ABSTRACT
The structural information described above has been combined with the known struc ture of human ceruloplasmin to construct a model for FVIIIa [16]. This model replaces an earlier model [47] based on the structure of NR, which, though useful in terms of overall shape prediction, suffers from the defect that the arrangement of the subdomains is almost certainly incorrect. In the model of FVIII domain Al, A2 and A3 correspond to the pairs of hCP domains 1 and 2, 3 and 4, and 5 and 6, respectively. Although models are not substitutes for high-resolution structures, in the absence of other structural information, they can often provide a means not only of understanding structure-function relationships but of suggesting further experi mental approaches. The model of FVIII based on hCP enables a number of observa tions to be made. Thus, five of the six disulfide bridges within the A-type domains are homologous to those in hCP (Al: 153-179 and 248-239; A2: 528-554 and 630-711; A3: 1832-1858) and will therefore be located near the bottom, flat surface of the FVIII molecule. The sixth is nonhomologous and occurs toward the top of the molecule in domain A3, 1899-1903. The three acidic peptide regions and the large B domain can also be predicted to lie near the bottom surface and will therefore be accessible to serine proteases such as thrombin and activated protein C. FVIII appears to contain one copper atom [48] and this can be predicted to be of type II (Sect. 4.1.2). Finally, and of prime importance, the FVIII model readily enables plausible mechanisms explaining why various mutations give rise to clinical disease [16].
