Collagen IV tx3, c 4, and ct5 Chains in Rodent Basal Laminae: Sequence, Distribution, Association with Laminins, and Developmental Switches

Collagen IV is a major component of vertebrate basal laminae (BLs). Studies in humans have revealed a family of genes encoding or-or6 collagen IV chains and implicated ot3-ot6 in disease processes (Goodpasture and Alport syndromes and diffuse leiomyomatosis). To extend studies of these components to an experimentally accessible animal, we cloned cDNAs encoding partial collagen t~3, u4, and ot5(iv) chains from the mouse. Ribonuclease protection assays showed that all three genes were expressed at highest levels in kidney and lung; ot5(iv) was also expressed at high levels in heart. We then made antibodies specific for each collagen IV chain. Immunohistochemical studies of several tissues revealed many combinations of collagen IV chains; however, t~3 and a4 (IV) were always coexpressed, and only appeared in BLs that were ot5(iv) positive. The ot3-ol5(iv) chains were frequently but not exclusively associated with the S (/32) chain of laminin, as were the or, 2 (IV) collagen chains with laminin B1 (/~1). An analysis of developing rat kidney BLs showed that newly formed (S-shaped) nephrons harbored collagen od and t~2(iv) and laminin B1; maturing (capillary loop stage) BLs contained collagen cd-ot5(iv) and laminin B1 and S-laminin; and mature glomerular BLs contained mainly collagen ot3-ot5(iv) and S-laminin. Thus, collagen otl and ot2(IV) and laminin B1 appear to be fetal components of the glomerular BL, and there is a developmental switch to collagen ot3-ot5(iv) and S-laminin expression. M AN Y cells in both vertebrates and invertebrates bear a thin, insoluble layer of extracellular matrix called a basement membrane or basal lamina (BL) ~. The major components of most BLs are two multimeric glycoproteins, collagen IV and laminin (reviewed in Rohrbach and Timpl, 1993). Each of these components was initially isolated from tumor tissues as single trimeric species: (od)2(t~2)l for collagen IV, and A-B1-B2 (also called al-/~l-~/1; Burgeson et al., 1994) for laminin. Recently, however, diversity has been revealed in the subunits that make up these trimers, with the discovery of the ct3, or4, ct5, and t~6(IV) collagen chains (reviewed in Hudson et al., 1993) and the S, M, K, and B2t (also called ~2, t~2, c~3, and 3'2) laminin subunits (Burgeson et al., 1994). Moreover, BLs are now know to vary in the collagen IV and laminin isoforms they contain (Kleppel et al., 1989a, b; Sanes et al., 1990; Engvall et al., 1990). Thus, all BLs appear to contain some Address all correspondence to J. R. Sanes, Department of Neurobiology, Washington University School of Medicine, 660 S. Euclid, St. Louis, MO 63110. Fax: (314) 362-3446. 1. Abbreviations used in this paper: BL, basal lamina; BS, Bluescript; GBM, glomerular basement membrane; NC1, noncollagenous domain 1; nt, nucleotide; RNase, ribonuclease; RT, reverse transcript; TBM, tubular basement membrane. collagen IV and some laminin, but in different isoform combinations, suggesting that the functional diversity of BLs arises in part from the particular collagen IV and laminin isoforms they contain. Laminin subunit diversity was first demonstrated by the discoveries of S-laminin, a homologue of the B1 subunit (Hunter et al., 1989b), and merosin M, a homologue of the A subunit (Ehrig et al., 1990). More recently, K-laminin/nicein/kalinin/epiligrin laminin variants have been identified in subsets of epithelial BLs (Marinkovich et al., 1992; Kallunki et al., 1992). The availability of numerous immunological and nucleic acid reagents is rapidly leading to an understanding of how these individual laminin subunits can be assembled. In tissues and in vitro, the B2 subunit associates with either the A or M heavy subunit plus either the B1 or S subunit, producing a heterogeneous family of laminin trimers (Engvall et al., 1990; Green et al., 1992). At the cellular level, immunohistochemical studies have shown that most BLs contain either A or M, either B1 or S, and B2. For example, renal glomerular basement membrane (GBM) contains A, S, and B2, whereas extrasynaptic muscle BL contains M, B1, and B2 (Sanes et al., 1990). For the collagens IV, in contrast, details of chain assembly have been difficult to determine because the ot3-a6(IV) chains have been studied primarily in the context of diseased © The Rockefeller University Press, 0021-9525/94/11/879/13 $2.00 The Journal of Cell Biology, Volume 127, Number 3, November 1994 879-891 879 on A uust 8, 2017 jcb.rress.org D ow nladed fom human tissue. The u3(iv) collagen chain was discovered as the antigen in Goodpasture syndrome, an autoimmune nephritis which targets the GBM in kidney and the alveolar BL in lung (Butkowski et al., 1987; Sans et al., 1988). Attempts to purify tx30V) resulted in the discovery of the collagen ~x4(IV) chain, which appears to be associated with it (Gunwar et al., 1990; Johansson et al., 1992). This work led to the hypothesis that either the collagen ix3 or a4(IV) chain was mutated in X-linked Alport syndrome, a hereditary glomerulonephritis known to involve defects in glomerular collagens. In fact, analysis of the mutant allele revealed that it encoded yet another chain, ct5(IV) (Hostikka et al., 1990; Barker et al., 1990; Tryggvason et al., 1993). Most recently, an tx6(IV) gene was identified next to the tx5(IV) gene and shown to be mutated in several X-linked cases of Alport syndrome in which mutations in ct5(IV) could not be detected (Zhou et al., 1993; Oohashi et al., 1994). In addition, deletions that removed parts of both a5 and tx6(IV) were found when Alport syndrome was accompanied by diffuse leiomyomatosis, a benign proliferation of smooth muscle. The juxtaposed or5 and ot60V) genes are arranged in a headto-head orientation on the X chromosome, as are the coregulated txl and Ot2(IV) genes on human chromosome 13 (Poschl et al., 1988). The o13 and tx4(IV) genes both map to human chromosome 2q35-37 and may be similarly arranged (Morrison et al., 1991; Turner et al., 1992; Mariyama et al., 1992b; Kamagata et al., 1992). Immunohistochemical studies have shown that the ot3ot5(IV) collagen chains have a restricted distribution in human tissues. For example, they are highly enriched in the GBM (consistent with the Goodpasture and Alport syndrome phenotypes) and are also found in a subset of tubular basement membranes (TBMs) (Kleppel et al., 1989; Hostikka et al., 1990; Sanes et al., 1990; Kleppel et al., 1992; Hudson et al., 1992). (No studies on the distribution of ot6(IV) have yet been reported.) On the other hand, the cd and ix2 chains of collagen IV are abundant in all TBMs and in the glomerular mesangial matrix, but are scarce in GBM (Kleppel et al., 1989; Kashtan and Kim, 1992; Sanes et al., 1990; Kleppel et al., 1992). In human muscle, the ot3 and ot4(IV) collagen chains are restricted to the synaptic basal lamina at the neuromuscular junction, while the txl and c~2(IV) chains are found extrasynaptically (Sanes et al., 1990). Interestingly, in both renal glomerular and muscle synaptic BLs the substitution of collagen oe3-a5(IV) chains for od and tx20V) is accompanied by a substitution of the S subunit of laminin for the B1 subunit. Taken together, these results raise several questions concerning BL structure and function: Are there any general rules governing the patterns of collagen IV subunit expression in BLs? Is there a special association between collagen od-o~2(iv) and laminin B1, or between collagen o~3-o~5(IV) and S-laminin? Do collagens tx3-cz5(IV) have special roles in the GBM or in synaptic BL? Are particular BL isoforms (such as the linked pairs of collagen genes) coregulated during development? Here, to begin to address these questions, we have cloned cDNAs encoding partial collagen ot3-ot5(IV) chains from the mouse, prepared recombinant proteins from the cDNAs, and generated antibodies to the proteins. With these reagents, we performed RNase protection and immunohistochemical studies to analyze the expression patterns of the collagen IV chains in rodents. Of particular interest is the finding that the complement of collagen IV and laminin chains in the GBM changes systematically as development proceeds. Materials and Methods Polymerase Chain Reaction The PCR was used to synthesize chain-specific collagen IV probes for screening mouse eDNA libraries. For ct3 and c~4(IV) collagen, bovine kidney poly A + RNA (Clontech, Palo Alto, CA) was reverse transcribed and amplified using the GeneAmp RNA PCR Kit (Perkin-Elmer Cetus, Norwalk, CT) with primers based on the published bovine sequences (Morrison et ai., 1991; Mariyama et al., 1992a). Primers were: c~3(IV) sense, 5AA CCI"GGAGACACTGGACCACCTGC 3'; ct3(IV) antisense, 5'GTGCTTGCCCAGCACCCTCCGAAC 3'; ot4(IV) sense, 5'CCTGGATACCTCAGTGGCTTCCTCC 3'; and t~4(IV) antisense, 5'CAGGAACGGTGCGGCTCTGAAATCC 3'. Thermal cycler conditions were: 95°C, I rain; 62°C, 1.5 rain; 72°C, 2 rain (+ 2 s/cycle), 33 cycles. PCR products were not visible by agarose gel electrophoresis after 33 cycles, so 5% of the sample was reamplified for 17 cycles, after which products of the expected length were detected. For collagen tx5(IV), we used adult mouse lung total RNA for reverse transcript (RT)-PCR with degenerate primers based on the published human sequence (Hostikka et al., 1990). The primers were: sense, 5AA(AG)GGNCA(AG)AG(CT)AT(ACT)CA(AG)CC 3'; antisense, 5'CTATC(GT)(CT)TTCAT(AG)CANAC(CT)TG(AG)CA 3'. Thermal cycler conditions were three cycles of 95°C, 1 rain; 51°C, 2.5 rain; 72°C, 2.5 rain, followed by 31 cycles of 950C, 1 rain; 55°C, 2.25 rain; 720C, 2.5 rain (+1 s/cycle). PCR products were ligated into the pCR II vector using a TA Cloning Kit (Invitrogen Corp., San Diego, CA) and analyzed by restriction enzyme digestion or sequencing. cDNA Library Screening The collagen IV fragments were liberated from the pCR II vector by digestion with EcoRI and isolated by agarose gel electrophoresis. The fragments were 32p-dCTP labeled with a Random Primed DNA Labeling Kit (Boehringer Mannheim Bit)chemicals, Indianapolis, IN). The bovine t~3 and ¢x4(IV) collagen fragments were used to screen a hgtll mouse kidney eDNA library (Clontech) at low stringency. Hybridization conditions were: 30% formamide, 900 mM NaCl, 90 mM sodium citrate, 50 mM NaPO, (pH = 6.5), 0.25 % nonfat dry milk, at 42°C overnight. The mouse collagen ct5(IV) fragment was used to screen a hgtll mouse muscle eDNA library (prepared and provided by Mafia J. Donoghue in our laboratory) as above, except in 50 % formamide. The inserts of hybridizing phage were subcloned into Blue, script (BS) II SK+ (Stratagene Cloning Systems, La Jolla, CA) and sequenced with a Sequenase 2.0 Sequencing Kit (United States Biochemical Corp., Cleveland, OH). Probes for RNase Protections The collagen cd(IV) riboprobe was derived from the plasmid pCIV-l-C87, a eDNA clone from an Engelbreth-Holm-Swarm library (Wood et al., 1988; obtained from the American Type Culture Collection, Rockville, MD). Its 676-bp StyI fragment was blunted, subcloned into the EcoRV site of BS II SK+, cut with BstElI, and transcribed with T7 RNA Polymerase to synthesize a 279-nucleodde (nt) probe that produced a 224-nt protected band. To synthesize the a3(IV) collagen riboprobe, the 5' EcoRI fragment (nt 1-552) in BS H SK+ was cut with StyI (nt 187) and transcribed with T7 RNA Polymerase to make a 424-nt probe which produced a 365-nt protected band. The tx4(IV) collagen probe, also in BS II SK+, was cut with StyI (nt 694) and transcribed with T7 to make a 311-nt probe that was protected to the EcoRI site (nt 946) to produce a 252 nt band. For the collagen ~x5(IV) probe, the original RT-PCR product, cloned into the pCR II vector, was used. That plasmid was cut with AccI (nt 12 of the sense PCR primer) and transcribed with SP6 RNA Polymerase to produce a 461-nt probe and an ,'v381-nt protected band. RNA Isolation and Analysis RNA was prepared from mouse tissues by acid guanidinium phenol/chloroform extraction (Chomczynski and Sacchi, 1987). Tissues were disrupted in the guanidinium solution with a Polytron. For RNase protection assays, 7/~g total RNA were hybridized with 1-5 x 109 probe molecules. SingleThe Journal of Cell Biology, Volume 127, 1994 880 on A uust 8, 2017 jcb.rress.org D ow nladed fom stranded RNA was digested with 1 U/ml RNase I"1 (United States Biochemical Corp.) and 0.4 ng/ml RNaseA (Sigma Chemical Co., St. Louis, MO). For details, see Miner and Wold (1991). Production of Fusion Proteins and Antisera To produce collagen c~3-ct5(IV) proteins, fragments of their cDNAs coding for noncollaganous domain 1 (NC1) segments were cloned in frame into the proper pET-3 vector (Rosenberg et al., 1987), all of which contain a common short leader sequence. The a3(IV) collagen fusion protein contained the final 184 amino acids of c~3(IV) collagen, the t~4(IV) collagen fusion protein contained its final 185 amino acids, and the et5(1V) fusion protein contained amino acids 120-248 (see Fig. 1). The pET-3 expression constrncts were transformed into the BL21(DE3) host strain (Novagen, Inc., Madison, WI) and then grown and induced for protein expression according to the manufacturer's instructions. Induced bacteria from a 50 rnl culture were pelleted, solubilized in sodium dodecyl sulfate loading buffer containing dithiothreitol, boiled for 5 rain, and electrophoresed through a preparative 10% SDS-polyacrylamide gel (Sambrook et al., 1989). Proteins were visualized in the gel with 0.05 % Coomassie brilliant blue in water (Harlow and Lane, 1988) and excised with a razor blade. Gel slices were shipped to Cocalico Biologicals, Inc. (Reamstown, PA), where they were used to immunize rabbits. A second collagen ct4(iv) fusion protein, containing its final 151 amino acids, was used on Western blots (see Fig. 4) but not for immunization. Antibodies and Immunohistochemistry After initial characterization, one antiserum each to collagen c~3, ct4, and ct5(IV) was used for subsequent studies. The ct3 and ot4(IV) antisera initially recognized all three fusion proteins on Western blots, but subsequent experiments suggested that this was mainly due to reaction with the common l l-amino acid leader sequence. Thus, much of the cross-reactivity could be abolished by incubating diluted antisera with inclusion bodies containing the pET-3 leader sequence fused to a portion of S-laminin (pET-36; Hunter et al., 1989a), which is unrelated to collagen IV. Most of the remaining cross-reactivity was removed by adsorption with inclusion bodies containing the noncognate collagen IV fusion proteins, followed by cantrifugation. Goat antiserum to human collagen cd, c~2(IV) was purchased from Southern Biotechnology Associates, Inc. (Birmingham, AL). This antiserum reacted with mouse, rat, and human proteins. The collagen cd, ct2(IV) monoclonal M3F7 (Foellmer et al., 1983), which reacts with rat and human proteins, was purchased from ICN Immunochemicals (Lisle, IL). Polyclonal (GPS) and monoclonal (C4) antibodies to S-laminin and monoclonal antibodies to laminin 131 ((221) and 132 (D18) have been described previously (Sanes et al., 1990). Rabbit anti-human c~5(IV) collagen peptide antiserum was a kind gift of M. Kleppel (University of Minnesota Medical School, Minneapolis, MN). Mouse and rat tissues were frozen in isopentane and sectioned at 4-8 pm on a cryostat. Human muscle biopsy material was provided by Kenneth Kaiser and Michael Brooke, then of the Department of Neurology, Washington University Medical Center (St. Louis, MO). In experiments involving the rabbit collagen IV antibodies, frozen sections were fixed in 100% ethanol for 5 rain at -20°C, rinsed in PBS, and treated with 6 M urea-O.1 M glycine, pH 3.5, for 1 h at 4°C, before antibodies were applied. The acid-urea treatment effectively exposed hidden collagen IV epitopes (Yoshioka et al., 1985), but it also greatly increased background in rodent (but not human) tissues. This background was reduced but not eliminated by applying antibodies in PBS containing 5-10% nonfat dry milk. Fluoresceinand rhodamine-conjugated secondary antibodies to rabbit, goat, mouse, and guinea pig were obtained from Boehringer Mannfieim Biochemicals, Sigma Chemical Co., or Cappel/Organon Teknika (Durham, NC), and were also diluted in milk and applied for 1-2 h. In cases where muscle sections were not treated with urea-glycine, rhodamine-c~-bungarotoxin was added with the second antibody to label neuromuscular junctions.