5659
Development 126, 5659-5667 (1999) Printed in Great Britain © The Company of Biologists Limited 1999 DEV2494
Graded interference with FGF signalling reveals its dorsoventral asymmetry at the mid-hindbrain boundary Matthias Carl and Joachim Wittbrodt* Developmental Biology Programme, European Molecular Biology Laboratory, Meyerhofstrasse 1, PO Box 10.2209, 69012 Heidelberg, Germany *Author for correspondence (e-mail:
[email protected])
Accepted 1 October; published on WWW 24 November 1999
SUMMARY Signalling by fibroblast growth factors (FGFs) at the midhindbrain boundary (MHB) is of central importance for anteroposterior neural patterning from the isthmic organiser. Graded suppression of FGF signalling by increasing amounts of a dominant negative FGF receptor provides evidence that in addition to anteroposterior patterning, FGF signalling is also involved in patterning along the dorsoventral axis at the MHB. FGF signalling at the MHB is required for the activation of the HH target
gene spalt at the MHB. Our results indicate that FGF signalling mediates the competence of the MHB to activate spalt in response to SHH. This interdependence of the two signalling pathways is also found in the outbudding optic vesicle where HH requires functional FGF signalling to activate spalt in the proximal eye region.
INTRODUCTION
activity, is also required for the growth and regionalisation of the midbrain-hindbrain region (Joyner, 1996; McMahon and Bradley, 1990; McMahon et al., 1992; Thomas and Capecchi, 1990). In zebrafish, two mutations affecting the FGF-signalling pathway have been identified. Homozygous acerebellar (ace) mutant embryos lack Fgf8 function and fail to develop the cerebellum and the MHB, but not the tectum (Reifers et al., 1998). The transcription factor Pax2.1 is inactivated in the no isthmus mutant (noi). These embryos fail to maintain Fgf8 expression and lack the tectum, the MHB and the cerebellum (Lun and Brand, 1998). Most studies involving FGF signalling in the developing brain have focused on its patterning activity along the anteroposterior (AP) axis. In this study, we examine the effect of graded interference with FGF signalling by a dominantnegative FGF receptor (XFD) along the AP and dorsoventral (DV) axes. We show that, at the MHB, En2 and Wnt1 expression is reduced by XFD along the AP and the DV axis in a dose-dependent manner. This indicates an extended role for FGF signalling in DV patterning of the neural tube. Within the neural tube, a prominent signal involved in DVpatterning is the secreted molecule SONIC HEDGEHOG (SHH) which emanates from mesoderm (notochord) and floorplate (Hammerschmidt et al., 1997; Placzek, 1995). SHH was reported to act in concert with FGF signalling in the developing limb bud and in the ventral midbrain and forebrain (Crossley et al., 1996b; Ye et al., 1998). We provide evidence for an early interaction between the FGF- and HH-signalling pathways at the MHB. We show that the expression of the HH target gene spalt (sal) is suppressed in a dose-dependent
During embryonic brain development, the fibroblast growth factors (FGFs) play a role in neural induction and the caudalisation of the forming neuroectoderm (Cox and Hemmati-Brivanlou, 1995; Kengaku and Okamoto, 1993, 1995; Lamb and Harland, 1995). Later FGFs are essential for the regionalisation and patterning of the brain (Bally-Cuif and Wassef, 1995; Wassef and Joyner, 1997) and for axon outgrowth and guidance (McFarlane et al., 1996; Saffell et al., 1997). One prominent source of FGF-signals involved in regionalisation of the brain is the boundary separating the midbrain and hindbrain (MHB), also referred to as isthmus. The MHB is an organising centre as shown by transplantation experiments in the chick. Grafts of tissue originating from the MHB into the caudal forebrain or hindbrain lead to the induction of ectopic midbrain or cerebellar structures, respectively (Martinez et al., 1991, 1995). FGF8 is able to mimic the organising activity of the MHB as shown by FGF8 bead implantation experiments (Crossley et al., 1996a). In mice, FGF8 signalling in the MHB is essential for the growth and polarity of the midbrain, cerebellum and the MHB itself (Lee et al., 1997; Meyers et al., 1998). The crucial role of the FGF-signalling pathway for the development of the MHB region was further demonstrated by targeted inactivation of genes that are regulated by FGF8. One target of FGF signalling is the transcription factor En1. In mice homozygously mutant in EN1, the formation of midbrain and hindbrain structures is impaired (Joyner et al., 1991; Joyner, 1996; Wurst et al., 1994). The secreted signalling molecule WNT1, which regulates EN
Key words: Medaka, MHB, FGF signalling, XFD, Dorsoventral asymmetry, shh, sal
5660 M. Carl and J. Wittbrodt
Fig. 1. Expression patterns of FGFR1, FGFR2, FGFR3 and FGFR4 during medaka embryonic development. Embryos are stained with probes indicated at the bottom. Dorsal views, anterior is to the left. (A-D) Embryos at late neurula (stage 18), (E-H) at early somitogenesis (stage 21) and (I-L) at early organogenesis stages (stage 24). (A,E,I) FGFR1 is expressed exclusively in posterior regions; (I) higher magnification of the tail shows expression of FGFR1 in the lateral somites (arrowhead). (B,F,J) FGFR2 is expressed adjacent to the MHB from early neurula stages onwards. In the developing brain FGFR3 (C,G,K) and FGFR4 expression (D,H,L) is detected in the midbrain region from early neurula stages onwards and at subsequent stages also in the anterior hindbrain. de, diencephalon; ele, equatorial region of the lens epithelium; hb, presumptive anterior hindbrain; le, lens epithelium; mb, presumptive midbrain; ov, optic vesicle; plf, primary lens fibre cells; pm, presomitic mesoderm.
manner by XFD. We demonstrate in coinjection experiments that apparently functional FGF signalling is necessary for HHdependent sal expression at the MHB. A similar interaction is seen in the forebrain, where HH acts in proximodistal patterning of the outbudding optic vesicles. Also here we find a requirement for functional FGF signalling for HH to activate sal. MATERIALS AND METHODS Medaka stocks Wild-type Oryzias latipes from a closed stock at the MPI for Biophysical Chemistry and the EMBL were kept as described (Köster et al., 1997). Isolation of Fgf8 and Wnt1 partial cDNAs A 545 bp Fgf8 PCR fragment was RT-PCR amplified from total RNA isolated from organogenesis stage embryos (Iwamatsu, 1994) using PCR primers specific for Fgf8: 5′-CAGCATGTGAGKGAGCARAG and 3′-GGAAGATCTACGCTCTCCTGAGTAGCG GGTG. Cycling conditions were: 3 minutes at 94°C, 35 cycles at 94°C, 30 seconds; 50°C, 1 minute; 72°C, 2 minutes. The resulting PCR product was cloned into the TopoTA vector (Invitrogen) and sequenced. EMBL database accession number: AJ243210. A 789 bp Wnt1 PCR fragment was RT-PCR amplified from total RNA isolated of organogenesis stage embryos. Primers: 5′GCARTTYMGIAACMGIMGITGGA and 3′-CAISWSACRTGRCARCACCARTGRAA. Cycling conditions: 3 minutes at 94°C, 10 cycles at 94°C, 30 seconds; 55°C, 1 minute; 72°C, 2 minutes, 25 cycles with annealing at 60°C. The resulting PCR product was cloned into the TopoTA vector and sequenced. EMBL database accession number: AJ243208.
Plasmids The dominant negative FGF-receptor plasmid, XFD, and the control plasmid, HAV, were provided by E. Amaya (Amaya et al., 1991, 1993). The respective RNA was transcribed using the SP6 mMESSAGEmMACHINE (Ambion). RNA was injected at concentrations between 150 and 500 ng/µl for XFD and 370 ng/µl for HAV. XFD RNA was injected at 250 ng/µl for subsequent analysis. RNA encoding hGFP was transcribed and injected at 60 ng/µl as described (Loosli et al., 1999). shh RNA was injected at 125 ng/µl as described (Köster et al., 1997). All injections of purified RNAs (Qiagen RNeasy) into one blastomere of embryos at the 2- to 4-cell stage (Köster et al., 1997) were carried out as described (Loosli et al., 1999). Whole-mount in situ hybridisation Whole-mount in situ hybridisation was performed using digoxigeninand fluorescein-labelled RNA probes as described (Loosli et al., 1998). The entire cloned cDNAs of FGFR1, FGFR2, FGFR3, FGFR4 (Emori et al., 1992), sal, Pax2, Fgf8, Wnt1 and En2 (Ristoratore et al., 1999) were transcribed for RNA antisense riboprobes.
RESULTS Medaka FGF-receptor expression during medaka embryonic development To investigate which of the four known medaka FGF receptors (FGFRs) is expressed at the MHB and thus can mediate FGF signalling, we examined their temporal and spatial expression patterns during medaka embryonic development by wholemount in situ hybridisation. At neurula stages (stage 17/18; Iwamatsu, 1994), FGFR1 is weakly expressed in the tailbud and the presomitic mesoderm (Fig. 1A). From early
DV asymmetry of FGF signalling at MHB 5661 somitogenesis stages (stage 21) onwards, expression is found in the somites and transiently in a transverse stripe at the level of rhombomere seven (Fig. 1E). At subsequent stages, FGFR1 is expressed exclusively in posterior regions of the developing embryo (lateral tips of the anterior somites, notochord), suggesting that this receptor is not involved in FGF signalling at the MHB (Fig. 1I). At late neurula stages (stage 18), the expression domains of FGFR2, FGFR3 and FGFR4 overlap in the region of the presumptive midbrain (Fig. 1B-D). In addition, FGFR2 is expressed in a transverse stripe in the region of the presumptive anterior hindbrain. Thus, with the exception of FGFR1, the FGF receptors are expressed in transverse stripes adjacent to the presumptive MHB long before the MHB is morphologically visible. At early somitogenesis stages (stage 21), the initially overlapping anteriormost expression domains of FGFR2, FGFR3 and FGFR4 in the presumptive midbrain are refined to partially overlapping regions (Figs 1F-H, 2). At this stage, expression of all three receptors is detectable in the anterior hindbrain. At early organogenesis stages (stage 24), FGFR2, FGFR3 and FGFR4 are expressed in the posterior forebrain and the anterior midbrain (Fig. 1J-L). Expression of FGFR2 and FGFR4 is also detected in the posterior part of the midbrain. FGFR3 is expressed in the entire hindbrain (Figs 1K, 2C), with the exception of the anteriormost region, similar to FGFR4 (Figs 1L, 2D). FGFR2 expression is detectable in the entire hindbrain and is only excluded from the dorsalmost part (Fig. 1J). Expression of all three FGFRs is excluded from a narrow domain at the MHB, where Fgf8 is expressed. From neurula stages onwards, FGFR3 and FGFR4 are also expressed in the presomitic mesoderm (Fig. 1B,C) and expression persists in the three anteriormost somites during subsequent somitogenesis. Weak FGFR2 expression is found in all somites (Fig. 1F-H). Complementary expression is detected in the developing eye (Fig. 1J-L): FGFR2 is expressed in the lens epithelium, FGFR4 in the equatorial region of the lens epithelium and FGFR3 in primary lens fibre cells. In the eye cup, FGFR2 and FGFR4 are expressed in the pigmented retina, FGFR2 is expressed in the ciliary margin of the neural retina. To determine the exact location of the respective FGFR expression domains within the MHB, Pax2 expression was used as a landmark (Fig. 2). Pax2 is expressed in the MHB and the posterior midbrain from late neurula stages onwards and is prominent at early somitogenesis stages (stage 21) (Köster et al., 1997). At neurula stages, FGFR2-expressing cells are located directly adjacent to and partially overlapping with Pax2-expressing cells in the anterior part of the Pax2 expression domain (Fig. 2A). At stage 21, FGFR2 expression in the midbrain overlaps posteriorly with the Pax2 expression, while FGFR3 and FGFR4 are expressed further rostral in the midbrain and do not overlap with Pax2 (Fig. 2B-D). In the anterior hindbrain, FGFR2 is expressed directly adjacent to Pax2, while the overlapping expression domains of FGFR3 and FGFR4 are located further caudally at the level of the second rhombomere (Fig. 2B-D). Dose-dependent interference with FGF signalling by a dominant negative FGF receptor To suppress FGF signalling, RNA encoding a dominant
Fig. 2. Expression analysis of FGFR2, FGFR3 and FGFR4 within the MHB. FGFRs (purple) in double whole-mount in situ analysis with Pax2 (red) as a landmark for the MHB region. Anterior is to the left, dorsal (A) and lateral views of the brain region (B-D). (A) FGFR2 is expressed overlapping with the anteriormost Pax2expressing cells from neurula stages (stage 18) onwards. (B) At early somitogenesis (stage 21), FGFR2 expression overlaps the Pax2 expression domain in the midbrain. (C,D) FGFR3 and FGFR4 expression is detected more rostrally in their midbrain and, more caudally, hindbrain expression domains. Abbreviations, see Fig. 1; ey, eye; r, rhombomere.
negative FGF receptor (XFD) lacking the intracellular tyrosine kinase domain (Amaya et al., 1991) was injected into one cell of 2- to 4-cell-stage medaka embryos. For control, RNA encoding a non-functional form of XFD (HAV; Amaya et al., 1993) was injected under identical conditions. Injection of HAV RNA did not affect embryonic development (Table 1). Coinjection of RNA encoding the green fluorescent protein (GFP) as a lineage tracer revealed that the injected RNA was evenly distributed in the embryo (data not shown). Injection of XFD RNA results in a severe reduction or absence of trunk and tail (Table 1). Moreover, anterior brain defects and strong cyclopia were observed. To investigate the effects of suppressing FGF signalling in more detail, we performed a dose-response analysis. Injection of XFD RNA at
5662 M. Carl and J. Wittbrodt Table 1. Embryos at early somitogenesis stage (stage 21) injected with XFD RNA show an underdevelopment or absence of trunk and tail structures and anterior brain defects % embryos showing defects RNA concentration No (ng/µl) effect 370 (XFD) 250 (XFD) 370 (HAV)
2 13 81
Brain affected, tail underdeveloped 69 64 14
Brain Number of affected, Not embryos no tail gastrulated injected 15 8 0
14 15 5
99 715 63
HAV, non-functional form of the XFD construct.
high concentrations (370 ng/µl) leads to severe trunk, tail and anterior brain defects. Embryos injected with RNA at an intermediate concentration (250 ng/µl) show less severe alterations. We observed a weaker reduction of tail structures and partial ventral fusion of eye structures (mild cyclopia). The resulting effects are highly consistent and tightly correlated with the amount of RNA injected. At all XFD concentrations analysed, the region of the MHB was morphologically unaffected. Injection of higher RNA concentrations (>500 ng/µl) lead to severe gastrulation defects, while XFD injections at lower concentrations (
