2.2. Non-fibril forming collagens

Grouped under the non-fibrillar collagens are all collagens that fall outside the category of the fibril-forming collagens. Except for the occurrence of one or more interruptions in the collagenous sequence of all these collagens, this group is structurally and functionally very heterogeneous. Several subfamilies can be distinguished, though, according to similarities in the domain organizations, supramolecular structures, and types of extracellular networks they form: A) collagens that are located on the surfaces of fibrils and are called fibril-associated collagens with interrupted triple helices (abbreviated as FACITs and including the structurally related collagens, types IX, XII, XIV, XVI, and XIX); B) collagens that form hexagonal networks (types VIII and X); C) basement membrane (BM) type IV collagen; D) type VI collagen that forms beaded filaments; D) type VII collagen that forms anchoring fibrils of BM; E) collagens with transmembrane domains (types XIII and XVII; and F) the family of type XV and XVIII collagens. (For reviews, see Vuorio & de Crombrugghe, 1990; van der Rest & Garrone, 1991; Burgeson & Nimni, 1992; Hulmes, 1992; Mayne & Brewton, 1993; Fukai et al., 1994; Pihlajaniemi & Rehn, 1995; Prockop & Kivirikko, 1995; Myllyharju & Kivirikko, 2001). Furthermore, the newly identified collagen types XX and XXIII are likely to belong to the non-fibrillar group of collagens (Myllyharju & Kivirikko, 2001).

Although the association with collagen fibrils has only been shown for type IX, XII, and XIV collagens, collagens XVI and XIX are also classified as FACITs, based on sequence homology (for reviews, see Fukai et al., 1994). Type IX collagen molecules consist of three collagenous (COL) and four noncollagenous (NC) domains, with the COL domains representing most of the molecule. Type XII and XIV collagens consist of two COL domains, while a substantial part of these molecules is noncollagenous. The association with collagen fibrils occurs via COOH-terminal COL domain(s), whereas one COL and one NC region project out of the fibril. This latter region is believed to be functionally important in bridging the collagen fibers with other matrix components. Type IX collagen is covalently cross-linked on the surface of type II collagen fibrils (Wu et al., 1992) and is expressed in cartilage and other noncartilagous tissues containing this collagen. Types XII and XIV collagens are found predominantly in tissues containing type I collagen. Collagen types IX, XII, and XIV contain glycosaminoglycan side chains in some tissues and are thus proteoglycans (Bruckner et al., 1985; Huber et al., 1986; McCormick et al., 1987; Watt et al., 1992). The collagenous regions of type XVI and XIX collagens are very interrupted, as they consist of ten and five COL domains, respectively (Pan et al., 1992; Inoguchi et al., 1995; Sumiyoshi et al., 1997). Type XVI collagen is widely distributed throughout the body. It is found close to the cells, but is not associated with collagen fibrils (Grassel et al., 1996; Lai & Chu, 1996). The type XIX collagen protein also has a widespread distribution in many BM zones (Myers et al., 1997). Specific roles in development, and especially in myogenesis, have been suggested for type XIX collagen (Sumiyoshi et al., 1997; Myers et al., 1999; Sumiyoshi et al., 2001 and see below).

The structurally homologous collagen types VIII and X consist of a single COL domain flanked by NC domains (for reviews, see Fukai et al., 1994). Both assemble into hexagonal lattices - type VIII collagen e.g. in the specialized BM of the corneal endothelium called Descemet´s membrane and type X collagen in the hypertrophic zone of cartilage. Based on the highly organized supramolecular assemblies of these collagens, a structural role has been suggested for them. Type VIII may play a role in resisting compression for example in the eye, while type X collagen may be involved in the formation of a scaffold during replacement of cartilage by bone in the process of endochondral ossification. This latter hypothesis has indeed gained support by the consequences of mutations in the type X collagen gene in murine models and in humans (reviewed by Chan & Jacenko, 1998, and see later). Mutations in the type VIII collagen gene encoding the α2 chain has been recently associated with two forms of inherited disorders of cornea (Biswas et al., 2001). In addition, the consequences of the mutations in both of these collagen types also support their roles in the determination of cellular phenotypes, in cell migration, and differentiation (Shuttleworth et al., 1997; Sutmuller et al., 1997).

Type IV collagen molecules are the major constituents of all BMs. They aggregate into network-like structures and thus provide an essential scaffolding for cells and other BM components to attach. The α-chains have a central long collagenous region, flanked by very small N-terminal and longer C-terminal NC domains. The α-chains are highly glycosylated. The six different α-chains differ considerably with respect to their tissue distribution, the α1(IV) and α2(IV) chains being ubiquitous components of all BMs and α3(IV)- α6(IV) having a more restricted distribution (Hudson et al., 1993). In addition to the obvious structural role of type IV collagen in providing integrity to tissues and in serving as molecular sieves, e.g. in glomerular filtration, the NC domains of α2(IV), α3(IV), and α6(IV) have recently been shown to be implicated in the inhibition of angiogenesis (Petitclerc et al., 2000; Maeshima et al., 2000, 2001).

The collagenous region of type VI collagen is short, with NC domains comprising most of the molecule. Type VI collagen molecules aggregate to form special fibrils, which resemble beaded filaments, and are present in all connective tissues close to the BM and cells. The important bridging function of type VI collagen (between BMs and matrix) has recently gained evidence, as mutations in its gene were identified in Bethlem myopathy (Jöbsis et al., 1996) and as similar symptoms were reproduced in Col6a1-deficient mice (Bonaldo et al., 1998, and see below).

Type VII collagen has the largest triple-helical region among the collagens flanked by a large tridentate N-terminal and smaller C-terminal NC domains. Type VII collagen forms the anchoring fibrils, which attach the BM of the epithelium to the underlying stroma, and are present beneath the squamous epithelium of the skin, for example. The importance of anchoring fibrils is attested to by the observation that alterations in their number or morphology results in the disruption of the dermal-epidermal connection. This leads to one form of blistering disease of skin, dystrophic epidermolysis bullosa (EB), which is caused by mutations in type VII collagen (see below).

Although not homologous in their primary structures, the type XIII and XVII collagens belong to the same subfamily. They both contain transmembrane domains in their noncollagenous N-terminus and are located in the plasma membrane with a cytoplasmic N-terminal end and an extracellular C-terminal collagenous region (for reviews, see Pihlajaniemi & Rehn, 1995). Type XIII collagen is mostly collagenous, consisting of three COL domains flanked by short NC domains. The collagenous region of type XVII collagen is highly interrupted, consisting of 15 COL domains flanked by a large N-terminal and short C-terminal NC domain. A remarkable feature of type XIII collagen is the complex alternative splicing of its transcripts, which affects both NC and COL domains. The type XIII collagen triple helix formation has recently been shown to occur from the N- to C-terminus and thus in the opposite orientation to that of the fibrillar collagens (Snellman et al., 2000). Type XIII collagen localizes to adhesive structures and along the BM zone in cells and tissues, and has been suggested to have a role in mediating cell-matrix adhesion (Hägg et al., 2001). Type XVII collagen, also known as BP180 and BPAG2, is a major component of the hemidesmosomes, structures involved in the adherence of stratified epithelia to the underlying BM, and one of the two autoantigens implicated in the aetiology of autoimmune disease, bullous pemphigoid (Pulkkinen & Uitto, 1998).

In addition, there are four α-chains for which the complete cDNA sequences have been characterized, and that could be included in the collagen superfamily. These include one chain that is homologous with fibril-forming collagens, two FACIT collagen-like chains, and one chain homologous with type XIII collagen (Myllyharju & Kivirikko, 2001). As their inclusion is not well founded and they are not well-characterized proteins, they are not further discussed here.