Abnormal alteration in the DNA methylation status of CpG islands (CGIs) represents a stable tumor-specific marker that can be readily detected in body fluids. Methylation status varies within a CGI, hence, establishment of methods that allow screening of multiple CpG dinucleotides individually is necessary. Likewise, a quantitative measurement of methylated DNA molecules is needed to prevent biased analysis in clinical application. Oligonucleotide microarrays are known as powerful tools for methylation profiling of multiple genes simultaneously. However, application of oligonucleotide-based microarrays is challenged by cross-hybridization of unspecific targets to probes, which is due to low sequence variety of CGIs. In our approach, genomic DNA is hybridized to oligonucleotide probes subsequent to a treatment with sodium bisulfite. Two probes are used to analyze methylation status of each CpG dinucleotide; one of which is complementary to the unmethylated target (U) and the other one is matched to the methylated target (M). After calculating the median hybridization signal from at least four experimental replications, the intensity ratio M/(M+U), which is hereafter called methylation index, is computed for each oligomer pair and is considered as methylation level for the relevant CpG dinucleotide. To assess the effect of probe length on cross-hybridization, oligonucleotide probes with lengths of 17 and 25 nucleotides were generated for CGIs of BRCA1, ER, TWIST and HIN-1 genes by photo-controlled in situ synthesis using the Geniom One technology of febit biotech. We performed a calibration for each probe based on hybridizations with control samples of 0%, 50% and 100% methylation. Methylation indexes were calculated for each probe pair and probes which exhibited a methylation index of less than 0.2 for 0% methylated target, between 0.4 and 0.6 for 50% methylated target and more than 0.8 for 100% methylated target, were selected. Totally 10.5% of examined probes for all four genes could pass our selection criteria when using 17mer length, while this proportion increased to 23.4% using 25mer probes. In other words, using 25mer probes, more CpG dinucleotides within the candidate genomic regions could be analysed quantitatively. In addition, to evaluate the reproducibility of the performance of the probes, we synthesised two array sets with 17, 21 and 25mer probes for different genes but probes for only one gene were applied in both array sets. After hybridization with their target mixtures, methylation indexes of the same probes were compared between two array sets. The minimum differences were observed for probes with 25mer length. Taken together, we conclude that 25mer probes provide more specificity and reproducibility in quantitative oligonucleotide-based array methylation analysis.

[First AACR International Conference on Molecular Diagnostics in Cancer Therapeutic Development, Sep 12-15, 2006]