Abstract
The histone acetyltransferases p300 and cAMP-responsive element-binding protein-binding protein (CBP) are required for the execution of critical biological functions such as proliferation, differentiation, and apoptosis. Both proteins are believed to regulate the activity of a large number of general and cell-specific transcription factors. Here we demonstrate a dramatic decrease in the total cellular levels of p300 and CBP with increasing population doublings of human normal melanocytes. We show that one consequence of p300 depletion is transcriptional down-regulation of the cyclin E gene, caused by deacetylation of histones at its promoter. The cyclin E promoter was activated by p300 and the histone deacetylase inhibitor trichostatin A. Conversely, the cyclin E promoter was repressed by wild-type Retinoblastoma tumor suppressor p105 protein (pRB) and by a dominant negative p300 mutant (DN p300) that lacks histone acetyltransferase activity. We also provide evidence of the alternative recruitment of p300 and histone deacetylase 1 to the cyclin E promoter in proliferating and senescent melanocytes, respectively. The biological significance of these results was established by showing that block of p300 activity by overexpression of DN p300 or by Lys-CoA, a specific chemical inhibitor of p300, resulted in growth inhibition, down-regulation of cyclin E, and activation of the senescence-associated β-galactosidase marker in human melanocytes and melanoma cells. Together, these results provide evidence for the essential role of p300 in the regulation of proliferation and senescence in cells from melanocytic origin.
INTRODUCTION
Cellular senescence is widely thought to be a tumor-suppressive mechanism and has also been proposed to contribute to aging (1). Senescent cells exhibit multiple changes in gene expression, including down-regulation of cell cycle-regulatory and stress response genes and up-regulation of matrix-remodeling proteins (reviewed in Ref. 1). This complex phenotype suggests that high order changes in chromatin structure may be involved in the generation of the senescent phenotype. Experiments to support this hypothesis include the findings that the HDAC3 inhibitor sodium butyrate rapidly induces a senescent-like phenotype in human fibroblasts (2) and that reduced levels of Sir2, a protein with HDAC activity, causes accelerated aging in yeast cells (3). In the presence of HDAC inhibitors, fibroblasts enter a state of senescence after several PDs (2). This state resembles replicative senescence in terms of morphology, saturation density, and cell cycle distribution, including the accumulation of hypophosphorylated RB and induction of the CDKI p21Waf-1 (4). In yeast cells, overexpression of Sir2 extends life span in a manner dependent on nicotinamide-adenine dinucleotide-dependent HDAC activity (reviewed in Ref. 5). It is presently unknown whether mouse or human Sir2 homologues have similar activity.
The chromosome structure is dynamic and can undergo extensive remodeling, leading to activation or repression of transcription. Chromatin remodeling is regulated by at least two highly conserved mechanisms, posttranslational modification of histone tails and ATP-dependent reorganization of nucleosome positioning. p300 and CBP are highly homologous global transcriptional coactivators that acetylate nucleosomal histones (reviewed in Ref. 6; see Refs. 7 and 8). Conversely, deacetylation of histone tails is achieved by HDACs, several of which have been identified (9), and, in general, results in transcriptional repression.
The complexity of the protein-protein interactions in which p300 and CBP participate has led to the idea that they are molecular integrators of the transcriptional control of cell cycle progression, differentiation, and tumorigenesis (reviewed in Refs. 10, 11, 12, 13, 14). Both p300 and CBP are targeted by viral oncoproteins, mutated in certain forms of cancer, and phosphorylated in a cell cycle-dependent manner (reviewed in Ref. 12). In addition, p300 and CBP interact with numerous transcription factors including p53, E2F, micropthalmia transcription factor (MITF), and RB (15) and form a complex with cyclin E-CDK2 (16). Recent evidence indicates that the HAT activities of p300 and CBP are required for the G1-S transition (17). p300 and CBP can stimulate either transactivation or repressor functions of certain transcription factors, suggesting that the interacting proteins, promoter, and cellular context are critical determinants of p300/CBP function (reviewed in Refs. 13 and 18). Repression has been proposed to result, at least in some cases, from competition between transcription factors for limited amounts of CBP and p300 in the nucleus (13). Unexpected insights into the functions of CBP and p300 have come from gene knockout studies in mice. The CBP−/− and p300−/− embryos die between days 9 and 11.5. Their main defects are severe developmental retardation, reduced size, failed neural tube closure, and altered cardiac ventricular trabeculation (reviewed in Ref. 13). Importantly, p300-deficient fibroblasts show slow proliferation and rapid senescence, suggesting that p300 is involved in maintaining the proliferative state of cells (19).
Sustained activation of the cAMP pathway is required for optimal melanocyte proliferation (reviewed in Ref. 20). However, cAMP also activates the melanogenic pathway, resulting in up-regulation of melanin synthesis. Melanogenesis correlates with withdrawal from the cell cycle at ∼13–15 PDs and acquisition of a senescent phenotype characterized by large, flat pigmented cells and expression of the SA-β-Gal marker (reviewed in Ref. 21; see Ref. 22). We have shown that the replicative senescence of human melanocytes is accompanied by up-regulation of the CDKI p16INK4a, down-regulation of cyclin E, down-regulation of CDK4 and CDK2 activities, and hypophosphorylation of the RB protein (reviewed in Ref. 21). In contrast to fibroblasts, however, the CDKIs p21Waf-1 and p27Kip-1 are also down-regulated. These changes appear to be important for replicative senescence because they do not occur in melanocytes that overexpress the catalytic subunit of the enzyme telomerase (hTERT) or in melanomas, which are tumors that originate from melanocytes or melanoblasts.
Here we show that as a consequence of p300 down-regulation, the cyclin E gene is transcriptionally silenced due to loss of histone acetylation at its promoter. We also provide evidence for the alternative recruitment of p300 and HDAC1 to the cyclin E promoter in proliferating and senescent melanocytes, respectively. We propose that p300/CBP insufficiency triggers a senescence checkpoint by permanently relocating genes in the heterochromatin compartment.
MATERIALS AND METHODS
Melanocyte Cultures.
Neonatal human melanocytes were isolated and cultured in a manner similar to that described previously (23, 24). The culture medium consists of MCDB-153 medium supplemented with 1 ng/ml recombinant human basic fibroblast growth factor, 5 μg/ml insulin, 50 μg/ml transferrin, 4% fetal bovine serum (Sigma, St. Louis, MO), 6.5 μg/ml pituitary extract (BioWhittaker, Walkersville, MD), 10 nm cholera toxin, and 0.1 mm 3-isobutyl-1-methylxanthine.
Permeabilization of Melanocytes in Vivo.
Proliferating human normal melanocytes (2 × 106) were trypsinized and suspended in 1 ml of ICB. [10 mm HEPES (pH 7.0), 0.14 m KCl, 0.01 m NaCl, and 2.4 mm MgCl2]. Twenty-five μl of an ICB solution containing 0.5% trypan blue ± Lys-CoA were added to 250 μl of cell aliquots, followed by 25 μl of transport reagent (1.2 mg/ml sphingosylphosphorylcholine) and incubation at 37°C. After 10 min, 50 μl of stop reagent (20% fatty acid-free BSA in ICB) were added to each tube. Cells were seeded in 24-well dishes containing coverslips and supplemented with 750 μl of melanocyte culture medium. This medium was changed after 1 h to dilute chemicals. The cells were fed with fresh medium the next day and fixed after 60 h. All reagents were from Sigma.
Antibodies and Other Reagents.
Anti-pRB, anti-p300, and anti-E2F-1 mouse monoclonal were purchased from Labvision Inc. (Fremont, CA). Anti-HDAC1, anti-acetylated histone H4, anti-histone H4, anti-CBP, and anti-cyclin E rabbit polyclonal antibodies were from Upstate Biotechnology, Inc. (Lake Placid, NY). Anti-β-actin monoclonal Ab was from Sigma. All other Abs were from Santa Cruz Biotechnology (Santa Cruz, CA). Abs were used at a dilution recommended by the manufacturer.
Preparation of Cell Extracts.
Whole cell extracts were prepared from melanocytes by adding twice the cell pellet’s volume of lysis buffer [50 mm Tris-Cl (pH 7.5), 0.4% NP40, 120 mm NaCl, 1.5 mm MgCl2, 2 mm phenylmethylsulfonyl fluoride, 80 μg/ml leupeptin, 3 mm NaF, and 1 mm DTT] and incubated in ice for 30 min with occasional vortexing. Lysates were collected after centrifugation at 14,000 × g for 10 min. The resulting supernatants were used for immunoblotting and immunoprecipitation.
Northern Blotting.
Total RNAs were extracted from the cells using the RNeasy kit (Qiagen). Three μg of total RNA were separated in 1% agarose gel and transferred to Hybond N+ membrane (Amersham). The membrane was probed with a 32P-labeled, 900-bp BamHI-SstI fragment from the cyclin E coding region. The blot was washed three times in high stringency washing buffer for 15 min each time and exposed to Kodak Biomax MR autoradiography film at −80°C.
Immunofluorescence.
Sixteen h after transfection, the cells were fed with fresh medium and incubated for 48 h. Afterward, the cells were washed twice with PBS and fixed in 4% formaldehyde for 1 h, permeabilized with 0.5% Triton X-100 for 30 min at room temperature, and blocked with TBST (20 mm Tris, 150 mm NaCl, and 0.2% Tween 20) + 5% nonfat dry milk at 37°C for 1 h. The cells were then incubated with a 1:30 dilution of a polyclonal anti-cyclin E Ab (Upstate Biotechnology) or with a monoclonal p300 Ab (Lab Vision Inc.) at 37°C for 1 h, followed by three washes with TBST and incubation with an antirabbit secondary Ab conjugated with tetramethylrhodamine isothiocyanate or with an antimouse Ab conjugated with FITC (Santa Cruz Biotechnology) at 37°C for 1 h. After a final wash, the slides were mounted with anti-fade mounting media (Vector Laboratories) and visualized under a fluorescence microscope.
Immunoblotting, Immunoprecipitation, and Expression of SA-β-Gal.
Equal amounts (50 μg) of whole cell lysates were electrophoresed in SDS-polyacrylamide gel and transferred to nitrocellulose membranes. Membranes were washed with TBS, blocked with 5% milk in TBS for 1 h at room temperature, and incubated overnight at 4°C in TBS containing appropriate dilutions of primary Abs. The membranes were then washed three times and incubated with a horseradish peroxidase-conjugated donkey antirabbit or sheep antimouse IgG Ab (1:3,000) in TBS for 1 h at room temperature. After washing, the target proteins were detected using the enhanced chemiluminescence immunoblotting detection kit (Amersham). For immunoprecipitation, 600 μg of total cell extracts were immunoprecipitated with an anti-RB Ab at 4°C for 3 h. Protein G-Sepharose beads were used to immobilize the Ab complex. The Sepharose-bound immunocomplexes were washed three times with 10 volumes of lysis buffer, boiled in SDS-Laemmli sample buffer, and electrophoresed in SDS-PAGE gels. SA-β-Gal activity was determined as described previously (25).
ChIP Assay.
We used a ChIP assay kit from Upstate Biotechnology. Briefly, 2 × 107 cells (either proliferative or senescent) were cross-linked with 1% formaldehyde for 10 min at 37°C. The cells were then washed in PBS, harvested, and lysed in 1% SDS +10 mm EDTA (pH 8.0) + 50 mm Tris-Cl (pH 8.0). The lysates were sonicated to reduce the DNA length to 200-2000 bp and diluted 10 times with ChIP dilution buffer [0.01% SDS, 1.1% Triton X-100, 1.2 mm EDTA (pH 8.0), 16.7 mm Tris-Cl (pH 8.0), and 167 mm NaCl]. One percent of the diluted lysate was used as total input. The remaining lysates were precleared with 80 μl of salmon sperm DNA/protein A-agarose slurry and immunoprecipitated with 5 μl of anti-acetylated histone H4 Ab or anti-histone H4 overnight at 4°C. An anti-actin Ab was used as a negative control. The immunocomplexes were collected using salmon sperm DNA/protein A-agarose slurry at 4°C for 1 h. The immobilized complexes were washed and then eluted with 1% SDS and 0.1 m NaHCO3. Cross-links were reversed, and chromatin-associated proteins were digested with 2 μl of proteinase K (10 mg/ml). The immunoprecipitated DNA was recovered by ethanol precipitation and subjected to PCR. PCR primers for the cyclin E promoter region were 5′-GGCGGGACGGGCTCTGGG-3′ and 5′-CCTCGGCATGATGGGGCTG-3′. This primer combination was used to amplify the −97 to +601 region of the cyclin E promoter. PCR amplification was performed using Taq DNA polymerase from Life Technologies, Inc. A total of 20 μCi of [32P]dCTP was used to label the PCR products. The PCR products were resolved in 2% agarose gel and exposed to Kodak Biomax MR film.
Histone Acetylation.
Histones were isolated from melanocytes by HCl extraction and acetone precipitation as described previously (26). Isolated histones were separated in 15% polyacrylamide gel containing 8 m urea, 5% acetic acid, and 0.375% Triton X-100; transferred to nitrocellulose; and immunoblotted with anti-acetylated histone H4 Ab (Upstate Biotechnology).
HAT Assay.
HAT assays were performed using the HAT-check kit (Pierce). A synthetic peptide corresponding to the first 23 amino acids of histone H4 (SGRGKGGKGLGKGGAKRHRKVLR), coupled through a GSSG linker sequence to a biotin molecule, was used as HAT substrate. This substrate was incubated with 30 μg of nuclear extracts from proliferative, senescent, and hTERT-overexpressing melanocytes in the presence of [14C]acetyl-CoA at 30°C for 1 h. The radioactively labeled, biotin-conjugated substrate was immobilized on streptavidin-agarose beads and washed, and 14C was counted.
HDAC Assay.
pRB immunoprecipitated complexes were used to determine pRB-associated HDAC activity (HDAC assay kit; Upstate Biotechnology). A SGRGKGGKGLGKGGAKRHRKVLR peptide substrate was prelabeled with [3H]sodium acetate using benzotriazol-1-yloxytris-(dimethylamino)-phosphonium hexafluoruphosphate. The immunoprecipitates were incubated with the labeled peptide overnight at room temperature. Radioactivity released by the labeled peptide (HDAC activity) was measured after ethyl-acetate extraction.
Plasmids.
A 2.1-kb SacI-SacI genomic DNA fragment encompassing the promoter region of cyclin E (27) was subcloned in SacI sites of pGL3 to construct a cyclin E-luciferase plasmid. The HDAC1 expression plasmid was constructed by subcloning a BglII-XhoI fragment from pOZ-HDAC1 (28) into the BamHI-XhoI site of pPCDNA3.1hygro. The p3001572–1903 plasmid was provided by N. La Thangue. RB expression plasmids (RB-WT and RBN757F) were provided by T. T. Chen and J. Y. J. Wang. A human E2F-1 expression plasmid was provided by J. Campisi.
Luciferase Assay.
Proliferating normal human melanocytes were transfected using FuGENE 6 reagent, as specified by the manufacturer (Boehringer Mannheim). Transfections used 0.5 μg of pGL-3 or cyclin E promoter-luciferase plasmids and combinations of 0.5 μg of E2F-1, pRB, or pRBN757F expression plasmids, together with a β-galactosidase expression plasmid, for normalization of transfection efficiencies. Luciferase and β-galactosidase activity were measured 48 h after transfection using the luciferase assay system kit (Promega). When indicated, TSA was added 5 h after transfections and incubated for 31 h before performing the luciferase assay.
RESULTS
Down-Regulation of p300/CBP Protein Levels and HAT Activity in Melanocytes undergoing Increased PDs.
To determine whether the HATs p300 and CBP are important for senescence of human normal melanocytes, we measured the levels of these proteins in lysates from cells that underwent an increasing number of PDs. Both proteins declined markedly as the cells underwent replicative senescence (Fig. 1,A). p300 can be unstable under some conditions used to make cell extracts. Hence, we confirmed p300 down-regulation by immunofluorescence (Fig. 1,B). It is likely that p300 down-regulation is senescence specific because early-passage, growth factor-deprived quiescent melanocytes and proliferating, late-passage “telomerized” melanocytes (hTERT) continued to express considerable amounts of p300 (Fig. 1 C). The p300 Ab consistently detected two bands in extracts from quiescent melanocytes, which may correspond to phosphorylated and underphosphorylated p300 or p300 proteolysis. However, additional analysis will be necessary to interpret these results because in contrast to CBP, p300 phosphorylation or proteolysis has not been studied in detail.
To determine whether loss of p300 and CBP during replicative senescence altered the levels of histone H4 acetylation, we precipitated total cellular histones by acid extraction (“Materials and Methods”), followed by electrophoresis in polyacrylamide gels containing 8 m urea, transfer to nitrocellulose membranes, and immunoblotting with an anti-acetylated H4 Ab. Histone H4 isolated from senescent melanocytes showed a dramatic decrease in acetylation levels, compared with proliferating or telomerized melanocytes (Fig. 1,D, top panel). These results were further confirmed by conventional PAGE and immunoblotting with an anti-acetylated H4 Ab (Fig. 1,D, middle panel). Importantly, the total levels of histone H4 remained essentially unchanged among proliferating, senescent, and telomerized melanocytes (Fig. 1,D, bottom panel). Consistent with the loss of histone H4 acetylation, down-regulation of p300 and CBP coincided with a 50% reduction in the levels of total cellular HAT activity in senescent melanocytes (Fig. 1,E). The telomerized melanocytes, which proliferate more slowly than unmodified control cells from the same donor (21), exhibited intermediate HAT activity (Fig. 1 E). Together, these results suggest that down-regulation of p300/CBP may cause altered activity of essential cell cycle-regulatory genes and consequently result in cell cycle arrest and senescence.
Down-Regulation of p300/CBP Correlates with Cyclin E Transcriptional Repression and Loss of Acetylated Histones at Its Promoter.
Cyclin E is a critical regulator of senescence because its overexpression is sufficient to trigger escape from RAS- and BRG1-induced senescence (29, 30). Given that p300/CBP regulates expression of many cell cycle genes, we examined whether its down-regulation could result in the transcriptional repression of the cyclin E gene. We found that a decline in cyclin E protein levels paralleled mRNA levels (Fig. 2, A and B). In contrast to cyclin E, cyclin D1 was modestly down-regulated at comparable PDs (Ref. 23; Fig. 2 A).
The cyclin E promoter in human tumor cells and mouse embryonic fibroblasts is regulated by acetylation/deacetylation of histone tails (15, 31, 32). Hence, we asked whether transcriptional repression of cyclin E could result from loss of acetylated chromatin at its promoter. For this purpose, we performed ChIP assays with Abs to acetylated histone H4 or total histone H4 (33). PCR primers were designed to span the region between 1098 and 1795 bp of the cyclin E promoter (27), which includes three E2F consensus binding sites (Fig. 2,C). This region was shown to be an active cyclin E promoter, containing a full complement of upstream sequences needed for regulation throughout the cell cycle (27). ChIP assays demonstrated that acetylated histone H4 was abundant in the cyclin E promoter in proliferating melanocytes. By contrast, this region of the promoter was essentially devoid of acetylated H4 in senescent melanocytes, despite total levels of histone H4 that were essentially identical between proliferating and senescent melanocytes (Fig. 2 D). Our results suggest that inefficient histone H4 acetylation may result in condensed heterochromatin and consequent transcriptional silencing of the cyclin E gene.
A Dominant Negative p3001572–1903 Protein Represses the Cyclin E Promoter and Reduces E2F-1-associated HAT Activity in Vivo.
To investigate in more detail how the cyclin E promoter is regulated by histone acetylation, we constructed a human cyclin E promoter luciferase reporter. We transiently transfected normal human melanocytes with this vector, alone or together with expression vectors encoding E2F-1, a WT p300 protein, and a dominant negative p300 mutant (p3001572–1903) that lacks HAT activity but retains the ability to interact with E2F (34). p3001572–1903 apparently competes with endogenous p300/CBP proteins for the E2F-5 domain. Because E2F-1 interacts with the third zinc finger domain of p300/CBP (12), it is likely that p3001572–1903 could also interfere with its activity. As expected from previous results (34), we found that E2F-1 and p300 strongly activated the cyclin E promoter (Fig. 3,A). However, p3001572–1903 prevented E2F-1-mediated activation (Fig. 3,B). Furthermore, p3001572–1903 repressed, in a dose-dependent manner, E2F-1 and p300 activation of the cyclin E promoter (Fig. 3,C). This suggests that p3001572–1903 also competes with endogenous p300/CBP proteins for the E2F-1 domain. Such competition should result in reduced HAT activity associated with E2F-1. To investigate such a possibility, we transfected 293 cells with WT p300 and p3001572–1903 vectors. These cells were selected because of their high transfection efficiency (∼65%) compared with normal human melanocytes (∼10%). We found that whereas total levels of HAT activity remained essentially unchanged 2 days after transfection, E2F-1-associated HAT was dramatically reduced in cells transfected with p3001572–1903 (Fig. 3 D). These results appeared to be specific because cotransfection of p3001572–1903 with WT p300 restored and augmented E2F-1-associated HAT activity. We conclude that p300/CBP is a critical regulator of cyclin E transcription in vivo.
p3001572–1903 Down-Regulates Cyclin E Expression and Induces a Senescent Phenotype in Human Melanoma Cells.
To link our findings on p300/CBP down-regulation and replicative senescence, we transiently transfected p3001572–1903 in human melanoma cells. We used these cells instead of human normal melanocytes because expression of cyclin E in melanoma cells is not restricted to the G1-S window (35, 36). Constitutive cyclin E expression was necessary to avoid false positive results in cotransfection experiments with a GFP and p3001572–1903 expression plasmids (“Materials and Methods”). Melanoma cells were cotransfected with an empty vector (plasmid control) or p3001572–1903 together with a GFP-expressing plasmid at a ratio of 10:1. Cell cycle profiles of GFP-sorted cells showed a dramatic reduction (48% versus 30%) in the number of S-phase cells 48 h after transfection (Fig. 4 A).
Immunofluorescence of individual cells also showed a dramatic reduction of cyclin E protein in GFP-positive cells that received the p3001572–1903 plasmid (Fig. 4,B), compared with GFP-positive cells that received the control vector. The effect of p3001572–1903 on cyclin E down-regulation is specific because cotransfection with a WT p300 plasmid at a 1:1 ratio, rescued cyclin E expression (Fig. 4,B). Expression of p300 alone did not have any effect on cyclin E expression. We noticed that p3001572–1903 induced a flat cell phenotype in cells (Fig. 4,B). Furthermore, some GFP-positive, cyclin E-negative cells showed a very large and flat cell phenotype (Fig. 4,C, arrowheads mark cell boundaries). Such morphological changes have been associated with cells that have reached replicative senescence (37). To determine whether p3001572–1903 induced a replicative senescent phenotype, the transfected cells were analyzed for the presence of the SA-β-Gal marker (25). Cultures that received the p3001572–1903 plasmid showed enlarged, SA-β-Gal-positive cells (Fig. 5,A), compared with cells that received the vector control. The senescent, SA-β-Gal-positive cell phenotype could be rescued by coexpression with a WT p300 plasmid (Fig. 5 A). p300 alone did not have any effect on SA-β-Gal expression. Together, these results suggest that down-regulation of p300 levels or competition by a dominant negative p300 mutant caused growth arrest and expression of a senescent-like phenotype in melanocytes and melanoma cells.
Lys-CoA Induces a Senescent Phenotype in Human Normal Melanocytes.
We also used an alternative method to demonstrate that lack of p300 protein and/or activity regulates senescence in human melanocytes. Lys-CoA was recently shown to be a potent p300 HAT inhibitor (38). Lys-CoA has a slow onset of inhibition, therefore, p300 HAT-dependent processes may not shut off immediately after addition of this inhibitor. Importantly, Lys-CoA causes minimal inhibition of p300/CBP-associated factor, a HAT not related to p300/CBP (39). Normal human melanocytes were permeabilized in the presence or absence of Lys-CoA, seeded in coverslips, and monitored 60 h later. Permeabilization per se did not substantially alter viability or proliferation because in the absence of Lys-CoA, we observed high levels of BrdUrd-positive melanocytes and background levels of SA-β-Gal (Fig. 5,B). However, Lys-CoA almost completely abolished proliferation, induced high levels of SA-β-Gal, and inhibited cyclin E expression in a dose-dependent manner (Fig. 5 B).
The Cyclin E Promoter Is Repressed by RB and HDACs.
Competition of HATs and HDACs for E2F-1 regulates cyclin E promoter activity (40). We observed that basal and E2F-1-induced activation of the cyclin E promoter was notably increased by TSA in a dose-dependent manner (Fig. 6, A and B).
Several E2F-responsive genes are known to be repressed by RB family members through the recruitment of HDACs (41). HDAC1 preferentially binds to the repressive, hypophosphorylated form of RB. This interaction appears to be direct and mediated by the sequence motif LXCXE (reviewed in Ref. 42). We found that E2F activity on the cyclin E promoter was counteracted by coexpression of WT RB and, to a lesser extent, by RBN757F, a RB mutant deficient in HDAC binding (Fig. 6 C). RB-dependent repression of the cyclin E promoter was sensitive to TSA, a HDAC inhibitor. The fact that TSA stimulated cyclin E transcription in the presence of RBN757F suggests that endogenous RB-HDAC complexes also compete for binding to this promoter.
Increased Association of HDAC1 to RB in Senescent Melanocytes Correlates with HDAC1 Recruitment to the Cyclin E Promoter.
To investigate the role of HDAC1 and RB in melanocyte senescence, we immunoprecipitated RB complexes from proliferating and senescent cells. We found increased levels of HDAC1 bound to unphosphorylated RB in senescent compared with proliferating melanocytes (Fig. 7,A) and a consequent increase in HDAC activity of the RB immunoprecipitates (Fig. 7,B). A decrease in p300/CBP and recruitment of HDACs to the cyclin E promoter might be necessary to maintain the irreversibility of the senescent state. To analyze the possibility of alternative recruitment of p300 and HDAC1 onto the cyclin E promoter, we used ChIP assays with Abs against p300 and HDAC1. In proliferating melanocytes, p300 was relatively abundant in the region between 1098 and 1795 bp of the cyclin E promoter (Fig. 7,C), which showed high levels of acetylated histone H4 (Fig. 2,D). By contrast, this region of the promoter was essentially devoid of p300 in senescent cells. Conversely, whereas no significant amounts of HDAC1 were detected in the cyclin E promoter in proliferating melanocytes, HDAC1 association increased in the senescent cells (Fig. 7 C). We observed that a larger fraction of the input was retained with anti-histones (positive control) than with anti-p300 or anti-HDAC1 Abs. As was noted by others, p300 and HDAC do not directly bind to DNA; therefore, detection of these molecules depends on the efficiency of the cross-linking technique (39).
To determine the biological significance of these results, we overexpressed a human HDAC1 cDNA in human melanomas. The percentage of transfected cells ranged between 50% and 60%, as estimated in parallel infections using the GFP. The vectors coexpressed a hygromycin resistance marker that allowed selection for pure populations of transfected cells within 4 days. Expression of ∼2–3 times the levels of basal HDAC1 (Fig. 7,E) produced dramatic changes in human melanoma cells. By day 12, virtually all cells had acquired an enlarged morphology, become BrdUrd negative, and expressed the SA-β-Gal marker (Fig. 7 D). In contrast, high levels of BrdUrd incorporation, insignificant SA-β-Gal positivity, and no changes in cell morphology were observed in melanoma cell populations transfected with the empty vector (pcDNA3.1). Together, our results suggest that to maintain cellular homeostasis, HAT and HDAC protein levels and activity must be tightly regulated, and disruption of this regulation results in growth arrest and senescence.
DISCUSSION
Our results establish a fundamental role for p300 and histone acetylation in the maintenance of the proliferative state of human normal melanocytes. Our findings are consistent with studies using fibroblasts from animals nullizygous for p300, which show rapid replicative senescence in culture, and with the requirement of p300/CBP for the G1-S transition and hence for cell proliferation (17).
p300/CBP appears to have all of the functions of a master regulatory protein and molecular switch because it is required for cell and tissue function during embryonic development, cell differentiation in vitro (39), and DNA repair (43). Although very little is known regarding p300 expression and function in terminally differentiated cells and mature tissues, p300 may also be required to maintain tissue homeostasis. Supporting this concept, p300 depletion in cardiocytes impairs cell-specific gene transcription (44). Thus, it is likely that reduction in p300 levels will have deleterious effects not only for cellular proliferation but also for tissue-specific gene activity.
We focused on understanding the consequences of p300 depletion on the regulation of cyclin E expression. Cyclin E is perhaps the most fascinating of the known cyclins. In association with CDK2, cyclin E controls three major S-phase events: (a) DNA replication; (b) centrosome duplication; and (c) histone gene expression (reviewed in Ref. 45). Overexpression of cyclin E promotes S-phase entry, increases the frequency of centrosome duplication and genetic instability (45), and induces escape from Ras-induced senescence in mouse embryonic fibroblasts (29). Conversely, a decrease in cyclin E-CDK2 activity results in inhibition of the G1- to S-phase transition. Our results indicate that cyclin E down-regulation appears to be a direct consequence of p300/CBP depletion (Figs. 2 and 7). To our knowledge, this is the first example of physiological p300 depletion in human cells and its direct effect on the transcription of a cell cycle regulatory protein.
It has been proposed that errors in maintenance of repressive heterochromatin domains accumulate during the proliferative life span of normal human cells, ultimately triggering a senescence checkpoint and consequent irreversible cell cycle exit (4). Intriguingly, HDAC inhibitors also induce a senescent-like phenotype in human fibroblasts (2). How can these apparently paradoxical results be reconciled with our findings? A likely explanation could be the necessity of the cells to maintain a critical balance between acetylated and deacetylated chromatin domains to sustain proliferation. Whereas hyperacetylation caused by loss of HDAC activity could activate antiproliferative genes in chromatin regions (4), loss of acetylation may shift the balance toward repressive heterochromatin, causing silencing of genes associated with cell cycle progression.
Local perturbations of chromatin structure could specifically alter the accessibility and/or function of transcriptional regulatory proteins that bind DNA sequences in the region where histone acetylation or deacetylation occurs (46). We postulate that chromatin modifications resulting from p300 down-regulation and a consequent increase in HDAC activity cause senescence in pigment cells. Whereas in proliferating cells, recruitment of p300/CBP and HDACs to promoters may be a cyclic event, a decrease in p300 levels could favor HDAC1 recruitment (Fig. 7,C), which in turn may drive the recruitment of other repressor proteins such as MeCP2 (reviewed in Refs. 47 and 48) and SUV39H1/HP1 (49, 50) to cyclin E and other cell cycle-regulatory gene promoters (Fig. 8). Considering that even a 25% decrease in p300/CBP proteins is detrimental for normal development (14, 19), a progressive decline in HAT levels with increased PDs could lead to altered gene regulation and finally to the activation of a senescent checkpoint by permanently relocating genes in the heterochromatin compartment.
HAT activity is required for p53-mediated transactivation and senescence induced by Ras (51). One immediate implication of p300 down-regulation is that it could potentially alter p53 acetylation levels and, consequently, cell cycle checkpoints regulated by this protein. Paradoxically, both p534 and p21Waf-1 are almost completely down-regulated in melanocytes undergoing extensive telomere attrition and replicative senescence. The promyelocytic leukemia protein interacts with CBP, and its recruitment to the NH2 terminus of p53 might lead to COOH terminus acetylation and stabilization by preventing MDM2-mediated ubiquitination (reviewed in Ref. 52). Thus, it is conceivable that p300/CBP down-regulation causes decreased p53 protein stability. Although the role of p53 in melanocyte senescence needs to be explored in more detail, the above-mentioned findings raise the question as to whether this tumor suppressor is dispensable for melanocyte senescence. Evidence in transgenic animals overexpressing a stable, “hyperactive” p53+/m mutant supports this possibility (53). The p53+/m mice retain one full-length allele and one that harbors a deletion of the first six exons of p53, resulting in a shortened protein that apparently stabilizes the WT p53 protein. Although these animals show accelerated aging in several tissues, pigmentation in hair (dermal melanocytes) or skin (intraepidermal melanocytes) does not show appreciable changes. This indicates that the p53+/m melanocytes are still fully functional and proliferative because melanocyte migration, proliferation, and differentiation are required during cyclic hair follicle regeneration (54, 55).
Finally, our results could have implications for novel melanoma therapies. Although it has been proposed that p300/CBP is a tumor suppressor (reviewed in Ref. 14), we have shown here that reduction of p300 protein levels and/or activity down-regulated cyclin E obliterated melanoma proliferation and induced a senescent phenotype. These results are in concordance with other p300/CBP functions, including stimulation of G1-S transition (17) and activation of E2F-regulated promoters (34). Together, our results suggest that p300/CBP could be a potential target for novel approaches in the treatment of melanoma.
Down-regulation of p300/CBP protein and HAT activity with increased PDs. A, p300 and CBP protein levels at different PDs determined by Western immunoblotting. β-Actin was used as a control for protein loading. B, immunofluorescence analysis of p300 protein in proliferating (top panel) and senescent (bottom panel) melanocytes. C, p300 levels in quiescent (Q) and hTERT melanocytes (immunoblot). D, levels of acetylated histone H4 in protein extracts from proliferating (PD 3), senescent (PD 13), and hTERT melanocytes. Blots from polyacrylamide gels containing 8 m urea (top panel) or PAGE gels (middle panel) were probed with an anti-acetylated histone H4 Ab. Total levels of histone H4 (bottom panel) were used as controls. E, HAT activity in extracts from proliferating (PD 3), senescent (PD 13), and hTERT melanocytes. The mean ± SE of three independent experiments is shown. Number in brackets indicates the percentage of BrdUrd-positive melanocytes (14). hTERT are normal human melanocytes expressing the catalytic subunit of telomerase. The hTERT melanocytes have an extended life span (>60 PDs) compared with nontelomerized melanocytes (12–14 PDs; Ref. 5).
Down-regulation of p300/CBP protein and HAT activity with increased PDs. A, p300 and CBP protein levels at different PDs determined by Western immunoblotting. β-Actin was used as a control for protein loading. B, immunofluorescence analysis of p300 protein in proliferating (top panel) and senescent (bottom panel) melanocytes. C, p300 levels in quiescent (Q) and hTERT melanocytes (immunoblot). D, levels of acetylated histone H4 in protein extracts from proliferating (PD 3), senescent (PD 13), and hTERT melanocytes. Blots from polyacrylamide gels containing 8 m urea (top panel) or PAGE gels (middle panel) were probed with an anti-acetylated histone H4 Ab. Total levels of histone H4 (bottom panel) were used as controls. E, HAT activity in extracts from proliferating (PD 3), senescent (PD 13), and hTERT melanocytes. The mean ± SE of three independent experiments is shown. Number in brackets indicates the percentage of BrdUrd-positive melanocytes (14). hTERT are normal human melanocytes expressing the catalytic subunit of telomerase. The hTERT melanocytes have an extended life span (>60 PDs) compared with nontelomerized melanocytes (12–14 PDs; Ref. 5).
Transcriptional silencing of the cyclin E gene by promoter deacetylation. A, cyclin E and D1 protein levels at different PDs determined by Western immunoblotting. B, cyclin E mRNA levels decrease with increased PDs. Equal amounts of RNA were loaded and hybridized using a cyclin E probe (“Materials and Methods”). A glyceraldehyde-3-phosphate dehydrogenase probe was used as control for RNA loading. C, scheme of the human cyclin E promoter. Black boxes indicate the three E2F-binding sites. Hatched boxes indicate exons. A thick black line indicates the cyclin E promoter region amplified by PCR. D, ChIP assays, demonstrating α-32P-labeled PCR products, were performed using chromatin isolated from proliferating (P) and senescent (S) melanocytes. Abs to acetylated histone H4 were used to precipitate acetylated chromatin. Abs against α-actin and total histone H4 (α-H4) were used as negative and positive controls, respectively.
Transcriptional silencing of the cyclin E gene by promoter deacetylation. A, cyclin E and D1 protein levels at different PDs determined by Western immunoblotting. B, cyclin E mRNA levels decrease with increased PDs. Equal amounts of RNA were loaded and hybridized using a cyclin E probe (“Materials and Methods”). A glyceraldehyde-3-phosphate dehydrogenase probe was used as control for RNA loading. C, scheme of the human cyclin E promoter. Black boxes indicate the three E2F-binding sites. Hatched boxes indicate exons. A thick black line indicates the cyclin E promoter region amplified by PCR. D, ChIP assays, demonstrating α-32P-labeled PCR products, were performed using chromatin isolated from proliferating (P) and senescent (S) melanocytes. Abs to acetylated histone H4 were used to precipitate acetylated chromatin. Abs against α-actin and total histone H4 (α-H4) were used as negative and positive controls, respectively.
p3001572–1903 represses the cyclin E promoter and reduces E2F-1-associated HAT activity in vivo. Normal human melanocytes (NHM) were transfected with a cyclin E-luciferase reporter (A−C). A, cells cotransfected with an E2F-1 expression plasmid (1 μg) and increased concentrations of a p300 expression vector; B, cells cotransfected with E2F-1 plus a dominant negative p3001592–1903 vector; C, cells cotransfected with E2F-1 plus WT p300 and p3001592–1903 vectors. pCMV-βgal (100 ng) was transfected as an internal control. Values shown are means ± SD of triplicate readings and represent the fold induction of normalized luciferase activity. Similar results were obtained in three independent experiments. Transfections of p300 vectors in the absence of E2F-1 induced minimal activation (data not shown). pGL3 was used as a control plasmid. Empty vectors and experimental conditions have been balanced for DNA mass. D, total and E2F-1-associated HAT activity in 293 cells transfected with pcDNA3.1 (control), p300, and p3001592–1903 vectors, alone or in combination. The mean ± SE of two independent experiments id shown.
p3001572–1903 represses the cyclin E promoter and reduces E2F-1-associated HAT activity in vivo. Normal human melanocytes (NHM) were transfected with a cyclin E-luciferase reporter (A−C). A, cells cotransfected with an E2F-1 expression plasmid (1 μg) and increased concentrations of a p300 expression vector; B, cells cotransfected with E2F-1 plus a dominant negative p3001592–1903 vector; C, cells cotransfected with E2F-1 plus WT p300 and p3001592–1903 vectors. pCMV-βgal (100 ng) was transfected as an internal control. Values shown are means ± SD of triplicate readings and represent the fold induction of normalized luciferase activity. Similar results were obtained in three independent experiments. Transfections of p300 vectors in the absence of E2F-1 induced minimal activation (data not shown). pGL3 was used as a control plasmid. Empty vectors and experimental conditions have been balanced for DNA mass. D, total and E2F-1-associated HAT activity in 293 cells transfected with pcDNA3.1 (control), p300, and p3001592–1903 vectors, alone or in combination. The mean ± SE of two independent experiments id shown.
Overexpression of p3001572–1903 results in down-regulation of cyclin E expression in vivo. A, cell cycle profiles of UCD-Mel-J cells transfected with a plasmid control or with p3001572–1903, together with a GFP expression plasmid at a 10:1 ratio. GFP-positive cells were sorted and analyzed for cell cycle distribution 48 h after transfection. The blue dome indicates cells in the S phase of the cell cycle. B, immunofluorescence analysis of UCD-Mel-J cells transfected with a pcDNA3.1 and GFP expression plasmid at a 10:1 ratio, with p3001572–1903 vector and GFP, with p3001572–1903 plus WT p300 (1:1) and GFP, and with a WT p300 plasmid and GFP. C, arrowheads indicate cell boundaries of an enlarged, flat, cyclin E-negative cell transfected with p3001572–1903. Cells were fixed, incubated with an anti-cyclin E Ab, and processed for immunofluorescence as described in “Materials and Methods.”
Overexpression of p3001572–1903 results in down-regulation of cyclin E expression in vivo. A, cell cycle profiles of UCD-Mel-J cells transfected with a plasmid control or with p3001572–1903, together with a GFP expression plasmid at a 10:1 ratio. GFP-positive cells were sorted and analyzed for cell cycle distribution 48 h after transfection. The blue dome indicates cells in the S phase of the cell cycle. B, immunofluorescence analysis of UCD-Mel-J cells transfected with a pcDNA3.1 and GFP expression plasmid at a 10:1 ratio, with p3001572–1903 vector and GFP, with p3001572–1903 plus WT p300 (1:1) and GFP, and with a WT p300 plasmid and GFP. C, arrowheads indicate cell boundaries of an enlarged, flat, cyclin E-negative cell transfected with p3001572–1903. Cells were fixed, incubated with an anti-cyclin E Ab, and processed for immunofluorescence as described in “Materials and Methods.”
Overexpression of p3001572–1903 or exposure to the p300 HAT inhibitor Lys-CoA inhibits proliferation and induces the SA-β-Gal marker. A, top panel, UCD-Mel-N cells transfected with the plasmids described in Fig. 4 and stained for SA-β-Gal activity (“Materials and Methods”). Bottom panel, phase microscopy showing SA-β-Gal-positive, large flat cells (arrowheads). B, induction of a senescent phenotype in normal human melanocytes by Lys-CoA. Top panel, dose-dependent repression of BrdUrd incorporation by Lys-CoA. BrdUrd incorporation (arrowheads) was reduced from 31% in control cells to 4% and 2% in cells treated with 0.1 and 1 mm Lys-CoA, respectively. Middle panel, induction of SA-β-Gal activity. The number of SA-β-Gal cells increased from 11% in control cells to 74% and 93% in cells treated with 0.1 and 1 mm Lys-CoA, respectively. Bottom panel, repression of cyclin E expression. Untreated controls demonstrate high levels of nuclear cyclin E reactivity. Note that cyclin E positivity in melanocytes exposed to 0.1 mm Lys-CoA (arrowheads) is diminished compared with untreated controls. No cyclin E-positive melanocytes were observed with 1 mm Lys-CoA. Normal human melanocytes were permeabilized in the presence or absence of Lys-CoA (“Materials and Methods”) and seeded in coverslips. After 3 days, the cells were pulse-labeled with BrdUrd, fixed, and stained with an anti-BrdUrd Ab or stained for SA-β-Gal activity (“Materials and Methods”). For cyclin E expression, coverslips were placed in 10 mm sodium citrate (pH 6.0) for 10 min. Antigen retrieval was performed as described previously (35), followed by incubation with a monoclonal anti-cyclin E Ab (BD PharMingen). Identical results were obtained in at least two independent experiments.
Overexpression of p3001572–1903 or exposure to the p300 HAT inhibitor Lys-CoA inhibits proliferation and induces the SA-β-Gal marker. A, top panel, UCD-Mel-N cells transfected with the plasmids described in Fig. 4 and stained for SA-β-Gal activity (“Materials and Methods”). Bottom panel, phase microscopy showing SA-β-Gal-positive, large flat cells (arrowheads). B, induction of a senescent phenotype in normal human melanocytes by Lys-CoA. Top panel, dose-dependent repression of BrdUrd incorporation by Lys-CoA. BrdUrd incorporation (arrowheads) was reduced from 31% in control cells to 4% and 2% in cells treated with 0.1 and 1 mm Lys-CoA, respectively. Middle panel, induction of SA-β-Gal activity. The number of SA-β-Gal cells increased from 11% in control cells to 74% and 93% in cells treated with 0.1 and 1 mm Lys-CoA, respectively. Bottom panel, repression of cyclin E expression. Untreated controls demonstrate high levels of nuclear cyclin E reactivity. Note that cyclin E positivity in melanocytes exposed to 0.1 mm Lys-CoA (arrowheads) is diminished compared with untreated controls. No cyclin E-positive melanocytes were observed with 1 mm Lys-CoA. Normal human melanocytes were permeabilized in the presence or absence of Lys-CoA (“Materials and Methods”) and seeded in coverslips. After 3 days, the cells were pulse-labeled with BrdUrd, fixed, and stained with an anti-BrdUrd Ab or stained for SA-β-Gal activity (“Materials and Methods”). For cyclin E expression, coverslips were placed in 10 mm sodium citrate (pH 6.0) for 10 min. Antigen retrieval was performed as described previously (35), followed by incubation with a monoclonal anti-cyclin E Ab (BD PharMingen). Identical results were obtained in at least two independent experiments.
The cyclin E promoter is repressed by RB and HDACs. Normal human melanocytes (NHM) were transfected with a human cyclin E-luciferase reporter. A, 5 h after transfection, melanocytes were treated with increasing concentration of TSA for a total of 51 h. B, melanocytes were cotransfected with an E2F-1 expression plasmid (1 μg) in the presence of increasing concentrations of TSA. C, melanocytes were transfected with RB or with the mutant RBN757F in the presence or absence of TSA. A pCMV-β-gal vector was transfected as control for transfection efficiency. Values shown are means ± SD of triplicate readings and represent luciferase activity normalized to β-galactosidase activity. Similar results were obtained in at least two independent experiments. PGL3 was used as a control plasmid. Empty vectors and experimental conditions have been balanced for DNA mass.
The cyclin E promoter is repressed by RB and HDACs. Normal human melanocytes (NHM) were transfected with a human cyclin E-luciferase reporter. A, 5 h after transfection, melanocytes were treated with increasing concentration of TSA for a total of 51 h. B, melanocytes were cotransfected with an E2F-1 expression plasmid (1 μg) in the presence of increasing concentrations of TSA. C, melanocytes were transfected with RB or with the mutant RBN757F in the presence or absence of TSA. A pCMV-β-gal vector was transfected as control for transfection efficiency. Values shown are means ± SD of triplicate readings and represent luciferase activity normalized to β-galactosidase activity. Similar results were obtained in at least two independent experiments. PGL3 was used as a control plasmid. Empty vectors and experimental conditions have been balanced for DNA mass.
Alternative recruitment of the cyclin E promoter by p300 and HDAC1 in proliferating and senescent melanocytes. A, increased association of RB to HDAC1 in immunoprecipitates from senescent melanocytes. B, HDAC1 activity assayed in RB immunoprecipitates (“Materials and Methods”). C, ChIP assays. ChIP assays demonstrating α-32P-labeled PCR products were performed using chromatin isolated from proliferating (P) and senescent (S) melanocytes. Chromatin was immunoprecipitated with anti-p300 and HDAC1 Abs. Abs against α-actin, and total histone H4 (α-H4) were used as negative and positive controls respectively. ∗ indicates the position of the PCR products from p300 and HDAC1 immunoprecipitated promoter fragments. +ve indicates PCR products from the cyclin E promoter-containing plasmid. D, overexpression of HDAC1 represses BrdUrd incorporation and induces SA-β-Gal-positive cells. UCD-Mel-N melanoma cells were transfected with pcDNA or HDAC1 expression plasmid, followed by selection with hygromycin (see text). E, Western blot indicating levels of HDAC1 protein after transfection with pcDNA3.1 or HDAC1 expression vectors.
Alternative recruitment of the cyclin E promoter by p300 and HDAC1 in proliferating and senescent melanocytes. A, increased association of RB to HDAC1 in immunoprecipitates from senescent melanocytes. B, HDAC1 activity assayed in RB immunoprecipitates (“Materials and Methods”). C, ChIP assays. ChIP assays demonstrating α-32P-labeled PCR products were performed using chromatin isolated from proliferating (P) and senescent (S) melanocytes. Chromatin was immunoprecipitated with anti-p300 and HDAC1 Abs. Abs against α-actin, and total histone H4 (α-H4) were used as negative and positive controls respectively. ∗ indicates the position of the PCR products from p300 and HDAC1 immunoprecipitated promoter fragments. +ve indicates PCR products from the cyclin E promoter-containing plasmid. D, overexpression of HDAC1 represses BrdUrd incorporation and induces SA-β-Gal-positive cells. UCD-Mel-N melanoma cells were transfected with pcDNA or HDAC1 expression plasmid, followed by selection with hygromycin (see text). E, Western blot indicating levels of HDAC1 protein after transfection with pcDNA3.1 or HDAC1 expression vectors.
Hypothetical mechanism for senescence in human melanocytes mediated by loss of p300 and recruitment of HDACs on cell cycle-regulatory gene promoters. Competition of p300 and HDACs for binding to promoters of cell cycle genes may determine whether a cell remains in the cell cycle or enters senescence.
Hypothetical mechanism for senescence in human melanocytes mediated by loss of p300 and recruitment of HDACs on cell cycle-regulatory gene promoters. Competition of p300 and HDACs for binding to promoters of cell cycle genes may determine whether a cell remains in the cell cycle or enters senescence.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by NIH Grants 1P01 AG19254 and R01 CA84282. D. B. was supported by a fellowship from the American Federation for Aging Research.
The abbreviations used are: HDAC, histone deacetylase; HAT, histone acetyltransferase; PD, population doubling; CBP, cAMP-responsive element-binding protein-binding protein; CDK, cyclin-dependent kinase; SA-β-Gal, senescence-associated β-galactosidase; CDKI, CDK inhibitor; pRB, Retinoblastoma tumor suppresspr protein p105; hTERT, human telomerase reverse transcriptase subunit. Ab, antibody; TBS, Tris-buffered saline; ChIP, chromatin immunoprecipitation; TSA, trichostatin A; WT, wild-type; GFP, green fluorescent protein; BrdUrd, bromodeoxyuridine.
E. E. Medrano and D. Bandyopadhyay, unpublished results.
Acknowledgments
We thank A. Harel-Bellan and A. Polesskaya for guiding us with the cell permeabilization protocol. We also thank N. LaThangue for the p3001572–1903 plasmid, Jean Y. L. Wang and Tung-Ti Chen for WT and mutant RB plasmids, and Nick Timchenko and Judy Campisi for critical reading of the manuscript.