What can AlphaFold do for antimicrobial amyloids?

Area of Science:

• Structural biology
• Biophysics
• Computational biology

Background:

• Protein and peptide assemblies, known as amyloids, play vital roles in biological processes, both normal and disease-related.
• Understanding amyloid structures is challenging but crucial for biomedical and technological applications.
• Antimicrobial peptides (AMPs) and human amyloids exhibit diverse structural configurations, including cross-β and cross-α structures.

Purpose of the Study:

• To evaluate the accuracy of AlphaFold2 ColabFold in predicting the structures of various antimicrobial amyloids.
• To investigate the structural preferences of AI models when predicting complex amyloid formations, including mixed cross-α and cross-β structures.
• To assess the potential of AI-driven structure prediction for understanding amyloid functions and therapeutic applications.

Main Methods:

• Utilized the AlphaFold2 ColabFold method for structure prediction.
• Analyzed eight antimicrobial peptides (AMPs) with known structures and distinct amyloidogenic features.
• Included human amyloids (amyloid-β and islet amyloid polypeptide) due to their disease relevance and antimicrobial properties.

Main Results:

• AlphaFold2 ColabFold models predominantly predicted α-helical structures for the tested amyloids.
• The AI models successfully identified α-helical mated sheets and hydrophobic cores characteristic of cross-α configurations.
• The predictions suggest a tendency for AI algorithms to favor monomeric helical assemblies or membrane-active helical peptide forms.

Conclusions:

• AI-based structure prediction methods like AlphaFold2 ColabFold show a preference for α-helical conformations in amyloid prediction.
• These findings offer insights into how AI might interpret or model the monomeric or membrane-interacting states of peptides.
• The study highlights the potential and limitations of current AI tools in deciphering complex amyloid structures.