Stabilization of mouse haploid embryonic stem cells with combined kinase and signal modulation

Mammalian haploid embryonic stem cells (haESCs) provide new possibilities for large-scale genetic screens because they bear only one copy of each chromosome. However, haESCs are prone to spontaneous diploidization through unknown mechanisms. Here, we report that a small molecule combination could restrain mouse haESCs from diploidization by impeding exit from naïve pluripotency and by shortening the S-G2/M phases. Combined with 2i and PD166285, our chemical cocktail could maintain haESCs in the haploid state for at least five weeks without fluorescence-activated cell sorting (FACS) enrichment of haploid cells. Taken together, we established an effective chemical approach for long-term maintenance of haESCs, and highlighted that proper cell cycle progression was critical for the maintenance of haploid state.


RDF inhibits self-diploidization of haESCs.
To effectively maintain the haploid state of haESCs, we adopted a chemical screen strategy to identify small molecules that could regulate diploidization of haESCs. We first tested a group of chemicals that are involved in activating or inhibiting certain signaling pathways ( Fig. 2A). However, all tested chemicals alone had no effect on the self-diploidization of haESCs ( Supplementary Fig. S1C). Then whether chemical combinations could inhibit self-diploidization was investigated. Indeed, a six-chemical combination, referred to as VCRDFG (consisting of V, VPA, HDACs inhibitor; C, CHIR99021, GSK-3 kinases inhibitor; R, Repsox, inhibitor of TGF-β pathways; D, DMH1, inhibitor of BMP4 pathway; F, Forskolin, adenylate cyclase activator; and G, Compound E, γ-secretase inhibitor) significantly inhibited both AG-and PG-haESCs from self-diploidization, as revealed by the higher ratio of 1 N cells in the presence of VCRDFG (Fig. 2B). In order to identify the essential chemicals required for haploidy stabilization, we further examined different combinations of the six chemicals in VCRDFG. One three-chemical combination (Repsox, DMH1 and Forskolin, termed RDF) was the minimum combination that exhibited similar haploidy stabilization effect as VCRDFG (Fig. 2C,D). In contrast, none of the two-chemical combinations from these three chemicals were sufficient to suppress the diploidization of haESCs (Fig. 2C).
To determine whether RDF impairs the developmental potential of haESCs, we injected wild-type blastocysts with RDF-treated haploid cells that were FACS-purified either from a HG165-derived cell line expressing a RFP reporter or from a AGH-OG-3-derived cell line carrying an Oct4-EGFP transgene 6 . Both RDF-treated cells lines contributed to different tissues ( Supplementary Fig. S1D), and more importantly germline cells in the gonads of embryonic day 13.5 (E13.5) embryos (Fig. 2E), demonstrating the developmental potential of haESCs was not impaired by RDF treatment. Our findings therefore demonstrated that RDF treatment could effectively suppress the diploidization of haESCs, without compromising their pluripotency.

RDF regulates cell cycle transition of haESCs.
Previous studies have shown that the progression of cell cycle plays an important role in the exit from naïve pluripotent state of ESCs 23,25,[35][36][37] . Therefore, we attempted to determine whether RDF could modulate the cell cycle profile of haESCs. Consistent with our RNA-seq data (Fig. 3C,D), RDF notably shortened the doubling time of haploid ESCs (Fig. 5A), without affecting the apoptosis of haESCs ( Supplementary Fig. S3A,B). Interestingly, only the length of S-G2\M phases was shortened by RDF, while the G1-phase length maintained (Fig. 5B,C). In both AG-and PG-haESCs, RDF treatment significantly increased the percentage of S-phase cells and decreased the percentage of G2/M-phase cells, without influencing the percentage of G1-phase cells ( Fig. 5D-G). On the contrary, RDF did not alter the cell cycle of haESC-derived diploid ESCs or normal diploid ESCs (Supplementary Fig. S3C-F). These data demonstrated that RDF facilitated haESC proliferation by accelerating S-G2\M-phase progression.
RNA-seq analyses also showed that RDF regulated the expressions of S-phase genes (Mki67, Ccng2, Orc1, E2f2, Gadd45g, Gadd45a, Cdc6, Rrm2) and G2/M-phase genes (Ccna1, Rad21, Rb1, Nek1, Cenpf, Myt1) (Fig. 3E). We further carried out Q-PCR analyses to validate the expression change of cell cycle regulators. Indeed, RDF treatment significantly reduced the expression of Myt1, a negative regulatory gene of G2/M-phase transition, and increased the expression of Cdc6, an essential regulator of DNA replication in S phase ( Supplementary  Fig. S4A,B,C). Although RDF and 2i had comparable effect on promoting pluripotent gene expressions ( Supplementary Fig. S2G), only RDF significantly increased the expression level of Cdc6 ( Supplementary  Fig. S4D,E), suggesting that RDF but not 2i might stabilize haESCs through regulating cell cycle progression. Since it has been reported that promoter DNA methylation of key regulatory genes could critically regulate cell cycle progression and ESC pluripotency 38-40 , we next examined whether RDF treatment could change the promoter DNA methylation of differentially expressed cell cycle genes. Among the genes we examined, RDF treatment resulted in a decrease of DNA methylation at the Cdc6 promoter but an increase of DNA methylation at the Fgf5 promoter ( Supplementary Fig. S4F), consistent with their expression changes ( We subsequently examined the effectiveness of RDF in the presence of 2i. Surprisingly, RDF decreased cell proliferation of haESCs in 2i condition (Supplementary Fig. S6A-C), leading to significant cell death after prolonged culture. However, when both RDF and PD166285 were added to the 2i culture medium, the cells could proliferate normally probably due to the growth promotion effect of PD166285 ( Supplementary Fig. S6A-C). More importantly, the combination of RDF and PD166285 functioned better than RDF alone in stabilizing haploidy ( Supplementary Fig. S6D). Therefore, RDF/PD166285/2i appeared to be an ideal culture condition for haESCs. To further evaluate the effectiveness of RDF/PD166285/2i, 1N cells freshly purified from haESCs were immediately cultured in 2i medium supplied with different chemicals for a relatively long period (Fig. 6A). We found that haESCs treated with RDF/PD166285/2i exhibited a very high ratio of haploid 1N cells and an evidently low ratio of diploid 4 N cells (Fig. 6B and F). Karyotype analyses also showed that haESCs treated with RDF/PD166285/2i displayed a higher percentage of haploid karyotype cells compared with haESCs treated with PD166285/2i or 2i alone (Fig. 6C,G; Supplementary Fig. S6E), consistent with FACS profiles. The haploid state could be well maintained for five weeks after RDF/PD166285/2i treatment without FACS enrichment (Fig. 6D,E). In addition, single colony assay revealed that the survived colonies after RDF/PD166285/2i treatment for 35 days contained a significant higher percentage of haploid 1N cells, compared with the survived colonies cultured in 2i medium supplied with or without PD166285 (Fig. 6H). Interestingly, RDF/PD166285/2i not only increased Cdc6 expression, but also upregulated the expression of Cdk2, which is essential in S-and G2-phase ( Supplementary Fig. S6F). Because Wee1 kinase has been reported to inhibit Cdk2 41 , RDF/PD166285/2i treatment might up-regulate Cdk2 through inhibiting Wee1 kinase by PD166285. Importantly, haESCs cultured in RDF/PD166285/2i condition could efficiently contribute to somatic and germline cells in E13.5 chimeric embryos ( Fig. 6I and Supplementary Fig. S6G). Thus, our data suggested that the RDF/PD166285/2i combination was suitable for long-term maintenance of haESCs.  Fig. 1F, Fig. 1G, and Fig. 5B, respectively. The haESCs used were the HG165 line, and the diploid ESCs were derived from HG165 haESCs. Data are shown as means ± sem. * P < 0.05, ** P < 0.

Discussion
Abnormal cell cycle progression, such as cell cycle arrest, induces ESC differentiation or apoptosis 42,43 . Compared to most other cells, ESCs exhibit a shortened cell cycle, characterized by a short G1-phase and a high proportion of S-phase cells. However, the lengths of G2-and M-phases are comparable between ESCs and other cells [44][45][46] . Our results showed that haESCs exhibited a prolonged G2/M phase compared to diploid ESCs, and RDF could shorten the duration of S-G2/M phases from 19 hours to 13 hours, suggesting that RDF might inhibit haESC diploidization by reducing the errors in S-G2/M phases and shortening the duration of S-G2/M phases. This is consistent with our recent report that prolonging prometaphase/metaphase of M phase by nocodazole or STLC treatment increased the diploidization rate of haESCs, whereas shortening prometaphase/metaphase by Aurora B overexpression significantly decreased self-diploidization rate 47 .
Environmental cues can trigger different responses in ESC cultures, leading to distinct gene expression networks under different culture conditions 48,49 . In this study, we found that RDF/PD166285 could facilitate long-term maintenance of haESCs in the 2i culture medium, but not in the ES medium, indicating that the culture condition is also critical in stabilizing the haploid state. It has been reported that PD166285 not only targets Wee1 kinase, but also inhibits the tyrosine kinase c-Src. While c-Src is involved in stimulating mouse ESC proliferation in the ES medium 50 , its activation promotes differentiation in the 2i medium 51 . Therefore, by inhibiting c-Src, PD166285 might block RDF's proliferation promotion effect on haESCs in the ES medium, but functions together with RDF in the 2i culture medium to promote cell proliferation and naïve pluripotency, resulting in an effective stabilization of haESC haploidy.
Self-diploidization is an intrinsic feature of haESCs, and the diploidization rate varies among different cell lines. For example, PG-haESC line 319 showed a higher diploidization rate than two AG-haESC lines (AGH-OG-3 and HG165). And among the AG-haESC lines used in this study, the A7 line which was derived from an old mouse exhibited higher diploidization rate than the other two lines derived from young mice (AGH-OG-3 and HG165), suggesting that the age of mice from which haESCs were derived might also affect self-diploidization rate. Thus, it is important to use multiple independent haESC lines to study the mechanism of self-diploidization. Using multiple haESC lines, we found that both AG-and PG-haploid ESCs exhibited lower proliferation rates than diploid ESCs. Interestingly, such proliferation difference between haploid and diploid ESCs was not observed in previous reports 1,7 , therefore examining the haESC lines used in those studies might provide further insight into the mechanism of haESC self-diploidization.
ESC differentiation provides a competitive strategy for tissue regeneration and cell therapy 52,53 . Recently, it has been reported that mouse ESCs could generate functional haploid gametes (i.e., haploid spermatid-like cells and female mature oocytes) in vitro 54,55 . However, such in vitro gametogenesis of diploid ESCs required gonadal somatic cells derived from embryos, limiting the application of this approach in human clinical research. Given the haploid nature of haESCs, it is highly possible that under appropriate conditions haESCs could differentiate into haploid gametes efficiently without the need of embryo-derived gonadal somatic cells. Despite their unstable nature, haESCs also provide new possibilities in many aspects, including facilitating large-scale genetic screen and understanding genome evolution and functions. Thus, our study not only reports an effective chemical cocktail that supports long-term maintenance of haESC haploidy, but also promotes future applications of haESCs in many exciting research areas.

Methods
Derivation of haESCs. All animal methods were performed in accordance with the guidelines of the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. All experimental procedures were authorized by the Animal Care and Use Committee of the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. Mouse haESCs were established as previously described 1,2,4,6 . All AG-and PG-haESCs used in this study were established from C57BL/6 mice. Androgenetic A7-haploid cell line was derived from elderly mouse, while androgenetic haploid cell lines AGH-OG-3 and HG165 were generated from young mice. Androgenetic AGH-OG-3 and parthenogenetic 319 haESCs were established from OCT4-EGFP transgene mice. For chimera formation experiments, HG165 haESCs were transfected with a piggyBac transposon vector to stably express RFP.
Flow cytometry analysis and cell sorting. HaESCs were trypsinized, and stained with Hoechst33342 for 30 mins at 37 °C. For cell sorting, haploid G1-phase and diploid G2\M-phase cells were purified on BD FACSAria TM II machine. For flow cytometry, the DNA contents of haESCs with or without chemicals treatment were detected on BD LSR II machine. Cell apoptosis assay was carried out following the manufacturer's protocol (Biolegend, 640930). Data were analyzed with the FlowJo software (Tree Star).
Real-time quantitative PCR. Total RNA was isolated from haESCs with TRIzol (Sigma) reagent, and then reverse transcribed with M-MLV Reverse Transcriptase (Promega) following manufacturers' instructions. Real-time quantitative PCR was performed on Stratagene MX3000P (Agilent Technologies) with 2 × JumpStart TaqReadyMix (Sigma) and EvaGreen Dye (Biotium). The relative expression values were normalized to the internal control Gapdh. Primer sequences are listed in Table 1.
Clonogenicity assay and alkaline phosphatase staining. For clonogenicity assay, 600 haESCs were plated in one well of a 6-well plate coated with gelatin, and cultured in ES medium added with DMSO or RDF for 3 days. These cells were fixed with 4% paraformaldehyde solution and then stained for alkaline phosphatase activity. The procedures of alkaline phosphatase staining were performed as described previously 56  Immunofluorescence staining. HaESCs cultured in Chamber Slides were fixed by 4% paraformaldehyde solution for 15 mins at room temperature. Blocking and permeabilization buffer (1% BSA, 0.5% Triton X-100 in PBS) was used to permeabilize the cells for 1 hour at room temperature. Cells were then incubated overnight with Oct-3/4 (sc-5279; Santa Cruz) and Nanog (sc-33760; Santa Cruz) antibodies at 4 °C, and then incubated with fluorescent conjugated secondary antibodies for 1 hour at room temperature. Nuclei were counterstained with DAPI. Images were captured with Zeiss LSM 710 Confocal Scanning Microscope.
Metaphase Chromosome Spread Analysis. HaESCs were cultured in 6-well plates for 2 days prior to the preparation of chromosome spreads. Cells were treated with 0.1 μg/ml Colcemid at 37 °C for 2 hours and then trypsinized. Cells were incubated with hypotonic solution (0.56% KCl) for 6 mins and fixed by 3:1 methanol:acetic acid for 10 mins at room temperature. Fixation was repeated three times following centrifugation and resuspension in fixative solution. Meataphase spreads were prepared on slides and stained with DAPI.
RNA-seq library preparation and data analysis. Total RNA was purified with an RNeasy Plus Mini kit (Qiagen). 150 ng of purified RNA was subjected to mRNA isolation and library preparation with a VAHTS Stranded mRNA-seq Library Prep Kit for Illumina (Vazyme) following manufacturer's instructions. Libraries were pooled and sequenced on an Illumina HISEQ 2500. RNA-seq data analysis was performed with Tophat and Cufflinks using the UCSC mm9 annotation. The Gene Expression Omnibus accession number for the RNA-seq data generated in this study is GSE104519.
DNA methylation analysis. Genomic DNA was extracted from haESCs with TIANamp Genomic DNA Kit following manufacturer's instructions. Promoter CpG methylation of primed pluripotent and cell cycle genes was analyzed as previously described 57 . The sequences of bisulfite PCR primers are provided in Table 2.
Statistical Analysis. Quantitative results were presented as mean ± SEM. One-way ANOVA or Student's t test was used for multiple-or two-sample-comparisons, respectively.