The LG1-3 Tandem of Laminin α5 Harbors the Binding Sites of Lutheran/Basal Cell Adhesion Molecule and α3β1/α6β1 Integrins*

The laminin-type globular (LG) domains of laminin α chains have been implicated in various cellular interactions that are mediated through receptors such as integrins, α-dystroglycan, syndecans, and the Lutheran blood group glycoprotein (Lu). Lu, an Ig superfamily transmembrane receptor specific for laminin α5, is also known as basal cell adhesion molecule (B-CAM). Although Lu/B-CAM binds to the LG domain of laminin α5, the binding site has not been precisely defined. To better delineate this binding site, we produced a series of recombinant laminin trimers containing modified α chains, such that all or part of α5LG was replaced with analogous segments of human laminin α1LG. In solid phase binding assays using a soluble Lu (Lu-Fc) composed of the Lu extracellular domain and human IgG1 Fc, we found that Lu bound to Mr5G3, a recombinant laminin containing α5 domains LN through LG3 fused to human laminin α1LG4–5. However, Lu/B-CAM did not bind other recombinant laminins containing α5LG3 unless α5LG1–2 was also present. A recombinant α5LG1-3 tandem lacking the laminin coiled coil (LCC) domain did not reproduce the activity of Lu/B-CAM binding. Therefore, proper structure of the α5LG1-3 tandem with the LCC domain was essential for the binding of Lu/B-CAM to laminin α5. Our results also suggest that the binding site for Lu/B-CAM on laminin α5 may overlap with that of integrins α3β1 and α6β1.

Laminins are a family of heterotrimeric glycoproteins that assemble into basement membranes. The laminin-511 trimer (␣5␤1␥1) is a major isoform widely expressed in fetal and adult tissues. Many of the biological functions of laminin-511 are mediated through the ␣5 subunit. Mice lacking laminin ␣5 die during late embryogenesis with several developmental defects, including defects in neural tube formation, digit separation, placentation, glomerulogenesis, lung lobe separation, intestinal smooth muscle development, and tooth morphogenesis (1)(2)(3)(4)(5)(6). A grafting experiment to bypass embryonic lethality has shown that laminin ␣5 is also required for hair follicle development (7). Deletion of ␣5 during lung development causes a delayed differentiation of distal epithelial cells and a reduced capillary density associated with a decrease in lung vascular endothelial growth factor production (8). Moreover, a hypomorphic mutation in laminin ␣5 causes polycystic kidney disease (9).
All five laminin ␣ chains have a laminin-type globular (LG) 2 domain at their COOH termini, which consists of five homologous domains (LG1-LG5). Molecular dissection of the laminin ␣5 LG domain in vivo revealed that the LG1-2 tandem harbors most of the functionality of the ␣5LG domain with regard to mediating developmental processes, whereas the ␣5LG3-5 tandem appears to be involved in glomerular filtration (5,10).
In previous studies, we have reported that laminin ␣5 (but not the other four laminin ␣ chains) is a ligand for Lu/B-CAM (17,21). We have also reported that the Lu/B-CAM binding site maps within the ␣5LG domain and that the ␣5LG3 domain is essential for Lu/B-CAM binding to laminin ␣5. Similarly, several groups have attempted to identify binding sites for inte-grins, syndecans, and ␣-dystroglycan within laminin ␣ chain LG domains (22,23). Yu and Talts (24) reported that ␣5LG1-3 and ␣5LG4-5 recombinant fragments had cell-adhesive activity dependent on ␣3␤1/␣6␤1 integrins and ␣-dystroglycan, respectively. Recently, Ido et al. (25) produced recombinant laminin trimers containing a modified ␣5 chain with serial deletions of LG1-5 and found that ␣5LG3 was essential for binding of the ␣3␤1 and ␣6␤1 integrins. These results lead to the conclusion that ␣5LG3 is critical for the binding of Lu/B-CAM and integrins ␣3␤1 and ␣6␤1. However, the precise structure of the binding sites for these receptors within ␣5LG remains unclear.
Here we produced a series of recombinant laminins to narrow the binding site of Lu/B-CAM within the laminin ␣5LG domain. Solid phase binding assays to these laminins were performed with a soluble recombinant Lu (Lu-Fc) composed of the extracellular domain of Lu fused to IgG1 Fc. Lu/B-CAM binding to laminin required the ␣5LG1-3 tandem, as did integrins ␣3␤1 and ␣6␤1 binding to laminin. Furthermore, we examined competition between Lu/B-CAM and ␣3␤1/␣6␤1 integrins for binding to ␣5LG1-3. Our results indicate that the binding site for Lu/B-CAM on ␣5LG1-3 overlaps with that of the ␣3␤1/ ␣6␤1 integrins.

EXPERIMENTAL PROCEDURES
Cell Culture-Human embryonic kidney (HEK293) and human fibrosarcoma (HT1080) cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA) and Health Science Research Resources Bank (Osaka, Japan), respectively. Both cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum. K562 human erythroleukemia cells transfected with cDNA encoding human integrin ␣3A or ␣6A subunits were kindly provided by Dr. Arnoud Sonnenberg (The Netherlands Cancer Institute) and maintained in RPMI 1640 medium containing 10% fetal calf serum and 1 mg/ml G418 (Sigma) (26).
Construction of Expression Vectors-Laminin chain expression vectors were prepared as follows. cDNA clones encoding full-length mouse laminin ␣5 with the FLAG sequence at the COOH terminus (generated in our laboratory) and human laminin ␣1 (provided by Dr. Karl Tryggvason, Stockholm, Sweden) were used to construct expression vectors encoding the chimeric laminin ␣ chains Mr51, Mr5G1, Mr5G2, Mr5G3, Mr5G13, and Mr5G23. PCR was used to introduce restriction sites at the appropriate locations and to seamlessly join amplified fragments with overlapping sequences by sequential PCR (30). For all PCR, Vent polymerase (New England Biolabs, Beverly, MA) was used according to the manufacturer's instructions. Preparation of Mr5, Mr51, Mr5G2, and Mr5G3 has been described previously (17). For generating Mr5G1, we took advantage of a unique AgeI site at the end of the sequence encoding ␣5LG1. To construct Mr5G1, human ␣1LG2-5 was amplified with sense (5Ј-AGGCGCGCCTCTAGAGCACCG-GTTCCCAGAATGAAGACCCTTCC-3Ј) and antisense (5Ј-AGGCGCGCCCGTACGTCAGGACTCGGTCCCAGGAC-3Ј). This segment was ligated to the cDNA encoding ␣5 domains LN through LG1. To construct Mr5G13, we used AgeI and SpeI sites in sequences encoding the ␣5LG domain. Human ␣1LG2 was amplified with sense (using the same sense primer to generate Mr5G1) and antisense (5Ј-AGTCATGGTGCG-TCCCACTAGCAGGAAGCTAACACTCCGGAT-3Ј); mouse ␣5LG3 was amplified with sense (5Ј-ATCCGGAGTGTTAGC-TTCCTGCTAGTGGGACGCACCATGACT-3Ј) and antisense (5Ј-CCGACCACATGTCGGATGCGA-3Ј). These two products were mixed and subjected to PCR again for 20 cycles to join them. The amplified PCR product was inserted into the AgeI and SpeI sites of Mr5G3. To construct Mr5G23, we used the Mr51 cDNA to make a chimeric segment of ␣5LCC and ␣1LG1. The chimeric segment was amplified with sense (5Ј-GCCCTGA-ATGAGCTGGCATC-3Ј) and antisense (5Ј-ACCGGTCTGGG-AGCTTCCGAAGCACCCA-3Ј). The amplified PCR product was inserted into the MluI and AgeI sites of Mr5G3. The fulllength mouse laminin ␣5 with the COOH-terminal FLAG tag and chimeric ␣ chains were then cloned into pcDNA3.1Zeo(ϩ) (Invitrogen) and used to produce recombinant laminins. cDNA clones encoding mouse laminin ␤1 chains (provided by Dr. Albert E. Chung, University of Pittsburgh, Pittsburgh, PA) and ␥1 (provided by Dr. Peter D. Yurchenco, Robert Wood Johnson Medical School, Piscataway, NJ) were cloned into pcDNA3.1Neo(ϩ) (Invitrogen) and pIREShyg2 (BD Biosciences Clontech), respectively.
The expression vector containing ␣5LG1-3 was prepared as follows. The DNA sequence encoding the LG1-3 domain of laminin ␣5 was amplified by reverse transcription-PCR using RNA from mouse embryonic endothelial cells and the primer pair sense (5Ј-GTCAGCTAGCCGCCAGCAAGGTCAAGG-3Ј) and antisense (5Ј-GTCACTCGAGCTACCTTGAGGTCT-CGATGAG-3Ј). The PCR product was inserted into the NheI-XhoI site of the episomal expression vector pCEP-Pu containing the BM-40 signal peptide (31).
An expression vector containing Lu-Fc was prepared as follows. A DNA fragment encoding human IgG1 Fc was prepared from the CD4-Ig vector (32) and inserted into the BamHI-XbaI sites of pcDNA3.1Neo(ϩ). A full-length human Lu cDNA was purchased from Invitrogen and used as a template for PCR. DNA sequences encoding the extracellular domain of Lu/B-CAM was amplified using sense (5Ј-CGGGATCCGCCACCA-TGGAGCCCCCGGACGCACCG-3Ј) and antisense (5Ј-CGGG-ATCCACTCCAGCCTGGGAGGTCTG-3Ј). The PCR product was inserted into the BamHI site of the above human IgG1 Fc expression vector.
Expression and Purification of Recombinant Proteins-HEK293 cells were transfected with the mouse laminin ␥1 expression vector using Lipofectamine 2000 (Invitrogen), and stable clones were selected using 400 g/ml hygromycin (Invitrogen). All further cell culture and clonal expansion were carried out in the presence of the relevant antibiotics for positive selection. A clone highly expressing mouse laminin ␥1 was then transfected with the mouse laminin ␤1 expression vector, and stable clones were selected using 800 g/ml G418 (Sigma). A clone highly expressing laminin ␥1 and ␤1 was transfected with mouse laminin ␣5 or chimeric ␣ chain expression vectors, and stable clones were selected using 400 g/ml Zeocin (Invitrogen). The selected clones were grown to confluency in 24-well culture plates with DMEM containing 10% fetal bovine serum. The confluent cells were incubated in serum-free DMEM for four days. The conditioned media were harvested and clarified by centrifugation at 10,000 revolutions/min for 20 min. They were screened for the expression of recombinant laminins by immunoblotting with antibody against the laminin ␣5 domain LN/LEa. The clones showing high secretion were grown to confluency in culture dishes with DMEM containing 10% fetal bovine serum. The confluent cells were incubated in serum-free DMEM for four days. The conditioned media were harvested and clarified by sequential centrifugation at 500 revolutions/min for 10 min and 10,000 revolutions/min for 20 min. The collected media were precipitated with ammonium sulfate at 80% saturation. The resulting precipitates were collected by centrifugation at 10,000 revolutions/min for 30 min and then dissolved in and dialyzed against PBS(Ϫ). The 30-fold concentrated media were used for purification. The recombinant laminins were purified from the conditioned culture media by anti-FLAG or anti-human laminin ␣1LG4 -5 monoclonal antibody coupled to agarose. The eluted fractions were pooled and dialyzed against 10 mM Tris-HCl, pH 7.5, 150 mM, 0.1% CHAPS, 1 mM EDTA, and 1% sucrose.
Human embryonic kidney EBNA-293 cells were transfected with the ␣5LG1-3 expression vector and the serum-free media were collected as described previously (31). The serum-free media were passed over a heparin-Sepharose column (2 ϫ 25 cm) equilibrated in 50 mM Tris-HCl, pH 7.4, and eluted with 0 -0.5 M NaCl gradient. The second purification was achieved on Superose 12 HR16/50 (GE Healthcare Biosciences) equilibrated with 0.2 M ammonium acetate. The pooled fraction was further purified on MonoQ HR5/5 (GE Healthcare Biosciences) equilibrated with 20 mM Tris-HCl, pH 8.0, and eluted with a linear gradient of 0 -0.6 M NaCl.
HEK293 cells were transfected with Lu-Fc expression vector using Lipofectamine 2000 (Invitrogen), and stable clones were selected using 1 mg/ml G418 (Sigma). The conditioned media were prepared as described above. Recombinant protein was purified from the conditioned media by protein-A-Sepharose (GE Healthcare Biosciences). The eluted fractions were pooled and dialyzed against PBS(Ϫ).
Electrophoretic Analysis and Immunoblotting-SDS-PAGE was carried out on 5 or 7.5% gels under reducing or non-reducing conditions. Proteins were visualized with silver staining or Coomassie Brilliant Blue. For immunoblotting, sample proteins were separated by SDS-PAGE and transferred onto polyvinyli-dene difluoride membranes. Proteins on the membrane were reacted with first antibodies followed by incubation with secondary antibody conjugated with horseradish peroxidase (GE Healthcare Biosciences). Bound antibodies were visualized with ECL Western blotting detection reagents (GE Healthcare Biosciences).
Solid Phase Binding Assay-Solid phase binding assays were carried out with recombinant proteins coated onto the plastic surface of microtiter plates. Plates were blocked with PBS(Ϫ) containing 1% bovine serum albumin and incubated with Lu-Fc at 37°C for 1 h. After washing with PBS(Ϫ), the bound Lu-Fc was detected with a biotinylated anti-human IgG1 Fc monoclonal antibody (Sigma). After further washing, the bound antibodies were detected by the addition of streptavidin-conjugated horseradish-peroxidase (GE Healthcare Biosciences) followed by the addition of 1 mg/ml o-phenylenediamine and 0.012% H 2 O 2 . The absorbance was measured at 450 nm with a Microplate Reader Model 550 (Bio-Rad).
Cell Adhesion Assays-Cell adhesion assays were performed as described previously (13). Briefly, 96-well microtiter plates (Nunc, Roskilde, Denmark) were incubated with recombinant laminins at 37°C for 1 h and then blocked with PBS(Ϫ) containing 1% bovine serum albumin for another hour. K562 transfectants were suspended in serum-free RPMI 1640 medium at 4 ϫ 10 5 cells/ml and preincubated with or without 1 g/ml anti-integrin ␤1 monoclonal antibody TS2/16 at room temperature for 10 min. 50 l of cell suspension were added to the laminin-coated wells and incubated at 37°C for 1 h. The attached cells were fixed with 4% formaldehyde, stained with Diff-Quik (International Regents Corp., Kobe, Japan), and counted under the microscope.
Cell adhesion inhibition assays were performed on the basis of the adhesion assay described above. HT1080 cells were suspended in serum-free DMEM at 4 ϫ 10 5 cells/ml and preincubated with 1 g/ml anti-integrin ␤1 monoclonal antibody TS2/16 at room temperature for 10 min. The cells were further preincubated with 10 g/ml monoclonal antibodies against different integrins and 20 g/ml recombinant Lu-Fc protein at room temperature for 10 min. TS2/16-stimulated K562 transfectants were also preincubated with 20 g/ml recombinant Lu-Fc protein at room temperature for 10 min. The preincubated cells were transferred to plates coated with 10 g/ml of the recombinant laminins. After incubation at 37°C for 20 min, the attached cells were stained and counted as described above.

RESULTS
Production of Recombinant Laminins-In a previous study, we reported that ␣5LG3 is essential for Lu/B-CAM binding to laminin ␣5 (17). Although our results showed that the ␣5LG4-5 tandem is not required for Lu/B-CAM binding, whether or not ␣5LG1-2 is required remains an open question. To further define the Lu/B-CAM binding site on the laminin ␣5 LG domain, we produced a series of recombinant laminin trimers containing chimeric ␣5 chains in which all or part of laminin ␣5LG was replaced with analogous segments of human laminin ␣1 (Fig. 1). Although the functional structure of laminins has in some cases been examined by producing deletion mutants (25,33), the deletion approach often leads to disruptions in the ter-tiary structure. Therefore, we undertook a chimeric approach using laminin ␣5 and ␣1 chains. The human laminin ␣1LG domain does not bind to Lu/B-CAM, and it also bears an epitope in the ␣1LG4-5 tandem recognized by a monoclonal antibody.
In our previous study, we investigated the binding site of Lu/B-CAM through analysis of chimeric laminin chains produced in transgenic mice in vivo (17). However, it is not feasible to continue to generate and maintain transgenic mice expressing chimeric ␣ chains. Two groups have already reported that it is possible to produce recombinant laminin-511 by sequential transfection of cDNAs encoding human laminin ␣5, ␤1, and ␥1chains into HEK293 cells (25,34,35). Similarly, here we have produced recombinant laminins containing full-length ␣5 or chimeric ␣5/␣1 chains in HEK293 cells. Secretion of endogenous human laminins was not detected (data not shown). Recombinant proteins secreted into the culture media were screened by immunoblotting with a polyclonal anti-laminin ␣5 domain LN/LEa serum. Conditioned media were subjected to SDS-PAGE and separated under non-reducing conditions. Immunoblotting revealed a band migrating at ϳ800 kDa, showing that the recombinant laminins were produced as disulfide cross-linked heterotrimers (data not shown) (10). The recombinant proteins were purified from the culture media using agarose conjugated with either anti-FLAG or anti-␣1LG4 -5 monoclonal antibodies. The purified proteins were subjected to SDS-PAGE under reducing conditions (Fig. 2). All recombinant proteins gave three bands upon silver staining; one corresponded to the ␣ chain, with a relative molecular mass of 350 kDa, and the two lower bands corresponded to the ␤1 and ␥1 chains. 300 ml of conditioned culture medium yielded ϳ40 g of recombinant laminin. Furthermore, all recombinant laminins reproduced heparin binding activity that is a common feature of laminin ␣ chains (22) (data not shown).

Mapping of the Lu/B-CAM Binding Site on the Laminin ␣5
LG Domain-To examine the binding of Lu to the recombinant laminins, we prepared a dimerized soluble recombinant protein (Lu-Fc) that is composed of the Lu extracellular domain and human IgG1 Fc portion. Lu-Fc dimerized via the Fc region was expected to bind to laminin ␣5 more effectively than the monomeric Lu (Sol-Lu) used in our previous study (17). By SDS-PAGE, purified Lu-Fc migrated at a relative molecular mass of 120 kDa under reducing conditions (Fig. 3A) and at 240 kDa under non-reducing conditions (data not shown). The purified protein reacted with anti-human Lu monoclonal antibodies BRIC 108 and BRIC 221 (data not shown).
Solid phase binding assays were performed using the purified recombinant laminins and Lu-Fc. Consistent with our previous study, Lu-Fc bound to recombinant laminin trimers containing Mr5G3 but not Mr5G2 (Fig. 3B). The straightforward interpretation is that Lu/B-CAM binds to ␣5LG3. However, the other recombinant laminins containing ␣5LG3 (Mr5G13 and Mr5G23) did not bind Lu-Fc. These results indicate that Lu/B-   CAM binding requires not only ␣5LG3 but also ␣5LG1-2. We conclude that the entire ␣5LG1-3 tandem structure is essential for Lu/B-CAM binding to laminin ␣5.
To examine the contribution of the LCC domain to Lu/B-CAM binding, we produced a recombinant ␣5LG1-3 tandem lacking the LCC domain in HEK293 cells. As shown in Fig. 4A, the purified protein migrated at 80 kDa as a single band under non-reducing conditions. The recombinant ␣5LG1-3 tandem exhibited weak Lu-Fc binding activity at high concentrations (Fig. 4B). Therefore, Lu/B-CAM binding activity likely requires a proper ␣5LG1-3 conformation in connection with the preceding LCC domain.
That both the integrins and Lu/B-CAM bind to ␣5LG1-3 suggests that the binding of Lu/B-CAM to ␣5LG1-3 might compete with binding by ␣3␤1/␣6␤1 integrins. To test this possibility, cell adhesion inhibition assays in the presence of Lu-Fc were performed using human fibrosarcoma HT1080 cells stimulated with activating anti-integrin ␤1 monoclonal antibody TS2/16 (Fig. 6). The adhesion of unstimulated and TS2/16stimulated HT1080 cells to laminin ␣5 was inhibited by a combination of anti-integrin ␣3 and ␣6 antibodies, in agreement with our previous study (13). Lu-Fc could partially inhibit the adhesion of unstimulated HT1080 cells to laminin ␣5, but the adhesion of TS2/16-stimulated cells was not inhibited any longer by Lu-Fc. The inhibitory effects of Lu-Fc were dependent on the activation state of integrins, suggesting that the binding of Lu/B-CAM on laminin ␣5 competed with that of ␣3␤1/␣6␤1 integrins. Furthermore, to determine whether the binding of Lu/B-CAM to laminin ␣5 occupied or obscured the binding site for ␣3␤1/␣6␤1 integrins, we performed adhesion assays using K562 cells transfected with integrin ␣3 or ␣6 in the presence of Lu-Fc. Lu-Fc could efficiently inhibit the adhesion of the transfectants to laminin ␣5 (Fig. 7). Therefore, we conclude that the binding site for Lu/B-CAM on laminin ␣5 is close to that for ␣3␤1/␣6␤1 integrins.  . Adhesion of K562 transfectants expressing integrin ␣3␤1 or ␣6␤1 to recombinant laminins. K562 non-transfectants (K562) and transfectants expressing ␣3␤1 (␣3A/K562) or ␣6␤1 (␣6A/K562) were preincubated with or without the activating anti-integrin ␤1 antibody TS2/16. The unstimulated (white columns) or stimulated (gray columns) cells were added to 96-well microtiter plates coated with 10 g/ml recombinant laminins and incubated at 37°C for 1 h. Adherent cells were stained and counted as described under "Experimental Procedures." Each column represents the mean Ϯ S.D. of triplicate assays.

DISCUSSION
In our previous studies, we generated transgenic mice expressing modified laminin ␣5 chains and investigated the roles of the ␣5LG domain in various developmental and physiological processes (5,10). This transgenic mouse approach also allowed us to discover the region of laminin ␣5 that binds Lu/B-CAM (17). In this study, we undertook an in vitro approach to accelerate the identification of the Lu/B-CAM binding site on the ␣5LG domain. We first produced recombinant mouse laminin-511 in HEK293 cells. The purified recombinant laminin-511 was apparently assembled and folded properly and displayed cell adhesion activity mediated through ␣3␤1/␣6␤1 integrins and Lu/B-CAM in a manner similar to that of intact human laminin-511. We also produced a series of recombinant laminins containing chimeric ␣ chains to identify the location of the binding site for Lu/B-CAM. Because TS2/16-stimulated HT1080 cells adhered to recombinant laminin containing Mr51 through integrin ␣6␤1 as well as they adhered to laminin-111 (see supplemental figure), the human ␣1 LG domain seems to be appropriately functional. All chimeric ␣ chains also contained the ␣1LG4-5 tandem, a major heparin binding region within ␣1LG (22). Because the purified proteins exhibited hep-arin binding activity, we presume that they form trimers and fold properly.
A recombinant human Lu extracellular domain fused with the Fc portion of human IgG1 (Lu-Fc) was produced in mammalian cells, as described in other studies (15,38). The recombinant proteins dimerized at the Fc region. As expected, dimerized Lu-Fc bound to laminin ␣5 more effectively than did monomeric Lu (Sol-Lu) (17). Solid phase binding assays showed that recombinant laminin containing Mr5G3 but not Mr5G2 bound to Lu-Fc. In agreement with our previous study, the ␣5LG3 domain is definitely required for Lu/B-CAM binding. However, the other recombinant laminins containing ␣5LG3 (Mr5G13 and Mr5G23) were not bound by Lu-Fc, suggesting that the ␣5LG3 domain is necessary but not sufficient to impart binding activity. These results allowed us to conclude that the entire structure of the ␣5LG1-3 tandem is essential for Lu/B-CAM binding to laminin ␣5.
The LG1-3 tandem of laminin ␣ chains is predicted to form the shape of a cloverleaf in contact with the LCC domain (22). The binding site of Lu/B-CAM could be formed at the tertiary structure level. Moreover, because the recombinant isolated ␣5LG1-3 tandem did not bind to Lu-Fc, the LCC domain may be required for the proper folding of ␣5LG1-3 within the LG domain. In addition, the Lu-Fc binding activity of Mr5G3 was lower than that of the full-length ␣5 chain, suggesting that the ␣5LG4-5 tandem (replaced by ␣1LG4 -5 in Mr5G3) is involved in Lu/B-CAM binding. Accumulating evidence indicates that ␣LG4-5 tandems are involved in laminin binding to ␣-dystroglycan and syndecan (22,23). It is unlikely that the manner of Lu/B-CAM binding is similar to that of ␣-dystroglycan and syndecans. Because structural analysis predicts that LG4 -5 is spatially located over the cloverleaf of LG1-3 (22,33), the possibility exists that ␣5LG4 -5 regulates the binding affinity of Lu/B-CAM to ␣5LG1-3.
Integrin-mediated cell adhesion is one of the hallmarks of the biological functions of laminin LG domains. Many studies that have functionally dissected LG domains have addressed the regions of various laminin isoforms that are responsible for integrin-mediated cell adhesion (22,24,25). Recently, Ido et al. (35) reported that the entire ␣5LG1-3 tandem was essential for the binding of integrin ␣6␤1. We also found that the binding sites for ␣3␤1/␣6␤1 integrins were within the ␣5LG1-3 tandem. In a recent study, we have reported that soluble recombinant Lu inhibits the adhesion of human corneal epithelial cells  to laminin-511 as well as function blocking antibodies to ␣3 and ␤1, suggesting that the binding sites for Lu and integrin ␣3␤1 overlap (39). Here, Lu-Fc inhibited ␣3␤1/␣6␤1 integrin-mediated cell adhesion to laminin ␣5. The most straightforward interpretation is that the binding site of Lu/B-CAM on ␣5LG1-3 overlaps with that of ␣3␤1/␣6␤1 integrins. Alternative explanations that cannot be ruled out are that 1) Lu/B-CAM binding causes a change in the structure of ␣5LG1-3 that prevents integrin binding or 2) steric hindrance prevents binding of both Lu/B-CAM and the integrin. We have also reported that Lu/B-CAM mediates the adhesion of human mesangial and endothelial cells to laminin ␣5 in collaboration with integrins containing the ␤1 subunit (5,40). As Lu-Fc did not inhibit the adhesion of HT1080 cells activated with anti-integrin ␤1 monoclonal antibody, the activation status of integrins on cells may affect the binding of Lu/B-CAM to ␣5LG1-3.
Distribution studies in various tissues have shown that Lu/B-CAM and ␣3␤1/␣6␤1 integrins are localized at the basal surface of cells, facing a basement membrane containing laminin ␣5 (20,(41)(42)(43). We have also observed that Lu/B-CAM is colocalized with ␣3␤1/␣6␤1 integrins in some mouse tissues (21). 3 Therefore, it is likely that Lu/B-CAM competitively interacts with ␣3␤1/␣6␤1 integrins for binding to the ␣5LG1-3 domain in vivo. Lu (but not B-CAM) contains the 40 amino acids of the cytoplasmic domain that carries the Src homology 3 binding motif and potential phosphorylation sites that could be involved in intracellular signaling pathways (20). Hines et al. (44) show that the physiologic stress mediator epinephrine, acting through the ␤2-adrenergic receptor, increased the adhesion of sickle red blood cells to laminin ␣5 via inside-out signals. Recently, it has been reported that cAMP signaling can promote adhesion to laminin ␣5 via Lu through two distinctive signaling pathways (45,46). The activated binding of Lu to laminin ␣5 may modulate intracellular signaling from the ␣5 chain via ␣3␤1/␣6␤1 integrins.
This study provides new insights into both integrin-and non-integrin-mediated cell adhesion to basement membranes containing laminin ␣5. Although we have narrowed the binding site of Lu/B-CAM on the laminin ␣5LG domain to the LG1-3 tandem, in the future it will be important to analyze the binding site at the level of tertiary structure. Also, Lu-Fc may serve as an important tool to identify integrin binding sites on the ␣5LG1-3 tandem.