Inhibition of BMPs by follistatin is required for FGF3 expression and segmental patterning of the hindbrain

Developmental Biology 324 (2008) 213–225

Contents lists available at ScienceDirect

Developmental Biology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / d eve l o p m e n t a l b i o l o g y

Inhibition of BMPs by follistatin is required for FGF3 expression and segmental patterning of the hindbrain Karen Weisinger a, David G. Wilkinson b, Dalit Sela-Donenfeld a,b,⁎ a b

Koret School of Veterinary Medicine, The Hebrew University, Faculty of Agriculture, Food and Environmental Quality Sciences, P.O. Box 12, Rehovot 76100, Israel Division of Developmental Neurobiology, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK

a r t i c l e

i n f o

Article history: Received for publication 17 January 2008 Revised 4 September 2008 Accepted 5 September 2008 Available online 18 September 2008 Keywords: Follistatin BMP4 FGF3 Hindbrain Rhombomere Segmentation Chick

a b s t r a c t A network of molecular interactions is required in the developing vertebrate hindbrain for the formation and anterior–posterior patterning of the rhombomeres. FGF signaling is required in this network to upregulate the expression of the Krox20 and Kreisler segmentation genes, but little is known of how FGF gene expression is regulated in the hindbrain. We show that the dynamic expression of FGF3 in chick hindbrain segments and boundaries is similar to that of the BMP antagonist, follistatin. Consistent with a regulatory relationship between BMP signaling and FGF3 expression, we find that an increase in BMP activity due to blocking of follistatin translation by morpholino antisense oligonucleotides or overexpression of BMP results in strong inhibition of FGF3 expression. Conversely, addition of follistatin leads to an increase in the level of FGF3 expression. Furthermore, the segmental inhibition of BMP activity by follistatin is required for the expression of Krox20, Hoxb1 and EphA4 in the hindbrain. In addition, we show that the maintenance of FGF3 gene expression requires FGF activity, suggestive of an autoregulatory loop. These results reveal an antagonistic relationship between BMP activity and FGF3 expression that is required for correct segmental gene expression in the chick hindbrain, in which follistatin enables FGF3 expression by inhibiting BMP activity. © 2008 Elsevier Inc. All rights reserved.

Introduction Subdivision of the neural epithelium into domains along the anteroposterior (AP) axis, each with a distinct regional identity, is a crucial early phase in the development of the vertebrate nervous system. The establishment of such subdivisions involves signals, initially from adjacent tissues and later from within the neural epithelium itself, that induce the spatially-restricted expression of transcription factors that specify regional identity. Understanding how the nervous system is patterned requires elucidation of this regulatory network of cell signalling and transcription factors (Lumsden and Krumlauf, 1996; Shimamura et al., 1997). Significant progress has been made in identifying genes that underlie regional specification in the hindbrain (Lumsden, 2004; Lumsden and Krumlauf, 1996; Moens and Prince, 2002). At early stages of nervous system development, the hindbrain is subdivided along the AP axis into a series of segments, the rhombomeres (r), each a cell lineage-restricted compartment with a distinct regional identity (Fraser et al., 1990). This segmentation of the hindbrain organises patterning of neurons and neural crest cells in the branchial region of the head (Graham et al., 1996; Keynes et al., 1990; Trainor and Krumlauf, 2000; Trevarrow et al., 1990). A network of transcription factors, including Krox20, vHnf1, Kreisler/val and Hox family members, governs the ⁎ Corresponding author. E-mail address: [email protected] (D. Sela-Donenfeld). 0012-1606/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2008.09.005

formation and AP specification of rhombomeres (Aragon et al., 2005; Barrow et al., 2000; Gavalas et al., 2003; Giudicelli et al., 2003; Krumlauf et al., 1993; Manzanares et al., 1999; Moens et al., 1998; Prince et al., 1998; Schneider-Maunoury et al., 1997; Seitanidou et al., 1997; Voiculescu et al., 2001) and regulates expression of members of the Eph and Ephrin families of cell surface receptors that restrict cell movement between adjacent segments (Cooke et al., 2001; Mellitzer et al., 1999; Theil et al., 1998; Xu et al., 1995, 1999a). FGF signaling proteins have been found to be involved in the upstream regulation of this complex genetic network (Aragon et al., 2005; Bel-Vialar et al., 2002; Marin and Charnay, 2000; Maves et al., 2002; Roy and Sagerstrom, 2004; Walshe et al., 2002; Wiellette and Sive, 2003, 2004). Studies in zebrafish have shown that the expression of FGF3 and FGF8 in r4 is required to upregulate Krox20 and Kreisler/val expression in r5 as well as establish segmentation of the r5/r6 region (Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003). Similarly, the results of blocking or activating FGF signaling in the chick hindbrain reveals that FGFs are required for Krox20 and Kreisler expression (Aragon et al., 2005; Marin and Charnay, 2000). Furthermore, these manipulations disrupted the normal expression of Hox, Eph receptor and Ephrin genes, and resulted in the loss of hindbrain segmentation (Marin and Charnay, 2000; Walshe et al., 2002). Although it is not known which members of the FGF family account for this role in hindbrain patterning in higher vertebrates, a candidate is FGF3, which has dynamic segmental expression in the chick and mouse hindbrain (Mahmood et al., 1995, 1996; Powles et al., 2004; Wilkinson et al.,

214

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

1988). FGF3 is required for induction and patterning of the adjacent otic vesicle (Alvarez et al., 2003; Leger and Brand, 2002; Mansour et al., 1993; Maroon et al., 2002; Phillips et al., 2001; Zelarayan et al., 2007), but as hindbrain segmentation is normal in FGF3 null mutants (Wright and Mansour, 2003) it is currently unclear whether it also has a role within the hindbrain, perhaps acting redundantly with other FGFs. Studies of patterning in a number of tissues have revealed antagonistic relationships between the FGF and BMP signaling systems. This antagonism can regulate cell fate, such as in neural induction when BMP signaling is necessary for epidermal cell specification, while FGF signaling and BMP inhibitors are both required for ectodermal cells to acquire a neural fate (De Robertis and Kuroda, 2004; Delaune et al., 2005; Furthauer et al., 2004; Stern, 2002). Inhibition of FGF expression by BMPs can regulate the establishment of regional domains, for example in the forebrain, in which this underlies reciprocal expression of FGF8 and BMP4 in distinct domains (Anderson et al., 2002; Ohkubo et al., 2002; Shimogori et al., 2004). Furthermore, the expression of the BMP inhibitor noggin in the telencephalon enables the upregulation of FGF8 and is itself negatively regulated by BMPs (Anderson et al., 2002; Ohkubo et al., 2002; Shimogori et al., 2004). In view of the antagonistic relationships between FGF and BMP signaling being required for patterning elsewhere in the nervous system, it is intriguing that there appear to be similarities in the expression of the BMP inhibitor follistatin (Albano et al., 1994; Connolly et al., 1995; Graham and Lumsden, 1996) and of FGF3 (Mahmood et al., 1995, 1996) in the chick and mouse embryo hindbrain. Previous studies have shown that BMP induces apoptosis of r3/r5 neural crest cells (Graham et al., 1994), and that Noggin prevents this activity in r4 (Smith and Graham, 2001), but the role of follistatin in the hindbrain has not been elucidated. We have therefore addressed the possibility that antagonism between BMP and FGF contributes to the segmental regulation of gene expression in the hindbrain. In this paper we report that FGF3 expression in specific rhombomeres requires the segmental inhibition of BMP activity by follistatin and that this interaction regulates segmental gene expression in the hindbrain. Materials and methods Embryos Fertile Loman chick eggs were incubated at 38 °C until embryos reached the desired somite-stage (ss). Before performing experimental procedures, eggs were windowed and embryos were visualised by injecting black ink below the blastodisc. Following manipulations, embryos were incubated to the required stage, fixed in 4% paraformaldehyde, dehydrated in 100% methanol, and stored at −20 °C. In ovo electroporation FITC-conjugated follistatin or control antisense morpholino (MO) oligonucleotides were diluted in PBS to a working concentration of 2 mM. Two sets of follistatin-MO were designed according to the manufacturer's criteria (GeneTools, OR USA), one targeted to prevent translation of the protein (follistatin-MO) and the other to create a misspliced protein (Ex2 Int2-MO) by blocking the splice region between exon 2 and intron 2. The sequences used are as follows: follistatin-MO: 5′-TAACATCCTCAGCTTAGCAGGGAGT-3′, Ex2 Int2-MO: 5′-TCCACATTCTCACATGTTTCTTTGC-3′, control-MO: 5′-CCTCTTACCTCAGTTACAATTTATA-3′. Full-length mouse BMP4 cDNA was inserted upstream of an internal ribosomal entry site (IRES) and nuclear localization sequence-tagged GFP in pCAGGS expression vector plasmid (a gift from J. Briscoe). pCAGGS-BMP-IRES-GFP and control pCAGGS-IRES-GFP were diluted in 10 mM Tris pH 8.5 to a working

concentration of 100 ng/μl or 2 μg/μl, respectively. Full-length Xenopus follistatin cDNA in CMV expression vector (a gift from N. Itasaki) was diluted in 10 mM Tris pH 8.5 to a working concentration of 2 μg/μl. MO-oligonucleotides or cDNAs were injected into the hindbrain lumen of embryos by using a pulled glass capillary. Following injection, electrodes were placed either above and below the embryo to obtain bilateral hindbrain transfection or at both the left and right sides of the hindbrain to obtain unilateral and dorsal transfection. Electroporation was performed using a BTX 3000 electroporator with four 45-millisecond pulses of 16–18 V and pulse intervals of 300 ms (Itasaki et al., 1999). Bead implantation Heparin acrylic beads were soaked at 4 °C for 2 h in FGF3 (1 mg/ml), follistatin (1 mg/ml), or BMP4 (10 ng/μl) (R&D systems, MN USA) all diluted in PBS-0.3% BSA, or in PBS-0.3% BSA solution as a control. AGX100 beads were soaked at room temperature for 2 h in SU5402 (200 μM, Calbiochem, CA USA) or DMSO. Beads were implanted in the hindbrain lumen of 4–8ss embryos, which were then incubated for further 3–20 h. Beads were removed before in situ hybridisation procedure. Whole-mount in situ hybridization and immunohistochemistry Whole-mount in situ hybridization was performed as described (Sela-Donenfeld and Kalcheim, 1999), using probes for chick FGF3 (EST clone 812g6, MRC Geneservice), follistatin (Connolly et al., 1995) BMP5 (EST clone 248c4, MRC Geneservice), BMP4 (Sela-Donenfeld and Kalcheim, 1999), Pax6, Nkx6.2 (a gift from J. Briscoe), Krox20 (a gift from P. Charnay) (Marin and Charnay, 2000), Hoxb1 (a gift from R. Krumlauf) and Msx1 (a gift from A. Graham). For double in situ hybridization, probes were either labelled with digoxigenin (DIG)-UTP or fluorescein-UTP and detected using alkaline phosphatase-coupled antibody (1:2000, Roche, Basel Switzerland). First, the DIG labelled probe was detected using NBT/BCIP as substrate (Roche, Basel Switzerland), after which embryos were treated with 100% methanol to inactivate alkaline phosphatase. Following the second antibody addition, embryos were stained with Fast Red (Roche, Basel Switzerland) to reveal the fluorescein-labelled probe. Whole-mount immunohistochemical localisation of proteins was carried out following some in situ hybridizations. Briefly, embryos were incubated in PBS with 0.1% Tween20, 5% goat serum for 2 h and then incubated with the following antibodies overnight: rabbit antiGFP (1:400, Molecular Probes, OR USA), sheep anti-fluorescein to detect FITC-conjugated MO-oligonucleotides (1:2000, Roche, Basel Switzerland)., rabbit anti-EphA4 (1:250) (Irving et al., 1996), and mouse anti-chondroitin sulphate proteoglycan (CSPG, 1:50, Sigma, MO USA). Following PBS washes, the following secondary antibodies were added: anti-rabbit or anti-mouse Alexa 488, anti-rabbit Alexa 594 (all 1:400, Molecular Probes, OR USA), anti-rabbit-HRP (1:250, Sigma, MO USA). Depending on the reagents used, the embryos were then stained with DAB (diaminobenzidine tetrahydrochloride, Sigma, MO USA) to reveal HRP activity, with Fast Red (Roche, Basel Switzerland) to reveal FITC-MO expressing cells, or visualised for fluorescent labelling. Cell death detection assay Cell death in whole mount embryos was detected by terminal transferase UTP nicked end labelling (TUNEL) according to a modified manufacturer's protocol (Roche, Basel Switzerland). Briefly, embryos were fixed in 4% PFA for 1 h at room temperature, washed in PBS, blocked in 3% H2O2 (in methanol) for 1 h at room temperature, and again washed in PBS. Embryos were permeabilized (0.1% Triton, 0.1% sodium citrate) on ice for 15 min and washed in PBS. Embryos were

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

assayed by TUNEL using In Situ Cell Death Detection Kit, POD (Roche, Basel Switzerland). POD labelled cells were visualized using AEC substrate system (Lab Vision Corporation, CA USA). Results Expression pattern of FGF3, follistatin and BMPs in the developing chick hindbrain Based on previous studies, FGF3 and follistatin appear to be expressed in the chick and mouse hindbrain (Albano et al., 1994; Connolly et al., 1995; Graham et al., 1994; Graham and Lumsden, 1996; Mahmood et al., 1995, 1996; Oh et al., 1996), but the expression patterns of these genes have not been directly compared. We therefore compared in detail the dynamic expression of FGF3 and follistatin during chick hindbrain development. As early as the 2 somite-stage (ss), FGF3 is expressed in presumptive r4 throughout the dorsal–ventral (DV) extent of the segment (Supplementary Figs. 1A, A'). This expression continues in 4ss–8ss embryos with an expansion into presumptive r5 and r6 (Supplementary Figs. 1B,B',C and A). In comparison, in 2ss–4ss embryos follistatin mRNAs are expressed dorsally along the entire length of the neural folds, whereas in presumptive r4–r6, it is also distributed slightly more ventrally (Supplementary Figs. 1D–E'). In 6ss embryos follistatin expression is more evident throughout the DV extent of r4–r6, similar to FGF3, and its dorsal expression in r3 is no longer apparent (Supplementary Fig. 1F). In 8ss embryos, FGF3 and follistatin continue to be expressed in r4–r6 (Figs. 1A,G), and dorsal follistatin expression remains in r2, but is downregulated in the more rostral domains of the hindbrain (Fig. 1G).

215

In 12ss embryos, overlapping r4–r6 expression of FGF3 and follistatin continues and becomes more ventrally restricted, whereas FGF3 is now present also in r2 (Figs. 1B, H). A double in situ hybridization of FGF3 and Krox20 (expressed in r3 and r5) at this stage shows the specific segmental expression of both genes (Fig. 1S, FGF3 in blue and Krox20 in red). Subsequently, the expression of FGF3 and follistatin is progressively downregulated from the rhombomeres while being upregulated in the inter-rhombomeric boundary cells, while they remain excluded from the dorsal neuroepithelium (Figs. 1C–F, I–L). At 18ss, both genes are expressed in r2, r4 and r6 and are disappearing from r5 (Figs. 1C, I), while at 22ss the expression in r2 disappears and expression in r2/3, r3/4, r4/5 and r5/6 boundary cells is now evident (Figs. 1D, J). By 27ss, rhombomeric expression of FGF3 and follistatin is no longer detected in r4, is still present in r6, and their expression in boundary cells has become more prominent (Figs. 1E, K). In 30–45ss embryos, FGF3 and follistatin transcripts are confined to the hindbrain boundary cells and are absent from all rhombomere bodies (Figs. 1F, L and data not shown). At these stages, follistatin is also expressed in a DV restricted stripe along the hindbrain that may correlate with the alar–basal border of the neural tube (our unpublished data). Since follistatin acts as an inhibitor of several TGFβ family protein members such as activins and BMP4–7 (Iemura et al., 1998; Liem et al., 1997; Nakamura et al., 1990; Yamashita et al., 1995), we examined the expression patterns of these genes in the hindbrain of 2–40ss embryos and found that they are largely excluded from follistatin and FGF3 domains. In 2ss embryos BMP4 is dorsally restricted in the anterior nervous system (Supplementary Figs. 1G, G'). In 4ss embryos the dorsal expression of BMP4 is expanded posteriorly to the anterior part of the hindbrain (Supplementary Figs. 1H, H') and is further caudally

Fig. 1. Expression of FGF3, follistatin, and BMPs in the chick hindbrain. In situ hybridization was carried out to detect the expression of (A–F) FGF3, (G–L) follistatin, (M–O) BMP4, (P–R) BMP5 and (S) FGF3 (blue) and Krox20 (red) at the stages indicated, shown in flat mount preparations of the hindbrain. Dynamic segmental expression of FGF3 and follistatin occurs in a similar pattern, at early stages in rhombomeres and later in hindbrain boundary cells. BMP4 and BMP5 are expressed in the dorsal hindbrain. Rhombomeres are numbered, probes are indicated, floor plate (FP) is marked and anterior is at the top.

216

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

expanded to encompass the whole dorsal hindbrain in 6ss embryos (Supplementary Fig. 1I). In 8ss and 12ss embryos BMP4 is expressed along the dorsal AP axis of the hindbrain and is found at slightly higher levels in r3 and r5 as compared to the adjacent rhombomeres (Figs. 1M, N). At later stages the dorsal expression of BMP4 is evenly distributed along the entire hindbrain (Fig. 1O). The expression of BMP5 is reminiscent of that of BMP4 with higher levels of dorsal expression in r3 and r5, although for BMP5 the r3 expression expands ventrally in 12ss embryos (Figs. 1P, Q). The segmental differences of both BMP4 and BMP5 mRNA levels in these stages are consistent with previous studies, where BMPs were shown to control neural crest cell death in r3, r5 (Farlie et al., 1999; Graham et al., 1994) and locus coeruleus differentiation in r1 (Vogel-Hopker and Rohrer, 2002), although we observe dorsal expression of BMP4 and BMP5 also in even numbered rhombomeres (Figs. 1M, N, P, and Q, and see also Oh et al., 1996), probably due to differences in the hybridisation sensitivity. At later stages BMP5 expression is again similar to BMP4 and is evenly distributed along the entire dorsal hindbrain (Fig. 1R). This dorsal expression is consistent with roles of BMPs in DV patterning (Hornbruch et al., 2005; Lee and Jessell, 1999; Liem et al., 1997; Timmer et al., 2002). BMP7 and ActivinA are not expressed

in the hindbrain (data not shown, and see also Connolly et al., 1995; Oh et al., 1996), although BMP7 transcripts are detected in the adjacent otic vesicle and ectoderm as well as in ventral midline cells of more anterior regions (Dale et al., 1999; Oh et al., 1996), suggesting BMP7 to be an additional candidate for follistatin inhibition. Taken together, these data indicate that in early embryos of 2–4ss, the restricted AP expression of FGF3 is more prominent than follistatin in the hindbrain, a stage where BMP is not yet expressed in the caudal hindbrain region. From 6–40ss, FGF3 and follistatin show a more similar spatial and temporal segmental expression in the rhombomeres and later in boundary cells, an expression which gradually becomes restricted to the ventral side of the neuroepithelium, distal from the domain of BMP expression. FGF3 expression requires follistatin The expression profiles of FGF3 and follistatin raise the possibility that the segmental expression of FGF3 is dependent upon inhibition of BMPs by follistatin. To test this, we first examined whether loss of follistatin protein in the hindbrain affects FGF3 expression. FITCconjugated morpholino antisense oligonucleotides (MO) directed

Fig. 2. Inhibition of follistatin leads to loss of segmental FGF3. (A–H) Expression of FGF3 in the hindbrain of embryos electroporated with FITC-conjugated control-MO (A, B, E, and G), follistatin-MO (C, D, and F) or Ex2Int2-MO (H) at the indicated stages and further incubated for 14 h prior to fixation. Embryos electroporated with control-MO show normal expression of FGF3, whereas embryos electroporated with follistatin-MO or Ex2Int2 MO show reduced FGF3 expression. (I–L) Expression of Nkx6.2 (I, J) and Pax7 (K, L) in embryos electroporated at the indicated stages with control-MO (I,K) or follistatin-MO (J, L) and incubated for 16 h prior to fixation. Follistatin-MO does not alter either the normal ventral expression of Nkx6.2 or dorsal expression of Pax7 mRNAs. Probes are indicated, rhombomeres are numbered and anterior is at the top.

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

against the 5′ UTR of chick follistatin sequence to prevent translation (follistatin-MO), or a control FITC-conjugated MO (control-MO), were electroporated into the hindbrain of ∼4ss or ∼ 10ss embryos and the expression of FGF3 was analyzed 14 h later. FGF3 expression was normal in embryos electroporated with control-MO (Figs. 2A, B, and E, 14/14 embryos for each stage), but greatly reduced in the hindbrain of embryos electroporated with follistatin-MO (Figs. 2C, D, and F, 16/16 embryos for each stage). To verify our results, we electroporated embryos with an additional MO to counter follistatin, but in this instance the MO targeted the splice region between the second exon– intron junction (Ex2Int2-MO) thereby producing a misspliced protein (Eisen and Smith, 2008). Embryos were incubated for an additional 14 h. Our results show that, similarly to the previously used MO (Figs. 2C, D, and F), FGF3 expression was reduced in the embryos transfected with the Ex2Int2-MO (Fig. 2H n = 9/10) as opposed to the control (Fig. 2G n = 0/7). We noticed that in some cases the level of FGF3 expression was lowered even in domains which do not contain follistatin-MO. This may be due to more cells expressing follistatin-MO than our detection system can detect or it may also be possible that follistatin has a wide diffusion radius and that its concentration is limiting, such that follistatin knockdown by the MO causes it to be less available also in areas that are outside of the knockdown domain. To demonstrate that the effect of the follistatin-MO on the segmental expression of FGF3 is specific, we electroporated embryos with follistatin-MO, incubated for a further 14 h and thereafter analysed the expression of two non-segmentally expressed genes, one ventral, Nkx6.2, and one dorsal, Pax7. In both cases the expression pattern of the follistatin-MO electroporated embryos (Fig. 2J for Nkx6.2 n = 6/6 and 2L for Pax7 n = 13/16) was not different from that of the control embryos (Fig. 2I for Nkx6.2 n = 5/5 and 2K for Pax7 n = 9/9). The lack of effect on both genes further confirms the specific AP restricted activity of follistatin. Moreover, this indicates that although Pax7 is a target of dorsally-

217

derived BMPs (Timmer et al., 2002), its homogenous DV distribution along the hindbrain does not involve segmental inhibitory activity of follistatin. Together, these results indicate that follistatin is required for rhombomeric expression of FGF3. To test whether follistatin is sufficient for segmental FGF3 expression, we analyzed the effect of ectopic follistatin on FGF3 expression. Follistatin or PBS/BSA soaked beads were implanted into the hindbrain lumen of 4ss embryos and analyzed for FGF3 expression 14 h later (Figs. 3A–D). Embryos treated with follistatin had significantly stronger expression of FGF3 in the hindbrain compared to PBS/BSA treated embryos (compare Figs. 3A to B–D, n = 18/26 for follistatin, n = 18/20 for control). Furthermore, ectopic expression of FGF3 was evident in some embryos in the floor plate, dorsal regions or in r3 (Figs. 3B, C), domains which normally do not express FGF3 (Figs. 1A–F). These experiments indicate that excess follistatin increases FGF3 levels in endogenous domains and also induces FGF3 expression in ectopic hindbrain regions. In an additional approach, we electroporated embryos with a follistatin overexpression construct and then analysed its effect on the expression of FGF3. Similarly to the exogenous addition of follistatin, we noted an ectopic expression of FGF3, but in this case, in the dorsal domain of the hindbrain (Fig. 3F), where the construct was electroporated. This ectopic expression was not observed in the control embryos (Fig. 3E). To verify these results we performed a rescue experiment. We prevented the expression of endogenous follistatin by electroporation of follistatin-MO and then implanted either a control or a follistatin-soaked bead. Our results show that, as previously seen, FGF3 is reduced in follistatinMO electroporated embryos (Fig. 3G n = 7/7), but that the expression of FGF3 is largely re-established in the rescue experiment (Fig. 3H n = 5/6). Taken together, our results indicate that inhibition or application of follistatin results in loss or overexpression of FGF3, respectively, suggesting that the segmental expression/activity of follistatin is necessary for segmental expression of FGF3 in the hindbrain.

Fig. 3. Addition of follistatin causes the upregulation of FGF3. (A–D) Expression of FGF3 in 4ss embryos implanted with control (A) or follistatin (B–D) beads in their hindbrain lumen and incubated for a further 14 h prior to fixation. Follistatin treated embryos have increased FGF3 expression levels compared to control embryos and also ectopic FGF3 in ventral and dorsal regions as well as in r3 (black arrows). Since follistatin-treated embryos had significantly higher expression of FGF3, staining of the experimental and control embryos was stopped much sooner than in other experiments, resulting in a weaker FGF3 staining in control embryos in (A) compared to controls in (Figs. 2A, B). (E, F) Expression of FGF3 in embryos dorsally electroporated at ∼3ss with a control (E) or follistatin overexpression construct (F) and allowed to develop for a further 14 h. Follistatin electroporated embryos show ectopic expression of FGF3 (black arrows) as compared to control embryos. (G, H) Expression of FGF3 in the hindbrain of embryos treated at ∼ 3ss with either electroporation of follistatin-MO (G) or electroporation of follistatin-MO in conjunction with follistatin bead implantation (H). Embryos were incubated an additional 14 h prior to fixation. Embryos given the combined treatment of follistatin and follistatin-MO show a rescue of FGF3 expression relative to embryos treated with follistatin-MO alone. The red staining corresponds to MO expressing cells detected with anti-fluorescein antibody. Probes are indicated, rhombomeres are numbered, localization of bead implantation is marked and anterior is at the top.

218

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

Overexpression of BMP4 suppresses FGF3 expression To further test whether FGF3 expression requires inhibition of BMP signaling, we overexpressed BMP4 in the hindbrain to override the inhibition by follistatin. Embryos were electroporated unilaterally and dorsally with a vector that drives expression of either GFP alone or of BMP4 and GFP. The BMP4 construct was used at a low concentration of 100 ng/μl to drive BMP4 overexpression without generating excessive neuroepithelial cell death (Timmer et al., 2002), which we confirmed by performing a TUNEL assay on ∼15ss embryos after either GFP or BMP4 electroporation at ∼4ss. Both control embryos (Fig. 4F n = 7/7) and BMP4 electroporated embryos (Fig. 4G n = 7/8) showed similar amount of cell death. Rhombomeric FGF3 expression was strongly suppressed in embryos 14 h after electroporation of BMP4 vector at ∼ 4ss or ∼10ss (Figs. 4A–E, n = 23/25 for BMP4, 0/15 for control). The amount of FGF3 inhibition by BMP4 was dependent on the overexpression level indirectly detected by anti-GFP staining: limited transfection of BMP4 correlated with unilateral effects of FGF3 while more extensive overexpression resulted in bilateral suppression of FGF3 (Fig. 4, compare transfected cells in B versus C). Misexpression of BMPs in ventral regions of the neural tube was shown to broaden the expression of dorsal genes and to downregulate ventrally-expressed genes in the CNS (Liem et al., 1997; Timmer et al., 2005, 2002). Based on the results of previous studies (Hornbruch et al., 2005; Timmer et al., 2005), the dorsal and lower overexpression level

of BMP4 used in our experiments is predicted not to cause such dorsalisation. Since dorsalisation would lead to a loss of ventral FGF3 expression, it was therefore important to ascertain whether or not dorsalisation occurs in our experiments. ∼ 10ss embryos were electroporated unilaterally in the dorsal region of the hindbrain with BMP4 or control plasmids (at the same concentrations as in Figs. 4A–G), and the expression of two ventral genes, Nkx6.2 and Pax6, was analysed 24 h later. We found that expression of both Nkx6.2 (n = 12/ 15) and Pax6 (n = 5/5) in BMP4-expressing embryos was either indistinguishable from control embryos (n = 6/6 for Nkx6.2 and for Pax6) or only slightly reduced (compare Figs. 4I, M to Figs. 4H, L). Furthermore, double in situ hybridization for FGF3 and Nkx6.2 was carried out in other BMP4-electroporated embryos and revealed that within the same embryos, overexpression of BMP4 is not affecting Nkx6.2 expression, (compare dark blue staining in Fig. 4K, n = 6/7 with control in Fig. 4J, n = 4/4) whereas FGF3 expression was downregulated (compare red staining in Fig. 4K, n = 6/7 with control in Fig. 4J, n = 0/4). In addition, we examined the location of BMP4 activity by analysing the expression of Msx1, a downstream target gene of BMP signaling. 4ss embryos were electroporated with either control GFP or BMP4 and incubated 20 h prior to fixation. BMP4 electroporated embryos showed a slightly stronger Msx1 expression level as well as a unilateral ventral expansion (Fig. 4O n = 14/16) compared to the control embryos (Fig. 4N n = 1/14). However, Msx1 medialisation did not extend to regions where FGF3, Pax6 or Nkx6.2 are expressed,

Fig. 4. Overexpression of BMP4 inhibits FGF3 expression. (A–E) Expression of FGF3 in the hindbrain of embryos unilaterally electroporated at the indicated stages with control pCAGGS-IRES-GFP (A, D) or BMP4-IRES-GFP (B, C, and E) plasmids and allowed to develop for 14 h (electroporated side marked by an asterisk). GFP expression in electroporated cells was detected with HRP-conjugated antibody (brown stain). FGF3 expression occurs at normal levels in control electroporated embryos and is absent or at low levels in embryos electroporated with BMP4 expression vector. The effect of BMP4 on FGF3 inhibition is stronger with higher levels of BMP4 overexpression. (F, G) embryos electroporated at ∼3ss with either control pCAGGS-IRES-GFP (F) or BMP4-IRES-GFP (G), allowed to develop for 20 h and thereafter analysed for cell death revealed by TUNEL assay. Cell death is similar in both control and BMP4 electroporated embryos. (H–M) Expression of Nkx6.2 (H–K; dark blue in J, K), FGF3 (J, K; red staining) and Pax6 (L, M) in embryos unilaterally electroporated at ∼10ss with control GFP (H, J, and L; insert in J shows electroporated area) or BMP4 (I, K, and M; insert in K shows electroporated area) and incubated for 24 h prior to fixation. Overexpression of BMP4 does not alter the normal ventral expression of Nkx6.2 and Pax6 mRNAs, whereas FGF3 expression is inhibited. (N, O) Expression of Msx1 in embryos unilaterally electroporated at ∼ 3ss with control GFP (N) or BMP4 (O) and incubated for 20 h prior to fixation. Msx1 expression appears stronger and more ventral in BMP4 electroporated embryos. Note that the ventral expression of Msx1 is slightly more pronounced on the electroporated side (compare width of arrows on the electroporated side). (P–U) FGF3 expression in embryos implanted at ∼8ss with PBS/BSA (P, R, and T) or BMP4 (Q, S, and U) soaked beads and allowed to develop for 3, 6 and 15 h, as indicated. FGF3 expression is at normal levels in control embryos (P, R, and T) and 3 h after addition of BMP4 (Q), but is lower in BMP4-treated embryos after 6 h (S) and absent after 15 h (U). Rhombomeres are numbered, otic vesicle (OV) is marked, probes are indicated, beads position is marked and anterior is at the top.

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

219

arguing against the possibility that the inhibition of FGF3 expression by BMP4 in our experimental conditions is due to general dorsalisation of the hindbrain. To determine how rapidly BMP4 downregulates FGF3 expression, we examined the time course of the effect by implanting BMP4soaked beads into the hindbrain of 8ss embryos. After 3 h of treatment, control (n = 6/8) and BMP4-treated embryos (n = 5/7) showed similar FGF3 expression (Figs. 4P, Q although in a few BMP4-treated embryos a slight reduction was detected (n = 2/8, data not shown). 6 h after BMP addition, the expression of FGF3 was significantly downregulated in the hindbrain (n = 6/8), in contrast to control embryos (n = 0/6) (Figs. 4R, S). This downregulation was enhanced in embryos receiving BMP4 beads for 15 h (n = 8/10 for BMP4, n = 0/6 for control) (Figs. 4T, U). FGF signaling maintains FGF3 expression in the hindbrain In a number of embryonic tissues FGFs maintain their own expression in a positive feedback loop (Agarwal et al., 2003; Canning et al., 2007; Mandler and Neubuser, 2004; Sekine et al., 1999; Trokovic et al., 2003; Xu et al., 1999b). We therefore addressed whether such autoregulation may contribute to FGF3 expression in the hindbrain. Beads soaked in SU5402 (a pharmacological inhibitor of FGF signaling) or DMSO as a control, were implanted into the hindbrain lumen of ∼ 4ss or ∼12ss embryos, which were fixed 16 or 20 h later, respectively. SU5402 treatment prevented FGF3 expression in embryos at both stages (Figs. 5B, D; n = 7/8 for 4ss embryos, n = 12/15 for 12ss embryos). In contrast, embryos receiving DMSO-beads expressed FGF3 in the normal pattern in r4–6 (Figs. 5A; n = 8/8) and in r6 and boundary cells (Fig. 5C; n = 15/15,). Moreover, similarly treated embryos showed no effect on follistatin expression in the hindbrain (Figs. 5E, F), suggesting that FGF signalling does not positively regulate follistatin expression, as well as indicating that FGF3 downregulation in SU5402 treatment is not due to toxicity. This result indicates that general blocking of FGF receptors prevents FGF3 expression. To further examine whether FGF3 autoregulates its expression, ∼12ss embryos were treated with FGF3 or PBS/BSA soaked beads and fixed 20 h later. FGF3 treated embryos had significantly higher expression levels of FGF3 (n = 8/11) compared to control embryos (n = 0/10) (Figs. 5G, H). The high levels of FGF3 expression were evident in both the boundary regions and in r4/r6, whereas in control embryos at this stage FGF3 has been downregulated from rhombomeres and confined to rhombomere boundaries. These data indicate that FGF3 expression is regulated by a positive feedback loop. Thus, we suggest that a combination of FGF autoregulation together with the inhibition of BMP activity by follistatin are involved in FGF3 expression in specific hindbrain domains. Segmental expression of Krox20 is obstructed by the inhibition of follistatin activity The segmentally-expressed transcription factor Krox20 regulates the formation and AP specification of odd-numbered rhombomeres (Frohman et al., 1993; Giudicelli et al., 2003; Lumsden and Krumlauf, 1996; Manzanares et al., 1999; Moens et al., 1998; SchneiderMaunoury et al., 1997; Seitanidou et al., 1997; Voiculescu et al., 2001). Based on evidence that the expression of this gene requires FGF signaling (Aragon et al., 2005; Marin and Charnay, 2000; Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003), we analysed whether it is affected by disruption of the FGF3–follistatin–BMP network. First, we confirmed that, as described previously (Marin and Charnay, 2000; Maves et al., 2002; Walshe et al., 2002, Aragon, 2005), Krox20 expression requires FGF signaling, since implantation of SU5402 beads into the hindbrain of ∼4ss embryos led to a loss of Krox20 expression in r3 and r5 20 h later (Fig. 6E, n = 13/17) in contrast

Fig. 5. FGF3 maintains its own expression. (A–F) Embryos implanted at the indicated stages with beads soaked in DMSO (A, C, and E) or SU5402 (B, D, and F), incubated for 16–20 h, fixed and stained for FGF3 (A–D) and follistatin (E, F). FGF3 is absent in SU5402 treated embryos (B, D) compared to control (A, C) while follistatin expression is similar in both SU5402 treated (E) and control (F) embryos. (G, H) FGF3 expression in the hindbrain of embryos implanted at the indicated stages with PBS/BSA (G) or FGF3 (H) beads and incubated for 20 h prior to fixation. FGF3 treated embryos (H) have higher levels of FGF3 expression as compared to control embryos (G). Rhombomeres are numbered, probes and bead positions are indicated and anterior is at the top.

to control embryos where Krox20 expression remained intact (Fig. 6A, n = 10/10). We next analyzed whether the disruption of endogenous follistatin protein expression affects Krox20 expression. ∼4ss embryos were electroporated with control or follistatin-MOs and fixed 20 h later. Embryos electroporated with control-MO had normal Krox20 expression in r3 and r5, with no differences between the electroporated side and the contra-lateral side (Fig. 6B n = 12/12). Notably, MO-labelled cells were commonly found within Krox20 domains, without affecting its expression. In contrast, in follistatin-MO electroporated embryos, Krox20 expression was reduced to very low levels or eliminated (Figs. 6F, G n = 9/12). Importantly, when electroporation of follistatin-MO was restricted to the dorsal-most region of the hindbrain, where endogenous follistatin is not expressed (Fig. 1), Krox20 expression was not disrupted (Fig. 6C n = 4/4), as compared to the ventrally-electroporated embryos (Figs. 6F, G), further confirming the specificity of follistatin-MO effect on Krox20. To verify our results we preformed a rescue experiment on the expression of Krox20. We electroporated ∼ 4ss embryos with follistatin-MO and

220

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

analyzed Krox20 phenotype, after adding an FGF3-soaked bead in conjunction with follistatin-MO electroporation. Our results show that, as previously shown, Krox20 is reduced in follistatin-MO electroporated embryos (Fig. 6D n = 7/7), but that the expression of Krox20 is recuperated in the rescue experiment (Fig. 6H n = 5/6). The effect of follistatin-MO on Krox20 expression was long-range as Krox20 was suppressed also in the contra-lateral side that expresses less MO (Figs. 6E, F). This result may indicate that more cells express follistatin-MO and induce Krox20 downregulation, than the cells stained by our detection system. Alternatively, it is possible that the non-cell-autonomous effect of follistatin-MO corresponds to the release of hindbrain BMPs from inhibition by follistatin, which in turn downregulate FGF3 expression. To further test this possibility, ∼8ss embryos were unilaterally electroporated with BMP4-GFP expression vector and analyzed 20 h later. While in control GFP electroporations, Krox20 expression occurred in normal r3/r5 domains (Fig. 7A, n = 10/10), it was substantially or totally inhibited in BMP4-transfected embryos (Figs. 7B, C, n = 15/17). Similar to our previous experiments, the effect of BMP4 was non-cell-autonomous and dependent upon the extent of electroporation: lower levels of Krox20 expression and bilateral downregulation correlated with a higher amount of BMP4/GFPexpressing cells (compare green-labelled cells in Fig. 7B' with C'). Together, these and the previous results show that blockage of FGF signaling, knockdown of follistatin, or overexpression of BMPs, all suppress Krox20 expression, consistent with the effect of the same manipulations on FGF3 expression. EphA4 is a direct target of Krox20 transcriptional activity (Theil et al., 1998) that interacts with ephrin ligands to restrict cells from crossing between rhombomeres (Xu et al., 1995, 1999a), and enables the formation of inter-rhombomeric boundaries (Cooke et al., 2005; Xu et al., 1995). We therefore analyzed whether the expression of EphA4 protein and the formation of boundary cells are also affected by BMP4 overexpression. We found that embryos overexpressing BMP4

showed decreased EphA4 expression on the electroporated side of the hindbrain (Fig. 7G, n = 7/9), in contrast to control embryos (Fig. 7F, n = 0/10). To analyse boundary cell formation, embryos ectopically expressing BMP4 were incubated to later stages when Chondroitin Sulphate Proteoglycan (CSPG) normally accumulates in chick hindbrain boundaries (Heyman et al., 1995). CSPG staining in hindbrain boundaries is disrupted in BMP-transfected embryos (Fig. 7I; n = 5/5). In contrast, control embryos displayed normal rhombomere boundaries marked by CSPG (Fig. 7H, n = 5/5). Previous works have suggested that neither the expression of Hoxb1 in r4 requires FGF signaling within the hindbrain (Bel-Vialar et al., 2002; Marin and Charnay, 2000; Maves et al., 2002) nor is it affected in Krox20 null mice (Schneider-Maunoury et al., 1993). While follistatin addition or inhibition up- or downregulates FGF3 and Krox20 expression, respectively, (Figs. 2, 3, and 6) we anticipated that follistatin activity would not alter Hoxb1 expression. We implanted 4ss embryos with either PBS- or follistatin-soaked beads and analysed Hoxb1. Our results show that, as expected, both, control and follistatin treated embryos display a similar expression levels of Hoxb1 (Supplementary Fig. 2). This result indicates that Hoxb1 is not affected by follistatin activity, which decreases BMP signaling and leads to an increase in FGF3 levels (Figs. 2, 3). Moreover, this suggests that the increase in FGF3 levels, observed by the addition of follistatin (Fig. 3), is not likely to be mediated by Hoxb1. Surprisingly, the opposite approach of BMP4 overexpression, which suppresses FGF3 and Krox20 (Figs. 4, 7) caused a decrease in Hoxb1 expression in r4 and in the caudal hindbrain, (Fig. 7E; blue signal Hoxb1; red signal Krox20; n = 6/6), as compared to control embryos (Fig. 7D, n = 0/8). Interestingly, these embryos also showed Hoxb1 upregulation in the floor plate of the posterior rhombomeres (Fig. 7E and data not shown). This unexpected effect of excess BMP on Hoxb1 may suggest that excessive BMP alters the expression of other factors that regulate Hoxb1 gene expression in the hindbrain, not mediated by follistatin, as further elaborated in the discussion.

Fig. 6. Inhibition of follistatin reduces segmental Krox20 expression. (A–H) Expression of Krox20 in embryos manipulated at ∼ 4ss and fixed 20 h later. (A, E) Embryos were implanted in the hindbrain lumen with beads soaked in DMSO (A) or SU5402 (E). Krox20 expression is reduced upon SU5402 treatment. (B, C, F, and G) Embryos electroporated with control-MO (B) or follistatin-MO (C,F,G). Krox20 expression is inhibited in follistatin-MO transfected embryos (F, G), but not in either control-MO (B) or dorsally electroporated follistatin-MO (C). (D, H) Embryos treated with either electroporation of follistatin-MO (D) or electroporation of follistatin-MO in conjunction with FGF3 bead implantation (H). Embryos given the combined treatment of FGF3 and follistatin-MO show a rescue of Krox20 expression relative to embryos treated with follistatin-MO alone. The red staining corresponds to MOexpressing cells labelled with anti-fluorescein antibody. Rhombomeres are numbered and anterior is at the top.

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

Discussion In this study, we have investigated regulatory interactions between follistatin, BMPs and FGF3 in early chick hindbrain. We found that follistatin is co-expressed with FGF3 in a dynamic pattern in hindbrain segments and subsequently in hindbrain boundaries. The results of gain- and loss-of-function experiments revealed that the local inhibition of BMP activity by follistatin is required for the expression of FGF3. In addition, FGF3 expression requires FGFR activation, suggesting that it acts in an autoregulatory loop. The segmental inhibition of BMP activity by follistatin is also required for the segmental expression of Krox20, which in turn regulates Eph receptors that underlie boundary formation in the hindbrain. These findings reveal a novel role for follistatin in establishing FGF3 expression and segmental patterning in the chick hindbrain, and indicate that an antagonistic interaction between BMP and FGF in rhombomeres is involved in regional specification in the hindbrain. Mechanisms that regulate FGF expression A number of studies have found that BMPs inhibit the expression of specific FGFs in other regions of the developing nervous system. At early stages of the anterior chick and mouse forebrain, BMP4 and FGF8 are expressed in reciprocal domains, and increase or inhibition of BMP4 signaling represses or enhances FGF8 expression, respectively (Anderson et al., 2002; Ohkubo et al., 2002). Moreover, increased BMP signaling also represses Shh in the prosencephalon, and Shh mutant mice have enhanced BMP signaling and loss of FGF8 expression (Anderson et al., 2002; Ohkubo et al., 2002). This antagonistic crossregulation of BMP4, FGF8 and Shh signaling centres regulate regional specification in the telencephalon and optic vesicles. Similarly, at later developmental stages in the cerebral cortex of mouse embryos, BMPs arising from the cortical hemisphere constrain FGF8 expression to the anterior telencephalon (Shimogori et al., 2004). This cross-inhibition is critical for Wnt1 expression and signaling (Gunhaga et al., 2003; Shimogori et al., 2004), and is integrated by the transcription factor

221

Emx2 in the early cortical primordium. Emx2 mutant mice have an expansion of Noggin and FGF8 expression, together with reduction of Wnt and BMP signaling, leading to severe hippocampal defects (Fukuchi-Shimogori and Grove, 2003). These findings suggest that Emx2 regulates Noggin and BMP expression and activity, which in turn controls FGF8 and Wnt1 signaling. It is therefore interesting that the Wnt antagonist molecule, SFRP2, is co-expressed with FGF3 and follistatin in even numbered rhombomeres in the chick hindbrain (Ellies et al., 2000), while Wnt1 is localized dorsally, similar to BMPs (Hollyday et al., 1995). Further investigation will be required to elucidate whether there is a regulatory relationship between FGF, BMP and Wnt expression in the hindbrain. Our findings show that FGF3 expression in the hindbrain is autoregulated by FGF signaling, as SU5402 or FGF3 application represses or enhances FGF3 expression, respectively. This data is at apparent variance with Aragon et al., (2005), where SU5402 was reported not to downregulate FGF3 expression. However, careful examination of their study shows that some loss of FGF3 expression may be seen upon FGFR inhibition, albeit to a lower extent than in our study. Moreover, in both studies Krox20 expression, which has been suggested to be regulated by FGF signaling in chick and in zebrafish (Marin and Charnay, 2000; Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003), is suppressed by SU5402 treatment. The discrepancy between these studies regarding the FGF3 autoregulation activity may be explained by the differences in the concentration or time exposure to the inhibitor, and in the ex-vivo conditions used in Aragon et al., (2005) as compared to the in-vivo approach used here. Positive feedback on FGF expression has been also found in several other tissues. For example, In the midbrain–hindbrain boundary FGF8 autoregulates its expression (Canning et al., 2007; Trokovic et al., 2003) and this positive feedback loop is essential for the formation and differentiation of the cerebellum and anterior hindbrain (Canning et al., 2007; Foucher et al., 2006; Jukkola et al., 2006; Mason et al., 2000; Roy and Sagerstrom, 2004; Trokovic et al., 2003). Moreover, during tooth morphogenesis, FGF3 expression and activity is regulated by FGF signaling (Jackman et al., 2004; Kettunen et al., 2000).

Fig. 7. Overexpression of BMP impairs the expression of segmental genes. (A–E) Embryos unilaterally electroporated at ∼8ss with control GFP (A, D) or BMP4-IRES-GFP (B, C, and E) plasmids and incubated for 20 h (electroporated side is marked by an asterisk). Krox20 expression is decreased in BMP4 electroporated embryos (B, C) compared to controls (A). Distribution of GFP-expressing cells which corresponds to the electroporated side are shown in panels A'–C' and are respective to images A–C. Similarly, BMP4 overexpression leads to decreased expression Hoxb1 (E, blue signal; Krox20 is red signal) compared to controls (D). (F–I) embryos were electroporated at ∼ 8ss with control (F, H) or BMP4 (G, I) plasmids and fixed 40 h later. Immunocytochemistry was carried out to detect EphA4 (F, G) or chondroitin-sulphate proteoglycan (CSPG) (H, I) proteins. BMP4 electroporated embryos have decreased expression of EphA4 in r3 and r5 (G), and of CSPG in hindbrain boundaries (I), compared to control transfected embryos (F,H). Rhombomeres are numbered, probes and antibodies are indicated at the top of each row and anterior is at the top.

222

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

Intriguingly, in feather placode initiation, FGF10 induces its own expression in the presence of low, but not high, BMP signaling (Mandler and Neubuser, 2004). We propose that an analogous situation occurs in the hindbrain, in which an autoregulatory loop maintains FGF3 expression in the presence of follistatin, but this loop is not established when FGF3 expression is inhibited by high BMP activity. Previous studies have found that several transcription factors are involved in the regulation of hindbrain FGF3 expression. Ectopic expression of the homeobox-containing gene vHnf1 resulted in the upregulation of chick FGF3 mRNA in both cell-autonomous and noncell-autonomous (Aragon et al., 2005). Also, FGF3 expression in the posterior mouse hindbrain was disrupted in Kreisler as well as in Hoxa1 mutants, (Carpenter et al., 1993; Frohman et al., 1993; Mark et al., 1993; McKay et al., 1996; Pasqualetti et al., 2001). The disruption of FGF3 expression in Hoxa1−/− embryos was partially rescued by addition of retinoic acid (RA), indicating that RA signaling is involved in the regulation of FGF3 expression in the posterior hindbrain (Pasqualetti et al., 2001). Other data from zebrafish embryos shows that Pbx and Meis homeodomain proteins interact with Hox group 1 genes to induce FGF signals in r4 (Waskiewicz et al., 2002). However, our results shows that Hoxb1 does not mediate FGF3 expression by follistatin. Further research will be required to determine whether and how follistatin activity to regulate FGF3 expression is combined with that of vHnf1, MafB/Kreisler and Hox and Pbx proteins, as well as with RA signaling pathways. An important question raised by our results is whether BMP signaling directly downregulates FGF3 expression, or acts via an intermediary factor. Our finding that FGF3 expression is downregulated within 6 h of implanting BMP4 beads is consistent with either possibility. However, the medial expansion of the BMP readout gene Msx1, observed in the BMP4 misexpressing embryos, does not overlap with the expression regions of FGF3, which are more ventral. This result favours the possibility that the segmental inhibition of FGF3, observed by overexpression of BMP4, is mediated through unknown factor(s) downstream of BMPs. Alternatively, several studies regarding the promoter/enhancer analysis of FGF3 gene have identified positive and negative regulatory regions, some of which affect the expression of FGF3 in the hindbrain (Murakami et al., 1993, 2001, 2004, 1999; Powles et al., 2004). Important clues regarding BMP repression mechanism on FGF3 will come from either detailed mapping of these and other regulatory regions of the FGF3 gene as well as from uncovering downstream molecule/s responsible for mediating the BMP signal. Potential function of FGF3 in the hindbrain Previous studies have implicated FGF signaling in hindbrain segmentation. The expression of FGF3 and FGF8 in r4 in the zebrafish hindbrain is essential for Krox20 expression in r5 and Kreisler/Val expression in r5/r6, which are required for segmentation of the posterior hindbrain (Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003). In contrast to zebrafish, FGF8 is not expressed in the chick or mouse hindbrain, while FGF3 has dynamic expression in r2, r4–r6 and hindbrain boundaries (Mahmood et al., 1995, 1996; McKay et al., 1994; Powles et al., 2004; Wilkinson et al., 1988). Although a direct role for FGF3 in patterning the chick hindbrain was not shown yet, different FGFRs and several cytoplasmic FGF signaling components (such as ERK1/2 and Mkp3) have been shown to be expressed in the hindbrain of early chick embryos (Lunn et al., 2007; Walshe and Mason, 2000), suggesting a role for FGF signaling in the hindbrain. Moreover, FGFR activation was demonstrated to be essential for Krox20 and Kr/MafB expression in the chick (Aragon et al., 2005; Marin and Charnay, 2000). However, inactivation of the FGF3 gene in the mouse has revealed roles in inner ear induction (Alvarez et al., 2003; Mansour et al., 1993) but did not affect hindbrain segmentation

(Wright and Mansour, 2003). Although it is currently unclear whether FGF3 also has a role within the mammalian hindbrain, perhaps acting redundantly with other FGFs, the expression pattern of FGF3 does not seem consistent with a role only in inner ear induction, but also in hindbrain organization, as shown in the zebrafish (Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003). Our findings that knockdown of follistatin or overexpression of BMP leads to loss of FGF3 and Krox20 expression, as inhibition of FGFR does, is most simply interpreted as the inhibition of BMPs in specific hindbrain segments being essential for upregulation of FGFs that may include FGF3. Intriguingly, in chick the expression of Krox20 in r3, but not in r5, is dependent upon neighbour interaction of r4 (Graham and Lumsden, 1996), and Krox20 inhibits follistatin expression in r3 (Giudicelli et al., 2001; Seitanidou et al., 1997). This suggests a model in which follistatin in r4 enables expression of FGFs that promote Krox20 expression and consequently downregulate follistatin in r3. Similarly, follistatin may be required for FGF signals that are essential for r5 development. However, unlike the situation in r3, follistatin and FGF3 are expressed at early stages in r5, and follistatin subsequent downregulation from this segment is not dependent on Krox20 expression (Seitanidou et al., 1997). Our results show that excess BMP4 inhibits EphA4 protein expression in r3 and r5. As EphA4 transcription is regulated by Krox20 (Theil et al., 1998), the simplest explanation is that loss of FGF3 due to BMP overexpression results in a lack of Krox20 expression, and consequently a failure to upregulate EphA4. The lack of boundary cells in these embryos is in agreement with studies in zebrafish showing that EphA4 signaling is required for boundary cell formation (Cooke et al., 2005; Xu et al., 1995). Unexpectedly, we found that the expression of Hoxb1 in r4 is reduced upon overexpression of BMP4. In contrast, addition of follistatin did not increase the levels of Hoxb1, which would have been expected were it regulated by BMP-inhibition mediated by follistatin. These results may indicate that ectopic BMP downregulates signals other than FGF3 that are required for Hoxb1 expression, whereas decreased BMP levels (upon follistatin bead application) does not disrupt Hoxb1 as the required signals are still present. Other studies have shown that either repression of FGF signaling (Marin and Charnay, 2000; Maves et al., 2002) or of Krox20 (Schneider-Maunoury et al., 1993) does not alter Hoxb1 expression. Thus, the lack of effect on Hoxb1 by follistatin bead suggests that follistatin activity which regulates FGF3, which in turn regulates Krox20, is not involving Hoxb1. However, another study in zebrafish has found that knockdown of FGFs leads to loss of both Krox20 and Hoxb1 (Walshe et al., 2002), Further studies are required to elucidate how the cross talk between FGF-BMP-follistatin regulate Hox genes. Roles of BMPs in dorsoventral and segmental patterning It is well established that the dorsally-expressed TGFβ family members, including BMPs, have crucial roles in the dorsoventral (DV) specification of cell types in the neural tube (Chizhikov and Millen, 2005; Lee and Jessell, 1999; Timmer et al., 2002). An important question raised from our studies is how the role of follistatin in segmental patterning does not interfere with the involvement of BMPs in DV specification. One possibility is that different TGFβ family members are involved in segmental versus DV patterning and that follistatin specifically inhibits only the former members. Smith and Graham (2001) have suggested that some BMPs are required for hindbrain neural crest formation, while others induce segmental neural crest apoptosis at r3 and r5 (Smith and Graham, 2001). The TGFβ family member Vg1 is one candidate that was shown to be restricted to odd-numbered segments in the chick hindbrain (Shah et al., 1997), suggesting that follistatin may antagonise its activity in a specific AP manner. GDF7, which is

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

expressed along the dorsal neural tube in the trunk and head regions, is another candidate that has been shown to specify dorsal fates in the spinal cord without being inhibited by follistatin (Lee et al., 1998). Our results that follistatin-MO does not alter DV markers further argues against a role of follistatin to affect DV patterning in the hindbrain. However, the finding that overexpression of follistatin in the chick spinal cord causes a dorsal to ventral switch in cell identity (Liem et al., 2000), suggests that BMPs inhibited by follistatin also contributes to DV specification in the trunk. Another possible explanation is that follistatin and BMPs have stage-specific roles in the hindbrain, initially in segmental patterning and later in DV specification. Indeed, BMPs are implicated in the specification of first-order relay sensory neurons that develop in the hindbrain from around 28ss (Hornbruch et al., 2005), after the downregulation of the segmental expression of follistatin. Notably, at late stages follistatin becomes expressed in a longitudinal column in r2–r7 that may correspond to the alar–basal plate border and could antagonise dorsally-expressed BMPs. In agreement with this possibility, somite-derived follistatin is required for DV patterning in the spinal cord by preventing dorsalisation of ventral domains by BMPs (Liem et al., 2000). Our data on the early expression patterns of BMP, follistatin and FGF3 shows that the expression of the two latter genes begins earlier that that of BMP4, suggesting that follistatin may be needed before BMPs in order to ensure the lack of segmental BMP activity on FGF3 when BMP is starting to be expressed in the hindbrain. Notably, FGF3 earliest segmental expression is preceding that of follistatin, suggesting that FGF3 initiation is independent of BMPs. Finally, it is possible that there are some overlaps in roles of follistatin in segmental and DV patterning, in which it both enables FGF3 expression required for segmentation and underlies segmental differences in DV specification of specific neurons. Since very little is known about hindbrain DV specification as compared to the trunk, further analysis of the role and timing of BMP activities in the hindbrain will be required to distinguish between these possibilities. Acknowledgments We thank James Briscoe, Robb Krumlauf, Nobue Itasaki and Patrick Charnay for plasmids, and James Briscoe also for critical reading of the manuscript. This work was supported by the Medical Research Council, UK, The Hebrew University Innovative Research Grants, and the Israel Science Foundation. DSD was a recipient of an EMBO postdoctoral fellowship. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2008.09.005. References Agarwal, P., Wylie, J.N., Galceran, J., Arkhitko, O., Li, C., Deng, C., Grosschedl, R., Bruneau, B.G., 2003. Tbx5 is essential for forelimb bud initiation following patterning of the limb field in the mouse embryo. Development 130, 623–633. Albano, R.M., Arkell, R., Beddington, R.S., Smith, J.C., 1994. Expression of inhibin subunits and follistatin during postimplantation mouse development: decidual expression of activin and expression of follistatin in primitive streak, somites and hindbrain. Development 120, 803–813. Alvarez, Y., Alonso, M.T., Vendrell, V., Zelarayan, L.C., Chamero, P., Theil, T., Bosl, M.R., Kato, S., Maconochie, M., Riethmacher, D., Schimmang, T., 2003. Requirements for FGF3 and FGF10 during inner ear formation. Development 130, 6329–6338. Anderson, R.M., Lawrence, A.R., Stottmann, R.W., Bachiller, D., Klingensmith, J., 2002. Chordin and noggin promote organizing centers of forebrain development in the mouse. Development 129, 4975–4987. Aragon, F., Vazquez-Echeverria, C., Ulloa, E., Reber, M., Cereghini, S., Alsina, B., Giraldez, F., Pujades, C., 2005. vHnf1 regulates specification of caudal rhombomere identity in the chick hindbrain. Dev. Dyn. 234, 567–576. Barrow, J.R., Stadler, H.S., Capecchi, M.R., 2000. Roles of Hoxa1 and Hoxa2 in patterning the early hindbrain of the mouse. Development 127, 933–944.

223

Bel-Vialar, S., Itasaki, N., Krumlauf, R., 2002. Initiating Hox gene expression: in the early chick neural tube differential sensitivity to FGF and RA signaling subdivides the HoxB genes in two distinct groups. Development 129, 5103–5115. Canning, C.A., Lee, L., Irving, C., Mason, I., Jones, C.M., 2007. Sustained interactive Wnt and FGF signaling is required to maintain isthmic identity. Dev. Biol. 305, 276–286. Carpenter, E.M., Goddard, J.M., Chisaka, O., Manley, N.R., Capecchi, M.R., 1993. Loss of Hox-A1 (Hox-1.6) function results in the reorganization of the murine hindbrain. Development 118, 1063–1075. Chizhikov, V.V., Millen, K.J., 2005. Roof plate-dependent patterning of the vertebrate dorsal central nervous system. Dev. Biol. 277, 287–295. Connolly, D.J., Patel, K., Seleiro, E.A., Wilkinson, D.G., Cooke, J., 1995. Cloning, sequencing, and expressional analysis of the chick homologue of follistatin. Dev. Genet. 17, 65–77. Cooke, J., Moens, C., Roth, L., Durbin, L., Shiomi, K., Brennan, C., Kimmel, C., Wilson, S., Holder, N., 2001. Eph signalling functions downstream of Val to regulate cell sorting and boundary formation in the caudal hindbrain. Development 128, 571–580. Cooke, J.E., Kemp, H.A., Moens, C.B., 2005. EphA4 is required for cell adhesion and rhombomere-boundary formation in the zebrafish. Curr. Biol. 15, 536–542. Dale, K., Sattar, N., Heemskerk, J., Clarke, J.D., Placzek, M., Dodd, J., 1999. Differential patterning of ventral midline cells by axial mesoderm is regulated by BMP7 and chordin. Development 126, 397–408. De Robertis, E.M., Kuroda, H., 2004. Dorsal–ventral patterning and neural induction in Xenopus embryos. Annu. Rev. Cell Dev. Biol. 20, 285–308. Delaune, E., Lemaire, P., Kodjabachian, L., 2005. Neural induction in Xenopus requires early FGF signalling in addition to BMP inhibition. Development 132, 299–310. Eisen, J.S., Smith, J.C., 2008. Controlling morpholino experiments: don't stop making antisense. Development 135, 1735–1743. Ellies, D.L., Church, V., Francis-West, P., Lumsden, A., 2000. The WNT antagonist cSFRP2 modulates programmed cell death in the developing hindbrain. Development 127, 5285–5295. Farlie, P.G., Kerr, R., Thomas, P., Symes, T., Minichiello, J., Hearn, C.J., Newgreen, D., 1999. A paraxial exclusion zone creates patterned cranial neural crest cell outgrowth adjacent to rhombomeres 3 and 5. Dev. Biol. 213, 70–84. Foucher, I., Mione, M., Simeone, A., Acampora, D., Bally-Cuif, L., Houart, C., 2006. Differentiation of cerebellar cell identities in absence of Fgf signalling in zebrafish Otx morphants. Development 133, 1891–1900. Fraser, S., Keynes, R., Lumsden, A., 1990. Segmentation in the chick embryo hindbrain is defined by cell lineage restrictions. Nature 344, 431–435. Frohman, M.A., Martin, G.R., Cordes, S.P., Halamek, L.P., Barsh, G.S., 1993. Altered rhombomere-specific gene expression and hyoid bone differentiation in the mouse segmentation mutant, kreisler (kr). Development 117, 925–936. Fukuchi-Shimogori, T., Grove, E.A., 2003. Emx2 patterns the neocortex by regulating FGF positional signaling. Nat. Neurosci. 6, 825–831. Furthauer, M., Van Celst, J., Thisse, C., Thisse, B., 2004. Fgf signalling controls the dorsoventral patterning of the zebrafish embryo. Development 131, 2853–2864. Gavalas, A., Ruhrberg, C., Livet, J., Henderson, C.E., Krumlauf, R., 2003. Neuronal defects in the hindbrain of Hoxa1, Hoxb1 and Hoxb2 mutants reflect regulatory interactions among these Hox genes. Development 130, 5663–5679. Giudicelli, F., Taillebourg, E., Charnay, P., Gilardi-Hebenstreit, P., 2001. Krox-20 patterns the hindbrain through both cell-autonomous and non cell-autonomous mechanisms. Genes Dev. 15, 567–580. Giudicelli, F., Gilardi-Hebenstreit, P., Mechta-Grigoriou, F., Poquet, C., Charnay, P., 2003. Novel activities of Mafb underlie its dual role in hindbrain segmentation and regional specification. Dev. Biol. 253, 150–162. Graham, A., Lumsden, A., 1996. Interactions between rhombomeres modulate Krox-20 and follistatin expression in the chick embryo hindbrain. Development 122, 473–480. Graham, A., Francis-West, P., Brickell, P., Lumsden, A., 1994. The signalling molecule BMP4 mediates apoptosis in the rhombencephalic neural crest. Nature 372, 684–686. Graham, A., Koentges, G., Lumsden, A., 1996. Neural crest apoptosis and the establishment of craniofacial pattern: an honorable death. Mol. Cell. Neurosci. 8, 76–83. Gunhaga, L., Marklund, M., Sjodal, M., Hsieh, J.C., Jessell, T.M., Edlund, T., 2003. Specification of dorsal telencephalic character by sequential Wnt and FGF signaling. Nat. Neurosci. 6, 701–707. Heyman, I., Faissner, A., Lumsden, A., 1995. Cell and matrix specialisations of rhombomere boundaries. Dev. Dyn. 204, 301–315. Hollyday, M., McMahon, J.A., McMahon, A.P., 1995. Wnt expression patterns in chick embryo nervous system. Mech. Dev. 52, 9–25. Hornbruch, A., Ma, G., Ballermann, M., Tumova, K., liu, D., Logan, C., 2005. A BMPmediated transcriptional cascade involving Cash1 and Tlx-3 specifices first-order relay neurons in the developing hindbrain. Mech. Dev. 122, 900–913. Iemura, S., Yamamoto, T.S., Takagi, C., Uchiyama, H., Natsume, T., Shimasaki, S., Sugino, H., Ueno, N., 1998. Direct binding of follistatin to a complex of bone-morphogenetic protein and its receptor inhibits ventral and epidermal cell fates in early Xenopus embryo. Proc. Natl. Acad. Sci. U. S. A. 95, 9337–9342. Irving, C., Nieto, M.A., DasGupta, R., Charnay, P., Wilkinson, D.G., 1996. Progressive spatial restriction of Sek-1 and Krox-20 gene expression during hindbrain segmentation. Dev. Biol. 173, 26–38. Itasaki, N., Bel-Vialar, S., Krumlauf, R., 1999. ‘Shocking’ developments in chick embryology: electroporation and in ovo gene expression. Nat. Cell Biol. 1, E203–207. Jackman, W.R., Draper, B.W., Stock, D.W., 2004. Fgf signaling is required for zebrafish tooth development. Dev. Biol. 274, 139–157. Jukkola, T., Lahti, L., Naserke, T., Wurst, W., Partanen, J., 2006. FGF regulated geneexpression and neuronal differentiation in the developing midbrain–hindbrain region. Dev. Biol. 297, 141–157.

224

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225

Kettunen, P., Laurikkala, J., Itaranta, P., Vainio, S., Itoh, N., Thesleff, I., 2000. Associations of FGF-3 and FGF-10 with signaling networks regulating tooth morphogenesis. Dev. Dyn. 219, 322–332. Keynes, R., Cook, G., Davies, J., Lumsden, A., Norris, W., Stern, C., 1990. Segmentation and the development of the vertebrate nervous system. J. Physiol. (Paris) 84, 27–32. Krumlauf, R., Marshall, H., Studer, M., Nonchev, S., Sham, M.H., Lumsden, A., 1993. Hox homeobox genes and regionalisation of the nervous system. J. Neurobiol. 24, 1328–1340. Lee, K.J., Jessell, T.M., 1999. The specification of dorsal cell fates in the vertebrate central nervous system. Annu. Rev. Neurosci. 22, 261–294. Lee, K.J., Mendelsohn, M., Jessell, T.M., 1998. Neuronal patterning by BMPs: a requirement for GDF7 in the generation of a discrete class of commissural interneurons in the mouse spinal cord. Genes Dev. 12, 3394–3407. Leger, S., Brand, M., 2002. Fgf8 and Fgf3 are required for zebrafish ear placode induction, maintenance and inner ear patterning. Mech. Dev. 119, 91–108. Liem Jr., K.F., Tremml, G., Jessell, T.M., 1997. A role for the roof plate and its resident TGFbeta-related proteins in neuronal patterning in the dorsal spinal cord. Cell 91, 127–138. Liem Jr., K.F., Jessell, T.M., Briscoe, J., 2000. Regulation of the neural patterning activity of sonic hedgehog by secreted BMP inhibitors expressed by notochord and somites. Development 127, 4855–4866. Lumsden, A., 2004. Segmentation and compartition in the early avian hindbrain. Mech. Dev. 121, 1081–1088. Lumsden, A., Krumlauf, R., 1996. Patterning the vertebrate neuraxis. Science 274, 1109–1115. Lunn, J.S., Fishwick, K.J., Halley, P.A., Storey, K.G., 2007. A spatial and temporal map of FGF/Erk1/2 activity and response repertoires in the early chick embryo. Dev. Biol. 302, 536–552. Mahmood, R., Kiefer, P., Guthrie, S., Dickson, C., Mason, I., 1995. Multiple roles for FGF3 during cranial neural development in the chicken. Development 121, 1399–1410. Mahmood, R., Mason, I.J., Morriss-Kay, G.M., 1996. Expression of Fgf-3 in relation to hindbrain segmentation, otic pit position and pharyngeal arch morphology in normal and retinoic acid-exposed mouse embryos. Anat. Embryol. (Berl) 194, 13–22. Mandler, M., Neubuser, A., 2004. FGF signaling is required for initiation of feather placode development. Development 131, 3333–3343. Mansour, S.L., Goddard, J.M., Capecchi, M.R., 1993. Mice homozygous for a targeted disruption of the proto-oncogene int-2 have developmental defects in the tail and inner ear. Development 117, 13–28. Manzanares, M., Trainor, P.A., Nonchev, S., Ariza-McNaughton, L., Brodie, J., Gould, A., Marshall, H., Morrison, A., Kwan, C.T., Sham, M.H., Wilkinson, D.G., Krumlauf, R., 1999. The role of kreisler in segmentation during hindbrain development. Dev. Biol. 211, 220–237. Marin, F., Charnay, P., 2000. Hindbrain patterning: FGFs regulate Krox20 and mafB/kr expression in the otic/preotic region. Development 127, 4925–4935. Mark, M., Lufkin, T., Vonesch, J.L., Ruberte, E., Olivo, J.C., Dolle, P., Gorry, P., Lumsden, A., Chambon, P., 1993. Two rhombomeres are altered in Hoxa-1 mutant mice. Development 119, 319–338. Maroon, H., Walshe, J., Mahmood, R., Kiefer, P., Dickson, C., Mason, I., 2002. Fgf3 and Fgf8 are required together for formation of the otic placode and vesicle. Development 129, 2099–2108. Mason, I., Chambers, D., Shamim, H., Walshe, J., Irving, C., 2000. Regulation and function of FGF8 in patterning of midbrain and anterior hindbrain. Biochem. Cell. Biol. 78, 577–584. Maves, L., Jackman, W., Kimmel, C.B., 2002. FGF3 and FGF8 mediate a rhombomere 4 signaling activity in the zebrafish hindbrain. Development 129, 3825–3837. McKay, I.J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A., Lewis, J., 1994. The kreisler mouse: a hindbrain segmentation mutant that lacks two rhombomeres. Development 120, 2199–2211. McKay, I.J., Lewis, J., Lumsden, A., 1996. The role of FGF-3 in early inner ear development: an analysis in normal and kreisler mutant mice. Dev. Biol. 174, 370–378. Mellitzer, G., Xu, Q., Wilkinson, D.G., 1999. Eph receptors and ephrins restrict cell intermingling and communication. Nature 400, 77–81. Moens, C.B., Prince, V.E., 2002. Constructing the hindbrain: insights from the zebrafish. Dev. Dyn. 224, 1–17. Moens, C.B., Cordes, S.P., Giorgianni, M.W., Barsh, G.S., Kimmel, C.B., 1998. Equivalence in the genetic control of hindbrain segmentation in fish and mouse. Development 125, 381–391. Murakami, A., Grinberg, D., Thurlow, J., Dickson, C., 1993. Identification of positive and negative regulatory elements involved in the retinoic acid/cAMP induction of Fgf-3 transcription in F9 cells. Nucleic Acids Res. 21, 5351–5359. Murakami, A., Thurlow, J., Dickson, C., 1999. Retinoic acid-regulated expression of fibroblast growth factor 3 requires the interaction between a novel transcription factor and GATA-4. J. Biol. Chem. 274, 17242–17248. Murakami, A., Ishida, S., Thurlow, J., Revest, J.M., Dickson, C., 2001. SOX6 binds CtBP2 to repress transcription from the Fgf-3 promoter. Nucleic Acids Res. 29, 3347–3355. Murakami, A., Shen, H., Ishida, S., Dickson, C., 2004. SOX7 and GATA-4 are competitive activators of Fgf-3 transcription. J. Biol. Chem. 279, 28564–28573. Nakamura, T., Takio, K., Eto, Y., Shibai, H., Titani, K., Sugino, H., 1990. Activin-binding protein from rat ovary is follistatin. Science 247, 836–838. Oh, S.H., Johnson, R., Wu, D.K., 1996. Differential expression of bone morphogenetic proteins in the developing vestibular and auditory sensory organs. J. Neurosci. 16, 6463–6475.

Ohkubo, Y., Chiang, C., Rubenstein, J.L., 2002. Coordinate regulation and synergistic actions of BMP4, SHH and FGF8 in the rostral prosencephalon regulate morphogenesis of the telencephalic and optic vesicles. Neuroscience 111, 1–17. Pasqualetti, M., Neun, R., Davenne, M., Rijli, F.M., 2001. Retinoic acid rescues inner ear defects in Hoxa1 deficient mice. Nat. Genet. 29, 34–39. Phillips, B.T., Bolding, K., Riley, B.B., 2001. Zebrafish fgf3 and fgf8 encode redundant functions required for otic placode induction. Dev. Biol. 235, 351–365. Powles, N., Marshall, H., Economou, A., Chiang, C., Murakami, A., Dickson, C., Krumlauf, R., Maconochie, M., 2004. Regulatory analysis of the mouse Fgf3 gene: control of embryonic expression patterns and dependence upon sonic hedgehog (Shh) signalling. Dev. Dyn. 230, 44–56. Prince, V.E., Moens, C.B., Kimmel, C.B., Ho, R.K., 1998. Zebrafish hox genes: expression in the hindbrain region of wild-type and mutants of the segmentation gene, valentino. Development 125, 393–406. Roy, N.M., Sagerstrom, C.G., 2004. An early Fgf signal required for gene expression in the zebrafish hindbrain primordium. Brain Res. Dev. Brain Res. 148, 27–42. Schneider-Maunoury, S., Topilko, P., Seitandou, T., Levi, G., Cohen-Tannoudji, M., Pournin, S., Babinet, C., Charnay, P., 1993. Disruption of Krox-20 results in alteration of rhombomeres 3 and 5 in the developing hindbrain. Cell 75, 1199–1214. Schneider-Maunoury, S., Seitanidou, T., Charnay, P., Lumsden, A., 1997. Segmental and neuronal architecture of the hindbrain of Krox-20 mouse mutants. Development 124, 1215–1226. Seitanidou, T., Schneider-Maunoury, S., Desmarquet, C., Wilkinson, D.G., Charnay, P., 1997. Krox-20 is a key regulator of rhombomere-specific gene expression in the developing hindbrain. Mech. Dev. 65, 31–42. Sekine, K., Ohuchi, H., Fujiwara, M., Yamasaki, M., Yoshizawa, T., Sato, T., Yagishita, N., Matsui, D., Koga, Y., Itoh, N., Kato, S., 1999. Fgf10 is essential for limb and lung formation. Nat. Genet. 21, 138–141. Sela-Donenfeld, D., Kalcheim, C., 1999. Regulation of the onset of neural crest migration by coordinated activity of BMP4 and Noggin in the dorsal neural tube. Development 126, 4749–4762. Shah, S.B., Skromne, I., Hume, C.R., Kessler, D.S., Lee, K.J., Stern, C.D., Dodd, J., 1997. Misexpression of chick Vg1 in the marginal zone induces primitive streak formation. Development 124, 5127–5138. Shimamura, K., Martinez, S., Puelles, L., Rubenstein, J.L., 1997. Patterns of gene expression in the neural plate and neural tube subdivide the embryonic forebrain into transverse and longitudinal domains. Dev. Neurosci. 19, 88–96. Shimogori, T., Banuchi, V., Ng, H.Y., Strauss, J.B., Grove, E.A., 2004. Embryonic signaling centers expressing BMP, WNT and FGF proteins interact to pattern the cerebral cortex. Development 131, 5639–5647. Smith, A., Graham, A., 2001. Restricting Bmp-4 mediated apoptosis in hindbrain neural crest. Dev. Dyn. 220, 276–283. Stern, C.D., 2002. Induction and initial patterning of the nervous system — the chick embryo enters the scene. Curr. Opin. Genet. Dev. 12, 447–451. Theil, T., Frain, M., Gilardi-Hebenstreit, P., Flenniken, A., Charnay, P., Wilkinson, D.G., 1998. Segmental expression of the EphA4 (Sek-1) receptor tyrosine kinase in the hindbrain is under direct transcriptional control of Krox-20. Development 125, 443–452. Timmer, J.R., Wang, C., Niswander, L., 2002. BMP signaling patterns the dorsal and intermediate neural tube via regulation of homeobox and helix–loop–helix transcription factors. Development 129, 2459–2472. Timmer, J., Chesnutt, C., Niswander, L., 2005. The activin signaling pathway promotes differentiation of dI3 interneurons in the spinal neural tube. Dev. Biol. 285, 1–10. Trainor, P.A., Krumlauf, R., 2000. Patterning the cranial neural crest: hindbrain segmentation and Hox gene plasticity. Nat. Rev. Neurosci. 1, 116–124. Trevarrow, B., Marks, D.L., Kimmel, C.B., 1990. Organization of hindbrain segments in the zebrafish embryo. Neuron 4, 669–679. Trokovic, R., Trokovic, N., Hernesniemi, S., Pirvola, U., Vogt Weisenhorn, D.M., Rossant, J., McMahon, A.P., Wurst, W., Partanen, J., 2003. FGFR1 is independently required in both developing mid- and hindbrain for sustained response to isthmic signals. EMBO J. 22, 1811–1823. Vogel-Hopker, A., Rohrer, H., 2002. The specification of noradrenergic locus coeruleus (LC) neurones depends on bone morphogenetic proteins (BMPs). Development 129, 983–991. Voiculescu, O., Taillebourg, E., Pujades, C., Kress, C., Buart, S., Charnay, P., SchneiderMaunoury, S., 2001. Hindbrain patterning: Krox20 couples segmentation and specification of regional identity. Development 128, 4967–4978. Walshe, J., Mason, I., 2000. Expression of FGFR1, FGFR2 and FGFR3 during early neural development in the chick embryo. Mech. Dev. 90, 103–110. Walshe, J., Maroon, H., McGonnell, I.M., Dickson, C., Mason, I., 2002. Establishment of hindbrain segmental identity requires signaling by FGF3 and FGF8. Curr. Biol. 12, 1117–1123. Waskiewicz, A.J., Rikhof, H.A., Moens, C.B., 2002. Eliminating zebrafish pbx proteins reveals a hindbrain ground state. Dev. Cell. 3, 723–733. Wiellette, E.L., Sive, H., 2003. vhnf1 and Fgf signals synergize to specify rhombomere identity in the zebrafish hindbrain. Development 130, 3821–3829. Wiellette, E.L., Sive, H., 2004. Early requirement for fgf8 function during hindbrain pattern formation in zebrafish. Dev. Dyn. 229, 393–399. Wilkinson, D.G., Peters, G., Dickson, C., McMahon, A.P., 1988. Expression of the FGFrelated proto-oncogene int-2 during gastrulation and neurulation in the mouse. EMBO J. 7, 691–695. Wright, T.J., Mansour, S.L., 2003. Fgf3 and Fgf10 are required for mouse otic placode induction. Development 130, 3379–3390. Xu, Q., Alldus, G., Holder, N., Wilkinson, D.G., 1995. Expression of truncated Sek-1 receptor tyrosine kinase disrupts the segmental restriction of gene expression in the Xenopus and zebrafish hindbrain. Development 121, 4005–4016.

K. Weisinger et al. / Developmental Biology 324 (2008) 213–225 Xu, Q., Mellitzer, G., Robinson, V., Wilkinson, D.G., 1999a. In vivo cell sorting in complementary segmental domains mediated by Eph receptors and ephrins. Nature 399, 267–271. Xu, R.H., Ault, K.T., Kim, J., Park, M.J., Hwang, Y.S., Peng, Y., Sredni, D., Kung, H., 1999b. Opposite effects of FGF and BMP-4 on embryonic blood formation: roles of PV.1 and GATA-2. Dev. Biol. 208, 352–361.

225

Yamashita, H., ten Dijke, P., Huylebroeck, D., Sampath, T.K., Andries, M., Smith, J.C., Heldin, C.H., Miyazono, K., 1995. Osteogenic protein-1 binds to activin type II receptors and induces certain activin-like effects. J. Cell Biol. 130, 217–226. Zelarayan, L.C., Vendrell, V., Alvarez, Y., Dominguez-Frutos, E., Theil, T., Alonso, M.T., Maconochie, M., Schimmang, T., 2007. Differential requirements for FGF3, FGF8 and FGF10 during inner ear development. Dev. Biol. 308, 379–391.