Hybrid quantum / classical simulations of enzymes

Background.

Enzymes are omnipresent in biological systems and are key components of circular bioengineering efforts. Catalyzing chemical reactions, they invariably are involved in the making and breaking of chemical bonds. Moreover, many of the enzymes that are of interest in BioToP are metalloenzymes or driven by either light or electron transfer. All of these properties require the inclusion of quantum mechanical methods in appropriate in silico descriptions. Arguably, enzymes are most efficiently described by hybrid quantum and classical mechanical approaches (QM/MM). In this project we will further develop the description of QM/MM Hamiltonians using advanced QM methods and machine-learned potentials.

The computational inefficiency of most quantum mechanical methods hamper either a full description of enzymes at the QM level or a sufficiently extensive sampling to describe both enthalpic and entropic effects appropriately. QM/MM approaches offer solutions that combine the best of both worlds. Molecular dynamics simulations can be coupled with a plethora of QM methods (Poliak, 2025), specifically suitable for specific questions. Recently tremendous progress has been made in the use of machine-learned potentials, which learn quantum mechanical interactions from an appropriate training set. The inclusion of these methods in QM/MM settings, furthermore, allows us to resolve interfacial limitations of traditional QM/MM methods (Lier, 2022; Crha, 2025). Inclusion of reactivity and enzymatic properties of proteins through electrostatic and dynamic effects are among the next challenges to tackle computationally.

Research objectives.

• Description of enzymatic reactions via QM/MM methods
• Inclusion of dynamic effects in quantum mechanical descriptions of heme proteins
• Inclusion of enzymatic reactivity in machine-learned hybrid QM/MM calculations tures. Experimental validation of predictions may be performed in collaboration with other doctoral candidates in the program.

Collaborations.

This project is also part of the Cluster of Excellence Circular Bioengineering. We will model and describe various enzymatic systems of interest within Cluster of Excellence and within BioToP: Enzymes that degrade or modify lignocellulosic materials (Roland Ludwig, Doris Ribitsch); Enzymes responsible for regio- and stereospecific hydroxylation by e.g. a-ketoglutarate dependent enzymes (Wolfgang Kroutil), and/or the mono-oxygenase AlkB (Rober Kourist); Heme containing enzymes (Stefan Hofbauer).

REFERENCES
Crha, R., Poliak, P., Gillhofer, M., Oostenbrink, C. (2025) Alchemical Free-Energy Calculations at Quantum-Chemical Precision. J Phys Chem Lett 16, 863-869. doi: 10.1021/acs.jpclett.4c03213
Lier, B., Poliak, P., Marquetand, P., Westermayr, J., Oostenbrink, C. (2022) BuRNN: Buffer Region Neural Network Approach for Polarizable-Embedding Neural Network/Molecular Mechanics Simulations. J Phys Chem Lett 13, 3812-3818. doi: 10.1021/acs.jpclett.2c00654
Poliak, P., Bleiziffer, P., Pultar, F., Riniker, S., Oostenbrink, C. (2025) A Robust and Versatile QM/MM Interface for Molecular