Methylation of RASSF1A, RASSF2A, and HIN-1 Is Associated with Poor Outcome after Radiotherapy, but not Surgery, in Oral Squamous Cell Carcinoma

Purpose: Radiotherapy is the standard adjuvant treatment for oral squamous cell carcinoma (OSCC). The Ras/PI3K/AKT pathway is the major mechanism associated with radioresistance. To evaluate the potential significance on the outcome of radiotherapy in OSCC of the Ras/PI3K/AKT pathway with respect to methylation of negative regulators, we examined the methylation status of genes known to be involved in Ras/PI3K/AKT pathway and aberrantly methylated in human cancers together with the mutation status of K-ras/H-ras.

Experimental Design: PCR–denaturing high-performance liquid chromatography was used to examine the methylation status of the RASSF1A, RASSF2A, PTEN, and HIN-1 genes, and PCR-RFLP was used to determine the mutation status of K-ras/H-ras in 482 OSCCs. Associations between mutation, methylation, clinicopathologic parameters, and outcome were evaluated.

Results: The frequencies of K-ras/H-ras mutation and promoter methylation of the RASSF1A, RASSF2A, PTEN, and HIN-1 genes were 6.6%, 22.4%, 27.8%, 1.2%, and 7.3%, respectively. A combination of RASSF1A and RASSF2A methylation was found to be significantly associated with poor disease-free survival (DFS). Furthermore, a gene dosage effect of the activated Ras/PI3K/AKT signal on DFS was observed in patients treated with radiotherapy after surgery but not in patients treated with surgery alone. The Ras/PI3K/AKT pathway was activated in 140 primary OSCCs among 286 patients treated with radiotherapy after surgery and methylation of RASSF1A/RASSF2A (75.7%) was the most common mechanism.

Conclusion: Our study indicates that epigenetic silencing of tumor suppressor genes involved in the Ras/PI3K/AKT pathway plays an important role in OSCC radioresistance and this provides a rationale for exploring novel treatment strategies.

Oral squamous cell carcinoma (OSCC; ICD9 code 140-149, excluding 142 and 147) is the most rapidly increasing male cancer in Taiwan. The incidence rate has increased 18% from 2001 to 2005. In 2005, OSCCs was the fourth most common cancer in males (1). Despite improvements in surgical and radiotherapy treatment, the 5-year survival rate for oral cancer has remained almost unchanged at ∼50% for the past 30 years (2). The most important determinant of a poor outcome within the first 2 years after treatment of the primary tumor is a high recurrence rate for primary tumors or the presence of a second primary tumor (3); these are often difficult cases involving radical surgical resectioning and resistance to radiotherapy (4).

Radiotherapy is currently the standard adjuvant treatment for OSCC. However, radiotherapy is sometimes ineffective because cancer cells can become radioresistant. Several reports have indicated that alterations in the Ras signaling pathway play a pivotal role in aberrant proliferation, invasion, and resistance to ionizing radiation across many tumor cell lines (5–7). Recent studies have further indicated that the Ras/PI3K/AKT pathway is associated with radioresistance in several human cancers both in vivo and in vitro (8–13). Activation of the Ras/PI3K/AKT pathway can occur by many mechanisms, which include activation of Ras (14, 15), mutation or amplification of PI3K (16), amplification of AKT (16), and mutation/decreased expression of the tumor-suppressor genes PTEN (16–18) and HIN-1 (19, 20).

To the best of our knowledge, the potential relationship between activation of the Ras/PI3K/AKT pathways through methylation of negative regulators and radioresistance has not been systematically analyzed in male OSSC. In this study, we examined in 482 Taiwanese OSCCs and analyzed the methylation status of four genes (RASSF1A, RASSF2A, PTEN, and HIN-1) that are involved in Ras/PI3K/AKT pathway and have been identified as aberrantly methylated in human cancers; in addition, we analyzed the mutation status of K-ras/H-ras. The combination of RASSF1A and RASSF2A promoter methylation was found to be significantly associated with poor disease-free survival (DFS) in these OSCC patients. In addition, an aberrant gene dosage effect of the Ras/PI3K/AKT signal on DFS was also observed in patients treated with radiotherapy after surgery.

Tissue samples. This study was approved by the institutional review board of Chang Gung Memorial Hospital. OSCC tissue (n = 482) was collected from patients who received surgical treatment at Chang Gung Memorial Hospital (Taoyuan, Taiwan) with informed consent. All surgically removed tissues were immediately frozen in liquid nitrogen and stored at −80°C until DNA extraction. Information on the patients' history of cigarette smoking, alcohol drinking, and areca quid (AQ) chewing together with general demographic data were obtained from a questionnaire by a well-trained interviewer.

DNA extraction and sodium bisulfite treatment. Genomic DNA was isolated from OSCC tumor tissues as described previously (21). The sodium bisulfite modification procedure was slightly modified from the protocol by Herman et al. (22). In brief, 2 μg of genomic DNA were denatured in 0.2 mol/L NaOH at 37°C for 20 min, and then sodium bisulfite NaHSO₃ (130 μL; 7.0 mol/L; freshly prepared; Sigma) and hydroquinone (10 μL; 30 mmol/L; freshly prepared; Sigma) were added; the mixture was initially incubated at 98°C for 10 min and then at 65°C for 2.5 h. The modified DNA was desalted using the Wizard DNA Clean-up System (Promega) and denatured in 0.2 mol/L NaOH at 37°C for 15 min; this was followed by neutralization by ammonium acetate (0.75 mol/L) and precipitation by alcohol. The pellet was then dried and dissolved in 30 μL of deionized water.

Analysis of ras gene mutation. Mutation of the H-ras and K-ras genes was detected by the PCR-RFLP method as described by Scott et al. (23). The primer sequences and restriction enzymes used for PCR-RFLP are listed in Table 1. All DNA fragments were electrophoresed on a 6% polyacrylamide gel, stained with ethidium bromide, and directly visualized under UV illumination. Cases displaying a mutant fragment were reamplified in another reaction and directly sequenced by an ABI 3130-Avant Automated DNA Sequencer (Applied Biosystems) to confirm and characterize the nature of the mutation.

Promoter methylation analysis by PCR–denaturing high-performance liquid chromatography. Because methylation-specific PCR has been shown to be too sensitive and to result in the overestimation of gene methylation status, we applied PCR–denaturing high-performance liquid chromatography (PCR-DHPLC) to investigate the methylation status of our target genes in the present study. Primers were designed to amplify the promoter CpG-rich region of PTEN, HIN1, RASSF1A, and RASSF2A gene by Methprimer.⁶

The sequences of the primers are listed in Supplementary Table S1. All PCR reactions were done on PTC-100 thermocyclers (MJ Research) in a final volume of 25 μL containing bisulfite-modified DNA (2 μL), 1× PCR buffer, 1.5 mmol/L MgCl₂, 0.2 mmol/L deoxynucleotide triphosphate, 10 pmol of each primer, and 1 unit of Thermostar polymerase (ABgene). DNAs from the peripheral blood lymphocytes of the healthy subjects with and without treatment with SssI methylase (New England Biolabs) were used as the positive and negative controls.

The PCR products were purified using the Gel/PCR DNA Fragments Extraction kit (Geneaid) according to the manufacturer's instructions. The methylation status of RASSF1A, RASSF2A, PTEN, and HIN-1 was determined by DHPLC on a WAVE DNA Fragment Analysis System (Transgenomic) as described previously (24, 25). In brief, PCR products were denatured, renatured, and analyzed by DHPLC. DNA was eluted within 4.5 min at a flow rate of 0.9 mL/min using an appropriate linear acetonitrile gradient of buffer B (0.1 mol/L triethylammonium acetate; 25% acetonitrile) at the appropriate denaturing temperature (Supplementary Table S2). Because methylation can protect against the conversion of C to U and then to T in PCR products and maintains a higher C/G content after bisulfite treatment, leading to a higher denaturing temperature and a delayed retention time, the temperature for heteroduplex detection was determined using WAVEMAKER software (Transgenomic), based on the melting profiles of the respective promoter sequences after bisulfite treatment of methylated DNA. To further confirm the accuracy of the DHPLC, positive and negative control DNAs were also sequenced using an ABI 3130-Avant Automated DNA Sequencer (Applied Biosystems). As shown in Supplementary Fig. S1, DHPLC was able to detect and analyze methylation at multiple CpG sites between the respective primers.

Western blotting. A total of 30 available fresh frozen OSCC tissues were included. All tissues were homogenized in lysis buffer [20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, 1% Triton X-100, and 2.5 mmol/L sodium PPi] containing complete protease and phosphatase inhibitor cocktails (Panomics, Inc.). Protein concentration was determined by the Bradford assay (Bio-Rad Laboratories). Equal amounts (60 μg) of total proteins were applied to 12% polyacrylamide SDS gels (SDS-PAGE), separated electrophoretically, and transferred onto polyvinylidene fluoride membranes (Millipore). The membrane was incubated with anti-AKT (1:1,000; Cell Signaling Technology) and anti-phosphorylated Ser⁴⁷³ AKT (1:1,000; Cell Signaling Technology). Expression of AKT and phosphorylated Ser⁴⁷³ AKT was detected with horseradish peroxidase–conjugated goat anti-rabbit IgG (1:3,000; Millipore) and an enhanced chemiluminescence kit (Millipore) according to the manufacturer's instructions. Anti–glyceraldehyde-3-phosphate dehydrogenase antibody (1:5,000; Sigma-Aldrich Corp.) was used as the loading control.

Statistical analysis. Statistical analysis was done using SPSS version 8.0 (SPSS). The associations between the clinicopathologic parameters and K-ras/H-ras mutation and the methylation status of the PTEN, HIN1, RASSF1A, and RASSF2A genes were examined using the χ² test or Fisher's exact test. Survival curves were constructed using the Kaplan-Meier method and compared with the log-rank test. The Cox regression model was applied to estimate the hazard ratio (HR) and 95% confidence interval (95% CI) for the effect of the Ras/PI3K/AKT pathway on DFS. A two-sided value of P < 0.05 was considered statistically significant.

Characteristics of the K-/H-ras mutations in OSCCs. A total of 482 OSCC patients were enrolled in this study. Table 1 presents the demographic information and clinicopathologic characteristics of the patients. The median follow-up time for all patients was 44.0 months. In total, 32 (6.6%) of the 482 OSCCs had mutations in K-ras/H-ras (Table 2). Among these 32 mutations, 3 were K-ras (2 at codon 12 and 1 at codon 13) and 29 were H-ras (24 at codon 12 and 5 at codon 61). Most of the mutations were G to A mutations (n = 20). The K-ras and H-ras mutations were mutually exclusive. Because the frequency of K-ras mutation was very low, further statistical analysis was not possible. Mutation of H-ras was found to be associated with tumor site, tumor size, and lymph node metastasis, but not associated with age, cigarette smoking, AQ chewing, or alcohol drinking. Mutations of H-ras were significantly higher in bucca cancer than tongue cancer (9.2% versus 2.4%; OR, 4.12; 95% CI, 1.28-14.71), in tumor size ≥4 cm than <4 cm (9.1% versus 3.9%; OR, 2.48; 95% CI, 1.08-5.76), and in lymph node negativity versus lymph node positivity (8.5% versus 1.7%; OR, 5.35; 95% CI, 1.52-22.55).

Characteristics of RASSF1A, RASSF2A, PTEN, and HIN-1 promoter methylation in OSCCs. As shown in Table 2, the frequency of promoter methylation in the RASSF1A, RASSF2A, PTEN, and HIN-1 genes was 22.4% (108 of 482), 27.8% (134 of 482), 1.2% (6 of 482), and 7.3% (35 of 482), respectively. The promoter methylation in these four genes was not associated with age, cigarette smoking, AQ chewing, or alcohol drinking. On the other hand, promoter methylation in the RASSF1A gene was significantly higher in tongue cancer cases than other cases (29.5% versus 13.2%; OR, 2.75; 95% CI, 1.42-5.38) and in tumor size <4 cm versus ≥4 cm (26.4% versus 16.7%; OR, 1.79; 95% CI, 1.11-2.91). However, promoter methylation in the RASSF2A, PTEN, and HIN-1 genes was not significantly associated with any clinicopathologic parameters.

To determine whether methylation of RASSF1A/RASSF2A genes correlates with activation of the Ras/PI3K/AKT signaling pathway, the phosphorylated form Ser⁴⁷³ AKT was further assessed (Supplementary Fig. S1). We found that OSCC tumors that displayed RASSF1A/RASSF2A methylation tended to show higher levels of phosphorylated Ser⁴⁷³ AKT than those that did not display methylation [methylated 6 of 10 (60%) versus unmethylated 6 of 20 (30%)].

Correlation between mutation, promoter methylation, and DFS in OSCCs. The survivorship for OSCC patients with a K-ras/H-ras mutation and with promoter methylation of the RASSF1A, RASSF2A, PTEN, and HIN-1 genes was estimated by the Kaplan-Meier methods and compared using the log-rank test. Among K-ras/H-ras gene mutation and promoter methylation of the RASSF1A, RASSF2A, PTEN, and HIN-1 genes, only RASSF2A promoter methylation was found to be associated with a poor DFS even after adjustment for age and tumor stage (HR, 1.37; 95% CI, 1.01-1.85; Fig. 1A). Because RASSF1A and RASSF2A both belong to the RASSF family, the combined effect of RASSF1A and RASSF2A promoter methylation on DFS was further investigated. As shown in Fig. 1B, the combination of RASSF1A and RASSF2A promoter methylation was found to be significantly associated with a poor DFS. Using the multivariate Cox proportional hazard model, including age and tumor stage, a combination of RASSF1A and RASSF2A promoter methylation was identified as an independent prognostic factor for DFS (HR, 1.49; 95% CI, 1.12-1.98). Furthermore, it is interesting to note that RASSF1A promoter methylation also showed significant association with a poor DFS in tongue cancer patients (n = 166; HR, 2.09; 95% CI, 1.25-3.50; Fig. 1C) and that HIN-1 promoter methylation was significantly associated with a poor DFS in lymph node extracapsular spread–positive patients (n = 116; HR, 2.66; 95% CI, 1.30-5.45; Fig. 1D).

A number of pieces of evidence have indicated that Ras/PI3K/AKT activation is a contributor to the radiation survival of tumor cells (26). Therefore, the overall effect of activation of the Ras/PI3K/AKT signal on DFS was further evaluated. In the present study, either a K-ras mutation, a H-ras mutation (26), inactivation of RASSF family members by promoter methylation (27–29), or activation of the PI3K/AKT pathway by inactivation of the pathway negative regulators (18, 19) by promoter methylation can activate the Ras/PI3K/AKT signal. We found that a gene dosage effect on the activated Ras/PI3K/AKT signal was present for DFS in patients who had undergone radiotherapy after surgery; however, this gene dosage effect was not present in patients who had been treated with surgery alone (Fig. 2; Table 3). In the subset of patients who had received radiotherapy after surgery, the patients who were detected as having one mechanism of activation of the Ras/PI3K/AKT signal showed a significantly poorer DFS compared with those with a normal Ras/PI3K/AKT signal; importantly, those patients who were detected as having at least two mechanisms of activation of the Ras/PI3K/AKT signal showed a further worsening of their DFS compared with the other two groups.

The Ras/PI3K/AKT pathway has been implicated in all major mechanisms of radioresistance. In this study, we used PCR-DHPLC to examine the status of promoter CpG island methylation of four genes (RASSF1A, RASSF2A, PTEN, and HIN-1) involved in this pathway in addition to mutation of K-ras and H-ras mutation in 482 OSCCs. We found that the frequency of K-ras/H-ras mutation and promoter methylation of the RASSF1A, RASSF2A, PTEN, and HIN-1 genes was 6.6%, 22.4%, 27.8%, 1.2%, and 7.3%, respectively.

There was a remarkably low frequency of K-ras mutation (0.6%) in this series of OSCCs, which suggests that mutational activation of the K-ras gene may not play an important role in OSCC tumorigenesis in Taiwan. Although Kuo et al. (30) reported that 18% of AQ chewing–related OSCCs were found to have a K-ras mutation, our results are more similar to the results for HNSCCs in Western countries (31–33), for Japanese OSCCs (34), and for general Taiwanese OSCCs (35).

RASSF1A and RASSF2A are two negative effectors of Ras proteins. Promoter methylation has been reported to be the major mechanism for silencing RASSF1A and RASSF2A in HNSCC and OSCC (29, 36). The methylation of RASSF1A gene during the development and progression of HNSCC, including OSCC, has been reported, but the results are divergent (7.5-93%; refs. 36–40). Recently, Imai et al. (29) reported that epigenetic inactivation of RASSF2 was detectable in 12 of 46 (26%) primary OSCCs. However, they did not examine the prognostic significance of RASSF2A methylation. In the present study, we found that promoter methylation of RASSF1A and RASSF2A were 22.4% and 27.8%, respectively. In addition, RASSF2A promoter methylation was found to be associated with a poor DFS even after being adjusted for age and tumor stage (HR, 1.37; 95% CI, 1.01-1.85). The frequency of combined promoter methylation of RASSF1A and RASSF2A was much higher than the frequency of K-ras/H-ras mutation (38.4% versus 6.6%), which suggests that the former play a key role in OSCC tumorigenesis in Taiwan. This observation is similar to a study of Japanese OSCCs (29). Furthermore, a combination of RASSF1A and RASSF2A promoter methylation was an independent prognostic factor for DFS (HR, 1.49; 95% CI, 1.12-1.98) after adjustment for age and tumor stage.

Differences in the molecular alterations among the various OSCC anatomic sites have not been reported previously. In this study, we found that H-ras mutation was significantly higher in bucca cancer than tongue cancer (9.2% versus 2.4%; OR, 4.12; 95% CI, 1.28-14.71) and was associated with tumor size ≥4 cm versus <4 cm (9.1% versus 3.9%; OR, 2.48; 95% CI, 1.08-5.76); this contrasts promoter methylation of the RASSF1A gene, which was significantly higher in tongue cancer (29.5% versus 13.2%; OR, 2.75; 95% CI, 1.42-5.38) as well as showing an association with tumor size <4 cm versus ≥4 cm (26.4% versus 16.7%; OR, 1.79; 95% CI, 1.11-2.91). This suggests that genetic and epigenetic alterations in the H-ras signaling pathway are likely to be mutually exclusive in the OSCC subgroups, which supports the presence of two independent pathways for the development of OSCC with respect to molecular features. Mutual exclusion between RASSF1A methylation and K-ras mutation has been reported in some human solid tumors (41). Because K-ras mutation is rare in OSCC, the observation that RASSF1A methylation and H-ras mutation are mutually exclusive events in the development of certain subset of OSCCs could help to further support the function of RASSF1A as a negative effector of H-Ras in an alternative signaling pathway (42, 43).

PTEN is the central negative regulator of the PI3K signal transduction cascade. Alterations of PTEN at the genetic level have been frequently found in human cancers and are associated with a poor prognosis (18). However, deletions or point mutations are unlikely to play a key role in OSCC tumorigenesis (44). Shin et al. (45) reported that the contribution of PTEN inactivation by either mutation or loss of transcript to the pathogenesis of Korean oral cancer was low. Furthermore, Ishida et al. (46) found that methylation of PTEN in 49 primary Japanese OSCCs was not common either. In this study, we found that methylation of PTEN in Taiwanese OSCCs was rare (1.2%). Taking these results together, we suggest that inactivation of PTEN through methylation might not be an important mechanism in OSCC pathogenesis.

The HIN-1 tumor suppressor gene has previously been shown to be methylated in a variety of human cancers (20), but this has not been examined in OSCC. In this study, HIN-1 methylation was detected in 7.3% of the OSCCs analyzed and was found to be significantly associated with a poor DFS in lymph node extracapsular spread–positive patients (n = 116; HR, 2.66; 95% CI, 1.30-5.45). Recently, Yang et al. (47) reported that epigenetic aberrations of HIN-1 were significantly associated with decreased 3-year survival in neuroblastoma. It has been shown that HIN-1 has various biological functions, including inhibiting cell cycle reentry, suppressing migration and invasion, and inducing apoptosis in breast cancer cell lines; these effects seem to be mediated by a high-affinity cell surface receptor and to involve the modulation of the AKT signaling pathway (19). The results from this study and that of Yang et al. (47) provide evidence in vivo to support the breast cancer cell line observations, although the function of HIN-1 in vivo still remains unclear.

A number of pieces of evidence have indicated that Ras/PI3K/AKT activation is a contributor to the radiation survival of tumor cells (26). In the present study, activation of the Ras/PI3K/AKT pathway by K-ras/H-ras mutation (26), by inactivation of RASSF family members (RASSF1A and RASSF2A) by promoter methylation (27–29), or by activation of PI3K/AKT pathway through inactivation of their negative regulators (PTEN and HIN-1; refs. 18, 19) by promoter methylation, was examined and the effect of these changes on DFS was further explored within subsets of patients who had been stratified in terms of treatment by radiotherapy after surgery. We found that a gene dosage effect between the activated the Ras/PI3K/AKT signal and DFS was observable among patients who had been treated with radiotherapy after surgery. This was not present in patients who had been treated with surgery alone (Fig. 2; Table 3). Among the 286 patients who had received radiotherapy after surgery, the Ras/PI3K/AKT pathway was activated in 140 (49.0%) primary OSCCs and methylation of RASSF1A/RASSF2A (75.7%, 106 of 140) was the most common mechanism; this was followed by methylation of PTEN/HIN-1 (20.7%, 29 of 140); and K-ras/H-ras mutation was the least common mechanism (12.1%, 17 of 140). In addition, 11 OSCCs displayed activation of the Ras/PI3K/AKT pathway through at least two mechanisms. Those patients who had one aberration, in terms of a K-ras/H-ras mutation, methylation of RASSF, or methylation of PTEN/HIN-1, had a significantly poorer DFS compared with those with a normal Ras/PI3K/AKT signal; more importantly, those who had at least two aberrations showed an even worse DFS. These observations confirm that Ras/PI3K/AKT pathway contributes to the radioresistance of the tumor cells (48, 49) and epigenetic silencing of tumor suppressor genes would therefore seem to play an important role in the pathogenesis of OSCCs and other types of human cancer (50). Furthermore, epigenetic silencing of tumor suppressor genes involved in Ras/PI3K/AKT pathway could contribute to a certain extent to radioresistance in Taiwanese OSCC. These results provide a rationale for exploring treatment strategies that include demethylating agents combined with radiotherapy to minimize radioresistance.

No potential conflicts of interest were disclosed.