Stress Granule Protein Features Allow for Accurate Prediction of Biological Condensate Localization with GraPES

Biological condensates have been thrust to the forefront of molecular biology over the past decade for their implications in enzymatic activity, neurodegenerative diseases, and cellular organization. It is thought that many of these membraneless organelles are formed through a process known as liquid‐liquid phase separation, in which key protein or nucleic acid drivers seed the formation of protein‐rich foci within the cellular milieu. Stress granules are stress‐induced assemblies that are belong to this group of biological condensates and are of keen interest to the biomolecular research community due to being linked to both long‐term cell viability and a variety of protein aggregation‐based diseases. Recently, a large amount of proteomic data has been generated that provides unprecedented insight into stress granule composition and stands as fruitful ground for further analysis. Interrogation of this data revealed that stress granule proteins are enriched in features that favor protein liquid‐liquid phase separation. Proteins within stress granules were found to be more disordered, soluble, and abundant than their proteome and cytosolic controls while also having an increased potential for post‐translational modifications. Furthermore, these “stress granuleomes” were found to be enriched for multivalent character by possessing multiple ordered domains and being likely to interact with RNA maintaining a high level of protein‐protein interactions under basal conditions. Our findings are consistent with the notion that stress granule formation is driven by protein liquid‐liquid phase separation. Furthermore, stress granule proteins appear poised near solubility limits while possessing the ability to dynamically alter their phase behavior in response to external threat. We culminate results from our analysis into novel predictors for granule incorporation in yeast and mammalian cells that out performs similar computational tools. Using this predictor, we were able to correctly identify new stress granule components, two of which we validated with colocalization microscopy in mammalian tissue culture cells [1]. We then developed a second series of sequence‐base predictors and integrate all of these computational tools into a user‐friendly web‐based interface, called GraPES (Granule Protein Enrichment Server), where users can either look up a variety of pre‐calculated likelihood z‐scores for human and yeast proteins or obtain novel predictions from their own input FASTA formatted protein sequences.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.