Normal Human Keratinocytes Bind to the α3LG4/5 Domain of Unprocessed Laminin-5 through the Receptor Syndecan-1*

Basal keratinocytes of the epidermis adhere to their underlying basement membrane through a specific interaction with laminin-5, which is composed by the association of α3, β3, and γ2 chains. Laminin-5 has the ability to induce either stable cell adhesion or migration depending on specific processing of different parts of the molecule. One event results in the cleavage of the carboxyl-terminal globular domains 4 and 5 (LG4/5) of the α3 chain. In this study, we recombinantly expressed the human α3LG4/5 fragment in mammalian cells, and we show that this fragment induces adhesion of normal human keratinocytes and fibrosarcoma-derived HT1080 cells in a heparan- and chondroitin sulfate-dependent manner. Immunoprecipitation experiments with Na2 35SO4-labeled keratinocyte and HT1080 cell lysates as well as immunoblotting experiments revealed that the major proteoglycan receptor for the α3LG4/5 fragment is syndecan-1. Syndecan-4 from keratinocytes also bound to α3LG4/5. Furthermore we could show for the first time that unprocessed laminin-5 specifically binds syndecan-1, while processed laminin-5 does not. These results demonstrate that the LG4/5 modules within unprocessed laminin-5 permit its cell binding activity through heparan and chondroitin sulfate chains of syndecan-1 and reinforce previous data suggesting specific properties for the precursor molecule.

trimer (␣␤␥) by forming a triple-stranded coiled-coil structure through their ␣-helical domains. At present, 15 heterotrimers termed LN-1 to LN-15 have been described with different subunit composition selected from five individual ␣ chains (␣1-␣5), three ␤ chains (␤1-␤3), and three ␥ chains (␥1-␥3) (1,2). All LN ␣ chains comprise a large globular domain in their carboxylterminal region (G domain), which consists of five homologous globular subdomains of about 200 amino acids each (LG1-LG5). LN-5, with chain composition ␣3␤3␥2, is a component of basement membranes underlying specialized epithelia with secretory or protective function (3). In skin, LN-5 is synthesized by keratinocytes initially as a high molecular mass precursor protein of 460 kDa of which the ␣3 and ␥2 chains undergo specific processing to smaller forms after being secreted and deposited into the extracellular matrix (ECM) (4,5). Processing of the ␣3 chain consists of cleavage of the carboxyl-terminal globular domains 4 and 5 (LG4/5) (6,7). An additional cleavage within the aminoterminal domain IIIa of the ␣3 chain occurs subsequently and may also be important for LN-5 function (8,9). Processing of the ␥2 chain occurs at the amino terminus with loss of a 50-kDa domain (9 -11). Processed LN-5 is the major component of anchoring filaments in skin (12) where it mediates cell adhesion via interaction of the ␣3 carboxyl-terminal LG1-3 triplet domain with both ␣ 3 ␤ 1 and ␣ 6 ␤ 4 integrins (13)(14)(15)(16). Several studies have suggested that processed LN-5 functions both in the nucleation of hemidesmosome assembly through interaction with the ␣ 6 ␤ 4 integrin and as an adhesive factor that retards cell motility (6,(17)(18)(19)(20), while other findings indicate that LN-5 containing the unprocessed ␣3 chain is capable of inducing cell migration and impedes hemidesmosome assembly (20,21). Moreover keratinocytes have been shown to deposit unprocessed LN-5 while migrating on collagen I (22) or after stimulation by tumor growth factor-␤1 (23). In vivo, epidermal injury activates the transcription and deposition of LN-5 into the provisional matrix by the leading keratinocytes in the process of epidermal outgrowth and migration at the wound edge (24 -26). The LN-5 precursor is found in this provisional matrix but is absent from mature basement membranes (21,25). These data indicate that the presence of the LG4/5 domain in the ␣3 chain of LN-5 may influence cell behavior.
Several studies conducted with native and recombinantly expressed LG subdomains from various laminin isoforms have highlighted a potential cellular function of the LG4/5 domain. Purification of the cleaved LG4/5 domain from conditioned HaCat cells or normal human keratinocyte (NHK) media has permitted identification of the cleavage site in the spacer re-gion between the G3 and G4 domains (7,27). Heparin binding properties of this fragment, combined with its cellular interaction activities, have allowed identification of a heparan sulfate proteoglycan (HSPG)-type cellular receptor. A recent study shows that syndecan-2-and -4-overexpressing 293T cells interact with recombinant ␣3LG4 (28), this interaction being believed to be responsible for induction of matrix metalloproteinase-1 expression in keratinocytes and fibroblasts through the mitogen-activated protein kinase signaling pathway (29). A further report shows binding of syndecan-1 from human submandibular gland cells and mouse melanoma cells B16F10 to a synthetic peptide corresponding to residues 1399 -1410 derived from the LG4 sequence of murine LN-1 and LN-2 (30).
Which syndecans from normal skin keratinocytes bind to the LG4/5 domain and whether these syndecans bind unprocessed LN-5 remain important questions to be answered. To address these questions, we recombinantly expressed the human ␣3LG4/5 fragment, and we show that syndecan-1 from NHKs and other cells is the major receptor for this fragment. Syndecan-4 also bound to a much lesser extent. We purified unprocessed LN-5 and show for the first time that it binds syndecan-1, while the processed form does not.
Production of the Recombinant ␣3LG4/5 Fragment in Mammalian Cells-A DNA fragment encoding the human laminin ␣3LG4/5 domain (nucleotides 4057-5142) was generated by PCR. cDNA was obtained by reverse transcribing total RNA of SCC25 epithelial cells using the primer 5Ј-GCATTGGGTAATCAGTGCTTTTGAAGG. Two oligonucleotides flanking the desired sequence were designed. One, corresponding to the 5Ј-end of the domain, carried an SpeI site (underlined) (5Ј-CCTTACTAGTTTGCTCACCACTTCCCAAG), and the second, corresponding to the 3Ј-end of the domain, introduced a NotI site and a stop codon (5Ј-CCAAGCCTATTTCACAGCAAGGCGGCCGCTATG). The resulting PCR product of 1083 base pairs was restriction-digested with SpeI and NotI and was inserted in-frame with the BM-40 signal peptide in the mammalian expression vector pCEP-Pu (32). All inserts and borders were fully sequenced. Recombinant plasmid was introduced by electroporation into the human embryonic kidney cell line 293-EBNA (Invitrogen). Transfected cells were selected with 0.5 g/ml puromycin and grown to confluence. Secretion of the recombinant LG4/5 fragment (rLG4/5) into the medium was confirmed by SDS-PAGE of conditioned medium samples from transfected and wild type cells.
Purification of the Recombinant ␣3LG4/5 Fragment in Mammalian Cells-To produce the recombinant protein, 293-EBNA cells were cultured in serum-free Dulbecco's modified Eagle's medium, and the conditioned medium was collected every 48 h. The medium was kept at Ϫ80°C and proteolytic degradation was inhibited by the addition of 50 M N-ethylmaleimide and 50 M phenylmethanesulfonyl fluoride until use. For purification of rLG4/5, conditioned medium was applied to a 1-ml HiTrap heparin column (Amersham Biosciences), and elution was achieved with a linear NaCl gradient from 0.1 to 1.2 M. Protein purity was checked by SDS-PAGE where it was estimated to be greater than 90%. Protein concentration was determined by microprotein assay using bicinchoninic acid (BCA) (Pierce).
Purification of the Native ␣3LG4/5 Fragment as Well as the Processed and Unprocessed Native LN-5-The native LG4/5 was purified from conditioned NHK medium as described previously (27). Processed LN-5 was purified from the culture medium of human SCC25 cells as described previously (15), and unprocessed LN-5 was purified according to the protocol described by Amano et al. (9). Briefly, 1 liter of conditioned medium from NHKs cultured in KBM-2 medium with 0.015 mM CaCl 2 (Clonetics) was precleared with gelatin-Sepharose and applied to a 1-ml HiTrap heparin column. The bound materials were eluted with a linear NaCl gradient from 0.15 to 0.75 M. Fractions eluting around 0.5 M were precipitated using trichloroacetic acid and further analyzed for their unprocessed LN-5 by transfer to nitrocellulose followed by immunodetection with Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences) or by Coomassie staining of the gel (Biosafe, Bio-Rad).
Flow Cytometry-Monolayer cells were harvested following exposure to PBS containing 5 mM EDTA, resuspended in PBS containing 2% BSA, and centrifuged. Cells (3 ϫ 10 5 /sample) were suspended in staining solution (PBS, 1% BSA) containing saturating amounts of mAbs. After a 30-min incubation at 4°C, cells were washed and suspended in staining solution containing fluorescein isothiocyanate-goat anti-mouse IgG (Silenus, Eurobio, Les Ulis, France) or fluorescein isothiocyanategoat anti-rabbit IgG (Jackson Immunoresearch, Beckman Coulter, Roissy, France) depending on the primary antibody. After further incubation at 4°C for 30 min, cells were washed, fixed, and analyzed with a flow cytometer (FACScan, BD Biosciences). A 2 g/ml propidium iodide solution was added to one sample to determine the population of dead cells. Controls were carried out using isotype-matched antibodies or by omission of primary antibody.
Cell Adhesion and Inhibition Assays-Multiwell tissue culture plates (Costar, Dutscher, Brumath, France) were coated with the indicated concentrations of processed LN-5 or LG4/5 substrates by overnight adsorption at 4°C. Cells were detached by incubation with PBS containing 5 mM EDTA and rinsed in the corresponding serum-free medium prior to the experiment. After saturation of the wells with 1% BSA, the plates were used immediately for cell adhesion assays (at the indicated number of cells per well) in serum-free medium as detailed previously (15). The extent of adhesion was determined after fixation of adherent cells with 1% glutaraldehyde in PBS followed by staining with 0.1% crystal violet and absorbance measurements at 570 nm using an MR5000 enzyme-linked immunosorbent assay reader. A blank value corresponding to BSA-coated wells (Ͻ5% of maximal cell adhesion) was subtracted. Adherent cells were photographed using a phase-contrast microscope equipped with a camera (Olympus Corp., Lake Success, NY). For cell adhesion inhibition experiments with glycosaminoglycans (GAGs), the coated wells were preincubated with heparin, heparan sulfate (HS) from bovine intestinal mucosa, chondroitin 4-sulfate (C4S) from bovine trachea, or chondroitin 6-sulfate (C6S) from shark cartilage (Sigma) for 1 h at room temperature, and adhesion was measured in the presence of the inhibitors. In all experiments, each assay point was derived from triplicate measurements.
Treatment of Cells with Enzymes Prior to Cell Adhesion and Interaction Studies-HT1080 cells and NHKs were treated with 25 g/ml cycloheximide for 2 and 1 h, respectively. All the following processes contained the same concentrations of cycloheximide. After detachment with PBS containing 5 mM EDTA, the cells were washed four times in serum-free medium and then suspended in the same medium, and the calcium concentrations were adjusted to 2 mM followed by treatment with 8 milliunits/ml heparitinase I and/or 50 milliunits/ml chondroitinase ABC (Seikagaku America, Goger, Paris, France) for 1 h at 25°C. For control experiments, cells were incubated in parallel with the corresponding buffers only. In cell adhesion experiments, the cells then were inoculated into the 96-well plates in the presence (or absence for the controls) of the same enzymes. For interaction studies on nitrocellulose, cells were homogenized with RIPA buffer (50 mM Tris/HCl containing 0.15 M NaCl, 1% Nonidet P40, 0.5% deoxycholic acid, and 0.1% SDS, pH 7.4) (as below) in ice. Cell lysates were cleared by centrifuga-tion and immediately incubated with membranes containing LG4/5.
Immunoprecipitation-NHKs and HT1080 cells were metabolically labeled with 100 Ci/ml sodium [ 35 S]sulfate (Na 2 35 SO 4 ) (PerkinElmer Life Sciences) for 16 h. After labeling, the cell layers were washed with PBS and extracted with RIPA buffer on ice for 10 min. All the following procedures were performed at 4°C. Extracts were incubated for 2 h with Gamma Bind-G Sepharose beads (Amersham Biosciences) previously incubated overnight with anti-syndecan-1 pAb (H-174), anti-syndecan-4 mAb (5G9), or anti-nLG4/5 pAb-rLG4/5 complex. The latter complex was obtained by incubating rLG4/5 with the beads carrying anti-nLG4/5 pAb for 2 h. In some experiments heparin was added during incubation. Beads carrying the immune complexes were washed three times with RIPA buffer, and finally the complexes were separated by heating at 95°C in denaturing sample buffer followed by separation using a 3-10% SDS-polyacrylamide gradient gel under reducing conditions. The gel was subjected to autoradiography followed by image analysis with a Storm gel and blot imaging system (Amersham Biosciences). For reimmunoprecipitation experiments, immune complexes formed on the beads carrying the anti-nLG4/5 pAb-rLG4/5 complex were eluted by heating at 95°C in 50 mM Tris/HCl containing 1 M NaCl, pH 7.4. The salt concentration of the eluate was brought to 0.25 M by dilution followed by reimmunoprecipitation overnight at 4°C with beads carrying pAb H-174 and mAb 5G9 or beads carrying IgG from a preimmune rabbit. After incubation, the beads were washed with PBS-Tween (0.05%) and processed as above. In the experiments conducted with processed and unprocessed LN-5, the procedure was identical except that the mAb 6F12 against the laminin ␤3 chain was used to bind both forms of LN-5 to the Gamma Bind-G Sepharose beads.
For detection of syndecans by immunoblot, rLG4/5 and processed or unprocessed LN-5 were covalently bound to beads to avoid constraints produced by immunoglobulin detection. The ligands were conjugated to CNBr-activated Sepharose 4B (Amersham Biosciences) at 1 mg of protein/ml of resin as described by the manufacturer. Beads were incubated with cell extracts and washed as described above. Beads were finally transferred in digestion buffer (20 mM sodium acetate containing 5 mM CaCl 2 , pH 7.0) and incubated with 8 milliunits/ml heparitinase I and 50 milliunits/ml chondroitinase ABC for 2 h at 25°C. The samples were then prepared and subjected to electrophoretic migration in a 10 or 15% SDS-polyacrylamide gel under reducing conditions. Proteins were transferred to nitrocellulose followed by immunodetection with Western Lightning Chemiluminescence Reagent Plus.
Interaction of Syndecan-1 with Immobilized Proteins-BSA or rLG4/5 (1 g) were blotted on a nitrocellulose filter previously soaked in distilled water and, after blotting, saturated overnight with BSA (5 mg/ml). Membranes were rinsed in PBS and then incubated for 2 h at ϩ4°C in the various cell lysates. After repeated washes of the membranes with RIPA buffer and one wash with PBS, syndecan-1 was immunodetected with pAb H-174 as described below.
Biotinylation Procedure and Heparan Sulfate Immobilization-HS (from intestinal mucosa, Celsus) was biotinylated at the reducing end as described previously (33). Briefly HS resuspended in PBS at 1 mM was incubated for 24 h at room temperature with 10 mM biotin-long chain-hydrazine (Pierce). The mixture was then extensively dialyzed against water to remove unreacted biotin and freeze-dried. Two flow cells of an F1 sensor chip (Biacore) were activated with 50 l of a mixture of 0.2 M 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, 0.05 M N-hydroxysuccinimide before injection of 50 l of streptavidin (0.2 mg/ml in 10 mM acetate buffer, pH 4). Remaining activated groups were blocked with 50 ml of 1 M ethanolamine, pH 8.5. Typically this procedure allowed the coupling of ϳ2500 -3000 resonance units (RU) of streptavidin. Biotinylated HS, 50 g/ml in HBS-P buffer (10 mM Hepes, 0.15 M NaCl, 0.005% P20, pH 7.4), was then injected on one of the two streptavidin-activated surfaces (the other one being a negative control) to yield an immobilization level of 30 -40 RU. Both flow cells were then conditioned with several injections of 2 M NaCl. For binding and competitive analyses, rLG4/5, preincubated or not with GAGs, was injected over the HS surface at a flow rate of 80 l/min after which the complexes formed were washed with HBS-P buffer. The sensor chip surface was regenerated twice with a 1-min pulse of 0.05% SDS followed by a 2-min pulse of 2 M NaCl in HBS-P buffer.
Analytical Methods-The following procedures were performed as previously described: SDS-PAGE followed by the electrophoretic transfer of proteins to nitrocellulose with immunoblot analysis (15). Silver staining was carried out with the Silver Stain Plus kit according to the manufacturer's instructions (Bio-Rad).

RESULTS
Production and Purification of Recombinant ␣3LG4/5 Fragment-rLG4/5 was purified from the culture medium of 293-EBNA cells by affinity chromatography using a 1-ml HiTrap heparin column. Coomassie Blue staining of the fractions eluting at around 0.8 M NaCl revealed a band of about 40 kDa (Fig.  1, A and B), which corresponded to the predicted size of rLG4/5. Immunoblot analysis of these fractions, using pAb against nLG4/5 (27), revealed the peptide as the rLG4/5 domain (Fig.  1C). As previously described for the native fragment, rLG4/5 bound to the heparin affinity column required a NaCl concentration of 0.8 M to be eluted, demonstrating that a high affinity for binding to heparin is characteristic of this fragment. Comparison of native and recombinant LG4/5 by silver staining and immunoblot analysis revealed an identical electrophoretic pattern for both fragments (Fig. 1D).
Cell Binding Activity of the Recombinant ␣3LG4/5 Fragment-We tested the ability of rLG4/5 to induce adhesion of normal cells and different cell lines. As shown in Fig adhesion for all three cell types was obtained at concentrations of 10 g/ml. The strong binding capacity of rLG4/5 to heparin suggested that adhesion might be mediated by cell surface HSPGs. This was verified by adding soluble heparin, which strongly inhibited cell adhesion to rLG4/5 (Fig. 2B). Inhibition was dose-dependent with half-maximal inhibition achieved in the range of 0.05-0.1 g/ml. The involvement of cell surface HSPG receptors in cell adhesion to rLG4/5 was verified by the use of CHO cells known to express HSPG receptors as well as mutants deficient in GAG synthesis. Wild type CHO-K1 cells were able to adhere to rLG4/5 in a dose-dependent manner (Fig. 2C). Three mutant cell lines derived from CHO-K1 were tested in adhesion experiments: CHO-677, deficient in HS but synthesizing more CS, CHO-745, which lacks both HS and CS, and CHO-618, deficient in all the GAGs (31). While mutant CHO-618 cells did not adhere at all to rLG4/5, the two mutant cells CHO-677 and CHO-745 exhibited a moderate to very low interaction with rLG4/5, respectively (Fig. 2C).
Because NHKs are in direct contact with LN-5 in vivo and because they are normal cells, their adhesion capacity to rLG4/5 was analyzed and compared with that of HT1080 cells (Fig. 3). Both these cell types were chosen for further characterization of the receptor involved in the interaction. Integrinmediated adhesion to processed LN-5 lacking the LG4/5 domain was also performed. NHKs attached to rLG4/5 in a dosedependent manner (Fig. 3A), but the adhesion was lower compared with that obtained with processed LN-5. HT1080 cells revealed a similar profile (Fig. 3B). Microscopic analysis of the cells after staining with crystal violet revealed that cell morphology was markedly different (Fig. 3C). Cells plated on wells coated with processed LN-5 exhibited a typical spread phenotype with numerous cell-cell contacts. In contrast, NHKs and HT1080 cells attached but failed to spread on rLG4/5 resulting in few intercellular contacts. Occasionally both NHKs and HT1080 cells produced numerous cytoplasmic processes resembling filopodia as shown by arrows in Fig. 3C. Detachment of HT1080 cells and NHKs by trypsin treatment (Fig. 4A) or detachment with 5 mM EDTA followed by heparitinase I and chondroitinase ABC treatment prior to the cell adhesion experiment largely reduced their adhesion to rLG4/5 (Fig. 4B), suggesting a non-integrin-mediated interaction. The use of 5 mM EDTA for cell detachment preserved their adhesion capacity and was used as the control. As expected, the integrin-mediated NHK and HT1080 cell adhesion to LN-5 was totally inhibited by incubation of the cells with 10 mM EDTA (Fig. 4C). Adhesion was partially affected when the cells were subjected to adhesion to rLG4/5 in the same conditions. To the contrary, heparin totally inhibited adhesion of NHKs and HT1080 cells to rLG4/5, but it had no effect when cells were plated on purified LN-5, confirming involvement of different types of cellular interaction on these two related substrates.
For further characterization of the GAGs involved in the FIG. 2. Cell adhesion characteristics of the recombinant LG4/5 fragment. A, dose-dependent cell adhesion to rLG4/5. Multiwell plates were coated with different concentrations of rLG4/5, and HT1080, HBL100, A431, and A375 cells were seeded at a density of 10 5 cells/well. Cells were incubated for 30 min at 37°C, and their adhesion was measured as described under "Experimental Procedures." Each point represents the average of triplicate wells. B, effect of soluble heparin on adhesion of HT1080, HBL100, and A431 cells to rLG4/5. Multiwell plates were coated with rLG4/5 at 10 g/ml. After saturation with 1% BSA, the wells were incubated with the indicated concentration of heparin for 1 h at room temperature, and the cells were seeded in the presence of the same concentration of heparin. The cells were incubated as in A, and the extent of adhesion was measured as above and expressed as percentage of adhesion in the absence of heparin. C, doseresponse curves of different CHO cell lines adhered to rLG4/5. CHO-K1 is the control cell line, CHO-677 is deficient in HS and expresses more CS, CHO-745 lacks both HS and CS, and CHO-618 lacks all the GAGs. Wells were coated with rLG4/5 at the indicated concentrations, and after saturation with 1% BSA, the CHO cells were seeded at a density of 10 5 cells/well. Cells were incubated for 30 min at 37°C, and their adhesion was measured as described under "Experimental Procedures." Each point represents the average of triplicate determinations. Abs., absorbance. interaction, rLG4/5-coated wells were preincubated with free GAG chains prior to NHK and HT1080 cell adhesion. As shown in Fig. 5, A and B, the HS greatly inhibited adhesion of both cell types to rLG4/5 at the concentration of 100 g/ml. While C4S produced partial but significant inhibition of cell adhesion at the same concentration (42% for HT1080 cells and 58% for NHKs), C6S had no effect. To further characterize the roles of HS and CS in cell binding, rLG4/5-coated wells were preincubated with different amounts of free HS and CS alone or in combination prior to HT1080 cell adhesion (Fig. 5C). The concentration of HS (10 g/ml) used in this experiment produced 43% inhibition. Similar to results shown in Fig. 5A, about 20% of the binding was inhibited if 50 g/ml C4S alone was used, and combining HS and C4S completely abrogated adhesion suggesting an additive effect of the two tested GAGs. As expected, combining HS and C6S did not result in a stronger inhibiting effect than HS used alone. These results suggest that both HS and C4S could be involved in NHK and HT1080 cell adhesion to rLG4/5 and that each of them alone is sufficient for the interaction.
To further test this hypothesis, we examined HT1080 and NHK adhesion to rLG4/5 after digestion of HS by heparitinase I and CS by chondroitinase ABC (Fig. 5, D and E). For both cell types, heparitinase I treatment as well as chondroitinase treatment of the cells partially abrogated cell adhesion to rLG4/5; however, heparitinase I was more efficient. Incubation of rLG4/5 with C4S prior to cell adhesion totally abrogated adhesion of the heparitinase I-treated cells, while it did not modify that of chondroitinase ABC-treated cells. Treatment of the substrate with C6S had no effect in any case. Inversely incubation of rLG4/5 with HS totally inhibited cell adhesion of chondroitinase ABC-treated cells as well as that of heparitinase I-treated cells. These results suggest that HS and C4S may bind the same site with different affinities. The Biacore technology was used to confirm and to extend the analysis of the rLG4/5 interaction with HS in vitro. Surface plasmon resonance was used to measure changes in refractive index caused by the binding of rLG4/5 to immobilized biotinylated HS. Injection of rLG4/5 (1 g/ml, 27 nM) over an activated sensor chip containing 30 -40 RU of HS gave a signal of 350 RU, whereas injection of the protein over a control surface (containing streptavidin only) led to a signal of no more than 40 RU (not shown). Using this binding assay, inhibition analysis

FIG. 4. Inhibition of cell adhesion to recombinant LG4/5 fragment.
A, prior to cell adhesion to rLG4/5 (10 g/ml), HT1080 cells and NHKs were detached from the plastic plates by either trypsinization or incubation with PBS containing 5 mM EDTA as indicated on the graph. B, HT1080 cells and NHKs were pretreated with 25 g/ml cycloheximide and were detached with PBS containing 5 mM EDTA followed by treatment with 8 milliunits/ml heparitinase I and 50 milliunits/ml chondroitinase ABC for 1 h at 25°C as indicated. For controls, the cells were incubated with the corresponding buffers under the same conditions. C, HT1080 cells and NHKs were used for cell adhesion to rLG4/5 (10 g/ml) and processed LN-5 (5 g/ml) in the presence or absence of either heparin (10 g/ml) or EDTA (10 mM) as indicated on the graph. In all experiments, HT1080 cells and NHKs were seeded at densities of 6 ϫ 10 4 and 3 ϫ 10 4 cells/well, respectively, and were incubated for 30 min and 1 h, respectively. The extent of adhesion was measured as described under "Experimental Procedures" and expressed as percentage of adhesion in the absence of enzyme treatment or inhibitor.

FIG. 5. Binding to recombinant LG4/5 fragment involves both heparan and chondroitin sulfate GAGs.
A, B, and C, effect of GAGs on HT1080 and NHK cell adhesion to rLG4/5. The rLG4/5 coats (10 g/ml) were saturated and incubated with either free HS, C4S, or C6S for 1 h at room temperature at the indicated concentration prior to carrying out the assay. The extent of adhesion was measured as described under "Experimental Procedures" and expressed as percentage of adhesion in the absence of competitor. Each point represents the average of triplicate wells. D and E, implication of both heparan and chondroitin sulfate for HT1080 and NHK cell adhesion to rLG4/5. HT1080 cells were detached with PBS containing 5 mM EDTA followed by treatment with either 8 milliunits/ml heparitinase I or 50 milliunits/ml chondroitinase ABC for 1 h at 25°C as indicated on the graph. For the control, the cells were incubated with the corresponding buffers under the same conditions. Prior to the cell adhesion assay with the enzyme-treated cells, the rLG4/5 coats (10 g/ml) were saturated and incubated with either PBS, C4S, C6S, or HS at the indicated concentrations for 1 h at room temperature. HT1080 cells and NHKs were seeded at a density of 10 5 and 6 ϫ 10 4 cells/well, respectively, and were incubated for 30 min. The extent of adhesion was measured as described under "Experimental Procedures" and expressed as percentage of adhesion in the absence of enzyme treatment. was performed to identify which GAG was able to compete with immobilized HS to bind to rLG4/5. For this purpose, rLG4/5 was preincubated with a range of concentration of either HS, C4S, or C6S and then injected over the HS-activated sensor chip. As expected soluble HS inhibited the binding of rLG4/5 to immobilized HS, total inhibition being observed with HS concentration of 5 g/ml (Fig. 6A). C4S also inhibited the rLG4/ 5-HS interaction, but 100 g/ml of C4S was required for total inhibition (Fig. 6B), while C6S was much less active (Fig. 6C). Together these data are consistent with the view that HS and CS compete with one another to bind rLG4/5 with HS having a higher affinity than CS for the protein.
Identification of Syndecan-1 as the Major Receptor for ␣3LG4/5 Fragment-As cell adhesion to rLG4/5 was totally inhibited by soluble heparin or prevented by trypsinization or treatment of cells with heparitinase I and chondroitinase ABC, we looked for heparan sulfate-type cell surface receptors. We set up a cell lysate system in which receptors were solubilized while retaining the ability to interact with rLG4/5. Among the several lysis buffer systems tested, the RIPA buffer provided the best compromise between solubilization of the receptor of interest and the absence of precipitation. These lysates were therefore used in solid phase assays. Biotinylated rLG4/5 bound to coated HT1080 and NHK cell lysates in a dose-dependent fashion (not shown). Immunoprecipitation experiments were therefore performed to identify the receptor. Capture of the proteoglycan-type receptor was carried out by incubating Na 2 35 SO 4 -labeled NHK or HT1080 cell lysates with beads that were immunologically covered with rLG4/5. Autoradiographic analysis of the bound material revealed a unique diffuse band of ϳ250 kDa for both cell types (Fig. 7, A and B,  lane 1). To identify the bound receptor, immunoprecipitations were performed concomitantly with anti-syndecan antibodies, and an analysis of the electrophoretic pattern revealed syndecan-1 and possibly syndecan-4 as potential candidates (Fig. 7, A  and B, lanes 2 and 3). The identity of the rLG4/5-bound proteoglycan-type receptor as syndecan-1 and/or syndecan-4 was verified by dissociation of the complex followed by reimmunoprecipitation of the eluate with the corresponding antibodies (Fig. 7, A and B, lanes 5 and 6). For both NHKs and HT1080 cells, syndecan-1 was identified as the major rLG4/5 partner (Fig. 7, A and B, lane 5). While syndecan-4 from NHKs was also detected as bound to rLG4/5 (Fig. 7B, lane 6), that from HT1080 cells could not be detected (Fig. 7A, lane 6). Depletion of syndecan-1 from the HT1080 cell lysate did not improve the detection of the syndecan-4 binding to LG4/5 (not shown). As expected, binding of syndecan to rLG4/5 was prevented by heparin for both HT1080 and NHK cell lysates (Fig. 7C).
As syndecan-2 and -4 were shown to interact with recombinantly expressed ␣3LG4 (28), we analyzed by flow cytometry the level of expression of the four different syndecans at the cell surface of the NHKs and HT1080 cells used in this study (Fig.  8). As a control, we included the keratinocyte cell line HaCat. The expression pattern was analyzed and revealed that for the . Electrophoretic analysis of the bound material was performed on 3-10% gradient SDS-polyacrylamide gels under reducing conditions followed by autoradiography. C, soluble heparin prevents syndecan interaction with rLG4/5. Beads immunologically covered with rLG4/5 were incubated with HT1080 and NHK cell lysates in the absence (lanes 1 and 4) or presence of 10 g/ml (lanes 2 and 5) and 100 g/ml (lanes 3 and 6) heparin. three cells types, syndecan-1 was the most intensively expressed syndecan. Syndecan-4 was also expressed in the three cells types although in a much lower amount. While syndecan-2 was expressed in a very low amount in HaCat cells only, syndecan-3 could not be detected in any of the three tested cell types.
To confirm the identity of the syndecans bound to rLG4/5, we performed detection by immunoblotting after treatment of the bound proteoglycan receptor with heparitinase I and chondroitinase ABC (Fig. 9A). Immunoblotting with the pAb H-174 against syndecan-1 revealed an intense positive band migrating around 96 kDa, which corresponds to the molecular mass of the syndecan-1 core protein. Immunoblotting of the same material electrophoretically separated on a 15% gel with pAb H-140 against syndecan-4 and pAb M-140 against syndecan-2 did not reveal any positive band in the position of the core protein of either syndecan-4 or syndecan-2 (not shown), suggesting that these receptors were not present in sufficient amount to be detected. To study the mechanism of syndecan-1/LG4/5 binding, intact as well as HS-free, CS-free, and HSϩ CS-free GAGs from HT1080 and HaCat cells were compared to reveal the possible role of HS and CS in syndecan-1 binding (Fig. 9B). HT1080 and HaCat cells were either untreated or treated with heparitinase I, chondroitinase ABC, or both to degrade HS, CS, or both. BSA-or rLG4/5-containing membranes were incubated in intact or enzyme-treated cell lysates, and bound syndecan-1 was detected by immunoblotting. Treatment of the cells with either heparitinase I or chondroitinase ABC only partially prevented interaction of the syndecan-1 with LG4/5, while digestion of both HS and CS prevented the interaction. These results suggest that syndecan-1 expressed on these cells contains both HS and CS and confirm our cell adhesion experiment suggesting that both HS and CS bind LG4/5 independently.
Unprocessed LN-5 Is a Ligand for Syndecan-1-We then addressed the question of whether unprocessed LN-5, contain-  1, 3, and 5) or without (lanes 2, 4, and 6) rLG4/5. After washes, bound material was digested with heparitinase I and chondroitinase ABC. Electrophoretic analysis of the bound material was performed on a 8% SDS-polyacrylamide gel under reducing conditions followed by immunoblotting with the pAb H-174 against syndecan-1. The migration positions of molecular mass markers are shown on the right. B, effect of heparitinase I and chondroitinase ABC on syndecan-1-rLG4/5 interaction. HT1080 and HaCat cells were treated with heparitinase I and/or chondroitinase ABC to remove HS and/or CS side chains. Nitrocellulose containing BSA and LG4/5 (1 g) was incubated in solution containing lysates from untreated or treated cells as indicated in the figure. After washes, the bound syndecan-1 was detected by immunoblotting with the pAb H-174. ing the LG4/5 domain at the carboxyl-terminal extremity of its ␣3 chain, is a ligand for syndecan-1 and -4 in NHKs. Native unprocessed LN-5 from NHK-conditioned medium was separated from processed LN-5 and partially purified by binding to a HiTrap heparin column and subsequently eluting with PBS containing 0.45 M NaCl as reported previously (9). The unprocessed LN-5 content was analyzed by immunoblot (Fig. 10A,  lanes 1, 3, and 5) and Coomassie staining (not shown) and compared with purified ␣3-processed LN-5 (Fig. 10A, lanes 2, 4,  and 6)  from the processed control. These two forms of LN-5 with or without the LG4/5 domain were specifically immunocaptured on beads covered with mAb 6F12 and incubated with Na 2 35 SO 4 -labeled NHK lysate. Autoradiographic analysis of the bound material revealed a characteristic diffuse band of over 250 kDa in the unprocessed LN-5-bound fraction whose molecular mass was identical to that of syndecan-1 (Fig. 10B,  lane 3). No significant signal was detected in the processed LN-5-bound fraction (Fig. 10B, lane 2). This was further verified by detection of syndecan-1 by immunoblotting after treatment of the bound proteoglycan receptor with heparitinase I and chondroitinase ABC (Fig. 10C). Immunoblotting with the pAb H-174 against syndecan-1 specifically revealed a positive band corresponding to the syndecan-1 core protein in the material bound to rLG4/5 (lane 2) and unprocessed LN-5 (lane 3), while no signal was seen in the negative control (lane 1) and materiel bound to processed LN-5 (lane 4).

DISCUSSION
Numerous recent reports indicate that distinct biological cellular events can be assigned to LN-5 depending on the level of processing of its ␣3 and ␥2 chains. Fully processed LN-5 induces stable adhesion of NHKs and hemidesmosome formation, while unprocessed LN-5 is related to migratory situations (34). To understand the physiological significance of the carboxyl-terminal processing of the LN-5 ␣3 chain, we have recombinantly expressed the entire LG4/5 domain, which, in normal keratinocyte cultures, is rapidly removed after secretion and deposition of LN-5 into the ECM. 293-EBNA-expressed rLG4/5 shares the high affinity to heparin with the native domain and induced adhesion of several cell types including NHKs (7,27). Moreover, in a manner comparable to nLG4/5, cell adhesion to rLG4/5 was specifically inhibited by heparin, suggesting the involvement of cell surface HSPG. In addition, the abolition of NHK and HT1080 cell adhesion to rLG4/5 by trypsinization supported the speculation that the cell surface receptors are HSPGs and not integrins, which are trypsin-resistant. Inhibition studies showed that HT1080 and NHK cell adhesion to rLG4/5 occurs in a HS-and CS-dependent manner and that C4S participates in the interaction. This was suggested by cell adhesion studies using CHO cell mutants lacking the ability to introduce HS and CS chains upon their cell surface proteoglycans as well as by inhibition of NHK and HT1080 cell adhesion when they were treated with heparitinase I and chondroitinase ABC. We show that HS and C4S interact with LG4/5. Heparin alone blocked cell binding entirely, which indicates that the HS and C4S chains both bind to the same site in rLG4/5 as heparin. Moreover our interaction studies demonstrate that HS and C4S bind the same site in rLG4/5 with different affinities with the affinity of HS being higher than that of C4S. A low binding of C6S to rLG4/5 could also be detected with the Biacore approach but was not detected in cell adhesion studies suggesting a very low affinity of this interaction. We have subsequently identified syndecan-1 as a major receptor for rLG4/5 by immunoprecipitation of Na 2 35 SO 4 -labeled NHK and HT1080 cell lysates with beads carrying rLG4/5. We also detected syndecan-4 as a receptor for rLG4/5 in NHK cell lysates, a finding in accordance with a recent study suggesting that fibroblast-produced syndecan-4 binds to a recombinantly expressed ␣3LG4 fragment (28). Probably due to its insufficient amount, binding of syndecan-4 from HT1080 cells could not be detected in our experiments. Indeed  2, 4, and 6). Immunoblot analysis was performed with mAb BM165 against an epitope within LG1-3 (lanes 1 and 2), pAb against nLG4/5 (lanes 3 and 4), and pAb 4101 against ␣3, ␤3, and ␥2 chains of LN-5 (lanes 5 and 6). Reactions were visualized by chemiluminescence. After stripping, the same membrane was repetitively used for detection with the antibodies described above. Molecular mass markers are shown on both sides. Arrows on the right show molecular masses of unprocessed ␣3 chain (190 kDa), processed ␣3 chain (165 kDa), unprocessed ␥2 chain (155 kDa), ␤3 chain (140 kDa), and processed ␥2 chain (105 kDa). B, immunoprecipitation analysis of NHK cell lysate binding to unprocessed LN-5. A cell lysate of Na 2 35 SO 4 -labeled NHKs was immunoprecipitated with beads immunologically covered with unprocessed (lane 1) or processed LN-5 (lane 2). For molecular mass comparison, the lysate was also immunoprecipitated with beads carrying pAb H-174 against syndecan-1 (lane 3) or mAb 5G9 against syndecan-4 (lane 4). Precipitated samples were subjected to SDS-PAGE followed by autoradiography as described under "Experimental Procedures." C, NHK cell lysates were incubated with beads covered without protein (lane 1) or with rLG4/5 (lane 2), unprocessed LN-5 (lane 3), and processed LN-5 (lane 4). After washes, bound material was digested with heparitinase I and chondroitinase ABC. Electrophoretic analysis of the bound material was performed on an 8% SDS-polyacrylamide gel under reducing conditions followed by immunoblotting with the pAb H-174 against syndecan-1. The migration positions of molecular mass markers are shown on the right. evaluation of the cell surface expression level of the four different syndecans in HT1080 cells and NHKs revealed syndecan-1 as the most abundant, while the expression of syndecan-4 was low. We also found a high level of syndecan-1 in HaCat cells, a result at variance with previous findings, which reported an absence of syndecan-1 mRNA in these cells (28). Our result was confirmed by immunoblot detection of syndecan-1 from HaCat cell lysate as a major receptor bound to rLG4/5 in a manner comparable to that of HT1080 and NHK lysates.
Our results provide the first evidence that primary keratinocytes use syndecan-1 for adhesion to ␣3LG4/5. As expected from the cell adhesion study, the binding of Na 2 35 SO 4 -labeled syndecans from NHKs to rLG4/5 was prevented by heparin. As syndecan-1 core protein has been shown to carry either both HS and CS (36,37) or HS alone (38), our experiments show that these two GAG chains bind independently to rLG4/5 with different affinities. A recent report describes the characterization of a heparin-binding synthetic peptide corresponding to residues KPRLQFSLDIQT (amino acids 1399 -1410) derived from the murine ␣3LG4 sequence (39). This peptide induces adhesion of the melanoma B16F10 cells through an HSPG-type receptor that remains to be identified. This sequence differs from that proposed by others (28) where cell adhesive activity of the recombinantly expressed LG4 fragment is reported to be mainly located in the sequence corresponding to residues NS-FMALYLSKGR (amino acids 1412-1423) within the human ␣3 chain. An additional heparin binding activity has been detected within a bacterially expressed recombinant human ␣3LG5 fragment (16). Considering these distinct results, one cannot exclude the possibility that several heparin binding sites are present within the ␣3LG4/5 fragment. Site-directed mutagenesis within LG4/5 will allow the precise determination of the amino acid sequence involved in syndecan-1 interaction and will clarify the contribution to binding of its HS and C4S sulfated chains.
We addressed the question of whether syndecan-1 from NHKs binds to unprocessed LN-5. Several results strongly suggest that the LG4/5 fragment is effective when attached to the rest of the LN-5 molecule. For example, the free LG4/5 fragment is never found in the ECM of cultured NHKs or HaCat cells, while both the 190-and 165-kDa forms of the ␣3 chains are detected (7). In addition, free LG4/5 is not found in the ECM of migrating keratinocytes where only the 190-kDa unprocessed ␣3 chain is present (6,23,34). However, all previously published results have been obtained with LG4, LG5, or LG4/5 fragments. We show for the first time that unprocessed LN-5 binds syndecan-1, while processed LN-5 does not. We cannot exclude the possibility that syndecan-4 binds to LN-5. The low cell surface expression of syndecan-4 combined with the low concentration of the native ligands used in the experiment may have resulted in our assay not being sensitive enough.
While NHKs and HT1080 cells adhered to the rLG4/5 substrate, they failed to spread, although they did develop actincontaining processes resembling filopodia. Indeed syndecans have been shown to function as cell adhesion receptors (40), and syndecan-1 is known to promote actin cytoskeleton assembly (41). The conserved carboxyl-terminal tetrapeptide sequence present in all syndecans binds certain PDZ domaincontaining proteins, such as syntenin (42) and calcium/ calmodulin-dependent serine protein kinase (43), which may function as membrane scaffold proteins that bind signaling and structural proteins to the plasma membrane. Syndecans have also been implicated in adhesion-dependent signaling (35). Previous immunohistochemical analysis of syndecan-1 and -4 in NHK culture revealed that syndecan-1 and -4 are both ex-pressed by keratinocytes at cell-cell junctions and underneath the cell body suggesting involvement in adhesion (44,45). Previous studies have shown that unprocessed LN-5 is present exclusively in the ECM of growing and migrating keratinocytes (6,23,34). Recent data show that recombinantly expressed ␣3LG4 is responsible for induction of matrix metalloproteinase-1 synthesis in NHKs through interaction with a proteoglycan-type receptor (30). Future studies will determine whether syndecan-1 has a role to play in this process. Syndecans are known to be involved in numerous cellular processes including wound healing, and they also function as co-receptors, acting in concert with integrins and growth factor receptors (46,47). In addition, both syndecan-1 protein and mRNA levels are specifically and strongly induced in wound edge keratinocytes during wound healing (48 -51). This expression resembles that of unprocessed LN-5 deposited in the provisional basement membrane by leading keratinocytes during wound healing (6,25), reinforcing our results showing its interaction with syndecan-1.