Pax3 deficiency diminishes melanocytes in the developing mouse cochlea

Abstract Cochlear melanocytes are intermediate cells in the stria vascularis that generate endocochlear potentials required for auditory function. Human PAX3 mutations cause Waardenburg syndrome and abnormalities of melanocytes, manifested as congenital hearing loss and hypopigmentation of skin, hair and eyes. However, the underlying mechanism of hearing loss remains unclear. During development, cochlear melanocytes in the stria vascularis are dually derived from Pax3-Cre+ melanoblasts migrating from neuroepithelial cells including neural crest cells and Plp1+ Schwann cell precursors originated from also neural crest cells, differentiating in a basal-apical manner. Here, using a Pax3-Cre mouse line, we found that Pax3 deficiency causes foreshortened cochlea, malformed vestibular apparatus, and neural tube defects. Lineage tracing and in situ hybridization show that Pax3-Cre derivatives contribute to S100+ , Kir4.1+ and Dct+ melanocytes (intermediate cells) in the developing stria vascularis, all significantly diminished in Pax3 mutant animals. Taken together, these results suggest that Pax3 is required for the development of neural crest cell-derived cochlear melanocytes, whose absence may contribute to congenital hearing loss of Waardenburg syndrome in human.


Introduction
Pax3 regulates induction and differentiation from neural crest cells which are critical in the development of many organs such as eye, hair, skin, neural tube, cranium, face and inner ear [1][2][3][4] . During embryonic development, Pax3-Cre + neuroepithelial cells including neural crest cells migrate into the otic vesicle, which is primordium of the inner ear, and give rise to melanocytes and glial cells in the developing cochlea [5][6][7][8] .
Human PAX3 gene mutations cause Waardenburg syndrome which is characterized by telecanthus (widely spaced eyes), heterochromia iridis, patchy pigmentation of hair and skin, and profound sensorineural hearing loss [9][10][11] . It has been postulated that congenital hearing loss in Waardenburg syndrome results from developmental defects of cochlear melanocytes, which are located as In this study, we examined Pax3 Cre/Cre mice which models human Waardenburg syndrome with complete loss of Pax3 protein 18 and found that loss of Pax3 prevents formation of melanocytes (intermediate cells) in the developing cochlea. At E18.5, Pax3 knockout cochleae showed 2 major phenotypes: 57% animals showed smaller cochleae with normal organization of the organ of Corti, whereas 43% showed severely shortened cochlea in addition to gross organ dysgenesis such as exencephaly. Furthermore, with fatemapping and immunostaining for cochlear melanocytes, we uncovered that S100 + Pax3 Cre

Results
Auditory function and cytoarchitecture of the Pax3 Cre/+ mice To determine the cell fate of Pax3-Cre derivatives in the cochlea, we used a lineage tracing approach (Pax3-Cre; CAG-CAT-EGFP or Rosa-mTmG) where Pax3-Cre derived cells are labeled with EGFP. This approach has been used previously to fate-map derivatives of Pax3-Cre + neuroepithelial cells including neural crest cells in various organs including the cochlea 5,6,18,19 . We rst con rmed that Pax3 Cre -EGFP + cells were distributed primarily in the stria vascularis and modiolar regions, while occasionally were detected in the organ of Corti and greater epithelial ridge (GER) region in the E18.5 and postnatal day 1 (P1) Pax3 Cre/+ heterozygous cochleae ( Fig. 1A-B). We also con rmed that Pax3 Cre/+ heterozygous embryos had normal cochlear development ( Fig. 1A-B), consistent with the previous reports 5,6 . We further assessed auditory responses of adult P56 Pax3 Cre/+ heterozygous mice and found that they had auditory brain responses comparable to wildtype littermates at all frequencies tested (8, 16, 32 kHz) (Fig. 1C).
Like wildtype control cochlea, the P43 Pax3 Cre/+ heterozygous cochlea displayed normal cytoarchitecture of the stria vascularis, organ of Corti and spiral ganglion ( Fig. 1D-E). These data indicate that Pax3 Cre/+ heterozygous cochlea serves as a normal control for Pax3 Cre/Cre homozygous cochlea and also as an excellent model to examine the Pax3-Cre derived cellular populations in the cochlea.
Pax3 knockout embryos show cochlear and vestibular defects in the late embryonic period During normal development, the otic vesicle rapidly expands and gives rise to the cochlear duct and vestibular apparatus shortly after E10.5 [20][21][22] . Pax3 Cre/Cre homozygous embryos display several developmental anomalies including loop tail and spina bi da with complete loss of Pax3 protein 18 . The degrees of anomaly are classi ed as mild or severe, with the latter showing exencephaly and shortened cochlea at E12.5 and E14.5, and then Pax3 Cre/Cre homozygous embryos also die by P0 5,6,18 . Thus, we characterized the morphology of embryos shortly before birth at E18.5 (27 wildtype, 37 Pax3 Cre/+ heterozygous, 14 Pax3 Cre/Cre homozygous embryos from 7 litters). We found that 8 Pax3 Cre/Cre homozygous embryos displayed mild anomalies (mostly loop tail and spina bi da) (Supplementary Fig.   S1A-C, Table S1), and 6 were severe (visibly smaller, displayed exencephaly, loop tail, and spina bi da) ( Supplementary Fig. S1D, Table S1).
To further examine the morphology of the inner ear of Pax3 Cre/Cre homozygous mice, we performed paintll of cochleae from E15.5 Pax3 Cre/Cre homozygous embryos using wildtype as controls. The inner ear of E15.5 Pax3 Cre/Cre homozygous mice with a mild phenotype was smaller but exhibited similar morphology to wildtype control ( Fig. 2A-B). However, Pax3 Cre/Cre homozygous embryo with a severe phenotype showed several malformations of both vestibular and cochlear organs, including underdeveloped semicircular canals, vestigial endolymphatic duct, and foreshortened cochlear duct (Fig. 2C). Similarly, at E18.5, Pax3 Cre/Cre homozygous inner ear with severe phenotype was noticeably smaller than wildtype and Pax3 Cre/+ heterozygous ( Fig. 2D-G). These results indicate that development of the inner ear is grossly normal in Pax3 Cre/Cre homozygous embryo with mild phenotype but is severely perturbed in Pax3 Cre/Cre homozygous embryo with severe phenotype. The mature stria vascularis has three cell types: marginal cells, basal cells and intermediate cells, the latter of which are melanocytes. By contrast, the embryonic stria vascularis is composed of only marginal cells and intermediate cells 23 . Previously, Pax3 Cre/Cre homozygous cochlea has been shown to display a complete loss of Pax3-Cre derivatives and Dct + cochlear melanocytes in the stria vascularis at E15.5 6 . As melanocytes originate from both melanoblasts and Schwann cell precursors 8 , we hypothesize that Pax3 de ciency leads to a partial, and not complete, loss of cochlear melanocytes at the late embryonic period. First, we analyzed the distribution of the Pax3-Cre derivatives in the stria vascularis of the E18.5 Pax3 Cre/Cre homozygous cochlea. Kcnq1 is a marker for the marginal cells, and S100 marks both the marginal cell and intermediate cells in the stria vascularis 23,24 . In the Pax3 Cre/+ heterozygous cochlea, many S100 + Pax3 Cre -EGFP + intermediate cells were detected next to Kcnq1 + S100 + marginal cells in the stria vascularis of all turns (Fig. 3A). By contrast, in the E18.5 Pax3 Cre/Cre homozygous cochlea (mild phenotype), few or no S100 + Pax3 Cre -EGFP + intermediate cells were detected in the stria vascularis of the apical (Fig. 3B") and middle turns (Fig. 3B', B''''), while some S100 + Pax3 Cre -EGFP + intermediate cells appeared in the basal turn ( Fig. 3B''').
Next, we quanti ed intermediate cells and compared the Pax3 Cre/Cre homozygous (mild phenotype) cochlea with that of the Pax3 Cre/+ heterozygous or wildtype cochlea at E18.5. All of them displayed four cochlear regions (apex, mid-apex, mid-base and base) in cross-section (Fig. 1A, 3A-B, 4A-B, 5A-B, Supplementary Fig. S2A). In each cochlear turn, there were noticeably fewer S100 + Pax3 Cre -EGFP + intermediate cells in the stria vascularis of Pax3 Cre/Cre homozygous embryos with mild phenotype than Pax3 Cre/+ heterozygous embryos. The reduction is most dramatic in the apical turn relative to the base (Fig. 3C). These data suggest that loss of Pax3 prevents normal development of intermediate cells in the developing cochlea, with the apical and middle cochlear turns more severely affected than the base.
Characterizing melanocytes in the stria vascularis in the Pax3 knockout embryos To further characterize whether cochlear melanocytes were perturbed in Pax3 Cre/Cre homozygous mouse, we performed in situ hybridization for markers of melanocytes. Dct, a classical marker of melanocytes, was detected in all three turns in the E18.5 wildtype cochlea (Fig. 4A). In the Pax3 Cre/Cre homozygous cochlea (mild phenotype), we discovered markedly fewer Dct + melanocytes in all turns, with the greatest reduction observed in the apical turn (Fig. 4B, C).
Moreover, we examined expression of the inwardly rectifying potassium channel Kir4.1, whose expression in the stria vascularis is crucial for development of the endocochlear potential after P7 12,13,25 . Because Pax3 Cre/Cre homozygous mice are lethal perinatally, we investigated Kir4.1 mRNA expressions in the E18.5 cochleae 18 . Kir4.1 mRNA expression was detected in the stria vascularis of both control and the Pax3 Cre/Cre homozygous mild phenotype cochlea. Kir4.1 mRNA expression was also detected in the organ of Corti and spiral ganglion ( Fig. 5A-B). In the Pax3 Cre/Cre homozygous embryos with mild phenotype, there were signi cantly fewer Kir4.1 + cells in the stria vascularis in the mid-basal and basal turns than those in Pax3 Cre/+ heterozygous embryos, although the apical and mid-apical turns showed no signi cant reduction in the number of Kir4.1 + cells between those groups (Fig. 5C). Together, these data reveal that Pax3 de ciency perturbs development of cochlear melanocytes.

Discussion
Waardenburg syndrome is characterized by hearing loss and developmental abnormalities of melanocytes 11,26 . Genetic studies suggest that Waardenburg syndrome is caused by mutations of PAX3 and other genes such as MITF, SOX10, EDN3, EDNRB and SNAI2 1,9,27−30 . Here, we used a mouse model of Pax3 de ciency to study Waardenburg syndrome and found that loss of Pax3 causes a reduction of melanocytes in the developing cochlea, possibly stemming from a disruption to the distribution of neuroepithelial cells including neural crest cells. We showed that Pax3 Cre/Cre homozygous cochlea was short and vestibular system was malformed. Moreover, we revealed that loss of Pax3 decreased the number of melanocytes (intermediate cells) in the stria vascularis.  13,36 . Endocochlear potential inturn drives depolarization of hair cells and is required for hearing function 37 . It has been thought that the developmental disorder of cochlear melanocytes causes Waardenburg syndrome 6, 9,27,38,39 . In this study, we demonstrated with a model of Waardenburg syndrome that Pax3 is necessary for the correct distribution of cochlear melanocytes, whilst an intermediate cell marker, S100, expressing Pax3-Cre derivatives from neuroepithelial cells distributed and Dct + or Kir4.1 + cochlear melanocytes were localized as intermediate cells in the stria vascularis at late embryonic day. Furthermore, our study pointed out that small number of cochlear melanocytes can still development despite Pax3 de ciency, suggesting that they do not always require Pax3 and may develop with any redundant pathway. In normal development, it is newly reported that cochlear melanocytes develop from Schwann cell precursors derived from neural crest cells, starting to migrate into the stria vascularis around middle embryonic day (E15.5) 8 .
Interestingly, a previous report found no Pax3-Cre derivatives or Dct + melanocytes in the stria vascularis at E15.5, although the distribution of Pax3-Cre derivatives remained normal in the glial cell region 6 . In contrast, we observed Pax3-Cre derivatives in both the glial cell region and stria vascularis, and Dct + or

Genotyping
Mouse genomic DNA was isolated from collected tail tips by adding 180 µl of 50 mM NaOH and incubating at 98°C for 10 min, followed by the addition of 20 µl of 1 M Tris-HCl. PCR was performed to genotype transgenic mice with three speci c primers which sequences were described in a previous paper 18 .

Auditory physiology measurements
Auditory brainstem responses were recorded as described in a previous paper 42 . Brie y, P56 mice were anesthetized with a ketamine/xylazine mixture (100 mg/kg ketamine and 10 mg/kg xylazine, IP) and placed on a heating pad at 37°C. Auditory brain responses were measured with a needle electrode which was located inferior to the tympanic bulla, referenced to an electrode on the vertex of the head, and a ground electrode was inserted at the hind limb. Tone burst stimuli were delivered with frequencies ranging from 8 to 32 kHz (8.0, 16.0, 32.0 kHz) up to 90 dB sound pressure level (SPL) in 5 dB steps. At each frequency and SPL, 512 trials were tested and averaged.
H&E staining P43 mice were deeply anesthetized with pentobarbital sodium (50 mg/kg, IP), and perfused transcardially with 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS), pH 7.4. Temporal bones were harvested and additionally xed in 4% PFA overnight at 4°C. After PBS wash, inner ears were dissected from temporal bones and decalci ed in 0.125 M EDTA for 2 weeks. Inner ears were dehydrated, embedded in para n wax, and sectioned to 4 µm using microtome REM-710 (YAMATO). Then, sections were stained with Hematoxylin and Eosin.
In situ hybridization: Harvested E18.5 heads were xed in 4% PFA overnight at 4°C, embedded for cryosections and sliced into 10 µm sections as described above. RNA probe synthesis and section in situ hybridization were performed as previously described with some modi cations 44 . Brie y, a digoxigeninlabeled antisense RNA probe was synthesized using the DIG RNA Labeling Kit (Promega) with plasmids containing the following mouse genes: Dct (forward primer, TCCCGAGGCAACCAACATCT; reverse primer, CAGTAGGGCAACGCAAAGGA) and Kir4.1 (forward primer, GGACAAACCCTTATCTGATTCCA; reverse primer, TGCGCAATAAGAAGCACGAT). Slides were permeabilized with protein K (Roche, 5 µg/ml) in PBS with 0.1% Tween 20 for 10 min at 37°C, incubated with 1 µg/ml digoxigenin-labeled riboprobe in hybridization buffer for 16 hours at 70°C, blocked with 10% heat-inactivated sheep serum in Tris-buffered saline containing 0.1% Tween 20 for 30 min at RT, incubated with anti-digoxigenin antibody-conjugated alkaline phosphatase (1:2000, Roche) in Tris-buffered saline containing 0.1% Tween 20 and 1% heatinactivated sheep serum for 2 hours at RT. Antibody detection was performed by incubating slides with 0.2% nitroblue tetrazolium and 0. Imaging and cell quanti cation Section cochleae were captured using Axio Imager D1 (Zeiss) for bright eld images or LSM500/880 (Zeiss) for uorescent images. Image analyses were performed using Zen Software (Zeiss) and Photoshop CS6 (Adobe Systems). Cells were quanti ed from each turn (apical, mid-apical, mid-basal and basal turn) in the developing stria vascularis of section images. Melanocytes in the stria vascularis were identi ed as Dct + or Kir4.1 + cells immediately next to marginal cells.

Statistical analyses
Statistical analyses were performed using Microsoft Excel (Microsoft) and GraphPad Prism 7.03 (GraphPad). Two-way analysis of variance (ANOVA) was used for comparison with two independent variables. P < 0.05 was considered statistically signi cant.

Paint injection
Paint-lling of the inner ear was performed as previously described 22 . E15.5 mouse embryos were harvested and xed overnight in Bodian's xative. Samples were then dehydrated with ethanol and cleared with methyl salicylate. Glass micropipette was inserted in the utricle, and then inner ears were visualized by injecting white latex paint in 0.1% methyl salicylate into the membranous labyrinth. Samples were captured with stereomicroscope SZ61 (Olympus).