A unique sequence of the laminin alpha 3 G domain binds to heparin and promotes cell adhesion through syndecan-2 and -4.

Laminin-5, consisting of the alpha 3, beta 3, and gamma 2 chains, is localized in the skin basement membrane and supports the structural stability of the epidermo-dermal linkage and regulates various cellular functions. The alpha chains of laminins have been shown to have various biological activities. In this study, we identified a sequence of the alpha 3 chain C-terminal globular domain (LG1-LG5 modules) required for both heparin binding and cell adhesion using recombinant proteins and synthetic peptides. We found that the LG3 and LG4 modules have activity for heparin binding and that LG4 has activity for cell adhesion. Studies with synthetic peptides delineated the A3G75aR sequence (NSFMALYLSKGR, residues 1412--1423) within LG4 as a major site for both heparin and cell binding. Substitution mutations in LG4 and A3G75aR identified the Lys and Arg of the A3G75aR sequence as critical for these activities. Cell adhesion to LG4 and A3G75aR was inhibited by heparitinase I treatment of cells, suggesting that cell binding to the A3G75aR site was mediated by cell surface heparan sulfate proteoglycans. We showed by affinity chromatography that syndecan-2 from fibroblasts bound to LG4. Solid-phase assays confirmed that syndecan-2 interacted with the A3G75aR peptide sequence. Stably transfected 293T cells with expression vectors for syndecan-2 and -4, but not glypican-1, specifically adhered to LG4 and A3G75aR. These results indicate that the A3G75aR sequence within the laminin alpha 3 LG4 module is responsible for cell adhesion and suggest that syndecan-2 and -4 mediate this activity.

Laminin-5 (␣3␤3␥2) is a component of anchoring fibrils and forms a complex with the hemidesmosome apparatus and stabilizes the basement membrane structure by forming supramolecular complexes with laminin-6 and -7, collagen VII (8,9), and fibulin-2 (10). Mutations in laminin-5 (11,12) cause congenital skin blister diseases junctional epidermolysis bullosa. Disruption of the laminin ␣3 gene in mice resulted in abnormal hemidesmosomes and blister formation in the skin (13). Laminin-5 is also shown to be an adhesive substrate for keratinocytes (14). In these processes, the ␣ 3 ␤ 1 integrin was identified as a cellular receptor for motility and ␣ 6 ␤ 4 integrin for anchorage (15)(16)(17)(18)(19). In cultured keratinocytes, laminin-5 is secreted in an unprocessed form containing a 200-kDa ␣3 chain, a 140-kDa ␤3 chain, and a 155-kDa ␥2 chain and then processed into a mature form containing of a 165-kDa ␣3 chain, a 140-kDa ␤3 chain, and a 105-kDa ␥2 chain (20). Proteolytic processing of the ␥2 chain by MT1-MMP is suggested to increase cell migration activity of laminin-5 (21). Similar to other laminin ␣ chains, the ␣3 chain contains a large C-terminal globular domain (G domain), which is composed of five tandem homologous modules (LG1-LG5) each of about 200 amino acids, forming an autonomous folding unit (22). The LG modules of the ␣1 and ␣2 chains have been identified as binding sites for heparin, nidogen-2, fibulin-1 and -2, ␣-dystroglycan, and syndecan-1 (23-26; for review, see Ref. 27). Recently, heparin binding activity of the ␣4 chain (28) and heparin-sensitive cell adhesion activity of the ␣5 chain have also been identified in the LG modules (29). It was shown that the C-terminal part of the G domain of the ␣3 (18,31) and ␣4 (32) chains was proteolytically cleaved and released from the ␣ chains after secretion into the cultured medium. The unprocessed laminin ␣3 chain (200 kDa) was found in cell layers of the provisional edge of the cell sheet in cultured keratinocytes. In vivo, the unprocessed ␣3 chain has been identified especially in the newly synthesized epidermal basement membrane in wounds (18, 31; for review, see Ref. 33). However, the biological function of the C terminus globular modules of the ␣3 chain is unknown.
In the present work, we used recombinant proteins to discover that the LG4 module of the laminin ␣3 chain has activity for heparin binding and cell adhesion. We delineated this activity to the sequence NSFMALYLSKGR by mutation analyses with recombinant LG4 and synthetic peptides. We demonstrated with affinity chromatography and DNA transfection assays that syndecan-2 and -4 are responsible for LG4-mediated cell adhesion.
Cultured Cells-Human neonatal dermal fibroblasts were obtained form Asahi Techno Glass Co., Tokyo, Japan. The HaCat human keratinocyte cell line was a kind gift of Dr. Fusenig (German Cancer Research Center, Heidelberg, Germany). 293T cells were derived from 293 cells as a highly transformed human renal epithelial line expressing two viral oncogenes, adenovirus E1a and SV40 large T antigen (34). These cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Life Technologies, Inc.) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 50 units/ml streptomycin.
Recombinant LG Domain Expression-cDNA was synthesized using mRNA from human fibroblasts and HaCat cells by the Superscript Preamplification System (Life Technologies, Inc.) and amplified by the Expand High Fidelity PCR System (Roche Diagnostics, Basel, Switzerland). A DNA sequence for the 27-amino acid signal peptide of the laminin ␥2 chain was inserted into the XhoI and NheI-BamHI sites of the MO90 vector, which expresses the human IgG Fc portion under the EF1-␣ promoter (35). cDNA for the LG modules of the human laminin ␣3 chain (GenBank TM L34155) was amplified and cut by AvrII and cloned into the NheI site of the above vector. The expression vector encodes a recombinant protein with the ␥2 signal peptide, a laminin ␣3 LG module, and IgG Fc. All constructs were confirmed by DNA sequencing using a thermo sequenase Cy5 dye terminator and ALF express II DNA sequencer (Amersham Pharmacia Biotech).
Purification of Recombinant LG Modules-Recombinant proteins were expressed in 293T cells by the Ca-P transfection kit (Invitrogen, Groningen, Netherlands). Conditioned media (CM) with 1.5% fetal bovine serum in DMEM were collected for 4 days at 12-h intervals. After addition of 2 mM phenylmethylsulfonyl fluoride and protease inhibitor mixture, including 7 g/ml pepstatin A (Sigma), 2 M leupeptin (Sigma), 10 mM benzamidine (Sigma), and 26 g/ml aprotinin (Sigma), CM were frozen and kept at Ϫ80°C until use. CM were precipitated with 50% ammonium sulfate and centrifugation at 15,000 ϫ g for 30 min at 4°C, followed by dialysis against PBS (phosphate-buffered saline), pH 7.4. The dialyzed samples were applied on protein A-Sepharose (Amersham Pharmacia Biotech) and washed with 0.1% Nonidet P-40 (Sigma), PBS and 1 mM EDTA, 20 mM Tris-HCl, pH 7.4. The bound materials were eluted with 0.1 M acetic acid and neutralized by 1 M Tris-HCl, then concentrated with Centricon YM-30 (Millipore, Bedford, MA). The protein concentration was calculated by the BCA protein assay kit (Pierce). Purity of proteins was determined by reducing 10% SDS-PAGE and Coomassie Brilliant Blue staining.
Heparin-Sepharose and Protein A-Sepharose Affinity Chromatography-After transfection, CM were collected, and 1 ml of CM was adjusted to 0.40 M NaCl and 0.5% Triton X-100 and incubated with 60 l of heparin-Sepharose (Amersham Pharmacia Biotech) for 3 h at room temperature or overnight at 4°C. The bound materials were eluted with SDS buffer for Western blotting after rinsing four times with 1 ml of 0.40 M NaCl, phosphate buffer, pH 7.4, and 0.5% Triton X-100. In protein A-Sepharose affinity chromatography, 60 l of protein A-Sepharose was mixed with 1.0 ml of CM. Competition assays were performed by incubation with various amounts of glycosaminoglycans (GAGs), including heparin, chondroitin sulfate A, dermatan sulfate, and hyaluronic acid (Seikagaku Kogyo, Tokyo, Japan).
Synthesis of Laminin ␣3LG4 Peptides-All peptides of the laminin ␣3LG4 were manually synthesized by the Fmoc (9-fluorenylmethoxycarbonyl)-based solid-phase methods with a C-terminal amide as described previously (6). Peptides were purified by reverse-phase high performance liquid chromatography using a Vydac 5C18 column and a gradient of water/acetonitrile containing 0.1% trifluoroacetic acid. Purity of the peptides was confirmed by analytical high performance liquid chromatography. Identity of the peptides was confirmed by a Sciex API IIIE triple quadrupole ion spray mass spectrometer.
Solid-phase Assays-Heparin-albumin (Sigma) was coated on Nunclon surface 96-well plates (Nalge Nunc International, Rochester, NY) and air-dried overnight. The amount of heparin-albumin was changed from 0 to 200 ng/well to obtain saturation curve and 2 ng/well of heparin-albumin was used as coated substrates for LG4 binding assays. After a 4-h block at room temperature with blocking buffer (10% nonfat dry milk, PBS, 0.1% Tween 20), purified recombinant proteins were incubated overnight at 4°C with gentle shaking in 50 l of blocking buffer containing 0.40 M NaCl. After rinse with PBS, 0.1% Tween 20, the wells were incubated with biotinylated anti-human IgG (ϫ2,000), followed by the incubation with alkaline phosphatase-conjugated streptavidin (ϫ5,000) for 1 h at room temperature. Bound recombinant proteins were detected with alkaline phosphatase substrate p-nitrophenyl phosphate using a microplate reader at A 405 nm (Toyo Soda, Tokyo, Japan). Background OD without heparin coating was subtracted from each OD value. For competition assays using synthetic peptides, 200 g/ml amounts of each peptide and 0.5 g/ml of LG4 were mixed during incubation on wells coated with heparin-albumin (2.0 ng/well).
Cell Adhesion Assays-Substrates used were BM165 affinity-purified processed laminin-5, which does not contain the LG4 -5 modules (kindly provided from Dr. Amano, Shiseido Co., Tokyo, Japan) (36), collagen I (Nitta zeratin, Tokyo, Japan), recombinant LG modules, and ␥2-s (laminin ␥2 N-terminal 617 amino acids). Surface 96-well plates were coated with substrates in distilled water overnight at 4°C, rinsed with water, and blocked with 1% bovine serum albumin (BSA) (Sigma) in DMEM for 1 h at room temperature. Cells were trypsinized and recovered in 10% fetal bovine serum, DMEM for 30 min at 37°C in 5% CO 2 in a humidified atmosphere followed by three rinses with 10 ml of 0.1% BSA, DMEM. After rinsing, 100 l of the cell suspension (2 ϫ 10 5 /ml) in 0.1% BSA, DMEM was plated on the wells for 1 h at 37°C in 5% CO 2 in a humidified atmosphere. The wells were rinsed once with PBS and stained with 0.1% crystal violet in 20% methanol (v/v) for 10 min. The attached cells were counted in three randomly selected fields (ϫ200). For competition assays, cells were preincubated for 10 min at room temperature with synthetic peptides or heparin, heparan sulfate, chondroitin sulfate A, dermatan sulfate, chondroitin sulfate C, and hyaluronate (Seikagaku Kogyo).
Heparitinase I Treatment-Enzyme treatment was performed as described previously (26). Cells were precultured for 2 h in the presence of 2 g/ml cycloheximide to inhibit protein synthesis. Then, cells were collected as described above in the presence of cycloheximide. Cells were incubated with 50 milliunits/ml heparitinase I (Seikagaku Kogyo) for 90 min at 37°C, with mixing every 15 min, followed by adhesion assays as described above.
Fibroblast Syndecan-2 Analysis-Early confluent dermal fibroblasts from three 75-cm 2 flasks were rinsed with PBS and scraped into 10 ml of PBS with a cell scraper and centrifuged, resuspended in 50 l of DMEM, and incubated at 37°C for 60 min in the presence or absence of heparitinase I (50 milliunits/ml). The cells were lysed with 30 l of lysis buffer (2% Triton X-100, PBS in the presence of 2 mM phenylmethylsulfonyl fluoride and protease inhibitor mixture) for 60 min on ice with occasional vortexing. The lysates were centrifuged for 30 min at 15,000 ϫ g at 4°C, and the supernatants were boiled in SDS buffer and separated on 15% SDS-PAGE, followed by Western blotting with antisyndecan-2 antibody (1:500) (Santa Cruz) and peroxidase-conjugated anti-goat IgG (Sigma) and ECL (Amersham Pharmacia Biotech). For affinity chromatography with recombinant fusion proteins, cells were scraped into PBS, washed, and lysed with lysis buffer as described above. The lysates were centrifuged, and the supernatants (20 mg/ml) were frozen at Ϫ80°C until use. The supernatant (80 l) was diluted with 500 l of 0.40 M NaCl, 0.1% Triton X-100, phosphate buffer, pH 7.4, and incubated with 3.0 g of GST-LG4-N or GST protein overnight at 4°C. The bound materials were eluted with 10 mM glutathione, 10 mM Tris-HCl, pH 8.0, followed by dialysis against distilled water. Heparan sulfate chains were removed by treatment with heparitinase I (1 milliunit/ml, 5 mM CaCl 2 , 20 mM sodium acetate, pH 7.0) at 42°C for 30 min twice. Western blotting was performed as described above after lyophilization of the samples.
For Western blotting of syndecan-2 and -4 expressed in 293T cells, the cell lysates were prepared as described above except for 20 passes through a 21-gauge needle before lysis on ice. The supernatants (20 mg/ml) were boiled for 10 min in 6 M urea/40 mM sodium acetate, pH 4.5, and 100 l was applied to DEAE-cellulofine (Seikagaku Kogyo). After washing with 0.1% Triton X-100, PBS, the bound materials were eluted with 1.0 M NaCl, 0.1% Triton X-100, PBS and dialyzed against water at 4°C. Heparitinase I digestion was performed as above, and Western blotting was done with anti-syndecan-2 and -4 antibodies (1:500). For Western blotting of glypican-1 and syndecan-2 overexpressed in 293T cells, heparitinase I treatment of intact cells was also performed before lysis as described above. Glypican-1 was probed with anti-neo-heparan sulfate monoclonal antibody 3G10 (0.2 g/ml). For immunocytochemistry of syndecans and glypican-1, cells expressing syndecan-2, synde-  (Table I)  The bound materials were eluted with 10 mM glutathione and digested with heparitinase I. Syndecan-2 was detected by Western blotting. Note the molecular size of syndecan-2 after affinity chromatography was shifted to ϳ33 kDa. can-4, or glypican-1 or parental 293T cells were placed on a cover glass for 24 h and fixed by 4% paraformaldehyde, PBS for 15 min at room temperature, blocked with 5% normal donkey serum (Chemicon International Inc., Temecula, CA), 1% BSA, PBS for 15 min, and incubated with FITC-conjugated anti-heparan-sulfate monoclonal antibody 10E4 (20 g/ml) for 2 h at 37°C. FITC-conjugated mouse IgM, k monoclonal immunoglobulin standard (anti-TNP) (PharMingen), was used as a negative control and showed no staining (data not shown). The pictures were taken by Nikon fluorescent microscopy.
Affinity Chromatography and Solid-phase Assays-Cell lysates (20 mg/ml) were obtained from the 293T syndecan-2-expressing clone syn-2-#3. 500 l was used for GST-LG4N affinity chromatography, and the bound materials were eluted and digested with heparitinase I. Syndecan-2 was probed with anti-syndecan-2 antibody as described above. For solid-phase assay, 5 l of cell lysates/well was used as a source of syndecan-2. The bound syndeca-2 was detected as described above.

Heparin Binding Activity of Recombinant Laminin ␣3LG
Modules-The expression vectors for the C-terminal G domain modules, LG1-5, LG1-2, LG3, LG4, and LG5, were transfected into 293T cells and recombinant proteins secreted in the CM were purified by protein A-Sepharose affinity chromatography. Purified LG1-2, LG4, and LG5 showed a single band in SDS-PAGE with the predicted molecular mass (Fig. 1A). LG1-5 and LG3 showed several bands with the largest corresponding for the predicted molecular size, while the smaller bands were partially degraded products (Fig. 1A). LG1-5, LG3, and LG4 bound heparin in heparin-Sepharose affinity chromatography, whereas LG1-2 and LG5 did not (Fig. 1B). Heparin binding activity of LG3 and LG4 was observed even in the presence of 0.40 M NaCl during affinity chromatography, indicating a strong interaction with heparin. We next examined specificity of heparin binding of LG4 by inhibition assays using various glycosaminoglycans (GAGs) as competitors (Fig. 1C). Increasing amounts of heparin inhibited LG4 binding to a heparin-Sepharose affinity column, but chondroitin sulfate A, dermatan sulfate, and hyaluronic acid failed to inhibit. Binding of LG4 to heparin was also examined by solid-phase assays (Fig. 1D). The LG4 binding was dose-dependent and half-maximal binding was achieved at 0.5 g/ml (ϳ8 nM). To delineate the active sequence within LG4 for heparin binding (Fig. 1E), overlapped synthetic peptides covering the entire sequence of LG4 were prepared (Table I). Competition assays with synthetic peptides demonstrated that peptide A3G75 efficiently blocked LG4 binding to heparin. These data indicated that the A3G75 sequence in the LG4 module was essential for heparin binding.
Cell Binding Activity of Laminin ␣3LG1-2, LG4, and LG5 Modules-Recombinant LG1-2, LG4, and LG5 were examined for cell binding activity (Fig. 2, A and B). We did not analyze cell binding for LG3 and LG1-5 because we could not obtain enough of these recombinant proteins for the assay. Both HaCat cells and dermal fibroblasts attached to LG4 in a dosedependent fashion but did not attach to LG1-2, LG5, and control ␥2-s substrates. Cell adhesion activity of LG4 reached The bound materials were eluted by 10 mM glutathione, followed by dialysis against distilled water and heparitinase I digestion. Samples were analyzed by 15% SDS-PAGE and Western blotting using anti-syndecan-2 antibody. Syndecan-2 bound the GST-LG4-N. Note the molecular size of heparitinase I-treated syndecan-2 after partial purification or affinity chromatography was shifted from ϳ46 to ϳ33 kDa as seen in the endogenous syndecan-2 of fibroblasts. D, GAGs competition for syndecan-2 binding to A3G75aR in solid-phase assays. Syndecan-2-containing cell lysates were added to the A3G75aR-coated wells (80 g/ml) in the presence of 500 g/ml of GAGs included. Hp, heparin; Hs, heparan sulfate; CsA, chondroitin sulfate A; DS, dermatan sulfate; CsC, chondroitin sulfate C. The bound syndecan-2 was detected by anti-syndecan-2 antibody, alkaline phosphatase-anti-goat IgG antibody, followed by colorization with p-nitrophenyl phosphate. Each value is depicted as percent of OD in the absence of GAGs and represents the mean Ϯ S.D. of three different determinations.
the maximum level at ϳ20 g/ml (ϳ320 nM) in both cell types. Collagen I as a positive control substrate showed cell adhesion activity. Laminin-5 from the conditioned media of cultured keratinocyte, which is processed and missing LG4 -5, showed strong cell adhesion activity for HaCat cells and dermal fibroblasts. A3G75 peptide as described below did not block cell adhesion to the processed laminin-5 (data not shown). These results indicated that the processed laminin-5 had a cell adhesion site(s) different from that of LG4 (15)(16)(17)(18)(19).
Identification of LG4 Sequences Active for Cell Binding-We next performed inhibition analysis in LG4 cell binding assays using synthetic peptides as competitors (Fig. 3, A and B). Peptides A3G75 and A3G76 strongly inhibited HaCat cell adhesion to LG4 (Fig. 3A). Peptide A3G83 had weak inhibitory activity for cell adhesion, whereas other peptides showed no inhibition. Similar inhibition patterns were observed with dermal fibroblasts (Fig. 3B). A dose-dependent inhibition study with these three active peptides demonstrated that A3G75 and A3G76 inhibited cell binding completely at 50 g/ml, whereas A3G83 required Ͼ500 g/ml to achieve complete inhibition (data not shown). These results indicate that HaCat and fibroblasts bound to the A3G75 and A3G76 sequences, and also weakly to the A3G83 sequence. Since a deletion of first Lys residue of A3G75 did not affect cell adhesion activity (data not shown) and the sequences of A3G75 and A3G76 were overlapped, we prepared another synthetic peptide A3G75aR, which is missing the first Lys of A3G75 but including the Arg of A3G76 (Table I). We prepared synthetic peptides by substituting these basic amino acids, Lys 1421 and Arg 1423 in A3G75aR with Ser and tested their inhibitory activity for cell adhesion (Fig. 3C). Fibroblast cell adhesion to LG4 was inhibited by A3G75aR and A3G75. Cell adhesion was partially inhibited by A3G75aR-SGR and A3G75aR-KGS but not at all by A3G75aR-SGS, in which both Lys and Arg were substituted. HaCat cells showed similar inhibitory profiles with these peptides (data not shown). Thus, we concluded that cell adhesive activity of LG4 was mainly located in the sequence A3G75aR, NSFMALYL-SKGR, and that the two basic residues Lys and Arg were critical for its activity.
We next analyzed heparin binding of the mutant recombinant LG4 proteins to study the correlation of cell adhesion and heparin binding of the LG4 module (Fig. 4). Compared with the wild-type LG4, LG4-SGR, and LG4-KGS showed weaker heparin binding, and LG4-SGS, with double substitutions, had little binding activity in the heparin affinity chromatography assay (Fig. 4B). Competition in the LG4 heparin binding solidphase assays showed strong inhibition by peptide A3G75aR and moderate inhibition by peptide A3G75aR-SGR and weak inhibition by peptide A3G75aR-KGS. There was no inhibition by peptide A3G75aR-SGS (Fig. 4C). These results suggest a close relation in heparin and cell binding of LG4. Although both basic residues are required for heparin binding activity of LG4, the lys 1421 appears to be more critical for the activity compared with the Arg 1423 .
GAG-dependent LG4 Cell Binding-Because the critical residues, Lys and Arg, in the A3G75aR sequence were common for cell adhesion and heparin binding, we looked for cell surface proteoglycan receptors for LG4 cell binding, such as syndecans or ␣-dystroglycan (␣-DG). Cell binding of HaCat cells and fibroblasts to LG4 was inhibited by heparin and weakly inhibited by heparan sulfate and dermatan sulfate (Fig. 5A). However, no significant inhibition was observed by chondroitin sulfate A, chondroitin sulfate C, and hyaluronic acid. Heparin inhibited ␣-DG (40)-and syndecan-1 (26)-dependent cell adhesion. However, the critical determinant of ␣-DG for laminin binding is not heparan sulfate, but o-mannosyl-type sialylated oligosaccharide (41). Since heparitinase I treatment had no effect to ␣-DG binding to laminin (42), we examined cell adhesion after heparitinase I digestion of cell surface heparan sulfate (Fig. 5B). Heparitinase I treatment of cells completely abolished cell adhesion to LG4 and to A3G75aR but not to the processed laminin-5 or collagen I. These results indicated that cell binding to LG4 was mediated by cell surface heparan sulfate proteoglycans.
Syndecan-2 of Fibroblasts Bound to LG4 in Affinity Chromatography-Since syndecans were shown to bind laminin (26), we first examined expression of syndecans in HaCat cells and dermal fibroblasts. RT-PCR analysis revealed that syndecan-4 mRNA was expressed in both HaCat cells and fibroblasts and syndecan-2 mRNA were detected only in fibroblasts. No syndecan-1 mRNA was observed in both cells (data not shown). Western blotting showed that syndecan-2 was present in fibroblasts (Fig. 6A). However, syndecan-4 could not be detected with specific antibody using the same amounts of samples from fibroblasts (data not shown). Using the bacterially expressed fusion proteins, we found that the N-terminal 101 amino acids of the LG4 containing the A3G75aR sequence (GST-LG4-N) was active for heparin binding in solid-phase assays, but the C-terminal LG4 was inactive (data not shown). Hence, we prepared a GST-LG4-N affinity column to identify a cell surface receptor for LG4, because bacterial recombinant proteins were easily obtainable. Fibroblast lysates were applied on GST-LG4-N affinity chromatography, and bound fractions were eluted, treated with heparitinase I, and analyzed by Western blotting with anti-syndecan-2 antibody (Fig. 6B). Syndecan-2 was eluted from the GST-LG4-N-Sepharose but not form the GST-Sepharose, indicating that syndecan-2 bound LG4.
Recombinant Syndecan-2 Expressed by 293T Cells Bound to LG4 -To confirm that syndecan-2 is a receptor for LG4 cell adhesion, we expressed recombinant syndecan-2 by stable transfection in 293T cells, which do not express endogenous syndecan-2 and do not bind LG4. One of the selected clones (syn-2-#3) showed cell surface localizations of heparan sulfate, which is most likely syndecan-2 (Fig. 7A). The size of exogenously expressed syndecan-2 was similar to endogenous syndecan-2 from fibroblasts (Fig. 7B, lanes 1 and 2). Endogenous syndecan-2 was not detected in control 293T cells by these two kinds of detection methods. Recombinant syndecan-2 expressed in 293T cells was able to bind GST-LG4-N in affinity chromatography similar to endogenous syndecan-2 of fibroblasts (Fig. 7C). The molecular size of recombinant syndecan-2 changed from ϳ46 to ϳ32 kDa after DEAE-cellulofine and GST-LG4-N affinity chromatography. A similar size change was also observed with endogenous syndecan-2 after GST-LG4-N affinity chromatography (Fig. 6B), suggesting that it may be due to the change of susceptibility to heparitinase I during chromatography. Solid-phase assays for binding of recombinant syndecan-2 to A3G75aR were performed in the presence of GAGs as a competitor (Fig. 7D). The binding was strongly inhibited by heparin and less effectively inhibited by heparan sulfate and dermatan sulfate. These results indicated that syndecan-2 is able to bind to A3G75aR through a heparan sulfate-, dermatan sulfate-, and heparin-dependent manner.
Syndecan-2-expressing 293T Cells Adhere to LG4 and A3G75aR-Syn-2-#3 cells expressing recombinant syndecan-2 attached to LG4, and to peptide A3G75aR, but parental 293T cells failed to bind to these substrates (Fig. 8A). The mutant peptide A3G75aR-SGS and collagen I were poor substrates for syn-2-#3 cell binding. Both syn-2-#3 and 293T cells bound the processed laminin-5. This is because cell binding of the processed laminin-5 is mediated through integrins, which recognize a site different from that of LG4. Inhibition analysis was performed in syn-2-#3 cell binding to LG4 and A3G75aR peptide-coated dishes in the presence of various GAGs or by A3G75aR (Fig. 8B). Syndecan-2-expressing 293T cell binding was strongly inhibited by heparin and A3G75aR, and moderately inhibited by heparan sulfate and dermatan sulfate, but not by other GAGs. Taken together, we concluded that syndecan-2 is a receptor that mediates cell binding of the LG4 module of the laminin ␣3 chain.
Syndecan-4-, but Not Glypican-1-, expressing 293T Cells Bind to LG4 -Since HaCat cells do not express syndecan-1 and -2, it is possible that syndecan-4 may be responsible for HaCat FIG. 9. Syndecan-4-, but not glypican-1-, expressing 293T cells adhere to LG4 and A3G75aR. A, syndecan-4-expressing cells (syn-4-#7) and glypican-1-expressing cells (GP-1-#5) on a cover glass were stained with FITC-anti-heparan sulfate antibody and then photographed (original ϫ200). B, the heparan sulfate proteoglycans partially purified by DEAE-cellulofine from the same amounts of cell lysates from syn-4-#7 (lane 1) and parental 293T (lane 2) were digested with heparitinase I and applied on 15% SDS-PAGE, followed by Western blotting with anti-syndecan-4 antibody. Parental 293T cells (lanes 3 and 4) and GP-1-#5 (lanes 5 and 6) were treated with (lanes 3 and 5) or without (lanes 4 and 6) heparitinase I, lysed, and the same amounts of supernatants were separated by 12% SDS-PAGE. Heparitinase I-digested heparan sulfate was detected with anti-⌬-heparan sulfate antibody 3G10. C, syn-4-#7 or GP-1-#5 cells were seeded on LG4 (16 g/ml), A3G75aR (10 g/ml), A3G75aR-SGS (10 g/ml), processed laminin-5 (4.0 g/ml), or collagen I (2.0 g/ml). Each value represents the mean Ϯ S.D. of three different determinations. D, inhibition of attachment of syn-4-#7 cells by 250 g/ml of GAGs or 100 g/ml of synthetic peptides. Each value is depicted as percentage of cell adhesion without competitor and represents the mean Ϯ S.D. of three different determinations. cell adhesion to LG4. To examine this possibility, syndecan-4expressing 293T cell clone (syn-4-#7) was obtained. We also obtained a glypican-1-expressing 293T cell clone (GP-1-#5) to test specificity of the LG4 interaction to syndecans. The surface localization of syndecan-4 or glypican-1 was confirmed by immunostaining with anti-heparan sulfate antibody 10E4 (Fig.  9A). Western blotting also revealed expression of syndecan-4 or glypican-1 after heparitinase I digestion (Fig. 9B). We found that syn4-#7 cells, but not GP-1-#5 cells, bound to LG4 and A3G75aR in a GAG-dependent manner similar to syndecan-2 ( Fig. 9, C and D). These results indicate that syndecan-4 is able to function as a receptor for cell adhesion to the laminin ␣3 chain LG4 module. Furthermore, cell adhesion to LG4 is not mediated by glypican-1, suggesting that heparan sulfate chains of syndecans are not sufficient for the interaction of LG4 with syndecans and that the syndecan protein core sequence is also involved in this interaction. DISCUSSION Using recombinant proteins, we showed that the LG3 and LG4 modules of the G domain of the laminin ␣3 chain bound to heparin. By mutation and synthetic peptide analyses, we subsequently identified the A3G75aR sequence NSFMALYLSKGR (1412-1423 amino acids) as a major heparin and cell binding site within the LG4 module. Heparin binding activity of the laminin ␣1 and ␣2 chains has been implicated in binding to extracellular matrix proteins, such as nidogen-2, fibulin-1, and fibulin-2 (23,43), and also in binding to a proteoglycan cell surface receptor component, ␣-dystroglycan (for review, see Ref. 44). Syndecan-1 was also identified as a laminin-1 receptor by affinity chromatography (26).
Various LG modules of the laminin ␣1-␣5 chains have been shown to bind to heparin. However, ligands that are involved in heparin binding of the ␣3, ␣4, and ␣5LG modules have not been identified. The ␣2LG5 module consists of a 14-stranded ␤-sheet sandwich structure, i.e. A-N sheet (45). The heparin binding sites are formed by basic residues on loops between each sheet of the sandwich structure and around the calcium ion binding site. When we align the A3G75aR site of the ␣3LG4 module using a structure-based sequence alignment, it corresponds to two ␤-strands (E and F), and a connecting loop in the crystal structure of the ␣2LG5 module (Fig. 10). The two critical Lys and Arg residues of A3G75aR are located on the connecting loop of the E and F strands and on part of the F strand and are both solvent-exposed at the edge of the 14-stranded ␤-sheet sandwich structure. However, these residues are not on the same side as the basic residues and the calcium binding site, which were previously shown to be essential for ␣-dystroglycan binding to the ␣2LG5 module (23,43). The calcium ion did not show any effects on heparin binding activity of ␣4LG4, in contrast to ␣2LG5 (28), and ␣4LG4 had a much weaker ␣-DG binding activity compared with ␣2LG4 -5 (32). Interestingly, the two positively charged residues, Lys 1421 and Arg 1423 , of A3G75aR are located on the same positions as His 1518 and Arg 1520 of the ␣4 LG4 heparin binding site (28).
In this study, we demonstrated that syndecans interact with the A3G75aR site of ␣3LG4. We found that heparitinase I digestion of cells inhibited cell binding to ␣3LG4, suggesting that the receptor for cell adhesion to ␣3LG4 is mediated by cell surface heparan sulfate chains. Our in vitro study using the GST-LG4-N affinity column and solid-phase assays with synthetic peptide showed that syndecan-2 bound to LG4. This was further confirmed by cell binding assays using recombinant, syndecan-expressing 293T cells. These results showed the specific involvement of syndecan-2 and -4 in cell adhesion to ␣3LG4. Syndecans have been shown to participate in extracellular matrix adhesion, cell motility, or focal adhesion assembly in several cell types (46, 47; for review, see Refs. 48 and 49). Our results, that both syndecan-2 and -4 mediate cell adhesion to ␣3LG4, suggest that cell adhesion to ␣3LG4 depends on which syndecan is expressed by cells. Recently, Iba et al. (50) demonstrated that mesenchymal cells attached to the ADAMs protein through any of syndecan-1, -2, or -4, but not glypican-1. We also found that glypican-1-expressing cells failed to attach to LG4. This suggests that heparan sulfate chains are not sufficient for the interactions and that the protein core sequence of syndecans is also required for specific interactions with LG4. GAGs are composed of numerous repeats of uronic acid and hexosamine. Sulfate is found as the N-sulfated form or o-sulfated ester. Heparan sulfate and dermatan sulfate were similarly active in blocking cell adhesion and also in syndecan binding to LG4 and A3G75aR. The structural feature common to heparin, heparan sulfate, and dermatan sulfate is that they contain the L-iduronic acid-2-sulfate. Hence, this sulfated saccharide may be important for syndecan binding to ␣3LG4.
Laminin-5 is up-regulated by keratinocytes at the wound edge. The deposition of unprocessed laminin-5 is observed only in the wound area, and unprocessed laminin-5 is not detected in the mature intact skin (18,33,51). When keratinocyte migration begins, unprocessed laminin-5 is found in the provisional basement membrane of the wound by leading keratinocytes (31,51,52). The cleavage of the C terminus of the laminin ␣3 chain by plasmin reduces cell migration and enhances the formation of hemidesmosomes (52). These observations suggest that the proteolytic cleavage process of the C terminus of the laminin ␣3 chain coincides with cell migration. It is conceivable that the LG4 -5 modules of the unprocessed laminin-5 may play an important role for cell migration during wound healing. Evidence that syndecan-4 is temporarily expressed mainly in basal keratinocytes at the wound of the skin (30) may implicate the significance of syndecan-laminin ␣3LG4 interaction in wound healing.  (1421) and Arg (1423) of A3G75aR (green color) that are essential residues for cell adhesion to ␣3LG4 are mapped on the structure of the ␣2LG5 module (protein data bank code 1qu0) (45). These residues are located on the connecting loop of the E and F strands and on part of the F strand and are both solvent-exposed at the edge of the 14-stranded ␤-sheet sandwich structure. The conserved position of the calcium ion of the ␣2LG5 module, which is not found in ␣3LG4, is depicted with gray color for orientation. The A3G75aR peptide sequence is aligned to Met 2984 -Met 2994 of the ␣2 sequence.