INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

TWO MORPHOLOGICALLY DISTINCT types of hair cells occur in the vestibular epithelia of mammals, birds, and certain reptiles. Type I cells are flask shaped, having a broad base, narrow neck region, and flared apex. The basolateral regions of the type I cell are encased by an afferent nerve calyx. Type II cells, present in all vertebrates, are generally more cylindrical in shape and receive bouton-type innervation at their basal pole. Displacement of cilia, projecting from the apical surface of the hair cells into the K⁺-rich endolymph, modulates the flow of mechanoelectric transduction current into the cells. Vestibular dark cells, located adjacent to the neuroepithelium, secrete K⁺ into the endolymph through apical K⁺ channels (17). The basolateral electrical membrane properties of type I and type II cells have been found to vary significantly, suggesting that these two cell types play different roles in the processing of mechanoelectric transduction signals (reviewed in Ref. 4). Vestibular hair cells maintain their characteristic morphology after isolation (26) but, under appropriate stimulation, may undergo somatic shape changes (6, 12, 16, 25, 36). Mammalian type I vestibular hair cells are reported to shorten by 0.5-1.0 µm in response to a variety of stimuli, including depolarizing steps in voltage clamp (12, 25), cell cooling (36), and external perfusion of isosmotic solutions with a raised K⁺ concentration (6, 16). The physiological significance of these changes is not known. However, because the membrane properties of type I hair cells are dominated by a large resting K⁺ conductance, it is likely that K⁺ levels in the restricted space between hair cell and calyx could rise substantially during hair cell stimulation (10, 11, 24).

Motile behavior has been described in outer hair cells of the mammalian cochlea (for review see Ref. 5). These cells can change length at acoustic frequencies when electrically stimulated as a consequence of the operation of voltage-sensitive membrane-bound molecular motors located along the length of the outer hair cell. Depolarization of the cell membrane results in cell shortening, and hyperpolarization results in elongation of the cell. The outer hair cell motor is neither Ca²⁺ nor ATP dependent, but its molecular identity is unknown. The simplest models in which the motor molecules bring about shape changes in the outer hair cell suggest that a specialized submembrane cortical lattice aids in force distribution during motility. The resulting cellular deformations alter the mechanics of the cochlear partition and result in sound amplification. Outer hair cells also undergo slow changes of shape, often occurring over several minutes. The underlying mechanism and physiological significance of this slow motility is less clear, but it may be Ca²⁺ and ATP dependent and under efferent control. There is no evidence to suggest that vestibular hair cells show a fast motility like that seen in outer hair cells; however, the somatic changes in vestibular cells take place on a time scale comparable with the slow motility of outer hair cells, suggesting a possible common mechanism.

The cytoskeleton of vestibular hair cells, which presumably defines cellular morphology, has been shown to consist of intermediate filaments and microtubules (32). A dense meshwork of actin filaments is present in the cuticular plate. Microtubules have also been found in the cuticular plate and in the cell neck, running parallel to the long axis of the cell. These filaments are thought to play a structural role, providing support to the cell and nucleus (32). Contractile proteins have been reported in the apex of mammalian vestibular hair cells, but their role, if any, in vestibular hair cell motility is not known (29).

In addition to the somatic shape changes described in vestibular hair cells, electrically induced and/or spontaneous deflections of hair bundles in the eel ampulla and in bullfrog saccular hair bundles also have been described (2, 28). Whether such deflections of the hair bundle in species having only type II hair cells are related to the somatic shape changes observed in isolated type I hair cells is not clear. However, because the type I cell is restrained by the calyx and neighboring cells in situ, it has been suggested that motile forces generated within the cell, perhaps under efferent influence, could result in a repositioning of the hair bundle and therefore alter the sensitivity of the cell (6, 12, 36).

We describe here experiments on type I hair cells isolated from guinea pig vestibular organs that were designed to investigate possible mechanisms underlying somatic shape changes. Perfusion with high K⁺ has been used to elicit type I hair cell shortening, as described previously (6, 16, 36). The effects of ion substitutions and pharmacological agents on K⁺-induced shape changes have been tested. Our results provide the first evidence that changes in response to high-K⁺ application indicate ion and water uptake through a bumetanide-sensitive Na⁺-K⁺-Cl cotransporter in these cells.

Members of the Na⁺-K⁺-Cl cotransporter family have been described in numerous cell types and are implicated in the maintenance of cell volume and in net transepithelial salt movements (13, 14, 22), for which the stoichiometry for the cotransporter is reported to be 1Na:1K:2Cl. Cotransporter activity is known to be regulated under certain conditions by mechanisms involving phosphorylation (13, 22). In vestibular dark cells, stimulation of the cotransporter with raised external K⁺ results in an increase in height (used as an index of volume) (34, 35). In many cells (e.g., epithelia, endothelial cells, nerve, and muscle), raised external K⁺ results in cell swelling. This is thought to be due to K⁺ or Cl entering through ion channels or coupled transport, resulting in water influx and a subsequent increase in volume. In type I cells, the Na⁺-K⁺-Cl cotransporter could remove excess amounts of K⁺ hypothesized to accumulate in the intercellular space during vestibular stimulation.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Isolation of hair cells. Guinea pigs (female albino, weight 200-400 g) were killed by rapid cervical dislocation. Both bullae were removed, and the end organs (semicircular canals and utricles) were dissected out. Gerbil type I hair cells were also used in three experiments to test high K⁺ and hyposmotic stimuli when guinea pigs were not available. Gerbil cells were dissociated, using methods described previously (24). End organs were removed from Mongolian gerbils (55-70 g) anesthetized with pentobarbital sodium (Nembutal, 50 mg/kg ip) and supplemented with ketamine (10 mg/kg im), followed by decapitation. Experiments were conducted in accordance with the "Guiding principles in the care and use of animals," as specified by the American Physiological Society. The otolithic membrane was mechanically removed from the utricle. End organs were placed in a low-Ca²⁺ phosphate-buffered solution, incubated with papain (0.17 mg/ml, 26-34 min), and washed in bovine albumin (0.5 mg/ml, normal phosphate-buffered solution), and the hair cells were mechanically dissociated as described previously (24). In some cases no enzyme was used, since papain is known to digest neurotransmitter cotransporters and certain ion channels (1, 23) and we wished to minimize alteration of membrane proteins. No obvious differences in responses were found between enzyme-treated and non-enzyme-treated cells. Cells were plated out onto glass chamber slides that had been coated with concanavalin A (0.5 mg/ml) so that cells adhered and did not wash away. After the isolation procedure, cells were left for ~20 min to allow settling to the base of the chamber. Thereafter, cells were perfused at ~0.7 ml/min with extracellular solution, using a peristaltic pump (Minipuls 2, Gilson International). Experiments were carried out at 22-24°C. As judged by morphological criteria (smooth opaque membranes, no nuclear membrane visible, and no granularity within the cytoplasm), hair cells remained healthy and were used in experiments for up to 7 h after extraction of the vestibular end organs from the labyrinth.

Image acquisition and cell measurements. Cells were viewed with ×40 [0.75 numerical aperture (NA)] or ×63 (0.9 NA) water immersion objectives on the stage of an upright microscope (Axioskop, Zeiss) equipped with Nomarski optics and a video camera (Wat-902, Watek). Images were recorded on an S-VHS videotape recorder (Panasonic AG 4700). During each experiment, up to four images, recorded at regular intervals over time, were digitally stored on an image processor (Arlunya, TF600; Dindima) to aid in motion detection. Before drug application, cells were identified according to their neck-to-plate ratio (NPR) and neck-to-body ratio (NBR) (27). This identification is based on previous measurements of avian and mammalian type I and type II cells in fixed tissue (where type I cells were identified by the presence of a nerve calyx) and measurements on dissociated cells (27). Hair cell measurements were made of the cell length, cuticular plate width, minimum neck width, and maximum cell body width perpendicular to the long axis of the cell. Cell length was measured as the distance from the center of the cuticular plate to the base of the cell. The error associated with length measurements was estimated to be 0.5%. Neck width was measured at the narrowest point below the cuticular plate (27). All type I cells studied here fell into the previously established group 1 category (NPR <0.70, NBR <0.64). Type II cells belonged to group 3 (NPR >0.70, NBR >0.58) (26, 27).

Solutions. The bath solution contained (in mM) 145 or 148 NaCl, 2 or 5 KCl, 1.1 CaCl₂, 2.0 MgCl₂, 3 D-glucose, and either 10 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid or 2 NaH₂PO₄ and 8 Na₂HPO₄. Unless otherwise stated, all solutions were adjusted to 305 mosM, and the pH was adjusted with NaOH or HCl to 7.4. High-K⁺ solution contained 125 mM K⁺ substituted (equimolar) for NaCl. Hyposmotic solution contained a lowered NaCl (120 mM). For the hyperosmotic solution, sucrose was added to the normal extracellular solution to increase the osmolarity to 350 mosM. Zero Ca²⁺ solution contained no added CaCl₂ and 1 mM ethylene glycol-bis(-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA). In 0 Na⁺ solution, Na⁺ was substituted (equimolar) with N-methyl-D-glucamine (NMDG). In 0 Cl solution, Na⁺, K⁺, and calcium gluconate and MgSO₄ were used to substitute (equimolar) for chloride salts.

Statistical analysis. Mean values ± SD are shown. For statistical analysis, logarithms of the percent values were taken and compared, using a paired Student's t-test. In cases of a nonnormal distribution, a Wilcoxon signed rank test was performed.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Hair cell shape changes in response to high-K⁺ application. An example of shortening and widening in response to high K⁺ in an isolated type I cell is shown in Fig. 1. Forty-five of fifty (90%) type I cells shortened by >0.5% in response to high K⁺. Two cells studied showed a very pronounced shortening in response to K⁺, to ~60%, but showed little recovery and have therefore been excluded from the sample. The mean length, as percent fraction of the control, was 96.7 ± 3.1%, representing an average decrease in length of 0.8 µm, and the mean width was 104.9 ± 5.7% of control in the presence of 125 mM K⁺ (n = 48 cells; Fig. 2A), an average increase of 0.4 µm. The cell length and cell width in high K⁺, shown in the distribution histograms of Fig. 2, B and C, were statistically significantly different from control (signed rank test and paired t-test, respectively, P < 0.001). Sixty seconds after the termination of the K⁺ pulse, cell length and width were 98.4 ± 2.8 and 101.4 ± 3.8% (n = 48), respectively, indicating a partial recovery. Recovery values for length and width were significantly different from values in high K⁺ but were also significantly different from control values (signed rank tests and paired t-tests used for length and width values, respectively; Fig. 2A). Approximately one-third (34.9%) of cells that shortened in high K⁺ recovered to lengths of 99.5% or greater during the first minute after K⁺ removal. During shortening, the cell neck length typically decreased and width increased, while the perinuclear region of the cell base increased in diameter. In some cases, neck shortening was asymmetric, resulting in a change in the angle of the cuticular plate as has been reported previously (6, 12, 36). These asymmetric shortenings were not analyzed further, because the cells were stuck down and it was not possible to determine whether the tilting was inherent to the cell. The width of the cuticular plate did not change during shortening. The mean control value for  was 0.38 ± 0.1. In high K⁺,  = 0.48 ± 0.1, which was statistically significantly different from the control (signed rank test, P = 0.02, n = 9), indicating that type I cells became significantly rounder during high-K⁺ application. Cells typically responded to repeated applications of high KCl (up to 3 times tested) over periods of up to 20 min. The time course of length changes for a cell during and after high-K⁺ application is shown in Fig. 3.

View larger version (87K): [in this window] [in a new window]   Fig. 1.   Shortening and widening of a type I cell in response to 125 mM K+. Cell is shown before high-K+ application (left) and 60 s into K+ application (right). Reference lines are included to show cell's original length.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   A: mean ± SD responses of 48 type I cells during application of 125 mM K+ and subsequent recovery 60 s after termination of K+ are shown for cell length (filled bars) and cell width (open bars). Measurements are normalized relative to starting cell length or width (100%). Both shortening and widening and their respective recoveries to near control are statistically significantly different from control values: * P < 0.05 and ** P < 0.001. Distributions of cell lengths (B) and widths (C) after application of high K+ are shown. Cell lengths appeared multimodal (B); cell widths conformed to a unimodal normal distribution (C).

View larger version (11K): [in this window] [in a new window]   Fig. 3.   Time course of cell shortening during application of 125 mM K+ for 60 s and recovery after K+ application. Images were taken at 1-s intervals from start of K+ application. Length of cell before K+ application was 18.7 µm. Change in cell length is indicated on y-axis.

Type I cell shortening is not dependent on Ca²⁺ influx. A previous report suggested that depolarization-induced shortening was dependent on external Ca²⁺, implicating a possible contractile mechanism in type I cells (12). Depolarization of the type I cell through elevation of external K⁺ could activate voltage-dependent Ca²⁺ channels, and the resulting Ca²⁺ influx might then trigger a Ca²⁺-dependent contraction of the cell. The effect of removal of external Ca²⁺ was therefore tested by using a solution containing high K⁺, nominally 0 Ca²⁺, and 1 mM EGTA. Removal of external Ca²⁺ has also been reported to reduce the concentration of intracellular Ca²⁺ in guinea pig type I hair cells (3). Ca²⁺ removal did not inhibit shortening in type I cells, as shown in Fig. 4A. Mean shortening in response to high K⁺ was 95.6 ± 4.0%, and cells subsequently recovered to 99.3 ± 3.0% (n = 4) of their original length. The same cells showed a mean shortening when external Ca²⁺ was absent to 95.9 ± 1.5%, which was not statistically significantly different from the high-K⁺ response, with recovery to 99.1 ± 2.4%.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   A: mean cell length in response to high-K+ solution and high-K+, 0 Ca2+, 1 mM EGTA solution (n = 4). Cell length in high K+ was not statistically significantly different from cell length in high K+, 0 Ca2+. B: mean cell length in response to high-K+ application, followed by mean cell length in response to high K+, 5 mM Ba2+.

Type I cell shortening is not blocked by externally applied Ba²⁺. Type I hair cells have a large K⁺ conductance that is active at potentials above approximately 90 mV and is blocked by externally applied Ba²⁺ (24). To investigate whether this conductance plays a part in type I cell shortening, external Ca²⁺ was replaced with 5 mM Ba²⁺. Ba²⁺ did not inhibit but in fact enhanced shortening in three cells tested (Fig. 4B). The mean cell length in high K⁺-Ba²⁺ solution was 76.4 ± 14.7%, whereas the mean length in high K⁺ was 97.4 ± 0.92%. Two of the cells studied shortened to <75% of their control length during application of high K⁺-Ba²⁺ solutions and did not recover when returned to control solution (mean length 78.7 ± 16.9%, n = 3).

Type I cell shortening to high K⁺ is blocked by removal of external Na⁺ or Cl. To investigate a possible role for Na⁺, external Na⁺ was replaced with NMDG. The mean response of nine cells tested with both high-K⁺ and high-K⁺-NMDG solution is shown in Fig. 5A. In the absence of external Na⁺, six of nine cells showed a small length increase when K⁺ was applied, and in the remaining three of nine cells, the response to high K⁺ was reduced. The length in K⁺-NMDG (100.1 ± 1.5%) was statistically significantly different from the response to normal high K⁺ (97.0 ± 1.2%, n = 9, signed rank test, P = 0.004). The mean width increase in the presence of K⁺-NMDG was 103.3 ± 6.9% (n = 9) and was not significantly different from the control mean width increase (108.8 ± 14.9%, n = 9; not shown).

View larger version (14K): [in this window] [in a new window]   Fig. 5.   A: mean responses of 9 type I hair cells to application of high K+ (filled symbols; 20 mM Na+) and high K+-N-methyl-D-glucamine (K+-NMDG; open symbols). Cell length during application of KCl-NMDG was significantly different from cell length during application of normal high K+ (* P < 0.005). B: responses of 6 type I cells to high K+ and a high-K+ solution in which all Cl was replaced by gluconate and sulfate are compared. Each cell is represented by a different symbol. Whereas high-KCl application consistently resulted in mean cell shortening, replacement of Cl resulted in a significant mean increase in cell length (* P < 0.05).

Type I cell shortening is reduced by bumetanide. The dependence on Na⁺ and Cl suggested that an Na⁺-K⁺-Cl cotransporter might be involved in type I cell shortenings. To investigate this further, bumetanide, a specific blocker of the cotransporter, was used. Control responses to high-K⁺ application and responses to high K⁺ with 10 µM (n = 4), 20 µM (n = 3), or 100 µM (n = 7) bumetanide are shown in Fig. 6. Results using the three different concentrations were pooled, since there were no significant differences in length or width values among the different bumetanide concentrations. In response to high K⁺ alone, cells shortened to 96.9 ± 2.0% of their original length and recovered to 98.4 ± 1.5%. In contrast, the mean shortening in the presence of K⁺ bumetanide was 99.1 ± 1.5% and recovery was 99.9 ± 1.4%. Cells widened to 105.4 ± 4.3% in the presence of K⁺ and recovered to 100.9 ± 3.1% (n = 14). Cells widened to 103.6 ± 4.6% in response to potassium bumetanide and showed no recovery (103.7 ± 6.7%, n = 14). Cell length in response to potassium bumetanide was significantly different from length in high K⁺ (paired t-test, P = 0.001), and, although 9 of 14 cells gave smaller responses to K⁺ bumetanide than to high K⁺, cell width was not statistically significantly different between the two groups. Application of control solution (normal K⁺) containing 100 µM bumetanide gave no statistically significant change in length or width compared with control (paired t-test, n = 4).

View larger version (17K): [in this window] [in a new window]   Fig. 6.   Mean responses of 14 cells to K+ application (open bars) and coapplication of high K+ and bumetanide (Bum) (filled bars). Bumetanide concentration ranged between 10 and 100 µM. Cell length in response to K+ bumetanide was statistically significantly different from length in high K+ (paired t-test, * P = 0.001).

Type I cell shortening is not blocked by ouabain. Ouabain, a blocker of Na⁺-K⁺-ATPase, did not block K⁺-induced shortenings. The average shortening to high K⁺ applied in the presence of 0.5 mM ouabain was 96.0 ± 3.9% (recovery 97.3 ± 2.3%, n = 3) compared with a control response to KCl of 97.8 ± 0.21% (recovery, 99.1 ± 1.6%, n = 3).

Response to changes in tonicity. External application of hyposmotic solution (255 mosM), shown in Fig. 7, resulted in type I cell shortening (mean value after 60 s was 92.1 ± 8.9%, n = 8; recovery 95.5 ± 5.5%, n = 4) and widening (112.5 ± 3.9%, n = 7; recovery 103.4 ± 5.2%, n = 3). Both shortening and widening in hyposmotic solution were statistically significantly different from control values (signed rank test, P < 0.05 and paired t-test, P < 0.001, respectively), whereas recovery values were not (paired t-test). In contrast, when the tonicity of the external solution was raised to 350 mosM, an increase in cell length accompanied by a decrease in diameter was observed (Fig. 7). The mean length after 60 s in hyperosmotic solution was 102.4 ± 2.2% and the mean width was 96.0 ± 2.5%. Both measurements were statistically significantly different from control (signed rank test and paired t-test, respectively, P < 0.05). One minute after return to normal osmotic strength, mean cell length and width were 98.6 ± 2.3 and 101.4 ± 3.4% (n = 8), respectively. Recovery values were not statistically significantly different from control. The  decreased from a control mean value of 0.40 ± 0.1 to 0.38 ± 0.08 in hyperosmotic solution, and this difference was statistically significant (P < 0.05, paired t-test, n = 7). In hyposmotic solution cells became more rounded. The  increased from a control value of 0.29 ± 0.04 to 0.33 ± 0.05, and this difference was also statistically significant (P < 0.05, paired t-test, n = 8).

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Type I cell length responses to hyperosmotic (open symbols, n = 8) and hyposmotic solution (filled symbols, n = 8) are compared. Mean ± SD values for length (top) and width (bottom) for control conditions, response to a 60-s osmotic challenge, and recovery 60 s after return to control solution are shown. * P < 0.05; ** P < 0.001.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Our results, in agreement with previously published observations (6, 16), show that vestibular type I cells respond to high K⁺ by a shortening and widening of the neck region and an increase in diameter of the base, resulting in an overall decrease in cell length. Based on the sensitivity of the K⁺-induced changes to bumetanide and their dependence on externally applied Na⁺ and Cl, our results indicate the presence of a bumetanide-sensitive Na⁺-K⁺-Cl cotransporter in these cells. In addition we show that type I cell shortening and widening were produced by hyposmotic solution and were likely due to an increase in cell volume through osmotic uptake of water (14). Although we did not measure cell volume directly, it appears that the shortening and widening in type I cells in response to hyposmotic solution and K⁺, as described in outer hair cells (7), involved an increase in cell volume. As type I cells swelled, they tended to become more spherical. The simplest scheme for the type I hair cell is one in which the neck region is modeled as a cylinder closed at the top by the cuticular plate and inserting at the base into a nearly spherical cell body (Fig. 8). In this configuration, an increase of total cell volume will decrease the length of the neck region, provided that the total membrane area is constant. There will be, to second order, a small increase in the diameter of the cell body as volume increases. Although the diameter of the neck and base of type I cells changed, the cuticular plate width did not; this may be due to the dense meshwork of actin and microtubules in this region (32).

View larger version (12K): [in this window] [in a new window]   Fig. 8.   A model for K+-induced type I hair cell shortening. K+ enters stereocilia through mechanoelectric transduction channels. Cell depolarizes and K+ exits through basolateral K+ channels, resulting in K+ accumulation in intercellular cleft. This increases driving force for net salt influx through Na+-K+-Cl cotransporter, and ion entry is accompanied by water influx. As a result of this influx via cotransporter, type I hair cell swells and shortens.

Hyperosmotic solution resulted in an increase in length and decrease in width, expected if the type I cell were to shrink in volume. It therefore appears that, as described in outer hair cells, cell length is inversely related to volume (7). The cells were only exposed to solutions for 1-min periods; this was insufficient time to determine whether type I cells are capable of volume regulatory mechanisms in the face of osmotic changes (14). Preliminary results indicated that type II cells did not shorten significantly in response to high K⁺.

In vestibular dark cells of the semicircular canals and utricle and in marginal cells of the stria vascularis in the cochlea, a basolateral Na⁺-K⁺-Cl cotransporter is thought to be involved in the uptake of K⁺ from the perilymph (33, 34). Secretion of K⁺ from the apical surfaces of vestibular dark cells occurs through a slowly activating K⁺ conductance (17). This contributes toward maintaining a high concentration of K⁺ in endolymph, the fluid that bathes the apical surfaces and hair bundles of hair cells. Both K⁺ uptake and secretion are blocked by basolaterally applied bumetanide (18, 33). After K⁺-induced swelling, mediated through Na⁺-K⁺-Cl cotransport, vestibular dark cells showed a decrease in volume (35). This was found to be based on K⁺ efflux through a K⁺ channel sensitive to Ba²⁺, lidocaine, quinine, quinidine, and 4-aminopyridine (35). The mechanism for recovery of type I cell length and width after termination of high K⁺ is not known. However, a large Ba²⁺-sensitive K⁺ conductance has previously been described in type I cells (24). After coapplication of K⁺ and Ba²⁺, type I cells showed large, irreversible shortenings. Our results suggest, therefore, that this K⁺ conductance is not required for K⁺-induced cell shortening in isolated cells but that it may be required for recovery.

Bumetanide reduced but did not completely inhibit K⁺-induced shortening. In plasma membranes isolated from dog kidney, radiolabeled bumetanide required ~20 min to reach maximal binding (9), and it is likely that the 60-s exposures to bumetanide used here were insufficient to produce a complete block in type I hair cells. Bumetanide is believed to inhibit cotransport activity by binding extracellularly and preventing the binding of Cl to an intracellular anion site (19). Although the affinity for bumetanide varies between cell types, the dissociation constant for mean saturable binding is generally in the range 1-10 µM (13). At concentrations one or two orders of magnitude greater than this, bumetanide may block certain Cl conductances and other transporters such as Na⁺-Cl cotransport (8, 22). However, we saw no significant differences in the effects of 10, 20, and 100 µM bumetanide on type I cell dimensions, suggesting that the effect was mediated through an Na⁺-K⁺-Cl cotransporter.

Hair cells from the goldfish saccule and lagena have been reported to lose Na⁺, K⁺, and Cl over a period of several minutes when bathed in a solution in which NMDG was substituted for Na⁺. However, no significant volume decrease was observed, and it was suggested that NMDG could permeate the cell membrane (21). A similar loss of osmolytes also may have occurred in guinea pig type I cells during exposure to NMDG. In addition, replacing external Na⁺ with NMDG would lead to an increase in intracellular Ca²⁺ through inhibition of the Na⁺/Ca²⁺ exchanger, thought to be present in guinea pig type I cells and in goldfish saccular hair cells (3, 20). The lack of type I cell shortening in the presence of NMDG also suggests that a Ca²⁺ influx is not involved. This is in agreement with results from a different study on guinea pig type I cells, in which K⁺-induced shortenings were observed when no external Ca²⁺ (or Mg²⁺ ) was present (16). In contrast, no voltage-induced shortenings were seen in type I cells bathed in 0 Ca²⁺ (12). A comparison of the K⁺-induced shortenings described here with voltage-induced shortenings (12) indicates that the two processes have a similar time course, but whether the underlying mechanism is the same remains to be tested. Because the cotransporter is electroneutral, this would seem unlikely. Our results do not support a Ca²⁺-dependent microtubule-driven contraction mechanism.

Ouabain had no effect on K⁺-induced shortening, indicating that Na⁺-K⁺-ATPase was not involved. In the guinea pig, immunocytochemistry using antibodies to the ₂- and/or ₃-subunit isoforms showed no evidence for Na⁺-K⁺-ATPase in vestibular hair cells, and immunoreactivity was confined to the calyx (15). Na⁺-K⁺-ATPase immunoreactivity was also reported in nerve terminals of the gerbil vestibular system (30). In contrast, immunoreactivity to the ₃- and ₁-subunits of the Na⁺-K⁺-ATPase has been reported in the sensory cells of the rat crista and otolithic organs (31).

Isolated type I cells have a large K⁺ current that is significantly activated at the zero-current potential of the cell (24). During positive deflection of the hair bundle, K⁺ will enter through transducer channels, leading to depolarization, and exit via basolateral K⁺ channels (Fig. 8). Estimates from mathematical modeling suggest that the K⁺ concentration in the restricted space between type I cell and the calyx nerve terminal, which envelops it in situ, could increase by up to 42 mM at the base of the hair cell (11). Although the concentration of K⁺ used in this study was somewhat higher than this, vestibular hair cell shortening has been reported in response to K⁺ concentrations as low as 25 mM (36). Osmolarity in the hair cell-calyx space could also increase significantly (10). In the intact epithelia, the consequences of type I cell changes in shape or volume are unclear. It has been suggested that depolarizations could drive active hair bundle deflections, altering mechanotransduction (6, 12, 36). An alternative view is that such changes in K⁺ concentration and osmolarity would be potentially damaging to both type I cells and their calyces. The concentration and regulation of K⁺ in the intercellular cleft remain to be determined experimentally. However, one possible function of the cotransporter, assuming it is located on the basolateral membrane of type I cells, could be reuptake of K⁺ into the cells to prevent continuous depolarization of the calyceal nerve terminals and deleterious effects on afferent coding.