Identification of the binding site for the Lutheran blood group glycoprotein on laminin alpha 5 through expression of chimeric laminin chains in vivo.

The Lutheran blood group glycoprotein (Lu), also known as basal cell adhesion molecule, is an Ig superfamily transmembrane receptor for laminin alpha5. Lu is expressed on the surface of a subset of muscle and epithelial cells in diverse tissues and is thought to be involved in both normal and disease processes, including sickle cell disease and cancer. Here we investigated the binding of Lu to laminin alpha5 in vivo and in vitro. We prepared a soluble recombinant Lu (sol-Lu) composed of the Lu extracellular domain and a His(6) tag. Sol-Lu bound specifically to laminin-10/11 (alpha5beta1/beta2gamma1) in enzyme-linked immunosorbent assays and bound to bona fide basement membranes containing laminin alpha5 in tissue sections. Sol-Lu did not bind to tissue sections of laminin alpha5 knockout embryos, despite the fact that the four other alpha chains were present. To identify the Lu-binding site on laminin alpha5, we prepared modified alpha5 cDNAs encoding chimeric laminins containing all or part of the laminin alpha1 G domain in place of the analogous alpha5 regions. These constructs were used to generate transgenic mice. Proteins derived from transgenes were detected in basement membranes and were assayed for their ability to bind Lu by examining the localization of endogenous Lu and the binding of sol-Lu applied to tissue sections. Our results demonstrate that the alpha5 LG3 module is essential for Lu binding to laminin alpha5.

The Lutheran blood group glycoprotein (Lu), also known as basal cell adhesion molecule, is an Ig superfamily transmembrane receptor for laminin ␣5. Lu is expressed on the surface of a subset of muscle and epithelial cells in diverse tissues and is thought to be involved in both normal and disease processes, including sickle cell disease and cancer. Here we investigated the binding of Lu to laminin ␣5 in vivo and in vitro. We prepared a soluble recombinant Lu (sol-Lu) composed of the Lu extracellular domain and a His 6 tag. Sol-Lu bound specifically to laminin-10/11 (␣5␤1/␤2␥1) in enzyme-linked immunosorbent assays and bound to bona fide basement membranes containing laminin ␣5 in tissue sections. Sol-Lu did not bind to tissue sections of laminin ␣5 knockout embryos, despite the fact that the four other ␣ chains were present. To identify the Lubinding site on laminin ␣5, we prepared modified ␣5 cDNAs encoding chimeric laminins containing all or part of the laminin ␣1 G domain in place of the analogous ␣5 regions. These constructs were used to generate transgenic mice. Proteins derived from transgenes were detected in basement membranes and were assayed for their ability to bind Lu by examining the localization of endogenous Lu and the binding of sol-Lu applied to tissue sections. Our results demonstrate that the ␣5 LG3 module is essential for Lu binding to laminin ␣5.
Laminins are a family of extracellular matrix proteins that are located primarily in basement membranes. They regulate various cellular functions such as adhesion, motility, growth, differentiation, and apoptosis through interaction with specific cell surface receptors (1,2). The three subunits of laminins, designated ␣, ␤, and ␥ chains, assemble to form what is typically a cross-shaped structure. Five ␣, four ␤, and three ␥ chains have been identified. To date, 15 different laminin heterotrimers have been found to be synthesized and secreted by cells (3)(4)(5)(6)(7), although many more combinations are theoretically possible. Of the three laminin chain types, only the ␣ chain has a large carboxyl-terminal globular (G) 1 domain consisting of a tandem array of five laminin-type G (LG) modules (LG1 through LG5) (8). These LG modules contain binding sites for ␤ 1 integrins and heparin, as well as ␣-dystroglycan in some isoforms (9).
The laminin ␣5 chain is a component of the laminin-10 (␣5␤1␥1) and laminin-11 (␣5␤2␥1) heterotrimers and is widely expressed (4, 10 -12). We have shown that mice lacking laminin ␣5 die during late embryogenesis with several developmental defects, including defects in neural tube closure, digit separation, placentation, and kidney and lung development (13)(14)(15). Laminin-10/11 is bound by several different receptors, including integrin ␣ 3 ␤ 1 , ␣ 6 ␤ 1 , and ␣ 6 ␤ 4 (16,17) and dystroglycan (18). Another potential non-integrin receptor for laminin ␣5 is the Lutheran blood group glycoprotein (Lu), which is a member of the Ig superfamily. A splice variant of Lu is known as basal cell adhesion molecule (B-CAM) (19,20). Lu/B-CAM has been studied primarily in the contexts of blood group antigens, sickle cell disease, and cancer (20 -26). Lee et al. (26) proposed that sickle red blood cells adhere to endothelial basement membranes by binding to laminin ␣5. They showed that sickle cells bind to laminin preparations containing the ␣5 chain, and an antibody to laminin ␣5 inhibits binding (26). Udai et al. (25) demonstrated that the major laminin receptor present on sickle cells is the Lu/B-CAM protein. Furthermore, K562 cells transfected with human Lu adhere to laminin-10/11 but not to laminins lacking the ␣5 chain (27). In our previous studies we made an antibody specific for mouse Lu, and we determined its expression pattern (28). Lu is expressed on the surface of a subset of muscle and epithelial cells in diverse tissues. In epithelial cells, Lu is concentrated on the basal surface adjacent to basement membranes containing laminin ␣5. In Lama5 Ϫ/Ϫ tissues, Lu is no longer localized to the basal surface, suggesting that Lu binds directly to ␣5. In transgenic mouse hearts that overexpress laminin ␣5, Lu levels are elevated, suggesting that the increased ␣5 in cardiomyocyte basement membranes recruits additional Lu to the cell surface through a direct interaction. Although the laminin-binding site on Lu has been mapped to a first approximation (20,27,29), there is little insight into the structural basis for Ig superfamily members binding to laminins. To understand better the interaction between Lu and the laminin ␣5 chain, it is important to determine the site of Lu binding on ␣5.
In this study we prepared a soluble recombinant protein containing the Lu extracellular domain (sol-Lu). Sol-Lu bound to laminin-10/11 in enzyme-linked immunosorbent assays and specifically recognized the laminin ␣5 chain on tissue sections. To identify the binding site for Lu on the laminin ␣5 chain in vivo, we produced transgenic mice expressing modified laminin ␣5 chains with LG module substitutions derived from laminin ␣1. These chimeric ␣ chains incorporated into basement membranes; sol-Lu was then used in tissue binding assays to narrow the Lu-binding site on ␣5 to the LG3 module.
Preparation of Sol-Lu-A cDNA expression plasmid containing the full-length human Lu coding region, a V5 tag, and a His 6 tag was purchased from Invitrogen. To remove sequences encoding the V5 tag and the transmembrane and intracellular domains, nucleotides 904 -1668 (GenBank TM accession number X83425) were amplified by PCR with Vent polymerase (New England Biolabs, Beverly, MA) following the manufacturer's instructions and using the primer combination: sense, 5Ј-GGCAGCCCCAGCCCGGAGTAT-3Ј; antisense, 5Ј-GGAAT-TCACCGGTCACTCCAGCCTGGGAGGTCTG-3Ј. The amplified fragment was digested and then ligated into the XhoI and AgeI sites of the expression vector. The resulting expression vector containing the Lu extracellular domain and a His 6 tag was transfected into COS-7 cells (American Type Culture Collection, Manassas, VA) using Lipo-fectAMINE (Invitrogen). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Invitrogen). The recombinant protein was purified from serum-free culture medium by nickel column chromatography. The eluted fractions were pooled and dialyzed against Ca 2ϩ and Mg 2ϩ -free phosphate-buffered saline (PBS(Ϫ)). The purity of recombinant protein was defined by SDS-PAGE ( Fig. 1).
In Vitro Binding Assays-Binding assays were carried out with various concentrations (0 -80 g) of laminin-10/11 and laminin-1 coated onto the plastic surface of microtiter plates. Plates were blocked with 1% BSA in PBS(Ϫ) and incubated with sol-Lu at 37°C for 1 h. After washing with PBS(Ϫ), the bound sol-Lu was detected with a human Lu-specific monoclonal antibody, BRIC108. After further washing, the bound antibodies were detected by addition of horseradish peroxidaseconjugated anti-mouse IgG1 (Roche Diagnostics), followed by addition of 1 mg/ml o-phenylenediamine and 0.001% H 2 O 2 . The absorbance was measured at 492 nm by VERSAmax (Molecular Devices, Sunnyvale, CA).
Preparation of Chimeric Laminin Constructs-cDNA clones encoding full-length mouse laminin ␣5 (generated in our laboratory) and human laminin ␣1 (provided by Dr. Karl Tryggvason, Stockholm, Sweden) chains were used to construct expression vectors encoding the chimeric laminin ␣ chains Mr51, Mr5G2, and Mr5G3. PCR was used to introduce restriction sites at appropriate locations and to seamlessly join amplified fragments with overlapping sequences by sequential PCR (36). For all PCR, Vent polymerase (New England Biolabs) was used according to the manufacturer's instructions. To construct Mr51, a BsiWI site was first engineered at the junction between ␣5 domain I/II and G. The G domain of human ␣1 was amplified with primers containing added BsiWI sites as follows: sense, 5Ј-CGGGATCCCGTACGCAAGCAGCTT-CTATTAAAGTCGCCG-3Ј, and antisense, 5Ј-GCTCTAGACGTACGGG-CGCGCCTCAGGACTCGGTCCCAGGAC-3Ј. This product was ligated to the cDNA encoding ␣5 domains VI through I/II. For generating Mr5G2 and Mr5G3, we took advantage of a unique AgeI site at the end of ␣5G1. To construct Mr5G2, ␣5LG2 was amplified with sense, 5Ј-A-AGCGCGCCTCTAGAGGGCGTTCAGGGGTACGACTG-3Ј, and antisense, 5Ј-GAAGCTAACACTTCCCACTAGCAGGTCAGCGGT-3Ј; ␣1LG3-5 was amplified with sense, 5Ј-CTGCTAGTGGGAAGTGTTAG-CTTCCTGAAAGGC-3Ј, and antisense, 5Ј-AGGCGCGCCCGTACGTCA-GGACTCGGTCCCAGGAC-3Ј. These two products were mixed and subjected to PCR again for 20 cycles to join them. To construct Mr5G3, ␣5LG2-3 was amplified with sense, same primer as for ␣5LG2, and antisense, 5Ј-CCGGGGCTCTCTGGCTGGTGTACAGCCTACGCT-3Ј; ␣1LG4-5 was amplified with sense, 5Ј-TGTACACCAGCCAGAGAGCC-CCGGGCTTTTCCA-3Ј, and antisense, same as ␣1LG3-5. These two products were mixed and subjected to PCR again for 20 cycles to join them. The segments encoding chimeric LG2-5 modules were ligated to the cDNA encoding ␣5 domains VI through LG1 to generate full-length chimeric cDNAs. These were then cloned into a modified widely active expression vector miw (30), which contains a fusion of the chicken ␤-actin promoter and the Rous sarcoma virus-long terminal repeat.
Generation of Knockout and Transgenic Mice-Production of Lama5 mutant mice and transgenic mice overexpressing full-length laminin ␣5 has been described (13,28). Transgenic mice expressing chimeric laminins were produced by the Mouse Genetics Core facility at Washington University School of Medicine by standard microinjection of DNA into pronuclei of (B6XCBA)F2 single-celled embryos. The desired transgenes were separated from plasmid vector sequences by digestion with NotI and agarose gel electrophoresis.
Immunohistochemistry-Mouse embryos from timed matings were frozen whole by immersing in OCT compound and quick-freezing in 2-methylbutane cooled in a dry ice/ethanol bath. Sections were cut at 7 m in a cryostat and air-dried. For staining, sections were blocked in 10% normal goat serum and then incubated with primary antibody. All antibody incubations were in PBS containing 1% BSA, and all washes were in PBS. Secondary antibodies were conjugated to fluorescein isothiocyanate (ICN, Costa Mesa, CA) or Cy3 (Chemicon, Temecula, CA). After several washes, sections were mounted in 90% glycerol containing 0.1ϫ PBS and 1 mg/ml p-phenylenediamine. Sections were examined through a Nikon Eclipse E800 microscope. Images were captured with a Spot 2 cooled color digital camera (Diagnostic Instruments, Sterling Heights, MI) using Spot Software version 2.1. Images were imported into Adobe Photoshop 5.0 and Adobe Illustrator 9.0 for processing and layout.
Sol-Lu Binding Assay on Tissue Sections-Sol-Lu was adjusted to 10 g/ml with 1% BSA/PBS(Ϫ). Sections were blocked in 10% normal goat serum and incubated with diluted sol-Lu. Bound sol-Lu was detected with a monoclonal antibody against human Lu (BRIC108) and methods described above.

Production of Recombinant Sol-Lu Protein and Its Binding to
Laminin-10/11-To examine the binding of Lu to laminin ␣5, we prepared a soluble recombinant protein that is composed of the Lu extracellular domain and a COOH-terminal His 6 tag. As shown in Fig. 1A, the purified recombinant protein (sol-Lu) migrated as a single band in SDS-PAGE. The identity of the purified protein was further defined by in enzyme-linked immunosorbent assay using a monoclonal antibody against human Lu (data not shown).
To test if sol-Lu binds to laminin-10/11 (␣5␤1/␤2␥1), we performed solid phase binding assays. Bound sol-Lu was detected by monoclonal antibody against human Lu. The binding of Lu to laminin-10/11 was observed at Ͼ5 g/ml of coating concentration (Fig. 1B). On the other hand, sol-Lu did not bind to laminin-1 (␣1␤1␥1). This specificity for laminins containing the ␣5 chain is consistent with published results (20,27). We therefore conclude that sol-Lu has binding properties similar to native cell surface Lu and is an appropriate tool to investigate and identify Lu-binding sites on the laminin ␣5 chain.
We also used sol-Lu to further characterize the nature of the interaction between Lu and laminin ␣5. Divalent cations are required for the binding of other laminin receptors such as integrins and dystroglycan (37, 38), and the binding of dystroglycan is affected by glycosaminoglycans (18,37). In contrast, the binding of sol-Lu to laminin ␣5 was not inhibited by EDTA or by heparin (Fig. 1C). However, high salt did inhibit the interaction (Fig. 1C), as is typical for biologically relevant protein-protein interactions.
Binding Specificity of Sol-Lu-There is a possibility that Lu also binds to other laminins or to unknown ligands. To test the specificity of Lu binding, we performed histochemistry using sol-Lu as a probe on tissue sections. Bound sol-Lu was detected by monoclonal antibody against human Lu. Sol-Lu bound to basement membranes containing the laminin ␣5 chain in tissue sections of an embryonic day (E) 13.5 mouse embryo (Fig. 2,  A and C). The pattern of sol-Lu binding was identical to the expression of laminin ␣5 chain in tissues such as lung, intestine, kidney, and pharynx (data not shown). There was no binding of sol-Lu to tissue sections from Lama5 Ϫ/Ϫ embryos (Fig. 2, B and D). However, laminins ␣1, ␣2, ␣3, and ␣4 were detected in Lama5 Ϫ/Ϫ basement membranes (Fig. 2, E-H). Together, these results suggest that Lu is a specific receptor for the laminin ␣5 chain and does not bind to other ␣ chains or to other basement membrane components.
Binding of Sol-Lu to the Laminin ␣5 Chain G Domain, in Vitro Assays-Although the laminin-binding site on Lu has been mapped to the first three of the five extracellular Ig domains, the structural basis for laminin ␣5 binding to Lu is unknown. To approach identification of the Lu-binding site, we prepared a chimeric construct encoding laminin ␣5 domains VI through I/II linked to the human laminin ␣1 G domain, designated Mr51 (Fig. 3). The construct encoding full-length laminin ␣5, Mr5, was also prepared as a positive control. To force expression of these transgenes in a variety of cell types, we used the miw expression vector, which directs widespread expression in transgenic mice (39). 2 The constructs were microinjected to generate transgenic mice. We obtained two independent lines of mice that expressed the full-length laminin ␣5 protein. Transgene-derived protein presumably trimerizes with ␤ and ␥ chains and assembles into basement membranes. During embryogenesis, transgene-derived laminin ␣5 levels were significantly increased in heart and skeletal muscle (28). Five founder mice harboring the Mr51 transgene were also generated. Transgene-positive offspring of the five founders were tested for expression using the anti-human laminin ␣1G domain monoclonal antibody, as well as our polyclonal antiserum to mouse laminin ␣5, domains IIIb and IVa. E13.5 embryos from all five lines expressed Mr51 protein in a similar fashion; high levels of chimeric protein were present in heart, and moderate levels were present in lung, kidney, skeletal muscle, airway epithelial, and brain pial basement membranes. 3 Because the expression of endogenous laminin ␣5 was very low in embryonic heart (Fig. 4A), and Mr5 and Mr51 were strongly expressed in embryonic heart (Fig. 4, C and E), we chose heart for sol-Lu binding assays. Sol-Lu bound to heart expressing Mr5 (Fig. 4D) but not to the control heart or to heart expressing Mr51 (Fig. 4, B and F). This indicates that the G domain of laminin ␣5 is required for sol-Lu binding, and we concluded that the G domain contains the Lu-binding site.
Binding of Lu to the Laminin ␣5 Chain G Domain, in Vivo Assay-Next transgenic mice expressing Mr5 or Mr51 on the Lama5 Ϫ/Ϫ genetic background were generated. Mouse genotypes were determined by PCR using appropriate primers. Mr5 was able to rescue all Lama5 Ϫ/Ϫ embryonic defects, but Mr51 could not. 3 Mr5-and Mr51-derived proteins assembled into basement membranes (Fig. 5, A and B). As before, Mr51 was detected with a monoclonal antibody against human ␣1LG4-5, 163DE4 (Fig. 5D), which did not cross-react with Mr5 (Fig. 5C). An antibody against the intracellular domain of Lu demonstrated that Lu was basally concentrated on epithelial cells in Lama5 Ϫ/Ϫ, Mr5 tissue (Fig. 5E), just as it is in wild-type (28). On the other hand, Lu was diffuse in Lama5 Ϫ/Ϫ, Mr51 tissue (Fig. 5F). These results show that Lu interacts with the laminin ␣5 G domain in vivo, because the G domain of ␣5, but not the G domain of ␣1, was able to polarize Lu.
Lack of Sol-Lu Binding to Endogenous Laminin ␣5 Expressed in Embryonic Skeletal Muscle-The basement membrane of embryonic skeletal muscle, both extrasynaptic and synaptic, is rich in the laminin ␣5 chain (40). Here we found that an antibody against laminin ␣5 LG4-5 stained the basement membranes of most E17.5 embryonic tissues but did not stain skeletal muscle (Fig. 6, A-D, and data not shown). This suggested that the COOH terminus of endogenous laminin ␣5 is either cleaved by protease or masked in embryonic skeletal muscle. The sol-Lu binding assay was performed on sections containing E17.5 embryonic lung and skeletal muscle. The sol-Lu bound to laminin ␣5 expressed in lung but not in skeletal muscle (Fig. 6, E and F). These results suggest that the Lu-binding site of laminin ␣5 is cleaved by protease in skeletal muscle and that the binding site may be present in LG4-5. However, the lack of reactivity with the anti-LG4-5 antiserum does not reveal where the putative cleavage event occurs, only that it is NH 2 -terminal to the most distal epitope recognized by the antiserum.
Binding of Sol-Lu to ␣5LG Modules-To attempt to more definitively narrow the binding site of Lu in laminin ␣5, we prepared two new chimeric laminin cDNAs encoding laminin ␣5 domains VI through either LG2 or LG3, linked to the human laminin ␣1 LG3-5 and LG4-5 domains, respectively. These were cloned into the miw expression vector and designated Mr5G2 and Mr5G3 (Fig. 7). The constructs were microinjected to produce transgenic mice. We obtained one founder for each construct that expressed the transgene. Tissues taken from 1-2-week-old mice were analyzed. The antibody against laminin ␣5 domain IIIb/IVa stained the basement membranes of skeletal muscle in all cases (Fig. 8, A, E, and I). As described The laminin ␣5 chain expressed in embryonic skeletal muscle was not reactive with antiserum to ␣5LG4-5 (D). E and F, when sol-Lu was applied to sections, it bound to lung (E) but not to skeletal muscle (F). These results suggest that laminin ␣5 in embryonic skeletal muscle lacks ␣5LG4-5 and the Lu-binding site. Bar, 100 m. above, the endogenous laminin ␣5 chain in skeletal muscle lacked immunoreactivity with anti-␣5LG4-5, as did the transgene-derived proteins (Fig. 8, B, F, and J). However, Mr5G2 and Mr5G3 could be detected with the monoclonal antibody against ␣1LG4-5, 163DE4 (Fig. 8, G and K), indicating that the chimeric proteins are expressed and incorporated into basement membranes. Sol-Lu bound to Mr5G3 (Fig. 8L) but not to Mr5G2 (Fig. 8H). This demonstrates that ␣5LG3 is crucial for Lu binding. DISCUSSION Lu is a member of the immunoglobulin superfamily and has five extracellular Ig-like domains, a transmembrane domain, and a cytoplasmic COOH-terminal domain of 40 amino acids (23). The cytoplasmic tail is absent in the human Lu isoform, B-CAM (21). In previous in vitro studies, the interaction between Lu and laminin-10/11 containing the ␣5 chain has been well demonstrated (19,20,25,27,41). The extracellular domain of Lu contains N-glycosylation consensus motifs. Recently it was reported that Ig-like domains 1-3 are involved in binding to laminin ␣5 (27,29), but the N-glycosylation motifs were not found to be involved in laminin binding. A bacterial Lu recombinant protein that we initially prepared did not bind to laminin-10/11 (data not shown), suggesting that proper glycosylation is required to ensure proper folding.
We also produced a recombinant human Lu extracellular domain (sol-Lu) in mammalian cells. Sol-Lu migrated at a higher molecular weight than predicted from the deduced amino acid sequence, suggesting that it was glycosylated. It reacted with a monoclonal antibody against human Lu and bound to laminin-10/11 but not to laminin-1 (Fig. 1). We therefore concluded that sol-Lu has binding properties similar to native cell surface Lu and is an appropriate tool to investigate the Lu-binding site on laminin ␣5. It is interesting that the binding of sol-Lu to laminin-10/11 was not inhibited by EDTA and heparin. Divalent cations are required for the binding of other laminin receptors such as integrins and dystroglycan (37, 38). The binding of dystroglycan is also affected by glycosaminoglycans (18,37). Lu therefore has laminin-binding properties significantly different from integrins and dystroglycan.
Our previous study demonstrated that Lu is localized on the basal surface of many epithelial cells and on the surface of a subset of muscle cells, in all cases adjacent to basement membranes containing laminin ␣5 (28). The localization of Lu suggested that Lu interacts with the laminin ␣5 chain in vivo. In the present study we examined whether there are other ligands for Lu. Soluble receptor binding assays on tissues is a proven method to reveal the presence of unknown ligands (42). Sol-Lu made it possible for us to perform binding assays on tissue sections. When applied to wild-type tissue sections, sol-Lu bound to basement membranes containing laminin ␣5. The binding of sol-Lu totally disappeared from laminin ␣5 knockout tissues, further demonstrating its specificity. The five laminin ␣ chains share similarities in sequence and domain structure (8,10). Although laminin ␣1Ϫ4 chains were detected in basement membranes in Lama5 Ϫ/Ϫ tissue, sol-Lu did not bind to them. This suggests that only laminin ␣5 bears the specific sequence or structure for Lu binding. Thus, whereas integrin ␣ 3 ␤ 1 , ␣ 6 ␤ 1 , ␣ 6 ␤ 4 , and dystroglycan are promiscuous laminin receptors (2), Lu is a specific receptor for laminin ␣5, at least during embryogenesis. Conditional laminin ␣5 knockout mice being generated in our laboratory will allow us to search for novel Lu ligands in adult tissues.
Until now, there has been no data addressing the structural basis of laminin ␣5 binding to Lu. We generated transgenic mice expressing chimeric laminin ␣5/␣1 chains that incorporated into basement membranes. This allowed us to search for the Lu-binding site in the context of a bona fide basement membrane. The ␣5LG3 module was identified as being required for Lu binding, and we suggest that it contains the binding site. However it is possible that Lu binding requires not only ␣5LG3 but also ␣5LG1-2.
LG modules are also important ligands for other cellular receptors, such as integrins and dystroglycan (9). Together with these and other laminin receptors, Lu could regulate growth, adhesion, and differentiation of epithelial cells. LG4-5 (C, G, and K); and sol-Lu (D, H, and L). Sol-Lu bound to Mr5G3 (L) but not to Mr5G2 (H), indicating that ␣5LG3 is critical for Lu binding. The endogenous laminin ␣5 chain in postnatal muscle lacked reactivity with the antiserum against ␣5LG4-5 (B) and lacked sol-Lu binding activity (C), similar to embryonic skeletal muscle (Fig. 6). Bar, 50 m.
Interestingly, we found that the endogenous laminin ␣5 chain expressed in embryonic skeletal muscle lacks reactivity with an ␣5LG4-5 antiserum and does not bind sol-Lu. The straightforward interpretation is that Lu binds to ␣5LG4-5. However, because sol-Lu binds to Mr5G3, which lacks ␣5LG4-5 but contains ␣1LG4-5 instead, the Lu-binding site must be in ␣5LG1-3. In skeletal muscle, it is possible that ␣5LG4-5 is released by protease, and this somehow affects the binding affinity of Lu for ␣5LG1-3. It has been shown that the G domains of laminin ␣2, ␣3, and ␣4 chains are cleaved by proteolytic processing (9,32,43). The link region between ␣5LG3 and LG4 modules contains an RRXR sequence, which is a furin-type cleavage site (8,10). This site is also conserved in the link region of human laminin ␣5 (44). Proteolytic processing appears to have occurred in this link region in skeletal muscle and perhaps affected Lu binding. Alternatively, protease cleavage sites may lie further upstream, resulting in removal of the critical LG3 module.
Adhesion of sickled red blood cells to laminin ␣5 via Lu is suspected to contribute to the painful vaso-occlusion episodes that occur in sickle cell patients (25,26). In addition, Lu/B-CAM overexpressed in some epithelial cancers may promote tumor metastasis (21,22). In the future, it is important to define at the amino acid level the critical structural determinants of laminin ␣5 and Lu that mediate their binding to each other. Toward this end, novel chimeric laminins are being generated in our laboratory. With additional details about the Lu-binding site, it may be possible to develop inhibitors that block the binding of Lu to laminin ␣5 in vivo. Such inhibitors may suppress vaso-occlusion in sickle cell disease and tumor cell metastasis in cancer.