Artificial induction of Sox21 regulates sensory cell formation in the embryonic chicken inner ear

Abstract

During embryonic development, hair cells and support cells in the sensory epithelia of the inner ear derive from progenitors that express Sox2, a member of the SoxB1 family of transcription factors. Sox2 is essential for sensory specification, but high levels of Sox2 expression appear to inhibit hair cell differentiation, suggesting that factors regulating Sox2 activity could be critical for both processes. Antagonistic interactions between SoxB1 and SoxB2 factors are known to regulate cell differentiation in neural tissue, which led us to investigate the potential roles of the SoxB2 member Sox21 during chicken inner ear development. Sox21 is normally expressed by sensory progenitors within vestibular and auditory regions of the early embryonic chicken inner ear. At later stages, Sox21 is differentially expressed in the vestibular and auditory organs. Sox21 is restricted to the support cell layer of the auditory epithelium, while it is enriched in the hair cell layer of the vestibular organs. To test Sox21 function, we used two temporally distinct gain-of-function approaches. Sustained over-expression of Sox21 from early developmental stages prevented prosensory specification, and abolished the formation of both hair cells and support cells. However, later induction of Sox21 expression at the time of hair cell formation in organotypic cultures of vestibular epithelia inhibited endogenous Sox2 expression and Notch activity, and biased progenitor cells towards a hair cell fate. Interestingly, Sox21 did not promote hair cell differentiation in the immature auditory epithelium, which fits with the expression of endogenous Sox21 within mature support cells in this tissue. These results suggest that interactions among endogenous SoxB family transcription factors may regulate sensory cell formation in the inner ear, but in a context-dependent manner.

Figure 1. Expression of Sox21 during chicken inner ear development.

(A) Sox21 mRNA showed faint levels of expression at E3 in the otic vesicle. (B) Upregulation of Sox21 expression was observed in the presumptive posterior crista and the presumptive superior crista regions at E5. (C) At E7 Sox21 transcripts were detected in the basilar papilla, the utricle, the saccule and the posterior, superior and lateral cristae. (C’) in the utricle, Sox21 is strongly expressed in the striola region (entire utricle is marked with black dotted line). (C’’) Transverse cryosection through the basilar papilla reveals Sox21 transcripts are restricted to the support cell layer. (C’’’) Transverse cryosection through the utricle reveals that Sox21 transcripts are enriched in the hair cell layer. (D) At E15 Sox21 transcripts were still present in the basilar papilla, the utricle, the saccule and the posterior, superior and lateral cristae. D’) Transverse cryosection through the E15 basilar papilla reveals Sox21 transcripts remain restricted to the support cell layer. (D’’) Transverse cryosection through the E15 utricle reveals Sox21 transcripts are enriched in the hair cell layer. (D’’’) Transverse cryosection through the E15 cristae reveal Sox21 transcripts are enriched in the hair cell layer. Picture shown is of a lateral crista; all cristae exhibited the same expression pattern. ov: otic vesicle; sc: superior crista; pc: posterior crista; lc: lateral crista; ut: utricle; bp: basilar papilla; HC: hair cell layer; SC: support cell layer.

Figure 2. Over-expression of Sox21 at early stages of inner ear development leads to a loss of prosensory identity.

(A) Brightfield and Prox1/Myc/Dapi immunostained views of inner ear dissected from E9 unelectroporated controls and RCAS-Sox21 electroporated embryos. Inner ears over-expressing Sox21 exhibit morphogenesis defects, including a reduction in the size of the cristae (white asterisks) and the presumptive utricle (marked as ”ut/sac” because identification is based on its position, and the reduced patch may also represent the saccule; images are composites of projections of distinct confocal stacks). Box plots of vestibular sensory patch maximum span (B; dashed line in panel A illustrates span of the utricle) and widths and length of the basilar papilla (C) in control (n = 12) and Sox21 (n = 13) transfected whole-mount preparations. Minimum, first quartile, median, third quartile and maximum are displayed. Prox1 (D–D’) and Sox2 (E–E’) expression is reduced in both vestibular and auditory sensory cells over-expressing Sox21 (white arrowheads). Immunostaining for otoferlin/HCA (F–F’) and Tuj1 (G–G’) show that cells over-expressing Sox21 from E2 onwards do not form hair cells or neurons and are not innervated. sc: superior crista; pc: posterior crista; lc: lateral crista; ut: utricle; bp: basilar papilla.

Figure 3. Tol2 system for Dox-inducible gene expression and classification of cell phenotypes.

(A) The pTRE-Sox21-eEGFP Dox-inducible expression vector and experimental design for testing the effects of Sox21 induction at late stages of ear development. (B) MYC-tagged Sox21 was detected in the nuclei of all eGFP positive cells. (C) Classification of eGFP-positive cells into 4 phenotypes: uncommitted progenitor/support cell, in which otoferlin and HCA are not expressed, and with a cytoplasmic process contacting the basal lamina (arrows); immature hair cell type I (imHC (I)) in which HCA is present (arrowhead) but otoferlin is not, and which can exhibit a basal cytoplasmic process (arrows); immature hair cell type II (imHC (II)), with an elongated cell shape and both otoferlin and HCA expressed; Mature hair cell (HC), with a flask-shaped cell body and both otoferlin and HCA expressed.

Figure 4. Induction of Sox21 expression at late embryonic stages promotes hair cell formation in vestibular patches, but not in the basilar papilla.

(A) cultures of transfected inner ear tissue treated for 48 hours with Dox, then immunostained for otoferlin/HCA. With the pTRE-eGFP vector, eGFP-positive cells are mainly uncommitted progenitor/support cell types but include some hair cells (red asterisks). Induction of Sox21-eGFP in vestibular sensory epithelia for 48 hours results in a large majority of induced cells exhibiting a hair cell phenotype (red asterisks). In the basilar papilla, induction of Sox21-eGFP for 48 hours produces a phenotype indistinguishable from that of the control. (B–C) Phenotypic distribution (colors correspond to those used in Figure 3D) of eGFP-positive cells after 12, 24 and 48 hours Dox treatment. In vestibular sensory epithelia, Sox21 induction caused a strong shift towards hair cell phenotypes over time. Total cell counts from 3 separate ear samples are shown above bars.

Figure 5. Short induction of Sox21 reduces Sox2 expression in organotypic cultures of E10 utricle and E7 basilar papilla.

(B) Representative images of Sox2 immunostaining in pTRE-Sox21-eGFP samples after 12 hours of Dox treatment. Induced cells are marked with asterisks, and tend to exhibit reduced levels of Sox2 expression compared to neighbouring untransfected cells (C). Box plot of Z-score values [Z = (x-mean)/stdev] for levels of Sox2 expression in supporting/progenitor cells of the basilar papilla (B; 3 samples) or of the utricle (C; 5 samples), expressing either eGFP only or Sox21-eGFP. Outliers, minimum, first quartile, median, third quartile and maximum are displayed, n = numbers of transfected cells. Sox2 expression levels were significantly lower in Sox21-eGFP expressing cells than in eGFP expressing cells in both the auditory (Mann-Whitney U = 14873; p = 0.00) and vestibular (Mann-Whitney U = 32335; p = 0.00) epithelia.

Figure 6. Induction of sox21 causes a progressive reduction in levels of Notch activity.

(A) surface views of E10 vestibular patches transfected with the pT2K-Hes5::nd2eGFP reporter and either pTRE-FP635 (control) or pTRE-FP635-Sox21 and treated in vitro with Dox for 12 or 48 hours. In control samples, the proportion of FP635-positive cells that were either positive (arrowheads) or negative (arrow) for nuclear Hes5::nd2eGFP expression was comparable after 12 or 48 hours Dox treatment. In contrast, the number of Sox21-FP635 induced cells positive for Hes5::nd2eGFP expression was reduced at both 12 hours and 48 hours of Dox treatment. In the 48 hours example shown, all FP635-Sox21 are negative for Notch activity (white arrows); the transverse reconstruction of the same region (z-view panel) also demonstrate expression of otoferlin/HCA in FP635-Sox21 induced cells. (B) Graph showing the proportion of Notch-active (mean intensity of nuclear d2eGFP signal >500 above background in a 12-bits image) cells among cells induced for FP635-only (FP635) or FP635-Sox21 (Sox21) expression after 12 and 48 hours Dox treatment; n = total number of transfected cells analysed in 3 samples. Standard error bars are shown. (C) Box plots of Z-scores values for Hes5::nd2eGFP fluorescence levels in supporting/progenitor cells transfected with either pTRE-FP635 (n = 180) or pTRE-Sox21-FP635 (n = 178) and treated for 12 hours with Dox. Minimum, first quartile, median, third quartile, and maximum are displayed. There were significantly lower levels of Hes5::nd2eGFP fluorescence in Sox21-FP635 expressing cells than in FP635 expressing cells (Mann-Whitney U = 11232; p = .00; n = total number of transfected cells analysed in 3 samples).

Figure 7. Overexpression of Sox21 does not influence Delta1 and Serrate1 expression.

(A–C) After 12 hours Dox treatment, Delta1 expression does not differ between immature hair cells overexpressing Sox21, and control untransfected immature hair cells (white arrowheads denote the location of immature hair cells; white asterisk denotes Sox21 induced cell. Mature hair cells overexpressing Sox21 do not continue to exhibit Delta1 expression (yellow arrowhead), consistent with untransfected mature hair cells. (D–F): Apical view of Serrate1 expression shows no obvious changes in expression levels between regions of induced Sox21 overexpression and untransfected regions.

Figure 8. Fluorescence of the pT2K-Atoh1::nTomato reporter is not elevated in Sox21-induced cells.

(A) Schematic representation of the Tol2 Dox-inducible Sox21 and the Tol2 Atoh1 reporter constructs. (B) A vestibular crista transfected with pT2K-Hes5::d2eGFP and the pT2K-Atoh1::nTomato reporters and immunostained with otoferlin antibodies; the Atoh1::nTomato fluorescence is strong in hair cells (asterisks), but not in Hes5::d2eGFP positive cells surrounding them (arrowheads). (C–D) Surface views of a BP transfected with pTRE-Sox21-eGFP and pT2K-Atoh1::nTomato at E2 and treated in ovo for 24 hrs with Dox at E6. Atoh1::nTomato-positive cells were found primarily in the central-distal region of the BP (asterisk), but a few Sox21-induced cells in the lateral wall were also positive (arrowheads in C–C’). At higher magnification (D–D’), note that the levels of Atoh1::nTomato fluorescence varied greatly in both Sox21-induced (white arrows and arrowheads) and non-induced (yellow arrows and arrowheads) cells. (E) Box plots of Z-scores values for Atoh1::nTomato fluorescence levels in untransfected (ø) versus Sox21-induced cells in the basilar papilla. Minimum, first quartile, median, third quartile, and maximum are displayed. There was no significant difference between the two categories of cells (Mann-Whitney U = 9818; p = 0.305; n = number of cells analysed in 2 samples). (F–G) surface views of a crista transfected with pTRE-Sox21-eGFP and pT2-Atoh1::nTomato at E2 and treated in ovo for 24 hrs with Dox at E6. Variations in levels of Atoh1 reporter fluorescence are also clearly visible. (H) Box plots of Z-scores values for Atoh1::nTomato fluorescence levels in untransfected (ø) versus Sox21-induced cells in vestibular epithelia. Minimum, first quartile, median, third quartile, and maximum are displayed. There was no significant difference between the two categories of cells (Mann-Whitney U = 2560; p = 0.073; n = number of cells analysed in 2 samples).