Developmental GAD2 Expression Reveals Progenitor-like Cells with Calcium Waves in Mammalian Crista Ampullaris.

Sense of motion, spatial orientation, and balance in vertebrates relies on sensory hair cells in the inner ear vestibular system. Vestibular supporting cells can regenerate hair cells that are lost from aging, ototoxicity, and trauma, although not all factors or specific cell types are known. Here we report a population of GAD2-positive cells in the mouse crista ampullaris and trace GAD2 progenitor-like cells that express pluripotent transcription factors SOX2, PROX1, and CTBP2. GAD2 progenitor-like cells organize into rosettes around a central branched structure in the eminentia cruciatum (EC) herein named the EC plexus. GCaMP5G calcium indicator shows spontaneous and acetylcholine-evoked whole-cell calcium waves in neonatal and adult mice. We present a hypothetical model that outlines the lineage and potential regenerative capacity of GAD2 cells in the mammalian vestibular neuroepithelium.

Introduction

Vestibular organs comprise one of the smallest and most complex sensory systems in gnathostomes orchestrating neural inputs from five organs, including three semicircular canals that provide a sense of linear and angular motion in three-dimensional space. Mechanotransduction takes place in specialized stereocilia projecting from the apical face of type I and type II sensory hair cells (HCs). The HCs in anterior and posterior semicircular canals sense rotation in the vertical plane. These cristae have a unique anatomical region in the center of their crista called the eminentia cruciatum (EC), which is absent in the horizontal canal in most gnathostomes and not observed in adult primates. Electron microscopy studies described the EC in the anterior and posterior canals cotaining cells without cilia across several species including fish, turtles, birds, and mice (Igarashi and Yoshinobu, 1966; Igarashi and Alford, 1969; Harada, 1972, 1983; Collazo et al., 2005; Chagnaud et al., 2017). Lack of an anatomically distinct EC in primates and in the horizontal canals of other species resulted in a limited number of studies, leaving knowledge about cells located in the EC, their regulation, and physiological function(s) unknown.

In the crista and EC, cells lacking stereocilia are grouped as nonsensory supporting cells (SCs). When the crista is damaged, SCs begin to divide, which can lead to differentiation into either new HCs or new SCs in birds, fish, and mammals (Corwin and Cotanche, 1988; Ryals and Rubel, 1988; Forge et al., 1993; Warchol et al., 1993;Bermingham-McDonogh and Rubel, 2003; Brignull et al., 2009; Jiang et al., 2014; Cruz et al., 2015; Scheffer et al., 2015; Kniss et al., 2016; Slowik and Bermingham-McDonogh, 2016; Burns and Stone, 2017; Lush and Piotrowski, 2014, 2019; Wan et al., 2020). It is not known whether a specific population of SCs is responsible for regenerating HCs, SCs, or both in mouse cristae.

There is evidence that glutamic acid decarboxylase (GAD) is expressed in the SCs of mammalian cristae, although the specific physiological role remains uncertain. In a GAD67 transgenic mouse, SCs expressing GAD67-GFP were located primarily in the peripheral zone (PZ) of cristae (Tavazzani et al., 2014). In the PZ, afferent neurons predominantly make direct synaptic contacts with type II HCs via boutons instead of calyceal synapses with type I HCs in the central zone (CZ; Lindeman, 1969; Lim, 1976; Baird et al., 1988; Lysakowski and Goldberg, 1997; Fernández et al., 1998, 1995; Desai et al., 2005a), but it is not known what role SCs may play in synaptic transmission between HCs and afferent terminals or how GAD might be involved. Two GAD isoforms are responsible for synthesizing GABA, namely, GAD67 and GAD65, which are expressed from genes GAD1 and GAD2, respectively. In the central nervous system (CNS), GAD67 synthesizes GABA primarily for synaptogenesis, whereas GAD65 synthesizes GABA for neurotransmission (Kanaani et al., 1999). GAD67 is expressed during early development, whereas GAD65 is expressed at later ages, reflecting functional differences in each GAD isoform in the CNS (Esclapez et al., 1994; Bowers et al., 1998). These findings motivated analysis of GAD-positive SCs in the crista during postnatal development.

We examined vestibular neuroepithelium from the first filial generation of GAD2-IRES-Cre crossed with a dual reporter GCaMP5G-tdTomato line (Holman et al., 2019). In this study, GAD2-tdTomato (GAD2-tdT) cells were observed in the EC and specific zones within vertical cristae . Two unique GAD2-tdT EC cell types were identified based on their location within the EC and their ACh-evoked Ca2+ transients. During early postnatal development GAD2-tdT EC cells initially form mosaics that eventually organize into rosettes around an EC plexus, a core structure with extending branches in the middle of the EC. We report genetic and biochemical evidence for EC GAD2 progenitor-like cells with acetylcholine- and muscarine-evoked calcium waves reversibly blocked by atropine and 2-aminoethoxydiphenyl borate (2-APB) during postnatal development.

Results

Tracking GAD2-tdT Cells in the Crista and EC

We hypothesized that a GAD2-IRES-Cre driver that targets astrocytes and GABAergic neurons in the CNS would similarly target GABAergic SCs in the semicircular canal cristae. GAD2-IRES-Cre driver was crossed with a dual GCaMP5G-tdTomato reporter line and whole mount cristae were examined. At day of birth, the earliest age examined in this study, GAD2-tdTomato (GAD2-tdT) cells were observed throughout the neuroepithelium (P0, Figure 1A). Illustrations in Figure 1B represent the optical imaging plane and age corresponding to the confocal images shown in Figure 1A. Representative whole-mount images of GAD2-tdT cells in cristae from ages P0 to P318 are shown (Figures 1A and 1C–1G). Immunolabeling with the HC marker MyoVIIa shows the location of developing HCs and a subpopulation of HCs with GAD2-dT in the first postnatal week. We observed an absence of MyoVIIa immunolabeling in the EC during development (Figure 1A) and in adults (Figures 1C and 1E–1G), further demonstrating a lack of HCs in the EC.

Figure 1

GAD2 Cells in EC and Crista

(A) Confocal microscopy of whole-mount fixed anterior canal cristae from transgenic mice (PC::G5-tdT), in the first postnatal week (A); day of birth (P0, k = 6), postnatal day 2 (P2, k = 3), and P4 (k = 5) with a GAD2 cell population (red) throughout the crista and hair cells (HCs) with MyoVIIa (white). Centrally located EC GAD2-tdT cells do not label with MyoVIIa (red; open arrowhead, clino2 cell; closed arrowheads, clinocytes).

(B) Illustrations are provided to orientate the plane in the corresponding confocal images.

(C) Phase contrast image overlaid with fluorescent confocal maximum intensity projection (MIP) gives relative position of GAD2-tdT cells throughout the crista and EC at postnatal day 12 (P12-14, k = 6). (i) Digitally zoomed ROI of the EC with an individual clino2 cell and clinocytes.

(D) Cell morphology and model rendering of a clino2 cell and two clinocytes (cts).

(E and F) The clino2 cell (open arrow) and cts (closed arrows) maintain their relative positions within the EC in adult mice (P97–P318, k = 12).

(G) A rotated view from (F) shows a clino2 cell (open arrow) and clinocyte (closed arrow) with surrounding HCs (cyan; MyoVIIa), and three HCs with GAD2-tdT (∗).

(H) Average size of GAD2-tdT cells at different ages based ion segmentation with same relative size. Data are represented as mean ± SEM; ∗p < 0.05.

Two GAD2-tdT cell types with distinct locations are present in the center and slope (i.e., clino) of the EC and referred to as clinocytes and clino2 cells, respectively (arrows, Figures 1A, 1Ci, and 1D; Table 1). GAD2-tdT cells with similar morphologies located more internally within the EC are referred to as clinocytes (cts; Figures 1A, 1Ci, and 1D, filled arrows), whereas GAD2-tdT cells located on the edge of the EC with consistent and highly expressed tdT are referred to as clino2 cells (open arrow). During the second postnatal week GAD2-tdT EC cell boundaries are better defined (Figures 1C–F and 1G). In Figure 1C, phase contrast and fluorescence images are overlaid to visualize the size, shape, and position of the clino2 cell, clinocytes, and HCs in an anterior canal EC relative to the crista. Three-dimensionally reconstructed surfaces of the clino2 cell and clinocytes are shown (Figure 1D). Surface renderings of the clino2 cells and clinocytes provide additional volume and cell morphology analysis (see Methods). Spatial relationships and volume ratio of the clino2 cell and clinocytes remain the same relative to the crista from neonate to adult ages (Figures 1D and 1H). In a mature mouse (P318), a subpopulation of GAD2-tdT HCs (red) persists among the HC population (cyan) in the crista with clinocytes and clino2 cells in the EC (Figure 1F). Figure 1G is a rotated 90° view of the EC visualizing the topology between a clino2 cell (open arrow) and clinocyte in the EC (closed arrow) relative to the GAD2-tdT HCs in the adjacent CZ (Figure 1G).

Table 1

Summary of GAD2 Progenitor-like Clinocyte and Clino2 Cell Features

	Cell Description	Antigenic Phenotype
clinocyte	GAD2 progenitor-like cell, which organizes into rosettes in the eminentia cruciata (EC); located centrally in the EC.	SOX2, PROX1, CTBP2, GABA
clino2 cell	GAD2 progenitor-like cell located on the slope (i.e., clino), and edge of the EC.	SOX2, GABA

Average volumes of GAD2-tdT cells are shown in Figure 1H. Data are represented as mean ± SEM. Clino2 cells have an average volume of 1,408 μm3 (4,180 voxels) (n = 1), clinocytes have an average volume of 1,172 μm3 (3,273 voxels; n = 3), and GAD2-tdT HCs have an average volume size of 998 μm3 (3,701 voxels; n = 4). clino2 cell volume increases proportionately to the crista with age; statistical analysis of P2 to P12 gives p values of p < 0.098 and between ages P318 and P533 p < 0.082. The close cell-to-cell junctions among clinocytes did not allow volume rendering or analysis of those cells.

GAD in the EC

To determine whether GAD2-tdT cells express glutamic acid decarboxylase, GAD65 or GAD67, immunohistochemistry was performed on cristae at different ages. GAD65 is present in the EC and co-labels clinocytes on day of birth (P0; blue, large arrow; Figure 2A). Immunolabeling with GAD65 reveals an additional structure in the middle of the EC, herein named EC plexus (small arrow). GAD67 expression is observed in cells near the EC, but does co-localize with GAD2-tdT cells or GAD65 at this age. By postnatal day 1 (P1), select clino2 cells at the edge of the EC co-label with GAD65 and GAD67 (arrow, Figure 2B). GAD65 immunolabelling regularly recognizes an EC plexus and labels branches extending outward toward clino2 cells and clinocytes.

Figure 2

Immunohistochemistry of GAD65 and GAD67 in the EC

(A) On day of birth, GAD65 (blue) co-labels GAD2-tdT cells (red; small arrow), and a central plexus structure in the EC (large arrow). Immunolabeling of GAD67 (green) occurs in cells on the side of the EC (k = 2).

(B) EC GAD2 cells with star-shaped morphologies along the edge of the EC at P1 label with GAD65 and GAD67. The central EC structure (EC plexus) labels with anti-GAD65 antibody (blue).

(C) At P1, clinocytes begin to form mosaics (k = 2). Clino2 cells (red) are adjacent to the clinocyte mosaic.

(D) At P10, the EC has many GAD2-tdT cells surrounding GAD65 and GAD67 puncta (k = 3). The image is rotated 90° on the x axis for visualizing GAD65, GAD67, and GAD2-tdT within the EC.

(E) Clinocytes (red) organize to form rosettes (i, ii) surrounding GAD65 (blue) and GAD67 (green) puncta at age P10 in the EC; (i) a cartoon depicting a clinocyte rosettes formation with GAD65 and GAD67 puncta in the center.

(F) A month old, a rosette with GAD65 puncta (blue, closed arrowhead) shows predominant GAD2-tdT expression in two opposing clino2 cells (red, open arrowhead) (P27-28, k = 3).

One feature of clinocytes (red) is their ability to form multicellular mosaics with adjacent GAD65 puncta (blue, Figure 2C). In the second postnatal week, GAD65 and GAD67 co-label several branches of the EC plexus (P10, Figure 2D), and GAD65 uniquely labels additional puncta throughout the EC. A rotated view (right) is provided to visualize the extent of branching from the EC plexus (Figure 2D). By postnatal day 10, clinocytes form rosettes consisting of at least six cells with GAD65 and GAD67 puncta in the middle (Figure 2E, box). Due to the three-dimensional structure of the EC, visualization of multiple rosettes and branches from the EC plexus are not apparent (Figure 2D). A deep dive of datasets from multiple whole-mount tissues (z stacks) using image rendering techniques in FluoRender revealed rosettes from multiple cristae (k = 10). At age P27, clinocyte rosettes continued to form with GAD65 puncta in the center (Figure 2F). In this example, prominent tdT expression in two opposing clino2 cells are positioned 180° apart around a clinocyte rosette (cartoon, right panel). These two clino2 cells may represent a different stage in cell cycle and an exit from the rosette.

Pluripotent Transcription Factors and GABA in EC Cells

We examined protein markers in clinocytes to determine what factors may be contributing to the organization of rosettes. Immunohistochemistry with known pluripotent transcription factors SOX2, PROX1, CTBP2, and ATOH1 was performed on whole-mount fixed tissues. Several studies have shown that SOX2 plays a pivotal role in the expansion of progenitor cells and differentiation of HCs during development in cochlear and vestibular organs (Dabdoub et al., 2008; Neves et al., 2013; Kempfle et al., 2016; Puligilla and Kelley, 2017; Atkinson et al., 2018; Yang et al., 2019; Steevens et al., 2019; Wan et al., 2020; Dvorakova et al., 2020). We observed SOX2 expression in clinocytes and clino2 cells in the first postnatal week (P1, Figure 3A).

Figure 3

Clino2 Cells, and Clinocytes SOX2, PROX1, and CTBP2

(A) Immunolabeling of SOX2 in GAD2-tdT cells located in the central and distal regions of the EC shown in a 112-μm confocal z stack MIP (k = 2).

(B) Immunolabeling of PROX1 (green) in clinocytes (red) forming rosettes at age P8 (box; k = 1, n = 3 cristae).

(C) CTBP2 (green) co-localizes with clinocytes (red) forming rosettes (box). ATOH1 (blue) labeling is observed in the EC at P14, but does not colocalize with CTBP2 or GAD2-tdT (P5-14, k = 3).

(D) An ATOH1-Cre Tg with transgenic ATOH1 expression in EC cells (P6, k = 2). A population of EC cells label with GABA (green) in cells forming a rosette (i, box), whereas GAD65 (blue) labels the EC plexus. (i) Zoomed image of GABA-labeled cells forming rosettes; scale bar, 5 μm.

(E and F) (E) GAD2-tdT cells and EC cells not expressing GAD2-tdT label with GABA (green, star) during postnatal development (P6–P8, k = 2; E) and in adult mice (P191, k = 2; F). GABA also labels puncta close to GAD2-tdT cells on the sides of the EC (E, arrowheads) and beneath a clino2 cell in an adult mouse (F).

Another core transcription factor, PROX1, has been shown in SCs and HCs of the developing cochlear epithelium (Bermingham-McDonough and Rubel, 2003; Kirjavainen et al., 2008; Fritzsch et al., 2010; Liu et al., 2018). Clinocytes forming a rosette (box) also express PROX1 during postnatal development (Figure 3B).

At age P14 more rosettes are visible with clinocytes expressing the C-terminal binding protein 2 (CTBP2, green; Figure 3C, box). This multifunctional core CTBP2 protein plays an important role in HC ribbon synapses, whereas it is a transcriptional coregulator in other cell types. We report that CTBP2 is expressed in clinocytes, but not in clino2 cells located along the edges of the EC. This suggests an important transcriptional role for CTBP2 in these cells (arrows; red).

Previous lineage tracing of ATOH1 has demonstrated that it is crucial in the development of SCs and HCs in the organ of Corti, although it is not the only factor in determining HC fate (Fritzsch et al., 2005; Driver et al., 2013). Immunolabeling with an anti-ATOH1 antibody (blue, Figure 3C) showed ATOH1 in the EC at P14, but expression in clino2 cells or clinocytes was not observed. An ATOH1-Cre driver was used as a control to further test ATOH1 expression in the EC. The ATOH1-Cre driver was crossed with the same dual GCaMP5G-tdTomato reporter used in the GAD2-Cre cross. We observed ATOH1-tdT cells in the EC during early postnatal development at P6 (Figure 3D). Without a non-transgenic marker for clinocytes or clino2 cells at this time we are unable to confirm ATOH1 in clinocytes and clino2 cells. Using a polyclonal GABA antibody (A2052, Sigma), we observed GABA (green) in EC cells, including putative ATOH1-tdT HCs (arrow). With a monoclonal GABA antibody (A0310, Sigma), we observed GABA immunolabeling structures adjacent to GAD2-tdT HCs (red), and in putative synapses along the EC plexus (Figure 3E). Immunolabeling with GAD65 and GABA in the same tissue gave distinct patterns that often did not co-label, which may reflect a tight regulation of GAD65 or conformational changes of this enzyme (Kass et al., 2014).

In an adult mouse, GABA was present in a clino2 cell at the apex of the EC and in adjacent cells (P191, Figure 3F; GABA: A2052, Sigma). GABA puncta also appeared beneath the basolateral surface of the clino2 cell toward the EC plexus. A phase contrast overlay panel is provided for orientation of the EC (right).

ACh Evoked Ca2+ Transients in Clino2 Cells and Clinocytes

After characterizing GAD2-tdT expression patterns from neonates to old age mice, we imaged GCaMP5G (G5) to monitor calcium dynamics in these cells. Semi-intact vestibular organs were placed in a bath with continuous perfusion of media and imaged with swept field confocal microscopy (SFC). Puff application of 100 μM ACh and muscarine resulted in large G5 fluorescence modulation (ΔF/F0), whereas no detectable changes in G5 were observed from GABA application (100 μM; data not shown). Therefore, ACh- and muscarine-evoked calcium transients in GAD2-G5::tdT cells remained the focus of this study.

Figures 4A–4C reports the resting G5 fluorescence (F0, blue) merged with peak changes in G5 (ΔF/F0, green). Spontaneous intracellular calcium transients appeared as waves with specific directionality from the apex of the cell closest to the cupula propagating toward the clino2 cell (c) base. Peak changes in GCaMP5G (G5) fluorescence during rest are shown in Figure 4B, ΔF/F0 (green), which were primarily seen in the clino2 cells (c), whereas G5 fluorescence changes in clinocytes were an order of magnitude smaller. A single 500-μs puff of 100 μM ACh evoked Ca2+ bursts in the clino2 cell (Figures 4A–4E) and clinocytes (Figures 4F–4I). These Ca2+ bursts started with a ∼14 s latency and returned to baseline after approximately 90 s (Figures 4D–4J). In both clino2 cells and clinocytes, Ca2+ G5 fluorescence propagated as a wave from the apex to base at a speed of ∼25 m/μs. Intracellular calcium wave speeds, G5 fluorescence intensity modulation, latency to the first ACh-evoked Ca2+ transient, and rate of evoked bursting were plotted for clino2 cells (c) and clinocytes (cts) (Figure 4J; also see Video S1). A significant difference of G5 fluorescence intensity modulation demonstrates that clino2 cells have a larger Ca2+ response compared with clinocytes (ΔF/F: clino2 cell = 0.29 ± 0.052; cts = 0.077 ± 0.0079). Furthermore, the burst rate in clinocytes at this postnatal age is significantly faster than in clino2 cells (rate: clino2 cell = 0.0534 ± 0.00054 s−1; cts = 0.077 ± 0.0079 s−1).

Figure 4

ACh-Evoked Calcium Transients in Clino2 Cells and Clinocytes at P1

(A–C) (A) Peak change in G5 fluorescence (B: ΔF/F0, green) merged with resting G5 fluorescence (C: F0, blue) in the central region of the anterior canal crista.

(D and E) A 500-ms puff of 100 μM ACh triggered intracellular Ca2+ bursts in clino2 cells (c) with a latency of ~14 s and returning to baseline after ~90 s.

(F–I) Amplification of ΔF/F0 (green) revealed smaller ACh-evoked Ca2+ bursts in clinocytes (cts), occurring at faster rates. In both cell types (c and cts), Ca2+ initially increased at the apex of the cell facing the endolymph and traveled toward the base at a speed of ~25 μm-s−1.

(J) Bar graphs show the wave speed, fluorescence intensity modulation, latency to the first ACh-evoked Ca2+ transient, and rate of evoked bursting (see Video S1). Data are represented as mean ± SEM; ∗p < 0.05.

In Köllicker's organ and the organ of Corti, ATP activates purinergic receptors in surrounding SCs, inner HCs, and dendrites of primary auditory neurons (Glowatzki et al., 1997; Housley et al., 1998, 1999; Järlebark et al., 2000, 2002; Parker et al., 1998; Salih et al., 1999; Sueta et al., 2003; Szücs et al., 2004; Telang et al., 2010; Wang et al., 2003; Yan and Gu, 2013; Zhao et al., 2005; Chen et al., 1998). ATP can increase membrane conductance, cause rises in Ca2+, and significantly change SC morphologies during development (Berekméri et al., 2019). Therefore, we tested the effects of ATP on possible purinergic signaling during development in clinocytes and clino2 cells. It was observed that 100 μM ATP evoked significant Ca2+ transients in multiple cell types in both the developing Köllicker's organ and organ of Corti, whereas there were no detectable ATP-evoked Ca2+ transients in clinocytes or clino2 cells of the same mouse (both ears were tested, data from one ear are shown; Figure S1 and Figure S2; Videos S1A and S1B).

Spontaneous Ca2+ Transients Evoked by ACh and Muscarine Blocked by Atropine

We next tested the effect of muscarinic receptor agonist atropine for its ability to block ACh- and muscarine-evoked Ca2+ transients in clino2 cells and clinocytes. Before ACh, muscarine, or atropine drug application two types of spontaneous Ca2+ events were observed and recorded (P1, Figures 5A–5C). One population of cells had slow spontaneous Ca2+ transients (Figure 5B, c) with a G5 fluorescence rise time constant of 2.80 ± 0.40 s, fall time constant of 17.1 ± 1.33 s, G5 half-width duration of 16.4 ± 1.35 s, ΔF/F 0.27 ± 0.020, and a frequency rate of 0.009 ± 0.0008 s−1. Another cell population had fast spontaneous Ca2+ transients (Figure 5C, cts) with a G5 fluorescence rise time constant of 1.54 ± 0.080 s, G5 fall time constant of 2.87 ± 0.065 s, G5 half-width duration of 1.89 ± 0.055 s, ΔF/F 0.487 ± 0.033, and G5 frequency of 0.0256 ± 0.0026 s−1 (Figures 5F–5H). Muscarine (100 μM) evoked fast Ca2+ transients in clinocytes whereas clino2 cells had large and longer lasting Ca2+ transients (Figures 5D and 5E).

Figure 5

ACh- and Muscarine-Evoked Ca2+ Transients Blocked by Atropine at P1

(A) Resting G5 fluorescence (F0, blue) merged with peak change in G5 fluorescence (ΔF/F0, green) occurring spontaneously in an excised preparation.

(B–E) (B and C) Waterfall plots of spontaneous Ca2+ transients in clino2 cells (c, n = 43; B and E–G), with long-lasting spontaneous transients, bursting in some cases (c.f. Figures 4 and 6). (C and E–G) Spontaneous Ca2+ transients in clinocytes (cts, n = 57) exhibited short events. (D and E) A 1-s puff of 100 μM muscarine evoked bursts of short Ca2+ transients in one cell population (cts, black traces), whereas another population had long-lasting Ca2+increases (red traces).

(F–H) Muscarine evoked short Ca2+ transients in clinocytes and bursts of long-lasting transients in clino2 cells. Data are represented as mean ± SEM, ∗p < 0.05. (F–H) Spontaneous and muscarine-evoked Ca2+ transients in clino2 (c) and clinocytes (cts): (F) peak ΔF/F0, (G) event half-widths; (H) latency to first muscarine-evoked response. (F) Atropine (ATR) reversibly blocked responses to muscarine. (H) Latency to the first pooled muscarine-evoked Ca2+ transients before 50 μM ATR (Musc.) and after washout (Musc. second) was statistically indistinguishable.

(I–K) ACh-evoked Ca2+ transients in the control condition of a clino2 cell (c) and clinocytes (cts) at P1 were blocked by atropine (50 μM ATR, K2), and partially recovered after washout (K3). Relatively small ACh-evoked Ca2+ transients were also present in G5-expressing hair cells (HC) and their hair bundles (hb; also see Videos S2, S3, S4A, and S4B).

Figure 6

Muscarine-Evoked Ca2+ Transients at P10

(A and B) Resting G5 fluorescence (F0, blue) merged with peak change in G5 fluorescence (ΔF/F0, green) evoked by 100 μM muscarine at two different focal planes (Z1, Z2) in the anterior crista.

(C and D) Short Ca2+ transients were evoked by muscarine in clinocytes (cts) in focal plane Z1, and longer sustained or bursting transients in clino2 cells (c).

(E and F) (E) Clinocytes (cts) covering the apex of the central crista (focal plane Z2) responded to muscarine with rapid Ca2+ transients. Transients are sorted by peak values for focal plane Z1 (D, n = 29) and focal plane Z2 (F, n = 28).

(G) Kinetics of G5-detected Ca2+ transients were similar to those in P1 mice (c.f. Figure 3), but with reduced onset latency in P10. Data are represented as mean ± SEM; ∗p < 0.05. H) A clino2 cell (clino2, green) and hypothetical clino2 cell (h. clino2, red) from Z1 and Z2 with different muscarine evoked Ca2+ transients (from graphs C and E).

(also see: Videos S5A and S5B).

Sensitivity to a puff application of 100 μM ACh in the control condition (Figures 5I–K1) was blocked by 50 μM atropine (ATR; Figure 5K2) and partially recovered after washout (Figure 5K3). Relatively small ACh-evoked Ca2+ transients were also present in G5-expressing hair cells (HC) and their hair bundles (hb) (Videos S2, S3, S4A, and S4B).

Muscarine-Evoked Ca2+ Transients in Second Postnatal Week

To determine whether ACh- and muscarine-evoked Ca2+ transients persist through postnatal development we examined EC cells during the second postnatal week. ACh-evoked Ca2+ responses were observed where resting G5 fluorescence (F0, blue) merged with peak changes in G5 fluorescence (ΔF/F0, green) recorded at two different focal planes (P10; Figure 6 Z1(A), Z2(B)). Short Ca2+ transients were evoked by 100 μM muscarine in clinocytes (cts) in focal plane Z1, along with longer transients in clino2 cells (C; Figures 6A, 6C, and 6D). Clinocytes (cts) comprising the apex of the central crista and EC (focal plane Z2) responded to muscarine with rapid Ca2+ transients (Figures 6B, 6E, and 6F). Kinetics of G5-detected Ca2+ transients were similar to those in P1 mice (c.f. Figure 3), but with reduced onset of latency (∼6.5 s), fall (∼1.45 s), half-width (∼2.65 s), and frequency (∼0.0385 s−1) in the P10 compared with P1 tissues tested (Figures 6G and 6H; also see: Videos S5A and S5B). Data shown in Figure 6 are representative of muscarine-evoked Ca2+ transients from 3 mice at similar ages (P10, P14, and P15). At this age and in adult mice, clinocytes and clino2 cells continue to have spontaneous and ACh-evoked Ca2+ transients (Figure S3).

Horizontal Canal Ca2+ Transients in GAD2-tdT Cells

As the mouse horizontal canal does not have an anatomically distinct EC, we examined the horizontal canal crista for GAD2-tdT cells with a morphology similar to that of clinocytes and clino2 cells (Figures S4A–S4E). Horizontal canal cristae from two representative ages, P1 (Figures S4C and S4D) and P35 (Figures S4A, S4B, and S4E), show GAD2-tdT cells throughout horizontal cristae. G5 resting fluorescence in GAD2-tdT cells responded to ACh (100 μM) with large G5 ΔF/F0 transients consistent with calcium transients observed in clino2 cells of anterior and posterior cristae (Figures S4C and S4D). Other GAD2-tdT cells with low G5 resting fluorescence responded to ACh with brief Ca2+ transients consistent with clinocytes (E; see Videos S2A and S2B). This raises the possibility that clinocytes and clino2 cells might be present in horizontal canal crista and cristae from primates in the absence of an EC (F). Immunolabeling of horizontal canal cristae with the HC marker MyoVIIa (green) and neurofilament marker NF200 (blue) show several GAD2-tdT type I HCs with calyces (ages P5, top panel, and P6, bottom panel). Hypothetical clino2 cells and clinocytes cluster near the planum (arrow, P5).

2-APB Blocks ACh-Evoked Ca2+ Transients in Clino2 Cells and Clinocytes

The IP3 receptor inhibitor 2-APB, which blocks the IP3 receptor in frog HCs (Rossi et al., 2006) and store-operated intracellular Ca2+ transients (Bootman et al., 2002) was tested for its effects on Ca2+ transients in clinocytes and clino2 cells. A semi-intact preparation of an anterior crista from a P10 mouse revealed a clino2 cell (c; solid green outline) and hair cells (HCs) with low levels of G5 fluorescence at rest (F0, blue, Figure 7A). Clinocytes (cts) showed low levels of resting G5 fluorescence in the same tissue (dotted outline, EC boundary, Figure 7A). A puff of 100 μM ACh evoked large Ca2+ transients in clino2 cells (c) and modest Ca2+ transients in clinocytes (Figures 7B and 7E). ACh-evoked Ca2+ transients were blocked with a single puff of 50 μM 2-APB in clinocytes and largely, but not completely, blocked in the clino2 cell (Figures 7C and 7F). The relative magnitude of ACh-evoked Ca2+ transients and extent of the 2-APB block is shown by merging Figures 7A–7C (Figure 7D). ACh-evoked Ca2+ transients in control conditions were sorted by evoked Ca2+ transient peak values, and in control conditions, were sorted by peak values plotted as waterfall charts (Figure 7E). The largest G5 fluorescence observed was from a clino2 cell with Ca2+ transients throughout the cell, starting in the apex (1), then middle (2), and finally base (3) of these cells (green, Figure 7E). Black traces represent multiple regions of interest in clinocytes and HC-calyx complexes. Application of 50 μM 2-APB significantly reduced ACh-evoked Ca2+ transients in all cells (Figure 7F). Responses in clino2 cells were significantly blocked by 2-APB (green; 1, 2, 3). Some HC-calyx complexes continued to show ΔF/F0 modulation evoked by ACh application, whereas other cells had small evoked short low-level Ca2+ transients appearing as spontaneous events unrelated to the ACh application (Figures 7C and 7F, arrows; also see Videos S6A and S6B).

Figure 7

Blocking Action of 2-APB on ACh-Evoked Ca2+ Transients in the Anterior Crista at P10

(A) Clino2 cell (solid green outline) and hair cells with G5 fluorescence at rest (F0, blue), whereas clinocytes (supporting, dotted green outlines) exhibit very modest G5 fluorescence at rest.

(B) ACh (100 μM) evokes intense Ca2+ transients in clino2 cell (c), and modest transients in clinocytes.

(C) ACh-evoked transients were blocked by 50 μM 2-APB in clinocytes and largely blocked in the clino2 cell.

(D) Relative magnitude of ACh-evoked transients and extent of 2-APB block is illustrated by merging (A–C).

(E) ACh-evoked transients in the control condition sorted by peak values. The largest transients were in the clino2 cell, with Ca2+ transients in the apical (1), central (2), and basal (3) regions of the cell (green). Black traces are clinocytes and hair cell–calyx complexes.

(F) Same cells as (E) in the presence of 50 μM 2-APB shown on a ~10× log scale. Responses in clino2 cells were largely blocked by 2-APB (green: 1, 2, 3). Some HC-calyx complexes continued to have low-frequency ΔF/F0 modulation following ACh application, whereas other cells had brief Ca2+ transients that appeared to be spontaneous (F: arrows) (also see Videos S6A and S6B).

Vesicular GABA Transporter in the EC and Crista

With large calcium responses to ACh and muscarine we examined the presence of vesicular transporter mechanisms for ACh and GABA in the EC of wild-type C57BL/6J and GAD2-IRES-Cre transgenic mice. Efferent neurons in the cristae utilize ACh as the primary transmitter; however, many studies have reported an absence of efferent innervation in the EC. Neither choline acetyltransferase (CHAT) nor vesicular acetylcholine transporter were observed in the EC (data not shown). Taken together, this suggests that the endogenous source of ACh acting on clinocytes and clino2 cells does not arise directly from efferent synaptic release. Although the application of GABA did not evoke detectable Ca2+ transients in clinocytes or clino2 cells, we observed GABA and vesicular GABA transporter (VGAT) in the EC (Figure 8). During early postnatal development small VGAT puncta (green) are present near clino2 cells (red) and in cells around the EC (Figure 8A). VGAT continued to label puncta at the sides of the EC (P10, Figure 8B). The majority of VGAT immunolabeling appeared in puncta surrounding putative GAD2-tdT HCs in the crista (Figure 8C). In this example, four putative GAD2-tdT type I HCs align in a row within the crista (dotted box). When the boxed image is rotated (i) three putative GAD2 type I HCs are apparent with several surrounding VGAT puncta (2–4), presumably from synaptic contacts; one putative HC without GAD2-tdT is outlined with VGAT puncta (∗). A GAD2-tdT cell with an SC morphology is devoid of synaptic VGAT puncta (red; 1). To verify the VGAT immunolabeling in GAD2-tdT transgenic mice we examined aged-matched wild-type C57BL/6J cristae. We observed VGAT puncta (green) near the apex and cuticular plates of cells (P8, Figure 8D). GAD65 puncta were present at the cell membranes in the apical, middle, and basolateral regions (blue). A large GAD65 foci also was observed (∗). These data further demonstrate that immunolabeling with GAD65 may not reflect all GABA activity.

Figure 8

Vesicular GABA Transporter (VGAT) in the Crista not EC

(A) VGAT (green) co-labels with GAD65 (blue) in small puncta in the EC near clinocytes (red; P5–P7, k = 2). GAD65 also labels large central regions in the EC plexus.

(B) A top-down view of clinocytes (red) in the middle and a clino2 cell on the edge of the EC (P8–P10; k = 2).

(C) In the crista, VGAT puncta surround GAD2-tdT HCs (box, i). A row of 4 GAD2-tdT cells appear as 3 putative type I HCs (2–4) and a putative SC (left). The GAD2-tdT cells (2–4) have VGAT puncta around cell surfaces. The GAD2-tdT cell (1) with a different morphology lacks VGAT puncta.

(D) In a cross section of a crista, VGAT puncta are present near the apex of cells labeled with GAD65 puncta (blue), near the cell membrane. An extracellular structure also labeled with GAD65 (star).

ACh-Evoked Ca2+ Transients in EC GAD2 Progenitor-like Cells in Mature Mice

Previous studies have shown that C57BL/6J mice have age-related hearing loss (Ison et al., 2007; Mock et al., 2016), but the vestibular function in these mice is normal (Jones et al., 2006). Because C57BL/6J mice have a decline in cochlear function with age we examined whether clino2 cells and clinocytes maintained ACh- and muscarine-evoked Ca2+ transients in old age. Semi-intact vestibular organs were placed under SFC for live cell imaging with continuous perfusion. Resting G5 fluorescence (F0, blue) merged with peak changes in G5 fluorescence (ΔF/F0, green) evoked by either 100 μM ACh (Figure 9A) or 300 μM muscarine (Figure 9B) in an anterior crista from a 2-year-old mouse (P825) was recorded. In this tissue clinocytes had three distinct ACh-evoked Ca2+ transients (1, 2; green traces, and 3, red trace). Responses in clinocytes were nearly identical between ACh and muscarine treatments (Figures 9A-9C and 9E-9G). Putative calyces (N) in the sensory HC domain of the crista (Figure 9A) responded to 100 μM ACh in mature mice with long-lasting Ca2+ transients (Figure 9D) consistent with previous studies examining mAChRs in calyces (Holt et al., 2017). The latency to the first evoked Ca2+ transient in this mouse was equivalent to that observed at P10 (c.f. Figure 4), suggesting that clinocytes are fully functional by age P10. Statistically significant differences between ACh and muscarine were recorded in the G5 fluorescence half-life, but the dose was not titrated to match strengths of the stimuli (Figure 9H; also see Videos S7A and S7B). The ACh-evoked Ca2+ transients in this mature mouse were apparent rhythmic bursts; however, bursting was also observed in clino2 cells and clinocytes in younger mice (e.g., Figure 6), ruling out age as a required factor. It has been suggested that small changes in IP3 signaling in astrocytes can lead to highly diverse intracellular calcium events ranging from a solitary pulse to rhythmic bursts (Taheri et al., 2017), and it is possible that similar mechanisms are present in clino2 cells and clinocytes. It is not yet known if bursting or the precise temporal waveform plays an important role in the physiological function of these cells.

Figure 9

ACh- and Muscarine-Evoked Ca2+ Transients in Clinocytes in Mature Mice

(A and B) Resting G5 fluorescence (F0, blue) merged with peak change in G5 fluorescence (ΔF/F0, green) evoked by 100 μM ACh (A, n = 19) and in the same cells 300 μM muscarine (B, n = 19) in the anterior crista of a P825 mouse.

(C–G) Clinocytes responses were nearly identical between ACh and muscarine treatments. (D) Putative calyces (N, n = 12) in the sensory hair cell domain of the crista (A) respond to 100 μM ACh in adult mice with long-lasting Ca2+ transients distinct from those evoked by muscarine or ACh in clinocytes. Latency to the first calcium transient in adult and mature mice was similar to that observed at P10 (c.f. Figure 4), suggesting that clinocytes are fully functional by P10.

(H and I) Bar graphs quantify the evoked Ca2+ transientsfrom ACh (grey) and muscarine (Musc., white)Data are represented as mean ± SEM; ∗p < 0.05. (I) The latency of evoked Ca2+ transients and rates of evoked Ca2+ transients after ACh or Musc were not signifianctly different, but peak changes in G5 fluorescence (ΔF/F0) was dignificantly less in Musc (also see: Videos S7A and S7B).

EC Plexus

Just below clino2 cells and clinocytes is a distinct anatomical structure with unknown function, denoted here as the EC plexus. To characterize the plexus, immunolabeling of CHAT, muscarinic acetylcholine receptor 1 (M1), and βIII tubulin (TUBB3) antibodies was tested. CHAT and M1 antibodies did not recognize epitopes within the EC (data not shown). Labeling with a fluorescently conjugated βIII tubulin antibody revealed a putative afferent fiber close to the EC (Figure 10A, inset, i). This βIII tubulin fiber appeared to initiate from one side of the crista, but was not observed within the EC, providing further evidence that the EC region is largely devoid of neurons. A split fluorescent image is shown of TUBB3-labeled afferent calyces (right side) and GAD2-tdT cells (left side) of a crista. This experiment was repeated in three mice from two age groups young (P0–P14) and mature (P100–P720), and βIII tubulin was consistently absent from the EC plexus. In Figure 10B, GAD65 immunolabels the EC plexus in a mature mouse aged P602 (Figure 10B), suggesting a functional relevance of the plexus throughout life. A mature mouse revealed individual clinocytes (arrows) contacted by branches from the EC plexus, along with the clino2 cell (∗) at the EC periphery. Although GABA did not evoke detectable Ca2+ transients in clinocytes or clino2 cells with the current transgenic model, we tested an anti-GABAA receptor antibody as it has been shown to play a role in regulating neural progenitor cells via GABA signaling (Wang et al., 2003). Immunolabeling of GABAA receptor (GABAA R, blue) appears in structures surrounding the outer membrane of clinocytes (arrows) and a clino2 cells (star) in a mature mouse (P936; Figure 10C; also see Video S8). Immunolabeling with GABAA R also appears in puncta surrounding HCs (white, MyoVIIa) in the crista during postnatal development (P6, Figure 10D).

Figure 10

EC Plexus

(A) In a posterior canal crista from a mature mouse (P318, k = 1), βIII tubulin calyces (green) surround putative GAD2-tdT HCs (red). Few non-calyceal βIII tubulin fibers are observed near the EC with GAD2-tdT cells (red; box and inset, i).

(B) Anti-GAD65 antibody (blue) immunolabels the EC plexus in a mature mouse (P602, k = 1), where branches from the EC plexus contact a clino2 cell (∗) and clinocytes (arrows).

(C) GABAA R immunolabeling (blue) within the EC near clino2 cell (star) and clinocytes (arrowheads) in a mature mouse (>2 years old, k = 1).

(D) GABAA R puncta near HCs (white; MyoVIIa) in a developing mouse crista (P6; k = 2; also see Video S8).

Discussion

We show here that GAD2 progenitor-like cells in the EC of vertical semicircular canal cristae are present in neonates and maintained in adult mice. ACh and muscarine evoke atropine-sensitive Ca2+ transients that distinguish two GAD2 progenitor-like cell types: clinocyte and clino2 cell. Our data provide evidence for a role of GAD2 during development in the neuroepithelium of cristae and GAD2 progenitor-like cells in the EC forming contacts with a central EC structure, the EC plexus.

GAD2 Progenitor-like Cells and Rosettes

Formation of multicellular rosettes occurs in many organs and species including zebrafish lateral line primordium, the Drosophila epithelium and retina, and adult neural stem cell niches (Mirzadeh et al., 2008; Elkabetz and Studer, 2008). To date, two mechanisms of cytoskeletal rearrangements have been described for rosette formation: apical constriction and planar polarized constriction (Harding et al., 2014). Extracellular cues that trigger these rearrangements in vivo are less well known and potentially more diverse.

In the CNS, neural precursor cells form rosettes in the ventral subventricular zone (V-SVZ) and act as a niche for neural stem cells, giving rise to specific subtypes of olfactory neurons and glial cells (Merkle et al., 2007; Goldberg and Hirschi, 2009; Giachino et al., 2014a,2014b). Unlike the transient embryonic rosettes seen in many systems, V-SVZ rosettes are maintained throughout adulthood, although there is a decrease in the number of rosettes with age (Shook et al., 2012). These rosettes provide a niche environment for proliferation and differentiation of adult neural stem cells.

V-SVZ rosettes are composed of ependymal cells that contact the cerebrospinal fluid of the ventricle surface and astrocyte-like cells that contact blood vessels (Ihrie and Alvarez-Buylla, 2011; Fuentealba et al., 2012). Clinocytes may hypothetically contact both the cupula and the surface of the EC plexus. This previously undescribed structure located in the center of the EC resembles a ventricle with its location, immunolabeling, and significant branching. The EC plexus may provide signaling mechanisms similar to the V-SVZ region described above and in other CNS regions (Tochitani and Kondo, 2013). Future studies examining the cellular composition and physiology of the plexus will determine its function.

Clinocytes and Clino2 Cells: GAD2, SOX2, PROX1, and CTBP2

Several studies have reported the important role of SOX2 in the organ of Corti during development (Dabdoub et al., 2008; Gu et al., 2016; Cheng et al., 2019; Kempfle et al., 2016; Yang et al., 2019; Locher et al., 2013; Wilkerson et al., 2019; Neves et al., 2013; Kiernan et al., 2005; Dvorakova et al., 2020). Similarly, SOX2 plays a vital role in the development of SC and HCs in the vestibular system. We demonstrate that clinocytes and clino2 cells express SOX2 during early postnatal development. Examining additional pluripotent factors including the stem cell marker Lgr5 and the Notch/Wnt signaling molecules would provide further insight into the mechanisms of these cells.

The transcription factor PROX1 is induced and regulated by SOX2 and is essential for controlling crista size (Dabdoub et al., 2008; Liu et al., 2018; Fritzsch et al., 2010). In this study, we observed PROX1 in clinocytes at a single age (P8); however, further studies are required to determine its potential regulatory mechanism in these cells.

CTBP2 is a transcription factor that has been less studied for its potential role in transcriptional regulation in cells of the vestibular epithelium. This is likely because of the significant role that CTPB2 plays in HC ribbon synapses. CTBP2 contributes to cell fate in mouse embryonic stem cells (mESCs) through the Oct4-interacting proteins, suggesting that CTBP2 is necessary for Oct4 function in establishing ESC identity (Ding et al., 2012; Tapia et al., 2015; van den Berg et al., 2010; Kim et al., 2015). Cells lacking CTBP2 have demonstrated delays in differentiation (Tarleton and Lemischka, 2010). Other studies report that CTBP2 acts as a transcriptional repressor facilitating the Notch signaling pathway (Suh et al., 2018). Published data suggest the presence of diffuse CTBP2 immunolabeling in mouse inner HC nuclei and adult mouse Lgr5 cell colonies (Kujawa and Liberman, 2009; Sergeyenko et al., 2013; McLean et al., 2017). Future studies examining the role of CTBP2 in clinocyte regulation would enhance our understanding of these cells and may provide information for the differential gene expression among clinocytes and clino2 cells.

We propose a hypothetical model for the EC GAD2 progenitor cell lineage where clinocytes (CTs) organize into rosettes in a quiescent state (Figure 11). Expression of GAD2 in CTs leads to GABA synthesis with SOX2 and unknown transcription factors (TF1). GABA along with other signaling molecules (SM1) signals back to CTs, promoting their self-renewal. If GABA synthesis is reduced or blocked, proliferation of CTs may undergo cell division and differentiate into clino2 cells (CC) with SOX2 with additional transcription factors (TF2) and signaling molecules (SM2). Transcriptional regulation in clino2 cells (TF3) and signaling to clino2 cells (SM3) may promote division into SCs, and further differentiation to GAD2 HCs (TF4) with other signaling molecules (SM4).

Figure 11

Hypothetical Model of Clinocyte and Clino2 Cell Lineage

A hypothetical model depicting GAD2 progenitor-like clinocyte cells (CTs) and their organization into rosettes, which are maintained in a quiescent state (far left). When GAD2 is transcribed in clinocytes and GAD65 is expressed these cells can synthesize GABA. Hypothetical release of GABA from CTs may provide a signaling mechanism that allowing a quiescent state and continuous rosette formation. Signaling molecules (SM1-4) and transcriptional factors (TF1-4), represent factors in the hypothetical clinocyte cell lineage. Regulation of GABA synthesis in CTscould maintain TFs and SMs keeping clinocytes in a quiescent state. Transient GABA could promote cell division of clinocytes into clino2 cells (CC; TF2, SM2). Transcriptional factors CTBP2, PROX1, and SOX2 in clinocytes may be actively involved at this stage. Differentiation of clino2 cells with key transcription factors (TF3) and signaling molecules (SM3) may lead to immature supporting cells (SC) that express ATOH1. A possible mechanism for generating GAD2 hair cells (HCs; MyoVIIa) with TF4 and SM4.

GABA and ACh Signaling

GABA is a conserved signaling molecule and adapted to serve as the predominant inhibitory neurotransmitter in the CNS. In the adult hippocampus SGZ neurogenic niche of radial-like glial cells maintains quiescence by tonic GABA release from interneurons. Once interneurons differentiate into neural progenitor cells, GABA continues to regulate their development into mature granule cells (Jin et al., 2001; Tozuka et al., 2005; Coulter and Carlson, 2007; Houser, 2007; Kim et al., 2012; Catavero et al., 2018). As clino2 cells have detectable levels of GABA and the EC plexus immunolabels with GABAA R we hypothesize that similar signaling mechanisms may contribute to proliferation of clinocytes and differentiation into clino2 cells in the EC (Figure 11). Further lineage tracing and single-cell analysis of EC GAD2 cells would provide information on the temporal control of GAD2 in clinocytes and clino2 cell progeny.

We demonstrate the presence of GABAergic machinery in the EC; however, many functional questions remain. Given the muscarinic ACh-evoked Ca2+ transients reported here, we hypothesize that ACh may modulate non-vesicular GABA released by clinocytes and clino2 cells and provide a mechanism for signaling to the EC plexus. The absence of innervation in the EC is consistent with previous reports (Desai et al., 2005a, 2005b; Lysakowski, 1996), and a source for ACh in the EC remains unknown. Evidence from other systems suggests that ACh plays a role in local cell signaling and autocrine function in embryos and adults (Grando, 1997; Wessler et al., 1998, Wessler and Kirkpatrick, 2008; Williams et al., 2004). Non-neuronal mouse and human embryonic stem cells express CHAT and muscarinic ACh receptors (Serobyan et al., 2007; Paraoanu et al., 2007a,2007b; Landgraf et al., 2010; Takahashi et al., 2014), and in mice, ACh mobilizes Ca2+, which increases cell viability, decreasing cell proliferation.

Although previous studies determined an EC in phylogenetically diverse species including birds, fish, frogs, turtles, bats, cats, dogs, rats, and mice (Igarashi and Yoshinobu, 1966; Lewis et al., 1985; Fritzsch et al., 2002), an EC has not been observed in all mammals. One possibility is that cell types identified here in the EC of mice are present, but not organized in an anatomically distinct EC. Based on this, further examination of GAD2, GABA, and the EC including clinocytes, clino2 cells, and the EC plexus is necessary to determine whether there are equivalent cells and mechanisms in human and non-human primates.

Limitations of the Study

This study utilized transgenic mice with a C57BL/6 strain of origin (Taniguchi et al., 2011; Gee et al., 2014). Previous studies reported C57BL/6 mice with early-onset hearing loss (Ison et al., 2007; Mock et al., 2016); however, their vestibular function was normal (Jones et al., 2006). Future studies will be directed toward GAD2 progenitor-like cells in other strains and species. Although immunohistochemical experiments of intact whole-mount vertical cristae reveal an EC plexus, its origin, cellular composition, and function remain unknown.

Resource Availability

Lead Contact

Holly Holman (holly.holman@utah.edu).

Materials Availability

This study did not generate new unique materials.

Data and Code Availability

The published article includes all data generated or analyzed during this study.

Data analysis software FluoRender and code are available on Github.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.