Abstract Details

Poster 5: Scaffold-Focused Virtual Screening to Identify Novel Kinase inhibitors

Sarah R. Langdon1, Nathan Brown1, Julian Blagg1
1In Silico Medicinal Chemistry, Cancer Research UK Cancer Therapeutics Unit, The Institute of Cancer Research
Kinases control many cellular processes including, but not limited to, signal transduction, cell cycle, apoptosis and transcription.1 Their deregulation can allow cancer cells to evade normal physiological constraints,2 therefore kinase have become a major target for cancer therapy.

The majority of kinase inhibitors target the adenosine triphosphate (ATP) binding site by forming hydrogen bonds to the amino acid backbone of the hinge region.2 This hinge region is highly conserved throughout the kinase family.2 Kinases have also become one of the most pursued classes of drug target in recent years.3 For these two reasons kinase inhibitor chemical space has been extensively mined leading to crowded intellectual property (IP) space.

Scaffold Hopping is a technique used to identify compounds with similar activity to known bioactive compounds, but with a novel structure.4 Our aim is to develop in silico tools that can be used by a medicinal chemist to find novel kinase inhibitor scaffolds derived from existing scaffolds or compounds of interest. In principle, such a technique could also be applied to inhibitor scaffolds across multiple target classes.

We have developed and applied a scaffold hopping protocol based on Scaffold Trees.5 This protocol includes a similarity search performed between a query scaffold and compound library represented as Level 1 scaffolds from the Scaffold Tree. A prospective validation of this protocol identifies active compounds that are structurally differentiated from the query.

We identified some limitations of Level 1 scaffolds as applied to our scaffold hopping protocol, and therefore sought an alternative scaffold definition for application to scaffold-focused virtual screens. We require a scaffold representation that more consistently identifies the most information rich scaffold; for example a method that avoids the rupture of fused, bridged or spiro ring systems or loss of highly connected and functionalised ring systems. The identity of scaffold substituents has an important influence on biochemical activity, (i.e. an unsuitable substituent can lead to loss of activity of an otherwise active scaffold); therefore, we also sought a method to annotate scaffolds with the position and properties of appended substituents.

This work builds on our previous virtual screens to provide improved in silico methods for identifying structurally novel kinase inhibitors and includes the development of a scaffold focused virtual screen as well as the development of a novel medicinal chemistry relevant scaffold representation for use in virtual screening.

We have developed SID (Scaffold Identification and Description), a new method for representing the scaffold of a molecule and describing the scaffold substituents.

SID uses simple scores designed to retain information-rich ring systems. These scores are based on how connected the ring system is, the number of heteroatoms in the ring system and the number of sp3 hybridized atoms. Once the scaffold has been identified the cut points at which the scaffold is attached to the rest of the molecule are annotated with dummy atoms. In order to retain more information on the substituents at the cut points, these dummy atoms represent 7 different atom types that describe the nature of the atom or functional group closest to the cut point.

We will present the implementation of SID in a scaffold focused virtual screen in which a similarity search is performed between a query compound of known activity and a vendor collection; the query and compounds in the vendor collection are represented as scaffolds using SID. We will present retrospective and prospective validations using this virtual screen method and results of these validations will be compared to those of the equivalent virtual screen using Level 1 scaffolds and a whole molecule similarity virtual screen.


1. P. Bamborough, D. Drewry, G. Harper, G. K. Smith, and K. Schneider. Assessment of chemical coverage of kinome space and its implications for kinase drug discovery. Journal of Medicinal Chemsitry, 2008, 51, 7898-7914.
2. J. Zhang, P. L. Yang, and N. S. Gray. Targeting cancer with small molecule kinase inhibitors. Nature Reviews Cancer, 2009, 9, 28-39.
3. P. Cohen. Protien kinases - the major drug targers of the twenty-first century? Nature Reviews Drug Discovery, 2002, 1, 309315.
4. S. R. Langdon, P. Ertl, and N. Brown. Bioisosteric Replacement and Scaffold Hopping in Lead Generation and Optimization. Molecular Informatics, 2010, 29, 366385.
5. A. Schuffenhauer, S. Roggo P. Ertl, S. Wetzel, M. A. Koch, and H. Waldmann. The Scaffold Tree - Visualization of the Scaffold Universe by Hierarchical Scaffold Classification. Journal of Chemical Information and Modeling, 2006, 47, 4758.

Return to Programme