|Home | About | Journals | Submit | Contact Us | Français|
To understand the mechanism by which canonical Wnt signaling sets boundaries for pattern formation in the otic vesicle (OV), we examined Tbx1 and Eya1-Six1 downstream of activated β-catenin. Tbx1, the gene for velo-cardio-facial syndrome/DiGeorge syndrome (VCFS/DGS), is essential for inner ear development where it promotes Bmp4 and Otx1 expression and restricts neurogenesis. Using floxed β-catenin gain-of-function (GOF) and loss-of-function (LOF) alleles, we found Tbx1 expression was downregulated and maintained/enhanced in the two mouse mutants, respectively. Bmp4 was ectopically expressed and Otx1 was lost in β-catenin GOF mutants. Normally, inactivation of Tbx1 causes expanded neurogenesis, but expression of NeuroD was downregulated in β-catenin GOF mutants. To explain this paradox, Eya1 and Six1, genes for branchio-oto-renal (BOR) syndrome were downregulated in the OV of β-catenin GOF mutants independently of Tbx1. Overall, this work helps explain the mechanism by which Wnt signaling modulates transcription factors required for neurogenesis and patterning of the OV.
The inner ear in mammals is an exquisitely complex system comprised of six sensory organs. The vestibular system is located within the dorsal half of the inner ear and is needed for proper balance. It is composed of three semicircular canals with associated sensory cristae, as well as the utricle and saccule, containing sensory maculae. The sixth and most complex sensory organ is the Organ of Corti contained within the cochlea. The coiled cochlea is located in the ventral part of the inner ear and is responsible for hearing. All of the inner ear sensory organs are innervated by the cochleovestibular ganglion (CVG).
The inner ear and CVG are formed from a region of thickened epithelium on the surface of the embryo called the otic placode. The otic placode invaginates to form the otic cup and then the otic vesicle (OV). Once induced by Fibroblast growth factors, Wnts and Notch pathway members, the developing OV expresses molecular markers that will roughly define three axes: anteroposterior (AP), dorsoventral (DV) and mediolateral (ML) (reviewed in Bok et al., 2007). The AP axis is identified coincident with the expression of neurosensory markers as the first step in morphogenesis of the inner ear. Neuroblasts, located in the anteroventral region of the otic cup and later OV, delaminate from the otic epithelium to form the CVG. The proneural gene Neurogenin 1 (Ngn1) is a member of the basic helix-loop-helix (bHLH) transcription factor family and is expressed in cells with both a neural and sensory fate, thereby giving rise to neurons as well as future hair cells and supporting cells of sensory epithelia (Ma et al., 2000, Raft et al., 2007). Another bHLH transcription factor gene, NeuroD, acts downstream of Ngn1 and is expressed in neuroblasts as they begin to delaminate (Raft et al., 2007). Genetic studies have shown that Ngn1 and NeuroD are both required for neurogenesis (Ma et al., 1998; Fritzsch, 2003). Opposite to the neurogenic region is the posterior part of the OV, where Tbx1 is expressed. Tbx1 encodes a T-box transcription factor required for inner ear development (Vitelli et al., 2003; Raft et al., 2004). Haploinsufficiency of TBX1 is responsible for the etiology of a human congenital anomaly disorder termed velo-cardio-facial syndrome/DiGeorge syndrome (VCFS/DGS; MIM # 92430/188400). Many VCFS/DGS patients have hearing loss mostly due to conductive hearing impairment resulting from chronic otitis media, while 10% have sensorineural hearing loss (Reyes et al., 1999; Ford et al., 2000; Vantrappen et al., 1998; 2003). Inactivation of Tbx1 in mice results in a smaller OV and a hypoplastic inner ear lacking both a cochlea and vestibular system (Vitelli et al., 2003; Raft et al., 2004). In contrast to this, the genes regulating the neurogenic domain are duplicated resulting in an enlarged CVG rudiment (Raft et al., 2004). Tbx1 is required to prevent loss of OV sensory cells (Vitelli et al., 2003, Xu et al., 2007) and to repress neurogenesis (Raft et al., 2004). The expression domain of Tbx1 is dynamic and complementary to that of Ngn1 and NeuroD, and acts to refine the AP axis of the OV.
The dorsoventral (DV) axis is the second to form and occurs as a result of signals from surrounding tissues (Bok et al., 2005). In the mouse, the DV axis is defined by morphogen signaling such as the Wnt and Shh pathways. Canonical Wnt signaling from the dorsal neural tube in the region of the future hindbrain activates β-catenin in the OV and is needed for the specification of the dorsal OV that will form the vestibular system (Riccomagno et al., 2005; Bok et al., 2005). Sonic hedgehog (Shh) emanates from the floor plate and notochord, and is required for specification of the ventral OV that will form the cochlea (Riccomagno et al., 2002). When canonical Wnt signaling is constitutively activated in the ectoderm near the future otic placode, it promotes an otic fate leading to an expanded otic placode domain and enlarged OV (Ohyama et al., 2006). In addition, forced activation of Wnt signaling to the otic cup causes an expansion of dorsal OV markers such as Dlx5 and Gbx2 in the enlarged OV (Riccomagno et al., 2005). Antagonism between Wnt and Shh signaling pathways in the OV acts to specify a DV boundary, likely with the help of other signaling molecules (Riccomagno et al., 2002; 2005).
It is possible that genes required to form the DV axis may modulate or restrict expression of those for other axes, to form specific borders, since many genes are not expressed in a simple manner. For example, Wnt and Shh act as DV signals that converge on the medial wall of the OV, for which Tbx1 is expressed in a complementary manner on the lateral wall. Wnt and Shh may help define the complex borders of Tbx1 expression in the OV and Tbx1, in turn, regulates expression of downstream genes. However, expression of Tbx1 in the OV is not altered in Shh loss- or gain-of-function mutant mouse embryos (Riccomagno et al., 2002).
In response to Wnt signaling, β-catenin remains unphosphorylated and shuttles to the nucleus to regulate transcription of downstream genes. Nuclear β-catenin activates transcription of target genes via the binding of TCF/LEF transcription factors. A reporter containing TCF/LEF sites termed TOPGAL, has been used as a readout of canonical Wnt signaling in the OV (Riccomagno et al., 2005). TOGPAL activity in response to Wnt signaling begins closest to the neural tube, in the dorsomedial region of the otic cup and then expands to the entire dorsal region of the OV and beyond (Riccomagno et al., 2005). Previous work showed that Tbx1 mRNA was absent from mutant embryos in which β-catenin was constitutively activated in the OV using the Pax2-Cre allele to drive recombination. There are two possible hypotheses to explain this. The first is that changes in Tbx1 expression are a result of early dorsalization of the OV and concomitant loss of ventral OV cells (Ohyama et al., 2006). The second is that loss of expression is due to specific regulation by canonical Wnt signaling.
In this report, we provide evidence for the second hypothesis. We show that Tbx1 protein has early overlapping expression with β-catenin in the OV and is subsequently restricted in a complementary manner to activated β-catenin. Supporting this hypothesis, we found that constitutive activity and loss of β-catenin results in loss and expansion of Tbx1 expression, respectively, using a different Cre driver than was used in previous studies. The use of the Foxg1-Cre allele demonstrates genetically that β-catenin functions directly in restricting borders of expression in the OV. It was previously shown that constitutively active β-catenin in the presumptive otic placode results in loss of Ngn1, NeuroD and Lunatic Fringe (Lfng) expression and failure to develop CVG neuroblasts (Ohyama et al., 2006). Inactivation of Tbx1 results in a duplicated CVG rudiment (Raft et al., 2004). To explain why neurogenesis was diminished and not expanded when Tbx1 was inactivated in β-catenin GOF mutants, we examined expression of Eya1 and Six1. The Eya1 gene encodes a transcriptional co-activator that acts with Six1 in the ventral OV to promote neurogenesis (Zou et al., 2004). Mutations in both are responsible for a human genetic syndrome named branchio-oto-renal (BOR) syndrome (MIM 113650). Both Eya1 and Six1 were downregulated in β-catenin GOF mutants supporting a model for a pathway in which, independent of Tbx1, β-catenin represses the Eya1-Six1 pathway upstream of neurogenesis. This work provides further insights into how multiple signaling pathways can modulate complex gene expression domains of important human disease genes.
To determine whether β-catenin was co-localized with Tbx1 protein, we performed immunofluorescence studies (Figure 1). Tbx1 protein is initially expressed on the posterolateral wall of the otic cup at E9.25 (Figure 1A) and persists through E10-E10.5 (Figure 1B, C). Activated β-catenin was present in the dorsomedial region as is consistent with TOPGAL reporter expression previously reported (Riccomagno et al., 2005). Minimal overlap in expression of β-catenin and Tbx1 in the ventral OV was detected. On the other hand, there was overlap in the anterodorsal part of the OV where β-catenin and Tbx1 are normally co-localized from E9.25-E10 (Figure 1A, B). By E10.5 (Figure 1C) this colocalization is diminished as the domains are more segregated along the DV axis. These results demonstrate that canonical Wnt signaling is largely complementary to Tbx1 expression in the OV, and suggests that it restricts Tbx1 to the posterolateral OV during OV patterning.
In the absence of Wnt signaling, β-catenin is localized to the cytoplasm where serine and threonine residues encoded by exon 3 of the β-catenin gene (Catnb) are phosphorylated resulting in targeted degradation of the protein. When canonical Wnt signaling is activated in a cell by the binding of Wnt to Frizzled transmembrane receptors, degradation of β-catenin is blocked thus allowing it to translocate to the nucleus. β-catenin can be constitutively activated in CatnbΔEx3/+ mice by removal of exon 3 using the Cre-loxP system, since exon 3 is not functionally required for the protein product (Harada et al., 1999). Previous work showed that constitutively active β-catenin as driven by the Pax2-Cre allele results in an enlarged OV and complete loss of Tbx1 expression by E9.5 (Ohyama et al., 2006), see Supplemental Figure 1. We used the Foxg1-Cre allele to drive recombination of loxP sites because it is active slightly later in time in the presumptive otic placode (Figure 2) (Hebert et al., 2000, Pauley et al., 2006). The rationale was to bypass ectoderm-otic placode decisions and focus on the roles of Wnt signaling in early OV patterning.
To compare the temporal activity of Foxg1-Cre to that of Pax2-Cre, we crossed these mice with a floxed GFP reporter line (RCEEGFP/EGFP) in which expression of GFP is conditionally activated by Cre recombinase and driven by a CAGG sequence together with the endogenous Rosa26 promoter (Sousa et al., 2009, Batista-Brito et al., 2009). At E8.5 (7SS), RCEEGFP/+; Pax2-Cre embryos express GFP in the presumptive otic placode as detected by direct fluorescence (Figure 2A). At the same stage, RCEEGFP/+; Foxg1-Cre embryos do not express GFP in the otic placode, however reporter expression can be detected in other Foxg1-Cre+ tissue such as the pharyngeal endoderm, as expected (Figure 2B, white arrowheads). By E9.25, both Foxg1-Cre and Pax2-Cre drive reporter expression in the otic cup (Figure 2A, B).
CatnbΔEx3/+;Foxg1-Cre gain-of-function mutants, hereafter referred to as BcatGOF mutants, survive to E10.5 and die shortly thereafter with defects in tissues expressing Foxg1 (Figure 3A). Some of the early functions of β-catenin in the pre-otic ectoderm are bypassed as these mutants are able to develop a fully closed OV by E10.5 (Figure 3A). Normally, Tbx1 is strongly expressed in the posterolateral OV at E9.5–E10.25 (Figure 3B, C). In BcatGOF mutants, Tbx1 is expressed at reduced levels in the posterolateral OV at E9.5 and E10 (Figure 3B, C). Expression of Tbx1 is completely lost by E10.25 in the OV of BcatGOF mutants suggesting that ectopic activation of nuclear β-catenin might repress expression of Tbx1 in the OV. The loss of Tbx1 expression at this stage is specific to the OV, as expression is still present in the pharyngeal region, albeit reduced due to morphological defects (Figure 3B, C). Changes in gene expression in the OV may also be in part due to morphological defects. For example, in Figure 3B, there is a slight delay in otic pit closure compared to the control (asterisk) at E9.5 but recovery occurs by E10.25. This phenotype is less severe than that of BcatGOF mutants generated using Pax2-Cre in which the OV is more dysmorphic and remains unclosed at E10.5 (Supplemental Figure 1). Pax2-Cre mutants also fail to initiate Tbx1 expression in the OV due to dorsalization (Ohyama et al., 2006). Therefore, it is unlikely that morphological defects explain the downregulation of Tbx1 at E10.25 in BcatGOF mutants, especially considering that Tbx1 protein is present in the lateral OV at E10 (Figure 3C). When taken together (Figures 1 and and3),3), it is possible that Wnt signaling may normally serve in part to directly restrict expression of Tbx1.
Tbx1 is required for proper expression of Bmp4 and Otx1 in the OV (Raft et al., 2004). The transcription factor Otx1 is required for inner ear development, and loss of Otx1 results in the absence of sensory organs (Morsli et al., 1999; Burton et al., 2004). The expression of Otx1 marks the future site of the lateral vestibular region and cochlea, and its expression partially overlaps with that of Tbx1 (Figure 4A). It was previously shown that Otx1 is downregulated when Tbx1 is inactivated, and the same is observed in BcatGOF mutants (Figure 4A). Changes in expression are consistent with the downregulation of Tbx1 in the OV by ectopically activated β-catenin.
Tbx1 expression in part, coincides with strong Bmp4 expression (Figure 4A). The Bmp4 gene encodes a soluble bone morphogen protein that is a member of the transforming growth factor gene superfamily. It is expressed in specific stripes within the anterior and posterior OV, and is required to form the superior, lateral and posterior cristae (Oh et al., 1996; Morsli et al., 1999) as well as the resulting semicircular canals (Chang et al., 2008). Bmp4 expression is reduced in Tbx1−/− mouse embryos, suggesting that it might act downstream of Tbx1 (Raft et al., 2004, Arnold et al., 2006). In BcatGOF mutants, Bmp4 expression was increased at E10.5 along the dorsal OV (Figure 4A, sections) and is ectopically activated in the ectoderm (Figure 4A, Supplemental Figure 2). Loss of Tbx1 cannot explain the strong ectopic Bmp4 expression in BcatGOF mutants.
Constitutive activation of Wnt signaling in Pax2+ presumptive otic placode ectoderm at E8.5 promotes an otic fate and results in dorsalization of otic tissue which only expresses dorsal markers such as Dlx5, Gbx2, and Pax8 (Ohyama et al., 2006). When canonical Wnt signaling is forcibly activated in otic tissue at a later stage (E9.25) by lithium chloride treatment, it acts to selectively promote expression of the dorsal otic markers Gbx2 and Dlx5 without complete dorsalization as ventral markers Pax2 and Ngn1 are still present (Riccomagno et al., 2005). Therefore, it seems that canonical Wnt signaling has separate and distinct functions in the pre-otic ectoderm versus the OV. To determine if the OV in BcatGOF mutants in our study undergoes dorsalization, we analyzed mRNA expression of dorsal markers Wnt2b, Lmx1a, Dlx5, and Msx1 (Figure 4B) at E10.5. We observed an increase in intensity of dorsal marker expression, however Wnt2b was the only marker whose expression domain was expanded ventrally. Dlx5, Lmx1a, and Msx1 show a disruption of normal expression due to differences in morphology, however they were not expanded ventrally demonstrating that the OV does not undergo dorsalization in BcatGOF mutants. In addition, Lmx1a and Dlx5 in BcatGOF mutants were ectopically activated in other tissues such as the forebrain, pharyngeal region, and ectoderm (Supplemental Figure 2). Therefore, loss of ventral markers such as Tbx1 and NeuroD at E10.5 cannot be attributed to dorsalization of the OV in Foxg1-Cre BcatGOF mutants (Figure 3B, ,6A).6A). Alternatively, there is a limited time frame in which dorsal and ventral genes are subject to patterning by Wnt signaling. If so, we would expect that Tbx1 and NeuroD would be lost at E9.25 when changes in expression of the ventral markers Eya1 and Six1 are evident (Figure 7). Altogether, this supports a distinct role for Wnt signaling in directly regulating specific gene expression patterning of the OV.
Tbx1 mRNA expression in β-catenin loss-of-function mutants was investigated using the Cre-loxP system which results in excision of exons 2–6 of the Catnb gene (Harada et al., 1999). Foxg1-Cre drives loss of β-Catenin mRNA expression in the forebrain, pharyngeal endoderm and head ectoderm (Hebert and McConnell, 2000). At E10.5, Tbx1 is expressed in the posterolateral region of the OV (Figure 5). The OV is much smaller when β-catenin is lost, as driven by Pax2-Cre, due to loss of otic placode specification in the ectoderm together with reduced proliferation and increased apoptosis in the otic cup (Ohyama et al., 2006). Similar results were generated using the Foxg1-Cre allele (mutants called BcatLOF; Figure 5) in which the smaller OV may also be attributed to reduced survival of cells. Tbx1 was strongly expressed in the small OV of BcatLOF mutants at E10.5. To validate our results from the BcatGOF model and support the hypothesis that Tbx1 may be a direct downstream target of Wnt signaling, Tbx1 expression should be upregulated in the OV of BcatLOF mutants. However, accurate quantitative analysis is not possible due to the small amount of otic tissue. Therefore, we can only conclude that lack of active β-catenin or low levels, may either have no effect or promote expression of Tbx1.
Ngn1 functions upstream of NeuroD and is expressed in the proximal cranial ganglia. Loss of Tbx1 results in ectopic expression of Ngn1 and NeuroD in the posterior half of the OV, creating a duplicated CVG rudiment and eventual loss of the mature CVG due to the absence of sensory epithelium (Raft et al., 2004). Constitutive activation of β-catenin, as driven by the Pax2-Cre allele, resulted in failure to initiate NeuroD expression by E9.5 (Ohyama et al., 2006). In Foxg1-Cre BcatGOF mutants, expression of Ngn1 and NeuroD was reduced at E9.5 then lost by E10.5 in the OV and developing CVG (Figure 6A, B). Whole mount immunohistochemistry with an antibody to neurofilament labels all cranial ganglia and confirms loss of proximal ganglia, including the CVG, as well as disorganization of the remaining ganglia in BcatGOF embryos (Figure 6B). This is in contrast to Tbx1−/− embryos, in which anti-neurofilament labels cranial ganglia that appear expanded (Xu et al. 2007). Loss of cranial ganglia neurogenesis indicates that expression of Ngn1 and its downstream targets is altered by constitutively activated β-catenin, either directly or indirectly.
To understand why inactivation of Tbx1 results in reduction of neurogenesis in BcatGOF mutants, we examined expression of Eya1 and Six1. Mutations in EYA1 and its transcription factor partner, SIX1 cause branchio-oto-renal (BOR; MIM 113650) syndrome (Abdelhak et al., 1997; Ruf et al., 2004). The Eya1 gene is expressed in ectodermal placodes and plays key roles in each, including the otic placode (Xu et al., 1999; Zou et al., 2004). In Eya1−/− and Six1−/−mutant mouse embryos, neurogenesis is initiated normally from E8.5-E9 as determined by expression of Ngn1, but then expression of neurogenic markers is attenuated by E10.5 resulting in degeneration of the CVG (Zou et al., 2004, Zheng et al., 2003). This indicates that the genes are not required for initiation of neurogenesis but for maintaining this process or promoting cell survival. NeuroD expression was also lost, concomitant with an increase in apoptosis of the CVG rudiment (Zou et al., 2004). Therefore, neurogenic failure of the CVG in the absence of Eya1 and Six1 is attributed to decreased proliferation and failed survival of neural precursors. Early neurogenesis was also arrested in the epibranchial placodes indicating that Eya1 and Six1 are required for epibranchial placodal neurogenesis. To determine whether Eya1 and Six1 were altered in BcatGOF mutants, we performed WMISH analysis at E9.25 and E9.5 (Figure 7).
As expected at E9.25, Eya1 was strongly expressed in controls in the ventromedial two-thirds of the otic cup but reduced in mutants (Figure 7). Eya1 was widely expressed in the OV at E9.5 in control embryos and again, the expression was reduced in BcatGOF mutant littermates (Figure 7). These results demonstrate that changes in expression of otic markers in response to activated β-catenin can occur as early as E9.25 in Foxg1-Cre BcatGOF mutants and that downregulation of Eya1 precedes changes in Tbx1 and NeuroD expression. The expression pattern and changes of Six1 in BcatGOF mutants were similar to that of Eya1 at these stages (Figure 7). As expected, neurogenesis is initiated regardless of the presence of Eya1 and/or Six1, however the proliferative and survival capabilities of these neural precursors is compromised if there is reduction or loss of Eya1 (Zou et al., 2004). This is also true for the Eya1bor/bor hypomorphic mutants which still express approximately 40% of the normal level of Eya1, indicating the requirement for Eya1 in CVG development is truly dosage-dependent (Friedman et al., 2005, Zou et al., 2008).
In BcatGOF mutants, Eya1 expression was eliminated from the OV by E10.5 (Figure 6), as compared to the control. Expression patterns of NeuroD and Eya1 were compared at E10.5. At this stage, expression of NeuroD was diminished in the cranial ganglia of BcatGOF mutants with complete elimination of the CVG rudiment (Figure 6). Eya1 and Six1 are normally expressed in the endoderm of the pharyngeal pouches as well as the posterior ectoderm of the mandibular arch (Zou et al., 2006). Eya1 was misexpressed in the pharyngeal region and downregulated in the pharyngeal pouches at E10.5 (Figure 6). At this stage, morphology changes were detected in the mutant embryos, which could explain some, but not all of the changes in Eya1 gene expression. Based upon this analysis, constitutively active β-Catenin may repress neurogenesis, in part, via the Eya1-Six1 pathway.
Another way to determine if the Eya1-Six1 pathway may be responsible for loss of neurogenesis is to examine the olfactory placode, a second site of expression of NeuroD that is also expressing Cre recombinase. Expression of all NeuroD, Eya1 and Six1 was eliminated in the olfactory placode of BcatGOF mutants, however, we did not observe characteristic epithelial thickening of the ectoderm (Supplemental Figure 3, bottom). This suggests that constitutively active β-catenin disrupts formation of the olfactory placode and this may be responsible for changes in gene expression due to the loss of proper tissue.
In this report, we showed that Wnt signaling sets complex boundaries for pattern formation in the otic vesicle (OV) by restricting expression of Tbx1, and Eya1-Six1, three genes associated with human congenital malformation syndromes, VCFS/DGS and BOR, respectively. Constitutive activation of β-catenin throughout the OV resulted in loss of Tbx1 expression. Previously, it was demonstrated that Tbx1 directly or indirectly inhibits neurogenesis and in its absence, the CVG rudiment is expanded (Raft et al., 2004). Surprisingly, in BcatGOF mutants in which Tbx1 is downregulated, neurogenesis is diminished resulting in failed development of the CVG along with other cranial ganglia. This demonstrates that canonical Wnt signaling may regulate neurogenesis independent of Tbx1. This can occur, among other possible mechanisms, by altered expression of another gene that may override Tbx1 transcriptional control. One possible gene is Eya1 because it is required for CVG survival. Supporting this, Eya1 and Six1 genes were downregulated in BcatGOF mutants. Based upon these findings, we propose a model shown in Figure 8 to build a genetic pathway downstream of Wnt/β-catenin signaling with respect to Tbx1 and neurogenesis in normal inner ear development.
In the model presented in Figure 8A, Tbx1 is normally expressed dynamically in the posteroventrolateral part of the OV at E10.5. Ngn1 is expressed in the anteroventral region where delamination of neuroblasts occurs to form the CVG. Wnt signaling initially begins in the dorsomedial OV (Riccomagno et al., 2005), a region complementary to Tbx1 expression. Some overlap takes place in the posterodorsal OV (Figure 8A). Based upon existing data, the lowest level of Wnt signaling is present in the region of highest Tbx1 expression. In the model, BcatGOF mutants express constitutively active β-catenin in the Tbx1 expression domain, thus reducing the level of Tbx1 mRNA and protein (Figure 1D) as well as that of Ngn1 (Figure 6). In contrast, loss of responsiveness to Wnt by inactivation of β-catenin results in maintenance or a spread of Tbx1 expression throughout the smaller OV (this report), as well as maintenance of Ngn1 (Ohyama et al., 2006). The phenotype observed in the β-catenin mutants is in part due to early changes in the region of otic competence (GOF = expanded otic placode; LOF = smaller otic placode) and the effect on downstream genes later.
We observed ectopic Bmp4 expression in BcatGOF mutants. It has been proposed that Bmp4 may lie downstream of Tbx1 since its expression is lost in Tbx1−/− null mutants (Raft et al., 2004). At E11, in Tbx1+/− embryos, where morphology of the OV appears normal, Bmp4 was diminished in expression, suggesting that Tbx1 alters cell fate rather than survival of cell populations (Raft et al., 2004). In BcatGOF mutants, Bmp4 is ectopically expressed in the OV despite a loss of Tbx1, therefore Bmp4 may act directly downstream of canonical Wnt signaling (Shu et al., 2005). This is further supported by ectopic activation of Bmp4 in tissues other than the OV such as the ectoderm and pharyngeal region where Tbx1 is either not expressed or unaltered. TCF/LEF sites were found in the Bmp4 promoter and are regulated by Wnt/β-catenin signaling in cell culture (Shu et al., 2005). However, there are no consensus T-box binding sites in the Bmp4 locus thus making it likely that Tbx1 may indirectly regulate Bmp4. This suggests a complex regulation of Bmp4 gene expression, downstream of nuclear β-catenin and Tbx1.
While it is known that Bmp4 is required for development of the sensory cristae and semicircular canals of the vestibular system, the role of Bmp4 in regulating neurogenesis of the CVG has not been described. While Eya1 and Six1 are required for CVG survival, we cannot exclude a possible function for Bmp4 in this process. It has been shown that activated β-catenin can induce Bmp4 expression in neural precursors in which Bmp4 promotes a glial fate at the expense of neurogenesis (Kasai et al., 2005, Bonaguidi et al., 2005). Taken together with the expression pattern of Bmp4 adjacent to the prosensory domain of the OV, it is possible that Bmp4 may directly repress neurogenesis of the CVG and that this may contribute to loss of Ngn1 expression in BcatGOF mutants.
The canonical Wnt signaling pathway activates transcription by binding of β-catenin and TCF/LEF proteins to transcriptional control regions of target genes. There are several possible mechanisms to explain repression of Tbx1 by β-catenin. One possible mechanism would be via direct transcriptional repression by TCF/LEF proteins on TCF/LEF consensus sequences within the Tbx1 locus. The Groucho/TLE family of co-repressors could interfere with the binding of β-catenin to TCF/LEF sites thereby inhibiting activation of expression (Fisher et al., 1998; Range et al., 2005; Arce et al., 2009). Co-expression of these genes in tissues responsive to high levels of Wnt signaling would support this model. In fact, a subset of Groucho genes is expressed in the otic vesicle, at least in Xenopus (Molenaar et al., 2000). In this model, low levels of Wnt signals might activate Tbx1 via the binding of activating complexes onto TCF/LEF sites in the Tbx1 genomic locus. Groucho/TLE family proteins might be enhanced and would then compete with nuclear β-catenin for binding to TCF/LEF proteins. Interestingly, alteration of Groucho/TLE gene expression levels in zebrafish resulted in abnormalities in the OV including misexpression of Eya1 and Six1, two genes in particular (Bajoghli et al., 2005) that are relevant to this study. Thus, a similar mechanism could explain the repression of Eya1 and/or Six1 genes. Whole mount expression analysis of Tle3, the homologue of the Drosophila Groucho gene, in BcatGOF mutants did not reveal any noticeable changes in expression (data not shown), however conclusions were complicated by seemingly ubiquitous expression of Tle3. In addition, there are four Groucho/Tle family members in mice (Bajoghli, 2007) that may interact with up to four members of the TCF/LEF family of co-regulators (Arce et al., 2006). Understanding possible functions of Groucho/Tle and TCF/LEF transcription factors upstream of Tbx1 and Eya1-Six1 will require further detailed expression analysis.
Besides direct effects of nuclear β-catenin, there are alternative possibilities for indirect transcriptional repression downstream of canonical Wnt signaling via activation of a gene(s) that represses Tbx1, Eya1, or Six1 expression. One candidate is the gene Sp5, a Zinc finger transcription factor and member of the very well known Sp1 gene family related to the Drosophila segmentation gene Kruppel (KLF). All Sp1 family members bind to GC-rich promoter elements and recognize the same consensus sequence (Suske, 1999; Black et al., 2001). Recently, it was shown that Sp5 is a direct target gene of Wnt signaling (Lee et al., 2000; Fujimura et al., 2007). Specifically, the proximal promoter of Sp5 contains five TCF/LEF binding sites that mediate direct regulation of its expression (Fujimura et al., 2007). Sp5 is expressed in the dorsomedial OV in the same region for which canonical Wnt signaling is strongest (Harrison et al., 2000; Treichel et a., 2001). Of interest, the founding T-box gene Brachyury contains GC-rich binding sites, and a genetic interaction between Sp5 and Brachyury has been demonstrated in the mouse (Harrison et al., 2000). The Tbx1 locus contains very GC-rich regions, in particular, several CpG islands surround the first coding exon. Putative cis-acting sites for the Sp1 family lie within the Tbx1 enhancer (Brown et al., 2004): a total of 9 sites are predicted in the 3,572 bp mouse regulatory region (chr16:18601154–18604726; Genomatix software). This suggests that Sp5, or another Sp1 co-repressor related family member, may repress Tbx1 downstream of canonical Wnt signaling. Expression analysis of Sp5 by whole mount RNA in situ hybridization shows strong up-regulation of Sp5 in tissues in which β-catenin has been constitutively activated, such as the pharyngeal region, forebrain, and ectoderm (Supplemental Figure 2), however we did not observe up-regulation of Sp5 in the OV. Support for function of Sp5 in the OV would be the presence of inner ear defects in Sp5−/− mouse mutants, however Sp5 knockout mice do not have a lethal phenotype (Harrison et al., 2000), although the inner ear in particular was not examined. The lack of a severe overall phenotype might be due to functional redundancy with other Sp1 family members that could lie downstream of Wnt signaling in the OV (Zhang et al., 2008). The functionality of binding sites for the Sp5 gene or family members within the Tbx1 locus will need to be investigated to determine if repression could be regulated by this mechanism.
Thus far, we have focused on the pathways that may restrict expression of Tbx1. It has not been established what pathways act upstream of Tbx1 to activate its expression in the OV. There have been two other pathways discussed with respect to establishing axes in the OV; the Bmp (Riccomagno et al., 2005) and Notch signaling pathways (Jayasena et al., 2008). Bmp morphogens are expressed in the roof plate of the hindbrain early during development (Bok et al., 2007). Since there are examples in which Bmp and Shh have been shown to oppose one another during development, it was suggested that this is another pathway that could influence OV patterning (Riccomagno et al., 2005). However, if Bmps derive from the dorsal OV, they likely act in cooperation with Wnt signals to restrict Tbx1 expression since it is expressed in the opposite domain. Another is the Notch signaling pathway for which many genes are expressed in the otic cup and OV. Recently, it was shown that the Notch pathway reinforces Wnt signaling, such that there is an apparent reciprocal positive interaction between the two pathways (Jayasena et al., 2008). In the OV, activated Notch1 and the Notch receptor Delta-like 1 (Dll1) are expressed in the neurogenic region of the OV opposite to Tbx1. In Tbx1ΔE5/LacZ null mutants, expression of activated Notch1 and Dll1 are expanded concomitant with the expansion of the neurogenic domain (Xu et al., 2007). While this implies that the Notch pathway lies downstream of Tbx1, there may also be an autoregulatory relationship similar to that which exists between the Wnt and Notch pathways. Based upon this data, it is possible that Notch also serves to restrict the expression of Tbx1.
Another pathway of interest to consider upstream of Tbx1 promoting its expression is the retinoic acid pathway (Roberts et al., 2005; and Bok and Wu, 2008, ARO Abstract). Vitamin A becomes metabolized to active retinoic acid (RA) by three enzymes, Raldh1-Raldh3, and becomes inactivated by the Cyp26a, b and c class enzymes (Vermot et al., 2003, Romand, 2003; Romand et al., 2006a; 2006b). RA levels are tightly regulated because altered dosage of RA results in fetal abnormalities that are similar to craniofacial and cardiac defects that occur in patients with VCFS/DGS (Roberts et al., 2005). Relevant to this study, it has been shown that the RA pathway acts both upstream and downstream of Tbx1 for embryonic development (Roberts et al., 2005; Zhang et al., 2006; Okano et al., 2008; Caterino et al., 2009).
During inner ear development, Raldh and Cyp26 genes are expressed in the OV and the surrounding periotic mesenchyme (POM) (Romand, 2003; Romand et al., 2006b). Loss of Tbx1 in the POM as well as other mesodermal tissues results in an increase in Raldh2 gene expression and downregulation of Cyp26a1, Cyp2b1 and Cyp26c1 genes (Guris et al., 2006). Raldh2 expression is expanded from the posterior pharyngeal apparatus towards the rostral region near the OV and results in an increase of locally active RA as determined using retinoic acid reporter mice (Guris et al., 2006). Inactivation of Tbx1 in the mesoderm using a T-Cre allele results in increased RA levels in the POM (Braunstein et al., 2008; Braunstein et al., 2009). The inner ear is abnormal with a hypoplastic cochlea due to multiple functions of Tbx1 in the POM, however these effects are later in development and not during early OV patterning. Similar results were found using the Mesp1-Cre allele (Xu et al., 2007). Nonetheless, altered levels of RA due to knocking out Tbx1 in the mesoderm did not dramatically alter Tbx1 expression in the OV in the mouse.
Results in this study suggest that the Eya1-Six1 pathway may act downstream of canonical Wnt signaling as well. This could explain the loss of Ngn1 and NeuroD expression in BcatGOF mutant embryos since this pathway is needed to maintain survival of neural precursors. Thus far, the transcriptional regulation upstream of Eya1 has not been completely delineated. However, there is a recent report that described a series of enhancers of Eya1 (Ishihara et al., 2008). One putative enhancer region, termed mEya1-5, had a TCF/LEF site, suggesting that nuclear β-catenin might act directly upstream of Eya1 via the Groucho/TLE co-repressors, as with Tbx1. Additionally, the Drosophila homolog of Eya1, eyes absent, is important for eye specification and has also been reported to be repressed by Wingless, the homolog of Wnt1 (Baonza and Freeman, 2002) as part of a negative feedback loop (Hazelett et al., 1998).
It is also possible that the Eya1-Six1 pathway lies downstream of Shh signaling since these genes are strongly expressed in the ventral OV. However, it has been reported that expression of Six1 in the OV is not regulated by Shh, but depends on unknown signals from the ventral hindbrain (Ozaki et al., 2004). Based upon our results, it is possible that as with Tbx1, low levels of Wnt signaling (and Wnt antagonists) might be required to activate transcription.
Tbx1 in the OV has a dynamic pattern of expression and has major functions for inner ear development. In this report we showed that its expression is modulated or restricted by the canonical Wnt signaling pathway, which simultaneously promotes Bmp4 expression (Figure 8C). High levels of activated β-catenin may repress Tbx1, while loss of Wnt signaling has no effect or promotes expression perhaps by relieving repression by another source. Similarly, the same signaling pathway also represses expression of Eya1 and its downstream gene, Six1, and by doing so, restricts the zone of neurogenesis to the anteroventral OV (Figure 8C). Further work needs to be done to identify positive regulators of Tbx1, Eya1 and Six1.
β-catenin Catnb gain-of-function (BcatGOF), β-catenin floxdel loss-of-function (BcatLOF), and Foxg1-Cre mouse lines have been previously reported and were obtained from Dr. Jean Hebert (Harada et al., 1999, Brault et al., 2001, Hébert and McConnell, 2000). Pax2-Cre mice have also been previously reported and were supplied by Dr. Andrew K. Groves (Ohyama and Groves, 2004). The RCEEGFP/+ reporter mice were a kind gift from Dr. Gordon Fishell at New York University Langone Medical Center (Sousa et al., 2009, Batista-Brito et al., 2009). All mouse lines were maintained in a Swiss Webster (SW, Taconic) background. Genotyping of β-catenin floxdel mice using RM41, RM42, and RM43 primers has been previously described (Brault et al., 2001). Genotyping of β-catenin Catnb mice was done using the following primers: HP-E2 (5′-GCTGCGTGGACAATGGCTACTCAA-3′), HP-E4 (5′-ACGTGTGGCAAGTTCCGCGTCATC-3 ′ ) a n d H P-P G K ( 5 ′-CCACTTGTGTAGCGCCAAGTGCCA-3′). The Foxg1-Cre and Pax2-Cre mice were genotyped using the following primers: EBCreFwd (5′-CAATGCTGTTTCACTGGTTATG-3′) and EBCreRev (5′-CATTGCCCCTGTTTCACTATC-3′). The RCEEGFP/+ reporter mice were genotyped using the following primers: GFP-Fwd (5′-TAAACGGCCACAAGTTCAGC-3′) and GFP-Rev (5′-GAACTCCAGCAGGACCATG-3′) to amplify the targeted allele. The wildtype Rosa26 locus was detected using the following primers: RO1F (5′-GCAATACCTTTCTGGGAGTT-3′) and GFP-wt-R (5′-CAATGCTCTGTCTAGGGGTT-3′). PCR for genotyping was done using the FastStart High Fidelity PCR System (Roche 03 553 361 001). Timed matings were determined by checking for vaginal plugs. The day of vaginal plug was considered to be E0.5, and embryos were staged according to somite number (Kaufmann, 1995). All animals were maintained in a 12 hr dark/12 hr light cycle. Wholemount images of mouse embryos were captured using a Zeiss Discovery V12 stereomicroscope.
Embryos were fixed in 4% paraformaldehyde at 4°C. Fixation times varied according to embryonic stage: 2 hours for E8.5, 3 hours for E9-10.25, and 4 hours for E10.5. Embryos were washed in 0.1M phosphate buffered saline (PBS) and sunk in 30% sucrose in PBS at 4°C overnight. Embryos were embedded in O.C.T. compound (Tissue-Tek) on dry ice and stored at −80°C. Transverse cryosections of the tissue were cut at 20μm. Tissue sections were washed in PBS and permeabilized in PBS/0.5% TritonX-100 for 5 minutes. They were then washed in PBS followed by PBS/0.1% TritonX-100 and blocked in 5% goat serum (Sigma-Aldrich G9023) in PBS/0.1% TritonX-100 for 1 hour at room temperature (RT) followed by incubation with primary antibodies diluted in block for 1 hour at RT. Primary antibodies were as follows: rabbit polyclonal α-Tbx1 (Zymed) 1:500 and mouse monoclonal β-catenin (Sigma-Aldrich C7207, 15B8 ascites fluid) 1:200. Sections were then washed 3 times in PBS/0.1% TritonX-100 and incubated with secondary antibodies diluted in block together with DAPI (1:500) for 1 hour at RT. Secondary antibodies were as follows: Alexa Fluor 568 goat α-rabbit IgG (Invitrogen A-11011) 1:500 and Alexa Fluor 488 goat α-mouse IgG (Invitrogen A-11001) 1:500. Sections were washed 3 times in PBS/0.1% TritonX-100 followed by brief washes in PBS and then water. For dual color immunofluorescence, primary antibodies were incubated on tissue sections at the same time, and secondary antibodies were subsequently incubated on sections at the same time. Slides were mounted in Vectashield hard-set mounting medium (Vector Labs H-1400) and stored at 4°C. Images were captured using a Zeiss Axio Observer.
Imaging of the RCEEGFP/+ reporter was captured by direct GFP fluorescence and did not require antibody staining. Wholemount images of Foxg1-Cre; RCEEGFP/+ embryos were taken immediately following dissection. Embryos were then fixed, embedded and cryosectioned for additional immunochemistry as described above.
Embryos were fixed in 4% paraformaldehyde at 4°C overnight. They were then dehydrated in a series of Methanol/PBS/0.1% Tween-20 dilutions to 100% Methanol and stored at −20°C. Upon rehydration to 0.1%PBS/0.1% Tween-20, in situ hybridization was carried out as previously described (Franco et al., 2001). Anti-sense digoxigenin-labeled RNA probes to Tbx1 (Funke et al., 2001), NeuroD (Lee et al., 1995), Bmp4 (Morsli et al., 1998), Otx1 (Raft et al., 2004), Msx1 (Bok et al., 2007), Lmx1a (Koo et al., 2009), Eya1 and Six1 (Dr. R. Maas, Harvard Medical School) were generated from plasmids using standard protocols. Additional RNA probes templates were generated from amplified E9.5 mouse cDNA using the following primers: Wnt2b-FWD (5′-GGGGAATTAACCCTCACTAAAGGGTGGAATTGCACCACACTGGA-3′), Wnt2b-REV (5′-GGGGTAATACGACTCACTATAGGGGTGTTTCTGCACTCCTTGCA- 3′), Dlx5-FWD (5′-GGGGAATTAACCCTCACTAAAGGGTCAGGAATCGCCAACTTTGC-3′), Dlx5-REV (5′-GGGGTAATACGACTCACTATAGGGGGTGGGAATTGATTGAGCTG-3′), Ngn1-FWD (5′-GGGGAATTAACCCTCACTAAAGGGAACCGCATGCACAACCTCAA-3′), Ngn1-REV (5′-GGGGTAATACGACTCACTATAGGGTAAAGTACCCTCCAGTCCAG-3 ′ ), S p 5-FW D (5′-GGGGAATTAACCCTCACTAAAGGGCAGACTTTTCCACCCTTGGA-3′), and Sp5-REV (5′-GGGGTAATACGACTCACTATAGGGCTCAGCGACTTTGAGCTTCT-3′). PCR-generated RNA probe templates contain T7 RNA polymerase binding sites used to generate antisense probes and T3 RNA polymerase binding sites to generate sense probes. E9.5 NeuroD and E10.5 Bmp4 embryos were sunk in 30% sucrose/PBS overnight at 4°C and embedded in O.C.T. compound (Tissue-Tek) on dry ice and cryosectioned coronal at 16μm. Sections were dehydrated and coverslipped in Permount.
Embryos were fixed in 4:1 methanol/DMSO overnight at 4°C. Endogenous peroxidase activity was blocked in 4:1:1 methanol/DMSO/30% H2O2 for 6 hours at room temperature. Embryos were transferred to 100% methanol and then rehydrated in 50% methanol/PBS for 30 minutes followed by a PBS wash for 30 minutes at room temperature. Nonspecific activity was blocked in 2% milk/0.5% Triton X-100/PBS (PBSMT) for 1 hour then incubated in anti-neurofilament (DSHB mAb2H3 monoclonal, 1:200) diluted in PBSMT overnight at 4°C. Embryos were washed five times for an hour each in PBSMT at 4°C and then incubated with [HRP]-sheep α-mouse IgG (Amersham NA931V, 1:500) diluted in PBSMT overnight at 4°C. Embryos were washed again five times for an hour each in PBSMT at 4°C and then washed for 30 minutes in 0.2% BSA (Sigma A4503)/0.5% Triton X-100/PBS and incubated in DAB substrate with chromagen (DAKO K3466) for 15 minutes. Embryos were then post-fixed in 4% paraformaldehyde overnight at 4°C and dehydrated in a methanol series. Imaging was performed after incubation in 1:2 Benzyl alcohol:Benzyl Benzoate (BABB) at room temperature. Embryos were stored long term in BABB in the dark at room temperature.
Lateral views of RNA in situ hybridization to Tbx1 at E10.5 in BcatGOF mutants generated using Pax2-Cre (A) as compared to Foxg1-Cre (B). Pax2-Cre BcatGOF mutants have fusion of the 1st (I) and 2nd (II) pharyngeal arches as can be detected by continuous Tbx1 expression in the pharyngeal mesoderm (black arrow). In addition, the OV (red dotted outline) is highly dysmorphic due to early activation of β-catenin in the otic placode, and has failed to undergo complete closure of the otic cup by E10.5 (white dotted outline). In comparison, Foxg1-Cre BcatGOF mutants have a fully developed OV (red dotted outline) by E10.25 despite a slight delay in otic cup closure at E9.5 (see Figure 3B, asterisk).
Lateral views of mRNA expression as detected by whole-mount RNA in situ hybridization on control (left) and BcatGOF mutant (right) littermates at E10.5. Bmp4, Dlx5, Sp5, and Msx1 are all known downstream targets of Wnt signaling and exhibit ectopic expression in the forebrain and pharyngeal region of BcatGOF mutants, consistent with the activation domain of Foxg1-Cre. Bmp4 is also ectopically expressed in the OV while Sp5 does not appear to be altered in this tissue. Dlx5 and Msx1 have increased expression in the dorsal OV, but do not undergo complete dorsalization.
High magnification lateral views of embryos with whole-mount RNA in situ hybridization using probes to NeuroD, Eya1 and Six1. All three genes are no longer expressed in the olfactory epithelium of BcatGOF mutants at E9.5 compared to controls. Since it is possible that there may be loss of tissue in this region in BcatGOF mutants, we analyzed cryosections of the embryos to determine if the olfactory placode was present. In the control, there is a clear thickening of the ectoderm (black bracket) that is expressing NeuroD (epithelium is defined between solid and dotted red lines). Development of the olfactory placode fails in BcatGOF mutants where a thickening of the ectoderm cannot be detected.
Grant Sponsor: NIDCD; Grant number: DC005186
We especially thank Dr. Evan Braunstein for preliminary experiments and intellectual advice. We thank Dennis Monks, Raquel Castellanos, Stephania Macchiarulo, and Dr. Silvia Racedo for critically reading the manuscript. We thank Dr. Jean Hebert for providing us with the Foxg1-Cre allele and Dr. Andrew K. Groves for the Pax2-Cre allele. We thank Dr. Gordon Fishell for kindly providing the RCEEGFP/+ reporter mice. We also thank Dr. Jean Hebert and Hunki Paek for providing the β-catenin Catnb (Makoto M. Taketo) mice, β-catenin floxdel mice (Jackson stock no. 004152), and genotyping primers. We thank Dr. Sonja Nowotschin and Dr. Vimla Aggarwal for supplying RNA probes and genotyping primers. We would also like to thank Dr. Doris Wu for the Lmx1a and Msx1 RNA probe plasmids. This work was supported by NIDCD R01, DC005186.