Expression of the mouse Fgf15 gene is directly initiated by Sonic hedgehog signaling in the diencephalon and midbrain

Authors

  • Hirotomo Saitsu,

    1. Department of Anatomy and Developmental Biology, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto, Japan
    Search for more papers by this author
  • Munekazu Komada,

    1. Department of Anatomy and Developmental Biology, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto, Japan
    Search for more papers by this author
  • Misao Suzuki,

    1. Center for Animal Resources and Development (CARD), Kumamoto University, Kumamoto, Japan
    Search for more papers by this author
  • Rika Nakayama,

    1. Laboratory for Animal Resources and Genetic Engineering (LARGE), Center for Developmental Biology (CDB), RIKEN Kobe, Chuou-ku, Kobe, Japan
    Search for more papers by this author
  • Jun Motoyama,

    1. Molecular Neuropathology Group, Brain Research Institute, RIKEN, Wako, Saitama, Japan
    Search for more papers by this author
  • Kohei Shiota,

    1. Department of Anatomy and Developmental Biology, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto, Japan
    2. Congenital Anomaly Research Center, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto, Japan
    Search for more papers by this author
  • Makoto Ishibashi

    Corresponding author
    1. Department of Anatomy and Developmental Biology, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto, Japan
    • Department of Anatomy and Developmental Biology, Graduate School of Medicine, Kyoto University, Yoshida, Sakyo-ku, Kyoto 606-8501, Japan
    Search for more papers by this author

Abstract

Sonic hedgehog (Shh) is a secreted molecule that is thought to regulate tissue growth and patterning in vertebrate embryos. Although it has been reported that Gli transcription factors mediate Shh signaling to the nucleus, little is known about developmental target genes of Gli. In the previous genetic study, we showed that Shh is required for Fgf15 expression in the diencephalon and midbrain. Here, we examined whether Fgf15 is a direct target of Shh signaling through Gli. Shh was expressed in the midline cells and Fgf15 in the medial region of the diencephalon/midbrain by the seven-somite stage. The Fgf15 expression domain coincided with that of Gli1 and overlapped with that of Gli2 at this stage. Fgf15 expression in the diencephalon/midbrain was greatly reduced in the seven-somite Shh mutant embryos. Transgenic analysis showed that the 3.6-kb 5′-flanking region of the Fgf15 gene is sufficient for induction of Fgf15 in the medial/ventral diencephalon/midbrain. Luciferase assay showed that the 3.6-kb Fgf15 enhancer/promoter was activated by Gli2. A Gli-binding site was located 1 kb upstream of the transcription start site and was required for expression in the medial/ventral diencephalon/midbrain in transgenic embryos and for activation in luciferase assay. These findings indicate that Fgf15 is directly regulated by Shh signaling through Gli proteins. Developmental Dynamics 232:282–292, 2005. © 2004 Wiley-Liss, Inc.

INTRODUCTION

In the developing central nervous system (CNS), neural precursor and/or stem cells are proliferating to prepare a cellular template for later patterning events. Cell survival and specification also contribute to control of size and shape of the tissue. Several secreted molecules are expressed and are thought to regulate local growth rates and cell specification. Some kind of regulatory network among these factors might coordinate cell proliferation, survival, and specification between adjacent areas to maintain tissue integrity. Molecular mechanisms of the regulatory network should be elucidated for further understanding of morphogenesis.

Sonic hedgehog (Shh) is a member of the Hedgehog (Hh) family of signaling molecules in vertebrates. It has been reported that Shh is involved in patterning of various tissues (Riddle et al.,1993; Chiang et al.,1996; Ericson et al.,1996; Goodrich et al.,1997; Hebrok et al.,2000; Ramalho-Santos et al.,2000). In addition, accumulating evidence suggests that Shh also plays important roles for cell proliferation (Dahmane and Ruiz i Altaba,1999; Wallace,1999; Wechsler-Reya and Scott,1999) and survival (Teillet et al.,1998; Borycki et al.,1999). Hh signals are transduced into cytoplasm by two transmembrane proteins, Patched (Ptch) and Smoothened (Alcedo et al.,1996; Chen and Struhl,1996; Marigo et al.,1996; Stone et al.,1996; van den Heuvel and Ingham,1996; Fuse et al.,1999). Then, Ci/GLI, zinc-finger transcription factors mediate the signals to the nucleus (Alexandre et al.,1996; Dominguez et al.,1996; Aza-Blanc et al.,1997; Hepker et al.,1997; Chen et al.,1999). Drosophila Ci is a pivotal switch for Hh signaling. In the presence of Hh, Ci remains unprocessed and activates transcription of target genes. On the contrary, the processed form of Ci represses expression of target genes in the absence of the ligand (Aza-Blanc et al.,1997; Chen et al.,1999). In vertebrates, Hh signal transduction is mediated by at least three Gli transcription factors (Gli1, Gli2, Gli3; Hui et al.,1994). Gli1 is itself a transcriptional target of Shh signaling and is thought to function only as a transactivator. Gli2 and Gli3 have both transactivation and repression domains and activate target genes in the presence of Shh signals as Drosophila Ci does (Dai et al.,1999; Sasaki et al.,1999; Ingham and McMahon,2001). Genetic studies indicate that Gli2 is a major mediator for initial responses to Shh signaling and that Gli3 plays a crucial role in cell specification along the dorsoventral axis of the neural tube (Ding et al.,1998; Matise et al.,1998; Theil et al.,1999; Litingtung and Chiang,2000; Tole et al.,2000; Bai et al.,2002,2004; Motoyama et al.,2003), whereas Gli1 is dispensable (Park et al.,2000).

Although the Shh pathway has been examined extensively in terms of development and human disease (Goodrich and Scott,1998; Muenke and Cohen,2000), only a few developmental regulatory genes have been identified as direct targets of Shh signaling. In mouse embryos, HNF-3β is an early developmental regulator of notochord and floor plate formation (Ang and Rossant,1994; Sasaki and Hogan,1994; Weinstein et al.,1994), and its expression in the floor plate is controlled by Shh and Gli2 (Ding et al.,1998; Matise et al.,1998) through a Gli-binding site is in its enhancer region (Sasaki et al.,1997).

We previously demonstrated that Shh signaling is required for normal development of both dorsal and ventral regions of the diencephalon and anterior midbrain (Ishibashi and McMahon,2002). In the other areas of the neural tube, Shh is indispensable for only the ventral half of development (Chiang et al.,1996). The expression analysis of Ptch1 suggested that Shh does not signal directly to the dorsal regions. Instead, the data suggested that Fgf15 transmits a ventrally derived Shh signal to the dorsal regions (Ishibashi and McMahon,2002). Although it has been reported that Fgf15 is expressed in the developing CNS (McWhirter et al.,1997; Gimeno et al.,2003), its roles in the neural tube has yet to be elucidated. The detailed analyses showed that Fgf15 is initially expressed in the ventral parts of the diencephalon and midbrain and the expression domain expands dorsally at later stages (Ishibashi and McMahon,2002). In Shh mutant embryos, Fgf15 expression was not detectable in the diencephalon and midbrain between the 12- and 16-somite stages (Ishibashi and McMahon,2002). However, these studies did not resolve whether Fgf15 is a direct target of Shh signaling, or alternatively, whether there is another signal between Shh and Fgf15. In this study, we examined the involvement of Gli proteins in the activation of Fgf15 expression in the diencephalon/midbrain. The expression domains of Gli1 and Gli2 overlapped with that of Fgf15 at the time of its induction. Luciferase assay demonstrated that Gli2 activates the Fgf15 enhancer/promoter through a Gli-binding site. This binding site was required for medial/ventral diencephalic and midbrain expression in transgenic embryos. Our data suggest that Fgf15 is a direct target of Shh signaling through Gli transcription factors.

RESULTS

Induction of Fgf15 Expression in the Medial/Ventral Diencephalon and Midbrain Is Greatly Reduced in Shh Mutants

Previously, we reported that Fgf15 was expressed in the ventral parts of the diencephalon and midbrain of wild-type embryos by the 12-somite stage. This expression was undetectable in Shh mutants, suggesting that Shh signaling is required for Fgf15 expression at this and later stages (Ishibashi and McMahon,2002). To further clarify the detailed expression pattern of Fgf15 in relation to Shh signaling, we performed in situ hybridization at earlier stages. Because it has been reported that Shh expression is detectable in the head fold of the eight-somite stage (Echelard et al.,1993), we examined the expression of Shh and Fgf15 at the three- and seven-somite stages. Neither were expressed at the three-somite stage (data not shown). At the seven-somite stage, Shh was expressed in the midline cells of the diencephalon/midbrain (Fig. 1C). On the other hand, Fgf15 was expressed in the medial (ventral at later stages) region but not in the midline (Fig. 1A, arrowheads, A′). Of interest, Fgf15 expression showed a medialhigh–laterallow gradient pattern (Fig. 1A,A′). The expression was greatly reduced to almost background level in the diencephalon/midbrain of Shh mutants at the same stage (Fig. 1B, bracket; B′), whereas the expression was preserved in the hindbrain (Fig. 1B, arrow). Taken together, this evidence suggests that Fgf15 is induced by midline cell-derived SHH in a dose-dependent manner in the diencephalon/midbrain. Because Fgf15 expression is excluded from the midline cells, it seems that repressors against Fgf15 expression are induced by the highest levels of Shh signaling (see Discussion section).

Figure 1.

Comparison of Fgf15 expression with Shh, Gli1, Gli2, and Gli3 in the diencephalon and midbrain at the time of its induction. A dorsal view of the head neural fold of wild-type (A,C–F) and Shh mutant (B) mouse embryos stained by whole-mount RNA in situ hybridization with probes as indicated at the seven-somite stage. A′ and B′ indicate transverse sections at the midbrain level of A and B, respectively. A,A′: In wild-type (WT) embryos, Fgf15 expression was observed in the medial diencephalon/midbrain (arrowheads) except for the midline cells. Note that intensity of the staining showed a medialhigh–laterallow gradient. B,B′:Fgf15 expression was greatly reduced in the diencephalon/midbrain of Shh mutant embryos (bracket), whereas the expression was preserved in the hindbrain (arrow). C:Shh expression in wild-type embryos was initiated in the midline cells of the diencephalon/midbrain by this stage. D:Gli1 was expressed abundantly in the medial region. E:Gli2 was expressed rather evenly throughout the diencephalon/midbrain. Expression of both Gli1 and Gli2 overlapped with that of Fgf15. F: The expression pattern of Gli3 exhibited a gradient opposite to that of Fgf15. Anterior is toward the top.

Fgf15 Expression Domain Coincides With That of Gli1 and Overlaps With That of Gli2 in the Diencephalon and Midbrain

Although mechanisms of intracellular transduction of Shh signaling are not fully understood, accumulating evidence supports that Gli transcription factors, vertebrate homologues of Drosophila Ci, play major roles in mediating Shh signals to the nucleus (Ingham and McMahon,2001; Ruiz i Altaba et al.,2002). To identify whether Gli transcription factors are involved in the induction of Fgf15 in the diencephalon and midbrain, we examined the expression patterns of Gli1, 2, and 3 in relationship to the Fgf15 expression domain. Robust expression of all the three Gli genes was observed in the diencephalon and midbrain at the seven-somite stage. Their expression patterns were similar to those shown in previous reports (Hui et al.,1994; Lee et al.,1997; Sasaki et al.,1997) and all of them were excluded from the midline cells (Fig. 1D–F). Expression of Gli1 showed a similar pattern to that of Fgf15 in the medial part (Fig. 1D), suggesting that Gli1 mediates induction of Fgf15 by Shh. The expression domain of Gli2 spread throughout the diencephalon/midbrain and accordingly covered that of Fgf15 (Fig. 1E). Because Gli2 activates target genes only in the presence of Shh signals (Sasaki et al.,1999; Mill et al.,2003), this is also consistent with the possibility that Gli2 mediates induction of Fgf15 by Shh. Gli3 showed a complementary pattern to Gli1 (Fig. 1F). Thus, these data are consistent with the possibility that Gli1 and/or Gli2 may be involved in induction of Fgf15 in the diencephalon and midbrain.

The 3.6-kb 5′-Flanking Region of the Fgf15 Gene Contains Sufficient Positive Regulatory Elements for Fgf15 Expression in the Diencephalon and Midbrain

To identify regulatory elements for Fgf15 expression in the diencephalon and midbrain at embryonic day (E) 8.5, we first isolated the 12-kb and 3.6-kb fragments upstream of the transcription initiation site (McWhirter et al.,1997; Fig. 2). Then the 12-kb and 3.6-kb fragments were tested for cis-activity by means of generation of transgenic mouse embryos carrying the reporter lacZ gene. Transgene expression analysis was transiently conducted at E8.5; with the 12-kb and 3.6-kb fragments, two and three embryos showed lacZ expression, respectively (Figs. 3, 6D; data not shown). In all transgenic embryos, lacZ showed a similar expression pattern to endogenous Fgf15 in the diencephalon and midbrain at the time of Fgf15 induction, except that it was also expressed in the ventral midline cells (Fig. 3, compare left and right columns; data not shown). Expression of lacZ in the ventral midline cells suggests lack of negative regulatory elements, because the expression of endogenous Fgf15 was excluded from this region. It is also suggested that positive regulatory elements are activated in response to Shh signals from the notochord and/or floor plate in the absence of negative elements. At the 12-somite stage, lacZ was expressed in the dorsal parts of the diencephalon and midbrain as well (Fig. 3C, filled arrowhead). This expression pattern was almost the same as that of endogenous Fgf15 (Fig. 3D, filled arrowhead). These findings indicate that the 3.6-kb fragment includes sufficient positive regulatory elements for Fgf15 expression in the diencephalon and midbrain. Taken together with Gli expression patterns, the data raise the possibility that the 3.6-kb enhancer/promoter of the Fgf15 gene may contain Gli-binding sites.

Figure 2.

Organization of the Fgf15 locus and location of a putative Gli-binding site. The transcriptional unit and its direction are indicated by boxes and an arrow, respectively. The 12-kb and 3.6-kb fragments of the Fgf15 5′-flanking region were used for transgenic study and luciferase assay. A candidate Gli-binding site (GliBSWT) was located 1 kb upstream of the Fgf15 transcription start site. The core region of the Gli-binding site (box) contained only one nucleotide mismatch to the consensus Gli-binding sequence (Kinzler and Vogelstein,1990) and is identical to the Gli-binding site in the mouse and rat Osteopontin promoter (Yoon et al.,2002). Nucleotide substitution mutations were introduced into this Gli-binding sequence (GliBSMut) to disrupt its Gli-binding capacity

Figure 3.

The 3.6-kb Fgf15 enhancer/promoter contains sufficient positive regulatory elements for Fgf15 expression in the diencephalon and midbrain. A–D: A lateral view of the 3.6-kb Fgf15 enhancer/promoter::lacZ transgenic embryos stained for β-galactosidase activity (A,C) and wild-type embryos stained by whole-mount RNA in situ hybridization (ISH) using the Fgf15 antisense probe (B,D) at the seven-somite (7so, A,B) and 12-somite (12so, C,D) stages. A′,B′: Transverse sections at the midbrain level of A and B, respectively. lacZ expression showed a similar pattern to endogenous Fgf15 expression in the medial/ventral diencephalon and midbrain except for ectopic expression of midline cells (compare the left panels with the right panels). Note that both lacZ and endogenous Fgf15 expression in 12-somite embryos exhibited dorsal expansion around the diencephalon/midbrain boundary (C,D, filled arrowhead). There was also ectopic lacZ expression in the hypothalamus (A,C). The open arrowheads indicate the midbrain–hindbrain boundary.

Figure 6.

The Gli-binding site is essential for Fgf15 induction in the medial/ventral diencephalon and midbrain in transgenic embryos. A–C: Lateral (A–C) and dorsolateral (A′) views of transgenic embryos stained for β-galactosidase activity at the indicated stages. Transgenic embryos with the mutant 3.6-kb Fgf15 enhancer/promoter showed no lacZ expression in the medial/ventral diencephalon and midbrain (white brackets), while lacZ expression in the dorsal diencephalon/midbrain and in the hindbrain was still observed. Open and filled arrowheads indicate the midbrain–hindbrain and the diencephalon–midbrain boundary, respectively. D: Summary of a transient transgenic mouse experiment of the 3.6-kb Fgf15 enhancer/promoter::lacZ transgene with GliBSWT and GliBSMut. Note that medial/ventral diencephalon/midbrain expression of the GliBSMut transgene was not detectable in all transgenic embryos.

Gli2 Directly Activates Fgf15 Expression in C3H10T1/2 Cells Through a Gli-Binding Site in the 3.6-kb Enhancer/Promoter Region

To examine whether the 3.6-kb enhancer/promoter is regulated by Gli transcription factors, we performed luciferase assays using C3H10T1/2 cells. This cell line is known to be a good responder to Shh signals, suggesting that a battery of the molecular machinery for Shh signaling is present within it (Nakamura et al.,1997; Frank-Kamenetsky et al.,2002). Indeed, when the 8×3′GliBS-Luc reporter (Sasaki et al.,1997), which contains the tandemly repeated Gli-binding site of HNF-3β gene, was transfected into this cell line with the expression vectors of Gli1 and Gli2, luciferase activity was up-regulated by 62-fold and 8-fold, respectively (Fig. 4A). Luciferase reporters containing the 430-bp Fgf15 promoter (McWhirter et al.,1997) or 3.6-kb Fgf15 enhancer/promoter were transfected into the cells with or without the expression vectors of Gli1, Gli2, and hGLI3. All three Gli transcription factors were shown to not activate the 430-bp Fgf15 promoter (Fig. 4B). In contrast, Gli2 caused robust transactivation of the 3.6-kb enhancer/promoter (Fig. 4B). Unexpectedly, Gli1 did not up-regulate luciferase activity, suggesting that Gli1 is not sufficient for Fgf15 induction. The same results were obtained with the 12-kb enhancer/promoter (data not shown). Moreover, hGLI3 did not activate the 3.6-kb enhancer/promoter and did not inhibit the Gli2-mediated activation of the 3.6-kb enhancer/promoter either (data not shown). When we transfected the expression vector of Gli2 or Gli2ΔN2, which encodes a constitutive activator form of Gli2 (Sasaki et al.,1999; Mill et al.,2003), up-regulation of the endogenous Fgf15 gene was observed in C3H10T1/2 cells (Fig. 4C). These indicate that Gli2 induces expression of Fgf15 in vitro.

Figure 4.

Gli2 activates the 3.6-kb Fgf15 enhancer/promoter through a putative Gli-binding site and up-regulates endogenous Fgf15 in C3H10T1/2 cells. A: The 8x3′GliBS-Luc reporter as a positive control (A; Sasaki et al.,1997) and the 430-bp Fgf15 promoter::luciferase reporter, the 3.6-kb Fgf15 enhancer/promoter::luciferase reporters with wild-type (GliBSWT), and mutant (GliBSMut) Gli-binding site (B) were cotransfected into C3H10T1/2 cells with the Gli1, Gli2, or hGLI3 expression vectors. Luciferase activities are indicated as fold activation relative to the luciferase activity with the control expression vector. Whereas the positive control 8x3′GliBS-Luc reporter was activated by both Gli1 and Gli2, the wild-type enhancer/promoter of the Fgf15 gene was activated by Gli2 but not by Gli1 and hGLI3. This activation by Gli2 was significantly reduced by mutation of the putative Gli-binding site in the enhancer/promoter (P = 0.0198 by unpaired Student t-test). The 430-bp Fgf15 promoter::luciferase reporter was not activated by all three Gli proteins. The results represent the average of three experiments with standard deviation depicted with error bars. C: C3H10T1/2 cells were transfected with the Gli2 or Gli2ΔN2 expression plasmids and then analyzed for endogenous Fgf15 expression by reverse transcriptase-polymerase chain reaction. Compared with mock transfection, expression of Fgf15 was increased by Gli2 and Gli2ΔN2 expression plasmids. Mouse EF-1α was used as an internal control to verify the amount of template. The polymerase chain reaction amplification of mouse EF-1α was not saturated at the 20th cycle.

To figure out if Gli2 directly up-regulates Fgf15, we searched the 3.6-kb sequence for the human GLI-binding consensus (Kinzler and Vogelstein,1990). At 1 kb upstream of the transcription start site, there was a similar sequence (1-base mismatch) to the GLI-binding consensus (Fig. 2). Then, to examine whether this sequence is required for up-regulation of Fgf15, we introduced base-substitution mutations that destroy the core Gli-binding sequence (see Fig. 2). The mutations resulted in a significant decrease of activation by Gli2 (Fig. 4B), indicating that this site is required for activation of the 3.6-kb enhancer/promoter of Fgf15.

Next, we performed an electrophoretic mobility shift assay to test whether Gli1 and Gli2 proteins directly bind to the putative binding site (GliBSWT) in vitro. Because GLI proteins bind to DNA through their zinc-finger domains, which consist of tandemly repeated C2H2 motifs (Pavletich and Pabo,1993), bacterial lysate containing a fusion protein of glutathione-S-transferase and the zinc-finger domain of Gli1 (Gli1-ZF) or Gli2 (Gli2-ZF) was used for this test. Glutathione-S-transferase (GST) -Gli1-ZF and GST-Gli2-ZF but not GST alone bound to the GliBSWT probe (Fig. 5A). The binding of GST-Gli1-ZF and GST-Gli2-ZF protein to the GliBSWT probe was greatly reduced by excess amounts of the unlabeled GliBSWT oligonucleotide but not by an excess of the GliBSMut oligonucleotide with the same base-substitution mutations as described above. Although Gli1 was not able to transactivate the 3.6-kb enhancer/promoter, Gli1 bound to GliBSWT sequence in this assay (see Discussion section).

Figure 5.

The zing-finger domain of Gli1 and Gli2 protein bound to the putative Gli-binding site in the 3.6-kb Fgf15 enhancer/promoter and the multimerized Gli-binding site showed enhancer activity. A: Electrophoretic mobility shift assay showed that the glutathione-S-transferase (GST) -fused zinc-finger domain of Gli2 (GST-Gli2) and Gli1 (GST-Gli1) bound to the 32P end-labeled 39-bp oligonucleotide probe of the GliBSWT. A 90-fold excess amount of the unlabeled GliBSWT oligonucleotide greatly reduced the GST-Gli2 and GST-Gli1 binding to the probe, whereas the GliBSMut oligonucleotide did not. B, protein-bound probe; F, free probe. B: The luciferase reporter with 8× multimerized Gli-binding site (8xGliBSWT) from the 3.6-kb Fgf15 enhancer/promoter was cotransfected into C3H10T1/2 cells with the Gli2 expression vector. Results are indicated as fold induction relative to luciferase activity with the control expression vector and represent the average of three experiments, with standard deviation depicted with error bars. Gli2 up-regulated the 8xGliBS reporter by 150-fold, whereas the 8xGliBSMut was not activated at all.

Furthermore, to confirm that transactivation of the Fgf15 enhancer/promoter by Gli2 is mediated by this Gli-binding site, the luciferase reporter plasmid with eight copies of the GliBSWT sequence (8xGliBSWT) or GliBSMut sequence (8xGliBSMut) was cotransfected with the Gli2 expression vector into C3H10T1/2 cells. A 150-fold increase of luciferase activity by Gli2 was observed with the 8xGliBSWT reporter. In contrast, the 8xGliBSMut reporter showed no increase of luciferase activity (Fig. 5B). These biochemical and functional assays clearly demonstrated that Gli2 transactivates the 3.6-kb Fgf15 enhancer/promoter by directly binding to this GliBSWT in vitro.

Gli-Binding Site Is Essential for Fgf15 Induction in the Medial/Ventral Diencephalon and Midbrain in Transgenic Embryos

The significance of the Gli-binding site in Fgf15 regulation in vivo was investigated by transient transgenic analysis. For this experiment, the GliBSWT was replaced with the GliBSMut in the 3.6-kb Fgf15 enhancer/promoter of the transgene. This mutation of the Gli-binding site completely abolished expression of lacZ in the medial/ventral diencephalon and midbrain (Fig. 6A–C, white brackets) in all transgenic embryos examined (Fig. 6D). Therefore, the Gli-binding site in the 3.6-kb Fgf15 enhancer/promoter has an essential positive regulatory function for Fgf15 expression in the medial/ventral diencephalon and midbrain. In contrast, the expression of the transgene remained in the hindbrain with the mutated enhancer/promoter, which is consistent with the evidence that Fgf15 was still expressed in the hindbrain of Shh mutants (Fig. 1B, arrow). Also, the mutation of the Gli-binding site spared the dorsal expression of Fgf15 in the diencephalon/midbrain, suggesting indirect roles of Shh in the dorsal regions (see Discussion section). These data support the proposition that Fgf15 expression is directly initiated by Shh signaling through Gli transcription factors in the medial/ventral diencephalon and midbrain in vivo.

Fgf15 Expression in the Medial/Ventral Diencephalon and Midbrain Is Not Decreased in Gli2 Mutants

To ascertain whether Gli2 is the sole mediator of Fgf15 induction in the medial/ventral diencephalon and midbrain in vivo, we examined Fgf15 expression in Gli2 mutant embryos. Fgf15 expression in the medial/ventral diencephalon and midbrain was not decreased in the Gli2 mutant embryos (Fig. 7B), indicating that Gli2 is not the sole activator for Fgf15 induction. The midline region in the diencephalon and midbrain, which does not express Fgf15 in wild-type embryos tended to be reduced in the Gli2 mutant embryos and this may be related to the lack of floor plate differentiation in Gli2 mutant embryos (Ding et al.,1998; Matise et al.,1998).

Figure 7.

Fgf15 expression in the diencephalon and midbrain in Gli2 mutant embryos at the 10-somite stage. A,B: A dorsal view of the head neural fold of wild-type (WT, A) and Gli2 mutant (B) mouse embryos stained by whole-mount RNA in situ hybridization. Fgf15 expression in the medial/ventral diencephalon and midbrain was not decreased in the Gli2 mutant embryos (B).

DISCUSSION

We previously demonstrated that Shh signaling is critical for the proliferation and survival of neural precursors in the diencephalon and anterior midbrain (Ishibashi and McMahon,2002). Unlike other areas of the neural tube, Shh signaling is required for normal development of both the dorsal and ventral regions of the diencephalon and anterior midbrain, and analysis of Ptch1 expression suggested that Shh does not signal directly to the dorsal regions (Ishibashi and McMahon,2002). In this context, Fgf15 seems to be a key molecule that mediates a ventrally derived Shh signal and coordinates growth of dorsal and ventral neural tissues (Ishibashi and McMahon,2002). However, it remained to be elucidated whether Fgf15 regulation by Shh is direct or indirect.

In the present study, we showed that a Gli-binding site in the 3.6-kb Fgf15 enhancer/promoter is essential for induction of Fgf15 in the medial/ventral diencephalon and midbrain in transgenic embryos. In addition, the 3.6-kb Fgf15 enhancer/promoter::lacZ transgene exhibited a dorsal expansion of the expression domain in the diencephalon/midbrain in 12-somite embryos. This dynamic expression pattern suggests that Fgf15 signaling connects the ventral and dorsal regions of the diencephalon/midbrain (Ishibashi and McMahon,2002). Whereas Fgf15 expression in Shh mutants was affected in both the ventral and dorsal diencephalon/midbrain, mutation of the Gli-binding site extinguished only the medial/ventral expression in the diencephalon and midbrain but not the dorsal expression. These data indicate that the induction in the medial/ventral region is directly regulated by Shh signaling through Gli, but by contrast, the dorsal expansion is indirectly regulated. One possible explanation is that FGF15 secreted from the ventral parts diffuses dorsally to induce Fgf15 expression; accordingly, dorsal expression is not observed in Shh mutants. Although the Ptch1 expression pattern suggested that SHH protein does not reach the dorsal parts, we cannot exclude the possibility that there are other Gli-responsive enhancers that are activated by undetectable levels of Shh signaling. The nature of dorsal expansion of Fgf15 expression will be understood by further detailed analysis of the 3.6-kb Fgf15 enhancer/promoter.

Our findings demonstrated that the induction of Fgf15 in the medial/ventral diencephalon/midbrain is regulated by Gli proteins. Therefore, the next question is, which Gli transcription factors regulate expression of Fgf15 in vivo. Previous genetic studies showed that Gli2 is a major mediator of initial Shh signaling (Ding et al.,1998; Matise et al.,1998). Although our luciferase assay showed that only Gli2 can transactivate the 3.6-kb enhancer/promoter in vitro, Fgf15 expression in the medial/ventral diencephalon and midbrain was not decreased in the Gli2 mutant embryos. These data suggest that Fgf15 is regulated redundantly by Gli proteins in vivo, although it is not known whether Gli1 and/or Gli3 activate Fgf15 expression through the 3.6-kb enhancer/promoter or other elements in Gli2 mutant embryos. Expression analyses in the diencephalon/midbrain revealed that the expression pattern of Gli1 was similar to that of Fgf15. In fact, the previous report suggested that Gli1 can transactivate the same set of genes as Gli2 in vivo (Bai and Joyner,2001). Thus, we suggest that not only Gli2 but also Gli1 mediates Shh signals to induce Fgf15 expression. On the other hand, previous studies suggested that Gli3 could also activate target genes in the presence of Shh (Motoyama et al.,2003; Bai et al.,2004). Although Gli3 showed a pattern complementary to Gli1 in the diencephalon/midbrain, we cannot exclude the possibility that a very small amount of Gli3 might be able to transactivate Fgf15 expression.

The preference of Gli2 in the activation of the 3.6-kb Fgf15 enhancer/promoter in C3H10T1/2 cells also provides an excellent model to understand the molecular mechanisms for differential uses of Gli proteins. In this study, the luciferase assay showed that Gli2 but not Gli1 can transactivate the 3.6-kb Fgf15 enhancer/promoter. Because Gli1 can transactivate the 8x3′GliGS-Luc reporter more intensively than Gli2 in C3H10T1/2 cells (Sasaki et al.,1997), this finding is not due to absence of cofactors that are required for activator function of Gli1. The Gli-binding site of the Fgf15 enhancer/promoter is identical to that in the Osteopontin promoter (Yoon et al.,2002). They reported that human GLI1 binds to the same Gli-binding sequence (5′-GACCTCCCA-3′) in vitro. Although we also demonstrated that Gli1 can bind to this sequence in vitro, our preliminary data showed that Gli1 cannot compete against the activation of 3.6-kb Fgf15 enhancer/promoter by Gli2, suggesting that Gli1 does not bind to this sequence in the context of 3.6-kb Fgf15 enhancer/promoter (unpublished data). Further analysis of this issue currently is being investigated.

Although mutation of the Gli-binding site in the 3.6-kb Fgf15 enhancer/promoter led to a complete loss of expression in the medial/ventral diencephalon and midbrain in vivo, the same sequence was still up-regulated by Gli2 in luciferase assays. We introduced mutations into several other candidate Gli-binding sites, in which two of nine bases were mismatched, but we did not observe any significant decrease of luciferase activity (data not shown). Recent studies showed that human GLI2 can activate the reporter expression even when the GLI2-binding cis-element is removed (Browning et al.,2001; Smith et al.,2001). Thus, there is a possibility that Gli2 can act as a cofactor in the absence of the Gli-binding site.

Another interesting question is how Fgf15 expression is excluded from the Shh expression domain, where the highest level of Shh signals is believed to exist. Although expression of Gli1 also exhibits a similar pattern to that of Fgf15, it has been reported that Gli1 is once expressed in the presumptive floor plate region and down-regulated in the same region later in development (Lee et al.,1997; Sasaki et al.,1997). Unlike Gli1, Fgf15 has not been observed to be expressed in the midline cells. There are two possible explanations for this expression pattern: (1) Fgf15 is induced directly by the intermediate level of Shh signals not by the highest level, and (2) repressors against Fgf15 expression are induced by the highest level of Shh signals; therefore, Fgf15 is not induced in the midline cells. The transgenic embryos with the wild-type enhancer/promoter demonstrated lacZ expression in the midline cells, which suggests the presence of negative regulatory elements. From our results, we prefer the latter explanation. Understanding this issue would provide us with some clues to elucidate how different levels of Shh signaling regulate the expression of different genes.

In summary, our data provide evidence that Fgf15 is directly initiated by Shh signaling through Gli transcription factors in the medial/ventral diencephalon and midbrain. This molecular interaction would be the basis for the coordination of tissue growth between the ventral and dorsal parts of the brain.

EXPERIMENTAL PROCEDURES

Mice

Generation of Shh and Gli2 mutant mice has been described previously (Mo et al.,1997; St-Jacques et al.,1998). Shh and Gli2 heterozygous mutants were maintained on the ICR and mixed 129/Sv and CD1 background, respectively. Noon of the day the vaginal plug was found was designed as E0.5. To score embryonic stages more precisely, somite numbers were counted.

Whole-Mount RNA In Situ Hybridization

Whole-mount RNA in situ hybridization of embryos was performed as previously described (Parr et al.,1993). Digoxigenin (DIG) probes were synthesized by using the DIG RNA labeling kit (Roche). The probes used in this study were as follows: Fgf15, Shh (Ishibashi and McMahon,2002), Gli1 (Sasaki et al.,1999), Gli2, Gli3 (Hui and Joyner,1993). The embryos were sectioned at a thickness of 20 μm to analyze the detailed staining patterns.

Cloning of Genomic DNA

A mouse C57BL/6J BAC clone RP23-332B13 (GeneBank Accession Number AC073752) was purchased from Research Genetics (Huntsville, AL). Restriction fragments containing the Fgf15 promoter were identified by Southern blotting using a 430-bp XhoI–HindIII fragment of Fgf15 promoter (McWhirter et al.,1997) as a probe. A 10-kb HindIII fragment (−8 kb to +2 kb) was subcloned into pBluescript (Stratagene). Subsequently, a 6-kb EcoRI fragment (−12.7 kb to −6.5 kb) was identified by Southern blotting by using a 1.4-kb HindIII–EcoRI fragment as a probe and subcloned into pBluescript (see Fig. 2).

Plasmid Construction and Mutagenesis

To subclone the Fgf15 5′-flanking region into reporter vectors, polymerase chain reaction (PCR) amplification by KOD-Plus (TOYOBO, Japan) was performed to produce a −675-kb to +65-bp XhoI fragment of Fgf15 transcription start site. For transgenic analysis, the 12-kb and 3.6-kb Fgf15 5′-flanking regions (SpeI–XhoI fragment and NsiI–XhoI fragment, respectively) were fused with the lacZ reporter. The vector sequence was removed before microinjection. For luciferase assays, the 12-kb and 3.6-kb Fgf15 5′-flanking regions (EcoRI–XhoI fragment and NsiI–XhoI fragment, respectively) were subcloned into pGL3-Basic reporter (Promega). The 8xGliBSWT and 8xGliBSMut::luciferase reporter plasmids were generated by subcloning eight tandemly repeated copies of GliBSWT (5′-CTCGAGCAGACAGACCTCCCATCACCAGTCGAC-3′) and GliBSMut (5′-CTCGAGCAGACAGACagCtgATCACCAGTCGAC-3′) into the pGL3-δ51 luciferase reporter, which was constructed by inserting BglII–HindIII fragment of chicken δ-crystallin promoter (Kamachi and Kondoh,1993) into BamHI–HindIII site of pGL3-Basic reporter (Promega). Mutations were introduced into the Gli-binding site by PCR-based in vitro mutagenesis using KOD-Plus, and the sequence was verified with an ABI 310 DNA sequencer (Applied Biosystems). The expression plasmids of pcDNA3.1HisB-Gli1, Gli2, Gli2ΔN2, hGLI3 were kindly provided by Dr. Hiroshi Sasaki (Sasaki et al.,1999).

Production of Transient Transgenic Embryos and β-Galactosidase Staining

Pronuclear injection of transgene DNA was commissioned to CARD (Kumamoto, Japan) and LARGE, RIKEN (Kobe, Japan), and transient transgenic embryos were dissected at E8.5. The genotype of embryos was determined by PCR amplification of a lacZ sequence from genomic DNA extracted from extra embryonic tissues. Primers used were forward 5′-GGTAGCAGAGCGGGTAAACT-3′ and reverse 5′-ATCTGACGGGCTCCAGGAGT-3′. Detection of β-galactosidase activity in dissected embryos was performed as previously described (Hogan et al.,1994). The embryos were sectioned at thickness of 20 μm to analyze the detailed staining pattern. The stained embryos were photographed in 80% glycerol in PBS.

DNA Transfection, Luciferase Assay, and Reverse Transcriptase-PCR

DNA transfection was performed with FuGene6 (Roche) according to the manufacturer's instructions. For luciferase assay, C3H10T1/2 cells were plated onto 12-well plates at 3 × 104 cells/well and cultured in DMEM supplemented with 5% fetal bovine serum. After overnight culture, cells were cotransfected with 450 ng of the expression vector of Gli1, Gli2, or hGLI3 and 50 ng of the reporter plasmid containing the 430-bp Fgf15 promoter (McWhirter et al.,1997); the 3.6-kb, 12-kb Fgf15 enhancer/promoter; 8xGliBSWT; or 8xGliBSMut. The 8×3′GliGS-Luc reporters (Sasaki et al.,1997) were used as a positive control. A total of 0.5 ng of Renilla luciferase plasmid (pRL-SV40, Promega) was also cotransfected to normalize transfection efficiency. The cells were lysed 24 hr after transfection and luciferase activity was measured with a Lumat LB9507 luminometer (Berthold). All luciferase assay experiments were repeated three times, and transfections were performed in duplicates. Statistical analyses were performed by using unpaired Student's t-test (two-tail).

For Fgf15 induction studies, C3H10T1/2 cells were plated onto 6-cm dishes at 2 × 105 cells/dish. Cells were transfected with 2 μg of the expression plasmid of Gli2 or Gli2ΔN2, which encodes constitutively active form of Gli2 (Sasaki et al.,1999; Mill et al.,2003), and RNA was prepared after 30 hr using TRIzol reagent (GIBCO/BRL). Three micrograms of total RNA from each dish was subjected to reverse transcription using Superscript First-Strand Synthesis System for reverse transcriptase-PCR with random hexamers (Invitrogen). One microliter of the reaction was used for PCR in 20 μl using Hotstartaq (Qiagen). Primers for mouse Fgf15 (forward, 5′-CAGCAATCCCAGTCTGTGTCAG-3′; reverse, 5′-GGCCTGGATGAAGATGATATGG-3′) and mouse EF-1α (forward, 5′-TTCTGGTTGGAATGGTGACA-3′; reverse, 5′-GAGAACACCAGTCTCCACTCG-3′) were used for amplification. The conditions of the PCR were 94°C for 30 sec, 55°C for 30 sec, 72°C for 1 min for 35 cycles (Fgf15) or 20 cycles (EF-1α). Then, products were electrophoresed on 2% agarose gel. The experiments were repeated twice.

Production of GST Fusion Proteins

Fusion proteins of GST and the zinc-finger domain of Gli1 (Gli1-ZF) or Gli2 (Gli2-ZF) were produced in Escherichia coli using pGEX-4T-3 expression vector (Amersham Pharmacia Biotech). The plasmid pGEX-Gli1-ZF included the 1.3-kb EcoRI–BamHI fragment of pcDNA3.1HisB-Gli1 (Sasaki et al.,1999). The plasmid pGEX-Gli2-ZF included the 1.1-kb EcoRI–ApaI fragment of pcDNA3.1HisB-Gli2-ΔN2 (Sasaki et al.,1999). To induce protein production, 0.1 mM IPTG was added to bacterial culture. After 3 hr of induction, bacteria were harvested by centrifugation and disrupted by sonication in PBS containing 1% Triton X-100. The sonicated solution was cleared by centrifugation at 9,000 rpm for 20 min, and the cleared lysate was used as GST fusion protein solution.

Electrophoretic Mobility Shift Assay

A 32P end-labeled, double-strand GliBSWT oligonucleotide (5′-CCGCTCGAGCAGACAGACCTCCCATCACCAGTCGACGCC-3′) was used as a probe. The binding reaction was set up as follows: 0.5 μl of cleared E. coli lysate, 250 fmol of the probe, 2 μg of poly (dI-dC), and 22.5 pmol of the nonlabeled competitor were combined in 10 μl of 1× binding buffer (10 mM Tris HCl, pH 7.5, 50 mM NaCl, 5 mM MgCl2, 0.5 mM dithiothreitol, 4% glycerol) on ice. GliBSWT or GliBSMut (5′-CCGCTCGAGCAGACAGACagCtgATCACCAGTCGACGCC-3′) oligonucleotide was used as a competitor. After incubation for 30 min on ice, protein–DNA complexes were run on a 4% acrylamide gel containing 0.25× TBE and visualized by autoradiography.

Acknowledgements

We thank Dr Murray Smith for critical reading of the manuscript, Dr. Andrew P. McMahon for Shh mutant mice; Dr. Chi-chung Hui for Gli2 mutant mice and Gli1, Gli2, Gli3 probes; Dr. Kenji Seki for mEF-1αprimer; Dr. Hiroshi Sasaki for Gli1, Gli2, Gli2ΔN2, hGLI3 expression plasmids, 8×3′GliBS-Luc reporter and technical advice; Dr. Bert Vogelstein for human GLI3 cDNA; Dr. Yusuke Kamachi for pΔ1lucII; Dr. Masanori Uchikawa for ptkEGFP; Dr. Hiroshi Ohizumi for technical support; Dr. Yasumasa Bessho for C3H10T1/2 cells and technical advice; and Dr. Simon Aspland and Dr. Cornelis Murre for pGL3-Fgf15 promoter.

Ancillary