We bred Lmx1a heterozygotes [Lmx1a drJ, Jackson Laboratory strain #000636 (Chizhikov et al., 2006)] to generate null, heterozygote, and control mice [Lmx1a ^+/+ (WT), Lmx1a ^dr/+ (Het), and Lmx1a ^dr/dr (KO)]. According to JAX, a total of 26 mutations of Lmx1a are known, with more details available for 16 spontaneous mutations (https://www.informatics.jax.org/marker/MGI:1888519). Eight spontaneous mutations are among those associated with hearing and vestibular defects. We know that a transversion in exon 2 resulted in an aspartate-to-valine substitution, effectively creating null alleles for Lmx1a (Chizhikov et al., 2006). Mice were collected at E12.5, E13.5, E15.5, E18.5, P2, P8, P10, P14, and P21; genotyped; and fixed in 4% PFA in 0.1 M P0₄ with 300 mM sucrose. Fixation was reduced to 0.4% PFA in 0.1 M PO₄ with 300 mM sucrose before shipping the fixed mice to UNMC for processing. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Tennessee Health Sciences Center [IACUC #15-057 and 18-037] and Creighton University (IACUC #10-35).

Animals were dissected to isolate the inner ear of control and Lmx1a KO mice. Frozen cochlear whole mounts were blocked and permeabilized with a 30% normal goat serum solution and 0.3% Triton X-100 in 1X PBS. Primary antibodies were incubated overnight in 1% normal goat serum solution and 0.1% Triton X-100 in 1X PBS. Cochlear and vestibular samples were immunostained with anti-pendrin (1:100, #2842). Alexa Fluor 488 (Thermo Fisher Scientific Cat#A-21202, RRID: AB-141607) was used as a secondary antibody at a 1:1,000 concentration for 12 h at room temperature (Koh et al., 2023). DAPI (Life Technologies) was used to highlight the expression of all cell nuclei. We also used a recently developed antibody against pendrin [Supplementary Figure S2 (Roux et al., 2023)].

Antibodies were applied to whole mounts flattened after Reissner’s membrane was cut open. In addition, sections were taken at 50-µm thickness using a vibratome, allowing for detailed expression of DAPI and pendrin. Images were captured using a Zeiss 700 microscope at 10x (0.45 NA), 20x (0.8 NA), and 40x (1.3 NA) magnification. Imaging was processed using Zen 3.8 to generate z-stacks and single images.

Dye tracing was performed by inserting dye into the brainstem to label afferents leading to the cochlea, as described by Elliott et al. (2023) and Nichols et al. (2008).

Samples were rinsed with PBS thrice, then placed in 30% sucrose overnight, and embedded in OCT. Sections were performed at −20°C with a 12 µm thickness on a cryostat (Leica). Slides were dried at room temperature, then rinsed with PBS, and permeabilized with Triton X-100 1% for 10 min, followed by blocking in 10% donkey serum for 30 min at room temperature. Primary antibodies diluted in 5% serum/PBS were placed on the slides overnight at 4°C [BSND (Rabbit AB196017-1001; 1/100); CD44 (Rat MA4405; 1/250); KCNQ1 (Guinea-pig APC-022-GP; 1/200) anti-pendrin (1:100, #2842) (Renauld et al., 2021; Renauld et al., 2022)]. The primary antibodies were washed three times, with each washing lasting 5 min, in PBS, and the secondary antibodies were placed at room temperature for 1 h (Alexa anti-rabbit 488; Alexa anti-guinea-pig 488 and Alexa anti-rat 555). Phalloidin-635 (A34054; 1/1,000) was used in conjunction with the secondary antibodies. Images were taken using a Zeiss 700 confocal imaging microscope at 10x (0.45 NA), 20x (0.8 NA), and 40x (1.3 NA) magnification. Imaging was processed using Zen 3.8 to generate z-stacks and single images.

Whole-mount in situ hybridizations were carried out according to the standard procedures (Pauley et al., 2003) using previously characterized digoxigenin-labeled riboprobes for dopachrome tautomerase (Dct) (Steel et al., 1992) and Tbx18 (Trowe et al., 2011; Trowe et al., 2008). Anti-digoxigenin-AP antibody and BM Purple (Roche) were used for colorimetric signal detection. Stained tissue was embedded in soft epoxy resin, sectioned, and documented as described above. Minimizing exposure to 100% EtOH and propylene oxide during embedding is critical for preserving this stain. Hybridizations omitting probes yielded a uniformly pale background stain. Where two ears are described as being simultaneously stained, all steps in the staining procedure were conducted in the same reaction vial. Identical camera settings were used to image different samples.

Animals were anesthetized with ketamine (16.6 mg/mL) and xylazine (2.3 mg/mL) and supplemented as needed to maintain a surgical plane of anesthesia. The core temperature was maintained at 38°C with a heating pad. An incision was made in the inferior portion of the right postauricular sulcus. The bulla was perforated, allowing for exposure of the stapedial artery and the cochlea’s basal and upper turns. A hole was made in the wall of the cochlea near the basal turn using a fine drill. A glass capillary pipette electrode (10–20 MΩ) filled with 150 mM KCl was mounted on a hydraulic micromanipulator and advanced until a stable potential was observed that did not change with increased electrode depth. The ground electrode was implanted in the dorsal neck muscles. The biological signals (filtered at 1 kHz) were amplified under current-clamp mode using an Axopatch 200B amplifier (Molecular Probe, Sunnyvale, CA) and acquired using pCLAMP 9.1 software (Molecular Probe) and Digidata 1322B. The voltage changes during entry into the endolymph were continuously recorded under the gap-free model using Clampex in the pCLAMP software package (version 9.2, Molecular Probe). The sampling frequency was 10 kHz. Data were analyzed using Clampfit and Igor Pro (WaveMetrics, Portland, OR). Six animals from each age/genotype were used.

Previous publications have shown the expression of Lmx1a during cochlea development. Lmx1a is expressed as early as E8.5, and by E10.5, it is expressed in almost the entire otocyst. After initial expression, Lmx1a becomes more absorbed in the endolymphatic sac, which is absent in E18.5-old mice, and the lateral side of the cochlear duct (Mann et al., 2017; Nichols et al., 2020; Nichols et al., 2008). To understand the role of Lmx1a in the lateral wall formation, we analyzed the cochlear ducts of Lmx1a WT and KO at E18.5 and P2 through histological sections (Figure 1). To recognize the different cell layers of the stria vascularis, we used well-known markers such as BSND (barttin) and KCNQ1 (potassium voltage-gated channel subfamily Q member 1) for the marginal cell layer and CD44, a cell surface adhesion receptor for the intermediate cell layer (Morris et al., 2006; Rickheit et al., 2008; Rohacek et al., 2017; Sakagami et al., 1991). KCNQ1, a channel protein essential for extruding K⁺ into the endolymph and ordinarily present at the luminal surface of the marginal cells, is absent in Lmx1a ^−/− mice (Figures 1A, A′). In Lmx1a ^+/+ mice, BSND labels the marginal cells and CD44 labels the intermediate cell layer. Phalloidin allows the delineation of the stria vascularis, which is rich in actin (Figures 1B, B′). In Lmx1a ^−/− mice, BSND expression is absent, and phalloidin labeling highlights a superficial epithelial layer in opposition to the three layers present in the WT (Figures 1C, C′, middle panel). The absence of landmark proteins in the marginal and intermediate cells of the stria vascularis, along with the morphology of single-cell epithelium, shows that the cellular architecture of the stria vascularis is severely compromised in Lmx1a ^−/− KO mice (Supplementary Figure S1).

During the development of the inner ear, neural crest cells that give rise to the intermediate cell layer can be observed lining up above the roof of the cochlear duct as early as E12.5 (Renauld et al., 2022). At E15.5, the melanoblast cells start to ingress into the single-cell layer that starts differentiating into marginal cells. In the absence of Lmx1a, we observe the migration of melanoblasts along the membranous labyrinth from the base to the apex of the cochlear coil (E14.5 in Figures 2A–C), but 2 days later (E16.5), Dct (a melanoblast marker)-positive cells are sparse (Figure 2D, asterisk) and lost by E18.5 (Figure 2E), probably due to the absence of either intrinsic factors or signaling factors required by marginal cells in the epithelial layer of the stria vascularis. On the transverse section at E13.5, the Dct-positive cells are visible above the cochlear roof epithelium in the Lmx1a WT and KO mice (Figure 2B, arrows). At E14.5, Lmx1a ^−/− cochlea has one cochlear turn instead of two (Figure 2C, arrowhead versus arrow; Koo et al., (2009); Nichols et al., (2008)).

Antibodies against several proteins can be used to label the pigment cells, including microphthalmia-associated transcription factor (MITF) and a cluster of differentiation 44 [CD44 (Renauld et al., 2022)]. We used CD44 labels, which are highly positive for pigment cells within the stria vascularis (Supplementary Figure S1B), and also labeled other cells, such as the lateral wall of the Claudius cells (CCs) (Supplementary Figure S1B′). Brainstem dye insertion shows the spiral ganglion neurons (SGNs) that radiate to reach out to the inner and outer hair cells (Supplementary Figure S1A) by radial fibers (RFs) and the modiolus. Lmx1a KO mice revealed poor segregation of the saccule and cochlear basal turn (Nichols et al., 2008) and showed a different pattern of radial fibers in the base (Supplementary Figures S1C, C′). Massive innervation extends to the posterior canal (PC), while the utricle (U) is incompletely fused between the anterior and horizontal canals (AC and HC). In contrast to the control mice, which show pigment cells in the stria vascularis region (Supplementary Figure S1B), Lmx1a KO mice exhibit a large cluster of CD44-positive cells (potentially pigment cells) that do not extend to innervate the stria vascularis (Supplementary Figures S1B, D). Similar to the WT mice, Lmx1a KO mice have a slight stretch of CCs labeled with CD44 (Supplementary Figures S1B′, D′).

Since we observed the absence of marginal cell markers and the failure of the melanoblasts to integrate into the stria vascularis, we studied whether the mesenchymal portion of the lateral wall underneath the stria vascularis was also affected. Previous papers have shown that Tbx18 is essential for condensation and mesenchymal–epithelial transition into the basal cell layer of the stria vascularis (Trowe et al., 2008). At E18.5, the in situ hybridization showed the expression of Tbx18 in the lateral wall above the developing stria vascularis (Figure 3A). In the absence of Lmx1a, the expression of Tbx18 is either exceptionally low or absent (Figure 3B). Further work using quantification is needed. Nevertheless, the near absence of Tbx18 expression and the absence of pigment cells (unlabeled in Figure 3A) in the mesenchymal cells above the stria vascularis may explain the absence of basal layer formation, as observed in Figure 1B′.

Previous publications of Lmx1a KO have shown that in the absence of a functional Lmx1a protein, there is pendrin-positive cell expression expansion (Mann et al., 2017; Nichols et al., 2008). To verify whether this expansion changes the fate of the cells at the stria vascularis level, we analyzed the features on cochlear roof epithelium and outer sulcus cells with specific markers. In previous research, in the absence of Esrp1, the cochlear duct showed an expansion of Reissner’s membrane (Rohacek et al., 2017). In this study, in the absence of Lmx1a, we observed an expansion of the outer sulcus/spiral prominence positivity for pendrin, visible in the whole mount of the organ of Corti and lateral wall, showing intense positive staining for pendrin (Figure 4A). In control mice, a sharp positive signal contrasts with the broader and gradual reduction in pendrin expression in Lmx1a KO mice. Counterstaining with DAPI (Figures 4A′′, B′′) shows a reduction in the number of OHCs from three to one, adjacent to IHC (HC), as previously reported (Nichols et al., 2020; Nichols et al., 2008; Steffes et al., 2012). The density difference in the cells positive for pendrin likely corresponds to the root and the spindle cells (Figures 4A′, B′). In the whole mount, we noticed specific labeling for the acellular tectorial membrane (TM) that is, in part, removed (Figures 4A, B). The extracellular matrix of the TM can be labeled by specific and less specific labeling such as TECTA, collagen, lectin, and Emilin2 (de Sousa Lobo Ferreira Querido et al., 2023; Fritzsch et al., 2024; Jean et al., 2023; Niazi et al., 2024). Pendrin is labeled with one antibody (Figure 4) but remains unlabeled with another antibody (Supplementary Figure S2), suggesting non-specific labeling. Overall, the expression of pendrin is not only observed in the spiral prominence but also in the root cells and outer sulcus (Figure 4A). Immunostaining on sections shows the highly positive expression of pendrin for the spiral prominence with no labeling on the stria vascularis (SV; Figures 5A–C). In contrast to pendrin-positive cells that end in the spiral prominence in the WT sections (Figures 5A–C), pendrin-positive expression continuous between the outer sulcus to the expected site for Reissner’s membrane in Lmx1a KO mice (Figures 5B′, B′′, D). A more detailed investigation will be required to differentiate whether it corresponds to root or spindle cells.

We examined the functional status of the Lmx1a mutant stria by comparing wild-type, heterozygote, and mutant EP (endocochlear potential) at P12 and P21. Six animals from each age/genotype were used. Figure 6 illustrates representative EP waveforms. Without a functional stria vascularis, the EP, a direct measurement of stria function, was not generated, leading to hearing loss. In this Lmx1a KO model, we measured the EP at P12 (during development) and P21 (mature EP) and confirmed a total absence of EP remained at less than 6-mV, as expected in the absence of a differentiated stria vascularis (Figures 6A, B). This absence of EP aligned with previous publications showing a profound hearing loss in Lmx1a KO mice. Thus, a functioning stria vascularis capable of producing the endocochlear potential is never present in Lmx1a mutant mice.

The molecular development of the lateral wall is not fully understood, but its failure to appropriately develop leads to profound hearing loss due to its essential role in endolymph homeostasis (Bovee et al., 2024; Thulasiram et al., 2024; Wangemann, 1995). During embryonic development, the cochlear roof epithelium expresses OC90 as early as E10.5 (Hartman et al., 2015). More recently, Qin et al. showed by RNAseq that OC90-positive cells can be divided into two populations. One is the Wnt4-positive population, which will give rise to Reissner’s membrane, and the other is the GSC-positive population, which will give rise to the future marginal cells (Munnamalai and Fekete, 2020; Qin et al., 2024). Examples of genes essential for forming the stria vascularis have been shown through mutations, such as integrating a retrotransposon in chromosome 11 and creating a cochlear duct without a lateral wall. In this mutant, the spiral prominence attaches directly to Reissner’s membrane, creating a truncated cochlear duct, where the scala media is smaller due to the relatively low height at which Reissner’s membrane is attached (Song et al., 2021). In the knockout for a splicing protein Espr1, the stria vascularis is absent and replaced by a longer Reissner’s membrane due to an ectopic Fgf9 signaling, creating a cell fate switch from marginal cells to Reissner’s membrane cells (Rohacek et al., 2017). In this study, we observed a different cell fate switch toward spiral prominence instead of Reissner’s membrane identity (Figures 1, 5). A similar cell fate has also been reported in the absence of ERR-beta/NR3B2, which led to a partial conversion of marginal cell fate toward the neighboring pendrin-expressing cells (Chen and Nathans, 2007). Further studies are required to elucidate the signaling pathway through which the lateral wall takes the spiral prominence identity.

During normal development, after marginal cells’ differentiation and intermediate cells’ recruitment, the mesenchymal cells aggregate and form the basal cell layer. This allows for the isolation of the stria vascularis interstitial space from the rest of the inner ear, and this separation is essential for hearing function (Gow et al., 2004; Nin et al., 2008; Souter and Forge, 1998; Wangemann, 2006; Xie et al., 2023) In the Lmx1a ^−/− mouse model, we note a single layer of epithelial cells at the level of the stria vascularis, which is visible with phalloidin staining on postnatal sections. Furthermore, the absence of the basal cell layer may be mediated by an absence of Tbx18, a transcription factor essential for mesenchymal aggregation onto the basal cell layer (Figure 3) during the development (Trowe et al., 2008). This absence of proper development into marginal, intermediate, and basal cell layers explains the absence of the endocochlear potential (Figure 6) and compromised hearing in Lmx1a KO mice, as previously reported for this mouse model (Steffes et al., 2012).

To develop properly, the stria vascularis requires the recruitment of pigment cells that will become intermediate cells and extrude K⁺ (Renauld et al., 2022; Steel and Barkway, 1989; Wangemann, 2006). In the absence of Lmx1a, we observed the migration of the melanoblasts (positive for Dct at E13.5), but later, these cells were absent from the cochlear roof (Figure 2). It would be interesting to study more systematically the disappearance of these cells to determine whether they died due to the absence of a survival signal as observed in other strial-defect-related deafness models, such as for the MITF mutation, where MITF has been suspected to be an important melanocyte survival factor (Ni et al., 2013; Steel and Barkway, 1989). A noticeable difference is the timeline of melanoblast disappearance occurring between P1 and P7 in the MITF mutation versus the current model, where the melanoblasts are almost inexistent as early as E18.5 (Figure 2E). This intrinsic defect would also explain the lack of pigmentation in the thoracic region of mutant mice (Ni et al., 2013). Another explanation for the disappearance of the melanoblasts on the roof of the cochlear duct could be that the cells integrate into the glial cells of the spiral ganglion in the absence of recruitment signals from the marginal cell layer as 80% of the intermediate cells originate from glial cell precursors (Renauld et al., 2022). Notably, many CD44-positive cells are clustered around the spiral ganglion neurons (Supplementary Figure S1), but further lineage tracing experiments are required to confirm whether these cells come from the Dct-positive cells present above the cochlear roof at E13.5.

A recent analysis of the pendrin null mutation (Slc26a4 ^−/− ) revealed a role of pendrin in the development of the cochlea, where it is needed for the formation of spindle cells and the spiral ligament, which contain extrinsic cellular components, a factor that enables cell-to-cell communication (Koh et al., 2023). The loss of pendrin disrupts the pH homeostasis mechanism, leading to acidic pH in the endolymphatic potential (Koh et al., 2023). As the stria vascularis is absent and pendrin expression is expanded, it would be interesting in future experiments to assess the pH present in Lmx1a KO mice endolymph. So far, few genes influencing pendrin have been identified in inner ear development. Pendrin expression is dependent on Foxi1, which regulates endolymphatic sac homeostasis and controls the hydrops formation (Hulander et al., 2003; Szeto et al., 2022). Without Lmx1a expression, the inner ear lacks the formation of an endolymphatic duct (Nichols et al., 2008) that is highly positive in pendrin (Figures 4, 5, 7; Supplementary Figure S2). Foxi1 has been shown to regulate pendrin expression in the endolymphatic sac, but its absence does not prevent pendrin expression in the cochlea (Hulander et al., 2003). Pendrin is also not dependent on Pax2, a highly expressed transcription factor in the lateral wall of the developing cochlear duct (Bouchard et al., 2010; Burton et al., 2004; Hosoya et al., 2022). In this study, the expression of non-functional Lmx1a showed increased pendrin expression, extending from the spiral prominence to Reissner’s membrane (Figures 4, 5; Supplementary Figure S2). Future studies on the molecular interaction between Lmx1a and pendrin may be warranted.

It was shown that SOX9 causes deafness via distinct mechanisms in the endolymphatic sac (ES)/duct and cochlea. SOX10 is downregulated, and there is developmental persistence of progenitors, resulting in fewer mature cells. In the postnatal stria vascularis, there is impaired normal interaction of SOX9 and SOX10, repressing the expression of the water channel Aquaporin 3, thereby contributing to endolymphatic hydrops (Szeto et al., 2022). In contrast, to expand hydrops, the ear remains small in the absence of the endolymphatic sac (Nichols et al., 2008; Roux et al., 2023), which does not form a simple sac without canal cristae that continue between the saccule and basal turn in Lmx1a KO mice (Nichols et al., 2020).

In conclusion, the absence of Lmx1a profoundly alters cochlear morphology and function, preventing the formation of a functional stria vascularis and cochlear potential. This disruption leads to an abnormal auditory–vestibular continuum, shedding light on the role of Lmx1a in establishing the distinct ionic environments necessary for auditory and vestibular function. These findings provide important insights into congenital hearing and balance disorders linked to developmental disruptions in inner ear compartmentalization and ion homeostasis. In the absence of Lmx1a, we observe an absence of stria vascularis markers due to the failure of marginal cells to differentiate correctly, the inability to recruit migrating melanoblasts, and the absence of aggregation of basal cells. Furthermore, we also observed an expansion of pendrin expression (schematic, Figure 7), which highlights the importance of Lmx1a in the cell fate determination and differentiation of the lateral wall of the cochlear duct, which does not segregate by a ductus reuniens. Moreover, the absence of the ∼80 mV endocochlear potential in Lmx1a KO mice requires the three layers of the stria vascularis in the cochlea, while it forms a simpler two-layer structure in the lateral wall, resembling the vestibular system with a ∼1 mV potential.