The Requirement of the Glutamic Acid Residue at the Third Position from the Carboxyl Termini of the Laminin γ Chains in Integrin Binding by Laminins*

Laminins are the major cell-adhesive proteins in the basement membrane, consisting of three subunits termed α, β, and γ. The putative binding site for integrins has been mapped to the G domain of the α chain, although trimerization with β and γ chains is necessary for the G domain to exert its integrin binding activity. The mechanism underlying the requirement of β and γ chains in integrin binding by laminins remains poorly understood. Here, we show that the C-terminal region of the γ chain is involved in modulation of the integrin binding activity of laminins. We found that deletion of the C-terminal three but not two amino acids within the γ1 chain completely abrogated the integrin binding activity of laminin-511. Furthermore, substitution of Gln for Glu-1607, the amino acid residue at the third position from the C terminus of the γ1 chain, also abolished the integrin binding activity, underscoring the role of Glu-1607 in integrin binding by the laminin. We also found that the conserved Glu residue of the γ2 chain is necessary for integrin binding by laminin-332, suggesting that the same mechanism operates in the modulation of the integrin binding activity of laminins containing either γ1 or γ2 chains. However, the peptide segment modeled after the C-terminal region of γ1 chain was incapable of either binding to integrin or inhibiting integrin binding by laminin-511, making it unlikely that the Glu residue is directly recognized by integrin. These results, together, indicate a novel mechanism operating in ligand recognition by laminin binding integrins.

Integrins play important roles in cell-matrix adhesion and signaling events regulating proliferation and differentiation of cells. Among the various integrin family members, ␣6␤1, ␣6␤4, ␣3␤1, and ␣7␤1 have been shown to be the major laminin receptors expressed in many cell types (10). Binding sites for these integrins have been mapped to the C-terminal globular (G) 3 domain of the laminin ␣ chains (6,(11)(12)(13)(14)(15). The G domain consists of five tandemly repeated LG modules of ϳ200 amino acid residues, designated LG1 through LG5. By analogy with the identification of the Arg-Gly-Asp (RGD) cell-adhesive motif in fibronectin, many attempts have been made to identify specific sequences mimicking the integrin binding activity of laminins. However, neither recombinant fragments of the G domain nor synthetic peptides modeled after the sequences in the G domain have shown any significant cell-adhesive activity comparable with that of intact laminins (15)(16)(17)(18). Previously, we found that deletion of the LG4-5 modules did not compromise the ability of laminin-511 4 (␣5␤1␥1) to bind ␣3␤1/␣6␤1 integrins, but further deletion up to the LG3 module resulted in the loss of its integrin binding ability (19). These results indicate that the LG3 module is required for integrin binding by laminin-511, although LG4-5 modules are dispensable. Nevertheless, LG3 or other LG modules of the ␣5 chain did not exhibit any significant activity to bind the ␣6␤1 integrin when recombinantly expressed alone or in tandem with adjacent modules (20), suggesting that G domain per se is not sufficient to recapitulate the integrin binding activity of laminins.
Despite the importance of the G domain in the ␣ chain, several lines of evidence indicate that heterotrimerization with ␤ and ␥ chains is required for laminins to exert their integrin binding activities. Thus, the "E8 fragment" produced by brief elastase digestion of mouse laminin-111 retained almost the full activity of the parental molecule to mediate integrin-dependent cell-substratum adhesion but lost its activity when its ␣1 chain * This study was partly supported by Grants-in-Aid for Scientific Research fragment containing LG1-3 modules was separated from the disulfide-linked ␤1-␥1 dimer fragment (21). Cell adhesive activity was restored to the ␣1 chain fragment upon in vitro recombination and refolding with the ␤1-␥1 dimer (21,22). The importance of heterotrimerization with the ␤ and ␥ chains in the integrin binding activity of laminins has also been demonstrated with the recombinant E8 fragment of laminin-332, which was reconstructed by combining the ␣3 chain fragment containing LG1-3 modules with the disulfide-bonded ␤3-␥2 dimer segment (23). These results indicate that not only the G domain of the ␣ chain but also heterotrimerization of the ␣ chain with the ␤ and ␥ chains within their coiled-coil domains is required for integrin binding by laminins, although the molecular basis of the requirement for the ␤ and ␥ chains remains to be elucidated.
To address the role of ␤ and ␥ chains in integrin binding by laminins, we focused on the putative interface between the G domain of the ␣ chain and the heterotrimerized coiled-coil domains. The coiled-coil domain of mouse laminin-111 has been visualized by electron microscopy as cylinders (24 -26); thus, it is conceivable that the C-terminal regions of the ␤ and ␥ chains are in close proximity to and possibly interact with the G domain, thereby regulating the integrin binding activity of the G domain. Given the extension of nine amino acids in the ␥1 chain against only one amino acid in the ␤1 chain, both of which are C-terminal to the conserved disulfide bond between the ␤1 and ␥1 chains (Fig. 1A), we focused our attention on the role of the C-terminal 9-amino acid extension of the ␥1 chain in integrin binding by laminin-511, a potent ligand for ␣6␤1 and ␣3␤1 integrins. In the present study we examined the integrin binding activities of a series of recombinant laminin-511 mutants with deletions or substitutions within the C-terminal region of the ␥1 chain. Our results show that Glu-1607 at the third position from the C terminus is a prerequisite for laminin-511 recognition by integrins.

EXPERIMENTAL PROCEDURES
Antibodies, Proteins, and Peptides-mAbs against the human laminin ␣5 chain (5D6; Ref. 27), the human laminin ␥1 chain (C12SW), and the human laminin ␣3 chain (2B10) were produced in our laboratory. The function-blocking mAb against integrin ␣6 subunit (AMCI 7-4) was kindly provided by Dr. Masahiko Katayama (Eisai Co., Ltd, Tsukuba, Japan) (28). Hybridoma cells secreting the function-blocking mAb against the integrin ␤1 subunit (AIIB2), developed by Dr. Caroline Damsky (University of California, San Francisco, CA), were obtained from the Developmental Studies Hybridoma Bank (University of Iowa, IA). The function-blocking mAb specific for integrin ␣5 subunit (8F1) was produced in our laboratory (29). Horseradish peroxidase-conjugated anti-His 6 and anti-FLAG mAbs were purchased from Qiagen and Sigma, respectively. A rabbit horseradish peroxidase-conjugated anti-hemagglutinin tag polyclonal antibody was purchased from QED Bioscience (San Diego, CA). A polyclonal antibody against the ACID/BASE coiled-coil peptides was generously provided by Dr. Junichi Takagi (Institute for Protein Research, Osaka University). Anti-GST mAb and horseradish peroxidase-conjugated streptavidin were purchased from Zymed Laboratories Inc. (San Francisco, CA). Human vitronectin was purified from human serum according to Yatohgo et al. (30). Human laminin-332 was purified from the conditioned medium of MKN45 human gastric carcinoma cells. The synthetic peptides modeled after the ␥1 C-terminal sequence (NTPSIEKP; designated as ␥1C peptide) and its mutant form (NTPSIQKP; designated as ␥1C(EQ) peptide) were purchased from Biologica (Nagoya, Japan).
Expression vectors for human laminin ␣3 chain lacking LG4-5 modules (nucleotides 1-4128), ␤3 chain, ␥2 chain, and ␥2 chain mutants were prepared as follows. A full-length cDNA encoding the ␣3 subunit lacking LG4-5 modules (GenBank TM accession number NM_000227) was amplified by reverse tran-scription-PCR as a series of ϳ1-kilobase fragments. After sequence verification, error-free cDNA fragments were ligated in tandem, and the resulting cDNA of the ␣3 chain lacking LG4-5 modules was inserted into the expression vector pcDNA3.1 (Invitrogen). A full-length cDNA encoding the ␤3 subunit (GenBank TM accession number NM_001017402) was purchased from Open Biosystems (Huntsville, AL) and inserted into the expression vector pcDNA3.1. A partial cDNA encoding the ␥2 subunit (GenBank TM accession number NM_005562) was purchased from Open Biosystems, and the remaining portion of the cDNA was amplified by reverse transcription-PCR. After sequence verification, error-free cDNA fragments were ligated in tandem to construct a cDNA encompassing the whole open reading frame, and the resulting cDNA was inserted into the expression vector pcDNA3.1. cDNAs encoding ␥2 mutants were amplified by PCR, and PCR products were inserted into the ␥2 expression vector. The list of primer sequences is available upon request.
Expression and Purification of Recombinant Proteins-Recombinant laminin-511 mutants were produced using the Free-Style TM 293 Expression system (Invitrogen) and purified from conditioned media as described previously (19). Recombinant laminin-311 and laminin-332, both lacking LG4-5 modules, and their mutants were produced using the same expression system. The resultant conditioned media were applied to immunoaffinity columns conjugated with an anti-human laminin ␣3 mAb 2B10. The columns were washed with TBS (50 mM Tris-HCl, pH 7.4, 150 mM NaCl), and bound laminins were eluted with 0.1 M triethylamine, neutralized, and dialyzed against TBS. Recombinant ␣6␤1, ␣3␤1, and ␣V␤3 integrins were produced using the same expression system and purified from the conditioned media using anti-FLAG columns (Sigma) described previously (20).
The recombinant E8 fragment of laminin-511 and its mutants were produced using the Free-Style TM 293 Expression system. The conditioned media were applied to nickel nitrilotriacetic acid affinity columns (Qiagen), and bound proteins were eluted with 200 mM imidazole. The eluted proteins were further purified on anti-FLAG columns (Sigma). After dialysis against TBS, the purity of these recombinant proteins was verified by SDS-PAGE followed by Coomassie Brilliant Blue staining or immunoblotting with anti-His tag (for ␣5E8), anti-hemagglutinin tag (for ␤1E8), or anti-FLAG tag (for ␥1E8) mAbs.
GST fusion proteins (GST-␥1(8AA), GST-RGD, and GST-RGE) were induced in Escherichia coli with 0.1 mM isopropyl-␤-D-thiogalactopyranoside and purified on glutathione-Sepharose 4B columns (Amersham Biosciences) after lysis of the cells by sonication. Bound proteins were eluted with 50 mM Tris-HCl, pH 8.0, containing 10 mM glutathione. After dialysis against TBS, the purity of these recombinant proteins were verified by SDS-PAGE followed by Coomassie Brilliant Blue staining or immunoblotting.
Binding Assays for ␣6␤1, ␣3␤1, and ␣V␤3 Integrins-Solidphase integrin binding assays of recombinant laminins, their mutants, or GST fusion proteins were performed using purified recombinant ␣6␤1, ␣3␤1, and ␣V␤3 integrins. 96-well microtiter plates were coated with recombinant proteins for testing at the indicated concentrations. The amounts of the recombinant proteins adsorbed on the plates were quantified with mAb 5D6 (for laminin-511, E8, and their mutants), mAb 2B10 (for laminin-311, laminin-332, and their mutants), or anti-GST antibody (for GST fusion proteins) to confirm the equality of the amounts of adsorbed proteins. After blocking with 3% BSA, plates were incubated with the 20 nM recombinant ␣6␤1, ␣3␤1, or ␣V␤3 integrin in the presence of 5 mM Mn 2ϩ at 37°C for 1 h. ␣6␤1, ␣3␤1, and ␣V␤3 integrins were used without proteolytic unclasping, since the binding of Mn 2ϩ to the integrin can overcome the structural constraint imposed by an artificially introduced interchain disulfide bond (32). After washing with TBS containing 5 mM Mn 2ϩ and 0.05% Tween 20, bound proteins were quantified after sequential incubations with the biotinylated anti-ACID/BASE antibody and horseradish peroxidaseconjugated streptavidin (20). For integrin binding inhibition assays using function-blocking mAbs against integrin subunits or synthetic peptides, adhesive proteins coated on the microtiter plates were incubated with ␣6␤1 integrin in the presence of mAbs (100 g/ml) or ␥1C or ␥1C(EQ) peptide at 37°C for 1 h before incubation with biotinylated anti-ACID/BASE antibody. Bound integrins were detected as described above. For determination of the apparent dissociation constants by saturation binding assays, plates were coated with laminins (5 nM). After blocking with BSA, serially diluted ␣6␤1 integrin was added to the plates, and bound integrin was quantified. The apparent dissociation constants were determined as described previously (34).
Cell Adhesion Assay-Cell adhesion assays were performed as described previously (19) using K562 human leukemia cells transfected with cDNA encoding an ␣6 integrin subunit (kindly provided by Dr. Arnoud Sonnenberg, The Netherland Cancer Institute, Amsterdam). The cells were maintained in RPMI1640 supplemented with 10% fetal calf serum and used without pretreatment with Mn 2ϩ to fully activate integrins. After fixation with formaldehyde and subsequent staining with Diff-Quik (International Reagents Corp., Kobe, Japan), attached cells were counted in three independent wells using Scion Image software (Scion Corp., Frederick, MD).

Deletion of the C-terminal Region of the ␥1 Chain Abrogates
Integrin Binding Activity-To address the potential involvement of the C-terminal region of the ␥1 chain in integrin bind-ing by laminin, we produced a mutant laminin-511 in which the C-terminal eight amino acids were deleted from the ␥1 chain (designated LN511-␥1⌬8AA; see Fig. 1A). The mutant laminin-511 also lacked the LG4-5 modules that are required for ␣-dystroglycan binding but not for integrin binding (19). A recombinant laminin-511 containing an intact ␥1 chain and the ␣5 chain deleted of LG4-5 (designated LN511-␥1) was used as a control laminin-511 throughout this study. The authenticity of the recombinant proteins as well as other mutant laminin-511 proteins with deletions or amino acid substitutions (see below) was verified by SDS-PAGE under reducing and non-reducing conditions (Fig. 1B). Under reducing conditions, the recombinant proteins gave three bands upon Coomassie Brilliant Blue staining, one corresponding to the ␣5 chain lacking LG4-5 modules and two lower bands corresponding to the ␤1 chain and intact or mutant ␥1 chains, the latter specifically detected by immunoblotting with a mAb against the ␥1 chain (Fig. 1B). Under nonreducing conditions, LN511-␥1 and its deletion mutants gave a slowly migrating band near the top of the gel, confirming that they were purified as trimers of the ␣5, ␤1, and ␥1 chains. Solid-phase binding assays with the ␣6␤1 integrin showed that the control laminin-511 (LN511-␥1) was fully active in binding to the ␣6␤1 integrin, whereas the mutant laminin-511 lacking the C-terminal eight amino acids of the ␥1 chain (LN511-␥1⌬8AA) exhibited only marginal activity comparable with that of another mutant laminin-511 lacking the LG3-5 modules of the ␣5 chain (designated ⌬LG3-5; Fig. 2A), which have been shown to be devoid of integrin binding activity (19). Because LN511-␥1 and LN511-␥1⌬8AA have common ␣5 and ␤1 subunits, the significant reduction in integrin binding activity would be ascribable to the deletion of C-terminal eight amino acids in the ␥1 chain, indicating its involvement in integrin binding by laminin-511.
To narrow down the amino acid residues involved in integrin binding, we produced a series of mutant laminin-511 with shorter deletions within the C-terminal region of the ␥1 chain and examined their integrin binding activities (Fig. 1A). Mutant laminin-511 lacking the C-terminal 1 or 2 amino acids (designated LN511-␥1⌬1AA and LN511-␥1⌬2AA, respectively) retained the integrin binding activity but were less active than control laminin-511 ( Fig. 2A). Reduced integrin binding affinity of these deletion mutants was confirmed by determination of apparent dissociation constants by saturation integrin binding assays; the estimated dissociation constants of LN511-␥1⌬1AA and LN511-␥1⌬2AA for ␣6␤1 integrin were 2.4 Ϯ 0.41 and 2.2 Ϯ 1.4 nM, respectively, ϳ5-fold lower than that of control laminin-511 (K d , 0.43 Ϯ 0.1 nM) (Fig. 3). Further deletion up to the third amino acid residue, i.e. Glu-1607, resulted in a significant abrogation of the integrin binding activity of laminin-511. A dramatic loss of the integrin binding activity upon deletion of the C-terminal three but not two amino acids from the ␥1 chain indicated that Glu-1607 at the third position from the C terminus played a critical role in the potent integrin binding by laminin-511. The C-terminal eight amino acid residues of the ␥1 chain are shaded. C-terminal amino acid sequences of intact ␥1 chain and its mutants are shown in the box. The cysteine residues are underlined. LN511-␥1, recombinant laminin-511 lacking LG4-5 modules but containing intact ␥1 chain; LN511-␥1⌬8AA, LN511-␥1 lacking the ␥1 C-terminal 8 amino acids; LN511-␥1⌬3AA, LN511-␥1 lacking the ␥1 C-terminal 3 amino acids; LN511-␥1⌬2AA, LN511-␥1 lacking the ␥1 C-terminal 2 amino acids; LN511-␥1⌬1AA, LN511-␥1 lacking only the most C-terminal amino acid of the ␥1 chain; LN511-␥1EQ, LN511-␥1 in which the glutamic acid residue (E) was replaced by glutamine (Q, closed box). B, purified LN511-␥1 and its mutants were analyzed by SDS-PAGE on 4% gels under reducing (left and center panels) and non-reducing conditions (right panel) followed by Coomassie Brilliant Blue (CBB) staining (left and right panels) or immunoblotting (IB) with a mAb against the laminin ␥1 chain (center panel). Under reducing conditions, each recombinant protein gave three bands, one corresponding to the ␣5 chain and two lower bands corresponding to the ␤1 and ␥1 chains; the latter was detected by an mAb against the ␥1 chain. Under nonreducing conditions, they gave a single band migrating at ϳ800 kDa, confirming that they were purified as trimers of the ␣5, ␤1, and ␥1 chains. The positions of molecular size markers are shown in the left margin.
To address the role of the Glu-1607 in integrin binding by laminin-511, we produced a mutant laminin-511 in which Glu-1607 of the ␥1 chain was replaced with glutamine (designated LN511-␥1EQ; see Fig. 1A). As expected, the laminin-511 with E1607Q substitution was almost devoid of integrin binding activity (Fig. 2B), confirming the importance of Glu-1607 in integrin binding by laminin-511.
Integrin Binding Activity of the E8 Fragment of Laminin-511 and Its Mutant Forms-The failure in the disulfide bond formation between the ␤1 and ␥1 chains near the C terminus has been shown to destabilize the conformation of the coiled-coil domain of laminin (26). Because the deletions and amino acid substitution introduced to the ␥1 chain were very close to the conserved cysteine residue near the C terminus, these mutations could compromise the disulfide bond formation between ␤1 and ␥1 chains, resulting in the destabilization of the coiledcoil domain of the mutant proteins and, hence, the abrogation of their integrin binding activities.
To examine this possibility, we produced E8 fragments of control and mutant laminin-511 consisting of three truncated subunits modeled after the E8 fragment of laminin-111 (Fig. 4), since the truncated fragments of ␤1 and ␥1 chains contain only one Cys residue, allowing us to assess the effect of the mutations introduced to the C-terminal region of the ␥1 chain on the disulfide bond formation between ␤1 and ␥1 chains. The authenticity of the resulting recombinant E8 fragments of laminin-511 was verified by SDS-PAGE and immunoblotting with a mAb against the FLAG tag, which was added to the N terminus of the truncated ␥1 chain. Under reducing conditions, each recombinant protein gave three bands upon Coomassie Brilliant Blue staining, one corresponding to the truncated ␣5 chain and two lower bands corresponding to the truncated ␤1 and ␥1 chains (Fig. 4B). Control and mutant E8 fragments gave two bands under nonreducing conditions, one corresponding to the truncated ␣5 chain and another corresponding to the heterodimer of the truncated ␤1 and ␥1 chains. Because the truncated ␤1 and ␥1 chains contain only one Cys residue near the C terminus, these results confirmed that the disulfide bond formation between these two chains was not compromised by the mutations introduced near the C terminus of the ␥1 chain.
The E8 fragment of control laminin-511, designated E8-␥1, exhibited potent integrin binding activity (Fig. 4C). Saturation integrin binding assays demonstrated that the K d of E8-␥1 for the ␣6␤1 integrin was 1.5 Ϯ 0.02 nM, roughly comparable with that of the parental protein (i.e. LN511-␥1). However, two E8 mutants with either deletion of the C-terminal three amino acids (E8-␥1⌬3AA) or substitution of Gln for Glu-1607 (E8-␥1EQ) were almost devoid of integrin binding activity, further confirming the importance of Glu-1607 in integrin binding by laminin-511. These results also demonstrated that the loss of integrin binding activity by deletion of the C-terminal three amino acids or E1607Q substitution was not due to a defect in disulfide bond formation between the ␤1 and ␥1 chains at their C termini, but rather, it was due to the critical role of the region containing Glu-1607 in sustaining the active conformation of the G domain or to its involvement in integrin binding by comprising a part of the integrin recognition surface. The integrin binding to E8-␥1 was strongly inhibited by function-blocking mAbs against integrin ␣6 and ␤1 subunits. It was completely blocked by a combination of anti-␣6 and anti-␤1 mAbs (Fig.  4D), confirming the integrin binding specificity of E8-␥1.

Reduced Cell-adhesive Activity of Laminin-511 with E1607Q
Substitution-The importance of Glu-1607 in integrin binding was further examined by cell adhesion assays using K562 cells stably transfected with ␣6 integrin (35,36). To avoid the auxiliary interactions of laminin-511 with ␣V␤3 and other integrins recognizing the Arg-Gly-Asp sequences within domain IVa (also designated domain L4b) of the ␣5 chain (37), we examined the cell-adhesive activities of the E8 fragments of control and mutant laminin-511 proteins. Mutant E8 fragments either lacking the C-terminal three amino acids or having E1607Q substitution were almost devoid of cell-adhesive activity, whereas the control E8 fragment was fully active in mediating cell adhesion of K562 cells expressing the ␣6␤1 integrin (Fig. 5). Control K562 cells untransfected with the ␣6 integrin did not adhere to any of the recombinant E8 fragments, confirming the specificity of the ␣6␤1 integrin-dependent cell adhesion. These results further support our conclusion that the C-terminal region of the ␥1 chain, in particular Glu-1607 at the third position from the C terminus, played an important role in integrin binding by laminin-511. increasing concentrations of LN511-␥1 and E8 fragments of laminin-511 (E8-␥1) and its mutant proteins (E8-␥1⌬3AA and E8-␥1EQ), blocked with BSA, and then incubated with 20 nM ␣6␤1 integrin. D, inhibition of integrin binding to E8-␥1 using anti-integrin mAbs. 96-well microtiter plates were coated with E8-␥1 (5 nM). ␣6␤1 integrin (20 nM) was preincubated with function-blocking mAbs against integrin subunits and then added to the precoated wells. The amounts of bound ␣6␤1 integrin were quantified as described under "Experimental Procedures." Each column and bar represents the mean of triplicate assays and the S.D., respectively. The C-terminal Region of the ␥1 Chain Does Not Function as an Integrin-binding Site-The binding profiles of the ␥1⌬3AA and ␥1EQ mutants of laminin-511 toward ␣6␤1 integrin raised the possibility that the C-terminal region of the ␥1 chain could comprise part of the binding site for ␣6␤1 integrin. To explore this possibility, we expressed the C-terminal eight amino acid residues of the ␥1 chain in bacteria as a GST fusion protein (GST-␥1(8AA): Fig. 6A) and examined whether it could bind to the ␣6␤1 integrin. Neither GST alone nor GST-␥1(8AA) showed any significant integrin binding activity, even at the coating concentrations as high as 200 nM, although control LN511-␥1 bound a significant amount of the integrin even at 5 nM (Fig. 6B), making it unlikely that the C-terminal peptide segment of the ␥1 chain is directly recognized by the ␣6␤1 integrin. The failure of GST-␥1(8AA) to bind to the ␣6␤1 integrin could arise from the bulky GST domain, blockading direct interaction of the ␥1 peptide with the integrin, although the RGD-but not RGE-containing peptide segment fused to GST exhibited significant binding activity toward the ␣V␤3 integrin (Fig. 6C), making this possibility unlikely.
To further explore the possibility that the ␥1 C-terminal peptide directly binds to the integrin, we examined whether a synthetic peptide modeled after the ␥1 C-terminal sequence was capable of inhibiting integrin binding. Thus, binding of the soluble ␣6␤1 integrin to immobilized LN511-␥1 was determined in the presence of increasing concentrations of the synthetic 8 amino acid peptide derived from the C-terminal region of the ␥1 chain (designated ␥1C peptide) or its mutant form, in which Glu-1607 was replaced by glutamine (␥1C(EQ) peptide; Fig. 7) The ␥1C peptide did not exert any significant inhibitory effect on integrin binding to LN511-␥1 even at 2.5 mM (125,000-fold molar excess to the integrin added) as compared with the control peptide with the E1607Q substitution. These results argue against the direct recognition of the C-terminal region of the ␥1 chain by the integrin and support the possibility that the ␥1 chain modulates the integrin binding activity of laminin-511 through interaction with the G domain of the ␣5 chain, possibly stabilizing the conformation of the G domain active in integrin binding.
Glutamic Acid Residue at the Third Position from the C Terminus of the ␥2 Chain Is Required for Integrin Binding by Laminin-332-The glutamic acid residue at the third position from the C terminus is conserved between laminin ␥1 and ␥2 chains (Fig. 9). The presence of the conserved glutamic acid residue (Glu-1191) in the ␥2 chain raised the possibility that this Glu residue is also required for integrin binding of the ␥2 chain-containing laminin, i.e. laminin-332 (␣3␤3␥2). To address this possibility, we expressed and purified laminin-332 lacking LG4-5 modules (designated LN332-␥2) and its mutant proteins either lacking the C-terminal three amino acids of the ␥2 chain (LN332-␥2⌬3AA) or having E1191Q substitution within the ␥2 chain (LN332-␥2EQ) and examined their ␣3␤1 integrin binding activities. Recombinant LN332-␥2 was found to possess potent ␣3␤1 integrin binding activity, comparable with that of laminin-332 purified from a conditioned medium of MKN45 human carcinoma cells (Fig. 9). However, neither ␥2 chain mutants, LN332-␥2⌬3AA and LN332-␥2EQ, showed any significant binding to the ␣3␤1 integrin. These results indicate that Glu-1191 within the ␥2 chain is involved in integrin binding by laminin-332 and provide evidence that the C-terminal region of the ␥ chains, particularly the Glu residue at the third position from the C terminus, has a critical role in integrin binding by laminins containing either ␥1 or ␥2 chains.

DISCUSSION
Despite accumulating evidence of the requirement of heterotrimerization of the ␣ chain with a disulfide-bonded dimer of the ␤ and ␥ chains in integrin binding by laminins (21)(22)(23), no studies have before directly addressed the roles of ␤ and ␥ chains in integrin binding through production of mutant laminin isoforms with deletions and/or amino acid substitutions in the ␤ and/or ␥ chains. In the present study we provide evidence that three amino acids within the ␥1 C-terminal region, particularly Glu-1607 at the third position from the C terminus, are required for the integrin binding activity of laminin-511. This conclusion is based on the following observations. First, deletion of the C-terminal three but not two amino acids within the ␥1 chain completely abrogated the integrin binding activity of laminin-511, as was the case with substitution of Gln for Glu-1607 at the third position from the ␥1 C terminus. Second, the  C-terminal region of the ␥1 chain is required for not only LN-511 but also LN-311 to bind to integrins. Third, the Glu residue at the third position from the C terminus is conserved for both ␥1 and ␥2 chains, and deletion of the C-terminal three amino acids within the ␥2 chain or substitution of Gln for Glu-1191 in the ␥2 chain completely abrogated the integrin binding activity of laminin-332.
One of the likely explanations for the requirement of the ␥1 C-terminal region, particularly the Glu residue at the third position from the C terminus, in integrin binding by laminins could be that Glu-1607 comprises part of the integrin recognition site of ␥1 chain-containing laminins (Fig. 10A). Because the metal ion bound to the MIDAS (metal ion-dependent adhesion site) motif within integrins directly coordinates the side chain of acidic residues in ligands (39), acidic residues such as Asp or Glu are considered to act as key residues in ligands to interact with integrins (40,41). However, despite the importance of the Glu-1607 in integrin binding by ␥1 chain-containing laminins, the C-terminal peptide segment of the ␥1 chain fused to GST and did not show any significant activity to bind the ␣6␤1 integrin, and the synthetic peptide modeled after the C-terminal peptide sequence of the ␥1 chain did not exert any significant inhibitory effect on integrin binding by laminin-511 even at concentrations as high as 2.5 mM. These results argue strongly against the possibility that the C-terminal segment of the ␥1 chain including Glu-1607 functions as an integrin-binding site, although we cannot exclude the possibility that the ␥1 C-terminal segment comprises part of the integrin recognition site, whose contribution to integrin binding by laminins is rather auxiliary and not sufficient to be detected in direct integrin binding or integrin binding competition assays. Another potential explanation for the requirement of the C-terminal region of the ␥1 chain in integrin binding by laminins could be that deletion or replacement of Glu-1607 may result in instability in laminin chain assembly and interferes with disulfide bond formation between ␤ and ␥ chains near their C termini. Because the conformation of the coiled-coil domain of laminin has been shown to be destabilized by disruption of the C-terminal disulfide bonds between ␤ and ␥ chains (26), deletion of the C-terminal three amino acids from the ␥1 chain or E1607Q substitution could impair ␤-␥ dimer formation through disulfide bonding, thereby leading to destabilization of the conformation of laminin-511 and, hence, inactivation of its integrin binding activity. However, SDS-PAGE analysis of purified E8-␥1⌬3AA and E8-␥1EQ under non-reducing conditions showed that a disulfide bond was spontaneously formed between truncated ␤1 and ␥1 chains (Fig. 4B), making this possibility unlikely.
The third explanation for the requirement of the C-terminal region of the ␥1 chain in integrin binding is that the Glu residue at the third position from the C terminus of ␥1 chain directly interacts with the G domain and alters its local or global conformation, thereby exposing the putative integrin-binding site within the G domain that otherwise remains cryptic. A hypothetical model for the structure of the entire G domain has been proposed (14) based on the three-dimensional structure of LG4-5 modules of the ␣2 chain and the known length of interdomain linkers (42). The model predicts that LG1-3 modules have the shape of a cloverleaf and that the coiled-coil domain adjacent to the LG1-3 modules seems to make direct contact with the cloverleaf. This model together with the results obtained in this study raises the possibility that the Glu residue in the C-terminal region of ␥ chains is in direct contact with one of the LG modules and alters either the local conformation of individual LG1-3 modules or the global configuration of the tandem array of LG1-3 modules, thereby exposing the putative integrin-binding site within LG1-3 modules (Fig. 10B). Previously, we demonstrated that the LG3 module is indispensable for binding to the ␣6␤1 integrin (19), although neither the LG3 module alone nor a tandem array of LG1-3 modules was capable of binding to the ␣6␤1 integrin (20), suggesting that a tandem array of the LG1-3 modules per se is not sufficient to recapitulate the integrin binding activity of laminins. It is tempting, therefore, to speculate that the integrin binding activity depends on both strictly regulated conformation of LG1-3 modules and Glu-1607 of the ␥1 chain, the latter interacting with LG1-3 modules and thereby stabilizing their active conformation, so that deletion or replacement of the Glu-1607 results in abrogation of the integrin binding activity of LG1-3 modules through destabilization of their active conformation. Further investigation is needed to confirm this possibility through defining the putative site within the G domain that interacts with Glu-1607.
We also showed that Glu-1191 within the ␥2 chain, which is located at the third position from the C terminus, is also required for the integrin binding activity of laminin-332, extending the importance of the Glu residue at the third posi- In model A, the Glu residue within the C-terminal region of the ␥ chain is directly involved in integrin binding by laminin as a critical amino acid residue recognized by integrin. In model B, the C-terminal Glu residue of the ␥ chain interacts with the G domain and induces (or exposes) the putative integrin-binding site within the G domain through conformational modulation. Alternatively, the C-terminal Glu residue may also comprise part of the integrin recognition site, whose contribution to integrin binding by laminins is rather auxiliary and not sufficient to be detected in direct integrin binding assays or integrin binding inhibition assays.
tion from the C terminus to the ␥2 chain. Previously, a novel variant ␥2 chain arising from alternative splicing of the 3Ј most exon was identified and designated ␥2 transcript variant 2 (GenBank TM accession number NM_018891; Refs. 43 and 44). This variant ␥2 chain is shorter than the normal ␥2 chain and lacks Glu-1191. It seems likely, therefore, that the laminin isoform containing the variant ␥2 chain is defective in binding to the ␣3␤1 integrin due to the absence of Glu-1191 that is required for the integrin binding by laminin-332. The transcript for the variant ␥2 chain was detected in cerebral cortex, lung, and distal tubules of the kidney (43). Generation of two splice variants of laminin ␥2 chains with and without integrin binding activity may be a novel mechanism modulating the interaction of cells with the basement membrane containing laminin-332 in these tissues.
It is interesting to note that a novel ␥ chain isoform, ␥3 (GenBank TM accession number NM_006059), also lacks the Glu residue that is conserved between the ␥1 and ␥2 chains (45). Furthermore, the C-terminal region of the ␥3 chain consists of only four amino acid residues after the conserved Cys residue, significantly shorter than the corresponding regions of other ␥ chains. These differences between ␥3 and other ␥ chains raise the possibility that the laminin isoform containing the ␥3 chain is defective in binding to integrins. Recently, Yan and Cheng (46) reported, however, that the ␥3 chain forms a heterotrimer with ␣3 and ␤3 chains and serves as a ligand for the ␣6␤1 integrin at the apical ectoplasmic specialization in rat testes. Thus, the ␥3 chain-containing laminin may well retain integrin binding activity despite the absence of the conserved C-terminal Glu residue in the ␥3 chain. Heterotrimerization of the ␥3 chain with ␣3 and ␤3 chains suggests that the laminin containing the ␥3 chain may circumvent the loss of the Glu residue, so that the G domain of laminin-333 can maintain active conformation in the absence of the conserved Glu residue in the ␥3 chain. However, laminin-333 has been neither purified nor examined directly for its cell-adhesive and integrin binding activities. It remains to be determined whether laminin-333 is capable of binding to ␣6␤1 and other laminin-binding integrins and, if so, how potent the binding is as compared with the binding of laminin-332 to ␣6␤1 and other laminin binding integrins.
Little is known about the role of the laminin ␤ chain in integrin binding by laminins. Although the YIGSR sequence within the short arm of the ␤1 chain has been proposed as an active site that mediates cell attachment, migration, and receptor binding (47)(48)(49), the biological significance of this sequence remains to be confirmed. In contrast to the nine amino acid extension C-terminal to the conserved Cys residue in ␥1 and ␥2 chains, the C-terminal region of the ␤1, ␤2, and ␤3 chains contains only one amino acid extension C-terminal to the conserved Cys residue, making it less likely that the C-terminal region of ␤ chains is directly involved in modulation of the integrin binding activity of laminins, although we cannot yet exclude the possibility. Recently, Nishimune et al. (50) demonstrated that the laminin ␤2 chain, a component of the synaptic cleft at the neuromuscular junction, binds directly to calcium channels that are required for neurotransmitter release from motor nerve terminals. Furthermore, they showed that a C-terminal 20-kDa frag-ment of the ␤2 chain, in particular leucine-arginine-glutamate tripeptide within this fragment, is necessary and sufficient for laminin interactions with voltage-gated calcium channels, suggesting that the C-terminal region of the ␤ chain functions as the binding site for protein(s) other than integrins.
In conclusion, our data showed that the Glu residue within the C terminus of the ␥1 chain is necessary for binding to ␣6␤1/ ␣3␤1 integrins and is potentially involved in maintenance of the active conformation of the G domain. It is interesting to note that this Glu residue is highly conserved from nematodes to mammals, suggesting that molecular mechanisms underlying integrin binding to laminins are evolutionarily conserved throughout metazoans. Our results provide new insights into the modulatory role of the ␥ chain in integrin binding by laminins.