Adult male and female Long-Evans rats (8–9 weeks old) were purchased from Janvier (le Genest Saint Isle, France). Animals were used directly or mated at the local animal facility to obtain descendants. For this purpose, additional pregnant females were also obtained from the same supplier. Eight rats (6 female, 2 male) were used for the characterization of the HCs in the adult sensory organs. Each estimate was obtained on several organs from at least two different animals as described throughout the text. To characterize the developmental maturation of the HCs in vivo, 24 newborns from three litters from different progenitors were used for histology at postnatal days 0 (P0), P1, P4, P7, P10, P14 or P28. Other newborns were used at postnatal day 1 (P1) for tissue culture. The present in vitro data were obtained on tissues from 57 pups taken from 12 litters. For tissue collection, rats were euthanatized by decapitation, after prior anesthesia of P14 and older animals. For direct histology, temporal bones were quickly immersed in fixative (4% paraformaldehyde in phosphate buffered saline (PBS), pH 7.2) and the vestibular epithelia were dissected using a binocular microscope under a fume hood. The tissues were fixed for 1 h in the same fixative, then rinsed twice for 20 min in PBS and immersed in a cryopreservative solution (34.5% glycerol, 30% ethylene glycol, 20% PBS, 15.5% distilled water) for storage at −20°C until further processing (Maroto et al., 2021).

Vestibular epithelia were collected from P1 rats. The dissection was performed in Leibovitz’s L-15 medium (Gibco, cat. # 11415064) in sterile conditions. Otoconial membranes were removed from utricles and saccules and vestibular nerves were trimmed, but the membranes roofing the endolymphatic spaces were not. The epithelia were then cultured free floating for up to 27 days (referred as to 28 DIV, corresponding to P1+ 27 days in culture) in a standard culture medium composed of Dulbecco’s Modified Eagle’s Medium-Nutrient Mixture F12 (DMEM/F12) (Gibco, cat. #11320033) supplemented with 25 mM HEPES (Gibco, cat.# 15630080), 2% N2 supplement (Cell Therapy Systems, cat.# A1370701), 1% GlutaMAX (Gibco, cat.# 35050061), 2 g/L glucose (Sigma, cat.# 49163) and 1.5 g/L penicillin (ThermoFisher, cat.#J63032.14), as described for adult mouse utricle cultures (Cunningham, 2006; Rua et al., 2013). The culture plates were incubated in a 5% CO₂/95% air environment at 37°C. In one experiment, several modifications to enrich these standard conditions were evaluated for their potential to increase HC survival, as hypothesized from literature data (Koeler et al., 2017; Meas et al., 2018; Lou et al., 2023). These were the addition of 1% Matrigel (Corning, cat.# 356237), the simultaneous addition of 1% -B27 (Gibco, cat. # 17504044) and 50 ng/mL mouse recombinant EGF (Gibco, cat.# PMG8041), and the application of slow orbital rocking (5 rmp, 7°, using a Biosan 3D Sunflower mini-shaker). These modifications were applied single or in combination for the entire period (28 days) or for the first half (days 1–14) of the period only. In another experiment, the culture medium was enriched for the four initial days only (DIV4) with growth and protectant factors. Thus, the standard medium was enriched with 50 ng/mL EGF (Ding et al., 2016; Preprotech, cat.# AF100-15), 0.1 μg/mL FGF2 (Jeong et al., 2021; Sigma, cat.#F0291) and 0.1 μg/mL FGF10 (Zelarayan et al., 2007; Preprotech, cat.# AF-100-26), as well as with 1 mM N-acetyl-L-cistein (Sigma, cat.#A9165), 10 mM nicotinamide (Sigma, cat.#N0636) and 1% insulin-transferrin-selenium-ethanolamine (ITS-X) supplement (Gibco, cat. # 51500056). The enriched medium was changed at DIV1 and DIV3 and filtered at DIV2 and DIV4. From DIV5 onwards, the cysts were maintained in standard medium, changed every other day. For histological assessment, vestibular cultures were fixed and stored as described above.

Whole-mount immunohistochemical labeling of the vestibular epithelia was performed using 3 or 4 of the primary antibodies listed in table 1, and suitable combinations of secondary antibodies conjugated with Day-Light-405 or Alexa-405, Alexa-488, Alexa-555 and Alexa-647 fluorophores (Invitrogen), according to published protocols (Lysakowski et al., 2011; Maroto et al., 2021a,b). In brief, the specimens were first rinsed with PBS and then permeated and blocked with 4% Triton-X-100% and 20% donkey serum in PBS for 1 h at room temperature. Next, they were incubated with the primary antibody mixture containing 1% donkey serum and 0.1% Triton-X-100 in PBS, for 24 h at 4°C. Secondary antibodies were incubated overnight in the same conditions plus protected from light. The specimens were rinsed with PBS between each incubation step and gently rocked at all steps. Fluoromount (Sigma-Aldrich cat.#F4680) was used for whole-mounting the samples.

The vestibular epithelia were imaged in a Zeiss LSM880 spectral confocal microscope. The tile scan function was used to obtain images comprising the entire epithelium with a ×40 objective (NA 1.3). The scan was adjusted to acquire images with a 10% overlap thus facilitating the accurate construction of the merged image. From these images, we obtained the number of HCs showing one specific label or one particular combination of labels in the complete epithelium by manually marking them using the Multipointer function or the Cell Counter plugin of the ImageJ software (National Institute of Mental Health, Bethesda, Maryland, United States). The images of epithelia in the figures are shown in pseudocolors selected to be accessible to people with colorblindness using the David Nichols’ website (https://davidmathlogic.com/colorblind/#%23F93535-%234071F9-%23FFFF07-%234FEF17).

Data are shown as mean ± SE or as individual points. The same data are also provided in table format in the Supplementary Table File. Group comparisons were made by the Student’s t-test or one-way ANOVA, as appropriate, using the GraphPad Prism 9 software. The Tukey test was used for multiple comparisons after significant ANOVA. The alpha error level was set at 0.05.

Total numbers of HCs in the vestibular epithelia of young adult Long-Evans rats were determined using specimens immunolabelled with antibodies against MYO7A, osteopontin and calretinin. MYO7A is a well described marker of all HCs (Hasson et al., 1997) and by counting MYO7A + cells we obtained the following mean numbers: 4827 in the utricle, 4384 in the saccule, 2987 in the anterior/posterior crista and 2395 in the lateral crista (Figure 1). These mean values were from one male and two female rats and the counts from the three animals were in the same range. Also, in two of the three animals examined, the anterior and posterior crista were identified, and no difference in cell numbers was apparent between these sensory organs (rat 1: anterior 2556, posterior 2721; rat 2: anterior 3409, posterior 2951).

Figure 2 shows the cell counts using osteopontin and calretinin as molecular markers. Only 0.2%–0.3% of the MYO7A+ HCs were counted to express both osteopontin and calretinin, indicating that these two markers identify two different populations of HCs. Therefore, in the adult rat, differently from neonates, MYO7A+ HC that express osteopontin do not express calretinin, and those that express calretinin do not express osteopontin. Besides these two populations, a third sizeable cell population was that of HCs expressing neither one of these two markers, but only the pan-HC marker MYO7A (MYO7A-only HCs). The proportion of each of these 3 cell kinds was alike in all types of sensory epithelia and the overall percentages from the 52,738 HCs identified were 65.3% osteopontin+ HCs, 30.1% calretinin+ HCs, and 4.3% MYO7A-only (osteopontin-, calretinin-) HCs.

Osteopontin has been described as good marker for HCI in the mouse (McInturff et al., 2018) while calretinin has been reported to identify a large but somewhat variable proportion of HCII in several species (Dechesne et al., 1991; Desai et al., 2005a; Desai et al., 2005b; McInturff et al., 2018). To corroborate their cell type specificity in the adult rat, we determined their association with calyx afferents, identified by the calyx junction adhesion protein CASPR1 (Sousa et al., 2009; Lysakowski et al., 2011; Sedó-Cabezón et al., 2015). As shown in Figures 3A–A', the data (n = 26,343 HCs from 10 epithelia derived from 2 different animals, one male and one female) clearly demonstrated the association of osteopontin + cells with CASPR1+ calyces (67.3%), whereas calretinin+ cells are not encased by calyces (31.8%). Other possible combinations of these three markers resulted in very small counts (<0.3% each). Therefore, osteopontin and calretinin selectively label HCI and HCII, respectively, in the adult rat.

Besides the osteopontin+ HCI and the calretinin+ HCII, we aimed to reveal the identity of the MYO7A-only HCs, that is, those expressing neither osteopontin, nor calretinin. Two approaches were used. First, we used the association with CASPR1+ calyces as a criterion to identify bona fide HCI. In epithelia labeled with anti-MYO7A, anti-osteopontin, anti-calretinin and anti-CASPR1 antibodies, the cell counts demonstrated that the MYO7A+/osteopontin-/calretinin- HCs do not associate with CASPR1+ calyces (Figure 3B-B′, n = 1,202 HCs, from 8 epithelia from 2 different animals), indicating that they are not HCI. In agreement with this observation, their cellular shape matched the known features of HCII (Figure 3B, arrow). The second approach to confirm this identity was to examine whether these cells express the transcription factor SOX2, demonstrated to determine HCII identity (Stone et al., 2021). In epithelia labeled with anti-MYO7A, anti-osteopontin, anti-calretinin and anti-SOX2 antibodies, we counted the number of MYO7A+/osteopontin-/calretinin- HCs per epithelium, and compared it with the number of MYO7A+/osteopontin-/calretinin-/SOX2+ cells. As shown in Figure 3C-C′, the same numbers were obtained when the SOX2 label was added to the MYO7A+/osteopontin-/calretinin- identification (paired t-test, t_8df = 2.23, ns, n = 9, one utricle, one saccule and one crista each from 3 different animals). This confirmed that most or all of the MYO7A+/osteopontin-/calretinin- HCs also express the positive HCII marker SOX2. Therefore, these MYO7A+/osteopontin-/calretinin- HCs are a subpopulation of HCII that do not express calretinin.

Other research in the laboratory suggested that the plasma membrane calcium-transporting ATPase 2 (PMCA2), which is expressed in the stereocilia (Yamoah et al., 1998), could be selective for HCI. We therefore evaluated whether PMCA2 associated with either osteopontin or calretinin expression. The data, shown in Figure 3D-D′, demonstrated that HCI, not HCII, express PMCA2 in their stereocilia bundles (n = 32,064 PMCA2+ HCs from 3 saccules, 3 utricles, 2 lateral crista and 6 anterior/posterior crista, derived from 2 male and 2 female rats). Thus, PMCA2 is also a specific marker of HCI.

Significant part of the maturation of the rodent vestibular epithelium occurs during the postnatal period (Nordemar, 1983; Rusch et al., 1998; McInturff et al., 2018). To describe the maturation process with the selective molecular markers used above for the adult tissues, we obtained cell counts in epithelia from postnatal day 0 (P0) to P28 (Figure 4). A detailed time course analysis was performed in the utricle (Figures 4A–E). The total number of MYO7A+ HCs increased during the first postnatal week and remained stable thereafter. At P0, very few HCs expressed only osteopontin, while about 40% of the cells showed an immature phenotype and expressed both osteopontin and calretinin. The number of these osteopontin+/calretinin+ cells declined with time and that of HCs expressing only osteopontin increased largely. Interestingly, the small population of MYO7A cells expressing neither calretinin nor osteopontin previously identified in adult rats was similarly present at all postnatal time points. At day 28, the utricle displayed a nearly mature composition, and no significant differences were found between P28 and P60 utricles in most parameters: total number of HCs (t_4df = 0.575; p = 0.596), number of osteopontin + HCI (t_4df = 1.266; p = 0.274), number of calretinin + HCII (t_4df = 2.062; p = 0.108), and number of MYO7A-only HCII (t_4df = 0.465; p = 0.566). However, a small (1.3%) but significant (t_4df = 5.139; p = 0.007 compared to P60) number of immature HCs, expressing both osteopontin and calretinin, was still present in the P28 utricles.

HC percentages were obtained from total cell counts assessed in P1 and P14 saccules and lateral cristae (Figure 4F). Like the utricles, these epithelia showed at P1 a large percentage of immature HCs expressing both osteopontin and calretinin. At this stage, comparison of the percentages of cell types among the three epithelia indicated that the crista had a significantly larger proportion of immature cells than the saccule (52% versus 38%, p < 0.05) and a smaller proportion of calretinin cells than both the saccule and the utricle (39% versus 56% and 53%, respectively, p < 0.05). However, these differences were reduced with time, and the only difference found at P14 was in the percentage of osteopontin cells, smaller (p < 0.05) in the utricle (46%) than in the crista (54%). Similarly to the utricle, HCs expressing osteopontin and not calretinin (i.e., HCI) were already the most abundant cell type at this time point in both the saccule and the lateral crista.

In preliminary experiments, we observed that most vestibular epithelia obtained at P1, including utricles, cristae, and saccules, successfully closed a fluid-filled cavity within 2–3 days of placement in free-floating conditions into standard culture medium. Cyst formation efficiency declined over the first postnatal week and did not require the presence of an extracellular gel (data not shown). Therefore, all subsequent studies were done with free-floating epithelia obtained at P1. For an initial evaluation of the viability of these cysts, we cultured epithelia in these standard conditions until day 28 (28 DIV, meaning P1+ 27 days in vitro). As shown in Figures 5A–C, cysts cultured in free-floating conditions in standard culture medium had a macroscopic appearance resembling the morphology of the corresponding epithelium in vivo. HC maturation in vitro was assessed by comparing utricles cultured for 10, 14, and 28 DIV (Figures 5D–F). The total number of HCs at these days was lower than those obtained at the same days in vivo (76%, 77%, and 57%, respectively; p < 0.05 in all three comparisons, Figure 5D). However, no significant differences in the number of total HCs occurred when comparing across days 10, 14, and 28 in culture. When HC types were determined (Figures 5E,F), we observed that the numbers of immature HCs, expressing both calretinin and osteopontin declined with time (from 26.0% to 4.1%), while the number of cells expressing only osteopontin or only calretinin increased. At 10DIV, the proportion of HCs expressing only MYO7A (15.0%) was larger than in vivo, but this number declined to 5.1% at 28DIV.

We also evaluated the potential improvement of the cyst viability by the addition of several modifications of the standard protocol, alone or in combination for 14 or 28 days. When the effect of 1% Matrigel, orbital rocking (7°, 5 rpm), and trophic enrichment (EGF + B27) were evaluated either alone or combined, the macroscopic appearance worsened in all cases with respect to the appearance shown by cysts maintained in standard conditions. For a quantitative assessment, we immunolabelled utricular cysts to count HCs and their cell type specification. The confocal microscopy images in Figure 6 illustrate the high density of HCs in the 28DIV utricles cultured in standard conditions (Figures 6A–D), and the reduced viability of utricles cultured in enriched conditions (Figures 6E–H). Quantitatively (Figure 6M), utricles maintained in standard medium had 2,688 ± 84 (X±SE) HCs at 28DIV, representing 57% of the total number of HCs obtained at P28 in vivo (Figure 4). The number of HCs was significantly reduced by all conditions we evaluated for 14 or 28 days as candidate modifications to improve differentiation and survival (p < 0.001 for all conditions, Tukey’s test after significant ANOVA, F (8,18) = 35.54, p < 0.001). Comparing the types of cells in the utricles cultured until 28 DIV (Figure 6N), we observed the following proportions for the standard culture conditions: 49.4% osteopontin+, 39.6% calretinin+, 8.4% MYO7A-only, and 2.6% expressing both osteopontin and calretinin. These proportions resembled much more those determined on the P28 utricles in vivo (61.2%, 33.4%, 4.4%, and 1.3%, respectively, Figure 4F) than after any of the enriched conditions, in which we observed higher proportions of calretinin+ HCs (up to 62.0% in utricles given the F28/M28/R28 condition) or MYO7A-only HCs, up to 78.3% in F28/R28.

In a second experiment, the culture medium was enriched with several growth and protectant factors for the initial 4 days in culture only, as these are key days for epithelium growth and maturation in vivo (Burns et al., 2012). The data obtained showed an improvement in epithelium differentiation (Figure 6O). Thus, the percentages of HCI and calretinin+-HCII (55% and 32%, respectively) in the enriched medium did not differ significantly from the in vivo values (61% and 33%), while a significant difference was recorded for the utricle cultures in standard medium for the whole 28 days (42% and 49%, p < 0.05 Tukey’s test after significant ANOVA). The ANOVA values were F (2,6) = 6.22, p = 0.034 for percent HCI and F (2,6) = 13.39 for percent calretinin+-HCII. Nevertheless, the total number of HCs in the utricles cultured with enriched medium for 4 days (2601 ± 56.8) was similar to the one in utricles cultures with standard medium (2586 ± 299), and both values were significantly lower than the number of HC in utricles from P28 rats (4679 ± 207.9) (Tukey test after significant ANOVA, F (2, 6) = 32.02, p = 0.0006).