Biophysical Society Thematic Meeting | Trieste 2024

Emerging Theoretical Approaches to Complement Single-Particle Cryo-EM

Poster Abstracts

17-POS Board 17 REFINING AND ASSIGNING METAL IONS OF RNA STRUCTURES Karolina Mitusinska 1 ; Alessandra Magistrato 1 ; 1 Scuola Internazionale Superiore di Studi Avanzati, Trieste, Italy

The presence of metal ions in the biomolecular structure is crucial to its function. For complex RNA-based structures, magnesium ions are of utmost importance due to their role in catalysis, folding and stability. They are also the most often found ion in such structures. However, around 40% of the ribosomal structures deposited in the Protein Data Bank database do not contain a single Mg2+ site. Moreover, the number of Mg2+ ions in RNA-based structures is mostly insufficient to neutralize the negative charge of the nucleic acids of the system. Identification of Mg2+ ion sites is challenging due to the fact that magnesium has the same number of electrons as water and sodium, and neither of which can be distinguished from Mg2+ by difference electron density maps alone. Therefore, in our study, we employed a machine learning (ML) approach to first identify incorrectly assigned ions and to predict the correct placement of metal ions in a set of ribozyme structures. All ion sites were examined using a ML approach in order to determine if the ion type was assigned correctly. Classical molecular dynamics (MD) simulations were used to observe the behavior of the metal ions combined with the same ML approach to determine the ion type during the course of the simulation. Finally, the metal ion positions were superimposed with ML predictions of metal ions placement in order to corroborate the observations from MD simulations. Surprisingly, we were able to identify metal ions which most likely are incorrectly assigned. Poorly coordinated metal ions were more likely to dissociate from their binding site, while those which were assigned correctly and coordinated by more ligands were more likely to stay in their site during the simulation.

45