Skip to main content
eLife logoLink to eLife
. 2021 Mar 5;10:e61092. doi: 10.7554/eLife.61092

Retrograde ERK activation waves drive base-to-apex multicellular flow in murine cochlear duct morphogenesis

Mamoru Ishii 1, Tomoko Tateya 2, Michiyuki Matsuda 1,3, Tsuyoshi Hirashima 1,4,5,
Editors: Doris K Wu6, Marianne E Bronner7
PMCID: PMC7935486  PMID: 33667159

Abstract

A notable example of spiral architecture in organs is the mammalian cochlear duct, where the morphology is critical for hearing function. Genetic studies have revealed necessary signaling molecules, but it remains unclear how cellular dynamics generate elongating, bending, and coiling of the cochlear duct. Here, we show that extracellular signal-regulated kinase (ERK) activation waves control collective cell migration during the murine cochlear duct development using deep tissue live-cell imaging, Förster resonance energy transfer (FRET)-based quantitation, and mathematical modeling. Long-term FRET imaging reveals that helical ERK activation propagates from the apex duct tip concomitant with the reverse multicellular flow on the lateral side of the developing cochlear duct, resulting in advection-based duct elongation. Moreover, model simulations, together with experiments, explain that the oscillatory wave trains of ERK activity and the cell flow are generated by mechanochemical feedback. Our findings propose a regulatory mechanism to coordinate the multicellular behaviors underlying the duct elongation during development.

Research organism: Mouse

Introduction

Spiral shapes are a widely occurring motif in many varied biological tissues and organisms, including shells, horns, and plants, but it has remained unclear how spiral shapes are formed spontaneously (Thompson, 1942). The general principle of spiral formation is differences in growth rate between the outer and inner tissues of the extending organ, with the growth rate of the outer tissue being faster than that of the inner one, which has been theoretically and experimentally demonstrated in shells and plants (Johnson et al., 2019; Raup and Michelson, 1965; Smyth, 2016; Wada and Matsumoto, 2018). The cellular processes causing this differential tissue growth are unique to each organ (Johnson et al., 2019; Saffer et al., 2017), and so identifying the organ-specific mechanisms underlying the differential tissue growth is crucial to understanding the developmental process of spiral morphogenesis.

An example of a spiral organ is the mammalian cochlear duct, which is a tonotopically organized auditory organ in the inner ear (Figure 1A). During murine development, the cochlear duct, composed of epithelial cells, elongates, bends, and coils to form a spiral. The molecular basis for the morphogenesis of the cochlear duct has been the subject of several previous studies. Gene knockout studies have clarified that the elongation of the cochlear duct requires sonic hedgehog (SHH) signaling from the cochleovestibular ganglion in the conical central axis of the cochlea (Bok et al., 2013; Liu et al., 2010; Tateya et al., 2013), fibroblast growth factor (FGF) signaling of epithelial cells (Pauley et al., 2003; Pirvola et al., 2000; Urness et al., 2018; Urness et al., 2015), and non-canonical Wnt–planar cell polarity (PCP) signaling of prosensory cells (Mao et al., 2011; Montcouquiol and Kelley, 2020; Qian et al., 2007; Saburi et al., 2008; Wang et al., 2005). Deletion of Shh expression leads to a shortening of the cochlear duct, and it was proposed that the SHH signaling promotes growth of the cochlear duct mainly in the base and middle regions (Bok et al., 2013). In Fgf10 null mutant mice, the cochlear duct is remarkably shorter, but cell proliferation is unaffected (Urness et al., 2015), suggesting that cell proliferation and other cellular processes regulate ductal outgrowth. From embryonic day (E) 14.5 onward, the mediolateral active migration of prosensory cells, during which these cells intercalate radially with their neighbors (known as convergent extension), contributes to longitudinal duct extension in a Wnt–PCP pathway-dependent manner (Chen et al., 2002; Cohen et al., 2020; Driver et al., 2017; Yamamoto et al., 2009). This cell intercalation drives ductal elongation; however, it simply cannot explain the duct bending before E14.5 without an asymmetric mode of cell intercalation (Sato et al., 2015). Although the underlying signaling pathways are well characterized, the physical cellular mechanisms underlying spiral morphogenesis of the cochlear duct remain elusive. In the present study, we aimed to identify the multicellular dynamics giving rise to elongating, bending, and coiling of the developing cochlear duct using a combination of live-cell imaging, Förster resonance energy transfer (FRET) quantitation, and mathematical modeling.

Figure 1. Quantification of morphology and cell proliferation in developing murine cochlear duct.

(A) Schematic diagrams showing the tissue axis and labels of the cochlear duct. (B) Immunofluorescence images of anti-E-cadherin staining in the murine developing cochlea from E12.5 to E14.5. The lower rows are magnified images of the dotted squares in the upper rows. Yellow dotted lines represent the edges of the epithelial layer. Scale bar, 200 µm. (C) Schematic diagram showing regions used for morphological quantification. (D) Curvature and thickness as a function of the arc length from the apex tip along the lateral epithelial layer from E12.5 to E14.5. Mean ± standard deviation (s.d.) N = 3. (E) Maximum projection images of stained anti-E-cadherin (white) and EdU (magenta) in the roof and floor region of cochlear duct at E12.5. Representative images showing EdU signals in the cochlear duct are shown. Arrows indicate the region of EdU signal gradient from the medial to the lateral side of the duct. Arrowheads indicate the base region where the EdU intensity is concentrated. Scale bar, 100 µm. (F) Heatmap of the sample-mean of maximum EdU intensity projection in the roof and floor region of the duct. Arrows and arrowheads are the same as in (E). N = 3. Heatmaps of summed and mean intensity projection are shown in Figure 1—figure supplement 1. (G) Cell density distribution along the mediolateral axis in the floor and roof region, respectively. Color indicates the different samples. N = 3.

Figure 1—source data 1. Curvature and thickness from E12.5 to E14.5.
Figure 1—source data 2. EdU intensity profiles.
Figure 1—source data 3. Cell density along the medial–lateral axis.

Figure 1.

Figure 1—figure supplement 1. Heatmaps of EdU (EdU: 5-ethynyl-2′-deoxyuridine, DMSO: dimethyl sulfoxide, PFA: paraformaldehyde, PBS: phosphate-buffered saline) intensity.

Figure 1—figure supplement 1.

Heatmaps of EdU intensity normalized to the maximum value of each graph using various projection methods: maximum intensity projection (left); summed intensity projection (center); and mean intensity projection (right). Note that, in the mean intensity projection, the EdU intensity is subject to increase toward the edge of the cochlear duct because the thickness of the epithelial layer along the projected axis is lesser in the edge than in the middle of the cochlear duct.

Results

Cell proliferation profile suggests cellular inflow to the lateral side of the apex for elongation of bending duct

We first examined the morphology of the developing cochlear duct from E12.5 to E14.5 by staining an epithelial marker, E-cadherin, followed by organ-scale 3D imaging. During this developmental period, the cochlear duct elongates and coils without changes to the mediolateral width at a horizontal section of the roof–floor axis (Figure 1B). The curvature and thickness of the epithelial layer was quantified on the lateral side of the cochlear duct (Figure 1C). We found that they remained almost constant and small variation along the arc length from the apex tip (Figure 1D, Figure 1—source data 1), suggesting a cellular mechanism for the developing duct to elongate while maintaining its curvature.

Next, we examined the spatial distribution of proliferating cells in the duct as it directly contributes to the local tissue growth. The spatial distribution of nuclei labeled with EdU for 30 min was measured on the roof and floor side of the cochlear duct at E12.5 (Figure 1E). For quantification, the image domain of the cochlear duct was divided into interrogation regions and the averaged fluorescence intensity of labeled EdU was measured within each region. On the floor side, EdU-positive cells were more abundant in the medial side than in the lateral side around the apex (arrows, Figure 1E, F, left, Figure 1—source data 2). However, on the roof side of the cochlear duct, EdU-positive cells were distributed evenly, with slightly fewer cells at the base than at the apex and without a significant bias along the mediolateral axis (Figure 1E, F, right, Figure 1—source data 2). Moreover, nuclear density showed no significant difference along the mediolateral axis in either the floor or the roof side (Figure 1G, Figure 1—source data 3). These observations do not strongly support the possibility that cell proliferation rates on the lateral side of the cochlear duct could drive mediolateral differential tissue growth and cause duct bending. Of note, the spatial map of EdU intensity shows that cell proliferation rates were higher in the floor–base region (arrowheads, Figure 1E, F, left). Supposing that cell proliferation is the main driver of local tissue growth, the higher volumetric growth observed in the medial side than in the lateral side would contribute to the duct bending inward at the lateral side, which contradicts the innate cochlear morphogenesis. We thus hypothesized that the cells in the lateral side of the growing apex may be supplied by the proliferation ‘hot spot’ in the base region of the cochlear duct, which could resolve the observed mismatch between tissue growth rates in the medial and lateral sides of the duct.

ERK inactivation resulted in cochlear duct shortening

We then investigated how the cells are supplied from the base to the apex of the cochlear duct. Earlier studies reported that FGF signaling is critical for cochlear duct outgrowth (Pirvola et al., 2000; Urness et al., 2015). Therefore, we focused on extracellular signal-regulated kinase (ERK)/MAP kinase, a downstream kinase in the FGF signaling pathway, and investigated the impact of FGFR–ERK signaling axis on the cochlear duct morphogenesis.

The dissected cochlea at E12.5 were cultured ex vivo for 2 days by treating with either PD0325901, a specific ERK kinase (MEK) inhibitor (Barrett et al., 2008), or SU5402, a multitargeted receptor tyrosine kinase inhibitor including FGFR (Sun et al., 1999). Both inhibitor treatments resulted in obvious impairments of the cochlear duct growth together with the reduction of EdU signals in the entire cochlear duct, indicating that the FGF–ERK signaling promotes the cell proliferation (Figure 2A–A’’). The arc length of the cochlear duct on the lateral side from the apex tip to the adjacent point of the saccule was significantly shorter both in the PD0325901 and SU5402 treatments than in the control (Figure 2B, Figure 2—source data 1). These results suggest that the cell proliferation would contribute to the duct elongation. Moreover, we found that those inhibitor treatments remarkably altered the curvature of the cochlear duct. As the size of cochlear duct was different in each treatment, the curvature profile along the arc length was normalized by the total arc length (Figure 2C, Figure 2—source data 2). To assess the curvature over the lateral edge of cochlear duct, we calculated total curvature over the normalized arc length, s, and found that the total curvature was larger both in the PD0325901 and SU5402 treatments than in the control (Figure 2D). Provided that cell proliferation is the only driving factor for the cochlear duct bending, the curvature profile should be almost constant along the lateral edge as observed at E12.5 (Figure 1D). However, the inhibitor treatments for the FGFR–ERK signaling resulted in the clear curvature gradient with larger curvature in the apex region than in the base region (Figure 2C). Therefore, the growth impairment of the cochlear duct by inhibition of FGFR–ERK signals would be caused not only due to the cell proliferation but also other cellular processes, and we suspected that the FGF–ERK signaling would play a role in the cell supply from the base to the growing apex region in the cochlear duct.

Figure 2. Pharmacological inhibition of FGFR–ERK signaling axis resulted in the impairment of cochlear duct growth.

Figure 2.

(AA’’) Maximum projection images of EdU labeling (magenta) with DAPI (white) nuclear counterstaining over the entire cochleae after 2 days of ex vivo culture from E12.5 in the treatment with DMSO (A), PD0325901 at 1 µM (A’), and SU5402 at 30 µM (A’’). Scale bar, 100 µm. (B) The arc length of the cochlear duct in the treatment with DMSO (red), PD0325901 (blue), and SU5402 (green). N = 3. Two-sample t-test without assuming equal population variances. p=0.023 for DMSO-PD0325901 and p<0.001 for DMSO-SU5402. (C) The curvature of the cochlear duct along the normalized arc length s. Mean ± s.d. N = 3. (D) Total curvature over the normalized arc length in the treatment with DMSO (red), PD0325901 (blue), and SU5402 (green). Two-sample t-test without assuming equal population variances. p=0.021 for DMSO-PD0325901 and p<0.0033 for DMSO-SU5402. ERK: extracellular signal-regulated kinase.

Figure 2—source data 1. Longitudinal length for treatments with DMSO, PD0325901, and SU5402.
Figure 2—source data 2. Curvature over arc length from the tip for treatments with DMSO, PD0325901, and SU5402.

Retrograde helical ERK activation waves drive base-to-apex multicellular flow

To examine the spatiotemporal ERK activity on the developing cochlea, we used a reporter mouse line that ubiquitously expresses a FRET-based biosensor for ERK activity in the cytosol (Harvey et al., 2008; Komatsu et al., 2018; Komatsu et al., 2011). 3D FRET imaging using two-photon microscopy revealed that ERK is preferentially activated in the lateral-roof side of the cochlear duct, including the outer sulcus and stria vascularis (Figure 3A, A’, Figure 3—video 1), which is consistent with the previously reported distribution of Fgfr2 expression (Urness et al., 2015).

Figure 3. Retrograde helical extracellular signal-regulated kinase (ERK) activation waves drive base-to-apex multicellular flow.

(A) 3D ERK activity map in the cochlear duct cultured ex vivo for 1 day from E12.5. (A’) Cross-sectional view (medial–lateral and roof–floor plane) of (A). Orange dotted line indicates the floor plane shown in (B) and (C). Scale bar, 50 µm. (B) Time-lapse snapshots of ERK activity maps in the floor plane. Time indicates the elapsed time of live imaging. Yellow arrowheads indicate the ERK activity peak. Scale bar, 100 µm. (C) Time-lapse snapshots of tissue flow speed obtained by particle image velocimetry in the floor plane. Scale bar, 100 µm. (D) Schematic diagram showing the axis, the apex–base line for kymography, and regions of interests (ROIs). (E) Representative kymograph of ERK activity. The horizontal axis indicates the position on the apex–base line shown in (D), and the vertical axis indicates the elapsed time of live imaging. Dotted lines represent oscillatory waves from the apex to the base. Scale bar, 100 µm. (F) ERK wave speed with mean and s.d. n = 5 from N = 3. (G) Time-series ERK activity rate and extension-shrinkage rate in representative three different ROIs. (H) Cross-correlation between the extension-shrinkage rate and ERK activity rate. n = 12. Mean ± s.d. (I, J) Tissue flow speed before and after the PD0325901 treatment at 1 µM, the SU5402 treatment at 30 µM (I), and the cyclopamine treatment at 30 µM (J). n = 285. Confirmed by N = 2. Two-sample t-test, p<0.001. (K) Time-lapse snapshots of surface-rendered ERK activity maps in the cochlear duct at E12.5. The green corners correspond to the green corner on the images shown in (B) and viewed from the left-bottom corner of (B). Circles indicate the position of ERK activity peaks, and the connecting dotted lines indicate a trace of the peak shift. The timescale is the same as in (B). (L) Schematics for the ERK activity waves and cell flow.

Figure 3—source data 1. Extracellular signal-regulated kinase activity rate and extension-shrinkage rate.
Figure 3—source data 2. Particle image velocimetry speed before and after the treatments with PD0325901, SU5402, and cyclopamine.

Figure 3.

Figure 3—figure supplement 1. Extracellular signal-regulated kinase (ERK) activity waves and cell flows.

Figure 3—figure supplement 1.

(A) Roof plane (orange dotted line) in the cross-sectional view of 3D ERK activity map in the cochlea at E12.5. Scale bar, 50 µm. (B) Time-lapse snapshots of ERK activity map in the roof plane. Time indicates the elapsed time of live imaging. Scale bar, 100 µm. (C) Time-lapse snapshots of tissue flow speed obtained by the particle image velocimetry in the roof plane. The multicellular flows direct toward the elongation direction around the apex tip despite winding in the region away from the tip in the roof side. Scale bar, 100 µm. (D) Kymograph of ERK activity for the two samples. The horizontal axis indicates the position on the apex–base line, and the vertical axis indicates the elapsed time of live imaging. Dotted lines represent the oscillatory wave trains from the apex to the base. Scale bar, 100 µm. (E) Time-series data of ERK activity and the tissue flow speed for elongation in the three different regions of interests. (F, F’) ERK activity before and after the MEK inhibitor PD0325901 treatment at 1 µM. Scale bar, 20 µm.
Figure 3—video 1. 3D extracellular signal-regulated kinase (ERK) activity map of the cochlear at E12.5, related to Figure 3.
Download video file (1.2MB, mp4)
Color indicates the ERK activity level shown in Figure 3A. Views from the roof side at initial and ones from the apex side at last.
Figure 3—video 2. Time-lapse movie of the extracellular signal-regulated kinase (ERK) activity and multicellular tissue flow, related to Figure 3.
Download video file (7.5MB, mp4)
Color in the left panel indicates the ERK activity level shown in Figure 3B and that in the right panel indicates the particle image velocimetry speed shown in Figure 3C. Views at the floor plane indicated in Figure 3A’. The time interval is 12 min.
Figure 3—video 3. Time-lapse movie of extracellular signal-regulated kinase (ERK) activity upon treatment with cyclopamine, related to Figure 3.
Download video file (6.3MB, mp4)
Exactly 30 µM of cyclopamine was added immediately before time zero. Color indicates the ERK activity level. Views at the floor plane indicated in Figure 3A’. The time interval is 12 min.
Figure 3—video 4. 3D time-lapse imaging of extracellular signal-regulated kinase (ERK) activity in the lateral side of the cochlear duct at E12.5, related to Figure 3.
Download video file (4.3MB, mp4)
Color indicates the ERK activity level shown in Figure 3K. Views from the lower-left corner of Figure 3B. Green arrow indicates the ERK activation peak.

For continuous observation during ductal outgrowth, we established an explant culture method in which the capsule above the apex tip was partially removed, allowing 3D organ-scale long-term imaging of ERK activity. Surprisingly, the time-lapse images of the cochlea dissected at E12.5 revealed that ERK activation propagates intercellularly as oscillatory waves from the apex to the base of the floor side (Figure 3B, Figure 3—video 2), while ERK is constitutively activated around the apex tip of the roof side (Figure 3—figure supplement 1A, B). We next quantified multicellular tissue flow by particle image velocimetry (PIV) at the supracellular (4–5 cell length) scale and found that cells coherently move as clusters of ~100 µm diameter from the base to the apex of the floor side, again as oscillatory waves, and similarly to the ERK activation waves of the floor side (Figure 3C, Figure 3—video 2). In the roof side, the multicellular tissue flows directly toward the direction of elongation around the apex tip (Figure 3—figure supplement 1C). Kymography of ERK activity along the apex–base line of the lateral-floor side (Figure 3D) shows oscillatory retrograde ERK activity waves (Figure 3E, Figure 3—figure supplement 1D), which proceed at a speed of 0.42 ± 0.078 µm min−1 (mean ± standard deviation) in space-fixed coordinates (Figure 3F).

Since ERK activation can be induced by cell extension – an increase in the projected area along the cellular apicobasal axis – during collective cell migration (Hino et al., 2020), we then calculated the ERK activity rate (time derivative of ERK activity) and the extension-shrinkage rate, that is, the local tissue strain rate along the apex–base line, by using time-series data of ERK activity and PIV speed, respectively (Figure 3—figure supplement 1E). We found that both the ERK activity rate and the extension-shrinkage rate oscillate across the time course (Figure 3G, Figure 3—source data 1). Moreover, cross-correlation analysis revealed that the local tissue deformation, represented by the extension-shrinkage rate, precedes the ERK activity rate by 24 min on average (Figure 3H).

The role played by ERK was confirmed via inhibitor assays at E12.5. Treatment with either PD0325901 or SU5402 resulted in the significant decrease of the tissue flow speed, as well as ERK inactivation (Figure 3—figure supplement 1F, F’), that is, the median speed decreased before and after the administration by 42% and 48%, respectively (Figure 3I, Figure 3—source data 2). These results corroborated that the FGFR–ERK signaling axis contributed to the base-to-apex multicellular tissue flow. On the other hand, we found that cyclopamine treatment, an inhibitor for the Shh signaling pathway, resulted in almost no change in the ERK activity (Figure 3—figure supplement 1F, F’, Figure 3—video 3) and affected the cell flow speed (16% reduction) to a lesser extent compared with PD0325901 and SU5402 treatment within 3 hr after administration (Figure 3J, Figure 3—source data 2). This suggested that ERK activity was not the only factor but it played a major role in regulating cell migration. Interestingly, we found that the cell flow was decreased by 62% along with ERK inactivation 8 hr after the administration (Figure 3J, Figure 3—source data 2, Figure 3—video 3). This long-term effect both on the cell flow and ERK activity might be caused by the lack of cell supply from the base to the apex due to the suppression of the Shh-induced cell proliferation.

Finally, we extended the analysis to 3D dynamics of ERK activity and cell movement in the developing cochlear duct. Surface rendering of the ERK activity map in the cytosol indicated that ERK activity peaks shift from the apex–roof to the base–floor in the lateral side of the cochlear duct (Figure 3K, Figure 3—video 4). Concomitantly with helical ERK activity waves, coherent cell movements can be observed from the base–floor to the apex–roof in the opposite direction to the ERK waves (Figure 3—video 4). This observation suggests that ERK-mediated helical collective cell movement on the lateral side could drive 3D duct coiling underlying the spiral morphogenesis of the cochlear duct (Figure 3L).

ERK-mediated mechanochemical feedback explains cell flow and ERK waves

The cross-correlation analysis and ERK inactivation assay suggested a regulatory regime of coupling between the ERK activation and cell migration; the extension-triggered ERK activation promotes cell contraction and pulling the neighboring cells, which eventually evokes transmission of ERK activation to the neighboring cells, that is, mechanochemical feedback (Boocock et al., 2021; Hino et al., 2020). To further verify this regime underlying the developing cochlear duct, we first suppressed actomyosin cell contraction by treating the E12.5 cochlea with blebbistatin. As anticipated, the cochlear duct was extended immediately after blebbistatin treatment and ERK was activated (Figure 4A, Figure 4—video 1). Moreover, the cell flow was significantly decreased by 79% (Figure 4B, Figure 4—source data 1), indicating that the active cellular contraction was required for the cell flow.

Figure 4. Extracellular signal-regulated kinase (ERK)-mediated mechanochemical feedback explains cell flow and ERK waves.

(A) ERK activity maps in the floor plane before (−12 min) and after (+36 min) the treatment with 30 µM of blebbistatin. Time indicates the timing of blebbistatin treatment. Scale bar, 100 µm. (B) Tissue flow speed before (−3 hr) and after (+3 hr) blebbistatin treatment. N = 258. Two-sample t-test, p<0.001. (C) Snapshot of the time-lapsed ERK activity map on the lateral side of the cochlear duct (green dotted line), showing coexistence of the ERK activity peak and the tissue curvature peak (green arrowhead). Scale bar, 50 µm. (D) Time-series ERK activity and the tissue curvature in the representative regions of interest. (E) Cross-correlation between the tissue curvature and ERK activity. n = 9. Mean ± s.d. (F) Schematics for the model mechanochemical coupling, in which ERK activity and cell deformation are reciprocally regulated. (G) Kymograph of the ERK activity in the model simulation. Scale bar, 100 µm. (H) Simulated time series of ERK activity rate and extension-shrinkage rate in the mechanochemical coupling regime. (I) Cross-correlation between the extension-shrinkage rate and the ERK activity rate. The lag time is −28 min. (J) Schematics for a counterpart regime of the mechanochemical coupling, in which the ERK activity unidirectionally regulates the cell deformation without closed feedback. (K) Kymograph of the ERK activity in the uncoupling model simulation. Scale bar, 100 µm. (L) Simulated time series of ERK activity rate and extension-shrinkage rate in the uncoupling regime. (D) Cross-correlation between the extension-shrinkage rate and the ERK activity rate. The lag time is −2 min.

Figure 4—source data 1. Particle image velocimetry speed before and after the treatments with blebbistatin.
Figure 4—source data 2. Extracellular signal-regulated kinase activity and tissue curvature.

Figure 4.

Figure 4—figure supplement 1. Multicellular tracking of the time-lapse images.

Figure 4—figure supplement 1.

A representative trace line for seven neighboring cells around the future prosensory region at E12.5 on the apex–base and medial–lateral plane (left), and the apex–base and luminal–basal plane (right). Color represents different cells. The trace lines do not intersect each other within 3 hr. Scale bar, 10 µm.
Figure 4—video 1. Time-lapse movie of extracellular signal-regulated kinase (ERK) activity upon treatment with blebbistatin, related to Figure 4.
Download video file (2MB, mp4)
Exactly 30 µM of blebbistatin was added immediately before time zero. Color indicates the ERK activity shown in Figure 4A. Views at the floor plane indicated in Figure 3A’. The time interval is 12 min.

Next, we noticed local deformation of the cochlear duct as a possible consequence of cellular contraction. Our live imaging system from E12.5 exhibited that the epithelium of the cochlear duct became concave locally in the high ERK activity region (Figure 4C), suggesting that the ERK activation promoted contraction. We then calculated the tissue curvature of the ductal lateral edge using time-lapse images and found that both the ERK activity and the tissue curvature similarly oscillated over the time (Figure 4D, Figure 4—source data 2). Cross-correlation analysis revealed that the curvature change was delayed by 24 min on average due to the changing ERK activity (Figure 4E).

Finally, we confirmed the plausibility of the mechanochemical coupling via mathematical modeling (Figure 4F). Our minimal mathematical model of the mechanochemical coupling reproduced multiple ERK activity propagations well (Figure 4G) as observed experimentally (Figure 3E). The mechanochemical model produced a 28 min lag from the extension-shrinkage rate to the ERK activity rate (Figure 4H, I). In a counterpart uncoupling model (Figure 4J), however, it was a 2 min lag under which the ERK activation waves regulated cell deformation unidirectionally (Figure 4K–M). Together, the model analysis supported the plausibility of mechanochemical coupling rather than uncoupling regulation.

Discussion

Previous genetic studies have revealed the molecular basis of cochlear duct elongation during development, but have been unable to explain the physical mechanisms by which the duct bends because of its severe phenotype (Bok et al., 2013; Groves and Fekete, 2012; Urness et al., 2018; Urness et al., 2015). Motivated by these earlier studies, we have visualized the cochlear duct development by two-photon microscopy under ex vivo culture condition. We found that the coherent multicellular flow occurs from the base to the apex exclusively on the lateral side of the growing cochlear duct and also elucidated that the multicellular flow was accompanied by retrograde ERK activation waves (Figure 3L). Thus, our long-term deep tissue imaging has illuminated unprecedented dynamics of cells and kinase activity underpinning the bending of developing cochlear duct.

In the present study, the establishment of long-term imaging techniques and biosensors for protein kinase activity has led to the discovery of unexpected spatiotemporal patterns of cell movement and ERK activity in the developing cochlear duct. Previously, we and others have observed intercellular ERK activation waves in the epithelium, such as migrating Madin–Darby canine kidney (MDCK) cells (Aoki et al., 2017; Hino et al., 2020), developing Drosophila tracheal placode (Ogura et al., 2018), and wounded murine skin (Hiratsuka et al., 2015). We have also proposed an ERK-mediated mechanochemical feedback system, in which cell extension activates ERK followed by ERK-triggered cell contraction (Boocock et al., 2021; Hino et al., 2020) that can explain the coordination between cell movement and ERK activity. The ERK activation wave speed in the developing cochlear duct was 0.42 µm min−1 (Figure 3F), which is slower than in MDCK cells and wounded mouse epidermis, where it proceeds at 2.5 µm min−1 and 1.4 µm min−1, respectively (Aoki et al., 2017; Hino et al., 2020; Hiratsuka et al., 2015). Interestingly, when normalized to the cell lengths of the developing cochlear duct (4 µm), MDCK cells (20 µm), and the basal cells of the mouse epidermis (10 µm), the wave speed of the developing murine cochlear duct, 6 cells diameter h−1, is comparable with that of MDCK cells, 7 cells diameter h−1, and that of wounded adult murine skin, 8 cells diameter h−1. Hence, our findings on the coupling between the multicellular flows and the ERK activation wave trains in the cochlear duct, together with the earlier studies (Boocock et al., 2021; Hino et al., 2020), support the existence of a general regulatory mechanism for the collective cell migration during tissue morphogenesis.

Recently, much slower intercellular ERK activation wave propagation was found in the regeneration of zebrafish scales (De Simone et al., 2021); the ERK activation wave speed is about at 0.17 µm min−1, that is, 1 cell diameter h−1 with the average osteoblast size (10 µm). It is proposed that these slow ERK waves were generated by a reaction–diffusion system including diffusible ligand-based activators and inhibitory regulators, such as dual-specificity phosphatases and sprout proteins. Of note, the reactions in this system consist of the ERK-mediated transcriptional and translational processes that characterize the slower timescale of the dynamics in the cells and the ERK activity compared with the other reported phenomena. Moreover, the diffusion of an Fgf ligand as hypothesized for scale regeneration in zebrafish was not considered in the present study since the mechanochemical cell-to-cell communications, which transmit the signals across the cells via diffusion of mechanical stress, can sufficiently explain the ERK activation waves for the cochlear duct development. As the ligand diffusion-based signal transmission is directly affected by 3D tissue geometry, especially in the curved epithelial tissues, the mechanical signal transmission would achieve a more robust regulatory system underlying the morphogenesis of the spiral cochlear duct.

Our 3D time-lapse imaging revealed coherent helical cell flow from the base–floor to the apex–roof in the lateral side of the cochlear duct. Cell flow analysis revealed that the rate of base-to-apex cell flow (0.24 µm min−1, Figure 3—figure supplement 1E) exceeds that of the duct elongation speed (0.13 µm min−1, Figure 3—video 2). Thus, the cell flow rate may be sufficient to compensate for the lateral tissue growth. We speculate that this ERK-mediated cell advection originating from the heterogeneity of cell proliferation causes consistent mediolateral differential growth at the tissue scale and results in cochlear bending. In support of this, knockout of Shh causes a significant decrease in the number of proliferating cells at the base of the cochlear duct and shortens the cochlear duct (Bok et al., 2013). Moreover, we showed that the impact of cyclopamine treatment on the cell flow became less evident immediately after its administration (Figure 3J), although the slight reduction of cell speed implied that the Shh signaling could be involved in controlling machineries for cell migration. As SHH is secreted mainly from the spiral ganglions located in the central axis of the cochlea (Bok et al., 2013), further investigations on the interaction between the cochlear duct and these ganglions will be helpful to clarify the overall cell flow in the elongating duct.

We reported that the cell flow depended on myosin activity (Figure 4B) and caused the local deformation of the basal edge (Figure 4C). Therefore, the cells generated the mechanical forces and actively migrated to coordinate the multicellular behavior at supracellular scale in the lateral side of the cochlear duct. The epithelium in the lateral side is simple cuboidal rather than pseudostratified observed in the floor and medial side, thereby the epithelial cells would attach to the basement membrane and load forces onto the basement membrane via the subcellular structures, such as cryptic lamellipodia and cellular protrusions as observed in the floor side (Driver et al., 2017). However, it was unfeasible to identify the single-cell behavior due to mainly our fluorescence labeling for cytosol in this study. We also attempted to detect the F-actin cortex on the inner face of the cell membrane using Lifeact-EGFP transgenic mice (Riedl et al., 2010), but we were not able to recognize the cell membrane dynamics clearly enough owing to its faint signals in the deep region. This problem will be solved by further improvements in fluorescence markers, spatiotemporal high-resolution microscopy, and organ culture methods.

Although we focused on the cell flow on the lateral side, there are other cellular events that can potentially underlie the duct morphogenesis. For example, the cartilaginous capsule may play an essential role in cochlear morphogenesis, especially as a physical restriction to avoid outward duct growth. Without volumetric growth of the capsule, the cochlear duct would not be able to grow due to a lack of appropriate space. Moreover, complete removal of the capsule led to failure in the elongating and bending of the cochlear duct cultured under the ex vivo condition in our observations. Therefore, the detailed balance of overall tissue growth between the duct epithelium and the capsule will be essential and needs to be elucidated. Another potential but less likely mechanism is the active cell intercalation that occurs from E14.5 onward as demonstrated previously (Chen et al., 2002; Cohen et al., 2020; Driver et al., 2017; Yamamoto et al., 2009). Depending on our careful observation and manual cell tracing of time-lapsed imaging data from E12.5 to E14.5, there was no clear mediolateral intercalation (Figure 4—figure supplement 1). However, the mediolateral cell intercalation might contribute to the maintenance of the cochlear duct width through polarized cellular mechanoresponse as proposed in an elongating epithelial duct (Hirashima and Adachi, 2019). As our live imaging data were insufficient for automatic 3D cell tracking due to a lack of organelle-specific marker and weak fluorescence signals, further improvement of imaging systems and single-cell tracking will clarify these aspects of multicellular complexity.

Overall, we visualized multicellular behavior underlying the elongating and bending of the cochlear duct during development using deep tissue live imaging. This contributes to a better understanding of symmetry breaking in tissue morphogenesis during development and in the generation of inner ear organoids (Koehler et al., 2017; Koehler et al., 2013). The live imaging technique used in the present study forms the basis for further analysis of the interplay between morphogenesis and cell fate decisions during cochlear development (Cohen et al., 2020; Tateya et al., 2019).

Materials and methods

Experiments

Animals

For FRET imaging, we used the transgenic mice that ubiquitously express an ERK biosensor with a long flexible linker (hyBRET-ERK-NLS) reported elsewhere (Harvey et al., 2008; Komatsu et al., 2018; Komatsu et al., 2011). Otherwise, we used ICR mice purchased from Japan SLC, Inc. We designated the midnight preceding the plug as embryonic day 0.0 (E0.0), and all mice were sacrificed by cervical dislocation to minimize suffering. All the animal experiments were approved by the local ethical committee for animal experimentation (MedKyo 19090 and 20081) and were performed in compliance with the guide for the care and use of laboratory animals at Kyoto University.

Antibodies

The following primary and secondary antibodies were used for immunofluorescence: anti-E-cadherin rat antibody (Thermo Fisher Scientific, #13-1900, 1:100 dilution) and Alexa Fluor 546-conjugated goat anti-rat IgG (H+L) antibody (Thermo Fisher Scientific, #A11081, 1:1000 dilution).

Small-molecule inhibitors

The following chemicals were used: blebbistatin (FUJIFILM Wako Pure Chemical Corporation, #021-17041), cyclopamine (FUJIFILM Wako Pure Chemical Corporation, #038-19311), SU5402 (FUJIFILM Wako Pure Chemical Corporation, #197-16731), and PD0325901 (FUJIFILM Wako Pure Chemical Corporation, #162-25291).

Whole-tissue staining and imaging

The cochleae were gently freed from the capsule, and the staining and clearing were performed according to an earlier study (Hirashima and Adachi, 2015). Briefly, the samples were fixed with 4% PFA in PBS overnight at 4°C and then blocked by incubation in 10% normal goat serum (Abcam, #ab156046) diluted in 0.1% Triton X-100/PBS (PBT) for 3 hr at 37°C. The samples were treated with primary antibodies overnight at 4°C, washed in 0.1% PBT, and subsequently treated with secondary antibodies conjugated to either Alexa Fluor 546 or Alexa Fluor 647 overnight at 4°C. For counter staining of nucleus, we used DAPI (Dojindo Molecular Technologies, #D523-10, 1:200 dilution). The samples were mounted with 10 µL of 1% agarose gel onto a glass-based dish (Greiner Bio-One, #627871) for stable imaging. Then, the samples were immersed with the CUBIC-R+ (Tokyo Chemical Industry Co., #T3741) solution for optical clearing. Images were acquired using the confocal laser scanning platform Leica TCS SP8 equipped with the hybrid detector Leica HyD with the ×40 objective lens (NA = 1.3, WD = 240 μm, HC PL APO CS2, Leica) and the ×20 objective lens (NA = 0.75, WD = 680 µm, HC PL APO CS2, Leica) and the Olympus FluoView FV1000 with the ×30 objective lens (NA = 1.05, WD = 0.8 mm, UPLSAPO30XS, Olympus).

EdU assay

For EdU incorporation to embryos, 200 µL of 5 mg/mL EdU in PBS was intraperitoneally injected to pregnant mice 30 min prior to dissection. For the incorporation to dissected cochleae, 10 µM of EdU was treated into the samples 1 hr prior to the chemical fixation. Before EdU detection, whole-tissue immunofluorescence of E-cadherin and counter nuclei staining with DAPI were performed. EdU was detected using the Click-iT EdU Imaging Kits (Thermo Fisher Scientific, #C10340). The samples were optically cleared with CUBIC-R+, and images were acquired by confocal microscopy as described above.

Explant cultures

We cultured the dissected cochleae without removing the capsule unless otherwise noted. The cochleae were mounted on a 35-mm glass-based dish (Iwaki, #3910-035) with 1 μL of growth factor reduced Matrigel (Corning, #356231), and filled with 2 mL of a culture medium including FluoroBrite DMEM Media (Thermo Fisher Scientific, #A1896701) with 1% GlutaMAX (Thermo Fisher Scientific, #35050061) and 1% N2 Supplement with Transferrin (Holo) (FUJIFILM Wako Pure Chemical Corporation, #141-08941). The samples were incubated at 37°C under 5% CO2.

Live imaging for explants

For long-term organ-scale imaging, we partially cut off the capsule adjacent to the apex tip of cochlear duct using tweezers carefully and the semicircular canals were removed. The isolated cochlea was put onto the dish as described above. For microscopy, we used an incubator-integrated multiphoton fluorescence microscope system (LCV-MPE, Olympus) with a × 25 water-immersion lens (NA = 1.05, WD = 2 mm, XLPLN25XWMP2, Olympus). The excitation wavelengths were set to 840 nm for CFP (InSight DeepSee, Spectra-Physics). Imaging conditions for the FRET biosensor were as follows: scan size: 800 × 800 pixels; scan speed: 10 μs/pixel; IR cut filter: RDM690 (Olympus); dichroic mirrors: DM505 and DM570 (Olympus); and emission filters: BA460-500 for CFP and BA520-560 for FRET detection (Olympus).

Quantification and analysis

FRET image analysis

The median filter of 3 × 3 window was processed to remove shot noises, and the background signal was subtracted each in the FRET channel and the CFP channel. Then, the ratio of the FRET intensity to the CFP intensity was calculated by a custom-made MATLAB (MathWorks) script. In the scale bar, color represents the FRET/CFP ratio and brightness represents the fluorescence intensity of the FRET channel.

Measurement of layer curvature and thickness

For 2D measurement of curvature and thickness, we first performed whole-mount immunofluorescence of E-cadherin to visualize the cochlear epithelium and acquired z-stack images by confocal microscopy as described above. Next, we manually traced the apical and basal sides of epithelial cells on the middle horizontal section of the roof–floor axis. The extracted epithelial layer was named as the lateral epithelial layer according to the side based on a manually chosen apex tip point. Then, the curve of the epithelial layer was obtained by the iterative skeletonization, and discrete points (xi,yi) were sampled along the curves at regular intervals of 15 µm. Finally, fitting the discrete points with a cubic spline function, the function Si at an interval [xi,xi+1] is denoted as

Six=aix-xi3+bix-xi2+cix-xi+di

Due to a definition of curvature κ(x)=S''(1+S'2)-3/2, the curvature from the spline function was calculated as

κi(x)=6ax-xi+2b(1+3ax-xi2+2bx-xi+c2)3/2.

The curve of the convex/concave to the duct lumen was assigned as positive/negative in κ. We defined the layer thickness as a linear length connecting to the luminal and basal edge, which is vertical to the curve of the epithelial layer at sampling points.

EdU intensity mapping

First, we separated 8-bit staining image stacks for E-cadherin and EdU into two regions, roof and floor, based on z position at the middle point, and performed three different projection methods onto the xy plane, including (1) maximum intensity projection, (2) summed intensity projection, and (3) mean intensity projection averaged within the cochlear duct. The mean intensity projection evaluates the EdU signals averaged only within the cochlear duct epithelium so that the denominators can be changed depending on the epithelial thickness on each measurement region. Therefore, there would be different results between the mean intensity projection and summed intensity projection. Next, we binarized immunostaining signals for E-cadherin using Otsu’s method with morphological operations and detected periphery of the cochlear duct with the MATLAB function ‘bwperim’. The medial curve was defined by connecting between the apex tip and the end of medial edge, both of which were given manually, along the duct periphery. Similarly, the lateral curve was defined by connecting between the apex tip and the end of lateral edge along the duct periphery. Then, we marked points to make 20 bins at a constant distance each along the medial curve and lateral curve. By connecting the marked point of the medial curve and that of the lateral curve indexed by an order from the apex tip and dividing the lines into 10, we partitioned the cochlear duct into small regions for measurement. Finally, we measured the averaged intensity of EdU signal within each region and normalized by 255. The EdU intensity distribution was worth being considered as a marker of local tissue growth at a fixed moment in time due to cell proliferation and homogeneous cell density.

Cell density measurement

Five equally divided sections along the mediolateral axis were set on the floor or the roof of the cochlear duct at E12.5. In each section, the supracellular region including more than hundred cells that overlapped each other was manually chosen. Then, the number of cells and area were measured within each region. The center of mass of the region was regarded as the position of that region.

Tissue flow and ERK activity

To calculate velocity fields of cells in cochleae, we performed PIV-based image processing using a free code MatPIV (a GNU public license software distributed by Prof. Kristian Sveen in University of Oslo) that was applied to time-lapse images of the CFP channel. Velocity fields at time T were computed by displacement between T and T+Δt. Δt was set as the sampling rate, 12 min. The size of the interrogation window was set to 40 pixels, approximately 25 µm, corresponding to 4–5 cell diameter, and the window overlap was set to 50%. The obtained velocity data were then smoothened via median filtering to eliminate peaky noises. We then obtained the 'tissue flow speed for the elongation' from PIV velocity vector projected onto the apex–base line depicted in Figure 3D and calculated the spatial derivative of the tissue flow speed for the elongation between two adjacent interrogation windows on the apex–base line according to the definition of a diagonal component of the strain rate. This quantity was smoothened using the MATLAB function 'smooth' to eliminate high-frequency components and defined as the extension-shrinkage rate. As for the ERK activity, we set thresholds in CFP images using Otsu’s method within each interrogation window to extract the cytoplasmic region and calculated the mean FRET/CFP ratio in the binarized region. The ERK activity rate was calculated as the time derivative of the ERK activity. Cross-correlation analysis was performed using the MATLAB function 'xcorr'.

Tissue curvature and ERK activity

The interrogation squared windows, each of which had 20 pixels (approximately 13 µm) per side, were set along the manually traced lateral edge of the cochlear duct. For the calculation of tissue curvature, the reference point on the traced lateral edge was determined as the nearest from each interrogation window. Four points centered on the reference point were sampled along the lateral edge at regular intervals of 19 µm, and the curvature at the reference point was calculated as described above (see Measurement of layer curvature and thickness). For the ERK activity, the measurement was performed as described above (see Tissue flow and ERK activity).

Statistical analysis

The number of cells or region of interests analyzed (n) and the number of biological replicates (N) are indicated in the figure legends. No particular statistical method was used to predetermine the sample size. A minimum of N = 3 independent experiments was performed based on previous studies in the field (Cohen et al., 2020; Driver et al., 2017; Tateya et al., 2019). No inclusion/exclusion criteria were used, and all analyzed samples were included in the analysis. No randomization was performed. Statistical tests, sample sizes, test statistics, and p-values are described in the main text. p-Values of less than 0.05 were considered to be statistically significant in two-tailed tests and were classified into four categories: *p<0.05, **p<0.01, ***p<0.001, and n.s. (not significant, i.e., p≥0.05).

Software

For digital image processing, we used MATLAB (MathWorks) and Image J (National Institute of Health). For graphics, we used MATLAB (MathWorks), Imaris (Bitplane), and ImageJ (National Institute of Health). For statistical analysis, we used MATLAB (MathWorks).

Mathematical model

Modeling oscillatory ERK activation waves and cell flows

We built a minimal 1D mechanochemical coupling model for the collective cell migration based on our previous studies (Boocock et al., 2021; Hino et al., 2020). Cells, each indexed as j=1,...,N, are represented as a chain of springs, whose junctions including the boundaries are labeled as i=1,...,N+1, with elastic constant k, and each cell generates contractile force at the rear side of the cell to move to the front with a force F. That is, the cell contractile force with j=N (Fj=N) is regarded as the force at the junction i=N (Fi=N). Because the epithelial cells adhere to neighboring cells and thus transmit the elastic force with viscous frictions ηc, the dynamics of cell collectives is represented as

ηcx˙i=k(εjεj+1)+Fj=i,εj=(xi+1xi)/L1,fori=1...Nandj=1...N (1)

where ε is the cell strain and L the typical cell length, that is, 5 µm. At the front edge of the cells, that is, i=N+1, a self-propelling force Ftip is generated, reflecting an elongation of the apex tip. Since the cells respond to stretching as activating the ERK, a coupling between the cell kinematics and the ERK activity is formulated as

ηEE˙j=tanh(αεj)Ej, (2)

where α denotes the sensitivity parameter and ηE the timescale of the dynamics. Then, the ERK activity is converted to the self-contractile force represented as dynamics of the

ηFF˙j=λEjFj, (3)

where λ denotes the controlling parameter of amplitude and ηF the timescale.

As for the uncoupling regime, ERK activity was given as the following traveling waves instead of Equation (2):

EX=sinπXw+vt+1×0.5 (4)

where w is the characteristic length of ERK activation and v the ERK activation speed.

The parameter w was set as 84 µm, that is, the wavelength of the ERK activity is 168 µm, from Figure 3E, and v was set as 0.42 µm min−1, from Figure 3F.

Numerical simulation

The ordinary differential equations were numerically solved by the forward Euler method with time step 0.01 using MATLAB. The number of cells N was set as 1000, and one boundary i=1 was fixed and the other i=N+1 was the moving boundary condition. The biologically plausible parameter set was determined as ηc = 40 (nN min µm−1), k = 20 (nN), Ftip = 6 (nN), α = 3, ηE = 30 (min), ηF = 10 (min), and λ = 9 (nN) according to the present study and a previous study (Serra-Picamal et al., 2012).

Code availability

The MATLAB code is available at https://github.com/tsuyoshihirashima/2020_cochlearduct/blob/main/erk1d.mIshii, 2021; copy archived at swh:1:rev:e81398f7827e8a7f91171191224e269cea4685f4.

Acknowledgements

This work was supported by the JSPS KAKENHI 17KT0107 and 19H00993, the JST PRESTO JPMJPR1949 and CREST JPMJCR1654, and the Medical Research Support Center of Kyoto University. We thank Akane Kusumi for technical assistance, and Edouard Hannezo, Naoya Hino, and Yoshiko Takahashi for fruitful discussion.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Tsuyoshi Hirashima, Email: hirashima.tsuyoshi.2m@kyoto-u.ac.jp.

Doris K Wu, NIDCD, NIH, United States.

Marianne E Bronner, California Institute of Technology, United States.

Funding Information

This paper was supported by the following grants:

  • Japan Society for the Promotion of Science 17KT0107 to Tsuyoshi Hirashima.

  • Japan Society for the Promotion of Science 19H00993 to Michiyuki Matsuda, Tsuyoshi Hirashima.

  • Japan Science and Technology Agency JPMJPR1949 to Tsuyoshi Hirashima.

  • Japan Science and Technology Agency JPMJCR1654 to Michiyuki Matsuda.

Additional information

Competing interests

No competing interests declared.

Author contributions

Resources, Data curation, Software, Formal analysis, Validation, Investigation, Methodology, Writing - original draft.

Conceptualization, Methodology, Writing - review and editing.

Resources, Supervision, Funding acquisition, Writing - review and editing.

Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing.

Ethics

Animal experimentation: All the animal experiments were approved by the local ethical committee for animal experimentation (MedKyo 19090 and 20081) and were performed in compliance with the guide for the care and use of laboratory animals at Kyoto University.

Additional files

Transparent reporting form

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files are provided.

References

  1. Aoki K, Kondo Y, Naoki H, Hiratsuka T, Itoh RE, Matsuda M. Propagating wave of ERK activation orients collective cell migration. Developmental Cell. 2017;43:305–317. doi: 10.1016/j.devcel.2017.10.016. [DOI] [PubMed] [Google Scholar]
  2. Barrett SD, Bridges AJ, Dudley DT, Saltiel AR, Fergus JH, Flamme CM, Delaney AM, Kaufman M, LePage S, Leopold WR, Przybranowski SA, Sebolt-Leopold J, Van Becelaere K, Doherty AM, Kennedy RM, Marston D, Howard WA, Smith Y, Warmus JS, Tecle H. The discovery of the benzhydroxamate MEK inhibitors CI-1040 and PD 0325901. Bioorganic & Medicinal Chemistry Letters. 2008;18:6501–6504. doi: 10.1016/j.bmcl.2008.10.054. [DOI] [PubMed] [Google Scholar]
  3. Bok J, Zenczak C, Hwang CH, Wu DK. Auditory ganglion source of sonic hedgehog regulates timing of cell cycle exit and differentiation of mammalian cochlear hair cells. PNAS. 2013;110:13869–13874. doi: 10.1073/pnas.1222341110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Boocock D, Hino N, Ruzickova N, Hirashima T, Hannezo E. Theory of mechanochemical patterning and optimal migration in cell monolayers. Nature Physics. 2021;17:267–274. doi: 10.1038/s41567-020-01037-7. [DOI] [Google Scholar]
  5. Chen P, Johnson JE, Zoghbi HY, Segil N. The role of Math1 in inner ear development: uncoupling the establishment of the sensory primordium from hair cell fate determination. Development. 2002;129:2495–2505. doi: 10.3410/f.1006295.78812. [DOI] [PubMed] [Google Scholar]
  6. Cohen R, Amir-Zilberstein L, Hersch M, Woland S, Loza O, Taiber S, Matsuzaki F, Bergmann S, Avraham KB, Sprinzak D. Mechanical forces drive ordered patterning of hair cells in the mammalian inner ear. Nature Communications. 2020;11:5137. doi: 10.1038/s41467-020-18894-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. De Simone A, Evanitsky MN, Hayden L, Cox BD, Wang J, Tornini VA, Ou J, Chao A, Poss KD, Di Talia S. Control of osteoblast regeneration by a train of erk activity waves. Nature. 2021;590:129–133. doi: 10.1038/s41586-020-03085-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Driver EC, Northrop A, Kelley MW. Cell migration, intercalation and growth regulate mammalian cochlear extension. Development. 2017;144:3766–3776. doi: 10.1242/dev.151761. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Groves AK, Fekete DM. Shaping sound in space: the regulation of inner ear patterning. Development. 2012;139:245–257. doi: 10.1242/dev.067074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Harvey CD, Ehrhardt AG, Cellurale C, Zhong H, Yasuda R, Davis RJ, Svoboda K. A genetically encoded fluorescent sensor of ERK activity. PNAS. 2008;105:19264–19269. doi: 10.1073/pnas.0804598105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hino N, Rossetti L, Marín-Llauradó A, Aoki K, Trepat X, Matsuda M, Hirashima T. ERK-Mediated mechanochemical waves direct collective cell polarization. Developmental Cell. 2020;53:646–660. doi: 10.1016/j.devcel.2020.05.011. [DOI] [PubMed] [Google Scholar]
  12. Hirashima T, Adachi T. Procedures for the quantification of whole-tissue immunofluorescence images obtained at single-cell resolution during murine tubular organ development. PLOS ONE. 2015;10:e0135343. doi: 10.1371/journal.pone.0135343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hirashima T, Adachi T. Polarized cellular mechano-response system for maintaining radial size in developing epithelial tubes. Development. 2019;146:dev181206. doi: 10.1242/dev.181206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hiratsuka T, Fujita Y, Naoki H, Aoki K, Kamioka Y, Matsuda M. Intercellular propagation of extracellular signal-regulated kinase activation revealed by in vivo imaging of mouse skin. eLife. 2015;4:e05178. doi: 10.7554/eLife.05178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Ishii M. 2020_cochlearduct. swh:1:rev:e81398f7827e8a7f91171191224e269cea4685f4Software Heritage. 2021 https://archive.softwareheritage.org/swh:1:dir:72666fd3c72ac48ce0b0242b6afc0fbbf5301e56;origin=https://github.com/tsuyoshihirashima/2020_cochlearduct/;visit=swh:1:snp:17519469e0b7a561b0f29367b2d27d03666f2106;anchor=swh:1:rev:e81398f7827e8a7f91171191224e269cea4685f4/
  16. Johnson AB, Fogel NS, Lambert JD. Growth and morphogenesis of the gastropod shell. PNAS. 2019;116:6878–6883. doi: 10.1073/pnas.1816089116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Koehler KR, Mikosz AM, Molosh AI, Patel D, Hashino E. Generation of inner ear sensory epithelia from pluripotent stem cells in 3D culture. Nature. 2013;500:217–221. doi: 10.1038/nature12298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Koehler KR, Nie J, Longworth-Mills E, Liu XP, Lee J, Holt JR, Hashino E. Generation of inner ear organoids containing functional hair cells from human pluripotent stem cells. Nature Biotechnology. 2017;35:583–589. doi: 10.1038/nbt.3840. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Komatsu N, Aoki K, Yamada M, Yukinaga H, Fujita Y, Kamioka Y, Matsuda M. Development of an optimized backbone of FRET biosensors for kinases and GTPases. Molecular Biology of the Cell. 2011;22:4647–4656. doi: 10.1091/mbc.e11-01-0072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Komatsu N, Terai K, Imanishi A, Kamioka Y, Sumiyama K, Jin T, Okada Y, Nagai T, Matsuda M. A platform of BRET-FRET hybrid biosensors for optogenetics, chemical screening, and in vivo imaging. Scientific Reports. 2018;8:8984. doi: 10.1038/s41598-018-27174-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Liu Z, Owen T, Zhang L, Zuo J. Dynamic expression pattern of sonic hedgehog in developing cochlear spiral ganglion neurons. Developmental Dynamics. 2010;239:1674–1683. doi: 10.1002/dvdy.22302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mao Y, Mulvaney J, Zakaria S, Yu T, Morgan KM, Allen S, Basson MA, Francis-West P, Irvine KD. Characterization of a Dchs1 mutant mouse reveals requirements for Dchs1-Fat4 signaling during mammalian development. Development. 2011;138:947–957. doi: 10.1242/dev.057166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Montcouquiol M, Kelley MW. Development and patterning of the cochlea: from convergent extension to planar polarity. Cold Spring Harbor Perspectives in Medicine. 2020;10:a033266. doi: 10.1101/cshperspect.a033266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ogura Y, Wen FL, Sami MM, Shibata T, Hayashi S. A Switch-like activation relay of EGFR-ERK signaling regulates a wave of cellular contractility for epithelial invagination. Developmental Cell. 2018;46:162–172. doi: 10.1016/j.devcel.2018.06.004. [DOI] [PubMed] [Google Scholar]
  25. Pauley S, Wright TJ, Pirvola U, Ornitz D, Beisel K, Fritzsch B. Expression and function of FGF10 in mammalian inner ear development. Developmental Dynamics. 2003;227:203–215. doi: 10.1002/dvdy.10297. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Pirvola U, Spencer-Dene B, Xing-Qun L, Kettunen P, Thesleff I, Fritzsch B, Dickson C, Ylikoski J. FGF/FGFR-2(IIIb) signaling is essential for inner ear morphogenesis. The Journal of Neuroscience. 2000;20:6125–6134. doi: 10.1523/JNEUROSCI.20-16-06125.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Qian D, Jones C, Rzadzinska A, Mark S, Zhang X, Steel KP, Dai X, Chen P. Wnt5a functions in planar cell polarity regulation in mice. Developmental Biology. 2007;306:121–133. doi: 10.1016/j.ydbio.2007.03.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Raup DM, Michelson A. Theoretical morphology of the coiled shell. Science. 1965;147:1294–1295. doi: 10.1126/science.147.3663.1294. [DOI] [PubMed] [Google Scholar]
  29. Riedl J, Flynn KC, Raducanu A, Gärtner F, Beck G, Bösl M, Bradke F, Massberg S, Aszodi A, Sixt M, Wedlich-Söldner R. Lifeact mice for studying F-actin dynamics. Nature Methods. 2010;7:168–169. doi: 10.1038/nmeth0310-168. [DOI] [PubMed] [Google Scholar]
  30. Saburi S, Hester I, Fischer E, Pontoglio M, Eremina V, Gessler M, Quaggin SE, Harrison R, Mount R, McNeill H. Loss of Fat4 disrupts PCP signaling and oriented cell division and leads to cystic kidney disease. Nature Genetics. 2008;40:1010–1015. doi: 10.1038/ng.179. [DOI] [PubMed] [Google Scholar]
  31. Saffer AM, Carpita NC, Irish VF. Rhamnose-Containing cell wall polymers suppress helical plant growth independently of microtubule orientation. Current Biology. 2017;27:2248–2259. doi: 10.1016/j.cub.2017.06.032. [DOI] [PubMed] [Google Scholar]
  32. Sato K, Hiraiwa T, Maekawa E, Isomura A, Shibata T, Kuranaga E. Left-right asymmetric cell intercalation drives directional collective cell movement in epithelial morphogenesis. Nature Communications. 2015;6:10074. doi: 10.1038/ncomms10074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Serra-Picamal X, Conte V, Vincent R, Anon E, Tambe DT, Bazellieres E, Butler JP, Fredberg JJ, Trepat X. Mechanical waves during tissue expansion. Nature Physics. 2012;8:628–634. doi: 10.1038/nphys2355. [DOI] [Google Scholar]
  34. Smyth DR. Helical growth in plant organs: mechanisms and significance. Development. 2016;143:3272–3282. doi: 10.1242/dev.134064. [DOI] [PubMed] [Google Scholar]
  35. Sun L, Tran N, Liang C, Tang F, Rice A, Schreck R, Waltz K, Shawver LK, McMahon G, Tang C. Design, synthesis, and evaluations of substituted 3-[(3- or 4-Carboxyethylpyrrol-2-yl)methylidenyl]indolin-2-ones as Inhibitors of VEGF, FGF, and PDGF Receptor Tyrosine Kinases. Journal of Medicinal Chemistry. 1999;42:5120–5130. doi: 10.1021/jm9904295. [DOI] [PubMed] [Google Scholar]
  36. Tateya T, Imayoshi I, Tateya I, Hamaguchi K, Torii H, Ito J, Kageyama R. Hedgehog signaling regulates prosensory cell properties during the basal-to-apical wave of hair cell differentiation in the mammalian cochlea. Development. 2013;140:3848–3857. doi: 10.1242/dev.095398. [DOI] [PubMed] [Google Scholar]
  37. Tateya T, Sakamoto S, Ishidate F, Hirashima T, Imayoshi I, Kageyama R. Three-dimensional live imaging of Atoh1 reveals the dynamics of hair cell induction and organization in the developing cochlea. Development. 2019;146:dev177881. doi: 10.1242/dev.177881. [DOI] [PubMed] [Google Scholar]
  38. Thompson DW. On Growth and Form. Dover Publications; 1942. [DOI] [Google Scholar]
  39. Urness LD, Wang X, Shibata S, Ohyama T, Mansour SL. Fgf10 is required for specification of non-sensory regions of the cochlear epithelium. Developmental Biology. 2015;400:59–71. doi: 10.1016/j.ydbio.2015.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Urness LD, Wang X, Doan H, Shumway N, Noyes CA, Gutierrez-Magana E, Lu R, Mansour SL. Spatial and temporal inhibition of FGFR2b ligands reveals continuous requirements and novel targets in mouse inner ear morphogenesis. Development. 2018;145:dev170142. doi: 10.1242/dev.170142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Wada H, Matsumoto D. Twisting Growth in Plant Roots Plant Biomechanics. In: Gitmann A, Grill J, editors. Structure to Function at Multiple Scales. Springer International Publishing; 2018. pp. 127–140. [DOI] [Google Scholar]
  42. Wang J, Mark S, Zhang X, Qian D, Yoo SJ, Radde-Gallwitz K, Zhang Y, Lin X, Collazo A, Wynshaw-Boris A, Chen P. Regulation of polarized extension and planar cell polarity in the cochlea by the vertebrate PCP pathway. Nature Genetics. 2005;37:980–985. doi: 10.1038/ng1622. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Yamamoto N, Okano T, Ma X, Adelstein RS, Kelley MW. Myosin II regulates extension, growth and patterning in the mammalian cochlear duct. Development. 2009;136:1977–1986. doi: 10.1242/dev.030718. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: Doris K Wu1
Reviewed by: Andres Collazo2, David Sprinzak3

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

It is a huge challenge to study the morphogenetic events that give rise to the snail-shaped mammalian cochlea. Through live imaging of FRET-based mouse cochlear explants and mathematical modeling, this study demonstrated that a concomitant wave of an oscillatory wave of ERK activity and cell flow in the opposite direction contributes to the extension of the developing cochlear duct. This work provides a conceptual framework for considering factors that are required for cochlear morphogenesis and contributes to the general understanding of complex organ formation.

Decision letter after peer review:

Thank you for submitting your article "Interplay between medial nuclear stalling and lateral cellular flow underlies cochlear duct morphogenesis" for consideration by eLife. Your article has been reviewed by Marianne Bronner as the Senior Editor, a Reviewing Editor, and three reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Andres Collazo (Reviewer #2); David Sprinzak (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

The mammalian cochlear duct is a spiral-shaped organ. This study investigated the mechanisms underlying the bending of the cochlear duct. Using two-photon live imaging and mathematical modeling, it was reported that the bending of the cochlear duct is caused by stalling of nuclei in the luminal side of the medial cochlear duct during interkinetic nuclear migration. Using FRET-based imaging, cochlear duct elongation is attributed to an oscillatory wave of ERK activity originating from the cochlear tip.

All three reviewers were impressed by the imaging results. Although the reviewers and editors find the concept and approach interesting, blocking cell proliferation may be too crude a method to address the authors' hypothesis and many questions were raised by the results of blocking cell division. Second, the relationship between cell proliferation and ERK-driven migration is also unclear. Please see comments from reviewer #2 and #3 for specifics. Third, what is the relationship between SHH-induced proliferation and ERK activation as suggested by the authors (see comments from reviewer #1)? Additionally, it is problematic to illustrate a difference in the bending force between medial and lateral cochlear duct that is presumably occurring at E12.5 and E14.5 with a cochlear dissection at E17.5. The tissue architecture is completely different between E12.5 and E17.5. The surgery basically removed a specific region of the cochlear duct, the stria vascularis, rather than medial versus lateral halves of the cochlear duct.

We hope the authors will be able to address these questions and comments and submit a revision. The full reviews are appended below for your information.

Reviewer #1:

This is a fascinating manuscript that explores for the first time the potential mechanisms underlying cochlear morphogenesis. The authors have used a combination of modeling, beautiful imaging and ERK-FRET reporter mice reporter mice to suggest at least two processes may be at play in cochlear shaping – differential interkinetic nuclear migration and a cellular flow that appears to correlate with ERK activation.

I have no major concerns with this lovely piece of work. The imaging and quantification is meticulous, and the observations made by the authors are novel and will of great interest to cell biologists interested in morphogenesis, no just aficionados of the inner ear.

The one suggestion I would make is for the authors to clarify the relationship between cell proliferation and ERK activation. When they reference the inner ear literature, they point out that FGF pathway mutants have deficient cochlear morphogenesis and proliferation, and they hypothesize that FGF-induced ERK activation may be responsible for their propagating waves. However, they also reference work suggesting that cellular extension during collective migration can also induce ERK activation and also suggest SHH-induced proliferation as another causative factor in promoting ERK activation through proliferation. I think the authors should try and clarify this – both in their explanation, but also by comparing the effects of the MEK inhibitor PD0325901 on ERK activity and tissue flow speed (Figure 4I and Figure 3—figure supplement 1F) with the effects of the FGFR inhibitor SU5402, and also Shh inhibitors like cyclopamine. If the effects they see are directly due to FGF signaling, one would expect a change in ERK activation and cell flow with the same kinetics as with PD0325901. However, if Shh-induced proliferation is responsible, the change in ERK activation would take much longer to achieve. I think these experiments should be possible to do in a relatively short period of time.

Reviewer #2:

The paper by Hirashima and colleagues shows some interesting cellular mechanisms they conclude drive the spiraling and outgrowth of the mammalian cochlea. The two cellular mechanisms they propose are supported by experiments and modeling. The spiraling ERK wave and the contrasting movement of lateral cells was very intriguing. However, the ERK wave and lateral cell movements seem disconnected from the bending forces discussed. Are the authors saying that the ERK mediated lateral cell movements are important for cochlear growth while the MEL is important for the bending? The two mechanisms they discuss seem insufficient to explain all of cochlear spiraling. Other cellular mechanisms such as cell proliferation and convergent extension are mentioned but their roles are not incorporated into their discussion. Are they not required? How do they complement their results?

1) While the authors talk about bending forces, the paper has no measurements of the forces generated by different tissues. I also feel there are other cellular mechanisms that are mentioned but never incorporated into their proposed explanation for duct coiling such as convergent extension and actomyosin based basal shrinkage. Proliferation is discussed quite a bit but seems to be dismissed as a force. In the introduction they mention how Shh mediated proliferation is required for duct elongation while Fgf10 null mutants have a shortened duct yet normal proliferation. So what is the role for proliferation? Maybe they can answer this in the context of their interesting observation that there is more proliferation in the roof than the floor which would be predicted to bend the cochlea along that axis. When combined with the medial lateral bending could these two forces result in the spiraling? It also seems like this differential proliferation between the floor and roof was in more than just the epithelium, correct? Could the cartilaginous capsule around the duct guide the bending as well? In their culture experiments, if too much of the capsule was removed then normal duct development was disrupted.

2) Their demonstration that the bending forces are in the medial half is interesting but the only tissue whose mechanism is studied is the MEL. Could convergent extension in other medial tissues such as the prosensory domain (which Wang et al., showed was occurring in this tissue) and surrounding mesenchyme be the main force generator for the bending of the medial half of the cochlear duct? Does the MEL cultured by itself bend? They say that cell intercalation can drive ductal elongation but not bending (Introduction) but can't convergent extension occur asymmetrically in the tissue? Such as by occurring in the overlying medial mesenchyme but not in the medial epithelium. It should be noted that the bending by the epithelium does not have to provide high forces as long as the force provided by other tissues are similar across the medial lateral axis, the bending in the epithelium could bias the mass of tissue to bend.

3) The mathematical modeling for the luminal bending is less convincing than the mathematical modeling for the ERK and Cell flow coupling. The simulated curves in Figure 2K are quite different from the Experimental measure in Figure 2M, especially for the Mitomycin C condition. I feel that the values plugged in for the Numerical simulation, the standard parameter set were not well justified. What happened to the simulations as these values changed? Was the parameter space for acceptable values broad? In contrast the parameters for the numerical simulation of the ERK activation waves and cell flows were well justified. The parameters chosen might explain the big differences between simulation and experimental in Figure 2.

4) For the cell tracking experiments in the lateral region the resolution was 4-5 cells. The resulting cell flow patterns were very interesting but why didn't the authors track single cells? Segmenting individual cells via cytoplasmic labeling is much trickier but the nuclei are identifiable and the Imaris software they used in the paper has a cell tracking feature for such labeling. I would think that individual cell movements might provide more insights. In subsection “Retrograde helical ERK activation waves drive base-to-apex multicellular flow” they say they can see cell contractions which I assume is for individual cells? How were cell contractions identified? Video 5 was excellent and very informative. Do the cell flows correlate at all with the proliferation seen with Edu staining?

Reviewer #3:

The manuscript by Ishii et al., focuses on understanding how cellular dynamics drive the spiral shape of the cochlear duct in mammals. The authors use live imaging of inner ear explants to follow dynamics of interkinetic nuclear migration (IKNM) and ERK activity (using ERK FRET sensor) to track some of the processes that give rise to tissue bending during spiral duct formation. On the imaging side, the manuscript presents a technical tour de force, showing remarkable two photon imaging capabilities that provide insights into the dynamics underlying cochlear extension. These experiments reveal several new observations: (1) Medial epithelial layer (MEL) tends to bend more than the lateral epithelial layer (LEL) despite being more proliferative. (2) That nuclei of cells in the curved region of the cochlea tend to stay in the luminal side, following cell division, rather than migrate back to the basal side. (3) The cells migrate towards the apical lateral roof. (4) That there are orchestrated ERK waves that correlate with cell migration. Based on these observations and on mathematical modeling, the manuscript has two main claims: (1) that nuclear stalling on the luminal side following cell division leads to increased curvature which gives rise cochlear duct bending, and (2) that multicellular flow mediated by ERK signaling waves pushes cells towards the growing apex, supplying the cells required for luminal expansion. While the observations in the manuscript are certainly interesting, I worry however, that some of the claims are not sufficiently substantiated, and also the connection between the two observations is rather weak. Here are the detailed concerns:

1) The authors argue that cell cycle arrest results in a decrease in the curvature of the cochlear duct, which supports the hypothesis that luminal nuclear stalling promotes MEL bending. This is fine, but luminal nuclear stalling can be a result and not a cause. Since in a bent region, the basal side is more packed, this density gradient can be the cause of nuclei stalling at the luminal side. The fact that the curvature decreased but not diminished after cell cycle arrest could suggest that nuclear stalling is not required for bending, but rather reinforces it.

2) Since the authors discuss both cell proliferation and nuclear stalling, and cell migration, as forces that can drive bending and coiling, it hard to interpret the results of the mitomycin C experiment. Could it be that the tissue is less curved because there are less cells to supply the elongation tissue rather than less nuclear stalling? The authors should consider inhibiting either cell migration or the cytoskeletal machinery required for IKNM to dissect these effects.

3) The authors present a mathematical model to demonstrate that nuclear stalling in the luminal side results in bending. To model nuclear motion they use a parameter, γ, which controls the degree of basalward movement after IKNM. Modeling in such way means that other than γ=1, the nucleus never fully returns to the basal side, but if I understand correctly this is not the case, as even if the nuclei that stall at the luminal side, eventually return to the basal side.

4) Furthermore, for luminal nuclear stalling, the authors tracked only the nuclei of dividing cells. This makes the data in Figure 2D' much clearer. However, in their model the authors show only these nuclei and not all nuclei. In addition, they show many crowded nuclei in the model, yet this is not observed in the images provided in the manuscript. Therefore, it seems the model does not represent the morphology of the tissue properly. The authors should model the process with non-dividing cells at the basal side.

5) In subsection “Spatial heterogeneity of cell proliferation suggests cellular inflow to the lateral side of the apex to realize cochlear bending” the authors claim that the higher volumetric growth measured at the MEL should cause an opposite curvature relative to the innate one. This is true if EdU intensity is proportional to volumetric growth, but cells in the MEL and LEL may not be the same size. For example, cells in the MEL could be smaller than cells in the LEL. The authors should therefore measure the nuclei number density and the volumetric cell density to clarify this. If the number density of the nuclei is indeed higher at the MEL, it may also explain the higher structural integrity of the MEL relative to the LEL demonstrated in Figure 1C.

6) The authors show the effect of ERK inhibition on tissue flow speed. This is a very important observation and raises several important questions. What is effect of ERK inhibition on curvature? On tissue length? On proliferation? These will provide a more complete understanding of the effect of RK inhibition.

7) The authors should also test the effect of mitomycin C on cell flow and ERK activity. As mentioned above, it is not clear whether the effect of mitomycin C is a result of less nuclear stalling or perhaps less cells that flow towards the apex.

8) In Figure 3 the authors analyze the EdU distribution over the cochlear duct. This analysis is done using the maximum intensity projection of the stack. It seems that a more accurate way to quantify would be to use the summed intensity image rather than the maximum intensity image. This may reveal additional details that were missed by throwing away all other layers except the one at maximum intensity.

9) In Figure 4 the colors used for the ERK activity analysis are very hard to see for color-blind people. It would be easier for this audience if the authors changed one of these colors to green/red/yellow.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Interplay between medial nuclear stalling and lateral cellular flow underlies cochlear duct morphogenesis" for consideration by eLife. Your article has been re-reviewed by Marianne Bronner as the Senior Editor, a Reviewing Editor, and three reviewers.

The reviewers and the Reviewing Editor have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

All three reviewers have judged that your manuscript is of interest and represents an advancement to the field. However, the manuscript cannot be accepted at its current form because of the issue whether IKNM described is causal or consequential to cochlear bending. This issue needs to be better resolved. Either the mathematical model is revised to accommodate for fewer cells stalling at the luminal side but could still account for the cochlear bending or more cells can be observed to stall at the luminal side that could account for the cochlear bending. Alternatively, the entire manuscript can be totally revised to accommodate for these two possibilities in an unbiased manner, starting with the Title.

Since many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Specific comments from two of the reviewers are listed below as guidelines for your revision:Reviewer #2:

The authors have addressed most of my comments. The experiments with the MEK inhibitor PD0325901 are not totally convincing. I worry about nonspecific effects of this pharmacological reagent. Another inhibitor of cell proliferation like a Cochlea from a KO mouse would provide a second piece of evidence. How did the FGF inhibitor SU5402 effect curvature? In Figure 4C, I was not sure why do the measurements stop at 0.8 mm? It seems like the trend was going down which would make it not significant?

In the Discussion the authors state they "provided the first experimental evidence that nuclei stall at the luminal side of the pseudostratified epithelium during IKNM in normal development". I go back to the discrepancy between the simulation and experimental results in Figure 2J-M. They address this in the Materials and methods but some mention of it here would be appropriate.Reviewer #3:

The authors address many of the points raised in my review and the other reviewers. The section on the ERK waves and the mechanochemical feedback has improved considerably and is actually very nice. In particular the new data on the effect of blebbistatin is very nice and supports the model.

I do, however, have some problem with the first part of the manuscript on the interkinetic nuclear movement (IKNM). The main problem here is that the model they suggest ignore an essential aspect of the system which is that only some small fraction of the cells perform the luminal stalling at a given time. The model suggests that the cause for bending is that luminal stalling leads to an inverted wedge like morphology of the cells and thus leads to bending. However, in the model, luminal stalling is assumed to happen in all the cells. This does not seem to match the images in Figure 2A-C showing that most nuclei are closer to basal position. Hence, to model that the authors should have assumed that there are only few cells whose nuclei are stalled at the laminal position. It seems to me, that it is unlikely that such a model would produce significant bending in that case. In their response, the authors argue that their model like all mathematical models is a simplification of the real system. While it is true that models always simplify, it seems to me that the assumption that all cells stall is an essential divergence from the real system and cannot be simplified.

As I suggested previously, the nuclear stalling may actually be a result of bending and not the cause. I understand that this is hard to test experimentally. However, I feel that the current model maybe misleading.

Since I like the second part and find it interesting and more convincing, it may be worth for the authors to restructure their manuscript so that the ERK waves are at the beginning and are the main focus. The IKNM is an interesting observation that can be added with the two potential interpretations.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Retrograde ERK activation waves drive base-to-apex multicellular flow in murine cochlear duct morphogenesis" for consideration by eLife. Your article has been reviewed by Marianne Bronner as the Senior Editor, a Reviewing Editor, and three reviewers. No reviewers found for this submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

The spiral shape of the mammalian cochlear duct is tightly coupled to its function in sound detection. However, the morphogenetic process that gives rise to the spiral structure is poorly understood. In this paper, by live-imaging of the developing cochlea from a FRET-based reporter mouse strain, Hiroshima et al. demonstrated elegantly that one of the mechanisms in cochlear duct elongation is driven by an oscillatory wave of ERK activity originating from the roof of the cochlear apex towards the base and a concomitant flow of cells from the base of the cochlear floor towards the roof of the apex.

All three reviewers felt the revised manuscript is much improved and concise by removing nuclear stalling data. The only suggestion is for you to consider a 3D schematic summary that illustrates direction of ERK oscillation, cell movement flow and the hot spot of cell proliferation in the floor of the cochlea. This will help readers outside of the ear field to appreciate your beautiful work.

Second, the citation of Bok et al., for reduction of cell proliferation at the base of the cochlear duct is incorrect. It is ok to state that cochlear duct growth requires Shh signaling but it is incorrect to say there is significant decrease in cell proliferation at the base of the cochlea in the Shh conditional mutants. First, only EdU-labeled hair cells were quantified in that paper, which could hardly account for the shortened cochlear duct phenotype. More importantly, those experiments were not designed to look at proliferation but cell cycle exit. They were conducted by injecting Edu at E13.5 and E14.5 and EdU-labeled hair cells were analyzed at birth. Heavily labeled Edu-positive cells at birth were cells that exited cell cycle shortly after EdU administration. Therefore, a reduced EdU labeling could mean there were more cells dividing at the base at E13.5 and E14.5 rather than less proliferation and the label were diluted and no longer detected by birth.

eLife. 2021 Mar 5;10:e61092. doi: 10.7554/eLife.61092.sa2

Author response


All three reviewers were impressed by the imaging results. Although the reviewers and editors find the concept and approach interesting, blocking cell proliferation may be too crude a method to address the authors' hypothesis and many questions were raised by the results of blocking cell division.

We understand that blocking cell proliferation is a rough experiment to address our hypothesis and admit its weakness as a methodological aspect in our study. We pondered over other alternatives among our current experimental techniques, but unfortunately, this particular method was the only possible approach. In response to the comments, we mitigated the claim in the section on nuclear behaviors and mentioned this limitation in the Discussion section.

Second, the relationship between cell proliferation and ERK-driven migration is also unclear. Please see comments from reviewer #2 and #3 for specifics.

We apologize for the lack of clarity on this. In response to the comments, we added the corresponding paragraph in the Discussion section.

Third, what is the relationship between SHH-induced proliferation and ERK activation as suggested by the authors (see comments from reviewer #1)?

In response to the comment by reviewer #1, we added the new results in Figure 5 as requested. Thank you for this suggestion. We believe that this point is made clear in the present manuscript.

Additionally, it is problematic to illustrate a difference in the bending force between medial and lateral cochlear duct that is presumably occurring at E12.5 and E14.5 with a cochlear dissection at E17.5. The tissue architecture is completely different between E12.5 and E17.5. The surgery basically removed a specific region of the cochlear duct, the stria vascularis, rather than medial versus lateral halves of the cochlear duct.

Thank you very much for your constructive comments. As the tissue structure is different between the E12.5 and E17.5, we also agree that it might be difficult to directly connect the surgery results obtained from the E17.5 cochlea and the force balance of the developing cochlea at earlier stages. Here, we would like to emphasize that we do not intend to link the ablation experiments and conjecture of mechanical forces between different stages. Rather, in this beginning section, we aim to introduce a motivation for examining the differences in the tissue architecture along the mediolateral axis in detail. We apologize for the confusion, and our original submission was unclear regarding this point. In the present manuscript, we added the following sentences: “This result motivated us to explore the differences in physical properties between the medial and the lateral side in earlier stages.”

Reviewer #1:

This is a fascinating manuscript that explores for the first time the potential mechanisms underlying cochlear morphogenesis. The authors have used a combination of modeling, beautiful imaging and ERK-FRET reporter mice reporter mice to suggest at least two processes may be at play in cochlear shaping – differential interkinetic nuclear migration and a cellular flow that appears to correlate with ERK activation.

I have no major concerns with this lovely piece of work. The imaging and quantification is meticulous, and the observations made by the authors are novel and will of great interest to cell biologists interested in morphogenesis, no just aficionados of the inner ear.

The one suggestion I would make is for the authors to clarify the relationship between cell proliferation and ERK activation. When they reference the inner ear literature, they point out that FGF pathway mutants have deficient cochlear morphogenesis and proliferation, and they hypothesize that FGF-induced ERK activation may be responsible for their propagating waves. However, they also reference work suggesting that cellular extension during collective migration can also induce ERK activation and also suggest SHH-induced proliferation as another causative factor in promoting ERK activation through proliferation. I think the authors should try and clarify this – both in their explanation, but also by comparing the effects of the MEK inhibitor PD0325901 on ERK activity and tissue flow speed (Figure 4I and Figure 3—figure supplement 1F) with the effects of the FGFR inhibitor SU5402, and also Shh inhibitors like cyclopamine. If the effects they see are directly due to FGF signaling, one would expect a change in ERK activation and cell flow with the same kinetics as with PD0325901. However, if Shh-induced proliferation is responsible, the change in ERK activation would take much longer to achieve. I think these experiments should be possible to do in a relatively short period of time.

We appreciate you raising an important point. As you mentioned, we have argued that ERK activation requires cell migration flow and that SHH-induced proliferation is another underlying causative factor in promoting the ERK-migration feedback system. We agree that this point was unclear in our original submission.

In accordance with your comment, we performed an inhibitor assay with the FGFR inhibitor SU5402 and the Shh signaling inhibitor cyclopamine. Treatment with SU5402 resulted in a significant decrease in tissue flow speed along with ERK inactivation immediately after administration (3 hours), and the level was almost the same as that of treatment with the MEK inhibitor PD0325901. This indicates that ERK-driven cell migration is directly due to FGF signaling. However, for the cyclopamine treatment, we found that the degree of decrease in cell flow was not as significant as that of the MEK inhibitor treatment within 3 hours after administration and also observed no change in ERK activation. It took 8 hours to reach the same degree of speed decrease as the MEK inhibitor treatment. These results support that SHHinduced proliferation was less influential on ERK activation in a short time period but was prominent through cell proliferation over longer periods. We have added the corresponding new results in Figure 5 and the comment in the Discussion in response to your comments.

Reviewer #2:

The paper by Hirashima and colleagues shows some interesting cellular mechanisms they conclude drive the spiraling and outgrowth of the mammalian cochlea. The two cellular mechanisms they propose are supported by experiments and modeling. The spiraling ERK wave and the contrasting movement of lateral cells was very intriguing. However, the ERK wave and lateral cell movements seem disconnected from the bending forces discussed. Are the authors saying that the ERK mediated lateral cell movements are important for cochlear growth while the MEL is important for the bending? The two mechanisms they discuss seem insufficient to explain all of cochlear spiraling. Other cellular mechanisms such as cell proliferation and convergent extension are mentioned but their roles are not incorporated into their Discussion. Are they not required? How do they complement their results?

We thank the reviewer for bringing up many important questions, and we agree that our proposal might not be clear enough in the original submission. We read several questions and concerns that were rephrased in the following detailed comments. Please see our reply below. We believe that our present manuscript has been made more convincing and hope it meets your criteria.

1) While the authors talk about bending forces, the paper has no measurements of the forces generated by different tissues. I also feel there are other cellular mechanisms that are mentioned but never incorporated into their proposed explanation for duct coiling such as convergent extension and actomyosin based basal shrinkage. Proliferation is discussed quite a bit but seems to be dismissed as a force. In the introduction they mention how Shh mediated proliferation is required for duct elongation while Fgf10 null mutants have a shortened duct yet normal proliferation. So what is the role for proliferation? Maybe they can answer this in the context of their interesting observation that there is more proliferation in the roof than the floor which would be predicted to bend the cochlea along that axis. When combined with the medial lateral bending could these two forces result in the spiraling? It also seems like this differential proliferation between the floor and roof was in more than just the epithelium, correct? Could the cartilaginous capsule around the duct guide the bending as well? In their culture experiments, if too much of the capsule was removed then normal duct development was disrupted.

As you point out, the previous manuscript included fewer arguments of active force generation despite the fact that it should be a key link between cellular processes and tissue morphogenesis. In the revised manuscript, we improved this point by supplementing new data on cellular force generation.

Regarding convergent extension, we manually tracked each cell movement in 3D and examined the trace lines of neighboring cell clusters. As a result, we were not able to find any evidence that mediolateral convergent extension occurs during stages E12.5 to E14.5. This result is in accordance with previous studies demonstrating the existence of convergent extension only after E14.5 as described in the Introduction. We added this to Figure 5—figure supplement 3. However, we were unable to perform the complete automatic 3D single-cell tracking as described in the reply to your comment #4, meaning that we cannot completely deny the existence of convergent extension in our observation. We included this concern in the Discussion as a possible extension of this study.

Meanwhile, we have argued that actomyosin-based shrinkage plays a central role in cell migration. Thanks to your comments, we have added a new section and dataset, showing the active shrinkage during cell migration from the base to the apex in Figure 6. We believe that this provides further understanding of the mechano-chemical feedback system of ERK activation and cell flow during the development of murine cochlear ducts.

This reviewer gives a thought-provoking comment about spiral formation. We consider that cell proliferation underlies global tissue growth through volumetric local growth but is not the sole factor that produces mechanical forces underlying the asymmetry along both the mediolateral axis and the roof-floor axis. Rather, we have argued that the “hotspot” where the remarkable cell proliferation occurs in the basal region plays a role as a source of material supply to the growing apex side of the cochlear duct as mentioned in the text.

We consider that the cartilaginous capsule may play an essential role in cochlear morphogenesis, especially as a physical restriction to avoid the duct growing outward. However, in this case, a detailed balance of overall tissue growth between the epithelium and the capsule is essential, and elucidation of this mechanism is beyond the scope of this paper. In addition, we do not have data that strongly support that the capsule controls the spiral formation. Therefore, we refrain from pursuing further on this point, and included these comments in the Discussion section to provide our idea to the readers.

2) Their demonstration that the bending forces are in the medial half is interesting but the only tissue whose mechanism is studied is the MEL. Could convergent extension in other medial tissues such as the prosensory domain (which Wang et al., showed was occurring in this tissue) and surrounding mesenchyme be the main force generator for the bending of the medial half of the cochlear duct? Does the MEL cultured by itself bend? They say that cell intercalation can drive ductal elongation but not bending (Introduction) but can't convergent extension occur asymmetrically in the tissue? Such as by occurring in the overlying medial mesenchyme but not in the medial epithelium. It should be noted that the bending by the epithelium does not have to provide high forces as long as the force provided by other tissues are similar across the medial lateral axis, the bending in the epithelium could bias the mass of tissue to bend.

We attempted to isolate the medial half of the cochlear duct from E12.5 to E14.5 but failed to properly dissect the tissues via manual surgery. We also tried to determine the dynamics of surrounding mesenchymal cells, but there appeared to be an insufficient number of mesenchymal cells to be perceived as group cell behaviors. As far as our observations, we could not see the asymmetric cell intercalation as reported in the development of Drosophila’s genital disc (Sato et al., 2015). Again, we have no positive evidence to support the existence of convergent extension in the cochlear epithelium as replied to your comment #1. Additionally, there are no reports on asymmetric convergent extension of the overlying medial mesenchyme as far as we know. As discussed above, we only mentioned this issue in the Discussion section.

3) The mathematical modeling for the luminal bending is less convincing than the mathematical modeling for the ERK and Cell flow coupling. The simulated curves in Figure 2K are quite different from the Experimental measure in Figure 2M, especially for the Mitomycin C condition. I feel that the values plugged in for the Numerical simulation, the standard parameter set were not well justified. What happened to the simulations as these values changed? Was the parameter space for acceptable values broad? In contrast the parameters for the numerical simulation of the ERK activation waves and cell flows were well justified. The parameters chosen might explain the big differences between simulation and experimental in Figure 2.

Thank you for your constructive suggestions. Although we have already chosen the standard set of parameter values under which the simulated results satisfactorily mimic the experimental observations, we did not show the degree to which the values were justified as you point out. According to your suggestions, we demonstrated the similarity of the simulated results to the experimental observation by introducing the root-mean-square error (RMSE), which measures the differences in datasets between simulations and experiments. As a result, the RMSE at the standard parameter set showed a minimal value and small variance, which indicates that the parameter values we used are plausible for the conditions of numerical simulation. These results are shown in Figure 2—figure supplement 1B.

We have also been concerned with the quantitative difference in the curves between experiments and simulations for the Mitomycin C condition. Because all parameters are determined either by measurement or fitting for the control condition, we consider that this difference largely stems from the boundary conditions of the model. In the simulations, the edges of cells at the tissue boundary were set to be free in the finite window size, which was determined as a practical condition of the numerical investigation. On the other hand, in a real experimental situation, the boundary cells are constrained by the other cells outside the corresponding window. We admit that this point is important to the reader. We have included these statements in the Materials and methods section.

4) For the cell tracking experiments in the lateral region the resolution was 4-5 cells. The resulting cell flow patterns were very interesting but why didn't the authors track single cells? Segmenting individual cells via cytoplasmic labeling is much trickier but the nuclei are identifiable and the Imaris software they used in the paper has a cell tracking feature for such labeling. I would think that individual cell movements might provide more insights. In subsection “Retrograde helical ERK activation waves drive base-to-apex multicellular flow” they say they can see cell contractions which I assume is for individual cells? How were cell contractions identified? Video 5 was excellent and very informative. Do the cell flows correlate at all with the proliferation seen with Edu staining?

We worked on single-cell tracking from nuclei reporter images (ubiquitous H2B-mCherry expression) but found that it failed to acquire reliable tracking data mainly due to weak signals and densely packed nuclei, even using Imaris software. As you suggest, we realize that singlecell tracking should provide more insights despite the challenge, and included this comment in the Discussion section.

In accordance with the comment regarding cell contraction, we added a new section and dataset for an inhibitor treatment of myosin activity and local tissue deformation owing to cell contraction as shown in Figure 6. We believe that this supplementation strengthens the validity of our claim. We appreciate your constructive suggestion.

As for the correlation between the cell flows and proliferation, we agree that it will provide further profound information to the readers. The corresponding movie is indeed informative, but it is almost impossible to recognize cell proliferation. In addition, because it is not easy to link static EdU images with movies, we refrain from adding results in the text.

Reviewer #3:

1) The authors argue that cell cycle arrest results in a decrease in the curvature of the cochlear duct, which supports the hypothesis that luminal nuclear stalling promotes MEL bending. This is fine, but luminal nuclear stalling can be a result and not a cause. Since in a bent region, the basal side is more packed, this density gradient can be the cause of nuclei stalling at the luminal side. The fact that the curvature decreased but not diminished after cell cycle arrest could suggest that nuclear stalling is not required for bending, but rather reinforces it.

We like this idea, and in fact, we mentioned the possibility that luminal nuclear stalling can be a result in the previous manuscript: “…while it remains unclear whether luminal nuclear stalling results from the MEL curvature, that is, the convexity of the luminal side”. It might be difficult to completely diminish the curvature because the epithelial tissue has hysteresis to some extent, even with cell cycle arrest. Luminal nuclear stalling promotes MEL bending, and the nuclear density gradient along the luminal-basal axis can cause luminal nuclei stalling. Therefore, we consider that there would be feedback in the nuclear movement and tissue geometry. We agree with your idea, and thus supplemented some sentences in the Discussion section.

2) Since the authors discuss both cell proliferation and nuclear stalling, and cell migration, as forces that can drive bending and coiling, it hard to interpret the results of the mitomycin C experiment. Could it be that the tissue is less curved because there are less cells to supply the elongation tissue rather than less nuclear stalling? The authors should consider inhibiting either cell migration or the cytoskeletal machinery required for IKNM to dissect these effects.

We admit that blocking cell proliferation is a rough experiment and is weak to distinguish which cellular processes are critical to morphogenetic mode at the tissue scale. We carefully considered other alternatives among our current experimental techniques, but unfortunately, we had no elegant method to suppress nuclear movement in the MEL, cell proliferation in the specific region, or cell migration from the base to the apex. We apologize for not being able to experimentally respond to your comment. Please understand that this is a very challenging experiment in our current situation. We have softened the claim in the section on nuclear behaviors and added a new paragraph concerning the methodological limitations of this study in the Discussion section.

3) The authors present a mathematical model to demonstrate that nuclear stalling in the luminal side results in bending. To model nuclear motion they use a parameter, γ, which controls the degree of basalward movement after IKNM. Modeling in such way means that other than γ=1, the nucleus never fully returns to the basal side, but if I understand correctly this is not the case, as even if the nuclei that stall at the luminal side, eventually return to the basal side.

Thank you for your careful reading. We set γ=0.9 as the standard parameter value in the numerical simulation by considering the size of the nucleus and based on observation. This means that the nuclei fully return to the basal side according to the cell cycle length in the simulation, which occurs in the experiment only in the uncurved region of the MEL. However, as shown in Figure 2D’, we never observe nuclear return to the edge of the basal side in the curved region within our observation period. To clarify this point, we added the above sentence in the Materials and method section.

4) Furthermore, for luminal nuclear stalling, the authors tracked only the nuclei of dividing cells. This makes the data in Figure 2D' much clearer. However, in their model the authors show only these nuclei and not all nuclei. In addition, they show many crowded nuclei in the model, yet this is not observed in the images provided in the manuscript. Therefore, it seems the model does not represent the morphology of the tissue properly. The authors should model the process with non-dividing cells at the basal side.

This comment is important for considering the use of the mathematical model. It should be emphasized that we do not aim to incorporate the physical situation precisely but rather utilized it to demonstrate the concept in some ideal situations. In any case, mathematical modeling is an abstraction of reality, and we hope you will agree with this. Although our model does not include detailed tissue architectures and physical properties, it does include some experimentally acquired parameters, such as the dimensions of the cell. Hence, we are convinced that the model captures the essential aspects of this complex process and is useful for delivering the concept.

Another possible approach to express individual nuclear dynamics is a particle-based model, which ignores cell shape and focuses only on the center of mass of the nuclei. For this model, it is not easy to link cell dynamics and mechanical integrity of the epithelial tissues with densely packed nuclei–this is a challenge for future studies. We have included these comments in the Materials and methods section. We believe that this response to your comment will be helpful to our readers, especially those interested in the use of mathematical models.

5) In subsection “Spatial heterogeneity of cell proliferation suggests cellular inflow to the lateral side of the apex to realize cochlear bending” the authors claim that the higher volumetric growth measured at the MEL should cause an opposite curvature relative to the innate one. This is true if EdU intensity is proportional to volumetric growth, but cells in the MEL and LEL may not be the same size. For example, cells in the MEL could be smaller than cells in the LEL. The authors should therefore measure the nuclei number density and the volumetric cell density to clarify this. If the number density of the nuclei is indeed higher at the MEL, it may also explain the higher structural integrity of the MEL relative to the LEL demonstrated in Figure 1C.

This is an important issue when considering the relationship between local volumetric growth and tissue morphogenesis. In accordance with this comment, we measured the nuclei density on both the floor and roof sides and found that the nuclei density was not significantly different along the mediolateral axis. We included the data in Figure 3. This observation supports the idea that the EdU intensity distribution is worthy of consideration as a marker of local tissue growth at a fixed moment in time due to cell proliferation. However, EdU intensity measurement is not the sole way to evaluate local tissue growth over time. We included this concern in the Materials and methods section. We would also like to emphasize that the conclusion of the EdU experiments section is not a positive statement to support the main driver of cochlear morphogenesis; rather, we regard this section as a bridge passage.

6) The authors show the effect of ERK inhibition on tissue flow speed. This is a very important observation and raises several important questions. What is effect of ERK inhibition on curvature? On tissue length? On proliferation? These will provide a more complete understanding of the effect of RK inhibition.

Since we have agreed that these comments are critical to understanding the ERK activity of cochlear morphogenesis, we added a new section with a dataset regarding the effect of ERK inhibition on the multiple aspects in Figure 4 of the present manuscript. We believe that these findings provide fundamental information on the long-term effects of ERK inactivation on cochlear duct morphogenesis.

7) The authors should also test the effect of mitomycin C on cell flow and ERK activity. As mentioned above, it is not clear whether the effect of mitomycin C is a result of less nuclear stalling or perhaps less cells that flow towards the apex.

We carefully considered whether our current experimental techniques are truly able to resolve your concerns. Mitomycin C treatment certainly results in both less nuclear stalling and less cell flow towards the apex, as we argue that cell flow is driven by the heterogeneous cell proliferation profile. Unfortunately, this experiment is the best possible option for us at the present time. We have accepted the weakness of the mitomycin C experiments, and this concern was included in the Discussion section.

8) In Figure 3 the authors analyze the EdU distribution over the cochlear duct. This analysis is done using the maximum intensity projection of the stack. It seems that a more accurate way to quantify would be to use the summed intensity image rather than the maximum intensity image. This may reveal additional details that were missed by throwing away all other layers except the one at maximum intensity.

In response to this comment, we made three different versions of the projection data: (1) maximum intensity projection, (2) summed intensity projection, and (3) mean intensity projection averaged within the cochlear duct. All processed data for the three experiments were added to Figure 3—figure supplement 1. As a result, we found no qualitative differences among different projections. Since we are not sure whether the summed intensity projection is better than the maximum intensity projection and the readers can refer to the different versions in the supplementary figure, we maintained the previous version in the current manuscript aside from the mapped graphs, which were replaced with the sample mean data over three samples instead of the single data.

9) In Figure 4 the colors used for the ERK activity analysis are very hard to see for color-blind people. It would be easier for this audience if the authors changed one of these colors to green/red/yellow.

This is an important comment for a wide range of readers. We believe that the current coloring is the best for effectively illustrating the spatio-temporal patterns of ERK activity to the readers. However, we admit that it is difficult to see some figures, especially the change of color in Figure. Therefore, we included another version for color-blind people below the corresponding figures.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

All three reviewers have judged that your manuscript is of interest and represents an advancement to the field. However, the manuscript cannot be accepted at its current form because of the issue whether IKNM described is causal or consequential to cochlear bending. This issue needs to be better resolved. Either the mathematical model is revised to accommodate for fewer cells stalling at the luminal side but could still account for the cochlear bending or more cells can be observed to stall at the luminal side that could account for the cochlear bending. Alternatively, the entire manuscript can be totally revised to accommodate for these two possibilities in an unbiased manner, starting with the Title.

We would like to thank the reviewers for their thoughtful responses and efforts toward improving our manuscript. We have responded to all concerns and revised the manuscript according to the suggestions. As described in the cover letter, we agree with the reviewers’ suggestions and totally changed the whole structure of our manuscript. In the revised version, the part of apical nuclear stalling was removed and the ERK activation wave becomes at the center. With this, we changed the Title as “Retrograde ERK activation waves drive base-to-apex multicellular flow in murine cochlear duct morphogenesis”. We believe that this change makes the main point of the paper clearer, and the revised manuscript meets with your approval.

Since many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Specific comments from two of the reviewers are listed below as guidelines for your revision:

Reviewer #2:

The authors have addressed most of my comments. The experiments with the MEK inhibitor PD0325901 are not totally convincing. I worry about nonspecific effects of this pharmacological reagent. Another inhibitor of cell proliferation like a Cochlea from a KO mouse would provide a second piece of evidence. How did the FGF inhibitor SU5402 effect curvature? In Figure 4C, I was not sure why do the measurements stop at 0.8 mm? It seems like the trend was going down which would make it not significant?

We appreciate your suggestion. We performed pharmacological assays using the SU5402 and found that the cochlear duct length became shorter and the curvature profile became overall larger compared with the control as observed in the PD0325901 treatments (Figure 2A-D). This reviewer pointed out the difference of measurement range between the control group and the inhibitor treatment group. This comes from the fact that the cochlear duct length was significantly shortened by the inhibitor treatments. In the current version, we adopted the curvature profile along the relative arc length normalized by the total arc length as the main figure (Figure 2C) because the size of cochlear duct became each treatment. We believe that this graph representation would be easier for readers to evaluate. We included these statements in the Introduction and Figure 2.

In the Discussion the authors state they "provided the first experimental evidence that nuclei stall at the luminal side of the pseudostratified epithelium during IKNM in normal development". I go back to the discrepancy between the simulation and experimental results in Figure 2J-M. They address this in the Materials and methods but some mention of it here would be appropriate.

In response to the previous reviewer comments, we deleted the whole part of apical nuclear stalling because we admit that the experimental evidences on apical nuclear stalling in the medial side of cochlear duct are rather weak to support it as a main driving factor for the duct bending. With this, this comment is not applicable anymore. We thank for your efforts on this comment.

Reviewer #3:

The authors address many of the points raised in my review and the other reviewers. The section on the ERK waves and the mechanochemical feedback has improved considerably and is actually very nice. In particular the new data on the effect of blebbistatin is very nice and supports the model.

I do, however, have some problem with the first part of the manuscript on the interkinetic nuclear movement (IKNM). The main problem here is that the model they suggest ignore an essential aspect of the system which is that only some small fraction of the cells perform the luminal stalling at a given time. The model suggests that the cause for bending is that luminal stalling leads to an inverted wedge like morphology of the cells and thus leads to bending. However, in the model, luminal stalling is assumed to happen in all the cells. This does not seem to match the images in Figure A-C showing that most nuclei are closer to basal position. Hence, to model that the authors should have assumed that there are only few cells whose nuclei are stalled at the laminal position. It seems to me, that it is unlikely that such a model would produce significant bending in that case. In their response, the authors argue that their model like all mathematical models is a simplification of the real system. While it is true that models always simplify, it seems to me that the assumption that all cells stall is an essential divergence from the real system and cannot be simplified.

As I suggested previously, the nuclear stalling may actually be a result of bending and not the cause. I understand that this is hard to test experimentally. However, I feel that the current model maybe misleading.

Since I like the second part and find it interesting and more convincing, it may be worth for the authors to restructure their manuscript so that the ERK waves are at the beginning and are the main focus. The IKNM is an interesting observation that can be added with the two potential interpretations.

In response to your comments, we determined to change the whole structure of the manuscript. We admit that the experimental evidences on apical nuclear stalling in the medial side of cochlear duct are rather weak to support it as a main driving factor for the duct bending. We also agreed your idea that ERK waves should be the main focus in the revised manuscript. We believe the updated manuscript is more convincing and meets your criteria.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The spiral shape of the mammalian cochlear duct is tightly coupled to its function in sound detection. However, the morphogenetic process that gives rise to the spiral structure is poorly understood. In this paper, by live-imaging of the developing cochlea from a FRET-based reporter mouse strain, Hiroshima et al. demonstrated elegantly that one of the mechanisms in cochlear duct elongation is driven by an oscillatory wave of ERK activity originating from the roof of the cochlear apex towards the base and a concomitant flow of cells from the base of the cochlear floor towards the roof of the apex.

All three reviewers felt the revised manuscript is much improved and concise by removing nuclear stalling data. The only suggestion is for you to consider a 3D schematic summary that illustrates direction of ERK oscillation, cell movement flow and the hot spot of cell proliferation in the floor of the cochlea. This will help readers outside of the ear field to appreciate your beautiful work.

Thank you for your suggestion. We modified Figure 3L, which will be helpful for readers.

Second, the citation of Bok et al., for reduction of cell proliferation at the base of the cochlear duct is incorrect. It is ok to state that cochlear duct growth requires Shh signaling but it is incorrect to say there is significant decrease in cell proliferation at the base of the cochlea in the Shh conditional mutants. First, only EdU-labeled hair cells were quantified in that paper, which could hardly account for the shortened cochlear duct phenotype. More importantly, those experiments were not designed to look at proliferation but cell cycle exit. They were conducted by injecting Edu at E13.5 and E14.5 and EdU-labeled hair cells were analyzed at birth. Heavily labeled Edu-positive cells at birth were cells that exited cell cycle shortly after EdU administration. Therefore, a reduced EdU labeling could mean there were more cells dividing at the base at E13.5 and E14.5 rather than less proliferation and the label were diluted and no longer detected by birth.

We revised as follows according to Discussion of the referenced paper:

“Deletion of Shh expression leads to a shortening of the cochlear duct, and it was proposed that the SHH signaling promotes growth of the cochlear duct mainly in the base region (Bok et al., 2013).”.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Figure 1—source data 1. Curvature and thickness from E12.5 to E14.5.
    Figure 1—source data 2. EdU intensity profiles.
    Figure 1—source data 3. Cell density along the medial–lateral axis.
    Figure 2—source data 1. Longitudinal length for treatments with DMSO, PD0325901, and SU5402.
    Figure 2—source data 2. Curvature over arc length from the tip for treatments with DMSO, PD0325901, and SU5402.
    Figure 3—source data 1. Extracellular signal-regulated kinase activity rate and extension-shrinkage rate.
    Figure 3—source data 2. Particle image velocimetry speed before and after the treatments with PD0325901, SU5402, and cyclopamine.
    Figure 4—source data 1. Particle image velocimetry speed before and after the treatments with blebbistatin.
    Figure 4—source data 2. Extracellular signal-regulated kinase activity and tissue curvature.
    Transparent reporting form

    Data Availability Statement

    All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files are provided.


    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES