|Home | About | Journals | Submit | Contact Us | Français|
To understand the etiology of congenital hearing loss, a comprehensive understanding of the molecular genetic mechanisms underlying normal ear development is required. We are identifying genes involved in otogenesis, with the longer term goal of studying their mechanisms of action, leading to inner ear induction and patterning. Using Agilent microarrays, we compared the differential expression of a test domain (which consisted of the pre-otic placodal ectoderm with the adjacent hindbrain ectoderm and the underlying mesendodermal tissues) with a rostral control domain (which included tissue that is competent, but not specified, to express inner ear markers in explant assays). We identified 1,261 transcripts differentially expressed between the two domains at a 2-fold or greater change: 463 were upregulated and 798 were downregulated in the test domain. We validated the differential expression of several signaling molecules and transcription factors identified in this array using in situ hybridization. Furthermore, the expression patterns of the validated group of genes from the test domain were explored in detail to determine how the timing of their expression relates to specific events of otic induction and development. In conclusion, we identified a number of novel candidate genes for otic placode induction.
Congenital hearing loss is among the most common birth defects, affecting approximately 3 out of every 1000 infants born each year in the United States. Genetic causes account for about 50% of all congenital hearing loss, but the roles of only a small number of genes directly linked to hearing loss are known. To better understand the mechanisms underlying congenital hearing loss, a clearer understanding of ear development and the genes required for otogenesis are needed.
Through candidate gene approaches and subsequent gene mis-expression experiments, a small number of secreted proteins have been identified in the initiation and early patterning of the otic placode and vesicle, such that a rudimentary signaling network is beginning to be established. The signaling proteins thus far identified in this cascade belong mainly to the FGF, WNT, SHH, and Notch protein families (Ladher et al., 2000, 2005; Vendrell et al., 2000; Adamska et al., 2001; Phillips et al., 2001, 2004; Leger and Brand, 2002; Maroon et al., 2002; Liu et al., 2002, 2003; Riccomagno et al., 2002, 2005; Alvarez et al., 2003; Wright and Mansour, 2003; Ohyama et al., 2006; Sun et al., 2007; Zelarayan et al., 2007; Jayasena et al., 2008; Park and Saint-Jeannet, 2008). Inductive signals from all three germ layers of the gastrulating/neurulating embryo play roles in otic induction. For example, in chick, Ladher and colleagues (2000, 2005) have shown that endodermal FGF8 induces the expression of mesodermal Fgf19. Subsequently, FGF19 induces Wnt8a (formerly called Wnt8c) expression in the hindbrain ectoderm. FGF19 and WNT8A then work together to signal the adjoining pre-placodal ectoderm to form otic tissue. Other studies have shown that Notch signaling also cooperates with the WNT pathway, in this case specifying the size of the otic placode (Jayasena et al., 2008).
Although some signals have been identified in early induction and patterning of the otic epithelium, it is unclear how these signals work in concert to induce the otic placode or what additional signals may exist. For example, FGF19 (from the mesoderm) and WNT8A (from the hindbrain) are both required for otic induction in the chick (Ladher et al., 2000). In contrast, Phillips and colleagues (2004) reported that in zebrafish, hindbrain WNT8 is dispensable in otic induction, but it plays a role in regulating the timing of hindbrain FGF3 expression, a factor required for otic induction in fish. Wright and Mansour (2003) showed that the mesodermal signal in mouse consists of both FGF3 and 10, in contrast to chick, but that the mesodermal inducing signals in both species are themselves induced by FGF8 (Ladher et al., 2005). Other studies have indicated that FGF-signaling is necessary to establish an otic precursor field, but must be subsequently attenuated for otic progression to occur (Freter et al., 2008). It is likely that the induction of the otic placode and its subsequent development involve additional signals and downstream effectors that remain to be identified, and that the exact players differ among species. At present, the candidate gene approach has largely exhausted known candidates. To better elucidate the molecular mechanisms of otogenesis, new candidate genes involved in the initiation and early patterning process need to be identified. To address this goal, we have performed a microarray analysis comparing a test domain (in which both the inducing and responding tissues for otic induction are localized) to a more rostral control domain that includes tissue (surface or non-neural ectoderm) that is competent, but not specified, to form otic tissue in explant assays.
To identify genes that are potentially involved in otic placode induction, we used microarray hybridization techniques to compare the test and control domains. To eliminate or at least to reduce the potential of identifying general placodal competency factors and to restrict our findings to genes that are specific to the ear-forming region, we selected a control domain, in Hamburger and Hamilton (1951) stage 6–7 chick embryos, that includes surface or non-neural ectoderm that is competent, but not specified, to form otic placode in explant assays (Fig. 1A: b; Ladher et al., 2000: area “d”). Using Agilent chick genome microarray chips, we directly compared the test and control domains (Fig. 1A: a, b, respectively), identifying 1,261 transcripts differentially expressed at a change of 2-fold or greater: 463 transcripts were upregulated and 798 transcripts were downregulated in the test domain as compared to the control domain.
Several genes known from previous studies to be involved in otic placode induction and/or early development of the inner ear were upregulated, as expected, in the test domain, including Fgf19, Fgf3, Gbx2, Hoxa1, and Hoxb1, demonstrating the validity of our microarray approach for identifying candidate genes for inner ear development. Also as expected, two additional inducers of otic placode, Fgf8 and Wnt8a, were not differentially expressed in our microarray. Differential expression of Fgf8 was not expected, because previous in situ hybridization studies (Ladher et al., 2005) showed that Fgf8 is expressed in both the test and the control domains. Moreover, we did not expect Wnt8a to be differentially expressed, because our previous in situ hybridization studies (Ladher et al., 2000) showed that it is not expressed in the test domain until stage 8, just after the stages used to collect tissues for microarray analysis.
As a control, we took a subset of embryos in which the test and control domains were removed on one side and processed them for in situ hybridization with Fgf19 (Fig. 1B). To identify the test domain during our dissections of living embryos, we used the first somite as a landmark for the pre-otic caudal boundary, and we micro-dissected a piece of tissue just rostral to the first somite that included the pre-placodal region, adjacent hindbrain, and underlying mesendoderm. The sizes of the test and control isolates were based on our earlier explant study (respectively called area “a” and area “d” in Ladher et al., 2000). Loss of most of the Fgf19 expression domain, after micro-dissection and subsequent in situ hybridization, demonstrated that the correct tissue for the test domain was reliably collected.
Thirty genes with differential expression of 2-fold or greater were selected for validation. Expression patterns were visualized first by in situ hybridization from pre-placode development—or during the stages of otic induction (stages 4–9)—through otic placode formation (stages 10–12). Those genes selected for validation consisted of signaling molecules and transcription factors. We focused on these because the majority of genes involved in otic induction reported in the literature belong to these two classes. Several signaling molecules were differentially expressed, including members of the FGF-, WNT-, and Notch-signaling pathways (Table 1). In addition, a number of transcription factors, including several Hox genes, were also differentially expressed.
In our microarray, three members of the FGF family were upregulated in the test domain: Fgfs3, 4, and 19. One FGF receptor, Fgfr3, was down regulated. Fgf3 was expressed at 4-fold higher, Fgf4 at 2.5-fold higher, and Fgf19 at 55-fold higher levels in the test domain as compared to the control domain. Conversely, Fgfr3 was expressed 4-fold higher in the control domain.
FGF-signaling plays a well-documented role in inner ear development and disease. FGF3 (Wright and Mansour, 2003; Phillips et al., 2004) and FGF19 (Ladher et al., 2000) are known inducers of the otic placode; mutations in Fgf3 cause congenital hearing loss and inner ear agenesis (Tekin et al., 2007, 2008; Alsmadi et al., 2009). Although the various skeletal dysplasia syndromes associated with FGF-receptor mutations are usually associated with conductive hearing loss, Kallmann syndrome 2 is associated with sensorineural hearing loss when caused by mutations in FGFR1 (Dode et al., 2003). Kallmann syndrome with hearing loss has also been reported in association with mutations in FGF8 (Falardeau et al., 2008). Moreover, a specific mutation in FGFR3 causes Muenke syndrome, consisting of coronal craniosynostosis and congenital sensorineural hearing loss (Muenke et al, 1997; Holloway et al., 1998; Lowry et al., 2001). Mutations in the tyrosine kinase domain of FGFR2 and 3 cause lacrimoauriculodentodigital syndrome (LADD), a component finding of which is mixed hearing loss (Milunsky et al., 2006; Rohmann et al., 2006).
Expression patterns of these genes, visualized by in situ hybridization, validated the specificity and sensitivity of our microarray results (Fig. 2A–P). During placode induction stages, Fgfs3, 4, and 19 were all expressed as early as stage 4 in the node and primitive streak (Fig. 2A–C), and expression of all three occurred in the test domain at stages 6–7 (Fig. 2E–G). The first expression of Fgf19 occurred in the mesoderm (Fig. 2G’), whereas Fgf3 and Fgf4 were first expressed in both the mesoderm and the overlying neuroectoderm of the developing caudal hindbrain (Fig. 2E’, F’). Expression of Fgf4 in the test domain was transient and disappeared by stage 9. Shortly after, however (~stage 11), Fgf4 was expressed in the endoderm of the second pharyngeal pouch subjacent to the developing otic placode. Expression of all three FGF ligands persisted at the ear-forming region through stages of early placode formation, and the primitive streak expression of these genes became consolidated into the developing tail bud (Fig. 2M–O). Although Fgf3 and 19 have described roles in otic formation, Fgf4 has not been identified previously as a potential candidate and studies of its potential role in ear induction are underway (Urness et al., in preparation).
In contrast to the three FGF ligands that were upregulated, the expression of Fgfr3 was downregulated in the test domain as compared to the control domain. Fgfr3 was the only FGF receptor that was identified in our microarray as differentially expressed between the test and control domains at stages 6–7. Fgfr3 was expressed only in the rostral half of the embryo at stages 4–5 (Fig. 2D), and it persisted in the most rostral portion of the embryo throughout stages 6–7 (Fig. 2H). From stages 6–9, Fgfr3 was expressed throughout the neural plate, extending to the level of the developing first somite. Fgfr3 expression was much weaker just rostral to the first somite, namely, in the region of the test domain, than in the adjoining regions (Fig. 2H, L, arrows). During stages of early placode formation, Fgfr3 was broadly expressed in the neural tube adjacent to the ear-forming region, extending both rostrally and caudally (Fig. 2P). Hindbrain expression of Fgfr3 at late otic placode induction and early placode formation stages could play a role in the induction and/or maintenance of the otic placode via potential effects on hindbrain specification. As discussed earlier, the hindbrain secretes growth factors, such as Wnt8a, that are important for otic induction, in responds to FGF signaling by the adjacent head mesoderm of the ear-forming region.
In our microarray, several members of the Wnt pathway were upregulated in the test domain, including both ligands and receptors. Two ligands, Wnt5b and Ndp—the latter, a non-traditional Wnt-signaling molecule—and one Wnt receptor, Frizzled (Fzd)10, were upregulated in the test domain; another Wnt receptor, Fzd8, was downregulated. Wnt5b was expressed at 3.5-fold higher, Ndp was expressed at 19-fold higher, and Fzd10 was expressed at 5-fold higher in the test domain, whereas Fzd8 was expressed at 11.5-fold higher in the control domain. Like the FGF pathway just described, the WNT-signaling pathway is also involved in otic placode induction and patterning of the inner ear (Ladher et al., 2000; Ohyama et al., 2006). Ndp is the gene responsible for Norrie disease (oculoacousticocerebral dysplasia), a syndrome that includes progressive hearing loss of cochlear origin with onset in late childhood as a component finding (Warburg, 1966; Berger, 1992).
In situ hybridization of these genes validated our microarray results (Fig. 3A–T). At stages 4–5, Wnt5b was expressed throughout the primitive streak and extended laterally in ingressing cells (Fig. 3A). Through stages 6–7, expression was maintained in the caudal two-thirds of the embryo, with its rostral boundary located near the level of the test domain (Fig. 3F). By stage 8, discrete expression occurred in bilateral patches (Fig. 3K, arrows) just rostral to the first somite, corresponding to the test domain. In addition, expression of Wnt5b persisted throughout the regressing primitive streak and in the segmented somites (Fig. 3K). By placodal stages, Wnt5b expression was restricted to the caudal-most portion of the embryo in the forming tail bud and flanking mesoderm (Fig. 3P).
At stages 4–5 Ndp was expressed in the caudal half of the primitive streak (Fig. 3B). By stages 6–7, Ndp was expressed in discrete bilateral patches of the neural plate ectoderm in the test domain, extending caudally to overlie the forming somites (Fig. 3G, G’). As development progressed, this expression extended caudally throughout the area of the neural plate overlying the segmenting somites (Fig. 3L). By placodal stages, Ndp was expressed throughout the neural tube medial to the somites, and in rhombomere 4, adjacent to the otic placode (Fig. 3Q: arrowheads).
Ndp expression was also studied at post-placodal (i.e., otocyst) stages to ask whether its pattern of expression might suggest a potential role in inner ear development. In whole mounts, Ndp appeared to be expressed at these stages throughout most of the length of the neural tube, as well as in the otic vesicle (Fig. 4A–C). However, sections revealed that much of the neural tube staining was trapping, except for localized regions of weak expression in the neuroepithelium, presumably marking cell columns (Fig. 4C’: arrows), and ventral neural tube around the floor plate (Fig. 4C’, C”). Similarly, staining in the otocyst largely consisted of trapping (Fig. 4C’). By stage 22, however, Ndp was robustly expressed in the dorsal region of the developing mesonephric kidneys (Fig. 4C”).
Although Fzd4 was not differentially expressed in our microarray, we looked at its expression pattern as well, because it is reported to be the only receptor that binds with Ndp (Smallwood et al., 2007). Furthermore, progressive hearing loss was observed in a Fzd4 null mouse (Wang et al., 2001). Fzd4 was expressed, although faintly, in bilateral mesodermal patches within the test domain (Fig. 3H, H’, M). The rostral end of the embryo, corresponding to our control domain, also lightly expressed Fzd4, which accounts for it not being detected as differentially expressed in our microarray. In placodal stages, Fzd4 was expressed throughout the head, including the level of the ear-forming region (Fig. 3R). Thus, within the ear-forming region there is overlap in the pattern of expression of both the Wnt ligand and its receptor (cf., Figs. 3Q, R), suggesting a potential role NDP signaling in early otic development.
Fzd10 was expressed at stages 4–5 throughout the primitive streak including Hensen’s node; to a lesser degree, expression of Fzd10 in ingressing cells also flanked the primitive streak laterally (Fig. 3D). By stages 6–7, Fzd10 was also expressed slightly more rostrally, extending up into the region of the test domain (Fig. 3I). Sections revealed that expression of Fzd10 was restricted to the epiblast (Fig. 3I’). Slightly later, at stage 8, Fzd10 was expressed mainly in the caudal portion of the embryo. However, discrete bilateral patches of expression occurred in the hindbrain portion of the test domain (Fig. 3N: arrows). The expression of Fzd10 persisted in the hindbrain into the placodal stages of development, but also extended rostrally into the developing midbrain, and caudally throughout the length of the spinal cord and into the tail bud and flanking mesoderm (Fig. 3S).
Expression of Fzd10 in the neural tube persisted through post-placodal stages and extended rostrally from the tail bud and flanking mesoderm, and throughout the spinal cord, up to the caudal limit of the forebrain (Fig. 4D, E). With formation of the hindlimb buds, expression extended into the hindlimb mesenchyme (Fig. 4E). Sections revealed that Fzd10 was expressed in the dorsal two-thirds of the neural tube and was not expressed in the otic vesicle (Fig. 4E’).
For comparison we visualized the expression of Fzd8, which was downregulated in our microarrays in the test domain as compared to the control domain. Consistent with our microarray results, Fzd8 was strongly expressed in the rostral midline of the embryo as early as stage 5, but was completely absent more caudally during the stages we used for our microarrays (Fig. 3E, J). Expression persisted in the developing head fold at stages 8–9 and extended into the forebrain, midbrain, and hindbrain regions (Fig. 3O). Fzd8 was expressed throughout the brain at the placode stage, but expression was considerably weaker than at earlier stages (Fig. 3T).
In our microarray, three genes of the Notch-signaling pathway were differentially expressed in the test domain. All three were expressed throughout stages of induction. One of these, Jagged1 (Jag1), is a ligand of the Notch receptor and was expressed 2-fold higher in the test domain as compared to the control domain. Mutations in JAG1 cause Alagille syndrome, but have also been associated with nonsyndromic familial deafness and congenital heart defects (Le Caignec et al., 2002). JAG1 is required for normal sensory development of the inner ear (Kiernan et al., 2001). The other two Notch-signaling genes identified in this study were Hairy and Enhancer-of-Split (Hes)5, expressed at 9-fold higher, and Hes6, expressed at 19-fold higher, in the test domain as compared to the control domain. The Hes transcription factors are downstream targets of Notch-signaling. Hes5 and Hes6 have both been implicated in sensory development of the inner ear at later stages in otogenesis: Hes5 is required for inner hair cell formation (Zine et al., 2001), whereas Hes6 is an inner ear marker that delineates the inner hair cells (Qian et al., 2006), although its role in their development is unknown.
In situ hybridization analysis of these genes validated our microarray results (Fig. 5A–L). At stages 4–5, Jag1 was expressed throughout the primitive streak and adjacent areas (Fig. 5A). At stages 6–7, Jag1 was expressed bilaterally in the test domain and extended caudally to either side of Hensen’s node, where it was expressed in both neuroectoderm and more lateral mesoderm (Fig. 5D, D’). Two additional bilateral patches of expression occurred in the developing heart field (Fig. 5D, G). During later otic induction and placodal stages, Jag1 was restricted mainly to the developing heart tube (Fig. 5G, J).
Jag1 expression was maintained at post-placodal stages. At stage 14, Jag1 was expressed in many other areas including the atria and ventricles of the heart, lens vesicles, mesonephric kidneys, spinal ganglia, mesencephalon, hindbrain, rostral spinal cord, and otic vesicles (Fig. 5M, M’). At stage 14, Jag1 was expressed throughout the entire extent of each otic vesicle, whereas by stage 22 it expression was restricted to patchy ventral domains marking the vesicle’s rostral, caudal, medial, and lateral walls (Fig. 5N, N”). Moreover, expression in the neural tube at most rostrocaudal levels was restricted to discrete cell columns (Fig. 5N, N’, N”).
Hes5 was not expressed at stage 4, but Hes6 was expressed at this time in the rostral half of the primitive streak (Fig. 5B, C). However, by stages 6–7, Hes5 and Hes6 were both expressed in bilateral patches in the test domain and in Hensen’s node (Fig. 5E, F). In the test domain, expression of both was restricted to the neuroectoderm (Fig. 5E’, F’). At placodal stages, Hes5 was expressed throughout a large rostrocaudal extent of the neural tube, extending rostrally throughout the developing cephalic region (Fig. 5K). This expression extended throughout the developing floor plate but was excluded from the developing roof plate (Fig. 5K’). In addition to neural tube expression, Hes5 was expressed in bilateral patches near the otic placodes (Fig. 5K: arrows). Sections revealed that Hes5 was expressed in the pharyngeal pouch endoderm subjacent to the otic placodes (Fig. 5K’). Hes6 expression was more restricted, with Hes6 being expressed medial to the somites, in the caudal hindbrain/rostral spinal cord, and in the metencephalon up to the midbrain boundary (Fig. 5L). To date there is no known role for Hes5 or 6 in otic induction.
In our microarray, several of the Hox genes, particularly the Hoxa gene family, were upregulated in the test domain. Four Hoxa genes (Hoxa1–4) were identified in our microarray data, along with Hoxb1. In the test domain, Hoxa1 was expressed at 6-fold higher, Hoxa2 was expressed at 8.5-fold higher, Hoxa3 at 3.5-fold higher, Hoxa4 at 3-fold higher, and Hoxb1 at 36-fold higher than in the control domain.
In situ hybridization of these genes validated our microarray data (Fig. 6A–P). At stages 4–5, Hoxa1–3, and Hoxb1 were all expressed in the primitive streak, with Hoxa members being restricted to the caudal half or less of the primitive streak (Fig. 6A–D; Hoxa4 was not visualized by in situ hybridization in this study). At stages 6–7, Hoxa1 was expressed more rostrally than the other Hoxa genes, completely overlapping the test domain, whereas Hoxa2 and 3 overlapped approximately the caudal half of the test region (Fig. 6E–G). Hoxa1 and Hoxb1 were expressed only in the neuroectoderm, whereas Hoxa2 and 3 were expressed in the both the neuroectoderm and underlying mesoderm (Fig. 6E’–H’). Hoxb1 is of particular interest because it is expressed in two discrete patches (Fig. 6H, arrow) in the test region that are separated from the broad caudal expression that extends only as far rostrally as the caudal border of the first somites. Furthermore, Gavalas and colleagues (1998) demonstrated that Hoxa1 and Hoxb1 act synergistically in hindbrain formation, which subsequently provides signals for inner ear development, as we discussed above. They found that Hoxa1 null mice had cochlear and vestibular malformations, whereas in double mutant mice in which both Hoxa1 and Hoxb1 were disrupted the entire inner ear was absent. Recessive loss of human HOXA1 causes Athabaskan brainstem dysgenesis syndrome with associated sensorineural deafness due to inner ear malformations, including bilateral absence of the cochlea (Erickson 1999; Tischfield et al., 2005). Mutations in human HOXA2 cause microtia, mixed hearing loss, and cleft palate (Alasti et al., 2008). From stages 8–12, all 4 Hox genes studied here were expressed throughout the ear-forming level of the neuraxis (Fig. 6I–P).
Six other transcription factors differentially expressed in our microarrays were chosen for in situ hybridization studies. Five of those (Gbx2, Meox1, Nkx6.2, Tbx22, and Tshz3) were upregulated in the test domain, and one (Dlx5) was downregulated, as compared to the control domain. Gbx2 was expressed in the test domain at 5-fold higher, Meox1 at 23-fold higher, Nkx6.2 at 13-fold higher, Tbx22 at 2.5-fold higher, Tshz3 at 5.5-fold higher, and Dlx5 at 3-fold higher than in the control domain.
In situ hybridization of these genes validated our microarray data (Fig. 7A–Y). With the exception of Tbx22 and Tshz3 at stages 4–5 (Fig. D, E), all of the other transcription factors were expressed during all placode induction (Fig. 7A–C, F–R) and placodal (Fig. 7S–Y) stages.
Gbx2 expression has previously been described in early chick development (Shamim and Mason, 1998). In summary, Gbx2 was expressed in the test domain neuroectoderm and adjacent medial non-neural ectoderm (Fig. 7G’) and along the cardiac crescent and subsequent heart tube during the stages of induction (Fig. 7G, M). Gbx2 expression persisted at the level of the otic placode (Fig. 7S; Shamim and Mason, 1998). Gbx2 expression is required for normal endolymphatic duct formation in both chick (Miyazaki et al., 2006) and mouse (Lin et al., 2005). Whether the otic placode itself was abnormal at earlier stages in mutants was not reported.
Meox1, originally called Gmox1, was expressed bilateral to the node as early as stages 4–5 (Fig. 7B). At stages 6–7, the expression of Meox1 extended caudally from the test domain through the first somite (Fig. 7H). At this time, expression in the test domain was restricted to the mesoderm (Fig. 7H’). As development proceeded, Meox1 expression extended caudally down the length of the embryo, persisting in the developing somites (Fig. 7N, T).
Nkx6.2 was expressed in faint bilateral patches flanking the node as early as stage 5 (Fig. 7C). At stages 6–7, its expression extended through the test domain from the developing first somite to about the level of the midbrain (Fig. 7I). Expression within the test domain was restricted to the neuroectoderm (Fig. 7I’). By stages 8–9, and into placodal stages, Nkx6.2 expression extended more rostrally along the rostrocaudal axis and was found mainly in the developing mid- and hindbrain regions (Fig. 7O, U), and transiently in the rostral spinal cord (Fig. 7O).
The expression of Nkx6.2 throughout the rostrocaudal extent of the neural tube persisted into post-placodal stages of development (Fig. 8A, B). At stage 15, Nkx6.2 expression was restricted to the ventral midline of the neural tube, from the spinal cord extending rostrally to the isthmus at the juncture of the metencephalon and mesencephalon (Fig. 8A, A’, A”). This expression pattern persisted through stage 18 (Fig. 8B). Nkx6.2 was not expressed in the otic vesicles, although Nkx6.2 was expressed in the ventral portion of the hindbrain adjacent to the vesicles (Fig. 8A”).
Tbx22, a member of the T-box family of transcription factors, was first expressed at stages 6–7, bilaterally in small faint patches within the test domain (Fig. 7J). Sections revealed that this expression was restricted to the mesoderm (Fig. 7J’). At stages 8–9, expression was more pronounced and expanded rostrally and caudally, extending throughout the caudal forebrain and mid- and hindbrain regions, and into the spinal cord, as well as being expressed in the somites (Fig. 7P). Later, at stages 10–11, expression persisted in the somites, but was absent from the mid- and hindbrain regions (Fig. 7V). However, Tbx22 was expressed in midline rostral head mesenchyme at stage 10, just rostral to the tip of the forebrain (Fig. 7V; inset).
During post-placodal stages, expression of Tbx22 became restricted to two areas: the somites and bilateral patches of ventral head mesenchyme (Fig. 8C–D). At stage 14, Tbx22-expressing cells flanked the floor of the developing midbrain as two distinct stripes (Fig. 8C-C’). By stage 16, Tbx22-expressing cells were more broadly distributed ventral and lateral to the mesencephalon and diencephalon, surrounding the developing pituitary rudiments (Fig. 8D, inset). Tbx22 expression persisted in the developing somites throughout post-placodal stages, at least until stage 22 (not shown), the latest stage studied.
Loss of function mutations in TBX22 cause X-linked submucous cleft palate with associated conductive hearing loss (Braybrook et al., 2001), but defects of the inner ear have not been reported.
The Teashirt zinc-finger 3 gene (Tshz3) was first expressed at stages 6–7 (Fig. 7K) in the primitive streak and in discrete bilateral patches in the neuroectoderm of the test domain (Fig. 7K’). By stages 8–9, Tshz3 ectodermal expression was restricted largely to the neural folds, except for a broader area of neuroectodermal expression of the level of the developing hindbrain/rostral spinal cord (Fig. 7Q). At stage 12 Tshz3 was expressed in the neural tube adjacent to the somites, with a bilateral patch of expression in rhombomere 4 (Fig.7W: arrowheads).
Tshz3 expression persisted into post-placodal stages. At stage 17, Tshz3 was expressed in a striped pattern along the anterior-posterior axis of the body. Tshz3 was expressed throughout the length of the neural tube, extending rostrally up into the caudal hindbrain, adjacent to the otic vesicles (Fig. 8E, inset). Laterally and parallel to the neural tube, Tshz3 was expressed in two bands, one on each side of the embryo, in the mesoderm flanking the developing gut tube (Fig. 8E’). By stage 21, Tshz3 was expressed throughout much of the embryo, including the limb buds and neural tube (Fig. 8F). Sections through the otic vesicles revealed that Tshz3 was expressed in the anterior-most portion of the otic vesicle (Fig. 8F’) and in localized regions in the ventral neural tube where neurites accumulate. The region of expression in the otocyst likely coincides with the position of the developing statoacoustic ganglion and its projections.
Tshz3 is a good candidate for a gene involved in otic induction and patterning. Tshz3 is a known down-stream target of FGF8 signaling in tendon development (Manfroid et al., 2006). This is of interest because FGF8 is a known inducer of otic development (Ladher et al., 2005), suggesting that Tshz3 may also be a target of FGF8 signaling during otic induction by endodermal FGF8.
Shown for comparison, Dlx5 was downregulated in the test domain as compared to the control domain. Dlx5 was expressed as early as stages 4–5 in the rostral portion of the developing embryo (Fig. 7F). At stages 6–7, Dlx5 was expressed in an oval-shaped region overlying the lateral plate mesoderm, just lateral to the developing neural plate and primitive streak (Fig. 7L). At this time throughout the head region, including within the test domain, Dlx5 was expressed at the lateral edge of the neural plate and in the adjacent non-neural ectoderm, as well as in the underlying mesoderm (Fig. 7L’). Based on its early ectodermal expression, Dlx5 (and other Dlx members) have been used as markers of the common pre-placodal domain (e.g., Brugmann and Moody, 2005; Streit, 2007). Although Dlx5 was downregulated in the test domain as compared to the control domain, it was still expressed there, where it likely plays a role in otic development. By stage 8–9, Dlx5 was expressed in the cranial portion of the embryo, in the heart tube, and in the developing tail bud and flanking mesoderm (Fig. 7R). Subsequently, it was expressed in several areas, most notably for the topic of this paper, in the developing eyes and otocyst (Fig. 7Y: arrowheads; Brown et al., 2005). Targeted disruption of Dlx5 in mice resulted in craniofacial and inner ear defects (Acampora et al., 1999; Depew et al., 1999). In addition, disruption of both Dlx5 and Dlx6 also resulted in limb malformations that phenocopied Split hand/foot and mouth syndrome (SHFM) in humans (Robledo et al., 2002). The Dlx5 and 6 genes are located in the SHFM type 1 locus on human 7q21, and approximately 35% of those with SHFM1 also have sensorineural, or mixed hearing loss (Elliott and Evans, 2006; Saitsu et al., 2009).
In conclusion, we have identified a number of novel candidate genes for otic placode induction. Although several of these candidate genes have roles at later stages of inner ear development, to our knowledge, this is the first time that the expression of most of them has been reported prior to otic placode formation. Several of the genes reported here had discrete bilateral patches of expression specific to the test domain and some surrounding tissues, indicative of potential roles in otic placode induction. Furthermore, a number of these genes have been associated with human or murine inner ear malformations and hearing loss.
Fertilized White Leghorn chicken eggs were obtained from Utah State University (Logan, UT, USA). They were incubated in humidified chambers at 38°C until embryos reached the desired stages, using the criteria of Hamburger and Hamilton (1951).
For differential gene expression analysis by microarray, we collected from stage 6–7 chick embryos two domains, each consisting of all three germ layers: a test domain, and a more rostral control domain (Fig. 1A). The test domain, located just rostral to the first somite, consisted of the pre-otic placodal ectoderm with the adjacent hindbrain ectoderm and the underlying mesendodermal tissues. The control domain, located near the rostral end of the embryo, included tissue (surface or non-neural ectoderm) that is competent but not specified to express inner ear markers in explant assays (e.g., Ladher et al., 2000). Tissue isolates consisted of all three germ layers because previous studies showed that signals for otic placode induction arise from the endoderm, mesoderm, and hindbrain neuroectoderm of the ear-forming region (Ladher et al., 2000, 2005). Bilateral dissections from seven embryos were pooled for each (of four) biological replicates. RNA was isolated as described by Eisenhoffer et al. (2008) for micro-amounts of tissue. After microdissection of the test and control domains in a subset of embryos, we visualized Fgf19 expression, one of the earliest expressed otic-inducing genes, to confirm our ability to reliably collect the test domain.
Cy-3 and Cy-5 fluorescently labeled cRNA was synthesized and microarray hybridizations were performed at the University of Utah Microarray Core Facility. Labeled cRNA was generated using the Agilent Two-Color Low RNA Input Linear Amplification Kit according to the manufacturer’s protocol. Briefly, cDNA was synthesized from total RNA with MMLV-RT and dT/T7 RNA polymerase promoter oligonucleotide sequences. Labeled cRNA was then generated using T7 RNA polymerase and dye-labeled nucleotides. The cRNA was purified using the Qiagen RNeasy mini kit. The microarray analysis was performed using the Agilent 4×44k chicken genome array according to microarray methods established by Agilent. The raw data files were uploaded to the GEO Database and are accessible through GEO Series accession number GSE16918 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE16918).
Microarray data was analyzed with the web-based GeneSifter® Analysis Edition (www.genesifter.net/web/) analysis software. Using this software, we performed a basic t-test that included: a quality cutoff value of 1, correction according to Benjamini and Hochberg, a minimum threshold limit of a 2-fold, and log2 transformation of the data. Our results included are presented based on the adjusted p-value assigned by the Benjamini and Hochberg correction.
We performed whole mount in situ hybridization on stage 4–22 chick embryos as previously described (Chapman et al., 2002) using the following primers for digoxigenin probe design:
We thank Dr. Victoria Prince for the Hoxa1 probe and Dr. Katherine Yutzey for the Hoxa3 probe. We also wish to thank Dr. Brian Dalley, Brett Milash, and the University of Utah Microarray Core facility for their assistance in hybridization and processing of the arrays. This work was funded in part by the Deafness Research Foundation and the National Institutes of Health (DC004185, DK066445, and DC009236).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.