Expression of the tumour suppressor gene CADM1 is ... - Nature

Oncogene (2008) 27, 3329–3338

& 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Expression of the tumour suppressor gene CADM1 is associated with favourable outcome and inhibits cell survival in neuroblastoma S Nowacki1, M Skowron1, A Oberthuer1, A Fagin2, H Voth1, B Brors3, F Westermann4, A Eggert2, B Hero1, F Berthold1 and M Fischer1 1 Department of Pediatric Oncology and Hematology, Children’s Hospital, University of Cologne, Cologne, Germany; 2Department of Pediatric Oncology and Hematology, University Hospital of Essen, Germany; 3Department of Theoretical Bioinformatics (B080), German Cancer Research Center, Heidelberg, Germany and 4Department of Tumour Genetics (B030), German Cancer Research Center, Heidelberg, Germany

Cell adhesion molecule 1 (CADM1) is a putative tumour suppressor gene, which is downregulated in many solid tumours. In neuroblastoma, loss of CADM1 expression has recently been found in disseminated tumours with adverse outcome, prompting us to investigate its role in neuroblastoma tumour progression. Oligonucleotidemicroarray analysis of 251 neuroblastoma specimens demonstrated that CADM1 downregulation is associated with unfavourable prognostic markers like disseminated stage 4, age >18 months, MYCN amplification and chromosome 11q alterations (Po0.001 each). Furthermore, low CADM1 expression was significantly correlated with unfavourable gene expression-based classification (Po0.001) and adverse patient outcome (Po0.001). Bisulphite sequencing and genetic analysis of 18 primary neuroblastomas suggested that neither haploinsufficiency nor hypermethylation is regularly involved in CADM1 gene silencing in neuroblastoma, which is in contrast to results obtained in other malignancies. In addition, no mutations disrupting the CADM1 reading frame were found in 25 primary neuroblastomas. Over-expression of CADM1 in neuroblastoma cells resulted in significant reduction of proliferation, viability and colony formation in soft agar. Collectively, our results suggest that downregulation of CADM1 tumour suppressor gene expression is a critical event in neuroblastoma pathogenesis resulting in tumour progression and unfavourable patient outcome. Oncogene (2008) 27, 3329–3338; doi:10.1038/sj.onc.1210996; published online 17 December 2007 Keywords: CADM1; neuroblastoma; tumor suppressor; outcome assessment (health care); cell proliferation; disease progression

Correspondence: Dr M Fischer, Department of Pediatric Oncology and Hematology, Children’s Hospital, University of Cologne, Kerpener Street 62, Cologne 50924, Germany. E-mail: matthias.fi[email protected] Received 6 September 2007; revised 24 October 2007; accepted 21 November 2007; published online 17 December 2007

Introduction Neuroblastoma is a paediatric tumour arising from the developing sympathetic nervous system. The biological and clinical behaviour of this malignancy is remarkably heterogeneous ranging from aggressive progression to spontaneous regression, and several clinical and molecular parameters have been identified to be associated with either phenotype. Although patients with metastatic disease (stage 4) and an age above 2 years in most cases have a poor outcome, spontaneous tumour involution is regularly observed in infants o1-year old with localized stages or with the special stage 4S, which is defined by the age of the patient (o1 year) and dissemination limited to bone marrow, liver and/or skin. It has been shown that characteristic cytogenetic aberrations are associated with these contrasting subtypes of neuroblastoma. Aggressive tumours of patients with poor survival often show either an amplification of the oncogene MYCN together with a 1p deletion, or a deletion of chromosomes 11q and 3p, whereas favourable tumours usually lack gross structural chromosomal aberrations. Moreover, the expression of various genes such as NTRK1 and FYN have been described to be indicative of neuroblastoma tumour behaviour (Nakagawara et al., 1993; Berwanger et al., 2002). Recently, prognostic gene expression signatures were shown to even more accurately predict the natural courses of this malignancy (Ohira et al., 2005; Oberthuer et al., 2006). Together, these data strongly suggest that spontaneously regressing and aggressive neuroblastoma represent two distinct subtypes of the disease. However, the exact molecular mechanisms underlying these contrasting phenotypes and the genes involved in spontaneous regression and tumour progression remain to be determined. To identify genes involved in spontaneous regression of neuroblastoma, we have previously generated gene expression profiles from spontaneously regressing stage 4S and fatal stage 4 disease using serial analysis of gene expression (SAGE). Differential expression of a number of transcripts was confirmed by quantitative real-time RT-PCR in these stages, including the putative tumour suppressor gene cell adhesion molecule 1 (CADM1),

CADM1 is a tumour suppressor gene in neuroblastoma S Nowacki et al

3330

which is also known as TSLC1, IGSF4, SynCAM and Necl-2 (Fischer et al., 2006). This gene maps to chromosome 11q23.2, a region frequently deleted in tumours such as malignant melanoma (Katoh, 2003, 2004), lung cancer (Gomyo et al., 1999) and neuroblastoma (Mertens et al., 1997), and encodes an immunoglobulin superfamily transmembrane molecule primarily involved in epithelial cell adhesion (Murakami, 2005). Loss of CADM1 expression has been described in many malignancies, including liver, pancreatic, breast, brain and prostate cancers, particularly in those showing invasion or metastasis (Kuramochi et al., 2001; Fukuhara et al., 2002; Jansen et al., 2002; Houshmandi et al., 2006; Heller et al., 2007). Promoter hypermethylation was reported to account for downregulation of CADM1 in several human cancers (Murakami, 2005; Ehrlich et al., 2006; Kikuchi et al., 2006; Worsham et al., 2006). Moreover, high expression levels of CADM1 from a recombinant adenovirus vector were shown to inhibit cell proliferation and to induce apoptosis in the non–small cell lung cancer cell line A549 (Mao et al., 2004). Together, these data suggested that loss of CADM1 expression might represent a critical event in tumour progression of unfavourable neuroblastoma. We therefore examined the role of CADM1 expression in neuroblastoma pathogenesis by analysing mRNA levels of CADM1 in a cohort of 251 primary tumours and their correlation with prognostic markers and patient survival. To address whether promoter hypermethylation might account for CADM1 gene silencing, the methylation status of this genomic region was determined in 18 primary neuroblastomas, and the CADM1 coding region was sequenced in 25 tumours to identify inactivating mutations,. Finally, CADM1 expression was restored in four neuroblastoma cell lines by adenoviral gene transfer to assess its functional role in regulating tumour growth of neuroblastoma. Results Low CADM1 transcript levels are correlated with markers of poor outcome in neuroblastoma Comparison of expression profiles of five stage 4S and three stage 4 neuroblastoma generated by SAGE revealed that transcript levels of the putative tumour suppressor gene CADM1 are substantially decreased in stage 4 tumours, which was confirmed in a cohort of 76 neuroblastoma samples (stage 4S, n ¼ 27; stage 4, n ¼ 49) using quantitative real-time RT-PCR (Fischer et al., 2006). To address the question whether downregulation of CADM1 is a marker of adverse outcome in general in this tumour, its expression levels were evaluated in a cohort of 251 neuroblastoma samples (stage 1, n ¼ 69; stage 2, n ¼ 44; stage 3, n ¼ 40; stage 4, n ¼ 67; stage 4S, n ¼ 31) reflecting the whole spectrum of the disease using the microarray technology (Oberthuer et al., 2006), and associations of CADM1 transcript levels with five prognostic markers were examined. Similar ranges of CADM1 transcript levels were observed for stage 1–3 Oncogene

tumours and stage 4S tumours, but both subgroups exhibited significantly higher expression levels in comparison to stage 4 (Po0.001; Figure 1a). CADM1 levels were substantially decreased in patients older than 18 months at diagnosis (Po0.001; Figure 1b) as well as in tumours with MYCN amplification (Po0.001; Figure 1c) and chromosome 11q aberrations (defined as deletion or imbalance; Po0.001; Figure 1d). Furthermore, low CADM1 transcript levels significantly correlated with unfavourable gene expression-based classification (Po0.001; Figure 1e) according to a classifier that we have previously defined using the prediction analysis for microarrays algorithm (PAM; Oberthuer et al., 2006). This predictive signature consists of 144 genes, but does not include CADM1. Taken together, these results clearly demonstrate that downregulation of CADM1 is significantly associated with unfavourable markers in neuroblastoma. Low CADM1 expression is associated with adverse outcome of neuroblastoma patients To determine whether CADM1 expression levels are related to differences in patients’ outcome, Kaplan– Meier estimates for event-free survival (EFS) and overall survival (OS) were calculated and compared by log-rank test. As expected from the strong correlation of CADM1 transcript levels with established prognostic markers, it turned out that low CADM1 expression was correlated with adverse clinical outcome (Figure 2). When using the 25th percentile of CADM1 mRNA levels as a cut-off for low expression (n ¼ 62), patients with intermediate (defined as >25th and o75th percentile, n ¼ 127) and high (defined as >75th percentile, n ¼ 62) CADM1 transcript levels were found to have a significantly better EFS as compared to patients with low expression (high and intermediate CADM1 levels, 5-year EFS 86±5 and 786±4%, respectively vs low CADM1 levels, 5-year EFS 366±7%; Po0.001; Figure 2a). Similar results with high statistical significance were observed for OS of these patients (high and intermediate CADM1 levels, 5-year OS 986±2 and 896±3%, respectively vs low CADM1 levels, 5-year OS 526±8%; Po0.001; Figure 2b). In addition, a trend towards a significantly differing EFS and a significant difference in OS was detected between patients with high and intermediate CADM1 expression levels (P ¼ 0.130 and 0.047, respectively). These results demonstrate that CADM1 transcript levels discriminate neuroblastoma patients with beneficial and adverse outcome. No evidence for hypermethylation of the CADM1 promoter region in neuroblastoma To investigate whether hypermethylation of the CADM1 promoter is involved in regulating CADM1 expression levels in neuroblastoma, eight CpG sites located in the promoter region that have been demonstrated to be methylated in several tumours (Fukami et al., 2003; Steenbergen et al., 2004; Kikuchi et al., 2006) were tested for methylation by bisulphite sequencing. Hypermethylation was defined as methylation of a


3331

Figure 1 Correlation of cell adhesion molecule 1 (CADM1) expression levels with prognostic markers. Association of (a) tumour stage, (b) age at diagnosis, (c) MYCN amplification status, (d) chromosome 11q status and (e) gene expression-based classification (PAM classifier) with relative CADM1 transcript levels in 251 neuroblastoma samples as determined by oligonucleotide microarray experiments. Boxes, median expression values (horizontal line) and 25th and 75th percentiles; whiskers, distances from the end of the box to the largest and smallest observed values that are o1.5 box lengths from either end of the box; open circles, outlying values; N, patient numbers.

specific CpG site in >50% of the clones sequenced. Tumour samples from 18 patients with various CADM1 expression levels were investigated (high IGFS4 expression as defined above, n ¼ 5; intermediate CADM1 expression, n ¼ 7; low CADM1 expression, n ¼ 6). In this cohort, the same methylation pattern was found throughout the whole group of patients (Figure 3; Supplementary data). In addition, the finding of low CADM1 expression levels in five samples without 11q alterations on the one hand, and intermediate expression levels in two samples with 11q deletions on the other (Figure 3) suggest that CADM1 expression is not directly affected by chromosome 11q aberrations. Together, these data suggest that neither global promoter hypermethylation nor biallelic inactivation by loss of heterozygosity and methylation nor haploinsufficiency is essential for repression of CADM1 in neuroblastoma.

No evidence for inactivating mutations in the coding region of the CADM1 gene in neuroblastoma To investigate whether inactivating mutations can be found in the coding sequence of CADM1 in neuroblastoma, cDNA sequence analysis was performed in 20 primary tumours with low expression and five primary tumours with high expression levels. Within the group of neuroblastoma with low CADM1 expression, a 6 bp inframe deletion in exon 1 was found in a single tumour, resulting in loss of Leu26 and Arg27, and a point mutation in exon 7 was observed in three tumours, resulting in an Asp285Glu replacement. All mutations were confirmed on the genomic level. As none of these mutations disrupts the reading frame of the transcript, and as aspartate and glutamate represent similarly structured hydrophilic acidic amino acids, it appears questionable whether these sequence alterations represent loss-of-function mutations. Oncogene


3332

Figure 2 Survival curves of patients with neuroblastoma according to expression of cell adhesion molecule 1 (CADM1) mRNA. The Kaplan–Meier curves show the probability of (a) event-free (EFS) and (b) overall survival (OS), respectively, in terms of the level of CADM1 expression as determined by microarray analysis. ‘CADM1 high’, ‘CADM1 intermediate’ or ‘CADM1 low’ indicate patients with CADM1 mRNA levels defined as high (>75th percentile), intermediate (>25th and o75th percentile) or low (o25th percentile), respectively.

Adenovirus-mediated expression of CADM1 in neuroblastoma cell lines To examine whether CADM1 functions as a tumour suppressor in neuroblastoma, CADM1 expression was reconstituted in neuroblastoma cell lines. For this purpose, we selected two human neuroblastoma cell lines with MYCN amplification (IMR-32 and Kelly) and two with normal MYCN status (SH-SY5Y and SH-EP). Cells were infected with either CADM1 or LacZ transducing adenovirus. Two days post-infection, saturation of infection was monitored using b-Gal staining, while no cytopathic effect was observed using Trypan blue staining (data not shown). Three days postinfection, northern and western blot analysis demonstrated that CADM1 expression was effectively restored in all of the cell lines, resulting in recombinant protein levels in range of physiological levels observed in patients (Figure 4). Peak expression of CADM1 mRNA was evident between days 2 and 3, with detectable levels by northern blot until day 10 post-infection (data not shown). CADM1 expression inhibits proliferation of neuroblastoma cells To investigate whether CADM1 is involved in the regulation of proliferation in human neuroblastoma cell lines, IMR-32, Kelly, SH-SY5Y and SH-EP cells were infected with adenovirus transducing either CADM1 or LacZ as a negative control, and cell proliferation was examined for up to 10 days using the XTT colorimetric assay. At day 10, cell proliferation was significantly decreased in all cell lines expressing CADM1 as compared to LacZ-infected controls (Figure 5a). In SH-EP cells, the most severe effect was detected with a Oncogene

reduction of proliferation to merely 20% at day 4 postinfection and to about 10% from day 7 on. Steadily decreasing cell proliferation was observed in IMR-32, Kelly and SH-SY5Y cells as well, even though slightly delayed as compared to SH-EP cells. Similar results were obtained for each cell line in two independent experiments, each of which was performed in triplicate. These findings indicate a critical role of CADM1 in regulating growth properties in human neuroblastoma cells in general, as not only a single cell line with certain clonal characteristics, but a panel of four different neuroblastoma cell lines showed a similar decline in proliferation following CADM1 over-expression. Decrease in cell viability in response to CADM1 over-expression Given the established role of CADM1 expression in inducing apoptosis when expressed transiently in A549 cells (Mao et al., 2004), we tested whether the antiproliferative effect of CADM1 observed in the XTT assay could be due to decreased cell viability by examining transgene-mediated changes in lactate dehydrogenase (LDH) release. Leakage of LDH into the culture supernatant inversely correlates with the number of viable cells (Racher et al., 1990). Starting 5 h postinfection, LDH released into the culture medium was measured for the following three days. Under these experimental conditions, a significantly increased LDH release in all four neuroblastoma cell lines was observed when comparing CADM1-transduced to LacZ-infected control cells (Figure 5b). CADM1 increased LDH release rates to about 20–60% above negative control levels throughout the measurement interval. These findings are consistent with results of the proliferation


3333

Figure 4 Physiological cell adhesion molecule 1 (CADM1) expression levels and re-expression of CADM1 in neuroblastoma cell lines IMR-32, Kelly, SH-SY5Y and SH-EP as determined by (a) northern and (b) western blot analysis. Cells were transduced with either adLacZ or adCADM1, and harvested 3 days postinfection. In northern blot analysis, only mRNA levels of the recombinant CADM1 transcript are shown, which is shorter than the endogenous transcript. Representative samples of patients with high (patient A) and low (patient B) CADM1 expression levels demonstrate that re-expression of CADM1 resulted in protein levels similar to physiological levels observed in patients with high CADM1 expression. In CADM1-transduced SH-EP cells, the bactin control was undetectable, which was consistent with northern blot analysis results (data not shown).

Figure 3 Cell adhesion molecule 1 (CADM1) expression, methylation status of the CADM1 promoter in a CpG island and 11q status in 18 neuroblastoma patients. (a) Long vertical bars indicate the eight CpG sites examined upstream of the predicted TATA box sequence. The open box and asterisk indicate exon 1 and the predicted TATA box sequence, respectively. ATG indicates the initial methionine codon. White and black circles represent unmethylated and methylated CpGs, respectively (detailed information of each sequenced clone are given in Supplementary Figure 1 and in the Supplementary sequencing files). Normalized log ratios of CADM1 expression levels were determined by oligonucleotidemicroarray analysis and defined as high (>75th percentile; n ¼ 62), intermediate (>25th and o75th percentiles; n ¼ 127) and low (o25th percentile, n ¼ 62). Status of chromosome 11q was defined as normal (n), imbalance (im) or deleted (del). Representative results of interphase fluorescence in situ hybridization (FISH) of (b) deletion and (c) retention of 11q23 alleles.

assay and demonstrate that a reduced number of viable cells might account for CADM1-induced decline in proliferation. CADM1 expression reduces the ability of colony forming in soft agar The results obtained from the analysis of proliferation and viability support the hypothesis of CADM1 as a

tumour suppressor gene in neuroblastoma. To further investigate the effect of CADM1 expression on tumorigenicity, we compared anchorage-independent growth rates of CADM1-transduced to LacZ-infected control cells. Cells were grown in soft agar for 7 days starting 2 hours post-infection. In three of four cell lines analysed (SH-EP, SH-SY5Y, Kelly), a significantly reduced colony-forming ability in soft agar (20–50%) was observed after 7 days, whereas the decline of 20% observed in IMR-32 cells was not significant (Figure 5c). Together, these results demonstrate that reconstitution of CADM1 expression attenuates the malignant phenotype of neuroblastoma cells in vitro.

Discussion The putative tumour suppressor gene CADM1 has been described to be inactivated in numerous solid human tumours including liver, pancreatic, breast, brain and prostate cancers (Kuramochi et al., 2001; Fukuhara et al., 2002; Jansen et al., 2002; Houshmandi et al., 2006; Heller et al., 2007), and there has been considerable interest in understanding the biological significance of CADM1 loss in tumourigenesis. Previous investigations have supported the hypothesis of CADM1 as a general tumour suppressor-mediating apoptosis (Mao et al., 2004), differentiation (Wakayama et al., 2003) and cell cycle regulation (Ito et al., 2003b). We recently found Oncogene


3334

Figure 5 Cell adhesion molecule 1 (CADM1)-induced changes in (a) cell proliferation (XTT assay) and (b) cell viability (LDH (lactate dehydrogenase)-release assay) and (c) soft agar colony formation in neuroblastoma cell lines IMR-32, Kelly, SH-SY5Y and SH-EP. Cells were transduced with either adLacZ or adCADM1, and adLacZ-transduced controls were set as 100%. The result of a representative experiment from two independent experiments with similar results is shown. Each value represents the mean of triplicate determinations; error bars are SD for triplicate analyses. Most of the error bars are too small to be visible. P-values for end-point measurements were calculated using the unpaired, two-sided Student’s t-test.

CADM1 to be differentially expressed with strong statistical significance between stage 4S and 4 neuroblastoma that usually correspond to regressing and progressing phenotypes, respectively (Fischer et al., 2006). High CADM1 transcript levels in stage 4S neuroblastoma samples may suggest that CADM1 expression is related to spontaneous regression of disseminated neuroblastoma. Furthermore, this gene has been highlighted as a candidate tumour suppressor gene in human neuroblastoma, because it was found to be downregulated in neuroblastoma cells in comparison with foetal neuroblasts (De Preter et al., 2006). These data prompted us to investigate the role of CADM1 in Oncogene

neuroblastoma tumour behaviour. First, CADM1 expression was correlated with established prognostic variables and patient outcome in a cohort of 251 neuroblastomas. In general, a strong association of low CADM1 mRNA levels with unfavourable prognostic markers like stage 4, age >18 months, MYCN amplification and chromosome 11q alterations was found. Furthermore, low CADM1 transcript levels significantly correlated with unfavourable gene expression-based classification according to the PAM classifier we have defined previously (Oberthuer et al., 2006). In line with these results, low and intermediate/high CADM1 expression levels discriminated patients with beneficial and adverse 5-year survival, respectively, thereby supporting the assumption that CADM1 may serve as a tumour suppressor in neuroblastoma. Similarly, an inverse correlation of CADM1 expression with increased malignancy grade and reduced patient survival has been reported for meningioma (Surace et al., 2004). One might therefore suggest that this tumour suppressor requires a critical transcript level to prevent aggressive behaviour of malignant cells in solid tumours resulting in progressive and advanced cancer. Promoter hypermethylation of CADM1 has been observed in many malignancies (Fukami et al., 2003; Li et al., 2005; Ehrlich et al., 2006; Kikuchi et al., 2006; Worsham et al., 2006) and has thus been suggested to represent the most common molecular mechanism responsible for gene silencing of CADM1 in human cancers. In contrast to these reports, we could not detect any associations of CADM1 expression level with the methylation status of the CADM1 promoter region in a set of 18 primary neuroblastomas. Therefore, the molecular mechanism underlying transcriptional repression in these cases is unlikely to be hypermethylation. Although we cannot rule out the possibility that hypermethylation of a distinct set of CpG sites in the promoter region might have occurred as described for cervical cancer (Li et al., 2005), the eight CpG sites chosen for our analysis have been considered to be representative for global hypermethylation in several studies (Fukami et al., 2003; Steenbergen et al., 2004; Kikuchi et al., 2006). Nevertheless, hypermethylation can be indicative, but is not necessarily involved in downregulation of CADM1 expression as suggested from previous analysis of two prostate cancer cell lines (Fukuhara et al., 2002). Alternatively, loss of CADM1 expression could be a result of haploinsufficiency, as deletion of the chromosomal region 11q23 is frequently observed in unfavourable neuroblastoma. However, in our analysis, low CADM1 expression levels were observed in five samples despite normal status of chromosome 11q (Figure 3). On the other hand, intermediate CADM1 expression levels were detected in two tumours with chromosome 11q deletions (Figure 3). These results indicate that neither hypermethylation nor haploinsufficiency nor biallelic inactivation by a combination of loss of heterozygosity and methylation is regularly involved in CADM1 downregulation in neuroblastoma. Furthermore, sequence analysis of 20 tumours with low and five


3335

tumours with high CADM1 expression revealed no evidence of inactivating mutations of the CADM1 coding region. Although a total of four tumours with low CADM1 mRNA levels were found to harbour mutations within the coding region, none of these resulted in disruption or termination of the CADM1 reading frame, which may indicate that they probably represent polymorphisms of the gene. Together, we therefore speculate that different mechanisms of gene silencing (for example, downregulation of essential activators or enhancers, changes in chromatin structure or upregulation of specific repressors) are responsible for decreased CADM1 expression in neuroblastoma, which needs to be examined by further investigations. To determine whether CADM1 directly contributes to the suppression of an aggressive phenotype in neuroblastoma, we examined its effect on growth properties of neuroblastoma cells after in vitro re-expression. Low endogenous CADM1 protein expression was found in three of four neuroblastoma cell lines, which corresponds well to the range of CADM1 transcript levels in patients with poor outcome (Figures 1 and 4). Different levels of endogenous CADM1 expression have been reported for several human cancer cell lines (Kuramochi et al., 2001; Ito et al., 2003b), and were considered to be a result of a complex regulatory network of important molecules, such as APC (Naishiro et al., 2005), ErbB4 (Ojeda et al., 2006) and MITF (Ito et al., 2003a). The significant protein level of endogenous CADM1 in Kelly cells might thus result from individual molecular characteristics of this cell line such as high MYCN expression. In comparison to LacZ-transduced controls, proliferation was significantly reduced in XTT turnover assays in all four CADM1-transduced cell lines irrespective of endogenous CADM1 expression. These results were in line with previous reports in several cancers (Surace et al., 2004; Houshmandi et al., 2006). As the cell lines used in this study differ in several biological and molecular characteristics like doubling time, morphology, MYCN status or tissue (metastatic or primary tumour tissue) they were derived from (Tumilowicz et al., 1970; Schwab et al., 1983; Ciccarone et al., 1989), our results suggest a strong antiproliferative effect of CADM1 expression on neuroblastoma tumour behaviour in general. In line with these results, a substantial reduction of cell viability was found in an LDH release assay, suggesting that over-expression of CADM1 in human neuroblastoma cell lines does not only reduce proliferation, but might also function as a proapoptotic or pronecrotic regulator. CADM1 over-expression is known to induce apoptosis in A549 non–small cell lung cancer cells (Mao et al., 2004), but many human neuroblastoma cell lines show considerable resistance to apoptosis when exposed to common apoptosis inducing triggers (Teitz et al., 2000; Lombet et al., 2001). Therefore, further experiments are necessary to more precisely delineate the mechanisms by which CADM1 triggers changes in growth properties and loss of cell viability. Finally, CADM1 significantly decreased anchorage-independent clonal growth in short-term soft agar assays in three of four neuroblastoma cell lines

investigated, which underlines its attenuating effect on in vitro tumorigenicity. In conclusion, this study demonstrates that expression of the putative tumour suppressor gene CADM1 is downregulated in aggressive neuroblastoma in general. The mechanisms that regulate CADM1 expression in neuroblastoma remain elusive, however, mutations in the coding region, promoter hypermethylation and haploinsufficiency apparently do not contribute to a significant extent for this event. The finding of impaired proliferation, reduced cell viability and decreased colony-forming ability after reconstitution of CADM1 expression in neuroblastoma cell lines supports its role as a molecule with critical regulatory functions in solid tumour progression. Further analysis of this gene is expected to contribute to the understanding of the molecular processes underlying the biological and clinical differences observed in neuroblastoma.

Materials and methods Microarray data and patients’ characteristics Gene expression profiles from 251 patients (stage 1, n ¼ 69; stage 2, n ¼ 44; stage 3, n ¼ 40; stage 4, n ¼ 67; stage 4S, n ¼ 31) of the German Neuroblastoma Trials NB90-NB2004 were generated as dye-flipped dual-colour replicates using customized 11 K oligonucleotide-microarrays as previously described (Oberthuer et al., 2006). Patients’ age at diagnosis ranged from 0 to 296 months (median age, 15 months). Median follow-up for patients without fatal events was 4.5 years (range, 0.8–15.6 years). Stage was classified according to the International Neuroblastoma Staging System (Brodeur et al., 1993). All raw and normalized microarray data are available at the ArrayExpress database (http://www.ebi.ac.uk/ arrayexpress; accession: E-TABM-38). Data analysis and statistics Statistical analysis was performed using SPSS software version 11.0.1 for Windows (SPSS GmbH Software, Munich, Germany). Levels of CADM1 mRNA determined by microarray analysis were compared in patient groups defined by MYCN status (normal vs amplified), age (o18 months vs >18 months), tumour stage (stage 1–3 and 4S vs stage 4), status of chromosome 11q (normal vs deletion/imbalance) and gene expression-based classification (favourable vs unfavourable) according to the published PAM classifier (Oberthuer et al., 2006). Two-tailed nonparametric tests (Wilcoxon, Mann– Whitney U, and Kruskal–Wallis test) were used where appropriate to test whether CADM1 expression levels differed significantly in these groups. Kaplan–Meier estimates for OS and EFS were calculated and compared by log-rank test. Recurrence, progression and death from disease were considered as events. Bisulphite sequencing CADM1 promoter methylation was analysed by bisulphite sequencing as described elsewhere (Waha et al., 2003). Sodium bisulphite-modified genomic DNA isolated from primary neuroblastoma specimens (Puregene DNA isolation Kit; Gentra Systems, Minneapolis, MN, USA) was tested for methylation at eight CpG sites located in the CADM1 promoter region that have been demonstrated to be differentially methylated in several tumours (Fukami et al., 2003; Oncogene


3336 Steenbergen et al., 2004; Kikuchi et al., 2006). Briefly, 1 mg DNA was diluted in 50 ml H2O, denatured with 5.5 ml 2 M NaOH for 10 min at 37 1C and incubated with 30 ml 10 mM hydroquinone and 520 ml 3 M sodium bisulphite at 50 1C for 16 h. After denaturation with 5.5 ml 3 M NaOH, the DNA was precipitated with 10 M ammonium acetate and 100% ethanol, washed with 70% ethanol and resuspended in 30 ml HPLCH2O. Bisulphite-treated DNA was then amplified using internal primers for the promoter region of CADM1 as previously described (Steenbergen et al., 2004). The PCR mixtures contained 20 ng of modified genomic DNA, primers at 0.5 mM, each of the four deoxynucleotide triphosphates at 200 mM, 1.5 mM MgCl2 and 1.25 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Weiterstadt, Germany) in a total volume of 50 ml. Amplification conditions were 95 1C for 7 min, followed by 30 cycles of 94 1C for 30 s, 59 1C for 30 s and 72 1C for 45 s, and a final elongation step at 72 1C for 10 min. Finally, the resulting PCR products were subcloned for sequencing (Seqlab, Go¨ttingen, Germany) and sequencing data were compiled and analysed using BiQ Analyzer software (Bock et al., 2005; see Supplementary data for detailed information). Sequencing of the coding region of CADM1 After extraction of total RNA from primary neuroblastoma specimens, cDNA was generated using Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA, USA) as previously described (Oberthuer et al., 2006). The complete coding region was then amplified by PCR, and amplificates were directly sequenced using the BigDye Terminator sequencing Kit (Version 3.1, Applied Biosystems). Mutations that were found in the cDNA were verified on the DNA level by amplification and direct sequencing of the respective genomic regions. Primer sequences for amplification and sequencing are available from the authors upon request. Cell culture The neuroblastoma cell lines SH-SY5Y, Kelly and SHEP were obtained from American Tissue Culture Collection (ATCC, Rockville, MD, USA), and the neuroblastoma cell line IMR32 was purchased from DSMZ-German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). All neuroblastoma cell lines were maintained in RPMI-1640 (PAA, Co¨lbe, Germany) supplemented with 10% fetal calf serum (Invitrogen) and 5 mg ml1 ciprofloxacin (Bayer, Leverkusen, Germany), which is referred to as RPMI-1640 throughout the article. Human embryonic kidney cells HEK293A (Invitrogen) were grown in Dulbecco’s modified Eagle’s medium (PAA) supplemented with 10% fetal calf serum (Invitrogen). All cell lines were maintained in a humidified incubator at 37 1C with 5% CO2 and passaged at 90% confluence using Accutase (PAA). Plasmids and adenoviruses Adenovirus transducing CADM1 cDNA was prepared with the ViraPower Adenovirus Expression System (Invitrogen) according to the manufacturers instructions. A pAd/CMV/V5GW/lacZ vector supplied with the kit served as a positive control for infection efficiency. A human CADM1 full openreading frame Gateway Shuttle Clone (pDONR201-CADM1, RZPDo834A0712D2) was obtained from the RZPD (Deutsches Ressourcenzentrum fu¨r Genomforschung GmbH, Berlin-Charlottenburg, Germany). The nucleotide sequence was confirmed by direct sequencing using the BigDye Terminator sequencing Kit (Applied Biosystems). The adenoviral plasmid pAd/CMV/V5-CADM1 was generated by Oncogene

Gateway LR Clonase-mediated recombination using pAd/ CMV/V5-DEST (Invitrogen) as acceptor and the pDONR201CADM1 plasmid as donor. The resulting plasmid was purified (NucleoBond Kit PC 500; Macherey-Nagel, Du¨ren, Germany) and sequenced. Cell supernatants containing recombinant adenovirus transducing either LacZ (adLacZ) or CADM1 (adCADM1) were produced in HEK-293A cells according to the manufacturer’s protocol (Virapower Adenovirus Expression System; Invitrogen). Viral titres were determined using the Adeno-X Rapid Titer Kit (Clontech, Heidelberg, Germany). Neuroblastoma cells were infected with either adLacZ (control) or adCADM1 using equal amounts of virus particles per cell. Transduction efficiencies were monitored by b-Gal staining by using a b-Gal Staining Kit (Invitrogen). Northern blot analysis CADM1 mRNA levels were analysed 72 h after adenoviral infection of neuroblastoma cell lines with either adLacZ or adCADM1. Total RNA was isolated using TRIzol reagent (Invitrogen), and northern blotting with 3–10 mg of total RNA was performed as previously described (Oberthuer et al., 2004). A sequence verified PCR product of 624 bp generated with CADM1-specific primers (forward: 50 -GTGAACTCAAAG TATCATTGACAA-30 ; reverse: 50 -CAGCGGTATGTAC CATTATCTG-30 ) served as template for the CADM1 probe. Western blot analysis CADM1 protein levels were analysed 72 h after adenoviral infection of neuroblastoma cell lines with either adLacZ or adCADM1. The NuPAGE electrophoresis and blotting system (Invitrogen) was used for protein analysis. About 10 mg of protein obtained from transduced cell lines or primary tumour samples were separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis with 4–12% bis-Tris gels and transferred to nitrocellulose membranes by tank blotting. The membranes were blocked with 5% dry milk powder in 0.05% Tween 20/phosphate-buffered saline before incubation with primary antibody (SynCAM antibody ab3910, dilution 1:1000; Abcam, Cambridge, UK) and horseradish peroxidase-labelled secondary goat anti-rabbit antibody (dilution 1:2000; Dako, Glostrup Denmark). The antigen–antibody complex was detected with Visualizer Spray & Glow (Upstate, Schwalbach, Germany). In vitro growth properties assays The effects of CADM1 expression on cell proliferation were assessed using the Cell Proliferation Kit II (XTT; Roche Diagnostics GmbH, Mannheim, Germany) as described by the manufacturer. Neuroblastoma cells were seeded in 24-well plates at a density of 30 000 cells per well in 800 ml RPMI-1640 and transduced the next day with either adLacZ or adCADM1. To assure continuous supply of nutrients throughout the measurement interval, 400 ml fresh RPMI1640 per well were added on day 5. XTT turnover was measured at days 4, 7 and 10 post-infection by adding XTT labelling mixture and incubation for 7 h at 37 1C to allow colour development. The formation of formazan was quantified in 96-well plates measuring the absorbance at 450 nm against a reference wavelength of 690 nm with an enzymelinked immunosorbent assay reader (Multiscan Ascent plate reader; Thermo Labsystems, Helsinki, Finland). Cell viability was determined by measuring cell membrane integrity, as indicated by the leakage of LDH into the culture supernatant (Decker and Lohmann-Matthes, 1988). The LDH activity in the extracellular medium was quantified using the colorimetric Cytotoxicity Detection Kit Plus (Roche


3337 Diagnostics GmbH) according to the manufacturer’s instructions. Neuroblastoma cells were seeded in 96-well plates at a density of 4500 cells per well in 100 ml RPMI-1640. On the next day, the medium was changed to contain 1% bovine serum albumin (Roth, Karlsruhe, Germany) instead of 10% fetal calf serum and cells were transduced with either adLacZ or adCADM1. LacZ-transduced cells were used as a control to determine background release of LDH caused by viral infection. At 5, 24, 48 and 72 h post-infection, LDH substrate mixture was added according to the manufacturer’s instructions, followed by incubation in the dark for 30 min at room temperature. The reaction was terminated by the addition of stop solution, and absorbance was measured at 450 nm against a reference wavelength of 690 nm with a Multiscan Ascent plate reader (Thermo Labsystems). The ability of CADM1 to inhibit soft agar colony forming of neuroblastoma cells was measured using the CytoSelect 96well Cell Transformation Assay (Colorimetric, Cell Recovery Compatible; Cell Biolabs, San Diego, CA, USA) according to

the manufacturer’s instructions. In this short-term assay, CADM1- and LacZ-transduced cells were grown in soft agar for 7 days starting 2 hours post-infection. For end-point quantification of soft agar growth, the soft agar matrix was solubilized, cells were stained using MTT, lysed and measured at 570 nm against a reference wavelength of 690 nm. All assays were conducted in triplicate and reproduced in two independent experiments. The statistical significance of differences in end-point measurements was evaluated by the unpaired, two-sided Student’s t-test. Acknowledgements We thank Yvonne Kahlert for excellent technical assistance. This work was supported by the Deutsche Krebshilfe (grants 50-2719-Fi1 and 106847), the Bundesministerium fu¨r Bildung und Forschung through the National Genome Research Network 2 (NGFN2 grants 01GS0456 and 01GR0450) and a grant of the Kind-Philipp Stiftung.

References Berwanger B, Hartmann O, Bergmann E, Bernard S, Nielsen D, Krause M et al. (2002). Loss of a FYN-regulated differentiation and growth arrest pathway in advanced stage neuroblastoma. Cancer Cell 2: 377–386. Bock C, Reither S, Mikeska T, Paulsen M, Walter J, Lengauer T. (2005). BiQ Analyzer: visualization and quality control for DNA methylation data from bisulfite sequencing. Bioinformatics 21: 4067–4068. Brodeur GM, Pritchard J, Berthold F, Carlsen NL, Castel V, Castelberry RP et al. (1993). Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment. J Clin Oncol 11: 1466–1477. Ciccarone V, Spengler BA, Meyers MB, Biedler JL, Ross RA. (1989). Phenotypic diversification in human neuroblastoma cells: expression of distinct neural crest lineages. Cancer Res 49: 219–225. De Preter K, Vandesompele J, Heimann P, Yigit N, Beckman S, Schramm A et al. (2006). Human fetal neuroblast and neuroblastoma transcriptome analysis confirms neuroblast origin and highlights neuroblastoma candidate genes. Genome Biol 7: R84. Decker T, Lohmann-Matthes ML. (1988). A quick and simple method for the quantitation of lactate dehydrogenase release in measurements of cellular cytotoxicity and tumor necrosis factor (TNF) activity. J Immunol Methods 115: 61–69. Ehrlich M, Woods CB, Yu MC, Dubeau L, Yang F, Campan M et al. (2006). Quantitative analysis of associations between DNA hypermethylation, hypomethylation, and DNMT RNA levels in ovarian tumors. Oncogene 25: 2636–2645. Fischer M, Oberthuer A, Brors B, Kahlert Y, Skowron M, Voth H et al. (2006). Differential expression of neuronal genes defines subtypes of disseminated neuroblastoma with favorable and unfavorable outcome. Clin Cancer Res 12: 5118–5128. Fukami T, Fukuhara H, Kuramochi M, Maruyama T, Isogai K, Sakamoto M et al. (2003). Promoter methylation of the TSLC1 gene in advanced lung tumors and various cancer cell lines. Int J Cancer 107: 53–59. Fukuhara H, Kuramochi M, Fukami T, Kasahara K, Furuhata M, Nobukuni T et al. (2002). Promoter methylation of TSLC1 and tumor suppression by its gene product in human prostate cancer. Jpn J Cancer Res 93: 605–609. Gomyo H, Arai Y, Tanigami A, Murakami Y, Hattori M, Hosoda F et al. (1999). A 2-Mb sequence-ready contig map and a novel immunoglobulin superfamily gene IGSF4 in the LOH region of chromosome 11q23.2. Genomics 62: 139–146. Heller G, Geradts J, Ziegler B, Newsham I, Filipits M, MarkisRitzinger EM et al. (2007). Downregulation of TSLC1 and DAL-1 expression occurs frequently in breast cancer. Breast Cancer Res Treat 103: 283–291.

Houshmandi SS, Surace EI, Zhang HB, Fuller GN, Gutmann DH. (2006). Tumor suppressor in lung cancer-1 (TSLC1) functions as a glioma tumor suppressor. Neurology 67: 1863–1866. Ito A, Jippo T, Wakayama T, Morii E, Koma Y, Onda H et al. (2003a). SgIGSF: a new mast-cell adhesion molecule used for attachment to fibroblasts and transcriptionally regulated by MITF. Blood 101: 2601–2608. Ito T, Shimada Y, Hashimoto Y, Kaganoi J, Kan T, Watanabe G et al. (2003b). Involvement of TSLC1 in progression of esophageal squamous cell carcinoma. Cancer Res 63: 6320–6326. Jansen M, Fukushima N, Rosty C, Walter K, Altink R, Heek TV et al. (2002). Aberrant methylation of the 50 CpG island of TSLC1 is common in pancreatic ductal adenocarcinoma and is first manifest in high-grade PanlNs. Cancer Biol Ther 1: 293–296. Katoh M. (2003). Identification and characterization of TPARM gene in silico. Int J Oncol 23: 1213–1217. Katoh M. (2004). Identification and characterization of human TMEM25 and mouse Tmem25 genes in silico. Oncol Rep 12: 429–433. Kikuchi S, Yamada D, Fukami T, Maruyama T, Ito A, Asamura H et al. (2006). Hypermethylation of the TSLC1/IGSF4 promoter is associated with tobacco smoking and a poor prognosis in primary nonsmall cell lung carcinoma. Cancer 106: 1751–1758. Kuramochi M, Fukuhara H, Nobukuni T, Kanbe T, Maruyama T, Ghosh HP et al. (2001). TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer. Nat Genet 27: 427–430. Li J, Zhang Z, Bidder M, Funk MC, Nguyen L, Goodfellow PJ et al. (2005). IGSF4 promoter methylation and expression silencing in human cervical cancer. Gynecol Oncol 96: 150–158. Lombet A, Zujovic V, Kandouz M, Billardon C, Carvajal-Gonzalez S, Gompel A et al. (2001). Resistance to induced apoptosis in the human neuroblastoma cell line SK-N-SH in relation to neuronal differentiation. Role of Bcl-2 protein family. Eur J Biochem 268: 1352–1362. Mao X, Seidlitz E, Truant R, Hitt M, Ghosh HP. (2004). Reexpression of TSLC1 in a non-small-cell lung cancer cell line induces apoptosis and inhibits tumor growth. Oncogene 23: 5632–5642. Mertens F, Johansson B, Hoglund M, Mitelman F. (1997). Chromosomal imbalance maps of malignant solid tumors: a cytogenetic survey of 3185 neoplasms. Cancer Res 57: 2765–2780. Murakami Y. (2005). Involvement of a cell adhesion molecule, TSLC1/IGSF4, in human oncogenesis. Cancer Sci 96: 543–552. Naishiro Y, Yamada T, Idogawa M, Honda K, Takada M, Kondo T et al. (2005). Morphological and transcriptional responses of untransformed intestinal epithelial cells to an oncogenic beta-catenin protein. Oncogene 24: 3141–3153. Oncogene


3338 Nakagawara A, Arima-Nakagawara M, Scavarda NJ, Azar CG, Cantor AB, Brodeur GM. (1993). Association between high levels of expression of the TRK gene and favorable outcome in human neuroblastoma. N Engl J Med 328: 847–854. Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R et al. (2006). Customized oligonucleotide microarray gene expressionbased classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol 24: 5070–5078. Oberthuer A, Hero B, Spitz R, Berthold F, Fischer M. (2004). The tumor-associated antigen PRAME is universally expressed in highstage neuroblastoma and associated with poor outcome. Clin Cancer Res 10: 4307–4313. Ohira M, Oba S, Nakamura Y, Isogai E, Kaneko S, Nakagawa A et al. (2005). Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas. Cancer Cell 7: 337–350. Ojeda SR, Lomniczi A, Mastronardi C, Heger S, Roth C, Parent AS et al. (2006). Minireview: the neuroendocrine regulation of puberty: is the time ripe for a systems biology approach? Endocrinology 147: 1166–1174. Racher AJ, Looby D, Griffiths JB. (1990). Use of lactate dehydrogenase release to assess changes in culture viability. Cytotechnology 3: 301–307. Schwab M, Alitalo K, Klempnauer KH, Varmus HE, Bishop JM, Gilbert F et al. (1983). Amplified DNA with limited homology to myc cellular oncogene is shared by human neuroblastoma cell lines and a neuroblastoma tumour. Nature 305: 245–248.

Steenbergen RD, Kramer D, Braakhuis BJ, Stern PL, Verheijen RH, Meijer CJ et al. (2004). TSLC1 gene silencing in cervical cancer cell lines and cervical neoplasia. J Natl Cancer Inst 96: 294–305. Surace EI, Lusis E, Murakami Y, Scheithauer BW, Perry A, Gutmann DH. (2004). Loss of tumor suppressor in lung cancer-1 (TSLC1) expression in meningioma correlates with increased malignancy grade and reduced patient survival. J Neuropathol Exp Neurol 63: 1015–1027. Teitz T, Wei T, Valentine MB, Vanin EF, Grenet J, Valentine VA et al. (2000). Caspase 8 is deleted or silenced preferentially in childhood neuroblastomas with amplification of MYCN. Nat Med 6: 529–535. Tumilowicz JJ, Nichols WW, Cholon JJ, Greene AE. (1970). Definition of a continuous human cell line derived from neuroblastoma. Cancer Res 30: 2110–2118. Waha A, Koch A, Meyer-Puttlitz B, Weggen S, Sorensen N, Tonn JC et al. (2003). Epigenetic silencing of the HIC-1 gene in human medulloblastomas. J Neuropathol Exp Neurol 62: 1192–1201. Wakayama T, Koami H, Ariga H, Kobayashi D, Sai Y, Tsuji A et al. (2003). Expression and functional characterization of the adhesion molecule spermatogenic immunoglobulin superfamily in the mouse testis. Biol Reprod 68: 1755–1763. Worsham MJ, Chen KM, Meduri V, Nygren AO, Errami A, Schouten JP et al. (2006). Epigenetic events of disease progression in head and neck squamous cell carcinoma. Arch Otolaryngol Head Neck Surg 132: 668–677.

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

Oncogene