Insights into phenotypic differences between humans and mice with p.T721M and other C-terminal variants of the SLC26A4 gene

Recessive variants of the SLC26A4 gene are an important cause of hereditary hearing impairment. Several transgenic mice with different Slc26a4 variants have been generated. However, none have recapitulated the auditory phenotypes in humans. Of the SLC26A4 variants identified thus far, the p.T721M variant is of interest, as it appears to confer a more severe pathogenicity than most of the other missense variants, but milder pathogenicity than non-sense and frameshift variants. Using a genotype-driven approach, we established a knock-in mouse model homozygous for p.T721M. To verify the pathogenicity of p.T721M, we generated mice with compound heterozygous variants by intercrossing Slc26a4+/T721M mice with Slc26a4919-2A>G/919-2A>G mice, which segregated the c.919-2A > G variant with abolished Slc26a4 function. We then performed serial audiological assessments, vestibular evaluations, and inner ear morphological studies. Surprisingly, both Slc26a4T721M/T721M and Slc26a4919-2A>G/T721M showed normal audiovestibular functions and inner ear morphology, indicating that p.T721M is non-pathogenic in mice and a single p.T721M allele is sufficient to maintain normal inner ear physiology. The evidence together with previous reports on mouse models with Slc26a4 p.C565Y and p.H723R variants, support our speculation that the absence of audiovestibular phenotypes in these mouse models could be attributed to different protein structures at the C-terminus of human and mouse pendrin.

Parallel to the transgenic mouse models, studies on cell lines have provided crucial insights into the pathogenicity of SLC26A4 variants. It has been reported that different SLC26A4 variants may result in different degrees of protein misexpression and/or dysfunction [15][16][17] . Among the SLC26A4 variants, whose pathogenicity has been investigated in cell lines, the p.T721M variant is especially interesting. The expression of pendrin in cell lines with p.T721M could not be rescued after salicylate treatment, but that in cell lines with p.H723R could, indicating that p.T721M might confer a stronger pathogenicity than p.H723R 15 . Based on these results, we hypothesized that mice with the Slc26a4 p.T721M variant might exhibit auditory phenotypes milder than those of c.919-2A > G mice but more severe than those of p.H723R mice. In this study, we generated a knock-in mouse model with the p.T721M variant, as well as compound heterozygous (Slc26a4 919-2A>G/T721M ) mice in which we tried to manipulate the severity of phenotypes by abolishing the other functional Slc26a4 allele 9 . We then characterized the audiovestibular phenotypes and inner ear pathology in these mouse models (Fig. 1).
We then examined the expression of pendrin in the cochlea of Slc26a4 T721M/T721M and Slc26a4 919-2A>G/T721M mice by immunolocalization (Fig. 3F). In both strains, pendrin was normally distributed in the spiral prominence and root cells, indicating that the expression of pendrin was normal in the p.T721M mice. In addition, the contour of the stria vascularis (indicated by white dashed lines in the figure) showed atrophic changes in Slc26a4 919-2A >G/919-2A>G mice, but not in Slc26a4 T721M/T721M and Slc26a4 919-2A>G/T721M mice.

Discussion
The knock-in mouse generated in this study, the Slc26a4 T721M/T721M with the Slc26a4 p.T721M variant, demonstrated normal audiovestibular phenotypes and inner ear morphologies, similar to wild-type mice. To investigate whether the p.T721M variant could contribute to SNHI in mice through the haplo-insufficiency mode, we further generated mice with compound heterozygous variants (Slc26a4 919-2A>G/T721M ) by intercrossing Slc26a4 +/T721M mice with Slc26a4 919-2A>G/919-2A>G mice, which segregated the c.919-2A > G variant with abolished function. Compound heterozygous mice for p.T721M and c.919-2A > G (i.e., Slc26a4 919-2A>G/T721M mice) also had normal audiovestibular phenotypes, indicating that a single p.T721M allele was sufficient to maintain normal inner ear physiology in the mice.
The SLC26A4 p.T721M variant has been documented sporadically in hearing-impaired families worldwide, including two Mediterranean families 18 , two Iranian families 19 , two Japanese family 20,21 , three Chinese families 22,23 , and one Taiwanese family 24 . Although this variant is widely distributed across different populations, its prevalence is relatively low compared to other SLC26A4 variants, such as c.919-2A > G or p.H723R. According     20 . These typical clinical manifestations in p.T721M homozygotes and compound heterozygotes suggest that p.T721M is a pathogenic SLC26A4 variant in humans.
In previous cell line studies, a number of SLC26A4 variants, including p.P123S, p.M147V, p.L236P, p.S657N, p.T721M, and p.H723R, have been demonstrated to confer pathogenicity by affecting the trafficking process, rather than the expression level of pendrin 15,17 . However, the affected trafficking process can be rescued by certain treatments 15,17 . Low-temperature incubation has been reported to rescue the trafficking of pendrin with p.H723R, which originally accumulated in the endoplasmic reticulum, but not the trafficking of pendrin with p.L236P, which originally accumulated in centrosomal regions 17 . Similarly, salicylate treatment could rescue the trafficking of pendrin with p.P123S, p.M147V, p.S657N, and p.H723R, and restore the function of pendrin as an anion exchanger, but not that of pendrin with p.T721M 15 . These lines of evidence also support the view that the SLC26A4 p.T721M variant is pathogenic and implies that p.T721M is more pathogenic compared to other missense SLC26A4 variants, such as p.H723R.
To our surprise, the pathogenicity of p.T721M as predicted by the ACMG guidelines and evidenced by the cell line studies was not observed in our mouse model with the Slc26a4 p.T721M variant. These findings are consistent with our previous studies in mouse models with Slc26a4 p.H723R 10 and p.C565Y 12 variants. In the cochlea of mice, pendrin is expressed at the spiral prominence and outer sulcus cells 27 , which is almost the same as its expression in the cochlea of primates 28 . Therefore, the position of pendrin expression does not appear to be a major factor contributing to the phenotypic discrepancy between species.
Alternatively, the inter-species phenotypic discrepancy may be attributed to the structural differences between mouse and human pendrin.
To date, five mouse models with missense Slc26a4 variants, including p.L236P 11 , p.S408F 8 , p.C565Y 12 , p.T721M (this study), and p.H723R 10 , have been documented in the literature (Fig. 5A). Abnormal audiovestibular phenotypes were observed in mice with p.L236P and p.S408F, but not in mice with p.C565Y, p.T721M, and p.H723R. Notably, p.L236P and p.S408F are located in the transmembrane domains of pendrin 27,28 . In contrast, p.C565Y, p.T721M, and p.H723R are located in the C-terminus of pendrin comprised of amino acid residues 508-780 27,28 (Fig. 5B). From an evolutionary perspective, the amino acid sequence of the pendrin C-terminus is less conserved, sharing only 86% identity between mice and humans. In contrast, the amino acid sequence of the transmembrane domains shared 92% identity between the two species (https:// www. expasy. org/&https:// www.nature.com/scientificreports/ www. unipr ot. org/). Accordingly, we speculate that the absence of phenotypes in mice with Slc26a4 p.C565Y, p.T721M, and p.H723R variants could be attributed to different protein structures at the C-terminus of human and mouse pendrin. Another line of evidence that supports our speculation is the hH723R Tg mouse model 14 . Instead of creating the variant on murine genomic DNA, the authors delivered a sequence of human cDNA harboring the p.H723R variant into pronuclear-stage mouse embryos. This humanized transgenic mouse model revealed profound SNHI 10 , suggesting that C-terminus SLC26A4 variants identified in humans might require a human peptide backbone to exhibit their pathogenicity.
Rapp et al. analyzed the locations of SLC26A4 mutations and identified a high density of SLC26A4 mutations in the anion-binding transmembrane domains (TMs), including TM1, 3, 8, and 10 29 . The findings indicate that when present in these transmembrane domains, SLC26A4 variants are more likely to be pathogenic. This could possibly explain the higher evolutionary conservation of amino acid sequences between human and mouse pendrin at the transmembrane domains, as well as the clear phenotypic manifestation of profound SNHI in mice with Slc26a4 variants at the transmembrane domains, such as Slc26a4 L236P/L236P and Slc26a4 loop/loop mice (i.e., mice with p.S408F).
In summary, using a genotype-driven approach, we generated a knock-in mouse model segregating the deafness-associated SLC26A4 p.T721M variant in humans. Surprisingly, mice with the Slc26a4 p.T721M variant exhibited a normal audiovestibular phenotype and inner ear morphology. Because there might be differences in the pathogenicity of specific SLC26A4 variants between humans and mice, caution is needed while extrapolating the results of animal studies to humans. Construction of Slc26a4 T721M/T721M knock-in mice. Transgenic mice were generated by the Transgenic Mouse Models Core (TMMC, Taiwan) using the CRISPR technology-associated RNA-guided endonuclease Cas9 to mutate the Slc26a4 gene and generate the Slc26a4 +/T721M mouse line. Specific guiding RNAs (sgRNAs) were developed to target exon 15 of the Slc26a4 gene in C57BL/6 mice. The sgRNA and CRISPR/Cas9 RNA were delivered into the mouse zygote to generate founders. The two male founder mice that were obtained from each, harbored the p.T721M (c.2162C > T) variant in the Slc26a4 gene. After germline transmission of the targeted variant allele, we produced the congenic Slc26a4 +/T721M mouse line used in this study by repeated backcrossing into the C57BL/6 inbred strain for 6-10 generations. Mice homozygous for the variant (Slc26a4 T721M/T721M ) were obtained by intercrossing heterozygous mice (i.e., Slc26a4 +/T721M ) (Fig. 1). Reverse transcription-PCR (RT-PCR) of mRNA of inner ear extract and direct sequencing confirmed a pure non-chimeric genetic background without unintentional wild-type Slc26a4 expression in Slc26a4 T721M/T721M mice.

Auditory evaluations.
Mice were anesthetized and placed in a head-holder within an acoustically and electrically insulated and grounded test room 10 . We used an evoked potential detection system (Smart EP 3.90; Intelligent Hearing Systems, Miami, FL, USA) to measure auditory brainstem response (ABR) thresholds in mice. Click sounds, as well as 8, 16, and 32 kHz tone bursts at various intensities were generated to evoke ABRs. Response signals were detected using subcutaneous needle electrodes. The active electrodes were inserted into the vertex and the ipsilateral retroauricular region with a ground electrode on the back of each mouse. The ABRs were recorded from postnatal 12-44 weeks, to trace changes in auditory function. ABRs were measured bilaterally. As there were no significant differences between the two ears, only the data of the left ears were presented and analyzed.
Inner ear morphology studies. For light microscopy studies, tissues were stained with hematoxylin and eosin (H&E). The morphology of each sample was examined using a Leica optical microscope 10 . First, inner ear tissues from adult mice (P28-P30) were fixed by perilymphatic perfusion with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) through round and oval windows, and a small fenestra in the apex of the cochlear bony capsule. The specimens were decalcified for one week. The samples were then dehydrated and embedded in paraffin. Subsequently, serial Sects. (7 mm) were stained with H&E. ImageJ software (NIH, http:// imagej. nih. gov/ ij/ downl oad. html) was used to quantify the cross-sectional area of the middle turn scala media.

Vestibular function evaluation.
Mice were subjected to a series of tests at 8 weeks, including the swimming and rotarod tests. For the swimming test, the swimming performance of the mice was scored from 0 to 3, with 0 representing normal swimming and 3 representing underwater tumbling 31 . For the rotarod tests, the mice were placed on the rotating rod for a maximum of 180 s. The speed of the rods was accelerated from 5 rpm to a maximum speed of 20 rpm in one min. The length of time each mouse remained on the rotating rod was recorded 32 .
Statistical analyses. Data are presented as the mean ± standard deviation. Statistical analyses were conducted using an unpaired Student's t-test with Bonferroni correction for continuous variables. Statistical significance was set at p < 0.05. All analyses were performed using SPSS software (version 15.0; SPSS Inc., Chicago, IL, USA).