© 2003 Nature Publishing Group http://www.nature.com/naturegenetics
NEWS AND VIEWS
Bared essentials of CDK2 and cyclin E James M Roberts & Charles J Sherr The cyclin E–CDK2 complex was thought to be an essential regulator of the mitotic cell cycle. Surprisingly, disabling the critical genes in the mouse has fewer effects than anticipated. Oscillations in the activities of cyclin-dependent kinases (CDKs) dictate orderly progression through the cell division cycle. In the simplest case, a progressive rise in the activity of a single cyclin–CDK complex can initiate DNA synthesis and then mitosis, and the subsequent fall in CDK activity resets the system for the next cell cycle1. In most organisms, however, the cell cycle machinery relies on multiple cyclin–CDKs, whose individual but coordinated activities are each thought to be responsible for just a subset of cell cycle events. For the mammalian somatic cell cycle, the current thinking is that cyclins E and A, in combination with CDK2, initiate chromosome and centrosome duplication, whereas subsequent activation of cyclin A– and cyclin B–driven CDK1 complexes governs later events that terminate with mitosis (Fig. 1). Activation of this intrinsic cell cycle program is linked to extracellular proliferative cues through D-type cyclins and their kinase partners CDK4 and CDK6. Their mitogen-dependent assembly and activation sequesters the CDK2 inhibitor, p27 (also called Kip1), and inactivates Rb-family proteins, thereby promoting synthesis of cyclins E and A and activa-
J.M.R. is in the Division of Basic Sciences, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., Room A3-023, Seattle, Washington 98104, USA, and is also affiliated with the Howard Hughes Medical Institute. C.J.S. is in the Department of Genetics & Tumor Cell Biology, St. Jude Children’s Research Hospital, 332 N. Lauderdale, Memphis, Tennessee 38105, USA, and is also affiliated with the Howard Hughes Medical Institute. e-mail:
[email protected] or
[email protected] Published online 24 August 2003; doi:10.1038/ng1234
tion of CDK2 later in G1 phase2. New work addressing the functions of CDK2 in Nature Genetics3 and of cyclin E in Cell4 and The EMBO Journal5 now call key features of this model into question. Together, these papers not only suggest that the basic cell cycle engine in mammalian cells need not be as elaborate as we had imagined, but they also present a surprising paradox whose resolution may uncover new mechanisms of cell cycle regulation. CDK2 surprises Things started to unravel earlier this year, when Tetsu and McCormick reported that CDK2 is not required for proliferation of certain tumor cell lines in vitro6. Cell cycle regulation in tumor cells is abnormal, with frequent overexpression of cell cycle activators and repression of inhibitors. Hence, the
fact that CDK2 could be discarded in tumor cells was unexpected but could perhaps be explained by the compensatory action of other misregulated cell cycle pathways. Barbacid and colleagues3 have now eliminated this ambiguity and carried the analysis to a new level by engineering a Cdk2 knockout mouse that is fully viable with no developmental or cell cycle abnormalities, except for unanticipated defects in meiosis. Cells isolated from Cdk2-null embryos proliferate normally in vitro, although they enter senescence early and are less likely to undergo spontaneous immortalization or oncogene-induced transformation. Moreover, acute deletion of Cdk2 in primary cultured cells by Cre-loxP-mediated recombination caused no overt problems in cell proliferation, eliminating developmental ‘plasticity’ as an explanation.
Inhibiting cell cycle inhibitors RB-family members p27 APC
Cyclin E
Activating cell cycle activators Replication origin licensing- MCM Origin firing? Centrosome duplication- NPM Histone gene transcription- NPAT Chromatin structure- p300/CBP; SWI/SNF
G1
S
G2
M
Figure 1 Cyclin E–CDK2: the way they were. The essential cell cycle functions that have been attributed to cyclin E–CDK2 and, in some cases, the particular protein substrates that are thought to mediate those functions are shown.
NATURE GENETICS VOLUME 35 | NUMBER 1 | SEPTEMBER 2003
9
© 2003 Nature Publishing Group http://www.nature.com/naturegenetics
NEWS AND VIEWS A key question is whether CDK2 ordinarily has an essential function that can be compensated by other CDKs or whether CDK2 is simply dispensable for typical mitotic cell cycles. The fact that many regulatory signaling pathways, which are thought to couple cell cycle progression to extrinsic anti-mitogenic or DNA damage signals, target CDK2 for inhibition supports the idea that CDK2 normally has an essential function. For example, the p53-dependent DNA damage checkpoints use p21 (also called Cip1) and Cdc25A to inhibit CDK2 and block the G1/S transition, and TGF-β signaling blocks Sphase entry by inhibiting both CDK4 and CDK2. Clearly, this interpretation requires that inhibition of CDK2 (as occurs during anti-mitogenic signaling) be different from loss of CDK2 (as in the knock-out mouse), because the former causes the cell cycle to arrest but the latter does not. Perhaps the presence of inhibited CDK2 impedes the ability of other CDKs to substitute for it. It will be important to determine whether these regulatory pathways operate correctly in the absence of CDK2. Cyclin E confounds If CDK2 is not essential—the Cdk2-null mouse is viable and overtly normal—then its main cyclin partners, the E-type cyclins (E1 and E2) and the A-type cyclins (A1 and A2), may be unnecessary as well. This is the case for cyclin A1, but it is not true for the others. Cyclin A1 is expressed exclusively in germ cell lineages, and the Ccna1-knockout mouse has a normal phenotype except for a defect in male meiosis I that resembles the defect reported for the Cdk2 knockout (surprisingly, female meiosis is normal in Ccna1-null mice; ref. 7). In contrast, cyclin A2 is expressed in all somatic cells, and its deletion results in early embryonic lethality8. But cyclin A2 associates both with CDK1 and CDK2, suggesting that the cyclin A2–CDK1 complex may execute the essential function of cyclin A2 in the Cdk2 knockout mouse.
10
Cyclins E1 and E2 are expressed together in all somatic cells, and deletion of either one alone has little effect on mouse development or cell proliferation in vitro. In new work from the labs of Peter Sicinski and Bruno Amati, mice engineered to lack both E-type cyclins have placental and cardiac defects but otherwise develop normally, and cells isolated from these mice undergo seemingly normal mitotic cell cycles in vitro4,5. But three notable cell cycle abnormalities are observed in these Ccne1- and Ccne2-null cells. The first is the inability of trophoblast giant cells and megakaryocytes to undergo endocycles in which DNA is replicated through successive S phases with no intervening mitoses. The second is the inability of quiescent (G0) cells to reenter the cell cycle, and the third is the resistance of cells lacking both E-type cyclins to undergo oncogenic transformation in vitro. Unexpectedly, then, the cell cycle in vertebrates may be able to proceed with just CDK1 and two cyclins: A2 for S phase and B for mitosis. Revising the orthodoxy In Drosophila, unlike the mouse, cyclin E is necessary for normal mitotic cell cycles9. Regulation of the anaphase-promoting complex (APC), which controls the degradation of A- and B-type cyclins, may underlie this difference10. In flies, cyclin E–CDK2 is the primary kinase that inactivates the APC at the end of G1, allowing accumulation of cyclin A and initiation of DNA replication. In mammals, however, cyclin A–CDK2 carries out this function, rendering cyclin E dispensable. Why is cyclin E uniquely required in endocycles and on exit from G0? Cyclin E seems to be necessary for assembly of the replication initiation complex, specifically for loading the MCM replicative helicase onto chromosomal DNA11. Notably, this function of cyclin E was shown in Drosophila endocycles12 as well as in nuclei prepared from mouse cells exiting quiescence, two settings in which cell cycle defects are observed in Ccne1- and Ccne2-null mice. Moreover,
cyclin E is seemingly not required for MCM loading as cells exit mitosis during a normal mitotic cycle13,14, which perhaps explains the normal phenotype in the Ccne1- and Ccne2null mice. Still, it is not understood why MCM loading proceeds by two different pathways depending on whether the cell approaches S phase from G0 or from mitosis. And what role does cyclin E have in this? These requirements for E-type cyclins but not CDK2 are perplexing. As far as we know, E-type cyclins associate exclusively with CDK2 in vivo and, indeed, seem to lack any associated kinase activity in Cdk2-null mice3. (E-type cyclins can interact with CDK3, but the gene Cdk3 is nonfunctional in the mouse15.) Although other explanations could still be entertained, it now seems possible that essential functions of cyclin E are not CDK-dependent. A rewriting of the cell cycle textbook may be in order. CDK2 can seemingly be discarded, and cyclin E is also dispensable under most circumstances. Neither seem to have the central role in S phase previously ascribed to them, but they may yet prove to be important in coupling the intrinsic cell cycle program to extrinsic regulatory inputs. In particular, the essence of cyclin E remains an enigma. 1. Fisher, D.L. & Nurse, P. EMBO J. 15, 850–860 (1996). 2. Sherr, C.J. & Roberts, J.M. Genes Dev. 13, 1501–1512 (1999). 3. Ortega, S. et al. Nat. Genet. advance online publication 17 August 2003 (doi:10.1038/ng1232). 4. Geng, Y. et al. Cell advance online publication, 15 August 2003 (doi:10.10160092867403006457). 5. Parisi, T. et al. EMBO J. (in the press). 6. Tetsu, O. & McCormick, F. Cancer Cell 3, 233–245 (2003). 7. Liu, D. et al. Nat. Genet. 20, 377–380 (1998). 8. Murphy, M. et al. Nat. Genet. 15, 83–86 (1997). 9. Knoblich, J. et al. Cell 77, 107–120 (1994). 10. Vodermaier, H. & Peters, J.-M. Nat. Cell Biol. 4, 119–120 (2002) 11. Coverley, D., Laman, H. & Laskey, R.A. Nat. Cell Biol. 4, 523–528 (2002). 12. Su, T.T. & O’Farrell, P.H. J. Cell Biol. 140, 451–460 (1998). 13. Dimitrova, D.S. & Gilbert, D.M. Mol. Cell 4, 983–993 (1999). 14. Mendez, J. & Stillman, B. Mol. Cell. Biol. 20, 8602–8612 (2000). 15. Ye, X., Zhu, C. & Harper, J.W. Proc. Natl. Acad. Sci. USA 98, 1682–1686 (2001).
VOLUME 35 | NUMBER 1 | SEPTEMBER 2003 NATURE GENETICS
