Introduction
It is estimated that there were 140,000 new cases and 63,500 deaths from head and neck squamous cell carcinoma (HNSCC) in Europe in 2012 [
1]. The 5-year age-standardised relative survival rate was approximately 40% for Europe overall with survival rates lowest in Eastern European and highest in Central, Western, and Northern European counties for all sub-categories of HNSCC such as oral cavity, oropharyngeal, laryngeal, and hypopharyngeal cancer [
2]. Head and neck squamous cell carcinoma (HNSCC) is frequently (54%) detected at an advanced stage and so often has a poor prognosis with less than 55% overall 5-year survival following diagnosis [
2,
3]. HPV infection is responsible for virtually all cervical cancers and approximately 35% of diagnosed oropharyngeal cancers worldwide, although there is a significant region to region variation globally [
4]. Chaturvedi et al. predict that the number of HPV-positive HNSCC cases will exceed cases of cervical cancer by 2020, demonstrating the urgent need for a better understanding of the disease process and potential prevention strategies [
5].
Folate is an essential dietary component required for the synthesis of nucleotides and for the synthesis and repair of DNA. There are several mechanisms by which folate status affects cancer risk. Low folate status has been shown to cause uracil mis-incorporation into DNA [
6], cytogenetic damage [
7], and impaired DNA repair [
8] all of which increase the risk of neoplastic change. Indeed, epidemiological studies have shown that low dietary folate intake is associated with an increased risk of HNSCC carcinogenesis [
9‐
11]. Conversely, other studies have shown that a high folate status may drive the growth of established tumours [
12,
13]. In addition, there is evidence that folate status affects HPV infection and behaviour. Studies by our group and others have shown that low folate status is associated with an increased risk of cervical HPV infection [
14,
15] and cervical intraepithelial neoplasia or invasive cancer [
15]. The direction of influence of folate status on HPV persistence is not clear [
14,
15]. Recently, Xiao et al. showed that low folate conditions increased HPV integration into the keratinocyte host genome and reduced viral capsid production in human keratinocytes in vitro [
16]. Folate and other diet-derived methyl donors (choline, betaine, methionine) may also influence cancer risk and progression through epigenetic effects. The epigenetic control of gene expression is regulated, in part, by the methylation of cytosine–guanine dinucleotides within gene promoter regions, and dysregulated methylation in the promoter sequences of tumour-suppressor or oncogenes can promote carcinogenesis in HNSCC [
17,
18]. For the methylation of DNA, methyl groups are transferred from the ultimate methyl donor,
s-adenosyl methionine (SAM), onto cytosine residues in DNA by DNA methyltransferases (DNMTs). Whilst the factors that regulate DNMTs are not well understood, there is evidence that methyl donor depletion may lower DNMT expression in cervical cancer cells enzymes. DNMT1 preferentially adds methyl groups to hemi-methylated DNA and is responsible for maintenance of DNA methylation following cell division [
19]. DNMT3a and 3b add methyl groups to CpG sites and are required for de novo DNA methylation [
20]. In contrast, Tet methylcytosine dioxygenase 1 has been implicated in demethylation of DNA [
21]. Poomipark et al. recently demonstrated methyl donor status altered the expression of DNMTs in cervical cancer cells [
22]. Appropriate gene methylation is important for the regulation of many processes, including those involved in cancer progression such as cell cycle control and apoptosis. In addition, cell stress due to lack of essential nutrients may also lead to increased expression of key pro-apoptotic genes such as death-associated protein kinase-1 (DAPK1) [
23] and p53 upregulated modulator of apoptosis (PUMA) [
24], and so alterations in methyl donor availability may have profound effects on cell behaviour.
Here, we have developed an in vitro model of methyl donor-deficient HNSCC cells, with particular emphasis on HPV-positive HNSCC. We show, for the first time, that methyl donor deficiency significantly alters the phenotype of HNSCC cells by decreasing their proliferation and migratory capacity whilst up-regulating the expression of pro-apoptotic genes and increasing levels of apoptosis.
Discussion
There is evidence to suggest that altered folate status influences the risk of cancer and there are plausible mechanisms to support this [
39,
40]. Studies of methyl donors other than folate have indicated a complex interaction between methyl donor availability and function, whereby deficiency in one methyl donor may be compensated for by another, to maintain a functional methyl cycle [
41,
42]. However, very few studies have examined effects of depletion of more than a single methyl donor, where the opportunity for such compensation is minimised. In this study, we successfully developed a model of functional methyl donor deficiency that was characterised by an increased concentration of homocysteine in the extracellular medium and decreased intracellular levels of methyl donors, indicating disturbance to the methyl cycle [
36,
37]. The time-related change in extracellular homocysteine was different in the two cell lines used. This may reflect cell-specific sensitivity to methyl donor depletion, including differential up-regulation of folate receptors and/or homocysteine metabolism [
43,
44].
Cancer cell motility is important for tumour cell invasion during HNSCC progression and subsequently leads to the development of local or distant site metastases. Our data demonstrate for the first time that HNSCC cell migration is significantly impaired in methyl donor depleted conditions. Moreover, the cellular response is unlikely to be due to metabolic shock as the metabolic status of deplete and control cells was similar. This is an important finding as increased cell migration has been associated with poor prognosis in HNSCC [
45] and the ability to immobilise cancer cells is an important target for cancer drug development. There has been relatively little research into the effects of methyl donor depletion on tumour cell migration and findings suggest that effects may be cell specific. Human HCT116 colon carcinoma cells showed increased migration in folate-deplete conditions [
46], whereas Graziosi et al. reported reduced cell migration in methionine-depleted gastric cancer cells [
47]. The effects on migration may be mediated by epigenetic alterations in the expression of genes important for cell locomotion. The actin cytoskeleton plays a central role in cell movement and studies on the effects of methyl donor depletion in various cell types suggest that genes associated with the regulation of the actin cytoskeleton are invariably down regulated. Duthie et al. reported reduced expression of proteins important for cytoskeleton organization in folate-depleted human colonocytes [
8], whilst Jhaveri et al. showed down-regulation of cytoskeleton 14 in folate-depleted human KB (HeLa) cells [
48].
The observed reduction in cell proliferation in methyl-donor deplete UD-SCC2 cells could, in part, be explained by a shortage of purine and pyrimidines that are essential for DNA production, as methyl donors are required for the synthesis of these molecules. An alternative explanation may be that methyl donor-depleted cells are arrested at G1 in the cell cycle, as has previously been reported in human and murine B cells with reduced levels of methionine and SAM [
49]. The effect on cell proliferation observed in this study was dose-dependent and the rate of cell proliferation was restored following repletion. The decrease in cell proliferation rate detected in methyl donor-deficient cultures did not account entirely for the reduction in cell number observed. Cell death, through apoptosis, also increased in the methyl donor depleted conditions, while cell necrosis remained unchanged.
The role that folate plays in DNA synthesis and cell division is well understood and this is the basis for the use of anti-folate drugs in cancer treatment. A decrease in methyl donor availability is expected to lead to a fall in cell proliferation and this is what we observed. Evidence linking folate status with apoptosis is limited and inconsistent [
50‐
52]. An increase in apoptosis is understood to have beneficial effects in cancer and a low level of apoptosis in HNSCC is associated with poor prognosis [
53]. Low availability of methyl donors, in particular folate, may lead to reduced DNA synthesis and thus inhibit tumour cell proliferation of established tumours.
DAPK1 and
PUMA are pro-apoptotic genes known to be up-regulated when cells encounter stress [
54,
55]. We hypothesised that activation of pro-apoptotic genes mediates the increase in apoptosis observed in methyl donor depleted cells.
DAPK1 and
PUMA mRNA levels significantly increased following methyl donor depletion and, in the case of DAPK1, this was associated with an increase in protein levels as well, a response that was reversed upon methyl donor repletion. Interestingly, cells that were methyl donor deficient also displayed reduced levels of phosphorylated DAPK1, the inactive form of the protein, compared to cells cultured in complete media. These data suggest that upon prolonged methyl donor depletion, cells respond by not only increasing the gene expression of
DAPK1 but also by increasing the active, non-phosphorylated form of the protein that then drives the pro-apoptotic pathway leading to the programmed cell death of HNSCC cells.
A functionally impaired methyl cycle was associated with effects on the expression of DNMTs and TET1. The expression of
DNMT1, the DNMT involved in maintaining DNA methylation status following cell division [
19], was not influenced by a reduction in methyl donor status in UD-SCC2 cells. Reports of effects of folate depletion on DNMT1 expression in cultured cells are inconsistent; Stempak et al., [
56] reported a down-regulation in mouse fibroblasts and human colon cancer cells, whilst Hayashi et al. [
32] observed an up-regulation in response to folate depletion of the same human colon cancer cells. In contrast to
DNMT1, expression of
DNMT3a was increased in methyl donor deplete UD-SCC2 cells and this effect was reversed upon methyl donor repletion. A similar direction of change was seen for
DNMT3b, although the effect was not statistically significant. Others have reported an increase in the expression of
DNMT3a in response to methyl donor depletion in animal models and the relative resistance of
DNMT3b to methyl donor depletion has also been documented in these studies [
57,
58]. However, colon cancer cells depleted of folate alone are reported to show a decrease in
DNMT3a [
32,
56], and Poomipark et al. reported a decrease in
DNMT3a and
DNMT3b in a folate and methionine-depleted cervical cancer cell line C4II [
22], adding to the evident inconsistency of response to single or multiple methyl donor depletion in different experimental systems. The fact that we were able to demonstrate reversibility of the effect on
DNMT3a expression is a strong indicator of the causal relationship between methyl donor availability and
DNMT3a expression in HNSCC cells. It is plausible that in methyl donor deplete HNSCC cells, expression of
DNMT3a and
TET1 (responsible for DNA demethylation) is increased as an adaptive response, to scavenge cellular sources of methyl groups and maintain epigenetic control of gene transcription. Deletion of
DNMT3a has been shown to promote lung tumour progression [
59] and expression has been shown to be higher in leukaemia compared with control cells [
60]. Sun et al. recently showed that inhibition of
DNMT3a in cervical cancer cells induced apoptosis, suggesting a cancer promoting effect of
DNMT3a [
61]. In our study, increased expression of
DNMT3a and
TET1 was associated with increased apoptosis; whether these responses to methyl donor depletion are causally linked is not clear.
Methylation of both host and viral DNA is thought to be important in HPV-positive cancers, with distinct methylation patterns observed in HPV-positive and negative disease [
62]. It has previously been shown that the
DAPK1 promoter is hyper-methylated in oral cancer cells and the prevalence of
DAPK1 methylation in oral cancer ranges from 7 to 68% [
18]. This led us to hypothesise that in methyl donor deplete conditions, the lack of readily available methyl groups would result in loss of methylation within the
DAPK1 promoter, leading to increased
DAPK1 gene expression, increased protein production and activity, and, therefore, increased apoptosis. The previous studies have shown variations in gene methylation patterns following folate depletion in vitro depending on the cell type and gene of interest [
39]. However, we found that the
DAPK1 promoter was not methylated in UD-SCC2 HPV-positive cells or in several other HNSCC/dysplastic (HPV-negative) cell lines. This was not due to a lack of specificity of the PCR primers used as SiHa cervical carcinoma cells displayed methylated
DAPK1 as previously described [
63]. These data suggest that other mechanisms regulating
DAPK1 gene expression and activation are at play in these cells or that the
DAPK1 gene is under epigenetic control at other sites within the promoter region that were not covered by our primer sequences.
The evidence presented from this study indicates that methyl donor depletion of HNSCC cell lines results in a less aggressive cancer cell phenotype. Cancer cells which display reduced proliferation, increased apoptosis, and decreased migration are highly likely to produce less aggressive tumours
in vivo. While low folate consumption is considered to be a risk factor for cancer at various sites, this may not be true for established cancers or in cells that contain microscopic neoplastic foci where folate could promote rather than prevent carcinogenesis through the provision of nucleotide precursors required for growth of rapidly dividing cancer cells [
64,
65]. Indeed, there is increasing evidence from studies in animals and humans to suggest that whilst adequate folate status may be protective against cancer initiation, folate deficiency can slow cancer growth, whilst high intakes may enhance growth of certain cancers, including colorectal, breast, and prostate cancer [
64‐
66]. Recent evidence extends this observation to other nutrients involved in the methyl cycle [
67].
This study has demonstrated that depleting HNSCC cells of methyl donors has beneficial effects in terms of the cancer phenotype and provides mechanistic data to support evidence suggesting that high intakes of methyl donors may have adverse effects in established cancers or where pre-cancerous lesions exist.