Introduction

Deafness is a common and highly disabling disease in humans that has not been effectively prevented or controlled. According to a survey report by the World Health Organization, the number of people suffering from deafness worldwide reached 466 million, accounting for more than 5% of the global population. By 2050, nearly 2.5 billion people are projected to have some degree of hearing loss. Deafness has become a major concern for global health. According to the 2016 population survey in China, approximately 70 million people experienced hearing loss. Among children in Jilin, Gansu, Guangdong, and Shanxi provinces of China, the prevalence rate of disabling hearing loss is 0.85%, and it is expected that this prevalence rate will increase year by year [1]. According to pathogenic factors, about 60% of deafness is caused by a genetic defect known as hereditary deafness. Investigating the genetic identification and molecular mechanisms of hereditary deafness would reveal the occurrence and development of the disease, which is the cornerstone of accurate diagnosis and prevention of deafness. Geneticists estimate that the total number of genes responsible for deafness is more than 1000. So far, around 156 non-syndromic deafness genes have been identified [2], and there are still a large number of new deafness genes that need to be identified.

Cholesterol is the most abundant sterol molecule in mammalian cells. It constitutes the cell membrane and is crucial in synthesizing important hormones, facilitating synapse formation, and mediating cell signaling transduction [3–6]. Previous studies have shown that dysregulated intracellular cholesterol homeostasis in auditory cells is involved in auditory development defects [6–8], hereditary hearing loss [9–13], noise-induced hearing loss [14–16], ototoxic hearing loss [17, 18], and age-related hearing loss [19–21]. It has been demonstrated that HSD17B7 converts zymosterone to zymosterol during the cholesterol synthesis process, and is expressed in the brain, heart, eye, and ear [22, 23]. However, the function of HSD17B7 in the auditory system remains unexplored. No pathogenic HSD17B7 mutations have been reported to be associated with hearing loss. Here, we show that HSD17B7 is highly expressed in sensory hair cells in mice and zebrafish. Our findings highlight the critical role of HSD17B7 in the auditory system, positioning it as a potential therapeutic target for hearing loss.

Results

Hsd17b7 expressed in sensory hair cells of zebrafish and mice

To assess the role of HSD17B7 in the auditory system, we first investigated the expression of the hsd17b7 in the developing zebrafish. Single-cell data analysis (accession no. GSE221471) [24] categorized neuromast hair cell (cluster 0), supporting cell (cluster 1), macula hair cell (cluster 2), crista hair cell 1 (cluster 3), mantle cell (cluster 4) and crista hair cell 2 (cluster 5) using the Seurat 4.0.1 platform (Figure 1A). The feature plot and violin plot showed that hsd17b7 was expressed in sensory hair cells, especially in neuromast hair cells (NM HCs) and crista hair cells (Figure 1B, 1C). Then, NM HCs, supporting cells, and mantle cells were ordered along pseudotime using Monocle 3 (Figure 1D, 1E), and a heatmap was generated to visualize the expression dynamics of marker genes in NM HC (Figure 1F). These data indicate that the expression of hsd17b7 gene was gradually increased along the NM HC development trajectory (Figure 1D, 1E). Consistently, whole-mount in situ hybridization (WISH) in zebrafish showed that hsd17b7 was specifically enriched in the NM HCs and crista HCs at 72 hours post-fertilization (hpf) and 96 hpf (Figure 1G). To further explore the localization of hsd17b7 in zebrafish HCs, we generated a hsd17b7-egfp fusion construct driven by a hair cell-specific promoter, myosin 6b [25, 26]. We injected this construct into a fertilized egg and expressed the hsd17b7-egfp fusion in the HCs. Confocal imaging analysis revealed that Hsd17b-EGFP was localized to the cytoplasm in both crista HCs and NM HCs, exhibiting a punctate distribution (Figure 1H, 1I).

Figure 1.

Meanwhile, to further elucidate the expression of HSD17B7 in the mammal auditory system, we analyzed the expression levels of Hsd17b7 in three single-cell data sets of the mouse [27–29], we found that Hsd17b7 were highly expressed in outer hair cells (OHCs), utricles hair cells, and inner hair cells (IHCs), but less in crista hair cells (Figure 1J). Next, we assessed the expression of HSD17B7 in C57BL/6 wild-type mice. Immunostaining analysis revealed that HSD17B7 was expressed in both outer hair cells (OHCs) and inner hair cells (IHCs) (Figure 1K), confirming the scRNA-seq data. These results demonstrate that HSD17B7 is highly expressed in the sensory hair cells of zebrafish and mice.

In addition, phylogenetic and sequence alignment analysis revealed that vertebrate HSD17B7s share a significant similarity in amino acid sequences (Figure S1A). UniProt sequence alignment analysis revealed that Danio rerio Hsd17b7 (NP_001070796.1) shares 58.2% identity with Homo sapiens HSD17B7 (NP_057455.1) and 67.1% identity with Mus musculus HSD17B7 (NP_034606.3), respectively (Figure S1B). These data suggest that HSD17B7 is conserved and expressed in sensory hair cells across vertebrate species.

Hsd17b7 deficiency impaired auditory and MET function in zebrafish

To investigate the Hsd17b7 function in the auditory system, we generated hsd17b7 mutant alleles using CRISPR/Cas9-mediated mutagenesis (Figure S2A, S2B). The target site fragment was PCR-amplified and sequenced (Figure S2C, S2D). Sequence analysis revealed that the mutant alleles included a 4-bp deletion and two 6-bp deletions in exon 3 (Figure S2E). Subsequently, the 4-bp deletion mutant line was chosen for further experiments, which led to a reading frame shift and premature termination of protein translation. The mutant protein is predicted to translate to 96 amino acids, then adds 4 code-shifted residues (Figure S2F).

We first examined the startle response evoked by auditory stimuli to assess the hearing abilities of zebrafish larvae [30]. We found that the movement trajectory, swimming velocity, and distance of the hsd17b7 mutants were significantly decreased compared to those of the controls at 5 dpf (Figure 2A-2C), indicating that depletion of hsd17b7 leads to significant hearing loss. To verify that the hearing loss was specifically caused by hsd17b7 deficiency, we attempted to rescue the impaired auditory function by injecting the hsd17b7 mRNA into hsd17b7 mutants. It was shown that the microinjection of hsd17b7 mRNA successfully restored the compromised movement trajectory, swimming velocity, and distance of the hsd17b7 mutants (Figure 2A-2C), indicating that Hsd17b7 is crucial for the auditory function in zebrafish.

Figure 2.

Furthermore, we designed antisense morpholino oligonucleotides (hsd17b7 Mo) to block the splicing of hsd17b7 RNA at the junction of exon 1/intron 1 to knock down Hsd17b7 to verify the phenotype (Figure S3A, S3B). RT-PCR revealed that natural hsd17b7 mRNA expression was significantly reduced, and mis-spliced mRNA was observed after injection of hsd17b7 Mo (Figure S3C). Additionally, we evaluated the knockdown efficiency and rescue efficiency of hsd17b7 by immunoblotting. As expected, the protein levels of Hsd17b7 decreased significantly after hsd17b7 Mo injection, and mRNA injection effectively rescued the Hsd17b7 protein level (Figure S3D). We subsequently found that the movement trajectory, distance, and swimming velocity of the hsd17b7 morphants were significantly decreased compared to those of controls at 5 dpf (Figure S3E-S3G). Moreover, hsd17b7 mRNA effectively rescued the defective startle response of hsd17b7 morphants (Figure S3E-S3G). These results demonstrate that Hsd17b7 is required for auditory function.

Since Hsd17b7 deficiency impaired auditory function, we investigated whether the mechanotransduction (MET) function of Hsd17b7 mutants is compromised. We generated hsd17b7 mutant alleles in the Tg(Brn3c: mGFP) transgenic line [24, 31], in which the membrane of sensory HCs are labeled with GFP. Confocal microscopy imaging analysis revealed that the average number of HCs in hsd17b7 mutants was decreased compared to that of controls (∼16% reduction) (Figure 2D, 2E). FM4-64 (N-(3-Triethylammoniumpropyl)-4-(6-(4-(Diethylamino) Phenyl) Hexatrienyl) Pyridinium Dibromide) is a fluorescence dye that labels HCs and can rapidly permeate HCs through traversing open mechanosensitive channels, which were commonly used as a proxy for MET function in mature HCs [26, 32, 33]. We found that at 5 dpf, the labeling intensity of FM4-64 in the HCs of the lateral line neuromasts in hsd17b7 mutants was markedly reduced compared with that of controls (∼40% reduction) (Figure 2F, 2G). Meanwhile, we also used the FM4-64 staining assay to examine the MET function of hsd17b7 morphants. The results showed that the FM4-64 intensity in hsd17b7 morphants was also significantly decreased compared to controls at 5 dpf (∼25% reduction) (Figure S4A, S4B). Given that the reduction in HC numbers (∼16%) is much smaller than the decrease in FM4-64 labeling intensity (∼40%), these results suggest that the loss of hsd17b7 function primarily affects the MET function of HCs in zebrafish rather than the number of HCs.

To verify that hsd17b7 is essential for the MET function, we performed a rescue experiment by injecting hsd17b7 mRNA. It was demonstrated that microinjection of the hsd17b7 mRNA successfully restored the FM4-64 intensity and the HC numbers (Figure 2D-2G, S4A, S4B). Collectively, these data demonstrate that hsd17b7 is critical for MET function in zebrafish.

HSD17B7 regulated cholesterol levels in vitro and in vivo

HSD17B7 converts zymosterone to zymosterol, participating in cholesterol biosynthesis (Figure 3A) [23]. Previous research has indicated that abnormally high and low cholesterol levels are detrimental to hearing [7, 34, 35]. To further explore the physiological mechanism underlying how HSD17B7 affects auditory function, we first verified whether HSD17B7 regulates cholesterol. It was shown that HSD17B7 knockdown in HEI-OC1 cells led to a decrease in total cellular cholesterol (Figure 3B, 3C). The minimal cholesterol-binding domain 4 (D4H) cholesterol binding probe derived from the toxin Perfringolysin O (PFO) was widely used to monitor cholesterol at the cytoplasmic leaflet of the plasma membrane (PM) [36, 37]. D4H-mCherry was used to monitor the cholesterol in HEI-OC1 cells. Immunostaining results showed that D4H-mCherry intensity was significantly increased in HEI-OC1 cells overexpressing HSD17B7 compared with control cells (Figure 3D, 3E), suggesting HSD17B7 regulates cholesterol.

Figure 3.

To further assess the functional role of Hsd17b7 in regulating cholesterol in vivo, we injected the construct expressing D4H-mCherry into one-cell-stage embryos of control and hsd17b7 mutant Tg(Brn3c: mGFP) transgenic line to evaluate the cholesterol in zebrafish (Figure 3F). We found that D4H-mCherry was enriched in the hair bundle (Figure 3G), which is consistent with previous reports [36]. Compared with controls, the intensity of D4H-mCherry in crista HCs and NM HCs of hsd17b7 mutants at 4 dpf was significantly reduced (Figure 3G, 3H). These results demonstrate that HSD17B7 regulated cholesterol synthesis in virto and in vivo. Given that abnormal cholesterol homeostasis is detrimental to hearing [7, 34, 35], the deficiency of Hsd17b7 led to compromised MET and auditory function, suggesting that Hsd17b7 regulates auditory function by influencing cholesterol levels in HCs.

A nonsense Variant in HSD17B7 identified in human deafness

Given the established role of Hsd17b7 in auditory function in zebrafish, the human HSD17B7 gene was screened in individuals with hearing loss (HL) who remained undiagnosed after analysis of 201 known HL genes using whole-genome sequencing (WGS)[38]. The proband failed the newborn hearing screening, was enrolled in a special education school during early childhood, and presented at 8 years of age with bilateral profound hearing loss. Physical examination revealed two preauricular appendages anterior to the right ear but no other systemic abnormalities. The proband Ⅱ 1’s parents had normal hearing and no family history of HL but declined genetic testing (Figure 4A, 4B). A heterozygous nonsense variant, c.544G>T (p.E182*) in HSD17B7 (NM_016371.4) was identified and confirmed by Sanger sequencing (Figure 4C). This mutation introduces a premature stop codon at position Glu182 (Figure 4C).

Figure 4.

The multiple sequence alignment showed that Glu182 is evolutionarily conserved in eight vertebrates, from zebrafish to humans (Figure 4D), implying that the nonsense mutation may have a potential pathogenetic effect on HI. According to UniProt, the encoded HSD17B7 protein contains two topological domains (extracellular domain: 1-229 aa, cytoplasmic domain: 251-341 aa) and a transmembrane region (230-250 aa). The novel nonsense mutations in the extracellular domain of HSD17B7 (c.544G>T) changed residue 182 from Glu to Ter (p.E182*) and resulted in the production of truncated proteins (Figure 4E). In addition, the three-dimensional protein structure of HSD17B7 is first modeled to predict the effect of the nonsense mutation on the protein using the SWISS-MODEL protein structure database (https://swissmodel.expasy.org) (Figure 4F). This data analysis suggests that this nonsense mutation may have a potential pathogenic effect on hearing loss.

E182* mutation mRNA of HSD17B7 failed to rescue the impaired MET function and auditory behavior of hsd17b7 mutants

To assess the impact of E182* nonsense mutation of HSD17B7 on MET function, we microinjected the human HSD17B7 wild-type or E182* mutant mRNAs separately into the fertilized one-cell stage embryos of hsd17b7 KO mutants in Tg(Brn3c: mGFP) transgenic line. It was shown that wild-type HSD17B7 significantly increased the FM4-64 intensity in NM HCs of hsd17b7 mutants, whereas the HSD17B7 E182* mutant mRNA failed to do so (Figure 5A, 5B), indicating that the E182* nonsense mutation of HSD17B7 has no effect in restoring normal MET function in zebrafish. Furthermore, the startle response assay demonstrated that wild-type HSD17B7 successfully rescued the abnormal auditory behaviors (movement trajectory, distance, and swimming velocity) of hsd17b7 KO mutants, whereas the E182* mutant did not (Figure 5C-5E).

Figure 5.

Heterozygous E182* mutation of HSD17B7 has a negative effect on the auditory function

To explore the pathogenetic role of HSD17B7 E182* heterozygous mutation in the patient’s hearing loss, we mimicked the proband Ⅱ heterozygote in vitro and vivo. We transiently expressed Flag-tagged HSD17B7 wild-type (HSD17B7^WT) or E182* nonsense mutation (HSD17B7^E182*) in HEI-OC1 cells to examine their subcellular localization. Since cholesterol synthesis mainly occurs in the endoplasmic reticulum (ER) [39, 40], we supposed that HSD17B7, an enzyme catalyzing cholesterol synthesis, might localize in the ER. Immunostaining revealed that HSD17B7^WT co-localized with Calnexin (an ER marker), whereas HSD17B7^E182* did not and instead exhibited a spot-like aggregate distribution (Figure 6A). These results were confirmed by matching fluorescence signal intensity profiles (Figure 6B, 6C). Given that the patient was a heterozygote mutant, we suspected that the HSD17B7^E182* mutant might impair the subcellular localization of HSD17B7^WT. Therefore, we performed immunostaining by co-transfecting HEI-OC1 cells with Flag-HSD17B7^WT and HSD17B7^E182*-Myc expression constructs. The results showed that the presence of the HSD17B7^E182* protein did not alter the subcellular localization of HSD17B7^WT (Figure 6D), as confirmed by matching two fluorescence signal intensity profiles (Figure 6E).

Figure 6.

To further investigate the underlying cause of hearing loss in the patient carrying the heterozygous HSD17B7^E182* mutation, we injected human wild-type (HSD17B7^WT) or mutant (HSD17B7^E182*) HSD17B7 mRNAs into one-cell stage embryos of the wild-type in Tg(Brn3c: mGFP) transgenic line to assess whether the overexpression of HSD17B7^E182* impairs MET function and auditory behaviors. The FM4-64 assay revealed that the overexpression of HSD17B7^WT did not result in a significant difference in fluorescence intensity compared to the control. In contrast, the overexpression of HSD17B7^E182* led to a significant reduction of fluorescence intensity (Figure 7A, 7B). In addition, HSD17B7^WT overexpression did not alter the startle response compared to the control. In contrast, HSD17B7^E182* overexpression resulted in abnormal movement trajectory, along with reduced mean distance and peak velocity in response to auditory stimulation (Figure 7C-7E). These data suggest that overexpression of HSD17B7^E182* has a negative effect on normal auditory function.

Figure 7.

Our studies have shown that HSD17B7 regulates cholesterol in vivo and in vitro (Figure 3), and previous research has demonstrated that proper cholesterol distribution in hair cells is essential for maintaining auditory function [36]. Therefore, we hypothesize that the heterozygous HSD17B7^E182*mutation may lead to hearing loss by altering cholesterol distribution. Next, we co-transfected HEI-OC1 cells with Flag, Flag-HSD17B7^WTor Flag-HSD17B7^E182* constructs along with D4H-mCherry (cholesterol probe). Immunostaining results revealed that in cells expressing Flag or Flag-HSD17B7^WT, the cholesterol was distributed in the same way around nuclei. In contrast, the nonsense mutation HSD17B7^E182* significantly altered the distribution of D4H-mCherry, forming spot-like aggregation (Figure 7F). Furthermore, HSD17B7^E182* colocalized with D4H-mCherry, verified by the fluorescence intensity profile (Figure 7G-7I), suggesting that HSD17B7^E182* may bind cholesterol and alter its intracellular distribution. Altogether, these results demonstrate that HSD17B7^E182* has a negative effect via altering cholesterol distribution in HCs, leading to compromised MET function and auditory behaviors.

Nonsense mutation (p.E182*) of HSD17B7 disrupted the interaction with RER1 and caused the altered localization and cholesterol distribution

To further explore the pathological mechanism of HSD17B7^E182*, we transfected HEI-OC1 cells with Flag, Flag-HSD17B7^WT, or Flag-HSD17B7^E182* constructs and captured the target proteins by co-immunoprecipitation (Figure 8A). We subsequently performed liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of the captured proteins (Figure S5). The proteins that interact with HSD17B7^WT and HSD17B7^E182*, respectively, are shown in the Venn diagram (Figure 8B). We systematically identified the proteins specifically associated with HSD17B7^WT and HSD17B7^E182*. Consistent with the role of HSD17B7, proteins specifically associated with HSD17B7^WT were strongly enriched for cholesterol metabolic process, while proteins specifically associated with HSD17B7 ^E182* were strongly enriched for chromatin remodeling (Figure 8C). Considering that HSD17B7^WT and HSD17B7^E182* have different subcellular localizations, we examined the cellular component of HSD17B7^WT-specific and HSD17B7^E182*-specific interacting proteins. It was found that the HSD17B7^WT-specific interacting proteins were enriched in the ER, while the HSD17B7^E182*-specific interacting proteins were enriched in the nucleus and cytoplasm (Figure 8D), consistent with the immunostaining results (Figure 6A-6C). Among these ER-associated proteins, RER1 has the top hit (Figure 8E). RER1 is involved in protein retention in ER [41, 42]. We verified the direct interaction between HSD17B7^WT and RER1 by an in vitro binding assay. Glutathione S-transferase (GST)-HSD17B7, but not GST, is bound directly to RER1 fused His (RER1-His) (Figure 8F). Next, we co-transfected HEI-OC1 cells with RER1-Myc along with Flag-HSD17B7^WT or Flag-HSD17B7^E182* constructs. The immunostaining results indicated that RER1 co-localized with HSD17B7^WT, whereas it did not co-localize with HSD17B7^E182* (Figure 8G), as confirmed by the intensity profiles (Figure 8H, 8I). These data demonstrate that RER1 interacts with HSD17B7^WT but not with HSD17B7^E182*, suggesting that the localization of HSD17B7^WT in the ER is due to its interaction with RER1. In contrast, HSD17B7^E182* cannot interact with RER1, resulting in abnormal subcellular localization.

Figure 8.

During transfection of Flag-HSD17B7^WT and Flag-HSD17B7^E182* constructs in HEI-OC1 cells, we noticed that the number of positive cells expressing HSD17B7^E182* was very low (Figure S6A). Additionally, the IP results also suggested that the nonsense mutation E182* in HSD17B7 led to a significant decrease in protein levels compared to HSD17B7^WT (Figure 8A).

To explain the phenomenon, we transfected equal amounts of Flag-HSD17B7^WT and Flag-HSD17B7^E182* constructs to confirm their protein levels. Immunoblotting revealed that the levels of E182* mutation protein were significantly lower compared to those of the wild-type (Figure S6B), consistent with our aforementioned findings. Next, we measured the mRNA level of HSD17B7^WT and HSD17B7^E182*. RT-qPCR results showed that the mRNA levels of the E182* mutation were significantly lower compared to those of the wild-type (Figure S6C). Considering that the E182* mutation decreased not only the protein levels but also the mRNA levels, we subsequently examined the effect of the mutation on the half-life periods of mRNA (Figure S6E). Interestingly, the E182* mutation shortened the mRNA half-life periods of HSD17B7 (Figure S6D). Overall, the E182* mutation reduced the stability of HSD17B7 mRNA, resulting in significantly decreased mRNA and protein levels. The abnormally low levels of HSD17B7^E182* cannot bind with RER1, preventing proper localization of HSD17B7 to the ER. This, in return, disrupts cholesterol synthesis and distribution, resulting in abnormalities in MET function and auditory behaviors.

Discussion

This study reveals that HSD17B7 is enriched in sensory hair cells in zebrafish and mice, regulating cholesterol levels in hair cells, and is essential for normal mechanotransduction and auditory function. Loss of hsd17b7 markedly reduces cholesterol in hair cells, causing abnormal MET function and hearing behaviors, suggesting that HSD17B7 is a novel candidate gene for sensory hearing loss. This was further supported by a case of sporadic deafness, in which we identified a previously undescribed variant (c.544G>T, p.E182*) in HSD17B7 through whole-exome sequencing and Sanger sequencing analyses. The residual HSD17B7^E182* protein alters subcellular localization and disrupts normal cholesterol distribution. The E182 mutation in HSD17B7 reduces the mRNA half-life, decreases mRNA abundance, and significantly lowers protein expression levels. Collectively, our findings suggest that the heterozygous c.544G>T (p.E182*) in the HSD17B7 (NM_016371.4) gene disrupts auditory function through potential pathogenic mechanisms. First, the heterozygous E182* mutation altered the intracellular distribution of cholesterol. Second, the haploinsufficiency resulting from reduced HSD17B7 protein levels disrupts cholesterol biosynthesis. These mechanistic insights highlight the critical role of cholesterol homeostasis in auditory function, providing a molecular basis for the development of potential targeted therapeutic strategies.

HSD17B7 was first identified in the rat prolactin receptor-associated protein [43], which is thought to play a key role in the production of estradiol E2 [44, 45]. Subsequent in vitro experiments demonstrated that HSD17B7 participated in cholesterol synthesis by catalyzing the reduction of zymosterone to zymosterol [46–48]. Both in vivo and in vitro, a deficiency of HSD17B7 leads to decreased cholesterol levels, while overexpression of HSD17B7 results in increased cholesterol levels, consistent with the pathway of cholesterol synthesis. Previous studies have shown that HSD17B7 is expressed in the liver, heart, brain, eye, and ear [22, 23, 49]. Although the role of the HSD17B7 gene in ovarian and breast cancer has been extensively explored in the literature [50–54], the role of the HSD17B7 gene in auditory function remains unclear. Our study highlights that HSD17B7 is essential for normal hearing in vertebrates. This conclusion is supported by the high expression of HSD17B7 in sensory hair cells across vertebrate inner ears, as revealed by the results of whole-mount in situ hybridization and immunostaining in our current study, as well as single-cell RNA sequencing data from our previous publication and those of others [24, 27–29]. Given that loss of hsd17b7 causes abnormal hearing behaviors, our research identifies HSD17B7 as a novel candidate gene for sensory hearing loss.

Cholesterol is a component of biological membranes, and it plays both structural and functional roles in the bio-membranes [55, 56]. Previous studies have shown that abnormally high or low cholesterol levels, or abnormal cholesterol distribution, are detrimental to the auditory system [7, 10, 11, 34–36, 57, 58]. However, they have not explored the effect of cholesterol synthesis by hair cells on this system. Smith–Lemli–Opitz syndrome (SLOS) is a common autosomal recessive disorder caused by mutations in the DHCR7 gene. It is characterized by multiple congenital malformations and intellectual disability resulting from an inborn error of cholesterol synthesis. The syndrome exhibits a wide spectrum of manifestations, ranging from mild behavioral and learning difficulties to perinatal lethality. Additionally, patients commonly present with sensorineural hearing loss; however, auditory function has not been thoroughly investigated [59]. Our study revealed that HSD17B7, an enzyme involved in cholesterol synthesis, is enriched in sensory hair cells and regulates cholesterol levels in these cells, playing a crucial role in maintaining normal auditory function. We found that knocking down or knocking out HSD17B7 resulted in low cholesterol levels, leading to auditory deficiency, whereas overexpressing HSD17B7, which caused elevated cholesterol, did not affect auditory behaviors. Given that the stereocilia links of hair cell kinocilia require cholesterol anchoring, we hypothesize that the hair cell function requires cholesterol to maintain a minimum critical concentration. Cholesterol may stiffen biological membranes [60], a biophysical feature that can be directly involved in MET [61, 62]. Conserved protein-lipid interactions in the TMC1 and TMC2 complex structures of MET-related proteins have been highlighted by cryo-EM [63, 64]. Our research found the truncated variant HSD17B7^E182* disturbed cholesterol abundance and distribution in vitro. Considering that the absence of Hsd17b7 or the overexpression of HSD17B7^E182* can lead to abnormal MET function in vivo, we hypothesized that HSD17B7 might regulate MET function by affecting the abundance or distribution of cholesterol in hair cells. Our study suggests that cholesterol might regulate hearing function by affecting MET function. Therefore, how cholesterol abnormalities caused by HSD17B7 deficiency or mutations regulate MET function requires further investigation.

Dominant non-syndromic hearing loss is common in patients with hereditary deafness [65–72]. Typically, the dominant allele is inherited from one parent, and de novo mutations should be considered in all cases with a negative family history. Considering that the parents of the proband Ⅱ 1 had normal hearing but were not willing to provide blood samples for sequencing analysis, and combined with the Sanger sequencing analysis of the proband Ⅱ 1, it can be inferred that the proband Ⅱ 1 may have de novo dominant non-syndromic hearing loss. However, this conclusion requires further confirmation through additional clinical cases. Some examples of dominant-negative effects in deafness have been reported, such as KCQN4 and GJB2 [73–76]. Here, we report a case of sporadic deafness with an HSD17B7 (c.544G>T, p.E182*) heterozygous mutation, characterized by bilateral profound deafness with no other clinical features. We found that the mRNA of HSD17B7 (c.544G>T, p.E182*) could not rescue the abnormal auditory function caused by the hsd17b7 KO in a zebrafish. Interestingly, the overexpression of the HSD17B7 mutant mRNA in wild-type zebrafish impaired auditory function, indicating that this nonsense mutation has a negative effect. Our results suggest that the HSD17B7^E182* mutant may be due to the altered cholesterol distribution.

In conclusion, we demonstrated that Hsd17b7 is required for auditory function by regulating cholesterol synthesis in sensory hair cells. Hsd17b7 deficiency leads to cholesterol abnormalities, disrupting the MET and auditory function. Additionally, we identified a sporadic nonsense mutation in the HSD17B7 gene in patients with deafness. Our research provides new insights into the critical role of cholesterol in hearing preservation and lays a molecular function for targeted therapeutic strategies.

Materials and Methods

Mouse and zebrafish husbandry

For mouse husbandry, mice in the C57BL/6J background were purchased from the Laboratory Animal Center of Nantong University. Mice were maintained in a barrier facility at 25℃ on a regular 12-hours light and 12-hours dark cycle. The day of vaginal plug observation was embryonic day 0.5 (E0.5), and the day of birth was considered postnatal day 0 (P0). For zebrafish husbandry, zebrafish (Danio rerio) were raised and maintained at 28.5°C. Wild-type AB zebrafish line and the transgenic zebrafish line Tg(Brn3c: mGFP) were used in the study described in our previous study [77]. The embryonic stage is defined as described in the literature [78]. Embryos were collected after natural spawns and moved to E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, and 0.33 mM MgSO₄, pH 7.2). After 20 hours post-fertilization (hpf), embryos were moved to a medium containing 0.2 mM phenylthiourea (PTU) to prevent pigmentation for in situ hybridization and imaging. All animal experiments were performed under the guidelines of the Institutional Animal Care and Use Committee of Nantong University.

Cell culture

Mice HEI-OC1 were obtained from Prof. Renjie Chai. The cells were authenticated by short tandem repeat (STR) profiling (Shanghai Biowing Applied Biotechnology Co., Ltd). The cell lines were routinely tested negative for mycoplasma using the MycoBlue Mycoplasma Detection Kit (Vazyme Biotech, #D101-01), and cell aliquots from early passages were used. HEI-OC1 cells were cultured in DMEM (Wisent, #319-005-CL) medium supplemented with 10% FBS (Sigma-Aldrich, #F8318) in an incubator with 10% CO₂ at 33℃.

Human HEK293T (sex unknown) cells were obtained from the Shanghai Institute of Biochemistry and Cell Biology (SIBC) and authenticated by short tandem repeat (STR) profilin (Shanghai Biowing Applied Biotechnology Co.,Ltd). The cell lines were routinely tested negative for mycoplasma using the MycoBlue Mycoplasma Detection Kit (Vazyme Biotech, #D101-01), and cell aliquots from early passages were used. Cells were cultured in DMEM (Wisent, #319-005-CL) medium supplemented with 10% FBS (Sigma-Aldrich, #F8318) in an incubator with 5% CO₂ at 37℃.

Antibodies

Anti-MYO7A (1:150, Developmental Studies Hybridoma Bank, #138-1); anti-HSD17B7 polyclonal antibody (1:150-1:1000, proteintech, #14854-1-AP); anti-MYC tag polyclonal antibody (1:200, proteintech, #16286-1-AP); anti-His antibody (1:2500, proteintech, # 66005-1-Ig); anti-DYKDDDDK tag mouse monoclonal antibody (1:200-1:5000, proteintech, #66008-4-Ig); anti-calnexin polyclonal antiboy (1:100, proteintech, #10427-2-AP); tubulin monoclonal antibody (1:5000, proteintech, #66031-1-Ig); goat anti-rabbit IgG(H+L) HRP (1:5000, MULTI SCIENCES, #GAR007); goat anti-mouse IgG(H+L) HRP (1:5000, MULTI SCIENCES, #GAM007); FITC goat anti-mouse IgG (H+L) (1:200, ABclonal, #AS001); Cy3 goat anti-rabbit IgG (H+L) (1:200, ABclonal, #AS007); anti-Digoxigenin-AP Fab fragments (1:1000, Roche, #11093274910).

RNA Isolation, Reverse Transcription (RT), and Quantitative Real-time PCR (RT-qPCR)

Total RNAs of all samples were extracted using TRIzol reagent (Invitrogen, #15596026), and genomic DNA contamination was removed by DNase I (Vazyme, #EN401) following the manufacturer’s directions. The RNA yield was determined using NanoDrop ND-2000 (Thermo Fisher Scientific, USA), and integrity was checked on a 1% agarose gel.

The cDNA was synthesized by using HiScript III RT SuperMix (Vazyme, #R323-01) following the manufacturer’s directions. Subsequently, quantitative PCR was performed using the ChamQ SYBR qPCR Master Mix (Vazyme, #Q341-02) with the specified primers (HSD17B7 primer, F: 5’- GACAAGCTTGGATCCATGCGAA-3’, R: 5’- ACCTGGACAATGGTGACCTC-3’; Gapdh primer, F: 5’- CACAGTCAAGGCCGAGAATGGGAAG-3’, R: 5’- GTGGTTCACACCCATCACAAACATG) with a final volume of 20 μL under the following condition: 15 min at 50°C, 5 min at 95°C, and then 40 cycles at 95°C for 15 s and 60°C for 30 s. Relative expression levels for the Hsd17b7 gene were calculated using the 2^-△△CT method, normalized to the Gapdh. All reactions were repeated in triplicate for each sample. The results were analyzed using the GraphPad Prism software (version 9.4.0).

Whole-mount in situ hybridization and image acquisition

The whole-mount in situ hybridization (WISH) of zebrafish was performed according to the following standard procedures. A 403 bp cDNA fragment of zebrafish hsd17b7 gene was amplified via PCR using designed primers (F: 5’-GACGTCCTCCAGTAATGCCC-3’, R: 5’- CATCTTGCTTGGTCGGGTGT-3’) and was cloned into the pGEM-T-easy vector. After linearization of the pGEM-T-easy vector inserting the hsd17b7 fragment, the DIG RNA Labeling Kit (SP6) (Roche, #11175025910) was used to prepare digoxigenin-labeled hsd17b7 antisense mRNA probes through transcription in vitro. Subsequently, embryos at different developmental stages were hybridized with an hsd17b7 mRNA probe overnight after a series of treatments, including fixation in 4% paraformaldehyde in PBS, digestion in proteinase K (Roche, #3115879001), and incubation with a pre-hybridized mix. Finally, the alkaline phosphatase (AP)-conjugated antibody against digoxigenin (Roche, #11093274910) and the AP-substrate NBT/BCIP solution (Roche, #11681451001) were used to detect the hsd17b7 expression. The WISH images of zebrafish were acquired using a stereomicroscope (Olympus, MVX10, Japan).

Plasmids construction

The HSD17B7^WT and RER1 full-length cDNAs were PCR amplified from the HEK293T cDNA and cloned into the pGEX-TEV and pET-23b vector, respectively. HSD17B7^E182* was generated by site-directed mutagenesis. For cell transfection, the cDNA of HSD17B7^WT and HSD17B7^E182* were subcloned into the pCMV2-Flag vector, respectively. The cDNA of RER1 and HSD17B7^E182* were subcloned into the pcDNA3.1-Myc-His A vector, respectively. eGFP was subcloned into pCMV2-Flag-HSD17B7^WT and pCMV2-Flag-HSD17B7^E182*, respectively. pCS2-D4H-mCherry was obtained from G. Peng’s lab. For plasmid injection of zebrafish, myosin 6b promoter was subcloned into the p-mTol2 vector, followed by D4H-mCherry was subcloned into the p-mTol2 vector. The cDNAs encoding the human protein of HSD17B7^WT-eGFP and HSD17B7^E182*-eGFP were inserted into the p-mTol2-myo6b vector, then linking P2A-D4H-mCherry by overlap extension. For mRNA injection of zebrafish, the cDNAs encoding the human protein of HSD17B7^WT and HSD17B7^E182* were inserted into the pCS2⁺ and likewise, the cDNAs encoding the zebrafish full-length protein of Hsd17b7 were inserted into the pCS2⁺, which were linearized using NotI and then transcribed in vitro to generate mRNA. All constructs were verified by DNA sequencing.

Cell transfection

HEI-OC1 cells were transfected with plasmids using an X-treme GENE HP DNA transfection reagent (Roche, #6366236001) according to the manufacturer’s protocol. After 36 hours or 48 hours transfection, the cells were fixed or collected and subjected to further analyses, including western blots, Co-IP, IF, or RT-qPCR.

HEI-OC1 cells were transfected with small interfering RNA. The mouse Hsd17b7 siRNA oligos (F: 5’-GGAGGUGUUUGAAACCAAUTT-3’, R: 5’- AUUGGUUUCAAACACCUCCTT-3’) were synthesized from GenePharma (Shanghai, China). Cells were transfected with siRNA oligos to knock down mHSD7B7 using Lipofectamine RNAiMAX (Invitrogen, #13778-150) according to the manufacturer’s protocols. Approximately 1×10⁵ cells per well were placed in six-well plates. When the cell density reached 70-80%, 4 μL of control siRNA or mHsd17b7 siRNA were added to 150 μL of opti-MEM (Gibco, #31985070), then 5 μL Lipofectamine RANiMAX was added, incubated at room temperature for 5 min, then added them to the cell culture medium. After 48 hours transfection, the transfected cells were harvested for western blots or cholesterol assay.

mRNA stability assay

HEI-OC1 cells were transfected with plasmids using an X-treme GENE HP DNA transfection reagent. After 24 hours, the cells were treated with 10 µM actinomycin D (MCE, #HY-17559) for 0, 2, 4, 6 and 8 hours. Then collected cells and isolated RNA for RT-qPCR [79].

Immunostaining and image acquisition

Cultured cells grown on coverslips, after 36 hours transfection, cells were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature and washed 3 times with PBS, then followed by permeabilized in 0.3% PBST (0.3% Triton X-100 in PBS) for 20 min. After three times washes with PBS, the cells were blocked with 10% FBS in PBS, followed by incubation with anti-primary antibodies overnight at 4°C, and then incubated with fluorescent dye-labelled secondary antibodies for 2 hours at room temperature before mounting using the mounting solution with DAPI (1:500, SouthernBiotech, #011-20).

The dissected mouse cochleae were fixed with 4% paraformaldehyde in PBS for 1 hours at room temperature. After 3 times washes with 0.1% PBST, the cochleae were briefly blocked with a blocking medium (PBS containing 1% Triton X-100 and 10% heat-inactivated donkey serum, pH 7.2) for 1 hours at room temperature, followed by incubation with the anti-primary antibody overnight at 4°C. Then the samples were washed in 0.1% PBST and incubated with the fluorescent dye-labelled secondary antibodies for 2 hours at room temperature before mounting using the mounting solution with DAPI (1:500, SouthernBiotech, #011-20).

The immunostaining images of stained HEI-OC1 cells and cochlea were acquired by a Nikon confocal microscope with NIS-Elements software. For zebrafish live imaging, the larvae were anesthetized with tricaine MS-222 (Sigma, #A5040) and mounted in 0.6% low-melting agarose (Invitrogen, #16520050) with a lateral view. All the reconstructed three-dimensional images and adjusted contrast were used with Imaris X64 software (version 9.0.1).

FM4-64 labeling

To investigate the basal activity of neuromast hair cells, the FM4-64 (1:500, Invitrogen, #T13320) vital dye was used to specifically label functional HCs in the neuromasts. The staining procedures were carried out as described previously [25, 26, 32, 33]. The free-swimming larvae were incubated in 3 μM FM4-64 vital dye for 15 s at room temperature in the dark. Afterward, the fish were rinsed 3 times using a PTU medium and imaged. The images of stained hair cells were acquired by a Nikon confocal microscope with NIS-Elements software.

Measurement of total cholesterol

HEI-OC1 cells were homogenized in chloroform/methanol (2:1) to extract lipids and then centrifuged at 20,000 × g for 10 min. The organic phase was harvested and dried using nitrogen flow. Total cellular cholesterol was quantified using an Amplex Red cholesterol assay kit (Invitrogen, #A12216) through a multilabel reader (PerkinElmer, USA) according to the manufacturer’s instructions.

Morpholino and mRNA injections

For inhibiting the expression of hsd17b7, hsd17b7-specific splicing-blocking morpholinos were designed and procured from Gene Tools, Inc. and the precise sequence was (5’- TGCAAACAGGTAACAAAACTGTGTG-3’). The powder of morpholino was dissolved and diluted in RNase-free water to obtain the work solution with a 0.3 mM final concentration for subsequent operations. About 2 nL dose of morpholino work solution was microinjected into zebrafish embryos at the one-cell stage. To examine the morpholino efficiency, embryos injected with morpholino were collected to extract RNA which then was subjected to reverse transcript cDNA. The designed primers franking on exon 1 and exon 2 (F: 5’- TACACAGGCAAACGTTAGAAGC-3’, R: 5’-CTTGAACTCCTCTGCACCCTT-3’) were used to amplify the fragment containing the mis-splicing target site which was located at the connection of exon 1 and intron 1. For rescue experiments, exogenous mRNA was first synthesized by transcription in vitro. Briefly, the designed primers (zebrafish hsd17b7 mRNA primer, F: 5’-CGCGGATCCATGAAGAAAGTAGTTTTGGT-3’, R: 5’- CCGGAATTCTCACATTCCATTTCTTTCTT-3’; human HSD17B7^WT mRNA primer, F: 5’- CGGGATCCATGCGAAAGGTGGTTTTGATC-3’, R: 5’- GCTCTAGATTATAGGCATGAGCCACTGA-3’; human HSD17B7^E182* mRNA primer, F: 5’-CGGGATCCATGCGAAAGGTGGTTTTGATC-3’, R: 5’-GCTCTAGACTAGAGGCTGAAATTAGATT-3’) were used to amplify target DNA containing the coding sequence. Then, the pCS2 vector inserted into the amplified fragment was linearized as a template for the transcription of mRNA utilizing the SP6 mMESSAGE mMACHINE Kit (Invitrogen, #AM1340). After purifying using the RNeasy Mini Kit (Qiagen, #74104), mRNA with a 70 ng/μL concentration was co-injected into one-cell stage embryos with morpholino for rescue experiments.

sgRNA/ Cas9 mRNA synthesis and injection

sgRNA was designed against the hsd17b7 gene (ENSDARG00000088140) using the CRISPR design tool (https://www.crisprscan.org). The sgRNA specifically targeting exon 3 of hsd17b7 (5’-GGGAATCATGCCTAATCCCA-3’) was first synthesized as follows. The GenCrispr sgRNA synthesis kit is used to generate a gRNA DNA template with a T7 promoter and synthesize gRNA upon in-vitro transcription (Genscript Biotech Co.Ltd., Nanjing, China). The pXT7-zCas9 plasmid was linearized using XbaI and then transcribed in vitro to generate Cas9 mRNA using the mMESSAGE mMACHINE T7 kit (Invitrogen, #AM1344). At the one-cell stage, embryos were injected with a 2-3 µL solution containing 300 ng/µL Cas9 mRNA and 150 ng/µL sgRNA. The injected embryos were then grown in an E3 medium. At 24 hpf, genomic DNA was extracted from injected embryos, and potential CRISPR-induced mutations were identified by PCR. Primers used for identification were designed around the hsd17b7 sgRNA target sites (F: 5’-AAAAACTTATTTTATTCCAGCCCAA-3’, R: 5’- TTTCCATGCAGCACTATCAAACAATT-3’).

Startle Response Test

The acoustic startle reflex was performed as described previously [30, 80]. Briefly, A plastic plate attached to a mini vibrator was used to place 20 normal larvae at 5 days post-fertilization (dpf), while an infrared digital video tracking system was used to monitor their swimming behavior. The 60 Hz tone bursts with two different sound levels of 9 dB re. m s⁻² were applied to the amplifier to drive the vibrator. The acoustic vibration stimuli lasting for 30 ms with a 180 s inter-stimulus interval were set and applied. Each sound vibration stimulus level was repeated 20 times and the locomotion behavior of the larvae with C-shape motion to this stimulus was recorded. Finally, the movement typical parameters of mean distance and peak velocity were analyzed to assess the startle response of larvae to sound vibration stimuli.

In vitro binding assay

GST and GST-HSD17B7^WT fusion proteins were immobilized to GSH resins were incubated with purified His-RER1 proteins overnight at 4℃ and extensively washed with washing buffer (0.3% PBST). Bound proteins were separated using SDS-PAGE and visualized with Coomassie Blue staining and Western blots.

Western blots

Frozen cells were lysed in RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% sodium deoxycholate, 1% w/v protein inhibitor (Roche, #4693132001)) for 30 min and centrifuged at 12,000 g for 15 min at 4℃. Protein concentrations were determined using a BCA protein assay kit (Thermo Fisher, #23227).

The proteins were separated on SDS-PAGE gels and subsequently transferred to a PVDF membrane (Millipore, USA). The membranes were blocked in blocking buffer (1 × TBS containing 0.5% milk and 0.5% Tween 20) for about 45 min at room temperature, then incubated overnight with primary antibodies in TBS containing 4% BSA, 1% Tween20, and 0.05% NaN₃. Next day, the PVDF membranes were incubated with the appropriate HRP-conjugated secondary goat antibodies for 2 hours at room temperature after extensive washing with a blocking buffer. The blots were detected using ECL TM Western Blotting Detection Reagents (GE, #RPN2106), and acquired images were analyzed with ImageJ software (version 1.8.0).

Co-IP, Mass spectrum, and Gene Ontology (GO) analysis

Cell lysates from Flag, Flag-HSD17B7^WT, and Flag-HSD17B7^E182* transfected cells were incubated with 30 μL of beads conjugated to anti-Flag antibodies (Smart-Lifesciences, #SA042001) respectively overnight at 4℃. The next day, Flag beads were washed with lysis buffer 3 times, then the immune complexes were eluted with 2 × SDS sample buffer and subjected to three samples for SDS-PAGE and mass spectrum analysis.

For mass spectrum, the immunoprecipitated HSD17B7^WT and HSD17B7^E182*-associated proteins were washed and dissolved in ammonium bicarbonate buffer (25 mM, pH 8.0), followed by trypsin digestion at 37℃ for 16 hours. The nano-LC-MS/MS experiments were performed with LTQ-Orbitrap MS (Thermo Fisher, USA) equipped with a Nano electrospray ion source.

For Venn diagram and GO analysis, microarray analysis identified HSD17B7^WT and HSD17B7^E182*-specific associated proteins by mass spectrum (peptide ≥ 1, unique peptide ≥ 1, P value < 0.05). Cellular component GO term enrichment analysis were performed using the web-based DAVID software [81]. Ontology networks were further investigated and visualized using R (version 4.4.1).

ScRNA-seq and data processing

Zebrafish scRNA-seq data was downloaded from NCBI Gene Expression Omnibus (GSE221471) [24]. The basic procedure for single-cell integrated data analysis was carried out in the Seurat 4.0.1 platform [82]. We selected five cell populations for re-analysis based on previous reports, including neuromast hair cells (cluster 0), supporting cells (cluster 1), macula hair cell (cluster 2), cirsta hair cell 1 (cluster 3), mantle cells (cluster 4), and crista hair cell 2 (cluster 5). Gene ontology (GO) and KEGG pathway enrichment analysis were performed using the R package clusterProfiler with the marker genes identified by Seurat V4.0.1 as the input, and the results were visualized using the enrich plot method [83].

To construct single cell trajectories, Monocle 3 (version 1.0.1) was used to cluster cells and reduce the dimensionality of gene expression matrices using UMAP. Monocle 3 package documentation was downloaded from GitHub. Subsets of cells that lead to the hair cell lineage were chosen using the “choose_cells” function for downstream analysis. The cells were ordered in pseudotime according to their development progress using the “order_cells” function and re-clustered in Monocle 3 for pseudotime and differential expression analysis. Cell paths were predicted by the “learn_graph” function of Monocle 3. The clusters identified as mantle cells furthest from the hair cells were chosen as the ‘‘roots’’ of the trajectory. Hair cells (cluster 0), supporting cells (cluster 1), and mantle cells (cluster 2) paths were selected separately (choose_cells) to plot their marker expression along pseudotime (plot_genes_in_pseudotime). Mouse scRNA-seq data was downloaded from previous articles (GSE71982, (GSE168901, GSE202920) [27–29]. The basic procedure for single-cell integrated data analysis and correction of batch effects were carried out in the Seurat 4.0.1 platform [82]. We selected four cell populations for re-analysis, including inner hair cells, outer hair cells, utricle hair cells and crista hair cells.

Whole-exome sequencing and Sanger sequencing analyses

Whole-exome sequencing was performed in the proband Ⅱ 1. Genomic DNA samples were extracted from whole blood samples of the proband Ⅱ 1 (Figure 4A).

The exomes and flanking intronic regions from whole blood DNA samples were captured by Agilent SureSelect Human All Exon Kit (Agilent Technologies, USA). The captured DNA was sequenced on Illumina HiSeq 4000 sequencing platform (Illumina, USA). Bioinformatics were aligned to NCBI build37/hg19 assembly using the BWA (version 0.7.12) software. Each sample was covered to an average sequencing depth of at least 100×. SNPs and indels were identified using GATK HaplotypeCaller software. Candidate pathogenic variants were defined as nonsense, missense, splice-site, and indel variants with allele frequencies of 0.001 or less in public variant databases dbSNP, 1000 Genomes, ESP6500, nci60, GnomAD and in disease database of COSMIC, Clinvar, OMIM, GWAS. Genotypes distributed in every 0.3 cM of genomic region were chosen for calculation of the logarithm of odds (LOD) scores using the Merlin v. 1.1.2 parametric linkage analysis package. The proband Ⅱ 1 was genotyped by Sanger sequencing analyses with designed primers (F: 5’-GTACTCTGATTGGTGACGGGTGAG-3’, R: 5’- GACAGTCATAGTTCATAGTTTATT-3’)

Quantification and statistical analysis

GraphPad Prism 9 (version 9.4.0) supported the whole statistical analysis. All data were repeated more than three times and presented as mean ± SD. An unpaired two-tailed student’s t-test was performed for two-group comparisons, while multiple comparisons were illustrated using one-way ANOVA. The p-value less than 0.05 (p ≤ 0.05) was considered as significantly different. p ≤ 0.05, p ≤ 0.01, p ≤ 0.001, and p ≤ 0.0001 were symbolized with “*, **, ***, and ****”, respectively, and “ns” represented no significance, p > 0.05.

Supplementary Figures

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Acknowledgements

This work was supported by the National Natural Science Foundation of China Grants (32200783 to Y.Q.S; 92368104 and 32350017 to D.L.) and the Natural Science Foundation of Jiangsu Province Grants (BK20220607 to Y.Q.S).

Additional information

Resource and Data Availability

The plasmids and zebrafish lines generated in this study are available on request to the Lead Contact. The published article includes all datasets generated or analyzed during this study. Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

Author Contributions

Conceptualization, Y.Q.S., Z.Y.W., and D.L.; methodology, Y.Q.S., Z.Y.W., M.J.Z., Xu.W., F.P.Q.; investigation, Y.Q.S., Z.Y.W., and D.L.; writing - original draft, Y.Q.S., Z.Y.W., Xi.W., and D.L.; writing - review & editing, Y.Q.S., Z.Y.W., and D.L.; funding acquisition, Y.Q.S., and D.L.; resources, Z.M.J., and J.C.; supervision, D.L and J.C.