Introduction

Hearing loss is one of the most common sensory disorders worldwide and represents a major global health burden. According to the World Health Organization, more than 466 million people currently suffer from disabling hearing loss, and this number is expected to increase substantially in the coming decades [1]. Genetic factors account for approximately 60% of congenital hearing loss, and mutations in genes that regulate the development and function of sensory hair cells (HCs) are the main cause of hereditary deafness. Although more than 150 non-syndromic deafness genes have been identified [2], the molecular mechanisms underlying HC dysfunction remain incompletely understood, and many pathogenic genes remain to be discovered.

HCs are the primary mechanoreceptors of the auditory and vestibular systems, converting mechanical stimuli into electrical signals through mechanotransduction (MET). This process critically depends on the integrity and biophysical properties of hair cell membranes, particularly those of the stereocilia. Cholesterol is the most abundant sterol molecule in mammalian cells and plays essential roles in maintaining membrane structure, synthesizing important hormones, facilitating synapse formation, and mediating cell signaling transduction [3–6]. Previous studies have shown that dysregulated intracellular cholesterol homeostasis contributes to auditory defects [6–8], including hereditary [9–13], noise-induced [14–16], ototoxic [17, 18], and age-related hearing loss [19–21]. HSD17B7 is a member of the hydroxysteroid dehydrogenase family and functions as a key enzyme in the cholesterol biosynthetic pathway by catalyzing the conversion of zymosterone to zymosterol. In addition to its role in steroid metabolism [22] and its involvement in hormone-related cancers [23, 24], HSD17B7 has been reported to be expressed in multiple tissues, including the brain, eye, and inner ear [25, 26]. Notably, single-cell RNA sequencing (scRNA-seq) and immunostaining analyses have suggested that HSD17B7 is enriched in mouse vestibular HCs [27]. Despite this expression pattern, the functional role of HSD17B7 in HCs and MET has not been investigated, and no pathogenic variants have been reported to be associated with hearing loss.

Here, our findings reveal a previously unrecognized link between hair cell–intrinsic cholesterol biosynthesis and the function of sensory hair cells, and identify HSD17B7 as a candidate gene underlying sensory hearing disorders.

Results

Hsd17b7 is 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 hsd17b7 in the developing zebrafish. scRNA-seq data analysis (accession no. GSE221471) [28] categorized lateral line hair cells (LLHCs) (cluster 0), supporting cell (cluster 1), macula hair cells (MHCs) (cluster 2), crista hair cells (CHCs) 1 (cluster 3), mantle cell (cluster 4), and CHCs 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 HCs, especially in LLHCs and CHCs (Figure 1B and C). Quantitative comparison of average expression levels within LLHCs indicates that hsd17b7 is expressed at a level comparable to several known MET-associated genes (e.g., tmc1 and lhfpl5a) (Figure 1D). To examine the temporal dynamics of hsd17b7 expression during LLHCs differentiation, LLHCs, supporting cells, and mantle cells were ordered along pseudotime using Monocle 3 (Figure 1E). A heatmap was generated to visualize the expression dynamics of marker genes along the LLHC developmental trajectory (Figure 1F). These analyses suggested a gradual increase in hsd17b7 expression during LLHC maturation (Figure 1E and F). Consistently, whole-mount in situ hybridization (WISH) further showed that hsd17b7 was specifically enriched in neuromast and crista regions at 72 and 96 hours post-fertilization (hpf) (Figure 1G). To further explore the subcellular localization of hsd17b7 in zebrafish HCs, we generated a hsd17b7-egfp fusion construct driven by a HC-specific myo6b promoter [29, 30]. Confocal imaging revealed that hsd17b7-EGFP localized predominantly to the cytoplasm in both CHCs and LLHCs, exhibiting a punctate distribution pattern (Figure 1H and I).

Figure 1.

To further elucidate Hsd17b7 expression in the mammalian auditory system, we analyzed three published scRNA-seq datasets from mouse inner ear tissues [31–33]. These analyses revealed that Hsd17b7 was enriched in sensory hair cells, including outer hair cells (OHCs), inner hair cells (IHCs), and utricular hair cells (UHCs), while lower expression levels were observed in CHCs (Figure 1J). Next, we performed immunostaining to assess Hsd17b7 protein expression in the organ of Corti. Hsd17b7 was detected in both OHCs and IHCs at P1, P4, and P7 (Figure 1K). The specificity of the Hsd17b7 antibody was further supported by independent validation experiments in HEI-OC1 cells (Figure S1A and B). These results indicate that Hsd17b7 is expressed and enriched in HCs in both zebrafish and mice.

In addition, phylogenetic and sequence alignment analysis revealed that vertebrate HSD17B7s share a significant similarity in amino acid sequences (Figure S2A). 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 S2B). These data suggest that Hsd17b7 is conserved and expressed in HCs across vertebrate species.

Hsd17b7 deficiency impaired acoustic startle responses and MET activity in zebrafish

To investigate the hsd17b7 function in the auditory system, we generated hsd17b7 mutant alleles using CRISPR/Cas9-mediated genome editing (Figure S3A–D). Sequence analysis identified multiple mutant alleles, including a 4-bp deletion and two 6-bp deletions in exon 3 (Figure S3E). The 4-bp deletion allele was selected for further analysis, as it led to a frameshift and premature stop codon, producing a severely truncated protein lacking most of the conserved catalytic domain required for hsd17b7 enzymatic activity (Figure S3F).

We first examined acoustic startle responses [34] at 5 dpf (Figure 2A). Compared with control larvae, hsd17b7 mutants exhibited significantly reduced movement trajectories, peak swimming velocity, and total travel distance following acoustic stimulation (Figure 2A–C), indicating impaired startle response. Microinjection of hsd17b7 mRNA into mutant embryos significantly restored these behavioral phenotypes (Figure 2A–C), confirming that the phenotype was specifically caused by loss of Hsd17b7.

Figure 2.

To validate these findings, antisense morpholino oligonucleotides (hsd17b7 Mo) targeting the junction of exon 1/intron 1 were designed to knock down hsd17b7 expression (Figure S4A and B). RT-PCR and immunoblotting analyses confirmed effective disruption of normal splicing and a substantial reduction in Hsd17b7 protein levels, both of which were rescued by co-injection of hsd17b7 mRNA (Figure S4C and D). Consistent with the mutant phenotype, hsd17b7 morphants displayed significantly reduced acoustic startle responses at 5 dpf, including decreased movement trajectories, swimming distance, and peak velocity, all of which were effectively rescued by hsd17b7 mRNA injection (Figure S4E–G).

In the Tg(Brn3c: mGFP) background [28, 35], confocal imaging revealed a significant reduction in the number of LLHCs in hsd17b7 mutants (∼16% reduction) (Figure 2D and E). To assess whether hsd17b7 loss affects the MET activity, we performed FM4-64 uptake assays, which label HCs through functional MET channels [26, 32, 33]. Quantification of FM4-64 fluorescence intensity per HC revealed a pronounced reduction in MET-dependent dye uptake in hsd17b7 mutant HCs at 5 dpf (∼40%; Figure 2F and G). A similar decrease was observed in hsd17b7 morphants (∼25%; Figure S5A and B). Importantly, hsd17b7 mRNA injection restored both FM4-64 uptake and HC number in mutants and morphants (Figure 2D–G, Figure S5A and B), indicating that hsd17b7 is required for normal MET activity in HCs.

Hsd17b7 deficiency disrupted cholesterol-associated transcriptional programs in hair cells

To investigate intrinsic transcriptional changes in HCs caused by hsd17b7 deficiency, we performed scRNA-seq on FACS-isolated mGFP⁺ HCs from control and hsd17b7 mutant (Figure 3A). Clustering analysis identified major inner ear (IE) and lateral line (LL) cell populations, including HCs and SCs (Figure 3B; Figure S6 A and B). Focusing on LLHCs, pseudobulk differential expression analysis revealed extensive transcriptional remodeling in hsd17b7 mutants, with 1355 genes differentially expressed, including 853 upregulated (63%) and 502 downregulated (37%) genes relative to controls (Figure 3C). Gene Ontology enrichment analysis of differentially expressed genes highlighted pathways related to lipid metabolism, membrane organization, and sensory perception, indicating broad alterations in cellular homeostasis upon disruption of cholesterol biosynthesis (Figure 3D and E).

Figure 3.

To further assess cholesterol- and MET-associated transcriptional programs, we performed module score analysis in both LLHCs and IEHCs. In mutant IEHCs, MET, cholesterol biosynthesis, and cholesterol uptake-related gene sets were significantly upregulated, whereas gene modules associated with tip-link components and cholesterol efflux were concurrently reduced (Figure 3F and G).

In mutant LLHCs, module score analysis revealed a partially overlapping transcriptional response. Gene modules associated with tip-link components and cholesterol efflux were reduced, whereas modules representing the core MET machinery and cholesterol biosynthesis enzymes did not show significant global changes, consistent with selective remodeling rather than uniform pathway activation or repression. At the gene level, cholesterol regulatory programs appeared partially uncoupled, with downregulation of sterol-sensing transcriptional regulators (e.g., srebf2, insig1, and mbtps1) accompanied by variable expression changes among cholesterol biosynthetic enzymes. In line with this pattern, gene set enrichment analysis (GSEA) indicated an overall attenuation of cholesterol biosynthesis–associated pathways in mutant LLHCs (Figure 3H).

Together, these findings indicate that hsd17b7 deficiency induces selective and context-dependent remodeling of cholesterol- and MET-associated transcriptional programs across HC populations.

Hsd17b7 is required for maintaining cholesterol levels in hair cells

Hsd17b7 converts zymosterone to zymosterol, participating in cholesterol biosynthesis (Figure 4A) [26]. Given that both elevated and reduced cholesterol levels are detrimental to auditory function [7, 36], we first verified whether hsd17b7 regulates cholesterol homeostasis in hair cells.

Figure 4.

In vitro, knockdown of Hsd17b7 in HEI-OC1 cells resulted in a significant reduction in total cellular cholesterol levels (Figure 4B and C). To further examine cholesterol distribution, we employed the D4H cholesterol-binding probe derived from perfringolysin O, which selectively labels cholesterol at the cytoplasmic leaflet of the plasma membrane [37, 38]. Overexpression of human HSD17B7 markedly increased D4H-mCherry fluorescence intensity in HEI-OC1 cells compared with controls (Figure 4D and E), indicating that HSD17B7 positively regulates cellular cholesterol abundance.

To further assess whether hsd17b7 regulates cholesterol in vivo, we expressed D4H-mCherry in control and hsd17b7 mutant zebrafish hair cells using the Tg(Brn3c:mGFP) background (Figure 4F). Consistent with previous reports [37], D4H-mCherry was enriched in the hair bundle (Figure 4G). Notably, cholesterol levels, as indicated by D4H-mCherry intensity, were significantly reduced in both CHCs and LLHCs of hsd17b7 mutants at 4 dpf (Figure 4G and H). A comparable reduction was observed in hsd17b7 morphants (Figure S5C–F).

Together, these results demonstrate that hsd17b7 is required for maintaining cholesterol levels in HCs, both in vitro and in vivo. Given the essential role of cholesterol homeostasis in MET and auditory function [7,36,39], loss of Hsd17b7 likely compromises MET activity and acoustic startle responses by disrupting cholesterol availability in HCs.

Identification of a candidate HSD17B7 nonsense variant associated with hearing loss

Given the established role of Hsd17b7 in HCs, we investigated whether variants in the human HSD17B7 might be associated with hearing loss. Whole-genome sequencing (WGS) was performed in individuals with hearing loss who remained undiagnosed after screening of 201 known deafness-associated genes [39]. We identified a heterozygous nonsense variant in HSD17B7 (NM_016371.4), c.544G>T (p.E182*), in a child with bilateral profound congenital hearing loss (Figure 5A–C). The proband failed the newborn hearing screening, required special education during early childhood, and was diagnosed with bilateral profound hearing loss by 8 years of age. Physical examination revealed two preauricular appendages anterior to the right ear, without additional systemic abnormalities. The proband Ⅱ 1’s parents had normal hearing and no family history of hearing loss, but declined genetic testing, precluding segregation analysis (Figure 5A and B).

Figure 5.

The p.E182* variant introduces a premature stop codon at position Glu182, resulting in a truncated HSD17B7 protein (Figure 5C). The multiple sequence alignment showed that Glu182 is evolutionarily conserved across vertebrates, from zebrafish to humans (Figure 5D). According to UniProt annotation, HSD17B7 contains an extracellular domain (1-229 aa), a transmembrane region (230-250 aa), and a cytoplasmic domain (251-341 aa). The p.E182* variant is predicted to remove the transmembrane region and cytoplasmic domain (Figure 5E). Together, these analyses identify p.E182* as a candidate variant in HSD17B7 that may be associated with hearing loss.

The HSD17B7^E182* truncation failed to rescue MET activity and auditory-related behaviors in hsd17b7 mutants

To assess the functional impact of the p.E182* variant, human HSD17B7 or truncation (HSD17B7^E182*) mRNAs were microinjected into the fertilized one-cell stage embryos of hsd17b7 mutants. Injection of HSD17B7 mRNA significantly increased the FM4-64 uptake in LLHCs of hsd17b7 mutants, whereas the injection of HSD17B7^E182* mRNA failed to restore FM4-64 labeling (Figure 6A and B), indicating that the truncated protein is unable to rescue normal MET activity. Consistently, acoustic startle response assays revealed that HSD17B7 mRNA rescued behavioral responses of hsd17b7 mutants, as assessed by movement trajectory, swimming distance, and peak velocity. In contrast, the p.E182* variant showed no rescuing effect on these behavioral deficits (Figure 6C–E).

Figure 6.

The HSD17B7^E182* truncation disrupts intracellular cholesterol distribution and MET activity

To explore the potential contribution of HSD17B7^E182* truncation to the patient’s hearing loss, we compared the cellular activation of the wild-type and truncated HSD17B7 in vitro and vivo. We analyzed the subcellular localization of Flag-tagged HSD17B7 and HSD17B7^E182* in HEI-OC1 cells. Since cholesterol synthesis mainly occurs in the endoplasmic reticulum (ER) [40, 41], HSD17B7 was expected to localize to this compartment. Consistent with this expectation, immunostaining in HEI-OC1 cells showed that Flag-tagged HSD17B7 co-localized with the ER marker Calnexin. In contrast, HSD17B7^E182* failed to localize to the ER and instead displayed a spot-like aggregate distribution (Figure 7A), a pattern further supported by fluorescence intensity profile analyses (Figure 7B and C).

Figure 7.

Given that the patient is heterozygous for the variant, we next examined whether HSD17B7^E182* interferes with HSD17B7 subcellular localization. Co-expression of Flag-HSD17B7 and HSD17B7^E182*-Myc in HEI-OC1 cells showed that HSD17B7 retained its ER localization, indicating that the truncated protein does not significantly alter the subcellular distribution of the wild-type protein (Figure 7D and E).

To further investigate the function of HSD17B7^E182* expression in vivo, we overexpressed human HSD17B7 or HSD17B7^E182* in zebrafish HCs. The FM4-64 uptake assay revealed that the overexpression of HSD17B7 did not significantly affect FM4-64 fluorescence intensity in LLHCs compared to the control. In contrast, the expression of HSD17B7^E182* led to a significant reduction of fluorescence intensity (Figure 8A and B). Consistently, startle response assays showed that HSD17B7 overexpression showed comparable movement trajectories, distance, and velocity in response to auditory stimulation, whereas larvae expressing HSD17B7^E182* reduced startle-associated behaviors (Figure 8C-E). These data suggest that expression of HSD17B7^E182* negatively impacts auditory-related behavior in vivo.

Figure 8.

Given our findings that Hsd17b7 regulates cholesterol homeostasis (Figure 4) and previous research that proper cholesterol distribution is critical for HC function [37], we next examined whether the HSD17B7^E182* truncation disrupts intracellular cholesterol organization. HEI-OC1 cells were co-transfected with cholesterol probe D4H-mCherry together with pCMV-Flag, pCMV-Flag-HSD17B7, or pCMV-Flag-HSD17B7^E182*. While cholesterol distribution appeared as a diffuse pattern around the nucleus in the control and HSD17B7-expressing cells, HSD17B7^E182* resulted in a spot-like aggregation of D4H-mCherry (Figure 8F). Line-scan intensity analysis further confirmed the aberrant colocalization of HSD17B7^E182* with cholesterol-enriched compartments (Figure 8G-I).

To validate these observations in vivo, HSD17B7 or HSD17B7^E182* was expressed in zebrafish HCs by injecting pDest-myo6b-HSD17B7-EGFP or pDest-myo6b-HSD17B7^E182*-EGFP plasmids into the Tg(myo6b: D4H-mCherry) transgenic line (Figure 8J). In HCs expressing HSD17B7, the protein exhibited a punctate intracellular distribution and did not noticeably alter the D4H-mCherry cholesterol signal compared with neighboring non-expressing hair cells. In contrast, HSD17B7^E182* showed a diffuse localization throughout the cytoplasm and nucleus, accompanied by a marked reduction or near-complete loss of the D4H-mCherry signal in the stereocilia position. These data suggest that HSD17B7^E182* with aberrant subcellular localization may bind cholesterol and alter its intracellular distribution. Altogether, these results demonstrate that HSD17B7^E182* has a negative effect by altering cholesterol distribution in HCs, thereby compromising MET function and impaired startle responses.

HSD17B7^E182* truncation disrupted the interaction with RER1

To further explore the pathological consequences of HSD17B7^E182*, we compared the interaction landscapes of wild-type and mutant in HEI-OC1 cells (Figure 9A). Proteomic analysis of co-immunoprecipitated complexes revealed distinct interaction profiles between HSD17B7 and HSD17B7^E182* (Figure 9A–C; Figure S7). Proteins uniquely associated with HSD17B7 were strongly enriched for cholesterol metabolic processes, whereas those preferentially associated with HSD17B7^E182* were enriched for chromatin remodeling–related pathways.

Figure 9.

Consistent with their subcellular localizations, cellular component analysis showed that HSD17B7-interacting proteins were predominantly enriched in the endoplasmic reticulum (ER), whereas HSD17B7^E182*-associated proteins were enriched in the nucleus and cytoplasm (Figure 9D), in agreement with immunostaining results (Figure 7A–C). Among ER-associated interactors, RER1 emerged as the top HSD17B7-specific binding partner (Figure 9E). RER1 is known to mediate ER retention of membrane-associated proteins [42, 43]. Direct binding between HSD17B7 and RER1 was confirmed by in vitro binding assays, whereas the HSD17B7^E182* failed to interact with RER1 (Figure 9F). Immunofluorescence analysis further demonstrated robust co-localization of RER1 with HSD17B7 but not with HSD17B7^E182* (Figure 9G–I), indicating that the p.E182* mutation disrupts RER1-mediated ER retention. Given that HSD17B7 is normally localized to the ER, loss of RER1 interaction provides a mechanistic explanation for the aberrant subcellular distribution of the mutant protein.

In addition to altered localization, the p.E182* mutation markedly reduced HSD17B7 expression levels. Compared with HSD17B7, the mutant protein exhibited substantially decreased abundance, accompanied by a significant reduction in mRNA levels (Figure S8A– C). mRNA stability assays revealed a shortened half-life of HSD17B7^E182* transcripts relative to wild-type (Figure S8D and E), indicating that the nonsense mutation compromises transcript stability and consequently reduces protein expression.

Discussion

This study reveals that HSD17B7 is enriched in sensory hair cells in zebrafish and mice. Using the zebrafish model, we show that loss of hsd17b7 markedly reduces cholesterol levels in hair cells, leading to impaired MET function and abnormal hearing behaviors. These findings identify HSD17B7 as a previously unrecognized regulator of hair cell physiology and a candidate gene for sensory hearing loss. In addition to the animal models, we identified a previously undescribed heterozygous nonsense variant in HSD17B7 (c.544G>T, p.E182*) in a patient with sporadic deafness. Functional analyses revealed that the residual HSD17B7^E182* protein alters subcellular localization and disrupts normal intracellular cholesterol distribution. Moreover, the mutation reduces mRNA half-life, decreases mRNA abundance, and significantly lowers protein expression. Collectively, these results suggest that the heterozygous c.544G>T (p.E182*) variant contributes to auditory dysfunction through potential pathogenic mechanisms: haploinsufficiency caused by reduced HSD17B7 expression and functional impairment due to altered cholesterol distribution. These mechanistic insights highlight the critical role of cholesterol homeostasis in MET and auditory-related function.

HSD17B7 was first identified as a prolactin receptor-associated protein in rats [44] and was proposed to be involved in estradiol biosynthesis [45, 46]. Subsequent biochemical studies demonstrated that HSD17B7 participated in cholesterol synthesis by catalyzing the reduction of zymosterone to zymosterol [47–49]. Consistent with this role, both in vivo and in vitro studies have shown that loss of HSD17B7 reduces cholesterol levels, whereas its overexpression increases cholesterol abundance. Previous studies have shown that HSD17B7 is expressed in the liver, heart, brain, eye, and ear [25, 26, 50], and scRNA-seq datasets have suggested its expression in mouse vestibular HCs [27]. Although the role of the HSD17B7 gene in ovarian and breast cancer has been extensively explored in the literature [51–55], the role of the HSD17B7 gene in auditory function remains unclear. Our results fill this gap by demonstrating that HSD17B7 is enriched in HCs, as supported by whole-mount in situ hybridization, immunostaining, and scRNA-seq analyses from our study and others [28, 31–33]. Given that loss of hsd17b7 causes abnormal auditory-related behaviors, these data establish HSD17B7 as a conserved and essential regulator of hair cell function.

Cholesterol is a fundamental component of biological membranes, contributing to both their structural integrity and functional properties [56, 57]. Previous studies have shown that either excessive or insufficient cholesterol levels, as well as abnormal cholesterol distribution, are detrimental to the auditory system [7, 10, 11, 36, 37, 58, 59]. However, previous studies have largely focused on systemic or membrane cholesterol alterations and have not addressed the contribution of hair cell–intrinsic cholesterol biosynthesis. Smith–Lemli–Opitz syndrome (SLOS), caused by mutations in DHCR7, exemplifies how impaired cholesterol synthesis can lead to multisystem defects, including sensorineural hearing loss, although auditory function in these patients has not been extensively investigated [60]. Our study revealed that HSD17B7, a cholesterol biosynthetic enzyme, is enriched in sensory hair cells and plays a crucial role in maintaining cholesterol levels required for auditory-related function. It should be noted that some localization and cholesterol distribution analyses were performed using C-terminal EGFP-tagged HSD17B7 constructs, which may influence protein behavior. However, the central conclusions of this study are supported by complementary loss-of-function models, rescue experiments using untagged mRNA, and independent cholesterol perturbation assays, reducing reliance on any single overexpression or fusion-based approach.

Here, we found that loss of Hsd17b7 or expression of the truncated HSD17B7^E182* variant reduced cholesterol abundance and disrupted its intracellular distribution, resulting in defective MET function and auditory-related impairment. In contrast, overexpression of wild-type HSD17B7 elevated cholesterol levels but did not impair auditory-related behavior, suggesting that hair cell function requires a minimum critical cholesterol threshold rather than being sensitive to moderate cholesterol excess. Freeze-fracture studies have shown that the stereociliary membrane is densely enriched in cholesterol [61], supporting the notion that cholesterol is a key determinant of hair bundle membrane properties. Cholesterol is known to stiffen biological membranes [62], a biophysical feature that may influence MET channel gating [63, 64]. Consistent with this idea, cryo-EM studies have revealed conserved protein– lipid interactions within the TMC1/TMC2-containing MET channel complex [65, 66]. Together with our findings, these observations support a model in which HSD17B7 regulates MET function by maintaining appropriate cholesterol abundance within hair cells. How cholesterol perturbations caused by HSD17B7 deficiency or mutation mechanistically alter MET channel activity remains an important question for future studies.

Dominant non-syndromic hearing loss is a common form of hereditary deafness [67–74] [75–78], and de novo mutations are frequently observed in patients with a negative family history. In the present study, both parents of the proband exhibited normal hearing and declined genetic testing, suggesting that the HSD17B7 (c.544G>T, p.E182*) variant may represent a de novo dominant mutation. This interpretation is consistent with previous reports showing that homozygous deletion of Hsd17b7 in mice is embryonic lethal, whereas heterozygous phenotypes have not been systematically characterized [79]. Notably, the mutant mRNA failed to rescue auditory defects in hsd17b7 mutants, and overexpression of HSD17B7^E182* in wild-type animals impaired startle response. Although additional clinical cases will be required to confirm pathogenicity, our functional analyses provide experimental support for a negative effect of the HSD17B7^E182* variant. Together with the marked reduction in mRNA and protein stability, these findings suggest that the HSD17B7^E182* variant primarily acts through loss of function, with the possibility that aberrant intracellular localization and misdistribution of cholesterol further exacerbate HC dysfunction.

In summary, our study identifies HSD17B7 as a critical regulator of cholesterol synthesis in sensory hair cells and as an essential factor in normal MET and sound-evoked sensory responses. Additionally, we identified a sporadic nonsense mutation in the HSD17B7 gene in patients with deafness and provided a mechanistic explanation. These findings uncover a previously unrecognized link between cholesterol biosynthesis and HC MET function and provide a molecular framework for exploring cholesterol-targeted therapeutic strategies for hearing loss.

Materials and Methods

Mouse and zebrafish husbandry

C57BL/6J mice were purchased from the Laboratory Animal Center of Nantong University and maintained in a barrier facility under a 12h light/12h dark cycle at 25℃. The day of vaginal plug observation was embryonic day 0.5 (E0.5), and the day of birth was defined as postnatal day 0 (P0). Zebrafish (Danio rerio) were raised and maintained at 28.5°C. Wild- type AB strain and the transgenic line Tg(Brn3c: mGFP) were used in the study described in our previous study [80]. The transgenic line Tg(myo6b: D4H-mCherry) was generated in this work. The embryonic stage is defined as described in the literature [81]. Embryos were collected following natural spawns and cultured in 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 incubated in E3 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

The mouse auditory cell line HEI-OC1 was obtained from Prof. Renjie Chai and 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℃.

HEK293T (sex unknown) cells were obtained from the Shanghai Institute of Biochemistry and Cell Biology (SIBC) and 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. 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

Primary antibodies used in this study included: anti-MYO7A monoclonal antibody (1:150, Developmental Studies Hybridoma Bank, #138-1); anti-HSD17B7 polyclonal antibody (1:150-1:1000, proteintech, #14854-1-AP); anti-Parvalbumin monoclonal antibody (1:500, sigma, #sab4200545); 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 were extracted using TRIzol reagent (Invitrogen, #15596026) and treated with DNase I (Vazyme, #EN401) to remove genomic DNA contamination. RNA concentration was measured using a NanoDrop ND-2000 (Thermo Fisher Scientific, USA), and integrity was verified by agarose gel electrophoresis.

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’-ACCTGGACAA TGGTGACCTC -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

The whole-mount in situ hybridization (WISH) of zebrafish was performed according to the following standard procedures. A 403 bp cDNA fragment of the zebrafish hsd17b7 gene or a 472 bp cDNA fragment of the zebrafish tmc1 gene was amplified via PCR using designed primers (hsd17b7-F: 5’-GACGTCCTCCAGTAATGCCC-3’, hsd17b7-R: 5’-CATCTTGCTTGGT CGGGTGT-3’; tmc1-F: 5’-TTGGGCAGTGATGTGCTGTA-3’, tmc1-R: 5’-GATGCTGTTT CTGCGTTGCT-3’) and was cloned into the pGEM-T-easy vector. After linearization of the pGEM-T-easy vector, inserting the hsd17b7 or tmc1 fragment, the DIG RNA Labeling Kit (SP6) (Roche, #11175025910) was used to prepare digoxigenin-labeled hsd17b7 or tmc1 antisense mRNA probes through transcription in vitro. Subsequently, embryos at different developmental stages were hybridized with an hsd17b7 or tmc1 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 or tmc1 expression. The WISH images of zebrafish were acquired using a stereomicroscope (Olympus, MVX10, Japan).

Plasmids construction

The HSD17B7 and RER1 full-length cDNAs were PCR amplified from the HEK293T cDNA and cloned into the pGEX-TEV and pET-23b vectors, respectively. HSD17B7^E182* was generated by site-directed mutagenesis. For cell transfection, the cDNA of HSD17B7 and HSD17B7^E182* were subcloned into the pCMV-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 pCMV-Flag-HSD17B7 and pCMV-Flag-HSD17B7^E182*, respectively. pCS2-D4H-mCherry was obtained from G. Peng’s lab. For plasmid injection of zebrafish, the myo6b promoter was subcloned into the pDestTol2 vector, followed by D4H-mCherry subcloned into the pDestTol2 vector. The cDNAs encoding the human protein of HSD17B7-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 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 of 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 was added to 150 μL of opti-MEM (Gibco, #31985070), then 5 μL Lipofectamine RANiMAX was added, incubated at room temperature for 5 min, and then added to the cell culture medium. After 48 hours of 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, the collected cells and isolated RNA were used for RT-qPCR [82].

Immunostaining and image acquisition

Cultured cells grown on coverslips, after 36 hours of transfection, were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature and washed 3 times with PBS, then 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 in 4% paraformaldehyde in PBS for 1 hour 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 hour at room temperature, followed by incubation with the anti-primary antibody overnight at 4°C. The samples were then washed in 0.1% PBST and incubated with fluorescent dye-labelled secondary antibodies for 2 hours at room temperature, before mounting in mounting solution containing 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 reconstructed three-dimensional images and contrast adjustments were processed using Imaris X64 (version 9.0.1).

FM4-64 labeling

To investigate the basal activity of HCs, 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 [29, 30, 83, 84]. 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.

Quantification of the relative fluorescence intensity of FM4-64 and D4H

Quantitative analysis of FM4-64 uptake in zebrafish lateral line hair cells and D4H-mCherry fluorescence in HEI-OC1 cells was performed at the single-cell level. For zebrafish experiments, hair cells were identified based on Tg(Brn3c:mGFP). Individual hair cells within each neuromast were manually segmented using ImageJ, and the mean FM4-64 fluorescence intensity per hair cell was measured. Background fluorescence was determined from adjacent cell-free regions and subtracted from each measurement. For each larva, the average FM4-64 intensity was calculated by averaging values from all analyzed hair cells.

For D4H-mCherry cholesterol sensor analysis in HEI-OC1 cells, individual cells were segmented based on cell morphology, and the mean intracellular D4H-mCherry fluorescence intensity per cell was quantified. Background fluorescence was measured from cell-free regions within the same field of view and subtracted to normalize for imaging variability. Fluorescence intensities were then averaged across cells for each condition and used for statistical analysis. All fluorescence measurements were performed on raw images acquired under identical imaging settings across experimental groups.

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 morpholino powder was dissolved in RNase-free water to prepare the working solution at a final concentration of 0.3 mM for subsequent operations. About 2 nL dose of morpholino work solution was microinjected into zebrafish embryos at the one-cell stage. To assess morpholino efficiency, embryos injected with morpholino were collected, and RNA was extracted and reverse-transcribed into cDNA. The designed primers flanking 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 transcribed in vitro. Briefly, the designed primers (zebrafish hsd17b7 mRNA primer, F: 5’-CGCGGATCCATGAAGAAAGTAGTTTTGGT-3’, R: 5’-CCGGAATT CTCACATTCCATTTCTTTCTT-3’; human HSD17B7 mRNA primer, F: 5’-CGGGATCCATGCGAAAGGTGGTTTTGATC-3’, R: 5’-GCTCTAGATTATAGGCATGA GCCACTGA-3’; human HSD17B7^E182* mRNA primer, F: 5’-CGGGATCCATGCGAAAG GTGGTTTTGATC-3’, R: 5’-GCTCTAGACTAGAGGCTGAAATTAGATT-3’) were used to amplify target DNA containing the coding sequence. Subsequently, the pCS2 vector inserted into the amplified fragment was linearized as a template for mRNA transcription using 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 to synthesize gRNA via 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’-TTTCCAT GCAGCACTATCAAACAATT-3’).

Startle Response Test

The acoustic startle reflex was performed as described previously [34, 85]. 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. 60 Hz tone bursts at two sound levels of 9 dB re. 1 m·s ⁻² were applied to the amplifier to drive the vibrator. Acoustic vibration stimuli lasting 30 ms, with an inter-stimulus interval of 180 s, were set and applied. Each sound vibration stimulus level was repeated 20 times, and the locomotion behavior of the larvae with C-shape motion in response to this stimulus was recorded. Finally, the movement’s 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 fusion proteins were immobilized on GSH resins and 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. The 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 the acquired images were analyzed using ImageJ (version 1.8.0).

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

Cell lysates from pCMV-Flag, pCMV-Flag-HSD17B7, and pCMV-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 3 times with lysis buffer, and 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 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 the Venn diagram and GO analysis, microarray analysis identified HSD17B7 and HSD17B7^E182*-specific associated proteins by mass spectrum (peptide ≥ 1, unique peptide ≥ 1, P value < 0.05). Cellular component GO term enrichment analysis was performed using the web-based DAVID software [86]. Ontology networks were further investigated and visualized using R (version 4.4.1).

Public scRNA-seq datasets and expression profiling of hsd17b7

Zebrafish scRNA-seq data were downloaded from NCBI Gene Expression Omnibus (GSE221471) [28]. The basic procedure for single-cell integrated data analysis was performed using Seurat 4.0.1 [87]. Cell clusters corresponding to LLHCs, MHCs, CHCs, supporting cells, and mantle cells were selected based on previously published annotations. Feature plots, violin plots, and average expression analyses were used to examine the expression pattern of hsd17b7 across cell types. To assess dynamic changes in hsd17b7 expression during HC differentiation, trajectory analysis was performed using Monocle 3 (v1.0.1). Cells were ordered along pseudotime, with mantle cells as the root population, and gene-expression changes were visualized along the inferred developmental trajectory.

Mouse scRNA-seq data were downloaded from previous articles (GSE71982, GSE168901, GSE202920) [31–33]. The basic procedure for single-cell integrated data analysis and batch-effect correction was performed using Seurat 4.0.1 [87]. We selected four cell populations for re-analysis, including inner hair cells, outer hair cells, utricle hair cells, and crista hair cells.

Single-cell RNA sequencing of control and hsd17b7 mutant hair cells

To investigate genotype-dependent transcriptional remodeling in sensory hair cells, we performed single-cell RNA sequencing on fluorescence-activated cell sorting (FACS)-isolated hair cells from control and hsd17b7 mutant zebrafish.

Isolation of hair cells and single-cell RNA sequencing

mGFP-positive hair cells were isolated from control and hsd17b7 mutant zebrafish larvae by fluorescence-activated cell sorting (FACS). Dissociated cells were gated to exclude debris and doublets, and viable mGFP⁺ cells were collected for downstream analysis. Single-cell RNA sequencing libraries were generated using the 10× Genomics Chromium platform according to the manufacturer’s instructions. Libraries were sequenced on an Illumina platform to obtain paired-end reads. Raw sequencing data were processed using the Cell Ranger pipeline (10× Genomics) to generate gene expression matrices for each sample. Downstream analyses were performed in R using the Seurat package.

Data preprocessing, integration, and cell type annotation

Cells with low gene counts or high mitochondrial transcript proportions were excluded to remove low-quality cells. Gene expression matrices from control and hsd17b7 mutant samples were normalized and integrated using Harmony to correct for batch effects while preserving biological variation.

Dimensionality reduction was performed using principal component analysis (PCA), followed by Uniform Manifold Approximation and Projection (UMAP) for visualization. Clustering was conducted using a shared nearest neighbor (SNN) graph-based approach. Cell clusters were annotated based on established marker genes for hair cells and supporting cells. Hair cells were identified by robust expression of myo6b, while inner ear and lateral line hair cell subtypes were distinguished by lhfpl5a and lhfpl5b, respectively. Supporting cell populations were annotated using stm as a pan–supporting cell marker, with otogl and irg1l marking inner ear and lateral line supporting cells, respectively. Cell type annotations were validated using both violin plots and Nebulosa density visualizations

Pseudobulk differential expression analysis

To assess genotype-dependent transcriptional changes while accounting for cell-to-cell variability, pseudobulk differential expression analysis was performed for hair cells. Gene expression counts from individual cells were aggregated by genotype to generate pseudobulk profiles. Differential expression between control and hsd17b7 mutant HCs was assessed using a generalized linear model framework. Genes with an adjusted p-value < 0.05 were considered differentially expressed. Results were visualized using volcano plots, with upregulated and downregulated genes highlighted accordingly.

Gene Ontology and pathway enrichment analysis

Differentially expressed genes were subjected to Gene Ontology (GO) biological process enrichment analysis using curated annotation databases [88]. Enrichment analyses were performed separately for upregulated and downregulated gene sets. Significantly enriched GO terms were visualized, and representative genes contributing to each term were illustrated using Sankey diagrams to highlight functional relationships between gene sets and biological processes.

Gene set enrichment analysis (GSEA) was additionally performed to assess coordinated changes in cholesterol biosynthesis–associated pathways at the transcriptome-wide level. Ranked gene lists derived from pseudobulk differential expression results were used as input, and enrichment significance was assessed using normalized enrichment scores and adjusted p-values.

Module score analysis of MET- and cholesterol-related gene sets

Module scores were computed using the AddModuleScore function in Seurat, based on normalized expression values from the RNA assay. Module score distributions were examined separately in lateral line hair cells (LL_HCs) and inner ear hair cells (IE_HCs). Differences between control and hsd17b7 mutant samples were assessed using the Wilcoxon rank-sum test. Results were visualized using heatmaps, violin plots, and ridge plots. Ridge plots were generated using z-scored module scores to illustrate population-wide shifts in transcriptional states rather than the emergence of discrete subpopulations.

Visualization and statistical analysis

UMAP feature plots, heatmaps, violin plots, and ridge plots were generated using Seurat and associated R visualization packages. Nebulosa density plots were used to visualize spatial expression patterns of marker genes across low-dimensional embeddings. Statistical analyses were performed using R, and exact p values or significance levels are reported in the corresponding figure panels and legends.

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 5A).

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 the Illumina HiSeq 4000 sequencing platform (Illumina, USA). Bioinformatics were aligned to the 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 the 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 databases (COSMIC, ClinVar, OMIM, GWAS). Genotypes distributed 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 presented as mean ± SD or mean ± SEM. 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 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.

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. Sequencing data have been deposited in GEO under accession codes GSE319132.

Supplementary Figures

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Figure S7.

Figure S8.

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

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.

Funding

National Natural Science Foundation of China (32200783)

• Yuqian Shen

National Natural Science Foundation of China (92368104)

• Dong Liu

National Natural Science Foundation of China (32350017)

• Dong Liu

Natural Science Foundation of Jiangsu Province Grants (BK20220607)

• Yuqian Shen