cell lines (Figures 1 and S2); thus, we decided to focus exclusively on Y‐treated CCs. In addition, in this study, we initially tested two different myocardial cell lines, CMs and coronary artery smooth muscle cells. So far, the results showed a better outcome of CMs, both in in vitro expansion, differentiation marker, and transcriptome despite the similar fibrosis inhibition function. In order to get closer to a more realistic outcome, further experiments focused solely on Y‐CMs (HCM cells).

3.4. EVs revert the fibroblast activation signalling pathways

Next, we aimed to understand the signalling pathways involved in fibroblast activation (Figure 4a) and the pathways altered during Y‐CM‐derived EV treatment (Figure 4b) by comparing the fibroblast transcriptome of “activated fibroblasts” with “untreated fibroblasts”; and “activated fibroblasts with EV treatment” with “activated fibroblasts”.

FIGURE 4.

Basal fibroblast expression is restored by EVs. (a) The increased and inhibited signalling pathways by fibroblast activation. (b) The increased and inhibited signalling pathways by EV treatment. Highlighted pathways are already described to be related to fibrosis, in orange increased, and in blue inhibited pathways. (c) Protein expression of representative molecules from the most inhibited signalling pathways. EV significantly reduced Smad2 (p‐value = 0.048), mTOR (p‐value = 0.048), FOXO1 (p‐value < 0.01), Ras (p‐value = 0.046), and Akt protein expression (p‐value = 0.041). n = 3. Y, Y‐treated CM‐derived EVs.

During fibroblast activation, multiple fibrosis‐related pathways were altered (Figure S9a). Specifically, the interferon, and interleukin signalling pathways, including NFkB, IL17A, IL1B, TLR4 and MYD88, were inhibited, while the TGFβ signalling pathway was highly increased, represented by the enhancement of TFGB1, TGFB3, TGFBR1, TGFB2 and SMAD2, among others. These same pathways were reversed after EV treatment, suggesting a rescued phenotype (Figures S9a and  4b). TCF7L2, Tgf beta and SMAD4 were inhibited, while IFNL1, IRF7, IFNA2, IFNB1, IFR1 and STAT1 were strongly activated. Despite the increase in several interferon signalling pathways, the overall inflammatory signalling pathways were generally inactivated (Figure S9b). Additionally, to revert the fibroblast activation phenotype, EV treatment triggered the gene expression of most of the Matrix metalloproteinase (MMP) family members (Figure S9c), which are the main effectors of ECM degradation (Lu et al., 2011).

In addition, the effect of EVs on those pathways was corroborated at protein level. Y‐CM‐derived‐EV treatment reduced the protein expression of Smad2, mTOR, FOXO1, Ras and Akt (Figure 4c).

3.5. EVs partially recovered cardiac function in vivo

The following steps included animal experiments in which EVs were injected intracardially to assess their antifibrotic effects. However, due to the reduced size of the mouse heart, the administered volume of EVs is very limited. To overcome this problem, after collecting collected EVs with ultracentrifugation, we concentrated the samples 100 times using a pressure evaporator (Figure S10a). After the concentration, the main size peak remained the same, although secondary smaller peaks appeared, doubling the size of the initial peak (Figure S10b) and suggesting few fusions of EVs. As shown by cryo‐TEM, the EV structure was maintained (Figure S10c), although some fussed EVs were found (Figure S10d). However, internalization by CFs did not appear to vary (Figure S10e). Thus, we proceeded with animal experiments using concentrated EVs.

A mouse model of chronic hypertension caused by the continuous release of angiotensin II was used (Figure S11a). Angiotensin II is a potent profibrotic molecule associated with atherosclerosis, hypertension, cardiac hypertrophy, and heart failure (Kim & Iwao, 2000). Preliminary experiments determined that angiotensin II induced cumulative cardiac fibrosis until the fourth week, arriving at the logarithmic phase. After four weeks of treatment, the myocardium was highly compromised with fibrosis (Figure S11b,c), and the cardiac function was decreased. EVs were then intracardially administered (Figure S11d), and two weeks later, the fibrosis levels were analysed.

Hematoxylin‐eosin staining, followed by αSMA staining, showed apparent hyperplasia in the angiotensin II‐treated hearts (Figure S12a), accompanied by an increase in αSMA‐positive regions (Figure S12b), similar to previous in vitro results. Sirius red staining showed a significant decrease in fibrotic area in the EV treatment group (Figure 5a,b), which was translated into an improvement of cardiac function analysed by cardiac echography (Figure S13a), as represented by an improved ejection fraction (Figure 5c) and fractional shortening (Figure 5d), of 5.4% and 3.9%, respectively. In contrast, the nontreated group (PBS) presented a functional impairment of −3.1% in ejection fraction and −1.7% in fractional shortening (Figure 5c,d). This improvement in cardiac function was condensed in the first week after the EV treatment, but persisted for at least two weeks, which was the length of the experiment (Figure S13b). The collagen deposition was also stained using the antibodies anti‐collagen type I (Figure S14a), which comprises 80% of all collagen types in the heart (Weber, 1989), and Vimentin (Figure S14b), which is expressed in activated fibroblasts. In both cases, the stained areas corresponded to the stained areas by Sirius red, with a decrease in the EV‐treated group compared to the PBS‐treated group.

FIGURE 5.

EV reduces fibrosis, increases angiogenesis, and improves overall function in vivo. (a) Sirius red/fast green staining, indicating collagen areas on mouse heart in purple and general protein in green. Scale: 500 μm. Two areas per cardiac section were magnified. Scale: 200 μm. (b) The fibrotic area was reduced in the EV treatment group by 18.5% in comparison to the angiotensin+PBS group. Sham vs. PBS p‐value < 0.001; sham vs. EV p‐value < 0.042; PBS vs. EV p‐value < 0.043. (c and d) Left ventricle ejection fraction (LVEF) and left ventricle fractional shortening (LVFS) showed improved cardiac function upon EV treatment. LVEF: 4 w sham vs. PBS p‐value = 0.011; 5 w sham vs. PBS p‐value < 0.001; 5 w PBS vs. EV p‐value < 0.001. LVFS: 4 w sham vs. PBS p‐value = 0.02; 5 w sham vs. PBS p‐value = 0.033; PBS vs. EV p‐value = 0.032. (e) Representative image of CD31 staining; arrowheads indicate the blood vessels. Scale: 500 μm. One area was magnified. Scale: 200 μm. (f and g) Microvessel area density (MVD) and number (MV number) were increased in the EV treatment group. MVD was increased 1.17‐fold and MV number 1.53‐fold compared to the PBS group. MVD: sham vs. EV p‐value = 0.022; PBS vs. EV p‐value = 0.04. MV num: sham vs. EV p‐value = 0.002. (h) Blood vessel size distribution showed an increase in blood vessel diameter but also novel small microvessels appearing in the analysed area. (i) Number of microvessels separated into distal (D) and proximal sections (P). Angiogenesis promotion was concentrated near the injection site. Sham D vs. EV P p‐value < 0.001; sham P vs. EV P p‐value < 0.001; PBS D vs. EV P p‐value < 0.001; PBS P vs. EV P p‐value = 0.024; EV D vs. EV P p‐value = 0.001. Sham and EV treatment group were composed by five mice, while PBS group only three due to death during the intracardiac injection. Each heart was cut into three sections and two of them were used as replicates. PBS, phosphate buffered saline.

Angiogenesis, an essential process for cardiac regeneration, was also investigated (Figure 5e). The total area and the number of microvessels were increased in the EV treatment group (Figure 5f,g). In fact, we observed an increase in blood vessels of all sizes (Figure 5h). Interestingly, each heart was cut twice, with one section proximal to the EV injection point and the other one at a distal point (Figure S15). The increase in microvessels was concentrated in the section proximal to the injection site, supporting that, in fact, EVs promoted angiogenesis (Figure 5i).

Altogether, EV treatment ameliorated the cardiac fibrosis effects, improving cardiac function. In particular, the reduction in fibrosis and the increase in angiogenesis appeared to be independent of blood pressure. Blood pressure rapidly increased after osmotic pump insertion and did not vary during the six weeks of the treatment, since angiotensin was continuously secreted. Furthermore, the angiotensin + PBS group and angiotensin + EV group showed no significant difference in blood pressure (Figure S16). Overall, this suggested that the effect of EVs directly targeted the heart.

3.6. EVs were identified within mouse cardiac fibroblasts

To corroborate the causal effect of EVs, we aimed to detect EVs within the cardiac tissue and more specifically within fibroblasts. On this occasion, after the generation of the fibrotic mouse model, the mice were sacrificed 6 h after EV injection, and the heart was collected and frozen rapidly to maintain the structure of EVs (Figure S17a). Although typical EV markers, such as CD9 or CD63, are expressed in the cell membrane of all cell types, making it impossible to distinguish EVs from membrane proteins if the host and EV are from the same species, an anti‐human CD9 antibody was used to identify human derived EVs within mouse tissue (Escola et al., 1998). This method was previously corroborated in vitro using the anti‐human CD9 (h‐CD9) antibody in mouse embryonic fibroblasts. EVs were stained with PKH67 (green colour) and added to the cells for an internalization assay. Later, the cells were fixed and stained red for h‐CD9. Most EVs, shown in green, were found to be co‐stained red. Red colour alone was not found, indicating high specificity of the antibody (Figure S17b).

In vivo, EVs were stained red, while several structures concordant with blood vessels were stained green to identify the cell type (Figure 6a). The hypertension mouse model shows diffuse and perivascular fibrosis. Moreover, it is highly possible that once injected, EVs may infiltrate blood vessels and disperse throughout the myocardium; thus, we focused on vascular structures. Although, it was not possible to quantify EVs due to low levels of found EVs, those could be in smooth muscle cells and fibroblasts (Figure 6b, left and middle). Additionally, the CD31⁺ area showed EVs within endothelial cells (Figure 6b, right). However, EVs were not found within myocytes (Figure S18a). In vitro analysis also showed a two‐fold internalization preference by CF in comparison to CMs (Figure S18b). Hence, an apparent predilection for CM‐derived EVs to be internalized by stromal cells was observed. Furthermore, it was observed that those same EVs were more than fourfold internalized by CFs in comparison with dermal and lung fibroblasts, while no significant difference was found in MSC‐derived EVs (Figures 6c and S19). This finding indicates a specificity and emphasizes the importance of the use of adult CM as the source of EVs instead of other cell types for the treatment of cardiac fibrosis.

FIGURE 6.

EVs were found in stromal cells from the heart. (a) Stained structures surrounding blood vessels (Vim⁺) included in concrete: endothelial cells (CD31⁺), smooth muscle cells (αSMA⁺) and ECM (collagen type I⁺). Scale: 20 μm. (b) Identification of EVs in several cell types in mouse hearts. Arrowheads signpost the EV. The asterisk marks the lumen of a blood vessel. Scale: 20 μm (c) Differential internalization of CM‐EVs and MSC‐EVs, showing specificity for cardiac fibroblasts exclusively in the case of CM‐derived EVs. Four random fields were selected, and all EVs within cells were counted. The EV number was normalized to the cell number. Scale: 20 μm. CM‐EVs: Cardiac vs. Dermal p‐value < 0.001; Cardiac vs. Lung p‐value < 0.001. n = 3. CM, cardiomyocytes; ECM, excessive extracellular matrix; EV, extracellular vesicle.

3.7. CM‐derived EVs are rich in antifibrotic microRNAs

To identify the concrete molecules inside EVs related to fibrosis inhibition, we examined their microRNA expression with small RNA sequencing. Over 2300 microRNAs were found. Given the antifibrotic effect of EVs, in which NT‐EVs presented some antifibrotic effects, but Y‐EVs had even a greater effect, the microRNAs enriched in the NT group and with at least 1.2‐fold higher expression in the Y group were selected. Within the top 50 enriched microRNAs, 12% were already described as CM‐specific microRNAs. Moreover, 54% of the microRNAs have already been described to have antifibrotic functions (Figure S20 and Table S3).

With a stricter criterion (expression level > 8.5; fold change Y‐NT > 1.5) 18 microRNAs were selected (Figure 7a). Through in silico analysis, we found 561 target genes that have been experimentally corroborated. Of those genes, approximately 16% are closely associated with cardiac diseases, including heart failure (Figure 7a), indicating a cardiac‐specific EV cargo while the main inhibited signalling pathway was TGFβ, the predominant pathway in fibrosis (Ruiz‐Villalba et al., 2020) (Figure 7b). The specific cargo of differentiated cells can also be inferred by other comparative studies with undifferentiated cells. Specifically, we selected a study that focused on undifferentiated iPS cells, and their differentiated CMs (iCMs) (Figure S21). Analysing the microRNA enrichment of a predefined gene marker list, based on the literature (Table S4), iCMs presented a more cardiac‐specific cargo. Cardiac differentiation markers, as well as microRNAs related to cardiac function and regeneration, were more enriched in iCMs than in undifferentiated iPSs.

FIGURE 7.

For microRNA content of EVs, the combination was optimal. (a) Selection of Y‐EV microRNAs with expression levels higher than 8.5 and differential expression of at least 1.5 compared with NT‐EVs. Next, their target genes were inferred. (b) EV target genes interact mainly through the TGFβ signalling pathway. Orange boxes indicate pathways related to fibrosis. (c) Selection of concordant microRNA target genes and decreased genes in fibroblasts. A total of 39 interconnected genes were found. (d) Scheme of the interconnected molecules; black surrounded circles indicate microRNA direct target genes. (e) Effect of individual microRNAs on suppressing the fibroblast activating marker αSMA by immunoblotting. Individual microRNA effects were not as good as the combination of all of them together, recapitulating the abundance in EVs (all relative), and indicated in a yellow dash. The equal mix of all microRNAs (all equally) was not significantly different from activated fibroblasts as well. Significantly different microRNAs were compared to TGFβ (+), and p‐values were TGFβ (‐) = 0.002; let‐7a = 0.007; miR‐125b = 0.023; miR‐490 = 0.049; miR‐486 = 0.021; all (relative to EV abundance) = 0.038. n = 3.

To evaluate the affected pathways after EV treatment, the concurring microRNA's target genes and genes decreased in fibroblasts after EV treatment were selected (Figure 7c). Out of the 197 concurring genes, 39 were highly interconnected genes that participated in a decrease in the mitogen‐activated protein kinase (MAPK), mammalian target of rapamycin (mTOR), janus kinase/signal transducers and activator of transcription (JAK/STAT), transforming growth factor beta (TGFβ) and phosphatidylinositol 3‐kinase/protein kinase B (PI3K/Akt) signalling pathways (Figure 7d), which were also downregulated at the protein level by EV treatment (Figure 4c). These signalling pathways are known to decrease cell cycle progression and cell survival (Steelman et al., 2011). The complete table for selected microRNA target genes and the transcriptome of fibroblasts can be found in Tables S5 and S6, respectively. Furthermore, IPA on the 196 concurring genes confirmed not only a decrease in the previously specified pathways, especially the TGFβ signalling pathway (Figure S22a) but also the phenotypical pathways, including a shortening in cell proliferation, reorganization and movement (Figure S22b), features hyperactivated in activated fibroblasts (Frangogiannis, 2021).

Subsequently, the antifibrotic effect of the previously selected microRNAs in silico was tested. Since miR‐142, miR‐143, miR‐126 and miR‐1246′s target genes were not found altered in the fibroblast transcriptome analysis, these microRNAs were removed from further analyses. The αSMA expression of CFs was analyzed after the addition of all microRNA mimics. Four microRNAs significantly reduced αSMA levels: let‐7a, miR‐125b, miR‐490 and miR‐486 (Figure 7e). Remarkably, all those microRNAs have been described to possess antifibrotic effects in previous reports. Importantly, no microRNA presented significantly stronger αSMA inhibition in comparison to the combination of all microRNAs mixed in the same ratio as found in EVs (all relative in Figure 7e).

To further support this evidence, the inhibitory effect of the individual microRNAs was also tested on several proteins from the previous mentioned inhibited signalling pathways. The expression of Smad2, mTOR, FOXO1, PTEN, Ras and Akt was analysed after 20 μM of microRNA mimic addition, finding that only a few microRNAs reduced them (Figure S23). Let‐7i and miR‐221 reduced the mTOR expression; let‐7i and miR‐486 significantly reduced FOXO1 expression; miR‐26b inhibited PTEN and miR‐199a shirked Ras expression. Previously, we showed that EV treatment reduced the expression of all those proteins (Figure 4c). Taken together, these results suggest that the combination of several microRNAs within EVs is indeed more beneficial than the use of individual microRNAs for overall cardiac regeneration.

4. DISCUSSION

EVs have been proposed as a novel cell‐free therapy for the treatment of cardiac diseases due to their proved benefits in intracardially transplanted cells studies. Here, we propose the use of adult CMs as the source of EVs for the treatment of cardiac fibrosis due to their specific cargo. We developed a novel system for adult‐CMs by the use of the Rock inhibitor Y‐27632 that ensures a large collection of EVs without compromising the cellular physiological integrity. EVs derived from CMs reduced fibroblast activation markers and ECM deposition in vitro. Consecutively, when injected intracardially in a cardiac fibrosis mouse model, EVs reduced the fibrotic area and increased angiogenesis, corresponding to improved cardiac function. CM‐EVs contained not only multiple antifibrotic microRNAs but also multiple cardiac‐specific microRNAs that may contribute to overall cardiac recovery.

NGS analysis followed by signalling pathway analysis showed that EV treatment reverted the transcriptome of activated fibroblasts, approaching the basal state transcriptome. EV treatment reduced the hallmarks of fibrosis by impairing the mTOR/Akt/FOXO1/Ras axis pathways. The EV cargo was enriched in many antifibrotic microRNAs. In essence, these microRNAs separately did not present as strong antifibrotic effect as all microRNAs combined in the same ratio as their original abundance in EVs due to the several antifibrotic pathways (Hu et al., 2019; Xiao, 2021; Zhang et al., 2019) targeted by different microRNAs (Figure 7e and S23). In addition, it is assumed that the combination of microRNAs enclosed within EVs, even though apparently not essential for cardiac fibrosis inhibition, could be essential for overall cardiac regeneration. Indeed, among the top 50 EV‐microRNAs, 50% of microRNAs have been previously described to promote cardiac regeneration. Likewise, multiple microRNAs, such as miR‐486 (Li et al., 2021) and miR‐126‐3p (Chen et al., 2017; Chong et al., 2014), promote angiogenesis, an essential process for cardiac regeneration (Singh et al., 2022).

Fibroblast activation is caused by cell proliferative activation, myofibroblast phenotype acquisition and enhanced ECM secretion (Frangogiannis, 2021), and most of these processes were decreased by Y‐EV treatment. Specifically, the most diminished signalling pathways were MAPK, mTOR, JAK/STAT, TGFβ and PI3K/Akt. The TGFβ signalling pathway was the most inhibited pathway and genes in the whole cascade were affected. This included the ligands BMP, Nodal and TGFβ and their receptors, which were targeted by multiple microRNAs: Let‐7i‐5p, miR‐21‐5p, and miR‐26‐5p targeted TGFβ, and miR‐125b‐5p and miR‐21‐5p targeted BMP receptors. Intermediaries in the pathway, such as SMAD and MYC, were also reduced, while the pathway inhibitor extracellular signal‐regulated kinase (ERK) was increased. The TGFβ signalling pathway affects multiple functions and is known to be the main effector in fibrosis by activating the cell cycle, exerting antiapoptotic effects and inducing cytoskeleton reorganization and ECM component secretion (Fine & Goldstein, 1987; Midgley et al., 2013). Unexpectedly, the upstream analysis by NGS in EV‐treated EVs also indicated that several inflammatory pathways were activated (Figure 4). However, the overall inflammatory pathways were inhibited in fibroblasts (Figure S9). Inflammatory inhibition is important since it is known that inflammation promotes fibrosis (Wynn & Ramalingam, 2012). Nonetheless inflammatory pathways must be activated in inflammatory cells such as macrophages, which are in charge of secreting cytokines that will, in turn, activate fibroblasts, favouring fibrosis (Wynn & Ramalingam, 2012). Therefore, the activation of the interleukin pathway in fibroblasts does not directly indicate an inflammatory response. For instance, the expression of IFNB in fibroblasts, which was increased after EV treatment, inhibits αSMA through STAT1 and STAT2 (Bolivar et al., 2021). Finally, the inhibition of ECM secretion and its degradation was also supported by the inhibition of collagen gene expression and the enhanced MMP family member expression (Table S6, Figure S9). ECM components such as collagen genes were found to be the target of let‐7i‐5p. Likewise, FN1 was also targeted by miR‐199a‐3p. In addition, ECM disruption, probably promoted by the increased expression of MMPs, supported the reduction in fibrotic area in in vivo experiments. Therefore, the microRNA assembly allowed fibroblast inactivation and decreased ECM throughout multiple signalling pathways working in synergy. In a general overview, this collective inhibition of microRNAs reduced cell growth, ECM secretion, and cytoskeletal reorganization, the hallmarks of fibrosis (Travers et al., 2017) (Figure S24), explaining the antifibrotic effects of EVs and their strong potential therapeutic usage.

The proposal of EVs as a therapeutic agent for cardiac remodelling is not necessarily new. However, most studies have used undifferentiated cell‐derived EVs, due to the short life span of primary cells, including of adult‐CMs. Currently we know that the whole cardiac microenvironment, which contains multiple cell types, is essential for proper functioning (Brutsaert, 2003; Ivey & Tallquist, 2016; Zhang et al., 2012). Here, we initially tested CMs and smooth muscle cells, but soon discarded the second one, due to the better outcomes of CMs. Differentiated cells also secrete EVs with regenerative microRNAs (Liu et al., 2018; Saha et al., 2020); for instance, Y‐CM‐EVs contained high levels of miR‐21 (Wang et al., 2015). Furthermore, since EV content partially mirrors the content of the cell of origin (Jenjaroenpun et al., 2013), CM‐derived EVs may be more beneficial to restore cardiac‐related functions. Indeed, in other research, when we compared the microRNAs in iCMs with iPS‐derived EVs, the proportion of cardiac‐function and regenerative‐related microRNAs was higher (Liu et al., 2018) (Figure S21). Furthermore, the authors found that CMs differentiated from iPSs generated EVs enriched with microRNAs that modulate cardiac‐specific processes. They even showed that iCM‐EVs had a greater regenerative effect than iPS‐EVs (Liu et al., 2018). Nonetheless, further analyses between iCMs and adult CMs have to be achieved since foetal, neonatal and adult CMs are different (Vujic et al., 2020). There are changes in the expression of ion channel and contractile protein isoforms during development, and thus, the results may not be extrapolated to adult specimens (Mitcheson et al., 1998). With this issue in mind, we utilized CMs from primary cell line, subtracted from adult individuals (see Materials and Methods). Based on our previous studies (Katsuda et al., 2017, 2019; Kawamata & Ochiya, 2010), we tested several small molecules and found that Y extended adult CM cell culture and retarded terminal differentiation and the expression of senescence markers (Figure 1). Furthermore, possible malignant alterations were discarded after NGS analysis followed by signalling pathway analysis. The concrete mechanisms by which Y improves CM cell culture remain unknown, but it is known that Y is a Rock inhibitor, a pathway necessary for CM terminal differentiation (Guo et al., 2018; Lim et al., 2007). In addition, we found that in Y‐CM, the Notch, Wnt and Yap signalling pathways were activated. During infarct healing, Wnt is activated and works along with Yap/Taz and Notch to induce CM proliferation (Blankesteijn, 2020), providing a partial explanation for the improved in vitro lifespan.

The importance of using EVs derived specifically from adult‐CMs is not only due to the EV cargo but also the EV membrane composition. EV membrane composition affects organotropism and internalization. Y‐EVs were 3‐fold more effectively internalized by CF than by dermal and lung fibroblasts (Figure 6). In contrast, MSC‐derived EVs were equally taken up by the three types of fibroblasts, indicating the specificity of CM‐derived EVs. Fibroblasts are highly tissue‐specific, only 12% of genes are shared among different parts of body fibroblasts (derived from the heart, colon and bladder, and skeletal muscle) (Muhl et al., 2020). Indeed CF are transcriptionally closer to CMs than to fibroblasts of unrelated sources (Fu et al., 2013), and thus CF are a specialized myocardial cell type and not a generic mesenchymal cell (Furtado et al., 2016). Since the EV membrane composition is entirely derived from the cell membrane, it is to be expected that two cell types that reside in the same microenvironment, such as CMs and CF, would be more receptive to one another rather than to unrelated cells. Similarly, Lai et al. showed that cardiosphere‐derived EVs showed some cardiac organotropism (Cambier et al., 2018; Lai et al., 2014), although the mechanisms were not further investigated. In the near future, we plan to further examine the membrane composition of Y‐EVs and try to associate them with the cellular receptor and membrane component to understand this interesting topic. In addition, we plan to study the possible use of EVs as a systemic administered therapy due to this incipient specificity.

Despite the benefits of EVs compared to cellular treatments, the use of EVs as a therapeutic agent has not yet been approved by the FDA. EV treatment has been proven to not increase the frequency of arrhythmia (Khan et al., 2015; Sayed et al., 2016) and enables dose flexibility that is not limited by transplanted viability (Rogers et al., 2020). Moreover, EV administration is immune‐tolerable due to the acellular component in mice (Zhu et al., 2017), pigs (Gallet et al., 2017) and humans (Makkar et al., 2020). Furthermore, the low immunogenicity allows nonpersonalized treatment. Ultimately, EVs were proven to be safely administered to humans in a clinical trial (NCT04491240), although their function still has to be proven. To date, no clinical trials have been performed with adult CM‐derived EVs. However, the results of our animal studies are auspicious. The acute MI model, while used extensively to assess bioactivity, cannot distinguish cardioprotective effects from genuine healing (Malliaras et al., 2013). Our experimental design takes this issue into account as we used a mouse model that developed cardiac fibrosis and a posteriori administered the EV treatment. This limits the scope of this study since it is known that there are several types of fibrosis: replacement, interstitial and perivascular; and hypertension mainly causes interstitial and perivascular fibrosis (Khan et al., 2015). Hypertension generally causes cardiac dysfunction without an apparent loss of ejection fraction, and this may be compromised only in late stages (Frangogiannis, 2021). In our scenario, the ejection fraction was reduced and improved by EV treatment, indicating a more critical cardiac damage close to MI damage. Another limitation of our study for the use of Y‐EV as a treatment is the necessity of EV processing. Even though we established a novel culture medium that allows a longer culture of CMs, these cells should be transferred to bioreactor culture since, in general, EV recovery is very low and large quantities of medium are needed (Silva et al., 2021). Thus, CMs should be able to be adapted to novel culture conditions, and this process is challenging in primary culture cells (Bertolino et al., 2022; Silva et al., 2021). This is an essential step that should be addressed to bring this project to fruition.

EVs made a strong impression as an emergent therapeutic effector due to the elegant incipient results in in vivo studies. Among many regenerative treatments, they are currently being studied for cardiovascular diseases. However, most EVs currently utilized originate from nonspecific, undifferentiated cells. Here, we propose an alternative, adult CMs as the source of EVs. CM‐derived EVs exhibit an enriched cardiac‐function microRNA cargo and presented specificity for cardiac fibroblast internalization. Due to the specialized cargo, these EVs showed antifibrotic results in vitro and in vivo, partially recovering cardiac function. While regulatory considerations remain for the use of EVs as a therapeutic product, the path toward more specific cells as the source of EVs, while being a nonpersonalized treatment, holds great promise for revolutionizing cardiovascular therapies and providing hope to those affected by these conditions.

AUTHOR CONTRIBUTIONS

Marta Prieto‐Vila: Conceptualization (equal); data curation (lead); formal analysis (lead); investigation (lead); methodology (lead); project administration (lead); resources (equal); validation (lead); visualization (lead); writing—original draft (lead); writing—review and editing (lead). Yusuke Yoshioka: Methodology (supporting); resources (supporting); supervision (supporting); visualization (supporting). Naoya Kuriyama: Data curation (supporting); formal analysis (supporting). Akihiko Okamura: Data curation (supporting); formal analysis (supporting); investigation (supporting). Yusuke Yamamoto: Investigation (supporting); project administration (supporting); resources (supporting); supervision (supporting); visualization (supporting). Asao Muranaka: Funding acquisition (supporting); validation (supporting); visualization (supporting). Takahiro Ochiya: Conceptualization (equal); funding acquisition (lead); project administration (supporting); supervision (equal).

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Prieto‐Vila, M. , Yoshioka, Y. , Kuriyama, N. , Okamura, A. , Yamamoto, Y. , Muranaka, A. , & Ochiya, T. (2024). Adult cardiomyocytes‐derived EVs for the treatment of cardiac fibrosis. Journal of Extracellular Vesicles, 13, e12461. 10.1002/jev2.12461