Spontaneous hair cell regeneration in the neonatal mouse cochlea in vivo

Abstract

Loss of cochlear hair cells in mammals is currently believed to be permanent, resulting in hearing impairment that affects more than 10% of the population. Here, we developed two genetic strategies to ablate neonatal mouse cochlear hair cells in vivo. Both Pou4f3(DTR/+) and Atoh1-CreER™; ROSA26(DTA/+) alleles allowed selective and inducible hair cell ablation. After hair cell loss was induced at birth, we observed spontaneous regeneration of hair cells. Fate-mapping experiments demonstrated that neighboring supporting cells acquired a hair cell fate, which increased in a basal to apical gradient, averaging over 120 regenerated hair cells per cochlea. The normally mitotically quiescent supporting cells proliferated after hair cell ablation. Concurrent fate mapping and labeling with mitotic tracers showed that regenerated hair cells were derived by both mitotic regeneration and direct transdifferentiation. Over time, regenerated hair cells followed a similar pattern of maturation to normal hair cell development, including the expression of prestin, a terminal differentiation marker of outer hair cells, although many new hair cells eventually died. Hair cell regeneration did not occur when ablation was induced at one week of age. Our findings demonstrate that the neonatal mouse cochlea is capable of spontaneous hair cell regeneration after damage in vivo. Thus, future studies on the neonatal cochlea might shed light on the competence of supporting cells to regenerate hair cells and on the factors that promote the survival of newly regenerated hair cells.

Keywords: Atoh1; Diphtheria toxin; Direct transdifferentiation; Fate mapping; Lgr5; Mitotic regeneration.

Fig. 1.

Progressive HC death in the Pou4f3^DTR/+ model. Projection images of Myo7a immunofluorescence in cochlear whole-mounts of control wild-type mice at P2 (A-C) and Pou4f3^DTR/+ mice at P2 (D-F), P5 (G-I) and P7 (J-L) after diphtheria toxin (DT) injection at P1. Repopulation of HCs was most robust in the apical turn at P7 (J). OHC, outer hair cells; IHC, inner hair cells. Scale bar: 50 μm.

Fig. 2.

Progressive HC death in the Atoh1DTA model. (A-O) Projection images of Myo7a immunofluorescence in cochlear whole-mounts from control mice (lacking either the Cre or DTA allele) at P2 (A-C) and Atoh1DTA mice at P2 (D-F), P4 (G-I), P7 (J-L) and P15 (M-O). Repopulation of HCs was most robust in the apical turn at P4 (G). (P,Q) 3D reconstruction of confocal z-stack images with SC nuclei labeled by Sox2 (green) and HCs labeled by Myo7a (magenta) in the middle turn of control (P) and Atoh1DTA (Q) cochleae at P7. Scale bar: 50 μm.

Fig. 3.

Fate mapping of SCs in the Pou4f3^DTR/+ model. Confocal images of tdTomato⁺ (magenta) HCs (Myo7a, green) in the apical cochlear turn of control (Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+}) (A) and Pou4f3^DTR/+; Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+} (B) mice at P7. (C) Confocal image of tdTomato⁺/Myo7a⁺ HCs that also express Sox2 (blue) in the apical turn of Pou4f3^DTR/+; Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+} mice at P7. (D-G) Cross-section focused on the tdTomato⁺/Sox2⁺ HC indicated by the arrow in C. Note that GFP expression from the Lgr5^CreER/+ allele is much weaker than Sox2 labeling. Number of double (Myo7a⁺/tdTomato⁺ or Myo7a⁺/Sox2⁺) or triple (Myo7a⁺/Sox2⁺/tdTomato⁺) labeled cells in the apical (H), middle (I) and basal (J) turns of Pou4f3^DTR/+; Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+} mice and control littermates. Data are expressed as mean ± s.e.m., n=3. ^*P<0.05, ^***P<0.001 compared with control number of the corresponding turn as determined by a two-way ANOVA followed by a Student’s t-test with a Bonferroni correction. Scale bars: A,B, 20 μm; C-G, 10 μm.

Fig. 4.

Fate mapping of SCs in the Atoh1DTA model. (A,B) X-Gal staining (blue) in Hes5-nlsLacZ cochlea at P1. Cochlear turns are labeled as apical (A), middle (M) and basal (B). (C-H) Confocal images of the apical turn of Hes5-nlsLacZ mice at P1. lacZ expression is detected with anti-β-gal antibody (green) and is specific to SCs. HCs are labeled by parvalbumin (PVALB; magenta). No β-gal⁺ cells were detected in control samples lacking the Hes5-nlsLacZ allele. Deiters’ cells (DC), outer pillar cells (OPC) and inner pillar cells (IPC) are labeled by Prox1 (blue). (I-K) Confocal images of β-gal⁺ (green) HCs (Myo7a, magenta) in the apical turn of Atoh1DTA; Hes5-nlsLacZ^+/- mice at P2. (L) Cross-section focused on the β-gal⁺ HC labeled by the arrow in I-K. (M) Transverse section of a littermate control (lacking either the Cre or DTA allele) at P2, in which all β-gal⁺ cells are in the SC nuclear layer. Scale bars: 200 μm in A,B; 10 μm in C-M.

Fig. 5.

Mitotic HC regeneration in the neonatal mouse cochlea. Confocal images of EdU (blue) incorporation in Sox2⁺ SCs (green, A-C) and Myo7a⁺ cells (magenta, D-F) in the apical turn of Pou4f3^DTR/+ mice at P7 after EdU injections at P3-P5. Some EdU⁺ HCs (Myo7a⁺) were also co-labeled with Sox2 (G-I). EdU⁺ SCs (Sox2, J,K) and EdU⁺/Myo7a⁺ cells (L,M) were also observed in the apical turn of Atoh1DTA mice 24 hours after EdU injection at P5. (N-O) Cross-section focused on the EdU⁺ nucleus of the cells indicated by the arrowhead and arrow in M. (P-R) Confocal images of EdU⁺ (blue) HCs (Myo7a, magenta) co-labeled with Sox2 (green) in the apical turn of Atoh1DTA mice 24 hours after EdU injection at P4. Scale bars: 20 μm in A-F; 10 μm in G-O.

Fig. 6.

Both mitotic regeneration and direct transdifferentiation occur in the neonatal mouse cochlea. (A-C) Confocal images of tdTomato⁺ (magenta) HCs (Myo7a, green) that are labeled by EdU (blue) in the apical turn of Pou4f3^DTR/+; Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+} mice at P7 after EdU injections at P3-P5. (D-G) Cross-section focused on the tdTomato⁺/EdU⁺ HC indicated by the arrow in A. Note that GFP expression from the Lgr5^CreER/+ allele is much weaker than EdU labeling. In the apical turn of the same organs, there were also EdU^-/Myo7a⁺/tdTomato⁺ cells (H-J). (I,J) Higher magnification of the boxed region in H. Scale bars: 20 μm.

Fig. 7.

Regenerated HCs are not derived from original, differentiated HCs. (A,B) Confocal images of tdTomato⁺ (magenta) HCs (Myo7a, green) in the apical turn of Atoh1-CreER™; ROSA26^CAG-tdTomato mice at P6 after tamoxifen injection at P0/P1. Nuclei are labeled by Hoechst (blue). Inset is a high-magnification image of tdTomato-labeled HCs. (C-E) Confocal images of EdU (blue) incorporation in Myo7a⁺ cells in the apical turn of Atoh1-CreER™; ROSA26^{DTA/CAG-tdTomato} mice 24 hours after EdU injection at P4. DTA^- HCs were traced with tdTomato (magenta). All EdU⁺/Myo7a⁺ cells were tdTomato^-. Scale bars: 100 μm in A,B, 50 μm in inset; 10 μm in C-E.

Fig. 8.

Myo7a⁺ cells are active in the cell cycle. (A-D) Confocal image of EdU incorporation (green) in Myo7a⁺ cells (magenta) in the Atoh1DTA model 4 hours after EdU injection at P4. (E-H) A pH3⁺ (green) HC (Myo6, magenta) was observed at P4 in the Atoh1DTA model. (G-H) Cross-section focused on the pH3⁺ HC indicated by the arrow in E. Note that Myo6 expression of this cell is less robust than in adjacent cells. (I-K) An EdU⁺ (green) HC (calbindin, magenta) with mitotic figures (Hoechst, grayscale) was observed in the Atoh1DTA model 24 hours after EdU injection at P4. (L,M) Control tissue (lacking either the Cre or DTA allele) shows expression of calbindin (magenta) in HC nuclei in tissues that were processed for EdU staining. (N,O) A pH3⁺ (green) HC (Myo7a, magenta) was observed at P7 in the Pou4f3^DTR/+ model. Scale bars: 10 μm.

Fig. 9.

Regenerated HCs are similar to endogenous HCs. Confocal images of EdU⁺ (green) cells co-labeled with HC markers in the apical turn of Atoh1DTA mice. (A-C) Four hours after EdU injection at P5, EdU⁺ cells also express calbindin (magenta). (D-F) Two days after EdU injection at P4 (at P6), EdU⁺ cells express parvalbumin (PVALB; magenta). (G-I) Six days after EdU injection at P4 (at P10), EdU⁺ cells express prestin (magenta). Note that prestin is expressed in the cytoplasm of the EdU⁺ HC indicated by the arrowhead, which is characteristic of a young HC (Mahendrasingam et al., 2010). (J-L) Eleven days after EdU injection at P4 (at P15), EdU⁺ cells (blue) have espin⁺ (green) stereocilia bundles. (L) Cross-section focused on the EdU⁺ HCs indicated by the arrow in J and K. (M-O) Scanning electron micrographs of the apical turn in Atoh1DTA mice at P15. (N) High-magnification image of M, showing an immature stereocilia bundle. The immature bundle in O still has a kinocilium. (P-R) Confocal images of EdU⁺ (green) HCs (Myo7a, magenta) in the apical turn of Atoh1DTA mice at P10 after EdU injection at P4. Scale bars: 10 μm in A-L,P-R; 1 μm in M-O.

Fig. 10.

No signs of HC regeneration when HC loss occurs 1 week after birth in the Pou4f3^DTR/+ model. Projection images of Myo7a immunofluorescence in cochlear whole-mounts of control wild-type mice at P9 (A-C) and Pou4f3^DTR/+ mice at P9 (D-F) and P11 (G-I) after DT injection at P6. Confocal images of tdTomato⁺ (magenta) HCs (Myo7a, green) in the apical turn of control (Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+}) (J) and Pou4f3^DTR/+; Lgr5^CreER/+; ROSA26^{CAG-tdTomato/+} (K) mice that were given tamoxifen at P1, DT at P6, and analyzed at P9. Scale bars: 50 μm in A-I; 20 μm in J,K.

Fig. 11.

No signs of HC regeneration when HC loss occurs 1 week after birth in the PrestinDTA model. (A-C) Prestin-CreER^T2; ROSA26^{CAG-ZsGreen/+} mice induced with tamoxifen from P6-P8 and analyzed at P15 have robust Cre activity in all outer HCs. HCs are labeled with Myo7a (magenta). Projection images of Myo7a immunofluorescence in cochlear whole-mounts of control mice (lacking the Cre or DTA allele) at P8 (D-F) and PrestinDTA mice at P8 (G-I) and P11 (J-L) given tamoxifen from P6-P8. Scale bars: 50 μm.