Date Published: April 01, 2016
Publisher: International Union of Crystallography
Author(s): Oleg Borbulevych, Roger I. Martin, Ian J. Tickle, Lance M. Westerhoff.
XModeScore determines the correct protomeric/tautomeric state or mode of active-site residues along with any bound ligand(s) using quantum-mechanics-based X-ray refinement followed by post-refinement scoring based on a combination of energetic strain (or ligand strain) and rigorous difference electron-density analysis.
Within structure-guided drug discovery (SGDD) and structure-based drug discovery (SBDD), accurate understanding of the protein–ligand complex structure, including the relevant proper protonation, is significant for obtaining meaningful results from docking/scoring, thermodynamic calculations, active-site exploration, lead optimization and, ultimately, medicinal chemistry (Pospisil et al., 2003 ▸). The most ubiquitous element in the universe is hydrogen, and these protons are critical for exploring the chemistry within the active site. For example, in the drug Mirapex, which is used to treat the symptoms of Parkinson’s disease, the important chemical activity is conferred by a single aminothiazole tautomeric state rather than an alternative imino tautomer (Varga et al., 2009 ▸); thus the selection of the wrong state during drug design would lead to irrelevant findings. This situation is not uncommon, and drug discovery frequently hinges on the determination of one state versus another (Martin, 2010 ▸).
In order to properly guide SBDD efforts, it is necessary to identify the correct tautomer/protomer state of the molecule in the bound state (Martin, 2009 ▸; Pospisil et al., 2003 ▸). The building blocks of common drug and drug-candidate small molecules include 5,6-membered heterocycles and various functional groups that make proton migration from one part of the molecule to another possible. Prototropy or proton-shift tautomerism represents the most common type of molecular rearrangement relevant to SBDD. Keto–enol, imine–enamine and other equilibrium types lead to hydrogen transfer between hydrogen-donor groups (e.g. —OH, —NH2) and hydrogen-acceptor atoms (e.g. =O, =N—) (Warr, 2010 ▸). While the tautomerism changes neither the molecular formula nor the molecular charge, each tautomer is a distinct chemical structure with unique physico-chemical properties. The key point is that different tautomers exist in an equilibrium in solution where the ratio between possible states is affected by the pH, temperature, concentration, ionic strength and other factors (Raczyńska et al., 2005 ▸). The general view is that protein receptors are capable of selectively binding a certain tautomeric form or forms from the mixture of several possible states (Pospisil et al., 2003 ▸). For example, the antibiotic tetracycline can exist and react in one of 64 possible tautomeric forms adapting to various chemical environments (Duarte et al., 1999 ▸). A growing body of evidence indicates that sometimes an unexpected tautomer form, or a form which does not correspond to the energy minimum of the tautomer set in vacuum, is found to react with the protein receptor (Martin, 2009 ▸).
With the calculations performed to date involving protomer/tautomer-state determination, XModeScore has shown itself to be versatile and robust. Further, while the method could be used with either QM-based refinement or conventional refinement, the significance of the QM-based results clearly appears to be noticeably higher than that observed in conventional refinement even when advanced types of ligand restraints (e.g.Mogul CIF) are employed. Another related area of interest is in the exploration of heavy-atom flip-state ambiguity often observed in macromolecular X-ray crystallography. X-ray studies of protein–ligand complexes reliably reveal only the configuration of heavy atoms of the structure, with the caveat that elements with similar atomic numbers, such as N and O, are often indistinguishable at modest resolutions. This leads to ambiguous orientations of molecule fragments capable of flipping, such as imidazole rings, amide groups and so on. Serious challenges in assigning the correct ligand orientation/flipping in X-ray macromolecular crystallography have been well documented and recognized (Malde & Mark, 2011 ▸). Often, the hypothetical flip state is chosen based upon its agreement with the hydrogen-bond network and van der Waals contacts with the residue in question (Word et al., 1999 ▸). Not only does our method offer an entirely new X-ray data-driven approach for selecting flip states, but broadly speaking, any docking/placement of a ligand within the ‘blob’ of electron density can be addressed using our method. Further studies of this phenomenon will be explored in subsequent work.