The insulator binding protein CTCF associates with the nuclear matrix

Available online at www.sciencedirect.com R

Experimental Cell Research 288 (2003) 218 –223

www.elsevier.com/locate/yexcr

The insulator binding protein CTCF associates with the nuclear matrix Katherine L. Dunn, Helen Zhao, and James R. Davie* Manitoba Institute of Cell Biology, University of Manitoba, Winnipeg, Manitoba, R3E 0V9 Canada Received 11 November 2002, revised version received 13 March 2003

Abstract Nuclear DNA is organized into chromatin loop domains. At the base of these loops, matrix-associated regions (MARs) of the DNA interact with nuclear matrix proteins. MARs act as structural boundaries within chromatin, and MAR binding proteins may recruit multiprotein complexes that remodel chromatin. The potential tumor suppressor protein CTCF binds to vertebrate insulators and is required for insulator activity. We demonstrate that CTCF is associated with the nuclear matrix and can be cross-linked to DNA by cisplatin, an agent that preferentially cross-links nuclear matrix proteins to DNA in situ. These results suggest that CTCF anchors chromatin to the nuclear matrix, suggesting that there is a functional connection between insulators and the nuclear matrix. We also show that the chromatinmodifying enzymes HDAC1 and HDAC2, which are intrinsic nuclear matrix components and thought to function as corepressors of CTCF, are incapable of associating with CTCF. Hence, the insulator activity of CTCF apparently involves an HDAC-independent association with the nuclear matrix. We propose that CTCF may demarcate nuclear matrix-dependent points of transition in chromatin, thereby forming topologically independent chromatin loops that may support gene silencing. © 2003 Elsevier Science (USA). All rights reserved. Keywords: CTCF; Nuclear matrix; Nuclear matrix proteins; Histone deacetylase; Insulators

Introduction CTCF, previously referred to as NeP1, contains a central 11 zinc-finger DNA binding domain [1]. Binding sites for CTCF (CCCTC-binding factor) are extremely divergent. Different combinations of the 11 zinc-fingers participate in binding a variety of DNA elements, including promoters, silencers, and insulators [2]. For example, CTCF contains a transcriptional activation domain and binds to the essential activator domain APB␤ in the amyloid ␤-protein precursor promoter [3,4]. CTCF participates in transcriptional repression while bound to the second module of the chicken lysozyme silencer [1] and also negatively regulates transcription at the human c-myc gene [2]. CTCF is also an insulator binding protein: all identified vertebrate insulators contain CTCF binding sites [5]. Transcription of the two imprinted genes H19 and Igf2 is controlled in part by a CTCF binding insulator located between them [6,7]. Bind* Corresponding author. Manitoba Institute of Cell Biology, 675 McDermot Avenue, Winnipeg, MB, R3E 0V9 Canada. Fax: ⫹204-787-2190. E-mail address: [email protected] (J.R. Davie).

ing of CTCF to this insulator is sensitive to DNA methylation [7]. H19 is active on the maternal chromosome, where the unmethylated insulator binds CTCF and blocks activation of Igf2 [6]. On the paternal chromosome, methylation of the insulator prevents CTCF binding, resulting in Igf2 activation [6]. Thus, CTCF appears to have multiple roles in gene regulation. The major isoform of CTCF has a molecular mass of 82 kDa [8]. However due to amino acid sequences in the Nand C-termini, CTCF migrates aberrantly in SDS–PAGE gels [8]. Full-length CTCF migrates at approximately 130 kDa, although several reports in the literature observe a band of approximately 140 kDa [5,8,9]. This protein appears to have more than one form. A 70-kDa isoform of CTCF has been identified in which the C-terminus is truncated [8]. As well, multiple isoforms of CTCF with molecular masses of 55, 70, 73, 80, 97, and 130 kDa have been observed in chicken [10]. It is uncertain, however, if these faster migrating isoforms of CTCF are isoforms or proteolytic fragments. CTCF is a ubiquitously expressed, highly conserved nuclear protein, with the full-length protein being 93% iden-

0014-4827/03/$ – see front matter © 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0014-4827(03)00185-X

K.L. Dunn et al. / Experimental Cell Research 288 (2003) 218 –223

tical in chicken and human. The zinc-finger DNA binding domain is 100% identical in amino acid sequence between chicken, mouse, and human. These facts point to a fundamental role for CTCF in the nucleus. It has been suggested that CTCF is part of a conserved mechanism of transcriptional regulation in vertebrates [5]. CTCF is the only known mediator of enhancer-blocking activity in vertebrates [5]. Three other enhancer-blocking proteins, namely Su(Hw), BEAF, and Zw5, have been identified in Drosophila melanogaster [11–13]. These proteins bear no sequence similarity to each other and interact with unique DNA binding sites, suggesting that there is more than one mechanism by which insulators block enhancer– promoter communication [14]. Insulators can be defined as DNA elements that are able to protect a region of chromatin from transcriptional activation by an inappropriate enhancer and/or silencing from the spread of condensed chromatin. CTCF binds to insulators that separate regions of poorly acetylated chromatin from regions enriched in acetylated H3 [15]. An in vitro GST-CTCF pull-down assay produced evidence that CTCF associates with the Sin3A HDAC complex containing HDAC1 and HDAC2 [16]. We tested in an immunoprecipitation assay whether CTCF was associated with HDAC1/2 in breast cancer cells and found that this was not the case. The mechanism by which insulators protect chromatin from these effects is not understood. Many models have been proposed to explain the enhancer-blocking activity of insulators. Insulators may interact with each other to create distinct, functionally independent chromatin domains [17]. Distinct chromatin domains could also be achieved by an interaction between insulators and the nuclear matrix in which case the nuclear matrix would serve as an anchor point in the creation of functional chromatin domains [18]. Several insulators, including the human apolipoprotein B 5⬘ insulator, are located beside matrix-associated regions (MARs) [19]. Furthermore, a segment of DNA mediating insulator activity 5⬘ of the chicken lysozyme promoter is able to bind the nuclear matrix [20]. Given the proximity of several CTCF binding sites to sequences able to interact with the nuclear matrix, we speculated that CTCF may associate with the nuclear matrix. Here we demonstrate that CTCF is a nuclear matrix-associated protein.

Materials and methods Cellular fractionation with Triton X-100 MCF-7 (T5) cells were fractionated as previously described [21]. Briefly, cells were resuspended in TNM buffer (100 mM NaCl, 300 mM sucrose, 10 mM Tris–Cl, pH 8.0, 2 mM MgCl2, and 1% thiodiglycol) plus 1 mM PMSF and 1 ⫻ protease inhibitor cocktail (Roche). Following centrifugation to collect the cytosol and nuclei, the nuclei were resuspended

219

in TNM buffer containing 0.5% Triton X-100, yielding Triton X-100-soluble and -insoluble nuclear fractions. Isolation of proteins cross-linked to DNA by cisplatin in situ Proteins were cross-linked to DNA with cisplatin in MCF-7 breast cancer cells as described previously [22]. Cell pellets (1 ⫻ 106 cells/ml) were incubated with 1 mM cisplatin for 2 h at 37°C. Lysis buffer (5 M urea, 2 M guanidine hydrochloride, 2 M NaCl, and 0.2 M potassium phosphate, pH 7.5) was then added to the cells followed by hydroxylapatite (Bio-Rad Laboratories, Hercules, CA). RNA and proteins not cross-linked to DNA were removed from the resin by three washes with lysis buffer. The protein–DNA cross-links were reversed by adding 1 M thiourea. Isolation of nuclear matrix proteins Nuclear matrices were isolated from MCF-7 breast cancer cells as described previously [23]. Briefly, MCF-7 (T5) cell pellets (1 ⫻ 107 cells) were resuspended in 5 ml cold TNM buffer (100 mM NaCl, 300 mM sucrose, 10 mM Tris–Cl, pH 8.0, 2 mM MgCl2, and 1% (v/v) thiodiglycol) containing 1 mM PMSF, 10 ␮g/ml aprotinin, and 1 ␮g/ml leupeptin. Triton X-100 was added to a final concentration of 0.25% (v/v). Nuclei (20 A260/ml) were digested with 100 ␮g/ml DNase 1 (D-5025; Sigma Chemical Company, St. Louis, MO) for 2 h at room temperature in DIG buffer (50 mM NaCl, 300 mM sucrose, 10 mM Tris–Cl, pH 7.4, 3 mM MgCl2, 1% (v/v) thiodiglycol, and 0.5% (v/v) Triton X-100) with 1 mM PMSF, 10 ␮g/ml aprotinin, and 1 ␮g/ml leupeptin. Ammonium sulfate (4 ␮, Ultra pure; ICN Biomedicals, Aurora, OH) was added slowly to a final concentration of 0.25 M to facilitate the removal of chromatin. The sample was then centrifuged at 9600g for 10 min at 4°C, and the supernatant, termed the S3 fraction, was saved. The pellet, containing the nuclear matrix and associated intermediate filaments (termed the NM1-IF fraction) was resuspended in 8 M Urea. Cell lysate preparation Cell lysate was prepared as described previously [24]. MCF-7 (T5) cells (1 ⫻ 106 cells) were collected in 500 ␮l of IGEPAL buffer (150 mM NaCl, 1% (v/v) IGEPAL, 50 mM Tris–Cl, pH, 8.0, 1 mM PMSF, and 1 ⫻ protease inhibitor cocktail from Roche). Cells were passed through a 25-gauge syringe four times and then sonicated at 40% output for 15 s twice. The supernatant (cell lysate) was removed and protein concentration determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories). Immunoprecipitation of HDAC1 and HDAC2 MCF-7 (T5) cell lysate (500 ␮g of total protein) prepared as described above was added to immunoprecipitation (IP) buffer (50 mM Tris–Cl, pH 8.0, 150 mM NaCl, 0.5%

220

K.L. Dunn et al. / Experimental Cell Research 288 (2003) 218 –223

Results CTCF is a nuclear matrix protein

Fig. 1. CTCF is tightly bound in the nucleus. Nuclei isolated from MCF-7 (T5) breast cancer cells were extracted with 0.5% Triton X-100. Equal volumes of cell fractions, cell lysate (lane 1), cytoplasm (lane 2), nuclei (lane 3), Triton-soluble (lane 4), and Triton-insoluble (lane 5), were run on SDS 10% polyacrylamide gels and transferred to nitrocellulose membrane. An anti-CTCF antibody immunochemically stained membrane is shown. CTCF was detected in the cell lysate, nuclear, and Triton-insoluble fractions (lanes 1, 3, and 5).

IGEPAL, 50 mM NaF, 1 mM EDTA, and 1 ⫻ proteinase inhibitor cocktail from Roche) to give a final volume of 1 ml. Cell lysate was precleared with 50 ␮l of a 1:1 slurry of Protein A sepharose (Pierce) (in IP buffer) for 30 min at 4°C. Five micrograms of polyclonal antibody specific for HDAC1 (Affinity Bioreagents Inc (ABR)) or HDAC2 (ABR) was incubated with the precleared sample overnight at 4°C with rotation. A negative control was done in which no antibody was added. A polyclonal antibody specific for ubiquitinated proteins was used as a nonspecific control. After incubation with polyclonal antibody, 50 ␮l of Protein A sepharose (Pierce) was added and the samples were incubated for 2 h at 4°C with rotation. Beads were washed three times with 1 ml of IP buffer and resuspended in SDS–PAGE loading buffer.

The subcellular distribution of CTCF was determined. MCF-7 (T5) breast cancer cells were fractionated in TNM buffer; the nuclei were then extracted with 0.5% Triton X-100. This resulted in a Triton-soluble fraction (loosely bound in the nucleus) and a Triton-insoluble fraction (tightly bound in the nucleus). These fractions were analyzed along with cellular, nuclear, and cytosolic fractions for the presence of CTCF by immunoblot analyses. Fig. 1 shows that the major 155-kDa and a less abundant 90-kDa isoform of CTCF were present in the nuclear and Triton X-100-insoluble fraction, indicating that CTCF is tightly bound to a nuclear structure. The nuclear matrix is one possible structure that could retain CTCF in the Triton-insoluble fraction. Nuclear matrices were isolated from MCF-7 (T5) breast cancer cells. Immunoblot analyses of the resulting fractions detected the presence of CTCF in all fractions except the cytoplasmic fraction (Fig. 2). The NM1-IF (nuclear matrix) fraction contained a significant amount of the major 155 kDa isoform of CTCF. The minimum percentage of the major nuclear 155-kDa isoform of CTCF associated with the nuclear matrix was determined to be 21.2% (data not shown). The 90-kDa isoform of CTCF was present in the chromatin released fraction (0.25 M ammonium sulfate extract), but not in the NM1-IF fraction (Fig. 2, lanes 5 and 6). Immunoblot analysis was also performed with an antibody specific for human HDAC1, which we have shown previously to be associated with the nuclear matrix [21] (Fig. 2). The 155-kDa isoform of CTCF was sensitive to proteolysis during the DNase I digestion at room temperature. Immunoblot analyses of CTCF in nuclei before and after the

Immunoblot analyses Proteins from the various fractions were separated by SDS 10% PAGE and analyzed by immunoblotting with antibodies specific for the C-terminal 570 –727 residues of CTCF (Upstate Biotechnology), HDAC1 (ABR), and HDAC2 (ABR) [21,25]. Quantification of proteins in the various fractions was done as described previously [25]. A standard curve was generated by running increasing volumes of a known concentration of NM1-IF on an SDS– PAGE gel and immunoblotting as previously described. The standard curve blot was scanned into the Image Station 440CF (Kodak Digital Science). The net intensity of the CTCF band, as determined using 1D Image Analysis Software (Kodak Digital Science), was plotted against the volume loaded to obtain the best trendline (Microsoft Excel). The intensity of the CTCF band in cellular and NM1-IF fractions was then used with the equation for the trendline to determine the relative amounts of CTCF in cellular and NM1-IF fractions.

Fig. 2. CTCF associates with the nuclear matrix. The nuclear matrix was isolated from MCF-7 (T5) breast cancer cells. Aliquots of whole cells (lane 1), nuclei before and after DNase 1 digestion (lane 2 and 3), cytosol (lane 4), and nucleoplasm (lane 5) were saved. Samples including the nuclear matrix fraction (NM1-IF) (lane 6) were analyzed by SDS 10% PAGE. Immunoblot analyses were performed with antibodies specific for CTCF and HDAC1.

K.L. Dunn et al. / Experimental Cell Research 288 (2003) 218 –223

221

ing that the majority, if not all, of CTCF does not interact with HDAC1 or HDAC2 in situ.

Discussion

Fig. 3. Cisplatin cross-links CTCF to DNA in situ. MCF-7 breast cancer cells were incubated with 1 mM cisplatin and the proteins cross-linked to DNA isolated. Two samples of cisplatin cross-linked proteins (20 ␮g) (lane 2 and 3) and 40 ␮g of MCF-7 cell lysate (lane 1) were separated by SDS 10% PAGE and transferred to nitrocellulose membrane. The membrane was immunochemically stained with antibodies specific for CTCF.

The nuclear matrix is a dynamic structure consisting of protein and RNA [30]. This nuclear substructure is the site of important cellular processes such as DNA replication and transcription [31,32]. Further, transcription factors and chromatin-modifying enzymes such as HDACs and histone acetyltransferases are associated with the nuclear matrix [21,33,34]. CTCF is involved in transcriptional activation and repression, insulator activity, and imprinting genetic

room temperature incubation demonstrated that the 90-kDa isoform of CTCF was not detectable before incubation, suggesting that proteolysis occurred during the incubation (Fig. 2, lanes 2 and 3). This band appears in other preparations that have not been treated with DNase 1 (see lane 1 in Figs. 1 and 3); therefore, it not likely due to a protease present in the DNase 1 preparation. To provide evidence that the nuclear matrix isoform of CTCF was associated with DNA in situ, we incubated cells with cisplatin, which preferentially cross-links nuclear matrix proteins to MAR DNA [26 –28]. The proteins crosslinked to DNA in situ with cisplatin were analyzed beside cell lysate proteins by immunoblotting (Fig. 3). The major (155 kDa) isoform and a minor (90 kDa) isoform of CTCF can be seen in the cell lysate (Fig. 3, lane 1). The 155-kDa isoform of CTCF was detected in the two cisplatin crosslinked samples (Fig. 3, lane 2 and 3), identifying the 155kDa isoform of CTCF as potential MAR binding protein. In addition to the 155-kDa isoform of CTCF, we reproducibly observed three other bands of 135, 120, and 75 kDa in the cross-linked protein sample (Fig. 3, lanes 2 and 3). The 90-kDa isoform of CTCF was not detected in the crosslinked protein sample. CTCF does not associate with HDAC1 or HDAC2 In vitro studies provided evidence that CTCF was associated with the transcriptional corepressor Sin3A and HDAC activity, suggesting that CTCF interacted with the Sin3A HDAC complex, which contains HDAC1 and HDAC2 [16,29]. To provide further evidence that HDAC1 and HDAC2 were associated with CTCF in situ, we immunoprecipitated HDAC1 and HDAC2 complexes from MCF-7 (T5) cells. As controls, we either left out the primary antibody or added an isotype-matched antibody against ubiquitin-conjugated proteins. Fig. 4 shows that CTCF was not detected in the fractions immunoprecipitated by antibodies against HDAC1 (Fig. 4A) and HDAC2 (Fig. 4B). CTCF was detected in the unbound fractions, indicat-

Fig. 4. CTCF does not associate with HDAC1 or HDAC2. Five hundred micrograms of MCF-7 (T5) breast cancer cell lysate was incubated with 5 ␮g anti-HDAC1 polyclonal antibody or 5 ␮g anti-HDAC2 polyclonal antibody. The immunoprecipitated (bound) and immunodepleted (unbound) fractions were collected. A negative control to which no antibody was added was also included. Five hundred micrograms of MCF-7 (T5) breast cancer cell lysate was also incubated with a polyclonal antibody specific for ubiquitinated proteins to check the specificity of binding. (A) Samples of the HDAC1 immunoprecipitation were separated by SDS 10% PAGE and analyzed by immunoblot analyses. The efficiency of immunoprecipitation was confirmed by immunoblot analyses with an HDAC1specific antibody (see lanes 4 and 7). The presence of CTCF in unbound fractions (lanes 5–7) was determined by immunoblot analyses with an antibody specific for CTCF. (B) HDAC2 was immunoprecipitated from MCF-7 (T5) breast cancer cell lysate. Samples were separated by SDS 10% PAGE. The efficiency of immunoprecipitation was confirmed by immunoblot analyses with an HDAC2-specific antibody (see lanes 4 and 7). The presence of CTCF in unbound fractions (lanes 5–7) was determined by immunoblot analyses with an antibody specific for CTCF.

222

K.L. Dunn et al. / Experimental Cell Research 288 (2003) 218 –223

information [2–5,7]. We demonstrated that CTCF is a nuclear matrix protein using two different protocols: nuclear matrix isolation and cisplatin cross-linking. Extraction of cells with Triton X-100 demonstrated that CTCF is not loosely bound in the nucleus. Therefore, CTCF is tightly associated with a nuclear structure. Isolation of the nuclear matrix demonstrated the association of CTCF with this structure, similar to the nuclear matrix protein HDAC1. Using cisplatin cross-linking, we found that CTCF is crosslinked to DNA by cisplatin, which preferentially cross-links nuclear matrix proteins to DNA. Our results demonstrate that at least 21.2% of the 155kDa isoform of CTCF is associated with the nuclear matrix. This is likely an underestimate. Complicating the analyses was the high susceptibility of the 155-kDa isoform of CTCF to proteolysis. Proteolysis of CTCF occurred under conditions in which HDAC1 and estrogen receptor were not degraded (Fig. 2 and data not shown). Attempts to prevent this degradation by lowering incubation temperatures and adding protease inhibitors were unsuccessful (data not shown). It is possible that there may exist a CTCF-specific protease that is insensitive to serine and cysteine protease inhibitors. Using an antibody specific for the C-terminus (amino acids 570 –727) of CTCF, we repeatedly observed a 90-kDa isoform of CTCF in several experiments (Figs. 1–3). This isoform does not correspond to the 70-kDa C-terminal truncation observed by Klenova et al. [8] and appears to be a product of proteolysis. Other proteolytic products were observed in samples of proteins cross-linked to DNA by cisplatin (Fig. 3). The bands detected had molecular masses of 75, 120, and 135 kDa but did not include the 90-kDa isoform detected earlier. Cisplatin cross-linking of CTCF to DNA may enrich a different set of DNA-bound CTCF proteolytic products or select for different isoforms of CTCF. Exclusion of the 90-kDa isoform could also be due to further proteolysis of this protein, resulting in the 75-kDa product. The proteolytic 90-kDa isoform of CTCF was not associated with the nuclear matrix. As the antibody against CTCF recognized the C-terminal region of CTCF, the results suggest that the CTCF nuclear matrix targeting domain resides in the N-terminal region of the protein. CTCF present in the chromatin-containing nucleoplasm may indicate the presence of two populations of CTCF that exist in equilibrium with each other. CTCF in the nucleoplasm may bind to promoters and silencers to activate or repress transcription, while nuclear matrix-bound CTCF could be responsible for insulator activity. Two copies of Su(Hw) flanking an enhancer block enhancer activity more effectively than a single insulator interposed between enhancer and promoter [35]. This discovery led to proposals of insulator–insulator interactions creating loop domains [35]. It is possible that insulators from different regions of chromatin interact to prevent enhancer–promoter communication [35]; however, the positioning of insulators on either side of enhancers is not often

observed in nature. Identification of CTCF as a nuclear matrix-associated protein and a MAR binding protein provides another possible mechanism for the formation of independently regulated chromatin loops. The nuclear matrix may serve as an anchor point that allows insulators to separate enhancers from heterologous promoters. Previous studies provided evidence that CTCF was associated with the Sin3 HDAC complex, which is associated with HDAC1 and HDAC2 [16,36]. As CTCF binds to an insulator element that establishes a boundary between condensed chromatin with unacetylated histones and a decondensed, DNase 1-sensitive chromatin region with acetylated histones, it was an intriguing possibility that CTCF would recruit the nuclear matrix-associated HDAC1 and 2 [36,37]. Under conditions in which we found that Sp1 and Sp3 were associated with HDAC1 and HDAC2, we were unable to find an association of CTCF with these chromatin-modifying enzymes in human breast cancer cells [24]. Further, immunodetection of HDAC2 showed that it coimmunoprecipitated with HDAC1, indicating that protein complexes were not disrupted under the conditions of our immunoprecipitation (data not shown). The CTCF-bound insulator also establishes a boundary between condensed chromatin with H3 methylated at lysine 9 and a decondensed chromatin region highly enriched in H3 methylated at lysine 4 [38]. In contrast to histone acetylation, histone methylation is not reversible [39]. Considering the interplay between histone methylation and histone acetylation, H3 site-specific methylation may play a role in deciding the acetylation state of a domain [40 – 42]. In this model it is conceivable that CTCF recruits specific histone methyltransferases. Acknowledgments We thank Virginia Spencer for providing a nuclear matrix sample for preliminary analysis. This research was supported by a grant from the Canadian Institutes of Health Research (MT-9186). A Mona and David Copp studentship award from the CancerCare Manitoba Foundation Inc. to K.L.D. is also gratefully acknowledged. References [1] M. Burcin, R. Arnold, M. Lutz, B. Kaiser, D. Runge, F. Lottspeich, G.N. Filippova, V.V. Lobanenkov, R. Renkawitz, Negative protein 1, which is required for function of the chicken lysozyme gene silencer in conjunction with hormone receptors, is identical to the multivalent zinc finger repressor CTCF, Mol. Cell Biol. 17 (1997) 1281–1288. [2] G.N. Filippova, S. Fagerlie, E.M. Klenova, C. Myers, Y. Dehner, G. Goodwin, P.E. Neiman, S.J. Collins, V.V. Lobanenkov, An exceptionally conserved transcriptional repressor, CTCF, employs different combinations of zinc fingers to bind diverged promoter sequences of avian and mammalian c-myc oncogenes, Mol. Cell Biol. 16 (1996) 2802–2813. [3] A.A. Vostrov, W.W. Quitschke, The zinc finger protein CTCF binds to the APBbeta domain of the amyloid beta-protein precursor promoter: evidence for a role in transcriptional activation, J. Biol. Chem. 272 (1997) 33353–33359.

K.L. Dunn et al. / Experimental Cell Research 288 (2003) 218 –223 [4] A.A. Vostrov, M.J. Taheny, W.W. Quitschke, A region to the Nterminal side of the CTCF zinc finger domain is essential for activating transcription from the amyloid precursor protein promoter, J. Biol. Chem. 277 (2002) 1619 –1627. [5] A.C. Bell, A.G. West, G. Felsenfeld, The protein CTCF is required for the enhancer blocking activity of vertebrate insulators, Cell 98 (1999) 387–396. [6] A.C. Bell, G. Felsenfeld, Methylation of a CTCF-dependent boundary controls imprinted expression of the Igf2 gene, Nature 405 (2000) 482– 485. [7] A.T. Hark, C.J. Schoenherr, D.J. Katz, R.S. Ingram, J.M. Levorse, S.M. Tilghman, CTCF mediates methylation-sensitive enhancerblocking activity at the H19/Igf2 locus, Nature 405 (2000) 486 – 489. [8] E.M. Klenova, R.H. Nicolas, U.S.A.F. Carne, R.E. Lee, V.V. Lobanenkov, G.H. Goodwin, Molecular weight abnormalities of the CTCF transcription factor: CTCF migrates aberrantly in SDS–PAGE and the size of the expressed protein is affected by the UTRs and sequences within the coding region of the CTCF gene, Nucleic Acids Res. 25 (1997) 466 – 474. [9] N. Saitoh, A.C. Bell, F. Recillas-Targa, A.G. West, M. Simpson, M. Pikaart, G. Felsenfeld, Structural and functional conservation at the boundaries of the chicken beta-globin domain, EMBO J. 19 (2000) 2315–2322. [10] E.M. Klenova, R.H. Nicolas, H.F. Paterson, A.F. Carne, C.M. Heath, G.H. Goodwin, P.E. Neiman, V.V. Lobanenkov, CTCF, a conserved nuclear factor required for optimal transcriptional activity of the chicken c-myc gene, is an 11-Zn-finger protein differentially expressed in multiple forms, Mol. Cell Biol. 13 (1993) 7612–7624. [11] S.M. Parkhurst, D.A. Harrison, M.P. Remington, C. Spana, R.L. Kelley, R.S. Coyne, V.G. Corces, The Drosophila su(Hw) gene, which controls the phenotypic effect of the gypsy transposable element, encodes a putative DNA-binding protein, Genes Dev. 2 (1998) 1205–1215. [12] K. Zhao, C.M. Hart, U.K. Laemmli, Visualization of chromosomal domains with boundary element-associated factor BEAF-32, Cell 81 (1995) 879 – 889. [13] M. Gaszner, J. Vazquez, P. Schedl, The Zw5 protein, a component of the scs chromatin domain boundary, is able to block enhancer– promoter interaction, Genes Dev. 13 (1999) 2098 –2107. [14] A.G. West, M. Gaszner, G. Felsenfeld, Insulators: many functions, many mechanisms, Genes Dev. 16 (2002) 271–288. [15] F. Recillas-Targa, M.J. Pikaart, B. Burgess-Beusse, A.C. Bell, M.D. Litt, A.G. West, M. Gaszner, G. Felsenfeld, Position-effect protection and enhancer blocking by the chicken beta-globin insulator are separable activities, Proc. Natl. Acad. Sci. USA 99 (2002) 6883– 6888. [16] M. Lutz, L.J. Burke, G. Barreto, F. Goeman, H. Greb, R. Arnold, H. Schultheiss, A. Brehm, T. Kouzarides, V. Lobanenkov, R. Renkawitz, Transcriptional repression by the insulator protein CTCF involves histone deacetylases, Nucleic Acids Res. 28 (2000) 1707–1713. [17] A.C. Bell, G. Felsenfeld, Stopped at the border: boundaries and insulators, Curr. Opin. Genet. Dev. 9 (1999) 191–198. [18] C. Bonifer, A. Hecht, H. Saueressig, D.M. Winter, A.E. Sippel, Dynamic chromatin: the regulatory domain organization of eukaryotic gene loci, J. Cell Biochem. 47 (1991) 99 –108. [19] T.J. Antes, S.J. Namciu, R.E. Fournier, B. Levy-Wilson, The 5⬘ boundary of the human apolipoprotein B chromatin domain in intestinal cells, Biochemistry 40 (2001) 6731– 6742. [20] L. Phi-van, W.H. Stratling, Dissection of the ability of the chicken lysozyme gene 5⬘ matrix attachment region to stimulate transgene expression and to dampen position effects, Biochemistry 35 (1996) 10735–10742. [21] J.M. Sun, H.Y. Chen, J.R. Davie, Effect of estradiol on histone acetylation dynamics in human breast cancer cells, J. Biol. Chem. 276 (2001) 49435– 49442. [22] V.A. Spencer, A.S. Coutts, S.K. Samuel, L.C. Murphy, J.R. Davie, Estrogen regulates the association of intermediate filament proteins

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30] [31] [32]

[33]

[34]

[35] [36]

[37] [38]

[39] [40]

[41] [42]

223

with nuclear DNA in human breast cancer cells, J. Biol. Chem. 273 (1998) 29093–29097. S.K. Samuel, M.M. Minish, J.R. Davie, Nuclear matrix proteins in well and poorly differentiated human breast cancer cell lines, J. Cell Biochem. 66 (1997) 9 –15. J.M. Sun, H.Y. Chen, M. Moniwa, D.W. Litchfield, E. Seto, J.R. Davie, The transcriptional repressor Sp3 is associated with CK2phosphorylated histone deacetylase 2, J. Biol. Chem. 277 (2002) 35783–35786. H. Zhao, L.L. Hart, U. Keller, L.T. Holth, J.R. Davie, Characterization of stably transfected fusion protein GFP-estrogen receptor in MCF-7 human breast cancer cells, J. Biol. Cell Biochem. 86 (2002) 365–375. L.T. Holth, D.N. Chadee, V.A. Spencer, S.K. Samuel, J.R. Safneck, J.R. Davie, Chromatin, nuclear matrix and the cytoskeleton: role of cell structure in neoplastic transformation, Int. J. Oncol. 13 (1998) 827– 837. S.K. Samuel, V.A. Spencer, L. Bajno, J.-M. Sun, L.T. Holth, S. Oesterreich, J.R. Davie, In situ cross-linking by cisplatin of nuclear matrix-bound transcription factors to nuclear DNA of human breast cancer cells, Cancer Res. 58 (1998) 3004 –3008. A. Baiker, C. Maercker, C. Piechaczek, S.B. Schmidt, J. Bode, C. Benham, H.J. Lipps, Mitotic stability of an episomal vector containing a human scaffold/matrix-attached region is provided by association with nuclear matrix, Nat. Cell Biol. 2 (2000) 182–184. C.D. Laherty, W.-M. Yang, J.-M. Sun, J.R. Davie, E. Seto, R.N. Eisenman, Histone deacetylases associated with the mSin3 corepressor mediate Mad transcriptional repression, Cell 89 (1997) 349 –356. J. Nickerson, Experimental observations of a nuclear matrix, J. Cell Sci. 114 (2001) 463– 474. T. Misteli, D.L. Spector, The cellular organization of gene expression, Curr. Opin. Cell Biol. 10 (1998) 323–331. W. Schul, L. de Jong, R. van Driel, Nuclear neighbours: the spatial and functional organization of genes and nuclear domains, J. Cell Biochem. 70 (1998) 159 –171. J.R. Davie, S. Samuel, V. Spencer, L. Bajno, J.-M. Sun, H.Y. Chen, L.T. Holth, Nuclear matrix: application to diagnosis of cancer and role in transcription and modulation of chromatin structure, Gene Ther. Mol. Biol. 1 (1998) 509 –528. G.S. Stein, M. Montecino, A.J. van Wijnen, J.L. Stein, J.B. Lian, Nuclear structure– gene expression interrelationships: implications for aberrant gene expression in cancer, Cancer Res. 60 (2000) 2067– 2076. H.N. Cai, P. Shen, Effects of cis arrangement of chromatin insulators on enhancer-blocking activity, Science 291 (2001) 493– 495. R. Ohlsson, R. Renkawitz, V. Lobanenkov, CTCF is a uniquely versatile transcription regulator linked to epigenetics and disease, Trends Genet. 17 (2001) 520 –527. V.A. Spencer, J.R. Davie, Role of covalent modifications of histones in regulating gene expression, Gene 240 (1999) 1–12. M.D. Litt, M. Simpson, M. Gaszner, C.D. Allis, G. Felsenfeld, Correlation between histone lysine methylation and developmental changes at the chicken beta-globin locus, Science 293 (2001) 2453– 2455. A.J. Bannister, R. Schneider, T. Kouzarides, Histone methylation: dynamic or static?, Cell 109 (2002) 801– 806. J.C. Rice, C.D. Allis, Histone methylation versus histone acetylation: new insights into epigenetic regulation, Curr. Opin. Cell Biol. 13 (2001) 263–273. S.I. Grewal, S.C. Elgin, Heterochromatin: new possibilities for the inheritance of structure, Curr. Opin. Genet. Dev. 12 (2002) 178 –187. Y. Zhang, D. Reinberg, Transcription regulation by histone methylation: interplay between different covalent modifications of the core histone tails, Genes Dev. 15 (2001) 2343–2360.