In the mouse, insights into germ cell development have been gleaned from the analysis of the impact of altered gene activity on the specification, the proliferation, and the migration of primordial germ cells (PGCs). For example, embryos that lack bone morphogenetic protein (Bmp)-4 activity do not activate Prdm1 (Blimp1) in the epiblast (Ohinata et al.,2005) and completely lack PGCs (Lawson et al.,1999). After PGCs are formed, the localization of PGCs in different germ layers is influenced by the expression of interferon induced transmembrane protein (Ifitm)-1 (mil-2/fragilis2) and Ifitm3 (mil-1/fragilis). Ifitm1, which elicits homotypic cell-to-cell interaction, is expressed in the mesoderm of the gastrula-stage embryo. PGC precursors in the mesoderm express Ifitm1, but those that have been relocated to the endoderm stop expressing this gene (Tanaka and Matsui,2002). This down-regulation of Ifitm1 is postulated to have generated a repulsive interaction between the mesoderm and the PGC precursors and facilitates their translocation from the Ifitm1-expressing mesoderm to the endoderm, which does not express Ifitm1 (Tanaka et al.,2005). PGCs descended from these precursors that reside in the endoderm of the hindgut remain responsive to the Ifitm1-mediated repulsive activity (Tanaka et al.,2005). The retention of the PGCs in the gut endoderm, prior to the re-initiation of migration from the gut through the mesentery to the genital ridges, may be a response of the PGCs to the repulsive activity elicited by the Ifitm1-expressing mesenchyme adjacent to the endoderm (Tam et al.,2006).
Activities of several transcription factor genes such as Brachyury (T), Foxa2, and Lhx1 (Lim1) are required for PGC formation, although the molecular pathway through which these factors might act is not known (Tsang et al.,2001). Loss of Lhx1 leads to the absence of PGCs in some mutant embryos, while in others PGCs are present in reduced numbers. In some Lhx1-null embryos, a cluster of alkaline phosphatase-positive PGCs is found ectopically in the extraembryonic mesoderm away from the primitive streak (Tsang et al.,2001). These cells are less likely to contribute to the PGC population within the embryo and might account for the reduction of the PGC population in Lhx1-null embryos (Tsang et al.,2001). These findings suggest that Lhx1 may have separate roles in the formation of the PGCs and their proper localization. Consistent with the latter role, Lhx1 activity has been shown to be required for the morphogenetic movement of axial mesendoderm in the Xenopus embryo, and of mesoderm and endoderm cells in the mouse embryo (Hukriede et al.,2003; Shimono and Behringer,2003; Tam et al.,2004). A confounding factor in the analysis of the impact of loss of Lhx1 on the PGC formation is that germ layer formation is also severely disrupted in the Lhx1-null mutant embryo (Tsang et al.,2000). Chimeric embryos with wild-type extraembryonic tissues and Lhx1-null mutant epiblast derivatives show no discernible defects of germ layer formation and morphogenesis, suggesting that Lhx1 activity in the non-epiblast derivatives, such as the visceral endoderm, is required for enabling gastrulation (Shawlot et al.,1999). In the present study, we dissected the specific effect of the loss of Lhx1 function on the PGCs by overcoming the confounding effect of abnormal gastrulation by studying embryos that lack Lhx1 function primarily in the epiblast, via Meox2-Cre-mediated excision of the floxed Lhx1 allele (Kwan and Behringer,2002). We show that Lhx1 activity in the epiblast and its derivatives does not affect the formation of PGCs but influences their retention in the hindgut endoderm, resulting in the loss of PGCs subsequently. We propose that the premature exodus of PGCs from the hindgut may be associated with the failure to elicit the requisite Iftim1-mediated repulsive activity that restricts the emigration of PGCs to the mesoderm surrounding the gut endoderm.
RESULTS AND DISCUSSION
Ablation of Lhx1 Activity in the Epiblast Derivatives
To visualize the pattern of Cre-activity in the conditional mutant, early streak stage Z/EG; Meox2-Cre embryos (see Experimental Procedures section) were examined for enhanced green fluorescent protein (EGFP) expression. In these embryos, the Z/EG reporter transgene that was activated by Cre activity in the epiblast remained active in the derivatives of the Cre-recombined epiblast cells. EGFP expression was found in about 75% (range 65–95%) of the epiblast cells and not in the visceral endoderm and extraembryonic ectoderm (Fig. 1A, A′). A similar frequency of EGFP-expressing cells was found in the hindgut endoderm of E8.25–E9.5 embryos (mean 65%, range 55–90%). A mosaic pattern of EGFP expression was also seen in the extraembryonic mesoderm derived from the epiblast (data not shown). In the Lhx1flox/del; Z/EG; Meox2-Cre (CKO) embryo, EGFP expression in epiblast-derived cells indicated that the Cre activity has activated the transgenic reporter and would also have ablated the floxed Lhx1 allele such that these cells become null for Lhx1 function. A similar mosaic pattern of EGFP-expression was observed in the epiblast and its derivatives in the CKO embryo and the Z/EG; Meox2-Cre embryos. CKO embryos with mosaic Lhx1 activity in the epiblast developed normally through gastrulation but displayed abnormal morphology during organogenesis (Fig. 1B–C′; Kwan and Behringer,2002). They developed proboscis-like rostral structures and foreshortened facial primordia, apparent first at the 8–12 somites stage (embryo day E 8.75; Fig. 1B, B′), culminating in a truncation of the head by the organogenesis stage (E9.5; Fig. 1C, C′). Other body parts were intact but overall development of CKO embryos was retarded (see also Kwan and Behringer,2002). Lhx1flox/del; Meox2-Cre embryos (i.e., without the Z/EG reporter gene, obtained by crossing Lhx1+/del; Meox2-Cre and Lhx1flox/flox mice) displayed a similar mutant phenotype. Therefore, subsequent analyses were performed on embryos of these two CKO genotypes and the results were compared with those of heterozygous embryos (see below).
Ablation of Lhx1 activity in the epiblast-derived cells was confirmed by immunostaining of Lhx1 protein in the E7.5 Lhx1flox/del; Meox2-Cre mutant embryos. Lhx1 was present in the mesendoderm and in the prospective lateral mesoderm in the E7.5 late streak to early head-fold stage wild-type embryo (Fig. 1D′ on the left; Shawlot and Behringer,1995; Tsang et al.,2000), whereas the mesoderm of the conditional Lhx1 mutant displayed a much weaker staining reaction (Fig. 1D′ CKO on the right). Confocal microscopy further highlighted the mosaic staining pattern of Lhx1 in the epiblast-derived cells in the mutant embryo. In the early streak-stage wild-type embryo, Lhx1 protein was present in the anterior visceral endoderm and the nascent mesoderm, whereas in the mutant embryo Lhx1 activity was absent from many of the mesoderm cells that would have expressed Lhx1 (Fig. 1E–F′ and magnified images G–H′). Based on the results of immunostaining and EGFP reporter expression (in Z/EG; Meox2-Cre embryos), we estimated that about 70% of epiblast-derived cells would be deficient in Lhx1 activity in the Lhx1flox/del; Meox2-Cre mutant embryo.
Lhx1 Activity in the Epiblast Is Dispensable for PGC Formation But Is Required for Population Expansion
The formation of PGCs was assessed by scoring the number of alkaline phosphatase (AP)-positive cells. Embryos of the following genotypes were examined: Lhx1+/flox (designed as wild-type); Lhx1flox/del (Lhx1-heterozygous), Lhx1+/flox; Meox2-Cre (conditional Lhx1-heterozygous) and Lhx1flox/del; Meox2-Cre (conditional Lhx1-knock out: CKO). At late-bud and neural plate stages, CKO embryos were phenotypically similar to the heterozygous embryos (Fig. 2A–E and data not shown). The number of AP-positive cells present in the E7.5 CKO mutant was not different from that in the wild-type or heterozygous embryos (Fig. 2F). These AP-positive cells expressed a PGC signature, the Dppa3/Pgc7/Stella protein (Fig. 2E, n = 5; Saitou et al.,2002; Sato et al.,2002) and the number of Dppa3-positive PGCs in the CKO mutants was not different from the heterozygotes (data not shown). Thus, loss of Lhx1 activity in the epiblast did not affect the formation of PGCs. In the CKO embryos, the population of PGCs in the hindgut endoderm did not increase at the same rate as in the heterozygous embryos after the 6–8 somite stage (Fig. 2G). Loss of Lhx1 activity in a majority of the epiblast-derived cells, therefore, leads to depletion of the PGC population after initial specification. The remaining Lhx1-positive epiblast-derived cells were unable to compensate for any loss of Lhx1 effect on the PGCs.
Reverse transcription polymerase chain reaction (RT–PCR) analysis showed that Lhx1 was not expressed in the Pou5f1- and Ifitm3-expressing PGCs in the E9.5 hindgut (Fig. 3A). This finding, together with that Lhx1 is not expressed in the PGC precursors (Saitou et al.,2002; Yabuta et al.,2006) is consistent with the concept that Lhx1 activity is not required autonomously for PGC development. However, it is possible Lhx1 function is critical for the allocation of PGC precursors in the epiblast and that the reduction in the PGC population is caused by the loss of the Lhx1-deficient precursors. The use of Meox2-Cre, which produces a mosaic pattern of recombinase activity in the epiblast, provides a unique opportunity to test this hypothesis. In the Lhx1flox/del; Z/EG; Meox2-Cre embryo, Lhx1-deficient PGCs may be distinguished by the expression of EGFP reporter from other PGCs in the embryos, which have no EGFP activity and, therefore, may still have a floxed (functional) allele. In the definitive endoderm of the E7.5 CKO embryo, both EGFP-positive and EGFP-negative Dppa3-expressing PGCs were found (Fig. 3B–C″). Overall 64% (29/45) of Dppa3-expressing PGCs in the CKO embryos showed EGFP activity (i.e., potentially with a deleted Lhx1 allele). The fraction of GFP-positive cells in the Dppa3-positive PGC population remained unchanged in the hindgut of the E9.5 CKO embryos: 67% (86/128) of the Dppa3-positive PGCs expressed EGFP (Fig. 3D–E″). By comparison, in the Lhx1+/flox; Z/EG; Meox2-Cre (heterozygous) embryo, 66.3% (177/267) of the Dppa3-positive PGCs expressed EGFP (i.e., with activated Cre but heterozygous for Lhx1 activity; Fig. 3F–G″). Collectively, these findings show that expression of Meox2-Cre activity alone or together with the deletion of Lhx1 activity has no cell-autonomous effect on the formation of the PGCs and the ability of the PGCs to populate the hindgut.
The Lhx1 transcription factor forms ternary complexes with LIM-binding domain proteins such as LDB1. Whereas Lhx1 activity in the epiblast is dispensable for PGC formation (this study; and Ifitm3-expressing PGCs are found in the E7.5 Lhx1-null embryos, data not shown), loss of Ldb1 is associated with the absence of PGCs (Mukhopadhyay et al.,2003). Ldb1 might, therefore, have a role in PGC formation independent of Lhx1 activity. The difference between Lhx1 and Ldb1 function might be related to the effect of BMP signaling, which is known to be essential for the induction of PGC precursors. Bmp4 is down-regulated in the Ldb1-null mutant (Mukhopadhyay et al.,2003), but is still expressed in the extraembryonic tissues of the Lhx1-null embryo (Kinder et al.,2001).
Lhx1 Activity Is Required for Retaining PGCs in the Hindgut
In CKO mutant embryos at the late bud (Fig. 2A–B′), neural plate (Fig. 2C–E), and early somite (1–5 somites; Fig. 4A–B′) stages, AP-positive PGCs were localized in the endoderm of the posterior gut, like those in the heterozygous embryos. However, by the 8-somite stage, PGCs in the CKO embryos were no longer confined to the gut but had dispersed to the adjacent mesenchyme (Fig. 4D, D′, arrowheads), in contrast to PGCs in the heterozygous embryos, which were predominantly found in the gut endoderm (Fig. 4C, C′). In the E9.0 (11-somite) CKO embryo, PGCs were scattered to the mesoderm tissues (Fig. 4E–F′, arrowheads). By E9.5, very few PGCs remained in the mutant hindgut (Fig. 4G–H′), with an average of 42.7 Dppa3-positive PGCs (n = 3) in the CKO hindgut compared with 134.5 (n = 2) in the heterozygous counterpart. In Sox17−/− embryos that were depleted of gut endoderm, it was found that hindgut morphogenesis is instrumental for the congregation of the PGCs to the gut endoderm (Hara et al.,2009). However, in the CKO mutant embryo, the hindgut and associated structures appeared to develop normally (Figs. 3D,F; 4I–K; 5C,C′; see also Kwan and Behringer,2002). The dispersion of PGCs from the hindgut of the CKO embryo is, therefore, unlikely due to defective formation of the gut endoderm.
To elucidate the ectopic PGC localization in the CKO embryo, a histological analysis of the 7–8-somite-stage embryos was performed. In the CKO embryos ( 4J–K″), many AP-positive PGCs were localized ectopically to the mesoderm (arrowheads). Fewer PGCs were found in the hindgut endoderm (arrows) of the CKO embryos (Fig. 4J–K”) when compared with the Lhx1 heterozygous embryo (Fig. 4I–I”). Overall 40.3% of AP-positive PGCs were localized outside of the gut in the 7- to 8-somite stage CKO embryos (total of 134 PGCs, n = 3 embryos), whereas only 6.9% of PGCs (total of 159 PGCs, n = 3 embryos) were not in the gut in the Lhx1 heterozygous embryos (Table 1; P < 0.001 by Chi-squared test). PGCs are, therefore, not confined to the hindgut in the CKO embryo.
Table 1. The Population of Alkaline Phosphatase (AP)-Positive Primordial Germ Cells (PGCs) Within and Outside Gut Endoderm of the Heterozygous (Lhx1+/flox, Lhx1flox/del, or Lhx1+/flox; Meox2-Cre) and Conditional-Null (Lhx1flox/del; Meox2-Cre) Embryos at the 7–8 Somite Stagea
No. of PGCs outside the gut (%)
No. of PGCs within the gut (%)
Numbers of PGCs were scored in serial sections of three embryos for each genotype.
Statistically different between the genotypes (χ2 = 46.9; P < 0.001 Chi-squared test).
Steel factor (Sl) function in the hindgut endoderm is involved in the regulation of PGC localization in the early-somite stage embryo (Gu et al.,2009). Confocal images of the E8.75 CKO embryos, which were subject to whole-mount immunostaining for Sl, revealed that Sl was strongly expressed in the mutant gut endoderm similar to that in the heterozygous embryo (Fig. 4L–M″). This result suggests that the dispersal of PGCs in the Lhx1 CKO embryo cannot be accounted for by changes in Sl activity in the gut endoderm. The functions of chemokine Sdf1 and the adhesion molecule E-cadherin in the embryonic tissues through which PGCs migrate are crucial for PGCs movement from the gut to the genital ridge (Ara et al.,2003; Molyneaux et al.,2003; Bendel-Stenzel et al.,2000; Gu et al.,2009). However, no discernible change in the expression of either Sdf1 or E-cadherin was detected in the E9.5 Lhx1-CKO embryos (see Supp. Fig. S1, which is available online). A proper c-kit/Sl signalling activity is an essential for PGC survival (Besmer et al.,1993). The early exit of PGCs from the Sl-expressing gut endoderm to the mesoderm where steel factor is not strongly expressed might expose the PGCs to a situation with reduced c-kit/Sl signalling activity. This may lead to the demise of the ectopically localized PGCs.
Ectopic Localization of PGCs Is Likely to Be Associated With Altered Ifitm1 Expression
We have noted that PGCs and the gut endoderm of E9.5 embryos did not express Lhx1 (Fig. 3; Shawlot and Behringer,1995; Tsang et al.,2000). The action of Lhx1 on the localization of PGCs is, therefore, likely to be mediated by signals emanating from the lateral plate mesoderm where Lhx1 is expressed (Shawlot and Behringer,1995; Tsang et al.,2000; Kwan and Behringer,2002). A potential candidate is Ifitm1, a member of the Ifitm family of proteins. Ifitm proteins that are localized on the cell surface can elicit homotypic cell-cell interaction. The over-expression and siRNA-mediated knockdown of Ifitm1/IFITM1 in cultured cell line cells changes cell migration and invasive capacity (Yang et al.,2005; Hatano et al.,2008). Different Ifitm proteins may be involved in navigating the movement or directing the localization of PGCs in the mouse embryo. For example, a repulsive action is generated between cells that differ in Ifitm1 activity. The repulsive forces produced by the Ifitm1-expressing somatic cells on the PGCs are postulated to facilitate the translocation of the PGCs from the mesenchyme in the primitive streak to the gut endoderm.
Ifitm1 expression is down-regulated in PGCs and it is not expressed in the gut endoderm, whereas it is expressed in the mesodermal tissues adjacent to the gut endoderm (Fig. 5B,C; Tanaka and Matsui,2002; Tanaka et al.,2005). This differential pattern of Ifitm1 expression in the two tissue compartments may facilitate the retention of the PGCs in the hindgut till they initiate migration through the mesentery to the genital ridges (Tam et al.,2006). To demonstrate the effect of Ifitm1 activity in influencing inter-compartment movement of cells, we ectopically expressed Ifitm1 in the endoderm of the mouse embryo from the late-streak stage onwards by lipofection (Supp. Fig. S2A). In these embryos, ectopic expression of Ifitm1 in the gut endoderm resulted in the absence of AP-positive PGCs from the endoderm (Supp. Fig. S2A′, B, C), suggesting that the translocation of PGCs from the mesoderm to the gut endoderm has been disrupted. Localized expression of Ifitm1 in the hindgut resulted in the regionalized exclusion of the AP-positive PGCs from the specific gut segment but not premature exit of the PGCs to the mesoderm, in contrast to the situation in the CKO embryo (Supp. Fig. S2D–F′; Tanaka et al.,2005). Ifitm1 activity in the immediate environment of the PGCs, therefore, has a significant impact on the localization of the PGCs.
To test if the dispersion of the PGCs from the hindgut in the Lhx1-deficient embryo may be related to aberrant Ifitm1 expression, in situ hybridization was performed on E7.5 and E8.75 CKO embryos. Ifitm1 expression in the E7.5 CKO embryos was not different from that of heterozygous embryos (Fig. 5A, A′), consistent with the normal localization of the PGCs to the posterior endoderm in the CKO embryos (Fig. 2A–E). In the early-somite (E8.75) stage wild-type embryo, Ifitm1 was differentially expressed in the hindgut: weak to absent in the endoderm and strong in the adjacent mesoderm (Fig. 5B, C; Tanaka et al.,2005). In contrast, in the CKO embryo, weak expression of Ifitm1 was found uniformly in the hindgut (and also in the paraxial and lateral plate mesoderm, Fig. 5B, C′). The abrogation of differential Ifitm1 expression between tissue compartments might undermine the ability to restrain the PGCs to the gut endoderm and thereby allows the premature departure of PGCs from the gut (Fig. 5D). This may account for the reduction of the resident PGC population in the gut and the ectopic localization of the PGCs outside the gut. The molecular mechanism whereby Lhx1 affects the expression of Ifitm1 is not known. However, in view of the fact that Ifitm1 activity is downstream of β-catenin (Lickert et al.,2005) and that the LIM1(LHX1)/LDB1/SSDP1 transcriptional complex may regulate the expression of certain WNT antagonists (Nishioka et al.,2005), it is plausible that Lhx1 might be connected with Ifitm1 expression via canonical WNT signaling.
A loss-of-function study of the Ifitm family of genes has been conducted in mice harboring an engineered chromosome with a 120-Kb deletion of the region containing Ifitm1and other family genes: Ifitm2, Ifitm3, Ifitm5 and Ifitm6. Surprisingly, losing the function of multiple Ifitm genes had no detectable effect on mouse PGC development or the generation of fertile mice (Lange et al.,2008). On the other hand, Iftim1 embryos derived from short hairpin RNA (shRNA) knock-down ES cells showed ectopic localization of PGC (Lickert et al.,2005; Tanaka et al.,2005). To resolve this discrepancy, it is imperative to analyze the PGC phenotype of the Ifitm1 knockout mice rather than the complex mutant with deletion of multiple family members. Our over-expression experiments in cultured mouse embryos have shown that the repulsive activity of Ifitm1 on PGC can be annulled when it was co-expressed with Ifitm3 (Tanaka et al.,2005). Therefore, it is possible that the combinatorial action of the Ifitm family proteins is crucial for delivering a specific cellular function and that the loss of Ifitm1 function can be compensated by the inactivation of other Ifitm proteins in the complex mutant. Examining PGC development in the single knockout mouse embryo will allow us to address the issue of the specific role of Ifitm1 in germ cell development.
Mice harboring a floxed (“flox”) Lhx1 allele with the coding region flanked by loxP sequences (Kwan and Behringer,2002) were crossed first to CMV-Cre mice to generate Lhx1+/del mice (“del” means deleted allele). Lhx1+/del mice were then crossed to Meox2-Cre mice, which express the Cre transgene in the epiblast at gastrulation (Tallquist and Soriano,2000) to produce Lhx1+/del; Meox2-Cre mice. The Lhx1flox/flox mice were crossed to Z/EG mice (CMV-pBA-loxPlacZloxP-EGFP; Novak et al.,2000) to generate Lhx1+/flox carrying a Z/EG allele, and then Lhx1+/flox; Z/EG mice were crossed to Lhx1flox/flox mice to generate Lhx1flox/flox carrying a Z/EG allele. The mating of Lhx1+/del; Meox2-Cre male and Lhx1flox/flox; Z/EG female mice produces embryos of the genotype Lhx1flox/del; Z/EG; Meox2-Cre, in which the floxed Lhx1 allele is ablated by Cre-recombinase activity (Tallquist and Soriano,2000) in the epiblast derivatives. Cells that have expressed Cre activity and potentially have lost Lhx1 activity can be identified by the expression of the EGFP reporter. EGFP-positive cells were counted using fluorescence microscopy with DAPI nuclear staining (Sigma-Aldrich, St. Louis, MO). Embryos of Pou5f1-GFP transgenic mice (Tanaka and Matsui,2002) were also used for this study. The use of animals was approved by the Animal Care and Ethics Committee of the Children's Medical Research Institute and the Children's Hospital at Westmead.
Histochemical Staining and In Situ Hybridization
To demonstrate AP activity, embryos were fixed with 2% paraformaldehyde (PFA) overnight at 4°C. After washing with phosphate buffered saline (PBS; pH 7.4) containing 0.1% Tween 20 (PBS-T), embryos were treated with 70% ethanol for at least 2 hr and then reacted with fast red and α-naphthyl-phosphate (Sigma-Aldrich) containing 5% borax (Sigma-Aldrich) in the staining solution. The specimens were fixed with 4% PFA overnight and embedded in OCT compound (Tissue-Tek, Torrance, CA). Then, frozen 10-μm sections were cut on a cryostat (Leica, Exton, PA). For the whole-mount immunostaining, embryos were fixed with 4% PFA in PBS overnight at 4°C. After being washed with PBS-T, embryos were treated with blocking solution (2% Boehringer blocking reagent and 5% fetal bovine serum in PBS containing 0.1% TritonX100) and then incubated with either an anti-Dppa3/Pgc7/Stella PGC-specific antibody (diluted at 1:1,000; Sato et al.,2002), an anti-Lhx1antibody (diluted at 1:100; Santa Cruz Biotechnology, Santa Cruz, CA), an anti-Steel factor (SCF) antibody (diluted at 1:50; R&D Systems, Minneapolis, MN), anti-E-cadherin antibody (diluted at 1:50; BD Biosciences, San Jose, CA) or anti-SDF1α antibody (diluted at 1:100; Abcam, Cambridge, MA) overnight at 4°C. The secondary antibodies were Alexa Fluor 594-conjugated anti-rabbit, Alexa Fluor 594-conjugated anti-goat, and Alexa Fluor 488-conjugated anti-mouse (diluted at 1:750; Molecular Probes, Eugene, OR) and fluorescence imaging was performed using an Olympus BX51 microscope and an Olympus DP72 image capturing system. Whole-mount in situ hybridization using a specific probe for Ifitm1 (240 bp) was as described (Tanaka and Matsui,2002). After fixation with 4% PFA overnight, stained specimens were embedded in paraffin wax (Paraplast; Oxford Labware) and sectioned at 10 μm.
Procedures for generating cDNA pools of PGCs and neighboring endoderm cells and whole gastrula embryo were as described (Tanaka and Matsui,2002). In brief, to generate cDNA pools of either PGCs or neighboring somatic cells in the hindgut endoderm, GFP-positive or -negative cells were collected from the dissociated cells of the hindgut tissue of the E9.5 Pou5f1-GFP transgenic mouse embryos using a micromanipulator (Narishige, Tokyo, Japan). About 50 collected cells were lysed by heat treatment at 6°C and first-strand cDNA was synthesized using a mixture of Moloney murine leukemia virus (Gibco/BRL, Gaithersburg, MD) and avian myeloblastosis virus reverse transcriptase (Roche, Nutley, NJ) with a poly-d(T) primer at 37°C for 15 min. Poly-d(A) was added onto the 3′ end of the first-strand cDNA by terminal transferase (Roche) at 37°C for 15 min and then amplified by PCR with poly-d(T) containing primer. PCR amplification in cDNA pools was performed using Taq DNA Polymerase (Roche) with 55 cycles of 30 s at 94°C, 30 s at 55°C, and 45 s at 72°C for Lhx1, Pou5f1 and Ifitm3, respectively. PCR amplification of Hprt was performed with 48 cycles of 30 s at 94°C, 30 s at 55°C, and 45 s at 72°C. The PCR products increased logarithmically with the cycles examined and the products were verified by direct sequencing. Primer sets for PCR amplification were as follows:
Lhx1, 5′-agcgaaggatgaaacagctaagcg-3′ and 5′-aagctcgtctctgtacaaccacg-3′;
Ifitm3, 5′-catcctttgcccttcagtgctgcc-3′ and 5′-acttcaggaccggaagtcggaatcc-3′;
Pou5f1, 5′-tcgagtatggttctgtaaccg-3′ and 5′-aatgatgagtgacagacaggc-3′;
Hprt, 5′-ggatttgaaattccagacaag-3′ and 5′-gcatttaaaaggaactgttgac-3′.
Whole-Embryo Electroporation and Lipofection
Procedures for plasmid construction of the CMV-Ifitm1-Ires-EGFP and for whole-embryo electroporation were described previously (Davidson et al.,2003; Tanaka et al.,2005). Procedure for whole-embryo lipofection was described elsewhere (Tanaka et al.,2011). In brief, to introduce the transgene into the surface layer of the embryo, such as a broad area of the endoderm layer, E7.5 embryos were soaked for 3 min in the lipofection solution, a mixture of the expression vector plasmid and LipofectAMINE 2000 (Invitrogen, Carlsbad, CA). Then, embryos were cultured for 24 hr in DMEM (Gibco/BRL) with 50% rat serum in 5% CO2 in air at 37°C, and the transgene expression was monitored by EGFP expression.
We thank David Loebel for comments on the manuscript and Bonny Tsoi for advice on in situ hybridization. Our work was supported in part by Mr. James Fairfax and a Grant-in-Aid for Scientific Research of Japan from the Ministry of Education, Science, Sports and Culture of Japan. P.P.L.T. is a Senior Principal Research Fellow of the National Health and Medical Research Council of Australia.