Functional Interplay Between P53 and E2F Through Co-Activator P300

Chang-Woo Lee, Troels S Sùrensen, Noriko Shikama and Nicholas B La Thangue

Division of Biochemistry and Molecular Biology, Davidson Building, University of Glasgow, Glasgow G12 8QQ, UK

Both E2F and p53 are sequence speci®c transcription early cell cycle progression where, in the appropriate factors that regulate early cell cycle progression. The conditions, cells become committed to the division pathway of control mediated through E2F governs the cycle. The transition from G1 into S phase is governed transition from G1 into S phase whereas p53 in response by a number of proteins with established roles in to genotoxic stress can facilitate cell cycle arrest or controlling proliferation, such as the retinoblastoma apoptosis. The mechanisms which in¯uence the outcome tumour suppressor protein (pRb), which is believed to of p53 induction are not clear, although transcription of be a molecule of pre-eminent importance in regulating the p53 target gene, encoding the cdk-inhibitor p21Waf1/ this transition (Weinberg, 1995), a role exerted Cip1, correlates with p53-mediated cell cycle arrest. Here principally through an ability to modulate the activity using a combination of biochemical and functional assays of transcription factors such as E2F (La Thangue, we identify p300 as a co-activator required for p53- 1994; Nevins, 1992). Indeed, E2F is a pivotal point of dependent transcriptional activation of Waf1/Cip1. control in the G1/S phase transition, allowing the Furthermore, we show that the cdk-inhibitor p21Waf1/Cip1 activity of proteins such as pRb to integrate with the autoregulates in a positive fashion transcription through activity of target genes required for cell cycle modulating the activity of the p53/p300 complex, whilst progression (La Thangue, 1994). negatively regulating the activity of E2F by preventing The activity of pRb is regulated at the level of cdk-dependent phosphorylation of pRb. Consistent with a phosphorylation by cyclin-dependent kinases (cdks). role for p21Waf1/Cip1 in the autoregulation of p53- Hypophosphorylated pRb, which is active in negative dependent transcription, p300 augments the ability of growth control and binding to E2F (Weinberg, 1995), p53 to cause G1 arrest and, conversely, cells undergoing is progressively phosphorylated during early cell cycle p53-dependent apoptosis are rescued by p300. Thus, our through the sequential activation of cdks (Pines, 1995; data suggest that the ability of p300 to interact with p53 Sherr, 1993). For pRb, cyclin D/cdk4 and cyclinE/cdk2 in¯uences the physiological consequence of p53 activa- are believed to be the predominant cdks involved in its tion. From previous studies it is known that cells phosphorylation (Dowdy et al., 1993; Ewen et al., expressing aberrant levels of E2F-1 can undergo p53- 1993; Hinds et al., 1992). In turn, the activity of cdk dependent apoptosis. In addition, we ®nd that E2F-1 can complexes is subject to aerent control through cause apoptosis in p537/7 tumour cells and further p300, another group of proteins, the cdk-inhibitors (Sherr which also functions as a co-activator for the E2F/DP and Roberts, 1995), which fall into two general classes heterodimer, enhances the apoptotic activity of E2F-1. In that, in each case, are exempli®ed by p21Waf1/Cip1 or conditions where E2F-1 and p53 co-operate in apoptosis p16INK4a (Sherr and Roberts, 1995). Members of the E2F-1 can eectively compete for p300, causing a p21Waf1/Cip1 group are antagonists of dierent cdk reduction in p53-dependent transcription. Thus, a complexes, including cyclin D and E-dependent functional interaction between p300 and either p53 or kinases (Harper et al., 1993; Xiong et al., 1993), E2F-1 has a profound impact on early cell cycle whereas the p16INK4a group preferentially regulate cyclin progression, speci®cally in regulating the contrasting D-dependent kinase (Serrano et al., 1993). The cdk- outcomes of cell cycle arrest and apoptosis. These results inhibitors impede cell cycle progression in part by suggest a critical role for p300 in integrating and co- preventing the phosphorylation of pRb (Koh et al., ordinating the functional interplay between the pathways 1995; Lukas et al., 1995). of growth control mediated by E2F and p53. Another important point of control during early cell cycle progression requires the p53 protein which Keywords: p53; E2F; p300; apoptosis; transcription remains in a latent state but, following genotoxic or other forms of stress, can cause cell cycle arrest or apoptosis (Ko and Prives, 1996). A primary function of p53 which is believed to correlate with cell cycle arrest Introduction relates to its role as a sequence-speci®c transcription factor (Crook et al., 1994; Pietenpol et al., 1994) for Progression through the mammalian cell cycle requires which a number of target genes have been identi®ed. that gene expression is co-ordinated with the activity of One of these encodes p21Waf1/Cip1 (El-Deiry et al., 1993; cell cycle control proteins. A critical period for Harper et al., 1993) which, as discussed, inhibits the integrating growth regulating signals occurs during activity cdks and thus is believed to maintain pRb and related proteins in a hypophosphorylated growth- suppressing state. The activity of p21Waf1/Cip1 is sucient to cause cell cycle arrest (Dulic et al., 1994; El-Deiry et al., 1993; Harper et al., 1993; Waldman et al., 1995) and, consequently, its activation likely to be Correspondence: NB La Thangue Received 8 October 1997; revised 23 December 1997; accepted 23 important in p53-induced cell cycle arrest (Chen et al., December 1997 1996; Macleod et al., 1995). However, it is possible that Early cell cycle control by p300 C-W Lee et al 2696 p53 regulates other targets required for cell cycle arrest E2F by preventing cdk-dependent phosphorylation of because in p21Waf1/Cip1 knockout cells G1 arrest pRb. Consistent with a role for p21Waf1/Cip1 in the following irradiation is only partially abolished autoregulation of p53-dependent transcription, p300 (Brugarolas et al., 1995; Deng et al., 1995), whereas augments the ability of p53 to cause G1 arrest and, in p537/7 cells G1 arrest is completely absent conversely, cells undergoing p53-dependent apoptosis (Donehower et al., 1992). are rescued by p300. Our data therefore suggest that The mechanisms involved in p53-regulated apoptosis the role of p300 as a co-activator for p53 contributes to are less clear. In some situations, transcriptional an ecient G1 arrest. Moreover, we ®nd that both E2F activation by p53 is dispensable for the induction of and p53 utilize p300 as a transcriptional co-activator apoptosis since mutants in p53 which fail to activate and that p300 enhances the apoptotic activity of E2F- transcription can still induce apoptosis (Chen et al., 1. Thus, the interaction of p300 with either p53 or 1996; Haupt et al., 1995b) and p53-dependent E2F-1 has a profound impact on early cell cycle apoptosis can occur in the presence of inhibitors of progression. Indeed, competition between the activa- transcription and translation (Caelles et al., 1994). tion domains of p53 and E2F-1 for p300 may be Other studies have suggested a role for transcriptional instrumental in determining which physiological out- activation by p53 (Attardi et al., 1996; Chen et al., come ensues. These results suggest a critical role for 1996; Sabbatini et al., 1995), such as for the Bax gene p300 in orchestrating early cell cycle progression (Miyashita and Reed, 1995), the product of which can through the pathways of control mediated by E2F antagonize the anti-apoptotic activity of Bcl-2 (White, and p53. 1996). Thus, for p53-dependent cell cycle arrest there is a good correlation with sequence speci®c transcriptional activation, although both transcription-dependent and independent mechanisms appear to contribute a to p53-dependent apoptosis. A considerable body of evidence argues that the p53 and pRb/E2F pathways are integrated not only at the level of cell cycle arrest, via the induction of p21Waf1/Cip1 and subsequent control of pRb activity, but through pathways which govern apoptosis. For example, mice carrying a targeted disruption in the Rb gene die in utero at about 13.5 days of gestation which, in part, is due to extensive apoptosis in a variety of tissues (Clarke et al., 1992; Jacks et al., 1992). Similarly, in tissue culture cells inactivation of pRb through the adenovirus E1a protein induces apoptosis (Debbas and White, 1993; Sabbatini et al., 1995), and targeting the HPV E7 protein to lens ®bre cells promotes an apoptotic outcome (Pan and Griep, 1994). Subsequent inactivation of p53, either by co-expression of the HPV E6 protein (Pan and Griep, 1994) or targeted b 4.5 30301.5 4.5 1.5 :Chase (h) disruption of the p53 gene (Morgenbesser et al., E2F–1 HA1 1994), overcomes the apoptosis evident in conditions +++ +++++– :p53 of Rb loss. Additional evidence for the integration between the p53 and E2F pathways has arisen from studies on E2F-1. Thus, although increased levels of E2F-1 can cause quiescent cells to enter S phase (Johnson et al., 1993), it has become increasingly clear p53 that usually this is followed by apoptosis (Qin et al., 1994; Wu and Levine, 1994) which, moreover, is enhanced by the presence of wild-type p53 (Wu and Levine, 1994). Furthermore, several studies have shown 123456789 that components within either pathway physically Figure 1 Regulation of p53-dependent transcription by E2F-1 associate with each other (Martin et al., 1995; and pRb. (a) The p53 reporter pTG13-GL was introduced into O'Conner et al., 1995; Sùrensen et al., 1996). Studies SAOS-2 cells together with expression vectors encoding p53 (1 mg; such as these suggest that the apoptotic outcome tracks 2 ± 17), in the presence of E2F-1 (0.5 mg in tracks 3, 9, 10 and 11; 1.0 mg in tracks 4, 12, 13 and 14 or 2.0 mg in tracks 5, 15, caused through aberrant control of the pRb-E2F 16 and 17), and/or pRb (1.0 mg in tracks 6, 9, 12 and 15, 6 mgin pathway is integrated with p53. tracks 7, 10, 13 and 16, and 12 mg in tracks 8, 11, 14 and 17). The Here, we have studied p53-dependent transcriptional data presented re¯ect averages from two sets of values. activation and identify p300 as a co-activator required Throughout pCMV-bgal was included in each transfection as an internal control. The relative activity of luciferase to b- for this process, a conclusion supported by recent galactosidase is presented. (b) The in¯uence of E2F-1 on p53 studies (Avantaggiati et al., 1997; Gu et al., 1997; Lill protein level was assessed in SAOS-2 cells transfected with et al., 1997; Somasundaram and El-Diery, 1997). We expression vectors encoding p53 (3 mg; tracks 1 ± 8), E2F-1 (3 mg; Waf1/Cip1 tracks 1 ± 4), or the control vector HA1 (3 mg; tracks 5 ± 8), show that the product of the p53 target gene p21 35 autoregulates in a positive fashion transcription metabolically labelled with S-methionine as described and thereafter chased for the indicated times. The levels of p53 were through modulating the activity of the p53/p300 resolved by immunoprecipitation. A non-transfected treatment is complex whilst negatively regulating the activity of shown in track 9 Early cell cycle control by p300 C-W Lee et al 2697 involved by evaluating the eects of E2F on p53- Results dependent transcription. Thus, we introduced wild-type p53 into the osteosarcoma-derived cell lines SAOS2 and Regulation of p53-dependent transcription by E2F and U2OS, and thereafter monitored the activity of a p53 pRb reporter which was dependent upon p53 transcriptional To address the interplay between E2F and p53, we activity. The reporter of choice, pTG13-GL, contains p53 considered that we may gain insight into the mechanisms binding sites immediately upstream of the herpes virus

Figure 2 Regulation of p53 activity requires the E2F-1 and p53 activation domains. (a) The E2F reporter pDHFR-luc was introduced into SAOS-2 cells together with expression vectors encoding E2F-1 (0.2 mg; tracks 2 and 3) or E2F-1D413 (0.2 mg; tracks 4 and 5) together with DP-1 (2 mg; tracks 3 and 5). The data presented re¯ect averages from two sets of values. Throughout pCMV- bgal was included in each transfection as an internal control, the relative activity of luciferase to b-galactosidase being presented. (b) The p53 reporter pTG13-GL was introduced into SAOS-2 cells together with expression vectors encoding E2F-1 (1 mg in track 3, and 3 mg in 4 and 7) or E2F-1D413 (1 mg in tracks 5, and 3 mg in tracks 4, 6 and 8). Values were calculated as described in (a). (c) Either the Gal4 reporter pG5-luc (1 mg; tracks 1 ± 3) or pTG13-GL (1 mg; tracks 4 ± 9) was introduced into SAOS-2 cells together with expression vectors encoding Gal4-p53 (1 mg; tracks 2 and 3), p53 (1 mg; tracks 5 and 6) or p53-VP16 (1 mg; tracks 8 and 9) in the presence of E2F-1 (2 mg; tracks 3, 6, and 9). The data presented re¯ect averages from two sets of values. pCMV-bgal was included in each transfection as an internal control, and the relative activity of luciferase to b-galactosidase activity is presented Early cell cycle control by p300 C-W Lee et al 2698 minimal thymidine kinase promoter together with the Certain control treatments were performed to SV40 enhancer driving the luciferase gene, and in SAOS2 establish the speci®city of E2F-1 on p53. For possessed minimal activity in the absence of p53 but example, the eects on p53 transcriptional activity responded favourably to exogenous wild-type p53 (for were dependent upon exogenous p53 since without p53 example Figure 1a, compare tracks 1 and 2); similar data no changes in activity on the pTG13-GL reporter in were obtained from transfection studies in U2OS cells. the presence of E2F-1 or pRb were apparent (data not To establish the in¯uence of the E2F pathway on shown). The level of exogenous p53 protein was not p53, we introduced increasing levels of E2F-1 or pRb aected by co-expression of E2F-1 (Figure 1b, compare and thereafter assayed p53 activity. Increased levels of tracks 1 ± 4 with 5 ± 8) and, further, each transfection E2F-1 caused a concomitant decline in the transcrip- treatment was internally controlled to rule out general tional activity of p53 (Figure 1a, compare tracks 2 , 3, eects of p53, E2F-1 or pRb on the transcription 4 and 5) whereas, in contrast, increased levels of pRb apparatus (see Materials and methods). Having stimulated p53-dependent transcription (Figure 1a, established the speci®city of the E2F-1 eect upon compare tracks 2 ± 6, 7 and 8). Further support for p53-dependent transcription, we went on to investigate this idea was provided by co-expressing E2F-1 in the the mechanism involved. presence of pRb. In these conditions, pRb antagonised the eects of E2F-1 so that a reduction in p53- The regulation of p53 activity requires the E2F-1 and dependent transcription was no longer apparent p53 activation domains (Figure 1a, compare tracks 3 ± 9, 10 and 11). At the same time, the stimulation of p53 activity by pRb was An outcome of the interaction between pRb and E2F-1 less when co-expressed with E2F-1 (Figure 1a, compare is the inactivation of the C-terminal transcription tracks 6 ± 8 to 9, 10 ± 11), and a titration of E2F-1 activating domain, which is integrated with the levels in the presence of pRb provided additional domain recognised by pRb (Flemington et al., 1993; evidence for this view, since upon increased levels of Helin et al., 1993; Kaelin et al., 1992). Since the E2F-1 there was a proportional decline in the regulation of p53 activity by E2F-1 was overcome by transcriptional activity of p53 (Figure 1a, compare pRb (Figure 1a), we reasoned that the E2F-1 activation tracks 9 ± 11 to 12 ± 17). Thus, the transcriptional domain may be in¯uential in regulating p53 activity. activity of p53 is in¯uenced by the levels of E2F-1 To assess this possibility, a mutant E2F-1 protein and, moreover, the presence of pRb overcomes this which lacks the C-terminal activation domain, due to eect, suggesting that the p53 and pRb/E2F pathways truncation up to amino acid residue 413 (E2F-1D413), of growth control are functionally integrated. was studied. As expected E2F-1D413 had greatly

a b

Figure 3 p300 rescues E2F-1 inactivated p53 and enhances the transcriptional activity of p53 and E2F-1. (a) The p53 reporter pTG13-GL was introduced into SAOS-2 cells together with expression vectors encoding p53 (1 mg; tracks 2 ± 7) in the presence of E2F-1 (2 mg; tracks 3, 4 and 5) with p300 (6 and 12 mg in tracks 4 and 5, and 6 and 7 respectively). The data presented re¯ect averages from two sets of values. Throughout pCMV-bgal was included within each transfection as an internal control, and the relative activity of luciferase to b-galactosidase presented. (b) The E2F reporter pDHFR-luc was introduced into SAOS-2 cells together with expression vectors encoding E2F-1 (0.2 mg; tracks 2, 3 and 4) together with p300 (2 and 10 mg in tracks 3 and 4, and 5 and 6 respectively). Values were calculated as described in (a) Early cell cycle control by p300 C-W Lee et al 2699 reduced ability to stimulate the transcription of the p300 physically associates with p53 and E2F-1 E2F-responsive DHFR promoter in the absence or presence of DP-1 (Figure 2a, compare tracks 1 ± 5). Having established a functional relationship between When assessed for its eect upon p53 activity, and in p53 and p300, and E2F-1 and p300, we wished to contrast to wild-type E2F-1, E2F-1D413 failed to cause determine if this was due to the direct physical a reduction but rather increased p53-dependent interaction between each pair of proteins. To test this transcription (Figure 2b, compare tracks 3 and 4 with possibility we used a p300 derivative, pG4-p300611 ± 2284, 5 and 6), the reasons for which remain to be explored. in which the region between amino acid residue 611 and Thus, the integrity of the transcriptional activation 2284 is fused to the Gal4 DNA binding domain. We domain is necessary for the reduction in p53 activity by chose pG4-p300611 ± 2284 for these studies because it is E2F-1. eciently expressed in a variety of mammalian cells Since the regulation of p53 required the E2F-1 (data not shown). Extracts were prepared from U2OS transcriptional activation domain, we reasoned that cells expressing pG4-p300611 ± 2284 and the interaction the activation domain in p53 may be involved. between p53 and pG4-p300611 ± 2284 assessed using a Therefore, we tested whether a fusion protein binding assay where in vitro translated and radiola- containing the N-terminal transcription activation belled p53 or the control luciferase was added to the domain of p53 was sensitive to the eects of E2F-1. U2OS cell extract and the association with p300 In a similar fashion to wild-type p53, the activity of determined by immunoprecipitation using an anti- Gal4-p53 was reduced by E2F-1 (Figure 2c, compare Gal4 antiserum (Figure 4a, compare tracks 5 and 6). tracks 2 and 3). An interaction between p53 and pG4-p300611 ± 2284 was It was important to determine the speci®city of the detected but not between luciferase (Figure 4a, compare eect of E2F-1 on p53. To this end, we studied tracks 3 and 4) suggesting a speci®c association between whether E2F-1 caused a reduction in the activity of p53 and p300. The detection of p53 in the immunopre- other transcription activation domains. A variety of cipitate was dependent upon immunoprecipitation since activation domains were assessed as Gal4 fusion in the absence of the anti-Gal4 antiserum p53 was no proteins, of which the p53 activation domain was by longer apparent (Figure 4a, compare tracks 2 and 4). far the most sensitive (data not shown). Further, the To con®rm that p300 and p53 can physically inactivation of p53 was not caused through general associate in vivo, we co-expressed G4-p300611 ± 2284 and eects, such as protein stability, because a hybrid p53 in U2OS cells, immunoprecipitated with anti-Gal4 protein in which the VP16 activation domain was followed by immunoblotting with anti-p53 monoclonal fused to the wild-type p53 protein became far less antibody 421. Again, p53 was speci®cally detected in sensitive to the eects of E2F-1 (Figure 2c, compare the p300 immunocomplex (Figure 4b, compare tracks tracks 5 and 6 with 8 and 9). 2, 3 and 4) establishing that p300 and p53 can physically associate in mammalian cells. Further evidence was obtained by taking a p300 relieves the inactivation of p53 by E2F-1 two-hybrid approach in U2OS cells in which The results so far raise the possibility that competition G4-p300611 ± 2284, which is transcriptionally inactive in for a shared transcription target is responsible for the these conditions, was used as the `bait' (Figure 4c, eect of E2F-1 on p53-dependent transcription. The compare tracks 2 ± 7). For the `prey', we fused wild- potential identity of such a molecule was addressed by type p53 to VP16 which, when expressed with the p300 determining which proteins, when co-expressed with hybrid, caused a signi®cant increase in the transcrip- E2F-1 and p53, relieved the repression of p53 activity. tional activity of the reporter pG5-luc (Figure 4c, Expression of the co-activator p300 (Shikama et al., compare tracks 7 ± 9). Transcription was dependent 1997), a molecule previously implicated in E2F-1 upon the hybrid p53-VP16 protein since neither p53 mediated transcriptional activation (Trouche et al., nor VP16 caused a similar increase (Figure 4c, compare 1996), eectively relieved the repression of p53 (Figure tracks 8, 9 and 10). Importantly, the eect of p53-VP16 3a, compare tracks 3, 4 and 5) and enhanced in p53- required p300 protein sequence as there was no dependent transcription in the absence of E2F-1 apparent eect on the Gal4 DNA binding domain (Figure 3a, compare tracks 3, 4, 5, 6 and 7). These alone (Figure 4c, compare tracks 2 ± 5). Similar data data argue that p300 functions as a transcriptional co- from the two-hybrid assay were observed in SAOS2 activator for p53 and, further, suggest that a limiting cells (data not shown). level of p300 is in part responsible for the inactivating An immunoprecipitation strategy was used to eect of E2F-1 on p53. determine if E2F-1 and DP-1 associate with p300. To con®rm that p300 can provide a co-activator After immunoprecipitating G4-p300611 ± 2284, immuno- function to E2F-1, we assayed the eect of p300 and blotting was performed with antisera against either E2F-1 on the DHFR promoter. As expected, co- E2F-1 or the haemagglutin (HA) epitope to detect the expression of p300 enhanced transcriptional activity in DP-1 protein. Both E2F-1 and DP-1 could associate an E2F-1-dependent fashion (Figure 3b, compare with p300 since either protein co-immunoprecipitated tracks 2 ± 6). Thus, the combined conclusion from with G4-p300611 ± 2284 (Figure 5a, compare tracks 1 ± 3 these studies suggests that p300 is a shared co-activator and 5 ± 7). utilized by p53 and E2F-1 and therefore a likely target We con®rmed the interaction by taking the two- involved in mediating the eect on p53 by E2F-1. That hybrid approach in which G4-p300611 ± 2284 was assessed co-expression of pRb relieved the eect of E2F-1 on for an interaction with GAD-E2F-1, a construct in p53 (Figure 1a) is consistent with this result since pRb which the Gal4 activation domain was fused to E2F-1. inactivates the transcriptional activity of E2F-1 There was a signi®cant increase in activity when pG4- (Flemington et al., 1993; Helin et al., 1993). p300611 ± 2284 and GAD-E2F-1 were expressed together Early cell cycle control by p300 C-W Lee et al 2700 (Figure 5b, compare tracks 6 and 8). The enhanced To con®rm the interpretation from the earlier transcription was speci®c for the p300 (Figure 5b, experiments, namely, that the activation domain of compare tracks 4 and 8) and E2F-1 (Figure 5b, p53 and E2F-1 is a target for p300, we studied in the compare tracks 8 and 9) sequence in each hybrid two-hybrid assay the interaction between a mutant p53 protein. The overall conclusion from the biochemical protein which lacked the N-terminal activation and two-hybrid based assays strongly suggests that p53 domain, p53D159-VP16, and a mutant E2F-1 protein and E2F-1 can form a stable physical complex with which lacked the C-terminal activation domain, E2F- p300. 1D413-VP16. Neither of the mutant hybrid proteins

a b G4–p300 G4–p300 –– + + : α Gal4 p53- lucp53 luc p53 luc p53 : IVT M– p53 vp16

92 —

68 — — p53 - VP16

46 —

12 34 5 6 1234

Figure 4 p300 physically associates with p53. (a) The expression vector encoding G4-p300611 ± 2284 was introduced into U2OS cells and after 48 h harvested as described. Either in vitro translated 35S-methionine-labelled luciferase (tracks 1 and 3) or p53 (tracks 2 and 4) was added to the extract, thereafter immunoprecipitated with the anti-Gal4 antiserum and further resolved by gel electrophoresis. Tracks 5 and 6 show the input in vitro translates for luciferase (track 5) or p53 (track 6). (b) The expression vector encoding G4-p300611 ± 2284 (tracks 2, 3 and 4) together with either p53 (track 3) or p53-VP16 (track 4) was introduced into U2OS cells and after 48 h immunoprecipitation performed with the anti-Gal4 antiserum (tracks 2, 3 and 4) followed by gel electrophoresis. Immunoblotting was performed with the p53 monoclonal antibody 421. Note that the common band present across tracks 2, 3 and 4 results from non-speci®c activity for the immunoglobulin used in the ®rst immunoprecipitation which obscures the wild-type p53 polypeptide in track 3 (data not shown). Track 1 shows the standard molecular weight polypeptides. The p53-VP16 polypeptide is indicated. (c) Two-hybrid assay in U2OS cells or SAOS2 cells (data not shown) where the Gal4 reporter pG5-luc was introduced together with expression vectors pG4 encoding either the Gal4 DNA binding domain (1 mg; tracks 2, 3, 4 and 5) or pG4-p300611 ± 2284 (1 mg; tracks 7, 8, 9 and 10) together with either wild-type p53 (5 mg; tracks 3 and 8), p53-VP16 (5 mg; tracks 4 and 9) or VP16 (5 mg; tracks 5 and 10). The data presented re¯ect averages from two sets of values. Throughout pCMV-bgal was included within each transfection as an internal control, and the relative activity of luciferase to b-galactosidase presented Early cell cycle control by p300 C-W Lee et al 2701

a IP α G4 / IB α E2F-1 IP α G4 / IB α HA

G4-p300 G4-p300

+ –– + –– : E2F-1 ––+ ––+ : HA-DP-1

— 92 — 92

— 68 — 68 E2F-1 — DP-1 — — 46 — 46

1234 5678

b c

Figure 5 p300 physically associates with E2F-1 and DP-1. (a) The expression vector encoding G4-p300611 ± 2284 (20 mg; tracks 1, 2, 3, 5, 6 and 7) together with either E2F-1 (20 mg; tracks 1 and 5) or HA-DP-1 (20 mg; tracks 2 and 6) were introduced into U2OS cells and after 48 h immunoprecipitation performed with the anti-Gal4 antiserum followed by gel electrophoresis. Immunoblotting was performed with either anti-E2F-1 (tracks 1 ± 4) or anti-HA (tracks 5 ± 8) as described in Materials and methods. Note that the common band present across tracks 1 ± 3, and 5 ± 7, results for non speci®c activity for the immunoglobulin used in the immunoprecipitation. (b) Two-hybrid assay in U2OS cells where the Gal4 reporter pG5-luc was introduced together with expression vectors pG4 (1 mg; tracks 2, 3, 4 and 5) or pG4-p300611 ± 2284 (1 mg; tracks 6, 7, 8 and 9) together with either wild-type E2F-1 (5 mg; tracks 3 and 7), pGAD-E2F-1 (5 mg; tracks 4 and 8) or GAD (5 mg; tracks 5 and 9). The data presented re¯ect averages from two sets of values. Throughout pCMV-bgal was included within each transfection as an internal control, and the relative activity of luciferase to b-galactosidase presented. (c) Two-hybrid assay in U2OS cells or SAOS2 cells (data not shown) where the reporter pG5-luc was introduced together with expression vectors for pG4-p300611 ± 2284 (1 mg; tracks 2 ± 6), p53-VP16 (5 mg; track 3), p53D159-VP16 (5 mg; track 4), E2F-1-VP16 (5 mg; track 5) or E2F-1D413-VP16 (5 mg; track 6). The data presented re¯ect averages from two sets of values, and throughout pCMV-bgal was included as an internal control Early cell cycle control by p300 C-W Lee et al 2702 were capable of interacting with pG4-p300611 ± 2284 transcriptional activity of the pG5-reporter (Figure 7a, compared to the wild-type hybrid protein (Figure 5c, compare tracks 3, 4, 5 and 6). The eect of p21Waf1/Cip1 compare tracks 3 and 4, and 5 and 6); similar two- was dependent upon the co-expression of p300 and p53 hybrid assay results were obtained from SAOS2 cells because there was only a marginal eect on G4- (data not shown). Furthermore, the co-expression of p300611 ± 2284 by p21Waf1/Cip1 (Figure 7a, compare track 2 p300 with p53 or E2F-1 failed to in¯uence the levels of with 7, 8 and 9). These data indicate that co-expression either protein (Figure 6a and b), consistant with the of p21Waf1/Cip1 enhances the activity of the p53/p300 idea that p300 enhances the transcriptional activity of interaction. p53 and E2F-1. Thus, we conclude that the interaction The conclusion that p21Waf1/Cip1 enhances the interac- of p300 with either p53 or E2F-1 is mediated through tion between p53 and p300 rests on the two-hybrid binding to the activation domain. assay. Therefore, it was necessary to demonstrate the eect of p21Waf1/Cip1 on the activity of transcriptionally active wild-type p53. Increasing levels of p21Waf1/Cip1 p21 enhances p53-dependent transcription by regulating were introduced together with wild-type p53 and p53- the p53/p300 interaction dependent transcription assayed. In the presence of The p21Waf1/Cip1 gene is transcriptionally activated by p21Waf1/Cip1 there was a marked stimulation of p53 p53, containing functional p53 binding sites within its transcriptional activity (Figure 7b, compare tracks 2, 3, transcriptional control region (El-Deiry et al., 1993; 4 and 5), an eect dependent upon the presence of Macleod et al., 1995). We were interested to test the wild-type p53 (Figure 7b, compare tracks 1 with 6, 7 idea that p21Waf1/Cip1 may act in an autoregulatory and 8). We conclude that the expression of p21Waf1/Cip1 fashion upon the p53/p300 interaction and thereby causes an increase in the transcriptional activity of stimulate transcription. To assess this possibility, we wild-type p53, and that a likely mechanism based on used the two-hybrid assay between G4-p300611 ± 2284 and the earlier results is through potentiating the activity of p53-VP16 (Figure 7a). Under conditions in which there the p53/p300 co-activator complex. was a clear interaction between G4-p300611 ± 2284 and p53-VP16 (Figure 7a, compare tracks 2 and 3), the p300 is required for the activation of Waf1/Cip1 levels of p21Waf1/Cip1 were titrated and eects on transcription measured. As the level of p21Waf1/Cip1 To investigate the role of p300 in the regulation of increased, there was a concomitant increase in the cellular genes, we studied the p53-responsive promoter taken from Waf1/Cip1 (El-Deiry et al., 1993). The expression of the adenovirus E1a protein caused a signi®cant reduction in transcriptional activity (Figure a p300 8a, compare tracks 1 and 2), consistent with the p53 – ++++ ++ ability of E1a to sequester p300 (Arany et al., 1995). In contrast, a mutant E1a protein that lacks the N- terminal p300 binding domain failed to reduce the transcriptional activity of Waf1/Cip1 (Figure 8a, 68 — compare tracks 1 and 3), consistent with a role for — p53 p300 in regulating the transcription of Waf1/Cip1. 45 — Conversely, in these experimental conditions the promoter taken from Bax, another p53 responsive gene (Miyashita and Reed, 1995), was only margin- M 1234567 ally aected by E1a (Figure 8a, compare tracks 4 b and 5), suggesting that the role of p300 in the transcription of Bax was less compared to its – p300 signi®cant role in regulating the transcriptional activity of Waf1/Cip1. To identify the region in p300 that physically 68 — E2F–1 interacts with p53, a panel of hybrid proteins were prepared in which dierent domains of p300 were fused to the Gal4 DNA binding domain and co-expressed with the hybrid protein p53-VP16 (Figure 8b). The p53 45 — interaction domain mapped to the C-terminal region of p300, speci®cally within the sequence from residue M 123456 1572 ± 1903, since co-expression of pG4-p3001572 ± 1903 Figure 6 p300 does not aect p53 or E2F-1 protein levels. (a) with p53-VP16 caused a six-fold induction of The expression vector encoding p53 (2 mg) was introduced into transcription (Figure 8b). A similar level of induction SAOS2 cells, (tracks 2, 3, 4, 5, 6 and 7) together with increasing was apparent with pG4-p3001572 ± 2284 but not for pG4- levels of the p300 vector (1, 3, 6, 12 and 18 mg) as indicated. p300611 ± 1572 or pG4-p3001302 ± 1572 (Figure 8b). Thus, p300 Extracts were prepared and immunoblotted with the 421 monoclonal antibody. Track 1 shows untransfected SAOS2 contains a C-terminal domain which is dedicated to a cells, and M the standard molecular weight markers. (b) The functional interaction with p53. Signi®cantly, this expression vector encoding E2F-1 (2 mg) was introduced into region in p300 is known to bind to the E1a protein U2OS cells together with increasing levels of the p300 vector (1, 3, (Eckner et al., 1996), suggesting that a competitive 6, 9 and 12 mg) as indicated. Extracts were prepared and mechanisms may account for the inactivation of p53- immunoblotted with the KH95 monoclonal antibody, and M shows the standard molecular weight markers. A tenfold dependent transcription by E1a through the sequestra- induction of E2F-1 was apparent in transfected cells tion of p300. Early cell cycle control by p300 C-W Lee et al 2703

Figure 7 p21Waf1/Cip1 enhances the activity of the p53/p300 complex and p53-dependent transcription. (a) Two-hybrid assay in SAOS-2 cells where the Gal4 reporter pG5-luc was introduced together with pG4-p300611 ± 2284 (1 mg; tracks 2 ± 9) and p53-VP16 (5 mg; tracks 3 ± 6) in the presence and increasing levels of p21Waf1/Cip1 (1 mg in tracks 4 and 7, 5 mg in tracks 5 and 8 and 10 mgin tracks 6 and 9). The data presented re¯ect averages from two sets of values. Throughout pCMV-bgal was included within each transfection as the internal control, and the relative activity of luciferase to b-galactosidase presented. (b) The p53 reporter pTG13- GL was introduced into SAOS-2 cells together with expression vectors encoding p53 (1 mg; tracks 2 ± 5) in the presence of increasing levels of p21Waf1/Cip1 (1 mg in tracks 3 and 6, 5 mg in tracks 4 and 7 and 10 mg in tracks 5 and 8). Values were calculated as described in (a)

We conclude that p300 can facilitate cell cycle arrest by Physiological consequences of p300 interacting with p53 p53, and that a possible mechanism involves increased and E2F transcription of p53 target genes, such as the p21Waf1/Cip1 The induction of transcription of the gene encoding gene, as a result of p300 enhancing the transcriptional p21Waf1/Cip1 by p53 correlates with cell cycle arrest by activity of p53. p53 (Chen et al., 1996; El-Deiry et al., 1993; Macleod Next, we explored the role of p300 in apoptosis. For et al., 1995) and the expression of p21Waf1/Cip1 is this, we used SAOS2 or U2OS cells that had been sucient to prevent cell cycle progression (Dulic et cultured in serum starvation conditions, condition al., 1994; Waldman et al., 1995). We reasoned that the favour that apoptosis (White, 1996). The data ability of p300 to interact with and enhance the presented were derived from SAOS2 cells, although transcriptional activity of p53 may facilitate cell cycle very similar results were obtained from U2OS cells arrest. (data not shown). Three assays were used to measure To test this idea we introduced p53 into asynchro- the level of apoptotic cells, namely TUNEL, ¯ow nous cultures of SAOS2 cells and monitored the cytometry (cells with sub-genomic levels of DNA) and proportion of transfected cells in G1, S and G2/M, the cell death detection assay (see Materials and both in the presence and absence of co-expressed p300 methods). Data derived from TUNEL assays are (Figure 9). Transfected cells were identi®ed by the shown (Figure 9b and summarized in Table 1), similar expression of the cell surface protein CD20 and conclusions being made from the other two assays. subsequent staining with an anti-CD20 monoclonal In SAOS2 cells, the introduction of wild-type p53 antibody, and the percentage of cells in each phase of caused a signi®cant level of apoptosis (Figure 9b and the cell cycle measured after treatment with propidium Table 1) consistant with previous reports (Chen et al., iodide. 1996). When p300 was assessed in the same cells, We performed the ¯ow cytometry experiments in little dierence was apparent from the untreated cells, SAOS2 where the introduction of p53 caused an although the introduction of p53 together with p300 accummulation of cells in G1, usually resulting in an sign®cantly reduced the proportion of apoptosing cells increase in G1 size of about 20% (Figure 9a); similar (Table 1). Thus, the ability of p53 to favour results were obtained in U2OS and 3T3 cells (data not apoptosis or cell cycle arrest is in¯uenced by the shown). Under these conditions we expressed p300 level of p300. either alone or together with p53. When p300 was co- We went on to study the eect of E2F-1 in SAOS2 expressed with p53 there was a marked increase in the cells. Interestingly, E2F-1 when expressed alone caused proportion of cells in G1, compared to p53 alone signi®cant levels of apoptosis (Figure 9b and Table 1) (Figure 9a); p300 had little eect in the absence of p53. indicating that the apoptotic activity of E2F-1 is not Early cell cycle control by p300 C-W Lee et al 2704 absolutely dependent upon the presence of wild-type Overall, these results suggest that p300 plays a critical p53. In these conditions, the apoptotic activity of E2F- role in regulating the early cell cycle, speci®cally in 1 was enhanced by p300 (Figure 9b and Table 1), in in¯uencing the contrasting outcomes of cell cycle arrest contrast to the eect of p300 on p53 where it increased and apoptosis. the eciency of p53-dependent cell cycle arrest. Further, a clear stimulation in the proportion of apoptotic cells was apparent when E2F-1 and p53 Discussion were expressed together (Figure 9b and Table 1), a result which corroborates earlier reports (Qin et al., p300 in the control of E2F and p53-dependent 1994; Wu and Levine, 1994). That limiting levels of transcription p300 may inpart be responsible for this enhanced apoptosis was suggested from introducing p300 into A considerable body of research evidence has suggested cells undergoing E2F-1/p53-dependent apoptosis, upon that the pathways regulated by E2F and p53 are which the level of apoptosis was signi®cantly reduced integrated in a way that allows p53 to sense aberrant (Figure 9b and Table 1). Thus, as observed earlier for control of E2F activity. A particularly clear example of the eect of p300 on p53-dependent apoptosis, p300 this phenomenon relates to the induction of p53- could partially rescue E2F-1/p53-dependent apoptosis. dependent apoptosis in conditions where E2F-1 is

Figure 8 (a) p300 is required for the activation of Waf1/Cip1. The promoter from either Waf1/Cip1 or Bax fused to luciferase, referred to as pWWP (tracks 1, 2 and 3) and pBax (tracks 4, 5 and 6) respectively, was introduced into U2OS cells together with expression vectors for either wild-type E1a (0.5 mg; tracks 2 and 5) or E1aD2-36 (0.5 mg; tracks 3 and 6). The data presented re¯ect averages from two sets of values, with pCMV-bgal included throughout as the internal control. (b) A C-terminal domain in p300 dedicated to an interaction with p53. Two-hybrid assay performed in U2OS cells where the Gal4 reporter pG5-luc was introduced together with expression vectors encoding the indicated Gal4-p300 hybrids (1 mg) with or without p53-VP16 (4 mg). The data presented re¯ect averages from two sets of values normalized to the activity of the internal control pCMV-bgal Early cell cycle control by p300 C-W Lee et al 2705 expressed at high levels in tissue culture cells (Qin et dependent upon p53 (Morgenbesser et al., 1994), al., 1994; Wu and Levine, 1994). Other studies, support this general idea. including those where the expression of viral oncopro- In this study, we have identi®ed a mechanism which teins which inactivate pRb occurs in de®ned cells, such can explain the functional interplay between E2F and as lens ®bre cells (Pan and Greip, 1994), and the p53 and which may, in part, in¯uence the physiological targetted disruption of Rb where the apoptosis is outcome in conditions of aberrant levels of E2F. A

b p300 – +

–

p53

E2F-1

p53/E2F-1

Figure 9 Eect of p300 on cell cycle arrest and apoptosis. (a) Flow cytometry was performed on asynchronous cultures of SAOS2 cells transfected with the CD20 expression vector (10 mg) together with p53 alone or p53 and p300 together. 15 mg of each expression vector was introduced, the total amount of DNA for each treatment being equivalent and made up with empty vector. Transfected cells were identi®ed by staining with anti-CD20 and the cell cycle pro®le was assessed by staining with propidium iodide. The percentage change compared to the control treatment (CD20 expression vector alone) is shown. The data show a representative example from at least four dierent data sets. (b) The eect of p300 on p53-dependent and independent apoptosis was assessed in both SAOS2 and U2OS cells transfected in conditions of serum starvation (see Materials and methods) as indicated with 4 mg of p53 and/or E2F-1 expression vector in the absence or presence of the p300 expression vector, the total amount being made up with empty vector. Results for SAOS2 cells are shown when assayed by TUNEL assay (left hand side of each treatment) and, for comparison, a DAPI stain of the same ®eld of cells. The level of apoptosis for each treatment is summarized in Table 1 Early cell cycle control by p300 C-W Lee et al 2706 Table 1 Eect of p300 on apoptosis mutated and possess the properties of tumour suppressor genes (Muraoka et al., 1996). No. of cells Percent No. of TUNEL examined apoptotic cells Treatment positive cells (DAPI stain) (%) Positive autoregulation of p53-dependent transcription by ± 9 564 1.6 p21Waf1/Cip1 p300 12 540 2.2 p53 69 389 17.1 Cell cycle arrest mediated by p53 is thought to rely on p53/p300 45 381 11.8 the ability of p53 to directly activate the transcription E2F-1 34 452 7.5 Waf1/Cip1 E2F-1/p300 63 321 19.6 of target genes, such as p21 , which functions as p53/E2F-1 104 356 29.2 an inhibitor of cyclin-dependent kinases (Chen et al., p53/E2F-1/p300 72 305 23.6 1996; El-Deiry et al., 1993; Harper et al., 1993; A summary of the level of apoptosis caused by each treatment in Macleod et al., 1995) and thus likely occurs through SAOS2 cells. The percentage of apoptotic cells was determined by the inactivation of G1 cdk activity and the main- counting the number of TUNEL positive cells compared to the total tenance of pRb and related proteins in a hypopho- cells examined from DAPI staining. The numbers represent the average from several independent experiments. The expression of sphorylated state. Our results de®ne a further level of transfected plasmids was con®rmed and the transfection eciency control in the regulation of p21Waf1/Cip1, exerted through established by indirect immuno¯uorescence (data not shown) a positive autoregulatory feedback mechanism whereby p21Waf1/Cip1 enhances the transcriptional activity of the p53/p300 co-activator complex. Although the precise mechanism of action remains to be determined, this mechanism was suggested when we observed that co- eect of p21Waf1/Cip1 suggests that the activity of the p53/ expression of E2F-1 caused a concomitant reduction in p300 co-activator complex is in¯uenced in a negative p53-dependent transcription, an eect which was fashion by cdk phosphorylation. The recent demon- overridden by co-expression of pRb. This, combined stration that p300 associates with cyclin E/cdk2 with the evidence that the trans activation domains of (Perkins et al., 1997) is consistent with this idea. E2F-1 and p53 were involved, raised the possibility A recurring theme in the control of the cell cycle is that competition for a rate limiting transcription target that of positive autoregulation, which is believed to may be responsible for the down-regulation of p53. provide a mechanism for enabling the rapid accummu- The p300 protein, which possesses the properties of a lation of regulatory proteins involved in mediating cell transcriptional co-activator and is involved in regulat- cycle transitions. Numerous examples are known, such ing the activity of a variety of sequence speci®c as in Saccharomyces cerevisiae where a positive transcription factors (Shikama et al., 1997), overcame autoregulatory feedback control on transcription by the down-regulation of p53 and, moreover, enhanced the G1 cyclins favours the rapid induction of cdk the activity of both E2F-1 and p53-dependent activity and thus progress into S phase (Johnston, transcription, an eect dependent on a physical 1992). Similarly, the mammalian E2F-1 gene is association between the trans activation domain of autoregulated by E2F activity (Johnson et al., 1994) each protein and p300. Thus, our data strongly suggest and the product of the cyclin E gene, which is that the reduced level of p53 transcriptional activity in regulated by E2F, may positively regulate transcrip- the presence of E2F-1 is exerted through each trans tion of its own gene by antagonizing the repressive activation domain competing for the p300 co-activator. eects of pRb and related proteins (Botz et al., 1996). That co-expression of pRb neutralized the eect of The autoregulation of p53/p300 activity by p21Waf1/Cip1 E2F-1 on p53 makes good sense in the light of the provides another interesting example of such a regulation of the E2F-1 activation domain by pRb mechanism which likely assists the rapid accummula- (Flemington et al., 1993; Helin et al., 1993; Kaelin et tion of p21Waf1/Cip1, thus augmenting cell cycle arrest. al., 1992). Indeed, it is possible that other studies where Since it is widely believed that p53 responds to an E2F-dependent reduction in p53 transcriptional environmental stress, a rapid cell cycle arrest clearly activity has been reported (O'Connor et al., 1995) is a desirable response. could involve a similar mechanism as the competition for p300 proposed in the present study. p300 in the control of apoptosis It is interesting that p300 functions as a co-activator for p53 and E2F-1. The p300 family of proteins, which Our conclusions relating to the biological signi®cance in mammalian cells includes one other molecularly of the interaction between p53 and p300 suggest that it characterized member called CBP, was de®ned by is critical in determining the physiological consequence studies on the mechanisms through which the oncogene of p53 induction. Thus, p300 can enhance the ability of products of DNA tumour viruses subvert normal p53 to cause G1 arrest and, moreover, reduce the cellular growth control (Arany et al., 1995). Within proportion of cells undergoing p53-dependent apopto- the adenovirus E1a protein, p300 interacts with a sis. These observations are consistent with a role for domain which is required for E1a to exert a range of p300 as a p53 co-activator in the transcription of biological eects on cells, such as, the induction of p21Waf1/Cip1, and in turn suggest that p53 induction and DNA synthesis and apoptosis (Moran, 1993). Recent limiting levels of p300 may be a signal which causes evidence indicates that the p300 family is responsible cells to enter apoptosis. It is known that transcription- for regulating the activity of a range of DNA sequence- ally inactive p53 proteins can induce apoptosis (Caelles speci®c transcription factors, particularly those which et al., 1994; Chen et al., 1996; Haupt et al., 1995b) function in dierentiation (Shikama et al., 1997). whereas transcriptional activity correlates with cell Further reports have connected p300 genes with cycle arrest (Crook et al., 1994). Further evidence diseases such as cancer where they can become suggests that the level of p53 may dictate the cellular Early cell cycle control by p300 C-W Lee et al 2707 response, with low levels of p53 causing cell cycle arrest Although apoptosis is enhanced when p53 and E2F- and high levels of apoptosis (Chen et al., 1996). Given 1 are expressed together, nevertheless E2F-1 alone was the data presented in this study, it is possible at high capable of inducing apoptosis in p537/7 tumour cells, levels of p53 that p300 becomes limiting which, suggesting that p53-dependent and independent me- suggested by the data presented here, provides an chanisms contribute towards E2F-1 induced apoptosis apoptotic signal. Further, the induction of p53- (this study and Hsieh et al., 1997; Phillips et al., 1997). dependent apoptosis by adenovirus E1a requires the In p537/7 cells, the co-expression of E2F-1 and p300 integrity of the N-terminal p300-binding domain enhanced apoptosis, in contrast to the eect of p300 in (Querido et al., 1997) and thus is likely to be mediated cells expressing p53 and E2F-1 when there was a by a transcriptionally compromised form of p53. signi®cant decrease in the proportion of apoptotic cells. Of course, some studies have suggested that Although there was not a complete rescue from transcription-dependent mechanisms are involved in apoptosis, this result is nevertheless consistent with p53-dependent apoptosis, and by no means is our the proposed model that limiting levels of p300 favour study incompatible with this view. For example, the Bax p53-dependent apoptosis. Indeed, we would have gene has been implicated as a direct target for p53 predicted that p300 could not completely rescue cells (Miyashita and Reed, 1995). However, although Bax7/7 from apoptosis through the enhanced activity of p53 thymocytes fail to respond to certain apoptotic stimuli, because, clearly, in SAOS2 cells E2F-1 can induce they are not de®cient in p53-dependent DNA damage apoptosis in the absence of p53. Moreover, it is induced apoptosis (Knudson et al., 1995). It is likely noteworthy that the partial rescue of apoptosis by therefore that p53 can in¯uence apoptosis through p300 is compatible with earlier reports indicating that dierent mechanisms. The model proposed in this study p53-mediated apoptosis can be overcome by pRb suggesting that apoptosis occurs in conditions of p53 (Haupt et al., 1995a) since, as explained, the presence induction with limiting levels of p300 may account for of p300 will directly favour the maintenance of one such mechanism. However, we note that other hypophosphorylated pRb. recent studies have suggested that p300 plays a positive role in regulating p53-dependent apoptosis (Avantag- A central role for p300 in early cell cycle control giati et al., 1997; Lill et al., 1997; Somasundaram and El-Deiry, 1997) and, although further experiments are That p300 functions as a co-activator involved with required to clarify these issues, it is possible that the E2F and p53-dependent transcription and can in¯uence dependence of p53 activity on p53 concentration (Ko the physiological consequence of p53 induction and Prives, 1996) is a contributory factor. suggests that p300 plays a central role in co-ordinating

ab c

Figure 10 Model for cell cycle control by p300. It is envisaged that during normal cell cycle progression (a) p53 remains in a latent state and that p300 functions as a co-activator for transcription factors such as E2F-1. Upon activation and cell cycle arrest by p53 (b), p300 functions as a co-activator for p53-dependent transcription, such as in the activation of the p21Waf1/Cip1 gene. The activity of the p53/p300 co-activator complex is enhanced by p21Waf1/Cip1 through a positive autoregulatory feedback loop whilst the activity of E2F declines due to the presence of hypophosphorylated pRb. For (c), in conditions which are inappropriate for cell cycle progression aberrantly high levels of E2F compete for p300, leaving p53 in a transcriptionally compromised state. It is envisaged that this form of p53, which may also arise upon increased levels of p53, together with E2F provides a signal for apoptosis Early cell cycle control by p300 C-W Lee et al 2708 early cell cycle events. The results presented in this and p53. Our data strongly support the idea that p300 study, together with the documented role of p53- functions as a negative regulator of cell cycle dependent transcription in cell cycle arrest, are progression and that p300 levels are instrumental in consistent with a model in which the productive in¯uencing whether cell cycle progression, G1 arrest or interaction between p53 and p300 and consequent apoptosis occurs. The recent identi®cation of tumour activation of target genes like p21Waf1/Cip1 is important in suppressor-like mutations in p300 genes in human controlling cell cycle arrest. However, we note that a tumour cells provides support for such ideas (Muraoka p21Waf1/Cip1 de®ciency does not completely abbrogate G1 et al., 1996). arrest due, for example, to irradiation (Brugarolas et al., 1995; Deng et al., 1995) whereas in p537/7 knock out animals G1 arrest is completely absent (Done- Materials and methods hower et al., 1992). Thus, p53 likely has other targets in addition to p21Waf1/Cip1 which confer cell cycle arrest. Transfections Relating our results to the model, we propose that in The following plasmids have been previously described; normal cell cycle conditions E2F utilises co-activator pCMV-pRb wt (Zamanian and La Thangue 1992), pCMV- p300 in the regulation of genes required for the G1/S p300 (Eckner et al., 1996), pCMV-E2F-1 (Kaelin et al., phase transition (Figure 10a), circumstances in which 1992), pCMV-p21Waf1/Cip1 (El-Deiry et al., 1993), pCMV- p53 remains in a latent and non-active state. When the bgal (Zamanian et al., 1992), and pGal4-p53 (Fields and induction of p53 causes cell cycle arrest, for example, Jang, 1990). To construct pTG13-GL, ®rst, the BamHI ± in response to genotoxic stress, we suggest that p53 BglII fragment of the HSV1-tk minimal promoter element activates transcription in concert with p300 as a co- (781 to +52) was inserted to the BglII site of pXP2 vector (Nordeen, 1988) resulting in pT81-luciferase. Then, the 13 activator (Figure 10b). Positive autoregulation by copies of the consensus p53-binding sites from pG13-CAT p21Waf1/Cip1 enhances the transcriptional activity of the (Kern et al., 1992) was subcloned into the HindIII and p53/p300 co-activator complex, thus facilitating the SmaI sites of pT81-luciferase. For the Gal4 reporter rapid induction of p21Waf1/Cip1, and thereafter cell cycle construct, the Gal4 DNA binding elements and E1b arrest. At the same time, the inhibition of cdk activity TATA minimal elements from pG5E1b-CAT (Fields and will retain pRb and related proteins in a hypopho- Jang, 1990) was excised by XhoIandBamHI digestion was sphorylated state and thus, by virtue of excluding the cloned into the pGL3 (Promega). pDHFR-luc has been E2F-1 activation domain, enhance the levels of p300 described previously (Sùrensen et al., 1996). pBax-luc and available for p53. In apoptotic conditions, for example, pWWP-luc were a gift from M Oren (Haupt et al., 1955b) caused by high levels of E2F-1 in conditions which are and B Vogelstein (El-Deiry et al., 1993), respectively. To construct pCMV-HA-DP1, three copies of HA1 inappropriate for cell cycle progression or increased epitope sequences from pGTEP-1 (a gift from B Futcher) levels of p53 we suggest that p300 becomes limiting was ampli®ed by PCR and cloned into the pCDNA3 thus facilitating an apoptotic signal. In these circum- (Invitrogen) with a consensus Kozak sequence introduced stances, and based on the studies presented here, E2F immediately 5' of the initiating ATG codon to create pCMV- and p53 are likely to compete for p300 leaving p53 in a HA1. The DP-1 cDNA sequence (Girling et al., 1993) in transcriptionally compromised or de®cient state frame was introduced into pCMV-HA to generate an (Figure 10c). However, we would like to emphasize expression plasmid encoding an amino-terminal HA-tagged that our data do not rule out the possibility that there full-length DP-1. are p53 target genes which need to be activated, For the mammalian two-hybrid assay, several plasmids perhaps through other co-activators, in order to were constructed as follows: the PCR fragment of VP16 activation domain, amino acid 423 ± 490, from Gal4-VP16 achieve apoptosis. With respect to this idea, we note (Fields and Jang, 1990) was subcloned in-frame to the that p53 has been shown to utilize other co-activator pCMV-HA to generate pCMV-VP16. Two p53 PCR molecules in transcriptional activation (Lu and Levine, fragments encoding amino acids 1 ± 393 and 159 ± 393 from 1995; Thut et al., 1995) and thus it is possible that the pCMV-p53, and two E2F-1 PCR fragments encoding amino regulation of p53 activity by p300 relates to a speci®c acids 1 ± 437 and 1 ± 413 from pCMV-E2F-1 were fused up- group of genes. stream of VP16 activation domain in pCMV-VP16 to It is noteworthy that the ability of E2F-1 to generate pCMV-p53/VP16, pCMV-p53D159/VP16, pCMV- in¯uence the physiological outcome of p53 activation E2F1/VP16, and pCMV-E2F1D413/VP16, respectively. may bear on the phenotype of the E2F-1 knockout pCMV-GAD was made by the insertion of HindIII and mice which suer thymic hyperplasia due to reduced EcoRI fragment encoding Gal4 activation domain amino acid 768 to 881, from pACT II (Clontech) into the pCDNA3. The levels of apoptosis during thymocyte ontogeny, NdeI and XbaI fragment from pCMV-E2F1 was fused in- together with an increased incidence of tumours in frame into pCMV-GAD to make the pCMV-GAD/E2F1. To older animals (Field et al., 1996; Yamasaki et al., construct pG4-p300 (611 ± 2284), the NdeI fragment from 1996). Based on our studies, one scenario which may pCMV-p300 was fused in-frame to the Gal4 DNA binding contribute to this phenotype in the absence of E2F-1 domain sequence of pG4m-polyII vector (Bandara et al., would be that a signal (provided by E2F-1) is lost 1993). pCMV-E2F-1D413 was made by internal deletion from which drives p53-dependent apoptosis, subverting the pCMV-E2F1. pCMV-p53 was made by the insertion of p53 normal apoptotic process in thymocytes and thus cDNA from php53 (Zakut-Houri et al., 1985) into the giving rise to hyperplasia. Similarly, the loss of an pCDNA3. To construct the Gal4-p300 derivatives, pG4- 611 ± 1257 1302 ± 1572 1572 ± 2284 1572 ± 1903 apoptotic signal will encourage tumour progression, in p300 , -p300 , -p300 , and -p300 were made by internal detection from pG4-p300611 ± 2284, respec- a similar fashion to the phenotype observed in the tively. Two E1A expression vectors, pCMV-E1A and pCMV- p537/7 mice (Donehower et al., 1992). E1AD2-36 were a gift from JR Nevins (Kraus et al., 1992). In conclusion, we have de®ned a central role for the For transfection (including ¯ow cytometry), U20S and p300 co-activator in co-ordinating the interplay SAOS2 cells were grown in Dulbecco's modi®ed Eagles between the pathways of control mediated by E2F medium (DMEM) supplemented with 10% fetal calf serum Early cell cycle control by p300 C-W Lee et al 2709 (FCS). Cells were plated out 24 h before transfection at (50% v/v) was added and the incubation continued for 26105 cells per 6 cm dish (for luciferase and b-gal assay) or another 1 h at 48C. The agarose beads were collected by 16106 cells per 10 cm dish (for immunoprecipitation and centrifugation for 30 s at 5 000 g, the supernatant removed ¯ow cytometric analysis). Transfections were performed by and the pellet washed three further times in TNN. SDS the calcium phosphate precipitation method. Whenever loading buer was added to the ®nal pellet, the sample required, pCDNA3 or pSG5 were used to maintain a denatured and loaded onto a 8~10% SDS-polyacrylamide constant amount of DNA in each sample. The protein gel. Western blotting was subsequently performed with expression and stability of each construct was veri®ed by either anti-mouse p53 (421, kindly supplied by Julian Western blotting (data not shown). Luciferase and b- Gannon), anti-mouse HA1 (12CA5, Boehringer Man- galactosidase assays were performed as described previously nheim), or anti-mouse E2F-1 (KH95, Santa-Cruz) mono- (Jooss et al., 1995). Each treatment was performed in clonal antibody. duplicate. In vitro transcription and translation was carried To assess the eect of E2F-1 on p53 stability (Figure 1b), out in a TNT T7/SP6 coupled reticulocyte lysate system 16 h after transfection the medium was changed and, after a (Promega). further 8 h, replaced with methionine-free medium. Cells were metabolically labelled with 35S-methionine for 1 h, washed three times, and chased for 0, 1.5, 3.0 or 4.5 h in Flow cytometry normal medium. Cells were harvested in TNN and For ¯ow cytometry analysis about 10 mg of an expression immunoprecipitated with the 421 monoclonal antibody. The vector for the cell surface protein CD20 was co-transfected eect of p300 on the level of p53 and E2F-1 (Figure 6) was into cells (grown in 10% FCS) together with 15 mgofthe assessed by performing the transfection as indicated and each expression vector for p53 or p300 (details given in monitoring the level of exogenous protein by immunoblotting Figure 8). Cells were washed and refed after 16 h and with either 421 or KH95. harvested 40 h later by washing in PBS and thereafter treating with cell dissociation medium (Sigma) for 15 min. Apoptosis assays Cells were washed in DMEM by centrifuging at 2000 r.p.m. and resuspended in a small volume of DMEM SAOS2 and U2OS cells were grown in serum starvation containing the anti-CD20 antibody leu16 (Becton Dick- conditions on 3 cm-diameter dishes with coverslips (for inson) coupled to ¯uorescein isothiocyanate (FITC). Cells TUNEL assay) or without (for cell death detection ELISA) were incubated upon ice for 45 min, further washed twice and transfected with 4 mg of p53, E2F-1, or p300 in PBS and then resuspended in 50% PBS in ethanol at expression plasmids. After a 16 h incubation with the 208C for 30 min. Cells were collected by centrifugation and precipitates, the cells were washed twice with DMEM and treated with RNase (125 U/ml) for 30 min, harvested by further incubated for 24 h in DMEM supplemented with centrifugation and suspended in propidium iodide (20 mg/ 0.2% foetal calf serum. ml) in PBS at 48C for 1 h. Flow cytometry was performed For TUNEL (TdT-mediated dUTP nick end labelling) on a Becton Dickinson ¯uorescence activated cell sorter. analysis, cells were washed and ®xed in 4% paraformalde- The intensity of propidium iodine staining was analysed in hyde for 15 min at room temperature, rinsed and cell populations that were positive for FITC staining to permeabilized in PBS cantaining 0.1% Triton X-100 and determine the cell cycle pro®le of the transfected 0.1% sodium citrate for 2 min on ice. Subseqently, cells population using the Consort 30 software. The data were incubated in a Ca2+ reaction buer containing presented show a representative example from multiple ¯uorescein-dUTP and dNTP, and terminal deoxynucleotidyl assays. transferase (Boehringer Mannheim) at 378C for 1 h in a humid chamber. And then, cells were further incubated in DAPI (2 mg/ml, Sigma) for DNA staining. Cover slips were Immunoprecipitation washed three times in PBS, mounted, and viewed under the U2OS cells were transfected with plasmids encoding Gal4- ¯uorescence microscope. p300, HA-DP1, E2F-1 or p53 (as indicated in ®gure legend) by the calcium phosphate procedure. After 48 h, cells were washed twice in PBS and harvested by scraping into cold TNN buer (50 mM Tris-HC1 pH 7.4, 120 mM Acknowledgements NaCl, 5 mM EDTA, 0.5% NP40, 50 mM NaF, 1 mM DTT, We thank our colleagues for helpful comments on the 1mM PMSF, 0.2 mM sodium orthovanedate, leupeptin manuscript and Marie Caldwell for help in its preparation. (0.5 mg/ml), protease inhibitor (0.5 mg/ml), trypsin inhibi- We thank R Bernards, P Jansen-DuÈ rr, R Eckner, J tor (1.0 mg/ml), aprotinin (0.5 mg/ml) and bestatin (40 mg/ Gannon, W Krek, DM Livingston, E Moran, R MuÈ ller, ml) and incubated on ice for 30 min. The cell extract was JR Nevins, M Oren and B Vogelstein, for plasmids and centrifuged for 10 min at 12 000 g and pre-cleared by reagents. This research was supported by the Medical incubating with protein G agarose for 1 h at 48C with Research Council. Chang-Woo Lee gratefully acknowl- constant rotation. The supernatant was harvested and 2 ml edges support from the British Council and ORS Scheme, of anti-mouse Gal4 monoclonal antibody (100 mg/ml, and Noriko Shikama was supported by an EMBO Fellow- Santa-Cruz) added for 1 h at 48C. Protein A agarose ship.

References

Arany Z, Newsome D, Oldread E, Livingston DM and Brugarolas J, Chandrasekaran C, Gordon JI, Beach D, Jacks Eckner R. (1995). Nature, 374, 81 ± 84. T and Hannon GJ. (1995). Nature, 377, 552 ± 557. Attardi LD, Lowe SW, Brugarolos J and Jacks T. (1996). Caelles C, Helmberg A and Karin M. (1994). Nature, 370, EMBO J., 15, 3693 ± 3701. 220 ± 223. Avantaggiati ML, Ogryzko V, Gardner K, Giordano A, Chen X, Ko LJ, Jayaraman L and Prives C. (1996). Genes & Levine AS and Kelly K. (1997). Cell, 89, 1175 ± 1184. Dev., 10, 2438 ± 2451. Botz J, Zerfass-Thome K, Spitkovsky D, Delius H, Vogt B, Clarke AR, Mazndag ER, Van Roon M, Van der Lugt NMT, Eilers M, Hatzigeorgiou A and Jansen-DuÈ rr P. (1996). Van der Valk M, Hooper MJ, Berns A and Te Riele H. Mol. Cell. Biol., 16, 3401 ± 3409. (1992). Nature, 359, 328 ± 330. Early cell cycle control by p300 C-W Lee et al 2710 Crook T, Marston NJ, Sara EA and Vousden KH. (1994). Lu H and Levine AJ. (1995). Proc. Natl. Acad. Sci., 92, Cell, 79, 817 ± 827. 5154 ± 5158. Debbas M and White E. (1993). Genes & Dev., 7, 546 ± 554. Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss Deng C, Zhang P, Harper JW, Elledge S and Leder P. (1995). M, Peters G and Bartek J. (1995). Nature, 375, 503 ± 506. Cell, 82, 675 ± 684. Macleod KF, Sherry N, Hannon G, Beach D, Tokino T, Donehower LA, Harvey M, Slagle BL, McArthur MJ, Kinzler K, Vogelstein B and Jacks T. (1995). Genes & Dev., Montgomery CA, Butel JS and Bradley A. (1992). 9, 935 ± 944. Nature, 356, 215 ± 221. Martin K, Trouche D, Hagemeier C, Sùrensen TS, La Dowdy SF, Hinds PW, Louie K, Reed SJ, Arnold A and Thangue NB and Kouzarides T. (1995). Nature, 375, 691 ± Weinberg RA. (1993). Cell, 73, 449 ± 511. 694. Dulic V, Kaufmann WK, Wilson SJ, Tlsty TD, Lees E, Miyashita T and Reed JC. (1995). Cell, 80, 293 ± 299. Harper JW, Elledge SJ and Reed SI. (1994). Cell, 76, Moran E. (1993). Curr. Opin. Genet. Dev., 3, 63 ± 70. 1013 ± 1023. Morgenbesser SD, Williams BO, Jacks T and DePinho RA. Eckner R, Yao T-P, Oldread E and Livingston DM. (1996). (1994). Nature, 371, 72 ± 74. Genes & Dev., 10, 2478 ± 2490. Muraoka M, Konishi M, Kikuchi-Yanoshita R, Tanaka K, El-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons Shitara N, Chong J-M, Iwama T and Miyaki M. (1996). R, Trent JM, Lin D, Mercer WE, Kinzler K and Oncogene, 12, 1565 ± 1569. Vogelstein B. (1993). Cell, 75, 817 ± 825. Nevins JR. (1992). Science, 258, 424 ± 429. Ewen ME, Sluss HK, Sheer CJ, Matsushime H, Kato JY and Nordeen SK. (1988). Biotechniques, 6, 454 ± 456. Livingston DM. (1993). Cell, 73, 487 ± 497. O'Conner JD, Lam EW-F, Gin S, Zhong S, Leighton C, Field SJ, Tsai F-Y, Kuo F, Zubiaga AM, Kaelin Jr, WG, Burbidge SA and Lu X. (1995). EMBO J., 14, 6184 ± 6192. Livingston DM, Orkin SH and Greenberg ME. (1996). Pan H and Griep AE. (1994). Genes & Dev., 8, 1285 ± 1299. Cell, 85, 549 ± 561. Perkins ND, Felzien LK, Betts JC, Leung K, Beach DH and Fields S and Jang SK. (1990). Science, 249, 1046 ± 1049. Nabel GJ. (1997). Science, 275, 523 ± 527. Flemington EK, Speck SH and Kaelin Jr, WG. (1993). Proc. Phillips AC, Bates S, Ryan KM, Helin K and Vousden KH. Natl. Acad. Sci., 90, 6914 ± 6918. (1997). Genes & Dev., 11, 1853 ± 1863. Girling R, Partridge JF, Bandara LR, Burden N, Totty NF, Pietenpol JA, Tokino T, Thiagalingam S, El-Deiry W, Hsuan JJ and La Thangue NB. (1993). Nature, 362, 83 ± Kinzler KW and Vogelstein B. (1994). Proc. Natl. Acad. 87. Sci., 91, 1998 ± 2002. Gu W, Shi X-L and Roeder RG. (1997). Nature, 387, 819 ± Pines J. (1995). Seminars in Cancer Biology, 6, 63 ± 72. 823. Qin X-Q, Livingston DM, Kaelin Jr, WG and Adams PD. Harper JW, Adami GR, Wei N, Keyomarsi K and Elledge (1994). Proc. Natl. Acad. Sci., 91, 10918 ± 10922. SJ. (1993). Cell, 75, 805 ± 816. Querido E, Teodoro JD and Branton PE. (1997). J. Virol., Haupt Y, Rowan S and Oren M. (1995a). Oncogene, 10, 71, 3526 ± 3533. 1563 ± 1571. Sabbatini P, Lin J, Levine AJ and White E. (1995). Genes & Haupt Y, Rowan S, Shaulian E, Vousden KH and Oren M. Dev., 9, 2184 ± 2192. (1995b). Genes & Dev., 9, 2170 ± 2183. Serrano M, Hannon GJ and Beach D. (1993). Nature, 366, Helin K, Harlow E and Fattaey AR. (1993). Mol. Cell. Biol., 704 ± 707. 13, 6501 ± 6508. Sherr CJ. (1993). Cell, 73, 1059 ± 1065. Hinds PW, Mittnacht S, Dulic V, Arnold A, Reed SI and Sherr CJ and Roberts JM. (1995). Genes & Dev., 9, 1149 ± Weinberg RA. (1992). Cell, 70, 993 ± 1006. 1163. Hsieh J-K, Fredersdorf S, Kouzarides T, Martin K and Lu Shikama N, Lyon J and La Thangue NB. (1997). Trends Cell. X. (1997). Genes & Dev., 11, 1840 ± 1852. Biol., 7, 230 ± 236. Jacks T, Fazeli A, Schmitt EM, Bronson RT, Goodell MA Somasundaram K and El-Deiry WS. (1997). Oncogene, 14, and Weinberg RA. (1992). Nature, 359, 295 ± 300. 1047 ± 1057. Johnson DG, Schwartz JK, Cress WD and Nevins JR. Sùrensen TS, Girling R, Lee C-W, Gannon J, Bandara LR (1993). Nature, 365, 349 ± 351. and La Thangue NB. (1996). Mol. Cell. Biol., 16, 5888 ± Johnson DG, Ohtani K and Nevins JR. (1994). Genes & Dev., 5895. 8, 1514 ± 1525. Thut CJ, Chen JL, Klemin R and Tjian R. (1995). Science, Johnston LH. (1992). Trends Cell. Biol., 2, 353 ± 357. 267, 100 ± 104. Jooss K, Lam EW-F, Bybee A, Girling R, Muller R and La Trouche D, Cook A and Kouzarides T. (1996). Nucleic Acids Thangue NB. (1995). Oncogene, 10, 1529 ± 1536. Research, 24, 4139 ± 4145. Kaelin Jr, WG, Krek W, Sellers WR, DeCaprio JA, Waldman T, Kinzler KW and Vogelstein B. (1995). Cancer Ajchanbaum F, Fuchs CS, Chittenden T, Li Y, Farnham Res., 55, 5187 ± 5190. PJ, Blanar MA, Livingston DM and Flemington EK. Weinberg RA. (1995). Cell, 81, 323 ± 330. (1992). Cell, 70, 351 ± 364. White E. (1996). Genes & Dev., 10, 1 ± 15. Kern SE, Pietenpol SA, Thiagalingam S, Seymour A, Kinzler Wu X and Levine AJ. (1994). Proc. Natl. Acad. Sci., 91, KW and Vogelstein B. (1992). Science, 256, 827 ± 830. 3602 ± 3606. Knudson CM, Tung KSK, Tourtellotte WG, Brown GAJ Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R and and Korsmeyer SJ. (1995). Science, 270, 96 ± 99. Beach D. (1993). Science, 366, 701 ± 704. Ko LJ and Prives C. (1996). Genes & Dev., 10, 1054 ± 1072. Yamasaki L, Jacks T, Bronson R, Goillot E, Harlow E and Koh J, Enders GH, Dynlacht BD and Harlow E. (1995). Dyson N. (1996). Cell, 85, 537 ± 547. Nature, 375, 506 ± 510. Zakut-Houri R, Bienz-Tadmor B, Givol D and Oren M. Kraus VB, Moran E and Nevins JR. (1992). Mol. Cell. Biol., (1985). EMBO J., 4, 1251 ± 1255. 12, 4391 ± 4399. Zamanian M and La Thangue NB. (1992). EMBO J., 11, La Thangue NB. (1994). Trends Biochem. Sci., 19, 108 ± 114. 2603 ± 2610. Lill NL, Grossman SR, Ginsberg D, DeCaprio J and Livingston DM. (1997). Nature, 387, 823 ± 827.