You are hereAugust 18, 2011 | Pluripotent Stem Cells
New insight into how a somatic past shapes the future of human iPSCs
From Nature Cell Biology
Paper commentary by Carla Mellough
The differences between human embryonic stem cells (hESC) and their somatically derived counterparts, induced pluripotent stem cells (iPSC), have been under close scrutiny following the accumulation of various reports indicating greater disparity between the two cell types than originally envisaged (for example see iPSC don’t forget their origins). Conflicting results have been reported and remain unresolved, for example the transcriptional signature of iPSCs, although ascribed to partial memory retention of their somatic origin, does not always correlate with the differences in gene expression between iPSCs and hESCs. This has led to some of the biases being attributed to interlaboratory methodological variation. Functional disparity between differentiating hESCs and iPSCs has also been reported, with iPSC derivatives showing limited differentiation potential and early senescence and perhaps indicating that iPSCs do not hold a comparable clinical value to hESCs.1 This, alongside numerous reports highlighting the remarkable similarity of both cell types2 has resulted in some confusion regarding the applicability of iPSCs to translational research. For this reason, elucidation of the true likeness between iPSCs and hESCs has been somewhat limited. A recent study published in Nature Cell Biology from various centres at the University of California by Ohi et al.3attempts to address these limitations by systematically comparing human iPSC lines derived from multiple somatic cells types and under the same methodology, in parallel.
In their study, Ohi et al.3 generated iPSCs from adult hepatocytes (of endodermal origin), newborn foreskin fibroblasts (mesoderm) and adult melanocytes (ectoderm) and validated the pluripotency of all three lines. Using Affymetrix ST 1.0 microarrays they performed transcriptional profiling on each line in triplicate alongside three hESC lines (H1, H7, H9). Cluster analysis showed that all of the iPSC lines grouped together with the hESCs demonstrating that reprogramming of the somatic transcriptional profile back to an ESC-like state had occurred. Differential gene expression patterns were investigated by plotting differences in gene expression between the iPSC and hESC lines against the differential gene expression between the somatic cell of origin and hESCs. This analysis revealed that some of the genes found to be highly expressed in somatic cells, although repressed in the respective iPSC lines, in fact remained partially elevated in all three iPSC lines when compared with hESCs. Ohi et al.3 then used differential expression via distance synthesis (DEDS) analysis, a robust statistical method, which confirmed that a very significant proportion of these genes represented a transcriptional memory of the parental somatic cells. Interestingly, further investigation revealed that the nature of this memory was not related to master lineage-specifying transcription regulators, such as hepatocyte nuclear factor, but that the majority of differentially expressed genes between iPSCs and hESCs were also differentially expressed across two or three of the somatic cell types, indicating that this memory was not cell-type specific but represented the transcriptional memory of a general differentiated state. Importantly, the authors also demonstrate that early passage iPSCs of endodermal origin produced no greater proportion of endodermal derivatives upon differentiation when compared to iPSCs of mesodermal origin, contrary to previous reports.4,5,6
The authors go on to examine the contribution of DNA methylation to the differential expression observed between iPSCs and hESCs using bisulphite sequence analysis. This revealed that although DNA methylation can partially explain somatic cell gene expression in iPSCs, differential expression relating to differential DNA methylation states was not dependant on the somatic cell-type of origin, but instead indicated that incomplete initiation of new promoter DNA methylation of CpG islands during reprogramming may be the causal factor. To investigate this, Ohi et al.3 examined the CpG promoter methylation of the top four (C9orf64, TRIM4, COMT and CSRP1) somatic genes which remained elevated across the iPSC lines. This showed that the promoter of C9orf64, TRIM4 and COMT genes lacked CpG methylation in somatic cells and iPSCs but showed heavy methylation in hESCs, while CSRP1 had high to low CpG methylation in hESCs, iPSCs and somatic cells, respectively. Impressively, to validate this result the authors then introduce an additional four hESC and four independent iPSC lines to the study for analysis, incorporating iPSCs that had been generated by different methods including RNA transfection. Indeed, their results held true following this comparison and, furthermore, hypomethylation of C9orf64, TRIM4 and COMT could also be demonstrated in 6 late passage (>30) iPSC lines when compared with late passage hESC lines, confirming that somatic gene hypomethylation can persist in iPSCs and correlates with the differential expression of iPSCs to hESCs. This is in contrast to another report indicating that early passage mouse iPSCs only retain a ‘hangover’ from their differentiated state which becomes diluted out following 10 or more passages7, but is in accordance with another study done in human iPSCs which indicates that although iPSCs become more similar to hESC with increased passage, their signature gene expression remains distinct.8
So how do these expression patterns compare with other published datasets? To establish this, the authors performed meta-analysis using eight external independent studies and two recent large pooled data set studies which compare multiple iPSC with multiple hESC lines. This confirmed that the differential expression of C9orf64 and TRIM4 (the most incompletely silenced genes observed in their own data set) could also be found within the top differentially expressed genes between iPSCs and hESCs in other independent data sets, and meta-DEDS analysis not only found a correlation between expression levels but a significant overlap of differentially expressed genes across all of the data sets studied. Further, they detected a significant correlation between transcription and DNA methylation levels in the pooled data sets, indicating that differences in DNA methylation levels at some somatic cell genes may in fact underlie the somatic memory of low passage iPSCs. This was not caused by an insufficiency in DNA methyltransferases (DNMTs), as Ohi et al.3 found DNMT levels to be equivalent levels across the iPSC and hESC lines tested. They did, however, find a pattern relating to the genomic location of incompletely silenced genes; these were separated from other genes that undergo successful silencing, indicating a lag or insufficiency in the silencing machinery at somatic genes that persist in their expression following reprogramming.
To determine any influence of incomplete silencing of the top differentially expressed gene in iPSCs on the reprogramming process, the authors performed RNA-mediated interference (RNAi) against C9orf64 during reprogramming. Interestingly, this interference decreased the formation of iPSC colonies that were immunopositive for the pluripotent stem cell marker Tra1-81, revealing a role for this gene in reprogramming.
This study is a conscientious and comprehensive effort to evaluate the comparability of iPSCs that were generated from multiple germ layer derivatives to hESCs and in fact proves a great deal of similarity between both cell types. Nonetheless, this work does confirm that a transcriptional ‘hangover’ exists from an iPSC’s somatic history, which appears to be related to a general differentiated state and not, as previously thought, distinct and germ layer specific paternal somatic gene expression. The RNAi data is particularly interesting and indicates that, in some cases, persistent somatic gene expression may in fact actively aid the reprogramming process, indicating a complex interplay between DNA demethylation and methylation during reprogramming. This certainly prompts further study, given that the role and protein domains of C9orf64 remain unknown. How the proximity of genes undergoing silencing effects epigenetic and transcriptional reprogramming is also particularly interesting for continued investigation, as in this study while the repression of multiple genes in close proximity appeared to enhance local silencing, more isolated genes fell short of this silencing machinery synergy and remained incompletely repressed. Importantly, further elucidation of the effects of incompletely silenced genes associated with somatic cell memory in iPSCs may provide some insight into cancer progression, as several of these are deleted or epigenetically silenced in numerous cancers.
1 Feng Q, Lu SJ, Klimanskaya I et al. Hemangioblastic derivatives from human induced pluripotent stem cells exhibit limited expansion and early senescence. Stem Cells. 2010 Apr;28(4):704-12.
2 Armstrong L, Tilgner K, Saretzki G et al. Human induced pluripotent stem cell lines show stress defense mechanisms and mitochondrial regulation similar to those of human embryonic stem cells. Stem Cells. 2010;28(4):661-73.
3 Ohi Y, Qin H, Hong C et al. Incomplete DNA methylation underlies a transcriptional memory of somatic cells in human iPS cells. Nat Cell Biol. 2011;13(5):541-9.
4 Marchetto MC, Yeo GW, Kainohana O et al. Transcriptional signature and memory retention of human-induced pluripotent stem cells. PLoS One. 2009;4(9):e7076.
5 Ji H, Ehrlich LI, Seita J et al. Comprehensive methylome map of lineage commitment from haematopoietic progenitors. Nature. 2010;467(7313):338-42.
6 Kim K, Doi A, Wen B et al. Epigenetic memory in induced pluripotent stem cells. Nature. 2010;467(7313):285-90.
7 Polo JM, Liu S, Eugenia M et al. Cell type of origin influences the molecular and functional properties of mouse induced pluripotent stem cells. Nature Biotechnology 2010;28:848–855.
8 Chin MH, Mason MJ, Xie W et al. Induced pluripotent stem cells and embryonic stem cells are distinguished by gene expression signatures. Cell Stem Cell 2009;5:111–123.