Deciphering the preference and predicting the viability of circular permutations in proteins

Area of Science:

• Protein engineering
• Computational biology
• Biotechnology

Background:

• Circular permutation (CP) involves relocating protein termini, offering unique applications in protein engineering and fusion protein creation.
• Current limitations in CP application stem from the difficulty in identifying viable permutation sites.
• Understanding CP site preferences is crucial for advancing its use in research and biotechnology.

Purpose of the Study:

• To analyze the sequence, structural, and dynamical properties of known circular permutation sites.
• To develop a robust prediction system for identifying viable circular permutation sites.
• To explore new applications of circular permutation in protein engineering and biotechnology.

Main Methods:

• Comprehensive statistical analyses of known CP sites.
• Molecular dynamics simulations to assess protein properties.
• Development and integration of four machine learning methods (ANN, SVM, Random Forest, hierarchical feature integration) for prediction.

Main Results:

• Circular permutation preferentially occurs at Gly, Pro, Asp, and Asn residues located in flexible, solvent-exposed regions like coils and loops.
• Disfavored sites include Cys, bulky hydrophobic residues, and those within helices or protein cores.
• The developed prediction system achieved an AUC of 0.9 on an independent dataset, demonstrating high accuracy.

Conclusions:

• The study reveals specific sequence and structural preferences for viable circular permutation sites.
• An advanced prediction system for viable CP sites has been developed, outperforming existing methods.
• This work facilitates broader application of circular permutation in protein engineering and biotechnology, identifying new potential uses.