Mammalian Collagen IV

Evolution of Assembly Question

In 1981, the first network model for the organization of collagen IV was reported (Timpl et al., 1981). Using rotary shadowing technique, it was demonstrated that individual collagen IV molecules appeared as 390-nm long strands with a globular domain which was termed noncollagenous domain (NC1). Two of such molecules could form a dimer through interaction of their NC1 domain. In protein fractions that had been solubilized by pepsin digestion, the NC1 domains disappeared and the remaining molecules mainly consisted of four triple-helical strands joined together through the amino terminal 7S domains. Thus it was demonstrated that collagenous network of BMs consists of collagen IV molecules with two different interacting sites; a 7S domain which connects four molecules and the NC1 domain which connects two molecules. Soon after, it was shown that the NC1 domain consisted of ∼30,000 molecular weight polypeptide located at the carboxyl end of the molecule and contained intramolecular disulfide bridges (Fessler and Fessler, 1982).

Until mid 80s, only two α-chains of collagen IV were known to exist. However, based on chemical analysis of the GBM, the existence of minor but different collagen IV subunits was known for several years. Following digestion of the GBM with bacterial collagenase and chemical analysis of isolated NC1 domains, evidence for the existence of two novel α-chains, named α3(IV) and α4(IV), with different amino acid sequences than those derived from the classical α1(IV) and α2(IV) chains were reported (Butkowski et al., 1985, 1987; Wieslander et al., 1985). Based on partial peptide sequences derived from isolated NC1 domains of human GBM (Butkowski et al., 1990; Gunwar et al., 1990; Kleppel et al., 1986; Saus et al., 1988), partial cDNA clones were soon isolated by PCR amplification (Mariyama et al., 1992; Morrison et al., 1991a,b). Shortly, this was followed by characterization of the complete primary structure of the human α3(IV) and α4(IV) chains and their genes were localized to human chromosome 2 (Leinonen et al., 1994; Mariyama et al., 1994; Quinones et al., 1992; Turner et al., 1992). With the discovery of α5(IV) and α6(IV) chains by molecular cloning, the fundamental question of chain-specificity was revived. The ability of collagen IV chains to self-assemble through interactions at their NC1 domains had already been reported in early 80s (Timpl et al., 1981). Pepsin-treated molecules which lacked their NC1 domains were also shown to lack the ability to form network (Yurchenco and Furthmayr, 1984). Using rotary shadowing technique, a time-dependent reassembly of heat-denatured collagen IV molecules was clearly visualized, demonstrating a zipper-like refolding fashion of the triple-helical molecule starting from the C-terminal NC1 domains and proceeding toward the N-terminal ends (Dolz et al., 1988). Direct evidence that the chain specificity of collagen assembly is encoded by the NC1 domains was first demonstrated using purified NC1 hexamers isolated from BMs of different tissues (Boutaud et al., 2000). It was shown that acid-dissociated NC1 hexamers had the ability to reassemble back to native configuration with the original molecular stoichiometry. More interestingly, it was shown that hexamer formation from the α3, α4, and α5 NC1 monomers required the presence of all three NC1 domains, further suggesting that mutations in individual NC1 domains may interrupt the assembly of the α3α4α5 network. All these findings clearly highlighted the critical role of the NC1 domain in the assembly of collagen IV molecules, and suggested that the NC1 domains contain recognition sequences for chain selection.

Using detailed kinetics and crystal structure analysis, we have recently reported the molecular mechanism of chain-selection in the assembly of collagen IV (Khoshnoodi et al., 2006b). Using a surface plasmon resonance biosensor, the kinetic of interaction and assembly of isolated α1(IV) and α2(IV) NC1 domains were determined. The analysis showed that the α2(IV) NC1 domain had a prominent role in regulating the stoichiometry of chains and assembly of α1α1α2 hetero-trimer, by displaying a significantly higher affinity for heterodimer formation with α1(IV) NC1 domain than with itself. As the chain-specific assembly of the collagen IV molecules is initiated by the early interactions of NC1 domains in the ER, this characteristic differential affinity of certain NC1 domains for dimer formation may play a major regulatory role in determining the stoichiometry of chains in a particular collagen IV molecule. Moreover, based on data derived from the atomic structure of the NC1 hexamer, two prominent regions which together provide specificity for chain selection were identified. These regions comprised a 13-residue long β-hairpin motif involved in the domain-swapping mechanism stabilizing the intermolecular interaction between two NC1 domains, and a 15-residue long region with genetic hypervariability (VR3) which mediates chain-specific recognition between two NC1 domains (Fig. 2) (Khoshnoodi et al., 2006b). Thus it is most likely that chain-specific assembly of triple-helical collagen IV molecules in the ER is governed mainly by two factors: (i) specific recognition sites located on each NC1 domain and (ii) differential affinity of particular α-chains for dimer formation. Based on sequence identity, both α4(IV) and α6(IV) chains belong to the α2-like subtype and also occur only once in α3α4α5 and α5α5α6 trimers, respectively. Therefore, it is likely that these chains also play a regulatory role in the assembly of their corresponding triple-helical molecules.

Integrin Receptors

Integrins comprise a large number of structurally and functionally related transmembrane glycoproteins which are the major cell surface receptors for the ECM (Hynes, 1987, 1992, 2002). On the cell surface, they are present as noncovalent heterodimers consisting of an α- and a β-subunit. Each subunit has a large extracellular domain, a short transmembrane segment, and a C-terminal cytoplasmic tail with affinity for cytoskeletal proteins. Thus the term integrins reflect their ability to “integrate” cells into their immediate environment by connecting the cytoskeleton with the extracellular components. So far, 18α- and 8β-subunits have been identified in mammalian cells and are known to interact in a tissue-restricted manner to form 24 distinct family members (Humphries et al., 2006; Hynes, 2002). This diversity is further expanded by alternative splicing, posttranslational modifications, and interaction to other cell surface receptors and intracellular molecules (de Melker and Sonnenberg, 1999; Green et al., 1998; Porter and Hogg, 1998).

An important characteristic feature of integrins is their ability to recognize short peptide sequences within large ECM proteins. The classical example of integrin binding site is the Arg-Gly-Asp (RGD) sequence originally identified in fibronectin, which is a common ligand for at least eight different integrins (Ruoslahti, 1996). Although multiple RGD sequences exist within the triple-helical region of several collagen IV α-chains, several studies have shown that cell adhesion to collagen IV is RGD-independent (Herbst et al., 1988; Kim et al., 1994; Kramer and Marks, 1989). A possible explanation is that due to their triple-helical nature, these RGD sequences are not accessible to integrins. However, cell adhesion to collagen IV is predominantly dependent on triple-helical state of collagen IV, since reduction of disulfide bonds followed by heat-denaturation dramatically decreases cell binding and spreading (Aumailley and Timpl, 1986; Perris et al., 1993; Santoro, 1986).

The major collagen receptor include the β₁ subgroup of integrins, namely α₁β₁ and α₂β₁ (Aumailley and Gayraud, 1998; Leitinger and Hohenester, 2007; Paulsson, 1992). These two integrins bind both collagen IV and collagen I, however, with distinct specificity as follows: integrin α₁β₁ has a higher affinity for collagen IV, while α₂β₁ binds stronger to collagen I (Kern et al., 1993; Tulla et al., 2001a). It was shown that deletion of α₁β₁ integrin by the homologous recombination resulted in dramatic decrease in adhesion and migration of fibroblasts and smooth muscle cells to collagen IV substrate (Gardner et al., 1996), while the importance of α₂β₁ integrin was demonstrated by the decrease of adhesion and morphogenesis on collagen IV after using antisense mRNA (Keely et al., 1995). A major binding site for α₁β₁ and α₂β₁ integrins has been originally identified within a triple-helical cyanogen bromide-derived fragment CB3, located 100 nm away from the amino-terminus of the collagen IV (Fig. 3) (Vandenberg et al., 1991). The triple-helical conformation of this particular region of the molecule is stabilized through interchain disulfide bridges between the α1(IV) and α2(IV) chains. Using antibodies to CB3 fragment, it was shown that about 80% of cell adhesion to collagen IV was inhibited, suggesting that it comprises a major cell binding region. By affinity chromatography using immobilized CB3 fragments and cellular lysates, α₁β₁ and α₂β₁ were isolated as the only integrins eluted from the affinity column (Vandenberg et al., 1991). Using purified integrins and shorter tryptic fragments of CB3, one binding site for α₁β₁ integrin and two for α₂β₁ integrin were located in different but adjacent positions on CB3 fragment (Kern et al., 1993). Further refinement of the α₁β₁ recognition site was achieved by digestion of one of the tryptic peptides with thermolysin (Eble et al., 1993). Thus, the recognition site within CB3(IV) fragment was located within a conformational-dependent region, where the spatial vicinity of Asp461 on α1(IV) chain to Arg461 on α2(IV) chain showed to be critical for binding of α₁β₁ integrin. Finally, functional activity of this site for α₁β₁ integrin binding has been demonstrated directly using synthetic triple-helical peptide containing 12 amino acid residues (457–468) from both α1 and α2 chains of collagen IV stabilized with artificial cysteine knot (Renner et al., 2004).

Similar to α₁β₁, a conformational-dependent binding to collagen I and IV has been shown for α₂β₁ integrin (Knight et al., 1998). The α₂β₁ integrin plays an important role in platelet adhesion to collagens and hemostasis in the blood vessel wall (Sixma et al., 1997). It has been shown that a short sequence of GFOGER peptide (O being hydroxyproline) represents a binding site for the α₂β₁ integrin (Knight et al., 2000). The sequence represents a high-affinity binding site in both collagen I and IV, and is entirely dependent on the native triple-helical conformation. Replacement of phenylalanyl or glutamyl residues abolished the binding, indicating their important role for recognition. Triple-helical GFOGER peptide is a potent inhibitor of cell adhesion to collagen IV. Moreover, this sequence is present in a1 chain of CB3(IV) fragment (385–390), where it might represent one of the two binding sites for α₂β₁ integrin (Fig. 3).

In addition to α₁β₁ and α₂β₁, other integrins may also participate in cell binding to collagen IV. Cell adhesion to collagen IV mediated by α₃β₁ integrin was shown for small lung carcinoma cell line expressing low levels of other β₁ integrins (Elices et al., 1991). In addition, α₃ subunit antisense oligonucleotide partially inhibited adhesion of prostate carcinoma cells to collagen IV (Leung-Hagesteijn et al., 1994). One binding site for α₃β₁ integrin has been identified on α1 chain of collagen IV (residues 531–543) using synthetic peptide GEFYFDLRLKGDK, which promoted adhesion of several tumor cell lines and induced α₃β₁-mediated signal transduction (Lauer et al., 1998; Miles et al., 1994, 1995). However, the role of α₃β₁ as a collagen IV receptor remains controversial since integrin-specific antibody failed to block the adhesion of various cells (Elices et al., 1991; Melchiori et al., 1995), and overexpression of α₃β₁ does not increase adhesion to the collagen IV (Delwel et al., 1994; Weitzman et al., 1993). Most surprisingly, genetic ablation of α₃ integrin increases adhesion and migration of keratinocytes on collagen IV leading to the hypothesis that α₃β₁ is a trans-dominant inhibitor of collagen IV binding integrins (Hodivala-Dilke et al., 1998). Finally, the α₃β₁ integrin complex, either as recombinant or purified from cells, has failed to bind collagen IV in vitro (Eble et al., 1998; Pochec et al., 2003).

Recently, binding of two new integrins, α₁₀β₁ and α₁₁β₁, to collagen IV have been reported (Tiger et al., 2001; Tulla et al., 2001b). The α₁₁ I domain binds stronger to collagen I through the same GFOGER motif as α₂β₁ integrin (Zhang et al., 2003), while binding of α₁₀ is more specific for collagen IV similar to α₁ I domain (Tulla et al., 2001b). Nevertheless, structure of binding sites for these two integrins on collagen IV remains unknown. Expression of α₁₀β₁ and α₁₁β₁ is spatially and temporally restricted to chondrocytes and fetal muscle cells, suggesting their specific role in development.

Thermal denaturation or proteolytic degradation results in exposure of novel binding sites on collagen molecules. This phenomenon was first described for collagen I, which after thermal denaturation binds fibronectin with increased affinity (Engvall and Ruoslahti, 1977; Schor and Court, 1979), and later for collagen IV (Aumailley and Timpl, 1986). These results led to suggestion that cell adhesion to denatured collagen could be mediated by fibronectin-α₅β₁ integrin bridge (Tuckwell et al., 1994). More recent studies show the exposure of a cryptic site within the triple-helical part of the collagen IV after proteolytic cleavage (Xu et al., 2001). This cleavage occurred in blood vessels during angiogenesis and was associated with a loss of α₁β₁ and concomitant gain of α_vβ₃ integrin binding. Moreover, monoclonal antibodies to denatured collagen IV inhibited angiogenesis and tumor growth in vivo. These results suggest that degradation and proteolytic remodeling of BMs might expose cryptic sites within collagen IV with altered integrin specificity and biological functions.

In addition to the triple-helical part, NC1 domain of collagen IV also promotes cell adhesion (Chelberg et al., 1989; Herbst et al., 1988; Setty et al., 1998), stimulates outgrowth of embryonic neurons (Lein et al., 1991), and inhibits morphogenesis in Hydra vulgaris (Zhang et al., 1994). The latter finding led to the suggestion that NC1 domain could inhibit angiogenesis by disrupting the assembly of vascular basement membrane (Sarras and Hudson, 1997). Subsequent studies with purified recombinant NC1 domains demonstrated that α2, α3, and α6 NC1 domains promote the adhesion and migration of endothelial cells, while α1, α4, and α5 NC1 domains were inactive (Petitclerc et al., 2000). Interestingly, systemic administration of α1, α2, and α3 NC1 domains also inhibits tumor growth and angiogenesis in vivo (Colorado et al., 2000; Maeshima et al., 2000b; Petitclerc et al., 2000). Inhibition of cell adhesion with neutralizing antibodies suggests that different NC1 domains may bind distinct integrins: α₁β₁ for α1 NC1, α_vβ₃, α_vβ₅, and α₃β₁ for α2 NC1, and α_vβ₃ for α3 and α6 NC1 domains (Fig. 3). Somewhat different results were obtained for melanoma and ovarian carcinoma cells which interact with α2 NC1 domain through α₁β₁ and α_vβ₃ integrins (Roth et al., 2005). Recently, we directly identified α_vβ₃ and α_vβ₅ integrins as endothelial receptors for α3 NC1 domain by affinity chromatography and confirmed binding of purified integrins in vitro (Pedchenko et al., 2004). So far, specific integrin binding sites have been characterized only for α3 NC1 domain (Han et al., 1997; Maeshima et al., 2000a). One region was initially identified within the first half of α3 NC1 domain (residues 54–132), since corresponding recombinant protein inhibited proliferation of endothelial cells, induced apoptosis, and decreased tumor growth in nude mice (Maeshima et al., 2001a). Subsequently, shorter synthetic peptide derived from this region (residues 74–98) was shown to inhibit proliferation of endothelial cells presumably via interaction with α_vβ₃ integrin in a RGD-independent manner (Maeshima et al., 2001b). This results in inhibition of the Cap-dependent protein synthesis mediated by focal adhesion/phosphatidylinositol-3/Akt/mTOR kinases (Maeshima et al., 2002; Sudhakar et al., 2003). A second cell-binding site was ascribed to amino acids 185–203 of α3 NC1 domain based on the inhibition of melanoma cells proliferation by the corresponding synthetic peptide (Han et al., 1997). This peptide binds α_vβ₃ integrin and CD47/IAP from the cell lysate through specific interaction of β₃ subunit with amino acid triplet SNS (Pasco et al., 2000; Shahan et al., 1999). In addition, we recently found that endothelial cell adhesion to recombinant α3 NC1 is critically dependent on the interaction of the adjacent amino terminal RGD motif with α_vβ₃ integrin as deletion of RGD sequence significantly decreased cell adhesion, and the two above-mentioned peptides were unable to recapitulate cell adhesion to α3 NC1 domain (Pedchenko et al., 2004). Thus, NC1 domains of collagen IV act as ligands for distinct integrins through several binding sites that may mediate their anti-angiogenic and anti-tumor activities.

Interaction of cells with 7S domain of collagen IV has been reported by several groups, suggesting the existence of an additional yet uncharacterized binding site for cell adhesion receptors. Purified 7S domain efficiently blocks cell aggregate development in Hydra, suggesting that it may play an important role in morphogenesis (Zhang et al., 1994). In another report, the 7S domain potently stimulated chemotaxis in neutrophils, and this response is potentially mediated by a 67-kD lectin-like protein similar to the elastin receptor (Senior et al., 1989).

Nonintegrin Receptors

Evidence has also been reported for cell-collagen IV interaction through nonintegrin receptors, such as cell-surface heparin sulfate proteoglycans, glycoprotein VI and mannose receptor family (Leitinger and Hohenester, 2007). Melanoma cells expressing the chondroitin sulfate proteoglycan CD44 bind directly to collagen IV, and cell interaction and migration could be inhibited by anti-CD44 antibodies (Knutson et al., 1996). Using synthetic triple-helical peptides, the binding site for CD44 has been localized to a 14-residue long region within the α1(IV) chain (Lauer-Fields et al., 2003). In addition, three heparin-binding sequences have also been found in the triple-helical and NC1 domains of α1(IV) and α2(IV) chains (Koliakos et al., 1989). A synthetic peptide (TAGSCLRKFSTM) corresponding to a sequence from α1(IV) NC1 domain has been shown to promote adhesion and spreading of endothelial cells (Tsilibary et al., 1990). Another heparin-binding peptide, Hep-III (GEFYFDLRLKGDK), derived from the interruption of α1 chain promoted adhesion of keratinocytes and mesangial cells (Setty et al., 1998; Wilke and Furcht, 1990). Interestingly, the same peptide mediates binding of α₃β₁ integrin and may also interact with α₂β₁ integrin depending on integrin expression levels (Kim et al., 1994; Setty et al., 1998). However, the degree of cell spreading on Hep-III is significantly lower than on full-length type IV collagen or its triple-helical domain, and Hep-III only partially inhibited cell motility on collagen IV, indicating rather limited contribution of proteoglycans to mechanisms of cell adhesion and migration.

Another group of collagen-binding proteins called discoidin domain receptors (DDRs) has recently been discovered (Shrivastava et al., 1997; Vogel et al., 1997). DDRs belong to the large family of receptor tyrosine kinases, and two members of this family, DDR1 and DDR2, are widely expressed in mammalian cells. Their most distinctive feature is the stimulation of tyrosine kinase activity by native triple-helical collagens that sustain over an extended period of time. Interestingly, collagen IV stimulates phosphorylation of DDR1, but not DDR2. Activation of DDR1 was not affected in the β₁ integrin null cells, suggesting that DDR1 represents independent receptor directly interacting with collagen (Vogel et al., 2000). This does not exclude a possibility of further downstream crosstalk between integrin and DDR signaling pathways. Deletion of DDR1 led to decreased collagen adhesion, migration of vascular smooth muscle cells, and enhanced proliferation of mesangial cells (Curat and Vogel, 2002; Hou et al., 2001). In addition, DDR1 knockout mice develop localized thickening of GBM and proteinuria, suggesting that collagen IV-DDR1 interaction plays an important role for structural integrity and filtration function in the kidney (Gross et al., 2004).

Alport's Syndrome

In 1927, Arthur Cecil Alport described the clinical syndrome of a hereditary nephritis characterized by hematuria and sensorineural deafness (Alport, 1927). For decades, the syndrome was diagnosed solely based on its clinical manifestations. However, with advances in electron microscopy in the early 1970s, the underlying pathological clues to the renal dysfunction, which was ultrastructural changes with regional thinning and thickening of the GBM, were reported from several laboratories (Churg and Sherman, 1973; Hinglais et al., 1972; Kinoshita et al., 1969; Spear and Slusser, 1972). The presence of a unique collagen in the GBM had already been reported by Kefalides in the late 60s (Kefalides, 1966, 1968), and it was thought to consist of three identical α-chains forming the scaffold of the GBM (Kefalides, 1971). This inspired Spear to suggest that mutations in the gene for this unique structural GBM component could cause Alport's syndrome (Spear, 1973). However, it was not until 1990 that the genetic basis for what came to be known as X-linked Alport's syndrome was described (Barker et al., 1990; Hostikka et al., 1990; Myers et al., 1990; Pihlajaniemi et al., 1990). Since then, a large number of different mutations in the COL4A5 gene as well as in the COL4A3 and COL4A4 genes have been reported (Jais et al., 2000, 2003; Longo et al., 2002). These mutations not only include nonsense and missense mutations, but also mutations that affect splicing, stability of triple-helical structure, posttranslational modifications, and the assembly of chains into heterotrimeric molecules. While mutations in the COL4A3 or COL4A4 genes cause autosomal recessive form of the syndrome affecting both males and females equally (Lemmink et al., 1994; Mochizuki et al., 1994), ∼85% of patients with Alport's syndrome carrying mutations in the COL4A5 gene and therefore are classified as X-linked with the highest prevalence in males (Barker et al., 1990; Martin et al., 1998).

The expression of genes encoding the α3(IV), α4(IV), and α5(IV) are subjected to developmental switch in the GBM. Thus, disregard of which gene is affected mutations that alter normal expression, posttranslational modifications or assembly of the α3α4α5 triple-helical molecule will cause an arrest in the developmental switch and persistence of the α1α1α2 network in the GBM with Alport's syndrome as consequence. Differences in the physical and biological properties between the classical α1α1α2 network and the α3α4α5 network are most likely to underlie the disability of the former network to compensate for the lack of the latter in the Alport's syndrome patients. Relative to α1α1α2 network, the α3α4α5 network is highly crosslinked because of its larger degree of inter- and intrachain disulfide bonds and has different physical properties than the classical α1α1α2 network (Gunwar et al., 1998). During the embryonic development of the kidney and the increased plasma filtration cross the glomerular capillaries, the levels of circulating proteins including serum proteases constantly increases. The higher susceptibility of the embryonic α1α1α2 network to proteolytic activities makes the developmental switch to the more resistant α3α4α5 network necessary to withstand higher glomerular pressure and proteolytic degradation of the GBM. Patients with Alport's syndrome probably become more sensitive to gradual proteolytic degradation resulting in uneven thickness of the GBM characterized in all renal samples of Alport's syndrome patients.

Patients with Alport's syndrome can benefit from renal transplantation as the development of nephritis due to generation of alloantibodies is not very common. Sensitive assays have shown that the posttransplant alloantibodies are always directed against multiple epitopes along the α3(IV), α4(IV), or α5(IV) chains (Hudson et al., 1992; Kalluri et al., 2000). Circulation of such antibodies is a characteristic feature of another collagen IV-related renal disease known as Goodpasture's syndrome.

Goodpasture's Syndrome

Goodpasture's syndrome is an autoimmune disease characterized by a rapidly progressive glomerulonephritis which is often associated with lung hemorrhage. The disease, which occurs primarily in young men usually in their late 20s and in both men and women in their 60s, is caused by circulating autoantibodies directed to specific epitopes located on the α3(IV) NC1 domain (Saus et al., 1988; Wieslander et al., 1984). Although immunoreactivity to epitopes located on the NC1 domains of other collagen IV chains have been reported, almost all anti-GBM antibodies from patients are reactive to the α3(IV) NC1 domain (Kalluri et al., 1995). In this regard, the immunospecificity of antibodies to α3(IV) NC1 domain is puzzling, since antibodies to NC1 domains of other chains, such as α1(IV) and α2(IV), do not cause anti-GBM nephritis (Bolton et al., 1995; Kalluri et al., 1996; Sado et al., 1998). Genetic factors have been reported indicating that patients with certain allelic variations (HLA-DRB1*1501 and DRB1*1502) have increased susceptibility to develop the disease, whereas other alleles (HLA-DR7 and DR1) are protective (Phelps and Rees, 1999). The role of autoreactive CD4+ Tcells in mediating the disease has been reported, clearly indicating the importance of cell-mediated autoimmunity in the pathogenesis (Salama et al., 2001; Wong et al., 2001).

The pathogenesis of Goodpasture's syndrome is characterized by glomerulonephritis which is initiated via high-affinity autoantibodies, usually of the IgG class. Two dominant epitopes (E_A and E_B) within the α3(IV) NC1 domain have been identified (Borza et al., 2000; David et al., 2001; Gunnarsson et al., 2000; Hellmark et al., 1999; Netzer et al., 1999). Located in close proximity to each other and near the triple-helical junction (Fig. 2), these epitopes do not normally show reactivity with the antibodies unless the NC1 hexamer is dissociated (Borza et al., 2002). Therefore, it is presumed that environmental factors such as exposure to hydrocarbons or tobacco smoke may be needed to reveal the cryptic epitopes to the immune system (Donaghy and Rees, 1983; Stevenson et al., 1995). Disregard of the cause, binding of immunoglobulins to the GBM activates the complement system and a cascade of proteases, which results in disruption of the glomerular filtration barrier followed by proteinuria and formation of glomerular crescent. The cell-mediated immunoreactivity by CD4+ and CD8+ T cells is soon followed by migration of macrophages and neutrophils into the glomerulus causing further inflammation and increased proteinuria which lead to interstitial nephritis (Kitching et al., 2002; Remuzzi and Bertani, 1998; Timoshanko et al., 2001; Wu et al., 2001, 2002).

Treatments for Goodpasture's patients are limited to plasmapheresis which removes the harmful antibodies from the circulations and/or immunosuppression. Kidney transplantation is possible, however, because there is a risk for recurrence the procedure should wait until anti-GBM antibodies are undetectable in the serum (Netzer et al., 1998a).

Rare Genetic Disorders of Collagen IV

Recently, it was shown that mutations in the COL4A1 gene causes type 1 or encephaloclastic porencephaly, which is a rare neurological disease typically manifested in infants (Gould et al., 2005). The disease is characterized by the presence of degenerative cavities and cerebral lesions in the brain leading to severe pathological disorders such as poor speech development, epilepsy, mental retardation, and cerebral palsy (Pasternak et al., 1980). Using a mouse model generated by random mutagenesis, Gould et al. characterized a novel mouse mutant with severe perinatal cerebral hemorrhage similar to those previously described in a large human family with autosomal dominant type 1 porencephaly (Aguglia et al., 2004; Smit et al., 1984). The causative gene in the mouse mutant was mapped and identified as COL4A1 gene. Sequence analysis revealed that a splice site mutation which resulted in excision of an exon generated a semidominant mutant α1(IV) chain, which upon assembly inhibited secretion of the wild-type and mutant collagen IV (Gould et al., 2005). Electron micrographs of the mutant mice clearly demonstrated structural defects in the cerebral vasculature such as uneven BMs with variable thickness, inconsistent density, focal disruptions, and herniation which presumably reflect a weaker BM structure. Because not all mutant mice developed porencephaly, the authors proposed that in addition to the mutations environmental factors such as birth trauma may be needed to cause the disease (Gould et al., 2005).

The above finding tempted the authors to investigate potential mutations in the human COL4A1 gene in two ethnically separate families with autosomal dominant porencephaly. Sequence analysis revealed two mutations (Gly to Arg and Gly to Ser) leading to replacement of a conserved Gly residue in the Gly-Xaa-Yaa repeats of the triple-helical collagenous domain. Such mutations are known to disrupt the triple-helical formation during the assembly of collagens leading to protein defect (Engel and Prockop, 1991). Interestingly, in addition to porencephaly, one of the adult patients was found to suffer from recurrent hemorrhagic strokes cased by weakened vascular BM. This finding indicates that such dominant mutations in the COL4A1 gene may also contribute to hemorrhagic stroke in adults (Gould et al., 2005). Indeed, such indications have been reported in an adult patient with autosomal dominant porencephaly who showed only minor neurological dysfunction, but suffered from recurrent strokes (van der Knaap et al., 2006). These findings suggest that certain mutations or polymorphisms in the COL4A1, COL4A2, or the genes encoding other BM components may conspire with environmental factors to underlie recurrent hemorrhagic strokes in adults.