Elastic Fiber Production in Cardiovascular Tissue-Equivalents

Matrix Biology 22 (2003) 339–350

Elastic fiber production in cardiovascular tissue-equivalents

Jennifer L. Long, Robert T. Tranquillo*

Department of Chemical Engineering & Materials Science and Department of Biomedical Engineering, 7-114 BSBE, 312 Church St SE, University of Minnesota, Minneapolis, MN 55455, USA

Received 10 January 2003; received in revised form 30 April 2003; accepted 30 April 2003

Abstract

Elastic fiber incorporation is critical to the success of tissue-engineered arteries and heart valves. Elastic fibers have not yet been observed in tissue-engineered replacements fabricated in vitro with smooth muscle cells. Here, rat smooth muscle cells (SMC) or human dermal fibroblasts (HDF) remodeled collagen or fibrin gels for 4 weeks as the basis for a completely biological cardiovascular tissue replacement. Immunolabeling, alkaline extraction and amino acid analysis identified and quantified elastin. Organized elastic fibers formed when neonatal SMC were cultured in fibrin gel. Fibrillin-1 deposition occurred but elastin was detected in regions without fibrillin-1, indicating that a microfibril template is not required for elastic fiber formation within fibrin. Collagen did not support substantial elastogenesis by SMC. The quantity of crosslinked elastic fibers was enhanced by treatment with TGF-b1 and insulin, concomitant with increased collagen production. These additives overcame ascorbate’s inhibition of elastogenesis in fibrin. The elasticfibers that formed in fibrin treated with TGF- b1 and insulin contained crosslinks, as evidenced by the presence of desmosine and an altered elastin labeling pattern when b-aminopropionitrile (BAPN) was added. These findings indicate that in vitro elastogenesis can be achieved in tissue engineering applications, and they suggest a physiologically relevant model system for the study of three-dimensional elastic structures. ᮊ 2003 Elsevier B.V.yInternational Society of Matrix Biology. All rights reserved.

Keywords: Elastin; Fibrillin; Tissue engineering; Fibrin; Smooth muscle cell

1. Introduction component. Approaches to date have included building tissue replacements on an elastin scaffold isolated from Elastic fibers are a critical structural component in cadaveric tissues (Berglund et al., 2001), supplying cardiovascular tissues. Present in arteries and heart soluble tropoelastin to a cell culture (Stone et al., 2001), valves, they dictate mechanics at low strains before and designing biocompatible synthetic elastic polymers stiffer collagen fibers are engaged (Bank et al., 1996; (Urry and Pattanaik, 1997). Evaluation of spontaneous Vesely, 1998). They also confer elasticity, preventing elastogenesis in tissue-engineered constructs has been dynamic tissue creep by stretching under load then limited. A rolled tube of vascular cell-derived matrix recoiling to their original configurations after load demonstrated elastin immunolabeling after three months release (Ross and Bornstein, 1971). Elastin knock-out of culture, but in the internal interfaces of the fibroblastic studies and clinical observations have revealed an essen- adventitial layer rather than within the muscular medial tial regulatory function during artery development. In layer (L’Heureux et al., 1998). Smooth muscle cells the absence of extracellular elastin accumulation, smooth produced significantly less elastin mRNA in a non- muscle proliferation stenoses arteries (Karnik et al., fibrillar collagen sponge than in a poly-glycolic acid 2003; Urban et al., 2002). scaffold (Kim et al., 1999). Transplanted or host cells To ensure appropriate mechanical function and pre- may deposit elasticfibers after tissue-engineered con- vent these serious complications, successful cardiovas- struct implantation (Berthod et al., 2001; Shum-Tim et cular tissue replacements must incorporate an elastic al., 1999; Stock et al., 2001), but this necessitates in *Corresponding author. Tel.: q1-612-625-6868; fax: q1-612-626- vivo maturation and precludes in vitro study of elasto- 6583. genesis. Elastic fiber formation by vascular cells E-mail address: [email protected] (R.T. Tranquillo). entrapped within three-dimensional fibrillar collagen or

0945-053X/03/$30.00 ᮊ 2003 Elsevier B.V.yInternational Society of Matrix Biology. All rights reserved. doi:10.1016/S0945-053XŽ03.00052-0 340 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 fibrin culture has not been shown to our knowledge. SMC (Merrilees et al., 2002). Beneficial proteoglycans Those two biopolymers are studied here as the basis for may protect elastin binding protein function as it shuttles a completely biological arterial media substitute. tropoelastin to nascent elastic fibers. Disruption of this Traditional monolayer cell culture studies have iden- system interferes with elastogenesis in multiple clinical tified several factors that influence elastogenesis. Tro- conditions (Hinek et al., 2000; Hinek and Wilson, poelastin expression in the strongly elastogenicaortic 2000). The recent studies illustrate the complexity of smooth muscle cell (SMC) is developmentally regulat- elastogenesis, which is now understood to encompass ed, with maximal secretion by cells freshly isolated from gene transcription, post-transcriptional regulation, and the neonate and virtually none by adult cells (Johnson coordinated assembly of multiple molecules within a et al., 1995; McMahon et al., 1985). Inhibitors include receptive extracellular milieu. Despite the increasing bFGF, which decreases elastin mRNA transcription appreciation of extracellular elastogenesis modulation, (Carreras et al., 2002), and ascorbate, which destabilizes the basic question of which environmental scaffolds elastin mRNA while improving collagen mRNA stability promote elastogenesis has not been previously (Davidson et al., 1997). TGF-b1 and insulin-like growth addressed. We compare collagen and fibrin scaffolds, factor-1 both increase tropoelastin mRNA and protein measuring elasticfibers directly, rather than intermedi- by various cultured cells (Davidson et al., 1993; Sauvage ates such as RNA, to account for the many steps in et al., 1998; Wolfe et al., 1993). TGF-b1 has also been mature elasticfiber production. found to upregulate the elastin gene promoter in genet- Monolayer cell culture has been informative, and can ically modified mice (Katchman et al., 1994) and to produce structures many cell layers thick. However, increase lysyl oxidase activity in cultured cells (Shanley reproducing the organization of native tissues is chal- et al., 1997). Lysyl oxidase crosslinks tropoelastin into lenging, if not impossible, with the monolayer approach. its mature fully functional form, but can be inhibited by This work demonstrates and quantifies elastogenesis in nitricoxide donors suchas b-aminopropionitrile (Smith- a totally biological tissue replacement that can mimic Mungo and Kagan, 1998). the characteristic alignment of native tissues. This align- Dermal fibroblast elastogenesis has also been studied ment confers the mechanical strength of tissue and extensively due to elastin’s importance in skin. Its provides a template for organized extracellular matrix regulation by soluble molecules is similar to that of deposition. We have previously reported that SMC SMC (Kahari et al., 1992). Unlike SMC, dermal fibro- compact tubes of fibrin gel around a non-adhesive blasts maintain elastin mRNA synthesis until mid-adult- mandrel, producing strong circumferential alignment of hood (Fazio et al., 1988) but deposit little insoluble both the initial fibrin and the subsequent cell-produced elastin (Narayanan et al., 1976). This discrepancy may collagen. The alignment in that case matches that of the be due to fibroblast incompetence in depositing elastic arterial medial layer (Grassl et al., 2002). Constructs fibers. A skin substitute co-culturing keratinocytes and prepared by constrained cell-induced compaction of fibroblasts on a collagen-GAG-chitin sponge produced highly hydrated and entangled native protein fibril net- elastin and fibrillin-1, while cultures of fibroblasts alone works (as depicted in Fig. 1) are termed ‘tissue-equiv- did not (Duplan-Perrat et al., 2000). Because of their alents’ (Tranquillo, 1999). While this study uses accessibility and robust collagen production, dermal disc-shaped tissue-equivalents for simplicity, the method fibroblasts have also been used in cardiovascular tissue can be extended to more relevant geometries for arteries engineering including heart valve-equivalents (Neidert and valves as has been reported by this laboratory et al., 2003). Elasticfiber productionin biopolymer (Grassl et al., 2002, 2003; Neidert et al., 2003).We culture without keratinocytes is critical for the valve show here that the organized elasticfibers are produced application, so is examined here. in this system in addition to substantial collagen, thus Extracellular matrix (ECM) likely influences elasto- approaching the ECM composition of small diameter genesis since the vasculature develops within a three- arteries. dimensional matrix rather than from a monolayer of Important regulators identified in the traditional cell cells. In one series of studies, cells cultured on decel- culture studies described above are examined in the lularized elastin-containing tissue layers incorporated three-dimensional gel culture system for tissue-equiva- new elastin into the existing matrix, producing more lent formation. Neonatal and adult SMC are compared, than on tissue-culture plastic alone (Mecham, 1981; TGF-b1 stimulation is tested, and ascorbate’s influence Parks et al., 1988). Numerous recent papers address the is studied. As indicated earlier, dermal fibroblasts are role of microfibrillar proteins and microfibril-associated also studied because of their noteworthy differences glycoproteins in elastic fiber assembly (Hill et al., 2002; from SMC in traditional culture and because they may Nakamura et al., 2002; Penner et al., 2002; Yanagisawa be used in cardiovascular tissue-equivalents. Each factor et al., 2002). ECM proteoglycans not associated with is examined in both type I collagen and fibrin. This microfibrils also play a role; overexpression of a versi- laboratory found that insulin, plasmin and TGF-b1 can variant has been shown to increase elastogenesis in synergistically increased collagen synthesis by human J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 341

Fig. 1. Schematic diagram of tissue-equivalent preparation and maturation. (a) Tissue cells, (e.g. smooth muscle cells or fibroblasts) are mixed with solubilized biomonomers (either pepsin-digested type I collagen or fibrinogen). (b) Addition of NaOH (to collagen) or thrombin (to fibrinogen) induces fibrillogenesis, entrapping the cells within a hydrated network of entangled fibrils. (c) Over days, cells contract the network, aligning it in response to applied mechanical constraints. (d) Over weeks, cells remodel the network, resorbing the original biopolymer and depositing cross-linked collagen fibrils and elasticfibers.

Fig. 2. Brightfield images of neoSMC tissue-equivalent sections after four weeks standard culture with 10% FBS, and ACA in fibrin. (A) Fibrin gel culture; and (B) collagen gel culture. Cell nuclei, visualized Fig. 3. Immunolabeling in fibrin-neoSMC tissue-equivalents shows with DAPI under UV light, are overlain on corresponding brightfield elastin and fibrillin-1 deposition. All were treated with 10% FBS and ( ) images. Bars50 mm. ACA. A, B, E, F, G, H additionally treated with TGF-b1 and insulin; (C, D) no additional treatment; (E, F) also treated with BAPN. (A, dermal fibroblasts (Neidert et al., 2002); thus, these C, E) rabbit anti-rat elastin antibody; (B, D, F): mouse anti-rat fibril- ( ) ( ) three molecules were combined in the fibroblast studies lin-1 antibody; G rabbit nonspecific IgG control; and H mouse nonspecific IgG control. Bars50 mm. here. Because neoSMC more aggressively degrade fibrin, plasmin was omitted from neoSMC cultures and natal rat SMC (neoSMC) are shown in Fig. 2. Both the fibrinolysis inhibitor ´-aminocaproic acid was added. types of tissue-equivalent demonstrated strong contrac- The influence of lysyl oxidase was also studied by both tion from their original thicknesses of approximately 1 quantifying desmosine crosslinks and inhibiting cross- mm. Cells remained distributed throughout the tissue- linking with b-aminopropionitrile. equivalent thickness, ultimately on the order of 100 mm. 2. Results Tissue-equivalents formed with the other cell types studied also contracted dramatically and with homoge- 2.1. Elastic fiber immunolabeling in tissue-equivalents neous cell distribution. Brightfield images of sections from collagen and Immunofluorescence on similar sections demonstrated fibrin tissue-equivalents cultured for 4 weeks with neo- elastin and fibrillin-1 in fibrin constructs (Fig. 3). Elastin 342 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350

Gel culture of neonatal human dermal fibroblasts (neoHDF) demonstrated fibrillar labeling of both elastin and fibrillin-1 (Fig. 6). Elastin appeared fibrous throughout the fibrin thickness, although large deposits noted in fibrin-neoSMC tissue-equivalents were absent (panel 6A). Fibrillin-1 was more limited spatially, appearing only in a few parallel bands (panel 6C). NeoHDF in collagen produced more and better-organized elastin and fibrillin-1 than neoSMC in collagen. Labeling for both proteins appeared fibrillar (panel 6B and 6D) although

Fig. 4. Immunolabeling in collagen-neoSMC tissue-equivalents shows little elastin or fibrillin-1 deposition. All were treated with 10% FBS, TGF-b1, and insulin. (A) Rabbit anti-rat elastin antibody; (B) mouse anti-rat fibrillin-1 antibody; (C) rabbit nonspecific IgG control; and (D) mouse nonspecific IgG control. Bars50 mm. and fibrillin-1 antibodies labeled throughout the entire thickness of tissue-equivalents treated with TGF-b1 and insulin (panels 3A and 3B). Elastin labeling was less pronounced in untreated tissue-equivalents (panels 3C and 3D). The lysyl oxidase inhibitor BAPN qualitatively decreased elastin while increasing fibrillin-1 (panels 3E and 3F). BAPN produced a more homogeneous elastin appearance, but still more pronounced than in controls labeled with non-specific IgG. In all cases non-specific IgG produced minimal labeling clearly distinguished from elastin or fibrillin-1 antibody labeling. Somewhat increased rabbit IgG (control for elastin antibody) background was noted in constructs treated with TGF-b1 and insulin (panel 3G) vs. the untreated case which produced no labeling. Mouse IgG (control for fibrillin- 1 antibody) produced no labeling in any case (panel 3H). Other SMC culture cases did not produce mature elasticfibers of the type seen in Fig. 3. Collagen gel culture of neoSMC greatly decreased elastin and fibrillin-1 immunolabeling relative to fibrin gel culture (Fig. 4). Although elastin was detected in collagen-neoSMC tissue-equivalents (panel 4A), only part of the tissue- equivalent thickness was labeled and the pattern was amorphous. Likewise, fibrillin-1 labeling was question- able and did not appear fibrous (panel 4B). Adult rat SMC (aSMC) also produced minimal elastic fibers (Fig. Fig. 5. Immunolabeling in aSMC tissue-equivalents shows little elas- 5). In both fibrin (panels 5A through 5D) and collagen tin or fibrillin-1 deposition. All were treated with 10% FBS, TGF-b1, ( ) and insulin (plus ACA for fibrin). (A–D) fibrin; (E–H) collagen. (A, panels 5E through 5H , elastin and fibrillin-1 labeling E) rabbit anti-rat elastin antibody; (B, F) mouse anti-rat fibrillin-1 appeared clumpy or disorganized and was less clearly antibody; (C, G) rabbit non-specific IgG control; and (D, H) mouse distinguished from non-specific IgG control labeling. nonspecific IgG control. Bars50 mm. J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 343

and fibrin neoSMC tissue-equivalents with TGF-b1 and insulin, closely matches that reported for rat aortic elastin (Starcher and Galione, 1976). Glycine, alanine and proline together comprised 66% of the amino acids extracted from fibrin and 65% of those extracted from collagen, consistent with the assertion that these non- polar amino acids contribute two-thirds of human aortic elastin residues (Labella and Vivian, 1967). Hydroxy- proline, while a known minor component of elastic fibers, was not resolved in the amino acid analysis performed here. TGF-b1 and insulin were required to produce this classic profile; omitting the growth factors increased the proportion of polar amino acids. Following extraction and hydrolysis, elastic fibers were quantified by a colorimetric reaction of free amine groups with ninhydrin. Fig. 8 shows elastin measured in 4-week-old tissue-equivalents with rat SMC, each of equal initial volume and gel protein content of approximately 1 mg. Elastin was increased by using neonatal cells rather than adult, culturing in fibrin rather than collagen, treating with TGF-b1 and insulin, and when all those conditions were met, by increasing the number of cells. Elastin production was not proportional to initial cell number when the other conditions were not optimized. Furthermore, significantly decreased elastin was noted in neoSMC-collagen tissue-equivalents with the highest initial cell number. Those constructs contracted to approximately 1y10 of their original volume within three days, whereas all other cases contracted more modestly. Hydroxyproline content, proportional to the collagen Fig. 6. Immunolabeling in neoHDF tissue-equivalents shows elastin deposited in fibrin, was measured by oxidation following and fibrillin-1 deposition. All were treated with 10% FBS, TGF-b1, hydrolysis of the non-elastin fraction of tissue-equiva- insulin, and plasmin. (A, C, E) neoHDF in fibrin; (B, D, F) neoHDF lents. Hydroxyproline data are shown in Fig. 8 for the ( ) ( ) in collagen. A, B Mouse anti-human elastin antibody; C, D Mouse same tissue-equivalents as the elastin data. Regulation anti-human fibrillin-1 antibody; and (E, F) Mouse nonspecific IgG control. Bars50 mm. background labeling from non-specific IgG was higher in collagen than in fibrin (panel 6F vs. panel 6E). To summarize the immunolabeling results, neoSMC in fibrin with TGF-b1 and insulin produced the most dramaticevidenceof elastin and fibrillin-1. Little elastin or fibrillin-1 was detected in collagen with neoSMC or in any aSMC culture. In contrast, neoHDF produced elastin and fibrillin-1 in both collagen and fibrin gels. Elastin extended beyond fibrillin-1 in neoHDF-fibrin cultures, although those labeling areas overlapped in neoSMC-fibrin tissue-equivalents.

2.2. Collagen and elastin quantification in tissue- Fig. 7. Amino acid analysis of elastic fibers. Open bars: Rat aortic equivalents elastin (Starcher and Galione, 1976). Filled bars: Elastin extracted from fibrin-neoSMC tissue-equivalents. Stippled bars: Elastin extracted from collagen-neoSMC tissue-equivalents. Tissue-equivalents were Elasticfibers were isolated from tissue equivalents by treated for four weeks with 10% FBS, TGF-b1, and insulin, plus ACA hot alkaline extraction and hydrolyzed. The resulting in fibrin. Experimental bars represent the mean of determinations from amino acid profile, shown in Fig. 7 for both collagen two separate samples. 344 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350

Fig. 8. Elastin and collagen incorporated into fibrin-SMC tissue-equivalents. Open bars: controls treated with 10% FBS and ACA. Filled bars: tissue-equivalents additionally treated with TGF-b1 and insulin. Initial cell densities are noted on the abscissa. (a, c, e) Bar equals mean of three tissue-equivalents, error bar equals standard deviation. (b, d, f): Bar equals mean of two tissue-equivalents. *: P-0.02 vs. control culture. q: P-0.02 vs. collagen gel culture. ‡: P-0.02 vs. 2E6 cellsyml initial density. §: P-0.05 vs. neoSMC. of collagen deposition differed from that of elastin 8. Overall amounts measured were comparable for HDF (panels 8E and 8F). Neither growth factor-stimulated and SMC. nor basal collagenesis differed significantly between Ascorbate treatment of neoSMC-fibrin tissue-equiva- adult and neonatal cells (P)0.05). NeoSMC in the lents mildly increased hydroxyproline content and absence of ascorbate were insensitive to growth factor decreased elastin content in a dose-independent manner stimulation of collagen production. (Fig. 10). TGF-b1 and insulin restored elastogenesis in Elastin and hydroxyproline incorporation into HDF these cultures and greatly enhanced collagenesis. Ascor- tissue-equivalents is summarized in Fig. 9. Neonatal bate treatment of neoSMC-collagen tissue-equivalents HDF did not improve elastogenesis over adult HDF, but did not achieve statistically significant differences in they did produce more hydroxyproline, contrary to the either elastin or hydroxyproline content. behavior of SMC. Growth factor treatment was ineffec- Desmosine crosslinks in elastic fibers were measured tive in fibrin but did boost elastogenesis in collagen, by radioimmunoassay. Fibrin tissue-equivalents with again in contrast to the results shown for SMC in Fig. 0.5E6 neoSMCyml and treated with TGF-b1, insulin, J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 345

ing (Grassl et al., 2002; Neidert et al., 2003). It is also a relevant in vitro model of elastogenesis in development or wound healing. Whereas some of our findings con- firm prior knowledge, noteworthy differences have been found between traditional cell culture and this more physiologicmatrix culture. In concordance with work demonstrating progressive destabilization of vascular cell tropoelastin mRNA with donor rat age (Johnson et al., 1995), neonatal vascular cells produced much more elastin than adult cells in this system. Some elastin was detected in adult cell cultures, but immunohistochemistry demonstrated that it was disorganized and hence unlikely to function optimally. Thus, adequate elastogenesis from adult donor cells is still problematic; using smooth muscle cells isolated from neonatal human umbilical arteries may be the most feasible option in cardiovascular tissue engineering. Neonatal aorticSMC producedmore elastin in fibrin gel culture than in collagen gel, building upon other work in this laboratory and elsewhere showing that type I collagen gel culture downregulates collagen production relative to fibrin gel culture (Clark et al., 1995; Grassl et al., 2002; Neidert et al., 2002). The much-improved elastogenesis in fibrin vs. collagen may explain in part the lack of reported success thus far regarding mechanical properties of collagen-based cardiovascular tissue- equivalents. Further study will clarify whether tropoelastin secretion or elastic fiber assembly is Fig. 9. Elastin and collagen incorporated into tissue-equivalents by enhanced by fibrin culture relative to collagen. Supple- HDF. Open bars: controls treated with 10% FBS. Filled bars: tissue- menting fibrin-neoSMC cultures with TGF-b1 and insu- equivalents additionally treated with TGF-b1, insulin, and plasmin. lin additionally increased elastogenesis, probably due to Bar equals mean of three tissue-equivalents; error bar equals standard multiple growth factor actions as discussed in the Intro- deviation. (a) Matrix type and cell source are noted on the abscissa. (b) All are in fibrin gel culture; cell source is noted on abscissa. *: duction. These growth factors were found to be essential P-0.02 vs. control culture. q: P-0.02 vs. collagen gel culture. §: P-0.05 vs. neoHDF. and 50 mgyml ascorbate contained 962"341 pmol desmosine per milligram total protein (mean"S.D. ns 5). Collagen gel culture of neoSMC reduced, but did not eliminate desmosine; only one construct was examined so statistical significance cannot be determined. No desmosine was detectable in aSMC tissue-equivalents or in fibrin-neoSMC tissue-equivalents additionally treated with BAPN.

3. Discussion

Studies of elastogenesis to date have been performed almost exclusively in developing organisms or in mon- Fig. 10. Elastin and hydroxyproline incorporated into fibrin-neoSMC olayer cell culture. Here we present a three-dimensional tissue-equivalents with ascorbate. Each bar equals the ratio of the matrix culture based on cell entrapment in a fibrin gel mean of 3 tissue-equivalents to the mean of 3 control tissue-equiva- in which neonatal smooth muscle cells and both neonatal lents of the same treatment but without ascorbate. Error bar equals and adult fibroblasts produce mature elastic fibers. This standard deviation. ‘FBS’ groups were treated with 10% FBS and ACA. ‘Factors’ groups were additionally treated with TGF-b1 and approach, in which the cells contract, align, degrade and insulin. Hproshydroxyproline. Ascorbate doses in mgyml: open bars replace the fibrin, allows formation of more complex zero ascorbate control, stippled bars10, filled bars50, hatched bars geometries as needed in cardiovascular tissue engineer- 100. *: P-0.05 vs. zero ascorbate control. 346 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 for producing the amino acid profile expected of mature lagen supported elastogenesis by HDF, and the response elastin after alkaline extraction. to growth factor stimulation in collagen surpassed that Fibrin’s superiority over collagen here lends support in fibrin. Collagen’s different influence on SMC and to the fibrin model of vasculogenesis. It has been HDF is consistent with developmental variation between proposed that vasculogenesis occurs through a fibrin dermal and vascular tissues in vivo. Unlike vasculature, exudate from leaky vessel precursors in development embryonic skin consists primarily of collagen during the (Hiraoka et al., 1998; Van Hinsbergh et al., 2001).In elastin deposition phase (Smith and Holbrook, 1986), vitro study of vessel development has been largely an inherent difference perhaps contributing to the suc- limited to endothelial tube formation, but the elastic cessful elastogenesis by HDF in collagen gels here. medial wall arises shortly thereafter. Thus, the fibrin Furthermore, the variable elastogenesis by different cell matrix may indeed be a more physiological milieu for types in collagen shown here may explain the findings smooth muscle cell-mediated elastogenesis than colla- of L’Heureux et al. in a different system. They demon- gen. Closer examination of the elastogenicmicroenvi- strated elastin in the rolled extracellular matrix synthe- ronment in vivo would address this hypothesis. Integrins, sized de novo by HDF but not in matrix similarly which are differentially engaged by fibrin and collagen, synthesized by neonatal SMC. Both matrices in that have many functions in vasculogenesis and may be the system were produced completely by the plated cells mechanism for improved elastogenesis in fibrin. For and were composed predominantly of cell-derived col- example, the avb3 integrin mediates both SMC and lagen (L’Heureux et al., 1998), in contrast to the cell endothelial cell interaction with fibrin but not collagen, entrapment and fibrin remodeling approach used here. and is known to be critical in vasculogenesis (Moiseeva, While the robust HDF elastogenesis is promising, SMC 2001; Dallabrida et al., 2000; Feng et al., 1999; Rupp were studied more extensively here because they would and Little, 2001). Further basicstudy of vasculogenesis supply vasocontractile function in a tissue-engineered and integrins will by necessity proceed comcomitantly artery. and should more fully define the elastogenicenviron- Ascorbate produced somewhat different results in gel ment. Since the system described in this study produces culture than in traditional monolayer studies. It has long substantial elastin, it may be useful for studying both been recognized to increase collagen synthesis, at least integrin effects on elastogenesis and overall vessel to some degree via stabilization of ferrous iron to development in fibrin. improve hydroxylation of lysine and proline residues Cell density appears to be a secondary influence on (Padh, 1991). In fibrin tissue-equivalents, we found a elastogenesis. In the inferior cases of collagen gel or significant but small increase in hydroxyproline content aSMC culture, elastin deposition was not proportional when ascorbate was added to constructs treated with to initial cell number. This suggests that lower density 10% serum only. The modest increase suggests that cells produced more tropoelastin on a per cell basis, collagen incorporation in the matrix is already near the perhaps due to better nutrition or oxygenation or the maximum possible with serum alone. Combining ascor- lack of paracrine signalling. In the highly elastogenic bate with TGF-b1 and insulin, however, synergistically case of neoSMC in fibrin, elastin deposition was pro- stimulated collagen incorporation. TGF-b1 increases portional to initial cell density. Previous studies in this protein synthesis overall while ascorbate increases the laboratory found an inverse relation between neoSMC collagen fraction of protein synthesis in monolayer proliferation in fibrin gel culture and initial cell density, culture (Phillips et al., 1992); such a shift in protein so that initial differences in cell density were equalized synthesis could contribute to the increased collagenesis by 4 weeks (Grassl, 2002). Persistent difference in shown here in the fibrin gel culture system. We further elastin at 4 weeks despite equalizing cell numbers postulate that the TGF-b1 and insulin treatment implicates early elastogenic behavior as key to the final improves conditions for matrix remodeling (for example amount incorporated. Indeed, real-time PCR found by upregulating lysyl oxidase (Shanley et al., 1997)) so strong tropoelastin gene expression at one and two that increased collagen production via ascorbate can weeks culture in fibrin-neoSMC tissue-equivalents, with translate to greater collagen deposition. tropoelastin mRNA reduced to 4% of maximum at 4 Importantly, elastin was downregulated with ascorbate weeks (Ross and Tranquillo, 2003). The per-cell elastin in this study only in serum-only cases; this does agree secretion then must be approximately equal at all cell with monolayer experiments showing less tropoelastin densities, indicating that neoSMC in fibrin are insensi- transcription in post-confluent ascorbate-supplemented tive to the signals which decreased per-cell elastin cultures (Davidson et al., 1997). TGF-b1 and insulin- incorporation with increasing cell density in the other stimulated elastogenesis was, however, maintained at all cases. ascorbate doses in fibrin gel culture. The precise mech- HDF behavior differed from SMC. Adult and neonatal anism by which ascorbate impacts elastogenesis has not fibroblasts produced comparable elastin amounts, which yet been defined, nor has the combined effect of ascor- were within the range of aorticSMC production.Col- bate and TGF-b1 on elastin been studied in other J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 347 systems. These results demonstrate that supplementing including microfibril-associated glycoproteins and ECM fibrin-neoSMC tissue-equivalents with TGF-b1, insulin, proteoglycans, have not been examined here but are and ascorbate can generate high levels of both elastin likely to be present and offer another point of potential and collagen deposition. control. Finally, fibrin culture of neoSMC with TGF- Desmosine RIA confirmed that substantial elastin b1, insulin, and 50 mgyml ascorbate, the condition crosslinking occurred in the optimized neoSMC-fibrin optimizing elastin and collagen production in this study, tissue-equivalents. Comparing amounts of desmosine is known to yield tissue-equivalents with 18- to 20-fold and elastin in similar samples yields an estimated 62 greater elasticmodulus and ultimate tensile stress than pmol desmosine per microgram elastin, which at a controls (Grassl et al., 2003). While it is difficult to tropoelastin molecular weight of 72 000 suggests 4.5 separate elastin’s influence from that of collagen and desmosines per tropoelastin molecule. This compares other matrix components, the elastic fibers demonstrated very favorably with the calculation of 3.6 desmosines here represent a significant advance in the in vitro per tropoelastin in aorticelasticfibers (Ross and Born- replication of arterial microstructure and are likely to stein, 1969). The dramaticchangein elastin immunola- contribute to the overall mechanical behavior. beling with BAPN treatment confirms that lysyl oxidase-mediated cross-linking is a key factor in fibrin 4. Experimental procedures tissue-equivalent maturation. Fibrillin-1 immunolabeling here increased with BAPN treatment, paralleling animal 4.1. Tissue-equivalent fabrication experiments with systemiclysyl oxidase inhibition (Gigante et al., 2001) and autopsy findings from a Adult and neonatal human dermal fibroblasts (aHDF patient with decreased lysyl oxidase activity secondary and neoHDF) were purchased from BioWhittaker to Menkes disease (Pasquali-Ronchetti et al., 1994). (Walkersville, MD). Adult and neonatal Fisher rats were Three-dimensional gel culture provides a relevant housed and handled in accordance with University of model system in which to study ECM assembly of Minnesota IACUC regulations. Rats were sacrificed, greater complexity than monolayer cell culture. Both thoracic aortae were harvested, and smooth muscle cells neoSMC and neoHDF here deposited elastin throughout (aSMC and neoSMC) were isolated using a digestive the entire thickness of fibrin gel, whereas fibrillin-1 was enzyme series. Cleaned, minced aortae were incubated found only near the lower surface closest to the plastic in 0.3% collagenase (Sigma, St. Louis, MO), then 0.1% culture well. The presence of elastin in apparently pancreatic elastase (Sigma), then 0.3% collagenase plus organized structures without corresponding fibrillin-1 0.1% elastase, each in phosphate-buffered saline for 1 h supports the emerging view that a microfibril template at 37 8C. All cells were maintained in DMEM supple- is not critical for elastogenesis (Marque et al., 2001). mented with 10% fetal bovine serum, 1% penicilliny

We propose that a surrogate molecule such as fibrin or streptomyocin and 1% fungizone in a 5% CO2 a serum protein that adheres to fibrin adopts the role of environment, were passed approximately weekly, and guiding tropoelastin alignment. We did not assess fibril- were used between passages 5 and 9. lin-2, so cannot comment on whether other microfibril- Collagen gel solution consisted of Vitrogen 100 pep- lary proteins may be more prevalent in this system. sin-digested bovine Type I collagen solution (Cohesion Fibrillin-1’s more limited spatial distribution does raise Technologies, Palo Alto, CA), mixed 8:1:1 with 10X the question of differential regulation of the multiple MEM culture medium and 0.1 M NaOH. Collagen gel components of elastic fibers, although we cannot rule solution was mixed 5:1 with cell suspension for an out that fibrillin-1’s rate of assembly within the fibrin ultimate collagen concentration of 2 mgyml. Fibrin gels gel may merely occur much faster than its diffusion were formed by mixing bovine fibrinogen solution 4:1:1 through the gel. Alternatively, fibrillin-1 may preferen- with bovine thrombin solution and cell suspension. tially self-aggregate rather than attaching to the fibrin Fibrinogen (Sigma) solution at 5 mgyml in 20 mM fibrils, leading to a limited deposition area. From a HEPES-buffered saline (HBS) was passed through a 0.2 tissue engineering standpoint, the mechanical behavior mm filter before use to remove clumps. Thrombin of elastin without corresponding fibrillin-1 is of para- (Sigma) solution was made at 2 Uyml in HBS with 15 mount interest. The biophysics of elastic fiber assembly mN CaCl2 . Cells were suspended in M199 medium with is a burgeoning field with further study required to 10% FBS at a concentration appropriate for a final cell address these and other questions. density of either 0.5E6, 1E6 or 2E6 cellsyml in the gel. This work demonstrates that in vitro elastogenesis for For sectioning and imaging, 100 ml adherent gels with tissue-engineering applications is feasible and control- 1E6 cellsyml were formed within circles scored in the lable to some extent by manipulating medium additives wells of a 24-well plate. For elastin extraction and to fibrin gel culture of neoSMC. Human dermal fibro- amino acid analysis, 480 ml gels were formed as thicker blasts also produced significant elastin and collagen. adherent disks. Unless noted otherwise, cell concentra- Other molecules known to be involved in elastogenesis, tion was 1E6 cellsyml. 348 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350

Gels were allowed to solidify for1hat378C before antibody solution: elastin ratio. After gently rotating adding culture medium to the wells. All gels were overnight at 4 8C, the incubation tubes were centrifuged maintained in M199 medium with 10% fetal bovine and the supernatant antibody solutions removed. Immu- serum, 1% penicillinystreptomyocin and 1% fungizone nolabeling was then performed as described above. in an air environment. In growth factor experiments, 2 Elastin labeling disappeared when using primary anti- ngyml TGF-b1 (R and D Systems, Minneapolis, MN) bodies, which had been pre-incubated with elastin iso- and 2 mgyml insulin (Sigma) were added to the culture lated from human umbilical artery. Fibrillin-1 labeling medium. Plasmin (R and D Systems) was included at activity was retained after antibody incubation with 0.01 Uyml only with HDF, because the SMC rapidly extracted elastin, verifying that the antibody does not degraded fibrin. To slow this degradation, medium recognize elastin and that microfibrils are destroyed in overlying SMC in fibrin was supplemented with ´- the alkaline extraction. amino-caproic acid (ACA) at 1 mgyml (Sigma).In another set of experiments, ascorbate (Sigma) was 4.3. Collagen and elastin quantification included at 10, 50 or 100 mgyml. b-aminopropionitrile, mono-fumarate salt (BAPN, Sigma) was included at 10 Elasticfibers were extractedfrom 480 ml tissue- mM in a set of neoSMC-fibrin tissue-equivalents. All equivalents by hot alkali treatment (Lansing et al., media were changed at three or four-day intervals. After 1952). After heating for 45 min at 98 8C in 0.1 M 4 weeks, the tissue-equivalents were rinsed with PBS, NaOH, insoluble residue was centrifuged and the super- removed from the wells, and frozen. natant removed. Elastin pellets were washed twice with distilled water, hydrolyzed for 24 h in 6 N HCl at 110 4.2. Assesment of elastin production in tissue-equivalents 8C, and lyophilized. Control acellular collagen and fibrin gels produced minimal alkaline-insoluble matter. Alka- One hundred microliter tissue-equivalents were cry- line supernatant contains all the other solubilized pro- osectioned at 10 mm and stored frozen. Immediately teins from the tissue-equivalent. Supernatants were before immunolabeling, sections were thawed at room lyophilized, hydrolyzed for 24 h in 6 N HCl at 110 8C, temperature, fixed for 10 min with 2% paraformalde- lyophilized again, and subjected to hydroxyproline hyde, and washed well with PBS. Sections were blocked measurement as an indicator of collagen content. for 30 min with 5% donkey serum (Jackson Immuno- Insoluble elastin was quantified in a modification of research Laboratories, West Grove, PA) in PBS. To the procedure described by Starcher (Starcher and further reduce non-specific binding to ECM components, Mecham, 1981). Hydrolyzed elastin was diluted in all antibodies were preabsorbed with paraformaldehyde- double distilled water, and 50 ml of each sample or a fixed acellular collagen and fibrin gels overnight at 4 hydrolyzed a-elastin standard (Elastin Products Com- 8C. Primary antibodies were applied for 60 min. Primary pany) was added to a 96-well plate. 100 ml of ninhydrin antibodies and dilutions included rabbit anti-rat elastin solution (2% wyv ninhydrin, 75% vyv ethylene glycol, at 1:100 (Elastin Products Company, Owensville, MO), 25% vyv 4N sodium acetate in 20% acetic acid; mixed mouse anti-human elastin at 1:50 (Chemicon Interna- with 2.5% vyv of 10% stannous chloride in ethylene tional, Temecula, CA), mouse anti-human or rat fibrillin- glycol) was added to each well. The plate was floated 1 at 1:50 (Chemicon) and pre-immune mouse or rabbit ona568C water bath for 1 h, conditions which were IgG (Jackson) at 25 mgyml, a concentration approxi- found to achieve complete color development. Total mately equivalent to the antibody solutions. Sections amino acid content was determined by colorimetric were washed in PBS, and secondary antibodies were absorbance at 550 nm. The elastin standard curve was applied for 60 min in the dark. Secondary antibodies consistently linear over a range of 0.3–10 mg. and dilutions were donkey anti-rabbit or mouse at 1:400, Desmosine RIA was performed on elastin isolated conjugated with Cy3 fluorophore (Jackson). Sections from selected tissue-equivalents, using a magnetized were washed, and coverslips were mounted with Vecta- antibody separation process (Starcher and Conrad, Shield plus DAPI (Vector Laboratories, Burlingame, 1995). Amino acid analysis was performed on elastin CA). Sections were viewed within 48 h with an Olym- isolated from selected tissue-equivalents, using a Dionex pus confocal microscope. Images presented are recon- ion-exchange separation column with ninhydrin structions of eleven optical sections taken at 1 mm detection. intervals. Hydroxyproline was measured by chloramine-T oxi- Acellular fibrin or collagen gels produced no immu- dation and DMBA reaction (Stegemann and Stalder, nolabeling. Primary antibody specificity was further 1967). Supernatant hydrolysates were diluted in citrate confirmed by preabsorption. Antibodies against elastin buffer, passed through charcoal to remove caramelized and fibrillin-1 were incubated with elastin extracted carbohydrates, and 100 ml added to a 96-well plate. from human umbilical arteries by the Lansing hot alkali Fifty microlitres each of 1.41% chloramine-T and 15% procedure (Lansing et al., 1952) at a 10:1 diluted DMBA reagents were added to each well. After incu- J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350 349 bating 30 min at 37 8C, hydroxyproline content was and organization of the elastin network in a skin equivalent. J. determined by colorimetric absorbance at 550 nm. Invest. Dermatol. 144, 365–370. Hydroxyproline background was low in the supernatant Fazio, M.J., Olsen, D.R., Kuivaniemi, H., Chu, M.L., Davidson, J.M., Rosenbloom, J., et al., 1988. Isolation and characterization of fluid of acellular fibrin gels, whereas the expected human elastin cDNAs, and age-associated variation in elastin gene amount of hydroxyproline was recovered in the super- expression in cultured skin fibroblasts. Lab. Invest. 58, 270–277. natant of acellular collagen gels. Feng, X., Clark, R.A., Galanakis, D., Tonnesen, M.G., 1999. Fibrin Standard deviation was calculated for data sets of at and collagen differentially regulate human dermal microvascular least three samples. Data from each cell type were endothelial cell integrins: stabilization of alphavybeta3 mRNA by analyzed by two-way ANOVA between gel type, growth fibrin1. J. Invest. Dermatol. 113, 913–914. Gigante, A., Chillemi, C., Quaglino, D., Miselli, M., Pasquali- factor treatment status, ascorbate concentration or cell Ronchetti, I., 2001. DL-penicillamine induced alteration of elastic density. Sets demonstrating P-0.05 at as0.05 were fibers of periostium-perichondrium and associated growth inhibi- further analyzed by pairwise 2-tailed t-tests assuming tion. J. Ortho Res. 19, 398–404. unequal variances to detect differences in means. The Grassl, E., 2002. Enhancing the properties of the medial layer of a same procedure was used to compare adult and neonatal bioartificial artery. Doctoral thesis. Department of Chemical Engi- SMC between corresponding conditions. neering and Materials Science. University of Minnesota, Minne- apolis, MN. Grassl, E.D., Oegema, T.R., Tranquillo, R.T., 2002. Fibrin as an Acknowledgments alternative biopolymer to type-I collagen for the fabrication of a media equivalent. J. Biomed. Mater. Res. 60, 607–612. We gratefully acknowledge Dr Barry Starcher for Grassl, E.D., Oegema, T.R., Tranquillo, R.T., 2003. A fibrin-based performing the desmosine RIA. This work has been arterial media equivalent. J. Biomed. Mater. Res. in press. ( ) Hill, C., Mecham, R., Starcher, B., 2002. Fibrillin-2 defects impair supported by NHLBI 1R01 HL60495 R.T.T. , a Whi- elastic fiber assembly in a homocysteinemic chick model. J. Nutr. taker Foundation Graduate Fellowship (J.L.L.), and a 132, 2143–2150. BakkenyMcKnight M.D.yPh.D. Endowment (J.L.L.). Hinek, A., Smith, A.C., Cutiongco, E.M., Callahan, J.W., Gripp, K.W., Weksberg, R., 2000. Decreased elastin deposition and high References proliferation of fibroblasts from Costello syndrome are related to functional deficiency in the 67-kD elastin-binding protein. Am. J. Hum. Genet. 66, 859–872. Bank, A.J., Wang, H., Holte, J.E., Mullen, K., Shammas, R., Kubo, Hinek, A., Wilson, S.E., 2000. Impaired elastogenesis in Hurler S.H., 1996. Contribution of collagen, elastin and smooth muscle disease: dermatan sulfate accumulation linked to deficiency in to in vivo human brachial artery wall stress and elastic modulus. elastin-binding protein and elasticfiber assembly. Am. J. Pathol. Circulation 94, 3263–3270. 156, 925–938. Berglund, J., Nerem, R., Sambanis, A., 2001. Intact elastin scaffolds Hiraoka, N., Allen, E., Apel, I.J., Gyetko, M.R., Weiss, S.J., 1998. in collagen-based tissue engineered vascular grafts. In: L.V. McIn- ( ) Matrix metalloproteinases regulate neovascularization by acting as tire Ed. , Proceedings of the Annual Fall Meeting of the Biomed- pericellular fibrinolysins. Cell 95, 365–377. ical Engineering Society, (Durham, NC: Biomedical Engineering Johnson, D.J., Robson, P., Hew, Y., Keeley, F.W., 1995. Decreased Society), pp. S-149. elastin synthesis in normal development and in long-term aortic Berthod, F., Germain, L., Li, H., Xu, W., Damour, O., Auger, F.A., organ and cell cultures is related to rapid and selective destabili- 2001. Collagen fibril network and elasticsystem remodeling in a zation of mRNA for elastin. Circ. Res. 77, 1107–1113. reconstructed skin transplanted on nude mice. Matr. Biol. 20, Kahari, V.M., Olsen, D.R., Rhudy, R.W., Carrillo, P., Chen, Y.Q., 463–473. Uitto, J., 1992. Transforming growth factor-beta up-regulates elastin Carreras, I., Rich, C., Panchenko, M., Foster, J., 2002. Basic fibroblast gene expression in human skin fibroblasts. Evidence for post- growth factor decreases elastin gene transcription in aortic smooth transcriptional modulation. Lab. Invest. 66, 580–588. muscle cells. J. Cellular Biochem. 85, 592–600. Karnik, S.K., Brooke, B.S., Bayes-Genis, A., Sorensen, L., Wythe, Clark, R.A.F., Nielsen, L.D., Welch, M.P., McPherson, J.M., 1995. J.D., Schwartz, R.S., et al., 2003. A critical role for elastin signaling Collagen matrices attenuate the collagen-synthetic response of in vascular morphogenesis and disease. Development 130, cultured fibroblasts to TGF-beta. J. Cell. Sci. 108, 1251–1261. 411–423. Dallabrida, S.M., De Sousa, M.A., Farrell, D.H., 2000. Expression of antisense to integrin subunit beta 3 inhibits microvascular endothe- Katchman, S.D., Hsu-Wong, S., Ledo, I., Wu, M., Uitto, J., 1994. lial cell capillary tube formation in fibrin. J. Biol. Chem. 275, Transforming growth factor-beta up-regulates human elastin pro- 32 281–32 288. moter activity in transgenic mice. Biochem. Biophys. Res. Com- Davidson, J.M., LuValle, P.A., Zoia, O., Quaglino, D., Giro, M., mun. 203, 485–490. 1997. Ascorbate differentially regulates elastin and collagen bio- Kim, B.S., Nikolovski, J., Bonadio, J., Smiley, E., Mooney, D.J., synthesis in vascular smooth muscle cells and skin fibroblasts by 1999. Engineered smooth muscle tissues: Regulating cell phenotype pretranslational mechanisms. J. Biol. Chem. 272, 345–352. with the scaffold. Exp. Cell. Res. 251, 318–328. Davidson, J.M., Zoia, O., Liu, J.M., 1993. Modulation of transforming L’Heureux, N., Paquet, S., Labbe, R., Germain, L., Auger, F.A., 1998. growth factor-beta 1 stimulated elastin and collagen production and A completely biological tissue-engineered human blood vessel. proliferation in porcine vascular smooth muscle cells and skin FASEBJ 12, 47–56. fibroblasts by basicfibroblast growth factor, transforming growth Labella, F.S., Vivian, S., 1967. Amino acid composition of elastin in factor-alpha, and insulin-like growth factor-I. J. Cell. Phys. 155, the developing human aorta. Biochim. Biophys. Acta 133, 189–194. 149–156. Lansing, A.I., Rosenthal, T.B., Alex, M., Dempsey, E.W., 1952. The Duplan-Perrat, F., Damour, O., Montrocher, C., Peyrol, S., Grenier, structure and chemical characterization of elastic fibers as revealed G., Jacob, M., et al., 2000. Keratinocytes influence the maturation by elastase and by electron microscopy. Anat. Rec. 114, 555–570. 350 J.L. Long, R.T. Tranquillo / Matrix Biology 22 (2003) 339–350

Marque, V., Kieffer, P., Gayraud, B., Lartaud-Idjouadiene, I., Ramirez, Rupp, P., Little, C.D., 2001. Integrins in vascular development. Circ. F., Atkinson, J., 2001. Aortic wall mechanics and composition in Res. 89, 566–572. a transgenicmouse model of marfan syndrome. Arterioscler. Sauvage, M., Hinglais, N., Mandet, C., Badier, C., Deslandes, F., Thromb. Vasc. Biol. 21, 1184–1189. Michel, J.B., et al., 1998. Localization of elastin mRNA and TGF- McMahon, M.P., Faris, B., Wolfe, B.L., Brown, K.E., Pratt, C.A., beta1 in rat aorta and caudal artery as a function of age. Cell. Toselli, P., et al., 1985. Aging effects on the elastin composition in Tissue Res. 291, 305–314. the extracellular matrix of cultured rat aortic smooth muscle cells. Shanley, C.J., Gharaee-Kermani, M., Sarkar, R., Welling, T.H., Krie- In Vitro Cell. Dev. Biol. 21, 674–680. gel, A., Ford, J.W., et al., 1997. Transforming growth factor-beta 1 Mecham, R.P., 1981. Effects of extracellular matrix upon elastoge- increases lysyl oxidase enzyme activity and mRNA in rat aortic nesis. Connect. Tissue Res. 8, 241–244. smooth muscle cells. J. Vasc. Surg. 25, 446–452. Shum-Tim, D., Stock, U., Hrkach, J., Shinoka, T., Lien, J., Moses, Merrilees, M., Lemire, J., Fischer, J., Kinsella, M., Braun, K., Clowes, M., et al., 1999. Tissue engineering of autologous aorta using a A., et al., 2002. Retrovirally mediated overexpression of versican new biodegradable polymer. Ann. Thorac. Surg. 68, 2298–2305. V3 by arterial smooth muscle cells induces tropoelastin synthesis Smith, L.T., Holbrook, K.A., 1986. Embryogenesis of the dermis in and elasticfiber formation in vitro and in neointima after vascular human skin. Ped. Derm. 3, 271–280. injury. Circ. Res. 90, 481–487. Smith-Mungo, L.I., Kagan, H.M., 1998. Lysyl oxidase: properties, Moiseeva, E., 2001. Adhesion receptors of vascular smooth muscle regulation and multiple functions in biology. Matrix Biol. 16, cells and their functions. Cardiovas. Res. 52, 372–386. 387–398. Nakamura, T., Lozano, P.R., Ikeda, Y., Iwanaga, Y., Hinek, A., Starcher, B., Conrad, M., 1995. A role for neutrophil elastase in the Minamisawa, S., et al., 2002. Fibulin-5yDANCE is essential for progression of solar elastosis. Connect. Tissue Res. 31, 133–140. elastogenesis in vivo. Nature 415, 171–175. Starcher, B.C., Galione, M.J., 1976. Purification and comparison of Narayanan, A.S., Sandberg, L.B., Ross, R., Layman, D.L., 1976. The elastins from different animal species. Anal. Biochem. 74, 441–447. smooth muscle cell III. Elastin synthesis in arterial smooth muscle Starcher, B.C., Mecham, R.P., 1981. Desmosine radioimmunoassay cell culture. J. Cell. Biol. 68, 411–419. as a means of studying elastogenesis in cell culture. Connect. Neidert, M.R., Oegema, T.R., Tranquillo, R.T., 2003. Tissue engi- Tissue Res. 8, 255–258. neered valves with commissural alignment. Submitted. Stegemann, H., Stalder, K., 1967. Determination of hydroxyproline. Neidert, M., Lee, E., Oegema, T., Tranquillo, R.T., 2002. Enhanced Clinica Chimica Acta 18, 267–273. fibrin remodeling in vitro for improved tissue-equivalents. Bioma- Stock, U.A., Wiederschain, D., Kilroy, S.M., Shum-Tim, D., Khalil, terials 23, 3717–3731. P.N., Vacanti, J.P., et al., 2001. Dynamics of extracellular matrix production and turnover in tissue engineered cardiovascular struc- Padh, H., 1991. Vitamin C: New insights into its biochemical tures. J. Cell. Biochem. 81, 220–228. function. Nutr. Rev. 49, 65. Stone, P.J., Morris, S.M., Griffin, S., Mithieux, S., Weiss, A.S., 2001. Parks, W.C., Whitehouse, L.A., Wu, L.C., Mecham, R.P., 1988. Building elastin. Incorporation of recombinant human tropoelastin Terminal differentiation of nuchal ligament fibroblasts: characteri- into extracellular matrices using nonelastogenic rat-1 fibroblasts as zation of syntheticproperties and responsiveness to external stimuli. a source for lysyl oxidase. Am. J. Resp. Cell. Mol. Biol. 24, Dev. Biol. 129, 555–564. 733–739. Pasquali-Ronchetti, I., Baccarani-Contri, M., Young, R.D., Vogel, A., Tranquillo, R.T., 1999. Self-organization of tissue-equivalents: the Steinmann, B., Royce, P.M., 1994. Ultrastructural analysis of skin nature and role of contact guidance. Biochem. Soc. Symposia 65, and aorta from a patient with Menkes disease. Exp. Mol. Pathol. 27–42. 61, 36–57. Urban, Z., Riazi, S., Seidl, T.L., Katahira, J., Smoot, L.B., Chitayat, Penner, A., Rock, M., Kielty, C., Shipley, J., 2002. Microfibril- D., et al., 2002. Connection between elastin haploinsufficiency and associated glycoprotein-2 interacts with fibrillin-1 and fibrillin-2 increased cell proliferation in patients with supravalvular aortic suggesting a role for MAGP-2 in elasticfiber assembly. J. Biol. stenosis and Williams–Beuren syndrome. Am. J. Hum. Genet. 71, Chem. 277, 35 044–35 049. 30–44. Phillips, C.L., Tajima, S., Pinnell, S., 1992. Ascorbic acid and Urry, D.W., Pattanaik, A., 1997. Elasticprotein-based materials in transforming growth factor-beta 1 increase collagen biosynthesis tissue reconstruction. Ann. NY Acad. Sci. 831, 32–46. via different mechanisms: coordinate regulation of pro alpha1(I) Van Hinsbergh, V.W., Collen, A., Koolwijk, P., 2001. Role of fibrin and Pro alpha 1(III) collagens. Arch. Biochem. Biophys. 295, matrix in angiogenesis. Ann. NY Acad. Sci. 936, 426–437. Vesely, I., 1998. The role of elastin in aortic valve mechanics. J. 397–403. Biomech. 31, 115–123. Ross, J., Tranquillo, R., 2003. ECM gene expression correlates with Wolfe, B.L., Rich, C.B., Goud, H.D., Terpstra, A.J., Bashir, M., in vitro tissue growth and development in fibrin gel remodeled by Rosenbloom, J., et al., 1993. Insulin-like growth factor-I regulates neonatal smooth muscle cells. Matr. Biol. accepted. transcription of the elastin gene. J. Biol. Chem. 268, Ross, R., Bornstein, P., 1969. The elasticfiber: separation and partial 12 418–12 426. characterization of macromolecular components. J. Cell. Biol. 40, Yanagisawa, H., Davis, E.C., Starcher, B.C., Ouchi, T., Yanagisawa, 366–381. M., Richardson, J.A., et al., 2002. Fibulin-5 is an elastin-binding Ross, R., Bornstein, P., 1971. Elasticfibers in the body. Sci.Am. protein essential for elasticfibre development in vivo. Nature 415, 224, 44–52. 168–171.