Molecular Cloning and Tissue-specific Expression of a Novel Murine Laminin γ3 Chain*

A novel laminin γ3 chain was identified from the expressed sequence tag data base at the National Center for Biotechnology Information. A complete cDNAderived peptide sequence reveals a 1592-amino acid-long primary translation product, including a tentative 33-amino acid-long signal peptide. Comparison with the laminin γ1 chain predicts that the two polypeptides have equal spatial dimensions. In addition, the well conserved domains VI and III(LE4) predict that γ3 containing laminins are able to integrate to the laminin network and also via nidogen connect to other protein networks in the basement membranes. Combination of Northern analysis and in situ hybridization experiments indicate that expression of the γ3 chain is highly tissue- and cell-specific, being significantly strong in capillaries and arterioles of kidney as well as in interstitial Leydig cells of testis.

There are two laminin ␥ chains established to date (7,8,12). The ␥2 chain is expressed restrictedly and is mainly confined to epithelial cells (12,30). Of the eleven characterized isoforms, only laminin-5 (␣3␤3␥2) contains the ␥2 chain (31)(32)(33). In contrast, the ␥1 chain is widely expressed (12,22,34,35) and takes part in all the other known isoforms (36 -41). This dominant position differs from the situation among the ␣ and ␤ chains, both classes having at least one widely expressed, alternative chain (22,38,(42)(43)(44)(45). We, therefore, considered the possibility of the existence of a third ␥ chain. Such a chain could be particularly interesting with respect to the size of the laminin family. An alternative ␥ chain would nearly double the number of potential laminin isoforms and, more importantly, proportionally increase the number of identifiable components embedded in the protein meshwork in basement membranes. Clearly, this would significantly improve the resolution of our current basement membrane models that are based on a limited number of distinct proteins (46 -49).
In a primary sequence, a coiled coil motif is seen as a sevenamino acid-long repeat (a, b, c, d, e, f, g) in which the residues at positions a and d are predominantly hydrophobic (50,51). Once the coiled coil is formed, these residues are effectively protected from the polar environment of tissue fluids. While the hydrophobic interactions greatly favor the coiled coil assembly, the specificity of the interaction is often determined by the polar residues running parallel on either side of the hydrophobic core as well as by the polar and charged residues embedded in the core (52,53).
In this study, we predicted the hydrophobic profile of a ␥ chain core in the C-terminal part and used this profile to discriminate against other coiled coil sequences that we found from the translated EST 1 data base at National Center for Biotechnology Information (54). Using this strategy, we could identify and further characterize a novel laminin ␥3 chain.

MATERIALS AND METHODS
Data Base Search-The EST data base at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov) was searched with tblastn sequence comparison program (55) installed at the National Center for Biotechnology Information server. A C-terminal sequence from the human laminin ␥1 chain (8) was used as a query. The probability of finding a match by chance was set to E ϭ 1000, since the sequence homology between laminin chains in the coiled coil regions is usually low, typically about 20 -30% identical residues (10, 16 -18). Previously uncharacterized translations matching the query were first selected for the presence of a cysteine within the 10 C-terminal residues. To compare the amphipathicity and the hydrophobic profile along the hydrophobic core, the remaining translations were analyzed with the helical wheel program from the Genetics Computer Group software package (56). This permits easy visualization of the aligned residues in a and d positions of successive turns of the helix. A murine EST clone (373279) from the WashU-HHMI Mouse EST Project at the Washington University (57) was found to have a similar hydrophobic profile to the laminin ␥1 and to the ␥2 chains. Furthermore, the sequence of this clone (GenBank TM accession number W64443) was found to be similar to that of the laminin ␥1 chain by the WashU-HHMI Mouse EST Project. Clone 373279 was purchased from Genome Systems and used * This work was supported in part by grants from the Swedish Medical Research Council, the Hedlund Foundation, the Academy of Finland, the Faculty of Veterinary Medicine from the University of Helsinki, and European Union Grant EU-BIO4-CT96 -0537. 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 as a probe for Northern analysis and for screening a mouse testisspecific cDNA library (CLONTECH).
Molecular Cloning-Two methods were used for obtaining additional clones from the same gene as the 373279 clone. First, a mouse testisspecific cDNA library was screened with gene-specific restriction fragments (58). Second, various gene-specific primers and a custom anchor primer 5Ј-tcgagcggcctcccgggcaggt-3Ј were used in rapid amplification of cDNA ends for amplifying flanking fragments from embryonic mouse cDNA (E15, CLONTECH). For sequencing, the cDNA fragments were ligated into pBluescript or pCRScript vectors (Stratagene). The cycle sequencing reactions were done with Taq-derived polymerase using termination chemistry (Perkin-Elmer) and the extension products resolved with Applied Biosystems 377 and Applied Biosystems 310 automated sequencers. All the oligonucleotides used in this study were synthesized with Applied Biosystems 392 DNA/RNA oligosynthesizer. The sequence data obtained were analyzed with the Genetics Computer Group software (56).
Northern and in Situ Hybridization Analyses-An RNA blot with 2 g/lane of poly(A)-RNA from various adult mouse tissues (CLONTECH) was hybridized according to standard procedures (58). Two probes giving essentially identical results were sequentially applied: a 0.7-kb fragment from the coiled coid region (nucleotides 3585-4263) and a 0.45-kb segment from the 3Ј-UTR. Random priming strategy was applied in labeling the probes (Prime-a-gene, Promega). The filters were hybridized at ϩ42°C for 16 -24 h in a solution containing 50% formamide, 2% SDS, 50 g/ml of polyvinylpyrrolidone (Pharmacia), 50 g/ml bovine serum albumin, 50 g/ml Ficoll 400 (Pharmacia), and 5 ϫ SSPE (750 mM NaCl, 50 mM sodium phosphate, pH 7.4, 5 mM EDTA). After washing in 1 ϫ SSC ϩ 0.1% SDS for 2 ϫ 15 min and in 0.1 ϫ SSC ϩ 0.1% SDS for 15 min, Kodak X-Omat films were exposed to the filterbound radioactivity for several days at Ϫ70°C using intensifying screens.
In situ hybridization was carried out on tissues from adult mouse essentially as described elsewhere (44,59,60). Briefly, tissues were fixed in the fresh 4% paraformaldehyde, dehydrated by a gradient of ethanol, embedded in paraffin, and sectioned. Following postfixation in 4% paraformaldehyde, the sections were incubated in phosphate-buffered saline containing 0.1% active diethyl pyrocarbonate (Sigma), equilibrated in 5 ϫ SSC and prehybridized for 2 h at 55°C. Hybridization was then carried out overnight at 55°C. For generation of singlestranded RNA probes, a 450-base pair PstI fragment from the 3Ј-UTR of the cDNA was cloned into pGEM3Z (Promega). Antisense and sense riboprobes were labeled with digoxigenin-11-UTP by in vitro transcription with T7 and SP6 RNA polymerase (Roche Molecular Biochemicals), respectively. After washing in 50% formamide and standard sodium citrate, the sections were incubated with an alkaline phosphatase coupled anti-digoxigenin antibody and stained using nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate solutions (Roche Molecular Biochemicals).

RESULTS AND DISCUSSION
Molecular Cloning-The specificity of laminin chain assembly lies at the coiled coil domain. According to our hypothesis, this is reflected in the chain-specific hydrophobic profiles in the coiled coil core. Similarly, the intraheptad position of the Cterminal cysteine residues in the coiled coil domain is chainspecific. Starting from these premises, the EST data base at the National Center for Biotechnology Information was searched as described under "Materials and Methods." The EST clone 373279 from the mouse EST project at the Washington University (see "Materials and Methods") fulfilled these criteria. Furthermore, the sequence of this clone was found to be significantly similar to that of the murine ␥1 chain by the EST project. We then used the clone 373279 as a starting point in the search for overlapping clones. Together with 373279, the identified clones encoded for an open reading frame of 4829 nucleotides that included 53 base pairs of tentative 5Ј-UTR (Fig. 1 (3). Using the ␥1 chain as a model, some predictions about the properties of the ␥3 chain can be made. First, it is likely that the dimensions of the two chains are similar. The lengths of the coiled coil domain are 579 (␥1) and 569 (␥3) residues. The globular domains VI and IV are of similar size as well. Furthermore, both chains have 11 copies of laminin repeat motifs (63,64). Second, the domains to which specific functions have been mapped in the ␥1-chain appear to be conserved: domain VI with a calcium-dependent laminin self-assembly activity (49,65) and nidogen binding site in the fourth laminin repeat in the III domain (66) (see Fig. 3). Taken together, this suggests that the ␥3 chain possesses a full networking capability similar to the ␥1, but different from the ␥2 chain.
Expression of the Laminin ␥3 Chain-To investigate the spatial expression in mouse tissues, an RNA blot containing RNA from various adult organs (CLONTECH) was hybridized serially with two laminin ␥3 chain specific probes (Fig. 4). The strongest signals were obtained from testis, kidney, and lung, with low level of expression in brain, skeletal, and heart muscle and intestines. Both probes revealed two transcripts of about 6 and 8 kb. The source for the different mRNAs remains un- A, staining is seen in a blood vessel identified as either an afferent or efferent arteriole (arrow). G, glomerulus. B, a strong staining is observed in an arteriole adjacent to a glomerulus (arrow). Weak signals are seen in some glomerular cells (small arrow). C and D, strong staining is also observed in a medium size artery of the cortex (C), as well as in capillaries located in between renal tubuli (arrows in D). E, negative control using a sense probe.
known. 3Ј-rapid amplification of cDNA ends using a primer at position 4748 -4771 of the full-length laminin ␥3 chain cDNA yielded an about 1.3-kb-long product (not shown). In the case of the ␥1 chain, the 3Ј-UTR contains several polyadenylation signals that probably are alternatively used (67) and thus account for the two mRNAs observed (8).
To further characterize the tissue compartments expressing the ␥3 chain, in situ hybridization experiments were performed on adult kidney and testis tissues. In the kidney, strong signals were seen in either afferent or efferent arterioles of the glomeruli (Fig. 5A), and in endothelial cells of other arteries (Fig. 5, B and C), while cells of the actual glomeruli only showed modest staining above background (Fig. 5, A and B). Tubular or interstitial cells did not exhibit signals, but capillaries located between tubular structures were clearly positive for expression (Fig. 5D). In the testis, the strongest signals were seen in the Leydig cells located between the seminiferous tubuli (Fig. 6, A-C). The sertoli and spermatogenic cells were negative. Staining with the sense probe did not yield any distinct staining (Fig. 6D).
The expression pattern of the laminin ␥3 chain overlaps with that of the other ␥ chains. In kidney, several cell types express the ␥1 chain (12,68). The ␥2 chain expression is confined to the epithelial cells (12,68). In testis, the laminin ␥1 chain is found in the seminiferous tubules (69). However, little is known about the expression of laminin ␥ chains by Leydig cells.
In general, the results of the in situ hybridization analyses indicate that endothelial cells of arteries and capillaries, as well as interstitial cells in certain tissues, express the ␥3 chain. However, expression is apparently not significant in all endothelial cells, as tissues rich in blood vessels and capillaries, such as cardiac and skeletal muscles, do not exhibit strong expression, e.g. as is the case for the laminin ␣4 chain (22,44,45). Further studies are needed to show in detail the expression pattern in epithelia. The present results indicate that, with the exception of Leydig cells in the testis, mesenchymal cells are not major producers of laminin containing the ␥3 chain.
By the virtue of their coexpression with the ␥3 chain, the ␣2, ␣3, ␣4, and ␣5 chains could take part in the assembly of a ␥3 chain containing laminin (22,30,38,42,44,45). Since both the ␤1 and the ␤2 chains are also expressed at these sites (34,35,38,42,43), at least eight hitherto uncharacterized laminin isoforms could collectively reside in these tissues. These tentative isoforms are likely to integrate into the basement membrane protein meshwork via the conserved networking promoting domains VI (49,65) and III(LE4) (66,70). FIG. 6. Expression of the laminin ␥3 chain in adult mouse testis as determined by in situ hybridization. A, staining with the antisense probe is seen between the seminiferous tubules (arrow). Sertoli or spermatogenic cells are negative. S, seminiferous tubule. B, schematic illustration of the section in A, showing several seminiferous tubuli (S) containing Sertoli and spermatogenic cells located in the tubules above the basement membrane. Leydig cells are located in the interstitial space filling the interstitium between adjacent tubules (dark pink areas). C, magnification of the box in A shows that the expression is present in Leydig cells. D, staining with the negative control sense probe.