Effects of depletion of CREB-binding protein on c-Myc regulation and cell cycle G1-S transition.

We recently reported that the transcriptional coactivator and histone acetyltransferase p300 plays an important role in the G(1) phase of the cell cycle by negatively regulating c-myc and thereby preventing premature G(1) exit (Kolli, et al. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 4646-4651; Baluchamy, et al. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 9524-9529). Because p300 does not substitute for all CREB-binding protein (CBP) functions, we investigated whether CBP also negatively regulates c-myc and prevents premature DNA synthesis. Here, we show that antisense-mediated depletion of CBP in serum-deprived human cells leads to induction of c-myc and that such cells emerge from quiescence without growth factors at a rate comparable with that of p300-depleted cells. The CBP-depleted cells contained significantly reduced levels of the cyclin-dependent kinase inhibitor p21 and low levels of p107 and p130 (but not pRb) phosphorylation, suggesting that these factors, along with elevated levels of c-Myc, contribute to induction of DNA synthesis. Antisense c-Myc reversed the phosphorylation of p107 and p130 and the induction of S phase in CBP-depleted cells, indicating that up-regulation of c-myc is directly responsible for the induction of S phase. Furthermore, the serum-stimulated p300/CBP-depleted cells did not traverse beyond S phase, and a significant number of these cells died of apoptosis, which was not related to p53 levels. These cells also contained significantly higher levels of c-Myc compared with normal cells. When c-myc expression was blocked by antisense c-Myc, the apoptosis of the serum-stimulated CBP-depleted cells was reversed, indicating that high levels of c-Myc contribute to apoptosis. Thus, despite their high degree of structural and functional similarities, normal levels of both p300 and CBP are essential for keeping c-myc in a repressed state in G(1) and thereby preventing inappropriate entry of cells into S phase. In addition, both these proteins also provide important functions in coordinated cell cycle progression.

p300 and the CBP 1 are two large highly homologous and conserved transcriptional adapter proteins containing histone acetyltransferase activity that link transcription with chromatin remodeling (1). CBP and p300 also acetylate a number of sequence-specific transcription factors, including p53, E2F, and T-cell factor-4, affecting their transcriptional activation and DNA binding activities (2)(3)(4). Mice in which both alleles coding for either CBP or p300 are disrupted die at the embryonic stage due to multiple developmental abnormalities, suggesting that both the p300 and CBP genes are required for the normal cell proliferation and development (5,6). Patients with haploinsufficiency of the CBP gene display severe abnormalities characteristic of Rubinstein-Taybi syndrome, comprising mental retardation and physical deformities (7). These patients also have an increased incidence of cancer. Mutations in the p300 and CBP genes have been reported in several cancers, including gastric and colon cancers, and certain forms of leukemia, indicating that they may have tumor suppressor properties (reviewed in Refs. 1 and 8). The intrinsic ubiquitin ligase activity of p300 plays a role in the proteasomal degradation of p53 (9,10). p300 and CBP also function as coactivators for tumor suppressor proteins such as p53, suggesting an indirect role for these proteins in tumor suppression (2,11,12).
Previous adenovirus (Ad) E1A studies have indicated that p300 may play a role in maintaining cells in the G 1 phase of the cell cycle. For example, E1A can stimulate DNA synthesis in quiescent baby rat kidney cells, which is dependent on its binding to p300 (13)(14)(15). E1A can also re-stimulate DNA replication in terminally differentiated cardiac myocytes through its p300/CBP-binding domain (16). There is evidence that p300 is involved in terminal differentiation in several cell types, including muscles, neurons, and enteroendocrine cells (reviewed in Ref. 1). During terminal differentiation, p300 transactivates p21 by cooperating with Sp1, Sp3, or tissue-specific transcription factors, suggesting that p300 may play a role in keeping cells in G 0 /G 1 (1,17,18). However, fibroblasts from p300 knockout mice are unable to replicate and appear to undergo cell cycle arrest, indicating that some p300 function may be necessary for cell cycle progression (5). Furthermore, CBP has been shown to be associated with Cdk2 in vivo and may be a substrate for cyclin E/Cdk2 (19). p300 provides a coactivator function for both growth stimulatory transcription factors such as E2F (3) and growth inhibitory proteins such as p53 (2,11,12). p300 is underphosphorylated during G 0 /G 1 and hyperphosphorylated during S and G 2 /M phases (20). Therefore, the overall roles played by p300/CBP at different stages of the cell cycle remain to be determined.
Although, in most studies, p300 was able to substitute for CBP functions, not all p300 functions can be substituted by CBP (reviewed in Ref. 1). For example, in F9 cells, both differentiation and cell cycle arrest induced by retinoic acid are dependent on normal levels of p300, but not of CBP (21). Mice lacking one copy of the CBP gene display some of the characteristics found in Rubinstein-Taybi syndrome patients, including the increased development of hematological neoplasm (6). In striking contrast, p300 heterozygous mice do not display this phenotype (5). These proteins also appear to play distinct roles in maintaining normal hematopoiesis (22). Mouse p300 Ϫ/Ϫ fibroblasts are also defective in retinoic acid-induced transcription despite the presence of normal level of CBP in these cells. In contrast, these cells maintain normal levels of CREB-activated transcription (5). In transient assays, transcription of the cyclin D 1 gene can be induced by p300, but not by CBP (23). Other studies related to functional differences between p300 and CBP include the involvement of p300 (but not CBP) acetyltransferase activity during myogenesis (24) and complexes containing p300 (but not CBP) and the nuclear proto-oncoprotein SYT in the proper activation of ␤ 1 integrin and control of cell adhesion (25).
Recently, we showed that quiescent human cells depleted of p300 enter S phase without serum and that this premature DNA synthesis is due to up-regulation of c-myc (26). Similarly, we also showed that conditional overexpression of p300 represses c-myc and S phase entry (27). In view of the abovenoted functional differences between p300 and CBP, we investigated the effects of depletion of CBP on c-myc regulation and G 1 -S transition. We now show that reduced synthesis of CBP in quiescent cells also results in the induction of c-myc and S phase entry. However, p300-or CBP-depleted cells with or without serum stimulation did not undergo normal cell cycle progression. Such cells expressed c-myc to much higher levels compared with control cells, and a significant portion of them died of apoptosis after serum stimulation, which did not appear to be related to p53 levels. Furthermore, we also show that in growth-arrested p300/CBP-depleted cells, the Cdk inhibitor p21 is dramatically down-regulated, and p107 and p130 (but not pRb) are hyperphosphorylated at low levels, suggesting that in addition to the induction of c-myc, the alterations in these negative regulators of the cell cycle may also contribute to the induction of S phase. Finally, we show that the inappropriate up-regulation of c-myc in CBP-depleted cells is primarily responsible for the induction of S phase and apoptosis, as inhibition of c-myc expression reverses both these effects. Thus, our results suggest that despite their structural and functional similarities, normal levels of both p300 and CBP are required for keeping c-myc in a repressed state and maintaining cells in the G 1 phase of the cell cycle. These two proteins may have a universal role as checkpoint proteins, preventing the untimely onset of DNA synthesis in senescent or differentiated cells. Our results also indicate that normal levels of both these proteins are essential for coordinated cell cycle progression.

EXPERIMENTAL PROCEDURES
Cells and Viruses-MCF10A cells are a nontransformed human breast epithelial cell line that arose as a spontaneous immortalization of normal breast epithelial cells derived from a reduction mammoplasty specimen (28). The growth conditions for these cells were described previously (29). AS-p300 is a replication-defective Ad vector that expresses 1784 bp of p300 cDNA 5Ј-sequence in an antisense orientation from the cytomegalovirus promoter in the E1A region (26). AS-CBP is identical to AS-p300 except that it expresses 2.0 kb of human CBP cDNA sequence (30) and was constructed as described (31). Ad␤-gal expresses ␤-galactosidase from the cytomegalovirus promoter (26). AdM4 is an Ad vector that contains a promoter-luciferase reporter cassette with four copies of the c-Myc-binding site (E-box element) cloned upstream of the herpes simplex virus thymidine kinase minimal promoter and linked to the luciferase reporter gene (27). AS-Myc is an Ad vector that expresses antisense c-myc sequences (26).
Cell Cycle Analysis-Cells were seeded overnight at a density of 1 ϫ 10 6 /100-mm dish and serum-starved. Thirty-six hours after serum starvation, they were infected with appropriate Ad vectors at a multiplicity  of infection of 200 and maintained under serum-starved conditions for  the duration of the experiments shown in the figure legends. For the  serum stimulation experiments, cells were stimulated with serum 18 h  after vector infection and harvested at the indicated time points. For experiments in which cell cycle analysis was carried out under continued serum starvation, cells were maintained in serum-free medium for the duration of the experiments and then harvested and processed for FACS analysis as described (29).
RNA Analysis-MCF10A cells were seeded, starved, and infected with Ad vectors as described above; starvation was continued for 18 h after virus infection; and the cells were stimulated with growth medium containing 5% horse serum. Total RNA was isolated from cells using an RNA isolation kit (Promega catalog no. Z3100). Poly(A)-containing RNA was isolated from total RNA using a poly(A) ϩ RNA isolation kit (QIA-GEN catalog no. 72012). Equal quantities of RNA were analyzed by Northern blotting using a human c-myc probe corresponding to 1 kb of N-terminal coding sequence.
Apoptosis Assays-Apoptosis assays for p300-or CBP-depleted serum-starved cells were carried out using the annexin V binding assay to assess the loss of phospholipid asymmetry of the plasma membrane (32) and by staining the mitochondria with chloromethyl-X-rosamine to assess the changes in the mitochondrial membrane potential (33). Cells were serum-starved, infected with Ad vectors, and serum-stimulated as described above for the cell cycle experiment. At the indicated times, the adherent cells were harvested after trypsinization and pooled with the previously collected floating cells, which were washed once with calcium/magnesium-free phosphate-buffered saline. Annexin V binding assay was then carried out using a commercially available kit (Immunotech catalog no. PN IM3546). Briefly, the phosphate-buffered salinewashed cell pellet was resuspended in 100 l of 1ϫ binding buffer at a concentration of 5 ϫ 10 6 /ml. One microliter of FITC-conjugated annexin V and 5 l of propidium iodide were then added from the stock solutions provided in the kit. The cell suspension was mixed gently and incubated in the dark for 10 min on ice. It was then diluted to 5-fold with 1ϫ binding buffer and analyzed by flow cytometry within 1 h using a Beckman Coulter Epics XL-MCL counter. Cells were analyzed at an excitation emission wavelength of 488 nm and at detection emission wavelengths of 620 nm for propidium iodide-positive cells and 525 nm for FITC-conjugated annexin V-positive cells.
Mitochondrial membrane potential was assessed by staining the mitochondria with chloromethyl-X-rosamine using a commercially available kit (Molecular Probes catalog no. M-7512). Briefly, cells were infected and harvested at the indicated times as described above for the annexin V binding assays, resuspended in 3 ml of culture medium containing 100 nM MitoTracker Red CMX-Ros (chloromethyl-X-rosamine), and incubated for 20 min at 37°C in a 5% CO 2 incubator. MitoTracker Red CMX-Ros is a modified chloromethyl-X-rosamine (as described in the Molecular Probes product information sheet). This modified chloromethyl-X-rosamine labels mitochondria with a longer wavelength fluorescence emission that can be separated from FITCconjugated annexin V fluorescence. The cell suspension containing chloromethyl-X-rosamine was spun down, washed once with calcium/magnesium-free phosphate-buffered saline, and resuspended in 100 l of annexin V binding buffer. One microliter of FITC-conjugated annexin V was then added, and the cell suspension was incubated on ice in the dark for 10 min. At the end of the 10-min incubation, the cell suspension was diluted to 5-fold with 1ϫ annexin V binding buffer and analyzed by flow cytometry at an excitation emission wavelength of 488 nm and at detection emission wavelengths of 620 nm for chloromethyl-X-rosamine-positive cells and 525 nm for FITC-conjugated annexin V-positive cells (33).
Myc Activity Assays-Serum-starved MCF10A cells were infected with Ad vectors as described above for cell cycle analysis and superinfected 1 h later with AdM4 at 10 plaque-forming units/cell. Infection was continued for another 18 h, and cells were either stimulated with serum or maintained under starvation conditions as appropriate. At the indicated times, cells were harvested in lysis buffer, and the luciferase activity was measured using a luciferase assay kit (Promega catalog no. E1501) as described (27). A parallel infection was carried out to assay the distribution of the vector-infected cells in G 1 , S, and G 2 /M phases to ensure that the p300-or CBP-depleted cells exited G 1 as expected.
Western Blot Analysis-To determine the levels of p300 and CBP, quiescent MCF10A cells were infected with antisense vectors, lysed, and immunoblotted using anti-p300 or anti-CBP antibodies as described (26). c-Myc, p53, p21, and the three Rb family proteins were analyzed by Western immunoblotting using appropriate antibodies (see figure legends for antibodies and other details). Cdk Assays-To determine cyclin D kinase activity, antisense-expressing cells were lysed at the indicated times, immunoprecipitated with anti-cyclin D antibody, and assayed for kinase activity using a recombinant truncated Rb protein as a substrate (Upstate Biotechnology catalog no. 12-439) (34). Cyclin E kinase activity was assayed using histone H1 as a substrate (27).

RESULTS
Quiescent Cells Depleted of CBP Exit G 1 without Serum at Levels Similar to Those Depleted of p300 -Our previous study showed that an antisense Ad vector expressing 1.7 kb of p300 cDNA 5Ј-sequence (AS-1; referred to here as AS-p300) effectively reduces the levels of p300 in quiescent cells (26). To block the synthesis of CBP in quiescent cells, we used a replicationdefective Ad vector expressing 2 kb of human CBP cDNA sequence from the cytomegalovirus promoter (AS-CBP). To determine whether AS-CBP would reduce CBP synthesis without affecting p300 levels, serum-starved MCF10A cells (see "Experimental Procedures" for a description of these cells) were infected with antisense vectors; and 20 h after infection, the levels of p300 and CBP in the cells were quantified by Western immunoblotting using appropriate antibodies. An Ad vector expressing ␤-galactosidase from the cytomegalovirus promoter (Ad␤-gal) was used as a control in these and all other experiments in this work. Fig. 1A shows that AS-CBP infection resulted in a significant reduction in CBP levels without affecting p300 levels. Consistent with our previous report, infection of cells with AS-p300 resulted in a considerable reduction in p300 levels, but not CBP levels. This result suggests that our antisense vectors are effective in reducing the synthesis of respective gene products with a high degree of specificity despite the fact that these two proteins are highly homologous (35)(36)(37).
To determine whether serum-starved cells infected with AS-CBP would exit G 0 /G 1 in a manner similar to AS-p300-infected cells, MCF10A cells were serum-starved and infected with the antisense vectors or Ad␤-gal, and serum starvation was continued (see Fig. 1B for the time course of infection and harvesting). After a total of 52 h of serum starvation, cells were harvested at 4-h time intervals up to 36 h post-infection. Fig.  1C shows that both CBP-and p300-depleted cells exited G 1 at similar rates, although AS-CBP was slightly more effective in this regard. For example, at 36 h after infection, ϳ35% of the AS-CBP-infected cells were in S phase fractions, whereas at this time point, ϳ28% of the AS-p300-infected cells were found in S phase. In contrast, the number of Ad␤-gal-infected cells in S phase at this time point was Ͻ5%, ruling out the effects of genes expressed by the vector backbone contributing to the induction of S phase. Both AS-CBP-and AS-p300-infected cells steadily accumulated in S phase, but did not traverse into G 2 /M, suggesting that these cells lack the ability to complete the cell cycle. This aspect is addressed below. Thus, our results show that both p300 and CBP provide an activity that allows the cells to remain in G 0 /G 1 phase.
Surprisingly, we did not observe a significant increase in the number of S phase-specific cells when cells expressed both types of antisense sequences simultaneously as opposed to cells expressing only antisense p300 or CBP sequences (data not shown). At present, we do not know the reasons for the lack of an additive or synergistic effect.
c-myc Is Up-regulated at Similar Levels in Antisense CBPand p300-expressing Cells-Our previous study showed that premature DNA synthesis that occurs in cells expressing antisense p300-specific sequences is due to up-regulation of c-myc (26). To determine whether the S phase entry of cells infected with AS-CBP was also due to up-regulation of c-myc, we assayed the levels of c-Myc protein in serum-starved cells expressing antisense CBP (see Fig. 2A for the time course of infection and harvesting). As a comparison, the levels of Myc in cells expressing antisense p300 were also determined in parallel. Myc levels increased significantly in both antisense CBP and p300-expressing cells compared with control cells infected with Ad␤-gal (Fig. 2, B and C, respectively). To determine the endogenous Myc transcriptional activation activity, we used an Ad vector (AdM4) that expresses a promoter-luciferase reporter cassette containing four copies of the c-Myc protein-binding site (E-box element) (27). We showed previously that an increase in luciferase activity in AdM4-infected cells directly correlates with an increase in c-Myc protein levels (27). To assay the induction of c-myc in p300-depleted quiescent cells, serumstarved cells were first infected with AS-CBP, AS-p300, or Ad␤-gal and again infected 1 h later with AdM4. Seventeen hours following AdM4 infection and every 2 h thereafter, they were harvested, and the luciferase activities of the infected cells were determined. As shown in Fig. 2 (D and E), the reporter activity in the antisense-expressing cells began to rise at 19 h post-infection and continued to increase up to 25 h (later time points were not tested). The luciferase activity of Ad␤-galinfected cells increased only at moderate levels, indicating that the 3-4-fold increase in luciferase activity observed in antisense-infected cells was due to an increase in Myc protein levels. Thus, we conclude that depletion of either p300 or CBP leads to up-regulation of c-myc. Furthermore, the above results indicate that both proteins are essential for the negative regulation of c-myc and the prevention of cells from exiting G 1 . Using antisense c-Myc, we have shown previously that the induction of DNA synthesis in p300-depleted cells is due to c-myc up-regulation (26). Using this strategy, we show below that S phase induction in CBP-depleted cells is also due to c-myc up-regulation. Consistent with the lack of additive effect of the two antisense viruses on DNA synthesis discussed above, we did not observe an increase in Myc activity when cells were infected simultaneously with two antisense vectors in our several independent experiments (data not shown).
p21 Levels Decrease Significantly in Antisense p300/CBPexpressing Cells-We showed previously that p300 negatively regulates c-myc expression in G 1 , which is primarily responsible for preventing early entry of cells into S phase (26,27). However, we were curious to know whether there are any changes in the status of the well known negative regulators of the cell cycle in p300/CBP-depleted cells that would contribute to DNA synthesis, including the cyclin/Cdk activity inhibitor p21 and pRb, p107, and p130, which sequester E2F. Because p300/CBP is a coactivator of transcription factors that control transcription of the p21 gene, we determined whether reduced synthesis of CBP affects the levels of p21 in antisense-expressing cells. The levels of p21 in cell extracts prepared from growth-arrested cells expressing antisense CBP sequences were determined by Western blotting (see Fig. 3A for the time course of infection and harvesting of cells). The data presented in Fig. 3B show that a significant reduction of p21 levels was observed as early as 18 h after AS-CBP infection compared with control samples, and the levels dropped further by 28 h post-infection. Interestingly, using the same cell lysates, we observed that at the 18-h time point, there was only a marginal induction of c-myc ( Fig. 3H; note that same cell lysates were used for the data shown Figs. 3 (B-H) and 8A). This indicates that the down-regulation of the p21 genes occurs before a significant rise in Myc levels. We have shown previously that p21 RNA levels decline in antisense AS-p300-expressing cells (26). Thus, the p21 gene is down-regulated in both CBP-and p300-depleted cells.
p107 and p130 (but Not pRb) Are Hyperphosphorylated at Low Levels in Antisense p300/CBP-expressing Cells-A normal growth-stimulated G 1 -S transition involves phosphorylation of pRb and the related proteins p107 and p130, resulting in the release of E2F from the chromatin repressor complexes, which is essential for S phase progression (reviewed in Refs. 38 -40). Thus, we were interested to know whether phosphorylation of pRb and the related proteins would occur in p300/CBP-depleted cells, which might contribute to the induction of DNA synthesis. The Western blot data presented in Fig. 3C show that there were negligible amounts of slower migrating forms of pRb in lanes corresponding to cell extracts prepared from quiescent cells infected with AS-CBP for 24 or 28 h (the B-gal ϩserum lane is a positive control and represents phosphorylation of pRb in serum-stimulated cells undergoing a normal S phase progression, and the phosphorylated form of pRb is shown by the arrow; see legend to Fig. 3C). Studies have shown that in G 0 and early G 1 , p130 appears as a hypophosphorylated doublet representing two phospho forms in Western blots, and a new hyperphosphorylated form appears in late G 1 and S phases (41)(42)(43). Consistent with these reports, p130 migrated as a doublet in control Ad␤-gal lanes, and a new band migrating slower than the two phosphorylated forms was evident in serum-stimulated cells, which served as a positive control (Fig.  3D, phosphorylated form shown by the arrow). A faint slower migrating band was evident in AS-CBP-infected cells at 24 h after infection, suggesting a low level phosphorylation of p130 ( Fig. 3D; the 28-h sample could not be completed in this experiment). Similarly, we observed a faint band migrating slower than the p107 band in AS-CBP-infected cells at the 24-and 28-h time points, indicating that p107 was also phosphorylated at low levels. Note that in serum-stimulated cells serving as a positive control, the band corresponding to phosphorylated p107 was merged with the p107 band (Fig. 3E). Because pRb was not hyperphosphorylated in antisense-expressing cells, we determined whether the cyclin D/Cdk activity was elevated in these cells. As shown in Fig. 3F, there was no change in the levels of cyclin D/Cdk activity at 20 and 24 h post-infection. In contrast, AS-CBP-infected cells showed elevated levels of cyclin E/Cdk activity beginning 20 h after virus infection (Fig. 3G). We believe that cyclin A/Cdk activity is also elevated in these cells, as we showed previously that both cyclin E and cyclin A kinase activities are induced in antisense p300-expressing cells (26). The lack of induction of cyclin D/Cdk activity in these cells could be attributed to the lack of growth factors in the medium because induction of cyclin D is dependent on growth factors (reviewed in Ref. 44).
Antisense c-Myc Reverses the Induction of S Phase and the Phosphorylation of p107 and p130 -To determine whether the induction of S phase and phosphorylation of the Rb family proteins in CBP-depleted cells was the result of induction of c-myc, we inhibited c-myc expression by expressing antisense c-Myc sequences as described previously (26), followed by monitoring the capacity of the CBP-depleted cells to exit G 1 . Serumstarved cells were co-infected with AS-CBP and an Ad vector that expresses antisense c-myc sequences (AS-Myc) (26). To maintain a constant multiplicity of infection, the Ad␤-gal vector was included in the protocol where appropriate. To confirm the effectiveness of AS-Myc in blocking the synthesis of c-Myc, c-myc transcriptional activity was monitored using the reporter assay described above. As shown in Fig. 4B (Fig. 4A shows the time course for infection and harvesting of cells), at 16 h postinfection, the luciferase activity in AS-CBP-infected cells was comparable with that detected in different control samples. In contrast, at 25 h post-infection, the AS-CBP-infected cells expressed ϳ3-fold more luciferase activity compared with cells infected with Ad␤-gal or with Ad␤-gal and AS-Myc. Importantly, in cells co-infected with AS-Myc and AS-CBP, Myc activity was reduced to ϳ50% of that observed in AS-CBPinfected cells, indicating that AS-Myc was effective in blocking c-myc expression.
Next, the S phase entry of cells co-infected with AS-CBP and AS-Myc was compared with that of AS-CBP-infected cells. As shown in Fig. 4C, at 16 h post-infection, Ͻ5% of the cells were detected in S phase in all samples. At 36 h post-infection, ϳ33% of the AS-CBP-infected cells accumulated in S phase, whereas Ͻ5% of the Ad␤-gal-and Ad␤-gal/AS-Myc-infected cells accumulated in S phase. Importantly, the number of S phasespecific cells co-infected with AS-CBP and AS-Myc was reduced to Ͻ10%, indicating that AS-Myc efficiently reversed the induction of S phase in CBP-depleted cells. Thus, we conclude that the induction of S phase in CBP-depleted cells is the result of induction of c-myc.
To determine whether AS-Myc would also reverse the phosphorylation of p107 and p130, cell extracts were prepared from cells infected with AS-CBP and with AS-CBP and AS-Myc for 28 h along with appropriate controls as shown in Fig. 4C. The phosphorylation status of p107 and p130 was then determined by Western immunoblotting as described in the legend to Fig.  3. Fig. 4D shows that p107 was not phosphorylated in cells infected with Ad␤-gal or with Ad␤-gal and AS-Myc. p107 was phosphorylated at moderate levels in AS-CBP-infected cells, which was reversed by AS-Myc (Fig. 4D Serum-starved p300-and CBP-depleted Cells Fail to Progress beyond S Phase after Serum Stimulation-Our FACS analysis of the serum-starved antisense-expressing cells suggested to us that these cells accumulated in S phase steadily, but did not traverse beyond S phase (data not shown). As growth factor-stimulated activation of cyclin D is essential for the G 1 -S transition of serum-starved cells, we considered it possible that serum stimulation of the cells accumulating in S The cell lysates were analyzed as described for C using anti-p130 monoclonal antibody (Santa Cruz Biotechnology catalog no. sc-317). E, Western blot showing the phosphorylation status of p107. Cell lysates were analyzed as described for C using anti-p107 monoclonal antibody (Santa Cruz Biotechnology catalog no. sc-318). The phosphorylated forms of pRb, p130, and p107 are shown by arrows in C-E. F, cyclin D kinase activity in serum-starved AS-CBP-infected cells. The cyclin D-Cdk complex was immunoprecipitated using anti-cyclin D antibody (Santa Cruz Biotechnology catalog no. sc-753), and the kinase activity in the immunoprecipitated complex was assayed using a recombinant truncated Rb protein as a substrate (see "Experimental Procedures"). G, cyclin E kinase activity in serum-starved AS-CBP-infected cells. The cyclin E-Cdk complex was immunoprecipitated using anti-cyclin E antibody (Santa Cruz Biotechnology catalog no. sc-248), and the associated kinase activity was determined using histone H1 as a substrate as described under "Experimental Procedures." H, Western blot showing the levels of c-Myc protein in cell lysates. Serum-starved cells were infected with antisense or Ad␤-gal vectors as shown in A. At the indicated time points, 60 g of protein from each cell lysate was Western-immunoblotted on an SDS-7.5% polyacrylamide gel using anti-c-Myc monoclonal antibody 9E10 (gift of J. DeCaprio). These data are taken from Fig. 8A after cropping, and tubulin levels are shown in Fig. 8A. phase would allow them to progress into G 2 /M and complete the cell cycle. Accordingly, cells were serum-starved for 36 h and infected with antisense vectors or Ad␤-gal, and starvation was continued for another 18 h as described above (Fig. 5A). Cells were then stimulated with serum, and the distribution of cells in G 1 , S, and G 2 /M phase fractions was determined by flow cytometry. Fig. 5B shows that control cells exited G 1 steadily beginning at 14 h after serum stimulation, continued to accumulate in S phase up to 26 h, and then declined. Quantification of these cells in the G 2 /M phase fraction indicated that at 26 h after serum stimulation, the control cells began to accumulate in the G 2 /M phase fraction, and this accumulation continued up to 29 h (Fig. 5C). After ϳ29 h, the number of cells in the G 2 /M phase fraction began to decline, with a corresponding increase in the G 1 phase fraction (data not shown). This pattern of accumulation was comparable with that observed for mockinfected cells, indicating that the Ad␤-gal-infected cells cycled normally. However, the distribution of antisense-expressing cells during these time points was dramatically different. For example, significantly more antisense-expressing cells appeared in S phase at 14 h after serum stimulation compared with control cells and continued to accumulate, reaching ϳ50% by 35 h (Fig. 5B). Surprisingly, unlike control cells, even after prolonged serum stimulation, neither antisense p300-nor CBP-expressing cells accumulated in the G 2 /M fraction (Fig.  5C), indicating that these cells were unable to traverse beyond S phase. These results suggested to us that a substantial number of the antisense-expressing cells did not survive beyond S phase and might be undergoing apoptosis.
A Considerable Portion of p300-or CBP-depleted Cells Die of Apoptosis after Serum Stimulation-To determine whether antisense-expressing cells die of apoptosis after serum stimula-FIG. 6. Apoptosis assay using annexin V/propidium iodide and chloromethyl-X-rosamine staining methods. All apoptosis assays were repeated twice with reproducible results, and data shown are from one such experiment. A-C, bar diagrams showing a comparison of the percentage apoptosing cells in serum-starved and serum-stimulated MCF10A cells infected with AS-p300, AS-CBP, and Ad␤-gal (B-gal) viruses. Serum-starved and serum-stimulated MCF10A cells infected with different Ad vectors were harvested at the indicated times and stained with annexin V/propidium iodide. The annexin V-stained cells were analyzed by flow cytometry as detailed under "Experimental Procedures." D, scatter diagrams for distribution of AS-p300-, AS-CBP-, and Ad␤-gal-infected MCF10A cells that are undergoing apoptosis at 0 and 16 h after serum stimulation. E, bar diagram showing the percentage apoptosing cells based on chloromethyl-X-rosamine staining assay. Serum-starved MCF10A cells were infected with AS-p300, AS-CBP, and Ad␤-gal viruses and stimulated with serum as described for all cell cycle analyses. The cells were harvested, stained with annexin V/chloromethyl-X-rosamine, and analyzed by flow cytometry as described under "Experimental Procedures." tion, annexin V binding assays were carried out to assess the loss of phospholipid asymmetry of the plasma membrane that occurs during the initial stages of apoptosis (32). Cells were also assayed for the loss of mitochondrial membrane potential that is characteristic of apoptosis by staining cells with the fluorescent stain trimethyl-X-rosamine (33). Cells were serumstarved for 36 h and infected with Ad␤-gal, AS-p300, or AS-CBP at a multiplicity of infection of 200 and maintained under serum starvation conditions for a total of 52 h. Cells were then stimulated with growth medium containing 5% serum as described above for the cell cycle analysis. At the indicated times shown in the figure legends, cells were incubated with FITCconjugated annexin V and propidium iodide as described under "Experimental Procedures" and analyzed by flow cytometry (Fig. 6, A-D). The signals that are registered in the upper right quadrangle (propidium iodide-positive and annexin V-positive; late stages of apoptosis), which are indicative of necrotic and apoptotic cells, and those in the lower right quadrangle (propidium iodide-negative and annexin V-positive; early stages of apoptosis), which are indicative of apoptotic cells, together constitute the percentage of apoptotic cells in a total population (Fig. 6D). These data indicate that the percentage of Ad␤-galinfected cells undergoing apoptosis before serum stimulation (0-h time point) (Fig. 6, A and D) was ϳ6 -8%, which is comparable with the percentage of AS-p300-and AS-CBP-infected cells undergoing apoptosis before serum stimulation (Fig. 6, B-D; scatter diagrams for the 9-and 12-h samples are not shown). The percentage of cells undergoing apoptosis upon Ad␤-gal infection remained ϳ10% at the 9-, 12-, and 16-h time points in the presence or absence of serum (Fig. 6, A and D). The percentage of AS-p300-and AS-CBP-infected cells undergoing apoptosis in the absence of serum stimulation also remained Ͻ10%. In contrast, at 16 h after serum stimulation, the percentage of apoptotic cells expressing antisense sequences increased up to 18 -20% (Fig. 6, B-D).
Using trimethyl-X-rosamine, which specifically stains mitochondria, we assessed the loss of mitochondrial membrane potential that occurs during apoptosis (33). The percentage of control cells and cells expressing antisense p300-or CBP-specific sequences undergoing apoptosis after serum stimulation based on this assay is shown in Fig. 6E. These data indicate that ϳ10% of the control cells underwent apoptosis during the time points tested, whereas the percentage of antisense-expressing cells undergoing apoptosis increased to ϳ20% at 16 h after serum stimulation. These results are in agreement with the annexin V binding assay results. Thus, we conclude that at least a considerable number of serum-starved antisense-expressing cells undergo apoptosis within 12-16 h after serum stimulation, and this may contribute at least in part to their inability to travel beyond S phase. Both annexin V binding assays and trimethyl-X-rosamine staining assays were repeated twice, and the data were very reproducible.
c-Myc Is Further Induced in Quiescent p300-or CBP-depleted Cells after Serum Stimulation-Studies have shown that high levels of c-Myc can cause apoptosis in several cell types (45,46). Our previously published data (26) and those presented above clearly show that depletion of p300 or CBP in serum-starved cells results in the induction of c-myc. Thus, we considered it possible that c-myc that is already induced in the serumstarved antisense-expressing cells is further induced upon serum stimulation. The presence of high levels of c-Myc in cells deprived of growth factors and a further increase in its levels after serum stimulation would lead to non-physiological levels of Myc in the cell, which may contribute to the apoptotic effects. We show below that c-myc induced as a result of p300/CBP depletion in antisense-expressing cells was further induced by serum stimulation.
First, increases in c-Myc RNA levels in antisense-expressing FIG. 7. c-Myc RNA levels in serumstarved and serum-stimulated MCF10A cells expressing antisense p300 or CBP sequences. A, time course used for serum starvation, vector infection, serum stimulation, and harvesting. B, autoradiograms showing c-Myc and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNA levels. Cells were serum-starved; infected with Ad vectors expressing AS-p300, AS-CBP, and Ad␤-gal (B-gal); and then stimulated with serum as shown in A. Poly(A) ϩ RNA was isolated at the time points indicated and analyzed by Northern blot hybridization. Glyceraldehyde-3-phosphate dehydrogenase was analyzed after reprobing the filters used for c-myc analysis. One microgram of poly(A) containing RNA was used in each case. cells were determined by Northern blot hybridizations. Cells were serum-starved for 36 h and infected with AS-CBP, AS-p300, or Ad␤-gal, and serum starvation was continued. After a total of 54 h of starvation (18 h after antisense virus infection), one set of plates was stimulated with serum, and the other set was maintained under starvation conditions (schematically shown in Fig. 7A). Two and 4 h after stimulation, serumstimulated and unstimulated cells were harvested, and poly(A) ϩ RNA was isolated and subjected to Northern blot hybridizations using a human c-myc probe. Fig. 7B shows that Myc RNA levels began to rise by ϳ18 h post-infection in AS-CBP-and AS-p300-infected cells in the absence of serum (0-h time point) (Fig. 7B). At this time point, Myc RNA was barely detectable in control cells. Myc RNA levels began to rise by ϳ2 h after serum stimulation in cells infected with Ad␤-gal. Quan-titation of RNA bands of the autoradiogram using a densitometer indicated that the RNA levels in unstimulated AS-p300and AS-CBP-infected cells increased by ϳ2-3and 4 -5-fold, respectively, compared with the serum-stimulated Ad␤-gal-infected cells. After serum stimulation, the antisense-expressing cells showed a further increase of 2-3-fold in Myc RNA levels compared with those without serum stimulation (Fig. 7B, compare the 0, 2, and 4 lanes in the Unstimulated panels with those in the Stimulated panels). These RNA levels were much higher than those observed in Ad␤-gal-infected cells at these time points and were also higher than those found in antisenseexpressing cells without serum stimulation. Thus, we conclude that c-myc induced in serum-starved p300/CBP-depleted cells is further induced upon serum stimulation.
Next, by Western immunoblotting, we determined the Myc  Fig. 7A. At the indicated time points, 60 g of protein from each cell lysate was Western-immunoblotted on an SDS-7.5% polyacrylamide gel using anti-c-Myc monoclonal antibody 9E10. Tubulin levels were determined by reprobing the membrane used for Myc analysis with anti-tubulin antibody (Sigma catalog no. T5168). The numbers 0, 2, 4, and 6 represent time points with respect to serum stimulation. The numbers in parentheses refer to hours after vector infection. B, assay of Myc activity in cells expressing antisense CBP sequences. Quiescent cells infected with AS-CBP and AdM4 were serum-stimulated, and the luciferase (LUC) activity in starved and stimulated cells was assayed at the indicated time points using 5 g of protein as described under "Experimental Procedures." AU, activity units. C, assay for Myc activity in cells infected with AS-p300 with and without serum stimulation. Assays in both B and C were carried out in triplicate, and the means Ϯ S.D. are shown. This experiment was repeated twice with highly reproducible results. protein levels in cell lysates prepared at the time points described above (see Fig. 7A), and the data are shown in Fig. 8A. The numbers in parentheses refer to hours after vector infection, whereas the numbers 0, 2, 4, and 6 refer to hours after serum stimulation, with 18 h after vector infection as 0 h (see Fig. 7A). These results show that in serum-starved AS-CBPinfected cells, c-myc was induced transiently beginning at 20 h post-infection (Fig. 8A, lanes 2, 4, 8, and 12), whereas Myc was barely detectable in control samples (lanes 1, 3, 7, and 11). However, when these cells were stimulated with serum, c-myc was induced in control cells as expected (2-, 4-, and 6-h samples; lanes 5, 9 and 13, respectively). Importantly, there was a further increase in Myc levels in samples corresponding to AS-CBP-expressing cells at 2, 4 and 6 h after serum stimulation compared with AS-CBP-expressing cells not treated with serum (compare lanes 6, 10, and 14 with lanes 4, 8, and 12). Thus, serum-stimulated AS-CBP-infected cells contained the highest levels of Myc compared with any other vector-infected cells, indicating that even though c-myc was already induced in antisense-infected cells before serum stimulation, they remained responsive to serum stimulation.
To determine whether the induction pattern of c-myc correlates with its transcriptional activation activity, an experiment identical to that described above was performed, but including AdM4 infection following AS-CBP infection as shown in Fig This would translate into an overall increase of 2.5-fold in Myc activity in the serum-stimulated antisense CBP-expressing cells compared with the unstimulated antisense-expressing cells and an 8-fold increase compared with the unstimulated control cells. This pattern of increase in Myc activity was also observed 10 h after serum stimulation (data not shown). This experiment was repeated using the AS-p300 vector. The results shown in Fig. 8C indicate that cells infected with AS-p300 also showed a similar pattern of increase in Myc activity, with the highest level of Myc activity in serum-stimulated AS-p300infected cells. These results were reproducible in at least three independent experiments. In summary, these results show that the c-myc induction pattern was similar in both antisense AS-CBP-and AS-p300-expressing cells and that both cells contained higher levels of Myc after serum stimulation compared with serum-stimulated control cells undergoing normal G 1 -S transition.
Antisense c-Myc Reverses Apoptosis of CBP-depleted Cells-If the superinduction of c-myc observed in serum-stimulated p300/CBP-depleted cells contributes to apoptosis, then inhibition of c-myc expression should reverse this effect. To test this prediction, serum-starved cells were infected with AS-CBP and with AS-CBP and AS-Myc along with appropriate control infections as shown in Fig. 9B (Fig. 9A shows the time course of infection and harvesting). Cells were harvested at 8 and 16 h after serum stimulation, and the percentage of apoptotic cells in the total population was determined by annexin V binding assays as described under "Experimental Procedures." As shown in Fig. 9B, at 8 h after serum stimulation, ϳ7-8% of the cells in the control samples (Ad␤-gal and Ad␤-gal ϩ AS-Myc) showed apoptosis. In contrast, the percentage of apoptosing cells in the CBP-depleted population increased to 15%. At 16 h post-infection, ϳ10% of the cells in control samples showed apoptosis, whereas the percentage of apoptotic cells in CBPdepleted population increased to ϳ25%. Importantly, at both time points, the percentage of apoptotic cells coexpressing antisense CBP-and c-Myc-specific sequences decreased to control levels, indicating that inhibition of c-Myc expression prevented the apoptosis of the CBP-depleted cells. These results were reproducible in two independent experiments. Thus, we conclude that high levels of Myc are primarily responsible for the apoptosis of CBP-depleted cells.
p53 Levels Do Not Increase in p300-or CBP-depleted Cells after Serum Stimulation-Because the p300-and CBP-depleted cells died of apoptosis, we made an attempt to identify the likely candidates that may cause apoptosis in these cells. As p53 levels increase in apoptosing cells in most cases (reviewed in Refs. [47][48][49], we determined the levels of p53 in antisense p300/CBP-expressing cells before and after serum stimulation. Serum-starved antisense-expressing cells were serum-stimulated as described above for the cell cycle experiments and were harvested at 4 h before serum stimulation and at 0, 8 h, and 16 h after stimulation (Fig. 10A). Western immunoblot assays indicated that p53 levels were comparable in control cells before and after serum stimulation. p53 levels in antisenseexpressing cells were comparable with those in control cells before serum stimulation and decreased after serum stimulation (Fig. 10B). Thus, we conclude that apoptosis of serum-stimulated p300-or CBP-depleted cells is not related to p53 levels.

DISCUSSION
We previously showed that p300 plays an important role in the negative regulation of c-myc in the G 1 phase of cell cycle. A decrease in p300 levels in G 1 leads to induction of c-myc and DNA synthesis (26), whereas an increase in p300 levels leads to repression of c-myc and inhibition of DNA synthesis (27). Although p300 and CBP are highly homologous proteins and can substitute for each other in most assays (1), several previous studies indicated that there may be some unique functions that these two proteins provide in cell growth (summarized in the Introduction). These observations prompted us to ask whether CBP also negatively regulates c-myc and whether reduced synthesis of CBP in G 1 would cause the cells to enter S phase prematurely. In this study, we showed that reduced synthesis of CBP also leads to induction of c-myc and that such cells escape G 1 without growth factor stimulation. Thus, although both proteins are highly homologous and can substitute for each other in many functions, they do not appear to be redundant, and normal levels of both p300 and CBP are required to keep c-myc in a repressed state in G 1 .
To gain insight into the mechanism of induction of S phase in antisense-expressing cells, we determined the levels of p21, a general inhibitor of cyclin/Cdk activities (50), and the phosphorylation of three pocket proteins, pRb, p130, and p107, which sequester E2F (51). We showed that p21 levels are reduced in antisense-expressing cells before a rise in c-Myc levels. The reduction of p21 levels might have contributed to the S phase induction, as cyclin-Cdk complexes are not under the influence of p21. Besides, p21 also has been shown to interact with proliferating cell nuclear antigen and to inhibit DNA replication (52). Therefore, decreased levels of p21 in antisense-expressing cells would provide a favorable environment for DNA synthesis. Down-regulation of the p21 genes most likely occurs at the transcriptional level, as we showed previously that p21 RNA levels are reduced in antisense p300-infected cells (26). Because p53 levels did not significantly change in serumstarved antisense-expressing cells (Fig. 10), coactivation of other transcription factors related to p21 gene activation might have been impaired (17). Furthermore, at later time points after antisense virus infection, high levels of Myc that accumulate in the cell also might contribute to repression of the p21 gene (53).
We have shown that in antisense-expressing cells, cyclin D/Cdk activity did not change, whereas cyclin E and A/Cdk activities were elevated to considerable levels. Furthermore, pRb was not phosphorylated, and p107 and p130 were phosphorylated in antisense-expressing cells at low levels. Because pRb is a major player in sequestering E2F, this suggests that only limited amounts of E2F might have been released from the repressor complexes, presumably by the phosphorylation of p107 and p130. Thus, E2F might play a limited role in the induction of S phase. For reasons that we cannot explain at present, even though there was a considerable induction of cyclin E and cyclin A kinases in antisense p300/CBP-expressing cells (Fig. 3G and data not shown) (26), there were negligible levels of pRb phosphorylation.
It is likely that induction of c-myc in p300/CBP-depleted cells is an initial event that leads to the phosphorylation of p107 and p130 and the induction of S phase. This is based on our observation that blocking the induction of c-myc in antisense CBPexpressing cells reversed S phase induction as well as phosphorylation of p107 and p130 (Fig. 4, C and D). These results are in agreement with our previous report in which we showed that inhibiting the induction of c-myc can reverse the induction of S phase in p300-depleted cells (26). Similarly, inhibition of G 1 exit in p300-overexpressing cells as a result of repression of c-myc can be reversed by overexpression of c-myc (27). A number of recent reports suggest the existence of two pathways that promote mitogen-stimulated G 1 -S transition. One mechanism involves induction of c-myc, which then transcriptionally targets several cell cycle-related genes, including cyclin E, cdc25A, and the genes that are responsible for DNA synthesis (reviewed in Refs. 54 -57). In the second well understood mechanism, mitogenic signals inactivate Rb and release E2F, which then activate cyclin E, cdc25A, and other E2F target genes involved in DNA synthesis. Both these pathways operate independently and mutually cooperate in inducing DNA synthesis and coordinated cell cycle progression. The contribution of both these pathways is essential for efficient S phase induction (58 -61). Thus, we believe that in antisense-expressing cells, DNA synthesis is primarily due to the Myc pathway, and insufficient contribution by the Rb/E2F pathway likely results in inefficient S phase induction.
In this study, we have also shown that the antisense-expressing cells that accumulated in S phase in the absence of serum stimulation did not transit into G 2 /M phase. Serum stimulation of these cells also did not promote their transition from S to G 2 /M phase. In contrast, cell cycle progression was normal in cells infected with the Ad␤-gal control. For many reasons, the antisense-expressing cells may have failed to transit from S to G 2 /M phase. For example, both p300 and CBP provide coactivator functions for a number of transcription factors, many of which are directly involved in the activation of genes that control DNA synthesis. These include activator protein-1 (cyclin D 1 ) (23), E2F (cyclins E and A) (62), and activating transcription factor (cyclin A) (63). These coactivators also acetylate nucleosomes of transcriptionally active promoters and stimulate transcription. The histone acetyltransferase activity of these proteins is also known to acetylate E2F and thereby increase its DNA binding and transcriptional activation activities (3). Other studies have suggested that the histone acetyltransferase activity of CBP increases during the G 1 -S transition (64). Therefore, lack of p300 or CBP, combined with lack of a contribution by the Rb pathway discussed above, would result in incomplete DNA synthesis, and such cells would not transit to G 2 /M phase.
We have shown that a significant number of serum-starved cells expressing antisense p300/CBP sequences underwent apoptosis upon serum stimulation. At 16 h after serum stimulation, the number of apoptotic cells increased by ϳ2-fold for antisense p300-or CBP-expressing cells compared with Ad␤gal-expressing cells (Fig. 6). We provided genetic evidence that apoptosis of the CBP-depleted cells could be reversed if expression of c-myc is blocked by antisense c-Myc sequences (Fig. 9). This shows that c-Myc plays an important role in the induction of apoptosis in CBP-depleted cells. A number of studies have shown that high levels of c-Myc can cause apoptosis (reviewed in Refs. 57, 65, and 66). In serum-starved antisense p300-and CBP-expressing cells, c-myc was induced at considerable levels. This induced c-myc was further induced upon serum stimulation ( Fig. 7 and 8). For example, at 4 h after serum stimulation, the AS-p300-and AS-CBP-infected cells contained ϳ3-fold higher levels of c-Myc activity compared with serum-stimulated control cells undergoing normal cell cycle progression (Fig. 8, B and C). It is conceivable that the inappropriate induction of c-myc and the unscheduled DNA synthesis that occur in serum-starved cells prime these cells for apoptosis. Further induction of c-myc by serum and perhaps other yet to be identified factors contribute to apoptosis. This interpretation is consistent with the view that induction of apoptosis by E1A, Myc, or E2F is the result of a conflict between the growthpromoting action of the oncoprotein and simultaneous growth inhibitory signals such as low serum (67,68). We believe that p53 does not play a role in this apoptosis, as the levels of p53 in antisense-expressing cells dropped beginning at 4 -6 h after serum stimulation (Fig. 10B). This is consistent with published studies suggesting that c-myc can induce apoptosis in certain cell types by both p53-dependent and p53-independent mechanisms (69 -71). Other studies have shown that during mycinduced apoptosis, Myc targets the pro-apoptotic gene bax, which mediates apoptosis (72,73). Western blot experiments did not detect a significant increase in the levels of Bax protein in antisense-expressing cells after serum stimulation (data not shown). Thus, at the present time, we do not know the Myc target genes in antisense-expressing cells that may be responsible for apoptosis.
In summary, normal levels of both p300 and CBP are essential to keep c-myc in a repressed state in G 1 and to prevent inappropriate DNA synthesis. Similarly, normal levels of both these proteins are also needed in S phase. Previously knockout mice studies showed that embryos lacking p300 are significantly smaller than control embryos (5). In vivo bromodeoxyuridine labeling studies showed a striking reduction of the number of cells in S phase in the embryos of these mice (5). Fibroblasts derived from p300 Ϫ/Ϫ mice also grow slowly and cease to divide after a few generations (5). These data indicate a defect in cell cycle progression in these mice and are in agreement with the data presented here.