CORRESPONDENCE

Cyclin-dependent kinases: a family portrait

Marcos Malumbres1,10, Edward Harlow2, Tim Hunt3, Tony Hunter4, Jill M. Lahti5, Gerard Manning6, David O. Morgan7, Li-Huei Tsai8 and Debra J. Wolgemuth9

To the Editor, tion with some cyclin-like regulatory subunit. (CrkRS) and Ched (Cdc2L5), were recently Cyclin-dependent kinases (CDKs) are protein Known family members were then renamed renamed CDK12 and CDK13, as they were kinases involved in critical cellular processes, CDK1–6, and further members (CDK7– reported to interact with cyclin L1 and cyclin such as cell cycle or transcription, whose activ- CDK10) were cloned and characterized in the L2 (refs 3, 4). ity requires association with specific cyclin sub- following years1. CDK11 has been used to refer Additional members of the family, such units. Based on sequence similarity, the human to the protein encoded by three different human as PCTAIRE and PFTAIRE proteins, CCRK genome contains 21 genes encoding CDKs and loci (CDC2L1, CDC2L2 and CDC2L6) in differ- and CDC2L6, although being very similar to five additional genes encoding a more distant ent publications. CDC2L1 and CDC2L2 are two other CDKs in their primary sequence (Fig. 1) group of proteins known as CDK-like (CDKL) highly similar human genes, which originated and their predicted cyclin-binding domains kinases. The current nomenclature for CDK by duplication in human chromosome 1, and (Supplementary Information, Fig. S1), still proteins includes 11 classical CDKs (CDK1–11), they encode almost identical protein kinases. maintain their original names (Supplementary two newly proposed family members (CDK12 Each of these loci encodes at least two major Information, Table S1). Ironically, some of these and 13) and additional proteins whose names peptides: a protein kinase of relative molecu- proteins have not received CDK names because are based on the presence of a cyclin-binding lar mass 110,000 (Mr 110K), and a smaller 58K their association with cyclins has not been element (PFTAIRE and PCTAIRE proteins) isoform that is expressed following alternative demonstrated, and yet they are known by their or simply based on a sequence relationship initiation of translation. To make the situation predicted cyclin-association element sequence with the original CDKs, such as CDC2-like more complex, some publications, including the (PFTAIRE or PCTAIRE; Supplementary kinases (CDC2L) or cell cycle-related kinases original description of the human kinome2, refer Information, Fig. S1). Although the strict ‘cyclin- (CCRK)1. Here we propose the use of ‘CDK’ to CDK11 as the protein encoded by CDC2L6 dependent’ rule was useful during the gradual for all CDK family members (CDK1–20) on (Supplementary Information, Table S1). The characterization of the CDK family, we feel the basis of similarities in sequence and func- single Cdc2l1 gene in the mouse genome that there is now enough evidence to group all tion as described below. This nomenclature will encodes a protein more similar to human these CDK-related proteins into a single protein facilitate the rational and comparative study of CDC2L2 (91% identity) than CDC2L1 (87%). family with sufficient sequence and functional the poorly understood family members and the We therefore propose the use of Cdk11 for the similarities to justify the use of the ‘CDK’ term analysis of their relevance to human disease. mouse gene and CDK11A and CDK11B for the for all family members. PFTAIRE (PFTK) and During the Cold Spring Harbor Symposium human CDC2L2 and CDC2L1 loci, respectively PCTAIRE (PCTK) proteins are encoded by on the Cell Cycle in 1991, a group of interested (Fig. 1; Supplementary Information, Table S1). five genes that are more related to CDK1 than scientists proposed that members of this kinase The corresponding human proteins should be other classic family members such as CDK4 family should be called cyclin-dependent kinases referred to as CDK11Ap58 or CDK11Ap110 for the or CDK6 (Fig. 1). Recent studies indicate that (CDKs). The consensus was that no kinase CDC2L2-encoded proteins, and CDK11Bp58 or PFTK1 is activated by cyclin D3 and cyclin Y, should be called a ‘CDK’ until it was proven CDK11Bp110 for the CDC2L1-encoded proteins. and phosphorylates the retinoblastoma protein rigorously that its activity depended on associa- Two additional kinases, formerly known as Crk7 (pRb; Supplementary Information, Table S1)5,6 . PFTK1 should therefore be renamed CDK14. 1Cell Division and Cancer Group, Spanish National Cancer Research Center (CNIO), 28029 Madrid, Spain. 2Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA PFTK2 displays 60% identity with CDK14/ 02115, USA. 3Cancer Research UK, Clare Hall Laboratories, South Mimms, Herts EN6 3LD, UK. 4Molecular PFTK1, but no studies of this protein have been and Cell Biology Laboratory, The Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037, USA. 5Department of Genetics and Tumor Cell Biology, St Jude Children’s Research Hospital, Memphis, TN 38105, published since its identification in 2001. Based USA. 6Razavi Newman Center for Bioinformatics, Salk Institute, La Jolla, CA 92037, USA. 7Department of on conservation of the PFTAIRE motif and sur- Physiology, University of California, San Francisco, CA 94158, USA. 8Howard Hughes Medical Institute, rounding sequences, as well as overall similarity, Picower Institute for Learning and memory, Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Stanley Center for Psychiatric Research, Broad Institute of Harvard and MIT, PFTK2 seems very likely to have a cyclin part- Cambridge, MA 02119, USA. 9Departments of Genetics and Development and Obstetrics and Gynecology, ner and should therefore be renamed as CDK15. The Institute of Human Nutrition, Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, NY 10032, USA. The related PCTAIRE kinases (PCTK1–3) are 10Correspondence should be addressed to M.M. (e-mail: [email protected]). also poorly studied, but retain a cyclin-binding nature cell biology VOLUME 11 | NUMBER 11 | NOVEMBER 2009 1275 © 2009 Macmillan Publishers Limited. All rights reserved.

CORRESPONDENCE

CDK1 CDC2 CDK2 CDK2 therapeutic value of the lesser-known CDKs, CDK3 CDK3 and also a failure to examine these protein CDK5 CDK5 kinases for specificity when developing CDK CDK14 PFTK1 CDK15 PFTK2 inhibitors. Protein names like PFTAIRE‑1 do CDK16 PCTK1 not encourage the consideration of the thera- CDK17 PCTK2 peutic value of these proteins, which are usu- CDK18 PCTK3 CDK4 CDK4 ally not tested for therapeutic purposes. In CDK6 CDK6 summary, we believe that sequence and func- CDK7 CDK7 tional relationships suggest that all of the 21 CDK20 CCRK CDK11A CDC2L2 human CDKs should be given the ‘CDK’ pre- CDK11B CDC2L1 fix. This expanded CDK family nomenclature CDK10 CDK10 has also been agreed on by the HUGO Gene CDK12 CRK7 CDK13 CDC2L5 Nomenclature Committee (HGNC) and the CDK9 CDK9 Mouse Genomic Nomenclature Committee CDK family CDK8 CDK8 (MGNC), and will be listed in all the major CDK19 CDC2L6 CDKL1 CDKL1 databases for the human and mouse genes CDKL4 CDKL4 encoding the CDK family proteins. The anal- CDKL2 CDKL2 ysis of sequence and functional trends in the CDKL family CDKL3 CDKL3 CDKL5 CDKL5 CDK family may facilitate new research (for example, compensatory functions in loss- Figure 1 Human CDK and CDKL proteins. The proposed nomenclature for CDK (purple) proteins is of-function studies) directed toward their shown in bold, whereas current symbols are indicated in the right column (black). CDK-like proteins relevance in the cell cycle, transcription or (green) cluster together in a separate branch close to MAP kinases and GSK3 kinases2. The absence of evidence for cyclin association precludes us from considering these proteins as CDKs. The sequences of other cellular processes. In addition, this sim- the kinase domains were obtained from UniProt (http://www.uniprot.org/) and analysed using Clustal2 plified nomenclature may increase awareness (http://www.ebi.ac.uk/Tools/clustalw2/) by the neighbour-joining method. of their potential importance as therapeutic targets in human disease. Researchers and motif and have some evidence for cyclin interac- CDK2 (refs 9, 10). We therefore propose the use drug discoverers should keep the whole fam- tion. PCTK3 was found to interact with cyclin K of CDK20 for this protein. ily picture in mind. 7 in a large-scale interaction study . The two other Although a few of these kinases do not 1. Malumbres, M. & Barbacid, M. Mammalian cyclin- PCTK kinases also interact with p35, the major have an identified cyclin partner, we believe dependent kinases. Trends. Biochem. Sci. 30, 630–641 (2005). activator of CDK5 (Supplementary Information, that the CDK term should be applied to the 2. Manning, G., Whyte, D. B., Martinez, R., Hunter, T. & Table S1). Given their interaction with cyclin whole family. First, it is quite likely that all Sudarsanam, S. The protein kinase complement of the K and p35, and the high similarity between of these proteins are activated by cyclins or human genome. Science 298, 1912–1934 (2002). 3. Chen, H. H., Wang, Y. C. & Fann, M. J. Identification PCTAIRE proteins and CDK5, we suggest that cyclin-related proteins, many of which are and characterization of the CDK12/cyclin L1 complex these proteins should be renamed as CDK16, still poorly studied. Second, even if some involved in alternative splicing regulation. Mol. Cell. Biol. 26, 2736–2745 (2006). CDK17 and CDK18. The same rationale can be family members are not activated by cyclins, 4. Chen, H. H., Wong, Y. H., Geneviere, A. M. & Fann, applied to the remaining members of the family, this does not invalidate the use of the ‘CDK’ M. J. CDK13/CDC2L5 interacts with L-type cyclins and regulates alternative splicing. Biochem. Biophys. Res. CDC2L6 and CCRK. CDC2L6 is highly simi- root to denominate all family members. There Commun. 354, 735–740 (2007). lar to CDK8, contains an identical SMSACRE are multiple examples in gene nomenclature 5. Shu, F. et al. Functional characterization of human PFTK1 as a cyclin-dependent kinase. Proc. Natl Acad. motif and it is also a component of the Mediator where some family members do not share Sci. U SA 104, 9248–9253 (2007). complex8, suggesting not only sequence homol- the major features that originally defined 6. Stanyon, C. A. et al. A Drosophila protein-interaction map centered on cell-cycle regulators. Genome Biol. 5, ogy, but also functional similarities. CDC2L6 that family. For example, 45 human protein R96 (2004). forms complexes with cyclin C, although the kinases lack at least one of the conserved cata- 7. Rual, J. F. et al. Towards a proteome-scale map of the human protein-protein interaction network. Nature 2 interaction has not been described in detail lytic residues and may not have kinase activ- 437, 1173–1178 (2005). (Supplementary Information, Table S1). CDK19 ity. Kinase activity has not been detected for 8. Sato, S. et al. A set of consensus mammalian mediator subunits identified by. should therefore be used instead of CDC2L6. several CDK-related proteins, but they are still multidimensional protein identification technology. Finally, CCRK (also known as CAKp42) is known as PFTK2, PCTK2 or CCRK, where Mol. Cell 14, 685–691 (2004). 9. Liu, Y., Wu, C. & Galaktionov, K. p42, a novel cyclin- selectively similar to CDK7, the catalytic subunit ‘K’ stands for ‘kinase’. Finally, most ‘cyclins’ dependent kinase-activating kinase in mammalian of the CDK-activating kinase (CAK) complex do not cycle. cells. J. Biol. Chem. 279, 4507–4514 (2004). 10. Ng, S. S. et al. Cell cycle-related kinase: a novel candi- (Fig. 1), and it has also been suggested to func- The current nomenclature has probably date oncogene in human glioblastoma. J. Natl Cancer tion similarly by phosphorylating and activating contributed to the lack of consideration of the Inst. 99, 936–948 (2007).

1276 nature cell biology VOLUME 11 | NUMBER 11 | NOVEMBER 2009 © 2009 Macmillan Publishers Limited. All rights reserved.