LOW-RESOLUTION TRIPLET PHASES FOR TWO PROTEINS ESTIMATED FROM THREE-BEAM DIFFRACTION EXPERIMENTS - ACQUISITION AND POSSIBLE APPLICATIONS
F. Mo1, R.H. Mathiesen1, P.M. Alzari2, J. Lescar3, B. Rasmussen3
1Dept. of
Physics, NTNU, N-7034, NORWAY,
2Institut Pasteur, Immunologie
Structurale, 25 rue du Dr. Roux, F-75724 Paris Cedex 15, FRANCE,
3ESRF, B.P. 220, F-38043 Grenoble Cedex,
FRANCE.
Keywords: three-beam diffraction,
experimental phases, direct methods, protein structure
determination.
The direct physical acquisition of phase information from three-beam interference experiments in non-perfect crystals has been demonstrated with crystals for a wide range of molecular complexity, including until recently 4 - 5 proteins [1]. The potential of physically estimated triplet phases (PETP's) in macromolecular crystallography depends critically on their availability from protein crystals. We have made successful physical phase estimation (PPE) with two other proteins: guinea-fowl hexagonal lysozyme in space group P6122, a = b = 89.02 A, c = 61.80 A, unit-cell volume V ~ 424,100 A3, and C. thermocellum cellulase CelA, space group P212121, a = 50.05 A, b = 63.50 A, c = 104.75 A, V ~ 333,000 A3. The structures of both these glycosidases have been determined recently to high resolution [2, 3]. The average difference |DF3| between the estimated triplet phases and those calculated from the crystallographic refinements was 17.9o and 15.9o, respectively.
Whereas the experimental requirements and the actual procedure for PPE are now quite well established, this is not so regarding the application of experimental phases for the solution of macromolecular structure. Clearly, to make the method practical, the experimental work must follow a strategy designed to give the information needed for solving the structure from a minimal number of PETP's. Feasibility studies on the use of PETP's for the re-determination of two small protein structures, rubredoxin at 1.54 A resolution, and pancreatic trypsin inhibitor at resolution in the range 1.55 - 2.0 A, have been successful and very instructive [4, 5]. In both cases small sets of triplet phases, assumed known with a mean error of 20o, were used together with single phases derived from these triplets as input to direct methods for phase refinement and expansion. Similar or even better results could well arise if this phase information was fed into other direct methods procedures, or maximum entropy methods.
PETP's can be chosen so as to contain a limited number of
strong reflections at low resolution. These reflections carry
information on the coarse structure that may be used to define
the molecular envelope in the solvent matrix. Identification of
the molecular envelope from a limited PPE, and the use of this
information in phase extension algorithms need to be explored.