Transfections and control design

All transfections were done with lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. The ratios of co-transfected plasmids were as follows: 1 gRNAs: 2 px330-dCas9: 1 pcDNA3-Dnmt3A-3 L (test) or pcDNA3 (control) for qPCR; 1 gRNAs: 1 pCAG-dCas9-5xPlat2AflD: 1 pCAG-scFvGCN4sfGFPTET1CD (test) or pcDNA3 (control) for qPCR; 19 pGL3-promoter or pGL3-promoter-enhancer: 1 pRL-TK for luciferase assay; and 5 gRNAs: 10 Px330-dCas9: 5 pcDNA3-Dnmt3A-3 L (test) or pcDNA3 (control): 19 pGL3-promoter or pGL3-promoter-enhancer: 1 pRL-TK for luciferase assay.

Above “pcDNA3 (control)” is a control for dCas9 targeting, in which dCas9 co-transfected with empty pcDNA3 without DNMT3A-3 L/TET1. The same conclusion as shown in Figs. ​Figs.3d3d and ​and4c4c can be reached by using scrambled sgRNA as the control for the qPCR test (Additional file 1: Figure S8c). dCas9 targeting specificity was confirmed with off-target test by bisulfite sequencing of non-targeted sites (WNT5B-sgRNA in Additional file 1: Figure S7a vs Fig. ​Fig.3c,3c, and NDRG1-sgRNA in Additional file 1: Figure S7b). dCas9 targeting specificity was also confirmed with qPCR quantifying other non-targeted genes with WNT5B-sgRNA (Additional file 1: Figure S7c). Furthermore, we performed experiments by using “untargeted” or catalytically inactive DNMT3A-3 L/TET1 to rule out the possibility of off-target due to overexpression DNMT3A-3 L/TET1 (Additional file 1: Figure S8a,b). The “untargeted” constructs were generated by removal of the PUFa linker from DNMT3A-3 L-fusion, or removal of scFv linker from TET1-fusion (Additional file 1: Figure S6). For these “untargeted,” catalytically active DNMT3A-3 L/TET1 was overexpressed but targeted to nowhere due to the deletion of “linker” domain. The “untargeted” or catalytically inactive DNMT3A-3 L/TET1 did not result in any significant change of the target gene expression (Additional file 1: Figure S8a, b).