Gene trap screening as an effective approach for identification of Wnt-responsive genes in the mouse embryo

Authors

  • Yoshifumi Yamaguchi,

    1. Okazaki Institute for Integrative Biosciences, National Institutes of Natural Sciences, Myodaiji, Okazaki, Aichi, Japan
    2. Graduate School of Biostudies, Kyoto University, Kitashirakawa, Sakyo-ku, Kyoto, Kyoto, Japan
    Search for more papers by this author
  • Satoshi Ogura,

    1. Okazaki Institute for Integrative Biosciences, National Institutes of Natural Sciences, Myodaiji, Okazaki, Aichi, Japan
    2. Graduate School of Biostudies, Kyoto University, Kitashirakawa, Sakyo-ku, Kyoto, Kyoto, Japan
    Search for more papers by this author
  • Mio Ishida,

    1. Okazaki Institute for Integrative Biosciences, National Institutes of Natural Sciences, Myodaiji, Okazaki, Aichi, Japan
    2. Graduate School of Biostudies, Kyoto University, Kitashirakawa, Sakyo-ku, Kyoto, Kyoto, Japan
    Search for more papers by this author
  • Mika Karasawa,

    1. Department of Cell Biology, The Cancer Institute, Kamiikebukuro, Toshima-ku, Tokyo, Japan
    Search for more papers by this author
  • Shinji Takada

    Corresponding author
    1. Okazaki Institute for Integrative Biosciences, National Institutes of Natural Sciences, Myodaiji, Okazaki, Aichi, Japan
    2. The Graduate University for Advanced Studies (SOKENDAI), Myodaiji, Okazaki, Aichi, Japan
    • Okazaki Institute for Integrative Biosciences, National Institute of Natural Sciences, 5-1 Higashiyama, Myodaiji, Okazaki, Aichi, 444-8787, Japan
    Search for more papers by this author

Abstract

In this study, we examined whether gene trap methodology, which would be available for systematic identification and functional analysis of genes, is effective for screening of Wnt-responsive genes during mouse development. We screened out two individual clones among 794 gene-trapped embryonic stem cell lines by their in vitro response to WNT-3A proteins. One gene was mainly expressed in the ductal epithelium of several developing organs, including the kidney and the salivary glands, and the other gene was expressed in neural crest cells and the telencephalic flexure. The spatial and temporal expression of these two genes coincided well with that of several Wnt genes. Furthermore, the expression of these two genes was significantly decreased in embryos deficient for Wnts or in cultures of embryonic tissues treated with a Wnt signal inhibitor. These results indicate that the gene trap is an effective method for systematic identification of Wnt-responsive genes during embryogenesis. Developmental Dynamics 233:484–495, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

The development of multicellular organisms requires a concerted and sequential expression of a variety of genes. Secreted signal proteins, including bone morphogenetic protein (BMP), fibroblast growth factor (FGF), and Wnt, activate various cellular signaling pathways, which then trigger intracellular molecular networks of gene expression during animal development. To gain more insight into the mechanism underlying a variety of developmental events in which these signalings are involved, further identification and functional analysis of components in the molecular cascades and networks activated by these signal are required.

Canonical Wnt signaling is involved in several aspects of development, such as axis determination, cell differentiation, and cell proliferation (Cadigan and Nusse, 1997; Nelson and Nusse, 2004). WNT proteins, which are cysteine-rich secreted molecules, bind to their transmembrane receptors, the Frizzled proteins, and its coreceptor, the LDL receptor-related protein 5 and 6 (LRP5/6; Mao et al., 2001a, b), and transduce a signal into the cell, leading to the translocation of β-catenin to the nucleus and to subsequent activation of target genes by transcriptional complex of β-catenin and TCF/LEF-1 (Behrens et al., 1996; Huber et al., 1996; Schneider et al., 1996). For an understanding of the molecular mechanism governing each developmental phenomenon regulated by the Wnt signaling, identification and functional analysis of Wnt-responsive genes are indispensable. Thus far, many Wnt-responsive genes have been identified in vertebrates; however, only a few of these genes have been characterized in the context of embryogenesis (Danielian and McMahon, 1996; Yamaguchi et al., 1999; Lickert et al., 2000; Kratochwil et al., 2002; Kioussi et al., 2002; Morkel et al., 2003).

The gene-trap methodology is a powerful strategy for the systematic identification and functional analysis of genes in the postgenomic era, because this methodology offers the identification of a novel gene, the analysis of its expression pattern, and the generation of its functional mutation in a single experimental approach as described below. In this methodology, random insertion of a gene-trap vector leads to the tagging, and frequently to the disruption, of genes across the genome (Friedrich and Soriano, 1991; Joyner et al., 1992; Skarnes et al., 1992; Hicks et al., 1997; Stanford et al., 2001). Therefore, if embryonic stem (ES) cells are used for the generation of insertion events, embryonic and adult whole bodies containing tagged and disrupted alleles can be produced. Such an insertion event affords the following advantages for identification and functional analysis of a trapped gene: (1) Nucleotide sequences of a trapped transcript and an insertion site can be determined by 5′ rapid amplification of cDNA ends (RACE) and plasmid rescue method, respectively. The process of identification of a trapped gene has been substantially eased by the recent completion of the genome database. (2) The expression pattern of a trapped gene during development can be easily monitored by the expression of protein tag derived from a gene-trap vector in embryos generated from trapped ES cells. (3) An insertion event has the potential to be mutagenic. Because of these strong advantages, the gene trap methodology has been applied for several studies, including large-scale insertional mutagenesis programs (Wurst et al., 1995; Stoykova et al., 1998; Stanford et al., 2001; Hansen et al., 2003; Stryke et al., 2003).

Although the gene-trap methodology basically involves the random integration of a gene-trap vector, this strategy has also been expected to be available for identification and characterization of genes regulated by particular signals. In a modified gene-trap strategy, called induction gene trap, genes are selected by their response to specific secreted signal proteins, including BMP2, activin, nodal, and FGF, added to the culture of gene-trapped cells (Forrester et al., 1996; Harrison and Miller, 1996; Stoykova et al., 1998; Mainguy et al., 2000; Akiyama et al., 2000; Medico et al., 2001; Kluppel et al., 2002; Tateossian et al., 2004). Although the in vivo correlation between selected genes and signals has remained unclear in these cases, in vitro prescreening of trapped ES cell lines with secreted signal proteins would be effective for enrichment of genes regulated by specific signals even in normal development.

Here, we used the induction gene-trap approach to identify Wnt-responsive genes during mouse development. We screened 794 trapped ES lines and recovered two ES cell lines that contained trapped genes responsive to WNT-3A protein. The embryonic expression, as well as the requirement of Wnt genes for their in vivo expression, was precisely examined in these two clones. These results strongly indicate that this approach is a powerful strategy for identification, and probably for functional analysis, of genes responsive to Wnt signaling during embryogenesis. Our results are the first clear indication that induction gene trapping can be used for identification of genes regulated by secreted signal proteins during normal development.

RESULTS

Isolation of Wnt-Responsive Gene-Trap Lines

To determine the culture period adequate for screening Wnt-responsive genes in ES cells, we first examined the time course of response to Wnt signal after the addition of medium conditioned by Wnt-3a–expressing cells (Wnt-3a C.M.; Shibamoto et al., 1998). Because Wnt-3a is known to be a typical signal to induce the canonical Wnt pathway, stimulation by Wnt-3a C.M. would be representative of that by many Wnt molecules that can induce this pathway, for instance, Wnt-1, Wnt-3, Wnt-8, Wnt-7b, and some other Wnt family members. The response to the Wnt signal was examined by monitoring EGFP reporter expression, which was driven from a promoter containing a tandem repeat of seven TCF-binding sites (Ueda et al., 2002). In the absence of Wnt-3a C.M., as well as in the presence of control C.M., the enhanced green fluorescent protein (EGFP) expression level was very low in almost all cells, indicating that the endogenous activation level of the canonical Wnt signaling is very low in ES cells (Fig. 1). On the contrary, the number of cells that expressed the EGFP signal gradually increased after 6 hr of incubation with Wnt-3a C.M. and reached its maximum after 24-hr incubation (Fig. 1). In parallel to this activation, the expression of brachyury, which has been shown to be activated by Wnt signaling in ES cells (Arnold et al., 2000), was also gradually induced (data not shown). Thus, a 24-hr incubation was considered to be long enough for screening Wnt-responsive genes with maximum sensitivity in ES cells.

Figure 1.

Induction of a Wnt reporter gene in embryonic stem (ES) cells. Wnt-3a conditioned medium (Wnt) and control conditioned medium (C) were added to ES cells stably containing an enhanced green fluorescent protein (EGFP) reporter gene that was driven from a promoter containing a tandem repeat of seven TCF-binding sites. The response to Wnt signal was examined by monitoring EGFP reporter expression under the fluorescence microscope at 1, 3, 6, 12, and 24 hr after the addition of conditioned medium. Phase-contrast views are shown above fluorescence views at each time point. The number of cells that expressed the EGFP signal gradually increased after 6 hr of incubation with Wnt-3a C.M. and reached maximum by 24 hr.

For screening Wnt-responsive genes by the gene-trap approach, we established several ES cell lines in which gene-trap vectors were randomly integrated into the chromosomal DNA. The vector used in this study, called pLSAβgeo, contains an engrailed-2 splicing acceptor sequence linked to the βgeo reporter gene (a lacZ-neomycin phosphotransferase fusion gene) with an internal ribosome entry sequence (IRES) at its 5′ end. ES cells were electroporated with this trap vector, and then G418-resistant ES cell colonies were picked up and split into two 96-well plates. The cells on the first plate were frozen for storage, whereas those on the second plate were allowed to grow, after which they were further divided into two gelatinized plates. For screening these clones in terms of their response to Wnt proteins, the divided cells on the second plates were cultured for 24 hr with either Wnt-3a C.M. or control C.M. Before the addition of the C.M., the ES cells on the second plates were allowed to differentiate for 24 hr without the addition of leukemia inhibitory factor (LIF), because we aimed at obtaining Wnt-responsive genes that play roles in early differentiated cells. As a result of screening of 794 individual transfectants, we obtained two cell lines (clone 43 and clone 5) in which β-gal activity was strongly increased by Wnt-3a C.M. (Fig. 2; Table 1). On the other hand, β-gal activity was decreased by Wnt-3a C.M in 1 cell line, whereas its level was maintained without any drastic change in 694 other cell lines (Table 1).

Figure 2.

Wnt-responsive expression of gene trapped lines. AD: Two trapped embryonic stem (ES) cell lines (clone 5, A,B; and clone 43, C,D) were incubated with medium containing control C.M. (A,C) or Wnt-3a C.M. (B,D) for 24 hr and were stained with 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal) for monitoring β-galactosidase activity.

Table 1. Summary of Gene Trap Screening in Embryonic Stem Cells
 Number of colonies
Resistant for G418794
 Positive for β-gal activity657
  Increased β-gal activity by Wnt-3a2
  Decreased β-gal activity by Wnt-3a1
  Unchanged654
 Negative for β-gal activity137

Molecular Characterization of the Trapped Genes

To identify the trapped genes molecularly and to confirm that their expression was actually regulated by Wnt signal, we cloned the trapped genes responsible for Wnt-dependent expression in clones 43 and 5. Genomic Southern blot analysis using a lacZ probe, which can detect restriction fragments containing junctions between the trap vector and chromosomal DNA, indicated that the trap vector was integrated into a single locus of the genome in both trapped lines (data not shown). Then, the integration sites were determined by the plasmid rescue method, and transcripts from the trapped loci were characterized by 5′-RACE on RNA isolated from each trapped ES cell line.

In the case of clone 43, the trap vector was inserted in the second intron of the CP2-related-transcription-repressor-1 (CRTR-1) locus (Fig. 3A; Rodda et al., 2001). Characterization of a transcript from the clone 43 trapped allele by 5′-RACE demonstrated that a fusion transcript had been generated by a splicing between the splicing donor site at the end of exon 2 of CRTR-1 and the engrailed-2 splicing acceptor in the trap vector.

Figure 3.

Characterization of the trapped alleles and expression of the trapped genes. A: Schematic representation of the two trapped alleles. B: Genomic sequences around the insertion site of the gene-trap vector in the clone 5 trapped line (cl.5). An underline indicates the sequences obtained by 5′-cap trapper rapid amplification of cDNA ends. An arrowhead points at the 5′-cap site. Sequences of the endogenous transcript confirmed by a reverse transcriptase-polymerase chain reaction (RT-PCR) experiment (GenBank accession no. AB193402) are enclosed by a box. The presumptive amino acid sequence is indicated below the box. Arrows show sequences used for the RT-PCR in C,D. Bases denoted in gray were deleted upon insertion of the trap vector into the genome. C: Increase in mRNA of CRTR-1 and cl.5 in embryonic stem cells cultured with Wnt-3a C.M.(WNT) for 24 hr compared with those with control C.M (con) or those before incubation (0h). Because the RT-PCR product of cl.5 is indistinguishable in size from that amplified from its genomic DNA, we examined it by comparison of with (+RT) and without (−RT) reverse transcriptase in the reaction. Hypoxanthine phosphoribosyltransferase (HPRT) as an internal control. Right panel: Quantitative representation of mRNA of CRTR-1 and cl.5 normalized to HPRT mRNA. The asterisk indicates P vs. 0 hr < 0.005, the dagger P vs. control (L.C.M.) < 0.005. D: The endogenous transcript of cl.5 was detected with RT-PCR on total RNA from day 11.5 embryonic heads by using primers shown in B.

In the case of clone 5, the gene-trap vector was inserted in chromosome 13 (Fig. 3A,B). Upon this insertion event, the splicing acceptor and a part of the 5′ region of the IRES sequence had been deleted from the vector, whereas the lacZ gene remained intact. Thus, the lacZ gene in this trapped allele was supposedly expressed without splicing. The genomic DNA around the insertion site was also deleted upon this insertion. A modified 5′-RACE method that can capture only 5′-capped mRNA revealed that a transcript in this trapped allele was transcribed starting 24-bp upstream from the integration site without splicing (Fig. 3B). In addition, reverse transcriptase-polymerase chain reaction (RT-PCR) analysis indicated that a transcript containing a sequence from the initiation site to 247 bp downstream to the integration site was generated from this locus in E11.5 embryos as well as in the wild-type ES cells (Fig. 3B–D). One presumptive open reading frame encoding a protein with a molecular weight of 5.8 kD was found in this transcript, and this protein showed no similarity in amino acid sequence to known proteins in the data base.

To examine whether the expression of the endogenous genes at the trapped loci was actually up-regulated by Wnt signaling, we monitored their expression in wild-type ES cells in the absence or presence of WNT-3A proteins by RT-PCR. The mRNA amounts of CRTR-1 and of the clone 5 gene were up-regulated by the addition of Wnt-3a C.M. by 2.5- and 1.5-fold, respectively, compared with their amount obtained with control C.M. (Fig. 3C). Thus, expression of the endogenous genes at the trapped loci was actually induced by Wnt signaling as in the case of the trapped alleles.

Close Correlation Between the Trapped Genes and Wnts in Their Expression Patterns During Embryogenesis

To examine the expression of these trapped genes during embryogenesis and the correlation between their expression and that of Wnt genes, we generated mice carrying these trapped alleles. Heterozygotes for these trapped alleles were produced by intercrossing of wild-type females with chimeric male mice that had been generated by injection of the trapped ES cells into C57BL/6 blastocysts. Both lines exhibited particular spatiotemporal lacZ expression patterns at mid-gestation stages in heterozygous embryos. Furthermore, in the case of clone 43, in situ hybridization analysis also indicated that the expression pattern of the lacZ reporter was identical to that of the endogenous gene as far as was examined. On the other hand, in the case of clone 5, no endogenous gene expression was detectable by in situ hybridization, probably because the expression level was too low and the length of this mRNA was too short. Therefore, we further observed the embryonic expression of clone 43 and clone 5 trapped genes precisely by monitoring LacZ-positive cells in addition to in situ hybridization analysis of clone 43.

Although the expression of CRTR-1, the trapped gene in clone 43, in the kidney at E16.5 and in adulthood was already described (Rodda et al., 2001), its precise expression pattern has remained unclear. Therefore, we examined its expression pattern at several embryonic stages. The expression of CRTR-1 was temporarily observed in the inner cell mass of blastocysts (data not shown). Later, CRTR-1 was specifically expressed in ductal structures in the mid-gestation stages. This gene was expressed during several aspects of the ductal morphogenesis in kidney development. The urogenital expression of CRTR-1 was observed first in the nephric duct and mesonephric tubules at E10.5 (Fig. 4A) and later in the Wolffian duct, the ureter, the collecting duct, and the distal portion of tubules connected to the collecting duct in the metanephros (Fig. 4B,C,E). The expression was restricted to the epithelium, i.e., was not found in the mesenchyme, in these organs (Fig. 4E,F). CRTR-1 was also expressed in the ductal epithelium of the salivary glands (submandibular, sublingual, and parotid glands) at E 15.5 (Fig. 4F,G and data not shown).

Figure 4.

Embryonic expression of the clone 43 trapped gene. A,B: During urogenital development, the lacZ reporter of clone 43 was expressed first in the nephric duct and mesonephric tubules at embryonic day (E) 10.5 (A); at E13, it was also expressed in the Wolffian duct, mesonephric tubules, the ureter, and the collecting duct (B). C,D: CRTR-1 mRNA (C) was coexpressed with Wnt-7b in the Wolffian duct (D, arrows). E: In situ hybridization with CRTR-1 antisense riboprobe on E14.5 kidney sections. F: A section of 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal) -stained submandibular gland (SMG) at E15.5. The LacZ reporter of clone 43 was expressed in the duct of the developing kidney and SMG. G,H: Expression of CRTR-1 mRNA (G) and Wnt-5b (H) in the SMG and sublingual gland at E13.5.

Of interest, the expression of several Wnt genes was also observed in the ductal structures in which CRTR-1 was expressed. In urogenital development, Wnt-7b, which has been reported to be expressed in the Wolffian duct, the ureter, and the collecting duct at E13.5 (Kispert et al., 1996; Patterson et al., 2001), was coexpressed with this trapped gene (Fig. 4C,D). On the other hand, in the submandibular gland (SMG), the expression of several Wnt genes, including Wnt-2, 2b, 3, 4, 5b, 6, 10b, 14, 16, was detected by RT-PCR analysis (data not shown). Among these Wnt genes, Wnt-5b exhibited an expression pattern spatially and temporally correlated with that of CRTR-1. Wnt-5b was expressed at mid-gestation stages, for instance E13, in the stalks of the SMG and the sublingual gland (SLG), where CRTR-1 was expressed (Fig. 4G,H). Thus, the ductal expression of CRTR-1 was well correlated with the expression of Wnt-7b and Wnt-5b in the kidney and the salivary gland, respectively.

The expression pattern of the clone 5 trapped gene, also suggested its close correlation with that of Wnt genes. The expression of the clone 5 trapped gene was first observed at E8.5 in rhombomere 5 (Fig. 5A). At E9.5, the expression of this gene was still detected in rhombomere 5, although it was relatively dispersed (Fig. 5B). Rostral to rhombomere 5, the lacZ-positive cells straggled in the superior membrane of the rhombencephalon. At E10.5–11.5, the expression was observed with strong intensity in scattered neural crest cells over the dorsal diencephalon and mesencephalon (Fig. 5C,D), as well as in relatively weak intensity over the dorsal spinal cord. Of interest, these scattered signals were strong along the dorsal midline adjacent to the roof plate where Wnt-1 and Wnt-3a were expressed (Roelink and Nusse, 1991; Parr et al., 1993). The clone 5 trapped gene was also expressed in mesenchyme in the telencephalic flexure (Figs. 5E, 6A). This mesenchymal expression represents another example of correlation between this gene and Wnt, because Wnt-3a is expressed in the neuroepithelium, adjacent to the mesenchymal cells, in the telencephalic flexure (Fig. 5F; Roelink and Nusse, 1991; Lee et al., 2000). The expression of the clone 5 trapped gene was also observed in the meninx at E13.5 (Fig. 5G). Taken together, the close correlation between gene expression of the two trapped genes and that of several Wnt genes strongly suggests that the trapped genes screened by their in vitro response to WNT proteins were also responsive to Wnt signals in vivo.

Figure 5.

Embryonic expression of the clone 5 trapped gene. A–E,G: LacZ reporter expression of the clone 5 trapped gene was examined by 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal) staining of heterozygous embryos for the clone 5 trapped allele. AD: A dorsal view at embryonic day(E) 8.5 (A), a dorsal view at E9.5 (B), a sagittal view at E10.5 (C), and a transverse section around hindbrain at E11 (D). E: A sagittal section around the telencephalic flexure at E11. In the telencephalic flexure, the clone 5 trapped gene was expressed in mesenchyme but not in neuroepithelium or surface ectoderm. F: For comparison of the expression of the clone 5 trapped gene with that of Wnt-3a, in situ hybridization with the Wnt-3a probe was performed. A frontal view of Wnt-3a expression at E10.5 is shown. Note that the telencephalic mesenchyme was surrounded by the cortical hem where Wnt-3a was expressed. G: A transverse section of the hindbrain at E13.5. The clone 5 trapped gene was expressed in the meninx. ov, otic vesicle; di, diencephalon; te, telencephalon.

Figure 6.

Down-regulation of the expression of the clone 5 trapped gene in Wnt-3a and Wnt-1/Wnt-3a mutants. AD: Frontal views of the expression of the lacZ reporter for the clone 5 trapped gene in the telencephalic flexure (arrow) at embryonic day (E) 11.5. EH: Dorsal views of the expression of the lacZ reporter for the clone 5 trapped gene in the hindbrain at E11.5. A–H: Wild-type (A,E), Wnt-3a null mutant (B,F), Wnt-1 null mutant (C,G), and Wnt-1 and Wnt-3a double-null mutant (D,H).

Requirement of Wnt Signaling for the Expression of the Trapped Genes

To examine whether Wnt genes are actually required for the in vivo expression of the trapped genes, a genetic analysis would be powerful. Because the expression of the clone 5 trapped gene correlated well with that of Wnt-1 and Wnt-3a, we next examined the expression of this trapped gene in mouse embryos deficient for Wnt-1, Wnt-3a, or both of these Wnts. The positive LacZ signal of the clone 5 trapped gene was absent in neural crest cells migrating from the hindbrain and the spinal cord in the Wnt-1/Wnt-3a double mutant (Fig. 6H). This absence of LacZ-positive cells appears to have been caused by decreased expression of the clone 5 trapped gene, not by lack of cells that normally express this gene, because the neural crest cells were definitely present, although the number of migrating neural crest cells was decreased in this mutant (Ikeya et al., 1997). Alternatively, it is also possible to speculate that Wnt-1 and Wnt-3a might act on a certain population of neural crest cells through the activation of the clone 5 trapped gene, in which case the absence of these Wnts might result in a selective decrease in this population. In contrast, the lacZ signal was present in the wild-type and in Wnt-1 and Wnt-3a single mutant embryos (Fig. 6E–G). These results strongly suggest that Wnt-1 and Wnt-3a were redundantly required for the expression of the clone 5 trapped gene in this region. In addition, the 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal) staining in the mesenchymal cells in the telencephalic flexure was significantly reduced in the Wnt-3a null mutant, as well as in Wnt-1/Wnt-3a double mutant, compared with that in the wild-type and Wnt-1 single mutant embryos (Fig. 6A–D). Thus, the expression of the clone 5 trapped gene in the mesenchyme was dependent on Wnt-3a. Because Wnt-3a is expressed in the neuroepithelium adjacent to the mesenchyme (Fig. 5F), but not in the mesenchyme, in the telencephalic flexure, this result suggests an inductive interaction between the neuroepithelium and the mesenchyme. Together, these results indicate that expression of the clone 5 trapped gene required the activity of Wnt even in the in vivo situation.

To investigate whether Wnt signaling is also required for the expression of CRTR-1 during organogenesis, we cultured submandibular and sublingual glands isolated from E13 embryos with or without CKI7, a chemical inhibitor of canonical Wnt signaling (Peters et al., 1999). In this culture, the expression of CRTR-1 was markedly diminished in comparison with the expression of keratin18, which is also expressed in the ductal epithelium, in the duct of the SMG and the SLG treated with CKI7 (Fig. 7A–D). Quantitative PCR experiments showed that the expression of CRTR-1 mRNA in cultures with CKI7 was decreased to 40% of that of control cultures, whereas that of keratin18 was not significantly changed in this condition (data not shown). Thus, Wnt signaling was required for the normal expression of CRTR-1 in the SMG and the SLG. Taken together, our results indicated that the in vivo expression of the two trapped genes, which were screened by their in vitro response to Wnt, was also dependent on Wnt activity.

Figure 7.

Down-regulation of CRTR-1 expression in cultured salivary gland by a Wnt inhibitor. An inhibitor for the Wnt signaling, CKI7, caused down-regulated CRTR-1 expression. A,B: In situ hybridization with the CRTR-1 probe of embryonic day 13 submandibular gland and sublingual gland cultured for 12 hr with dimethyl sulfoxide (DMSO, A) or CKI7 (B). C,D: The expression of keratin18, which was detected in the ductal epithelium, was not altered between cultures treated with DMSO (C) and CKI7 (D).

Evidence That a Gene Trap Event Is Mutagenic

To examine whether the trapped gene was required for proper development of cells or organs where it was expressed, we generated homozygous mutants for the CRTR-1 trapped allele by crossing the heterozygous males and females. In homozygotes for this allele, the expression of this gene was almost completely diminished, whereas the expression of a truncated transcript fused to the β-geo gene in the trap vector was strongly detected (Fig. 8A). Of interest, 70% of the mice homozygous for the CRTR-1 trapped allele died before 5 weeks after birth. Most of these mice exhibited hypoplasia of the kidney at postnatal day 30, whereas the heterozygous and wild-type mice in the same litters showed no obvious defect in their kidney (Fig. 8B,C). Furthermore, in homozygous mice, tubules in the cortex were occasionally dilated and the papilla appeared apart from the ureter. Thus, CRTR-1 is likely to be required for proper development of the kidney, in which this gene is expressed. More extensive analysis of this mutant phenotype at molecular and physiological levels will be described elsewhere.

Figure 8.

Defect in the kidney of homozygotes for the clone 43 trapped allele. A: Northern blot analysis to indicate the expression of CRTR-1 mRNA in the kidney of wild-type (+/+), heterozygote (+/tra), and homozygote (tra/tra) for the clone 43 trapped allele at postnatal day (P) 0. In homozygotes, CRTR-1 mRNA was almost diminished, whereas the expression of truncated transcript, in which the first two exons were fused to β-geo gene, was strongly detected. B,C: Histological sections of kidneys of a heterozygote (B) and a homozygote (C) for the clone 43 trapped allele at P0. Most of in the CRTR-1 homozygous mice exhibited hypoplasia.

Thus, the in vivo expression of the two trapped genes, which were screened by their in vitro response to Wnt, was also dependent on the Wnt activity. Furthermore, homozygotes for a trapped allele showed a morphological phenotype in the kidney, where the trapped gene was expressed overlappingly with Wnt-7b. These results indicate that an inductive gene trap in ES cells is likely to be effective for screening and functional analysis of genes induced by Wnt signaling during embryogenesis.

DISCUSSION

Gene Trap Screening as an Efficient Approach for Identification and Functional Analysis of Wnt-Responsive Genes

To identify genes regulated by Wnt signaling during embryogenesis, we established a gene-trap screening system, in which gene-trapped ES cell lines were selected in terms of their response to Wnt proteins. Among 794 clones screened, two clones exhibited a strong increase in in vitro reporter gene expression in response to Wnt-3a C.M. (Fig. 2). We examined temporal and spatial expression of these genes in embryos and found that the expression patterns of these trapped genes correlated well with those of several Wnt genes (Figs. 4, 5). Furthermore, the expression of the trapped genes was regulated by Wnt signaling not only in vitro but also in vivo (Figs. 6, 7). Thus, the gene-trap approach coupled with a primary screening with WNT proteins appears to be effective for identification of genes actually regulated by Wnt signaling during embryogenesis.

Furthermore, the gene-trap approach has the strong advantage of not only being a screening method, but also being a tool for mutagenesis. Mutant mice can be generated from trapped ES cells if integration of the trap vector blocks the production of a gene product with normal function. Actually, homozygotes for the clone 43 trapped allele showed abnormal development of the kidney (Fig. 8). Thus, the gene-trap approach is a powerful and systematic strategy not only for identification but also for functional analysis of genes regulated by Wnt signals during mouse development. Further extensive studies should reveal the precise function of this gene in kidney development and its relation to Wnt-7b, which is overlappingly expressed with CRTR-1 in the developing kidney.

Variety of Genes That Can Be Obtained by Gene-Trap Screening With Wnt-Treated ES Cells

Wnt signaling plays roles in several different aspects of embryogenesis and is operative even after birth. Different sets of genes should be induced or repressed by this signal, depending on the cellular context. Thus, the variety of Wnt-responsive genes that we can obtain by gene-trap screening would reflect the characteristics of cells used for the screening. ES cells, which we used for screening of Wnt-responsive genes in this study, possess self-renewing activity as stem cells, as well as pluripotency to differentiate into all cell types. Under the in vitro culture conditions used in the present study, brachyury and cdx-1, both of which are known to be induced by Wnt signals in mesodermal cells in early embryonic stages (Yamaguchi et al., 1999; Arnold et al., 2000; Lickert et al., 2000; Ikeya and Takada, 2001; Prinos et al., 2001), were induced by WNT proteins (data not shown). In addition to these mesodermal genes, we also detected Wnt-induced expression of the light polypeptide of neurofilaments, which is expressed in neural cells from an early differentiated stage (data not shown). Thus, the culture conditions used for ES cells in this study appear to mimic some events at early embryonic stages and to be effective for obtaining genes responsive to Wnt signaling during early embryogenesis. Actually, CRTR-1 was expressed in the inner cell mass of blastocysts, and the clone 5 trapped gene was first expressed in rhombomere 5 at E8.5, suggesting that biological events in blastocysts and in early neural development seem to be reproduced at least to some extent under the in vitro culture conditions used for this screening. However, it also seems true that these in vivo events are not exactly reproduced under this in vitro condition, especially in terms of the temporal regulation of gene expression. For instance, expression of the clone 5 trapped gene was increased in ES cells cultured for 24 hr with Wnt-3a C.M.; although its earliest in vivo expression was not detected at the appropriate time expected by the in vitro expression, but later, at E8.5. Thus, the in vitro culture conditions for this screening might accelerate and/or bypass some events in early development; thus, the screening strategy in this study may have a bias for a variety of certain genes. In addition to the characteristics and culture conditions of the cells, the characteristics of the gene-trap vector affects the variety of genes that can be obtained. The gene-trap vector used in the present study can capture genes expressed in undifferentiated ES cells, whereas it cannot those exhibiting no expression in these cells. Further improvement of cell culture conditions and design of the gene-trap vector should be effective for obtaining genes with different characteristics, for instance, those expressed in late embryonic stages.

One of the important applications of the present gene-trap approach is a selective screening of genes involved in particular biological phenomena regulated by Wnt signal. As mentioned above, ES cells possess pluripotency to differentiate into all cell types. Recently, several lines of evidence have suggested that activation of Wnt signaling in human and mouse ES cells leads to inhibition of differentiation and maintenance of pluripotency (Aubert et al., 2002; Kielman et al., 2002). Thus, by modifying the culturing conditions of gene-trapped ES clones, some Wnt target genes involved in the machinery for maintaining pluripotency might be identified. On the other hand, several culture conditions in which ES cells efficiently differentiate into particular cell types, including neuronal, mesodermal, and endodermal cells, have been established (Nishikawa et al., 1998; Kawasaki et al., 2000; Mizuseki et al., 2003; Kubo et al., 2004). Thus, a gene-trap approach with more refined and lineage-restricted in vitro screening could be available for identifying Wnt target genes in some particular cell lineages.

Roles of the Trapped Genes in Clone 43 and Clone 5 During Mouse Embryogenesis

The trapped gene of clone 43 was identified as CRTR-1, a member of the CP2/LSF/Grh transcription factor family (Rodda et al., 2001; Venkatesan et al., 2003). In vertebrates, this family consists of several members, including CRTR-1 (known as LBP-9 in human; Huang and Miller, 2000), CP2 (also known as LBP-1c and LSF; Lim et al., 1992; Shirra et al., 1994), NF2d9 (known as LBP-1a in human; Sueyoshi et al., 1995; Yoon et al., 1994), mammalian Grainyhead (MGR), brother of MGR (BOM; Wilanowski et al., 2002), sister of MGR (SOM)/GET1 (Kudryavtseva et al., 2003; Ting et al., 2003b), but their developmental roles have been largely unknown (Ramamurthy, 2001; Ting et al., 2003a). We found that CRTR-1 was specifically expressed in the ductal epithelium of the kidney and the salivary glands (Fig. 4). The developmental processes of these organs share many similarities; for instance, epithelial–mesenchymal interaction, branching morphogenesis, and lumen formation. Thus, CRTR-1 may be commonly involved in the development of different organs by playing a role in some of these phenomena. Of interest, a Drosophila member of the CP2/LSF/Grh family, Grainy head (Uv et al., 1994), controls luminal elongation of the airways. Overgrowth of the apical membrane in grainy head mutants leads to lumen elongation without affecting epithelial integrity, whereas overexpression of the gene limits luminal growth (Hemphala et al., 2003). Thus, CRTR-1 may have a role in luminal development during vertebrate organogenesis, as in the case of its homolog in Drosophila.

It has been reported that several Wnt genes are expressed in the ductal epithelium where CRTR-1 was expressed. In the SMG and the SLG, Wnt-5b was expressed in the stalk (Fig. 4H), where CRTR-1 was expressed, whereas Wnt-2b were expressed in the mesenchyme (Lin et al., 2001). On the other hand, in the developing kidney, Wnt-6 and Wnt-7b (Kispert et al., 1996; Patterson et al., 2001; Itaranta et al., 2002) are expressed in the collecting duct, whereas Wnt-2b is expressed in the surrounding mesenchyme around ureteric buds. Also, Wnt-4, which is required for tubulogenesis, is expressed in the condensed mesenchyme and in the newly formed distal tubule (Stark et al., 1994). Among these Wnt genes, Wnt-7b showed expression tightly coincident with that of CRTR-1 in the Wolffian duct, the ureter, and the collecting duct (Patterson et al., 2001). Since Wnt-7b has been reported to induce canonical Wnt signaling in mammalian cells in culture, the overlapped expression of CRTR-1 with Wnt-7b suggests that the CRTR-1 expression induced by Wnt-3a, a typical inducer for the canonical Wnt signaling, was also induced through canonical Wnt signaling by Wnt-7b (Zhang et al., 2004). Thus, Wnt-7b seems to be the best candidate signal to regulate CRTR-1 expression in the developing kidney. However, because the role of Wnt-5b and Wnt-7b in salivary gland and kidney development remains unclear, we cannot now speculate on the role of CRTR-1 based on studies of Wnt genes.

The clone 5 trapped gene was expressed in a part of the migrating neural crest cells caudal to the diencephalon and in the mesenchyme in the telencephalic flexure. In neural crest development, Wnt signaling plays several important roles. Wnt-1 and Wnt-3a, which are expressed in the roof plate of the neural tube, redundantly regulate proper expansion of neural crest precursor cells, and β-catenin–mediated Wnt signaling also regulates the specification of subtypes of neural crest cells (Ikeya et al., 1997; Dorsky et al., 1998; Lee et al., 2004; Lewis et al., 2004). Because the expression of the clone 5 trapped gene in the hindbrain and around the dorsal neural tube in the trunk was significantly reduced in the Wnt-1/Wnt-3a double mutant (Fig. 6H), this gene may be involved in some aspect of neural crest development activated by both Wnt-1 and Wnt-3a. In addition, the expression at a later stage was found in the meninx, which may originate from the neural crest (Fig. 5G; Couly et al., 1993), suggesting that the clone 5 trapped gene might be involved in the development of the meninx. On the other hand, clone 5 expression in the mesenchyme in the telencephalic flexure of the Wnt-3a null mutant and in the Wnt-1/Wnt-3a double mutant was significantly reduced compared with that in wild-type embryo and Wnt-1 null mutant, respectively (Fig. 6A–D). This finding clearly showed that the expression of the clone 5 trapped gene in the mesenchyme required Wnt-3a, which is expressed in the neuroepithelium of the telencephalic medial wall (Roelink and Nusse, 1991). It has already been described that Wnt-3a acts locally in the telencephalic flexure as an autocrine signal to regulate the expansion of the neuroepithelial cells from which the hippocampus develops (Lee et al., 2000), but our results have revealed a new role of Wnt-3a, as a paracrine signal in this region.

The results of this study strongly suggests that the induction gene trap approach in ES cells is an effective one for screening the downstream target genes of Wnt signaling during embryogenesis. We have been generating and analyzing mice homozygous for each trapped gene to investigate their function in vivo. Extensive studies with these mutants should reveal the roles of these Wnt-responsive genes and molecular machinery activated by Wnt signal in several aspects of vertebrate development.

EXPERIMENTAL PROCEDURES

Gene Trap Vector, Selection of Wnt-Responsive Cell Lines, and Generation of Mice

pLSAβgeo was modified from SA-IRESβgeo (Mountford et al., 1994). Briefly, lox71 sequences (Araki et al., 2002) were inserted into a BamHI site immediately upstream of the en-2 splicing acceptor.

CJ7 ES cells were maintained on primary embryonic fibroblasts in DMEM containing 15% fetal calf serum (FCS) and LIF (1,000 U/ml). ES cells (1 × 107) were electroporated with 25 μg of linearized gene-trap vector DNA in 1 ml of phosphate buffered saline (PBS), by applying two pulses of 0.23 V, 500 μF. After 5-min incubation on ice, the cells were plated in 10-cm dishes and allowed to recover for 24 hr before adding 200 μg/ml G418 (Invitrogen) for selection of neomycin-resistant colonies. After 7–10 days, single neomycin-resistant colonies were picked and grown in duplicate 96-well dishes: one for freezing and the other for screening.

For screening of Wnt-responsive clones, the cells from the screening plate were split into two gelatin-coated dishes and allowed to grow for 24 hr without LIF. The medium conditioned by either Wnt-3a–expressing L cells or parental L cells (Shibamoto et al., 1998; Muroyama et al., 2004) containing 5% FCS was then added to individual dishes. After an additional 24 hr, lacZ expression was detected by staining the cells for β-galactosidase activity and examining them microscopically. Clones in which the expression of the lacZ reporter was activated by treatment with Wnt-3a C.M. were chosen, recovered from frozen stocks, and re-tested repeatedly for their responsiveness to Wnt-3a C.M.. The concentration of Wnt-3a protein in Wnt-3a C.M. was 400 μg/ml (Shibamoto et al., 1998), and 100 μl of C.M. was added to each well.

The selected ES cell lines were used for generating mouse chimeras by blastocyst injection. Chimeric mice were checked for germ-line transmission and used for generation of transgenic mice heterozygous for the trapped gene. For examining the time course of response to Wnt signal, an ES cell line stably containing an EGFP reporter gene whose expression was driven from a promoter containing a tandem repeat of seven TCF-binding sites (Ueda et al., 2002) was established.

β-Galactosidase Staining and In Situ Hybridization

Cells were washed once in PBS, fixed for 10 min in 0.2% glutaraldehyde in PBS, washed twice in PBS, and stained in staining solution [PBS containing 1 mg/mmol X-gal, 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, and 2 mM MgCl2] at 30°C overnight. Embryos were dissected and fixed in a solution consisting of 1% formaldehyde, 0.2% glutaraldehyde, 0.02% NP40, 2 mM MgCl2, 5 mM ethylenediaminetetraacetic acid (EDTA) in PBS (for embryos older than E12.5, 0.4% NP40 was used and 0.1% sodium deoxycholate was added) for 10 min (∼E11.5) to 60 min (E16.5); and washed twice in PBS containing 2 mM MgCl2 for 10–60 min. They were then stained in a solution containing 20 mM Tris pH 6.8 (for embryos older than E12.5, 0.4% NP40 and 0.1% sodium deoxycholate were added) at 30°C until sufficient color had developed. Whole-mount in situ hybridization was performed as described (Wilkinson, 1992). Full-length cDNAs for Wnt-7b (Kispert et al., 1996) and Wnt-5b (Gavin et al., 1990) were labeled with digoxigenin to prepare riboprobes.

Identification of Trapped Locus

RNA preparation was performed with an RNeasy kit (QIAGEN). Cloning of fusion transcripts by 5′-RACE was performed with a Smart RACE kit (BD Clonetech) or GeneRacer kit (Invitrogen) according to the manufacturer's instructions. The gene-specific primer sequences were as follow: GSP1, 5′-TGGCGAAAGGGGGATGTGCTG-3′; nested GSP1, 5′-GATGTGCTGCAAGGCGATTAAG-3′; nested GSP2, 5′-CTCAGCCTTGAGCCTCTGGAGCTGCTC-3′. RACE–PCR products were cloned and sequenced.

Plasmid rescue was done as follows: Briefly, genomic DNA was extracted from ES cell clones with lysis buffer (0.1 M Tris-HCl pH 8.0, 5 mM EDTA, 0.2% sodium dodecyl sulfate, 0.2 M NaCl with 0.2 mg/ml of proteinase K). Ten micrograms of DNA was digested with NcoI overnight, cleaned, and ligated at a concentration of 5 μg/ml. The reactions were incubated overnight at 16°C. Transformation of DH10B (GIBCO) was done by electroporation with 1 μg of ligated DNA. Colonies were selected for ampicillin resistance, and the rescued plasmids were sequenced. In addition to the plasmid rescue method, DNA Walking SpeedUp Kit (Seegene) was used for cloning of the transition site between the gene-trap vector and genomic DNA according to the manufacturer's instructions.

Organ Culture

The SMG rudiments at E13 were dissected in Hanks' balanced salt solution (Umeda et al., 2001). The rudiments were placed on Millipore filters and cultured for 12 hr with DMEM/10% FCS containing 0.3mM of CKI7 (Seikagaku Kogyo, Japan) dissolved in dimethyl sulfoxide. Total RNA used for cDNA synthesis was extracted from two rudiments with Trizol (Invitrogen), treated with DNase, and cleaned up by use of an RNeasy mini kit (QIAGEN).

Quantitative RT-PCR

Quantitative RT-PCR was done with LightCycler (Roche). Complementary DNA was created from DNase-treated total RNA by using SuperScriptIII (Invitrogen) with random oligo (6mer) primer. Minus RT controls were also prepared similarly. One microgram of RNA was included in each reaction in a total volume of 20 μl. The amplification reaction was performed using the following thermocycler conditions: 95°C for 10 min followed by 40 cycles of 95°C for 15 sec, 55 °C for 5 sec, and 72°C for 15 sec. Plus RT, minus RT, and no template controls were tested for primer and MgCl2 concentration optimization. The combination of forward and reverse primer concentrations was selected based on the presence of a single band only in the plus RT sample. Expression levels of all samples were normalized to the level of mouse hypoxanthine phosphoribosyltransferase (HPRT) in each sample. The primer sequences used were as follows: CRTR-1 (forward), 5′-ATCTTCCTGGAAGAGCTGAC-3′; CRTR-1 (reverse), 5′-TCAGGATGATGTGGTAGCCATC-3′; clone 5U(forward) 5′-TTCCCTAAAGACCAATCAGT-3′; clone 5L(reverse), 5′-CCACATGGAGCCAGATCAATG-3′; HPRT (forward), 5′-GCTGGTGAAAAGGACCTCT-3′; and HPRT (reverse), 5′-CACAGGACTAGAACACCTGC-3′. Each experiment was performed in triplicate.

Northern Blot Analysis

Ten micrograms of total RNA prepared from P0 kidney of the wild-type, heterozygous, and homozygous mice for the clone 43 trapped allele were separated by agarose gel electrophoresis according to standard procedures. Blotting and hybridization were performed according to the manufacturer's protocol (Roche). The riboprobe for CRTR-1 was generated from an EcoRI fragment of CRTR-1 cDNA containing exons 1 to 15.

Histology

For histological examination, kidneys from 30-day-old mice were fixed in Bouin's fixative, dehydrated, embedded in paraffin, and sectioned at 6 μm. The sections were dewaxed, rehydrated, and stained by the periodic acid/Schiff reaction according to standard procedures.

Acknowledgements

We thank W. Wurst for the gift of pGT1 vector, A.P. McMahon for Wnt-7b and Wnt-5b riboprobes, M. Hijikata for the reporter plasmids, and H. Niwa for the EB5 cell. We also thank Y. Hieda for expert help in explant cultures, H. Hijikata and R. Takada for preparing Wnt-3a–expressing L cells, N. Takeda and H. Watanabe for blastocyst injection, M. Ogawa for technical advice, and M. Futamata and M. Sawada for technical support. We thank all the members of the S.T. lab and the Takeichi lab for helpful discussions. This work was supported by a grant-in-aid for scientific research from the Ministry of Education, Science, Culture, and Sports of Japan and grants from the Japan Science and Technology Corporation, and Mitsubishi Foundation to S.T.

Ancillary