DNA methylation dynamics in human induced pluripotent stem cells over time

Abstract

Epigenetic reprogramming is a critical event in the generation of induced pluripotent stem cells (iPSCs). Here, we determined the DNA methylation profiles of 22 human iPSC lines derived from five different cell types (human endometrium, placental artery endothelium, amnion, fetal lung fibroblast, and menstrual blood cell) and five human embryonic stem cell (ESC) lines, and we followed the aberrant methylation sites in iPSCs for up to 42 weeks. The iPSCs exhibited distinct epigenetic differences from ESCs, which were caused by aberrant methylation at early passages. Multiple appearances and then disappearances of random aberrant methylation were detected throughout iPSC reprogramming. Continuous passaging of the iPSCs diminished the differences between iPSCs and ESCs, implying that iPSCs lose the characteristics inherited from the parent cells and adapt to very closely resemble ESCs over time. Human iPSCs were gradually reprogrammed through the "convergence" of aberrant hyper-methylation events that continuously appeared in a de novo manner. This iPS reprogramming consisted of stochastic de novo methylation and selection/fixation of methylation in an environment suitable for ESCs. Taken together, random methylation and convergence are driving forces for long-term reprogramming of iPSCs to ESCs.

Figure 1. Pluripotent stem cells are significantly more hyper-methylated than their parent cells.

(A) The human cell origins used for generation of iPSCs. (B) Morphology of the parent cells (upper panels) and iPSCs (lower panels). (C) Unsupervised hierarchical clustering analysis based on DNA methylation. (D) Distribution of 24,273 CpG sites with their methylation scores in the parent cells, iPSCs and ESCs. (E) The average number of high (>0.6) methylated CpG sites. The iPSCs have more highly methylated sites than the parent cells.

Figure 2. Defining stem cell-specific DMRs as novel epigenetic iPS markers.

(A) Venn-like diagram showing overlapping CpG sites among ESCs, iPSCs and their parent cells. The 220 overlapping sites are stem cell-specific differentially methylated regions (DMRs). Notably, neither overlapping iPSCs-specific DMRs nor inherited regions in iPSCs from the parent cells were observed. (B) Proportion of the hyper- and hypo-methylated stem cell-specific DMRs and GO analysis. Approximately 80% of the regions were hyper-methylated in iPSCs, compared with that of the parent cells. (C) Proportion of the regions associated with CpG islands and non-CpG islands in the hypo-methylated stem cell-specific DMRs. The hypo-methylated regions were biased to CpG islands, whereas the hyper-methylated regions were biased to non-CpG islands. (D) DNA methylation levels in the 8 representative genes determined by Illumina Infinium HumanMethylation27 assay and Bio-COBRA. These 8 genes were defined as SS-DMRs with significant changes of expression and were described in Table S6. The relative amount of methylated and unmethylated DNA ratio is indicated as the black and white area, respectively, in the pie chart. (E) Expression of the 8 genes. Expression of the 8 genes had an inverse correlation with DNA methylation level. (F) Bisulfite sequencing analysis of the 8 genes in endometrial cells (UtE1104), UtE-iPS-11 and HUES-8 cells. (Top) Schematic diagram of the genes. Arrows, open boxes and open circles represent transcription start site, first exon and position of CpG sites, respectively. (Bottom) Open and closed circles indicate unmethylated and methylated sites, respectively. Red and blue arrowheads represent the position of CpG sites in Infinium assay and COBRA assay, respectively.

Figure 3. Aberrant methylation in human iPSCs.

(A) Comparison of DNA methylation states of each iPSC line or each parent cell line with that of ESCs. The DMRs between ESCs and iPSCs are designated as ES-iPS-DMRs, and the DMRs between ESCs and parent cells are designated as ES-parent-DMRs. (B) The number of ES-iPS-DMRs and ES-parent-DMRs on whole genome (top), autosomes (middle) and X chromosome (bottom). Ratios of number of inherited regions in iPSCs from parent cells (blue) and aberrant regions in iPSCs that differ from ESCs and parent cells (red) in the ES-iPS-DMRs are shown in bars. Female iPSCs were demonstrated to carry high number of EiP-DMRs on X chromosome. (C) Number of overlapped ES-iPS-DMRs frequency in iPSCs. No overlapping ES-iPS-DMRs in all 22 iPSC lines. (Inlet) A small number of overlapping ES-iPS-DMRs of the frequency from 15 to 22. Overlapping frequency of each gene is indicated in parentheses. (D) Proportion of the hyper- and hypo-methylated ES-iPS-DMRs. More than 75% of the ES-iPS-DMRs were hyper-methylated in iPSCs. (E) Proportion of the ES-iPS-DMRs associated with CpG islands and non-CpG islands in ach iPSC line. ES-iPS-DMRs were biased to CpG islands.

Figure 4. Effect of long-term cultivation on ES-iPS-DMRs.

(A) Decrease in the number of the ES-iPS-DMRs with continuous passaging. Upper panels show change of the number of the ES-iPS-DMRs (left), the inherited regions (middle) and aberrant regions (right) on whole genome. Lower panels show change in the number of the ES-iPS-DMRs (left), inherited regions (middle) and aberrant regions (right) on X chromosome. The number of the ES-iPS-DMRs in XX-iPSCs approached zero with continuous passaging on X chromosome. In contrast, XY-iPSCs had few ES-iPS-DMRs on X chromosome throughout the passages. (B) The number of the ES-parent-DMRs with continuous passaging. (C) No expression of the transgenes in iPSCs at each passage was detected by RT-PCR.

Figure 5. Number of the ES-iPS-DMRs and ES-parent-DMRs with passaging.

(A) Number of the ES-iPS-DMRs with passaging. Red line plots indicate total number of the ES-iPS-DMRs. Blue bars indicate the number of the ES-iPS-DMRs that appeared at the earliest passage. Orange, green and red bars indicate the number of the ES-iPS-DMRs that appeared secondarily at later passages. Appearance/disappearance of the ES-iPS-DMRs and inherited regions were repeated, but the number of newly-appeared ES-iPS-DMRs was decreased with passaging. (B) Number of the ES-parent-DMRs with passaging. Blue bars indicate the number of the ES-parent-DMRs at P5 (or P7). Orange and green bars indicate de novo ES-parent-DMRs at P11 and P16, respectively.

Figure 6. Hyper-methylation in the ES-iPS-DMRs and ES-parent-DMRs.

ES-iPS-DMRs that disappeared in UtE-iPS-11 and Edom-iPS-2 at the latest passage (upper) were analyzed and the methylation score of each ES-iPS-DMR was plotted on bar graph (middle). To clearly compare methylation scores, difference value were estimated by subtracting the scores of ESCs from that of each sample (lower). Red and blue bars represent hypo- and hyper-methylated regions, respectively, in the parent cells, compared with ESCs. Notably, almost all the regions, even though their difference values were hypo-methylated in the parent cells, became hyper-methylated in iPSCs at the early passage, and then their methylation levels were adjusted to the level of ESCs with passaging, i.e. subtracted methylation score became close to zero. This transiently-induced hyper-methylation was not detected in parent cells.

Figure 7. Model of mechanism for transgene-independent reprogramming.

During reprogramming from somatic cells to iPSCs, the cells undergo dynamic change of methylation of SS-DMRs and genome. The cells with incomplete reprogramming or excessive hyper-methylation of the genome fail to maintain pluripotency at early passages. Human iPSCs are transgene-independently reprogrammed gradually through “convergence” of periodic aberrant hyper-methylation and become closer to ESCs upon continuous passaging. Due to the sensitivity to aberrant methylation on X chromosome, XY-iPSCs become close to ESCs faster than XX-iPSCs do.