Characterizing the existing and potential structural space of proteins by large-scale multiple loop permutations.

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein engineering

Background:

• Protein structure space expansion has stalled despite global genomics efforts.
• Experimental methods for discovering new protein folds are largely unsuccessful.
• Previous computational studies on protein structure space mapping were limited in scope.

Purpose of the Study:

• To computationally generate and characterize novel protein structures using multiple loop permutations (MLP).
• To assess the completeness of the observed protein structure space and identify potential new folds.
• To explore the utility of MLP for protein engineering and design.

Main Methods:

• Extended naturally occurring circular permutations to multiple loop permutations (MLP) to create novel protein structures.
• Clustered generated structures using the TM-score for structural similarity.
• Applied MLP to 2936 protein domains from the Structural Classification of Protein Domains database.

Main Results:

• MLP successfully reproduced existing structural clusters (63%) and identified novel clusters.
• Significant numbers of potentially stable novel clusters were found, particularly for medium-to-large single-domain proteins (>100 residues).
• MLP offers a method to engineer and design novel protein structures, including knotted proteins.

Conclusions:

• Multiple loop permutations (MLP) are a powerful computational tool for discovering and engineering novel protein folds.
• The study suggests that nature may not have fully explored the potential protein structure space.
• MLP has implications for improving protein structure prediction techniques, especially for new-fold targets.