Abstract
Hydrophobins (HFB) find application in various industries including biotechnology and medical devices; therefore, it is imperative to elucidate and learn more about their folded structures. Few fungal HFB protein structures are available in the Protein Data Bank (PDB), and fewer have been elucidated using homology modeling or short molecular dynamics (MD) simulations in the literature. Moreover, many homology modeling algorithms will only model the region with sequence identity. Therefore, we turned towards the state-of-the-art, artificial intelligence powered AlphaFold. It performed well in predicting the core β-barrel, a characteristic of HFBs, except for HFB9A which was unfolded with low confidence scores. These initial structures were then prepared for accelerated MD simulation in the hope of observing higher protein folding. With 500 ns long aMD simulations, we were able to obtain folded and energetically stable conformations for all the proteins except HFB9A, which exhibited much higher disorder, connected with higher atomic fluctuation, lowest hydrophobicity and overall compactness, and lesser secondary structure formation as visualized during the aMD simulation. The underlying intrinsic disorder in the HFBs was found to be the basis of harder-to-reach folding by AF2 which can be compensated by enhanced sampling MD simulations like the aMD technique. The characteristic two disordered loops of class I HFBs were obtained for SC3, while HFB9A showed that not all class I HFBs contain them. Class II HFBs were more stable, folded and compact with secondary structure motifs conserved throughout the trajectory which can be correlated with their comparatively much lower intrinsic disorder.