Global methylation screening in the Arabidopsis thaliana and Mus musculus genome: applications of virtual image restriction landmark genomic scanning (Vi-RLGS)

Abstract

Understanding the role of 'epigenetic' changes such as DNA methylation and chromatin remodeling has now become critical in understanding many biological processes. In order to delineate the global methylation pattern in a given genomic DNA, computer software has been developed to create a virtual image of restriction landmark genomic scanning (Vi-RLGS). When using a methylation- sensitive enzyme such as NotI as the restriction landmark, the comparison between real and in silico RLGS profiles of the genome provides a methylation map of genomic NotI sites. A methylation map of the Arabidopsis genome was created that could be confirmed by a methylation-sensitive PCR assay. The method has also been applied to the mouse genome. Although a complete methylation map has not been completed, a region of methylation difference between two tissues has been tested and confirmed by bisulfite sequencing. Vi-RLGS in conjunction with real RLGS will make it possible to develop a more complete map of genomic sites that are methylated or demethylated as a consequence of normal or abnormal development.

Figure 1

Application of Vi-RLGS to the Arabidopsis genome. (a) Vi-RLGS profile and (b) real RLGS profile of Arabidopsis nuclear DNA (NotI–EcoRV–MboI). Vi-RLGS spots are defined in the profile with the chromosome number (the first bold and underlined numerical number) and clone ID (the following five to six digit code). Red arrows in (b) indicate invisible or very faint spots in the real RLGS that were predicted from virtual RLGS. A yellow arrow indicates an example where no corresponding spot is seen in the Vi-RLGS. This may be due to incomplete sequence information. (c) PCR amplification designed for two unmethylated (in 4T16H5 and 4F13C5) and six methylated (in 2F26H6, 2F12C20, 3T28A8, 4T24H24, 5T20L15 and 5MTH12) NotI sites whose methylation status in the genome is predicted by subtraction between Vi-RLGS and real RLGS. The BAC/PAC ID numbers following the chromosome number are indicated on each lane. PCRs are performed for NotI digested (–NotI) and undigested (–) Arabidopsis genomic DNAs. Primers designed to amplify an ∼1-kb sequence containing the NotI sites predicted by Vi-RLGS were synthesized. If amplification products are produced (+) in NotI digested DNA, and the products have a NotI site, this means that the NotI sites are methylated and the virtual RLGS spot will not be detected in the real RLGS pattern. (d) Comparison between real RLGS patterns of normal (top) and hypomethylated (bottom) Arabidopsis genome. Three out of four invisible spots (red arrows in control) predicted from virtual RLGS appear on the real RLGS pattern of hypomethylated Arabidopsis genome DNA (black arrows).

Figure 2

A NotI physical map of Arabidopsis with the methylation status predicted by Vi-RLGS and real RLGS. Methylated spots were confirmed in triplicate by PCR assays. Loci with asterisks have not yet been confirmed by the PCR method. When there are two or three NotI sites within one BAC/PAC, they are numbered in order from centromere to telomere.

Figure 3

Application of Vi-RLGS to the mouse genome. (a) A region corresponding to 6–8 kb in the first dimension, and 80–110 bp in the second dimension is shown. The chromosome number is underlined followed by the alphabetical identification. The virtual spots are color coordinated according to chromosome. The real profile is from C57BL/6J kidney. (b) PCR analysis using each sequence-specific primer set. Templates from C57BL/6J genomic DNA and eluted spot DNAs are indicated on each lane. Spots C and D overlap in the real RLGS profile and could not be purified separately. Therefore, the spot DNA was amplified by both sets of primers (C and D). (c) Sequence trace of C57BL/6J normal and bisulfite-treated genomic liver DNA. Bisulfite sequencing indicates that the two CpGs at NotI sites are highly methylated (underlined). BLAST analysis indicates that there is only one LTR with 100% homology to the LTR sequence-specific primers used for the PCR.

Figure 4

Correspondence between mouse real (bottom) and virtual RLGS (top) profiles in the region corresponding to 4–8 kb in the first dimension and 110–180 bp in the second. The PCR results for 17 RLGS spots in four divided areas (a, b, c and d) are shown. C57BL/6J genomic DNA and eluted spot DNAs are amplified by each sequence-specific primer set in each area. The sequences of the PCR products are then confirmed by direct sequencing (data not shown). The black arrow indicates an unusual spot that is present in the real profile but is absent from the virtual profile (see text for explanation).

Figure 5

Differential methylation in tissue development. Real RLGS profiles from C57BL/6J liver and C57BL/6J kidney genomic DNA. The spot indicated by the arrow is very faint in liver genomic DNA. The spot DNA sequence was confirmed by PCR of spot DNA eluted from the kidney genomic DNA gel using primers determined from the virtual sequence. Bisulfite sequencing of the liver genomic DNA indicates that the two CpGs at NotI sites and two adjacent sites are highly methylated (underlined).