Evaluation of AlphaFold3 for Predicting Human Heme-Binding Protein Structures

Area of Science:

• Biochemistry
• Structural Biology
• Computational Biology

Background:

• Heme is a vital cofactor in biological processes, and understanding its interaction with heme-binding proteins (HBPs) is crucial for elucidating protein function and developing new applications.
• Artificial intelligence (AI) tools like AlphaFold3 (AF3) are increasingly used for protein structure prediction, but their accuracy for heme-binding proteins requires evaluation.

Purpose of the Study:

• To assess the reliability of AlphaFold3 (AF3) in predicting the structures of human heme-binding proteins (HBPs) in their heme-bound states.
• To compare AF3-generated HBP structures with experimentally determined structures, focusing on heme-binding configurations and their functional implications.

Main Methods:

• Generated apo and holo structures of four human HBPs (SO-b5, Ncb5or-b5, CYB5B, NGB) using AlphaFold3 (AF3).
• Compared AF3 predictions with existing experimental HBP structures, analyzing heme positioning, binding pocket geometry, and coordination environments.
• Utilized molecular dynamics simulations to evaluate the conformational flexibility and stability of AF3-predicted structures, particularly for neuroglobin (NGB).

Main Results:

• AF3 generally predicted accurate overall heme molecule positions within the binding pockets of HBPs.
• Significant differences were observed in heme-binding configurations, including pocket geometry and coordination, impacting functional interpretation.
• The AF3-predicted NGB model lacked a crucial disulfide bond present in the experimental structure, leading to altered local folding and heme-binding environment.
• Molecular dynamics simulations revealed distinct conformational dynamics for the AF3-predicted NGB compared to its experimental counterpart.

Conclusions:

• AlphaFold3 (AF3) demonstrates potential for modeling heme-binding proteins (HBPs), particularly in predicting overall heme placement.
• Limitations exist in AF3's ability to capture fine details of heme-binding configurations and local structural features crucial for protein function.
• Further refinement and validation are necessary before AF3 can be fully relied upon for detailed functional analysis of heme-binding states in HBPs.