Figure 5—figure supplement 2. AtCMT3-induced genic CG methylation is maintained at higher levels than background following loss of AtCMT3 expression. ; (A) Comparison of the % methylation of CG sites that were not found to be methylated in the non-transgenic wild type (Shandong ecotype) parent of AtCMT3 transgenic lines. Shown is the % CG methylation calculated in each lineage across hyper CHG DMRs identified in AtCMT3-L2T4 that overlap CHG-gain genes identified in AtCMT3-L2T4 (black bars; same regions assessed in Figure 5B) compared to the average % CG methylation of an equal amount of sequence space extracted from five randomly chosen sets of unmethylated genes that did not gain CHG methylation in AtCMT3-L2T4. The number of genes chosen in each set was equal to the number of AtCMT3-L2T4 CHG gain genes. Note that AtCMT3-L2T5 and T6 exhibited silencing of the AtCMT3 transgene, yet still maintained CG methylation levels higher than that detected on unmethylated genes. Methylation over the same regions was also assessed in additional, non-transgenic E. salsugineum accession (Yukon) to demonstrate that the levels of CG methylation over CHG gain genes in transgenic lines were unlikely to have occurred independently of AtCMT3. (B) Comparison of the % methylation of CG sites that were not found to be methylated in the non-transgenic wild type (Shandong ecotype) parent of AtCMT3 transgenic lines. Shown is the % CG methylation calculated in each lineage across hyper CHG DMRs identified in AtCMT3-L1T5 that overlap CHG-gain genes identified in AtCMT3-L1T5 (black bars; same regions assessed in Figure 5—figure supplement 1E) compared to the average % CG methylation of an equal amount of sequence space extracted from five randomly chosen sets of unmethylated genes that did not gain CHG methylation in AtCMT3-L1T5. The number of genes chosen in each set was equal to the number of AtCMT3-L1T5 CHG gain genes. AtCMT3-L1T5 was crossed to wild type (non-transgenic) and three F2 progeny were assessed: one progeny that retained the transgene (L1T5XWT F2 (+CMT3)) and two where the transgene segregated out (L1T5XWT F2 (-CMT3) #1 and #2). Note that the F2 progeny where the transgene segregated out still maintained CG methylation levels higher than that detected on unmethylated genes. As in (A), methylation over the same regions was also assessed in additional, non-transgenic E. salsugineum accession (Yukon) to demonstrate that the levels of CG methylation over CHG gain genes in transgenic lines were unlikely to have occurred independently of AtCMT3. Error bars are ± one standard deviation of the mean.

Figure 5—figure supplement 2. AtCMT3-induced genic CG methylation is maintained at higher levels than background following loss of AtCMT3 expression. ; (A) Comparison of the % methylation of CG sites that were not found to be methylated in the non-transgenic wild type (Shandong ecotype) parent of AtCMT3 transgenic lines. Shown is the % CG methylation calculated in each lineage across hyper CHG DMRs identified in AtCMT3-L2T4 that overlap CHG-gain genes identified in AtCMT3-L2T4 (black bars; same regions assessed in Figure 5B) compared to the average % CG methylation of an equal amount of sequence space extracted from five randomly chosen sets of unmethylated genes that did not gain CHG methylation in AtCMT3-L2T4. The number of genes chosen in each set was equal to the number of AtCMT3-L2T4 CHG gain genes. Note that AtCMT3-L2T5 and T6 exhibited silencing of the AtCMT3 transgene, yet still maintained CG methylation levels higher than that detected on unmethylated genes. Methylation over the same regions was also assessed in additional, non-transgenic E. salsugineum accession (Yukon) to demonstrate that the levels of CG methylation over CHG gain genes in transgenic lines were unlikely to have occurred independently of AtCMT3. (B) Comparison of the % methylation of CG sites that were not found to be methylated in the non-transgenic wild type (Shandong ecotype) parent of AtCMT3 transgenic lines. Shown is the % CG methylation calculated in each lineage across hyper CHG DMRs identified in AtCMT3-L1T5 that overlap CHG-gain genes identified in AtCMT3-L1T5 (black bars; same regions assessed in Figure 5—figure supplement 1E) compared to the average % CG methylation of an equal amount of sequence space extracted from five randomly chosen sets of unmethylated genes that did not gain CHG methylation in AtCMT3-L1T5. The number of genes chosen in each set was equal to the number of AtCMT3-L1T5 CHG gain genes. AtCMT3-L1T5 was crossed to wild type (non-transgenic) and three F2 progeny were assessed: one progeny that retained the transgene (L1T5XWT F2 (+CMT3)) and two where the transgene segregated out (L1T5XWT F2 (-CMT3) #1 and #2). Note that the F2 progeny where the transgene segregated out still maintained CG methylation levels higher than that detected on unmethylated genes. As in (A), methylation over the same regions was also assessed in additional, non-transgenic E. salsugineum accession (Yukon) to demonstrate that the levels of CG methylation over CHG gain genes in transgenic lines were unlikely to have occurred independently of AtCMT3. Error bars are ± one standard deviation of the mean.