SEARCH

SEARCH BY CITATION

Keywords:

  • Fgf10;
  • enhancer analysis;
  • cis-element;
  • mouse genome;
  • chicken genome;
  • transgenic mouse;
  • inner ear development, semicircular canals;
  • Fgf10 knockout mouse

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES

Fibroblast growth factor (FGF) signaling is crucial for the induction and growth of the ear, a sensory organ that involves intimate tissue interactions. Here, we report the abnormality of Fgf10 null ear and the identification of a cis-regulatory element directing otic expression of Fgf10. In Fgf10 null inner ears, we found that the initial development of semicircular, vestibular, and cochlear divisions is roughly normal, after which there are abnormalities of semicircular canal/cristae and vestibular development. The mutant semicircular disks remain without canal formation by the perinatal stage. To elucidate regulation of the Fgf10 expression during inner ear development, we isolated a 6.6-kb fragment of its 5′-upstream region and examined its transcriptional activity with transgenic mice, using a lacZ-reporter system. From comparison of the mouse sequences of the 6.6-kb fragment with corresponding sequences of the human and chicken Fgf10, we identified a 0.4-kb enhancer sequence that drives Fgf10 expression in the developing inner ear. The enhancer sequences have motifs for many homeodomain-containing proteins (e.g., Prx, Hox, Nkx), in addition to POU-domain factors (e.g., Brn3), zinc-finger transcription factors (e.g., GATA-binding factors), TCF/LEF-1, and a SMAD-interacting protein. Thus, FGF10 signaling is dispensable for specification of otic compartment identity but is required for hollowing the semicircular disk. Furthermore, the analysis of a putative inner ear enhancer of Fgf10 has disclosed a complicated regulation of Fgf10 during inner ear development by numerous transcription factors and signaling pathways. Developmental Dynamics 233:177–187, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES

The vertebrate inner ear forms a highly complex sensory structure responsible for the detection of sound and balance. During embryonic development, the inner ear arises from a simple epithelium adjacent to the hindbrain, the otic placode, which is specified through inductive interactions with surrounding tissues. Embryological evidence shows that the induction of the otic placode is a multistep process, which requires sequential interaction of different tissues, the adjacent neuroectoderm, and underlying mesoderm with the future otic ectoderm. Recent progress has been made to identify some of the molecular players involved in the developmental processes of the inner ear.

Owing to their gene expression patterns and various experimental manipulations, members of the fibroblast growth factor (FGF) gene family, including FGF2, FGF3, FGF8, FGF9, FGF10, and FGF15/19 have been implicated in different stages of inner ear formation in different species (Pirvola et al., 2004; Wright et al., 2004; for reviews, Baker and Bronner-Fraser, 2001; Rinkwitz et al., 2001; Noramly and Grainger, 2002). These FGFs are involved in induction of the otic vesicle or in later development of the inner ear. The involvement of FGF3 in the formation of the otic vesicle has been demonstrated in chicks (Represa et al., 1991; Vendrell et al., 2000). However, knockout of FGF3 in mice still permitted formation of the inner ear (Mansour et al., 1993). Because both FGF3 and FGF10 are present in the early developing otic vesicle with FGF receptor type 2b (FGFR2b; Pirvola et al., 2000), an effective receptor for both ligands (Ornitz et al., 1996), it was speculated that these two FGFs may have redundant roles by means of FGFR2b during induction of the otic vesicle. We and others reported that Fgf10 null ears were still normal until otic vesicle formation and exhibited a mild abnormality in inner ear development (Ohuchi et al., 2000; Pirvola et al., 2000; Pauley et al., 2003). Recently, double-mutant mice with FGF3 and FGF10 were generated, in which the formation of the otic vesicle was severely reduced. It is thought, thus, that these FGFs act in combination with each other as neural signals for otic vesicle formation (Wright and Mansour, 2003; Alvarez et al., 2003).

Here, we addressed the role of FGF10 in later development of the inner ear by studying the inner ear phenotype in Fgf10 null mice. Although FGF10 and other FGFs have multiple and redundant roles in inner ear development, it is crucial to identify the individual role of each Fgf gene during organogenesis. On the other hand, the morphology of the inner ear is so complicated that deciphering the molecular mechanisms for transcriptional regulation of Fgf10 will help to elucidate important aspects of the mechanisms involved in inductive events during inner ear formation. So far, there are several reports on regulatory sequences to target the expression of a gene in the developing inner ear. For example, a promoter sequence specific to the inner ear sensory cells (hair cells) has been reported in the gene of unconventional myosin VIIA (Boeda et al., 2001), and enhancers for Sox2 expression in the nasal/otic placodes were determined (Uchikawa et al., 2003). A regulatory element of the Fgf3 promoter region was also identified, but whether this element could drive the expression in the otic vesicle was not decisively determined in vivo (Murakami et al., 2001). Taken together with these accumulating studies, the identification of ear enhancer sequences would open up the possibility for the analysis of developing ear transcriptome. We previously reported a transcriptional regulation of the Fgf10 gene during limb formation by identifying the 2.1-kb 5′-upstream region of Fgf10 that drives the expression in the limb bud and cartilages (Sasaki et al., 2002). In this study, we further elucidated an upstream cis-element regulating–specific expression of Fgf10 in the inner ear primordia by analyzing its enhancer activity in transgenic mice.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES

Ears in Fgf10 Null Mutants

FGF10 mutant mice have been reported to display relatively mild defects in inner ear development (Ohuchi et al., 2000; Pauley et al., 2003). At embryonic day (E) 10.5, inner ears of mutants appear morphologically normal; the endolymphatic duct that is absent from Fgfr2b null mice (De Moerlooze et al., 2000) was formed in Fgf10null otic vesicle. However, the mutant otic capsule was smaller, indicating some structural defects in the inner ear formation (Ohuchi et al., 2000). This preliminary result led us to examine the ear phenotype at later stages, just before birth, as Fgf10 null mice die at birth due to the absence of lungs. To better understand the complex morphology of the inner ear, we reconstructed three-dimensional images of the inner ear from histological sections using a computational program (Fig. 1A–C). We found that the major domains of the inner ear, an endolymphatic duct, semicircular canals, a central vestibule, and a coiled cochlea appeared to form in the mutant. However, semicircular canal formation was markedly disrupted in the Fgf10 null ear; semicircular disks remained without canal formation. The disk showed roughly L-shaped profile with single vertical disk facing medially with a lateral protrusion of the horizontal plate (Fig. 1B). Serial sectioning of the mutant ear showed two separate neuroepithelial layers: one of them was positioned closed to the vestibule or fused with the macula utriculi and the other in the lateral protrusion of the disks (Fig. 1F,F′). Therefore, they may correspond to anterior and lateral (horizontal) crista, respectively. However, we could not find the presence of the posterior crista in the mutant, whereas three distinct sensory epithelia with cupula were observed in the wild-type (Fig. 1D–E′). On the other hand, the sensory area of the vestibule developed morphologically normally in the mutant (Fig. 1G). The length of the cochlea duct and morphogenesis of endolymphatic duct also appeared normal in Fgf10 null mutants (Fig. 1B,C).

thumbnail image

Figure 1. Disturbed inner ear development in Fgf10 null mutant mice at late embryogenesis. A–C: Three-dimensional reconstruction of a membranous labyrinth of embryonic day (E) 19 wild-type (A) and age-matched Fgf10 knockout mice (B,C); medial (A,C) and anteromedial (B) views. The right side is forward the rostral. A: The horizontal semicircular duct (hsd) and utricle (u) cannot be distinguished from this angle. B: Semicircular disks remain without canal formation (asterisk). C: The mutant cochlear duct coils in 1.5 or less than 2 turns as observed in the normal. D–G: Transverse sections of E19 otic capsules stained with hematoxylin and eosin. E′,F′ are larger magnifications of E,F, respectively. D,F,F′,G: Typical morphology of semicircular ducts (as shown in D) is missing, and the ducts are replaced by a dilated cavity (asterisks) in the mutant (F,F′,G). D,E,E′,F′: Whereas three distinct sensory epithelia with cupula develop in the wild-type crista ampullaris (D,E,E′), only two of them are observed in the mutant, deformed horizontal crista (hc) and abnormal anterior crista (ac) fused with the macula utriculi (mu). G: The sensory areas of the utricle and saccule develop normally in the mutant. asd, anterior semicircular duct; cc, crus commune; co, cochlea; ed, endolymphatic duct; hp, horizontal plate; ms, macula sacculi; pc, posterior crista; psd, posterior semicircular duct; s, saccule. Scale bars = 100 μm in D–F, 200 μm in E′,F′,G.

Download figure to PowerPoint

We next investigated the developmental stage that is affected by Fgf10 inactivation and that leads to the absence of semicircular canal formation. First, the observation was made at E13.5, when the semicircular ducts are well formed in the normal inner ear (Morsli et al., 1998; Fig. 2A,B). The semicircular ducts emerge as bilayered outpocketings of the dorsal vestibular epithelium. Subsequently, the two opposing walls of these outpocketings approach each other to form a so-called fusion plate, in which the epithelial cells first intercalate to form a single layer and then disappear, creating a hollow duct, which will grow and obtain its adult form (Martin and Swanson, 1993). In the Fgf10 mutants, the walls of the outpocketing generating the anterior semicircular duct remained relatively dilated even at E13.5 (Fig. 2D). Careful observation of serial sections revealed that the formation of anterior and horizontal fusion plates was initiated (Fig. 2E). However, we could not find the formation of the posterior fusion plate. In contrast, the development of the saccule, cochlear duct, and cochleovestibular ganglion appeared normal in the mutant at E13.5 (Fig. 2C,F).

thumbnail image

Figure 2. Defects at the early stages of inner ear development in Fgf10 null mutants. A–F: Hematoxylin and eosin–stained transverse sections of embryonic day (E) 13.5 inner ears; the mutants (D–F) show that prominent hypoplasia of the vestibular mesenchyme (m), which is associated with the absence of semicircular ducts, compared with normal littermates (A–C). A,B: In normal embryos, the opposing walls of the outpocketing giving rise to the anterior semicircular duct (asd) are approaching one another at E12.0, the corresponding fusion plate is formed by E12.5, and semicircular ducts are well formed by E13.5. D: In the mutants, the walls of the outpocketing generating the anterior semicircular duct remain relatively dilated even at E13.5. E: Serial sections show that the formation of anterior and horizontal (inset) fusion plates is initiated. C,F: The development of the saccule (s), cochlear duct (co), and cochleovestibular ganglion (VIII) appears normal in the mutant at E13.5. G–J: RNA in situ analysis of netrin 1 expression in normal (wild-type/heterozygote) (G,H) and Fgf10 null (I,J) ears at E12.5. Transverse sections through anterior (H,J) and posterior (G,I) outpocketings, counterstained with nuclear red. G,H:Netrin 1 expression (in blue) is detected in the otic epithelium forming the fusion plates. I,J: In the mutant, the extent of the area expressing netrin 1 is decreased and the level of netrin 1 expression appears lower in the epithelium. K,L: RNA in situ analysis of Nkx5.1/Hmx3 expression in normal (wild-type/heterozygote, K) and Fgf10 null (L) ears. Nkx5.1 is expressed in the developing vestibular ducts of both controls and mutants at E13.5. Nkx5.1 expression appears down-regulated in most of the dilated ducts of the mutant, whereas a section of a mutant horizontal semicircular duct (inset in L) exhibits a distinct expression of Nkx5.1. ed, endolymphatic duct; fp, fusion plate; hsd, horizontal semicircular duct; psd, posterior semicircular duct; u, utricle. Scale bars = 100 μm in A–L, 50 μm in inset in L.

Download figure to PowerPoint

To further dissect the phenotype of Fgf10 null inner ear, we compared the expression of two genes, netrin 1 and Nkx5.1 (also termed as Hmx3) in Fgf10 mutants and controls. Netrin 1 is a laminin-related protein, whose inactivation leads to the absence of semicircular ducts due to the blockage of fusion plate formation (Salminen et al., 2000). In normal embryos, the opposing walls of the outpocketing giving rise to the anterior semicircular duct are approaching one another at E12.0, and the corresponding fusion plate is formed by E12.5. As reported, netrin 1 expression was detected in the otic epithelium forming the fusion plates in the normal littermate (Fig. 2G,H). In the mutant, netrin 1 was expressed, but the extent of the area expressing netrin 1 seemed decreased and the level of netrin 1 expression appeared lower in the epithelium (Fig. 2I,J). On the other hand, Nkx5.1 is a homeobox gene, which is expressed in the developing vestibular structures (Wang et al., 1998). Mice carrying the Nkx5.1 null mutation exhibit severe malformations of the semicircular canals (Hadrys et al., 1998; Wang et al., 1998). In situ hybridization analysis showed that Nkx5.1 was expressed in the developing vestibular ducts of both controls and mutants at E13.5 (Fig. 2K,L). However, Nkx5.1 expression appeared down-regulated in the dilated ducts of the mutant (Fig. 2L).

These results suggest that FGF10 is required for semicircular canal morphogenesis, especially in later stages after fusion plate formation, and indispensable for the subsequent removal of the fused cells to proceed with the hollowing of the center of each semicircular plate. Expression of netrin1 and Nkx5.1 in the inner ears of Fgf10 mutants indicates that FGF10 is not required for the initiation of gene transcription of netrin1 or Nkx5.1.

Role of FGF10 in Inner Ear Development

The detailed expression patterns of Fgf10 and its major receptor Fgfr2b have already been reported: Fgf10 and Fgfr2b mRNAs exhibit distinct, complementary expression patterns in the undifferentiated otic epithelium (Fgf10 in the ventral epithelium and Fgfr2b in the dorsal epithelium; Pirvola et al., 2000). Subsequently, Fgf10 mRNA becomes confined to the presumptive cochlear and vestibular sensory epithelia and to the neuronal precursors and neuron, while Fgfr2b mRNA is expressed in the nonsensory epithelium of the otocyst that gives rise to structures such as the endolymphatic and semicircular ducts (Pirvola et al., 2000). From these expression patterns, it has been concluded that inner ear development depends on paracrine signals that operate within the epithelium. Because Fgfr2b is expressed by the presumptive semicircular epithelium and the Fgf10 null ear exhibits malformation of semicircular canals, FGF10–FGFR2b signaling is likely to mediate some inductive processes from sensory epithelium to nonsensory epithelium during semicircular canal formation. This type of tissue interaction is reminiscent of that observed between the Rathke's pouch and the infundibulum during pituitary development: Fgf10 is expressed in the infundibulum, whereas Fgfr2b is expressed in the Rathke's pouch (Takuma et al., 1998). The anterior pituitary derived from Rathke's pouch is absent from Fgf10- or Fgfr2b-deficient mice (Ohuchi et al., 2000; De Moerlooze et al., 2000). Furthermore, in otic placode induction, FGF10 has a redundant role with FGF3, as revealed by severely reduced otic vesicles in double mutant mice with Fgf3 and Fgf10 (Wright and Mansour, 2003; Alvarez et al., 2003). At this phase of inner ear development, mouse Fgf10 is expressed in the mesenchyme underlying the prospective otic placode (Wright and Mansour, 2003), indicating that FGF10 signaling mediates epithelial–mesenchymal interactions in early otic development. Thus, it seems likely that FGF10-FGFR2b signaling is involved in dual aspects of tissue interactions during inner ear development.

Inner ear defects of another Fgf10 null mouse (Min et al., 1998) were examined by Pauley et al. (2003). Our characterization of the Fgf10 null mutation further has revealed several points. In the absence of FGF10, (1) semicircular plates remain without the hollowing of the center of each semicircular plate; (2) morphogenesis of the cochlear duct (turning, length) occurs normally; (3) expression of netrin 1 and Nkx5.1 can be observed in the developing semicircular primordium, showing fusion plate formation initially takes place. Compared with the previously published data, there are the similarities in the inner ear phenotype of these Fgf10 null mutants as follows; (1) Fgf10 null mutants show complete agenesis of the posterior crista and the posterior semicircular canal; (2) Fgf10 mutants have deformations of the anterior and horizontal cristae and reduced formation of the anterior and horizontal canals. Taken together, FGF10 is required for proceeding the removal of the fused cells after semicircular plate formation, especially for the posterior canal formation. The fusion plate cells are thought to be recruited back into the duct epithelium in the mouse (Martin and Swanson, 1993), whereas programmed cell death has been shown to play an important role in removing them in the chick embryo (Fekete et al., 1997). Therefore, FGF10 may be involved in cell migration and death during the removal of semicircular fused cells at the cellular level. Furthermore, recent studies have shown that a canal genesis zone is present adjacent to each prospective crista (Chang et al., 2004) and that FGFs, including FGF10 in the crista promote canal development (Chang et al., 2004; Pirvola et al., 2004). Because FGF10 is an extracellular protein with a high affinity to heparin, it is conceivable to think that FGF10 secreted from the developing crista has effects on the neighboring prospective canal epithelium by means of such proteins with heparin-binding domains as netrin 1.

Sequence Analyses and Identification of the Transcriptional Regulatory Region of the Fgf10 Gene

The dynamic expression pattern of Fgf10 during intricate inner ear morphogenesis (Alvarez et al., 2003; Pauley et al., 2003; Wright and Mansour, 2003) prompted us to study the regulatory element that drives the expression in the inner ear anlage. To elucidate the regulatory network involved in organ-specific Fgf10 expression, we compared the 5′ sequences from the human, mouse, and chicken genomes. Although human and mouse sequences are relatively well-conserved up to 6.6 kb from the initiation codon (Fig. 3A), comparison with the chicken sequence shows a highly conserved sequence (more than 75%) up to 3.7 kb (2.9- to 6.6-kb region in Fig. 3B). There are four regions where mouse sequences are highly homologous to the corresponding chicken ones, indicating the presence of conserved enhancers. The conserved region up to 2.1 kb from the initiation codon, containing regions “b”, “c”, and “d” is involved in limb bud-specific expression of the Fgf10 gene, as reported previously (Sasaki et al., 2002). A region further upstream (2.9–3.3 kb; designated as region “a” in Fig. 3A,B) appears to have an inner ear-specific enhancer element leading to Fgf10 transcription in the otic primordia, as described below.

thumbnail image

Figure 3. A,B: Homology analysis of 5′ fragment of Fgf10 between mouse and human (A) and between mouse and chicken (B). There are four conserved regions indicated by the squares in the fragment. Region “a” is an ear enhancer; regions “b”, “c”, and “d” were previously reported (Sasaki et al., 2002). There is a repetitive sequence (RS) element in the minus 4.5- to 5.1-kb region of the mouse Fgf10 gene that is absent in the human genome. C: Schematic illustration of the constructs for generating transgenic mice. The 5′ fragment (F) was digested by restriction enzymes into three smaller fragments (KpnI for K; ApaI for A, BamHI for B, 0.2 kb). The restriction enzyme fragments containing regions “a” to “d” were designated as 3-1, 9-1, 10-1, and 8-1, respectively.

Download figure to PowerPoint

Identification of the Mouse Cis-Control Elements of Fgf10 for the Transgene Expression in the Developing Ear

To identify the minimal cis-control elements regulating Fgf10 expression in the developing embryo, we obtained DNA fragments containing conserved sequences designated as 3-1, 9-1, 10-1, and 8-1 (Fig. 3C). We generated transgenic mice that had a lacZ reporter gene under the control of 6.6- (designated as construct F), 2.1- (K), 0.7- (A), or 0.3- (B) kb 5′ fragments of Fgf10. We could generate two lines of transgenic mice with construct F: lines 20 and 53. In both lines, transgene expression was similarly detected in the otic vesicle at E9.5 (Fig. 4A; line 20; not shown for line 53 at E9.5). A strong FGF10 signal was found in the otic epithelium, except for its dorsal portion, and in the emerging ganglion cells (Fig. 4B). These expression domains of lacZ correspond well with the localization of Fgf10 mRNA previously reported (Pirvola et al., 2000). At E11, lacZ expression was found in the sensory epithelia of the semicircular, vestibular, and cochlear regions, as well as in the vestibulocochlear ganglion (Fig. 4E–H). The two transgenic lines exhibit a distinct lacZ expression in the endolymphatic duct, where Fgf10 mRNA was not usually detected (Fig. 4E,H,J; Pirvola et al., 2000). On the other hand, in the transgenic embryo with construct K, the transgene was not expressed in the developing ear (n = 8; Fig. 4D;Sasaki et al., 2002). Thus, the enhancers for expression in the ear anlage seemed to be localized between −6.6 and −2.1 kb of the 5′ fragment of Fgf10. Therefore, we generated transgenic mice with construct 3-1, which contains the conserved “a” region (Fig. 3). In relatively early embryos with construct 3-1, to date, we could not detect lacZ expression in the otic anlage as revealed by whole-mount X-gal staining (Fig. 4C, n = 2 for ∼E8.5, n = 2 for ∼E9.5). By E11, lacZ expression was detected in the developing inner ear (Fig. 4I). However, the expression domains driven by the short construct were more restricted to a portion of the inner ear anlage compared with those of construct F (Fig. 4J,K): construct 3-1 has driven the expression in one side of the endolymphatic duct and developing horizontal crista, as revealed by whole-mount X-gal staining, whereas in the embryo with construct F lacZ expression was found in the whole domain of the endolymphatic duct and all three crista. At E12, the lacZ expression by construct 3-1 was observed in more restricted manner in the horizontal and posterior cristae, compared with construct F (Fig. 4L,O,P). Serial sections of a 3-1 transgenic embryo showed lacZ expression in the developing crista and macula, ganglion VIII neurons, and emerging organ of Corti of the cochlea (Fig. 4M,N). We found that construct 3-1 drives a distinct expression in the developing horizontal crista, because all of nine embryos examined at E12 exhibited intense lacZ staining in the horizontal crista as revealed by whole-mounts. These results indicate that the enhancers for expression in the sensory compartment of developing inner ears, at least in the horizontal crista, are contained in the conserved “a” region of the sequence 3-1 (Fig. 3C). Because construct F can drive the lacZ expression in broader domains of the inner ear primordium than construct 3-1 while construct K cannot drive the expression in the developing ear, additional enhancers for Fgf10 expression in the inner ear anlage may be present other than the K and 3-1 sequences.

thumbnail image

Figure 4. Expression of the lacZ reporter gene in transgenic embryos with the construct F/lacZ (A,B,E–H,J,O), 3-1/lacZ (C,I,K,L–N,P), and K/lacZ (D). Embryonic days of the mouse embryos are indicated in the right bottom corner of each panel. The arrows in (A,C,D,E,I) show the otic vesicle or ear primordium. A,C: By E9.5, construct F drives lacZ expression in the otic vesicle (A), while construct 3-1 cannot drive a distinct expression of lacZ at this stage (C), as revealed by whole-mount X-gal staining. B: Transverse section of the stained transgenic embryo at E10 (construct F, line 53), showing lacZ expression in the otic vesicle (ov). Because of greater sensitivity of β-galactosidase detection, the expression seems to expand to a bit more dorsal domain of the otic vesicle than reported on localization of Fgf10 mRNA (Pirvola et al., 2000). The inset shows a serial section of the same embryo with lacZ expression in the emerging ganglion cells (arrowhead). D: Construct K cannot drive lacZ expression in the otic vesicle. E: The E11 embryo (construct F, line 20, also in J) shows lacZ expression in the otic vesicle, eyelid, and limb region. F–H: Transverse sections of the stained embryo in E. F,G: The lacZ expression is found in the sensory epithelia of the semicircular duct and vestibular region (arrows in F,G) and ganglion cells (g in F). H: The line 20 embryo also exhibits the expression in the wall of the endolymphatic duct (ed), utriculo-saccular space (us), and cochlear duct (cd). I: Construct 3-1 (line 1) drives the expression in the restricted area of the inner ear by E11, compared with construct F. J: High-power view of the otic vesicle in E. The lacZ expression is found in the endolymphatic duct (arrowhead) and developing crista ampullaris. K: High-power view of the otic vesicle in I. Construct 3-1 drives the expression in one side of the endolymphatic duct (arrowhead) and developing horizontal crista. L: At E12, more restricted expression in the inner ear is driven by construct 3-1. M: Transverse sections of the stained embryo in L. LacZ expression in the developing crista (arrow), and ganglion VIII neurons. N:LacZ expression in the developing macula (arrow), ganglion neurons, and emerging organ of Corti of the cochlea (co). O: The lacZ expression driven by construct F (line 20) in the endolymphatic duct (arrowhead) and in all three crista. P: Construct 3-1 drives lacZ expression in more restricted portion of the developing inner ear. ac, anterior crista; asd, anterior semicircular duct; hb, hindbrain; hc, horizontal crista; hv, primary head vein; p, pharynx; pc, posterior crista. Scale bars = 100 μm in B,F–H,M,N.

Download figure to PowerPoint

Transcriptional Regulation of the Fgf10 Gene

Figure 5A shows the alignment of mouse, human, and chicken nucleotide sequences for fragment 3-1. A highly conserved sequence among these three species has been detected in the first half of the sequence, approximately 400 base pair long. In this 0.4-kb fragment, there are many potential enhancer elements that are presumed to be activated by homeodomain-containing proteins, POU-domain factors, zinc-finger transcription factors, TCF/LEF-1, and a SMAD-interacting protein (Table 1; Fig. 5B). Numerous mouse mutations have been identified that show defects in morphology of the inner ear, its sensory organs, or the vestibulocochlear ganglion neurons (for a review, see Fekete, 1999). For example, the semicircular canals fail to develop in Nkx5.1 null mutants or in mice with a double knockout of Prx1 and Prx2 (Wang et al., 1998; Hadrys et al., 1998; ten Berge et al., 1998). The Hoxa1 knockouts exhibit varying degrees of severity in inner ear dysmorphogenesis (Lufkin et al., 1991; Chisaka et al., 1992; Mark et al., 1993). This and other (Pauley et al., 2003) studies have shown that FGF10 is required for semicircular canal formation. Although we suggests a possibility that Fgf10 might be transcriptionally regulated by some Prx, Hox, and Nkx proteins, further work is needed to elucidate the relationship between Fgf10 and these transcription factors, such as detailed comparison of their expression domains and mutagenesis of the putative binding sites.

thumbnail image

Figure 5. A: Comparison of nucleotide sequences corresponding to fragment 3-1 among mouse, human, and chicken. The asterisks indicate the identical nucleotide among the three; the dot indicates where two of the three species have the same nucleotide. The 5′-half of the sequences is highly conserved among mouse, human, and chicken. B: Positions of DNA binding motifs conserved among mouse, human, and chicken in the 0.4-kb element of the mouse Fgf10. Corresponding to the 5′-half of the fragment 3-1, analyzed with MatInd and MatInspector (Quandt et al., 1995). For details, see Table 1.

Download figure to PowerPoint

Table 1. Conserved Elements of the Fgf10 Promotera
No.Name of familyFurther informationOpt.Position from-toStr.Core sim.Matrix sim.Sequence
  • a

    Conserved elements in the 5′ region of the Fgf10 promoter between mouse, human, and chicken. Numbers in the left-most column correspond to those in Figure 5B. Opt. (optimized matrix threshold): This matrix similarity is the optimized value defined in such a way that, at most, three matches are found in 10,000 bp of random DNA sequences. Position: This is shown by the number of nucleotide in the mouse, starting from the first nucleotide of fragment 3-1. Str., Strand. Core sim. (core similarity): The “core sequence” of a matrix is defined as the highest consecutive conserved positions (usually four) of the matrix. The core similarity is calculated as described in the MatInspector paper (Quandt et al., 1995). The maximum core similarity of 1.0 is only reached when the highest conserved bases of a matrix match exactly in the sequence. Matrix sim. (matrix similarity): Matrix similarity is calculated as described in the MatInspector paper (Quandt et al., 1995). A perfect match with the matrix gets a score of 1.00 (when each sequence position corresponds to the highest conserved nucleotide at that position in the matrix), a “good” match to the matrix usually has a high conservation profile (ci- value > 60). Base pairs in capital letters denote the core sequence used by MatInspector.

1S8Binding site for S8 type homeodomains0.9751–59(+)10.995aacaATTAa
2MSXHomeodomain proteins MSX-1 and MSX-20.9749–61(−)10.978tatTAATtgtttc
3BRIGHTBright, B cell regulator of IgH transcription0.9250–62(+)10.947aaacaATTAataa
4SOX5Sox-50.8748–64(+)10.981agaaaCAATtaataaag
5BRN3POU transcription factor Brn-30.7849–65(+)0.750.815gaaACAAttaataaagg
6HOXA5Hox a-5, vertebrate homeobox protein0.8350–66(−)10.887tcctttATTAattgttt
7LHX3Homeodomain binding site in LIM/homeodomain factor LHX30.8153–63(+)10.851caaTTAAtaaa
8NKX25Homeodomain factor Nkx-2.5/Csx, tinman homolog low affinity sites0.8865–77(−)10.903tctTAATgggttc
9TST1POU-factor Tst-1/Oct-60.87115–129(+)10.873gtagAATTtcagtcc
10LEF1TCF/LEF-1, involved in the Wnt signal transduction pathway0.94124–140(−)10.982gttacttCAAAggactg
11HLFHepatic leukemia factor0.84125–145(−)10.85tctcaGTTActtcaaaggact
12VBPPAR-type chicken vitellogenin promoter-binding protein0.86130–140(−)10.903gTTACttcaaa
13E4FGLI-Krueppel—related transcription factor, regulator of adenovirus E4 promoter0.82131–143(+)0.7890.881ttgAAGTaactga
14MOK2Ribonucleoprotein-associated zinc-finger protein MOK-2 (mouse)0.74130–150(−)0.750.767atcattctcagttACTTcaaa
15VMYBv-Myb0.9135–145(+)0.8760.912agtAACTgaga
16AP1Activator protein 10.95145–155(+)0.8460.961aatgaTTCAgg
17PIT1Pit1, GHF-1 pituitary-specific pou domain transcription factor0.86145–155(+)10.919aatgATTCagg
18MEIS1Homeobox protein MEIS1 binding site0.79146–158(+)10.793aTGATtcaggcct
19PAX6Pax-6 paired domain binding site0.75148–166(+)0.7540.821gattcAGGCctcattacag
6HOX1-3Hox-1,3, vertebrate homeobox protein0.83154–170(+)10.853ggcctcATTAcagagat
20ISL1Pancreatic and intestinal LIM-homeodomain factor0.82152–172(−)10.846atatctctgTAATgaggcctg
21GATA3GATA-binding factor 30.91164–176(+)10.958cagAGATataatc
22FAST1FAST-1 SMAD interacting protein0.81169–183(−)0.9830.869gaatgtaGATTatat
23CART1Cart-1 (cartilage homeoprotein 1)0.84170–186(+)10.862taTAATctacattcata
17PIT1Pit1, GHF-1 pituitary-specific pou domain transcription factor0.86180–190(+)0.820.892attcATACatt
24CREBP1cAMP-responsive element binding protein0.8183–203(−)0.7880.811aatcactaACATaaatgtatg
25HNF1Hepatic nuclear factor 10.8193–209(+)10.865tGTTAgtgattcaaatc
16AP1Activator protein 10.95197–207(−)0.8840.95tttgaATCAct
17PIT1Pit1, GHF-1 pituitary-specific pou domain transcription factor0.86197–207(+)10.879agtgATTCaaa
26CHRCell cycle gene homology region (CDE/CHR tandem elements regulate cell cycle dependent repression)0.92198–210(−)10.971agatTTGAatcac
27PTX1Pituitary homeobox 1 (Ptx1)0.79198–214(−)0.7890.825caatagaTTTGaatcac
28CDPCR3Cut-like homeodomain protein0.75199–215(+)0.9750.752tgattcaaatctATTGg
21GATA3GATA-binding factor 30.91201–213(−)10.937aatAGATttgaat
29HNF6Liver enriched Cut - Homeodomain transcription factor HNF6 (ONECUT)0.82203–217(+)0.7850.83tcaaaTCTAttggga
30STATSignal transducers and activators of transcription0.87222–240(−)10.909gctatttaacGGAAtccaa
15VMYBv-Myb0.9226–236(−)10.979tttAACGgaat
31EN1Homeobox protein engrailed (en-1)0.77225–241(−)10.785cgctaTTTAacggaatc
32BARBIEBarbiturate-inducible element0.88302–316(+)10.904atgcAAAGtggtggg
33OCTOctamer-binding factor 10.9371–385(−)0.9440.926tATATctcattcttc
34GATA2GATA-binding factor 20.9376–388(+)10.935atgaGATAtaatc
35VMAFv-Maf0.82388–412(−)10.878aggaaaagcTGACatagctttttag
36RORA2RAR-related orphan receptor alpha20.82392–408(+)0.750.831aaagctaTGTCagcttt
37MYT1MyT1 zinc-finger transcription factor involved in primary neurogenesis0.75398–410(−)0.750.756gaaAAGCtgacat

Mutations that affect the cell types and tissue of the inner ear have also been identified. The inner ear houses the sensory organs for both hearing and balance, which use the same type of cell for sensory transduction: the hair cell. This study on the Fgf10 enhancer analysis has verified Fgf10 expression in the sensory portion of developing otic epithelium (Pauley et al., 2003). Pauley et al. (2003) further described that the Fgf10 null inner ear exhibited some defects in cilia formation of hair cells. Thus, it is crucial to identify upstream regulators for Fgf10 in establishment of the otic sensory system. A mutation of Brn3c results in a failure of hair cells to appear in the inner ear, with subsequent loss of cochlear and vestibular ganglia (Erkman et al., 1996). Brn3a and Brn3b are also expressed in subsets of spiral and vestibular ganglion neurons (Xiang et al., 1998). On the other hand, a zinc-finger transcription factor, GATA3 is expressed in the sensory domain of the developing inner ear and involved in specification of auditory neurons (Karis et al., 2001; Lawoko-Kerali et al., 2004). Taken together with the findings of this study, it is most likely that Brn3 and GATA3 may directly regulate the expression of Fgf10 in establishment of the otic sensory system.

Analysis of the Fgf10 ear enhancer sequence suggests the involvement of Wnt- and BMP-signaling pathways in the regulation of Fgf10 expression. Bmp2 is expressed in the prospective semicircular canals (Chang et al., 2002), whereas it has been proposed that Wnt signaling gives rise to planar polarity of the outer hair cells in mammalian cochlea (Dabdoub et al., 2003). Interestingly, it was reported that the synergistic interactions of a member of FGF, FGF19, and Wnt8c initiate inner ear development (Ladher et al., 2000). Thus, it will be intriguing to clarify the specific interactions between FGF10 and these signals during inner ear development.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES

Mice

Fgf10 knockout mice were generated on a C57BL/6 X CBA background and genotyped as described (Sekine et al., 1999).

RNA In Situ Hybridization

Mouse embryos were fixed in 4% paraformaldehyde and dehydrated before embedding in paraffin or cryoprotected with 30% sucrose before embedding in O.C.T. compound. In situ hybridization was performed on 5-μm-thick paraffin sections or 18-μm-thick cryosections as described (Schaeren-Wiemers and Gerfin-Moser, 1993). A DNA fragment coding for the amino acids 25–185 of the Nkx5.1 (Hmx3) protein (Bober et al., 1994) was generated by reverse transcriptase-polymerase chain reaction (RT-PCR) from total RNA isolated from an E12.5 mouse head, cloned into pGEM-T-Easy vector (Promega), and used for preparing the probe. The template for the mouse netrin 1 probe (nucleotides 1431–1679) was obtained by RT-PCR and cloned into pBluescript vector (Stratagene). The digoxigenin-labeled RNA probes were prepared according to the standard procedure. The corresponding sense probes were used in parallel with antisense probes as negative controls.

Three-Dimensional Reconstruction of the Inner Ear

Eight-micrometer-thick coronal sections of the temporal region were cut with a microtome, and every sixth section was collected onto albumin-coated glass slides. They were stained with hematoxylin and eosin. The inner ear pictures were taken using a DP70 charge coupled device camera (Olympus Co., Tokyo, Japan). After adjustment of the axis of the temporal bone, outlines of the epithelial vesicle (which develops into the membranous labyrinth) were traced for digitizing with the use of a three-dimensional reconstruction software program (Tri software; Ratoc System Co., Ltd., Tokyo, Japan).

Isolation of the 5′-Flanking Region of Fgf10

Mouse Fgf10 genomic clones were isolated from a TT2 ES cell genomic library by plaque hybridization using full-length rat Fgf10 cDNA as a probe (Sekine et al., 1999). Chicken Fgf10 genomic clones were isolated from a chicken genomic library as previously described (Sasaki et al., 2002). We compared sequences of the mouse 6.6-kb 5′ fragment of the Fgf10 promoter region with the corresponding regions of human and chicken Fgf10 as a percentage identity plot (pip), which shows both the position in one sequence and the degree of similarity for each aligning segment between the two sequences (obtained with PipMaker; Schwartz et al., 2000), as illustrated in Figure 3.

Generation of Transgenic Mice and Staining for β-Galactosidase Activity

The isolated genomic clone containing a 6.6-kb 5′ upstream fragment of Fgf10 was designated as ”F” (Fig. 3C). The “F” fragment was digested with KpnI, ApaI, and BamHI into 2.1- (K), 0.7- (A), 0.3- (B) kb fragments (Fig. 3C; Sasaki et al., 2002). According to homology analysis (Fig. 3A,B), the clones 3-1, 9-1, 10-1, and 8-1 were obtained by subcloning of the restriction fragments (Fig. 3C). The analysis of clones 9-1, 10-1, and 8-1 will be reported elsewhere. To construct transgenes, each fragment was inserted into a lacZ reporter vector containing a promoter of the heat shock protein 68 (hsp68) and the Shine-Dalgarno-Kozak (SDK) sequence (Sasaki and Hogan, 1996).

After excision of a transgene from the vector, the concentration of the transgenes was adjusted to 500 molecules/pl and the solution was microinjected into the male pronuclei of fertilized eggs derived from superovulated BDF1 (C57BL/6 X DBA2 F1) female mice crossed with males of the same strain. Oviduct implantation of the surviving injected embryos into pseudopregnant MCH/ICR female mice was carried out according to the standard protocol. After implantation, pregnant mice were put to sleep and the embryos from E8 to E14.5 were treated with 2% paraformaldehyde for 30 min. Reporter activity was analyzed by X-gal staining overnight at 37°C. X-gal stained embryos were post-fixed in 2% paraformaldehyde, dehydrated, placed in xylene, and embedded in paraffin. The sections with a thickness of 20 μm were prepared for analysis of signals at the cellular level (Sasaki et al., 2002). The integration of the transgene into the mouse genome was detected by PCR, using primers established at the sequence of LacZ cassette and DNA extracted from tail snips of 3-week-old offspring by the proteinase K/SDS method. With construct F, two stable transgenic lines (lines 20 and 53) were obtained and had essentially the same lacZ expression pattern in the developing inner ear, although the expression pattern in other sites were different between the two transgenic lines. With construct K, none of transient transgenic embryos (n = 8; Fig. 4D) showed lacZ expression in the developing ear. As for construct 3-1, all eight of the transient transgenic embryos and one stable line obtained so far (line 1) exhibited a similar lacZ expression in the developing inner ear.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES

The authors thank Dr. R. Shigemoto and his lab members for allowing us use of their equipment. We also thank C. Komaguchi for her technical assistance. The sequence of the mouse FGF10 inner ear enhancer has been deposited with DDBJ/GenBank under the accession no. AB176670, and the chicken Fgf10 genomic sequence, including the inner ear enhancer and exon 1, under the accession no. AB176671. The materials for chicken Fgf10 genomic DNA and mouse inner ear enhancer should be requested from T.N. Funding was provided to S.N., T.N., and H.O. by the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS AND DISCUSSION
  5. EXPERIMENTAL PROCEDURES
  6. Acknowledgements
  7. REFERENCES