Coronary artery mechanics induces human saphenous vein remodelling via recruitment of adventitial myofibroblast-like cells mediated by Thrombospondin-1

Rationale: Despite the preferred application of arterial conduits, the greater saphenous vein (SV) remains indispensable for coronary bypass grafting (CABG), especially in multi-vessel coronary artery disease (CAD). The objective of the present work was to address the role of mechanical forces in the activation of maladaptive vein bypass remodeling, a process determining progressive occlusion and recurrence of ischemic heart disease. Methods: We employed a custom bioreactor to mimic the coronary shear and wall mechanics in human SV vascular conduits and reproduce experimentally the biomechanical conditions of coronary grafting and analyzed vein remodeling process by histology, histochemistry and immunofluorescence. We also subjected vein-derived cells to cyclic uniaxial mechanical stimulation in culture, followed by phenotypic and molecular characterization using RNA and proteomic methods. We finally validated our results in vitro and using a model of SV carotid interposition in pigs. Results: Exposure to pulsatile flow determined a remodeling process of the vascular wall involving reduction in media thickness. Smooth muscle cells (SMCs) underwent conversion from contractile to synthetic phenotype. A time-dependent increase in proliferating cells expressing mesenchymal (CD44) and early SMC (SM22α) markers, apparently recruited from the SV adventitia, was observed especially in CABG-stimulated vessels. Mechanically stimulated SMCs underwent transition from contractile to synthetic phenotype. MALDI-TOF-based secretome analysis revealed a consistent release of Thrombospondin-1 (TSP-1), a matricellular protein involved in TGF-β-dependent signaling. TSP-1 had a direct chemotactic effect on SV adventitia resident progenitors (SVPs); this effects was inhibited by blocking TSP-1 receptor CD47. The involvement of TSP-1 in adventitial progenitor cells differentiation and graft intima hyperplasia was finally contextualized in the TGF-β-dependent pathway, and validated in a saphenous vein into carotid interposition pig model. Conclusions: Our results provide the evidence of a matricellular mechanism involved in the human vein arterialization process controlled by alterations in tissue mechanics, and open the way to novel potential strategies to block VGD progression based on targeting cell mechanosensing-related effectors.


Ex vivo SV tissue stimulation
For bioreactor-based ex vivo stimulations, we employed exclusively SV segments obtained from patients undergoing CABG intervention (Table S1). To minimize the damage due to vessel manipulation in the surgery theatre, SVs employed for CABG implantation were harvested using a 'no touch' procedure [1,2], which maintains almost integrally the adventitia. This is a standard procedure adopted by our cardiac surgery Teams to maintain patency in implanted SV grafts and maximize the clinical outcome. In order to maximize the vitality of the SV, the vessels were maintained hydrated in saline solution for the whole duration of the surgery.
After vessels were released from the surgery room, they were stored in DMEM supplemented with 10% FBS, 1% L-Glutamine (L-Glut), 1% P/S (all by Lonza) at 4 °C until ex vivo tissue stimulation into bioreactors. As a quality checking process, they were measured in length and calibre at the two extremities. Indeed a variable calibre and length of the vessels may result into significant variability in the flow/pressure patterns experienced especially in the CABG. In any instance, SVs shorter than 5 cm and with a calibre < 5 mm were excluded from the study. Additionally, to avoid confounding boundary effects, the parts of the SVs considered to analyse the effect of mechanical forces were derived from the central portions (~3 cm in length) while the two extremities that were anchored to inlet/outlet of the stimulation systems by vessel loops were always discarded. SV segments were mounted in an EVCS mimicking coronary hemodynamics (CABG-like stimulation, luminal pressure: 80 -120 mmHg; pulse frequency: 1 Hz; mean flow rate: ~150 mL/min) [3] or venous conditions as controls (VP, continuous luminal pressure: 5 mmHg; flow rate: 5 mL/min) [4,5] and cultured for 7 or 14 days.
DMEM supplemented with 10% FBS, 1% L-Glut, 1% P/S, and containing 3.5% Dextran (450000-650000 kDa, Sigma-Aldrich) was used to mimic blood-like viscosity of 3.2 cP in CABG-like stimulation. Medium (1/2 of the volume) was changed every 4 days. At the end of the stimulation period, SV samples were recovered. Portions of the stimulated SV were in part processed for histology and immunofluorescence and in part for proteomic analyses. Untreated tissue rings of each SV sample were fixed and snap-frozen upon arrival from the surgery theatre and stored to serve as baseline controls. In addition, to validate our venous control condition, we determined intima thickness after conventional passive culture of SV segments, a common model of intimal hyperplasia described before [6][7][8], in which rings are cultured in medium (DMEM supplemented with 10% FBS, 1% L-Glut, 1% P/S) in the absence of mechanical stimuli.

Tissue morphometry and immunohistochemistry
Fixed tissue rings were embedded in paraffin and cut at 5 µm using a rotary microtome (Leica). Lumen perimeter and the distance of the outer part of the media were measured on sections stained with Masson's trichrome (MT) staining. Intima thickness was measured on sections stained with Weigert van Gieson (WvG, both from Bio-Optica Milano, Italy), and media thickness was determined as the difference between the two measurements. Digital images were acquired using a light microscope and dedicated software (AxioVision Bio Software, Carl Zeiss, Germany). For immunohistochemistry, after heat-induced epitope unmasking (citrate buffer, pH 6, 10 min) and quenching with hydrogen peroxide (0.6%, 20 min), nonspecific binding was blocked with bovine serum albumin (BSA 3%, 45 minutes, Sigma-Aldrich) and sections were incubated overnight with a primary antibody against TSP-1 (mouse anti-human A6.1, 2 μg/mL, Invitrogen). Subsequently, sections were incubated with a secondary antibody (rabbit anti-mouse IgG HRP, Invitrogen) for 1 h, after which colour was developed with diaminobenzidine (ImmPACT DAB, DBA, Italy) and nuclei were counterstained with hematoxylin. Apoptosis was determined on sections stained with the Deadend Colorimetric TUNEL System (Promega, Italy) according to the manufacturer's protocol and with a hematoxylin counterstain.

Tissue and Cell Immunofluorescence
Immunofluorescence (IF) staining for different markers was performed on sections after epitope unmasking and blocking with BSA (3%, 1 h). Sections were incubated overnight at 4 ˚C with primary antibodies and subsequently with appropriate AlexaFluor-conjugated secondary antibodies (Invitrogen) for 1 h at RT (Table   S3). Cells positive for the different markers, counted in at least 3 fields per section, were expressed as percentage of total cells in the media. For cellular IF, after fixation with 4% paraformaldehyde (Sigma-Aldrich), cells were permeabilized for 30 min at RT with PBS containing 3% (w/v) BSA and 0.2% (v/v) Triton X-100 (AppliChem), followed by primary antibody incubation at 4 °C overnight. Negative control cells were incubated in a PBS solution containing 3% BSA. AlexaFluor labelled secondary antibodies were employed to detect primary antibodies. Digital images were obtained using an ApoTome fluorescence microscope or LSM-710 confocal scanning microscope (both Carl Zeiss, Germany). All measurements and quantifications were performed using ImageJ (version 1.46r, National Institutes of Health, USA). Analysis of 3 fields per section was found to be a good representation of markers expression in the whole tissue, based on the magnification used (10 X), and the average size of the sections. Four sections representative of four consecutive portions of the stimulated vessels were quantified for each condition and for each marker. Positive cells were counted in image files blinding the type of stimulation to the examiner. Marker + cells were generally scored just when they clearly exhibited a complete staining around (membrane markers) or inside cytoplasm (cytoskeleton) or nuclei (proliferation markers). This staining pattern was visible only in tissue sections stained with the primary/secondary antibody combinations and not in control staining that were included in every set of immunofluorescence labelling.

Isolation and culture of SV-derived cells
SVPs and SMCs for in vitro experiments were isolated from SVs of patients subjected to unilateral saphenectomy (Table S1). SVP isolation was performed as described previously [9]. In brief, the vein was mechanically minced and digested for 4 h at 37 °C with 3.7 mg/mL Liberase 2 (Roche). Remaining aggregates were removed through filtration with 70 μm and 40 μm cell strainer. CD34 POS /CD31 NEG cells were isolated by magnetic bead-assisted cell sorting (MACS, Miltenyi Biotec). Cells were grown in a humidified atmosphere (95% air, 5% CO2) at 37 °C in Endothelial Growth medium (EGM-2) supplemented with 2% Fetal Bovine Serum (FBS) and 1% Penicillin/Streptomycin (P/S, all by Lonza). For SMCs isolation, after removing the external tissues, including the adventitia, veins were cut longitudinally to remove the endothelium by gentle scraping and then finally minced. Tissue fragments were incubated in Dulbecco Modified Eagle's Medium

In silico modelling of strain responses in SV media and adventitia
An in-silico model was implemented with the aim of quantifying the deformation experienced by the cells embedded in different extracellular matrices, namely the adventitial layer or the medial layer. The cell was assumed as a 10 µm-diameter sphere embedded in an ECM volume. The cell and the ECM perfectly adhere at their interface. A 40 µm-edge cubic ECM volume was modelled around the cell, with two additional lateral extension volumes that were used to smooth the edge effects caused by the application of lateral displacements as the mechanical loading condition. The overall volume (ECM plus cell) was meshed with about ~ 230.000 tetrahedral elements. A linear elastic constitutive model was used for the materials representing the ECM and the cell. Young's moduli were set according to the literature: Young's moduli were set according to literature: 1 kPa for cells [10] 2 kPa for the adventitia (which was treated as a soft collagen matrix [11]) and 88 kPa for the media layer [12,13]. Lateral displacements were applied to the lateral surfaces of the extension volumes, in such a way as to simulate a 16% strain for the cubic ECM volume. Such strain value was meant to represent the circumferential strain at which the vessel matrix is subjected when it is loaded with an arterial-like pressure.
It was estimated by means of the Laplace law, considering blood pressure in the range 80-120 mmHg, a vessel diameter of 2.5-3 mm, a media thickness of 0.52 mm (as measured from histological slices of 9 SV samples), and the media Young modulus of 88 kPa, under the assumption that the pressure load is supported entirely by the medial layer.
Static controls were provided by seeding an equal amount of cells into the same FN-coated plates, keeping them under the same atmospheric conditions but without mechanical stimulation.

Western and ELISA analyses
Western blot analyses were performed according to standard procedures. Cells were lysed in a buffer containing 10 mM Tris-Cl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% (v/v) Triton X-100, 1% (w/v) sodium deoxycholate, 0.1% (w/v) sodium dodecyl sulfate and 1% (v/v) protease and phosphatase inhibitor mixture (Sigma-Aldrich). Whole cell lysates were sonicated, centrifuged for 15 min at 14 000 g; cell supernatants were then collected. Proteins were quantified by BCA protein assay kit (Pierce Chemical Co). Cell lysates (30 μg per lane) were diluted in Laemli sample buffer, heated at 95 °C for 5 min, run onto 4-12% gradient SDSpolyacrylamide gels (Invitrogen), and transferred to nitrocellulose membranes. The blots were blocked with Tris Buffered-saline containing 5% (w/v) nonfat dried milk (AppliChem) at RT for 1 h. Overnight incubation at 4 °C with primary antibodies listed in Table S4 was performed to examine individual protein expression.
Membranes were finally incubated with appropriate secondary antibodies for 20 min. Images were taken by LI-COR Odyssey and band densities were quantified using ImageJ software. An enzyme-linked immunosorbent assay (ELISA, #BMS2100, Invitrogen) was performed on conditioned medium of strained vs.
control SMCs according to the manufacturer's instructions to detect the levels of human TSP-1. Calibration curves were prepared using purified standards for the protein assessed and curve fitting was accomplished by regression following the manufacturer's instructions.

mRNA analysis
Total RNA was extracted from cell lines using TRIzol (Invitrogen), quantified with NanoDrop-1000 spectrophotometer (Thermo Fisher Scientific) before integrity assessment with an Agilent 2100 Bioanalyzer (Agilent Technologies). Superscript III (Thermo Fisher Scientific) was used for reverse transcription.
Quantitative real-time PCR analysis were performed with Power SYBR Green PCR Master Mix (Applied Biosystems) in an ABI 7900 Fast thermal cycler to detect TSP-1, TAGLN, TGFβR, Coll1A gene amplification products (primers details in Table S5). The reported expression levels were calculated relative to GAPDH mRNA, used as an internal standard control. The fold change of the genes in strained condition vs. control samples was calculated as 2 -ΔΔCT and the statistical analysis was done on the ΔCT values.

Mass spectrometry of culture supernatants
For the analysis of cell secretome by label-free mass spectrometry, the conditioned media were collected from strained or not-strained SMCs and processed as described [14,15]. Quantitative label-free LC-MSE was performed on a hybrid quadrupole-time of flight mass spectrometer (Synapt-MS, Waters Corporation, Milford, USA) as previously described [16,17]. The proteins were identified and quantified using Progenesis QIP for optical density (550 nm) was measured using Infinite M200 PRO reader (Tecan). To inhibit TSP-1-dependent migration, cells were incubated at 37 °C for 30 min in serum-free EGM2 in presence or absence of function blocking antibody to CD47 (clone B6H12, 20 μg/mL, Invitrogen) and its isotype control (IgG1, 20 μg/mL, Invitrogen) before subjecting them to the transwell assay.

In vivo porcine arterialization SV model
Anaesthesia was induced with intramuscular ketamine administration (0.1 mg/Kg). After endotracheal intubation, anaesthesia was maintained using halothane, the animals ventilating spontaneously throughout. A 12-15 cm of the left long saphenous vein was isolated using a 'no touch' technique [18][19][20][21], the vein divided and stored in iso-osmotic sodium chloride solution (containing 2 IU/mL heparin (CP Pharmaceuticals Ltd, Wrexham, UK) and 50 µg/mL-glyeryl trinitrate (Schwarz Pharma, Bucks, UK; room temperature) until required, to prevent spasm. The animal was heparinized by intravenous administration of 100 IU/Kg of heparin. Both common carotid arteries were exposed via longitudinal neck incisions medial to the sternomastoid muscle. End-to-end interposition grafts were created in both common carotid arteries (using continuous 7-0 Prolene or Surgipro), with reversed 45°-bevelled 3 cm segments of saphenous vein replacing 45°-bevelled 1 cm excised segments of carotid artery. The proximal and distal anastomoses were approximately 4 cm apart. The order of performing the grafts and the site of insertion (left or right) were randomized between procedures and the vein segments. Neck and leg wounds were closed in layers, and the animals given antibiotic cover (ampicillin (200 mg amfipen (MSD Animal Health, UK) in sterile water i.m.) and appropriate levels of analgesia (buprenorphine (Vetergesic (Ceva Animal Health Ltd, UK) 15 μg/Kg i.m.), repeated as necessary). Animals were observed continually during recovery, which always took less than 1 h.
Once recovered, the animals were inspected daily and fed a normal diet and given water ad libitum. Analgesia was continued as required, but no animal required additional antibiotic cover.    that was adopted to assess the mechanical responses of the cells to cyclic elongation as already described by us. Namely, the frequency of the angles (θ) comprised between the major axis of the cell nuclei and an ideal orthogonal direction to the stretching were calculated and represented. From the graphical representation [23], it is evident that mechanically stimulated cells acquired a preferentially orthogonal specific orientation of their nuclei (red bars) demonstrating cytoskeleton rearrangements and mechano-sensitivity. Figure S6. Immunofluorescence staining with antibodies for contractile/secretory phenotype in SV-derived SMCs subjected to mechanical strain for 72 h. In panel A, a double staining with Phalloidin-TRITC (red) and αSMA antibody (green) is shown in control cells and cells subjected to strain. Panel B represents the distribution of the focal adhesion contacts, as detected with Vinculin antibody (green dots connected to the Factin cytoskeleton, red); arrows indicate cells with a lower organization of the F-Actin cytoskeleton and a reduced number of focal contacts. Panel C shows the expression and the intracellular distribution of Vimentin (green fluorescence). In keeping with previous data [24] and biochemical results shown in Figure 6B, the lower polymerization and organization of the αSMA + fibres (A), the reduced number of focal contacts (B) and the higher presence of Vimentin (C), these staining confirm the phenotypical transition between the contractile (static) and the secretory (dynamic) phenotype of SV-SMCs. Figure S7. The upper picture shows the region of interest (ROI) in the vessel (comprising the media and the intima layers) where TSP-1 was quantified (see also Figure 6D). The lower panels show the Image J-based system used to quantify the expression of TSP-1 in the tissue. In order to make possible statistical comparisons (Figure 6D), the area (TSP-1 + area) occupied by black pixels was quantified and expressed as a percentage of the total area. Figure S8. Immunofluorescence staining with antibodies recognizing SM22α (green fluorescence) and TSP-1 (red fluorescence) in SV samples before (T0) and after culture under venous (VP) or coronary (CABG) flow/pressure pattern for 14 days. It is evident that coronary flow mechanics increased expression of TSP-1 in cells expressing (yellow arrows) or not expressing (rose arrows) SM22α in the SV media. White arrows indicate SM22α + in the picture showing the media of VP-treated SV sample. Figure S9. Characterization of human SVPs. Cells were culture amplified as described in methodological section and in reference publications [9,25], exploiting their CD34 + /CD31antigenic repertoire. As shown in the FACS histogram plots, SVPs expanded in culture expressed high levels of mesenchymal markers CD90, CD105 and CD44. NG2, a pericyte marker characterising cells in the SV vasa vasorum was, at least in part, downregulated. Bar graph represents average expression level and the relative SE of each marker in n = 37 SVPs preparations. Figure S10. Effect of TSP-1 and TGF-β treatment on variations of SM22α expression in SVPs exposed to single or combined treatment (See also Figure 7E). * indicate P < 0.05 by one-way ANOVA with Newman-Keuls's comparison test. Bar graph represents mean and SE.    Table S5. PCR oligo sequences of primers employed in Q-RT-PCR analysis