Current Compound Coverage of the Kinome - Journal of Medicinal

Perspective pubs.acs.org/jmc

Current Compound Coverage of the Kinome Miniperspective Ye Hu,† Norbert Furtmann,†,‡ and Jürgen Bajorath*,† †

Department of Life Science Informatics, B-IT, LIMES Program Unit Chemical Biology and Medicinal Chemistry, Rheinische Friedrich-Wilhelms-Universität, Dahlmannstrasse 2, D-53113 Bonn, Germany ‡ Pharmaceutical Institute, Rheinische Friedrich-Wilhelms-Universität, An der Immenburg 4, D-53121 Bonn, Germany S Supporting Information *

ABSTRACT: Publicly available kinase inhibitors have been analyzed in detail. Nearly 19000 inhibitors have been identified with activity against 266 different kinases. Thus, about half of the human kinome is currently covered with active small molecules. The distribution of inhibitors across the kinome is uneven. Most available kinase inhibitors are likely to be type I inhibitors. By contrast, type II inhibitors are rare but usually have high potency. Kinase inhibitors generally display high scaffold diversity. Activity cliffs with an at least 100-fold difference in potency are only found for inhibitors of 106 kinases, which is partly due to only small numbers of compounds available for many kinases, in addition to scaffold diversity. Moreover, kinase inhibitors are less promiscuous than often thought. More than 70% of available inhibitors are only annotated with a single kinase activity, and only ∼1% of the inhibitors are active against five or more kinases.

■

hydrophic region adjacent to the ATP site.9,10 Thus, these inhibitors stabilize the inactive state of kinases. Furthermore, there are type III and IV inhibitors that bind to different regions outside the ATP site and act through more or less pronounced allosteric effects.7,9 For example, CI-1040 is a well-known type III inhibitor of mitogen-activated protein kinase kinase11 and GNF-2 is an allosteric inhibitor that binds to the four-helix bundle of ABL kinase.12 Allosteric inhibitors are anticipated to have the highest potential to yield kinase selective or specific inhibitors.7 However, to date, only very few such inhibitors have been characterized. For example, Ashwell et al. identified an allosteric imidazopyridine-based inhibitor that was selective for protein kinase B over ∼300 other kinases.13 Recently, publicly available X-ray structures of kinases in complex with type I and type II inhibitors have been systematically analyzed and compared,9 hence providing a comprehensive structural basis for studying inhibitor binding characteristics. Type II inhibitors are often thought to be more selective than classical type I inhibitors because the “DFG-out” induced subsite they bind to is less conserved across different kinases than other regions of the ATP site.7,9 However, kinase profiling experiments using type II inhibitors have questioned such assumptions, revealing that many type II inhibitors are also promiscuous.9 Hence, much remains to be learned about kinase inhibitor selectivity, further emphasizing the need to explore allosteric or other mechanisms of action potentially

INTRODUCTION Over the past 2 decades, protein kinases have been (and continue to be) among the most popular therapeutic targets,1 and kinase inhibitors are intensely investigated, especially for cancer treatment.1,2 In 2001, imatinib was the first kinase inhibitor approved as a drug, and today there are at least 23 marketed kinase inhibitors. Most kinase inhibitors reported thus far, including marketed drugs, target the ATP (cofactor) binding site that is largely conserved in kinases.3−6 ATP site directed kinase inhibitors have often been found to display limited selectivity,3−6 consistent with the conservation of the ATP binding site. Promiscuity of such inhibitors might often be functionally relevant and is thought to be a hallmark for cancer therapy.1,3 However, if kinase inhibitors are to be used for other therapeutic applications, especially for the treatment of chronic (inflammatory) diseases, a high level of target selectivity is likely to be required.7,8 Accordingly, increasing attempts are being made to generate kinase inhibitors that have further improved selectivity or bind to less conserved regions in the catalytic domain of kinases outside the ATP site.7,8 Currently, four different categories or types of kinase inhibitors are distinguished according to different modes of actions elucidated by X-ray crystallography.7,9 Type I inhibitors are competitive ATP site directed inhibitors that bind to the active forms of kinases that are characterized structurally by the so-called “DFG-in” conformation of the activation segment proximal to the active site.9 These type I inhibitors represent the “classical” category of kinase inhibitors. By contrast, type II inhibitors bind to the inactive “DFG-out” conformation of the activation segment that makes different subsites accessible in a © XXXX American Chemical Society

Special Issue: New Frontiers in Kinases Received: May 28, 2014

A

dx.doi.org/10.1021/jm5008159 | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry

■

Perspective

KINASE INHIBITORS We identified a total of 18951 kinase inhibitors with activity against 266 human kinases that formed a total of 27063 kinase−inhibitor interactions, as reported in Table 1. These

leading to specific kinase inhibition. However, in this context, it should also be noted that promiscuity and selectivity of kinase inhibitors have thus far only been investigated for limited numbers of inhibitors,4 although general conclusions have frequently been drawn from such studies. While available structural data and kinase inhibition modes have been intensely studied,9,10,14 little is currently known about the global distribution of inhibitors across the kinome. The human kinome contains more than 500 different kinases15 and for about 450 kinases, enzyme assays are available,8 hence providing extensive opportunities for inhibitor discovery. Two decades of intense drug discovery efforts targeting kinases have yielded large amounts of compound activity data, and much of this information has by now entered the public domain.14,16,17 Given the proprietary nature of many discovery projects, one can essentially never expect to obtain a complete picture of compounds available in a given therapeutic or target area such as the kinome. However, the kinase field has matured to the point that it currently represents one of the major sources of publicly available discovery data. As a contribution to the themed kinase issue of this journal,8 we have determined the current compound coverage of the human kinome and characterized available inhibitors and their activity profiles. As a data source, ChEMBL16 was utilized that currently represents the major public repository of compound data from medicinal chemistry sources and includes activity data originally provided in other public databases.17

Table 1. Kinase Inhibitorsa ChEMBL (release 18) number of

total

Ki

IC50

inhibitors kinases interactions BM scaffolds CSKs

18951 266 27063 7698 3616

1760 94 2654 761 442

17775 264 25043 7343 3503

a

For ChEMBL, release 18, the number of kinase inhibitors, the number of kinases these inhibitors were active against, and the total number of kinase−inhibitor interactions are reported. In addition, the number of Bemis−Murcko (BM) scaffolds and cyclic skeletons (CSKs) from all kinase inhibitors are given. Furthermore, corresponding statistics are provided for Ki and IC50 value-based kinase inhibitor subsets.

kinase inhibitors represented 7698 unique BM scaffolds and 3616 corresponding CSKs (a given CSK covers topologically equivalent scaffolds). On average, each BM scaffold represented only ∼2.5 kinase inhibitors and each CSK ∼5.2 inhibitors. Hence, there was overall unexpectedly high scaffold diversity among kinase inhibitors. We then distinguished the kinase inhibitors on the basis of different activity measurements. The composition of the Ki and IC50 value-based subsets is also reported in Table 1. There were 10+ times more inhibitors available in the IC50 than the Ki subset. Inhibitors with equilibrium constants were active against 94 kinases, whereas the IC50 subset covered nearly all kinases for which inhibitors were available (i.e., 264 of a total of 266). IC50 measurements covered ∼92% of all kinase−inhibitor interactions (i.e., 25043 of 27063). Hence, the IC50 subset contained the most kinase activity data available in the public domain. Figure 1 reports the distribution of inhibitors over different kinases. The average number of inhibitors per kinase target was ∼102 for the global kinase data set, ∼95 for the IC50 subset, and ∼28 for the Ki subset. However, for ∼49% (Ki) and ∼30% (IC50) of the kinase targets, only one to five inhibitors were available. Thus, for a significant number of kinases, inhibitors are currently rare.

■

COMPOUND ACTIVITY DATA From the latest version of ChEMBL (release 18), compounds with direct interactions (i.e., target relationship type “D”) with human kinase targets at the highest confidence level (i.e., target confidence score 9) were extracted. Two different types of potency measurements were considered including (assayindependent) equilibrium constants (Ki) and (assay-dependent) IC50 values. To ensure high data confidence, approximate measurements such as “>”, “10 sum

1198 366 106 66 17 1 2 1 1 2 0 1760

68.07 20.80 6.02 3.75 0.97 0.06 0.11 0.06 0.06 0.11 0 100

13,694 2719 753 280 121 68 45 27 14 7 47 17775

77.04 15.30 4.24 1.58 0.68 0.38 0.25 0.15 0.08 0.04 0.26 100

a

For the Ki- and IC50-based subsets, the number and proportion of kinase inhibitors with activity against increasing numbers of kinases are reported.

■

POTENCY DISTRIBUTION Figure 4 monitors the potency distribution of kinase inhibitors. The majority of kinase inhibitors fell into the logarithmic

Figure 5. Potency distribution of promiscuous inhibitors. Potency value distributions of all inhibitors and promiscuous inhibitors active against more than three kinases are reported in box plots for the Ki (blue, all inhibitors; light blue, promiscuous inhibitors) and IC50 (orange, all inhibitors; light orange, promiscuous inhibitors) subsets, respectively. Each box plot provides the lowest potency value within the 1.5 interquartile range of the lower quartile (bottom line), lower quartile (lower boundary of the box), median value (thick line), upper quartile (upper boundary of the box), and the highest value within the 1.5 interquartile range of the upper quartile (top line). Potency values falling outside these ranges are indicated by circles.

further refined set of seven hydrogen bonding fragments and a set of 10 hydrophobic tail fragments were proposed to characterize type II inhibitors.9 In order to screen the kinase inhibitor universe for likely type II inhibitors, we generated all 70 possible combinations of these hydrogen bonding and hydrophobic tail fragments. To ensure reproducibility, the substructure search was exclusively based on the original library of 70 type II kinase inhibitor signature fragments, without further structural modifications. The complete set of these combined signature fragments is provided in Figure S1 of the Supporting Information. Database searches for type II-like inhibitors were limited to these intuitive substructure matches, despite their approximate nature. For a thorough characterization of type I and type II inhibitors, structural data need to be considered on a case-by-case basis, different from our statistical analysis. All signature fragments were mapped to inhibitors from the Ki and IC50 subsets, and inhibitors were identified that contained a terminal signature fragment. When a signature fragment occurred as a central substructure in an inhibitor, the match was not further considered. Inhibitors that contained one of these fragments as a terminal substructure were designated as “type II-like inhibitors”. The remaining inhibitors that were not matched were considered as “type I-like inhibitors”. In addition to exact fragment matches, we also accepted approximate matches where one to maximally three additional small substituents such as halogen atoms, methyl, or carbonyl groups were permitted in the tail fragment. Inhibitors yielding such approximate matches were also considered type II-like. Examples of inhibitors with exact and approximate matches

Figure 4. Potency distribution. Reported are the compound potency distributions within the Ki (blue) and IC50 value-based (orange) kinase inhibitor subsets.

potency range from 6 to 9. Although the small Ki subset contained a larger proportion of inhibitors in the low nanomolar range and the large IC50 subset a larger proportion of inhibitors in the high micromolar range, the potency distributions were overall similar. Furthermore, the potency distribution of kinase inhibitors was also comparable to the distribution of all bioactive compounds.22 In addition, Figure 5 compares the potency distribution of promiscuous inhibitors with the global distribution. As can be seen, the potency distribution of promiscuous inhibitors closely resembled the global distribution for both the Ki or IC50 subsets. For this comparison, promiscuous inhibitors with activity against more than three kinases were considered.

■

MAPPING OF TYPE II INHIBITOR SIGNATURE FRAGMENTS On the basis of extensive structural studies of type I and type II kinase inhibitor binding modes, many type II inhibitors were found to contain a “type I head” fragment and share structural features that distinguished them from type I inhibitors.10 These features included a hydrogen bond donor−acceptor pair (e.g., urea or amide group) and a hydrophobic tail moiety (binding into the DFG pocket below the α-C-helix).10 Recently, a E



Perspective

Figure 6. Compounds containing type II inhibitor signature fragments. Mapping of two representative type II inhibitor signature/tail fragments is shown including (a) fragment 38 and (b) fragment 52. For each fragment, the total number of inhibitors with exact and approximate matches is reported. For example, “4|6” means that there were four and six inhibitors with exact and approximate matches to fragment 38, respectively. For each fragment, two exemplary compounds with an exact (top) or approximate match (bottom) are shown. Tail fragments are highlighted in red. For compounds with an approximate match, the modification of the tail fragment is highlighted in green.

However, it should be noted that the anilinoquinazoline substructure belonged to the “type I head” fragment and was not among the type II inhibitor signature fragments.

are shown in Figure 6. Figure 6b shows an inhibitor containing an anilinoquinazoline substructure, which is also a common structural feature of several type I inhibitor such as lapatinib. F



Perspective

Accordingly, lapatinib did not produce a match for type II fragments. In addition, the well-known type II inhibitor imatinib was not matched, although it contained an amide group and a hydrophobic tail moiety, similar to but distinct from fragment 34 in Figure S1. Therefore, matching of imatinib would have required the definition of a new fragment, in addition to the originally proposed set. Similarly, BIRB-796, another type II inhibitor, was not matched because the topology of the tail fragment differed from original fragments. For consistency, only the 70 original fragments were used for substructure searching. In addition, the fragments were only mapped to inhibitors in the Ki or IC50 subsets of ChEMBL with high-confidence activity data. Care must be taken to evaluate search results in the absence of data confidence criteria. For example, when fragment 29 in Figure S1 was searched in PubChem, a total of 266 hits were obtained, irrespective of the position of fragment 29 in these compounds. Of these 266 hits, only three compounds were designated as active in one or more assays, and all of these three compounds were present in ChEMBL. However, if the compound selection criteria described above were applied, no qualifying compound was detected. Similar observations were made for many fragments. Surprisingly, no type II-like inhibitors were detected in the Ki subset. In the IC50 subset, 65 and 231 inhibitors yielded exact and approximate matches of signature fragments, respectively. Hence, a small set of only 231 type II-like inhibitors was detected among 17775 kinase inhibitors in the IC50 subset. Table 5 reports the target distribution of type I- and type II-like

Figure 7. Potency distribution of type I-like and type II-like inhibitors. Distributions of potency values of type I-like (orange) and type II-like (light orange) inhibitors are reported in box plots for the IC50 subset. Box plot representations are according to Figure 5

Table 6. MMP-Cliff Statistics for Kinase Inhibitorsa no. MMPs no. MMP-cliffs % MMP-cliffs no. inhibitors forming MMP-cliffs % inhibitors forming MMP-cliffs no. kinases with MMP-cliffs

Table 5. Type I-like and Type II-like Inhibitorsa no. inhibitors no. kinases

type I-like

type II-like

total

1 2 3 4 5 6 7 8 9 10 >10 sum

13503 2685 751 279 120 68 45 26 14 7 46 17544

191 34 2 1 1 0 0 1 0 0 1 231

13694 2719 753 280 121 68 45 27 14 7 47 17775

Ki

IC50

11989 614 5.12 376 21.36 31

108591 4482 4.13 3062 17.23 106

a

For the Ki- and IC50-based subsets, the number of MMPs, the number and proportion of MMP-cliffs, the number and proportion of kinase inhibitors forming MMP-cliffs, and the number of kinase data sets containing MMP-cliffs are reported.

Table 7. Single-Target and Multitarget MMP-Cliffsa no. MMP-cliffs

a

For the IC50 kinase subset, the numbers of type I-like and type II-like inhibitors with activity against increasing numbers of kinases are reported.

degree (number of kinases)

Ki

IC50

1 2 3 4 5 6

589 22 3 0 0 0

4008 338 78 24 10 24

a

The number of MMP-cliffs with activity against different numbers of kinases is reported. For example, a degree of 2 means that two inhibitors forming an MMP meet the potency difference threshold for two kinases.

inhibitors that had exact or approximate matches to signature fragments. Of the 231 type II-like inhibitors, 191 were reported to be active against a single kinase, 34 against two kinases, and only three against five or more kinases. Interestingly, type IIlike inhibitors generally had high potency, with a median pIC50 value close to 8, as reported in Figure 7.

transformation size restricted matched molecular pair (MMP)24,25 and a potency difference of at least 2 orders of magnitude were required, respectively, corresponding to the definition of (structurally conservative) MMP-cliffs.25 An MMP is defined as a pair of compounds that are only distinguished by a structural change at a single site.24 Table 6 reports the MMP and MMP-cliff statistics for the Ki and IC50 subsets. In the IC50 and Ki subset, 4482 and 614 MMP-cliffs were detected, respectively. These cliffs included 4−5% of all transformation size restricted MMPs and were

■

ACTIVITY CLIFFS As an indicator of SAR information content, we also determined all activity cliffs23 formed by kinase inhibitors, i.e., pairs of structurally similar/analogous compounds having a large difference in potency. As the similarity and potency difference criteria for activity cliffs, the formation of a G



Perspective

Figure 8. Chirality cliffs. (a) Shown are three chirality cliffs formed by inhibitors from the Ki subset. Four stereoisomers were active against protein kinase C (PKC) theta. For each inhibitor, the potency (pKi) value is given. Highly and weakly potent cliff partners are shown on a green and red background, respectively. (b) Shown are two chirality cliffs for inhibitors from the IC50 subset that had significant differences in potency against two and three kinases, respectively (i.e., these inhibitors formed multitarget chirality cliffs). For each inhibitor, potency (pIC50) values for different targets are provided.

formed by on average every fourth to fifth kinase inhibitor. These numbers were very close to averages calculated for currently available compound activity classes.23 In the IC50

subset, only 36 of all 4482 MMP-cliffs were formed by type IIlike inhibitors. Furthermore, in the Ki and IC50 subsets, MMPcliffs were formed by inhibitors active against 31 and 106 H



Perspective

Figure 9. MMP-cliffs. (a) Shown is a MMP-cliff identified for three subtypes of phosphatidylinositol 3-kinase (PI3K) of the Ki subset. For each inhibitor, corresponding pKi values are provided. Structural modifications are highlighted in red. Highly and weakly potent cliff partners are shown on a green and red background, respectively. (b) Shown is a MMP-cliff formed by inhibitors active against tyrosine−protein kinase JAK1, JAK3, and TYK2 of the IC50 subset. For each inhibitor, corresponding pIC50 values are provided.

small molecule inhibitors. There were a few heavily populated kinases (including popular oncology targets), while for many others, only limited numbers of inhibitors were available. Furthermore, our analysis of activity profiles revealed that kinase inhibitor promiscuity was lower than often assumed. Only ∼1% of kinase inhibitors were found to be active against five or more targets. Taken together, these findings suggest that there should be ample opportunities for kinase drug discovery going forward. In addition, results of signature fragment mapping indicated that type II kinase inhibitors are currently still rare (in addition to very few allosteric inhibitors that have thus far been characterized). Therefore, investigating compounds with alternative mechanism of action should continue to be a focal point of kinase discovery, not necessarily to reduce compound promiscuity but to generate highly efficacious inhibitors. Last but not least, ∼5000 well-defined activity cliffs formed by kinase inhibitors that were identified for more than 100 kinases (with >90% of these being target-specific cliffs) should provide a wealth of SAR information for further advancing ATP site directed inhibitors of these targets.

different kinases, respectively. Hence, cliffs were detected for ∼40% of all kinases for which inhibitors were available. This was in part due to low compound numbers available for a significant number of kinases, in addition to scaffold diversity. Table 7 reports the distribution of single-target and multitarget MMP-cliffs. Consistent with the low promiscuity rates discussed above, the vast majority of MMP-cliffs represented single-target cliffs. In the IC50 subset, only 474 of 4482 MMPcliffs were multitarget cliffs (∼10.6%), with 338 of these being dual-target cliffs. Figure 8 shows exemplary activity cliffs formed by kinase inhibitors that were only distinguished by the configuration at one or more chiral centers, so-called chirality cliffs,26 including single-target cliffs from the Ki (Figure 8a) and dual- and triple-target cliffs from the IC50 subset (Figure 8b). Furthermore, Figure 9 shows two triple-target MMP-cliffs with small structural modifications, including the change of a heteroatom in the core structure (Figure 9a) and the replacement of a substituent (Figure 9b).

■

CONCLUDING REMARKS Herein we have reported a detailed survey of currently available protein kinase inhibitors, their target distribution, and activity profiles. This data-driven analysis has been concisely presented to serve as a reference for the kinase field. Nearly 19000 kinase inhibitors were found to be associated with high-confidence activity data, hence providing a rich source of information. However, analysis of the target distribution revealed that only about half of the human kinome is currently populated with

■

ASSOCIATED CONTENT

S Supporting Information *

Figure S1 containing the complete list of type II kinase inhibitor signature fragments. This material is available free of charge via the Internet at http://pubs.acs.org. I



■

Perspective

Wodicka, L. M.; Zarrinkar, P. P. A Quantitative Analysis of Kinase Inhibitor Selectivity. Nat. Biotechnol. 2008, 26, 127−132. (5) Cheng, A. C.; John Eksterowicz, J.; Geuns-Meyer, S.; Sun, Y. Analysis of Kinase Inhibitor Selectivity Using a ThermodynamicsBased Partition Index. J. Med. Chem. 2010, 53, 4502−4510. (6) Metz, J. T.; Johnson, E. F.; Soni, N. B.; Merta, P. J.; Kifle, L.; Hajduk, P. J. Navigating the Kinome. Nat. Chem. Biol. 2011, 7, 200− 202. (7) Gavrin, L. K.; Saiah, E. Approaches To Discover Non-ATP Site Kinase Inhibitors. Med. Chem. Commun. 2013, 4, 41−51. (8) Laufer, S.; Bajorath, J. New Frontiers in Kinases: Second Generation Inhibitors. J. Med. Chem. 2014, 57, 2167−2168. (9) Zhao, Z.; Wu, H.; Wang, L.; Liu, Y.; Knapp, S.; Liu, Q.; Gray, N. S. Exploration of Type II Binding Mode: A Privileged Approach for Kinase Inhibitor Focused Drug Discovery? ACS Chem. Biol. 2014, 9, 1230−1241. (10) Liu, Y.; Gray, N. S. Rational Design of Inhibitors That Bind to Inactive Kinase Conformations. Nat. Chem. Biol. 2006, 2, 358−364. (11) Ohren, J. F.; Chen, H.; Pavlovsky, A.; Whitehead, C.; Zhang, E.; Kuffa, P.; Yan, C.; McConnell, P.; Spessard, C.; Banotai, C.; Mueller, W. T.; Delaney, A.; Omer, C.; Sebolt-Leopold, J.; Dudley, D. T.; Leung, I. K.; Flamme, C.; Warmus, J.; Kaufman, M.; Barrett, S.; Tecle, H.; Hasemann, C. A. Structures of Human MAP Kinase Kinase 1 (MEK1) and MEK2 Describe Novel Noncompetitive Kinase Inhibition. Nat. Struct. Mol. Biol. 2004, 11, 1192−1197. (12) Adrián, F. J.; Ding, Q.; Sim, T.; Velentza, A.; Sloan, C.; Liu, Y.; Zhang, G.; Hur, W.; Ding, S.; Manley, P.; Mestan, J.; Fabbro, D.; Gray, N. S. Allosteric Inhibitors of Bcr-Abl-Dependent Cell Proliferation. Nat. Chem. Biol. 2006, 2, 95−102. (13) Ashwell, M. A.; Lapierre, J. M.; Brassard, C.; Bresciano, K.; Bull, C.; Cornell-Kennon, S.; Eathiraj, S.; France, D. S.; Hall, T.; Hill, J.; Kelleher, E.; Khanapurkar, S.; Kizer, D.; Koerner, S.; Link, J.; Liu, Y.; Makhija, S.; Moussa, M.; Namdev, N.; Nguyen, K.; Nicewonger, R.; Palma, R.; Szwaya, J.; Tandon, M.; Uppalapati, U.; Vensel, D.; Volak, L. P.; Volckova, E.; Westlund, N.; Wu, H.; Yang, R. Y.; Chan, T. C. Discovery and Optimization of a Series of 3-(3-Phenyl-3H-imidazo[4,5-b]pyridin-2-yl)pyridin-2-amines: Orally Bioavailable, Selective, and Potent ATP-Independent Akt Inhibitors. J. Med. Chem. 2012, 55, 5291−5310. (14) van Linden, O. P.; Kooistra, A. J.; Leurs, R.; de Esch, I. J.; de Graaf, C. KLIFS: A Knowledge-Based Structural Database to Navigate Kinase−Ligand Interaction Space. J. Med. Chem. 2014, 57, 249−277. (15) Manning, G.; Whyte, D.; Martinez, R.; Hunter, T.; Sudarsanam, S. The Protein Kinase Complement of the Human Genome. Science 2002, 298, 1912−1934. (16) Gaulton, A.; Bellis, L. J.; Bento, A. P.; Chambers, J.; Davies, M.; Hersey, A.; Light, Y.; McGlinchey, S.; Michalovich, D.; Al-Lazikani, B.; Overington, J. P. ChEMBL: A Large-Scale Bioactivity Database for Drug Discovery. Nucleic Acids Res. 2011, 40, D1100−D1107. (17) Bento, A. P.; Gaulton, A.; Hersey, A.; Bellis, L. J.; Chambers, J.; Davies, M.; Krüger, F. A.; Light, Y.; Mak, L.; McGlinchey, S.; Nowotka, M.; Papadatos, G.; Santos, R.; Overington, J. P. The ChEMBL Bioactivity Database: An Update. Nucleic Acids Res. 2014, 42, D1083−D1090. (18) UniProt Consortium.. Reorganizing the Protein Space at the Universal Protein Resource (UniProt). Nucleic Acids Res. 2012, 40, D142−D148. (19) Bemis, G. W.; Murcko, M. A. The Properties of Known Drugs. 1. Molecular Frameworks. J. Med. Chem. 1996, 39, 2887−2893. (20) Xu, Y. J.; Johnson, M. Using Molecular Equivalence Numbers To Visually Explore Structural Features That Distinguish Chemical Libraries. J. Chem. Inf. Comput. Sci. 2002, 42, 912−926. (21) Hu, Y.; Bajorath, J. High-Resolution View of Compound Promiscuity. F1000Research 2013, 2, 144 (DOI: 10.12688/f1000research.2-144.v2). (22) Hu, Y.; Bajorath, J. Growth of Ligand−Target Interaction Data in ChEMBL Is Associated with Increasing and Activity MeasurementDependent Compound Promiscuity. J. Chem. Inf. Model. 2012, 52, 2550−2558.

AUTHOR INFORMATION

Corresponding Author

*Phone: 49-228-2699-306. E-mail: [email protected]. Notes

The authors declare no competing financial interest. Biographies Ye Hu studied clinical medicine at the Southeast University, China, from 1999 to 2004. In 2006, she joined the Life Science Informatics Master program at the University of Bonn, Germany, and obtained her Master’s degree in 2008. In October 2008, she began her Ph.D. studies in the group of Prof. Jürgen Bajorath, focusing on systematic computational analysis of molecular scaffolds of bioactive compounds and associated characteristics. Since July 2011, she has been working as a Postdoctoral Fellow in the Department. Her current research interests include large-scale mining of ligand−target interaction data and structure−activity relationship analysis. Norbert Furtmann studied pharmacy at the University of Bonn, Germany, from 2006 to 2011. In January 2012, he started his Ph.D. studies in the groups of Prof. Jürgen Bajorath and Prof. Michael Gütschow, focusing on computer-aided drug design as well as synthesis and biological evaluation of enzyme inhibitors. Jürgen Bajorath studied biochemistry at the Free University, Berlin, Germany. He currently is Professor and Chair of Life Science Informatics at the University of Bonn, Germany, and also an Affiliate Professor in the Department of Biological Structure at the University of Washington, Seattle. His research interests include computer-aided medicinal chemistry, chemical biology, and drug discovery. For further details, see http://www.lifescienceinformatics.uni-bonn.de.

■ ■

ACKNOWLEDGMENTS N.F. is supported by a fellowship from the Jürgen Manchot Foundation, Düsseldorf, Germany. ABBREVIATIONS USED AGC, protein kinase A, protein kinase G, and protein kinase C containing families; BM, Bemis−Murcko; CAMK, calcium/ calmodulin-dependent protein kinase; CK1, casein kinase 1; CMGC, cyclin-dependent kinase, mitogen-activated protein kinase, glycogen synthase kinase 3, and dual specificity protein kinase containing families; CSK, cyclic skeleton; MMP, matched molecular pair; PI3/PI4, phosphatidylinositol 3/4kinase; SAR, structure−activity relationship; STE, homologues of yeast sterile 7, sterile 11, sterile 20 kinases; TK, tyrosine kinase; TKL, tyrosine kinase-like

■

REFERENCES

(1) Ward, R. A., Goldberg, F. W., Eds. Kinase Drug Discovery; RSC: Cambridge, U.K., 2011. (2) Levitzki, A. Tyrosine Kinase Inhibitors: Views of Selectivity, Sensitivity, and Clinical Performance. Annu. Rev. Pharmacol. Toxicol. 2013, 53, 161−185. (3) Fabian, M. A.; Biggs, W. H., 3rd; Treiber, D. K.; Atteridge, C. E.; Azimioara, M. D.; Benedetti, M. G.; Carter, T. A.; Ciceri, P.; Edeen, P. T.; Floyd, M.; Ford, J. M.; Galvin, M.; Gerlach, J. L.; Grotzfeld, R. M.; Herrgard, S.; Insko, D. E.; Insko, M. A.; Lai, A. G.; Lélias, J. M.; Mehta, S. A.; Milanov, Z. V.; Velasco, A. M.; Wodicka, L. M.; Patel, H. K.; Zarrinkar, P. P.; Lockhart, D. J. A Small Molecule-Kinase Interaction Map for Clinical Kinase Inhibitors. Nat. Biotechnol. 2005, 23, 329−336. (4) Karaman, M. W.; Herrgard, S.; Treiber, D. K.; Gallant, P.; Atteridge, C. E.; Campbell, B. T.; Chan, K. W.; Ciceri, P.; Davis, M. I.; Edeen, P. T.; Faraoni, R.; Floyd, M.; Hunt, J. P.; Lockhart, D. J.; Milanov, Z. V.; Morrison, M. J.; Pallares, G.; Patel, H. K.; Pritchard, S.; J



Perspective

(23) Stumpfe, D.; Hu, Y.; Dimova, D.; Bajorath, J. Recent Progress in Understanding Activity Cliffs and Their Utility in Medicinal Chemistry. J. Med. Chem. 2014, 57, 18−28. (24) Kenny, P. W.; Sadowski, J. Structure Modification in Chemical Databases. In Chemoinformatics in Drug Discovery; Oprea, T. I., Ed.; Wiley-VCH: Weinheim, Germany, 2004; pp 271−285. (25) Hu, X.; Hu, Y.; Vogt, M.; Stumpfe, D.; Bajorath, J. MMP-Cliffs: Systematic Identification of Activity Cliffs on the Basis of Matched Molecular Pairs. J. Chem. Inf. Model. 2012, 52, 1138−1145. (26) Hu, Y.; Bajorath, J. Extending the Activity Cliff Concept: Structural Categorization of Activity Cliffs and Systematic Identification of Different Types of Cliffs in the ChEMBL Database. J. Chem. Inf. Model. 2012, 52, 1806−1811.

K


Current Compound Coverage of the Kinome - Journal of Medicinal

Recommend Documents