From Data or Dogma? The Myth of the Ideal Helix

Daniel J. Kuster, Sergio Urahata, Jay W. Ponder and Garland R. Marshall
Washington University in St. Louis, Saint Louis, MO, USA

Biophysical Journal, 96, A5 (2009)

Abstract:

In the course of building detailed surface representations for
peptidomimetics, we were motivated to analyze the detailed surfaces of
helices in proteins. Unexpectedly, we observed few ideal helical forms
in high-resolution protein struc- tures. Instead of a bimodal distribution
matching the well-known alpha and 3(10) helical forms, we observed a smooth,
single-peaked population, characterized by intermediate helices with shared
hydrogen bonds. Bifurcated, or three- center hydrogen bonds, have been
well-documented in small molecules and peptides, but they’re rarely
highlighted in the context of folded proteins. The data suggests shared
hydrogen bonds are a major component of helices in proteins. Contrary to
the Pauling-Corey-Branson models, we did not restrict our analysis to single
hydrogen bonds–shared three-center hydrogen bonds were included. High
resolution (<2.0 Ang), electron density data is sharp enough that a helical
structure should be unambiguous and accurately modeled. These helices are
centered on an intermediate helical form. At poorer resolutions (2.0-5.0 Ang),
electron density is ambiguous; refinement fitting methods are employed to
model ideal structures into the data. Here there is an enriched population
of ideal structures. A structurally representative subset of proteins reveals
the same trends as the whole PDB. The data support the observation that ideal
helical parameters do not accurately describe the distribution of real helices
in proteins. Hydrogen bonds are a polar (and polarizable) moiety and an
accurate model must account for this. We present data comparing molecular
dynamics simulations using popular monopole force fields (OPLS-AA and CHARMM)
with a next generation force field (AMOEBA) implementing polarizability and
multipole electrostatics. AMOEBA simulations are shown to quantitatively
reproduce the experimentally observed trends in helical populations. These
results emphasize the importance of using appropriate force field potential
models when simulating hydrogen bonded structures in proteins.