Fate map of the chicken otic placode

Modeling any developing organ of an organism is an attempt to rationally systematize a complex network of molecular and cellular interactions. It has been shown that self-enhancement and long-range inhibition mechanisms can control position-dependent gene activations and repressions in pattern-forming systems (Meinhardt, 2008), managing cell segregation mechanisms devoted to the creation of lineage compartments in growing tissues (Dahmann and Basler, 1999). The vertebrate inner ear is one of the most intricate examples of organogenesis, and is an excellent model for scientific inquiry into the molecular kit implemented in complex sensory organogenesis. This elaborate sensory organ arises from the otic placode, a thickened, two-dimensional portion of ectoderm located on each side of the developing hindbrain, dorsal to the pharyngeal branchial pouches. The otic placode invaginates and pinches off from surface ectoderm to form a simple saccular otocyst which, after diverse inductive and morphogenetic events, then gives rise to the three-dimensional inner ear complex, comprising an interconnected epithelial labyrinth with vestibular and auditory sensory specializations. The vestibular part is composed of utricle, saccule, three semicircular canals (with the respective macular and ampullary sensory organs) plus the non-sensory endolymphatic system, while the auditory part forms the cochlea. The corresponding sensory ganglion cells are produced locally as well.

The morphogenesis of this organ is governed by signals diffusing from nearby tissues, or spreading planarly within the otic epithelium itself. Despite various breakthroughs in the molecular genetic mechanisms involved in the specification and shaping of the membranous labyrinth, many questions remain about its early patterning, which might start as early as the placodal stage. Thus, it would be very useful to construct precise fate maps that relate the otic placode to the derived regionalized vesicle.

A few fate mapping studies have been carried out on the developing vertebrate inner ear, mostly using the chick embryo as a model. These works used either injections of the fluorescent lipophilic vital dyes DiI/DiO or retrovirus-mediated lineage analysis (Abelló et al., 2007; Bell et al., 2008; Brigande et al., 2000a; Kil and Collazo, 2001; Kozlowski et al., 1997; Lang and Fekete, 2001; Li et al., 1978; Pieper et al., 2011; Satoh and Fekete, 2005; Stevenson et al., 2012; Streit, 2002; Xu et al., 2008). Some of the studies were performed before the otic and trigeminal placodes were constituted as two separate identities (Kozlowski et al., 1997; Pieper et al., 2011; Streit, 2002), and a single study of the frog inner ear was carried out exactly at the otic placode stage (Kil and Collazo, 2001). Other studies addressed otic cup stages (Abelló et al., 2007; Bell et al., 2008; Brigande et al., 2000a) or otocyst stages (Kil and Collazo, 2001; Li et al., 1978), i.e. examined otic primordia that had already initiated important morphogenetic changes. The resulting differences in methods and data sets have generated some unresolved questions, such as the possible existence of compartments or of cell lineage dispersion within the otic epithelium (Bell et al., 2008; Kil and Collazo, 2002).

Here, we present a detailed fate mapping study of the avian otic placode at the 10-somite stage, using the well-known chick-quail grafting model (Alvarado-Mallart and Sotelo, 1984; Le Douarin, 1969), with an experimental operating grid of 12 sectors [three square dorsoventral (DV) areas each at four anteroposterior (AP) levels, using neighboring hindbrain rhombomeres as positional landmarks; Vaage, 1969]. The fates were evaluated at stages HH29-HH34 using various specific markers. The relative topological position of the presumptive territory for all sensory and non-sensory components of the membranous labyrinth could thus be identified. We found that the avian otic placode is constituted by three band-like AP domains arranged dorsoventrally. The dorsalmost domain gives rise to the endolymphatic system, whereas the ventralmost band develops into the ampullary cristae and associated semicircular canals. The intermediate domain generates the maculae and the macular-derived basilar papilla, together with their attached non-sensory epithelia. This detailed fate map should be of help in the search for a model that considers dose-dependent and multi-step actions of diverse inductive signals released from nearby tissues, to achieve the proper specification of each constituent part of the avian inner ear in a mosaic-like fashion.

To obtain a detailed fate map of the otic placode at the 10-somite stage, we first analyzed the possible contribution of a broad portion of the cephalic ectoderm to the otic placode (Fig. 1A). Preliminary homotopic quail-chick grafts of large pieces of embryonic skin next to particular rhombomeres (rh) showed that only the ectoderm facing rh4 to prospective rh7 contributes to the inner ear (Fig. 1A). To analyze with greater resolution the fate of the portions of ectoderm associated topographically with each of these rhombomeres, three types of grafting experiments were performed at each of these four AP sites (Fig. 1A,B). In all cases, the dorsal border of the grafted territory coincided with the embryonic midline (Fig. 1A). Type 1 grafts were limited to the dorsal third of the ectodermal strip at each AP level (dT; Fig. 1A,B). Type 2 grafts included the dorsal two-thirds of the respective tissue strips at each level, therefore including the type 1 graft domain (Fig. 1A,B). The consequently expanded extent of type 2 grafted territories was analyzed at the same stages, noting differences with type 1 grafts. The supplementary quail-derived area was assigned to the fate of the middle third of the grafted ectoderm (mT; Fig. 1B). Type 3 grafts comprised the entire cephalic ectoderm next to a specific rhombomere, thus encompassing the territories already taken in grafts of types 1 and 2, plus a further ventral domain. Comparing the type 2 and type 3 grafts, the supplementary QCPN-positive territory was associated with the fate of the ventral third of the relevant transverse strip (vT; Fig. 1B). The fate map we obtained is thus based on otic-derived structure determined at stages HH29-HH34, projected backwards upon a 4×3 grid of square areas, adjacent to rh4-rh7 at stage HH10.

First, the dorsal third of the ectoderm facing rh4 was transplanted (rh4-ect type 1; n=5; Fig. 2A). In an E6 (HH29) chimeric inner ear, the QCPN-positive quail cells were mostly localized in the anterolateral portion of the endolymphatic sac and duct, forming a DV band (Fig. 2B-D). Grafted cells were also observed in a small area immediately rostral to the endolymphatic duct insertion, in the dorsal aspect of the vestibule, far from the anterior and posterior cristae (Fig. 2E). Ventrally, the graft increased in size, extending rostrally towards the Fgf10-positive anterior crista (Fig. 2F), but without contacting it (Fig. 2F,G). The graft also extended ventrally toward the macula utriculi and macula sacculi. Some chimeric embryos (n=3/7) showed a small number of QCPN-positive cells in the dorsalmost aspect of both maculae (not shown).

Next, a type 2 graft was performed, transferring the two dorsal and middle thirds of the ectoderm facing rh4 (rh4-ect type 2; n=6; Fig. 2H). Grafted cells were detected in the anterolateral portion of the endolymphatic system (not shown), as already observed in rh4-ect type 1 experiments (Fig. 2B-D). In rh4 type 2 chimeric embryos, the entire macula utriculi was always occupied by QCPN-positive cells (Fig. 2I-K). Only a small portion of the utricule wall was devoid of quail cells (Fig. 2I). The lateral border of the grafted domain was commonly very near the medial limit of the lateral crista (Fig. 2I,J). The macula sacculi showed some quail cells exclusively in its dorsalmost part (Fig. 2I,J), but not in the rest of this macula (Fig. 2K). The epithelial area lying between the macula utriculi and macula sacculi was devoid of grafted cells (Fig. 2I,J).

Finally, the fate of the entire ectoderm opposite rh4 was determined (rh4-ect type 3; n=10; Fig. 2L), corroborating the results obtained from the previous type 2 experiments (Fig. 2H-K; dT plus mT) and deducing the fate of the remaining ventralmost portion of the grafted ectoderm (vT). In type 3 chimeric embryos, the entire wall of the utricule was now included in the transplanted area. Remarkably, the medial half of the anterior crista, but not its lateral half, was constituted by quail cells (Fig. 2M,N). The medial wall of the anterior portion of the developing vertical pouch (a-vp), a non-sensory structure that develops into the anterior semicircular canal, was also immunopositive for QCPN (Fig. 2M). In some cases, quail cells were also detected in the medial portion of the lateral crista (n=3/10; not shown; see Fig. 2O,P). The wall of the utricule was clearly constituted by grafted cells and the area between the utricular and saccular maculae was constituted by quail cells (Fig. 2O,P). Fig. 2Q,R summarizes all these results in a stage E6 inner ear.

When the dorsal third of this ectoderm was transplanted (rh5-ect type 1; n=5; Fig. 3A), the quail-derived clones were observed in the anteromedial portion of the endolymphatic system, forming a DV band (Fig. 3B,C), similar and contiguous to the endolymphatic territory transplanted in rh4-ect type 1 experiments (Fig. 2B-D). In the vestibule, grafted cells were detected in a small area immediately medial to the insertion of the endolymphatic duct (not shown). Moreover, the macula sacculi contained quail cells in a small medial area (Fig. 3D,E). The QCPN-positive area invaded slightly into the proximal part of the basilar papilla in the cochlear duct, exclusively in its medial part (Fig. 3F,G).

The middle-third portion of the ectoderm opposite rh5 was then grafted in addition to the dorsal third (rh5-ect type 2; n=6; Fig. 3H). The Fgf10-positive macula sacculi showed QCPN-positive cells almost throughout (Fig. 3I,J), except in its dorsalmost portion (not shown; see Fig. 3I,J). A portion of the wall of the saccule was now also included in the transplanted area (Fig. 3I,J). The QCPN-labeled territory extended more laterally in the proximal cochlear duct, with the Fgf10-positive basilar papilla uniformly showing quail cells (Fig. 3K,L). In the proximal cochlear duct, the graft extended laterally far from the basilar papilla (Fig. 3K,L). The rest of the cochlear duct showed no grafted cells (Fig. 3S,T).

The entire cephalic ectoderm facing rh5 was then transplanted (rh5-ect type 3; Fig. 3M; n=11). In all cases, the type 3 graft extended rostrally towards the well-defined, Fgf10-positive anterior crista (Fig. 3N,O), therefore including the lateral wall of the anterior vertical pouch (Fig. 3N). Interestingly, the lateral half, but not the medial half (labeled in rh4 grafts), of the anterior crista was constituted by grafted rh5 cells (Fig. 3N,O). It is also significant that the entire horizontal pouch, the future lateral semicircular canal, was constituted by grafted cells (Fig. 3N). The Fgf10-positive lateral crista was now also included in the grafted epithelium (Fig. 3P). The macula sacculi was QCPN positive (complementing significantly the rh4-labeled medial portion), but not the macula utriculi (Fig. 3P,Q). In the proximal cochlear duct, the whole anterior half of its wall was QCPN immunopositive (Fig. 3R). Fig. 3S,T summarizes the respective distribution of cells derived from the three ectodermal sites examined (dT, mT and vT) opposite rh5.

To determine the fate of the ectoderm facing prorhombomere C (RhC), which will divide into rh6 and rh7, we first focused on its rostral half, which would roughly correspond to future rh6 (Fig. 4). When the dorsal third of this ectoderm was transplanted (rh6-ect type 1; n=6; Fig. 4A), the quail-derived grafts formed a DV band located in the caudal aspect of the chimeric inner ear. QCPN-positive cells were detected in the caudal portion of the endolymphatic system (Fig. 4B). In the vestibule, quail cells were present in an epithelial area situated caudal to the insertion of the endolymphatic duct (Fig. 4C). In the cochlear duct, a narrow band formed by grafted cells contributed slightly to the caudalmost portion of the Fgf10-positive basilar papilla (Fig. 4D,E). A few scattered quail cells appeared dispersed a short distance outside of the grafted area (arrow in Fig. 4D, arrowhead in its inset). The macula lagena was located immediately caudal to the QCPN-immunostained area (Fig. 4F,G).

When the dorsal and middle thirds of the ectoderm opposite rostral RhC were grafted (rh6-ect type 2; n=10; Fig. 4H), a small portion of the posterior vertical pouch (p-vp) was labeled in addition to the caudomedial portion of the vestibule (Fig. 4I); the p-vp is directly implicated in the formation of the posterior semicircular canal. By contrast, the horizontal pouch was completely devoid of QCPN-positive cells (Fig. 4I). The ventral half of the macula neglecta was constituted by quail cells (Fig. 4K,L), whereas its dorsal half was devoid of QCPN-positive cells (Fig. 4Q). The rest of the vestibular sensory epithelia, including the posterior crista (Fig. 4I,J), did not contain quail cells in these experiments (Fig. 4I,K,L). In the distal cochlear duct, the quail transplant occupied a large portion of its wall (Fig. 4M,N). Thus, the distal basilar papilla and the entire macula lagena were constituted by grafted cells (Fig. 4M,N).

The entire ectoderm facing the rostral half of the RhC was then grafted (rh6-ect type 3; n=11; Fig. 4O). This kind of transplantation confirmed that the ventral third of the local ectoderm (vT) will contribute to the posterior portion of the vertical pouch (p-vp in Fig. 4P,Q). The posterior crista was completely included in the grafted territory as well (Fig. 4Q,R). The dorsal part of the macula neglecta was devoid of grafted cells (Fig. 4Q), although some quail cells were observed in its ventral portion (Fig. 4K). Fig. 4S,T summarize the distribution of cells resulting from the three ectoderm domains (dT, mT and vT) examined opposite the prospective rh6.

Finally, we considered the ectoderm grafts facing the caudal half of the RhC (prospective rh7). When the dorsal third of this ectoderm was transplanted (rh7-ect type 1; n=6; Fig. 5A), the quail grafts were found exclusively in a very small caudolateral portion of the endolymphatic apparatus (not shown; see Fig. 5B). Chimeric embryos with rh7-ect type 2 grafts (n=6; Fig. 5A) showed additional labeling in the dorsal half of the macula neglecta (Fig. 5E,F), whereas its ventral half was devoid of QCPN-positive cells (Fig. 5G,H; see also rh6-ect type 2, Fig. 4K,L). The grafted middle-third territory was also observed in the vestibule (Fig. 5C), but far from the posterior crista (Fig. 5C,D). In addition to the rh7-ect type 2 results, rh7-ect type 3 experiments showed labeling of a small portion of the caudal vertical pouch (n=9; not shown; see Fig. 5I,J). Fig. 5I,J summarize all these results in a stage E6 inner ear.

In order to better define the borders of the otic placode within our 4×3 experimental grid, DiI labeling experiments were performed at the 10-somite stage and analyzed at the otic cup stage, determining whether they fell inside or outside of the presumptive otic vesicle. DiI crystals were placed in the peripheral area of the ectoderm strips, the central part of which included the otic placode. We observed that a narrow peripheral band of the analyzed ectoderm does not contribute to chick inner ear formation (Fig. 6). These results agreed with the fate map reported by Brigande et al. (2000a).

Fig. 7 summarizes the fate of the 12 ectoderm domains that contribute to the otic placode (Fig. 7A), as read out in the inner ear at E6 (Fig. 7B), using the same color-code to map each set of experiments. Regarding the sensory patches of the developing inner ear, our preparations revealed that: (1) the macula utriculi and a part of the anterior crista, as well as a small portion of the macula sacculi and lateral crista, originated from the ectoderm facing rh4; (2) the rest of the macula sacculi and lateral crista, as well as the complementary portion of the anterior crista, in addition to the proximal part of the basilar papilla, originated from the ectoderm opposite rh5; (3) the entire macula lagena and posterior crista, as well as one half of the macula neglecta, and the distal basilar papilla, came from the ectoderm facing the rostral RhC (presumptive rh6); and (4) the remaining half of the macula neglecta originated opposite the caudal RhC (presumptive rh7). The origin of the non-sensory elements was also determined (Fig. 8). Remarkably, the endolymphatic system clearly derives from the dorsalmost part of the otic placode (Fig. 8).

To confirm the fate map obtained from the 12 grafting experiments presented above, three new kinds of graft have now been performed (supplementary material Fig. S1). (1) Similar portions (types 1-3) of two contiguous rhombomeres were grafted and the fates analyzed (n=25). (2) The dorsal half of the ectoderm facing a rhombomere was also grafted and the fate was analyzed (n=12). (3) To remove any possible doubts, we performed additional experiments but now cutting the cephalic ectoderm along a line perpendicular to the neural tube passing through the middle of a rhombomere. Graft sizes equivalent to two or three contiguous rhombomeres were also employed (n=17). Out of this additional round of grafting experiments new chimeric cases were analyzed (n=10; supplementary material Figs S2-S4). In all these new grafts, the location of the graft borders was absolutely coincident with the fate mapping study obtained above, confirming once again the illustrated fate map showing the location of sensory and non-sensory organs within a Cartesian coordinate positional system.

The chick otic placode is morphologically first perceptible around the 10-somite stage on both sides of the hindbrain. Previous early fate mapping studies, which involved placing the fluorescent dyes DiI and/or DiO in the chick epiblast (Streit, 2002), did not investigate the 10-somite stage (the subject of this work), and focused instead on a preceding developmental period, namely the head process and 4-somite stage (HH8). At stage HH8, the otic placode precursors are scattered and intermingled with those of other nearby embryonic tissues, in particular the trigeminal placode (see also Stevenson et al., 2012; stage HH7). Extensive cell movements bring otic precursors gradually to convergence within a defined otic placode, and are then entirely isolated, especially with respect to the trigeminal placode. This result has been confirmed in Xenopus, zebrafish and chick embryos (McCarroll et al., 2012; Stevenson et al., 2012). The trigeminal Otx2-positive and the otic Gbx2-positive domains showed reduced overlap at stage HH7 and were exclusive soon thereafter, strongly suggesting the existence of a repressor-modulated molecular mechanism for the segregation of sensory progenitors into different placodal fates (Stevenson et al., 2012; see also Hidalgo-Sánchez et al., 2000). Our studies with DiI labeling at the 10-somite stage rule out any contribution of the cephalic ectoderm facing r3 to the otic placode, and also make clear the contribution of ectoderm facing r4 to the future inner ear rudiment. Therefore, results of the present work and those of Stevenson et al. (2012) disagree with the fate map reported for the ophthalmic and maxillomandibular trigeminal placodes in the chick embryo (Xu et al., 2008), which concludes that cells labeled at the cephalic ectoderm at the level of rh3-4, even at the 15/16-somite stage, will be incorporated into the otic vesicle (Xu et al., 2008).

It is possible (as suggested by results from this work) that some reported fate maps in the chick overestimate cell mixing between the trigeminal and otic placodes at the 10-somite stage. A similar conclusion was reached recently for Xenopus embryos using both grafts and DiI/DiO labeling experiments for studies of the origin and segregation of cranial placodes (Pieper et al., 2011; see also Schlosser, 2010; Stevenson et al., 2012). The fact that the epibranchial placode precursors are also restricted along the AP axis of the Pax2-positive ectodermal domain at early stages of somitogenesis strongly supports this contention (McCarroll et al., 2012).

It is well known that the inner ear and the otic and epibranchial placodes arise from a common progenitor domain, termed the posterior placodal field (Ladher et al., 2010), which is Pax2 positive (Baker and Bronner-Fraser, 2000; McCarroll et al., 2012; Streit, 2002). FGF signaling from the underlying segmented paraxial mesoderm causes its early induction in the competent ectoderm (reviewed by Ohyama et al., 2007; Schimmang, 2007; see also Padanad et al., 2012; Yang et al., 2013). Thus, early patterning of the head mesoderm, governed by a dynamic combination of extrinsic cues (Bothe et al., 2011), plays a pivotal role in early otic-epibranchial induction, separating this primordium from other ectodermal derivatives (Baker and Bronner-Fraser, 2001). Once this large otic-epibranchial field has been specified in the cephalic ectoderm, an elaborate network of additional signals is involved in further patterning processes within it.

Regarding the developing inner ear, our data suggest a DV specification of the otic-epibranchial placodal field into stripe-like longitudinal domains with distinct fates; this pattern might depend on a DV gradient of a morphogen that diffuses from the dorsal midline of the embryo. It is known that canonical WNT signaling from the dorsal neural tube and/or dorsal ectoderm (Riccomagno et al., 2005) acts after the induction of the posterior placodal area by FGF (Bok et al., 2007; Ladher et al., 2010). This molecular mechanism is directly involved in the placodal versus epibranchial-epidermis fate decisions, conferring otic identity upon the medial (dorsal) part of the Pax2-expressing domain (Ohyama et al., 2006). Attenuation of mitogenic FGF activity and activation of the differentiation-inducing WNT signaling pathway leads to expanded otic commitment and local repression of the epibranchial fate (McCarroll et al., 2012; Ohyama et al., 2006). In addition, the BMP signaling pathways are also relevant in early regionalization of the otic-epibranchial placodal field (Abelló et al., 2010; Begbie et al., 1999; Holzschuh et al., 2005). Hence, the spatiotemporally orchestrated actions of several signaling pathways possibly determine the final cell fates of otic progenitors, at least as regards their organization into superposed longitudinal stripe-like domains with distinct fates.

According to the present data, the cells of the dorsalmost band, probably influenced by higher concentrations of dorsal morphogens, will give rise to the endolymphatic system, whereas the ventralmost band, located far from the hypothetical dorsal morphogen source, will develop into the cristae and their associated semicircular canals. Between these two bands, a third domain will give rise to most of the maculae plus the macular-derived basilar papilla, accompanied by their attached non-sensory epithelia (Fig. 8). The fact that Bmp7 and Lmx1b expression patterns and Wnt/β-catenin activity characterize longitudinal territories located in the dorsal part of the otic placode (8- to 9-somite stage) strongly supports this stripe-like DV patterning model (Abelló et al., 2010; Riccomagno et al., 2005). A fourth AP band, located ventral to the otic placode, would represent the domain from which the epibranchial placode arises (Ladher et al., 2010). This fourth band is possibly specified by the same gradient mechanism.

Other gradient effects oriented orthogonally to the DV mechanism must control the specification of AP differences within the otic placode. Various authors have likewise postulated the existence of driving forces that govern the patterning of the otic placode along its AP axis (Abelló et al., 2007; Bok et al., 2007; Léger and Brand, 2002). A relevant transverse organizing center might be established rostrally. Interestingly, chick embryos show a narrow band of Fgf8 expression that appears at the border of the Gbx2-positive otic placode with the Otx2-positive trigeminal placode (Hidalgo-Sánchez et al., 2000; for comments on Otx2/Gbx2 mutual repression, see Stevenson et al., 2012). Similarly, as is widely accepted for the isthmic organizer, the confrontation of antagonistic Otx2-expressing and Gbx2-expressing epithelial domains might induce the creation of an intercalated otic/trigeminal organizing center that expresses Fgf8 and releases FGF8 as a diffusing morphogen (Hidalgo-Sánchez et al., 2005). The gradient-governed interaction of DV and AP organizing centers would establish a Cartesian coordinate positional system for differential molecular signaling in the otic placodal field (Meinhardt, 2008).

However, we cannot rule out other possibilities that have been suggested for DV and AP patterning, in particular the relevance of Shh and retinoic acid signals (Bok et al., 2007, 2011; Brown and Epstein, 2011; Hammond and Whitfield, 2011; Riccomagno et al., 2005; Wu et al., 1998). Restricted spot-like signaling centers from the segmented hindbrain, or even from the pharyngeal endoderm and subjacent mesoderm, might also be involved directly or indirectly in otic placode specification, representing potential sources of additional differential AP signals (Abelló et al., 2010; Alsina et al., 2009; Baker et al., 2008; Begbie et al., 1999; Chen and Streit, 2013; Graham, 2008; Grocott et al., 2012; Ladher et al., 2010; Lleras-Forero and Streit, 2012; McCabe and Bronner-Fraser, 2009; Schlosser, 2006, 2010; Schneider-Maunoury and Pujades, 2007; Whitfield and Hammond, 2007).

We visualize the otic placode as a two-dimensional field organized along the AP and DV axes, which are probably defined by early gradient-distributed inductive and positional signals, later complemented by various point-like signaling effects. By the otic cup stage, additional intrinsic and extrinsic signaling pathways, secondary to the preliminary mechanisms acting at the placodal stage, will trigger and modulate the multi-step (combined) mechanism for the specification of distinct sensory and non-sensory patches within the developing otic rudiment (reviewed by Bok et al., 2007). Thus, subsequent morphogenetic phenomena – for instance the differential rate of proliferation and cell death in the posterior placodal field (Washausen et al., 2005) – would result in the creation of the three-dimensional structure of the otic vesicle around a fluid-filled cavity.

The understanding of otic morphogenetic events now allows us to employ in any morphological description either two-dimensional topological Cartesian coordinates derived from the placodal stage or polar coordinates corresponding to the cup or vesicular stages. We believe that such a conversion would reduce possible interpretational mistakes when different developmental stages are being compared, or when the topology of a signaling effect is unclear. In this sense, current understanding of the fate of the primary axes of the otic rudiment from the otic cup stage onwards should be revised, particularly in the case of the DV axis.

This fate mapping study has systemized for the first time otic differential fates in a flat Cartesian positional model of intercrossing longitudinal epithelial stripes and transverse paraneuromeric domains. Distinct otic epithelial compartments, presumptively created by position-dependent activation or repression of specific genes already at placodal stages, could define subsequent borders of lineage restriction, gene expression and fate within the developing otic anlage. These regional DV and AP molecular identities, plus theoretically predicted antagonistic local activation and long-range inhibition phenomena (Meinhardt, 2008), could culminate in the species-specific distribution of sensory epithelia in the membranous labyrinth (Fekete, 1996; Fekete and Wu, 2002). Note that transient developmental boundaries produced between initial compartments may generate new signaling mechanisms (Hidalgo-Sánchez et al., 2005; Meinhardt, 2008). At the otic cup stage, or even later, new inductive or inhibitory signals generated in the otic epithelium itself (such as sources of FGFs, BMPs, WNTs and retinoic acid) are apparently decisive for the final modeling of the inner ear in amniotes (Romand et al., 2006; Sánchez-Guardado et al., 2009,, 2013; Sienknecht and Fekete, 2009; Whitfield and Hammond, 2007). Our results are therefore consistent with an open-ended compartmental model (Fekete, 1996; Brigande et al., 2000b; Fekete and Wu, 2002).

In summary, this fate mapping study promises to aid in the quest to understand how diverse molecular agents control asymmetric expression patterns of various regulatory genes, with important morphogenetic consequences in differential growth, differentiation and histogenesis. This experimental fate map of the chick otic placode also helps to anchor the plethora of existing descriptive and experimental data about otic epithelium patterning in a new developmental context. Moreover, the placodal otic fate map allows developmental biologists to design innovative functional studies to examine the molecular mechanisms involved in the specification of all the vestibular and auditory structures in the developing vertebrate inner ear.

Fertilized White Leghorn chick (Gallus gallus) and Japanese quail (Coturnix coturnix japonica) eggs were incubated in a humidified atmosphere at 38±1°C. Chimeric embryos were fixed and processed for cryostat sectioning as previously described (Sánchez-Guardado et al., 2009).

Chick and quail embryos were used to obtain chimeras by exchanging specific portions of the cephalic ectoderm adjacent to the hindbrain (Vaage, 1969) at stage HH10 (exactly 10 somites; Hamburger and Hamilton, 1992) in unilateral homotopic and isochronic transplants (Le Douarin, 1969). The constrictions between rhombomeres (or prorhombomeres) served as landmarks. A grid inserted into one ocular of the operating microscope helped to define the three DV divisions of the otic placode. In all cases, the chick embryo was the host and the quail embryo the donor. The microsurgical procedure in ovo was that described in detail by Alvarado-Mallart and Sotelo (1984). A small portion of the alar plate of the neural tube and the intervening mesenchyme was also jointly transplanted to obtain a more solid graft and ensure the integration of the grafts in the host (Fig. 1). After transplantation, the host eggs were closed with Parafilm, sealed with paraffin, and kept at 38±1°C. The resulting chimeric embryos were analyzed at 6-8 days of incubation (stages HH29-34), when major morphogenetic changes had already taken place (Bissonnette and Fekete, 1996; Sánchez-Guardado et al., 2013).

The grafts of quail tissue were perfectly integrated in the chick host in less than 3 hours. After this period, the quail graft always occupied exactly the holes made in the host chick embryo in such a way that it was not possible to identify without specific labeling the donor/host interface (Hidalgo-Sánchez et al., 1999). Therefore, since the experimental injury healed very well, any possible alteration caused by the wound is obviated.

In situ hybridization was performed on cryosections as described by Sánchez-Guardado et al. (2011). The chick Fgf10 subclone used in this study is described elsewhere (Sánchez-Guardado et al., 2013). To visualize the grafted cells, the QCPN monoclonal mouse antibody (Developmental Studies Hybridoma Bank; 1/100) was used, which specifically identifies quail cells among unstained chick tissues (for the procedure, see Hidalgo-Sánchez et al., 1999). Immunoreaction with the 3A10 antibody [Developmental Studies Hybridoma Bank (DSHB), mouse, monoclonal, #3A10, 1:40] is a useful tool to label the axons of the acoustic-vestibular ganglion neurons and to aid in the identification of presumptive sensory epithelia (for the procedure, see Sánchez-Guardado et al., 2013).

Small lipophilic dye (DiI) crystals were selectively placed in ovo at the predicted border of the otic placodal ectoderm, using the tip of a pulled glass micropipette (Bell et al., 2008). These experimental embryos were fixed at the otic cup stage (HH14-15; just prior to the closure of the otic pore) in 4% formaldehyde in PBS for 6 h, washed in PBS, and prepared for cryosectioning. The histological location of the DiI-labeled cells was then imaged using epifluorescence. Only cases in which the placement of the DiI crystal was restricted to a small group of cells were analyzed.

We are grateful to Rosa-Magda Alvarado-Mallart for inspiring us to carry out this work, which is dedicated to her memory.