Molecular cloning of a novel laminin chain, alpha 5, and widespread expression in adult mouse tissues.

We have identified a fifth member of the α subfamily of vertebrate laminin chains. Sequence analysis revealed a close relationship of α5 to the only known Drosophila α chain, suggesting that the ancestral α gene was more similar to α5 than to α1-4. Analysis of RNA expression showed that α5 is widely expressed in adult tissues, with highest levels in lung, heart, and kidney. Our results suggest that α5 may be a major laminin chain of adult basal laminae.

Accordingly, we undertook a search for additional laminin ␣ chains. Using the polymerase chain reaction, we have identified laminin ␣5, a novel murine ␣ chain. Sequence analysis reveals that ␣5 is more similar in domain structure to Drosophila laminin A (34,35) than it is to mammalian ␣1-4; the ancestral vertebrate ␣ chain may, therefore, have been more similar to ␣5 that to ␣1-4. RNA analysis demonstrates that ␣5 is widely expressed in adult tissues and thus may be a major laminin ␣ chain of adult basal laminae.

MATERIALS AND METHODS
Degenerate primers were designed based on sequences conserved between mouse laminin ␣1 (7) and human laminin ␣2 (36). Reverse transcription-PCR 2 was performed on RNA from postnatal day 7 mouse kidney using a GeneAmp kit (Perkin-Elmer). One pair of primers from the amino terminus of domain V (sense, 5Ј-GGNAGTTGYATHT-GYTAYGGNC-3Ј and antisense, 5Ј-TRCANTGYTCRCARTTDATNCC-3Ј, where N ϭ A/G/C/T, H ϭ A/T/C, and D ϭ G/A/T) produced a fragment of appropriate length (ϳ330 base pairs). The band was excised from agarose (SeaPlaque GTG, FMC Bioproducts, Rockland, ME), reamplified with the same primers, purified with a Wizard PCR Preps kit (Promega Corp., Madison, WI), and incubated with BglI to digest laminin ␣1 products, thereby preventing their further amplification. The remaining full-length product was reamplified and ligated into the pCRII vector (Invitrogen Corp., San Diego, CA).
One resulting clone, DB2, bore an insert related to but distinct from laminins ␣1 and ␣2. The DB2 insert was labeled with [ 32 P]dCTP by the random primed DNA labeling kit (Boehringer Mannheim) and used to screen an adult mouse lung oligo(dT)ϩ random primed ZAP II cDNA library (Stratagene, La Jolla, CA). Subsequent cDNA library screening was performed as a "walk" using selected restriction fragments of hybridizing phage to obtain overlapping clones. Clones were sequenced on an ABI 373A DNA sequencer using a Taq DyeDeoxy Terminator cycle sequencing kit (Applied Biosystems Inc., Foster City, CA). All sequences were determined from both strands. Data base homology searches were performed on the BLAST server at the National Center for Biotechnology Information (37), and sequences were compared using Genetics Computing Group programs (38).
For Northern analysis, a filter containing poly(A)-selected RNA from several adult mouse tissues (Clontech) was hybridized with a probe comprising nucleotides 7855-9361 of the ␣5 sequence. RNase protection analysis was performed as described previously (24).

RESULTS AND DISCUSSION
Molecular Cloning of Laminin ␣5-We used PCR to amplify a 334-base pair fragment from mouse kidney that encoded a novel laminin-like sequence. This fragment was used as the starting point for isolation of a series of overlapping cDNA clones that spanned Ͼ11 kb. Sequence analysis revealed that the cDNAs encoded a single open reading frame of 3610 amino acids (Fig. 1). The sequence of the predicted protein is related to but clearly distinct from those of previously reported laminin chains (7-19, 34 -36). As detailed below, the novel sequence is more closely related to laminin ␣ chains than to ␤ or ␥ chains. * This work was supported by a National Institutes of Health grant. 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  Because four mammalian laminin ␣ chains have been described to date (7,11,(15)(16)(17)(18), we have named the novel chain laminin ␣5.
The first amino acid of the deduced sequence is aspartic acid rather than methionine, indicating that the cDNAs do not reach the 5Ј end of the coding sequence. Repeated efforts to obtain additional cDNAs failed. Based on homology to other laminin ␣ chains, we believe that the mature protein contains ϳ3630 amino acids, of which we have identified 3610.
Comparison with Other Laminin Chains-The laminin ␣5 chain contains eight domains, based on predicted secondary structure and homology to the laminin ␣1 chain (Fig. 2A). Their nomenclature follows that for ␣1, which was in turn designed to maintain consistency with the ␤1 and ␥1 chains (1, 7). The carboxyl-terminal half of the protein contains a large (ϳ100 kDa) globular domain (G) and an ␣-helical segment (I/II). The amino-terminal half contains three cysteine-and glycine-rich regions predicted to form rigid, rodlike structures (IIIa, IIIb, and V) and three smaller globular regions (IVa, IVb, and VI).
Five of the domains are characterized by repeating structures: heptad repeats with hydrophobic residues in the first and fourth position of domain I/II; "EGF-like" repeats of ϳ50 amino acids each in domains IIIa, IIIb, and V; and five tandem "G" repeats of ϳ186 amino acids each in G.
Several features of ␣5 identify it as an ␣ chain ( Fig. 2A). First, it contains a G domain, which is present in all ␣ but no ␤ or ␥ chains. Second, ␣5, like ␣1 and ␣2, contains three sets of EGF-like repeats (IIIa, IIIb, and V) and three globular regions (IVa, IVb, and VI) in its amino-terminal half, whereas no ␤ or ␥ chain has more than two of each. Third, ␣5 lacks the ␣-insert between domains I and II that characterizes ␤ chains.
Of the four vertebrate laminin ␣ chains characterized to date, two (␣1 and ␣2) contain domains G-VI, whereas the other two (␣3 and ␣4) are truncated and contain only domains G-IIIa ( Fig. 2A). An alternatively spliced product of the ␣3 gene, ␣3B, which contains domains IIIb and IV, has been identified but not yet fully sequenced (12). In its domain structure, ␣5 is more similar to ␣1 and ␣2 than to ␣3 or ␣4, and it can therefore be classified as a "full-length" ␣ chain. On the other hand, within the domains shared by all ␣ chains, the laminin ␣5 sequence is more similar to ␣3 and ␣4 than to ␣1 or ␣2 (Fig. 2B). Thus, sequence analysis reveals an apparent discrepancy between relationships based on primary and secondary structure.
More surprising is that ␣5 is more similar in domain structure to the only known Drosophila A chain (␣D (34,35)) than it is to any of the vertebrate ␣ chains. Both ␣5 and ␣D contain 11 EGF repeats in domain V, 4 in domain IIIb, and 7 in domain IIIa, whereas ␣1 and ␣2 have 4, 9, and 4 repeats in domains V, IIIb, and IIIa, respectively ( Fig. 2A). Likewise, domain IVb is much larger in ␣D and ␣5 (558 and 577 amino acids) than in ␣1 or ␣2 (196 amino acids).
Together, these results suggest the evolutionary scheme diagrammed in Fig. 2C. We propose than an ancestral ␣ gene, similar in domain structure to ␣D and ␣5, was duplicated early in the vertebrate lineage. One of the daughter genes evolved the ␣1/2 domain structure, perhaps by recombination (39), and was then duplicated again to generate ␣1 and ␣2. The other product of the original duplication was also reduplicated. One daughter evolved into ␣5 without further rearrangement, while the other suffered truncation and then duplicated yet again to generate ␣3 and ␣4. Although speculative, this scheme accounts for the otherwise puzzling findings that ␣5 resembles ␣1 and ␣2 in secondary structure but is most closely related to ␣3 and ␣4 in primary sequence.
Expression of Laminin ␣5-RNA from a panel of adult mouse tissues was subjected to Northern analysis with an ␣5 cDNA. High levels of ␣5 mRNA were detected in heart, lung, and kidney (Fig. 3A). Longer exposures revealed lower but significant levels of ␣5 mRNA in brain, muscle, and testis (Fig. 3B). In all of these tissues, an RNA species of ϳ12 kb was detected; an additional RNA of 9 kb, visible only in testis, would be unable to encode full-length ␣5. A more sensitive RNase protection assay revealed significant levels of ␣5 RNA in liver, as well as in gut and skin (not shown). Moreover, RNase protection and in situ hybridization analyses indicated that ␣5 is expressed in many tissues by embryonic day 11. 3 Thus, the ␣5 gene is widely expressed.
The broad distribution of ␣5 RNA in adult tissues stands in marked contrast to the restricted patterns of expression of the ␣1, ␣3, and ␣4 genes (16,18,30,41). 4 Laminin ␣2 RNA is widely distributed in adult tissues but is predominantly ex-  (38), is shown to the left of the other ␣ chains. Numbers of EGF repeats, rounded to the nearest integer, are indicated within domains III and V. Domain structures of ␤1 and ␥1 are shown to indicate the basis for assigning ␣5 to the ␣ subfamily. ␣D, Drosophila A. B, relationships among mammalian and Drosophila ␣ chains, based on sequence alignment performed by the PILEUP program (38). Comparisons were based on domains G-IIIa, which all ␣ chains contain. C, evolutionary scheme for vertebrate ␣ chains, incorporating comparisons of primary sequences (from B) and predicted secondary structure (from A). In this scheme, the ancestral ␣ gene was more similar in domain structure to ␣5 than to ␣1-4. See text for details. pressed by mesenchymal or mesodermal cells (36). Taken together, these results suggest that laminin ␣5 is a major ␣ chain of adult epithelial and/or endothelial basal laminae. Moreover, it seems possible that reports of ␣1-like immunoreactivity in tissues such as kidney, lung, and muscle (20,23,(25)(26)(27)(28), which contain little ␣1 RNA, reflect cross-reactivity of anti-␣1 antibodies with ␣5. We are currently preparing monospecific antibodies to evaluate this possibility.