Sox2 interacts with Atoh1 and Huwe1 loci to regulate Atoh1 transcription and stability during hair cell differentiation

Abstract

Stem cell pluripotency gene Sox2 stimulates expression of proneural basic-helix-loop-helix transcription factor Atoh1. Sox2 is necessary for the development of cochlear hair cells and binds to the Atoh1 3' enhancer to stimulate Atoh1 expression. We show here that Sox2 deletion in late embryogenesis results in the formation of extra hair cells, in contrast to the absence of hair cell development obtained after Sox2 knockout early in gestation. Sox2 overexpression decreased the level of Atoh1 protein despite an increase in Atoh1 mRNA. Sox2 upregulated E3 ubiquitin ligase, Huwe1, by direct binding to the Huwe1 gene. By upregulating its cognate E3 ligase, Sox2 disrupts the positive feedback loop through which Atoh1 protein increases the expression of Atoh1. We conclude that Sox2 initiates expression, while also limiting continued activity of bHLH transcription factor, Atoh1, and this inhibition represents a new mechanism for regulating the activity of this powerful initiator of hair cell development.

Fig 1. Conditional knockout of Sox2 gives rise to extra inner hair cells.

(A) Diagram shows the surface and section views of the postnatal cochlea. The sensory epithelium is composed of 3 rows of outer hair cells and 1 row of inner hair cells (green cells expressing Atoh1) surrounded by supporting cells (white cells with red nucleus expressing Sox2). (B) Timeline of tamoxifen (TAM) injection and assessment of the effect of Sox2 knockout. (C) Extra hair cells were apparent (yellow asterisks) after Sox2 knockout at E18 in Sox2-CreERT/+;Sox2^fl/+ embryos (also referred to as Sox2 knockouts) and Sox2-CreERT/+ embryos (also referred to as Sox2 hypomorphs). The additional hair cells were apparent in a whole mount of the cochlea in a Sox2 knockout, whereas a Cre-negative littermate control had only a single row of inner hair cells and a heterozygous knockout had a small number of extra inner hair cells. Myosin VIIa labels hair cells. Scale bar = 100 μm. (D) Quantification of extra hair cells. Extra hair cells from control and Sox2 conditional knockout cochlea from (B) are shown (n = 4 for each group). Error bars indicate SEM. **p < 0.01, ***p < 0.001. (E) Examination of the cochlea at E21 after Sox2 deletion at E18 showed an extra row of hair cells (arrowheads). Hair cells expressed myosin VIIa (white). Nuclei were labeled with DAPI in blue. A littermate control (Cre-negative) with a single allele of Sox2 had normal hair cells. White bracket indicates outer hair cells and yellow bracket indicates inner hair cells. Scale bar = 25 μm.

Fig 2. Sox2 stimulates degradation of Atoh1 through ubiquitin E3 ligase, Huwe1, and knockout of Sox2 or Huwe1 results in supporting cell division.

(A) qRT-PCR analysis of Atoh1 after Sox2 overexpression in cochlear cells. OC-1 cells were transfected with the indicated concentration of Sox2. Lysates were subjected to qRT-PCR. Error bars indicate SEM (n = 3 experiments). (B) Downregulation of Atoh1 protein by Sox2 overexpression. OC-1 cells were transfected with the indicated concentrations of Sox2. Lysates were processed for Western blotting with Atoh1 or β-actin (loading control) antibodies. Atoh1 levels (determined by densitometry) are shown below each lane. (C) qRT-PCR analysis of Huwe1 in OC-1 cells. The samples from (B) were analyzed by qRT-PCR. (n = 3 experiments). (D) Timeline of tamoxifen and EdU induction. Tamoxifen (TAM) was injected for two consecutive days, followed by EdU. (E) EdU incorporation in Sox2-deleted cochlea at P5. A Sox2-CreERT/+;Sox2^fl/+ cochlea examined at P5 after deletion of Sox2 by administration of tamoxifen at P0 and EdU at P2 had numerous proliferated supporting cells. White bracket indicates outer hair cells and yellow bracket indicates inner hair cells. An orthogonal scan was performed at the position indicated by the white dotted line. Scale bars = 25 μm. (F) Quantitative analysis of EdU-positive cells. Error bars indicate SEM (Sox2^+/-, n = 4 experiments; Sox2-CreERT/+;Sox2^fl/+, n = 3 experiments; ***p < 0.001). (G) Extra hair cells were apparent (yellow asterisks) after Huwe1 knockout in Sox2-positive supporting cells at P0, examined at P5. The additional hair cells were apparent in a whole mount of the cochlea in a Huwe1 knockout. Myosin VIIa labels hair cells. Scale bar = 100 μm. (H) EdU staining of Sox2-CreERT/+;Huwe1^fl/y at P5. Tamoxifen was induced at two consecutive days, P0 and P1 followed by EdU at P2 and P3. White bracket indicates outer hair cells and yellow bracket indicates inner hair cells. An orthogonal scan was performed at the position indicated by the white dotted line. EdU positive cells could be observed in Sox2-CreERT/+;Huwe1^fl/y cochlea. Scale bar = 25μm. (I) Quantitative analysis of EdU positive cells in the pillar cell area. Error bars indicate SEM (n = 4 experiments; ***p < 0.001).

Fig 3. Sox2 is upstream of Atoh1 and Huwe1 in embryonic development and Huwe1 knockdown results in hair cell generation from dividing supporting cells.

(A) qRT-PCR showed that Huwe1 and Sox2 were expressed in the cochlea at E12. Atoh1 was not detected at that time point but was detected at E15. Sox2 was expressed before Atoh1 and was maintained throughout the downregulation of Atoh1. Huwe1 was expressed prior to Atoh1, and expression was also maintained. Relative expression levels were compared with E15 and normalized to Gapdh. Error bars indicate SEM (n = 3 experiments). (B) Huwe1 expression was downregulated after conditional knockout of Sox2 in the embryonic cochlea. Female Sox2^fl/+mice were mated with Sox2-CreERT/+ males and received tamoxifen at E17.5 and E18.5. Embryonic cochlea was analyzed by qRT-PCR at E21. Error bars indicate SEM (n = 3 experiments). (C) Timeline of tamoxifen and EdU induction. (D) The EdU-labeled cells were seen at P5 in the hair cell layer after Huwe1 deletion in a Sox2-CreERT/+;Huwe1^fl/y cochlea induced by tamoxifen at P0. An orthogonal scan was performed (D’) at the position indicated by the white dashed line. Scale bar = 25 μm. (E) Optical section at the supporting cell level shows EdU-positive supporting cells (yellow arrowheads) after administration of EdU at P2 with tamoxifen at P0. An orthogonal scan was performed (E’) at the white dashed line. The yellow bracket indicates newly generated supporting cells. Scale bar = 25 μm. (F) Neonatal organ of Corti is divided into 4 regions. (G) Effect of Huwe1 knockdown on the organ of Corti. Organs of Corti treated with Huwe1 siRNA (100 nM) for 72 hours have increased numbers of hair cells, which are marked by myosin VIIa. The supernumerary hair cells are positive for phalloidin, a hair bundle marker. The scale bar is 100 μm. (H) Quantification of hair cell counts in the apex, mid-apex, mid-base, and base (mean ± SEM per 100 mm; *p < 0.05, ***p<0.001, n = 7 for both groups). (I) Double-labeled cells positive for the tdTomato reporter (red) and myosin VIIa (blue) were found in the pillar area in the mid-apex region of cochlear tissue from neonatal mice carrying the Sox2-CreERT as well as the tdTomato reporter 3 days after treatment with Huwe1 siRNA (100 nM). Td-Tomato reporter in myosin VIIa-positive cells indicates hair cells that arose from supporting cells. Scale bar = 25 μm. (J) Effect of Huwe1 knockdown was significant. Quantification of the tdTomato reporter-positive hair cells in scrambled and Huwe1-siRNA treated explants showed significantly more reporter-labeled hair cells after Huwe1 siRNA treatment (mean ± SEM per 100 mm; *p < 0.05, n = 4 for both groups).

Fig 4. Sox2 direct interaction with mouse Huwe1 and human Huwe1 genes.

(A) Schematic diagram shows four putative Sox2-binding motifs (1–4) in the region 100-kb upstream of the translation start site, and three putative Sox2 binding motifs (5–7) in the region 50-kb downstream of the stop codon of mouse Huwe1. A random motif downstream of the translation stop codon of mouse Huwe1 (8) was the control (ctrl). (B) Chromatin immunoprecipitation for Sox2 binding sites at the mouse Huwe1 locus in cochlear organoids which were proliferated for 10 days and differentiated for 2 days. The fold enrichment for each site after immunoprecipitation with an antibody to Sox2 compared to normal goat IgG is shown. Error bars indicate SEM (n = 3 experiments; *p<0.05, **p<0.01, ***p<0.005 calculated using T test). (C) Schematic diagram shows two putative Sox2-binding motifs (A, B) in the region 42-kb upstream of the translation start site, and three putative Sox2 binding motifs (C-E) in the region 10-kb downstream of the stop codon of human Huwe1. A random motif downstream of the translation stop codon of human Huwe1 (F) was the control (ctrl). (D) Chromatin immunoprecipitation for Sox2 (HA) binding sites at the human Huwe1 locus after overexpression of Sox2-HA in HEK 293T cells. The fold enrichment for each site after immunoprecipitation with an antibody to HA (Sox2) compared to mouse IgG is shown. Error bars indicate SEM (n = 3 experiments; *p<0.05, **p<0.01, ***p<0.005 calculated using T test).

Fig 5. Huwe1 is required for Sox2 stimulated degradation of Atoh1.

(A) Atoh1 mRNA upregulation by Sox2. HEK-293T cells were transfected with the indicated concentrations of Sox2 followed by qRT-PCR. Error bars indicate SEM (n = 3 experiments; **p < 0.01, ***p < 0.001). (B) Lysates from (A) were processed for Western blotting with FLAG antibody (Atoh1), or β-actin antibody (loading control) and quantified by densitometry (Atoh1 levels are shown below the lanes). (C) Downregulation of Atoh1 by Sox2 overexpression is rescued by proteasome inhibition. HEK 293T cells were transfected with HA-Sox2 and/or treated with proteasome inhibitor, MG132 (10 μM), for 6 hours. After 24 hours of transfection, lysates were processed for Western blotting with FLAG antibody, Sox2 antibody, or β-actin antibody (loading control). (D) Huwe1 knockdown prevented Atoh1 degradation like proteasome inhibitor, MG132. Both the baseline and the Sox2-induced decrease in Atoh1 in HEK 293T cells were prevented by Huwe1 shRNA and MG132. Atoh1 levels (determined by densitometry) are shown below each lane. Lysates were processed for Western blotting with FLAG (Atoh1), Sox2, or β-actin (loading control) antibodies. (E), (F) qRT-PCR of Atoh1 and Huwe1. Sox2 overexpression in HEK 293T cells upregulated both Atoh1 and Huwe1, while Huwe1 shRNA and MG132 did not affect Atoh1 mRNA. Error bars indicate SEM (n = 3 experiments). (G), (H) The decreased Atoh1 protein half-life upon Sox2 overexpression could be rescued by Huwe1 knockdown in a cycloheximide chase (CHX) assay in HEK 293T cells transiently transfected with FLAG-HA-Atoh1, HA-Sox2, and Huwe1 shRNA. Error bars indicate SEM (n = 3 experiments). (I) Sox2 induces Atoh1 transcription by binding to the Atoh1 3’ enhancer. The Atoh1 protein binds to its 3’ enhancer to upregulate transcription in a positive feedback loop (1) and activates downstream genes (2). Sox2 also activates transcription of E3 ligase Huwe1, which binds Atoh1 (3) and initiates proteasomal degradation by ubiquitylation. Degradation of Atoh1 breaks the positive feedback loop through which Atoh1 upregulates Atoh1 transcription.

Fig 6. Cochlear phenotypes obtained after Sox2 deletion.

Sox2 is expressed in cochlear supporting cells and hair cells, both of which are derived from prosensory cochlear progenitors. The time of treatment is plotted from left to right and shows that prosensory cochlear progenitor development requires Sox2 and that after Atoh1 expression, Sox2 switches from its role in keeping proneural genes inactive and allowing the expansion of progenitors to a proneural role by activating transcription of Atoh1. At E18, Sox2 is needed to prevent overexpression of Atoh1 protein through activation of Huwe1, thus exemplifying its role in the initiation of cell fate. Sox2 is silenced shortly after the establishment of the hair cells at birth.