Oncogene (2006) 25, 5807–5822 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ONCOGENOMICS Promoter CpG hypermethylation and downregulation of XAF1 expression in human urogenital malignancies: implication for attenuated p53response to apoptotic stresses

M-G Lee1,4, J-S Huh2,4, S-K Chung1, J-H Lee1, D-S Byun1, B-K Ryu1, M-J Kang1, K-S Chae1, S-J Lee3, C-H Lee3, JI Kim3, S-G Chang3 and S-G Chi1

1School of Life Sciences and Biotechnology, Korea University, Seoul, Korea; 2Department of Urology, College of Medicine, Cheju National University, Cheju, Korea and 3Department of Urology, School of Medicine, Kyung Hee University, Seoul, Korea

XIAP-associated factor 1 (XAF1) is a new candidate Introduction tumor suppressor, which has been known to exert proapoptotic effects by interfering with the caspase- Apoptosis plays an essential role for elimination of inhibiting activity of XIAP. To explore the XAF1’s defective or potentially dangerous cells and provides a candidacy for a suppressor in urogenital tumorigenesis, we defense against malignant transformation (Thompson, investigated the XAF1 status in a series of cancer cell lines 1995). The inhibitor of apoptosis (IAPs) are a and primary tumors derived from the bladder, kidney and family of intrinsic cellular antiapoptotic proteins (De- prostate. Expression of XAF1 transcript was undetectable veraux and Reed, 1999; Sun et al., 1999). The human or extremely low in 60% (3/5) of bladder, 66% (10/15) of IAP family includes cIAP-1, cIAP-2, XIAP, NAIP, kidney, and 100% (3/3) prostate cancer cell lines. survivin, apollon, ILP2, and livin, and several members Abnormal reduction of XAF1 was also found in 33% of human IAP family, including XIAP, c-IAP-1 and c- (18/55) of primary bladder and 40% (8/20) of primary IAP-2, have been shown to prevent apoptosis by binding kidney tumors, and showed a correlation with advanced to caspase-3, -7 and -9 and thereby protecting them from stage and high grade of bladder tumor. Hypermethylation cleavage activation (Deveraux et al, 1997; Roy et al., at 14 CpG sites in the 50 proximal region of the XAF1 1997; Green, 2000). Deregulation of IAPs plays a key promoter was highly prevalent in cancers versus adjacent role in the aberrantly increased cell viability normal or benign tissues and tightly associated with and resistance to the anticancer therapy in human reduced expression. XAF1 expression enhanced the cancers whereas overexpression appears to suppress apoptotic response of tumor cells to chemotherapeutic apoptosis against a large variety of triggers including agents, such as etoposide or 5-FU. While XAF1 expres- tumor necrosis factor (TNF), Fas, staurosporin, sion did not influence the subcellular distribution or etoposide and growth factor withdrawal (Duckett expression of XIAP, it elevated the stability of et al., 1996; Liston et al., 1996; Ambrosini et al., 1997; p53and its target gene expression. Moreover, the Li et al., 1998). Of the known human IAP proteins, apoptosis-sensitizing and growth suppression function of XIAP is the most potent and versatile inhibitor of XAF1 was markedly impeded by blockade of p53 caspases and apoptosis. XIAP mRNA levels are function. Collectively, our study demonstrates that relatively high in the majority of cancer cell lines and epigenetic alteration of XAF1 is frequent in human high levels of XIAP protein are generally found in urogenital cancers and may contribute to the malignant human cancers including bladder and renal cell carci- progression of tumors by rendering tumor cells a survival nomas (Fong et al., 2000; Tamm et al., 2001; Bilim et al., advantage partially through the attenuated p53response 2003). XIAP protects tumor cells from apoptosis to apoptotic stresses. induced by many triggers, including irradiation, serum Oncogene (2006) 25, 5807–5822. doi:10.1038/sj.onc.1209867; withdrawal and chemotherapeutic compounds (Holcik ONCOGENOMICS published online 7 August 2006 et al., 2001). It has been shown that XIAP regulates Akt activity and caspase-3-dependent cleavage, and XIAP Keywords: XAF1; promoter hypermethylation; CpG downregulation induces apoptosis in chemoresistant site; urogenital cancer; apoptosis; p53 human ovarian cancer cells (Sasaki et al., 2000; Asselin et al., 2001). The caspase-inhibiting activity of XIAP can be suppressed by two mitochondrial proteins, Smac/DIA- Correspondence: Dr S-G Chi, School of Life Sciences and Biotechno- BLO and HtrA2, which are released from mitochondria logy, Korea University, 136-701 Seoul, Republic of Korea. into the cytoplasm during apoptosis, where they bind to E-mail: [email protected] the hydrophobic pocket in BIR3 of XIAP (Du et al., 4These authors contributed equally to this article. Received 17 November 2005; revised 24 February 2006; accepted 27 June 2000; Verhagen et al., 2000; Suzuki et al., 2001). 2006; published online 7 August 2006 Recently, XIAP-associated factor 1 (XAF1), which is Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5808 located at 17p13.2, was identified as a novel negative Results regulator of XIAP based on its ability to bind XIAP (Liston et al., 2001). XAF1 encodes 33.1 kDa protein Altered expression of XAF1 in human urogenital cancer with seven zinc fingers and its mRNA is ubiquitously cell lines and primary tumors expressed in all normal adult and fetal tissues, but To investigate the candidacy of XAF1 as a suppressor in absent or present at very low or undetectable levels in urogenital tumorigenesis, we initially characterized the various cancer cell lines. XAF1 sensitizes tumor cells to expression status of XAF1 transcript in five bladder, 15 the proapototic effects of etoposide and prevents the kidney and three prostate cancer cell lines. XAF1 XIAP-mediated inhibition of caspase-3. Moreover, the expression was undetectable or very low in three of five antiapoptotic ability of XIAP in serum withdrawal or (60%)bladder (T24, HT1197 and 253J/BV),10 of 15 etoposide-induced apoptosis was restored in tumor cells (66%)kidney (UOK107, UOK108, UOK109, UOK112, by infection of adeno-XAF1 antisense. These observa- UOK115, UOK122, UOK123, UOK130, ACHN and tions thus suggest that deregulation of apoptosis WWC1), and all of three (100%) prostate (LNCaP, through the loss of XAF1 expression might be important DU145 and PC-3)cancer cell lines (Figure 1a).By for malignant cell survival, and a high level of XIAP to contrast, Smac/DIABLO and HtrA2, two other XIAP XAF1 expression in cancer cells may provide a survival antagonists showed easily detectable expression in all advantage through the relative increase of XIAP cell lines tested. Expression levels (Target/GAPDH)of antiapoptotic function. XAF1, Smac/DIABLO, and HtrA2 in cancer cell lines XAF1 was originally reported as a nuclear protein were observed within the range of 0.00–1.40 (mean; to exert its proapoptotic effect by directly interacting 0.68), 1.04–1.36 (mean; 1.18) and 0.82–1.22 (mean; with XIAP and inducing XIAP sequestration in 0.98), respectively (Figure 1b). XAF1 levels determined nuclear inclusions (Liston et al., 2001). A recent tissue by semiquantitative RT–PCR were fairly consistent with microarray assay showed a significant reduction of the results from Northern blot assay (Figure 1a). It has XAF1 expression in 15 of 16 human melanoma cell lines been suggested that a high level of XIAP to XAF1 due and a large fraction of primary melanomas (Ng et al., to either elevated expression of XIAP or reduction of 2004). While high level of XAF1 was detected in XAF1 may provide a survival advantage for tumor cells both nucleus and cytoplasm in normal melanocyte, through the relative increase of XIAP antiapoptotic the percentage of XAF1 positive cells and the staining function (Liston et al., 2001). However, in contrast to intensity of XAF1 were significantly decreased in a XAF1, the majority of urogenital tumor cells we tested majority of melanoma tissues compared to benign exhibited similar expression of XIAP at both mRNA melanocytic nevi. Given the fact that XAF1 could and protein levels (Figure 1c). trigger the relocalization of XIAP in the cytoplasm, it Next, we evaluated XAF1 expression in 55 bladder was proposed that loss of XAF1 expression in transitional cell carcinoma (TCC)and 20 renal cell tumor may increase the functional pool of cytoplasmic carcinoma (RCC)tissues. In both tumor types, a XIAP, which in turn deregulate the normal apoptotic substantial fraction of tumors showed markedly reduced process and contribute to the malignant nature of XAF1 expression (mean; 0.74 and 0.64 in TCC and tumors (Ng et al., 2004). Interestingly, it has been RCC, respectively)compared to its noncancerous reported that the natural resistance of the adult counterparts (mean; 1.20 and 1.16 in bladder and motoneurons to axotomy is abolished when the ratio kidney, respectively)( Po0.005)(Figure 2a and b).In of XAF1 to XIAP is increased, and that XAF1 is matched tissue sets examined, 50% (10 of 20)of TCC increased and redistributed during reperfusion after and 60% (12 of 20)of RCC cases revealed tumor- focal cerebral ischemia in rats, which is accompanied by specific reduction of XAF1 (Figure 2c). We arbitrarily diffuse redistribution of XIAP and activation of classified tumors with expression levels less than a half caspase-3 (Perrelet et al., 2004; Siegelin et al., 2005). (o0.60 in TCC; o0.58 in RCC)of normal means as Therefore, the balance between XIAP and XAF1 may abnormally low expressors. Abnormal reduction of play an essential role for adult motoneuron protection XAF1 was identified in 33% (18 of 55)of TCC and and implicated in the pathophysiological mechanisms of 35% (seven of 20)of RCC (Figure 2b).In bladder reperfusion injury as well as tumorigenesis. tumor, abnormal reduction of XAF1 was significantly In the present study, we demonstrates that XAF1 higher in muscle-invasive tumors (T2–T4, 13 of 28 expression is absent or significantly downregulated (46%)) compared to superficial tumors (Ta–T1, five of in a substantial fraction of urogenital malignancies 27 (19%)) (Po0.001), and more frequent in high grade by aberrant promoter CpG sites hypermethylation, tumors (grade II–III, 14 of 37 (38%)) than low grade and that abnormal reduction of XAF1 correlates tumors (grade I, four of 18 (22%)) (Po0.01) with advanced stage and high grade of bladder (Figure 2d). However, XAF1 expression showed no tumor. Our data also show that XAF1 expression correlation with morphologic patterns of tumors greatly increases tumor cell sensitivity to apoptosis (papillary, 14 of 44 (32%); non-papillary, four of 11 triggered by chemotherapeutic drugs, suggesting that (36%)), age (o40, two of six (33%); 41–60, five of 16 epigenetic inactivation of XAF1 might provide a (31%); >61, 11 of 33 (33%)), sex (male, 15 of 47 (32%); selective advantage for tumor cell survival and thus female, three of eight (38%)), and smoking status (none, contribute to the malignant progression of human seven of 17 (41%); >10 years, 11 of 31 (36%)) of urogenital cancers. the bladder cancer patients. XAF1 association with

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5809

Figure 1 Expression status of XAF1, Smac/DIABLO, HtrA2 and XIAP in human bladder, kidney and prostate cancer cell lines. (a)Semiquantitative RT-PCR and Northern blot analysis of XAF1, Smac/DIABLO, and HtrA2 expression in cancer cell lines. (b)Expression levels of XAF1, Smac/DIABLO, and HtrA2 transcripts in cancer cell lines. Quantitation was achieved by densitometric scanning of RT-PCR products in ethidium bromide-stained gels and absolute area integrations of the curves representing each specimen were compared after adjustment for GAPDH. Data represent means of triplicate assays. (c)XIAP expression in urogenital cancer cell lines. Expression of XIAP mRNA was analysed by semiquantitative RT-PCR. For Western blot assay of XIAP protein expression, 30 mg of total protein was fractionated using 10% SDS–PAGE and XIAP was detected using an anti-XIAP monoclonal antibody and enhanced chemiluminescence.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5810

Figure 2 Expression status of XAF1 in human primary bladder and kidney carcinomas. (a)Semiquantitative RT–PCR analysis of XAF1, Smac/DIABLO, and HtrA2 expression in primary TCC of the bladder, primary RCC, and adjacent normal tissues. NB, normal bladder tissue; NK, normal kidney tissue. (b)Expression levels of XAF1, Smac/DIABLO, and HtrA2 transcripts in primary carcinomas. RT–PCR was repeated at least three times for each specimen and the means were obtained. Bar indicates the mean expression level of each specimen group (c)XAF1 expression in matched tissue sets. Expression level of XAF1 mRNA in cancer and adjacent noncancerous tissues were compared using matched sets obtained from the same cancer patients (P). N, normal tissue; T, tumor tissue. (d)Comparison of XAF1 expression levels between superficial (Ta–T1)and invasive (T2–T4)tumors and low- and high-grade tumors of the bladder.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5811 clinicopathologic features of RCC was not defined in no expression. Likewise, methylation was detected in 16 this study due to the small number of cases analyzed. of 18 (89%)TCC and six of seven (86%)RCC tissues, which show low expression, while all noncancerous tissues and tumors with normal XAF1 expression were Tumor-specific downregulation of XAF1 by promoter unmethylated. In addition, 35% (seven of 20)of primary CpG sites hypermethylation prostate carcinomas but none of 10 BPH specimens The XAF1 gene is located at the 17p13.2, approximately were identified to be methylated at these regions. 3 cM to the p53 tumor suppressor, which undergoes frequent allelic losses in many human malignancies. To Aberrant CpG sites hypermethylation of the XAF1 define whether altered expression of XAF1 is associated promoter with gene deletion, we tested genomic level of XAF1 and To further define the relationship between aberrant p53 in the cell lines and tumor tissues. Five cancer cell hypermethylation and gene expression, we characterized lines (J82, T24, UOK122, UOK123 and PC-3), which the methylation status of 14 CpG sites located in the 50 have been known to carry allelic deletion of p53, proximal region (nucleotides –20 to –695)of the XAF1 exhibited low genomic level of p53, while none of the promoter using bisulfite DNA sequencing analysis cell lines displayed detectable reduction of XAF1 gene (Figure 4a). Bisulfite-modified genomic DNA, which level (Figure 3a). Similarly, all 75 primary tumors (55 were extracted from nine cancer cell lines (two normal TCCs and 20 RCCs)tested showed XAF1 gene level and seven abnormal expressors)and 14 primary comparable to that of normal tissues while some tumors carcinomas (six TCCs, four RCCs and four prostate revealed low level of p53. To further define the genomic tumors), were subjected to PCR amplification for the status of XAF1, loss of heterozygosity (LOH)of promoter region spanning the 14 CpG sites and 10 PCR 17p13.1–13.2 region was evaluated using three poly- clones from each specimens were sequenced to deter- morphic markers (D17S796, D17S1832 and D17S1828) mine methylation frequency at individual CpG sites (Figure 3b). Among 40 matched sets (20 TCCs and 20 (complete methylation; 80–100%, partial methylation; RCCs)we tested, nine (23%)were informative for the 20–60% and unmethylation; 0%). The majority (12–14 centromeric marker (D17S796), which is located approx- sites; 86–100%)of the sites was completely or partially imately 2.5 cM telomeric of the p53 locus, and LOH was methylated in cells with no or extremely low expression detected from five (56%)cases. Intriguingly, four (80%) (253J, LNCaP, ACHN, DU145 and PC-3)and 9–11 of the five LOH tumors showed a marked decrease in (64–79%)sites were partially methylated in low p53 gene level, while none of the tumors exhibited expressors (HT1197 and T24), whereas only 2–3 detectable reduction of XAF1 at both genomic and (14–21%)sites showed partial methylation in normal transcriptional levels. In addition, none of 10 (25%) expressors (J82 and HT1376)( P 0.001)(Figure 4b). cases informative for telomeric markers (D17S1832 and o Likewise, six primary tumors (four TCCs and two D17S1828)revealed LOH, suggesting that LOH at RCCs)with abnormally low expression harbored D17S796 might be associated with allelic loss of p53, complete or partial methylation at 10–12 (71–86%) and genomic deletion at this region including p53 locus CpG sites, whereas its adjacent noncancerous tissues as might rarely extend into the XAF1 gene. well as four tumors with normal expression carried Next, to elucidate whether aberrant DNA methyla- partial methylation at less than four sites (P 0.001) tion is associated with XAF1 downregulation, tumor o (Figure 4c). Consistent with the result of MSP assay, cells with no or low expression (T24, HT1197, 253J, two MSP-positive prostate tumors were found to have UOK123, ACHN, LNCaP, DU145 and PC-3)were complete or partial methylation at 10–12 (71–86%)CpG treated with the demethylating agent 5-aza-20-deoxycy- sites, whereas two MSP-negative tumors and two BPH tidine (5-Aza-dC). XAF1 expression was reactivated or tissues were detected to have only partial methylation at elevated in these cells following 5-Aza-dC treatment in a 2 (14%)sites. Together, these results demonstrate that dose-dependent manner (Figure 3c and d). To determine methylation of promoter CpG sites is tightly associated the overall frequency of XAF1 hypermethylation in with downregulation of XAF1 expression in urogenital tumors, we carried out methylation-specific PCR (MSP) cancer cell lines and primary tumors (supplementary analysis of the XAF1 promoter sequences. Methylation- data; Table 1). specific primers (MS and MAS)and unmethylation- specific primers (US and UAS)were designed to amplify nucleotides À687 to À458 and À354 to À13, respec- Absence of somatic mutations of XAF1 tively, based on our previous finding of a tight To explore the presence of mutational alteration of correlation of CpG sites hypermethylation in these XAF1, we performed RT–PCR–SSCP analysis for regions with decreased XAF1 expression in gastric XAF1 transcripts expressed from 11 tumor cell lines cancers (Byun et al., 2003). Methylation was detected (four bladder and seven kidney cell lines)and 75 from all 16 cell lines having no or abnormally low XAF1 primary tumor tissues (55 TCCs and 20 RCCs)and expression, but not found from the seven cell lines with DNA-PCR-SSCP analysis of the XAF1 gene for 20 normal expression (Figure 3e). Both methylation and primary prostate tumors. The entire coding region of unmethylation signals were observed from T24 and XAF1 transcript was amplified using seven sets of exon- HT1197 cells with low expression, and only methylation specific primers for RT–PCR-SSCP and eight sets signal was identified in ACHN and PC-3 cells showing of intron-specific primers for DNA-PCR–SSCP. To

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5812

Figure 3 LOH and methylation analysis of XAF1 in cancer cell lines and primary tumors. (a)Genomic levels of XAF1 and p53 in cancer cell lines and tumor tissues. Exon 6 of XAF1 and exon 8 of p53 were amplified by intron-specific primers. Ten ml of the PCR products were resolved on a 2% agarose gel. GAPDH was used as an endogenous control. NB, normal bladder tissue; NK, normal kidney tissue. (b)No association of LOH at D17S796 with genomic status of XAF1. At centromeric marker D17S796, 5 tumor specimens exhibited LOH in the tumor DNA (T)compared with its corresponding normal DNA (N).Genomic level of XAF1 and p53 were evaluated using quantitative DNA-PCR. (c)Effect of 5-Aza-dC treatment on XAF1 expression in urogenital cancer cell lines. Cancer cell lines with no or low XAF1 expression were treated with 5-Aza-dC (5 mM)for 5 days and XAF1 expression were evaluated by RT-PCR. C, control; T, treated. (d)Representative example of a dose-dependent reactivation of XAF1 mRNA expression by 5-Aza-dC in cancer cells. A kidney cancer cell line ACHN was treated with increasing doses of 5-Aza-dC (2.5–20 mM)for 5days and expression level of XAF1 mRNA was determined by RT-PCR. (e)Methylation-specific PCR analysis of the XAF1 promoter in cancer cell lines and tissues. Fifty nanograms of bisulfite-modified genomic DNA was subjected to PCR amplification of the XAF1 promoter sequences using unmethylation (U)-specific and methylation (M)-specific primer sets, which cover nucleotides À354 to À13 (341 bp)and À687 to À458 (230 bp), respectively. NB, normal bladder tissue; NK, normal kidney tissue; CaP, carcinoma of the prostate.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5813

Figure 4 Bisulfite DNA sequencing analysis of the XAF1 promoter. (a)A map of the CpG sites of the 5 0 upstream region of the XAF1 gene. Fourteen CpGs (nucleotide À23 to À692)analysed by bisulfite DNA sequencing are represented by vertical lines with circle and numbered 1–14. The first nucleotide of ATG start codon is indicated by an arrow at þ 1. (b)Methylation status of the 14 CpG sites in the XAF1 promoter region of cancer cell lines. The promoter region comprised of 14 CpGs was amplified by PCR using primers MS7 and MS2. The PCR products were cloned and 10 plasmid clones were sequenced for each cell line. Percent methylation was determined from the number of clones containing a methylated CpG at each position relative to the total number of clones analysed. Black, gray, and white circles represent complete methylation (>7 clones), partial methylation (1–6 clones), and unmethylation, respectively. (c) Methylation status of the 14 CpGs in primary tumor and its adjacent noncancerous tissues. NB, normal bladder tissue; NK, normal kidney tissue; CaP, carcinoma of the prostate.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5814

Figure 5 Effect of XAF1 expression on cell proliferation and apoptosis. (a)[3H]thymidine uptake assay of XAF1 effect on DNA synthesis. Cells were transfected with increasing doses of XAF1 expression plasmid or siRNA against XAF1. After 44 h transfection, 3 cells were pulse-labeled for 4 h with 1 mCi/ml [ H]thymidine. The radioactivity incorporated into trichloroacetic acid-precipitable materials was counted by a liquid scintillation counter. Data represent means of triplicate assays (Bars, s.d.). (b)No effect of XAF1 expression on cell cycle progression. 253J cells were transfected with GFP-XAF1 expression plasmids, and the distribution of cell cycle phases of GFP-positive cells were analyzed using flow cytometry. (c)TUNEL assay for XAF1 effect on tumor cell apoptosis. Cells transfected with XAF1 expression plasmid or XAF1 siRNA were treated with etoposide (10 mM)or 5-FU (10 mM)for 36 h. Apoptotic cell death was measured by using TUNEL assay. Data represent means of triplicate assays (***Po0.001; **Po0.01; *Po0.05).

improve mutation detection sensitivity, the same PCR detect any types of mutation leading to amino-acid products were digested using different restriction en- substitutions or frameshifts, whereas all of the pre- donuclease(s)and SSCP was carried out under two viously reported p53 mutations in the cell lines were different running conditions. However, we failed to detected under the same conditions.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5815 Increased cellular response to apoptosis-inducing cells with varying expression levels of XAF1 are chemotherapeutic drugs by XAF1 expression comparable to those of vector-transfected or untrans- Frequent downregulation of XAF1 in tumors suggests fected controls (Figures 5b and 6c). that XAF1 may play an important role in tumorigenesis. To explore the growth-suppressive role for XAF1, we initially assessed XAF1 effect on cell proliferation. The Induction of p53 by XAF1 through the regulation of XAF1-nonexpressing 253J cells were transfected with protein stability expression plasmids designed to produce XAF1 protein We next investigated whether the apoptosis-sensitizing with a Flag tag at the N-terminus (Flag-XAF1), and effect of XAF1 is associated with activation of p53, XAF1-expressing HT1376 cells were transfected with which plays a critical role in apoptosis induction siRNA against XAF1 to knockdown its endogenous triggered by a variety of stresses. Intriguingly, we found XAF1 expression (Supplementary data; Figure s1). that p53 protein level is markedly elevated in XAF1- [3H]thymidine uptake analysis showed that both transfected 253J cells following etoposide treatment XAF1-restored 253J and XAF1-disrupted HT1376 have (Figure 7a and Figure s1). XAF1 expression led to a no detectable change in DNA replication rate compared dose-associated induction of p53 protein and its target to empty vector- or control siRNA-transfected cells gene expression, including p21Waf1, PUMA and NOXA. (Figure 5a). Flow cytometric analysis also revealed that Moreover, a higher and more sustained induction of p53 ectopic overexpression of XAF1 in 253J cells does not protein was detected in XAF1-versus vector-transfected cause detectable change in the distribution of cell cycle cells, indicating that the presence of functional XAF1 phases, while it slightly increases basal level of apoptosis causes a prolonged activation of p53 in response to (Figure 5b). Restoration of XAF1 expression in 253J apoptotic damage (Figure 7b). To define the role for cells led to a slight increase of apoptosis in the absence XAF1 in p53 accumulation, the cells were exposed to of etoposide or 5-FU treatment but greatly activated the protein synthesis inhibitor cyclohexamide (CHX) apoptosis induction following treatment (Figure 5c). and XAF1 effect on p53 stability was determined. As Likewise, siRNA-mediated XAF1 knockdown in seen in Figure 7c, p53 accumulation in etoposide-treated HT1376 cells attenuated the apoptotic response of the cells was higher in XAF1-transfected cells compared to cells to the drugs. These results indicate that expression vector-transfected cells, supporting that XAF1 upregu- of XAF1 strongly enhances the cellular sensitivity to lation of p53 occurs at the post-translational level. To apoptosis-inducing stimuli. determine whether functional p53 is required for the proapoptotic action of XAF1, we utilized PC-3 prostate cancer cells, which express extremely low level of XAF1 No visible effect of XAF1 expression on subcellular and no functional p53 protein. While XAF1 transfection localization of XIAP alone has a negligible effect on both basal and etopo- To elucidate whether the proapoptotic action of XAF1 side-induced apoptosis, co-transfection with wild-type is accompanied with nuclear translocation of XIAP, p53 led to a dramatic elevation of apoptosis (Figure 7d). the effect of XAF1 overexpression on the subcellular TUNEL assay demonstrated that after 36 h exposure to distribution of XIAP protein was investigated etoposide (20 mM), approximately 69% of PC-3 cells using fractionation assay. XAF1-transfected 253J cells display apoptotic death by co-transfection of XAF1 and were treated to etoposide for 24–72 h to induce p53, while only 18 and 29% of the cells undergo apoptosis and equivalent amounts of protein from apoptosis by transfection of XAF1 and p53, respec- cytoplasmic and nuclear fractions were assessed for tively. Requirement of functional p53 for XAF1’s XIAP and XAF1 by Western blot analysis. In the proapoptotic effect was further characterized using absence of apoptotic stimuli, XAF1 protein was 253J and its subline (253J/p53–175D)expressing a found in both the nucleus and the cytoplasm dominant negative mutant form of p53 (175D), which and endogenous XIAP protein was observed ablates apoptotic function of wild-type p53 (Ryan and predominantly in the cytoplasm in both vector- and Vousden, 1998). Flow cytometric analysis of the sub-G1 XAF1-transfected cells (Figure 6a). As predicted, phase cells demonstrated that XAF1 expression strongly etoposide treatment resulted in a significantly higher enhances apoptosis of parental cells, but its effect is apoptosis in XAF1-transfected cells compared to vector- markedly abrogated in the 253J/p53–175D cells. On this transfected cells (Figure 6b). However, XAF1-expres- basis, we tested the growth suppression activity of sing 253J cells undergoing apoptosis exhibited no XAF1 using colony formation assay. 253J cells were detectable nuclear translocation of XIAP up to 72 h stably transfected with expression plasmids encoding after etoposide treatment, indicating that cytosolic sense or antisense XAF1, and XAF1 expression in the XIAP proteins are not sequestered to the nucleus by transfectants was verified by RT–PCR and immuno- XAF1 expression, which is sufficient to stimulate bloting using anti-XAF1 antibody (data not shown). As apoptosis induction. Similar results were obtained for seen in Figure 7e, introduction of sense XAF1 resulted the XIAP localization from the XAF1-restored LNCaP in approximately threefold decrease in the number of prostate cancer cells (data not shown). Additionally, neomycin-resistant colonies, and the XAF1-mediated etoposide-induced apoptosis was markedly activated by inhibition of colony formation was significantly im- XAF1 transfection in a transfection dose-dependent peded by co-transfection of a dominant negative mutant manner, but mRNA and protein levels of XIAP in these form of p53 (175D).

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5816

Figure 6 Effect of XAF1 expression on cellular localization and expression of XIAP. (a)Fractionation assay for XAF1 effect on cellular distribution of XIAP protein in cells undergoing apoptosis. XAF1- or empty vector-transfected 253J cells were treated with etoposide. Nuclear and cytoplasmic extracts of the cells were prepared at the indicated time point and separated by 10% SDS–PAGE. Endogenous XIAP protein was detected by immunoblotting assay using an anti-XIAP antibody. U1 SnRNP 70 and b-tubulin were used as a control for purity of nuclear and cytoplasmic extracts, respectively (***Po0.001; **Po0.01). (b)Elevated tumor cell response to apoptotic stimuli by XAF1 expression. 253J cells transfected with Flag-XAF1 expression (3 mg)or empty vector (3 mg)were treated with etoposide (10 mM)for 24–72 h. Number of apoptotic cell was determined using TUNEL assay. Data represent means of triplicate assays. Bars, s.d. (c)No visible effect of XAF1 on mRNA and protein expression level of XIAP during apoptosis. 253J cells transfected with increasing doses of XAF1 expression vectors were treated with etoposide (20 mM)for 48 h and mRNA and protein levels of XIAP and p53 were determined by semiquantitative RT–PCR and Western blot assay.

Discussion inactivation, based on the microsatellite analysis of the NCI 60 cell line panel showing a decreased hetero- The 17p13 region, where the XAF1 gene is located, zygosity within the XAF1 region at 17p13.2 (Fong et al., undergoes frequent allelic losses in a variety of human 2000). XAF1 is located at approximately 3 cM telomeric malignancies including bladder cancer, and the tumor- of the p53 tumor suppressor, which is frequently specific loss or downregulation of XAF1 expression associated with LOH, underscoring the importance of suggests a possible role for XAF1 in the suppression of resolving the basis for the XAF1 downregulation in malignancy (Fong et al., 2000; Liston et al., 2001; Steidl cancer cells. However, our previous study showed that et al., 2002). However, the mechanism by which XAF1 is LOH at the centromeric region of the XAF1 locus is downregulated in human cancers has been poorly rarely extends into the XAF1 gene in human gastric characterized. Allelic loss of the XAF1 gene has been cancers (Byun et al., 2003). Moreover, tumors with initially predicted as a possible mechanism for XAF1 LOH at telomeric markers of the p53 gene exhibited low

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5817

Figure 7 XAF1 upregulation of p53 protein stability. (a)Induction of p53 and its target gene expression by XAF1. 253J bladder cancer cells were transfected with increasing does of Flag-XAF1 expression vectors and treated with etoposide (20 mM)for 48 h. Protein level of p53 was analysed by Western blot assay and mRNA expression level of p53 target p21Waf1, PUMA and NOXA were determined by semi-quantitative RT–PCR. (b)Elevated and sustained induction of p53 by XAF1. XAF1- or vector-transfected 253J cells were pulse-treated with etoposide (50 mM)for 6 h and p53 protein level was determined at the indicated times after treatment. (c)Enhancement of p53 protein stability by XAF1. The cells were pretreated with CHX (10 m M)for 1 h before etoposide treatment (20 mM)and p53 protein level was examined 12 h after etoposide treatment. ( d)Requirement of functional p53 for the apoptotic action of XAF1. PC-3 (p53-deficient), 253J (wild-type p53), or 253J/p53–175D cells were transfected separately or simultaneously with expression vectors encoding XAF1 or wild-type p53 and its effect on apoptosis was determined using TUNEL assay (PC-3)or flow cytometric analysis of sub-G1 fraction (253J). Data represent means of triplicate assays (Bars, s.d.; ***Po0.001; **Po0.01; *Po0.05). (e)XAF1 inhibition of tumor cell growth and its dependency on functional p53. 253J cells (5 Â 104)transfected the indicated transgene were seeded in soft agar and maintained in the presence of neomycin for 4 weeks. Neomycin-resistant colonies were stained with crystal violet. Assays were performed in triplicate and average number of colonies and s.d. were calculated (***Po0.001).

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5818 genomic levels of p53 while none of these tumors show blocked by anti-TRAIL antibody whereas sensitivity to detectable reduction of XAF1 gene level, supporting that IFN-a in IFN-resistant bladder cancer cells was LOH at the 17p13 region is associated with allelic increased by treatment with a TRAIL-sensitizing agent deletion of p53 but not of XAF1. Consistent with this, all such as Bortezomib. It was also demonstrated that urogenital cancer cell lines and primary tumors tested in patients who responded to BCG therapy have signifi- this study have normal level of the XAF1 gene, while a cantly higher urine TRAIL levels compared with substantial fraction of the same set of tumor specimens nonresponders and TRAIL plays a crucial role in shows significantly reduced genomic level of p53, BCG-induced antitumor effects (Ludwig et al., 2004). indicating that genomic deletion of XAF1 might be These findings thus suggest that IFN-induced apoptosis infrequent in human urogenital cancers. in bladder cancer cells involves autocrine TRAIL It has been well documented that hypermethylation in production and overcoming TRAIL resistance may be CpG-rich promoter or transcription regulatory region is very effective in restoring IFN sensitivity in human strongly associated with transcriptional silencing and a bladder tumors (Papageorgiou et al., 2004). In this critical event leading to the epigenetic inactivation of context, it could be suspected that bladder cancers with tumor suppressor genes in human cancers. In this study, XAF1 inactivation might be more resistant to BCG or we found that XAF1 expression is reactivated or IFN combined immunotherapy than cancers with upregulated in no or low expressor tumor cells following normal XAF1 expression, and restoration of functional 5-Aza-dC treatment and aberrant hypermethylation at XAF1 could be effective in overcoming TRAIL 14 CpG sites located within the 50 proximal region resistance by sensitization of tumor cells to TRAIL- (nucleotides À23 to À692)of the promoter is tightly induced apoptosis. Therefore, it will be valuable to associated with downregulation of XAF1 expression in examine that expression status of XAF1 could be a both cancer cell lines and primary tumors. Thus, this clinically useful marker for cancer treatment including suggests that hypermethylation of the 50 proximal region IFN and BCG therapy. might be critical for the transcriptional silencing of XAF1 was originally identified as a nuclear protein XAF1 in urogenital cancers and possibly other human that could inactivate anticaspase function of XIAP via malignancies. sequestering XIAP protein to the nucleus (Liston et al., Very recently, XAF1 was identified as a novel 2001). However, we could not detect the nuclear interferon (IFN)-stimulated gene that contributes to translocation of XIAP protein in tumor cells undergoing IFN-dependent sensitization of cells to TRAIL-induced apoptosis by XAF1 overexpression. XIAP protein was apoptosis (Leaman et al., 2002). XAF1 mRNA was predominantly observed in the cytoplasm and its level upregulated by IFN-a2 and IFN-b and high levels of was decreased in apoptosis-undergoing cells after etopo- XAF1 protein were induced predominantly in cell lines side treatment. Additionally, restoration of XAF1 sensitive to the proapoptotic effects of IFN-b. It was expression in XAF1-deficient cells markedly accelerated also found that XAF1-expressing melanoma cells are apoptosis induction by etoposide, but no visible effect sensitive to TRAIL-induced apoptosis compared with on XIAP protein level was identified in the cells. This nonexpressing cells and the degree of sensitization is observation raises the possibility that the apoptosis- correlated with the level of XAF1 expressed. Intrigu- sensitizing function of XAF1 could be exerted, in part, ingly, we found that epigenetic inactivation of XAF1 is through the XIAP-independent pathway. In this con- significantly high in muscle-invasive bladder tumors text, our finding that XAF1 activates p53 protein compared with superficial tumors and more frequent in accumulation lends support to the notion that XAF1’s high-grade tumors than low-grade tumors, suggesting its proapoptotic effect may not be solely dependent on its contribution to the malignant progression of human XIAP-interfering activity. Furthermore, we found that bladder tumors. Instillation of Bacillus Calmette-Guerin XAF1 enhances the protein stability of p53, thus leading (BCG)is one of the most successful immunotherapies to a prolonged activation of p53 and its target gene for superficial bladder cancer and carcinoma in situ. expression under stress condition, and the proapoptotic BCG seems to induce tumor regression by producing a effect of XAF1 was dramatically increased in the proinflammatory Th1 cytokine response, including presence of functional p53. It is therefore conceivable IFN-g,TNF-a and TRAIL (Jackson et al., 1995). IFNs that inactivation of XAF1 may decrease the apoptotic induce apoptosis and/or growth arrest of bladder cancer sensitivity of tumor cells by the attenuated p53 response cells and combination of IFN with BCG results in an to various apoptotic stimuli. This is in line with our additive or synergistic effect against clinical bladder result that XAF1-mediated suppression of colony- cancers (Luo et al., 1999; Benedict et al., 2004). forming ability of cancer cells is substantially impeded However, some bladder cancer cells are resistant against by blockade of p53 function. Thus, functional p53 might to BCG- or IFN-induced tumor regression, leading to be critical for the apoptosis-sensitizing action of XAF1 the conjecture that defective response to growth although the p53-independent proapoptotic role for suppression effect of BCG and IFNs may give the XAF1 cannot be excluded. In fact, the existence of p53- tumors a selective advantage and abet escape from independent function of XAF1 was suggested by our T-cell antitumor response. Recently, IFN-a was found data showing that the apoptotic sensitivity of p53- to stimulate marked increases in TRAIL expression and deficient cells is reduced by knockdown of XAF1 induce early activation of caspase-8 (Papageorgiou expression using siRNA. Taken together, these findings et al., 2004). IFN-a-induced apoptosis was significantly support that XAF1 might be a novel effector of p53 in

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5819 signaling apoptosis and tumor suppression function of PCR was performed for 30–34 cycles at 951C (1 min), 58–621C XAF1 might stem partially from its capability to (0.5 min)and 72 1C (1 min)in 1.5 m M MgCl2-containing enhance the p53 apoptosis-signaling pathway. reaction buffer (PCR buffer II, Perkin Elmer). For quantita- In conclusion, the data presented here clearly tive genomic PCR, genomic DNA was extracted from the same demonstrate that XAF1 undergoes epigenetic silencing cells of the tissues or cell lines from the DNA phase after RNA was extracted. Two-hundred ng of genomic DNA were used in a considerable proportion of human urogenital for amplification of the exon 6 region of XAF1, exon 8 of p53, malignancies. An association of abnormal XAF1 ex- and intron 5 of GAPDH using intron-specific primers as pression with advanced tumor stage and higher grade of previously described (Chi et al., 1999). RT- and genomic PCR bladder tumor further supports its implication in the products (10 ml)were resolved on 2% agarose gels. Quantita- malignant progression of human tumors. Additionally, tion was achieved by densitometric scanning of the ethidium our study raises the possibility that loss or abnormal bromide-stained gels. Absolute area integrations of the curves dowwnregulation of XAF1 expression might render representing each specimen were then compared after adjust- tumor cells a survival advantage by attenuating the ment for GAPDH. Integration and analysis was performed apoptotic response of p53 to various stress conditions. using Molecular Analyst software program (Bio-Rad, Her- It will be therefore valuable to explore the possible cules, CA, USA). Quantitative PCR was repeated at least three times for each specimen and the mean was obtained. Gene application of XAF1 as a molecular marker for expression data was analysed by w2 test and Student’s t-test. detection and treatment of human urogenital malig- Po0.05 was considered statistically significant. nancies. Immunoblotting assay Cells were lysed in a lysis buffer containing 20 mM Tris (pH Materials and methods 7.4), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium phosphate, 1 mM b-glycerolphosphate, 1 mM Primary tumor tissues and cancer cell lines Na3VO4,1mg/ml leupeptin, and 1 mM PMSF. The cell lysate A total of 95 primary tumor specimens, including 55 TCC and was clarified by centrifugation and 20 mg of total protein were corresponding normal bladder mucosa, 20 RCC and adjacent supplemented with Laemmli buffer and loaded on a 10% noncancerous kidney tissues, and 20 prostate carcinomas were sodium dodecyl suflate–polyacrylamide gel (SDS–PAGE)for obtained by surgical resection in the Kyung Hee University electrophoresis. Western analyses were performed using Medical Center (Seoul, Korea). Ten benign prostatic hyper- antibodies specific for XIAP (BD Biosciences, San Diego, plasia tissues were also obtained from patients with no CA, USA), p53 (Santa Cruz Biotechnology, CA, USA), evidence of cancer. Tissue specimens were snap-frozen b-tubulin (Sigma, St Louis, MO, USA), and U1 SnRNP 70 immediately in liquid N2 and stored at À801C until used. (Santa Cruz Biotechnology, CA, USA). Antibody binding was Signed informed consent was obtained from each patient. Bits detected by enhanced chemiluminescence (Amersham Bio- of primary tumors and adjacent portions of each tumor were sciences, Piscataway, NJ, USA)using a secondary antibody fixed and used for hematoxylin and eosin staining for conjugated to horseradish peroxidase. For stripping, the blots histopathological evaluation. Tumor specimens composed of were incubated in a stripping buffer (0.2 M glycine (pH 2.2), at least 70% carcinoma cells and adjacent tissues found not to 0.1% SDS, 1% Tween-20)at room temperature for 60 min. contain tumor cells were chosen for molecular analysis. Six human bladder cancer cell lines (T24, J82, HT1197, HT1376, Loss of heterozygosity analysis 253J and 253J-BV)and three human prostate cancer cell lines LOH was determined using three polymorphic CA markers (LNCaP, DU145 and PC-3)were obtained from American (D17S796, D17S1832 and D17S1828)localized at Type Culture Collection (Rockville, MD, USA)or Korea Cell 17p13.1–13.2. PCR amplification was performed on each Line Bank (Seoul National University, Seoul, Korea). Total tumor and normal DNA sample pair obtained from 40 cellular RNA and genomic DNA specimens extracted from patients using primers as previously described (Byun et al., 15 human RCC cell lines (UOK101, UOK105, UOK107, 2003). PCR products (10 ml)were electrophoresed on non- UOK108, UOK109, UOK110, UOK112, UOK115, UOK121, denaturing 6% polyacrylamide gels. Signal intensity of UOK122, UOK123, UOK124, UOK130, ACHN and WWC1) fragments and the relative ratio of both tumor and normal were provided by Dr deVere White (University of California at allele intensities were determined by scanning desitometry. As Davis, CA, USA). certain numbers of noncancerous cells might be present in tumor tissues, LOH was assigned when the intensity ratio of Semi-quantitative RT- and genomic PCR analysis the two tumor alleles differed by at least 50% from that The strategy for our semi-quantitative RT–PCR was pre- observed on its corresponding normal DNA. The same PCR viously described (Chi et al., 1994; Tricoli et al., 1996). Briefly, products were also subjected to nonisotopic SSCP analysis for 1 mg of RNA was converted to cDNA by reverse transcription verification of LOH. using random hexamer primers and MoMuLV reverse transcriptase (Life Technologies Inc., Gaithersburg, MD, Nonisotopic RT–PCR–SSCP analysis USA). For quantitative evaluation of XAF1 expression levels To detect possible sequence alterations in XAF1, we performed by RT–PCR, we initially performed the PCR reaction over a nonisotopic RT–PCR–SSCP analysis. The XAF1 transcript range of cycles (20–40 cycles)using serially diluted cDNA, and was amplified with seven sets of primers that were designed to 1:4 diluted cDNA (12.5 ng/50 ml PCR reaction)undergoing cover the entire coding region of the gene. Sequences of the 24–36 cycles was observed to be within the logarithmic phase primers used will be obtained upon request. The PCR products of amplification and yielded reproducible results with primers of over 300 bp in lengths were digested with endonuclease(s)to all primers used for XAF1, Smac/DIABLO, HtrA2, XIAP, p53, increase the sensitivity of SSCP analysis. RT–PCR–SSCP p21Waf1, PUMA, NOXA and an endogenous expression analysis of p53 transcripts was performed as described standard gene GAPDH (Chi et al., 1998; Byun et al., 2003). previously (Chi et al., 1994). PCR products (10 ml)mixed with

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5820 10 ml of 0.5 N NaOH, 10 mM EDTA, and 15 ml of denaturing Institute for Cancer Research, Glagsow, UK). Small interfer- loading buffer (95% formamide, 20 mM EDTA, 0.05% ing RNA (siRNA)duplex against XAF1 (50-ATGTTGTCCA bromophenol blue and 0.05% xylene cyanol). After heating GACTCAGAG-30)and control siRNA duplex served as at 951C for 5 min, samples were loaded in wells precooled to negative control were synthesized by Dharmacon Research 41C and run using 8% nondenaturating acrylamide gels (Lafayette, CO, USA). For transfection, 1 Â 105 cells were containing 10% glycerol at 4–81C and 18–221C. plated on 60-mm-diameter dishes 24 h before transfection. The cells were incubated with a siRNA-Oligofectamine mixture at 5-Aza-dC treatment 371C for 4 h, followed by addition of fresh medium containing To assess reactivation of XAF1 expression, cell lines showing 10% fetal bovine serum. no or low levels of XAF1 transcript were plated in six-well tissue plates 48 h before treatment. 5-Aza-dC (Sigma, St Louis, Cell proliferation and DNA synthesis assays MO, USA)was added to the fresh medium at concentration of Cells were seeded in six-well plate at the density of 2 Â 104 cells 2.5–20 mM in duplicate and cells were harvested after 5 days of per cell in duplicate and were maintained in the presence of treatment (Lee et al., 2001). Using cDNA obtained from 10% FBS. The following day, cells were transfected with cancer cell lines, mRNA transcript of XAF1 before and after expression vector or siRNA as described above. Cell numbers 5-Aza-dC treatment were analysed by RT–PCR. were counted using a hemocytometer for 4 days at 24-h intervals. DNA synthesis was measured by determining the Methylation-specific PCR analysis incorporation of [3H]thymidine. After 44 h transfection, cells 3 Genomic DNA (1 mg)in a volume of 50 ml was denatured by were pulse-labeled for 4 h with 1 mCi/ml of [ H]thymidine NaOH (final concentration 0.3 M). Thirty micorliters of 10 mM (Amersham Pharmacia Biotech, NJ, USA)and the radio- hydroquinone and 520 mlof3M sodium bisulfite (pH 5.0)were activity incorporated into trichloroacetic acid-precipitable added and incubated at 551C for 16–20 h. DNA samples were materials was counted by a liquid scintillation counter. For purified using Wizard DNA clean-up system (Promega Corp., flow cytometry analysis, cells were harvested 48 h after Madison, WI, USA), again treated with NaOH at 371C for transfection and fixed with 70% ethanol and resuspended in 15 min, precipitated with ethanol, and resuspended in distilled 1 ml of PBS containing 50 mg/ml RNase and 50 mg/ml water. Fifty nanograms of bisulfite-modified or unmodified propidium iodide (Sigma, St Louis, MO, USA). The assay DNA were subjected to PCR amplification. PCR was was performed on a FACScan flow cytometer (Becton performed separately with a methylation-specififc primer set, Dickinson, San Jose, CA, USA), and the cell cycle profile MS (sense; 50-TTTATTTTATTGGTAGACGTTACG-30)and was analysed using MultiCycle software (Phoenix Flow MAS (antisense; 50-ATAACTCCTAAACTTCCAAACG-30) Systems, San Diego, CA, USA). and a nonmethylation-specific primer set, US (sense; 50- TTTGGAAAAGGGATGGAGATTTAGATG-30)and UAS TUNEL assay (antisense; 50-ACAAACTTTCAATTAAATTTCA-30)for 38 TUNEL assay was performed to evaluate apoptotic effect of cycles at 951C for 1 min, at 60–631C for 1 min, and 721C for XAF1. Briefly, 253J and HT1376 cells were transfected with 1 min. Ten micoliters of PCR product were visualized on a 2% XAF1-expressing vector and siRNA for XAF1, respectively, agarose gel. and the cells were treated with etoposide (25 mM)or 5-FU (25 mM)for 24–72 h. The cells were then fixed with 4% Bisulfite DNA sequencing analysis for the XAF1 promoter paraformaldehyde in PBS, and the buffer containing 3% Fifty nanograms of bisulfite-modified DNA were subjected to bovine serum albumin and 0.1% Triton X-100 was added and PCR amplification of the XAF1 promoter region using primers incubated for 15 min at 41C. The cells were labeled by the MS7 (sense; 50-AATTTTGTTATTTTTTTTAGAG-30)and terminal deoxynucleotidyltransferase-mediated dUTP-biotin MS2 (antisense; 50-CATATTCTACTCTCTACAAAC-30)for nick end labeling (TUNEL)reaction mixture using the In Situ nucleotides À700 to À20. The PCR products were cloned into Cell Death Detection kit (Roche, Mannheim, Germany). pCRII vectors (Invitrogen, Carlsbad, CA, USA)and 10 clones TUNEL signals of cells were visualized directly under of each specimen were sequenced by automated fluorescence- microscopy. based DNA sequencing to determine the methylation status. Preparation of cytosolic and nuclear extracts Construction of expression plasmids, siRNA and transfection Nuclear and cytosolic extracts from XAF1-or vector-trans- Expression vectors encoding sense or antisense XAF1 were fected cells were prepared for immunoblotting assay for XAF1 constructed by a PCR based approach using primers XAF1- and XIAP at the indicated time points. Cells were harvested in FLAG-14 (sense; 50-ATGGATTACAAGGATGACGACGA 1 ml of phosphate-buffered saline with a rubber policeman. TAAGATGGAAGGAGACTTCTCGGT-30)and XAF1–15 Samples were centrifuged for 1 min at 6000 g (41C), and the (antisense; 50-GTTAAAAGTGAAATCTTTTGAATT-30). resulting cell pellets were resuspended in 100 ml of low salt The PCR products were first ligated to the pCR2.1-TOPO buffer (20 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM NaVO4, vector (Invitrogen, Carlsbad, CA, USA)and then subcloned 1mM EDTA, 1 mM EGTA, 0.2% Nonidet P-40, 10% glycerol, into the pcDNA3.1-FLAG vector (Invitrogen, Carlsbad, CA, supplemented with a set of proteinase inhibitors, Completet). USA). Approximately, 1 Â 105 cells were plated per six-well After 10 min of incubation on ice, the samples were centrifuged plate in media containing 10% fetal bovine serum to give at 13 000 g for 2 min (41C), and the supernatants (cytosolic 50–60% confluence, and the transfection of constructs was extracts)were immediately frozen in a dry ice/ethanol bath. performed using FuGene6 (Roche, Mannheim, Germany) Pelleted nuclei were resuspended in 60 ml of high salt buffer according to the manufacturer’s protocol. The transfection (20 mM HEPES, pH 7.9, 420 mM NaCl, 10 mM KCl, 0.1 mM efficiency was monitored using a fluorescence microscopy for NaVO4,1mM EDTA, 1 mM EGTA, 20% glycerol, supple- FLAG or CAT assay (Roche, Mannheim, Germany). Expres- mented with Completet), and nuclear proteins were extracted sion vector encoding human mutant p53 (R175D), which give by shaking on ice for 30 min. Samples were centrifuged at rise to a protein with impaired apoptotic and cell cycle arrest 13 000 g for 10 min (41C), and the supernatants were taken as functions, was kindly provided by Dr KH Vousden (Beatson nuclear extracts.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5821 Colony formation assay difference. The w2 test was used to determine the statistical 253J human bladder carcinoma cells were transfected with significance of expression and methylation levels between expression vectors encoding sense XAF1, antisense XAF1, tumor and normal tissues. A P-value of o0.05 was considered wild-type 53 or mutant p53 (R175D)using FuGene6 (Roche, significant. Mannheim, Germany). Transfected cells (5 Â 104)were main- tained grown in soft agar in the presence of G418 for 4 weeks. Colonies were stained with 0.5% crystal violet in 20% ethanol. Acknowledgements

Statistical analysis This work was supported in part by grants from Korea Science The results of DNA replication, apoptosis, colony forming and Engineering Foundation (R02-2003-000-10031-0), Korea assays were expressed as mean7s.d. A Student’s t-test was Research Foundation (2003-070-C00031), and the National used to determine the statistical significance of the Cancer Center (0420230), Korea.

References

Ambrosini G, Adida C, Altieri DC. (1997). A novel anti- Holcik M, Gibson H, Korneluk RG. (2001). Apoptotic brake apoptosis gene, survivin, expressed in cancer and lymphoma. and promising therapeutic target. Apoptosis 6: 253–261. Nature Med 3: 917–921. Jackson AM, Alexandroff AB, Kelly RW, Skibinska A, Asselin E, Mills GB, Tsang BK. (2001). XIAP regulates Akt Esuvaranathan K, Prescott S et al. (1995). Changes in activity and caspase-3-dependent cleavage during cisplatin- urinary cytokines and soluble intercellular adhesion mole- induced apoptosis in human ovarian epithelial cancer cells. cule-1 (ICAM-1)in bladder cancer patients after bacillus Cancer Res 61: 1862–1868. Calmette-Guerin (BCG)immunotherapy. Clin Exp Immunol Benedict WF, Tao Z, Kim CS, Zhang X, Zhou JH, Adam L 99: 369–375. et al. (2004). Intravesical Ad-IFNa causes marked regression Leaman DW, Chawla-Sarkar M, Vyas K, Reheman M, Tamai of human bladder cancer growing orthotopically in nude K, Toji S et al. (2002). Identification of X-linked inhibitor of mice and overcomes resistance to IFN-a protein. Mol Ther apoptosis-associated factor-1 as an interferon-stimulated 10: 525–532. gene that augments TRAIL Apo2L-induced apoptosis. Bilim V, Kasahara T, Hara N, Takahashi K, Tomita Y. J Biol Chem 277: 28504–28511. (2003). Role of XIAP in the malignant phenotype of Lee MG, Kim HY, Byun DS, Lee SJ, Lee CH, Kim JI et al. transitional cell cancer (TCC)and therapeutic activity of (2001). Frequent epigenetic inactivation of RASSF1A in XIAP antisense oligonucleotides against multidrug-resistant human bladder carcinoma. Cancer Res 61: 6688–6692. TCC in vitro. Int J Cancer 103: 29–37. Li J, Kim JM, Liston P, Li M, Miyazaki T, Mackenzie AE Byun DS, Cho K, Ryu BK, Lee MG, Kang MJ, Kim HR et al. et al. (1998). Expression of inhibitor of apoptosis proteins (2003). Hypermethylation of XIAP-associated factor 1, a (IAPs)in rat granulosa cells during ovarian follicular putative tumor suppressor gene from the 17p13.2 locus, in development and atresia. Endocrinology 139: 1321–1328. human gastric adenocarcinomas. Cancer Res 63: 7068–7075. Liston P, Fong WG, Kelly NL, Toji S, Miyazaki T, Conte D Chi SG, Chang SG, Lee SJ, Lee CH, Kim JI, Park JH. (1999). et al. (2001). Identification of XAF1 as an antagonist of Elevated and biallelic expression of p73 is associated with XIAP anti-Caspase activity. Nat Cell Biol 3: 128–133. progression of human bladder cancer. Cancer Res 59: Liston P, Roy N, Tamai K, Lefebvre C, Baird S, Cherton- 2791–2793. Horvat G et al. (1996). Suppression of apoptosis in Chi SG, deVere White RW, Meyers FJ, Siders D, Lee F, mammalian cells by NAIP and a related family of IAP Gumerlock PH. (1994). p53 in prostate: frequent expressed genes. Nature 379: 349–353. transition mutations. J Natl Cancer Inst 86: 926–933. Ludwig AT, Moore JM, Luo Y, Chen X, Saltsgaver NA, Chi SG, Kim HJ, Park BJ, Min HJ, Park JH, Kim YW et al. O’Donnell MA et al. (2004). Tumor necrosis factor-related (1998). Mutational abrogation of the PTEN/MMAC1 gene apoptosis-inducing ligand: a novel mechanism for Bacillus in gastrointestinal polyps in patients with Cowden disease. Calmette-Guerin-induced antitumor activity. Cancer Res 64: Gastroenterology 115: 1084–1089. 3386–3390. Deveraux QL, Reed JC. (1999). IAP family proteins-suppres- Luo Y, Chen X, Downs TM, DeWolf WC, O’Donnell MA. sors of apoptosis. Genes Dev 13: 239–252. (1999). IFN-a2B enhances Th1 cytokine responses in Deveraux QL, Takahashi R, Salvesen GS, Reed JC. (1997). bladder cancer patients receiving Mycobacterium bovis X-linked IAP is a direct inhibitor of cell-death proteases. bacillus Calmette-Guerin immunotherapy. J Immunol 162: Nature 388: 300–304. 2399–2405. Du C, Fang M, Li Y, Li L, Wang X. (2000). Smac, a Ng KCP, Campos EI, Martinka M, Li G. (2004). XAF1 mitochondrial protein that promotes cytochrome c-depen- expression is significantly reduced in human melanoma. dent caspase activation by eliminating IAP inhibition. Cell J Invest Dermatol 123: 1127–1134. 102: 33–42. Papageorgiou A, Lashinger L, Millikan R, Grossman HB, Duckett CS, Nava VE, Gedrich RW, Clem RJ, Van Dongen Benedict W, Dinney CP et al. (2004). Role of tumor necrosis JL, Gilfillan MC et al. (1996). A conserved family of cellular factor-related apoptosis-inducing ligand in interferon-in- genes related to the baculovirus iap gene and encoding duced apoptosis in human bladder cancer cells. Cancer Res apoptosis inhibitors. EMBO J 15: 2685–2689. 64: 8973–8979. Fong WG, Liston P, Rajcan-Separovic E, St Jean M, Craig C, Perrelet D, Perrin FE, Liston P, Korneluk RG, MacKenzie A, Korneluk RG. (2000). Expression and genetic analysis of Ferrer-Alcon M et al. (2004). Motoneuron resistance to XIAP-associated factor 1 (XAF1)in cancer cell lines. apoptotic cell death in vivo correlates with the ratio between Genomics 70: 113–122. X-linked inhibitor of apoptosis proteins (XIAPs)and Green DR. (2000). Apoptotic pathways: paper wraps stone its inhibitor, XIAP-associated factor 1. J Neurosci 24: blunts scissors. Cell 102: 1–4. 3777–3785.

Oncogene Epigenetic inactivation of XAF1 in urogenital tumors M-G Lee et al 5822 Roy N, Deveraux QL, Takahashi R, Salvesen GS, Reed JC. of the inhibitor-of-apoptosis protein XIAP. Nature 401: (1997). The c-IAP-1 and c-IAP-2 proteins are direct 818–822. inhibitors of specific caspases. EMBO J 16: 6914–6925. Suzuki Y, Imai Y, Nakayama H, Takahashi K, Takio K, Ryan KM, Vousden KH. (1998). Characterization of structur- Takahashi R. (2001). A serine protease, HtrA2, is released al p53 mutants which show selective defects in apoptosis but from the mitochondria and interacts with XIAP, inducing not cell cycle arrest. Mol Cell Biol 18: 3692–3698. cell death. Mol Cell 8: 613–621. Sasaki H, Sheng Y, Kotsuji F, Tsang BK. (2000). Down- Tamm I, Kornblau SM, Segall H, Krajewski S, Welsh K, regulation of X-linked inhibitor of apoptosis protein induces Kitada S et al. (2001). Expression and prognostic signifi- apoptosis in chemoresistant human ovarian cancer cells. cance of IAP-family genes in human cancers and myeloid Cancer Res 60: 5659–5666. leukemias. Clin Cancer Res 6: 1796–1803. Siegelin M, Touzani O, Toutain J, Liston P, Rami A. (2005). Thompson CB. (1995). Apoptosis in the pathogenesis and Induction and redistribution of XAF1, a new antagonist of treatment of disease. Science 267: 1456–1462. XIAP in the rat brain after transient focal ischemia. Tricoli JV, Gumerlock PH, Yao JL, Chi SG, D’Souza SA, Neurobiol Dis 20: 509–518. Nestok BR et al. (1996). Alterations of the retinoblastoma Steidl C, Simon R, Burger H, Brinkschmidt C, Hertle L, gene in human prostate adenocarcinoma. Genes Chromo- Bocker W et al. (2002). Patterns of chromosomal aberrations somes & Cancer 15: 108–114. in urinary bladder tumours and adjacent urothelium. J Verhagen AM, Ekert PG, Pakusch M, Silke J, Connolly LM, Pathol 198: 115–120. Reid GE et al. (2000). Identification of DIABLO, a Sun C, Cai M, Gunasekera AH, Meadows RP, Wang H, mammalian protein that promotes apoptosis by binding to Chen J et al. (1999). NMR structure and mutagenesis and antagonizing IAP proteins. Cell 102: 43–53.

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

Oncogene