The paired sensory organs of the head develop from ectoderm lying close to the neural plate (Baker and Bronner-Fraser, 2001). This ectoderm becomes organized into a series of craniofacial placodes, from which will arise the olfactory epithelium, the lens of the eye, trigeminal and epibranchial ganglia, and the entire inner ear. A variety of experiments show that much of this early ectoderm is competent to give rise to different placodes if grafted to the appropriate location (Jacobson, 1963, 1966; Baker et al., 1999; Groves and Bronner-Fraser, 2000). In addition, several genes have been shown to mark the border between the neural plate and epidermis (Yang et al., 1998; Quint et al., 2000; Streit, 2002). These data have led to a model in which all craniofacial placodes arise from a common “pan-placodal” or “preplacodal” domain marked by these border genes. Initial inductive events are proposed to establish this pan-placodal domain, followed later by more focal signaling events that induce specific placodes (Baker and Bronner-Fraser, 2001; Brown et al., 2003; but see also Begbie and Graham, 2001 for an alternative view).
To understand the molecular mechanisms responsible for competence and cell fate specification in pan-placodal ectoderm, we wished to find more genes expressed specifically in this region. In particular, certain types of transcription factors have proved to be useful markers of the pan-placodal cell state and of specific sensory placodes. Two examples of pan-placodal gene markers are Dlx5 and 6, which are expressed broadly in the surface ectoderm from late embryonic day (E) 6.5 to the late presomite stages in mouse (Quint et al., 2000; Streit, 2002; S. Brown and A.K.G., unpublished data). This broad expression may correlate with competence to form cranial placodes at this stage. At early somite stages, Dlx5 and 6 expression becomes restricted to the frontonasal ectoderm and the otic placode (Yang et al., 1998). At this time, Pax2 and 8 start to be expressed specifically in the presumptive otic region from the late presomite stage onward (Ohyama and Groves, 2004). These dynamic expression patterns suggest that the induction of the specific placodes or other organs derived from the pan-placodal ectoderm starts from the late presomite stages. In an attempt to find more genes that mark the pan-placodal domain and specific cranial placodes, we focused on the Foxi gene family.
Since the discovery of the highly conserved motif between the Drosophila forkhead gene (Weigel et al., 1989; Weigel and Jackle, 1990) and the mammalian HNF-3 transcription factor (Lai et al., 1990, 1991), more than 100 members of the Fox (forkhead box) gene family have been identified from yeast to human (Kaestner et al., 2000). Fox proteins have diverse roles in organisms from embryonic development to the control of metabolism in differentiated tissues (reviewed in Kaufmann and Knochel, 1996; Carlsson and Mahlapuu, 2002). Foxi class genes have been identified and analyzed in several species. Mouse Foxi1, previously called Fkh10, is expressed in the otic vesicle at the E9.5 and the expression becomes restricted to the epithelium of endolymphatic duct and sac at the later stages. Targeted mutation of Foxi1 causes an abnormal expansion of the membranous labyrinth, and the resulting mutant mice suffer from hearing impairment and vestibular dysfunction (Hulander et al., 1998, 2003). In addition, mouse Foxi1 is also expressed in the embryonic and adult kidney from E16.5, although the mutant mice show no phenotype in the kidney (Overdier et al., 1997). Zebrafish foxi1 is expressed in the early otic placode and branchial arches, and the zebrafish foxi1 mutant (hearsay) and morpholino knock-down experiments show that zebrafish foxi1 regulates dlx and pax gene expression in the early otic placode and branchial arches (Lee et al., 2003; Nissen et al., 2003; Solomon et al., 2003a). Xenopus foxi1a and b are mainly expressed in the neuroectodermal and mesodermal lineage during early embryogenesis (Lef et al., 1994), and foxi1c is expressed in the epidermal ring around the neural field and subsequently localized in placodal precursor cells (Pohl et al., 2002). Recently, Solomon and colleagues have identified three other foxi class genes in zebrafish. Zebrafish foxi2 is expressed in the chordamesoderm during early somitogenesis and the retina and branchial arches during later stages. Zebrafish foxi3a and b are expressed in the epidermal mucous cells throughout embryogenesis and early larval stages. Zebrafish foxi2, foxi3a, and b are not expressed in the otic lineage (Solomon et al., 2003b).
Here, we show the expression patterns of the mouse Foxi class genes during early embryogenesis. Mouse Foxi genes are expressed in an early pan-placodal ectodermal domain, and also later in particular cranial placodes. Their expression shows some similarities, but also significant differences with the expression of Foxi genes in other vertebrate species.
RESULTS AND DISCUSSION
Identification of the Mouse Foxi2 and Foxi3 Genes
A homology search of the mouse Foxi1 cDNA against the entire mouse genome by using the Ensembl Web site (www.ensembl.org) revealed two more Foxi class genes in the mouse genome. We designated these as Foxi2 and 3 according to the nomenclature of Solomon and colleagues (Solomon et al., 2003b). Expressed sequence tag (EST) clones of both Foxi2 (Ensembl gene ID ENSMUSG00000048377) and Foxi3 (NCBI gene ID LOC232077) exist in several databases of mouse embryonic EST libraries such as RIKEN FANTOM database, indicating their mRNAs are expressed during embryonic development.
Foxi2 Is Expressed in Cranial Ectoderm
Expression analysis from E6.5 to E10.5 revealed that mouse Foxi2 is expressed faintly in the ectoderm of the midbrain–hindbrain boundary region at the late presomite stage (ss). This expression is no longer visible at later stages (data not shown). At the 4–5ss, more distinct but patchy expression is visible in the cranial ectoderm (Fig. 1A). At this stage, the Foxi2 expression domain overlaps the Pax2 expression domain, which marks the presumptive otic region in the cranial ectoderm adjacent to rhombomere 3 to 5 (Fig. 1B). By the 10–12ss, the ectoderm expression has become stronger and extends toward the dorsal half of the head region and the dorsal part of the branchial arches (Fig. 1C). The posterior expression is visible in the ectoderm around the point of dorsal closure, notably excluding the otic placode (arrowheads in Fig. 1C,F). Transverse sections of whole-mount embryos revealed that Foxi2 is not expressed in the thickening otic placode, whereas it is expressed in the more ventral ectoderm that is still one-cell-layer thick (Fig. 1D,E,G, H). As the otic placode invaginates, it becomes easier to distinguish the Foxi2-negative otic cup from the foxi2-positive ectoderm. At E9.5, Foxi2 expression in the craniofacial ectoderm becomes more robust (Fig. 2A–E). However, the expression in the first and second branchial arches is gradually restricted to the posterior edges at E10.5 (Fig. 2J). Sections of double whole-mount in situs with Foxi2 and Neurogenin-2 (Fig. 2F,G) or NeuroD (Fig. 2H,I) showed that the Neurogenin-2–positive cells in the epibranchial placodes are double-labeled with Foxi2 (arrowheads in Fig. 2G), whereas NeuroD-positive cells delaminating from the epibranchial placodes (Fode et al., 1998) are not Foxi2-positive (arrows in Fig. 2I), indicating that Foxi2 is down-regulated when the placodal cells delaminate from the cranial ectoderm.
Recently, a fate mapping study in chick showed that ectodermal cells in the presumptive otic region intermingle with future neural, neural crest, epidermal, and other placodal cells (Streit, 2002). Both chick and mouse Pax2 expression in the presumptive otic ectoderm covers a wide area up to the level of rhombomere 1 at the 4–5ss (Fig. 1B, see also Groves and Bronner-Fraser, 2000; Ohyama and Groves, 2004). Mouse Foxi2 is expressed in a salt-and-pepper manner in the cranial ectoderm at the 4–5ss and overlaps the Pax2 expression domain. At this stage, the patchy expression of Foxi2 makes it hard to determine whether the Foxi2 domain is a subset of the Pax2 domain, or vice versa. In the absence of Foxi2 antibodies, we were unable to determine whether Foxi2 and Pax2 are expressed in the same cells. The Foxi2 expression domain excludes the otic placode at later stages (Fig. 1C,F). Our results lead us to propose two contrasting hypotheses. First, Foxi2-positive cells in the Pax2 domain may be future epidermal and epibranchial cells, not otic placodal cells, and migrate away from the otic region as the otic placode forms. Alternatively, some of the Foxi2-positive cells in the Pax2 domain are future otic placodal cells and Foxi2 expression is down-regulated in these cells as otic placode forms. To test these hypotheses requires lineage analysis, for example by crossing Foxi2-Cre mice with Cre-loxP reporter mouse lines. These experiments are in progress.
Foxi3 Is Expressed in the Surface Ectoderm and the Pharyngeal Endoderm
Mouse Foxi3 is expressed in a broad region of the surface ectoderm adjacent to the neural plate from late E6.5 (not shown) to presomite stages (Fig. 3A–C). The Foxi3 expression domain appears to be very similar to that of Dlx5 (Fig. 3C–F), which is expressed mainly in the anterior surface ectoderm (Yang et al., 1998; Quint et al., 2000). However, the border of the neural plate and surface ectoderm is not clearly defined morphologically at this stage. Double in situ hybridization experiments confirmed that the expression domains of Foxi3 and Dlx5 overlap (not shown), but the resolution of this technique did not allow us to determine whether every single Foxi3-expressing cell also expressed Dlx5. During early somite stages, the ectoderm expression is gradually down-regulated and then restricted to the branchial arches by the 8ss (Fig. 4A). Foxi3 is also expressed in the pharyngeal endoderm. At the late presomite stages, it starts to be expressed in the endoderm adjacent to the prospective cranial region (arrowhead in Fig. 3C inset). At the 8ss, the endoderm expression resolves into two domains between the first and second, and second and third arches, overlying the two domains of the Foxi3 expression in the cranial ectoderm (Fig. 4A). Between E9.5 and E10.5, Foxi3 expression becomes restricted to the region between the maxilla and mandible and the branchial pouches (Fig. 4D,G). Sections of these embryos showed that it is expressed both in the ectoderm and in the endoderm at the pouches where the ectoderm and the endoderm touch (Fig. 4B,C). Foxi3 expression is also detected in the dorsal part of the optic vesicle at late E9.5 to E10.5 (arrowhead in Fig. 4G).
We found mouse Foxi3 is an early marker of the pan-placodal domain in addition to Dlx5 and 6. Since zebrafish foxi1 has been shown to regulate dlx3b, 4b, and 5a in the otic placode (Nissen et al., 2003; Solomon et al., 2003a), mouse Foxi3 may regulate Dlx5 and 6 expression in the pan-placodal domain during early embryonic development. From early somite stages, Foxi3 expression in the pan-placodal domain is down-regulated except for the future branchial pouch region. In the otic region, for example, this down-regulation is correlated with the up-regulation of the otic placode-specific gene markers such as Pax2 and 8. Therefore, it is possible that some of the placode-specific genes down-regulate Foxi3 in the nonbranchial pouch lineages. The restricted expression of Foxi3 both in the ectoderm and in the endoderm of the branchial pouches from E9.5 to E10.5 suggests that Foxi3 may be involved in the pharyngeal organ development such as parathyroid, ultimobranchial body and thymus derived from the branchial pouches (Kaufman and Bard, 1999). Finally, Foxi3 expression in the endoderm from the late presomite stage may be one of the earliest markers of the branchial pouch lineage, and this finding suggests that the specification of future branchial pouches may begin at this early stage.
Although the expression pattern of mouse Foxi3 at the neural plate stage is not similar to any of the zebrafish foxi class genes, it is similar to the Xenopus foxi1c expression with the exception that foxi1c is not expressed in the posterior half of the pan-placodal plate (Pohl et al., 2002). At later stages, mouse Foxi3 is expressed in the branchial pouches in a similar pattern to the zebrafish foxi class genes (Nissen et al., 2003; Solomon et al., 2003a) and Xenopus foxi1c (Pohl et al., 2002). These conserved expression patterns suggest that Foxi3 may play a significant role in the establishment of pan-placodal domain and branchial pouch development.
Foxi1 Expression in the Inner Ear
In addition to the Foxi2 and 3 analysis, we attempted to reproduce the expression pattern of mouse Foxi1 shown by Hulander and colleagues (1998). They showed that mouse Foxi1 is expressed in the entire otic vesicle at E9.5. However, we did not observe any expression from E6.5 to E9.5. Instead, the earliest expression we observed was an intense signal in the lateral side of the endolymphatic duct at E10.5 (Fig. 5). We have obtained similar results with two different Foxi1 probes, including that used by Hulander and colleagues (not shown). We believe the difference in Foxi1 expression reported here is due to differences in hybridization conditions. The inner ears of Foxi1 mutant mice develop morphological abnormalities at mid- to late embryonic stages. These are attributed to defects in the endolymphatic duct (Hulander et al., 2003). These authors attributed the lack of an earlier ear phenotype in the Foxi1 mutants to redundancy. Our present data suggest that because no other Foxi genes are expressed in the inner ear, it is unlikely that they can compensate for the loss of Foxi1. Moreover, the endolymphatic duct phenotype displayed by Foxi1 mutant mice may reflect expression of this gene exclusively in the endolymphatic primordium.
By scanning the mouse genome, we found new gene markers of the pan-placodal domain and the cranial ectoderm, mouse Foxi3 and Foxi2, respectively. Unlike zebrafish foxi1, none of the three mouse Foxi class genes were early, unique markers of the otic placode. However, the similarity of the mouse Foxi3 expression pattern with zebrafish and Xenopus foxi class genes, and the unique expression pattern of the mouse Foxi2 suggest a variety of roles in the early embryonic development of the head and cranial placodes. In the future, these hypotheses will be tested by lineage analysis, and by gain- and loss-of-function experiments in mice.
The genomic DNA sequences of mouse Foxi1, Foxi2, and Foxi3 were retrieved from the Ensembl Web site (www.ensembl.org) and the homology search was done using the TIGR BAC end sequencing database (www.tigr.org) to obtain the information of the BAC clones that include the mouse Foxi1, Foxi2, and Foxi3 genes. BAC clones RPCI-23-328P6 (for Foxi1), 150K14 (for Foxi2), and 315G17 (for Foxi3) are obtained from BACPAC resources at Children′s Hospital Oakland.
Whole-Mount In Situ Hybridization
Whole-mount in situ hybridization was performed by a standard protocol (Myat et al., 1996). Detailed protocols are available upon request. Partial exon2 fragments of mouse Foxi1, Foxi2, and Foxi3 were amplified from the BAC clones described above by high fidelity Pfu polymerase chain reaction (PCR). Primers used for the PCR are 5′-AGATTCATCCTCCAGCACCAG-3′, 5′-GGAAACCAAGCAACGCTTCTC-3′ (Foxi1, 1.7 kb; 1.2-kb region divided by XbaI site was used for probe), 5′-CTGTGCTCCTGGCATCATCAG-3′, 5′-GAGTATGGGTTGACTGTAAGC-3′ (Foxi2, 0.7 kb), and 5′-GGAAGGGTAATTACTGGACTC-3′ and 5′-ATGAGGCTGTTGACCATGCTG-3′ (Foxi3, 0.6 kb). PCR products were subcloned into pGEM-T Easy (Promega) for digoxigenin- or fluorescein-labeled probes. Other probes used in this study were Pax2 (provided bt Greg Dressler), Pax8 (provided by Meinrad Busslinger), Dlx5 (provided by Thomas Lufkin), Krox20 (provided by David Wilkinson), Neurogenin-2 (provided by David Anderson), and NeuroD (provided by Michael Mulhesien). After the color reaction, embryos were washed with phosphate-buffered saline (PBS) containing 1 mM ethylenediaminetetraacetic acid and dehydrated and rehydrated through the diluted series of methanol to reduce the background staining. For double in situ staining, rehydrated embryos were blocked again, incubated with the second antibody (anti-digoxigenin or fluorescein), and washed and incubated with INT/BCIP solution. For the histological analysis, stained embryos were equilibrated in PBS containing 15% sucrose and embedded in 7.5% gelatin (300 bloom, Sigma) and 15% sucrose in PBS as previously described (Groves and Bronner-Fraser, 2000). The 12- to 14-μm sections were collected on Superfrost Plus slides (Fisher). Slides were washed in PBS at 37°C and incubated in 3 μM 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; Molecular Probes) in PBS before being mounted in Fluoromount-G (Southern Biotechnology). Whole embryos and sections were photographed by Zeiss Stemi SV11 Apo and Axiophot2 with an Axiocam digital camera. Brightfield and fluorescent images of the sectioned embryos were taken to show the in situ signal and nuclear DAPI staining.
We thank Suzi Mansour for the discussion of our data. We thank Rebecca Ferreira and her staff for animal maintenance; Juan Llamas, Welly Makmura, and Sheri Juntilla for colony management; and Xiomara Padilla for excellent technical support. We thank all Groves lab members and CMB members for help and advice. A.K.G. is supported by the House Ear Institute and by the National Institutes of Health.