ウシ網羅的メチル化解析におけるヒトメチル化チップ利用の可能性

抄録

DNA methylation is important for regulating gene expression during many biological processes. There are two popular techniques for detecting genome-wide DNA methylation profiles: next-generation sequencing and DNA arrays. We conducted a pilot study for data filtering using Human Methylation arrays to evaluate the genome-wide methylation profiles using bovine DNA samples. Five tissues of two female Japanese Black cattle were analyzed in this study. We focused on the detection P-value as an indicator of the reliability of data. Here, detection P-values were used for examining array performance, which indicate the proportion of 600 negative control probes with higher signal intensity than measured for each probe. We first investigated the distribution of detection P-values using bovine DNA, and next evaluated bovine sequences with high similarity to human sequences on the probes by a BLAST search. A total of 75,207 CpG sites had a detection P-value of 0 in all examined samples (15.5%). The detection P-values for bovine DNA were markedly worse than for humans and monkeys, but better than for mice. The CpG sites with low signal intensity apparently had low quality. The total signal intensity level at each CpG site was also shown to be very important to improve array performance, similarly to the detection P-value. Among the CpG sites that fulfilled the thresholds of both detection P-value and mean total signal intensity, it was determined that approximately 70% had been evaluated properly. Therefore, it was expected that useful methylation data with good quality were obtained for approximately 50,000 CpG sites (about 10% of all probes). These results show that the Human Methylation arrays have potential for evaluating a few tens of thousands of CpG sites in cattle by appropriate filtering to ensure the quality of the data.