The Future of Structural Biology: Artificial Intelligence and in Situ Structures

Inselhalle

Conference Room 1–3

Moderator: Wolfgang Lubitz
Max Planck Institute for Chemical Energy Conversion, Germany

Johann Deisenhofer

It has been known since the 1960s that the three-dimensional structures of proteins should be deductible from their amino acid sequence and physical principles. However, until recently, the computational requirements exceeded by far the available resources. The accumulation of experimental 3-D structures in the Protein Data Bank and of a vast number of sequences allowed the use of statistical methods that brought the goal within reach. The remaining gap between predicted and experimental structures was closed by the addition of Artificial Intelligence methods, as documented in the recent rounds of CASP (Critical Assessment of Structure Prediction).

Hartmut Michel

The precise knowledge of the atomic structures of biological macromolecules like enzymes and of the intermediates of their catalytic cycles forms the basis to understand their mechanism of action. Our main interest lies in cytochrome c oxidase, a membrane bound enzyme which reduces oxygen to water generating electric voltages and pH gradients across membranes. Surprisingly, the two major methods to determine the high-resolution structures, namely X-ray crystallography and single particle electron cryomicroscopy, lead to different, incompatible results. These will be presented and discussed.
An interesting future of structural biology lies in the determination of atomic structures of biological macromolecules in their native environment inside biological cells ("in-situ" structures). Recent progress will be reviewed.