An analysis of incorrectly folded protein models. Implications for structure predictions

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein structure prediction

Background:

• Homologous proteins share similar amino acid sequences and three-dimensional structures.
• Protein structure prediction often assumes homologous structures can be adapted by side-chain substitution and minor adjustments.
• Understanding native protein characteristics is crucial for accurate structure prediction.

Purpose of the Study:

• To examine the validity of predicting protein structures using homologous modeling.
• To isolate and identify characteristics specific to native protein structures.
• To evaluate the accuracy of computational modeling in protein structure prediction.

Main Methods:

• Constructed two incorrectly folded protein models by swapping side-chains between sea-worm hemerythrin (alpha-helices) and mouse immunoglobulin K-chain (beta-sheets).
• Utilized an automatic computer procedure for side-chain substitution.
• Performed energy minimization using the CHARMM program and compared resulting structures with native forms.

Main Results:

• Incorrect side-chains were incorporated with only minor structural adjustments (0.7–0.9 Å shift).
• Misfolded models achieved comparable potential energy values to native structures post-minimization.
• Detailed analysis revealed less stabilizing electrostatic, van der Waals', and hydrogen-bonding interactions in misfolded models, with increased solvent exposure and altered interior packing.

Conclusions:

• Absence of steric clashes is insufficient for validating model-built protein structures.
• Homologous structure modeling necessitates thorough quantitative evaluation.
• Misfolded models reveal native protein characteristics through their absence.