Multiscale Modeling of Enzymatic Reactions and Bioimaging Probes Abstract This project addresses a critical technology/software gap — the accessibility of ab initio quantum mechan- ical molecular mechanical (QM/MM) multiscale modeling tools in the enzymology and bioimaging fields — by developing machine-learning-based and physics-based electronic structure and molecular simulation methods. In our R01 funding period, we made several methodological advances, especially (a) the first !-learning poten- tial for simulating the enzyme reaction free energy (as demonstrated with the modeling of chorismate mutase catalysis), (b) an advance to the multiple time-step free energy simulation methodology, and (c) an analysis tool for interpreting/predicting the substitution effect on chromophore emission wavelength based on chromophore- substituent orbital interactions. Building on these method developments, we will further develop robust active-learning protocols for training !-learning potentials for ground and excited electronic states for systems in the macromolecular or solvent en- vironments. These !-learning potentials, when validated against advanced physics-based models, will enable routine (a) enzyme reaction simulations and (b) optoacoustic and other bioimaging probe modeling in our lab and the larger community. CRISPR-Cas proteins will be used as our primary test enzyme systems. We will employ the !-learning potentials in molecular dynamics simulations and seek an atomistic understanding of (a) the effect of SpyCas9 conformational changes on HNH- and RuvC-domain catalyzed DNA cleavage activity, and (b) the mechanism of AsCas12a and FnCas12a RuvC-domain catalyzed non-target strand DNA cleavage. For bioimaging probes, we will seek general guiding principles for designing optoacoustic imaging probes. It will be achieved by performing ab initio-QM/MM-quality multiscale modeling to explore the impact of struc- tural modifications (especially the attachment of halogen atoms and flexible alkane chains) on the molecular absorbance and nonradiative decay rate, both key factors in the optoacoustic signal generation.

Project Narrative This project aims to develop quantum-mechanics-based multiscale computational methods to quickly and accurately model enzymatic reactions and biomedical imaging probes. It will lead to reliable and efficient computational tools for use by the general scientific community. It will facilitate the study of enzymatic reaction mechanisms and the computer-aided design of new bioimaging probes.