Temporal coupling between specifications of neuronal and macular fates of the inner ear

Abstract

The inner ear is a complex organ comprised of various specialized sensory organs for detecting sound and head movements. The timing of specification for these sensory organs, however, is not clear. Previous fate mapping results of the inner ear indicate that vestibular and auditory ganglia and two of the vestibular sensory organs, the utricular macula (UM) and saccular macula (SM), are lineage related. Based on the medial-lateral relationship where respective auditory and vestibular neuroblasts exit from the otic epithelium and the subsequent formation of the medial SM and lateral UM in these regions, we hypothesized that specification of the two lateral structures, the vestibular ganglion and the UM are coupled and likewise for the two medial structures, the auditory ganglion and the SM. We tested this hypothesis by surgically inverting the primary axes of the otic cup in ovo and investigating the fate of the vestibular neurogenic region, which had been spotted with a lipophilic dye. Our results showed that the laterally-positioned, dye-associated, vestibular ganglion and UM were largely normal in transplanted ears, whereas both auditory ganglion and SM showed abnormalities suggesting the lateral but not the medial-derived structures were mostly specified at the time of transplantation. Both of these results are consistent with a temporal coupling between neuronal and macular fate specifications.

Introduction

During early stages of inner ear development in both chicken and mice, cells within the otic epithelium can be categorized into neuronal, sensory and non-sensory fates. Based largely on gene expression patterns, neuronal and sensory precursor domains overlap in the anterior otic cup region and are collectively named as the neural-sensory competent domain (NSC) (Fekete and Wu, 2002). Neuronal-fated cells delaminate from this NSC to form the neurons of the cochleovestibular ganglion (CVG), which later divides into the vestibular and auditory ganglion. Cells that remained in the NSC after neuroblast delamination are thought to split into various sensory organs over time (Fekete and Wu, 2002).

In the chicken otic placode, cell-tracing experiments using lipophilic dye suggest that the sensory cells of an individual sensory organ and its innervating neurons are derived from the same region of the placode (Bell et al., 2008). More specifically, cell lineage studies using replication incompetent retrovirus in chicken indicate that neurons of the CVG and cells within the utricle share a common lineage (Satoh and Fekete, 2005). Furthermore, genetic fate mapping of the neuronal lineage in the mouse inner ear using the Neurogenin1-cre (Neurog1-cre) strain indicates that the vestibular and auditory neurons share a common lineage with the two maculae, utricular macula (UM) and saccular macula (SM) (Raft et al., 2007). However, the relationships between the neuronal and macular fate specifications are not known.

Among the three primary fates of the inner ear, the neuronal fate is likely to be the first to be specified, since the earliest morphological change evident in the shallow otic cup is the delamination of neuroblasts from the anterior-ventral region (Alvarez and Navascues, 1990; Carney and Couve, 1989). Within the NSC, there are also spatial and temporal differences in the generation of neuronal subtypes: the lateral NSC gives rise to vestibular neuroblasts (Fig. 1A, red colors) at an earlier time than the medial NSC gives rise to auditory neuroblasts (Fig. 1A, purple colors) (Bell et al., 2008; Koundakjian et al., 2007). While these spatial and temporal differences suggest that neuroblasts may have already acquired some of their identities prior to exiting from the otic epithelium, the timing of specification for each neuronal is not clear. Based on the evidence of shared lineage between neurons and the two maculae in both chicken and mice (Raft et al., 2007; Satoh and Fekete, 2005), we hypothesized that location within the NSC not only dictates neuronal subtype fate but also specifies the type of macula that develops subsequently in the same region. This hypothesis predicts that the lateral NSC specifies the fate of both vestibular neurons and the laterally-positioned UM, whereas the medial NSC specifies the formation of auditory neurons and the medially-positioned sensory organs such as the SM (Fig. 1). More importantly, this model predicts that specification of both neuronal and macula fates are coupled and this coupled specification occurs early, possibly at the time of neuroblast delamination.

Figure 1.

Schematic diagram of an otic cup (A), otocyst (B), and a mature inner ear (C). (A) and (B) illustrate the consensus that the vestibular neurogenic region (red color) is located lateral to the auditory neurogenic region (purple color), where respective neuroblasts delaminate. Based on the spatial locations of the sensory organs in the mature inner ear, we propose that after neuroblasts delamination, the lateral neurogenic region develops into the utricular macula (red color, C) and the medial region develops into the saccular macula (purple color, C). For simplification, the vestibular and auditory ganglia are not shown in (C). Abbreviations: ASC, anterior semicircular canal; CD, cochlear duct; ED, endolymphatic duct; ES, endolymphatic sac; LSC, lateral semicircular canal; PSC, posterior semicircular canal; SM, saccular macula; UM, utricular macula. Orientations: A, anterior; D, dorsal; L, lateral; M, medial; P, posterior; V, ventral.

To address the above hypothesis, we first fate mapped the lateral rim of the neurogenic region in the chicken otic cup using lipophilic dyes (Fig. 2). Our model predicts that labeling this region, which presumably gives rise to the vestibular neurons (Bell et al., 2008), will also label the UM (Fig. 1). After identifying a region that exclusively gives rise to the vestibular neurons and lateral sensory organs including the utricle, we surgically inverted the medial-lateral (M/L) axis of a chicken otic cup in ovo and asked whether inverting the relative positions of the lateral and medial NSC domain simultaneously affected both the neuronal and macular fates in the corresponding region. We reasoned that if neuronal fates are established (i.e. specified) prior to delamination and if neuronal and macular fates are indeed coupled, neither of these fates should be affected by this axial inversion.

Figure 2.

Neurod1 expression in the chicken otic cup. (A-D) Dorsal and lateral views of an otic cup at 19ss. (C) and (D) are higher magnification of the otic cup shown in (A,B). By aligning the ventral tip of the otic cup (C, arrow) as the 6 o'clock position of a clock face, the Neurod1 domain at the rim of the otic cup always falls between 4 to 6 o'clock positions (D). (E) Schematic diagram of the neurogenic domain, its delaminating neuroblasts and locations of dye injections. Scale Bars: 100μm.

Since the otic cup is slightly deepened at the time of transplantation, the surgery was effectively a dual inversion of both M/L and dorsal-ventral (D/V) axes. Our results indicate that the identities of both lateral NSC-derived structures, the vestibular ganglion and UM were largely unchanged after this dual axial inversion, suggesting that these fates were specified at the time of transplantation. In contrast, the identity of the structures derived from the medial NSC, the auditory ganglion and SM, were affected, suggesting that these structures were plastic and not yet specified at the time of transplantation but that timing of their specification may well be coupled to each other. Taken together, our results support the hypothesis of a lateral to medial timing of the UM and SM specification, which corresponds to the timing of lateral to medial vestibular and auditory neuronal fate specifications, at stages that is well ahead of any overt sensory differentiation.

Materials and Methods

Fate mapping and Transplantation Surgery

Fertilized chicken eggs (B&E farm, Maryland) were incubated at 39°C for various days and staged according to Hamburger and Hamilton (HH; 1992). Incubated eggs were windowed and injected with black India ink (Pelican) underneath the embryo to enhance contrast. For fate mapping study, at HH St13 (19-20 somite stage (ss), Embryonic day 2 (E2)), lipophilic tracers, CM-DiI or FAST DiO (Molecular Probes, # C-7000 and D-3898), was injected at designated locations on the rim of the otic cup according to a clock face grid (Fig. 2; (Brigande et al., 2000)). Working solution for both dyes was prepared by 1:10 dilution of CM-DiI (1mg/ml) or DiO (2mg/ml) stock solution prepared in 70% dimethylformamide.

For transplantation surgeries, an E2 donor embryo was transferred to a Sylgard dish. The left otic cup was injected with dyes and isolated using a tungsten needle and a homemade microblade. Then, the right otic cup of an age-matched host embryo was removed and replaced with a donor's left otic cup aligned to the same anterior-posterior axis as the host. Digital photographs were taken before, during, and after the surgery for documentation. After surgery, Tyrode's solution was applied to the embryo before sealing the egg and returning to the incubator until harvest.

Summary of samples processed

Various locations of dye were tested initially and two injection positions, 4 o'clock and 5 o'clock on the rim of the otic cup, were chosen for reproducibility in targeting the neurogenic domain. The survival rates for the operated embryos were around 80%, 50% and 30% after 24, 48, and 96 hours, respectively. A total of 58 embryos were used for the fate mapping study, among which 29 embryos were successfully harvested at 24 (E3; n=11; 6/11 dual-labeled), 48 (E4; n=6; 5/6 dual-labeled), and 96 (E5.5; n=12, 4/12 dual-labeled) hours post-surgery.

For the transplantation study, a total of 52 embryos were harvested, 19 of these were from transplants conducted at various ages as pilot studies to test the feasibility of the transplantation procedures. After we determined that 19-20ss (St13) otic cups were optimal for transplants, we successfully harvested 33 embryos at 24 (E3, n=11), 48 (E4, n=13) and 96 (E5.5, n=9) hours post-surgery. Analyses were only conducted on specimens with good dye positioning and in situ hybridization results.

Probes and in situ hybridization

To identify molecular markers specific for the vestibular and auditory ganglion, we focused on several candidates according to the gene profiling results described for the mouse such as Synaptophysin (Syp), Gata3, Tbx3, Neurexin 3, and NetrinG1 (Appler et al., 2013). We determined that Syp and Gata3 are the best markers for distinguishing the respective vestibular and auditory ganglion for the stage of analysis at E5.5. Neurexin3 and NetrinG1 expression patterns were weak and not restricted to a specific ganglion. In contrast, Tbx3, while not exclusively expressed in either ganglion, is a good marker for the lagenar macula and utricular macula. Information for generating chicken anti-sense RNA probes for Crabp1, Gata3, Lunatic fringe (Lfng), Neurod1, Neurofilament 68 (NF68), Ngfr, Msx1, Pax2, Syp, and Tbx3 can be found as described (Wu and Oh, 1996) or provided upon request. Section and whole mount in situ hybridization were conducted as described (Wu and Oh, 1996).

Quantification of dye-labeled cells in the ganglia

To analyze the distribution of lipophilic dye labeled cells in the auditory and vestibular ganglion, numbers of dye-labeled cells (in pixels) were estimated using square paintbrush on low-power micrographs in Photoshop CS3. Gata3 and Syp hybridization signals were used to identify the auditory and vestibular ganglion, respectively. A minimum of 7 sections evenly interspersed per specimen were counted, which spanned a total thickness of 400 to 450μM.

Results

Gene expression patterns and fate mapping of the neurogenic region

In the developing chicken inner ear, neuroblasts start to delaminate from the otic epithelium at E2 (16-17ss), and this process continues beyond E4 (Bell et al., 2008). To address whether the vestibular neurogenic region also gives rise to the UM, we first fate-mapped the lateral rim of the neurogenic domain at the 19ss, which presumably gives rise to the vestibular neurons (Bell et al., 2008). At this age, Neurod1-positive neuroblasts are delaminating from the anterior ventral quadrant of the otic cup (Fig. 2A-B, E). We divided the rim of the otic cup into the face of a clock as described previously (Brigande et al., 2000), marking the ventral tip of the otic cup as 6 o'clock (Fig. 2C, arrow) and the widest points of the otic cup along the anterior-posterior axis as 3 and 9 o'clock, respectively (Fig. 2D). Using these criteria, the majority of the Neurod1-positive cells lie between 4 and 6 o'clock positions (Fig. 2C).

Figure 3 illustrates an inner ear that was injected with DiI and DiO at the respective 4 and 5 o'clock positions (Fig. 3A-B) within the projected Neurod1-positive region (Fig. 2D-E). Both 4 and 5 o'clock dye injections in the otic cup were associated with the Neurod1-positive neurogenic region by the otocyst stage (Fig. 3E-H′), confirming proper targeting of the neurogenic region at the otic cup stage. However, 4 o'clock injected cells were always located more medially than cells injected at the 5 o'clock position (E3, n=6; E4, n=5; dual-labeled samples). Additionally, the Neurod1-positive neurogenic domain in the otocyst (Fig. 3E-H, blue bracket) can be delineated into a lateral Crabp1-positive and a medial Pax2-positive region (Fig. 3D). Dye injected at the 5 o'clock position of the otic cup consistently labeled a lateral, Crabp1-positive and Pax2-negative part of the neurogenic region at E4 (Fig. 3E-H′, green bracket; n=8, E3 and E4), whereas dye injected at the 4 o'clock position labeled more medial (Crabp1-positive but also weakly Pax2-positive) cells (Fig. 3E-H′, red bracket; n=6, E3 and E4). Dye-labeled cells from both positions took on a medial trajectory after leaving the epithelium. Five o'clock injections that were broader or closer to the 5:30 o'clock position yielded similar results as 5 o'clock injections and these results were grouped together with 5 o'clock injections for analyses at these time points (n=4/8) as well as subsequent analyses of 96-hour harvests (n=3/9, Table 1). Dye injections closer to the 3 o'clock position often resulted in labeled cells in the dorsal region of the `otocyst and these samples were not analyzed further.

Figure 3.

Fate mapping the lateral edge of the neurogenic region of the otic cup. (A) Schematic diagram showing coordinates of a 19ss otic cup. (B) An otic cup injected at the rim with DiI (red) and DiO (green) in the 4 and 5 o'clock position, respectively. (C) Drawing of an otocyst indicating the level of sections shown in (E-H) of the neurogenic domain (dotted region) at E4, 48 hours after injections. (D) Schematic summary of gene expression patterns and the approximate locations of dye-labeled regions in (E-H′) are shown. (E-H) Adjacent sections showing hybridization signals of Neurod1, Pax2, and Crabp1 transcripts at E4. The Neurod1-positive neurogenic region (blue bracket) is delineated by a lateral, Crabp1-positive and a medial, Pax2-positive region. (E′-H′) are the same sections as (E-H) showing DiI and DiO labeling in the otic epithelium 48 hours after injections. DiO injected at 5 o'clock labels a lateral, Pax2-/Crabp1+ region in the neurogenic region (green bracket), whereas DiI injected at the 4 o'clock position labels a neurogenic region that is Crabp1+ but also weakly Pax2+ (red bracket). Scale Bars: 100 μm.

Table 1.

Summary of lipophilic dye labeling in control ears at 96 hours post surgery.

	Dye-labeled cells at E5.5
Location of otic cup injections	Ganglion		Utricle		Saccule	Crista			Between CC and Saccule	ES/ED
						AC	LC	PC
4 o'clock	Robust and in AG VG		In Sensory	Lateral of sensory	n=4/7	n=0/7	n=0/7	n=4*/7	n=4/7	n=3/7
	n=6/7		n=7/7	n=7/7
5 o'clock	Only trace amount in AG	Robust on lateral of VG	Medial, close to saccule	Lateral, close to lateral crista	n=0/9	n=2/9	n=7/9	n=1*/9	n=1/9	n=5/9
	n=8/9	n=9/9	n=6/9	n=7/9

^*Only trace amount of labeling found.

Analyses of dye-labeled cells in the inner ear 96 hours after injections

By 96 hours after dye injection, the embryos were staged closer to E5.5 (HH St28) rather than E6. At this age, the auditory and vestibular ganglia are distinguishable entities and can be identified based on the respective expression of Gata3 and Syp (Fig. 4B′-C′, D; see Materials and Methods). The DiO-labeled green cells that were injected at the 5 o'clock position of the otic cup and located in the Crabp1-positive lateral neurogenic domain at E4 (Fig. 3D, G, G′, green bracket) were primarily associated with the vestibular ganglion by E5.5 (Fig. 4A-D; Table 1). In contrast, DiI-labeled red cells in the 4 o'clock region of the otic cup, which weakly expressed Pax2 in addition to Crabp1 at the otocyst stage (Fig. 3D, F, F′, red bracket), were located in both vestibular and auditory ganglia at E5.5 (Fig. 4A-D; Table 1). The distributions of dye-labeled cells in the ganglia from both injection locations were quantified (Fig. 5). Together, these results indicate that we have identified an epithelial location (5 o'clock position) in the otic cup that preferentially gives rise to cells in the vestibular ganglion. In addition, injections slightly dorsal to this location (4 o'clock injections) resulted in labeling in a location that was more medial by the otocyst stage and gave rise to cells in both auditory and vestibular ganglia in a 3:2 ratio (Fig. 5). The inclusion of auditory ganglion labeling with more medially located dye is consistent with the known relationships between auditory and vestibular neurogenic regions within the NSC (Bell et al., 2008; Lawoko-Kerali et al., 2004).

Figure 4.

Dye-labeled cells in the inner ear and ganglia 96 hours after otic cup injections. (A) A 20ss otic cup injected with DiI and DiO at 4 and 5 o'clock, respectively. (B-D) Adjacent sections of an E5.5 inner ear 4 days after the injections in (A). (B,C) DiI and DiO labeling in the ganglia. DiO-labeled green cells at 5 o'clock in the otic cup are associated with the Syp-positive, vestibular ganglion (VG, B-C′). In contrast, 4 o'clock DiI-labeled red cells are associated with both the vestibular and auditory ganglia (AG), which are positive for Syp and Gata3, respectively (B-D). (B′) and (C′) are the same sections as (B) and (C), respectively. (E-I) An inner ear injected with DiI and DiO at 4 and 5 o'clock of a 19ss otic cup (E) and harvested at 96 hours later at E5.5 (F-I). (F) Lfng is expressed in all sensory epithelia. Five o'clock injections at the otic cup stage consistently label the lateral crista (LC) and non-sensory regions on both sides of the utricular macula (UM) (F-I, asterisk), whereas 4 o'clock injections are associated with the Crabp1+ UM (G-H, white arrows) and Pax2+ saccular macula (SM) (G, G′, arrowheads). Red arrows in (I) indicate the 4 o'clock dye associated with the endolymphatic duct. The non-sensory region adjacent to SM is Crabp1-positive (H, yellow brackets). Scale Bars: 100 μm.

Figure 5.

Distribution of dye-labeled cells in the auditory and vestibular ganglia 96 hours after lipophilic dye injections. In controls, 4 o'clock dye-labeled cells are distributed in a 3:2 ratio between auditory (AG) and vestibular ganglia (VG) (n=3), whereas 5 o'clock dye-labeled cells are mostly associated with the VG (n=3). In contrast, 5 o'clock dye-labeled cells in transplanted ears resulted in a 1:2 ratio between AG and VG (n =6). Error bars represent standard deviation. Distribution of dye-labeled cells in individual specimens are also listed in Supplementary Table 1.

Next, we investigated whether the dye-labeled cells at the two injected positions described above were also associated with the two maculae in the inner ear as predicted by our hypothesis. By E5.5, all the sensory organs in the inner ear can be clearly identified based on Lfng expression (Fig. 4F). Pax2, which labels the medial neurogenic region at the otocyst stage (Fig. 3F), is mainly expressed in the SM by E5.5 (Fig. 4G′), whereas Crabp1, a marker for the vestibular neurogenic region, is primarily expressed in the UM (Fig. 4I) and non-sensory regions flanking the SM (Fig. 4I, yellow brackets). Four o'clock injections in the otic cup resulted in red-labeled cells in both the maculae of the utricle (arrows) and saccule (arrowheads) at this stage (Fig. 4E-I; Table 1). In contrast, 5 o'clock injections labeled more lateral structures of the inner ear: the lateral crista and non-sensory regions of the utricle spanning both lateral and medial side of the UM (Fig. 4E-I, asterisk; Table 1) but not the saccule. The non-sensory region between the lateral crista and UM has been shown to be lineage related to the neurogenic region in both the chicken and mouse (Raft et al., 2007; Satoh and Fekete, 2005).

Although dye labeling was associated with other sensory organs as well (Table 1), a majority of epithelial labeling was associated with the two maculae, which supports the evidence of shared lineage among cells in the maculae and ganglia (Raft et al., 2007; Satoh and Fekete, 2005). Furthermore, the concomitant inclusion of auditory neurons and saccule with more medial labeling of the NSC (4 o'clock injections) is correlated with our hypothesis linking specification of auditory neurons with the saccule and specification of vestibular neurons with the utricle. However, some labeled cells from a 4 o'clock injection were often found to intersperse between two domains of 5 o'clock labeled cells (Fig. 4G-H, asterisks; n=5/9), suggesting a dynamic cell movement/displacement within the neurogenic domain during inner ear morphogenesis.

Axial inversion of the neurogenic region

To investigate whether the vestibular and auditory neuronal fates were specified at the time of delamination, we inverted the M/L axis of the developing otic cup in ovo with the intent of switching the position of the vestibular and auditory neurogenic regions. We then asked whether such axial inversion affected the specification of the neuronal fates. The transplantations were conducted at the 19-20ss (HH St13, E2), which is shortly after neurogenic transcripts such as NeuroM and NeuroD are first detectable in the otic epithelium (approximately 9 hours earlier at the 16ss (HH St12)) (Bell et al., 2008). The axial inversion was accomplished by replacing the right otic cup in a host embryo with a left donor otic cup that was first injected with DiO at the 5 o'clock position (Fig. 6A-B). The dye labeling served as a marker for the original vestibular neurogenic region. As a result of the surgery, the M/L axis of the transplanted ear was inverted relative to the M/L axis of the host embryo. Since the otic cup at 19ss is deepening, this inversion also resulted in a concomitant D/V axial inversion as evident by the 5 o'clock dye location in the donor ear translocated to approximately the 2 to 3 o'clock positions in the host (Fig. 6A-B). Consistently, the dye remained positioned dorsally on the host otocyst at 24 and 48 hours after surgery (Fig. 6C-D). Cryo-sections of 48 hour-specimens through the neurogenic region indicate that the dye-labeled cells at the 5 o'clock injection were medially displaced after this surgical manipulation (Fig. 6F′-H′), compared to control otocysts (Fig. 3E′-H′). Notably, the delaminated dye-labeled cells showed a slightly lateral trajectory (Fig. 6F′-H′) in contrast to the medial trajectory of controls (Fig. 3E′-H′). In addition, the dye-labeled cells within the epithelium remained associated with the weak Pax2 and Crabp1-positive Neurod1 domain (Fig. 6E-H, yellow bracket), suggesting that the molecular identity of the neurogenic region may already be fixed at the time of transplantation. In contrast, beyond the neurogenic region, re-specification of Pax2 expression in the medial region and Crabp1 in the lateral region were observed (Fig. 6F, H, double arrowheads), compared to normal ears (Fig. 3F-G).

Figure 6.

Medial-lateral inversion of the otic cup at 19-somite stage. (A) Schematic diagram illustrating the otic cup transplantation procedure whereby the right ear of a host embryo is replaced with a left ear of an age-matched donor. The DiO labeling at 5 o'clock of a counterclockwise clock face of the left donor otic cup at 19ss (B) is located in a relatively dorsal position after transplantation (C, D). The straight line in (D) indicates the level of sections in (F-H′). (E) Location of the dye-labeled region and summary of gene expression patterns of the neurogenic region in control and transplanted ears at E4. (F-H) Gene expression patterns of the transplanted inner ear in (B) after 48 hours at E4 showing a M/L inversion of the neurogenic domain (blue bracket) with Crabp1 expression domain (yellow bracket) medial to the Pax2 domain (purple bracket), compared to the control shown in Fig. 3E-H. Next to the Neurod1-positive region, there appears to be re-specification of Pax2 expression in the medial and Crabp1 in the lateral region (red arrowheads, F, H), in addition to endogenous expression patterns found prior to transplantation (arrows). (F′-H′) Pictures of the same sections as (F-H) taken prior to in situ hybridization, respectively. The dye-labeled cells remained associated with Pax2-/Crabp1+ region of the neurogenic domain in the transplanted ear. Scale Bars: 100 μm.

Transplanted ears with dual color labeling confirmed a similar M/L inversion of the otocyst (Supplementary Fig. 1). Additionally, in sham-operated transplants, in which the right ear of the host was replaced with a right donor ear, the 5 o'clock labeling remained in the lateral region of the neurogenic domain (Supplementary Fig. 2), similar to 5 o'clock injected non-transplanted controls (Fig. 3E-H′). Taken together, these results indicate that the left to right ear transplantations successfully inverted the M/L axis of the neurogenic domain. The inversion along the D/V axis was not characterized molecularly in the transplants due to lack of known molecular markers that distinguish this axis within the neurogenic domain.

The vestibular ganglion of transplanted ears shows similar position and molecular identity as controls

In a normal ear, the Syp-positive vestibular ganglion is located in a more dorsal and lateral position relative to the Gata3-positive auditory ganglion at E5.5 (HH St28; Fig. 7A-D, I). This M/L relationship seems to be maintained in the transplanted ear with Syp-positive region located more laterally than the Gata3-positive ganglion (Fig. 7E-H), despite the M/L inversion of the neurogenic domain in the transplanted ear at E4 (Fig. 6F-H′). Syp-positive fibers were also found adjacent to some vestibular sensory organs (Supplementary Fig.3E-F). Furthermore, DiO that was injected to the 5 o'clock position of a donor otic cup prior to transplantation appeared more widely distributed between the two ganglia but still were primarily associated with the Syp-positive ganglion in a AG:VG ratio of 1:2 (Fig. 7E-F′, n=6/6; Fig. 5, Supplementary Table 1), compared to over 90% of labels in VG of non-transplanted controls. Nevertheless, the combined gene expression and dye-labeling results confirmed that the correctly positioned vestibular ganglion in the transplanted ear was, at least in part, derived from the original vestibular neurogenic region of the donor ear. In contrast, the Gata3-positive auditory ganglion, though positioned medial to the vestibular ganglion as in controls, did not seem to extend ventrally by the cochlear duct (Fig. 7G-H, J; n=6/6), suggesting that the auditory ganglion may be compromised by the axial inversion.

Figure 7.

Distribution of dye-labeled cells in ganglia of transplanted ears 96 hours post surgery. (A-D) Gata3 (A,C) and Syp (B,D) are expressed in the medial auditory and lateral vestibular ganglion, respectively (AG, VG). (E-H) In transplanted ears, the M/L relationship between AG and VG is similar to controls, based on the expression pattern of Gata3 (E, G) and Syp (F, H). However, the AG does not extend ventrally to the cochlear duct (CD), compared to controls (C-D). (E′,F′) DiO labeling at 5 o'clock of the otic cup stage remains associated with Syp-positive VG at E5.5. (I-J) Schematic drawings showing levels of sections in control and transplanted inner ears, the ganglia are highlighted with pink color. Outlines of ganglia shown in this figure were independently verified with adjacent sections probed for neurofilament transcripts, NF68 (not shown). (E′) and (F′) are the same sections as (E) and (F), respectively. Scale Bars: 100 μm.

Identity of maculae in transplanted ears

Next, we investigated the identity of maculae, which are associated with the neurogenic region. The transplanted inner ears were malformed based on serial sections and 3D reconstruction analyses. Figures 7 to 9 illustrate two of the transplanted specimens. Most of the samples show a common chamber in the middle of the inner ear that opened up ventrally into the cochlear duct and dorsally into the vestibule.

Figure 9.

Sensory organ analyses of transplanted inner ears 96 hours post surgery. (A-E) Adjacent sections of a control E5.5 ear showing various Lfng-positive sensory organs: the anterior crista (AC, insets) and lateral crista (LC) are Tbx3- and Ngfr+, the basilar papilla (BP) is Gata3+ and Tbx3-, and the lagenar macula (LM) is Tbx3+ and Ngfr- (B, C, arrows). Both saccular macula (SM) and utricular macula (UM) is Ngfr- but UM is weakly Tbx3+. (F-J) In the transplanted ear at E5.5, both presumed UM (orange brackets) and SM (purple brackets) are Ngfr- but Tbx3+ (H, I). The presumed BP (blue brackets), which is Gata3+ (G) but it is Tbx3+ as well (H) and is fused to the SM (F, J). Additionally, at the distal end of the BP, an Lfng-positive sensory patch showing similar molecular identity as the LM, which is Tbx3+ and Ngfr- (H, I). (F′-J′) Same sections as (F-J) respectively, showing DiO labeling associated with: lateral region (arrowheads) of the UM (orange brackets), SM (purple brackets), BP (blue brackets), auditory ganglion (Gata3+, (G)) and vestibular ganglion. Scale Bars: 100 μm.

Using Lfng as a pan-sensory marker, two macula-like sensory organs were often located in the medial and lateral wall of the common chamber/cochlear duct (Fig. 8F-J; Fig. 9F-J). The criteria we used to identify individual sensory organs are as follow: Crabp1, Gata3 and Tbx3 positive cells defined the UM (Fig. 8A-E, 9A-E; (Lillevali et al., 2007)); Pax2 positive but Gata3, Crabp1, and Tbx3 negative cells defined the SM (Fig. 8A-E, 9A-E; (Chervenak et al., 2013; Li et al., 2004)); Gata3 positive but Crabp1 and Tbx3 negative cells defined the basilar papilla (BP, Fig. 8A-E, 9A-E). All six transplanted ears showed the lateral sensory organs expressing markers that represent a normal UM (Fig. 8F-J, 9F-J, orange brackets; Table 2; Supplementary Table 1). On the medial side, whereas the presumed SM (Table 2, n=4/6) expressed strong Pax2 (Fig. 8G-H, n=4/4) similar to the SM of controls (Fig. 8A-E), some also expressed weak Gata3 (Fig. 8I, Supplementary Table 1; n=1/4), Tbx3 (Fig. 9H; Supplementary Table 1; n=4/4) and Crabp1 (Supplementary Table 1; n=1/4), which are not present in controls (Fig. 8B,D and Fig. 9B,C). Therefore, these presumed SM showed a mixed macula identity in transplants and were present at a lower frequency than UM (Table 2; Supplementary Table 1). In addition, some of these SM were fused to a small patch of Gata3-positive basilar papilla (Fig. 9F-G, J, purple and blue brackets; n=2/4).

Figure 8.

Tissue morphology and distribution of dye-labeled cells in transplanted ears 96 hours post surgery. (A-E) In a normal E5.5 inner ear, Lfng is expressed in all sensory organs (A, E). Additionally, the utricular macula (UM) is Crabp1+ and Gata3+ (B, D), whereas the saccular macula (SM) is Pax2+, Crabp1+ and Gata3- (B, C, D). (F-J) In the transplanted ear at E5.5, the presumed UM (F-J, orange brackets) is also Crabp1+ and Gata3+ (G, I). The presumed SM (F-J, purple brackets) is Pax2+ and Crabp1- (G-H), similar to controls (B-C), but unlike controls, it is also Gata3+ (I). The location of the presumed UM is ventrally displaced relative to the presumed SM (F-J). (F′-J′) Fluorescent micrographs of the outlined area in (F-J) taken prior to in situ hybridization. DiO-labeled cells (green bracket) are located laterally in the utricle but overlap partially with the UM (F-G′, orange brackets). A-E and F-J are serial-sections. Scale Bars: 100 μm.

Table 2.

Frequency of identifiable sensory organs and 5 o'clock dye labeling in transplanted ears at 96 hours post surgery.

Presumed sensory organs	Ganglion		Utricle		Saccule	Lagena	Crista			BP	ES/ED
	A G	V G					AC	LC	PC
Confirmed by molecular markers*	6/6	6/6	6/6		4/6	5/6	5/6	1/6	6/6	3/6	ND
Presence of 5 o'clock dye labeling	5/6	6/6	Sensory	Non-sensory	Sensory		Sensory	Non-sensory		Sensory	Later al
			3/6	4/6	2†/4	0/6	3/5	2/6	0/6	2/3	4/6

^*Molecular criteria for sensory organ identification: AC and PC: Lfng+, Msx1+^§, Ngfr+^§, Gata3+ in cruciatum only, and Crabp1-. AG: Gata3+ and Syp–. BP: Lfng+, Gata3+, Ngfr-^§ and Crabp1 Unlike normal BP, these BP are Pax2+, Tbx3+. LC: Lfng+, Msx1+^§, Ngfr+^§, Gata3-, Pax2-, Tbx3- and Crabp1-. LM: Lfng +, Pax2+, Tbx3+, Gata3+, Msx1+^§, Crabp1- and Ngfr-^§; SM: Lfng+, Pax2+, Crabp1-, Msx1-^§, Ngfr-^§. Unlike normal SM, these SM are either Tbx3 or Gata3 positive. UM: Lfng+, Gata3+, Crabp1+, Tbx3+, Pax2-, Msx1-^§ and Ngfr-^§; VG: Syp+ and Gata3-.

^§Msx1 and Ngfr were probed in selected samples only.

^†Both of these specimens are Tbx3 positive.

To determine whether the UM was specified before transplantation, we traced the dye labeling in the transplants. In a normal inner ear, 5 o'clock injection resulted in labeled cells spanning both sides of the UM (Fig. 4G-H, asterisks). In the transplanted ears, a stream of dye-labeled cells was observed in the dorsal region of the sensory domain where the two maculae were sometimes fused (Supplementary Fig. 4C′-D′, green brackets). Labeled cells were also observed to spread ventrally along the medial and lateral side of the common chamber and extended into the cochlear duct (Fig. 9F′-J′; Supplementary Fig. 4E′). Similar to controls, not all 5 o'clock injections resulted in labeling in the utricle (Table 1 and 2; Supplementary Table 1). However, among the 4 transplants with utricular labeling, all showed non-sensory labeling similar to controls and 3 transplants showed additional labeling in the sensory tissue. In contrast, two of the transplanted specimens showed labeling in the SM, which was not observed in controls. Although the sample size is small and the morphogenesis of this region in the transplants is poor, it is possible that dye-labeling in the SM of transplants represents a fate change on some levels. Nevertheless, a majority of the labeled cells remained associated with the utricle-like sensory organ and its vicinity in the transplants. Thus, both the dye labeling and gene expression results support the notion that the UM is specified at the time of otic cup transplantation.

Despite the M/L axial inversion, the M/L relationship of the presumed SM (Pax2-positive) and UM (Crabp1-positive) in the transplanted ears appeared relatively normal. However, the position of the UM was ventrally displaced relative to the SM in two out of the four specimens that had both end organs (Fig. 8F-J). This was in contrast to the UM located dorsally to the SM in controls (Fig. 8A).

Sensory organ analyses of transplanted ears

Additional gene expression analyses were conducted to distinguish the dye-labeled maculae from other sensory organs such as the crista and lagena. Msx1 is a pan-marker for both cristae and lagenar macula (Wu and Oh, 1996), whereas Ngfr is exclusively expressed in the cristae (Fig. 9D) and Tbx3 is preferentially expressed in the lagenar macula and weakly in the utricle but not the cristae (Fig. 9C). Both SM and UM associated with dye in the transplanted ears did not express the cristae markers, Ngfr (Fig. 9I-J) and Msx1 (Supplementary Table 1), and only weakly expressed a lagenar macula marker, Tbx3 (Fig.9C). Collectively, these gene expression results confirmed that the dye-labeled sensory organs are most likely SM and UM and not crista or lagenar macula.

Two Ngfr-positive and Tbx3-negative sensory cristae were usually found in the anterior (Fig. 9F-J, AC; Fig. 10C; n=5/6) and posterior region (Fig. 10D; n=6/6) of transplanted ears. While the size of the anterior-located crista was sometimes bigger than controls (Fig. 10, AC, PC), only one crista-like sensory patch was usually found in the anterior suggesting that with the exception of one sample, the lateral crista was either absent or fused with the anterior crista in transplanted ears (Table 2). Additionally, a small Lfng-positive sensory patch was present within the ventral cochlear duct (Fig. 9F-J, arrows), which co-expressed lagenar macula markers Tbx3 (Fig. 9H) and Msx1 (not shown), and was negative for the crista marker Ngfr (Fig. 9I), resembling that of the lagenar macula in controls (Fig. 9A-D, n=5/6). A thin bundle of Syp-positive nerve fibers in the vicinity of the presumed lagenar macula could be traced to the ganglion as well (Supplementary Fig. 3E-F).

Figure 10.

Three-dimensional reconstruction of a control and a transplanted inner ear 96 hours post surgery. (A, C) Anterior and (B, D) posterior views. Various Lfng-positive sensory patches are shown in green. In the transplanted ear (C, D), the presumed saccular macula (SM) is located more dorsally than the utricular macula (UM) as compared to the control. The lateral crista (LC) and basilar papilla (BP) are absent in this transplanted specimen. The anterior and posterior cristae (AC, PC) in the transplant appear larger than those in the control. The endolymphatic duct (ED) in the transplanted ear is truncated. Scale bar: 200 μm.

Two rudimentary endolymphatic sac/endolymphatic duct (ES/ED) outpouches can be seen emerging from both medial and lateral regions of five out of six transplanted ears but the molecular identities of these two structures were not evaluated (Supplementary Fig. 3G-H, asterisks). However, dye-labeled cells were often found within ES/ED of controls (Table 1) and they were found to be associated with the lateral instead of medial ES/ED of transplants (not shown), suggesting the lateral ES/ED is the one specified prior to transplantation.

Discussion

Specification of the neuronal fates in transplanted ears

In this study, we investigated the timing of neuronal specification in the inner ear by inverting the primary axis of the otic cup at the time of neuroblast delamination. Our results showed that both vestibular and auditory ganglia in transplants formed fairly well in their respective lateral and medial locations and maintained their respective expression of Syp and Gata3. While Syp-positive fibers were observed to come in close contact with some vestibular organs (Supplementary Fig. 3E-F), the auditory ganglion failed to extend ventrally to the basilar papilla (Fig. 7). In addition, more than half of the dye labeling remained associated with the vestibular ganglion in transplanted ears (Fig. 5), suggesting that the original vestibular neurogenic region that was labeled in the donor and relocated medially, maintained its identity and gave rise to vestibular neurons after transplantation. These results suggest that at least the vestibular ganglion is mostly determined at the time of transplantation. If the fates of the neurogenic region were plastic at the time of transplantation, one would expect the medial-displaced dye-labeled region to give rise to auditory neuroblasts and the auditory ganglion should be the main ganglion labeled with dye. This was not the case and we excluded the possibility that the neurogenic region is plastic at the time of transplantation. However, approximately one-third of the dye-labeling was found to be associated with the auditory ganglion in the transplants, which was not observed in controls. This could represent a re-specification of some of the neuroblasts and/or artifacts generated by the dual-axial transplantation procedure. A partial re-specification is not too surprising since neuroblast delamination is an ongoing process at the time of transplantation.

Extrapolating from the notion that the vestibular ganglion is specified at the time of transplantation, one might expect the M/L relationship of the two ganglia to be inverted in the host embryo rather than being relative normal and similar to controls. The trajectory of delaminated neuroblasts towards the lateral direction (opposite to controls) in transplanted ears provided insights into how the normal ganglia locations could be acquired. The periotic mesenchyme could also be providing the cues for the correct relocation for the two ganglia in the transplants. We tested the role of periotic mesenchyme in this context by including the anterior mesenchyme in our transplants but these experiments were unsuccessful due to the lack of integrity of the mesenchymal tissues and poor reproducibility in taking a defined region of mesenchyme between experiments (n=9).

In short, the most likely explanation of our results is that the vestibular ganglion is mostly specified at the time of transplantation but not the auditory ganglion. This notion is consistent with previous reports of vestibular neurons developing ahead of the auditory neurons normally (Bell et al., 2008; Koundakjian et al., 2007; Ruben, 1967). Alternatively, it remains a possibility that the auditory neuronal fates are specified at the same time as the vestibular neurons but the poor development of a basilar papilla within the cochlear duct in most transplants might have affected the final position and projections of the auditory ganglion.

Relationships between neuronal and sensory primordia in the inner ear

Neuronal and sensory precursors sharing a common domain (NSC) in the developing inner ear is a phenomenon found among zebrafish, chicken and mice (Bell et al., 2008; Raft et al., 2004; Sapede et al., 2012). However, the position of the NSC within the developing inner ear is not the same among the three species. In the zebrafish, the NSC is located in the posterior medial region, which gives rise to both neurons and hair cells in the posterior macula/saccular macula (Sapede et al., 2012). In mice, the neurogenic region is located in the antero-ventral region of the otocyst and it shares a common lineage with the two maculae (Fekete and Wu, 2002; Raft et al., 2007; Satoh and Fekete, 2005). In chicken, a shared lineage among neurons in the ganglia and cells in the utricle has been demonstrated using replication-incompetent retrovirus (Satoh and Fekete, 2005). However, the NSC in the chicken could be larger than in the mouse since fate mapping of the otic placode links precursors of each sensory organ to its corresponding innervating neurons (Bell et al., 2008). Nevertheless, taken together these results, one or both maculae appear to be the common sensory organ(s) sharing a lineage with the neurogenic region in all three species. Our fate mapping results largely support these findings. The two dye injections at 4 and 5 o'clock of the otic cup stage resulted in a M/L relationship respective to each other by the otocyst stage. Five o'clock injections (marking the lateral NSC) primarily labeled the vestibular neurons and lateral sensory organs such as the utricle and lateral crista. In contrast, medial labeling of NSC (4 o'clock injections) resulted in increased labeling in the medially located structures: the auditory ganglion and SM, but not the lateral crista. These labeling patterns are consistent with our hypothesis that the neurogenic region and presumptive maculae are spatially related, in that the vestibular neurogenic region is linked to the UM and lateral crista, and the auditory neurogenic region is linked to the SM. Nevertheless, given that the NSC is a loosely defined region in the developing inner ear, it would not be surprising that if other sensory organs are related to the neurogenic region as well and these relationships may vary among different species.

Specification of neuronal and macular types may be coupled

Under the assumption that specification of vestibular neurons is more advanced than that of the auditory neurons at the time of transplantation, similar relationships were observed for the UM and SM. In all the transplanted ears, the UM showed appropriate expression of all the utricle-specific markers found in the controls (Crabp1 and Gata3-positive and Pax2-negative). In contrast, the identity of the SM was compromised in the transplanted ears. A small sensory patch expressed a signature gene of the SM, Pax2, but this area also weakly expressed Gata3 or Tbx3, which are not normally expressed in the control SM. The medially-located SM in the transplanted ears showed a partial saccular sensory identity compared to controls, similar to the auditory ganglion. Thus, the robustness of the UM development and the partial identity of the SM are similar to the properties displayed by the respective vestibular and auditory ganglia in the transplanted ears. This suggests that in addition to the shared spatial locality, the timing of specification of neuronal and macula fates is coupled. If further validated, this is quite remarkable since overt morphological differentiation of hair cells in the utricle and saccule is not evident until after E5 (Bartolami et al., 1991), at least three days after the time of transplantation.

If the UM was indeed specified at the time of transplantation, then why was it not located medially in the inner ear? We offer a similar explanation as the vestibular ganglion. Comparable to controls, the dye was associated with the vicinity of UM in the transplants (Figs. 8, 9, Table 2, Supplementary Table 1), suggesting that this is the original UM that was induced prior to transplantation. The mechanism(s) involved for re-positioning the two maculae to their correct M/L relationship after axial inversion remains a puzzle. This could be mediated by peri-otic mesenchyme and/or epithelium surrounding the neurogenic domain. Mixed gene expression patterns in the epithelial regions adjacent to the neurogenic domain after transplantation lend support to this notion (Fig. 6F-H).

Order of sensory organ specification

Previous studies of axial inversion of the chicken otocyst show that the anterior-posterior axis of the sensory organs is specified prior to the D/V axis (Wu et al., 1998). Results from previous M/L inversions were inconclusive because the inner ears were severely malformed after such inversions. Our transplantation study here extended the previous results in several aspects. First, the specification of the M/L axis in respect to the type of neurons and maculae occurs early at the otic cup stage. Second, despite the fact that formation of the lagenar macula, anterior and posterior cristae were found in the correct D/V relationship relative to the maculae in the transplants, the position of the UM was ventrally misplaced in two of the specimens (Fig. 8). While these numbers are small, this result raises the possibility that some aspects of the D/V axis may be determined early, which has not been appreciated previously (Wu et al., 1998). Third, the lateral crista failed to develop in most of the transplanted ears, and the SM and basilar papilla have mixed molecular identities. These results suggest that specification of these organs was not complete at the time of transplantation and the host environment is not sufficient to specify formation of a new lateral crista or basilar papilla.

Taking previous and current results together, we propose that the two maculae are the first to be specified in the NSC domain and in a lateral to medial direction: utricular macula before saccular macula. At the fringe of the NSC, the presumptive cristae, lagenar macula and basilar papilla are induced. The low frequency of the basilar papilla and lateral crista forming in the transplants suggests that they are the last to be specified among all the sensory organs.

In summary, while we have not exhaustively characterized the molecular identity of individual ganglia and sensory organs, our results indicate that the NSC was inverted in the transplants. Yet, the vestibular ganglion and UM developed reasonably well despite of the axial inversions but the auditory ganglion and SM showed abnormalities and mixed identities. The preferential association of dye-labeled cells with the vestibular ganglion and utricle similar to controls suggests that these were the original structures induced prior to transplantation. The presence of dye-labeled cells in AG and SM, which were not found in controls suggests that re-specification of fates could be occurring on some levels but it is unlikely to have a major role in VG and UM formation. Furthermore, our results suggest that specification of the vestibular fate is ahead of the auditory neuronal fate. Both of the neuronal fates are likely to be determined by the time of their neuroblast exit from the otic epithelium and these fates also pre-determine the respective UM and SM that develop subsequently. Future experiments will focus on the molecular identities that define individual ganglion and sensory organ fates.

Highlights.

• Medial displacement of cells from the lateral rim of the chicken otic cup within the neurogenic domain.

• The lateral rim of the neurogenic domain at the otic cup stage gives rise to both auditory and vestibular neurons.

• Neuronal fates of the inner ear are most likely specified at the time of neuroblast delamination, with the vestibular neuronal fate established prior to the fate of auditory neurons.

• The lateral to medial wave of vestibular to auditory neuronal fate specification is most likely coupled to the specification of utricular macula and saccular macula fates, respectively.