*2.3. Transgenic Process Leads to a Higher Number of Hyper-DMRs Than Hypo-DMRs*

We analyzed the differentially methylated regions (DMRs) of wild-type and transgenic lines and found that there were more hypo-DMRs (71%) in CK lines but more hyper-DMRs (78%) in transgenic lines in *pOsNAR2.1-OsNAR2.*1 under WYJ7. Similar results were observed in *pUbi-OsNAR2.1* and RNAi under NP (71% and 73%) and, for the SDWT line, the percentage of the hypo- and hyper- DMRs were very close (49% and 51%) (Figure 2a–c,f,g). In spite of global DNA methylation caused by the transgenic process being background sensitive, the insertion of the transgene was able to induce more hyper-DMRs than hypo-DMRs in both backgrounds. CK and the three transgenic lines contained 5122, 3539, 5253, and 4239 DMRs, respectively, which were considerably more comparable with the 2732 DMRs of SDWT. There were few common DMRs between these hypo- and hyper-DMRs in CK and *pOsNAR2.1-OsNAR2.1* lines, nor in SDWT and *pOsNAR2.1-OsNAR2.1* lines (Figure 2d,e). However, there were nearly 2000 common hyper-DMRs and nearly 500 common hypo-DMR between the *pUbi-OsNAR2.1* and RNAi lines (Figure 2h), caused by insertion of the transgene, rather than by the expression of *OsNAR2.1*.

**Figure 2.** Characteristics of DMRs. (**a**-**c**) Total number of DMRs and breakdown of hyper- and hypo-DMRs in CK, SDWT and *pOsNAR2.1-OsNAR2.1* samples. (**d**) Venn diagram of unique and shared hyper- and hypo-DMRs in *pOsNAR2.1-OsNAR2.1* and CK samples. (**e**) Venn diagram of unique and shared hyper- and hypo-DMRs in *pOsNAR2.1-OsNAR2.1* and SDWT samples. (**f**,**g**) Total number of DMRs and breakdown of hyper- and hypo-DMRs in *pUbi-OsNAR2.1* and RNAi samples. (**h**) Venn diagram of unique and shared hyper- and hypo-DMRs in *pUbi-OsNAR2.1* and RNAi samples.

#### *2.4. Tissue Culture Process Causing Random Epigenetic Changes*

We analyzed the differentially CpG methylated regions of five comparison groups: WYJ7 and SDWT, WYJ7 and CK, WYJ7 and *pOsNAR2.1-OsNAR2.1*, SDWT and *pOsNAR2.1-OsNAR2.1*, and CK and *pOsNAR2.1-OsNAR2.*1. We found that the mutual DMRs differed among them and calculated the CpG methylation level in these DMRs in different replicates of different samples. The clustered heatmap is shown in Figure 3. The heatmap shows that the DMR clustering pattern is similar between wild-type

and lines subjected to transgenic manipulation (SDWT and *pOsNAR2.1-OsNAR2.1*), whereas a more variable pattern is observed for CK replicates. The result shows that different CpG clusters of common DMRs were detected in different CK lines regenerated from the same tissue culture process with the same explant WYJ7 seeds (Figure 3). This suggests that the tissue culture process could randomly alter the epigenetic status. On the other hand, a tissue culture process followed by a transgenic process tends to lead to a more consistent CpG methylation level clustering pattern (Figure 3), suggesting that the influence of the transgenic process on the rice epigenome is stronger than the influence of the tissue culture process.

**Figure 3.** Heatmap for DMRs methylation level cluster. Heatmap representation of hierarchical clustering based on CG methylation levels within DMRs. Rows represent all DMRs identified and columns represent the samples.
