ABSTRACT
ABSTRACT
Epigenetic control of gene expression is a major determinant of tumor phenotype and has been found to influence sensitivity to individual chemotherapeutic agents. Glutathione peroxidase 3 (GPX3, plasma glutathione peroxidase) is a key component of cellular antioxidant regulation and its gene has been reported to be methylated in specific tumor types. GPX3 role in oxidative damage has been associated with sensitivity to platinums in other tumors but its importance in colorectal cancer (CRC) has not been determined. We examined the role of GPX3 methylation in colorectal carcinoma in determining sensitivity to platinum drugs using primary tumor specimens, cell lines, knockdown cell lines, and tumor cell line xenografts. We find GPX3 promoter region methylation in approximately one third of CRC samples and GPX3 methylation leads to reduced GPX3 expression and increased oxaliplatin and cisplatin sensitivity. In contrast, in cell lines with high baseline levels of GPX3 expression or with the ability to increase GPX3 expression, platinum resistance is increased. The cisplatin IC50 in GPX3-methylated cell lines is approximately 6-fold lower than that in GPX3-unmethylated lines. Additionally, knockdown cell lines with essentially no GPX3 expression require N-acetylcysteine to survive in culture underscoring the importance of GPX3 in redox biology. In vivo, GPX3 methylation predicts tumor xenograft sensitivity to platinum with regression of GPX3 knockdown xenografts with platinum treatment but continued growth of GPX3 wild type xenografts in the presence of platinum. These studies demonstrate the importance of GPX3 for CRC cells resistance to platinums and the potential utility of GPX3 methylation status as a predictive biomarker for platinum sensitivity in CRC.
Abbreviations
| 5FU | 5-fluorouracil |
| CHFR | checkpoint with forkhead and RING finger domains |
| CRC | colorectal cancer |
| GPX3 | glutathione peroxidase 3 |
| MGMT | O-6-methylguanine-DNA methyltransferase |
| MMR | mismatch repair |
| MSI | microsatellite unstable |
| MSP | methylation-specific PCR |
| MSS | microsatellite stable |
| MLH1 | mutL homolog 1 |
| qPCR | quantitative real-time PCR |
Introduction
An estimated 50,000 people die from colorectal cancer (CRC) each year in the US.1 Molecular heterogeneity in CRC is an important factor in determining sensitivity or resistance to individual therapies for metastatic disease, including the association of mutations of KRAS with resistance to cetuximab.2 Epigenetic changes are critical determinants of normal and tumor cellular phenotypes, and represent heritable changes in gene expression not caused by changes in nucleotide sequence.3 These changes can also be associated with therapeutic sensitivity.3
In particular, studies have demonstrated the importance of epigenetic changes in determining tumor resistance or sensitivity to specific chemotherapeutic agents through targeting pathways that are epigenetically dysregulated. Esteller et al. demonstrated that the O-6-methylguanine-DNA methyltransferase (MGMT) promoter is hypermethylated in 40% of high grade gliomas and this was associated with tumor regression, increased disease-free survival, and overall survival in these patients when treated with a carmustine based regimen.4 Subsequent studies have confirmed MGMT hypermethylation as a predictor of response to the alkylating agent temozolomide and MGMT methylation testing is now available to guide treatment decisions in high grade gliomas.5-7
Epigenetic dysfunction is well described in CRC. For example, promoter hypermethylation of the DNA mismatch repair gene MLH1 leads to microsatellite instability (MSI) in a subset of sporadic CRCs8 that we have previously reported may predict sensitivity to gemcitabine.9 In addition, we recently demonstrated that promoter hypermethylation of checkpoint with FHA and ring finger (CHFR), a mitotic checkpoint gene that causes delayed entry into metaphase under conditions of mitotic stress,10 is predictive of docetaxel sensitivity in CRC cells.9 These studies suggest the potential to uncover a subset of patients who may be sensitive to therapies that have not thus far been effective in unselected populations of CRC. However, this same approach might also identify patients with CRC who are more sensitive or resistant to therapies already in clinical practice, and thereby allow better selection among standard therapies. Since the majority of patients with metastatic CRC are treated with a fluoropyrimidine together with irinotecan or oxaliplatin, a way to choose among these later agents would be clinically useful.
Epigenetic silencing of glutathione peroxidase 3 (GPX3), a member of a family of selenoproteins with important antioxidant roles,11 has been reported in a variety of tumor types including gastric, prostate, head and neck, ovarian, and bladder tumors.12-14 This antioxidant role has led to exploration of GPX3 as a determinant of chemotherapeutic activities, with evidence for a relationship to platinum agent sensitivity in ovarian clear cell cancers.15 These compounds, particularly oxaliplatin, are commonly used in the treatment of gastrointestinal malignancies, including CRC, but not all patients respond to these agents, and most patients develop resistance over time.16-18 An important role for GPX3 has been shown in CRC development, in that studies of GPX3 −/− mice develop increased numbers of colonic tumors and a reduction in GPX3 expression in vitro was associated with increased reactive oxygen species (ROS), DNA damage, and apoptosis following oxidative stress.19 These features would also be expected to alter sensitivity to platinum agents, since, in addition to platinum-DNA adduct formation,20 these drugs disrupt cellular redox systems as another mechanism of activity.21 With this important role of GPX3 in managing the oxidative stress response, we hypothesized that loss of expression of GPX3 is associated with GPX3 methylation and might predict platinum sensitivity in CRC.
Results
The Cancer Genome Atlas (TCGA) analysis: GPX3 methylation represses gene expression and is associated with MSI status
As a first examination of GPX3 alterations in colorectal cancer, we queried The Cancer Genome Atlas (TCGA) data for mechanisms of inactivation. To explore genetic alterations, using cBioPortal,22,23 we found no evidence for mutational inactivation of GPX3, a result consistent with the TCGA colorectal publication.24 Only 3 of 224 sequenced colorectal carcinomas have mutations, and all were missense mutations of neutral or medium predicted functional impact score. We then analyzed GPX3 expression and alterations in DNA methylation in the promoter region. Five promoter region probes: cg21504918, cg26638444, cg21516478, cg10802379, and cg17820459 all show low level methylation in normal colon tissue and increased methylation in a subset of colorectal tumors. The prevalence of GPX3 methylation in 292 colorectal cancer TCGA samples is 30% (87 of 292 with β value higher than 0.2, a value higher than all normal tissues). We calculated the Spearman's rank correlation coefficient for GPX3 promoter DNA methylation with expression, using probe cg26638444, which is the closet to the transcriptional start site. Fig. 1A shows that colorectal cancers with promoter region methylation have lower levels of GPX3 mRNA expression relative to normal colon and colon tumors without DNA methylation. Since colorectal cancer includes both microsatellite stable (MSS) and instable (MSI), and these phenotypes may be associated with response to chemotherapies including oxaliplatin,25,26 we examined whether GPX3 promoter methylation differed between these phenotypes. Indeed, we found that GPX3 methylation was associated with MLH1 promoter methylation (Fig. 1B), which results clinically in microsatellite instability (MSI+). This would suggest that if GPX3 silencing was associated with oxaliplatin sensitivity, this might differ between MSI and MSS patients.
Published online:
21 July 2017Figure 1. A: GPX3 promoter methylation and expression are inversely correlated in primary colorectal cancer. TCGA dot plot demonstrating the relationship between GPX3 methylation (β value on X axis) at cg2663844 located in the promoter region CpG island and expression (y axis) in TCGA CRC samples. Note, all normal colon tissues (triangles) express high levels of GPX3 and are unmethylated (low β value), while a subset of colon cancers have β values above normal (> 0.2) and have reduced expression (Spearman's correlation = −0.43). B: GPX3 promoter methylation correlates with MSI phenotype. TCGA dot plot at cg2663844 demonstrating the association between GPX3 methylation (Y axis) and MSI phenotype (by MLH1 promoter methylation status) (X axis) in TCGA colon samples. Normal tissue at left, Microsatellite (MSI+) positive samples with higher median methylation (β value) compare with microsatellite stable (MSI-), T-test P value = 2.32e−06.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0001_oc.gif"})
Figure 1. A: GPX3 promoter methylation and expression are inversely correlated in primary colorectal cancer. TCGA dot plot demonstrating the relationship between GPX3 methylation (β value on X axis) at cg2663844 located in the promoter region CpG island and expression (y axis) in TCGA CRC samples. Note, all normal colon tissues (triangles) express high levels of GPX3 and are unmethylated (low β value), while a subset of colon cancers have β values above normal (> 0.2) and have reduced expression (Spearman's correlation = −0.43). B: GPX3 promoter methylation correlates with MSI phenotype. TCGA dot plot at cg2663844 demonstrating the association between GPX3 methylation (Y axis) and MSI phenotype (by MLH1 promoter methylation status) (X axis) in TCGA colon samples. Normal tissue at left, Microsatellite (MSI+) positive samples with higher median methylation (β value) compare with microsatellite stable (MSI-), T-test P value = 2.32e−06.
GPX3 expression correlates with methylation status in primary xenografts specimens
We examined colorectal cancer patient-derived tumor xenografts, established as previously described,27,28 for GPX3 promoter methylation status and expression. The use of xenografts allows the remaining normal tissue present in the tumor to not contaminate expression or methylation analysis. Methylation specific PCR (MSP) was performed on DNA isolated from patient xenograft tissue and assessed for GPX3 promoter methylation. GPX3 was methylated at a frequency similar to primary tumors (3/12) (Fig. 2A). As we had observed in TCGA tumors, GPX3 methylation correlates with markedly reduced or absent expression by qPCR (Fig. 2B), with complete loss of expression in xenografts 10 and 24. Unmethylated xenografts expressed varying amounts of GPX3 mRNA, but was statistically higher than the methylated xenografts (P = 0.01, two-tailed Student T-test).
Published online:
21 July 2017Figure 2. A: GPX3 expression correlates with methylation status in primary xenograft specimens. GPX3 msp on gDNA isolated from patient-derived CRC xenograft tissue. PCR products were run on an agarose gel for visualization. IVD: in-vitro methylated lymphocyte DNA, BH2O: bisulfite-treated distilled water, H2O: untreated distilled water. The IVD, BH20, and H2O lanes are from the same gel as the other samples but are from a different row. MSP for xenograft #18 not shown on this gel because the gel was run separately due to timing of DNA preparation. However, it demonstrated that xenograft #18 was unmethylated. DNA methylation of GPX3 promoter region is observed in xenografts 10, 21, and 24. B: GPX3 expression correlates with methylation status in primary xenograft specimens. GPX3 expression by qPCR on colorectal cancer patient tumor xenograft samples. Methylation status (U: unmethylated, M: methylated) as determined by MSP indicated above the bars from panel A. Xenografts with methylation (10, 21, and 24) have reduced expression compared with unmethylated xenografts (P-value calculated using a two-tailed Student's T-test, P = 0.01). Samples were run in triplicate and each sample was normalized to GAPDH expression. Error bars indicate standard deviation.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0002_oc.gif"})
Figure 2. A: GPX3 expression correlates with methylation status in primary xenograft specimens. GPX3 msp on gDNA isolated from patient-derived CRC xenograft tissue. PCR products were run on an agarose gel for visualization. IVD: in-vitro methylated lymphocyte DNA, BH2O: bisulfite-treated distilled water, H2O: untreated distilled water. The IVD, BH20, and H2O lanes are from the same gel as the other samples but are from a different row. MSP for xenograft #18 not shown on this gel because the gel was run separately due to timing of DNA preparation. However, it demonstrated that xenograft #18 was unmethylated. DNA methylation of GPX3 promoter region is observed in xenografts 10, 21, and 24. B: GPX3 expression correlates with methylation status in primary xenograft specimens. GPX3 expression by qPCR on colorectal cancer patient tumor xenograft samples. Methylation status (U: unmethylated, M: methylated) as determined by MSP indicated above the bars from panel A. Xenografts with methylation (10, 21, and 24) have reduced expression compared with unmethylated xenografts (P-value calculated using a two-tailed Student's T-test, P = 0.01). Samples were run in triplicate and each sample was normalized to GAPDH expression. Error bars indicate standard deviation.
Varying sensitivity to platinum drugs among CRC cell lines correlates with GPX3 methylation status
To examine the relationship between GPX3 expression and platinum sensitivity, we used MTS assays to establish CRC cell lines sensitivity to cisplatin. These lines differed in platinum sensitivity, and these differences correlated with GPX3 methylation status and GPX3 mRNA expression (Fig. 3A and B). Methylated and partially methylated cell lines with low level or absent GPX3 expression, have cisplatin IC50 from 2.5 to 6 micromolar, while unmethylated cell lines with higher GPX3 expression are relatively resistant (IC50: 15–50 micromolar). We also tested sensitivity to oxaliplatin and found similar patterns in terms of oxaliplatin sensitivity and GPX3 methylation (Fig. 3A). Together, the data indicate a relationship between GPX3 promoter methylation, decreased GPX3 expression, and increased platinum sensitivity.
Published online:
21 July 2017Figure 3. A: Platinum IC50 associates with GPX3 expression and the presence of DNA methylation in CRC cell lines. qPCR indicating relative GPX3 mRNA levels for 9 CRC cell lines. Samples were run in triplicate and each sample was normalized to its own gapdh expression. For comparison between cell lines, Colo205 expression was set to 1. M and P denote GPX3 promoter methylation or partial methylation and U denotes samples with an unmethylated GPX3 promoter, as determined in our laboratory. Note, cell lines with partial methylation have repressed levels of GPX3 expression compare with unmethylated cell lines, and GPX3 expression is absent in SW48 and HCT116 with complete methylation. Cisplatin and oxaliplatin IC50 values are shown beneath the cell line names and were determined by MTS cell proliferation assays. Methylated and partially methylated cell lines have cisplatin IC50: 2.5–6 micromolar, while unmethylated cell lines with high GPX3 expression are relatively resistant (IC50: 15–50 micromolar). Error bars indicate standard deviation. B: Cisplatin sensitivities among CRC cell lines. Y axis indicates percentage alive based on MTS assays with increasing cisplatin concentration (x axis), used to generate IC50 values in Fig. 3A. Black color line indicates cell lines with unmethylated GPX3, red/light color line indicates cell lines with methylated or partially-methylated GPX3.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0003_oc.gif"})
Figure 3. A: Platinum IC50 associates with GPX3 expression and the presence of DNA methylation in CRC cell lines. qPCR indicating relative GPX3 mRNA levels for 9 CRC cell lines. Samples were run in triplicate and each sample was normalized to its own gapdh expression. For comparison between cell lines, Colo205 expression was set to 1. M and P denote GPX3 promoter methylation or partial methylation and U denotes samples with an unmethylated GPX3 promoter, as determined in our laboratory. Note, cell lines with partial methylation have repressed levels of GPX3 expression compare with unmethylated cell lines, and GPX3 expression is absent in SW48 and HCT116 with complete methylation. Cisplatin and oxaliplatin IC50 values are shown beneath the cell line names and were determined by MTS cell proliferation assays. Methylated and partially methylated cell lines have cisplatin IC50: 2.5–6 micromolar, while unmethylated cell lines with high GPX3 expression are relatively resistant (IC50: 15–50 micromolar). Error bars indicate standard deviation. B: Cisplatin sensitivities among CRC cell lines. Y axis indicates percentage alive based on MTS assays with increasing cisplatin concentration (x axis), used to generate IC50 values in Fig. 3A. Black color line indicates cell lines with unmethylated GPX3, red/light color line indicates cell lines with methylated or partially-methylated GPX3.
Chronic azacitidine treatment alters CRC cell lines sensitivity to oxaliplatin
The demethylating agent 5-azacitidine can be used to re-express silenced genes and reverse the silenced phenotype.8 However, the acute cytotoxic effects of this drug can complicate determinations of changes in sensitivity to other chemotherapies, including the platinum agents which we wished to study. To avoid this confounding issue, we used long-term azacitidine treatment at lower doses to achieve gradual gene demethylation without a large degree of cell death. For these studies, Colo205 (GPX3 unmethylated), SW480 (GPX3 partially methylated), and HCT116 (GPX3 methylated) were chronically-exposed to azacitidine and were examined using oxaliplatin MTS assays. We hypothesized that GPX3 demethylation in HCT116 would result in greater resistance to oxaliplatin, while the unmethylated cell line Colo205 basally expressing GPX3 would experience no change in sensitivity. We also examined the partially methylated cell line SW480, which has lower expression. Rather than showing no change in sensitivity, treatment of SW480 and Colo205 with azacitidine resulted in increased cell death in the presence of oxaliplatin compared with their non-azacitidine-treated counterparts (Fig. 4A). When GPX3 expression analysis was compared by qPCR (Fig. 4A), performed on cells harvested prior to the addition of oxaliplatin, we found that GPX3 expression was not significantly changed in SW480 and Colo205 with chronic azacitidine exposure (minimal increase in SW480 and no change in Colo205). We wondered whether epigenetic changes at other loci caused by azacitidine treatment might be reducing sensitivity to oxaliplatin independent of GPX3 status. We examined the pro-apoptotic gene APAF1, which has been reported to be epigenetically repressed,29 and found that expression of this gene increased in both SW480 and Colo205, following azacitidine treatment (data not shown). An increase in APAF1 expression following chronic azacitidine treatment in SW480 and Colo205, could partially explain increased cell death in these cell lines in that the net impact of having no significant increase in GPX3 expression combined with an increase in APAF1 expression was increased cell death. In contrast, in HCT116, GPX3 expression was increased with azacitidine treatment and, despite a concomitant increase in APAF1 expression (data not shown), HCT116 was more resistant to oxaliplatin (Fig. 4A). We confirmed that the change in expression was due to DNA demethylation, using MSP performed on the same cell samples harvested prior to oxaliplatin treatment (Fig. 4B). This suggests that in HCT116, a demethylation of GPX3 and a corresponding increase in GPX3 expression contributed to increased resistance to and growth in oxaliplatin despite changes in APAF1 expression that would have contributed to cell death.
Published online:
21 July 2017Figure 4. Chronically azacitidine-treated SW480 and Colo205 are more sensitive to oxaliplatin but chronically azacitidine-treated HCT116 is more resistant to oxaliplatin. A: MTS assays indicating increased sensitivity to oxaliplatin after treatment of SW480 and Colo205 with azacitidine and decreased sensitivity to oxaliplatin after treatment of HCT116 with azacitidine. Next to each cell lines' MTS assay are the GPX3 expression data as determined by qPCR. The degree of increased oxaliplatin resistance in azacitidine-treated HCT116 varies with the extent of GPX3 demethylation and re-expression. qPCR samples were run in triplicate and each sample was normalized to its own gapdh expression. MTS error bars indicate SEM and qPCR error bars indicate standard deviation. B: MSP data demonstrating no detectable demethylation of GPX3 in either SW480 or Colo205 but detectable demethylation of GPX3 in HCT116 after treatment with azacitidine. DKO (DNMT double knockout HCT116 cells) and NL (normal lymphocytes) serve as negative controls. The red line indicates that the controls for SW480 and Colo205 were on a 2nd gel (run at the same time) due to space constraints.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0004_oc.gif"})
Figure 4. Chronically azacitidine-treated SW480 and Colo205 are more sensitive to oxaliplatin but chronically azacitidine-treated HCT116 is more resistant to oxaliplatin. A: MTS assays indicating increased sensitivity to oxaliplatin after treatment of SW480 and Colo205 with azacitidine and decreased sensitivity to oxaliplatin after treatment of HCT116 with azacitidine. Next to each cell lines' MTS assay are the GPX3 expression data as determined by qPCR. The degree of increased oxaliplatin resistance in azacitidine-treated HCT116 varies with the extent of GPX3 demethylation and re-expression. qPCR samples were run in triplicate and each sample was normalized to its own gapdh expression. MTS error bars indicate SEM and qPCR error bars indicate standard deviation. B: MSP data demonstrating no detectable demethylation of GPX3 in either SW480 or Colo205 but detectable demethylation of GPX3 in HCT116 after treatment with azacitidine. DKO (DNMT double knockout HCT116 cells) and NL (normal lymphocytes) serve as negative controls. The red line indicates that the controls for SW480 and Colo205 were on a 2nd gel (run at the same time) due to space constraints.
Knockdown of GPX3 by shRNA decreases GPX3 expression in CaCO2, HT29, and HCT116 and increases platinum sensitivity in cell lines with basal GPX3 expression
Because of the confounding effects of azacitidine in testing the specific role of GPX3 in platinum sensitivity, we created GPX3 shRNA knockdown constructs in CaCO2 (high baseline GPX3 expression), HT29 (moderate baseline expression GPX3), and HCT116 (no or minimal baseline GPX3 expression). Fig. 5A demonstrates the knockdown constructs in CaCO2 cells and the essentially complete repression of GPX3 with construct #1 in CaCO2. Consistent with GPX3 role in redox biology, CaCO2 cells, with the near complete loss of expression in knockdown construct #1, required the antioxidant N-acetylcysteine (NAC) in the culture media for growth beyond 1–2 weeks. GPX3 qPCR (Fig. 5B) confirms markedly reduced GPX3 expression at the RNA level in the knockdown CaCO2 and HT29 lines, and although barely detectable in comparison to the other cell lines, minimal HCT116 expression is further decreased in HCT116 #1 cells. The knockdown cell lines do not grow significantly slower than their wild type counterparts (data not shown). MTS assays performed to assess the impact of GPX3 on oxaliplatin sensitivity and representative plots are shown in Fig. 5. Knockdown of GPX3 in CaCO2 sensitizes these cells to oxaliplatin in both knockdown #1 and #3. There appears to be no impact of knocking down GPX3 in HCT116, consistent with the essentially absent expression in wild type HCT116 at baseline. While to a more limited extent than seen in CaCO2, knocking down GPX3 in HT29 still increases sensitivity to oxaliplatin. We observed no compensatory changes in GPX1 and GPX2 expression in the GPX3 knockdown cells (data not shown), suggesting that GPX3 is the key glutathione peroxidase protein associated with sensitivity to oxaliplatin.
Published online:
21 July 2017Figure 5. A: shRNA knockdowns of GPX3 in CaCO2. Western blot demonstrating absent or reduced GPX3 protein expression after lentiviral infection of CaCO2 with different (numbered) shRNA constructs. Constructs #1 and #3 (in bold) were used in subsequent experiments with construct #1 used to knock down expression in HCT116 and HT29 as well. The vertical red line denotes a gap in the gel where a 5th knockdown construct was run but then cropped from the image because no staining was present in that lane even for the Lamin B control. WT: wild type. B: GPX3 qPCR confirms markedly reduced GPX3 expression in the knockdown CaCO2, HCT116, and HT29 lines. Representative GPX3 qPCR data demonstrating significantly reduced GPX3 expression in the knockdown CRC lines. Samples were run in triplicate and each sample was normalized to GAPDH expression. Each line was normalized to its baseline GPX3 expression to show relative change. SCR indicates scramble construct. Error bars indicate standard deviation. C: shRNA oxaliplatin MTS assays demonstrating a minimal increase in the sensitivity of CaCO2 GPX3 knockdown lines to oxaliplatin. Error bars indicate SEM. D: Oxaliplatin exposure induces GPX3 after acute exposure. Acute oxaliplatin exposure induces GPX3 expression in wild type CaCO2 but not in knockdown CaCO2 lines. CaCO2 cells were plated and then grown in oxaliplatin (3.13 uM final) for 48 h before harvesting for RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression. The oxaliplatin concentration used was based on prior IC50 measurements. For comparison between the different lines, CaCO2 WT GPX3 fold change was set to 1. Ox: oxaliplatin. Error bars indicate standard deviation.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0005_oc.gif"})
Figure 5. A: shRNA knockdowns of GPX3 in CaCO2. Western blot demonstrating absent or reduced GPX3 protein expression after lentiviral infection of CaCO2 with different (numbered) shRNA constructs. Constructs #1 and #3 (in bold) were used in subsequent experiments with construct #1 used to knock down expression in HCT116 and HT29 as well. The vertical red line denotes a gap in the gel where a 5th knockdown construct was run but then cropped from the image because no staining was present in that lane even for the Lamin B control. WT: wild type. B: GPX3 qPCR confirms markedly reduced GPX3 expression in the knockdown CaCO2, HCT116, and HT29 lines. Representative GPX3 qPCR data demonstrating significantly reduced GPX3 expression in the knockdown CRC lines. Samples were run in triplicate and each sample was normalized to GAPDH expression. Each line was normalized to its baseline GPX3 expression to show relative change. SCR indicates scramble construct. Error bars indicate standard deviation. C: shRNA oxaliplatin MTS assays demonstrating a minimal increase in the sensitivity of CaCO2 GPX3 knockdown lines to oxaliplatin. Error bars indicate SEM. D: Oxaliplatin exposure induces GPX3 after acute exposure. Acute oxaliplatin exposure induces GPX3 expression in wild type CaCO2 but not in knockdown CaCO2 lines. CaCO2 cells were plated and then grown in oxaliplatin (3.13 uM final) for 48 h before harvesting for RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression. The oxaliplatin concentration used was based on prior IC50 measurements. For comparison between the different lines, CaCO2 WT GPX3 fold change was set to 1. Ox: oxaliplatin. Error bars indicate standard deviation.
Acute oxaliplatin exposure induces GPX3 expression in wild type CaCO2 but not in knockdown CaCo2 lines
While DNA methylation resulted in diminished expression in primary colorectal tumors (Fig. 1A), primary colorectal xenografts (Fig. 2) and colorectal cancer cell lines (Fig. 3), we noted that some primary tumors and xenografts had relatively low expression without the presence of DNA methylation and complete silencing. Since one of the functions of GPX3 may be to reduce ROS induced damage, we questioned whether induction of GPX3 expression would be in the cellular response to acute exposure to DNA damage. To test this hypothesis, CaCO2 cells were exposed to oxaliplatin for 48 h and examined for GPX3 expression. Both wild type and scramble CaCO2 acutely increase GPX3 expression in response to oxaliplatin (Fig. 5D), suggesting the importance of GPX3 expression in oxidative damage response. However, as expected, CaCO2 #1 and CaCO2 #3 knockdown lines are unable to significantly increase GPX3 expression in response to oxaliplatin due to the shRNA inhibition. In contrast, the GPX3-methylated cell line HCT116, when acutely challenged with oxaliplatin as above, resulted in continued absence of GPX3 expression after oxaliplatin exposure (data not shown), suggesting that epigenetic silencing prevented transcription of this gene and thus precluded the normal response of GPX3 increased expression in response to damage produced by oxaliplatin.
GPX3 expression is induced during the development of oxaliplatin resistance in chronically-exposed HCT116 and HT29
We next questioned whether long-term treatment could alter GPX3 expression. To do so, we treated two colorectal cell lines with varying levels of GPX3 expression, either completely silent (HCT116) or unmethylated and expressed (HT29), with increasing concentrations of oxaliplatin over many weeks. The resulting oxaliplatin-resistant HCT116 and HT29 cells were examined for GPX3 expression (Fig. 6), and both HCT116 and HT29 expressed significantly increased levels of GPX3 compare with untreated parental controls. While the levels of GPX3 expression in HCT116 remain low relative to HT29, there was an approximate 3.5-fold induction. This suggests that the resistance to oxaliplatin occurs at least in part due through increasing GPX3 expression.
Published online:
21 July 2017Figure 6. Oxaliplatin exposure induces GPX3 after chronic exposure. GPX3 expression is induced during the development of oxaliplatin resistance in chronically exposed wild type HCT116 and HT29. HCT116 and HT29 cells were exposed to oxaliplatin over a period of weeks to create oxaliplatin-resistant lines. Cells were then harvested for RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression. For comparison between the different lines, HCT116 WT (parental) GPX3 fold change was set to 1. To better assess the degree of GPX3 in the HCT116 lines, arrow demonstrates just the HCT116 lines. P: parental, OXR: oxaliplatin resistant. Error bars indicate standard deviation.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0006_oc.gif"})
Figure 6. Oxaliplatin exposure induces GPX3 after chronic exposure. GPX3 expression is induced during the development of oxaliplatin resistance in chronically exposed wild type HCT116 and HT29. HCT116 and HT29 cells were exposed to oxaliplatin over a period of weeks to create oxaliplatin-resistant lines. Cells were then harvested for RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression. For comparison between the different lines, HCT116 WT (parental) GPX3 fold change was set to 1. To better assess the degree of GPX3 in the HCT116 lines, arrow demonstrates just the HCT116 lines. P: parental, OXR: oxaliplatin resistant. Error bars indicate standard deviation.
Tumor xenografts from GPX3 knockdown line demonstrate increased sensitivity to cisplatin
To determine the impact of GPX3 on platinum sensitivity in vivo, mice implanted with cell line xenografts of CaCO2 wild type and CaCO2 #1 were treated with cisplatin and tumor sizes were monitored (Fig. 7). Untreated wild type and CaCO#1 knockdown xenografts grew similarly. CaCO2 is intrinsically relatively resistant to cisplatin with a high cisplatin IC50; accordingly, cisplatin-treated wild type xenografts had a growth rate decline but not regression compare with untreated wild type xenografts. However, tumor xenografts from GPX3 knockdown cells were markedly more sensitive to cisplatin compared with their wild type counterparts with essentially no measurable tumor after 2 weeks, and maintained a similar trend beyond until study termination. These data, using xenografted, isogenic cell lines differing only in their expression of GPX3, validate the importance of GPX3 for platinum sensitivity.
Published online:
21 July 2017Figure 7. Tumor xenografts from GPX3 knockdown line demonstrate increased sensitivity to cisplatin. Parental CaCO2 cells and construct #1 knockdown cells were injected into immunodeficient mice. Following initial tumor growth, mice were treated with cisplatin or untreated, as described in methods section. There was no significant effect of GPX3 knockdown on tumor growth rate in the untreated mice, and no significant growth inhibition in the parental CaCO2 tumors when treated with cisplatin. However, in knockdown cells, cisplatin resulted in tumor regression.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/kepi20/2017/kepi20.v012.i07/15592294.2016.1265711/20171127/images/medium/kepi_a_1265711_f0007_oc.gif"})
Figure 7. Tumor xenografts from GPX3 knockdown line demonstrate increased sensitivity to cisplatin. Parental CaCO2 cells and construct #1 knockdown cells were injected into immunodeficient mice. Following initial tumor growth, mice were treated with cisplatin or untreated, as described in methods section. There was no significant effect of GPX3 knockdown on tumor growth rate in the untreated mice, and no significant growth inhibition in the parental CaCO2 tumors when treated with cisplatin. However, in knockdown cells, cisplatin resulted in tumor regression.
Discussion
CRC is the 3rd leading cause of cancer deaths among US men and women.1 In addition to new treatments, molecular-based strategies to optimize current treatments are needed. Epigenetic changes can be used to predict chemosensitivity in tumor tissues to help improve clinical outcomes.
Here we demonstrate that GPX3 promoter methylation, which is present in one third of TCGA CRC samples and correlates with MSI status, is associated with decreased GPX3 expression in patient tumor xenografts and in CRC cell lines. We first confirmed methylation status prevalence with primary specimens, choosing primary tumor xenografts to ensure that neoplastic cells were purely human, thus avoiding contamination from normal tissue. Using CRC cell lines, GPX3 promoter methylation and expression directly correlated with platinum drug sensitivity and, when azacitidine treatment was used to demethylate GPX3 leading to its re-expression, platinum sensitivity decreased. Thus, in HCT116, azacitidine treatment led to an approximately 100-fold increase in GPX3 expression and resulted in resistance to oxaliplatin. In contrast, in SW480 and Colo205, when azacitidine treatment did not result in increased GPX3 expression, subsequent treatment of those cells with oxaliplatin resulted in cell death. These data suggest the functional importance of GPX3 silencing in cancer cells on sensitivity to platinums.
Consistent with GPX3 role in cells antioxidant system, when GPX3 was essentially completely knocked down in CaCO2 (a line which has unmethylated and high expression of GPX3 at baseline), the cells required the antioxidant NAC to survive. NAC was not needed if there was any residual GPX3 expression in CaCO2. This reflects the importance of at least a minimal amount of GPX3 expression for CaCO2 growth and maintenance in cell culture, a ROS-rich environment. We also demonstrate that GPX3 is upregulated after platinum exposure; although, expectedly, the degree varies with the cells' ability to express GPX3 at baseline; knockdown lines and methylated lines have blunted or delayed upregulation responses. The azacitidine experiments and the oxaliplatin challenge experiments demonstrate that the capacity to increase GPX3 expression, not merely baseline GPX3 expression, is important for mediating platinum resistance. In the acute setting, azacitidine can demethylate and allow re-expression of GPX3; meanwhile, chronic exposure to platinum may allow for demethylation and upregulation of GPX3 in the absence of azacitidine. When tested in cell line tumor xenografts, the GPX3 knockdown CaCO2 line was strikingly more sensitive to platinum treatment than its wild type counterpart, again confirming the importance of GPX3 for platinum resistance.
Although not the only alteration that determines MSI status, MLH1 methylation leads to an MSI+ phenotype. Interestingly, there is a strong association between MLH1 methylation and GPX3 methylation. Methylation of GPX3 is found in 87 of 292 (30%) CRC samples. GPX3 methylation is seen in 28 of 48 (58%) MLH1 methylated tumors compared with 59 of 244 (24%) MLH1 unmethyated tumors (Fisher test P-value = 8.161e-06). Additionally, however, GPX3 methylation is strongly associated with the CpG island methylator phenotype-high (CIMP-H) phenotype in that 54 of 87 (62%) CIMP-H tumors have GPX3 methylation while 33 of 205 (16%) CIMP low or negative have GPX3 methylation (Fisher test P-value = 2.143e-14). Whether GPX3 methylation is directly related to MLH1 silencing through mechanisms related to subsequent microsatellite instability or is associated with the CIMP-H phenotype cannot be determined through these association studies. MLH1 methylation and the resultant MSI+ phenotype also correlates with the CpG island methylator-high phenotype (CIMP-H); as such, it is unclear if MLH1 methylation directly leads to GPX3 methylation, through microsatellite instability, or if GPX3 methylation is a consequence of the CpG island methylator-high phenotype (CIMP-H) in general. Although CIMP is a well recognized phenomenon in colorectal cancer, it was not examined further here for association with platinum sensitivity because a precise definition of CIMP has not been established and published reports of CIMP in CRC use many different panels and criteria for scoring CIMP, including CIMP-negative, CIMP-high and CIMP-low. These issues are described in a review by Hughes et al.30 Additionally, this study focused on a specific epigenetic change and the potential association of the silencing of GPX3 to directly alter a phenotype that might be predicted by the change in DNA damage response.
Previous analyses have explored the complex potential association of MSI+ with chemosensitivity and resistance. Early clinical studies suggested that MSI+ tumors were less sensitive to nucleoside analog 5-flurouracil (5FU) than MSI- tumors; of note, 5FU has been the backbone of CRC chemotherapy regimens for the past 3–4 decades. However, with the addition of oxaliplatin to 5FU, now the standard adjuvant regimen, more recent studies demonstrate that MSI+ tumors do benefit from this combination just as MSI- tumors do.25,26 These data suggest that the platinum compound may be responsible for the sensitivity of MSI+ tumors to the current adjuvant standard of care regimen of 5FU and oxaliplatin. The correlation of GPX3 methylation with the MSI+ phenotype may be one underlying factor for the sensitivity of MSI+ tumors to 5FU combined with oxaliplatin when they are resistant to 5FU alone.
Although the impact of GPX3 methylation on platinum sensitivity could just be a surrogate for other processes, such as global methylation changes impacting drug sensitivity, the magnitude of our in vivo data suggests GPX3 is itself affecting the phenotype. Though it is likely not the sole determinant of platinum resistance, and other groups have explored alternative genes (such as BRCA1 and ERCC1, and the association of their methylation with platinum sensitivity),31 taken together, the GPX3 data presented here demonstrate the importance of GPX3 for platinum resistance in CRC. GPX3 promoter methylation predicts this relationship and potentially can be used as a predictive biomarker. Interestingly, though a decrease in GPX3 expression has also been linked to chemosensitivity in other tumors as well,15 GPX3 methylation was associated with chemoresistance in head and neck cancer.12 A potential application is in the choice between first-line oxaliplatin or irinotecan in stage IV CRC. Both oxaliplatin and irinotecan are approved in conjunction with 5-fluorouracil for first-line treatment of patients with metastatic colon cancer.32 If a patient's tumor demonstrates methylation and silencing of GPX3, it may be more likely to respond to oxaliplatin. However, if a patient's tumor has unmethylated (and highly expressed) GPX3, it may be less likely to respond to oxaliplatin and thus the patient may be better served by a regimen not containing platinum but instead containing another agent, such as irinotecan. Thus, the methylation status of GPX3 could inform treatment strategies and could be tested prospectively in a first-line metastatic trial.
Materials and methods
The Cancer Genome Atlas (TCGA) analysis
We analyzed 292 samples of colorectal carcinoma and 38 adjacent normal samples from the Cancer Genome Atlas project (TCGA)24 that had both DNA methylation. TCGA used Illumina Infinium HumanMethylation450 BeadChip (Illumina)33 for DNA methylation profiling. Two hundred and fifty-five of 292 samples have mRNA expression measured using RNAseq for mRNA expression profiling. We calculated the Spearman's rank correlation coefficient for GPX3 promoter DNA methylation, using probe cg26638444, and mRNA expression. Samples with β value higher than 0.2 were considered as methylated. MSI status of a sample was derived using MLH1 promoter methylation status (probe cg00893636) and threshold 0.2.
Analyses described above as well as plots including boxplots, and scatterplots were performed in R34 using standard methods and customized routines.
Cell lines and culture conditions
Cell lines (RKO, SW48, LOVO, HCT116, SW480, SW620, COLO205, CACO2 and HT29) were obtained from ATCC. Cells were maintained in McCoys 5A media with 10% fetal bovine serum at 37°C in 5% CO2.
Methylation specific PCR (MSP)
Genomic DNA was extracted using a DNA extraction kit (Qiagen, # 13,343). The extracted DNA was quantified using a NanoDrop spectrophotometer and up to 1 µg of extracted genomic DNA was used for bisulfite treatment. Bisulfite conversion was done using the EZ DNA Methylation kit (Zymo Research, #D5001). GPX3 MSP primers are as previously described.12 PCR was performed as previously described 35 in a 25 µl reaction volume containing RedTaq DNA polymerase (Sigma, #D4309) and template DNA, with 35 cycles at 60°C or 65°C annealing temperature. DKO cells (DNMT double knockout HCT116 cells), normal human peripheral lymphocytes (NL), and bisulfite treated water (BH2O) serve as negative methylation controls. IVD serves as the positive methylation control.
Quantitative real time PCR (qPCR)
cDNA was prepared from isolated RNA (Qiagen RNeasy kit, # 74,104) using the iScript cDNA synthesis kit (Bio-Rad, # 1,708,890). qPCR was performed using SsoAdvanced SYBR Green Supermix (Bio-Rad, #1,725,270) and a “My iQ iCycler” thermal cycler (Bio-Rad). PCR was performed in triplicate for each sample: 1 µl of cDNA product, 7.5 µl of SsoAdvanced SYBR Green master mix, 0.6 µl of combined forward and reverse primers (10 μM each), and up to a total volume of 15 µl with sterile water per PCR reaction. PCR was performed with an initial incubation step at 95°C for 3 min followed by 40 cycles at 95°C for 30 s and 60°C for 30 s. These 40 cycles were followed by melting curve analysis. GPX3 qPCR primers are as previously described.15 Apaf-1 primers were designed with the help of the Primer3 program36: forward, AACCAGGATGGGTCACCATA and reverse, ACTGAAACCCAATGCACTCC. Each cell line sample was normalized to its own GAPDH control using the following primers to amplify GAPDH in the samples: forward, GAAGGTCGGAGTCAACGGATTT and reverse, ATGGGTGGAATCATATTGGAAC. Calculations were made using the ΔΔCt method. Primers were synthesized by IDT.
In vitro cytotoxic assays
Cytotoxicity of the drugs was determined by MTS assay (Promega, # G5421). Drugs tested were oxaliplatin (Sigma, # O9512) and cisplatin (Sigma, # 479,306). In brief, cells were plated in each well of 96 well plates 24 h prior to the addition of drug. After 24 h, drug was added in increasing concentrations; each drug concentration was done in triplicate and allowed to incubate for two days. After two days, MTS assays were performed per manufacturer instructions. Raw absorbance values were background subtracted with wells containing only media. Cell viability was calculated by the following formula (absorbance of treated well/absorbance of mock*100) for each drug concentration. Cell viability was plotted against drug concentrations and the concentration at which 50% of the cells are viable is taken as IC50.
Cell growth rate assays
To measure cell growth rates, the CellTiter-Glo (Promega, #G7570) system was used. Cells from each line were plated in triplicate on day 1 of the assay. CellTiter reagent was added on day 2 and baseline luminescence measurements were read 60 min after substrate addition. Subsequent measurements were made at 24 h intervals.
Re-expression of GPX3 in colorectal cancer cell lines
SW480, COLO205, and HCT116 were maintained as above but with the addition of azacitdine to a final concentration of 500 nM 5-azacitidine (Sigma, # A2385) during twice weekly media changes. This treatment strategy was performed on these 3 cell lines continuously, over a period of months. Untreated cell lines were similarly maintained and passaged. Oxaliplatin cytotoxicity assays were performed using MTS reagent. For these assays, 3,000 (SW480) or 4,000 (COLO205, HCT116) cells were plated per well into 96-well plates 24 h prior to oxaliplatin addition. Each drug concentration was performed in replicates of five. On the day of plating, additional samples of these azacitidine-treated and untreated cells were pelleted, frozen on dry ice, and stored at −80°C for molecular studies. After 48 h of incubation with oxaliplatin, the MTS in vitro cytotoxic assays were performed.
Nucleic acid from the frozen cell pellets corresponding to each MTS assay was isolated using the ZR-Duet DNA/RNA MiniPrep kit (Zymo Research, # D7001) or separately using Qiagen DNA and RNA isolation kits. DNA bisulfite conversion was performed using the EZ DNA Methylation kit and MSP was performed as described above. cDNA was prepared from the isolated RNA using the iScript cDNA synthesis kit and qPCR was performed as described above.
GPX3 knockdown
The Sigma MISSION pLKO.1-puro lentiviral shRNA system (Sigma, #NM_0,02084) was used to repress GPX3 expression in CRC cell lines, following the manufacturer's instructions. The Sigma pLKO clones TRCN0000008678, TRCN0000273647, TRCN0000273648, and TRCN0000273651 correspond to constructs #1, #2, #3, and #4, respectively. A pScramble construct was used as a control vector. The knockdown lines were cultured as above but with the addition of puromycin (Sigma, # P8833) for the duration of the studies (CaCO2 at 8 μg/ml, HT29 at 3 μg/ml, and HCT116 at 1 μg/ml). CaCO2 infected with construct #1 virus (CaCO2 #1) was not viable beyond approximately 2 weeks. To minimize any toxicity from the transfection itself, the cell lines were infected again with construct #1 virus at a lower titer. Additionally, CaCO2 GPX3 knockdown line #1 required N-acetylcysteine (NAC) (Sigma, # A9165) to a final concentration of 1 mm for maintenance in culture. CaCO2 #3 was able to be maintained in culture without the addition of NAC, though this knockdown did still have some minimal GPX3 expression detectable. Though these lines were maintained in puromycin +/− NAC, these drugs were not included during growth rate, MTS, or drug-challenge assays. Anti-GPX3 antibody at 1:500 (Abcam,# ab27325) and Anti-Lamin B antibody at 1:2,500 (Santa Cruz, # sc-6,217) were used for western blots.
Acute oxaliplatin challenge experiments
CaCO2 cells were plated and then grown in oxaliplatin (3.13 μM final) for 48 h before harvesting for RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression, as above. The oxaliplatin concentration used was based on prior IC50 measurements.
Chronic oxaliplatin exposure in HCT116 and HT29
HCT116 and HT29 cells were cultured for weeks in increasing concentrations of oxaliplatin, based on the lines' IC50 values to develop oxaliplatin-resistant cell lines. RNA isolation, cDNA synthesis, and qPCR to assess GPX3 expression were performed, as above.
Xenograft models
Antitumor activity of cisplatin was evaluated in xenograft models. GPX3 knockdown and wild type CaCO2 cells (2 × 106) in Phosphate Buffered Saline and matrigel in the ratio of 1:1 were injected subcutaneously into the two flanks of homozygous female athymic nude mice (Simonson Laboratories). Five mice were used for each experimental condition. Cisplatin (5 mg/kg) was administered once tumors became palpable. The mice were treated only until the time point when the mock animals were sacrificed due to tumor volume. Length (L) and width (W) of tumors were measured with a caliper until the tumors reached 2 cm. Tumor volume (TV) was calculated with the following formula TV = 1/2*(Length*width2). Matrigel for xenograft studies was obtained from BD PharMingen (#354,234). Cisplatin was obtained from the Johns Hopkins Hospital pharmacy. All small animal experiments described conformed to the guidelines of the Animal Care and Use Committee of Johns Hopkins University. Mice were maintained in accordance with the guidelines of the American Association of Laboratory Animal Care.
Related Research Data
Acknowledgements
We would like to thank Yi Cai, Tina Largent, Weijie Poh, and Michelle Vaz for their help and suggestions with this project and Scott Kern for his constructive comments and advice regarding this manuscript. This work was supported by a T32 institutional grant to the Johns Hopkins University School of Medicine, Medical Oncology Fellowship Program (L.P.) and an American Cancer Society grant 127343-RSG-15–068–01-TBG (N.A.).
Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.
References
- Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin 2016; 66(1):7-30; PMID:26742998; https://doi.org/10.3322/caac.21332 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Lee MS, Kopetz S. Current and Future Approaches to Target the Epidermal Growth Factor Receptor and Its Downstream Signaling in Metastatic Colorectal Cancer. Clin Colorectal Cancer 2015; 14(4):203-18; PMID:26077270; https://doi.org/10.1016/j.clcc.2015.05.006 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Herman JG, Baylin SB. Gene silencing in cancer in association with promoter hypermethylation. N Engl J Med 2003; 349(21):2042-54; PMID:14627790; https://doi.org/10.1056/NEJMra023075 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V, Baylin SB, Herman JG. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med 2000; 343(19):1350-4; PMID:11070098; https://doi.org/10.1056/NEJM200011093431901 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Hegi ME, Diserens AC, Godard S, Dietrich PY, Regli L, Ostermann S, Otten P, Van Melle G, de Tribolet N, Stupp R. Clinical trial substantiates the predictive value of O-6-methylguanine-DNA methyltransferase promoter methylation in glioblastoma patients treated with temozolomide. Clin Cancer Res 2004; 10(6):1871-4; PMID:15041700; https://doi.org/10.1158/1078-0432.CCR-03-0384 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M, Kros JM, Hainfellner JA, Mason W, Mariani L, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 2005; 352(10):997-1003; PMID:15758010; https://doi.org/10.1056/NEJMoa043331 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Paz MF, Yaya-Tur R, Rojas-Marcos I, Reynes G, Pollan M, Aguirre-Cruz L, García-Lopez JL, Piquer J, Safont MJ, Balaña C, et al. CpG island hypermethylation of the DNA repair enzyme methyltransferase predicts response to temozolomide in primary gliomas. Clin Cancer Res 2004; 10(15):4933-8; PMID:15297393; https://doi.org/10.1158/1078-0432.CCR-04-0392 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, Markowitz S, Willson JK, Hamilton SR, Kinzler KW, et al. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci U S A 1998; 95(12):6870-5; PMID:9618505; https://doi.org/10.1073/pnas.95.12.6870 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Pelosof L, Yerram SR, Ahuja N, Delmas A, Danilova L, Herman JG, Azad NS. CHFR silencing or microsatellite instability is associated with increased antitumor activity of docetaxel or gemcitabine in colorectal cancer. Int J Cancer 2014; 134(3):596-605; PMID:23873170; https://doi.org/10.1002/ijc.28390 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Scolnick DM, Halazonetis TD. Chfr defines a mitotic stress checkpoint that delays entry into metaphase. Nature 2000; 406(6794):430-5; PMID:10935642; https://doi.org/10.1038/35019108 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Brigelius-Flohe R. Tissue-specific functions of individual glutathione peroxidases. Free Radical Biol Med 1999; 27(9-10):951-65; PMID:10569628; https://doi.org/10.1016/S0891-5849(99)00173-2 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Chen B, Rao X, House MG, Nephew KP, Cullen KJ, Guo Z. GPx3 promoter hypermethylation is a frequent event in human cancer and is associated with tumorigenesis and chemotherapy response. Cancer Lett 2011; 309(1):37-45; PMID:21684681; https://doi.org/10.1016/j.canlet.2011.05.013 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Yu YP, Yu G, Tseng G, Cieply K, Nelson J, Defrances M, Zarnegar R, Michalopoulos G, Luo JH. Glutathione peroxidase 3, deleted or methylated in prostate cancer, suppresses prostate cancer growth and metastasis. Cancer Res 2007; 67(17):8043-50; PMID:17804715; https://doi.org/10.1158/0008-5472.CAN-07-0648 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Zhang X, Yang JJ, Kim YS, Kim KY, Ahn WS, Yang S. An 8 gene signature, including methylated and down-regulated glutathione peroxidase 3, of gastric cancer. Int J Oncol 2010; 36(2):405-14; PMID:20043075. [PubMed], [Web of Science ®], [Google Scholar]
- Saga Y, Ohwada M, Suzuki M, Konno R, Kigawa J, Ueno S, Mano H. Glutathione peroxidase 3 is a candidate mechanism of anticancer drug resistance of ovarian clear cell adenocarcinoma. Oncol Rep 2008; 20(6):1299-303; PMID:19020706 [PubMed], [Web of Science ®], [Google Scholar]
- Andre T, Boni C, Mounedji-Boudiaf L, Navarro M, Tabernero J, Hickish T, Topham C, Zaninelli M, Clingan P, Bridgewater J, et al. Oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment for colon cancer. N Engl J Med 2004; 350(23):2343-51; PMID:15175436; https://doi.org/10.1056/NEJMoa032709 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Conroy T, Desseigne F, Ychou M, Bouché O, Guimbaud R, Bécouarn Y, Adenis A, Raoul JL, Gourgou-Bourgade S, de la Fouchardière C, et al. FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer. N Engl J Med 2011; 364(19):1817-25; PMID:21561347; https://doi.org/10.1056/NEJMoa1011923 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- de Gramont A, Figer A, Seymour M, Homerin M, Hmissi A, Cassidy J, Boni C, Cortes-Funes H, Cervantes A, Freyer G, et al. Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer. J Clin Oncol 2000; 18(16):2938-47; PMID:10944126; https://doi.org/10.1200/jco.2000.18.16.2938 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Barrett CW, Ning W, Chen X, Smith JJ, Washington MK, Hill KE, Coburn LA, Peek RM, Chaturvedi R, Wilson KT, et al. Tumor suppressor function of the plasma glutathione peroxidase gpx3 in colitis-associated carcinoma. Cancer Res 2013; 73(3):1245-55; PMID:23221387; https://doi.org/10.1158/0008-5472.CAN-12-3150 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Shen DW, Pouliot LM, Hall MD, Gottesman MM. Cisplatin resistance: a cellular self-defense mechanism resulting from multiple epigenetic and genetic changes. Pharmacol Rev 2012; 64(3):706-21; PMID:22659329; https://doi.org/10.1124/pr.111.005637 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Jungwirth U, Kowol CR, Keppler BK, Hartinger CG, Berger W, Heffeter P. Anticancer activity of metal complexes: involvement of redox processes. Antioxidants Redox Signal 2011; 15(4):1085-127; PMID:21275772; https://doi.org/10.1089/ars.2010.3663 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discovery 2012; 2(5):401-4; PMID:22588877; https://doi.org/10.1158/2159-8290.CD-12-0095 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Science Signaling 2013; 6(269):pl1; PMID:23550210; https://doi.org/10.1126/scisignal.2004088 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Network TCGAR. Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012; 487(7407):330-7; PMID:22810696; https://doi.org/10.1038/nature11252 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Tougeron D, Mouillet G, Trouilloud I, Lecomte T, Coriat R, Aparicio T, Des Guetz G, Lécaille C, Artru P, Sickersen G, et al. Efficacy of Adjuvant Chemotherapy in Colon Cancer With Microsatellite Instability: A Large Multicenter AGEO Study. J Natl Cancer Inst 2016; 108(7):1-9; PMID:26839356; https://doi.org/10.1093/jnci/djv438 [Crossref], [PubMed], [Google Scholar]
- Zaanan A, Cuilliere-Dartigues P, Guilloux A, Parc Y, Louvet C, de Gramont A, Tiret E, Dumont S, Gayet B, Validire P, et al. Impact of p53 expression and microsatellite instability on stage III colon cancer disease-free survival in patients treated by 5-fluorouracil and leucovorin with or without oxaliplatin. Ann Oncol 2010; 21(4):772-80; PMID:19833818; https://doi.org/10.1093/annonc/mdp383 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Gallmeier E, Hermann PC, Mueller MT, Machado JG, Ziesch A, De Toni EN, Palagyi A, Eisen C, Ellwart JW, Rivera J, et al. Inhibition of ataxia telangiectasia- and Rad3-related function abrogates the in vitro and in vivo tumorigenicity of human colon cancer cells through depletion of the CD133(+) tumor-initiating cell fraction. Stem Cells 2011; 29(3):418-29; PMID:21308861; https://doi.org/10.1002/stem.595 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Jimeno A, Feldmann G, Suarez-Gauthier A, Rasheed Z, Solomon A, Zou GM, Rubio-Viqueira B, García-García E, López-Ríos F, Matsui W, et al. A direct pancreatic cancer xenograft model as a platform for cancer stem cell therapeutic development. Mol Cancer Ther 2009; 8(2):310-4; PMID:19174553; https://doi.org/10.1158/1535-7163.MCT-08-0924 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Soengas MS, Capodieci P, Polsky D, Mora J, Esteller M, Opitz-Araya X, McCombie R, Herman JG, Gerald WL, Lazebnik YA, et al. Inactivation of the apoptosis effector Apaf-1 in malignant melanoma. Nature 2001; 409(6817):207-11; PMID:11196646; https://doi.org/10.1038/35051606 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Hughes LA, Khalid-de Bakker CA, Smits KM, van den Brandt PA, Jonkers D, Ahuja N, Herman JG, Weijenberg MP, van Engeland M. The CpG island methylator phenotype in colorectal cancer: progress and problems. Biochim Biophys Acta 2012; 1825(1):77-85; PMID:22056543; https://doi.org/10.1016/j.bbcan.2011.10.005 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Moutinho C, Martinez-Cardus A, Santos C, Navarro-Pérez V, Martínez-Balibrea E, Musulen E, Carmona FJ, Sartore-Bianchi A, Cassingena A, Siena S, et al. Epigenetic inactivation of the BRCA1 interactor SRBC and resistance to oxaliplatin in colorectal cancer. J Natl Cancer Inst 2014; 106(1):djt322; PMID:24273214; https://doi.org/10.1093/jnci/djt322 [Crossref], [PubMed], [Google Scholar]
- NCCN. NCCN Clinical Practice Guidelines in Oncology, Colon Cancer. 2015;Version 2.2015. [Google Scholar]
- Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, Delano D, Zhang L, Schroth GP, Gunderson KL, et al. High density DNA methylation array with single CpG site resolution. Genomics 2011; 98(4):288-95; PMID:21839163; https://doi.org/10.1016/j.ygeno.2011.07.007 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- computing Rffs. R: A language and environment for statistical computing 2004. [Google Scholar]
- Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A 1996; 93(18):9821-6; PMID:8790415; https://doi.org/10.1073/pnas.93.18.9821 [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Rozen S, Skaletsky H. Primer3 on the WWW for general users and for biologist programmers. Methods Mol Biol 2000; 132:365-86; PMID:10547847 [PubMed], [Google Scholar]