As part of a program to determine the roles of individual amino acids in stabilizing protein structures, the structures of a series of mutant T4 phage lysozymes have been determined (e.g. Grütter, Weaver, Gray & Matthews, 1983; Alber, Grütter, Gray, Wozniak, Weaver, Chen, Baker & Matthews, 1986). The refinement of each mutant (unpublished results) is in principle very similar to the refinement of an enzyme-inhibitor complexes. One starts with the refined structure of the native protein, locally modified to correspond to the structure of the mutant. In the early stages of the refinement the stereochemical restraints are kept weak to allow the starting model to relax to conform to the diffraction data observed for the mutant structure. When there appear to be no systematic shifts in the coordinates from one refinement cycles to the next, the stereochemical restraints are strengthened in order to enforce the `ideal' geometry. Some representative refinement statistics are summarized in Table 4. Experience with refining these mutant structures at resolutions in the range 1.9 - 1.7Å suggests that the refinement procedure is capable of successfully moving both main-chain and side-chain atoms through distances of 1.0Å.
24pt Several of our colleagues have contributed to the development of the refinement package by applying it to their own problems and suggesting modifications and improvements; in particular, we thank Drs M. A. Holmes and M. F. Schmid. Dr Holmes also suggested `TNT' as the name for the package. We also thank Dr N. Isaacs for providing detailed information on the Lifchitz modification of the Agarwal fast Fourier transform algorithm. The work was supported in part by grants from the national Institutes of Health (GM 20066; GM 21967; GM 35114), the National Science Foundation (DMB 8611084) and the Murdock Charitable Trust.