The full phenotypic range of OI is caused by mutations in the two chains that comprise type I collagen, the major protein of the extracellular matrix of bone, skin and tendon. [3] It is a heterotrimer composed of two copies of the α1 chain, encoded by the COL1A1 gene on chromosome 17, and one copy of the α2 chain, encoded by COL1A2 on chromosome 7. The two alpha chains are similar in sequence organization; they are composed of 338 uninterrupted repeats of the sequence Gly-X-Y, where gly is glycine, X is often proline and Y is often hydroxyproline. A glycine residue in every third position along the chain is crucial for helix formation; glycine’s small size allows it to be tucked into the sterically constricted internal aspect of the helix. The collagen genes are organized with each exon coding for the helical region beginning with a glycine codon and ending with codon for a Y position; therefore the skipping of a helical exon does not cause a frameshift in the collagen transcript.

Over 850 mutations in both chains of type I collagen have been described in OI patients. One general correlation between genotype and phenotype has emerged. Type I OI, the mild form, is caused by quantitative defects in collagen. Only half the normal amount of collagen is produced but all the collagen produced is structurally normal. This is almost always due to a null allele of COL1A1. (15) On the other hand, types II, III and IV OI, the clinically significant forms, are caused by structural defects in either of the type I collagen chains. About 80-85% of these structural mutations cause the substitution of another amino acid, with a charged, polar or bulky side chain, for one of the obligatory glycine residues occurring in every third position along the chain. Glycine substitution mutations temporarily block helix formation and cause overmodification (glycosylation) of the chains of the trimer. About 15-20% of structural mutations are single exon skipping defects, which are incorporated into the trimer because the frame of the transcript remains intact. [17] Essentially all OI-causing mutations are dominant negative mutations. They exert their effects by being secreted and incorporated into the matrix, causing a weakened higher order structure.

For structural mutations of type I collagen, the relationship between genotype and phenotype has been elusive. A lethal mutation is more likely to be located in the α1 chain, in which about 1/3 of known glycine substitutions cause lethal OI, than in the α2 chain, in which only 1/5 are lethal. [18] Nonetheless, both chains contain substantial numbers of mutations causing the full range of the OI phenotype. The two chains have different patterns of lethal and non-lethal mutations along the helical region, supporting different roles for the two chains in matrix. Lethal and non-lethal clusters alternate along the α2(I) chain. The clusters are quite evenly spaced, suggesting that they may play a role in regularly repeating interactions of collagen with non-collagenous matrix molecules. This Regional Model empirically predicts the lethality of 86% of mutations along the α2(I) chain, but binding data supporting the functional role of the regions has yet to be presented. In the α1(I) chain, the mutations may disrupt the stability of the collagen helix itself. [19] Two regions of uninterrupted lethal mutations in the carboxyl end of α1(I) coincide with the major ligand binding region (MLBR) for integrins, fibronectin, and COMP. [18]

The phenotype-genotype relationship in OI is complicated by multiple examples of variable expression. Individuals with the same genotype have a different phenotype, an interesting feature of many dominant disorders. In the α1(I) chain, there are currently about 33 [18, 20, 21] examples of extreme variable expression of the same mutation; these glycine substitutions are found in both lethal and non-lethal forms of OI. A more frequent occurrence in both chains is substantial variation in severity between family members or unrelated individuals with the same mutation. For example, phenotype can range from type III to IV OI. One explanation for this interesting feature may be the existence of discrete modifying genes. Understanding modifying factors may provide new approaches to treatment.

Recently generated mouse models will shed new light on the pathophysiology of OI as well as on modifying factors. The brittle mouse (Brtl) is a knock-in model for type IV OI. [22] It contains a classic glycine substitution at α1(I)G349C, which causes dominant negative OI. The Brtl mouse reproduces the phenotype, histology, biochemistry and biomechanics of the disorder. It also has variable phenotypic expression, which may lead to an understanding of modifying factors. In addition, there is a naturally occurring mouse model for type III OI, the oim mouse. [23] This mouse is atypical of OI in that it has recessive inheritance. The collagen defect is a mutation in the α2(I) chain that prevents incorporation into heterotrimer and leads to the production of an α1(I) homotrimer.