Regensburg 2025 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 3: Computational Biophysics I
BP 3.4: Vortrag
Montag, 17. März 2025, 10:15–10:30, H44
Uncovering the Non-Canonical RNA Binding site on the Immune Sensor OAS2 by combining AI, MD simulations and experiments. — •Adrian F. Schnell1, Veronika Merold2, Indra Bekere2, Carina C. de Oliveira Mann2, and Nadine Schwierz1 — 1Institute of Physics, University of Augsburg — 2Department of Bioscience, Technical University of Munich
Molecular dynamics (MD) simulations and machine learning provide powerful tools to predict protein-RNA interactions, but their predictions require experimental verification. In this talk, we showcase an advancement in understanding the immune sensor 2'-5'-oligoadenylate synthetase 2 (OAS2) by combining AlphaFold 3, MD simulations, cryo-electron microscopy (cryo-EM), and cellular assays. Although the structure of the OAS2 has been resolved through cryo-EM, the precise mechanisms underlying its activation and the RNA binding site remained elusive.
To fill this gap, we combined all-atom MD simulations based on cryo-EM structures and AlphaFold 3 predictions to identify non-canonical RNA binding interfaces on the catalytically deficient OAS2 domain. By integrating mutagenesis studies and contact data from MD simulations, we uncovered critical structural details of RNA binding and OAS2 activation. Importantly, our findings reveal how OAS2 domains discriminate RNA length, providing new insights into its function and regulatory mechanisms. These results enhance our understanding of OAS2's antiviral immune role and offer a foundation for developing antiviral strategies targeting the OAS-RNase L pathway.
Keywords: Molecular Dynamics Simulations; Machine Learning; Viral Sensor; Protein-RNA Interactions; Immune Sensor