Precise alternating cellular pattern in the inner ear by coordinated hopping intercalations and delaminations

Abstract

The mammalian hearing organ, the organ of Corti, is one of the most organized tissues in mammals. It contains a precisely positioned array of alternating sensory hair cells (HCs) and nonsensory supporting cells. How such precise alternating patterns emerge during embryonic development is not well understood. Here, we combine live imaging of mouse inner ear explants with hybrid mechano-regulatory models to identify the processes that underlie the formation of a single row of inner hair cells (IHCs). First, we identify a previously unobserved morphological transition, termed "hopping intercalation," that allows cells differentiating toward IHC fate to "hop" under the apical plane into their final position. Second, we show that out-of-row cells with low levels of the HC marker Atoh1 delaminate. Last, we show that differential adhesion between cell types contributes to straightening of the IHC row. Our results support a mechanism for precise patterning based on coordination between signaling and mechanical forces that is likely relevant for many developmental processes.

Fig. 1.. A single row of IHCs is formed from an initially disordered salt-and-pepper pattern.

(A) Schematic of the mammalian inner ear. The dashed line represents the location of the OoC along the cochlea. A zoom-in on the OoC with marked OHCs, IHCs, SCs, and IPCs is presented on the right. (B) Schematic of the transgenic mice used for imaging. Atoh1-mCherry (HC marker) mice are crossed with ZO1-EGFP (tight junction marker) mice to produce double-reporter mice. (C) Snapshots from movies of cochlear explants of a ZO1-EGFP (green)/Atoh1-mCherry (gray) mouse showing initial (left, E15.5) and advanced (right, E17.5) patterns of the IHC row. Double arrow displays the medial-lateral axis. (D) Snapshots from the initial (left) and final (right) time points of a lateral inhibition simulation over a narrow strip of cells (movie S2). Color bar represents the Notch activity level. Initial pattern is resolved to a salt-and-pepper pattern of low Notch activity HCs (red) and high Notch activity SCs (green). The IPCs (blue), the cells below the IHC domain (dark gray), and the cells above the IPCs (light gray) do not participate in the lateral inhibition process. Snapshots shown are sections from a 25-cell-wide simulation window. (E) A filmstrip from a movie of an E15.5 cochlear explant showing the gradual differentiation and patterning of the IHC row (movie S3). Scale bars, 10 μm.

Fig. 2.. Hopping intercalation of Atoh1⁺ cells toward the IPC row.

(A to C) A filmstrip from an E15.5 cochlear explant showing an apical view of a hopping intercalation event with (A) both Atoh1 and ZO1 markers, (B) only the ZO1 marker, and (C) a segmentation of (B) highlighting intercalating cell (light gray) and IPCs (dark gray) (movie S6). New apical surface opens up [yellow arrow in (B)] and merges with the original apical surface [red arrow in (B)]. (D) A cross section showing a side view of the intercalating cell. The dashed line marks the apical surface of the tissue. Red and yellow arrows mark the original and new apical surfaces of the cell, respectively. (E) Change in apical area of hopping cells. The graph shows the relative change in apical area (area after/area before hopping) for hopping cells. For control, we measured the change in apical area of non-hopping neighboring IHCs at similar times. (F) A cross section of the filmstrip in (A) to (C) [zoomed out with respect to (D)] showing a ZO1 punctum propagating toward the apical surface. Time stamps show time with respect to the initiation of the hopping event (at 00:00 hours:min). (G) A cross section schematic of a hopping intercalation. Statistical test in (E): Welch’s t test, plot shows means ± SEM. n = 12 and 11 repeats of neighbors and hopping cells, respectively. ****P < 0.0001. Scale bars, 5 μm

Fig. 3.. A hybrid modeling approach incorporating hopping intercalations accounts for reduction in patterning defects.

(A) Schematic of the hybrid modeling approach incorporating a feedback between a lateral inhibition circuit (left) and a mechanical 2D vertex model (right; see Materials and Methods for details). (B) A filmstrip of a simulation showing hopping intercalation of an HC. Asterisks mark an SC:SC defect being resolved by hopping intercalation (see Materials and Methods) (movie S7). (C) Comparison between circularity of IHCs at E15.5 and E17.5. (D) A filmstrip of a simulation combining lateral inhibition and hopping intercalations (see hybrid1 model in table S2) (movie S8). The white arrowhead shows an intercalating cell. Black arrowheads show out-of-row IHC defects. Snapshots shown are sections from a 25-cell-wide simulation window. (E to H) Quantitative comparison between a model with lateral inhibition only (L.I. only) and a hybrid model of lateral inhibition and hopping intercalations (hybrid 1 model; see Materials and Methods). Circularity of IHCs (E), number of out-of-row IHC defects (G), and number of SC:SC defects (H) are shown for the two models (average of n = 50 simulations). (F) shows relative change in apical area (area after/area before hopping) for hopping cells in the hybrid 1 model. For control, we measured the change in apical area of non-hopping IHCs in the simulation. Statistics: (C, E, and F) Welch’s t test; (G and H) Mann-Whitney test; all plots show means ± SEM. Repeats: (C) n = 5 independent movies for each time point, (E) n = 50, (F) n = 175 and 457 for hopping cells and neighbors, respectively, and (G and H) n = 50 simulations. ****P < 0.0001 and *P < 0.05.

Fig. 4.. The IHC pattern is refined by delaminations of low Atoh1⁺ cells.

(A) A filmstrip showing delaminations (arrowheads) of low-Atoh1⁺ cells (movie S9). (B) Comparison of normalized Atoh1 levels in delaminating Atoh1⁺ cells versus nondelaminating Atoh1⁺ cells. (C) Comparison of the percentage of IHCs that touch/do not touch the IPC row for delaminating and nondelaminating Atoh1⁺ cells. (D) Normalized Atoh1 level in HCs that touch the IPC row relative to HCs that do not touch the IPC row. (E) Simulation of the patterning process of IHCs that incorporates both hopping intercalation and delaminations of IHCs that do not touch the IPC row (hybrid 2 model) (movie S12). Arrowheads mark two out-of-row IHC defects that were resolved by the end of the simulation. Snapshots shown are sections from a 25-cell-wide simulation window. Statistics: (B and D) Mann-Whitney test; plots show means ± SEM. (C) χ² test. Repeats: (B) n = 15 delaminating cells and 150 neighbors, (C) n = 15, and (D) n = 43 and 106 for cells not touching the IPC row and cells touching, respectively. ****P < 0.0001. Scale bar, 5 μm.

Fig. 5.. Differential adhesion underlies straightening of the IHC:IPC boundary.

(A) Two images from E15.5 and E17.5 cochlear explants showing the increase in straightness of the IPC row. (B) Quantification of IPC alignment measured by the reduced χ² (see Materials and Methods). (C) Comparison of the mean length of IHC:IPC and SC:IPC contact at E15.5. (D) Comparison between hybrid models that include (hybrid 3) or do not include (hybrid 2) adhesion rules (movie S13). In the hybrid 3 model, tension (IPC:SC) = tension (IHC:IHC) > tension (IPC:IHC) > tension (IPC:IPC). Snapshots shown are sections from a 25-cell-wide simulation window. (E) Quantification of IPC alignment for the two models. (F) Ratio between the mean length of IHC:IPC and SC:IPC boundaries in the two models. (G) Filmstrip of a simulation from the hybrid 3 model, where IPC row is repaired (dark blue) after a single IPC was ablated (red “x”) (movie S14). (H) Fraction of simulations that resulted in repair of the IPC row in the two models. (I) A filmstrip from a laser ablation experiment performed at the base region of an E17.5 cochlea, where a single IPC was ablated (red “x”) (movie S16). (J) Fraction of IPC ablation experiments that resulted in the repair of the IPC row by neighboring IPC. Statistics: (B and F) Student’s t test, (C) Welch’s t test, and (E) Mann-Whitney test; plots show means ± SEM. (H and J) χ² test. Repeats: (B) n = 7 for E15.5 and n = 6 for E17.5; (C) 20 boundaries of each type from n = 4 movies, (E and F) n = 50, (H) n = 150, and (J) n = 17 mid and n = 12 base. ****P < 0.0001 and *P < 0.05. Scale bars, 10 μm (A) and 5 μm (I).

Fig. 6.. Nectin-3 accumulation at the IHC:IPC boundary correlates with onset of patterning.

(A) Antibody staining of an E15.5 cochlea with ɑ-Nectin-3 antibody. Nectin-3 becomes localized to the IHC:IPC boundary in the more developmentally advanced base region. (B) An image of a sub-apical plane in the mid region, showing a colocalized punctum of ZO1 with Nectin-3. The dashed line marks the IPC boundary, separating between the IHCs and OHCs. (C) Schematic of the localization of Nectin-1 and Nectin-3 in the apical and sub-apical domains. Scale bar, 5 μm.