Epigenetic regulation of Atoh1 guides hair cell development in the mammalian cochlea

Abstract

In the developing cochlea, sensory hair cell differentiation depends on the regulated expression of the bHLH transcription factor Atoh1. In mammals, if hair cells die they do not regenerate, leading to permanent deafness. By contrast, in non-mammalian vertebrates robust regeneration occurs through upregulation of Atoh1 in the surviving supporting cells that surround hair cells, leading to functional recovery. Investigation of crucial transcriptional events in the developing organ of Corti, including those involving Atoh1, has been hampered by limited accessibility to purified populations of the small number of cells present in the inner ear. We used µChIP and qPCR assays of FACS-purified cells to track changes in the epigenetic status of the Atoh1 locus during sensory epithelia development in the mouse. Dynamic changes in the histone modifications H3K4me3/H3K27me3, H3K9ac and H3K9me3 reveal a progression from poised, to active, to repressive marks, correlating with the onset of Atoh1 expression and its subsequent silencing during the perinatal (P1 to P6) period. Inhibition of acetylation blocked the increase in Atoh1 mRNA in nascent hair cells, as well as ongoing hair cell differentiation during embryonic organ of Corti development ex vivo. These results reveal an epigenetic mechanism of Atoh1 regulation underlying hair cell differentiation and subsequent maturation. Interestingly, the H3K4me3/H3K27me3 bivalent chromatin structure observed in progenitors persists at the Atoh1 locus in perinatal supporting cells, suggesting an explanation for the latent capacity of these cells to transdifferentiate into hair cells, and highlighting their potential as therapeutic targets in hair cell regeneration.

Keywords: Epigenetics of Atoh1 regulation; Epigenetics of inner ear development; Mouse; Sensory hair cell differentiation.

Fig. 1.

Micro-chromatin immunoprecipitation (µChIP) shows that the Atoh1 gene is bivalent (H3K27me3+ and H3K4me3+) in prosensory progenitors of the organ of Corti, and that H3K27me3 levels are strongly reduced in differentiating hair cells. (A) Relative Atoh1 mRNA levels in the cochlea increase during hair cell differentiation. Levels peak at E17.5 and then decrease during postnatal maturation. Atoh1 mRNA levels extracted from whole cochleae were analyzed at each time point by qPCR. Atoh1 expression levels are normalized to Gapdh as internal reference. Results are mean±s.e.m. (n=3). (B) Representative fluorescence images of whole-mount cochleae and FACS gating used to purify E14.5 progenitors (left, p27^Kip1-GFP transgene), and nascent E17.5 hair cells (right, Atoh1-GFP transgene). (C) Relative Atoh1 mRNA levels in FACS-purified progenitors and hair cells, and in control cell types E14.5 progenitor cells (PG), E17.5 hair cells (HC), mouse embryonic stem cells (mESC) and P1 cerebellar granule cell precursors (GCPs). Atoh1 expression levels are normalized to Gapdh as internal reference. Results are mean±s.e.m. (n=3). (D) µChIP analysis of the Atoh1 locus indicates a change in relative H3K27me3 levels between Atoh1-expressing hair cells (FACS-purified from P1 Atoh1-GFP transgenic mice), compared with prosensory progenitors (FACS-purified from P1 p27^Kip1-GFP transgenic mice). Schematic shows the locations across the Atoh1 locus (sites 1, 2 and 3; triangles) analyzed by µChIP qPCR for the relative abundance of H3K4me3 and H3K27me3. Results are mean±s.e.m., *P<0.05 (n=3).

Fig. 2.

Atoh1 expression and hair cell differentiation is dependent on Atoh1-associated histone acetylation. (A) Schematic shows the locations across the Atoh1 locus (sites 1, 2 and 3; triangles) analyzed by µChIP qPCR for the presence of H3K9ac in FACS-purified E14.5 progenitors (PG) and E17.5 hair cells (HC), as well as control mouse embryonic stem cells, (mESC), cerebellar granule cell precursors (GCPs), and astrocytes. H3K9ac increases upon hair cell differentiation across the Atoh1 locus (sites 1, 2 and 3). Results are mean±s.e.m., *P<0.05 (n=3). (B-D″) HATi blocks Atoh1 upregulation and hair cell differentiation in embryonic organ of Corti explant cultures. Representative images of whole-mount cochlear epithelia from Atoh1-GFP transgenic organ of Corti. Time-mated E13.0 cochlear ducts (B) were cultured in the absence (C,D) or presence (C′,D′) of HAT inhibitor (curcumin) for 24 or 48 h as indicated. Alternatively, organs were cultured in HATi for 24 h, after which the inhibitor was washed out, and organs analyzed after another 24 h (D″). Scale bars: 100 µm. Arrows indicate the approximate origin and direction of progressive Atoh1-GFP expression. (E) Quantification of hair cell number as shown in B-D″. At time point 0 h there were no Atoh1-GFP+ cells. Control organs (black bars) have on average 105±42 Atoh1-GFP+ hair cells by 24 h and 1063±195 cells by 48 h. The presence of HAT inhibitor decreases the number of Atoh1-GFP+ cells (gray bars) to 32±30 by 24 h, and to 113±114 by 48 h of treatment relative to control. Washing out the inhibitor at 24 h restores the number of hair cells to 235±52 (yellow bars) by 48 h. Results are mean±s.e.m., *P<0.05 (n=3). (F) Atoh1 mRNA levels at the 0 h, 24 h, and 48 h time points, and following inhibitor washout, in control and HATi cultures. HATi decreases the relative Atoh1 mRNA levels 95% (from 40.0±3.4 to 12.7±10.0) at 24 h of treatment, and 66% (from 82.0±15.4 to 27.6±16.4) at 48 h compared with control organs. Washout of inhibitor at 24 h, increased Atoh1 mRNA levels 1.9-fold at 48 h compared with time-matched control (from 27.6±16.4 to 52.7±9.6). Atoh1 gene expression is normalized to Gapdh as internal reference. Results are mean±s.e.m., *P<0.05 (n=3). (G) H3K9ac-µChIP qPCR assessment of the Atoh1 transcription start site (TSS) at 0 h, 24 h or 48 h in HATi-treated organotypic cultures. HAT inhibition in nascent hair cells reduced the H3K9ac levels at the Atoh1 transcription start site 37% (from 5.9±1.4% to 3.7±1.1% of input) at 24 h, and 19.8-fold (from 20.9±6.2% to 1.0±0.1%) by 48 h. Washing out the inhibitor increases the H3K9ac level 15.2-fold (from 1.0±0.1% to 16.0±1.0%), to a level almost as high as that of the control organs at 48 h. Results are mean±s.e.m., *P<0.05 (n=3).

Fig. 3.

Downregulation of Atoh1 during hair cell maturation depends on H3K9 de-acetylation. Hair cells (Atoh1-GFP transgenic mice) were FACS-purified for µChIP analysis from E17.5, P1 and P6 mice. (A) Relative Atoh1 mRNA levels were measured and analyzed by qPCR. Expression levels are normalized to Gapdh internal reference. Results are mean±s.e.m. (n ≥3). (B) Schematic shows the locations across the Atoh1 locus (sites 1, 2 and 3; triangles) analyzed by µChIP qPCR for the presence of H3K9ac and H3K9me3 in FACS-purified hair cells at E17.5, P1 and P6. H3K9ac decreases from E17.5 to P6, whereas H3K9me3 levels increase over this period. Data are presented as the percent input DNA in µChIP reactions (% input). Results are mean±s.e.m., *P<0.05 (n=3). (C) HDAC inhibitor (VPA) blocks age-dependent reduction of Atoh1 mRNA (left panel) and H3K9ac levels at the Atoh1 TSS (right panel) in purified hair cells. Atoh1 mRNA relative expression increases 2.6-fold (from 4.8±0.5 to 12.6±0.7; grey bars) at 24 h of treatment compared with control (black bars). Atoh1 expression levels were normalized to Gapdh as internal reference. Results are mean±s.e.m., *P<0.05 (n=3). HDAC inhibition increases the levels of H3K9ac at the Atoh1 TSS 2.3-fold (from 7.2±1.0% to 16.5±2.6%) by 6 h, and 6.6-fold (from 3.1±1.0% to 20.0±8.5%) by 24 h of treatment. Results are mean±s.e.m., *P<0.05 (n=3).

Fig. 4.

Bivalency marks H3K27me3 and H3K4me3 are maintained in perinatal supporting cells. Schematic shows the locations across the Atoh1 locus (sites 1, 2 and 3; triangles) analyzed by µChIP qPCR for the presence of H3K27me3 and H3K4me3. Hair cells (Atoh1-GFP transgenic mice) and supporting cells (p27^Kip1-GFP transgenic mice) were FACS-purified from P1 and P6 mice. H3K27me3 levels remain relatively unchanged in P6 supporting cells, relative to P1. H3K4me3 is also maintained at similar levels in both P1 and P6 supporting cells. Moreover, both marks are maintained at higher levels in supporting cells at P1 and P6 compared with hair cells. Red bars, H3K27me3; green bars, H3K4me3. Data are presented as percentage of input. Results are mean±s.e.m., *P<0.05 (n=3).