p53 Chromatin Epigenetic Domain Organization and p53 Transcription

Abstract

Epigenetic organization represents an important regulation mechanism of gene expression. In this work, we show that the mouse p53 gene is organized into two epigenetic domains. The first domain is fully unmethylated, associated with histone modifications in active genes, and organized in a nucleosome-free conformation that is deficient in H2a/H2b, whereas the second domain is fully methylated, associated with deacetylated histones, and organized in a nucleosomal structure. In mitotic cells, RNA polymerase is depleted in domain II, which is folded into a higher-order structure and is associated with H1 histone, whereas domain I conformation is preserved. Similar results were obtained for cells treated with inhibitors of associated regulatory factors. These results suggest that depletion of RNA polymerase II is the result of a physical barrier due to the folding of chromatin in domain II. The novel chromatin structure in the first domain during mitosis also suggests a mechanism for marking active genes in successive cell cycles.

DNA methylation and chromatin organization constitute the major epigenetic control mechanisms of eukaryotic gene expression (14, 16, 23, 28). Methylation of sequences in the promoter region has been shown to repress transcription through the recruitment of methylcytosine-binding protein and chromatin-remodeling enzymes (2, 5). Gene expression is also regulated by histone modifications that alter the conformation of chromatin (13, 29, 44). In spite of our knowledge of the role of epigenetic modifications in the promoter region, few studies have addressed the epigenetic modifications in the intragenic region and the role of intragenic epigenetic modifications in gene expression. In order to elucidate the biological functions of intragenic epigenetic modifications in gene expression, we analyzed the epigenetic organization of the entire gene region of the mouse p53 gene with respect to DNA methylation, histone modifications, association with regulatory factors, chromatin folding, and transcription. The p53 gene was found to be organized into two epigenetic domains with differential DNA methylation characteristics, histone modifications, and compositions. We showed that in mitotic cells and in cells treated with inhibitors, the loss of RNA polymerase II (pol II) and regulatory factors in the second epigenetic domain was associated with higher-order folding and association with H1 histone in this domain. This result suggests that higher-order folding provides a physical barrier for the elongation of transcription into the second domain. The first domain remained in a loose chromatin conformation, in association with regulatory factors, and deficient in H2a/H2b in mitotic cells, suggesting that this special chromatin organization serves as a memory of active genes to be transcribed in the successive phases of the cell cycle.

MATERIALS AND METHODS

Cell culture and transfection.

PT67 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and 1× gentamicin. PT67 cells were grown to confluence for most experiments. Cells were transfected using Lipofectamine (Invitrogen) according to the instructions of the manufacturer. For analysis of p53 transcription in mitotic cells, cells were cultured in DMEM containing fetal bovine serum (FBS; 0.1%) for 30 h to synchronize cells in the G₀ phase. Fresh DMEM supplemented with 10% FBS was then added to induce the cells to reenter the cell cycle. After 18 h, nocodazole (0.4 μg/ml) was added before the cells entered the M phase. For topoisomerase (Topo) inhibition experiments, cells were treated with 10 μM to 25 μM camptothecin (Topo I inhibitor) for 10 min to 2 h or 10 μM to 50 μM etoposide (Topo II inhibitor) for 10 min to 2 h.

Bisulfite methylation assays.

Genomic DNA (2 μg) from PT67 cells or from different organs of 4-week-old mice (gift of T. F. Tsai) was digested overnight with the XhoI restriction enzyme. Bisulfite treatment was carried out as described by Paulin et al. (34) The bisulfite-treated DNA was amplified by PCR. Sequencing of the PCR-amplified product was performed using forward and reverse primers. An α-³³P-labeled dideoxynucleoside triphosphate terminator kit (Amersham Biosciences) was used for sequencing. The sequencing gel was dried and exposed to an X-ray field. Methylation analysis was carried out by quantifying the intensity of the C and T bands using PhosphorImager analysis (Molecular Dynamic) and by calculating the percentage of C bands [C/(C+T)] with ImageQuant 2.0 software. For a list of primers used in the bisulfite methylation assay, see Table S1 in the supplemental material.

Nucleus preparation.

Cells grown to confluence were scraped and pelleted at 1,500 rpm for 10 min. Cells were washed twice in ice-cold 1× phosphate-buffered saline. The cell pellet was resuspended in ice-cold lysis buffer (10 mM Tris [pH 7.4], 10 mM NaCl, 3 mM MgCl₂, 0.5% NP-40, 0.15 mM spermine, 0.5 mM spermidine) and incubated on ice for 10 min (11) Nuclei were pelleted at 1,200 rpm for 10 min and then washed in micrococcal nuclease (MNase) digestion buffer (10 mM Tris [pH 7.4], 15 mM NaCl, 60 mM KCl, 0.15 mM spermine, 0.5 mM spermidine).

MNase digestion.

MNase digestion proceeded as described by Richard-Foy and Hager, with minor modifications (37). In brief, the cell pellet was resuspended in ice-cold MNase digestion buffer with 0.1 mM CaCl₂ and 10 to 75 U/ml MNase (Pharmacia), and the reaction mixtures (each in a total volume of 100 μl) were then incubated for 5 min at 37°C to allow digestion. The reactions were terminated by adding 80 μl MNase digestion buffer and 20 μl stop solution (100 mM EDTA, 100 mM EGTA, 3 μl proteinase K [25 mg/ml], and 10 μl of 20% sodium dodecyl sulfate [SDS]) and incubated overnight at 37°C. Genomic DNA was purified with phenol-chloroform extraction for PCR and Southern blotting analyses.

ChIP assays.

Chromatin immunoprecipitation (ChIP) assays were performed according to the manufacturer's protocol (Upstate Biotechnology, Inc., Lake Placid, NY) except that the sonication conditions were changed to five times for 30 s each time at 10% output. Specific antibodies for immunoprecipitations were anti-H3K9ac, anti-H4K16ac, anti-H3K4me2, anti-H3K4me3, anti-H3K9me2, anti-H3K36me, or anti-HDAC1 (Upstate Biotechnology); anti-histone H1, anti-Dnmt1, anti-pol II, anti-Brg1, anti-CTCF, anti-Topo I, anti-Topo IIα, or anti-PARP1 (Santa Cruz Biotechnology, Inc.); or anti-histone H2B (Abcam). Nonspecific rabbit polyclone antibody was used as a negative control. After protein-DNA cross-links in the immunoprecipitation complexes were reversed, DNA was extracted for subsequent PCR analysis. For a list of primer sets for PCR, see Table S2 in the supplemental material. PCR products were analyzed by agarose gel electrophoresis, and the relative intensities of DNA bands were calculated using ImageQuant 2.0 software.

3C assays.

We used chromatin conformation capture (3C) assays (38) with some modifications. We fixed the PT67 cells in 2% formaldehyde for 10 min at room temperature and quenched them with 0.125 M glycine. After centrifugation for 15 min at 3,500 rpm, the cells were suspended in lysis buffer (10 mM Tris-HCl [pH 8.0], 10 mM NaCl, 0.2% Nonidet P-40, and 1:500 Complete protease inhibitor cocktail; Roche) for 90 min on ice. The nuclei were pelleted by centrifugation for 15 min at 2,500 rpm, resuspended in 500 μl of 1× NEB buffer 4 plus 0.3% SDS, and incubated at 37°C for 1 h. After the addition of Triton X to a final concentration of 1.8% to sequester the SDS and incubation at 37°C for 1 h, 800 U of NcoI was added to digest chromatin overnight at 37°C. The reaction was terminated by adding SDS to a final volume of 1.6%, and the solution was heated to 65°C for 20 min. Ligation of DNA in situ was carried out with 0.5 to 2.0 ng/μl of chromatin in 800 μl of ligation buffer (NEB) plus 1% Triton X and 30 Weiss units of T4 ligase (NEB) for 4 h at 16°C. After cross-links were reversed with proteinase K digestion at 65°C overnight, DNA was purified by phenol-chloroform extraction and ethanol precipitation. Ligation products were detected by PCR using primers located near NcoI cutting sites. PCR parameters were as follows: initial denaturation at 95°C for 5 min, followed by a three-step cycle (98°C for 30 s, 65°C for 30 s, 72°C for 30s) for 35 cycles and a final extension step of 72°C for 10 min. The PCR products were purified from agarose gels, cloned, and sequenced.

RESULTS

p53 contains two epigenetic domains with differential DNA methylation characteristics and histone modifications.

Mouse p53 is about 11.8 kb in length and contains 11 exons (Fig. 1A). We analyzed the distribution of CpG methylation of the mouse p53 gene and its surrounding sequences in mouse PT67 cells by using a bisulfite sequencing technique. As shown in Fig. 1A, all the 136 CpG sites from nucleotide −1729 to nucleotide +2112 relative to the transcription initiation site were found to be completely unmethylated. This region includes the promoter, the first exon, and part of the first intron. Bisulfite sequencing also showed that the rest of the gene, containing 187 CpG sites, was fully methylated. This result showed that p53 is organized into two DNA methylation domains with a sharp methylation boundary located in the first intron at nucleotide +2112. This organization of DNA methylation was also found in the DNA extracted from mouse adult and fetal mouse liver, brain, heart, and kidney tissues, as well as testis tissues. Thus, the dual methylation pattern is preserved in different fetal and adult organs. We designated these two regions of the p53 gene epigenetic domains I and II. The first domain encompasses the promoter region, noncoding exon 1, and part of intron 1, whereas the second domain comprises the majority of the p53 gene (Fig. 1A).

FIG. 1.

The p53 gene is organized into two epigenetic domains by DNA methylation characteristics and histone modifications. (A) Schematic representation of the p53 gene showing the position of the promoter and exons as well as the methylation boundaries. The gene is subdivided into 25 regions for PCR analyses of the chromatin structure and ChIP assays. The distribution of CpG methylation was analyzed by both methylation-sensitive PCR and bisulfite sequencing (M, methylated region; Um, unmethylated region; T, upstream methylation transition zone). The positions of CpG sites are shown below the methylation data. (B to E) Distribution of histone modifications in the p53 gene. Specific antibodies used in the ChIP experiments were anti-acetylated H3-K9 (H3K9ac; B), anti-acetylated H4-K16 (H4K16ac; C), anti-dimethylated H3-K4 (H3K4me2; D), and anti-trimethylated H3-K4 (H3K4me3; E). DNA isolated from immunoprecipitated chromatin was subjected to PCR to amplify DNA fragments from the murine p53 locus in the 25 regions specified in panel A. Nonimmune immunoglobulin G (IgG)-immunoprecipitated DNA was used as the control. To determine the relative levels of acetylation and di- and trimethylation in these regions (1 through 25), primers specific for the murine GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene promoter with known histone modification patterns were included in the PCRs as the internal control. The degree of enrichment is calculated relative to the ratio of signals obtained in the input DNA fraction. Data are the means ± standard errors of the means (SEM) from at least three independent experiments.

Further analysis showed that there is another methylation boundary at the 5′ end of the p53 promoter. Beyond nucleotide −2225, the genomic sequence again showed full CpG methylation as determined by bisulfite sequencing. However, between nucleotides −2225 and −1729, there is a 500-bp region with nine CpG sites that showed partial methylation patterns that varied among the mouse tissues and between fetal and adult tissues (see Fig. S1 in the supplemental material). However, there was no simple correlation between the degree of methylation in this region and the level of p53 expression in the mouse adult and fetal tissues. The biological significance of this tissue-specific partially methylated region remains to be analyzed.

Since modified histones have been found to associate with differentially methylated DNA sequences, we then used ChIP techniques to analyze the nature of histones associated with the two epigenetic domains. We divided the p53 gene and its surrounding sequences into 25 segments and analyzed the presence of acetylated and methylated H3 as well as acetylated H4 histones in the entire region. As depicted in Fig. 1, domain I was found to be associated with H3K9ac (Fig. 1B), H3K4me2 (Fig. 1D), and H3K4me3 (Fig. 1E) as well as H4K16ac (Fig. 1C), whereas these modified histones were absent in domain II. H3K36me, which is associated with transcription elongation (20, 27), was found enriched in the promoter region (data not shown). No H3K9me, which is associated with inactive chromatin, was found in the gene. These results indicate that p53 is organized into two epigenetic domains containing differentially modified histones associated with differentially methylated DNA. Domain I contains no methylated CpG and is associated with modified histones that have been found to be associated with active genes (8, 24, 29), whereas domain II contains extensively methylated DNA but none of the modified histones described above.

Unmethylated epigenetic domain I is hypersensitive to nuclease and deficient in H2a/H2b.

Chromatin with acetylated histones is usually organized in a loose conformation. To see if the two epigenetic domains were indeed organized into different chromatin conformations, we treated the nuclei with MNase under limited digestion conditions. We used PCR to probe the region hypersensitive to nuclease digestion. The nuclei from PT67 cells were digested for 5 min at 37°C with different amounts of nuclease, and the sensitivities of different regions of p53 to nuclease digestion were assayed by PCR. We reasoned that DNA cleaved between the PCR primer sites would not be amplified by PCR; therefore, regions more sensitive to nuclease digestion would show decreased amounts of PCR product. This analysis showed that unmethylated domain I was much more sensitive to nuclease digestion than domain II (Fig. 2A). A Southern blot analysis of MNase-digested DNA also showed that domain I was rapidly cleaved into small fragments of a size smaller than that of the nucleosomal core and without a nucleosomal DNA ladder, suggesting that domain I is not organized into a regular nucleosome structure, whereas under the same digestion conditions, the domain II region showed a regular nucleosomal ladder pattern for the bulk of the chromatin (Fig. 2B). Under more mild digestion conditions (at 30°C or 20°C for 5 min), we were able to map three nuclease-hypersensitive sites in the p53 promoter region, around bp −1046, −1246, and −1446, respectively, by using an indirect end-labeling technique (see Fig. S2 in the supplemental material).

FIG. 2.

Unmethylated domain I region of p53 is hypersensitive to MNase digestion and devoid of H2B. (A) PCR analysis of different regions of the p53 gene after PT67 nuclei were digested with increasing amounts of MNase (0, 10, 15, 20, 25, 35, and 75 U/ml) at 37°C for 5 min. After purification, genomic DNA was analyzed by PCR amplification using the primer pairs 1 to 13 as indicated in the p53 map. The unmethylated region is marked and is clearly found to be nuclease hypersensitive. MN conc., MNase concentration. (B) Southern blotting analysis of the nucleosomal organization of epigenetic domains I and II. DNA was extracted from nuclei digested with MNase for 5 min at 37°C with increasing amounts of nuclease (from 10 to 75 U/ml). The DNA was electrophoresed in a 1.2% agarose gel. Left, result of Southern blot analysis using probe A as indicated in the upper portion of the panel. Right, Southern blot analysis of nucleosomal organization of epigenetic domain II using probe B and showing the nucleosomal DNA ladders. conc., concentration. (C) Distribution of H2B of p53 chromatin in the PT67 cells (n = 3; means ± SEM are shown).

Since MNase digestion of domain I resulted in the rapid generation of DNA fragments smaller than the nucleosomal core, we further investigated the nature of chromatin structure in epigenetic domain I. Since the H2a/H2b dimer has been shown to be displaced during transcription (19), we tested the possibility that H2a/H2b was released from this part of the p53 gene, resulting in the alteration of nucleosomal structure. ChIP analysis using an antibody to H2a or H2b indeed showed that epigenetic domain I was deficient in H2a/H2b histones relative to levels of these histones in domain II (Fig. 2C). This result suggests that the hypersensitivity to nuclease digestion in epigenetic domain I is the result of altered histone-DNA composition and structure with a deficiency in the H2a/H2b dimer.

Association of chromatin modification and transcription regulatory factors with promoter and intragenic regions of p53.

The novel organization of distinct epigenetic domains in p53 suggests that this organization may play a role in the regulation of gene expression. To examine this possibility, we first analyzed the location of factors involved in transcription regulation relative to the epigenetic domains. We examined the distribution of seven regulatory factors, CTCF, PARP1, Topo II, Topo I, HDAC1, Dnmt1, and Brg1, along the genes by using a ChIP technique. These analyses showed that five regulatory factors, CTCF, Topo II, PARP1, HDAC1, and Dnmt1, were associated with epigenetic domain I in the promoter/first intron region, whereas all seven regulatory factors were associated with methylated domain II near the domain boundary (Fig. 3).

FIG. 3.

Association of several regulatory factors with the p53 epigenetic domains. ChIP assays using anti-CTCF, anti-Topo II, anti-HDAC1, anti-Dnmt1, anti-Brg1, and anti-Topo I antibodies were performed to determine the binding sites of these regulatory factors. Input chromatin and nonspecific IgG precipitate are shown as controls. CTCF, HDAC1, Dnmt1, and Topo II were found in domains I and II, whereas Brg1 and Topo I is found only in domain II. CTCF and Topo II were located in region 13; HDAC1, Brg1, and Topo I were associated with region 12; and Dnmt1 was associated with both region 12 and region 13. α-, anti-.

The results obtained showed that there were two CTCF chromatin boundary binding regions in the p53 gene, one located in the p53 promoter region from bp −1673 to +290 in epigenetic domain I and the other in the second intronic epigenetic domain from bp +3205 to +3887 (Fig. 3). Interestingly, these sites are located near the upstream and downstream boundaries surrounding epigenetic domain I. The presence of CTCF binding sequence was further verified using a gel mobility shift assay (see Fig. S3A and B in the supplemental material). We used a reporter gene assay (41) to demonstrate that this CTCF binding site could indeed block enhancer function. The upstream CTCF binding sequence of p53 was cloned into the pGL2 vector (as depicted in Fig. 3C in the supplemental material). Beta-globin insulator sequence (12) was used as a control to demonstrate the enhancer-blocking effect. The plasmids were linearized by digestion with SacI restriction endonuclease and transfected into PT67 cells. The known beta-globin insulator sequence, when inserted between the promoter and the enhancer, could suppress the promoter activity about sixfold, whereas insertion outside the promoter-enhancer pair resulted in a twofold reduction in promoter activity. The p53 CTCF binding site could suppress promoter activity about 3-fold when inserted between the promoter and enhancer, whereas insertion outside the promoter-enhancer suppressed the promoter only about 1.4-fold. These results indicate that the CTCF binding site in p53 indeed possesses insulator activity.

Depletion of RNA pol II in epigenetic domain II in mitotic phase-arrested cells with loss of regulatory factors from the methylated epigenetic domain.

To examine the role of dual epigenetic organization in p53 gene expression, we analyzed the distribution of RNA pol II in different phases of the cell cycle by using a ChIP technique. Since transcription in general is downregulated during G₂/M (33, 39), we first checked the transcription status of p53 in cells arrested in mitosis by treatment with nocodazole. Since Ser10-phosphorylated H3 is a marker of the mitotic chromosome (43), we used ChIP analysis to verify that this modified histone had indeed become associated with p53 chromatin in mitotic cells (see Fig. S4 in the supplemental material).

Reverse transcription-PCR analysis showed that the expression of p53 was downregulated in mitotic phase-arrested cells. To understand the mechanism of downregulation, we used a ChIP technique to analyze the distribution of RNA pol II in the p53 gene. The analysis showed that in mitotic phase-arrested cells, RNA pol in epigenetic domain II became depleted (Fig. 4C), suggesting that downregulation of transcription is due to depletion of RNA pol II at the epigenetic domain boundary. In contrast, the enzyme was more or less uniformly distributed in the p53 gene in interphase cells (Fig. 4B).

FIG. 4.

Depletion of RNA pol II with loss of regulatory factors in the methylated epigenetic domain of p53 gene chromatin in G₂/M-arrested cells as analyzed by ChIP. (A) Structure of the p53 gene showing the positions of the promoter, the exons, and the two epigenetic domains. (B and C) Distribution of RNA pol II of p53 in the nonsynchronized cells or in the G₂/M-phase cells, respectively. RNA pol II is depleted in the epigenetic domain II in mitotic phase-arrested cells accumulated by treatment with nocodazole (G2/M) compared with the nonsynchronized cells (untreated). (D and E) Loss of regulatory factors from the intragenic region of the p53 gene in mitotic phase-arrested cells. ChIP assays using anti-CTCF, anti-Topo II, anti-PARP1, anti-Brg1, and anti-Topo I antibodies were performed to determine the binding sites of these regulatory factors. α-, anti-.

We employed run-on experiments to further confirm the depletion of transcription in epigenetic domain II and the nature of transcription in domain I. In order to deal specifically with run-on nascent RNA, we labeled the run-on RNA with biotin and affinity purified the run-on products. The nature of the affinity-purified nascent RNA was then analyzed by PCR using primers across the p53 gene. The result indeed shows that the run-on RNA obtained from G₂/M nuclei was derived from the 5′ part of the gene (see Fig. S5 in the supplemental material). In contrast, biotin-labeled run-on RNA from interphase nuclei is derived from sequences across the gene.

To examine the role of regulatory factors bound in p53 in the regulation of transcription in mitotic cells, we analyzed the presence of these factors in the two domains of p53 chromatin. A ChIP analysis of p53 chromatin in the mitotic cells showed that downregulation of transcription of p53 in mitotic cells was accompanied by the loss of the CTCF boundary element in the methylated domain II but not in the unmethylated domain I (Fig. 4D). Interestingly, we found that other regulatory factors, including PARP1, Topo I, and Brg1, were also lost from the methylated domain II of p53 chromatin (Fig. 4E). The behavior of Topo II differed from that of other factors in that it was lost from both domain I and domain II in the mitotic cells. The loss of these regulatory factors and depletion of RNA pol in domain II in the mitotic cells suggests that these factors may be involved in the regulation of transcription in epigenetic domain II of p53.

Inhibition of PARP1 and Topos resulted in the depletion of RNA pol in epigenetic domain II and the loss of regulatory factors.

Since PARP has been implicated in the regulation of CTCF-dependent chromatin insulation as well as in transcription (48), the observation of the presence of this enzyme in the same CTCF binding site regions at the domain boundaries of p53 (Fig. 3) and the loss of this factor in domain II in mitotic cells (Fig. 4D) suggest that PARP may also be involved in the regulation of transcription of p53. To examine this possibility, we studied the effect of 3-aminobenzamide (3-AB), a potent inhibitor of PARP, on the expression of p53. 3-AB treatment inhibited the expression of p53, as shown by reverse transcription-PCR analysis (data not shown). Consistent with this finding, 3-AB treatment resulted in the depletion of RNA pol II in the methylated domain II, as was observed for the mitotic phase-arrested cells (Fig. 5B). ChIP analysis showed that inhibition of transcription by 3-AB treatment was accompanied by the loss of PARP1 as well as of CTCF, Topo II, Brg1, and Topo I from the p53 chromatin domain II (Fig. 5G and H). This result suggests that the PARP enzyme located in the domain boundary is involved in the regulation of transcription elongation into domain II and the organization of these factors at the epigenetic domain boundaries of the p53 gene.

FIG. 5.

Inhibition of PARP1 and Topos resulted in the depletion of RNA pol II in the second epigenetic domain and the loss of regulatory factors. (A) Structure of the p53 gene showing the positions of the promoter, the exons, and the two epigenetic domains. (B) Depletion of RNA pol II in domain II in the p53 gene in 3-AB-treated PT67 cells. (C) Treatment with the Topo I inhibitor camptothecin for 30 min resulted in depletion of pol II in domain II of p53. (D) Treatment with the Topo I inhibitor camptothecin for 2 h resulted in an accumulation of pol II in region 12, where Topo I is located (Fig. 3). (E) Treatment with the Topo II inhibitor etoposide for 30 min in vivo resulted in the depletion of pol II in domain I of p53 and the 3′ half of domain II. (F) Treatment with the Topo II inhibitor etoposide for 2 h resulted in the depletion of pol II in the entire gene (G and H). After 3-AB treatment, loss of CTCF, Topo II, PARP1, Brg1, and Topo I from p53 chromatin occurred. Input chromatin and nonspecific IgG precipitate are shown as controls. α-, anti-.

The transcription process generates topological strain (10, 46), and relaxation of the supercoiling tension by Topos is required for transcription to proceed on the chromatin template (26). The presence of both Topo I and Topo II in the boundary of epigenetic domains suggests that these enzymes may also be involved in the regulation of transcription elongation near the boundary. To examine this possibility, we treated cells with Topo I and Topo II inhibitors and analyzed the distribution of RNA pol II in the p53 gene by using a ChIP technique.

Treatment with a low concentration of Topo I inhibitor camptothecin (5 μm) for 30 min also caused a rapid depletion of pol II in the methylated epigenetic domain II (Fig. 5C). Longer treatment resulted in an accumulation of pol II at the domain boundary (Fig. 5D). This result again suggests that Topo I is involved in the regulation of transcription elongation of p53.

Treatment with a low concentration of the Topo II inhibitor etoposide for 30 min resulted in a rapid depletion of pol II in the unmethylated epigenetic domain I and the 3′ half of the methylated domain (Fig. 5E). This result suggests that inhibition of Topo II located at sequences flanking epigenetic domain I causes depletion of pol II in epigenetic domain I and blocking of transcription in domain II. Longer treatment resulted in the depletion of pol II in the entire gene (Fig. 5F). Inhibition of Topo II has been shown to cause double-stranded cleavage of DNA (7, 25). In our experiment, we also found that treatment with etoposide for 30 min produced some double-stranded DNA breaks at nucleotides −837, +413, and +1663, near Topo II binding sites in the p53 gene (data not shown). However, the limited extent of cleavage of chromatin could not account for the extensive depletion of RNA pol II in the chromatin.

Association of histone H1 with epigenetic domain II of p53 chromatin during blocking of transcription elongation.

The depletion of RNA pol II at the epigenetic boundary as described above suggests a physical barrier for elongation into the methylated epigenetic domain II. Since compaction of chromatin has been shown to be a hallmark of the silenced gene (18, 30), we tested the possibility that the higher-order folding of the methylated epigenetic domain II of p53 results in a blocking of transcription elongation at the epigenetic domain boundary. Since histone H1 is involved in the first step of folding of nucleosomal fiber (6, 36) and inhibition of chromatin remodeling (17, 35), we analyzed the distribution of H1 histone in p53 chromatin in nonsynchronized and mitotic cells accumulated by nocodazole treatment. As shown in Fig. 6, H1 histone was found in the p53 chromatin in mitotic phase-arrested cells but not in nonsynchronized cells. Furthermore, H1 histone in mitotic cells was found in epigenetic domain II and the region upstream of epigenetic domain I, whereas epigenetic domain I remained free of H1. These results suggest that the folding/condensation of epigenetic domain II during metaphase or when cells were treated with inhibitors is responsible for the blocking of transcription elongation before domain II.

FIG. 6.

Histone H1 levels appear enriched on epigenetic domain II of p53 chromatin in G₂/M-arrested cells as analyzed by ChIP. Distribution of histone H1 of p53 in the nonsynchronized cells or in the G₂/M-phase cells, respectively (n = 3; means ± SEM are shown).

Change of higher-order chromatin structure in epigenetic domain II and preservation of extended conformation and active histone modification codes of epigenetic domain I in mitotic cells and in cells treated with inhibitors.

The association of histone H1 in domain II suggests that domain II is in higher-order chromatin conformation in mitotic cells when transcription is blocked at the epigenetic domain boundary. A consequence of higher-order folding of domain II in p53 chromatin is that the distal 3′ part of the gene will be topologically close to the promoter region. To test this possibility, we employed the 3C technique (38) to examine whether the 3′ part of p53 chromatin could become topologically proximal to the promoter as the result of chromatin folding in the mitotic cells. As shown in Fig. 7, 3C analysis showed that the promoter sequence of p53 was found to be joined to epigenetic domain II sequences, including the distal 3′ end, in mitotic cells. No such joining was found in interphase or in G₁ cells (Fig. 7). These results thus are consistent with the folding of epigenetic domain II in metaphase cells. 3C analysis also showed a folding of domain II when cells were treated with camptothecin or 3-AB, suggesting that the folding of chromatin caused inhibition of transcription elongation into this region of p53 chromatin.

FIG. 7.

Epigenetic domain II sequences including the distal 3′ end interact with the p53-proximal promoter region. (A) Schematic of the p53 gene with the orientation of primers (arrows) and their distances to NcoI sites in the p53 locus are shown. (B) 3C assay shows the presence of the specific 3C products (1-2, 1-3, 1-4, and 1-5) in different drug-treated PT67 cells (G2/M, nocodazole-treated PT67 cells; 3-AB, 3-AB-treated PT67 cells; CPT, camptothecin-treated PT67 cells; G1/S, aphidicolin-treated PT67 cells) under conditions in which formaldehyde (CH₂O) cross-linking and ligation of the samples were carried out. The loading control primers (gapdh promoter) showed that all of the samples contained similar amounts of DNA.

The absence of H1 in domain I in mitotic cells suggests that this region of p53 chromatin is not folded in a higher-order structure. Nuclease digestion experiments indeed showed this region to remain hypersensitive to nuclease digestion, as in nonsynchronized cells. We also showed that H3K9ac, H3K9me2, H3K9me3, and H4K16ac remained with domain I in mitotic cells. Furthermore, domain I remains deficient in H2a/H2b histones (see Fig. S6 in the supplemental material). These results support the model that p53 chromatin is differentially folded in domain II but not in domain I in mitotic cells and when transcription elongation is blocked by treatment with inhibitors.

DISCUSSION

In this report, we show that p53 is organized into two epigenetic domains with distinct chromatin structures, CpG methylation, histone compositions, and modifications, as well as associated regulatory factors. The open chromatin conformation, unmethylated CpG, and histone modifications associated the first domain are characteristic of an active chromatin, whereas the characteristics of the second domain, with full CpG methylation, nuclease resistance, and unacetylated histones, are reminiscent of inactive chromatin.

The promoter region has been known to be organized into a nucleosome-free structure (32), but whether this structure is due to the absence of histones or to a novel organization of the histone-DNA complex has not been investigated. The ChIP technique allows us to address this question, and we show that nuclease hypersensitivity and the absence of a nucleosomal DNA ladder in epigenetic domain I are associated with the depletion of the H2a/H2b dimer. Previous studies have shown that the H2a/H2b dimer could be displaced from chromatin when DNA is positively supercoiled (22) or when chromatin is remodeled by SWI/SNF in vivo (42) or in vitro (3). In the case of displacement of H2a/H2b in the mouse mammary tumor virus promoter (42), only the nucleosome under remodeling and not the adjacent nucleosome is affected. Our results, however, show that displacement of H2a/H2b is extensive in the 3.7-kb region of epigenetic domain I. We believe this is the first example of the depletion of the H2a/H2b dimer in an extended chromatin region in vivo.

Previous studies have shown that the H3/H4 tetramer is the main component in the assembly of nucleosome, but without H2a/H2b, the structure cannot be folded into a higher-order structure (15). A previous study using in vitro-reconstituted chromatin showed that chromatin containing two H3-H4 tetramers without H2a/H2b is transcribed much more efficiently than chromatin containing H2a/H2b (9). Transcription by RNA pol II is also known to cause displacement of H2a/H2b (19, 21). The novel chromatin structure in domain I with modified histones and the deficiency in H2a/H2b therefore would provide a chromatin structure conducive to transcription by RNA pol. Such a structure would be ready for transcription when the cells enter a new cell cycle.

It has been shown using cross-linking techniques that histone H1 interacts with histone H2a in chromatin at specific sites in the proteins (4). The depletion of H2a/H2b in epigenetic domain I therefore would allow domain I to escape compaction by H1 histone during the condensation of chromatin in metaphase, thus explaining the persistent nuclease hypersensitivity and the association of regulatory factors in domain I in mitotic cells. This would provide a mechanism for marking the active gene for transcription in the next cell cycle. The mechanism for establishing this novel chromatin structure with depletion of H2a/H2b in domain I is currently unknown. During DNA replication, H3 and H4 are deposited immediately, followed a few minutes later by H2a and H2b (45). If some modifying factors could bind to the nascent nucleosome before the deposition of H2a/H2b, then a special chromatin structure would be established. We are now using the special characteristics of this novel chromatin structure to search for active genes in the cell.

During our run-on analysis, we noticed that PCR analysis of biotin-labeled and affinity purified run-on products obtained from G₂/M nuclei always failed to produce a product in the region containing nucleotides +971 to +1535 (see Fig. S5 in the supplemental material, region labeled #9) even after prolonged PCRs. No such phenomenon was observed for nascent RNA obtained from interphase nuclei. The experiments strongly suggest that there is a new transcription initiation site inside the intron and that there are two transcription termination events, one starting at the regular p53 promoter and terminating a short distance after initiation (before +971) and the other starting at a new initiation site and terminating around +3500. Since we found only one 5′ end of mature mRNA by 5′ rapid amplification of cDNA ends (RACE), we suspected that the new putative promoter was used to transcribe some new RNA species. To examine this possibility, we performed 5′ plus 3′ RACE using an intronic primer near the putative novel promoter. Unexpectedly, this experiment revealed four short poly(A)-terminated RNA species starting from the novel promoter with the body of RNA in intron 1. An examination of the DNA sequence of p53 indeed showed that there are two AAUAAA sites [+3344 and +3507; human p53 contains one such poly(A) signal at the corresponding position] located near the 3′ end of the novel short RNAs in intron 1. Thus, we have uncovered novel short poly(A)-RNA species inside p53 genes that are initiated by a new promoter downstream of the regular promoter. The result suggests that p53 transcript and new RNA species are derived from domain I and not across the gene from G₂/M nuclei. Our run-on experiment using mitotic cell nuclei also suggested that these novel RNA species are continuously synthesized in mitotic cells, and we are now investigating the biological functions of these novel RNA species.

Another interesting finding of this work is the demonstration of bimodal binding of several factors involved in chromatin remodeling and transcription with p53 chromatin. The presence of these factors in the second epigenetic domain suggests that they play some role in the regulation of transcription through this domain. The nucleosome structure presents an obstacle for transcription by RNA pol II, and chromatin remodeling is required to allow transcription elongation through the chromatin-containing nucleosomal structure (1). Our observation that chromatin-remodeling factor Brg1 is located in intron I at the beginning of domain II-containing regular nucleosomal organization but not in the domain I nucleosome-free region suggests that this factor is involved in the control of chromatin remodeling in the second domain for transcription elongation through domain II. The loss of this factor from domain II when transcription is blocked is consistent with this interpretation.

It is interesting to observe that the two Topos have different distributions in p53 chromatin as well as different effects on p53 transcription. Topo I is located only in the intronic region, and inhibition of Topo I with camptothecin resulted in the rapid loss of RNA pol II in domain II, whereas a longer treatment caused the accumulation of pol II at the Topo I binding site in domain II. This result is consistent with the role of Topo I in transcription elongation through domain II but not in transcription initiation. On the other hand, Topo II is found in both the promoter and intronic regions, and the inhibition of Topo II resulted in a rapid depletion of RNA pol II in domain I, including the promoter region. This result is consistent with the role of Topo II in the initiation of p53 transcription.

We hypothesize that the depletion of transcription in epigenetic domain II is the result of impedance of transcription through domain II due to higher-order folding of chromatin in domain II in metaphase. 3C analysis indeed shows that the promoter region of p53 becomes topologically proximal to domain II and the distal 3′ end, most likely as the result of a folding of the chromatin in the second domain. Recently, Tan-Wong et al. presented a 3C analysis showing that activation of the BRCA1 gene by estrogen resulted in the opening of a chromatin loop formed between the promoter and terminator regions of the BRCA1 gene (40). Thus, higher-order folding of chromatin seems to be important for the regulation of transcription. Our results further suggest that higher-order folding of p53 domain II is mediated by histone H1. Higher-order folding in domain II could also result in the loss of regulatory factors associated with domain II. The higher-order folding of domain II is reversed concomitant with the reassociation of regulatory factors in domain II when the cells enter the G₁ cell cycle.

It is interesting that in mitotic cells, epigenetic domain I remains in a nuclease-hypersensitive nucleosome-free conformation, contains active histone modification codes, is free from H1 histone, and is still associated with regulatory factors in mitotic cells. These observations are similar to those reported for the mouse globin gene cluster, namely, that erythroid-specific factor and active histone modifications remained associated with the gene in mitotic cells (47). This result suggests that domain I may serve as a chromatin memory mark of p53 for transcription in the subsequent cell cycle without the need for new initiation (31, 47).