Vascular grafts are often used in the treatment of cardiovascular disease, such as atherosclerosis, aneurysms, and congenital heart defects1 and in the construction of vascular accesses for hemodialysis treatment. In the small diameter range (<5 mm), synthetic grafts fail due to thrombosis or intimal hyperplasia and are outperformed by autografts.2 Tissue engineering approaches have been used for the regeneration of functional blood vessels,2–7 but, despite significant advances, the ideal vascular graft for small diameter vessels is yet to be found. Processing choices during the manufacturing of a biologic scaffold, such as decellularization media and time and cross-linking procedures, can markedly influence the downstream remodeling events.8–12 This seems to be the case of extracellular matrix (ECM) vascular grafts made from small intestinal submucosa (SIS). Small intestinal submucosa have been found to produce promising although inconsistent results in vivo, with patency rates varying between 0% and 100%,3,5,13–19 concurrently with variability in luminal surface modification treatments5,20 and hydration state.16,19
Manipulation of the luminal surface through the preservation of the stratum compactum layer of the intestine—a dense collagen layer naturally present in the small intestine3,14, or through the deposition of a thin layer of dense bovine collagen5—has been observed in some studies to increase patency from 0–13% to 88–100%.3,5,14 Yet, the presence of such dense collagen luminal layer has also been observed to produce poor cellular infiltration,21 which ultimately hinders the regeneration of the tissue. On the other side, hydration might also play a role in host response. While hydrated planar ECMs have shown in vitro a threefold improvement in cellular adhesion,21,22 dehydrated SIS sheets are still of interest, since they have also produced promising results in tissue regeneration in vivo in other anatomical locations,8,11,23 and are easier to manipulate. Furthermore an increased adhesion might also relate to adverse outcomes, such as hyperplasia (a recurrent cause of vascular graft failure).
Our group previously found that the two mentioned fabrication parameters affect the mechanics, microstructure, and micromechanical environment of SIS.24 In this study, we used a 7d in vivo model to investigate and compare the effects of these parameters on early in vivo patency and regeneration, in an effort to predict the combination of these two parameters that would achieve successful blood vessel regeneration. Our 7d model accounted for early failure due to thrombogenesis25 and included the analysis of the macrophage profile as a predictor of mid-term regeneration. Macrophages are a subset of the mononuclear cell population that migrates to the material immediately after implantation, is predominant after 2–10 days, and is capable of modulating its polarization between M1 and M2 phenotypes in response to local stimuli in the process of wound healing.9,11,12,23,26 M1 phenotype macrophages have been associated to the proinflammatory pathway that leads to inflammation, fibrosis, and material encapsulation, whereas M2 macrophages have been identified to be of an anti-inflammatory phenotype that triggers cellular recruitment, differentiation and proliferation, angiogenesis and degradation of bioresorbable materials.8,26,27
Materials and Methods
Four differently fabricated SIS vascular grafts were obtained according to a two-factor, two-level factorial experimental design as shown in Table 1. All specimens were prepared from the jejunum portion of three small intestines harvested from market weight swine within 10 minutes of euthanasia and randomly assigned to the study groups. The tissues were immediately placed in 0.9% saline solution and kept at 4°C. Fabrication of SIS in which the dense collagen luminal layer was preserved (P grafts) was performed according to the methods described previously by Badylak et al.22 In brief, intestinal contents were rinsed and the intestine was split open longitudinally to form a rectangular membrane, with its longitudinal axis parallel to the longitudinal direction of the intestine. The tunica mucosa, muscularis externa, and tunica serosa layers were removed by mechanical scraping. The remaining tissue was comprised of the stratum compactum, muscularis mucosa, and submucosa layers. The tissue was then disinfected with a 0.1% peracetic acid solution (Sigma Aldrich, St. Louis, MO) and rinsed thoroughly in phosphate buffered saline (PBS, pH = 7.0) and type I water. Similarly, SIS in which the dense collagen luminal layer was removed (R grafts) was fabricated by mechanical scraping, but by disinfecting with a proprietary sodium hypochlorite and hydrogen peroxide solution (both reagents from Sigma Aldrich) that removed the stratum compactum, and finishing by rinsing in PBS and autoclaved deionized water. It should be noted that both disinfection procedures had a sterilization effect on the samples. After disinfection and rinsing, all SIS membranes were stored in autoclaved deionized water at 4°C until shaping into cylindrical tubes. The face of the membranes that was closest to the original luminal surface of the intestine (i.e., the face closest to the tunica mucosa) was designated and marked as luminal and used as such in the cylindrical grafts. Evidence of the presence or removal of the stratum compactum, as well as of effective decellularization, was provided by hematoxilin and eosin (H&E) staining (Figure 1A, C, respectively). Further microstructural characteristics were described elsewhere.24
Dehydrated vascular grafts (D grafts) were fabricated by wrapping an SIS membrane around a 4.5 mm diameter cylindrical mandrel, air drying in a laminar flow hood for 2 hours and re-sterilizing with ethylene oxide. Grafts in a hydrated state (H grafts) were shaped into cylindrical tubes by the surgeon in the surgical room before implantation, by wrapping the same mandrel with the hydrated membrane, suturing along the edge with 7-0 prolene suture (Ethicon, Somerville, NJ) and keeping hydrated in a solution of heparin (125 UI/ml; Figure 1B, D, respectively).
In vivo Model
All the procedures were performed in accordance to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health and approved by the Institutional Review Board of the Fundación Cardioinfantil. For implantation surgery, 25 kg Yorkshire swine were provided intramuscular sedation with xylazine (1 mg/kg), midazolam (0.5 mg/kg), and atropine (0.04 mg/kg), received cefalotin antimicrobial prophylaxis (20 mg/kg) and tramadol for analgesia (1 mg/kg). Induction and maintenance of anesthesia was administered intravenously with propofol (30 mg/kg/h). Grafts were implanted in the right and left external carotid arteries of eight animals (1.5–2.0 cm long, n = 4, statistical power of 0.74). The right external carotid artery was exposed first through a longitudinal incision in the neck. Heparin sodium was provided (80 UI/kg) 1 minute before arterial clamping, and a randomly chosen graft was then placed as an interposition graft with 7-0 prolene suture (Ethicon). Dehydrated (D) grafts were rehydrated 3 minutes before implantation in a solution of heparin (125 UI/ml). The excised native carotid artery (NCA) was stored in saline at 4°C for biaxial testing control. Upon completion of the anastomoses and re-establishment of arterial flow, the left external carotid artery was exposed and another interposition graft was implanted in the same manner. Closure was performed by layers. A prophylactic postoperative treatment of antiaggregant (aspirin, 100 mg) and anticoagulant (enoxaparin sodium, 20 mg) was provided daily during a 7 day implantation period, after which animals were euthanized by providing sedation with the same cocktail used for implantation, and then administering a solution of pentobarbital sodium and diphenylhydantoin sodium (Euthanex, Invet S.A., Bogotá, 0.2 ml/kg). Grafts were collected with adjacent segments of the carotid artery and tested for patency by verifying the outflow of saline supplied with a syringe at the proximal end of the sample. Specimens were finally cut longitudinally in half to provide samples both for mechanical testing and histology/immunohistochemistry.
Histology and Immunohistochemistry
Half-sections of the explanted specimens were fixed in buffered formalin and embedded in paraffin. Paraffin blocks were cut into 4 μm thick sections, mounted into glass slides and deparaffinized in xylene and a graded series of ethanol solutions (100–80%). Hematoxilin and eosin and Masson’s Trichrome were used for histology staining to determine patency. For immunohistochemistry, slides were washed in water, blocked in 6% hydrogen peroxide for 5 minutes, and washed again in water. Antigen retrieval was performed in a citrate-based buffer (Novocastra RE7113, Leica Biosystems, Newcastle, pH 6.0) for macrophage phenotypes at boiling temperature for 1 minute or in a Tris/EDTA solution (EnVision FLEX Target Retrieval Solution High pH K8000, Dako, Glostrup, ph 9.0) for endothelium at 97°C for 20 minutes, after which samples were allowed to cool down for 20 minutes. Following 6 minute washes in water and a Tris Buffered Saline with Tween (TBST) wash buffer (S3006, Dako, Glostrup, pH 7.0, 0.05 mol/L) were performed. The slides were then incubated with one of the following primary antibodies: rabbit anti-CCR7 (Novus, Littleton, CO) for M1 macrophage phenotype (1:500, 60 minutes), mouse anti-CD206 (Novus) for M2 macrophage phenotype (1:100, 60 minutes) or polyclonal rabbit anti-human von Willebrand Factor FVIII (Dako) for endothelial cells (ECs) (1:300, 60 minutes). After four washes with the TBST wash buffer, the sections were incubated in the secondary antibody (biotinylated link antibody and peroxidase-labelled streptavidin, K0690, Dako) for 30 minutes, washed again in TBST, incubated in 3,3′-diaminobenzidine chromogen solution (Dako) and washed in distilled water. Counterstaining with H&E was performed afterwards for 1 minute followed by immersion in ammonia water (0.037 mol/L). Tissues were finally dehydrated with graded solutions of ethanol (80–100%) and xylene and coverslipped with HistoResin mounting media (Leica, Buffalo Grove, IL).
Quantification of Early Regeneration
Regeneration outcome was assessed by a blinded pathologist using a scoring methodology in which the result for each criterion was classified into one of four possible regeneration score (RS) ranges (Table 2). A lower RS indicated a response in the scarring pathway and a higher RS a response in the regenerative pathway. Observations were performed on three 40× magnification fields located after the proximal anastomosis, at the middle portion of the graft and before the distal anastomosis. Macrophage phenotype was evaluated with the number of CCR7+ and CD206+ cells in the same 40× fields and the CD206+/CCR7+ ratio was calculated. Given the short length of the implantation period, which was meant to account for early failure mechanisms, the presence of ECs was considered to be better assessed qualitatively as present or absent.
Biaxial Mechanical Testing
Biaxial mechanical testing was performed on a Biaxial Testbench device (Bose Electroforce, Eden Prairie, MN). In brief, square 10 × 10 mm samples were mounted to tissue holders, using hooks attached to 4-0 suture lines set up in loops that encircled five rods within the tissue holder. The loops around the rods assured that forces were evenly distributed in the lines and along the sides of the samples. The sides of the samples were aligned to the longitudinal (X1) and circumferential (X2) directions of the graft. Samples, hooks, and pulleys were immersed in a water bath at 37°C to maintain hydration and physiological temperature of all samples during testing. Tests were displacement-controlled up to a maximum motor displacement of 6.5 mm. Loads were monitored with two load cells (1 mN/0.1 g resolution) and strains were digitally calculated from the displacement of five black markers affixed to the surface of the sample. Data was transformed into stresses and stretches (stretch = (final length)/(initial length)) for analysis. An anisotropy ratio (AR) defined as (X2 maximum stretch)/(X1 maximum stretch)24 was also calculated to assess changes in the preferential fiber orientations over time.
Regeneration scores and macrophage phenotype quantitative data are reported as mean ± standard error of the mean. Statistically significant differences were determined with two-way ANOVA and Tukey’s test (p value < 0.05). Differences between parameters (i.e., P vs. R and H vs. D) were analyzed with Student’s t-test. All analyses were performed in GraphPad Prism 6.
Patency Outcome and Macroscopic Appearance
Assessment of outflow through the explanted samples indicated patency rates of 100% (4/4) for PD grafts, 100% (4/4) for RD grafts, 50% (2/4) for PH grafts, and 75% (3/4) for RH grafts. Patent grafts had white, glistening areas distributed throughout the luminal surface (Figure 2, PD, RD, and RH). Occluded specimens had dark and stiff thrombi present mainly at the distal anastomoses (Figure 2, PH). Macroscopically, PD grafts sharply kept their tubular shape but were stiff and fragile. One PD specimen had laminar (nonocclusive) thrombi close to the anastomoses and along the length of the graft. RD samples had a soft feeling. A laminar (non occlusive) thrombus was observed in one RD specimen and a slight stenosis at the distal anastomoses was observed in two specimens. The two patent PH grafts had a minor stenosis and one of the occluded specimens was surrounded by inflammatory exudate. One of the patent RH grafts had a minor stenosis caused by tissue ingrowth and a laminar thrombus.
Histological Assessment and Quantification of the Regeneration
Overall, the early host response comprised a generally populous infiltration of inflammatory cells and fibroblasts inside the scaffold and tissue growth on the abluminal periphery of all the grafts (Figure 3). Differences in RS between groups were more marked in vascularization and fibroblast population of the scaffold (Figure 4). Laminar thrombi composed of a fibrin mesh, sometimes with entrapped red blood cells, were found covering the luminal surface of a large proportion of the specimens (black arrows in Figure 3), as quantified by the thrombus length criteria (RS 2, Figure 4). Nonetheless, the short thrombi diameters proved their mainly nonocclusive character, especially in PD scaffolds which had the best outcome in thrombogenicity related to early-failure. Overall, the inflammatory response to all the scaffolds was given an RS of 3. Monocytes were observed inside RD, PH, and RH scaffolds and in the tissue surrounding PD grafts (white arrow heads in Figure 3). PD scaffolds had the best outcome for polymorphonucleocytes, which were predominantly present close to the distal anastomoses. Very few giant cells were found in the observed fields (seven in total across all groups, five in one RD specimen, and two in one RH specimen). Vascularization (black arrow heads in Figure 3) was lowest in PD (RS 1), higher in RD and PH (RS 2), and highest in RH (RS 3). Fibroblast population was lowest in PD (RS 1) and had an RS 3 in all the other groups (white arrows in Figure 3). Capillaries and fibroblasts were observed distributed throughout RD, PH and RH scaffolds.
PD scaffolds had a scarce macrophage infiltration both of CCR7+ (M1) and CD206+ (M2) phenotypes (Figure 3, gray arrows, and Figure 5A). CCR7+ macrophages were found in a similar amount in the grafts with a removed dense collagen layer (RD and RH groups, Figure 5A), while PH scaffolds had the highest count of this phenotype. The CD206+ count was highest in RD scaffolds and similar between hydrated scaffolds (PH and RH). PD, RD, and RH scaffolds had an M2 dominant profile, whereas PH had a M1-dominant profile, as indicated by the CD206+/CCR7+ ratio (Figure 5B). The M2 predominance was significantly stronger in D scaffolds than in H scaffolds. Endothelial cells were found on the luminal surface close to the anastomosis in two RD, one PH, and one RH specimen (Figure 6). No ECs were found in PD specimens. FVIII staining was also found around vasa vasorum and capillaries within the regenerated vascular wall.
Biaxial Mechanical Properties
The biaxial mechanical properties of the four SIS materials before implantation (Figure 7A) were previously obtained24 and were used here to analyze the mechanical evolution of the grafts after early in vivo implantation. Before implantation, only the X2 data of PH and RH samples was beyond a 1.024 mm/mm stretch. Grafts in the same hydration state had a similar mechanical biaxial behavior, with H grafts more elastic than D grafts, and X1 was the preferential direction in all scaffolds (AR > 1). After implantation, P grafts were found to be stiffer than R grafts (Figure 7B) and there were no clear differences between the elasticity of H and D groups. PD samples had the stiffest behavior and RD grafts were the most elastic. R grafts were found to have maximal stress-stretch values outside the initial stretch range (shaded areas in Figure 7). The elastic behavior of RD samples was the closest to that of NCA. Anisotropy remained the same in RD grafts, had a marked increase in RH grafts, and, surprisingly, became inverted in P grafts: the behavior in the X1 direction after the 7d period evolved into that for the initial X2 direction and vice versa, making the X2 direction preferential in P scaffolds after implantation and changing their AR to a <1 value (Figure 8). Although these values did come closer to the AR measured for NCA, P grafts still remained highly stiff and did not stretch beyond their initial stretch range.
Vascular grafting of small diameter blood vessels is still only possible through the use of autografts. To get closer to the development of an off-the-shelf option, we herein used a predictive approach to perform a comparison on the early in vivo patency and regeneration outcomes of 4.5 mm ID SIS grafts manufactured by four different fabrication procedures. Our methodology allowed us to select a promising fabrication procedure for future mid-term animal studies.
Two of the four fabrication procedures were associated with a 100% patency (PD and RD). However, PD grafts were fragile and mechanically stiff, probably due to the marked lack of cellular infiltration. In contrast, RD grafts showed an extensive macrophage and fibroblast infiltration (which could play a role on the gradual degradation of SIS and the deposition of new ECM), a regenerative phenotype dominant macrophage profile, the early presence of ECs and mechanics rapidly evolving to the elasticity of native vascular tissue. Thus, RD grafts are recommended for future mid-term studies on the basis of our findings.
Overall, the P parameter had the main effect of inducing a stiffest elasticity compared to R, whereas the D parameter was associated with a more dominant regenerative macrophage profile compared with H. As mentioned before, hydration has been observed to increase the adhesion of cells in SIS. However, our results indicate that this increased adhesion could also hinder overall regeneration through the increased adhesion of adverse cellular phenotypes, such as CCR7+.21,22
Some of the grafts in this study were found to have minor anastomotic stenoses. Our observations suggest that this might be associated with the vasoconstriction that occurs during surgical manipulation and the initial mechanical properties of SIS, which allow the creasing of the scaffold to adapt to the contracted native vessel’s diameter, instead of forcing the vessel to stretch up to the graft’s diameter. In our study, the macroscopic inspection of the graft suggested that the standard continuous suture might have not relaxed enough along with the native vessel upon closure of the incision, creating a narrow anastomosis at implantation. A possible solution to this problem would be the use of a discontinuous suture that provides the anastomosis with freedom to stretch with the native vessel. Nonetheless, 13/16 grafts stayed patent after 7 days. Four had migration of ECs close to the anastomoses. Migration of ECs may be a consequence of peptides that derive from SIS degradation which are chemoattractants to primary ECs.27 The location of ECs close to the anastomoses at such an early time point observed herein could indicate that endothelialization of the graft begins transanastomotically, and is worth of further investigation to improve the understanding of the regenerative process.
It should be noted that this work did not investigate the impact of the fabrication parameters on the detachment of growth factors, glycosaminoglycans and proteglycans present in the intestinal ECM.28 Nonetheless, such a study would be of a complementary nature from an early response perspective, as it would provide elucidation on other mechanisms underlying the differences in patency and regenerative outcome induced by fabrication. In addition, this study did not evaluate the specific role of the anticoagulant/antiaggregant prophylactic therapy used, which would be of interest for future studies, especially on the most successful RD grafts.
The results of our predictive animal model indicate that the two studied fabrication parameters were of SIS major importance on patency and regeneration in SIS vascular grafts. Overall, removing the stratum compactum layer of the small intestine and dehydrating SIS produced the best outcome after 7 days of in vivo evaluation, in terms of patency, cellular infiltration, macrophage phenotype, early endothelialization, and mechanics. This study is a contribution to the development of an off-the-shelf regenerative small diameter vascular graft.
The authors thank Sergio Galvis, DVM and Guillermo Ruiz, DVM for assistance with animal procedures, the Departments of Pathology of Hospital Universitario Fundación Santa Fe de Bogotá and Fundación Cardioinfantil for collaboration on histology and immunohistochemistry and Professor Rigoberto Gomez from the Department of Chemistry of Universidad de los Andes for support on the synthesis of peracetic acid.
1. Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics 2011 update: A report from the American Heart Association. Circulation. 2011;123:e18–e19
2. Hong Y, Ye SH, Nieponice A, Soletti L, Vorp DA, Wagner WR. A small diameter, fibrous vascular conduit generated from a poly(ester urethane)urea and phospholipid polymer blend. Biomaterials. 2009;30:2457–2467
3. Lantz GC, Badylak SF, Hiles MC, et al. Small intestinal submucosa as a vascular graft: A review. J Invest Surg. 1993;6:297–310
4. Robotin-Johnson MC, Swanson PE, Johnson DC, Schuessler RB, Cox JL. An experimental model of small intestinal submucosa as a growing vascular graft. J Thorac Cardiovasc Surg. 1998;116:805–811
5. Huynh T, Abraham G, Murray J, Brockbank K, Hagen PO, Sullivan S. Remodeling of an acellular collagen graft into a physiologically responsive neovessel. Nat Biotechnol. 1999;17:1083–1086
6. Naito Y, Shinoka T, Duncan D, et al. Vascular tissue engineering
: Towards the next generation vascular grafts
. Adv Drug Deliv Rev. 2011;63:312–323
7. Courtney T, Sacks MS, Stankus J, Guan J, Wagner WR. Design and analysis of tissue engineering scaffolds that mimic soft tissue mechanical anisotropy. Biomaterials. 2006;27:3631–3638
8. Badylak SF, Valentin JE, Ravindra AK, McCabe GP, Stewart-Akers AM. Macrophage phenotype as a determinant of biologic scaffold
remodeling. Tissue Eng Part A. 2008;14:1835–1842
9. Badylak SF, Gilbert T, Myers-Irvin Jvan Blitterswijk C. The extracellular matrix as a biologic scaffold
for tissue engineering. Tissue Engineering. 2008 Burlington, VT Academic Press:121–143
10. Valentin JE, Badylak JS, McCabe GP, Badylak SF. Extracellular matrix bioscaffolds for orthopaedic applications. A comparative histologic study. J Bone Joint Surg Am. 2006;88:2673–2686
11. Brown BN, Valentin JE, Stewart-Akers AM, McCabe GP, Badylak SF. Macrophage phenotype and remodeling outcomes in response to biologic scaffolds with and without a cellular component. Biomaterials. 2009;30:1482–1491
12. Keane TJ, Londono R, Turner NJ, Badylak SF. Consequences of ineffective decellularization of biologic scaffolds on the host response. Biomaterials. 2012;33:1771–1781
13. Badylak SF, Lantz GC, Coffey A, Geddes LA. Small intestinal submucosa as a large diameter vascular graft in the dog. J Surg Res. 1989;47:74–80
14. Lantz GC, Badylak SF, Coffey AC, Geddes LA, Blevins WE. Small intestinal submucosa as a small-diameter arterial graft in the dog. J Invest Surg. 1990;3:217–227
15. Sandusky GE Jr, Badylak SF, Morff RJ, Johnson WD, Lantz G. Histologic findings after in vivo
placement of small intestine submucosal vascular grafts
and saphenous vein grafts in the carotid artery in dogs. Am J Pathol. 1992;140:317–324
16. Sandusky GE, Lantz GC, Badylak SF. Healing comparison of small intestine submucosa
and ePTFE grafts in the canine carotid artery. J Surg Res. 1995;58:415–420
17. Badylak SF. Xenogeneic extracellular matrix as a scaffold
for tissue reconstruction. Transpl Immunol. 2004;12:367–377
18. Kim SS, Kaihara S, Benvenuto MS, Kim BS, Mooney DJ, Vacanti JP. Small intestinal submucosa as a small-caliber venous graft: A novel model for hepatocyte transplantation on synthetic biodegradable polymer scaffolds with direct access to the portal venous system. J Pediatr Surg. 1999;34:124–128
19. Badylak SF, Freytes DO, Gilbert TW. Extracellular matrix as a biological scaffold
material: Structure and function. Acta Biomater. 2009;5:1–13
20. Crapo PM, Gilbert TW, Badylak SF. An overview of tissue and whole organ decellularization processes. Biomaterials. 2011;32:3233–3243
21. Freytes DO, Tullius RS, Valentin JE, Stewart-Akers AM, Badylak SF. Hydrated versus lyophilized forms of porcine extracellular matrix derived from the urinary bladder. J Biomed Mater Res A. 2008;87:862–872
22. Badylak S, Liang A, Record R, Tullius R, Hodde J. Endothelial cell adherence to small intestinal submucosa: An acellular bioscaffold. Biomaterials. 1999;20:2257–2263
23. Valentin JE, Stewart-Akers AM, Gilbert TW, Badylak SF. Macrophage participation in the degradation and remodeling of extracellular matrix scaffolds. Tissue Eng Part A. 2009;15:1687–1694
24. Sánchez-Palencia DM, D'Amore A, González-Mancera A, Wagner WR, Briceño JC. Effects of fabrication on the mechanics, microstructure and micromechanical environment of small intestinal submucosa scaffolds for vascular tissue engineering
. J Biomech. 2014;47:2766–2773
25. Kumar VA, Brewster LP, Caves JM, Chaikof EL. Tissue engineering of blood vessels: Functional requirements, progress, and future challenges. Cardiovasc Eng Technol. 2011;2:137–148
26. Piterina AV, Cloonan AJ, Meaney CL, et al. ECM-based materials in cardiovascular applications: Inherent healing potential and augmentation of native regenerative processes. Int J Mol Sci. 2009;10:4375–4417
27. Brown BN, Londono R, Tottey S, et al. Macrophage phenotype as a predictor of constructive remodeling following the implantation of biologically derived surgical mesh materials. Acta Biomater. 2012;8:978–987
28. Li F, Li W, Johnson S, Ingram D, Yoder M, Badylak S. Low-molecular-weight peptides derived from extracellular matrix as chemoattractants for primary endothelial cells. Endothelium. 2004;11:199–206