TGFβ2-induced senescence during early inner ear development

Embryonic development requires the coordinated regulation of apoptosis, survival, autophagy, proliferation and differentiation programs. Senescence has recently joined the cellular processes required to master development, in addition to its well-described roles in cancer and ageing. Here, we show that senescent cells are present in a highly regulated temporal pattern in the developing vertebrate inner ear, first, surrounding the otic pore and, later, in the otocyst at the endolymphatic duct. Cellular senescence is associated with areas of increased apoptosis and reduced proliferation consistent with the induction of the process when the endolymphatic duct is being formed. Modulation of senescence disrupts otic vesicle morphology. Transforming growth factor beta (TGFβ) signaling interacts with signaling pathways elicited by insulin-like growth factor type 1 (IGF-1) to jointly coordinate cellular dynamics required for morphogenesis and differentiation. Taken together, these results show that senescence is a natural occurring process essential for early inner ear development.


Material and Methods
Chicken and mouse embryos. Chicken embryos were obtained from fertilized eggs from a local farm (Granja Santa Isabel, Cordoba, Spain) and they were incubated in a humidified atmosphere at 37.8 °C. Embryos were staged according to Hamburger and Hamilton's (HH) criteria 28 . Mouse embryos were obtained from C57BL/6J mice that were purchased from Charles River (Barcelona, Spain) and housed under standard conditions. All experiments were approved by the CSIC Bioethics Committee and carried out in full accordance with the guidelines of the European Community (2010/63/EU) and the Spanish regulations (RD 53/2013).
Embryo and tissue preparation for in situ hybridization and immunofluorescence. Whole embryos or tissues were dissected in phosphate-buffered saline (PBS) and fixed overnight in 4% (w/v) paraformaldehyde (PFA; Merck, Darmstadt, Germany) in PBS at 4 °C. Subsequently, embryos were cryoprotected overnight in 15% sucrose/PBS at 4 °C and then embedded at 37 °C in 15% sucrose/10% gelatine in PBS. Gelatine-embedded tissues were frozen in isopentane at −80 °C and sectioned (20 μm) at −25 °C in a cryostat (Cryocut 1900; Leica Microsystems, Deerfield, IL). The sections obtained were used for in situ hybridization or immunofluorescent staining as described 6,17,19 . Isolation, organotypic culture and treatment of otic vesicles. Embryos at stage HH18 (65 h of incubation) were obtained and the otic vesicles were dissected from the surrounding mesenchymal tissue with sharpened tungsten needles, they were transferred into four-well culture-plates (Nunc Roskilde, Denmark) and then incubated at 37 °C in a water-saturated atmosphere containing 5% CO 2 . The standard culture medium was M199 medium with Earle's salts (Sigma Aldrich Química, Madrid, Spain) supplemented with 2 mM glutamine (Gibco, Paisley, UK) and antibiotics [50 IU/ml penicillin (Ern, Barcelona, Spain) and 50 mg/ml streptomycin (CEPA, Madrid, Spain)]. For immunostaining and TUNEL labelling otic vesicles were fixed for 2 h in PFA at 4 °C. For SAβG staining the otic vesicles were fixed for 10 min with the fixative solution provided by the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology, Danvers, Massachusetts, USA).
Endolymphatic duct and dorsal areas were measured from light microscopy images using the FIJI package image processing based on ImageJ software (Wayne Rasband, National Institutes of Health, Bethesda, MD). Results are presented as the mean ± SEM of the percentage of the area of interest per total otic vesicle area. Values were normalized to the 0S explants.
Quantitative RT-PCR. Otic vesicles were dissected out from chicken embryos and pooled to obtain RNA at different developmental stages: closure of the otic cup (HH17, n = 45-50), formation and growth of the otic vesicle (HH18, n = 30-40), endolymphatic duct emergence (HH19, n = 25-30) and a more differentiated stage (HH23, n = 5-10). Microdissection of HH19 otic explants was performed to isolate dorsal and ventral regions. Three to four independent pools of RNA from each stage were isolated with TRIZOL (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. The integrity and concentration of the RNA was assessed with an Agilent Bioanalyzer 2100 (Agilent Technologies) and cDNA was generated by reverse transcription, RT, (High Capacity cDNA Reverse Transcription Kit; Applied Biosystems). Quantitative PCR (qPCR) of each pool was performed in triplicate using specific oligonucleotides for the indicated genes. Specific primers for chicken genes would be provided upon request. Eukaryotic 18S rRNA and HPRT1 were used as endogenous housekeeping genes. SYBR Green qPCR was performed on a 7900HT Real-Time PCR System and gene expression was estimated as 2 −ΔΔCt .
Immunohistochemistry and immunofluorescence. Antibodies used for immunofluorescence are shown in Supplementary Information Table S1. Procedures used have been described elsewhere 6,17,19 . Levels of PH3 immunostaining and otic vesicle areas were quantified using FIJI software, in compiled confocal microscopy projections as reported 17,19 . Colour channels of the signals of interest were converted into grey scale images. Subsequently both the area and the intensity of the signal were measured, and normalized to the 0S condition, which was given an arbitrary value of 100. 3-6 samples per condition were assayed from 2-6 independent experiments.
Western blotting. Otic vesicles (HH18, n = 25) were isolated and cultured for 20 h for protein extraction and analysis as reported 17,19 . Membranes were incubated overnight at 4 °C, with the primary antibodies indicated in Supplementary Information Table S1. Images of the blots were captured using an ImageQuant LAS4000 mini digital camera (GE Healthcare Bio-Sciences), and densities of the immunoreactive bands were quantified by densitometry using ImageQuant TL software (GE Healthcare Bio-Sciences).
Apoptosis. Apoptotic cell death patterns were studied by Tdt-mediated dUTP nick-end labelling (TUNEL) of fragmented DNA using the kit Dead-EndTM Fluorometric TUNEL System (Promega, Madison, WI) as described by the manufacturer in frozen sections as reported 5 , or adapted to whole organ labelling 6,17,19 . Otic vesicles were mounted in Prolong Gold/DAPI and visualized by confocal microscopy. TUNEL-positive nuclei were quantified from compiled confocal microscopy projections by FIJI software and the results are presented as the mean ± SEM of the positive cells per total area. Values were normalized to those of the 0S explants.
Senescence-associated beta-galactosidase staining. SAβG was performed by using the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology). After culture and fixation, otic vesicles were washed in PBS and incubated with the X-Gal solution at pH 6 at 37 °C protected from light. For SAβG in whole embryos at stages HH17, HH18, HH19 and HH20, the embryos were fixed for 15 min and incubated in X-Gal solution. To perform the SAβG in chicken embryo tissue sections, the embryos were obtained, fixed for 15 min, frozen and sectioned. Sections were post-fixed for 2 min and incubated with the X-Gal solution.
EdU labelling and SAβG staining of otic vesicles. Otic vesicles (HH18) were isolated and cultured for 20 h. EdU (5-ethynyl-2′-deoxyuridine, Invitrogen) was diluted in PBS-DMSO 1% and added for the last hour of the culture at a final concentration of 100 µM. Otic vesicles were fixed for 20 min and incubated with the X-Gal solution pH 6 at 37 °C. Then, otic vesicles were fixed with PFA, permeabilized with PBS-T, washed with 2% PBS-BSA and incubated with the Click-iT Reaction Cocktail (Invitrogen) with Alexa-488 or Alexa Fluor 594 (5 µM, Invitrogen) at RT in darkness for 30 min. Finally, they were washed again with 2% PBS-BSA and mounted in Vectashield with DAPI (Vector, Peterborough, UK).
SAβG staining levels and EdU labelling were quantified using the FIJI image processing software based on ImageJ. The colour channels of the signals of interest were converted into grey scale images. The blue staining area was analysed relative to the otic vesicle total area. Levels of EdU signal from compiled confocal microscopy were measured and relativized to otic vesicle total area. Data were normalized to the 0S condition, which was given an arbitrary value of 100.
In situ hybridization. In situ hybridization with digoxigenin-labelled antisense RNA probes (1 mg/ml) was performed essentially as described previously with minor modifications 17 . Three HH18, HH20 and HH25 embryos were tested for Tgfβ2 and three HH18 embryos were tested for Tgfβr2 expression in at least two independent experiments. Specificity of the signal was assessed by hybridization of a sense riboprobe 29 . Statistics. Results are shown as mean ± SEM and the statistical significance was determined with the Student's t-test. A p < 0.05 was considered significant. *p < 0.05; **p < 0.01; ***p < 0.001. www.nature.com/scientificreports www.nature.com/scientificreports/

Results
Cellular senescence during early development of the inner ear. β-galactosidase staining (SAβG) was used to study the presence of senescent cells in the developing inner ear in whole-mount HH17 to HH20 embryos ( Supplementary Information Fig. S1). HH17 embryos showed marked staining of SAβG in the otic epithelium, with the highest levels observed around the otic pore (Fig. 1A), an area of known intense apoptosis 30 . Later in development (HH19-HH20), SAβG staining was intense in the endolymphatic duct (Fig. 1B,C). Dissected otic vesicles from mouse E10 and HH19 chicken embryos showed that senescence is highly restricted to the developing endolymphatic duct in both species (Fig. 1D,E), suggesting that otic senescence is dynamic and developmentally-regulated. To assess the time-window of otic senescence induction, expression levels of p21, Btg1 and Btg2 were analysed by RT-qPCR in HH18 (white bars) and HH19 (black bars) otic vesicle explants. While p21 expression was not significantly increased, Btg1 and Btg2 showed a significant increase (2.7-fold and 2-fold respectively) in HH19 compared to HH18 (Fig. 1F).
Isolated HH18 otic vesicles were cultured for 24, 48 and 72 h and the intense SAβG staining at the endolymphatic duct showed that senescence could be studied ex vivo ( Fig. 1G-I, arrows). Endolymphatic duct SAβG-labelled senescent cells did not show EdU incorporation (Fig. 1J, arrow), thus, increased senescence was associated with reduced proliferation. Finally, TUNEL assay confirmed the presence of increased apoptosis in the endolymphatic duct (Fig. 1K, arrow). Figure 1L schematizes otocysts patterns of the above described senescent, apoptotic and proliferative cells. Taken together, these data suggest that developmental senescence contributes to sculpt the otocyst.
Otic morphological changes are associated with modulation of senescence. Otic vesicles were next incubated with navitoclax that specifically promotes apoptosis of senescent cells 31 . Navitoclax-treated otic vesicles showed a 70% reduction of SAβG staining ( Fig. 2A,E, quantification in M), a rounded morphology and a reduction in the endolymphatic duct area (0.8-fold, Fig. 2B,F, asterisk, quantification in N). EdU incorporation and immunofluorescence for mitotic cells with PH3 showed an increment of cells in the S-phase in navitoclax-treated otic vesicles (1.3-fold, Fig. 2C,D,G,H, quantification in P-Q). The elimination of senescent cells with navitoclax also increased the number of apoptotic cells ( Supplementary Information Fig. S2). Conversely, reduction of apoptosis by treatment with the pan-caspase inhibitor BOC increased SAβG staining ( Supplementary Information Fig. S2). These results suggest a link between apoptosis and senescence during endolymphatic duct formation.
To promote senescence, explants were treated with palbociclib 32 . After 20 h in culture, palbociclib treatment promoted a 2.9-fold increment of SAβG staining ( Fig. 2A,I, quantification in M) in otic vesicles showing an expanded and thinned dorsal region (Fig. 2B,J, insets, quantification in O). Induction of senescence with palbociclib was associated with the absence of proliferation markers (Fig. 2C,D,K,L, quantification in P-Q). To further document navitoclax and palbociclib actions, we examined gene expression levels of p21, Btg1 and Btg2, which were upregulated by palbociclib (1.52-, 1.57-and 1.48-fold respectively; Fig. 2R). Finally, considering that it has been reported that the highly secretory senescent cells require induction of autophagy to restore cell homeostasis 33 , we measured Lc3b (microtubule-associated protein light chain 3 beta, widely used as an autophagy biomarker) transcription levels that indeed were increased in otic vesicles incubated with palbociclib (1.38-fold; Fig. 2R).
In summary, we show that otic vesicle morphogenesis can be modulated by drugs that increase or reduce cellular senescence, which particularly impact the formation of the endolymphatic duct.
tGFβ family factors and their receptors are expressed in the developing inner ear. Senescent cells have a secretory phenotype that is essential to modify their cellular microenvironment, with TGFβ family members featuring among the secreted factors 15 . To explore the mediator that induces senescence in the endolymphatic duct, we considered the actions of TGFβ/BMP signalling with a special interest in TGFβ2, a factor that has not been explored in this context before.
The expression of Tgfβ2, Tgfβ3, Tgfβr1, Tgfβr2, Tgfβr3 and Eng was studied by RT-qPCR in inner ear extracts from HH17-HH19 and HH23 embryonic stages (Fig. 3A). All the analysed transcripts were expressed in the studied stages, but with distinct temporal patterns. Tgfβr1 and Tgfβr3 expression levels did not vary, but, in contrast, Tgfβ2, Tgfβ3 and the receptor Tgfβr2 transcripts showed a significant increase at HH23, whereas the co-receptor Eng increased its expression at stage HH19 that persisted to HH23. These data were further confirmed as the expression of TFGβ-system genes in HH23 inner ear explants was associated with areas of increased SAβG staining (data not shown).
The cellular expression pattern of the transcript for Tgfβ2 was further studied by in situ hybridization (Fig. 3B,C). Tgfβ2 mRNA presented a localized expression at stage HH18 in the endolymphatic duct precursor area (Fig. 3B first panel, arrow). Consistent with TGFβ2 signalling, we found typical nuclear immunopositivity for the phosphorylated SMAD2 effector protein (pSMAD2) in the otic epithelium of the otic vesicle and in the endolymphatic duct anlagen (Fig. 3B, third panel). Tgfβr2 was found to be strongly expressed in the dorsal part  www.nature.com/scientificreports www.nature.com/scientificreports/ of the otic epithelium, which includes the endolymphatic duct anlagen (Fig. 3B, second panel, arrow). We also investigated the presence of Tgfβ2 at later developmental stages and found strong expression in the developing endolymphatic duct at stages HH20 and HH25 (Fig. 3C, arrows).
Thus, our results show that the TGFβ signalling is active in the otic epithelium during early inner ear development, with specific expression of TGFβ2 in the endolymphatic duct area. tGFβ2 promotes senescence in the otic vesicle. Otic vesicles (HH18) were isolated and cultured with or without exogenous TGFβ2 (Fig. 4). TGFβ2 significantly increased SAβG by 2-fold (Fig. 4A,B, quantification in F) and increased Btg1 and Btg2 expression levels (1.18-fold and 1.33-fold respectively; Fig. 4G). To further confirm the role of TGFβ2 in the regulation of otic senescence, otic vesicle explants were treated with the inhibitor of TGFβ signalling SB431542 34 . TGFβ2-induced SMAD2 phosphorylation was reduced by co-treatment with www.nature.com/scientificreports www.nature.com/scientificreports/ SB431542 (Fig. 4E), and SB431542 also reduced TGFβ2-induced SAβG staining (Fig. 4C,D, quantification in F). In contrast, SAβG staining in the control condition (0S) was not modified by SB431542.
To explore if TGFβ2 modulates endolymphatic duct morphogenesis we analysed the morphology of the otocyst as well as the expression levels of Wnt2b, which is specifically expressed in the mouse endolymphatic duct 35 . We confirmed that Wnt2b is expressed in the chicken endolymphatic duct by showing its upregulation (1.47-fold) in HH19 otic vesicle explants. Concordantly, Wnt2b expression was higher (7.3-fold) in dissected dorsal regions of HH19 otic explants compared to ventral regions (Fig. 4H). In parallel, the ventral-expressed Neurog1 was downregulated in HH19 (0.52-fold) consistently with its role in early neurogenesis 3 and the expression levels were upregulated (2-fold) in the ventral region compared to the dorsal region ( Supplementary Information Fig. S3). The association between cellular senescence and endolymphatic duct development was further confirmed by the differentiation of the endolymphatic duct observed following TGFβ2 treatment (Fig. 4I,J, asterisk, quantification in K). This was consistent with the significant increase of Wnt2b found in otic vesicles treated with either TGFβ2 (3-fold) or palbociclib (1.47-fold; Fig. 4L). Therefore, we conclude that TGFβ2 plays an inductive role in endolymphatic duct morphogenesis by promoting cellular senescence.

PI3K/AKT and RAF-MEK-ERK pathways modulate senescence in the otic vesicle.
Based on previous findings from our lab 17,19 , IGF-1 was used as an activator of otic PI3K/AKT and RAF-MEK-ERK pathways. Figure 5 shows a significant decrease (70%) of SAβG staining in the presence of the RAF kinase inhibitor sorafenib (Fig. 5A,B, quantification in H), whereas an increase was observed when the activity of AKT was inhibited (7.7-fold; Fig. 5A,C, quantification in H). Treatment with AKTi or sorafenib inhibited the phosphorylation of AKT and ERK, respectively (Fig. 5G). To further understand the interplay between the extrinsic factors that regulate these pathways, cellular senescence was studied following IGF-1-stimulation ( Fig. 5D-F). SAβG staining was increased by IGF-1 (1.9-fold), which was able to partially overcome the effects of sorafenib but not those of AKTi that were even potentiated (Fig. 5E,F quantification in 5 H). In contrast, palbociclib-induced senescence showed reduced Igf1 transcript levels (0.52-fold; Fig. 5I) but increased Igf1r expression (1.37-fold). These results are in agreement with the arrested proliferative status of senescent cells (Figs 1J and 2Q,R) and the reported proliferative role of IGF-1 36 . Upregulation of IGF1R transcription is a typical compensatory cell response to reduced IGF-1 signalling 37,38 .

Discussion
Here we show that senescence plays a role in the early development of the endolymphatic duct, site of intense programmed cell death and reduced proliferation. Elimination of senescent cells and induction of senescence in the otic vesicles showed opposing morphogenetic defects. Our results show that senescence is induced by TGFβ2 in the otic vesicle, where TGFβ pathway components are expressed at these early embryonic stages. In addition, IGF-1 targeted kinases AKT and RAF modulated SAβG staining in the otic vesicle.
Developmental senescence in the otic vesicle is confirmed by specific SAβG staining, arrested proliferation and senescence related gene transcript upregulation. Btg1 and Btg2 appeared to be particularly relevant as senescence read-outs in chicken. Both genes are differentially regulated in limb samples of species that eliminate the interdigital tissue by senescence (chicken) compared to those that doesn´t (duck). Btg1 and Btg2 are both upregulated during interdigital tissue regression (7.5-8d) in chick limbs, while in the duck limbs Btg1 is upregulated but Btg2 is downregulated 14 . Our results show that Btg1 and Btg2 expression is modulated during otic development, being both upregulated at HH19 at the time of endolymphatic duct formation. Btg1 and Btg2 roles in senescence was further confirmed by their induction with palbociclib and TGFβ2.
Senescence is very intense at the edges of the pores of structures that arise from ectoderm invaginations which eventually close up, such as the otic pore and the lens vesicle pore. Apoptosis is also present at pore edges during the otic cup stage, which associates with the folding and subsequent closure and separation of the otic vesicle from the ectoderm 4 . These results suggest a connection between senescence and apoptosis in promoting the closure of the otic cup to form the otocyst. The concurrence of senescence and apoptosis during development has been recently demonstrated in the vertebrate limb, particularly in the AER and the elimination of the interdigital tissue 13,14 . Our work also shows that senescence staining and TUNEL labelling coincides in specific areas of the endolymphatic duct also showing signs of cell cycle arrest. Therefore, our data support the presence of developmental senescence concomitant with reported areas of apoptosis in the otocyst 4 .
Research on senescence during development published to date shows that underlying molecular mechanisms and function vary depending on the structure. Developmental senescence in the interdigital webs of the developing limb is followed by apoptosis and subsequently removed by macrophage-mediated phagocytosis 14 . Mesonephros elimination is due to senescent cells that are also subsequently removed by macrophage-mediated phagocytosis 12 . This cell-clearing function might be conserved in all vertebrates studied from teleost fish to mice 39 . Thus, senescence might be a conserved mechanism that facilitates the elimination of the excess of cells generated during development.
Work from us and others show developmental senescence at regions that undergo morphological changes and generate three-dimensional structures such as the dorsal region of the pharyngeal slits, the edges of the optic lens vesicle pores, the pineal gland and the otic vesicle 13 . Moreover, senescence has been reported to be a remodelling mechanism in Xenopus laevis cement gland 40 . The morphological changes that the HH18 otocyst undergoes include the protrusion of the endolymphatic duct. Elimination of senescent cells abolished endolymphatic duct morphogenesis. It would therefore be reasonable to propose here that senescence facilitates the acquisition of the three-dimensional otic vesicle shape, first by blocking proliferation and second by promoting the slow removal of cells. www.nature.com/scientificreports www.nature.com/scientificreports/ Our results also suggest a role for developmental senescence in otic dorso-ventral patterning. The phenotype of otic vesicles treated with palbociclib showed a dorsalized otocyst with increased SAβG staining in the ventral region. This result might be explained by the ectopic secretion in the ventral region of SASP component(s) that could promote dorsalization. In fact, the palbociclib-induced phenotype is related to that of the effect of BMP4 overexpression which also induces thinning and ventral expansion of the dorsal epithelium 41 .
Later in development, senescence facilitates the maturation of the endolymphatic sac by favouring the presence of pendrin-positive cells 12 , which are essential for the maintenance of the endolymph in the adult inner ear 42 . Thus, senescence during the developing inner ear plays different roles sequentially, first sculpting the endolymphatic duct in the otocyst and later defining endolymphatic sac cell populations.
TGFβ2 plays an inductive role in programmed senescence during mouse embryonic development 12 . Our data show that TGFβ2 also promotes senescence during inner ear development. Tgfβ2 is expressed in the otic epithelium with the highest levels observed in the dorsal part of the otocyst, including the endolymphatic duct anlagen. Its target pSMAD2 was also present in these areas. In our hands, TGFβ2-treatment potentiates endolymphatic duct formation through senescence. These results are particularly interesting because inhibition of BMP-ligands reduces differentiation of dorsal otic areas without affecting endolymphatic duct formation 41 . Therefore, our work proposes a specific role for TGFβ2 on the endolymphatic duct morphogenesis by means of inducing senescence. The specific expression of TGFβ2 in the endolymphatic duct at HH18 and HH20 reinforce the proposed actions for TGFβ2 in the formation of the endolymphatic duct. Further supporting this view is the fact that TGFβ2 deficient mice show morphological defects and alterations in the homeostasis of endolymph 43 .
Surprisingly, in the absence of an exogenous source of TGFβ2, blocking endogenous TGFβ signalling with SB431542 does not suffice to decrease otic senescence. This highlights the fact that the inhibitor may not be completely inhibiting its target, TGFβR1. The small reduction in the magnitude of the phenotype upon addition of both SB431542 and TGFβ2 might explain why we do not observe a significant reduction in the senescence with the inhibitor alone as compared to the control condition (0S).
However, we cannot exclude that other senescence-modulating players could be playing a role in compensating the absence of complete TGFβ signaling. One strong candidate might be IGF-1.
Our results showed that palbociclib inhibited cell proliferation and reduced Igf1 expression. IGF-1 is a crucial survival and proliferation factor for otic precursors in a variety of species from chicken to mammals 16,44 . IGF-1 actions in the developing inner ear are mainly mediated by the RAF-MEK-ERK cascade and the PI3K/AKT pathway 17,19 . Inhibition of PI3K/AKT increased senescence in the otocyst, suggesting that this pathway is repressing senescence as reported during mouse development 12 . On the contrary, the inhibition of RAF kinases by sorafenib reduces senescence, suggesting that this pathway promotes both proliferation and senescence. Accordingly, Storer et al., 2013 showed the non-autonomous involvement of the RAF-MEK-ERK pathway in the induction of developmental senescence in the mouse AER. These data suggest that senescence could be one of the mechanisms activated secondarily to proliferation to control the extent of cell division needed to sculpt the otocyst. B-RAF expression shows a highly specific pattern in the chicken developing inner ear that is very intense in the endolymphatic duct 19 , thus our data also suggest that the degree of activation of RAF kinases might be the trigger that switches cell fate from proliferation to senescence. Accordingly, it has been reported that increased B-RAF activity may correlate with cell cycle arrest and the expression of typical cell senescence markers 45 .
IGF-1 is a survival factor whose proliferative actions have been well-documented 36 . Our data show that IGF-1 induced proliferation and survival but also an increase in the number of cell cycle-arrested senescent cells. Therefore, our data suggest that as a consequence of IGF-1 strong induction of proliferation there is an immediate induction of senescence to sculpt the otocyst in those areas where a more active remodelling is needed, as in the endolymphatic duct. Indeed, IGF-1 treatment during 24 h turns a rounded otic vesicle into a pear-shaped structure with the primordium of the endolymphatic duct evidenced at the most dorsal region 46 . The fact that IGF-1, IGF1R and B-RAF are expressed at the endolymphatic duct during chicken otic development 16,19 , is consistent with this hypothesis.
In summary, our work shows that cellular senescence occurs in the developing inner ear. TGFβ2 is a potent inductor of senescence that may be instructing endolymphatic duct morphogenesis. We also show here that cell proliferation induced by IGF-1 via activation of RAF kinases modulates developmental senescence and, in turn, morphogenetic events.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.