SEARCH

SEARCH BY CITATION

Keywords:

  • Mouse embryonic stem cells;
  • Mouse epiblast stem cells;
  • Mouse induced pluripotent stem cells;
  • Epigenetics;
  • Imprinting

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Mouse epiblast stem cells (EpiSCs) derived from postimplantation embryos are developmentally and functionally different from embryonic stem cells (ESCs) generated from blastocysts. EpiSCs require Activin A and FGF2 signaling for self-renewal, similar to human ESCs (hESCs), while mouse ESCs require LIF and BMP4. Unlike ESCs, EpiSCs have undergone X-inactivation, similar to the tendency of hESCs. The shared self-renewal and X-inactivation properties of EpiSCs and hESCs suggest that they have an epigenetic state distinct from ESCs. This hypothesis predicts that EpiSCs would have monoallelic expression of most imprinted genes, like that observed in hESCs. Here, we confirm this prediction. By contrast, we find that mouse induced pluripotent stem cells (iPSCs) tend to lose imprinting similar to mouse ESCs. These findings reveal that iPSCs have an epigenetic status associated with their pluripotent state rather than their developmental origin. Our results also reinforce the view that hESCs and EpiSCs are in vitro counterparts, sharing an epigenetic status distinct from ESCs and iPSCs. STEM CELLS 2012; 30:161–168.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

ICM-derived mouse embryonic stem cells (ESCs) are maintained by LIF/signal transducer and activator of transcription (STAT)-3 signaling, can be dissociated into single cells with high replating efficiency, and contribute extensively to chimeras, including their germline [1]. By contrast, mouse epiblast stem cells (EpiSCs) derived from the postimplantation epiblast of embryos rely on Activin and basic fibroblast growth factor signaling for self-renewal, are refractory to single cell passaging, and contribute poorly or not at all to chimeras, although they can give rise to all three primary germ layers in teratomas [2, 3]. Both ESCs and EpiSCs express key pluripotency factors, Oct4, Sox2, and Nanog, and can differentiate into ectoderm, endoderm, and mesoderm lineages in vitro. Thus, ESCs and EpiSC each fulfil certain criteria of pluripotency, but questions remain about what mechanisms are responsible for their distinct properties.

Recently, a minor EpiSC subpopulation characterized by its expression of an Oct4 promoter-driven transgene was shown to contribute to chimeras (including their germline), suggesting that an underlying property related to gene regulation distinguishes between pluripotency of EpiSCs and ESCs [4]. Mouse ESCs and EpiSCs are derived from successive developmental stages that have distinct potency and epigenetic properties [5], raising the question whether there is a special epigenetic state corresponding to each pluripotent state. Genomic imprinting is controlled by epigenetic mechanisms that can distinguish parental alleles and result in the expression of genes in a parent-of-origin specific manner, making it a useful measure of epigenetic status, particularly where epigenetic alterations permit expression of normally repressed alleles. Moreover, reprogramming of differentiated cells to a pluripotent state can erase their epigenetic memory [6] and reset their epigenetic state to that of pluripotency.

Extensive epigenetic modification takes place in mouse embryos before and immediately after the blastocyst stage [7, 8]. This appears to confer an unstable epigenetic state on ESCs, in which imprinted genes can become biallelically expressed even after relatively short culture periods [9]. In contrast, EpiSCs are derived from the late epiblast layer of the postimplantation embryo, when the process of epigenetic modification is largely completed and random X-chromosome inactivation has occurred in females [10]. EpiSCs share several features with human ESCs (hESCs), including growth factor requirements, colony morphology, growth rate, and X-chromosome inactivation, and on this basis EpiSCs are hypothesized to be orthologs of hESCs. As imprinting status in hESCs is generally stable [11–13], a similar stability can be predicted for mouse EpiSCs. As mouse induced pluripotent stem cells (iPSCs) are derived from fetal fibroblasts or later cells with a mature, stable epigenetic state but undergo epigenetic alterations during reprogramming [14] it is less clear what to expect of their genomic imprinting.

To examine imprinting status in mouse EpiSCs, ESCs, and iPSCs, we measured allele-specific expression of nine imprinted genes and assayed the overall extent of DNA methylation for six differentially methylated regions (DMRs) in 24 cell lines (Supporting Information Tables S1 and S2). Genes regulated either by maternal or paternal differential methylation were included. The reciprocally imprinted expression of Igf2 and H19 is controlled by H19 DMR. The maternally methylated KvDMR regulates the expression of Kcnq1ot1, Cdkn1c, and Cd81. Three paternally expressed imprinted genes, Snrpn, Peg3, and Mcts2 are controlled by Snrpn DMR1, Peg3 DMR, and Mcts2 DMR accordingly. The intergenic germline IG-DMR regulates maternal expression of Gtl2. Early and later passage numbers of both male and female lines were assessed. To evaluate the effects of genetic background and passage history on the stability of genomic imprinting, all three types of stem cell lines were derived anew from reciprocal crosses of two mouse strains.

We found that mouse EpiSCs generally had monoallelic expression of imprinted genes, while ESCs and iPSCs tend to lose their allele-specific expression. Female ESC and iPSC lines were more prone to lose their imprinted expression than male lines at specific imprinted loci, while the stability of imprinting in female EpiSCs was similar to male EpiSCs. Consequently, mouse EpiSCs seemingly resemble hESCs, whereas mouse iPSCs closely resemble ESCs in their epigenetic properties, thereby distinguishing two epigenetic states that correspond to the two distinct types of pluripotent stem cells.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Culture of ESCs, EpiSCs, and iPSCs

F1 fetuses, mouse embryonic fibroblasts (MEFs), EpiSCs, ESCs, and iPSCs were generated from reciprocal C57BL/6J female × Mus musculus castaneus male (bc) and Mus musculus castaneus female × C57BL/6J male (cb) crosses. Mouse blastocysts (E3.5) and pregastrula epiblast layers (E6.5) were isolated as described [2], except that MEF-conditioned medium was used for derivation of EpiSCs. ESCs were generated as previously described [15]. iPSCs were derived as described by others [14, 16] using vsv-pseudotyped retroviruses individually expressing human KLF4, OCT4, SOX2, and cMYC. Standard ESC medium for derivation of ESCs and iPSCs consisted of Dulbecco's modified Eagle's medium plus 20% Knock Out (KO) Serum Replacer (Life Technologies, Grand Island, NY, USA, www.invitrogen.com) and 10 ng/ml LIF. ESCs and iPSCs formed typical rounded colonies on feeders. For depletion of feeder cells before DNA and RNA isolation, ESCs and iPSCs were cultured on gelatin-coated dishes in N2B27 medium plus LIF (10 ng/ml) and BMP4 (10 ng/ml) as described [17]. For feeder- and serum-free culture, EpiSCs were grown in chemically defined medium (CDM), supplemented with Activin A (20 ng/ml) and FGF2 (12 ng/ml). Eight cell lines for each type of pluripotent stem cells were analyzed, including both male and female lines. Cells were analyzed at early (p9–p11) and at later (p18–p20) passages. The care of animals was in accordance with institutional guidelines.

ESCs were converted to EpiSC-like cells (cEpiSCs) by culturing on gelatin-coated dishes in feeder-free conditions for 2 days in standard ESC medium containing LIF at 10 ng/ml. Subsequently, medium was changed to CDM containing Activin A (20 ng/ml) plus fibroblast growth factor-2 (FGF) (12 ng/ml). Cultures were passaged as small clumps using collagenase-based dissociation methods and grown in standard mEpiSC medium. Loss of expression of Rex1 by polymerase chain reaction (PCR) methods, loss of alkaline phosphatase staining, and stable growth pattern as flat, spread out colonies confirmed their cEpiSC status. EpiSCs were reverted to ESC-like cells (rESCs) by culture in standard ESC medium containing 10 ng/ml LIF on feeder-coated dishes. Cells reverted to typically rounded colonies on feeders and were passaged using trypsin dissociation methods. They expressed mES pluripotent markers, alkaline phosphatase, and regained Rex1 expression. For depletion of feeder cells before analysis, rESC lines were cultured for two passages on gelatin-coated dishes in N2B27 medium containing LIF at 10 ng/ml and BMP4 at 10 ng/ml.

Allelic Expression Analysis

The cDNA samples were subjected to PCR amplification in a 30 μl reaction volume containing 200 nM forward and reverse primers, 0.625 Units of Taq polymerase, 0.1 mM each dNTP, 1× PCR buffer, 5 μl cDNA template, 1.5 mM MgCl2. PCR cycle conditions consisted of an initial denaturation at 94°C for 2 minutes, followed by 40 cycles of 94°C for 30 seconds, specific annealing temperature for 30 seconds, and extension at 72°C for 30 seconds, followed by a final extension at 72°C for 5 minutes. Pyrosequencing was performed on a PSQ HS96 system using enzymes and substrates from the PyroGold Q96 reagent kit (Qiagen, Crawley, United Kingdom; www.qiagen.com).

Bisulfite Conversion and Pyrosequencing

DNA (1 μg) was treated using the EZ-96 DNA methylation Kit (Zymo Research, Irvine California, USA; www.zymoresearch.com) in accordance with the manufacturer's instruction. Bisulfite-treated DNA was eluted in 30 μl of elution buffer. Amplicons were generated in a 25 μl reaction volume containing 100 nM forward and reverse primers, 1.25 Units of HotstarTaq DNA Polymerase (Qiagen), 0.2 mM dNTPs, and 5 μl of bisulfite-treated DNA. PCR cycle conditions consisted of an initial activation step of 95°C for 15 minutes, followed by 50 cycles of 94°C for 30 seconds, specific annealing temperature for 30 seconds, and extension at 72°C for 30 seconds, followed by a final extension at 72°C for 10 minutes. Pyrosequencing was carried on PSQ HS96 System using PyroGold Q96 SQA Reagents (Qiagen). The degree of methylation at CpG sites (without distinguishing between maternal and paternal alleles) was determined by pyro-Q CpG software.

Statistics

Because of the small sample sizes and the lack of normal distributions, nonparametric statistics were used for analysis of significant differences between fetal tissue controls and pluripotent stem cell samples. Kruskal–Wallis test was used to compare groups and Dunn's post hoc test was used to make pairwise comparisons. Bonferroni correction was used to adjust the significance level for multiple comparisons. The significance level, based on six pairwise comparisons, was set at 0.0083. Association between DNA methylation and loss of imprinting was examined by Spearman's rank correlation test and expressed as Spearman r, with significance level p = .05. All statistic analyses were performed using Prism 5.01 (GraphPad Software, La Jolla, CA, USA, www.graphpad.com).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Establishment and Characterization of Three Types of Mouse Pluripotent Stem Cells

We cultured mouse ESCs, EpiSCs, and iPSCs to early (p9–p11) and later (p18–p20) passages under chemically defined conditions. EpiSCs expressed the pluripotent markers Oct4, Sox2, and Nanog but not the ICM and ESC marker, Rex1. EpiSCs also differed from ESCs and iPSCs by upregulation of the late epiblast marker, Fgf5 (Fig. 1A). Furthermore, EpiSCs formed flattened colonies whereas ESCs and iPSCs shared a characteristic dome-shaped morphology (Fig. 1B). Expression of Oct4, Nanog, and SSEA1 was maintained in ESCs, EpiSCs, and iPSCs at later passages (Fig. 1C), suggesting that all three stem cell types remained pluripotent. Immunofluorescence revealed a characteristic H3K27me3 signal associated with the inactive X chromosome in female EpiSCs but not in female ESCs or iPSCs (Fig. 1D). Consistent with other reports [2, 3, 18], EpiSCs possessed key features of pluripotency but nevertheless differed from ESCs and iPSCs, confirming that stem cells derived from early postimplantation mouse embryos are readily distinguishable, both molecularly and epigenetically, from those derived from preimplantation embryos and reprogrammed fibroblasts.

thumbnail image

Figure 1. Characterization of pluripotent stem cells. (A): Q-PCR analyses were performed on two independent cell lines of each cell type. Relative expression of each gene was normalized to Pbgd. Data represent mean of three independent experiments and error bars indicate SEM. (B): Distinct morphology of colonies in ESCs, EpiSCs, and iPSCs. Scale bar = 100 μm. (C): Oct4, Nanog, and SSEA1 expression of colonies was analyzed at passage 20 by immunofluorescence. Scale bar = 100 μm. Nuclei are shown by Hoechst staining. (D): Immunofluorescence revealed a prominent intranuclear signal for H3K27me3 in the female EpiSCs, which was absent from female ESCs and iPSCs indicating their X-active state. Abbreviations: EpiSCs, epiblast stem cells; ESCs, embryonic stem cells; iPSCs, induced pluripotent stem cells; PCR, polymerase chain reaction; SSEA1, stage-specific embryonic antigen-1.

Download figure to PowerPoint

Assessment of Imprinted Gene Expression Patterns in Pluripotent Stem Cells

To assess the effect of genetic background on the stability of imprinting [19, 20], we first analyzed imprinted gene expression in E15 fetuses and E13.5 primary MEFs using pyrosequencing to determine the parental contribution to transcripts from F1 hybrids obtained by mating C57BL6/J females × Mus musculus castaneus males (bc cross) and reciprocally by mating Mus musculus castaneus females × C57BL6/J males (cb cross). Seven genes showed strict paternal (Snrpn, Peg3, Igf2, Kcnq1ot1) or maternal (Gtl2, H19, Cdkn1c) allele-specific expression in fetal and fibroblast samples (Supporting Information Fig. S1). Mcts2 was expressed predominantly from the paternal allele [21]. The placental-specific imprinted gene Cd81 was, as expected, biallelically expressed in fetuses and in MEFs. Thus, genetic background did not affect the stability of genomic imprinting in the interspecies hybrid fetuses and primary fibroblasts. Accordingly, data from the two crosses have been combined in Figures 2, 3 and Supporting Information Figures S2 and S3. Next, we determined the absolute expression levels of each imprinted gene in pluripotent stem cells and control fetuses using Q-PCR. Except for H19, Igf2, and Cdkn1c, imprinted genes showed expression levels comparable to or above E15 fetal controls, reflecting robust expression levels in pluripotent stem cells (Supporting Information Fig. S2).

thumbnail image

Figure 2. Status of imprinting in pluripotent stem cells. Pyrosequencing was used to analyze allele-specific expression of genes in ESCs, EpiSCs, and iPSCs at early passages (combined data for both bc and cb crosses). Fetuses (E15) were used as controls. Polymorphisms in the genomic DNA were identified, and cDNA of each gene was pyrosequenced to examine allelic expression. The percentage of total gene expression contributed by the repressed allele (expected on the basis of each gene's typical pattern of monoallelic maternal or paternal expression) was plotted on the y-axis. Each cell line is shown as a dot; red dots represent female cells and black dots represent male cells. Bar indicates median and dashed lines indicate the maximum value in controls except for Mcts2 and Cd81, where the dashed lines include the entire range of values in controls. Asterisk means significantly different from controls, p < .0083. Abbreviations: EpiSCs, epiblast stem cells; ESCs, embryonic stem cells; iPSCs, induced pluripotent stem cells.

Download figure to PowerPoint

thumbnail image

Figure 3. Methylation levels of DMRs in pluripotent stem cells. Methylation levels of DMRs were examined by pyrosequencing after bisulfite conversion. ESCs, EpiSCs, and iPSCs at early passages were analyzed (combined data for both bc and cb crosses). Fetuses (E15) were used as controls. Each cell line is shown as a dot; red dots represent female cells and black dots represent male cells. Bar indicates median and dashed lines indicate the entire range of values in controls. Asterisk means significantly different from controls, p < .0083. Abbreviations: CpG, cytosine-phosphate-guanine; DMR, differentially methylated region; EpiSCs, epiblast stem cells; ESCs, embryonic stem cells; iPSCs, induced pluripotent stem cells.

Download figure to PowerPoint

Seven out of nine imprinted genes studied here (Snrpn, Gtl2, Peg3, Mcts2, H19, Kcnq1ot1, and Cdkn1c) showed predominantly monoallelic expression in EpiSCs (Fig. 2A–2G). Exceptionally, expression of Igf2 was biallelic in all EpiSC lines examined (Fig. 2H). The placental-specific imprinted gene Cd81 also showed biallelic expression in all EpiSC samples (Fig. 2I), but this may reflect the norm at this developmental stage [22]. These findings indicate that other than Igf2, imprinted expression in EpiSCs was maintained, similar to the behavior of these genes in fetuses and fibroblasts.

In contrast to the generally monoallelic expression pattern observed in EpiSCs, most ESC and iPSC lines showed loss of imprinting of four imprinted genes (Peg3, H19, Kcnq1ot1, and Cdkn1c) (Fig. 2C, 2E–2G). Interestingly, the normally repressed maternal Igf2 allele was also active in all ESC and iPSC lines (Fig. 2H). Several lines of ESCs and iPSCs exhibited monoallelic expression of Snrpn and Gtl2 genes (Fig. 2A, 2B), but the interline variation of their derepression in ESCs and iPSCs was higher than that observed in EpiSC lines. In addition, Mcts2 and Cd81 showed biallelic expression in most ESC and iPSC lines (Fig. 2D, 2I), reflecting the normal status of imprinting of these genes in the cell types from which they arose. Overall, expression of imprinted genes showed a strong tendency to become biallelic in both ESCs and iPSCs, contrasting with the generally monoallelic pattern of EpiSCs and suggesting that a distinct epigenetic state exists in ESCs and iPSCs when compared with EpiSCs.

Determination of DNA Methylation Levels at DMRs for Imprinted Genes

To understand the potential mechanisms responsible for genomic imprinting in these two epigenetically distinguishable states (ESC/iPSC vs. EpiSC), we used pyrosequencing to investigate methylation of known DMRs regulating the genes studied here. DMRs of control fetal DNA showed approximately 50% methylation, as expected for normal parent-specific methylation properties (Fig. 3A–3F). Most EpiSC lines showed differential methylation of Snrpn DMR1, IG-DMR, and Peg3 DMR, correlating with their monoallelic expression of the Snrpn, Gtl2, and Peg3 genes, respectively (Fig. 3A, 3B, 3C, Supporting Information Fig. S3A, S3B, S3C). EpiSCs showed hypermethylation at Mcts2 DMR and KvDMR (Fig. 3D, 3F); however, this was not accompanied by changes in the corresponding Mcts2, Kcnq1ot1, or Cdkn1c genes, which were expressed similarly to controls (Supporting Information Fig. S3D, S3G, S3H). EpiSCs also showed hypermethylation of the H19 DMR (Fig. 3E) and had increased expression of Igf2, which was biallelic, but had no change in H19 expression, which remained monoalleleic (Supporting Information Figs. S2, S3E, S3F). Hypermethylation of the H19 DMR would be expected to cause biallelic expression of Igf2 and biallelic silencing of H19 [23]. The low overall levels of H19 expression in EpiSCs (when compared with controls) could indeed reflect hypermethylation at the H19 DMR, but we also could not exclude biallelic silencing of H19 (or of Kcnq1ot1 and Cdkn1c), since pyrosequencing only detects expressed alleles. Some discordance between DMR methylation and allelic expression also occurs in hESCs [13]. Regardless of its consequences, the hypermethylation observed for the Mcts2 DMR, KvDMR, and H19 DMR was a distinguishing feature of EpiSCs, especially when compared with ESCs and iPSCs.

The analyses of DMR methylation in ESCs and iPSCs revealed a variety of levels (Fig. 3A–3F). Hypomethylation was seen for the Snrpn DMR1, Peg3 DMR, Mcts2 DMR, and H19 DMR in ESCs and iPSCs, and this correlated with derepression of the respective genes, particularly for Snrpn and Peg3 in the female lines (Fig. 3A, 3C, Supporting Information Fig. S3A, S3C, S3D, S3E). While the normally silent allele of Gtl2 and Igf2 was derepressed in ESCs and iPSCs, this did not correlate with IG-DMR and H19 DMR methylation levels, respectively (Supporting Information Fig. S3B, S3F). Hypomethylation of the KvDMR correlated with activation of the repressed allele of Kcnq1ot1 and Cdkn1c in ESCs (Supporting Information Fig. S3G, S3H) and with Kcnq1ot1 derepression in iPSCs, but it did not correlate with derepression of Cdkn1c in iPSCs (Supporting Information Fig. S3G, S3H). These results may reflect the relatively low expression levels of Cdkn1c and Kcnq1ot1 in iPSCs, as suggested for EpiSCs (above). In summary, these data indicate that vulnerability to loss of DMR methylation is markedly greater in ESCs and iPSCs than in EpiSCs, emphasizing the existence of different epigenetic states associated with the two distinct pluripotent states.

To gain further insight into the relationship between DMR methylation and imprinted gene expression, we subjected an EpiSC line (EP2BC, male) to ESC culture conditions to select for reversion to an ESC-like (rESC) state [24] and subjected an ESC line (ES9CB, male) to EpiSC conditions to convert them to an EpiSC-like (cEpiSC) state [18]. Data from the rESC and cEpiSC lines showed relatively modest changes in allele-specific transcript levels for most imprinted genes (Supporting Information Fig. S6). Reversion of EpiSCs to an ESC-like state modestly decreased methylation of H19 DMR and KvDMR but did not substantially alter Snrpn DMR1, IG-DMR, or Peg3 DMR methylation. However, reversion to an ESC-like state substantially decreased methylation of the Mcts2 DMR. Conversion from ESC to EpiSC-like cells increased methylation levels of some DMRs (H19 DMR, KvDMR) but did not substantially alter others (Snrpn DMR, IG-DMR, Peg3 DMR, and Mcts2 DMR). However, alterations in methylation levels were often accompanied by changes in imprinted gene expression: silent allele expression either correlated inversely with methylation (Mcts2, H19, Kcnq1ot1, and Cdkn1c in rESCs) or correlated directly with methylation (Igf2 in rESCs and cEpiSCs). These observations reinforced the findings with the unmodified ESC and EpiSC stem cell lines, together indicating a relationship between DMR methylation and imprinted gene expression in both pluripotent states, with the possible exception of genes showing low transcript levels that could reflect biallelic silencing.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Our results demonstrate the contrasting status of genomic imprinting in different types of pluripotent stem cells represented by ESCs/iPSCs and by EpiSCs (Supporting Information Figs. S4, S5); namely, EpiSCs showed higher stability of imprinted gene expression and methylation than either ESCs or iPSCs. Accordingly, EpiSCs resemble hESCs in having relatively higher stability of genomic imprinting than either mouse ESCs or iPSCs [11–13]. However, previous studies of epigenetic status in hESCs have differed in their interpretation on the status of imprinting in hESCs. One perspective is that the stability of genomic imprinting is gene specific in human ESCs. For instance, 11 out of 22 imprinted genes showed interline variability of allelic expression [13], and similar variability was observed in another study [25]. While the variable imprinting status of several genes reported in human ESC lines may due to variations between lines, it could also reflect the selection of genes analyzed in those previous studies. For example, four genes selected for analysis [13], TP73, IGF2R, WT1, and SLC22A18, showed biallelic expression in all informative human ESC lines. However, in humans those genes are not imprinted at all or show polymorphic imprinting. Similarly, three genes that showed biallelic expression in several hESC lines (MEST isoform 2, ATP10A, and PHLDA2) have tissue-specific and polymorphic patterns of allele-specific expression [26–29]. Finally, polymorphic loss of imprinting of IGF2 was observed in a Japanese population [29] where loss of imprinting of IGF2 is estimated to occur in 5%–10% of the population. Thus, interpretation of imprinting status should be associated with their imprinting status in vivo which we have sought to do in this study.

Our findings raise the question of whether the relative stability of EpiSC imprinting lies in their postimplantation origin or has another cause. In addition to the differing developmental origins of the two pluripotent stem cell types, their culture conditions could affect the vulnerability to imprinting instabilities as previously observed for preimplantation embryos [20]. We used CDM containing Activin A and FGF2 to culture EpiSCs and medium containing N2B27 plus LIF and BMP4 to grow ESCs and iPSCs. While these differences between media represent marked variations in culture conditions, the two pluripotent states cannot be maintained otherwise in chemically defined conditions, thus it is impossible to eliminate these variables. Interestingly, the rESCs and cEpiSCs did not closely match the DMR methylation and imprinted expression patterns in ESCs and EpiSCs, respectively. This may suggest that the specific imprinting status seen in rESCs and cEpiSCs is determined less by the changes in their culture conditions than by their initial pluripotent state (either EpiSCs or ESCs, respectively). Moreover, when different culture media were used to grow hESCs, the allelic expression of imprinted genes was generally retained [30], suggesting that the stability of genomic imprinting was inherent in the type of pluripotent cells rather than attributable to culture conditions. Notably, most alterations in allelic expression or DMR methylation patterns of imprinted genes took place at early passages and did not change progressively during further passages (Supporting Information Figs. S4, S5), suggesting that the majority of epigenetic changes, once established, were stably maintained in culture as seen with hESCs [31]. Accordingly, the primary determinant of imprinted gene expression appeared to be the type of pluripotency achieved during the derivation process itself.

We found that the genetic sex of the cells also played an important role in the stability of imprinting of Snrpn and Peg3 in both ESCs and iPSCs, accounting for much of the variation between lines in each case. Importantly, female ESC and iPSC lines had a higher tendency for loss-of-imprinting than male cells and had lower levels of methylation for most DMRs assayed in ESCs. This confirms the lower global and DMR methylation previously reported for female ESCs [32] and also seen for female mouse primordial germ cells [33]. EpiSCs by contrast did not differ in loss-of-imprinting and DMR methylation between female and male cells. The derepression of silent alleles in female ESCs and iPSCs, which have two active X-chromosomes, contrasts with the maintenance of imprinting in female EpiSCs, which have only one. This appears to support the hypothesis of an X-linked dosage-dependent repressor of the methyltransferase family that inhibits de novo DNA methylation, thereby resulting in reduced methylation levels in female cells [32]. However, the identity of such repressor(s) remains to be determined and other sex-chromosome-specific mechanisms cannot be ruled out.

In addition, we found that imprint stability varied between different genes, reflecting diverse mechanisms of imprinting for the genes selected for analysis. However, this does not appear to reflect the differing germline origins of the methylation, as both maternal- and paternal-derived methylation imprints were affected. The biallelic expression of Igf2 in all pluripotent cell types may reinforce the hypothesis of selective growth advantage of Igf2 expression in self-renewing cultured stem cells [30] or may reflect the normal imprinting patterns seen for this gene in early stages of development [34]. The methylation pattern of IG-DMR and the absolute expression levels of Gtl2 in the iPSCs studied here were comparable with those in ESCs and were thus similar to the “good” iPSCs clones that contributed to high-grade chimeras and gave all-iPSC mice [35]. Importantly, two genes, Cd81 and Mcts2, which are expressed monoallelically only in certain fetal and postnatal tissues, were biallelically expressed in ESCs and iPSCs, suggesting that their normal expression pattern is maintained during derivation and culture.

In summary, our results demonstrate marked epigenetic differences between ESCs/iPSCs and EpiSCs, which are reflected by differing patterns of allelic expression and DMR methylation. The close link between the epigenetic state and the pluripotent state indicates that different epigenetic states are associated with certain properties of these distinct pluripotent states [5, 36]. The mechanism(s) underlying this correlation remains unclear. The different phenotypes represented by ESC/iPSC and EpiSC pluripotent states may reflect the interaction between transcriptional activities of key pluripotency genes and epigenetic states. The striking similarity of imprint status in EpiSCs to that of hESCs suggests that an orthologous pluripotent state is conserved between mice and humans despite considerable divergence in the composition and placement of regulatory elements for many of the core pluripotency network genes they express [37].

These observations also underscore the issue of epigenetic memory, which has major implications for developmental and therapeutic normality of iPSCs. Recent studies show that mouse iPSCs retain epigenetic signatures of their cell type of origin, with nonfibroblast cell types more likely to yield iPSCs in which Gtl2 is repressed [35], and iPSCs in general likely to retain epigenetic signatures favoring differentiation along their originating lineage [6]. By contrast, there was little tendency for such epigenetic memory in mouse ESCs derived from somatic cell nuclear transfer (SCNT) [38, 39]. However, mouse EpiSCs derived from SCNT showed perturbed expression of some imprinted genes, when compared with EpiSCs from fertilized embryos [40]. Together with these findings our observations lead to the hypothesis that the epigenetic state achieved by reprogramming to human iPSCs is likely to resemble that of hESCs and EpiSCs, rather than mouse ESCs and iPSCs. The consequences of the epigenetic state resulting from reprogramming to iPSCs will have to be taken into account in any human translational applications.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We found that EpiSCs have generally monoallelic expression of imprinted genes, in contrast to ESCs and iPSCs, which tend to express the same genes biallelically. Loss of imprinting in iPSCs reflects their gain of an ESC-like epigenetic state. The maintenance of imprinting in EpiSCs reveals a similarity in their epigenetic state to hESCs, which also show generally stable imprinting and a tendency for X-chromosome inactivation in female lines. Intriguingly, we found that female cells were more likely to lose imprinting in ESCs and iPSCs but not in EpiSCs. Overall, our results suggest that different epigenetic states are associated with each of the two distinct classes of pluripotent mammalian stem cells.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We thank Harwell MRC Mammalian Genetics Unit for Mus musculus castaneus mice; M. Alexander, K. Woodfine, and D. Gray for their support in the cell culture and mouse genetic aspects of this study; and M. Charalambous for her invaluable discussions and advice. This work was supported by MRC studentship (B.S.), U.S. National Institutes of Health/National Institute of Child Health and Human Development, EU FP7 Grant HEALTH-F4-2009-223485/PluriSys (R.A.P.), MRC program grants (R.A.P., and A.F.S.), Wellcome Trust grant (A.F.S.), MRC Senior Research Fellowship (L.V.), CRUK Senior Cancer Research Fellowship and AICR grant (A.M.), and by Cambridge University Hospitals National Institute for Health Research Biomedical Research Centre and Evelyn Trust funding (R.A.P. and L.V.). B.S. is currently affiliated with Shanghai Jiao Tong University School of Medicine, Shanghai, China.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
STEM_793_sm_suppldata.pdf692KSupplementary Figures and Tables

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.