Extracellular Vesicles Enhance the Remodeling of Cell-Free Silk Vascular Scaffolds in Rat Aortae

Abstract

Vascular tissue engineering is aimed at developing regenerative vascular grafts to restore tissue function by bypassing or replacing defective arterial segments with tubular biodegradable scaffolds. Scaffolds are often combined with stem or progenitor cells to prevent acute thrombosis and initiate scaffold remodeling. However, there are limitations to cell-based technologies regarding safety and clinical translation. Extracellular vesicles (EVs) are nanosized particles released by most cell types, including stem and progenitor cells, that serve to transmit protein and RNA cargo to target cells throughout the body. EVs have been shown to replicate the therapeutic effect of their parent cells; therefore, EVs derived from stem or progenitor cells may serve as a more translatable, cell-free, therapeutic base for vascular scaffolds. Our study aims to determine if EV incorporation provides a positive effect on graft patency and remodeling in vivo. We first assessed the effect of human adipose-derived mesenchymal stem cell (hADMSC) EVs on vascular cells using in vitro bioassays. We then developed an EV-functionalized vascular graft by vacuum-seeding EVs into porous silk-based tubular scaffolds. These constructs were implanted as aortic interposition grafts in Lewis rats, and their remodeling capacity was compared to that observed for hADMSC-seeded and blank (non-seeded) controls. The EV group demonstrated improved patency (100%) compared to the hADMSC (56%) and blank controls (82%) following eight weeks in vivo. The EV group also produced significantly more elastin (126.46%) and collagen (44.59%) compared to the blank group, while the hADMSC group failed to produce significantly more elastin (57.64%) or collagen (11.21%) compared to the blank group. Qualitative staining of the explanted neo-tissue revealed improved endothelium formation, increased smooth muscle cell infiltration, and reduced macrophage numbers in the EV group compared to the controls, which aids in explaining this group’s favorable pre-clinical outcomes.

Graphical Abstract

1. INTRODUCTION

Cardiovascular disease is the primary cause of death globally and refers to disorders of the heart and blood vessels,^1,2 often requiring revascularization treatment of the affected arteries to restore blood flow to distal tissues.^3,4 Stent deployment can be used to reopen localized obstructions; however, severe or diffuse obstructions require bypass surgery. Autologous saphenous vein and internal mammary artery are the current gold standard bypass conduits for the small diameter (<6 mm) vessels of the peripheral and coronary arteries, respectively. However, saphenous veins are often unavailable or unsuitable and frequently fail due to intimal hyperplasia,⁵ while the failure of internal mammary artery grafts remains a common issue.⁶ As synthetic grafts are not feasible options for small diameter grafts due to acute thrombosis,^7,8 a different approach is required.

Vascular tissue engineering is aimed at restoring vessel function by bypassing defective sections of the vascular network using a tissue engineered vascular graft (TEVG) that exhibits a regenerative capacity. TEVGs are typically comprised of a tubular biodegradable scaffold that is intended to be resorbed by the host and replaced with functional vascular neo-tissue. The scaffold is often used in combination with a therapeutic base to prevent acute thrombosis and increase the likelihood of positive scaffold remodeling (as reviewed in ref 9). Stem and progenitor cells have been used extensively by our group as a therapeutic base for TEVGs as these cells have been shown to prevent acute thrombosis and initiate scaffold remodeling through paracrine signaling to endogenous cells.^10–15 Although this approach has shown promising results, there are limitations and potential regulatory hurdles to cell-based technologies regarding safety (e.g., mutated cell DNA) and clinical translation (e.g., off-the-shelf storage).¹⁶ Alternative therapeutic bases for vascular scaffolds are therefore required to ensure that biologically active scaffold-based treatments for cardiovascular disease are effectively translated to the clinic.

Extracellular vesicles (EVs) are cell-derived nanoparticles that possess a phospholipid bilayer membrane and serve to transmit protein and RNA cargo to target cells throughout the body.¹⁷ Numerous studies have demonstrated the therapeutic effects of free-EV injections in cardiovascular research. For example, EVs derived from stem cells have been shown to reduce intimal hyperplasia in vein grafting,¹⁸ increase angiogenesis following cerebral artery occlusion,^19,20 and reduce infarct size following myocardial infarction.^21–23 EVs have also been used in tissue engineering for bone,²⁴ adipose,²⁵ and skin regeneration,^26,27 therefore demonstrating the potential of EVs as an alternative therapeutic base to cells in vascular scaffolds.

Recently, two studies have explored the use of EVs in vascular scaffolds.^28,29 Wei et al. and Chen et al. developed EV-based TEVGs using a soak-loading method and observed increased patency and reduced calcification in rat models of arterial interposition grafting. However, there are limitations to the EV–TEVG systems that were developed in these studies. First, neither study quantified the matrix deposition within the scaffold, a key indicator of long-term graft survival.³⁰ Second, in both cases EVs were solely presented on the scaffold’s outer surfaces due to the soak-loading approach employed, meaning that the inner wall (the future tunica media) would be unlikely to support EV-based recruitment of endogenous smooth muscle cells (SMCs). As in vitro EV exposure has a positive effect on SMC recruitment (as shown here), and exposing SMCs to EVs increases matrix production (as in a forthcoming publication using existing techniques³¹), designing an EV-rich scaffold that facilitates SMC infiltration throughout the entire scaffold structure is a promising approach to improving TEVG patency, matrix remodeling, and long-term graft survival.

To this end, we employ a bilayered silk-based scaffold with a porous inner layer that supports endogenous cell infiltration. In contrast to soak-loading, we employ our established vacuum-rotation seeding method that incorporates EVs throughout the wall of the scaffold, thereby maximizing the endogenous cell–EV interactions.³² The EV-seeded scaffolds are here shown to outperform blank and hADMSC-seeded scaffolds in both patency and matrix remodeling. Qualitative examination of the explanted neo-tissue reveals improved endothelium formation, SMC infiltration, and macrophage clearance in the EV group compared to the controls, which aids in explaining this group’s favorable pre-clinical outcomes. Our novel EV scaffold system highlights the clinical potential of cell-free TEVGs for small diameter arterial grafting applications.

2. METHODS

2.1. Human Adipose-Derived Mesenchymal Stem Cell (hADMSC) Culture.

hADMSCs were obtained from RoosterBio, Inc. (RoosterVial-hAD-1M MSC lot no. 00097) and cultured in supplemented growth media (RoosterBio, SU0005, GM) for two passages. The hADMSCs were then frozen in a freezing medium (90% FBS and 10% DMSO). To harvest EVs, hADMSCs were cultured in a non-conditioned medium (nCM) consisting of a protein-free basal medium (RoosterBio SU-006) supplemented with 10% FBS, 1% penstrep, 0.1% fungizone, and 10 μL/L dexamethasone. The FBS (Atlanta Biologicals) was depleted of EVs via ultracentrifugation at 120 000g for 18 h at 4 °C.

A stock of one million hADMSCs were thawed in a 175 cm² flask (Falcon) and cultured for three days in 15 mL of GM. Medium was replaced after 16 h to remove any residual DMSO. After three days of culture, six million cells were passed into two five-layer tower flasks (equivalent to approximately 3500 cells/cm²) and cultured for 18 h in 75 mL of GM. The GM was then removed and the cells were washed in phosphate-buffered saline (Gibco pH 7.4, PBS). Next, 110 mL of nCM was added to each tower, and the cells were cultured for a further 48 h. The medium was collected after 48 h and referred to as conditioned medium (CM) (Figure 1A).

Figure 1.

Characterization of extracellular vesicles (EVs) derived from conditioned media (CM) generated by adipose-derived mesenchymal stem cells (hADMSCs). (A) Schematic of EV isolation that was achieved using differential ultracentrifugation and filtration of hADMSC CM. (B) Morphology of EVs as observed using transmission electron microscopy. The left and right panels depict separate areas of the same EV isolate at 60 000 and 80 000× magnification, respectively. Examples of EVs are highlighted with arrows. (C) Size distribution of particles within the EV isolate as determined using dynamic light scattering in terms of intensity and (D) number of readings. Each curve represents the analysis of a separate EV isolate. (E) Total protein content of the EV isolate before and after lysis. (F) Presence of the EV-enriched tetraspanin CD63 as determined using Western blot; * represents p < 0.05.

2.2. EV Isolation.

The CM was immediately subjected to centrifugation at 250g (20 min at 4 °C) to remove the nonadherent cells, followed by 2500g (20 min at 4 °C) to remove the cell debris. The CM was then filtered through 0.22 μm filters (Millex, Express PES Membrane) to remove the apoptotic bodies, and 34 mL of the processed CM was added to each ultracentrifuge tube (Beckman Coulter, polypropylene 25 × 89 mm, six tubes total). The tubes were inserted in metal cases, hung on a swinging bucket rotor (Beckman Coulter, SW28.1, 117.1 g cases), and centrifuged at 100 000g (70 min at 4 °C) to pellet the EVs using an ultracentrifuge (Beckman Coulter, L8–70M). The tubes were overturned on a sterile drape to drain the remaining supernatant, and residual droplets within the tubes were removed with a vacuum tip. The pelleted EVs were resuspended in 100 μL of 4 °C PBS (100 μL/tube) for 30 min. During this time, the pellet was briefly agitated by hand to assist EV resuspension. For in vitro experiments, the EV isolate was used at the resuspended concentration (600 μL of total EV isolate obtained from 204 mL of CM). For scaffold seeding and in vivo experiments, the EV isolate was diluted to 1 mL in PBS.

2.3. EV Staining.

To stain EVs, 100 μL of Cell Mask Deep Red plasma membrane dye (Invitrogen, C10046, working concentration of 5 μg/mL in PBS) was added to each ultracentrifuge tube immediately following EV resuspension (1:1 dilution of Cell Mask/EV isolate in each tube). The EVs and Cell Mask were allowed to react for 15 min. Then, each tube was combined in a fresh ultracentrifuge tube, and the reaction mixture was diluted to 30 mL with PBS. The stained EVs were then centrifuged at 100 000g (70 min at 4 °C) to remove any unbound dye. Stained EVs were resuspended in 100 μL of PBS and diluted to 1 mL in PBS prior to being seeded in scaffolds. Cell Mask was also added in a 1:1 dilution with PBS in a separate fresh ultracentrifuge tube and processed in the same manner as the stained EVs to act as a “Cell Mask Only” control that contained the dye in the absence of EVs.

2.4. EV Characterization.

2.4.1. Transmission Electron Microscopy.

Transmission electron microscopy was performed using a JEM-1011 (JEOL) to characterize the EV morphology. Briefly, 5 μL of fresh EV isolate was applied to a 3 mm carbon coated grid, and excess liquid was wicked away using filter paper. A 1% uranyl acetate solution (5 μL) was applied to the grid to stain lipids and increase the contrast of the EV membrane relative to the grid. Excess uranyl acetate solution was wicked away using filter paper.

2.4.2. Dynamic Light Scattering.

Dynamic light scattering was performed using a Nano-ZS90 Zetasizer (Malvern Panalytical) to characterize the EV size. Briefly, 80 μL of fresh EV isolate was added to an ultramicro 8.5 mm cuvette (Brandtech Scientific). The cuvette was inserted into the Zetasizer and allowed to equilibrate to 25 °C for 60 s prior to particle measurement at an automatically determined attenuation and duration. EVs were presumed to have the same refractive index and absorption as proteins suspended in water. Data were processed using the protein analysis model included with the Zetasizer software (ver. 7.13). This model allows for the clear identification of adjacent peaks and interpretation of the measurements.

2.4.3. Total Protein Concentration of the EV Isolate.

The total protein content of the EV isolate was quantified using a micro bicinchoninic acid protein assay (Thermo Scientific). Briefly, 50 μL of fresh EV isolate was diluted to 500 μL in either PBS or sodium dodecyl sulfate (SDS, final concentration of 2% in PBS). Samples were then vortexed for 30 s. SDS and vortexing was employed to lyse intact EVs and release encapsulated protein.³³ Next, 150 μL of the vortexed sample was added to the wells of a 96-well plate in triplicate. Then, 150 μL of the working solution was added to each well to initiate the colorimetric reaction. Two standard curves were plated in triplicate for bovine serum albumin suspended in either PBS or 2% SDS at concentrations ranging from 0 to 200 μg/mL. The plate was incubated at 37 °C for 2 h and then cooled to room temperature, and protein concentration was quantified by measuring the absorbance at 562 nm with a microplate reader (BioTek, Synergy HT).

2.4.4. Western Blot.

Western blot was employed to identify CD63, a tetraspanin commonly used to confirm the presence of EVs.¹⁷ Briefly, 30 μL of the hADMSC cell lysate or EV isolate was boiled at 95 °C in combination with 10 μL of reducing sample buffer and separated by SDS/PAGE. The protein was transferred to nitrocellulose (Bio-Rad) using wet Dunn carbonate transfer buffer (10 mM NaHCO₃, 3 mM Na₂CO₃, 20% methanol pH 9.9). The blot was incubated in a blocking solution (5% dry milk and 0.1% Tween-20 in PBS) for 1 h and then incubated with primary rabbit polyclonal CD63 antibody (EXOAB-CD63A-1, System Biosciences, 1:2000) overnight at 4 °C. The blot was then probed with HRP-conjugated secondary antirabbit IgG (System Biosciences, 1:5000) and developed using enhanced chemiluminescence.

2.5. In Vitro Experiments.

2.5.1. Vascular Cell Culture.

SMCs were obtained from ATCC (no. PCS-100-012), cultured in a supplemented basal medium (no. 311 K-500, Cell Applications Inc., SBM), and used between passages 7–9. Human coronary artery endothelial cells (ECs) were obtained from Cell Applications (no. 300–05a), cultured in a supplemented basal medium (no. 212 K-500, Cell Applications Inc., SBM), and used in passage 4. The treatments applied to these cells were PBS (vehicle control) and EVs (50 and 150 μL of EV isolate to examine the dose dependency). The unsupplemented basal medium (BM) and SBM were used as negative and positive controls, respectively.

2.5.2. Proliferation Assays.

Here, 48-well plates were coated with collagen (type I rat tail collagen, Gibco A10483) dissolved in 0.02 M acetic acid. Briefly, 500 μL of a 50 μg/mL collagen solution was added to each well and placed at room temperature in sterile conditions for 1 h. The wells were then washed three times in PBS, and either SMCs or ECs were plated at concentrations of 4000 or 5000 cells per well, respectively, in 1 mL of cell-specific SBM. After 16 h, the SBM was removed, and the wells were washed with PBS. A baseline reading of cellular activity was obtained by adding 300 μL of BM and 30 μL of alamar blue (Invitrogen, DAL1100) to each well for a final concentration of 11:1 and incubating the plate at 37 °C and 5% CO₂ for 4 h. Next, 100 μL of solution was removed from each well and transferred to a 96-well plate, where the absorbances of the product and reactant were read at 570 and 600 nm, respectively, using a plate reader. Readings are presented as 570/600 nm minus blank wells. Each well was washed with PBS, and treatments comprised of 150 μL of BM and 150 μL of treatment were applied to each well in triplicate. Following incubation for 24 h, treatments were removed, and each well was washed with PBS. Cell activity was measured using alamar blue as outlined above, and the cell activity after 24 h of treatment was normalized to the baseline cell activity to account for well-to-well variance. Cell proliferation for each treatment is presented relative to the BM treatment group.

2.5.3. Wound Closure Assays.

SMCs or ECs were plated in 24-well plates at concentrations of 50 000 or 90 000 cells per well, respectively, in 2 mL of cell-specific SBM. After 16 h, 1 mL pipet tips were used to introduce a scratch in each well. The SBM was removed, and treatments comprised of 350 μL of BM and 150 μL of treatment were applied to wells in triplicate. The plate was transferred to a closed stage-top incubator (Tokai Hit Co.) atop the motorized stage of an inverted Nikon TiE fluorescent microscope (Nikon, Inc.) equipped with a 10×, 0.5 NA plan apochromatic lens (Nikon, Inc.) and maintained at 37 °C and 5% CO₂. Wells were imaged every hour for 48 h. The wound area was measured using ImageJ, and wound closure is presented as the final would area relative to the initial wound area. SMC and EC wound closure were calculated at 24 and 48 h, respectively.

2.6. Scaffold Fabrication.

Silk fibroin protein was obtained from mulberry Bombyx mori silk cocoons following previously described protocols.³⁴ Briefly, silk cocoons (from a local farmhouse) were chopped into small pieces, degummed in 0.02 M Na₂CO₃, dissolved in 9.3 M LiBr (Sigma-Aldrich, no. 213225), and dialyzed against distilled water to obtain regenerated silk fibroin protein. The concentration of silk fibroin protein was determined by drying 500 μL of silk solution at 50 °C overnight and weighing the resulting film. The silk protein solution was stored at 4 °C. A 6% silk protein solution was added to tubular molds measuring 2.5 mm in diameter with an inner mandrel measuring 1.5 mm in diameter as previously described.³⁵ The solution was placed at −20 °C for 1 h, after which the top mold cap and mandrel were removed. The solution remained at −20 °C for a further 18 h and was then lyophilized for 24 h. The scaffold was removed from the mold, and water annealed by immersion in 80% ethanol. A 1:1 (v/v) solution of 10% (w/v) silk and 10% (w/v) polycaprolactone (Sigma-Aldrich, no. 440744), both of which were dissolved in hexafluoroisopropanol (Sigma-Aldrich, no. 105228), was electrospun around the scaffold. The scaffold was rotated at 200 rpm with a linear displacement of 50 mm/s under a 15 kV potential. Finally, 200 μL of the solution was applied to each scaffold, and the scaffolds were stored in 80% ethanol at 4 °C. Details regarding the physical characteristics of the bilayered silk scaffold used in this study can be found in our previously published work.³⁵

2.7. Scaffold Seeding.

Scaffolds were washed three times in sterile PBS for 10 min on a shaker plate prior to incorporating either hADMSCs or EVs. The following two forms of EV incorporation were examined: (1) soak-loading and (2) bulk-loading via infusion (seeding). Scaffolds were soak-loaded with EVs by submersing the scaffolds in 1 mL of EV isolate, obtained from a single harvest, and leaving the submersed scaffold on a shaker plate for 2 h. Scaffolds were infused with EVs by first mounting the scaffolds on tubing (Aligent Technologies, PTFE, 0.5 mm inner diameter, 1 mm outer diameter) fed through a 10 μL pipet tip and secured using 4–0 sutures (Sofsilk, S-403). Scaffolds were then mounted in a vacuum chamber and primed with 500 μL of PBS prior to infusion with 3 mL of PBS at 1 mL/min. During infusion, scaffolds were rotated by hand at approximately 15 rpm under a vacuum of −20 kPa. Scaffolds were then primed with 500 μL of EV isolate and infused with 500 μL of EV isolate. The EV isolate that exited the scaffold via transmural flow during infusion was gathered from the chamber, and the infusion was repeated five times. A micro bicinchoninic acid protein assay was performed on the EV isolate before and after seeding (n = 3 scaffolds) to quantify the protein being retained by the scaffold. EV-seeded scaffolds intended for implantation were stored in the EV isolate for 2 h prior to implantation.

Scaffolds were seeded with hADMSC as previously outlined.³² Briefly, scaffolds were primed with 500 μL of cell suspension and then seeded with three million hADMSCs suspended in 3 mL of GM using the rotational vacuum approach. The hADMSC-seeded scaffolds were placed in 5 mL of fresh GM and incubated overnight at 37 °C and 5% CO₂ prior to surgical implantation to allow for cell attachment.

2.8. Scaffold Fluorescent Imaging.

Fluorescent imaging of scaffolds seeded with stained EVs was performed to quantify EV coverage of the scaffolds (n = 3 for each group). Following EV seeding, scaffolds were washed three times in sterile PBS for 10 min on a shaker plate. The scaffolds were then bisected, and one half was infused with distilled water under physiological flow conditions (120 mmHg over 80 mmHg) using a peristatic pump (Biomedicus, 520D extracorporeal blood pump) for 1 h to examine the EV retention. Both halves of the EV-seeded scaffolds were then placed in an optimum cutting temperature medium (Fisher Healthcare, no. 4585) and frozen at −40 °C. The scaffolds were sectioned using a cryotome (Cryotome Electronic, Thermo Shandon) into 15 μm slices and imaged using a fluorescent microscope (Nikon Digital Eclipse 90i). Both the FITC and Cy5 channels were imaged at 200 ms exposure time to visualize the auto-fluorescence of the scaffolds and the stained EVs, respectively. Percentage coverage of the stain was calculated using ImageJ as Cy5 area coverage relative to FITC area coverage.

2.9. Scanning Electron Microscope (SEM) Imaging of Scaffold.

Scaffolds were seeded with unstained EVs, cut in half, and frozen at −80 °C in the EV isolate. Scaffolds were then lyophilized for 24 h and coated with palladium for 60 s using a sputter coater (Cressington Scientific Instruments Ltd.). Scaffolds were imaged with an Apreo field emission SEM (Thermofisher Scientific). Blank scaffolds were infused with PBS, cut in half, and frozen at −80 °C in PBS, followed by lyophilization, palladium coating, and imaging to act as blank controls.

2.10. Rat Abdominal Aortic Interposition Graft Model.

Animal procedures were performed using a protocol approved by the University of Pittsburgh Institutional Animal Care and Use Committee. Scaffolds were surgically implanted as infrarenal abdominal aortic interposition grafts in male Lewis rats as described previously.^{10–13,15,36} Briefly, rats were anesthetized using ketamine (30 mg/kg) and kept under sedation for the duration of the surgery using nose cone administration of isoflurane (1.5 L/min in oxygen). The abdominal wall was incised, and the infrarenal abdominal aorta was isolated from the vena cava using blunt dissection. Microclamps (Vascu-Statt II Plus, Scanlan International) were applied to the aorta prior to bisection with a clamping force of 20g, and a 1 cm scaffold was placed as an aortic interposition graft using interrupted 10–0 Prolene sutures (Ethicon, Somerville, NJ). Animals received a blank scaffold (n = 11), a hADMSC-seeded scaffold (n = 16), or an EV-seeded scaffold (n = 10). The clamps were removed, and patency was verified by observation of the distal pulse pressure. The abdominal opening was closed with 3–0 polyglactin sutures (McKesson, Richmond, VA). Buprenorphine hydrochloride (500 μL per rat) was administered postoperatively every 12 h for the first 72 h, and animals were maintained on anticoagulants (100 mg of dipyridamole and 100 mg of aspirin) for 4 weeks.

After 8 weeks, rats were euthanized using isoflurane and a single intracardiac injection of 10 units of heparin and KCl per kilogram of rat. The left ventricle was catheterized and injected with a Renografin X-ray contrast agent for imaging. An angiography was performed to determine graft patency (GE, OEC 9800 Plus). The graft and adjacent aortic tissue were excised for further analysis. The distal and proximal ends of the explant connected to the aorta were removed with a microtome blade. A 500 μm slice was removed from each end of the explant and frozen at −80 °C for ninhydrin and hydroxyproline analysis. The remaining explant portion was placed in a 4% paraformaldehyde solution (Sigma-Aldrich, no. 158127) prior to being macro-imaged and paraffin embedded for histological analysis.

2.10.1. Explant Staining.

Explants were sliced into 5 μm sections, and immuno-histological staining for cell, collagen, and elastin distribution was performed using Hemotoxylin and Eosin (H&E), Picrosirius-Red, and Verhoeff van Gieson (VVG) staining, respectively. Staining was performed at the Histology Core at the McGowan Institute for Regenerative Medicine. Stained sections were imaged using a Nikon 90i upright microscope. Immuno-fluorescent staining for von Willebrand factor (Abcam ab8822, VWF), α-smooth muscle actin (Sigma no. A5228, αSMA), calponin (Abcam, ab46794), and CD68 (Bio-Rad, MCA341GA) was also performed.

2.10.2. Elastin and Collagen Content Assays.

Explant sections intended for elastin and collagen analysis were thawed and processed for ninhydrin and hydroxyproline assays, respectively, as previously described.^31,37,38 Base hydrolysis was achieved via incubation in 0.1 M NaOH (1 h at 98 °C). The resulting solution was centrifuged at 400g (5 min at 4 °C, repeated twice) to separate the insoluble elastin protein from the soluble non-elastin protein. Acid hydrolysis was achieved via incubation in 6 N HCl (24 h at 110 °C). Assay quantification of both the soluble and insoluble fractions (ninhydrin-based for insoluble elastin and total protein and hydroxyproline-based for collagen) was used to quantify the protein deposition within each explant.

2.11. Statistics.

Statistical analysis was performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA). Data normality was examined using Shapiro–Wilk tests. Significant differences were identified between groups of continuous variables using two-tailed unpaired t tests for normally distributed data and exact two-tailed Mann–Whitney U-tests for non-normally distributed data. Ordinary one-way ANOVA was used to identify significant differences between more than two groups of variables. A p-value of less than 0.05 was considered statistically significant.

3. RESULTS

3.1. EV Characterization.

Electron microscopy revealed nanoparticles in the EV isolate that exhibit the cup-like morphology typical of EVs¹⁷ (Figure 1B). Dynamic light scattering revealed the presence of two particle-size populations within the EV isolate (Figure 1C), with max peaks at 42.67 ± 4.09 and 215.94 ± 34.78 nm, which are indicative of exosomes and microvesicles, respectively.¹⁷ Figure 1D displays the number–diameter distribution of particles within the EV isolate as a percentage of the total number of measurements wherein the two peaks are reduced to a single peak at 38.56 ± 5.78 nm. The single peak in Figure 1D indicates that although the EV isolate contains two populations of particles (exosome- and microvesicle-sized particles) the exosome-sized particles occur far more frequently by number.

Total protein determination analysis (Figure 1E) revealed that the protein concentration of the lysed EV isolate increased relative to the intact EV isolate (170.47 ± 3.58 vs 159.31 ± 5.03 μg/mL, p < 0.05). This supports the preceding electron microscope and dynamic light scattering evidence that EVs are present in the EV isolate and that the EVs are encapsulating proteins. Furthermore, Western blot analysis revealed the presence of the EV-enriched tetraspanin CD63 in the EV isolate, which is commonly used to confirm the presence of EVs¹⁷ (Figure 1F).

3.2. In Vitro Experiments.

3.2.1. Proliferation.

Application of EV isolate to SMCs and ECs increases proliferation relative to vehicle control (PBS) in a dose-dependent manner (Figure 2A). An increase in SMC proliferation was observed between the EV 150 μL group and PBS (4.27 ± 0.25 vs 1.85 ± 0.3, p < 0.0001). An increase in EC proliferation was observed between the EV 50 μL group and PBS (1.34 ± 0.08 vs 0.63 ± 0.13, p < 0.005) and also between the EV 150 μL group and PBS (1.61 ± 0.06 vs 0.63 ± 0.13, p < 0.0001).

Figure 2.

(A) Proliferation and (B) migration of smooth muscle cells (SMCs) and endothelial cells (ECs) when exposed to extracellular vesicle (EV)-based treatments; * represents p < 0.05, ** represents p < 0.005, and *** represents p < 0.0001.

3.2.2. Migration.

Application of EV isolate to SMCs and ECs increases migration relative to vehicle control (PBS) in a dose-dependent manner (Figure 2B). An increase in SMC migration was observed between the EV 50 μL group and PBS (47.28 ± 3.3 vs 28.01 ± 2.27%, p < 0.05) and also between the EV 150 μL group and PBS (57 ± 7.45 vs 28.01 ± 2.27%, p < 0.005). An increase in EC migration was observed between the EV 50 μL group and PBS (32.7 ± 3.05 vs 18.46 ± 2%, p < 0.05) and also between the EV 150 μL group and PBS (44.2 ± 6.07 vs 18.46 ± 2%, p < 0.005). Representative images of the initial and final wound area for each group are shown in Figure S1.

3.3. Scaffold Seeding.

Panels A and B of Figure 3 provide an overview of the EV infusion (seeding) approach performed in this study. Infusing scaffolds with EV isolate results in a reduction of the protein content within the EV isolate (163 ± 13.35 vs 89.85 ± 4.76 μg/mL, p < 0.0001) and indicates that 45.43 ± 4.6% of the total protein content is being retained within the scaffold (Figure 3C). Infusion of the scaffolds with stained EVs significantly increases EV coverage compared to soak-loading (5.64 ± 1.66 vs 81.7 ± 12.7%, p < 0.0001) (Figure 3D). EV coverage was maintained after applying a physiological flow profile to the EV infused group for 1 h (81.7 ± 12.7 vs 69.4 ± 10.78%, p > 0.05). Representative images depicting the EV coverage before and after exposure to flow for each group are displayed in Figure 3E, with green representing the FITC auto-fluorescence of the scaffold (left of the diagonal dashed line) and magenta representing Cy5 staining of the EVs in that same scaffold (right of the diagonal dashed line). The images reveal the localization of EV coverage to the outer and inner surfaces of the scaffold following EV soak-loading and near-complete coverage of the scaffold cross section following EV infusion. Overlaid images of the scaffold auto-fluorescence and EV staining are shown in Figure S2. SEM analysis of scaffolds infused with unstained EVs confirms the presence of nanoparticles within the scaffold that exhibit the size and morphology typically displayed by EVs (Figure 3F, indicated with arrows). The nanoparticles are absent from blank scaffolds examined in the same manner (Figure 3G).

Figure 3.

Development of the EV–TEVG system. (A) Schematic depicting the seeding of silk-based vascular scaffolds with hADMSC-derived EVs. (B) The rotational vacuum device used to seed scaffolds with EVs.³² (C) The total protein content of the EV isolate (intact and lysed) before and after seeding of the scaffolds. (D) Percentage coverage of the scaffold with Cy5 fluorescent dye following seeding of the scaffolds with stained EVs using the different methods. (E) Fluorescent images of blank scaffolds, scaffolds soaked in spun down Cell Mask dye, scaffolds infused with Cell Mask dye, scaffolds soaked in stained EVs, and scaffolds infused with stained EVs. Green represents the FITC auto-fluorescence of the scaffold (left of the diagonal dashed line), and magenta represents Cy5 staining of the EVs in that same scaffold (right of the diagonal dashed line). Dotted lines have been added to the scaffolds with minimal Cy5 staining to indicate the inner and outer boarders. Scale bars depict 500 μm. (F) SEM imaging of scaffolds seeded with unstained EVs vs (G) unseeded blank scaffolds; *** represents p < 0.0001.

3.4. Scaffold Implants.

Figure 4A provides an overview of the implants performed in this study. EV-seeded scaffolds were compared to both blank and hADMSC-seeded scaffolds. Eight weeks after implantation, patency rates were determined using angiography. Panels B and C of Figure 4 show the angiogram and gross imaging of a patent and occluded graft, respectively. Patency rates of the scaffold groups were 100% for EV-seeded scaffolds, 82% for blank scaffolds, and 56% for hADMSC-seeded scaffolds (Figure 4D). Two blank scaffolds occluded due to intimal hyperplasia after eight weeks, with one of the occluded scaffolds also exhibiting aneurysm. Three hADMSC-seeded scaffolds failed due to acute thrombosis within one week, and a further four failed due to intimal hyperplasia at eight weeks. Rats in the hADMSC-seeded group that experienced graft failure due to acute thrombosis also exhibited nephritis of the kidneys, which is often caused by a foreign body response reaction.³⁹ As this response was not observed in the other groups, it suggests that the high graft failure rates observed in the hADMSC-seeded group may be due to an overactive immune response to the presence of human cells. Such an immune response to human cells was not anticipated, as our Lewis rat model of aortic interposition grafting is tolerant of human cells implanted in vascular scaffolds (as reviewed in ref 9).

Figure 4.

Graft patency following eight-week implantation. (A) Schematic representation of scaffold seeding and implantation as an aortic interposition graft in a rat model. (B) Angioplasty and gross imaging of a representative patent explant and (C) angioplasty and gross imaging of a representative occluded explant. Scale bars depict 500 μm. Red arrows indicate the graft location during angioplasty. (D) Patency rates of each group as well as the manner of graft occlusion.

Figure 5 displays the quantitative and qualitative analysis that was performed on graft explants. Panels A and B of Figure 5 display the elastin and collagen present within each graft following explantation, respectively. The EV-seeded group produced significantly more elastin (0.92 ± 0.55 vs 0.41 ± 0.15%, p < 0.005) and collagen (1.48 ± 0.55 vs 1.02 ± 0.34%, p < 0.05) compared to the blank group, which amounts to increases of 126.46 and 44.59% on average, respectively. The hADMSC-seeded group did not produce significantly more elastin (0.64 ± 0.38 vs 0.41 ± 0.15%, p > 0.05) or collagen (1.13 ± 0.32 vs 1.02 ± 0.34%, p > 0.05) compared to the blank group, which amounts to increases of 57.64 and 11.21% on average, respectively.

Figure 5.

Remodeling of implants following eight-week implantation. (A) Elastin and (B) collagen content of each explant as a percentage of the total explant protein. (C) Immuno-histological and immuno-fluorescent images representing explants from each group. Scale bars depict 100 μm; * represents p < 0.05, and ** represents p < 0.005.

Figure 5C displays immuno-histological and immuno-fluorescent staining of the explanted tissue. H&E staining revealed the formation of cellular neo-tissue around the lumen in all groups. VVG staining showed the presence of dense elastic fibers (black color) throughout the thickness of the neo-tissue in the blank and EV-seeded groups; however, this was limited to the internal elastic lamella in the hADMSC-seeded group. Picosirius Red staining showed the presence of a collagenous capsule around the explants in all groups as well as aligned collagen fibers within the luminal neo-tissue. Immuno-fluorescent staining confirmed the presence of an endothelial layer along the neo-tissue lumen (VWF), the presence of SMCs within the neo-tissue (αSMA and calponin), and a number of macrophages within the neo-tissue of each group. Qualitatively, the endothelial staining (VWF) appears to be more prominent in the EV-seeded and blank groups compared to the hADMSC-seeded group, which may aid in explaining the improved patency exhibited by the blank and EV-seeded groups. Furthermore, a lower number of macrophages and a higher number of SMCs appear to be present in the EV-seeded group compared to the hADMSC-seeded and blank groups, which may aid in explaining the improved matrix remodeling exhibited by the EV-seeded group.

4. DISCUSSION

Our study is the first to report a TEVG system that incorporates EVs throughout the thickness of a porous scaffold. Our system is designed to increase endogenous cell–EV interactions in order to fully recruit, and populate the scaffold with, endogenous cells from the adjoining vessel. In vitro exposure to EVs increases the migration and proliferation of ECs and SMCs (Figure 2), cells which must be recruited in order for successful TEVG performance in vivo. Implanting our EV–TEVG system (Figure 3) as an aortic interposition graft in a rat model resulted in improved patency and matrix deposition compared to the blank and parent cell-seeded scaffolds (Figures 4 and 5). Qualitative immuno-fluorescent staining of the explanted neo-tissue revealed improved endothelium formation, increased SMC infiltration, and reduced macrophage numbers in the EV–TEVG system compared to the controls, which points towards the mechanisms by which our system achieves improved vascular graft performance (Figure 5).

Numerous studies have observed and characterized the positive effect that stem cell inclusion has on vascular scaffold patency and remodeling (as reviewed in ref 9). However, the capacity of autologous stem cells to induce positive remodeling of a TEVG decreases when they are sourced from individuals at high cardiovascular risk, thereby limiting the potential of an autologous stem cell–scaffold approach.^12,40 Cell-free approaches that directly incorporate the stem cell-conditioned media into TEVGs have not generated promising pre-clinical data;⁴¹ however, incorporating EVs derived from stem cell-conditioned media into vascular scaffolds presents a potential alternative. Previous in vivo evaluations^28,29 utilized soak-loading of EVs, which can confine EVs to the outer surfaces. Our system builds on these findings by ensuring a uniform distribution of EVs throughout the scaffold (Figure 3), therefore facilitating an increased number of EV–endogenous cell interactions. Furthermore, our study is the first to establish the effect of EV incorporation on matrix deposition in vivo, a critical aspect of neo-vessel formation and long-term graft survival.

There are some key limitations of the present study. First, this study employs differential ultracentrifugation to yield an EV-rich isolate, while alternative methods of EV isolation have been shown to obtain purer EV isolates.^42,43 As there is no “gold standard” method for isolating EVs, the method chosen should reflect the intended purpose of the EVs.⁴⁴ We therefore employ one ultracentrifugation step because it allows for a large volume of CM to be processed in a relatively short time frame. Repeat ultracentrifugation, intended to increase EV isolate purity, was avoided, as this method has been shown to decrease EV yield due to the incomplete sedimentation and aggregation of EVs.⁴⁵ Future studies wishing to refine our system could attempt alternative methods that allow for the isolation of a purer EV fraction.

Second, the approach used to develop and evaluate our novel EV–TEVG system does not employ nanoparticle tracking analysis to estimate EV numbers within the isolate. Nanoparticle tracking would provide a more accurate estimate of EV retention within the scaffold and also allow the EV dose to be standardized by particle number rather than protein content. Future studies should identify the optimal number of EVs required to generate the desired clinical effects by employing nanoparticle tracking analysis to better refine the performance of our EV–TEVG system.

Third, this study is limited to the qualitative analysis of the endogenous cells that infiltrate our grafts using immuno-fluorescent staining. Future studies should characterize the explanted grafts using cell extraction, followed by cell sorting and analysis techniques to obtain quantitative data on the effects that EV incorporation has on the number and type of endogenous cells that repopulate the graft. This would allow for the mechanism underlying the EV-based recruitment of endogenous cells to be determined. Furthermore, this approach would allow for the response of immune cells, and their role in patency and remodelling, to be characterized in a spatiotemporal manner.

Finally, our study does not identify the specific cargo within the EV isolate that is responsible for the improved patency and matrix production generated by our EV–TEVG system. However, previous work has demonstrated that hADMSC-derived EVs are enriched for miRNAs including miR-183, miR-378, miR-140, miR-222, and miR-255, mRNAs including TRPS1, ELK4, KLF7, and NRIP1, as well as 277 proteins including glycoproteins, extracellular matrix proteins, and proteins involved in blood coagulation, the inflammatory response, TGF-β signaling, and angiogenesis.⁴⁶ Furthermore, it has been shown that hADMSC-derived EVs are specifically enriched in cargo that is involved inextracellular matrix remodeling, such as MMP9 and TGFβ−1.^47,48 Our group is conducting ongoing studies to identify the mechanisms by which EVs elicit a matrix synthesizing effect in vascular SMCs as well as how EVs modulate the immune response during the TEVG remodeling process.

With respect to future practice, our novel EV–TEVG system advances the current state-of-the-art systems by facilitating an increased number of endogenous vascular cell–EV interactions within the TEVG. The performance of our novel system demonstrates that EVs can act as a cell-free therapeutic base for TEVGs that not only increases patency rates but also encourages the increased deposition of the matrix proteins essential to functional neo-tissue formation and long-term graft survival. Based on the positive results of this study, we feel that the performance of our novel EV–TEVG system warrants the scaling up of our technology for pre-clinical investigation in a large animal model as the next step.

5. CONCLUSIONS

In this work, we hypothesized that hADMSC-derived EVs would provide a cell-independent positive effect on TEVG remodeling. In vitro, we showed a positive dose-dependent effect of EVs on EC and SMC proliferation and migration. EVs were then uniformly seeded throughout scaffolds and demonstrated superior performance in vivo, including improved patency and matrix deposition compared to the parent cell-seeded and blank scaffolds. The findings of this study support our hypothesis that EVs can serve a bioinstructive role in the fabrication of viable cell-free small diameter vascular grafts.

ABBREVIATIONS

EV

extracellular vesicle

hADMSC

human adipose-derived mesenchymal stem cell

GM

growth medium

FBS

fetal bovine serum

DMSO

dimethyl sulfoxide

PBS

phosphate-buffered saline

CM

conditioned medium

nCM

non-conditioned medium

SBM

supplemented basal medium

BM

basal media (unsupplemented)

SDS

sodium dodecyl sulfate

SMC

smooth muscle cell

EC

endothelial cell

H&E

hematoxylin and eosin

VVG

verhoeff van gieson

VWF

von Willebrand factor

αSMA

α-smooth muscle actin

EV50

50 μL of EV isolate

EV150

150 μL of EV isolate