Prolonged Dexamethasone Exposure Enhances Zebrafish Lateral-Line Regeneration But Disrupts Mitochondrial Homeostasis and Hair Cell Function

Experimental procedures were performed with approval from the Washington University School of Medicine or NIH Institutional Animal Care and Use Committee, and in accordance with NIH guidelines for use of zebrafish.

Adult zebrafish were raised under standard conditions at 27–29 °C in the Washington University or NIH Zebrafish Facility. Embryos were raised in incubators at 28 °C in embryo media (EM; 15 mM NaCl, 0.5 mM KCl, 1 mM CaCl₂, 1 mM MgSO₄, 0.15 mM KH₂PO₄, 0.042 mM Na₂HPO₄, 0.714 mM NaHCO₃) with a 14 h:10 h light:dark cycle. At 3–4 days post-fertilization (dpf), larvae were anesthetized in 0.02 % tricaine methanesulfonate (MS-222) and screened for transgenic fluorophores using a Leica MZ10 F stereomicroscope with fluorescence equipped with a GFP and a DSR filter set. After screening, 4 dpf, larvae were raised in 100–200 ml EM media in 250-ml plastic beakers and fed rotifers daily. Sex of the animal was not considered for this study because sex cannot be determined in larval zebrafish.

The following transgenic lines were used in this study: nl1Tg (TgBAC(neurod1:EGFP)) [13], w78Tg (Tg(myo6b:GCaMP3)) and w119Tg (Tg(myo6b:mitoGCaMP3)) [14], w200Tg (Tg(mpeg1:yfp)) [15], y229Gt (Tg(tnks1bp1:EGFP)) [16], and idc1Tg (Tg(myo6b:GCaMP6sCAAX)) [17].

To ablate lateral-line neuromast hair cells and induce afferent neurite retraction, 5 dpf larvae were exposed to 3 µM CuSO₄ (Sigma-Aldrich, Cat# 451,677) diluted in EM in cell strainers (Corning; Cat# 431,752) for 1 h. This concentration and time course of CuSO₄ exposure were verified as the optimal minimum dose to induce complete neuromast hair cell loss (mean hair cell number per neuromast: 11 ± 3 (control) vs. 0 ± 1 (3 uM CuSO₄₎, 14 fish per condition, 3 trials). After exposure, larvae were rinsed, and then allowed to swim freely during a 2 h recovery in EM. Immediately following this 2 h recovery, larvae were moved to 6-well plates (Falcon, Cat# 351,146) containing 4 ml EM per well with either 0.1 % DMSO (drug carrier alone) or 0.1 % DMSO + 10 µM dexamethasone (Sigma-Aldrich, Cat# D1756) for 48 h. The dexamethasone concentration used in this study (10 µM) was based on the concentration used in a previous study that identified dexamethasone as an enhancer of lateral-line hair cell regeneration [10]. Larvae were maintained at 28 °C throughout CuSO₄ exposure and dexamethasone post-treatments.

Macrophages are effector cells of the innate immune system, and in a previous study, we observed robust recruitment of macrophages to neomycin-injured neuromasts [18]. To test whether dexamethasone treatment inhibited macrophage recruitment following hair cell injury, 5 dpf w200Tg (Tg(mpeg1:yfp)) larvae were pre-treated with 0.1 % DMSO or 1–50 µM dexamethasone in EM for 24 h. Following pretreatment, larvae were rinsed twice and placed in cell strainers, and then exposed to 100 µM neomycin (Sigma-Aldrich, Cat# N1876) for 30 min. Following neomycin exposure, fish were rinsed and moved to 6-well plates with 4 ml of EM to recover for 1 h. Exposure and recovery times for neomycin treatment were chosen based on previous studies evaluating aminoglycoside toxicity in zebrafish lateral-line hair cells [19, 20].

To test whether dexamethasone pretreatment sensitized hair cells to ototoxic insults, 5 dpf wild-type larvae were pre-treated with 0.1 % DMSO or 10 µM for 48 h, then exposed to 0–100 µM neomycin as described above.

Free-swimming larvae were exposed to DAPI (1:2000, 4 min; D3571, Thermo Fisher) to label hair cell nuclei, then sedated on ice for 4 min. For GFP immunolabeling to label the afferent process or for Otoferlin immunolabeling to label hair cells, larvae were fixed (4 % paraformaldehyde, 4 % sucrose, 150 µM CaCl₂, 0.1 M phosphate buffer) in 2-ml Eppendorf tubes overnight at 4–8 °C. For synaptic labeling, larvae were fixed for 5 h at 4–8 °C.

Larvae fixed overnight were blocked in phosphate-buffered saline (PBS: 8 g NaCl, 0.2 g KCl, 0.24 g KH₂PO₄, 14.4 g Na₂HPO₄ per 1 l distilled H₂O) with 5 % normal horse serum (NHS; Sigma-Aldrich, Cat# H0146), 1 % DMSO, 1 % Triton X-100 (Sigma-Aldrich, Cat# T8787) for 2 h at room temperature. Primary antibody exposure was performed in PBS with 2 % NHS, 0.1 % Triton-X100. Secondary antibody exposure was performed in PBS with 2 % NHS. Following all antibody exposures, larvae were rinsed several times in PBS with 0.1 % Tween-20 (Sigma-Aldrich, Cat# P9416).

For synaptic labeling, larvae were fixed for 5 h, permeabilized in ice-cold acetone (−20 °C, 5 min, Sigma-Aldrich, Cat# 179,124), then blocked in PBS with 2 % goat serum (Sigma-Aldrich, Cat# G9023), 1 % bovine serum albumin (BSA; Sigma-Aldrich, Cat# A9647), and 1 % DMSO for 2 h. Primary and secondary antibody solutions and all post-antibody rinses were performed in the same PBS, BSA, and DMSO concentration mix used for blocking.

Primary antibodies:

Secondary antibodies (used at 1:1000):

All fish were rinsed in PBS prior to being mounted on glass slides in elvanol (13 % w/v polyvinyl alcohol, 33 % w/v glycerol, 1 % w/v DABCO (1,4 diazobicylo[2,2,2] octane) in 0.2 M Tris, pH 8.5) and #1.5 cover slips.

To evaluate hair cell mechanotransduction [21], 7 dpf larvae were anesthetized in EM containing 0.01 % tricaine (EMT), exposed to 3 µM FM1-43 (N-(3-triethylammoniumpropyl)-4-(4-(dibutylamino)-styryl) pyridinium dibromide,Thermo Fisher, Cat# T3163) for 20 s, rinsed in EMT, and then mounted on a FluoroDish (World Precision Instruments, Inc., Cat# FD3510) in low-melt agarose (Invitrogen, Cat# 16,520) containing 0.01 % tricaine (LMAT). The agarose was allowed 2.5 min to solidify before covering in EMT and imaging.

Most confocal images were acquired via an ORCA-Flash 4.0 V3 camera (Hamamatsu) using a Leica DM6 Fixed Stage microscope with an X-Light V2TP spinning disc confocal (60-micron pinholes) equipped with an LDI-7 laser diode illuminator (89 North). Image acquisition was controlled by MetaMorph software. For fixed imaging, z-stack images ~ 10–15 µm depth with a z-step of 0.2 μm were acquired using a × 63/1.4 N.A. oil immersion objective. For live imaging, larvae were anesthetized with 0.01 % tricaine in EM, then mounted lateral side up on a thin layer of 1 % low-melt agarose in a tissue culture dish with a cover-glass bottom (FluoroDish; World Precision Instruments) and covered in EMT media. z-stack images of neuromast L3-5 were acquired with a × 63/0.9 N.A. water immersion objective. z-acquisition parameters with X-light spinning disc: z step of 1 μm, laser “20 % power” for all wavelengths, 100 ms per frame for fixed imaging, 120 ms per frame for live imaging. Images in Fig. 5 were acquired using an LSM 700 laser scanning confocal microscope with a × 63 1.4 N.A. Plan-Apochromat oil immersion objective (Carl Zeiss). Confocal z-stacks of ~ 15 µm depth were acquired with a z-step of 1 µm.

Fiji (ImageJ software) was used to process and analyze most confocal images [22]. To analyze macrophage activation and inflammation, confocal images were reconstructed and analyzed using the Volocity software (Quorum Technologies). Subsequent image processing for display within figures was performed using the Illustrator software (Adobe).

Hair cell counts were quantified from hair cell–specific DAPI-labeled nuclei or from cells immunolabeled for Otoferlin (HCS-1; Fig. 5). To quantify hair cell synapses, raw images containing single immunolabel were converted to 8 bit, then subtracted for background using a 20-pixel (CtBP) or 10-pixel (MAGUK) rolling ball radius. Puncta were defined as regions of immunolabel with pixel intensity above a determined threshold: threshold for Ribeye label was calculated using the Isodata algorithm and threshold for MAGUK label was calculated using the Yen algorithm. Intact synapses were defined as CtBP immunolabel juxtaposed with MAGUK and quantified by scrolling through image z-stacks.

To measure neurite length, the ImageJ plugin Simple Neurite Tracer [23] was used on z-stack images of GFP-labeled afferent neurons. Following ImageJ’s Basic Instructions, which can be found in the Help drop-down menu of the Tracer plugin, neurites were traced by scrolling through the z-stack and creating paths. All neurite structures making up the basket-like processes that surround the neuromast were included in the paths, as well as the stalk at the base of the neuromast where innervating neurites emerged from the main afferent neuronal tract and any meandering projections coming off the dendritic basket. The Simple Neurite Tracer automatically detected the path set between two designated points and calculated a measured length in microns. Any branches emerging from and between paths were measured as separate paths. Once fully traced, all paths within the dendritic basket were then added together to get the total neurite length within the neuromast.

Macrophage density was quantified by scrolling through image z-stacks and counting the number of macrophages within 25-µm radius of a neuromast (using a circle inscribed on the neuromast). The number of macrophages contacting a neuromast was determined by scrolling through the x–y planes of each image stack (1-µm interval between x–y planes, 15-µm total depth) and counting macrophages that were in direct contact with Otoferlin-labeled hair cells. Finally, the number of macrophages that had internalized Otoferlin-immunolabeled material (hair cell debris) was counted and assumed to reflect the number of phagocytic events. For each metric, the recorded number reflected the activity of a single macrophage, i.e., a macrophage that made contacts with multiple hair cells and/or had internalized debris from several hair cells was still classified as a single “event.”

To measure the baseline fluorescence intensity of indicators across whole neuromasts, images containing single channels were background subtracted using a rolling ball radius of the following sizes: images containing cytoGCaMp3 or mitoGCaMP3 (50 pixel) and images containing FM1-43FX or TMRE label (100 pixel). Whole neuromasts were delineated based on hair cell–specific DAPI label in maximum-intensity projections, and mean intensity of the indicator was measured. Measurements from each experimental trial were normalized to the control median value. To accommodate nested data and ensure statistical rigor, we performed subsequent statistical analysis on average value of measurements from multiple neuromasts (L3, L4, L5) within an individual fish, such that each fish provided a single data point.

GCaMP6s-based calcium imaging was used to examine evoked signals in zebrafish hair bundles and presynapses, as described previously [24]. Briefly, after 48 h dexamethasone treatment, individual 7 dpf larvae were first anesthetized with EMT. Pins were used to restrain larvae onto a Sylgard-filled recording chamber. To suppress the movement, alpha-bungarotoxin (125 μM, Tocris) was injected into the cavity of the heart. Larvae were then immersed in extracellular imaging solution (in mM: 140 NaCl, 2 KCl, 2 CaCl₂, 1 MgCl₂, and 10 HEPES, pH 7.3, OSM 310 + / − 10) without tricaine. A fluid jet was used to mechanically stimulate hair bundles (L1–L4) with posterior fluid flow for 500 ms.

To quantify the magnitude of evoked GCaMP6s signals, images were processed in FIJI. Z-stacks were max projected, and images at each timepoint were aligned using StackReg (Thevenaz et al. 1998). A 5-μm and 1.31-μm-diameter circular ROI was drawn over each hair cell base and hair bundle, respectively, to make intensity measurements. To plot evoked changes in GCaMP6s intensity, the baseline (F₀, signal during the 2-s pre-stimulus period) was used to plot the relative change in fluorescent signal from baseline (∆F/F₀). The magnitude of evoked calcium signals was determined using ∆F/F₀ plots. The baseline GCaMP6s signal F₀ was used to estimate the resting calcium levels in each hair bundle or presynaptic region. The response magnitudes or resting GCaMP6s levels for each neuromast were averaged to obtain a single point per neuromast.

Statistical analyses were performed using Prism 9 (GraphPad Software Inc). Datasets were confirmed for normality using the D’Agostino-Pearson test—all the datasets in this study were normally distributed. Statistical significance between two groups was determined by an unpaired Student’s t-test if the two populations had the same variance, or an unpaired Welch’s unequal variance t-test if the two populations did not have the same variance. Statistical significance between multiple groups with normal distributions was determined by one-way ANOVA and appropriate post hoc tests. For datasets dependent on multiple independent variables, statistical significance was determined using two-way ANOVA and appropriate post hoc tests.

In two separate small molecule screens using zebrafish larvae, dexamethasone has been identified as an enhancer of hair cell regeneration in lateral-line neuromasts and an inhibitor of peripheral fin nerve regeneration [10, 11]. Dexamethasone has also been reported to enhance axon regeneration in rats, following injury [25]. We therefore addressed whether dexamethasone influences afferent reinnervation and hair cell synapse regeneration in zebrafish lateral-line neuromasts. To induce injury, we exposed animals to copper sulfate (CuSO₄), which induces loss of lateral-line hair cells and considerable retraction of afferent nerve terminals [26‐28]. Five-day-old zebrafish larvae were treated in either 3 µM CuSO₄ in embryo media (EM) or EM alone for 1 h. After CuSO₄ exposure, the animals were allowed to recover for 2 h, and then transferred to EM containing either 10 µM dexamethasone or drug carrier alone (0.1 % DMSO) for 48 h (Fig. 1A). After 48 h, we assessed the extent of afferent reinnervation and hair cell and synapse regeneration. We focused on posterior lateral-line neuromasts L3–L5 for analysis (Fig. 1B). To examine afferent reinnervation and hair cell regeneration, we visualized afferent nerve fibers labeled with GFP as well as hair cell nuclei labeled with DAPI (Fig. 1C). In agreement with previous observations of lateral-line organs injured with neomycin, we observed a significantly greater number of regenerated hair cells per neuromast in larvae exposed to dexamethasone during regeneration compared to controls (Fig. 2C–E,orange points, ****adjusted P < 0.0001,Tukey’s multiple comparisons test). Additionally, among fish that were not exposed to CuSO₄ (unlesioned), significantly more hair cells per neuromast were observed in fish that were exposed to dexamethasone (Fig. 2A, B, E,gray points). This observed increase in hair cell number per neuromast has been previously shown to be due to increased proliferation of hair cell precursors rather than prevention of hair cell turnover [10]. These results therefore confirm that dexamethasone enhances hair cell regeneration and stimulates hair cell precursors to divide, even in the absence of injury.

To determine whether dexamethasone treatment enhanced or inhibited afferent innervation of hair cells, we measured the length of innervating afferent processes in both unlesioned and lesioned neuromast after regeneration (Fig. 3A–D). In neuromasts that had regenerated after CuSO₄-induced lesions, we observed that the average length of afferent processes per neuromast was significantly greater in animals that recovered in dexamethasone compared to animals exposed to drug carrier alone (Fig. 3C–E (orange box plots); ****adjusted P < 0.0001; Tukey’s multiple comparisons test). We reasoned that increased neurite length in dexamethasone-treated animals was related to the greater number of regenerated hair cells per neuromast. To verify this idea, neurite length was normalized relative to the number of hair cells in each neuromast. Following normalization, we observed no difference in relative neurite length per hair cell in animals recovered in dexamethasone vs. those in drug carrier alone (Fig. 3F (orange box plots); adjusted P = 0.5630). In unlesioned animals treated with dexamethasone, we observed no change in afferent neurite length per neuromast (Fig. 3A, B, E (gray box plots); adjusted P = 0.8215) and a modest reduction in the relative neurite length per hair cell that was not statistically significant (Fig. 3F (gray box plots); adjusted P = 0.2959). These data indicate that dexamethasone treatment enhanced the formation of afferent processes in regenerating neuromasts, perhaps to accommodate a greater number of regenerated hair cells. By contrast, in unlesioned neuromasts, dexamethasone did not promote the addition of afferent neurites to accommodate additional hair cells.

In addition to quantifying hair cell numbers and neurite length, we also quantified the number of intact synapses per hair cell (Fig. 4A–D; defined as presynaptic ribbons adjacent to postsynaptic densities). We observed no difference in synapse counts between dexamethasone treated and 0.1 % DMSO controls when comparing unlesioned neuromasts or when comparing CuSO₄-lesioned neuromasts following regeneration (Fig. 4E). Notably, we did observe significantly more intact synapses per hair cell in all regenerating neuromasts relative to unlesioned neuromasts (Fig. 4E; adjusted ***P = 0.0003 (DMSO), *P = 0.0202 (Dex); Tukey’s multiple comparisons test). Similar observations have been reported in regenerating neuromast hair cells [26, 29] and may reflect ongoing synaptic refinement akin to that observed in developing cochlear hair cells [30]. Overall, these results suggest that dexamethasone treatment enhances afferent neurite outgrowth in regenerating neuromasts and does not impact synapse formation in newly formed hair cells in regenerating or unlesioned neuromasts.

Dexamethasone inhibits inflammation and macrophage activation, and various studies indicate dexamethasone’s anti-inflammatory action can influence nerve regeneration [31]. Previous work has shown that macrophages are recruited to lateral-line organs after injury induced by the ototoxin neomycin [18]. We therefore assessed whether dexamethasone inhibited macrophage entry and phagocytosis after application of the ototoxin neomycin. We quantified macrophage entry and phagocytosis in response to neomycin treatment after pretreating larvae for 24 h with 1, 10, or 50 µM dexamethasone. Larvae that were not exposed to either neomycin and/or dexamethasone served as controls (Fig. 5A). All experiments followed the same protocol that we had previously used to characterize macrophage recruitment and phagocytosis after neomycin toxicity [18]. Dexamethasone treatment did not induce any significant changes to macrophage contacts in the absence of neomycin (Fig. 5B) or to the subsequent macrophage response following neomycin lesion (Fig. 5C–F). Quantification revealed that the number of macrophage-hair cell contacts (Fig. 5G) and number of phagocytotic events (Fig. 5H) were similar in all neomycin treatment groups. These observations indicate that pre- and co-treatment with 10 µM dexamethasone does not suppress macrophage activation in response to neuromast injury.

Numerous studies have indicated that dexamethasone can induce mitochondrial dysfunction in various tissues, including adipocytes [32], skeletal muscle [33], and neurons [34]. We previously reported reduced hair cell uptake of the cationic dye FM1-43 in zebrafish mutants with impaired hair cell mitochondrial function (mpv17^−/−), indicating that these fish possessed reduced driving force for cations through mechanotransduction channels, and elevated hair cell mitochondrial calcium [35]. We therefore evaluated whether 48 h dexamethasone exposure affected hair cell uptake of FM1-43 and calcium homeostasis. We observed no differences in FM1-43 uptake in unlesioned neuromasts that were treated in dexamethasone vs. those incubated in DMSO carrier (Fig. 6A, B, E (gray circles)). In regenerating neuromasts, dexamethasone treatment appeared to cause a modest increase in FM1-43 uptake, but this trend did not reach statistical significance (Fig. 6C–E (orange circles),adjusted P = 0.1786, Tukey’s multiple comparisons test). Notably, hair cells in regenerating DMSO-treated neuromasts (Fig. 6C) displayed significantly reduced FM1-43 uptake relative to unlesioned DMSO-treated neuromasts (Fig. 6A, E,adjusted *P = 0.0226), indicating reduced mechanotransduction in regenerated control neuromast hair cells. By contrast, there was no significant difference in FM1-43 uptake in regenerating vs. unlesioned dexamethasone-treated neuromast hair cells (Fig. 6B, D, E,adjusted P = 0.2747), suggesting that dexamethasone treatment during regeneration may enhance hair cell regeneration and may also slightly impact the maturation of mechanotransduction.

We next examined hair cell cytosolic and mitochondrial calcium homeostasis using the genetically encoded calcium indicator GCaMP3 [14, 36]. When comparing unlesioned and CuSO₄ lesioned neuromasts following regeneration, we observed average cytosolic calcium levels did not differ in neuromast hair cells (Fig. 7A–E), while mitochondrial calcium levels appeared significantly higher in regenerating neuromast hair cells relative to unlesioned in both DMSO and dexamethasone treatment conditions (Fig. 7F–J, adjusted ***P = 0.0006 (DMSO), **P = 0.0019 (Dex); Tukey’s multiple comparisons test). Notably, dexamethasone treatment had no effect on cytosolic or mitochondrial calcium levels relative to control in unlesioned and regenerating neuromasts (Fig. 7E, J). Cumulatively, these data suggest that prolonged dexamethasone exposure does not significantly impact hair cell mechanotransduction or affect calcium homeostasis in either intact or regenerating neuromasts.

To determine whether 48 h dexamethasone exposure impacted mitochondrial function in lateral-line neuromasts, we examined mitochondrial membrane potentials (ΔΨm) in unlesioned neuromast hair cells and supporting cells using tetramethylrhodamine methyl ester (TMRE)—a cell-permeant cationic dye used to assess mitochondrial function [37, 38]. Dexamethasone-exposed neuromasts had significantly higher TMRE mean fluorescence intensity relative to controls (Fig. 8A, B, C,**** P < 0.0001; Welch’s t-test), indicating that mitochondria in dexamethasone-exposed neuromast cells are more hyperpolarized. Elevated TMRE fluorescence was apparent not only in the mitochondria of hair cells (Fig. 8A′, B′, D; EGFP-negative cells) but also in the surrounding supporting cells (Fig. 8A″, B″, E; EGFP-positive cells), suggesting prolonged dexamethasone exposure disrupts normal mitochondrial function in both hair cells and supporting cells. Given that the hair cells in dexamethasone-treated larvae had more negatively charged ΔΨm, we predicted that these hair cells would be more vulnerable to neomycin-induced hair cell death [35, 37]. We therefore compared neomycin-induced hair cell lesion in fish pre-treated with dexamethasone or DMSO carrier alone (Fig. 9A–J). When treated with higher doses of neomycin (50–100 µM), we observed significantly more lateral-line hair cell loss in fish pre-exposed for 48 h to dexamethasone, relative to controls (Fig. 9K, adjusted P = 0.1407 (30 µM), *P = 0.0353 (50 µM), ****P < 0.0001 (100 µM); Šídák’s multiple comparisons test). Thus, while prolonged dexamethasone does not appear to impact cytosolic or mitochondrial calcium homeostasis, it does alter neuromast mitochondrial function and increases hair cell sensitivity to neomycin-induced death.

Hair cells respond to sensory stimuli via deflection of apical hair bundles and opening of mechanotransduction channels. Influx of potassium and calcium occurs through these channels which depolarizes the hair cell, leading to influx of calcium through basally localized presynaptic voltage-gated calcium channels, and triggering release of synaptic vesicles [39‐41]. To determine if prolonged dexamethasone exposure influenced hair cell responses to mechanical stimulation, we measured evoked calcium influx at hair bundles and hair cell synapses in response to bundle deflection using a transgenic line expressing plasma-membrane-targeted GCaMP6s (Fig. 10A [17]). In response to a 500-ms step deflection, calcium influx at the hair bundle was significantly reduced in dexamethasone-treated larvae relative to controls (Fig. 10B, B′,**P = 0.0052,unpaired t-test); an unexpected result given that uptake of the cationic dye FM1-43 was not reduced in dexamethasone-treated larvae (Fig. 5A, B, E). By contrast, presynaptic calcium influx was not significantly different in dexamethasone-treated fish vs. controls (Fig. 10C, D). Baseline GCaMP6sCAAX measurements at neuromast hair bundles and hair cell presynaptic regions were comparable between dexamethasone-treated and control animals (Fig. 10E, F), indicating that calcium levels at rest were not altered in these two subcellular compartments. Overall, these results provide evidence that prolonged dexamethasone exposure does not affect hair cell calcium homeostasis but does negatively influence calcium flow through the hair bundle in response to stimulus.

Exposure to dexamethasone significantly increased the total length of afferent neurites per regenerating neuromast (Fig. 3C–E). The addition in nerve fiber length corresponds to the increased number of hair cells (Fig. 3F), suggesting that nerve fiber length was increased in regenerating neuromasts to accommodate a larger number of newly formed hair cells. Alternatively, it is possible that the additional neurites may be involved in facilitating hair cell generation. A recent study characterizing functional lateral-line neuromast regeneration following CuSO₄ ablation has reported that afferent nerve terminals formed initial contacts with supporting cells during regeneration, suggesting that afferent nerve contacts may help facilitate transdifferentiation of supporting cells into hair cells [26]. However, while dexamethasone treatment caused regenerating neuromasts to add additional afferent dendrite length, no such increase was observed in unlesioned neuromasts (Fig. 3A, B, E), suggesting that the generation of additional hair cells in unlesioned neuromasts is not facilitated by additional nerve fiber contacts. Given that dexamethasone treatment did not increase the number of afferent synapses per hair cell in unlesioned neuromasts (Fig. 4E, gray circles), we posit that afferent neurons innervate and form synapses with newly generated hair cells in unlesioned larval neuromasts without adding additional processes.

We also did not observe inhibition of macrophage activation in response to ototoxic damage following dexamethasone pretreatment (Fig. 5). This result provides additional support to a previously reported observation that dexamethasone likely enhances neuromast regeneration via a mechanism other than immunosuppression [10]. Instead, dexamethasone may be acting through glucocorticoid receptors to enhance division of a subset of hair cell precursors. Further studies will be needed to explore the cellular pathways underlying dexamethasone-induced hair cell generation, and the mechanistic differences between afferent nerve innervation and synapse formation in regenerating vs. intact neuromasts.

Glucocorticoid receptors reside in the cytosol of cells. When activated by a ligand such as cortisol or dexamethasone, they regulate gene expression by localizing to the cell nucleus and either directly binding to DNA or altering the activity of transcription factors or various kinases [42]. Activated glucocorticoid receptors have also been shown to translocate into the mitochondria and regulate mitochondrial gene expression [43], and previous studies of mouse liver and rat distal colon have reported enhanced expression of mitochondrial genes involved in oxidative phosphorylation following dexamethasone treatment [44, 45]. The glucocorticoid receptor gene nr3c1 has been shown to be expressed in larval zebrafish neuromast hair cells and all populations of supporting cells [46]. Speculatively, prolonged dexamethasone exposure may be modulating the expression of mitochondrial DNA-encoded genes involved in oxidative phosphorylation, thereby altering mitochondrial metabolism in neuromast hair cells and supporting cells.

A recent study assessing the effects of cortisol exposure on hair cell sensitivity to neomycin further supports the hypothesis that activated glucocorticoid receptors regulate expression of genes that modulate mitochondrial metabolism [47]. Those authors reported that pretreatment of zebrafish larvae with cortisol for 24 h sensitized neuromast hair cells to neomycin-induced hair cell death, which is comparable to what we observed in dexamethasone-treated fish (Fig. 9). Cortisol-mediated sensitization to neomycin was dependent on glucocorticoid receptor activation and could be blocked by inhibiting protein translation. It is worth noting that Hayward et al. did not report a similar sensitization to neomycin following 24 h dexamethasone exposure, but it is likely dexamethasone and cortisol are acting through a shared mechanism. The difference between their reported observations and our results is probably due to two factors. First, the concentration of neomycin used in Hayward et al. following dexamethasone was 25 µM, which was twofold less than the dose at which we first observed significant effects ([47], Fig. 9K). Also, the dexamethasone pretreatment doses (0.001–1 µM vs. 10 µM) and exposure time (24 vs. 48 h) used by Haywood et al. (2019) were less than those used in the present study.

While we did observe hyperpolarization of mitochondrial membrane potential in dexamethasone-exposed neuromasts (Fig. 8), we did not observe significant changes in cytosolic or mitochondrial calcium homeostasis within hair cells relative to controls (Fig. 7). By contrast, in response to water-jet stimuli, evoked calcium influx at the hair bundle was significantly reduced (Fig. 10B), suggesting calcium influx through activated MET channels is reduced. This observed effect on calcium influx specifically at the hair bundle may be a consequence of altered mitochondrial metabolism. Mitochondria are important regulators of local intracellular calcium levels in sensory hair cells, and in addition to directly taking up calcium, mitochondria provide ATP to drive clearance of calcium through plasma membrane Ca²⁺-ATPases (PMCA1 and PMCA2), with PMCA2 shown to be densely localized in hair cell stereocilia [48]. Zebrafish homologues of PMCA1 and PMCA2 are strongly expressed in mature neuromast hair cells of larval zebrafish [46]. It is possible that altered mitochondrial metabolism may be influencing the activity of PMCA2 in dexamethasone-treated hair cells, thereby modifying PMCA2 regulation of calcium levels at the hair bundle. It is also conceivable that activated glucocorticoid receptors may be modulating expression of genes involved in mechanotransduction channel calcium permeability. Further experiments will be needed to determine the mechanism underlying dexamethasone’s specific effect on calcium influx through stimulated hair bundles.

In conclusion, our observations from zebrafish lateral-line neuromasts indicate that 48 h dexamethasone treatment enhances regeneration and afferent reinnervation but alters mitochondrial homeostasis and evoked calcium influx through hair bundles. Dexamethasone is a steroid commonly used for treating hearing loss, and identifying strategies to retain high levels of dexamethasone for long periods within the inner ear is currently an active area of research. Considering our observations showing that prolonged exposure to dexamethasone had an adverse effect on hair cell mitochondrial metabolism and impacted hair bundle calcium influx, we believe future studies on the effects of long-term, localized dexamethasone treatment on mammalian cochlear function are indicated.