Kinase inhibition remains an area of significant interest across academia and in the pharmaceutical industry. There are now many marketed drugs which target kinases and a significant number of compounds are currently in various stages of clinical development. Although there have been a number of publications on kinase inhibition, this is the first to examine the future opportunities and challenges in targeting this important family of enzymes. The book is forward-looking and focuses on a number of key areas for kinase inhibition over the coming years.
About the Author
Dr Richard A Ward is a Computational Chemist, Oncology iMed, at AstraZeneca, UK. He received his BSc (Hons) in Chemistry with Bio-organic Chemistry at The University of Birmingham and gained a PhD in Computational Chemistry also at The University of Birmingham, under the supervision of Dr John Wilkie. His experience is in target selection, lead identification, lead generation and lead optimisation against kinase and non-kinase targets with specialisations in fragment-based lead generation along with library design and collection enhancement activities. He has publications in lead generation, virtual screening, druggability assessments, collection enhancement activities using computational ring enumeration along with reagent enhancement and has published supporting papers with a biological focus. He is also named as an inventor on a number of small molecule patents. Dr Frederick W Goldberg is a Medicinal Chemist at AstraZeneca, UK. He received his MSci (Hons) in Natural Sciences (Chemistry) at Cambridge University and then gained a PhD in Organic Chemistry at Imperial College, London, under the supervision of Dr Alan Armstrong. Subsequently he completed a Postdoc (AstraZeneca Fulbright scholarship) at the University of Texas at Austin, USA on "Formal synthesis of Diazonamide A", under the supervision of Dr Philip Magnus. He is presently a lead chemist on various kinase and non-kinase targets, working within the oncology lead generation group and diabetes lead optimization groups. He has publications in kinase lead generation and has filed 8 patents as primary inventor.
Read an Excerpt
Kinase Drug Discovery
By Richard A. Ward, Frederick Goldberg
The Royal Society of ChemistryCopyright © 2012 Royal Society of Chemistry
All rights reserved.
The Kinome and its Impact on Medicinal Chemistry
DAVID H. DREWRY, PAUL BAMBOROUGH, KLAUS SCHNEIDER AND GARY K. SMITH
The "Kinome" describes the protein kinase component of the human genome, exhaustively compiled in 2002 by Manning et al. A search of available human sequence sources using a hidden Markov model identified 478 ePK (eukaryotic protein kinase) genes, 491 ePK domains, and 40 "atypical" protein kinases. The tree-like classification developed from this has become the defining image in the field of protein kinase pharmaceutical research and has been used as the framework for the display of numerous properties. The ePKs are subdivided into eight main groups (Table 1.1), extending the previous classification of Hanks and Hunter.
For the last few years protein kinases have been among the most actively studied pharmaceutical targets. In 2005 it was estimated that around one in three discovery efforts targeted protein kinases. Federov et al. surveyed the landscape of kinase inhibitor publications and patents, and noted that the number of patents declined from 2006–2009, perhaps suggesting that industrial research is moving on to other areas, although this could also show that it is becoming increasingly difficult to differentiate novel compounds from prior art. However, it was also found that not all kinases have received equal attention and that the majority of protein kinases have not been studied at all. A similar trend is seen when the count of compounds from journals and patent literature, rather than the number of publications, is plotted on the kinome tree as shown in Figure 1.1 [Illustration reproduced courtesy of Cell Signaling Technology, Inc. (http://www.cellsignal.com)]. Certain tyrosine kinases, CMGC kinases, and AGC kinases have been intensively studied. In contrast, the majority of kinases have no published activity data whatsoever. A significant number of targets have no published inhibition data apart from the small number of compounds arising from high throughput profiling efforts. It is also apparent that these "untouched" kinases are unevenly distributed and that there are entire branches of the kinome tree for which small molecule inhibitors are unknown. Targets for which inhibitors exist with the selectivity necessary to make them useful tool compounds are even fewer. This distribution probably mirrors the body of literature describing the functions of these kinases, of which ~50% are said to be largely uncharacterised. Many kinases still have unknown function, and as a result there has been little incentive to seek to develop inhibitors.
This pattern is partly repeated when the analysis is restricted to the ten kinase inhibitors (Figure 1.2) that have been approved and marketed (Table 1.2). At least so far as it is possible to judge from the published stories of their discovery, nine out of the ten drugs were originally conceived as inhibitors of various tyrosine or tyrosine-like kinases. With evolution in screening technology it is possible to profile these clinical compounds against an ever expanding list of kinases. They are generally less selective than initial reports suggested, but range from the highly selective, such as Lapatinib, which targeted EGFR and ErbB2, to the fairly promiscuous, such as Sunitinib, which bound to multiple kinases.
These compounds share another feature in common besides the intent to inhibit tyrosine kinases. Nine of the ten kinase drugs are approved for oncology indications. Whilst the clinical attrition rate of kinase compounds is significantly lower than that for other antitumor agents, there are few examples of success for other indications to date. Fasudil, an inhibitor of ROCK (an AGC family kinase) is the exception, approved for acute cardiovascular disease in Japan. To some extent this may indicate that there has been a greater focus on cancer indications in kinase R&D organizations, perhaps because these are now precedented targets. This cannot be the whole story, since considerable effort has been expended in other areas, notably for inflammatory diseases.
On the positive side, if drugs for chronic conditions can be developed, they should not suffer from the problems of emerging resistance that have plagued kinase oncology drugs. However, there are significant challenges that must be overcome before kinase inhibitors can be useful for chronic diseases. The first limiting factor is the understandable unwillingness to accept risks of side-effects in chronic disease that may be acceptable in severe acute illness. All medications may suffer from on-target side-effects, but kinase drugs seem more likely than most to suffer from unexpected off-target effects because of their closely related binding sites and well-established cross-activity. A secondary problem is that a chronic disease treatment may place a greater importance on oral, twice-daily dosing in a tablet form than acute disease. To some extent these two factors may be related. Many of the cancer compounds listed in Table 1.2 have relatively high molecular weights compared to the average for oral drugs of 337. Presumably this has been brought about in part by the need to build in extra features in order to achieve greater kinase selectivity. Compounds targeting the inactive DFG-out and C-helix-out conformations are often especially large. Larger molecules generally have poorer pharmacokinetic properties, so if indeed larger molecules are needed to attain the requisite selectivity, it may be harder to find orally active compounds with appropriate selectivity for chronic diseases.
This may be made harder still depending on the choice of primary kinase target. It has been shown that tyrosine kinase inhibitors are more likely to show cross-inhibition of moderately closely related kinases than are inhibitors of CMGC kinases, for example. In the case of chronic diseases demanding a low-risk profile and greater selectivity, the historical emphasis on tyrosine kinases as targets might have contributed to the low success rate. Yet, with a few exceptions, target validation for most kinases on other kinome branches has been slower to emerge.
Table 1.3 lists some promising compounds in development for inflammation, and structures are shown in Figure 1.3. The furthest advanced of these is Tasocitinib, a JAK inhibitor currently in Phase III trials for rheumatoid arthritis (RA). Following behind are several compounds in Phase II trials for various diseases, notably SYK. Both JAK and SYK are tyrosine kinases. One of the most intensively studied targets that is not a tyrosine kinase is p38 MAP kinase.
It is instructive to consider the reasons for the failure of clinical p38 inhibitors, a target which has been extensively pursued for RA for many years without success. Early failures can be attributed to the use of a limited number of closely related chemical series and to incompletely determined selectivity profiles. For example, early pyridinyl imidazoles such as SB-203580 were thought to be p38-selective but are now known to inhibit JNKs and other targets. When truly selective molecules such as VX-702 finally entered Phase II clinical trials for RA, side-effects were relatively minor. The surprising result was the lack of efficacy of bioavailable inhibitors for reasons that remain unclear. It is striking that early signs of efficacy on inflammatory biomarkers diminished over time, suggesting that an unknown compensation mechanism engages in the RA disease process, and calling into question the established RA disease models. These results may have finally invalidated p38 as a target for RA, but its suitability as a target for other diseases is perhaps enhanced by the lack of severe side-effects. Interestingly, a p38 inhibitor has achieved positive proof of concept in a study of dental pain.
This example reinforces the critical importance of reliable target validation for a given disease. It seems likely that other kinases yet to emerge could impact chronic diseases but not suffer from such unforeseen target/disease related problems. If these are not tyrosine kinases, it may be easier to obtain selective inhibitors to avoid off-target effects while remaining in drug-like chemical space. While research continues into tyrosine kinases for chronic diseases, other kinases with fewer close homologues are emerging as targets, for example mTOR, IKKβ, MK2, and lipid kinases, and these may prove more tractable, provided their intrinsic functions allow for a safe therapeutic window.
As mentioned above, nine of the ten approved kinase inhibitor drugs were originally conceived as inhibitors of various TK or TKL kinases. If success in kinase drug discovery is measured by drugs on the market, then success has been limited to oncology, and to this branch of the kinome. In order to move towards a wider exploitation of kinases as drug targets, we need small molecule inhibitors with suitable potency, selectivity and cell activity that can be used to elucidate the biology of unexplored kinases. To identify these chemical probes we need to have assays to understand inhibition profiles, chemical starting points for optimization, measurements of selectivity, and strategies to modulate selectivity. We will now touch on each of these areas.
1.2 Kinome Scale Assays
Understanding the affinity of new inhibitors across multiple kinases will be important to exploring the therapeutic value and liability of the unexplored kinase targets. The next section of this review will focus on the commercial assays available to assess broad kinase inhibitor potency. These assays are currently expanding the breadth of the kinase world and in the future will enable production of inhibitors for most of the kinases on the kinome.
The publication of the kinome tree dramatically increased the challenge and potential for kinase therapeutics. This challenge was reflected in both the recognition that selectivity within the protein kinase family was defined to a universe of about 518 individuals and in the recognition that an assay for each individual kinase was needed to assess the selectivity of compounds in development. It was immediately understood that the presence of a highly conserved ATP binding site throughout the kinome would make achieving selectivity challenging.
In contrast, recognition of the conserved ATP site also led to the realization that "lead hopping" from an inhibitor of one kinase to an inhibitor of another kinase, closely or more distantly related, was feasible. It may be possible to find ligands for each member of the tree via this lead hopping and cross screening through the kinome. Indeed GSK has used this strategy to develop a number of kinase leads and candidates. For example, Figure 1.4 shows the source of GSK kinase leads or Starts of Chemistry from 2002-1Q2009.
The data show that from 2002 to 1Q2009 the largest source of leads or SoC over this period consistently came from the kinase cross-screen (X-screen). The X-screen process is depicted schematically in Figure 1.5. Over half (53%) of all leads identified derived from this approach. For our purposes, a hit is an active compound that comes out of a screen, and a lead is a more advanced compound that has significant promise such that one has confidence that an optimization effort will lead to a clinical candidate. The other two target class based screening methodologies produced 17% and 20% of the leads/SoC, respectively. In total, about 90% of kinase leads/SoC from 2002 to 1Q2009 derived from target class based focused screening – a combination of X-screening, focused screening, and knowledge based design. HTS accounted for about 11% of the leads/SoC.
Stavenger and colleagues described an excellent example of the power of the cross screen to identify a starting point that can be turned into a useful tool molecule. In this case the molecule shown on the left in Figure 1.6 was made for our MSK1 program. The Rho kinase 1 (Rock1) program routinely screened all new kinase chemistry and found this compound to be a 19 nM IC50 hit. However, the molecule was not selective for Rock1. It exhibited mid to low nM inhibition of other AGC kinases including RSK1, p70/S6K, MSK1 and sub-micromolar inhibition of a number of other kinases. Application of several rounds of medicinal chemistry to the hit provided the elaborated molecule shown on the right of Figure 1.6 with enhanced selectivity and useable pharmacokinetic parameters that the team utilized for in vivo target validation experiments.
We believe that the target class based focused screening was so successful because of the highly conserved ATP binding site in all protein kinases (and many lipid kinases). The Kinase Compound Set (KCS) was designed to target this ATP site, and the X-Screen exposed the target to the newest GSK kinase chemistry thought processes (largely targeting this site). These two approaches provided 70% of chemical starting points. Finally, knowledge based screening of compounds designed for this site adds another 20% of historical chemical starting points.
These on-going kinase medicinal efforts at GSK and other companies increased the demand for additional kinase selectivity assays to thoroughly assess development and tool potential for these compounds coming from corporate and academic screens. To address this within GSK, a small, representative kinome panel (~15% of the kinome at any time) was developed and screened weekly along with targets. In general we found this quite predictive of broader kinome selectivity. However, we did not find it feasible to develop a panel of kinase assays approaching the full scale of the kinome for selectivity or hit identification, and because of known and unknown liabilities associated with inhibiting the other ~85% of the kinome, therapeutic project teams sought the means to assay the broader kinome. This limitation at GSK and other pharmaceutical companies was one impetus for the development of commercial kinome panels.
The table in Appendix 1.1 shows the status of kinome assay panels (as of October 2010) that have been developed (and are still growing) for nine of the major contributors to this field. In the table, the first column lists the kinases from the Sugen kinome paper, plus some lipid kinases, mutant kinases and specific constructs or activation states. The next five columns include gene names and aliases along with family information from the Sugen publication. The remaining columns list the kinase assays commercially available from the suppliers; these assays are aligned to the Sugen names. Within the individual supplier columns, the nomenclature of the supplier was maintained where feasible in order to facilitate using this information when communicating with suppliers.
The sizes of these panels vary from about 200 kinases at Caliper to over 400 at Ambit. The assay counts include the lipid kinases, mutant kinases and kinases that are available in more than one assay condition (the number of wild type kinases in any of the panels is less than the total). These assay panels enable assessment of broad selectivity and liability information needed for lead and candidate characterization. They also enable broad screening for chemical starting points for kinases across the kinome in parallel. Indeed, we reported doing this in 2008 against the 203 kinases available in the Ambit panel at that time, which will be discussed elsewhere in this review.
The kinome panels at the nine suppliers in Appendix 1.1 are not all the same, and the supplier diversity provides value. They differ in composition, kinase constructs, assay technology, assay conditions, kinase and ATP concentration, and in kinase substrate selection. This diversity enables the kinome scientist to select the kinase and, sometimes, the assay and conditions to answer specific questions.
By combining assays from the nine providers in Appendix 1.1, inhibitor binding for up to 407 wild type protein or lipid kinases can be assessed. The data in Appendix 1.1 shows that the kinase families can be well covered by a combination of the commercial assay panels. Where one panel may be deficient, other panels provide coverage.
All of the providers offer screening at a single concentration of compound (giving %Inhibition at that concentration) as well as IC50 or Kd determination, and they all also provide some mechanism of action work for all kinases in the panel. In addition, all providers offer screens against a number of clinically important mutant protein or lipid kinases. Most providers also enable the scientist to view target profiling and selectivity data graphically, with a visualization tool. Some also have nonhuman and non-mammalian kinases available (not included in this review or Appendix 1.1).
Excerpted from Kinase Drug Discovery by Richard A. Ward, Frederick Goldberg. Copyright © 2012 Royal Society of Chemistry. Excerpted by permission of The Royal Society of Chemistry.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
An Introduction to Kinase Inhibition; The Kinome and its Impact on Medicinal Chemistry; Contemporary Approaches to Kinase Lead Generation; The Learning and Evolution of Medicinal Chemistry against Kinase Targets; The Mechanism and Kinetics of Kinase Inhibitors; Kinase mutations and resistance in cancer; Non-Protein Kinases as Therapeutic Targets; The Drug Discovery and Development of Kinase Inhibitors Outside of Oncology; Allosteric Activators of Glucokinase (GK) for the Treatment of Type 2 Diabetes; Drug Discovery and Non-Human Kinomes; The future of kinase drug discovery