Introduction

Methylation of CpG dinucleotides plays an essential role in regulation of gene expression. Five proteins that contain methyl-CpG-binding domains (MBDs), namely MeCP2, MBD1, MBD2, MBD3, and MBD4, are considered to be critical for interpretation of the DNA-methylation signal (Ballestar and Wolffe, 2001). The MBD domain is essential for these proteins to bind to methylated-DNA with little specificity for flanking nucleotide sequences. In addition, a methyl-CpG-binding protein lacking the MBD domain, Kaiso, was reported (Prokhortchouk et al., 2001; Daniel et al., 2002). Kaiso was identified as a p120^ctn-interacting protein (Daniel and Reynolds, 1999), a member of MeCP1 complex (Prokhortchouk et al., 2001) as well as that of the N-CoR complex (Yoon et al., 2003). Kaiso can bind a specific consensus sequence, but can also bind to methyl-CpG dinucleotides by its zinc-finger (Daniel et al., 2002).

In earlier studies, we noted an interesting correlation between the methylation pattern of intron 1 of EGR2 and the expression level of the gene (Unoki and Nakamura, 2003a). Through a line of analyses designed to clarify the transcriptional regulation of EGR2, we identified inverted-CCAAT-box-binding protein of 90 kDa (ICBP90) as a novel methyl-CpG-binding protein and as a candidate of transcriptional regulator of EGR2, but we could not conclude that ICBP90 was a transcriptional regulator of EGR2. However, since the protein seemed to play important roles in transcriptional regulation in general by its unique feature, we decided to further examine its biological functions.

ICBP90 was first isolated as a protein that bound to a CCAAT box in the promoter region of the topoisomerase II gene (Hopfner et al., 2000, 2002). ICBP90 localizes in nuclei and contains an ubiquitin-like (UbL) domain, a leucine zipper, a zinc-finger of the PHD-finger type, SRA domain, two nuclear localization signals (NLSs), and a zinc-finger of the ring-finger type. Its discoverers found abundant expression of ICBP90 mRNA in actively proliferating tissues; similarly, ICBP90 protein was highly expressed in fibroblasts at the active proliferative stage, but not after the cells reached confluence (Hopfner et al., 2000). Overexpression of ICBP90 was observed in breast carcinomas, indicating a possible role in human mammary carcinogenesis; those experiments also suggested that E2F-1 was an upstream regulator of ICBP90 (Mousli et al., 2003). Hopfner et al. (2000) also predicted several putative phosphorylation sites for cAMP/cGMP-dependent kinase in the ICBP90 protein, and recently the protein kinase A was shown to phospholylate one of the predicted phosphorylation sites (Trotzier et al., 2004). Moreover, expression of ICBP90 decreases after DNA damage involving the p53/p21^WAF1-dependent pathway, and its downregulation contributes to arrest of the cell cycle at the G1/S transition (Arima et al., 2004).

Np95, a 95-kDa nuclear protein in the mouse, is considered the murine counterpart of human ICBP90 (Fujimori et al., 1998; Uemura et al., 2000; Miura et al., 2001). Recently, it was revealed that Np95 and NIRF, another protein in the same family as ICBP90 (Mori et al., 2002), had ubiquitin ligase activity (Citterio et al., 2004; Mori et al., 2004). Np95 ubiquitinates histones, especially histone H3, through its RING finger domain (Citterio et al., 2004). Expression of Np95 is induced by E1A protein and serum, and repression of Np95 expression suppresses cell proliferation (Bonapace et al., 2002). NIRF ubiquitinates a function-unknown protein, PCNP (Mori et al., 2004), and induces G1 arrest through the association with Cdk2 (Li et al., 2004).

In the work reported here we demonstrated that ICBP90 binds with high affinity to methylated CpGs through its SRA domain and recruits HDAC1 through the same domain. ICBP90, whose expression is directly regulated by E2F-1, targets methylated promoter regions of various tumor suppressors. A hypothesis proposed here involves E2F-1, ICBP90, HDAC1, and tumor suppressors in a model pathway that could be important for carcinogenesis.

Results

Purification and identification of a novel property of ICBP90, ability to bind methyl-CpG

We isolated ICBP90 (approved symbol is UHRF1) as a 90-kDa molecule binding to a methylated 1.2-kb DNA fragment corresponding to the CpG islands of intron 1 of the EGR2 gene by DNA–protein pull-down assay (Unoki and Nakamura, 2003a) and subsequent LC/MS-MS analysis. No binding to ICBP90 occurred when those sites were unmethylated (data not shown). We subsequently examined ICBP90 function on EGR2 transcription, but failed to find any specific effects on it. However, since all known methyl-CpG-binding proteins have been shown to have important biological functions (Ballestar and Wolffe, 2001), we decided to investigate the biological effect of ICBP90 as a novel methyl-CpG-binding protein.

To delineate the specific binding target of ICBP90, we prepared three 400-bp portions of the EGR2 fragment (F, M, B), and examined their binding affinities for ICBP90 (Figure 1a). As ICBP90 bound to all three fragments with almost the same affinity, ICBP90 did not appear to have any obvious sequence specificity for other than methyl-CpG dinucleotides. To further determine the binding affinity of ICBP90, we performed EMSA experiments combined with immunodetection, using methylated or unmethylated 25-bp oligonucleotides (Figure 1b) and anti-ICBP90 antibody. The band detected in a lane of protein/oligonucleotide mixture is in fact the ICBP90/oligo complex because the shift band was disappeared by addition of the antibody, although the band obtained from the experiment using unmethylated-CpG oligonucleotide and anti-ICBP90 antibody cannot be distinguished from that obtained from the experiment using methylated-CpG oligonucleotide. As the result, we found that ICBP90 required at least one symmetrically methyl-CpG dinucleotides as its recognition sequence, and again showed little specificity for flanking nucleotide sequences (Figure 1b, c). These experiments revealed, moreover, that this protein has significantly stronger binding affinity for methyl-CpG dinucleotides than it does for its reported target sequence in the topoisomerase II promoter (Figure 1b, c: oligo7, TopoII ICBP2).

Figure 1.

Target sequence of ICBP90. (a) Schematic representation of CpG sites in EGR2 gene, and sequence of a 1.2-kb portion of intron 1. Proteins binding to each methylated or unmethylated biotin-labeled DNA probe were purified on streptavidin–sepharose, and immunodetected by anti-ICBP90 antibody. UM, unmethylated; M, methylated. (b) EMSA–Western blotting analysis of oligonucleotides containing various numbers of CpGs. Each sequence of the oligonucleotides is described at the bottom. UM, unmethylated; M, methylated. Anti-FLAG antibody was for negative control. We used the nuclear extract prepared from cells transfected with full-length ICBP90-expressing plasmid. The left-end lane does not contain either oligonucleotides or antibodies. (c) Comparison of binding of ICBP90 to various methylated oligonucleotides on a gel. Anti-ICBP90 antibody was used for detection of the oligonucleotide/protein complex

Full figure and legend (223K)

Identification of the methyl CpG-binding domain of ICBP90

We constructed various plasmid clones, each of which encoded a part of the ICBP90 protein (Figure 2a), checked the sizes of the expressed proteins (Figure 2b), and then examined their binding ability to methyl-CpGs in double-stranded oligonucleotides (oligo-1) by EMSA–Western analysis. As shown in Figure 2c, the SRA domain seemed to be essential for binding; the mutant construct including the SRA domain (ICBP90 SRA-FLAG) could bind to the methylated oligonucleotides (Figure 2c upper right panel). We examined the binding affinity using another oligonucleotide and found that the ICBP90 SRA-FLAG could bind to the methylated one, but not the unmethylated one. Since EMSA–Western analysis was performed using nondenaturing gels, only proteins with low isoelectric points (PI) could enter the gel and be electrophoresed without binding to DNA (negative charge). Therefore, ICBP90SRARING (PI: 5.10) and ICBP90UbL (PI: 5.09) were detected without binding to oligonucleotides (Figure 2c, lower panels, the left lane in each panel). We judged that these proteins did not bind to methylated oligonucleotides, since we observed no bands reflecting different mobilities (Figure 2c, lower panels, the right lane in each panel). NIRF, a protein related to ICBP90, and Np95, the murine homologue of ICBP90, also bound to oligonucleotides containing methylated CpGs through their own SRA domains (Figure 2d).

Figure 2.

Methyl-CpG-binding motif of ICBP90. (a) Deletion mutants of ICBP90 and their isoelectric points (PI). (b) Expression level and size of the mutant proteins. (c) EMSA–Western blotting assay using methylated oligo-1. Cellular proteins that include each of the mutant protein were used for determining the binding region. The oligonucleotide/protein complex was detected by anti-ICBP90 antibody or anti-FLAG antibody. The lane (-) indicates the mixture containing neither oligonucleotides nor antibody; middle lane (UM) indicates the mixture of the proteins and the unmethylated oligonucleotides; and the right lane (M) indicates the mixture of the proteins and the methylated oligonucleotides. (d) Schema of SRA-containing constructs derived from NIRF and Np95 (upper), and expression levels and sizes of these proteins detected by anti-Myc antibody (left). EMSA–Western blotting assay was performed using methylated oligo-1. The lane (-) indicates the mixture containing neither oligonucleotides nor antibody; middle lane (UM) indicates the mixture of the proteins and the unmethylated oligonucleotides; and the right lane (M) indicates the mixture of the proteins and the methylated oligonucleotides

Full figure and legend (149K)

Targets of endogenous ICBP90

Since the promoters of various tumor suppressor genes are highly methylated in several cancer-cell lines, we performed chromatin immunoprecipitation (ChIP) analysis to investigate the biological function of ICBP90. After confirming the absence of homozygous deletions, except in the p16^INK4A and p14^ARF genes in A549 cells (Miki et al., 2000), we determined the methylation status of promoter in six tumor suppressor genes by methylation-specific PCR, using sodium bisulfite-treated genomic DNAs and published primer sets (Supplementary Table S1, Figure 3a). We found no discrepancies with previously reported results in terms of methylation status, except for the p14^ARF gene in SW480 cells (Burri et al., 2001). Then, we performed ChIP analysis using anti-ICBP90 antibody. Figure 3b shows that endogenous ICBP90 bound to methylated promoters of all six genes, but not when they were unmethylated with two exceptions; one was the promoter region of RAR gene in HCT116 cells and the other was the promoter region of p14 gene in SW480 cells; these promoters were methylated, but the binding of ICBP90 was not detected. The status of methylation and the results of ChIP assay were highly concordant, indicating that ICBP90 would be likely to bind to the methylated DNAs in vivo. As we had shown earlier, the affinity of ICBP90 to the promoter region of topoisomerase II was very low (Figure 1b).

Figure 3.

Binding of ICBP90 to methylated promoters of various tumor suppressors. (a) Methylation-specific PCR experiments using three cancer cell lines. Conditions of the PCR are described in Supplementary Table S1. (b) ChIP assay using anti-ICBP90 antibody and the three cancer cell lines. As a negative control, no antibody immunoprecipitation (-) and anti--actin immunoprecipitation (-actin) were performed. The PCR conditions are also described in Supplementary Table S1

Full figure and legend (78K)

Formation of ICBP90–HDAC1 complex via the SRA domain

Since other methyl-CpG-binding proteins are involved in HDAC complexes (Ballestar and Wolffe, 2001; Prokhortchouk et al., 2001), we measured HDAC activity involved in ICBP90 complex (Figure 4a). ICBP90 immunocomplexes prepared from lysates of HEK293T cells expressing exogenous ICBP90 revealed higher HDAC activities than the control, a normal mouse IgG complex. Since the substrate for the assay is effectively deacetylated by HDAC1 and HDAC2 (manufacturer's data), we investigated whether ICBP90 could form complexes with either or both of those major histone deacetylases. Immunoprecipitation experiments (Figure 4b) revealed that endogenous ICBP90 could interact with endogenous HDAC1 but not with HDAC2. We then determined which domain(s) of ICBP90 had bound to HDAC1, using plasmid constructs encoding parts of the ICBP90 protein. Among these constructs, ICBP90RING and ICBP90 SRA-FLAG could bind to HDAC1 (Figure 4c, upper panels), while neither ICBP90SRARING nor ICBP90UbL could bind to HDAC1 (lower panels), indicating that ICBP90 binds to HDAC1 through its SRA domain. The SRA domains of Np95 and NIRF proteins revealed binding ability toward HDAC1 as well (Figure 4d).

Figure 4.

Formation of complex with HDAC1 via the SRA domain of ICBP90. (a) ICBP90 was immunoprecipitated by anti-ICBP90 antibody (left) and HDAC activity including the immunocomplexes was measured (right). The light emission of deacetylated substrate was quantified at 405 nm. Error bars, s.d. (Scheffe's F-test). (b) Endogenous ICBP90 was immunoprecipitated with anti-ICBP90 antibody and immunodetected with anti-ICBP90, anti-HDAC1, or anti-HDAC2 antibodies. (c) HEK293Tlls were transfected with full-length ICBP90 or a series of its deletion mutants and with HDAC1-Myc/His plasmids. ICBP90 and its deletion mutants were immunoprecipitated by anti-ICBP90 or anti-FLAG antibody, while HDAC1 was immunoprecipitated by anti-Myc antibody, and immunodetected by anti-Myc or anti-FLAG antibodies. (d) Exogenous Myc-tagged SRA domains of Np95 and NIRF, and FLAG-tagged HDAC1 were immunoprecipitated with anti-Myc or anti-FLAG antibodies, and immunodetected by anti-FLAG or anti-Myc antibodies

Full figure and legend (116K)

Downregulation of ICBP90

To clarify the role of ICBP90 in cell proliferation, we suppressed expression of ICBP90 by siRNAs. Transfection of three plasmids, each designed to express siRNA against ICBP90, into HEK293 cells suppressed expression of ICBP90 protein (Figure 5a), and inhibited cell growth compared with cells transfected with siRNA against EGFP (Figure 5b). The same result was obtained when we used HCT116 cells. Therefore, ICBP90 expression seems to be correlated with cell-proliferation status.

Figure 5.

Endogenous ICBP90 associates with cell growth. Three siRNA expression plasmids against ICBP90 were transfected. (a) Expression of ICBP90 protein was decreased by the siRNAs, comparing with control EGFP-siRNA. (b) siRNAs against ICBP90 suppressed cell growth, while no growth suppression was observed with control EGFP-siRNA. Cell numbers (% of control) were assessed by MTT dye conversion. Error bars, s.d. (Scheffe's F-test)

Full figure and legend (37K)

Cell-proliferation effect of ICBP90

Since ICBP90 is expressed at a high level in cells during the proliferation phase and is low or absent in cells at confluence (Hopfner et al., 2000), we checked the expression status of this protein in various cancer cell lines. We chose three lines that revealed contact inhibition (LNCap.FGC, PC3, and DBTRG-05MG) and two (HCT116 and Ishikawa3-H-12) that did not, according to their cellular morphologies in confluent conditions (Figure 6a). Contact inhibition-positive cells stopped growing but sustained their shape and size, while negative cells kept growing even after contact with neighboring cells, and became smaller. As shown in Figure 6b, expression of ICBP90 protein was decreased in the three contact inhibition-positive lines at the confluent stage compared to the growing phase, but was unchanged in the two negative lines. The suppression of ICBP90 expression was confirmed by RT–PCR (Figure 6c). Expression levels of two ICBP90-related genes, E2F-1 and topoisomerase II, correlated well with that of ICBP90 but expression levels of p16^INK4A and p14^ARFdid not. Although downregulation of ICBP90 by p53/p21-dependent DNA-damage checkpoint signals was reported recently (Arima et al., 2004), expression of p21^WAF1 was not always increased by contact-inhibition signals (Figure 6c), suggesting that different signals might be inducing downregulation of ICBP90 under confluent conditions. A scratching assay using contact inhibition-positive DBTRG-05MG cells (Figure 6d) revealed a significant accumulation of endogenous ICBP90 protein during the proliferative phase. This accumulation was confirmed in other contact inhibition-positive cells (data not shown).

Figure 6.

Expression of ICBP90 in proliferating, confluent cell lines. (a) Morphology of three cells that show contact inhibition, and two cells that do not. (b) Expression level of endogenous ICBP90 protein at proliferative and confluent status. -Actin served as an internal control. (c) RT–PCR results of ICBP90, E2F-1, topoisomerase II, p21^WAF1, p16^INK4A, and p14^ARF. The integrity of each RNA template was controlled through amplification of GAPDH. (d) Scratching assay using DBTRG-05MG cells. Endogenous ICBP90 was detected by DAB staining

Full figure and legend (185K)

Identification of an additional binding partner of ICBP90

Moreover, we identified an approximately 170 kDa novel ICBP90-binding partner designated ICBP90-binding protein 1 (UHRF1BP1) by immunoprecipitation using anti-ICBP90 antibody and subsequent LC/MS-MS analysis (Accession numbers; NM_017754 and AB126777: Figure 7a). The gene encoding UHRF1BP1 is located at chromosome 6p21 and spans an 81.3-kb genomic region with 21 exons (gi|27498529); its product has an NLS in the N-terminal region, a leucine-zipper motif, and a coiled-coil domain in its C-terminal region by computer prediction. We confirmed binding between exogenous ICBP90 and FLAG-tagged UHRF1BP1 by immunoprecipitation (Figure 7b, left). FLAG-tagged UHRF1BP1 also bound to myc-tagged HDAC1 (Figure 7b right). UHRF1BP1 also formed a complex with itself (Figure 7c). Expression levels of UHRF1BP1were unchanged at the confluent or proliferative stage in both the contact inhibition-positive and -negative cells (Figure 6c).

Figure 7.

Identification of an ICBP90-binding partner, UHRF1BP1. (a) Result of immunoprecipitation using anti-ICBP90 antibody. Anti-HA and anti-FLAG antibodies were used as negative controls. The gel was stained by silver after performing 6% SDS–PAGE. A band of approximately 170 kDa was observed in the lane containing proteins binding to endogenous ICBP90; this band was identified as UHRF1BP1 by subsequent LC/MS-MS analysis. (Right): schema of UHRF1BP1 genomic structure and protein, which has an NLS, a leucine zipper, and a coiled-coil motif. (b) Flag-tagged UHRF1BP1 shows binding affinity to both ICBP90 and HDAC1. (c) UHRF1BP1 also binds to itself. (d) Proposed model for the signaling pathway involving ICBP90

Full figure and legend (122K)

Increased expression of ICBP90 in primary breast cancers

The data shown above, combined with reports of overexpression of ICBP90 in breast cancer samples (Hopfner et al., 2002; Mousli et al., 2003), indicate a presumptive importance of ICBP90 in human carcinogenesis. Our own cDNA-microarray data (Table 1, Nishidate et al., 2004) have indicated overexpression of ICBP90 in 22 of 32 moderately and poorly differentiated breast cancers (histological types a2 and a3), although only six of 25 well-differentiated breast cancers or ductal carcinomas in situ (histological types a1 and 1a) revealed upregulation of ICBP90 (P=0.0008: ²-test). Estrogen and progesterone receptor status, age, or menopausal status was not associated with levels of ICBP90 expression. We confirmed the microarray data by RT–PCR using 12 scirrhous breast carcinomas (Supplementary Figure 1a). The result was also confirmed by immunohistochemistry (Supplementary Figure 1b). Figure 7d illustrates a model for the ICBP90 signaling pathway, from proliferation signals to carcinogenesis, proposed on the basis of previous reports and our observations described here.

Table 1 - Expression level of ICBP90 of breast cancer patients and their clinicopathological features.

Full table

Discussion

Methylation of CpG-dinucleotides is receiving increasing attention because of its important role in the regulation of gene expression. Among the five methyl-CpG-binding proteins that have been reported so far, MBD1, MBD2, MBD4, MeCp2, and Kaiso, the last is the only one that binds to methylated CpGs through its zinc-finger domain; the others bind through 'MBD' domains (Ballestar and Wolffe, 2001, Prokhortchouk et al, 2001). We have reported here the second example of a methyl-CpG-binding human protein, ICBP90, which has no MBD domain; this molecule binds through an SRA domain (Baumbusch et al., 2001).

The SRA domain of the murine homologue of ICBP90, Np95, has histone H3-binding activity (Citterio et al., 2004). Since our experiments revealed that Np95 could also bind methyl-CpGs, we suggest that methylated DNA twisted around histone H3 might be the primary target for Np95 and ICBP90 in vivo. This hypothesis is supported by the results of our ChIP assay, which showed that ICBP90 did not bind to unmethylated tumor suppressor genes in vivo. We demonstrated that the protein could bind to DNA segments containing even a single symmetrically methyl-CpG dinucleotides. Sequences flanking methyl-CpG did not appear to be very important but did affect binding affinity to some extent, as is the case with other methyl-CpG-binding proteins.

ICBP90 was previously reported to be a transcriptional regulator of topoisomerase II through interaction with the inverted CCAAT box (ICB2) in the gene's promoter (Hopfner et al., 2000). However, we found that ICBP90 had greater binding affinity for methyl-CpG than for the reported sequences, indicating that ICBP90 could play a much wider role in transcriptional regulation, through binding to methylated-DNA. Following up on that idea, we examined binding of ICBP90 to the promoter region of tumor suppressor genes where MBD2 and MeCp2 are known to bind (Magnaghi-Jaulin et al., 1998; Nguyen et al., 2001) and found that ICBP90 also bound to the methylated promoter in those genes, with two exceptions. These discrepancies were due to the effects of chromatin structures, or the presence of other binding proteins that competitively work to the same region. Since some methyl-CpG-binding proteins are constituents of HDAC complexes (Ballestar and Wolffe, 2001; Prokhortchouk et al., 2001), we examined the possibility that ICBP90 also could be involved in the complexes. Our experiments revealed that complexes containing ICBP90 did possess HDAC activity, and that ICBP90 was interacting with HDAC1 through its SRA domain. Therefore, ICBP90 is likely to bind to methylated promoter of some tumor suppressor genes in cells at the proliferative stage, and suppress their expression by cooperation with HDAC1. However, at present, we have no direct evidence that the binding of ICBP90 suppresses the expression of these genes. Methylated promoters of tumor suppressor genes are also targets of other methyl-CpG-binding proteins such as MeCP2 and MBD2. Hence, it makes very difficult to examine the effect of ICBP90 alone in vivo because cells are likely to have several alternative pathways to suppress gene expressions thorough the methylated promoter. Moreover, since expression of p16^INK4A and p14^ARF was not increased under the confluent status at which the expression of ICBP90 was decreased, we need to investigate downstream genes specific to ICBP90 by comprehensive expression analysis combined with methylation status analysis.

We examined ICBP90 expression in cancer cell lines and found significant decreases even in cancer cell lines whose growth was suppressed by contact inhibition. Therefore, ICBP90 seemed to express in proliferative cell phase by some proliferative signals that were conveyed continuously after reaching to the confluence stage in contact-inhibition-negative cells. We successfully demonstrated by scratching assay the expressional differences of the protein in the proliferative cells and the quiescent cells. This result indicates that the expression of ICBP90 may be a useful marker for proliferative stage of cells. On the other hand, ICBP90 expression was significantly downregulated by serum starvation or by DNA damage (Supplementary Figure 2a, 2b), and downregulation of ICBP90 by siRNAs caused growth suppression. Hence, ICBP90 might be a good target for development of cancer treatment, although it is required to find out the way to avoid the growth suppression of normal proliferative cells such as born marrow where ICBP90 is expressed.

We also isolated an additional protein, UHRF1BP1, that is one of the members to constitute the ICBP90 complex, by immunoprecipitation using anti-ICBP90 antibody. We presented the data indicating that the interaction of ICBP90 with UHRF1BP1 caused relocation of ICBP90 (Supplementary Figure 3a), and that overexpression of UHRF1BP1 might induce inhibition of cell growth (Supplementary Figure 3b). Therefore, we considered that UHRF1BP1 might be a negative regulator of cell growth like a tumor suppressor, but further detailed investigation should be required to define the function of this potein. Recently Arima et al. (2004) reported that downregulation of ICBP90 by adriamycin treatment was involved in p53/p21^WAF1-dependent growth-arrest pathway. As a candidate of transcriptional regulator of ICBP90, E2F-1 has been reported (Mousli et al., 2003). We found by ChIP analysis that E2F-1 bound to intron 1 of ICBP90, which contains two E2F-1-binding motifs (Supplementary Figure 4). Based on our observation and previous reports, we propose a model for the ICBP90 signaling pathway involved in E2F-1 regulation (Figure 7d).

Overexpression of ICBP90 in breast cancers (Hopfner et al., 2002; Mousli et al., 2003) was also confirmed by us. The combined results underscored a significant role of ICBP90 in methylation-dependent transcriptional regulation, and suggested potential involvement of ICBP90 in human carcinogenesis.

Materials and methods

Cell lines

Ishikawa3-H-12 was obtained from Dr M Nishida at Tsukuba University (Tsukuba, Japan). SW480, A549, DBTRG-05MG, HCT116, LNCap.FGC, PC-3, HEK293, and HEK293T cells were obtained from the ATCC (Manassas, VA, USA). All cell lines were grown in monolayer in appropriate media supplemented with 10% FBS.

Clinical breast cancer samples and microarray

Primary breast cancers had been obtained with informed consent from 57 patients who were treated at the Department of Breast Surgery, Cancer Institute Hospital, Tokyo. Clinical information was obtained from medical records and each tumor was diagnosed by pathologists according to histopathological subtype and grade. Tumor tissues were also evaluated according to the Japanese Cancer Society classification; clinical stage was judged according to the Japanese Breast Cancer Society's TNM classification. Expression of estrogen and progesterone receptors was determined by EIA; ER was judged as negative when it was less than 13 fmol/mg protein. A mixture of RNAs isolated from normal breast ductal cells of 15 premenopausal breast cancer patients served as a normal control. The microarray experiment using the samples was performed as described previously (Nishidate et al., 2004).

Purification of methyl-CpG-binding proteins

Proteins binding to methylated CpGs were purified using a methylated-DNA probe derived from intron 1 of EGR2. The protocol was described in our previous report (Unoki and Nakamura, 2003a). The purified proteins were separated and stained by CBB. Specific bands were cut out and analysed by LC/MS-MS (APRO Life Science Institute, Tokushima, Japan).

Antibodies

Antibodies used here included mouse monoclonal antibody to ICBP90 (612264; BD Transduction Laboratories, San Diego, CA, USA), mouse monoclonal antibody to -actin (AC-15; SIGMA, St Louis, MO, USA), rabbit polyclonal antibody to HDAC1 (H-51; Santa Cruz), rabbit polyclonal antibody to HDAC2 (H-54; Santa Cruz), FLAG M2 mouse monoclonal antibody (F-3165; SIGMA), FLAG rabbit antibody (F-7425; SIGMA), mouse monoclonal antibody to Myc tag (9E-10; Santa Cruz), rabbit monoclonal antibody to Myc tag (A-14; Santa Cruz), rabbit polyclonal antibody to E2F-1 (C-20, Santa Cruz), and normal mouse IgG (sc-2025).

Construction of expression plasmids

All sequences for cloning were amplified by PCR using KOD-Plus (TOYOBO, Tokyo, Japan). The entire coding sequence of ICBP90, and of deletion mutants designated ICBP90RING, ICBP90RINGSRA, and ICBP90UbL, were cloned into pcDNA3.1(+) (Invitrogen, Carlsbad, CA, USA). The entire coding sequence of UHRF1BP1 and ICBP90 SRA were cloned into pFLAG^®-CMV-5a (SIGMA). The entire coding sequence of HDAC1 was cloned into pcDNA3.1/Myc-His^©A (Invitrogen). The entire coding sequence of UHRF1BP1 and the SRA domains of NIRF and Np95 were cloned into pCMV-Myc (Clontech, Palo Alto, CA, USA). HDAC1-FLAG expression vector was provided by Dr S Ishii (RIKEN, Tsukuba, Japan).

EMSA combined with Western blotting

HEK293T cells were plated and each culture was transfected by lipofection (FuGENE^TM6 Transfection Regent; Roche) with 8 g of an indicated plasmid. After 48 h, the cells were harvested and whole-cell proteins were lysed in high salt buffer (20 mM ES at pH 7.6, 20% glycerol, 500 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM DTT, 0.1% NP40). The extracts were incubated for 1 h at room temperature under usual EMSA condition with appropriate nonlabeled, double-stranded oligonucleotides, and indicated antibody. Double-stranded oligonucleotides were methylated by incubation with SssI. Each protein–oligonucleotide complex was separated in native 5% polyacrylamide gel, transferred to a nitrocellulose membrane, and immunoblotted using appropriate antibodies.

Sodium bisulfite modification and methylation analysis

Genomic DNAs were treated with sodium bisulfite according to a method described previously (Unoki and Nakamura, 2003a). For methylation-specific PCR, the treated DNA fragments were amplified under the conditions described in Supplementary Table S1.

ChIP analysis

ChIP assays were performed using acetyl-histone H3 ChIP Assay Kits (Upstate Biotechnology, Waltham, MA, USA) as recommended by the manufacturer, except that antibodies against ICBP90, E2F-1, and -actin were used in this study. PCR primer sets and annealing temperatures for the analysis are described in Supplementary Table S1.

HDAC colorimetric activity assay

HEK293T cells were transfected with ICBP90 pcDNA3.1(+) and harvested after 24-h culture. Cellular proteins were extracted in NP40-based lysis buffer (0.1% NP40, 150 mM NaCl, 50 mM Tris-HCl), and immunoprecipitated by anti-ICBP90 antibody or by normal mouse IgG as a negative control. The immunocomplexes were collected in protein A/G PLUS-Agarose (sc-2003, Santa Cruz), washed, and measured for HDAC activity using HDAC Colorimetric Activity Assay/Drug Discovery Kits (AK-501, BIOMOL^® Research Laboratories, Plymouth Meeting, PA, USA) according to the manufacturer's recommendations. The light emission of deacetylated substrate was quantified at 405 nm.

Immunoprecipitation and immunoblotting

HEK293T cells were seeded 12–24 h before transfection and transfected with 8 g of plasmid in FuGene 6 regent (Roche). MG132 (10 ug/ml) was added 6 h prior to harvest; 48 h after transfection, the cells were lysed in NP40-based lysis buffer (0.1% NP40, 150 mM NaCl, 50 mM Tris-HCl). Immunoprecipitation experiments were performed using the indicated antibodies with protein A/G PLUS-Agarose (sc-2003, Santa Cruz). Each precipitate was washed and proteins were eluted with sample buffer. Immunoblotting was performed by standard protocols.

siRNA expression-plasmid vectors against ICBP90

The corresponding nucleotides of ICBP90 cloned into an siRNA expression vector, psiU6BX3, constructed previously by us (Shimokawa et al., 2004) were as follows: si-1: 1480–1500 si-2: 857–875, and si-3: 2123–2141. Each plasmid construct was transfected to HEK293 cells, and the cells were selected by G418 (900 ug/ml). At 48 h after transfection, the cells were harvested for Western blotting; after 9 days, and the cell viability was evaluated by MTT assay (Unoki and Nakamura, 2003b).

Diaminobenzidine tetrahydrochloride (DAB) staining

DBTRG-05MG cells were cultured at confluent status for 3 days and scratched by a disposable plastic PCR tip. At 20 h after scratching, the cells were fixed by 4% paraformaldehyde and treated for 3 min with 0.1% Triton X-100. DAB staining was carried out using DAKO EnVision Systems (K1390, DakoCytomation, Carpinteria, CA, USA) according to the manufacturer's recommended protocol. Anti-ICBP90 antibody was used to detect endogenous ICBP90.

Identification of UHRF1BP1

Endogenous ICBP90 protein in HEK293T cells was immunoprecipitated by anti-ICBP90 antibody; proteins binding to ICBP90 were separated and stained by Silver stain MS kit (Wako Pure Chemical Industries, Japan). Specific bands detected only in lanes containing proteins immunoprecipitated by anti-ICBP90 antibody were cut out and analysed by LC/MS-MS (APRO Life Science Institute).

References

1. Arima Y, Hirota T, Bronner C, Mousli M, Fujiwara T, Niwa S, Ishikawa H and Saya H. (2004). Genes Cells, 9, 131–142. | Article | PubMed | ISI | ChemPort |
2. Ballestar E and Wolffe AP. (2001). Eur. J. Biochem., 268, 1–6. | Article | PubMed | ISI | ChemPort |
3. Baumbusch LO, Thorstensen T, Krauss V, Fischer A, Naumann K, Assalkhou R, Schulz I, Reuter G and Aalen RB. (2001). Nucleic Acids Res., 29, 4319–4333. | Article | PubMed | ISI | ChemPort |
4. Bonapace IM, Latella L, Papait R, Nicassio F, Sacco A, Muto M, Crescenzi M and Di Fiore PP. (2002). J. Cell Biol., 157, 909–914. | Article | PubMed | ISI | ChemPort |
5. Burri N, Shaw P, Bouzourene H, Sordat I, Sordat B, Gillet M, Schorderet D, Bosman FT and Chaubert P. (2001). Lab. Invest., 81, 217–229. | Article | PubMed | ISI | ChemPort |
6. Citterio E, Papait R, Nicassio F, Vecchi M, Gomiero P, Mantovani R, Di Fiore PP and Bonapace IM. (2004). Mol. Cell. Biol., 24, 2526–2535. | Article | PubMed | ISI | ChemPort |
7. Daniel JM and Reynolds AB. (1999). Mol. Cell. Biol., 19, 3614–3623. | PubMed | ISI | ChemPort |
8. Daniel JM, Spring CM, Crawford HC, Reynolds AB and Baig A. (2002). Nucleic Acids Res., 30, 2911–2919. | Article | PubMed | ChemPort |
9. Fujimori A, Matsuda Y, Takemoto Y, Hashimoto Y, Kubo E, Araki R, Fukumura R, Mita K, Tatsumi K and Muto M. (1998). Mamm. Genome, 9, 1032–1035. | Article | PubMed | ISI | ChemPort |
10. Hopfner R, Mousli M, Jeltsch JM, Voulgaris A, Lutz Y, Marin C, Bellocq JP, Oudet P and Bronner C. (2000). Cancer Res., 60, 121–128. | PubMed | ISI | ChemPort |
11. Hopfner R, Mousli M, Oudet P and Bronner C. (2002). Anticancer Res., 22, 3165–3170. | PubMed | ISI | ChemPort |
12. Li Y, Mori T, Hata H, Homma Y and Kochi H. (2004). Biochem. Biophys. Res. Commun., 319, 464–468. | Article | PubMed |
13. Magnaghi-Jaulin L, Groisman R, Naguibneva I, Robin P, Lorain S, Le Villain JP, Troalen F, Trouche D and Harel-Bellan A. (1998). Nature, 391, 601–605. | Article | PubMed | ISI | ChemPort |
14. Miki K, Shimizu E, Yano S, Tani K and Sone S. (2000). Oncol. Res., 12, 335–342. | PubMed |
15. Miura M, Watanabe H, Sasaki T, Tatsumi K and Muto M. (2001). Exp. Cell. Res., 263, 202–208. | Article | PubMed |
16. Mori T, Li Y, Hata H and Kochi H. (2004). FEBS Lett., 557, 209–214. | Article | PubMed |
17. Mori T, Li Y, Hata H, Ono K and Kochi H. (2002). Biochem. Biophys. Res. Commun., 296, 530–536. | Article | PubMed | ISI | ChemPort |
18. Mousli M, Hopfner R, Abbady AQ, Monte D, Jeanblanc M, Oudet P, Louis B and Bronne C. (2003). Br. J. Cancer, 89, 120–127. | Article | PubMed | ISI | ChemPort |
19. Nguyen CT, Gonzales FA and Jones PA. (2001). Nucleic Acids Res., 29, 4598–4606. | Article | PubMed | ISI | ChemPort |
20. Nishidate T, Katagiri T, Lin ML, Mano Y, Miki Y, Kasumi F, Yoshimoto M, Tsunoda T, Hirata K and Nakamura Y. (2004). Int. J. Oncol. (in press).
21. Prokhortchouk A, Hendrich B, Jorgensen H, Ruzov A, Wilm M, Georgiev G, Bird A and Prokhortchouk E. (2001). Genes Dev., 15, 1613–1618. | Article | PubMed | ISI | ChemPort |
22. Shimokawa T, Furukawa Y, Sakai M, Li M, Miwa N, Lin YM and Nakamura Y. (2004). Cancer Res., 63, 6116–6120.
23. Trotzier MA, Bronner C, Bathami K, Mathieu E, Abbady AQ, Jeanblanc M, Muller CD, Rochette-Egly C and Mousli M. (2004). Biochem. Biophys. Res. Commun., 319, 590–595. | Article | PubMed | ISI | ChemPort |
24. Uemura T, Kubo E, Kanari Y, Ikemura T, Tatsumi K and Muto M. (2000). Cell. Struct. Funct., 25, 149–159. | Article | PubMed | ISI | ChemPort |
25. Unoki M and Nakamura Y. (2003a). FEBS Lett., 554, 67–72. | Article |
26. Unoki M and Nakamura Y. (2003b). Oncogene, 22, 2172–2185. | Article | PubMed | ChemPort |
27. Yoon HG, Chan DW, Reynolds AB, Qin J and Wong J. (2003). Mol. Cell, 12, 723–734. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We thank Y Miki, F Kasumi, and M Yoshimoto at The Japanese Foundation of Cancer Research (Tokyo, Japan) for offering breast cancer samples, S Ishii (RIKEN, Tsukuba, Japan) for obtaining HDAC1-FLAG expression plasmid, M Nishida (Kasumigaura National Hospital, Tsuchiura, Japan) for obtaining Ishikawa3-H-12 cells, and all our colleagues, in particular, T Katagiri and A Konuma for analysis of breast cancer samples, and T Shimokawa for constructing psiU6BX3 vector. This work was supported in part by Research for the Future Program Grant #00L01402 from the Japan Society for the Promotion of Science.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc).