Allosteric mechanism of the circadian protein Vivid resolved through Markov state model and machine learning analysis

The fungal circadian clock photoreceptor Vivid (VVD) contains a photosensitive allosteric light, oxygen, voltage (LOV) domain that undergoes a large N-terminal conformational change. The mechanism by which a blue-light driven covalent bond formation leads to a global conformational change remains unclear, which hinders the further development of VVD as an optogenetic tool. We answered this question through a novel computational platform integrating Markov state models, machine learning methods, and newly developed community analysis algorithms. Applying this new integrative approach, we provided a quantitative evaluation of the contribution from the covalent bond to the protein global conformational change, and proposed an atomistic allosteric mechanism leading to the discovery of the unexpected importance of A’α/Aβ and previously overlooked Eα/Fα loops in the conformational change. This approach could be applicable to other allosteric proteins in general to provide interpretable atomistic representations of their otherwise elusive allosteric mechanisms.Author summary: Allostery is an important but elusive property that governs critical functionality of many proteins. Quantitative analysis is needed to provide significant insight into protein allostery and lead to better prediction power of this ubiquitous phenomenon. We developed machine learning methods based on robust Markov state model to delineate allosteric mechanism of Vivid as an allosteric protein in the filamentous fungus Neurospora crassa, regulating circadian rhythm of this organism. We accurately reconstructed the equilibrium distributions for two allosteric configurations of Vivid, and determined structural differences among these states. Intriguingly, the novel community analysis derived from machine learning methods reveals the importance of two loop regions for Vivid allostery through quantitative evaluations with statistical significance.