SOLVING SMALL PROTEIN STRUCTURE USING EXPERIMENTAL TRIPLET PHASES IN DIRECT METHODS. Mathiesen, R.H. & Mo, P. Dept. of Physics, Norwegian University of Science and Technology-NTNU, N7034 Trondheim, Norway
Triplet phases (TP's) can be determined from 3-beam diffraction experiments with protein crystals [1]. We are investigating the potential for using such measurements in combination with direct methods to solve protein structures. Unlike the MIR and MAD techniques, this method does not require the presence of heavy atoms.
Pancreatic trypsin inhibitor, a small protein with 58 amino acid residues (V=50 111 Å3), was originally solved from 1Å resolution data [2]. In our study small sets of TP's, each assigned a value [[phi]]3=n.[[pi]]/4 closest to its refined equivalent were used to initiate phase expansion and refinement; [[phi]]3 defines the mean direction of the corresponding von Mises distribution. Structure solution, revealing 95% of the backbone atoms, was obtained in a recycling process both with data at 1.55 and 1.75Å. By employing single phases obtained from linear equations in the assigned [[phi]]3-values, the problem of numerous possible solutions was avoided leaving only 4 different phase models to be examined in both cases. Crucial in this process was the derivation of a new figure of merit which is sensitive both for the initial selection and the ensuing expansion of a structure fragment.
References:
[1] Werkert, E., Schwegle, W. & Hümmer, K. (1993). Proc. Roy. Soc. Lond. A442, 33-46.
[2] Walter, J. & Huber, R. (1983). J. Mol. Biol. 167, 911-917.