The 19th Mini-Symposium on Liquids 2026

Abstract

Insulin fibrillation is a pH-dependent process that renders protein therapeutics pharmacologically inactive and potentially cytotoxic. While experimental data confirms that aggregation is favored at low and near-neutral pH, the atomic-level link between microscopic protonation behavior and macroscopic thermodynamic stability remains elusive. In this work, we employ pH-Replica Exchange Molecular Dynamics (pH-REMD), combining molecular dynamics sampling of protein-solvent coordinates with Metropolis Monte Carlo sampling of discrete protonation states, to achieve insight into the correlation between protonation fluctuations and self-association. To bridge the gap between simulation and thermodynamics, we introduce PRIME (Protonation-state Reweighting in Multi-pH Ensembles), a minimal-variance free energy estimator equivalent to MBAR, yet utilizing naturally discrete, site-unresolved titration values, PRIME allows us to derive the pH-dependent free energy of fibrillation. By evaluating the differential proton affinity between the fibril and monomer, we demonstrate that fibril formation is characterized by a significant proton retention effect at weakly acidic conditions. This creates a free energy landscape where fibrillation becomes increasingly favorable as the assembly grows ($n$). However, because the transformation from site-unresolved titration to microscopic $pK_a$ values is non-unique, we investigated the site-unresolved titration curves to investigate the microscopic mechanism. We show that insulin’s titration is defined by strong electrostatic coupling, where specific Glu residues undergo complex, multi-step re-protonation events. Structural analysis reveals that maximum inter-monomer contact is not found in the fully deprotonated state, but is instead mediated by an alternating protonation network at weakly acidic conditions. This network relieves the electrostatic frustration that otherwise leads to massive charge repulsion at high pH. In general; our findings provide a general multiscale framework for understanding how collective ionization behavior drives protein self-assembly and thermodynamic stability.

Stefan Hervø-Hansen

Specially Appointed Researcher

My research interests include molecular simulation methods, free energy calculations, and rational macromolecular design.