Transforming Growth Factor-β1 Modulates Extracellular Matrix Production, Proliferation, and Apoptosis of Endothelial Progenitor Cells in Tissue-Engineering Scaffolds
Background— Valvular endothelial cells and circulating endothelial progenitor cells (EPCs) can undergo apparent phenotypic change from endothelial to mesenchymal cell type. Here we investigated whether EPCs can promote extracellular matrix formation in tissue engineering scaffolds in response to transforming growth factor (TGF)-β1.
Method and Results— Characterized ovine peripheral blood EPCs were seeded onto poly (glycolic acid)/poly (4-hydroxybutyrate) scaffolds for 5 days. After seeding at 2×106 cells/cm2, scaffolds were incubated for 5 days in a roller bottle, with or without the addition of TGF-β1. After seeding at 15×106 cells/cm2, scaffolds were incubated for 10 days in a roller bottle with or without the addition of TGF-β1 for the first 5 days. Using immunofluorescence and Western blotting, we demonstrated that EPCs initially exhibit an endothelial phenotype (ie, CD31+, von Willebrand factor+, and α–smooth muscle actin (SMA)−) and can undergo a phenotypic change toward mesenchymal transformation (ie, CD31+ and α-SMA+) in response to TGF-β1. Scanning electron microscopy and histology revealed enhanced tissue formation in EPC-TGF-β1 scaffolds. In both the 10- and 15-day experiments, EPC-TGF-β1 scaffolds exhibited a trend of increased DNA content compared with unstimulated EPC scaffolds. TGF-β1–mediated endothelial to mesenchymal transformation correlated with enhanced expression of laminin and fibronectin within scaffolds evidenced by Western blotting. Strong expression of tropoelastin was observed in response to TGF-β1 equal to that in the unstimulated EPC. In the 15-day experiments, TGF-β1–stimulated scaffolds revealed dramatically enhanced collagen production (types I and III) and incorporated more 5-bromodeoxyuridine and TUNEL staining compared with unstimulated controls.
Conclusions— Stimulation of EPC-seeded tissue engineering scaffolds with TGF-β1 in vitro resulted in a more organized cellular architecture with glycoprotein, collagen, and elastin synthesis, and thus noninvasively isolated EPCs coupled with the pleiotropic actions of TGF-β1 could offer new strategies to guide tissue formation in engineered cardiac valves.
Eight of 1000 infants are born with congenital heart disease (www.americanheart.org). Of these patients, 15% to 20% require pulmonary valve replacement and/or right ventricular outflow tract (RVOT) reconstruction. Although clinically approved homograft valves are initially acceptable, they cannot grow with pediatric patients and yield a high frequency of reoperations. Thus, we have undertaken a tissue engineering (TE) approach toward cardiovascular replacement tissues. Our group has demonstrated encouraging results in animals culminating in single leaflet1 and trileaflet heart valve2 replacements using vascular wall-derived cells. Toward clinical applications, we are investigating less invasively accessible cells, including bone marrow-derived mesenchymal stem cells and peripheral blood-derived endothelial progenitor cells (EPCs).3,4
Evidence suggests that subsets of endothelial cells (ECs) can transdifferentiate to a mesenchymal phenotype. Studies using mature ECs from bovine aortic and pulmonary artery, ovine pulmonary valve, and human umbilical vein5–7 demonstrated α–smooth muscle actin (SMA) expression in response to growth factors such as transforming growth factor (TGF)-β1 or reduced concentration of serum. EPC respond similarly to TGF-β1 on TE scaffolds.8 Hence, EPCs are conceptually appealing for use in constructing autologous TE heart valves, because they could potentially provide interstitial and endothelial functions from a single cell source. In this study, we investigated whether EPCs can secrete extracellular matrix (ECM) when seeded onto synthetic polymer scaffolds in response to TGF-β1.
Animal care procedures were approved by the Animal Care Committees of the Children’s Hospital Boston and Harvard Medical School.
Isolation and Culture of EPCs
EPCs were isolated from 30 mL of heparinized (2.5 mL of 1000 IU/mL) ovine (age, 4 weeks; weight, 10 to 15 kg) peripheral blood as described.3 The leukocyte fraction was obtained by Histopaque gradient (density, 1.077 g/mL; Sigma) centrifugation for 20 minutes at 2000g. Isolated mononuclear cells were transferred to EBM-2 medium supplemented with EGM-2 SingleQuots (Clonetics), antibiotics, and 20% FBS (Hyclone) but without hydrocortisone, and placed on a 6-well plate coated with 1% gelatin (Sigma) for 24 hours and then transferred to new gelatin-coated plates. After 4 days in culture, nonadherent cells were removed and the culture maintained through days 7 to 10. Ovine vascular ECs (VEC) and smooth muscle cells (SMCs) were prepared as described previously.2 At &80% confluence, cells were harvested with 0.05% trypsin-EDTA (Gibco) and replated in 150-cm2 polystyrene dishes (Corning) at 104 cells/cm2. After 3 to 4 weeks, significant cell counts necessary for seeding were obtained. Cells were characterized before seeding.
Preparation of the Tissue Engineered Constructs
Nonwoven polyglycolic acid scaffolds (thickness, 1.0 mm; specific gravity, 69 mg/cm3; Albany International) were cut into 1-cm2 pieces, dip coated for 30 seconds in 1% wt/vol poly-4-hydroxybutyrate (P4HB; molecular weight, 1×106 PHA 4400; Tepha) in tetrahydrofuran, and cold gas sterilized with ethylene oxide. Dynamic rotational seeding and culturing in a laminar flow fluid system were performed as described.9 EPCs (2×106 per cm2 and 15×106 per cm2; passages 4 to 6) were seeded onto scaffolds (n =5) in a disposable scintillation vial (Wheaton, 20 mL) rotated at 6 rpm at 37°C for 5 days. Culture medium (10 mL) was changed after 4 hours then every 8 to 12 hours thereafter. After 5 days of dynamic seeding, seeded scaffolds were cultured in a polystyrene roller bottle (Corning, 850 cm2) rotated at 0.2 cycles/min at 37°C for 5 and 10 days. Medium was changed based on the pH (pH was maintained at pH >7.0), and air was displaced with 5% CO2 (compressed gas) for 40 seconds. Endothelial-to-mesenchymal transdifferentiation was induced with 10 ng/mL of TGF-β15,8 for 5 days.
Analysis of EPC Cultures and Tissue Engineered Constructs
First passage and seeded cells were plated into 2-well chamber slides and fixed with −20 °C methanol for 20 minutes as described.3,5 Slides were blocked with 1% BSA in PBS for 30 minutes and incubated with primary antibodies against CD31 (1:1000, Santa Cruz), von Willebrand factor (1:500, DAKO), and α-SMA (1:2000, Sigma) followed by species-specific fluorescein-conjugated secondary antibodies (Vector Laboratories). Slides were examined and photographed under a fluorescence microscope (Nikon Eclipse TE2000). Ovine vascular ECs and SMCs were prepared and stained in the same manner and used as positive and negative controls, respectively.
Histology and Immunohistochemistry
Histological analysis and characterization of cell phenotypes by immunohistochemistry were performed as described.3,10 Representative portions of scaffolds were fixed in 10% buffered formalin and embedded in paraffin. Serial sections (6 μm) were stained with hematoxylin/eosin for morphology and antibodies specific for α-SMA (Sigma) and CD31 (DAKO) for cell phenotype and collagen types I and III (Novus Biologicals). Immunostaining was performed by the Avidin-Biotin-peroxidase-Complex (ABC) method with biotin-labeled secondary antibodies (Vector Laboratories) and using 3-amino-9-ethyl carbazole (DAKO) as a substrate. Sections were counterstained with Gill’s hematoxylin solution (Sigma). For specificity controls, an appropriate nonimmune IgG was substituted for the primary antibody.
Scanning Electron Microscopy
Samples were fixed in 2.5% (v/v) glutaraldehyde (Sigma) and 1% (wt/vol) osmium tetraoxide (Sigma), dehydrated in graded series of ethanol (Sigma), dried and sputter coated with gold-palladium alloy (&3 nm thickness; Cressington Scientific Instruments). Each scaffold sample was imaged at 5 kV on a JEOL JSM-6330F field emission Scanning electron microscopy (JEOL USA).
Measurement of Cellular Proliferation and Apoptosis
The EPC seeded scaffolds were labeled in vitro with 5-bromodeoxyuridine ([BrdU] Sigma, 16 ng/mL per day) for 3 days before the end of incubation.11 BrdU and TUNEL followed the manufacturer’s protocols (BrdU Staining kit, BD Biosciences; Apop Tag Peroxidase In Situ Apoptosis Detection kit, Intergen) with diaminobenzidine as a chromogen. Sections were counterstained with methyl green solution (DAKO). For negative controls, BrdU antibody and terminal deoxynucleotidyltransferase were omitted from the labeling mixture. Cells containing dark nuclear BrdU staining were considered to be BrdU+ and to have undergone DNA synthesis during labeling. Apoptotic nuclei were defined as TUNEL+ nuclei in cells with morphological features of apoptotic cell death (cell shrinkage, aggregation of chromatin into dense masses, and cell fragmentation). Overall proliferation and apoptosis data were obtained by counting the number of BrdU+ and TUNEL+ cells out of the total cell number in histological sections.
Cells were lysed with 4 mol/L urea, 0.5% sodium dodecyl sulfate, 0.5% Nonidet P-40, 100 mmol/L Tris, and 5 mmol/L ethylenediaminetetraacetic acid (pH 7.4; Boston Bioproducts, Inc), containing protease inhibitors (Roche). Protein concentrations were determined by the manufacturer’s protocol (BCA Protein Assay Reagent kit, Pierce Biotechnology), and immunoblotting of cells and seeded scaffold extracts (n =5) using CD31 (1:500, Santa Cruz), α-SMA Clone 1A4 (1:1000, Sigma), and α-Tubulin Clone DM 1A (1:500, Sigma) were performed as described.5,8 Detection of ECM protein expression in seeded scaffolds was performed using polyclonal antibodies: rabbit anti-human collagen types I and III (1:2500, Novus Biologicals,), fibronectin (1: 5000, Novus Biologicals), laminin (1:5000, Chemicon), tropoelastin (1:1000, Elastin Products), and elastin monoclonal Clone BA-4 (1:500, Sigma).
Quantitative Biochemical Matrix Analysis
Quantification of total cellular DNA was performed as described.12 Representative portions of scaffolds were weighed and extracted in a solution of 0.125-mg/mL papain and 10-mmol/L l-cysteine dihydrochloride (Sigma) in phosphate-buffered EDTA (Sigma) in a 60°C water bath for 10 hours. Total DNA was measured according to the manufacturer’s protocol (Picogreen dsDNA Quantitation kit, Molecular Probes). Collagen and sulfated glycosamino-glycan (S-GAG) content were assayed as described.12 Representative portions of scaffolds were weighed. Total collagen was extracted in 0.5-mol/L acetic acid and pepsin (1 mg/mL Pepsin A, P-7000). Proteoglycans and S-GAG were extracted in 4 mol/L of guanidine-HCl and 0.5 mol/L of sodium acetate. Samples were rotated at 4°C overnight. Total collagen and S-GAG content were measured according to the manufacturer’s protocol (Sircol and Blyscan assay kits, Biocolor Ltd).
All of the results are expressed as mean ± SEM. Comparisons between groups were made with a Student’s t test (Sigma Stat, Jandel Scientific). If measurements failed the normality test, the nonparametric Mann-Whitney rank sum test was used. A P value ≤ 0.05 was considered significant.
Statement of Responsibility
All 6 authors have full access to the data and take full responsibility for its integrity. We have read and agree to the article as written.
Circulating EPCs Demonstrate Endothelial Phenotype Before Seeding
The isolated EPCs exhibited clear endothelial morphology (cobblestone monolayer) as described previously3,5,8 and confirmed by immunostaining with antibodies specific for endothelial markers (Figure 1a and 1b). Immunofluorescence of the isolated pure EPC revealed expression of CD31 at the cell-cell membrane borders and von Willebrand factor in cytoplasmic granules similar to Weibel palade bodies (Figure 1a and 1b). No immunoreactivity was observed for α-SMA in EPCs (Figure 1c) or for isotype-matched control IgG (Figure 1d and 1e). Ovine vascular SMCs serve as control for SMC markers (Figure 1f). To further characterize the endothelial phenotype, whole cell extracts were run on SDS-PAGE, immunoblotted, and confirmed to express CD31 and were negative for α-SMA at appropriate molecular weights (Figure 1g). Ovine carotid artery-derived SMCs and ECs served as controls, demonstrating the appropriate presence and absence of immunoreactivity for α-SMA and CD31 markers, respectively.
EPC-TGF-β1 Scaffolds Reveal Dense Tissue Formation and Confluent Smooth Surfaces
Scanning electron microscopy examination was performed on scaffolds seeded with EPCs (Figure 2a), EPC seeded scaffolds treated with TGF-β1 (Figure 2b), and unseeded controls (Figure 2c). EPC-TGF-β1 scaffolds demonstrated a smooth and confluent surface layer (Figure 2b). In contrast, a less homogenous construct with isolated individual cells attached to polymer fibers was seen in the unstimulated controls (Figure 2a).
Enhanced Tissue Formation in the EPC-TGF-β1 Scaffolds
Expression of CD31 and α-SMA in EPC-TGF-β1 Scaffolds
Immunohistochemical staining demonstrated CD31 (Figure 4a and 4b) and α-SMA (Figure 4c through 4e) positive cells throughout the EPC-TGF-β1 scaffolds. Lysates of the tissue samples were run on SDS-PAGE, immunoblotted, and showed CD31 and α-SMA at appropriate molecular weights (Figure 4f). Tubulin served as a loading control.
Effect of EPC-TGF-β1 Scaffolds on DNA, S-GAG, and Collagen Content
Between days 10 and 15 of incubation, the total collagen content in EPC-TGF-β1 scaffolds changed substantially. In contrast, no collagen content was detected at day 10 both in EPC-TGF-β1 scaffolds and unstimulated EPC controls by biochemical assay, immunoblot, and immunostaining (data not shown). However, total collagen content is significantly higher at day 15 in EPC-TGF-β1 scaffolds compared with the unstimulated EPC controls (Figure 5a). Both collagens (types I and III) were detected in both stimulated and unstimulated scaffolds (Figure 5b). However, EPC-TGF-β1 scaffolds demonstrated more enhanced collagen expression compared with unstimulated EPC controls (Figure 5b). Immunostaining of both scaffolds revealed collagen (types I and III) predominately in the “interstitium” of the scaffold with lesser immunoreactivity in the surface layer of the scaffold.
Effect of EPC-TGF-β1 Scaffolds on ECM Production
Expression of fibronectin and laminin were higher in EPC-TGF-β1 scaffolds compared with unstimulated scaffold samples (20- and 2.5-fold; P<0.05 versus control), although expression of tropoelastin was similar in the EPC-TGF-β1 scaffolds and in unstimulated EPC controls (Figure 6).
EPC-TGF-β1 Scaffolds Demonstrate Enhanced Proliferation and Apoptosis
We assessed the BrdU and TUNEL labeling index in these TE constructs. Polyglycolic acid/poly-4-hydroxybutyrate composite scaffolds were seeded with either EPCs at either 2×106 per cm2 or 15×106 per cm2 at days 10 and 15, respectively. Immunohistochemical analysis of TE constructs revealed that BrdU incorporation during day 10 did not differ significantly between EPC-TGF-β1 scaffolds and unstimulated controls (Figure 7a). By day 15, BrdU incorporation in EPC-TGF-β1 scaffolds increased significantly from that at 10 days and compared with unstimulated controls, whereas rates of incorporation into unstimulated controls was unchanged at longer incubation (Figure 7a).
At 10 days, rates of apoptosis in TE constructs assessed by TUNEL staining were virtually identical between scaffolds (EPC-TGF-β1 scaffolds versus unstimulated controls). At 15 days, the frequency of apoptotic cells in EPC-TGF-β1 scaffolds was more than 6 times (P<0.05) that in unstimulated controls (Figure 7b). After 15 days of incubation, the frequency of apoptotic cells in EPC-TGF-β1 scaffolds and unstimulated controls exceeded those at 10 days, and both had increased similarly (13.8 versus 2.3) (Figure 7b).
Current homografts or prosthetic grafts used in RVOT reconstruction do not grow in pediatric patients, ultimately contributing to a high frequency of reoperation. In attempting to develop clinically feasible TE biomaterials for use in pediatric RVOT reconstruction, we investigated the ability of noninvasively isolated EPCs to secrete ECM on scaffolds in response to TGF-β1. Investigators demonstrated previously that EPCs can undergo apparent phenotypic change from endothelial toward a mesenchymal cell type in response to TGF-β1 on scaffolds.8 The current study confirmed that EPCs can undergo phenotypic change to a cell type that has some characteristics of valvular interstitial cells,10 with ECs and SMC-like cells throughout the surface and interstitium of the scaffold. Of potential functional consequence, stimulation of EPC-seeded scaffolds with TGF-β1 also resulted in significant ECM production.
The ECM composition determines the biomechanical characteristics of TE cardiovascular constructs. The quantitative ECM analysis of both unstimulated and TGF-β1–stimulated scaffolds at days 10 and 15 demonstrated values slightly lower but generally consistent with TE structure-based arterial wall cells.12 In both experiments, DNA content of the EPC-TGF-β1 scaffolds exhibited a trend toward increased values when compared with unstimulated EPC-seeded scaffolds. Consistent with known pleiotropic actions of TGF-β1, stimulated scaffolds at longer incubation and increased cell density revealed enhanced collagen production (types I and III), incorporated more BrdU, and had more TUNEL staining, indicating an increase in proliferation and apoptosis, respectively, and providing a likely mechanism for the continuing process of tissue remodeling and maturation in vitro. Interestingly, S-GAG production that is thought to be regulated by TGF-β1 at day 10 demonstrated lower levels compared with the unstimulated cells. However, at day 15, there was a trend toward increased S-GAG production in the TGF-β1–stimulated scaffolds. Therefore, it is possible that a larger number of TE constructs per group may have revealed a significant difference. Moreover, EPC-secreted tropoelastin and adhesive glycoproteins (laminin and fibronectin) are appropriate for semilunar valve tissue.
What cell type participates in ECM formation in EPC-TGF-β1 scaffolds in vitro? Previous reports established relationships between α-SMA expression and ECM synthesis in vascular SMCs.13 Whether the presence of collagen, elastin, and glycoproteins in EPC-TGF-β1 scaffolds stems directly from SMC-like cells or indirectly from the paracrine effect of TGF-β1 on the SMC-like cells remains unclear. Because mechanical stimulation has been shown to elicit enhanced expression of TGF-β1 in other mesenchymal cell types, further studies are warranted to determine the effects of biomechanical stimulation (eg, via a cyclic flexure bioreactor12). Collectively, the results of this study suggest that EPCs coupled with the pleiotropic actions of TGF-β1 may offer a strategy for enhancing tissue formation in engineered cardiac valves.
There are several limitations within the present study. We did not perform any purification of primary EPC cultures by flow cytometry. We isolated the EPCs using 2 different methods: for the first set of experiments at day 10, single cells were derived from clonal populations of ovine peripheral blood; at day 15, they were derived from a single population. We partially purified this by Histopaque gradient centrifugation as described previously by our group.3 We confirmed the endothelial phenotype of these cells derived from 2 different methods with expression of CD31 and von Willebrand factor and no detection of α-SMA. Previous studies by our group and others demonstrated that clonal populations of native valvular endothelial cells and circulating EPCs expressed α-SMA in response to TGF-β1 or reduced concentrations of serum.5,8 On the basis of the data on the characterization of circulating EPCs in our study using immunofluorescence and immunoblotting, we can predict that our primary EPC cultures are devoid of mesenchymal/fibroblast like cells. Second, we were unable to quantify the collagen types (I and III) in our constructs. However, we believe that our results showing enhanced collagen expression in EPC-TGF-β1 scaffolds compared with unstimulated scaffolds based on our qualitative detection will likely result into a significant collagen types (I and III) content. Third, as published previously,12 the S-GAG values presented herein represent the combined amounts of free and proteoglycan-bound S-GAG and thus provide an indicator of ECM production. Although certain proteoglycans (eg, small leucine-rich proteoglycans) are known to modulate ECM mechanical properties, the focus of the current study was simply to determine whether EPCs have the capacity to produce ECM. However, further studies on the exploration and quantitation of various types of S-GAGs and collagen produced are warranted.
The effect of TGF-β1 on remodeling is highly relevant to potential techniques in the clinical situation, which could be used in developing TE cardiovascular structures by affecting the rate of production of ECM components. This type of intervention could allow more precise engineering of these types of structures.
We acknowledge important advice and contributions of Drs David Martin (Tepha Inc) and Marsha Moses (Children’s Hospital Boston), excellent technical assistance of Tonora Archibald (Children’s Hospital Boston), and editorial expertise of Dr Michael Blaze.
Sources of Funding
This work was supported by National Institutes of Health grants HL-60463 (J.E.M.), HL-68816 (M.S.S.), and Tepha/National Institute of Standards and Technology grant NANB2H3053 (J.E.M.); Gross Cardiovascular Fund (J.E.M.); Center for Integration of Medicine and Innovative Technology (CIMIT) (J.E.M); and National Research Service Award (NRSA)/National Institute of Biomedical Imaging and Bioengineering (NIBIB) grant (F32 EB003353-01 to B.A.M.). G.C.E. was supported by American Heart Association Predoctoral Fellowship 0415406U, PA/DE Affiliate. J.A.J. is a recipient of a Summer Fellowship from the American Heart Association, Northeast Affiliate.
Presented at the American Heart Association Scientific Sessions, Dallas, Tex, November 13–16, 2005.
Hoerstrup SP, Sodian R, Daebritz S, Wang J, Bacha EA, Martin DP, Moran AM, Guleserian KJ, Sperling JS, Kaushal S, Vacanti JP, Schoen FJ, Mayer JE Jr. Functional living trileaflet heart valves grown in vitro. Circulation. 2000; 102: III44–III49.
Sutherland FW, Perry TE, Yu Y, Sherwood MC, Rabkin E, Masuda Y, Garcia GA, McLellan DL, Engelmayr GC Jr, Sacks MS, Schoen FJ, Mayer JE Jr. From stem cells to viable autologous semilunar heart valve. Circulation. 2005; 111: 2783–2791.
Paranya G, Vineberg S, Dvorin E, Kaushal S, Roth SJ, Rabkin E, Schoen FJ, Bischoff J. Aortic valve endothelial cells undergo transforming growth factor-beta-mediated and non-transforming growth factor-beta-mediated transdifferentiation in vitro. Am J Pathol. 2001; 159: 1335–1343.
Frid MG, Kale VA, Stenmark KR. Mature vascular endothelium can give rise to smooth muscle cells via endothelial-mesenchymal transdifferentiation: in vitro analysis. Circ Res. 2002; 90: 1189–1196.
Ishisaki A, Hayashi H, Li AJ, Imamura T. Human umbilical vein endothelium-derived cells retain potential to differentiate into smooth muscle-like cells. J Biol Chem. 2003; 278: 1303–1309.
Dvorin EL, Wylie-Sears J, Kaushal S, Martin DP, Bischoff, J. Quantitative evaluation of endothelial progenitors and cardiac valve endothelial cells: proliferation and differentiation on poly-glycolic acid/poly-4-hydroxybutyrate scaffold in response to vascular endothelial growth factor and transforming growth factor beta1. Tissue Eng. 2003; 9: 487–493.
Matsubayashi K, Fedak PW, Mickle DA, Weisel RD, Ozawa T, Li, RK. Improved left ventricular aneurysm repair with bioengineered vascular smooth muscle grafts. Circulation. 2003; 108 (suppl 1): II219–II225.