Assessment of inconsistencies in the solvent-accessible surfaces of proteins between crystal structures and solution structures observed by LC-MS.

Structural proteomics techniques are useful for identifying the binding sites of proteins. The surface of a target protein with and without a bound binding partner is artificially labeled using a hydroxy radical, deuterium, or a low-molecular-weight chemical, and the difference in the label strength with and without the bound partner is determined. Label strength maps are then prepared on the Protein Data Bank (PDB) structure to identify the binding surface. However, the surface-accessible sites determined using such structural proteomics methods are frequently inconsistent with those calculated based on PDB structures, speculating that the measurement determines chemical accessibility rather than solvent accessibility. In this study, the solvent-accessible surface of human serum albumin was analyzed using covalent protein labeling with varying concentrations of CH2O and then compared to surfaces derived from 27 PDB structures. The results indicated that inconsistencies in solvent-accessible surface area values calculated from PDB structures are not caused by the limited capabilities of liquid chromatography-mass spectrometry coupled with covalent protein painting but instead are due to the lack of PDB data representing the structures in solution.