Oncogene (2010) 29, 2217–2227 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Epigenetic regulation of the human p53 gene promoter by the CTCF transcription factor in transformed cell lines E Soto-Reyes and F Recillas-Targa Instituto de Fisiologı´a Celular, Departamento de Gene´tica Molecular, Universidad Nacional Auto´noma de Me´xico, Me´xico D.F., Me´xico Epigenetic silencing of tumor suppressor gene promoters cataloged as a low, intermediate or high CpG island has become a more frequent phenomenon in cancer than (Weber et al., 2007). High CpG islands have been linked previously anticipated. In this study we addressed the to housekeeping gene promoters that are normally mechanisms involved in the protection of the p53 tumor unmethylated, whereas low CpG islands are associated suppressor gene against epigenetic silencing in human with tissue-specific gene promoters that can be methy- transformed cell lines. We characterized a binding site for lated, apparently without affecting their regulatory the CCCTC-binding factor (CTCF) in the human p53 potential. This classification demands novel interpreta- gene promoter that contributes to its transcriptional tions depending on the nature of the CpG island, as a expression, and has the ability to maintain this regulatory distinct combination of epigenetic silencing mechanisms element in a local open chromatin configuration. In the can now be expected (Weber et al., 2007; Kondo et al., absence of CTCF we observe the incorporation of 2008; Meissner et al., 2008). For example, the high repressive histone marks, such as H3K9me3, H3K27me3 CpG island promoters are more frequently silenced and H4K20me3, in different sub-domains of the upstream by DNA methylation in abnormal conditions, even regulatory sequence. This evidence suggests that CTCF though histone repressive marks, such as H3 lysine 9 protects the p53 gene promoter against repressive histone trimethylation (H3K9me3), can also be incorporated marks. Notably, no apparent direct correlation between (McGarvey et al., 2006; Kondo et al.,2008).Incontrast, repression and DNA hypermethylation has been detected. intermediate and low CpG islands can be silenced Together, we present evidence supporting the relevant role by covalent histone modifications such as H3 lysine 27 of CTCF in the epigenetic regulation of tumor suppressor trimethylation (H3K27me3) (Kondo et al.,2008), genes and cancer. We propose that CTCF is a strategic and not by DNA methylation. Thus, H3K27me3 and component responsible for the maintenance and segrega- DNA methylation can exert an effect independently, and tion of epigenetic traits. their aberrant promoter targets seem to be defined by the Oncogene (2010) 29, 2217–2227; doi:10.1038/onc.2009.509; CpG content (Weber et al., 2007; Kondo et al.,2008). published online 25 January 2010 In this study we assessed the mechanisms of epigenetic silencing of the human p53 tumor suppressor gene in Keywords: tumor suppressor gene; CTCF; epigenetics; cancer cells. On the basis of recently established DNA methylation; p53 human promoter and histone definitions, we consider the human p53 promoter to covalent modifications correspond to an ICP island that is apparently not frequently silenced by DNA methylation (Pogribny et al., 2000; Kang et al., 2001; Amatya et al., 2005). We hypothesize that different classes of CpG islands Introduction might be regulated differently, not only at the level of DNA methylation and/or Polycomb family members From an epigenetic perspective, the study of tumor but also in terms of selective and differential use of suppressor genes has become notably relevant due to the transcription factors. In accordance, in this study we fact that deregulation of these genes is associated with address whether different kinds of CpG island promo- tumorigenic processes (Jones and Baylin, 2007). In ters are epigenetically regulated by alternative mechan- particular, 70% of human genes are linked to promoters isms (De La Rosa-Vela´zquez et al., 2007). We found that contain CpG islands that are the target of abnormal that the multifunctional CCCTC-binding site factor epigenetic events (Antequera, 2003; Issa, 2004). On the (CTCF) is involved in p53 epigenetic regulation by basis of recent proposals, the CpG island can be harboring an open chromatin configuration. Remark- ably, p53 gene promoter epigenetic silencing is not Correspondence: Dr F Recillas-Targa, Instituto de Fisiologı´a Celular, dependent on DNA methylation. Instead, repressive Departamento de Gene´tica Molecular, Universidad Nacional Auto´n- histone modifications are enriched and represent the oma de Me´xico, Apartado Postal 70-242, Mexico City, D.F., 04510, main cause of silencing in the presence of reduced levels Me´xico. E-mail: [email protected] of CTCF. We propose a model in which CTCF can Received 18 May 2009; revised 3 December 2009; accepted 13 December overcome epigenetic silencing not only by counter- 2009; published online 25 January 2010 acting DNA methylation (as in the case of Rb and other Epigenetic regulation of the p53 tumor suppressor gene E Soto-Reyes and F Recillas-Targa 2218 tumor suppressor gene promoters) but also by blocking Surprisingly, when we performed a sodium bisulfite the incorporation of repressive covalent histone conversion assay and sequencing using primers located modifications. in the core promoter we were not able to detect any DNA methylation (Figure 3a, site 4, and Supplementary Figure 1). On the basis of such unexpected result we decided to survey the DNA methylation status of a Results larger genomic area in T98G cells (Figure 3a). In all, seven DNA sub-fragments were analyzed using sodium Gene expression and epigenetic profiles of the human bisulfite conversion and sequencing, including upstream p53 gene in epithelial cell lines genomic sequences in relation to the p53 transcription p53 On the basis of predicted differences in gene start site and CpG-rich intronic sequences (Figure3a and expression in transformed epithelial cell lines, we Supplementary Figure 1). As positive controls we tested decided to explore these differences and correlate them two unrelated genomic sequences known to possess with their chromatin structural status with the aim of different levels of DNA methylation and in vitro finding a model system for p53 epigenetic silencing. We methylated genomic DNA (Supplementary Figure 2). first characterized the expression profile of the p53 gene One more time, this extended analysis did not allow the using reverse transcriptase PCR (RT–PCR) in SW480, identification of any DNA methylation. These results HeLa, Kb epithelial cell lines and in a primary culture of confirm on one side that p53 gene expression can be human fibroblasts obtained from neonatal foreskin reactivated using the DNA methylation inhibitor in (Figure 1a). We found that the colorectal adenocarci- glioma cell lines, and on the other, that the core noma SW480 cells show the highest levels of expression, promoter and surrounding sequences in a range of followed by HeLa and Kb cell lines, when compared 3 kb are hypomethylated. Therefore, these two glioma with the primary culture of normal fibroblasts. cell lines may represent a cancer cell context in which p53 To analyze the relationship between gene expres- p53 epigenetic silencing can be studied, and whether the sion and its epigenetic entourage, we next evaluated the presence of CTCF contributes to such a process can also DNA methylation status of the p53 gene promoter by be explored. the sodium bisulfite conversion assay (Figure 1b). No DNA methylation was detected at the p53 gene core promoter in any of the cell lines studied. We then In vitro and in vivo binding of CTCF to the upstream evaluated the patterns of histone covalent modifications region of the human p53 gene promoter over the p53 gene promoter. We found that the histone On the basis of our previous report suggesting that H3K4me2 modification and the polyacetylation of CTCF is able to counteract epigenetic silencing events in histone H3 and H4 are drastically enriched in SW480 the human Rb gene promoter, we decided to verify cells (Figure 1c). To corroborate the in situ chromatin whether such a mechanism is conserved in other tumor status of the promoter we performed a restriction suppressor genes, and particularly for p53 (De La Rosa- enzyme accessibility assay (Escamilla-Del-Arenal and Vela´zquez et al., 2007). We found a putative CTCF- Recillas-Targa, 2008) (Figure 1d). The results of these binding motif in a DNA fragment located at B700 bp experiments are in accordance with the previous assays, upstream of the p53 transcription start site (Figure 3a) and they show that the p53 gene promoter has an open (De La Rosa-Vela´zquez et al., 2007; Kim et al., 2007). chromatin conformation in the SW480 cell line, whereas We performed a gel shift analysis using total extracts the HeLa and Kb cell lines show a more compacted from HeLa cells to determine whether CTCF associates chromatin structure (Figure 1d). Taken together, these with this sequence in vitro (Figure 3b). We next carried results suggest that p53 gene expression regulation in out a chromatin immunoprecipitation (ChIP) assay normal and cancer cells can be influenced not only by using an antibody against human CTCF and by using ubiquitous or stage-specific transcription factors, but primers that flank the novel CTCF-binding sequence in also by the chromatin configuration of its promoter. several cell lines. As a negative amplification control we used the exon 27 of the Rb gene, as reported previously Epigenetic silencing model for the p53 gene promoter (De La Rosa-Vela´zquez et al., 2007). Interestingly, we in glioma cell lines observed enrichment of the CTCF immunoprecipitated We next decided to explore other cell types that have fraction in SW480 and HeLa cell lines, but not in the been reported to show DNA methylation over the p53 U87MG and T98G glioma cells lines in which p53 gene gene promoter (Amatya et al., 2005).