Front. Neurosci., 04 January 2010 | http://dx.doi.org/10.3389/neuro.15.005.2009
In silico enhanced restriction enzyme based methylation analysis of the human glioblastoma genome using Agilent 244K CpG Island microarrays
Department of Neurology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
The Henry E Singleton Brain Cancer Research Program, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Department of Neurosurgery, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Department of Human Genetics, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Genome wide methylation profiling of gliomas is likely to provide important clues to improving treatment outcomes. Restriction enzyme based approaches have been widely utilized for methylation profiling of cancer genomes and will continue to have importance in combination with higher density microarrays. With the availability of the human genome sequence and microarray probe sequences, these approaches can be readily characterized and optimized via in silico modeling. We adapted the previously described HpaII/MspI based Methylation Sensitive Restriction Enzyme (MSRE) assay for use with two-color Agilent 244K CpG island microarrays. In this assay, fragmented genomic DNA is digested in separate reactions with isoschizomeric HpaII (methylation-sensitive) and MspI (methylation-insensitive) restriction enzymes. Using in silico hybridization, we found that genomic fragmentation with BfaI was superior to MseI, providing a maximum effective coverage of 22,362 CpG islands in the human genome. In addition, we confirmed the presence of an internal control group of fragments lacking HpaII/MspI sites which enable separation of methylated and unmethylated fragments. We used this method on genomic DNA isolated from normal brain, U87MG cells, and a glioblastoma patient tumor sample and confirmed selected differentially methylated CpG islands using bisulfite sequencing. Along with additional validation points, we performed a receiver operating characteristics (ROC) analysis to determine the optimal threshold (p ≤ 0.001). Based on this threshold, we identified ∼2,400 CpG islands common to all three samples and 145 CpG islands unique to glioblastoma. These data provide general guidance to individuals seeking to maximize effective coverage using restriction enzyme based methylation profiling approaches.