Deregulation of Caspase 8 and 10 Expression in Pediatric Tumors and Cell Lines¹

The caspases, including CASP8³ and CASP10, are cysteine proteases involved in apoptosis. CASP8 and CASP10 are upstream or initiator caspases, involving death receptors and their ligands such as TRAIL and TNF. Binding of death receptors to their ligands results in recruitment and activation of CASP8 and CASP10, formation of the death-inducing signaling complex (DISC) and subsequent initiation of the apoptotic cascade (1). CASP8 may play an essential role in apoptosis induced by chemotherapeutic agents and irradiation (2, 3). The role of the closely related gene CASP10, located in the same gene complex at 2q33 (4), is more controversial. Our recent studies indicate that each caspase can initiate apoptosis independently of the other (5). A novel molecule called c-FLIP (FADD like interleukin-1 converting enzyme inhibitory protein) located at 2q33, resembles CASP8 in overall structure but is proteolytically inactive (6). In contrast to CASP8 and 10, overexpression of c-FLIP has been shown to protect against apoptosis mediated by FasL and TRAIL in several cancer cell lines in vitro (6).

Recent studies have reported that CASP8 was frequently inactivated by a combination of methylation and allelic deletion at 2q33 (the chromosomal location of the gene) in neuroblastomas with MYCN amplification (7, 8, 9, 10). CASP8 was methylated with a loss of gene expression in 55% of medulloblastomas (11). Except for a small number of Ewing’s sarcoma and rhabdomyosarcoma cell lines (which retained expression and lacked methylation; Refs. 9, 12), other pediatric tumors have not been examined. We have recently demonstrated that CASP8 is methylated and inactivated in neuroendocrine lung carcinomas (small cell and bronchial carcinoids) but not in non-small cell carcinomas (13). In addition, loss of CASP10 protein expression may occur in breast and lung carcinomas that retain message expression (5).

In this report, we examined the methylation status of the gene in 181 tumors representing the major forms of pediatric tumors. In addition, we investigated expression (at the message and protein level) and CASP8 activity and their relationship to methylation of the gene and expression of CASP10 and c-FLIP in pediatric tumor cell lines.

A total of 181 and 23 pediatric tumors and cell lines were examined. Most of the tumor samples (n = 144) were obtained from Children’s Hospital Medical Center, Dallas, TX, after obtaining Institutional Review Board approval and informed consent. Twenty hepatoblastoma samples were obtained from the Pediatric Oncology Group Hepatoblastoma Tumor Bank. The retinoblastomas (n = 17) were from the University of Siena, Siena, Italy. Rhabdomyosarcoma samples included alveolar (n = 6), embryonal (n = 7), and anaplastic types (n = 2). Eighteen acute lymphoblastic (ALL) and two acute myelogenous leukemia (AML) samples were included in the acute leukemia cases. Nonmalignant samples included corresponding histologically normal kidney (n = 11) from Wilms’ tumor cases and liver (n = 1) from a hepatoblastoma case, peripheral blood lymphocytes (n = 2) from healthy adults, and autopsy tissues from patients without cancer (six muscle, six lung, and five liver samples; Tables 1 and 2).

Twenty-three pediatric tumor cell lines were obtained from the American Type Culture Collection (ATCC; Manassas, VA; Table 1). Cell lines were grown in RPMI 1640 (Life Technologies, Inc., Rockville, MD) supplemented with the optimal concentration (10–20%) of fetal bovine serum and incubated in 5% CO₂ at 37°C.

Genomic DNA was isolated from frozen tissues and cell lines by SDS and proteinase K digestion, phenol/chloroform extraction, and ethanol precipitation (14). DNA was treated with sodium bisulfite as described previously (15). Two μl of bisulfite-modified DNA was amplified by PCR, using primers that were specific for methylated or unmethylated sequences of CASP8 gene as described previously (7). PCR-amplified products were electrophoresed on 2% agarose gel and visualized under UV illumination.

Total RNA was extracted from cell lines using TRI Reagent (Molecular Research Center Inc., Cincinnati, OH) following the manufacturer’s instructions. Two μg of total RNA was reverse-transcripted by use of SUPERSCRIPTII First-Strand synthesis (Invitrogen, Carlsbad, CA), and then 1 μl of cDNA was used for the amplification of mRNA of CASP8 and CASP10. Because multiple isoforms of CASP8 have been described, we used two sets of primers for its expression. One was as described previously (7) for exons 8 and 9, and the other set was designed by us for exons 1 and 3: sense, 5′-GGGAAGTGTTTTCACAGGTT-3′ and antisense, 5′-TTCTTGCTTCCTTTGCGGAAT-3′. Primer sequences for CASP10 were: sense, 5′-AAGAAACAGATGCCCCAGCCTG-3′ and antisense, 5′-CGAGACTACAGTGAGCCGTGATTG-3′. GAPDH, a housekeeping gene, was used as a control for RNA integrity, as described previously (16). PCR products were electrophoresed on 2% agarose gel and visualized under UV illumination.

Cell lines lacking gene expression were treated with 5-aza-CdR at a concentration of 0.5–2 μg/ml for 5 days. TSA treatment was performed at a concentration of 300 nm for 24 h. Duplicate flasks before and after treatment were harvested and tested for gene expression.

Cells were placed in lysis buffer [0.5% NP40, 10 mm Tris HCl (pH 8.0), 150 mm NaCl, 3 mm MgCl₂, and 2 mm phenylmethylsulphonyl fluoride]. Samples normalized for total protein content (30 μg/lane) were separated by 10% SDS-PAGE, electroblotted on nitrocellulose, and immunostained. Monoclonal antibodies to CASP8 (Oncogene, Boston, MA) and CASP10 (MBL International Corporation, Watertown, MA) were used according to the manufacturers’ instructions. A rabbit monoclonal antibody to c-FLIP was generated by one of us (P. M. C.). Detection of immunocomplex was performed using Supersignal (Pierce, Rockford, IL).

The activity of CASP8 was determined using ApoAlert Caspase8 colorimetric assay kit (Clontech, Palo Alto, CA) by following the manufacturer’s instructions. Briefly, CASP8 was activated in pediatric tumor cell lines by the treatment of 100 ng/ml human TNF-α (Clontech) for 24 h. Untreated cells were used as a control for each cell line. Cells were counted (3–5 × 10⁶ cells) and centrifuged (3,000 rpm for 10 min). The pellets were resuspended in 50 μl of cell lysis buffer and incubated on ice for 10 min, and cell lysates were centrifuged at 14,000 rpm for 10 min at 4°C. Fifty μl of 2× reaction buffer containing 10 mm DTT and 5 μl of 4 mm IETD-pNA substrate were added to the supernatants, and the reaction mixtures were incubated at 37°C for 2 h in the dark. The release of pNA as the measure of CASP8 activity was read at 405 nm. The increase of CASP8 activity was determined by comparing the results with untreated controls. A calibration curve was generated using different concentrations of pNA (Sigma, St. Louis, MO). The slope of the curve was used to calculate the units of CASP8 activity.

Bisulfite-modified DNA was amplified by PCR using two sets of primer in the CASP8 gene. Primer sequences were: for CASP8SQ1; sense, 5′-TGTAAGAAAGAATGGTATATTA-3′, and antisense, 5′-ATAAAAACACTTCCCTCCAAC-3′; and for CASP8SQ2, sense, 5′-AGAGTTAGGGTGGTTATTGA-3′, and antisense, 5′-AATAAAATCTTCTAAACTCTCC-3′. The PCR products were cloned into pCR2.1-TOPO vector (Invitrogen); 10 clones from each of the samples were selected, and plasmid DNAs were purified using QIAprep Spin Miniprep kits (Qiagen, Valencia, CA). DNA sequencing was performed by ABI PRISM 377 (Applied Biosystems, Foster City, CA) to determine the methylation status of CASP8.

Statistical analyses for differences between groups were performed using the χ² and the Fisher’s exact tests and the nonparametric Mann-Whitney t test. Kaplan-Meier log-rank test was performed to calculate overall survival.

Methylation of CASP8 was present in 58 (32%) of 181 pediatric tumors and 13 (57%) of 23 cell lines (Tables 1 and 2; Figs. 1 and 2 A). Rhabdomyosarcomas (83%), medulloblastomas (81%), retinoblastomas (59%) and neuroblastomas (52%) were frequently methylated. However, the rate of methylation of Wilms’ tumors was 19%, and methylation was absent in hepatoblastomas, acute leukemias, osteosarcomas, and Ewing’s sarcomas. CASP8 methylation was present in the three major subtypes of rhabdomyosarcomas (4 of 6 alveolar, 9 of 10 embryonal, and 2 of 2 anaplastic types). In contrast to neuroblastomas, the differentiated, much-less-aggressive ganglioneuromas (n = 6) lacked CASP8 methylation. Despite the high rate of CASP8 methylation in primary rhabdomyosarcoma samples (83%), only one of five cell lines (20%) was methylated. Methylation was absent in all of the normal samples examined.

We examined the relationship between methylation of CASP8 and clinicopathological parameters including sex, age, stage, existence of metastasis, and outcome. Neuroblastoma patients with methylation of CASP8 were significantly older than patients whose tumors lacked methylation (2.31 ± 1.37 years old versus 0.87 ± 1.40 years old, respectively; P = 0.002). There was no relationship between methylation status and other clinicopathological parameters.

We previously reported that the RASSF1A gene, a newly described tumor suppressor gene located at 3p21.3 (17), was frequently methylated in pediatric tumors (18). Therefore, we examined the relationship between CASP8 and RASSF1A methylation. There was a significant relationship between them in all of the pediatric tumors (P < 0.0001), in retinoblastoma (P < 0.0001), and in neuroblastoma (P = 0.0039). Amplification of the N-MYC gene in neuroblastoma is a negative prognostic factor and has been associated with CASP8 methylation (8). As a requirement for enrollment onto Pediatric Oncology Group clinical protocols, the patients with neuroblastic tumors had the status of the N-MYC gene in their tumor tissues examined by St. Jude Children’s Research Hospital, Memphis TN. Most of the cases were early stage and only two of the neuroblastomas and ganglioneuromas were amplified for N-MYC by fluorescence in situ hybridization analysis (i.e., copy number >10), and neither of these cases was methylated for CASP8.

We examined expression of CASP8 mRNA by RT-PCR in 23 pediatric tumor cell lines (Table 1; Figs. 2,A and 3). Expression was present in 14 cell lines (61%). Of these, 10 expressed only the unmethylated alleles, whereas 4 cell lines had both methylated and unmethylated alleles. Expression was absent in nine (39%) cell lines. These cell lines had only methylated alleles and lacked unmethylated alleles of CASP8 except for one neuroblastoma line, SK-N-DZ, which had both alleles. We also performed Western blot analysis for CASP8 in 13 cell lines. We detected the two isoforms, caspase-8a and -8b (58,000 and 56,000 daltons, respectively) as described previously (19). There was good concordance between RT-PCR and protein expression, although cell lines SK-N-SH and SK-NEP-1 expressed mRNA but no protein.

We also performed an assay for CASP8 activity after TNF-α stimulation in 13 cell lines (Fig. 3). Although CASP8 activity was variable, it was concordant with protein expression in 11 cell lines. Cell lines WERI-RB-1 and D341 MED had no detectable protein expression but had weak activity.

To determine whether methylation of CASP8 is associated with transcriptional silencing, we examined expression of CASP8 mRNA before and after treatment with 5-aza-CdR, a demethylating agent, by RT-PCR in 11 cell lines (Fig. 2 B). We found restoration of CASP8 mRNA after 5-aza-CdR treatment in six (SH-SY5Y, BE(2)-M17, SK-N-BE(2), D283, Y-79, WERI-RB-1) of seven methylated cell lines but not in D341 MED. Four cell lines (SK-N-FI, SK-N-SH, RD, and SK-NEP-1), which had both alleles and expression, were also treated with 5-aza-CdR. An increase of expression of CASP8 mRNA was detected in three of these four cell lines except for SK-N-SH. Restoration of CASP8 mRNA was not detected in two cell lines (D341 MED, SK-N-SH).

We also treated the 11 cell lines with TSA, an inhibitor of histone deacetylase activity. TSA treatment failed to restore expression in the seven methylated cell lines lacking expression of CASP8. In only one cell line, RD, having both methylated and unmethylated alleles and expression, did the intensity of mRNA expression increase slightly after treatment.

In the original report on methylation of CASP8 in neuroblastomas, Teitz et al. (7) reported that they examined the 5′ flanking region of the gene, presumably the site of the promoter region. However, according to sequencing data of CASP8 (GenBank accession no. AF210257), the region examined is downstream from the TATA box and is located within the 5′ untranslated region of the gene. As we have discussed previously (13), this region only partially satisfies the criteria of a CpG island (20), and the total number of CpG sites (seven) within the MSP amplicon is relatively small. Despite these findings, we found that methylation of this region of the CASP8 gene was associated with CASP8 silencing in both lung cancer (13) and pediatric tumor cell lines. To further investigate this relationship, we amplified two parts (319 bp and 456 bp) of this region that encompassed 15 CpG sites (and that included the MSP amplicon), and cloned the resultant amplicons. As demonstrated in Fig. 2 C, cell lines BE(2)-M17 and D283, which lacked gene expression and had methylated bands by MSP, were heavily methylated at most CpG sites in all of the clones. Cell line SAOS-2, which only had an unmethylated band by MSP assay, completely lacked methylation at all sites. Cell line SK-N-FI, which had both methylated and unmethylated bands by MSP and was expression positive, demonstrated considerable clonal and site variation in its methylation pattern.

The CASP10 gene is located ∼20–30 kb 5′ of the CASP8 gene (4) and CASP10 expression has been reported in CASP8-negative pediatric tumors and cell lines (2, 7, 21). We have previously reported the absence of CASP10 protein in lung and breast cell lines expressing mRNA. For these reasons, we compared mRNA and protein expression of these two functionally related and closely located genes (Fig. 2,A). The intensities of expression of CASP10 mRNA varied, although expression was present in all 13 cell lines examined. However, protein expression was present in only one cell line, SK-N-FI. c-FLIP resides in the same gene cluster at 2q33 as CASP8 and CASP10 (6). All of the cell lines expressed c-FLIP protein (Fig. 2 A).

In this study, we demonstrated that methylation of CASP8 was present in several types of pediatric tumors and its methylation was associated with gene silencing in pediatric tumor cell lines. Rhabdomyosarcomas (83%), medulloblastomas (81%), retinoblastomas (59%), and neuroblastomas (52%) were frequently methylated, whereas the frequencies of methylation were low or absent in other types of pediatric tumors. These results were not discrepant with previous reports (7, 9, 11, 12, 21, 22, 23, 24) except for methylation in Ewing’s sarcomas. Fulda et al. (12) showed that 13 of 20 Ewing’s sarcoma tumors were methylated, but in our study, methylation of CASP8 was absent in Ewing’s sarcoma tumors. However, Fulda et al. found protein expression (by immunostaining) in several of the methylated tumors, making their findings difficult to interpret. Methylation of CASP8 was absent in corresponding nonmalignant tissues and in ganglioneuromas, which are considered to be a differentiated, benign form of neuroblastoma. These findings indicate that methylation of CASP8 is specific for certain pediatric tumors.

Recently, we found that the RASSF1A gene, a putative tumor suppressor gene, is frequently methylated in some types of pediatric tumors (18). We noted a high concordance rate between methylation of CASP8 and RASSF1A. Concordant methylation of both genes may contribute to the pathogenesis of several types of pediatric tumors.

Because the MSP amplicon used in this and other studies does not target the promoter region of the gene, and because this site does not rigorously fulfill the criteria of a CpG island, we performed a detailed correlation between methylation, mRNA expression, protein expression, and protein function in pediatric cell lines. In general, there was a negative concordance between the presence of methylation (by MSP assay) and mRNA expression. Data from bisulfite-modified sequencing were consistent with these results. However, five cell lines had both methylated and unmethylated alleles, and four of these had mRNA expression, which indicated monoallelic inactivation with a functional unmethylated allele, or heterogeneity of methylation. Sequencing studies of one of these four cell lines demonstrated considerable clonal and positional variation in methylation. After treatment with 5-aza-CdR, expression of CASP8 was restored in six of seven methylated cell lines lacking expression, but restoration was not detected after TSA treatment. These findings are further supportive evidence that methylation of CASP8 was associated with gene silencing. Whereas methylation in the region examined is unlikely to be the direct cause of gene silencing, methylation in this region may reflect methylation in the promoter region.

Of 13 cells lines tested, there was concordance between expression of mRNA and protein, but protein expression was absent in 2 lines having mRNA expression. This finding suggests that posttranscriptional control may be responsible for gene silencing in some cases. On the other hand, there was good concordance between expression of protein and enzymatic activity of CASP8.

We also examined the state of the closely related CASP10 gene. Although all of the 13 cell lines expressed mRNA, the intensity of the band was highly variable. However, by Western blot, only one cell line expressed protein. These findings are similar to our previous findings in lung and breast cancer cell lines (5). Teitz et al. (7) found variable intensities of protein expression in neuroblastoma cell lines. We have noted that some commercial antibodies cross-react with the similar-sized heat shock protein 60, including the antibody used by Teitz et al. (5), possibly accounting for this apparent discrepancy. c-FLIP is a molecule closely related to CASP8 but proteolytically inactive and may function as an inhibitor of apoptosis. By Western blot, all of the cell lines expressed variable amounts of c-FLIP. There was no apparent relationship between expression of CASP8, CASP10, and c-FLIP.

We have demonstrated that the loss of protein expression and enzymatic activity of the CASP8 gene are common in pediatric tumor cell lines. In many but not all cases, silencing appears to be related to methylation of the 5′ noncoding region of the gene. Methylation of the gene is common in several but not all types of pediatric tumors, and is concordant with methylation of the RASSF1A gene. Loss of expression of the closely related but functionally independent CASP10 gene is also frequent in many pediatric tumor cell lines. Resistance to apoptosis is one of the hallmarks of cancer (25), and our findings suggest that deregulation of the death receptor pathway to apoptosis is frequent in many types of pediatric tumors and their cell lines.