Dissertation/Thesis Abstract

Quantifying Protonation State Effects in Protein-ligand Interaction
by Rustenburg, Ariën Sebastiaan, Ph.D., Weill Medical College of Cornell University, 2019, 308; 27543710
Abstract (Summary)

The existence of many easily-accessible protonation state and tautomers is a ubiquitous phenomenon in proteins, nucleic acids, and small molecules. Many of the standard amino acids, for example, have more than one accessible protonation state at physiological pH.

The prevalence of multiple accessible tautomers and protonation states in biochemistry is even more widespread for small molecule drugs, many of which feature easily titratable moieties. Despite this, the majority of biomolecular simulation techniques model only a single fixed protonation/tautomeric state — due to a combination of both technical limitations and lack of appreciation for the importance of protonation state and tautomeric effects. When a system is studied computationally, it begs the question as to what the quantitative effect of incorporating (or neglecting) protonation states and tautomers into molecular simulation is, specifically in the context of protein-ligand interaction.

The aim of this work is to contribute to knowledge and the toolset available to the computational chemist for accounting for protonation state effects computationally. One part of this involves the assessment of pKa prediction tools, and philosophical discussion of the means by which we should be assessing performance of pKa prediction methods. Another contribution discussed in this work is the development of a strategy for simulation of protein-ligand complexes with explicit treatment of protonation states by combining existing algorithms into the protons simulation package. This novel simulation package is built on the foundations of the open source, GPU-accelerated codebase of OpenMM. For the validation of computational predictions of protonation state effects, the availability of reliable experimental data is a necessity. To this end, I have worked on the improvement of the experimental analysis procedures of isothermal titration calorimetry (ITC) experiments. The enthalpic information that can be obtained from ITC experiments is frequently used for determining protonation state effects, but the estimation of enthalpies from ITC is remarkably sensitive to concentration errors. I have added to the reliability of ITC data with the help of baseline integration methods, and the development of Bayesian models which include prior information on experimental error. This should lead to lower errors, while also making more reliable estimation of the uncertainty in the experimental parameters. Additionally, I've performed distribution coefficient (log D) measurements which can be used as references to parametrize and assess the performance of computational models. These physical-chemical properties are sensitive to protonation states, and provide a great simple model system to test hypotheses about protonation states against.

Indexing (document details)
Advisor: Chodera, John D
Commitee: Boudker, Olga, Accardi, Alessio, Aksay, Emre, Gunner, Marilyn
School: Weill Medical College of Cornell University
Department: Physiology, Biophysics & Systems Biology
School Location: United States -- New York
Source: DAI-B 81/4(E), Dissertation Abstracts International
Subjects: Computational chemistry, Analytical chemistry
Keywords: Distribution coefficients, Isothermal titration calorimetry, Molecular dynamics, pKa, Protonation
Publication Number: 27543710
ISBN: 9781392776865
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy