Conceived and designed the experiments: TD GF. Performed the experiments: TD. Analyzed the data: TD GF. Contributed reagents/materials/analysis tools: TD NL. Wrote the paper: TD GF.
The authors have declared that no competing interests exist.
X chromosome inactivation (XCI) is a dosage compensation mechanism essential for embryonic development and cell physiology. Human embryonic stem cells (hESCs) derived from inner cell mass (ICM) of blastocyst stage embryos have been used as a model system to understand XCI initiation and maintenance. Previous studies of undifferentiated female hESCs at intermediate passages have shown three possible states of XCI; 1) cells in a pre-XCI state, 2) cells that already exhibit XCI, or 3) cells that never undergo XCI even upon differentiation. In this study, XCI status was assayed in ten female hESC lines between passage 5 and 15 to determine whether XCI variations occur in early passages of hESCs. Our results show that three different states of XCI already exist in the early passages of hESC. In addition, we observe one cell line with skewed XCI and preferential expression of X-linked genes from the paternal allele, while another cell line exhibits random XCI. Skewed XCI in undifferentiated hESCs may be due to clonal selection in culture instead of non-random XCI in ICM cells. We also found that
Human embryonic stem cells (hESCs) are an invaluable tool for regenerative medicine and a model for early human embryogenesis
In mice,
A recent study in pre-implantation human embryos reported that
Since differentiation of hESCs can be used to model human embryogenesis
It is therefore better to evaluate the status of XCI in early passages of undifferentiated hESCs that have been minimally exposed to culture effects. Hereby we report the status of XCI in ten lines of female hESC at the earliest passages available. Our results indicate that the three distinct states of XCI can be observed even in minimally passaged hESCs. In addition, we investigated the pattern of XCI in two cell lines- one showed random XCI reminiscent of mESCs, while the other showed non-random XCI. Consistently, we found that the methylation pattern of the
Recent studies have identified three distinct states of XCI in a variety of female hESCs
Relative expression levels of
(A)
It has been shown that XCI initiation correlates with low levels of pluripotency related factors in mouse ES cells
(A) CSES1 shows positive punctate staining for H3K27me3 (indicated by the arrow heads). CSES3 cells do not show punctate staining pattern for H3K27me3 in (B) late or (C) early passages. (D) Shown is
To visualize XCI states at a single cell resolution, we assayed for
Low level of
Finally, we also observed the transition from the
During mouse embryogenesis, imprinted XCI of the paternal X chromosome occurs prior to the blastocyst formation whereas random XCI is initiated upon differentiation. However, previously we observed non-random XCI in later passages of female hESC lines
We chose to analyze the pattern of XCI in two cell lines, CSES1 and CSES8, because both showed XCI markers. In order to verify the parental origin of the active X chromosome, we used maternal DNA that was extracted from granulosa cells, which are somatic cells that surround the oocyte during maturation. These cells are normally removed at the time of egg retrieval during IVF procedure in order to expose the oocyte surface for fertilization. Frozen granulosa cells for CSES1 and CSES8 were used for DNA extraction. Following DNA amplification these samples were hybridized to Affymetrix SNP array (250k Sty) side by side with the corresponding hESC DNA. The results from this analysis enabled us to choose SNPs within X-linked genes that were identified as heterozygous in the hESC lines and homozygous in the maternal DNA.
In order to verify the maternity of the granulosa DNA sample with the corresponding hESC sample, we performed identical by state (IBS) analysis to show that our samples are genetically related. For instance, comparison of CSES1 and its corresponding maternal sample showed that in 71.7% of the SNPs both alleles are shared (IBS = 2) and in 28% of the SNPs one of the alleles is shared (IBS = 1). This indicates that 99.7% of SNPs show at least one allele shared between the samples. Overall 85.7% of SNPs are shared between CSES1 and its maternal sample. This indicates close genetic relationship between the two samples (
SNP ID | Band | chromosome location | Gene name | CSES1 genotype | Granulosa genotype | CSES1 p5 expression | CSES1 p14 expression |
rs5914796 | Xp11.21 | 56807583 | DKFZp686L07201 | A/T | T/T | A | A |
rs4828327 | Xq21.1 | 84236784 | SATL1 | A/C | C/C | A | - |
rs6620161 | Xq21.33 | 96027150 | DIHPA2 | A/G | G/G | A/G | A/G |
rs2428212 | Xq24 | 118985598 | UPF3B | A/G | G/G | A | A |
rs6641482 | Xq28 | 147887801 | AFF2 | G/A | A/A | G | - |
rs41537046 | Xq26.2 | 132470683 | GPC4 | A/G | G/G | A | A/G |
rs895744 | Xq28 | 153998985 | BRCC3 | G/T | T/T | G | - |
SNP ID | Band | Chromosome location | Gene name | CSES8 genotype | Granulosa genotype | CSES8 p5 expression | CSES8 p14 expression |
rs3747276 | Xp22.11 | 21985464 | SMS | A/G | G/G | A/G | A/G |
rs6628886 | Xp21.1 | 34654777 | TMEM47 | A/G | G/G | A/G | A/G |
rs6625472 | Xq13.1 | 68739542 | TMEM28 | A/G | G/G | A/G | A/G |
rs479640 | Xq13.2 | 73668829 | SLC16A2 | C/T | T/T | C/T | C/T |
rs717689 | Xq21.1 | 77379935 | PGK1 | A/G | G/G | A/G | A/G |
rs1204399 | Xq22.1 | 99886830 | TSPAN6 | A/G | G/G | A/G | A/G |
rs2294504 | Xq23 | 109552667 | AMMECR1 | C/T | T/T | C/T | C/T |
rs42890 | Xq24 | 119578513 | LAPM2 | T/G | G/G | T/G | T/G |
rs5977910 | Xq26.2 | 132901293 | GPC3 | T/G | T/T | T/G | T/G |
Our results indicate discrepancies between the two hESC lines with regard to the pattern of XCI. In the CSES8 cell line at both p5 and p14, we consistently observed bi-allelic expression of X-linked genes, indicating that random XCI has occurred. On the other hand, CSES1 at p5 showed preferential expression of the paternal allele suggesting skewed XCI with six out seven X-linked genes exhibiting mono-allelic expression from the paternal X. Theoretically, the probability of having maternal or paternal allele expressed per SNP is 50%. According to binomial distribution B(x = 6; n = 7, p = 0.5), the cumulative probability of getting as many as six out of seven SNPs with paternal expression is 0.0625. Since this probability is very small, it argues against the hypothesis of random XCI in this case. Therefore, the preferential expression from the paternal allele is due to skewed maternal X inactivation. Moreover, upon prolonged culture of CSES1 cell line, we observe partial reactivation of the second allele that coincides with the loss of
The variability of XCI status in early passages may indicate epigenetic modifications of the
Schematic of
In this study, we attempted to minimize the
Here we show that variations of XCI are already present from early passage (p5) to later passage (p14) of the same cell line (CSES8), supporting the notion that XCI is highly affected by culture conditions
Concerning the pattern of XCI in hESCs at the early passages, our SNP analysis revealed random XCI for one line of hESCs (CSES8) while skewed XCI for the other cell line (CSES1) with preferential expression from the paternal X chromosome. Skewed XCI is known to happen during mouse embryogenesis at the two cell stage until the blastocyst stage when the inactive paternal X becomes reactivated. In this case, the inactivation is imprinted and the paternal X chromosome is inactivated. In hESCs, methylation specific analysis of a polymorphic tri-nucleotide repeat at the
The mechanism underlying the initiation and loss of XCI markers in hESCs is largely unknown. It is known that
In this study, we propose that XCI occurs in undifferentiated hESC in a random manner and the observations of skewed XCI are probably a result of a clonal selection occurring in hESC culture. It was recently shown that mouse induced pluripotent cells (miPSC) recapitulate XCI patterns of mESCs
Overall the variability in the status of XCI among the cell lines indicates either rapid epigenetic culture effect or the potential heterogeneity of the original ICM cells. It seems that the cells, even in early passages, tend to undergo XCI and later on lose the XCI markers. Thus, XCI process is highly affected by culture conditions and inactivation of one of the X chromosomes may provide an advantage in the current culture condition. We propose that a careful re-examination of XCI status in human ICM will shed light on the status and pattern of XCI in these cells. It is known that normal XCI is critical in the embryonic development
This research was approved by UCLA Embryonic stem cells research oversight (ESCRO) committee.
Cedars Sinai Embryonic Stem cell lines (CSES) were derived as described
Total RNA was extracted using RNeasy mini kit (Qiagen, Valencia, CA
RNA from undifferentiated hESCs were analyzed by a Taqman© based assay, using the human stem cell pluripotency array (ABI, foster city,CA). Delta Ct values were obtained by identifying the number of amplification cycles needed to reach the common threshold (Ct) value for each gene. Then, these values were normalized by subtraction of the Ct values obtained for a control gene (beta-Actin) for the same sample. In order to show that
Genomic DNA from the hESC lines was extracted using DNeasy kit (Qiagene). Due to low amount of starting cells, DNA for Granulosa cells was extracted with the same kit and was subjected to whole genome amplification using REPLI-g mini kit (Qiagene). DNA from CSES1 (p14), CSES8 (p12) and their corresponding maternal granulosa cells were hybridized to Affymetrix 250k Sty SNP array (Affymetrix, Santa Clara, CA
Genomic DNA samples for CSES1, 2, 3, 5 and 7 were digested with
Cells for immunostaining were first fixed with 4% PFA/PBS for 20 minutes at room temperature and washed with PBS. Cells were then permeabilized by 0.4% Triton-X in TBST for 20 min and blocked in 10% milk and 1% normal goat serum for 1 hour. Cover slips were incubated 1 hour at room temperature with primary antibodies diluted in 3% BSA in TBST [monoclonal mouse anti-OCT4 (1∶20, Santa Cruz) and polyclonal rabbit anti-H3K27me3 (1∶1000, a gift from Yi Zhang, University of North Carolina, Chapel Hill, NC)]. After being washed three times with PBS, cover slips were incubated in fluorochrome-conjugated secondary antibodies for 1 hour at room temperature with protection from light. Hoechst dye #33342 was used to label cell nuclei.
Primer sequences.
(0.05 MB DOC)
Percentage of undifferentiated cells positively stained for H3K27me3. Intermediate state (state II) cells show significant number of cells with punctate staining for H3K27me3 CSES1 p15 (n = 511, 94%), CSES8 p8 (n = 257, 87%) CSES10 p15 (n = 257, 32%) and CSES11 p12 (n = 655, 84%).
(4.55 MB TIF)
Identical by state (IBS) analysis for CSES1 and its granulosa cells. Proportion of IBS = 0, 0.3% (no shared alleles), IBS = 1, 28% (one shared allele) and IBS = 2, 71.7% (two shared alleles). Overall, 85.7% of the alleles are shared, clearly indicating for close genetic relationship between the samples.
(8.34 MB TIF)
SNP sequences for CSES1 samples. SNP rs6620161 shows biallelic expression in p5 with one of the alleles more prominently expressed. However, in p14 both alleles are expressed at the same level. rs41537046 shows monoallelic expression at p5, but at p14 both of the alleles are already expressed. SNP rs5914796 shows expression of the paternal allele both at p5 and p14.
(9.12 MB TIF)
SNP sequences for CSES8 samples. Representing SNP sequences for CSES8 cell line. SNPs rs1204399, rs42890 and rs6625472 all show biallelic expression both at p5 and p14.
(9.24 MB TIF)
Induction of XCI in CSES upon Retinoic Acid differentiation. XCI detected by immunostaining for H3K27me3 and pluripotency detected by staining for OCT4 were tested in the three different classes of CSES cells. In CSES2 and CSES7 (A–D), representing class I cells, we were able to detect induction of XCI upon differentiation in CSES2 (A, B) but not for CSES7 (C, D). CSES3 representing class II cells were not able to induce XCI upon differentiation (E, F). In CSES8, we were able to detect XCI markers both in the undifferentiated and differentiated cells (G, H).
(9.89 MB TIF)
We greatly thank Professor Nissim Benvenisty for reagents and many helpful discussions during the course of this study and Professor Rita Cantor for statistical analysis of the data presented. We would also like to thank Kavita Narwani and Dr. Juan-Carlos Biancotti for their help in the derivation and characterization of the CSES cell lines, Shaily Shah for her help in the SNP analysis and Kevin Huang and Thuc Le for proofreading this manuscript.