Down-regulation of Rap1GAP via Promoter Hypermethylation Promotes Melanoma Cell Proliferation, Survival, and Migration

Melanoma is the most serious, highly aggressive form of skin cancer with recent dramatic increases in incidence. Current therapies are relatively ineffective, highlighting the need for a better understanding of the molecular mechanisms contributing to the disease. We have previously shown that activation of Rap1 promotes melanoma cell proliferation and migration through the mitogen-activated protein kinase pathway and integrin activation. In the present study, we show that expression of Rap1GAP, a specific negative regulator of Rap1, is decreased in human melanoma tumors and cell lines. Overexpression of Rap1GAP in melanoma cells blocks Rap1 activation and extracellular signal-regulated kinase (ERK) phosphorylation and inhibits melanoma cell proliferation and survival. In addition, overexpression of Rap1GAP also inhibits focal adhesion formation and decreases melanoma cell migration. Rap1GAP down-regulation is due to its promoter methylation, a mechanism of gene silencing in tumors. Furthermore, treatment of melanoma cells with the demethylating agent 5-aza-2'-deoxycytidine reinduces Rap1GAP expression, followed by decreased Rap1 activity, ERK phosphorylation, and cell proliferation and survival—changes that are significantly blunted in cells transfected by small interfering RNA–mediated Rap1GAP knockdown. Taken together, our findings indicate that down-regulation of Rap1GAP via promoter hypermethylation promotes melanoma cell proliferation, survival, and migration. [Cancer Res 2009;69(2):449–57]

Melanoma is the most serious, highly aggressive form of skin cancer whose incidence has increased at a rate greater than that of any other cancer type in the United States.⁵

Recently, targeted melanoma therapies have been developed based upon molecular insights into tumor pathogenesis (such as receptor kinase inhibitors) but have shown relative ineffectiveness to date when used as a monotherapy (1). The lack of therapeutic response to currently available treatment highlights the importance of improved understanding of the complex molecular mechanism(s) that contribute to melanoma development.

The RAS/BRAF/extracellular signal-regulated kinase (ERK) pathway plays a central role in the development of most types of melanoma (2–4). We previously reported that activation of Rap1, a member of the Ras family of small GTPases, promotes melanoma cell proliferation and migration through the mitogen-activated protein kinase (MAPK) pathway and integrin activation, highlighting an oncogenic role of Rap1 in melanoma (5). Like other small GTPases, Rap1 functions as a molecular switch cycling between an active GTP-bound form and an inactive GDP-bound form. Rap1 is turned on by guanine-nucleotide-exchange factors (GEF), which replace GDP for GTP, and cycled off by GTPase-activating proteins (GAP), which accelerate the ability of Rap1 to hydrolyze GTP into GDP (6). Rap1GAP belongs to a family of proteins with GAP function specific for Rap1/2, which also includes Rap1GAP2, SPA-1, and E6TP1. Recent evidence supports a role for Rap1GAPs in oncogenesis. Down-regulation of E6TP1 contributes to the development of cervical cancer associated with chronic human papillomavirus infection (7, 8). Deficiency of murine SPA-1, selectively expressed in lymphoid tissues (9), results in the development of chronic myeloproliferative or myelodysplastic disease (10). In addition, tuberin, the product of the tuberous sclerosis type 2 gene that shares sequence homology with Rap1GAP, functions to stimulate Rap1GTPase activity (11, 12), and its reduced expression is associated with the development of human gliomas (13, 14).

The Rap1GAP gene is located at chromosome 1p36.1-p35, a locus where deletions or mutations have been identified in many malignant tumors, including melanoma, (15, 16) and, more recently, breast cancer (17). Loss of heterozygosity (LOH) of the Rap1GAP gene was recently reported in pancreatic cancer (18), and decreased expression has been identified in thyroid cancer (19, 20) and oropharyngeal squamous cell carcinoma (21). In these studies, the loss of Rap1GAP expression was associated with tumor cell proliferation and migration and tumor growth, leading to the proposal that Rap1GAP may function as a tumor suppressor gene in these tumors.

Loss of function of tumor suppressor genes is a common theme in tumorigenesis. The transcription of tumor suppressor genes can be silenced by gene mutations, LOH, or epigenetic mechanisms, including gene promoter methylation (22–24). Promoter methylation, catalyzed by DNA methyltransferases, occurs via covalent modification of cytosines and typically affects CpG-rich areas known as CpG islands located in the promoter region of many genes (24) and has been identified to play an important role in carcinogenesis (25–27). Approximately 50 genes that are silenced by promoter hypermethylation in melanoma have been identified (28), including methylthioadenosin phosphorylase (25), PTEN (26), and Apaf-1 (27). Furthermore, DAB2IP, a GAP specific for Ras (29), and DLC-1, a Rho family member GAP (30), is down-regulated in certain human tumors via aberrant promoter methylation.

Here, we report that the expression of Rap1GAP is decreased in human melanoma tumors and cell lines. Overexpression of Rap1GAP in melanoma cells blocks Rap1 activation and ERK phosphorylation and inhibits melanoma cell proliferation and survival. In addition, overexpression of Rap1GAP inhibits focal adhesion formation and decreases melanoma cell migration. We show that Rap1GAP was down-regulated in melanoma via promoter methylation. Furthermore, treatment of melanoma cells having low constitutive Rap1GAP expression with the demethylating agent 5-aza-2'-deoxycytidine (5-aza) results in reinduction of Rap1GAP expression and decreases Rap1 activity, ERK phosphorylation, and cell proliferation and survival. These effects are blunted in small interfering RNA (siRNA)–mediated Rap1GAP knockdown melanoma cells, confirming the specificity of the effects. Our findings contribute to further understanding of the molecular mechanisms underlying melanoma development and may lend further support to the exploration of treatment targeting epigenetic gene regulation in melanoma.

Melanoma cell lines and human tumors. Fresh frozen and formalin-fixed, paraffin-embedded human melanoma tumors and human melanoma cell lines (A375, MeWo, A2058, 501 mel, LOX, M14, DM6, DM 13, Sk mel 5, DM14, and LH) were used in accordance with an approved institutional protocol (HRPO 05-0794). Detailed descriptions can be found in Supplementary Materials and Methods.

Cell transfections. Melanoma cell tranfections were carried out with either a pcDNA3.1 vector containing Flag-Rap1GAP (generous gift of Dr. P. Stork, Oregon Health Sciences University) or Rap1GAPsiRNA (Santa Cruz Biotechnology) according to the manufacturer's instructions. Experimental details are listed in Supplementary Materials and Methods.

Western blot and Rap1 activation assays. Western blot analyses and Rap1 activation assays using cell protein lysates were performed, as described previously (5). Further experimental details, including specific antibodies, are described in Supplementary Materials and Methods.

Reverse transcription-PCR. Total RNA was isolated from melanoma cells using RNA STAT-60 (Tel-Test, Inc.) and quantified by spectrophotometry (Eppendorf). Primer sets and PCR conditions are listed in Supplementary Materials and Methods.

Real-time PCR. Total RNA was isolated from melanoma cell lines, as described above. cDNA was synthesized by SuperScript First-Strand Synthesis System for reverse transcription–PCR (RT-PCR) from Invitrogen. Primer sets, PCR conditions, and details of quantitation analysis are listed in Supplementary Materials and Methods.

Immunohistochemistry and immunofluorescence. Formalin-fixed, paraffin-embedded melanoma tumors, and benign nevi were sectioned (5 μm), dehydrated, and immunohistochemistry for Rap1GAP was performed using primary and secondary antibodies, as listed in Supplementary Materials and Methods. For immunofluorescence, cells were fixed in 4% paraformadyhyde for 30 min and permealized with 0.1% Triton X-100, and immunostaining was performed with antibodies, as listed in Supplementary Materials and Methods.

Cell proliferation and apoptosis assays. BrdUrd incorporation was used to determine cell proliferation. Cell apoptosis was detected by terminal deoxyribonucleotydyl transferase–mediated dUTP nick-end labeling assay. Experimental details can be found in Supplementary Materials and Methods.

Migration assay. Transfected LH melanoma cells were plated onto 8-μm pore size transwell inserts (Corning Costar), and cell migration into the bottom of the serum-containing well was assessed at 24 h, as previously described (5).

Methylation-specific PCR. To investigate whether Rap1GAP promoter methylation could be active in regulation of Rap1GAP expression, methylation-specific Rap1GAP PCR primers were designed within the promoter region using methylprimer (31) located at −561 to −332, containing 20 CpG sites, and methylation-specific PCR (MSP) was performed. Details, including primer sets, can be found in Supplementary Materials and Methods.

DNA demethylation. Experimental details can be found in Supplementary Materials and Methods. 5-Aza, a demethylating agent, was used in the present study.

Quantitative DNA methylation analysis. Genomic DNA extraction and DNA bisulfite modification were performed. Sequenom technology (Sequenom) was used for quantitative Rap1GAP gene methylation at a core facility at our institution.⁶

Three pairs of primers for three CpG islands identified in the Rap1GAP promoter were designed using Epidesigner software (Sequenom) and included the CpG islands that were initially examined by MSP. Further experimental details, including primer sets and methylation analyses, can be found in Supplementary Materials and Methods.

BRAF, NRAS, Rap1A, Rap1B, and RAP1GAP gene mutation analysis. Genomic DNA or total RNA was extracted from melanoma cell lines and frozen human melanoma tumors. Details of sequence analysis can be found in Supplementary Materials and Methods.

Statistical analysis. Statistical analysis was performed using Student's t test. For comparisons between multiple treatments, significance was determined using Bonferroni's post hoc ANOVA test where appropriate. All values were represented as mean ± SD. Significance was determined at P < 0.05.

Decreased Rap1GAP expression in melanoma tumors and cell lines. To characterize the role of Rap1GAP in melanoma tumorigenesis, we first examined its expression in human melanoma tumors and cell lines. Rap1GAP expression and Rap1 activity in eight human melanoma tumors were determined by Western analysis and Rap1 activation pull-down assay, respectively. Decreased or absent Rap1GAP expression was found in cutaneous (15138 and 33221) and metastatic (14583, 16151, 11771, 13488, and 14150) melanoma tumors compared with human epidermal melanocytes (HEM; Fig. 1A). An increased level of Rap1GTP was associated with decreased Rap1GAP expression in the majority of tumors, consistent with negative regulation of Rap1 activity by Rap1GAP. In addition, decreased Rap1GAP expression was found in tumors that were either wild-type BRAF or BRAF^V600E. All tumors examined were wild type for NRAS. We found no association between decreased Rap1GAP expression and the BRAF and NRAS mutations (Fig. 1A). Interestingly, tumor 19480 had higher levels of Rap1GTP in spite of high Rap1GAP expression. We therefore examined whether this tumor harbored Rap1GAP mutations, as recently identified in two human breast cancers (17). We had previously sequenced the entire Rap1GAP gene (21 exons) in seven human melanoma tumors and in two melanoma cell lines (A375 and MeWo) and had been unable to identify significant mutations.⁷

Similarly, we were unable to identify mutations 769T>G, 1826A>G in tumor 19480 at the two sites identified in breast cancer (data not shown). The mechanism of this dissociation of Rap1GAP expression and Rap1 activity needs to be further identified.

Rap1GAP expression was determined in formalin-fixed, paraffin-embedded melanoma tumors (n = 11) and nevi (n = 8) by immunohistochemistry. Five samples were paired malignant (melanoma) and benign (melanocytic nevi) from the same patient. High expression of Rap1GAP protein in the suprabasal epidermis served as positive internal control (Fig. 1B). In seven of eight nevi, Rap1GAP was expressed in the cytoplasm of dermal nevus cells (Fig. 1B,, top). In contrast, in all of the melanoma tumors examined (n = 11), Rap1GAP was not detected in the tumor cells (Fig. 1B , middle). These results showed decreased Rap1GAP expression in melanoma but not in benign nevi.

Next, Rap1GAP level in 11 melanoma cell lines was determined by Western blot. Eight of 11 cell lines had decreased or absent Rap1GAP expression compared with HEM (Fig. 2A). Again, there was no correlation between BRAF/NRAS mutations and Rap1GAP expression. Taken together, our findings of down-regulation of Rap1GAP in a high percentage of both the melanoma tumors and the cell lines investigated indicate a role for Rap1GAP in melanoma tumorigenesis.

Rap1GAP blocks ERK phosphorylation, inhibits melanoma cell proliferation and survival, and is associated with decreased Rap1 activity. We had previously found that Rap1 promotes melanoma cell proliferation and ERK activation, an effect that was blunted by dominant-negative Rap1A or Rap1 siRNA (5). Because Rap1GAP inhibits Rap1 activity, we asked whether Rap1GAP expression affects ERK activity and melanoma cell proliferation. Two melanoma cell lines with low or nondetectable Rap1GAP expression and high-baseline Rap1 activity (LH and Skmel 5) were chosen for further experiments. We wanted to assure that the constitutively high Rap1 activity in these two cell lines was not due to activating mutations in Rap1 (32). Direct sequencing of cDNA for Rap1A and Rap1B was performed. No mutations were found in either cell line (data not shown). We then proceeded to overexpress Rap1GAP in these cells by transfection with FLAG-tagged Rap1GAP. Consistent with its GAP function, Rap1GAP overexpression also greatly reduced Rap1GTP level, which in turn decreased ERK phosphorylation (Fig. 2B) and inhibited cell proliferation by nearly 50% compared with empty vector-transfected control cells (Fig. 2C). Rap1GAP overexpression also significantly induced apoptosis (Fig. 2D). These data show that increased Rap1GAP expression suppresses Rap1 activity, as well as ERK activation and melanoma cell proliferation and survival.

Rap1GAP suppresses focal adhesion formation and cell migration. Rap1 regulates integrin activation, focal adhesion formation, and cell polarization (33). Phosphorylation of Src, a nonreceptor tyrosine kinase that is autophosphorylated and activated upon engagement of α_vβ₃ integrin to vitronectin (34), is a surrogate marker for increased α_vβ₃ activation. Overexpression of Rap1GAP decreased basal Src phosphorylation in LH melanoma cells and blocked serum-induced Src phosphorylation (Fig. 3A).

We then investigated the effect of Rap1GAP on two integrin-dependent events, focal adhesion formation, and filamentous actin (F-actin) assembly. Serum-starved empty vector and Rap1GAP-transfected LH cells were stimulated with serum for 1 hour and stained for F-actin and vinculin. After serum stimulation, empty vector-transfected cells spread, and vinculin was translocated from the cytoplasm to the focal adhesions at the leading edge of the cell membrane (Fig. 3B,, top left). F-actin was organized into numerous filaments oriented parallel to the long axes of the cells (Fig. 3B,, top middle). In contrast, Rap1GAP-overexpressing cells remained in a rounded morphology, and F-actin and vinculin were localized to the cytoplasm after serum stimulation (Fig. 3B , bottom).

The formation of focal adhesions is required for cell migration (33–35). We then determined the effect of overexpression of Rap1GAP on melanoma cell migration. Serum-stimulated Rap1GAP-overexpressing LH cells exhibited decreased cell migration compared with empty vector control cells (P < 0.05; Fig. 3C and D). These findings, taken together with the effect of Rap1GAP overexpression on Src phosphorylation, focal adhesion formation, and F-actin filament formation, provide functional data to support our proposal that loss of Rap1GAP expression promotes processes that may favor tumor growth and metastasis.

Down-regulation of Rap1GAP is mediated by its promoter methylation. The expression of tumor suppressor genes may be silenced through gene mutations, LOH, or epigenetic mechanisms, including gene promoter hypermethylation (22–24). We have been unable to identify significant Rap1GAP gene mutations in human melanoma tumors and cell lines in our earlier investigations.⁸

We, thus, focused on other potential mechanisms of Rap1GAP down-regulation, in particular, epigenetic gene silencing via promoter hypermethylation. We initially used methylation-specific PCR for Rap1GAP to detect the methylation status of the Rap1GAP promoter in both human melanoma tumors and cell lines. As shown in Fig. 4A and B, Rap1GAP promoter methylation was found in four of the melanoma tumors (14583, 11771, 15138, 14150) and in A375, LOX, M14, DM14, DM13, LH, SK mel5 melanoma cell lines that had decreased Rap1GAP expression, as shown earlier by Western blot (Figs. 1A and 2A).

We next wanted to determine whether the demethylating agent 5-aza was effective in demethylating the Rap1GAP promoter in melanoma cells (LOX, A375, and LH). In fact, 5-aza treatment of all three cell lines resulted in demethylation of the Rap1GAP promoter compared with 5-aza–untreated cells (Fig. 4C). To further confirm Rap1GAP promoter methylation and the effect of 5-aza, we analyzed the degree of methylation of three CpG islands within the Rap1GAP promoter in two melanoma tumors and three melanoma cell lines using Epityper software (Sequenom). Human melanoma samples that showed either promoter hypermethylation (14150) or hypomethylation (19480) by MSP (Fig. 4A) were examined along with three melanoma cell lines (LOX, LH, and A375) pre–5-aza treatment and post–5-aza treatment. Results are reported as the relative methylation of each CpG unit expressed as a ratio of methylated/unmethylated (range, 0–1) for all evaluated islands (36). Relative methylation of each CpG unit (each CpG unit is composed of one or several CpG sites, which collectively constitute a CpG island) in three CpG islands is shown in tabulated form in Supplementary Table S1. Sequenom mass signal for methylation was sufficient to evaluate 84% of the CpG units within CpG island 1, 86% of the CpG units within island 2, and 86% of the CpG units in island 3 in tumor 14150 and LOX, A375, and LH cells. Mass signal was too low to quantitate the methylation of CpG island 3 in HEM, tumor 19480, and 5-aza–treated melanoma cells.

In melanoma tumor 14150, relative methylation for all three CpG islands was >0.7 and significantly increased compared with HEM in CpG islands 1 and 2 (Supplementary Table S1; Fig. 1A and B). In tumor 19480, relative CpG methylation was <0.07 for both CpG islands 1 and 2. There was no significant difference for these two CpG islands between tumor 19480 and HEM (Supplementary Table S1; Fig. 1A and B). Overall, tumor 14150 showed significantly higher methylation of the Rap1GAP promoter than HEM (P < 0.05), whereas tumor 19480 showed similar methylation to HEM (P = 0.48; Fig. 4D , left), consistent with our MSP data.

In all three melanoma cell lines, there was a significant increase in the level of relative methylation of the CpG units in CpG islands 1 and 2 compared with HEM, and 5-aza reversed this methylation (Supplementary Table S1; Fig. 1C and D). The average methylation level of CpG island 3 was >0.7 in control LOX, A375, and LH cells. Data for post 5-aza–treated cells and HEM were not available due to low mass signal (Supplementary Table S1). Taken together, melanoma cell lines not only showed higher levels of methylation of three CpG islands within the Rap1GAP promoter region compared with HEM but the relative methylation of the Rap1GAP promoter was essentially reversed by 5-aza in these three melanoma cell lines, respectively (Fig. 4D , right). These results were again consistent with the results of MSP and extended our findings to provide a more extensive, quantitative measure of Rap1GAP promoter methylation (37).

5-Aza reinduction of expression of Rap1GAP suppresses melanoma cell proliferation and survival. We tested whether 5-aza treatment restores Rap1GAP expression in three melanoma cell lines. Indeed, 5-aza treatment reinduced expression of Rap1GAP mRNA in LOX cells in a time-dependent manner (Supplementary Fig. S2) and reinduced expression of Rap1GAP mRNA in A375 and LH cells compared with controls (Fig. 5A). This was confirmed by real-time PCR, which showed >200-fold increases in Rap1GAP mRNA expression in 5-aza–treated LOX, A375, and LH melanoma cells compared with untreated controls (Supplementary Table S2). [A single band PCR amplicon for each primer pair was verified by RT-PCR for Rap1GAP (149 bp) and glyceraldehyde-3-phosphate dehydrogenase (94 bp; Supplementary Fig. S3)]. We then investigated whether 5-aza reinduction of Rap1GAP expression could reproduce the downstream effects of Rap1GAP overexpression that we found after Rap1GAP transfection. In fact, after 5-aza treatment, Rap1GAP protein reinduction was associated with decreased Rap1 activity, ERK phosphorylation (Fig. 5B), cell proliferation (Fig. 5C), and increased apoptosis (Fig. 5D) compared with untreated controls.

To confirm the specificity of Rap1GAP induction by 5-aza, we used siRNA to specifically knockdown Rap1GAP. As shown in Fig. 6A, Rap1GAP siRNA transfection significantly decreased Rap1GAP mRNA reinduction in LOX cells after 5-aza treatment compared with 5-aza–treated cells with or without control siRNA transfection. More importantly, 5-aza reinduction of Rap1GAP protein expression was blunted, concomitant with increases in Rap1 activity and ERK phosphorylation in LOX cells transfected with Rap1GAP siRNA compared with control siRNA-transfected, 5-aza–treated cells (Fig. 6B). As a result, 5-aza inhibition of cell proliferation was reversed after Rap1GAP siRNA transfection compared with control siRNA-transfected, 5-aza–treated cells (Fig. 6C). These findings confirmed that deregulation of Rap1GAP expression via epigenetic mechanisms specific to the RAP1GAP gene contributes to Rap1 activation, ERK phosphorylation, and other processes important to melanoma tumorogenesis—cell proliferation and apoptosis.

The majority (∼85%) of melanomas harbor activating mutations in either BRAF or NRAS, acting upstream of ERK/MAPK pathway. Either mutation alone, however, seems insufficient for melanoma development, and ∼15% to 20% of melanomas do not harbor mutations in either gene. To this point, it has been shown that inhibition of ERK activity alone does not correlate with inhibition of melanoma cell growth (38). It has also been determined that acral, mucosal, and subungual melanomas uncommonly carry BRAF and NRAS mutations and may have mutations in other oncogenic genes, such as kit (39). Recently, through a systematic interrogation of melanoma cell lines lacking mutations in either of these two genes, a molecular distinct subset was identified characterized by p53 inactivation, reduced ERK activity, and increased expression of epithelial cell markers (40). Reports, such as these, highlight the genetic heterogeneity of melanoma (2, 3) and support the contribution of additional molecular events, acting either through the ERK/MAPK pathway or distinct from it, to melanoma tumorigenesis.

In this study, we have shown that Rap1GAP regulates processes important in melanoma tumorigenesis and show an epigenetic mechanism of Rap1GAP regulation. More specifically, Rap1GAP expression was down-regulated in most melanoma tumors and cell lines examined, and immunohistochemical analysis showed that Rap1GAP is expressed in the cytoplasm of benign nevus cells but not in malignant melanoma cells. Second, overexpression of Rap1GAP in human melanoma cell lines that have very low or absent Rap1GAP expression have greatly reduced Rap1GTP levels and diminished to absent ERK phosphorylation. Third, Rap1GAP inhibits melanoma cell proliferation, survival, focal adhesion formation, and migration. We also report that the mechanism of Rap1GAP decreased expression is via promoter methylation in a significant number of melanoma tumors and cell lines. Finally, we show the specificity of the demethylating agent 5-aza in reinduction of Rap1GAP expression in melanoma cell lines using siRNA specific to Rap1GAP.

Rap1GAP is located on chromosome 1p35-36, a locus frequently implicated as a putative tumor suppressor gene in several cancer types, including chronic myelocytic leukemia (41), gliomas (42), and oral squamous carcinoma (15) via mechanisms including LOH and chromosomal deletion. Until recently, the identification of a specific tumor suppressor gene residing at this site has remained elusive. Using murine models engineered for gain and loss of a region corresponding to human 1p36, CHD5 was identified as having tumor suppressor activity (43). LOH studies in melanoma have found 1p36 allelic loss in a significant number of melanoma tumors (44) and may represent another gene in addition to CHD5.

The role of other Rap1 GAP family members in tumor development has been reported, including tuberin in tuberous sclerosis (11–14), E6TP1 in cervical cancer (7, 8), and SPA-1 in murine myelodysplastic syndrome (9, 10). More specifically, a tumor suppressor role for Rap1GAP has been recently reported in pancreatic cancer (18), thyroid cancer (19, 20), and oropharyngeal squamous cell carcinoma (21). Although Rap1GAP expression is decreased in all of these tumors, the mechanism(s) of Rap1GAP down-regulation may be cell type–specific. For instance, treatment of pancreatic cancer cell lines with absent Rap1GAP expression with the DNA methyltransferase inhibitor 5-aza failed to induce Rap1GAP expression. However, lack of Rap1GAP expression correlated well with LOH status of the Rap1GAP gene (18). In addition, recent mutations in the Rap1GAP gene have been identified in two breast cancers (17), although we have been unable to identify similar mutations in the melanoma tumors or cell lines examined. We did, however, detect a high frequency of promoter methylation in the Rap1GAP gene, supporting a role for promoter hypermethylation in Rap1GAP down-regulation in melanoma. Furthermore, 5-aza treatment reinduced Rap1GAP expression in several melanoma cell lines, with consequent decreased Rap1 activity, ERK phosphorylation, and inhibition of melanoma cell proliferation and survival. The specificity of 5-aza–induced demethylation on Rap1GAP was shown using Rap1GAP siRNA-transfected cells, where 5-aza–reinduced Rap1GAP expression was blunted together with increased Rap1 activity, ERK phosphorylation, and cell proliferation. Taken together, these findings highlight the role of the Rap1GAP/Rap1 pathway in processes important to melanoma tumorogenesis.

Several studies have shown that, dependent upon the cell and tissue type, Rap1 can either induce or inhibit ERK activation (6, 45). Early work found that Ras, as opposed to Rap1, mediated the activation of ERK in melanocytes and B16 murine melanoma cells (46). Activated Rap1 can associate with BRAF, forming a complex that activates ERK in certain cells (47). In addition, distinct cellular pools of Rap1 exist, and the potential to form a Rap1/BRAF complex leading to activation of ERK is dependent on the capability of distinct Rap1-GEFs to redirect Rap1 to the plasma membrane (48).

In the present study, we found that decreased Rap1GAP expression in melanoma tumors and cell lines with increased Rap1 activity did not correlate with either BRAF or NRAS mutational status. Moreover, in both LH and Skmel 5 melanoma cells having very low or absent Rap1GAP expression but distinct BRAF and NRAS gene mutational status, overexpression of Rap1GAP inhibited ERK phosphorylation and cell proliferation and survival. This is consistent with our previous work showing that inhibition of Rap1 activity using Rap1 siRNAs and dominant-negative Rap1 inhibits ERK phosphorylation and melanoma cell growth independent of BRAF and NRAS mutation status (5). Based upon these findings, we propose that the Rap1GAP/Rap1/ERK pathway may function either in a synergistic manner or independent from the RAS/RAF/ERK pathway in melanoma tumorigenesis.

In squamous cell carcinoma, Rap1GAP expression was found to decrease activation of Rap1 and ERK and inhibit tumor growth (21), consistent with our findings. Seemingly conflicting cell type–specific findings, however, have been reported with respect to the effect of Rap1GAP expression on cell migration and invasion. For example, Rap1GAP expression in squamous cell carcinomas induced metalloproteinase expression (MMP-2 and MMP-9) and increased cell invasion (49), whereas antisense oligonucleotides to Rap1GAP had no effect on ovarian cancer cell migration through Matrigel (50). We had found that Rap1 activation increases activation of α_vβ₃ integrin and promotes cell migration in melanoma cells (5). In the present study, we show that overexpression of Rap1GAP in melanoma cells with low to absent Rap1GAP expression blocks serum-induced cell spreading, focal adhesion formation, F-actin cytoskeletal arrangement, and integrin activation—processes integral to cell migration (34, 35). In fact, we show that Rap1GAP overexpression does inhibit melanoma cell migration, consistent with findings in pancreatic (18) and thyroid (20) carcinoma, wherein overexpression of Rap1GAP not only inhibited tumor cell growth but also suppressed tumor cell invasion and migration. Again, we propose that these effects are cell type–specific and tumor type–specific. Further studies using in vitro assays and murine xenograft models to investigate the effect of Rap1GAP expression on melanoma cell invasion, tumor growth, angiogenesis, and metastasis are currently under way in our laboratory.

In addition, the squamous cell carcinoma study showed that Rap1GAP expression suppressed tumor growth and increased invasive capability, and matrix metalloproteinase activity in vivo was associated with early (low-N stage) squamous cell carcinoma (49). In our studies, the primary human melanoma tumors examined by immunohistochemistry did not express Rap1GAP protein, in contrast to their benign counterparts (melanocytic nevi). In addition, the majority of both the cutaneous and metastatic tumors that we examined by Western blot had decreased to absent Rap1GAP expression. Interestingly, the findings in squamous cell carcinomas were dependent upon tumor stage, and we do not have clinical correlation or patient follow-up for the melanoma tumors that we investigated. Determination of whether a correlation between Rap1GAP expression and pathologic or clinical stage in melanoma will obviously require further study.

Taken together, our findings support a role for Rap1GAP in the regulation of processes important to melanoma tumor formation and progression and show the role of promoter methylation in the regulation of gene expression. These data may contribute to the further understanding of the molecular mechanisms underlying melanoma development and may have implications for melanoma treatment. Our findings may also lend further support to the exploration of treatments targeting epigenetic gene regulation in melanoma.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Anne M. Bowcock for her comments and the use of Core Facilities for gene mutation analysis (http://hg.wustl.edu/gcore/methylation.html). We also thank Yuefang Huang and Dr. Laurin Council for their technical assistance.