Division of Basic Sciences phase whereas CDK1 primarily with mitosis. However, assays using defined peptides reveal only subtle differ-Fred Hutchinson Cancer Research Center and Howard Hughes Medical Institute ences in substrate preference between CDK1 and CDK2 (Holmes and Solomon, 1996), and there is no compelling 1100 Fairview Avenue North Seattle, Washington 98104 evidence that CDK2 and CDK1 are intrinsically different with regard to their ability to initiate S and M phases. Indeed, when either of these genes are expressed in budding yeast they can rescue cdc28 mutations, per-The first molecular models of cell cycle regulation focused on how a single enzymatic oscillator, comprised forming both its S and M phase functions (Elledge and Spottswood, 1991). of a B-type cyclin and a cyclin-dependent kinase (CDK) controlledmitoticentryandexit. However, theseelegant A different explanation for the specific actions of CDKs is that their regulatory subunits, the cyclins, are and relatively simple models were soon recognized to be insufficient to account for the roles that CDKs play the major determinants of their biological specificity. All eukaryotes express multiple cyclins which assemble in multiple cell cycle events, including DNA replication, mitosis, centrosome/spindle pole body duplication, and into complexes with CDKs in various combinatorial arrangements, controlling the patterns and level of CDK cell morphogenetic changes. Various explanations were then considered for how activity, and perhaps also playing roles in substrate selection (Figure1A). Thus, eachstageofthecellcyclecan CDKs could control the orderly operation of the cell cycle, especially the regular alternation of S and M be characterized by its unique collection of cyclin–CDK complexes. It was not unreasonable to think that each phases. An early debate was over whether the ability to induce temporally specific events was intrinsic to the different cyclin–CDK complex would execute distinct cell cycle events and that the sequential assembly and CDK holoenzyme, or alternatively whether the changing state of the cell determined the particular cell cycle activation of these complexes could explain, at least in part, the order in which cell cycle events occur. events that happened in response to CDK activation. Both factors probably contribute, but there is strong In fact, there is little doubt that different cyclins can have intrinsically different biological actions. On the one evidence that the CDK is critically important (Stern and Nurse, 1996 and references therein). Thus, premature hand, there is a very clear distinction between the events that can be induced by the G1 versus S/G2 cyclins. In CDK activation in G1 can cause a cell to skip S phase and proceed directly to mitosis. Other manipulations of both budding and fission yeast, the S/G2 cyclins can initiate S phase and mitosis whereas the G1 cyclins CDK activity (for instance, turning the enzyme off and back on again) can cause a G2 cell to reenter S phase cannot (Schwob and Nasmyth, 1993; Fisher and Nurse, 1996). Conversely, the G1 cyclins can perform functions without an intervening mitosis. The idea that biological specificity resides in the CDK that the B-type cyclins cannot, one example being the unique ability of the G1 cyclins in budding yeast to reholoenzyme itself presents a paradox. In yeast, a single CDK is responsible for initiating all cell cycle transitions press the mating pathway (Oehlen et al., 1998) and another the inability of the yeast B-type cyclins to activate (CDC28 in S. cerevisiae and CDC2 in S. pombe, referred to here generically as CDK1). What explains its specific the G1 transcriptional program (see …