Background
There are two features of melanoma that suggest a defect in DNA repair, an extremely high mutation load, indicative of unrepaired UV-induced DNA damage [
1], and innate resistance to DNA damaging agents such as platinum based chemotherapies [
2]. These can both be connected to the nucleotide excision repair (NER) pathway, the DNA repair system that is responsible for the removal of DNA damage that distorts the DNA helix, including UV photoproducts and platinum strand crosslinks [
3].
The NER pathway consists of approximately 30 proteins that remove helix distorting lesions through for steps: a) damage recognition, b) unwinding of the DNA locally around damage, c) incision of damaged DNA by endonucleases, and d) DNA resynthesis and ligation [
4]. There are two branches of damage recognition that lead to the common repair pathway: transcription coupled repair (TCR) and global genome repair (GGR). TCR is linked to active gene transcription and is initiated when RNA polymerase is stalled at DNA damage during transcription. GGR however is not dependent on transcription and scans the entire genome including both active and silent genes, and non-transcribed regions using DNA damage binding proteins XPC and UV-DDB (DDB1 and DDB2) [
5].
Several previous studies have found an association between high or low levels of NER protein and mRNA levels and platinum chemoresistance (reviewed in [
6]). The NER component ERCC1 has been extensively studied as a predictive biomarker for response to chemotherapy. To date, high levels of ERCC1 before platinum chemotherapy have been associated with poor response in melanoma [
7], non-small cell lung cancer [
8,
9], head and neck [
10], gastric [
11,
12], bladder [
13] and oesophageal cancer [
14]. Small molecule inhibitors of the ERCC1-XPF complex have been developed and shown to potentiate cisplatin efficacy in the A375 melanoma cell line [
15] and H460 and H1299 lung cancer cell lines [
16]. Further to this, in a melanoma mouse xenograft model loss of ERCC1 resulted in sensitivity to cisplatin [
17]. To date, the study of NER in relation to platinum chemoresistance has largely focused on ERCC1.
In addition to the evidence supporting ERCC1 as a biomarker of platinum chemoresistance, our previous research has shown that the GGR damage detection branch of NER, does not function correctly in melanoma. We have found that the three GGR components XPC, DDB1 and DDB2 do not respond to UV treatment in melanoma cell lines, resulting in reduced repair of UV-induced DNA damage [
18,
19]. We also identified that melanoma tumours with low
XPC expression have significantly shorter survival [
18]. The functional loss of NER in melanoma has also been reported by Belanger et al. [
20] and could account for the high UV mutation signature of melanoma. This was further supported by analysis of melanoma genomes, that concluded somatic mutations active gene promoters is caused by a decrease in the levels of nucleotide excision repair (NER) activity [
21,
22]. We have also shown that these same GGR transcripts do not respond to the platinum agent cisplatin in melanoma compared to normal melanocytes, which may be responsible for resistance to this treatment [
23].
A role for GGR in melanoma development and chemotherapy resistance may come from the broad functions it has in controlling the DNA damage response. Damage recognition by XPC and DDB2 leads to activation of other pathways that control cell cycle and apoptosis, in addition to NER. XPC and DDB2 are involved the activation of the checkpoint signalling protein ATR in response to UV-induced DNA damage [
24]. Both proteins also play a role in apoptosis in response to DNA damage [
25,
26]. Additionally, XPC deficient cells have a significantly reduced cisplatin-mediated p53 and apoptotic response [
27,
28], suggesting that DNA damage recognition is an important part of cisplatin induced apoptosis. Therefore, loss of GGR, in particular XPC, in melanoma could play a role in resistance to platinum chemotherapies.
The underlying mechanism that is responsible for the GGR deficiency seen in melanoma is yet to be identified. To date, somatic mutations in XPC, DDB1 and DDB2 have rarely been reported in melanoma tumours. We reported that upstream regulators of GGR including p53, BRCA1 and PCNA are not responsible [
18,
23,
29]. One possible mechanism affecting the expression of these genes is dysregulation of epigenetics such as DNA methylation. Aberrant changes in DNA methylation patterns are a key feature of many cancers including melanoma, where global hypomethylation increases DNA instability and local hypermethylation of promoter CpG islands can silence the expression of many tumour suppressor genes [
30]. DNA methylation is one of the best studied epigenetic modifications and has high potential in cancer research as a target due to DNA methyltransferase inhibitors such as decitabine (5-aza-2’deoxycytidine) that can demethylate and reverse silencing of genes [
31].
To date there has only been only one study to investigate the methylation of XPC in melanoma. A mouse model found that melanocytes with BRAF
V600E and p14ARF
−/− background developed melanoma in response to UV radiation, with impaired DNA repair capacity due to reduced XPC expression from promoter hypermethylation [
32]. However, as this study only examined the methylation of three CpG sites within the XPC promoter, further investigation is warranted.
More recently, the importance of CpG island shore methylation altering the expression of genes in cancer [
33], has been reported. Methylation patterns within the CpG island shores of XPC have not been investigated. Methylation in these regions has a strong effect on the expression of genes and several studies have now identified changes altering expression of genes in various cancers [
34‐
36].
As there is evidence of silencing of XPC by methylation in melanoma the aim of this study was to investigate the methylation pattern of the XPC promoter region, including the CpG island and flanking shores, and its effect on gene expression in our melanoma cell lines that display reduced GGR. We also examined if methylation patterns could be altered by demethylation and restore XPC function, therefore reinstating platinum chemotherapy sensitivity.
Methods
Cell culture and treatment
Four melanoma cell lines were used in this study: MM200, Sk-mel-28, Me4405 and Mel-RM. The source, tumour status and p53 status of each cell line has been previously described [
37‐
39]. A human neonatal, medium pigment HEMn-MP melanocyte cell line was purchased (Cascade Biologics, USA and ThermoFisher, USA). Cell lines were authenticated as previously described [
23] using GenePrint 10 (Promega). Mycoplasma was tested and not detected at 6 month intervals using MycoSEQ mycoplasma detection kit (Thermo Fisher Scientific). Melanoma cell lines were cultured in high glucose DMEM (5% FBS) (Gibco, Thermo Fisher Scientific) and HEMn-MP was cultured in Medium 254 (Gibco, USA). All cells were incubated at 37 °C 5% CO
2 (Hera Cells 240, Thermo Scientific).
Carboplatin (Sigma-Aldrich) and decitabine (5-aza-2′-deoxycytidine) (Sigma-Aldrich) were resuspended in MilliQ H
2O at 10 mg/mL and 1 mM respectively, with decitabine stored at − 80 °C. For treatment decitabine was diluted in cell culture medium to either 10 μM or 0.26 μM where indicated. Cell lines were treated with decitabine for 72 h with cell culture media replaced every 24 h with fresh media and decitabine. Carboplatin was diluted in cell culture medium to 8 μg/mL and cells treated for 48 h. For combination treatment cell lines were treated first with 0.26 μM decitabine for 72 h followed by 48 h of 8 μg/mL carboplatin. These doses were chosen as they were based on plasma concentrations of each drug when used as chemotherapy agents [
40,
41].
Global DNA methylation quantification
Global DNA methylation levels were quantified using a 5-mC DNA ELISA Kit (Zymo Research) as per manufacturer’s instructions. DNA from melanoma cell lines before and after decitabine treatment was extracted by Quick-DNA Universal Kit (Zymo Research). 100 ng of genomic DNA and methylated standards were bound to an ELISA plate and methylated DNA was detected with antibodies to 5-methylcytosine, quantified by colorimetric analysis.
Gene expression analysis
Before and following treatment at specified time points RNA was collected by phenol chloroform extraction. RNA (1 μg) was reverse transcribed in triplicate using the High Capacity Reverse Transcription Kit (Thermo Fisher Scientific) and the resultant cDNA was diluted 1:20 as previously described [
23]. Relative expression of
XPC was measured in triplicate and normalised to the geometric mean of three housekeeping genes,
GAPDH,
ACTB and
18S rRNA using TaqMan gene expression assays and a Viia7 system (Applied Biosystems). Relative expression was calculated using 2
-ΔCt.
Western blotting
Nuclear protein fractions were obtained using the NucBuster protein extraction kit (Merck Millipore). Protein lysate (40 μg) was added to 4X SDS-sample loading buffer (250 mM Tris-HCl, pH 6.8, 4% LDS, 40% (w/v) glycerol, 0.02% bromophenol blue, 15% beta-mercaptoethanol) and denatured by boiling for 5 min. Samples were loaded onto 4-20% TGX precast polyacrylamide gels (Bio-Rad Laboratories) and run at 150 V (constant voltage) in Tris-Glycine buffer (25 mM Tris, 192 mM glycine, 0.1% SDS). Proteins were transferred onto nitrocellulose using the TransBlot Turbo system (high-MW 10 min; Bio-Rad Laboratories) and visualised using Ponceau S (0.1% (w/v) Ponceau S in 5% acetic acid; Sigma-Aldrich). Following transfer, blots were blocked in 5% skim milk for 1 h at room temp. XPC was detected using anti-XPC rabbit polyclonal antibody (H-300) (1:200; sc-30,156 Santa Cruz Biotechnology, Inc.) and anti-TATA binding protein (TATA-BP) (1:1000 ab51841, Abcam) was used as a nuclear loading control. Primary antibodies were incubated at 4 degrees overnight. Blots were washed three times for 5 min in PBS-T then incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (goat anti-rabbit 170-6515, Bio-Rad Laboratories). Blots were washed as done previously then proteins detected by chemiluminescence using Clarity Western ECL reagent (Bio-Rad) and imaged using the ChemiDoc MP system (Bio-Rad Laboratories). Image processing and densitometry analysis was performed on all blots using ImageJ. Data was normalised to TATA-BP and expressed as fold induction from baseline.
Bisulfite sequencing of XPC
DNA was bisulfite converted using an EZ DNA Methylation Kit (Zymo Research) according to manufacturer’s instructions. The CpG island and surrounding shores of
XPC promoter was amplified by PCR using Taq Polymerase (Invitrogen) and the primers (Additional file
1: Table S1). All PCRs were performed in triplicate for all cell lines before and after treatment with 0.26 μM decitabine. PCR products were cleaned with Exonuclease I and Alkaline Phosphatase (Thermo Fisher Scientific). For sequencing, fragments for each sample were pooled and libraries prepared using the TruSeq Nano DNA Library Prep kit (Illumina). Sequencing was performed on an Illumina MiSeq and analysed using Bismark [
42].
Flow cytometry
After treatment with 0.26 μM decitabine, 8 μg/ml carboplatin and in combination, both attached and detached cells were collected. Apoptotic cells were quantified after drug treatment using an Annexin V Apoptosis Detection Kit (BD Biosciences) following manufacturers instruction performed on a BD FACSCanto II flow cytometer (BD Biosciences). 1 × 106 cells before and after treatment were washed and stained with 7-AAD and PE conjugated Annexin-V for 15 min in the dark. Data was analysed on FlowJo v10 (FlowJo, LLC). Apoptotic cells were quantified as the percentage of cells that stained positive for Annexin-V and double Annexin/7-AAD positive cells.
Cell proliferation and senescence detection
Cellular proliferation after treatment was measured using a CellTitre-Glo Luminescent Cell Viability Assay kit (Promega) according to manufacturer’s instructions. Cells were seeded in 96-well plates at 5 × 103 cells per well overnight before drug treatment and luminescence measured on a Cytation 3 plate reader (BioTek). After combination treatment senescence was measured by β-galactosidase staining using an Abcam Senescence Detection Kit (Abcam). Cells were plated and fixed in 6-well plates after combination treatment and stained with X-gal. Positively stained cells were identified under a light microscope.
siRNA knockdown
The expression of XPC was knocked down after decitabine treatment using siRNA purchased from Dharmacon (siGENOME Human XPC, D-016040-04-0010). Transfections were carried out in the last 24 h of decitabine treatment with 25 nM of XPC siRNA in OptiMEM medium (Gibco) using Lipofectamine RNAiMAX (Invitrogen) according to manufacturer’s instruction. An NTC siRNA (siCONTROL Non-targeting siRNA #4, Dharmacon) was used as a control for transfection in identical conditions to XPC siRNA.
Statistical analysis
Statistical analysis was performed using GraphPad Prism 6 (GraphPad Software). Non-parametric Mann-Whitney tests were used to assess differences between groups. A p-value of < 0.05 was considered statistically significant.
Discussion
We have previously reported that XPC does not respond to DNA damage in melanoma [
18,
19,
23], which may be a key component of melanoma development from UV exposure and resistance to platinum chemotherapies. The cause of this loss has not been discovered, but some evidence exists of DNA methylation altering XPC expression [
32]. In this study, we investigated the effect of DNA methylation on the expression of
XPC in melanoma. Here we have shown that, while methylation may not be the cause of reduced XPC expression, treatment of melanoma cell lines with decitabine can restore expression, and allow for strong induction in response to carboplatin. The sequential treatment of melanoma cell lines with decitabine and carboplatin also increased apoptosis and decreased cell proliferation, suggesting that this combination can overcome platinum resistance in vitro.
Bisulfite sequencing revealed that the CpG island of the XPC promoter was not methylated in the cell lines. This methylation patterns in our study are consistent with data available from the Cancer Genome Atlas (TCGA) showing that the
XPC CpG island is not methylated in melanoma tumours. The same pattern of methylation was also seen in a melanocyte cell line (Additional file
3: Fig. S2) suggesting it is a lineage-specific epigenetic pattern.
Methylation within the CpG island shores was present in melanocytes and the 4 melanoma cell lines in this study. The methylated regions were partially demethylated by decitabine in the melanoma cell lines. As melanocytes have low replication rates in vitro and do not replicate in vivo the blocking of methylation does not occur to the same extent as in melanoma cells, therefore the effect of decitabine is not seen in melanocytes. More research is needed to confirm that demthylation of the XPC CpG island shores is responsible for increased XPC expression in response to decitabine. It is possible that demethylation of the shores allows for some other elements such transcription factors access to XPC which may result in the increased expression in response to decitabine. One possible explanation is that an upstream regulator of XPC expression is also being demethylated by decitabine and can induce XPC expression.
Two doses of decitabine were used in this study. The first (10 μM) was chosen as it is a high dose that would ensure demethylation across the genome, while the second (0.26 μM) represented a pharmacological dose [
40]. 10 μM decitabine induced a greater increase in
XPC when compared with 0.26 μM. This may have been due to the fact that high doses of decitabine can induce DNA damage by prolonged binding of DNMT1 leading to double stranded breaks [
47]. It could be that in combination with demethylation, this type of damage signalled for
XPC upregulation. The lower dose of 0.26 μM should not have induced as great of levels of damage and as such could explain why only two cell lines Mel-RM and Me4405 increased
XPC expression after 0.26 μM decitabine.
To identify if platinum-induced DNA damage would induce XPC following demethylation, we treated the cells with carboplatin following demethylation with 0.26 μM decitabine. Combined treatment resulted in a significantly greater XPC response in 3 of 4 cell lines. While low dose decitabine was not enough to increase XPC expression alone it may demethylate a particular region of XPC or an upstream target that then allows XPC to respond to the DNA damage signal caused by carboplatin.
CpG island shores have been gaining more consideration in the past few years after being confirmed as one of the major regions for differential methylation in cancer [
33,
48]. Although the specific function of shore methylation has not yet been identified, changes in methylation are reported to affect expression of genes. The
HOX10 gene has CpG island shore methylation that is associated with transcriptional repression in breast cancer [
36]. Methylation in the shores varied from 5 to 95% and was inversely correlated with expression; those with higher shore methylation had lower expression. A similar pattern was found in the caveolin-1 (
CAV1) gene [
34]. This study found a negative relationship between
CAV1 shore methylation and expression in breast cancer. This relationship has been found further in other genes in other cancers [
49,
50]. All this data surrounding shore methylation indicates that shore methylation is associated with transcriptional repression but the exact molecular mechanisms of this relationship are not understood.
Regardless of the dynamics of
XPC demethylation and expression, this study revealed some exciting results with translational potential. Demethylation with decitabine increased the sensitivity of melanoma cells to the growth inhibitory and apoptotic effects of carboplatin, which is typically ineffective in melanoma [
51]. The cell lines that showed an increased XPC expression also had significantly higher levels of apoptosis and cell death. This was combined with a decreased rate of cell proliferation, and senescence in some cell lines. These results were much greater than those compared to carboplatin alone. Sk-mel-28 did not induce
XPC or apoptosis in response to decitabine and carboplatin. Sk-mel-28 carries a mutation that results in constitutively activated p53 [
52] which may account for the lack of response, but requires further investigation. Carboplatin is a slower acting drug compared to other platinum chemotherapies such as cisplatin and does not induce as much DNA damage [
53]. This means an even greater response could be seen if a second carboplatin dose was added after demethylation.
The importance of XPC in the response to combined treatment was investigated. When XPC was knocked down with siRNA during decitabine treatment minor changes were seen in apoptosis and proliferation. Given the effect of XPC knockdown on apoptosis was minor, despite robust attenuation of XPC, the effect of exogenous overexpression of XPC expression is required to further clarify if the effects that are being observed are indeed related to XPC expression.
The data reported herein suggests
XPC is not the only driver of the responses seen. This could be expected as there are numerous and complex pathways that are involved in cell death and proliferation, which combined with the fact decitabine demethylates globally, may lead to many changes in various pathways. It is unlikely that one single demethylated target will drive the response alone. The melanocyte-lineage specific master regulator, MITF may additionally contribute to the apoptotic response to decitabine and carboplatin. Although this requires further investigation, MITF is involved in melanoma proliferation and survival (reviewed in [
54]) and has been associated with DNA repair in melanoma [
55]. Regardless of the role of the
XPC response and its importance, here we have identified a potential combination treatment in melanoma.
While some studies have examined the effect of decitabine on melanoma growth and apoptosis, ours is the first study to examine the combination of demethylation with platinum therapy within melanoma. A study in 2011 showed that decitabine has a cytotoxic effect on melanoma that is independent of apoptosis [
56]. While apoptosis remained low from treatment, decitabine induced G2/M cell cycle arrest, inhibiting the growth of a melanoma cell line. Decitabine was also able to induce differentiation, marked by the formation of melanocyte-like dendritic structures. Similar results are seen in another study that expands treatment to a panel of cell lines and a mouse xenograft model [
57]. In response to decitabine the majority of cell lines showed reduced proliferation and markers of melanocytic differentiation, including dendritic formation, decreased nuclear-cytoplasmic ratio, and increased melanin production. Again, apoptosis was not a significant response to decitabine treatment. The response was discovered to be independent of p53 and
CDKN2A. Differentiation was driven by demethylation of the melanocyte differentiation driver
SOX9 promoter and subsequent increased expression. Exit from the cell cycle and reduced proliferation was a result of upregulated p21 (
CDKN1A), and p27 (
CDKN1B). Whole transcriptome analysis in response to decitabine [
58] confirms a role for p21 and proposes a role for the WNT signalling and β-catenin pathway in both resistance to decitabine and supressing apoptosis from decitabine treatment alone.
While a new idea in melanoma, combining demethylation treatment with platinum chemotherapies has been studied in other cancer types with positive results. One particular study examining multiple combinations found that decitabine had the greatest combination potential with platinum compounds, out of a screen of 16 drugs [
59]. Decitabine can also rescue cisplatin resistance in head and neck squamous cell carcinoma (HNSCC) by demethylating of panel of genes associated with HNSCC resistance, and enhancing the cytotoxic and apoptotic effects of cisplatin [
60]. This combination is also showing promise in ovarian cancer tumours. One study showed combined treatment resulted in a high response rate and progression free survival in platinum resistant patients [
61]. In this study pathways enriched for demethylation include pathways in cancer, WNT signalling, and apoptosis, and some demethylated genes such as
HOXA10 and
RASSF1A correlated with progression free survival. Similar results have been found in neuroblastoma [
62], renal carcinoma [
63] and non-small cell lung cancer [
64]. This makes combined decitabine and carboplatin treatment worthy of investigation in melanoma as it shows potential in other platinum resistant cancers.
The benefits of a combination decitabine and carboplatin treatment are not limited to DNA damaging and apoptotic mechanisms. One major challenge in cancer treatment is the ability of tumours to escape from immune detection and prevent efficient T-cell response to cancer cells [
65]. Emerging research shows both carboplatin and decitabine may also have therapeutic benefits by enhancing the immunogenicity of tumour cells, allowing the immune system to target the cancer. This offers an interesting potential to be combined as a priming regime for immunotherapy.
Platinum chemotherapy compounds including carboplatin can induce a combination of stress and cell death that can initiate a tumour-specific immune response. Firstly, platinum therapy induces endoplasmic reticulum (ER) stress that results in calreticulin exposure on the cell surface, acting as a signal for dendritic cells [
66]. Secondly, cells undergoing apoptosis from platinum compounds release ATP which acts as a chemoattractant for dendritic cells and macrophages to the tumour site [
67]. Thirdly, during cell death, high mobility group protein 1 (HMGB-1) is released from the nucleus. This binds to toll like receptor 4 (TLR4) on dendritic cells leading to cytokine secretion and cross presentation [
68]. Altogether these factors result in the attraction and maturation of dendritic cells. Additionally drugs such as carboplatin, but not other chemotherapies, have been shown to downregulate the programmed death ligand 2 (PDL-2) on both dendritic cells and tumour cells, resulting in enhanced antigen specific T-cell activation, through the IL-4/STAT6 pathway [
69]. This can result in tumour antigen presentation in the lymph nodes leading to increased activated tumour-specific T-cells which will mount an immune response against the tumour [
70].