FLASH/casp8ap2 Is Indispensable for Early Embryogenesis but Dispensable for Proliferation and Differentiation of ES Cells

FLICE/caspase-8-associated huge protein (FLASH)/casp8ap2 is involved in various cellular functions, such as cell cycle progression, transcriptional regulation, the regulation of apoptosis, and the regulation of histone gene expression. The down-regulated expression of FLASH has been shown to inhibit cell cycle progression in the S phase in many kinds of mice and human cell lines and the inhibition of cell cycle progression may be attributed to the suppressed expression of replication-dependent histone genes. We here demonstrated that the induced knockout of FLASH never affected cell cycle progression in ES cells, in which the expression of core histone genes was decreased to levels similar to those in human KB cells sensitive to the knockdown of FLASH. In addition, the FLASH conditional knockout ES cells could differentiate normally into not only mesodermal and endodermal cells, but also trophoblasts. In order to investigate the function of FLASH in early embryogenesis in vivo, we also examined a FLASH mutant mouse, in which FLASH mutant allele did not express FLASH mRNA in embryos and most adult organs, except for the testis. FLASH mutant embryos died between E3.5 and E8.5. Furthermore, the in vitro cultivation of FLASH mutant embryos generated by in vitro fertilization showed embryonic lethality at the pre-implantation stage by inhibiting the hatching of embryos and their adherence to substrates. Taken together, these results indicate that FLASH plays an important role in early embryogenesis, but is not essential for either the proliferation or differentiation of ES cells.


Introduction
FLICE/caspase-8-associated huge protein (FLASH)/casp8ap2, which was originally identified as a caspase-8-assoiated protein, is a high molecular weight protein with both a putative nuclear export signal (NES) and nuclear localization signal (NLS) [1]. FLASH is conserved in humans, mice, and chickens, and possibly also in Drosophila [2] and Xenopus [3]. Previous studies showed that FLASH was mainly localized in the nucleus, formed nuclear foci, and interact with other nuclear proteins with speckled localization, such as NPAT and ARS2 [4][5][6][7][8]. FLASH was reported to play a role in various cellular processes, including cell cycle progression especially in the S phase [4,6,9], processing of histone mRNA [2], regulation of apoptosis [1,5,7,10], and transcriptional control [3,[11][12][13][14][15]. The down-regulated expression of FLASH by an RNAi or shRNA-expression method is known to induce cell cycle arrest in the S phase. FLASH was also identified as a prognostic marker of acute lymphoblastic leukemia [16], and the monoallelic deletion of FLASH was observed in 18% of T-cell lymphoblastic lymphoma patients who had a poor prognosis [17].
A previous study demonstrated that FLASH was mainly localized in the nucleus and in nuclear bodies with NPAT [4]. Although FLASH was initially reported to localize in Cajal bodies [18], recent studies showed that FLASH did not localize exclusively in Cajal bodies; a fraction of FLASH-containing nuclear bodies were associated with Cajal bodies [5][6][7][8]10]. FLASH was also shown to localize in Histone Locus Bodies (HLBs) together with NPAT and Hinf-P, and to be an essential structural component of HLBs [4,6,9].
FLASH is known to play an essential role in the replicationdependent 39-end processing of histone pre-mRNAs, and the down-regulated expression of FLASH by an RNAi or shRNAexpression method induced cell cycle arrest within the S phase [2,6,9]. When the expression of FLASH was suppressed, the expression levels of core histones, such as histone H2, H3, and H4, were also decreased [6,9,19], which, in turn, induces the arrest of cell cycle progression within the S phase.
FLASH was recently identified as an essential molecule in early embryogenesis, and the expression of histone H4 was downregulated at both the mRNA and protein levels in FLASH mutant embryos [20]. In the present study, we generated and analyzed FLASH conditional knockout (KO) ES cell clones. These cells proliferated and differentiated normally into primitive cell lineages and the trophectoderm (TE) lineage, whereas FLASH-deficient mice showed an embryonic lethal phenotype at the preimplantation stage, as previously reported. These results indicate that FLASH plays an important role in early embryogenesis, but is not essential in either the proliferation or differentiation of ES cells.

FLASH was dispensable for the proliferation of ES cells
To examine the function of FLASH in early embryogenesis, we generated FLASH conditional knockout (KO) mouse ES cell clones using a gene targeting technique with the Cre-loxP system ( Figure 1A). The FLASH gene encodes 12 exons, and exon 2 contains a translation initiation site. We initially generated FLASH flox/+ ES clones in which exon 2 of FLASH in one allele was flanked by loxP sites. FLASH flox/-ES clones were then generated by deleting exon 2 of FLASH in the other allele. By using an expression vector for Mer-Cre-Mer (Mouse Estrogen Receptor-Cre-Mouse Estrogen Receptor), the treatment of FLASH flox/-ES cells with 4-OHT led to the FLASH gene being biallelically deleted ( Figure 1B). The expression of the FLASH protein was suppressed 4 days after the treatment with 4-OHT ( Figure 1C). To investigate the function of FLASH in ES cells, we examined the effects of a more than 10-day treatment with 4-OHT on the growth of FLASH conditional KO ES cells. In contrast to previous findings in various human and mouse cell lines [4,6,9], the induction of FLASH KO did not affect the proliferation of ES cells ( Figure 1D).

FLASH was dispensable for the differentiation of ES cells
The induced knockout of the FLASH gene did not affect the proliferation of ES cells. We then investigated whether FLASH knockout affected the differentiation of ES cells induced by Embryoid Body (EB) formation. No significant difference was observed in EB formation activity between FLASH KO ES cells and WT ES cells ( Figure 1E). The expression levels and patterns of both an undifferentiated ES cell marker, Oct3/4, and differentiated ES cell markers to the mesoderm and endoderm, Brachyury and GATA6, respectively, were the same between FLASH WT and FLASH KO ES cells ( Figure 1F). These results suggested that FLASH may not affect the early development of ES cells.
We then examined the differentiation activity of FLASH KO ES cells. FLASH KO ES cells could differentiate into Tuj-1positiive neural cells, and no significant differences were observed in neuronal differentiation activity between WT and FLASH KO ES cells ( Figure S1). We then generated beating cardiac muscle cells from WT and FLASH KO ES cells after the 10-day in vitro cultivation of EB. FLASH KO ES cells differentiated normally into cardiac muscle cells and no significant differences were detected between cardiac muscle cells derived from WT and FLASH KO ES cells (Data not shown). These results suggested that FLASH may not play an essential role in not only the proliferation, but also the differentiation of ES cells in vitro. In addition, FLASH may dispensable for the survival and proliferation of differentiated cells derived from ES cells.

FLASH was dispensable for the differentiation of ES cells into trophoblasts
Although FLASH was dispensable for the in vitro proliferation and differentiation of ES cells, it was previously shown to be essential during early embryogenesis in the mouse. We speculated that the discrepancy in the phenotype between ES cells and zygotes that do not express FLASH may have been caused by the ability to form a placenta. The trophectoderm (TE) is essential for hatching from the zona pellucida and implantation, and the differentiation activity of zygotes to trophoblasts is subsequently indispensable for ontogenesis from zygotes. Cdx-2 is an essential transcription factor that regulated the formation and function of the TE [21]. The exogenous expression or activation of Cdx-2 in ES cells was previously shown to trigger differentiation into the trophectoderm lineage [22]. To investigate the role of FLASH in differentiation into trophoblasts or the survival of trophoblasts, we generated an inducible TE differentiation system in FLASH KO ES cell lines. A Cre recombinase-expression vector was introduced into a FLASH flox/-(heterozygous FLASH KO) ES cell clone by electroporation and a single FLASH KO ES clone was produced. Transgenic FLASH wt, FLASH flox/-, and FLASH KO ES cells, all of which constitutively expressed the fusion protein of Cdx2 and the Estrogen Receptor (ER) (Cdx2ER), were then generated (Figure 2A The expression of FLASH was absent in most tissues in the FLASH mutant mouse While FLASH mutant mice have been reported to die in the early embryonic stage [20], FLASH KO ES cells was shown to proliferate and differentiate normally in vitro. To examine the effects of FLASH during early embryogenesis, we obtained and analyzed FLASH mutant mouse from Lexicon Pharmaceuticals in which an OmniBank gene trapping vector was inserted in the FLASH allele ( Figure 3A). Mouse ES cells carrying a gene trapping retroviral vector in the FLASH gene were found in the OmniBank, a library of gene-trapped ES cell clones identified by a corresponding OmniBank sequence tag (OST). Mice derived from the ES cell clone corresponding to OST 97730, matching the mouse FLASH sequence, were analyzed using inverse genomic PCR analysis and Southern blot analysis, and the results obtained showed the insertion of the gene trapping retroviral vector in intron 1 between the 789 th and 790 th bases of the mouse FLASH gene (Figure 3 A, B and C).
The insertion of the trapping vector into the FLASH gene generated a fusion transcript between the Neomycin-resistant gene with the translational termination codon and exon 1 of the FLASH gene under the control of the FLASH promoter ( Figure 4A). This fusion transcript only produced the Neomycinresistant protein because exon 1 of the FLASH gene did not encode the translational initiation codon of FLASH. The trapping vector encoded a PGK promoter followed by a small piece of the BTK gene connected to a splice donor signal. Splicing from the donor site to the splicing acceptor site in exon 2 of the FLASH gene, which encoded the translational initiation codon of FLASH, generated fusion mRNA (mutant FLASH mRNA) containing a piece of BTK and exons 2-11 of FLASH, which were presumed to produce the complete FLASH protein encoded by exons 2-11 of the FLASH gene ( Figure 4A). We then analyzed the expression of WT and mutant FLASH mRNA by RT-PCR in the various organs and embryonic fibroblasts (MEFs) of FLASH mut/+ mice.
WT FLASH mRNA was detected in all the organs examined and MEFs; however, mutant FLASH mRNA was only detected in the testis ( Figure 4B). These results suggested that mutant FLASH mRNA was not expressed from the FLASH mutant allele in most tissues, except for the testis. To confirm this result, the amounts of FLASH mRNA and the FLASH protein in FLASH +/+ and FLASH mut/+ MEFs were quantified using qRT-PCR and Western blot analyses, respectively. The expression levels of both FLASH mRNA and protein were almost 50% lower in FLASH mut/+ MEFs than in FLASH +/+ MEFs ( Figure 4C and D). These results demonstrated that the FLASH mutant allele did not express FLASH in most tissues, except for the testis.
FLASH was essential for early embryonic development FLASH mut/+ mice were healthy and did not appear to be different from their WT littermates (Data not shown). Genotyping of postnatal mice after intercrossing with heterozygote mice revealed the presence of WT mice and heterozygous mice (FLASH mut/+ ) at the expected 1:2 Mendelian ratio, whereas the homozygous FLASH mutant (FLASH mut/mut ) was not detected ( Table 1). The genotypes of the embryos at E8.5 (E = embryonic day) through E14.5 were analyzed by PCR after FLASH mut/+ mice were intercrossed, and the results obtained revealed the absence of FLASH mut/mut embryos. On the other hand, FLASH mut/mut embryos were detected at E3.5 according to Mendelian ratios. These results indicated that FLASH mut/mut embryos died between E3.5 and E8.5.
FLASH mut/mut embryos at the blastocyst stage could not attach to gelatin-coated dishes or hatch from their zona pellucida To further investigate the lethality of FLASH mut/mut early embryos, embryos were cultured in vitro in gelatin-coated dishes containing the culture medium for ES cells without leukemia inhibitory factor (LIF). These embryos were generated by the in vitro fertilization (IVF) of sperm and oocytes from FLASH mut/+ male and female mice, respectively. Three days after fertilization, FLASH mut/mut embryos did not exhibit normal hatching from their zona pellucid and could not adhere to the culture dish, whereas FLASH +/+ and FLASH mut/+ embryos displayed proper hatching, appropriate adhesion to the dish, and normal progression ( Figure 5A) ( Table 2). The expression levels of WT and mutant FLASH mRNAs were examined in blastocyst stage embryos by RT-PCR ( Figure 5B). The results obtained indicated that mutant FLASH mRNA was not expressed in FLASH mut/+ or FLASH mut/mut embryos, while wild-type FLASH mRNA was expressed in FLASH +/+ and FLASH mut/+ embryos. Taken together, these results clearly demonstrated that FLASH was indispensable for the pre-implantation stage by supporting the hatching of embryos and their adherence to substrates.
The transcriptional deregulation of the core histone genes was observed in FLASH KO ES cells Previous studies showed that the down-regulated expression of FLASH by RNAi and shRNA-expression methods induced cell cycle arrest in the S phase and the cell cycle arrest may be attributed to the down-regulated expression of core histone genes such as histone H3 and H4 at the mRNA level in various cell lines including human KB cells [6,9]. In addition, the expression of both histone-H4 and H3 mRNA was shown to be lower in FLASH-aberrant embryos than in wild type embryos [20]. We examined the expression levels of histone H3 and H4 mRNA and histone H3 protein in FLASH KO ES cells in which cell cycle arrest was absent ( Figure 6). The expression of both histone-H3 and H4 was lower in FLASH KO ES cells than in wild type ES cells. Consistent with previous findings [6], the induced expression of the shRNA of FLASH suppressed the expression of histone H3 and H4 genes in KB cells in which cell cycle progression was inhibited at the S phase, and the expression levels of histone H3 and H4 genes in FLASH KO ES cells were suppressed to a similar extent as those in KB cells expressing the shRNA of FLASH ( Figure 6). These results indicated that the down-regulated expression of core histone genes was not correlated with cell cycle arrest in the S phase in FLASH down-regulated cells.
To investigate the function of FLASH in early embryogenesis, we generated inducible FLASH knockout ES cell clones. Previous studies showed that the suppression of FLASH expression by an RNAi or shRNA-expression method caused cell cycle arrest at the S phase in various cell lines [6,9,13]. However, our FLASH KO ES cells grew normally and cell cycle progression was normal ( Figure 1D). A deficiency in the FLASH protein was examined using Western blot analysis with both an anti-FLASH monoclonal antibody and anti-FLASH polyclonal antibody (Figure 2A, data not shown). The results obtained indicated that not only the fulllength FLASH protein, but also truncated forms of the FLASH protein were not generated in FLASH KO ES cells that could proliferate and differentiate normally. Previous studies demonstrated that the knockdown of FLASH caused a decrease in replication-dependent histone mRNA levels, and also that the down-regulated expression of core histones may cause cell cycle arrest at the S phase [4,6,9]. We then quantified the amounts of histone-H3 and H4 mRNAs and proteins using qRT-PCR and Western blot analyses, respectively, in FLASH KO ES cells ( Figure 6). The suppression of FLASH expression reduced the amounts of both histone-H3 and H4 to a similar extent in not only KB cells sensitive to FLASH knockdown, but also FLASH KO ES cells. These results suggested that the down-regulated expression of S phase-specific core histone genes was not correlated with cell cycle arrest at the S phase. The molecular mechanism underlying cell cycle arrest at the S phase in FLASH-deficient cells currently remains unknown. Therefore, the mechanisms by which FLASH is involved in S phase progression and/or how ES cells with the decreased expression of core histones can proliferate normally must be clarified. We speculated that cell-cycle-independent histone variants, including histone H3.3, might be involved in restoration of normal chromatin assembly in FLASH KO ES cells. Histone H3.3 was reported to be able to substitute for S-phase specific canonical histone H3.2 in histone H3.2-deficient Drosophila, and cells in histone H3.2-deficient Drosophila could divide and differentiate when histone H3.2 was replaced by S phase-expressed histone H3.3 [23]. It is necessary to analyze the expression levels and functions of cell-cycle-independent histone variants in FLASH KO ES cells.
To investigate the physiological function of FLASH in vivo, we examined a FLASH mutant mouse, generated by Lexicon Pharmaceuticals, Inc., that harbored the trapping vector between exons 1 and 2 in the FLASH gene. We demonstrated that this mouse was not a FLASH knockout mouse, but was a mutant mouse of the FLASH promoter (Figure 4). A similar FLASH mutant mouse was used as a FLASH knockout mouse in a recent study in which FLASH was shown to be essential for mouse early development; however, the expression of FLASH mRNA and the FLASH protein was not examined [20]. In the present study, we analyzed the expression of FLASH mRNA in various organs of the mutant mouse, and the mutant allele was shown to not express FLASH mRNA in most tissues and embryos, except for the testis. Although the silencing mechanism of FLASH expression currently remains unclear, we speculated that three termination codons in The genotypes of viable mice at the indicated postnatal (P) or embryonic (E) day were determined by PCR analysis. ND; genotype could not be determined. doi:10.1371/journal.pone.0108032.t001 Figure 5. In vitro culture of FLASH mutant embryos. (A) FLASH +/+ (WT), FLASH mut/+ (mut/+), and FLASH mut/mut (mut/mut) embryos 2 days after in vitro fertilization (IVF) or from the oviducts of FLASH mut/+ female mice mated 3.5 days before with FLASH mut/+ male mice were cultured for the indicated days in gelatin-coated dishes with ES medium, and were then observed under a phase-contrast microscope. ZP: zona pellucida, ICM: inner cell mass, and TCG: trophectoderm giant cell. Scale bar, 20 mm. (B) Genotyping of the embryos 5.5 days after IVF was carried out by PCR analysis with primers 5, 6, and 8, as indicated in Figure 3. One WT, two mut/+, and two mut/mut embryos were detected. (C) Total RNA was prepared from adult mut/+ testis, and two WT and two mut/+ embryos 5.5 days after IVF, and WT and mutant (Mut) FLASH mRNAs were detected by RT-PCR. doi:10.1371/journal.pone.0108032.g005 the part of the BTK gene that fused with the 59 end of FLASH mRNA may degrade mutant FLASH mRNA. This degradation of mRNA is known as nonsense-mediated decay (NMD) and was previously reported to be induced in many tissues, except for the testis [24]. The mutant FLASH mRNA was expressed in only testis conceivably through inhibition of NMD. Alternatively, we speculated that instability of nucleosomes containing a testis-specific histone variant, H3T, may be involved in the testis-specific expression of mutant FLASH mRNA, because histone H3Tcontaining nucleosomes were reported to be unstable [25,26], which could enhance gene expression. FLASH mut/mut mice did not survive until birth, and an analysis of FLASH mut/mut embryos revealed that they died between E3.5 and E8.5. Approximately 70% of FLASH mut/mut embryos  27 14 In vitro fertilized eggs were cultured for 3 days, and it was then determined whether the embryos had adhered to the culture dish. doi:10.1371/journal.pone.0108032.t002 appeared to form normal blastocysts, while the other 30% showed a little lag in development. These results suggested that FLASH may be essential for mouse early development. In addition, the cultivation of embryos obtained by the in vitro fertilization (IVF) of FLASH mut/+ mice showed that FLASH mut/mut embryos did not hatch from the zona pellucida or attach onto gelatin-coated dishes. In addition, approximately 30% of FLASH mut/mut embryos remained at the morulae stage, whereas the remainder developed into blastocysts. In this pre-implantation stage, embryos contain maternally-produced gene products, most of which disappear at the 2-8 cell or blastocyst stage. Therefore, maternal FLASH mRNA and protein may remain and function in the blastocyst stage in FLASH mut/mut embryos, and the passing of maternal FLASH may induce abnormal embryogenesis. Consistent with previous findings [20], we showed that FLASH played an essential role in embryogenesis around the pre-implantation stage. However, the mechanisms by which knockdown of FLASH expression disrupted the development of the embryo prior to implantation have not been clarified. We speculated that cell-cycle-independent histone variants, such as histone H3.3, might be involved in the mechanisms. Recently, histone H3.3 incorporation into nucleosomes was reported to be essential for parental genome reprogramming and reveal an unexpected role for ribosomal RNA transcription in the mouse zygote [27]. In FLASH KO embryo, down-regulated expression of S-phase-specific canonical histones might induce abnormal incorporation of histone H3.3 into nucleosomes and aberrantly-incorporated histone H3.3 might militate against early embryogenesis. We previously reported that FLASH interacted with ARS2 [6] and both molecules played a role in cell cycle progression [6,28]. In addition, Ars2 was recently reported to play a role in histone mRNA 39 end formation and expression [29], and ARS2-deficient mice were shown to die around the pre-implantation stage, similar to FLASH-deficient mouse [30]; however, the mechanism underlying the embryonic lethality of ARS2-deficient mice remains unclear. Both FLASH and ARS2 were essential for the development of mouse embryos in the pre-implantation stage; however, it has yet to be determined whether FLASH and ARS2 have the same function in the pre-implantation stage. A detailed examination of embryos from FLASH and ARS2 knockout mice is needed in order to clarify the physiological meaning of the interaction between the FLASH and ARS2 proteins.
In order to further understand the physiological roles of FLASH, FLASH conditional knockout mice must be generated and examined, and the molecular mechanism underlying cell cycle arrest at the S phase in FLASH-deficient cells must also be clarified.

Ethics statement
The animal experiments were carried out in strict accordance with the protocols approved by the Animal Experiment Committee of Graduate School of Biostudies, Kyoto University (Animal Experiment Protocol No. Lif-K12009). All efforts were made to minimize animal suffering.
Mice C57BL/6 mice were purchased form CLEA Japan. FLASH mutant mouse was obtained from Lexicon Pharmaceuticals in which an OmniBank gene trapping vector was inserted in the FLASH allele. The FLASH mut/+ mice on the C57BL/6 background were generated by backcrossing with C57BL/6 mice for more than 12 generations and maintained. Mice were maintained in Specific Pathogen-Free conditions.

RT-PCR and qRT-PCR
Total mRNA for RT-PCR and qRT-PCR was prepared using ISOGEN (Nippon Gene) or the RNeasy Mini kit (Qiagen), according to the manufacturer's instructions. RT products for RT-PCR and real-time PCR were generated using the Thermo Script RT-PCR system (Invitrogen) or ReverTra Ace qPCR RT Kit (TAKARA), and a high-capacity cDNA RT kit (Applied Biosystems), respectively. RT products were analyzed with ABI PRISM7500 (Applied Biosystems). Primer sequences are shown in Figure S2. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal standard for all samples. PCR was performed using the following primers for embryos and tail snips: primer 5 (59-GGCGATCAGTGACAGGGGCAGTTTCC-39) and primer 8 (59-AGCTTTGAAAAGGCCAGTCCCTAG-CACCA-39) to amplify a 808 bp fragment from the wild-type allele, and primer 5 and primer 6 (59-TGTTGTGCCCAGTCATAGCC-GAAT-39) to amplify a 1486 bp fragment from the mutant allele.

Cultivation of isolated embryos
Heterozygous FLASH mut/+ female mice were superovulated using an injection of pregnant mare serum (PMS) gonadotrophin (Asuka Pharmaceuticals). After 2 days, human chorionic gonadotrophin (HCG) (Asuka pharmaceuticals) was injected, and FLASH mut/+ female mice were then crossed with heterozygous FLASH mut/+ male mice. The next day was established as embryonic day 0.5 (E0.5) if successful mating was detected by the presence of a vaginal plug. E3.5 embryos were obtained by flushing oviducts and uteri with KSOM medium (Chemicon). E3.5 embryos were placed individually onto 0.1% gelatin-coated dishes, cultured in ES cell medium without LIF, and observed under an inverted microscope (Leica).

Cultivation of embryos obtained by in vitro fertilization (IVF)
PMS and HCG were injected into FLASH mut/+ female, and eggs were obtained for IVF by flushing the oviducts with KSOM. Sperm isolated from FLASH mut/+ male mice in HTF (human tubal fluid.) medium were added to the eggs. Fertilized eggs were placed individually onto 0.1% gelatin-coated dishes, cultured in ES cell medium without LIF, and observed under an inverted microscope (Leica).

Southern blot analysis
Genome DNA of the tail snips or ES cells was digested with EcoRI. The probe used in Southern blot analysis for the Neo r gene was generated by PCR with primers (59-CCGGTGCCCTGAAT-GAACTGCAGG-39 and 59-CCAACGCTATGTCCTGATAG-CGGT-39). Southern blot hybridization was performed at 68uC in Rapid-Hyb Buffer (Amersham) with the 32 P-labeled probes, followed by washing at room temperature and then at 68uC. Autoradiography was performed with FLA-7000 (Fuji Film).

Construction of the targeting vector and establishment of FLASH conditional knockout ES lines
The first and second targeting vectors described in Figure 1A were constructed using a C57Black6 BAC clone RP23-69g24 (BACPAC Resources Center). The first targeting vector contained two loxP sites at 7976 th bp and 12255 th bp of the FLASH gene. The targeting vector was introduced into ES cells by electroporation (0.3 KV, 75 mF; twice). After screening by G418 selection, recombinant ES clones (FLASH flox/+ ) were identified by PCR, and confirmed by Southern blot analysis. FLP recombinase encoded by an adenovirus vector was introduced into the ES clones in order to eliminate the Neo r gene. The second targeting vector containing a Neo r gene in exon 2 of the FLASH gene was introduced into the ES clones by electroporation, and FLASH flox/clones were identified after screening by G418 selection and PCR analysis. A pCAG-MerCreMer-IP vector encoding a fusion gene of the mouse estrogen receptor and Cre (Mer-Cre-Mer), IRES, and a puro r gene were then introduced into the FLASH flox/clones and FLASH flox/clones expressing MerCreMer were established after screening by puromycin.

Trophoblast differentiation
A pCAG-Cdx2ER-IP vector, which was gifted by Dr. Hitoshi Niwa, RIKEN CDB, was introduced into a FLASH flox/-ES cell clone, and FLASH flox/-ES cells expressing Cdx2ER were identified by selection with puromycin. A pIC-Cre vector was then introduced into the ES clone by electroporation (0.25 KV, 500 mF), and a FLASH -/-ES clone expressing Cdx2ER was established. The ES cells obtained were seeded at a low density (3610 3 cells/cm 2 ) on gelatin-coated dishes without feeder MEFs, and treated with 1 mM 4-OHT for 8 days to generate differentiated trophoblast cells.

Embryoid body formation
Embryoid bodies were generated using the standard hanging drop method as follows. ES cells were suspended in DMEM supplemented with 10% fetal calf serum, and 25 ml drops (1610 4 ES cells per one drop) were plated on the lid of petri dishes in regular arrays. The lid was inverted, placed over the bottom of petri dishes filled with PBS, and incubated at 37uC for 2 days. Embryoid bodies in the drops were transferred to bacterialgrade dishes and cultured.

Neural cell differentiation
The differentiation of ES cells into neural cells was conducted using the SDIA (stromal cell-derived inducing activity) method as follows. ES cells were cultured on PA6 cells with neural differentiation-specific medium (GMEM (Gibco) containing 10% KSR (Gibco), 0.1 mM non-essential amino acids (Gibco), 1 mM sodium pyruvate (Sigma), and 0.1 mM 2-mercaptoethanol (WAKO)) for 10 days. The culture medium was exchanged every 2 days.