SB1190-1923 began to experience more severe progressive SNHL in the high-frequency range starting in their 50 s. Ten years after the onset of significant SNHL, SB1190-1923 had developed profound deafness with a speech discrimination score of only 8% (Fig. 1). ES identified a potentially causative, novel variant of HOMER2 (c.1033 del:p.R345Efs*64) in a single heterozygous state, and the variant status was confirmed with Sanger sequencing.

Pedigree, audiometric profile, genetic analysis, and post-cochlear implantation speech outcomes from SB1190-1923 (F/61). a Pedigree of SB1190: the proband (F/61) carries the previously unreported, novel HOMER2 variant (c.1033 del (p.Arg345Glufs*64)) which is associated with DFNA68. Low pLI score (0.01) of HOMER2 suggests that alteration of HOMER2 exerts a pathogenic effect via a mechanism other than loss-of-function. Both parents of SB1190-1923 passed away in their early 70 s without showing any signs of significant hearing loss, and none of the SB1190-1923’s children exhibited any symptoms of hearing loss. b Preoperative pure tone audiograms show profound down-sloping sensorineural hearing loss with 8% speech discrimination score bilaterally. c Post-Cochlear Implantation speech outcome speech recognition improves significantly at 3- and 6-months post-implantation across various syllable and sentence conditions. F, female; pLI, intolerant; CI, cochlear implantation

The patient’s SNHL phenotype was matched perfectly with previous reports in the literature related to HOMER2 [1, 3, 5, 20, 21] (Table 1 and Fig. 1). This variant, which we categorize as a “frameshift extension variant,” features a cytosine deletion at Chr15:83,518,499, leading to a frameshift that alters R345 to E(Glu) and results in the addition of 63 amino acids after R344.

Based on the previously reported sixth variant [4], specifically the nonstop extension variant identified as pathogenic, we were reasonably confident that our variant, which results in a longer extension, would also be pathogenic. Therefore, we aimed to determine whether the pathogenic effect of our variant was primarily driven by the alteration of the 10 amino acids at the C-terminus or by the extension of 53 additional amino acids beyond the stop codon. To investigate this, we conducted molecular modeling and functional studies in zebrafish, comparing a hypothetical variant where the C-terminal 10 amino acids were truncated (HOMER2 p.R345*) with our actual frameshift extension variant (HOMER2 p.R345Efs*64).

The HOMER2 variant p.R345Efs*64 causes a frameshift extension affecting the C-terminal coiled-coil (CC) domain. The CC domain in the C-terminal region of HOMER2, near which p.R345 is located, is known as essential for interactions between Homer proteins [22]. In contrast, the EVH1 (Ena/Vasp Homology domain 1) domain, located relatively closer to the N-terminus of HOMER2, plays a crucial role in mediating interactions with other proteins, including the IP3 receptor and C/EBPβ [23, 24]. Structural predictions using AlphaFold2 suggest that the p.R345Efs*64 variant causes structural changes in the EVH1 domain (Fig. 2a), rather than directly affecting the CC domain. When predicting dimer structures with HOMER1 and HOMER2, the position of the EVH1 domain in the two variants (p.R345* and p.R345Efs*64) HOMER2 proteins was found to be significantly different compared to the WT protein (Fig. 2b). Upon observing the dimer structures from the side, it was noted that the EVH1 domains were more vertically shifted in the dimer with HOMER2 p.R345Efs*64 than with WT and p.R345* variant (Fig. 2c).

Predicted 3D structures of HOMER2 WT and HOMER2 p.R345Efs*64 using AlphaFold2. a Structural alignment of HOMER2 WT and HOMER2 p.R345Efs*64. The EVH1 domain is highlighted in the yellow area, and the coiled-coil (CC) domain is marked by the blue dashed line. a1 The hydrogen bond between N43 and A113 is disrupted in HOMER2 p.R345Efs*64 compared to HOMER2 WT. a2 The β-sheet structure in HOMER2 p.R345Efs*64 extends to F90, and a hydrogen bond forms between F74 and S71, which is absent in HOMER2 WT. b The structure of the HOMER1 [NP_004263.1] and HOMER2 dimer is shown. HOMER1 is depicted in gray, while HOMER2 WT, HOMER2 p.R345Efs*64, and HOMER2 p.R345* are represented in green, brown, and purple, respectively. The coiled-coil region of HOMER2, which can interact with other proteins, is located at amino acids 307–329 in the WT and is shifted to amino acids 275–297 in the HOMER2 variants. c Structural changes in the EVH1 domain were detected in the HOMER1 WT and HOMER2 variants dimer. The distance between each EVH1 domain of HOMER1 and HOMER2 p.R345Efs*64 was reduced (b) compared to the WT, which is shifted vertically. In HOMER1 WT and HOMER2 p.R345* dimer, the distance between each EVH1 was increased (b) and slightly shifted vertically. d The predicted tetramer structures were obtained using HOMER1 and HOMER2. Two HOMER1 proteins were combined with two HOMER2 WT, two HOMER2 p.R345Efs*64, or two HOMER2 p.R345* molecules. In the HOMER1 and HOMER2 p.R345Efs*64 tetramer, the location of the EVH1 domain of HOMER1 was altered, and the C-terminal of HOMER2 approached the EVH1 domain of HOMER2 (dashed black box). In the HOMER1 and HOMER2 p.R345* tetramer, HOMER1 and HOMER2 formed homodimers, and each homodimer formed a tetramer, unlike the other configurations. WT, wild-type; R, arginine; E, glutamate; fs, frameshift; EVH1, Ena/Vasp homology domain 1; F, phenylalanine; S, serine; N, asparagine; A, alanine

As previously reported, HOMER proteins form tetramers [25]. Similarly, tetramer structures were predicted using HOMER1. Interestingly, HOMER2 WT, HOMER2 p.R345* and HOMER2 p.R345Efs*64 formed dimers with HOMER1, and two of these dimers formed a tetramer unlike with HOMER2 p.R345* (Fig. 2d). Notably, a significantly interfered EVH1 domain was detected in the HOMER2 p.R345Efs*64-containing tetramer (Fig. 2d, black dashed box).

Through protein prediction modeling, it was predicted that both p.R345Efs*64 and p.R345* have pathogenic effects. Therefore, we decided to test both variants in a zebrafish model. Following the injection of HOMER2 mRNA of WT and both variants (hypothetical and patient-derived) into zebrafish embryos, larvae were observed to have heart malformations and overall morphological defects at 3 dpf. (Fig. 3a). The distribution of four categories was analyzed across five groups including HOMER2 variants. The percentage of larvae exhibiting heart malformations and abnormal morphology increased progressively in the HOMER2 variants group. Specifically, the proportion of malformations in both HOMER2 variant groups was significantly increased compared to the RFP control group (1.63-fold in p.R345* and 1.78-fold in p.R345Efs*64) (Fig. 3b), indicating a stronger effect of these variants on cardiac development.

Cardiac and morphological defects in zebrafish larvae injected with HOMER2 mRNA. a Representative images of zebrafish larvae at 3 dpf displaying varying degrees of heart malformations. The larvae were categorized into four groups: normal, mild, and severe heart defects based on the degree of cardiac enlargement, and an additional group with general abnormal morphology. The categories include normal, mild, and severe heart defects, with red arrowheads marking the regions of cardiac deformity. Additionally, larvae with abnormal overall morphology are shown. b Bar graph illustrating the distribution of larvae with heart and morphological abnormalities in the different experimental groups: uninjected control, RFP control, HOMER2 WT, HOMER2 p.R345*, and HOMER2 p.R345Efs*64. Chi-squared analysis was used to compare the proportions of defects between groups, with significant differences indicated. (****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05). n = 100 per group. dpf, days post-fertilization; RFP, red fluorescent protein; WT, wild-type; R, arginine; E, glutamate; fs, frameshift

To assess the potential effects of HOMER2 mRNA variants on auditory system development, the otic capsule area was measured (Fig. 4a). Quantification of the otic capsule area across five groups and its statistical analysis indicated no significant impact of HOMER2 variants on the size of the otic structure (Fig. 4b).

Otic capsule area in zebrafish larvae at 3 days post-fertilization. a Representative image of 3 dpf zebrafish larvae, with the otic capsule indicated by a red dashed circle. b Bar graph showing the measured otic capsule area in control, RFP control, HOMER2 WT, HOMER2 p.R345*, and HOMER2 p.R345Efs*64 groups. No statistically significant differences were observed between groups (p > 0.05). n = 15 per group. dpf, days post-fertilization; RFP, red fluorescent protein; WT, wild-type; R, arginine; E, glutamate; fs, frameshift

The number of hair cells in neuromasts of zebrafish larvae was analyzed to investigate the potential impact of HOMER2 variants on neuromast hair cell development. Hair cells from four neuromasts (supraorbital (SO1 and SO2), otic (O1), and occipital (OC1)) were examined, and the total number of hair cells was compared between groups (Fig. 5a). The statistical analysis from the four neuromasts revealed no significant differences in hair cell numbers between the groups (Fig. 5b). These results indicate that HOMER2 variants did not affect the development of the otic capsule and the number of neuromast hair cells.

Analysis of hair cell numbers in zebrafish neuromasts. a Representative image of the four neuromasts (SO1, SO2, O1, and OC1) from each group: control, RFP control, HOMER2 WT, HOMER2 p.R345*, and HOMER2 p.R345Efs*64. b Bar graph comparing each group's average number of hair cells across the four neuromasts. No statistically significant differences were observed between the groups (p > 0.05). n = 10 per group. SO1, supraorbital1; SO2, supraorbital2; O1, otic; OC1, occipital; RFP, red fluorescent protein; WT, wild-type; R, arginine; E, glutamate; fs, frameshift

To evaluate the impact of HOMER2 variants on neuromast hair cell function, FM1-43 uptake was assessed. In the control, RFP, and HOMER2 WT groups, robust FM1-43 uptake was observed within the neuromast hair cells. However, in the HOMER2 variant groups, FM1-43 uptake was markedly reduced (Fig. 6a). In the quantification assay, while no significant differences were detected in the mCherry fluorescence between the control, RFP, and HOMER2 WT, the HOMER2 variants exhibited a significant reduction in intensity compared to the RFP group (0.41-fold in p.R345* and 0.39-fold in p.R345Efs*64) (Fig. 6b). The data suggest that HOMER2 variants severely impaired the capacity of neuromast hair cells to uptake FM1-43, indicating disrupted hair cell function in variant larvae.

Comparison of FM1-43 uptake in zebrafish neuromast hair cells. a Representative image of neuromast hair cells in control, RFP control, HOMER2 WT, and HOMER2 variant groups. Hair cells are labeled with GFP, and FM1-43 uptake is indicated by mCherry fluorescence. b Bar graph comparing the fluorescence intensity of FM1-43 uptake across groups. While no significant differences were observed between the control, RFP, and HOMER2 WT groups, the HOMER2 variant groups showed a statistically significant reduction in intensity compared to the RFP group (*p < 0.05, **p < 0.01). n = 5 per group. RFP, red fluorescent protein; WT, wild-type

In the light–dark behavior test, the expression of HOMER2 variants did not alter the behavioral response of zebrafish larvae (Supplementary Fig. 2). For another behavioral test at 6 dpf, latency increased progressively in the HOMER2 p.R345Efs*64 group (Fig. 7a). In terms of the distance moved during the startle reflex, the HOMER2 p.R345Efs*64 group exhibited a significantly lower distance moved compared to both the HOMER2 WT and p.R345* groups (Fig. 7b). Additionally, the distance moved in the HOMER2 p.R345Efs*64 group was also significantly lower than that in the RFP control group, further underscoring the severe impact of the p.R345Efs*64 variant. The increased response time coupled with reduced movement suggests a functional impairment in neuromotor coordination in the HOMER2 variant larvae. Especially, p.R345Efs*64 variant showed greater functional deficits compared to the p.R345* variant, highlighting the variant’s stronger impact on neuromotor coordination and reflex amplitude.

Comparison of startle reflex in zebrafish larvae at 6 days post-fertilization. a Bar graph depicting the latency (time from stimulus application to the onset of movement) across the control, RFP control, HOMER2 WT, HOMER2 p.R345*, and HOMER2 p.R345Efs*64 groups. The latency was significantly longer in the HOMER2 p.R345Efs*64 group compared to HOMER2 WT (p < 0.05). b Bar graph showing the distance moved during the startle response. The HOMER2 p.R345Efs*64 groups exhibited a significant reduction in distance moved compared to HOMER2 WT and p.R345*, and showed a further reduction relative to the RFP control group (p < 0.05). n = 20 per group. RFP, red fluorescent protein; WT, wild-type; R, arginine; E, glutamate; fs, frameshift