Epigenetic Inactivation of RAS Association Domain Family Protein 1 (RASSF1A) in Malignant Cutaneous Melanoma

INTRODUCTION

Aberrant methylation of CpG islands of the promoter region of genes and the role that this epigenetic event plays in the development of various cancers have recently become an important area of investigation in assessing the mechanisms of tumor suppressor and regulatory gene inactivation. It has been shown that TSGs 4 can be transcriptionally silenced when their promoter region CpG islands contain methylated cytosines located 5′ to an adjacent guanine (1 , 2) . The recent advent of better methods to assess this epigenetic gene silencing event, such as utilization of MSP, has simplified the detection of hypermethylated CpG islands with limited amount of specimen DNA (3) . The methylation status of several TSG promoter regions has been investigated and profiled for a number of cancers (4, 5, 6, 7) . Studies have shown that hypermethylation of promoter regions of TSGs is quite common and a major alternative mechanism by which a tumor cell can shut off TSG expression in addition to deletions and specific mutations (1 , 2) . The majority of studies on hypermethylation of TSG promoter regions have been focused on carcinomas. To date, there is only a limited amount of work reported regarding the investigation of TSG promoter methylation in malignant cutaneous melanoma tumors (8, 9, 10, 11, 12) .

The Ras superfamily of small GTP-binding proteins plays various critical roles in intracellular signal transduction pathways (13) . In the active GTP-bound form, Ras GTPases interact with various downstream target proteins referred to as Ras effectors, which promote cellular events (14) . The Ras effectors include Raf-1, phosphatidylinositol 3′-kinase, AF-6, RalGDS, and Nore1 (15 , 16) . These Ras effectors have similarities based on sequence homology within a domain termed the Ras association domain (17) . One recently identified and cloned gene is the Ras effector homologue RASSF1 (18 , 19) . RASSF1 is located at chromosome 3p21.3, within a homozygous deletion region of 120 kb that frequently demonstrates LOH in both lung and breast cancers (20) . LOH of 3p21.3 is not limited to these two types of carcinomas, however, suggesting that this region may contain a more universal TSG (18, 19, 20, 21, 22, 23, 24) . The RASSF1 gene properties in various tumors suggest it is a TSG. The RASSF1 gene has been identified to have three major transcripts (A, B, and C), derived from alternative splicing and usage of different promoter sites. The RASSF1 gene encodes for two major transcripts, A and C, both of which are found in multiple normal tissues and contain independent CpG-rich promoter regions (18) . In previous work, Dammann et al. (18) made the initial discovery showing that hypermethylation of the RASSF1A CpG island promoter region correlated with the loss of gene expression in lung cancer cell lines (18) . The RASSF1A gene promoter CpG island has since been shown to have aberrant hypermethylation in a significant percentage of carcinomas including breast, prostate, ovarian, kidney, gastric, bladder, and nasopharyngeal carcinomas (18, 19, 20, 21, 22, 23, 24, 25, 26) . RASSF1A methylation has also been reported in colon carcinoma but not at significant levels (27) .

Understanding the physiological role of RASSF1A, which encodes a predicted M_r 39,000 peptide, is still in its early stages. The role of RASSF1A protein as a tumor suppressor has been suggested through various functional studies (28) . It has been shown that introduction of RASSF1A into lung cancer cell lines lacking RASSF1A gene expression can reduce colony formation, growth in soft agar, and tumorigenesis in nude mice (18 , 22) . The RASSF1A protein has a predicted RAS association domain at its COOH terminus and is 55% homologous to the Ras effector Nore1 protein, which associates with Ras-like GTPases (28) . The RAS-binding domain binds RAS in a GTP-dependent manner (28) . The domain has been shown to mediate an apoptotic response within the Ras signaling pathway (29) . Recently, RASSF1A has been shown to heterodimerize with Nore1 (28) .

The RAS gene family plays a significant role in various cell growth events, whereby when disruption occurs of its function, it is usually via mutation(s). However, the frequency of Ras mutation in melanomas is low [less than 25%, depending on the assay and specimens sampled (30)] . In this study, we investigated whether the inactivation of a Ras effector pathway could be due to RASSF1A gene silencing through an epigenetic event such as aberrant hypermethylation in malignant cutaneous melanoma. We demonstrate that RASSF1A is significantly inactivated due to hypermethylation of its CpG island in melanoma cell lines and tumor biopsies.

MATERIALS AND METHODS

Cell Lines and Tissues.

Eleven established melanoma cell lines were cultured in culture medium and prepared for DNA extraction as described previously (31) . Melanoma tumors were obtained from elective surgery performed at John Wayne Cancer Institute (Santa Monica, CA). Institutional review board approval of human subjects was through Saint John’s Health Center and John Wayne Cancer Institute.

Methylation Analysis.

DNA was isolated from cell lines and tissues using DNAzol Genomic DNA Isolation Reagent (Molecular Research Center, Inc., Cincinnati, OH) according to the manufacturer’s recommendations. The methylation status of the RASSF1A promoter region was determined by a bisulfite modification protocol (32 , 33) . Briefly, 1 μg of genomic DNA was denatured in 0.3 m NaOH for 15 min at 37°C. Cytosines were sulfonated in the presence of 3.26 m sodium bisulfite (Sigma, St. Louis, MO) and 5 mm hydroquinone (Sigma) in a water bath for 16–18 h at 55°C. Thereafter, the DNA samples were desalted using the Wizard DNA Clean-Up System (Promega, Madison, WI) and desulfonated in 0.3 m NaOH at 37°C for 15 min. Finally, the treated DNA samples were precipitated with ethanol and resuspended in 10 mm Tris-Cl and 1 mm EDTA (pH 7.6). DNA sequences were amplified by mixing 100 ng of bisulfite-treated DNA with 50 pmol of primer MU379 (5′-GTTTTGGTAGTTTAATGAGTTTAGGTTTTTT) and 50 pmol of primer ML730 (5′-ACCCTCTTCCTCTAACACAATAAAACTAACC) in a 100-μl reaction buffer containing 200 μm each dNTP and Taq polymerase at 95°C for 1 min, 55°C for 1 min, and 74°C for 2 min for 30 cycles. Seminested PCR was performed using one-fiftieth of the amplified products and internal primer ML561 (5′-CCCCACAATCCCTACACCCAAAT) and primer MU379 with similar PCR conditions as described above. The PCR products were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA). Products were directly sequenced using an automated DNA sequencer. Primers used for bisulfite sequencing were outside of the MSP-analyzed CpG-containing sequences and assessed as described previously (18) .

For MSP, two methods were used to assess the different regions of the RASSF1A CpG promoter island. In promoter region 1, two sets of RASSF1A promoter-specific primers similar to those described by Burbee et al. (22) were used to specifically amplify methylated and unmethylated DNA sequences, respectively, after treatment with sodium bisulfite. Bisulfite-treated genomic DNA (100 ng) was amplified with methylated DNA-specific primers [M210 (5′-GGGTTTTGCGAGAGCGCG-3′) and M211 (5′-GCTAACAAACGCGAACCG-3′)] or unmethylated DNA-specific primers [UM240 (5′-GGGGTTTTGTGAGAGTGTGTTTAG-3′) and UM241 (5′-TAAACACTAACAAACACAAACCAAAC-3′)] in a 100-μl reaction volume containing 200 μm each dNTP and Taq polymerase. PCR was performed with an initial incubation for 15 min at 95°C and followed by 45 cycles of denaturation at 94°C for 30 s, annealing (64°C for M210/M211, 62°C for UM240/241) for 50 s, and extension at 72°C for 30 s. Fifteen μl of PCR products were resolved on 2% Tris-borate EDTA-agarose gel.

The methylation status of RASSF1A region 2 was assessed using two sets of fluorescent-labeled primers specific for methylated MF (5′-GTGTTAACGCGTTGCGTATC-3′) and MR (5′-AACCCCGCGAACTAAAAACGA-3′) and unmethylated UF (5′-TTTGGTTGGAGTGTGTTAATGTG-3′) and UR (5′-CAAACCCCACAAACTAAAAACAA-3′) (25) modified DNA sequences. One hundred ng of bisulfite-modified DNA were amplified in a final reaction volume of 20 μl containing 0.8 mm dNTPs and Taq polymerase. PCR was performed with an initial 10-min incubation at 95°C, followed by 40 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 30 s, and a final 7-min hold at 72°C. PCR products were visualized using capillary array electrophoresis (CEQ 2000XL DNA Analysis System; Beckman Coulter, Fullerton, CA). The assay was set up in a 96-well microplate format. Multiple PCR products can be run in each well for comparisons. Multiple PCR products were visualized simultaneously by labeling forward primers with a choice of three Beckman Coulter WellRED Phosphoramidite-linked dyes. Forward methylated specific primer was labeled with D4pa dye (blue), and forward unmethylated specific primer was labeled with D2pa dye (black). One μl of methylated PCR product and one μl of unmethylated PCR product were mixed with 40 μl of loading buffer and 0.5 μl of dye-labeled size standard (Beckman Coulter). Labeling forward primers specific for methylated or unmethylated modified DNA distinguishes the respective products so that they may be analyzed simultaneously.

RT-PCR Analysis.

Total cellular RNAs were extracted from cells using the Trizol reagent (Life Technologies, Inc., Rockville, MD). RT-PCR was essentially performed as described previously (34) . Briefly, 100 ng of RNA were preassociated with of a lower primer from exon 4 of the RASSF1A gene (5′-TCCTGCAAGGAGGGTGGCTTC). After the reverse transcription reaction, an upper primer from exon 1α (5′-TGGTGCGACCTCTGTGGCGACTT) was used in the PCR. PCR conditions were 95°C for 30 s, 60°C for 30 s, and 74°C for 1 min (25 cycles for RASSF1A, and 18 cycles for the glyceraldehyde-3-phosphate dehydrogenase gene). These cycle numbers were chosen because they were in the exponential range of product amplification. PCR products were separated on 2% Tris-borate EDTA-agarose gels, blotted, hybridized with a radiolabeled probe covering exon 3, and visualized by autoradiography.

Reexpression of the RASSF1A Gene.

Several cell lines were grown for 4 days in T75 cm² tissue culture flasks in the presence of 0, 2, 5, and 10 μm 5-aza-2′-deoxycytidine (Sigma). RNA was isolated, and RT-PCR was performed as described above to analyze for RASSF1A gene expression.

RESULTS

RASSF1A Gene Methylation.

In this study, we investigated the methylation status of two regions of the RASSF1A CpG promoter island (Fig. 1) ⇓ . Two sets of primers similar to those described by Burbee et al. (22) were used to determine the methylation status of region 1 of the RASSF1A CpG island. Region 1 is located upstream of the transcription start codon and contains three Sp1 consensus binding sites. Two sets of MSP primers were also designed to investigate the methylation status of region 2 of the RASSF1A CpG promoter island. Region 2 is located within the first exon (1α) specific to the RASSF1A transcript.

Initially, the methylation status of 11 melanoma cell lines was assessed by MSP. Hypermethylation of region 1 was found in 7 of 11 (64%) cell lines (Table 1 ⇓ ; Fig. 2A ⇓ ). Analysis of region 2 revealed aberrant methylation in 9 of 11 (82%) melanoma cell lines (Table 1) ⇓ . All melanoma cell lines found to be hypermethylated in region 1 were also methylated in region 2. Two of the cell lines that exhibited no methylation in region 1 demonstrated partial methylation in region 2. Overall, 82% of the melanoma cell lines were hypermethylated in either region 1 or 2, and 64% of the cell lines were methylated at both regions. Controls for the methylation studies included known methylated and unmethylated cell lines, as well as healthy donor peripheral blood lymphocytes. Expression of the RASSF1A gene was analyzed in seven of the melanoma cell lines by RT-PCR. No RASSF1A gene expression was detected in cell lines exhibiting complete methylation in both regions 1 and 2. RASSF1A gene expression was detected in the nonmethylated cell line as well as the two cell lines showing partial methylation in region 2 only. No methylation was detected in available normal skin tissues or healthy donor peripheral lymphocytes. Most normal tissues have been shown to express RASSF1A gene transcripts (18) .

We next assessed 41 AJCC stage III/IV melanoma patients from whom 44 metastatic melanoma tumors were available (Fig. 2B ⇓ ; Table 2 ⇓ ). MSP was used to investigate the methylation status of the RASSF1A gene in 44 stage III/IV melanoma tumors (9 were from stage III patients, and 32 were from stage IV patients). Hypermethylation of RASSF1A region 1 was found in 18 of 44 (41%) tumors, 3 of 9 (33%) stage III patients, and 14 of 32 (44%) stage IV patients (Table 2) ⇓ . Analysis of RASSF1A region 2 revealed aberrant methylation in 22 of 44 (50%) tumors, 2 of 9 (22%) stage III patients, and 18 of 32 (56%) stage IV patients (Table 2) ⇓ . Of the 44 melanoma tumors analyzed, methylation occurred in both regions in 16 (36%) tumors [2 (18%) stage III and 14 (42%) stage IV tumors]. Two tumors demonstrated methylation isolated to region 1 [one (9%) stage III and one (3%) stage IV tumor]. Six tumors showed methylation isolated to region 2 [one (9%) stage III and five (15%) stage IV tumors]. There was significant concordance between hypermethylation of both promoter regions (κ analysis, 0.66; P < 0.001). Overall, 55% of the 44 melanoma tumors examined by MSP exhibited hypermethylation within the RASSF1A CpG island regions.

Methylation Site Sequencing.

Bisulfite sequencing was carried out to confirm the methylation status of the CpG island promoter regions that regulate RASSF1A transcription. Extracted DNA was treated with sodium bisulfite, which converts unmethylated cytosines to uracil. Thymine is then substituted for uracil during subsequent PCR. Methylated cytosines (5-methylcytosine) are protected from this process and remain unchanged. Accordingly, all cytosines present after sequence analysis represent methylated cytosines (Fig. 3) ⇓ .

Reexpression of RASSF1A.

Treatment of cells with the DNA methylation inhibitor 5′-aza-2′-deoxycytidine can reverse epigenetic transcriptional silencing caused by methylation (35) . Melanoma cell lines (n = 4) that expressed no RASSF1A were treated with various concentrations of this methylation inhibitor (0, 2, 5, and 10 μm) for 4 days. The DNA methylation inhibitor treatment induced RASSF1A mRNA expression in all treated cell lines (Fig. 4) ⇓ . A dose-dependent effect was observed in activation of RASSF1A mRNA expression.

DISCUSSION

Recent studies have shown that aberrant promoter methylation is associated with loss of gene function that can promote and maintain cancer cell growth. The hypermethylation of promoter regions of TSGs is now one of the most well-categorized epigenetic events to occur in tumors (1 , 2 , 4) . Amazingly, this type of mechanism of silencing TSG function is frequently found in multiple tumors and is possibly more frequent than mutation-induced disruption of TSG function. There are an increasing number of tumor-related genes, including TSGs, that have been reported to be silenced via hypermethylation of CpG-rich promoter regions. In many situations, promoter region hypermethylation of tumor-related genes is the major mechanism whereby gene function is lost. Examples of this include the candidate TSGs O6-methylguanine-DNA methyl-transferase (MGMT) (4 , 12) , cyclin-dependent kinase inhibitor 2 (CDKN2) (4 , 10) , and, most recently, the newly discovered gene RASSF1A (18) . This field of identification of aberrant hypermethylation has developed rapidly due to the advancement of molecular techniques to specifically assess this epigenetic event using minimal amounts of tissue specimen DNA. Although cell lines are easily accessible, they often skew results and lead to false interpretations of “genotype reality” in in vivo situations. Thus, the assessment of tumors is essential to confirm the frequency of hypermethylation in vivo.

Several major cancers such colon, bladder, breast, and lung carcinomas have been studied for aberrant methylation of promoter regions of TSGs and tumor-related genes (4 , 5) . Studies of epigenetic inactivation of TSGs or tumor-related genes in cutaneous melanoma have been limited. To date, the most studied gene for aberrant promoter region methylation in melanoma has been p16^INK4a (10 , 11) . The studies have predominately been focused on cell lines. Epigenetic silencing of p16^INK4a and MGMT is <30%. Interestingly, for TSGs or oncogenes identified in melanoma tumors, the frequency of mutations is low compared with major known carcinomas. For example, for p16^INK4a, p15^INK4b, p14^ARF, p53, and Ras, the frequency of mutations in general is <25% in melanomas. These observations have been a major enigma to those that generalize the significance of inactivation of specific TSGs as universal mechanisms in neoplastic transformation and tumor development. This suggests that there are potentially other TSGs and tumor-related genes that are inactivated. In this study, we demonstrated that the newly identified TSG RASSF1A is significantly silenced in malignant cutaneous melanoma.

Our studies demonstrated that the RASSF1A gene was hypermethylated in 55% of the malignant melanoma tumors analyzed. This is, to date, one of the most significant epigenetic aberrations of a TSG reported in cutaneous melanoma. Also, this is the most significant loss of a TSG expression reported to date in cutaneous melanoma. The studies indicate that inactivation of RASSF1A may play an important role in the selective advantage of malignant melanoma cells. Tumors of epithelial origin have demonstrated high frequency of RASSF1A promoter region methylation. This includes breast tumors (62%; Ref. 19 ), small cell lung carcinomas (79%; Refs. 18 and 22 ), nasopharyngeal carcinomas (67%; Ref. 25 ), renal cell carcinomas (56%; Ref. 26 ), and prostate carcinomas (63%; Ref. 23 ). Recently, studies have shown that tumors of nonepithelial origin exhibit significant RASSF1A gene methylation, as was demonstrated in 55% of neuroblastomas (36) . It is clear that many of the cell lines and melanoma tumors we analyzed exhibit only methylated alleles (Fig. 2) ⇓ , indicating that epigenetic inactivation of RASSF1A may play an important role in melanoma. A common polymorphism is not known in the region analyzed, and thus we could not determine whether both maternal and paternal alleles were methylated. Because allele loss in 3p21.3 is rare in melanomas, the available evidence points to frequent biallelic methylation of RASSF1A. In many types of epithelial cancers, RASSF1A inactivation is accomplished via chromosomal deletions (LOH) and/or aberrant methylation patterns (18, 19, 20, 21, 22, 23, 24, 25, 26) . However, similar to melanoma, allelic losses at chromosome 3p21.3 are also rare events in medulloblastoma, and yet methylation of RASSF1A has recently been reported in over 80% of these latter tumors (37) . Our study supports that the loss of RASSF1A function is not just limited to epithelial origin tumors. These studies support the universal inactivation of the RASSF1A gene in a variety of tumors, not just carcinomas.

In the analysis of aberrant methylation of RASSF1A, we assessed two different regions of the RASSF1A CpG island. The assays established were highly specific using MSP as well as bisulfite sequencing. Both regions have been analyzed in various cancers, but usually only one for any cancer is assessed. The concordance of hypermethylation of RASSF1A region 1 and region 2 was significant. As expected, melanoma cell line hypermethylation of both regions 1 and 2 was high (64%) compared with tumor specimens (36%). Reexpression studies demonstrated that the RASSF1A transcript could be induced by incubation with a DNA methylation inhibitor. These functional studies support the significance of hypermethylation of the RASSF1A CpG island in malignant melanoma. Hypermethylation was higher in RASSF1A region 2 in both cell lines and tumors. Verification of loss of RASSF1A gene expression by aberrant hypermethylation was demonstrated by RT-PCR analysis for mRNA. These studies demonstrated that methylation of both RASSF1A regions 1 and 2 turned off gene function. It is possible that the hypermethylation event initiates in region 2 (exon 1α) and then spreads into region 1, the upstream promoter region, which ultimately results in gene silencing. Patient 9 (Table 2) ⇓ was of particular interest in that multiple lesions were assessed. Three independent consecutive surgical resections (a, b, and c) of an aggressively growing tumor in the patient were performed at different sequential time points, respectively. Analysis of specimens a and b demonstrated no methylation, whereas c (a later recurrence) showed methylation at region 2 only.

In general, it is known that the frequency of Ras mutation in melanoma is <25% (30) . It has been puzzling in malignant melanoma to date why disruption of Ras signaling is not a predominantly observed event. The high frequency of inactivation of RASSF1A in melanomas suggests that this may be an important alternative pathway for affecting Ras signaling. Recently, RASSF1A has been shown to negatively regulate cyclin D1 accumulation, thereby regulating cell cycle progression (38) . Additional studies are needed to identify epigenetic factors that effect methylation of the RASSF1A gene and its regulatory role(s) in cutaneous melanoma tumorigenesis and progression. At present, the mechanisms of de novo methylation of TSGs and other regulatory genes are poorly understood. How significant methylation of the RASSF1A gene plays in melanoma progression remains to be determined.

The disruption of p16^INK4a and p53 genes is infrequent in melanoma and appears likely not to play a major role in tumor development or progression. The RASSF1A TSG may have a potential role in tumor initiation and progression. The clinicopathology utility of RASSF1A silencing is unknown in many tumor systems to date. Recently, in bladder cancer, silencing of RASSF1A was shown to be related to poorer disease outcome in a retrospective analysis (24) . In our study, there was a trend of poorer survival in patients with aberrant methylation of RASSF1A, although the difference was not significant. Additional follow-up studies are needed. The elucidation of RASSF1A protein function in tumors may open up new strategies for molecular targeted therapy. Additional studies are needed to determine at what stage in melanoma tumor development RASSF1A is inactivated and whether this is an early-stage or later-stage epigenetic silencing event.