p53 binding to nucleosomes within the p21 promoter in vivo leads to nucleosome loss and transcriptional activation

Abstract

It is well established that p53 contacts DNA in a sequence-dependent manner in order to transactivate its myriad target genes. Yet little is known about how p53 interacts with its binding site/response element (RE) within such genes in vivo in the context of nucleosomal DNA. In this study we demonstrate that both distal (5′) and proximal (3′) p53 REs within the promoter of the p21 gene in unstressed HCT116 colon carcinoma cells are localized within a region of relatively high nucleosome occupancy. In the absence of cellular stress, p53 is prebound to both p21 REs within nucleosomal DNA in these cells. Treatment of cells with the DNA-damaging drug doxorubicin or the p53 stabilizing agent Nutlin-3, however, is accompanied by p53-dependent subsequent loss of nucleosomes associated with such p53 REs. We show that in vitro p53 can bind to mononucleosomal DNA containing the distal p21 RE, provided the binding site is not close to the diad center of the nucleosome. In line with this, our data indicate that the p53 distal RE within the p21 gene is located close to the end of the nucleosome. Thus, low- and high-resolution mapping of nucleosome boundaries around p53 REs within the p21 promoter have provided insight into the mechanism of p53 binding to its sites in cells and the consequent changes in nucleosome occupancy at such sites.

In eukaryotic cells, genomic DNA is tightly associated with histones resulting in the organized and dynamic structure known as chromatin (1, 2). The primary unit of chromatin is the nucleosome, which is composed of approximately 146 bp DNA wound around the core histone octamer (3). The resulting higher-ordered structure helps to compact the DNA within the nucleus. At the same time it represents an accessibility barrier for specific transcription factors whose primary role is to regulate the multistep process of transcriptional activation of most gene promoters in response to pathway-activating stimuli. Many in vitro biochemical studies have revealed significant reduction in transcription factor binding affinities toward their cognate sites within nucleosomal DNA as compared to naked DNA. In vivo, a number of molecular mechanisms may promote specific and efficient interactions between a given transcriptional regulator and its binding site within DNA. Some of these mechanisms depend on the enzymatic activities of chromatin remodeling complexes that facilitate either nucleosome eviction or sliding (4), while others rely on cooperative binding between transcription factors (5), local histone modifications (6), and/or prior nucleosome interactions with so-called “pioneer” factors (7).

p53 is a sequence-specific transcriptional activator that exerts its tumor-suppressor activity primarily through regulation of transcription initiation of multiple downstream target genes (8). Mutations within the DNA-binding domain (DBD) of p53 and subsequent loss of specific DNA-binding activity are responsible for p53 inactivation in more than 50% of tumors. The p53 consensus binding site is quite complex and consists of two decameric half-sites, RRRCA/TT/AGYYY that are usually directly adjacent but can be separated by up to 13 bp (9–11). Multiple studies in recent years have focused on the interaction of p53 with its cognate binding sites in vivo and in vitro and subsequent gene transactivation (or transrepression). Here we have examined the nucleosomal status in vivo of p53 binding sites within one of its major target genes, p21, before and after induction of p53 and have also determined the extent to which p53 is able to interact with its cognate sites within nucleosomal context.

Results

p53-Dependent Loss of Nucleosomes Occurs at p53 Binding Sites Within the p21 Promoter.

We focused on p21 as it is one of the best characterized bona fide p53 target genes. The p21 promoter has two p53 binding sites (or response elements, REs) that conform to the p53 consensus binding sequence (Fig. 1A), the more distal (5′) site at -2283 that binds p53 relatively strongly and the more proximal (3′) site at -1391 that is more weakly bound by p53 (12–14). We first determined the nucleosome status at these sites between matched HCT116 colon carcinoma cell lines that either contain (+/+) or lack (-/-) full-length p53 (15). As expected, treatment with doxorubicin (dox) resulted in a significant increase of p53 levels in HCT116 (+/+) cells and this preceded increases in both p21 mRNA and protein (Fig. S1 in SI Appendix). Activation of p21 expression correlated with p53 binding to its 5′ and 3′ REs as measured by chromatin immunoprecipitation (ChIP) (Fig. 1B). Consistent with the known higher affinity of p53 for the 5′ RE than for the 3′ RE, there was significant basal binding to the former site that was even equivalent to binding to the 3′ site after 4 h of dox treatment. To examine the nucleosome status of the p53 REs within the p21 promoter, we determined the extent to which these regions were resistant to micrococcal nuclease (MNase) digestion as reported for other transcribed genes (16). Cells were treated or not with dox followed by cross-linking with formaldehyde (to preserve chromatin structure) and then nuclei were isolated and incubated with MNase. DNA recovered from the MNase-resistant mononucleosomal fraction was extracted from the gel and subjected to Q-RT-PCR analysis to assess the nucleosome status in the vicinity of the two p53 REs in the p21 promoter region as well as in two control regions: the TATA box at -20 bp and further downstream +11.4 kb. In unstressed HCT116 (+/+) or (-/-) cells both p53 REs had a relatively high nucleosome content when compared to the TATA region that was previously reported to be bound by RNA polymerase II and virtually nucleosome free (17) (Fig. 1C, time “0”). Notably, dox treatment resulted in a rapid p53-dependent loss of nucleosomal content within both p53 REs (Fig. 1 C and D), while we detected no p53-associated increase in MNase sensitivity within any of the two control regions (Fig. 1E).

Fig. 1.

Doxorubicin treatment leads to p53-dependent loss of nucleosomes at p53 binding sites within the p21 promoter. (A) p21 promoter region showing p53 5′ and 3′ REs, TATA region and +11.4 kb control region. Bold arrow represents the p21 transcription start site and head-to-head oriented pairs of arrows represent the sites that were assayed for MNase hypersensitivity by Q-RT-PCR amplification. (B) HCT116 (+/+) cells were incubated with 0.75 μM dox followed by processing for ChIP analysis of p53 binding to p21 5′ and 3′ REs and +11.4 kb region as a negative control. (C) HCT116 (+/+) cells were treated with 0.75 μM dox for indicated time periods as above, followed by formaldehyde cross-linking and processing to determine relative mononucleosomal occupancy at the p21 distal 5′ and proximal 3′ p53 REs as well as the p21 TATA region. MNase-digested chromatin were deproteinized and mononucleosomal DNA was gel-purified for use as templates in Q-RT-PCR with pairs of primers flanking indicated regions within the p21 promoter. (D) Same as in C performed using HCT116 p53 (-/-) cells. (E) MNase hypersensitivity data obtained for two control regions within p21 gene, with relatively high (11.4 kb site), and low (TATA) initial nucleosomal content.

Because dox can induce DNA strand breaks and thereby activate a number of intracellular processes associated with chromatin remodeling independently of p53 (18), we performed the same experiment using Nutlin-3 that disrupts the p53 interaction with Mdm2 and thereby stabilizes p53 in the absence of DNA damage (19). Nutlin-3 induced p21 RNA accumulation that was accompanied by a decrease in MNase-resistant DNA at both p53 REs (Fig. S2 A–C in SI Appendix). The kinetics of nucleosome loss from both p53 REs upon treatment with either p53 stabilizing agent was very similar. Note that dox-induced loss of nucleosomes within the vicinity of the 5′ RE in HCT116 (+/+) cells required the presence of p53, because introduction of p53 siRNA but not control siRNA into these cells prior to drug treatment resulted in significant loss of MNase hypersensitivity within that region (Fig. S3 in SI Appendix).

We asked whether the altered nucleosome occupancy at the p53 REs depended upon ongoing transcription of p21 RNA by performing a similar experiment to the one shown in Fig. 1 but in the presence or absence of the potent RNA Polymerase II inhibitor, α-amanitin (Fig. S4 in SI Appendix). While transcription of several p53-dependent targets such as p21, puma, and mdm2 was blocked by α-amanitin regardless of p53 activation status (Fig. S4A in SI Appendix), MNase hypersensitivity within the p21 REs remained unchanged by the inhibitor (Fig. S4B in SI Appendix). Thus, p53 does not require active transcription to occur in order for it to initiate eviction of nucleosomes within the regions of its binding sites.

Localized p53-Dependent Eviction of Nucleosomes Occurs at the p53 REs Within the p21 Promoter.

We extended our analysis of the nucleosome density changes within chromatin at the p53 REs of the p21 promoter after dox treatment. For this purpose, multiple primer pairs were designed covering 450–500 bp surrounding either the distal or proximal p53 REs (Fig. 2A and Fig. S5A in SI Appendix, respectively). The length of an average amplicon obtained in a Q-PCR reaction with either primer pair was approximately 85 bp and sequences of neighboring amplicons overlapped by approximately 10 bp. Unfortunately, we could not design primers that would cover approximately 60 bp gaps that were 5′ to either the distal or the proximal site (between the first and second primer pairs; see Fig. 2A and Fig. S5A in SI Appendix). For analysis of those, we employed the ligation-mediated (LM) PCR technique (see Fig. 5 below).

Fig. 2.

Doxorubincin treatment alters nucleosome distribution near the p21 distal p53 binding site. (A) Schematic representation of the approximately 500-bp region surrounding the p21 5′ p53 RE together with the corresponding amplicons. Each filled segment represents 20 bp and the open segment is the location of the 5′ p53 RE. Note that there is an approximately 60-bp gap between the first and the second amplicons that could not be assayed to due to difficulties with primer design. (B and C) MNase-resistant DNA in the amplified regions shown in A. Mononucleosomal DNA was prepared as described in the SI Materials and Methods in SI Appendix from either HCT116 (+/+) (B) or HCT116 (-/-) (C) cells treated with 0.75 μM doxorubicin for 0 (filled bars) or 8 h (open bars). (D) MNase-assisted ChIP (see SI Materials and Methods in SI Appendix) was used to measure the relative amounts of H3 histone associated with each amplicon in A in HCT116 (+/+) cells. Data are presented as the ratio of histone H3 at 8 hr post Dox/ histone H3 at 0 hr (untreated).

Fig. 5.

High-resolution mapping of MNase-sensitive sites indicate that the p21 distal p53 binding site is close to the end of a nucleosome in vivo. HCT116 (+/+) p53 cells were treated with 0.75 μM doxorubicin for 0 or 8 hours after which cells were cross-linked with formaldehyde followed by isolation of nuclei and treatment with three different concentrations of MNase. Deproteinized, purified DNA was subjected to LM PCR as described in SI Materials and Methods in SI Appendix. Left panel shows PhosphorImager scan of the 9% polyacrylamide gel that resolved the LMPCR products. AGCT lanes are DNA markers obtained in primer extension reactions with acyclo-ATP, -GTP, -CTP, and -TTP, respectively (New England Biolabs). Graph on right shows densitometry analysis of the gel products at 0 and 8 hours after dox treatment. Putative nucleosome boundaries deduced from MNase-sensitive sites and their locations relative to the p53 5′ RE are depicted below.

Unexpectedly, this low-resolution mapping experiment revealed the presence of two distinct (though closely spaced) sites of MNase sensitivity within the distal p53 RE locus in HCT116 +/+) cells (Fig. 2 B–D). The first coincides with the original p53 RE itself (primer pairs 2 and 3) and the second site is located about 150–160 bp 3′ to the RE (primer pair 5). These alterations required full-length p53 because there were no significant changes within the corresponding regions in HCT116 -/- cells (Fig. 2C). In HCT116 (+/+) cells the relative increases in MNase sensitivity upon treatment with dox were accompanied by a significant loss of histones within both regions as shown by an MNase-assisted ChIP experiment (Fig. 2D). Interestingly, between these two hypersensitive regions we also detected a site that became more resistant to MNase under conditions that activate p53, and MNase-assisted ChIP analysis of that region revealed that it, too, had a relative increase in local histone occupancy (primer pair 4).

Although p53 binds far more weakly to the proximal RE in the p21 promoter, the MNase sensitivity profile within that region was very similar to that seen at the stronger distal RE element (compare the data from Fig. 2B with Fig. S5B of SI Appendix). This region showed two MNase-sensitive sites (primer pairs 2, 3, and 6) separated by a region where there was no significant change in MNase-resistant DNA (primer pairs 4/5). There were also corresponding changes in histone H3 occupancy within this region. As with the 5′RE region, there were only very modest dox-dependent changes in nucleosome distribution at the p21 proximal RE in the absence of full-length p53 (Fig. S5C in SI Appendix).

p53 Is Bound to Nucleosomal DNA in Unstressed HCT116 Cells.

Although the above mapping experiments elucidated features of the chromatin organization within the distal and proximal p53 REs, they did not reveal whether p53 can bind directly to mononucleosomal DNA in vivo. To address this mononucleosomes prepared from formaldehyde-crosslinked chromatin by limited digestion with MNase were immunoprecipitated with either anti-p53 or anti-H3 antibodies. The recovered mononucleosomal DNA was amplified by Q-PCR using primer pairs specific to the p21 p53 distal and proximal REs, or the two negative control regions at the TATA and +11443 regions (Fig. 3). The results strongly indicate that in unstressed cells p53 is bound to MNase-resistant DNA and, depending on the site, such binding is either greatly decreased or completely lost upon treatment with dox, which is mirrored by significant loss of histones within the p53 REs. Supporting this likelihood, the signal at the REs was markedly greater than those with either the TATA or +11443 regions where there was also very little change after dox treatment. Note that while the amount of specific mononucleosomal DNA immunoprecipitated with the p53 antibodies was reduced upon dox-dependent p53 activation (as measured by MNase-assisted ChiP), total p53 binding to the p21 REs was increased over the same time period after dox treatment (compare the data from Fig. 3 and from Fig. 1B). Thus, loss of the signal upon p53 activation in the MNase-assisted ChiP assay simply reflects the loss of nucleosomes but not the loss of overall p53 DNA-binding per se.

Fig. 3.

p53 is bound to nucleosomal DNA in cells prior to treatment with doxorubicin. MNase-assisted ChIP was performed using HCT116 (+/+) cells treated with 0.75 μM doxorubicin for 0 (filled bars) or 8 (open bars) hours. Q-RT-PCR was used to measure MNase-resistant DNA immunoprecipitated with either anti-p53 or anti-H3 antibodies as indicated at (A) p21 5′ p53 RE, (B) p21 3′ p53 RE, (C) p21 TATA region, or (D) p21 +11.4 region. Dotted line represents the background signal for each experiment.

Stable Nucleosomal Binding by p53 Requires That Its Site Be Close to the End of Nucleosomal DNA.

With few exceptions biochemical studies on p53/DNA interactions have been examined in the context of naked DNA. We therefore sought to recapitulate our observations showing that p53 binds to its sites within chromatin at the p21 promoter by examining how p53 interacts with mononucleosomes in vitro. We performed DNase I footprinting using either naked or reconstituted mononucleosomal DNA spanning the p21 promoter p53 REs (Fig. S6 in SI Appendix). A strong nucleosome positioning sequence derived from the yeast HSP82 promoter region was incorporated into a chimeric 170 bp DNA construct that had the p21 distal p53 RE located 30 bp from the 5′-end (see SI Materials and Methods in SI Appendix for details). As expected, preincubation of naked DNA with increasing amounts of purified wild-type p53 protein resulted in strong protection from DNase I cleavage within the RE region (Fig. S6A in SI Appendix). As described previously (20), when the same DNA was present within a mononucleosome, DNase I cleavage yielded a characteristic approximately 10-bp periodicity pattern that revealed the contacts between the DNA and the core histone octamer (Fig. S6B in SI Appendix). p53 binding to this mononucleosomal DNA resulted in clear but relatively weak protection of the p21 gene distal RE (see densitometry analysis) along with an appearance of a DNase I hyper-sensitive site located at the 3′ edge of the RE sequence (indicated by asterisk).

An electrophoretic mobility shift assay (EMSA) was also performed to characterize p53 interactions with mononucleosomal DNA in vitro (Fig. 4). Our initial experiments revealed that p53 did not detectably bind to such DNA from the p21 promoter when the p53 5′ RE was centrally positioned (Fig. S7 in SI Appendix) so we assessed whether p53 might bind to its site at other positions within mononucleosomal DNA. To this end, we prepared a set of ten 170-bp chimeric DNA templates containing the p21 p53 5′ RE located at positions that differed by one base pair in their distance from the 5′-end of the DNA (Fig. 4A). EMSA analysis showed that p53 binding to mononucleosomes became detectable and increased as the RE was moved further from the center of the DNA. In fact, a 2-bp shift in the RE (170–30) resulted in an approximately fivefold increase in binding to mononucleosomal DNA when compared to a mononucleosome where the RE was located 32 bp from the end of the DNA (170–32), and a 9-bp shift in the RE toward the center (170–39) reduced binding by more than a factor of 10 (see Fig. 4 B and C, graph). Thus, the distance between the binding site and the end of the DNA represents a critical parameter influencing binding of p53 to mononucleosomal DNA. A similar dependence of binding to the site within nucleosomal DNA was demonstrated before for other sequence-specific proteins (e.g., for restriction endonucleases) (21).

Fig. 4.

Detectable p53 binding to the p21 distal RE requires the site to be positioned close to the end of a mononucleosome. (A) The location of p21 distal p53 REs inserted within 170 BP DNA fragments from yeast HSP82 sequence containing a strong positioning sequence assembled into mononucleosomal DNA by the octamer transfer method. (B) An electrophoretic mobility shift assay was performed with [³²P] labeled mononuclesomal DNAs shown in A and purified p53 protein. Four different DNA/mononucleosomes were run on one gel, and then construct (170–30) was re-run on each gel to correct differences between the experiments. Shown are PhosphorImager scans of three 4% native 0.5X TBE gels with 10 mononucleosomal constructs. PhosphorImager scans in B were analyzed using ImageQuant Software and presented as graphs (C).

Mapping the Location(s) of the Distal p53 Binding Site Within a Nucleosome in the p21 Promoter in Vivo.

We next sought to determine the location of the p21 distal p53 binding site within the relevant mononuclesome in vivo by performing linker-mediated PCR (LMPCR) in HCT116 + /+ cells (Fig. 5). This analysis revealed several major primer extension products (numbered 1–5) located at different distances from the p53 RE that we interpret as representing nucleosomal boundaries. In Fig. 5 products numbered 1 and 2, that are 107 and 68 bp from the p53 5′ RE, respectively, likely represent the boundaries of two neighboring nucleosomes. Additionally, three products of weaker intensity (numbered 3, 4, and 5) were situated much closer to the p53 RE at 41, 32, and 18 bp, respectively. Indeed, consistent with our in vitro binding data, the location of these products positions the p53 binding site close to the edge of the nucleosome, thus making it more accessible for p53. Further, all above-mentioned species were reduced upon dox treatment, which is in good agreement with the data obtained from the low-resolution mapping experiment shown in Fig. 2B. This pattern of LMPCR extension products suggests that the nucleosome at the distal p53 RE is somewhat heterogeneously positioned. We assume that formaldehyde cross-linking prior to MNase treatment helped us to catch a nucleosome that contains the p53 distal RE in unstressed cells in several most-probable positions (i.e., positions 2, 3, 4, and 5). Analysis of the proximal 3’ p53 RE by LMPCR also revealed multiple species (the most prominent one situated about 55 bp away from the RE) that again likely represent multiple potential nucleosome boundaries (Fig. S8 in SI Appendix). Here too there was a significant loss of intensity of the major band/boundary and almost total vanishing of minor bands 8 h after administration of dox.

Discussion

p21 was one of the first genes found to be positively regulated by p53 (12, 22). Although a number of studies have addressed the binding of p53 to its distal and proximal binding sites within the p21 promoter in vitro and in vivo, none has interrogated such binding in the context of mononucleosomal DNA in detail. In this study we have gained insight into chromatin organization in the p21 promoter, in particular within regions spanning its two p53 REs. By applying ChIP, MNase hypersensitivity assays, MNase-assisted ChIP, and LM PCR, we demonstrate that p53 REs within the p21 promoter in unstressed HCT116 cells are localized within nucleosomal DNA. Further, nucleosomes are rapidly lost upon activation of p53 with either the DNA-damaging agent doxorubicin or by blocking its degradation by Mdm2 through administration of Nutlin-3. Changes in hypersensitivity within both p53 REs of p21 promoter may be explained by either nucleosome sliding or nucleosome eviction. After p53 activation by dox, within the vicinity of the p21 5’ RE there is an apparent loss of two nucleosomes (Fig. 2, amplicons 2 and 5) and between them there is a relative increase in H3 occupancy within amplicon 4. Yet the distance covered by amplicon 4 is not long enough to contain a full nucleosome core particle (i.e., consisting of the cannonical histone octamer). Conceivably partial loss of a portion of the histones from that nucleosome and/or a subsequent clash between two neighboring nucleosomes has occurred. This is not unheard of—see, for example, Dechassa et al. (23). Using in vitro reconstituted mono-, di-, and three-nucleosome arrays in the presence of SWI/SNF, those authors demonstrated collision of two neighboring nucleosomes, as well as partial loss of H2A/H2B dimer. As discussed below, the presence of certain histone variants (H3.3 and H2A.Z, in particular) and their effect on nucleosome stability may also be in play here.

Does p53 bind to nucleosomal DNA and cause loss of nucleosomes (directly by destabilizing multiple histone-DNA contacts, or indirectly by bringing down components of chromatin remodeling machinery), or does nucleosome loss precede and is necessary for p53 sequence-specific binding? The necessity of nucleosome removal for gene activation was suggested more than two decades ago. Since then, numerous in vitro binding experiments performed on reconstituted nucleosomal substrates have almost exclusively supported the idea that efficient binding of a given transcription factor to its RE is greatly reduced in the context of a nucleosome. Factors that were shown to interact with nucleosomes (e.g., TFIIIA, Sp1) have significantly reduced binding affinities when compared to naked DNA (5). Given the complexity of p53 REs and multiplicity of contacts between the p53 core domain and DNA sequences within each RE (24, 25), we assumed that p53 would not bind detectably to its sites when wrapped around the histone octamer. Surprisingly, our MNase-assisted ChIP experiment performed on p21 distal and proximal REs demonstrated that p53 in unstressed cells is bound to nucleosomal DNA. What is the significance of such binding when p53 becomes activated by stress signals? We envision two possible scenarios. In the first, increasing amounts of stabilized p53 transiently compete with a nucleosome for the RE. In this case, when p53 levels are increased (e.g., after DNA damage) the relatively low affinity of p53 for its binding site within the nucleosome might still be enough to result in efficient transcriptional activation in vivo. Indeed, recent experiments using fluorescence recovery after photobleaching (FRAP) have revealed that the nature of transcription factor (TF) binding to chromatin is very transient, and the time that a given TF spends on its RE in vivo may be less than a minute (reviewed in ref. 26). This time frame may be sufficient to bring down the components of modifying and/or remodeling machinery leading to local nucleosome eviction/displacement, thereby culminating in formation of a competent preinitiation complex and subsequent activation of transcription. Experiments that employ FRAP and/or single molecule technology may provide estimates of the half-life of p53 bound to its REs within nucleosomes or chromatin.

The second scenario depends on the ability of nucleosomes to adopt multiple positions within certain regions, which is dictated both by the DNA sequence and components of the chromatin remodeling machinery (reviewed in ref. 27). In this case, initial p53 binding will occur under the most favorable conditions, such as positioning of the RE in relatively close proximity to the end of the nucleosome. Indeed, our LM PCR and EMSA experiments support this theory. Interestingly, large scale genome analysis of data on the distribution of nucleosomes within the p21 promoter in either A375 or MDA-kb2 cell lines indicates that both p53 REs are situated in regions with fading nucleosomal density, very close to a relatively nucleosome-depleted part of the promoter (UCSC Genome browser on Human, March 2006 Assembly; NCBI36/hg18). Our low- and high-resolution mapping experiments correlate well with these findings. We acknowledge that nucleosomal profiles differ between cell lines, which could be related to changes in DNA such as mutations, deletions, or amplifications, as well as epigenetic changes, all of which may contribute to the chromatin landscape within a given cell, tissue or organism.

In addition to showing that p53 is able to bind to nucleosomal DNA both in vitro and in vivo, our study raises the possibility of a potential second p53 binding site that would reside approximately 150–160 bp downstream of the bona fide distal p21 RE (Fig. 2, primer pair 5). This is supported by increased p53-dependent MNase sensitivity and p53-dependent loss of core histones as well as the presence of three overlapping weak p53 half-sites within that region. Whether the putative second binding region exists and is functional in vivo remains to be answered by future experiments.

Local histone modifications and/or histone variants may also affect p53 binding to nucleosomes. For example, the histone H2 variant, H2A.Z, is enriched at p53 REs within the p21 promoter and DNA-damage stress caused by doxorubicin leads to eviction of H2A.Z in a p53-dependent but transcription-independent manner (28). The presence of H2A.Z may positively influence p53 binding to mononucleosomes because the H2A.Z/H2B histone dimer is less stable than the regular H2A/H2B dimer and can be relatively easily released from nucleosomes (29). Experiments involving in vitro p53 binding to the RE localized within mononucleosome reconstituted with either H2A or H2A.Z may provide insight into the relative affinities of such substrates toward p53.

A different mechanism for p53-nucleosome interaction was proposed by Sahu et al. based on the results of an elegant in vitro study (30). The authors suggest that orientation of the p53 binding site on a nucleosome in conjuction with a nucleosomal positioning sequence may preset the p53 RE in a way that is easily accessible to p53. Though attractive, confirmation of this model requires support from experiments on p53 binding to bona fide REs in the context of their natural settings in vivo.

Certain p53 REs may localize within nucleosome-depleted regions. In these cases p53 binding may be a different and less complex process and could be regulated by different mechanism(s) such as relative strength of the RE, local concentrations of p53, modification status of p53, or cooperative binding with other transcription factors. Indeed, the results of MNase hypersensitivity analysis performed on several p53-dependent promoters in HCT 116 +/+ and -/- cells rather support this assumption (Fig. S9 in SI Appendix). The levels of basal nucleosome occupancy differ drastically at p53 REs within the mdm2P2, bax, puma, or PCNA promoters: Whereas the first two are virtually nucleosome free, bax and (even more so) puma p53 REs show strong MNase resistance. Interestingly as well, there is relatively higher basal nucleosome occupancy detected within p53 REs of these promoters in HCT116 (-/-) than in HCT116 (+/+) cells, particularly within the mdm2P2 and pcna promoters (2.8-fold and 2.1-fold, respectively). Moreover, the level of nucleosome occupancy within the TATA box region of the p21 promoter was significantly elevated in p53 - /- cells (approximately 2.0-fold to 2.3-fold). This indicates an important but poorly understood role of p53 in formation of the chromatin landscape within the promoters of p53-induced genes that merits further detailed investigation.

While this manuscript was in preparation, Lidor et al. published an interesting study in which, using a custom DNA microarray, they analyzed the distribution of approximately 2,000 p53 binding sites within chromatin in unstressed vs. stressed cells and their relative affinities to p53 (31). They made the unexpected observation that p53 binding sites reside preferentially within genomic regions with relatively high intrinsic nucleosome occupancy. They showed as well that upon DNA damage nucleosomes are partially and reversibly displaced from a region surrounding bound p53 sites. However, these authors did not directly address whether or how p53 binds directly to nucleosomes. Our study, though limited to the p21 p53 REs, has both delved in more detail into the nucleosomal organization of the p21 distal and proximal REs in vivo and revealed that p53 binds directly to nucleosomal DNA both in vivo and in vitro. Our work has also suggested possible mechanisms by which p53 does so and sets the stage for future mechanistic experiments that can reveal in greater detail the events related to p53 binding to its cognate sites in cells.

Methods

Details of methods used in this paper including cell culture, protein expression and purification, MNase based experiments, ChIP, ligation-mediated PCR, and in vitro nucleosome binding experiments are described in SI Materials and Methods in SI Appendix.