Systems Approach to Discovery of Therapeutic Targets for Vein Graft Disease

Supplemental Digital Content is available in the text.


Materials, Specimen and Reagents
implantation. Neointima plaque was estimated using wall thickness measurement tool from the VevoVasc TM analysis package of the VevoLab TM software version 1.6.0 build 6078 (Fujifilm Visualsonics), in order to choose the appropriate animal model for the study.

Vein graft vessel wall layer proteomics sample preparation
Vein graft implantation surgery.
Vein graft implantation was done at 12 weeks of age on two Ldlr-/-mice and two wild type Danvers, MA) with protease inhibitor cocktail, PIC, (Sigma) and processed for mechanical tissue disruption using Precellys beads (Precellys, Atkinson, NH) high speed multi-directional dual beadgrinding homogenization. Vein grafts of Ldlr-/-mice were isolated from their carotid attachments and dissected free of the P.E.E.K. cuffs and perivascular fat as much as possible. Thickened neointimal plaque (NEO) was micro-dissected under the microscope to cleanly separate it from the adventitial layer (ADV) of the graft. The NEO and ADV layers were immediately placed in separate containers with 400 µL RIPA buffer with PIC as above and minced using fine microdissecting scissors. Vein grafts of wild type mice (WVG) did not produce a visibly thickened plaque both by ultrasound imaging and by ocular inspection making vessel layer microdissection not feasible. Hence, they were cleaned of their minimal perivascular fat, harvested and minced as above without separating neointima and adventitia. All vein graft tissue/tissue layer samples were processed for mechanical disruption and lysis using Precellys system as with the IVC controls above.

Protein and peptide preparation.
Tissue homogenates were sonicated on ice in a 4 O C environment/room using Sonifire 450 After peptide digestion, RapiGest reagent was removed by repeated dilution with LC/MS grade water followed by evaporative vacuum centrifugation for a total of five cycles, while being careful to avoid complete drying-out of the samples. After which, sample volume was raised to Samples were further diluted 10-fold for mass spectrometry.

Vein graft time course proteomics.
Twelve Ldlr-/-mice underwent vein graft implantation as above. Each mouse was maintained on high triglyceride high cholesterol Clinton-Cybulsky diet (Research Diets, New Brunswick, NJ) one week prior to implantation and continued until designated time points for vein graft harvest. The four timepoints investigated were day 1, day 3, day 7 and day 28 post-vein graft implantation. Three mice for each time point were maintained on high fat diet and then were euthanized for vein graft (VG) and inferior vena cava (IVC) harvest as described above. Protein and peptide preparations steps were done as above with the following exceptions: the protein input was increased to 100 µg and Lysyl endopeptidase was used for proteolysis. The samples were diluted between 2 to 10-fold in order to acquire a constant total ion current. Proteins are then clustered together into specific groups or "cluster numbers" based on the similarity of their trend lines. For clustering, the model-based algorithm either optimizes the number of clusters at a minimum Bayesian information criteria (BIC) in an unbiased manner.

Mass spectrometry and proteomics data analysis
Clustering is performed using the "mclust" R package, which determines a covariance structure and the number of clusters at the best BIC, which in our case resulted in 30 clusters ( Figure 1F).
XINA provided statistical and graphical tools to investigate the coabundance patterns of proteins derived from this multiplexed (VG and IVC) time-series proteomics data. Further explanation of how the algorithm clusters the protein abundances can be found in the aforementioned methods report 6 . Trends may either be increasing, decreasing, fluctuating or relatively unchanging across the four time points (day 1, day 3, day 14, and day 28). We subdivided the trend patterns into two: early phase abundant or predominant proteins and late phase abundant or predominant proteins. Among these proteins found in these trend clusters, we identified proteins that have higher abundances in either day 1 or day 3 time point, relative to their abundances in day 14 and day 28 time points using a reference threshold (red dashed line in legend, Figure 1F) and classify these proteins as early phase predominant proteins (see purple trend lines, Figure 1F). On the other hand, protein trend clusters with higher abundances in either day 14 or day 28 time points are classified as late phase predominant proteins (see orange trend lines, Figure 1F). Of these proteins we exclude the ones that have either early phase or late phase trend patterns that are also found in the IVC samples. Proteins exhibiting early phase and late phase patterns found in vein graft samples alone, but have unchanging trend (gray trend lines, Figure 1F) found in IVC (control) samples, are considered for the analysis. The rationale behind this is that if proteins have the same trend in both vein graft samples and IVC non-arterialized or control samples, then these proteins may have little to no association with the arterialized phenotype of the vein. We also excluded proteins that have unchanging trend (clusters with gray trend lines) in our downstream analysis.
Using these statistically filtered data, we performed integrated networks-based analysis including pathways enrichment to find significant disease associations and potential therapeutic target(s) among 'hub' nodes using the proprietary computational algorithms of MetaCore™ (Clarivate Analytics/Thompson Reuters).

Target prioritization by network of pathways analysis.
Increasing differentially expressed protein (DEP) lists from each group (

Formulation of siRNA lipid nanoparticles (LNPs) for in vivo gene expression silencing
siRNA for firefly luciferase (control siRNA) and mouse PPAR were synthesized by Axolabs (Germany). We used 2 oligonucleotides XD-07623 and XD-07624 among a list of 10 candidate siRNA oligonucleotides listed in Supp "recipient" mice. At the day of surgery, VG recipient animals were assigned randomly to receive either the siControl group or siPPAR- group depending on the type of siRNA-LNP will be given.
Both trial analyst and surgeon were blinded to the groupings as assigned by a third party. Tail vein injections were given a week before surgery, right after surgery, and twice a week for the subsequent four weeks following VG surgery for a total of 11 doses. After 4 weeks, vascular ultrasound was done to assess vein graft plaque sizes. In a select set of animals, we tested for glucose uptake in the vein grafts using intravenously injected RediJect Glucose 24 hours prior to animal tissue harvesting. Glucose uptake is recorded by near-infrared fluorescence imaging using the Olympus MVX-10 macroview scope.

Double fluorescence immunohistochemistry of vein grafts
Mouse vein grafts were prepared as fresh frozen tissue using OCT (optimal cutting temperature) resin. Frozen tissue blocks are transversely sectioned as 7.0 micrometer thick sections onto Superfrost TM plus-plus glass slides. The sectioned slides were thawed and dried at room temperature for 30 minutes followed by fixation with 4% paraformaldehyde solution in phosphate buffered saline (PBS) for 15 minutes. After rinsing the fixative solution off the slides with three washes of PBS, the slides were blocked with 5% goat serum and 5% donkey serum in PBS solution (blocker solution) for 30 minutes before exposing to the primary antibody solution.
The primary antibodies that were used (e.g. CD68, -SMA, MMP-9, & MMP13) were diluted with a solution of 1% bovine serum albumin (BSA) in phosphate buffered saline (PBS) at final concentration of 1.0 ug/mL. Primary antibody incubation was done overnight at 4 O C followed by triple washes of the slide sections with room temperature PBS at 5 minutes each. Subsequently, the slides were incubated with a secondary antibody solution of fluorescent conjugated antibodies diluted at 1:500 ratio in 1% BSA solution (PBS) for 1 hour. Slides were washed in PBS three times for 5 minutes in room temperature. Each slide was then mounted using fluorescent mounting medium VECTASHIELD with DAPI (Vector Laboratories, Burlingame, CA). Imaging was done on a Nikon Ti Eclipse confocal microscope. Image quantification was done by using Nikon Elements and % positive area of the signal was quantified within the neointimal area.

Picrosirius staining of vein graft sections
All succeeding steps in this workflow were carried out in room temperature (

Hematoxylin and Eosin Staining
All succeeding steps in this workflow were carried out in room temperature (25 O C). Vein graft sections post-cryostat sectioning are left to dry for 30 minutes and were then fixed in 10% formalin solution (Fisher Scientific, Pittsburg, PA) for ten minutes. After five minutes of rinsing in tap water, slides were washed with distilled deionized water (ddH 2 O) twice for three minutes each.
Slides are then placed in a solution of Harris hematoxylin (Fisher Scientific, Pittsburg, PA) (prefiltered prior to use) for two minutes. Slides were then rinsed well in ddH 2 O. The slides were then de-stained with 0.5% acetic acid quickly by dipping each slide in the acidic solution 2-3 times immediately followed by a rinsing step of ddH 2 O. Slides are then incubated in "bluing" solution of ammonium water (0.2% ammonium hydroxide in water) for one minute. This bluing reagent is rinsed off with running tap water for five minutes. Subsequently, the slides are briefly rinsed in 70% ethanol followed by staining with 1% alcoholic eosin (Eosin Y, VWR, Visalia, CA) for 25 seconds with gentle agitation of the slides. Eosin stain is washed off by dipping the slides in 2 fresh washes of 95% ethanol solution. Slides are then dehydrated with 100% ethanol followed by 2 washes of xylene for 3 minutes. Slides are the mounted using SHUR/Mount™ mounting media (VWR, Visalia, CA).

Plasma lipid and glucose colorimetric assay
Venous blood was harvested from the animals under terminal sedation using heparinized phlebotomy needles. Blood samples underwent centrifugation at 700 g for 20 minutes to separate platelet poor plasma. Plasma samples were assayed for triglyceride levels, cholesterol levels and glucose levels. We used the LabAssay Triglyceride Assay Kit (Wako, Richmond, VA) to assay plasma triglycerides, the Cholesterol E Assay (Wako, Richmond, VA) to measure plasma total cholesterol, and the Autokit Glucose Assay (Wako, Richmond, VA) to measure plasma random glucose levels. All procedures for these colorimetric assays were done as per manufacturers' suggested protocols. Colorimetric absorbance measurements were done using the SpectraMax Gemini XPS (Molecular Devices, San Jose, CA).

Bone marrow derived mouse macrophage culture
Wild type C57BL/6 mice aged ~12 weeks were sacrificed ( At day 10 in culture, cells are used for functional experiments.

In vitro mRNA silencing experiments and real time quantitative PCR assay (RT-qPCR)
Silencing experiments with PPAR are carried out using magnetofection of siRNA of PPAR or a non-targeting siRNA (siControl) using SilenceMag transfection kit (OZ Biotechnology, San Diego CA). Protocols for silencing were based on manufacturer's recommendations and was described in a previous report 17 . We silenced PPAR expression using 40 nM of species specific PPAR siRNA SMART Pool ON-TARGET oligonucleotide or Non-targeting siRNA (Dharmacon, Lafayette, CO). To check for PPAR specificity of differential gene expression, we added pemafibrate 100 nM to either non-targeting control or PPAR silenced groups. For each silencing condition used, we nested the following additional conditions: (1) PBS for 6 hours for baseline macrophage status (plus 5 ppm dimethyl sulfoxide, DMSO), (2) lipopolysaccharide (Sigma-Aldrich, Burlington, MA), LPS, at 10ng/mL for 6 hours + DMSO (5 ppm), (3) LPS at 10ng/mL for 6 hours with 100 nM pemafibrate (Kowa, Tokyo). RT-qPCR was carried out using PerfeCTa® qPCR SuperMix kit (VWR, Visalia, CA) as per kit vendor's recommended protocols. Samples were then processed for single cell qPCR experiments. All experiments were carried out in biological triplicates using three buffy coat donors with PBMC isolations done at different occasions depending on the schedule of blood collections.

Single cell qPCR and analysis
Monocytes are isolated from peripheral blood mononuclear cells derived from buffy coats (Research Blood Products, Boston, MA) using EasySep human monocyte negative selection kit (STEMCELL Technologies, Cambridge, MA). Following manufacturer's recommended protocol, monocytes are isolated and plated for cell culture using serum free RPMI 1640 for 2 hours.
Subsequently, serum free media was discarded along with floating non-adherent monocytes and was replaced with complete growth media: RPMI 1640 with 5% human serum and supplemented with penicillin streptomycin. Cells were maintained in this culture media for ten days to allow cells into fully differentiating into adherent macrophages. Cells were divided into three conditions: (1) baseline non-stimulate macrophages, pre-treated with 5 ppm DMSO for 12 hours, (2) pre-treated macrophages with 5 ppm DMSO for 12 hours then stimulated with 10 ng/mL LPS treatment for 6 hours and (3)  commercial sequencing service (Novagene) and analysis of raw data was done using the following software: Cell Ranger, Loupe browser, SeqGeq, MetaCore, Plotly (python), and Excel.

Arteriovenous fistula disease mouse model creation and optimization
To assess a second maladaptive vein disease, we proceeded to create a surgical arteriovenous fistula (AVF) in fat fed Ldlr-/-mice to model AVF failure and disease. We based our surgical technique from previously reported mouse AVF models 26, 46 with some modifications.
The Institutional Animal Care and Use Committee at Beth Israel Deaconess Medical Center reviewed and approved all animal experimental protocols (protocol # 021-2017). Male LDL receptor-deficient (Ldlr-/-) mice in a C57BL/6 background were purchased from Jackson Laboratory (Bar Harbor, ME) and maintained with normal chow diet and water ad libitum. Twelve- week-old Ldlr-/-mice were converted to high fat diet containing 1.25% cholesterol (D12108CO; Research Diets, NJ). After a two-week high-fat diet feeding, arteriovenous fistula (AVF) was performed by an end-to-side anastomosis between the left carotid artery and the ipsilateral external jugular vein (EJV) using ten to eleven interrupted stitches of Ethilon 11-0 suture (Owens and Minor, Mechanicsville, VA) per anastomosis. The cut portions of the source EJV were ligated using Ethilon 8-0 sutures. This procedure generates the arterialized EJV (a-EJV). After the AVF procedure prior to closing the surgical site, blood flow and pulsation of the a-EJV was visually assessed to confirm successful fistula formation. The surgical site is then closed using 6-0 silk sutures.

Gain-of-function study of PPAR in the murine AVF disease model
To test if PPAR activation could produce a similar benefit to AVF disease model as it did for the vein graft disease model, we conducted a similarly randomized gain-of-function study on the mouse AVF model. We used 22 fat-fed Ldlr-/-male mice, aged 12 weeks old, randomized into two groups: (1) a control group fed high fat diet and (2)  "open" or "patent" due to high velocity turbulent flow of blood as detected by the color doppler. If there was no color doppler signal, we called this as "non-patent" or "closed". However, we are aware that absence of color doppler signal may also mean that the stenosis is significant enough to slow down the velocity of the blood flow to an extent below the threshold of detection by the color doppler, even at the lowest pulse repetition frequency (PRF) threshold of 1.0. VevoLab software version 1.6.0 build 6078 (Fujifilm Visualsonics) was used in the assessment. At seven weeks post -AVF surgery, the animals were sacrificed and the AVF venous limb, the a-EJV, was harvested and processed for histology. Majority of the neointima formation is in the proximal third of the a-EJV nearest to the anastomosis site. By taking this proximal third segment (~1.2 mm for every AVF study mice), we sectioned and took this proximal third segment's mid-length cross section and stained for Hematoxylin and Eosin (H&E). Histologic examination was done to assess for patency ("open") of the a-EJV lumen described as having minimum stenosis of 30% patency.
Otherwise, we designate that specimen as "occluded" or "closed". Statistical analysis was done using GraphPad Prism software calculating for the Fisher's exact as a measure for statistical significance of categorical variable (observations).

Seahorse metabolic assays
We assessed energy metabolism in LPS stimulated macrophages and the influence of perturbation via PPAR activation. The rate of glycolysis of macrophages is inferred by measuring the extracellular acidification rate (ECAR) and the rate of oxidative respiration is inferred by measuring the oxygen consumption rate (OCR

Metabolomic survey of human primary macrophages
We outsourced metabolomic profiling of human primary macrophages to Metabolon, Inc.
We availed of three analysis survey packages: (1) macrophage whole cell metabolomic survey using the HD4 platform, (2) whole cell lipidomic survey using the CLP platform and the (3) isolated macrophage mitochondria metabolomic survey using the HD4 platform. Human PBMC macrophages were cultured from LeukoPak buffy coat preparations (StemCell Technologies, Cambridge MA). For each of the three analyses surveys listed we set up five sample conditions: (1) baseline macrophage status, no LPS stimulation, 12 hours DMSO pretreatment (5 ppm), (2) 12 hours DMSO pretreatment (5 ppm), 1 hour 10 ng/mL LPS treatment + DMSO (5 ppm), (3) 12 hours 100 nM pemafibrate pretreatment, 1 hour 10 ng/mL LPS treatment + 100 nM pemafibrate, instructions. Whole cell specimens were harvested by washing the culture plates with ice cold PBS followed by scraping the cells using a cell scraper. Cells were immediately frozen in liquid nitrogen after discarding excess PBS post-centrifugation. Each mitochondria specimen was prepared by harvesting and pooling macrophage cells in culture from ten 10 cm 2 culture dishes, then isolating mitochondria using an isolation procedure that has been previously reported 46,47 .
In summary, this 72-sample study examined global metabolic profiles in human macrophages and mitochondrial isolates treated with either LPS alone, or LPS with an unknown drug. Samples were collected at baseline, one hour, and four hours. Global metabolic profiles were determined from the experimental groups outlined in the table below. Platform, samples were inventoried and immediately stored at -80 O C. Each sample received was accessioned into the Metabolon LIMS system and was assigned by the LIMS a unique identifier that was associated with the original source identifier only. This identifier was used to track all sample handling, tasks, results, etc. The samples (and all derived aliquots) were tracked by the LIMS system. All portions of any sample were automatically assigned their own unique identifiers by the LIMS when a new task was created; the relationship of these samples was also tracked.

Mitochondria -Global
All samples were maintained at -80C until processed. Samples are then prepared using the automated MicroLab STAR® system from Hamilton Company. Several recovery standards were added prior to the first step in the extraction process for quality control, QC purposes. To remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for two minutes (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions: two for analysis by two separate reverse phase    (n=5) and siPPAR (n=6) groups of Ldlr-/-mice were used. siRNA LNP were injected in the tail vein at a dose of 0.5 mg/kg/dose twice a week for three weeks. Animals were challenged with an intraperitoneal injection of 2% thioglycolate (w/v) for 48 hours (pain medications were appropriately given). Afterward, peritoneal lavage was done to harvest the peritoneal macrophages. B. Spleens were also harvested, and mononuclear cells were isolated by homogenizing the spleen and gradient centrifugation (Histopaque 1083). qPCR of peritoneal macrophages and spleen mononuclear preps were done to assess silencing. (*) p < 0.05, (n.s.) p > 0.05. C. Bodyweight monitoring between Pemafibrate and Control group, n=11 mice per group comparison D. Food consumption monitoring between Pemafibrate and Control group, n=11 mice per group comparison (per cage average).          conjugates. siControl (n=5) and siPPARα (n=6)         Mitochondria are isolated from THP-1 differentiated cells treated with LPS for 2 hours with or without pemafibrate (100 nM). Pemafibrate group was pretreated with 100 nM for 2 hours before LPS stimulation, while DMSO at 4 ppm was used for vehicle control.