Abstract
RNA-binding proteins (RBPs) are important for posttranscriptional RNA processing, including pre-mRNA alternative splicing, mRNA stability, and translation. Several RBPs have been shown to play pivotal roles in the inner ear, whose dysfunction leads to auditory and/or balance impairments. Epithelial splicing-regulatory protein 1 (ESRP1) regulates alternative splicing and mRNA stability, and mutations in ESRP1 gene have been associated with sensorineural hearing loss in humans. In Esrp1 knock-out mouse embryos, alternative splicing of its target genes such as Fgfr2 is impaired, which eventually result in cochlear development deficits. However, Esrp1 knock-out mice die soon after birth because of complications from cleft-lip and palate defects, impeding further investigations at later postnatal ages. In the present study, we explored the role of ESRP1 in hearing using zebrafish as a model. We showed that esrp1 and its paralog esrp2 are expressed in the inner ear and certain anterior lateral line (ALL) neuromasts. Furthermore, our data suggested that Esrp1 and Esrp2 are required for the mechano-electrical transduction (MET) function of hair cells. RNA sequencing results indicated a significant decrease in the levels of several mRNAs in esrp1/2 double knock-out larvae. Among the dysregulated genes are tmc1 and tmc2a, which encode essential subunits of the MET complex. Further investigations demonstrated that Esrp1/2 could directly bind to tmc1 and tmc2a mRNAs and affect their stability. Taken together, we showed here that Esrp1 and Esrp2 regulate the MET function of zebrafish sensory hair cells by modulating the stability of tmc1 and tmc2a mRNAs.
Significance Statement
ESRP1 is an important RNA-binding protein, whose malfunction has been associated with hearing loss in humans. Esrp1 knock-out affects alternative splicing of its target mRNAs such as Fgfr2, eventually leading to cochlear development deficits in mice. However, Esrp1 knock-out mice die soon after birth, precluding further investigations at later postnatal ages. In this study, we explored the role of ESRP1 in hearing using zebrafish as a model. Our results demonstrated that esrp1 and its paralog esrp2 are expressed in the zebrafish inner ear and that esrp1/esrp2 double knock-out compromised the mechano-electrical transduction (MET) function of hair cells. Additionally, we successfully identified tmc1 and tmc2a mRNAs as the targets of Esrp1/2, which encode essential subunits of the MET complex.
Introduction
Hair cells serve as the mechanosensitive receptor cells responsible for detecting sound and balance information (Kazmierczak and Müller, 2012). At the apical surface of each hair cell reside multiple F-actin-based stereocilia and a single tubulin-based kinocilium, collectively known as the hair bundle (Barr-Gillespie, 2015). The stereocilia play a central role in mechano-electrical transduction (MET) that converts the mechanical inputs into electrical signals (Hudspeth and Jacobs, 1979). The stereocilia are organized in rows of increasing height, and various types of extracellular links provide connections between individual stereocilia (Goodyear et al., 2005). At the tips of the shorter-row stereocilia localize the MET channels (Beurg et al., 2009). Deflection of the stereocilia by mechanical force changes the open probability of the MET channels, resulting in the influx of cations into hair cells (Hudspeth and Jacobs, 1979; Kazmierczak and Müller, 2012). Several proteins have been identified as integral components of the so-called MET protein complex, including TMC1/2, TMIE, LHFPL5, and CIB2/3 (Kawashima et al., 2011; Xiong et al., 2012; Pan et al., 2013; Zhao et al., 2014; Giese et al., 2017; Michel et al., 2017; Wang et al., 2017, 2023a; Pan et al., 2018; Cunningham et al., 2020; Jia et al., 2020; Liang et al., 2021; Fu et al., 2025; Giese et al., 2025).
RNA-binding proteins (RBPs) are important for posttranscriptional RNA processing, ranging from pre-mRNA alternative splicing and mRNA stability to translation (Hentze et al., 2018). Several RBPs have been identified as indispensable in the inner ear, whose dysfunction usually leads to auditory and/or balance deficits (Shi et al., 2022). For example, mutations in the gene encoding epithelial splicing-regulatory protein 1 (ESRP1) have been associated with sensorineural hearing loss in humans (Rohacek et al., 2017). ESRP1 belongs to a highly conserved family of RBPs that regulate alternative splicing in epithelial tissues (Derham and Kalsotra, 2023). In Esrp1 knock-out mouse embryos, alternative splicing of its target genes such as Fgfr2 is impaired, which eventually leads to cochlear development deficits (Rohacek et al., 2017). However, Esrp1 knock-out mice die soon after birth because of complications from cleft-lip and palate defects, precluding further analysis at later postnatal ages (Bebee et al., 2015).
Zebrafish have emerged as a powerful animal model for hearing research (Nicolson, 2005; Pickett and Raible, 2019). In the inner ear of zebrafish, sensory hair cells are located in the maculae of the utricle and saccule as well as in the cristae of the three semicircular canals (Nicolson, 2005). Meanwhile, hair cells also populate the neuromasts of the two lateral line systems—the anterior lateral line (ALL) and the posterior lateral line (PLL)—on the body surface of zebrafish (Nicolson, 2005; Pickett and Raible, 2019). Owing to the teleost-specific genome duplication, zebrafish possess more protein-coding genes than mammals (Meyer and Schartl, 1999; Howe et al., 2013). For instance, zebrafish harbor three tmc1/2 paralogs, namely, tmc1, tmc2a, and tmc2b, whose expression patterns vary across different hair cell types (Maeda et al., 2014; Smith et al., 2023). In line with this, when different tmc genes are inactivated in zebrafish, the MET function of different hair cell types is compromised (Chou et al., 2017; Chen et al., 2020; Smith et al., 2020; Zhu et al., 2021; Kindig et al., 2023). On the other hand, each esrp paralog (esrp1 and esrp2) has a single copy in zebrafish, paralleling the mammalian scenario (Burguera et al., 2017).
In the present study, we endeavored to explore the role of ESRPs in hair cells using zebrafish as a model. We showed that esrp1 and esrp2 are expressed in the inner ear and certain ALL neuromasts of zebrafish. Moreover, our data suggested that Esrp1 and Esrp2 are required for the MET function of hair cells. RNA sequencing (RNA-seq) results indicated that the levels of several transcripts including tmc1 and tmc2a were significantly decreased in esrp1/2 double knock-out (DKO) larvae. Further investigations revealed that Esrp1/2 could directly bind to tmc1 and tmc2a mRNAs and regulate their stability.
Materials and Methods
Zebrafish
All procedures were conducted in accordance with the guidelines approved by the Animal Ethics Committee of Shandong University School of Life Sciences. The esrp1 and esrp2 knock-out zebrafish was generated using the CRISPR/Cas9 technique. Briefly, genomic DNA sequences 5′-TGAACTCTCGGAAGAGTGTCAGG-3′ in exon 2 of the esrp1 gene and 5′-GTGCAAATGGGGGTAAACTGGGG-3′ in exon 1 of the esrp2 gene were selected as the targets for the guide RNAs (gRNAs). The Cas9 mRNA and gRNAs were in vitro transcribed and subsequently injected into one-cell stage embryos. Genomic DNA was extracted from the tail fins of F0 fish and used as template for polymerase chain reaction (PCR) with the following primers: 5′-ATTGGTTAAGGCCGGTCAGG-3′ and 5′-GGATGCAAGACTTGACGCAC-3′ for esrp1, 5′-CAGGTAGAGTCTCGTGCCG-3′ and 5′-TGTTTTGATATCTCAGACGGTG-3′ for esrp2. The PCR products were subjected to Sanger sequencing to identify successful knock-out fish. The brn3c:GFP transgenic zebrafish were generated as previously described (He et al., 2017).
In situ hybridization
The cDNAs of esrp1 (NM_001080576.1, 252–1,098 nt), espr2 (NM_199272.1, 2,175–2,974 nt), tmc1 (NM_001312681.1, 2,824–3,643 nt), tmc2a (NM_001302237.1, 2,680–3,257 nt), tmc2b (NM_001302223.1, 2,644–3,462 nt), and myo7aa (NM_152983.1, 4,594–5,308 nt) were cloned into the pZeroBack vector (Tiangen) and used as DNA templates for RNA probe synthesis. Digoxigenin-labeled antisense probes were in vitro transcribed using DIG RNA labeling mix (Roche, catalog #11277073910) according to the manufacturer's protocol. Zebrafish larvae were anesthetized with 0.17 mg/ml Tricaine (Sigma-Aldrich, catalog #MS-222), followed by fixation with 4% paraformaldehyde (PFA) at 4°C overnight. After incubation with 10 μg/ml proteinase K at 37°C for 10 min, samples were incubated with digoxigenin-labeled probes at 60°C overnight, followed by incubation with anti-digoxigenin antibody (Roche, catalog #11093274910) at 4°C overnight. After signal development in NBT/BCIP solution (Roche, catalog #11681451001), samples were mounted in 50% glycerol/PBS and imaged with an Olympus IX53 microscope.
Immunostaining and confocal microscopy
Zebrafish larvae were anesthetized and fixed as aforementioned. After incubating in acetone at −20°C for 10 min, the larvae were blocked with 10% goat serum at room temperature for 1 h. For visualization of the stereociliary F-actin core of inner ear hair cells, the larvae were incubated with TRITC-conjugated phalloidin (Sigma-Aldrich, catalog #P1951) at room temperature for 1 h. For visualization of the lateral line hair cells, the larvae were incubated with Tris-HCl (150 mM, pH 9.5) at 70°C for 15 min before incubation with acetone and blocking. After that, the larvae were incubated sequentially with rabbit anti-MYO7A antibody (Proteus BioSciences, catalog #25-6790) at 4°C overnight and Alexa Fluor 594-conjugated donkey anti-rabbit IgG (Thermo Fisher Scientific, catalog #R37119) at room temperature for 1 h. The samples were mounted in 1% low-melting agarose and images were taken using a confocal microscope (Zeiss, LSM900). Z-stack projections were obtained by using the z-projection function, and image volumes were acquired at 0.5 μm intervals along the z-axis. Hair cells were imaged with a 0.95 NA/40× Kort M27 objective lens. Higher-resolution images of hair cells were taken with a 0.95 NA/63× Kort M27 objective lens.
FM4-64 and FM1-43FX uptake assay
For inner ear hair cells, zebrafish larvae were anesthetized as aforementioned, and then 1 nl of FM4-64 (1.2 μM, MCE, catalog #HY-103466) was microinjected into the otic capsules. The labeled crista hair cells were visualized using a confocal microscope (Zeiss, LSM900). For lateral line neuromast hair cells, zebrafish larvae were incubated with 1.2 μM FM1-43FX (Thermo Fisher Scientific, catalog #F35355) in embryo medium for 30 s. After rinsing three times in fresh embryo medium, the larvae were anesthetized with Tricaine, and the labeled hair cells were visualized using a confocal microscope (Zeiss, LSM900).
RNA-seq analysis and reverse transcription (RT)-PCR
RNA-seq analysis was performed by Genewiz as previously described (Zhang et al., 2020). Briefly, total RNAs were extracted from the otic vesicles of zebrafish larvae at 34 hours postfertilization (hpf) using TRIzol reagent (Invitrogen, catalog #15596018CN), and mRNAs were enriched using the Poly(A) mRNA Magnetic Isolation Module (NEB), followed by mRNA fragmentation and cDNA synthesis. The cDNA libraries were constructed using Ultra RNA Library Prep Kit for Illumina (NEB), and 150 bp paired-end sequencing was performed on Illumina HiSeq. The sequencing results were filtered using Cutadapt (v1.9.1) and aligned with reference genome using Hisat2. Differential expression was analyzed using Htseq and DESeq2. At last, RT-PCR and quantitative PCR (qPCR) were performed to verify the sequencing results using primers specific for candidate genes. The sequences of primers are as follows: 5′-GAGGGCAGAAGAGGAGATTG-3′ and 5′-CAGCAGTTGTTCAGGAATCG-3′ for tmc1, 5′-AGATGAAGAGGATGAGTCCA-3′ and 5′-GATGCAAGCCGTTCTAAAAT-3′ for tmc2a, 5′-ACCCTCGAATCGCACTAAAG-3′ and 5′-TGGATGTCGGAAACAGTGAG-3′ for tmc2b, 5′-AGCCAAAAGGTCATTATCGT-3′ and 5′-CATCACTTCTTAGAGCGACA-3′ for gas2b, 5′-CAGCTTTCTGAGAGCCTACC-3′ and 5′-GCGTGAGTCCAGCTTGTAAT-3′ for ndrg1b, 5′-GTGTCTGACAGGTTACAGGC-3′ and 5′-CTCCATGGGAGCTGATTTGG-3′ for otop2, 5′-CACGCTCAATTTGCTTCTAT-3′ and 5′-CGCAGCTCATCATTAGGATT-3′ for zpld1a, 5′-AAATGGCTCAGACCTTTCGT-3′ and 5′-AGCTTGCGCATAAAATCAGC-3′ for si:ch73-59f11.3, 5′-GTACATGGTGGCATCAGGAG-3′ and 5′-CTTTACCCTTGAGCTCGGTC-3′ for gucy2d, 5′-GCTCGGGCATAAACAGC-3′ and 5′-CTGGAGTCTGACGACAC-3′ for fgfr2b, 5′-TTGGGATCTCCTATCACACT-3′ and 5′-CTGGAGTCTGACGACAC-3′ for fgfr2c, 5′-CACAGTGCTGTCTGGAGGTAC-3′ and 5′-GAGGGCAAAGTGGTAAACG-3′ for β-actin.
mRNA stability assay
HEK293T cells were cultured in the Dulbecco's modified Eagle's medium (DMEM, Invitrogen, catalog #C11995500BT) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Lonsera, catalog #S711-001S) and 1% penicillin/streptomycin (P/S) at 37°C in a 5% CO2 humidified atmosphere. Vectors expressing Esrp1/2 and tmc1/2a/2b mRNAs were transfected into cultured cells using LipoMax transfection reagents (Sudgen, catalog #32012). Twenty-four hours after transfection, cells were treated with actinomycin D (ActD; 5 μg/ml, Gene-View, catalog #AA007) for 8 h. Total RNA was extracted using TRIzol reagent (Invitrogen) and used as template for qPCR using primers specific for tmc1/2a/2b.
RNA immunoprecipitation (RIP)
RIP experiments were performed as previously described (Zhang et al., 2020; Wang et al., 2023b). Briefly, HEK293T cells were cultured in DMEM supplemented with 10% heat-inactivated FBS and 1% P/S at 37°C in a 5% CO2 humidified atmosphere. Vectors expressing full-length tmc1/2a/2b cDNAs or various tmc1 cDNA fragments were transfected into cultured cells together with Myc-Esrp1/2 expression vectors using LipoMax transfection reagents (Sudgen). The transfected cells were lysed in lysis buffer containing RNase inhibitor (Takara, catalog #2313A). After centrifugation, the supernatant was collected and then incubated with immobilized anti-Myc antibody (Sigma-Aldrich, catalog #E6654). The immunoprecipitated RNA was used as template for RT-PCR analysis using primers specific for tmc1/tmc2a/2b.
Statistical analysis
Data are shown as means ± standard error of mean (SEM). Student's t test was used for statistical analysis, and p < 0.05 was considered statistically significant.
Results
The espr1 and esrp2 transcripts are expressed in the inner ear as well as the ALL neuromasts of zebrafish
We first examined the expression patterns of the esrp1 and esrp2 transcripts in zebrafish larvae by performing whole-mount in situ hybridization. The esrp1 transcripts were detected in the otic vesicles and pharynx at 36 hpf (Fig. 1A–A’’). By 48 hpf, the expression of esrp1 in the otic vesicles and pharynx became more pronounced (Fig. 1B–B’’). When examined at 60 and 72 hpf, the expression level of esrp1 in the otic vesicles slightly declined; meanwhile, its expression in certain ALL neuromasts such as O1, D1, and D2 became evident (Fig. 1C–D’’).
The esrp1 and esrp2 transcripts are expressed in the inner ear and ALL neuromasts. A–D’’, Expression of esrp1 transcripts was examined by performing in situ hybridization of zebrafish embryos at 36 hpf (A–A’’), 48 hpf (B–B’’), 60 hpf (C–C’’), and 72 hpf (D–D’’). E–H’’, Expression of esrp2 transcripts was examined by performing in situ hybridization of zebrafish embryos at 36 hpf (E–E’’), 48 hpf (F–F’’), 60 hpf (G–G’’), and 72 hpf (H–H’’). The inner ear is indicated with arrows, and ALL neuromasts are indicated with arrowheads. Scale bars, 0.5 mm (A–H) and 0.2 mm (A’–H’, A’’–H’’). Extended Data Figure 1-1 for the in situ hybridization results of myo7aa in the zebrafish embryos at different developmental stages.
The esrp2 transcripts exhibited a similar yet not entirely identical expression pattern. At 36 hpf, it was only detected in the otic vesicles (Fig. 1E–E’’). At 48 and 60 hpf, it was robustly expressed in both the otic vesicles and pharynx (Fig. 1F–G’’). At 72 hpf, the expression of esrp2 in the pharynx was decreased (Fig. 1H–H’’). One prominent difference between the expression patterns of esrp1 and 2 was that only esrp1, but not esrp2, was detected in the ALL neuromasts at 60 and 72 hpf (Fig. 1C–D’’,G–H’’). Meanwhile, neither esrp1 nor esrp2 transcripts were detected in the PLL neuromasts up to 72 hpf, which were labeled by myo7aa probes (Fig. 1, Extended Data Fig. 1-1). Collectively, our data suggested that the esrp1 and esrp2 transcripts are expressed in the otic vesicles and certain ALL neuromasts of zebrafish.
Disruption of esrp1 and esrp2 affects otolith and semicircular canal development
To explore the function of esrp1/2 in hair cells, we generated esrp1 and esrp2 knock-out zebrafish employing the CRISPR/Cas9 technique. The sgRNAs were targeted at exon 2 of the esrp1 gene and exon 1 of the esrp2 gene. For esrp1, we selected a mutant allele with a 11-base-pair (bp) deletion (Fig. 2A). For esrp2, we selected a mutant allele with a 10-bp deletion (Fig. 2C). Both of these deletions would cause frameshift and premature translational termination, giving rise to truncated proteins that lack all the RRM domains (Fig. 2B,D). We cloned the wild-type (WT) and mutant esrp1/2 cDNAs into expression vectors to express Esrp proteins with a Myc epitope fused to their C terminal ends in cultured cells. Western blot analysis using an anti-Myc antibody detected a band with a molecular weight of ∼90 kDa from WT- but not mutant esrp1/2-expressing cells, confirming the successful knock-out (Fig. 2E). We crossed single knock-out zebrafish to each other to obtain esrp1/2 double knock-out (DKO) zebrafish. The esrp1/2 single knock-out zebrafish larvae demonstrated normal viability; the DKO zebrafish larvae survived until 8 dpf, but all died between 8 and 13 dpf (Fig. 2F). This provided us with a time window to assess the development and function of hair cells in the DKO larvae.
Construction of the esrp1 and esrp2 knock-out zebrafish. A, Schematic drawing of the strategy for construction of the esrp1 knock-out zebrafish. Exons are indicated with numbers. The target of sgRNA in the esrp1 gene is indicated with red color, and the PAM sequence is highlighted in yellow. The deleted nucleotide acids in the knock-out zebrafish are indicated with dashes. B, Top, Schematic drawing of the domain structure of Esrp1. Bottom, The amino acids of WT and mutant Esrp1. C, Schematic drawing of the strategy for construction of the esrp2 knock-out zebrafish. Exons are indicated with numbers. The target of sgRNA in the esrp2 gene is indicated with red color, and the PAM sequence is highlighted in yellow. The deleted nucleotide acids in the knock-out zebrafish are indicated with dashes. D, Top, Schematic drawing of the domain structure of Esrp2. Bottom, The amino acids of WT and mutant Esrp2. E, Western blot was performed to examine the expression of WT or mutant Esrp1/2 protein. The coding sequence of esrp1/2 with or without the deletion shown in A and C was inserted into expression vector to express Esrp1/2 protein with a Myc tag at the C terminus. Expression vectors were transfected into HEK293T cells, and western blot analysis was performed using an anti-Myc antibody. F, The survival rate of WT or knock-out zebrafish larvae. The numbers of larvae for each genotypes are indicated in the brackets.
Examination with a light microscope revealed that otolith development was perturbed in esrp1/2 knock-out larvae. In WT larvae at 36 hpf, two otoliths could be readily observed that remained separate up to 72 hpf (Fig. 3A,B). In contrast, a significant proportion of single knock-out or double heterozygous larvae had two otoliths in close proximity at 72 hpf (Fig. 3A,B). Furthermore, esrp1+/−;esrp2−/− or esrp1−/−;esrp2+/− larvae exhibited more severe phenotypes compared with the corresponding single knock-out larvae (Fig. 3A,B). Notably, in ∼10% of esrp1−/−;esrp2+/− larvae, the two otoliths were in contact. Finally, the DKO larvae manifested the most severe phenotype, with all of them having contacting otoliths at 72 hpf (Fig. 3A,B). Besides that, all the DKO larvae displayed delayed semicircular canal fusion (Fig. 3C). Given their consistently severe phenotypes, we focused on DKO zebrafish in the subsequent experiments.
Otolith formation and semicircular canal fusion are affected in the esrp1 and esrp2 knock-out zebrafish. A, Otolith formation in zebrafish larvae was examined using a light microscope at 36, 48, 60, and 72 hpf. Shown are different types of otolith formation observed in zebrafish from various genotypes. B, The percentage of each types of otolith formation from different genotypes at 72 hpf were calculated from results similar to A. C, Semicircular canal fusion in WT or esrp1;espr2 DKO zebrafish larvae was examined using a light microscope at 48 and 72 hpf. The numbers of larvae for each genotypes are indicated in the brackets. Scale bars, 50 μm.
Disruption of esrp1 and esrp2 compromises the MET function of hair cells
To examine whether disruption of esrp1 and esrp2 impairs hair cell development, we crossed DKO zebrafish with brn3c:GFP transgenic zebrafish that express GFP in the hair cells of both the inner ear and lateral line systems (He et al., 2017). Examination with a confocal microscope clearly revealed GFP-positive hair cell patches in the three cristae of WT larvae at 72 hpf (Fig. 4A). Similar numbers of hair cells were observed in the DKO larvae at the same age, suggesting that hair cell development in the cristae is largely unaffected by esrp1/2 disruption (Fig. 4A,B). We also examined the hair cells in the saccule and utricle. Consistent with the contacting otoliths observed with the light microscope, the sensory epithelia of the saccule and utricle were in contact in DKO larvae when examined at 72 hpf (Fig. 4C). Nevertheless, the numbers of GFP-positive saccular and utricular hair cells were comparable between WT and DKO larvae at this age, suggesting that hair cell development in the saccule and utricle is also unaffected by esrp1/2 disruption (Fig. 4C,D). The MET function of hair cells in the cristae was then evaluated by performing FM 4-64 dye uptake experiment. The results demonstrated that FM 4-64 dye uptake was significantly decreased in DKO larvae compared with WT larvae at 5 dpf, suggesting that the MET function of inner ear hair cells is compromised by disruption of esrp1/2 (Fig. 4E,F).
MET function is affected in the inner ear of esrp1/esrp2 DKO zebrafish. Hair cells in WT or esrp1;espr2 DKO larvae in a brn3c:GFP transgenic background were examined by GFP fluorescence at 72 hpf. MET function of hair cells in the cristae of semicircular canals was examined by performing FM 4-64 uptake experiment at 5 dpf. A, GFP-positive hair cells in the cristae of semicircular canals were examined using a confocal microscope. Phalloidin staining was performed to visualize stereociliary F-actin core. B, Numbers of GFP-positive hair cells in the cristae of three semicircular canals were calculated from results similar to A. C, GFP-positive hair cells in the sensory epithelia of the utricle (Utr) and saccule (Sac) were examined using a confocal microscope. Phalloidin staining was performed to visualize stereociliary F-actin core. D, Numbers of GFP-positive hair cells in the sensory epithelia of the utricle and saccule were calculated from results similar to C. E, FM 4-64 uptake of hair cells in the cristae of semicircular canals was examined using a confocal microscope. F, The FM 6-64 uptake per crista was quantified from results similar to E. The numbers of larvae for each genotypes are indicated in the brackets. AC, anterior canal; LC, lateral canal; PC, posterior canal. Scale bars, 20 μm (A, C), 10 μm (E). ns, not significant; **p < 0.01.
We also investigated whether esrp1/2 disruption affects lateral line hair cell development using brn3c:GFP transgenic zebrafish. The numbers of hair cells per neuromast in both ALLs and PLLs were not significantly different between WT and DKO zebrafish larvae when examined at 72 hpf, suggesting that lateral line hair cell development is unaffected by esrp1/2 disruption (Fig. 5A–D). However, FM1-43FX dye uptake was significantly decreased in certain ALL neuromasts of DKO larvae at 4 dpf (Fig. 5E,F). Similar results were obtained when examined at a higher magnification; meanwhile, FM1-43FX dye uptake was unaffected in PLL neuromasts of DKO larvae at the same age (Fig. 5G,H). Taken together, our present data suggest that the MET function is affected by esrp1/2 disruption in the inner ear and certain ALL neuromasts.
MET function is affected in certain ALL but not PLL neuromasts of esrp1/esrp2 DKO zebrafish. Hair cells in WT or esrp1;espr2 DKO larvae in a brn3c:GFP transgenic background were examined by GFP fluorescence as well as immunostaining using an anti-MYO7A antibody at 72 hpf. MET function of ALL or PLL neuromast hair cells in WT or esrp1;espr2 DKO larvae without brn3c:GFP was examined by performing FM 1-43FX uptake experiment at 4 dpf. A, GFP-positive hair cells in the ALL neuromast (M1) were examined using a confocal microscope. Hair cells were also labeled by immunostaining using an anti-MYO7A antibody. B, Numbers of GFP-positive hair cells in the ALL neuromast (M1) were calculated from results similar to A. C, GFP-positive hair cells in the PLL neuromast (L1) were examined using a confocal microscope. Hair cells were also labeled by immunostaining using an anti-MYO7A antibody. D, Numbers of GFP-positive hair cells in the PLL neuromast (L1) were calculated from results similar to C. E, FM 1-43FX uptake of ALL neuromast hair cells in WT or esrp1;espr2 DKO larvae without brn3c:GFP was examined using a confocal microscope. F, The FM 1-43FX uptake per neuromast was quantified from results similar to E. G, FM 1-43FX uptake of ALL (M1) and PLL (L1) neuromast hair cells in WT or esrp1;espr2 DKO larvae without brn3c:GFP was examined using a confocal microscope at higher magnification. H, The FM 1-43FX uptake per neuromast was quantified from results similar to G. The numbers of larvae for each genotype are indicated in the brackets. Scale bars, 10 μm (A, C, G), 200 μm (E). ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001.
Disruption of esrp1 and esrp2 leads to dysregulated gene expression
ESRP1 has been shown to regulate the alternative splicing of target genes such as Fgfr2, which plays crucial roles in mouse cochlear development (Rohacek et al., 2017). Consistently, our qPCR results confirmed that fgfr2b/c splicing was perturbed in esrp1/2 DKO larvae (Fig. 6A,B). To further identify Esrp1/2-downstream genes that might regulate hair cell function, we analyzed the transcriptome of otic vesicles from WT and DKO larvae at 34 hpf by performing RNA-seq. We identified 5,238 genes whose mRNA levels changed by >1.2-fold in DKO larvae compared with WT larvae, among which 2,453 genes were upregulated and 2,785 genes were downregulated (Fig. 6C,D). We selected several dysregulated candidates and performed RT-PCR and qPCR to examine their levels in the otic vesicles. The results indicated that the mRNA levels of gas2b, ndrg1b, otop2, and zpld1a were significantly decreased, while the mRNA levels of si:ch73-59f11.3 and gucy2d were significantly elevated in the otic vesicles of esrp1/2 DKO larvae at 34 hpf, confirming the RNA-seq results (Fig. 6E,F).
Identification of Esrp1/2 target mRNAs. A, Schematic drawing of the alternative splicing of fgfr2 mRNA. The positions of qPCR primers are indicated by arrows. B, qPCR results show the level of fgfr2b and fgfr2c in WT or DKO larvae at 34 hpf. C, Volcano plot of the dysregulated mRNAs in the inner ear of DKO larvae at 34 hpf revealed by RNA-seq. D, The heatmap of the top dysregulated mRNAs in the inner ear of DKO larvae at 34 hpf revealed by RNA-seq. E, Expression of potential Esrp1/2 target mRNAs in the inner ear of WT and DKO larvae at 34 hpf was examined by performing RT-PCR. F, Expression of potential Esrp1/2 target mRNAs in the inner ear of WT and DKO larvae at 34 hpf was examined by performing qPCR. G, Expression of tmc1, tmc2a, and tmc2b in WT and DKO larvae at 48 hpf (for tmc1) or 96 hpf (for tmc2a/2b) was examined by performing in situ hybridization. The otic vesicles are indicated by arrows. Scale bars, 0.2 mm. ns, not significant; **p < 0.01; ***p < 0.001; ****p < 0.0001.
Notably, RT-PCR and qPCR results corroborated that the mRNA levels of tmc1 and tmc2a, but not tmc2b, were significantly decreased in the otic vesicles of esrp1/2 DKO larvae (Fig. 6E,F). To further examine tmc1/2a/2b mRNA levels in the otic vesicles, in situ hybridization experiments were performed using probes specific for tmc1/2a/2b. The results showed that tmc1/2a/2b transcripts were readily detectable in the otic vesicles at 48 or 96 hpf (Fig. 6G). However, in the esrp1/2 DKO larvae, tmc1 and tmc2a were significantly decreased, while tmc2b remained unaffected (Fig. 6G), which was consistent with the RT-PCR and qPCR results. The tmc1/2a/2b transcripts have also been detected in ALL or PLL neuromasts by in situ hybridization or RNA-FISH (Maeda et al., 2014; Smith et al., 2023). However, we failed to observe their expression in the neuromasts, possibly due to the low affinity of the probes used in the current experiment. Taken together, our data suggest that tmc1 and tmc2a transcripts are downregulated in the otic vesicles of esrp1/2 DKO larvae, which might contribute to the impaired MET function of hair cells.
Esrp1/2 regulates the stability of tmc1/2a transcripts
The RNA-seq, RT-PCR, and in situ hybridization results suggest that Esrp1/2 might regulate the stability of tmc1 and tmc2a mRNAs. To test this hypothesis, we transfected HEK293T cells with Esrp1/2-expressing vectors along with tmc-expressing vectors. The transfected cells were treated with the transcription inhibitor ActD and the levels of tmc mRNAs at different time points were examined by performing RT-qPCR. The results revealed that in the presence of Esrp1 or Esrp2, the level of tmc1/2a was significantly elevated, indicating that Esrp1/2 enhances the stability of tmc1/2a mRNAs (Fig. 7A,B). In contrast, the stability of tmc2b mRNA was unaffected by the presence of Esrp1/2 (Fig. 7C).
Esrp1/2 regulates the stability of tmc1 and tmc2a mRNAs. A–C, The relative levels of tmc1 (A), tmc2a (B), and tmc2b (C) mRNAs in transfected HEK293T cells were normalized with the level of β-actin transcript and plotted along with time to calculate their relative half-life. HEK293T cells were transfected with tmc1/2a/2b-expressing vectors along with Esrp1/2 expression vectors. The transfected cells were treated with ActD and the level of tmc1/2a/2b mRNAs at different time points was examined by performing RT-qPCR. D, RIP results show that Myc-Esrp1/2 binds to tmc1/tmc2a/tmc2b mRNAs in transfected HEK293T cells. E, Schematic drawing shows the interaction between Esrp1 and tmc1 mRNA with various deletions revealed by RIP results. F, RIP results show that the interaction of Esrp1 with WT and mutant tmc1 mRNA. G, Quantification of results similar to F. ***p < 0.001.
We then performed RIP experiments to investigate whether Esrp1/2 could directly bind to tmc1/2a/2b mRNAs. Myc-tagged Esrp1/2 together with full-length tmc1/2a/2b cDNA was overexpressed in cultured HEK293T cells. After immunoprecipitation using immobilized anti-Myc agarose, the RNAs bound to Myc-Esrp1/2 were subjected to RT-PCR. The results revealed that tmc1 and tmc2a mRNAs were indeed associated with Esrp1/2, suggesting that they are Esrp1/2-binding mRNAs (Fig. 7D). Interestingly, the RIP results also indicated that tmc2b mRNA was associated with Esrp1/2 (Fig. 7D), although the stability of tmc2b mRNA was not regulated by Esrp1/2.
In order to identify the region in tmc mRNAs required for binding Esrp proteins, we expressed different fragments of tmc1 cDNAs together with Myc-Esrp1 in HEK293T cells and performed RIP experiments (Fig. 7E). The results showed that nucleotides 2,520–2,973 in the coding sequence (CDS) of tmc1 were associated with Esrp1 (Fig. 7E). Next, we expressed full-length tmc1 cDNAs with various deletions in this region and repeated RIP experiments. The results demonstrated that when nucleotides 2,540–2,645 were deleted, tmc1 mRNA was no longer associated with Myc-Esrp1 (Fig. 7E). Sequence analysis revealed that there are two potential ESRP-binding “UGG” motifs in this region (Fig. 7E). We then mutated “GG” into “AA” in these two motifs, and RIP experiments showed that this mutation significantly weakened the association of tmc1 mRNA with Myc-Esrp1 (Fig. 7E–G).
Discussion
ESRP1 and ESRP2 are RBPs that play pivotal roles in many cellular processes, ranging from regeneration and epithelial-mesenchymal transition to cancer progression (Derham and Kalsotra, 2023). It was reported recently that Esrp1 knock-out in mice led to cochlear development deficits at the early embryonic stage (Rohacek et al., 2017). However, the physiological roles of ESRP1 in the inner ear at later stages and/or hair cells remain unknown. In the present work, we employed zebrafish as a model and demonstrated that Esrp1 and its paralog Esrp2 play essential roles in the hair cells by regulating mRNA stability.
It is well known that ESRP1 and ESRP2 execute their functions by regulating alternative splicing (Derham and Kalsotra, 2023). However, reports also suggested that ESRP1 might regulate the stability of mRNAs such as Cyclin A2 and PHGDH (Chen et al., 2019; Gökmen-Polar et al., 2023). Herein, we present evidence that Esrp1/2 is important for regulating the stability of tmc1 and tmc2a mRNAs. Several lines of evidence support this conclusion. Firstly, RIP results revealed that Esrp1/2 directly binds to tmc1 and tmc2a mRNAs and that mutation within the potential binding site of tmc1 substantially attenuates this interaction. Secondly, Esrp1/2 overexpression augments the stability of tmc1 and/or tmc2a mRNAs in a heterologous system. Thirdly, RNA-seq, RT-PCR, and in situ hybridization results confirmed that the levels of tmc1 and tmc2a mRNAs are decreased in esrp1/esrp2 DKO zebrafish. Therefore, our data confirm that Esrp1 and Esrp2 are important RBPs that regulate both alternative splicing and mRNA stability.
As integral components of the MET complex, TMC1 and TMC2 have attracted considerable attention. Molecular, electrophysiological, and structural investigations have shed light on the mechanisms by which TMC1/2 regulates MET (Kawashima et al., 2011; Pan et al., 2013; Beurg et al., 2015, 2018; Pan et al., 2018; Jia et al., 2020; Jeong et al., 2022; Clark et al., 2024; Fu et al., 2025). Meanwhile, evidence has emerged elucidating how TMC1/2 is regulated at the posttranscriptional level. For instance, the mouse Tmc1 gene is subjected to alternative splicing, giving rise to different splice variants (Kawashima et al., 2011; Yamaguchi et al., 2020; Zhou et al., 2021). Alternative splicing of exon 2 results in two TMC1 isoforms that only differ within the first 5 amino acid (aa) residues at the N terminus, and both isoforms are capable of rescuing the MET deficits in Tmc1;Tmc2 knock-out mice (Kawashima et al., 2011). Alternative splicing of exon 9 or 14 impacts the inclusion of 33 aa in the first extracellular loop or 3 aa in the sixth transmembrane domain, respectively (Zhou et al., 2021). The biological significance of exon 9 and 14 splicing remains elusive. In the present work, we demonstrate that the stability of tmc1 and tmc2a mRNAs is regulated by Esrp proteins in zebrafish.
Interestingly, several studies have reported that TMC1/2 mRNAs or proteins exhibit varying levels across different hair cell types (Maeda et al., 2014; Beurg et al., 2018; Smith et al., 2023). It was reported that there is a tonotopic gradient of TMC1 proteins at the stereociliary tips in the mouse cochlea (Beurg et al., 2018). In the zebrafish inner ear, tmc1 and tmc2b transcripts are mainly expressed in extrastriolar hair cells, while tmc2a transcripts are predominantly expressed in striolar hair cells (Smith et al., 2023; Sun et al., 2024). Consistently, functional data suggest that there is differential reliance on tmc1/2a/2b genes in different hair cell types in zebrafish (Chou et al., 2017; Chen et al., 2020; Smith et al., 2020; Zhu et al., 2021; Kindig et al., 2023; Sun et al., 2024). However, it remains elusive how these differential expression patterns are established. Our in situ hybridization results revealed that esrp1 transcripts are mainly expressed in the inner ear and certain ALL neuromasts, and esrp2 transcripts are highly expressed in the inner ear. Neither esrp1 nor esrp2 transcripts were detected in PLL neuromasts. Meanwhile, our data demonstrated that Esrp1/2 regulates the mRNA stability of tmc1 and tmc2a but not tmc2b. Therefore, it seems plausible that Esrp1/2-mediated mRNA stability might contribute to the differential expression patterns of tmc1/2a/2b. The absence of esrp1/2 expression in PLL neuromasts suggests that tmc mRNA stability might be regulated by other RBPs in PLL neuromasts, which awaits further investigations.
Our RIP results indicated that tmc2b mRNA is also associated with Esrp1/2, albeit the stability of tmc2b mRNA is unaffected by Esrp1/2. It has been reported that ESRP1 regulates protein translation when bound to the 5′- or 3′-UTR of its target mRNAs (Leontieva and Ionov, 2009; Fagoonee et al., 2013; Peart et al., 2022). Currently, the binding site of Esrp1/2 in tmc2b mRNA remains elusive, and we are uncertain whether Esrp1/2 regulates tmc2b translation. Nevertheless, the hypothesis that Esrp1/2 regulates mRNA stability and/or translation of tmc1/tmc2a/tmc2b is intriguing and warrants further investigations.
Our current work suggests that esrp1/esrp2 knock-out also impacts otolith and semicircular canal development. Dysregulated otolith formation and/or semicircular canals have been associated with hair cell development deficits in several zebrafish mutants (Stooke-Vaughan et al., 2012; Whitfield, 2020). However, hair cell development appears to be unaffected in esrp1/esrp2 knock-out larvae. At present, we do not know the underlying mechanism by which esrp1/esrp2 knock-out affects otolith and semicircular canal development. RNA-seq and RT-PCR analyses revealed that otop2 mRNA is downregulated in the inner ear of esrp1/esrp2 DKO larvae. Otop2 belongs to an evolutionarily conserved gene family encoding proton-selective ion channels Otopetrin 1–3 (OTOP1–3; Tu et al., 2018). Among the three orthologous genes, otop1 has been shown to play essential roles in otolith formation in zebrafish (Hughes et al., 2004; Söllner et al., 2004). Further investigations are warranted to elucidate whether esrp1/esrp2 affects otolith and semicircular canal development by regulating the expression of otop2 and/or other target genes.
Footnotes
This work was supported by grants from the National Key R&D Program of China (2022YFE0131900 to Z.X.) and the National Natural Science Foundation of China (82192861 to Z.X.). We thank Sen Wang, Yuyu Guo, Xiaomin Zhao, and Haiyan Yu from the core facilities for life and environmental sciences, Shandong University for technical support in confocal microscopy.
The authors declare no competing financial interests.
- Correspondence should be addressed to Zhigang Xu at xuzg{at}sdu.edu.cn.
