Identification of Predictive DNA Methylation Biomarkers for Chemotherapy Response in Colorectal Cancer

Abstract

Resistance to 5-Fluorouracil (5-FU) is a major obstacle to the successful treatment of colorectal cancer (CRC) and posed an increased risk of recurrence. DNA methylation has been suggested as one of the underlying mechanisms for recurrent disease and its contribution to the development of drug resistance remains to be clarified. This study aimed to determine the methylation phenotype in CRC for identification of predictive markers for chemotherapy response. We performed DNA methylation profiling on 43 non-recurrent and five recurrent CRC patients using the Illumina Infinium HumanMethylation450 Beadchip assay. In addition, CRC cells with different genetic backgrounds, response to 5-FU and global methylation levels (HT29 and SW48) were treated with 5-FU and DNA methylation inhibitor 5-aza-2'-deoxycytidine (5-azadC). The singular and combined effects of these two drug classes on cell viability and global methylation profiles were investigated. Our genome-wide methylation study on the clinical specimens showed that recurrent CRCs exhibited higher methylation levels compared to non-recurrent CRCs. We identified 4787 significantly differentially methylated genes (P < 0.05); 3112 genes were hyper- while 1675 genes were hypomethylated in the recurrent group compared to the non-recurrent. Fifty eight and 47 of the significantly hypermethylated and hypomethylated genes have an absolute recurrent/non-recurrent methylation difference of ≥20%. Most of the hypermethylated genes were involved in the MAPK signaling pathway which is a key regulator for apoptosis while the hypomethylated genes were involved in the PI3K-AKT signaling pathway and proliferation process. We also demonstrate that 5-azadC treatment enhanced response to 5-FU which resulted in significant growth inhibition compared to 5-FU alone in hypermethylated cell lines SW48. In conclusion, we found the evidence of five potentially biologically important genes in recurrent CRCs that could possibly serve as a new potential therapeutic targets for patients with chemoresistance. We postulate that aberrant methylation of CCNEI, CCNDBP1, PON3, DDX43, and CHL1 in CRC might be associated with the recurrence of CRC and 5-azadC-mediated restoration of 5-FU sensitivity is mediated at least in part by MAPK signaling pathway.

Keywords: 5-aza-2′-deoxycytidine; 5-fluorouracil; DNA methylation; chemoresistance; colorectal cancer; epigenetics; recurrent cancer.

FIGURE 1

Volcano plot representing differentially methylated probes. The significantly hypermethylated probes are shown in red while hypomethylated probes are shown in blue (P < 0.05).

FIGURE 2

Description of the 450K DNA methylation array. (A) General overview of significantly differentially methylated probes in relation to CpG content and neighborhood context. (B) Hypermethylated probes and (C) Hypomethylated probes according to CpG content and neighborhood context.

FIGURE 3

Heatmap of 50 differentially methylated genes in recurrent versus non-recurrent cases. Samples were clustered based on 4787 significant differentially methylated genes at P < 0.05. The row represents individual genes and the column represents individual samples. The pink color indicates recurrent cases and purple color indicates non-recurrent cases. The color in each small boxes represents the methylation level of the genes. The red boxes indicate genes that are hypermethylated while blue boxes represent genes that are hypomethylated.

FIGURE 4

Top 10 of significantly differentially methylated genes. The blue circles represent average β-value of respective genes in non-recurrent samples while black diamonds represent average β-value of respective genes in recurrent samples.

FIGURE 5

Gene ontology (GO) enrichment analysis of (A) hypermethylated genes and (B) hypomethylated genes with. The genes were clustered according to the cellular component, molecular function and biological process.

FIGURE 6

Predicted function of the five selected genes. These genes were mostly involved in cell cycle, cell growth, and cell differentiation.

FIGURE 7

Growth studies and effects of combination treatments of 5-FU with 5-azadC. Treatment of (A) 5-Fluorouracil (5-FU), (B) 5-Aza-2′-deoxycytidine (5-AzadC), (C) combination of 5-FU and 5-azadC on SW48 and HT29 for 72 h. Data is presented as mean ± SD, n = 9. ^∗P < 0.05 compared to untreated cells.

FIGURE 8

Methylation profiling of SW48 and HT29 cell lines. Methylation profiling was carried out following incubation with 5-FU, 5-aza-dC and combination of both drugs. Untreated cells were included as control. (A) Heatmap of the 50 top significantly differentially methylated genes in SW48. Red represent hypermethylated genes while blue represent hypomethylated genes. (B) Heatmap of the top 50 significantly differentially methylated genes in HT29.