Treffer: Sequence-based prediction of intermolecular interactions driven by disordered regions.

Intrinsically disordered regions (IDRs) in proteins play essential roles in cellular function. A growing body of work has shown that IDRs often interact with partners in a manner that does not depend on the precise order of amino acids but is instead driven by complementary chemical interactions, leading to disordered bound-state complexes. However, these chemically specific dynamic interactions are difficult to predict. In this study, we repurposed the chemical physics developed originally for molecular simulations to predict this chemical specificity between IDRs and partner proteins using protein sequence as the only input. Our approach—FINCHES—enables the direct prediction of phase diagrams, the identification of chemically specific interaction hotspots on IDRs, the decomposition of chemically distinct domains in IDRs, and a route to develop and test mechanistic hypotheses regarding IDR function in molecular recognition. Editor's summary: Intrinsically disordered regions (IDRs) are common features in proteins that enable both specific and general interactions between biomolecules in cells. Understanding from their sequence how IDRs function is challenging given their lack of a defined three-dimensional structure. Ginell et al. developed a computational analysis based on coarse-grained force fields that enables the prediction of IDR associations with other IDRs and folded proteins. This method provides a rapid way to characterize IDRs directly from sequence and identify and prioritize potential cellular interactions for experimental testing. —Michael A. Funk INTRODUCTION: Intrinsically disordered regions (IDRs) are found in >70% of human proteins and can play crucial roles in many cellular processes. IDRs lack a stable three-dimensional structure yet often play key roles in mediating complex interactions with various cellular partners. These interactions can be sequence specific, leading to bound states with a structured interface, or chemically specific, leading to an ensemble of bound configurations. Although deep-learning models can predict sequence-specific interactions, predicting chemically specific molecular recognition remains challenging. RATIONALE: The past few years have seen substantial progress in the accuracy of simple (coarse-grained) molecular force fields for describing the biophysical properties of IDRs. Force fields are a set of equations and numbers that capture the chemical physics representing how residues in a disordered protein interact. Force fields were developed to perform molecular simulations, which can be slow and challenging, especially for larger IDRs. We reasoned that it might be possible to repurpose force fields to predict intermolecular interactions without performing simulations for cases where bound states lack residual structure. To test this idea, we developed a computational framework, FINCHES, which enables different force field functional forms and parameters to predict IDR-mediated chemical specificity. FINCHES enables direct predictions of which residues or regions in an IDR are expected to provide attractive and repulsive interactions with a partner, be that the surface of a folded domain or another IDR. RESULTS: We tested FINCHES-based predictions by focusing on three types of behavior: predicting IDR-IDR interactions, predicting phase separation propensity from sequence, and predicting IDR–folded domain interactions. In all cases, we found good qualitative or semiquantitative agreement between predictions and experiments across various proteins and systems. Because these predictions are analytical and based on an underlying energy function, they are fast, tunable, and fully interpretable. This enables proteome-scale interrogation in minutes. We used FINCHES to explore the chemical structure of IDRs across the human proteome, delineate large IDRs into subdomains, and systematically investigate the consequences of phosphorylation on IDR-mediated interactions at a proteome scale. Most importantly, FINCHES provides an easy and fast way to develop precise molecular hypotheses regarding the driving forces for interaction that may underlie IDR-mediated function. CONCLUSION: Although many caveats are associated with predicting chemical specificity in this way (e.g., sequence-specific interactions will not be captured, so structured interactions mediated by disordered regions are missed), we see FINCHES as providing a new and complementary set of information for understanding IDR-mediated function. FINCHES is fully open source and available as a Python package (https://github.com/idptools/finches) and through an easy-to-use web server (https://www.finches-online.com/). Intrinsically disordered regions can interact with partners driven by complementary chemistry.: FINCHES uses chemical physics from molecular force fields to analytically predict which regions and residues in an IDR can drive attractive or repulsive interactions with a partner protein (top). This program enables the de novo prediction of regions that drive IDR-mediated binding (bottom left), how sequence and environment cooperate to tune IDR-mediated phase separation (bottom middle), and how IDRs can interact with the surfaces of folded domains (bottom right). [ABSTRACT FROM AUTHOR]

Copyright of Science is the property of American Association for the Advancement of Science and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)