INTRODUCTION

Auditory hair cells convert sound into an electrochemical signal by a process called transduction. At the hair cell’s apical surface, actin-filled, filamentous stereocilia are arranged in parallel rows of increasing length that extend into the scala media (reviewed in Hudspeth 1997). Endolymphatic fluid surrounds the stereocilia and apical surface of hair cells. Sound causes deflection of stereocilia that, when directed towards the tallest stereocilium, opens cation-selective transduction channels, allowing endolymph potassium (K+) and calcium (Ca2+) to flow into and depolarize the hair cells (Lumpkin et al. 1997; Ricci and Fettiplace 1998).

In vitro studies indicate that Ca2+ regulates several hair cell functions. Bullfrog saccular maculae hair cell transduction channels have been shown to be selective for Ca2+ over monovalent cations (Lumpkin et al. 1997). In ionic environments approximating endolymph, Ca2+ carries as much as 20% of turtle auditory hair cell transduction currents (Ricci and Fettiplace 1998). Generation of stable transduction currents in chick vestibular hair cells requires minimum extracellular Ca2+ concentrations of 20 μM, and transduction currents are not elicitable in 10 μM extracellular calcium (Ohmori 1985). Bullfrog macular epithelium microphonic currents, produced by deflecting hair bundles, also require greater than 10 μM extracellular Ca2+ (Corey and Hudspeth 1979). Bundle stiffening is diminished (Ricci et al. 2000) and adaptation is slowed (Assad et al. 1989; Ricci and Fettiplace 1998) when extracellular calcium concentrations are reduced.

In vivo observations suggest that maintenance of endolymph Ca2+ is necessary for normal hearing. Application of EDTA, a Ca2+ chelating agent, to the scala media suppresses cochlear microphonics (Tanaka et al. 1979; Marcus et al. 1982). Rats with hypocalcemia secondary to vitamin D deprivation exhibit depressed cochlear microphonic amplitudes and elevated thresholds (Ikeda et al. 1987a). In humans, vitamin D deficiency (Brookes 1985; Ikeda et al. 1989), renal failure (Bergstrom et al. 1973; Johnson and Mathog 1976), and hypoparathyroidism (Ikeda et al. 1987b)—conditions that cause low serum and presumed low endolymph Ca2+ concentrations—are associated with sensorineural hearing loss.

In wild-type rats, endolymph Ca2+ concentrations have been reported as 20–30 μM (Bosher and Warren 1978; Salt 1986; Ferrary et al. 1988). The concentration of Ca2+ in mammalian endolymph is low relative to perilymph and other extracellular fluids, but given the strongly positive endocochlear potential (EP), which drives cations out of the endolymph, Ca2+ is present in concentrations much higher than predicted by passive distribution. In the rat, based on Bosher’s measured perilymph calcium concentration of 640 μM and the endocochlear potential, the Nernst equation predicts that the endolymph Ca2+ concentration should be 0.7 μM, but measured concentrations are 25–40 times higher. In the chinchilla, Ca2+ has been shown to have a 20.2 mV electrochemical potential gradient from the perilymph to the endolymph (Ikeda et al. 1988). These observations indicate that Ca2+ is actively transported into the endolymph, but neither the mechanism nor the site of transport is known.

Two families of proteins are known to extrude Ca2+ from cells: sodium/calcium exchangers and calcium ATPases (Carafoli 1987; Miller 1991). Sodium/Ca2+ exchangers rely on elevated extracellular sodium concentrations to transport Ca2+ out of cells, but endolymph sodium concentrations are low, approximating those of intracellular fluid. Furthermore, amiloride, a sodium/Ca2+ exchanger inhibitor, does not lower endolymph calcium concentrations (Ikeda et al. 1988). Application of vanadate, an ATPase antagonist, to the round window has been shown to decrease endolymph Ca2+ concentrations and eliminate the Ca2+ electrochemical potential gradient in chinchillas, suggesting that a calcium ATPase maintains endolymph Ca2+ (Ikeda et al. 1987a, 1988). Vanadate has also been shown to inhibit extrusion of Ca2+ from the apical surface of isolated bullfrog saccular hair bundles that normally lie in endolymph (Yamoah et al. 1998).

Molecules capable of transporting Ca2+ into the endolymph must be present in the apical surface of the endolumenal cell layer whose ion-impermeable tight junctions isolate the endolymph from the surrounding perilymph (Jahnke 1975). The endolumenal layer is comprised of interdental, hair, Hensen, Claudius, external sulcus, stria vascularis, and Reissner’s membrane cells.

PMCA is a plasma membrane protein that, in mammals, exists in four known types (PMCA1–4) encoded by four different genes each of which undergoes alternative exon splicing in regions A and C (Keeton et al. 1993). All four PMCA types have been demonstrated in the mammalian cochlea (Crouch and Schulte 1996). In the endolumenal layer, immunocytochemistry (ICC) has identified PMCA2a in inner and outer rat hair cell bundles (Dumont et al. 2001). In addition, in situ hybridization has demonstrated PMCA1 mRNA in rat stria vascularis (Furuta et al. 1998), and non-type-specific PMCA ICC has suggested that PMCA protein is present in gerbil and guinea pig stria vascularis (Crouch and Schulte 1995; Agrup et al. 1999).

The deafwaddler2J (dfw2J) mutant mouse exhibits profound deafness (Noben–Trauth et al. 1997), vestibular dysfunction, and a wobbling gait. The dfw2J phenotype results from a two-base-pair deletion in the Atp2b2 gene that encodes PMCA2. This mutation produces a frame shift and premature protein truncation after amino acid 471 (Street et al. 1998). Although anatomic abnormalities, including decreased numbers of stereocilia and compressed tunnels of Corti, have been observed in dfw2J (Street et al. 1998), the pathophysiology underlying the mutant’s phenotype is not fully understood.

Given the importance of Ca2+ for hair cell function, the evidence that a calcium ATPase maintains endolymphatic calcium, and the knowledge that mutation of PMCA2 produces a profoundly deaf mouse, we hypothesized that PMCA2 maintains endolymph Ca2+ and that dfw2J mice would have low endolymph calcium concentrations. To investigate this hypothesis, we used immunocytochemistry to determine which PMCA isoforms are present in endolumenal cells where they might pump calcium into the endolymph, and we measured the endolymph Ca2+ concentrations of dfw2J and control mice.

MATERIALS AND METHODS

Animals

Deafwaddler2J mice in the BALB/c background were obtained from Jackson Labs and transferred into the CBA background via serial backcrosses to CBA/CaJ mice for five generations to produce the CBA/CaJ-Atp2b2dfw2J congenic strain that we refer to here as dfw2J. CBA/CaJ (CB) mice were used as controls. For the anatomic and immunocytochemistry studies, all animals were postnatal day 30. For the calcium and EP measurement experiments, all animals were between 4 and 6 weeks old. All procedures were approved by the University of Washington Institutional Animal Care and Use Committee.

PMCA type localization

PMCA-type-specific antibodies NR1, NR2, and NR3 are polyclonal antibodies specific for PMCA1, PMCA2, and PMCA3, respectively (Filoteo et al. 1997). All NR antibodies were used with a goat-anti-rabbit biotinylated secondary antibody (Vector Laboratories, Burlingame, CA). JA9 is a monoclonal antibody specific for PMCA4 (Caride et al. 1996). 5F10 (Affinity Bioreagents, Inc., Golden, CO) is a monoclonal antibody that recognizes all known PMCA types. JA9 and 5F10 were used with a goat-anti-mouse biotinylated secondary antibody (Sigma, St. Louis, MO). PMCA1–PMCA4 antibodies were diluted 1:250, and 5F10 was diluted 1:500 in nonspecific blocking buffer (3% BSA, 3% goat serum, and 0.2% Triton X-100). All antibody and blocking buffer incubations were performed at 4°C.

For each antibody, one dfw2J and one CB cochlea were reacted in whole mount, embedded in plastic, and sectioned to provide high anatomic resolution. Two cochleae from each genotype were embedded in paraffin, sliced, and mounted prior to antibody exposure to allow use of multiple antibodies in each cochlea. Mice were killed by carbon dioxide exposure, and the cochleae were removed and fixed with 4% paraformaldehyde. The specimens were then decalcified in 0.5 M EDTA at 4°C for 4 days. All sections were analyzed by light microscopy.

Plastic sections.

After decalcification, a midmodiolar cut was made, and the cochleae were immersed in 0.25% H2O2 for 30 min. Cochleae were washed in phosphate-buffered saline (PBS), incubated in nonspecific blocking buffer overnight, and covered with primary antibodies for 96 h. Negative controls were prepared with no primary antibody. Following PBS rinses, cochleae were incubated in secondary antibody for 24–36 h. After rinsing with PBS, specimens were incubated in Vectastain Elite ABC solution (Vector Laboratories) for 24–36 h, washed in PBS, and developed with diaminobenzidine tetrahydrochloride (DAB) for 45 min. Then, the cochleae were embedded in Spurr’s plastic (Ted Pella, Inc., Reddington, CA), sectioned to 3 μm thickness, and analyzed.

Paraffin sections.

After decalcification, cochleae were dehydrated through a graded series of alcohols, embedded in Paraplast Plus (3 × 2 h at 58°C; Curtis Matheson, Houston, TX), sectioned at 7 μm, and collected onto glass slides. Sections were deparaffinized, rehydrated through a graded series of xylenes and alcohols, and incubated in 0.25% H2O2 for 30 min at room temperature. After washes in PBS, the sections were incubated in nonspecific blocking buffer for 30 min. Next, they were incubated with primary antibodies for 48 h rinsed with PBS, and incubated with secondary antibodies for 1 h. After PBS rinses, sections were incubated in ABC solution at room temperature for 1 h, rinsed with PBS, developed with DAB, and analyzed.

Anatomy analysis

Plastic sections from negative controls were examined to assess dfw2J and CB cochlear anatomy. Each section was evaluated for intactness of the cochlear duct, tunnel of Corti, stria vascularis, inner and outer hair and support cells, and stereocilia.

Endolymph Ca2+ and endocochlear potential measurements

A microaspiration/microelectrode (aspirode) device was designed to collect a sample of endolymph while simultaneously monitoring the endocochlear potential. The presence of an EP was used to verify placement of the aspirode in the scala media, and stability of the EP during aspiration was monitored to verify the integrity of the cochlear duct. Disruption of the cochlear duct would allow free exchange of ions between the endolymph and perilymph, rapidly depleting the EP and contaminating the sample with Ca2+ -rich perilymph.

Double-barrel, 1.5-mm-inner-diameter, glass capillaries with internal filaments (World Precision Instruments, Sarasota, Fl) were pulled with a David Kopf Instruments (Tujunga, CA) model 700c vertical puller at a solenoid setting of 20 and a current of 19 A. Tips were subsequently broken to an individual barrel inner diameter of 2.5 μm and a total width of 10 μm. One of the barrels was connected to a tuberculin syringe using polyethylene tubing (outer diameter: 0.61 mm, inner diameter: 0.28 mm) (Becton Dickson, Franklin Lakes, NJ) attached with epoxy resin glue. The other barrel was filled with 18.2 MΩ, purified water. Using a Hamilton gas-tight syringe, 0.4 μL of purified water was then dispensed onto a glass slide. This calibration sample was drawn into the aspirating barrel by applying negative pressure with the syringe, the height of the meniscus was marked, and the sample expelled. Aspirodes were allowed to dry overnight, and the electrode barrel was filled with 0.5 M KCl into which a 0.013 in. diameter silver chloride wire was inserted and attached to a voltmeter. A silver chloride ground wire was placed in the animal’s neck musculature and connected to the voltmeter.

A custom Ca2+ quantification well was fashioned from polystyrene and glass and filled with low-fluorescence immersion oil (Cargille Laboratories Inc., Cedar Grove, NJ). Fluorescent dye was aspirated to the calibration mark in the aspirode to be used and was injected into the bottom of the well forming a bubble. The ratiometric, Ca2+-sensitive dye Fura FF (Molecular Probes, Eugene, OR) and fluorescence microscopy were used to measure endolymph sample Ca2+ concentrations, based on a K d derived from a 1–500 μM calcium standard calibration curve (Molecular Probes).

Surgery

Mice were rapidly anesthetized with ketamine/xylazine, and endolymph was aspirated from both cochleae. First, the skull was fixed to a stabilization bar, and the pinna and posterior auditory canal wall were resected to expose the basal turn of the cochlea. A small hole was made in the otic capsule, exposing the spiral ligament, and using a micromanipulator, the aspirode tip was advanced to just touch the spiral ligament. The voltmeter was zeroed, the aspirode was advanced through the spiral ligament until a stable EP was measured, and fluid was aspirated to the calibration mark. The EP was monitored throughout the aspiration; if it changed more than 10 mV, the sample was discarded. Accepted samples were transferred to the bottom of the Fura FF preloaded well.

The well was placed on the stage of a fluorescence microscope, and the edge of the sample bubble was brought into sharp focus. The objective was centered in the bubble, and a 75 W xenon short arc lamp excited the sample at 340 and 380 nm. The resulting 510 nm fluorescence was measured and the R value determined with an intensified CCD camera (Hamamatsu Photonics, Hamamatsu, Japan) and Metafluor (SDR Clinical Technology, Sydney, Australia). The entire process of focusing, recentering, and determining the R value was repeated three times for each sample, and the R value average was used to calculate the Ca2+ concentration according to the formula: [Ca2+]= K d Q(RRmin)/(RmaxR)]. The concentration was then corrected for dilution in Fura FF.

Aspirodes were originally calibrated with 1 μL of water. Aspiration to this calibration mark resulted in a rapid decline of the EP presumably due to cochlear duct rupture. A smaller calibration volume of 0.4 μL was adopted, resulting in stable EPs. To more accurately determine the volume of endolymph withdrawn by aspirodes calibrated with this volume, aspirodes used in the experiments were filled to their calibration mark with water that was then ejected into a preweighed microfuge cap (Eppendorf, Hamburg, Germany) filled with mineral oil. This process was repeated ten times and the cap was reweighed. The original cap and mineral oil weight was subtracted from the result and the difference was used to calculate the sample’s total volume based on a water density of 0.997542 g/ml at 23°C. The result was divided by 10 to determine the individual sample volume. Volumes calculated from five different aspirodes used in these studies ranged from 0.21 to 0.25 μL, slightly larger than the cochlear duct volume measured by magnetic resonance imaging of a mouse (Thome et al. 1999).

Results

Anatomy

Microscopy revealed anatomic abnormalities of varying severity in each dfw2J cochlea. All cochleae contained areas that appeared normal with intact inner and outer hair cells, tunnels of Corti, support cells, and stereocilia (Fig. 1A, B). Variable portions of each dfw2J cochlea, however, exhibited collapsed tunnels of Corti, poorly defined cell types, and unidentifiable stereocilia (Fig. 1C, D). Reissner’s membrane was occasionally disrupted in both dfw2J and CB cochlea sections but was always intact in some sections from each cochlea.

Figure 1
figure 1

dfw2J anatomy. Examples of the most normal and abnormal cochlear anatomy in a single dfw2J mouse. A, B Sections with relatively intact anatomy, viewed at 10× and 40×. The tunnel of Corti, Reissner’s membrane, and inner and outer hair cells appear normal. C, D Sections with distorted anatomy, viewed at 10× and 40×. The tunnel of Corti and hair cells are not identifiable. All sections are embedded in plastic, 3 μm thick, and viewed with Nomarski optics. SM, scala media; SV, scala vestibuli; ST, scala tympani; arrow, Reissner’s membrane; large arrow head, inner hair cell; small arrow heads, outer hair cells; *, tunnel of Corti. Scale bar = 50 μm for A and C and 20 μm for B and D.

Immunohistochemistry

Wild-type mice demonstrated prominent PMCA1 labeling of inner hair cell basolateral membranes (Table 1; Fig. 2, NR1 CB; Fig. 4, CB). Though cellular resolution was somewhat limited by the intensity of staining, PMCA1 labeling appeared to be present around the entire plasma membrane of the stria vascularis marginal cells that surround the scala media (Table 1; Fig. 3A, NR1 CB; Fig. 4, CB). Interdental cells, lining the furrows between spiral limbus Hushke’s teeth, were also intensely labeled for PMCA1 (Table 1; Fig. 3B, NR1 CB; Fig. 4, CB). dfw2J mice demonstrated similar stria and interdental cell PMCA1 labeling (Table 1; Fig. 3A, NR1 dfw2J; Fig. 3B, NR1 dfw2J; Fig. 4, dfw2J), but inner hair cell labeling in the mutants was apical rather than basolateral (Table 1; Fig. 2, NR1 dfw2J; Fig. 4, dfw2J). dfw2J mice also exhibited apical outer hair cell staining not observed in wild-type mice (Table 1; Fig. 2, NR1 dfw2J; Fig. 4, dfw2J). Low-intensity, diffuse PMCA1 labeling was present occasionally in dfw2J and CB Hensen and external sulcus cells (Table 1; Fig. 4, CB and dfw2J).

Table 1 PMCA ICC summarya
Figure 2
figure 2

PMCA1, PMCA2, and PMCA4 staining in the organ of Corti. NR1 CB NR1 immunocytochemistry (ICC) for PMCA1 in CB organ of Corti. There is intense NR1 labeling of the inner hair cell basolateral membrane. There is no NR1 labeling of outer hair cells. NR1 dfw2J NR1 ICC for PMCA1 in dfw2J organ of Corti. NR1 labeling is concentrated at inner and outer hair cell apical surfaces with minimal basolateral labeling. NR2 CB NR2 labeling for PMCA2 in CB organ of Corti. There is prominent NR2 labeling of outer hair cell stereocilia (moderate labeling of inner hair cell stereocilia is more prominent in other sections). NR2 dfw2J NR2 labeling for PMCA2 in dfw2J organ of Corti. There is no NR2 labeling of the stereocilia. JA9 CB JA9 labeling for PMCA4 in CB organ of Corti. There is no JA9 labeling of the CB organ of Corti. JA9 dfw2J. JA9 labeling for PMCA4 in dfw2J organ of Corti. There is moderate labeling of inner hair cell stereocilia (labeling of outer hair cell stereocilia is more prominent in other sections). Large arrowhead, inner hair cell stereocilia; small arrowheads, outer hair cell stereocilia; +, inner hair cell nuclei; *, outer hair cell nuclei; TM, tectorial membrane. All slices are 3 μm thick, imbedded in plastic, and shown at 100× magnification with Nomarski optics. Scale bar = 10 μm.

Figure 3
figure 3

A PMCA1 and PMCA2 staining in the stria vascularis. NR1 CB NR1 ICC for PMCA1 in CB cochlea. There is intense NR1 labeling of the stria vascularis and low-intensity labeling of the external sulcus. NR1 dfw2J NR1 ICC for PMCA1 in dfw2J cochlea. There is intense NR1 labeling of the stria vascularis and low-intensity labeling of the external sulcus. NR2 CB NR2 ICC for PMCA2 in CB cochlea. There is no NR2 labeling of the stria. NR2dfw2J NR2 ICC for PMCA2 in dfw2J cochlea. There is no NR2 labeling of the stria. SV, stria vascularis; ES, external sulcus. All slices are imbedded in plastic, 3 μm thick, and shown at 100× magnification with Nomarski optics. Scale bar = 20 μm. B PMCA1 ICC of spiral limbus. NR1 CB NR1 ICC for PMCA1 in CB cochlea. There is intense NR1 labeling of spiral limbus interdental cells. NR1dfw2J NR1 ICC for PMCA1 in dfw2J cochlea. There is intense NR1 labeling of interdental cells. TM, tectorial membrane; IC, interdental cells. Slices are imbedded in plastic, 3 μm thick, and shown at 63× magnification with Nomarski optics. Scale bar = 20 μm.

Figure 4
figure 4

Summary of PMCA staining. The schematics depict PMCA ICC results in the endolumenal cell layer of CB and dfw2J mice. In CB mice, intense PMCA1 labeling was present in the stria vascularis, basolateral membranes of inner hair cells, and interdental cells. Low-intensity PMCA1 labeling was present occasionally in Hensen and external sulcus cells. Intense PMCA2 labeling was present in the outer hair cell stereocilia and apical surfaces. Moderate PMCA2 labeling was present in inner hair cell stereocilia. There was no staining for PMCA4 in CB mice. In dfw2J mice, intense PMCA1 labeling was present in the stria vascularis, apical surface of inner and outer hair cells, and interdental cells. Moderate PMCA1 labeling was present occasionally in Hensen and external sulcus cells. No PMCA2 labeling was present in the dfw2J endolumenal cell layer. Moderate PMCA4 staining was present in inner and outer hair cell stereocilia. +, inner hair cells; *, outer hair cells; EC, external sulcus cells; HC, Hensen cells; IC, interdental cells; RM, Reissner’s membrane; SV, stria vascularis; TM, tectorial membrane.

In CB mice, PMCA2 labeling was present at high levels in outer hair cell stereocilia and apical membranes and at moderate levels in inner hair cell stereocilia (Table 1; Fig. 2, NR2 CB; Fig. 4, CB). No PMCA2 immunostaining was observed in dfw2J mice (Table 1; Fig. 2, NR2 dfw2J; Fig. 3A, NR2 dfw2J; Fig. 4, dfwJ2). Given the NR2 antibody’s recognition of an epitope upstream of the mutant’s protein truncation, this result confirms that dfw2J is, effectively, a null mutation without expression of the truncated protein (Street et al. 1998).

Immunostaining for PMCA3 was not observed in CB or dfw2J cochleae (Table 1). Moderate PMCA4 labeling was present in the inner and outer hair cell stereocilia of dfw2J mice but was not observed in wild-type animals (Table 1; Fig. 2, JA9 CB and JA9 dfw2J ; Fig. 4, CB and dfw2J). Consistent with its recognition of all four PMCA types, 5F10 labeled all structures identified by NR1–3 and JA9 in both CB and dfw2J mice, with no labeling of other structures (Table 1). There was infrequent low-intensity labeling of Reissner’s membrane by all five antibodies but no consistent labeling with any single antibody. We interpret our inconsistent labeling of Reissner’s membrane as an artifact. However, others have reported prominent 5F10 labeling of Reissner’s membrane in the developing gerbil and low-level labeling in the adult (Crouch and Schulte 1995; Agrup et al. 1999). In addition, in situ hybridization experiments have suggested moderate levels of PMCA2 mRNA in the adult rat Reissner’s membrane (Furuta et al. 1998). These differences may be due to interspecies variation, differing techniques, or differences in threshold criteria.

Endolymph calcium concentrations

The mean endolymph Ca2+ concentration in control animals was 22.9 ± 3.5 μM (n = 8; all means are presented as mean ± standard error; Fig. 5A), consistent with previous measurements in the rat of 23 μM (Bosher and Warren 1978). With the possible exception of one Ca2+ measurement of 42.3 μM, there was no evidence of contamination by Ca2+ -rich perilymph, indicating that our technique is effective in selectively measuring endolymph calcium concentration. The mean endolymph calcium concentration of dfw2J mice was 6.6 ± 0.6 μM (n = 8, Fig. 5A). The difference between dfw2J and CB Ca2+ concentrations was statistically significant with a p value of 0.001 determined by the two-tailed Student’s t test. Of all the wild-type animals, the largest difference between the three measurements made of each sample was 11 μM. Most measurements were tighter and the next largest variabilities were 6.2 and 4.6 μM. Among dfw2J animals, the largest difference in measurements of a single sample was 2.6 μM.

Figure 5
figure 5

A Mean endolymph calcium concentrations. The graph depicts mean calcium concentrations of endolymph samples; n = 8 in each group. Standard error bars are shown. B Mean endocochlear potentials. The graph depicts mean endocochlear potentials; n = 8 in each group. Standard error bars are shown.

Endocochlear potentials

Including initial experiments to develop EP and Ca2+ measurement techniques, stable EPs were recorded in 58% of dfw2J (n = 24) and 61% of control (n = 28) mice (Fig. 5B). Among the mice in which endolymph Ca2+ concentrations were measured, the mean CB EP was 89.3 ± 7.9 mV, and the mean dfw2J EP was 61.9 ± 7.8 mV (n = 8 in each group). The difference between dfw2J and CB EPs was statistically significant with a p value of 0.031 determined by the two-tailed Student’s t test.

Discussion

Our measurements of low endolymph Ca2+ concentrations in dfw2J mutant mice lacking functional PMCA2 suggest that PMCA2 is necessary for maintenance of normal endolymph Ca2+ concentrations. Within the endolumenal cell layer where calcium transport into the endolymph must occur, our ICC results indicate that PMCA2 is present only in inner and outer hair cell stereocilia and outer hair cell apical membranes, indicating that these areas may be important sites of endolymph Ca2+ regulation.

dfw2J anatomy

The anatomic abnormalities observed in dfw2J mice are consistent with those previously reported in dfw2J (Street et al. 1998) and in mice homozygous for a targeted mutation of PMCA2 (Kozel et al. 1998). Anatomic distortion and absence of identifiable hair cells in the most affected areas of dfw2J cochleae likely contribute to the mutants’ deafness, but if the observed abnormal anatomy were the only defect, portions of the cochlea with apparently normal architecture should provide some hearing. Instead, dfw2J mice are profoundly deaf at all tested frequencies (Noben–Trauth et al. 1997), suggesting additional pathology. Sporadic disruption of dfw2J’s Reissner’s membrane likely represents a tissue preparation artifact rather than rupture of the cochlear duct in vivo given that stable EPs were recorded with near equal frequency in dfw2J and in control mice and that membrane disruption was also observed in control mice.

Immunocytochemistry

Previous immunocytochemistry experiments agree that PMCA1 is present in the stria vascularis but differ regarding localization of PMCA1 within marginal cells, the innermost stria cells that line the cochlear duct. PMCA non-type-specific labeling of marginal cells in the developing gerbil was reported as basolateral (Crouch and Schulte 1995) but, in the adult rat, labeling was demonstrated along the entire marginal cell plasma membrane (Agrup et al. 1999). Our results concur with the previous rat study, suggesting that PMCA1 is present on both apical and basolateral surfaces of marginal cells.

To our knowledge, there is no previous report of PMCA1 in spiral limbus interdental cells, which are thought to contribute to maintenance of the tectorial membrane (TM) (Spicer et al. 2000). Chelation of Ca2+ with EDTA causes swelling of the TM (Kronester–Frei 1979) and loss of its normally organized, striated structure (Hasko and Richardson 1988). Interdental cell PMCA1 might provide localized calcium to help establish or maintain the overlying TM.

Our ICC results indicate that PMCA2 is the primary PMCA type present in wild-type mice inner and outer hair cell stereocilia, consistent with findings in the rat, which further specify that the isoform is PMCA2a (Dumont et al. 2001). Occasional PMCA1 immunolabeling of outer hair cell apical surfaces has been reported in the rat (Dumont et al. 2001), but we observed no such labeling in the wild-type mouse. Mutant dfw2J mice, however, demonstrated prominent, apical, outer hair cell, PMCA1 labeling. In addition, dfw2J mice exhibited inner hair cell, apical, PMCA1 labeling that differed markedly from the basolateral labeling that we observed in wild-type mice and that others have reported in rats (Dumont et al. 2001). The abnormal localization of PMCA1 to the apical surfaces of inner and outer hair cells of dfw2J may be a compensatory response to the loss of functional PMCA2.

PMCA3 labeling was not observed in dfw2J or CB mice but has been reported in a ring surrounding rat inner hair cell cuticular plates and occasionally in basolateral membranes and stereocilia (Dumont et al. 2001). Negative PMCA3 ICC has been reported in gerbil cochleae (Crouch and Schulte 1995), and the differences in our observations from those made in the rat may reflect interspecies variability. In the case of pericuticular labeling, confocal microscopy’s ability to focus precisely on immunofluorescent labeling in the cuticular plane may have afforded Dumont et al. greater sensitivity than that provided by our light microscopy technique.

PMCA4 labeling was present in dfw2J IHC and OHC stereocilia but not in wild-type cochleae. Previous studies in the adult rat organ of Corti also report no PMCA4 ICC labeling (Dumont et al. 2001), but PMCA4 immunolabeling of inner and outer hair cells has been observed in immature PN 6, 10, and 15 CB mice organs of Corti (Muchinsky, unpublished data). In situ hybridization has also suggested transient PMCA4 expression in inner hair cells at PN12 (Furuta et al. 1998). These observations suggest that PMCA4 may have a developmental function and may persist in dfw2J mice in the absence of functional PMCA2.

Our immunocytochemical studies show that two different PMCA types, PMCA1 and PMCA4, both respond to the lack of PMCA2 in stereocilia of dfw2J mutants. In the case of PMCA1, expression is turned on in outer hair cells and membrane localization becomes apical in both inner and outer hair cells. For PMCA4, expression is maintained beyond its normal time of down-regulation during development. These changes may reflect shared but hierarchical sensitivity of response elements in the respective promoters, for instance to changes in cellular Ca2+ levels or flux in dfw2J mutant hair cells (West et al. 2002). Compensatory upregulation of related genes in a developmental pathway has been observed, classically in the Myf-5 to MyoD transition in muscle cell development (Weintraub 1993). While the mechanism(s) of the compensation seen in dfw2J remain unknown, that they occur and yet are insufficient to compensate for the loss of PMCA2 suggests that the rapid and potent pumping activity of PMCA2 is critical for its function in stereocilia (Caride et al. 2001).

PMCA2 and endolymph Ca2+

In adult wild-type mice, PMCA1 in the stria vascularis and interdental cells and PMCA2 in hair cells were the only PMCA types identified in the apical membranes of endolumenal cells where it is possible to transport Ca2+ across the tight junction barrier into the endolymph. Despite high levels of PMCA1 in the stria vascularis and interdental cells, abnormal localization of PMCA1 to hair cell apical surfaces, and the presence of PMCA4 in stereocilia, dfw2J endolymph Ca2+ concentrations are significantly lower than those of wild-type mice, suggesting that PMCA2 is necessary to maintain normal endolymph Ca2+ levels. Thus, our findings indicate that, unlike potassium whose endolymph concentration is believed to be established by the stria vascularis (Konishi et al. 1978; Wangemann et al. 1995), Ca2+ endolymph concentrations are maintained, at least in part, by inner and outer hair cells.

Lowered EPs may contribute to dfw2J’s hearing deficit but likely do not account fully for the mutant’s profound deafness. Compound action potentials (CAPs) reflect auditory nerve response to acoustic stimuli (Moller 1983). Gerbils whose EPs have been reduced to as low as 43 mV by chronic furosemide administration exhibit CAP thresholds as high as 80 dB SPL to 20,000 Hz stimuli and as low as 35 dB SPL to 3000 Hz stimuli (Schmiedt et al. 2002). dfw2J’s mean EP of 61.9 mV is not as reduced as the furosemide-treated gerbil. Thus, unless mouse CAPs are more sensitive to decreases in the endocochlear potential, dfw2J’s lowered EP does not account for the mutant’s profound deafness with no ABR responses to stimuli up to 100 dB SPL at 8000, 16,000, and 32,000 Hz (Noben–Trauth et al. 1997). Lower EPs in dfw2J reinforce the significance of low calcium concentrations, for an isolated decrease in the EP should increase the endolymph calcium concentration.

Although our experiments do not directly determine the cause of dfw2J’s deafness, they do suggest that low endolymph Ca2+ concentrations might contribute significantly. As discussed previously, in vitro experiments have demonstrated that chick vestibular hair cell transduction (Ohmori 1985) and bullfrog macular epithelium microphonic currents require extracellular Ca2+ concentrations greater than 10 μM (Corey and Hudspeth 1979). If mouse auditory hair cell extracellular Ca2+ requirements are similar, dfw2J’s 6.6 μM endolymph Ca2+ concentration is not adequate to support transduction, unless localized concentrations of calcium surrounding the hair cells are higher than the gross endolymph concentration that we measured. Additionally, low endolymph Ca2+ concentrations might interfere with the Ca2+-dependent hair cell processes of adaptation and hair bundle movement. Independent of the precise etiology of dfw2J’s deafness, our identification of PMCA2 in wild-type mice inner and outer hair cell stereocilia and outer hair cell apical membranes and the observation of low endolymph Ca2+ concentrations in dfw2J mice suggest that PMCA2 in inner and outer hair cells regulates endolymph Ca2+ concentrations.