Quantifying the Errors Introduced by Continuum Scattering Models on the Inferred Structural Properties of Proteins

Abstract

Atomistic force fields that are tuned to describe folded proteins predict overly compact structures for intrinsically disordered proteins (IDPs). To correct this, improvements in force fields to better model IDPs are usually paired with scattering models for validation against experiments. For scattering calculations, protein configurations from all-atom simulations are used within the continuum-solvent model CRYSOL for comparison with experiments. To check this approach, we develop an equation to evaluate the radius of gyration (Rg) for any defined inner-hydration shell thickness given all-atom simulation data. Rg based on an explicit description of hydration waters compares well with the reference value of Rg obtained using Guinier analysis of the all-atom scattering model. However, these internally consistent estimates disagree with Rg from CRYSOL for the same definition of the inner-shell. CRYSOL can over-predict Rg by up to 2.5 Angstroms. We rationalize the reason for this behavior and highlight the consequences for force field design.