A tissue boundary orchestrates the segregation of inner ear sensory organs

The inner ear contains distinct sensory organs, produced sequentially by segregation from a large sensory-competent domain in the developing otic vesicle. To understand the mechanistic basis of this process, we investigated the changes in prosensory cell patterning, proliferation and character during the segregation of some of the vestibular organs in the mouse and chicken otic vesicle. We discovered a specialized boundary domain, located at the interface of segregating organs. It is composed of prosensory cells that gradually enlarge, elongate and are ultimately diverted from a prosensory fate. Strikingly, the boundary cells align their apical borders and constrict basally at the interface of cells expressing or not the Lmx1a transcription factor, an orthologue of drosophila Apterous. The boundary domain is absent in Lmx1a-deficient mice, which exhibit defects in sensory organ segregation, and is disrupted by the inhibition of ROCK-dependent actomyosin contractility. Altogether, our results suggest that actomyosin-dependent tissue boundaries ensure the proper separation of inner ear sensory organs and uncover striking homologies between this process and the compartmentalization of the drosophila wing disc by lineage-restricted boundaries.


INTRODUCTION 1 2
The inner ear of vertebrates is composed of a series of liquid-filled chambers containing distinct 3 sensory organs. Each organ contains a sensory epithelium populated with mechanosensory 'hair 4 cells', separated from one another by supporting cells. In amniotes, the ventral part of the inner ear 5 forms the cochlea, which hosts an auditory epithelium called the basilar papilla in birds and 6 crocodiles, or the organ of Corti in mammals. The dorsal part forms the vestibular system, with three 7 cristae (anterior, posterior and lateral) and their associated semi-circular canals responding to the 8 angular rotations of the head, and the maculae of the utricle and saccule, sensitive to gravity and 9 linear acceleration. These sensory organs are separated from one another by non-sensory territories 10 with essential roles in inner ear function and homeostasis (Ekdale, 2016).

11
The development of the inner ear is a remarkable example of 3D tissue morphogenesis that starts 12 with the formation of the otic placode, an epithelial thickening located on both sides of the hindbrain.

13
The placode rapidly invaginates into the underlying mesenchyme to form the otic cup. Subsequently,

28
One aspect of inner ear development that results in the spatial separation of adjacent groups of cells 29 is the formation of its sensory organs. In the chicken or mouse otic vesicle, the prospective sensory 30 domains, or prosensory domains, are recognizable as thickened epithelial patches in the ventro-31 medial and posterior regions (Knowlton, 1967). The prosensory patches express the Notch ligand 32 Jag1, which promotes the maintenance of the prosensory fate by Notch-mediated lateral induction 33 (reviewed in Daudet and Żak, 2020;Kiernan, 2013). They also express the High-Mobility Group 34 transcription factor Sox2, which is essential for the formation of all inner ear sensory organs . This segregation process was first described in the 5 inner ear of fish and amphibians more than a century ago (Norris, 1892) and could have played an 6 essential role in the evolution of the vertebrate inner ear, by allowing new sensory organs to acquire 7 specific functions (reviewed in Fritzsch et al., 2002). Genetic studies in the mouse have shown that 8 mutations or absence of specific transcription factors and signalling molecules can lead to defects 9 in sensory organ segregation (reviewed in Alsina and Whitfield, 2017). One such factor is Lmx1a 10 (cLmx1b in the chicken), a LIM-homeodomain transcription factor whose orthologue Apterous acts 11 as a "selector gene" for the dorsal compartment of the Drosophila wing disc [ref]. In the mouse inner 12 ear, Lmx1a is expressed in the non-sensory territories separating the sensory organs, and its

21
To tackle this question, we examined the changes in cell morphology during the segregation of the 22 lateral (LC) and anterior (AC) cristae from the pan-sensory domain in the embryonic chicken and 23 mouse otocyst. We identified a population of prosensory cells with enlarged surfaces, which appear 24 at the interface of segregating organs and gradually elongate and re-align along the border of the 25 prospective cristae before down-regulating Sox2 expression. As segregation proceeds, these large 26 cells constrict basally at the interface of Lmx1a-positive (on the cristae side) and negative (on the 27 pan-sensory domain side) cells, strongly suggesting the formation of a lineage-restricted tissue 28 boundary. In Lmx1a-null mouse otocysts, the boundary domain is disrupted; the AC and the LC 29 initiate their segregation from the utricle but they remain fused to one another, indicating that Lmx1a 30 is required for the differentiation of the non-sensory cells separating the two cristae. Finally, we show 31 that the perturbation of ROCK-dependent actomyosin contractility, a common player in the 32 establishment of lineage-restricted boundaries, leads to defects in the organisation of the boundary 33 domain and abnormal sensory organ segregation.

34
Altogether, our results strongly suggest that a lineage-restricted tissue boundary forms at the 35 interface of segregating sensory organs, providing support to the compartment-boundary model of 36 inner ear development. They also uncover striking similarities in the genetic circuits and cellular 37 mechanisms of tissue segregation in the inner ear and the Drosophila wing disc.

A specialized boundary domain with a basal constriction forms between 2 segregating vestibular organs in the chicken otocyst 3
We focused in this study on the formation of the anterior (AC) and lateral (LC) cristae, using whole-4 mount preparations of the otic vesicle immunostained for the prosensory marker Sox2. In the chicken 5 inner ear, the AC is the first to segregate from the large anterior pan-sensory domain between stage 6 HH23-25 (E4-E5), followed by the LC between HH25-27 (E4.5-E5.5) (Figure 1a-c; see also Mann 7 et al., 2017). When we examined at high magnification phalloidin-stained samples, we noticed that 8 the cells located at the interface of the cristae and the pan-sensory domain had a larger apical 9 surface than those of surrounding territories.

10
Focusing on the LC, which could be identified from the earliest stages of its formation, we

28
Concomitant to these changes in cell surface morphology, we noticed a progressive enrichment in 29 F-actin staining at the base of the epithelium along the presumptive border of the LC (Figure 2). At

Lmx1a-negative cells in the mouse otocyst 17
To find out if a comparable interface domain forms in the mammalian inner ear, we analysed the 18 otocysts of Lmx1a GFP knock-in mouse embryos. In a previous study, we showed that the inner ear of             Altogether, these data indicate that the cellular correlates of cristae segregation are comparable in 1 the mouse and chicken inner ear, despite some differences in its dynamics and outcomes. In both 2 species, a specialized population of boundary cells with enlarged apical surfaces appears at the 3 interface of the two cristae and the utricle at early stages of their segregation. Furthermore, a basal 4 constriction and alignment of apical cell borders occur in the middle of the large cell domain, at the 5 precise interface of Lmx1a-expressing and non-expressing cells. This strongly suggests that a 6 lineage-restricted tissue boundary separates these two cell populations.

7
The boundary domain is disrupted in Lmx1a-null mice 8 The complete absence of Lmx1a function leads to severe defects in inner ear morphogenesis,     territories between the AC and LC and the maintenance of the boundary domain between the cristae 2 and the utricle. This is consistent with the notion that Lmx1a acts as a "selector" for the non-sensory 3 fate in the domain separating the two cristae. Lmx1a function is also required for the formation of the 4 large cell domain and its associated basal constriction at the limit of the Lmx1a-expressing domain, 5 but neither appear necessary for the initial separation of the utricle from the cristae.

2
We next compared pairs of contralateral HH26 otocysts cultured for 24h in either control medium or 3 20m Y-27631 (Figure 6m-r). Only three (out of Y-27631-treated samples could be analysed, due

9
EGFP localizes at the cell membrane (Figure 7a). The analysis of surface cell morphologies with 10 Epitool (Figure 7b-d) showed that RIIC1-EGFP causes a mild (approximately 23%) but statistically

18
EGFP proviral DNA and the embryos were incubated for 3 to 5 days post-EP. In controls (n=8; Figure   19 7e-f''), the LC was present in 6 out of 8 samples and separated from the utricle by distances ranging 20 from 55 to 155 m. We hypothesize that the absence of LC in two of the controls may be due to

17
We conclude that ROCK-dependent actomyosin contractility is required for the coordination of

A tissue boundary forms during the segregation of inner ear sensory organs 19
The morphogenesis of complex tissues depends on the harmonious coordination of cell proliferation,     14 Which mechanisms establish and maintain the boundary domain between segregating sensory 15 organs? In a previous study, we showed that Lmx1a is essential for the specification of the non-16 sensory territories separating sensory organs and its overexpression in the chicken inner ear 17 antagonizes prosensory specification, suggesting that it acts as a "selector" gene for the non-sensory               with tape and returned to incubation for 2 hours, then the embryos heads were collected and fixed for 30 min at room temperature in PBS with 4% Formaldehyde. Following several rinses in PBS and 1 PBT, the inner ear tissue was dissected and processed for Click-iT EdU reaction (Thermofisher UK) 2 according to the manufacturer's protocol. At the end of the reaction, the tissue was further processed 3 for Sox2 immunostaining as described above and labelled with DAPI before mounting. For the 4 mapping of mitotic figures (identified by DAPI staining), a 72x36m "region-of-interest" centred on 5 the boundary between the lateral crista and pan-sensory domain was selected from a confocal stack