Small Disulfide Proteins with Antifungal Impact: NMR Experimental Structures as Compared to Models of Alphafold Versions.

In response to the growth of emerging resistance to conventional antifungal drugs, antifungal proteins (AFPs) of filamentous Ascomycetes origin have been discovered in recent years. Understanding the structure of AFPs is crucial for elucidating their antifungal mechanisms and developing new therapeutic agents. While nuclear magnetic resonance (NMR) has proven effective in determining the structures of small proteins, some AFP structures remain unresolved, necessitating the use of alternative prediction methods. Through bioinformatics analysis and heatmaps of amino acid sequence identity and similarity matrix, we categorized AFPs into three major classes and six subcategories, revealing structural and bioactivity differences. We employed AlphaFold (AF) to predict the 3D structures of six different AFPs, with predictions compared to NMR-derived structures. The results demonstrated a high degree of consistency between AF and NMR structures, with AF excelling in structural quality assessment and accurately capturing complex disulfide bond patterns. Both AF2 and AF3 models outperform the NMR model in overall structural quality and coherence, with AF3 showing the best performance. However, the limitations of AF should be considered, including its reduced accuracy in predicting multi-metal ion complexes, suboptimal performance in highly flexible or disordered regions, and its inability to account for multiple conformers, as it generates only a single dominant structure. Moreover, while AF3 accurately predicts all disulfide bond patterns, AF2 falls short in this regard. This study verifies the reliability of AF in the structural prediction of cysteine-rich AFPs while highlighting these constraints, offering important support for the rational design of new protein-based antifungal drugs.