Reprogramming with Atoh1, Gfi1, and Pou4f3 promotes hair cell regeneration in the adult organ of Corti

Abstract

Cochlear hair cells can be killed by loud noises, ototoxic drugs, and natural aging. Once lost, mammalian hair cells do not naturally regenerate, leading to permanent hearing loss. Since the mammalian cochlea lacks any intrinsic ability to regenerate, genetic reprogramming of cochlear supporting cells that lie adjacent to hair cells is a potential option for hearing restoration therapies. We targeted cochlear supporting cells with three hair cell transcription factors: Atoh1, or Atoh1 + Gfi1, or Atoh1 + Gfi1 + Pou4f3 and found that 1- and 2-factor reprogramming is not sufficient to reprogram adult supporting cells into hair cells. However, activation of all three hair cell transcription factors reprogrammed some adult supporting cells into hair cell-like cells. We found that killing endogenous hair cells significantly improved the ability of supporting cells to be reprogrammed and regenerated numerous hair cell-like cells throughout the length of the cochlea. These regenerated hair cell-like cells expressed myosin VIIa and parvalbumin, as well as the mature outer hair cell protein prestin, were innervated, expressed proteins associated with ribbon synapses, and formed rudimentary stereociliary bundles. Finally, we demonstrate that supporting cells remained responsive to transcription factor reprogramming for at least 6 weeks after hair cell damage, suggesting that hair cell reprogramming may be effective in the chronically deafened cochlea.

Keywords: hair cells; inner ear; reprogramming.

Fig. 1.

Combined Atoh1, Gfi1, and Pou4f3 reprogram supporting cells into hair cell-like cells in the mature cochlea. A) Recombination in Lfng-Cre^ERT2::Rosa26^tdTomato in the mature organ of Corti following TM induction. tdTOMATO expression is observed in supporting cells adjacent to hair cells. CreER-mediated recombination was activated at 3 weeks of age with 2 TM injections (9 mg/40 g) 24 h apart. Samples were collected at 4, 5, or 6 weeks of age; the image is from a sample at 5 weeks of age. TM B) Myosin VIIA positive hair cells and SOX2-positive supporting cells are observed in the cochlear epithelium. Reprogrammed hair cells were observed in the cochlea after Atoh1, Gfi1, and Pou4f3 misexpression (arrows) but not in cochleas after Atoh1 misexpression alone. Arrows indicate reprogrammed cells, scale bar = 50 µm. C) Quantification of ectopic hair cells at 4, 5, or 6 weeks of age in control cochlea as well as reprogramming with either Atoh1, Gfi1 + Atoh1, or Gfi1 + Atoh1 + Pou4f3. Significantly more reprogrammed hair cells are observed with Gfi1 + Atoh1 + Pou4f3 reprogramming compared with Atoh1 or Gfi1 + Atoh1. In addition, significantly more reprogrammed cells are observed in the outer hair cell region compared with the inner hair cell region and more regenerated hair cells were observed at 5 and 6 weeks of age compared with 4 weeks. Ctrl = control, A = Atoh1, GA = Gfi1 + Atoh1, GAP = Gfi1 + Atoh1 + Pou4f3. Significance indicated by asterisks ****P < 0.0001.

Fig. 2.

Hair cell death increases the number of reprogrammed hair cell-like cells. A) Reprogramming and fate mapping of cochlear supporting cells was activated at 3 weeks of age with 2 TM injections (TM, 9 mg/40 g) ∼24 h apart. Endogenous hair cells were killed at 3 weeks of age by injecting DT into Pou4f3^DTR -positive animals ∼6 h after the second TM injection. Supporting cell fate was traced with tdTOMATO. Near complete loss of endogenous hair cells was observed following damage. Expression of Atoh1 (A) alone did not reprogram supporting cells into hair cell-like cells while Gfi1 + Atoh1 + Pou4f3 (GAP) reprogrammed significantly more supporting cells into hair cells thus regenerating them. Regenerated hair cells were observed both medial and lateral to the pillar cells. TM = tamoxifen, DT = diphtheria toxin, hrs = hours. A = Atoh1, GAP = Gfi1 + Atoh1 + Pou4f3. Scale bar = 20 µm. B) Significantly more hair cells were reprogrammed following GAP expression when hair cells are killed compared with when the original hair cells are present. Atoh1 expression alone was not sufficient to regenerate hair cells, however, GAP combined expression reprogrammed hair cells both medial and lateral to the pillar cells. C) Regenerated hair cells were observed throughout the entire length of the cochlea. D) Regenerated hair cells were observed in similar numbers following 2 weeks of reprogramming and 5 weeks of reprogramming. Scale bar = 200 µm. Significance indicated by asterisks **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 3.

Regenerated hair cells develop structural components of endogenous hair cells. A) Endogenous hair cells (Myosin VIIa) of the intact cochlea are innervated by spiral ganglion neurons (TUBB3). Following hair cell killing, hair cells and neurons are absent from the organ of Corti. Neurites reinnervate the organ of Corti, making contact with regenerated hair cells. B) Inner and outer hair cells in the intact organ of Corti express the ribbon synapse protein ribeye (anti-CTBP2). Hair cell killing results in the loss of both hair cells and CTBP2, while regenerated hair cells exhibit CTBP2-positive puncta. C) SEM image showing that hair cell killing results in the loss of stereocilia and minimal presence of microvilli on SC apical surfaces. Following induced Atoh1, Gfi1, and Pou4f3 expression in the SCs, disorganized and immature stereocilia are present. D) More regenerated hair cells in the medial compartment express parvalbumin at 5–6 weeks of age. Regenerated hair cells turn on Prestin by 6 weeks of age, but not 5 weeks of age. Scale bars in A, B, and D = 20µm.

Fig. 4.

Mature cochlear supporting cells are reprogrammed into hair cell-like cells 6 weeks after hair cell loss. A) Timeline for hair cell killing and reprogramming. Hair cells were killed at 6 weeks of age. Reprogramming and tdTomato were activated in supporting cells via TM injections at 12 weeks of age followed by analysis at 16 weeks of age. B) In control cochlea with no damage, hair cells are observed in the expected three rows of outer hair cells and one row of inner hair cells with adjacent tdTOMATO-positive supporting cells. Following DT mediated damage, myosin VIIa + cells are absent however supporting cells remain. Following combined hair cell damage + GAP reprogramming, regenerated hair cells were observed in both the medial and lateral compartment of the organ of Corti. Scale bar = 50 µm. C) Quantification of the total number of MYO7A/tdTOMATO-double positive cells in undamaged and damaged samples without reprogramming as well as samples with both damage and reprogramming. Significance is indicated by asterisks **P < 0.01, ***P < 0.001.