Loss of Genomic Imprinting in Mouse Parthenogenetic Embryonic Stem Cells

Authors

  • Takuro Horii,

    1. Laboratory of Genome Science, Biosignal Genome Resource Center, Institute for Molecular and Cellular Regulation, Gunma University, Gunma, Japan
    Search for more papers by this author
  • Mika Kimura,

    1. Laboratory of Genome Science, Biosignal Genome Resource Center, Institute for Molecular and Cellular Regulation, Gunma University, Gunma, Japan
    2. PRESTO, Japan Science and Technology Corporation, Saitama, Japan
    Search for more papers by this author
  • Sumiyo Morita,

    1. Laboratory of Genome Science, Biosignal Genome Resource Center, Institute for Molecular and Cellular Regulation, Gunma University, Gunma, Japan
    2. PRESTO, Japan Science and Technology Corporation, Saitama, Japan
    Search for more papers by this author
  • Yasumitsu Nagao,

    1. Educational Development, College of Arts and Sciences, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Izuho Hatada Ph.D.

    Corresponding author
    1. Laboratory of Genome Science, Biosignal Genome Resource Center, Institute for Molecular and Cellular Regulation, Gunma University, Gunma, Japan
    2. PRESTO, Japan Science and Technology Corporation, Saitama, Japan
    • Biosignal Genome Resource Center, Laboratory of Genome Science, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-22 Showa-machi, Maebashi, Gunma 371-8511, Japan. Telephone: 81-27-220-8057; Fax: 81-27-220-8110
    Search for more papers by this author

Abstract

In mammals, complementary contributions of both the maternal and the paternal genomes are required for normal development because of the parental-allele-specific modification of the genome, called genomic imprinting. Therefore, parthenogenetic embryos (PG) with two maternal genomes cannot develop to term, and PG chimeras show a restricted cell contribution of donor cells and reduced weight, although they can develop to term. On the other hand, parthenogenetic embryonic stem cells (PGES) chimeras are more normal in their tissue contribution of donor cells and body weight compared with PG chimeras. To elucidate the epigenetic mechanisms underlying this, we analyzed the imprint status in donor cells of PGES and PG chimeras. In somatic lineages, genomic imprinting was lost in some PGES chimeras, whereas those in PG chimeras were almost totally maintained. Moreover, loss of imprints correlated to the gene expression pattern of imprinted genes. Therefore, this loss of imprinting in PGES chimeras could improve the tissue contribution and body weight to a normal level. On the other hand, in germ lineages, both PGES and PG in chimeras showed normal erasure of imprints, indicating that the reprogramming in germ lineages is an inevitable event, regardless of the imprint status of primordial germ cells.

Disclosure of potential conflicts of interest is found at the end of this article.

Introduction

Mammalian parthenotes cannot develop normally to term. Mouse parthenogenetic embryos die by day 10 of gestation [1, 2]. Most notably, they fail in the trophectoderm and primitive endoderm, which results in failure of the extraembryonic tissues in the whole parthenogenetic conceptus [3]. Alternatively, viable parthenogenetic chimeras can be produced by normal host embryo rescue, and parthenogenetic cells can give rise to a functional germline [4, 5]. In the somatic lineages of chimeras, parthenogenetic cells are initially allocated randomly in the embryo proper [6, [7]–8], but this is followed by a progressive elimination of parthenogenetic cells, most notably between days 13 and 15 of gestation [9]. This can be explained by parent-specific epigenetic modification of the genome, which leads to the altered expressions of imprinted genes in parthenogenetic cells. Gene expressions of imprinted genes are greatly dependent on the methylation status in differentially methylated regions (DMRs) of imprinted genes. In parthenogenetic embryos with the two maternal genomes, the paternally expressed genes Peg1/Mest [10], Peg3 [11], Snrpn [12], and Igf2 [13] are silenced, whereas the maternally expressed genes Igf2r [14], p57kip2 [15], and H19 [16, 17] are expressed at twice the normal amount. It has been thought that the imprint pattern of parthenogenetic cells is maintained in parthenogenetic embryonic stem cells (PGES) chimeras, as well as in parthenogenetic embryo (PG) chimeras. Since the PGES closely resemble PG in their developmental potential, they have been used to study aspects of cell differentiation and commitment related to imprinting. In contrast, the classic phenotype of growth retardation normally observed in PG chimeras was not seen in PGES chimeras [18]. In addition, higher overall levels of PGES are detected than PG in terms of tissue contribution. These imply that the aberrant acquisition or loss of imprints occurs in PGES. In fact, Igf2r, one of the maternally methylated imprinted genes, was shown to be demethylated in PGES, and methylation was not restored after differentiation in vitro [19]. Therefore, we can expect that the phenotypic difference between PGES and PG could be caused by modified expressions of imprinted genes due to alteration of imprints in PGES. Thus, it seems very interesting to study the genomic imprinting of PGES derivatives in somatic lineages.

On the other hand, in germ lineages of normal female mice, erasure of genomic imprinting occurs around the time that primordial germ cells (PGCs) enter the gonad (embryonic day [E]10.5–E13.5), and the establishment of new imprints occurs in the postnatal growth phase of oogenesis [20, [21]–22]. The erasure process of paternal imprints in PGCs is necessary to reset and restore the new imprints in normal female germ line cells. In contrast, the erasure of parental imprints seems to be unnecessary in parthenogenetic female germ cells, because parthenogenetic cells have the same imprint pattern as mature oocytes; namely, both alleles are of maternal origin. A previous study has shown that normal offspring were born from PG- and PGES-derived germ cells [5, 18], implying that mature oocytes derived from parthenogenetic cells finally establish normal imprints. However, it has not been elucidated whether erasure of imprints occurs or old imprints continue through germ lineages. If the imprints of parthenogenetic cells are erased in germ lineages, we can conclude that the erasure of imprints is an inevitable event, even in parthenogenetic cells. Alternatively, if imprints are not erased, it means that the erasure of imprints is not necessary for parthenogenetic cells in germ lineages.

In the study of genomic imprinting in chimeras, the purity of parthenogenetic cells is important, because parthenogenetic derivatives and host embryo derivatives are mixed in tissues and organs. Selection by a drug, such as puromycin or neomycin, is a convenient method for isolating the drug-resistant parthenogenetic cells [18, 23], although it may be difficult to apply to the isolation of special cells, such as PGCs, oocytes, and sperm. However, green fluorescent protein (GFP) makes it possible to detect and isolate parthenogenetic derivatives under fluorescence microscopy without any staining procedures or drug selection [24].

In this study, we show the difference in epigenetic status in somatic lineages between PG and PGES chimeras, explaining why phenotypic differences between these chimeras occur. We also address whether the reprogramming of genomic imprinting is necessary for these parthenogenetic cells in germ lineages.

Materials and Methods

Generation of PG Chimeras

Diploid parthenogenetic embryos were produced as follows. C57BL/6-Tg (ACTbEGFP) 1Osb/J (B6-GFP; a kind gift from Dr. M. Okabe) females were superovulated by the injection of 5 units of pregnant mares' serum (Sankyo, Tokyo, http://www.daiichisankyo.co.jp) and 44–48 hours later by 5 units of human chorionic gonadotrophin (hCG; Sankyo). Oocytes at the metaphase stage of the second meiotic division (MII) were collected from the oviducts 18 hours after hCG, and then cumulus cells were removed by digestion with hyaluronidase (300 units/ml) in M2 medium. Artificial activation was performed by brief exposure to 10 mM SrCl2 and 5 μg/ml cytochalasin B in Ca2+-free M16 embryo culture medium for 6 hours. After activation, the embryos were cultured in M16 medium supplemented with 0.1 mM EDTA/2Na. PG chimeras were made by aggregating the 4–8 cell stage of B6-GFP parthenogenetic embryos with the same stage of C57BL/6J(B6) normal host embryos. After 2 days of culture, chimeric blastocysts were transferred to the uterine horns of recipient females (CD-1) on the 3rd day of pseudopregnancy, counting the day of finding the vaginal plug as day 1.

Establishment of PGES Cell Lines and the Generation of PGES Chimeras

B6-GFP and F1-GFP (CBA/Ca × B6-EGFP) females were superovulated, and MII oocytes were activated and diploidized as described above. Parthenogenetic embryos were cultured in vitro until expanded blastocysts started to hatch in M16 medium. Hatching blastocysts were transferred onto mitomycin C (0.2 mg/ml)-treated mouse embryonic fibroblast feeder cells in gelatinized tissue culture wells (2–3 blastocysts per well of a four-well multiplate; Nunc, Rochester, NY, http://www.nuncbrand.com) and cultured for 7 days in embryonic stem (ES) medium, Dulbecco's modified Eagle's medium (DMEM) containing 17.5% Knockout SR (Gibco, Grand Island, NY, http://www.invitrogen.com) following standard procedures [25, 26]. After 7 days, inner cell mass outgrowths were harvested in trypsin/EDTA (0.25%/1 mM; Gibco), disaggregated by mouth pipetting, and plated onto feeder cells in ES medium (passage 1). Clones resembling ES cells in morphology were then picked and disaggregated a second time. They were then expanded and passaged prior to freezing or use. PGES chimeric embryos were produced by B6-GFP PGES aggregated with four- to eight-cell stages of normal B6 embryos or by injecting F1-GFP PGES into the blastocoel cavity of CD-1 blastocysts [27]. Chimeric blastocysts were transferred to the uterine horns of pseudopregnant recipient females (CD-1), as described above.

Collection of Somatic and Germ Cells from Parthenogenetic Chimeras

For somatic cell collection from chimeras, primary mouse embryonic fibroblasts (MEFs) from E13.5 chimeras after 3 days of culture and kidney cells from newborn chimeras were sorted by FACS (FACSCalibur; Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). It is known that the GFP gene promoter is not subject to gene silencing and that GFP-positive cells can contribute to all internal tissues of chimeric mice, except for erythrocytes and hair [24]. For germ cell collection, genital ridges at E11.5 and E13.5 were cut out from the parthenogenetic chimeric fetuses. The tissue fragments of individual embryos were incubated with trypsin/EDTA for 20 minutes at 37°C and then were triturated with tweezers to suspend the single cells. PGCs can be detected by the undifferentiated cell marker stage-specific embryonic antigen (SSEA-1) [28]. Cells were suspended in 200 μl of DMEM, supplemented with 2% fetal calf serum containing 10 μg of the SSEA-1 antibody (Kyowa Hakko, Tokyo, http://www.kyowa.co.jp/eng/), and incubated at 4°C for 45 minutes. After centrifugation (900 rpm for 5 minutes at 4°C), the cell pellet was resuspended in 300 μl of medium containing 1 μg of the goat anti-mouse IgM antibody (ICN Biomedicals, Inc., Costa Mesa, CA, http://www.icnbiomed.com) conjugated with B-phycoerythrin (B-PE) followed by incubation at 4°C for 20 minutes. After centrifugation, the cell pellet was resuspended in M2 medium containing 5 μg/ml cytochalasin B (M2-CB medium). Oocytes from 1-day postpartum (dpp) chimeras were dissociated as previously described [29]. These PGCs and oocytes of parthenogenetic derivatives and host embryo derivatives were collected using a micromanipulator under the fluorescence microscope in M2-CB medium.

DNA Isolation and Methylation Analysis

DNA was isolated from at least 200 cells of each sample. Bisulfite treatment was carried out using a CpGenome DNA Modification kit (InterGen, Burlington, MA, http://www.intergen.com), according to the manufacturer's instructions. Polymerase chain reaction (PCR) amplification for Peg1/Mest, Snrpn, and Igf2r was carried out on each set of isolated cells in DMRs. The amplification consisted of a total of 35–40 cycles at 95°C for 10 seconds, 60°C for 30 seconds, and 72°C for 60 seconds in a GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). The primers [30] used were as follows: Peg1/Mest (promoter and exon 1), 5′-GGTTGGGTTTGGATATTGTAAAGT-3′ and 5′-TT-CCCTATAAATATCTTCCCATATTC-3′; Snrpn (DMR1), 5′-T-TTGGTAGTTGTTTTTTGGTAGGATAT-3′ and 5′-ACTAAA-ATCCACAAACCCAACTAAC-3′; Igf2r (DMR2), 5′-GAAGTT-GTGATTTTGGTTATGTTAAG-3′ and 5′-ACAATTTACACC-CTCAAAATACCTC-3′; and H19 (DMR), 5′-GGATATATGTA-TTTTTTAGGTTGGT-3′, and 5′-AAAAAAACTCAATCAATT-ACAATCC-3′.

PCR products were subcloned into the TA cloning vector (pCR 2.1; Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Approximately 8–12 positive clones in each sample were sequenced using the Big Dye terminator method (ABI Prism 3100, Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). For combined bisulfite restriction analysis (COBRA) [31], Peg1/Mest and Igf2r products were digested with TaqI at 65°C, Snrpn products were digested with HhaI at 37°C, and H19 products were digested with HinfI at 37°C for 3 hours. DNA fragments were separated on a 2.5% agarose gel. Quantification was conducted with a Fluor-S multi-imager (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Quantity One software (Bio-Rad) was used for densitometric analysis according to the manufacturer's instructions. Images were scanned with a 12-bit gray scale, a specific format of the software. Background subtraction was performed with the global method in the Quantity One software, in which the selected area was calculated as the background. Then, the density of bands before and after digestion was compared, and the percentages of methylation were calculated.

Quantitative Real-time Reverse Transcription-PCR Analysis

Total RNA purified from MEFs was reverse-transcribed using Superscript II (Takara Bio, Shiga, Japan, http://www.takara-bio.com) and oligo(dT)12–18 primer (Takara) in a total volume of 20 μl. Quantitative real-time PCR was performed using SYBR Premix Ex Taq (Perfect Real Time; Takara). The PCR mixture consisted of 2× SYBR Premix Ex Taq, 0.2 μM forward and reverse primers, 50× carboxy-X-rhodamine reference dye, and template cDNA in a total volume of 25 μl. The cocktail was initially heated at 95°C for 10 seconds to activate it. Subsequent PCR was carried out at 95°C for 5 seconds and at 60°C for 30 seconds for 40 cycles in an ABI 7700 sequence detector. In each run, the dilution series of cDNA from normal B6 MEFs was amplified to serve as a standard curve for the calculation of relative quantities of the target gene using Sequence Detector software, version 1.7 (comparative cycle threshold method). The Gapdh gene was used to standardize the data. All results were obtained from at least two independent experiments, and each assay was duplicated. PCR amplification was performed using the following primer sets, as described [32] except for a Gapdh gene: Peg1/Mest, 5′-ATTCGCAACAATGACGGC-3′ and 5′-TGAGGTGGACTATTGTGTCACC-3′; Snrpn, 5′-ATACTGGCATTGCTCGTGTG-3′ and 5′-TGGAGGAGGCATGCCTAT-AG-3′; Igf2r, 5′-TTCGACCTATAAGAAGCCTT-3′ and 5′-GG-GTACTTTGCTTTTGGGTA-3′; H19, 5′-TGTAAACCTCTTTG-GCAATGCTGCC-3′ and 5′-TATTGATGGACCCAGGACCT-CTGGT-3′; Igf2, 5′-CTACTTCAGCAGGCCTTCAAG-3′ and 5′-GATGGTTGCTGTACATCTCC-3′; Gapdh, 5′-AATGCATCCT-GCACCACCAA-3′ and 5′-GTGGCAGTGATGGCATGGAC-3′.

Statistical Analysis

Data are shown as averages and standard deviations. Student's or Welch's t test was used for body weight, methylation, and gene expression analyses. Pearson's correlation coefficient test was used for correlation among methylation, gene expression, chimerism, and body weight. A p value of <0.05 was considered significant. All statistical analyses were performed with Excel X and Statcel2 for Macintosh.

Results

Methylation Status of DMRs in Preimplantation PG and Undifferentiated PGES

The methylation status of PG and PGES was analyzed for the DMRs of maternally methylated imprinted genes Peg1/Mest, Snrpn, and Igf2r using COBRA and a bisulfite genomic sequencing method. At the eight-cell stage, normal embryos had both methylated and unmethylated alleles, whereas almost all alleles were methylated in PG (Fig. 1). However, the loss of imprinting was observed at parthenogenetic blastocysts in more 2 days of culture. Although the theoretical methylation level was 100% in these three genes of PG, it was actually approximately 60%–70% in this stage. The loss of imprints was also seen in normal cultured blastocysts. On the other hand, for the analysis of PGES, two B6-GFP PGES lines (B6#1 and B6#2) and three F1-GFP (CBA/Ca × B6-EGFP F1) PGES lines (F1#1, F1#2, and F1#3) were established. Demethylation at the same level as parthenogenetic blastocysts was also observed in undifferentiated PGES cell lines as a result of COBRA (Fig. 1). However, in the analysis of bisulfite genomic sequencing, demethylation was sparse in parthenogenetic blastocysts, whereas completely demethylated alleles existed in all PGES samples (Fig. 2). In contrast, the paternally methylated imprinted gene H19 was almost completely unmethylated in PG and PGES (Fig. 1).

Figure Figure 1..

Partial demethylation was observed in preimplantation PG and undifferentiated PGES. Analysis of DMR methylation of Peg1/Mest, Snrpn, Igf2r, and H19 genes by combined bisulfite restriction analysis (COBRA) in normal embryos (eight-cell and blastocyst), PG (eight-cell and blastocyst), and PGES (F1 and B6). Representative COBRA electrophoresis gels stained with ethidium bromide and visualized for fluorescent bands are shown. Both digested PCR products (+) and undigested products (−) were subjected to gel electrophoresis. Bar graphs show the methylation percentages, which are the average of three (blastocysts) or two (others) independent assays. Abbreviations: PG, parthenogenetic embryos; PGES, parthenogenetic embryonic stem cells.

Figure Figure 2..

Complete demethylated alleles existed in undifferentiated PGES, whereas only sparse demethylation was seen in PG. DNA methylation of normal embryos (eight-cell), PG (eight-cell and blastocyst), and PGES (B6 and F1) was analyzed by bisulfite genomic sequencing. The methylation status of the maternally methylated imprinted genes Peg1/Mest, Snrpn, and Igf2r is shown schematically (filled circles represent methylated sites). Abbreviations: DMR, differentially methylated region; PG, parthenogenetic embryos; PGES, parthenogenetic embryonic stem cells.

Next, these GFP transgenic PG and PGES were used to produce parthenogenetic chimeras to isolate parthenogenetic cells in vivo. Some of the chimeric embryos were collected by caesarean section at the prenatal stage. The productive efficiency of PG and PGES chimeras is summarized in Table 1. Growth retardation of newborns was not observed in B6 PGES chimeras (normal, 1.48 ± 0.21 g, vs. chimera, 1.48 ± 0.37 g; p = .45) and F1 PGES chimeras (normal, 1.84 ± 0.24 g, vs. chimera 1.81 ± 0.26 g; p = .7), as previously reported [18]. In our study, the proportion of males (39 of 56 = 69.6%) was significantly higher (p < .05) in PGES chimeras than normal control mice by the χ2 test for independence. The previous report [18] also implied a higher proportion of male chimeras. For PGES chimeras, male host embryos could have higher survival ability than female host embryos during the prenatal stage. The presence of the Y chromosome or X inactivation might be related, but we do not know the exact reason.

Table Table 1.. Production of parthenogenetic chimeras
original image

Methylation Status of DMRs in Somatic Lineages

To collect PG- and PGES-derived somatic cells, MEFs from E13.5 chimeras were generated because parthenogenetic cells contribute extensively to the majority of tissues at this stage [9]. PG and PGES derivatives were sorted by FACS with a purity of more than 95% in each sample. The methylation status of the imprinted genes was investigated by COBRA (Fig. 3A). In PG-derived MEFs, almost all methylation imprints were apparently restored compared with those of blastocysts. The average methylated percentages were restored in Peg1/Mest (66%→80%), Snrpn (64%→84%), and Igf2r (61%→81%). In contrast, in PGES-derived MEFs, some samples restored methylation imprints as well as PG derivatives did, whereas others did not. The average methylation percentages of PGES-derived MEFs were not significantly different from those of PG derivatives in each gene (p > .05); however, when the data of all genes were unified, a significant difference was seen between PG and PGES derivatives (p < .05). Moreover methylation of PGES derivatives was altered in samples originating from the same cell lines and showed higher standard deviations than PG derivatives. For instance, the methylation of all three maternally methylated imprinted genes was highly demethylated in B6#2-2, F1#3-2, and F1#3-4 MEFs and less demethylated in B6#2-3 and F1#3-3 MEFs, although these MEFs were all of B6#2 and F1#3 PGES origin, respectively. The result of bisulfite sequencing confirmed that complete demethylation was occurred in some alleles of PGES-derived MEFs (Fig. 3B).

Figure Figure 3..

Methylation status of somatic lineages in parthenogenetic cells. (A): Combined bisulfite restriction analysis of E13.5 MEFs of PG and PGES. Both digested PCR products (+) and undigested products (−) were subjected to gel electrophoresis. Among PGES-derived MEFs, samples 1-1 to 1-3 were B6 PGES 1 line-derived, samples 2-1 to 2-4 were B6 PGES 2-derived, and samples 3-1 to 3-4 samples were F1 PGES 3-derived. (B): Bisulfite genomic sequencing of PG- and PGES-derived MEFs. Abbreviations: E, embryonic day; MEF, mouse embryonic fibroblast; PG, parthenogenetic embryos; PGES, parthenogenetic embryonic stem cells.

Expression of Imprinted Genes in Somatic Lineages

To clarify whether demethylation is correlated to expression level of imprinted genes, we carried out quantitative real-time reverse transcription-PCR. Among maternally methylated imprinted genes, Peg1/Mest and Snrpn are expressed from the paternal allele, whereas Igf2r is expressed from the maternal allele. Therefore, in parthenogenetic cells, the loss of imprints leads to the upregulation of Peg1/Mest and Snrpn and the downregulation of Igf2r. In fact, hypomethylated B6#2-2, F1#3-2, and F1#3-4 MEFs were upregulated in Peg1/Mest and Snrpn expressions and downregulated in Igf2r expressions, compared with other PGES samples (Fig. 4). The average expression levels of each gene were not significantly different between PG and PGES derivatives (p > .05); however, there were significant correlations (p < .05) between the methylation status of DMRs and the gene expression level in these three genes (Fig. 5A). Accordingly, demethylation of PGES derivatives would be responsible for the change in the gene expression level. Next, correlations between the demethylation of imprinted genes and the tissue contribution of PGES-derived cells were examined. For E13.5 chimeras, there was low correlation between the percentage of Igf2r DMR2 methylation and the percentage of chimerism (R2 = 0.4719); however, this correlation was much higher in newborn chimeras (R2 = 0.6981; Fig. 5B). For Peg1/Mest and Snrpn, there were no significant correlations (p > .05), but we obtained higher correlation coefficients (R) in newborns than in E13.5 fetus (data not shown). We also examined the correlation between the percentage of methylation and body weights, but there were no significant correlations (data not shown). On the other hand, paternally imprinted genes H19 and Igf2, which are regulated by H19 DMR methylation, were generally unmethylated in both PG and PGES-derived cells (Fig. 3A), and the expression level of both genes did not differ between PG and PGES chimeras (Fig. 4).

Figure Figure 4..

Expression of imprinted genes in parthenogenetic mouse embryonic fibroblasts derived from PG, PGES, and N. Quantitative real-time polymerase chain reaction was performed for the Peg1/Mest, Snrpn, Igf2r, H19, and Igf2 genes. The Gapdh gene was used to standardize the data. Data are the average of two independent assays, and each assay was duplicated. Standard deviations are indicated by bars. Abbreviations: N, normal embryos; PG, parthenogenetic embryos; PGES, parthenogenetic embryonic stem cells.

Figure Figure 5..

Methylation in DMRs of imprinted genes regulates gene expression levels and the tissue contribution rate of parthenogenetic cells in chimeras. (A): The percentage of DMR methylation is plotted against the relative gene expression level for PG chimeras (filled circles; n = 6) and PGES chimeras (open circles; n = 11). Methylation level was determined by combined bisulfite restriction analysis, and chimerism was determined by the percentages of GFP-positive cells. Significant correlations (p < .05) were found for Peg1/Mest, Snrpn, and Igf2r. (B): Correlations between the percentage of Igf2r DMR2 methylation and the percentage of chimerism for PGES chimeras. Significant correlations (p < .05) were found for E13.5 and newborn chimeras. Low correlation was observed in E13.5 fetuses (n = 11), whereas higher correlation was observed in newborns (n = 7). Abbreviations: DMR, differentially methylated region; E, embryonic day; PG, parthenogenetic embryos; PGES, parthenogenetic embryonic stem cells.

Methylation Status of DMRs in Germ Lineages

Next, we investigated whether imprints of these parthenogenetic cells were erased in germ lineages. For the isolation of female PGCs, immunofluorescent staining using the PGC-specific marker SSEA-1 antibody-conjugated B-PE was performed. Both GFP-positive (green) and B-PE-positive (red) cells were collected as the parthenogenetic cell-derived PGCs, and only B-PE-positive cells were collected as the normal host-derived PGCs (supplemental online Fig. 1). In contrast, oocytes at 1 dpp can be distinguished from somatic cells by the difference in size; namely, the diameter of normal oocytes at this stage is approximately 15 μm. As expected, both parthenogenetic and normal oocytes were easily isolated depending on the GFP marker (supplemental online Fig. 2). Both PG- and PGES-derived germ cells contributed equally to female gonads, and their morphology was normal. Methylation status of DMRs in germ lineages was analyzed using these samples. Up to E11.5, we observed a substantial amount of methylated clones in Snrpn DMR1 in PG-derived PGCs, whereas demethylation has already started in Peg1/Mest and Igf2r (Fig. 6). At this developmental stage, there are no significant differences in the methylation status between PG and normal host embryos. Thereafter, the population of PGCs that was demethylated at CpG sites gradually increased, and 2 days later, at E13.5, there was a striking reduction in the overall methylation state of most DMRs in both PG (n = 2) and PGES (n = 3) chimeras. This timing of erasure was normal compared with the biparental normal PGCs. Erasure was maintained in 1-dpp oocytes, immediately before the acquisition of new imprints (PG chimeras: n = 3, PGES chimeras: n = 3). Thus, imprints of parthenogenetic PGCs from PG and PGES chimeras were normally reprogrammed, as were normal PGCs in germ lineages.

Figure Figure 6..

Normal erasure of imprinting in parthenogenetic cells that enter the germ lineages. PGCs at E11.5 and E13.5 and oocytes at 1 dpp from PG chimeras, PGES chimeras, and normal control mice were examined for the maternally methylated imprinted genes Peg1/Mest, Snrpn, and Igf2r. Abbreviations: DMR, differentially methylated region; dpp, day postpartum; E, embryonic day; PG, parthenogenetic embryos; PGC, primordial germ cell; PGES, parthenogenetic embryonic stem cells.

Discussion

Parthenogenetic embryos cannot develop to term in mammals [1, 2]. On the other hand, parthenogenetic chimeras rescued by normal embryos develop to term [4]. In parthenogenetic chimeras, PGES resembled PG in their developmental potential. However, PGES chimeras are more normal in body weight and tissue contribution than PG chimeras [18]. We postulated that this difference might be caused by the modified expressions of imprinted genes due to loss of imprinting in PGES. To investigate the epigenetic status of parthenogenetic cells in somatic and germ lineages, we produced parthenogenetic chimeras using PG and PGES.

Genomic Imprinting in Somatic Lineages of Parthenogenetic Cells

In mammals, methylation patterns are reprogrammed genome-wide in premeiotic germ cells and in blastocysts. However, demethylation of imprinted genes generally does not occur in the blastocyst stage in vivo. Therefore, maternally methylated imprinted genes are expected to be completely methylated in parthenogenetic preimplantation embryos. However, partial loss of imprinting was found in parthenogenetic blastocysts. This partial demethylation also occurred in in vitro-cultured normal blastocysts as reported previously [33]. Perhaps, demethylation was caused in parthenogenetic blastocysts as well as normal blastocysts in vitro. However, such demethylation is generally restored postimplantation. Mann et al. reported that the loss of imprints in normal blastocysts remained in placentas but was restored in fetuses postimplantation [33]. In our experiment, partial demethylation in parthenogenetic blastocysts (61%–66% methylation) was restored in chimeric fetuses (81%–84% methylation), indicating that the loss of imprints in PG was considerably restored postimplantation.

The partial loss of imprints was also observed in undifferentiated PGES. However, the loss was frequently not restored in PGES chimeras postimplantation. In some cases, the loss of imprints seemed to progress in PGES chimeras. Why did loss of imprints occur only in PGES chimeras? Regarding this point, we can make some explanations. The first explanation is that XX ES cells are more susceptible to demethylation than XO and XY ES cells [19]. A previous report indicates that XO and XY ES cells have the restoration ability of imprints, but XX ES cells do not [19]. Generally, PGES have two X chromosomes, so that hypomethylated PGES could progress the hypomethylation of imprinted genes in chimeras. The second explanation is that the level of DNA methyltransferase (Dnmt) expression may be insufficient for PGES. Dnmt3a and Dnmt3b are known to play a critical role as the de novo methyltransferase for the restoration of methylation postimplantation [34, 35]. Interestingly, the forced expression of Dnmt3a or Dnmt3b restores the methylation of XX ES cells [19], suggesting that the levels of Dnmt3a and/or Dnmt3b are not sufficient for remethylation in PGES. On the other hand, once methylation of imprinted genes is completely lost in XX ES cells, the loss is not restored by the forced expression of Dnmt3a or Dnmt3b in vitro [19]. According to the bisulfite genomic sequencing of PG and PGES, demethylation was sparse in parthenogenetic blastocysts, whereas completely demethylated alleles existed in undifferentiated PGES (Fig. 2), suggesting that completely demethylated alleles in PGES were not able to be remethylated in chimeras.

On the other hand, PGES-derived MEFs with various imprint statuses were obtained from the same ES cell line. Of course, it is possible that PGES have lost imprints during in vivo differentiation, as mentioned above. However, there is another explanation: it is possible that the cells that make up ES cells are not homogeneous but that a wide variation of epigenetic statuses is included in the same cell line. Certainly, the result of bisulfite sequencing of undifferentiated PGES, indicating the existence of various sequence patterns (Fig. 2), supports this hypothesis. In addition, the fluorescence in situ hybridization analysis of PGES clarified that both XX cells and XO cells were contained in each cell line (supplemental online Fig. 3). As described above, XO ES cells are stable but XX ES cells are unstable in methylation imprints [19]. This means that PGES with both stable and unstable methylation imprints were contained in the same cell line. When producing ES chimeras, chimeras are most often derived from one or two founder ES cells, even if many ES cells are introduced [36]. This indicates that if only a single ES cell that has lost its imprints or that is unstable in methylation imprints generates a chimera, methylation imprints are possibly lost in somatic cells. There may be several explanations for demethylation in PGES chimeras; however, we can at least say that PGES lose imprints in some alleles during long-term culture and differentiation.

Influence on Gene Expression Due to Loss of Imprints

In chimeras, higher overall levels of PGES than PG are detected, in terms of tissue contribution. Furthermore, no significant growth retardation is apparent in PGES chimeras, irrespective of their degree of chimerism or the PGES lines used [18]. Also, in our study, growth retardation was not found in B6 and F1 PGES chimeras. A phenotypic difference is expected to be caused by the difference in expressions of imprinted genes due to the loss of imprints in PGES derivatives. In some PGES chimeras, the expression of the Igf2r gene was reduced by demethylation of the Igf2r DMR2. The Igf2r gene encodes a nonmitogenic receptor that targets IGF-II to the lysosomes for degradation and, therefore, inhibits the mitogenic function of IGF-II [37]. Thus, the reduction of Igf2r expression in PGES seems to promote cell proliferation and fetal growth. Actually, the artificial reduction of Igf2r expression rescues the growth inhibition of parthenogenetic cells [23]. Thus, one of the reasons why body weight reduction is not seen in PGES chimeras seems to be the downregulation of the Igf2r gene expression. Besides the Igf2r gene, the loss of imprints was observed in Peg1/Mest and Snrpn genes, at least in our experiment. The Peg1/Mest gene is also related to embryonic growth [38], and the Snrpn gene regulates RNA splicing [39], indicating that the gene expression patterns were widely changed in PGES-derived somatic cells. Therefore, there is no doubt that these alterations of gene expressions cause the tissue contribution of PGES in chimeras. In PG chimeras, progressive elimination of PG occurs after day 13 of gestation [9]. We found higher positive correlations between demethylation and chimerism in newborns than in E13.5 fetuses, suggesting that demethylated parthenogenetic cells evaded progressive elimination from tissues after day 13 of gestation. In this study, we could not find significant correlations between methylation and body weight; however, demethylation, which occurred widely in DMRs of PGES, possibly affects the body weight of chimeras.

Genomic Imprinting in Germ Lineages of Parthenogenetic Cells

Next, we investigated whether the genomic imprinting of PG and PGES is reprogrammed in germ lineages. In normal female germ cells, the erasure of imprints occurs around the time that PGCs enter the gonad, and erasure is maintained until the postnatal growth phase of oogenesis [20, 21, 22]. This erasure process of parental imprinting in PGCs is necessary to reset and restore the new imprints in normal germ line cells. However, in parthenogenetic female germ cells, imprinting mechanisms, such as the erasure of parental imprints, seem to be unnecessary because parthenogenetic cells have the same imprint pattern as mature oocytes. We observed the almost complete erasure of imprints in PG- and PGES-derived germ cells, as well as in normal germ cells. Erasure of methylation imprints in parthenogenetic cells occurred more rapidly in Peg1/Mest and Igf2r than in Snrpn, in accordance with the previous reports about normal PGCs [40, 41]. This result indicates that the erasure and reprogramming of genomic imprinting is the inevitable event in germ lineages, regardless of imprint status of PGCs. The normal erasure of imprinting in parthenogenetic cells suggests that the paternally imprinted allele is not related to the erasure of imprints and the gene that controls the erasure of imprinting functions normally in parthenogenetic cells. It may be interesting to examine the germ cells of later stage for the acquisition of de novo methylation.

In summary, this study revealed that the variations of imprint status were frequently observed in somatic cells of PGES origins in chimeras. Demethylation occurring in PGES may reprogram the maternally methylated imprinted genes and improve the tissue contribution and possibly body weight to a normal level. In contrast, the uniparental imprinting in both PG and PGES is normally erased in germ cells, suggesting that reprogramming in germ lineages is an inevitable event and that paternal imprints do not relate to the erasure of imprints in germ lineages. Recently, it has been reported that uniparental ES cells can differentiate into transplantable hematopoietic progenitors in vitro that contribute to long-term hematopoiesis in recipients [42]. The PGES that have lost maternal methylation imprints and obtained more normal imprint pattern might be used for fertility treatment and regenerative medicine.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Acknowledgements

We thank Dr. M. Okabe for providing C57BL/6-Tg (ACTbEGFP)1Osb/J mice. This work was supported in part by grants from the Japan Society for the promotion of Science (Grant 17770182); Japan Science and Technology Corporation; the Ministry of Education, Culture, Sports, Science and Technology of Japan; the Ministry of Health, Labor and Welfare of Japan; and Japan Health Sciences Foundation.

Ancillary