© 2001 Nature Publishing Group http://structbio.nature.com news and views Mapping out the basement membrane © 2001 Nature Publishing Group http://structbio.nature.com Robert C. Liddington Recent structural studies suggest that not only the protein components but also the nature and location of the protein–protein interaction surfaces in basement membranes are highly conserved throughout the metazoa, suggesting that their precise ultrastructure will be invariant, and reflecting their critical role in early development. The cells of multicellular animals (metazoa) secrete and shape around them a network of proteins called the extracellular matrix (ECM). One specialized element of the ECM is called the basement membrane, a thin (<1,000 Å) sheet of protein and proteoglycan1. The recent sequencing of the Caenorhabditis elegans genome shows that the basement membrane is highly conserved throughout metazoan evolution, much more so than the rest of the ECM2,3. In the simplest of metazoa such as hydra, the basement membrane divides the inner layer of cells (endoderm) from the outer layer (ectoderm). In more complex organisms, it provides a mechanical support and barrier in blood vessels and other tissues, and regulates the growth, differentiation and migration of the adherent cells. For example, epithelial cells must be anchored to the basement membrane to survive; if contact is lost, they undergo programmed cell death (apoptosis) in a process that has been called anoikis, from the Greek word for homelessness4. Until recently, our structural knowledge of the basement membrane was limited to electron microscopic resolution. Unraveling the atomic resolution network of interactions and how they are regulated within the basement membrane is still in its infancy, but the recent work of Hohenester, Timpl and colleagues has started to shed some light. On page 634 of this issue of Nature Structural Biology, these authors describe the latest atomic resolution structure of a basement membrane component5, a ligand-binding module of nidogen, one of four proteins common to all basement membranes. They go on to show that a surface of the module that is conserved from the earliest metazoans provides a high affinity binding site for another basement membrane component, perlecan. These results, together with earlier work from the same groups, suggest a highly conserved structure for the basement membrane. Fig. 1 Cartoon diagram of nidogen. Nidogen contains three globular domains: G1, G2 and G3. G1 is of unknown structure and function. G3 has been predicted to fold into a six-bladed β-propeller7. G1 and G2 are connected by a flexible linker, whereas G2 and G3 are connected by a rigid rod that includes EGF-like modules. The structure of G2 is described in the text. perlecan and nidogen, which form a complex web of interactions. The nonfibrillar type IV collagen is a very long (∼2,000 Å) molecule consisiting of a collagen triple helix with N- and C-terminal globular domains. Laminin is a large (900 kDa) trimeric cruciform molecule that is long enough (∼700 Å) to span the entire width of the basement membrane. Perlecan is a five-domain protein (∼400 kDa) that has three large heparan sulfate chains attached to its N-terminus. Nidogen is the smallest of this quartet (150 kDa). One role of nidogen is to provide a link between laminin and type IV collagen. In C. elegans nidogen has also been shown to be required for the correct positioning of longitudinal nerves6. The structure of the second globular domain (G2) of nidogen revealed in this issue5 comprises an EGF-like domain fused to the top of a β-barrel (Fig. 1). The barrel is 45 Å tall and has a simple shape but a complex topology, with an α-helix inserted into the center of the barrel. This fold has been seen before, in the green fluorescent protein (GFP) from Aequorea victoria, a brightly luminescent jellyfish8. Although there was no detectable sequence identity (∼10%), the structural Basement membrane components homology is outstanding (root mean All basement membranes have four major square (r.m.s.) deviation of 2.5 Å for 195 constituents: type IV collagen, laminin, Cα atoms). The authors reasonably sugnature structural biology • volume 8 number 7 • july 2001 gest that the two proteins have a common ancestor, as it is hard to imagine such a complex fold evolving twice. Beyond the striking similarity in fold, however, there are no other similarities. Nidogen is not fluorescent and none of the fluorophore residues are conserved. GFP is not an extracellular protein: it is found in intracellular particles called lumisomes that are 2,000 Å in diameter. Sequence searches with either GFP or nidogen generate only closely related members of each family. Jellyfish are amongst the simplest of the metazoa, on one of the earliest branches of the metazoan tree. It is very likely that jellyfish have a bona fide nidogen, since the nematode C. elegans, on the next evolutionary branch, has one. So it seems likely that nidogen and GFP have diverged from a common ancestor, and the structural similarity is just a curiosity at this point. As new structures are solved, these unexpected structural homologies are becoming increasingly common. One of the main legacies of the structural genomics initiatives may be to reveal these evolutionary relationships that are not obvious by sequence comparisons. Interactions in the basement membrane The assembly of basement membranes occurs in part in a spontaneous fashion — 573 © 2001 Nature Publishing Group http://structbio.nature.com © 2001 Nature Publishing Group http://structbio.nature.com news and views laminin and collagen both self-assemble to form networks in vitro, the former in a Ca2+-dependent fashion; nidogen and perlecan spontaneously bind and bridge between laminin and collagen IV. Nevertheless, the active participation of cells is required for proper assembly and localization of the basement membrane, and this is dependent on two classes of transmembrane receptors, the integrins and the dystroglycan–dystrophin complex, both of which bridge and convey signals to and from the actin cytoskeleton9. Although the data are sketchy, it may not be too early to review what is known about these interactions and speculate on how they are regulated during biological processes such as development, wound healing, and angiogenesis, which require modeling or remodeling of the basement membrane. In 1996 Timpl, Huber and colleagues described the first crystal structure of a laminin module, comprising three tandem repeats of an EGF-like domain (LE domain)10. By mutagenesis, they showed that the binding site for nidogen involves a short loop containing the conserved sequence Asp-Pro-Asn-Ala-Val in a single LE domain11. This high affinity interaction (Kd = 0.5 nM) is critically dependent on three residues in a conserved motif — Asp, Asn and Val. These data suggest a highly specific interaction in which a ridge on laminin fits perfectly into a groove on the complementary region of nidogen. In 1999 Hohenester, Timpl and colleagues described a second laminin module, the LG (for laminin globular) domain12. This module connects the basement membrane to the dystroglycan–dystrophin complex. A metal ion site found on the surface, coordinated by conserved acidic groups on the domain surface and a sulfate ion, was suggested to resemble the 574 interaction of a sulfated dystroglycan carbohydrate, in which the Ca2+ ion forms an integral part of the interaction — that is, a metal bridge of the kind seen in integrin–ligand interactions13. The latest structural study by Hohenester and colleagues5 suggests a more conventional kind of protein–protein interaction. The authors show that a patch on the barrel surface of the nidogen G2 domain is strikingly conserved in all metazoan nidogens. They further show by mutagenesis that this surface is very likely the binding surface for one of the immunoglobulin (Ig)-like domains of perlecan. This is a high affinity interaction and appears to involve a large surface on the barrel, which presumably binds a concave surface on the Ig domain. Is the binding of nidogen to perlecan regulated? The G2 domain binds transition metals strongly, and metal binding inhibits perlecan binding. Although no metal ion was found in the crystals (zinc was included in the crystallization buffer), owing perhaps to the low pH of crystal growth, the authors suggest that a loop adjacent to the binding surface has the potential for zinc binding. Metal ion binding could provide a credible mechanism of inhibition by distorting the binding surface. Curiously, Zn2+ has been shown to enhance the interactions between laminin and both nidogen and collagen IV, perhaps via the formation of zinc finger-like structures14. Could metal ions play a regulatory role in the assembly of basement membranes? There are no data on changes in extracellular Zn2+ concentration. However, Ca2+ and Mg2+, which are normally present at a high and constant level (∼mM), are known to undergo rapid changes in cation concentrations at wound sites, for example, where basement membrane remodel- ing is required15. These circumstances could potentially regulate the protein– protein interactions discussed above, if the metal dissociation constants are commensurate. Perspective Now that high resolution structures of parts of the basement membrane components are available, the exciting challenges for the future clearly lie in the study of these parts in the context of the full length proteins, as well as in the protein complexes within the basement membrane. Integration of such information into lower resolution images of intact basement membranes derived from electron microscopy should allow detailed models of these important ECM elements to be built. Robert C. Liddington is Director of the Program on Cell Adhesion, The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, California 92037, USA. email: [email protected] 1. Yurchenco, P.D. & O’Rear, J.J. Curr. Opin. Cell Biol. 6, 674–81 (1994) 2. Hutter, H. et al. Science 287, 989–94 (2000) 3. Hynes, R.O. & Zhao, Q. J. Cell. Biol. 150, F89–96 (2000) 4. Frisch, S. & Ruoslahti, E. Curr. Opin. Cell Biol. 9, 701706 (1997). 5. Hopf, M., Göhring, W., Ries, A., Timpl, R. & Hohenester, E. Nature Struct. Biol. 8, 634–640 (2001) 6. Kim, S. & Wadsworth, W.G. Science 288, 150–154 (2000) 7. Springer, T.A. J. Mol. Biol. 283, 837–862 (1998) 8. Ormö, M. et al. Science 273, 1392–1395 (1996). 9. Lohikangas, L., Gullberg, D. & Johansson, S. Exp. Cell Res. 265, 135–144 (2001). 10. Stetefeld, J., Mayer, U., Timpl, R. & Huber, R. J. Mol. Biol. 257, 644–657 (1996). 11. Pöschl, E. et al. EMBO J. 15, 5154–5159 (1996) 12. Hohenester, E., Tisi, D., Talts, J.F. & Timpl, R. Mol. Cell 4, 783–792 (1999) 13. Emsley, J., Knight, C.G., Farndale, R.W., Barnes, M.J. & Liddington, R.C. Cell 101, 47–56 (2000) 14. Ancsin, J.B. & Kisilevsky, R. J. Biol. Chem. 271, 6845–6851 (1996) 15. Brown, E.M., Vassilev, P.M. & Hebert, S.C. Cell 83, 679–682 (1995) nature structural biology • volume 8 number 7 • july 2001
© Copyright 2026 Paperzz