Serum-derived extracellular vesicles (EVs) impact on vascular remodeling and prevent muscle damage in acute hind limb ischemia

Serum is an abundant and accessible source of circulating extracellular vesicles (EVs). Serum-EV (sEV) pro-angiogenic capability and mechanisms are herein analyzed using an in vitro assay which predicts sEV angiogenic potential in vivo. Effective sEVs (e-sEVs) also improved vascular remodeling and prevented muscle damage in a mouse model of acute hind limb ischemia. e-sEV angiogenic proteomic and transcriptomic analyses show a positive correlation with matrix-metalloproteinase activation and extracellular matrix organization, cytokine and chemokine signaling pathways, Insulin-like Growth Factor and platelet pathways, and Vascular Endothelial Growth Factor signaling. A discrete gene signature, which highlights differences in e-sEV and ineffective-EV biological activity, was identified using gene ontology (GO) functional analysis. An enrichment of genes associated with the Transforming Growth Factor beta 1 (TGFβ1) signaling cascade is associated with e-sEV administration but not with ineffective-EVs. Chromatin immunoprecipitation analysis on the inhibitor of DNA binding I (ID1) promoter region, and the knock-down of small mother against decapentaplegic (SMAD)1–5 proteins confirmed GO functional analyses. This study demonstrates sEV pro-angiogenic activity, validates a simple, sEV pro-angiogenic assay which predicts their biological activity in vivo, and identifies the TGFβ1 cascade as a relevant mediator. We propose serum as a readily available source of EVs for therapeutic purposes.


Nanoparticle tracking analysis
sEVs were analyzed using the Nanosight LM10 system (Nanosight Ltd., Amesbury, UK). Briefly, sEVs preparations were diluted (1:1000) in sterile 0.9% saline solution and analyzed by NanoSight LM10 equipped with the Nanoparticle Analysis System & NTA 1.4 Analytical Software. The number of total EVs for each patient was obtained by multiplying the value given by the instrument (microparticles/ml) by the dilution for the analysis and by the number of microliters in which sEVs were re-suspended 3 .

sEV angiogenic assay
In preliminary studies, a dose response curve was performed to evaluate the number of sEVs needed to obtain the best biological response in human microvascular endothelial cells (HMEC) 2 . The acronym ECs will be used throughout the study. It was found, using 4 different sEV samples, that 5x10 4 sEVs/target cells was the most effective sEV dose. 5x10 4 sEVs/target cells were therefore used throughout the in vitro study 4 . sEVs from single samples were thus evaluated for their pro-angiogenic activity using BrdU 5 and in vitro tubulogenesis assays 6 . Negative and positive controls were used to evaluate sEVs angiogenic potency (for BrdU assay: negative control was medium w/o FCS; positive control was with 10% FCS; in vitro angiogenesis assay: positive control was 10 ng/ml of VEGF). The following formula was applied: % effect = ( − (0%) (100%)− ) × 100. Values exceeding 50% of VEGF proangiogenic capability (for both assays) were considered as making sEVs efficient.

In vivo angiogenesis assay
Angiogenesis was assessed by measuring the growth of blood vessels from subcutaneous tissue into a solid gel of basement membrane, as previously described 2,7 . First, ECs (1×10 6 cells/injection) were incubated overnight with sEVs (5×10 10 EVs per 1×10 6 of ECs). Male severe combined immunodeficiency (SCID) mice (6 wks old) were then injected subcutaneously with 0.5 ml of ice-cold BD Matrigel Matrix Growth Factor Reduced (BD Biosciences), which had been mixed with pre-stimulated ECs. An equal number of non-stimulated ECs was used as a negative control. The Matrigel plugs were excised 7 days later and fixed in 4% paraformaldehyde for 8 hours. Matrigel-containing paraffin sections (5-8 μm thick) were stained using the trichrome stain method. 2 The vessel lumen area was determined as previously described. 2

Murine model of hind limb ischemia
C57 mice (Charles River Laboratories), age 7 to 8 weeks and weighing 18 to 22 g, were anesthetized via the intramuscular injection of zolazepam 80 mg/Kg. The animals were closely monitored postoperatively and analgesia with Ketorolac (5 mg/Kg) was administered if required. A small skin incision was made under sterile conditions overlying the middle portion of the left hind limb of each mouse. The proximal end of the left femoral artery and the distal portion of the saphenous artery were ligated, dissected and excised. The overlying skin was then closed using a sterilized 6-0 silk suture 8 . In preliminary experiments, different sEV numbers (ranging from 1x10 9 to 1x10 11 ) and e-sEV administration routes were evaluated (not shown). The latter included intravenous (iv) and intramuscular (im) alone or differently combined. We found that the best administration route and e-sEV number, in terms of animal distress and response to treatment, was 2x10 10 sEVs: 1x10 10 administrated immediately after intervention (T0) iv, 0.5x10 10 im on day one (T1) and day two (T2). Animals were sacrificed on day 7 (T7) for histological analysis.

Hind limb blood flow monitoring
After anesthesia, mice were placed on a heating plate at 37 °C for 5 min to minimize temperature variations. Hind limb blood flow was measured using a Laser Doppler Blood Flow (LDBF) analyzer (PeriScan PIM 3 System, Perimed, Stockholm, Sweden), immediately before and after surgery and at days 3 and 7 after surgery. LDBF analysis was performed on hind limbs and feet. Blood flow was reported as changes in the laser frequency using different color pixels. Images were analyzed to quantify blood flow using ROIs (regions of interest) drawn freehand. To avoid data variations that may be caused by ambient light and temperature, hind limb blood flow was expressed as the ratio of left (ischemic) to right (non-ischemic) LDBF 8 .

Evaluation of capillary density and inflammatory cells
Capillary density and inflammatory cell recruitment was quantified within gastrocnemius muscles using immunofluorescence analysis. Muscle samples were embedded in OCT compound (Bio-Optica) and snap-frozen in liquid nitrogen. Tissue slices (5 μm in thickness) were prepared and capillary endothelial cells identified by immunofluorescence using a goat polyclonal antibody against mouse CD31 (Santa Cruz Biotechnology Inc., Santa Cruz, CA); anti-goat Alexa Fluor 488 (Molecular Probe) was used as secondary antibody. Hoechst was added for nuclear staining. Fifteen randomly chosen microscopic fields, from three different sections in each tissue block, were examined for the capillary endothelial cell count. Capillary density was expressed as the number of CD-31-positive features per high power field (HPF) ± SEM (Magnification: x400). Cryosections of the ischemic limbs were stained with rat anti mouse CD14 primary antibody (PharMingen), while anti-rat Alexa Fluor Texas Red (Molecular Probe) was used as secondary antibody. Dapi was added for nuclear staining. Immunofluorescence was performed on six ischemic non-treated hind limbs and on six e-sEV-treated ischemic hind limbs. CD14 positive cells were counted in ten randomly chosen microscopic fields for each sample (magnification: x400) and expressed as mean per high power field (HPF) ± SEM.

Histology
The gastrocnemius muscle, from ischemic and non-ischemic limbs, was removed at day 7 after surgery, immediately fixed with 4% paraformaldehyde (Sigma) for 8 hours and then embedded in paraffin. Tissue slices were stained with hematoxylin and eosin. Slides were examined under light microscopy at ×200 magnification. Images were acquired from all the injured areas of the ischemic limb sections for the total muscle fiber count. Random images from the total, non-ischemic limb section were considered as a control 9 .

Protein array
Proteins from different sEV preparations were extracted using NP40 lysis buffer (150 mM NaCL, 50 mM TRIS-HCl pH 8, 1% NP40) and quantified on a BCA protein assay (Thermo Fisher). The angiogenic protein profile was performed using Quantibody® Human Angiogenesis array 1000 (Ray Biotech) which contains a combination of two non-overlapping arrays to quantitatively measure the expression of 60 human angiogenic proteins. 25 µg of protein were loaded for each sample. Data were subjected to background subtraction and protein concentration was obtained from the different standard curve of the array. Data were expressed as concentration (pg/ml) ±SD. Functional annotation enrichment analysis was performed using Funrich V3 software and the geneMANIA plug in on Cytoscape.

Western Blot analysis
Proteins from different s-EVs and from ECs, treated with either TGFβ1, e-sEV or i-sEV, were extracted using the RIPA buffer (Sigma), supplemented with proteases and phosphatase inhibitors (Sigma) and then quantified using a Bradford assay (Biorad). Thirty µg, for cells, and 10 µg, for EVs, were subjected to SDS-PAGE, transferred into nitrocellulose membranes and underwent immunoblotting with antibodies against anti-p ser 463-465-SMAD1-5 (Cell Signalling), SMAD1/5/9 (abcam), -actin and TGFβ1 (St. Cruz Biotechnology Inc., Santa Cruz, CA). Densitometric analyses, performed on Image Lab software (Biorad), were used to calculate the differences in the fold induction of protein levels, which were normalized to SMAD1/5/9 and -actin 10 . Values are reported as relative amounts.

mRNA-angiogenic microarray profile
Total RNA from 4 e-sEV and 4 i-sEV samples was isolated using the All in One (Norgen, Thorold, ON, Canada) extraction method. RNA concentration was measured using the NanoDrop1000 spectrophotometer. cDNA was synthesized using the RT 2 First Strand kit (SABiosciences) according to manufacturer's instructions. Gene expression profiling, using the Angiogenesis RT 2 Profiler PCR Array (PAHS 024, SA Biosciences), was performed by loading 200 ng of cDNA for each sEV sample. The expression profile of 84 key genes in angiogenesis was analyzed (list of genes available on website: http://www.sabiosciences.com). Quantitative RT-PCR (qRT-PCR) was conducted using the StepOne Plus TM System (AB Applied Biosciences). Relative gene expression was determined using the ΔΔC T method. The angiogenesis RT 2 Profiler PCR Array was also performed on tubule-like structures formed upon sEV administration. ECs were harvested and purified using the RNeasy Mini kit (Qiagen, Germany), and processed using the Angiogenesis RT 2 Profiler PCR Array. Five endogenous control genes; beta-2-microglobulin (B2M), hypoxanthine phosphoribosyltransferase (HPRT1), 60S acidic ribosomal protein P0 (RPLP0), glyceraldehyde-3phosphate dehydrogenase (GAPDH), and b-actin (ACTB), all present on the PCR array, were used for normalization. The fold-change for each treated sample corresponded to 2 -ΔΔCt , relative to the control sample. Changes in the gene expression of treated, with respect to untreated, ECs were reported as a fold increase/decrease ± SD. For e-sEV and i-sEV treatment, up-regulated transcripts with a fold increase ≥ 3 with respect to control samples (untreated ECs) were used for further investigation. Data were further analyzed using Expression Suite and Funrich V3 Software. Functional annotation enrichment analysis was performed using Funrich V3 software and DAVID GO.

Chromatin Immunoprecipitation (ChIP) Assay
A ChIP assay was performed on ECs treated with sEVs, as indicated in the Results section, using Magna ChIP A kit (Millipore), according to the vendor's instructions 4 . Briefly, ECs, treated as indicated, were cross-linked with 1% formaldehyde and quenched before harvest and sonication. The sheared chromatin was immunoprecipitated with either the anti-p ser 463-465-SMAD1-5 antibody or control IgG on protein G Sepharose magnetic beads. The eluted IP were digested with proteinase K, while DNA was extracted and underwent real time PCR with primers specific for ID1 promoter region: sense, 5'-CAGTTTGTCGTCTCCATG-3'; antisense, 5'-TCTGTGTCAGCGTCTGAA-3' and GAPDH: sense, 5'-TGGAAGGACTCATGACCACAGT-3'; antisense, 5'-CATCACGCCACAGTTTCCC-3' as the housekeeping gene control. RQ values were considered using untreated EC samples as internal control. ChIP analysis was also performed in ECs transfected with duplex siRNAs or with siRNA for SMAD1-5.

Statistics
Results are expressed as mean ± SD or ± SEM, unless otherwise reported. Statistical analysis was carried out using 1-way ANOVA, followed by Tukey's post hoc or multiple comparison, Student t tests for 2-group comparison and Newman-Keuls Multiple Comparison Test where appropriate. The cut-off for statistical significance was set at p<0.05 (*p<0.05, **p<0.01, ***p<0.001).      Figure S2. Analysis of TGF1 mRNA and protein content in e-sEVs isolated by floating density gradient. (A) Representative expression of the CD63 exosomal marker of e-sEVs isolated by floating density gradient on OptiPrep as described in Methods and subjected to Western blot analysis for CD63 exosomal marker. Mean of 3 different samples ± SEM (***p<0.001 30% fraction vs 15% and 60% fraction). (B) e--sEVs isolated with ultracentrifugation were used as the control (***p<0.001 30% fraction vs 15% and 60% fraction) (n=3). (C) e-sEVs isolated as above were lysed and analyzed by Western Blot for p<0.001 30% fraction vs 15% and 60% fraction) (n=3).  Supplementary TABLE S5   Table S5. List of genes regulated upon e-sEV and i-sEV treatment. The Table summarizes the list of genes that are up-regulated and down-regulated upon e-sEV and i-sEV treatment.