Identification of novel epigenetically modified genes in human melanoma via promoter methylation gene profiling


Address correspondence to Adam I. Riker, e-mail:


The inactivation of tumor-related genes through the aberrant methylation of promoter CpG islands is thought to contribute to tumor initiation and progression. We therefore investigated promoter methylation events involved in cutaneous melanoma by screening 30 genes of interest for evidence of promoter hypermethylation, examining 20 melanoma cell lines and 40 freshly procured melanoma samples. Utilizing quantitative methylation-specific PCR, we identified five genes (SOCS1, SOCS2, RAR-beta 2, TNFSF10C, and TNFSF10D) with hypermethylation frequencies ranging from 50% to 80% in melanoma cell lines as well as freshly procured tissue samples. Eighteen genes (LOX, RASSF1A, WFDC1, TM, APC, TFPI2, TNFSF10A, CDKN2A, MGMT, TIMP3, ASC, TPM1, IRF8, CIITA-PIV, CDH1, SYK, HOXB13, and DAPK1) were methylated at lower frequencies (2–30%). Two genes (CDKN1B and PTEN), previously reported as methylated in melanoma, and five other genes (RECK, IRF7, PAWR, TNFSF10B, and Rb) were not methylated in the samples screened here. Daughter melanoma cell lines showed identical methylation patterns when compared with original samples from which they were derived, as did synchronous metastatic lesions from the same patient. We identified four genes (TNFSF10C, TNFSF10D, LOX, and TPM1) that have never before been identified as hypermethylated in melanoma, with an overall methylation frequency of 60, 80, 50, and 10%, respectively, hypothesizing that these genes may play an important role in melanoma progression.


The number of melanoma cases worldwide is increasing faster than any other cancer (Lens and Dawes, 2004; Jemal et al. 2008). Patients diagnosed with metastatic melanoma (AJCC stage IV) have an overall poor prognosis, with few patients surviving longer than 2 years from the time of diagnosis (Balch et al., 2000). Most patients with advanced melanoma do not respond to available therapies, currently limited to only a few agents with limited activity for patients with stage III and IV melanoma. Development of more effective treatment options will require a greater understanding of the malignant transformation process and metastatic behavior of melanoma.

Irreversible changes that occur within the human genome, such as chromosomal deletions, amplifications, and gene mutations, have been shown to contribute to the development and progression of melanoma (Chin et al., 2006). Recently, there has been much emphasis on further identifying and analyzing the (epi)genetic events involved in melanoma progression and metastasis, including histone modification and DNA methylation. The molecular and cellular processes associated with promoter region CpG hypermethylation act as an alternate and/or complementary mechanism to gene deletion or mutation, resulting in the inactivation of specific gene expression and function. However, the aberrant hypermethylation of such genes is a ‘reversible’ phenomenon, unlike those gene changes seen with point mutations and chromosomal translocation events. Researchers continue to identify more genes that are epigenetically modified in melanoma, with a wide range of alternative gene functions, such as cell cycle control, apoptosis, cell signaling, tumor cell invasion and metastasis, angiogenesis, and immune recognition (Rothhammer and Bosserhoff, 2007). Such evidence further supports the notion that promoter region DNA methylation events of tumor-related genes play an important role in melanoma progression and metastasis. As methylation can be reversed by specific drugs, the use of DNA demethylating agents has been exploited to investigate their therapeutic ability to ‘re-activate’ such genes upon demethylation with subsequent gene expression contributing to the regression of established tumors (Appleton et al., 2007; Reu et al., 2006).

To date, several investigators have identified at least 50 genes that appear to be silenced secondary to promoter region hypermethylation (Rothhammer and Bosserhoff, 2007). In this study, we began by systematically examining the methylation events involved in advanced-stage, human melanoma samples, extensively screening the methylation status of the promoter regions of 30 ‘cancer-related’ genes utilizing a sensitive and quantitative methylation-specific PCR-based assay. The panel of genes was selected based upon their suspected involvement in key events in tumor cell progression and metastasis. They are either previously reported to be methylated in melanoma samples or other tumor histologies, exerting a significant functional impact upon gene re-activation. The majority of these genes have not been fully investigated in melanoma (Table 1). From this panel, we have identified five genes, SOCS1, SOCS2, RAR-β2, TNFSF10C, and TNFSF10D, which were found to have promoter regions with a high frequency of methylation in fresh melanomas tissue and cell lines compared with normal skin and cultured melanocytes. Seven genes (CDKN1B, PTEN, RECK, IRF7, PAWR, TNFSF10B, and nad Rb) that have previously been found to be highly methylated in other tumor histologies were not found to be significantly hypermethylated in the large panel of melanomas tissues and cell lines examined. Lastly, we have discovered the hypermethylation of several novel genes in melanoma, TNFSF10C, TNFSF10D, LOX, and TPM1, not previously reported and possibly implicating these genes as having an important role in tumor progression and metastatic potential of melanoma.

Table 1.   The proposed function of 30 genes in tumorigenesis and their methylation in tumors
GenesProposed function in tumorsReported methylation frequency in tumors
Genes involved in cell cycle and/or inducing apoptosis
 APCAdenomatous polyposis coli geneWNT signaling pathway antagonist, involved in cell proliferation and migration17% in melanoma (Worm et al., 2004)
 ASCApoptosis Speck-like protein containing a CARDPromote apoptosis directly by activation of downstream caspases50% in melanoma (Guan et al., 2003)
 CDKN2ACyclin-dependent kinase inhibitor 2ACell cycle inhibitor, inhibit cell proliferation10% in melanoma (Gonzalgo et al., 1997)
 CDKN1BCyclin-dependent kinase inhibitor 1BCell cycle inhibitor, inhibit cell proliferation6.5% in melanoma (Worm et al., 2000)
 DAPK1Death-associated protein kinase 1Positive mediators of apoptosis19% in melanoma (Hoon et al., 2004)
 TNFRSF10A (DR4) Tumor necrosis factor receptor superfamily, member 10aTransduces cell death signal and induces cell apoptosisMethylated in 2 melanoma cell lines (Bae et al., 2008)
 TNFRSF10B (DR5)Tumor necrosis factor receptor superfamily, member 10bTransduces cell death signal and induces cell apoptosis1% in prostate cancer (Suzuki et al., 2006);
 TNFRSF10C (DcR1)Tumor necrosis factor receptor superfamily, member 10c, decoy without an intracellular domainAntagonistic receptor that protects cells from TRAIL-induced apoptosis70% in breast cancer, 31% in lung cancer, 63% in malignant mesothelioma, 60% in prostate cancer, 42% in bladder cancer, 100% in cervical cancer, 43% in ovarian cancer, 41% in lymphoma,, 26% in leukemia, 56% in multiple myeloma (Shivapurkar et al., 2004)
 TNFRSF10D (DcR2)Tumor necrosis factor receptor superfamily, member 10d, decoy with truncated death domainAntagonistic receptor that protects cells from TRAIL-induced apoptosis
 PAWRPRKC, apoptosis, WT1, regulatorWT1-interacting protein, functions as a transcriptional repressor32% in endometrial cancer (Moreno-Bueno et al., 2007)
 PTENPhosphatase and tensin homologBlock cell cycle through negatively regulating the PI3K signaling pathway62% in melanoma (Mirmohammadsadegh et al., 2006)
 RASSF1ARAS association domain family protein 1ANegative regulator of cell proliferation through inhibition of G1/S-phase progression. Inactivation lead to resistance to chemotherapy and/or IFNs55% in melanoma (Spugnardi et al., 2003)
 RbRetinoblastoma geneCell cycle inhibitor, inhibit cell proliferation4% in brain tumor (Yin et al., 2002)
Genes involved cell adhesion, invasion, metastasis or angiogenesis
 CDH1E-cadherinIntercellular adhesion as well as a part of a complex signaling pathwayMethylated in 2 melanoma cell lines (Tsutsumida et al., 2004)
 RECKReversion-inducing-cysteine-rich protein with kazal motifsNegative regulator for matrix metalloproteinase-963.5% in lung cancer (Chang et al., 2007)
 TFPI2Tissue factor pathway inhibitor 2MMP inhibitor, suppress the invasiveness of tumor cells13.5% in melanoma (Nobeyama et al., 2007)
 TPM1Tropomyosin 1Involved in assembly and stabilization of actin filaments and control of cell motilityMethylated in breast and colon caner cell lines (Varga et al., 2005)
 TIMP3Tissue inhibitor of metalloproteinase 3MMP inhibitor, blocks the binding of VEGF to VEGFR2<15% in uveal melanoma (Moulin et al., 2008)
 TMThrombomodulinMediate cell adhension. Anticoagulant effect?60% in melanoma (Furuta et al., 2005)
Genes involved in immune response/interferon signaling pathway
 CIITA- PIVClass II transactivator, promoter IVInvolved in IFN-g inducible gene expressionMethylated in uveal melanoma
 IRF7IFN regulatory factor 7Regulated at the transcriptional level the expression of interferon genesMethylated in Hela cells (Lu et al., 2000)
 IRF8IFN regulatory factor 8Regulated at the transcriptional level the expression of interferon genesMethylated in colon carcinoma cell lines (Yang et al., 2007)
 HOXB13Homeobox B13Growth inhibition effects in melanoma cells20% in melanoma (Muthusamy et al., 2006)
 LOXLysyl oxidaseParadoxical role in tumor growth and metastasis27% in gastric cancer (Kaneda et al., 2004)
 RARβ2Retinoic acid receptor, beta isoform 2Limit cells growth by regulating gene expression.70% in melanoma (Hoon et al., 2004)
 MGMTO6-methylguanine DNA methylatransferaseMay associate with tumor responsiveness to alkylating agents34% in melanoma (Hoon et al., 2004)
 SOCS-1Suppressor of cytokine signaling 1Attenuators of cytokine-mediated processes75% in melanoma (Marini et al., 2006)
 SOCS-2Suppressor of cytokine signaling 2Attenuators of cytokine-mediated processes43% in melanoma
 SYKProtein-tyrosine kinase SYKGrowth inhibition effects in melanoma15% in melanoma (Muthusamy et al., 2006)
 WFDC1Wap 4-disulfide core domain 1Growth inhibition?5% in melanoma (Muthusamy et al., 2006)


Methylation profiling of human melanoma cell lines and epidermal melanocytes

A total of 30 genes were analyzed for evidence of aberrant methylation of CpG promoter regions, utilizing two normal human epidermal melanocyte (NHEM) cell lines, 17 daughter melanoma cell lines and three well-established melanoma cell lines (A375-Mel, Lox and C8161.9). Samples with a NIM ≥ 10% were defined as being methylated (see details in Material and methods). We identified SOCS1, SOCS2, RAR-beta 2, TNFSF10C and TNFSF10D as the most frequently methylated genes in the panel of samples examined, with a specific frequency of 90, 80, 60, 60, and 85%, respectively (Table 2). A panel of other genes (RECK, IRF7, PAWR, TNFSF10B, Rb, CDKN1B and PTEN) were found not to be methylated in any of the cell lines investigated. The methylation events involved in individual melanoma cell lines varied significantly, ranging from only 3/30 (10%) genes to as high as 19/30 (63%) genes methylated. The cell lines revealed a similar pattern of hypermethylation, which included synchronous sample pairs from the same patient such as MCC12A/MCC12F, MCC80a/MCC80b, and MCC74/MCC81. When comparing NHEM with melanoma samples, none of the examined promoter regions were found to be methylated in NHEM, except for a single gene, WFDC1, which was methylated at a low level in one NHEM cell line but not in a second.

Table 2.   Summary of promoter hypermethylation in cell lines as detected by Q-MSP Thumbnail image of

Methylation profiling of freshly procured melanoma specimens

We next analyzed 40 freshly procured and cryopreserved melanoma tumor specimens from patients with stage III and IV melanoma. Again, samples with a NIM ≥ 10% were defined as being methylated. Similar to the results obtained from the panel of melanoma cell lines, 18/30 (60%) promoter CpG islands regions were identified as hypermethylated with varying levels of frequency. The most frequently hypermethylated genes identified were SOCS1, SOCS2, RAR-ß2, TNFSF10C, and TNFSF10D (Table 3). While other genes, including HOXB13, CDH1, DAPK, CDKN1B, Rb, PTEN, CIITA-PIV, TNFSF10B, IRF7, IRF8, PAWR, and RECK, were found not to be methylated in any of the screened melanoma samples. The comparative frequency of gene methylation between cell lines and melanoma tissues shows a highly correlative pattern of methylation (Figure 1). Metastatic synchronous samples derived from the same patient showed similar methylation profiles, with little, if any, correlation between primary and metastatic samples from various sites. The total number of methylated genes within individual tumor samples varied from 0 to 12 genes, with most samples showing evidence of approximately 3–10 methylated genes. A total of five melanoma samples were found to have >10 significantly methylated genes, with five other samples showing only a low frequency of promoter methylation (Table 3). A parallel direct comparison of five original tumor samples and derived cell lines shows a similar pattern of gene methylation, with little suggestion of artifactual gene methylation change related to in vitro culturing.

Table 3.   Summary of promoter hypermethylation in clinical melanoma tissue specimens as detected by Q-MSP Thumbnail image of
Figure 1.

 Promoter methylation frequencies in melanoma tissues and cell lines. Samples with NIM ≥ 10% were designated as methylated ones. The promoter methylation frequencies were calculated as proportion of methylated samples of all the samples detected. The comparative frequencies of gene methylation between cell lines and melanoma tissues show a highly correlative pattern of methylation.

Correlation of promoter methylation status with melanoma gene expression

Our screening of 30 genes yielded four genes (TNFSF10C, TNFSF10D, TPM1, and Lox) not previously identified as being hypermethylated in melanoma. We then evaluated the expression of DcR1, DcR2, and TPM1 utilizing reverse transcription PCR in a panel of 15 well-established melanoma cell lines and correlated gene expression with promoter hypermethylation. We were able to detect all three gene transcripts within NHEM samples. Gene expression was also detected in the un-methylated cell lines and in some partially methylated cell lines. No gene transcripts were detected in cell lines exhibiting high level of promoter methylation (Table 4).

Table 4.   mRNA expression of DcR1, DcR2, and TPM1 in NHEM and 15 melanoma cell lines as evaluated by reverse transcription PCR
Cell linesDcR1DcR2TPM1
mRNA expressionPromoter methylationmRNA expressionPromoter methylationmRNA expressionPromoter methylation
  1. Promoter methylation status of the three genes was evaluated by Q-MSP (M, methylated; U, unmethylated).


The gene expression of LOX mRNA in cell lines was evaluated by real-time reverse transcription PCR (Figure 2). Compared with NHEM, both upregulation and downregulation of LOX mRNA expression were observed in melanoma cell lines. In melanoma cell lines with no evidence of promoter methylation, equal or significantly increased expression of LOX mRNA was observed compared with NHEM. LOX mRNA was also detectable in melanoma cell lines with partial promoter hypermethylation, except in cell line MCC067. Conversely, LOX gene expression was silenced in melanoma cell lines MCC005, TC072, A375, and Lox, which possessed a high level of promoter methylation, with a exception of MCC069B. On the whole, cell lines with un-methylated promoters express higher level of LOX mRNA than cell lines with methylated promoters. These results indicate that promoter methylation is one of the mechanisms that actively regulate the LOX mRNA expression levels in melanoma cells.

Figure 2.

 Relative quantitation of LOX mRNA expression in NHEM and 16 melanoma cell lines as evaluated by real-time RT-PCR. LOX promoter DNA methylation status was evaluated by quantitative MSP (Black boxes indicate promoter methylation with NIM ≥ 25%. Gray boxes represent NIM between 10% and 25%. White box represent NIM < 10%).

Reversal of gene silencing following demethylation of promoter region CpG islands

To further establish the relationship between promoter methylation and downregulation of DcR1, DcR2, TPM1, and LOX mRNA expression, we tested whether demethylation could reverse the suppression of these genes in melanoma cell lines. We treated the well-established metastatic melanoma cell lines Lox and C8161.9 with the demethylating agent, 5-aza-2-deoxycytidine (DAC), or with the histone deacetylase inhibitor (HDACI), trichostatin A (TSA). Utilizing quantitative methylation specific PCR (Q-MSP), we found a significant decrease in the level of promoter region methylation for those cells treated with DAC for all four genes tested (DcR1, DcR2, TPM1, LOX; Figure 3). We did not find a significant change in methylation status for the same cell lines treated with TSA only.

Figure 3.

 Melanoma cell line Lox was treated by DAC or TSA. After the treatment, genomic DNA was isolated and promoter methylation levels for genes TNFSF10C, DcR2, TPM1 and LOX were evaluated by Q-MSP. M: universally methylated human genomic DNA as positive control.

We then examined the same melanoma cell lines for evidence of re-expression of DcR1, DcR2, TPM1, and LOX following treatment with DAC or TSA. We found a direct and significant correlation with gene expression for all four genes upon demethylation with DAC, but no significant difference with TSA treatment (Figure 4). In the case of LOX gene, we examined its mRNA expression in three melanoma cell lines (MCC005, Lox, and C8161.9), before and after treatment with DAC or combined with TSA. LOX mRNA expression was not detectable in melanoma cell lines MCC005 and Lox, and was detectable in C8161.9 prior to DAC treatment. After DAC treatment, LOX mRNA was re-expressed in MCC005 and Lox and significantly upregulated in C8161.9. We did not observe significant synergy when DAC and TSA were combined (Figure 5). These in vitro studies are suggestive that hypermethylation of the promoter region of these genes is an active mechanism of silencing gene expression. Bisulfite sequencing of MSP products was also carried out to confirm the methylation status of these genes in melanoma cell lines. Representative examples of genomic sequencing results from the melanoma cell line, Lox, are shown in Figure 6, which demonstrate all CpG sites contained in the LOX MSP and TPM1 MSP products as methylated.

Figure 4.

 DcR1, DcR2, and TPM1 mRNA expression in two melanoma cell lines, Lox and C8161.9, as evaluated by RT-PCR. DcR1 was silenced in both Lox and C8161.9, while DcR2 and TPM1 were only silenced in Lox but expressed in C8161.9, consistent with their promoter methylation status. Only DAC treatment, but not TSA treatment, can induce gene re-expression in corresponding melanoma cell lines.

Figure 5.

 LOX mRNA was detectable in melanoma cell line C8161.9, but not in melanoma cell lines MCC005 and Lox. After DAC treatment, LOX mRNA expression was re-expressed in MCC005 and Lox, and significantly upregulated in C8161.9.

Figure 6.

 Representative PCR analysis of promoter region CpG island sequence of TPM1 and LOX from bisulfite-treated DNA obtained from melanoma cell line Lox.
Reported transcriptional startside was regarded as +1.


It is very important to continue the search for genes involved in the process of malignant transformation and progression of melanoma, as this may provide us with novel targets for therapy. We have aimed to illustrate that key gene promoter regions are hypermethylated in melanoma samples, able to be reversed effectively with an agent that is already in clinical use in human cancers. Although we only have a primitive understanding of human epigenetics in cancer, we must greatly enhance our current understanding of the complex mechanisms involved in global demethylation and regional hypermethylation in the genome of cancer cells.

Our current understanding as to the exact function of many tumor suppressor genes (TSG) is limited, however it appears that TSG expression play a key role in maintaining normal homeostatic cellular mechanisms and further providing control over cellular proliferation and metastasis in several different tumor histologies. We are quickly discovering many such ‘gene methylation markers’, however, few such markers are found with enough frequency and specificity to have any major impact upon predicting clinical outcomes in cancer patients (Mori et al., 2005). Furthermore, it remains unclear as to the specific genes involved in this process and additionally complicated by the lack of specificity of demethylating agents, often able to globally demethylate the human genome.

In this study, we have comprehensively examined the methylation profile of a selected group of 30 genes, chosen for their presumed roles in tumor progression and metastasis. Our current analysis of CpG island promoter regions in melanoma samples reveals three main categories of gene methylation: high frequency (SOCS1, SOCS2, RAR-beta 2, TNFSF10C, and TNFSF10D), intermediate to low frequency (ASC, APC, CAKN2A, MGMT, WFDC1, TNFSF10A, TM, TIMP3, TFPI2, LOX, IRF8, CIITA-PIV, and TPM1), and absence of methylation (CDKN1B, PTEN, RECK, IRF7, PAWR, Rb, and TNFSF10B). Although CDKN1B and PTEN have previously been reported as hypermethylated in melanoma (Mirmohammadsadegh et al., 2006; Worm et al., 2000), we were not able to corroborate these results in our melanoma panel. However, our results are in agreement with those obtained by Furuta et al. (2004), who also failed to detect methylation of CDKN1B and PTEN in all of the melanoma samples in their screening panel. The MSP primers for CDKN1B were designed within the promoter region which was previously analyzed in melanoma by methylation-specific sequencing (Worm et al., 2000). And the MSP primers for PTEN were utilized as reported, which were designed in a promoter region with scarce homology with the psiPTEN sequence (García et al., 2004). Thus, the MSP primers used here to interrogate the promoter methylation status of CDKN1B and PTEN are indeed consistent with the reported literature. The discrepancy of methylation frequency observed here is most likely not due to the wide range of possible CpG sites, but rather indicate that promoter methylation of CDKN1B and PTEN is a relatively infrequent event in melanoma. Indeed, Worm et al. (2000) examined the methylation status of the CDKN1B promoter region, detecting only four of 61 melanoma tissue samples as methylated. Lastly, we examined five genes (RECK, IRF7, PAWR, Rb, and TNFSF10B) that have been previously identified as methylated in non-melanoma tumor types, finding that these five genes were not methylated in the human melanoma samples examined in this study.

The gene, TPM1, has been previously identified as a TSG capable of inhibiting tumor metastasis in breast cancer, whose loss of expression has been shown to be the result of promoter DNA hypermethylation (Varga et al., 2005). In support of this, we have also found a high level of promoter region hypermethylation for the TPM1 gene in both melanoma cell lines and tissues. In the highly metastatic melanoma cell line, Lox, a high level of TPM1 promoter methylation is associated with silencing of TPM1 mRNA expression. We further show that upon demethylation, significant TPM1 mRNA re-expression can occur while the acetylation inhibitor, TSA, is unable to induce TPM1 mRNA expression. Silencing of the TPM1 gene by promoter hypermethylation, albeit at a lower frequency (10%), strongly implicates TPM1 as a potential TSG in melanoma. Recent evidence suggests that promoter methylation may not be a dominant mechanism for the loss of TPM1 expression and function, with TPM1 identified as a target of the microRNA, MIRN21, which functions as an oncogene in breast cancer (Zhu et al., 2007). Here, our analysis reveals that TPM1 mRNA can readily be detected in the majority of melanoma cell lines. Thus, it is tempting to speculate that other epigenetic events, such as microRNA modulation, may also be involved in the loss of function of TPM1 in melanoma. Further analysis of TPM1 protein expression and the interactions of microRNA’s are ongoing to address the exact function of TPM1 in melanoma.

A second newly identified methylation target in melanoma is the gene, lysyl oxidase (LOX), found secondary to a high frequency of promoter region hypermethylation in the majority of melanoma samples. Paradoxically, LOX gene expression is associated with both tumor suppression and tumor progression, with the LOX protein involved in vital biological processes such as cell migration, signal transduction and gene regulation (Payne et al., 2007). Indeed, loss of LOX gene expression in numerous cancers, such as gastric, colon, lung and ovarian cancer, in addition to several oncogene-transformed cell models has implicated LOX as a strong candidate TSG (Kaneda et al., 2004). The tumor suppressive activities of LOX were also recently demonstrated in both lung and pancreatic cancer cell lines, where Bcl-2 was identified as an essential target of LOX-mediated inhibition of the transformed phenotype (Wu et al., 2007). Re-express of LOX can effectively inhibit invasive phenotype of lung and pancreatic cancer cells. Min et al. (2007) also showed that in breast cancer, LOX propeptide (LOX-PP) was able to inhibit the transformation of breast cancer cells driven by Her-2/neu, further inhibiting Her-2/neu tumor formation in a nude mouse xenograft model. Conversely, LOX overexpression in tumors has also been observed, primarily in hypoxic tumor cells, implicating LOX as a metastasis promoting gene (Erler et al., 2006). These results create a conundrum within the LOX research field, with our data showing that the LOX gene appears to be hypermethylated in the majority of melanoma samples, with the functional significance unknown at this point.

In the case of human melanoma, both upregulation and downregulation of LOX gene expression have been reported by several investigators (Kirschmann et al., 2002; Kuivaniemi et al., 1986). Consistent with previous findings, we observed both reduced and increased expression of LOX mRNA in different melanoma cells lines when compared with NHEM (Figure 2). Melanoma cell lines with no or only partial promoter methylation demonstrated higher or equal levels of LOX expression compared to NHEM. Cell lines with heavily methylated promoter regions revealed a reduced or complete loss of LOX expression, with subsequent DAC treatment leading to the significant induction of LOX gene re-expression. These results indicate that promoter methylation is an important mechanism of LOX gene downregulation in melanoma samples. Other mechanism, such as loss of heterozygosity (Kaneda et al., 2004), might also contribute to the downregulation of LOX gene expression in melanoma. In particular, we observed undetectable levels of LOX mRNA for the cell line MCC067, finding only a partial methylation within the promoter regions.

An interesting phenomenon that we observed was that some melanoma cell lines with no promoter methylation seemed to express significantly higher levels of LOX mRNA, with other melanoma cells with partial promoter region methylation still expressing almost equal levels of LOX mRNA compared with NHEM. One melanoma cell line, MCC069B, expressed considerable level of LOX mRNA despite the heavily methylated promoter. This observation indicates that other mechanisms are involved with the regulation of LOX gene expression in melanoma cells. It has been reported that hypoxia can significantly induce LOX mRNA expression through transcription factors, such as the hypoxia-inducible factor (HIF), in hypoxic tumor cells (Erler et al., 2006). We speculate that this may also be occurring in melanoma cells. High activity of transcription factors may induce significant LOX mRNA transcription from the unmethylated alleles in melanoma cells. Till now, no evidence of gene mutations, amplifications or polymorphisms has been reported to be involved with the regulation of LOX gene expression. Transcriptional factors and promoter region hypermethylation may together result in the upregulation or downregulation of the LOX gene in melanoma cells. And the functional impact of LOX in melanoma cells may dependent on specific cellular location or transformation status (Erler et al., 2006).

Although re-activation of TSG and inhibition of tumor growth by demethylation agents are known to occur, there are always concerns in regards to the lack of specificity of current demethylating agents, resulting in the global hypomethylation of the human genome. Specifically, it may be possible to not only re-express previously silenced TSG, but also to activate ‘tumor-promoting genes’. This is exemplified in the current study, examining the death receptor superfamily, comprised of four receptors, DR4, DR5, DcR1, and DcR2, all capable of binding to the Tumor Necrosis Factor-Related Apoptosis-Inducing Ligand, or TRAIL. Whereas DR4 and DR5 are pro-apoptotic death receptors, the DcR1 and DcR2 receptors act as dominant-negative receptors for TRAIL, functioning as anti-apoptotic decoy receptors (Sheikh and Fornace, 2000). Most non-neoplastic cells express all four receptors and this balance prevents TRAIL-induced apoptosis.

It appears that most tumor cells lose DcR1 and DcR2 expression, resulting in the selective cytotoxicity of TRAIL upon tumor cells. Loss of DcR1 and DcR2 expression by promoter methylation has been shown in neuroblastoma, where downregulation of DcR1 seems correlated with TRAIL sensitivity of tumor cells (van Noesel et al., 2002). Our data here demonstrates that promoter hypermethylation of DcR1 and DcR2 is also a frequent event in metastatic melanoma, with methylation frequencies of 55% and 85%, respectively, subsequently leading to their reduced or complete inhibition in melanoma cells. The DR4 receptor was found to be methylated in only 5/40 (10%) melanoma samples, while none were found to have evidence of DR5 hypermethylation. These results show that TRAIL decoy receptor genes are much more frequently methylated than death receptors in melanoma, as is also the case in multiple other tumor types (Shivapurkar et al., 2004).

It has been shown that expression of DR4 by melanoma cells is critical to maintain their sensitivity to chemotherapy and/or immunotherapy (Bae et al., 2008), The overexpression of decoy death receptors (DcR1, DcR2) by tumor cells leads to a markedly reduced sensitivity to chemotherapy (Liu et al., 2005). Thus, the downregulation of the DcRs through promoter hypermethylation in tumor cells appears to render tumor cells more susceptible to TRAIL and chemotherapy induced apoptosis. This may represent a ‘physiologic’ protective response against tumor formation or progression (van Noesel et al., 2002). Nevertheless, due to the high frequency of methylated DcRs when compared with death receptors, our data strongly suggests that the demethylation of melanoma cells may be more likely to lead to the selective re-activation of DcR1 and DcR2 gene expression, rather than DR4 and/or DR5. This, in turn, may lead to the potential reduction in sensitivity of melanoma cells to select chemotherapeutic and immunotherapeutic agents. Hence, it will be crucial to gain a full understanding of the methylation events involved in melanomas before de-methylation agents can be properly utilized as effective anti-cancer therapies.

In conclusion, more genes are being identified as hypermethylated in human melanoma. Identification of novel methylation targets will greatly improve our current knowledge related to the complex process of melanoma tumorigenesis, progression and metastasis. Recent data suggest that methylation events within the human genome greatly contribute to melanoma initiation and progression, with their reversibility with demethylating agents a clear example of gene function regained. However, the lack of overall specificity of demethylation and the lack of understanding of methylation at the global level of the human genome limit the therapeutic utility of such agents. It is postulated that select patients with a ‘proper’ CpG island methylator phenotype (CIMP; Issa, 2004) may be the most likely to benefit from demethylation agents. Future treatment regimens involving the use of demethylating agents will most likely involve the accurate screening and measurement of multiple methylation events within the human genome, using Q-MSP in the sequentially procured tumor biopsies of individual patients prior to, during and after treatment. This will allow for the sequential analysis of the specific CIMP profile for each patient, further defining each patient’s response to demethylation therapy.

Materials and methods

Cell culture

Freshly excised melanoma samples were procured at the time of surgical resection, placing a small piece of macrodissected melanoma into standard cell culture media (RPMI 1640 + 5% FCS). We utilized previously published techniques for tissue procurement and in vitro melanoma cell line growth and expansion (Riker et al., 1999; Riker, 2004). All melanoma cell lines were grown in monolayer culture in RPMI 1640 with glutamine supplemented with 10% fetal bovine serum at 37°C in a humidified atmosphere consisting of 5% CO2 and 95% air. Two normal human neonatal melanocyte (NHEM; Cambrex, Baltimore, MD, USA) cell lines were cultured under the same incubator conditions but in specialized melanocyte media (Cambrex) according to the supplier’s recommendations. All established melanoma cell lines were further characterized by flow cytometry and/or cytospin preparation for cellular confirmation of melanoma cell purity of >99% (data not shown). The cell lines, MCC005, MCC13, MCC70, and MCC80a were derived from primary melanoma samples, while MCC80b was derived from a metastatic lymph node (from the same patient as MCC80a). The cell lines, MCC12A and MCC12F, were derived from two different synchronous subcutaneous metastatic melanoma nodules from the same patient. The cell line MCC74 was derived from a single lymph node replaced by melanoma, with MCC81 originally procured from an isolate melanoma metastasis to the brain from the same patient. The other 6 cell lines, MCC46, MCC66c, MCC69B, MCC72, MCC83, and MCC89, were all derived from bulky lymph nodes replaced with melanoma. Three human melanoma cell lines, A375 (American Type Culture collection; Manassas, VA, USA), Lox (established from a lymph node metastasis of a patient at the Norwegian Radium Hospital; Fodstad et al., 1988) and C8161.9 (a highly metastatic, amelanotic melanoma cell line derived from an abdominal wall metastasis; Welch et al., 1991, 1994), were also routinely utilized for many of the experiments.

Tumor procurement

Over a 3-year period, we surgically procured tumor samples from patients with primary cutaneous melanoma (PM) and metastatic melanoma (MM). All samples were obtained under an Investigational Review Board (IRB) approved tissue procurement protocol (MCC#13448, IRB#101751; PSM# 990914-JM, and 020318-JM). Upon surgical removal of the primary melanoma, a single surgical oncologist (A.I.R.) utilized a scalpel to macrodissect and procure a portion of the remaining primary tumor, with a similar technique utilized for grossly involved lymph nodes where the melanoma had completely replaced the lymph node. Samples were taken from non-necrotic areas of the tumor. The same process was performed for all distant metastases, careful to avoid surrounding tissues or stroma. All samples were cryopreserved in liquid nitrogen and stored within the Tissue Procurement Laboratory of the Moffitt Cancer Center, securely de-identified through a centralized database. We analyzed 8 thick PM’s (>4 mm), 32 MM samples, composed of 20 bulky, macroscopic (replaced) lymph node metastases, 10 subcutaneous, and two distant metastases (adrenal and brain). All MM samples were procured from patients that had failed multiple previous therapies, ranging from single agent Interferon, single or multi-agent chemotherapy, immunotherapy, biochemotherapy and/or other experimental treatment options. All primary cutaneous cancers were procured from previously untreated patients. Additionally, we included three samples of pathologically normal human skin samples.

5-aza-2-deoxycytidine and trichostatin A treatment of cells

Melanoma cells were seeded at a density of 5 × 105 cells/10 cm plate on day 1. Cell lines were treated with 5 μM DAC (Sigma, St Louis, MO, USA) for 96 h or 200 ng/mL trichostatin A (TSA; Sigma) for 24 h. For the combination treatment, cells were first treated with 5 μM DAC (Sigma) for 96 h, then 200 ng/mL TSA were added and cells were treated for an additional 24 h. Culture medium containing DAC or TSA was prepared fresh daily. At the end of treatment, the medium was removed, and the RNA and DNA were isolated for subsequent reverse transcription PCR and bisulfite analysis.

Nucleic acid isolation and bisulfite modification

Genomic DNA (gDNA) was isolated from tumor and skin samples, as well as cultured cell lines, using a Qiagen Blood and Cell Culture DNA kit (Qiagen, Inc., Valencia, VA, USA) and stored at −20°C before use. gDNA (0.5 μg) was used for bisulfite treatment with the EZDNAMethylation kit (Zymo Research, Orange, CA, USA) according to the supplier’s protocol.

Real-time quantitative methylation-specific PCR (Q-MSP)

Methylation of the promoter regions was determined through bisulfite modification of gDNA (Clark et al., 1994). Promoter methylation of human genes was investigated utilizing quantitative, methylation-specific PCR (Q-MSP; Herman et al., 1996). Briefly, bisulfide-treated DNA was amplified using real-time PCR with oligonucleotide primers complementary to a region of the β-actin sequence that does not contain CpG dinucleotides, but contains non-CpG cytosines. From this, we were able to ascertain the amount of converted input DNA template for each sample. Hypermethylation of the 30 CpG islands was then examined with real-time PCR amplification of bisulfite-modified DNA using oligonucleotide primers specific for methylated, bisulfide-converted regions of each CpG island. Primer oligonucleotide sequences, their position relative to transcriptional start site and PCR annealing temperatures are listed in Table 5. The MSP primers used were either reported previously or re-designed utilizing the software, ‘MethPrimers’ (, if no reported MSP primers are available, or if the reported primer pairs were not suitable for Q-MSP analysis. The newly designed primers covering the similar promoter regions of individual genes were previously interrogated. The performance of the Q-MSP primers was then evaluated by performing a standard curve and melting curve before they were applied for gene methylation analysis in tumor samples. All PCR reactions were carried out on an iQ5 real-time thermal cycler (Bio-Rad, Hercules, CA, USA) at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C or 64°C for 60 s. A subsequent dissociation curve analysis verified the specificity of each amplified product. Each PCR reaction was carried out in a 25 μl volume containing 2× SYBR Green PCR master mix (Applied Biosystems, Foster City, CA, USA), 200 nM each primer and 1 μl bisulfite-modified genome DNA. All reactions were carried out in duplicates, in order to diminish intersample variability. Bisulfide-converted, universally methylated human gDNA (Chemicon, Temecula, CA, USA) served as a positive control and bisulfide-converted, universally un-methylated human gDNA (Chemicon) and blank reactions with water served as negative controls. The normalized index of methylation (NIM) was defined as the ratio of the normalized amount of methylated templates at the target promoter to the normalized amount of converted β-actin templates in any given sample (Yegnasubramanian et al., 2004). A NIM cut-off of 10% was utilized to distinguish methylation positive (NIM ≥ 10%) from negative (NIM < 10%) samples. This was determined due to high level of correlation with the loss or reduced expression of corresponding mRNA in the present assay conditions (data not shown).

Table 5.   Primers for Q-MSP based on SYBR green
GeneForward primersReverse primersReferences or position AT (°C)
  1. Reported transcriptional start site was regarded as +1 concerning the primers positions. AT, annealing temperature.


Genomic sequencing

The MSP products from melanoma cell lines for DcR1, DcR2, TPM1, and LOX were gel purified using QIAquike Gel Extraction Kit (Qiagen, Valencia, CA, USA). Then the PCR fragments were ligated into pCR-TOPO Vectors (Invitrogen, Carlsbad, CA, USA) and transformed into DH5a-competent cells (Invitrogen). Four colonies were chosen randomly. Plasmid DNA was purified with Invitrogen mini plasmid Prep kit. Sequencing was carried out by MWG Biotech, Inc. (Huntsville, AL, USA).

RNA isolation and reverse transcription PCR

Cultured normal melanocytes and melanoma cells lines were collected and dissolved in TRIzol® (Invitrogen) and RNA was purified according to the manufacturer’s recommendations. First-strand cDNA synthesis was performed with 2 μg total RNA for each sample in a total volume of 20 μl using a high capacity cDNA archive kit (Applied Biosystems). The reverse transcription reaction was performed with random primers and incubated at 25°C for 10 min followed by 37°C for 120 min. RT-PCR analysis of DcR1, DcR2, and TPM1 transcripts was carried out utilizing the following primers: DcR1 forward, CCCAAAGACCCTAAAGTTCGTC, and reverse GCAAGAAGGTTCATTGTTGGA; DcR2 forward, ACCCCAAGATCCTTAAGTTCG, and reverse CAAGAAGGCAAATTGTTGGAA (van Noesel et al., 2002). TPM1 forward, GCTGGTGTCACTGCAAAAGA, TPM1 reverse: CTGCAGCCATTAATGCTTTC (Varga et al., 2005). β-actin was used as an equal loading control. All PCR reactions were performed in a 25 μl PCR reaction containing 2× GoTaq Green master mix (Promega, Madison, WI, USA), 1 μl of cDNA and 300 nM each specific primers. PCR conditions were as follows: one cycle, 2 min/95°C; 35 cycles, 30 s/95°C, 30 s/60°C, and 30 s/72°C; and 1 cycle, 10 min/72°C. PCR products were loaded on a 2% agarose gel stained with ethidium bromide and directly visualized under UV illumination. LOX transcripts were analyzed by real-time PCR using Assays-on-Demand Gene Expression Assays (Applied Biosystems; assay ID: Hs00184700_m1), with GAPDH (assay ID Hs99999905_m1) serving as an internal control. All PCR reactions were performed in a total volume of 25 μl, containing 2× TaqMan Universal PCR Master Mix (Applied Biosystems), 20× Assays-on-Demand Gene Expression Assay Mix and 40 ng cDNA. All assays were performed in triplicate and run on an iQ5 real-time thermal cycler (Bio-Rad) using the following conditions: 50°C for 2 min, 95 °C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. Relative quantitation of the amplified products was based upon Ct values.


This research was supported by the Abraham A. Mitchell Clinical Research Scholarship (A.I.R.).