The Wnt signaling pathway is capable of self-regulation through positive and negative feedback mechanisms. For example, the oncoprotein c-Myc, which is upregulated by Wnt signaling activity, participates in a positive feedback loop of canonical Wnt signaling through repression of Wnt antagonists DKK1 and SFRP1. In this study, we investigated the mechanism of Wnt inhibitory factor-1 (WIF-1) silencing. Mapping of CpG island methylation of the WIF-1 promoter reveals regional methylation (−295 to −95 bp from the transcription start site) that correlates with transcriptional silencing. We identified Miz-1 as a direct activator of WIF-1 transcriptional activity, which is found at WIF-1 promoter. In addition, we show that c-Myc contributes to WIF-1 transcriptional repression in a Miz-1-dependent manner. Although the transient repression mediated by Miz-1/c-Myc is independent of de novo methylation, the stable repression by this complex is associated with CpG island methylation of the critical −295 to −95-bp region of the WIF-1 promoter. Importantly, Miz-1 and c-Myc are found at WIF-1 promoter in WIF-1 non-expressing cell lines DLD-1 and 209myc. Transient knockdown or somatic knockout of c-Myc in DLD-1 failed to restore WIF-1 expression suggesting that c-Myc is involved in initiating rather than maintaining WIF-1 epigenetic silencing. In a genome-wide screen, DNAJA4, TGFβ-induced and TRIM59 were repressed by c-Myc overexpression and DNA promoter hypermethylation. Our data reveal novel insights into c-Myc-mediated DNA methylation-dependent transcriptional silencing, a mechanism that might contribute to the dysregulation of Wnt signaling in cancer.
Transcriptional silencing of tumor-suppressor genes associated with promoter hypermethylation is a key epigenetic event that contributes to the hallmarks of cancer (Herman and Baylin, 2003). However, the exact process through which this phenomenon occurs remains elusive for most genes. Transcriptional repression of RARβ by PML–RARα, the defining molecular change in acute promyelocytic leukemia, is thus far the best described example of epigenetic silencing of a target gene (Di Croce et al., 2002). The leukemia-promoting PML–RAR fusion protein directly recruits the DNA methyltransferases (Dnmts) 1 and 3a to the RARβ promoter, leading to DNA promoter hypermethylation and gene silencing of RARβ (Di Croce et al., 2002). The polycomb repressive complex 2 has also been implicated in this process, suggesting that Suz12 not only directly modifies histone tails of PML–RAR target genes but also mediates promoter hypermethylation of target genes (Villa et al., 2007).
MIZ-1 (Myc-interacting Zn finger protein-1) is a member of the POZ domain/zinc-finger transcription factor family (Peukert et al., 1997), which acts as a transcription activator of the cell cycle arrest genes P21CIP1 (Seoane et al., 2002; Wu et al., 2003) and P15INK4b (Seoane et al., 2001; Staller et al., 2001). Through its direct interaction with Myc, Miz-1 can also recruit the Myc/Max complex and prevent the recruitment of transcription activator such as P300 (Staller et al., 2001). In addition, epigenetic mechanisms might also mediate the repressive activity of the Miz-1/c-Myc complex, for example, through the recruitment of the Dnmt3a to gene promoter (Brenner et al., 2005).
Targets of the Miz-1/c-Myc repressive complex include P21CIP1 (Peukert et al., 1997; Seoane et al., 2002; van de Wetering et al., 2002; Wu et al., 2003), P15INK4B (Seoane et al., 2001; Staller et al., 2001), α6 and β1 integrins (Gebhardt et al., 2006). Nevertheless, Miz-1-independent mechanisms of gene silencing mediated by c-Myc also exist. For example, P21CIP1 gene silencing also occurs through the displacement of Sp1 from P21CIP1 promoter by c-Myc, independently of Miz-1 (Gartel et al., 2001). In addition, c-Myc was shown to induce apoptosis through the inactivation of Miz-1 and its target BCL2 (Patel and McMahon, 2006, 2007).
Many of the Wnt target genes are capable of enhancing or antagonizing Wnt pathway activity suggesting that feedback loop mechanisms contribute to its regulation (He et al., 1998; Gujral and MacBeath, 2009). Among these, c-Myc participates in a positive feedback loop in Wnt signaling through the repression of Wnt antagonists DKK1 and SFRP1, resulting in the transformation of mammary epithelial cells (Cowling et al., 2007). WIF-1 (Wnt inhibitor factor-1) encodes for a Wnt pathway antagonist (Hsieh et al., 1999) frequently methylated in lung (Mazieres et al., 2004), colon (Taniguchi et al., 2005), breast (Ai et al., 2006), bladder (Urakami et al., 2006) and nasopharyngeal cancers (Chan et al., 2007). In addition, c-Myc and Wif-1 protein expression inversely correlate in bladder cancer cells, which further argues for the existence of such feedback mechanism in Wnt signaling (Urakami et al., 2006; Tang et al., 2009).
In this study, we examined the mechanism through which WIF-1 transcriptional silencing occurs. We carried out an extensive characterization of the DNA methylation status of the WIF-1 promoter and determined the roles of Miz-1 and c-Myc on WIF-1 promoter activity and transcription. Our study defines a role for c-Myc in the epigenetic silencing of WIF-1 and other genes in cancer.
WIF-1 silencing is associated with DNA methylation of a specific region within WIF-1 promoter
In contrast to most colorectal cancer cell lines that lack basal WIF-1 expression (Taniguchi et al., 2005) (Figure 1a) and have methylation of the entire WIF-1 promoter (data not shown), non-small cell lung cancer (NSCLC) cell lines differed in WIF-1 expression levels (Figure 1a). Specifically, H1703, H920, H1435, H1299 and H358 express WIF-1 at high level. U1752, H125, H157, A549, H460 and normal human bronchial epithelial cells have detectable, but lower levels of WIF-1 (Figure 1a). H23, H1666, H1395, H838, H1993 and H1155 lack WIF-1 expression. Treatment with the demethylating agent 5-aza-2'-deoxycitidine led to an increase in WIF-1 expression for cell lines with low or absent basal WIF-1 expression (Figure 1a).
Genomic bisulfite sequencing was used to identify DNA methylation pattern associated with WIF-1 silencing in NSCLC cell lines. The regions −21/+209 bp from the transcription start site or TSS, and −910/−619 bp both showed discordance between methylation status and WIF-1 expression level (Figure 1b and Supplementary Figure 1, respectively). In contrast, bisulfite sequencing of the −401/−350 bp and the −320/−72-bp regions revealed that cell lines heavily methylated in both regions had no detectable levels of the WIF-1 transcript (Figure 1c), while cell lines with WIF-1 expression had incomplete methylation. More specifically, DNA methylation of the region −295 to −95 bp of WIF-1 promoter seems essential for WIF-1 transcriptional silencing.
Characterization of WIF-1 proximal promoter
Five deletion constructs (WIF-1-C1 to C5) were designed for functional characterization of the WIF-1 proximal promoter (Figure 2a). WIF-1-C3 had the highest relative luciferase activity (Figure 2b) providing a potential explanation for the importance of methylation of the −401/−350 bp and −320/−72-bp regions of WIF-1 promoter on the control of WIF-1 transcription. The minimal difference in luciferase activity between WIF-1-C3 and C4 constructs further refines this to −320/−72 bp as the key region for WIF-1 transcriptional regulation. As a direct evidence for the role of methylation in this region, in vitro methylation of WIF-1-C3 led to a 90% decrease in luciferase activity (Figure 2c).
The presence of active (H3K4me2) and inactive (H3K27me3) histone modifications at WIF-1 promoter was examined by real-time chromatin immunoprecipitation (ChIP) in H1703, H460 and H838 (Figure 2d). Enrichment of the active mark H3K4me2 (Figure 2d) correlates with WIF-1 expression status and extent of DNA methylation of the WIF-1 promoter. Specifically, the WIF-1-expressing cell lines H1703 and to lesser extent H460 have a higher enrichment of H3K4me2 than the WIF-1 non-expressing cell line H838. In contrast, the repressive mark H3K27me3 (Figure 2e) was most enriched in H838 and least enriched in H1703, with these changes most different at −250 to −50 bp.
Miz-1 directly activates WIF-1 promoter activity
Data describing the repression of Wnt antagonists SFRP1 and DKK1 by Miz-1/c-Myc (Cowling et al., 2007) and the reverse correlation between c-Myc and Wif-1 protein expression in bladder cancer (Urakami et al., 2006) prompted us to determine whether the Miz-1/c-Myc complex may contribute to WIF-1 silencing. We first assessed the effect of full-length Miz-1 on WIF-1 promoter activity in a luciferase-based assay. Co-transfection of 1 and 10 ng of a construct encoding full-length Miz-1 increased the relative luciferase activity of WIF-1-C3 by 1.4- and 4-fold, respectively, when compared with empty vector (Figure 3a). Similarly, Miz-1 overexpression in H1299 increased endogeneous WIF-1 mRNA level by 3.5-fold compared with empty vector (Figure 3b), and led to the enrichment of Miz-1 at WIF-1 promoter (Figure 3c). Transient knockdown of Miz-1 with small interfering RNA (siRNA) in H1299 led to a decrease in endogeneous WIF-1 mRNA level (Figure 3d). This data therefore show that Miz-1 functions as a direct activator of WIF-1 transcription.
Miz-1/c-Myc represses WIF-1 transcription
Although c-Myc alone led to a 52% reduction in P21CIP1 promoter activity (Supplementary Figure 2A), as previously reported (Brenner et al., 2005), it had no effect on WIF-1-C3 luciferase activity (Figure 4a). We therefore hypothesized that c-Myc-mediated repression of WIF-1 may require co-expression of Miz-1 in this assay. Indeed, overexpression of c-Myc led to a 50% reduction in WIF-1-C3 promoter activity driven by the overexpression of Miz-1 (Figure 4a). Repression by c-Myc was lost when using a mutant version of c-Myc (c-MycV394D), which cannot bind Miz-1. Transient expression of c-Myc in H1299 also modestly repressed endogenous WIF-1 mRNA, suggesting that this repression was present at the native promoter (Supplementary Figure 2B). WIF-1 repression was also observed in cells, which stably express or constitutively express c-Myc, 209myc and DLD-1, respectively (Figure 1a and Supplementary Figure 2C). Importantly, both Miz-1 and c-Myc were present at WIF-1 promoter in these WIF-1 non-expressing cell lines (Figure 4b).
Given that the Miz-1/c-Myc repressive complex was shown to physically interact with Dnmt3a at P21CIP1 promoter (Brenner et al., 2005), we next analyzed the contribution of de novo methylation to the c-Myc-mediated transient repression of WIF-1. Treatment with 5-aza-2'-deoxycitidine had no effect on c-Myc-mediated transient repression of Miz-1-driven WIF-1-C3 luciferase activity (Figure 5a). We also analyzed the effect of 5-aza-2'-deoxycitidine on WIF-1 expression and promoter hypermethylation in H209 and the stably derived c-Myc overexpressing cell line 209myc. WIF-1 was expressed in H209, lacking c-Myc expression, but fully repressed in 209myc cells (Figure 5b, upper panel). WIF-1 expression was restored in 209myc cells after treatment with 5-aza-2′-deoxycitidine (209myc AZA), showing that WIF-1 repression in 209myc is conferred by promoter hypermethylation. Bisulfite genomic sequencing showed that the −401/−350-bp region of WIF-1 promoter was methylated in both H209 and 209myc cells (Supplementary Figure 3A) while the −320/−72-bp region was heavily methylated only in 209myc (Figure 5b, lower panel). In particular, the increase in density of methylation of the −295 to −95 bp of WIF-1 promoter is highly similar to the methylation patterns observed in NSCLC cell lines not expressing WIF-1 (Figure 1c). The increase in methylation of −21/+209 bp region in 209myc (Supplementary Figure 3B) was less prominent. Finally, the re-expression of WIF-1 in 209myc AZA was associated with demethylation of the WIF-1 promoter (Figure 5b, lower panel and Supplementary Figure 3B).
Loss of c-Myc does not affect WIF-1 expression in DLD-1 cells
c-Myc was depleted through transient siRNA knockdown (Figure 5c) or somatic gene knockout in the colorectal cancer cell DLD-1 (DLD-1 c-Myc−/−) and the expression of WIF-1 and two other c-Myc targets, P21CIP1 and GADD45, analyzed. Although c-Myc depletion by siRNA was incomplete, we observed a 3.4- and 5.6-fold induction in p21CIP1 mRNA after c-Myc siRNA for 24 and 120 h, respectively (Figure 5d). Complete loss of c-Myc expression in c-Myc−/− resulted in a 47-fold increase in P21CIP1 transcript compared with parental DLD-1 (Figure 5d). GADD45, another gene repressed by c-Myc (Bush et al., 1998), showed a 30-fold increase in expression relative to wild-type DLD-1 cells (data not shown). However, WIF-1 expression remained undetectable by real-time reverse transcriptase (RT)–PCR and western blot analysis in DLD-1 cell treated with c-Myc siRNA for 24 or 120 h and in the complete genetic c-Myc knockout (DLD-1 c-Myc−/−) (data not shown). In addition, WIF-1 promoter remained heavily methylated in DLD-1 c-Myc−/− (Figure 5e) cells. This data therefore suggest that c-Myc might initiate but not maintain WIF-1 transcriptional silencing.
New targets of c-Myc-mediated DNA methylation-dependent gene silencing
Although the mode of repression of already known Miz-1/c-Myc targets seems to be transient and DNA methylation-independent, our data argue that the repression of some genes such as WIF-1 might be locked in place by DNA promoter hypermethylation. We therefore used a gene expression microarray to discover additional genes repressed by c-Myc and DNA promoter hypermethylation (Figure 6). 209myc was compared with parental H209 and to 209myc AZA, resulting in 185 Myc-repressed genes that could be reversed by DNA demethylation treatment (Figure 6a and Supplementary Table 3). Thirty-six were then validated for expression differences by gel-based and/or real-time RT–PCR (Figure 6b and data not shown), with many genes having only minimal changes in expression levels (Figure 6b, left panel). However, transforming growth factor (TGF)β-induced, DNAJA4 and TRIM59 were expressed in H209, silenced (that is, undetectable by gel-based RT–PCR) in 209myc and re-expressed in 209myc AZA (Figure 6b, right panel). Although we were able to see individual genes repressed by c-Myc and reactivated by 5-aza-2'-deoxyciditine, WIF-1 was not detectable above background signal using the Agilent microarray at baseline or in 209myc AZA. This has previously been observed for other known silenced genes in cancer (Schuebel et al., 2007). Methylation-specific PCR (Figure 7a) and bisulfite sequencing (Figures 7b–d) demonstrated that all three genes were unmethylated in H209, methylated in 209myc and had reduced methylation in 209myc AZA. There was no evidence of promoter hypermethylation of control genes (TP53INP1, SP5, DKK3 and ITM2C) in 209myc cells (data not shown).
Through a comprehensive bisulfite sequencing mapping of WIF-1 CpG island, we found a complex pattern of methylation in NSCLC not previously reported (Mazieres et al., 2004). In particular, DNA methylation at the WIF-1 promoter in some NSCLC cell lines remains incomplete compared with colorectal carcinoma (Taniguchi et al., 2005; Ai et al., 2006). This heterogeneous promoter hypermethylation defines a region within WIF-1 promoter spanning −295 to −95 bp relative to WIF-1 TSS as critical for complete transcriptional repression. Hence, while a previous study has suggested that a region 1.5 Kbp from the TSS was required for expression (Reguart et al., 2004), we found the WIF-1 proximal promoter spanning −436 to +120 bp to have maximal activity. The critical role of this region is consistent with our mapping of promoter hypermethylation at WIF-1 promoter and histone marks revealing active or repressive chromatin associated with variable WIF-1 expression. The mosaic methylation pattern observed in NSCLC cell lines suggests that distinct regions of methylation within gene promoters exist and that such regions of methylation have functional significance. This might also be the case in breast cancer, where a similar methylation pattern was reported (Ai et al., 2006).
We found that Miz-1 is recruited to WIF-1 promoter where it activates WIF-1 transcription. This is particularly interesting given that Miz-1 expression suppresses growth of neuroblastoma cell lines in vitro (Ikegaki et al., 2007). It would be interesting to see whether this effect might be mediated through the increase in WIF-1 therefore resulting in a decrease in Wnt signaling activity, which contributes to neuroblastoma progression (Liu et al., 2008).
We next showed that Miz-1/c-Myc represses WIF-1 transcription. First, in a transient assay we found that c-Myc repression of WIF-1 is Miz-1-dependent but is independent of DNA promoter hypermethylation. The latter is further suggested by our preliminary data showing that Dnmt3a reduces Miz-1-driven WIF-1-C3 luciferase activity independently of its Dnmt activity (Supplementary Figure 3C). This is particularly interesting given the previously reported ability of another Dnmt, Dnmt1, to participate in transcriptional repression through interactions with HDAC2 (Rountree et al., 2000).
Second, we found that Miz-1/c-Myc are recruited to WIF-1 in cells stably overexpressing c-Myc (that is, 209myc) but also in cells constitutively expressing c-Myc through activated β-catenin/TCF (that is, DLD-1). The latter therefore suggests that WIF-1 repression by Miz-1/c-Myc might indeed reflect a more physiological mechanism. However, and in contrast to what we observed in our transient assay, WIF-1 repression in 209myc cells is accompanied by extensive promoter hypermethylation, particularly in the −295 to −95-bp region. Interestingly, c-Myc-mediated immortalization of human fibroblasts is associated with DNA hypermethylation and silencing of ARF, and is found in immortalized but not early-passage cells (Benanti et al., 2007).
The absence of WIF-1 re-expression or promoter demethylation after c-Myc depletion in DLD-1 suggests that c-Myc is involved in the initiation, but not the maintenance of WIF-1 silencing. This supports a model where recruitment of c-Myc to the WIF-1 promoter by Miz-1 might, if sustained, trigger epigenetic events leading to the irreversible silencing of WIF-1. These include the targeted hypermethylation of the critical region of the WIF-1 promoter, mediated by Dnmts (Ai et al., 2006) and possibly changes in chromatin conformation as well (data not shown) (Tiwari et al., 2008). Hence, similar to the way that c-Myc dynamically relocates to preassembled transcription factories (Osborne et al., 2007), one could hypothesize that c-Myc might also be involved at repressive factories, which might preferentially locate to the nuclear periphery (Finlan et al., 2008), where the methyl-CpG-binding protein MeCP2 and the inner nuclear membrane protein LBR interact (Guarda et al., 2009).
Through genomic studies, we find that this process is likely not limited to WIF-1. Indeed we identified additional genes repressed by c-Myc and promoter hypermethylation include TGFβ-induced (TGFβi), the heat shock protein DNAJA4 and the RING finger 1-like TRIM59. The fact that WIF-1 was not identified in this screen suggests that there are likely additional Myc-repressed genes subsequently silenced by DNA methylation, which may have also been missed using this screening strategy.
Preliminary analysis suggests that the consensus Inr sequence (YYAN(T/A)YY), a candidate for Miz-1 binding, might be found at the promoter of TGFβi, TRIM59, DNAJA4 and WIF-1 which, if confirmed, would further suggest that Miz-1 is indeed a bona fide transcriptional regulator of these genes (Smale and Baltimore, 1989; Peukert et al., 1997; Bowen et al., 2002). These genes may have important roles in carcinogenesis. For example, given the prominent role of ubiquitylation in Wnt signaling, it would be interesting to further characterize the biological significance of c-Myc transcriptional repression on TRIM59 in this context. Loss of TGFβi was recently shown to predispose mice to tumor development (Zhang et al., 2009), and TGFβi downregulation is associated with promoter hypermethylation in lung and prostate cancer (Shah et al., 2008) as well as in MLL rearrangement leukemia (Li et al., 2009). Finally, TGFβi also appears to be a marker for drug sensitivity for paclitaxel in ovarian cancer cells (Ahmed et al., 2007) and decitabine in melanoma cells (Halaban et al., 2009).
Materials and methods
NSCLC and colorectal cancer cell lines used were obtained from ATCC (ATCC, Manassas, VA, USA). DKO, a double knockout for DNMT3b and DNMT1 has been previously described (Rhee et al., 2002). Normal human bronchoepithelial cell line was purchased from Cambrex Bio Science and grown in the Bullet kit media (Cambrex Bio Science, Walkersville, MD, USA). The DLD-1 homozygote (c-Myc−/−) DNA and cDNA were a gift from Dr Kurt E Bachman. The H209 and 209myc cell lines have been previously described (Barr et al., 2000). Treatment of cells with 2 μM of 5-aza-2′-deoxycitidine (Sigma-Aldrich, St Louis, MO, USA), DNA, RNA lands whole cell protein lysate extractions were carried out as previously described (Licchesi et al., 2008).
PCDNA3-c-Myc (wt), a Myc mutant that abolishes binding of c-Myc to Miz-1 (c-Myc V394D) and PCDNA3-Miz-1 were kindly provided by Dr Martin Eilers (University of Marburg, Germany). PCDNA3.1-Myc-Dnmt3a was obtained from Dr Riggs AD (Beckman Research Institute, Duarte, CA, USA). Dnmt3a was subcloned in pCMV-3xFLAG (Stratagene, La Jolla, CA, USA) (Li et al., 2006). A mutant version of this construct, targeting a catalytic cysteine important for its DNMTase activity (Chen et al., 2005), was also made (pCMV-3xFLAG-Dnmt3a-C487W).
WIF-1 luciferase constructs
Each of the five constructs engineered had the same 3′ end, ending at +120 bp from the TSS. Construct WIF-1-C1 spanned the −1391 to +120 bp; construct WIF-1-C2, −929 to +120; construct WIF-1-C3, −436 to +120; construct WIF-1-C4, −348 to +120 and construct WIF-1-C5, −50 to +120. All distance are relative to the TSS. It is to be noted that construct WIF-1-C1 covers the same region of WIF-1 promoter as WIF-1 construct 5 described by others (Reguart et al., 2004). Restriction sites NheI (5′) and HindIII (3′) were used to clone WIF-1 promoters constructs into the pGL3 basic vector (Promega, Madison, WI, USA). The full-length P21CIP1 luciferase construct was kindly provided by Dr Bert Vogelstein (Johns Hopkins University, Baltimore, MD, USA).
Primary antibodies used include c-Myc (N262), Miz-1 (H190), Wif-1 (K16), (Santa Cruz Biotechnology, Santa Cruz, CA, USA), p21Cip1 (SXM30) (BD Biosciences, San Jose, CA, USA) and β-actin (A5441) (Sigma-Aldrich). The enhanced chemiluminescence or enhanced chemiluminescence femto were used as detection reagents (Pierce Biotechnology, Rockford, IL, USA).
RT–PCR primers for WIF-1 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) have been previously described (Mazieres et al., 2004; Licchesi et al., 2008). Primers for genes discovered through microarray experiments and for c-MYC and P21CIP1 were designed with DNASTAR software (DNASTAR, Inc., Madison, WI, USA) (Supplementary Table 1). Real time RT-PCR: real-time RT–PCR was carried out in triplicate in 20 μl reactions containing 10 × PCR buffer, 10 mM dNTPs, 10 pmol of each of the primer, 0.1 unit of Hot Start DNA polymerase, 1 μl of Syber Green (Molecular Probes, Carlsbad, CA, USA) and 0.6 μl of fluorescein (Bio-Rad, Hercules, CA, USA) in an iCycler Optical Module (Bio-Rad). Data are reported as relative expression (1/(CtWIF-1-CtGAPDH) (Figure 1a) or as the relative fold change in expression and is equal to 2–(Sample1 ΔCt – Sample2 ΔCt), where ΔCt=average Ct (gene) – average Ct (GAPDH) (Figure 6b). Taqman RT–PCR: Taqman RT–PCR was carried out with the ABI7900 Taqman thermocycler (Applied Biosystems, Foster City, CA, USA). Taqman gene expression assays for WIF-1 (Hs00183662) (Applied Biosystems), and Tata box-binding protein were used (de la Roche et al., 2008). Values for gene expression in each case were calculated relative to a standard curve of Tata box-binding protein expression.
ChIP for Miz-1 and c-Myc was performed using a dual cross-linking ChIP protocol as previously described (Perini et al., 2005). H1299 transfected with 2 μg/T75 flask of empty vector or Miz-1 for 24 h, 209myc and DLD-1 cells were cross-linked with disuccinimidyl glutarate (2 mM final concentration) for 45 min at room temperature followed by 15 min with formaldehyde (1% final concentration). Cross-linking was stop with glycine (125 mM final concentration). After washing, samples were sonicated, centrifuged and the supernatant was pre-cleared for 2 h with Dynabeads protein G (Invitrogen Corporation, Carlsbad, CA, USA). In all, 50 μg of chromatin were used for each ChIP condition and incubated overnight with either 5 μg/ml of Miz-1 (H190) or Myc (N262) antibody (Santa Cruz Biotechnology). An immunoglobulin G control was also carried out as a negative control. The following day complexes were captured with magnetic beads, washed several times and the cross-linking was reversed overnight at 65 °C. DNA was then precipitated and resuspended in 50 μl of water. Input was diluted 1/100 and PCR was performed using the following condition. WIF-1 (−50/+120; annealing 69 °C, 33 cycles). Finally, PCR products were resolved on 2% agarose gel and stained with ethidium bromide.
Real-time ChIP was carried out as previously described (McGarvey et al., 2006). Antibodies against H3K4-me2 and H3K27me3 were purchased from Upstate (Millipore, MA, USA). ChIP primers for WIF-1 region −836/−608 region and −250/−50 were manually designed (Supplementary Table 1). Amplifications were performed in triplicate, and the enrichment was determined compared with input. Data were presented normalized to the value obtained for the H838 cell line (−836/−608 region).
293 cells were seeded in 24-well dish 24 h before transfection. In all, 200 ng of WIF-1 deletion construct (WIF-1-C1 to C5), pGL3-control vector or pGL3-basic vector constructs was co-transfected with 10 ng of pRL-TK vector using lipofectamine 2000 (Invitrogen). For co-transfection, various quantity of Miz-1 (1, 10, 100 ng), c-Myc (1, 10, 100 ng), Dnmts (1, 10, 50 ng) were used. In each experiment, the amount of plasmid transfected was kept constant by adding empty vector, and each data point was done is triplicate. Transfections were carried out for 24–48 h. Luciferase and renilla activity were determined using dual-luciferase (Promega) and is shown relative to control, which was set to 1.
In vitro DNA methylation
Construct pGL3-C3 was in vitro methylated as described (Yu et al., 2005). 200 ng of mock or SSI-treated vector were transfected in 293T cells. After 48 h, luciferase and renilla activity were assayed as described above.
Bisulfite modification, methylation-specific PCR (Licchesi et al., 2008) and bisulfite sequencing (Zinn et al., 2007) were carried out as previously described. For cloning experiments, 5–10 clones were sequenced at the Johns Hopkins Core sequencing facility. Primers used for methylation-specific PCR and bisulfite sequencing were manually designed (Supplementary Table 1).
c-Myc (MYC), Miz-1 (ZBTB17) SMARTpool ON-TARGETplus siRNA or control siRNA (Dharmacon Inc., Chicago, IL, USA) were transfected with lipofectamine 2000. Samples were harvested at 24, 48 or 72 h. For the time course c-Myc siRNA in DLD-1 cells, transfection with siRNA was repeated at 48 and 96 h. Protein and RNA were extracted at 24, 48, 96 and 120 h after the first transfection and analyzed for c-MYC and P21CIP1 expression by western blot and quantitative real-time RT–PCR or Taqman PCR.
The experimental procedure was as previously described (Schuebel et al., 2007). 209myc 5-aza-2′-deoxycitidine (209myc AZA) and 209myc mock samples were co-hybridized on a single array (Human 44K Agilent Technologies array platform, Agilent Technologies Inc, Santa Clara, CA, USA), as were H209 and 209myc mock. Data analysis identified 324 unique probes which had both at least a 1.41-fold increase in the H209 vs 209myc and at least a 1.41-fold increase in 209myc AZA vs 209myc (Supplementary Table 2). In total, 264 out of 324 (81%) of the probes either represented individual genes or had an annotation that could be confirmed by using a modified BLAST algorithm to identify short (nearly) identical matches and map all the probes on a transcriptome database as downloaded from EnsEMBL release 43 http://www.ensembl.org,European Bioinformatics Institute, Hinxton, UK). In all, 185 out of 264 (70%) of these genes were likely to have a CpG island in the upstream region of the promoter (−1 Kbp up to +200 bp from the putative TSS) (Supplementary Table 3).
We acknowledge Dr Martin Eilers (Marbourg University, Germany), Dr Arthur Riggs (Beckman Research Institute, Duarte, CA, USA) and Dr Bert Vogelstein (Johns Hopkins School of Medicine, USA) for plasmids, Dr Kurt Bachman (University of Maryland, USA) for the DLD-1 c-Myc −/− DNA and cDNA. We are grateful to Dr Wayne Yu at the Johns Hopkins Microarray Core facility, Dr Kornel Schuebel for his help with the microarray experiment and Dr Mariann Bienz for her support. Finally, we thank all the members of the Baylin/Herman laboratory for useful discussions and Kathy Bender for administrative assistance. This study was supported by NCI/SPORE grant CA058184 to James G Herman, AICR grant 07-0040 to Mariann Bienz (MRC-LMB) and NCI grant P30 CA06973-44 to Leslie Cope.
non-small cell lung cancer
transcription start site
Wnt inhibitory factor-1
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)