STRUCTURE DETERMINATION OF HELICAL AND HAIRPIN PEPTIDES WITH UP TO 30 RESIDUES PER ASYMMETRIC UNIT. Isabella L. Karle, Laboratory for the Structure of Matter, Naval Research Laboratory, Washington, D. C. 20375-5341, USA

Practical information for the successful application of vector search/translation function/tangent formula expansion using the known structure of a fragment from another crystal to derive the structures of molecules containing 100 to 300 C, N and O atoms will be presented. Large molecules containing S or heavier atoms will not be discussed since the heavier atoms usually significantly facilitate the structure solution by the Patterson function, direct phase determination or anomalous dispersion differences (see the 308 non-H atom structure, Konnert, Karle and Karle, Montreal ACA Abstracts 23, 2m6 A (1995)). The structure solutions of the following will be described: [[beta]]-turn/[[beta]]-sheet Boc-Leu-Val-Val-D-Pro-Gly-Leu-Val-Val-OMe (two conformers in P1, a=9.739 Å, b=11.519 Å, =26.253 Å, [[alpha]]=98.39, [[beta]]=91.45 and [[gamma]]=107.80deg.); the helix-linker-helix Boc-L-[Val-Ala-Leu-Aib-Val-Ala-Leu]-Acp-D-[Val-Ala-Leu-Aib-Val-Ala-Leu]-OMe (in P4₁, a=b=10.094 Å, c=93.383 Å); and the helix- reverse-helix Boc-L-[Val-Ala-Leu-Aib-Val-Ala-Leu]-D-[Val-Ala-Leu-Aib-Val-Ala-Leu]-OMe (two conformers in P2₁, a=10.016 Å, b=86.843 Å, c=10.036, [[beta]]=90.08deg.). The latter two crystals have very long axes (near 89 Å and 93 Å), that caused overlap of the diffraction spots; however, the less than ideal data measurements did not hinder the solution process. The data were measured to 0.9 Å resolution. Past experience and more recent experience in direct phase determination, as well as vector search procedures, strongly suggest that data with a resolution of at least 1.1 Å is needed for a successful solution. However, limiting the data used for the Patterson function to a maximum value of 60deg.-65deg. for 2[[theta]] (Cu radiation) has usually been necessary for a successful vector search of a fragment. The model used for the search fragment has consisted of 20-25% of the size of the unknown structure. The model generally contains only backbone and C^[[beta]] atoms from the more rigid portion of the known structure due to the inherent flexibility of peptide molecules. It may be desirable, but not necessary, to use a model from a known structure that has a sequence similar to that present in the unknown. Finally, the values of the combined figure of merit for the fit of the search model to the data of the unknown have ranged from 0.2 to 0.9 for successful trials.

The PATSEE program has been used for the above examples.