The crystal structure of bovine antithrombin III has been solved by molecular replacement, using a~-antitrypsin (30% sequence homology with human antithrombin III) as a model. The protein crystallizes with two molecules in the asymmetric unit. The use of different resolution ranges was essential in determining the true orientations of the two molecules, because these were the orientations that appeared most consistently in the rotation function, albeit with different scores and slightly different values. Accuracy and correctness of the orientations of the two independent molecules was crucial for the success of the translation function. Stripping off surface atoms from the trial model resulted in a better signal-to-noise ratio in the Crowther-Blow translation function, which turned out to be very discriminative, even though the translational search was performed with less than 30% of the asymmetric unit. In this way, the orientation and position of each of the two independent molecules was separately determined; the crystal packing of the corresponding dimer showed no bad contacts. Phases calculated with this model allowed the determination of heavy-atom sites in a derivative which had previously resisted interpretation. The current R factor after energy minimization and a l ps moleculardynamics simulation is 32% at 3.6 ,~ resolution.
In~oducfionAntithrombin III (Mr = 56 600) is a plasma antiserine protease involved in the coagulation process. Its physiological role is thought to be the inhibition of thrombin and/or factor Xa; factor Xa, together with factor Va and Ca 2÷ ions, helps cleave prothrombin into thrombin, which then cleaves fibrinogen into fibrin (Furie & Furie, 1988). Fibrin then polymerizes in molecular nets which, together with activated platelets, repair blood vessel damage around a wound. The rate of inhibition by antithrombin III (ATIII) is modulated by a polysaccharide, heparin (Rosenberg & Damus, 1973 (Blackburn, Smith & Sibley, 1983) and genetic studies (Chang & Tran, 1986), and is located in the N-terminal domain of the protein. Furthermore, a pentasaccharide that presents the same activity as heparin for the inactivation of factor Xa by antithrombin has been chemically synthesized (Choay, Petitou, Lormeau, Sinay, Casu & Gatti, 1983;Sinay et al., 1984). However, specific contacts at the molecular level between heparin and ATIII are unknown and this prevents any rational design of chemically synthesized analogs of heparin. In order to make such drug-design experiments and molecular modeling possible, the three-dimensional structure of ATIII has to be known. At a later stage, the complex between the pentasaccharide and ATIII will also have to be solved.In this article, we report the crystal structure determination of bovine ATIII. Our strategy was to pursue isomorphous replacement and molecular replacement methods separately. The latter method proved to be successful and helped interpret some results from the first. Because of some difficulties encountered in the application of the ...