Zebrafish (Danio rerio) were reared and maintained at 28.5°C as Westerfield described (Westerfield, 1995). Two zebrafish lines were used in the study, including the wild-type AB line and the transgenic line Tg(Brn3c:mGFP), where membrane-localized green fluorescent protein (GFP) is specifically expressed in the hair cells. Embryonic stages are defined as described (Kimmel et al., 1995). Embryos were collected after natural spawns. Embryos were moved to embryo medium containing 0.2 mM phenylthiourea at ∼20 h postfertilization (hpf) to prevent pigmentation. All animal procedures were performed according to protocols approved by the Animal Care and Use Committee of Nantong University and were consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

The DNA fragments of zebrafish dusp14 and eya1 were amplified by PCR using the primers (Supplementary Table 3). Then, they were subcloned into the pGEM-T Easy Vector (Promega, United States), and a gene-specific digoxigenin-labeled RNA probe was transcribed in vitro using the DIG RNA Labeling Kit (SP6&T7) (Roche) following the manufacturer’s instructions. The prefixed embryos were incubated with the probe overnight at 4°C. The alkaline phosphatase (AP)-conjugated antibody against digoxigenin (Roche) was used to detect the digoxigenin-labeled RNA probe. The AP-substrate NBT/BCIP solution (Roche) was added to the reaction system to stain the tissues specifically expressing dusp14 gene or eya1 gene in zebrafish.

According to the manufacturer’s instruction, morpholino antisense oligos (MOs; Gene Tools) were prepared at a stock concentration of 1 mM. We designed dusp14 splice-modifying morpholino (dusp14-MO) to knockdown the expression of dusp14 (Supplementary Table 3). MOs were diluted to 0.3 mM and injected into one-cell-stage embryos.

To generate the dusp14 gene mutant zebrafish, as described in our previous work (Gong et al., 2017), 2–3 nL of solution containing specific single-guide RNA (sgRNA) and Cas9 mRNA was injected into one-cell-stage embryos (primers used are listed in Supplementary Table 3).

dusp14 DNA fragments were synthesized by PCR using the primers, dusp14-BamHI-F and dusp14-EcoRI-R (Supplementary Table 3). The DNA fragments were subcloned into the pCS2 + vector, and the recombinant plasmid was linearized using the restriction endonuclease NotI (New England Biolabs). The linearized product was purified as a template and transcribed into mRNA in vitro using the mMESSAGE mMACHINE High Yield Capped RNA Transcription kit (Ambion). Then, the synthesized dusp14 mRNA was coinjected with dusp14-MO into one-cell-stage zebrafish embryos. The rate of rescued zebrafish after injection of dusp14 mRNA was analyzed at 72 hpf.

Total RNAs of all samples were extracted using TRIzol reagent (Invitrogen, United States), and genomic DNA contamination was removed by DNase I (Promega, United States). The RNA yield was determined using NanoDrop ND-2000 (Thermo Fisher Scientific, United States), and integrity was checked on a 1% agarose gel. The cDNA was synthesized using oligo-dT primers and Superscript RT-III (Takara, JP). Quantitative real-time PCR (qRT-PCR) was performed using a Plus Real-Time PCR System (Applied Biosystems, United States). SYBR Prime ScriptTM RTPCR kit (Takara, JP) was used for mRNA qRT-PCR. Data were analyzed using the relative 2^–ΔΔCT method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008). The primers for qRT-PCR are listed in Supplementary Table 3.

Acoustic or vibrational startle protocol was based on previous studies (van den Bos et al., 2017) with appropriate modifications. Larvae (6 dpf) were transferred from petri dishes to wells filled with 1 mL of E3 medium. Both the control and dusp14-MO zebrafish were tested. The protocol (lights on) consisted of 10-min acclimation, followed by 9 acoustic or vibrational stimuli (DanioVision intensity setting) with a 20-s interstimulus interval (ISI). Variable of interest to show the startle response was maximum velocity (mm/s) with 1-s intervals, since the startle response is a short burst of activity best captured by this parameter. When subjects did not show a clear response to the first stimulus (values lower than 15 mm/s), they were discarded from analysis.

The acoustic startle reflex was performed as described previously (Yang et al., 2017). The larvae (5 dpf) were put in a thin layer of culture media in a petri dish attached to mini vibrator. The response of larvae to sound stimulus (a tone burst 9 dB re. m s^–2, 600 Hz, for 30 ms) generated by the vibrator was recorded from above by an infrared camera over a 6-s period. The mean moving distance and peak speed were used to quantify the startle response.

For immunofluorescence staining, the embryos were anesthetized and then fixed using 4% paraformaldehyde. After washing three times with PBS-T, the embryos were incubated in the antigen retrieval solution (Beyotime Biotechnology, China, #P0088) for 15 min at 98°C. Non-specific binding was then blocked with 10% donkey serum (Solarbio, China, #SL050) in PBS-T. Next, specific primary antibodies against GFP (Abcam, #ab13970) and cleaved caspase-3 (CST, #9664), 5-bromo-2′-deoxyuridine (BrdU) (Sigma, #B5002) or SOX2 (Abcam, #ab97959) were added, and secondary antibodies were used to detect the primary antibodies.

TdT-mediated dUTP nick end labeling (TUNEL) assay was performed according to the manufacturer’s instructions (Alexa Fluor 640, cat#: 40308ES20, YEASEN Biotech Co. Ltd) to detect cell death in the HCs of neuromast. In brief, the embryos were anesthetized and then fixed using 4% paraformaldehyde. After washing three times with PBST, 20 μg/mL proteinase K (Roche) was used to treat the embryos. Next, Alexa Fluor 640-12-dUTP Labeling Mix was applied to label the apoptotic cells for at least 3 h. DAPI was applied to label the nucleus.

Images were taken with a Nikon confocal microscope A1R at 40× magnification and were analyzed by Nikon A1R NIS Elements. Exposure settings were adjusted to minimize oversaturation.

The p38 inhibitor (APExBIO, #C5248), with a working concentration at 50 ng/μL, was coinjected with dusp14-MO into the 1–2 cell-stage zebrafish embryos. Additionally, the injected zebrafish were raised at 28.5°C. The development status was recorded with a bright field microscope at about 72 hpf.

All data were analyzed using GraphPad Prism 8.3.0. One-way ANOVA, unpaired Student’s t-tests, and two-way ANOVA were used to determine statistical significance when comparing two groups. The value of p < 0.05 was considered as statistically significant. All data are presented as means with SEM, and all experiments were repeated at least three times.

To study gene expression changes after dusp14 knockdown in zebrafish, we performed transcriptome sequencing. The wild-type zebrafish and dusp14-MO injected zebrafish of 3 days old were prepared, respectively. Three independent replicates of the samples were analyzed for each treatment. The total RNA was extracted using TRIzol reagent (Invitrogen, United States) following the manufacturer’s procedure. The quantity and purity of total RNAs were checked using a Bioanalyzer 2100 and RNA 6000 Nano LabChip Kit (Agilent, CA, United States) with RIN value > 7.0. All RNA samples were submitted to GENEWIZ Science (Suzhou, China), and deep sequencings were performed on an Illumina Hiseq2500.

For RNA-seq data, the Cuffdiff was used to estimate differential expression between samples at the transcript level (Kim et al., 2013). The differentially expressed gene (DEG) was determined with log2 fold change > 1, p-value < 0.05. The R package Clusterprofer was used for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and gene ontology (GO) annotation.

First, to observe the conservation of DUSP14, we constructed an evolutionary tree of 14 species based on the amino acid sequence of DUSP14, including mouse, giant panda, domestic cat, human, big brown bat, Chinese horseshoe bat, sperm whale, sea lion, bottlenose dolphins, lizards, zebrafish, yeast, and nematodes (Figure 1A). The results showed that dusp14 is an extensive conserved gene in many species with DUSP (dual-specificity phosphatases) conserved domain.

Next, to investigate the spatiotemporal expression of dusp14 during the embryonic development, we performed the WISH using a dusp14 antisense probe. As shown in Figures 1B,B′, dusp14 is expressed in the neuromast at 48 hpf. At 72 h (Figure 1C), the dusp14 gene expression is pronounced in the lateral line (Figure 1C′) and the otic vesicle (Figure 1C′′). These results indicate that dusp14 may play an important role in inner ear development.

To explore the effect of the dusp14 gene on zebrafish behavioral response, we designed morpholino oligonucleotides to knockdown the expression of dusp14, and dusp14-MO was validated to effectively reduce dusp14 expression (Supplementary Figure 1). Then, we performed acoustic or vibrational startle test after dusp14-MO injection at 5 dpf. The results showed that the response of dusp14-MO zebrafish to percussion stimuli becomes sluggish, whereas the injection of dusp14 mRNA significantly rescued the phenotype (Figures 2A,B). In addition, there is no difference in movement rate among wide-type zebrafish, dusp14-MO zebrafish, and dusp14-mRNA and MO coinjected zebrafish during the total test (Figure 2C), which suggests that the injection did not affect locomotor activities of zebrafish.

To further investigate the effects of dusp14-MO on hearing dysfunction, we performed the acoustic startle reflex experiment (Figure 2D). The results showed that the swimming trajectory (Figure 2E), swimming distance (Figure 2F), and velocity (Figure 2G) of dusp14 morphants were reduced compared to the controls, which were reversed by dusp14-mRNA. These results indicate that dusp14 is necessary for hearing-related behaviors.

It is well known that hair cells in the inner ear of zebrafish are responsible for the balance perception and hearing. Therefore, to test the underlying cellular mechanisms of dusp14 gene on ear function, we knockdown the dusp14 expression by morpholino in the transgenic zebrafish line Tg(Brn3c:mGFP), in which the GFP were specifically expressed in hair cells (Figure 3A). The results showed that the number of hair cells in three different crista hair cell clusters (anterior crista hair cells: AC; lateral crista hair cells: LC; posterior crista hair cells: PC) was reduced after dusp14-MO microinjection at 72 hpf (Figures 3B,B′), which was neutralized by dusp14 mRNA treatment.

The hair cells in zebrafish are also present in the lateral line system containing the neuromasts that is important to perceive changes in the surroundings. Then, we further investigate the role of dusp14 gene on neuromast formation and found that the number of hair cell clusters in the posterior lateral line of dusp14 morphants was significantly decreased at 72 hpf (Figures 3C,C′). Moreover, the number of the hair cells in the remaining neuromast L3 was also decreased (Figures 3B,B′′). But the dusp14 mRNA treatment interfered these changes remarkably. In addition, we used the eya1 gene to label the neuromast cells (Grant et al., 2005; Whitfield, 2005) in lateral line by WISH and discovered that the number of neuromasts in the posterior lateral line of dusp14 morphants was significantly reduced (Figures 3D,D′). These results indicate that dusp14 gene affects the number of hair cells in zebrafish inner ear and lateral line.

To further validate the function of dusp14 during the hair cell development, the CRISPR/Cas9 system was utilized to knockout dusp14 in wild-type zebrafish. As shown in Supplementary Figure 2A, we chose a sgRNA target site near the translation start codon in the exon1 of dusp14 for CRISPR/Cas9-mediated mutation to abolish the protein translation. Similar to the results of the dusp14 morphants, the number of hair cell clusters and hair cells in the posterior lateral line of the dusp14 mutants was remarkably fewer than that of the control fish at 72 hpf, which was partially reversed by the dusp14 mRNA injection (Supplementary Figures 2C,C′,C′′).

As we know, hair cells of the inner ear and lateral line system in zebrafish regenerate from mitotic supporting cells (Williams and Holder, 2000; Kniss et al., 2016; Thomas and Raible, 2019). Since knockdown of the dusp14 gene caused the reduction of hair cells, we speculated that support cells, resource of hair cells, were also affected by dusp14 morpholino injection.

To test the hypothesis, we analyzed SOX2⁺ cells in the posterior lateral line of dusp14 morphants by immunostaining with anti-SOX2 antibody. The results showed that SOX2⁺ cell number was dramatically decreased after dusp14 morpholino injection, which was substantially reserved by dusp14 mRNA treatment (Figures 4A,A′). Furthermore, we performed the BrdU incorporation to measure the regeneration of supporting cells in the lateral line (Figures 4B,B′). Our results showed that the number of proliferating supporting cells was significantly reduced in the dusp14 morphants. Additionally, the dusp14 mRNA treatment partially interfered these changes. What is more, there is no difference in apoptotic signal between dusp14 morphants and the control, which was immunostained with TUNEL and cleaved caspase 3 in zebrafish neuromasts (Supplementary Figure 3). Taken together, we found that dusp14 gene may regulate the formation of hair cells and ultimately affect the hearing by modulating the proliferation of supporting cells in the zebrafish lateral line (Figure 4C).

To gain a further insight into the molecular mechanism by which dusp14 gene responsible for the regeneration of the supporting cells, we performed RNA sequencing (RNA-seq) of 72 hpf wild-type zebrafish and dusp14-MO zebrafish. Over 40 million valid reads per library on average were obtained after quality filtering, which was, respectively, mapped to about 90% of the zebrafish genome (Supplementary Table 1). The results revealed 2,418 DEGs that might be affected by the absence of dusp14 with 1,787 upregulated DEGs and 631 downregulated DEGs (Figure 5A and Supplementary Table 2). What is more, 12 genes (apoeb, gsta.q, pkma, meis2b, dld, prkcbb, mag, ush1ga, mkm2os.2, fbxo16, pcloa, and cyp26c1) were randomly selected for qRT-PCR analysis to confirm the quality of RNA-seq data. Notably, the expression changes of 6 among 12 genes were consistent with the results of RNA-seq data analysis (Supplementary Figures 4A,B), indicating the reliability of RNA-seq analysis.

Subsequently, we performed KEGG pathway enrichment analysis on the RNA-seq data. The results showed that knockdown dusp14 gene by morpholino mainly resulted in changes in various metabolic pathways (Supplementary Table 2), including carbon metabolism, tryptophan metabolism, glycine metabolism, serine metabolism, tyrosine metabolism, retinol metabolism, metabolism of xenobiotics by cytochrome P450, phenylalanine metabolism, and so on. Furthermore, the most significant GO analysis enrichment terms were oxidoreductase activity, which were assigned as molecular functions. The cell membrane was the most significant GO enrichment term assigned as cellular component, and regulation of transportation was the most significant GO enrichment term assigned as a biological process (Figure 5B).

The DUSP family targets to MAPK signaling pathway, which modulate diverse cellular functions, such as regeneration, differentiation, and apoptosis. Therefore, we observed the gene expression changes in p38 kinases, ERKs, and JNKs signaling pathway in RNA-seq data. As is shown in Figure 5C, the genes directly regulated by dusp14 dephosphorylation include mapk8b, mapk9, mapk12a, and mapk12b, which are mainly in the JNK and p38 signaling pathways. Among them, mapk12b belonging to p38 signaling pathway showed the most significant difference, which suggests that p38 signaling pathway may play a vital role in dusp14 regulation of hearing in zebrafish.

To further explore whether p38 signaling is responsible for dusp14 regulation of hearing in zebrafish, the wild-type and dusp14-MO zebrafish were treated with p38 inhibitor. As previously shown, although knockdown dusp14 gene expression by morpholino significantly decreased the number of hair cell clusters (Figures 6A,A’), the number of hair cells in inner ear cristae and neuromast L3 in the posterior lateral line, p38 inhibitor dramatically reversed these changes (Figures 6B,B’,B”). Additionally, p38 inhibitor also blocked the effect of dusp14-MO on the number of supporting cell and the proliferation of supporting cells (Figures 6C,C’,D,D’). These results indicate that p38 signaling pathway is necessary for the dusp14 regulation of hearing.