Epigenetic Regulation of the Human Retinoblastoma Tumor Suppressor Gene Promoter by CTCF Free

The retinoblastoma (Rb) gene is one of the most important tumor suppressor genes (1–4). Mutations and deletions of this gene have been associated with a number of inherited malignancies, mainly retinoblastoma. In recent years, it has become apparent that carcinogenesis is also driven by epigenetic mechanisms. A substantial number of investigations have shown that epigenetic alterations such as gain or loss of DNA methylation and altered patterns of histone modification occur in most cancers (5, 6).

The mechanisms that regulate the epigenetic control of Rb expression have not been elucidated, and it would be important to understand how the Rb promoter in normal cells remains unmethylated to guarantee its transcription and which molecular mechanisms might be affected in malignant transformation because the human Rb gene promoter is hypermethylated in several human tumors and the extent of methylation correlates with the progression of tumorigenesis (7). As a working model, we raise the possibility of the presence of a boundary element that would shield a CpG-island promoter, not only against DNA methylation but also against the incorporation of other repressive chromatin marks.

Recent evidence points toward a role for the 11-zinc finger CCCTC-binding factor (CTCF) in the establishment of DNA methylation free zones (8, 9). CTCF is involved in enhancer blocking by insulators, is also an important component in the determination of epigenetic control of diverse imprinted loci, and participates in promoter activation and repression (8, 10, 11). Recent reports indicate that CTCF can regulate the expression of cell cycle–related genes (12–14). Interestingly, a CTCF recognition motif has been identified at the p19^ARF and BRCA1 gene promoters (13, 14). In particular, CTCF defines a DNA methylation transition zone at the human BRCA1 gene promoter, but its contribution to regulation has not been addressed (13).

We investigated whether CTCF is involved in epigenetic regulation of the human retinoblastoma gene. Detailed promoter sequence analysis allowed us to identify a GC-rich sequence of 53 bp with significant homology to previous characterized CTCF binding sequences. Functional assays, including CTCF knockdown, showed that CTCF is an epigenetic regulator of Rb gene promoter activity. Rapid transgene extinction, when the site is mutated, supports a protective role for CTCF against epigenetic silencing. Consistent with this, when the promoter is hypermethylated, CTCF binding is lost, and the site is recognized by the methyl-CpG–binding protein Kaiso. These results establish that CTCF may represent an epigenetic component needed for the chromatin structure and functional integrity of CpG islands such as the human Rb gene promoter.

Plasmid constructs. Plasmids pGLRb containing the complete human Rb gene promoter (genomic positions 1634–2020, GenBank accession number L11910) and pGLRbΔCTCF (lacking the CTCF site in the promoter) were generated by amplification from human lymphocyte genomic DNA with Fp-Rupp or FTRb-Rupp primers and cloned in the pGL3basic vector. The plasmids RbΔCTCF+1×, RbΔCTCF+2× and RbΔCTCF-FII were constructed by cloning the RbCT or FII DNA sequences used for electrophoretic mobility shift assay (EMSA) upstream of the promoter in the pGLRbΔCTCF plasmid. Fp: 5′-CGGGATCCAGACTCTTTGTATAGCC-3′; Rupp: 5′-CGGGATCCCGAGCTGTGGAGGAG-3′; and FTRB: 5′-CGGGATCCTCGCGGACGTGACGC-3′. Mutant constructs pGLRbmutE, pGLRbmutG (luciferase reporters), and pERbmutE (GFP reporter) were generated by two-step PCR using EMSA oligonucleotides. All plasmid constructs were checked by DNA sequencing.

Cell culture. HeLa cells were grown in DMEM medium containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Life Technologies, Gaithersburg, MD). Human lymphocytes were obtained from blood of a healthy donor, isolated with Ficoll-Plaque Plus (Amersham, Uppsala, Sweden) following the manufacturer's instructions and cultured in DMEM plus 10% FBS and 1% penicillin/streptomycin. The lymphocytes were treated with phytohemaglutinin (Life Technologies) for 3 days to stimulate cell proliferation. K562 cells were grown in IMEM medium (Invitrogen, San Diego, CA) containing 10% FBS and 1% penicillin/streptomycin.

DNA methylation analysis. Bisulfite analysis was done as previously described (15). The PCR primers for stable cell lines were designed against the plasmid sequences to avoid amplification of the endogenous promoter. Nested PCR was done with primers EGFPbis1-EGFPbis2, and the second round of PCR amplification was done with EGFPbis3-EGFPbis4 primers. The product from the second PCR was gel purified, and a third PCR was done with Rb11-Rb12 primers specific for the Rb promoter (7). PCR products were cloned in pGEM-11zf vector (Promega, Madison, WI) for sequencing using the T7 primer. Primers used were EGFPbis1: 5′-TTTGGTTTTTTGTTGGTTTTTTGT-3′ and EGFPbis2: 5′-AAATAAACCAAAACACCAACAAC-3′; EGFPbis3: 5′-CGGGATCCTTTTTTTTGTGTTATTTTTTG-3′ and EGFPbis4: 5′-CGGGATCCAAATCAACTTACCCTAAATAAC-3′.

Transient and stable transfection of HeLa and K562 cells. In different experiments with constructs containing firefly and Renilla luciferase gene reporters, 80% confluent HeLa cells were transfected with 1 μg of each plasmid and 50 ng of the Renilla plasmid (pRL-CMV) for normalization. For CTCF cotransfection, 500 ng of the plasmid pCI-7.1 (kindly provided by Elena Klenova, University of Essex, Essex, United Kingdom) containing the full-length human CTCF cDNA was used. All transfections were carried out with LipofectAMINE 2000 (Invitrogen). Luciferase activities were determined 48 h later using the Dual Luciferase kit (Promega). Relative luciferase units were measured in a TD 20/20 luminometer (Turner Designs, Sunnyvale, CA). For HeLa and K562 stable transfection, linearized pERb, pERbΔCTCF, or pERbmutE containing the Fp-Rupp, FTRb-Rupp, or full promoter with mutation E fragments, respectively; the green fluorescent protein (GFP) and the neomycin resistance cassette (pEGFP-1; Clontech, Palo Alto, CA) were used. After selection, neomycin-resistant clones were isolated and analyzed by fluorescence-activated cell sorting (FACS). Clones were subsequently cultured in the absence of neomycin in medium for up to 23 weeks. The integrity of the transgene was checked by Southern blot (data not shown). For reactivation experiments, stable cell lines were treated with 5-aza-2'-deoxycytidine (3 μmol/L), trichostatin-A (5 ng/mL), or both for 3 days. At least three independent reactivation experiments were done.

Transient small interfering RNA transfection of HeLa cells. HeLa cells were treated with human CTCF small interfering RNA (siRNA) (Santa Cruz Biotechnology, Santa Cruz, CA) or siGFP-443 (kindly provided by Luis Vaca, Instituto de Fisiología Celular, Universidad Nacional Autónoma de Mexíco, Mexíco, D.F, Mexíco) every 24 h for 3 days using LipofectAMINE 2000. At day 3, pGLRb and pRL-CMV (Renilla) vectors were cotransfected. Cell pellets were harvested at day 4, and whole cell lysates were analyzed for luciferase activity.

Western blotting. Whole cell lysates (50 μg/sample) were analyzed for CTCF presence by Western blotting using goat anti-CTCF(N17) (1:500 dilution) and with rabbit anti-actin (H300; 1:500, both from Santa Cruz Biotechnology) as control.

In vitro transcription/translation. The full-length human Kaiso cDNA (kindly provided by Jieming Wong, Baylor College of Medicine, Houston, TX) was transcribed/translated in vitro using the TnT reticulocyte lysate-coupled system following the kit instructions (Promega).

Electrophoretic mobility shift assay. The EMSA assay and nuclear extract preparation were done as previously described (16, 17). Competitions were carried out with 200 pmol of gel-purified unlabeled oligonucleotides. In vitro methylation of the probe was carried out using the SssI methylase (New England BioLabs, Beverly, MA) corroborated by HpaII/MspI digestion. Supershift experiments were done using 1 μg of the following antibodies: CTCF(BD) (clone 48, BD Bioscience, San Jose, CA), cCTCF(86-233) (at high concentration, this antibody partially reacts with the human CTCF; ref. 16), CTCF(N-17), MBD2(N18), and Sp1(pep2) (all from Santa Cruz Biotechnology), Kaiso(clone 6F) and MeCP2 (Upstate, Charlottesville, VA). Primers used (showing only the top strand) were RbCTCF: 5′-CGCCCCAGTTCCCCACAGACGCCGGCGGGCCCGGGAGCCTCGCGGACGTGACG-3′; FII: 5-CCCAGGGATGTAATTACGTCCCTCCCCCGCTAGGGGGCAGCAGGCGCGCCT-3; and Sp1: 5′-ATTCGATCGGGGCGGGGCGAGC-3′.

Chromatin immunoprecipitation. The chromatin immunoprecipitation (ChIP) assay was done as previously reported (16) with 4 μg of antibodies against CTCF(N17), Kaiso(clone 6F), and acH3 and H3K4me2 antibodies from Upstate, and H3K9me1-3 and H3K27me3 antibodies were kindly provided by Thomas Jenuwein (Research Institute of Molecular Pathology, Vienna, Austria). Immunoprecipitated DNA was analyzed by PCR using primers specific for the endogenous Rb gene promoter (Fp-Rupp) or Fp-Hec02r primers for the promoter and plasmid DNA for the stable cell lines to discriminate ectopic and endogenous promoters and RTRb-F/RTRB-R for Exon 27. For semiquantitative ChIP assay, duplex-PCR was done as described in ref. 18. Primer used were HEC02R: 5′-ACCATGGTGGCGACC-3′; RTRb-F 5′-AAGTACCCATCTAGTACT-3′; and RTRb-R: 5′-AAGTTACAGCATCTCTAAA-3′. Primers sequences are available by request.

In vitro binding of CTCF to the conserved sequence on the human Rb promoter. To investigate the possible participation of CTCF in transcriptional control of the human Rb gene, we searched for CTCF binding sequences along 2 kb upstream of the Rb gene transcription start site. Fixing the 5′-CTAG-3′ motif from the chicken cHS4 β-globin insulator FII CTCF binding site as a reference (16, 19), we found a putative CTCF binding sequences in the human retinoblastoma promoter region located immediately upstream of the group of transcription factor binding sites responsible for Rb promoter activity (Fig. 1A).

To show whether the novel identified sequence is capable of binding CTCF in vitro, EMSA were employed. A 53-bp oligonucleotide (Rb-CTCF) was synthesized, and particular attention was paid to exclude any binding sites for nuclear factors located on its 3′ side (Fig. 1A). The Rb-CTCF probe was able to generate a retarded complex using increasing amounts of nuclear extracts from HeLa cells (Fig. 1B). The complex formed can be competed by the chicken FII CTCF sequence (Fig. 1B,, compare lanes 8 and 9). To show the specificity of in vitro CTCF binding, we carried out heterologous competition with a Sp1 consensus sequence (Fig. 1C and D). As expected, unlabeled Rb-CTCF and the FII oligonucleotides were able to compete CTCF-DNA complex formation, whereas the Sp1 consensus did not (Fig. 1C, lanes 3–5). Next, a supershift experiment was done using an antibody raised in our laboratory against CTCF [α-cCTCF(86-233); ref. 16] and an Sp1 antibody as heterologous control. Significant reduction of the intensity of the retarded complex was observed with the CTCF antibody, whereas no change was seen with the Sp1 antibody (Fig. 1C,, compare lanes 6 and 7). To confirm this result, two independent commercial antibodies against human CTCF [α-CTCF(N-17) and α-CTCF(BD)] were tested. The same result was reached, but in this case, a supershifted complex was seen using the α-CTCF(N-17) antibody (Fig. 1D , lane 6). In other studies, we observed DNase I footprints encompassing the 53-bp sequence that we defined as a potential CTCF binding motif (data not shown), consistent with other work (20). Thus, our in vitro data support the novel interaction of CTCF with a sequence present in the human Rb gene promoter.

In vivo occupancy of CTCF to the human Rb gene promoter. To investigate the in vivo association of CTCF with the human Rb gene promoter in transformed and normal cells, ChIP assay was done using antibodies against CTCF, employing primers amplifying the Rb promoter and the exon 27 of the Rb gene as negative control (Fig. 1E). Enrichment of the immunoprecipitated fraction was observed for HeLa cells, quiescent and stimulated human lymphocytes (Fig. 1E), MCF-7 and SW480 cells (data not shown). In conclusion, CTCF is interacting in vitro and in vivo at the human Rb promoter in cancer cell lines and in human lymphocytes.

Functional contribution of CTCF to human Rb promoter activity. The contribution of CTCF to Rb promoter activity was addressed by transient transfection assays with the Rb promoter driving the Luciferase reporter gene, and trans-activation experiment using a human CTCF cDNA. We observed a 2-fold increase in reporter gene expression induced by CTCF overexpression (Fig. 2A and C, Rb + CTCF). Next, the RbΔCTCF from which ∼200 bp of promoter sequence upstream of the RBF-1 binding motif, including the Rb-CTCF site, had been removed was transfected showing a modest reduction in reporter activity (Figs. 1A and 2A, RbΔCTCF). This result is in agreement with data showing that an equivalent deletion reduced human Rb promoter activity in transiently transfected mouse myoblasts (21). Surprisingly, when the RbΔCTCF construct was cotransfected with the CTCF cDNA, a positive trans-activation was observed (Fig. 2A; compare Rb + CTCF with RbΔCTCF + CTCFcDNA).

To further characterize the effect of CTCF on the human Rb promoter, the CTCF binding sequence previously tested by EMSA was fused to the RbΔCTCF promoter (Fig. 2A; RbΔCTCF+1× and RbΔCTCF+2×). One or two copies of the CTCF binding site were cloned on the 5′ side of RbΔCTCF promoter, and as an additional control, the FII motif was incorporated (Fig. 2A; RbFII; ref. 22). As expected, the reintroduction of the Rb-CTCF binding sequence restored promoter activity and its capacity to be trans-activated, in a linear manner, by CTCF cDNA (Fig. 2A; RbΔCTCF+1×+CTCFcDNA and RbΔCTCF+2×+CTCFcDNA). Notably, the FII sequence was unable to significantly contribute to the Rb promoter trans-activation, although it has been extensively shown that FII binds CTCF (15, 16, 22). These results suggest that CTCF requires a specific context within the human Rb promoter to perform its function; possibly along with cofactors interacting with CTCF and/or determining combinatorial use of zinc fingers by CTCF at this promoter.

To show the specificity of such interaction, we designed several mutations over the CTCF binding site to be tested in the context of the entire Rb promoter (Fig. 2B and C). Based on in vitro footprinting (data not shown) and the literature (20), we tested, in gel-shift assays, the capacity of seven different mutations to compete CTCF binding. Mutations E and G were not able to compete, and mutation E was incorporated in the context of complete Rb promoter (Fig. 2C). The RbmutE promoter showed around 2-fold reduction on reporter gene activity in transient transfections. The RbmutG was also tested with similar results (data not shown). The same construct was unable to be trans-activated by CTCF cDNA. Furthermore, human p53 core promoter alone and cotransfected with CTCF cDNA were tested as negative controls (Fig. 2C).

In summary, our series of transient transfection experiments supports a regulatory role of CTCF over the human Rb gene promoter.

Relationship between CTCF and the Rb gene promoter in an integrated context. Because CTCF seems to be a component of epigenetic regulation at distinct levels we decided to test its effect on the Rb promoter in a chromatin context (8, 23). To this end, a series of stable cell lines (single- and multicopy integrants) using GFP as reporter were generated. The activities of the constructs harboring complete Rb promoter, the RbΔCTCF, and the RbmutE promoters were compared (Fig. 2D). The mean expression was evaluated by flow cytometry (FACS), and we observed that when the CTCF sequence is removed (RbΔCTCF) or mutated (RbmutE), the mean transgene expression level is lower, and the cell fluorescence spreads along the logarithmic intensity axis, resembling a variegated pattern of gene expression (Fig. 2D; refs. 24, 25).

These results suggest that CTCF is inducing a local chromatin environment that contributes to the proper responsiveness of the promoter.

RNA interference against CTCF reduces the Rb gene promoter activity. With the aim to confirm the role of CTCF in human Rb gene expression, RNA interference (RNAi) against CTCF was done. Multiple rounds of transfection of RNAi against CTCF and GFP as negative control (mock) were carried out. A Western blot showed the decrease of CTCF peptide concentration but not for the control GFP (Fig. 3A). CTCF-depleted HeLa cells were transiently transfected with the pGLRb reporter plasmid carrying the entire Rb promoter, and consistent with previous results, we found a reduction of promoter activity when CTCF protein is depleted (Figs. 2A,, RbΔCTCF, and 3B).

Taken together, our functional results favor the idea that CTCF is not playing a classic cis-acting transcription factor role. At this point, we asked if there is any other role, in particular through chromatin structure modulation, for CTCF at the human Rb gene promoter. Then, we decided to explore an epigenetic function for CTCF.

Accelerated extinction of transgene expression in the absence of the CTCF motif. To gain insight into the function of CTCF at the human Rb gene promoter, we generated a set of stable clones of human erythroleukemic K562 cells containing the complete Rb, RbΔCTCF, and RbmutE promoters (Fig. 4). In the RbΔCTCF and RbmutE transgenes, rapid silencing of reporter gene expression was observed, whereas in control Rb lines, no extinction was observed over the time period tested (see d0 and d15 in Fig. 4). Extinguished lines were able to be partially reactivated when incubated with 5-aza-2'-deoxycytidine (data not shown), suggesting that CTCF protects the Rb promoter against progressive epigenetic silencing.

Generation of stable cell lines for the study of the human Rb gene promoter silencing. Because abnormal DNA methylation of tumor suppressor promoters is a frequent phenomenon in cancer, we decided to test whether CTCF is involved in Rb promoter regulation even under conditions of epigenetic silencing (26). To this end, we generated a stable HeLa cell line carrying the entire Rb promoter. The stable line was kept under continuous cell culture in the absence of drug selection for more than 100 days until a biphasic pattern of GFP gene expression was reached as determined by FACS (Fig. 5A). Our group and others have established that in continuous cell culture, there is a progressive extinction of transgene expression that is attributable to epigenetic changes in chromatin structure, such as histone deacetylation, histone methylation, and DNA methylation (24, 27, 28). Thus, the GFP-expressing and GFP-nonexpressing cell populations were sorted. Inactive HeLa cells were incubated with the histone deacetylase inhibitor, trichostatin-A (TSA), and the DNA methylation inhibitor, 5-aza-2'-deoxycytidine (5-aza-dC). The results showed that TSA is unable to reactivate the silenced transgene (Fig. 5B). In contrast, 5-aza-dC treatment resulted in a significant reactivation of reporter gene expression of around 60%, demonstrating that the reactivation remains partial, possibly reflecting the presence of other epigenetic marks not influenced by TSA (29). The same level of reactivation was reached in the RbΔCTCF and RbmutE silenced clones (Fig. 4A; data not shown). As predicted, reactivation experiments indirectly indicate that the nonexpressing cell population is silenced by epigenetic mechanisms.

To confirm DNA methylation of the Rb promoter, we did a sodium bisulfite sequencing analysis (Fig. 5C). We found an increase of DNA methylation in the nonexpressing cell population in comparison to the expressing one. Surprisingly, the methylation is not as generalized as expected, perhaps reflecting the contribution to silencing by other epigenetic modifications (29).

To complement the overview of the chromatin structure of the generated lines, we did a series of ChIP assays. Interestingly, we only found drastic differences between acH3 and H3K4me2 histone marks over the Rb gene promoter when we compared the positive and negative cell populations (Fig. 5D). Remarkably, we did not observe any enrichment of repressive marks (H3K9me1-3 and H3K27me3). The same results were found using primers located over the GFP gene (data not shown).

Thus, the H3 hypoacetylation, loss of H3K4me2, and slight increase on DNA methylation are the main observed epigenetic modifications on the silenced cell population.

DNA methylation–sensitive binding of CTCF and incorporation of Kaiso. With the aim to determine whether Rb-CTCF binding in vitro is sensitive to DNA methylation, we methylated in vitro the Rb-CTCF binding sequence and did EMSA using HeLa cell nuclear extracts (refs. 9, 10, 30; Figs. 1A and 6A). A methylated and nonmethylated mixed population of probe was obtained that allowed us to visualize, in addition to the CTCF complex, the appearance of a novel slow migrating complex (Fig. 6A,, lane 2). The slow migrating complex was specifically competed with unlabeled in vitro methylated probe (Fig. 6A,, lane 3). In contrast, when a nonmethylated Rb-CTCF probe was used as competitor, only the CTCF complex was competed and the slow migrating complex remained associated (Fig. 6A , lane 4). Together, these results suggest that CTCF binding is sensitive to DNA methylation, and notably, another factor or complex binds the same Rb promoter sequence in a DNA methylation-dependent manner.

Further analysis of the Rb-CTCF binding sequence revealed the presence of the previously identified 5′-C^mGC^mG-3′ sequence motif known to bind Kaiso protein in a methylation-dependent manner (31, 32). Kaiso is considered a methyl-CpG–binding protein that recruits the N-CoR repressor complex and is involved in epigenetic silencing (32). Thus, we decided to explore the possibility of Kaiso interaction at the methylated Rb-CTCF motif through EMSA using in vitro transcribed and translated full-length human Kaiso cDNA (Fig. 6B). A specific retarded complex was observed using increasing concentrations of the in vitro transcribed/translated Kaiso (Fig. 6B, lanes 2–4). The specificity of the interaction was directly confirmed by a supershift assay (Fig. 6B , lane 5).

Next, we carried out an EMSA with an Rb-CTCF probe that was extensively methylated by SssI in vitro using HeLa nuclear extracts (Fig. 6C). Although the retarded CTCF complex was competed with the unmethylated Rb-CTCF probe, the slow migrating complex is not competed (Fig. 6C,, lane 3). In contrast, self-competition using in vitro methylated sequence abolished the slow migrating complex, supporting its DNA methylation–dependent formation (Fig. 6C,, lane 4). In addition, no competition of the slow migrating complex was obtained using FII and Sp1 binding sequences (Fig. 6C,, lanes 5 and 6). To investigate the nature of the slow migrating retarded complex and considering its DNA methylation requirements, we did a supershift using antibodies against the methyl-CpG–binding proteins MBD2, MeCP2, and Kaiso (Fig. 6C, lanes 7–11). No supershift of the slow migrating retarded complex was seen with antibodies against CTCF, MBD2, MeCP2 or Sp1. However, a clear supershift was obtained when the antibody against Kaiso was used, supporting the idea that the slow migrating band corresponds to Kaiso and possibly an associated repressive complex.

Because we have suggested that CTCF could be involved in protection of the Rb promoter against DNA methylation, we explored its in vivo presence in the active and silenced stable cell populations by ChIP (Figs. 5A and 6D). As predicted, we found that CTCF is enriched on the active population, but it is absent in vivo on the silenced lines (Fig. 6D). Conversely, Kaiso is not present on the active population and is abundantly associated with the Rb promoter on the epigenetically silenced cell population (Fig. 6D).

In conclusion we have shown that the newly defined Rb-CTCF binding site on the human Rb gene promoter is able to bind the methyl-CpG–binding factor Kaiso when the sequence is DNA methylated. These results open new possibilities in the mechanism of CTCF protection of the Rb gene promoter against epigenetic silencing. In cancer, CTCF might be displaced by DNA methylation, in turn recruiting Kaiso and inducing a repressive epigenetic conformation at the human Rb gene promoter.

A number of epigenetic alterations, including promoter hypermethylation of tumor suppressor genes causing silencing, play a role in the etiology of human cancers. In the course of the identification of a novel epigenetic regulator of the human retinoblastoma gene promoter, we obtained insights into the mechanism responsible for maintaining the Rb CpG island in a DNA hypomethylated state. We found that CTCF is involved in Rb gene promoter epigenetic regulation, and we propose its participation in maintenance of the chromatin integrity and unmethylated state of this CpG-island. Functional assays showed that CTCF has a modest but positive influence on Rb promoter driven reporter gene expression in cultured cells. In vitro and in vivo assays determined CTCF binding to the 5′ distal part of the Rb core promoter. Removal and point mutations of the CTCF binding site showed a variegated reporter gene expression behavior in an integrated chromatin environment. Our investigation revealed that the association of CTCF with the Rb promoter is sensitive to DNA methylation. In stable lines with the RbΔCTCF or RbmutE transgenes, an unexpected and rapid reporter gene expression silencing was observed compared with the control wild-type promoter clones. Consistent with a protective role of CTCF, we found that in methylated and silenced Rb-reporter cell populations, CTCF no longer binds to the Rb promoter. Instead, Kaiso, a zinc-finger BTB/POZ protein known to bind methylated DNA and capable of recruiting the N-CoR repressive complex, is bound, contributing to epigenetic Rb promoter silencing.

The contribution of CTCF to human retinoblastoma epigenetic regulation could be linked to stable expression of the Rb gene and its protective role against aberrant DNA methylation. A large amount of data exist concerning the role of CTCF at imprinted genes, in particular, its ability to maintain one of the alleles at imprinted loci unmethylated and to maintain an open chromatin conformation in genes that escape X chromosome inactivation (ICR; refs. 10, 33–35). The most recent evidence of the protective function of CTCF against epigenetic silencing is at the DM1 locus, which is related to the myotonic dystrophy disorder. The data support a model in which CTCF restricts the production of a bidirectional intergenic transcript, histone H3-K9 methylation, incorporation of HP1γ and DNA methylation, maintaining the stability of the intergenic region and normal amounts of CTG repeats (35, 36).

These observations and the results presented here support a model in which CTCF might have the capacity to prevent DNA methylation or other epigenetic repressive marks, and we speculate that this could be part of the mechanism that maintains any CpG-island unmethylated (26, 35–37). Our model is consistent with the “CpG island methylator phenotype” proposed by Issa in which methylation centers are the nucleating point for DNA methylation followed by spreading and virtual invasion of a CpG-island (38). The CTCF function at the human retinoblastoma gene promoter is in agreement with the need to block methylation spreading (29, 35).

A novel aspect of our investigation is the observation of Kaiso interaction at the Rb-CTCF binding site when methylated (Fig. 6). This finding is relevant to the mechanisms of epigenetic silencing at the Rb promoter because previous studies showed the cofractionation of Kaiso with the N-CoR repressor complex (32). Our results are complemented by studies of the metastasis-associated gene 2 (MTA2), in which an in vivo association of the Kaiso/N-CoR repressor complex with the MTA2 gene promoter was observed along with H3-K9 methylation and DNA methylation (32). We suggest that N-CoR complex and its associated repressive chromatin remodeling activities may contribute to epigenetic silencing of the retinoblastoma gene promoter in human diseases such as cancer.

It will be relevant to better understand the influence of CTCF on epigenetic regulation of the human Rb gene and its potential role at gene components of the Rb pathway. We hypothesize that CTCF may represent a novel epigenetic regulatory factor involved in the control of cell cycle, senescence and cancer. Our findings attempt to draw one of the first pictures of how a CpG-island can be protected against some epigenetic silencing mechanisms. We propose that CTCF may play a relevant role in maintaining CpG-islands and genomic domains unmethylated. Furthermore, CTCF may contribute to an optimal chromatin conformation that influences Rb promoter responsiveness.

Grant support: Supported by the Dirección General de Asuntos del Personal Académico-Universidad Nacional Autónoma de México (IN203200, IX230104, and IN209403), Consejo Nacional de Ciencia y Tecnología (CONACyT; 33863-N and 42653-Q), the Third World Academy of Sciences (grant 01-055 RG/BIO/LA), and Fundacíon Miguel Alemán, A.C. (F. Recillas-Targa); CONACyT grant 38168-M (L. Benítez-Bribiesca); Ph.D. fellowship from CONACyT and Dirección General de estudios de Posgrado-Universidad Nacional Autónoma de México (I.A. De La Rosa-Velázquez and H. Rincón-Arano); Instituto Mexicano del Seguro Social fellowship (I.A. De La Rosa-Velázquez).

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.

We thank Ann Dean, Mayra Furlan, and Eria Rebollar for suggestions and critical reading of this manuscript and Ernesto Soto-Reyes for pGLp53 plasmid and constant discussions. We are particularly indebted to excellent technical assistance of Georgina Guerrero Avendaño and members of the Félix Recillas-Targa laboratory. We thank L. Ongay, G. Codiz, and M. Mora from the Unidad de Biología Molecular from the Instituto de Fisiología Celular, Universidad Nacional Autónoma de México for DNA sequencing and FACS facility.