A Molecular Mechanism for Electrical Tuning of Cochlear Hair Cells

See allHide authors and affiliations

Science  08 Jan 1999:
Vol. 283, Issue 5399, pp. 215-217
DOI: 10.1126/science.283.5399.215


Cochlear frequency selectivity in lower vertebrates arises in part from electrical tuning intrinsic to the sensory hair cells. The resonant frequency is determined largely by the gating kinetics of calcium-activated potassium (BK) channels encoded by the slogene. Alternative splicing of slo from chick cochlea generated kinetically distinct BK channels. Combination with accessory β subunits slowed the gating kinetics of α splice variants but preserved relative differences between them. In situ hybridization showed that the β subunit is preferentially expressed by low-frequency (apical) hair cells in the avian cochlea. Interaction of β with α splice variants could provide the kinetic range needed for electrical tuning of cochlear hair cells.

Acoustic analysis begins with the discrimination of individual frequency components within the cochlea. The sensory hair cells contribute to frequency selectivity through combinations of electrical and mechanical feedback whose relative importance differs among vertebrate species. For the low-frequency hearing of the turtle, electrical tuning of hair cells provides most of the filtering (1). This mechanism depends on the interplay between voltage-gated calcium channels and large-conductance calcium-activated potassium (BK) channels (2, 3). A 30-fold variation in BK channel kinetics determines the tuning frequency between 50 and 600 Hz (4). BK channels are encoded by theslo gene in chick hair cells (5), where they also support electrical tuning (6). It has been proposed that alternative splicing of the slo gene could provide the functional heterogeneity of hair cell BK channels, and numerous alternative exons of slo have been cloned from chick and turtle cochleas (5, 7–9). However, previous expression of several splice variants revealed little or no difference in steady-state or kinetic parameters (8, 10). We have since identified an alternative 183–base pair (bp) exon (AF076268) in chick hair cells [at splice site 2, hslo numbering (11)] that is identical to an exon cloned from the turtle cochlea (9) and is homologous to the Strex-2 exon from rat adrenal chromaffin cells (12). When this 61–amino acid exon was spliced into a full-length chick slo cDNA (cSlo1-U23821) (13), the encoded channels were kinetically distinct from the exonless form and were more calcium-sensitive. Furthermore, coexpression of these variantslo channels with accessory β subunits exaggerated their kinetic differences but diminished differences in calcium affinity.

Channels were studied by transient transfection of HEK 293 cells (14). α61 (the exon-added α variant) and α0 (cSlo1) (5) were examined by voltage-clamp analysis of excised inside-out patches exposed to different concentrations of calcium (15). Currents flowing across patches expressing either α0 or α61 in 5 μM calcium are overlaid for comparison in Fig. 1A. Tail current deactivation rates were consistently slower for α61. The average deactivation time constants for a range of voltages in 5 μM calcium are shown for these two splice variants in Fig. 1B. On average, α61decayed 2.5 times more slowly than did α0 (similar results were obtained in Ca2+ concentrations ranging from 1 to 20 μM).

Figure 1

α0 and α61 differ in both kinetics and calcium sensitivity. (A) Normalized current traces from IOPs exposed to 5 μM Ca2+ and depolarized to 80 mV (14). Single exponentials were fit to tail currents at a −60-mV repolarization voltage to determine deactivation time constants (α0: τ = 1.5 ms; α61: τ = 4.8 ms). (B) Average deactivation time constants (τ) of IOPs (5 μM Ca2+) plotted versus membrane voltage. (C) IOPs were exposed to 5 μM Ca2+, and g/g max was evaluated from steady-state conditions to generate a conductance-voltage plot. A Boltzmann fit yielded the following V 1/2 values: α0, V 1/2 = 32.5 mV; α61, V 1/2 = −4.4 mV (16). (D) Average V 1/2data from patches exposed to Ca2+ concentrations ranging from 0.2 to 20 μM. The data were fit with straight lines (21) using a least-squares method that produced Rvalues of 0.97 and 0.95 for α0 and α61, respectively.

The voltage dependence of the channel was determined by measurement of the steady-state current amplitude as a function of the command voltage (Fig. 1C). When corrected for driving force, these values provided normalized conductance as a function of voltage and were fit with Boltzmann relations (16). The half-activation voltage (V 1/2) plotted as a function of calcium concentration illustrates the interdependence of calcium and voltage incSlo channel gating. As calcium concentration rose, the channels activated at more negative membrane potentials. The resultingV 1/2 versus calcium plot for α0and α61 (Fig. 1D) shows that addition of the 61–amino acid exon caused a negative shift, equivalent to an increase in the channel's calcium sensitivity. These steady-state effects along with the slowing of deactivation rate were similar to those reported for the Strex-2 slo variant in adrenal chromaffin cells (12).

β subunits combine with and alter the calcium affinity and gating kinetics of mammalian slo-α channels (17), raising the question of whether avian hair cell slo behaves similarly. A presumptive quailslo-β (18) (U67865) was obtained and was coexpressed by mixture with cSlo-α cDNA before the transfection of HEK 293 cells (19). Combination of α0 with β resulted in BK channels that had prolonged deactivation kinetics (Fig. 2A) (20). Decay time constants were more than 10 times larger with β addition. Combination with β also affected the steady-state gating parameters. The V 1/2 of α0β in 5 μM calcium was more than 50 mV negative to that of α0 alone (Fig. 2B). Similar effects were seen when slo-β was combined with α61. Again, tail decay times were prolonged (Fig. 2C), and half activation occurred at more negative membrane potentials (Fig. 2D). These effects were observed at calcium concentrations from 1 to 20 μM.

Figure 2

Addition of β subunits shiftsV 1/2 to more negative voltages and slows tail currents. (A and C) Tail currents at −100 mV in IOPs containing β with either α0 or α61(14). Coexpression of slo-β with α0 (A) or α61 (C) results in a greater than 10-fold slowing of the deactivation rate. (B andD) g/g max was evaluated from instantaneous tail currents and plotted against the activation voltage. Addition of the β subunit shifts the V 1/2 of both splice variants to more negative voltages (V 1/2 = 32.5 and −4.4 mV for α0 and α61; V 1/2 = −20.9 and −22.7 mV with β subunits added to α0 and α61, respectively) (16). The intracellular calcium concentration was 5 μM for these experiments.

The effects of β coexpression on these α splice variants are summarized in Fig. 3. Coexpression with β subunits greatly prolonged the deactivation time of bothcSlo-α splice variants but preserved and even exaggerated kinetic differences between the variants. α61β was five times slower than α0β, compared to the 2.5-fold kinetic difference between α subunits alone. It appears that β coexpression acts to amplify intrinsic kinetic differences between the α subunits, providing more than a 50-fold range in deactivation kinetics between α0 and α61β. A still greater range of kinetic variability may be provided by other splice variants that have been described but remain to be characterized (7–9). For instance, it is expected that faster α forms will be required to match the fastest BK channels observed in turtle hair cells (4).

Figure 3

The addition of β subunits to the two splice variants (α0 and α61) generates four kinetically distinct channels. Open bars are the dissociation constants (K D) for Ca2+ binding at 0 mV.K Ds were calculated from linear fits toV 1/2 versus log[Ca2+] (21), shown in Fig. 1D. The time constant (τ, solid bars) of the decay of tail currents was measured in 5 μM Ca2+at −100 mV. α0 has the lowest affinity for Ca2+ and the fastest deactivation rate, whereas α61β has the highest affinity for Ca2+ and the slowest deactivation rate.

In addition, it should be noted that the α61β combination had even greater calcium sensitivity and slower kinetics than are found in turtle hair cell BK channels. This raises the possibility that the saturation of α61 with β produced channels that might not be found naturally in hair cells. Further insight into β modification of slo channels is derived from consideration of its effects on steady-state parameters. Combination of α subunits with slo-β decreased theK D (the calcium concentration needed to open half the channels at 0 mV) (Fig. 3) (21). When channel gating depends on both voltage and calcium, an increase in the calcium affinity of the channel's open state, relative to that of the closed state, could provide the slower deactivation and left-shifted voltage sensitivity seen with β coexpression.

Coexpression with β subunits exaggerated the kinetic differences among BK channels but minimized variations in calcium binding. The net effect was to give expressed channels the appearance of native BK channels whose kinetics, but not calcium affinity, vary with hair cell tuning (4). Additional support for the role of slo-β in hair cells was obtained with the use of reverse transcriptase polymerase chain reaction (RT-PCR) to amplify it from the quail cochlea (22). The expression pattern ofslo-β in hair cells was examined with in situ hybridization (23), which revealed that slo-β mRNA was found particularly in apical (low-frequency) hair cells of the basilar papilla (Fig. 4). This is consistent with the hypothesis that β subunits combine withslo-α to produce low-frequency tuning (24). A further prediction is that some α splice variants, such as α61, would experience lesser modulation by β if their expression were restricted to basal regions of the cochlea.

Figure 4

Slo-β mRNA is expressed in quail hair cells. In situ hybridization was performed on cochlear cross sections (23). Solid arrowheads point to the rows of hair cells in the basilar papilla. The tonotopic positions of the sections are shown on the schematic cochlear duct [low frequency (F), apical end on top] at left. The tonotopic axis of the chick extends from 100 to 5000 Hz. Hair cell slo-β decreases from lowest to highest frequency regions. Label intensity in the tegmentum vasculosum (open arrowheads) did not vary systematically.

The present results show that β coexpression extends the kinetic range of cSlo-α splice variants whose intrinsic gating differs only modestly. Thus the combination of alternative splicing and β modulation may provide a molecular basis for the functional heterogeneity of BK channels that supports electrical tuning. Only twoslo-α splice variants were studied here, and other alternate exons (7–9) may provide still further variation. Also, additional β subunits, as yet unknown, could provide still other forms of modulation to hair cell channels. The challenge remains to match particular channel proteins to the functional properties of an identified hair cell. Finally, the impressive conservation of channel function among the hair cells of amphibia, reptiles, and birds raises expectations that related molecular mechanisms will be found in the mammalian cochlea, where developmental changes in hair cell excitability (25) parallel the embryonic acquisition of BK channels in chick cochlear hair cells (26).

  • * To whom correspondence should be addressed. E-mail: pfuchs{at}


View Abstract

Navigate This Article