The Inactivating Potassium Currents of Hair Cells Isolated from the Crista Ampullaris of the Frog

JOURNAL OF NEIJROPHYSIOLOGY Vol. 68, No. 5. November 1992. Prinred in U.S.A.

The Inactivating Potassium Currents of Hair Cells Isolated From the Crista Ampullaris of the Frog

C. H. NORRIS, A. J. RICCI, G. D. HOUSLEY, AND P. S. GUTH Departments of Pharmacology and Otolaryngology, Tulane University School of IWedicine, New Orleans, Louisiana 70112

SUMMARY AND CONCLUSIONS sophila (Baumann et al. 1987; Jan et al. 1977; Kamb et al. 1. A-type outward currents were studied in sensory hair cells 1987; Papazian et al. 1987). Investigations have localized isolated from the semicircular canals (SCC) of the leopard frog the sites for voltage-dependence, ion selectivity, and the ki- (Rana pipiens) with whole-cell voltage- and current-clamping netics of inactivation (Catterall 1986; Greenblatt et al. techniques. 1985; Iverson and Rudy 1988; MacKinnon et al. 1988). 2. There appear to be two classes of A-type outward-conduct- Alternate splicing of the Shaker gene has demonstrated that ing potassium channels based on steady-state, kinetic, pharmaco- multiple mRNAs can be made that produce A-channels logical parameters, and reversal potential. that differ in inactivation kinetics as well as in the recovery 3. The two classes of A-type currents differ in their steady-state from inactivation (Iverson and Rudy 1988; Timpe et al. inactivation properties as well as in the kinetics of inactivation. 1988; Zagotta et al. 1989). The steady-state inactivation properties are such that a significant portion of the fast channels are available from near the resting According to Hudspeth ( 1986), the bullfrog’s saccular potential. hair cell has seven ionic currents, including an A current 4. The inactivating channels studied do not appear to be cal- (Z*) . This transient K+ current in the saccular hair cells is cium dependent. fully inactivated at the normal resting potential (Hudspeth 5. The A-channels in hair cells appear to subserve functions and Lewis 1988)) suggesting at best a minimal role in sac- that are analogous to IA functions in neurons, that is, modulating cular hair cell function. Alternatively, Sugihara and Furu- spike latency and Q (the oscillatory damping function). The A- kawa ( 1989) have suggested that the A-currents in saccule currents appear to temporally limit the hair cell voltage response hair cells may be involved in modulating the damping func- to a current injection. tion of the membrane oscillations typical of these cells. Murrow and Fuchs ( 1990) have described an A-type con- INTRODUCTION ductance in the bird basilar papilla that was found exclu- sively in the short hair cells and although not in a physio- The transient outward K+ current, which has come to be logically active range, could be enabled during hyper- called Z*, was first described by Connor and Stevens polarization induced by Acetylcholine, the major efferent ( 197 1). They separated Z* from the delayed rectifier transmitter. current by the use of a protocol that elicited all currents, In contradistinction to the auditory hair cells, in a prelim- then only the noninactivating currents. Subtraction of these inary study of the currents of hair cells isolated from the two currents revealed the existence of the A current. The posterior semicircular canal (SCC) of the frog, Rana pi- protocol that elicited both inactivating and noninactivating piens Housley et al. ( 1989), suggested that an active Z* currents employed a hyperpolarizing prepulse to enable the could be elicited by depolarization from normal resting po- inactivating conductances and thus allow the channels to tentials. Correia et al. ( 1989) and Rennie and Ashmore open when the cell was depolarized. The second protocol ( 199 1) have described an A-type current in bird and guinea used a depolarizing prepulse, which inactivated the A-type pig semicircular canal, respectively. Although the steady- conductances, leaving only the noninactivating conduc- state properties allow these channels to be active in a more tances. physiological range, the relative proportion of the A-chan- In addition, these two potassium currents can sometimes nels is significantly less than that found in the frog. be separated pharmacologically. In some cells, ZA is less sen- A-type currents may be of fundamental importance in sitive to inhibition by tetraethylammonium (TEA) and SCC hair cells. We, therefore, made a more rigorous study more sensitive to block by 4-aminopyridine (4.AP) than of these inactivating outward currents revealing the exis- the delayed rectifier (Rudy, 1988; Thompson, 1977). Thus tence of two classes of A-type channels, differing largely in the definition of A-type channels is that they are voltage- their inactivation properties. sensitive, rapidly-activating, rapidly-inactivating, outward- flowing potassium channels that are blocked by 4-AP. METHODS Several roles have been suggested for A currents. In neurons I) they could control the spike latency, 2) they could Zsolation of hair cells regulate interspike interval, and 3) they could contribute to The procedure adopted essentially follows that reported by action potential repolarization (Rudy, 1988). Lewis and Hudspeth (1983), Hudspeth and Lewis (1988), and A great deal of information about A-channels has come Housley et al. ( 1989) for the isolation of hair cells from frog vestib- from the discovery of the Shaker mutant genes of the dro- ular organs. The external medium used by Lewis and Hudspeth

1642 0022-3077192 $2.00 Copyright 0 1992 The American Physiological Society INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS 1643

TABLE 1. Media decapitated.With the head bathed in external medium (Table 1) and sectionedsagittally, the inner ear wasexposed by openingthe Dissociation, mM External, mM Internal, mM otic capsule.The semicircularcanals were dissectedfree from the rest of the membranouslabyrinth and trimmed leaving only the CaCIZ. 2Hz0 0.02 2 1 ampullae.These were openedand placed in dissociationmedium KC1 3 3 115 (Table 1) containing the proteolytic enzyme papain (0.1 mg/ml; MgC12 l HZ0 1 3 CalbiochemNo. 5 125) and L-cysteine(0.33 mg/ml) for 5 min. NaCl 122 119 The tissueswere then washedin dissociationmedium containing NaH,PO, 2 2 Na,HPO, 8 8 bovine serum albumin (0.5 mg/ml), and the cristaewere moved KH2P0,, 1 to a glass-bottomedbath on an inverted microscope.Mechanical K2HP04 4 twisting of the tissuesthen freed the individual hair cellsfrom the ATP 3 rest of the crista. The bathing medium was then changedfrom EGTA 11 dissociationmedium to external medium (Table 1). D-glucose 3 3 3 Studieswith whole-cell recordingsrequired the use of various 4-AP 10 solutions(Table 1). The bath (2.ml vol) containing the isolated TEA 10 hair cells was perfusedwith external medium at a rate of l-2 GTP 0.5 ml/min, and all experimentswere carried out at room tempera- ture (21°C). EGTA, ethylene glycol-bis(P-aminoethyl ether)-IV,N,N’,N’-tetraacetic acid; 4-AP, 4-aminopyridine; TEA, tetraethylammonium chloride; GTP, Cellswere selectedfor whole-cell patch clamping accordingto guanosine triphosphate. the criteria enumeratedby Housley et al. ( 1989).

( 1983) contains Hepes buffer. In our procedure, this has been Whole-cell recording replacedby phosphatebuffer ( Table 1) (Norris and Guth 1985) . Cellswere isolatedas previously described( Housley et al. 1989). Gigohm seals( 1- 15 GQ) were obtainedwith 1.5-mmOD boro- Leopard frogs (Rana pipiens) were chilled, pithed, and then silicateglass pipettes (Frederick Haer, capillary tubing No. 30-32-

000 pA Ii- 0 rn¶

00 PA Intermediate L 0 ms

FIG. 1. Current vs. time responses for three different cells (top to bottom) to 2 different whole-cell voltage-clamp protocols (A and B) . A : the cell is first voltage clamped to a conditioning prepulse of - 130 mV from a holding potential of -60 mV. After the 80 ms, - 130 mV prepulse the cell is depolarized for 200 ms to a given voltage step and then returned to the -60 mV holding potential. This cycle is repeated every 5 s with a 10 mV more positive step utilized each cycle. Depolarizing voltage steps ranged from - 130 to + 120 mV. This protocol is designed to elicit the inactivating currents as well as the noninactivating channel types such as the delayed rectifier, or the calcium-dependent potassium conductances. B: current responses to a similar protocol are depicted except that the conditioning prepulse has been changed to -20 mV. This protocol has been designed to elicit the noninactivating currents (see Fig. 5 ) . C: current responses are the differences (the inactivating currents) between those in A and those in the B. 1644 NORRIS, RICCI, HOUSLEY, AND GUTH 1) pulled to a resistanceof -3-5 Ma. A continuous single-elec- Drug solutions(4-AP, 1O-20 mM ) were made up in external trode patch-clamp amplifier ( Axopatch- 1B, Axon Instruments) medium (Table 1). The NaCl concentration was adjustedto ac- was usedto record from crista hair cellsduring voltage or current commodatethe drug and maintain osmotic pressure.4-AP was clamp (after Hamill et al. 1981). Unlessspecified, records were obtainedfrom Sigma.It wasapplied by bath substitutionfor 3 to 5 low-passfiltered at 5 kHz with a four-pole Besselfilter. Command min (i.e., 1.5-3~ bath volume) beforethe voltage-steppingproto- potentialsand currents werecontrolled with a 12.bit digital-to-an- col designedto elicit the inactivating current was applied to the alogconverter, and data were sampleddigitally at 2OO-psintervals cells( seeFig. 7 ) . Changesin inactivating currentscaused for exam- with a 12.bit analog-digitalconverter (Labmaster,Tekmar Indus- ple by 4-AP exposurewere expressedas percent of maximal inacti- tries) coupled to a microcomputer (PC’s Limited 80286, Dell vating current elicited in that cell. The currents were elicited once Computer). Clamp speed,which isdependent on uncompensated beforedrug application, at leasttwice during drug application,and seriesresistance and cell capacitance,was estimatedto be 15 ps. at least once during the drug washout period. Data were stored on hard media for off-line analysis(p-clamp version 5.0 1, Axon Instruments). The junction potential between RESULTS the pipette internal solution and the bath was nulled before seal formation. The ground electrodewas placed in a separatebath of Cell parameters internal solution and coupledvia an agarbridge to the recording chamber, so that no significant liquid junction potential should The input capacitance for all the cells used in this study exist betweenelectrodes. was 5.3 t 0.2 (SE) pF (n = 25 ) ; the series resistance was Cell conductanceswere then calculated from current records 9.3 t 0.7 MQ, and the zero current membrane potential resulting from various voltage-clamp step protocols. Each step was -50 t 2 mV. These properties agree very well with the changein voltage was separated by 5 s. In most of the cells, the data previously published by Housley et al. ( 1989). voltage error due to seriesresistance was partially ( 60-80s) can- celedelectronically. The residualuncompensated series resistance The major conductance is inactivating was at most l-3 MQ, which could causean averageof 3 mV of error in the voltage protocols for the largestcurrents recordedand This study will suggest that two classes of inactivating a smallererror for smallercurrents. channels exist in semicircular canal hair cells and that these

. 1 t t t p1116.326&=+134&-= ~=1490+12Obc=+51&= \.

y =A0 +Ale ‘-l +%e ‘2

Percentage of Slow Channels ($1 mo+J4p’L711 FIG. 2. A and B: examples of inactivating outward currents from 2 different cells obtained by applying the subtraction method shown in Fig. 1 to the responsesto the step to + 120 mV (maximal current response). Dotted lines represent the actual data. Each current trace was fitted with a double exponential equation (solid line, see middle of figure, and text). The double exponential was fitted to the data starting r2 ms after the start of the voltage-clamp step to avoid any clamping artifact. A : the fast portion of the double exponential predominates (A I > Al); B: the slower portion of the double exponential predominates (A, > A, ). C: histogram depicting the distribution of the fast (Al ) and slow (A,) coefficients; the bin width is 5 ms. The coefficients represent the relative proportion of a given current in a particular cell. The mean q is 4.4 ms. The mean r2 is 36.4 ms. INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS conductances, defined as fast and slow, are of the A-type. The two conductances exist in varying proportions between cells, so that any given cell will be dominated by either the 7wo O--0Sbw Cob (n-o) 0-0 Fast CdJo (n-l 1) fast or slow channel type, but in general will contain both -6ooo types. For the purpose of distinguishing between channel % U5000 characteristics, cells were grouped on the basis of the rela- % / . tive amount of the fast and slow component. That is, cells that contain mostly fast channels are fast cells, and cells that contain mostly slow channels are slow cells. Where rele- vant, intermediate cells are shown. The proportion of fast and slow conductances were measured from the coefficients of the double exponential kinetic fit to the inactivat- 0 ing current obtained from subtracted data described in Figs. -140 -100 -60 -20 20 60 loo 1 and 2. The ratio AI /(A, + A, + A,) = (R f) reflects the Membrane Voltage (mV) proportion of fast current in the cell, whereas the ratio AJ (A, + A, + AZ) = (R,) reflects the proportion of slow current. For fast cells Rf > R, and for slow cells R, > R f. It is important to realize that the classification is a tool for distinguishing channel types and does not necessarily imply two physiological cell types. Where comparisons between fast and slow channels were performed, cells were chosen that were dominated by either the fast or slow component :: 0.6 (>90%). Choosing the two extreme groups allows for a bio- E logical separation of the channel types. Where no differences were found between the fast and slow components, the cells were pooled and a composite figure is presented.

Preliminary evidence for two independent inactivating channels

KINETICS. All the inactivating current responses, derived -0.2 -150 -100 -50 SO 100 160 as in Fig. 1 A, were examined in 26 different cells. A least- Membrane Voltage squares curve fitting procedure (clampfit, P-clamp version 5.01, Axon Instruments) was applied to the rising phase FIG. 4. Current and chord conductance vs. membrane voltage. A: the peak of the inactivating current derived as in Fig. 1A is plotted as a func- (activating phase) and to the falling phase (inactivating tion of membrane voltage for fast cells (A, > A2 see RESULTS in text) and phase) of each current response to each voltage step. The for slow cells (A, > A, see RESULTS in text). Cells dominated by the slow channels show a much larger macroscopic current than do the cells dominated by the fast current. B: from these plots, normalized chord conduc- 200- 8 0-0 control rubtractlon tance (G/G,,,) was calculated and is plotted. The solid circles are the lSO*- 0-0 hlgh K wbtroctlon $ 2 calculated normalized conductance and for comparison, the solid line is a - (mean f SLM. n-3) / Boltzmann equation fitted to the data with a half-maximal voltage of -2 4 n loo-- 0 mV and a gating charge of - 1. (R = 0.985). so / t - zt- o-- TL 4/ P v /Ii 0 1’ kinetics of activation were fitted with a single exponential W 0y t 0 whose time constant was (1 ms. No detectable difference Q/ -50i-+ .9’f 00 ,,/ - a was found between fast and slow channels, however, both ) -loo:‘* ,/ A had significantly faster activation than reported for other hair cells (Hudspeth and Lewis 1988; Lang and Correia 0 -150*- ,/ - 1989; Rennie and Ashmore 199 1) . Analysis of these param- -200-t- : I- ,1 .I eters of activation are limited by the sampling and settling -130 -110 -80 -70 -60 -30 -10 times as dictated by the cells and the electronics. That is, STEPS (mv) often the curve fitting program applied to the rising phase of FIG. 3. Tail current responses to 2 voltage-clamp protocols. In the pro- the current response only had two or three valid points with tocol, cells are held at -60 mV and then clamped for an 80-ms long pre- which to work. pulse at - 130 mV. After the prepulse, the cells are briefly depolarized to +30 mV for 5 ms, then stepped back to a less depolarized level for 60 ms, The inactivating phase of the maximal current response and finally returned to the holding potential. The cycle is repeated each 5 s (i.e., the response to the step to + 120 mV) was fitted best with a - 10 mV increment. The protocol is repeated except that the pre- with a double exponential [y = A, + A, exp( -t/T,) + A2 pulse used is -20 mV. By subtracting the responses to the 2 protocols, the exp( -t/T2)] where R = 0.998 or better (Fig. 2). In this tail currents for the inactivating channels can be obtained. The results are plotted to show the reversal potential for the channels. Increasing the potas- equation, the mean fast-time constant (T] ) was 4.44 ms sium concentration in the external medium of the cells from 3 to 30 mM (n = 25), and the mean slow-time constant (Q) was 36.4 shifts the reversal potential to an -53 mV more positive level. ms ( n = 25) (Fig. 2C). No significant voltage dependence 1646 NORRIS, RICCI, HOUSLEY, AND GUTH of the time constants was observed. Analysis of the propor- was increased from 3 to 30 mM, the reversal potential be- tions between the coefficients of the two exponentials for came 53 mV less negative. The Nernst equation predicts a each double exponential fit indicated that the proportions change of 58 mV for a potassium selective channel. No of fast- and slow-inactivating channels in these cells varied. difference in this reversal potential was found for fast and Figure 2C is a histogram showing the range in the coeffi- slow channels. Initial comparisons between fast and slow cients, suggesting a range in the proportion of channel types channels were not different so the data has been pooled and found between cells. These results suggest that there are two presented in Fig. 3. inactivating channels, which vary in number but not charac- Additional evidence that potassium is the current carrier ter among cells. for these channels was provided by the observation that replacement of the potassium in the internal solution in the Potassium selectivity patching pipette with cesium eliminates the outward Evidence that potassium is the major current carrier in current response to the voltage-clamping protocol (n = 4, the inactivating channels was provided by tail current analy- data not shown). A similar finding was reported by Fuchs sis. Prepulse protocols, similar to those described in Fig. 1, et al. ( 1990) and in Table 1 of Rudy’s review ( 1988). were used to isolate the tail currents produced by the inactivating conductances (Fig. 3; see legend for protocol detail). Activation The responses to the two protocols were subtracted to Depicted in Fig. 4A is the peak of the inactivating current reveal the tail currents for the A-type inactivating channels. isolated as in Fig. 1 and plotted as a function of the mem- Then the tail current responses were plotted as a function of brane voltage with fast and slow groups of cells. The main the clamped tail voltage steps to reveal a reversal potential difference between fast and slow cells in this figure is that of -77 mV for the inactivating channels (Fig. 3). When the slow cells have almost six times more current than fast cells. potassium concentration in the medium bathing the cells Chord conductance allows for the evaluation of the volt- A OmV l-l 2OmV

-6OmV

-1JOrnV

C 1.0

0.8 0-0 Fad Cells (n-6) X A-A Intermediate Celh (n-17) 0 O-O Slow Cells (w6) 0.6 E

1 0.4

0.2

-1 0.0 -140 PREPULSE (mV) FIG. 5. Inactivation curves for 3 groups of cells. A : protocol used showing that cells are voltage-clamped and held at -60 mV and then stepped to a prepulse for 200 ms. After the prepulse, they are depolarized to 0 mV for 50 ms (test level), then stepped back to the prepulse level for 200 ms (tail current level), and finally returned to the holding level. C: peak current at the test level is plotted as a function of the prepulse level voltage. Examples of the current responses are shown in B where kinetically, fast cells (the /ej-most responses) have sharply inactivating currents, slow cells (the right-most responses) inactivate with much slower time constants, and intermediate cells (the center set of responses) have inactivating rates between the 2 extremes. A separate single Boltzmann equation is fitted to the fast and slow cells’ responses plotted in C, and a double Boltzmann equation is fitted to the intermediate cells’ responses (R = 0.99). The double Boltzmann equation is composed of elements identical to the scaled sum of the 2 single Boltzmann equations, suggesting that the hair cells contain both fast- and slow-inactivating channels in different proportions. INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS 1647

age dependence of activation of a channel independent mann equation was fitted to the data for fast cells ( Rf > from the changing driving force and is given as ga = ( ZA/ 90% ) and ‘to that for slow cells (R, > 90% ) . A double Boltz- ( V- IQ. Here ga reflects conductance at a given voltage; ZA mann equation was fitted to the data for intermediate cells. is the peak current plotted in Fig. 4A and obtained from the The elements for each half of this double Boltzmann equa- subtracted data described in Fig. 1; Vis the voltage step, and tion were not significantly different from the analogous ele- & is the reversal potential calculated from Fig. 3 so that the ments of the separate single Boltzmann equations. That is, difference ( V - VR) is equivalent to the driving force. The for very fast cells the voltage of half-maximal conductance, plot in Fig. 4 B is chord conductance (G) normalized to V0 = -59t l.OmVandtheslope(ze/KT),S= -7.8t0.8 maximal conductance ( G,,,) . No significant difference ( normalized current/volt). For very slow cells, V0 = -95 t was seen in the conductance plots between fast and slow so 0.5 mV and S = -13.4 t 0.4. For the two halves of the cells were grouped and analyzed together. The conductance double Boltzmann equation, the fast V0 = -63 t 0.7 mV plot was shown to fit a single Boltzmann function (-), with S = -5 t 0.8 and the slow V0 = - 10 1 t 5.0 mV with where G/G,,, = [ 1 + ecvo-v,Ze/KT]-l. Here V0 is the voltage S = - 18 + 3.0. The results then suggest that each cell can be of half-maximal conductance ( -2 mV), z is the effective represented by the scaled sum of two independent Boltz- gating charge ( 1 ), and e/ KT is a constant. The single Boltz- mann equations. Thus, these results provide further evi- mann fit (R = 0.985) suggests that the fast and slow chan- dence that there are two types of inactivating channels. As nels have similar activation gates, that the channels operate already seen with the kinetics, the difference between cells is independently, and that the channels exist either in an open the proportion of fast and slow channels present. The very or closed state (multiple closed states may exist) ( Adams et fast cells have mostly the fast inactivating type channel al. 1982; Belluzzi et al. 1985; Hille 1984; Hodgkin and (96% in Fig. 5), whereas the very slow cells have mostly the Huxley 1952). Here too the cells were pooled after a com- slower inactivating type channel (99% in Fig. 5), and the parison of fast and slow groups did not reveal a significant intermediate cells have a mixture of the two types (49% difference in the activation curves. slow and 5 1% fast in Figure 5 ). The difference in the steady- state inactivation properties between fast and slow cells Voltage dependence of inactivation (Fig. 5) is different from that described by Hudspeth and Inactivation was also studied in three groups of cells as Lewis( 1988),whoreportedashifiof-lOto-15mVinthe shown in Figure 5. With the use of TABLE CURVE V3.01 voltage dependence of activation and inactivation of Z* software from Jandel Scientific, a separate single Boltz- over time. They attributed this shift in inactivation to a

Holding C-dAjng

35cc 7 A

Prcpulsc Duration P~ul88 Dwctlcn C-C4QOOmt 0-04O!Xms A-A 1000ms ~-4 lGOOm8 A-A Mom3 A-A 4cCrr.S 0-0 1OOms 0-0 looms 0-O :Oms 0-O 1Oms

I 9 L 0’ 1L 1 1I t i of . . 4 -90 -80 30 -60 -50 -io -30 -90 -80 30 -60 -50 -40 -30 Prepulse Voltage (mV) Prepulse Voltage (mV) FIG. 6. Removal of inactivation from inactivating channels depends on both duration and amplitude of the cell’s potential before activation. Cells are clamped to 0 mV for long periods ( 1 to 3 s) to inactivate all the A-type type channels, and then they are clamped to a potential between -30 and -90 mV for conditioning periods of 10, 100,400, 1,000, or 4,000 ms before a voltage step to +30 mV for 120 ms (the test pulse). The cells were held close to their 0 current potential ( -60 mV) for 2 s between trials. The responses to slow cells are depicted in A and those to fast cells in B. 1648 NORRIS, RICCI, HOUSLEY, AND GUTH 1.200 O-Wow Cells (n=4) Donnan potential and as seen by Hudspeth and Lewis *-aFast Cells (n=7) ( 1988). Thus the differences described here are not the 1 .ooo same as those seen in Hudspeth and Lewis ( 1988) and ap- d pear to be of physiological significance, not due to record- 0.800 :: / i-i ing artifact. E 0.600 / 5 Time dependence of inactivation / / 0.400 I ~ The removal of inactivation of the A-type current is both 0.200 /I voltage and time dependent. This was studied with the paradigm described in the legend of Fig. 6. As inactivation of 0.000 1 I the Z* channels was removed by successively longer periods 1 10 100 1000 lE4 lE5 at membrane prepulse potentials more negative than -50 Prepulse Duration (ms) * mV, larger transient currents were produced during the step

l Prepulse potential = -9OmV to the +30 mV test pulse ( Fig. 6). This paradigm covering seven conditioning membrane potentials (which were ran- FIG. 7. Normalized peak inactivating current as a function of prepulse duration for a -90 mV prepulse. Data were obtained with the protocol of domized) and five durations for each conditioning pulse Figure 6 with the prepulse potential fixed at -90 mV and with an addi- was presented successfully to 10 cells (5 fast and 5 slow tional 3 durations (40, 10,000, and 20,000). types) l Activation of the Z* was noticeable in a few cells after dissipation of a Donnan potential across the tip of the elec- holding the membrane potentials at -40 mV for several trode. Our data, which shows a >50 mV difference between seconds. However, on average, significant removal of inac- fast and slow components cannot be explained as a dissipa- tivation of Z* occurred after 1 s at -50 mV. As the mem- tion of a Donnan potential. On the basis of a Donnan equi- brane was held at more negative potentials, the time re- librium, an internal potassium concentration of >200 mM quired to achieve a similar level of removal of inactivation would be predicted to account for such a large shift (an (enabling of the Z* channel) decreased. In several cells, it unlikely possibility). Also no differences were seen in the was possible to show that maximal removal of inactivation activation properties, as would be predicted by a shift in a of Z* occurred at potentials of -80 mV and more negative

FAST CELL SLOW CELL

r. -6OmV T - -

FIG. 8. Responses of fast and slow cells to 4-aminopyridine (4-AP). A: control current responses are shown with the voltage-clamp protocol inset in the center (similar to that used in Fig. 1A). B: current responses to the same protocol are shown after 10 mM 4-AP was added to the medium bathing the cells. C: responses in B subtracted from the responses in A and thus represents the currents that are blocked by 4-AP. The currents blocked by 4-AP have characteristics similar to those of both the fast- and slow-channel types separated by voltage-clamp protocol subtractions, suggesting that both the fast and slow channels are of the A-type. INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS 1649

*;$dp-” but is sensitive to block by 4-AP (Hudspeth and Lewis 0 ;t: ‘A 1988 ) . We therefore decided to test 4-AP in our prepara- 0.800 0 ‘0 t t1 1 O-0 VOLTAGE SUBTR. P$ 01 tion. The results are shown in Fig. 8. The results in Fig. 8C e- / 0 ‘r/r show that 4-AP blocks both fast- and slow-type channels. x 0.600 -* 4AP SUBTR. 0 0 r4 f/l With the currents as derived in Fig. 8C and a reversal poten- u #La sz I tial of -77 mV (Fig. 3)) conductance curves were calculated for each cell studied with 4-AP. These are normalized, 3 averaged, and plotted in Fig. 9 (0). For comparison, the 0.200 -- analogous, normalized conductance from Fig. 4 is also plotted on this figure (0). Thus, although the currents used were derived differently [one method with a differential -0.200 ! I I 1 ,I 1 1 prepulse protocol and the other with a pharmacological (4- -140 -100 -60 -20 20 60 100 AP) protocol], the activation conductances appear similar. Membrane Voltage Kinetic analysis of the 4-AP-sensitive currents give results FIG. 9. Normalized activation conductance was calculated for the similar to those obtained from the voltage clamp subtracted currents blocked by 4-AP as derived by the subtraction method of Figure 7 data suggesting that the inactivating currents are of the A- ( l ). Fast and slow cells were pooled because there was no difference be- type (Fig. 2). 4-AP results were not completely satisfactory. tween them. For comparison, the normalized activation conductance as The nonspecific actions of the drug made further character- shown in Figure 4 (prepulse voltage subtraction method) is also depicted (o ). Although the plots are similar, significant heterogeneity exists be- ization impossible. Apparently the drug increased a nonin- tween isolation procedures. This is most likely due to the nonspecific ac- activating component. The differences between the voltage- tions of 4-AP. current plots may be a reflection of the nonspecific action of the drug. after - 10 s. However, most cells could not withstand such It should also be noted that there are some inactivating treatment. Figure 6 also shows that fast and slow cells are currents that are not blocked by 4-AP (Fig. 8B) and in Fig. different in their responses to various prepulses. Figure 7 is 1 there are some inactivating currents that are not elimi- a plot showing how the inactivating currents vary with pre- nated with a -20 mV prepulse. We are not sure if any of pulse duration at one particular prepulse voltage. Again, these latter inactivating currents are of the A-type. How- differences between fast and slow cells can be seen. The data ever, the 4-AP data does suggest that the channels we de- suggests that the slow channels are more temporally depen- scribe as being fast and slow are of the A-type. dent. Antagonism of ,4-type channels by 4-AP Culcium dependence The inactivating outward potassium current seen in hair Protocols identical to those in Fig. 1 were used to test the cells is not sensit&e to 10 mM TEA (Housley et al. 1989) dependency of I* on external calcium. One example of the

-60 mV

c = A-B

A-A B-B C-C FIG. 10, Apparent lack of dependency of the inactivating channels on calcium. Current traces in the top TOWare control responses to the same voltage-clamp prepulse protocols as are used in Figure 1. This is a fastcell. The responses in the middle row are to the same protocols in the same cell after the external medium bathing the cells was replaced with one in which calcium was replaced by magnesium. The traces in the bottom YOWare those in the middle TOWsubtracted from those in the upperTOW. Slow cells responded in a like manner. 1650 NORRIS, RICCI, HOUSLEY, AND GUTH

h\ I J -1m

B 70 mV L-10 ms

FIG. 1 1. Responses to voltage- and current-clamp protocols in the same cell. A: current responses are derived with the same prepulse protocol as in Fig. 1A. This is classified as a fast cell. B: cell is held at 0 current and then clamped to a depolarizing current level. The voltage of the cell at 0 current was -50 mV. C: voltage responses to similar current-clamp steps are changed by the presence of a hyperpolarizing current prepulse. resulting raw data is shown in Fig. 10. Some of the non-A- dampening of the response may also be due to the action of type currents (middle column) are reduced or blocked, but the inactivating currents. Similar current-clamp responses none of the A-type inactivating currents (right-hand col- were seen in slow cells. The main difference was that the umn) were affected by the lack of calcium in the external rise to the second plateau was much slower. medium. As studied with this protocol, neither the fast nor the slow A-type channels (n = 6) appeared to be calcium DISCUSSION dependent. Two types of A-channels Current-clamp responses We have attempted to demonstrate by kinetic analysis, steady-state activation and inactivation and susceptibility The voltage-clamped current responses of a cell, which to 4-AP that there are two types of A-channels in hair cells were typical of many of the fast cells studied, are shown in isolated from the XC of R. pipiens. The following are used Fig. 11 A. This voltage-clamp protocol used the 80 ms, to describe A-channels: I ) voltage-sensitive, 2) rapidly acti- - 130 mV prepulse as described for Fig. 1 A. vating, 3) rapidly inactivating, and 4) sensitive to 4-AP In the current clamp responses of Fig. 11, B and C, the (Figs. 1,2, and 8). The separation into fast and slow catego- cell depolarized to a relatively constant potential at the low- ries is first seen in the kinetic analysis of Fig. 2. In addition est current steps. At the higher current steps, the cell contin- to the kinetic analysis, studies of the inactivating curves also ued to depolarize very slowly until the voltage suddenly show two different types of A-channels (Fig. 5 ). When the stepped to a much higher plateau. This plateau occurred effect of increasing the amplitude and duration of the pre- earlier at higher current steps. The voltage responses to the pulse used to enable the inactivating channels was exam- depolarizing current clamp steps in Fig. 11 C have the same ined ( Figs. 6 and 7 ), we again found evidence for two types general shape as in Fig. 11 B except that the step to the of channels. The differences found between the channels higher plateau is significantly delayed. The most likely were in the steady-state inactivation properties and in the cause of this delay is the recruitment of the inactivating kinetics of inactivation. channels by the hyperpolarizing current pulse. Also the The possibility that the apparent differences between the oscillatory-like wave at the beginning of the voltage-re- fast and slow channels was due to an error in the voltage sponse trace is narrower and taller in Fig. 11 C as compared protocol subtraction method was addressed pharmacologi- with Fig. 11 B. Similar findings were reported by Sugihara cally. It is possible that a slowly inactivating delayed recti- and Furukawa ( 1989). The sharpening and decreased fier type of channel was contaminating the subtraction tech- INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS 1651 nique. This type of conductance was described in semicir- A-current is dominant cular canal hair cells by Correia et al. ( 1989). 4-AP was used to selectively antagonize the rapidly inactivating A- In hair cells from the SCC of R. pipiens, the largest most type channels. The action of 4-AP (Fig. 8) on the cells rein- dominant current is clearly of the A-type (Fig. 1). Although forces the original hypothesis of two A-type channels and it is possible that a holding period at -60 mV followed by validates the voltage protocol subtraction method. When the prepulses used may not have completely removed inac- the responses of the cells after 4-AP application were sub- tivation (in the - 130 mV prepulse case) or completely in- tracted from those before 4-AP application, the currents activated the Z* (in the -20 mV prepulse case), the exam- blocked by 4-AP are revealed, and both types of A-currents ples demonstrate that these protocols allow effective isola- can be seen with characteristics as previously described with tion of the 1* (Fig. 1). It is also clear that the amount and a prepulse protocol (Fig. 8). Although further studies were character of the inactivating currents varies among hair planned with 4-AP, the nonspecificity of the drug, as re- cells isolated from the SCC. The resting membrane poten- viewed in Choquet and Korn ( 1992), made further investi- tial for the cells studied in this paper was between -45 and gations with 4-AP ambiguous. -70 mV. This means that somewhere between 20% of the Apparently, the reversal potential (Fig. 3) and the nor- slow channels and 50% of the fast A-channels are enabled malized activation conductance (Fig. 4) are the same for and can be activated from a resting potential. These values both fast and slow channels. However, the slower A-chan- are based on the steady-state inactivation plots described nels carry larger currents at a given membrane voltage as earlier. Although these currents can be activated from rest, compared with the faster A-channels. It is not possible to only 2- 10% may be on at the resting membrane potential, tell from this data whether the greater current is the result of based on the steady-state activation properties. Therefore more channels per cell or more current per channel. the inactivating currents do not appear to be involved in On the basis of the evidence presented here, we are sug- establishing the resting membrane potential of the semicir- gesting that the major inactivating conductances are of the cular canal hair cells. A-type. Two A-type conductances have been described, fast and slow. The differences between the two are the kinetics Current-clamp responses of inactivation and the steady-state inactivation properties that leave a higher proportion of the fast channels enabled In the current-clamp responses of Fig. 11, the differences at rest. noted between Fig. 11, A and B, are apparently caused by The idea that there can be two forms of 1A in one cell, one the enabling of A-channels in Fig. 11 C. Sugihara and Furu- with a fast and one with a slow inactivating component, is kawa ( 1989) have reported similar findings in hair cells of not new. Rudy ( 1988) suggested that close examination of the goldfish sacculus, which they also attribute to the A- data presented by Gustafsson et al. ( 1982), Halliwell et al. currents. Puil et al. ( 1989) also present evidence that the ( 1986), and Boyett ( 198 1) will reveal two transient out- A-current can significantly alter membrane oscillations in ward potassium currents that fit the classification of A-type. trigeminal root ganglion neurons. Hair cells are capable of One has fast inactivation characteristics and the other ex- generating spike-like activity (Evans and Fuchs 1987; alli- hibits slower inactivation characteristics. Greene et al. gator cochlea), (Hudspeth and Corey 1977-frog), (Art ( 1990) also clearly describe in detail fast and slow A- and Fettiplace 1987; turtle cochlea), and (Fuchs and Mann currents in histaminergic neurons from the rat hypothal- 1986; chick cochlea). The delay between the start of the amus. current step and the sharp voltage rise to a final plateau seen The differences observed between the fast and slow chan- in Fig. 11 appears to be the hair cell equivalent of spike-la- nels are localized to the inactivation properties. The steady- tency interval in neurons. Angelaki and Correia ( 199 1) state inactivation properties, including the voltage of half- have suggested that the membrane oscillations of semicir- maximal conductance as well as the slope function (a re- cular canal are generated by the interaction between a de- flection of the gating charge), are dramatically different. layed rectifier type of conductance and an A-type conduc- The kinetics of inactivation is also different. Activation, ion tance. Preliminary evidence here suggests that the fre- selectivity, and pharmacological sensitivity were not differ- quency of oscillation is not regulated by the A-current, ent. These results might then suggest that the differences rather the damping function (Q) is modulated. between the channels are localized to the inactivation gate or the portion of the protein responsible for inactivation. Comparison with other hair cells For Al-shaker channels this site has been localized to the NH,-terminal (Hoshi et al. 1990; Zagotta et al. 1989; Za- Although the 1A recorded by Hudspeth and Lewis ( 1988) gotta and Aldrich 1990). A ball and chain type of mecha- in frog saccular hair cells is unlikely to be active during nism has been described in which the amino terminus acts depolarizing changes from physiological resting membrane as a tethered inactivation particle that can block the inter- potentials, these authors suggest “Z* may become activated nal mouth of the channel ( Hoshi et al. 199 1; Zagotta et al. by a rapid depolarization after a long hyperpolarization, 1989; Zagotta and Aldrich 1990). It is possible, although such as might result from mechanical stimulation following speculative, that the differences between the fast and slow a burst of activity of the cell’s efferent nerve supply” (after channels are due to differences in the NH,-terminal of the Art et al. 1984; Murrow and Fuchs 1990) or from depolar- protein. Investigation of what regulates the preferential ex- izing mechanical stimulation after a strong hyperpolarizing pression of one type over the other may lead to important mechanical stimulation. Murrow and Fuchs ( 1990) have information regarding signal processing in the vestibular demonstrated that the A-type channels are found in short end organs. hair cells as opposed to tall hair cells in the basilar papilla. 1652 NORRIS, RICCI, HOUSLEY, AND GUTH No particular cell type in the frog appears to be dominated suggest that the fast inactivating current is involved in mod- by the A-type conductance. ulating the damping function of the membrane voltage Studying vestibular hair cells isolated from the chick, Oh- oscillation. mori ( 1984) did not record the presence of an A-current. In these cells, the major outward K current was an IK(ca). Like- We acknowledge the helpful criticism of this manuscript by Drs. C. W. wise, Lewis and Hudspeth ( 1983) reported that the largest Clarkson and G. G. Schofield. A special thanks goes to A. Alexander for current produced by depolarization of hair cells isolated outstanding technical support of this research. The authors are grateful for support provided by National Institutes of from the saccule of the bullfrog was the Ca*’ activated K+ Health Grants NS-2205 1 and DC-OO303-05 and the Southern Hearing and current. Lang and Correia ( 1989) recorded an A-type Speech Foundation. The New Zealand Deafness Research Foundation current in pigeon semicircular canal hair cells. However, in and the Medical Research Council of New Zealand are also thanked for the pigeon, the A-type current was only weakly active when assistance given to G. D. Housley. Present address of G. D. Housley: Dept. of Physiology, University of membrane depolarizations occurred after preconditioning Aukland, Private Bag, Aukland, New Zealand. pulses between -70 and -50 mV and was totally inactive Address for reprint requests: C. H. Norris, Dept. of Otolaryngology, after preconditioning pulses more positive than -40 mV. Head and Neck Surgery, Tulane University School of Medicine, 1430 Tu- Rennie and Ashmore ( 199 1) also report on A-currents in lane Ave., New Orleans, LA 70112. type II semicircular canal hair cells isolated from the guinea Received 7 May 199 1; accepted in final form 18 June 1992. pig. The current is a smaller proportion of the total current as compared with SCC cells from the frog and is only found REFERENCES in selected cells. Thus, semicircular canal hair cells of different species and from different organs may differ from one ADAMS, P. R., BROWN, D. A., AND CONSTANTI, A. M-currents and other another in regard to important electrophysiological proper- potassium currents in bull-frog sympathetic neurones. J. Physiol. Lond. 330: 537-572, 1982. ties. The differences found between different species is inter- ANGELAKI, D. AND CORREIA, M. Models of membrane resonance in pi- esting and future work will be geared toward determining geon semicircular canal type II hair cells. Biol. Cybern. 65: l- 10, 199 1. the significance of these differences. ART, J. J., CRAWFORD, A. C., AND FETTIPLACE, R. Electrical resonance Although hair cells from the SCC will show some electri- and membrane currents in turtle cochlear hair cells. Hear. Res. 22: 3 l- cal resonance, that is, a membrane voltage oscillation re- 36, 1986. ART, J. J. AND FETTIPLACE, R. Variation of membrane properties in hair sults from the injection of current (Correia et al. 1989; Ren- cells isolated from the turtle cochlea. J. Physiol. Lond. 385: 207-242, nie and Ashmore 199 1 ), this response is not nearly as ro- 1987. bust as has been seen in other hair cell systems (i.e., ART, J. J., FETTIPLACE, R., AND FUCHS, P. A. Synaptic hyperpolarization Crawford and Fettiplace 198 1). Angelaki and Correia and inhibition of turtle cochlear hair cells. J. Physiol. Lond. 356: 525- 550, 1984. ( 199 1) demonstrated that the interaction between a de- BAIJMANN, A., KRAH-JENTGENS, I., MULLER, R., MULLER-HOLTKAMP, layed rectifier and the A-type current could result in a F., SEIDEL, R., KECSKEMETHY, N., CASAL, J., FERRUS, A., AND PONGS, membrane voltage oscillation of low Q and low frequency. 0. Molecular organization of the maternal effect region of the Shaker The properties of the fast inactivating current reported here complex of Drosophila: characterization of an IA channel transcript with homology to vertebrate Na+ channels. Eur. Mol. Biol. Organ. J. 6: suggest that it may be involved in the hair cell response to a 34 19-3429, 1987. current injection; however, the slow conductance would re- BELLUZZI, O., SACCHI, O., AND WANKE, E. A fast transient outward quire an initial hyperpolarization for it to be enabled. Fu- current in the rat sympathetic neurone studied under voltage clamp ture experiments are being directed toward characterizing conditions. J. Physiol. Lond. 358: 9 108, 1985. the frog semicircular canal hair cell voltage response proper- BOYETT, M. R. A study of the effect of the rate of stimulation on the transient outward current in sheep cardiac purkinje fibers. J. Physiol. ties, with special emphasis on the role of the inactivating Lond. 319: l-22, 1981. conductances. CATTERALL, W. A. Voltage-dependent gating of sodium channels: corre- lating structure and function. Trends Neurosci. 9: 7- 10, 1986. CHOQUET, D. AND KORN, H. Mechanisms of 4-Aminopyridine action on Possible significance of the inactivating currents the voltage-gated potassium channels in lymphocytes. J. Gen. Physiol. Several roles may be suggested for the A-currents in ves- 99: 2 17-240, 1992. CONNOR, J. A. AND STEVENS, C. F. Voltage clamp studies of a transient tibular sensory cells. They could act as a hyperpolarizing outward membrane current in gastropod neural somata. J. Physiol. buffer that serves to rapidly oppose transductional depolar- Land. 2 13: 2 l-30, 197 1. izations of the hair cell membrane potential. It is important CORREIA, M. J., CHRISTENSEN, B. N., MOORE, L. E., AND LANG, D. C. to consider that this putative buffer then turns itself off by Studies of solitary semicircular canal hair cells in the adult pigeon. I. Frequency and time-domain analysis of active and passive membrane inactivating, thus allowing the cell to respond to prolonged properties. J. Neurophysiol. 62: 924-934, 1989. stimuli such as are appropriate for vestibular organs. They CRAWFORD, A. C. AND FETTIPLACE, R. An electrical tuning mechanism in can act to block out responses to sudden vibratory move- turtle cochlear hair cells. J. Physiol. Lond. 3 12: 377-412, 198 1. ments of the perilymph/endolymph fluids that might be EVANS, M. G. AND FUCHS, P. A. Tetrodotoxin-sensitive, voltagedepen- dent sodium currents in hair cells from the alligator cochlea. Biophys. J. important for auditory responses by other areas of the 52: 649-652, 1987. inner ear. FUCHS, P. A., EVANS, M. G., AND MURROW, B. W. Calcium currents in In summary, frog semicircular canal hair cells appear to hair cells from the cochlea of the chick. J. Physiol. Lond. 429: 553-568, have two distinct A-type conductances. The channel types 1990. differ in steady-state inactivation properties as well as in the FUCHS, P. A. AND MANN, A. C. Voltage oscillations and ionic currents in hair cells isolated from the apex of the chick’s cochlea. J. Physiol. Lond. kinetics of inactivation. Hair cells vary in the relative pro- 371: 31P, 1986. portion of each channel type found. The physiological sig- GREENBLATT, R. E., BLAT, Y., AND MONTAL, M. The structure of the nificance remains to be studied, but preliminary data would voltage-sensitive sodium channel. Inferences derived from computer- INACTIVATING CURRENTS IN VESTIBULAR HAIR CELLS 1653

aided analysis of the Electrophorus electricus channel primary structure KAMB, A., TSENG-CRANK, J., AND TANOUYE, M. Molecular characteriza- FEBS Lett. 193: 125-134, 1985. tion of Shaker, a drosophila gene that encodes a potassium channel. Cell GREENE, R. W., HASS, H. L., AND REINER, P. B. Two transient outward 50: 405-413, 1987. current in histamine neurones of rat hypothalamus in vitro. J. physiol. LANG, D. G. AND CORREIA, H. J. Studies of solitary semicircular canal hair Land. 420: 149- 163, 1990. cells in the adult pigeon. II. Voltage-dependent ionic conductances. J. GUSTAFSSON, B., GALVAN, M., GRAFE, P., AND WIGSTROM, H. Transient Neurophysiol. 63: 935-945, 1989. outward current in a mammalian central neuron blocked by 4-amino- LEWIS, R. S. AND HUDSPETH, A. J. Voltage- and ion-dependent conduc- pyridine. Nature Land. 299: 252-254, 1982. tances in solitary vertebrate hair cells. Nature Lond. 304: 538-54 1,1983. GUTH, P. S., NORRIS, C. H., AND SEWELL, W. F. Primary afferent trans- LUKASIEWCZ, P. AND WERBLIN, F. A slowly inactivating potassium mission in acousticolateralis organs. In: Auditory Biochemistry, edited current truncates spike activity in ganglion cells of tiger salamander by D. G. Drescher. Springfield, IL: Charles C. Thomas, 1985, p. 42-49. retina. J. Neurosci. 8: 4470-445 1, 1988. HALLIWELL, J. V., OTHMAN, L. B., PELCHEN-MATTHEWS, A., AND DOLLY, MACKINNON, R., REINHART, P. H., AND WHITE, M. M. Charybdotoxin J. 0. Central action of dendrotoxin selective reduction of a transient block of shaker K channel suggests that different types of potassium K-conductance in hippocampus and binding to localized acceptors. channels share common structural features. Neuron 1: 997- 100 1, 1988. Proc. Natl. Acad. Sci. USA 83: 493-497, 1986. MURROW, B. W. AND FUCHS, P. A. Preferential expression of transient HAMILL, 0. P., MARTY, A., NEHER, E., SAKMANN, B., AND SIGWORTH, potassium current (IA) by ‘short’ hair cells of the chicks cochlea. Proc. F. J. Improved patch-clamp techniques for high-resolution current re- R. Sot. Lond. B Biol. Sci. 242: 189- 195, 1990. cording from cells and ceil free membrane patches. Pfluegers Arch. 39 1: NORRIS, C. H. AND GUTH, P. S. Buffers may have pharmacological ac- 85-100, 1981. tions. Trends Pharmacol. Sci. 6: 15, 1985. HILLE B. Ionic Channels of Excitable Membrane, Sunderland, MA: OHMORI, H. Studies of ionic currents in the isolated vestibular hair cells of Sinauer Associates, 1984, p. 53-57. the chick. J. Physiol. Lond. 350: 56 l-58 1, 1984. HODGKIN, A. L. AND HUXLEY, A. F. A quantitative description of mem- PAPAZIAN, D., SCHWARTZ, T., TEMPEL, B., JAN, Y., AND JAN, L. Cloning brane current and its application to conduction and excitation in nerve. of genomic and complementary DNA from Shaker, a putative potas- J. Physiol. Land. 117: 500-544, 1952. sium channel gene from Drosophila. Science Wash. DC 237: 749-753, HOSHI, T., ZAGOTTA, W., AND ALDRICH, R. Biophysical and molecular 1987. mechanisms of Shaker Potassium channel inactivation. Science Wash. PITCHFORD, S. AND ASHMORE, J. F. An electrical resonance in hair cells of DC 250: 533-538, 1990. the amphibian papilla of the frog. Hear. Res. 27: 75-83, 1987. PUIL, E., MIURA, R. M., AND SPIGELMAN, I. Consequences of4-aminopyr- HOSHI, T., ZAGOTTA, W., AND ALDRICH, R. Two types of inactivation in Shaker K channels: effects of alterations in the carboxy-terminal region. idine applications to trigeminal root ganglion neurons. J. Neurophysiol. Neuron 7: 547-556, 1991. 62: 8 10-820, 1989. RENNIE, K. J. AND ASHMORE, J. F. Ionic currents in isolated vestibular HOUSLEY, G. D., NORRIS, C. H., AND GUTH, P. S. Electrophysiological properties and morphology of hair cells isolated from the semicircular hair cells from the guinea pig crista ampullaris. Hear. Res. 5 1: 279-292, canal of the frog. Hear. Res. 38: 259-276, 1989. 1991. RUDY, B. Diversity and ubiquity of K channels. Neuroscience 25: 729- HUDSPETH, A. J. The ionic channels of a vertebrate hair cell. Hear. Res. 749, 1988. 22: 21-27, 1986. SUGIHARA, I. ANLI FURUKAWA, T. Morphological and functional aspects HUDSPETH, A. J. AND COREY, D. P. Sensitivity, polarity and conductance of two different types of hair cells in goldfish sacculus. J. Neurophysiol. change in the response of vertebrate hair cells to controlled mechanical 62: 1330- 1343, 1989. stimuli. Prod. Natl. Acad. Sci. USA 74: 2407-24 11, 1977. THOMPSON, S. H. Three pharmacologically distinct potassium channels in HUDSPETH, A. J. AND LEWIS, R. S. Kinetic analysis of voltage- and ionde- molluscan neurones. J. Physiol. Lond. 265: 465-488, 1977. pendent conductances in saccular cells of the bull-frog, Rana Catesbei- TIMPE, L. C., SCHWARTZ, T. L., TEMPLE, B. L., PARAZIAN, D. M., JAN, ana. J. Physiol. Lond. 400: 237-274, 1988. Y. N., AND JAN, L. Expression of functional potassium channels from IVERSON, L. E. AND RUDY, B. The role of the divergent amino and car- shaker cDNA in Xenopus oocytes. Nature Lond. 33 1: 143-145, 1988. boxy1 domains on the inactivation properties of potassium channels ZA~~A, W. AND ALDRICH, R. Voltage-dependent gating of shaker A- derived from the shaker gene of Drosophila. J. Neurosci. 10: 2903- type potassium channels in Drosophila muscle. J. Gen. Physiol. 95: 29- 2916, 1988. 60, 1990. JAN, Y. N., JAN, L. Y., AND DENNIS, M. J. Two mutations of synaptic ZAGOTTA, W., HOSHI, T., AND ALDRICH, R. Gating of the single shaker K transmission in drosophila. Proc. R. Sot. Lond. B Biol. Sci. 198: 87- 108, channels in drosophila muscle and in oocytes injected with shaker mRNA. Proc. Natl. Acad. Sci. USA 86: 7243-7247, 1989.