Evolutionary selection of proteins with two folds

Area of Science:

• Biochemistry
• Structural Biology
• Bioinformatics

Background:

• Most globular proteins adopt a single stable structure, but some dynamically remodel their structures in response to cellular signals.
• Current protein structure prediction algorithms often fail to identify alternative functional folds in these dynamic proteins.
• These algorithms infer structure from coevolved amino acid pairs, potentially missing crucial signatures.

Approach:

• Developed a novel approach to analyze diverse protein superfamilies and subfamilies, including both single-fold and fold-switching variants.
• Searched for coevolutionary signatures masked by over-represented single-fold sequences.
• Applied the approach to identify coevolutionary signals corresponding to distinct protein conformations.

Key Points:

• Successfully identified coevolution of amino acid pairs corresponding to both conformations in 56/58 fold-switching proteins.
• Utilized predicted coevolutionary pairs to guide AlphaFold2 in predicting two experimentally consistent conformations for a protein with an unsolved structure.
• Demonstrated that fold-switching sequences are preserved by natural selection, indicating functional advantages.

Conclusions:

• Widespread dual-fold coevolution suggests that nature actively selects for proteins capable of adopting multiple stable structures.
• This discovery challenges existing prediction models and opens new avenues for understanding protein diversity.
• Paves the way for predicting multiple protein structures from a single amino acid sequence, enhancing our understanding of protein function and evolution.