Biophysical Society Thematic Meeting | Trieste 2024

Emerging Theoretical Approaches to Complement Single-Particle Cryo-EM

Tuesday Speaker Abstracts

ZERNIKE3D AND HETSIREN FOR CONFORMATIONAL AND COMPOSITIONAL HETEROGENEITY ANALYSIS OF CRYO-EM IMAGES Jose Maria Carazo 1 ; David Herreros 1 ; James Krieger 1 ; Marcos Gragera 1 ; Carlos Oscar S Sorzano 1 ; 1 CNB-CSIC, Department of Structural Biology, Madrid, Spain We present two different approaches to analyze Electron Microscopy (cryo-EM) images with the goal to detect conformational and/or compositional structural changes at the specimen level in a continuous manner. One makes use of the notion of flexibility fields and expansion into a certain bases (Zernikes3D), precisely modulating biologically relevant motions by integrating bonded and non-bonded sterochemical constraints in the motion fields. In contrast, the other method directly works at the image/volume level (HetSIREN). In both cases the optimization is carried on following an Encoder/Decoder approach. Zernikes3D is well suited for conformational heterogeneity, while HetSIREN addresses both conformational and compositional changes. It is to be noted that for Zernikes3D we can perform an inversion (reconstruction) process that explicitly takes into account the deformation field, while for HetSIREN this task is implicitly performed by the neural network. We will present results of Zernikes3D and HetSIREN in several systems, focusing on the study of the conformational changes happening on Epidermal Growth Factor Her2 upon binding of the therapeutic antibody Trastuzumab as well as on the complex GR-HsP90-FKBP51 in the context of a collaboration with Prof. David Agard laboratory. Additionally, we introduce a novel method for comparing conformational spaces with diverse characteristics within a unified consensus space. This approach not only simplifies the identification of potential discrepancies between spaces generated by different algorithms, but also enables the refinement of conformational states based on reliability metrics derived directly from the consensus space. Through this uniquely flexible consensus framework, we aim to facilitate researchers with tools to better interpret and trust the results produced by the latest conformational heterogeneity algorithms emerging in the field.

12