Role of DNA Methylation in Glioblastoma Development
Abstract
Glioblastoma (GBM) is the most common and malignant of the glial tumors. These tumors may develop from lower-grade astrocytomas (diffuse astrocytoma; grade II or anaplastic astrocytoma; grade III) through a progressive pathway, but, more frequently, they manifest de novo without any evidence of a pre-malignant lesion. The treatment of GBM includes surgery, radiotherapy, and chemotherapy with temozolomide. Despite improvements in treatment protocols, the median survival of GBM patients remains very low at 12-15 months.
The cause of glioma (either development or progression) can be genetic and epigenetic modification driven changes. In contrast to genetic modifications, where DNA sequence is changed, epigenetic modifications are those gene expression regulatory mechanisms which do not involve the change in the DNA sequence. It includes DNA methylation, chromatin modifications and miRNA mediated changes in gene expression. Aberrant DNA methylation is one of the common molecular lesions occurring in the cancer cell. The 5th position of cytosine (CpG) is the most preferred site of DNA methylation in mammalian cells. The methylated cytosines are prone to undergo oxidative deamination, and get mutated to thymine in DNA. Consequently, this led to decrease in CpG abundance in the genome. In normal conditions, promoters of majority of the genes escape methylation, because of which CpG of these regions remain same. This phenomenon led to the restriction of CpGs in the promoter regions of most of the genes. These CpG rich regions of the promoters are known as CpG islands, and the methylation status of these islands have a major role in regulating gene expression. The cancer genome is shown to undergo genome-wide hypomethylation whereas CpG islands undergo hypermethylation compared to normal tissue, resulting in net loss of total methylation, as the CpGs from non-island areas far exceed in number compared to the CpGs from islands.
The most studied change of DNA methylation in neoplasms is the silencing of the tumor suppressor genes by CpG island promoter hypermethylation. Apart from few studies, the role of DNA methylation in glioma development and progression is poorly known.
With this background, we have focused our study on DNA methylation changes in GBM. To identify GBM specific DNA methylation alterations, we have performed the genome wide methylation profile of 44 GBM and 8 normal samples using Infinium methylation array. Beta value, which is a measure of methylation, was calculated for all the CpG probes. Beta value ranges between 0-1 (from no methylation to complete methylation). We sought to understand the clinical importance, with particular importance to patient survival, of the DNA methylation pattern observed. We also undertook steps to understand the contribution of the differential DNA methylation and the associated gene expression changes in GBM development.
This work has been divided into three parts:
Part I –Identification of GBM specific methylome and development of a DNA methylation prognostic signature for GBM
To identify the differentially methylated genes in GBM, we compared the methylation levels of 27,578 CpGs between GBM and normal control samples using statistical methods. We then compared the list of differentially methylated genes with the expression data generated by The Caner Genome Atlas (TCGA) to find out genes whose expression oppositely correlates with the DNA methylation status. This resulted in the identification of 62 genes hypermethylated and down regulated, while 65 genes hypomethylated and up regulated. We believe that this set of differentially methylated genes may play important role in glioma development.
Next, to identify GBM specific DNA methylation survival signature, we correlated the survival data of 44 GBM patients with beta values of all the 27,578 probes. Using Cox regression method, we identified a set of 9 genes, whose methylation predicted the survival in GBM patients. A risk score was then calculated using methylation values and regression co-efficient of each of the genes. The methylation risk score was found to be an independent predictor of survival in a multivariate analysis in TCGA data set and the Bent et al data set (independent validation sets). Using methylation risk score, we were able to divide the patients into low and high risk groups with significant difference in survival. To discover the biology behind the difference in the survival of low and high risk groups, we performed network analysis, using differentially expressed genes between low and high risk patients, which revealed an activated NFkB pathway association with poor prognosis. The inhibition of NFkB pathway sensitized the glioma cells for chemotherapeutic drugs only in NFkB activated cell lines, suggesting a pivotal role for NFkB pathway imparting chemoresistance in poor surviving group.
Part II -NPTX2, a methylation silenced gene, inhibits NFkB through a p53-PTEN-PI3K-AKT signaling pathway
To understand the mechanism behind the prediction of survival by methylation of 9 genes, we took NPTX2 as a candidate gene for further investigation. NPTX2, a risky methylated gene, is highly methylated in high risk group with poor survival, which suggests that it may have a growth inhibitory activity in GBM. Bisulphite sequencing confirmed the hypermethylation status of NPTX2 promoter in GBM samples and glioma cell lines compared to normal brain tissue. As expected, NPTX2 transcript level was significantly down regulated in GBMs and glioma cell lines compared to normal samples, and could be re-expressed upon methylation inhibitor treatment in glioma cells. Exogenous over expression of NPTX2 inhibited proliferation, colony formation and sensitized glioma cells to chemotherapeutic drugs. Moreover, NPTX2 also inhibited soft agar colony formation in vitro, which confirms its growth inhibitory function in GBM. As NPTX2 was methylated and silenced in the high risk group, which has high activation of NFkB pathway, we then checked if NPTX2 could inhibit NFkB activity. Indeed, we observed that NPTX2 overexpression inhibited expression from NFkB dependent luciferase reporter, sequence-specific DNA-binding of NFkB, nuclear translocation of NFkB sub unit (p65) and it also significantly repressed key NFkB target genes. We also show that NPTX2 mediated inhibition of NFkB could be abrogated by co-expression
of constitutively active forms of PI3 kinase, AKT and IKKα, suggesting an involvement of PI3K-AKT-IKKα axis in NPTX2 mediated NFkB inhibition. Further, we found that NPTX2 repressed NFkB activity by inhibiting AKT through an ATM-p53-PTEN-PI3K dependent pathway. Thus, these results explain the need for hypermethylation and down regulation of NPTX2 in high risk GBM wherein the NFkB pathway is activated.
Part III -Methylation silencing of ULK2, an autophagy gene, is important for astrocyte transformation and cell growth
Among the differentially methylated genes (see part I), ULK2 was one of the most hypermethylated and down regulated genes. ULK2 is a known initiator protein in autophagy pathway, which is a type II cell death mechanism. There are many contradictory reports with respect to the role of autophagy in GBM development. For example, it has been shown that autophagy has a tumor suppressor activity and is essential for temozolomide mediated cell toxicity in GBM cells, whereas others studies implicate its involvement in tumor growth and progression. Hence, we carried out experiments to understand the role of ULK2 in GBM development. Using bisulphite sequencing, we validated ULK2 promoter hypermethylation status in GBM and glioma cell lines. In good correlation, ULK2 was found to be down regulated in GBMs and glioma cell lines, which was reexpressed by methylase inhibitor treatment in glioma cell lines.
The over expression of ULK2 was found to inhibit colony formation, proliferation and soft agar colony formation of glioma cells. As expected, ULK2 overexpressing cells showed higher autophagy, compared to control cells. Interestingly, we also found increased apoptosis in ULK2 overexpressing cells. The cell death caused by ULK2 overexpression was compromised when cells were treated with 3-MA (an autophagy inhibitor) or Z-VAD-FMK (a pan caspase inhibitor). However, ULK2 failed to inhibit cell growth in autophagy deficient cells (ATG5-/-), thereby suggesting the importance of autophagy in ULK2 induced cell death. Further, ULK2 overexpression, increased catalase degradation and Reactive Oxygen Species (ROS) generation, which suggests that increase in ROS may play a role in ULK2 dependent cell death. In good correlation, N-Acetyl Cysteine, a ROS inhibitor, treatment rescued the cells from ULK2 mediated cell death, confirming the role of ROS in ULK2 induced cell death. Kinase deficient ULK2 overexpression failed to induce cell growth inhibition, autophagy and apoptosis, suggesting kinase activity of ULK2 is important for ULK2 function. Co-transfection of ULK2 inhibited Ras mediated transformation of immortalized normal human astrocytes. Taken together, we have identified and validated ULK2 as one of the DNA methylation silenced genes in GBM. ULK2 was found to be growth inhibitory in GBM cells by increasing autophagy dependent apoptosis. ULK2 inhibited Ras mediated transformation, suggesting essentiality of DNA methylation mediated ULK2 down regulation in GBM.
In conclusion, the present work sheds light on the importance of methylation of genes in GBM progression. As observed, two of the genes, NPTX2 and ULK2 play as critical growth inhibitors in GBM. Also, we have identified a robust, independent and a highly sensitive 9 gene methylation signature, for GBM patient’s survival prediction.