The Role of Syndecan Cytoplasmic Domain in Basic Fibroblast Growth Factor-dependent Signal Transduction*

To determine the role played by syndecan-4 cytoplasmic domain in the mediation of basic fibroblast growth factor (bFGF) signaling, immortalized human cells (ECV) were used to generate cell lines expressing constructs encoding full-length sequences for syndecan-4 (S4), syndecan-1 (S1), glypican-1 (G1), or chimeric proteins consisting of the ectoplasmic domain of glypican-1 linked to the transmembrane/cytoplasmic domain of syndecan-4 (G1-S4c) and the ectoplasmic domain of syndecan-4 linked to the glypican-1 glycosylphosphatidylinositol (GPI) anchor sequence (S4-GPI). Vector-transduced cells (VC) were used as controls. Expression of all these proteoglycans (except for the vector control) significantly increased cell-associated heparan sulfate mass and the number of low affinity bFGF-binding sites. However, in low serum medium, the addition of bFGF stimulated growth and migration of cells expressing S4 and G1-S4c constructs but not G1, S1, S4-GPI, or VC cells. Similar results were obtained using Matrigel growth assays. Mutations of heparan sulfate attachment sites on S4 construct abolished syndecan-4-dependent augmentation of bFGF responses. We conclude that cytoplasmic tail of syndecan-4 plays an important role in bFGF-mediated signal transduction.

Cell surface and extracellular heparan sulfate (HS) 1 are presumed to play an important role in interaction of heparinbinding growth factors with their corresponding high affinity receptors (1,2). Although the exact nature of this interaction is not clear, proposed models entail formation of a multimolecular complex involving heparan sulfates, a heparin-binding growth factor, and the corresponding growth factor receptor with the involvement (direct or indirect) of heparan sulfates in high affinity receptor dimerization and alteration in growth factorreceptor conformation allowing for a better "fit" (3)(4)(5). In the case of bFGF, a prototypical heparin-binding growth factor, HS-bFGF interaction is thought to lead to dimerization of bFGF that facilitates growth factor binding to FGF receptor 1 that is in turn followed by receptor dimerization and activation of intracellular signaling cascade. Therefore any alteration of HS chain composition on the cell surface or in the extracellular matrix by means of altered synthesis, degradation, or modification of glycosaminoglycan (GAG) chains can, conceivably, affect heparin-binding growth factor signaling.
The majority of heparan sulfates on the cell surface is associated with one of two classes of core proteins, syndecans and glypicans. The syndecan family is composed of four closely related proteins (syndecans-1, -2, -3, and -4) coded by four different genes. Syndecans-1 and -4 show expression in a variety of cell types including epithelial, endothelial, and vascular smooth muscle cells, although expression in quiescent tissues is fairly low (6,7). Syndecan-2 (fibroglycan) is expressed at high levels in cultured lung and skin fibroblasts, although immunocytochemically it is barely detectable in most adult tissues (6). Syndecan-3 (N-syndecan) demonstrates a much more limited pattern of expression, being largely restricted to peripheral nerves and central nervous system tissues, although high levels of expression are present in neonatal heart (8). Syndecans are capable of carrying both HS and chondroitin sulfate (CS) chains although most of syndecan-associated biological effects, including regulation of blood coagulation, cell adhesion, and signal transduction are largely thought to be due to the presence of HS chains capable of binding growth factors, cell adhesion receptors, and other biologically active molecules (5,9). At the same time, relatively little attention has been paid to the function of core proteins themselves.
All syndecans are membrane-spanning proteins possessing highly conserved C-terminal and transmembrane domains. The cytoplasmic domain contains four universally conserved tyrosines, a conserved serine and a C-terminal EFYA sequence of amino acids comprising a PDZ-binding region. Several observations suggested that syndecan expression may be related to cell proliferation. Thus, mesenchymal cell growth during tooth organogenesis is also associated with transient induction of syndecan-1 gene expression (10). In adult tissues, expression of syndecan-1 and syndecan-4 is increased in arterial smooth muscle cells after balloon catheter injury (11), in healing skin wounds (12,13), and in the heart following myocardial infarction (14). The effects of changes in syndecan expression on cell behavior are not well understood. Overexpression of syndecan-1 in 3T3 cells led to inhibition of bFGF-induced growth (15), whereas in 293T cells it augmented serum-dependent growth (16). Furthermore, syndecan-1 overexpression increased intercellular adhesion in lymphoid (17) and Raji cells (18) while decreasing the ability of B-lymphocytes to invade collagen gels (19).
The glypican core protein family consists of five murine and human members and a Drosophila homologue, dally (5). Unlike syndecans, glypicans are fully extracellular proteins attached to the cell membrane via a glycosylphosphatidylinositol (GPI) anchor. Only one family member, glypican-1, has been identified in endothelial cells. Another unique feature of glypicans is that they carry essentially only HS chains (20). Whereas little is known about their biological function, glypicans appear able to stimulate FGF receptor 1 occupancy by bFGF and promote biological activity of FGF and TGF-␤ family members (21).
Although the effect of syndecan expression on cell function is thought to be largely mediated via extracellular HS chains, a number of observations point to a key role of cytoplasmic domains in syndecan function. Recent studies have shown that syndecan cytoplasmic domain can interact with a number of proteins including PKC-␣ (22), syntenin (23), and CASK/lin-2A (24,25). The purpose of the present study, therefore, was to assess the relative contributions of extracellular HS chains versus syndecan core protein expression in endothelial cells.

vector and
BamHI/EcoRI fragment of rat glypican-1 into BglII/EcoRI sites of the same vector. Syndecan/glypican chimeras were created via PCR mutagenesis, cloned into the pCDNA3 vector, sequenced, and shuttled into the MSCV2.2 vector. Syndecan-4-GPI (S4-GPI) construct was created by deleting C-terminal end of rat syndecan-4 sequence starting with Gln 247 and replacing it with the C-terminal sequence of rat glypican-1 starting with Ser 510 . Glypican-syndecan-4 cytoplasmic domain (G1-S4c) construct was created by replacing C-terminal sequence of rat glypican-1 starting with Ser 510 with amino acids 247-321 of the rat syndecan-4 sequence. The created chimera thus contains both transmembrane and cytoplasmic regions of syndecan-4. Syndecan-4 construct with mutated GAG attachment sites (S4⌬GAG) was created by point mutations of Ser 44 3 Ala, Ser 65 3 Ala, and Ser 67 3 Ala. Full-length rat syndecan-1 cDNA and syndecan-4 cDNAs were cloned into the pGRE (29) expression vector. Transduction of the MSCV2.2 vector alone as well as transfection of Neo R was used to generate control ECV cell populations.
Generation of Cell Lines-For retroviral transductions, the virus was produced by calcium phosphate transient transfection (30) of 10 g of each construct on amphotropic Phoenix packaging cells (ATCC). Viral supernatants were collected after 36, 48, and 72 h, sterile-filtered through 0.2-m filter, and then transferred to ECV304 cells at 32°C in the presence of 25 g/ml DEAE-dextran. Typical viral titers in the supernatant were approximately 68⅐10 5 infectious particles/ml. Virus exposure was repeated 4 times for each construct; following the last exposure the cells were cultured in 10% FBS/M199 supplemented with 400 g/ml active G418 (Sigma) for 2 weeks.
Stable transfections were carried out with supercoiled, Qiagen column-purified plasmid DNA using the calcium phosphate precipitation method. At 16 -220 h post-transfection, cells were trypsinized and seeded again. For isolation of stable transfectants, the cells were plated in duplicate 100-mm culture dishes. After 10 -14 days of growth in G418 containing medium (1.2 mg/ml), single colonies were picked and grown in 24-well plates before further growth in tissue culture flasks. Appropriate construct expression was documented using RT-PCR and Western analysis (see below). No induction of GRE promoter was used for functional studies.
Growth and Migration Assays-For growth assays, 100,000 cells were plated in 6-well cell culture plates and allowed to attach overnight. At that time, the cells were washed 3 times with phosphate-buffered saline (PBS), and the medium was changed to M199 supplemented with 0.25% FBS. Twenty four hours later, 25 ng/ml bFGF (Chiron Corp.) were added to the cell culture medium. Cell counts were then obtained at 24-h intervals starting with the time of exposure to bFGF by trypsinizing the well and counting cell suspension on a Coulter counter (Coulter Corp.).
Migration assays were carried out using modified Boyden chambers (Neuroprobe, Inc). ECV304 cells and derived clones were grown in 10% FBS/M199 supplemented with 25 ng/ml DiI ((3) 1,1-dioctadecyl-3,3,3Ј,3Ј-tetramethylindocarbocyanide perchlorate (DiIC 18 ), Molecular Probes) living cell fluorescent stain overnight. Following that, the cells were trypsinized, washed with M199, diluted in M199 supplemented with 0.5% FBS, and seeded in wells at 60,000 cells per well. The cell containing compartments were separated from the lower wells by 25 ϫ 80-mm polycarbonate filters with 8-m pores (Poretics Corp.). The lower chambers were filled with 0.5% FBS/M199 supplemented with 50 ng/ml bFGF, and the entire apparatus was incubated in a tissue culture incubator at 37°C, 5% CO 2 for 4.5 h. After that time non-migrating cells were removed by washing the upper wells with PBS; the upper surfaces of the filters were scraped with a plastic blade, and the filters were fixed in 4% formaldehyde for 1 min and placed on a glass slide. The migrated cells were imaged using a digital SesSys TM camera attached to a Nikon fluorescent microscope. For each slide, 10 non-overlapping high power fields were selected for analysis. Following image acquisition using PMIS image processing software (Photometrics, Ltd.) the number of cells was automatically determined using Optimas 6.0 software (Bioscan, Inc.) Matrigel Growth Assay-Growth factor-depleted Matrigel TM (Beckton Dickinson) plates were prepared by adding 0.5 ml of thawed Matrigel TM to a well of a refrigerated 24-well tissue culture plate. The gel was allowed to solidify for 1 h at 37°C and overlaid with 1 ml of 0.5% FBS/M199 containing 30,000 cells. The cell culture was carried out at 37°C in a humidified atmosphere supplemented with 5% CO 2 . The analysis of cell growth was carried out by obtaining low (ϫ 10) and high (ϫ 40) power images of the wells with a digital SesSys TM camera focused on the surface of the gel using an inverted Nikon fluorescent microscope. The cell-covered area was then determined using Optimas 6.0 software.
Core Protein Expression Level Studies-To document increased expression of appropriate core proteins in newly generated cell lines, Western blotting was used to determine the expression of syndecan-4 and syndecan-1 core proteins, respectively. In the case of syndecan-4 expression studies, cultured cells were washed with PBS and trypsinized for 10 min at 37°C (0.05% trypsin, 0.5 mM EDTA) to remove extracellular domains of syndecans. The cell suspension was again washed three times with PBS, lysed in a lysis buffer (50 mM HEPES, pH 7.4, 1% Triton X-100, 150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 20 mM NaF, 20 mM sodium pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 ), resuspended in Laemmli buffer, and subjected to SDS-PAGE on 16.5% Tris glycine gel (Bio-Rad). Western blotting was carried out using a cytoplasmic tail-specific antiserum for syndecan-4 (kind gift of Dr. N. Shworak, Beth Israel Deaconess Medical Center, Boston) as described previously (31). For syndecan-1 expression studies, cultured cells were washed with PBS, lysed as described above, and digested with Flavobacterium heparatinase 1, 2, and 3 (0.25 units/l, Sigma) and chondroitinase ABC (0.005 units/l, Sigma). Following another round of PBS washes, cell lysates were separated on 12% SDS-PAGE and subjected to Western blotting with anti-syndecan-1 ectodomain antibody (kind gift of Dr. A. Rapraeger, University of Wisconsin). Blotting with anti-␤-actin antibody (Sigma) was used to adjust syndecan-4 and syndecan-1 expression levels for protein loading.
Expression levels of glypican-1 and syndecan-4-GPI constructs were confirmed by RT-PCR analysis. To this end, total RNA was isolated using TRI Reagent TM (Sigma). The RNA pellet was dissolved in RNasefree water and ethanol-precipitated. For RT-PCR analysis, 0.2 g of total RNA were used for reverse transcription with a 15 pmol of oligo(dT) 20 primer, 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, 0.5 mM each dNTP in 50 mM Tris-HCl, pH 8.3, buffer. The mixture was heated to 70°C for 10 min and then cooled to 37°C, and 1 l of Super Script II reverse transcriptase (200 units/l, Life Technologies, Inc.) was added. The reaction was allowed to proceed for 1 h at 37°C and then terminated by heating to 65°C for 5 min followed by chilling to 4°C. 1 l of the RT reaction mixture was used for PCR amplification using specific primers. The PCR reaction was carried out in the presence of 1.5 mM MgCl 2 , 0.2 mM dNTP, 400 nM 3Ј and 5Ј primers, and 2.5 units of Taq DNA polymerase (Roche Molecular Biochemicals). The following specific primers were used: glypican-1, 5Ј CCC CGC CAG CAA GAG CCG GAG CT and 3Ј GTG AGG CTC TGG GCG AGT GGC GG; syndecan-4, 5Ј (with SacI restriction site) ATA GAG CTC TTG GAA CCA TGG CGC CTG TCT GCC and 3Ј (GPI) GGA ATT CCA GGT TTT ATT ATC TTT TTA TC.
For standardization purpose a conserved region of human and mouse glyceraldehyde-3-phosphate dehydrogenase gene was chosen for amplification as a control template. The following primers were used: 5Ј CGT ATT GGG CGC CGT GTC ACC AGG GC and 3Ј GGC CAT GAG GTC CAC CAC CCT GTT GC. All PCR reactions were carried out using GeneAmp PCR 2400 system (Perkin-Elmer) as follows: 94°C (1 min), 50 -55°C (30 s), 72°C (1.5 min). The additional final extension step was performed at 72°C for 7 min. A total of 30 cycles was done for each reaction. Following PCR amplification, reaction products were subjected to 1% agarose gel electrophoresis. All experiments were repeated three times.
Determination of Heparan Sulfate Mass in Cultured Cells-To determine the total mass of heparan sulfate chains, endothelial cell cultures are washed twice with PBS and incubated for 24 h with 2 mCi of Na 2 35 SO 4 in 2 ml of a modified basal Eagle's medium supplemented with 1% Neutrodoma-SP. At the end of labeling, cells are washed with cold PBS and incubated with a lysis buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris, pH 7.4, 1 mM EDTA, 1 mM MgCl 2 , 1 mM iodoacetate, 100 M phenylmethylsulfonyl fluoride, 0.02% NaN 3 ) on ice for 20 min, followed by centrifugation at 15,000 ϫ g for 10 min at 4°C. Total proteoglycans are isolated from the supernatant by DEAE chromatography as described (32). Glycosaminoglycans were then cleaved from the total proteoglycan pool by ␤-elimination, and the relative content of HS and CS was determined by appropriate enzyme digests with chondroitinase ABC or Flavobacterium heparatinase 1, 2, and 3 (Sigma). Preliminary experiments on microvascular endothelial cells demonstrated that the sum of HS and CS sulfate accounted for Ͼ98% of the total proteoglycan sulfate.
Scatchard Analysis of Low Affinity bFGF-binding Sites-For determination of the number and affinity of bFGF heparan sulfate-binding sites, endothelial cells were grown to near-confluence in 24-well dishes in 10% FBS/M199. After two washes with cold PBS, 200 l of binding buffer (25 mM HEPES, pH 7.4, 0.1% bovine serum albumin, 0.05% gelatin in M199 medium), 6 ϫ 10 6 cpm (0.5 ng/ml) 125 I-bFGF (DuPont, specific activity 2000 Ci/mmol), and increasing amounts (0 -600 ng/ml) of cold bFGF were added to each well. The cells were incubated at 4°C for 2 h with gentle agitation; at the end of that time, the cells were washed three times with 1 ml of PBS containing 0.1% bovine serum albumin and then incubated with 1% Triton X-100 in 5 ml of water supplemented with 0.01% bovine serum albumin (Sigma) for 30 min at room temperature with vigorous shaking. Following this, 0.4-ml aliquots were counted in 1272 CliniGamma counter (Amersham Pharma-cia Biotech). Cell counts determined by Coulter Counter were employed to establish the number of cells per well. Background counts were subtracted from all samples. Scatchard analysis of the specifically bound material versus the molar amount of cold competitor was carried out using Origin 5.0 software (Microcal Software, Inc.). All experiments were carried in triplicate and repeated at least twice.

Constructs and Cell
Lines-To investigate the role of cellassociated heparan sulfate carrying core proteins in cell function, we screened a number of immortalized cell lines for expression of syndecan-4 and glypican-1. Of these, spontaneously immortalized human cell line ECV304 cell line exhibited the lowest expression of these proteoglycans (data not shown) and was chosen for the study. ECV304 cells have a complex genotype combining markers of a human bladder carcinoma line T-24 and human endothelial cells. In particular, the endothelial markers include expression of vascular endothelial growth factor receptor Flt-1 2 and von Willebrand factor. Furthermore, these cells respond to endostatin with dose-dependent inhibition of migration (33).
ECV304 cells were transducted with retroviral constructs containing full-length cDNAs for either syndecan-4 or glypican-1. In addition, in order to differentiate potential biological effects secondary to increased mass of cell surface and/or extracellular heparan sulfates versus increased core protein expression themselves, two other chimerical constructs as well as a syndecan-4 mutant construct were created. In one, S4-GPI, syndecan-4 extracellular domain was linked to glypican-1 GPI anchor, and in the other, G1-S4c, the extracellular domain of glypican-1 was linked to the trans-membrane and cytoplasmic domains of syndecan-4 (Fig. 1). Cells transfected with a vector only construct (MSCV-VC) were used as control. In a separate series of studies, stable transfections were used to generate ECV304 cells lines overexpressing full-length syndecan-4, syndecan-1, or a construct encoding a syndecan-4 mutant incapable of carrying GAG chains (S4⌬GAG) under control of an inducible GRE promoter (Fig. 1). Vector-transfected (GRE-Neo) as well as wild type ECV cells were used as control.
Analysis of derived cell populations following G418 selection demonstrated appropriate expression of all constructs (Fig. 2) in both retrovirally transduced and stably transfected cell lines. Quantification of protein levels using syndecan-4 cytoplasmic domain-specific antibody demonstrated a 2.5-3.0-fold increase in expression of syndecan-4 in both MSCV-S4 and in MSCV-G1/S4c chimera ( Fig. 2A). A 2-fold increase in syndecan-1 protein (Fig. 2B) and syndecan-4 (not shown) was noted in GRE stables. Appropriate expression of glypican-1 S4-GPI constructs was demonstrated by RT-PCR in the absence of suitable antibodies (Fig. 2C).
Increased expression of syndecan-4, syndecan-1, and glypican-1 constructs as well as syndecan/glypican chimeras may be expected to result in larger numbers of heparan sulfate chains on the cell surface. To test this prediction, we determined the total cell-surface mass of heparan sulfate chains in the wild type as well as four newly generated ECV cell lines. Total heparan sulfate mass was significantly increased (per g of total cellular protein) in MSCV-S4, MSCV-G1, MSCV-S4-GPI, MSCV-G1-S4c, and GRE-S1 but not MSCV-VC or GRE-S4⌬GAG cells (Table I). The increase in the total HS cell mass (2-3-fold) correlated well with corresponding increase in core protein expression as determined by Western blotting.
In order to assess whether these changes in heparan sulfate expression resulted in selective alterations of binding of heparin binding growth factors, we examined the binding of bFGF. Scatchard analysis of the wild type and newly generated ECV cell lines showed that there were no significant changes in the affinity of bFGF binding to either its low (heparan sulfates) or high (FGF receptor) affinity receptors (Table I). At the same time, there was a 3-fold increase in the number of low affinity bFGF-binding sites in the GRE-S1, a 2.0 -2.5-fold increase in MSCV-S4 and MSCV-G1-S4c cells, and a somewhat smaller increase in MSCV-G1 and MSCV-S4-GPI cells (Table I). The smaller increase in the cell-associated heparan sulfate mass in glypican and syndecan-4 GPI overexpressors is expected given higher shedding rates for GPI-linked glypican compared with the transmembrane syndecan.
The increase in the number of bFGF-binding sites was of the same order as the increase in the total HS cell mass suggesting that there was no preferential creation of bFGF-binding sites. This is further demonstrated by no significant change in the bFGF-HS/HS ratio (calculated as ratio of a relative increase in the number HS-bFGF sites per cell and a relative increase in the total HS mass). Thus, for a MSCV-S4 clone compared with control, there was a 5.94⅐10 6 /2.32⅐10 6 ϭ 2.56-fold increase in the number of bFGF-HS sites and a 0.33/0.14 ϭ 2.36 increase in the total HS mass per g of protein, giving the HS-bFGF/HS ratio of 2.36/2.56 ϭ 0.92. Similar analysis demonstrated no significant change for HS-bFGF/HS ratio for any of the transductants.
Functional Impact of Core Protein Expression-To study the effect of syndecan-4, syndecan-1, and glypican-1 expression on ECV cell growth, we analyzed the ability of wild type and newly created ECV cell lines to grow in vitro in response to bFGF. Comparison of growth rates between ECV cells transfected with a GRE driven or transduced with a retroviral promoter syndecan-4 expression construct showed no differences between the two lines. Likewise, there was no difference in growth between respective ECV controls (GRE-transfected and MSCV-transduced ECV cells, data not shown). Therefore, only retrovirally transduced ECV cells were used for subsequent syndecan-4 expression studies.
Shed ectodomains of HS-carrying core proteins may influence growth behavior of EC due to their ability to bind bFGF. Therefore, additional experiments were carried out to demonstrate that a variable rate of shedding by various ECV cells lines used in these studies was not responsible for the observed growth rate differences. To this end, we tested growth rate of MSCV-S4 cells in cell culture medium conditioned by either wild type ECV cells or MSCV-S4-GPI, MSCV-S4, or MSCV-G1 cells in the presence of bFGF. In all cases, growth rate remained the same (Fig. 3B) suggesting that ectodomain shedding had no effect on the observed differences in growth rates between endothelial cell lines expressing various core protein constructs.
To test the effect of the expression of these constructs on the ability of the cells to form vascular structures, wild type and newly generated ECV clones were plated on growth factordepleted Matrigel plates in the presence of medium supplemented with bFGF. Three days later, the formation of definable structures (cords and rings) was assayed by light microscopy. As in the case of in vitro growth assays, MSCV-S4 and MSCV-G1-S4c cells formed more numerous and denser vascular structures compared with wild type ECV, MSCV-G1, or MSCV-S4-GPI cells (Fig. 4). GRE-S4⌬GAG and GRE-S1 cells actually demonstrated reduced ability to grow and form structures on Matrigel compared with control cells. Quantification of the cell density on the Matrigel confirmed increased growth of MSCV-S4 and MSCV-G1-S4c cells (Fig. 5). It should be noted, however, that such analysis may underestimate a true growth response because of the three-dimensional nature of the assay.
To analyze further the effect of syndecan, glypican, or syndecan/glypican chimeras expression on biological behavior of endothelial cells, we analyzed the wild type and generated ECV cell lines for migration toward serum and bFGF in Boyden chamber assays. Similar to growth assay results, cells lines expressing increased amounts of syndecan-4 or G1-S4c chimeras demonstrated significantly higher ability to migrate compared with wild type ECV or ECV expressing syndecan-1, glypican-1, or extracellular domain of syndecan-4 linked to the glypican-1 GPI anchor (Fig. 6). As in the case of growth assays, GRE-S4⌬GAG cells demonstrated somewhat enhanced migration compared with vector-transduced ECV cells. However, this increase in migration was significantly smaller than that of MSCV-S4 and MSCV-G1-S4c cells. DISCUSSION Heparin-binding growth factors require heparan sulfates on the cell surface for effective binding to their high affinity re-ceptors (6). In addition to facilitating growth factor-receptor interactions, cell surface and extracellular matrix heparan sulfates may act to sequester the growth factor that makes them inaccessible for receptor binding or limits accessibility to the receptor. Thus, changes in the cell surface and/or the extracellular matrix HS content may up-regulate or down-regulate the ability of heparin binding growth factors to induce cell growth and migration. Such changes in the cell surface/matrix HS content can occur either as a result of shedding (34,35), as a consequence of activity of physiologic degradation (36), or due to an increase in the expression of HS-carrying core proteins. Selective changes in the composition of HS chains may arise due to activity of various sulfotransferases (5). An increase in HS proteoglycan core protein expression is observed under a variety of conditions including various stages of tissue differentiation and development (7), wound healing (13), response to ischemic injury such as myocardial infarction (14), and as a result of growth factor stimulation (9). In the present study we set out to differentiate the effect of changes in the core protein versus HS expression. To this end, we examined the effect of overexpression of three key cell surface HS-carrying core proteins-syndecans 1 and 4 and glypican-1, in ECV cells.
To facilitate analysis of functional impact of changes in core protein versus heparan sulfate expression, we took advantage of unique features of syndecan-4 and glypican-1 structure to create chimeras designed to give insight into the function of these components. The choice of these two cores is dictated by a number of considerations. Syndecan-4 is a transmembrane protein, whereas glypican-1 is attached to the cytoplasmic membrane via a GPI anchor. However, despite these structural differences, both have been implicated in the bFGF signaling (21,37). This creates an opportunity, then, to dissect the contributions of cytoplasmic domain versus ectoplasmic domains and GAG chains. Syndecan-4 is the most widely expressed syndecan family member (5), whereas glypican-1 is the only member of glypican family detected in endothelial cells (38). Although glypican GAG chains are almost exclusively HS (39), syndecan-1 and syndecan-4 carry both HS and CS chains (32). Thus, expression of an S4-GPI chimera would be expected to produce a somewhat lower total HS mass then expression of glypican-1 itself, whereas expression of the G1-S4c chimera would be expected to generate somewhat larger number of HS chains than expression of syndecan-4. Analysis of HS chain expression in the transductants demonstrated expected increase in the HS cell mass in cells expressing native and chimeric proteoglycan constructs, whereas the expression of the retroviral vector itself had no effect on the HS mass. The increase in the HS cell mass in the MSCV-G1 and MSCV-S4-GPI clones was substantially less than in GRE-S1, MSCV-S4, and MSCV-G1-S4c clones. This difference is likely due to a higher shedding rate of GPI-anchored cores (39) compared with TABLE I Effect of S4, G1, and syndecan/glypican chimera construct expression on heparan sulfate cell mass and bFGF binding The abbreviations used are: HS-bFGF, HS sites capable of binding bFGF; ECV-VC, vector controls (MSCV2.2 transduced) ECV cells; ECV-S4, syndecan-4-transduced ECV cells; ECV-G1, glypican-1-transduced ECV cells; ECV-S4-GPI, syndecan-4-GPI chimera-transduced ECV cells; ECV-G1-S4c, glypican-1 syndecan-4 transmembrane/cytoplasmic domain chimera-transduced ECV cells; ECV-S1, syndecan-1-transfected ECV cells; ECV-S4⌬GAG, ECV cells transduced with syndecan-4 construct lacking GAG attachment sites; [ 35   affinity of HS chains for bFGF. Therefore, the overall increase in the HS mass did not result in preferential generation of bFGF-binding sites or a change in bFGF affinity. Overexpression of syndecan-4 gene in ECV cells resulted in a significant increase in the cell growth and migration in response to bFGF. However, overexpression of glypican-1 despite comparable, al-beit smaller, increase in the total HS mass and the number of bFGF-binding sites did not increase cell growth and migration in response to bFGF. On the other hand, expression of G1-S4c cytoplasmic domain chimera (MSCV-G1-S4c cells) closely paralleled enhanced growth and migration responses seen in syndecan-4 overexpressors, strongly suggesting that the presence of syndecan-4 cytoplasmic domain was needed for this effect. This deduction was further confirmed by a lack of increased growth or migration in cells expressing extracellular domain of syndecan-4 linked to the glypican GPI anchor.
One potential explanation for the difference in growth rates between various constructs in these studies is the difference in extent of shedding of ectodomains that can affect cell growth rate due to accumulation of HS chains in the matrix. However, exposure of fast growing ECV-S4 cells to media conditioned by other cell lines, including ECV-G1 (presumed to have the highest shedding), failed to influence its growth. We conclude, therefore, that ectodomain shedding could not account for the differences in growth seen in these studies.
To investigate whether the observed bFGF-dependent stimulation of ECV-S4 and ECV-G1-S4c clones growth is unique to syndecan-4 cytoplasmic domain and to better control for the effect of overall changes in the HS cell mass, we generated stable cell lines overexpressing syndecan-1 construct using the GRE promoter. To minimize differences between cell lines, additional cell lines were generated using GRE-S4 expression. Comparison of GRE-S4 and MSCV-S4 cells did not demonstrate any difference in any of functional assays including twoand three-dimensional growth and migration. Despite similar increases in the HS cell mass between clones overexpressing syndecan-4 and syndecan-1, MSCV-S1, unlike MSCV-S4 cells, did not demonstrate enhanced growth when challenged with bFGF in two-dimensional culture. Furthermore, MSCV-S1 clones actually appeared to be substantially inhibited in the Matrigel growth assay. This retardation of growth following syndecan-1 overexpression is consistent with prior observations (15). Thus, increased growth of MSCV-S4 and MSCV-G1-S4c overexpressors is attributable to the presence of unique sequences found syndecan-4 cytoplasmic domain.
In order to address the role of HS chains versus core protein expression per se, additional studies were carried out using S4⌬GAG construct. The magnitude of the increase in syndecan-4 cytoplasmic tail expression at the protein level was similar in GRE-S4⌬GAG and GRE-S4 or MSCV-S4 cells. At the same time, the expression of this construct did not affect overall HS cells mass or the number of low affinity bFGF-binding sites per cell. In both two-dimensional and Matrigel growth assays GRE-S4⌬GAG cells demonstrated significantly lower proliferative response to bFGF administration compared with wild type ECV cells. Similar results were seen in migration assays. We conclude from these experiments that the presence of HS chains is required for syndecan-4 mediation of bFGF response.
A number of previous studies have provided conflicting data with regard to the effect of syndecan overexpression on the bFGF responsiveness (15,16,18,21). A variety of factors including differences in cell types, bFGF amounts, and cell culture conditions can all contribute to the observed differences. In the current study, the chosen cell line ECV possesses a relatively low level of endogenous syndecan expression. In addition, bFGF stimulation of growth and migration was examined in low serum setting using growth-arrested cells.
Whereas syndecan-1, syndecan-4, and glypican-1 expression increased the total HS cell mass, there were no significant changes in the number of low affinity HS bFGF-binding sites as a proportion of total HS sites. Thus, increased responsiveness to bFGF in MSCV-S4 versus GRE-S1 or MSCV-G1 cells cannot be attributed to an increase in the number of HS-bFGF-binding sites but rather to increased expression of syndecan-4 cytoplasmic domain. This is best illustrated in the case of GRE-S1 cells that had the largest increase in the total HS mass (0.36 Ϯ 0.05 versus 0.14 Ϯ 0.026 [ 35 S]HS/mg of protein in MSCV-VC cells) and in the number of HS-binding sites per cell (7.01 Ϯ 1.08 ϫ 10 6 versus 2.32 Ϯ 0.40 ϫ 10 6 in MSCV-VC cells) but failed to demonstrate enhanced growth when stimulated with bFGF.
The suggestion that syndecan-4 cytoplasmic domain may play a unique role in response to bFGF-dependent stimulation of growth and migration of endothelial cells is not altogether unexpected. Unlike other syndecans, syndecan-4 cytoplasmic domain contains a unique 9-amino acid insert capable of binding and activating PKC-␣ (22,40,41) in Ser 183 phosphorylationdependent manner (42) which, in turn, is the subject of exquisite control by a bFGF-activated serine phosphatase and a novel PKC (31). Previous studies have shown that increased expression of PKC-␣ in endothelial cells is associated with a similar change in endothelial cell behavior, namely accelerated growth and migration (43), whereas inhibition of PKC activity prevents bFGF-dependent activation of endothelial cell growth (44).
Alternatively, increased bFGF responsiveness of cells carrying constructs containing syndecan-4 cytoplasmic domain could have been due to the ability of this domain to interact with cellular proteins other then PKC-␣. In particular, all syndecans share a PDZ binding domain, and PDZ-dependent interactions have recently been implicated in a number of cell signaling events. However, the absence of enhanced growth/migration response following syndecan-1 overexpression strongly argues against this possibility. bFGF was chosen in these studies as a prototypical heparinbinding growth factor. In principle, nothing so far suggests that increased responsiveness of syndecan-4 or G1-S4c chimeraexpressing cells is in any way specific to bFGF as opposed to any other heparin-binding growth factor. We did not observe any preferential increase in the proportion of HS sites capable of bFGF binding nor was there any change in the bFGF affinity of HS binding. Therefore, it is possible that the same results would have been obtained with any other heparin-binding growth factor capable of stimulation of cell growth and migration. On the other hand, the absence of increased bFGF responsiveness of the cells expressing glypican-1 or S4-GPI chimera does not rule out that responsiveness to other growth factors is not altered in these cells. In particular, TGF-␤ may be an especially likely candidate given studies suggesting glypican involvement in TGF-mediated signal transduction (45,46).
In summary, increased expression of syndecan-4 or glypicansyndecan-4 cytoplasmic domain in endothelial cells leads to increased growth and migration in response to bFGF. These observations suggest that the syndecan-4 cytoplasmic region may play an important role in the bFGF signal transduction.