The Periodic Down Regulation of Cyclin E Gene Expression from Exit of Mitosis to End of G1 Is Controlled by a Deacetylase

Home , Cycle (gene)

The periodic down regulation of Cyclin E gene expression from exit of mitosis to end of G1 is controlled by a deacetylase- and E2F-associated bipartite repressor element

Jolanta Polanowska1, Eric Fabbrizio1, Laurent Le Cam1, Didier Trouche3, Stephane Emiliani2, Raphael Herrera4 and Claude Sardet*,1

1Institut de Genetique Moleculaire UMR 5535 / IFR24 CNRS, 1919 Route de Mende 34293, Montpellier cedex 5, France; 2Institut de Genetique Humaine UPR 1142 - CNRS 141, rue de la Cardonille 34396 Montpellier cedex 5, France; 3LBME IBCG CNRS 118, Route de Narbonne 31062 Toulouse cedex, France; 4Baylor College of Medecine, Department of Molecular and Cellular Biology, Houston, Texas, TX 77030, USA

The expression of cyclin E and that of a few other bona following cycles. These include genes encoding various ®de cell cycle regulatory genes periodically oscillates actors in proliferation such as c-fos,c-myc, B-myb, every cycle in proliferating cells. Although numerous various cdk inhibitors and several members of the E2F experiments have documented the role of E2F sites and family. Their expression is depressed in G0, strongly E2F activities in the control of these genes as cells exit stimulated as cells move through the ®rst G1 and S from G0 to move through the initial G1/S phase phases, and then varies little during the next cycles. transition, almost nothing is known on the role of E2Fs The second class, encoding few cellular factors during the subsequent cell cycles. Here we show that a controlling the cell cycle clock, DNA replication and variant E2F-site that is part of the Cyclin E Repressor mitosis, are both growth-stimulated and cell cycle Module (CERM) (Le Cam et al., 1999b) accounts for regulated. Their expression periodically oscillates every the periodic down regulation of the cyclin E promoter cycle in proliferating cells, ensuring an ordered observed between the exit from mitosis until the mid/late progression through the cell cycle. Thus, the G1-speci®c G1 phase in exponentially cycling cells. This cell cycle- and periodic expression of several proteins involved in dependent repression correlates with the periodic binding DNA replication is controlled at the transcriptional of an atypical G1-speci®c high molecular weight p107- level and ensures the correct timing of initiation of E2F complex (Cyclin E Repressor Complex: CERC2) DNA replication. This limited subset of cell cycle- that diers in both size and DNA binding behaviors from regulated genes includes, cdc6, cdk2, PCNA, Mcm 2, known p107-E2F complexes. Notably, anity puri®ed Mcm 6 and cyclin E (Dyson, 1998; Leone et al., 1998; CERC2 displays a TSA-sensitive histone deacetylase Nevins, 1998; Ohtani et al., 1996; Sardet et al., 1997). activity and, consistent with this, derepression of the Their expression increases from the M/G1 boundary to cyclin E promoter by trichostatin A depends on the late G1 and then drops as cells progress through S, G2 CERM element. Altogether, this shows that the cell and M phases. Surprisingly, the transcriptional me- cycle-dependent control of cyclin E promoter in cycling chanism that enforces this periodic expression is largely cells is embroiled in acetylation pathways via the unknown, as most promoter studies concerning these CERM-like E2F element. Oncogene (2001) 20, 4115 ± genes did not discriminate between the cell cycle- and 4127. the growth-dependent components of their regulation. In cells synchronized by serum starvation the G0/S Keywords: cyclin E; E2F; promoter; transcriptional activation of these genes is mainly dependent on repression; cell cycle binding sites for the E2F transcription factors, which mediate repression when bound by E2F-pocket protein complexes and/or activation when associated with Introduction (free) E2Fs (Dyson, 1998; Leone et al., 1998; Moberg et al., 1996; Sardet et al., 1997). However, it remains The stimulation of quiescent mammalian cells leads to unclear whether E2F sites are also responsible for their the induction of two classes of genes. The ®rst one cell cycle-dependent ¯uctuations. This doubt is quite includes strictly growth stimulated genes which are justi®ed by the fact that it has been recently reactivated during the ®rst cell cycle after serum demonstrated that other E2F-target genes, such as addition and then remain fully activated during the E2F1 or B-myb, are only growth-regulated with the E2F sites in their promoter actively mediating repression in quiescent but not in cycling cells (Campanero et al., 1999; Leone et al., 1998; Smith et al., 1996). To *Correspondence: C Sardet Received 5 January 2001; revised 6 April 2001; accepted 9 April address the role of E2F sites in cell cycle regulation, we 2001 have now used the cyclin E gene as a model of a bona Periodic regulation of cyclin E gene expression J Polanowska et al 4116 ®de cell cycle-regulated gene containing functional E2F al., 1999b). Although CERC1 contains E2F4/DP1 and sites in its promoter (Botz et al., 1996; Geng et al., the pocket protein p130, it diers in both binding 1996; Le Cam et al., 1999b; Ohtani et al., 1995). behavior and size from known E2F/DP-p130 com- The cyclin E/cdk2 kinase is an essential, rate-limiting plexes. CERC1 sediments faster (700 KD) than other regulator of the G1 to S phase transition in the known E2F complexes in a glycerol gradient, suggest- mammalian cell cycle (Donnellan and Chetty, 1999; ing the presence of additional components. Moreover, Keyomarsi and Herliczek, 1997; Sherr, 1996). Its anity puri®ed-CERC1 binds to the cyclin E promoter activity provides a threshold for entry into S phase, element CERM but not to canonical E2F sites in acting as a switch for various cellular processes, EMSAs (Le Cam et al., 1999b). including pRB-dependent transcription (Dyson, 1998; Here we have determined whether this E2F/CERM/ Nevins, 1998; Sardet et al., 1997), initiation of DNA CERC-dependent system also functions to time cyclin replication (Duronio et al., 1998; Hua et al., 1997; E expression in cycling cells. To address this question Krude et al., 1997; Krude, 2000), histone biosynthesis we performed detailed in vivo and in vitro analyses of (Ma et al., 2000; Zhao et al., 2000) and centrosome CERM/CERC-dependent regulation of the cyclin E duplication (Hinchclie et al., 1999). Those events promoter in exponentially cycling K562 human control S phase entry not only at the initial G1/S leukemic cells (Alitalo, 1990) separated by centrifugal transition as cells emerge from a quiescent state, but elutriation into dierent cell cycle compartments also at each subsequent G1/S boundary as cells (Merrill, 1998). We provide compelling evidence that proliferate. Cyclin E mRNA and protein rise sharply CERM regulates the cell cycle dependent repression of during a narrow window of time that precedes each the gene from the exit of mitosis to the end of G1 entry into S phase (Donnellan and Chetty, 1999; through its periodic association with a high molecular Keyomarsi and Herliczek, 1997; Le Cam et al., weight E2F complex, CERC2. Furthermore, puri®ed 1999b). There are multiple evidence showing that CERC2 displays a TSA-sensitive deacetylase activity E2Fs are involved in the control of this expression. and, consistent with this, derepression of the cyclin E In Drosophila and mouse, genetic studies indicate that promoter by trichostatin A depends on the CERM the cyclin E gene is a downstream target of an E2F- element; both observations suggesting that cell cycle- mediated transcriptional event (Duronio and O'Farrel, dependent control of cyclin E transcription involves 1995; Duronio et al., 1998; Herrera et al., 1996; acetylation /deacetylation of other transcription factors Humbert et al., 2000; Hurford et al., 1997). In cultured and/or chromatin (Ng and Bird, 2000; Wole et al., cells, cyclin E is rapidly induced following over- 2000). expression of E2F (DeGregori et al., 1995; Ohtani et al., 1995) or disruption of E2F-pocket protein repressor complexes by the viral oncoproteins AdE1A Results and HPV16 E7 (Spitkovsky et al., 1994; Vogt et al., 1999; Zerfass et al., 1995). Consistent with this, the Expression of cyclin E during a cell cycle in proliferating GC-rich promoter region of the mouse and human K562 cells genes contains several functional binding sites for E2Fs whose mutation aects the pro®le of activation in G0/S We ®rst analysed the cell cycle-dependent expression of (Botz et al., 1996; Geng et al., 1996; Le Cam et al., the endogenous cyclin E gene in proliferating cells. 1999b; Ohtani et al., 1995). Moreover, the presence of Exponentially growing human immature K562 ery- E2F proteins on the cyclin E promoter in living cells throid cells (a cell line derived from a blastic crisis of was recently demonstrated using chromatin immuno- chronic myeloid leukemia) were fractionated into precipitation (Wells et al., 2000) and high resolution in dierent cell cycle phases by counter¯ow centrifugal vivo genomic footprinting techniques (Le Cam et al., elutriation. This procedure separates logarythmically 1999b). On the mouse cyclin E promoter, we previously growing cells into healthy G1, S, and G2+M phase- demonstrated that one of these E2F binding sites was enriched subpopulations without the use of drugs, important for the timing of expression of cyclin E (Le therefore causing only minimal distortion in protein Cam et al., 1999b). This variant E2F site is located expression or metabolic functions (Merrill, 1998). The immediately downstream the transcription start site DNA content of 11 narrow size fractions was measured and is part of a larger element that we termed the by staining with propidium iodide and FACScan Cyclin E Repressor Module (CERM). In cells emerging analysis. Figure 1a shows that homogenous G1,S, from quiescence CERM function involves this E2F site and G2+M phase subpopulations were obtained. Two and a contiguous upstream AT-rich sequence, which representative fractions of G1 cells were also reseeded cooperate to delay cyclin E expression until late G1. into complete medium to check for respective position Both sequences are required to detect a nuclear of the various fractions within the G1 phase. As CERM-binding activity, termed CERC1 for Cyclin E illustrated in Figure 1b, these two fractions reached the Repressor Complex 1, which gradually disappears next S phase with a distinct kinetics, showing that they upon progression through G0 ±G1 and S. The disap- corresponded to distinct G1 subpopulations, represent- pearance of CERC1 activity correlates kinetically with ing early and late G1 cells, respectively. RNAs and the loss of occupation of the CERM module and with proteins were prepared from each fraction (Figure 2). cyclin E transcriptional induction in vivo (Le Cam et Cyclin E mRNA was almost undetectable in early/mid

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4117

Figure 1 Synchronization of K562 cells. Logarithmically growing K562 cells were separated according to their dierent cell cycle phases by centrifugal elutriation. Eleven distinct cell fractions obtained were cyto¯uorometrically analysed for DNA distribution. (a) FACScan pro®le after propidium iodine staining. Composition of each fraction is represented as the percentage of three subpopulations : G1 S and G2/M. (b) Two G1 fractions were reseeded in complete media (RPMI/ 10% FCS) and the kinetics of their cell cycle progression were followed by analysis of their DNA content 4, 8 and 13 h after reseeding. The dierential timing of G1/S transition of these two G1 subpopulations demonstrates that they represent early and late G1 cells, respectively

G1, then transiently peaked at the G1/S phase boundary and ®nally returned to the basal level as the cells progressed through G2, M and into the G1 phase of the next cycle (Figure 2a). The tight control of cyclin E expression was also re¯ected by the transient accumulation of the cyclin E protein, which appeared in late G1 and S, and then became undetectable as cells passed through G2 and M phases (Figure 2b).

The repressor element CERM controls the timing of cyclin E promoter activity during a cell cycle in proliferating K562 cells The timing of cyclin E promoter activity in mouse ®broblasts exiting G0 is controlled via the regulated selective association of an atypical high molecular E2F complex (CERC) with the bipartite repressor module (CERM) located nearby the transcription start site of the cyclin E gene (Le Cam et al., 1999b). To explore whether this regulatory element might also control the transient activity of the cyclin E promoter at each G1/S transition in exponentially growing cells, we monitored the cell cycle-dependent activity of cyclin E reporter genes in proliferating cells. Wild-type (proCE) and CERM-mutated versions of the promoter in which Figure 2 Homogenous subpopulations of elutriated K562 cells either the E2FX site (TGTCCCGC?TGTAGAGC, were analysed for the presence of cyclin E mRNA and protein. (a) (proCE-DCERM/E2Fmut) or the AT-sequence (TTT- Northern blot revealing the transient expression of cyclin E TAAAT?TCTGCAAT, proCE-DCERM/ATmut) was mRNA at the G1/S boundary; GAPDH mRNA is used as an modi®ed were introduced into K562 cells. Transfec- internal control. (b) Western blot analysis performed on total tants stably expressing these constructs were fractio- cellular extracts showing that cyclin E protein is present in late nated by centrifugal elutriation and each fraction was G1 and can be detected throughout the S phase until S/G2 transition assayed for luciferase activity (Figure 3).

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4118 major transcription start site (Geng et al., 1996). In the immediate vicinity of the major initiation site, we observed protection of G-residues within the E2FX site of CERM (Figure 4a). Although DMS reacts pre- dominantly with G-residues, it also methylates adenine (N3-methyladenine) with a lower eciency. This allowed us to reproducibly detect a protection of the AT-rich sequence located ®ve nucleotides upstream from the E2FX site (Figure 4a), as previously seen with the mouse cyclin E promoter (Le Cam et al., 1999b). Both DNA-protein contacts were clearly visible when the endogenous cyclin E gene was not expressed (i.e. in early G1 phase. In contrast, the region was no longer protected when cyclin E was expressed (G1/S transition). This shows that CERM (AT and E2FX sequences together) is bound by proteins during early G1 phase in exponentially growing cells in vivo. Interestingly, it remains unoccupied during the subsequent inactivation of the cyclin E gene, i.e. during S, G2 and M phases (J Polanowska and C Sardet, unpublished data). Figure 3 Role of the CERM element in the control of the cyclin E promoter during the ongoing cell cycle. Exponentially growing K562 cells stably expressing ®re¯y luciferase reporter constructs The atypical E2F complex CERC2 binds CERM every G1 driven either by the wild type (WT) cyclin E promoter (proCE) or in proliferating cells by its CERM mutated version (proCE-DCERM), were elutriated and analysed for DNA content as described in Figure 1. All cells In mouse ®broblasts exiting quiescence, we previously were cotransfected with a CMV-driven Renilla luciferase construct. Each distinct subpopulation of cells was then analysed showed that CERM selectively bound the atypical for ®re¯y luciferase activity normalized to renilla luciferase E2F-containing complex CERC1 and that this binding activity (left axis). The results represent the average of three was disappearing upon progression through G0 ±G1 in experiments using independent populations of transfectants correlation with the transcriptional induction of the cyclin E gene. To explore whether CERC activity is also regulated during the cell cycle of proliferating The proCE contruct yielded a several fold increase in cells, EMSA were performed with a DNA probe luciferase activity between early G1 and late G1/S spanning the CERM sequence of the human cyclin E phase fractions that re¯ects the activation of the promoter (probe CERM) and nuclear extracts pre- endogenous cyclin E gene. In contrast, the proCE- pared from the various elutriated fractions of K562 DCERM/ATmut construct was expressed in early G1 cells (Figure 5a). As a control for `regular' E2F cells and was only weakly induced during G1 (Figure activity, the same extracts were also incubated with a 3). Strictly similar results were obtained with the other probe bearing the E2F binding site from the proCE-DCERM construct (proCE-DCERM/E2Fmut) adenovirus E2A promoter (Figure 5d). EMSAs (J Polanowska et al., unpublished data). This func- performed with the E2 probe show that the cell cycle tional analysis suggests that inactivation of either the distribution of canonical E2F complexes in K562 cells AT-rich or the E2FX sequence of CERM is sucient varies little (Figure 5d). Speci®city controls (Figure 5e), to abolish the cell cycle-dependent oscillation of the deoxycholate (DOC) sensitivity (J Polanowska and C cyclin E promoter construct in exponentially growing Sardet, unpublished data) and antibody reactivity cells. (Figure 5f) tests showed that the three prominent complexes detected with this E2 probe were primarily composed of E2F4/DP1-p107-cyclinA, E2F4/DP1-pRB Periodic in vivo occupation of CERM in proliferating cell and E2F4/DP1 complexes, respectively (Figure 5f). Using a Dimethyl Sulfate (DMS)/LMPCR-based in In contrast, EMSA complexes formed on the vivo genomic footprinting technique, we previously CERM/E2F probe behaves dierently. A CERM- showed that CERM was occupied in mouse ®broblasts speci®c DNA-protein complex (termed CERC2) was during G0 and early G1 but not in late G1 and S phases detectable in early and mid G1 phase extracts, but then (Le Cam et al., 1999b). To con®rm the role of CERM disappeared in extracts prepared from other phases of in proliferating cells, we footprinted the CERM region the cell cycle (Figure 5a). Notably, the detection of this in elutriated early G1 and S phase subpopulations of EMSA complex correlates with the in vivo protection K562 cells (Figure 4). In vivo DMS-methylated K562 of CERM and with down regulation of the cyclin E DNA was used to generate human cyclin E promoter- promoter. These observations indicate that CERC2 speci®c ladders by ligation-mediated PCRs (LMPCR) might represent the cell cycle-regulated repressor (Le Cam et al., 1999a,b) using primers (Figure 4b) controling the G1-dependent repression of the cyclin designed to investigate the region encompassing the E gene in proliferating cells.

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4119

Figure 4 Identi®cation of cell cycle-regulated protein-binding sites protected in vivo in the transcription start site region of the human cyclin E promoter. (a) In vivo footprinting of the mouse cyclin E promoter around the major transcription start site in DMS- treated K562 cells. Early G1 and S phase cell subpopulations were isolated by centrifugal elutriation and treated by the methylating agent DMS. Ligation mediated polymerase chain reactions (LMPCR) were performed on the genomic templates obtained from these cells (lanes labeled in vivo) with the set of primers described in b. The in vitro lane refers to similar LMPCRs carried out with DMS-methylated naked DNA. (b) Summary of in vivo protections identi®ed around the major transcription start site of human cyclin E promoter. Protected bases are marked with ®lled circles. A region of DNA-protein contacts centered around a variant E2F site, termed E2FX and a AT-rich sequence, collectively refered to as CERM, is protected in a cell cycle regulated manner. Oligonucleotides used for LMPCR are indicated (ASCE1, ASCE2, ASCE3)

Interestingly, CERM/CERC2 association displayed unpublished data) or directed against cyclin A (Figure characteristics similar to those of CERM/CERC1 (Le 5c). Cam et al., 1999b). It was insensitive to competition by Although CERC2 contains E2F4, DP1 and p107, its `regular' E2F sites and by oligonucleotides bearing binding behavior suggests that it likely represents a mutations within either the E2FX or the AT-rich subclass of the cellular p107-E2Fcomplexes. To address sequence (Figure 5b). Conversely, complexes formed this question, nuclear extracts prepared from early G1- on the E2 probe were poorly competed out by CERM enriched K562 cells were fractionated in native glycerol oligonucleotides (Figure 5e). These data suggest that gradients. Each fraction was tested by EMSA for both both the AT and E2FX sequences of CERM are CERM (CERM probe) and whole E2F (E2 probe) required to bind CERC2 and that the E2FX site itself binding activities (Figure 6). The bulk of E2F binding is a low anity binding site for `regular' E2F activity, containing mainly E2F4/DP1-p107-cyclinA complexes. Consitent with this, we have previously (Figure 5f) was detected in gradient fractions 9 ± 13, shown that E2FX binds poorly to recombinant E2F/ with a maximum in fraction 11. CERC appeared in DP-pocket protein complexes (Le Cam et al., 1999b). fraction 11 and persisted until fraction 16, with a Nevertheless, CERC2 from elutriated G1-K562 cells maximum in fraction 13. This clear dierence in contained E2Fs and E2F-associated proteins (Figure sedimentation pro®les strongly suggests that CERC2 5c). CERC/CERM complex was supershifted or is an atypical E2F complex containing protein(s) in disrupted by antibodies directed against p107, E2F4 addition to E2F/DP and p107. and DP1 proteins but not by antibodies directed To gain further information about CERC2 compo- against pRB, p130 (Figure 5c) or other members of nents, this complex was puri®ed by DNA anity the E2F and DP families, namely E2F1, E2F2, E2F3, chromatography from a large quantity of nuclear E2F5 and DP2 (J Polanowska and C Sardet, extracts prepared from elutriated G1-K562 cells. In

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4120

Figure 5 EMSAs showing protein complexes formed with either the cyclin E promoter CERM element or with the E2F site of the adenovirus E2 promoter during the ongoing cell cycle. Exponentially growing K562 cells were elutriated and each subpopulation was analysed for DNA content as described in Figure 1. Nuclear extracts were prepared from each fraction and used for EMSA with either the CERM probe (a, b, c) or the E2 probe (d, e, f) as indicated. (a) A speci®c protein complex (CERC) binds the CERM module in early and mid G1 cells. NS: CERM-independent complexes. (b) Competition experiment showing the speci®city of the CERC complex for CERM and the requirement of both the AT-rich and E2FX sequences for binding. Nuclear extracts from early G1 cells were used for EMSA in presence of a 256excess of the WT CERM cold oligonucleotide (CERM), of CERM oligonucleotides containing mutations in either the E2FX sequence (DE2FX) or the contiguous AT-rich sequence (DAT), or with the E2 oligonucleotide. (c) EMSA complex CERC from G1-elutriated K562 cells reacts with antibodies directed against p107, E2F4 and DP1, respectively but not with those directed against either pRB or p130. (d) E2F complexes formed with the E2 probe during the ongoing cell cycle. Nuclear extracts are the same than those used in a.(e) Competition experiment showing that E2F complexes formed on the E2 probe are poorly competed out by a 256excess of cold CERM oligonucleotide (CERM). (f) Antibody interference assay showing that at least three types of E2F complexes can be detected in nuclear extracts from elutriated G1 cells using the E2 probe. p107-E2F4/DP1-cycA (a), pRB-E2F4/DP (b) and free E2F/DP1

brief, after a gel ®ltration step CERC2 was puri®ed on an oligonucleotide column bearing either the WT CERM sequence or the mutated version of CERM (Figure 7a). Notably, consistent with our EMSAs, we ®rst observed that `regular' E2F complexes (recombinant or cellular) were poorly absorbed on the CERM column (J Polanowska and C Sardet, unpublished data). Bound complexes were eluted in a high salt buer and analysed by Western blotting (Figure 7b) and EMSAs (J Polanowska and C Sardet, unpublished data). Using this strategy we previously showed that puri®ed CERC1 was able to bind CERM, this Figure 6 Ten to ®fty per cent glycerol gradient fractionation of requiring both the AT and E2FX sequences, but not to `regular' E2F sites (Le Cam et al., 1999b). The same nuclear extracts prepared from elutriated G1 cells. Similar amounts of indicated fractions were used to perform EMSAs result was obtained here for CERC2 (J Polanowska with either the E2 or the CERM probe as indicated and C Sardet, unpublished data). Immunoblot analyses

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4121 detected p107, E2F4, cyclin E, and cdk2 in this and BRG1, two members of the SNF2 family of enriched CERC2 eluate (Figure 7b), whereas pRB, helicase (Sudarsanam and Winston, 2000) and RbAp48 p130 (Figure 7b), E2F1 and E2F3 (J Polanowska and (Qian et al., 1993; Nicolas et al., 2000), a component of C Sardet, unpublished data) were absent. In addition several distinct nucleosome-modifying complexes. we probed this eluate for other mammalian proteins Although all these proteins were abundant in K562 that associate with pocket proteins, including: Hbrm total nuclear extracts, only RbAp48 was detected in CERC2 (Figure 7b and J Polanowska and C Sardet, unpublished data). Thus, CERC2 contains a subset of the components present in the prominent E2F complex detected with an E2 probe, namely E2F4/DP-p107- cyclin E-cdk2, as well as RbAp48. However, CERC2 probably contains other components, such as the still- unidenti®ed protein responsible for the in vivo occupation of the AT-rich sequence of CERM and the enzyme responsible for the histone deacetylase activity of CERC2 (see below).

Affinity purified CERC2 displays a histone deacetylase activity Consistent with the presence of RbAp48 and the putative role of CERC2 in repression, the puri®ed CERC2 fractions described above displayed a histone deacetylase activity (HDAC) when assayed on [3H]acetylysine-labeled K562 histones (Figure 8a). In contrast, no activity was detectable in the proteins eluted from the control DCERM column (Figure 8a). The HDAC activity of CERC2 was inhibited by 70 nM trichostatin A (TSA), an inhibitor of all known histone deacetylases (Figure 8a). Moreover, consistent with a link between CERM/CERC and the action of HDACs, Figure 7 Analyses of the protein components of CERC2 when the puri®ed CERC2 fraction was cleared of p107 puri®ed by CERM anity chromatography from nuclear extracts with a polyclonal antibody directed against p107 (Le of G1 K562 cells. (a) General scheme of the puri®cation Cam et al., 1999b) trapped on protein A agarose beads, procedure. (b) Western blot analysis of nuclear extracts of G1 no deacetylase activity could be recovered in the cells (G1) and of proteins eluted from either a CERM oligonucleotide column (CERM) or from a control column made supernatant (E Fabbrizio and C Sardet, data not of oligonucleotides bearing mutations within the CERM sequence shown). The presence of this HDAC activty in CERC2 (DCERM) suggests that it might control the cyclin E promoter via

Figure 8 CERC2 contains histone deacetylase activity and CERM mediates the TSA-sensitivity of the cyclin E promoter. (a) Eluates from either the CERM anity column (CERC) or from the DCERM column (control) (as described in Figure 7) were tested for deacetylase activity using acetylated histones as a substrate. The observed activity of CERC is trichostatin A (TSA, 70 nM)- sensitive. (b) Slowly exponentially growing K562 cells stably expressing ®re¯y luciferase reporter constructs driven either by the wild type (WT) cyclin E promoter (proCE-CERM) or by its CERM mutated version (proCE-DCERM), were treated with TSA (70 nM). At the indicated times after addition of the drug luciferase activities were measured as described in Figure 3

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4122 deacetylation. This prompted us to test whether the those including cdc6, cyclin E, PCNA, cdk2, MCM2, CERM element was responsible for the previously MCM6 (Leone et al., 1998; Dyson, 1998; Sardet et al., described TSA-induced activity of the cyclin E gene. To 1997; Nevins, 1998). this end, we analysed the TSA-response of the proCE What are the subtle characteristics that distinguish (CERM) and that of the DCERM-proCE reporter these E2F sites? At least one parameter seems to be contructs stably transfected in K562 cells. Addition of their architecture. Indeed, we previously showed that TSA to slowly growing K562 cells (4% FCS) CERM is a two-component repressive element com- signi®cantly stimulated the activity of the WT cyclin prised of an E2F site and of a contiguous AT-rich E promoter but only weakly that of the CERM sequence (Le Cam et al., 1999b). This bipartite mutant. This indicates that the CERM negative architecture is also found in CDE (Cell Cycle- element plays a key role in the TSA-sensitivity of the Dependent Element)-CHR (Cell Cycle Homology cyclin E gene. This also supports a model where the region) repressor elements which regulate the growth- periodic expression of the cyclin E gene is controlled by dependent expression of the cyclin A, cdc25C and a deacetylase-dependent repressive system relying on possibly the cdc2, B-myb, survivin, Plk and p130 genes the presence of the CERC2-bound CERM motif. (Bennett et al., 1996; Fajas et al., 2000; Huet et al., 1996; Li and Altieri, 1999; Lucibello et al., 1995; Uchiumi et al., 1997; Zwicker et al., 1995). Like Discussion CERM, these elements are located nearby the transcription initiation site and involve the cooperation of Numerous experiments document the role of E2F sites an AT-rich CHR motif with a contiguous GC-rich and E2F activities in the control of a variety of genes CDE motif. Inactivation of either region of the during the initial G1/S transition as cells emerge from bipartite repressor module results in high constitutive quiescence (Dyson, 1998; Nevins, 1998; Sardet et al., transcriptional activity of these promoters throughout 1997). In contrast, very little is known on their role at the cell cycle (Zwicker and Muller, 1997). Thus, both each subsequent G1/S transition as cells continue to E2F- and CDE-mediated cell cycle-dependent repres- proliferate. Here we show that a variant E2F-site that sion appears to depend on promoter-speci®c, AT-rich is part of the CERM repressor element, is responsible corepressor elements and involves a mechanism that, in for cyclic down-regulation of the cyclin E promoter both cases, remains to be identi®ed. between the exit of mitosis to mid/late G1 in CERM-mediated repression functions after each exponentially growing cells. We also show that the mitosis through the G1-speci®c binding of a protein cell cycle-dependent repression mediated by this complex, as evidenced by the in vivo footprint detected element correlates with its periodic association with a on CERM in G1. Moreover, this correlates with the G1-speci®c, high molecular weight E2F complex detection of a single nuclear CERM-speci®c binding (CERC2) displaying histone deacetylase activity. Con- activity (CERC2) by EMSA. This activity rises as cells sistent with this, the TSA-sensitive repression of the exit mitosis and drops dramatically upon S phase cyclin E promoter was found to depend on the CERM entry. This ®nding is unusual in the present state of element. the art of the E2F ®eld. Indeed, while several studies In ®broblasts emerging from a quiescent state the have described variations in the in vivo occupation of bipartite E2F-AT/CERM element located nearby the E2F sites during the G0-S transition, none could transcription start site of the cyclin E gene was acting reproduce these variations in vitro using EMSA solely as a repressor element during G0 and G1 (Le (Campanero et al., 1999; Tommasi and Pfeifer, 1995, Cam et al., 1999b). We now show that the CERM/E2F 1999; Wells et al., 1996; Zwicker et al., 1996). We element is necessary for the ecient down regulation of previously hypothesized that CERM is a two-compo- the cyclin E promoter activity upon exit from mitosis nent element which displays a weak anity for E2Fs in exponentially growing K562 cells. While mutations unless they are associated with an unidenti®ed in CERM lead to precocious promoter activity in early partner(s) within the CERC complex. In agreement G1, they do not increase its maximal intrinsic activity. with this, we found that neither recombinant E2Fs nor Therefore, CERM functions solely as a G1-speci®c free cellular E2F/DP heterodimers bind to a CERM repressive element and does not seem to be involved in probe in EMSAs while they bind strongly to a genuine transactivation. This observation provides the ®rst E2F site (Le Cam et al., 1999b). Conversely, puri®ed clear evidence that an E2F site, albeit non-canonical, CERC cannot bind to canonical E2F sites but binds is not only growth-responsive but also cell cycle- tightly to CERM in a manner strictly dependent on regulated. Notably, this contrasts with the strictly the presence of both the E2F and AT-rich sequences growth-responsive role of E2F sites in the E2F1 and B- (Le Cam et al., 1999b). myb promoters (Campanero et al., 1999, Leone et al., In Swiss 3T3 cells arrested in G0, the CERM-binding 1998, Smith et al., 1996). This suggests that two types activity, CERC1, contains E2F4/DP1 and p130 but of E2F binding sites exist, those that are only growth- sediments faster than the abundant E2F4/DP1-p130 responsive and those that also control transcription complexes detectable with a genuine E2F probe (Smith during each cell cycle. Consistent with this, whereas all et al., 1996, Sardet et al., 1997), suggesting the presence E2F target genes are growth-regulated, only a small of additional components in CERC (Le Cam et al., subset of them displays cell cycle regulation features, 1999b).

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4123 Here we show that the CERC2 complex which reforms after each mitosis contains a subset of the prominent E2F-pocket protein complexes found in proliferating K562 cells, i.e. E2F4-p107-cyclinE/cdk2. This observation is consistent with the recent ®nding that the mouse cyclin E promoter region containing CERM can be detected in chromatin immunoprecipi- tation experiments (ChlPs) performed with antibodies directed against E2F4 and p107 (Wells et al., 2000). However, the sedimentation characteristics and binding behavior of CERC2 dier from that displayed by genuine p107-E2F complexes. Notably, E2F4-p107 complexes can be detected by EMSA with a canonical E2F probe throughout the cell cycle whereas CERC2 is present only during G1. Moreover, CERC2 sedimentation pro®le suggests that, other, as yet unidenti®ed DNA-binding protein(s) must be present in this complex, including a factor recognizing the AT-rich sequence of CERM. This leads us to hypothesize that an atypical nuclear subset of the p107-complexes, those present in CERC2, are involved in cell cycle control. Many biochemical and genetic arguments demon- strate that E2F1, E2F3 and pRB are also involved in cyclin E promoter regulation (Botz et al., 1996; Figure 9 Model for the transcriptional control of the cyclin E DeGregori et al., 1995; Duronio and O'Farrel, 1995; promoter by CERC/CERM as detailed in Discussion Duronio et al., 1998; Geng et al., 1996; Herrera et al., 1996; Le Cam et al., 1999b; Ohtani et al., 1995; Humbert et al., 2000; Hurford et al., 1997; Wells et al., 2000; Zhang et al., 2000). However, none of these In contrast, puri®ed CERC2 also reacts with factors are present in anity puri®ed CERC2 which antibodies directed against RbAp48, a mammalian suggests that they likely interact with other E2F sites protein that binds pRb (Qian et al., 1993) and which present in the cyclin E promoter essential for is a component of at least four distinct nucleosome- transactivation (Botz et al., 1996; Le Cam et al., modifying complexes, the nuclear histone deacetylases 1999b; Ohtani et al., 1995). Consistent with this, other (HDACs), the drosophila nucleosome-remodeling E2F sites in the cyclin E promoter seem to be factor NURF, the chromatin assembly factor 1 constitutively occupied in vivo during the cell cycle (CAF-1), and Hat1, a type B (cytoplasmic) histone (Le Cam et al., 1999b; Wells et al., 2000). These data acetylase involved in chromatin assembly (reviewed in lead us to propose a model (Figure 9) where: (i) Ng and Bird, 2000; Wole and Pruss, 1996). CERC/CERM functions as an `o-switch', controlling Moreover, E2F1 and RbAp48 also physically associ- the timing of expression of the cyclin E gene by ate in the presence of pRb and HDAC1, suggesting repressing its transcription from the exit of mitosis to that RbAp48 could be involved in transcriptional the end of G1, (ii) other E2F sites, bound by the pRB- repression of E2F-responsive genes. (Nicolas et al., controlled E2F1 and E2F3 factors, control the level of 2000). Consistent with this presence of RbAp48, we activation of the promoter. ®nd that anity puri®ed CERC2 displays a TSA- Other candidate proteins potentially present in sensitive histone deacetylase activity (HDAC). How- CERC2 were those belonging to the hSWI/SNF ever, we failed to detect HDAC1 (or HDAC2 and nucleosome remodeling complex. Indeed, it was HDAC3) in CERC2 puri®ed fraction (data not reported that pRb can associate simultaneously with shown) suggesting that CERC is a so far yet an histone deacetylase (HDAC) and the SNF2-like undescribed novel type of pocket protein-RbAp48- (BRG1 and hbrm) component of the mammalian HDAC complex. Nevertheless, the presence of a hSWI/SNF nucleosome remodeling complex, at least HDAC activity in CERC2 suggests that its repressive when proteins were co-overexpressed in cells (Zhang et eect on cyclin E promoter involves deacetylation al., 2000). Moreover, in cotransfection experiments, and thus potential changes in chromatin structure. this complex was seen to inhibit transcription of Accordingly, inhibitors of histone deacetylase activity reporter genes driven by the cyclin E promoter and stimulate transcription of the endogenous cyclin E to arrest cells in the G1 phase of the cell cycle (Zhang gene as well as that of reporter constructs driven by et al., 2000). However, we did not ®nd either BRG1 or the cyclin E promoter (Brehm et al., 1998; Kim et hbrm in puri®ed endogenous CERC2, despite the fact al., 1999; Sambucetti et al., 1999; Sandor et al., 2000; that these proteins were abundant in our K562 nuclear Zhang et al., 2000). Moreover, our data attribute a extracts. large part of this eect to CERM, since mutations in

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4124 this element strongly diminish the eect of TSA on Cell synchronization and elutriation cyclin E transcription. The presence of HDACs in CERC2 is also consistent Counter¯ow centrifugal elutriations were performed using the with recent reports describing direct interactions JE-5.0 elutriation system (Beckman Instruments) and a rotor between the pocket proteins, pRB, p107 and p130 equipped with a 4 ml standard separation chamber. We used Master¯ex Model 6411-16 silicone tubing and medium ¯ow and histone deacetylase 1 (HDAC1). This recruitment was controlled with a Master¯ex pump, with standard Model of HDAC might explain how pocket proteins trigger 7016-20 pump head (Cole-Parmer). Log-phase growing K562 repression of promoters to which they have been cells were resuspended at 2.26108 cells in 50 ml phosphate- anchored by E2Fs (Brehm et al., 1998; Ferreira et al., buered saline (PBS) pH 7.4, supplemented with 50% FCS, 1998; Harbour et al., 1999; Nicolas et al., 2000; 0.9 mM CaCl2, and 0.5 mM MgCl2. Cells were loaded into the Robertson et al., 2000; Stiegler et al., 1998; Takahashi chamber at a ¯ow rate of 20 ml/min at a constant rotation et al., 2000; Zhang et al., 2000). Nevertheless, other speed of 2456 r.p.m. (580 g), followed by a 40 min equilibra- E2F-regulated genes such as E2F1 or DHFR are not tion time at the same ¯ow rate and rotation speed. The induced by histone deacetylase inhibitors, suggesting elutriation medium was PBS supplemented with 0.9 mM that this eect depends on promoter's structure CaCl2 0.5 mM MgCl2 and 2% FCS. Homogenous populations of cells with speci®c sizes were eluted out of the rotor (Sambucetti et al., 1999). by increasing the pump speed of 2 ± 3 ml/min; 150 ml The CERC-associated deacetylase activity could also fractions were collected at each ¯ow rate. Fractions were modify the acetylation dependent-activity of surround- analysed using a CASY-1 cell counter (Model TTC, Scharfe ing tanscription factors, including E2F family members System GmbH) and monitored for DNA content by themselves (Martinez-Balbas et al., 2000; Marzio et al., FACScan (Beckton Dickinson) after propidium-iodine label- 2000). However, since the TSA sensitivity of the cyclin ing. The same fractions were used to prepare RNA, DNA E reporter constructs was higher when plasmids were and proteins. chromosomally integrated, we currently favor a chromatin-dependent eect (J Polanowska et al., Northern blot analysis 2001, unpublished data). Moreover, the mapping of Total RNA was isolated from K562 cells by guanidinium nucleosomes on the cyclin E promoter and an thyocyanate-phenol/chloroform extraction. Eight mg of total evaluation of their acetylation level during the cell RNA were resolved by electrophoresis on 1% agarose gel cycle have been performed. This unpublished analysis containing 5% formaldehyde and transferred to nylon clearly shows that serum induction increases acetyla- membrane (Hybond-N, Amersham). The same membrane tion of a nucleosome positionned at the transcriptional was successively hybridized to a-32P-labeled probes derived start site, i.e. where CERM/CERC is located (AJ from human cyclin E or glyceraldehyde-3-phosphate dehy- Morrison, C Sardet and RE Herrera (2001) submitted). drogenase (GAPDH) full length cDNAs as previously Altogether this suggests that the control of the periodic described (Le Cam et al., 1999b). oscillation of cyclin E expression in proliferating cells depends on the CERM/CERC-mediated acetylation Protein extracts and EMSAs status of a nucleosome located in the transcription start K562 nuclear extracts were prepared as follows: cells were site region of the cyclin E gene. packed in Eppendorf tubes at 5000 g and resuspended in two volumes of nuclei preparation buer (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA pH 8, 0.1 mM EGTA, 2m MgCl ,1mM DTT, 30 mM NaF, 20 mM b-glyceropho- Materials and methods M 2 sphate, 0.3 mM PMSF) and incubated on ice for 15 min. Cell cultures, stable transfections and reporter assays Nuclei were isolated by addition of NP-40 to a ®nal concentration of 0.1% and centrifugation at 5000 g. The K562 cells (Alitalo) were grown in RPMI 1640 medium pellet was resuspended in two volumes of extraction buer (GIBCO ± BRL) supplemented with 10% (v/v) fetal calf (20 mM HEPES, pH 7.9, 0.42 M NaCl, 1 mM EDTA, 1 mM serum (FCS) (Biomedia) and 10 mM HEPES at 378Cina DTT, 30 mM NaF, 20 mM b-glycerophosphate, 0.3 mM 6 5% CO2 containing atmosphere. 2610 cells were transfected PMSF), vortexed at 48C, and centrifuged at 10 000 g,48C with Lipofectamine (GIBCO ± BRL) in 900 ml of OptiMEM for 20 min. Protein concentration of supernatants was with a reporter plasmid containing the ®re¯y luciferase gene adjusted at 8 mg/ml with extraction buer and extracts were (Promega) placed under the control of either the wild type conserved in 35% glycerol at 7808C. Double stranded cyclin E promoter (pro CE) or the promoter bearing probes for EMSAs were: E2, GCATAAGTTTCG- mutations at the CERM site (proCE-DCERM/E2FXmut, CGCCCTTTCTCAG (E2F consensus site from the adeno- proCE-DCERM/ATmut) as described previously (Le Cam et virus E2 promoter), E2mut, GCATAAGTTTCGA- al., 1999b). Transfections were performed for 6 h at 378Cina TCCCTTTCTCAG; CERM, CAGCTCGATGACGCTGG- 5% CO2 atmosphere. Reporter plasmids were cotransfected GATTTTTAAATGTCCCGCTCGAAGTCATC; CERM/ with a neomycin resistance expression vector, pCl-neo mutAT/mutE2F, CAGCTCGATGACGCTGGGATTCTG- (Promega) and a û-gal expression vector (pCH110). Neomy- CAATGTAGAGCTCGAAGTCAT C. Probes were labeled cin resistant populations stably expressing reporters were either by ®lling in 5' overhangs with a-32P-dCTP with Klenow selected in RPMI containing 12.5% FCS and 1 mg/ml G-418 polymerase (GIBCO ± BRL) or end-labeled with T4 poly- for 2 weeks. Measurements of luciferase activities were nucleotide kinase (New England Biolabs) and g-32P-ATP. All performed on 16105 cells by using a luciferase kit (Promega); probes were puri®ed by chromatography on G25 sephadex values were normalized to total proteins and b-galactosidase columns (TE, 150 mM NaCl). EMSAs were performed as activities as previously described (Le Cam et al., 1999b). follows: 5 mg of nuclear extracts were pre-incubated for

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4125 15 min at 228Cin10ml of EMSA buer (10 mM Tris-HCl EDTA, 50 mM NaCl). Gradients were centrifuged at pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 5% 40 000 r.p.m. in a SW 60 rotor for 12 h at 48C. Twenty glycerol, 0.1% BSA) and 0.5 mg of calf thymus genomic fractions of 200 ml each were collected starting from the top DNA (Pharmacia). Ten femtomoles of probe (56104 c.p.m.) of the gradient. Two microliters of each fraction were used were then added and samples were incubated for another for EMSAs performed with E2 and CERM probes. 30 min at 228C. DNA-protein complexes were then separated on a 4.5% acrylamide-bis acrylamide (19 : 1) non-denaturat- Affinity purification of CERC ing gel in 0.256TBE at RT and 10 V:cm. Dried gels were analysed with a Phosphorlmager (Molecular Dynamics, Inc). Twenty mg of nuclear cell extracts prepared from G1 For antibody perturbation experiments antibodies were elutriated K562 cells were loaded onto a Superdex 200 added during preincubation. Antibodies were obtained as column (Pharmacia) equilibrated in buer A (25 mM Tris follows: pRB (Pharmigen), p130 (Rb2, Transduction Labora- pH 7.8, 100 mM NaCl, 2 mM EDTA and 2 mM DTT). tories), p107 (C-18/sc-318), cyclin E (HE12/sc-247, C19/sc- Elution was done in buer A at 1.5 ml/min. CERC activity 198), E2F4 (C-20/sc-866), and RbAp48 (N-19/sc-8270) from containing fractions were pooled, three-fold diluted in buer Santa Cruz Biotechnology, Inc. and cyclin A (C-4710) from B (10 mM Tris-HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, Sigma. 1mM DTT, 5% glycerol, 0.1% BSA) containing calf tyhmus DNA (Pharmacia) at 75 mg/ml, incubated 30 min at 48C and applied onto an anity column which was prepared as Western blot analysis follows. Sixty nmoles of 5'-biotinylated double-stranded wild Whole protein extracts for immunoblotting were prepared as type or mutated CERM oligonucleotides were linked to 1 ml follows: K562 cell pellet washed twice in PBS was streptavidin agarose beads (GIBCO-BRL). The column was resuspended in ®ve volumes of triple detergent lysis buer equilibrated in buer B, saturated with 20 ml of ultra pure (50 mM Tris-HCl pH 8, 150 mM NaCl, 0.02% sodium azide, BSA (Boerhinger) at 200 mg/ml in buer B and washed with 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate, 2 mg/ml 20 volumes of column equilibration buer before loading the aprotinin, 2 mg/ml leupeptin, 0.5 mM PMSF), incubated sample. After reloading ®ve times the ¯ow through fraction, 30 min on ice, and centrifuged at 5000 g for 10 min. the column was extensivly washed with 50 volumes of buer Supernatants were heat denaturated in SDS-loading buer B containing 200 mM NaCl and eluted twice with 0.5 ml of (125 mM Tris-HCl pH 6.8, 2% SDS, 20% glycerol, 5% b- buer B containing 500 mM NaCl. Eluted material was mercaptoethanol, 0.1% bromophenol blue); 15 mg of total collected into siliconized tubes and stored at 7808C after proteins were separated by 10% acrylamide SDS ± PAGE, adding glycerol up to 30%. SDS ± PAGE and western blotted to PVDF membrane (Schleicher & Schuell) and blotting of eluted fractions were performed after heat probed with antibodies. denaturation in Laemmli buer.

In vivo DMS-genomic footprinting Deacetylase activity and TSA treatment The Whole procedure was performed essentially as described Deacetylase activity of the puri®ed CERC fraction was previously (Le Cam et al., 1999a,b). Brie¯y, intact cells measured as previously described (Emiliani et al., 1998) using (26106) were resuspended in 5 ml of cell culture medium 3H-acetylysine-labeled K562 histones prepared using standart (RPMI, FCS) buered with HEPES (20 mM ®nal concentra- protocoles (Carmen et al., 1996). 16105 slowly growing K562 tion), pH 7.4 and treated with the methylating agent DMS at cells (maintained in 4% FCS) stably expressing cyclin E 0.2% for 5 min at RT. Cells were then washed three times reporters (see above) were treated with trichostatin A (70 nM with cold PBS containing 2% b-mercaptoethanol and ®nal concentration, RPMI/FCS). Cellular extracts were collected in 1 ml lysis buer (50 mM Tris pH 8.0, 20 mM prepared 4, 8, 12 and 18 h after TSA addition and luciferase EDTA, 1% SDS, 2% b-mercaptoethanol). Genomic DNA activity was measured as described above. was isolated by three phenol extractions followed by two rounds of precipitation in 4 M ammonium acetate with three volumes of ethanol. As a reference, genomic DNA from K562 cells was methylated in vitro with 0.5% DMS for 4 min Acknowledgments at RT. Then, samples were piperidine cleaved (958C, 30 min), This work was funded by grants from the french CNRS precipitated in ethanol and resuspended at 0.4 mg/ml in (ATIPE n83), from `l'Association pour la Recherche contre water. Two mg of cleaved DNA were used for LMPCR le Cancer (ARC)', from `La Ligue Contre le Cancer' and reactions carried out as described (Le Cam et al., 1999a,b). from the `Human Frontier in Science Program'. J Samples were loaded onto a 5% sequencing gel and run at Polanowska was supported by a Boehringer Ingelheim 50 W. Dried gels were analysed with a Phospholmager doctoral fellowship and L Le Cam by a predoctoral (Molecular Dynamics, Inc). fellowship from La ligue Contre le Cancer. We are grateful to Dr E Schwob for precious help on developing centrifugal elutriation, Dr O Coux and F Martin for help Native glycerol gradient analysis of CERC to purify CERC, and to Dr C Murchardt and J De Caprio Two mg of nuclear extracts prepared G1 fraction of elutriated for reagents. We would like to express special thanks to B K562 cells were applied to a 10 ± 50% glycerol gradient (8 ml) Hipskind, JM Blanchard, L Dirick and A Le Cam for in modi®ed EMSA buer (10 mM Tris-HCl pH 7.5, 1 mM critical reading of the manuscript.

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4126 References

Alitalo R. (1990). Leuk. Res., 14, 501 ± 514. MaT,VanTineBA,WeiY,GarrettMD,NelsonD,Adams Bennett JD, Farlie PG and Watson RJ. (1996). Oncogene, 13, PD, Wang J, Qin J, Chow LT and Harper JW. (2000). 1073 ± 1082. Genes Dev., 14, 2298 ± 2313. Botz J, Zerfass-Thome K, Spitkovsky D, Delius H, Vogt B, Martinez-Balbas MA, Bauer UM, Nielsen SJ, Brehm A and Eilers M, Hatzigeorgiou A and Jansen-Durr P. (1996). Kouzarides T. (2000). EMBO. J., 19, 662 ± 671. Mol. Cell. Biol., 16, 3401 ± 3409. Marzio G, Wagener C, Gutierrez MI, Cartwright P, Helin K BrehmA,MiskaEA,McCanceDJ,ReidJL,BannisterAJ and Giacca M. (2000). J. Biol. Chem., 275, 10887 ± 10892. and Kouzarides T. (1998). Nature, 391, 597 ± 601. Merrill GF. (1998). Methods Cell. Biol., 57, 229 ± 249. Campanero MR, Armstrong M and Flemington E. (1999). Moberg K, Starz MA and Lees JA. (1996). Mol. Cell. Biol., Mol. Cell. Biol., 19, 8442 ± 8450. 16, 1436 ± 1449. Carmen AA, Rundlett SE and Grunstein M. (1996). J. Biol. Nevins JR. (1998). Cell. Growth Dier., 9, 585 ± 593. Chem., 271, 15837 ± 15844. Nicolas E, Morales V, Magnaghi-Jaulin L, Harel-Bellan A, DeGregori J, Kowalik T and Nevins JR. (1995). Mol. Cell. Richard-Foy H and Trouche D. (2000). J. Biol. Chem., Biol., 15, 4215 ± 4224. 275, 9797 ± 9804. Donnellan R and Chetty R. (1999). FASEB. J., 13, 773 ± 780. Ng HH and Bird A. (2000). Trends Biochem. Sci., 25, 121 ± Duronio RJ and O'Farrel PH. (1995). Genes Dev., 9, 1456 ± 126. 1468. Ohtani K, DeGregori J and Nevins JR. (1995). Proc. Natl. Duronio RJ, Bonnette PC and O'Farrel PH. (1998). Mol. Acad.Sci.USA,92, 12146 ± 12150. Cell. Biol., 18, 141 ± 151. Ohtani K, DeGregori J, Leone G, Herendeen DR, Kelly TJ Dyson N. (1998). Genes Dev., 12, 2245 ± 2262. and Nevins JR. (1996). Mol. Cell. Biol., 16, 6977 ± 6984. Emiliani S, Fischle W, Van Lint C, Al-Abed Y and Verdin E. Qian YW, Wang YC, Hollingsworth RE, Jones D, Ling N (1998). Proc. Natl. Acad. Sci. USA, 95, 2795 ± 2800. and Lee EY. (1993). Nature, 364, 648 ± 652. Fajas L, Le Cam L, Polanowska J, Fabbrizio E, Servant N, RobertsonKD,Ait-Si-AliS,YokochiT,WadePA,JonesP Philips A, Carnac G and Sardet C. (2000). FEBS Lett., and Wole AP. (2000). Nat. Genet., 25, 338 ± 342. 471, 29 ± 33. SambucettiLC,FischerDD,ZabludoS,KwonPO, Ferreira R, Magnaghi-Jaulin L, Robin P, Harel-Bellan A Chamberlin H, Trogani N, Xu H and Cohen D. (1999). and Trouche D. (1998). Proc. Natl. Acad. Sci. USA, 95, J. Biol. Chem., 274, 34940 ± 34947. 10493 ± 10498. Sandor V, Senderowicz A, Mertins S, Sackett D, Sausville E, Geng Y, Ng Eaton E, Picon M, Roberts JM, Lundberg A, Blagosklonny MV and Bates SE. (2000). Br.J.Cancer,83, Giord A, Sardet C and Weinberg RA. (1996). Oncogene, 817 ± 825. 12, 1173 ± 1180. Sardet C, Le Cam L, Fabbrizio E and Vidal M. (1997). In Harbour JW, Luo RX, Dei Santi A, Postigo AA and Dean Progress in Gene Expression, Oncogenes as Transcriptional DC. (1999). Cell, 98, 859 ± 869. Regulators, Karin M, Ghysdael J and Yaniv M (eds) Herrera RE, Sah VP, Williams BO, Makela TP, Weinberg Birkhauser Verlag Publisher: Basel, Swiss. Vol 2, pp. 1 ± RA and Jacks T. (1996). Mol. Cell. Biol., 16, 2402 ± 2407. 62. Hinchclie EH, Li C, Thompson EA, Maller JL and Sluder Sherr CJ. (1996). Science, 274, 1672 ± 1677. G. (1999). Science, 283, 851 ± 854. Smith E, Leone G, DeGregori J, Jakoi L and Nevins JR. Hua XH, Yan H and Newport J. (1997). J. Cell. Biol., 137, (1996). Mol. Cell. Biol., 16, 6965 ± 6976. 183 ± 192. Spitkovsky D, Steiner P, Lukas J, Lees E, Pagano M, Schulze Huet X, Rech J, Plet A, VieÂ A and Blanchard JM. (1996). A, Joswig S, Picard D, Tommasino M, Eilers M and Mol. Cell. Biol., 16, 3789 ± 3798. Jansen-DuÈrr P. (1994). J. Virol., 68, 2206 ± 2214. Humbert PO, Verona R, Trimarchi JM, Rogers C, Stiegler P, De Luca A, Bagella L and Giordano A. (1998). Dandapani S and Lees JA. (2000). Genes Dev., 14, 690 ± Cancer Res., 58, 5049 ± 5052. 703. Sudarsanam P and Winston F. (2000). Trends Genet., 16, Hurford RK, Cobrinik R, Lee MH and Dyson N. (1997). 345 ± 350. Genes Dev., 11, 1447 ± 1463. Takahashi Y, Rayman JB and Dynlacht BD. (2000). Genes Keyomarsi K and Herliczek TW. (1997). Prog. Cell. Cycle Dev., 14, 804 ± 816. Res., 3, 171 ± 191. Tommasi S and Pfeifer GP. (1995). Mol. Cell. Biol., 15, Kim YB, Lee KH, Sugita K, Yoshida M and Horinouchi S. 6901 ± 6913. (1999). Oncogene, 18, 2461 ± 2470. Tommasi S and Pfeifer GP. (1999). J. Biol. Chem., 274, Krude T, Jackman M, Pines J and Laskey RA. (1997). Cell, 27829 ± 27838. 88, 109 ± 119. Uchiumi T, Longo DL and Ferris DK. (1997). J. Biol. Chem., Krude T. (2000). J. Biol. Chem., 275, 13699 ± 13707. 272, 9166 ± 9174. Le Cam L, Polanowska J, Fabbrizio E, Fajas L and Sardet C. Vogt B, Zerfass-Thome K, Schulze A, Botz JW, Zwerschke (1999a). Biotechniques, 36, 840 ± 843. W and Jansen-Durr P. (1999). J. Gen. Virol., 80, 2103 ± Le Cam L, Polanowska J, Geng Y, Fabbrizio E, Olivier M, 2113. Philips A, Ng Eaton E, Classon M and Sardet C. (1999b). Wells J, Held P, Illenye S and Heintz NH. (1996). Mol. Cell. EMBO. J., 18, 1878 ± 1890. Biol., 16, 634 ± 647. Leone G, DeGregori J, Yan Z, Jakoi L, Ishida S, Williams Wells J, Boyd KE, Fry CJ, Bartley SM and Farnham PJ. RS and Nevins JR. (1998). Genes Dev., 12, 2120 ± 2130. (2000). Mol. Cell. Biol., 16, 5797 ± 5807. Li F and Altieri DC. (1999). Cancer Res., 59, 3143 ± 3151. Wole AP and Pruss D. (1996). Cell, 84, 817 ± 819. Lucibello FC, Truss M, Zwicker J, Ehlert F, Beato M and Wole AP, Urnov FD and Guschin D. (2000). Biochem. Soc. Muller R. (1995). EMBO. J., 14, 132 ± 142. Trans., 28, 379 ± 386.

Oncogene Periodic regulation of cyclin E gene expression J Polanowska et al 4127 Zerfass K, Schulze A, Spitkovsky D, Friedman V, Henglein Zwicker J, Lucibello FC, Wolfram LA, Gross C, Trus M, B and Jansen-Dur P. (1995). J. Virol., 69, 6389 ± 6399. Engeland K and MuÈller R. (1995). EMBO J., 14, 4514 ± Zhang HS, Gavin M, Dahiya A, Postigo AA, Ma D, Luo 4522. RX, Harbour JW and Dean DC. (2000). Cell, 101, 79 ± 89. Zwicker J, Liu NS, Engeland K, Lucibello FC and Muller R. Zhao J, Kennedy BK, Lawrence BD, Barbie DA, Matera (1996). Science, 271, 5255 ± 5254. AG, Fletcher JA and Harlow E. (2000). Genes Dev., 14, Zwicker J and Muller R. (1997). Trends Genet., 1, 3±6. 2283 ± 2289.

Oncogene