Mechanical forces drive ordered patterning of hair cells in the mammalian inner ear

Abstract

Periodic organization of cells is required for the function of many organs and tissues. The development of such periodic patterns is typically associated with mechanisms based on intercellular signaling such as lateral inhibition and Turing patterning. Here we show that the transition from disordered to ordered checkerboard-like pattern of hair cells and supporting cells in the mammalian hearing organ, the organ of Corti, is likely based on mechanical forces rather than signaling events. Using time-lapse imaging of mouse cochlear explants, we show that hair cells rearrange gradually into a checkerboard-like pattern through a tissue-wide shear motion that coordinates intercalation and delamination events. Using mechanical models of the tissue, we show that global shear and local repulsion forces on hair cells are sufficient to drive the transition from disordered to ordered cellular pattern. Our findings suggest that mechanical forces drive ordered hair cell patterning in a process strikingly analogous to the process of shear-induced crystallization in polymer and granular physics.

Fig. 1. Hair cells gradually reorganize into a checkerboard-like pattern.

a Schematic of the organ of Corti. b Left: schematic of the mammalian inner ear. Right: Confocal image of the cochlea of a E17.5 mouse embryo marking the base, mid and apex regions. Cochleae are taken from transgenic mice expressing Math1-GFP (green) and are immunostained with α-ZO1 (red). Scale bar: 100 μm. c Representative images at different developmental times and different positions along the base-to-apex axis. Rows and columns correspond to different positions along the cochlear axis and different development times, respectively, as indicated. Scale bar: 10 μm. d Schematic of the definition of two order parameters: (i) The number of SC neighbors of each HC in OHC₂ (middle row) and (ii) hexagonal order parameter ${\psi }_6^{*}$ (bottom row). Yellow lines connecting HC centroids (orange dots) demonstrate higher hexagonal order at the base relative to the mid. ${\psi }_6^{*}$ values for each centroid cluster are as indicated. Scale bar: 5 μm. e–g Morphological and order parameters in different regions of the cochlea from apex to base (defined in inset) and at different developmental times (columns). Rows correspond to number of SCs neighbors (e), hexagonal order parameter ${\psi }_6^{*}$ (f), and ratio of HC to SC surface area (g). Local measures of order parameters associated with each HC are pooled by developmental age over n = 3,4,3 cochleae at E15.5, E17.5, P0, respectively, and then binned by cochlear position. Bars represent average on all local orders parameters within each bin. Error bars represent S.E.M. Schematic in a is modified with permission from Dror and Avraham.

Fig. 2. Shear motion and morphological transitions drive organization in the organ of Corti.

a, b A mid-apex region of a cochlear explant of a ZO1-EGFP mouse at E15.5. a Image showing the different regions along the medial-lateral axis of the Organ of Corti. Dotted line marks the pillar cell row. Dashed arrow separates the medial and lateral regions of the OHC region. b A filmstrip of the dashed square region in a, showing shear motion in the OHC region. A connecting line between an OHC (blue dot) and an IHC (gray dot) highlights the relative motion between the cells. The displacement of a HC (blue dot) and a SC (red square) is indicated by the distance from their initial positions (blue empty circle and red empty square, respectively). D_x, D_y are the total displacements in the x and y directions at the end of the movie compared to the initial position. Scale bars: 10 μm. Movie shown in Supplementary Video 1. c Displacement of apparent HCs and SCs from the movie shown in b. Displacements are calculated relative to the initial position of each cell. Cells from the medial (light red, light blue) and lateral (dark red, dark blue) OHC regions display different motion profiles. Shaded regions represent the boundaries of S.E.M. d, e Filmstrips showing d an intercalation process between two cell pairs (marked with red and blue dots), and e a delamination process of the cell marked with red dot. Bottom rows present segmented versions of the transitions. Movies shown in Supplementary Videos 3 and 4, respectively. f Rate of intercalations in the organ of Corti at E15.5 and E17.5. Gray dots correspond to individual data points obtained from n = 3 movies. Black dots and error bars represent average and S.E.M, respectively. Statistical analysis was done using two-sided two-sample t-test. g A filmstrip showing an event where a cell (blue dot) is “squeezed out” toward the top border of the OHC region. Bottom row presents a segmented version of the process. Movie shown in Supplementary Video 7. Scale bars for d, e, g: 5 μm. Observations in a, b, d, e, g were seen in all three repeats.

Fig. 3. A mechanical model based on global shear and local repulsion explains patterning of the OHCs.

a Schematic of the initial state of the model and the desired final state. The initial state for the simulation begins with the IHCs and pillar cells already pre-formed. A lateral inhibition mechanism is used to define a disordered pattern of HCs and SCs at a certain distance from the pillar cells. b Schematic of the three-dimensional structure of the organ of Corti showing the forces assumed in the model (see also image in Fig. S3c–e). HCs nuclei apply steric repulsion at the HC nuclei plane (yellow arrows) while lateral shear motion driven by movement of Hensen cells leads to shear and compression on HCs (red arrows). c Schematic of the key assumptions for the first stage of the simulation (compaction stage). HCs are subjected to global shear forces (red arrows) and local repulsion forces (yellow arrows). d A filmstrip of the first stage in the simulation. Global shear and local repulsion forces lead to the formation of a compact state of HCs. Movie shown in Supplementary Video 8. e Schematic of the key assumption for the second stage (refinement stage). The tension in SC:SC junctions is increased relative to that of HC:SC junctions. f A filmstrip of the second stage in the simulation. This stage begins at the end of the compaction stage. Movie shown in Supplementary Video 9. g, h Simulations capture the dynamics of the order parameters observed experimentally. Similar to the analysis in Fig. 1e–f, the number of SC neighbors decreases (g) and the HC organization exhibits higher hexagonal order (h). i Simulations capture the change in HCs and SCs areas. Gray dots correspond to individual data points of the ratio of HC to SC areas at the end of stage 1 and stage 2, obtained from n = 50 simulations. Black dots represent average. Statistical analysis was done using two-sided two-sample t-test. Error bars in g–i indicate S.E.M. Full description of the simulations is provided in the methods. Parameters used are provided in Supplementary Table 1.

Fig. 4. Laser ablation experiments show that SCs are more deformable than HCs.

a A filmstrip of a simulation modeling cell ablation (ablated cell marked with red “x”). Surrounding HCs (blue) and SCs (red) are marked. Movie shown in Supplementary Video 11. b Violin plot of the distributions of the relative change in area of HCs and SCs surrounding the ablated cells in the simulation. The presented data is taken from n = 825 simulations. c A filmstrip from a E17.5 ZO1-EGFP explant showing a laser ablation experiment. Ablation point is marked by red “x”. Bottom row shows a segmented version of the filmstrip with ablated area (gray), nearest neighbors HCs (blue), and nearest neighbor SCs (red) marked. Movie shown in Supplementary Video 12. Scale bar: 10 μm. d Normalized area of the cells shown in c. The area of each cell is normalized by its initial area before the ablation. e Violin plot of the distributions of the relative change in area of HCs and SCs, 10 min after ablation. The presented data is taken from n = 10 experiments. In the violin plots in b, e, median is indicated by white point, interquartile range (IQR) is indicated by thick line, Q3 + IQR and Q1 − IQR are indicated by the edges of the thin line and kernel density estimate of the data is shown in blue. In both plots, the p-value from two-tailed Kolmogorov–Smirnov test is as indicated.

Fig. 5. Model successfully predicts morphological changes induced by NMII inhibition.

a A filmstrip of a simulation modeling NMII inhibition. The addition of a NMII inhibitor, blebbistatin, was simulated by reducing and equating the tension for all junction types and overall contractility. Simulation starts at the end point of the simulations shown in Fig. 3. Movie shown in Supplementary Video 13. b Analysis of the ratio of HC area to SC area during the simulations. Model predicts a decrease in the ratio of HC to SC area. Data is averaged over n = 30 simulations. c A filmstrip from an E17.5 ZO1-EGFP explant treated with blebbistatin (10 µM). Blebbistatin was added 3 hours after the movie started. Movie shown in Supplementary Video 14. Scale bar: 10 µm. d Analysis of the ratio of HC area to SC area as a function of time demonstrated a decrease in the ratio of HC to SC areas in experiments where blebbistatin was added (red) compared to control experiments with no addition of blebbistatin (blue). Data is averaged over n = 3 experiments. Error bars in b, d indicate S.E.M.