Laminin α1 Chain LG4 Module Promotes Cell Attachment through Syndecans and Cell Spreading through Integrin α2β1*

The laminin α1 chain is a subunit of laminin-1, a heterotrimeric basement membrane protein. The LG4-5 module at the C terminus of laminin α1 contains major binding sites for heparin, sulfatide, and α-dystroglycan and plays a critical role in early embryonic development. We previously identified active synthetic peptides AG73 and EF-1 from the sequence of lamininα1 LG4 for binding to syndecan and integrin α2β1, respectively. However, their activity and functional relationship within the laminin-1 and LG4 as well as the functional relation between these sites and α-dystroglycan binding sites in LG4 are not clear. To address these questions, we created mutant recombinant LG4 proteins containing alanine substitutions within the AG73 (M1), EF-1 (M2, M3), and α-dystroglycan binding sites (M4, M5) and analyzed their activities. We found that recombinant proteins rec-M1 and rec-M5, containing mutations within M1 and M5, respectively, did not bind heparin or lymphoid cell lines expressing syndecans. These results suggest that LG4 binds to heparin and syndecans through M1 and M5. Rec-M1 and rec-M5 reduced fibroblast attachment, whereas mutant rec-M2 and rec-M3 retained cell attachment activity but did not promote cell spreading. Fibroblast attachment to rec-LG4 was inhibited by heparin but not by integrin antibodies. Spreading of fibroblasts on rec-LG4 was inhibited by anti-integrin α2 and β1 but not by anti-integrin α1 and α6. These results suggest that the M1 and M5 sites are necessary for cell attachment on LG4 through syndecans and that the EF-1 site is for cell spreading activity through integrin α2β1. In contrast, laminin-1-mediated fibroblast attachment and spreading were not inhibited by heparin or antiintegrin α2. Our findings indicate that LG4 has a unique function distinct from laminin-1 and suggest that laminin α1 LG4-5 may also be produced by a proteolytic cleavage in certain tissues where it exerts its activity.

Laminins are heterotrimeric basement membrane proteins that exert multiple biological functions through interactions with extracellular matrix molecules and with cell surface receptors. The regulation of these interactions are critical to many biological processes, including cell adhesion and migration, angiogenesis, tumor progression, and neurite outgrowth (1,2). The laminin family contains at least 11 chains (five ␣, three ␤, and three ␥ chains) (2). The five laminin ␣ chains share a large module at their C-terminal region (G domain), which contains five laminin G domain-like modules (LG1-5), and each module consists of about 200 amino acids (3,4).
Laminin ␣1 knock-out mice die around embryonic day E7.5 due to the lack of Reichert's membrane and defective epiblast polarization (23). Recently, mutant mice that express a truncated laminin-1 molecule that is specifically missing the LG4-5 portion of the laminin ␣1 chain were generated (24). The mutant mice die from defective Reichert's membrane, just like the laminin ␣1 knock-out mice. In addition, the ectoplacental cone of wild-type mice stain with antibodies specific to laminin ␣1 LG4 but not to the N terminus or to LG1-3, suggesting that laminin ␣1 may be cleaved to generate the LG4-5 molecule in vivo, as observed in laminin ␣2 (25,26), ␣3 (27,28), ␣4 (29), and ␣5 (30). Taken together, these results suggest that the laminin ␣1 LG4-5 tandem module is critical for mouse embryogenesis and may also serve as a distinct functional unit apart from the whole laminin-1 molecule in certain tissues.
In this report we prepared recombinant mouse laminin ␣1 LG4 proteins, which contained site-specific mutations within the AG73, EF-1, K1, and K3 sites and analyzed their activities for binding to heparin, syndecans, and integrins in relation to their activities for cell attachment and spreading. We found that the critical residues of the AG73 and ␣-dystroglycan binding sites are very closely located in the predicted three-dimensional structure and also bind to syndecan. On the other hand, the EF-1 site is not essential for cell attachment but is required for cell spreading through integrin ␣2␤1 binding. This site is positioned on the opposite side from the syndecan binding sites in the LG4 module. Inhibition assays revealed that the laminin ␣1 LG4 module has a unique function distinct from the whole laminin-1 molecule. Our results suggest that the AG73 and ␣-dystroglycan sites within LG4 have an overlapping function, and the EF-1 site exerts its activity for cell spreading in concert with the AG73 and ␣-dystroglycan sites.
The expression vector for the LG4 module, pCEP4-Mul-PURD, is a modified vector of pCEP4-WT-His (33), a derivative of pCEP4 (Invitrogen), and contains a BM-40 signal peptide, a hexahistidine tag, and a multicloning site along with the cytomegalovirus promoter and enhancer. In this vector the original hygromycin resistance gene was replaced by the puromycin resistance sequence from the pPUR vector (BD Bioscience). HindIII site in puromycin was deleted by the fill-in method using the DNA Blunting Kit (Takara, Ohtsu, Japan), and the hygromycin resistance gene was replaced in pCEP4 by PURD at SpeI and NruI. Various mutated laminin ␣1 LG4 module sequences were cloned into the HindIII and NotI sites of pCEP4-Mul-PURD.
Site-directed mutagenesis of LG4-5 was also performed similar to LG4. Three amino acids, Arg-2719, Lys-2720, and Arg-2721, within AG73 were substituted with alanine individually or all together. The mutant LG4-5 sequences were cloned into the HindIII and XhoI sites of pSecTag2/Hygro B (Invitrogen). The recombinant proteins were prepared as previously reported (34).
Expression and Purification of Recombinant Proteins-The recombinant LG4 expression vectors were transfected into 293 EBNA cells using FuGENE 6 (Roche Applied Science). The transfected cells were maintained for 2 days, then split at a 1:6 ratio in 10% FBS in Dulbecco's modified Eagle's medium and incubated with 5 g/ml puromycin (Sigma) for 3 days for selection. The selected cells were maintained with 0.5 g/ml puromycin. To prepare recombinant proteins, nearly confluent cells were cultured without serum for 3 days, and the condition medium was collected. The medium was cleared of debris by centrifugation, and Complete EDTA-free (Roche Applied Science) was added to decrease proteolysis. Nickel-charged agarose resins (Probond, Invitrogen) were equilibrated with Dulbecco's PBS (D-PBS; Invitrogen), pH 8.0, containing 10 mM imidazole and added to the conditioned medium. After incubation at 4°C for 60 min, the resins were transferred to a column and washed with D-PBS, pH 8.0, containing 20 mM imidazole. His-tagged laminin ␣1 LG4 module recombinant proteins were eluted with D-PBS, pH 8.0, containing 250 mM imidazole. Purified proteins were dialyzed against D-PBS and quantified using the BCA protein assay kit with bovine serum albumin as a standard (Pierce). Purity was determined by reducing SDS-PAGE followed by Coomassie G-250 Blue (GelCode Blue, Pierce) staining and judged to Ͼ95%.
Synthetic Peptides-The peptides AG73 and EF-1 were manually synthesized with an amide at the C terminus by the Fmoc (9-fluorenylmethoxycarbonyl) strategy as previously described (17,35). Purity and identity of the peptides were confirmed by an analytical high performance liquid chromatography and an electrospray ionization mass spectrometer at the Central Analysis Center, Tokyo University of Pharmacy and Life Science.
Cell Attachment Assay-96-Well round-bottomed microtiter plates (Immulon-2HB) were coated with recombinant proteins or laminin-1 from Engelbreth-Holm-Swarm (EHS) tumor (Sigma) in 50 l of D-PBS overnight at 4°C and then blocked for 60 min at RT with 200 l of heat-denatured (30 min at 70°C) Dulbecco's modified Eagle's medium containing 1% bovine serum albumin (Sigma, 1% blocking solution) and washed twice with 0.1% blocking solution. Protein coating efficiency of recombinant proteins was measured by the enzyme-linked immunosorbent assay using His-tag antibody. These results showed that all recombinant proteins bound dishes at almost equal coating efficiency (less than a 2% difference). ARH-77 cell lines were cultured in suspension, washed twice with 0.1% blocking solution, and resuspended in a concentration of 2 ϫ 10 5 cells/ml. Human foreskin fibroblasts were detached with 0.02% trypsin, EDTA (Invitrogen). Cell suspension was diluted with an equal volume of 10% FBS in Dulbecco's modified Eagle's medium to inhibit trypsin and resuspended and incubated in 10% FBS-containing media at 37°C for 20 min to recover the cells. Then cells were washed twice and resuspended in 0.1% blocking solution and used for attachment assays. 2 ϫ 10 4 cells/100 l cells were added to the wells and incubated for 60 min at 37°C in a humidified atmosphere of 5% CO 2 . Wells were stained for 10 min with 0.2% crystal violet (Sigma) in 20% methanol. After washing with H 2 O, plates were dried overnight at RT. Cells were dissolved in 1% SDS (150 l), and the absorbance at 570 nm was measured. Each sample was assayed in triplicate, and cells attached to the bovine serum albumin substrate were subtracted from all measurements. For inhibition of cell attachment with heparin, EDTA, peptides, and anti-integrin antibodies, fibroblasts were first incubated for 10 min at 37°C in the presence of either 10 g/ml heparin, 5 mM EDTA, 100 g/ml peptides, or 10 g/ml antibodies, respectively, and then incubated and measured as described above.
Cell Spreading Assay-Glass Chamber Slides (Nalge Nunc International, Naperville, IL) were coated with proteins (3 g/well) and prepared as described for the cell attachment assay. Human foreskin fibroblasts (2 ϫ 10 3 cells/well) were added to the wells and incubated for 150 min at 37°C in a humidified atmosphere of 5% CO 2 . After washing with prewarmed D-PBS containing calcium and magnesium (Invitrogen), cells were fix with 4% paraformaldehyde D-PBS for 10 min at 37°C. Fixed cells were mounted with GEL/MOUNT (Biomeda, Foster City, CA) and examined under an Axiovert 135 microscope (Carl Zeiss, Thornwood, NY) equipped with an AxioCam HRc CCD camera (Carl Zeiss). Digital images were obtained using MetaMorph 6.16 software.
Immunofluorescence Assay of Cultured Cells-Fibroblasts were prepared as described for the spreading assay. Cells were fixed and permeabilized with prewarmed D-PBS containing calcium and magnesium (Invitrogen), 4% paraformaldehyde, 0.5% Triton X-100, and 5% sucrose for 3 min at 37°C. Then, for further fixation, slides were treated with 2% paraformaldehyde containing 5% sucrose D-PBS with calcium and magnesium for 10 min at 37°C and blocked in D-PBS, 20% donkey serum (Jackson ImmunoResearch Laboratories, Inc.), and Mouse on Mouse (MOM) blocking reagent (Vector Laboratories, Inc., Burlingame, CA) as recommended by the manufacturer. Cells were incubated with anti-vinculin or integrin antibody in D-PBS, 0.05% Tween 20, and 7% MOM protein concentrate (Vector Laboratories, Inc.) to 10 g/ml for 60 min at RT or 4°C overnight. Bound antibodies and F-actin were visualized by secondary antibody diluted to 1:100 and Alexa Fluor 488 phalloidin (Invitrogen) diluted to 1:100 in D-PBS, 5% donkey serum, and 0.05% Tween 20 for 30 min at RT. After washing, cells were mounted with VECTASHIELD (Vector Laboratories) and examined under a LSM510 fluorescent confocal microscope (Carl Zeiss).

Mutant Recombinant Laminin ␣1
LG4 and LG4-5 Proteins-Laminin ␣1 LG4-5 contains two tandem repeat modules, LG4 and LG5. We first prepared recombinant LG4-5, LG4, and LG5 modules of laminin ␣1 and tested their activity for cell attach-Cell Attachment and Spreading on the Laminin ␣1 LG4 Module OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 LG4-5 and LG4 promoted cell attachment, but LG5 did not (see Fig. 3B and data not shown). These results suggest that LG4 plays a critical role for cell binding activity of laminin ␣1 LG4-5. We, therefore, primarily focused on the LG4 module in this study.
We prepared five mutant LG4 proteins (rec-M1 to M5) with alanine substitutions; rec-M1 has a mutation within the AG73 sequence, rec-M2/-M3 has a mutation within the EF-1 sequence, and rec-M4/-M5 has a mutation within ␣-dystroglycan binding regions (Fig. 1A, Table 1). Mutant rec-M1 protein has alanine substitutions at three consecutive positively charged residues, Arg-2719, Lys-2720, and Arg-2721 (M1 site), within AG73 that were shown to be critical for peptide AG73 cell attachment and heparin binding activity (14 -16). Mutant rec-M2 and rec-M3 proteins contained a single alanine substitution at Arg-2757 (M2 site) and Asp-2763 (M3 site) within the EF-1 sequence, respectively, which were important for peptide EF-1 integrin ␣2␤1 binding activity (17). In addition, two mutant proteins, rec-M4 and rec-M5, were created within ␣-dystroglycan binding sites by alanine substitutions at the three positively charged residues in each site, Lys-2766, Arg-2768, and Lys-2770 (M4 site) and Lys-2791, Arg-2792, and Lys-2793 (M5 site), respectively (22). We also created four mutations within the M1 site of AG73 in recombinant LG4-5; alanine was substituted at residues Arg-2719, Lys-2720, and Arg-2721 either singly or all together. These mutant proteins were purified from the conditioned media of 293 EBNA cells or Cos7 cells that had been transfected with the expression vectors as described under "Experimental Procedures." The purified proteins showed a single band with 95% purity in SDS-PAGE analysis under a reduced condition and also under a nonreduced condition (Fig. 1B).

Heparin Binding of rec-LG4 and Mutant
LG4 Proteins-First we tested heparin binding of the recombinant wild-type LG4 protein (rec-LG4) and the mutant proteins (rec-M1 to rec-M5) using biotinylated heparin in a solid phase binding assay (Fig. 2). Rec-LG4 strongly bound heparin, whereas heparin binding activity of rec-M1 and rec-M5 was significantly reduced. Rec-M2 and rec-M3 bound to heparin, whereas rec-M4 moderately decreased heparin binding. We obtained similar results LG4-5 tandem modules was constructed based on the crystal structure of the laminin ␣2 LG4-5 tandem modules (41; PDB accession number 1DYK). The molecule is shown from two angles, with a 90°rotation on the vertical axis, and the dotted lines indicate the linker region between LG3 and LG4 (4). The yellow spot shows the calcium binding site of the laminin ␣1 LG4 module. The molecular surface was generated with DeepView/Swiss PDB viewer, and final total energy was calculated to Ϫ14919 KJ/mol (53)(54)(55). See Table 1 for a summary of the sites and mutant proteins.
using a heparin beads affinity assay (data not shown). These results indicate that the M1 sequence of AG73 and M5 sequence of the ␣-dystroglycan site is critical for heparin binding of LG4 and suggest that the EF-1 site is not involved in heparin binding of LG4. All of the basic charged residues (Arg-2719, Lys-2720, and Arg-2721) of the M1 site are critical for heparin binding since mutant LG4-5 with single substitutions at each of these residues lost heparin binding activity (data not shown).
To identify syndecan binding sites of LG4, we examined cell attachment activity of mutant LG4. We found that mutant rec-M1, rec-M4, and rec-M5 proteins do not attach to any of the syndecan-expressing cell lines, ARH-Synd-1, ARH-Synd-2, and ARH-Synd-4 (Fig. 3A). In contrast, rec-M2 and rec-M3 attached to these syndecan-expressing cells. None of the mutant LG4s attach to ARH-Gpc-1 cells similar to rec-LG4. These results indicate that LG4 attaches to cells expressing syndecan-1, -2, or -4 but not to cells that express glypican-1. The M1, M4, and M5 sites are essential for these interactions, and a mutation at any of these sites destroys the activity of LG4 attachment to the syndecan-expressing lymphoid cells. Our data also suggest that the EF-1 site is not necessary for LG4 attachment to syndecan-expressing cells.
To confirm the significance of the M1 site within LG4-5 for syndecan binding, we analyzed cell attachment of mutant LG4-5 proteins using ARH-Synd-1 cells (Fig. 3B). We found that mutant LG4-5 containing a single substitution at each of three RKR residues of the M1 site failed to attach to the cells. These results indicate that each of these basic residues is critical for syndecan binding of LG4-5 and suggest that the LG5 module is not involved in this activity.
Fibroblast Attachment and Spreading on rec-LG4, Mutant LG4 Proteins, and Laminin-1-To identify the active sites of LG4 that are important for cell attachment and spreading and their functional relationship, we analyzed the activities of mutant LG4 proteins for cell attachment and spreading using human foreskin fibroblasts, which express both syndecans (18) and integrin ␣2␤1 (36). Fibroblasts attached to laminin-1, rec-LG4, rec-M2, and rec-M3 in a dose-dependent manner (Fig.  4A). However, fibroblasts did not attach to rec-M1, rec-M4, or rec-M5 substrates. These results indicate that heparin binding sites M1, M4, and M5 are all required for fibroblast attachment of rec-LG4, whereas the M2 and M3 sites are not necessary for fibroblast attachment.
We next analyzed heparin, EDTA, and peptide AG73 and EF-1 inhibition of fibroblast attachment on rec-LG4 and mutant LG4 proteins. Fibroblast attachment to rec-LG4 was significantly inhibited by heparin and the AG73 peptide but not by EDTA or the EF-1 peptide (Fig. 4B). In contrast, fibroblast attachment to laminin-1 was significantly inhibited by EDTA as previously reported (37) but not by heparin, AG73, or EF-1 (Fig.  4C). These results suggest that rec-LG4 attaches to fibroblasts through syndecans, and laminin-1-mediated fibroblast attach-  The sequence positions and sites are from Andac et al. (22). b These active peptides were derived from the laminin ␣1 LG4 module. The LG4 sequence is shown in Fig. 1A. ment is primarily through interactions with integrins such as ␣6␤1, whose binding sites are located within the C-terminal triple-helical region of laminin ␣1 (38).
We also assessed the cell spreading activity of recombinant LG4 proteins and laminin-1 (Fig. 5). Laminin-1 and rec-LG4 induced fibroblast cell spreading, although rec-LG4 was slightly less active than laminin-1 (Fig. 5, A and B). Both rec-M2 and rec-M3 promoted cell attachment as described above, but rec-M3 did not spread well, whereasrec-M2 showed weak cell spreading activity (Fig. 5, C and D). Rec-M1, rec-M4, and rec-M5 did not show cell attachment and spreading (Fig. 5, C, F,  and G). These results suggest that the M3 site is involved in the cell spreading activity of LG4.

Organization of Actin Stress Fibers and Vinculin Localization of Fibroblasts on rec-LG4, Mutant LG4 Proteins, and
Laminin-1-We next looked at the organization of actin stress fibers and vinculin localization to focal contacts of fibroblasts on laminin-1, rec-LG4, rec-M2, and rec-M3 by immunostaining (Fig. 6). Fibroblasts on laminin-1 formed well organized actin stress fibers and focal contacts containing vinculin (Fig. 6A). The cells on rec-LG4 also formed actin stress fibers and focal contacts but to a lesser degree than those on laminin-1 (Fig. 6B). Fibroblasts weakly spread on rec-M2, but there were a few organized stress fibers and focal contacts containing vinculin (Fig. 6C). Fibroblasts did not spread well on rec-M3 (Fig.  6D), and there was actin accumulation at the edge of cells but no accumulation of vinculin. These results indicate that the M2 and M3 sites are important for cell spreading and focal contacts.
Effect of Heparin, EDTA, and Peptides AG73 and EF-1 on Fibroblast Spreading on rec-LG4 and Laminin-1-We examined the effect of heparin and EDTA on fibroblast spreading on the rec-LG4 substrate and compared their effects on laminin-1. Heparin inhibited fibroblast attachment on rec-LG4 (Fig. 7C) but not on laminin-1 (Fig. 7D). EDTA inhibited cell spreading on rec-LG4 but not cell attachment (Fig. 7E). In contrast, fibroblast attachment on laminin-1 was completely inhibited by EDTA (Fig. 7F). This indicates that LG4mediated fibroblast attachment is dependent on heparin and syndecan interactions as shown in Fig. 4, but cell spreading on rec-LG4 is required for cation-dependent binding. Laminin-1-mediated cell attachment and spreading are required for cation-dependent binding.
Next we examined AG73 and EF-1 inhibition of cell spreading when fibroblasts were plated on rec-LG4 and laminin-1 (Fig. 7, G-J). Peptide AG73 reduced focal contacts and actin LG4 proteins and laminin-1. Attached cells were stained with 0.2% crystal violet, and the absorbance at 570 nm was measured. The effects of heparin, EDTA, and peptides AG73 or EF-1 on fibroblast attachment to rec-LG4 (B) and laminin-1 (C) were analyzed. In these assays 10 g/ml heparin, 5 mM EDTA, and 100 g/ml peptides were added to the cell suspension and incubated for 10 min at 37°C before cell attachment was assayed as in A. *, p Ͻ 0.001. stress fibers on rec-LG4 but not on laminin-1 (Fig. 7, G and H). In contrast, peptide EF-1 strongly inhibited cell spreading on rec-LG4 and weakly inhibited it on laminin-1 (Fig. 7, I and J).
These results indicate that the mechanisms of cell attachment and spreading are different between LG4 and laminin-1.
Fibroblast Spreading on rec-LG4 and Laminin-1 through Integrin-Cell-matrix interactions via integrins are critical for organizing actin stress fibers and focal contacts. We examined the effect of anti-integrin antibodies on fibroblast spreading on rec-LG4 and laminin-1 using double immunostaining for actin stress fibers and vinculin (Fig. 8A). Fibroblast spreading on rec-LG4 was inhibited by anti-integrin ␣2 and ␤1 antibodies but not by anti-integrin ␣1 or ␣6 antibodies. However, anti-integrin ␤1 and ␣6 antibodies inhibited cell attachment spreading on laminin-1, but neither anti-integrin ␣2 nor ␣1 antibodies inhibited cell spreading.
Immunostaining of fibroblasts with anti-integrin antibodies showed that integrin ␣2 was localized at focal contacts on the LG4 substrate but not the laminin-1 substrate, whereas integrin ␤1 was present at  focal contacts on both substrates (Fig. 8B). Taken together these results indicate that integrin ␣2␤1 plays a critical role in organizing actin stress fibers and focal contacts on LG4 and that integrin ␣6␤1 is important for fibroblast attachment and spreading on laminin-1 as previously reported (38,39).

DISCUSSION
The laminin ␣1 chain G domain interacts with extracellular matrix molecules and receptors and is implicated in cellular processes and assembly of the basement membrane (2). Several active sites of laminin ␣1 LG4 have been identified. It has been shown that the synthetic peptide AG73 promotes cell attachment and heparin and syndecan binding (12)(13)(14)(15)(16). More recently EF-1 was identified as an active peptide for integrin ␣2␤1 binding (17). However, it is not clear whether these sites are active in LG4 and laminin-1 proteins or the nature of their functional relationship. In addition, two active sites, M4 and M5, located at the C-terminal region of LG4 have been identified as essential for heparin and ␣-dystroglycan binding (22). However, their cell attachment activity, syndecan binding, and functional relations with peptides AG73 and EF-1 are unknown. Here, we report that the M1 site in the AG73 sequence and the M4 and M5 sites for ␣-dystroglycan binding are utilized for primary cell attachment through syndecan and that the EF-1 site is critical for cell spreading through integrin ␣2␤1. We also found that the mechanism of LG4-mediated fibroblast attachment is distinct from that of the whole laminin-1 in cell culture.
The mutation analyses revealed that the M1, M4, and M5 sites are all required for heparin binding of LG4 (though to a lesser extent with the M4 site) and that these sites are critical for the attachment of LG4 to lymphoid cells expressing syndecan-1, -2, and -4 but not to cells that express glypican-1. These results suggest that LG4 interacts with syndecans and not with glypican-1. These cell lines have been shown to produce heparan sulfate chains attached to the protein core of syndecans and glypican-1. The specific interaction of LG4 with syndecans may require not only heparan sulfate chains but also a core protein sequence of syndecans. It may also require different modifications of heparan sulfate chains between syndecans and glypican-1. The inhibition and mutation analyses also showed that the M2 and M3 sites in the EF-1 sequence are not essential for cell attachment but are required for fibroblast spreading on LG4. When LG4 was used as a substrate, anti-integrin ␣2 and ␤1 antibodies inhibited cell spreading but not cell attachment, whereas heparin inhibited cell attachment. These results suggest that the syndecan-binding sites M1, M4, and M5 are primarily utilized for cell attachment on LG4. A triple-amino acid substitution mutation may cause substantial conformational change that inactivates the biological function of mutant proteins. However, we found that even a single substitution mutation within each of the M1 sequences abolished heparin and cell binding activity of LG4-5 (Fig. 3B). In addition, the circular dichroism spectrum showed that there was no significant difference in conformation between the mutant LG4 proteins and wild-type LG4 protein (data not shown). These results suggest that the overall structure of the mutant LG4 proteins is likely maintained.
The crystal structure of the laminin ␣2 chain LG4-5 modules has been determined (40,41). Because ␣2 LG4-5 shows 41% sequence homology to ␣1 LG4-5, we can predict the threedimensional structure of the laminin ␣1 LG4 module (Fig. 1C). The M1 and M5 sites are located at the same edge in the three-  (G and H), or 100 g/ml EF-1 (I and J). Actin stress fibers and vinculin were visualized as in Fig. 6. Bar, 20 m.

Cell Attachment and Spreading on the Laminin ␣1
LG4 Module OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 dimensional structure of LG4 and share an exclusive surface. This information together with our results that mutations in either one of the sites reduced syndecan and heparin interactions of LG4 (Figs. 2 and 3) suggest that the M1 site in AG73 and M5 in the ␣-dystroglycan binding region work in concert for binding of the LG4 module to syndecans. On the other hand, the M3 site, which is crucial for cell spreading, is located at the opposite side from M1 and M5, suggesting these sites may function independently for spreading.
Cell attachment and spreading are essential cellular processes for cell growth, migration, and invasion (42,43). Cell spreading involves rapid rearrangement of actin stress fibers through integrins to form actin stress fibers. These signals are induced at integrin clustering points, called focal contacts, consisting of several molecules such as integrins, vinculin, paxillin, and actin, and integrins are thought to stabilize focal contacts by integrin-mediated "outside-in" signals (44,45). Fibroblasts on laminin-1 form organized actin stress fibers and focal contacts as previously reported (46). Fibroblasts on rec-LG4 also form actin stress fibers and focal contacts (Fig. 6B). Fibroblasts on rec-M2 decreased the formation of actin stress fibers and focal contacts, and fibroblasts on rec-M3 completely lost these activities. Furthermore, focal contacts in fibroblasts on rec-LG4 were eliminated by anti-integrin ␣2 and ␤1 antibodies, whereas anti-integrin ␣1 and ␣6 antibodies had no effect (Fig. 8A). When fibroblasts on rec-LG4 were stained with anti-integrin ␣2 or ␤1 antibodies, accumulation of integrin ␣2 or ␤1 at focal contacts was observed (Fig. 8B). These results suggest that integrin ␣2␤1 mediates fibroblast cell spreading on LG4. On the other hand, fibroblast attachment and spreading on laminin-1 were not inhibited by heparin and anti-integrin ␣2 antibody but were inhibited by anti-integrin ␣6 and ␤1 antibodies, suggesting that the integrin ␣6␤1 binding site located at the C-terminal region of laminin-1 is the primarily site for cell attachment and spreading when laminin-1 is used as a substrate. Thus, the mechanism of cell attachment and spreading is different in laminin-1 and LG4. Recently, laminin ␣1 LG4-5-deficient mice were created (24). In these mutant mice, Reichert's basement membrane was not formed, similar to whole laminin ␣1-deficient mice (23), despite the presence of the truncated N-terminal part of the laminin ␣1 chain, suggesting a critical role for laminin ␣1 LG4-5 in basement membrane assembly. The embryonic basement membrane between the visceral endoderm and epiblast was also not formed well, resulting in defects of epiblast development, suggesting an important function of laminin ␣1 LG4-5 for epiblast differentiation.
LG4, as a part of laminin-1, may exert its activity in different cell types and developmental stages and may function as an anchoring site between the cell and matrix to form basement membrane in vivo. FIGURE 8. Effects of integrin on actin stress fibers and focal contacts. A, fibroblasts were plated on 8-well glass chamber with 3 g/well of rec-LG4 (a, c, e, and g) or laminin -1 (b, d, f, and h) in the presence of 10 g/ml anti-integrin antibodies ␣1 (a and b), ␣2 (c and d), ␣6 (e and f), or ␤1 (g and h). Selected actin stress fibers and vinculin accumulation were visualized as in Fig. 6. B, immunostaining of fibroblasts plated on rec-LG4 (a and c) or laminin-1 (b and d) with anti-integrin ␣2 (a and b) and ␤1 antibodies (c and d). Integrin ␣2 was present in focal contacts on rec-LG4 but not on laminin-1. Integrin ␤1 was present in focal contacts on both rec-LG4 but not on laminin-1. Bar, 20 m.
Spreading of fibroblasts on rec-LG4 was inhibited by EDTA and anti-integrin ␣2 or ␤1 antibodies, whereas fibroblasts cultured on rec-M3 did not spread. In contrast, cell attachment of fibroblasts on rec-LG4 was inhibited by heparin, and rec-M1 and rec-M5 did not show cell attachment activity. These results may suggest that fibroblast attachment and spreading on LG4 occurs in two steps; fibroblasts initially bind to syndecans, which induces focal enrichment of integrin ␣2␤1 and then results in spreading. A similar stepwise mechanism for cell adhesion involving a cysteine-rich domain of ADAM12, a member of the transmembrane cell adhesion receptor ADAMs family, was proposed. A cysteine-rich domain of ADAM12 binds syndecan that promotes mesenchymal cell spreading through integrin ␤1 (47). In addition, several groups reported that the organization of actin stress fibers and focal contacts through integrin signals requires syndecans (48 -52). For example, melanoma cell attachment to fibronectin through integrin ␣5␤1 requires syndecan-4 clustering to organize focal contacts and actin stress fibers and up-regulation of protein kinase C␣ signaling (48). It has also been reported that integrin ␣v␤3 signaling requires the accumulation of syndecan-1 for carcinoma cell attachment to vitronectin (49), and integrin ␣5␤1 signaling requires the accumulation of syndecan-2 for carcinoma cell attachment to fibronectin (50). A similar cooperation between syndecans and integrin may regulate LG4-mediated cellular processes.
Cleavage of the laminin ␣ chain G domain by endogenous proteolytic processing has been reported (25)(26)(27)(28)(29)(30). Unprocessed laminins and cleaved G domain fragments may have distinct functions in vitro and in vivo. For example, laminin-5 containing the ␣3 chain without LG4-5 promotes keratinocyte adhesion, but the ␣3 chain with LG4-5 induces keratinocyte migration to the leading edge in the wound (27). The cleavage in the ␣2 LG3 module is required for clustering of acetylcholine receptors and for neuromuscular junction formation in concert with agrin (26). Laminin ␣1 may also be cleaved in the G domain since anti-LG4-5-specific antibody stained the ectoplacental cone of laminin ␣1 LG4-5-deficient mice, but antibodies specific to the N-terminal laminin ␣1 chain failed to stain, suggesting that the LG4-5 fragment is generated for ectoplacental cone development (24). Thus, the cleavage product, laminin ␣1 LG4, may have a unique functional role in development, cellular processes, and basement membrane formation. Laminin-1 is a large multifunctional domain protein. Its proteolytic cleavage would provide a mechanism by which new in vivo function was generated at specific times in development.