Introduction
Neuroblastoma (NB) is the most commonly occurring solid extra-cranial tumor in children accounting for 6% of cancer incidence and 9% of cancer deaths in children [
1]. It is a highly clinically and biologically heterogeneous cancer of the postganglionic sympathetic nervous system with tumors developing from immature or dedifferentiated neural crest cells [
1,
2]. Most tumors originate in the adrenal medulla or in paraspinal sympathetic ganglia. Common genetic alterations in NB tumors are
MYCN amplification, 17q gain, 1p deletion and loss of 11q [
2,
3]. The list of genes epigenetically silenced in cancer is growing and the inactivated genes represent all cellular pathways. Several genes have been reported as silenced by methylation in NB and one example is the Ras-associated family member
RASSF1A, located at chromosome 3p [
4]. CpG island methylation of
RASSF1A has been reported as a frequent event in NB tumors and cell lines [
4] and loss of heterozygozity (LOH) at 3p, i.e. the loci containing the
RASSF1A gene, has been reported in primary NB tumors [
5].
RASSF1A is also epigenetically silenced by promoter methylation in many other human tumors [
6]. The Ras proto-oncogenes belong to a super-family of GTPases that participate in a range of cellular processes such as cell growth, adhesion, migration, differentiation and apoptosis [
7], with defects in Ras signaling pathway resulting in disease and oncogenesis. The Ras proteins carry out their diverse functions via interaction with RASS effectors which have conserved Ras interacting domains. One of many such Ras interacting domains is the RA-domain, and the RA-domain is a common feature of the genes in the Ras-association domain family (RASSF). This family has ten members;
RASSF1-10, which are divided into two groups, the classical members
RASSF1-6 and the N-terminal members
RASSF7-10[
8]. The classical
RASSF family members have been reported to be involved in many biological processes such as microtubule stability, cell cycle control and apoptosis and are generally considered as tumor suppressors [
8]. Based on our previous data using IIumina 27K methylation arrays [
9] we noted that several of the
RASSF genes were methylated in NB. Eight of the
RASSF genes were included on the IIumina 27K methylation arrays
(RASSF1A, RASSF2, RASSF3, RASSF4, RASSF5, RASSF6, RASSF7 and
RASSF8). The following seven RASSF genes were chosen for further methylation analysis;
RASSF2, RASSF4, RASSF5, RASSF6, RASSF7, RASSF8 and
RASSF10). The two CpG sites in
RASSF3 were unmethylated in all NB tumors and this gene was therefore not investigated further.
RASSF8 however, we wanted to include in the verification analysis with BSP to see if surrounding CpG sites also were unmethylated since this gene has been reported as methylated in Childhood Leukemia cell lines [
10].
RASSF1A was not analyzed further as this gene is well known to be deregulated in NB due to DNA methylation. Recent published data have shown that
RASSF10 is methylated in other cancers which led us to include this gene in our analyses. In addition to the
RASSF1A gene, DNA methylation was found in six out of seven analyzed
RASSF genes (
RASSF2, 4, 5–7 and
10). Several of the
RASSF genes had reduced mRNA expression levels in NB cell lines and the methylation status of some of the
RASSF genes was able to significantly discriminate between biological subgroups of NB tumors.
Discussion
In this study, we investigated whether the
RASSF family genes are epigenetically silenced in NB. Data from our previously performed 27K methylation array showed that members of the
RASSF gene family were methylated in NB cell lines and tumors [
9]. The methylation status observed with the 27K methylation array was verified by DNA methylation analysis using bisulfite sequencing, MSP or COBRA, of the corresponding CpG island. DNA methylation was most commonly observed in
RASSF7 (eight out of nine cell lines methylated),
RASSF5 (six to seven out of nine cell lines methylated depending on region analyzed) and
RASSF6 (six out of nine cell lines methylated) (Figure
4A). Gene expression analysis performed on NB cell lines showed in general low to moderate expression of the
RASSF genes (Figure
4B).
RASSF4 had the highest expression in the NB cell lines whereas the mRNA levels of
RASSF6,
RASSF7 and
RASSF10 were either below detection levels or very low in most NB cell lines. Also, the mRNA levels of
RASSF2A and
RASSF5 were low in general. In order to see if
RASSF gene expression could be restored, NB cell lines were treated with 5-Aza-dC/or TSA. Gene expression of most of the
RASSF genes (
RASSF2A,
RASSF4,
RASSF5,
RASSF7 and
RASSF10) was up-regulated following epigenetic treatment, suggesting these genes may be epigenetically regulated. The concentration of each epigenetic drug and the treatment time will most certainly affect the results of up-regulation. The chosen conditions of drugs and treatment times were based upon optimization conditions where re-expression of a panel of known methylated genes occurred [
9]. More gene-specific optimizations of the epigenetic drug treatment conditions may be necessary in order to explore the re-activation potential fully. Also, up-regulation of a gene may be the result of their up-stream regulators being affected by treatment and it is also possible that other epigenetic mechanisms than DNA methylation are responsible for the up-regulation. Genes belonging to the
RASSF family are generally considered as TSGs and many of the members have been reported as silenced by promoter methylation in human cancers, (Table
5).
Table 5
Information of the
RASSF
genes studied in relation to tumorigenesis
RASSF1A
| 3p21 | Ras association domain family member 1 | NB tumors and cell lines [ 4] as well as various human cancers, reviewed in [ 6] | TSG involved in regulation of cell proliferation. Promotes apoptosis and cell cycle arrest, involved in migration and maintenance of genomic stability (reviewed in [ 6]). KO of RASSF1A in mice enhances spontaneous tumor formation [ 18]; [ 19] |
RASSF2
| 20p13 | Ras association domain family member 2 | Colorectal cancer [ 13], gastric cancer [ 20], nasopharyngeal carcinoma [ 21], breast-, lung and NSCLC tumors [ 22], thyroid cancer [ 23], pancreatic cancer [ 24] | Function as a TSG, reduces colony formation, promotes apoptosis and cell cycle arrest [ 25]; [ 15]; [ 21]; [ 22]; [ 23] |
RASSF4
| 10q11 | Ras association domain family member 4 | Kidney-, breast- and lung cancer cell lines, breast- and lung primary tumors [ 14] | Have growth inhibitory properties and promotes apoptosis in lung- and breast tumor cell lines [ 14] |
RASSF5
| 1q32 | Ras association domain family member 5 | NB cell lines [ 26]; [ 27], lung-, breast-, colorectal-, and kidney tumor cell lines and in primary NSCLC tumors [ 28], Wilms tumor [ 29], CCRCC [ 30], gastric cancer [ 20], colon cancer [ 13], squamous cell cancer of head and neck [ 31], hepatocellular carcinoma [ 32] | Also called NORE1 and forms heterodimers with RASSF1A [ 33]. Associate with microtubules and act growth inhibitory by a process involving p53. Promotes apoptosis when overexpressed or in the presence of activated Ras [ 34]. Neuroblastoma tumors, especially non MYCN-amplified, show suppressed NORE1A expression [ 27] |
RASSF6
| 4q13 | Ras association domain family member 6 | | Putative TSG in childhood leukemia [ 10]. Promotes apoptosis [ 35]. Downregulated at both mRNA and protein level in gastric cancer. Gastric cancer patients with RASSF6-negative tumors had worse outcome and higher recurrence rate than patients with RASSF6-positive tumors [ 36] |
RASSF7
| 11p15 | Ras association domain family (N-terminal) member 7 | | Centrosome associated protein necessary for spindle formation and completion of mitosis in the neural tube in Xenopus [ 8]. Required for completion of mitosis in human cells and KO results in mitotic arrest [ 37] |
RASSF8
| 12p12 | Ras association domain family (N-terminal) member 8 | Childhood leukemia cell lines [ 10] | TSG candidate in lung cancer [ 38]. KO enhances anchorage independent growth in soft agar and promotes tumor formation in mice [ 39] |
RASSF10
| 11p15 | Ras association domain family (N-terminal) member 10 | Childhood leukemia [ 10], thyroid cancer [ 40], primary glioblastomas and astrocytomas [ 41], malign melanoma [ 42] | Suggested as a regulator of mitosis. Over-expression decrease colony formation in soft agar [ 41] |
RASSF1A is a TSG involved in a range of cellular processes that are essential for normal cell growth control. Rassf1a is one of the most commonly inactivated proteins in cancer and inactivation by promotor hypermethylation is a common event in various human malignancies, including NB [
4]. Demonstrating the validity of our 27K methylation array data, we detected dense
RASSF1A methylation of NB primary tumors and cell lines, which is in agreement with published data.
RASSF2A mRNA expression was in the current study generally low in NB cell lines and up-regulation of RASSF2A was seen following a combined treatment with 5-Aza-dC and TSA (4/9 NB cell lines) even though methylation at the RASSF2A CpG island was not commonly observed.
RASSF4 mRNA expression was detected in all NB cell lines and 5-Aza-dC and TSA treatment resulted in up-regulation of RASSF4 mRNA levels in 5/9 NB cell lines. The strongest up-regulation was detected in the cell line IMR-32 which also showed the highest methylation level.
RASSF5, also called
NORE1 (Novel Ras Effector 1), is localized at 1q32.1 and has a 60% similarity to
RASSF1A, the most commonly described methylated gene in cancer so far. The
RASSF5 gene encodes at least three different isoforms due to different promoter usage and alternative splicing. Two of the
RASSF5 isoforms,
RASSF5A and
RASSF5C are broadly expressed in most normal tissues.
RASSF5A is the longest isoform transcribed from the most 5′-promoter and the isoform
RASSF5B is produced by alternative splicing. The shorter isoform
RASSF5C is transcribed from a more downstream promoter. Promoter methylation of
RASSF5A has been reported to not occur in primary NB tumors and there are some conflicting data concerning methylation of NB cell lines, where NB cell lines have been described as low methylated or unmethylated in different studies [
26,
27]. Interestingly
RASSF5 was recently shown to be demethylated and up-regulated in the NB cell line SK-N-BE during ATRA-induced differentiation [
43], suggesting that
RASSF5 could be aberrantly methylated in undifferentiated NB tumors cells but demethylated and re-expressed through differentiation. According to our 27K methylation array data, two CpG sites were methylated in NB primary tumors and cell lines. The methylated CpG sites were located in different
RASSF5 promotor CpG islands. The 27K methylation array site cg17558126 was located in the most 5′-promoter where transcription of
RASSF5A starts and cg02589695 were located in a downstream promoter were the
RASSF5C transcript starts. Bisulfite sequencing of the two regions revealed that both CpG sites present on the 27K methylation array were indeed methylated in most NB cell lines (Figure
4A). The methylation status of the CpG sites surrounding cg17558126 (
RASSF5A) was highly variable throughout the CpG island, but most sites showed partial methylation, whereas the CpG sites surrounding cg02589695 (
RASSF5C) were unmethylated in all NB cell lines. The variable methylation of CpG sites in this island might explain why there are conflicting published data regarding the methylation status of
RASSF5A in NB cell lines. Gene expression of
RASSF5A have also been described as low in NB cell lines, with the highest expression in SK-N-SH and absent expression in IMR-32 [
26], which is in agreement with our data (Figure
4A and
4B).
RASSF5 mRNA expression was in this study up-regulated for several NB cell lines. For example, the methylated cell line SH-SY-5Y was up-regulated after 5-Aza-dC treatment and even more up-regulated following a combined treatment with both 5-Aza-dC and TSA (Table
4). Two of the analyzed
RASSF5 CpG sites on the 27K methylation array were significantly more methylated in INRG stage M tumors compared to L tumors (Figure
1B). Also,
RASSF5A methylation was highly correlated to
MYCN amplification (Figure
1B).
RASSF5A mRNA expression have also been reported as frequently down-regulated in NB and pheochromocytoma primary tumors and lower
RASSF5A expression was seen in NB tumors without
MYCN-amplification compared to
MYCN-amplified tumors [
27]. The methylation beta-value for two of the
RASSF5 sites was significantly higher in
MYCN-amplified tumors compared to non-amplified tumors which contradicts an earlier report that showed lower expression of this gene in non-
MYCN-amplified tumors [
27].
RASSF6, located at chromosome region 4q13.3 has recently been suggested as a TSG candidate in childhood leukemia and was found to be silenced by heavy methylation across the whole CpG island in leukemia cell lines [
10]. In the current study,
RASSF6 promoter methylation was found in 6/9 NB cell lines and
RASSF6 expression was absent or just above detection level in the panel of NB cell lines (Figure
4A and
4B). High methylation of
RASSF6 was significantly correlated to unfavorable outcome (5-OS), 1p deletion and
MYCN amplification in our patient cohort (Figure
2B and
2C). Recently,
RASSF6 was shown to be down-regulated at both mRNA and protein level in gastric cancer tumors and loss of
RASSF6 expression correlated with poor survival and increased tumor recurrence rate [
36]. Functional studies have indicated that
RASSF6 is involved in promoting apoptosis [
35].
RASSF7, also known as
HRC1 (HRAS1 cluster 1), is located at chromosome region 11p15.5 and lacks the conserved SARAH domain present in
RASSF1-6. To our knowledge, there are yet no reports of epigenetic silencing of
RASSF7 in cancer but important functions have been reported (Table
5). In this study, bisulfite sequencing showed methylation of the
RASSF7 promotor CpG sites in 8/9 NB cell lines (Figure
4A). All four NB cell lines present on the 27K methylation array were heavily methylated (84-96%) at the analyzed CpG site. Interestingly, the mRNA expression of
RASSF7 was very low or absent in most NB cell lines (Figure
4B). According to our cDNA microarray analysis the methylated NB cell line SK-N-BE was strongly up-regulated following epigenetic treatment, which was verified with end-point RT-PCR (Table
4).
RASSF10, located at 11p15.2, has recently been reported as methylated and silenced in childhood leukemia [
10], thyroid cancer [
40] and in astrocytic glioma [
41]. In this study the
RASSF10 mRNA expression was absent or just above the detection level in NB cell lines but low methylation was found in only 1/9 NB cell lines (Figure
4A and B).
Collectively, the
RASSF family members have been demonstrated to have several tumor suppressive properties (Table
5). Although RASSF proteins lack catalytic activity, they are suggested to be non-enzymatic adaptors that are involved in growth and tumor suppression. The molecular mechanisms behind their growth suppressing properties are not yet elucidated but a number of reports show association with microtubules or centromeres indicating that the
RASSF genes are important in microtubule dynamics and mitosis. In addition, the
RASSF family genes participate in regulation of apoptosis and epigenetic silencing of
RASSF genes may contribute to cancer by preventing RAS induced-apoptosis. In a normal cell, there is an important balance between signaling pathways that promote survival and those who promote apoptosis. If
RASSF genes are silenced, the pro-apoptotic effects of RAS signaling may be lost which may favor the balance towards the pro-survival PI3 kinase pathway. Future studies regarding the exact function of the
RASSF family genes and their interacting partners are essential to elucidate the role that epigenetic silencing of
RASSF genes might play in NB and cancer in general.
Several genes from various cellular pathways have been reported as epigenetically silenced by DNA methylation in NB. For example, Caspase-8 (
CASP8) located at 2q33 was one of the first genes to be reported as methylated in NB [
44]. Aberrantly methylated genes could in the future be used in clinical patient stratification as biomarkers or as therapeutic targets. Our group and many others have shown that DNA methylation of single genes or a selected group of genes, are able to predict patient outcome, for a review see Decock et al., [
45]. Epigenetic inactivation of
RASSF1A has been reported as associated with high risk disease, age >1 year and poor survival for NB patients [
46]. Further,
RASSF1A hypermethylation in serum from patients with NB has been reported as a reliable prognostic predictor [
47].
In summary, in addition to RASSF1A which is already known as frequently methylated in NB, this study highlights the RASSF gene family members RASSF5, RASSF6 and RASSF7 as promising candidates for further analysis in NB. These three genes are targeted by DNA methylation in NB primary tumors and cell lines and show low levels of mRNA expression in NB cell lines. Also, CpG site specific DNA methylation of RASSF5 and RASSF6 was able to significantly discriminate between different subgroups of NB.
Authors’ contributions
AD carried out the experiments, analyzed the results and drafted the manuscript. HC planned and coordinated the study, performed experimental and statistical analysis and revised the manuscript. TM and PK provided clinical information. All authors read and approved the final manuscript.