Synthetic, Biochemical, X-ray Crystallographic, Computational and High-Throughput Screening Approaches Toward Anthrax Toxin Lethal Factor Inhibition

Abstract

The lethal factor (LF) enzyme secreted by Bacillus anthracis is chiefly responsible for anthrax-related cytotoxicity. In this dissertation, I present the computational design, synthesis, biochemical testing, structural biology, and virtual and high-throughput screening approaches to identify binding requirements for LF inhibition. To this end, we designed ~50 novel compounds to probe design principles and structural requirements for LF. Specifically, in Chapters 2 and 3, computational, synthetic, biochemical and structural biology methods to explore the underinvestigated LF S2′ binding subsite are described. We discovered that LF domain 3 is very flexible and results in a relatively unconstrained S2′ binding site region. Additionally, we found that the S1′ subsite can undergo a novel conformational change resulting in a previously unreported tunnel region, which we term S1′*, that we expect can further be explored to design potent and selective LF inhibitors. Using this novel LF configuration, we virtually screened ~11 million drug-like compounds for activity against LF and have identified a novel compound that inhibits LF with an IC50 of 126 μM. In the course of this work, we found that reliable representation of zinc and other transition metal centers in macromolecules is nontrivial, due to the complexity of the coordination environment and charge distribution at the catalytic center. In Chapter 7, I will present work on applying and optimizing quantum mechanical methods developed by the Truhlar group to accurately calculate bond dissociation energies at low computational cost for various representative Zn2+ and Cd2+ model systems. By analyzing errors, we developed a prescription for an optimal system fragmentation strategy for our models. With this scheme, we find that the EE-3B-CE method is able to reproduce 53 conventionally calculated bond energies with an average absolute error of only 0.59 kcal/mol. Therefore, one could use the EE 3B CE approximation to obtain accurate results for large systems and/or identify better parameters for Zn centers for use in virtual screening. Finally, we present the results of a large-scale in vitro HTS campaign of ~250,000 small-molecules against LF. After extensive validation, involving secondary assays and hit synthesis we were able to prioritize a key lead for further prosecution.