Fig. 4: Epigenetic editing shows a causative role for DNA methylation of INS, TH and GCG on gene expression in beta cells.
a, The location of a human alpha versus beta cell INS DMR (purple; see also Fig. 2e), gRNA binding site (pink), TF binding in islets (https://pasqualilab.upf.edu/app/isletregulome/; blue), promoter (green) and enhancers (red). The zoomed-in part (bottom) shows the sequence targeted by the gRNA (black) and the two CpG sites covered by the pyrosequencing assay (orange). The figure was made with the help of the UCSC Genome Browser. INS runs from right to left in this image. b–d, The gRNA binding sites (pink), enhancer (c; red) and CpG sites where DNA methylation changed after epigenetic editing (P < 0.05) within the targeted loci for INS (b), TH (c) and GCG (d) analysed by pyrosequencing (displayed in blue), or EPIC v2.0 array (displayed in black). e, The TH DMR with differential DNA methylation in human alpha versus beta cells, in the region targeted by epigenetic editing. q value was corrected for multiple testing. Locations of the methylation loci are shown with black lines at the bottom, and average, smoothed lines based on the dmrseq25 analysis are shown in pink (beta cells) or blue (alpha cells), including the standard deviation. The shaded area shows the DMR. Annotations to CpG island regions and genes are shown below the graph. f, Design of the epigenetic editing experiments performed to investigate the causal relationship between altered DNA methylation and changes in gene expression/cell function. g, Pyrosequencing shows that epigenetic editing with CRISPR–dCas9–DNMT3A and the gRNA targeting the INS DMR (INS-DMR-gRNA) increased DNA methylation of two analysed sites (CpG Chr11: 2161227 and CpG Chr11: 2161277, blue; b), compared with the negative control including DNMT3A but not the INS-DMR-gRNA in EndoC-βH1 beta cells. n = 4 biological replicates, **P < 0.01, based on one-tailed paired t-tests (left, P = 0.008; right, P = 0.006). h, The EPIC v2.0 array identified edited CpG sites between 10 kb upstream and downstream of INS, discovering absolute methylation changes of up to 40% between beta cells that had received both CRISPR–dCas9–DNMT3A and the INS-DMR-gRNA and the negative control (only DNMT3A). n = 4 biological replicates, *P < 0.05, **P < 0.01 and ***P < 0.001 based on one-tailed paired t-tests. i, Pyrosequencing data of two CpG sites with altered methylation after epigenetic editing with CRISPR–dCas9–DNMT3A together with TH-DMR-gRNA versus control (only DNMT3A). n = 4 biological replicates, *P ≤ 0.05 based on one-tailed paired t-tests (left, P = 0.05; right, P = 0.04). j, presents CpG sites identified by EPIC v2.0 array, where DNA methylation was altered after epigenetic editing with CRISPR–dCas9–DNMT3A and the TH-DMR-gRNA versus control (only DNMT3A). n = 4 biological replicates, *P < 0.05, **P < 0.01 and ***P < 0.001 based on one-tailed paired t-tests. k–m, CRISPR–dCas9–DNMT3A-based editing of INS or TH reduced INS (P = 0.001) and TH (P = 0.023) expression, respectively (quantitative PCR (qPCR), n = 4 biological replicates, k and l), and insulin content (P = 0.021, only after editing with INS-DMR-gRNA, n = 3 biological replicates, m) in EndoC-βH1 beta cells. *P < 0.05, **P < 0.01, based on one-tailed paired t-tests. n,o, Pyrosequencing and EPIC v2.0 array data show that epigenetic editing with CRISPR–dCas9–TET1 and the gRNA, GCG-DMR-gRNA, targeting the GCG DMR in Fig. 2g decreased DNA methylation of two CpG sites (CpG Chr2: 162152662 for pyrosequencing and cg05108984 for EPIC v2.0) versus the control (only TET1) in EndoC-βH1 beta cells. n = 6 biological replicates for pyrosequencing, *P = 0.05 based on one-tailed paired t-test. n = 4 biological replicates for EPIC v2.0 array, **P = 0.00741, based on one-tailed paired t-test. p,q, CRISPR–dCas9–TET1-based editing with GCG-DMR-gRNA caused increased GCG expression (*P = 0.047, qPCR, n = 6 biological replicates, p), and glucagon content (**P = 0.008, ELISA, n = 5 biological replicates, q) versus the control (only TET1) in EndoC-βH1 beta cells, based on one-tailed paired t-tests. The colours of the dots in g–m and n–q represent samples from the same experiments. Data are presented as the mean ± s.e.m. Bars shown in light grey include CRISPR–dCas9–DNMT3A or CRISPR–dCas9–TET1 plus gRNA, while white bars are controls only including CRISPR–dCas9–DNMT3A or CRISPR–dCas9–TET1, and no gRNA.
