Primary Structure, Developmental Expression, and Immunolocalization of the Murine Laminin α4 Chain*

The complete primary structure of the mouse laminin α4 chain was derived from cDNA clones. The translation product contains a 24-residue signal peptide preceding the mature α4 chain of 1,792 residues. Northern analysis on whole mouse embryos revealed that the expression was weak at day 7, but it later increased and peaked at day 15. In adult tissues the strongest expression was observed in lung and cardiac and skeletal muscles. Weak expression was also seen in other adult tissues such as brain, spleen, liver, kidney, and testis. By in situ hybridization of fetal and newborn tissues, expression of the laminin α4 chain was mainly localized to mesenchymal cells. Strong expression was seen in the villi and submucosa of the developing intestine, the mesenchymal stroma surrounding the branching lung epithelia, and the external root sheath of vibrissae follicles, as well as in cardiac and skeletal muscle fibers. In the developing kidney, intense but transient expression was associated with the differentiation of epithelial kidney tubules from the nephrogenic mesenchyme. Immunohistologic staining with affinity-purified IgG localized the laminin α4 chain primarily to lung septa, heart, and skeletal muscle, capillaries, and perineurium.

The laminins are a complex family of trimeric extracellular matrix proteins (1)(2)(3)(4)(5). They are composed of three types of chains, which are classified as ␣, ␤, and ␥. The chains associate with one another to form a triple-helical coiled coil. To date, the primary structure of 10 genetically distinct laminin chains, ␣1-␣5, ␤1-␤3, and ␥1 and ␥2 have been determined from mammals (6 -28). The different laminin chains are composed of specific modules that include domains with globular structures, rod-like domains containing cysteine-rich repeats, and domains participating in the coiled coil of the long arm. Additionally, the ␣ chains contain a large carboxyl-terminal globular domain with five internal repeat motifs that share homology with corresponding motifs in perlecan (29,30), agrin (31), neurexins (32), and Drosophila crumbs protein and the sex hormone-binding globulin (33).
Numerous functions have been proposed for laminins (for review, see Refs. 1, 4, and 34), but the physiological roles of the different isoforms are still largely unknown. However, the laminin subunit chains differ extensively with respect to their distribution suggesting tissue-specific functions. Insight into the functions of individual chains and isoforms has come particularly from studies on genetic diseases and by targeted gene disruption in mice. First, in epidermolysis bullosa junctionalis, a skin blistering disease, mutations have been described in all three human genes encoding the subunit chains of the laminin-5 isoform (␣3␤3␥2) (35)(36)(37)(38). This demonstrates the requirement of this hemidesmosomal protein for the attachment of epithelial cells to the underlying matrix. Secondly, defects involving the gene for the ␣2 chain (LAMA2) have been described in dy mice (39) and in patients with congenital muscular dystrophy (40). This suggests that this chain, a component of the laminin-2 isoform (␣2␤1␥1), is important for the differentiation or function of muscle fibers. Thirdly, the requirement of the laminin ␤2 chain for a functional glomerular filtration barrier has been demonstrated by the development of proteinuriaassociated nephropathy in mice with inactivated gene (LAMB2Ϫ/Ϫ) (41). However, the actual chain composition of the GBM laminin isoform containing the ␤2 chain is not known.
The development of nephrons, the filtration units of the kidney, involves a specific interplay of signaling between the epithelial cells of the ureter bud and the surrounding mesenchymal cells (42,43). During this process, the epithelial cells induce condensation and polarization of the mesenchyme at the tip of the ureter bud. This group of mesenchymal cells undergoes transepithelialization to form an S-shaped mass referred to as the S body. Later, the S body fuses to the ureter bud. Eventually, the S body gives rise to the proximal and distal parts of the tubular system, as well as to the epithelial component of the glomeruli, i.e. cells covering the inside of the Bowman's capsule and the podocytes that cover the outside of the glomerular capillaries. It has been shown earlier that the expression of laminin genes is subject to characteristic changes in the developing kidney. Specifically, ␣1 chain expression is absent in the undifferentiated mesenchyme, but it is turned on concomitantly with the start of differentiation and polarization of the mesenchyme (44). Furthermore, the ␣1 chain is expressed in the epithelial cells of the ureter bud and the tubules that form the loops of Henle (44).
The function and expression of the laminin ␣4 chain is still poorly known. We have previously reported the sequence of the human laminin ␣4 chain, but thorough studies on developmental expression of this protein are difficult in human tissues. In the present study we have cloned the murine ␣4 chain, determined its primary structure, and examined its developmental * This work was supported in part by grants from the Swedish Medical Research Council, Swedish Cancer Foundation, Academy of Finland, Sigrid Juselius Foundation, and Finland's Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U59865. expression in mouse tissues. Furthermore, polyclonal antibodies were made against the ␣4 chain and used for immunolocalization of the protein in murine tissues. The results showed the ␣4 chain to be located mainly in lung, around heart muscle fibers, and in capillary basement membranes as well as in perineurium.

EXPERIMENTAL PROCEDURES
Isolation and Characterization of cDNA Clones-A murine lung cDNA library in the gt11 cloning vector (CLONTECH) was screened using standard techniques (45) with a human laminin ␣4 chain cDNA clone covering about 1.1 kb 1 of the 3Ј-coding sequence (15). Clone ML1 was sequenced and shown to encode the mouse ␣4 chain. The ML1 clone was subsequently used as a probe to screen a mouse lung cDNA library, and in this manner a series of overlapping cDNA clones were identified. The 5Ј-end clone, ML89, was obtained using an amplified library sequence. A 50-l PCR was performed using 0.02 l of library stock as a template and gt11 reverse primer as one primer and (5Јctgggccaacac-cagcttcac3Ј) as the other. One l of the product was reamplified using the same vector primer and a nested specific primer (5Јcctcggggccaagt-tcctgattc3Ј). The resulting product was electrophoresed on agarose (SeaPlaque GTG). A pipette tip was stabbed through a 1-kb product and swirled in a PCR mixture. A following round of amplification was carried out using the same specific primer and biotinylated vector primer. Reasonably specific product was obtained and gel-purified, and its strands were separated using streptavidin-coated Dynabeads M-280 (Dynal). The product was identified by sequencing and used as a probe to find ML89. The 0.45-kb gap between ML89 and ML64A ( Fig. 1) was filled by a two-round PCR from the library using (5Јgaacgctgtgcacctg-gttac3Ј) as the 5Ј primer and the aforementioned 3Ј primers. To eliminate possible errors from PCR, products from separate reactions were sequenced in parallel. The cDNAs were subcloned into pBluescriptI-I.SKϪ (Stratagene), pUC18, or pUC19 vectors for sequencing by the chain termination method (46) using a T7 sequencing kit (Pharmacia Biotech Inc.). All sequence analyses were carried out using the GCG software (47). All primers in this study were synthesized using DNA/ RNA synthesizer model 392 (Applied Biosystems).
Northern Analysis and in Situ Hybridization-For Northern analysis, filters containing RNA from several mouse tissues (CLONTECH) were hybridized by standard procedures (45) with a probe spanning nucleotides 2,992-4,262 of the mouse laminin ␣4 chain cDNA and a ␤-actin probe as a control. For in situ hybridization with radiolabeled probes, a 0.5-kb fragment from the 5Ј end of ML54 was ligated into EcoRI-and SacI-digested pGEM-3Z vector (Promega), making use of the EcoRI site in the ML54 5Ј-linker sequence. In situ hybridization analysis was carried out using single-stranded RNA probes on murine fetal and neonatal tissues as described previously (48). For cardiac and skeletal muscle, the tissue samples were obtained from newborn and 8-day-old animals, fixed with 4% paraformaldehyde in phosphate-buffered saline and embedded in paraffin, and processed essentially as described by Breitschopf et al. (49). Briefly, fragment from the mouse laminin ␣4 cDNA clone was labeled with digoxigenin (Boehringer Mannheim), cut to about 150-base pair fragments by alkaline hydrolysis, and then used as a probe. The tissue sections were treated with 0.2 M HCl, 0.1 M triethanolamine buffer, pH 8.0, containing 0.25% (v/v) acetic anhydride and 100 g/ml proteinase K. The sections were hybridized with the probe at 62°C for 16 h. After rinsing in 50% formamide and standard sodium citrate, the probe was immunologically detected with an antibody to digoxigenin conjugated to alkaline phosphate enzyme (Boehringer Mannheim). The color was developed with nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate.
Preparation of Recombinant Laminin ␣4 Chain Antigens and IgG-For the generation of antibodies, recombinant protein containing base pairs 1221-2342 of the cDNA, corresponding to residues 408 -781 in domain I/II of the human ␣4 chain was expressed. The insert was cloned into the pGEX-1T vector (Pharmacia), expressed according to standard procedures, and the recombinant glutathione S-transferase fusion protein was then purified using glutathione-Sepharose affinity columns (Pharmacia). Two rabbits were immunized five times with standard procedures using 50 -100 g of antigen for each injection.
Laminin ␣4 chain-specific IgGs were affinity-purified as follows. The bacterial fusion protein, eluted from the glutathione-Sepharose matrix was further purified by reversed phase chromatography. Aliquots of the fusion protein (150 l) were applied onto a Delta Pak RP4 3.9 ϫ 150 mm column (Waters) and eluted with a CH 3 CN gradient (45-52% in 28 min, flow 0.8 ml/min) in 0.1% trifluoroacetic acid. The correct fractions were identified by SDS-polyacrylamide gel electrophoresis, concentrated in a SpeedVac, and redissolved in 2 M urea, 50 mM Na 2 HCO 3 , pH 8.3, 0.5 M NaCl. This material was coupled to CNBr-activated Sepharose 4B (Pharmacia) according to the manufacturer's instructions. The IgG was first preadsorbed with bovine serum albumin-coupled Sepharose, and then the nonbound fraction was incubated with the glutathione Stransferase-antigen fusion protein coupled to Sepharose. Following extensive washing with phosphate-buffered saline the bound IgG was eluted with 0.1 M glycine, pH 3.0, immediately neutralized, pooled, and dialyzed against phosphate-buffered saline. The affinity-purified IgG was used for immunostaining at a 15 g/ml concentration.
Western Blotting and Immunocytochemistry-Specificity of the antisera was examined by Western blotting analyses. Briefly, denaturing 6% polyacrylamide gels were electrophoresed under reducing conditions, after which the proteins were blotted onto nitrocellulose filters using a semidry blotting device (Bio-Rad) according to the manufacturer's instructions. The bound antibodies were detected using horseradish peroxidase-coupled secondary antibodies (Dako) and chemiluminescence (NEN Life Science Products).
For immunohistochemical analyses mouse tissues were embedded in paraffin following fixation in 10% formalin. Endogenous peroxidase activity was quenched with 3% H 2 O 2 for 30 min, and the samples were boiled for 10 min in 1 M urea. Skeletal muscle was fixed with 3% acetic acid, 70% ethanol. Staining was performed by indirect techniques using biotinylated secondary antibodies in conjunction with a horseradish peroxidase ABC kit (Dako). Peroxidase activity was detected with diaminobenzidine or aminoethylcarbazole.

Primary and Domain Structure of the Mouse Laminin ␣4
Chain-Six overlapping cDNA clones isolated and characterized in this study (Fig. 1) together covered a total of 5,824 nucleotides, including 208 bp of a 5Ј-untranslated region and 168 bp of a 3Ј-untranslated region, respectively (GenBank TM accession number U59865). The open reading frame of 5,448 nucleotides coded for a 1,816-residue laminin ␣-type chain (Fig.  2), characterized by the presence of a carboxyl-terminal G domain with five internal repeat motifs. The predicted sequence of the first 24 residues is characteristic of a hydrophobic signal peptide which is present in all other laminin chains, and the amino acid sequence around position 24 is characteristic for a signal peptidase cleavage site as described by von Hejne and co-workers (50,51). Thus, the mouse laminin ␣4 chain proper contains 1,792 residues. The calculated mass of the entire translation product is 201,818 and that of the processed ␣4 chain 199,164. Expression of the Laminin ␣4 Chain during Mouse Development-In our previous work (14) we showed by Northern hybridization studies that the human laminin ␣4 chain is expressed in a highly tissue-restricted manner, indicating that this chain is present in a laminin isoform with highly specific, but as yet unknown functions. However, in situ hybridization studies were only possible to perform on a limited number of embryonic brain, lung, and muscle tissues. These studies indicated that the expression of the ␣4 chain is mainly confined to cells of mesenchymal origin. The present cloning of the mouse ␣4 chain allowed a more detailed analysis of its expression in mouse tissues.
Expression of the laminin ␣4 chain was first studied for levels of mRNA using Northern analysis (Fig. 3). Using RNA from whole mouse embryos, the level of expression was low at embryonic day 7, but it later increased being highest at day 15. In adult, strong expression of a 6-kb mRNA was seen in lung and in skeletal and heart muscle tissues. Weaker expression was seen in all other tissue studies, i.e. brain, spleen, liver, kidney, and testis. In testis a second weak band of about 5 kb was also observed.
In situ hybridization of mouse embryos revealed expression of the ␣4 chain in mesenchymal cells of several tissues (Fig. 4). In the developing lung strong expression was seen in mesenchymal cells surrounding the branching lung epithelia (Fig. 4,  A and B). Expression was also seen in the villi and submucosa of the intestine (Fig. 4, C and D) and in the external root sheath of vibrissae follicles (Fig. 4, E and F). The expression in intestine and vibrissae may be associated with the endothelial cells of capillaries. Distinct expression was also shown to be present surrounding the nuclei of both cardiac and skeletal muscle fibers (Fig. 5). Some of the signals in muscle appeared to be present in endothelial cells of capillaries.
The expression pattern in the developing kidney was intense, but only transient. Transcripts were first observed in the newly formed mesenchymal condensates next to ureter buds (not shown). Expression was strong in the early kidney tubules differentiating from the mesonephric condensates (Fig. 6, A-F). Following maturation of the proximal nephron into distinct glomerular and tubular structures, the expression of the ␣4 chain diminished significantly. In the mature kidney of the adult mouse no expression could be seen (Fig. 6, G and H).
Characterization of Polyclonal Antibodies-The antiserum raised against a portion of domain I/II of the laminin ␣4 chain was found to be suitable for characterization by Western blotting. To examine for monospecificity of the antiserum against the ␣4 chain, Western blotting was carried out on extracts from cultured cells and tissues, such as heart and lung. As shown in Fig. 7, the antiserum, used at 1:1000 dilution, detected a distinct band corresponding to slightly over 200 kDa in the cell culture medium of human SK-LMS-1 leiomyosarcoma cells (ATCC HTB 88). The calculated size of the unglycosylated ␣4 chain is 199,164 daltons and the slightly larger size obtained here is therefore likely to correspond to a glycosylated polypeptide chain which contains several potential glycosylation sites as described above (Fig. 2). Consequently, it could be concluded that the antiserum is specific for the murine laminin ␣4 chain.
Immunolocalization of the Murine Laminin ␣4 Chain-Immunohistochemical staining was carried out on several embryonic and postnatal mouse tissues to establish the sites of deposition of laminin isoforms containing the ␣4 chain. Affinitypurified IgGs were used to ensure the specificity of the immunostaining and the staining was carried out using peroxidase-labeled ABC system. The results demonstrated highly restricted tissue distribution of the protein. Thus, the ␣4 chain was not detected in any significant amounts in basement membranes of epithelia and liver or those of glomeruli and tubuli of the kidney. In contrast, this laminin chain was generally located in mesenchymal tissues such as perineurium and skeletal and heart muscles, as well as in subendothelial basement membranes of capillaries. Lung septa also stained strongly.
Staining of adult cardiac muscle revealed positive reaction as a continuing lining of the cell membrane of most individual fibers (Fig. 8, A and B). Additionally, strong staining was observed in the subendothelial basement membranes of heart capillaries (Fig. 8, A and B). Staining of adult skeletal muscle also revealed reaction, but it appeared to be mainly present in capillary walls, while the sarcolemma stained very weakly if at all (Fig. 8C). This differs from what we observed in embryonic and newborn skeletal muscle which did show specific staining of the sarcolemma or its immediately adjacent extracellular matrix (not shown). In peripheral nerves peroxidase staining yielded clear reaction of the perineurium (Fig. 8D), and the Schwann cell sheets appeared to be positive also. Some axons showed positive reaction, but as it was not observed in all axons it is probably unspecific (data not shown). Examination of lung tissue showed strong specific staining of the alveolar septa (Fig. 8E). The staining appeared to be located both in the interstitium as well as in some subepithelial and subendothelial capillary basement membranes (Fig. 8E). Staining with preimmune control serum did not reveal any specific staining of the lung (Fig. 8F). DISCUSSION The present work describes the full sequence of the mouse laminin ␣4 subunit chain. Importantly, the clones isolated in this study facilitated thorough studies on the temporal and spatial expression of this chain, which was shown to have a highly restricted expression pattern during mouse development.
Comparison of the mouse ␣4 chain sequence with that of human one revealed a sequence identity of 88.1%. Of the 44 cysteines and 19 potential N-linked glycosylation sites in the mature polypeptide, 43 and 18 are conserved, respectively. It is of interest that although the total length of the mouse and human chains is identical, in the alignment there are two balancing 2-residue gaps located in the G-domain. The significance of these gaps is not clear, but it is possible that the structural constraints posed by the folded polypeptide chain allow some variation at these sites. The first gap is close to the boundary between domains I/II and G, and the second gap is located between the third and fourth subdomains of the G domain. In the drosophila ␣-chain, a spacer sequence of threonines and serines lies between these subdomains (52,53).
In general, tissue distribution of the laminin ␣4 chain was shown to extensively differ from that of the other laminin ␣ chains. The ␣4 chain is predominantly located in lung alveolar septa, around heart muscle fibers, in the perineurium, as well as in subendothelial basement membranes of capillaries. In contrast, the ␣1 chain is expressed mainly in the central nervous system, and in the developing kidney (44,54). Expression of the ␣1 chain is also seen in developing lung and gut epithelium (54), but the expression is never mesenchymal as was shown to be the case for the ␣4 chain in the present study. The ␣2 chain is expressed predominantly by skeletal and heart muscle cells (11,55,56). Although the ␣2 chain is expressed in mesenchymal cells in several tissues during human embryonic development, the expression pattern is different from that of the ␣4 chain (10). Here, the ␣4 chain was shown to differ from a2 as ␣4 is present around adult cardiac but not skeletal muscle fibers. However, the ␣4 chain is expressed in fetal skeletal muscle fibers as determined by in situ hybridization and is present around these fibers in embryos and newborn mice as shown by immunohistochemical staining. Therefore, the developmentally regulated ␣4 chain may be of importance for the development of skeletal muscle and for function during early development. The ␣2 and ␣4 chains also differ from each other in that the ␣2 chain is expressed in the developing human kidney in the mesenchyme adjacent to the condensing pretubular epithelia, while ␣4 is absent in this region (10).
Although the ␣4 chain is structurally most closely related to the ␣3 chain, their tissue distribution is very different. The ␣3 chain is epithelial. It is seen in the epithelium of skin, lung, and FIG. 8. Immunolocalization of the laminin ␣4 chain using peroxidase-labeled second antibodies. A, low magnification of adult cardiac muscle reveals linear staining of most muscle fibers and intense staining of capillary basement membranes. B, higher magnification of heart muscle emphasizes the intense subendothelial staining of capillaries (straight arrow), as well as the reaction around individual muscle fibers (bent arrow). C, staining of adult skeletal muscle yields staining in what appears to be capillaries, whereas the sarcolemma does not exhibit any strong staining (longitudinal section). D, staining of peripheral nerves shows distinct staining of the perineurium as well as Schwann cell sheaths. E, staining of lung reveals distinct, but discontinous staining of alveolar septa with anti-␣4 chain IgG, whereas preimmune IgG (F) is completely negative. digestive tract, but never in the mesenchyme (57). In contrast, the ␣4 chain is mainly mesenchymal and endothelial. In addition, the enamel organ shows strong signals for the ␣3 chain (57), but we did not see any signals above background for the ␣4 chain in the developing teeth (not shown). The newly discovered ␣5 chain is strongly expressed in the adult kidney (16) where the level of ␣4 chain expression is very low. There is little evidence for the presence of the other ␣ chains in several mesenchymal tissues, i.e. in lung and intestine, where we see the ␣4 chain expressed (10,(55)(56)(57)(58). These tissues do express ␤1 and ␥1 chains (54, 56), but not ␤3 or ␥2 (57). Furthermore, in the embryonic kidney the mesenchyme adjacent to the ureter bud lacks ␤2, ␤3, and ␥2 chain expression (41,57,58), while it gives strong signals for the ␤1 and ␥1 chains (44,54). Therefore, it is possible that the ␣4 chain associates in these tissues with the ␤1 and ␥1 chains to form the eighth laminin isoform. However, this and the other possible chain combinations involving the ␣4 chain remain to be determined.
The subendothelial location of the ␣4 chain suggests that the laminin isoform containing it may serve as an adhesion protein for endothelial cells in a similar manner as the ␣3 chain containing laminin-5 (59). However, it is apparent that ␣4 chain containing laminin isoforms have broader physiological functions, as this polypeptide chain is also so prominent in striatic muscle fibers and peripheral nerves. The unique expression pattern of the laminin ␣4 chain during a special period of nephrogenesis, i.e. the formation of the proximal nephron, implies specific temporal requirements for a laminin isoform containing the ␣4 chain. The ␣1 chain may also be important for the formation of the proximal nephron, but it is apparently also important for the formation and maturation of the entire nephron as it has a unique expression pattern in the distal parts of the tubuli (44). Expression of both chains decreases in the kidney with age.
Thus far, no diseases have been linked to the genes for the ␣1, ␣4, or ␣5 chains, as is the case for the ␣2 and ␣3 chain genes that are affected in congenital muscular dystrophy and epidermolysis bullosa, respectively (35)(36)(37)(38)(39)(40). However, based on the wide distribution of the laminin ␣4 chain, it can be anticipated that mutations in its gene can lead to severe complications and diseases. The generation of transgenic mice containing an interrupted LAMA4 gene may shed light on the role of this interesting laminin component.