Figure 1.
Methylation Profiles for Col and Ler Arabidopsis Ecotypes
(A) Microarray data from Chromosome 4 are displayed for a 125-kb region 9 Mb from the nuclear organizing region (Figure S1). Open reading frames from genes (yellow) and retrotransposons (green) are indicated, along with repeats predicted by RepBase and TandemRepeatFinder. Small RNA matches from massively parallel signature-sequencing (MPSS) data are indicated. Tiles that represent significant CNPs are highlighted in purple (CGH), while tiles that detect significant DNA methylation are highlighted in red for the two ecotypes (5 mC, Col and Ler). Examples of a gene CNP (a), a TE CNP (b), and a methylation polymorphism (c) are boxed.
(B) Significant methylation was detected by microarray analysis for two representative genes in Ler (At4g40980) and Col (At4g28850), respectively. This methylation was verified by digestion of genomic DNA by McrBC, followed by PCR amplification using primers specific for each gene (lower panels). Failure to amplify a product after digestion by McrBC indicates that the gene is methylated. Control primers from an unmethylated tile (ta25c11) and a methylated transposon (TA2) indicate complete digestion and amplification in each case.
Table 1.
CNP between Arabidopsis Ecotypes
Table 2.
Summary of Methylated Features
Table 3.
Methylation Polymorphism at TEs
Figure 2.
Variation in Methylation of Genes and Transposons
Methylation status of tiles detecting significant methylation but not overlapping CNPs (Table 1) was compared between ecotypes. Among these were tiles annotated as retrotransposons (1,551), transposons (623), tandem repeats (107), genes with repeats (904), genes (2,174), and unannotated (304). Methylation in Col (red), both ecotypes (green), and Ler (blue) is represented in the proportional bar graphs for each class of tile. A total of 45% of genic and unannotated tiles that detected significant methylation did so in only one of the two ecotypes, while 94% of TE tiles detected significant methylation in both (Table 2).
Figure 3.
Methylation is Localized within Genes
(A) Intermediate levels of methylation were detected by microarray analysis in Col and Ler for one representative gene (genome browser tracks are annotated as in Figure 1). Primer pairs 1–6 are indicated below each tile of the array. Amplification of McrBC digested Col DNA (+) and undigested Col DNA (−) was performed as in Figure 1. Failure to amplify digested DNA indicated that methylation was localized to regions 2, 4, and 6, rather than spanning the entire gene.
(B) The number of genic tiles detecting significant methylation was calculated at 10% intervals relative to the length of each gene and compared with the number expected if methylation was randomly distributed (black line). Only genes larger than 2 kb (three tiles or more per gene) were considered. Methylated tiles differing between ecotypes were also plotted in a similar way (light gray line). Genic methylation is largely concentrated in the middle of genes. For comparison, methylation distribution as a function of position was also calculated for TE-derived open reading frames (dark grey line) and is uniformly distributed.
Figure 4.
Heritability of Polymorphic Gene Methylation
The gene At4g28850 is methylated in Col but not Ler. Genomic DNA was prepared from F1 and F2 siblings derived from reciprocal crosses between Col and Ler and subjected to McrBC digestion and PCR amplification of this locus as in Figure 1. The amplification product (tb63b02) has a small deletion in Ler, enabling the parental alleles to be distinguished. DNA samples were digested (+) or mock-digested (−) with McrBC. Control primers were used to amplify a methylated retrotransposon (TA2) in each sample, as well as an unmethylated control tile (ta25c11). Samples of two F1 and ten F2 plants are shown for each cross (out of a total of eight and 40, respectively) (Table S3). In almost all cases, the Col allele is digested by McrBC, and the Ler allele is never digested. Two exceptions are indicated (a and b) in which the Col allele has lost all or most of its associated methylation.
Figure 5.
Genic Methylation among 96 Arabidopsis Accessions
Methylation polymorphism was surveyed at 18 genomic loci in 96 natural variation accessions of Arabidopsis [43], including the flowering time control locus fwa. Equal amounts of undigested (McrBC−) and digested genomic DNA (McrBC+) from each accession were amplified using PCR with primers specific for each locus. Methylation was scored as binary traits, represented in the graphical matrix by red (methylated) or black (unmethylated). Col-0 and Ler-1 are highlighted with an asterisk, and loci are arranged left to right in correspondence with their physical order on Chromosome 4. Per-accession methylation profiles were clustered using unweighted pair group method with arithmetic mean, and a support tree was generated using 10,000 bootstrapped replicates. The resulting tree bears no resemblance to a recent kinship-based phylogeny [42], and no major branches of the tree have significant support (unpublished data).