The inner ear, which mediates the sensations of hearing and equilibrium, develops from an ectodermal placode that is located adjacent to the developing hindbrain. Placodal ectoderm becomes committed to an otic fate after receiving inductive signals from the adjacent neurectoderm and underlying mesenchyme. In mouse embryos, the first morphologic evidence of otic placode induction occurs at the 7- to 8-somite stages (embryonic day [E] 8.5), when the committed ectoderm lateral to rhombomeres (r) 5 and 6 thickens. Subsequently, during the 13- to 20-somite stages (E9.0), the placode invaginates to form a cup. During this stage, the otic epithelium delaminates neuroblasts that will ultimately aggregate and differentiate to form the neurons of the eighth (cochleovestibular) ganglion. By the time the embryo has 21–29 somites (E9.5), the cup closes to form a roughly spherical vesicle known as the otocyst or otic vesicle. At E10.5, morphogenesis of the vesicle initiates with the dorsomedial protrusion of the precursor of the endolymphatic duct. During subsequent days, the vesicle undergoes a very complex morphogenesis to elaborate the auditory and vestibular compartments and also embarks on the process of differentiating the numerous sensory and nonsensory cell types found in the mature inner ear (Torres and Giraldez, 1998; Groves and Bronner-Fraser, 2000; Baker and Bronner-Fraser, 2001; Kiernan et al., 2002; Groves, 2003). Several families of intercellular signalling molecules, including the fibroblast growth factor (FGF) family are involved in induction, morphogenesis, and differentiation of the otic epithelium (Wright and Mansour, 2003b).
The Fgfs comprise a 22-member gene family encoding secreted, heparan sulfate-binding proteins that signal through a family of tyrosine kinase receptors, the fibroblast growth factor receptors (FGFRs), which are encoded by four genes. Fgfr1, -2, and -3 mRNAs are alternatively spliced within the region encoding the third extracellular immunoglobulin-like domain, resulting in the production of IIIb and IIIc isoforms of these receptors. Fgfr4 produces only the IIIc-type isoform. The alternative splicing is regulated in a tissue-specific manner and affects ligand binding specificity and affinity. FGF signalling plays numerous roles during development, including regulation of cell proliferation, migration, differentiation, and survival (Ornitz and Itoh, 2001).
Gain-of-function and loss-of-function studies implicate four members of the Fgf family, Fgf3, Fgf8, Fgf10, and Fgf19, in a variety of stages of otic development in different species (Mansour et al., 1993; Ladher et al., 2000; Vendrell et al., 2000; Phillips et al., 2001; Leger and Brand, 2002; Maroon et al., 2002; Pauley et al., 2003; Wright and Mansour, 2003a). There is also genetic evidence for transmission of FGF signals needed for several phases of otic development through FGFR1, FGFR3 (isoforms not established), and FGFR2b (Deng et al., 1994; Yamaguchi et al., 1994; Partanen et al., 1998; Pirvola et al., 2000; Yu et al., 2000; Eswarakumar et al., 2002; Mueller et al., 2002). In addition, expression profiles have been generated for some Fgfr family members during otic development (Orr-Urtreger et al., 1991, 1993; Peters et al., 1992, 1993; Yamaguchi et al., 1992; Pirvola et al., 2000; Pickles, 2001; Wright and Mansour, 2003a). Although these data suggest roles for Fgfr1, Fgfr2, and Fgfr3 during otic vesicle morphogenesis and differentiation, little is known about the function of the Fgfr isoforms during the early phases of otic induction and vesicle formation.
To determine whether there are other members of the mouse Fgf and Fgfr families that could participate in early otic development, we evaluated the expression patterns of 18 members of the Fgf family, as well as three members of the Fgfr family, at four time points during early otic development, spanning the preplacode to vesicle stages. Two members of the Fgf family, Fgf4 and Fgf16, and three members of the Fgfr family, Fgfr2c, Fgfr3c, and Fgfr4, were expressed in tissues relevant to inner ear development. Fgf4 was expressed in the preplacodal and placodal ectoderm and Fgf16 was expressed asymmetrically in the otic cup and vesicle. Fgfr2c was expressed in the neurectoderm that lies medial to the otic region throughout early otic development. Fgfr4 was expressed in a similar pattern but transcripts could be detected only during preplacodal and placodal stages. Temporally, the expression of Fgfr3c overlapped with Fgfr2c, but Fgfr3c was localised to the otic cup and vesicle and was not found in the neurectoderm. These patterns of gene expression suggest that there could be additional roles for FGF signalling during the early phases of otic development.
Expression Analysis of Fgf and Fgfr Family Members During Early Otic Development
To determine whether any Fgfs and Fgfrs in addition to those previously determined could play early roles in otic development in the mouse, we examined the normal expression patterns of 18 members of the Fgf family (Fgfs 1, 2, 4, 5, 6, 7, 9, 11, 12, 13, 14, 16, 17, 18, 20, 21, 22, and 23) and 3 members of the Fgfr family (Fgfr2c, 3c, and 4) by whole-mount in situ hybridization at four stages of development. These stages encompassed the early events in otic development that culminate in formation of the otic vesicle and included preplacodal ectoderm (E8.0, 4–6 somites), otic placode (E 8.5, 7–10 somites), otic cup (E9.0, 11–19 somites), and otic vesicle (E9.5, >20 somites). Of the 18 Fgfs examined, two family members, Fgf4 and Fgf16, were expressed in patterns relevant to the early development of the ear. We also detected expression of Fgfr2c, Fgfr3c, and Fgfr4.
Fgf4 Is Expressed in Preplacodal and Placodal Otic Ectoderm
As expected, Fgf4 transcripts were detected in the remnants of the primitive streak at the four-somite stage. In addition, Fgf4 was also expressed in a patch of surface ectoderm in these embryos (Fig. 1A,B; Niswander and Martin, 1992; Drucker and Goldfarb, 1993). To determine whether this region of the ectoderm was likely to encompass the presumptive otic placode, we compared its location with that of MafB (kreisler) transcripts by simultaneous hybridization of Fgf4 and MafB probes. MafB is expressed in r5 and r6 (Cordes and Barsh, 1994) and the otic placode forms lateral to this region (Groves and Bronner-Fraser, 2000; Fig. 1C). Observation of whole-mount embryos and coronal sections taken through eight-somite embryos hybridised with both Fgf4 and MafB probes showed that both genes had the same anterior and posterior expression boundaries, suggesting that Fgf4 is expressed in preplacodal otic ectoderm (Fig. 1D). This observation was confirmed by hybridizing an eight-somite embryo with Fgf4 (Fig. 1E), followed by staining with an antibody directed against PAX2, which is expressed in the otic placode. Colocalisation of Fgf4 and PAX2 in the ectoderm confirmed that Fgf4 is expressed in the otic placode (Fig. 1F). In embryos with nine somites, Fgf4 transcripts were still detected in the otic placode and were also detected in the pharyngeal endoderm (Fig. 1G,H). Fgf4 transcripts were not detected in otic tissue at the 11-somite and older stages but could be seen in the pharyngeal endoderm and tail bud as previously described (Niswander and Martin, 1992).
Fgf16 Is Expressed Asymmetrically in the Otic Cup and Vesicle
Otic expression of Fgf16 was initially detected very weakly throughout the thickened placode at the 10-somite stage, just before invagination of the placode (Fig. 1I,J). By 13 somites, Fgf16 transcripts were more strongly expressed and localised to the posterior region of the invaginating otic cup (Fig. 1K,L). At this stage, Fgf16 transcripts were also detected in the pharyngeal endoderm, branchial arches, and the olfactory placode (Fig. 1K). At 19 somites, Fgf16 transcripts were still detected in the posterior otic cup but they were no longer found in the branchial arches and pharyngeal endoderm (Fig. 1M,N). By 24 somites, the otic cup had closed to form an otic vesicle and Fgf16 continued to be expressed in the dorsolateral wall of the posterior half of the vesicle (Fig. 1O,P).
Expression of Fgfr2c, Fgfr3c, and Fgfr4 During Early Otic Development
If Fgf4 and Fgf16 participate in otic development, genes encoding the appropriate receptors should be expressed in the same or in adjacent tissues. FGF4 activates mitogenic signalling through all of the IIIc isoforms of the FGF receptors (Ornitz et al., 1996). Directly comparable studies are not available for FGF16, but this ligand activates proliferation of embryonic brown adipocytes, which express Fgfr1c, Fgfr2c, and Fgfr4 and it binds in vitro to FGFR4 but not to FGFR1c or FGFR2c, suggesting that FGFR4 can act as a receptor for FGF16 (Konishi et al., 2000). Previous studies have shown that at preplacodal and placodal stages, Fgfr1 transcripts (IIIb and IIIc isoforms not distinguished) are present in the neurectoderm and at E9.5 Fgfr1 is localised to the hindbrain and to the mesenchyme surrounding the otic vesicle (Yamaguchi et al., 1992; Pirvola et al., 2002; Wright and Mansour, 2003a). We, therefore, examined expression of Fgfr2c, Fgfr3c, and Fgfr4 in preplacodal to otic vesicle stages.
Fgfr2c Is Expressed in the Neurectoderm Throughout Early Otic Development
Fgfr2c transcripts were not found in the otic tissue itself, but were present in the adjacent neurectoderm medial to the preplacodal otic ectoderm at three somites (Fig. 2A,B). This expression persisted through otic placode, cup, and vesicle stages (Fig. 2C–H). By E9.5, Fgfr2c transcripts were also detected in the branchial arch mesenchyme, the forelimb bud, and nephrogenic cords as previously observed (Fig. 2G,H; Orr-Urtreger et al., 1993).
Fgfr3c Is Expressed in the Otic Cup and Vesicle
Expression of Fgfr3c was not detected in otic tissue until the cup stage. At 19 somites, Fgfr3c transcripts were detected throughout the otic cup (Fig. 2I,J). Expression persisted throughout the otic epithelium as the cup closed to form a vesicle, as seen in a 21-somite embryo (Fig. 2K,L). Fgfr3c expression was also detected in the tail bud, first branchial arch, and optic cup (Fig. 2I,K).
Fgfr4 Is Expressed in the Neurectoderm During Preplacodal and Placodal Otic Development
Just before somite formation, Fgfr4 was detected in the neurectoderm, extraembryonic endoderm, and caudal lateral plate (Fig. 2M,N). Expression in the neurectoderm persisted until the otic placode stage at eight somites (Fig. 2O,P). By 16 somites, when the otic cup had invaginated, expression of Fgfr4 in the neurectoderm was no longer detected (Fig. 2Q,R). At this stage, Fgfr4 transcripts were seen in the rostral somites and hindgut as previously described (Fig. 2Q; Stark et al., 1991).
The identification of five genes, two encoding FGF ligands and three encoding FGF receptors, which were not known previously to be expressed in tissues that are relevant to early otic development, suggests that there may be additional roles for FGF signalling at these stages. Fgf4 is expressed in the preplacodal and placodal otic ectoderm, coinciding with the PAX2 expression domain. FGF4 signals through the IIIc isoforms of the FGF receptors (Ornitz et al., 1996). We and others have demonstrated Fgfr1, Fgfr2c, and Fgfr4 expression in the neurectoderm during preplacodal and placodal stages (Stark et al., 1991; Yamaguchi et al., 1992; Wright and Mansour, 2003a). Therefore, one possible role of FGF4 may be to signal in a paracrine manner to the neurectoderm. A second possibility is that one of the IIIc receptor isoforms is expressed at low levels in the mesenchyme underlying the placode or in the placode itself and that in situ hybridisation is not sensitive enough to detect these levels. If this is the case, then FGF4 could signal in a paracrine manner to the mesenchyme or in an autocrine manner to the placode.
If Fgf4 is signalling to the hindbrain (or even to the mesenchyme) through any of its receptors, how might this affect otic development? The hindbrain and mesenchyme are sources of otic-inducing signals, including FGF3 and FGF10, respectively (Wright and Mansour, 2003a), so an FGF4 signal from the placode could be used to activate or maintain expression of otic-inducing factors. Similarly, if FGF4 signals in an autocrine manner to the placode, it may regulate or maintain the placodal response to otic induction. As Fgf4 null mutants are peri-implantation lethal (Feldman et al., 1995), conditional loss-of-function studies will be needed to address these issues.
Fgf16 is initially expressed throughout the invaginating otic placode before becoming localised to the posterior otic cup. Subsequently, Fgf16 transcripts are detected in the dorsolateral wall of the otic vesicle. Receptor binding studies have shown that FGFR4 but not FGFR1c or FGFR2c can serve as a receptor for FGF16 (Konishi et al., 2000). The potential for binding of FGF16 to FGFR3c was not tested in this study; however, given that FGF9, which is one of the FGFs most closely related to FGF16, activates FGFR3c, it would not be surprising if the same were true for FGF16 (Ornitz et al., 1996; Kim, 2001). Our expression analysis, moreover, shows that Fgfr3c is localized appropriately to serve as an FGF16 receptor. Fgfr3c is expressed throughout the otic cup and vesicle; therefore, Fgf16 could signal through Fgfr3c in either an autocrine or paracrine manner.
The expression pattern of Fgf16 suggests possible roles in otic cell fate decisions and/or axis formation. Fate maps generated by injecting chick otic cup with vital dyes suggest that there is an anterior to posterior lineage boundary dividing the endolymphatic duct that is in place by the otic cup stage. Thus, cells in the dorsal posterior portion of the cup give rise to the posterior part of the endolymphatic duct. Cells from more ventral regions of the posterior cup can adopt a variety of generally posterior fates (Brigande et al., 2000). Similar studies in the frog, however, do not support the existence of the developmental boundaries seen in the chick (Kil and Collazo, 2002). Fate maps of the mouse otocyst lineages have not been produced. As Fgf16 is one of the earliest regionally restricted transcripts identified in otic tissue to date and is the only transcript known so far to be polarized along the anteroposterior axis of the otic cup, it should be possible to exploit this finding to determine the fate of mouse posterior otic cup cells. Furthermore, targeted mutagenesis of Fgf16 will show whether it plays a role in specifying particular otic cell fates. Another possible role for Fgf16 could be to participate in otic axis formation. The anteroposterior otic axis is established before the dorsoventral axis, and in chick, this development occurs at the otic cup stage (Wu et al., 1998). Equivalent information is not available for the mouse. It will be very interesting, therefore, to observe the orientation of otic structures in Fgf16 null mutants.
In Situ Hybridisation
Expression patterns of Fgf and Fgfr genes were determined by using embryos generated by mating wild type CD-1 mice (Charles River Laboratories). Embryos were isolated on the indicated days after detection of a vaginal plug. Digoxigenin-labelled RNA probes were prepared, hybridised to the embryos, and detected as described (Henrique et al., 1995). cDNAs used to prepare probes for Fgfr2IgIIIc (accession no. NM010207, bp1599-1737; Kettunen et al., 1998), Fgfr3IgIIIc (accession no. L26492, bp1998-2125; Kettunen et al., 1998), Fgf4 (probe contained exon 3 and 3′UTR sequences, accession nos. NM010202, U43515; Moon et al., 2000), and Fgf16 (accession no. NM030614; Miyake et al., 1998) have been described previously. The Fgfr4 probe contained 700 bp from the 3′ untranslated region (mFgfr4, accession no. XM193720, bp2697-3291). An antisense probe was generated by digestion with XhoI, followed by transcription with T7 RNA polymerase.
After whole-mount in situ hybridisation with the Fgf4 antisense probe, embryos were fixed in 4% paraformaldehyde/phosphate buffered saline. For whole-mount immunohistochemical detection of PAX2, polyclonal antibodies (Covance), kindly provided by Drs. D. Wellik and M. Capecchi, were used at a 1:62 dilution. The protocol for fixation of the embryos and application and detection of the primary antibodies with horseradish peroxidase–labelled secondary antibodies and diaminobenzidine staining was carried out as described previously for phosphorylated histone H3 antibodies (Gavalas et al., 2001).
Embryos stained for analysis of gene expression were cryoprotected in sucrose and sectioned at 14 μm by using a Leica cryostat as described (Stark et al., 2000).
Whole embryos were photographed by using a Zeiss SV-11 dissecting microscope fitted with a digital camera (Kodak MDS120 or 240). Sections were photographed by using a Zeiss Axioscop fitted with DIC optics and a digital camera (AxioCam).
We thank Dr. Raj Ladher for help- ful discussions and prepublication data. cDNA clones were generously shared by Dr. P Kettunen (Fgfr2IgIIIc, Fgfr3IgIIIc), Dr. N. Itoh (Fgf16), and Dr. A. Moon (Fgf4). Drs. Wellik and Capecchi kindly provided the PAX2 antibodies. T.J.W. received funding from the Deafness Research Foundation, T.J.W., G.C.S, and S.L.M. were funded by the NIDCD, and H.K. and P.K. were funded by the Turkish Neurological Society.