MAL promoter hypermethylation as a novel prognostic marker in gastric cancer

T-lymphocyte maturation associated protein, MAL, has been described as a tumour-suppressor gene with diagnostic value in colorectal and oesophageal cancers, and can be inactivated by promoter hypermethylation. The aim of this study was to analyse the prevalence of MAL promoter hypermethylation and the association with mRNA expression in gastric cancers and to correlate methylation status to clinicopathological data. Bisulphite-treated DNA isolated from formalin-fixed and paraffin-embedded samples of 202 gastric adenocarcinomas and 22 normal gastric mucosae was subjected to real-time methylation-specific PCR (Q-MSP). Two regions within the MAL promoter (M1 and M2) were analysed. In addition, 17 frozen gastric carcinomas and two gastric cancer cell lines were analysed both by Q-MSP and real-time RT–PCR. Methylation of M1 and M2 occurred in 71 and 80% of the gastric cancers, respectively, but not in normal gastric mucosa tissue. Hypermethylation of M2, but not M1, correlated with significantly better disease-free survival in a univariate (P=0.03) and multivariate analysis (P=0.03) and with downregulation of expression (P=0.01). These results indicate that MAL has a putative tumour-suppressor gene function in gastric cancer, and detection of promoter hypermethylation may be useful as a prognostic marker.

hypermethylation of Cox2 has been shown to be an independent prognostic marker in gastric cancer (de Maat et al, 2007).
In other gastrointestinal cancers, that is, colorectal and oesophageal cancer, the T-lymphocyte maturation associated protein MAL, involved in glycolipid-enriched membrane-mediated apical transport, has been described to be inactivated by promoter hypermethylation (Puertollano and Alonso, 1999;Mimori et al, 2003;Kazemi-Noureini et al, 2004;Mori et al, 2006;Lind et al, 2007). Promoter hypermethylation of MAL was a frequent event in these two cancer types, but infrequent in normal mucosa. As promoter hypermethylation of MAL could already be detected in cancer precursor lesions, it has been suggested as a tumoursuppressor gene with diagnostic value (Lind et al, 2007;Mimori et al, 2007). To the best of our knowledge, MAL promoter hypermethylation has not yet been shown in gastric cancers. Aim of this study was therefore to analyse promoter hypermethylation of MAL in gastric cancers, its relation to gene silencing and to determine its clinical value as a prognostic marker.

Material
Two hundred and two formalin-fixed and paraffin-embedded (FFPE) gastric adenocarcinoma samples, randomly selected from the Leeds University archive and 22 normal gastric biopsy specimens, randomly selected from the archives of the VU University Medical Center, were included in this study. In addition, 17 snap-frozen gastric cancer tissue samples, obtained from the archives of the department of pathology of the VU University Medical Center (Weiss et al, 2003) were included. Patients did not receive chemotherapy, nor radiotherapy. Moreover, two gastric cancer cell lines, IPA220 and GP202 (Gartner et al, 1996), kindly provided by Professor Dr R Seruca (IPATIMUP, Porto, Portugal), the cervical cancer cell line SiHA, obtained from the American Type Culture Collection (Manassas, VA, USA), and primary human keratinocytes were included. The study was approved by the Institutional Review Board and was in accordance with medical and ethical guidelines in place in The Netherlands.

DNA and RNA isolation procedures
DNA of the primary gastric tumour tissues was isolated as described before using a commercially available DNA isolation kit (QIAamp microkit; Qiagen, Hilden, Germany) (Weiss et al, 1999;Buffart et al, 2007). Briefly, areas of at least 70% of tumour cells were marked on a 4 mm haematoxylin and eosin stained tissue section. Tumour tissue was macro dissected from adjacent serial 10 mm sections, using a needle. After deparaffinisation, an overnight incubation at 371C with sodium-thiocyanate (1 M) and a proteinase K treatment, DNA was extracted.
DNA and RNA of the 17 snap-frozen gastric carcinoma tissue samples and gastric cancer cell lines, SiHa cervical cancer cell line and primary keratinocytes was isolated using TRIzol reagent (Invitrogen, Breda, The Netherlands) according to the instructions of the manufacturer, with some modifications. Details are described elsewhere (Weiss et al, 1999(Weiss et al, , 2003 (http://www.english. vumc.nl/afdelingen/microarrays/). All DNA and RNA concentrations and purities were measured on a Nanodrop ND-1000 spectrophotometer (Isogen, IJsselstein, The Netherlands).

Bisulphite treatment and real-time methylation specific PCR
Of each DNA sample, 500 ng was used for bisulphite treatment using a commercially available DNA modification kit (EZ DNA Methylation Kitt; Zymo Research, Orange, CA, USA).
Real-time methylation specific PCR (Q-MSP) was performed using primer sets targeting two regions within the MAL promoter (i.e. from À680 to À573 bp (M1) and À92 to À7 bp (M2) relative to the first ATG). Both regions within the MAL promoter were selected within the CpG island of the MAL promoter. Amplicons were detected and quantified using Taqman probes. The housekeeping gene b-actin was chosen as a reference for total DNA input measurement.
Q-MSP reactions were carried out in a total reaction volume of 12 ml containing 50 ng bisulphite treated genomic DNA, 417 nM of each primer, 208 nM probe and 2 Â QuantiTect Probe PCR Kit master mix (Qiagen, Westburg, Leusden, The Netherlands). For both MAL M1 and M2 regions, the PCR reaction was performed for 45 cycles (15 s at 951C and 60 s at 591C) with an initial denaturation of 15 min at 951C. For each Q-MSP a standard curve of serial dilutions of bisulphite-treated DNA (50, 5, 2.5, 0.5 and 0.25 ng) of the SiHa cervical cancer cell line was used. All samples were analysed in duplicate. As a negative control, multiple water samples, unmodified genomic DNA obtained from SiHa cells and unmethylated DNA obtained from primary keratinocytes were included. To determine the relative quantity of methylation, we calculated the ratios between MAL M1 and MAL M2 methylated DNA vs b-actin DNA (average quantity of methylated MAL DNA/ average DNA quantity for b-actin Â 1000).

Real-time RT -PCR
Of each RNA sample, 1mg was reverse transcribed to cDNA using oligo(dT) 20 Primer (Invitrogen) with AMV reverse transcriptase (Promega, Leiden, The Netherlands). RT -PCR was performed in a total reaction volume of 25 ml, containing 22.5 ml master mix and 2.5 ml cDNA (25 ng). The master mix contained 12.5 ml of SYBR Green PCR master mix (Applied Biosystems, Nieuwerkerk a/d IJssel, The Netherlands) and 0.5 mM of each primer. All samples were analysed in duplicate in a 7300 Real-time PCR System (Applied Biosystems). Amplification was performed in 50 cycles of 951C for 15 s and an annealing temperature of 601C for 1 min, with an initial denaturation step of 5 min at 951C. Relative expression levels were determined from the obtained Ct values and the 2DDCt method, using snRNP U1A as reference (Livak and Schmittgen, 2001), and transformed into a ln scale. Primary keratinocytes and the SiHa cervical cancer cell line were used as positive and negative controls respectively. Primer sequences are described earlier (Wisman et al, 2006;de Wilde et al, 2008;Wilting et al, 2008).

Statistical analysis
Receiver operator characteristic (ROC) curves were analysed for assessing the best cutoff value for methylation, on all FFPE samples, assuming that normal gastric mucosae are unmethylated and gastric carcinomas are methylated. Cutoff points were chosen based on the point on the ROC curve showing 100% specificity. Positivity for each methylated promoter region was considered when a specific sample had a ratio of M1/b-actin Â 1000 or M2/b-actin Â 1000 above the respective cutoff value. A sign test was used for testing significance of differences in frequencies of M1 vs M2 methylation.
Mann -Whitney U-test was used to determine significance of differences in expression values between methylated and unmethylated gastric carcinomas. Survival analysis was performed using the Kaplan -Meier method, using the survival length starting from the day of surgery of the primary tumour to the date of death due to gastric cancer (event) or to the last day of clinical follow-up (censored). Differences in survival length were analysed using the log-rank test. Multivariate analysis was performed using a Cox's proportional hazard regression model in a forward stepwise method for variable selection. Gender, histological type, tumour stage (T-stage) and lymph node stage (N-stage) were entered into the analysis. w 2 test was used for calculating differences in methylation status and tissue type, gender of the patient, histological type of the tumour and tumour stage. t-Test was used to evaluate age related differences in methylation status (SPSS 14.0 for Windows, Chicago, IL, USA). P-values below 0.05 were considered to be significant.

Frequent MAL promoter methylation in gastric carcinomas
The chosen cutoff values of methylation for the M1 and M2 promoter regions, based on the ROC curve analysis, were ratios of relative methylated DNA quantities of M1/b-actin Â 1000 and M2/b-actin Â 1000 above 95 and 22, respectively, yielding a specificity of 100% for both promoter regions and a sensitivity of 71 and 80% for M1 and M2 promoter regions, respectively. ROC curves for both M1 and M2 promoter regions are shown in Figure 1.
Both gastric cancer cell lines, IPA220 and GP202, showed methylation of both M1 and M2 promoter regions. Of all 202 gastric adenocarcinomas tested, 143 carcinomas (70.8%) showed methylation of the M1 promoter region and 162 carcinomas (80.2%) showed methylation of the M2 promoter region. Methylation prevalences of M1 and M2 promoter regions were significantly different (P ¼ 0.004). Dense methylation, that is, methylation of both M1 and M2 promoter regions, was detected in 133 (65.8%) of the carcinomas. Thirty carcinomas (14.9%) were unmethylated. All normal gastric mucosa samples were unmethylated for both regions. w 2 test yielded a significant difference between methylation status of gastric carcinomas and normal gastric mucosa tissues (Po0.001). An overview of the methylation status for both promoter regions is given in Table 1.

Correlation of MAL promoter methylation and survival
Follow-up data was available for 200 out of 202 patients. Only patients without distant metastasis at the time of surgery (M0) were included, leaving 179 patients for the survival analysis. Patients with a gastric carcinoma methylated for the M2 promoter region had a significantly better survival compared with patients with tumours unmethylated for the M2 promoter region (P ¼ 0.03) ( Figure 2). No significant correlation was found between M2 promoter methylation and age or gender of the patient, histological type of the tumour, T-stage and N-stage (Table 2). In addition, no significant correlation between M1 promoter methylation status and clinicopathological characteristics, including survival, was found.
Multivariate analysis revealed only N-stage, T-stage and MAL M2 promoter methylation status, in order of significance, to have independent prognostic value ( Table 2).

MAL promoter methylation is associated with reduced gene expression
Reduced expression of MAL relative to the housekeeping gene snRNP U1A was observed in both gastric cancer cell lines compared with primary keratinocytes that did not show MAL promoter hypermethylation. Of the 17 gastric cancer tissue samples tested for MAL mRNA expression, 11 (64.7%) were methylated for M1, 11 (64.7%) for M2 and 9 (52.9%) samples showed methylation of both regions. Gastric carcinomas with M2 promoter methylation showed significantly lower expression of the MAL gene compared with M2 unmethylated gastric cancers (P ¼ 0.01) (Figure 3, Table 3). For the M1 region, no significant differences in MAL mRNA expression were found between methylated and unmethylated tumours.

DISCUSSION
Gastric cancer is a common disease with generally a poor prognosis (Parkin et al, 2005). Biomarkers can be used to predict prognosis and optimise therapeutic strategies. Hypermethylation of the MAL promoter has been shown in colorectal and oesophageal cancers and MAL has been proposed as a putative tumour suppressor in these cancer types (Mimori et al, 2003;Kazemi-Noureini et al, 2004;Mori et al, 2006). Moreover, it has been proposed as a candidate marker for early detection of these cancers, as methylation of MAL could already be detected in precursor lesions (Lind et al, 2007(Lind et al, , 2008Mimori et al, 2007). In this study we show that MAL might have a similar role in gastric cancers, as methylation of MAL is detected at high frequency in gastric cancers and not in normal gastric mucosa samples.
In this study, two regions within the MAL promoter, M1 and M2, were analysed. Q-MSP analysis showed methylation of both promoter regions in the two gastric cancer cell lines analysed, indicating that methylation actually occurs in gastric epithelial     cells. Earlier studies showed protein expression of MAL by immunohistochemistry in normal gastric mucosa. Strong expression of MAL was observed in parietal and chief cells, but not in muscle and submucosa cells (Marazuela et al, 2003). Gel-based MSP analysis revealed a small subpopulation of unmethylated cells for both the M1 and M2 promoter regions in these cell lines (data not shown). Expression of MAL was hardly detected in both cell lines indicating that methylation is probably the main mechanism of downregulation of this gene. In gastric cancers, methylation of the M2 region was more frequently observed compared with the M1 region (80.2 vs 70.8%). In consistence with what has been described by Lind et al (Lind et al, 2008), an unequal distribution of DNA methylation within the MAL promoter was observed in gastric cancers, with the highest frequencies of methylation in the region closest to the transcription start site. M2 region is located around the transcription start site.
Results of this study show that MAL may serve as a prognostic marker in gastric cancer, as patients with tumours methylated for the M2 region show significantly better survival compared with patients with tumours unmethylated for MAL or methylated only for the M1 region. This survival benefit was independent of other clinicopathological data such as age, gender, histological type of the tumour, tumour stage and lymph node status. Interestingly, also in Hodgkin's lymphoma patients a similar association was found with a significantly worse survival in patients whose tumours expressed the MAL protein compared with patients with tumours lacking MAL expression (Hsi et al, 2006), indicating a prognostic value of MAL also in other cancer models.
The finding that inactivation of MAL by methylation gives a better prognosis for the patients may seem contradicting with a putative tumour suppressor function of this gene in gastric cancer. However, this finding has been observed earlier for the mismatch repair gene hMLH1, which also has a tumour suppressor function. Inactivation of this gene leads to microsatellite instability of the tumour and patients with microsatellite instable tumours have a better prognosis compared with patients with microsatellite stable tumours (Ribic et al, 2003;Parc et al, 2004;Beghelli et al, 2006). Therefore, tumours without MAL methylation might have a different biology overall, which could relate to poorer clinical outcome, rather than the outcome being dependent on MAL itself. However, this study was performed retrospectively on archival material, and therefore, rather should be considered as hypothesis generating. Actual clinical implementation of MAL promoter hypermethylation as a diagnostic marker requires further validation in a prospective study.
To test the biological relevance of MAL promoter hypermethylation, in a subset of gastric cancers the association between MAL promoter hypermethylation and mRNA expression was analysed. Results showed lower expression of MAL in gastric cancers methylated for the M1 or M2 region. The association between promoter methylation and reduced mRNA expression strengthens the biological relevance of MAL methylation and supports a putative role as tumour-suppressor gene. However, correlation between reduced expression and methylation of MAL was only significant for the M2 region, indicating that M2 methylation would have more biological relevance. Two out of 17 gastric cancers showed reduced MAL mRNA expression whereas the gene was unmethylated for both promoter regions. This indicates that other regulatory mechanisms of MAL silencing also exist in gastric cancer, which may include DNA copy number loss, other epigenetic mechanisms, altered expression of transcription factors regulating MAL or microRNAs targeting the MAL gene.
In summary, this study shows frequent promoter hypermethylation of MAL in gastric cancers, and not in normal gastric mucosa samples. Promoter hypermethylation of MAL is associated with downregulation of its expression, especially when methylation occurs at the M2 region within the promoter. Methylation of this region within the MAL promoter correlates with a significantly better survival of the patients. Altogether, these results pinpoint MAL as a putative tumour-suppressor gene with a role in gastric cancer, which may serve as an independent prognostic marker for clinical outcome of gastric cancer patients.