Oncogene (2005) 24, 2746–2755 & 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc Cell cycle-dependent transcription in yeast: promoters, transcription factors, and transcriptomes Curt Wittenberg1 and Steven I Reed*,2 1Department of Molecular Biology, MB-3, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA; 2Department of Molecular Biology, MB-7, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA In the budding yeast, Saccharomyces cerevisiae,a 1983) allowed for the testing of a large number of genes significant fraction of genes (>10%) are transcribed with for cell cycle regulation of transcription. Not surpris- cell cycle periodicity. These genes encode critical cell cycle ingly, many genes involved in specific cell cycle processes regulators as well as proteins with no direct connection to were found to be expressed maximally during specific cell cycle functions. Cell cycle-regulated genes can be cell cycle intervals where synthesis of encoded products organized into ‘clusters’ exhibiting similar patterns of could be rationalized in terms of function (Wittenberg regulation. In most cases periodic transcription is achieved and Reed, 1991). Based on single-gene analyses, it could via both repressive and activating mechanisms. Fine- be predicted that a significant but limited set of genes tuning appears to have evolved by the juxtaposition of encoding proteins directly involved in periodic cell cycle regulatory motifs characteristic of more than one cluster functions would be subject to cell cycle regulation. within the same promoter. Recent reports have provided However, more recently, large-scale array analysis of the significant newinsight into the role of the cyclin-dependent yeast transcriptome has revealed that a much larger set kinase Cdk1 (Cdc28) in coordination of transcription with of genes (minimally 800 or >10% of protein-encoding cell cycle events. In early G1, the transcription factor genes) exhibit cell cycle regulation and that many of the complex known as SBF is maintained in a repressed state encoded proteins are not ostensibly linked to cell cycle by association of the Whi5 protein. Phosphorylation of functions (Cho et al., 1998; Spellman et al., 1998). This Whi5 by Cdk1 in late G1 leads to dissociation from SBF presumably reflects a more intricate and subtle relation- and transcriptional derepression. G2/M-specific transcrip- ship between cellular physiology and cell cycle progres- tion is achieved by converting the repressor Fkh2 into an sion than has been appreciated. Furthermore, high- activator. Fkh2 serves as a repressor during most of the resolution scrutiny has revealed that even within so- cell cycle. However, phosphorylation of a cofactor, Ndd1, called clusters of coordinately-regulated genes, there are by Cdk1 late in the cell cycle promotes binding to Fkh2 significant differences observable between individual and conversion into a transcriptional activator. Such members, suggesting individualized optimization of insights derived from analysis of specific genes when transcriptional programming according to function combined with genome-wide analysis provide a more (Cho et al., 1998; Spellman et al., 1998). In this article, detailed and integrated viewof cell cycle-dependent we will review advances based on detailed analysis of transcription. transcriptional control of individual genes, but also Oncogene (2005) 24, 2746–2755. doi:10.1038/sj.onc.1208606 discuss the implications of findings based on genome- wide analysis. Keywords: transcription; cell cycle; cyclin-dependent protein kinase The G1 gene cluster Cell cycle initiation during G1 phase is the consequence Introduction of transcriptional activation of greater than 200 genes, many of which participate in events specifically asso- With the advent of modern molecular biology, some 30 ciated with cell cycle progression (Cho et al., 1998; years ago, it became possible to efficiently analyse Spellman et al., 1998). The primary cis-acting regulators transcription patterns of individual genes. In yeast, the of the G1-specific gene cluster are targeted by two convergence of relatively straightforward cell cycle alternative heterodimeric transcription factors com- synchronization protocols (Amon, 2002) with rapid prised of unique DNA-binding components, Swi4 and efficient mutational cloning techniques (Struhl, (SBF) or Mbp1 (MBF), and a common component, Swi6, that participates in transactivation (Breeden, 1996). The optimal binding site for MBF is called the *Correspondence: SI Reed; E-mail: [email protected] Mlu Cell cycle Box (MCB), due to the presence of an Cell cycle-dependent transcription in yeast C Wittenberg and SI Reed 2747 MluI restriction site in the consensus sequence, and that of at least a portion of SBF and MBF target genes, for SBF is the Swi4 Cell cycle Box (SCB) (Breeden, 1996; including CLN1 and CLN2 (Epstein and Cross, 1994; Di Lee et al., 2002; Kato et al., 2004). Como et al., 1995; Wijnen and Futcher, 1999). The target genes for SBF and MBF have been The identification of a central role for Cln3/Cdk1 traditionally defined by the presence of redundant SCB in transcriptional activation (Cross and Tinklenberg, and MCB sequences in their promoters. Functionally, 1991; Dirick and Nasmyth, 1991), and the finding that SBF and MBF targets, respectively, fall roughly into Cdk1 is likely to act directly rather than via its effect on two classes. MBF targets include those involved in the the accumulation of other proteins (Marini and Reed, control or execution of DNA replication and repair 1992), has led to a protracted search for CDK targets (POL2, CDC2, RNR1, CLB5/6), whereas SBF targets involved in activation. Swi6 and Swi4 are established include those involved in cell morphogenesis, spindle targets in vitro and Swi6 is known to be phosphorylated pole body duplication and other growth-related func- in a Cdk1-dependent manner in vivo (Sidorova et al., tions (CLN1/2, PCL1/2, GIN4, FKS1/2). This apparent 1995; Ubersax et al., 2003; Geymonat et al., 2004). differentiation into distinct functional classes may have Elimination of Cdk sites by mutation leads to a defect in adaptive significance due to regulatory differences nuclear import of Swi6 (Sidorova et al., 1995; Geymonat between the two transcription factors. However, each et al., 2004). However, those mutations appear not to group also includes many members that do not fall have an effect on the timing, extent of transcriptional neatly into these categories. Consequently, it is possible activation or Cdk dependence of transcriptional activa- that this separation is merely a reflection of the tion, suggesting that phosphorylation of Swi6 is not rate evolutionary history of these transcriptional regulatory limiting for access to the nucleus. modules (see below). The lack of a functional link between phosphoryla- Whereas the optimal binding sites for SBF and MBF tion of the transcription factors and transcriptional are distinct, there is considerable overlap in their activation has led to the suggestion that Cdk1 acts specificity both in terms of the DNA sequence require- indirectly via phosphorylation of another factor (Wijnen ments for binding and the occurrence of such sites et al., 2002), perhaps a previously unrecognized tran- within promoters. Consequently, there is considerable scriptional activator or repressor. Although a number of ambiguity concerning the role of each factor in proteins have been shown to interact with Swi6 (e.g., regulating specific genes (Partridge et al., 1997). This is Hrr25 (Ho et al., 1997), Skn7 (Morgan et al., 1995), and apparent from the global analysis of the binding of the Stb1 (Costanzo et al., 2003)), none appears to play a specific transcription factors to promoters (Iyer et al., significant role in the Cln/Cdk-dependent activation of 2001; Simon et al., 2001; Lee et al., 2002) and may transcription. Recently, the search for additional factors explain the modest effect of inactivation of SBF or MBF regulating G1 transcription led to the discovery of Whi5 on the expression of G1-specific genes (Koch et al., as an SBF-associated transcriptional repressor (Costan- 1993; de Bruin and Wittenberg, unpublished). zo et al., 2004; de Bruin et al., 2004). Inactivation of Whi5, previously shown to cause cell cycle initiation at a Regulation of G1-specific transcription small cell size (Jorgensen et al., 2002; Zhang et al., 2002), suppresses the requirement for Cln3in the absence of G1-specific transcription is promoted, in large part, by Bck2 (Costanzo et al., 2004; de Bruin et al., 2004). G1 cyclin-associated CDK activity. In fact, deregulation Consistent with that phenotype, inactivation of WHI5 of that process was the basis for the discovery of the G1 leads to premature activation of G1-specific transcrip- cyclin Cln3(Cross, 1988; Nash et al., 1988). Mutations tion during the G1 phase. The Whi5 protein associates leading to stabilization of Cln3result in premature with G1-specific promoters in an SBF-dependent expression of G1-specific genes and, thereby, small cell manner and is released from DNA coincident with size and resistance to G1-phase arrest by mating transcriptional activation. It is phosphorylated at Cdk pheromone. Any of the three G1 cyclins (Cln1, Cln2, consensus sites in vivo and in vitro and can be or Cln3) in conjunction with Cdk1 (Cdc28) were later phosphorylated in vitro by both G1- and B-type shown to be sufficient to mobilize Cdk1-dependent cyclin-associated Cdk1. Consistent with a role for activation of G1-specific transcription (Cross and Cdk1 in activation of SBF-dependent transcription, Tinkelenberg, 1991; Dirick and Nasmyth, 1991; Marini phosphorylation of SBF/Whi5 complexes in vitro and Reed, 1992). However, whereas Cln1- and Cln2- promotes dissociation of the complex. The regulation containing CDK complexes are sufficient for transcrip- of SBF by Whi5 bears striking resemblance to the tional activation, Cln3/Cdk1 is the primary activator regulation of the metazoan G1/S transcription factor, under physiological conditions (Tyers et al., 1993; E2F, by the retinoblastoma tumor suppressor protein, Dirick et al., 1995; Stuart and Wittenberg, 1995).