The zebrafish used in this study was kindly provided by Dr. Liu Dong from Nantong University (Nantong, China). Zebrafish wild-type AB line and Tg (brn3c: GFP) line was fed twice a day and kept in a fully automated zebrafish rearing system (ESEN, China; temperature 28 ± 0.5°C, pH 7.0, and conductivity 500 μS) with a photocycle of 14-h light and 10-h dark. Embryos were placed in 0.003% (w/v) 1-phenyl-2-thiourea (PTU) within 24 h to prevent pigmentation.

The suckling mice used in this study were postnatal 3 days old (P3) C57BL/6 mice. The P3 mice were purchased from the Animal Center of Shandong University (Jinan, China). After inhalation anesthesia, mice were acutely decapitated with dissecting scissors, and then quickly used for dissection and further experiments. Laboratory animal Use Permit [SYXK(LU) 2019 0005].

The HEI-OC1 cells, a commonly used auditory cell line with the characteristics of hair cells, were used to investigate the molecular mechanism of auditory development (Gao et al., 2020; Zhong et al., 2020). The HEI-OC1 cells were cultured in the mixture of 5% volume of FBS without antibiotics and high-sugar DMEM (Gibco BRL, Grand Island, NY, USA) and kept in a 5% CO₂ incubator at 33°C.

The zebrafish immunohistochemistry experiments were performed based on the methods previously reported (Campbell et al., 2006). The antibodies used in this study included Alexa Fluor 488 antibody (1:1,000; Invitrogen), anti-myosin VIIa antibody (1:200; DSHB), Alexa’s anti-SOX2 antibody (1:500; abcam), and Fluor 594 secondary antibody (1:1,000; Invitrogen). Larvae were incubated with DAPI (1:500 dilution; Thermo Fisher Scientific) for 20 min to label the nucleus. Cells were fixed with 4% PFA for 30 min, then permeabilized with 1% Triton X-100 for 5 min and blocked with 1% BSA in phosphate buffered saline (PBS) for 1 h. Then, the cells were incubated with primary anti-Loxhd1 antibody (bs-18343R, Bioss), anti-BDNF antibody (orb318730, biorbyt), and anti-Myosin VIIa antibody (138-1, DSHB) at 4°C overnight, and then with both fluorescent secondary antibody and DAPI (D9542, Sigma) in dark for 1 h, and washed with PBS three times. The samples were observed under a laser scanning confocal microscope (Germany, LEICA).

The whole-body in situ hybridization (WISH) was performed based on the standard procedures (Nasevicius and Ekker, 2000; He et al., 2021). First, the Loxhd1b specific primers (forward primer 5′-GTGGTGGATGATGAGGAGATG-3′ and reverse primer 5′-TGTCCGACTCGATCACTCTG-3′) and the BDNF specific primers (forward primer 5′-AGGAGTTGCTTGAGGTGGAA-3′ and reverse primer 5′-TATCCATGGTAAGGGCTCGC-3′) were used to amplify the target fragments from zebrafish cDNA and transform into the pGEM-T Easy vector. Then, the DIG RNA labeling kit was used to transcribe the gene-specific digoxin-labeled RNA probe in vitro. The pre-immobilized embryos were incubated with the probe overnight at 4°C. Second, an alkaline phosphatase (AP)-conjugated anti-digoxigenin (Roche, #11093274910) antibody was used to detect the digoxin-labeled RNA probe. The samples were observed with pictures taken with the stereo microscope MVX10 (Japan, Olympus).

Studies have shown the successful method used to knock down targeted genes in zebrafish using morpholino (MO) (Summerton and Weller, 1997; Summerton, 1999; Leitner, 2014). Meanwhile, the MO-mediated gene downregulation method has been well applied in Xenopus and zebrafish embryos (Summerton and Weller, 1997; Summerton, 1999). However, many limitations in the application of MO are revealed, i.e., the downregulation of gene expression is not achieved through stable genetic manipulation, causing unstable efficiency with the relatively short maintenance time, only about 3 days (Summerton, 1999; Bill et al., 2009). Gene Tools was used to synthesize Loxhd1b MO (5′-CAGTGATGCTGAAAGTCTGACTTGC-3′). Embryos obtained from natural mating of Tg (Brn3c: GFP) zebrafish were used for microinjection. The MO was diluted with RNase-free water and injected into embryos at single-cell stage, which were then cultured in egg water medium at 28.5^°C.

Zebrafish hair cells were labeled with viability dye FM1-43FX (Molecular Probes, Invitrogen), which was a fixable analog of N-(3-triethylammonium propyl)-4-(4-(dibutyl) (Amino)-styryl) pyridinium dibromide (FM1-43). FM1-43FX was thought to mark hair cells by traversing mechanically sensitive channels (Liu et al., 2015; Yan et al., 2017). In this experiment, in order to label the hair cells in neuroma, the larvae were placed in embryo medium containing 0.3 μg/ml FM1-43FX for 30 s, and washed 3 times in fresh embryo medium. Finally, the fish was anesthetized with 0.02% Tricaine (MS-222; Sigma) to observe the hair cells.

The HEI-OC1 cells were cultured in a 100 mm cell culture dish, and after the treatment of RNA interference, the cells were collected by centrifugation. Then, PMSF was added to the cytoplasmic protein lysate, and the cells were transferred into 0.22 μm PVDF and incubated with primary antibody BDNF (ab108319; abcam), anti-TrkB (bs-3732R; bioss), anti-ERK (12754S; CST), anti-β-actin (ab8226; abcam) diluted 1:1000 (4°C, overnight), and anti-rabbit IgG secondary antibody (ZB-2301; zsbio) at 25°C for 1.5 h. After washing, the immune complexes were detected using a chemiluminescence imaging system (model: Al480; Bio-Rad).

The total RNA of the samples between Loxhd1b knock-down and wild type zebrafish were extracted with TRIzol reagent (Invitrogen) and analyzed by Agilent 2100 Bioanalyzer (Agilent Technologies, USA). The Illumina HiSeq platform was used to prepare and sequence libraries according to the manufacturer’s instructions (Illumina, USA). The sequences were processed and analyzed by GENEWIZ with the p-value set to < 0.05 to reveal differentially expressed genes (DEGs).

Total RNA was extracted from zebrafish embryos by TRIzol reagent according to the manufacturer’s instructions (Thermo Fisher Scientific, USA) with the genomic contaminations eliminated by DNaseI. Total RNA (2 μg) was reversely transcribed using a reversed first strand cDNA synthesis kit (Fermentas, USA) and stored at –20^°C. The sequences of PCR primers used for validating the splice-blocking MO were forward primer 5′-GAAAGCGGAGAAGGAGGACT-3′ and reverse primer 5′-CAGCCGTGAACACACTCACT-3′. The housekeeping gene ef1a amplified with forward primer 5′-CTTCAACGCTCAGGTCATCA-3′ and reverse primer 5′-CGGTCGATCTTCTCCTTGAG-3′ was used as internal control to produce a 126-bp amplification product.

The hearing and balance functions of zebrafish are mainly related to the development of the inner ear (De Rose and Farber, 2003; Huang et al., 2013). We performed a fast escape reflex and used two near-field pure tone stimuli with different sound intensity levels to perform hearing tests on MO-injected zebrafish (Kohashi and Oda, 2008). This experiment was tested in a 96-well plastic plate and recorded with a high-speed camera (Redlake, MotionScope M3, 1000 fps) under infrared light illumination. Larvae in each group was tested 15–18 times with the movement distance after the startled reaction calculated.

All data were analyzed using GraphPad Prism 8.0.2. Statistical analysis was performed by one-way analysis of variance (ANOVA) and Student’s t-test or Chi-square test with the p-value < 0.05 considered statistically significant. Each experiment was repeated with 3 or more biological replicates.

Initially, the results of WISH experiments showed that the expression of Loxhd1b in zebrafish was detected by digoxigenin-labeled Loxhd1b probe (Figure 1). The Loxhd1b gene was highly conserved and was expressed in zebrafish ear vesicles at 24 h post-fertilization (hpf) and over time, was also detected in the olfactory pores (Figure 1A). In addition, the results of real-time PCR (RT-PCR) (Figure 1B) showed that the expression of Loxhd1b was gradually decreased from 24 to 96 h over time, which were consistent with the results by the in situ hybridization.

Results of gene knockdown analysis showed that MO effectively affected the splicing of Loxhd1b precursor mRNA, resulting in deletion of exon 2 and reading frame shifted (Figures 2A,B). To further investigate the status of hair cells, we observed and counted the number of three inner ear hair cell clusters, i.e., the anterior crest (AC), posterior crista (PC), and lateral crest (LC) (Figure 2C). By using genetically modified fish, we visually evaluated the changes in zebrafish inner ear hair cells after the Loxhd1b gene knockdown. The confocal results showed that 3 days after the injection of Loxhd1b MO, the number of hair cells in each nerve cluster in the lateral line of zebrafish was significantly reduced (Figures 2D,G). In the 72-hpf zebrafish inner ear hair cells, we counted the total number of hair cell clusters of morphant (15) and WT (32) (Figures 2E,I). We also counted the total number of hair cell clusters in the inner ear of zebrafish at 96-hpf of MO (15) and Ctrl (32) (Figures 2E,J). Since the hair cells on the lateral line play a vital role in sensing changes in the surrounding environment, therefore, we further performed WISH experiments based on eya1 specific probes (Figures 2F,H). The results showed that the Loxhd1b gene knockdown caused decreased number of neuromas in zebrafish. Furthermore, we designed a C-shaped startle response to explore the effect of Loxhd1b on zebrafish hearing. Results showed that the movement distance of Loxhd1b-morphant zebrafish C-startle response was significantly lower than that of WT zebrafish (Figure 2K).

Because the sensory cells of the lateral neuromound are composed of hair cells and supporting cells, we then explored the effect of the knockdown of Loxhd1b in zebrafish on the formation and maturation of hair cells and supporting cells in lateral neuroma. The results showed that the number of hair cells in the lateral line neuroma labeled with FM1-43FX were significantly lower in Loxhd1b-morphant zebrafish than that in the WT group (Figures 3A,D,E). Meanwhile, we performed the immunohistochemical experiments to label Sertoli cells and hair cells using SOX2 antibody and myosin-7a antibody. The results showed that the number of hair cells in Loxhd1b-morphant zebrafish lateral line neuroma was significantly reduced (Figures 3B,F) as well as the number of support cells (Figures 3C,G).

To identify the factors that cause the reduction of zebrafish inner ear and lateral line hair cells, previous studies have reported that the loss of hair cells may be caused by the apoptosis of hair cells (Chen et al., 2003; Slattery and Warchol, 2010; Liu et al., 2016; Wang et al., 2017) with the TUNEL test (Stooke-Vaughan et al., 2015; Wang et al., 2017; Li et al., 2018; Yang et al., 2019; Zhang et al., 2019) for diagnosis of apoptosis; alternatively, the loss of hair cells may be due to the abnormal proliferation of support cells, while the hair cells were differentiated from support cells. Therefore, we used TUNEL to test the Loxhd1b knockdown zebrafish model, and found no TUNEL-positive cells, indicating no apoptosis of hair cells (Figure 4). Further studies showed that the abnormal function of Loxhd1b caused the hair cell dysplasia was not due to apoptosis of hair cells but by reduced proliferation of support cell.

It was reported previously that the abnormal development of otoliths led to balance disorders (Stooke-Vaughan et al., 2015). As pre-in situ hybridization showed significant expression of Loxhd1b in the ear vesicles, we further used confocal microscopy to observe the morphological development of ear vesicles and otoliths in zebrafish at 72 hpf. The results showed that the size of the 72 and 96 hpf ear capsules of Loxhd1b morphant was significantly smaller than the control group (Figures 5A,C). In addition, compared with the control group, the number and shape of otoliths in 77% of Loxhd1b morphant showed a significant reduction or defect (Figures 5B,D).

Transcriptome sequencing was performed based on the zebrafish embryos injected with Loxhd1b-morphant at 120 hpf and the control group. The results showed that compared with the control group, a total of 212 and 297 genes were significantly up- and down-regulated in Loxhd1b-morphant, respectively (Figure 6A). These results were further verified by real-time PCR analysis to reveal the difference between the BDNF gene at 72 and 96 hpf Loxhd1b-morphant and the control group, respectively (Figure 6B). The results of the enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (He et al., 2021) of these DEGs revealed a total of 30 metabolic pathways enriched, e.g., the MAPK signaling pathway (Figure 6C). In order to explore the relationship between the BDNF/TrkB/ERK signaling pathway and the effect of Loxhd1b on auditory abnormalities, we used the TrkB inhibitor K252a and the ERK inhibitor UO126 to act on the 293T cells. The results of western blot showed that BDNF regulated the activation of its specific receptor TrkB, i.e., the overexpression of BDNF gene enhanced the activation of TrkB, which was required to mediate BDNF’s involvement in Loxhd1b regulation of auditory development, whereas BDNF could not rescue Loxhd1b expression after TrkB was blocked. The results of western blotting experiments further showed that BDNF could regulate the activation of ERK, i.e., the overexpression of BDNF enhanced the activation of ERK, which was required for the regulation of auditory development by Loxhd1b. After the ERK was blocked, BDNF could not rescue the expression of Loxhd1b (Figures 6D–H). These results suggested that the regulation of Loxhd1b expression by BDNF blocked the TrkB/ERK signaling pathway, causing the auditory abnormalities, while Loxhd1b did not affect BDNF expression.

In order to further determine the relationship between BDNF and Loxhd1b, we first explored the expression location of BDNF in zebrafish tissue by whole embryo in situ hybridization. The results were consistent with those of Loxhd1b, i.e., both BDNF and Loxhd1b were expressed in the inner ear and olfactory pore of zebrafish (Figure 7A).

We also detect the localization of both Loxhd1b and BDNF by immunofluorescence analysis with the results observed under confocal microscopy, showing that Loxhd1b and BDNF proteins were highly enriched in hair cell cytosol and slightly distributed in nucleus (Figure 7B). Next, we used 3-day old C57BL/6 mouse basement membrane for immunostaining and specific immunostaining with myosin VII and Phalloidin. The observation under a confocal microscope showed that Loxhd1b and BDNF were specifically expressed in inner and outer hair cells (Figures 7C,D).

In order to verify the rescuing effect of the exogenous BDNF injection on the absent phenotype of hair cell development in Loxhd1b morphant, we injected both the synthetic BDNF mRNA and Loxhd1b MO into zebrafish embryos at the single-cell stage in vitro, and performed confocal observations in 3 days. Results showed that the number of hair cell clusters in the ear follicle and lateral line of Loxhd1b morphant were partially restored (Figures 8A–G). Similarly, compared with Loxhd1b morphant, co-injection of BDNF mRNA also partially rescued abnormal phenotypes, including smaller ear follicle size and a decrease in the number of hair cells in the ear follicle (Figures 9A–D).