Early Steps towards Hearing: Placodes and Sensory Development

Sensorineural hearing loss is the most prevalent sensory deficit in humans. Most cases of hearing loss are due to the degeneration of key structures of the sensory pathway in the cochlea, such as the sensory hair cells, the primary auditory neurons, and their synaptic connection to the hair cells. Different cell-based strategies to replace damaged inner ear neurosensory tissue aiming at the restoration of regeneration or functional recovery are currently the subject of intensive research. Most of these cell-based treatment approaches require experimental in vitro models that rely on a fine understanding of the earliest morphogenetic steps that underlie the in vivo development of the inner ear since its initial induction from a common otic–epibranchial territory. This knowledge will be applied to various proposed experimental cell replacement strategies to either address the feasibility or identify novel therapeutic options for sensorineural hearing loss. In this review, we describe how ear and epibranchial placode development can be recapitulated by focusing on the cellular transformations that occur as the inner ear is converted from a thickening of the surface ectoderm next to the hindbrain known as the otic placode to an otocyst embedded in the head mesenchyme. Finally, we will highlight otic and epibranchial placode development and morphogenetic events towards progenitors of the inner ear and their neurosensory cell derivatives.


Introduction
Ear and epibranchial placodes are originating from the ectoderm to generate neurons and sensory cells, including the underlying gene regulatory networks, and signaling pathways to make sensory neurons, hair cells and taste buds. Much early development has been characterized in animal models, mostly in mice, but it is incompletey described for humans, including CHARGE [1,2] and BOR [3][4][5] syndrome defects.
Hearing starts from the placode that will be transformed into the otocyst or otic placode. The prosensory domain of the otic placode gives rise to the vestibular system as well as the cochlea and the spiral ganglion neurons (SGNs) that will connect with the cochlear hair cells [6,7]. Once the connection between the SGNs and the cochlear nuclei is established, the central auditory system develops to reach out the inferior colliculi and the auditory cortex [8,9]. In contrast to the gain of hair cells, SGN, cochlear nuclei and the auditory system, we also have a decline of sensory and neuron losses that will reduce the sound input [10,11]. Hearing deficits will likely affect close to one billion people [12] who are living with Alzheimer's disease [13][14][15]. Likewise, the associated epibranchial placode that gives rise to distinct neuron development [16,17] can lose the sensory cells of taste in aging people [18][19][20]. Human placode development and restoration has remained a serious medical condition caused by the dysfunction of placode-derived tissues. The formation of new hair cells, vestibular ganglion neurons (VGN), SGNs, epibranchial neurons, taste buds, brainstem and cortex are a common problem in humans for which we need a solution to generate new sensory receptors and neurons. 2 of 19 In the context of the inner ear, cell-based strategies to replace damaged neurosensory cells by implantation of different cell types into the inner ear represent a challenge which is in continuous progress [21,22]. Transplantation of stem cell-derived otic neuronal and epithelial progenitor cells into the cochlear nerves or into the cochleae has been demonstrated in animal models of SNHL. Successful engraftment and integration have been observed into the target sites, although with a lower survival rate, and displayed molecular features of early neurosensory differentiation [23][24][25][26].
The results from these studies are a proof-of-principle, that transplantation of partially differentiated otic progenitors may be a useful for cell-based cell therapy therapeutic strategy to treat SNHL.
Despite significant progress in inner ear transplantation, methodologies will need to be refined to generate a homogenous population of differentiated otic progenitor cells in 2D and 3D culture systems and their optimal molecular characterization before in vivo implantation.
Stem cell-derived inner ear organoids are also in continuous progress for the generation of neurosensory-like cells. They harbor fairly ordered tissues closer to those in the in vivo developing ear than can be achieved in the 2D monolayer culture system, allowing for increased recapitulation of some developmental events. In these stem cellderived 3D cultures, otic vesicles form autonomously within the aggregates, following otic-epibranchial progenitor domain formation [27,28]. In a similar manner, initial cell culture studies with isolated otic placodes have shown that once induction and invagination is complete, sensory cell and neuronal differentiation is autonomous [29]. However, in the current otic differentiation models, major limitations of the 3D organoid cultures are related to the variable reproducibility of the system [30,31], and the requirements to optimize protocols specific for a given pluripotent stem cell line [27,32] have substantially limited the studies using 3D inner ear organoids.
Therefore, a deep comprehension of the earliest requirements for pre-placodal ectoderm, epibranchial placode and sensory organ development is of utmost clarification to refine human 3D otic cell differentiation protocols that will contribute to the next generation of cell-based therapeutic approaches to restore sensorineural hearing loss.
In this review, we describe how the neural plate forms to generate the ear and epibranchial placode. We will also highlight the otic and epibranchial placode development towards progenitor cells and their neurosensory cell derivatives including hair cells and taste buds.
The expression of Sox genes results in the transformation of these precursor cells into neural plate stem cells [34]. The SoxB1 family (Sox1, Sox2, Sox3), together with Gmnn, Foxd4 and Zic2, are essential for the continued proliferation of undifferentiated neural stem cells [34]. The explosive proliferation of these neural stem cells and the coordination with convergent-extension leads to the folding of the neural plate into the neural tube (Figure 1; Refs. [40,41]). The following transition from a neural stem cell into a neuronal progenitor cell requires the expression of Sox11 [34]. For a neuronal progenitor cell to exit the cell cycle and proceed to differentiate, the initial upregulation of Gmnn, Foxd4, Zics and Sox genes must be downregulated [37,38,42,43]. Following the downregulation of the various proliferative genes, bHLH genes involved in neuronal differentiation in the CNS become expressed in dorso-ventral expression patterns within the developing neural tube to regulate fates of neurons within various subdomains [44,45]. promotes the proliferation of neuronal precursors [35] and regulates expression of the Gmnn and Zic genes [34,39]. The neural plate is induced following the inhibition of the epidermally expressed BMPs and Wnts along with the upregulation of the Gmnn, Foxd4 and Zic genes [34,37,40]. The expression of Sox genes results in the transformation of these precursor cells into neural plate stem cells [34]. The SoxB1 family (Sox1, Sox2, Sox3), together with Gmnn, Foxd4 and Zic2, are essential for the continued proliferation of undifferentiated neural stem cells [34]. The explosive proliferation of these neural stem cells and the coordination with convergent-extension leads to the folding of the neural plate into the neural tube (Figure 1; Refs. [40,41]). The following transition from a neural stem cell into a neuronal progenitor cell requires the expression of Sox11 [34]. For a neuronal progenitor cell to exit the cell cycle and proceed to differentiate, the initial upregulation of Gmnn, Foxd4, Zics and Sox genes must be downregulated [37,38,42,43]. Following the downregulation of the various proliferative genes, bHLH genes involved in neuronal differentiation in the CNS become expressed in dorso-ventral expression patterns within the developing neural tube to regulate fates of neurons within various subdomains [44,45]. After the neural plate fuses to form the neural tube (Figure 1), the newly formed dorsal roof plate begins to express BMPs and Wnts while the ventral floorplate and underlying notochord expresses Shh [50,51]. Although many genes play a role in patterning the neural tube, BMPs, Wnts and Shh play a primary role [50]. Expression of Wnts and BMPs in the dorsal neural tube is driven by the dorsal expression of Lmx1a/b [44,46,52]. Lmx1a/b expression itself is in the roof plate and is regulated by Zic1, Zic3 and Zic4 [53]. While the timing of Wnt and BMP expression during neural tube formation varies between mice and frogs [38,54], these two genes, as well as Fgfs and retinoids, are critical in defining the dorsal part of the brainstem and spinal cord [55]. Loss of Wnt or BMP signaling negatively affects dorsal progenitor cells [46].
Determination of dorso-ventral identity within the brainstem is governed by the expression of various homeodomain and bHLH transcription factors in restricted areas along the dorso-ventral axis [45,55]. The dorsal brainstem is subdivided into eight domains of neuronal progenitor populations [55]. The bHLH gene, Atoh1, is expressed in the dorsalmost domain (dA1) throughout the brainstem and spinal cord [55,56]. Atoh1-positive cells generate the rhombic lip of the hindbrain, superficial migratory neuron streams and cerebellar granule cells, that contribute to auditory, vestibular, solitary tract and proprioceptive networks [57]. As expected from the expression pattern, loss of Atoh1 eliminates most cochlear nuclei neurons [58]. Additional genes are expressed in the Olig3 domain including Neurog1/2 (dA2), Ascl1 (dA3-dB1) and Ptf1a (dA4-dB1) [45,59]. For example, the dorsalmost subdomain (dA1), that also expresses Atoh1, expresses Pax3, the next subdomain (dA2) expresses Pax3 and Pax7 and the third subdomain (dA3) expresses Pax3, Pax6 and Pax7 [55]. Additional genes are uniquely expressed in progenitor subdomains, such as Barhl1 in dA1 progenitors, Lhx in dA2, Tbx in dA3, Foxd3/Foxp2 in dA2 and dA4 and Phox2b in dB2 [55]. Genes, such as Pou4f1, are expressed in multiple subdomains (dA1-dA3, dB3) [55]. Interestingly, Atoh1, which defines the dorsal-most progenitor population (dA1), is also expressed in a more ventral subpopulation (dB2) during their maturation [45,55,60]. Phox2b is expressed in rhombomere 2-6 of the hindbrain (dB2), but not in rhombomere 7 or the spinal cord [55]. This unique hindbrain domain (dB2) also later expresses Atoh1. A second unique domain in the hindbrain (dA4) expresses Ptf1a, Foxd3 and FoxP2, among other genes [55]. Loss of Ptf1a results in the loss of dA4 and dB1 neurons in combination with the expansion of dA3 and dB3 neurons, leading to the eventual misspecification of somatosensory and viscerosensory nuclei neurons in the hindbrain [61,62]. A combination of bHLH genes Ascl1, Neurog2 and Olig3 defines dA3, in the brainstem, which extends through the spinal cord (rostral to caudal) at a location which includes the developing solitary tract [55]. In situ hybridization showed co-expression of Rnx (Tlx3) and Phox2b in the neurons of the solitary tract; in the absence of Neurog1, non-taste ganglia are lost. Projection neurons of nST are entirely absent in both Tlx3 (Rnx) and Phox2b knockouts, indicating that these neurons are dependent on these factors for their development [63].
Transition from neuronal progenitor to neuronal differentiation requires the interaction of pro-neural bHLH genes and additional bHLH genes, the Class I Hes/Hey genes [42] and the Class V ID genes [64][65][66]. Proliferation of neurosensory precursors is driven by Hes, ID, Sox2 and Myc genes, and the transition to differentiated cells is controlled by the balance of Notch signaling molecules and the pro-neural bHLH genes. Oscillation of a pro-neural bHLH gene and a repressive one, such as Ascl1 and Hes1, through cross regulation of each other controls the timing of neurogenesis [42]. This oscillation maintains proliferation of neuronal progenitors, whereas the eventual loss of Hes1 expression and subsequent sustained expression of a pro-neural bHLH gene leads to neuronal differentiation [42]. The expression of downstream bHLH genes, such as Neurod1, adds to these complex interactions. For instance, Atoh1 expression extends along the roof plate to the cerebellum, parallel to the slightly more ventral expression of Neurog1/2, and the loss of dorsal neurons in Atoh1 null mice [57,58] results in a reduced cerebellum and auditory nuclei [52]. Neurod1 negatively regulates Atoh1 expression during cerebellum, and gut proliferation and manipulating Neurod1 expression that may help to counteract Medulloblastoma [67][68][69][70]. Atoh1 shows a much higher level of expression in the auditory nuclei and counteracts with Neurod1, indicating a differential regulation of expression in auditory nuclei [71].
In summary, an overview is provided of the neural placode that develops to generate the pre-placodal induction ( Figure 1) and develops the hindbrain region that defines the different longitudinal expression of bHLH and other gene expression.

Patterning of the Pre-Placode Epithelium and Restriction of Pan-Placodal Development
Once the neural plate is fused and generates the roof plate, neural crest and placodes develop from immediately adjacent ectodermal tissue [33,72]. Neural crest and placodal neurons give rise to the formation of all or part of the various sensory systems. The work of Northcutt and Gans provided a novel perspective of the organization of the neural crest and placodes [73]. Development of neural crest and placodes is highly conserved across vertebrate species [33,74].
Inner ear neurogenesis requires the induction of Eya1 by Sox2, which in turn downregulates Sox2 expression by Neurog1 [108] that provides a negative feedback. Neurog1 [109,110] upregulates Neurod1 [8,111,112] and several other bHLH genes [113,114]. Pou4f1 is also involved in neuronal development and in addition, for the proper pathfinding of inner ear afferent neurons [115,116]. Additional genes, such as Npr1, Prickle1, Fzd3/6, Ephrin's and Vangl2, have been implicated in inner ear afferent central and/or peripheral pathfinding [117][118][119][120][121][122]. Neural crest induction occurs at the lateral edge of the neural plate [123]. Following closure of the neural tube, neural crest cells migrate throughout the embryo, contributing to a wide range of tissues including neurons, craniofacial skeleton, smooth muscles and melanocytes [72,124]. BMP, Wnt, Fgf and Notch signaling are all essential for the formation of neural crest cells [124]. Upon induction of the neural crest, many crest-specific genes are expressed, including Sox5, Sox8, Sox9 and Sox10 [72,123]. While there are specific species differences in the onset and sequence of expression, these Sox genes are important for the specification, migration and differentiation of neural crest cells [123,125]. Additional genes that are upregulated include Snail, Slug, Pax3/7, Hairy2, Msx1/2, Dlx5 and Gbx2 [72,124].
In summary, the earliest induction of distinct derivatives for the formation is presented of pre-placodal to pan-placodal development.

Development of Distinct Placodal Derivatives of the Otic Placode and the Epibranchial Placode
A unique feature is forming by sinking in the placode to generate an otic cup (Figure 2; about E9 in mice, d24 in humans) that will pinch off to form an otic cyst (about E9.5 in mice, d26 in humans). At E10 in mice (d28 in humans), the otocyst is separate from the overlaying ectoderm. The next step is to develop an endolymphatic duct that elongates. Caveating of the otocyst will develop the three semicircular canals, the utricle and saccule and elongates to become the cochlear duct (Figure 3, Ref. [105]). In contrast, only neurons are generated from epibranchial placodes [16,126]. Specifically, the facial (VII), glossopharyngeal (IX) and vagal (X) neurons develop from placodal and neural crest cells [127]. These neurons project centrally to specific brainstem nuclei (solitary tract, VII, IX, X; Figure 3) and peripherally to sensory cells (taste buds). mice, d26 in humans). At E10 in mice (d28 in humans), the otocyst is separate from the overlaying ectoderm. The next step is to develop an endolymphatic duct that elongates. Caveating of the otocyst will develop the three semicircular canals, the utricle and saccule and elongates to become the cochlear duct (Figure 3, Ref. [105]). In contrast, only neurons are generated from epibranchial placodes [16,126]. Specifically, the facial (VII), glossopharyngeal (IX) and vagal (X) neurons develop from placodal and neural crest cells [127]. These neurons project centrally to specific brainstem nuclei (solitary tract, VII, IX, X; Figure 3) and peripherally to sensory cells (taste buds). Induction of the epibranchial placodes depend critically on the expression of Eya1/2/Six1/4. These genes are upstream of Pax2, which help define the caudal (Pax2/8, VII, IX, X) placodes [75,126]. Additional genes are important in the development of the placodes contributing to the different cranial ganglia. For instance, Neurog1 is necessary for trigeminal sensory ganglia [129], whereas Neurog2 is necessary for sensory ganglia that develop from epibranchial placodes (VII, IX, X) [130]. Downstream of Neurog1/2 are Neurod1, Isl1 and Pou4f1 genes, which promote the differentiation of placodally derived sensory neurons [111,115,[129][130][131]. In addition, the epibranchial placodally derived neurons (VII, IX, X) depend upon Phox2b expression [17], which is downstream of Eya1/Six1 [82]. Additional genes such as Notch, Hes, Rbpj, Fgf8, PDGF and Wnts are important for proliferation and differentiation of neurons [132].
In summary, we provide an overview of the placodal development that will incite future interactions to make different sensory neurons of the ear and epibranchial neurons first followed by the hair cells and taste buds that are forming later.

Hair Cells of the Inner Ear Have a Shared Developmental Program of Neurons
Vestibular hair cells generate at least two types, type I and type II (Figure 4 [76,133,134]). A unique development is defined for the mammalian cochlear hair cells, the inner hair cells (IHCs) and outer hair cells (OHCs). Upstream is the expression of Eya1 and Brg1 that are needed for the initiation of hair cell progenitors. Downstream is the expression of Sox2 to interact with specific bHLH genes to initiate cochlear hair cells [108]. The bHLH gene Atoh1 (Figure 4) is needed for vestibular and cochlear hair cell formation beyond undifferentiated progenitor hair cells [6,135]. Downstream are Pou4f3, Gfi1 and Barhl1 gene expression that are needed for hair cell maintenance [136].
An interaction is existing between the bHLH gene Neurog1 and Foxg1 that causes a reduced expression of Atoh1: The absence of Neurog1 or Foxg1 results into a short and wider cochlear set of hair cells that leads to more rows of OHCs (instead of three rows of OHCs; [84]). A misexpression of Neurog1 instead of Atoh1 results in IHCs and OHCs that are not functional and are converted as inner pillar cells into gaps between near normal IHCs [137]. An interaction also shows a shorter cochlea and converts OHCs into IHCs after deletion of Neurod1, a bHLH gene. Proliferation of more hair cells depend on n-Myc for normal differentiation [105]. Most recently, an effect of follistatin (Fst) is found in the apex [138]. Combined, the cochlear apex and base are differentially affected by Lmx1a, n-Myc, Fst, Neurog1 and Foxg1, among others.
Downstream of OHCs depend on Insm1 and Ikzf2 that are needed for OHC development; OHC development interacts with Neurod1 that also regulates OHCs instead converted to IHC [139]. In contrast, loss of Tbx2 converts IHC into OHC-like hair cells [140]. Furthermore, IHC depends on Fgf8 and Srrm3/4, and cochlea will lose nearly all IHCs in the absence of these two selectively expressed genes (Figure 4; Ref. [141]).  that have a mixed distribution of type I and type II HCs. In contrast in the cochlea, we have a single row of IHCs and three rows of OHCs. Tbx2, Srrm3/4 and Fgf's are needed for differentiation and viability of IHC. Insm1, Ikzf2 and Fgf20 are needed to differentiate OHCs or requires for forming three rows of OHCs. A very different sequence of genes is needed in taste buds. Starting is the upregulation Krt8/14 that is prior to Shh. Downstream is Sox2 that is required for taste bud differentiation. Overlapping are other genes some of which seem to differentiate into distinct taste sensory input. Compile with permission from Refs. [4,8,16,52,84,110,[140][141][142][143][144][145].
Several genes are needed for the cochlear development which results in the absence of all cochlear hair cells. In the absence of Gata3, Lmx1a/b and Pax2 (Figure 4), the cochlea do not develop any hair cell that likely interacts with Atoh1 for hair cell formation [6,52,85,93]. In addition, we have a delayed loss of all hair cells that initially develop after a loss of Bdnf/Ntf3 double deletions [146]. Moreover, Bdnf deletion will result in the loss of apical OHCs, while the loss of Ntf3 causes the loss of OHCs in the basal cochlear turn. Likewise, it is dependent and will degenerate in Cdc42 for IHCs first, followed by OHCs in a base to apex progression [147][148][149]. In addition, MANF deletion results in the loss of OHCs in the base of the cochlea [150].
There is a delay between SGNs and cochlear HCs: SGNs form first in the base (E10.5) to apex (E12.5). In contrast, cochlear HCs start to from in the apex around E12.5 and progresses to the base on E14.5 [110]. Atoh1 is needed for all HCs but has a different progression starting in the base at E14.5 and progresses to the apex around E18.5 [135]. Innervation is reaching IHCs starting at E15 that expand to reach OHC innervation by Type II fibers at E18 (Figure 4; Ref. [151]).
In summary, at least two types of cochlear HCs develop that have a unique inputoutput connection [7,152] that are likely unrelated to the types I and II of vestibular hair cells [6,134].

A Unique Set of Genes Is Needed for Taste Bud Development
Like the taste ganglia, taste bud development is orchestrated by a specific sequence of gene expression and trophic interactions with nerve fibers. In the tongue, taste buds are in the papillae, which develop from placodes that arise prior to innervation and taste bud formation [153,154]. The initial signals that orchestrate taste bud development and establish their patterns on the tongue arise from the tongue epithelium [154,155]. Interestingly, signal patterning of the location of taste buds differs depending on the tongue region (i.e., fungiform papillae vs. circumvallate papillae), likely because these tissues arise from different branchial arches. Before any other gene is expressed, the earliest expression is Krt8 followed by Krt14 [143]. Specifically, fungiform papillae ( Figure 4) are patterned during development by sonic hedgehog (Shh) and Wnt signaling pathways [156][157][158][159], whereas the circumvallate papilla is regulated by fibroblast growth factor 10 (FGF10) and its receptors Spry1-2 [88,160]. The cells within the developing taste epithelial placode that express Shh differentiate into taste buds during development [161]. Prior to differentiation, these Shh+ placodal cells become innervated, and this innervation is required to maintain Sox2 expression [162]. Both Sox2 expression and innervation are required for continued taste bud development [145,163]. A differentiation requires Lgr4, Six1, Hes6 and Foxa2 expression [143,164], among others. The loss of taste bud innervation following the knockout of either the neurotrophin, BDNF, or its receptor, TrkB, results in a loss of taste buds over time [165][166][167]. The factors that are produced by the neurons to support continued taste bud development are unclear; however, Shh and R-spondin are likely possibilities [168][169][170]. Although the taste system matures, some developmental processes continue into adulthood, including taste bud cell differentiation. One unique feature of taste buds is that cells are replaced every 8-20 days, depending on the specific taste bud types [171][172][173]. The stem cells that give rise to adult taste buds express Lgr6 in the fungiform papillae, but express Lgr5 within the circumvallate papillae [174,175]. Lgr5/6 cells give rise to all taste bud cell types [175]. The absence of Neurog2 leads to reduced Sox2 expression [145,176] and disrupts taste bud formation beyond a limited differentiation of small, single taste buds [163]. Taste bud cells that transduce bitter express Eya1 suggesting that it may be involved in the differentiation of this cell type [144]. Similarly, the differentiation of type II cells is dependent on the transcription factor, Pou2f3, while type III cells depend on Ascl1 for their differentiation [16,144,[177][178][179].
In summary, the development of taste buds is controlled by a series of gene expression events. Once the axons of peripheral neurons reach the taste epithelium, these two cell types become interdependent. Thus, the formation of the peripheral taste system represents the interaction between early and late genes that regulate cell fate and the trophic interactions that occur between taste buds and nerve fibers.

Development of Neuronal Genes Are Needed for Ears and Epibranchial Placodes
A set of unique genes is characterized so that it requires the vestibular, spiral and epibranchial ganglion neurons [6,16].
Vestibular ganglion neurons (VGNs) and spiral ganglion neurons (SGNs) are independently derived from a common origin, Eya1, Sox2 and Neurog1 ( Figure 5). Eya1 and Sox2 define SGN precursor populations, whereas Neurog1 is needed to initiate the proliferation and differentiation of SGNs [4]. Without Eya1, Sox2 or Neurog1, no sensory neurons would develop [4,108]. Downstream there is a segregation of genes that are Tlx3-positive for vestibular neurons but are negative for SGN [142]. Furthermore, Sall3 is in part positive for certain VGNs and negative for other VGNs [180].
Several selective genes can selectively induce losses of distinct SGNs. Conditional deletion of Gata3, Lmx1a/b, Dicer, Shh or Pax2 results in the complete loss of SGNs ( Figure 5), while many VGNs develop despite the loss of SGNs [52,85,93,[183][184][185]. Furthermore, the cochlea is reduced into a sac without forming a cochlear duct in several mutants [52,85,93]. 5), while many VGNs develop despite the loss of SGNs [52,85,93,[183][184][185]. Furthermore, the cochlea is reduced into a sac without forming a cochlear duct in several mutants [52,85,93].  A unique set of Tbx1, 2 and 3 deletions interact with Neurog1 to downturn VGNs and SGNs [142]. Tbx1 acts as a selector gene which controls neuronal fate in the otocyst. Ablation of Tbx2 leads to cochlear reduction, but it is unclear how many SGNs form. Likewise, absence of Tbx3 results in vestibular malformations that require the distribution of VGNs. Combined deletion of both Tbx2/3DKO results in incomplete and largely absent cochlear and vestibular structures in newborn mice for which we have no differentiation of VGNs and SGNs.
Downstream are neurotrophins that are needed for sensory neuron development and maturation [187]. The loss of all SGNs is documented for TrkB and TrkC that signal for Bdnf (TrkB) and Ntf3 (TrkC). There is an incomplete deletion in the basal cochlear turn that is dependent on Ntf3 and TrkC, while the loss of Bdnf and TrkB resulted in the reduction and loss of the apex SGNs: Bdnf depends on 95% of VGNs while only about 5% are lost in the SGN, mainly in the apex. In contrast, Ntf3 (NT3) is nearly lost of 95% of SGNs, has lost all basal turn in SGNs, but has only an additional loss of about 5% of VGNs [188].
Proliferation starts from E9-14 for VGNs and from E10.5-12.5 for SGNs. The VGNs delaminate and migrate to form a ganglion outside of the ear while the SGNs stay inside the cochlea. Developing neurons reach out first from VGNs at E10; whereas, the first SGNs fibers reach the auditory nuclei at E12.5. Deletion of certain genes [Neurod1, Isl1; Refs. [112,181]] as well as loss of Schwann cells after Sox10 or ErbB2 deletion show a migration of SGNs that mixed with the VGNs [125].
In summary, a set of genes is needed for the formation of the SGNs and depends on their development which diversifies into four SGN types.
Like we detailed for the origin of the otocyst, gustatory ganglion formation depends on the sequential expression of specific genes, resulting in epibranchial placode formation, delamination, migration and cellular differentiation [16]. Initial placode development depends on the expression of the transcription factor Foxi3, which is essential for the development of the ear and epibranchial placodes ( Figure 5). In the absence of Foxi3 expression, the ectoderm fails to thicken, and all placodes fail to form [48]. Six1/2/4 and Eya1/2 are essential for the specific formation of epibranchial placodes [33,72,126] downstream of Foxi3. The Sox2, which is ubiquitously expressed in the epibranchial placodes, is required for neuron development [108]. Epibranchial placodes are patterned into rostral and caudal domains by Notch signaling [189], which is regulated by Eya1 [190]. From the Sox2+ precursor pool, placode cells divide into a non-neural population and neuroblasts defined by the transcription factor Neurog2 in mammals [190]. Neurog2 is critical for neuronal development [163,191], and Neurog1 plays a similar role in chickens [128,192]. The activation of Notch signaling discourages neuronal fate [190] in favor of a non-neuronal cell fate. Downstream of Neurod1, Isl1, Pou4f1 and Phox2b interact to regulate neuronal migration and differentiation [128,163,193]. Although all neuroblasts express the pro-neural transcription factor Isl1, Phox2b is required specifically for a visceral sensory neuron (taste) fate [194]. Foxg1, Phox2a and Phox2b are expressed in the placodes during delamination [128], followed by the set of genes Coe1, Drg11 and Dcx, which are only activated after the migrating cells have left the placode [128,192]. Once the final position is reached (facial, glossopharyngeal, vagus ganglion), the neurons grow a single process that branches to innervate the targeted taste bud cells and sends the proximal innervation to reach distinct areas of the hindbrain to reach out the solitary tract [16,127,195].
In summary, we provide an overview of otic and epibranchial neuron formation from the earliest expression of precursor cells to the differentiation of vestibular, spiral and epibranchial neurons.

Concluding Remarks
A detailed characterization of the human embryonic and fetal inner ear development is important. Specifically, induction and the early morphogenesis of the inner ear require signaling cues deployed in both a spatially and temporally restricted pattern. Mimicking these signals under a chemically defined 3D growth system has been shown in many exciting studies to differentiate stem cells into "inner ear organoids" containing sensory epithelia and neurons. However, it should not be concluded that these neurons that differentiated in inner ear organoids only belong to the otic placode lineage. For instance, many of these neurons might resemble cells in the hindbrain or epibranchial placode lineage. Thus, a careful characterization of stem cell-derived neurons in organoid and 3D cell culture systems, principally regarding their specific lineage identities, is still unsatisfactory.
The last several years, developments, although promising for the derivation of human inner ear neurosensory cells from stem cells, emphasize the importance to further improve and build on our deep understanding of inner ear induction and early organogenesis. This knowledge can be effectively exploited to faithfully recapitulate in vitro the crucial developmental steps leading to otic progenitors and their neurosensory derivatives, which can be used to develop therapies for the human inner ear.
Author Contributions: A.Z. initially drafted the paper; B.F. helped with layout of the paper and drafting some of the paper; A.Z. and B.F. finalized the paper editing. All authors have read and agreed to the published version of the manuscript.
Funding: This research has been financially supported by EU-FP7 under the health topic, grant number 603029 and by la Fondation pour l'Audition (Paris) to A.Z. and NIA R01 AG060504 to B.F.

Institutional Review Board Statement:
Ethical review and approval were waived for this study due to providing a review, not primary data.

Informed Consent Statement:
Written informed consent has been obtained from the patient(s) to publish this paper, if applicable.