Modulation of cytoskeletal dynamics through growth factor signaling pathways is one of the main effectors of tissue morphogenesis during organ development. The auditory sensory epithelium, located within the cochlea of the inner ear, is a particularly appealing structure to examine how growth factor signaling imparts both the maturation of the cytoskeleton and the static material properties of the cell. The architecture of this tissue has a characteristic patterning of one row of sensory inner hair cells and three rows of outer hair cells with several interdigitating non-sensory supporting cell subtypes. In the mouse, the organization of this sensory epithelium occurs during embryonic development,¹ but the rearrangements of cytoskeletal components of individual cells and the formation of fluid-filled extracellular spaces in the sensory epithelium occur during postnatal development.² One such fluid-filled space between the two rows of microtubule-dense PCs, the tunnel of Corti, has been predicted to permit a fluid traveling sound wave³ that supports OHC amplification.⁴ Furthermore, the presence of cellular material within this tunnel has been reported in cases of Alport Syndrome with sensorineural hearing loss,⁵ suggesting a direct role for the tunnel of Corti in auditory function. Therefore, structural defects to extracellular spaces in the cochlea,^6-8 as well as supporting cell cytoskeletal defects, may have profound effects on auditory function.

One potential growth factor mediating the cellular architecture within the cochlea is the Fibroblast growth factor (Fgf) signaling pathway. The Fgf family of receptors regulates differentiation in both OHCs⁹ and PCs.¹⁰ Initial evaluations of cell stiffness indicate that Fgf-signaling also regulates cell surface mechanical properties within the developing organ of Corti.¹¹ Finally, mutations that alter the dosage of Fgf-receptor 3 result in cochlear impairment and profound hearing loss in both mice and humans.^12,13 Here we test the hypothesis that the genetic ablation of Fgf receptor 3 (Fgfr3) reduces cytoskeletal components that contribute to the development of both cell mechanical properties and the formation of fluid-filled structures necessary for hearing function.

We have previously shown that treatment with Fgf2, a strong ligand for all four Fgf-receptors, delays cell structural development and decreases stiffness at the OHC and PC surfaces in vitro.¹¹ To explore which Fgf receptor(s) could mediate the organization of the cytoskeleton and control the maturation of cell mechanical properties, we analyzed Fgfr3-deficient mice kindly provided by Dr. David Ornitz (Washington University in St. Louis, Missouri, USA). All genotypes were confirmed with PCR, which confirmed that the genetic deletion leads to a loss of both ligand binding and transmembrane domains of Fgfr3.¹⁴ Mice were bred on a C57 black 6 background to generate Fgfr3^−/− animals, which are viable but have a 60 dB hearing loss compared with wild-type littermates.¹⁴ To determine whether defects in OHC function might underlie the hearing loss in these mutants, we first assessed the electromotility of OHCs, a phenomenon crucial for cochlear amplification.¹⁵ We measured OHC voltage-dependent (nonlinear) capacitance, which results from translocation of the electrical charge across the plasma membrane during conformational changes of OHC membrane motors and is considered to be a “signature” of electromotility.¹⁶ Organ of Corti explants were dissected at postnatal day 4 (P4) and cultured for 4 to 5 days in vitro before measuring outer hair cell capacitance with whole-cell patch-clamp recording techniques as described previously.¹⁷ There was no statistically significant difference in nonlinear capacitance in OHCs of Fgfr3^−/− and Fgfr3^+/− mice (Fig. 1A). To compare between different cells, the capacitance was normalized to the cell surface area. No significant differences were observed in the maximal charge translocated across the plasma membrane, average linear (voltage-independent) capacitance, and the potential at the peak of nonlinear capacitance (Fig. 1B). We also noticed apparently normal mechano-electrical transduction in Fgfr3^−/− OHCs (data not shown). Therefore, we conclude that Fgfr3 deficiency does not affect the normal development of the sensory and motor functions of OHCs.

Despite normal OHC electromotility, Fgfr3^−/−mice have no detectable distortion product otoacoustic emissions,⁷ which indicates that the loss of cochlear amplification is due to other factors. For example, defects to surrounding supporting cell structure may disrupt mechanical properties of the cochlear partition.¹⁸ Therefore, we assessed how the loss of Fgfr3 impacts pillar cell structure. Cochleae from P3 mutants and heterozygous littermates were prepared for electron microscopy as previously described.¹¹ Consistent with previous findings,^2,11 three types of supporting cells, the inner phalangeal cell, inner pillar cell, and outer pillar cell, were clearly present at the lumenal surface of the sensory epithelium between IHCs and the first row OHCs of Fgfr3^+/− mice (Fig. 2A). Micrographs of Fgfr3 deficient mice showed a decrease in the supporting cell surface area between IHCs and OHCs relative to heterozygous littermates at P3 (Fig. 2A). In apical regions of Fgfr3 mutants, PCs protruding to the lumenal surface of the sensory epithelium were also smaller and lacked the characteristic bundles of microtubules (Fig. 2B) previously described in normally developing PCs.¹¹ To quantify this difference in tissue architecture, we calculated the average inner hair cell-to-outer hair cell (ITO) distance as previously described¹⁹ for 2 basal and 2 apical samples from 2 animals for each condition at 1 µm, 3 µm, and 5 µm distances from the lumenal surface of the pillar cell. Overall, a 50% decrease in the average ITO distance was observed at all three distances from the lumenal surface of the sensory epithelium (Fig. 2C). This difference could not be statistically tested with our small sample size. Therefore we also examined the lumenal surface of this epithelium by labeling P3 Fgfr3-mutant and heterozygous littermate supporting cells as previously described¹¹ with an antibody against calcium binding protein S100-A1 and phalloidin, which labels actin filaments. In these whole-mount preparations, we also observed a decreased distance between IHCs and the first row of OHCs in Fgfr3-deficient mice (Fig. 2D). These data indicate that while the same numbers of cells are present between the IHC and first row of OHCs in mutant and heterozygous cochleae, fewer cells protrude to the lumenal surface between the IHC and first row of OHCs in Fgfr3-deficient mice.

It is known that Fgf-signaling has direct implications on development of the actin cytoskeleton,²⁰ and that decreased Fgf-signaling can lead to patterning defects during cochlear development.^7,9,13,14 To examine the effects of Fgfr3-deficiency on cell surface mechanical properties, we measured the stiffness at different points across the surface of the sensory epithelium. Average Young’s modulus for regions corresponding to OHCs and PCs was calculated as previously described¹¹ and compared between Fgfr3 knockout mice and heterozygous littermates. Results indicate that at P3, OHC Young’s modulus (mean ± s.e.m) was decreased from 6.45 ± 0.95kPa to 2.59 ± 0.61kPa in Fgfr3^−/− mice and PC Young's modulus was decreased from 3.95 ± 1.06kPa to 1.68 ± 0.38kPa in Fgfr3^−/− mice relative to heterozygous littermates (Fig. 2E). Together, these results suggest that Fgfr3 does regulate microtubule formation in PCs, and that microtubules may contribute a significant part of the developing Young’s modulus at the cell surface.

Since Fgfr3^−/− mice have fewer microtubules in PCs, we set out to determine whether acute modulation of Fgf signaling would also influence microtubule formation in wild type mice. Therefore, cochlear explant cultures were established at E16, P0, and P3, and maintained 18 h in vitro with one of the following: Fgf2 and heparin sulfate to bind and activate Fgf-receptors, SU5402 to inhibit Fgf-receptor tyrosine kinase activity,²¹ or vehicle control as previously described.¹¹ To confirm the effectiveness of these treatments, an antibody against p75 neurotrophin receptor (p75ntr) was used to assess differentiation of PCs, which showed increased p75ntr expression and supporting cell de-differentiation induced by Fgf2 treatment, and decreased p75ntr expression under SU5402 treated conditions as previously described.^11,19 At postnatal day 0 (P0), β-Tubulin I&II localized to the lateral side of the OHC below the lumenal surface (Fig. 3A), and is consistent with previous expression profiling of β-Tubulin isotypes during development in vivo.^22,23 By comparison with control cultures, Fgf2- and SU5402-treated cultures showed decreased β-Tubulin I&II fluorescence (Fig. 3A). To quantify this difference, OHCs were counted as positive or negative for β-Tubulin I&II fluorescence (Fig. 3A), as untreated OHCs contained β-Tubulin I&II at all time-points measured. Then the percent of positively labeled OHCs per culture (mean ± s.e.m.) was calculated and compared between control and treated conditions. Cultures treated with Fgf2 had a significantly lower percentage of β-Tubulin I&II labeled cells at E16 and P3 when compared with control conditions, while SU5402-treated cultures showed a significant decrease in β-Tubulin I&II only at P3 (Fig. 3B). These data suggest that disruptions to Fgf-signaling lead to decreased levels of β-Tubulin in apical regions, which, over time, may contribute to the observed patterning disruption and auditory deficits in Fgfr3-deficient cochleae.

To identify potential downstream transcriptional targets of Fgf-dependent microtubule development in OHCs and PCs, cochleae were cultured and treated as previously described¹¹ for 2 days in vitro with one of the following: Fgf2, SU5402, or vehicle control. Six cochleae per condition were pooled for RNA extraction and cDNA synthesis using an RT² First Strand Kit (Qiagen). A gene expression panel of 84 cytoskeletal regulators was screened and repeated in triplicate for each condition using a commercially available PCR array (PAMM-088ZA, Qiagen). To express similarity among relative gene expression levels, a hierarchical clustering method was used (RT² Profiler^TM Software, Qiagen). This method applied the correlation coefficient as a similarity metric, as well as the absolute expression level and an average linkage method, to calculate the average distances between all pairs of genes. Based on this method, 18 clusters of genes were linked for these tissue samples (Fig. 3C), and one cluster of genes in particular (Aurkc, Msn, Cdk5, and Fscn2), which is involved in actin regulation and cell-cycle progression,^24,25 showed increased expression when treated with SU5402, and decreased expression when treated with Fgf2, relative to control conditions (Fig. 3C). Since no change in proliferation was observed in Fgfr3-deficient conditions^7,19 we examined the fold expression change normalized to a house-keeping gene panel of β2 microglobulin, glyceraldehyde 3-phosphate dehydrogenase, glucuronidase-β, and heat-shock protein-90, for each gene, and found significant differences between control and treated conditions in microtubule regulators (Fig. 3D). Cytoplasmic linker associated protein 2 (Clasp2) showed a significant 3-fold increase in expression in Fgf2 relative to control conditions (Fig. 3D). The protein product of this gene has been shown to stabilize a subset of microtubules²⁶ indirectly through binding with CLIPs and End-Binding (EB) proteins.^27,28 In SU5402-treated cochleae, statistically significant increases in expression of Crk, Cyfip1, Mapre2, and Mark2 were observed relative to control cultures (Fig. 3D, p-value < 0.05). Cytoplasmic tyrosine kinase adaptor molecule, Crk, had a relative 5-fold higher expression in SU5402-treated cultures. Crk is a member of an adaptor protein family that binds to several tyrosine-phosphorylated proteins,^29,30 including Epidermal Growth Factor (EGF) receptor, upon stimulation with EGF.³¹ These expression data suggest that a loss in Fgf-receptor tyrosine kinase activity may have a different regulatory cascade from stimulation of Fgf-receptors that can lead to upregulation of genes for adaptor molecules involved in receptor tyrosine kinase activities, as has been seen for mitogen-inducible genes.²⁰ Microtubule-associated protein RP/EB family member 2, Mapre2, had 3-fold higher gene expression in SU5402-treated relative to control conditions. Mapre2 encodes the protein EB-2, which binds the plus-end of growing microtubules.³² Enzyme in the ELKL Motif Kinase (EMK) family member Microtubule affinity regulating kinase 2, Mark2, had 4-fold higher gene expression in SU5402-treated relative to control conditions; it is a serine/threonine protein kinase involved in the control of cell polarity and is known to interact with doublecortin³³ and KIF13B.³⁴ The one exception to the characterization of cytoskeletal regulators involved in microtubule formation was Cyfip1, encoding a clathrin heavy chain binding protein that showed a 4-fold increase in gene expression and has been shown to interact with the product of the Fragile X Mental Retardation (FMR) gene.³⁵ FMR protein was first found to be responsible for Fragile X Syndrome, and it has been suggested that the phenotype observed in the central nervous system³⁶ is due to a disruption in protein synthesis, and, more recently, indirectly associated with remodeling of the actin cytoskeleton.³⁷ In summary, both stimulation and antagonism of the Fgf-signaling pathway alters the gene expression profile of different candidate microtubule regulators in the developing cochlea in vitro.

We have characterized the extent to which Fgfr3 contributes to developing microtubule structure and maturing cell surface mechanical properties. We show that while OHCs have normal electromotility in Fgfr3 knockout mice, there is a lack of microtubule formation in PCs, indicating a disruption of the supporting cell patterning. Both PCs and Deiter’s cells contain bundles of microtubules at postnatal stages, and disruptions to Fgfr3 lead to a decrease in microtubule formation. Therefore, it is reasonable to suggest that a lack of microtubules impairs the opening of extracellular spaces in this tissue. Indeed, we show that when microtubules are decreased, there is also a decrease in the distance between IHCs and the first row of OHCs. Fgfr3 knockout mice also have a 50% reduction in surface stiffness of OHCs and PCs. However, microtubules are oriented to polymerize from the basolateral to the lumenal cell surface, and contribute only a part of the cell surface mechanical properties. Therefore, the Young’s modulus may be different when measured from the lateral surfaces of these cells. Finally, we characterized the mechanism by which Fgf-signaling might regulate microtubule formation in the cochlea, and found 3 candidate cytoskeletal regulators previously shown to play a role in microtubule formation. However, because different gene expression profiles were significantly upregulated in Fgf2- and SU5402-treated conditions, we expect that this pathway may cross-talk with other signaling cascades in different cell types to affect microtubule formation. Taken together, our results suggest that the Fgfr3^−/− mouse model featuring impaired microtubule development in supporting cells can lend further insight into the architecture necessary for proper hearing function.