Research in the Li Group
Metal ions are indispensable for life, driving important processes such as respiration, photosynthesis, signal transduction, enzymatic catalysis, and electron transfer. Approximately 40% of all proteins require metal ions for activity, yet the mechanisms of many metalloproteins remain poorly understood. Modeling these systems is especially challenging due to the complex properties of metal ions. My group integrates theoretical and computational chemistry at the interface of biological and inorganic chemistry. We develop accurate and efficient metal ion models and apply them to fundamental questions in metalloprotein function. Our goals are twofold: (1) to gain mechanistic insight into metalloproteins, and (2) to leverage these insights for molecular design in biomedical, catalytic, and materials contexts. To achieve this, we create efficient and accurate computational tools tailored for metal ion-containing systems.
Direction 1. Accurate and Efficient Physical Models for Metal Ions
It is challenging to simulate the intricacies of metal ion-ligand interactions due to the involvement of d- and f-orbitals, multiple oxidation and spin states, and flexible coordination numbers.[Chem. Rev. 2017, 117, 1564-1686] Three factors—(1) ion size, (2) polarization, and (3) charge transfer—are central to these interactions but are often poorly represented in existing models. Over the past five years, my group has advanced the modeling of metal ions through several contributions: (1) Ion Size—We systematically characterized ion sizes, including van der Waals radii [J. Chem. Theory Comput. 2023, 19, 2064-2074.] and “absolute” radii,[J. Phys. Chem. A 2025, 129, 5118-5126] across the periodic table. (2) Polarization—We calibrated ion polarizabilities of ions across the periodic table and quantified their relationship with ion sizes, and developed a general atom-in-molecule scheme (i.e. the 12-6-4-NBFIX method) to incorporate ion-induced dipole effects.[J. Chem. Theory Comput. 2024, 20, 8505-8516.] (3) Charge Transfer—We clarified the relationship between the 12-6-4 model and the fluctuating charge model [Front. Chem. 2021, 9, 721960.] and developed a novel fluctuating charge model for transition metals.[J. Chem. Inf. Model. 2024, 64, 812-824; J. Phys. Chem. B 2024, 128, 10329-10338] In addition, we have validated the accuracy of the discrete–continuum model for simulating the hydration of divalent and highly charged metal ions.[J. Phys. Chem. B 2024, 128, 11904-11913] Building on these successes, our future research will focus on efficiently incorporating many-body interactions into force fields for metal ions, Together, these efforts provide a comprehensive toolkit for modeling metal ions in metalloproteins and other complex systems—advancing molecular modeling in my group and serving the broader scientific community.
Direction 2. Physics-Informed Machine Learning Models for Metal Ions
Metal-ligand interactions are ubiquitous in chemical, biochemical, and materials systems, yet modeling them requires a difficult trade-off between accuracy and efficiency. While recent advances in pure machine learning (e.g., kernel, tree, or neural network-based approaches) show promise, they often suffer from limited transferability outside their training data. To address this, we aim to develop physics-informed machine learning models (or machine learning-assisted physical models) that predict the energetic properties of metal-containing systems with high accuracy, computational efficiency, and superior transferability. Our primary focus is on constructing accurate Potential Energy Surfaces (PES) for molecular dynamics and predicting binding affinities. These enhanced models will accelerate discovery and design in fields ranging from biomedicine to materials science and catalysis.
Direction 3. Ion Selectivity in EF-hand Containing Proteins
About two dozen metal ions are biologically available, coexisting in intracellular and extracellular fluids, but only specific ones are selected for proper protein function. Mis-selection can impair efficiency or cause toxicity. Understanding ion selectivity is therefore critical for treating metal-related diseases, developing metallodrugs, designing metalloproteins, and engineering systems for ion separation. The EF-hand loop is a versatile ion-binding motif found in proteins such as Lanmodulin, Lanthanide-Binding Tags, Troponin C, and Calmodulin. Despite its importance, the molecular determinants of EF-hand ion selectivity remain elusive, partly due to the complexity of polarization and charge transfer effects in metal binding. By applying force field models, we are simulating EF-hand proteins to reveal the thermodynamic, structural, and dynamic bases for the ion binding.[J. Chem. Inf. Model. 2023, 63, 354-361] The recently awarded R35 MIRA grant (Award Number: R35GM160470) for this project is allowing my group to expand this research direction through (1) illuminating unique ion selectivity patterns across diverse EF-hand proteins, (2) elucidating the role of ligands in tuning ion affinity, (3) deciphering the influence of protein environment and dynamics on ion binding, and (4) unraveling mutational impacts on ion binding kinetics. These insights will enable the design of proteins with customized ion specificities for applications in medicine and materials science.
Direction 4. Proton-Coupled Electron Transfer (PCET) in Metalloproteins
Electron transfer (ET), proton transfer (PT), and proton-coupled electron transfer (PCET) are central to many biological and catalytic processes. Among them, PCET is especially complex, as it combines redox and acid–base chemistry, occurring either stepwise or concertedly. Quantum effects, arising from the small masses of protons and electrons, further complicate the modeling. Our group investigates ET, PT, and PCET mechanisms in metalloproteins using advanced theoretical and computational approaches. We have provided mechanistic insights into a series of important metalloproteins: (1) electron transfer in lignin peroxidase,[JACS Au 2023, 3, 536-549.] (2) PCET catalysis in glycosylated fungal lipoxygenase,[Biochemistry 2023, 62, 1531-1543] (3) proton pumping in cytochrome c oxidase,[Phys. Chem. Chem. Phys. 2023, 25, 25105-25115.] (4) redox properties of the copper center in azurin,[J. Inorg. Biochem. 2024, 259, 112651.] and (5) electron transfer in dihydroorotate dehydrogenase 1B.[Biochemistry 2025, 64, 3971–3985.] These studies not only illuminate fundamental processes but also guide the rational design of enzymes, catalysts, and inhibitors.
