STRUCTURAL AND DYNAMIC PROPERTIES OF A BETA-HAIRPIN-FORMING LINEAR PEPTIDE .1. MODELING USING ENSEMBLE-AVERAGED CONSTRAINTS

The linear peptide Y-Q-N-P-D-G-S-Q-A (one letter amino acid code) disp lays a high population of P-hairpin conformations in aqueous solution at 5 degrees C (F. J. Blanco et al. J. Am. Chem. Sec. 1993, 115, 5887- 5888), indicating that it should be a useful model system for elucidat ing local interactions that induce and stabilize beta-hairpins. Agains t this background, we have performed a detailed study of the peptide's conformational and dynamic properties using 2D NMR and computational modeling. Using the linear component of NOE buildup curves, 122 nuclea r Overhauser effect (NOE) distance constraints were derived for the ma jor (trans-Pro-4) isomer. These distance constraints and three dihedra l angle constraints were used in conjunction with simulated annealing (X-PLOR program) to produce a well-converged set of 24 solution struct ures. The individual structures all contain significant NOE constraint violations. The average RMS violation is 0.25 Angstrom, and the avera ge maximum violation is 1.02 Angstrom. This result prompted a further evaluation of the NOE distance constraints using the DISCON program, w hich iteratively removes spin diffusion contributions. Reduced, but st ill significant, violations were observed after reminimizing with redu ced repulsive interactions and with loosened distance constraints corr esponding to the widest bounds derived from the DISCON and NOE buildup calculations. To investigate whether the residual constraint violatio ns reflect conformational averaging, selected sets of constraints were treated in an ensemble-averaged fashion within the CONGEN molecular m odeling program. In addition to the 122 NOE constraints, 41 (3)J coupl ing constant constraints and 55 ''no NOE'' constraints were also used in the CONGEN calculations. Constraint satisfaction and physical energ ies improved significantly when ensemble-averaging was applied, The pr edominant solution conformation contains a type I beta-turn, with Pro- 4 and Asp-5 occupying the corners. Major stabilizing interactions incl ude backbone-backbone hydrogen bonds involving Asn-3, Gly-6, and Ser-7 , hydrogen bonds involving the Asn-3 side chain, and contacts between Tyr-1 and Ala-9. Thus, the peptide's conformation space is restricted by hydrogen bonding interactions with varying degrees of occupancy and by the presence of a proline residue. Hydrophobic, hydrogen bonding, and electrostatic interactions occur between Tyr-1 and Ala-9, but the ensemble-averaged calculations indicate that these interactions are tr ansient. These conclusions are supported by C-13 relaxation measuremen ts and a water-solvated unrestrained molecular dynamics simulation [M. S. Friedrichs et al. J. Am. Chem. Sec. 1995, 117, 10855-10864].