Modeling unfolded states of peptides and proteins

Abstract

The hydrophobic effect is the major factor that drives a protein molecule toward collapse and folding. In this process, residues with apolar side chains associate to form a solvent-shielded hydrophobic core. Often, this hydrophobic contribution to folding is quantified by measuring buried apolar surface area, reckoned as the difference in area between hydrophobic groups in the folded protein and in a standard state. Typically, the standard state area of a residue is obtained from tripeptide models, with the results taken to implicitly represent values appropriate for the unfolded state. Here, we show that a tripeptide is a poor descriptor of the unfolded state, and its widespread use has prompted erroneous conclusions about folding. As an alternative, we explore two limiting models, chosen to bracket the expected behavior of the unfolded chain between reliable extremes. One extreme is represented by simulated hard-sphere peptides and shown to behave like a homopolymer with excluded volume in a good solvent. The other extreme is represented by fragments excised from folded proteins and conjectured to approximate the time-average behavior of a thermally denatured protein at Tm, the midpoint of the unfolding transition. Using these models, it is shown that the area buried by apolar side chains upon folding is considerably less than earlier estimates. For example, upon transfer from the unfolded state to the middle of an alpha-helix, an alanine side chain buries negligible area and, surprisingly, a valine side chain actually gains area. Among other applications, an improved model of the unfolded state can be used to better evaluate the driving force for helix formation in peptides and proteins.