Main

Haploid genetics is useful for delineating genome function. Although haploid genetics has been mostly restricted to unicellular organisms, recent reports of haploid ES cells have extended it into animal species, including mammals1,2,3,4,5,6,7. Deriving haploid human ES cells by similar approaches has probably been hindered by the limited availability of human oocytes8. Artificial activation of unfertilized metaphase II (MII) human oocytes results in efficient development to the blastocyst stage and subsequent derivation of parthenogenetic ES (pES) cells9,10,11. In mouse parthenogenetic embryos haploidy usually persists at the blastocyst stage12,13, but diploid cells progressively dominate over increasing cell cycles due to spontaneous and irreversible diploidization13,14,15 (Fig. 1a). By estimation, even if diploidization occurs in 1 out of 10 cell cycles, 1% of ES cells may remain haploid at early passages (Extended Data Fig. 1a).

Figure 1: Derivation of haploid human ES cells.
figure 1

a, Schematic of putative haploidy in pES cells. b, DNA content profiles of haploid pES10, established by repeated enrichment of 1c cells. Top to bottom, unsorted diploid cells, partially purified (fourth sort) and mostly purified (sixth sort) haploid cells. c, pES10 karyotypes before and after 1c-cell enrichment. p, passage. d, e, DNA FISH (d) and centromere staining (e) in haploid-enriched pES10 cells. Magnifications show haploid nuclei with a single hybridization signal (d) and 23 centromere foci (e), respectively. Scale bars, 10 μm.

PowerPoint slide

Full size image

To explore the feasibility of deriving haploid human ES cells, we generated and analysed a collection of 14 early-passage human pES cell lines for the persistence of haploid cells. We initially used chromosome counting by metaphase spreading as a method for unambiguous and quantitative discovery of rare haploid nuclei. Among ten individual pES cell lines, a low proportion of haploid metaphases was found in one cell line, pES10 (1.3%, Extended Data Table 1). In four additional lines, we also used viable fluorescence-activated cell sorting (FACS) with Hoechst 33342 staining, aiming to isolate cells with a DNA content corresponding to less than two chromosomal copies (2c), leading to the successful enrichment of haploid cells from a second cell line, pES12 (Extended Data Table 2).

Two individual haploid-enriched ES cell lines were established from pES10 and pES12 (hereafter referred to as h-pES10 and h-pES12) within five to six rounds of 1c-cell enrichment and expansion (Fig. 1b and Extended Data Fig. 1b). These cell lines were cultured in standard conditions for over 30 passages while including cells with a normal haploid karyotype (Fig. 1c and Extended Data Fig. 1c). However, since diploidization occurred at a rate of 3–9% cells per day (Extended Data Fig. 1a; see Methods), sorting at every three to four passages was required to maintain haploid cells. Visualization of ploidy was further enabled by DNA fluorescence in situ hybridization (FISH) (Fig. 1d and Extended Data Fig. 1d) and quantification of centromere protein foci (Fig. 1e and Extended Data Fig. 1e; see Supplementary Notes and Extended Data Fig. 2). Besides having an intact karyotype, haploid ES cells did not harbour significant copy number variations (CNVs) relative to their unsorted diploid counterparts (Extended Data Fig. 1f), nor common duplications that would result in pseudo-diploidy, indicating that genome integrity was preserved throughout haploid-cell isolation and maintenance.

Both h-pES10 and h-pES12 exhibited classical human pluripotent stem cell features, including typical colony morphology and alkaline phosphatase activity (Fig. 2a, b). Single haploid ES cells expressed hallmark pluripotency markers as confirmed by centromere foci quantification in essentially pure (>95%) haploid cultures (Extended Data Fig. 3). Selective flow cytometry enabled the validation of the expression of two human ES-cell-specific cell surface markers (TRA-1-60 and CLDN6 (ref. 16)) in single haploid cells (Fig. 2c). Moreover, sorted haploid and diploid ES cells showed highly similar transcriptional and epigenetic signatures of pluripotency genes (Fig. 2d, e). In accordance with a parthenogenetic origin, these cells also featured distinct molecular profiles of maternal imprinting (Extended Data Fig. 4a–d).

Figure 2: Haploid human ES cells display pluripotent stem cell characteristics and enable loss-of-function screening.
figure 2

a, b, Colony morphology and alkaline phosphatase staining. Scale bars, 50 μm. c, Flow cytometry analysis of gated h-pES10 1c cells by co-staining DNA and cell surface markers TRA-1-60 and CLDN6. d, Expression levels of pluripotency genes in G1-sorted haploid and diploid ES cells (n = 4 each, two biological replicates per line, logarithmic scale). Mean indicated by bars. e, DNA methylation levels at pluripotency genes in biological duplicates of G1-sorted ES cells and control fibroblasts (Fib). f, Schematic of gene trapping and screening for 6-TG-resistance genes. g, NUDT5 insertions (red arrows) detected in three 6-TG-resistant colonies (indicated by TTAA). h, Metabolic pathway leading to 6-TG toxicity.

PowerPoint slide

Full size image

Haploid cells are valuable for genetic screening because phenotypically selectable mutants can be identified upon disruption of single-copy alleles. To demonstrate the applicability of this principle in haploid human ES cells, we generated a genome-wide mutant library using a gene trap transposon system (Fig. 2f and Extended Data Fig. 4e; see Methods), and screened for resistance to the purine analogue 6-thioguanine (6-TG). Out of six isolated and analysed 6-TG-resistant colonies, three harboured an identical gene trap insertion at the NUDT5 autosomal gene (Fig. 2g). The disruption of this gene was recently confirmed to confer 6-TG resistance in human cells17 (Fig. 2h). Detection of a loss-of-function phenotype due to an autosomal mutation thereby validates that genetic screening is feasible in haploid human ES cells.

The ability of human ES cells to exist as both haploids and diploids led us to investigate whether these two ploidy states differ in certain aspects of gene regulation and cell biology. To analyse haploid and diploid ES cells in the same cell cycle phase, we used FACS to isolate G1-phase haploid cells (1c) and compared them with isogenic G1-phase diploid cells (2c) from unsorted diploid cultures (Fig. 3a and Extended Data Fig. 5a; see Methods). We first aimed to uncover putative ploidy-associated differences by comparing the transcriptomes of haploid and diploid ES cells using RNA sequencing (RNA-seq), considering that observed changes in expression levels would be relative to the total gene expression of each ploidy state, rather than representing absolute differences. On the genome-scale, undifferentiated haploid and diploid ES cells clustered closely and separately from differentiated embryoid bodies (EBs) (Fig. 3b). Nonetheless, we identified 275 relatively upregulated and 290 relatively downregulated genes in haploids compared with diploids (greater than twofold change, false discovery rate (FDR) <0.05; Extended Data Fig. 5b).

Figure 3: Molecular and cellular comparisons of haploid and diploid ES cells.
figure 3

a, Experimental scheme. b, RNA-seq-based clustering analyses of isogenic G1-sorted haploid and diploid cells (two biological replicates per line) and diploid pES12-derived EBs. c, Pie chart of relatively downregulated and upregulated genes in haploid versus diploid ES cells. d–g, Differential X chromosome inactivation in haploid and diploid ES cells. d, Genome-wide expression moving median plot. e, XIST expression levels. f, H3K27me3 staining. Scale bar, 10 μm. g, X chromosome DNA methylation levels. h, Relative total RNA, cell volume and ratio of mitochondrial to genomic DNA (mtDNA/gDNA) between G1-sorted haploid and diploid ES cells. Number of replicates indicated in parenthesis. Error bars represent s.d. i, Mean expression levels ± s.e.m. of nuclear and mitochondrial oxidative phosphorylation genes in haploid and diploid ES cells (replicates as in Fig. 2d), and their schematic organization in this pathway. *P < 0.05; **P < 0.01 (two-tailed unpaired Student’s t test). Source data for h and i are available online.

PowerPoint slide

Source data

Full size image

Notably, X chromosomal genes were significantly overrepresented among the relatively upregulated gene set (40%, P < 0.001, χ2 goodness of fit test) (Fig. 3c), and the expression levels of X chromosomal genes alone clearly distinguished between haploid and diploid ES cells (Fig. 3b). These data correlate with an expected differential status of X chromosome inactivation in haploid and diploid human ES cells: whereas the single-copy X chromosome in haploids is transcriptionally active (Xa), one of the two X chromosomes in diploids is often inactivated (XaXi)18. Indeed, haploid human ES cells exhibited a relative increase in X chromosomal gene expression and lacked expression of the XIST transcript which drives X chromosome inactivation (Fig. 3d, e and Extended Data Fig. 5b–d), as in diploid XaXa human ES cells19. X chromosome inactivation is regulated by repressive histone modifications and DNA methylation. H3K27me3 foci were consistently observed in unsorted diploid ES cells, but not in the haploid-enriched counterparts (Fig. 3f). Moreover, methylome analysis showed that the X chromosome DNA methylation signature of haploid ES cells resembles that of diploid male ES cells (XaY), whose single-copy X chromosome is largely hypomethylated, as opposed to the composite pattern of a hypomethylated Xa and a hypermethylated Xi in diploid female cells (Fig. 3g). Interestingly, recently diploidized ES cells (see Methods) remained XaXa soon after diploidization by all assays mentioned earlier (Fig. 3a, d–g).

Normalization to total gene expression20 resulted in seemingly similar expression levels of autosomal genes, but higher levels of X-linked genes in haploid cells (Fig. 3d and Extended Data Fig. 5c). However, assuming that the absolute expression of X-linked genes in haploid Xa and diploid XaXi cells are equivalent, these data suggest a genome-wide autosomal gene expression level reduction in haploids (Extended Data Fig. 5e, f). In support of this, total RNA amounts isolated from haploid ES cells were significantly lower than those from equal numbers of diploid cells (Fig. 3h). A decrease in total gene expression implied that physical dimensions may also be altered. Indeed, the average haploid:diploid diameter ratio of G1-sorted ES cells was ~0.8, corresponding to ~0.7 surface area ratio and ~0.6 volume ratio (Fig. 3h and Extended Data Fig. 5g).

We subsequently focused on consistent differential regulation within autosomes (see Methods). By transcriptome and methylome analyses, genes relatively downregulated in haploid ES cells were significantly enriched for genes encoding signal-peptide-bearing proteins (Extended Data Fig. 5h). Remarkably, we also detected subtle yet significant relative upregulation of 11 genes involved in oxidative phosphorylation in haploid cells, including representatives encoding subunits of four out of the five complexes comprising this pathway (Fig. 3i and Extended Data Fig. 5i). Furthermore, all 13 mitochondrially encoded oxidative phosphorylation genes were consistently upregulated as well (Fig. 3i), indicating coordinated regulation between nuclear and mitochondrial genes. This coincided with a 32% increase in the mitochondrial DNA (mtDNA) to nuclear DNA ratio between haploids and diploids (Fig. 3h), suggesting that mitochondrial abundance relative to nuclear DNA content is higher in haploid cells.

We next sought to assess the differentiation potential of haploid human ES cells. The 21-day-old EBs generated by spontaneous differentiation of haploid-enriched and diploid ES cells could not be distinguished by their appearance (Fig. 4a), and the morphology of dissociated haploid-cell-derived EB cells was consistent with differentiation (Extended Data Fig. 6a). Notably, metaphase spreading revealed a haploid karyotype (Fig. 4b; 4/4 metaphases), and a largely haploid (~70%) DNA profile was confirmed by flow cytometry in both h-pES10-derived and h-pES12-derived EB cells (Fig. 4c and Extended Data Fig. 6b). We then compared the gene expression profiles of G1-sorted haploid ES cells and EB cells, focusing on 18 lineage-specific genes across nine cell types (Fig. 4d). Whereas expression levels were negligible in undifferentiated ES cells, all tissue-specific genes were expressed in haploid and diploid EB cells (Fig. 4d and Extended Data Fig. 6c). Haploid and diploid EB cells showed insignificant expression of pluripotency-specific genes, consistent with efficient acquisition of differentiated cell fates.

Figure 4: Differentiation of haploid human cells.
figure 4

a, EBs from haploid-enriched and diploid pES12 cells. Scale bar, 100 μm. b, Haploid EB cell karyotype. c, DNA content profile of h-pES10 EB cells. d, Expression of tissue- and pluripotency-specific genes in G1-sorted haploid and diploid ES and EB pES10 cells. e, Differential X chromosome inactivation in haploid and diploid EBs and NPCs by genome-wide expression moving median plot. f, g, i, j, Centromere and differentiation marker co-staining in h-pES12-derived neurons (f), cardiomyocytes (g), definitive endoderm cells (i) and pancreatic cells (j). Scale bars, 10 μm. h, k, DNA content profiles of h-pES12 cells differentiated into cardiomyocytes (h) and PDX1-positive pancreatic cells (k). l, TUJ1 (ectoderm), α-SMA (mesoderm), AFP (endoderm) and OCT4 (pluripotency) staining in an h-pES12-derived teratoma. Scale bars, 50 μm. m, DNA content profile of an h-pES10-derived teratoma. n, Serial h-pES12-derived teratoma sections analysed histologically (left; scale bar, 20 μm) and by DNA FISH (middle; scale bar, 20 μm). Haploid nuclei are shown in magnification (right; scale bar, 5 μm).

PowerPoint slide

Full size image

To extend this analysis to more specific cell types, we subjected haploid ES cells to directed differentiation assays. Haploid ES cells differentiated towards a neural fate for ten days remained haploid, while giving rise to NCAM1-positive neural progenitor cells (NPCs, with ~90% efficiency) (Extended Data Fig. 7a, b). Sorted haploid NPCs expressed multiple neural-lineage-specific genes but not pluripotency-specific genes (Extended Data Fig. 7c, d). X chromosome inactivation in differentiated diploid female cells results in dosage compensation of 1:2 between the X chromosome and autosomes. As haploid cells do not inactivate their single-copy X chromosome, an X:autosomes dosage imbalance of 1:1 should persist into the differentiated state. Indeed, both haploid NPCs and EB cells showed an Xa signature contrary to the XaXi signature of diploid EB cells as indicated by whole-genome expression analysis and XIST levels (Fig. 4e and Extended Data Fig. 7e).

Neuronal differentiation was not restricted to the progenitor stage as the cells also differentiated with high efficiency (>90%) into mature TUJ1 (also known as β-tubulin III)-positive neurons by 20 days, with persistence of haploid cells as shown by co-staining with centromeres (Fig. 4f; 47% haploids, n = 104) and FISH analysis (Extended Data Fig. 7f, g; 46% haploids, n = 200). Similarly, haploid cells differentiated into TNNT2-expressing cardiomyocytes (Fig. 4g; 32% haploids, n = 97) during an eleven-day protocol resulting in spontaneously beating clusters (Supplementary Video 1), and 39% (n = 31) of haploid cells sorted from the whole culture (25% 1c-cells) were confirmed as TNNT2-positive (Fig. 4h and Extended Data Fig. 7h). Next, we differentiated haploid-enriched cultures (~70% haploids) to the pancreatic lineage. Analysing two stages of differentiation by centromere foci analysis, we observed robust differentiation (>90%) of both haploids and diploids into FOXA2-positive definitive endoderm cells (Fig. 4i; 56% haploids, n = 112), and differentiation into PDX1-positive pancreatic cells (Fig. 4j; 13% haploid, n = 103), some of which were also positive for NKX6-1. In addition, the persistence of haploid PDX1-positive cells was confirmed by flow cytometry (Fig. 4k; 10% PDX1-positive 1c cells; and Extended Data Fig. 7i, j).

Finally, both haploid-enriched human ES cell lines gave rise to teratomas comprising cell types of ectodermal, mesodermal and endodermal origins (Fig. 4l and Extended Data Fig. 8a, b), meeting the most stringent criterion for human pluripotency in vivo. Importantly, no residual undifferentiated OCT4-positive cells were detected (Fig. 4l and Extended Data Fig. 8b). Upon dissection, DNA content analysis revealed that a considerable population of h-pES10-derived teratoma cells remained haploid (Fig. 4m). Combined analysis of serial sections from an independent, h-pES12-derived teratoma, by histology and FISH confirmed the existence of in vivo differentiated haploid human cells capable of contributing to an organized tissue structure while responding to developmental signals (Fig. 4n). Haploid cells were identified in all analysed teratomas (n = 4), albeit with variable proportions, which may be influenced by the initial amount of haploid cells and/or the duration of teratoma formation.

Haploid mammalian cells have proven invaluable for loss-of-function screens7. Using a genome-wide library of gene-trapped haploid human ES cells, we demonstrated their potential for biomedically relevant functional genomics by forward genetic screening (Supplementary Discussion). Whereas previous studies on non-human haploid ES cells mostly emphasized the similarity between haploids and diploids, here we also pointed to several transcriptional, epigenetic and physical properties that set them apart (Supplementary Discussion). Interestingly, we did not observe global transcriptional compensation in haploid cells, indicating that it is not required for cellular viability as long as an autosomal balance is preserved. In contrast, autosomal imbalance appears intolerable based on the strict absence of human autosomal monosomies in in vitro fertilization ES cells21. Remarkably, we found that a haploid human karyotype is not a barrier for ES cell differentiation. As observed in the mouse2, haploid human ES cells gave rise to NPCs while remaining haploid. However, while mouse studies showed that haploid cells are lost upon further differentiation2,14, we observed specification of human haploid cells into somatic cell fates of all three embryonic germ layers, despite persistent dosage imbalance between the X chromosome and autosomes (Supplementary Discussion).

Throughout evolution, mammalian genomes have been solidified by diploidy-dependent adaptations such as parental imprinting, which restrict the development of haploid uniparental embryos. Nonetheless, haploid cells are capable of directing development in certain animal species22. The surprising differentiation potential of haploid human genomes suggests that diploidy-dependent adaptations, rather than haploidy, pose the predominant barriers for uniparental development in humans. The discovery of haploid human ES cells should thus provide novel means to delineate basic aspects of human genetics and development.

Methods

Data reporting

No statistical methods were used to predetermine sample size. The investigators were not blinded to allocation during experiments and outcome assessment.

Human oocyte manipulation and parthenogenetic ES cell line derivation

Human oocyte donation and pES and swaPS cell line derivation procedures were described previously11,23. Oocyte donors gave informed consent. Experiments were approved by the embryonic stem cell research oversight committee and the institutional review board at Columbia University Medical Center. Briefly, mature MII oocytes were activated using a calcium ionophore and/or an electrical pulse, followed by 4 h of culture with puromycin. Polar body extrusion and the presence of a single pronucleus indicating haploidy were confirmed, and oocytes were allowed to develop to the blastocyst stage. swaPS cells were derived following activation of an oocyte whose nuclear genome had been swapped with that of another oocyte11. ES cell lines were derived by laser ablation of the trophectoderm24 and addition of ROCK inhibitor Y-27632 at 10 μM to the derivation medium23. Then 2–3 days after plating, remaining trophectoderm cells were laser ablated, and inner cell mass cells were allowed to grow for 10–14 days until manual picking of the outgrowth was feasible.

Cell culture

Unless otherwise stated, human ES cells were cultured on a feeder layer of growth-arrested mouse embryonic fibroblasts (MEFs) in standard human ES cell medium composed of Knockout Dulbecco’s Modified Eagle’s Medium supplemented with 15% Knockout Serum Replacement (KSR, Thermo Fisher Scientific), 2 mM l-glutamine, 0.1 mM nonessential amino acids, 50 units ml−1 penicillin, 50 μg ml−1 streptomycin, 0.1 mM β-mercaptoethanol and 8 ng ml–1 basic fibroblast growth factor (bFGF). Cells were free of mycoplasma and maintained in a humidified incubator at 37 °C and 5% CO2. Passaging was carried out either mechanically with gentle trypsinization using trypsin solution A without EDTA (Biological Industries), or enzymatically using TrypLE Express (Thermo Fisher Scientific) with addition of 10 μM ROCK inhibitor Y-27632 (Stemgent) for 1 day after splitting. Haploid ES cells could also be grown in feeder-free conditions on Matrigel-coated plates (Corning) in mTeSR1 (STEMCELL Technologies) or StemFitN.AK03 (Ajinomoto) media.

Isolation and maintenance of haploid human ES cell lines

Following identification of haploid cells in human parthenogenetic ES cell lines at passages 6–7 by either metaphase spread analysis or sub-2c-cell sorting (see below and Extended Data Tables 1 and 2), haploid ES cell lines were established by sorting the 1c-cell population, with diploid cells serving as a reference. Haploid ES cell cultures were further maintained by enrichment rounds of 1c-cell sorting every 3–4 passages.

Metaphase spread analysis

For induction of mitotic arrest, growing cells were incubated for 40 min in the presence of 100 ng ml–1 colcemid (Biological Industries), added directly to the culture medium in a humidified incubator at 37 °C with 5% CO2. The cells were then trypsinized, centrifuged at 1,000 r.p.m. at room temperature and gently resuspended in 37 °C warmed hypotonic solution (2.8 mg ml−1 KCl and 2.5 mg ml–1 sodium citrate) followed by 20 min of incubation at 37 °C. Cells were fixed by addition of fixative solution (3:1 methanol:acetic acid) and incubation for 5 min at room temperature. Fixation was repeated at least three times following centrifugation and resuspension in fixative solution. Metaphase spreads were prepared on slides and stained using the standard G-banding technique. Karyotype integrity was determined according to the International System for Human Cytogenetic Nomenclature (ISCN) based on the observation of a normal karyotype in at least 80% of analysed metaphases (minimum of 20 metaphases per analysis).

Live ES cell sorting by DNA content

Cells were washed with phosphate buffered saline (PBS), dissociated using either TrypLE Select or TrypLE Express (Thermo Fisher Scientific) and stained with 10 μg ml−1 Hoechst 33342 (ref. 2) (Sigma-Aldrich) in human ES cell medium at 37 °C for 30 min. Following centrifugation, cells were resuspended in PBS containing 15% KSR and 10 μM ROCK inhibitor Y-27632, filtered through a 70-μm cell strainer (Corning) and sorted using the 405 nm laser in either BD FACSAria III or BD Influx (BD Biosciences). For continued growth, sorted cells were plated with fresh medium containing 10 μM ROCK inhibitor Y-27632 for 24 h. For comparative analyses, G1-phase cells were sorted from isogenic haploid-enriched and unsorted diploid cultures. Cells that had undergone diploidization relatively recently in culture (within 3 passages after haploid cell enrichment) were isolated by sorting the 4c peak in haploid-enriched cultures and compared with 4c diploid cells from unsorted diploid cultures. Note that haploid-enriched cultures also consist of a mixed 2c-cell population of G2/M-phase haploids and G1-phase diploids. Sorting purity was confirmed by rerunning a fraction of sorted samples through the instrument.

Flow cytometry

All DNA content profiles were generated based on flow cytometry with Hoechst 33342 staining. Haploid cell proportion was estimated based on the percentage of 1c cells and the relative contribution of G1 cells with regards to other phases of the cell cycle. Estimation of diploidization rate was based on the proportion of haploid cells between consecutive enrichment rounds as well as experimental analysis of h-pES10 diploidization kinetics throughout 7 passages (30 days) by analysing the DNA content of 2–3 replicates at each passage using flow cytometry with propidium iodide in methanol-fixed and RNase-treated cells. Diploidization rate was estimated by fitting the data to an exponential decay curve. For simultaneous flow cytometry analysis of DNA content and cell surface molecules, cells were washed, dissociated and incubated on ice for 30 min in the presence of 10 μg ml−1 Hoechst 33342 (Sigma-Aldrich) and either a conjugated antibody or a secondary antibody diluted 1:200 following a 60 min incubation with a primary antibody. For simultaneous flow cytometry analysis of DNA content and intracellular PDX1, dissociated cells were treated as described for immunofluorescence procedures, with Hoechst 33342 for DNA staining. Primary antibodies are detailed in Supplementary Table 1. In all flow cytometry procedures, samples were filtered through a 70-μm cell strainer (Corning Life Sciences) and analysed with either BD FACSAria III or BD Influx (BD Biosciences).

DNA fluorescence in situ hybridization

DNA FISH was performed as described elsewhere25 using probes for human chromosomes 2 and 4 and DNA staining with 4′,6-diamidino-2-phenylindole (DAPI). Haploidy and diploidy were respectively determined per nucleus based on single or double hybridization signals. ES cells subject to FISH were grown on Matrigel-coated plates in StemFitN.AK03 medium for several passages before analysis.

Alkaline phosphatase and immunofluorescence staining

Alkaline phosphatase staining was performed using the Leukocyte Alkaline Phosphatase Kit (Sigma-Aldrich). For immunofluorescence staining, samples were washed with PBS, fixed with 4% paraformaldehyde for 10 min, and permeabilized and blocked in blocking solution (0.1% Triton X-100 and 5% donkey serum in PBS). Cells were incubated with primary antibodies (detailed in Supplementary Table 1) and secondary antibodies diluted 1:500 in blocking solution, and DAPI was used for DNA staining. Cells were washed twice with PBS subsequently to fixation and each incubation step. Images were taken using Zeiss LSM 510 Meta Confocal Microscope. Centromere quantification was carried out by manually counting centromere foci across individual planes along the z axis. EdU staining was performed using the Click-iT EdU Alexa Fluor 488 Imaging Kit (Thermo Fisher Scientific). ES cells subject to centromere staining in Fig. 1e and Extended Data Fig. 1e were grown on Matrigel-coated plates in StemFitN.AK03 for several passages before analysis.

6-TG resistance screen

To generate a gene trap mutant library, 9 replicates of approximately 4 × 106 haploid pES10 cells (within one passage after 1c-cell enrichment) were co-transfected with 20 μg 5′-PTK-3′ gene trap vector26 and 20 μg pCyL43 piggyBac transposase plasmid27 using Bio-Rad Gene Pulser (suspended in 800 μL Opti-MEM, 4-mm cuvettes, 320 V, 250 μF), and replated on a 100 × 20 mm dish with DR3 MEFs and ROCK inhibitor Y-27632. Selection for insertions into expressed loci was carried out using 0.3 μg ml−1 puromycin starting 48 h post transfection, followed by pooling into a single library, represented by approximately 16,000 resistant colonies. Transfection with 5′-PTK-3′ only was used as a negative control. To screen for 6-TG-resistant mutants, the mutant library was grown in the presence of 6 μM 6-TG (Sigma-Aldrich) on DR4 MEFs for 18 days, during which 6 resistant colonies were independently isolated and characterized. Analysis of a resistant clone showed persistence of haploid cells. Genomic DNA was extracted (NucleoSpin Tissue Kit, MACHEREY-NAGEL) and insertion sites were detected using splinkerette PCR as described previously28, followed by PCR product purification and Sanger sequencing (ABI PRISM 3730xl DNA Analyzer (Applied Biosystems)). Sequences were mapped to the human genome (GRCh38/hg38) using UCSC BLAT search tool.

Isolation of total DNA and RNA

Total DNA was isolated using the NucleoSpin Tissue Kit (MACHEREY-NAGEL). Total RNA was isolated using Qiagen RNeasy Kits according to the manufacturer’s protocols. To determine total RNA levels per cell, haploid and diploid cells were isolated from the same cultures by sorting the 1c (haploid G1) and 4c (diploid G2/M) populations, respectively. Following growth for 2 passages, cells were harvested and counted, and RNA was isolated from triplicates of 400,000 cells from each cell line and ploidy state (pES10 and pES12, haploid and diploid; 12 samples in total). RNA amounts were quantified using NanoDrop.

Genome integrity analysis

Copy number variation (CNV) analysis was carried out on DNA samples of G1-sorted haploid and diploid pES10 and pES12 cells (see Supplementary Table 2) using Infinium Omni2.5Exome-8 BeadChip single nucleotide polymorphism (SNP) arrays (Illumina) following the manufacturer’s protocols. Raw data were processed using Genome Studio Genotyping Module (Illumina) to obtain log R ratios values for analysis using R statistical programming language. As expected, diploid pES10 and pES12 cells were homozygous across all chromosomes.

RNA sequencing

For a detailed list of samples analysed by RNA-seq, see Supplementary Table 3. Total RNA samples (200 ng–1 μg, RNA integrity number (RIN) >9) were enriched for mRNAs by pulldown of poly(A)+ RNA. RNA-seq libraries were prepared using the TruSeq RNA Library Prep Kit v2 (Illumina) according to the manufacturer’s protocol and sequenced using Illumina NextSeq 500 to generate 85 bp single-end reads.

Transcriptome analysis

RNA-seq reads were aligned to the human reference genome (GRCh37/hg19) using TopHat (version 2.0.8b) allowing 5 mismatches. Reads per kilobase per million fragments mapped (RPKM) values were quantified using Cuffquant and normalized using Cuffnorm in Cufflinks (version 2.1.1) to generate relative gene expression levels. Hierarchical clustering analyses were performed on RPKM values using Pearson correlation and average linkage. Analysis of differential gene expression relative to total RNA in haploid and diploid human ES cells (n = 4 in each group) was carried out by two complementary strategies, as follows: first, we used Cuffdiff with default parameters, considering differences of greater than twofold with FDR <0.05 as significant; second, to identify possibly subtle yet consistent transcriptional differences, we tested for genes whose minimal expression levels across all replicates of a certain group were higher than their maximal expression level across all replicates of the other group. Statistical significance was then determined by two-tailed unpaired Student’s t-test. Functional annotation enrichment analysis was done by DAVID (using the Benjamini method to determine statistical significance). Imprinting analyses included 75 human imprinted genes (http://www.geneimprint.com/), listed in Supplementary Table 4. RNA-seq data from control ES cell line NYSCF1 were published elsewhere29 (GEO accession number GSE61657). Genome-wide gene expression moving median plots were generated using the R package zoo (version 1.7–12) after removal of genes that were not expressed in the averaged reference diploid sample by flooring to 1 and setting an expression threshold of above 1. RNA-seq data from different tissues were retrieved from the Genotype-Tissue Expression (GTEx) portal (http://www.gtexportal.org/)30. Colour-coded scales in Fig. 4d correspond to gene expression levels relative to the mean across tissues (left scale) and across each set of ES cell duplicate and EB sample (right scale). Expression microarray analysis was performed as previously31 by using Affymetrix Human Gene 1.0 ST arrays.

DNA methylation analysis

DNA methylation analysis was performed on genomic DNA from the samples detailed in Supplementary Table 2 using Infinium HumanMethylation450 BeachChips (Illumina) following the Infinium HD Methylation Protocol as described previously29. DNA methylation data from control ES cell line NYSCF1 were published before29 (GEO accession number GSE61657). Data were processed and normalized by using subset-quantile within array normalization (SWAN) and adjusted for batch effects using the R package ChAMP (version 1.4.0). DNA methylation levels at CpG sites associated with pluripotency-specific genes and iDMRs were analysed as described before29. For analysis of DNA methylation levels on the X chromosome, probes with average β values of less than 0.4 were filtered out. DMR analysis was facilitated by the lasso function in ChAMP using default settings. DMRs were then assigned to genes by proximity and analysed for functional annotation enrichment using DAVID (using the Benjamini method to determine statistical significance).

Cell size analysis

Following sorting of haploid and diploid cell populations in G1, the diameter (2r) of viable single cells was measured by Countess Automated Cell Counter (Invitrogen) and their surface area and volume were calculated as 4πr2 and 4/3πr3, respectively. Analysis included 7, 4, 8 and 4 technical replicates for 1n pES10, 1n pES12, 2n pES10 and 2n pES12, respectively.

Mitochondrial DNA abundance analysis

Relative mtDNA abundance was analysed by quantitative PCR (qPCR) by using primers for the mitochondrial gene MT-ND2 (forward primer: 5′–TGTTGGTTATACCCTTCCCGTACTA–3′; reverse primer: 5′–CCTGCAAAGATGGTAGAGTAGATGA–3′) and normalization to nuclear DNA by using primers for the nuclear gene BECN1 (forward primer: 5′–CCCTCATCACAGGGCTCTCTCCA–3′; reverse primer: 5′–GGGACTGTAGGCTGGGAACTATGC–3′), as described elsewhere32. Analysis was performed using Applied Biosystems 7300 Real-Time PCR System with PerfeCTa SYBR Green FastMix (Quanta Biosciences). Analysis included all G1-sorted samples detailed in Supplementary Table 2 (n = 4 for each group, with two biological replicates for each cell line).

Embryoid body differentiation

EB differentiation was carried out by detaching ES cell colonies with Trypsin solution A without EDTA (Biological Industries), followed by resuspension and further culture of cell aggregates in human ES cell medium without bFGF on low attachment plates. Differentiation of haploid ES cells was initiated within 2 passages after 1c-cell enrichment. After 21 days, EB RNA was extracted from unsorted and/or sorted EB cells in G1 following dissociation and staining with 10 μg ml−1 Hoechst 33342 (Sigma-Aldrich) at 37 °C for 30 min. Metaphase spread analysis was performed on dissociated EB cells plated on 0.2% gelatin and expanded in human ES cell medium without bFGF.

Differentiation into neural progenitor cells

NCAM1-positive ES cell-derived neural progenitor cells were obtained using a 10-days protocol for efficient neural differentiation33 with slight modification34. Differentiation was initiated within 2 passages after 1c-cell enrichment. RNA was extracted from sorted haploid NCAM1-positive cells in G1 by co-staining with Hoechst 33342 and an anti-human NCAM-1/CD56 primary antibody (see Supplementary Table 1) and a Cy3-conjugated secondary antibody (Jackson Immunoresearch Laboratories) diluted 1:200.

Neuronal differentiation

Differentiation into neurons was carried out by following a published protocol35 based on synergistic inhibition of SMAD signalling36 with modification, as follows: differentiation was initiated within 2 passages after 1c-cell enrichment with fully confluent ES cells cultured on Matrigel-coated plates in mTeSR1 by replacing the medium with human ES cell medium without bFGF, containing 10 μM SB431542 (Selleckchem) and 2.5 μM LDN-193189 (Stemgent) for 4 days. Subsequently, cells were kept in N2 medium35 supplemented with 10 μM SB431542 and 2.5 μM LDN-193189 for an additional 4 days, followed by 2 days in N2 medium supplemented with B-27 (Thermo Fisher Scientific) and 10 μM DAPT (Stemgent). The cells were then dissociated and replated on 0.01% poly-l-ornithine coated (Sigma-Aldrich) and laminin coated (4 μg ml−1, Thermo Fisher Scientific) plates in the presence of 10 μM ROCK inhibitor Y-27632 (Selleckchem), and further cultured in the same medium without Y-27632 for the next 4 days. Neuronal cultures were maintained in N2 medium supplemented with B-27 and 20 ng ml–1 BDNF (R&D) until analysis by immunostaining and FISH on day 20.

Cardiomyocyte differentiation

80–90% confluent ES cells grown on Matrigel-coated plates in mTeSR1 were subject to an 11-day regimen37 based on consecutive GSK3 and WNT inhibition with CHIR99021 and IWP-2 (Selleckchem), respectively. Differentiation was initiated within 2 passages after 1c-cell enrichment. On day 11 of differentiation, 1c cells were sorted and plated for immunostaining.

Differentiation towards the pancreatic lineage

The protocol used here was developed based on several recent publications38,39,40. ES cells grown in feeder-free conditions were differentiated into definitive endoderm by using STEMdiff Definitive Endoderm Kit (StemCell Technologies) for 3–4 days. Subsequent specification was achieved by a step-wise protocol involving treatment with recombinant human KGF/FGF7 (R&D Systems), LDN-193189 (Stemgent), KAAD-cyclopamine (Stemgent) and retinoic acid (Stemgent). On days 8–11, EGF (R&D System) was used to induce pancreatic cells. Differentiation was initiated within as few as 2 passages after 1c-cell enrichment.

Teratoma formation assay

All experimental procedures in animals were approved by the ethics committee of the Hebrew University. ES cells were trypsinized and approximately 2 × 106 cells were resuspended in 100 μl human ES cell medium and 100 μl Matrigel (BD Biosciences), followed by subcutaneous injection into NOD-SCID Il2rg−/− immunodeficient mice (Jackson Laboratory). 8–12 weeks after injection tumours were dissected and subjected to further analysis. Histological slides were prepared from tumour slices cryopreserved in O.C.T. compound (Sakura Finetek) using Leica CM1850 cryostat (Leica Biosystems, 10-μm sections), followed by immunostaining, haematoxylin and eosin staining or FISH analysis. Flow cytometry with Hoechst 33342 staining was performed on dissociated cells from freshly dissected tumours.