Ablation studies on the developing inner ear reveal a propensity for mirror duplications


  • Erik H. Waldman,

    1. Leslie and Susan Gonda (Goldschmied) Department of Cell and Molecular Biology, House Ear Institute, Los Angeles, California
    Current affiliation:
    1. Department of Otolaryngology & Communication Disorders, Children's Hospital Boston, 300 Longwood Avenue, LO367, Boston, MA 02115
    Search for more papers by this author
  • Aldo Castillo,

    1. Leslie and Susan Gonda (Goldschmied) Department of Cell and Molecular Biology, House Ear Institute, Los Angeles, California
    Search for more papers by this author
  • Andres Collazo

    Corresponding author
    1. Leslie and Susan Gonda (Goldschmied) Department of Cell and Molecular Biology, House Ear Institute, Los Angeles, California
    • Andres Collazo, Leslie and Susan Gonda (Goldschmied) Department of Cell and Molecular Biology, House Ear Institute, Los Angeles, CA 90057
    Search for more papers by this author


The inner ear develops from a simple ectodermal thickening known as the otic placode. Classic embryological manipulations rotating the prospective placode tissue found that the anteroposterior axis was determined before the dorsoventral axis. A small percentage of such rotations also resulted in the formation of mirror duplicated ears, or enantiomorphs. We demonstrate a different embryological manipulation in the frog Xenopus: the physical removal or ablation of either the anterior or posterior half of the placode, which results in an even higher percentage of mirror image ears. Removal of the posterior half results in mirror anterior duplications, whereas removal of the anterior half results in mirror posterior duplications. In contrast, complete extirpation results in more variable phenotypes but never mirror duplications. By the time the otocyst separates from the surface ectoderm, complete extirpation results in no regeneration. To test for a dosage response, differing amounts of the placode or otocyst were ablated. Removal of one third of the placode resulted in normal ears, whereas two-thirds ablations resulted in abnormal ears, including mirror duplications. Recent studies in zebrafish have demonstrated a role for the hedgehog (Hh) signaling pathway in anteroposterior patterning of the developing ear. We have used overexpression of Hedgehog interacting protein (Hip) to block Hh signaling and find that this strategy resulted in mirror duplications of anterior structures, consistent with the results in zebrafish. Developmental Dynamics 236:1237–1248, 2007. © 2007 Wiley-Liss, Inc.


The inner ear is a highly asymmetrical sensory structure with clearly defined anteroposterior, dorsoventral, and mediolateral axes. Its sensory organs, which contain the mechanosensory hair cells, are stereotypically located within the inner ear, which is important for their normal function in balance and hearing (Hudspeth,1989). The number of sensory organs contained in the inner ear varies between vertebrates, but all have at least six: five for the vestibular system plus one for the auditory system (Fritzsch et al.,2002). Approximately a century of descriptive and experimental embryological studies have provided a basic understanding of the development of the inner ear from the otic placode, an ectodermal thickening adjacent to the developing hindbrain (Fritzsch et al.,1998; Baker and Bronner-Fraser,2001; Rinkwitz et al.,2001; Fekete and Wu,2002).

To test when during development the three main axes (anteroposterior, dorsoventral, and mediolateral) of the inner ear were determined, embryological manipulations involving the rotation of the otic placode were done (Yntema,1955). Most of these studies were conducted in several different species of amphibians, but more recent work in avian embryos came to the same conclusion, that is, that the anteroposterior axis is determined before the dorsoventral axis (Wu et al.,1998). Interestingly, in a small percentage of cases in the amphibian rotation studies, mirror duplicated ears were generated (Harrison,1936). These embryological manipulations were the only known method of generating such duplications until recently.

A zebrafish study identified that members of the Hedgehog (Hh) signaling pathway are involved in anteroposterior patterning of the developing inner ear (Hammond et al.,2003). This finding contrasts with the role of Hh signaling in mouse inner ear development, where Sonic hedgehog (Shh) plays a role in dorsoventral patterning by promoting ventral identity and restricting Wnt genes to the dorsal region (Riccomagno et al.,2002,2005; Ohyama et al.,2006). In chickens, Shh has also been shown to be important for ventral patterning of the otocyst (Bok et al.,2005). Hh (a secreted and lipid-modified signaling protein) binds to its receptor patched, which releases inhibition of another transmembrane protein, smoothened, transducing the signal by means of the Gli transcription factors (Ingham,1998; McMahon et al.,2003; Panakova et al.,2005; Fuccillo et al.,2006). Two zebrafish mutants known to eliminate Hh signaling were found to have partial duplications of both anterior inner ear structures and gene expression in the posterior half, as well as reduction of posterior gene expression. One of these mutations is in the Hh signal transducer Smoothened and is one of the strongest Hh signaling mutants so far isolated in zebrafish (Hammond et al.,2003). These effects were phenocopied by overexpression of the patched1 mRNA, which also reduces Hh signaling. Augmenting Hh signaling by such manipulations as overexpressing Shh generated the reverse phenotype, duplicating both posterior structures and gene expression in the anterior half of the inner ear. Of the three hedgehog genes known in zebrafish, only Shh and Tiggy-winkle hedgehog were involved.

We discovered another embryological manipulation that results in the formation of mirror duplicated inner ears. In these experiments, either the anterior or posterior half of the placode is physically removed and the embryos are left to develop. Significant percentages resulted in duplicated inner ears, with two anterior or two posterior halves fused as if mirror images. These percentages were much higher than was ever seen with the amphibian rotation studies (Yntema,1955) or even the zebrafish Hh signaling pathway mutants (Hammond et al.,2003). Labeled grafting experiments confirmed that the regenerated tissue was from the remaining placode and not surrounding tissues. Complete extirpation experiments address the following question: given the removal of the placode or otocyst what is the potential of the surrounding tissues to generate a normal inner ear? The results of completely removing the placode or early otocyst reveal that the inner ear is much less regenerative than other placodes such as the olfactory (Stout and Graziadei,1980) and somewhat contradict earlier ablation studies in salamander (Kaan,1926). Lastly, to investigate the potential involvement of the hedgehog signaling pathway on anteroposterior inner ear development, experiments overexpressing Hedgehog interacting protein (Hip) were undertaken. Hip has been demonstrated to be a potent inhibitor of Hh signaling (Koebernick et al.,2003; Cornesse et al.,2005). Hip overexpression resulted in mirror anterior duplications. By showing that Hh signaling is involved in the anteroposterior patterning of the inner ear, these results suggest that Xenopus is more like zebrafish than mouse and chickens. They also suggest that, during the course of evolution, the role of Hh signaling in inner ear patterning changed from an inducer of posterior identity to an inducer of ventral identity in amniotes.


Complete Ablations Suggest That the Regenerative Ability of the Inner Ear Quickly Declines With Age

To test the ability of surrounding tissues to regenerate the inner ear, the placode or otocyst was completely removed leaving the surrounding tissue intact. The contralateral side served as a nonoperated control. Previous studies in salamanders found that the inner ear could still develop after ablation of the prospective otic placode, even when an area larger than the otic placode was ablated (Kaan,1926). Our complete extirpations in Xenopus embryos were done at stages ranging from 21 to 30 (Nieuwkoop and Faber,1967), which represent three categories of inner ear development: early placode (21–23), late placode (24–27), and otocyst (28–30; Fig. 1; Schlosser and Northcutt,2000).

Figure 1.

Results of complete ablations. Inner ear placodes or otocysts were ablated at stages representing three different categories encompassing several Nieuwkoop and Faber (1967) stages. The Y-axis lists the percentage of embryos (out of 142, see Table 1) that had the observed phenotype. The four categories are discussed in the text. Previously published results showing the percentage of complete regeneration seen after prospective olfactory tissues were ablated (nasal regeneration) are included for comparison (Stout and Graziadei,1980).

Table 1. Complete Ablations
Stage at ablationNo regenerationOtocysts without sensoryOtocysts with sensoryComplete regenerationTotals
Early placode (21-23)1123.9%1839.1%1123.9%613.0%46
Late placode (24-27)1634.0%2144.7%1021.3%00.0%47
Otocyst (28+)4898.0%00.0%12.0%00.0%49
Totals75 39 22 6 142

The morphology of inner ears resulting from complete ablation could be placed into four categories: (1) complete regeneration, (2) formation of abnormal otocysts without sensory cells, (3) formation of abnormal otocysts with sensory cells, and (4) complete absence of regeneration (Fig. 2; Table 1). The stage the ears were scored, typically 48, was a tadpole stage where the semicircular canals are present and five of the eventual eight sensory organs have developed (Kil and Collazo,2001; Bever et al.,2003). The ear was scored as completely regenerated if it appeared identical to nonmanipulated control ears. The abnormal inner ears in the other two categories were often smaller in size and did not have fully formed semicircular canals. The abnormal otocysts with sensory cells typically had fewer sensory organs than control ears, but we could not determine whether these findings resulted from fusions or absence of sensory organs. These sensory organs were often in ectopic positions and of variable sizes, so identification of these sensory organs as cristae or macula was not always possible. The complete absence of regeneration could be judged by blood vessels that normally ended at the lateral edge of the inner ear continuing to the hindbrain and by the apposition of anterior and posterior lateral line ganglia that are normally separated by the inner ear (Fig. 2D,D′).

Figure 2.

Examples of the phenotypes seen after complete ablation. A,A′: Complete regeneration (stage 21 ablation). B,B′: Otocyst without sensory organs (stage 25 ablation). C,C′: Otocyst with sensory organs (stage 24 ablation). D,D′: No regeneration (stage 21 ablation). A–D: Brightfield viewed at stages 47–48. A′–D′: Corresponding fluorescent image. L, left side ablated; R, right side control. Dorsal views, anterior to top. Tadpoles were imaged alive through an epifluorescent compound microscope using a low light level camera. Gray brackets indicate the length of inner ear. mb, midbrain; hb, hindbrain; sc, spinal cord; as, anterior semicircular canal (scc); ls, lateral scc; ps, posterior scc; uo, utricular otolith; so, saccular otolith; suo, saccular &/or utricular otolith; allg, anterior lateral line ganglion + facial; pllg, posterior lateral line ganglion; ac, anterior crista; lc, lateral crista; pc, posterior crista; um, utricular macula; sm, saccular macula; uso, unidentified sensory organ; ki, kidney autofluorescence; A, anterior; D, dorsal. Scale bar = 750 μm.

The ability of surrounding tissues to regenerate any type of inner ear quickly drops from placode to otocyst stages (Fig. 1). Complete regeneration of the inner ear was only seen at early placode stages (13%, Table 1). Removing the otic structure at placode stages typically resulted in some form of regeneration, although the ears were mostly abnormal. Xenopus embryos fail to regenerate ears if the extirpation occurs during the otocyst stages of development (48/49 showed no regeneration). The inner ear, unlike other placodally derived sensory structures, such as the olfactory placode (Stout and Graziadei,1980), has less regenerative potential, even though the olfactory and otic placode differentiate at similar stages (Fig. 1; Nieuwkoop and Faber,1967; Schlosser and Northcutt,2000).

The most common category of inner ear seen at early and late placode stages was otocysts without sensory organs (Table 1). Of the four categories of ears resulting from complete ablation, no regeneration was the second most common at late placode stages and tied with abnormal otocysts with sensory organs at early placode stages. The lack of regeneration apparent at otocyst stages, but also seen at placode stages, suggests that any regenerated tissue in partial ablation would likely be derived from the remaining placode and not surrounding tissues. In summary, completely normal inner ears are only seen at early placode stages, but even at this stage, no regeneration is more common, and by otocyst stages, complete ablation results in no regeneration.

Half Ablations Result in a High Percentage of Mirror Duplicated Inner Ears

Previous otic placode rotation experiments revealed a small percentage of embryos with an interesting phenotype: mirror duplicated (enantiomorphic) inner ears (Harrison,1936; Yntema,1955). These findings were either mirror anterior (with two sets of anterior and lateral cristae and utricular maculae but no posterior cristae) or mirror posterior (with two sets of posterior cristae, papilla amphibiorum, and lagenar maculae but no anterior and lateral cristae or utricular maculae). In the course of our ablation experiments, we discovered another embryological manipulation that results in mirror-duplicated ears at even higher frequencies than seen in previous studies.

We ablated either the anterior or posterior half of the otic placode at late placode to early otocyst stages (24–27; Table 2). Unlike complete ablations in which the inner ear often never regenerated or formed vesicles without sensory organs, these half ablations all (71/71) had sensory organs in the otocyst and many of these were mirror duplicated inner ears. The resulting ears could be placed in three categories, perfect mirror duplications, normal, and an intermediate category with either partial duplications of certain structures, but not all, or noticeably smaller but normal-looking ears (Fig. 3). Anterior mirror duplications contained a symmetrical inner ear with two anterior halves fused about the anteroposterior axis, whereas the presumed posterior mirror duplications seemed to contain a symmetrical inner ear with two posterior halves fused about the anteroposterior axis (Fig. 4). The identity of the cristae in what we call mirror posterior duplications could instead represent two lateral cristae, two anterior cristae, or an anterior and posterior crista. We think these ears have two posterior cristae because the resulting cristae have hair cells in the horizontal arrangement typical of anterior and posterior cristae rather than the denser and more vertically arranged hair cells in lateral crista. Also these ears are missing the utricle and its otoliths, which typically lie just below the anterior cristae. The intermediate category (called “other” in Fig. 3 and “other otocysts with sensory” in Table 2) for the half ablations mainly consisted of ears with partial but not complete duplications. For anterior half ablations, this included ears where the cristae were duplicated but not the macula with its otolith or vice versa, whereas for posterior half ablations, this included ears missing the endogenous utricle or either the lateral or anterior crista. The smaller normal-looking ears that were also part of this category were seen less often than the partial duplications discussed, especially in posterior half ablations.

Table 2. Partial Ablations
 Otocysts without sensoryOther otocysts with sensoryaMirror duplicationsComplete regenerationTotals
  • a

    For the half ablations, this category consisted of ears with partial but not complete mirror duplications or normal-looking ears that were significantly smaller than the control side.

Anterior half ablations00.0%1028.6%1131.4%1440.0%35
Posterior half ablations00.0%1438.9%1541.7%719.4%36
Ant/Post 1/3 ablations00.0%00.0%00.0%10100.0%10
Ant/Post 2/3 ablations420.0%1050.0%630.0%00.0%20
Middle 1/3 ablations00.0%342.9%00.0%457.1%7
Totals4 37 32 35 108
Figure 3.

Results of anterior (posterior left) or posterior (anterior left) half ablations. The resulting inner ears were placed in three categories: Mirror, Normal, and Other (an intermediate category). These three categories are explained in the text. Percentages are from values in Table 2.

Figure 4.

Mirror image duplicated inner ears. Dorsal views. A,A′: Normal unoperated ear. B,B′: Mirror anterior duplications after the posterior half was removed. C,C′: Mirror posterior duplications after the anterior half was removed. A–C′: Fluorescent (A–C) and brightfield (A′–C′) images at stage 48. All panels are to scale except for C, which was magnified to better visualize the hair cells. The figures are montages of multiple planes of focus. Tadpoles were imaged alive through an epifluorescent compound microscope using a low light level camera. ls, lateral semicircular canal (scc); ps, posterior scc; uo, utricular otolith; so, saccular otolith; ac, anterior crista; lc, lateral crista; pc, posterior crista; um, utricular macula; sm, saccular macula; A, anterior; L, lateral. Scale bars = 100 μm.

Anterior half ablations leaving the posterior half resulted in mirror duplicated posterior halves in 31% of cases. If we include those with partial but not perfect duplications, the percentage rises to almost 60%. Posterior half ablations, leaving the anterior half, resulted in mirror duplicated anterior halves in 42% of cases. Including those with partial but not complete duplications increases the percentage to almost 80%. The remaining ears for either half ablation looked completely normal, suggesting some ability to regenerate the normal pattern remains in the leftover half. The mirror duplicated ears after anterior half ablation were usually smaller than those after posterior half ablation (Fig. 4), suggesting that there is more potential for growth in the anterior half of the otic placode.

To determine the source of the regenerated tissue, we transplanted rhodamine-labeled half placodes into embryos in which the endogenous placode had been removed. The resulting regenerated inner ears consisted entirely of labeled tissue, demonstrating that the surrounding tissues do not contribute to the regenerated ear (Fig. 5). The example shown resulted in a normal inner ear; other examples had partial mirror duplications (data not shown). The rhodamine fluorescence was concentrated in the thicker tissues of the inner ear, such as the cristae with their densely packed hair cells and supporting cells (Fig. 5D). The otic ganglion neurons were also labeled.

Figure 5.

Results after transplanting fluorescent rhodamine-labeled half placode where endogenous placode was completely removed. The resulting inner ear was normal and consisted entirely of rhodamine-labeled cells. AD: Brightfield (A,C) and fluorescent images (B,D) at stage 48. Transplant done at stage 28. C,D: Higher magnification images of the left ear shown in A,B. Tadpole slightly tilted so the left ear saccular otolith is clearly visible while the one in the right ear is under the hindbrain. C: Montage of multiple planes of focus. L, left side transplanted; R, right side unoperated control. Dorsal views, anterior to top. The tadpole was imaged alive through an epifluorescent compound microscope using a low light level camera. hb, hindbrain; ls, lateral semicircular canal (scc); ps, posterior scc; uo, utricular otolith; so, saccular otolith; ac, anterior crista; lc, lateral crista; pc, posterior crista. Scale bars = 200 μm. A and B are at same magnification; C and D are at same magnification.

Results of Other Partial Ablations

To see if there was a dose response based on the amount of tissue left, we removed either one third or two thirds of the placode (Table 2). When one third of the anterior or posterior end was removed (n = 10; stages 24 to 27), the resulting ears were all normal. When two thirds of the placode was removed (n = 20; stages 24 to 27), the resulting ears were more variable, but complete regeneration was never seen, similar to what we observed after full ablation at these stages (Table 1). Unlike after complete ablation, two-thirds ablations resulted in a good number of mirror duplications (30%) that were consistent with the side being ablated as seen after anterior or posterior half ablations. It was also the only manipulation besides complete ablation that resulted in otocysts without any sensory organs. The abnormal otocysts with sensory organs seen after two-thirds ablations (category “Other otocysts with sensory” in Table 2; “Otocyst with sensory” in Table 1) were similar to those seen after complete ablation, with fewer or larger sensory organs in ectopic locations (Fig. 2). Although the smaller normal otocysts seen in this category after half ablation were not seen, two embryos did have partial duplications that were similar to what we described above in the half ablations.

These results suggest that the anterior or posterior-most half (maybe even third) has the information necessary to replicate a mirror image ear. To test if there was something special about the middle portion of the placode, we removed the middle third of the placode at stages 24 to 25 (Table 2). At least half of these looked entirely normal (four of seven), with the remainder having smaller ears. In one case, two separate small ears were formed. This finding was likely due to not pushing the two pieces together after removal of the middle third. The smaller ears had reduced or fused otoliths, missing canals, and fewer hair cells. All these results suggest that the placode is capable of regulating to form a normally patterned inner ear after a loss of only a third or even half of its tissue, but not if two thirds are missing.

Blocking Hedgehog Signaling Results in Mirror Anterior Duplications

In zebrafish, Hh signaling is required for posterior identity; when it is inhibited, anterior structures are duplicated posteriorly (Hammond et al.,2003). Hip is a vertebrate-specific inhibitor of Hh signaling that was discovered in mice (Chuang and McMahon,1999; Koebernick et al.,2003). We blocked Hh signaling by overexpressing the mouse Hip, which has been shown to act the same as the endogenous Xenopus hip (Cornesse et al.,2005), in half of the embryo, leaving the contralateral side as a control. The inner ears of the injected side had anterior mirror duplications in 30 to 35% of the embryos (Table 3; Fig. 6). The three examples shown include two where just the anterior and lateral cristae were duplicated and another in which both cristae and the utricular macula were duplicated. The ear on the noninjected side was always normal. The smaller eye phenotype previously described after Hip injection (Cornesse et al.,2005) was seen in 50 to 100% of the embryos injected (Table 3).

Table 3. Hip mRNA Injections
Concentration4 ng5 ng9 ng
  • a

    All embryos with mirror duplications also had the smaller eye phenotype on the same side.

Smaller eye880%15100%1995%
Mirror duplicationsa330%533%735%
Total10 15 20 
Figure 6.

Inhibiting Hedgehog (Hh) signaling results in mirror anterior duplications. Projection of multiple confocal sections. AC: Dorsal view of right (A) or left (B,C) ears. Three examples are shown. ac, anterior crista; lc, lateral crista; pc, posterior crista; um, utricular macula; A, anterior. Gray arrowhead, background fluorescence, mostly outside of the inner ear. Scale bars = 50 μm.

The extra sensory organs seen after Hip injection and half ablation do not result from the splitting of endogenous sensory organs but from the induction of new, extra hair cells. We counted the number of hair cells in endogenous and ectopic cristae in the posterior half of mirror image and control injected inner ears. The median number of hair cells normally found in the anterior and posterior cristae of normal tadpoles at stage 48 is the same, approximately 12. The number of hair cells in the ectopic “anterior” crista was not significantly different from the number found in posterior crista of control ears injected with a green fluorescent protein (GFP) mRNA (11.0 SD = 5.3; n = 5 vs. 11.7 SD = 0.8; n = 15, respectively; P < 0.63). If we add the hair cells in the ectopic “lateral” crista, the mean number of hair cells in the duplicated half increases to 23.8 (SD = 6.7; n = 5), which is significantly more than the total found in the single posterior crista of GFP mRNA injected controls (P < 0.001). The number of hair cells in the endogenous anterior crista of ears with posterior mirror duplications was not different from that of control embryos or the uninjected side of the same embryo. Our results suggest that the role of Hh signaling in anteroposterior patterning of Xenopus inner ears is similar to that in zebrafish.


Complete Ablation

Ablations studies have historically been important for understanding what a given cell or tissue forms or induces and for testing the ability of surrounding cells or tissues to reconstitute lost structures (DuShane,1935,1938; Scherson et al.,1993). Previous ablation studies in salamanders (Kaan,1926) found that most ears ablated at early and late placode stages could regenerate at least some otic tissue (76% and 72%, respectively). We found less regeneration after complete ablation then Kaan found and most of our experiments were done at comparable stages of inner ear development. Complete regeneration of normal ears was only seen when the experiments were done at early placode stages (13%), and by late placode stages, all resulting inner ears were abnormal, as were most resulting from early placode ablation (Fig. 1). Kaan found that, at early placode stages most of the regenerated ears were completely normal (9 of 14), even when an area greater than that of the otic placode was ablated. By late placode stages, she did find that most of the ears were abnormal, especially at the later otic cup stages. Most of the abnormal ears resulting from our complete ablations at early and late placode stages were otocysts with little or no invagination of the semicircular canals and lacked any hair cells that could be detected by our in vivo labeling. The otoliths, which are at least in part formed by the support cells of the sensory organ (Thalmann et al.,2001; although in zebrafish lacking support cells, very small otoliths can form; Haddon et al.,1999), were also often missing. These results led us to conclude that these abnormal inner ears lack sensory organs. The formation of empty otocysts devoid of sensory organs suggests that the formation of the nonsensory portion of the otocyst (the membranous labyrinth) may be uncoupled from that of the sensory organs. This finding is consistent with recent studies that separate the roles of genes involved in otic placode specification from those involved in later patterning (Groves and Bronner-Fraser,2000; Saint-Germain et al.,2004; Kil et al.,2005).

By otocyst stages, we found a complete lack of regeneration after ablation. This finding is a dramatic decrease in regenerative potential from late placode stages. Kaan did not do any ablations at these stages, but the only examples where she saw no regeneration were her latest otic cup ablations (Kaan,1926). That even at early and late placode stages we found no regeneration to be the second most common result further demonstrates the lack of regenerative potential of surrounding tissues. This finding was surprising given that other placodes in Xenopus such as the olfactory are highly regenerative at comparable stages of development (Stout and Graziadei,1980). Of interest, the otic placode also differs from the lens placode in that ear induction appears to begin later and end earlier than lens induction (Gallagher et al.,1996).

Two concerns arising from any ablation experiment are being certain that the tissue was completely removed and making sure that only the tissue of interest and not surrounding tissues are removed. Placodes and otocysts are relatively easy to distinguish and remove. That combined with our molecular marker, X-dll3, allowed us to be certain that we could reliably remove the whole inner ear, and just this tissue, for our total extirpation experiments (Fig. 7). Although some neural crests cells that adhere to the otocyst may also be removed, they would be few in number. Earlier ablation studies in amphibians removed the otic placode as well as tissue around and beyond the area of the placode (Kaan,1926). Because of the lack of regeneration in our complete extirpation experiments, it was fairly straightforward to be certain that we were just eliminating the inner ear and not surrounding tissues. Tadpoles with missing inner ears had all the other tissues expected around the ear such as the lateral line ganglia, blood vessels, branchial arches, and normal appearing hindbrain, suggesting that these tissues were not removed, as we confirmed visually when doing the extirpations. When removals were done before placode formation the embryos lost many structures such as the neural crest-derived branchial arches of that side (data not shown). This finding meant that we could not do ablations earlier than placode stages, so we could not determine whether, at earlier time points, the placode could regenerate more completely as might be extrapolated back from Figure 1. The fate map in chicken reveals that the prospective otic placode cells are intermixed with neural crest, hindbrain, and other placodes, so our results before placode formation are not surprising (Streit,2002).

Figure 7.

Examples of whole and partial ablations. A,B: After complete placode extirpation on one side of an albino embryo (stage 25), it was fixed for whole-mount in situ hybridization. The probe used was to Xenopus homeobox gene Distal-less 3 (X-dll3). A: Right-side lateral view of the unoperated side showing the inner ear. B: Left-side lateral view showing that the inner ear was completely removed. C,D: Partial ablation where the posterior half was removed. Pigmented superficial epidermis is peeled away to expose otocyst (edges indicated by arrows). Note how quickly it heals back in the few minutes between C and D. C: The unoperated otic placode at stage 26 (surrounded by arrowheads). D: The posterior half has been removed: arrowheads point to anterior half, while the dotted line indicates half removed. pr, pharyngeal region; a, anterior; d, dorsal (same orientation in B,C,D); A and B are at same original magnification; C and D are at same original magnification.

Mirror Duplications

Why do anteroposterior half ablations result in mirror image duplications? The simplest explanation might be that, in the case of posterior half ablations, the tissue that would respond to Hh signaling is being eliminated and the resulting regenerated ear only receives anterior identity signals. For anterior half ablations, the reverse would be true, although the identities of the molecules involved in anterior patterning are not known. These results have implications for how many signals are necessary to pattern the inner ear and suggest that at least two signals are required for anteroposterior patterning. A one-signal model would require a gradient of activity and should result in normal, if smaller, ears. That mirror image duplications can still form, even after 2/3 ablations, suggests that prospective patterning centers reside in or adjacent to the anterior and posterior-most tips of the otic placode. These patterning signals could be intrinsic and/or extrinsic to the inner ear. For example, the Hh signal for inner ear patterning in zebrafish, chicken, and mouse likely arises from adjacent tissues (Hammond et al.,2003; Bok et al.,2005; Riccomagno et al.,2005).

Although anterior versus posterior signals can explain why the regenerated half has the same identity as the original half, it less clear why the duplicated inner ear structures are in mirror image to those in the original half. One possibility is that, during regeneration, a central anteroposterior boundary could result in a third signal that subsequently polarizes each half. Another alternative is that the endogenous and reconstituted signals each produce lateral inhibitions that cause them to be distally located, resulting in a mirror image pattern. In the latter case, the anteroposterior boundary is not functionally important. Whereas our middle third ablations might suggest that a central anteroposterior boundary is not required for normal development, we cannot rule out the possibility that, after the middle third is removed, a new anteroposterior boundary is being reconstituted by the remaining anterior and posterior signals.

Our discussion up to this point has suggested that regional identity and axial polarity are the same, but this does not have to be the case. The establishment of polarity may be occurring first in the remaining placode and only later is the identity of that tissue in terms of anterior or posterior structures established. A key assumption of most patterning models is that the regenerated inner ear is only derived from the remaining placode and not surrounding tissues as we confirmed for our half ablations (Fig. 5). This suggests that, after an ablation, a “new” anterior or posterior edge is established along the cut portion of the remaining placode. It is necessary for the partially ablated placode to maintain or reestablish anterior or posterior polarity whether it is going to result in a normal or mirror image pattern. While the same molecules involved in anterior and posterior identity may also be involved in axial polarity, they could also be different.

The key manipulation for generating mirror duplications appears to be removal of half or more of the ear. Although two-thirds ablations can also produce mirror duplications, it is in smaller numbers when partial duplications are included. Also it is critical that the half removed for mirror duplications to occur has to be the anterior or posterior and not the dorsal or ventral halves. When dorsal or ventral halves are ablated, no mirror duplications occur and the abnormal phenotypes are quite different (Forristall and Collazo, unpublished data) but similar to the results of dorsal and ventral half ablations done in salamanders (Kaan,1926).

Fate map studies in frog indicate that cell mixing is occurring during the stages we did our ablation, yet our half ablation results are also consistent with a compartment model of inner ear development in that the remaining half often could only regenerate tissues with its own identity (Brigande et al.,2000; Kil and Collazo,2001). These two results may seem to contradict, but our ablations and the fate map experiments address different questions that are not comparable. The fate map experiments are looking at normal development, while the ablations are challenging the fate of the remaining tissues. The two results can be reconciled for example if we assume that cells in the anterior half that would normally contribute to posterior structures instead form anterior structures in the mirror duplicated ears. Also, many of the ears in the half ablations were normal, suggesting that cells with posterior identity could reside in the anterior half. As a review of the inner ear fate map literature suggests, genes with complex and dynamic expression patterns may also be important for inner ear patterning during development, which would be consistent with the extensive cell mixing observed (Kil and Collazo,2002).

Hedgehog Inhibition Results in Mirror Duplications

Almost all the components of the Hh signaling pathway are expressed in or around the developing inner of Xenopus. In addition to Sonic Hh (Shh), Xenopus has Banded (Bhh) and Cephalic hedgehogs (Chh), homologues of mammalian Indian and Desert hedgehogs, respectively, of which only Bhh is expressed directly in the otocyst (Ekker et al.,1995). Although Shh is not expressed in the ear, it is expressed in adjacent tissues. Also, the Hh receptor and transducers are expressed in the developing inner ear of Xenopus (Takabatake et al.,2000; Koebernick et al.,2001).

Our results inhibiting Hh signaling in Xenopus produced the same results seen in zebrafish: mirror anterior duplications. Whereas Hh signaling in zebrafish plays an important role in anteroposterior patterning, in mouse and chicken, it appears to play a different role. Loss of Hh signaling in mouse, as seen in the Shh knockout, results in loss of ventral molecular markers and structures, such as the cochlea and cochleovestibular (otic) ganglion, whereas ectopic activation of Hh signaling dorsally results in down-regulation of dorsal markers and the turning on of some ventral markers (Riccomagno et al.,2002). Hh signaling is also important for inhibiting the expression of more dorsal genes, such as Wnt, ventrally (Riccomagno et al.,2005; Ohyama et al.,2006). Similar morphological defects to those observed in mouse are seen after loss of Shh activity in chicken embryos (Bok et al.,2005). The mouse and zebrafish studies are not strictly comparable as the zebrafish study required that two Hedgehogs, Shh and Tiggy-winkle Hh (Twhh), be knocked down before any mirror duplications were noted. The strongest duplications were found to occur in the strongest inhibitor of Hh signaling in zebrafish, the Smo mutant, whose inner ear phenotype in mouse has not been described because it does not survive long enough (Zhang et al.,2001). Our data in Xenopus combined with the findings in zebrafish suggest that, during the course of evolution, there may have been a change in the role of Hh signaling in ear development from promoting posterior identity to promoting ventral identity in amniotes.

Gain and loss of function experiments in Xenopus suggested that Hh signaling was involved in otic placode induction (Koebernick et al.,2003). Either activation or inhibition of Hh signaling produced the same results, ectopic otic vesicles. One reagent used to inhibit Hh signaling, cyclopamine treatment, did not produce ectopic otic vesicles but instead enlarged the endogenous otocyst. Later work from the same group showed that the effects on otic induction may have been due to some of the reagents used, specifically Hip, disrupting Fgf and Wnt signaling (Cornesse et al.,2005), two pathways known to be involved in otic placode induction (Groves and Bronner-Fraser,2000; Noramly and Grainger,2002). Therefore, we cannot rule out that the mirror duplications we observed could be due to Hip inhibiting other cell signaling pathways such as Fgf and Wnt.

What is interesting about the results in zebrafish is that the mirror image duplications are the results of not just creating anterior structures in the posterior half (in the case of anterior duplications) but also of inhibiting posterior structures and gene expression. This suggests that the creation of mirror image duplications is a two-step process first involving the suppression of genes and structures in that half and then the promotion of genes and structures with the new identity. However, the percentage of embryos with complete duplications after Hh signaling was disrupted is smaller than the percentage seen after our partial ablations. This finding would suggest that, while Hh signaling is an important component of anteroposterior patterning, other molecules are likely to be involved. Our half ablations will provide a powerful assay for determining other molecules potentially involved in anteroposterior patterning by providing a ready source of material for further molecular analyses.


Experimental Manipulation of Otic Tissue

Xenopus laevis eggs were fertilized in vitro and staged according to the normal table of Nieuwkoop and Faber (Nieuwkoop and Faber,1967). The placode is first visible morphologically as early as stage 21 (Nieuwkoop and Faber,1967; Hausen and Riebesell,1991). The otic placode begins to invaginate at stages 22–23 and begins to form a vesicle at stage 24 but the otic vesicle does not completely separate from the surface ectoderm until stage 28 (Nieuwkoop and Faber,1967; Schlosser and Northcutt,2000).

Embryos were chemically dejellied for 10 min in a 2% cysteine-HCl solution in rearing solution (20 mM Instant Ocean: a commercial aquarium salt) at pH 8.2. For stages before hatching, the vitelline membranes were removed with sharpened watchmaker forceps (Dumont #5). Living embryos were anesthetized with tricaine methanosulfonate (Finquel), in rearing solution, at concentrations ranging from 1:10,000 to 1:2,500. To better visualize the otic placode (stages 21–27) and otocyst (stages 28–31), the pigmented epidermis was peeled away using quartz glass micropipettes pulled into fine needles using a P-2000 micropipette puller (Sutter Instrument Co.). Removing this epidermis does not disturb ear development (Kil and Collazo,2001); it heals quickly, and this most superficial epidermis does not contribute any tissue to the inner ear at placode or otocyst stages (Hausen and Riebesell,1991; arrows in Fig. 7C,D). After exposure, the placode or otocyst was ablated in total, half (anterior or posterior), or thirds leaving either nothing or the unablated portion (Fig. 7). Immediately after surgery, each partial ablation was scored poor, fair, good, or excellent, and only those embryos with good or excellent surgeries were analyzed further. These ablations were done with quartz micropipettes (described above), an eyelash knife, and/or Dumont #5 forceps. A small subset of ablations were also accomplished by aspiration with a mouth pipette (Sigma A-5177) attached to a micropipette broken to the appropriate tip diameter. Results were similar for all methods of removal used. Pictures of the partial ablations were taken on a Zeiss M2Bio dissecting microscope using a color Zeiss Axiocam and the Axiovision 2.0 software.

To make certain that the otocyst was completely removed; we practiced extirpating the placode on one side of albino embryos and fixed them immediately after the operation for whole-mount in situ hybridization (protocol below; Fig. 7A,B). As a probe, we used the Xenopus homeobox gene Distal-less 3 (X-dll3; actually Dlx5; Stock et al.,1996), which is strongly and globally expressed in the otic placode and otocyst (Papalopulu and Kintner,1993). We have found that X-dll3 is expressed in prospective otic regions well before placode formation (stage 13, which is just after gastrulation), making it one of the earliest otic genes expressed in Xenopus, although not earlier than the transcription factors Sox9 and Pax8 (Saint-Germain et al.,2004). These ablations were done until 100% of labeled embryos were completely missing their inner ears on the operated side, indicating that the experimenter had achieved proficiency. Only when the experimenter demonstrated and believed that he or she was proficient did we proceed to the actual experiments that were scored.


To determine the tissue source of regenerated ears, grafting experiments of labeled donor ears into unlabeled hosts were undertaken. To generate labeled donors, fertilized eggs of Xenopus were injected with a fluorescent label, lysinated rhodamine dextran (LRD; Molecular Probes D-1817), at a concentration of 50 to 100 mg/ml at the one-cell stage. Rhodamine-labeled anterior or posterior half placodes were transplanted into unlabeled animals whose placode had been completely removed. Animals were allowed to develop to Nieuwkoop and Faber stage 47–49 at which time any regenerated form was analyzed as described below. Regenerated tissue was scored for the presence or absence of rhodamine labeling.

Labeling Hair Cells In Vivo

Labeling of hair cells was done as previously described (Kil and Collazo,2001). Briefly, inner ears of stage 47–49 tadpoles, or any structure that had regenerated in its place, were injected with a solution of the styryl fluorescent dye 4-Di-2-ASP or FM1-43 (Molecular Probes) to visualize hair cells (Meyers et al.,2003). This process was accomplished using quartz micropipettes (described above) back-filled with the vital dye and a pico-injector, PLI-100 from Harvard Apparatus. At these stages, many of the sensory hair cells have differentiated and can be labeled with the dye. The five sensory organs (anterior, lateral, and posterior cristae and maculae of the utricle and saccule) visible at these stages and the two otoliths associated with these maculae were compared with the control side. Otoliths were scored before injection, as the injection process sometimes moved the calcium carbonate crystals of which they consist. Other sensory organs, like the amphibian papilla and the basal papilla, develop later (stage 50+; Nieuwkoop and Faber,1967; Quick and Serrano,2005) and so were not scored during our manipulations.

Imaging and Scoring of Manipulated Inner Ears

Living embryos were prepared for imaging by being placed on a bed of 2% Bactoagar just after injection at stages 47–49. Tadpoles are transparent at these stages, facilitating observation. Labeled cells were visualized on either a Zeiss Axiophot2 Mot epifluorescence microscope or Zeiss LSM 410 Scanning Laser Confocal microscope. Long working distance air or water objectives ranging up to a magnification of ×40 were used. Data were recorded digitally from a light-intensifying camera (Hamamatsu SIT) using MetaMorph 3.5 (Molecular Devices) on the epifluorescence microscope. Any regenerated structure was scored for the presence or absence of sensory cells, otoliths, semicircular canals, and the pattern these structures formed. Ears on the unoperated side were always normal. Adobe PhotoshopTM (Versions 6.0 to CS) was used to view and process images.

Whole-Mount In Situ Hybridization

The method of Knecht and workers (Harland,1991; Knecht et al.,1995) was used to perform whole-mount in situ hybridization (ISH). Antisense and sense probes were made to X-dll3. The sense probe served as a negative control. Pictures of the ISH were taken with a ProgRes 3012 digital color CCD camera using the Roche Diagnostic software running under Windows for Workgroup 3.11. Images were collected through either a Zeiss SV11 dissecting microscope or Zeiss Axioplan upright compound microscope.

Injection of Hip mRNA

Injection of mouse Hedgehog-interacting protein 1 (Hip) was used to block Hh signaling as has been previously described (Koebernick et al.,2003). Previous authors have found that Hip gives the same phenotype as the endogenous Xenopus hip (Cornesse et al.,2005). Hip mRNA was injected with the same apparatus described above for styryl dye injection into the ear. One cell at the two-cell stage was injected with 4 to 9 ng of mRNA, resulting in half the animal being affected, leaving the other half as a control. In initial experiments, a small amount of fluorescent lysinated rhodamine dextran (10,000 molecular weight, Molecular Probes) or GFP mRNA was co-injected so the experimental side could be identified by its fluorescence. The smaller eye phenotype previously described after Hip injection in Xenopus (Cornesse et al.,2005), always correlated with the labeled side, so we used this phenotype to identify the affected half in some later experiments. Embryos were scored by labeling the hair cells as described above and counting the number of hair cells in the anterior and posterior cristae. Statistical analyses were done using the paired student's t-test in Microsoft Excel (Office 2000) and the Web site http://statpages.org for unpaired t-tests.


We thank Drs. Olivier Bricaud, Donna Fekete, Caryl Forristall, Andy Groves, and Sung-Hee Kil for stimulating discussions and advice during our studies. We also thank Olivier Bricaud, Andy Groves, and two anonymous reviewers for helpful comments on earlier versions of this study. A.C. was funded by the NIH and NIDCD, and E.H.W. was supported by grants from the Deafness Research Foundation and Quota International. We also thank Drs. Katja Koebernick and Nancy Papalopulu for the mouse Hip and X-dll3 plasmids, respectively.