D824–D831 Nucleic Acids Research, 2009, Vol. 37, Database issue Published online 5 November 2008 doi:10.1093/nar/gkn832 MoKCa database—mutations of kinases in cancer Christopher J. Richardson1, Qiong Gao2, Costas Mitsopoulous2, Marketa Zvelebil2, Laurence H. Pearl1 and Frances M. G. Pearl1,* 1Section of Structural Biology and 2The Breakthrough Breast Cancer Research Centre, Institute of Cancer Research, Chester Beatty Laboratories, 237 Fulham Road, London SW3 6JB, UK Received August 14, 2008; Revised October 3, 2008; Accepted October 14, 2008 ABSTRACT the cell (and its progeny) that contain them. Knowledge of these mutations is key to understanding the biology of Members of the protein kinase family are amongst cancer initiation and progression, as well as to the devel- the most commonly mutated genes in human opment of more targeted therapeutic strategies. While cancer, and both mutated and activated protein there are rare examples of cancers driven by a single kinases have proved to be tractable targets for genetic alteration (1), in most solid tumours, tumourigen- the development of new anticancer therapies esis is a multistep process (2,3), reflecting the genetic The MoKCa database (Mutations of Kinases in alterations necessary to transform normal cells into malig- Cancer, http://strubiol.icr.ac.uk/extra/mokca) has nant derivatives. The crucial events include acquisition been developed to structurally and functionally of genomic instability, cell cycle deregulation, evasion of annotate, and where possible predict, the pheno- apoptosis, limitless replicative potential, angiogenesis and typic consequences of mutations in protein kinases metastasis (4). Regardless of the multiple mutations ulti- implicated in cancer. Somatic mutation data from mately required, a single mutation can initiate the process. tumours and tumour cell lines have been mapped Members of the protein kinase family are amongst the onto the crystal structures of the affected protein most commonly mutated genes in human cancer, and both domains. Positions of the mutated amino-acids are mutated and activated protein kinases have proved to be tractable targets for the development of new anticancer highlighted on a sequence-based domain picto- therapies (5). There are 518 documented mammalian gram, as well as a 3D-image of the protein structure, protein kinases (6) encoded in the human genome, which and in a molecular graphics package, integrated for together represent the largest family of human enzymes, interactive viewing. The data associated with each collectively termed the kinome. They play indispens- mutation is presented in the Web interface, along able roles in numerous cellular, metabolic and signalling with expert annotation of the detailed molecular pathways, in all cell types. Around 40% of the kinome functional implications of the mutation. Proteins have multiple splice variants and 10% of the total encode are linked to functional annotation resources and catalytically deficient enzymes that have been termed are annotated with structural and functional features pseudokinases. such as domains and phosphorylation sites. MoKCa Although there are several kinase classification schemes aims to provide assessments available from multiple (e.g. 6,7), the KinBase resource (http://www.kinase.com/ sources and algorithms for each potential cancer- kinbase) (6) reflects the currently accepted classification of associated mutation, and present these together eukaryotic protein kinases where the kinases are broadly in a consistent and coherent fashion to facilitate split into two groups: conventional protein kinases (ePKs) and atypical protein kinases (aPKs). The ePKs are the authoritative annotation by cancer biologists and largest group, and are subclassified into eight families structural biologists, directly involved in the genera- using the sequence similarity of the kinase domain, the tion and analysis of new mutational data. presence of accessory domains, and consideration of their modes of regulation. The aPKs are a smaller set of protein kinases that do not share clear sequence similar- INTRODUCTION ity with ePKs, but have been shown experimentally to Cancers arise due to the accumulation of mutations in have protein kinase activity. As the entries in KinBase critical target genes that confer a selective advantage on are filtered by stringent criteria, including verification by *To whom correspondence should be addressed. Tel: +44 (0)20 7153 5443; Fax: +44 (0)20 7153 5457; Email: [email protected] ß 2008 The Author(s) This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/ by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Nucleic Acids Research, 2009, Vol. 37, Database issue D825 cDNA cloning, the KinBase classification scheme is the repetitious sequencing in affected families of particular one favoured by experimentalists working on kinases kinases mutated in specific diseases, or from harvesting and signal transduction pathways. literature identifications of mutations observed in individ- Protein kinases are frequently found to be mis-regulated ual studies. Furthermore, the COSMIC database (17), is in human cancer, and the Cancer Genome Project and sim- undertaking to document all somatic cancer mutations ilar initiatives, have undertaken systematic re-sequencing reported in the literature. screens of all annotated protein kinases in the human By bringing together automatic assessments available genome, to attempt to identify commonly occurring muta- from multiple sources and algorithms for each potential tions that may play significant roles in a range of differ- cancer-associated mutation, and presenting these in a con- ent cancers (8–10). In all cases the key to understanding venient and coherent fashion, MoKCa aims to facilitate the contribution of a particular disease-associated kinase authoritative annotation by cancer biologists and struc- mutation to development and progression of cancer, comes tural biologists, directly involved in the generation and from an appreciation of the consequences of that muta- analysis of new mutational data. These ‘experts’ are then tion on the function of the affected protein, and the able to bring detailed insights into the biochemistry and impact on the pathways in which that protein is involved. biology of individual proteins and systems, that are vir- It is this that the MoKCa database described here, aims tually impossible to encapsulate in an algorithm, but are to facilitate. key to determining if and how a particular mutation will Changes in the nucleotide sequence of a gene can have a alter the biological function of a protein. Thus the MoKCa variety of consequences for the encoded protein, including database combines automated and ‘expert’ annotation of truncations and frameshifts that disrupt the protein struc- individual mutations, and is firmly directed towards the ture and/or reduce transcript levels via nonsense-mediated specific needs of the cancer research community. RNA decay. The most common mutational event is a single base change, leading to a missense mutation of a single amino acid. Cancer-associated somatic mutations BUILDING THE MoKCa DATABASE (CASMs) or missense variants are commonly identified Mutation data in somatic tumour DNA, but only a fraction directly contribute to oncogenesis. Distinguishing those that con- The original mutational data was provided by the Sanger tribute to cancer from those that do not is a difficult pro- Cancer Genome Project (CGP) Team, and comprises blem, potentially requiring detailed and protracted the mutations found in the large-scale re-sequencing of functional analysis. However the ability to make this the kinase complement from 210 human cancers. This determination, both rapidly and inexpensively, will be included samples from breast, lung, colorectal, gastric, essential to realization of ‘personalized therapy’ targeted testis, ovarian, renal, melanoma, glioma and acute lym- to the individual tumour. phoblastic leukaemia (ALL) cancers (9). Of the 210 Several approaches have been taken to predict which tumours studied, 169 were primary tumours, 2 were genes contain mutations that contribute to mutagenesis early cultures and 39 immortal cell lines. (i.e. drivers genes) rather than those genes that contain One-thousand and seven somatic mutations are docu- mutations that arise by chance but have no bearing on mented in the coding exons and splice junctions in the the disease (passenger mutations). Statistical models com- kinases of 137 of the tumours studied. Nine-hundred and paring the observed ratio of non-synonymous:synon- twenty one were single base substitutions, 78 were small ymous compared with that expected by chance, have insertions or deletions and 8 were complex changes. Of the been used to identify and estimate the number of cancer single base substitutions, 620 were missense changes, 54 drivers within a total set of identified genetic variation caused nonsense changes and 28 were observed at highly (11). These methods are excellent for predicting which conserved positions of splice junctions. There were also genes contain drivers, but do not identify the driver muta- 219 silent (synonymous mutations). Approximately one- tion alone. Consequently algorithms have been developed third of these mutations had previously been reported in that attempt to assess