2.1 Adipose stromal cell isolation and culture

ASC were isolated from subcutaneous adipose tissue obtained by liposuction from healthy human donors as previously described and characterized [12]. The ASC were cultured in Dulbecco's Modified Eagle Medium (DMEM; 1 g/L glucose, Gibco, Life Technologies, USA) supplemented with 10% Foetal Bovine Serum (FBS; Gibco, Life Technologies), 1 mM L-glutamine (Sigma-Aldrich, USA) and 0.1% Bovine Serum Albumin (BSA, A8806-5 g, Sigma, St Louis, United States), 100 U/mL penicillin/streptomycin (Sigma-Aldrich, USA) at 37 °C in a humidified atmosphere with 95% air and 5% CO2. Maximum passage five was used. The ASC from these batches were previously characterized and confirmed by FACS analyses for positive and negative surface markers, including CD45, CD31, CD29, CD90, and CD105 [16]. To compensate for donor heterogeneity, ASC were cultured as pools of five donors, all combined and consistently used in this fashion for all experiments.

2.2 ECs culture

Human umbilical vein endothelial cells (HUVECs) were obtained from the Endothelial Cell Facility (University Medical Center Groningen, Groningen, The Netherlands) and cultured on gelatine–precoated tissue culture flasks (Nunc ™ EasYFlask ™ Thermo Fisher, The Netherlands) in Endothelial Cells Culture Medium (ECM), RPMI (Lonza, Switzerland) supplemented with 20% FBS (Gibco, Life Technologies), 50 µg/mL crude ECGF solution, 2 mmol/L L-Glutamine (Sigma-Aldrich, St. Louis, MO, USA), 5 U/mL Heparin, 100 U/ml Penicillin, 100 µg/mL Streptomycin (Sigma-Aldrich, St. Louis, MO, USA) at 37C°, 5% CO2 HUVEC (max passage 5) were harvested by trypsinization (0.5% trypsin–EDTA, Sigma-Aldrich) at 80% confluence. Only Mycoplasma spp. free cells were used in these experiments.

2.3 Conditioned medium (CM) and EV isolation

Subconfluent ASC in 175 cm2 flasks were washed twice with Dulbecco’s phosphate-buffered saline (DPBS; BioWhittaker®, Walkersville, MD, USA). Cells were incubated in normoxia (N, 21% oxygen) and hypoxia (H, 1% oxygen) for 24 h in a serum-free medium (DMEM, Gibco, Life Technologies, USA). Next, the ASC-conditioned media (ASC-CM) was collected and concentrated 20-fold with 3 kDa MW cutoff Amicon filters (Sigma Aldrich UFC900308). The EVs were isolated by differential centrifugation according to Théry et al. [17] with minor modifications (Fig. 1A). Briefly, the medium was collected and centrifuged at 2,000 xg for 20 min to remove intact detached cells and debris. The supernatant was centrifuged at 10,000 xg for 40 min to remove big-sized EVs and other debris. Subsequently, the supernatant was ultracentrifuged at 110,000 xg for 3 h. The pellet containing the EVs was collected and washed with PBS and centrifuged again at 110,000 xg for 3 h. All centrifugation steps were performed at 4 °C. Ultracentrifugation was performed with a Type 45Ti rotor (K factor:133, #339,160, Beckman Coulter®, USA) with polycarbonate bottles (#355,622, Beckman Coulter®, USA) and a SW55Ti rotor (K factor:48, #342,196, Beckman Coulter®, USA) with tubes (#326,819, Beckman Coulter®, USA), with maximum acceleration and deceleration. The EV samples were aliquoted and stored at − 80 °C until further use. Every batch of CM or EVs is obtained from a batch of ACS cell culture. The supernatants of 15 T75 cell culture bottles, with 3 replicates. All EV batches were obtained from cells in passages 4–5.

Fig. 1

Characterisation of ASCs and ASCs-derived EVs in normoxia and hypoxia. A Workflow for EV isolation and procedures. The ASC-EVs production, isolation, characterisation, and bioassays in ECs and aorta segments. Made in biorender. B Immunofluorescence of ASC cultured in normoxia (N), hypoxia (H), and HeLa cells, stained for Hypoxia Inducible Factor (HIF-1α). HIF-1α was detected with Rabbit HIF-1α, and Alexa Fluor 594 anti-Rabbit nuclei were stained with Dapi. Scale bar 400 µm and 50 µm. C Hypoxia did not increase the accumulation of HIF-1α in ASC. Image quantification HIF-1α fluorescence intensity (FU) in arbitrary units for the stain control. D The expression of HIF-1α and EV-enriched proteins, including CD9 (no detected), CD63, and CD81 in ASC and ASC-derived EVs, was confirmed by western blot. HeLa cells cultured in hypoxia were used as control, * corresponding to EVs derived from HeLa hypoxia cells. E The mean size and F particle concentration of the ASCs-derived EVs were identified by NTA. Hypoxia did not alter G protein and H RNA concentration. Mean ± SD error bars were calculated based on at least three independent experiments (n = 3) performed in triplicates. Multiple comparisons p value < 0.05 (*); ns: not significant. Each point in the graphs represents an individual sample.

2.4 Decellularized extracellular matrix (dECM) hydrogel

The decellularized extracellular matrix of the left ventricle (LV dECM) was produced from healthy porcine hearts purchased from a local slaughterhouse (Kroon Vlees, Groningen, the Netherlands). The left ventricle was cut into small pieces (1-2mm3), then mechanically minced in a commercial kitchen blender (Bourgini 21,300, Breda, The Netherlands) to a homogeneous paste, which was digested with 0.05% trypsin (Thermo Fisher Scientific, Waltham, MA, USA) for 3 h. Afterwards, it was incubated with excess NaCl-saturated solution for 3 h. Next, the crude LV dECM extract was incubated for 24 h with 1% SDS (Sigma-Aldrich). Followed by 24 h incubation with 1% Triton X-100 solution (Sigma-Aldrich). The next day, the material was incubated for 24 h with 1% sodium deoxycholate (Sigma-Aldrich), followed by 24 h incubation with DNase solution (DNase, 30 µg/mL in 1.3 mM MgSO4 and 2 mM CaCl; Roche Diagnostics GmbH Mannheim, Germany). The extract was washed with 96% ethanol for 3 h. All previous steps were performed at 37C° with constant shaking; samples were extensively washed multiple times with sterile PBS between steps. Finally, the LV dECM was freeze-dried (Labconco, Kansas City, MO, USA) overnight and then ground into a powder with an ULTRA-TURRAX (IKA, Staufen, Germany). To produce the hydrogel, the LV dECM was digested with porcine pepsin (2 mg/mL, 3200 I.U., Sigma-Aldrich, St. Louis, MO, USA) in 0.01 M hydrochloric acid (HCl) for 6 h, at room temperature (RT) and constant stirring. Then, the pH was neutralized by adding 1/10th volume of 0.1 M sodium hydroxide (NaOH) and 1/10th volume of 10 × PBS to generate a neutral and isotonic hydrogel. Hydrogel has a final concentration of 10 mg/ml of LV dECM and was stored at 4 °C.

2.5 Nanoparticle tracking analysis (NTA)

The particle size distribution and concentration of EVs were determined by NTA equipped with a 405 nm laser with an LM14 module and the NTA software (Ver. 3.0.) to track each particle on a frame-by-frame basis (NanoSight NTA 3.0, Malvern Instruments, Amesbury, UK) used according to the manufacturer’s recommendations. The EVs were diluted tenfold and 100-fold in double-filtered Dulbecco’s phosphate-buffered saline (DPBS; BioWhittaker®, Walkersville, MD, USA) before analysis, and sole DPBS was used as control. The analysis settings were optimised, and the individual size distribution of five captures (60 s each) per sample was repeated 5 times with a minimum of 50 particles per frame. The data were analysed to obtain the mean, mode, median, and estimated concentration of each particle size per image.

2.6 Protein isolation and quantification

Cells and EVs were lysed in RIPA buffer (Pierce, Thermo Fisher Scientific Inc, Netherlands), supplemented with Protease (Sigma-Aldrich, P8340, 1:100 v/v) and Phosphatase (Thermo Fisher Scientific, 78,420, 1:250 v/v) inhibitor cocktails. The protein concentration of the cell lysates was determined using the BCA Protein assay kit (Thermo Fisher Scientific), while the protein concentration in conditioned medium and isolated EV samples was quantified using the micro-BCA Protein Assay Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The absorbance was read at 562 nm with Benchmark Plus™ microplate spectrophotometer system (Bio-Rad, Hercules, CA, USA). The protein concentration (µg/mL) was determined based on a calibration curve from a dilution series of bovine serum albumin (BSA, Thermo Fisher Scientific). DPBS served as a blank. n = 3 in triplicate.

2.7 Western blot (WB) analysis

A total of 20–40 μg of isolated protein from both cell extract and EVs was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes (Immobilon®-FL, Millipore, Germany) using a Trans-Blot Turbo Transfer System (Bio-Rad). The membranes were blocked for 1 h in 5% non-fat dry milk powder in TBST buffer (0.1% Tween 20 in Tris-PBS buffer). Then, the membranes were incubated overnight at 4 °C with the primary antibody. Primary antibodies used: anti-CD63 (1:1,000; ab68418, Abcam, UK), anti-CD81(1:800; ab109201, Abcam), rabbit anti-CD9 (1:1,000; ab92726, Abcam), mouse anti-GAPDH antibody (2118, Cell Signaling Technology; CST), rabbit anti-β-actin primary antibody (1:1,000, Cell Signaling; 4967L), and rabbit HIF-1α (1:1,000; 737R, Bioss). Next, membranes were washed in Tris-buffered saline (pH 7.4) and incubated for 1 h at RT with horseradish peroxidase (HRP)-conjugated donkey anti-mouse secondary antibody (1:5,000; 6410–05, Southern Biotech) or HRP-conjugated goat anti-rabbit secondary antibody (1:5,000; 4050–05, Southern Biotech). Immobilon® Forte Western HRP Substrate (Sigma-Aldrich) was used to visualise the chemiluminescent protein bands after exposure with ChemiDoc XRS + System (Bio-Rad). ImageJ software (Wayne Rasband NIH-USA v1.51 A, available at http:// rsb.info.nih.gov./ij/) was used to quantify the measured signals.

2.8 EV labelling

Isolated EVs derived from ASC cultured in normoxia and hypoxia were labelled with the lipophilic dye DiI (D3911, ThermoFisher Scientific, Life Technologies, USA) according to the manufacturer’s instructions. Briefly, the EVs were incubated with 1 µM DiI Cell-Labelling Solution at 37 °C for 1 h with gentle agitation and then were washed twice with filtered DPBS and ultracentrifuged at 110,000xg for 3 h at 4 °C to ensure removal of DiI excess. A DiI dye background control sample was prepared following the above procedures but without using any ASC-EVs. The stained EVs were quantified by NTA.

2.9 EV uptake

HUVEC cells were seeded into 12-well plates, 20,000 cells/cm2. After 24 h, a cold medium with DiI-stained ASC-EVs (1 × 103 EVs/cell) or without EVs (control) was added for 1 h at 4 °C. Cells were transferred to 37 °C, and uptake was determined after 0, 3, 6, 12, and 24 h incubation, using flow cytometry and fluorescence microscopy. Flow cytometry cells were washed twice with DPBS, trypsinised, and washed twice in DPBS supplemented with 5% FBS to remove extracellular EV. Flow cytometry was performed on a NovoCyte Quanteon System (Agilent, USA) with default gain control settings. Autofluorescence was compensated for gating the cell population above the 5% threshold. Forward scatter (FSC-HLog) versus red fluorescence (RED-HLog) were recorded and analysed using FlowJo software. For fluorescent microscopy, cells were washed three times with DPBS, fixed with 4% paraformaldehyde (PFA), and incubated with 4',6-Diamidino-2-phenylindole (DAPI, 1 µg/mL, D9564, Sigma) and phalloidin coupled to Alexa Fluor®488 (1:500, Life Technologies) for 1 h at 37 °C. Images were acquired with a Leica fluorescence microscope (DM4000B, Leica Microsystems, Wetzlar, Germany). Image analyses and quantifications were made using the ImageJ software (Wayne Rasband NIH-USA v1.51 A, available at http:// rsb.info.nih.gov./ij/). Data from three independent experiments were used to generate results.

2.10 Immunofluorescence

ASC (normoxic and hypoxic pre-conditioned) and HeLa, as control, were fixed with 4% PFA and incubated for 10 min with 50 mM NH4Cl. After permeabilisation (0.3% Triton-X for 5 min) and blocking (PBS with 5% donkey serum, Agilent Technologies; and 1% BSA, Sanquin) for 1 h, the antigens were probed with a rabbit anti-HIF-1α antibody (1:300; 0737R, Bioss), revealed with an Alexa Fluor 594-coupled donkey-anti rabbit antibody (1:500; A21207, Life Technologies). Nuclei were stained with DAPI (1 µg/mL, D9564, Sigma) and actin with Alexa-Fluor 488-phalloidin. Images were acquired with a Leica fluorescence microscope (DM4000B, Leica Microsystems, Wetzlar, Germany). Micrographs were analysed using ImageJ software (Wayne Rasband NIH-USA v1.51 A, available at http:// rsb.info.nih.gov./ij/. n = 3 in triplicate, a minimum of 3 random fields per replicate were quantified. The mean fluorescence intensity per cell was calculated.

2.11 Aorta ring sprouting assay

Aortas from male Hsd: Sprague Dawley® SD® rats (Envigo, The Netherlands), 250–300 g (8–10 weeks old), were debrided of fibrofatty tissue and cut into rings of ~ 2 mm in width under sterile conditions. The aortic rings were transferred to a 48-well plate and starved in reprogramming media (M199, cat. 11,150,059, Gibco, USA, supplemented with 1% FBS) for 24 h. The following day, the rings were transferred to wells of 48-well plates pre-coated with 100 μL Geltrex™ LDEV-Free growth factor-reduced basement membrane matrix (Gibco, Life Technologies, USA) or the previously described LV dECM. Next, the rings were embedded with an additional 100 μL Geltrex™ or LV dECM and were incubated in basal medium with ASC-EVs (1 × 106 EVs/ring/day) or with EVs-depleted conditioned medium (EDCM) for 10 days at 37 °C under 5% CO2. Rings cultured in the complete and basal medium were used as positive and negative controls, respectively. Media, with EVs or control, were replenished every other day. On days one, five, and ten, live-dead fluorescent staining was performed, and the tissue cultures were imaged as described below. Some aorta samples were lost during handling procedures and treatments.

2.12 Vital staining (calcein-AM/PI)

Aorta ring cultures were washed three times with DPBS and dark incubated with the Live/Dead working solution composed of 5 µM Calcein-AM (Life Technologies®, Eugene, USA), 2 µM propidium iodide (PI; Sigma-Aldrich) and Hoechst nuclear dye (1:50, Thermo Fisher Scientific, Waltman, MA, USA) in serum-free medium at 37 °C, 5% CO2 for 30 min. The cultures were washed with DPBS and imaged with EVOS® M5000 digital inverted microscope (Electron Microscopy Sciences, Hatfield, USA) using the filters: GPF (λex 470/22 nm/λem 525/50 nm), Texas Red (λex 585/29 nm/λem 628/32 nm) and DAPI (λex 357/44 nm/λem 447/60 nm) to visualise Calcein AM, PI, and Hoechst, respectively, at 4 × magnification. Images were analysed using ImageJ (https://imagej.nih.gov/ij/) with the Angiogenesis plug-in [18, 19] or AngioTool v.2 software. All images were transformed to 8-bit, and threshold and contrast were corrected to reduce hydrogel autofluorescence, facilitating particle segmentation and quantification. A background subtraction-based ROI extraction was performed (Suppl Fig. 1). To correct false positive angiogenesis results, day one measurements were labelled as zero and subtracted from the corresponding image measurements on days five and ten. Negative values were removed. This and manual errors resulted in some experimental groups having fewer repeats. Originally n = 6.

2.13 RNA isolation, sequencing, and analysis

Total rna in cells or evs was extracted using a miRNeasy Isolation Kit (Qiagen; Venlo, Netherlands) according to the manufacturer’s instructions. The RNA concentration was determined with a NanoDrop 1000 spectrophotometer. The isolated RNA from ASC-EVS was analysed using an Agilent 2100 Bioanalyzer, an RNA 6000 Nano Kit, and an RNA 6000 ladder (Agilent Technologies; Santa Clara, CA) following the manufacturer’s protocol. The miR profiling of isolated total EVs-RNA was outsourced to Genohub Inc. (Austin, TX, USA) and performed as follows. RealSeq-Dual libraries were prepared using 10 µl of 1 ng/µl RNA input and 25 cycles of PCR. The libraries were pooled and then purified, and the size was selected using a Pippin Prep (Sage Biosciences). The library pool was profiled using a Tapestation and Qubit before sequencing on the NextSeq 550. Sequencing was performed with single 75 bp reads. Raw read sequences in FASTQ files were evaluated for quality control using FastQC (0.11.2) and filtered for low-quality and abundant short reads (less than 15 bp) (Suppl Fig. 2).

The raw fastq files were processed, adapter sequences were removed, and reads were filtered based on length. Reads shorter than 5 bp were filtered first to determine the RNA degradation. The 3’ adapters (with length alignment greater than 25) were trimmed with fastx_clipper (0.0.13.2) using Cutadapt. Reads with a minimum 15 bp length were aligned to a reference. Those filtered reads were mapped to the human reference genome (Homo sapiens, release hg19—UCSC Genome Bioinformatics) using the Bowtie2 mapping algorithm [20]. SAM files were post-processed to the BAM format using SAMtools. A file with read counts corresponding to annotated miRNAs (from miRBase) was generated with the BEDtools package [21]. Read counts were normalised and adjusted to a negative binomial model with an EdgeR-Bioconductor package to identify miRNAs highly enriched in the EVs fraction. We used TarPmiR, a random-forest-based approach trained with 13 features to predict the probability of a miRNA candidate target site. The CluePedia Cytoscape plugin was used to calculate the linear and non-linear statistical dependencies from experimental data. Genes, proteins, and miRNAs linked based on in silico and/or experimental information are integrated into a network with ClueGO terms/pathways. Gene/miRNA enrichments evaluated interrelations within pathways and associations. A pathway-like visualisation was created using the Cerebral plugin layout.

2.14 Statistical analysis

Quantitative data were expressed as mean ± standard deviation (SD) based on at least three independent assays in duplicates or triplicates. EVs size, concentration, protein, and RNA concentrations were analysed using a two-tailed t-test. EVs uptake was analysed using one-way ANOVA and Tukey’s post hoc test. Analyses of tube formation and aortic ring sprouting assays were conducted using a mixed model. The nonparametric test Kruskal–Wallis test was used for multiple comparisons. For miR profiling, a Student’s t-test analysis was performed. All statistical analyses were performed using GraphPad Prism v9.1.0 (GraphPad Company, San Diego, USA). Graphs are presented as median with quartiles, mean ± SD or mean ± s.e.m. All p-values below 0.05 (*), 0.01 (**), and 0.001 (***) were considered statistically significant.