Abstract

Identification of ligand binding sites on
proteins can significantly enable chemical biology and drug discovery
applications. The best current methods to do this are based on screening large
libraries of lead-like or fragment molecules, a process that is both resource
intensive and time consuming.  We have
developed a novel, efficient and effective experimental approach to mapping
interaction sites on a protein using a designed set of small (typically ≤12
heavy atoms), halogenated compounds that express hydrogen-bonding pharmacophore
doublets, termed FragLites. The FragLites identify productive drug-like
interactions, which are identified sensitively and unambiguously by X-ray
crystallography, exploiting the anomalous scattering of the halogen
substituent.  This mapping of protein
interaction surfaces can be used (i) to assess the propensity of specific sites
on proteins to bind ligands, allowing assessment of druggability, (ii) to
identify efficient start points for the de novo design of hit molecules that incorporate
the interacting motifs present in the FragLites and (iii) as starting points in
functional studies to locate regulatory protein interaction sites. The validity
of this approach is illustrated by challenging crystals of cyclin-dependent
kinase 2 (CDK2) with a set of FragLites, successfully identifying orthosteric
(ATP-competitive) and allosteric ligand-binding sites. The orthosteric pocket
bound seven different FragLites, with six making one or more hydrogen bonding
contacts with the “hinge” region.  It is
well established that targeting the ATP-competitive site of CDK2 with drug-like
small molecules is feasible, with most such molecules forming hydrogen bonding
interactions with the hinge region. 
Hence, the observation that this region binds the FragLite molecules
with the highest frequency establishes that the mapping exercise can
successfully identify the most tractable site, and therefore contribute to
ranking and/or prioritising sites for further investigation. We are now extending our analysis to assess the
binding of the FragLite set to other drug discovery targets. We aim to confirm
whether the FragLite set identifies known protein interaction sites and can
identify potential novel protein interaction sites for further study.  In summary, the results show that FragLites
efficiently sample chemical space and are amenable to sensitive detection in
diffraction experiments.  This suggests
that the approach provides a new generic method for identifying protein-ligand
and potentially protein-protein interaction sites, which would allow for the
discovery of binding hotspots as well as assessing the tractability of a
potential drug target, to aid discovery of new leads. Fraglites are available for purchase from
Ximbio https://ximbio.com/reagent/154152/fraglites-screen-dmso

We would like to
thank the JGW Paterson Foundation (studentship award to DJW), Newcastle
University (studentship award to JDL-F) and Jordan University of Science &
Technology (studentship award to IA-K) Cancer Research UK and the Medical
Research Council for their support.