Journal of the American Chemical Society, Vol.121, No.21, 4952-4960, 1999
Iterative protein redesign
An iterative redesign protocol for the transformation of a non-native peptide into a series of nativelike proteins derived from elementary considerations of biological evolution coupled with H-1 NMR as an artificial selection criterion is presented. Each of three heptad d position leucines in the helix-helix interfaces of the prototype heme protein maquette, [H10H24](2) or (alpha-SS-alpha)(2), were replaced in a unit modification per helix by more conformationally restricted beta-branched and aromatic amino acids. The secondary structure content (evaluated by circular dichroism and infrared spectroscopies), solvent accessibility of the tryptophan residues (measured by fluorescence spectroscopy), global stability (quantitated by isothermal chemical denaturation), and degree of conformational specificity (determined by H-1 NMR spectroscopy) of the resultant peptides were determined. Improvement in the degree of conformational specificity was utilized as a selection criterion to choose three of the nine singly modified peptides for a second unit modification per helix. Five of the resultant seven doubly modified peptides were nativelike, as determined by NMR spectroscopy. One of the doubly modified peptides was chosen for a third unit modification per helix, which resulted in three triple variants with low conformational specificity. The 20 proteins synthesized fold into discrete, stable four-alpha-helix bundles but with differing stabilities (-Delta G(H2O) from 10.50 to 22.73 kcal/mol) and varying degrees of conformational specificity (multistructured to singular solution structure). The singly, doubly, and triply modified (per helix) peptides can be mapped onto a contiguous segment of sequence space, providing the first experimental map of this vast molecular terrain. The energetic contours of sequence space are revealed in terms of both global folding energies (-Delta G(H2O)) and degree of conformational specificity within the hydrophobic core. Remarkably, six of the peptides studied (30%) contain uniquely structured hydrophobic cores amenable for NMR structural determination. The map of sequence space readily identifies a plastic site within the protein, a position which can be occupied by various amino acids with retention of a uniquely structured global fold, thereby providing a possible route for iterative redesign toward chemical enzymatic function.