Laminin gamma3 chain binds to nidogen and is located in murine basement membranes.

Recently a novel laminin gamma3 chain was identified in mouse and human and shown to have the same modular structure as the laminin gamma1 chain. We expressed two fragments of the gamma3 chain in mammalian cells recombinantly. The first, domain VI/V, consisting of laminin N-terminal (domain VI) and four laminin-type epidermal growth factor-like (domain V) and laminin N-terminal modules, was shown to be essential for self-assembly of laminins. The other was domain III3-5, which consists of three laminin-type epidermal growth factor-like modules and is predicted to bind to nidogens. The gamma3 VI/V fragment was a poor inhibitor for laminin-1 polymerization as was the beta2 VI/V fragment. The gamma3 III3-5 fragment bound to nidogen-1 and nidogen-2 with lower affinity than the gamma1 III3-5 fragment. These data suggested that laminins containing the gamma3 chain may assemble networks independent of other laminins. Polyclonal antibodies raised against gamma3 VI/V and gamma3 III3-5 showed no cross-reaction with homologous fragments from the gamma1 and gamma2 chains of laminin and allowed the establishment of gamma chain-specific radioimmunoassays and light and electron microscopic immunostaining of tissues. This demonstrated a 20-100-fold lower content of the gamma3 chain compared with the gamma1 chain in various tissue extracts of adult mice. The expression of gamma3 chain was highly tissue-specific. In contrast to earlier assumptions, the antibodies against the gamma3 chain showed light microscopic staining exclusively in basement membrane zones of adult and embryonic tissues, such as the brain, kidney, skin, muscle, and testis. Ultrastructural immunogold staining localized the gamma3 chain to basement membranes of these tissues.

Basement membranes are specialized structures of the extracellular matrix with multiple functions (1). As thin condensed matrices, they divide the cells of the parenchymal tissues from the interstitial matrix. Their main components are collagen type IV variants, laminins, nidogens, and perlecan (2). Laminins constitute a family of heterotrimeric proteins (␣␤␥) which are mainly localized in basement membranes and are involved in cell matrix and various other protein interactions. Eleven different chains (␣1-␣5, ␤1-␤3, ␥1-␥3) have been identified and sequenced, and they assemble fifteen different iso-forms, laminin-1-15 (3,4,5). These chains share a 600-residue domain II/I that oligomerizes into a rod-like coiled-coil structure forming the long arm of laminins. The N-terminal short arms consist of rod-like elements on the basis of tandem arrays of laminin-type epidermal growth factor-like (LE) modules (6) and several globular domains, referred to as laminin N-terminal (LN), 1 L4 modules, and domains IVb (␣3B and ␣5 chains)/IV (␤1 and ␤2 chains), which have not been classified to date. All ␣ chains share a unique C-terminal G domain, which consists of five laminin G modules (7). Most of these modules are also shared by several other extracellular proteins, such as the proteoglycans perlecan and agrin.
Nidogen-1 is a ubiquitously expressed basement membrane protein consisting of three globular domains (G1-G3) connected by a link and a rod-like segment. It has been shown to bind several extracellular proteins through different domains and has been proposed to act as a connecting element for basement membrane assembly (1,13). The G3 domain of nidogen-1 binds to a single laminin-type epidermal growth factorlike (LE) module of the ␥1 chain with high affinity (14,15), and this binding is thought to be particularly important for basement membrane assembly. Nidogen-2 is another isoform and also binds several matrix proteins with affinities different from nidogen-1 (16,17), although the binding repertoire is largely overlapping. Site-directed mutagenesis demonstrates the crucial amino acids for the nidogen binding in the laminin ␥1 LE module ␥1 III4 (18) and subsequently these data are confirmed by the crystal structure of the complex between nidogen and laminin fragments (19). The ␥3 chain contains a ␥1-like nidogen binding motif with only a single conservative amino acid substitution (8,9), suggesting that ␥3-containing laminins should be capable of associating with other basement membrane molecules through nidogens.
The ␥3 chain has been localized at the apical surface of ciliated epithelial cells of lung, oviduct, epididymis, ductus deferens, and seminiferous tubules, and it was reported that the ␥3 chain is a non-basement membrane-associated laminin chain (8). The laminin isoforms containing the ␥3 chain are widely expressed in different compartments of the retina (5).
For this study, we prepared recombinant fragments of the laminin ␥3 chain and compared the binding properties with those of the ␥1 and ␥2 chains. These fragments allowed us to raise specific antibodies that are useful for quantitative analyses and examination of the distribution of the ␥3 chain in tissues. We systematically localized the laminin ␥1 and ␥3 chains during mouse organogenesis from day 12 to day 18 and in adult organs at the light microscopic level as well as at the ultrastructural level and found it to be a true basement membrane component. In addition, we performed immunogold double labeling and showed a co-localization of the laminin ␥3 chain and nidogen-1 in basement membranes in vivo.

Production of Recombinant Proteins
The templates used were mouse ␥3 cDNA clone ␥3_3pB provided by Prog. Dr. K. Tryggvasson for ␥3 III3-5 and RNA from mouse testis for ␥3 VI/V fragment. The sense and antisense primer for ␥3 III3-5 were 5-ЈGTCAGCTAGCGCCCTGTCCGTGCCCTGG-3Ј and 5-ЈGTCACTCG-AGCTAGCTCTGGCAGCCCCTCCC-3Ј, respectively, and for ␥3 VI/V, 5-ЈGTCAGCTAGCAGTGCAGAGCGGGG-3Ј and 5-ЈGTCACTCGAGCT-AGCTGCTGCAGCCCACTGG-3Ј, respectively. They were used for amplification by PCR. These primers introduced at the 5Ј end a NheI site and at the 3Ј end a stop codon followed by a XhoI site. A NheI/XhoI cDNA fragment for ␥3 III3-5 as well as that for ␥3 VI/V was inserted into the episomal expression vector pCEP-Pu containing the BM-40 signal peptide (26). These vectors were used to transfect 293-Epstein-Barr virus nuclear antigen-1 cells, and serum-free medium was collected from these cells (26).
Conditioned medium was passed over a DEAE-cellulose column equilibrated with 0.05 M Tris-HCl, pH 8.6, and eluted with a linear 0 -0.5 M NaCl gradient. Most of fragment ␥3 III3-5 did not bind to the column, whereas fragment ␥3 VI/V was eluted at 0.1-0.2 M NaCl. They were further purified on a Superose 12 column (HR 16/50) equilibrated with 0.2 M ammonium acetate, pH 6.8.

Protein Interaction Assays and Laminin-1 Polymerization Inhibition Assay
Solid phase assays with laminin fragments immobilized on the plastic wells of microtitre plates and soluble nidogens followed published procedures (27). Binding was detected by antibodies specific for each nidogen isoform. The radioligand inhibition assay with 125 I-labeled laminin fragment P1 and recombinant nidogen-1 (14) was also used. The laminin-1 polymerization inhibition assay as well as surface plasmon resonance assays (BIAcore 1000, Stevenage, Hertsfordshire, UK) were carried out as described previously (25).

Immunological Assays
Immunization of rabbits, affinity purification of antibodies, enzymelinked immunosorbent assay titration and inhibition radioimmunoassays were carried out using established protocols. Extraction of mouse tissue was performed with neutral buffer containing 10 mM EDTA followed by the same buffer containing detergents, and both buffers contained protease inhibitors (28).

Animals
Female New Mexico Research Institute (NMRI) mice were kept on a normal day/night cycle and received Altromin commercial food and water ad libitum. The day on which, at 11:00 a.m., a vaginal plug was detected after a mating period of 3 h, was designated as day 0 of gestation. On the respective days of gestation, beginning with day 12, pregnant mice were anesthetized with ether and sacrificed by cervical dislocation. After dissection of the uterine horns, the embryos (day 12) or fetuses (days 14, 16, and 18) were removed. Three embryos or fetuses of each developmental stage were investigated. For light microscopic analysis of adult tissues, the various organs from three different threemonth-old NMRI mice were taken.

Immunohistochemistry
Fixation and Preparation of Tissues-For light microscopy, all specimens were fixed by immersion in 4% paraformaldehyde in phosphate buffer, pH 7.2, at 4°C. They were then dehydrated in an ascending series of ethanol from 30 to 100% and embedded in paraffin (29). Serial sections of 5 m were cut with a Reichert's microtome. Every fifth section was stained with hematoxylin for topological orientation within the anatomical regions examined, and staging of the embryos or fetuses was achieved by comparison with the appropriate Theiler stages. For the ultrastructural approach, the tissue pieces were fixed in 4% paraformaldehyde and 0.5% glutaraldehyde for 15 min, dehydrated in a graded series of ethanol up to 70%, and embedded in the acrylic resin LR-Gold (London Resin Company, Reading, UK). Semithin (1-m) and ultrathin (0.8-m) sections were cut according to procedures previously described in detail (29).
Light Microscopic Immunohistochemistry-For the light microscopic immunohistochemistry, sections were deparaffinized, rehydrated, and rinsed for 10 min in PBS. Endogenous peroxidase was blocked by incubation in 3% H 2 O 2 in methanol for 45 min in the dark. Each of the reaction steps was followed by rinsing for 10 min in PBS. The sections were pretreated for 5 min with 10 g/ml protease XXIV (Sigma). The anti-laminin ␥1 antibody, the anti-laminin ␥3 III3-5 antibody, and the laminin ␥3 VI/V antibody were used at a dilution of 1:100 for 1 h at room temperature. The anti-nidogen-1 antibody was also used at a dilution of 1:100 for 1 h at room temperature. The peroxidase-anti-peroxidase method followed the previously described procedures (29). As negative controls, normal rabbit IgGs and the corresponding preimmune sera were used instead of the primary antibodies, at similar concentrations. No immunostaining was observed.
Immunogold Histochemistry-For single labeling using immunogold histochemistry, the tissue sections were incubated for 1 h at 4°C with the antibodies against laminin ␥3 VI/V and laminin ␥3 III3-5. The sections were rinsed in PBS and incubated with the 16-nm gold-coupled goat anti-rabbit IgG diluted 1:20 in PBS for 16 h at 4°C. Thereafter, the sections were rinsed with water and stained with uranyl acetate (10 min) and lead citrate (8 min). The sections were examined with a LEO 906E electron microscope. For double labeling, all sections were incubated for 5 min at room temperature with 1% bovine serum albumin in PBS and then rinsed in PBS. Thereafter, anti-laminin ␥3 antibody diluted 1:100 in PBS was applied for 1 h at 4°C. After a rinse with PBS, the gold-coated (16-nm) goat anti-rabbit antibody diluted 1:200 was applied for 20 min at room temperature. The sections were thoroughly rinsed with PBS. The gold-coated (8-nm) anti-nidogen-1 antibody diluted 1:100 in PBS was then incubated. Colloidal gold particles were prepared and coupled to the antibodies according to our standard protocols (13). All sections were finally rinsed with water and stained with uranyl acetate (10 min) and lead citrate (8 min). As the control for the double labeling experiments, we also applied a monoclonal anti-nidogen-1 antibody (JF4). This approach avoids possible cross-links between the two rabbit IgGs used in the double labeling described above and yielded identical results.

Expression and Purification of Recombinant Proteins-Pre-
vious recombinant studies with the LN module of the ␣1 (30) and the ␤1 and ␥1 chains 3 have shown that this module does not represent an autonomous folding unit and requires the addition of several adjacent, rod-like LE modules from domain V to achieve efficient production in mammalian cells. Therefore, we prepared the LN module of the ␥3 chain together with the complete domain V (four LE modules) (position 29 -488), as already performed for the ␥1 VI/V fragment. Domain ␥3 III3-5 (three LE modules) corresponds to amino acid 766 -927 of the laminin ␥3 chain, a region homologous to the laminin ␥1 chain exhibiting a high affinity binding site for nidogen-1. These fragments were produced in human 293-EBNA cells transfected with episomal expression vectors (26). These cells secreted the recombinant fragments in sufficient amounts into serum-free culture media, which were used for purification by conventional chromatography. The purified fragments showed a single 55-kDa band for domain ␥3 VI/V (Fig. 1, lane 1) and a 23-kDa band for domain ␥3 III3-5 (Fig. 1, lane 5) in SDSpolyacrylamide gel electrophoresis. The purity was also confirmed by a single N-terminal ADMGS(C)YDGV sequence for ␥3 VI/V, in agreement with the presence of a typical signal peptide and an APLAP(C)P(C)PGQ sequence for ␥3 III3-5, the first four residues being derived from the signal peptide cleavage region of the vector.
Inhibition of Laminin-1 Polymerization-A major function of LN modules is the promotion of self-assembly of laminins into non-covalent quasihexagonal networks in a Ca 2ϩ -dependent fashion, as shown for laminin-1 (10) and subsequently for ␣2 chain-containing laminins (11). In an earlier experiment, we developed a quantitative inhibition radioimmunoassay to study the interference of the different ␣ chain fragments (25). In the present study, we used this assay for the ␤ and ␥ chain fragments. The assay was designed to allow the polymerization of 30 -38% of laminin-1 (subsequently set to 100%), which was reduced to only a few percent in the presence of EDTA (Table I). The inhibitors were added in stoichiometric amounts (0.32 M) or in five-fold molar excess (1.6 M) prior to starting the polymerization. Lower activities for ␤3 VI/V and ␥3 VI/V and distinctly low activity for ␤2 VI/V at both concentrations were observed (Table I). As a positive control, we used the laminin E4 fragment corresponding to ␤1 VI/V, which was used in previous inhibition studies (10), and found activity approximately identical to that of recombinant ␤1 VI/V and ␥1 VI/V. Surface plasmon resonance assays were used to examine the binding to immobilized laminin-1. This demonstrated a similar binding of ␥1 VI/V (K d ϭ 0.245 M) and ␤1 VI/V (K d ϭ 0.275 M), and lower affinities were observed for ␤3 VI/V (K d ϭ 1.05 M) and for ␥3 VI/V (K d ϭ 1.75 M). No binding was detected for ␤2 VI/V up to the highest concentration used (4 M).
Immunological Assays-Rabbit antisera were prepared against recombinant ␥3 VI/V and ␥3 III3-5 fragments. These antisera had high titers against the antigen used for immunization and were only marginally cross-reactive with the fragments obtained from ␥1 and ␥2 chains by enzyme-linked immunosorbent assay (data not shown) and by inhibition radioimmunoassays (Fig. 4). Several tissues from laminin ␥3 knock-out mice tested for both antibodies did not show any staining (data not shown).
In inhibition radioimmunoassays, the half-maximal inhibitions were achieved at 0.1 nM for ␥3 VI/V (Fig. 4A) and at 0.06 nM for ␥3 III3-5 (Fig. 4B). A Ͼ1000-fold excess of homologous fragments obtained from other ␥ chains did not show inhibition. Some adult mouse tissues were extracted with EDTA and detergents and examined by this assay together with the ␥1 VI/V assay. However, mouse tissues showed inhibitions in the ␥3 VI/V assay but not in the ␥3 III3-5 assay. Therefore, the amount of the ␥3 chain was quantified by the ␥3 VI/V assay. The much lower amounts of the ␥3 chain (1-6% of the ␥1 chain of laminin) were found in the tissue extracts examined (Table II).
Immunolocalization of Laminin ␥1 and ␥3 Chains during Embryonic Development and in Adult Tissues-From day 12 to day 18, and in adult mouse tissue, staining for the laminin ␥1 chain was seen in almost all epithelial basement membrane zones in all consecutive stages of development, as already known for a long time (Table III). In the skin, for example, the dermal-epidermal basement membrane zone exhibited staining, as did the basement membrane zones of hair follicles, but neither the fibroblasts nor the keratinocytes showed any staining (Fig. 5A). In all stages of kidney organogenesis, only basement membrane zones of the consecutive stages of glomeruli development (comma, S-shaped, early glomeruli) and those of the tubules were positive for the ␥1 chain (Fig. 5B). In the small and large intestine, the epithelial cells were not stained, whereas the basement membrane zones underlying the developing epithelium were stained for the ␥1 chain (Fig. 5C).
From day 12 to day 18, and in adult mouse tissue, staining for the laminin ␥3 chain was seen in basement membrane zones of capillaries, neuroectoderm, and choroid plexus in all consecutive stages of brain development. In addition, the endothelial basement membrane zones of capillaries and larger blood vessels revealed staining for the protein. In the skin, the dermal-epidermal basement membrane zone and the basement membrane zones of hair follicles were positive for the ␥3 chain (Fig. 5D). The endoneurium of peripheral nerves also stained positive for the ␥3 chain. In the kidney, the basement membrane zones of developing glomeruli (comma, S-shaped, early  glomeruli) and tubules were positive for the ␥3 chain (Fig. 5E). The same was true for the basement membrane zones underlying the developing epithelium of the small and large intestine (Fig. 5F). In all stages of testis organogenesis, only basement membrane zones of tubuli seminiferi were stained for the ␥3 chain.
Ultrastructural Localization of Laminin ␥3 Chain during Mouse Development and in Adult Tissues-Affinity-purified antibodies specific for ␥3 VI/V and ␥3 III3-5 were also used for ultrastructural localization of laminin ␥3 chain in mouse tissues by immunogold staining (Fig. 6). The staining was observed in the basement membranes of developing brain arteriole (Fig. 6A), in those of developing tubules in the kidneys (Fig. 6B), and also in muscle basement membranes (Fig. 6C).
We also performed double labeling for a potential binding partner, nidogen-1. Laminin ␥3 chain was co-localized with nidogen-1 in the basement membranes of the adult mouse kidney (Fig. 7). A co-localization was classified as a potential molecular contact when the distance between the different sizes of gold label (8 or 16 nm) was Ͻ30 nm (31). Fig. 8A demonstrates the mean values of nidogen-1 (Fig. 8A, lane a), laminin ␥3 chain (lane b), and co-localization of nidogen-1 and the ␥3 chain (lane c). The ␥3 chain is nearly always co-localized with nidogen-1 (Fig. 8A, lane c). Fig. 8B visualizes the mean values of unspecific (Fig. 8B, lane a) and specific (Fig. 8B, lane b) reactions. The difference between the co-localization rates of the laminin ␥3 chain with nidogen-1 was statistically significant with p values Ͻ 0.001 (Fig.  8B, lane b, asterisk). DISCUSSION Recently, the ␥3 chain of laminin was identified by protein chemistry (8) and by a homology search from the sequence tag data base (9). Sequence comparison with the laminin ␥1 chain predicts that they have similar functional activities. We have now expressed two fragments of the ␥3 chain to compare them We always talk about an organ anlage when primordial developmental stages are seen. As soon as the characteristic features of an organ are found, we talk about developing structures without pointing out the exact embryonic names for each developmental stage. E, embryonic day; ECT, endocardial cushion tissue; BM, basement membrane zone; nd, not yet developed; re, regressed. with those of the ␥1 and ␥2 chains. These fragments were also used to raise specific antibodies against the ␥3 chain. The important step for the supramolecular organization of basement membranes is the formation of two independent networks of collagen type IV and laminins, which are connected to each other and stabilized by interactions with nidogens, perlecan, and other proteins (1,4,12). The polymerization was initially shown for laminin-1 from the Engelbreth-Holm-Swarm tumor. This process is dependent on the concentration (Ͼ0.1 M), temperature (Ͼ30°C), and Ca 2ϩ and could be reversed by EDTA and cooling, demonstrating a non-covalent interaction (32,33). At high concentrations (Ͼ1 M), laminin-1 forms a quasihexagonal network similar to those also found in situ. The N-terminal globular domains of all three laminin-1 chains are involved in the polymerization, as shown in inhibition studies with proteolytic fragments (10, 34 -36). Later, it was shown that ␣2 chain-containing laminins (laminin-2 and -4) can also self-assemble and co-polymerize with laminin-1, but laminin-5 (␣3A␤3␥2) and -6 (␣3A␤1␥1) cannot (11). Up to now, 10 more laminin isoforms are known, but they are not available to perform such analysis. Therefore, we have used domain VI/V fragments to evaluate their activity in laminin-1 polymerization inhibition assays. The previous study on VI/V fragments from different ␣ chains demonstrated a high activity of ␣3B VI/V and ␣5 VI/V fragments, 50 -60% inhibition of laminin-1 polymerization being achieved at equimolar concentrations of laminin-1. However, a five-fold molar excess was required for ␣1 VI/V, ␣2 VI/V, and ␤1 VI/V to produce comparable effects (25). The present data show that ␤2 VI/V can inhibit only ϳ10%, and 30 -40% inhibitions are found for ␤3 VI/V and ␥3 VI/V at five-fold molar excess. The data on ␤1 and ␥1 fragments agree well with previous data (10,11,35). These data suggest that laminin-13, -14, and -15, consisting of ␤2 and ␥3 chains together with different ␣ chains (␣3 chain for laminin-13, ␣4 chain for laminin-14, and ␣5 chain for laminin-15) may not co-polymerize, at least not with laminin-1. Recently, recombinant VI/V fragments from eight laminin chains (but not the ␣3B chain) have been expressed, and homophilic and heterophilic interactions have been analyzed using various techniques (37). It has been shown that most of the LN domains derived from ␣ chains and ␤ chains interact with each other homotypically, except for the ␤1 chain-derived LN domain, which exhibits no self-interaction. However, the ␥ chain LN domains showed limited interactions without self-interactions and especially the ␥3 LN domain interacted only with ␤2 and ␤3 LN domains. Our inhibition data indicated that ␤2, ␤3, and ␥3 LN domains have no binding or only low binding to ␣1, ␤1, and ␥1 LN domains. In this respect, our data contradict earlier assumptions (37). According to the three-arm interaction model for laminin self-assembly, they speculate that the ␥3 chaincontaining laminins may not form tight networks. This possibility should be tested with isolated laminin isoforms. In Table  II, the sum of the ␥1 and ␥3 chains extracted sequentially with neutral buffer containing EDTA followed by detergents is shown. 70 -80% of the ␥1 chain-containing laminins were extracted by EDTA, but only 30 -40% of the ␥3 chain was extracted in the same extract, except from skin and intestine (60 -75%). These data suggest that the ␥3 chain-containing laminins may integrate into tissues by a different mechanism.
The laminin ␥1 chain has a high affinity binding site for nidogen-1 (14), and the interface between the ␥1 IIILE modules and nidogen-1 G3 domain was revealed by the crystal structure of these complexes (19). The ␥1 IIILE4 module docks to an amphitheatre-shaped concave surface on the pseudo-six-fold axis of the ␤-propeller in nidogen-1 with a complementary shape and a small interface. The LE3 module binds over its rim. The key interactions are mediated by the LE4 module as predicted from mutagenesis (18), and the contribution of the LE3 module to the binding is rather small (14). The interaction sites in the ␥1 IIILE4 module are Asp-800, Asn-802, Val-804, Arg-809, and Arg-816, and the first three amino acids occupy about 60% of the interface. Asn-802 and Val-804 are changed to Ser in the human ␥2 chain, and Val-804 is changed to Ser in the mouse ␥2 chain, showing a low binding to nidogen-1 or none at all (21,24,38). These amino acids, except for the mutation R809P, are conserved in human and mouse ␥3 chains, and the interaction sites are also conserved in nidogen-1 and -2 from human and mouse, which is a reason to predict that the ␥3 chain binds to both nidogens with affinities similar to the ␥1 chain.
Contrary to the previous report that the laminin ␥3 chain is only localized apically (8), the systematic light microscopic localization of the laminin ␥3 chain from day 12 to day 18 and in adult tissues shows that the protein is very tissue-and cell-specific and can always be localized in close proximity to basement membranes. The laminin ␥3 chain can be localized in basement membrane zones of brain, skin, kidney, testis, and endoneurium (Table II). However, basement membranes are ultrastructures and only an ultrastructural method can undoubtedly confirm whether or not a protein is a true basement membrane component. Electron microscopic immunohistochemistry revealed that the laminin ␥3 chain is localized in tissue-and cell-specific basement membranes in adult tissues, as well as during development.
We performed immunogold double labeling of the laminin ␥3 chain and nidogen-1. This semi-quantitative method revealed FIG. 6. Electron microscopic localization of laminin ␥3 chain during mouse development. A, laminin ␥3 chain is found in the basement membrane of a day 16 mouse arteriole in the brain. l, lumen; e, erythrocyte. B, laminin ␥3 chain is stained in the basement membrane of the proximal tubule of adult mouse kidney. C, the protein is seen in the basement membrane of the adult mouse muscle. Basement membrane of myocyte (arrows), basement membrane of capillary (asterisk). Bars ϭ 0.25 m. that ϳ98% of the detectable laminin ␥3 chain is co-localized with nidogen-1 in basement membranes in vivo. In general, these results indicate that the laminin ␥3 chain has similar cell biological functions as the laminin ␥1 chain and might form laminin networks via nidogen-1 and -2 (29). However, despite the proposed similar functions of the laminin ␥1 and ␥3 chains, during early developmental stages, the laminin ␥3 chain cannot compensate for the absence of the laminin ␥1 chain, as laminin ␥1 knock-out mice die at a very early embryonic stage (39).