Selective expression and processing of biglycan during migration of bovine aortic endothelial cells. The role of endogenous basic fibroblast growth factor.

Repair of the vascular lumenal surface after injury requires a controlled endothelial cell response that includes cell migration, proliferation, and remodeling of the extracellular matrix. These cellular processes are modulated by growth factors that are released or activated following cell injury. When endothelial cell migration is stimulated in response to monolayer wounding in vitro, cells increase synthesis of small leucine-rich dermatan sulfate proteoglycans (PGs) (Kinsella, M. G., and Wight, T. N. (1986) J. Cell Biol. 102, 679-687). However, the identity of the PGs that are increased during cell migration and the factors that affect this modulation have not been identified. We now report that basic fibroblast growth factor (bFGF) is responsible for the transient increase of [35S]sulfate incorporation into PGs following monolayer wounding. SDS-polyacrylamide gel electrophoresis analysis revealed that bFGF-treated and wounded cultures increase both biglycan core protein synthesis and biglycan proteolytic processing, which results in the accumulation of a approximately 20-kDa N-terminal biglycan fragment in the culture media. Biglycan RNA steady-state levels also selectively increase 2- to 3-fold after wounding or bFGF treatment. Finally, immunocytochemical staining localizes biglycan to the tips and edges of lamellopodia on migrating cells, indicating that biglycan is found at loci at which the formation and dissolution of adhesion plaques occurs, consistent with hypotheses that predict involvement of biglycan in the control of cell migration. Taken together, these results suggest that release of endogenous bFGF is primarily responsible for altered biglycan expression, synthesis, and proteolytic processing as endothelial cells migrate after wounding.

Migration and proliferation of endothelial cells and the subsequent establishment of a stable monolayer are critical events in the repair of injured vessels and in angiogenesis and vasculogenesis during development, tumor growth, and tissue repair. Cell migration and proliferation are controlled by multiple factors, such as FGF 1 and transforming growth factor-␤ (1,2), and cell adhesive interactions, mediated by cell surface receptors such as integrins (3). Thus, initiation of cell migration may involve a change in cell-matrix interaction characterized by extracellular matrix remodeling and altered matrix receptor expression, which are processes directed in part by growth factors.
Proteoglycans are a heterogeneous group of protein families that bear anionic glycosaminoglycan (GAG) chains covalently bound to core proteins. These molecules are prominent constituents of both the extracellular matrix and the cell surface, where they are proposed to play roles in cell adhesion, growth factor interactions, and matrix assembly (4,5). Members of the small leucine-rich PG family, such as biglycan and decorin, which bear dermatan or chondroitin sulfate side chains, bind fibrillar proteins, such as collagens (6,7), and may regulate fibrillogenesis (8). In addition, biglycan and decorin may affect cell migration both by modulating interactions of cell surface receptors with their matrix ligands (9,10), and by influencing growth factor availability and function (5).
We have previously reported that the induction of migration in wounded endothelial cell monolayers is accompanied by increased dermatan sulfate PG synthesis, which is associated with cells at the wound edge (11). The principal dermatan sulfate PG synthesized by confluent cultured aortic endothelial cells is biglycan (12,13), consistent with immunochemical and in situ hybridization studies of endothelia in vessels (14,15). However, decorin expression, which is not detectable in monolayers of aortic endothelial cells (13), is induced in concert with type I collagen when cells undergo sprouting and tube formation in vitro (16). Thus, decorin and biglycan expression are differentially regulated as endothelial cells migrate, proliferate, or alter their phenotype. Although changes in the expression of small leucine-rich PGs may affect extracellular matrix assembly, cellular adhesion, or growth factor utilization that is critical to cell migration, the factors that control these changes are poorly understood. bFGF is clearly a candidate for such a role, as it alters migration, proliferation, and matrix protein deposition by endothelial cells (2), and is released from cultured monolayers after wounding in vitro (17). Therefore, we have investigated the role of endogenous bFGF release as a mechanism that may control the expression of PGs after wounding of endothelial monolayers in vitro. We find that increased expression and proteolytic processing of biglycan, which is present on the lamellopodia of migrating cells, is selectively induced by the release of endogenous bFGF after multiscratch wounding of monolayers, suggesting that biglycan metabolism may be involved in the control of cell migration. Two to three days after reaching confluence (8 -10 days in culture), some cell layers were multiscratch wounded with a nylon rake, singlescratch wounded with a razor blade as described previously (11), or treated with growth factors. For metabolic labeling of proteoglycans, cultures were labeled with 50 -100 Ci/ml carrier-free Na[ 35 S]O 4 (Amersham) for times up to 24 h, or with 50 Ci/ml [ 35 S]methionine/ cysteine (Tran 35 S-label, DuPont NEN) for times up to 6 h, by replacement with fresh Dulbecco's modified Eagle's medium (either sulfatefree or prepared without methionine, as appropriate) containing 10% fetal bovine serum (Sigma). Incorporation of radiosulfate into proteoglycans was determined from duplicate aliquots of culture medium or cell layer extracts by cetyl pyridinium chloride precipitation (19).
Electrophoretic Separation and Immunochemical Identification of Biglycan-For analytical separation of radiolabeled PGs on SDS-PAGE or preparative electrophoresis in SDS-agarose gels, samples were partially purified by application to a 0.5-ml DEAE-Sephacel column in 8 M urea with 0.5% Triton X-100, 0.1 M Tris-HCl, pH 7.5, and 0.25 M NaCl (urea buffer), washing with ϳ10 volumes of urea buffer, and then eluting bound macromolecules in urea buffer with 3 M NaCl. Portions of eluted material were precipitated twice by addition of 3.5 volumes of 95% ethanol containing 1.3% potassium acetate and dried. Dried samples were resuspended in deionized 8 M urea, either with or without prior digestion with 0.02 unit of chondroitin ABC lyase (ICN Pharmaceuticals) in enriched Tris buffer (20), pH 8.0, for 3 h at 37°C. Digested and undigested samples in urea were then boiled in SDS-containing sample buffer and applied to either 4 -12% gradient SDS-polyacrylamide gels (21) or agarose-SDS gels as described previously (22) and below. 14 C-Labeled or prestained protein molecular weight standards (Life Technologies, Inc.) were used to estimate PG size. The positions of labeled bands on SDS-PAGE gels were visualized by fluorography of fixed, dried gels treated with Enlightening (DuPont NEN), and exposed at Ϫ70°C on Kodak XAR-2 film. Quantitation of radiolabeled bands was by scanning densitometry of the fluorograph, as described previously (23). For Western blotting of core proteins, SDS-PAGE gels were equilibrated in 50 mM Tris, 40 mM glycine, pH 9.2 transfer buffer with 20% methanol and 0.0375% SDS (24), and transferred to nitrocellulose (BA83, Schleicher and Schuell, Inc.) for 50 min with a semidry transfer apparatus (Transblot SD, Bio-Rad Laboratories). Nitrocellulose membranes were blocked with 2% bovine serum albumin (Fraction V, Boehringer Mannheim Corp.) in Tris-buffered saline with 0.05% Tween 20, and exposed to primary antibodies (diluted 1:1000) overnight at 4°C. After incubation of the blot with alkaline phosphatase-conjugated secondary antibodies, bands that bound primary antibodies were visualized by an enzyme-linked chemiluminescence procedure (Tropix, Bedford, MA), or using 0.2 mg/ml nitro blue tetrazolium (Sigma) in 100 mM Tris-HCl, pH 9.7, with 5 mM MgCl 2 , 10 M ZnCl 2 , and 0.2 mg/ml 5-bromo-4-chloro-3-indolyl phosphate (Sigma).
SDS-agarose gels were used for electrophoretic separation and isolation of individual PG bands. Thus, mixtures of PG, prepared for electrophoresis as described above, were applied to horizontal 4% NuSieve GTG agarose (FMC BioProducts, Rockland, ME) gels in 90 mM Tris borate buffer, pH 8.3, with 2 mM EDTA and 0.1% SDS, and run at 50 V for 6 h. To isolate separated PGs, unfixed gels were sectioned horizontally using prestained protein standards as reference markers. Excised agarose sections were melted, diluted 10-fold in urea buffer and applied to DEAE-minicolumns as described above. After washing, bound PGs from each section were eluted in urea buffer with 3 M NaCl and dialyzed into PBS. Aliquots of isolated PGs were iodinated by the chloramine T procedure, using IODO-BEADS (Pierce), and applied to SDS-PAGE gels to determine the purity of each isolate.
cDNA Probes-Several cDNAs were used as probes for PG RNA on Northern blots, including full-length human biglycan cDNA (P16) and full-length bovine decorin cDNA, provided by Dr. L. Fisher (27) and Dr. M. Young (28), respectively, of the NIDR, NIH, Bethesda, MD; human perlecan cDNA (HS-1) from Dr. R. V. Iozzo, Thomas Jefferson University, Philadelphia, PA (29), and human versican cDNA (C10) from Dr. E. Ruoslahti (30). Northern blots were normalized for loading by comparison to hybridization of bovine 28 S rRNA cDNA that was kindly provided by Dr. E. H. Sage, University of Washington, Seattle, WA. Probes were 32 P-labeled by nick translation (Life Technologies, Inc.) or random priming (Amersham), using 5Ј-[␣-32 P]dCTP (Amersham), as described previously (13), and used to hybridize with RNA on Northern blots prepared as described above. Autoradiographs were prepared by exposure on Kodak XAR2 film at Ϫ70°C and then developed. Quantitation of labeled bands was as described above for SDS-PAGE. Statistical analysis of sample means was by the two-tailed t test.
Immunocytochemistry-Cells were cultured on coverslips until confluent, at which time portions of some monolayers were removed with a razor blade fragment as described previously (11). Coverslips containing cells were washed twice with phosphate-buffered saline (PBS) and fixed for 15 min in freshly prepared 3% paraformaldehyde in PBS, pH 7.4, at 4°C. After washing in PBS, cells on some coverslips were permeabilized with 0.1% Triton X-100 in PBS, and coverslips were blocked in 10% calf serum in PBS and then in 2% goat serum (Sigma, for rabbit primary antibodies) or in 2% rabbit serum (for chicken primary antibodies). Prior to staining with LF-96, coverslips were treated with 0.05 unit/ml chondroitin ABC lyase in enriched Tris buffer (20) for 3 h at 37°C. Cells were exposed to primary antibodies diluted in 10% calf serum at 4°C overnight, at dilutions ranging from 1:50 to 1:500. Control coverslips were exposed to PBS, normal rabbit serum (Zymed Laboratories, Inc., So. San Francisco, CA), or normal chicken serum (Accurate Chemical and Scientific Corp., Westbury, NY), diluted to match the IgG concentrations in the diluted primary antibodies. As an additional control, preadsorption of diluted LF-96 with the N-terminal bovine biglycan peptide it was raised against (kindly provided by Dr. L. Fisher, NIDR, NIH) was done at a concentration of 100 nM, overnight at 4°C, followed by centrifugation of the preadsorbed antibody. After washing with 10% chondroitin sulfate in PBS, and then PBS alone, coverslips were treated with fluorescein isothiocyanate-labeled goat anti-rabbit (Zymed Laboratories, Inc.) or rabbit anti-chicken IgG (Zymed Laboratories, Inc.) diluted 1:200 in 10% calf serum, for 1 h, washed, and mounted in 10% glycerol for fluorescent photomicroscopy.  (Fig. 1). When similarly labeled wounded cultures were compared at 2, 5, 24, and 48 h after wounding with confluent cultures treated with 10 ng/ml bFGF, a peak of incorporation ϳ2-fold higher than control was found (Fig. 2, A and B) after either treatment. This increase is apparent 24 h after wounding, in agreement with earlier studies (11), while the increase occurs by 5 h after bFGF treatment. bFGF treatment of cultures, like wounding, causes larger increases in PG synthesis over control in the media (Fig. 2, A and B) of the cultures than in the cell layers (not shown). In both culture treatments, the change in incorporation is transient and approaches control values by 48 h. Comparison of the relative proportions of [ 35 S]sulfate incorporated into either chondroitin/dermatan sulfate or heparan sulfate with time after wounding ( Fig. 3A) or bFGF treatment ( Fig. 3B), indicates that chondroitin/dermatan sulfate is specifically increased after either treatment, while heparan sulfate is either decreased (after wounding) or unchanged (after bFGF treatment). These experiments suggest that both wounding and exogenous bFGF result in a transient, specific increase in synthesis of dermatan sulfate PG, the bulk of which is rapidly secreted to the culture media.

Monolayer
Neutralizing Anti-bFGF Antibodies Inhibit PG Synthesis by Migrating Cells-A neutralizing antibody against bFGF (number 2373-1 (18)) was added to confluent cultures prior to multiscratch wounding or addition of exogenous bFGF to determine if release of endogenous bFGF contributes to increased PG synthesis in wounded cultures (Fig. 4). Antibody-treated, control confluent, and bFGF-treated and wounded cultures were pulsed with radiosulfate for 2 h at times for which increased incorporation had been demonstrated in preliminary experiments (10 -12 h   between 48 and 72 h after bFGF addition. Similar changes in the ϳ200,000 M r band are apparent after SDS-PAGE of cell layer samples (not shown), although the predominant band in the cell layer after 2 h pulse labeling is a large PG (M r Ͼ600,000) that is insensitive to chondroitin ABC lyase and has been identified as perlecan (31,32). 2 The PG band showing altered expression in wounded and bFGF-treated cultures has an apparent size (M r ϳ200 -300,000) consistent with its identification as biglycan (12,13,22,27). To confirm this identification, PGs were concentrated from culture media and the 200 -300,000 M r band was purified by SDS-agarose gel electrophoresis. After radioiodination and chondroitin ABC lyase digestion, this band yielded a ϳ50-kDa core protein, which was recognized on a Western blot by LF-96, a bovine biglycanspecific antibody (not shown, but see Fig. 6).
Another distinct and transient change in the radiolabeled band pattern in PGs present in media from both wounded and bFGF-treated cultures is the appearance of a new band between the 43-and 68-kDa standard bands (Fig. 5, arrows). The low M r band that is prominent in media samples is not apparent in cell layer extracts. To determine if the low M r , [ 35 S]sulfate-labeled band that appears after wounding or bFGF treatment includes a core protein, cultures treated with bFGF were labeled for 6 h with [ 35 S]methionine. After chondroitin ABC lyase treatment, SDS-PAGE revealed both the [ 35 S]methionine-labeled ϳ50-kDa core protein band of intact biglycan, as well as another core peptide band of ϳ20 kDa (Fig. 6A, arrows). In another experiment, PGs were partially purified by chromatography on DEAE-Sephacel from media of cultures treated for 12 h with bFGF (10 ng/ml). These isolates were then chondroitin ABC lyase-digested, run on SDS-PAGE, and Western blotted (Fig. 6B), using chicken anti-bovine biglycan (Ch105) or monoclonal antibodies (3B3 and 2B6) that recognize the terminal sulfated disaccharides that remain as "stubs" on the protein core after chondroitin lyase digestion. Both the anti-biglycan and the anti-stubs antibodies recognized bands on the Western blot of ϳ50 kDa and ϳ20 kDa. These results indicate that the diffuse lower M r band that appears after wounding or bFGF treatment of cultures contains a core protein fragment of biglycan, encompassing its N terminus, which contains the sites of dermatan sulfate chain addition (27). In a separate experiment, treatment of cultures with anti-bFGF neutralizing antibody prior to exposure to bFGF or wounding decreased the broad ϳ50-kDa band to 27% or 38%, respectively, of the percentage of radiolabel in the band in bFGF-treated or wounded cultures in the absence of the antibody (not shown). This observation suggests that the cleavage that results in the generation of biglycan N-terminal fragment is related directly to the induction of protease activity by bFGF.
Modulation of Biglycan Metabolism by bFGF or Wounding Involves Both Core Protein and RNA Expression-Biglycan synthesized after wounding or bFGF treatment is clearly larger in apparent M r than biglycan secreted by unstimulated cells (see Fig. 5). Similar increases in size of PGs by cells stimulated with several growth factors have been attributed to the syn- thesis of longer chondroitin sulfate chains on both small and large PGs (22,33). This observation raises the possibility that the described increase in [ 35 S]sulfate incorporation into biglycan is primarily an effect on GAG chain synthesis. Therefore, the time course of biglycan core protein synthesis was quantitated after pulse-labeling with [ 35 S]methionine at various times after multiscratch wounding or addition of 10 ng/ml bFGF to confluent cultures, to determine if biglycan core protein synthesis is increased (Fig. 7). Aliquots from bFGF-treated (Fig. 6A) or wounded cultures (not shown) run on SDS-PAGE gels showed the expected broad bands representing biglycan and its N-terminal fragment. Chondroitin ABC lyase digestion generates several core proteins including those of biglycan and biglycan fragment, at ϳ50 kDa and ϳ20 kDa, respectively. Digested cell layer aliquots contain less than 10% of the label present in comparable bands from media samples. Quantitation of radiolabel incorporated into total (intact ϩ fragment) biglycan core indicates that biglycan core protein increases ϳ4.2-fold compared with untreated control cultures, with a maximal stimulation occurring between 6 and 12 h after bFGF addition (Fig. 7B). However, the largest component of increased core production is the N-terminal fragment, which accounts for about 85% of total biglycan core protein (on a molar basis) at the peak of bFGF-stimulated synthesis. Similar results were obtained after quantitation of biglycan core protein synthesis following multiscratch wounding (Fig. 7A). However, the peak (ϳ3.3-fold increase over unwounded cultures) of core protein synthesis occurred later (18 h) than after bFGF addition, and N-terminal fragments comprised ϳ71% of total biglycan core. These results clearly indicate that overall synthesis of biglycan core protein is stimulated along with galactosaminoglycan chain synthesis after bFGF treatment or wounding. However, biglycan proteolytic processing also appears to be dramatically increased, with the result that the bulk of newly synthesized biglycan is rapidly cleaved after stimulation.
Northern blotting of total RNA extracted from wounded cultures at various times after wounding (Fig. 8) indicates that biglycan RNA expression is significantly up-regulated as cells migrate after wounding, with increased expression evident after 24 h. In contrast, expression of RNA for perlecan, which represents the other principal PG expressed by endothelial cells, changes little after multiscratch wounding, and versican expression decreases. Decorin RNA expression is barely detectable in endothelial cell cultures and remains at low levels at all times after wounding or growth factor treatment (not shown). A comparison of biglycan RNA expression after stimulation with bFGF or wounding (Fig. 9) indicates that biglycan RNA levels are elevated about 2-fold at 18 h after either treatment. However, expression of biglycan RNA remains elevated 24 h and 48 h after wounding, while in bFGF-treated cultures steady-state levels of biglycan mRNA decline at later time points. Biglycan mRNA expression appears to increase subsequent to both increased [ 35 S]sulfate incorporation into biglycan GAG chains (see Figs. 2 and 5) and labeled methionine incorporation into biglycan core protein (see Figs. 6 and 7). This unexpected observation may suggest that early changes in biglycan core protein synthesis are not controlled at the level of mRNA transcription (see "Discussion").
Biglycan Is Associated with the Lamellopodia of Migrating Endothelial Cells-Previous work has shown that dermatan sulfate PG synthesis is increased after endothelial cell monolayer wounding in vitro, and that this increase is primarily associated with migrating cells near the wound edge (11). The principal dermatan sulfate PG expressed (13) and regulated during cell migration after wounding by bovine aortic endothelial cells is biglycan. Therefore, the distribution of biglycan on nonpermeabilized cells in wounded and confluent monolayers of cells was examined immunocytochemically, using chicken anti-bovine biglycan (Ch105, Fig. 10) or rabbit antibodies against bovine biglycan-specific N-terminal peptide (LF-96, not shown) to address whether cell-associated biglycan was associated with particular cell surface domains. In these cultures, biglycan was distributed relatively uniformly at cell boundaries in confluent cultures (Fig. 10, C and D). In contrast, biglycan immunostaining on the surfaces of migrating cells at the wound edge was selectively associated with lamellopodia (Fig.  10, A and B). Control coverslips incubated with normal chicken serum instead of anti-biglycan antibody showed only light, diffuse fluorescence (not shown). This observation indicates that biglycan at the cell surface is associated with cellular structures that are important to cell movement and morphology.

DISCUSSION
As endothelial cells migrate after wounding, biglycan expression, synthesis, and proteolytic processing are selectively and transiently increased. Biglycan is a member of the family of FIG. 8. Northern blots of the expression of PG mRNA in confluent and multiscratch wounded cultures. Total RNA isolated from confluent (unWo) cultures and multiscratch wounded cultures at 2, 5, 24, and 48 h after wounding were separated on a denaturing agarose gel (15 g/lane), transferred to nitrocellulose, and probed with 32 P-labeled cDNAs for the prinicipal endothelial cell PGs, biglycan and perlecan, as well as versican, a large nonabundant chondroitin sulfate PG. A probe for 28 S ribosomal RNA was used to compare relative loading among the lanes. Autoradiographs were exposed at Ϫ70°C for 14 h (biglycan), 24 h (perlecan), 4 days (versican), and 12 h (28 S rRNA). Decorin expression (not shown), was barely detectable after a 9-day exposure.
small leucine-rich PGs, which contains at least 4 different proteins, including decorin. The members of this PG family are characterized by conserved cysteines that form disulfide--bonded loops near both termini of the protein core and highly homologous internal leucine-rich repeats, which comprise about 80% of the deduced sequences and have been postulated to mediate protein-protein interactions. Decorin (28) and biglycan (27) bear one or two dermatan sulfate chains, respectively, at the N-terminal serine residues that support GAG chain substitution. Leucine-rich PGs, such as biglycan and decorin, may regulate cellular migration through several proposed mechanisms. As constituents of the extracellular matrix, the core proteins of decorin and biglycan are thought to interact with several different matrix proteins and influence matrix assembly. For example, the interaction of decorin with interstitial collagens affects collagen fibrillogenesis (8). Biglycan may interact with fibronectin (23) and types I (7) and VI (34) collagen. The matrix proteins with which leucine-rich PGs interact provide binding sites for adhesion receptors on cells within the extracellular matrix. The interaction of decorin and/or biglycan with fibronectin modulates cellular adhesion (10,35), perhaps by influencing the binding of RGD-dependent cell surface receptors to sites on fibronectin. Such cooperative interactions have been proposed to destabilize focal adhesions, thereby modulating cellular migration (9). Also, biglycan and decorin bind to growth factors that may be important to vascular cell migration, including transforming growth factor-␤ (5,36) . This interaction may affect the availability or activity of these growth factors, which are important to the control of cell growth and migration (1,2). Taken together, the hypotheses engendered by these observations suggest that induction of biglycan expression or cleavage during cell migration may be important to modulate either cell-substrate adhesion or influence the availability and function of growth factors with a demonstrated role in cell migration and angiogenesis. Interestingly, the expression of decorin is not up-regulated as macrovascular endothelial cells migrate after wounding, in contrast to published studies of PG expression by these cells during in vitro angiogenesis (16). This observation suggests that processes involved in the formation and stabilization of endothelial tubes during in vitro angiogenesis, which may require the induction of decorin expression, are clearly distinct from processes involved in cell movement, during which decorin expression is not induced.
Because bFGF is known to be released after wounding of endothelial cell monolayers (17), and modulates cell phenotype and behavior, the effects of exogenous bFGF on PG synthesis were compared to the modulation of PG synthesis as cells migrate after multiscratch wounding. In these experiments, addition of bFGF had similar effects on biglycan synthesis as wounding, including transient increases in RNA expression, core protein synthesis, size of the intact PG monomer, and biglycan turnover. Since a neutralizing antibody against bFGF inhibits the stimulation of [ 35 S]sulfate incorporation into PG and proteolytic processing of biglycan in wounded or bFGFtreated cultures, we conclude that the release of endogenous bFGF is required for the modulation of these processes during cell migration. A notable difference in the effects of added exogenous bFGF and wounding on biglycan synthesis is that the response of the cells is delayed and persists for longer times in wounded cultures. This may suggest that release of bFGF from cell monolayers also persists for a relatively extended period of time after wounding.
In addition to increased sulfate incorporation into the GAG chains of biglycan, biglycan RNA expression and protein core synthesis are also elevated after bFGF treatment or wounding, indicating that the effect is not exclusive to stimulation of glycosyltransferase activity. However, the increased size of biglycan, which is attributable to the presence of longer GAG chains on the core protein (22,33), is apparent earlier than either up-regulation of protein core synthesis or RNA expression. Therefore, GAG chain synthesis and protein core expression may be independently regulated by cells. In addition, the peak of stimulation of biglycan core protein synthesis clearly precedes the peak increases in mRNA expression for biglycan after either bFGF treatment or monolayer wounding (compare Figs. 7 and 9). This observation suggests that the control of biglycan synthesis by bFGF may be post-transcriptional, and that the subsequent smaller increase in biglycan transcript levels may be an indirect effect, perhaps due to the stimulation of an intermediate factor. Recent studies have indicated that, when transfected into the tumor cell line, MG-63, biglycan promoter constructs as well as the transcription from the endogenous biglycan gene are not up-regulated by TGF-␤1 (39), despite an increase in biglycan RNA steady-state levels that is FIG. 9. Time course of steady-state RNA expression for biglycan after multiscratch wounding or bFGF treatment. RNA expression for biglycan, normalized to 28 S rRNA (Bar ϭ S.E.) was quantitated by scanning densitometry of autoradiographs in several experiments (n ϭ 4), in which total RNA was harvested at various times after multiscratch wounding (black) or 10 ng/ml bFGF treatment (gray) and Northern blots were performed (e.g. Fig. 8). *, p Ͻ 0.05; **, p Ͻ 0.01. induced by TGF-␤1 treatment. These data are consistent with the regulation of biglycan expression through post-transcriptional mechanisms, at least under those conditions. It remains possible, however, that activation of TGF-␤ in EC cultures after wounding may be responsible for the delayed up-regulation of biglycan RNA expression. Although a few studies have indicated that bFGF increases the expression of some PGs (40,41), cells may respond to other growth factors induced in wounded or bFGF-treated cultures. For example, transforming growth factor-␤1, which is an antagonist of bFGF, activates gene transcription and increases synthesis and deposition of fibronectin, collagens, and the PGs, versican and biglycan (1,22,33,(41)(42)(43) and is activated by proteases induced by bFGF (44). In recent work, Ku and D'Amore (45) have demonstrated that the upregulation of bFGF mRNA expression in response to release of bFGF following sublethal injury to endothelial cells can be abrogated by neutralizing anti-TGF-␤1 antibodies or antiprotease strategies that reduce the activation of TGF-␤. Because preliminary studies indicate that transforming growth factor-␤1 also increases biglycan synthesis in endothelial cells, 3 we cannot yet eliminate the possibility that such factors, activated or induced downstream from bFGF stimulation, are also involved in the control of biglycan metabolism.
Biglycan proteolytic processing, which results in the accumulation in the culture media of an N-terminal fragment of biglycan representing about one-half of the protein core, is dramatically stimulated in addition to increased biglycan synthesis by migrating or bFGF-treated cells. This observation raises several possibilities. First, if biglycan is required during the normal migration of cells after wounding, it may be that the functional portion of the molecule is one of the fragments that is generated, rather than the intact PG. Second, a migrationrelated cellular process may require the removal, rather than the increased production, of biglycan. While changes in biglycan expression and processing occur during cell migration, it remains to be tested whether biglycan processing at the cell surface is required for the modulation of cell-adhesive interactions or the utilization of growth factors important to endothelial cell movement.
The rapid degradation of biglycan probably occurs at or near the cell surface, since intact biglycan released to the media is not degraded with time, 3 and the N-terminal fragment accumulates in the culture supernatant, rather than in an intracellular compartment. Immunocytochemical staining indicates that cell-associated biglycan is distributed at the edges of lamellopodia of migrating cells. The localization of biglycan to these cellular structures is intriguing because these cell surface sites are critical both to cellular adhesive events and to the utilization of growth factors by cells, which are processes in which leucine-rich PGs may play a role. Urokinase plasminogen activator, which generates active plasmin from plasminogen, has been localized to focal adhesions (46), which are concentrated on cell processes. Both metalloproteinases and serine proteases, such as urokinase-plasminogen activators, are strongly induced in endothelial cells by bFGF (47), which promotes cell migration and angiogenesis (2). The induction of such proteases at the cell surface occurs concurrent to cell migration (1,2,48), although specific substrates are not fully known. However, substrates include matrix proteins, such as collagens (49), as well as cell surface-associated biglycan. Preliminary studies indicate that the serine protease inhibitor, aprotinin, inhibits the cleavage of biglycan in FGF-stimulated endothelial cell cultures, and that plasmin generates cleavage products from purified calf cartilage biglycan similar to those that are found in stimulated endothelial cell cultures, including a ϳ20-kDa N-terminal fragment. 3 These observations raise the possibility that plasmin, which is activated by urokinase induced by bFGF, may be responsible for biglycan processing during cell migration.
In summary, endothelial cell migration is controlled by a complex series of interactions between cells and both soluble growth regulatory molecules and the extracellular matrix. Evidence has accumulated that leucine-rich PGs, such as decorin and biglycan, are involved in the modulation of cellular adhesion and growth factor activity, and biglycan is present at cell surface sites that suggest a role in those processes. Clearly, biglycan expression and turnover are rapidly and transiently modulated as cells migrate following wounding. In addition, the alteration of biglycan metabolism after monolayer wounding is controlled by release of endogenous bFGF, which is a principal factor that regulates endothelial cell proliferation, migration, and differentiation (2). However, whether changes in biglycan expression or turnover are required for normal cell migration or the maintenance of a differentiated phenotype remain to be tested.