The experimental investigation on human-derived SVPs was approved by the local Ethical Committee at Centro Cardiologico Monzino, IRCCS. Twelve patients recruited for the study were, females or males undergoing surgical removal of saphenous vein because of varicosity. All tissues used in the study are surgical leftovers. Exclusion criteria include concomitant neoplastic, infectious, connective tissue or inflammatory diseases, pregnancy. The Research Ethics Committee approved the study which was performed according to the ethical principles recorded in the 1964 Declaration of Helsinki and later amendments. All the subjects gave written informed consent to participate. The main clinical characteristics of the participants are included in Supplementary Table 1.

Isolation of cells for in vitro experiments was performed as previously described, using two consecutive immunomagnetic selections to obtain CD31^negative/CD34^positive homogenous cell populations from saphenous vein, known as SVPs (21). After saphenectomy, veins were washed in PBS containing Penicillin/Streptomycin 100 U/mL (PBS + P/S). Saphenous vein walls were finely shredded using scissors and digested with Liberase Blendzyme 2 (Roche) diluted 2 mg/ml in Dulbecco Modified Eagle's Medium (DMEM, GIBCO) for 4h at 37°C. The minced tissue was serially filtered through 70 μm, 40 μm, and 30 μm cell strainers using PBS + P/S and then centrifuged at 300xg for 10 min at room temperature. SVPs were isolated as a result of two consecutive magnetic beads–assisted cell sorting (MACS, MiltenyiBiotec) using a MACS MS column, according to manufacturer's instructions. First selection was performed using CD31 magnetic beads (MiltenyiBiotec), in which the negative fraction (CD31^negative) was retained and subjected to a second magnetic sorting performed with CD34 magnetic beads (MiltenyiBiotec). After that, CD31^negative/CD34^positive cell fraction was centrifuged and resuspended in endothelial growth medium EGM-2 (Lonza). SVPs were plated on fibronectin/gelatin-coated plate (0.1% fibronectin - 0.4% gelatin in PBS) 3 × 10³ cells/cm² and cultured at 37°C in a cell culture incubator with 5% CO₂. Culture medium was replaced twice a week and once cells reached 80% confluency, they were trypsinized and split 1:3. Cells used in RNA-seq were between passage 3 and 5.

To validate the identity of isolated SVPs we performed immunocytochemical characterization of the cultured cells for expression of neural/glial antigen 2 (NG2), platelet derived growth factor receptor-β (PDGFRβ), GATA Binding Protein 4 (GATA4), CD31, CD146 and α-smooth muscle actin (α-SMA) (Supplementary Figure 1). SVPs were seeded at 5x10³ cells/cm² on fibronectin/gelatin-coated plate chamber slides, after 24 hours (h) cells were washed with PBS and fixed with 4% PFA in PBS for 15 min at room temperature. When required (GATA4 and α-SMA), the cells were permeabilized with 0.1% Triton X-100 (Sigma-Aldrich) in PBS for 10 min at room temperature. Non-specific staining was blocked with 5% Fetal Bovine Serum (FBS) (GIBCO) in PBS for 30 min at room temperature. Following elimination of excess serum, the cells were exposed to the unconjugated primary antibodies at 4°C for 16 h: NG2 (1:100, Millipore AB-5320), PDGFRβ (1:50, Santa Cruz SC-339), GATA4 (1:100, Abcam ab61767), CD31 (1:100, R&D BBA7), CD146 (1:100, Abcam ab75769), α-SMA (1:200, Dako M0851). After washing in PBS, the appropriate fluorescent secondary antibody (Alexa Fluor) diluted 1:200 in PBS was added to the cells for 60 min at 37°C. Nuclei were stained with DAPI (Thermo Fisher Scientific)1 μg/ml for 10 min at room temperature. Photos of random fields were taken at 20x magnification with Zeiss Observer Z1 inverted microscope.

To investigate the effect of mechanical strain on cultured cells, SVPs were subjected to cyclic strain using the FlexCell Tension Plus FX-5000T system. Before cell seeding, six-well uniaxial Bioflex plates were surface-coated with human fibronectin (10μg/ml) in PBS after covalent crosslinking with a crosslinking reagent (sulfosuccinimidyl 6-(4'-azido-2'-nitrophenylamino) hexanoate; Sulfo-SANPAH) at 0.2 mg/ml in Hepes 50 mM (pH 8.5), photo-activated by exposure to UV-light (365 nm). Cells were subjected to uniaxial cyclic deformation protocol (0–10% deformation, 1 Hz frequency), for 24 and 72 h [according to a protocol established in (19)], while static controls were provided by seeding an equal amount of cells, under the same atmospheric conditions, but without mechanical stimulation (Supplementary Figures 2A,B). For imaging of mechanically stimulated SVPs, cells were fixed after 72 h of uniaxial cyclic straining with 4% PFA in PBS, and thereafter stained with Phalloidin-TRITC (1:500, Sigma) and DAPI 1μg/ml (Thermo Fisher Scientific) at room temperature for 1 h.

For RNA-Seq analysis, total RNA was extracted from 5 different donors using RNeasy Mini kit (Qiagen). Cultured SVPs, approximately 4×10⁵ to 5×10⁵ cells per well, were resuspended in 700 μl of QIAzol Lysis Reagent (Qiagen) and residual DNA was removed by on-column DNase digestion. Total RNA was purified following the manufacturer's instructions and quantified by using NanoDrop-1000 spectrophotometer before integrity assessment with Agilent 2100 Bioanalyzer (RNA Integrity Number values >8).

Next-generation sequencing experiments, including samples quality control and bioinformatics analysis, were performed by Genomix4life S.R.L. (Baronissi, Salerno, Italy). Indexed libraries were prepared from 500 ng/ea purified RNA with TruSeqStranded total RNA Sample Prep Kit (Illumina) according to the manufacturer's instructions. Libraries were quantified using the Agilent 2100 Bioanalyzer (Agilent Technologies) and Qubit fluorometer (Invitrogen Co.), then pooled such that each index-tagged sample was present in equimolar amounts, with a final concentration of the pooled samples of 2 nM. The pooled samples were subject to cluster generation and sequencing using an Illumina HiSeq 2,500 System (Illumina) in a 2 × 100 paired-end format at a final concentration of 8 pmol. The raw sequence files generated (.fastq files) underwent quality control analysis using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and the quality checked reads were trimmed with cutadapt v.1.10 and then aligned to the human genome (hg38 assembly) using STAR v.2.5.2, with standard parameters. Differentially expressed mRNAs were identified using DESeq2 v.1.12. Gene annotation was obtained for all known genes in the human genome, as provided by GenCode (GRCh38.p7 release 25). Using the reads mapped to the genome, we calculated the number of reads mapping to each transcript with HT Seq-count v.0.6.1. These raw read counts were then used as input to DESeq2 for calculation of normalized signal for each transcript in the samples, and differential expression was reported as Fold Change along with associated adjusted p-values (computed according to Benjamini-Hochberg). Raw data of RNA-Seqthat support the findings of this publication have been deposited in NCBI's Gene Expression Omnibus (22) and are accessible through GEO Series accession number GSE192712 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE192712).

Upon inspection of quality control features with FastQC (version 0.11.8) [Andrews (2010). FastQC: a quality control tool for high throughput sequence data. Available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc] and MultiQC (version 1.8), the raw sequencing reads were trimmed to remove adaptor contaminations, using Cutadapt (version 1.18) with the following parameters: cutadapt-a AGATCGGAAGAGCACACGTCTGAACTCCAGTCA -A AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT –trim-n –pair-filter=any –minimum-length 20. Genome contamination screening was performed through FastQ Screen (version v0.14.0). The sequencing reads were, then, aligned to GENCODE's human reference genome (GRCh38 primary assembly v31) and quantified at the gene level using STAR (version2.7.3a), with default parameters and allowing up to 3 mismatches.

Evaluation of sequencing alignment data was performed through Qualimap application (v.2.2.2-dev) (23), using the following analysis types: Multi-sample BAM QC, RNA-seq QC, and Counts QC.

RNA-Seq analysis was performed using DESeq2 package (version 1.22.2) (24) in the R software (version 3.6.3) [R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/]. Samples were normalized for sequencing depth and RNA composition according to DESeq2 median of ratios method. Differential gene expression analysis was performed on genes that passed genefilter/DESeq2 independent filtering procedure (alpha = 0.05). Genes showing a BH-adjusted p-value ≤ 0.05 and log2FoldChange > 1.5 or log2FoldChange < −1.5 were considered differentially expressed (DEGs).

For visualization purposes, counts were transformed using DESeq2 regularized log-transformation. DESeq2 plotPCA function was used for PCA analysis. Samples correlations were computed using the cor.dat function from the stats package (version 3.6.3). Heatmaps were plotted using the pheatmap package (version 1.0.12) [RaivoKolde (2019). pheatmap: Pretty Heatmaps. R package version 1.0.12. https://CRAN.R-project.org/package=pheatmap] and volcano plots with the EnhancedVolcano package (version 1.4.0) [Kevin Blighe, Sharmila Rana and Myles Lewis (2019). EnhancedVolcano: Publication-ready volcano plots with enhanced coloring and labeling. R package version 1.4.0. https://github.com/kevinblighe/EnhancedVolcano] (BH-adjusted p-value cutoff = 0.05, FC cutoff = 1.5).

Motif-based TF prediction was performed in Cytoscape (version 3.8.0) running iRegulon App (25) on the complete set of differentially expressed genes in the dynamic vs. static conditions, with the following parameters: Motif collection = 10 K (9,713 PWMs), Track collection = 1,120 ChIP-seq tracks (ENCODE raw signals), min NEScore = 2, ROC threshold for AUC calculation (% = 3), max FDR = 0.05, Motif rankings database = 20 kb centered around TSS (7 species), Track rankings database = 20 kb centered around TSS (ChIP-seq derived).

DEGs from each contrast and the corresponding logFCs and adjusted p-values were uploaded to Ingenuity Pathways Analysis (IPA) software (version 60467501) (QIAGEN Inc.) (26). Core Analysis was performed using default settings: only protein-protein interaction networks, upstream regulators networks and Diseases and Biological Functions enriched pathways involving AMIGO2 were considered, along with evaluation of the IPA TGF-β signaling canonical pathway. Bubble plots were produced using ggplot2 package (version 3.3.2) [H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2016.] in R.

Pathway enrichment analysis on DEGs at 24 h and 72 h was performed using Metascape (27) (Express Analysis), only pathways involving AMIGO2 gene were considered.

In vivo studies were performed with Large White-Landrace cross pigs (weight 25 to 30 kg) using the saphenous vein-to-carotid artery interposition grafting model previously described (28). This investigation was performed in accordance with the Home Office Guidance on the operation of the Animals (Scientific Procedures) Act 1986 (HMSO, London, UK; PPL numbers 30/2585 and 30/3064) and was compliant with the EU Directive 2010/63/EU and principles stated in the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996).

The animals were subjected to unilateral or bilateral autologous saphenous vein (SV) into common carotid artery bypass grafting. In brief, pigs were anesthetized with ketamine (Ketaset, 100 mg/mL), intubated, and maintained on 1–3% halothane under spontaneous ventilation. The animal was heparinized by intravenous administration of 100 IU/kg of heparin. The long saphenous vein was harvested from the hind leg using the “no touch” technique (29), rinsed in a saline solution containing 2 IU/ml heparin and 50 μg/ml glyceryl trinitrate, and stored in the same solution at room temperature until needed. A 3-cm length of the vein was placed as an interposition graft to the internal carotid artery using continuous 7/0 Surgipro sutures. The flow was re-established and checked using a hand-held Doppler flow meter (Multidoplex II (model MD2), Huntleigh Diagnostics Ltd, Cardiff, UK), and the animals were given antibiotic (ampicillin) and analgesic (buprenorphine) before and during recovery. The pigs were anesthetized and the SV grafts were collected at the established time points (1, 3, 7, 14, and 90 days), followed by euthanasia using an intra-cardiac overdose of pentobarbital. The grafts were fixed in 4% paraformaldehyde for 16 h and then embedded in paraffin for immunohistochemical analyses. Paraffin-embedded tissue sections were dehydrated and after blocking with 6% BSA for 1 h, incubated for 16 h at 4°C with the primary antibodies: AMIGO2 (1:100, Abcam ab84416) and α-SMA (1:150, Dako M0851) to enable the identification between the medial and adventitial layer. Subsequently, sections were incubated with appropriate secondary antibodies diluted 1:200 for1 h at room temperature. Nuclei were stained with DAPI 1μg/ml for 10 min at room temperature. Digital images were obtained using an ApoTome fluorescence microscope or LSM-710 confocal scanning microscope (both Carl Zeiss, Germany). Cells positive for AMIGO2, counted in 3 fields per section, were expressed as a percentage of cells in the whole adventitia. Measurements and quantifications were performed using ImageJ (version 1.46r, National Institutes of Health, USA). Differences between time points were analyzed in GraphPad Prism 5 using ANOVA with Newman-Keuls post hoc test, with a significance level of 0.05.

SV segments for ex vivo culture were supplied from the Department of Cardiovascular Surgery at Centro Cardiologico Monzino. The veins were obtained from patients undergoing coronary artery bypass operations under protocols approved by the Ethical Committees of the Centro Cardiologico Monzino (Italy). Mechanical stimulation of SVs was performed using a custom-made bioreactor tailored to reproduce the coronary mechanics. SV were harvested with a “no-touch” technique (7) and stored at 4°C in DMEM supplemented with 10% FBS, 1% L-Glutamine, and 1% P/S. The ex vivo culture system exploited for veins stimulation was designed by Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano and Unità di Ingegneria Tissutale of Centro Cardiologico Monzino-IRCCS in Milan (30). This culture system allowed mimicking the arterial-like stimulation with a circumferential strain applied to the SV wall typical of the coronary circulation (31). Briefly, the arterial-like flow is accomplished in 4 independent phases: i) a loading step, ii) a pulsatile stimulation step, iii) an unloading step, and iv) a recirculation step. In the first phase, the vessel is filled with DMEM, 10% FBS, 1% L-Glutamine, and 1% P/S. During the pulsatile step, the medium is put under oscillating pressure between 80 mmHg and 120 mmHg and in the third phase the medium flows out of the vessels. The fourth phase is necessary to replace the medium inside the vein, thus maintaining stable nutrients and oxygen supply (17). The culture system was placed in a standard incubator at 37°C in a 5% CO 2 atmosphere and for a culture period of 14 days. Then, SV segments were un-mounted from the culture system, fixed in formalin 37% for 16 h, and then embedded in paraffin for immunohistochemical analyses. In brief, tissue sections were subjected to heat-induced antigen retrieval with 10 mM Sodium Citrate buffer pH 6.0 at 94°C for 30 min. Non-specific binding was blocked with 20% goat serum (Sigma-Aldrich) for 30 min at room temperature. Following the elimination of excess serum, the sections were exposed to the unconjugated primary antibodies at 4°C for 16 h:CD34 (1:50, R&D AF7227), AMIGO2 (1:200, Abcam ab84416), CD31 (1:50, R&D BBA7). After washing in PBS, the appropriate fluorescent secondary antibody (Alexa Fluor) diluted 1:200 in PBS was added to the cells for 60 min at 37°C. Nuclei were stained with DAPI (Thermo Fisher Scientific)1 μg/ml for 10 min at room temperature. Photos were taken at 20x magnification with Zeiss Observer Z1 inverted microscope and reconstructed using Photoshop (Adobe) to obtain the adventitia-lumen image.

To assess the effect of mechanical stress on SVP transcriptome, we subjected cells to cyclic uniaxial strain (defined as dynamic, dyn, from now on) for 24 and 72 h compared to standard culture (static, stat) and performed RNA-Seq. This in vitro system allowed us to identify the early alterations in the transcription, underscoring the mechanisms associated. First, we conducted an unsupervised clustering of the results based on the top 100 most variables expressed genes in the different culture conditions. This analysis and principal component analysis (PCA) showed no stretch-related groupings among replicates (Supplementary Figures 3A,B). However, since the sequencing depth allowed us to quantify the expression of 28,624 genes that represent <50% of the ones annotated for Homo sapiens and include genes expressed at low levels, results may be non-consistent when performing differential expression analysis. Therefore, we next conducted a selection of genes expressed at reliable levels according to DESeq2 independent filtering procedure. The four datasets resulting from this filtering were paired in four comparisons to perform the differential expression analysis as follows: 1) 24 h dyn vs. 24 h stat, and 2) 72 h dyn vs. 72 h stat assessed the effect of strain at the two different time points 24 and 72 h; 3) 72 h stat vs. 24 h stat, and 4) 72 h dyn vs. 24 h dyn evaluated the effect of time on the culture conditions. These comparisons led to the following number of differentially expressed genes (DEGs): 1) 103 DEGs (24 hdyn vs. 24 h stat), 2) 819 DEGs (72 h dyn vs. 72 h stat), 3) 245 DEGs (72 h stat vs. 24 h stat), and 4) 72 DEGs in (72 h dyn vs. 24 h dyn) (Supplementary Table 2). We used hierarchical cluster analysis to assess the relationships between the DEGs in the four experimental conditions (Figure 1). As shown, the heatmaps built using the four different datasets indicated a major prevalence of DEGs in the static vs. dynamic conditions at 72 h. In addition, coherent differential regulation of genes was already observed at an earlier time point in dynamically strained vs. control cells for 24 h (Figure 1A). Moreover, a time-dependent effect both in static and dynamic conditions was observed (Figures 1B,C). The higher number of DEGs resulting from the comparison between dynamic vs. static conditions at 72 h prompted us to further analyze the identity of the top-score genes up/downmodulated in this specific condition and to explore whether these genes were represented already at the earlier time point (24 h of stimulation). To this aim, we listed the 10 most up-regulated and the 10 most down-regulated genes in the 72 h dyn vs. 72 h stat condition (Table 1). These genes can be easily visualized in a volcano plot displaying the 819 DEGs arranged by fold change and Benjamini-Hochberg (BH)-adjusted p-value (Figure 2), they all had significantly higher expression (at least 1.5-fold more) than static controls along with a significant p-value (BH-adjusted p-value ≤ 0.05). In addition, Table 2 shows the top regulated DEGs in the 24 h dyn vs. 24 h stat conditions. Only five genes were present in the top regulated lists at both time points representing candidates consistently regulated by mechanical stress in SVPs: Adhesion Molecule with Ig Like Domain 2 (AMIGO2), Serine/threonine-protein kinase 38-like (STK38L), Caveolae Associated Protein 4 (CAVIN4), Growth differentiation factor 5 (GDF5) and Polypeptide N-acetylgalactosaminyltransferase 15 (GALNT15) (Table 3).

Among the genes that appear to be involved in vascular biology/pathology, our interest focused on AMIGO2, a gene encoding for a cell surface type I transmembrane receptor, which is a member of a novel class of leucine-rich repeat (LRR) and Ig superfamily proteins (32). AMIGO2 was identified for the first time in 2003, as an adhesion molecule necessary for the development of the axonal tract of neurons (33). Later, numerous other studies have reported AMIGO2 expression in gastrointestinal tract cancers, revealing its anti-apoptotic and cell adhesion activities which bestow to the tumor cells a higher metastasis formation capacity (34–38). Therefore, considering that cell adhesion and migration are two of the main mechanisms involved in the reorganization of the SV wall during the formation of the neointima, we explored the putative functions of AMIGO2 in mechanically stimulated SVPs and the signaling that may lie upstream of its differential expression at a transcriptional level.

First, we validated the RNA-Seq results using RT-qPCR on the same RNA pools used to perform the RNA-seq and on RNAs isolated from SVPs of different patients subjected to uniaxial strain. Results confirmed the overexpression of AMIGO2 in the SVPs when mechanically stimulated (Figure 3).

We then investigated the regulation of AMIGO2 in SVPs under mechanical stress by examining the DEGs in all data sets, taking into consideration the AMIGO2-associated expression processes, the interactions with other proteins, and the biological functions in which it is putatively involved. First, through iRegulon motif-based prediction we identified 8 transcription factors (TFs) potentially involved in AMIGO2 expression regulation. Hierarchical clustering showed a partial grouping among SVPs samples after 24 h of strain, which becomes robust when considering the samples cultured for 72 h (Figure 4). Interestingly, 5 out of 8 TFs were related to transforming growth factor β (TGF-β) pathways (RUNX1, PRDM1, SOX4, PPARG, and SMAD3), known for modulating mesenchymal phenotype acquisition (39–43). The TF CBFB acts in coordination with RUNX1 regulating the transcription of several genes, one of which is NOTCH3 (44). FOXD1 mediates gene expression of the cell during the reprogramming process (45). These activities combined with the negative effect of stretching on TEF expression suggest the switching of SVPs toward a proliferative phenotype (46). For a better understanding the effects of the mechanical stretching on the TGF-β pathway, we performed Ingenuity Pathway Analysis (IPA) at both 24 h dyn vs. 24 h stat (Supplementary Figure 4) and 72 h dyn vs. 72 h stat conditions (Figure 5). Through pathway enrichment analysis for TGF-β signaling at 72 h dyn vs. 72 h stat, we were able to identify 36 genes associated with the TGF-β pathway that were down-regulated and 45 genes up-regulated (Supplementary Figure 5). To show relevant relationships between modulated genes we performed IPA analysis in the complete dataset. The topmost identified networks are shown in Figures 6A,B. In both comparisons 24 h dyn vs. 24 h stat and 72 h dyn vs. 72 h stat, AMIGO2 expression was predicted to be controlled by NR3C1, the human glucocorticoid receptor gene, and NEUROG1, a regulator of neural progenitors' differentiation (47, 48). Connections were also found for the mesenchymal oncogenes FUS-DDIT3 at both time points and for RASSF1 at 72 h (49, 50).

To gain further insight into mechanical strain-associated changes in SVPs in relation to AMIGO2 we analyzed the human protein-protein interaction networks by IPA (Figures 7A,B). In the 24 h dyn vs. 24 h stat comparison, AMIGO2 was found connected to NFkB complex which was predicted up-regulated by the modulation of DEGs present in the network. After 72 h of culture, the specific modulation of several genes, including AMIGO2, suggested an interaction with Akt, which was predicted to be inhibited as a result of the stretch-dependent changes in the transcriptional profile of SVPs. This finding was in contrast with what has been reported in AMIGO2's activation mechanism in endothelial cells (ECs) (36), and suggests that in SVPs AMIGO2 could operate through different pathways, independent of PDK-Akt. Next, the effect of mechanical strain on the molecular and cellular functions of SVPs has been closely examined via Gene Ontology (GO) analysis of the 24 h dyn vs. 24 h stat and 72 h dyn vs. 72 h stat conditions. Table 4 shows the top 5 enriched pathways involving AMIGO2. The topmost function in SVPs stretched cells was cell-matrix adhesion (p-value 7.72 × 10⁻¹⁰), along with chemotaxis (p-value 6.92 × 10⁻¹⁰) and cell-cell adhesion via plasma-membrane adhesion molecules (p-value 2.07 × 10⁻¹⁰), describing a potential switching of the SVPs phenotype toward migration. Ultimately, diseases and biological function-related pathways involving AMIGO2 were obtained through IPA (Figures 8A,B). We observed an increase in the number of functions identified prolonging mechanical stress from 24 to 72 h. For example, the Cell Death and Survival macro-category after 24 h addressed only the regulation of apoptosis and necrosis, but after 72 h we also detected activation of cell viability and cell survival pathways. Of note, the cancer-related biofunctions were the most associated with AMIGO2 at both time points, reflecting its potential role in tissue invasion and remodeling (34, 35, 37, 38). Thus, the investigations performed with iRegulon and IPA as well as the AMIGO2-related GO findings shed a light on how mechanical stress could produce a phenotypic switching in the SVPs, mediating their activation and increasing their responsiveness to modifications occurring in the extracellular environment.

Given that the cyclic mechanical stress on SVPs produced a significant overexpression of AMIGO2 at the transcription level, we explored the effect of pulsatile coronary flow on AMIGO2 protein expression. This was assessed in two experimental systems that we previously used to validate Thrombospondin-1 as a relevant target of mechanical stress in the human SV arterialization process (19). These consisted of an in vivo SV arterialization model, performed by surgical SV interposition into carotid arteries in pigs, and of direct stimulation of human SVs using a coronary pulse duplicator that allows reproducing the mechanical conditions of the coronary circulation in vitro. Immunofluorescence staining was used to quantify the percentage of AMIGO2 positive cells in pig SV native conduits (T0) and at 1, 7, 14, and 90 days after grafting into the carotid artery (Figure 9A). <40% of the cells in the SV adventitia expressed AMIGO2 at T0 (37.13 ± 5.84%, N = 3) (Figure 9B). This percentage remained relatively unaffected during the following 7 days after surgery. On day 14, the percentage reached almost 70% of the total cells of the adventitia (69.35 ± 3.52%, N = 4), but this was not statistically different from the previous time points. Only at day 90 post-surgery, we observed a significant increase of AMIGO2 positive cells compared to T0, day 1, and day 7. In addition, we successfully identified numerous AMIGO2 expressing cells in human SV, both in the untapped conduit (T0) and following the application of a pulsatile pressure regimen for 14 days in ex vivo culture (Figure 9C). It is interesting to note that in both experiments AMIGO2 expression was not restricted to the adventitia layer, but was also detected in several cells in the tunica media, as well as ECs. Particularly significant was the detection of AMIGO2/CD34 positive cells in the proximity of the vasa vasorum, where the SVPs are normally localized (21). The quantification of AMIGO2 expression in SVs from 4 different donors did not highlight any significant difference in the comparison T0 vs. Day 14 (data not shown). These data consolidate AMIGO2 as an important effector in the response to pathologic mechanical stress. Thus, despite the early AMIGO2 upregulation driven by the mechanical strain at 24 and 72 h in the SVPs, the protein increase in the adventitia appears to be delayed when the entire vein wall undergoes pulsatile arterial flow.