Interruption of intrachromosomal looping by CCCTC binding factor decoy proteins abrogates genomic imprinting of human insulin-like growth factor II

The transcriptional regulator CCCTC binding factor (CTCF) controls the expression of several genes via chromatin insulation or enhancer blocking (Bell et al., 2001; Ohlsson et al., 2001; Dunn and Davie, 2003; Recillas-Targa et al., 2006). Through the use of different combinations of its 11 highly conserved zinc fingers (ZFs; Mukhopadhyay et al., 2004), CTCF binds to sites within promoters, silencers, and insulators of genes involved in growth, proliferation, differentiation, apoptosis, imprinting, and X chromosome inactivation (Filippova, 2008; Donohoe et al., 2009; Ohlsson et al., 2010). More than 15,000 CTCF binding sites have been identified throughout the genome (Kim et al., 2007). In addition, CTCF has recently been shown to act as a tethering protein, serving as a molecular glue to secure long-range intrachromosomal (Kurukuti et al., 2006; Li et al., 2008) and interchromosomal (Ling et al., 2006) interactions.

CTCF was initially shown to serve as an insulator in the chicken β-globin locus (Bell et al., 1999). The imprinting control region (ICR) of the two coordinately imprinted genes IGF2 and H19 was also identified as another CTCF target. The ICR, located in the 5′ flanking region of the H19 gene and 90 kb downstream of the Igf2 gene, is maternally unmethylated and paternally methylated (Mann et al., 2000; Reik et al., 2000; Sasaki et al., 2000; Arney, 2003; Engel and Bartolomei, 2003; Murrell et al., 2004) The binding of CTCF to the unmethylated maternal ICR creates a physical boundary, blocking the interaction of downstream enhancers with the remote IGF2 promoters and silencing the maternal allele (Bell and Felsenfeld, 2000; Hark et al., 2000; Kanduri et al., 2000).

Recent studies have demonstrated that CTCF regulates allelic expression of mouse Igf2 by forming a long-range intrachromosomal loop (Murrell et al., 2004; Li et al., 2008). This complex intrachromosomal loop is established on the maternal chromosome between the ICR and the Igf2 promoters, presumably through the formation of CTCF–CTCF dimers or multimers (Yusufzai et al., 2004; Li et al., 2008). CTCF also binds to components of the polycomb repressive complex-2 (PRC2), leading to silencing of the maternal allele through histone K27 methylation (Li et al., 2008), thereby allowing the exclusive expression of H19 from the maternal allele and IGF2 from the paternal allele (Mann et al., 2000; Wolffe, 2000; West et al., 2002; Bartolomei, 2003; Engel and Bartolomei, 2003). Biallelic expression of IGF2 may be seen in many cancer cells, although the molecular basis for this loss of imprinting is poorly understood.

In this work, we extend our examination of the role of CTCF in orchestrating long-distance intrachromosomal looping in the human IGF2/H19 imprinting domain. We synthesized CTCF decoy proteins that contain the protein’s evolutionarily conserved DNA binding domain to decipher the mechanisms through which CTCF controls allelic regulation of human IGF2.

CTCF recruits the PRC2 complex to induce allelic silencing through histone K27 methylation by interacting with SUZ12. We decided to characterize the CTCF domain that interacts with SUZ12. CTCF can be divided into three functional regions: the NT, DNA-binding ZF, and C-terminal (CT) domains. We constructed each domain in a GST vector and purified recombinant proteins using a GST column (Fig. 1 A). Recombinant proteins composed of each domain were incubated with recombinant SUZ12. After pulldown, the GST–protein complexes were separated on a PAGE gel and examined for their interaction with SUZ12 using Western blot analysis. As expected, SUZ12 was detected in the reaction with full-length CTCF (Fig. 1 B, lane 5), confirming the interaction between these two functional proteins as previously reported (Li et al., 2008). SUZ12 bound to the CTCF-NT domain (Fig. 1 B, lane 4), but neither the ZF domain, which is required for DNA recognition and binding, nor the CT domain interacted with SUZ12.

To study the role of the SUZ12 interaction in the maintenance of imprinting, we synthesized a decoy CTCF that maintains the ability to bind target DNA sequences but is unable to interact with SUZ12. For this purpose, the ZF domain was amplified by PCR and was linked in frame to Sss1 (CpG methyltransferase and ZF-Sss1) or EGFP (tracking gene and ZF-EGFP). The addition of Sss1 or EGFP allowed us to synthesize decoy CTCF proteins with a similar molecular mass as that of the full-length CTCF (83-kD native CTCF, 87-kD full-length ZF-Sss1, and 73-kD ZF-EGFP). Sss1 is a DNA methyltransferase that is able to methylate CpG dinucleotides in some DNA sequences near the region where it binds, and EGFP is useful in tracking the transfected cells as a marker. To examine whether the decoy CTCFs interact with SUZ12, the purified recombinant decoy proteins or wild-type CTCF (Fig. 1 C) was incubated with cell extracts, pulled down using Ni–iminodiacetic acid columns, and examined for SUZ12 binding by Western blotting. As expected, SUZ12 was detected in the reaction with full-length CTCF (Fig. 1 D, top panel, lane 2), but neither ZF-Sss1 nor ZF-EGFP interacted with SUZ12 (Fig. 1 D, top panel, lanes 3 and 4). As controls, full-length CTCF was detected by an antibody that specifically recognizes the NT portion of CTCF (Fig. 1 D, second panel, lane 2), and decoy ZF-EGFP was detected by a GFP antibody (Fig. 1 D, third panel, lane 4).

The CTCF-Sss1 and CTCF-EGFP genes were packaged into lentiviruses and transduced into human MCF7 (breast cancer), HBF1, and WSF7 (normal fetal skin fibroblasts) cells that maintain normal IGF2 imprinting, i.e., the exclusive expression from the paternal allele. We used two control cell lines, one with a mock virus carrying the empty vector and one without the virus. The expression of CTCF-Sss1 and CTCF-EGFP was measured by RT-PCR (Fig. 2, A and B) in stable cell clones selected with puromycin. Using EGFP as a tracking marker, we also observed the green fluorescence of the expressed ZF-EGFP fusion protein in MCF7 and WSF7 cells (Fig. 2 C). Using an anti-EGFP antibody, we demonstrated the expression of the predicted fusion protein (Fig. 2 D, lanes 3 and 6), which has a larger molecular weight than the native EGFP control (Fig. 2 D, lanes 2 and 5).

We then used a chromatin immunoprecipitation (ChIP) assay to detect the binding of the fusion proteins to the IGF2/H19 ICR. Because there are no available antibodies to Sss1, we took an approach in which two antibodies were used to immunoprecipitate different domains of CTCF in ZF-Sss1–transfected cells. The first anti-CTCF antibody that we used specifically recognizes the CT domain of CTCF, which was deleted in the decoy CTCFs; thus, this antibody only detects the binding of the endogenous (native) CTCF. As expected, the ChIP PCR showed that the endogenous CTCF binds to CTCF sites 1 and 3 in human fibroblast HBF1, WSF7, and breast cancer MCF7 cells, in which IGF2 imprinting is maintained (Fig. 2, E and F, lanes 7 and 8). However, in decoy CTCF-expressing cell clones, no wild-type CTCF binding was detected using an antibody that recognizes the CT domain of CTCF (Fig. 2, E and F, lanes 3 and 4), suggesting that the overexpressed ZF-Sss1 decoy outcompetes the endogenous CTCF for binding at the ICR.

The second antibody specifically recognizes the CTCF ZF domain that is present in both the fusion protein and the endogenous CTCF. ChIP with this antibody showed that the ZF-Sss1 binds to these two CTCF sites (Fig. 2, E and F, lanes 1 and 2).

We also used these two antibodies to examine the CTCF binding to the sixth CTCF site in the ICR region. Again, the data support the finding that the viral overexpressed ZF-Sss1 overrides the endogenous CTCF in binding CTCF site 6 in both MCF7 and HBF1 cells (Fig. 2 G, lanes 1, 2, 5, and 6).

In addition, we also confirmed the direct competition between the ZF-EGFP decoy and the endogenous CTCF by using an anti-EGFP antibody and an antibody against the CT portion of CTCF. As expected, no GFP binding to the ICR sixth CTCF site was detected in normal MCF7 and WSF7 cells (Fig. 2 H, middle panel, lanes 1 and 2). However, in cell clones expressing ZF-EGFP, the decoy directly binds to the ICR in both human fibroblast WSF7 and breast cancer MCF7 cells, as no binding of the native CTCF was detected (Fig. 2 H, lanes 3 and 4). Collectively, these data indicate that the overexpressed decoy proteins outcompete endogenous CTCF in binding to the IGF2/H19 ICR site.

We then determined where CTCF bound in the IGF2 promoter region using ChIP. After immunoprecipitation, IGF2 promoters in the anti-CTCF–precipitated DNA were examined by PCR (Fig. 3 A). In control cells, wild-type CTCF binds to at least three sites in the promoter region (Fig. 3 A, d–f sites, which are near each of the imprinted promoters; and Fig. 3 B, top panel, lanes 10–12), but not to the a site further upstream or the c site between promoter 1 and 2 (Fig. 3 B, top panel, lanes 7 and 9). In the CTCF-Sss1 gene–transfected MCF7 clones, very weak or nonendogenous CTCF binding could be seen in these regions (Fig. 3 B, lanes 2–6). We also detected very weak or non–wild-type CTCF binding near the IGF2 promoters in CTCF-Sss1–transfected fibroblast HBF1 cells (Fig. 3 C, lanes 2–6). The far upstream a site was included as a control site where Suz12 binds independently without the participation of CTCF (Fig. 3, B and C, lanes 1 and 7).

SUZ12 does not directly bind DNA, but it associates with DNA by interacting with the CTCF in some cell lines (Fig. 1; Li et al., 2008). We examined the recruitment of SUZ12 to the IGF2 promoters by ChIP analysis and found that SUZ12 interacted in the four IGF2 promoter regions (Fig. 3 B, second panel, lanes 8 and 10–12). A positive site (a) further upstream was also included, where SUZ12 binding is CTCF independent. In CTCF-Sss1–expressing cells, however, there was no SUZ12 binding at the IGF2 promoter region (Fig. 3 B, lanes 2–6), and SUZ12 binding persisted at the control site, a (Fig. 3 B, lanes 1 and 7). These data show that the CTCF-Sss1 protein, which lacks the CTCF NT domain, was unable to bind SUZ12.

The CTCF-SUZ12 chromosomal interaction is required for specific monoallelic methylation of histone 3 lysine 27 (K27) in IGF2 promoters (Li et al., 2008). We examined whether the loss of the SUZ12 interaction would affect H3K27 methylation in the IGF2 promoter region. In control MCF7 cells, we observed H3K27 methylation at the IGF2 promoters (Fig. 3 B, third panel, lanes 10–12). In contrast, no H3K27 methylation was observed in this region in cells transfected with a CTCF-Sss1 gene (Fig. 3 B, lanes 2–6). We also demonstrated that the CTCF-Sss1 protein abolished the SUZ12 interaction and H3K27 methylation in the IGF2 promoters in transduced normal human skin fibroblast HBF1 cells (Fig. 3 C).

We then examined CTCF-mediated intrachromosomal looping using the chromatin conformation capture technique (3C; Dekker et al., 2002; Li et al., 2008). Cloned cells were fixed with 1% formaldehyde, digested with restriction enzyme EcoR1, and then ligated with T4 DNA ligase. In cells in which imprinting is preserved, the IGF2 promoters directly interact with the ICR that is located 80 kb downstream (Fig. 4 A; Li et al., 2008; Vu et al., 2010). Control MCF7 breast cancer cells maintained the intrachromosomal interaction between the IGF2 promoters and the ICR (Fig. 4 B, lanes 3 and 4). In MCF7 cells that were transfected with the CTCF-Sss1 decoy gene, however, no intrachromosomal interaction was detected between the ICR and the IGF2 promoters (Fig. 4 B, lanes 1 and 2). The control T48/T49 DNA was detected in all samples. The CTCF-Sss1 decoy protein also interrupted the IGF2 promoter–ICR intrachromosomal interaction in human skin fibroblasts (Fig. 4 C).

Interestingly, we also noticed changes in the local chromatin structure around the IGF2 promoters after the expression of the CTCF-Sss1 protein. In control cells, interactions between promoter P2 (1b) and promoter P4 (3b), as well as among 4b, 5b, and 6b (differentially methylated region-2 [DMR2]), were observed (Fig. 5 A, lanes 3 and 4). In CTCF-Sss1–expressing cells, however, some local chromatin interactions were lost, whereas other new interactions appeared (Fig. 5 A, lanes 1 and 2).

To identify which allele is involved in those lost and new local interactions, we sought two single nucleotide polymorphism sites to distinguish between the parental alleles by endonucleases BfaI and NalIII (Fig. 5, B and C, top). In control cells, the 1b–3b interaction between P2 and P4 was biallelic (Fig. 5 B, lanes 7 and 8). In cells expressing the ZF-Sss1, the new interaction between P2 and P3 was also biallelic (Fig. 5 B, lanes 1 and 2). However, in the DMR2 region, the lost and new interactions were monoallelic (Fig. 5 C, lanes 3–6). These data suggest that CTCF may also be involved in maintaining local chromatin structure around the IGF2 imprinting locus, although these interactions may not directly participate in the regulation of allelic expression.

We then examined whether allelic expression of IGF2 was affected in MCF7 and HBF1 cells transfected with CTCF-Sss1. We detected monoallelic expression of IGF2 in control and mock-treated MCF7 cells (Fig. 6 A, lanes 9–12). In the cells transfected with CTCF-Sss1, IGF2 became biallelically expressed (Fig. 6 A, lanes 1–6). Similarly, the decoy protein also abolished IGF2 imprinting in human skin fibroblast cells (Fig. 6 B, lanes 1–6). To confirm the observations, we transduced cells with another CTCF analogue that fuses the CTCF ZF domain with EGFP. This CTCF-EGFP decoy protein also induced biallelic expression of IGF2 in both MCF7 cells (Fig. 6 C, lanes 1–3). Similar data were also seen in a second human fetal fibroblast cell line, WSF7 (Fig. 6 D, lanes 1–3).

In mice, the monoallelic expression of Igf2 is closely coordinated in a reciprocal fashion with H19 imprinting through CTCF-ICR insulation (Bartolomei et al., 1993; Bell and Felsenfeld, 2000; Hark et al., 2000). The CTCF insulator marks the boundary in the ICR that is differentially methylated on the two parental alleles. CTCF binds to the unmethylated maternal CTCF DMR and insulates the Igf2 promoter from the remote enhancer downstream of H19. However, allelic methylation of the paternal ICR prevents the binding of CTCF and thus allows the exclusive expression of H19 from the maternal allele and Igf2 from the paternal allele (Mann et al., 2000; Wolffe, 2000; West et al., 2002; Bartolomei, 2003; Engel and Bartolomei, 2003). We examined H19 imprinting in our cell lines and found that H19 was biallelically expressed in MCF7 cells transfected with CTCF-Sss1 (Fig. 7 A) and with CTCF-EGFP (Fig. 7 B) but remained monoallelically expressed in human fibroblast HBF1 (Fig. 7 C) and WSF7 (Fig. 7 D) cells. Thus, imprinting of the human IGF2 and H19 can be uncoupled in a manner similar to that as observed in human tumors (Feinberg, 1993; Cui et al., 2002; Ulaner et al., 2003; Chen et al., 2006).

Sss1 is a CpG DNA methylase. Using sodium bisulfite sequencing, we did not observe any alteration of DNA methylation at the ICR DNA (Fig. S1). Thus, as reported in human tumors, biallelic IGF2 expression induced by the decoy protein is not necessarily accompanied by altered DNA methylation in the ICR.

Loss of genomic imprinting is one of the factors that contributes to the increased expression of IGF2 in some tumors. We thus were interested in whether loss of IGF2 imprinting induced by decoy CTCFs would alter IGF2 expression. We quantitated IGF2 mRNA transcripts by real-time PCR (Fig. 8). IGF2 is monoallelically expressed in the breast cancer cell line MCF7 and in normal breast tissue. There was increased expression of IGF2 mRNA in the CTCF decoy–expressing cells, where IGF2 was biallelically expressed (Fig. 8, A and B). Similarly, loss of H19 imprinting also led to increased H19 abundance in MCF7 cells (Fig. 8 A). In normal fibroblasts, where H19 remained imprinted despite CTCF decoy expression, H19 gene expression did not increase. We also observed enhanced cell proliferation in human fibroblast cells that show biallelic expression of IGF2 (Fig. S2). There were no obvious changes in the cell morphology of the infected cells.

In our previous study (Li et al., 2008), we demonstrated that CTCF and Suz12 are coprecipitated from nuclear extracts and interact with each other in a two-hybrid system. RNAi knockdown of Suz12 also leads to reactivation of the maternal Igf2 allele and biallelic Igf2 expression. Our data in this study provide further evidence that CTCF is an active participant in controlling allelic expression of IGF2 (Fig. 9). CTCF binds to regions around each of the IGF2 promoters as well as to the unmethylated ICR on the maternal allele. CTCF molecules from these distant loci can dimerize, leading to the formation and reinforcement of intrachromosomal loops. By orchestrating chromatin loop structures, CTCF serves as a DNA-binding protein scaffold to recruit and bind polycomb repressive complexes and deliver the parent-specific H3K27 methylation signal to the remote IGF2 promoters, leading to suppression of maternal IGF2 expression. Through the direct interaction of SUZ12 with the DNA-bound CTCF (Li et al., 2008), the PRC2 complex is recruited specifically to the maternal promoters, where it methylates H3-K27, leading to the formation of a repressive chromatin state on the maternal allele. CTCF cannot bind to the methylated paternal ICR, and, thus, there is no scaffold to secure PRC2 to that site (Bell and Felsenfeld, 2000; Hark et al., 2000). In the absence of CTCF–PRC2 complex binding, a more relaxed chromatin structure is achieved, the paternal IGF2 promoters are able to access the downstream enhancers, and the gene is transcribed in an allele-specific manner.

CTCF plays a critical role in regulating the allelic expression of IGF2. By binding to its target sites at the promoter region and at the distant ICR, CTCF complexes can tether a long-range chromatin loop that delivers the parent-specific methylation signal in the ICR to the remote IGF2 promoters that do not carry any imprinting marks. Interruption of any components in this pathway, including aberrant ICR DNA methylation, decreased expression of PRC2 proteins, mutation of the ICR, and altered H3-K27 methylation, will cause a loss of IGF2 imprinting. Because CTCF regulates many genes by binding to promoters, enhancers, and silencers, it would be of great interest to explore whether CTCF-mediated recruitment of PRC2 proteins is a necessary common mechanism in regulating these targeted genes.

In mouse, Igf2 and H19 are tightly coordinated and reciprocally imprinted (Bartolomei et al., 1993) through the CTCF insulating effect on the unmethylated maternal CTCF DMR (Bell and Felsenfeld, 2000; Hark et al., 2000). However, allelic methylation of the paternal ICR abrogates the binding of CTCF and thus allows the exclusive expression of H19 from the maternal allele and Igf2 from the paternal allele (Mann et al., 2000; Wolffe, 2000; West et al., 2002; Bartolomei, 2003; Engel and Bartolomei, 2003). Deletion or mutation of the CTCF DMR relaxes the normally silent maternal Igf2 allele (Thorvaldsen et al., 1998; Srivastava et al., 2000; Szabó et al., 2004). In this study, however, we found that monoallelically expressed H19 and biallelically expressed IGF2 coexisted in decoy CTCF–transfected human fibroblast HBF1 and WSF7 cells (Figs. 6 and 7). Thus, in human IGF2/H19, the regulation of these two reciprocally imprinted neighboring genes can be uncoupled. This uncoupling has previously been reported in a variety of human tumors (Feinberg, 1993; Cui et al., 2002; Ulaner et al., 2003; Chen et al., 2006).

It should be emphasized that all genetic manipulations, including ICR deletions and/or point mutations in the ICR, designed to prevent CTCF binding, are affected before Igf2 imprinting is set up in early embryos. However, in human tumors, biallelic IGF2 expression occurs as a result of aberrant regulation of the imprinting maintenance mechanism after imprinting has already been established. In addition, all ICR knockout or mutation experiments have been conducted in mouse models, and the regulation of IGF2/H19 imprinting may depend on species-specific factors. Thus, the CTCF decoy approach provides an alternative strategy to explore possible mechanisms underlying loss of IGF2 imprinting in human cells.

The critical involvement of CTCF in maintaining normal monoallelic expression of IGF2 suggests that the CTCF regulatory pathway may be dysfunctional in human tumors in which IGF2 is biallelically expressed (Thorvaldsen et al., 1998; Schoenherr et al., 2003; Pant et al., 2004). A maternally transmitted microdeletion of two CTCF binding sites in the ICR results in biallelic IGF2 expression and H19 silencing in Beckwith-Wiedemann syndrome (Sparago et al., 2004). Using nuclear transfer, we previously showed that loss of IGF2 imprinting in human tumor cells was reversed by the imprinting machinery in normal fibroblast cytoplasm, leading to monoallelic expression of IGF2 in the reconstructed tumor cybrids or hybrids (Chen et al., 2006). Moreover, this epigenetic resetting of IGF2 imprinting in tumors was not accompanied by any changes in DNA methylation at any of the DMRs (DMR₀, ICR, and K_vDMR₁). Recently, we also showed that the CTCF-dependent intrachromosomal loop was lost in human fibroblasts in which IGF2 is biallelically expressed after cycloheximide treatment, despite the fact that DNA methylation in the ICR was not altered (unpublished data).

Similarly, we did not detect any alteration of DNA methylation in the ICR in CTCF-Sss1–transfected MCF7 and HBF1 cells (Fig. S1). Sss1 is a CpG dinucleotide methylase cloned from Spiroplasma monobiae strain MQ1 (Renbaum et al., 1990). It de novo methylates DNA exclusively at CpG sites in vivo and in vitro without sequence preference. In a previous study by Xu and Bestor (1997), Sss1 methylase, when fused to the CT domain of the Zif268 ZF domain or a Zif268 derivative that binds to the p53 binding sites, was able to induce de novo DNA methylation specifically at the target sites. In a parallel study with mouse fibroblasts, we have found that our CTCF-Sss1 construct is able to methylate CTCF binding sites at the K_vDMR₁ in the Kcnq1 imprinting locus (Fitzpatrick et al., 2007), which is ∼700 K from mouse Igf2 (unpublished data). Thus, the gene locus and its local chromatin structure may be critical factors that determine the ability of CTCF-Sss1 to methylate nearby DNA. Alternatively, the methylase, when fused with the ZF domain, may become inactive in this locus by an unknown protection mechanism. Collectively, these findings suggest that, unlike the mouse Igf2, loss of IGF2 imprinting in tumors may not necessarily be accompanied by changes in DNA methylation in known ICRs, but rather may be related to dysfunction of any of the components of the regulatory network, including the intrachromosomal loop, the ICR, CTCF, or the PRC2 complex. In support of this hypothesis, it is known that human tumors often show a lack of correlation between DNA methylation and IGF2 imprinting status (Moore et al., 1997; Sullivan et al., 1999; Cui et al., 2002; Chen et al., 2006).

It also should be noted that CTCF regulates many genes by binding to promoters, enhancers, and silencers. The CTCF–PRC2 complex may play a role in the regulation of many other genes that are regulated by CTCF, such as c-myc, β-globin, amyloid β-protein precursor, and X-inactivated genes. Furthermore, CTCF has been described as the master weaver of the genome, facilitating intrachromosomal as well as interchromosomal interactions (Phillips and Corces, 2009). For example, CTCF mediates an interchromosomal colocalization between the Igf2/H19 ICR on mouse chromosome 7 and Wsb1/Nf1 on mouse chromosome 11 (Ling et al., 2006). Thus, it would be of great interest to learn whether decoy CTCF proteins alter interchromosomal associations and, in addition, the expression of these CTCF-regulated genes. Furthermore, the model as described in Fig. 9 does not explain the involvement of the H19 enhancer. Further studies are needed to delineate how the enhancer is coordinated with the CTCF–Suz12–RPC2 complex in regulating allelic expression of IGF2.

To construct recombinant GST fusion proteins, the various CTCF cDNA fragments (full-length CTCF, CTCF-NT, CTCF-ZF, CTCF-ZFCT, and CTCF-CT) were generated from template pOBT7-CTCF vector by PCR amplification containing BamHI–XhoI restriction enzyme (Table S1) using pfu polymerase (Agilent Technologies) and were cloned into pGEX-4T-2 vector (Invitrogen). The PET-24b-SUZ12 construct was generated by cloning the full-length SUZ12 cDNA with SalI–NotI site digested from pCMV-SPORT6 (Invitrogen) into pET-24b vector.

The SssI DNA methyltransferase DNA was amplified from the genomic DNA of S. monobiae strain MQ1 (33825; American Type Culture Collection). To enhance translation in mammalian cells, four TGA codons that encode tryptophan residing in the S. monobiae were converted into TGG by PCR ligation. The cDNA fragment encoding the CTCF ZF domain was generated from the pOBT7-CTCF vector by PCR amplification. These two DNA fragments were then linked by SV40 NLS and a short linker sequence to produce the ZF-Sss1 construct, which was then cloned into the NheI–BamH sites in pCDH-CMV-MCS-EF1-Puro lentivirus vector (SBI). EGFP was amplified from EGFP-N1 vector (Takara Bio Inc.) and was used to replace Sss1 to generate pN1-CTCF ZF-EGFP construct (Table S2).

To construct recombinant His-tagged fusion proteins, the full-length CTCF and decoy CTCF (ZF-Sss1 and ZF-EGFP) fragments with NheI and NotI sites were generated from template pOBT7-CTCF vector pCDH-ZF-Sss1 and pCDH-ZF-EGFP by PCR, respectively. These fragments were then cloned into pET-24b vector by framing with His to generate His-tagged fusion proteins (Table S3).

All recombinant GST fusion, His-tagged, and SUZ12 proteins were expressed in KRX Escherichia coli strain according to the manufacturer’s protocol (Promega). The in vitro protein interaction assay was performed as described previously (Li et al., 2008). In brief, purified GST fusion and His-tagged proteins (10 µg) were incubated with 50 µl of cell-free supernatant containing recombinant SUZ12 in 200 µl of binding buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2.5 ng/ml BSA, 10 mM EDTA, 0.1% Triton X-100, 1 mM DTT, and 10% glycerol). After incubation for 3 h at 4°C, the glutathione particles (BD) or His tag isolated beads (Invitrogen) containing the protein complex were pulled down and washed three times. Binding proteins were eluted for Western blotting using anti-SUZ12, anti-CTCF, and anti-GFP antibodies (Abcam).

The lentiviruses were generated in 293SF-PacLV cells according to the protocol provided by the manufacturer. The viral supernatants were filtered through a 0.45-mm filter, concentrated by the PEG-IT kit (SBI), and aliquoted in a −80°C freezer for long-term storage. Human fibroblasts and tumor cells were seeded at 1.0 × 10⁵ cells per well of a 6-well plate 24 h before transduction. The medium was replaced with virus-containing supernatant containing 5 mg/ml polybrene (Sigma-Aldrich) and incubated overnight. 24 h after transduction, the virus-containing medium was replaced with fresh medium for further culture. 4 d later, the infected cells were harvested by trypsinization and replated on a new 100-mm dish. When cell confluence reached 20–30%, the culture medium was replaced by fresh medium containing 1 µg/ml puromycin for colony selection. The medium was changed every 3–4 d. After selection, colonies were chosen and expanded for further analyses.

The 3C assay was performed as described previously (Dekker et al., 2002; Li et al., 2008). In brief, 1.0 × 10⁷ cells were cross-linked with 2% formaldehyde and lysed with cell lysis buffer (10 mM Tris, pH 8.0, 10 mM NaCl, 0.2% NP-40, and protease inhibitors). Nuclei were collected, suspended in 1× restriction enzyme buffer in the presence of 0.3% SDS, and incubated at 37°C for 1 h. Triton X-100 was then added to a final concentration of 1.8% to sequester the SDS. An aliquot of nuclei (2 × 10⁶) was digested with 800 U of restriction enzyme EcoRI at 37°C overnight. After stopping the reaction by adding 1.6% SDS and incubating the mixture at 65°C for 20 min, chromatin DNA was diluted with ligation reaction buffer (New England Biolabs, Inc.), and 2 µg DNA was ligated with 4,000 U T4 DNA ligase (New England Biolabs, Inc.) at 16°C for 4 h (final DNA concentration of 2.5 µg/ml). After treatment with 10 mg/ml proteinase K at 65°C overnight to reverse cross-links and with 0.4 µg/ml RNase A for 30 min at 37°C, DNA was extracted with phenol-chloroform, ethanol precipitated, and used for PCR amplification of the ligated DNA products. PCR primers used in this study are listed in Table S4.

ChIP assays were performed as described previously (Li et al., 2008). In brief, five million cells were fixed with 1% formaldehyde and sonicated for 180 s (10 s on and 10 s off) on ice with a Branson sonicator with a 2-mm microtip at 40% output control and 90% duty cycle settings. The 1 ml of sonicated chromatin was clarified by centrifugation, aliquoted, and snap frozen in liquid nitrogen. To perform ChIP, 150 µl of sonicated chromatin was diluted 10-fold and purified with 2–5 µl of specific antiserum and 60 µl protein G–agarose. Antibodies to GFP, CTCF, SUZ12, and dimethyl-H3-K27 (lysine 27 of histone H3) were obtained from Abcam. DNA that was released from the bound chromatin after cross-linking reversal and proteinase K treatment was precipitated and diluted in 100 µl of low TE buffer (1 mM Tris and 0.1 mM EDTA). PCR reactions were performed under liquid wax in a reaction containing 1 µl ChIP (or input) DNA, 0.5 µM of appropriate primer pairs, 50 µM deoxynucleotide triphosphate, and 0.2 U Klen-TaqI (Ab Peptides). Standard PCR conditions were 95°C for 2 min followed by 35 cycles of 95°C for 15 s, 65°C for 30 s of annealing, and 72°C for 30 s of extension (Table S5). The PCR products were separated on a 5% polyacrylamide–urea gel and quantified by a PhosphorImager (Molecular Dynamics).

Total RNA was extracted from tissues by TRI-REAGENT (Sigma-Aldrich) according to the manufacturer’s guide, and cDNA was synthesized with RNA reverse transcription. Genomic imprinting of IGF2 was examined by PCR in cDNA samples as previously described using primers specific for two polymorphic restriction enzymes (Apa1 and Alu1) in the last exon of human IGF2. After PCR, two parental alleles were distinguished by the digestion of polymorphic restriction enzymes AluI and ApaI and were separated on 5% polyacrylamide gel (Hu et al., 1996, 1997; Chen et al., 2006). Note that in CTCF decoy–expressed cells, the imprinted A allele became activated, leading to biallelic expression of IGF2. Allelic expression of H19 was assessed by polymorphic restriction enzyme Rsa1. PCR primers used to measure allelic expression of IGF2 and H19 are listed in Table S6.

Quantitative real-time RT-PCR amplification was performed using QuantiTect SYBR green (QIAGEN) as previously described (Chen et al., 2006). Specifically, total RNA was extracted by TRIZOL reagent (Invitrogen), and cDNA was synthesized with RNA reverse transcription. The threshold cycle (Ct) values of IGF2 and H19 were quantitated by quantitative PCR in duplicate using a sequence detector (ABI Prism 7900HT; Applied Biosystems) according to the manufacturer’s protocol and was normalized over the Ct of the β-actin control (Table S7).

Total nucleic acids extracted from fibroblast and tumor cells were used to examine DNA methylation patterns. As previously described (Ulaner et al., 2003), total nucleic acids were treated with sodium bisulfite, and DNA in the IGF2/H19 DMR was amplified with DNA methylation-specific primers designed for CTCF binding sites (Table S8). After PCR, methylated and unmethylated DNAs were separated by MluI. To examine the status of DNA methylation in every CpG site in the key sixth CTCF binding region, the amplified PCR DNAs were cloned into TA vector (Invitrogen) and were sequenced using the vector primer.

Protein interaction was determined by Western blotting as previously described. Pull-down protein complexes were separated by SDS-PAGE in 12% (wt/vol) polyacrylamide gels and transferred to polyvinylidene fluoride membranes. Membranes were used in immunoblotting with anti-SUZ12, anti-CTCF, and anti-GFP antibody (Abcam).

Cells were seeded at 5,000 cells per well in flat-bottomed 96-well plates. At the end of the incubation time, 20 µl of 5-mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Sigma-Aldrich) in PBS was added to each well. After 4 h, media were discarded, and cells were lysed with 100 µl DMSO. Cells were incubated for a further 30 min at 37°C with gentle shaking. The optical density was determined with a microplate reader at 570 nm. Absorbance values were normalized to the values of control cells at day 1 to calculate the percentage of survival.

All experiments were performed in triplicate, and the data were expressed as mean ± SD. The comparative Ct method was applied in the quantitative real-time RT-PCR assay according to the Δ2ΔCt method. The data were analyzed with t test or by one-way analysis of variance, and results were considered statistically significant at P ≤ 0.05.

Fig. S1 shows DNA methylation at the CTCF ICR in the IGF2/H19 imprinting domain. Fig. S2 shows the growth characteristics of decoy CTCF–expressing cells. Table S1 shows primers used for construction of various GST-CTCF fusion proteins. Table S2 shows primers used for construction of decoy CTCF vectors. Table S3 shows primers used for construction of His-tagged fusion proteins. Table S4 shows primers used for 3C at the IGF2 locus. Table S5 shows primers used for ChIP assays. Table S6 shows primers used for allelic expression of IGF2 and H19 imprinted genes. Table S7 shows primers used for real-time PCR. Table S8 shows primers used for DNA methylation of the sixth CTCF binding site in the ICR of the IGF2/H19 locus.

The authors thank Weiwei Jiao for help and support in conducting part of these studies.

This work was supported by a National Institutes of Health grant (1R43 CA103553-01) and Department of Defense grant (W81XWH-04-1-0597) to J.-F. Hu, a Medical Merit Review from the Medical Research Service of the Department of Veterans Affairs to A.R. Hoffman, a National Key Program for Basic Research of China grant (2010CB529902) to G. Qian and S. Ge, a National Natural Science Foundation of China grant (30973663) to G. Qian, and a Shanghai Leading Academic Discipline Project grant (S30205) to S. Ge.