CHD1 acts via the Hmgpi pathway to regulate mouse early embryogenesis

During preimplantation development in the mouse, the first important gene expression event is zygotic gene activation (ZGA), the first transcription from the newly formed zygotic genome, which occurs between the late 1-cell and 2-cell stages and is required for normal development (Hamatani et al., 2004; Levey et al., 1977; Li et al., 2010; Minami et al., 2007; Schultz, 1993; Wang and Dey, 2006; Warner and Versteegh, 1974). One-cell embryos treated with α-amanitin, an RNA polymerase II inhibitor, arrest development at the 2-cell stage because ZGA is suppressed (Levey et al., 1977; Warner and Versteegh, 1974). The second transcriptional event is mid-preimplantation gene activation (MGA), which occurs between the 4-cell and 8-cell stages. During this period, genes required for cell fate specification, such as the transcription factors Pou5f1 (also known as Oct4), Nanog and Cdx2, are expressed; these genes are key regulators governing formation of the inner cell mass (ICM) and trophectoderm (TE) (Hamatani et al., 2004; Nichols et al., 1998; Niwa et al., 2005; Strumpf et al., 2005; Wang and Dey, 2006; Yoshikawa et al., 2006). Thus, to understand cell fate specification, we must discern the regulatory mechanisms underlying expression of lineage-specific marker genes such as Pou5f1, Nanog and Cdx2 at MGA. A previous study showed that a deficiency of Tead4, a transcription factor expressed during ZGA in mouse preimplantation embryos, led to a failure of cell fate specification due to suppression of Cdx2 expression at MGA and developmental arrest at the morula stage (Yagi et al., 2007). Other experiments have also identified genes that are involved in cell fate specification by regulating the expression of Pou5f1, Nanog or Cdx2 (Do et al., 2013; Elling et al., 2006; Home et al., 2012; Strumpf et al., 2005; Wang et al., 2010; Zhang et al., 2006, 2013).

During preimplantation development in mammals, dynamic changes occur in chromatin structure (Abdalla et al., 2009; Albert and Helin, 2010; Burton and Torres-Padilla, 2010; Corry et al., 2009; Morgan et al., 2005; Rasmussen and Corry, 2010; Shi and Wu, 2009). Previous studies showed that suppression of Brg1 (Smarca4 – Mouse Genome Informatics Database), a subclass of switch/sucrose non-fermentable (SWI/SNF) ATP-dependent chromatin remodelers, in mouse preimplantation embryos causes widespread expression of Pou5f1 in the TE and leads to early embryonic death (Kidder et al., 2009; Wang et al., 2010). This suggests that chromatin-remodeling factors play important roles in mouse cell fate specification. CHD1 (chromodomain helicase DNA binding protein 1) belongs to the family of ATPase-dependent chromatin remodeling factors (Woodage et al., 1997). It recognizes the trimethylated lysine 4 of histone 3 (H3K4me3) (Sims et al., 2005) and has been implicated in transcriptional activation in yeast (Simic et al., 2003), Drosophila (Stokes et al., 1996) and mammalian cells (Sims et al., 2007). CHD1 is involved in pre-mRNA maturation (Sims et al., 2007), the maintenance of mouse embryonic stem cell (ESC) pluripotency (Gaspar-Maia et al., 2009), the maintenance of optimal transcriptional output in mouse ESCs and epiblast cells (Guzman-Ayala et al., 2015), and incorporation of the variant histone H3.3 into paternal pronuclear chromatin at fertilization in Drosophila embryos (Konev et al., 2007). However, the function(s) of CHD1 in mammalian preimplantation development is not fully understood.

Here, we investigated whether CHD1 functions during preimplantation development in the mouse. We observed that Chd1 expression started to increase at the 2-cell stage, CHD1 was intensely localized in nuclei starting at the 2-cell stage and loss of CHD1 function by siRNA treatment led to embryonic lethality after implantation, due to suppression of Pou5f1, Nanog and Cdx2 expression at MGA. Additionally, the expression of Hmgpi and Klf5, which regulate the expression of Pou5f1, Nanog and Cdx2 during mouse preimplantation development (Ema et al., 2008; Lin et al., 2010; Yamada et al., 2010), was dramatically suppressed beginning at ZGA. However, Hmgpi mRNA microinjection in Chd1-knockdown embryos (Chd1-knockdown-Hmgpi-rescue embryos) rescued Pou5f1, Nanog and Cdx2 expression and postimplantation embryo development. Based on these results, we propose that CHD1 has important roles in the development of pre- and postimplantation embryos via the activation of Hmgpi expression at ZGA.

To investigate the roles of CHD1 during preimplantation development in the mouse, Chd1 expression and CHD1 localization were examined in preimplantation embryos (Fig. 1). Chd1 mRNA was expressed from the 1-cell to the blastocyst stage. Specifically, the expression increased at the 2-cell stage, peaked at the 8-cell stage and then dramatically decreased. CHD1 was localized in the nuclei of all blastomeres during preimplantation development and its staining intensity increased from the 2-cell stage onward.

Although the results of qRT-PCR and immunostaining revealed that the amounts of Chd1 mRNA and protein were dramatically reduced after the 2-cell stage when siChd1 was injected at the 1-cell stage (Fig. 2), Chd1-knockdown embryos exhibited morphologically normal growth until the blastocyst stage, and Chd1-knockdown blastocysts (E4.5) had normal numbers of cells in both ICM and TE compared with controls (Fig. 3). In addition, there were no differences in hatching percentages between control and Chd1-knockdown embryos (Fig. 3). However, outgrowth experiments showed that the percentage of ICM-derived colony formation was dramatically decreased in Chd1-knockdown embryos (Fig. 3). Furthermore, the litter size after embryo transfer was significantly reduced in Chd1-knockdown embryos (Table 1).

CHD1 is required for optimal transcriptional output in mouse ESCs (Guzman-Ayala et al., 2015). Therefore, to investigate the cause of litter size reduction in Chd1-knockdown embryos, we examined the effect of Chd1-knockdown on global gene expression in preimplantation embryos by BrUTP incorporation assay. The results showed that the reduction of global transcriptional activity does not occur in Chd1-knockdown embryos at the 2- and 4-cell stages (Fig. 4). To investigate the cause of litter size reduction in Chd1-knockdown embryos further, we assessed expression of the lineage-specific markers Pou5f1, Nanog and Cdx2 at the mRNA and protein levels by qRT-PCR and immunostaining, respectively (Fig. 5). In control embryos, Pou5f1 mRNA levels increased dramatically at the 4- and 8-cell stages, but subsequently decreased. By contrast, Pou5f1 mRNA expression was suppressed throughout these stages in Chd1-knockdown embryos. In control embryos, Nanog mRNA levels increased dramatically at the 4- and 8-cell stages, peaked at the morula stage and then decreased. By contrast, Nanog mRNA expression was suppressed throughout these stages in Chd1-knockdown embryos. With respect to Cdx2, in control embryos mRNA was first detected at the 8-cell stage and then gradually increased. Interestingly, Cdx2 mRNA expression was suppressed during these stages in Chd1-knockdown embryos. Furthermore, in Chd1-knockdown embryos, immunofluorescence detection of OCT4 (encoded by Pou5f1), NANOG and CDX2 revealed that maternal OCT4 and NANOG proteins were maintained until the 4-cell stage, whereas for newly synthesized proteins, OCT4 and NANOG were reduced starting at the 8-cell stage, and CDX2 was reduced starting at the morula stage, respectively. The localization of OCT4 in the ICM, NANOG in the epiblast and CDX2 in the TE did not change in Chd1-knockdown embryos (Fig. 5).

The mechanisms that regulate Pou5f1, Nanog and Cdx2 expression during mouse preimplantation development have gradually been elucidated. However, the only factors identified to date that regulate the expression of Pou5f1 and Nanog in ICM and Cdx2 in TE are HMGPI and KLF5, both of which are transcription factors (Ema et al., 2008; Lin et al., 2010; Yamada et al., 2010). Therefore, we hypothesized that CHD1 regulates the expression of Pou5f1 and Nanog in the ICM and Cdx2 in the TE via activation of Hmgpi and Klf5 expression. To investigate the effects of CHD1 on Hmgpi and Klf5 expression, the levels of Hmgpi and Klf5 mRNA and protein in Chd1-knockdown embryos were examined by qRT-PCR and immunostaining, respectively (Fig. 6). With respect to Hmgpi, in control embryos mRNA was first detected at the late 2-cell stage, peaked at the 4-cell stage and then gradually decreased. Conversely, Hmgpi mRNA expression was suppressed during these stages in Chd1-knockdown embryos. Immunofluorescent (IF) detection of HMGPI protein determined that its levels were also reduced from the 4-cell stage onward in Chd1-knockdown embryos (Fig. 6). With respect to Klf5, in control embryos mRNA was also first detected at the late 2-cell stage, peaked at the 8-cell stage and was maintained through the blastocyst stage. On the other hand, in Chd1-knockdown embryos, expression was suppressed at the 4- and 8-cell stages but recovered after the morula stage. IF detection of KLF5 demonstrated that the amount of KLF5 protein was remarkably reduced at the 4- and 8-cell stages; however, KLF5 protein levels gradually recovered after the morula stage in Chd1-knockdown embryos (Fig. 6).

To investigate whether Hmgpi regulates the expression of Pou5f1, Nanog and Cdx2 under the control of CHD1, we rescued HMGPI expression by injecting Hmgpi mRNA into Chd1-knockdown embryos. Outgrowth and embryo transfer experiments showed that normal ICM-derived colony formation and litter size were recovered in Chd1-knockdown-Hmgpi-rescue embryos (Fig. 7, Table 2). qRT-PCR results indicated that Chd1 mRNA levels were not restored in Chd1-knockdown-Hmgpi-rescue embryos at any stage (Fig. 7). Nonetheless, qRT-PCR and immunostaining results showed that Pou5f1 and Nanog mRNA levels recovered, starting at the 4-cell stage onward, and that OCT4 and NANOG protein expression levels recovered starting at the 8-cell stage onward (because maternal OCT4 and NANOG proteins remain until the 4-cell stage in Chd1-knockdown-Hmgpi-rescue embryos). Furthermore, Cdx2 mRNA levels were restored starting at the 8-cell stage onward, and CDX2 protein levels were recovered starting at the morula stage onward in Chd1-knockdown-Hmgpi-rescue embryos (Fig. 7). Meanwhile, qRT-PCR and immunostaining of Klf5 revealed that HMGPI rescue had no effect on the expression of Klf5 mRNA or protein (see supplementary material Fig. S1).

CHD1, a chromatin remodeling factor, recognizes H3K4me3 (Sims et al., 2005), facilitates pre-mRNA maturation (Sims et al., 2007) and is required for the maintenance of mouse ESC pluripotency (Fazzio and Panning, 2010; Gaspar-Maia et al., 2009). CHD1 is also necessary at fertilization in Drosophila embryos for the incorporation of the variant histone H3.3 into paternal pronuclear chromatin in the absence of transcription (Konev et al., 2007). Recent work showed that Chd1-knockout embryos undergo developmental arrest at E6.5 due to a failure to maintain the epiblast (Guzman-Ayala et al., 2015). However, because Chd1-knockout embryos were produced from heterozygous intercrosses, maternal Chd1 mRNA and protein are not suppressed at ZGA in Chd1-knockout embryos. Here, we showed that in Chd1-knockdown embryos, expression of Chd1 mRNA and proteins were suppressed at ZGA, and Chd1 knockdown led to embryonic lethality after implantation. Additionally, we showed that suppression of zygotic Pou5f1, Nanog and Cdx2 expression continued through the blastocyst stage via suppression of Hmgpi. However, we also showed that CHD1 was not required for global transcriptional activity at ZGA and MGA, suggesting that CHD1 is involved in selective regulation of gene expression that governs normal embryogenesis and the maintenance of pluripotency during mouse preimplantation development.

The expression of Pou5f1, Nanog and Cdx2 at MGA is regulated by Hmgpi and Klf5, both of which begin to be expressed at ZGA (Ema et al., 2008; Lin et al., 2010; Yamada et al., 2010). In embryos lacking functional Hmgpi or Klf5, developmental arrest occurs during peri-implantation development due to suppression of Pou5f1, Nanog and Cdx2 expression (Ema et al., 2008; Lin et al., 2010; Yamada et al., 2010), suggesting that Hmgpi and Klf5 regulate the initiation of cell fate specification. However, the detailed mechanisms underlying regulation of Pou5f1, Nanog and Cdx2 by HMGPI and KLF5 are unknown. The results of this study showed that in Chd1-knockdown embryos, suppression of Hmgpi expression continued until the blastocyst stage, whereas suppression of Klf5 expression was relieved by the morula stage, when Klf5 expression recovered spontaneously. Accordingly, we hypothesized that suppression of Hmgpi through the blastocyst accounts for suppression of Pou5f1, Nanog and Cdx2 expression, normal ICM-derived colony formation and the numbers of live offspring. Thus, we investigated the effects of Hmgpi-rescue in Chd1-knockdown embryos on Pou5f1, Nanog and Cdx2 expression and postimplantation development. We observed that Pou5f1, Nanog and Cdx2 expression, normal ICM-derived colony formation and the numbers of live offspring were all restored in Chd1-knockdown-Hmgpi-rescue embryos, suggesting that CHD1 plays important roles as a trigger for Pou5f1, Nanog and Cdx2 expression through the regulation of Hmgpi expression at ZGA. Furthermore, we showed that the rescue of HMGPI had no effect on Klf5 expression in Chd1-knockdown-Hmgpi-rescue embryos, suggesting that there is no direct interaction between Hmgpi and Klf5 during preimplantation development.

The maintenance of pluripotency depends on OCT4 and NANOG functions during preimplantation development (Chen et al., 2009; Mitsui et al., 2003; Nichols et al., 1998; Shao et al., 2008). Pou5f1 and Nanog, both of which are ICM markers, negatively interact with Cdx2, a TE marker, and these three genes are key regulators in cell fate specification (Chen et al., 2009; Niwa et al., 2005; Ralston et al., 2010; Strumpf et al., 2005). Pou5f1 or Nanog knockout mouse embryos can develop into morphologically normal blastocysts; however, developmental arrest occurs during the postimplantation period due to a loss of pluripotency (Mitsui et al., 2003; Nichols et al., 1998; Ralston et al., 2010). By contrast, Cdx2 knockout mouse embryos are arrested at the early blastocyst stage because they fail to maintain the blastocoel (Strumpf et al., 2005). Therefore, it is possible that suppression of Pou5f1, Nanog and Cdx2 could account for the exhibition of embryonic lethality. Although Chd1-knockdown embryos can develop into morphologically normal blastocysts despite suppression of Cdx2, it is probable that low levels of remaining CDX2 protein might function during TE development. Guzman-Ayala et al. (2015) reported that CHD1 is required for the maintenance of the epiblast at E5.5 (Guzman-Ayala et al., 2015); however, our results show that CHD1 is required for the expression of Hmgpi at ZGA and that global transcriptional activity is not changed at ZGA and MGA in Chd1-knockdown embryos. These results suggest that CHD1 selectively regulates transcription during mouse early preimplantation development. Together, these results suggest that CHD1 plays dual roles during embryogenesis: (1) resulting in Hmgpi and Klf5 expressions at ZGA during preimplantation development, and (2) acting as a regulator of global gene expression after implantation.

Developmentally important genes are marked by H3K4me3 at pre-ZGA and have a strong propensity to be activated after ZGA (Lindeman et al., 2011). In this study, we demonstrated that CHD1, a chromatin-remodeling factor that recognizes H3K4me3, is involved in the regulation of gene expression at ZGA and affects the development of mouse pre- and post-implantation embryos. Therefore, to understand the regulatory mechanisms of early embryogenesis, it is important to discern the transcriptional mechanisms in which H3K4me3 is involved at ZGA.

In conclusion, we demonstrated that CHD1 regulates the initiation of zygotic Oct4, Nanog and Cdx2 expression at MGA via activation of Hmgpi and Klf5 expression at ZGA. Thereafter, both HMGPI and KLF5, under the control of CHD1, regulate the expression of Pou5f1, Nanog and Cdx2, and thereby control initiation of cell fate specification. Subsequently, Klf5 expression escapes from the control of CHD1, and HMGPI regulates the expression of Pou5f1 and Nanog in the future ICM/EPI and the expression of Cdx2 in the future TE.

Eight- to ten-week-old ICR female mice (Japan SLC) were superovulated by injecting 5 IU of equine chorionic gonadotropin (eCG; ASUKA), followed by 5 IU of human chorionic gonadotropin (hCG; ASUKA) 48 h later. Unfertilized eggs were harvested 14 h after the hCG injection and placed in a 90-μl droplet of HTF supplemented with 4 mg/ml BSA (A3311; Sigma-Aldrich) (Minami et al., 2001). Spermatozoa were collected from the cauda epididymis of 11- to 15-week-old ICR male mice (Japan SLC) and cultured for 2 h in 100-μl of HTF supplemented with 4 mg/ml BSA. After preincubation, sperm were introduced into fertilization droplets at a final concentration of 1×10⁶ cells/ml. After a 3-h incubation, fertilized 1-cell embryos were collected and washed three times in KSOM supplemented with amino acids (Ho et al., 1995) and 4 mg/ml BSA, and then were either used for microinjection (39) or cultured until embryonic day 4.5 (E4.5) in the same medium under mineral oil (Sigma-Aldrich) at 37°C in an atmosphere of 5% CO₂.

Approximately 5-10 pl of 100 µM Chd1 siRNA (siChd1) (RNAi, Japan; 5′-GGUUUACUUAGGCGACAUUAA-3′) or the control scrambled sequence siRNA (siControl; RNAi) (5′-GCGUUUAUAGCAUAUUGCGAA-3′) in annealing buffer [consisting of 30 mM HEPES-KOH (pH 7.4), 100 mM KOAc and 2 mM Mg(OAc)₂] was microinjected into the cytoplasm of 1-cell embryos between 3 and 4 h after insemination. After injection, the embryos were cultured in KSOM medium supplemented with amino acids (Ho et al., 1995) with 4 mg/ml BSA under mineral oil (Sigma-Aldrich) at 37°C in an atmosphere of 5% CO₂. To examine developmental competency and hatching ability, embryos were observed at 36 (E1.5), 60 (E2.5), 84 (E3.5), 108 (E4.5) and 132 h (E5.5) after insemination. Furthermore, the embryos were harvested for quantitative RT-PCR (qRT-PCR), IF staining, outgrowth analysis or embryo transfer.

Embryos were harvested after culturing for 14, 20, 36, 48, 55, 76 and 108 h after insemination, when most of the oocytes had reached the 1-cell, early 2-cell, late 2-cell, 4-cell, 8-cell, morula and blastocyst stages, respectively. RNA extraction and qRT-PCR were performed as described previously (Suzuki et al., 2013). Total RNA from 30 embryos was extracted using TRIzol (Invitrogen). Transcription levels were determined on three different sets of 30 embryos per stage and normalized to H2afz, a stable reference gene used for normalization of gene expression in mouse preimplantation embryos (Jeong et al., 2005; Mamo et al., 2007); relative gene expression was calculated using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). The primers used for qPCR are listed in supplementary material Table S1.

Embryos for immunostaining were collected as described above. For staining of CHD1, HMGPI and KLF5, zona pellucida was removed from the embryo by acid Tyrode's solution (pH 2.5) and the embryos were fixed in phosphate buffered saline (PBS) containing 4% paraformaldehyde (Sigma-Aldrich) for 20 min at 4°C. After washing three times in PBS containing 0.3% polyvinylpyrrolidone (PVP K-30, Nacalai Tesque; PBS/PVP), embryos were treated with 0.5% Triton X-100 (Sigma-Aldrich) in PBS for 40 min at room temperature (RT), blocked in PBS containing 1.0% BSA (A9647, Sigma-Aldrich) for 1 h at RT (for CHD1 and KLF5) or 3.0% BSA overnight at 4°C (for HMGPI) and then incubated overnight at 4°C with a rabbit anti-CHD1 antibody (1:25 dilution; #4351, Cell Signaling) or a rat anti-KLF5 antibody (1:500 dilution; Shindo et al., 2002) or for 1 h at RT with a rabbit anti-HMGPI antibody (1:100 dilution; Yamada et al., 2010) in PBS containing 1.0% BSA (PBS/BSA). After washing with PBS/BSA, embryos were incubated in PBS/BSA containing a secondary antibody (Alexa Fluor 594 goat anti-rabbit IgG, 1:250 dilution, Invitrogen, R37117; Alexa Fluor 488 goat anti-rat IgG, 1:300 dilution, Invitrogen, A-11006; or Alexa Fluor 488 goat anti-rabbit IgG, 1:500 dilution, Invitrogen, A27034) for 1 h at RT. After washing three times in PBS/BSA for 15 min each, nuclei were stained in PBS/BSA containing 10 µg/ml Hoechst 33342 (Sigma-Aldrich) for 10 min. Immunostaining with normal rabbit IgG (sc-2027, Santa Cruz; 1:200 dilution) was included as a negative control for the specificity of the anti-CHD1 antibody. For double-staining of OCT4 and CDX2, immunofluorescence staining was performed as previously described (Isaji et al., 2013). For NANOG staining, embryos were fixed in PBS containing 4% paraformaldehyde for 10 min at RT. After three washes in PBS/PVP, fixed embryos were treated with 0.5% Triton X-100 (Sigma-Aldrich) in PBS for 30 min at RT, blocked in PBS containing 10% fetal bovine serum and 0.1% Triton X-100 (PBS/FBS) for 1 h at RT. Next, embryos were incubated overnight at 4°C with a rabbit anti-NANOG antibody (1:1000 dilution, 1 µg/ml; ab5731, Millipore) in PBS/FBS. Embryos were washed three times in PBS/FBS and then incubated for 1 h at RT with the appropriate secondary antibody diluted 1:750 (Alexa Fluor 594-conjugated goat anti-rabbit IgG, Invitrogen, A-11037). After staining, the samples were washed three times in PBS/BSA or PBS/FBS for 15 min, and nuclei were stained for 10 min at RT in PBS containing 10 µg/ml Hoechst 33342 (Sigma-Aldrich). For staining of BrUTP, immunofluorescence staining was performed as described previously (Aoki et al., 1997). After IF staining, embryos were mounted on slides in 50% glycerol/PBS and fluorescent signals were detected with a fluorescence microscope (BX50, Olympus). At least 30 oocytes were examined in each group. The numbers of ICM and TE cells were determined by counting OCT4⁺ and CDX2⁺ cells, respectively. Total embryonic cell numbers were obtained by adding the numbers determined for the ICM and TE cells. To count the cell number and observe the localization of OCT4, NANOG and CDX2 in Chd1-knockdown embryos, exposure time was extended relative to that used for control embryos.

Outgrowth analysis was carried out on E3.5 embryos after removal of the zona pellucida by acid Tyrode's solution (pH 2.5). The embryos were cultured in ES medium (Glasgow modification of Eagle's medium/GMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS), 1 mM sodium pyruvate (Sigma), 0.1 mM MEM non-essential amino acid (Invitrogen), 0.1 mM 2-mercaptoethanol (Wako), 0.1 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen) on a 0.1% gelatin-coated dish (Sigma-Aldrich) at 37°C in an atmosphere of 5% CO₂. After 4 days in culture, the embryos were photographed and the percentage of blastocysts that had undergone outgrowth was calculated.

Fifteen embryos that reached the 2-cell stage after microinjection were transferred into the oviducts of surrogate females (Japan SLC), which had been mated with vasectomized males the day before embryo transfer. These females were sacrificed at day 19 and pups were counted. The experiment was repeated three times.

Embryos were harvested after culturing for 36 and 48 h after insemination, when most of the oocytes reached the late 2- and 4-cell stages, respectively. BrUTP incorporation assay was performed by electroporation using the Super Electroporator NEPA 21 (NEPAGENE). Embryos were washed twice in PBS and then transferred in a line on the glass chamber between metal plates filled with PBS containing 10 mM BrUTP (Sigma-Aldrich). The poring pulse (voltage: 225 V, pulse length: 0.5 ms, pulse interval: 50 ms, number of pulses: 4, +) and the transfer pulse (voltage: 20 V, pulse length: 50 ms, pulse interval: 50 ms, number of pulses: 5, ±) were selected. The embryos were washed three times and cultured in KSOM supplemented with amino acids (Ho et al., 1995) and 4 mg/ml BSA under mineral oil (Sigma-Aldrich) at 37°C in an atmosphere of 5% CO₂ for 1 h. Subsequently, the embryos were treated with 0.2% Triton X-100 (Sigma-Aldrich) in PBS for 3 min at RT after removal of the zona pellucida by acid Tyrode's solution (pH 2.5) and fixed in PBS containing 4% paraformaldehyde (Sigma-Aldrich) overnight at 4°C. After washing three times in PBS/PVP, they were subjected to immunostaining. BrUTP-related signals were observed using a fluorescence microscope (BZ-X700, Keyence), and the fluorescence signal intensity was quantitated by using measurement module BZ-H3C (Hybrid Cell Count, Keyence).

The Hmgpi ORF was generated by PCR from mouse embryonic stem cell cDNA. For construction of an Hmgpi expression vector, the Hmgpi ORF was digested by AgeI and BamHI, and the fragment was cloned into the pAcGFP1-C1 vector plasmid (Clontech). The plasmid, digested by MluI, was used as template for in vitro transcription. RNA synthesis and poly(A) tailing were performed with a MEGAscript T7 kit (Invitrogen). Approximately 5-10 pl of 100 ng/μl Hmgpi mRNA in DEPC water (Invitrogen) was microinjected into the embryonic cytoplasm just after microinjection of siChd1. The primers used for cloning are listed in supplementary material Table S2.

Each experiment was repeated at least three times. All data are expressed as means±s.e.m. Statistical analysis of the data was performed by analysis of variance (ANOVA) with Student's t-test. P-values <0.05 were considered to be statistically significant.

All animal experiments were approved by the Animal Research Committee of Kyoto University (Permit number: 24-17) and performed in accordance with the guidelines of the committee.

We thank Dr Toshio Hamatani (Department of Obstetrics and Gynecology, Keio University School of Medicine, Tokyo, Japan) for kindly providing rabbit anti-HMGPI antibody.