Accurate Force Fields for Spectroscopic Studies of Protein–Ligand Interactions and Self-Assembly Structures

Sammanfattning: The computational prediction of complex molecular behaviors is an essen- tial component of modern chemistry, as it provides a faster and more cost- effective way to explore molecular interactions that may be difficult or even impossible to study experimentally. Molecular dynamics (MD) simulations of- ten serve as a valuable tool for such predictions; however, their accuracy is inherently dependent on the force field (FF) parameters employed. While the general amber force field (GAFF) is designed to provide reasonable results for a broad array of small molecules, it often requires further refinement when using it for a specific small organic molecule. Especially for ligands of the oligothiophene class, the dihedral potential representing the rotatable bond between the two thiophene rings (of the SCCS type) is inadequately described. An objective of this dissertation is to refine FF parameters for producing meaningful MD trajectories that capture key molecular interactions, binding modes, and thermodynamic properties, and subsequent accurate calculations of spectroscopic properties. The refined FF parameters were first tested by comparing the dihedral potential derived from the FF method to the density functional theory (DFT) based dihedral potential. They were then validated by assessing the relative energies of conformers optimized using both FF and DFT methods, and by comparing the transition wavelengths calculated based on geometries optimized with both FF and DFT approaches. Importantly, the errors in dihedral potential were kept below 1 kcal/mol, and the discrepancies in transition energies were less than 0.1 eV for molecular transitions around 5 eV. This FF parametrization methodology was used in research studies focus- ing on two classes of supramolecular systems: host-guest chemistry related to neurodegenerative diseases, and self-assembly systems for material development. Specifically, we examine host-guest interactions involving proteins such as amyloid-beta, tau, and transthyretin (TTR), which are associated with neu- rodegenerative diseases. Various fluorescent ligands are used for the detection of these proteins in pathological samples. Our results for these protein–ligand systems propose strong binding sites and modes, and include estimations of binding energies for different ligands interacting with the targeted proteins. Additionally, comparative studies among the ligands have been conducted. Interestingly, no fluorescence was observed when low binding energy ligands interacted with amyloid fibrils. In the case of bTVBT4 binding to tau associated with Alzheimer’s disease (AD), a unique binding site was identified. This site was not accessible in the tau fold found in Pick’s disease (PiD), thus explaining the specificity of bTVBT4 for AD-related tau. For self-assembly systems, our findings encompass spectral profiles altered by tyrosine substitutions in oligothiophenes, a stable self-assembly model formed by chiral sulfonimidamides that explains the involved interactions, and comparisons of experimental circular dichroism (CD) profiles to assign isomers of [4]cyclonaphthodithiophene diimides to specific spectral profiles. We also investigated the solvent effects on the spectroscopic properties of symmetric and asymmetric azaoxahelicenes. In conclusion, the methodological development of FF parameters provides a robust framework for accurately modeling the behavior of complex supramolecular systems. The improvements in the dihedral potential align closely with DFT-based calculations, thereby elevating the predictive power of MD simulations for both binding modes and subsequent spectroscopic properties. The research has direct applications in the detection of neurodegenerative diseases at an early stage by designing fluorescent ligands that specifically bind to targeted proteins. It also contributes to the creation of advanced materials with finely-tuned properties. Furthermore, the methodology employed can serve as a blueprint for future studies aiming to refine computational models for other classes of molecules.