Journal of Neurophysiology

Effects of Quinine on the Excitability and Voltage-Dependent Currents of Isolated Spiral Ganglion Neurons in Culture

Xi Lin, Shanping Chen, Daniel Tee


Lin, Xi, Shanping Chen, and Daniel Tee. Effects of quinine on the excitability and voltage-dependent currents of isolated spiral ganglion neurons in culture. J. Neurophysiol. 79: 2503–2512, 1998. This work examined how quinine, a drug that induces both hearing loss and tinnitus, interfered with the excitability of spiral ganglion (SG) neurons in cultures. The membrane potential changes and the modification of the action-potential waveform induced by quinine were studied in SG neurons under current clamp. The effects of the drug on voltage-dependent currents in SG neurons were also investigated by the voltage-clamp method. Quinine did not appreciably affect either resting membrane potentials or input resistance at rest. However, action potentials fired by SG neurons were significantly broadened by the presence of quinine. With higher concentrations of quinine (>20 μM), the amplitude of action potentials was also reduced. Voltage-clamp results demonstrated that quinine primarily blocked the whole cell potassium currents (I K) in a voltage-dependent manner. Up to 100 μM of quinine did not appreciably block I K evoked by a test pulse to −35 mV. In contrast, I K was significantly reduced with more positive test pulses, e.g., the concentration needed to obtain 50% inhibition (IC50) was 8 μM for a test pulse to 65 mV. At higher concentrations (>20 μM), quinine also reduced the size of sodium currents (I Na) in a use-dependent manner, while leaving calcium currents (I Ca) relatively unaffected. Compared with the potency of quinine's effects on other targets in the inner ear, the relatively low IC50 and the voltage-dependent nature of quinine inhibition on I K suggested that its modulation of the waveform and threshold of action potentials of SG neurons probably was primarily responsible for its ototoxic effects. From the point of view of how neural signaling process is affected by the drug, quinine-induced tinnitus may be explained by its broadening of action potentials while the drug's inhibition on I Na may result in hearing loss by making the conversion from excitatory postsynaptic potentials to the generation of action potentials more difficult.


Quinine is the major alkaloid derived from the bark of the cinchona tree (Webster 1985). Clinically, it has been widely used to treat malaria and nocturnal leg cramps (Jung et al. 1993; Wolf et al. 1992). Commercially, 40% of quinine production is used in tonic beverages as a bitter agent (Colley et al. 1989; Jung et al. 1993). Overdoses of quinine have widespread side effects, which is termed as cinchonism. A therapeutic dosage of quinine is capable of inducing reversible high-frequency hearing loss (Roche et al. 1990). Higher doses result in reversible deafness, vertigo, and tinnitus (Jung et al. 1993; Wolf et al. 1992). Cinchonism, however, is not limited to the auditory system. Other clinical manifestations include some general neurological effects: nausea, headache, and visual loss (Smilkstein et al. 1987; Wolf et al. 1992). In the most severe cases, patient death can result from cardiotoxicity (Bodenhamer and Smilkstein 1993; Wolf et al. 1992).

Past investigations suggest that quinine ototoxicity originates from the auditory periphery (Puel et al. 1990). Direct perfusion of quinine into the perilymph shifts the intensity function of the compound action potential (CAP) to the right. It also reduces cochlear microphonics (CM) and summating potential, while leaving the endocochlear potential intact (Puel et al. 1990). High doses of quinine (in the mM range) directly affect the mechanical tuning of the basilar membrane (Karlsson et al. 1991b) and the cupula of the fish lateral line (van Netten et al. 1994). It is also observed that the length of isolated outer hair cells (OHCs) is changed by quinine (Karlsson and Flock 1990). In guinea pigs exposed to quinine, vasoconstriction in the cochlear microcirculation has been observed (Smith et al. 1985). Because quinine induces tinnitus in human subjects (Jung et al. 1993; Wolf et al. 1992), it has also been used to construct an animal model of tinnitus (Jastreboff et al. 1991).

Several hypotheses have been proposed to explain quinine-induced ototoxicity, e.g., blocking action of the voltage-gated potassium channels of hair cells (HCs) (Jung et al. 1993; Puel et al. 1990; Wolf et al. 1992) based on reports that the drug inhibits various types of potassium channels in nonauditory cells (Fatherazi and Cook 1991; Glavinovic and Trifaro 1988), or damage to the subsurface cisternae in the OHCs (Karlsson and Flock 1990; Karlsson et al. 1991a). However, the cellular mechanism of quinine ototoxicity remains unclear. In this paper, the effect of quinine on the membrane of isolated spiral ganglion (SG) neurons in culture was examined by using the whole cell voltage-clamp technique (Hamill et al. 1981). Results indicated that increasing concentration of quinine affected the excitability of SG neurons by first broadening and then reducing the amplitude of action potentials. The resting membrane potential of the neurons, however, was not affected due to the voltage-dependent nature of the drug action. Results obtained from voltage-clamp experiments revealed that quinine blocked both potassium currents (IC50 around 10 μM) and sodium currents (IC50 around 85 μM), while leaving the calcium current relatively unaffected. The voltage-dependent nature of the drug action and the relatively low concentrations of quinine required to affect ion channels responsible for generating action potentials suggested that SG neurons may be the primary target of quinine's ototoxic effects. A neural signaling model based on how action-potential duration affects Ca2+ entry and transmitter release is presented in the discussion to explain quinine's ability to induce both hearing loss and tinnitus.


Cell culture of isolated SG neurons

Cell culture methods were the same as those used previously for gerbil SG neuron cultures (Lin 1997). Briefly, cochlear tissues dissected from newborn to 3-day postnatal mouse pups (P0–P3, CD-1 strain, Harlan Sprague Dawley, Indianapolis, IN) were used to obtain SG neurons. The tissues were cut into small pieces and treated with an enzymatic solution (collagenase type IV, 0.5 mg/ml and trypsin, 2.5 mg/ml) at 37°C for 30 min. They were then centrifuged at 2,500 rpm, and supernatant was replaced with culture medium containing 90% Dubelcco's modified eagle medium (DMEM), 10% fetal bovine serum (FBS), penicillin (50 U/ml) and streptomycin (50 μg/ml). Tissue was gently pipetted to mechanically dissociate cells that were subsequently plated at a density of two cochleae per 35-mm dish in a humidified CO2 incubator (5%) at 37°C. Cultures were maintained for no more than 24 h for the purposes of the current study. Adult SG neurons (30–40 g) were obtained acutely without culture. After treating the adult modiolus with the same enzymatic solution described above for 15 min, isolated SG neurons were obtained by pulling apart the tissue sample in random directions with two fine glass pipettes. Most cell bodies of the adult SG neurons were myelinated. Performing patch-clamp type of recordings from those neurons were not possible until some of the cell bodies spontaneously lost their myelin sheathes in ∼1–2 h after being kept in the in vitro environment. All animal care procedures were approved by the House Ear Institute Animal Care and Use Committee.

Electrical recordings

All experiments were carried out at room temperature (∼22°C). To minimize space-clamp problems, electrical recordings were done between 5 and 18 h after the cultures were started. Our previous results showed that the space-clamp error in SG neurons studied at this stage in culture was minimal (Lin 1997). Cultured SG neurons were easily identified by their large, round, and phase-bright cell bodies when viewed under phase-contrast optics of an inverted microscope (Axiovert 135 TV, Carl Zeiss, Jena, Germany). Their identity was further confirmed by the presence of sodium currents (I Na) and action potentials. As reported by others, the SG neurons were unmyelinated at this stage of development (Romand and Romand 1985). The morphology of cultured mouse SG neurons was similar to that previously observed in gerbils (Lin 1997); a detailed description was therefore omitted in this paper.

The whole cell recording configuration (Hamill et al. 1981) was established on SG neurons with tight seals (2–5 GΩ) formed on the cell bodies. Current (under voltage clamp) or voltage (under current clamp) signals were amplified using a patch-clamp amplifier (Axopatch 200B, Axon Instruments, Foster City, CA). Signals were filtered at 2 kHz and sampled at 100-μs intervals with the aid of an A/D and D/A converter (12 bit, Digidata 1200, Axon Instruments). For sodium current recordings, filter and sampling rate were set at 10 kHz and 10 μs, respectively. The experiments were conducted and data analyzed with the Pclamp software (version 6.0.2, Axon Instrument). Borosilicate glass capillary pipettes (World Precision Instruments, Sarasota, FL) were used. Pipettes used in voltage-clamp experiments usually had resistance of ∼1.5 MΩ when placed in the bath solution. The actual pipette access resistance was always below 3 MΩ when whole cell voltage-clamp configuration was established. At least 85% of this access resistance was compensated. Therefore the voltage-clamp error in most of the experiments was <1.4 mV (with a whole cell current of 3 nA). With larger whole cell current, the voltage-clamp error could be significant. When the whole cell current was >10 nA, the current-voltage (I-V) relations were constructed by using the corrected command voltage, which was calculated by the following equation: corrected command voltage = command voltage-current⋅pipette resistance. For current-clamp experiments, the pipette resistance was higher (between 3 and 4 MΩ) to prevent the oscillation of the amplifier. The resistance was compensated just before the recording of action potentials. Fast current-clamp mode of the Axopatch 200B amplifier was always used to record action potentials from SG neurons. Under our recording conditions, typical current-clamp speed is ∼10 μs, which is ≥50 times faster than the rise time of action potentials. Input resistance (R in) of SG neurons was measured by giving a 5-mV hyperpolarizing voltage step of 20 ms duration to neurons under voltage clamp. Steady-state currents (I) were measured, and R in was calculated by equation: R in = I/5. The action-potential width was defined as the time span during which the membrane potential was depolarized to more positive than −20 mV during an action potential.

Solutions and reagents

All solutions and drugs were purchased from Sigma Chemical (St Louis, MO). Action potentials were recorded from SG neurons bathed in the Hanks' balanced salt solution (HBSS), which contained (in mM) 137 NaCl, 0.2 Na2HPO4, 5.4 KCl, 0.4 KH2PO4, 0.8 MgSO4, 1.3 CaCl2, 5.6 glucose, and 10 N-[2-hydroxyethyl]piperazine-N′-[2-ethanesulfonic acid] (HEPES). Internal solution was (in mM) 120 KCl, 20 potassium fluride (KF), 2 NaCl, 2 MgCl2, 10 ethylene-bis(oxyethylenenitrilo) tetraacetic acid (EGTA), and 10 HEPES. All solutions were buffered to pH 7.4 with Trizma⋅OH. The pH of the solutions was not changed when the highest concentration of quinine hydrochloride (0.5 mM) was added. For I Na recordings, I K was blocked by replacing K+ with Cs+ in the pipette internal solution. For I K recordings, I Na was blocked by adding tetrodotoxin (TTX, 300 nM) to the bath solution. Nimodipine (1 μM) was also added to the bath to block calcium currents (I Ca) in both cases. Because calcium channels generally have higher permeability for barium ions (Ba2+) than for calcium ions (Fox et al. 1987), Ba2+ was used as the permeant ion in estimating I Ca. The external solution used for recording barium currents was (in mM) 90 BaCl2, 1.5 MgCl2, 7.3 glucose, 9.1 HEPES, and 0.0017 TTX. Solutions containing certain concentrations of the drug (given in the text when its uses are described) were delivered by a programmable superfusion device (ALA Scientific Instruments, Westbury, NY). Speed of the drug delivery system was checked by measuring the rate at which membrane potentials were depolarized by a solution in which NaCl in the HBSS was replaced by an equal amount of KCl (termed high K+ solution). Membrane potentials were depolarized rapidly (Fig. 1) to close to 0 mV, typically within 100 ms. In some experiments in which internally applied quinine effects on I K were examined, the drug was dissolved in the internal solution and applied to cells by the patch recording pipettes after the establishment of the whole cell recording configuration.

Fig. 1.

Effect of quinine on the resting membrane potential. The membrane potential of a spiral ganglion (SG) neuron was recorded under current clamp. The resting membrane potential of the cell was recorded because no current was injected, which was around −58 mV. Sodium currents were blocked by adding 1 μM tetrodotoxin to the bath solution. Change of the resting membrane potential induced by the presence of high K+ and quinine (at 50, 100, and 200 μM) was plotted. Black bars under the data trace represent the approximate time course of drug perfusions. Normal external bath solution [Hanks' balanced salt solution (HBSS)] was perfused in between drug applications.


Quinine does not affect the resting membrane potential and input resistance

On the basis of quinine's well-known blocking effects on potassium channels obtained from cells of nonauditory system (Fatherazi and Cook 1991; Glavinovic and Trifaro 1988), it has been proposed that one of the possible mechanisms for quinine-induced tinnitus is to excite cells in the organ of Corti by depolarizing their membrane potentials (Feldmann 1995; Jung et al. 1993). To test whether applications of quinine depolarize SG neurons, the drug effects on the resting membrane potential of SG neurons were measured from cells under current clamp (Fig. 1). In those experiments, action potentials were inhibited by adding TTX (1 μM) in the bath solution to block the I Na. Applications of a high K+ solution rapidly depolarized membrane potentials from the rest (about −58 mV) to zero, which served as a positive control for the following tests. In contrast, three subsequent applications of quinine (up to 200 μM) had little or no excitatory effects on the membrane potential. The input resistance (R in) around the resting membrane potential was not significantly changed by quinine as well. R in was 898 ± 02 MΩ (mean ± SD, n = 10) in the absence of the drug, and was 926 ± 136 MΩ, 848 ± 129 MΩ and 862 ± 98 MΩ (n = 10) in the presence of 50, 100, and 200 μM quinine, respectively. Results presented in the following section (Fig. 5) demonstrated that quinine inhibition on I K was voltage dependent; the drug only inhibited potassium conductances at membrane potentials much more depolarized from the rest.

Fig. 5.

Dose response of the quinine inhibition on the I K. Normalized current amplitude as a function of quinine concentration gave the dose response curve. Current amplitudes near the end of the 600-ms test pulses were compared. Data are presented as average (squares) ± SD (vertical bars). Two examples of the dose response curve at −35 mV (A) and −65 mV (B) test potentials are given. The smooth curve in B was the best fit to data using the function: % of control response = 100 * [1 − [drug]n/(IC50 + [drug]n)]. Parameter used in the fit was IC50 = 8.1 μM, n = 0.92. C: the IC50 is plotted as a function of the test pulse potential to show the voltage dependence of the drug inhibition.

Quinine primarily prolonged the duration of action potentials

To simulate the effect of quinine on spontaneous firing of the auditory nerve, the drug was given to SG neurons that were firing action potentials regularly. Figure 2 gives an example in which action potentials were elicited by injecting suprathreshold current pulses (5 nA in amplitude, 0.1 ms in duration) repetitively at 50 pulses/s to an SG neuron under the current clamp. Consistent with results demonstrated in Fig. 1, resting membrane potentials were not affected by quinine. On a slower time scale shown in Fig. 2, the most obvious change was a gradual decline of action-potential amplitudes caused by quinine. Time courses of the action-potential amplitude reduction were much slower than the speed of the drug delivery system, suggesting a cumulative effect of quinine on the action-potential amplitude. The effect was dose dependent and reversible. While 20 μM quinine reduced the amplitude by ∼10 mV at the steady state (Fig. 2 A), 200 μM quinine decreased the action-potential amplitude by >40 mV a few seconds after the drug application (Fig. 2 D). Effects of quinine were readily reversible, action-potential amplitudes returned to the control size within a few seconds (duration of drug application indicated by black bars above data). Shown on an expanded time scale, a more detailed comparison indicated that the action-potential duration was also dramatically broadened as the repolarization rate of the membrane potential was slowed down considerably in response to quinine applications (Fig. 3, A–D). The action-potential duration increased gradually from ∼1.2 times of the control in the presence of 20 μM quinine to >3.5 times of the control in the presence of 200 μM quinine (Fig. 3 E). If SG neurons were stimulated by a series of current pulses whose amplitude started below the threshold level, it was found that quinine elevated the threshold needed to elicit action potentials in SG neurons (data not shown).

Fig. 2.

Effects of quinine on action potentials. Action potentials were elicited by suprathreshold current pulse injections (20 pulses/s, 5 nA in amplitude, and 0.1 ms in duration) to an isolated SG neuron under current clamp. Black bars on top of data traces represent approximate time courses of quinine applications. Concentrations of the drug (A, 20 μM; B, 50 μM; C, 100 μM; D, 200 μM) were given on the top left of each bar. Data presented in this figure were obtained from the same neuron. Resting membrane potential of the neuron was stable (∼ −65 mV) during the entire course of the recording.

Fig. 3.

Detailed comparison of the waveform of action potentials revealed that it was broadened by quinine. Same data presented in the Fig. 2 were replotted on a faster time scale to construct this figure. Two segments in Fig. 2, A–D, before ( Graphic ) and during (⋅⋅⋅) drug applications were aligned and superimposed to give Fig. 3, A–D, correspondingly. Relative increase of the action-potential duration (defined arbitrarily as the time when the membrane potential was above −20 mV) as a function of quinine concentration is given in E.

Voltage-dependent sodium and potassium currents are two components responsible for generating action potentials (Hodgkin and Huxley 1952). One of many functions of the voltage-dependent calcium current is to couple action potentials to neurotransmitter release (Katz and Miledi 1967b). Similar to voltage-dependent currents recorded from gerbil (Lin 1997), whole cell currents obtained from SG neurons of mice displayed three major components. To understand how quinine affected excitability of SG neurons, effects of the drug on these currents were examined directly under the condition that currents were isolated pharmacologically in the next series of voltage-clamp experiments.

Quinine effects on the potassium currents (Ik)

General properties of I K recorded in mice were quite similar to those obtained in the gerbil (Lin 1997). For example, peak I K recorded in mice inactivated substantially at holding potentials more negative than that needed for I K activation. Accumulative inactivation of I K also resulted in a gradual broadening of action potentials elicited by a train of current injections in mouse SG neurons. Because of the similarities, those data are not shown in this paper.

The use of the fast perfusion device often allowed us to obtain whole sets of dose response data and the result of wash out in a single trial. Part of the results of the quinine inhibition on I K from one SG neuron are presented in the Fig. 4. In control tests (Fig. 4 A) a series of five voltage steps elicited outward I K that inactivated slowly during the test pulse (∼600 ms in duration). The inactivation was, however, more prominent at more positive test voltages whose time courses could be mostly described by a single exponential decay function. Subsequently, increasing concentrations of quinine were given to the neuron to obtain the dose response for the drug. The peak amplitude of I K was reduced in response to the presence of quinine in a dose-dependent manner. Inactivation rates of I K were also accelerated by quinine, especially those elicited by positive test potentials. An exponential decay function with two time constants (τshort ≅ 50–20 ms, τlong ≅ 150–300 ms) best described the I K inactivation time courses in response to test pulses at 15, 40, and 65 mV. I K recovered to levels comparable with those in the control within a few seconds when the drug was washed out (compare Fig. 4, A and F), indicating that quinine inhibition of I K was reversible. To allow the interaction between the channel and the drug to better reach the steady state, dose responses were quantified by comparing I K amplitudes near the end of the test pulses. A voltage dependence of quinine inhibition on I K was clearly revealed when the concentration needed to block 50% of controls (IC50) were calculated at the five test voltages. Amplitudes of I K in response to the test pulse at −35 mV (compare the lowest current traces in Fig. 4, A–F) were barely affected. No IC50 could be readily determined at this test potential (Fig. 5 A, n = 9). In contrast, I K obtained at more positive test pulses were affected much more dramatically (Fig. 5, B and C). IC50 declined gradually from 34.3 μM for test potential at −10 mV to 8.1 μM when test potential was increased to +65 mV (n = 9 for all cases, Fig. 5 C). When quinine was added in the internal solution, I K ran down much faster. While the size of I K was essentially constant during the first 10 min after the establishment of whole cell recording configuration in the controls, I K declined to 61 ± 8% (n = 5) of its initial amplitudes if 20 μM of quinine was included in the internal solution. This result suggested that quinine was also effective from the intracellular side.

Fig. 4.

Reversible inhibition of quinine on the I K. The outward I K of SG neurons was isolated pharmacologically and elicited from a voltage-clamped SG neuron by a series of 5 test pulses to −35, −10, 15, 40, and 65 mV. Holding potential was −100 mV. Waveform of the test protocol is given on the bottom of data traces. Data presented in this figure were obtained from the same cell. A variety of concentrations of quinine (B, 5 μM; C, 10 μM; D, 20 μM; E, 50 μM) were applied to the neuron after a control test (A) were taken. An additional test (F) was given 2 min after the drug test (50 μM quinine) to see whether the drug effect was reversible.

Quinine effect on the sodium current (INa) and barium current (IBa)

General properties of I Na are not reported in this paper because I Na recorded in mice were similar to those obtained previously from gerbils (Lin 1997). Figure 6, A–F (all data obtained from the same cell), demonstrates the reversible inhibition of various concentrations of quinine on I Na. In comparison with its inhibition on I K, a higher concentration of quinine was needed to achieve the same amount of block on I Na (Fig. 7). Calculated IC50 was around 85 μM at all test voltages (n = 7, Fig. 7 B), in sharp contrast to the voltage-dependent block by quinine on I K (Fig. 5 C).

Fig. 6.

Reversible inhibition of quinine on the I Na. The inward I Na was isolated pharmacologically and elicited by a series of 4 pulses to −20, 0, 20, and 40 mV. Time interval between pulses was 2 s. Data presented in this figure were obtained from the same cell. The holding potential was −100 mV. After a control test (A) was taken, a variety of concentrations of quinine (B, 20 μM; C, 100 μM; D, 200 μM; E, 500 μM) were delivered to the cell by superfusion. An additional test (F) was obtained 2 min after the 500 μM quinine was washed out.

Fig. 7.

Dose response of quinine blockage on I Na. Normalized current amplitude as a function of quinine concentration gave the dose response curve. A: peak current amplitudes in response to test pulses were compared. Data are presented as average (squares) ± SD (vertical bars). One example of the dose response curve obtained by comparing response elicited by the test potential to 20 mV is given. The smooth curve in A was the best fit to data using function: % of control response = 100 * [1 − [drug]n/(IC50 + [drug]n)]. Parameter used in the fit was IC50 = 88.4 μM, n = 1.1. B: the IC50 was plotted as a function of the test pulse potential to show the voltage dependence of the drug inhibition.

Similar to local anesthetics (Hille 1992), quinine inhibition on I Na was use dependent if sodium channels were opened repetitively at a fast pace. Twenty-four I Na in responses to repeated voltage pulses (given at 50 Hz) in the absence (Fig. 8 A) and presence (Fig. 8 B) of quinine are superimposed (similar results were obtained from 5 cells). In comparison with the control (Fig. 8 A), 50 μM quinine caused ∼40% of tonic block (1st response in Fig. 8 B). The inhibition was then accumulated pulse by pulse, with the first few pulses showing the largest decline of I Na. At the end of the 24th test pulse, I Na was reduced to ∼38% of the control level (Fig. 8 C).

Fig. 8.

Quinine blockage on I Na was use dependent. A: inward sodium currents obtained from a voltage-clamped SG neuron in the absence of quinine. Currents were elicited by test pulses repeated at 50 pulses/s and the I Na were superimposed. B: inward sodium currents obtained in the presence of 50 μM quinine in response to the same test pulses as that in A. Current responses were superimposed. C: normalized peak I Na (nth I Na in the presence of quinine/nth I Na response in the absence of quinine) was plotted as a function of the test pulse episode.

Barium currents (I Ba) recorded in mouse SG neurons showed typical L-type calcium current that inactivated little during the test pulse (60 ms in duration). Seven cells were tested; one typical example is given in Fig. 9. A series of test pulses (protocol illustrated on top of current traces in Fig. 9) at potentials more positive than −50 mV elicited slowly inactivating inward currents. The peak amplitude of currents increased gradually until the test potential reached 0 mV, then declined gradually and reversed direction at ∼50 mV. The presence of quinine (tested at 50 μM) did not significantly alter the I-V response (Fig. 9 C).

Fig. 9.

Effects of quinine on I Ba. The inward I Ba was isolated pharmacologically and elicited from a voltage-clamped SG neuron by using a test protocol shown above the data traces. A: a series of I Ba obtained without quinine. B: a series of I Ba obtained in the presence of 100 μM quinine. Data were obtained from the same neuron. C: voltage and the peak current amplitude relation in the absence (•) and presence (▪) of quinine.

Quinine broadened action potentials fired by adult SG neurons

Results presented so far were obtained from cultured postnatal SG neurons. To check whether adult SG neurons were similarly affected by quinine, the drug's effects were tested on cell bodies of acutely isolated adult SG neurons. Injections of current pulses of short duration (100 μs) elicited action potentials (Fig. 10 A, Graphic ), with averaged peak amplitude of 42.3 ± 4.7 mV and duration of 1.36 ± 0.1 ms (n = 5). Quinine reversibly reduced amplitude and prolonged the duration of action potentials. For example, peak amplitude and the duration of the action potential (Fig. 10 A, – – –) were changed to 23.7 ± 3.5 mV and 2.26 ± 0.2 ms, respectively (n = 5) in the presence of 50 μM quinine. The lowering of action-potential amplitude and repolarization rate probably were caused by quinine's blocking on the sodium and potassium channels, as demonstrated in the cultured postnatal SG neurons. When tested under the voltage-clamp configuration, adult SG neurons cell bodies demonstrated both the transient inward sodium current and the outward potassium currents when bathed in HBSS. The sizes of the outward current were especially large, often in excess of 20 nA. Both the sodium and potassium currents were reduced by quinine, with the outward potassium current being much more sensitive to the drug. One example of the effect of 10 μM quinine on the whole cell currents of adult SG neurons is presented in Fig. 10 B. At this concentration, the drug caused little reduction of the inward sodium current. However, the outward potassium currents were significantly reduced in a voltage-dependent manner. Although there was little change in response to the first test pulse at −33 mV, the outward current was reduced by ∼40% in response to the last test pulse at +55 mV.

Fig. 10.

Quinine effects on adult SG neurons. A: current injection elicited action potentials obtained from acutely isolated adult SG neurons in control ( Graphic ) and in the presence of 50 μM quinine (– – –). Diagram under the data traces describes the current pulse protocol, 100 μs in duration and 3 nA in amplitude. B: 5 whole cell currents recorded in response to test pulses to (clamp voltage corrected for series resistance) −33.3, −6.7, +9.7, +32.8, and + 54.7 mV. Currents obtained in the control ( Graphic ) and in the presence of 10 μM quinine (– – –) were superimposed.


Results of this study showed that, dependent on dosage, quinine had a broad spectrum of pharmacological effects on ion channels of SG neurons. Although its blocking effect on I K was expected, the sharp voltage dependence of its action was a novel finding of this study. This property of the interaction between the drug and the potassium channels made quinine essentially ineffective in affecting the resting membrane potential. Only at depolarized membrane potentials that occurred when SG neurons were firing action potentials, quinine effectively blocked I K (Figs. 4 and 5). This mode of quinine action broadened action potentials without significantly changing the resting membrane potential of cells (Figs. 1 and 2). At concentrations >20 μM, amplitudes of action potentials were also reduced. This effect was revealed to be due to the direct inhibition of I Na by quinine (Figs. 6 and 7).

The dual inhibition of quinine on both I K and I Na might provide a simple explanation for quinine's ability to induce both hearing loss and tinnitus. Inhibition of quinine on the postsynaptic I Na may make it harder for excitatory postsynaptic potentials to initiate the firing of action potentials because fewer sodium channels are available, thus providing a direct explanation for quinine-induced hearing loss. This hypothesis is consistent with the findings of Puel et al. (1990) that quinine causes a parallel shift of the CAP intensity function to the right. The search for a reasonable explanation for the etiology of tinnitus is, however, always more challenging. The following discussion attempts to present a plausible hypothesis to explain how the observed effects of quinine on the peripheral auditory targets may lead to the generation of tinnitus. More comprehensive reviews (Møller 1995) are available for those who want to get a more complete analysis on the various origins of tinnitus. Past efforts to correlate spontaneous firing of the auditory nerve and tinnitus generated ambiguous results. Increased spontaneous activities from the auditory nerve of cats were first reported by Evans' group (Evans and Borerwe 1982; Evans et al. 1981) after applying high doses of salicylate (400 mg/kg). The results, however, could not be repeated on guinea pigs (Mulheran 1990). Later work by Stypulkowski (1990) did not show any significant changes in spontaneous firing rate in cats after the animal was given salicylate. Other works that show increased activities (Kumagai 1992; Shehata-Dieler et al. 1994) have been questioned (Chen and Jastreboff 1995) because of the side effects caused by high doses of salicylate used in experiments. It seemed that, other than results obtained from neurons in some higher auditory centers in the brain stem (Chen and Jastreboff 1995; Kaltenbach and McCaslin 1996; Ochi and Eggermont 1997), no significant increase in spontaneous firing rate in the auditory nerve could be produced by tinnitus-inducing drugs at physiologically relevant dosages. This conclusion is consistent with the general finding of this study that quinine is not an excitatory agent for the peripheral auditory function.

Increased neural output of spontaneous activities of the auditory nerve need not to be solely dependent on increases in the number of spikes fired. Alternatively, it can be achieved by more reliable synaptic transmission through either presynaptic action-potential broadening (Abrams et al. 1984; Augustine 1990; Jackson et al. 1991) or postsynaptic modification of receptors responses (Bashir et al. 1991; Hawkins et al. 1993). The augmented neural output from the cochlea in the absence of sound may ultimately result in symptoms manifested as tinnitus. One of the major findings of the study was that quinine broadened action potentials of SG neurons in culture. At present, it is not clear whether in vivo action potentials arrived at the nerve terminals in the cochlear nuclei are similarly affected. However, it is conceivable that quinine is internalized at the auditory periphery and then transported to the terminals in the cochlear nuclei since the drug is an amphiphilic molecule (Zidovetzki et al. 1989). The waveform of presynaptic action potentials there may be similarly broadened because our results also showed that internally applied quinine was effective. Action-potential broadening tends to allow calcium channels to open for longer times. The lowering of action-potential amplitude caused by quinine's inhibition on I Na may increase the driving force for Ca2+ entry as well. The consequence of action-potential broadening on the Ca2+ entry was theoretically examined in Fig. 11 in which I Ca accompanying each action potential was calculated by a model first presented by Llinas et al. (1981). The effect of 200 μM quinine on the waveform of action potentials is replotted in Fig. 10 A from data presented in Fig. 3 D. The corresponding I Ca accompanying each action potential was calculated (Fig. 11 B). As a result of the action-potential broadening (Fig. 11 A, ⋅⋅⋅), I Ca lasted much longer and had greater amplitude during the repolarizing phase of action potentials. This effect combined with the increase in driving force due to a lowering of action-potential amplitude resulted in a dose-dependent increase in Ca2+ entry (Fig. 11 B). More calcium entry may have many effects on the function of SG neurons because intracellular Ca2+ homeostasis is related to a wide variety of cellular functions, including an augmentation of transmitter release (Augustine et al. 1985). Classical work by Katz (1967a, 1970) demonstrated that an increase in presynaptic free Ca2+ concentration through the opening of voltage-gated Ca2+ channels triggered by the arrival of action potentials is required for transmitter release. Augustine (1990) found that a small 30% increase in the presynaptic action-potential duration will increase presynaptic I Ca by 230%, and postsynaptic excitatory current by 190%. Recently, it has been demonstrated that the presynaptic action-potential waveform effectively modulates the synaptic strength in the CNS (Sabatini and Regehr 1997). From the point of view that more neural output means a greater neural signal, the end result of those quinine toxic effects could be the perception of tinnitus as normally nonaudible spontaneous firings in the auditory nerve become audible by the amplification of the action-potential broadening. It has been demonstrated that the width of action potentials is dependent on the resting membrane potential of SG neurons (Lin 1997); therefore it is plausible that the action-potential broadening may underlie cochlear-originated tinnitus other than the one induced by quinine because it is conceivable that the resting membrane potential may be depolarized in many pathological conditions.

Fig. 11.

Modeling results on how action-potential waveform affects Ca2+ entry. Action-potential waveform obtained in the absence ( Graphic ) and presence (⋅⋅⋅) of 200 μM quinine. The same data used in Fig. 3 A were used here. The corresponding action-potential–evoked calcium currents are calculated. Integration of I Ca gives the total Ca2+ entry in the absence( Graphic ) and presence (⋅⋅⋅) of quinine. Calculations were done using data from Fig. 3, A–D, and the results are given in B.

Where is the primary site of quinine's ototoxic actions? Clinical symptoms of quinine intoxication include hearing loss, tinnitus, decrease in visual acuity, and headache. These symptoms seem to be consistent with a general neurological effect of quinine intoxication. In general, quinine affects many types of voltage-gated channels with varying degree of potencies. Half inhibition concentrations of quinine on the potassium current (IC50 = 5.1 μM) and sodium current (IC50 = 64 μM) of the rat taste cells (Chen and Herness 1997) are comparable with results reported here. In other nonauditory cells, quinine has also been found to inhibit both calcium-dependent (IC50 = 300 μM) and calcium-independent (IC50 = 25 μM) potassium channels (Fatherazi and Cook 1991; Glavinovic and Trifaro 1988). On the single-channel level, quinine usually causes a flickery block in the open state without interfering with the normal gating process of the potassium channel (Glavinovic and Trifaro 1988; Takeuchi et al. 1995). This open-channel blocking scheme may give rise to the apparent acceleration of the inactivation of the whole cell potassium current, as those demonstrated in Fig. 4. The same effect, however, could be produced by differentially blocking noninactivating components in the whole cell potassium current. Exactly how quinine differentially affects each type of potassium channels in the SG neurons is not clear. However, the observation that quinine could reduce whole cell potassium current of SG neurons to almost zero at moderate concentrations (∼200 μM) indicated that major types of potassium channels in the SG neurons are effectively blocked by quinine.

In a study in which the degree of hearing loss and plasma concentration of quinine were correlated (Roche et al. 1990), peak quinine concentration was found to be in the range of 10–58 μM. This concentration of quinine is within the range where the drug may affect both potassium and sodium channels in the SG neuron membrane (Figs. 5 and 7), but far less than that needed to affect other targets in the inner ear. For example, 0.5–4 mM quinine was used to affect the cochlear mechanics in the isolated temporal bone preparation (Karlsson et al. 1991b), 1 mM quinine was observed to cause a flicking block on ion channels obtained from dissociated marginal cells of gerbil stria vascularis (Takeuchi et al. 1995). IC50 of quinine block on the I K of isolated OHCs (49 μM) (Lin et al. 1995) is also substantially higher than that of SG neurons (see Fig. 5). Both clinical symptoms and data obtained on the cellular level seem to agree that neurons in the sensory systems may be the primary targets affected by quinine intoxication. One alternative hypothesis, which remains to be tested, is that quinine may increase the transmitter leakage on the bases of inner hair cells by depolarizing the membrane potential. However, voltage dependence of quinine blockage on I K revealed in this work (Fig. 5) made this hypothesis unlikely.


The author is grateful for the helpful comments of Drs. David Lim, Neil Segil, and Federico Kalinec and proofreading by L. Gnerre. Two anonymous reviewers also contributed valuable suggestions to the manuscript.

This work was supported by National Institute of Deafness and Other Communication Disorders Grant DC-02567 and a grant from the National Organization for Hearing Research.


  • Address reprint requests to X. Lin.


View Abstract