ICBP90, an E2F-1 target, recruits HDAC1 and binds to methyl ... - Nature

Oncogene (2004) 23, 7601–7610

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

ORIGINAL PAPERS

ICBP90, an E2F-1 target, recruits HDAC1 and binds to methyl-CpG through its SRA domain Motoko Unoki1,2, Toshihiko Nishidate1 and Yusuke Nakamura*,1 1 Laboratory of Molecular Medicine, Human Genome Center, Institute of Medical Science, The University of Tokyo, 4-6-1 Shiorokanedai, Minato-ku, Tokyo 108-8639, Japan

ICBP90, inverted CCAAT box-binding protein of 90 kDa, has been reported as a regulator of topoisomerase IIa expression. We present evidence here that ICBP90 binds to methyl-CpG when at least one symmetrically methylated-CpG dinucleotides is presented as its recognition sequence. A SET and RING finger-associated (SRA) domain accounts for the high binding affinity of ICBP90 for methyl-CpG dinucleotides. This protein constitutes a complex with HDAC1 also via its SRA domain, and bound to methylated promoter regions of various tumor suppressor genes, including p16INK4Aand p14ARF, in cancer cells. It has been reported that expression of ICBP90 was upregulated by E2F-1, and we confirmed that the upregulation was caused by binding of E2F-1 to the intron1 of ICBP90, which contains two E2F-1-binding motifs. Our data also revealed accumulation of ICBP90 in breast-cancer cells, where it might suppress expression of tumor suppressor genes through deacetylation of histones after recruitment of HDAC1. The data reported here suggest that ICBP90 is involved in cell proliferation by way of methylation-mediated regulation of certain genes. Oncogene (2004) 23, 7601–7610. doi:10.1038/sj.onc.1208053 Published online 13 September 2004 Keywords: ICBP90; methyl-CpG-binding HDAC1; SRA domain; E2F-1; UHRF1BP1

protein;

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 *Correspondence: Y Nakamura; E-mail: [email protected] 2 Current address: Laboratory of Human Carcinogenesis, National Cancer Institute, NIH, Building 37, Room 3060, 37 Convent Drive, Bethesda, MD 20892-4255, USA Received 13 May 2004; revised 20 July 2004; accepted 20 July 2004; published online 13 September 2004

(Prokhortchouk et al., 2001; Daniel et al., 2002). Kaiso was identified as a p120ctn-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 IIa 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/p21WAF1-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

ICBP90 binds to methyl-CpG with HDAC1 M Unoki et al

7602

(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/MSMS 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

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 Oncogene


7603

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 IIa promoter (Figure 1b, c: oligo7, TopoII ICBP2). 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, ICBP90DSRADRING (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). 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)

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 Oncogene


7604

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, ICBP90DRING and ICBP90 SRAFLAG could bind to HDAC1 (Figure 4c, upper panels), while neither ICBP90DSRADRING 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 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-b-actin immunoprecipitation (ab-actin) were performed. The PCR conditions are also described in Supplementary Table S1

analysis to investigate the biological function of ICBP90. After confirming the absence of homozygous deletions, except in the p16INK4A and p14ARF 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 p14ARF 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 RARb 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 IIa was very low (Figure 1b). 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 Oncogene

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 cellproliferation status. 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 IIa, correlated well with that of ICBP90 but expression levels of p16INK4A and p14ARFdid not. Although downregulation of ICBP90 by p53/p21-dependent DNA-damage checkpoint signals was reported recently (Arima et al., 2004), expression of p21WAF1 was not always increased by contact-


7605

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 FLAGtagged HDAC1 were immunoprecipitated with anti-Myc or anti-FLAG antibodies, and immunodetected by anti-FLAG or anti-Myc antibodies

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). Identification of an additional binding partner of ICBP90

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)

inhibition signals (Figure 6c), suggesting that different signals might be inducing downregulation of ICBP90 under confluent conditions. A scratching assay using

Moreover, we identified an approximately 170 kDa novel ICBP90-binding partner designated ICBP90binding 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 UHRF1BP1Oncogene


7606

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. Discussion

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. b-Actin served as an internal control. (c) RT–PCR results of ICBP90, E2F1, topoisomerase IIa, p21WAF1, p16INK4A, and p14ARF. 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

were unchanged at the confluent or proliferative stage in both the contact inhibition-positive and -negative cells (Figure 6c). 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 ¼ 0.0008: w2-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– Oncogene

Methylation of CpG-dinucleotides is receiving increasing attention because of its important role in the regulation of gene expression. Among the five methylCpG-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 IIa 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 methylCpG 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 (MagnaghiJaulin 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-CpGbinding 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


7607

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

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 p16INK4A and p14ARF 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/p21WAF1dependent 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 Oncogene


7608 Table 1 Expression level of ICBP90 of breast cancer patients and their clinicopathological features ID no.

Age in operation

Classification

Stage

T

N

M

Histological type

Hormone receptor ER

PgR

Menopausal status

Fold change Cy5/Cy3

MMK010178 MMK010591 MMK010613 MMK010680 MMK010711 MMK010744

51 40 37 58 51 41

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

1a 1a 1a 1a 1a 1a

+ +

+ + + + + +

Pre Pre Pre Post Pre Pre

1.05 2.90 3.11 1.26 1.10 0.30

MMK010004 MMK010005 MMK010013 MMK010086 MMK010129 MMK010138 MMK010214 MMK010327 MMK010411 MMK010471 MMK010500 MMK010521 MMK010623 MMK010631 MMK010758 MMK010760 MMK010818 MMK010864 MMK010869

47 44 45 42 52 38 42 43 46 42 45 21 39 41 40 42 51 52 45

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

1 0 1 0 2 0 1 1 0 1 0 0 1 0 1 0 0 1 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2 2 2 2 3 2 2 2 2 2 2 2 2 2 2 2 2 2 2

a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1 a1

+ + + + + + + + + + + + +

+ + + + + + + + + + + + +

Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre

0.35 2.16 1.37 0.90 0.98 0.65 1.63 1.42 2.34 1.24 0.31 2.78 1.40 0.32 1.23 1.60 0.84 1.44 6.33

MMK010102 MMK010247 MMK010252 MMK010255 MMK010302 MMK010473 MMK010478 MMK010502 MMK010508 MMK010769 MMK010779 MMK010780

51 48 52 47 46 40 38 51 51 33 46 31

2 2 2 2 2 2 2 2 2 2 2 2

1 1 1 0 1 1 2 0 1 0 1 0

0 0 0 0 0 0 0 0 0 0 0 0

3 2 2 2 2 2 3 2 2 2 2 2

a2 a2 a2 a2 a2 a2 a2 a2 a2 a2 a2 a2

+ +

+ +

Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre

1.96 0.40 1.97 1.25 7.09 9.87 7.84 2.99 4.72 17.93 13.48 3.20

MMK010003 MMK010031 MMK010145 MMK010149 MMK010175 MMK010304 MMK010431 MMK010453 MMK010491 MMK010554 MMK010571 MMK010644 MMK010646 MMK010709 MMK010724 MMK010762 MMK010772 MMK010781 MMK010794 MMK010858

51 29 51 35 38 48 50 49 46 51 45 47 37 33 40 50 45 44 52 42

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

1 2 1 0 0 1 0 1 0 0 1 2 1 0 1 1 1 0 1 1

0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0

2 3 2 2 2 2 2 2 2 2 4 2 2 2 2 2 2 2 2 2

a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3 a3

+ + + + + + + + + + + + + + +

+ + + + + + + + + + + + + + +

Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre Pre

0.35 4.37 5.70 5.56 8.23 1.96 4.85 1.14 () 2.10 4.61 1.04 0.76 7.53 1.61 0.90 3.64 5.80 1.33 5.83

Expression level of ICBP90 of breast cancer patients and their clinicopathological features. All information were judged according to the Japanese Breast Cancer Society. T, Tumor stage; N, lymph-node metastasis status; M, distant metastasis. Histological type; 1a, ductal carcinoma in situ; a1, papillotubular carcinoma (well-differentiated); a2, solid tubular carcinoma (moderately differentiated); a3, scirrhous carcinoma (poorly differentiated). A mixture of RNAs isolated from normal breast ductal cells of 15 premenopausal breast-cancer patients was labeled by Cy3 and served as a normal control. RNAs isolated from each breast cancer samples was labeled by Cy5. This fold change indicates expression level of ICBP90 in breast cancer compared with normal tissue. () in Cy3/Cy5 column indicates no signal in Cy5 spot probably due to hybridization miss

Oncogene


7609

us. The combined results underscored a significant role of ICBP90 in methylation-dependent transcriptional regulation, and suggested potential involvement of ICBP90 in human carcinogenesis.

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).

Materials and methods

EMSA combined with Western blotting

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 b-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 KODPlus (TOYOBO, Tokyo, Japan). The entire coding sequence of ICBP90, and of deletion mutants designated ICBP90DRING, ICBP90DRINGDSRA, and ICBP90UbL, were cloned into pcDNA3.1( þ ) (Invitrogen, Carlsbad, CA, USA). The entire coding sequence of UHRF1BP1 and ICBP90 SRA were cloned into pFLAGs-CMV-5a (SIGMA). The entire coding sequence

HEK293T cells were plated and each culture was transfected by lipofection (FuGENETM6 Transfection Regent; Roche) with 8 mg 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, doublestranded oligonucleotides, and indicated antibody. Doublestranded 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 b-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( þ ) 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 antiICBP90 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, BIOMOLs 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 mg 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. Oncogene


7610 siRNA expression-plasmid vectors against ICBP90

Identification of UHRF1BP1

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).

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).

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.

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.

References Arima Y, Hirota T, Bronner C, Mousli M, Fujiwara T, Niwa S, Ishikawa H and Saya H. (2004). Genes Cells, 9, 131–142. Ballestar E and Wolffe AP. (2001). Eur. J. Biochem., 268, 1–6. 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. 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. 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. 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. Daniel JM and Reynolds AB. (1999). Mol. Cell. Biol., 19, 3614–3623. Daniel JM, Spring CM, Crawford HC, Reynolds AB and Baig A. (2002). Nucleic Acids Res., 30, 2911–2919. 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. 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. Hopfner R, Mousli M, Oudet P and Bronner C. (2002). Anticancer Res., 22, 3165–3170. Li Y, Mori T, Hata H, Homma Y and Kochi H. (2004). Biochem. Biophys. Res. Commun., 319, 464–468. Magnaghi-Jaulin L, Groisman R, Naguibneva I, Robin P, Lorain S, Le Villain JP, Troalen F, Trouche D and HarelBellan A. (1998). Nature, 391, 601–605.

Miki K, Shimizu E, Yano S, Tani K and Sone S. (2000). Oncol. Res., 12, 335–342. Miura M, Watanabe H, Sasaki T, Tatsumi K and Muto M. (2001). Exp. Cell. Res., 263, 202–208. Mori T, Li Y, Hata H and Kochi H. (2004). FEBS Lett., 557, 209–214. Mori T, Li Y, Hata H, Ono K and Kochi H. (2002). Biochem. Biophys. Res. Commun., 296, 530–536. Mousli M, Hopfner R, Abbady AQ, Monte D, Jeanblanc M, Oudet P, Louis B and Bronne C. (2003). Br. J. Cancer, 89, 120–127. Nguyen CT, Gonzales FA and Jones PA. (2001). Nucleic Acids Res., 29, 4598–4606. 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). Prokhortchouk A, Hendrich B, Jorgensen H, Ruzov A, Wilm M, Georgiev G, Bird A and Prokhortchouk E. (2001). Genes Dev., 15, 1613–1618. Shimokawa T, Furukawa Y, Sakai M, Li M, Miwa N, Lin YM and Nakamura Y. (2004). Cancer Res., 63, 6116–6120. 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. Uemura T, Kubo E, Kanari Y, Ikemura T, Tatsumi K and Muto M. (2000). Cell. Struct. Funct., 25, 149–159. Unoki M and Nakamura Y. (2003a). FEBS Lett., 554, 67–72. Unoki M and Nakamura Y. (2003b). Oncogene, 22, 2172–2185. Yoon HG, Chan DW, Reynolds AB, Qin J and Wong J. (2003). Mol. Cell, 12, 723–734.

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

Oncogene