Background
The metallothioneins (MT) are a group of low molecular weight, cysteine-rich intracellular proteins that are involved in maintaining intracellular metal homeostasis by binding transition metals such as zinc and copper. There are 10 functional isoforms of MTs described, which are divided into 4 classes, designated MT1 – 4, on the basis of small differences in protein sequence and charge characteristics [
1,
2]. The MTs have been proposed to play an important role in protecting against DNA damage, apoptosis and oxidative stress [
3].
Metallothionein 3 (MT3) was first identified as a growth inhibitory factor, expressed in normal brain, which inhibited the survival of neurones in culture and also neurite formation [
4]. Subsequent studies using glial [
5] or tumour[
6‐
8] cells, stably transfected with
MT3, showed that its endogenous over-expression could inhibit cell growth. More recently, down-regulation of
MT3 was reported as one of 17 changes in gene expression which was most likely to be associated with metastasis and poor clinical outcome in a range of solid tumours [
9]. One mechanism for reducing gene expression is methylation of the CpG island when present in the promoter region of the gene [
10]. Methylation of the
MT3 promoter has been observed and has been suggested to cause reduced expression in gastric cancer [
11].
The levels of
MT3 expression, the frequency of
MT3 methylation, and correlations between
MT3 methylation and clinical parameters, have not been investigated in oesophageal squamous cell carcinoma (SCC). In this study we used combined bisulfite restriction analysis (COBRA) [
12] to estimate the frequency of methylation at specific sites within the
MT3 promoter in oesophageal cancer cell lines, and investigated the relationship between methylation and its expression by quantitative real-time RT-PCR. We then measured
MT3 gene expression and the frequency of
MT3 methylation in primary oesophageal SCCs and, when available, the histologically normal, proximal resection margin from those patients.
Discussion
In this study we investigated the expression of MT3 and its control by DNA methylation in oesophageal SCC. We measured methylation of MT3 by COBRA in three separate regions of its promoter, and gene expression by quantitative real-time RT-PCR. In the oesophageal cell line OE33, which had complete methylation at all the CpG sites analysed, there was also complete transcriptional silencing of the MT3 gene. In a further 3 cell oesophageal cancer cell lines, each of which had partial methylation, there was a partial but not complete reduction in gene expression. Treatment of each of these cell lines with aza-dC, which reduced methylation, resulted in an increase in gene expression.
In those the cell lines which had partial methylation, the amount of methylation could vary between contiguous CpGs in the same cell line, and could vary in particular CpGs between cell lines. Thus, a particular CpG which might be unmethylated in one cell line could be partially or completely methylated in other cell lines expressing similar levels of
MT3. This is consistent with the reported variability of methylation of the p16 CpG island in primary human mammary epithelial cells during escape from growth arrest [
13], and the O6-methylguanine-DNA methyltransferase gene in human cell lines [
14]. The significance of this finding is that methods for the measurement of DNA methylation such as methylation specific PCR (MSP) which analyse methylation at one or two CpGs only may misrepresent the methylation pattern of a region. In addition these methods would mislead if they probed CpGs with low levels of methylation in all tissues analysed, such as the site 12 in intron 1 in our analysis of
MT3. This was methylated in normal lymphocytes and normal oesophageal tissue, as well oesophageal SCC, even though other CpGs were unmethylated and gene expression was high.
In order to show the relationship between methylation within the regions studied and MT3 gene expression, we incubated cultures of the oesophageal cell lines for approximately two cell divisions with aza-dC, a potent inhibitor of DNA methyltransferase. Because the rate of division varied for each of the cell lines, we determined the length of time required for each of the cell lines to divide twice in the presence of aza-dC using PKH-26 labelling [
15]. This stable red fluorescent dye inserts into the cytoplasmic membrane, and is distributed equally amongst each of the daughter cells at the time of cell division, such that the mean fluorescence intensity of the cell population is halved with each cell division. As expected, treatment with aza-dC for two cell divisions did not completely demethylate the cell lines. This is because it is only newly synthesised strands which are demethylated, so if in a cell a particular region of DNA is completely methylated on both alleles, after 2 divisions only 75% of the alleles will be demethylated. We found that there was a significant reduction in the amount of methylation at each of the methylated sites analysed following drug treatment. Interestingly, the amount of the reduction varied between different CpG sites within a cell line, and varied from cell line to cell line at a given site. The reasons for this cell line and regional variation in the extent of demethylation are unknown.
Previous studies have shown that the methylation of only a subset of available CpG sites within a region can be sufficient to reduce transcription [
16‐
18,
13], with a greater reduction in transcription as the density of CpG methylation increases [
19]. Complete methylation of all the sites in the regions which we studied correlated with lack of any gene expression, while methylation of only a subset of sites correlated with some
MT3 expression, and the further reduction in methylation resulting from demethylation with aza-dC always correlated with an increase in
MT3 expression. In this study we evaluated
MT3 expression only by RT-PCR, because an antibody specific for the MT3 protein was not commercially available. However, previous studies have shown that when the
MT3 transcript was able to be detected the protein was also detectable [
20,
21,
7,
8].
Our interest in
MT3 came from a report that solid tumours with a specific pattern of expression of 17 genes, including down-regulation of
MT3, were most likely to be associated with metastasis and poor outcome [
9]. Oesophageal SCC is an aggressive disease with a 5 year survival of about 20%, death most commonly due to secondaries. Molecular markers which assist in predicting metasases might help to tailor treatment options better.
MT3 was first identified as a growth inhibitory factor which decreased the survival of neuronal cultures [
4]. Subsequent studies using stably transfected cell lines showed that up-regulation of
MT3 was growth inhibitory in some, but not all, cell lines [
6‐
8].
The reported role of
MT3 in carcinogenesis is unclear. In gastric carcinomas
MT3 expression was found to be markedly reduced, but there was no indication of any relationship to outcome in this report [
22]. In contrast, levels of
MT3 protein were shown to be elevated in bladder [
20] and breast carcinomas [
21]. This elevated expression was a poor prognostic indicator, being associated with increased tumour stage in bladder carcinoma, and poor disease outcome in some breast carcinomas. In our series of 64 patients with oesophageal SCC we found a degree of methylation in 52% of primary tumours and 3% of the histologically normal proximal resection margin. We observed that
MT3 expression was frequently down-regulated in primary oesophageal SCC when compared to normal mucosa, and significant down-regulation was most commonly observed in those tumours that were methylated for
MT3, and not unmethylated tumours. Interestingly, 3 tumours which had methylated DNA also had normal expression of
MT3. This might reflect a heterozygous pattern of methylation, with only one allele methylated, or a mixed tumour cell population containing some cells which were methylated and some which were not. The techniques used in this study could not distinguish between these possibilities. Also, 3 patients with unmethylated DNA had low levels of
MT3 expression, perhaps reflecting mutation or other change in the tumour. We found no relationship between methylation status and survival in the subset of patients for whom data were obtainable, suggesting that in these patients there would also be no relationship between gene expression and survival. Methylation and reduced expression of
MT3 was also not associated with tumour stage or tumour size. Thus a change in
MT3 expression by itself does not appear to favour increased tumour growth, and methylations status is not associated with survival, in patients with oesophageal SCC.
Authors' contributions
ES isolated the DNA and RNA, performed the bisulphite modifications, COBRAs and RT-PCRs, designed the study, performed the statistical analysis and coordinated and drafted the manuscript. GCM aided with the RT-PCRs. ZQT and JFL collected the oesophagectomy material and clinical data. NJD performed the cell culture and edited the manuscript. AR confirmed the pathology. PD, DIW and GGJ supervised the work and edited the manuscript. All authors read and approved the final manuscript.