Computer-Aided Drug Design of G-quadruplex Structures: Harnessing Polarization for Rational Drug Design

Abstract

G-quadruplexes (GQs) are noncanonical nucleic acid structures formed by guanine-rich se- quences that adopt stable, planar arrangements of guanines stabilized by Hoogsteen hydro- gen bonding. These structures are found throughout the human genome and in numerous pathogens, including viruses and parasites. GQs play important roles in transcriptional and translational regulation, genome stability, and replication, making them attractive thera- peutic targets in cancer, neurodegeneration, and infectious diseases. However, the lack of structural selectivity in current GQ-targeting ligands has limited their clinical success. This dissertation focuses on improving the rational design of GQ-binding ligands by advancing computational methods that better capture the structural and electrostatic properties of GQs. Central to this effort is the use of the classical Drude oscillator polarizable force field model to more accurately describe GQ dynamics, ion interactions, and ligand binding ener- getics. Through an in-depth characterization of the HIV-1 LTR-III GQ, we reveal how local electric fields and base dipole moments influence its conformational behavior, reinforcing the importance of polarization in GQ modeling. Building on these insights, we developed a novel workflow, SILCS-Nucleic, to extend Site Identification by Ligand Competitive Sat- uration (SILCS) methodology to nucleic acids using the Drude force field. SILCS-Nucleic enabled fragment-based mapping and pharmacophore generation with enhanced electrostatic fidelity. The utility of this approach was validated across diverse DNA and RNA systems, and ultimately used to initiate an early drug discovery campaign targeting the HIV-1 LTR GQs. Computational methods were used to identify novel chemical scaffolds for both LTR GQs, while experimental methods were additionally used to determine the ability of these scaffolds to stabilize LTR-III. Altogether, this research demonstrates the value of physics- based computational modeling for guiding structure-based drug discovery of nucleic acid targets. The workflows developed here lay the groundwork for more selective, informed lig- and design, with the potential to broaden GQ-targeting therapeutics beyond oncology into the realm of infectious disease.