The inner ear develops from an ectodermal thickening, the otic placode, into a complex structure that is asymmetrical along all three axes (Bok et al., 2007a; Groves and Fekete, 2012). Unlike most vertebrate sensory structures, the inner ear has a relatively small percentage of its surface area devoted to sensory cells. The mechanosensory hair cells are housed in precisely positioned sensory patches or organs within an enclosed nonsensory epithelium (Torres and Giraldez, 1998; Fritzsch et al., 2002). Both sensory and nonsensory tissues display axial asymmetries. Embryological manipulations, mostly in amphibians but also in chicken, have shown that the anterior to posterior (A-P) axis is specified before the dorsal to ventral (D-V) axis (Yntema, 1955; Wu et al., 1998; Whitfield and Hammond, 2007).
Recent studies have provided a better understanding of the molecules involved in axial patterning, a separable and later event than otic placode induction (Bok et al., 2007a, 2011; Whitfield and Hammond, 2007; Groves and Fekete, 2012). Even though these events are separable, some of the same cell signaling molecules are involved in both developmental processes (Barald and Kelley, 2004; Ohyama et al., 2007). Patterning of the dorsal to ventral axis is well understood in mice, with Wnt signaling regulating dorsal fates while hedgehog (Hh) signaling (specifically Sonic Hedgehog, Shh) specifies ventral fates (Riccomagno et al., 2002, 2005; Brown and Epstein, 2011). These same genes may also play a role in the axial patterning of inner ears in other organisms. In Xenopus, the depletion of Sox9 that mainly affects dorsal otic structures also results in the loss of dorsal Wnt expression (Park and Saint-Jeannet, 2010). In chicken, loss of Shh activity leads to morphological defects similar to those seen in mouse (Bok et al., 2005). The need to inhibit Hh signaling for normal dorsal pattering seems to be a conserved feature of inner ear development, occurring in mice, chickens, and zebrafish (Bok et al., 2007b; Hammond et al., 2010).
In contrast, molecules involved in patterning the anterior to posterior axis appear to vary across vertebrate species. In zebrafish, Hh signaling is necessary and sufficient for posterior patterning while Fibroblast growth factor (Fgf) signaling is necessary and sufficient for anterior patterning (Hammond et al., 2003; Hammond and Whitfield, 2011). We have previously shown that in Xenopus laevis, as in zebrafish, Hh signaling is necessary for posterior patterning (Waldman et al., 2007). It is still not known which molecules are necessary for anterior patterning in Xenopus. In mice and chickens a gradient of retinoic acid (RA) confers anterior to posterior identity (Bok et al., 2011). Brief exposure to RA results in anterior structures while a prolonged exposure results in posterior structures. The downstream targets of RA signaling that specifically confer anterior or posterior identity are still unknown in amniotes. In zebrafish, RA has been shown to be involved in positioning the boundary between neurogenic and nonneurogenic regions of the developing inner ear (Radosevic et al., 2011).
A characteristic phenotype seen after molecules involved in posterior patterning are lost or inhibited is the formation of mirror symmetric anterior half duplications (Whitfield and Hammond, 2007). We discovered an embryological manipulation, removal of the posterior half of the otic placode, that results in even higher percentages of mirror anterior duplications than most of the genetic and molecular manipulations (Waldman et al., 2007). Cell/tissue ablation experiments provide a powerful tool for studying developmental and physiological processes, while growth replacing the ablated cells or tissues can provide insights into regenerative potential (Curado et al., 2007, 2008). Our partial ablations along the A-P axis revealed that the developing inner ear can regenerate lost structures at placode or otocyst stages, although the regenerated tissue often has an incorrect identity (Waldman et al., 2007).
In this study we decided to explore the effects of partial ablations along the D-V axis and investigate the molecules involved in this patterning. Ablation of the dorsal or ventral half resulted in severe loss of structures, such as reductions in the number of sensory organs, and, unlike partial ablations along the A-P axis, normal ears were not observed. Abnormal ears resulting from dorsal half ablations were different from those seen after ventral half ablation with dorsal ablations having a more severe phenotype, in regards to lost structures, than ventral half ablations. To determine when the D-V axis in the inner ear of Xenopus is specified we did rotation experiments inverting this axis at various placode and otocyst stages. The timing of D-V axis specification is just before the beginning of expression of Wnt3a mRNA in the dorsal otic placode (Wolda et al., 1993). This stage is at a point in inner ear development defined as late placode or early otocyst, (Xenopus stage 24–26) (Nieuwkoop and Faber, 1967; Hausen and Riebesell, 1991; Schlosser and Northcutt, 2000; Waldman et al., 2007). Given that Wnt signaling is important for dorsal patterning in mice and the start of Wnt3a expression is just after D-V axis specification, we decided to study the effects of blocking Wnt signaling on inner ear patterning. We used a known inhibitor of canonical Wnt signaling in Xenopus embryos, the well characterized morpholino to β-catenin, a crucial component for controlling Wnt target genes (Heasman et al., 2000; Logan and Nusse, 2004; Clevers and Nusse, 2012). Targeting of the morpholino to the developing inner ear and surrounding tissues resulted in a range of abnormally patterned inner ears that could be placed in one of three categories of decreasing severity, most of which resembled phenotypes seen after dorsal ablations. To specifically test whether Wnt3a alone was sufficient for D-V patterning, we placed beads soaked with Wnt3a protein at the location of the dorsal or ventral half ablation. We found that Wnt3a protein could almost completely rescue the loss of structures seen after dorsal but not ventral half ablation.
Partial Ablations Along the D-V Axis
Previous studies in Xenopus, physically removing portions of the otic placode along the A-P axis, discovered that half ablations resulted in enantiomorphic (mirror duplicated) inner ears (Waldman et al., 2007). To understand if there were differences in regenerative potential along different axes, partial ablations along the D-V axis were conducted. These embryological manipulations were done at the same stages as the A-P partial ablations, Nieuwkoop and Faber stages 24–32 (Nieuwkoop and Faber, 1967). Half ablations along the D-V axis produced repeatable, although quite different results from those along the A-P axis (Fig. 1) (Table 1) (Waldman et al., 2007). Dorsal or ventral half ablations almost never resulted in the mirror image duplications or normal ears we saw after A-P half ablations. Partial ablations along the D-V axis instead resulted in inner ear deficits, with the extent and number of structures lost differing between dorsal and ventral half ablations.
Table 1. Results of Partial Ablations Along the Dorsal–Ventral Axis
Type of ablation
No. of embryos
No. of otoliths
Endolymphatic duct present
No. of sensory organs (average)
Almost all examples had the normal number of sensory organs, five.
Dorsal ablations had a more severe phenotype, in regards to number of structures lost, than ventral half ablations. Removing the dorsal half (n = 44) resulted in ears of close to normal size but missing semicircular canals, one or both otoliths (the latter in at least 70% of ears) and the endolymphatic duct (42 of 44) (Fig. 1A–D,F). These ears had an average of 2.5 sensory organs (out of 5). The identity of these sensory organs was difficult to define relative to control ears (Fig. 1D,A′–C′,E). The stage ablations were done had relatively minor effects on number of sensory organs (an average of two at early stages, 25–27, n = 6, vs. 2.6 at later stages, 29–32, n = 5) as well as other structures. The two dorsal ablations that still had an endolymphatic duct present were done at early stages (between stages 25 and 27). As to the loss of otoliths, dorsal ablations done at later stages (stages 29–32) were more likely to have none (five of five) than those done at early stages (four with none, one with one, and one with two; stages 25–27).
Removing the ventral half (n = 26) resulted in small ears with at most two semicircular canals, the loss of the utricular but not saccular otolith in 85% of ears and an average of 3.7 sensory organs remaining (Fig. 1G–K′). These ears still usually formed the endolymphatic duct, a dorsal structure (11 of 16) (data not shown). Even though ears resulting from ventral half ablations were smaller, they typically had at least one partially formed semicircular canal and larger numbers of sensory organs than seen after dorsal half ablations (Fig. 1H–K). Relatively few of the ventral half ablations resembled mirror posterior duplicated inner ears (Fig. 1J,J′), although it was difficult to determine precise numbers given the malformed nature of the canals, decreased size and the typically asymmetrical distribution of sensory organs (Fig. 1H,H′,I,I′,K,K′). Ventral half ablations never resulted in the loss of both otoliths that was often seen after dorsal half ablation (Table 1). Unlike dorsal half ablations, there were no differences between ventral ablations done at early (stages 25–27) vs. late stages (stages 29–32) in terms of the number of sensory organs (an average of 3.5 at early stages, n = 6, vs. 3.6 at later stages, n = 5) or presence of an endolymphatic duct. While we see clear size differences after ablating half of the developing inner ear, it is important to note that seemingly normal sized inner ears from dorsal half removal may still have fewer cells because they lack semicircular canals and/or are more inflated than smaller ears with the same number of cells.
Removal of just the dorsal third (n = 18) resulted in malformed inner ears, although not as severe as those seen after dorsal half ablation (Fig. 2). Almost half of these ears formed an endolymphatic duct (8 of 18; Table 1 and data not shown). Semicircular canals ranged from almost entirely normal to completely malformed, although never entirely absent (Fig. 2A–H). Generally all the resulting ears, even those with the most severely malformed semicircular canals, had three cristae, including the lateral crista (Fig. 2D,D′,K,K′). However, resulting cristae were often smaller than those found in normal inner ears due to having fewer mechanosensory hair cells. Other sensory organs such as saccular maculae were almost always present but the utricular macula was either severely reduced in size, missing, or in an ectopic location with the latter two possibilities difficult to distinguish (Fig. 2C′,J′). Dorsal third ablations rarely resulted in the loss of both otoliths, although loss of one otolith (typically the one associated with the utricular macula) was the most common phenotype (56%) (Table 1) (Fig. 2F–H). The single otolith often had abnormally large, fused otoconia and was typically associated with the saccular macula. Labeled hair cells could still form in a utricular macula without an otolith (2B,B′,C,C′). There were some differences between third ablations done at early (stages 25–28) vs. later (stages 29–32) stages. Early ablations were more likely to be missing the endolymphatic duct than later ones (6 of 7 vs. 4 of 11). The distribution of otoliths also differed between early (zero with none, three with one, four with two) and later stages (one with none, seven with one, three with two).
D-V Axis Determination
Embryological manipulations reversing one axis but not the other, at subsequently later stages, were used to learn when the D-V axis is determined in amphibians and chickens (Yntema, 1955; Wu et al., 1998). We tested this experimentally by rotating the otic placode at different stages and raising the embryos to tadpole stages (47–48) to assay sensory organ pattern. Rotations were done at stages 23 (n = 2), 24 (n = 1), 26–27 (n = 3), and 28–31 (n = 4). Our experiments suggest that the D-V axis is not determined until stage 24 to 26 as rotations done at stage 23 resulted in normal ears (Fig. 3A,A′). Rotations done at stage 24 or later (a stage 28 rotation is shown) resulted in abnormal inner ear development but not a simple inversion of the axis (Fig. 3B,B′). A subset of embryos (n = 4 for rotations done at stages 27–29) were fixed 1 or 2 day(s) after inner ear rotation (stage 39 shown which was 2 days later) for whole-mount in situ analysis. This showed that Wnt3a expression is associated with the determination of the D-V axis, as after at least stage 27, expression of Wnt3a, normally dorsal, remained inverted (Fig. 3D).
Inhibiting Canonical Wnt Signaling Results in Abnormal Inner Ears
The role of canonical Wnt signaling in vertebrate development has been well characterized in Xenopus (Moon et al., 2002; Clevers and Nusse, 2012). Ectopic Wnt activation by injection of Wnt ligands into ventral cells at the four to eight cell stage, leads to axis duplication through β-catenin activation, and results in two-headed embryos (McMahon and Moon, 1989; Semenov et al., 2001). To inhibit canonical Wnt signaling while minimizing nonear specific effects, we injected a morpholino to β-catenin previously used in Xenopus (Heasman et al., 2000) into half the embryo at the two-cell stage or into one cell at later stages (up to 32 cell stage, Nieuwkoop and Faber stage 6), destined to give rise mainly but not exclusively to the inner ear (Moody, 1987a, 1987b). The resulting inner ear phenotypes were somewhat variable but could be divided into one of three categories, I–III, of decreasing severity with category I being the most common (49%) (Table 2; Fig. 4). Ears in the least severe category (III, the next most common after I) had three cristae-like sensory organs but were often missing one or both maculae (Table 2) (Fig. 4B,C). The reduced number of sensory organs observed and decreased number of semicircular canals that were often not fully formed resembled the less severe phenotypes seen after dorsal third ablations (Fig. 2). While the invaginations of developing semicircular canals were always present, these partially formed semicircular canals were more severely malformed than those seen after dorsal third ablations. Also, category III ears, like dorsal third ablations, were often missing one of two otoliths. In the 17 category III animals examined for the presence of an endolymphatic duct, all but four had it.
Table 2. Results of Blocking Canonical Wnt Signaling by the Injection of β-Catenin Morpholino.
Appearance at Stage 45–48
Stage one cell injected and the numbers and percentages resulting.
Endolymphatic Duct Present
Average Number of Sensory Organs
0.83 +/− 0.20
2.11 +/− 0.25
3.39 +/− 0.19
5.00 +/− 0
The intermediate phenotype, category II, had two nearly normal sized crista-like sensory organs and was missing the anterior, utricular macula, and otolith (Table 2) (Fig. 4D–F). The two crista-like sensory organs were often symmetrically located (32 of 37) and the absence of the anterior macula suggests that these cristae may have a “posterior” identity (Fig. 4D,D′,F,F′). These inner ears resembled the mirror posterior duplications seen after anterior half ablations as well as a few of the ventral half ablations with their two crista-like sensory organs, although they were not as small as ventral ablations (Fig. 1J′). Those with three crista-like sensory organs were not symmetrically located (5 of 37) and had two of the cristae positioned anteriorly (Fig. 4E,E′). Semicircular canals were more severely malformed than those of category III, having fewer and less extensive invaginations (Fig. 4D,F).
Category I ears were the smallest with no fully formed semicircular canals, at most one to two sensory organs and either no otoliths or a single one with fused otoconia (Fig. 4G–J). These resembled dorsal half ablations except for their size which was usually smaller than that seen after dorsal ablations (Fig. 1). Even the invaginations associated with partial formation of the semicircular canals were rarely visible (Fig. 4J). Any sensory organs present were of reduced size and either macula-like, usually associated with an otolith, or crista-like. There was not more than a single macula- or crista-like sensory organ (those few with two had one of each) and the latter could be located anteriorly or posteriorly.
Wnt3a Protein Alone is Sufficient to Rescue the Loss of the Dorsal Structures
Rescue of the severe deficits seen after dorsal half ablations with Wnt3a protein soaked beads was impressive (73% of 15 embryos) (Fig. 5). Not only was the pattern of sensory organs nearly normal but semicircular canals also formed (Fig. 5A,A′,B). That a single molecule can compensate for the loss of a presumably large number of genes normally expressed in this region is surprising. In contrast, Wnt3a could not rescue ventral half ablations (11% of 9 embryos). Control beads without Wnt3a did not rescue the dorsal half ablations (Fig. 5C,C′). Ablations and Wnt3a bead implants were done at stage 24–28 and confocal analyses were done at stage 45–48. Of interest, three of the embryos that rescued the dorsal half ablations had mirror anterior duplications, supporting our hypothesis that Wnt signaling may play a role in anterior patterning as well (Fig. 5D).
Ablation Results in Relation to Regenerative Potential and Inner Ear Fate Maps
Removal of the dorsal or ventral half results in the loss of structures presumed to originate from that region, for example, more semicircular canals were lost after dorsal than ventral half ablations. These results, when compared with our previous study (Waldman et al., 2007), suggest there is less regenerative potential along the D–V axis as compared to the A-P axis. This confirms the results of partial ablations done in salamanders that showed loss of inner ear structures thought to arise from the removed region (Kaan, 1926). This is also consistent with the compartment and boundary model of inner ear patterning that says otic structures arise from specific sub-regions of the placode or otocyst (Brigande et al., 2000b). The smaller ears observed after ventral vs. dorsal ablations could be explained by the extensive contribution of the ventral otic placode to the lateral wall of the otocyst observed in the chick otic placode fate map (Brigande et al., 2000a). This fate map found that a significant portion of the otocyst originated from the ventral otic placode so the loss of this region could result in a smaller otocyst.
However, the ablation results were complicated by the fact that not all the structures lost could be readily assigned to the region being removed. Other structures such as sensory organs were lost after either manipulation and while there were fewer sensory organs after dorsal half ablation, the type of sensory organs remaining did not seem to differ between these two manipulations. This is consistent with our inner ear fate map in Xenopus which showed there was cell mixing during development and that specific sensory organs could originate from different regions (Kil and Collazo, 2001, 2002). This observation is not unique to Xenopus, as what appeared to be cell mixing was also observed in the developing mouse otocyst (Riccomagno et al., 2005).
Comparisons Between Ablations and Morphants
Conditionally blocking canonical Wnt signaling with the β-catenin morpholino resulted in a broader range of phenotypes than those seen after dorsal or ventral partial ablations (Figs. 1, 4). These mainly resembled phenotypes seen after dorsal third and half ablations. Morphants of our category I resembled the class III phenotype seen in embryos depleted of Sox9 which also had reduced Wnt3a expression (Park and Saint-Jeannet, 2010). Morphants in the less severe categories II and III more resembled dorsal third than half ablations, although they had more severely malformed semicircular canals than dorsal third ablations. These results suggest that morphants show a similar dosage effect to that seen after partial dorsal ablations. Given Wnt signaling's role in dorsal patterning, it should not be surprising that the β-catenin morpholino phenotypes resembled dorsal partial ablations.
Although there were clear resemblances between morphant phenotypes and those seen after dorsal ablations, there were also some interesting differences. A few category II morphants resembled ventral half ablations, something not seen in categories I and III. This resemblance is mainly due to the smaller size of the ear and layout of the decreased number of sensory organs (which could resemble mirror posterior duplications with just two distally located crista-like sensory organs) observed in β-catenin morphants. While this resemblance to ventral ablations may seem unintuitive given Wnt's presumed role in dorsal patterning this could be explained if one considers the severity of the phenotype. Dorsal half ablations have a more severe phenotype than ventral half ablations so resemblance of the β-catenin morpholino phenotypes to the latter may just represent a less severe effect on inner ear patterning. Also, even though the abnormal pattern of sensory organs seen in type II morphants and ventral ablations could be similar, all three morphant phenotypes usually had more severe semicircular canal deficits than ventral half ablations. Such semicircular canal deficits are more consistent with dorsal partial ablations. As to the issue of smaller size, in mice Wnt signaling positively regulates the size of the otic placode by directly targeting otic specific genes and by interactions with the Notch signaling pathway (Jayasena et al., 2008). This role for Wnt may account for the decreased size observed after β-catenin knockdown especially for those category I inner ears that otherwise resembled dorsal half ablations except for their smaller size.
Possible Role for Wnt in Anterior as Well as Dorsal Patterning
The range of inner ear phenotypes seen after knocking down β-catenin is quite broad and includes a small number that appear to affect the A-P axis (the 27% type II morphants, Table 2). Type II morphants more resembled mirror posterior inner ears than did ventral half ablations and provide more direct evidence suggesting that Wnt signaling may also play a role in anterior patterning. Whether these do represent mirror duplications is not certain, especially in the case of those resulting from ventral ablations, which may be revealing a symmetrical prepattern without axial identity as has been observed early in zebrafish otic development (Hammond and Whitfield, 2011). The results of the Wnt3a bead experiments provide further evidence of a possible role for Wnt signaling in anterior patterning. Partial mirror image anterior duplications were sometimes seen in the rescued inner ears resulting from a Wnt3a bead placed where the dorsal half had been removed (Fig. 5D). Together these results suggest that Wnt signaling is not only necessary and sufficient for dorsal patterning but may also be involved in anterior patterning. The role of Wnt signaling in inner ear patterning of Xenopus may be the inverse of the role of Hh signaling in zebrafish that is thought to be involved in ventral and posterior patterning (Whitfield and Hammond, 2007).
Our study not only describes a role for canonical Wnt signaling in the dorsal patterning of the frog inner ear but also proposes that Wnt3a is the specific Wnt molecule involved. Beads containing Wnt3a protein were capable of rescuing the loss of inner ear structures seen after dorsal but not ventral half ablation, suggesting that Wnt3a protein alone is sufficient to rescue the severe deficits resulting from dorsal half ablations. In mice it has been shown that Wnt1 and Wnt3a are the Wnt molecules necessary and sufficient for dorsal patterning (Riccomagno et al., 2005). The role of Wnt signaling in dorsal patterning of the inner ear, much like the role of Hh signaling in ventral patterning, appears to be evolutionarily conserved across vertebrate taxa (Whitfield and Hammond, 2007; Hammond et al., 2010). While the specific Wnt molecules involved in dorsal patterning of the inner ear may vary across vertebrates, it appears that the role of Wnt3a in this process has also been conserved during vertebrate evolution. If there is a role for Wnt signaling in anterior patterning, it appears to be specific to Xenopus.
Differences in Axial Patterning Across Vertebrates
While there is a growing understanding of the molecular pathways involved in patterning the developing inner ear, there are seemingly contradictory findings across species (Whitfield and Hammond, 2007; Groves and Fekete, 2012). In chick and mice, RA is involved in anterior and posterior patterning (Bok et al., 2011). In contrast, A-P patterning in zebrafish depends on two different cell signaling pathways, which is likely the case for Xenopus as well (Hammond and Whitfield, 2011). Anterior patterning in zebrafish requires Fgf signaling while Hh signaling is required for posterior identity (Hammond et al., 2003; Hammond and Whitfield, 2011). As in zebrafish, Hh signaling is required for posterior patterning in Xenopus (Waldman et al., 2007). In zebrafish, Fgf and Hh signaling are thought to act on a symmetrical prepattern early in otic development which helps explain the mirror duplications seen after either signaling pathway is perturbed (Hammond and Whitfield, 2011). The specific Fgf molecule involved in anterior patterning has been thought to be fgf3 but while fgf3 is sufficient to induce anterior identity in the posterior half, loss of fgf3 does not completely eliminate anterior identity (Kwak et al., 2002; Hammond and Whitfield, 2011). Like posterior patterning in zebrafish, where more than one Hh molecule is involved, anterior patterning likely requires multiple Fgf ligands (Hammond et al., 2003; Hammond and Whitfield, 2011).
Other differences in axial patterning seen across species involve the source of the signaling molecules. In zebrafish, A-P patterning depends on signals from the hindbrain while in amniotes A-P patterning is independent of hindbrain signals and instead depends on signals from adjacent ectoderm (Groves and Fekete, 2012). There are multiple Fgf genes expressed in the anterior zebrafish inner ear during development (fgf3, fgf8, and fgf10a), but they are expressed after anterior specification occurs and so are more likely playing a role in reinforcing anterior identity (Hammond and Whitfield, 2011). In mouse, Wnt1 and Wnt3a are expressed in the dorsal hindbrain adjacent to the otocyst but not in the otocyst itself (Takada et al., 1994). This is in contrast to chick and frogs where Wnt3a is also expressed in the dorsal otocyst, although in mammals Wnt2b has this expression (Wolda et al., 1993; Hollyday et al., 1995; Hatch et al., 2007). The source of the Wnt signal involved in D-V patterning of the inner ear may represent another character that varies across vertebrates.
Some of the phenotypes we observed after β-catenin knockdown that did not resemble those seen after D-V half ablations may reflect an earlier role of Wnt signaling such as in separating otic from epidermal domains during placode induction (Lombardo and Slack, 1998; Ladher et al., 2000; Ohyama et al., 2007; Urness et al., 2010; Groves and Fekete, 2012). Even the conserved role of Wnt signaling in dorsal patterning across vertebrates does not tell the whole story. In mouse, both canonical (β-catenin dependent) Wnt signaling and Fgf3 signaling are necessary for dorsal patterning but how these two pathways interact to achieve this is not well understood (Riccomagno et al., 2005; Ohyama et al., 2006; Hatch et al., 2007). Of the two Wnt ligands known to be involved in mice, Fgf3 inhibits ventral but not dorsal expression of Wnt3a in the hindbrain but has no effect on Wnt1 (Hatch et al., 2007). Later roles for Wnt signaling involve the patterning of specific sensory organs within the inner ear. Ectopic activation of Wnt signaling in chick embryos results in the conversion of auditory hair cells to vestibular hair cells in the cochlear duct (Stevens et al., 2003). Other defects include ectopic hair cells and otoliths in the cochlear duct and gross morphological defects such as enlargements of the cochlear and endolymphatic ducts. While this chick study has suggested a role for Wnt in regulating the size of the sensory organ in the cochlea, these results might also be due to effects on cell proliferation (Groves and Fekete, 2012).
Partial ablations provide a powerful embryological manipulation for understanding axial patterning and the regenerative potential of the otic placode and otocyst. Combining these with molecular studies on the role of Wnt signaling in D-V patterning has provided some unique insights. Wnt3a protein alone is sufficient to almost completely compensate for the loss of the many genes expressed in the removed dorsal half. Partial ablations along the A-P axis revealed a dosage effect, with 2/3rd ablations having more abnormal ears than those after third or even half ablations (Waldman et al., 2007). A similar dosage effect was also seen in our partial ablations along the D-V axis with third ablations having a less severe phenotype than those seen after dorsal half ablations. However, unlike the anterior or posterior third ablations which were all normal, dorsal third ablations resulted in mostly abnormal ears with loss of structures such as otoliths and semicircular canals. Dorsal third ablations are removing most if not all of the Wnt3a mRNA expressing region which is restricted to the dorsal-most region of the otocyst (Wolda et al., 1993). Interestingly, the middle third ablations done along the A-P axis, unlike removal of the distal-most anterior or posterior third, resulted in significant numbers of abnormal inner ears that were not mirror duplications (42.9%) (Waldman et al., 2007). This partial ablation would have removed much of the dorsally expressed Wnt3a gene, which may account for some of the abnormal phenotypes seen. Other genes are expressed in the dorsal otocyst of the developing frog inner ear, but only Wnt3a expression begins just after the determination of the D-V axis.
Microsurgical Manipulations of Embryonic Inner Ear Tissues
Microsurgeries involving ablations and tissue grafting were done as in our previous study with the following modifications (Waldman et al., 2007). Partial ablations to determine the role dorsal and ventral regions play in inner ear patterning, were done by physically removing ½ or ⅓ of the otic placode or otocyst along the D-V axis. Stages used were similar to those used for our A-P half ablations, 24–32 (Waldman et al., 2007). There was no obvious correlation between stage and severity of phenotypes observed. For example, dorsal half ablations done at stage 26 resembled ones done at stage 31 (Fig. 1A,B).
Rotations of the inner ear placode or otocyst were done at multiple stages to understand when the D-V axis was determined. Manipulations inverting just the D-V axis involved removal of the placode or otocyst on one side (i.e., left) and the grafting of a rotated placode or otocyst from the other side (i.e., right) which would leave the A-P axis unchanged. Controls consisted of experiments where an unrotated placode or otocyst from the same side was grafted in and these did not result in abnormal development. A point labeling with Nile blue sulfate on the dorsal part of the placode or otocyst being transplanted helped keep track of orientation. Previously, we had shown that (1) complete removal of the otocyst results in loss of the inner ear and (2) regenerated tissues from partial ablations arise from the remaining placode and not surrounding tissues (Waldman et al., 2007).
Attempts to rescue the partial ablation defects were done by placing Wnt3a protein soaked (R&D Systems, 1324-WN/Carrier Free) heparin/acrylic beads (Sigma H-5236) where either the dorsal or ventral half had been removed. The methods used were adapted from bead experiments in chick embryos (Niswander et al., 1993). Appropriate sized beads (approximately 40 to 100 μm) were collected using forceps into a 100-μl drop of phosphate buffered saline (PBS) in a 35 mm Petri dish. In another 35-mm Petri dish, approximately fifteen 8-μl drops of PBS were placed in a circle throughout the dish, excluding the center, to maintain humidity when the lid was on. In the center, a 1- to 3-μl drop of protein was placed. Beads were picked up using forceps and transferred to the protein drop in the center of the second dish. More than twenty 100-μm beads fit in a 1-μl drop. To make sure no PBS was transferred and avoid protein dilution, capillary action was used to wick up excess PBS using almost closed forceps, working back and forth between a dry area on the plate and one of the 8-μl drops until the bead was almost dry. If the bead was too dry, static electricity would cause it to jump away from the forceps. Beads were soaked in protein for approximately 2 hr at room temperature or overnight at 4°C. Wnt3a protein concentration for bead soaking was 40 μg/ml, diluted in PBS as per the manufacturer's instructions. Beads were rinsed in PBS just before grafting into frog embryos by transfer to one of the fifteen 8-μl drops. Negative controls for bead experiments used beads soaked in just PBS which did not rescue half ablations.
We inhibited canonical Wnt signaling by injecting a morpholino to β-catenin whose function and phenotype has been well characterized in Xenopus (Heasman et al., 2000). In fact, this morpholino works so well that it is recommended as a positive control by Gene Tools LLC (sole commercial supplier of morpholinos). β-catenin mRNA is strongly expressed in the developing frog otocyst (DeMarais and Moon, 1992). Injecting the β-catenin morpholino into half the embryo at the two-cell stage or later helped avoid severe disruptions of development that were nonear specific. β-catenin morpholino injected at the 16- to 32-cell stages was targeted to a cell destined to give rise to the inner ear (Moody, 1987a, 1987b). It is important to note that while injection of the morpholino at all these stages would label the ear on one side, leaving the other side as a negative control, the Moody fate maps show that even one cell at the 32-cell stage would often result in the labeling of organs and structures adjacent to the otocyst. The morpholino experiments could not address the source of the Wnt signal as they are knocking down Wnt signaling both extrinsic and intrinsic to the inner ear. However, that in all but one of our β-catenin knockdowns (n = 139; Table 2), inner ears on the uninjected side were normal suggests that any reduction of Wnt signaling in the hindbrain was insufficient to disrupt development of the control inner ear and/or that any extrinsic Wnt signal is not very long range. The morpholino is fluorescently tagged so only embryos with expression in the otocyst were selected for detailed analyses (Fig. 6). Concentrations of morpholino injected (50 ng for the embryos shown) were higher than previously published because of our use of a fluorescently tagged morpholino which increased the molecular weight without increasing bioactivity. The morpholino was titrated through a range of concentrations and by injection into one cell from 2- to 32-cell stages. No increase in mortality was observed using concentrations ranging from 8 to 88 ng but over 88 ng we saw an increase in mortality (data not shown).
Analysis of Inner Ear Phenotypes
The analyses of inner ear phenotypes generated by our partial ablations were done at later stages when most of the sensory organs and all three semicircular canals should have formed. Embryos were raised until tadpole stages (mainly stage 47 or 48; 1 to 2 weeks of age) when at least five of the eventual eight sensory organs have developed (Kil and Collazo, 2001). We labeled hair cells in vivo with the styryl dyes FM1-43 or 4-Di-2-ASP (Meyers et al., 2003) to assess sensory organ pattern as we have published (Kil and Collazo, 2001; Waldman et al., 2007). Positions and sizes of otoliths were noted before when hair cells were labeled as injection of the vital dye into the otocyst can move otoconia. Images were collected with either a widefield Zeiss epifluorescence microscope or a laser scanning confocal microscope as described in our previous study with the following differences (Waldman et al., 2007). Confocal images were collected with either a Zeiss LSM 710 or a Leica SP5 and consisted of multiple focal planes in a Z-stack. Some images collected with the LSM 710 were taken using a two-photon laser at a wavelength of 900 nm and are noted in the figure legends. For some images taken with the epifluorescence microscope, multiple focal planes were combined to make one extended focus image using the FIJI version of the ImageJ software and extended depth of field plugin in either easy or expert mode, the latter using a complex wavelets algorithm (Forster et al., 2004). Most confocal images shown were maximum intensity projections of the collected Z-stack except for those in Figure 4 which used the extended depth of field plugin from FIJI.
RNA In Situ Hybridizations
Whole-mount in situ hybridizations of Wnt3a mRNA were done as described in our previous studies (Kil and Collazo, 2001; Waldman et al., 2007). This Xenopus Wnt3a probe was kindly provided by Dr. Randy Moon (Wolda et al., 1993). Expression of Wnt3a was studied in a subset of embryos with rotated inner ears to see if its region specific expression correlated with when the D-V axis was determined. Embryos to be analyzed were fixed 1 or 2 day(s) after rotation. Wnt3a expression was also studied in a subset of embryos in which the dorsal half (n = 6) or dorsal third (n = 2) was removed at stages 29–31. Such late stages were chosen because the earliest that we could detect dorsal Wnt3a expression was stage 26. For the dorsal half ablations, there was no Wnt3a expression in 4 of 6 embryos examined while none of the two dorsal third ablations examined expressed Wnt3a. Wnt3a expression was documented with a color Zeiss Axiocam, through a Zeiss M2Bio Dissecting scope, using the Zeiss Axiovision software.
This work was supported by RO1 Grant DC004061 to A. Collazo from the National Institutes of Health (NIH) and National Institute for Deafness and Other Communication Disorders (NIDCD) and by a National Organization for Hearing Research (NOHR) grant to C. Forristall. This work was also supported by the Oberkotter Foundation through the House Ear Institute/House Research Institute, where much of this work was done. We want to thank the many University of Redlands students who worked on this project. Specifically we acknowledge the work of John Gregorius, Darlynn Korns, Kristen Block, and Gerald Chavez. We also thank the members of the Andy K. Groves, Neil Segil, and Andres Collazo laboratories when they were at the House Ear Institute, for their stimulating discussions of the data presented here. Core resources were provided by the Ahmanson Foundation and NIH to D.J.L. The Xenopus Wnt3a probe was kindly provided by Dr. Randy Moon.