INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Behavioral responses to sensory stimuli are governed by mechanisms operating at the level of circuits, synapses, and intrinsic membrane properties, all of which contribute to neuronal excitability in sensory-motor pathways. Although experience-dependent changes in behavior are commonly thought to arise from changes in synaptic strength, a growing body of evidence indicates that ion channels controlling firing responses can be strongly influenced by neuronal activity (Aizenman and Linden 2000; Desai et al. 1999; Ganguly et al. 2000; Turrigiano et al. 1994). These findings raise the possibility that regulation of ion channels could contribute to modulation of behavioral responses.

The vestibuloocular reflex (VOR) provides a particularly tractable system for investigating the roles of intrinsic currents in behavioral signaling and plasticity. The function of the VOR is to stabilize images on the retina during self-motion by producing eye movements that compensate for motion of the head. This simple behavioral reflex is remarkably adaptable: both unilateral loss of peripheral vestibular function and persistent image motion during head motion in labyrinth-intact subjects result in adaptive recalibration of the VOR (reviewed in du Lac et al. 1995; Smith and Curthoys 1989). This recalibration is expressed as changes in the gain of the VOR (the ratio of evoked eye velocity to input head velocity). VOR gain changes are paralleled by changes in the strength of firing responses during head movement in vestibular nucleus neurons, which transform head movement signals into the appropriate oculomotor commands (Lisberger and Pavelko 1988; Lisberger et al. 1994; Newlands and Perachio 1990a,b; Partsalis et al. 1995). Whether the underlying cellular mechanisms involve alterations in synaptic strength or intrinsic membrane properties has yet to be determined. Recent evidence, however, suggests that adaptive changes in intrinsic membrane properties contribute to VOR plasticity (Cameron and Dutia 1997; Him and Dutia 2001; Ris et al. 1995, 2001).

The goal of this study was to identify ion channels in vestibular nucleus neurons that influence firing responses and could therefore be candidates for mediating behavioral gain control in the VOR. Neurons in the medial vestibular nucleus (MVN) transform their inputs into firing rates in a remarkably linear fashion over a wide range of input strengths and firing rates (Fig. 1) (du Lac and Lisberger 1995a,b). If regulation of particular classes of ion channels in MVN neurons were a mechanism that modulated the gain of the VOR, then modulations of those channels should influence firing response gain (the ratio of evoked firing rate to input current) without affecting response linearity or dynamic range. To identify such channels, we analyzed intrinsic firing responses during pharmacological manipulation of calcium and potassium currents and of kinases and phosphatases that could modulate them. We show that two distinct classes of calcium-dependent potassium channels control firing responses in MVN neurons and that response gain is actively modulated by constituent phosphorylation.

View larger version (36K): [in this window] [in a new window]   Fig. 1. Firing responses are linear in medial vestibular nucleus (MVN) neurons. A: response of an MVN neuron to intracellular injection of a step of current (175 pA). Membrane potential is plotted as a function of time at top of graph; injected current is plotted below. The neuron fired spontaneous action potentials; in response to current injection (at time = 0), firing rate increased and was sustained throughout the 1-s step. B: time course of firing rate changes in response to repeated steps of depolarizing current. Instantaneous firing rate (reciprocal of the interspike interval) is plotted as a function of time for the response shown in A and 2 additional responses of the neuron to the identical current step. The neuron fired reliably in response to input current and showed little spike frequency adaptation. C: relationship between input current and firing rate in the same neuron. Mean firing rate during each step is plotted as a function of step amplitude. For responses to 3 repetitions of each current step, SDs are smaller than the symbols. The current to firing rate relationship was linear (R2 = 0.999) with a gain of 149 (spikes/s)/nA. D: firing response gains of 64 neurons are plotted vs. input resistance for each neuron. Gains were poorly correlated with input resistance: correlation coefficient (R2) is 0.21.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slice preparation

Firing responses of MVN neurons were assessed with whole cell patch-clamp recordings from rat brain stem slices. Slices containing the MVN were prepared from 16- to 25-day old rats (Long Evans) as detailed previously (Murphy and du Lac 2001). Following pentobarbital sodium anesthesia, the brain stem was dissected from the skull in Ringer solution maintained at 4°C and bubbled with carbogen (95% O₂-5% CO₂). Transverse slices were prepared on a vibratome (Campden) at thicknesses of 300-400 µM and incubated in carbogenated Ringer solution at room temperature for 1 h prior to electrophysiological recordings. Carbogenated Ringer solution had a final pH of 7.4 and contained (in mM) 124 NaCl, 5 KCl, 1.3 MgSO₄, 26 NaHCO₃, 2.5 CaCl₂, 1.0 NaH₂PO₄, and 11 dextrose.

Electrophysiology

Brain stem slices were held in a submersion-style chamber during recording experiments. Carbogenated Ringer solution, warmed to 31-33°C, flowed through the chamber at a rate of 4 ml/min. Neurons were recorded intracellularly with whole cell patch pipettes pulled from borosilicate glass on a Flaming/Brown electrode puller. The electrodes had tips of diameter 1-2 µM and resistances of 5-10 M. Electrodes were filled with an solution containing (in mM) 122.5 K⁺-gluconate, 17.5 KCl, 8 NaCl, 10 HEPES, 0.1 EGTA, 2 MgATP, and 0.3 Na₂GTP (pH = 7.2; 280-285 mOsm). Electrophysiological recordings were made in bridge mode with an Axoclamp 2B amplifier (Axon Instruments). Neurons were visualized using differential interference contrast optics under infrared illumination. Voltage signals were amplified 50×, filtered (3 kHz), sampled at 20 kHz, and collected on a PowerMac computer (Apple) using software written in the laboratory with IgorPro (Wavemetrics). Voltage offsets were corrected after removal of the electrode from the cell. Membrane potentials were corrected for a liquid junction potential.

In all experiments, kynurenic acid (2 mM) and picrotoxin (100 µM) were added to the Ringer solution to block synaptic transmission. MgCl₂ was substituted for CaCl₂ to maintain external divalent cation concentration in low-external-Ca²⁺ experiments; Ca²⁺ concentrations refer to those added to the Ringer solution rather than to measured external Ca²⁺. All pharmacological agents except for EGTA, BAPTA, ATPS, microcystin LR, protein kinase inhibitor (5-24), -conotoxin GVIA, and -agatoxin were applied to the perfusing Ringer solution. BAPTA, EGTA, microcystin LR, and protein kinase inhibitor (5-24) were added to the intracellular recording solution. ATPS was substituted for ATP in the intracellular solution. -Conotoxin GVIA and -Agatoxin were made as 35× stocks in Ringer and added directly to the chamber with perfusion stopped to attain a final concentration of 2 µM and 200 nM, respectively. The toxins were allowed to diffuse for 2-5 min before perfusion was continued. Pharmacological occlusion experiments requiring complete BK channel blockade were performed by preincubating slices in iberiotoxin for 5-7 h, thereby enabling block of BK channels containing the 4 subunit that produce a very slow association rate between iberiotoxin and channel (Meera et al. 2000). Cadmium, nickel chloride, amiloride, KN-62, cyclosporin A, protein kinase inhibitor (5-24), chelerythrine chloride, H-89, ATPS, and nifedipene were obtained from Sigma; apamin, -conotoxin GVIA, microcystin LR, and -agatoxin were obtained from Calbiochem; iberiotoxin was obtained from Alomone Laboratories. Microcystin LR and KN-62 were first dissolved in DMSO (<0.1% DMSO, final concentration).

Data analysis

Neurons were excluded from the analysis if they could not maintain firing of 50 spikes/s during steps of injected current or if their action potentials were <40 mV from threshold to peak. Spontaneous firing rates varied during the course of some recordings, typically declining but occasionally increasing (Nelson et al. 2001). In contrast, responses to current injection were quite consistent throughout control experiments (see RESULTS); we therefore focused the analyses on evoked responses. Firing responses were obtained during intracellularly injected steps of current, 1 s in duration. Each of a range of current amplitudes was repeated three times to obtain firing response gains (the slope of the current to mean firing rate relationship). Firing response dynamics were quantified by an adaptation index, defined as the ratio of firing rates at the beginning (from peak firing rate to 100 ms after step onset) and end (final 200 ms) of the current step. Analyses of action potentials were performed on averages of 10 spikes, aligned at their peak membrane potential. Action potential threshold was defined the intersection between linear fits to the gradual depolarizing phase preceding the action potential (from 10 ms preceding the action potential peak) and the rapidly rising phase of the action potential (to the peak of the derivative of membrane potential) (Murphy and du Lac 2001). After-hyperpolarization (AHP) amplitude was defined as the difference between action potential threshold and the most negative membrane potential attained during the AHP. Input resistances were calculated from the change in membrane potential evoked by hyperpolarizing current steps (500 ms) when the neuron was held hyperpolarized with DC current (typically at about 75 mV) to preclude spontaneous firing. To minimize contribution from subthreshold, voltage-dependent conductances, stimuli consisted of small amplitude (10-50 pA) steps of current. Input resistance values were obtained from averages of responses to 6-10 steps of current.

Data are reported as means ± SE. Students' paired t-tests were used to evaluate the effects of pharmacological agents.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MVN neurons recorded in brain stem slices fired regular, spontaneous action potentials at rates 30 spikes/s. Depolarization by intracellular current injection evoked sustained increases in firing rate that depended linearly on current amplitude, as has been described previously (du Lac and Lisberger 1995a,b). Figure 1 shows an example in a representative MVN neuron that fired spontaneously at 10 spikes/s. In response to a 175-pA step of intracellularly injected current, firing rate rapidly increased and then returned to the spontaneous level following the offset of the current step. The evoked increases in firing rate declined little during maintained depolarization and were highly consistent from trial to trial as can be seen in plots of instantaneous firing rate versus time in response to three repetitions of the same current step (Fig. 1B). The mean firing rate during current injection increased linearly with the amplitude of injected current (R² = 0.999) with a slope of 149 (spikes/s)/nA (Fig. 1C). Because the current to firing rate relationship is linear in MVN neurons, its slope can be used to define the gain of the firing response. For any given MVN neuron, firing response gain was very consistent over time, changing by <1% during 30-60 min of recording (0.2 ± 2%). In contrast, spontaneous firing rate could vary over the course of an experiment either via apparent run-down or by activity-dependent potentiation (Nelson et al. 2001).

Firing response gain varied across MVN neurons, from 61 to 637 (spikes/s)/nA (n = 64). One potential source of this large diversity in gain is variation in membrane area or total number of open channels, which together contribute to input resistance. However, as shown in Fig. 1D, input resistance measured below spike threshold did not correlate well with gain (R² = 0.21). This finding suggests that gain is controlled primarily by currents active when MVN neurons are firing, such as those mediated by voltage-gated conductances activated by depolarization during action potentials or by conductances that are active during the interspike interval.

Firing response gain depends on voltage-sensitive Ca²+ currents

Ca²⁺ influx through voltage-sensitive Ca²⁺ channels that open during the action potential influences the firing properties of many types of neurons either directly or via Ca²⁺-dependent K⁺ channels (Sah 1996). To determine how Ca²⁺ influx contributes in MVN neurons, we analyzed firing rate responses to current injection in altered extracellular Ca²⁺ concentrations and in the presence of cadmium, a broad-spectrum blocker of voltage-sensitive Ca²⁺ channels (Tsien et al. 1988).

Manipulations of extracellular Ca²⁺ evoked dramatic changes in firing response gain in MVN neurons. Figure 2A shows the firing rate responses of an individual MVN neuron to a 200-pA step of current in control conditions ([Ca²⁺]_ext = 2.5 mM), in high external Ca²⁺ (10 mM), and in low Ca²⁺ (0 mM). In control conditions, the step increased firing rate to a mean of 38 spikes/s. High-external-Ca²⁺ concentrations reduced the mean firing response to 23 spikes/s, whereas the same stimulus in low external Ca²⁺ evoked an initial firing rate of 87 spikes/s that declined to 50 spikes/s during the 1-s step.

View larger version (18K): [in this window] [in a new window]   Fig. 2. Extracellular calcium influences firing response gain and the afterhyperpolarization (AHP). A: firing rate responses to 3 repetitions of a step of current (200 pA) are plotted vs. time for a single neuron in the presence of different concentrations of extracellular Ca2+: control (2.5 mM), low (0 mM), and high (10 mM). Evoked firing rate varied inversely with extracellular calcium concentration. B: relationship between input current and mean evoked firing rate in the same neuron. Evoked firing responses were linear in each condition (R2 values were 0.996, 0.992, and 0.994 in control, high Ca2+, and low Ca2+, respectively). Firing response gains varied as a function of extracellular calcium concentration [gains were 164, 89, and 266 (spikes/s)/nA in control, high, and low Ca2+, respectively]. C: spike shape varied with extracellular [Ca2+]. Lowering [Ca2+] produced modest increases in spike width and dramatic decreases in the AHP. Averages of 25-50 action potentials are shown for each condition. Membrane potential traces, corrected for junction potentials, have been aligned vertically for ease of comparison. Horizontal dotted line, 45 mV in control, 42 mV in low Ca2+, and 52 mV in high Ca2+.

Figure 2B plots mean firing rate during the injected current step as a function of stimulus amplitude for the neuron and conditions shown in Fig. 2A. In control conditions, the relationship between firing rate and current amplitude was linear with a gain of 164 (spikes/s)/nA. Responses in high extracellular Ca²⁺ were reduced over the entire range of stimulus amplitudes presented such that the gain dropped to 89 (spikes/s)/nA. Conversely, decreasing extracellular Ca²⁺ evoked an increase in gain to 266 (spikes/s)/nA. In each of seven neurons, firing response gain increased in low Ca²⁺ and decreased in high Ca²⁺ (Fig. 11). Manipulations of [Ca²⁺]_ext can affect neuronal excitability via changes in voltage-dependent gating of ion channels (Hille 2001). However, extracellular application of cadmium evoked an increase in gain similar to that seen when lowering extracellular Ca²⁺ (Fig. 11, n = 6). These results indicate that firing response gain is influenced by Ca²⁺ influx through voltage-sensitive Ca²⁺ channels.

The dynamics of the firing rate response were also affected by extracellular Ca²⁺ concentration such that the firing rate declined more steeply during sustained depolarization in low Ca²⁺ than in control conditions (Fig. 2A). Firing response dynamics were quantified by an adaptation index (AI), which measured the ratio of firing rates attained at the onset and offset of current injection (see METHODS). The AI increased slightly but significantly in low Ca²⁺ but was unaffected by high Ca²⁺ (Table 1). Cadmium (100 µM) had similar effects on dynamics as low Ca²⁺ (Table 1). Together, these results indicate that the spike frequency adaptation observed in MVN neurons is not mediated predominantly by Ca²⁺-dependent K⁺ channels.

To obtain insights into the mechanisms by which Ca²⁺ influx affects firing responses, we compared the membrane potential during firing in different concentrations of extracellular Ca²⁺ (Fig. 2C). Changes in external Ca²⁺ had relatively little effect on action potential width. In contrast, the AHP was strongly affected by external Ca²⁺. In low Ca²⁺, the magnitude of the AHP decreased by 7.7 ± 1.4 mV (P < 0.001, n = 7), whereas in high Ca²⁺, the AHP increased by 5.0 ± 0.5 mV (P < 0.005, n = 7). Cadmium (100 µM) produced a reduction of the AHP similar to that seen with low Ca²⁺ (by 6.6 ± 1.2 mV, P < 0.005, n = 6). These effects are likely to be mediated by potassium currents that are activated by Ca²⁺ influx during the action potential (Hille 2001).

SK Ca²+-dependent potassium currents regulate firing response gain in MVN neurons

Two primary classes of potassium channels are activated by Ca²⁺ influx into neurons: BK channels have been implicated in repolarization of the action potential, whereas SK channels are thought to control the membrane potential between action potentials and, consequently, neuronal excitability (Sah 1996). We assessed the contribution of SK-type Ca²⁺-activated K⁺ channels to firing responses with apamin, a specific blocker of a subset of SK channels (Kohler et al. 1996).

Blockade of SK channels with apamin evoked robust increases in firing response gain without affecting response linearity or dynamics. The effects of apamin on firing rate responses to intracellular injected current in an individual MVN neuron are shown in Fig. 3. Apamin increased evoked firing rate while having little effect on spike frequency adaptation (Fig. 3A and Table 1). Figure 3B plots mean firing rate evoked by a family of current steps for the neuron shown in Fig. 3A. The slope of the current to firing rate relationship increased from 147 (spikes/s)/nA in control conditions to 404 (spikes/s)/nA in the presence of apamin. Despite this dramatic increase in firing response gain, the relationship between input current and evoked firing rate remained linear up to firing rates of 150 spikes/s. The magnitude of the gain increase after apamin blockade of SK currents ranged widely across neurons from 1.6- to 4.6-fold (Fig. 11, n = 10) and was poorly correlated with gain in control conditions (R² = 0.15).

View larger version (16K): [in this window] [in a new window]   Fig. 3. Blockade of SK channels produces an increase in firing response gain without compromising linearity. A: firing rate responses of an MVN neuron to 3 repetitions of a 100-pA current step in control conditions () and in the presence of 200 nM apamin (). Response magnitude increased in apamin with no concomitant change in spike frequency adaptation. B: effects of apamin on the relationship between current amplitude and mean evoked firing rate in the same neuron. Firing responses were linear (R2 = 0.99) in control conditions with a gain of 147 (spikes/s)/nA. Apamin resulted in a dramatic increase in gain, to 404 (spikes/s)/nA, with no effect on response linearity (R2 = 0.996). C: averaged spike shape in control conditions () and in apamin (- - -). Apamin had little effect on action potential width or repolarization but decreased the AHP beginning within 2 ms of the peak of the action potential. · · · , 45 mV.

Blocking apamin-sensitive SK currents reduced the late component of the AHP (Fig. 3C). Apamin attenuated the AHP by 3.7 ± 0.7 mV (P < 0.001, n = 10). The membrane potential in apamin deviated from that in control conditions within 1.1-3.1 ms of the peak of the action potential. The ratio of firing response gains in apamin and control conditions were not correlated with either the ratios or the differences in the AHP (R² = 0.046 and 0.013, respectively). Taken together, these results indicate that firing response gain is influenced by the opening of apamin-sensitive SK channels during the AHP but that additional ionic currents contribute to the AHP.

BK Ca²+-dependent potassium currents regulate firing response gain in a subset of MVN neurons

To determine whether BK-type Ca²⁺ activated potassium channels also regulate firing in MVN neurons, we assessed firing responses in the presence of iberiotoxin (IBTX), which specifically blocks BK channels (Galvez et al. 1990). IBTX (150 nM) had variable effects on firing responses in MVN neurons, as shown in Fig. 4. Two classes of effects could be distinguished.

View larger version (30K): [in this window] [in a new window]   Fig. 4. BK channels contribute to firing response gain in some MVN neurons. A-C and D-F, respectively, indicate data from 2 different neurons. A: firing rate responses of an MVN neuron to 3 repetitions of a 200-pA step of current in control conditions () and after 15 min in 200 nM iberiotoxin (IBTX: ). B: effects of IBTX on the relationship between current amplitude and mean evoked firing rate in the same neuron. Firing responses were linear (R2 = 0.999) in control conditions with a gain of 179 (spikes/s)/nA. IBTX approximately doubled the gain, to 334 (spikes/s)/nA, with little effect on response linearity (R2 = 0.994). C: averaged spike shape in control conditions () and in IBTX (- - -) for the neuron shown in A and B. IBTX slowed the rapid repolarization of the action potential and reduced the AHP. D: firing rate responses to 3 repetitions of a 200-pA step of current in a different MVN neuron in control conditions () and after 15 min of 200 nM IBTX (). Spontaneous firing was slowed in IBTX but response magnitude was unaffected. E: mean firing rate vs. current plots for the neuron shown in D. IBTX reduced the range of firing rates that could be sustained during 1-s current depolarizations and resulted in a slightly sublinear response (R2 = 0.998 in control and 0.98 in IBTX). Control gain: 154 (spikes/s)/nA, IBTX gain: 135 (spikes/s)/nA. F: averaged spike shape in control () and IBTX (- - -) for the neuron shown in D and E. IBTX slowed spike repolarization and reduced the early component of the AHP. · · · in C and F indicate 45 mV.

In 10 of 15 neurons, IBTX evoked a pronounced increase in firing response gain, as exemplified by the neuron shown in Fig. 4, A and B. In the remaining five neurons, firing response gain was either little affected or slightly reduced in IBTX. For example, in the neuron shown in Fig. 4D, the mean change in firing rate in response to a 150-pA step of current was 29 spikes/s in control conditions and 30 spikes/s in IBTX. Firing rate responses to larger current steps were progressively smaller in IBTX than in control, producing a nonlinear current to firing rate relationship, as shown in Fig. 4E. This decline in incremental firing rate evoked by larger current steps resulted from an increase in spike frequency adaptation in IBTX relative to control (Table 1).

As exemplified in Fig. 4, C and F, the width of the action potential increased (by 0.32 ± 0.06 ms, P < 0.0001) and the magnitude of the AHP decreased (by 7 ± 1.2 mV, P < 0.005, n = 15) with IBTX application in all neurons tested. The membrane potential in IBTX deviated from that in control conditions within 0.25-0.4 ms of the peak of the action potential, indicating rapid activation of BK currents during the action potential. In neurons in which IBTX evoked increases in firing response gain, the late component of the AHP was also reduced (Fig. 4C). However, in the other neurons, the late component of the AHP was similar in magnitude in control conditions and in IBTX (Fig. 4F), indicating that in the absence of BK currents, other ionic currents contribute to membrane repolarization.

To investigate whether Ca²⁺ influx can influence firing response gain in MVN neurons by acting on mechanisms other than BK and SK channels, we performed the following experiment. First, apamin and IBTX were applied to MVN neurons at concentrations (200 nM) that produced saturating changes in gain. Then, extracellular Ca²⁺ was replaced with Co²⁺, which prevents Ca²⁺ influx through voltage-sensitive Ca²⁺ channels. This complete blockade of SK, BK, and voltage-gated Ca²⁺ channels depolarized MVN neurons and resulted in a reduced range of firing rate responses to input current (data not shown). However, the addition of Co²⁺ produced no additional increases in gain (n = 3). These data indicate that SK and BK channels are the predominant route by which Ca²⁺ influences firing responses in MVN neurons.

N-type Ca²+ currents influence firing response gain via SK channels

Having established that Ca²⁺ influx into MVN neurons regulates firing response gain and the AHP, we next investigated whether distinct Ca²⁺ sources supply the requisite Ca²⁺ to SK and BK channels, respectively, by using blockers of specific voltage-sensitive Ca²⁺ channels.

Neither blockade of P/Q-type Ca²⁺ channels with agatoxin (200 nM) nor of L-type Ca²⁺ channels with nifedipene (250 µM) had significant effects on firing response gains (Fig. 11). However, as shown in Fig. 5, -conotoxin (-CTX; 2 µM), which specifically blocks N-type Ca²⁺ channels, evoked pronounced increases in firing response gain (n = 7). -CTX reduced the late component of the AHP (Fig. 5C); the membrane potential in -CTX deviated from that in control conditions within 0.5-1 ms of the peak of the action potential, reducing AHP magnitude by 3.4 ± 0.5 mV (P < 0.005, n = 5). The effect of -CTX on the time course of the AHP was similar to that evoked by the SK channel blocker apamin (compare Figs. 5B and 3C), suggesting that N-type Ca²⁺ currents influence firing responses of MVN neurons via SK channels.

View larger version (15K): [in this window] [in a new window]   Fig. 5. Blockade of N-type Ca2+ channels evokes a linear increase in firing response gain. A: relationship between mean firing rate and input current in control conditions and in the presence of 2 µM -conotoxin GVIA (-CTX). Gain increased in -CTX from 162 to 403 (spikes/s)/nA, while linearity was unaffected (R2 = 0.993 and 0.996 in control and -CTX, respectively). B: averaged spike shapes in control () and in -CTX (- - -). -CTX had little effect on the action potential but reduced the late component of the AHP. · · · , 45 mV.

If Ca²⁺ influx through N-type Ca²⁺ channels affects firing responses of MVN neurons by supplying Ca²⁺ to SK channels, then SK channel blockade should occlude the effects of blocking N-type channels. Figure 6A shows that this prediction was confirmed. Firing response gain in control conditions was 139 (spikes/s)/nA in the neuron shown in Fig. 6A. Following saturating blockade of SK currents with apamin (200 nM), gain increased by more than threefold. Subsequent application of -CTX evoked no further change in gain (n = 4 neurons, P = 0.73). These results indicate that N-type Ca²⁺ channels influence firing response gain solely through SK-type Ca²⁺-dependent K⁺ channels.

View larger version (22K): [in this window] [in a new window]   Fig. 6. N-type calcium channels affect gain by providing Ca2+ to SK channels. A: blockade of SK channels occludes gain increases by -CTX. Firing rate-current relationships in an MVN neuron in control conditions [gain = 139 (spikes/s)/nA], in the presence of saturating concentrations (200 nM) of apamin [gain = 502 (spikes/s)/nA], and in 200 nM apamin combined with 2 µM -CTX [gain = 531 (spikes/s)/nA)]. B: N-type Ca2+ channels are not the only source of calcium for SK channels. Blockade of N-type channels with saturating concentrations of -CTX (4 µM) evoked increases in gain over control conditions from 172 to 324 (spikes/s)/nA. Addition of 200 nM apamin to the -CTX-Ringer evoked an additional increase in gain to 537 (spikes/s)/nA.

Do N-type currents provide the sole source of Ca²⁺ for SK channels? To investigate this, we performed the converse occlusion experiment. First, N-type channels were blocked with saturating concentrations of -CTX (4 µM). If N-type channels provided the only Ca²⁺ source for SK-type channels, then subsequent blockade of SK channels should evoke no further increases in gain. However, the results shown in Fig. 6B indicate that blocking N-type Ca²⁺ channels did not occlude increases in gain following SK channel blockade with apamin: gain increased by 191 ± 13% when apamin was applied subsequent to -CTX (P < 0.05, n = 3). These findings demonstrate that in MVN neurons, SK channels can be activated by Ca²⁺ supplied both by N-type channels and by Ca²⁺ sources other than P/Q- or L-type channels. Possibilities include R-type Ca²⁺ channels, for which no specific antagonist is currently available, and T-type Ca²⁺ channels.

Ni²+- and amiloride-sensitive currents Ca²⁺ currents influence firing response gain via BK channels

We assessed whether presumed T-type Ca²⁺ currents influence firing responses in MVN neurons with amiloride and low concentrations of Ni²⁺ (100 µM), each of which selectively blocks T-type currents in other neurons (Fox et al. 1987; Tang et al. 1988).

Ni²⁺ (100 µM) produced increases in firing response gain of 22-80% (n = 5, Figs. 7 and 11) without affecting response linearity (Fig. 7A) or dynamics (Table 1). As shown in Fig. 7B, the early component of the AHP was consistently reduced in Ni²⁺ (by 2.4 ± 0.7 mV, P < 0.005, n = 4). The membrane potential deviated from that in control conditions within 0.35 ms, indicating that Ni²⁺-sensitive currents are rapidly activated during the action potential. As was the case when blocking BK channels with IBTX, the Ca²⁺ channel blocker amiloride (250 µM) had variable effects on firing responses in MVN neurons: in 8/11 neurons, amiloride increased gain by 12-69%, while it had no effect on gain in the remaining 3 neurons. The AHP was consistently reduced in amiloride (by 4.7 ± 1.1 mV; P < 0.05, n = 6).

View larger version (18K): [in this window] [in a new window]   Fig. 7. T-type Ca2+ channels influence firing response gain. A: mean firing rate versus input current in control conditions and in the presence of 100 µM Ni2+. Ni2+ increased gain from 169 in control conditions to 206 (spikes/s)/nA. B: averaged spike shape in control () and Ni2+ (- - -). Ni2+ slowed action potential repolarization and reduced the AHP. · · · , 45 mV.

The effects of Ni²⁺ and IBTX channel blockade on the time course of the AHP (compare Fig. 7C with 4C) suggest that T-type channels influence firing response gain via BK channels. Pharmacological occlusion experiments were consistent with this hypothesis. Following saturating blockade of SK channels with apamin (200 nM), subsequent application of Ni²⁺ produced increases in gain of 25 ± 7% (Fig. 8A; n = 3). Similar results were found with amiloride following apamin occlusion with an average gain increase of 24 ± 4% (data not shown; n = 4). The magnitude of this increase was similar to that observed in the absence of apamin, indicating that presumed T-type Ca²⁺ currents contribute to gain via a non-SK mechanism.

View larger version (21K): [in this window] [in a new window]   Fig. 8. T-type Ca2+ channels affect gain primarily through BK channels. A: blockade of SK channels does not occlude gain increases by 100 µM Ni2+. Firing response gain increased from 126 to 322 (spikes/s)/nA in apamin, and in 100 µM nickel combined with apamin, the gain increased further to 379 (spikes/s)/nA (ANCOVA: P < 0.005). B: blockade of BK channels with saturating exposure to iberiotoxin (IBTX) occluded further increases in gain due to 100 µM nickel. Firing response gain in saturating IBTX was 67 and 71 (spikes/s)/nA with the addition of 100 µM Ni2+.

To assess whether Ni²⁺-sensitive current acts via the BK pathway, we first blocked BK channels with saturating IBTX (150 nM) and measured firing responses in the presence of Ni²⁺. As shown in Fig. 8B, pretreatment with IBTX occluded further increases in gain by Ni²⁺ (gain increased by 3 ± 12%, n = 3, P = 0.58). These results indicate that Ca²⁺ influx through presumed T-type channels influences firing response gain primarily via BK channels.

Ca²+-calmodulin-dependent kinase II actively controls firing response gain

Many ion channels are regulated by phosphorylation, including SK and BK channels (Levitan 1994). Kinases, especially Ca²⁺/calmodulin-dependent protein kinases, often act as mediators in activity-dependent processes by translating activity, which generates Ca²⁺ influx, into phosphorylation of ion channel targets, which in turn can modulate synaptic efficacy (reviewed in Schulman et al. 1992; Stevens et al. 1994) and intrinsic excitability (Muller et al. 1992). To address whether firing response gain in MVN neurons is regulated by phosphorylation, we examined the effects of agents that influence kinase and phosphatase activity.

Neither inhibition of cAMP-dependent protein kinase (with H-89, 10 µM, and protein kinase inhibitor 5-24, 100 µM) nor of protein kinase C (with chelerythrine chloride, 10 µM) evoked consistent changes in gain. However, a selective inhibitor of the Ca²⁺-calmodulin-dependent protein kinase II (CaMK II), KN-62 (10 µM), produced robust increases in gain (Figs. 9A and 11), while reducing the AHP. Both the fast and slow components of the AHP were affected, with an average reduction in the AHP of 4.1 ± 0.5 mV (n = 8, P < 0.005), suggesting the involvement of BK channels. These data indicate that the AHP and firing response gains are regulated by constitutive CaMK II activity.

View larger version (14K): [in this window] [in a new window]   Fig. 9. Ca2+-calmodulin kinase II (CaMKinase II) actively regulates firing response gain via BK channels. A: mean firing rate vs. input current in a control neuron and in the same neuron after 55 min of exposure to the specific CaMKinase II blocker KN-62 (10 µM). Gain increased in KN-62 from 136 to 321 (spikes/s)/nA. B: averaged spike shape in the same neuron shown in A in control conditions () and after treatment with KN-62 (- - -). KN-62 reduced both the rapid repolarization of the action potential and AHP magnitude. · · · , 45 mV. C: blockade of BK channels with saturating exposure to IBTX occluded increases in gain due to KN-62. Gain in IBTX was 457 (spikes/s)/nA, and after treatment with KN-62 for 58 min, the gain was no greater [435 (spikes/s)/nA].

To identify the gain control pathway regulated by CaMK II, we first blocked BK channels with saturating concentrations of IBTX (150 nM). As shown in Fig. 9C, subsequent application of KN-62 did not evoke further increases in gain (3 ± 7% increase over IBTX alone; n = 3). These results indicate that CaMK II regulation of firing response gain occurs via the BK channel pathway.

Phosphatases regulate firing response gain via BK channels

To further explore the regulation of firing response gain by phosphorylation, we employed several strategies. Blockade of phosphatases would be expected to increase the phosphorylation level of cellular proteins if kinases were constitutively active. Accordingly, we tested the effects of microcystin LR, a wide-spectrum phosphatase inhibitor. Microcystin LR (10 µM) produced a robust increase in gain over a period of 60-100 min of intracellular dialysis (n = 6, Figs. 10A and 11). Intracellular dialysis of the nonhydrolyzable ATP analogue ATPS, which acts as an effective phosphatase inhibitor, resulted in similar increases in gain (n = 9, Fig. 11). Both microcystin and ATPS reduced the fast and slow components of the AHP, similar to the effect of the BK channel blocker IBTX (compare Figs. 10B and 4B). With microcystin, the AHP was reduced by 7.6 ± 1.9 mV (P < 0.05, n = 6), and with ATPS, it was reduced by 5.8 ± 1.9 mV (P < 0.01, n = 9).

View larger version (16K): [in this window] [in a new window]   Fig. 10. Firing response gain is actively regulated by phosphorylation through BK channels. A: inclusion of the phosphatase inhibitor, microcystin LR (10 µM), in the recording pipette generated increases in firing response gain over time. Control gain [101 (spikes/s)/nA] was measured 1 min into the recording, before substantial dialysis and irreversible phosphorylation could occur. After 100 min of dialysis with microcystin LR, the gain increased to 234 (spikes/s)/nA. B: averaged spike shape in control conditions () and after intracellular dialysis with microcystin LR (- - -) for the neuron shown in A. Microcystin LR slowed the rapid repolarization of the action potential and reduced the AHP. C: blockade of BK channels with saturating exposure to IBTX (150 nM) occluded any further increases in firing response gain with the nonhydrolyzable ATP analogue, ATPS. In IBTX, gain was 253 (spikes/s)/nA, and following 50 min of dialysis with ATPS, the gain was nearly identical [246 (spikes/s)/nA].

View larger version (63K): [in this window] [in a new window]   Fig. 11. Summary of manipulations of firing response gain in MVN neurons. Pooled data indicate the mean and SE of the fractional changes in gain evoked by each pharmacological agent. Numbers of neurons for each experiment are denoted in parentheses. The vertical dotted line indicates no change in gain. Cadmium blocks voltage-gated calcium channels, apamin (200 nM) blocks SK-type calcium-activated potassium channels, IBTX (150-200 nM) blocks BK-type calcium-activated potassium channels, nifedipene (250 µM) blocks L-type calcium channels, -agatoxin (200 nM) blocks P/Q-type calcium channels, -CTX (2 µM) blocks N-type calcium channels, Ni2+ (100 µM) and amiloride (250 µM) block T-type calcium channels, KN-62 (10 µM) blocks the Ca2+/calmodulin-dependent protein kinase II, microcystin (10 µM) blocks phosphatases, and ATPS (2 mM) causes irreversible thiophosphorylation of cellular proteins. All manipulations except nifedipene and -agatoxin produced statistically significant differences from controls (P < 0.05).

To determine whether these phosphorylation-mediated increases in gain occur via the BK pathway, we first blocked BK channels with IBTX and then analyzed firing responses in the presence of ATPS. ATPS was used for the occlusion experiment because it acted more rapidly than did microcystin, evoking significant gain increases within 10-20 min of dialysis. To ensure that BK channels were completely blocked, including those containing beta subunits particularly resistant to blockade by IBTX (see Meera et al. 2000), brain stem slices were preincubated for 5-7 h in carbogenated Ringer containing 150 nM IBTX prior to recording. As shown in Fig. 10C, in the presence of saturating IBTX, ATPS failed to evoke changes in firing response gain (1 ± 8% increase over control, n = 4), indicating that increases in protein phosphorylation regulate gain via the BK channel pathway.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Calcium influx modulates neuronal firing responses to synaptic inputs via a variety of mechanisms that act over short and long time scales to influence both ongoing circuit function and long-term adaptive changes in behaviors. To identify calcium-dependent mechanisms that could influence the gain of vestibular reflexes, we examined how calcium influx into vestibular nucleus neurons affects intrinsic firing response gain. We showed that specific voltage-sensitive calcium channels supply calcium to two classes of calcium-dependent potassium channels, SK and BK channels, each of which can modulate firing response gain without compromising response linearity. Furthermore, blockade of ongoing kinase and phosphatase activity revealed that firing response gains are actively controlled by constitutive protein phosphorylation and dephosphorylation. These results indicate that the gain of vestibular reflexes could be modulated by regulation of calcium-dependent potassium channels or their calcium sources.

Gain control by SK currents

The SK channel blocker apamin dramatically increased firing response gains in MVN neurons by reducing the late component of the afterhyperpolarization that follows each action potential. Ca²⁺-dependent K⁺ channels that are sensitive to apamin are members of the recently cloned SK family of channels (Kohler et al. 1996). During repetitive firing in MVN neurons, calcium influx during the action potential activates apamin-sensitive currents within 1-3 ms of the action potential peak (Fig. 3C), suggesting that the Ca²⁺ channels providing the requisite calcium must be located close to SK channels (Sah 1992). Our pharmacological occlusion experiments indicate at least two distinct sources of Ca²⁺ for SK channels in MVN neurons: N-type Ca²⁺ channels and an unidentified source which may be toxin-resistant R-type Ca²⁺ channels (Ellinor et al. 1993).

SK currents were not measured directly in the present study but can be inferred to decay rapidly: if SK currents outlasted the interspike interval, then they would summate during successive action potentials and lead to spike frequency adaptation. In neurons with pronounced spike frequency adaptation, calcium-activated potassium currents last for tens to thousands of milliseconds (Sah 1996). The lack of effect of apamin on spike frequency adaptation in MVN neurons at firing rates 150 Hz suggests that calcium entering during each action potential is cleared within 10 ms, by either diffusion or active buffering. The finding that apamin-sensitive SK channels strongly influence firing response gain but do not influence response linearity or dynamics suggests that the predominant function of these channels in MVN neurons is to modulate gain.

Gain control by BK currents

BK channels demonstrate both voltage and calcium dependence (reviewed in Vergara et al. 1998), endowing them with a potential role in activity- and history-dependent processes. In marked contrast to the SK channel blocker apamin, which caused substantial increases in gain in every MVN neuron, the BK channel blocker IBTX increased firing response gain in some neurons, while having little effect on gain in others. This heterogeneity may have been caused by one or more of the following factors: the differential sensitivity of BK channel subunits to IBTX, differences in other intrinsic currents which partially compensate for the loss of BK currents, or localization of BK channels on dendrites, which can be cut off in brain stem slice preparations. The two distinct responses to IBTX identified in this paper may correspond to neuronal populations containing predominantly IBTX-sensitive or -insensitive BK channels (Reinhart et al. 1989), with large and small gain changes in response to IBTX, respectively. These populations may reflect variability in BK channel subunit composition. Addition of beta subunits confers several physiological changes on BK channels, including increased calcium sensitivity (McManus et al. 1995) and reduced apparent toxin sensitivity (Meera et al. 2000).

We have shown that Ni²⁺- and amiloride-sensitive currents, which presumably block T-type Ca²⁺ channels, are coupled to BK channels in MVN neurons, while currents through L-, N-, and P/Q-type channels are not. Due to the lack of selective antagonists for T-type calcium channels, it was not possible to assess whether these channels provide the only source of calcium for BK channels. Co-localization of specific calcium channels and calcium-dependent potassium channels has been demonstrated previously (Issa and Hudspeth 1994; Marrion and Tavalin 1998; Roberts et al. 1990), although the coupling of specific calcium and potassium channels varies across cell types (Callister et al. 1997; Davies et al. 1996; Perez et al. 1999; Pineda et al. 1998; Williams et al. 1997).

Regulation of gain by kinases and phosphatases

This study not only demonstrates the contribution of various calcium and potassium currents in the control of firing response gain but also shows that gain is actively maintained by constitutive kinase and phosphatase activity, thereby revealing a potential means of gain regulation during motor learning. Kinases and phosphatases are well known to regulate the activity of ion channels by altering channel open probability, expression, kinetics, or gating (Levitan 1994) and have a particularly prominent role in forms of neural plasticity, such as long-term potentiation (reviewed in Nicoll and Malenka 1999; Soderling and Derkach 2000).

Phosphorylation has been shown to modulate many of the ionic currents examined in this study (Doerner et al. 1988; Reinhart et al. 1991). Occlusion experiments employing the CaMK II inhibitor KN-62 and the BK channel antagonist IBTX indicate that CaMK II regulates gain via the BK channel pathway. Possible targets for phosphorylation include the BK channel itself, which can be regulated by CaMK II (Muller et al. 1996; Sansom et al. 2000) or its calcium sources: T-type  subunits are also regulated by CaMK II (Lu et al. 1994). However, our experiments do not distinguish which of these channels, or combination of them, is regulated by CaMK II.

While our results implicate CaMK II phosphorylation of channels in the reduction of gain, experiments using the functional phosphatase inhibitors microcystin LR and ATPS suggest the presence of an additional phosphorylation pathway, which leads to increases in gain. These findings appear contradictory, given that BK channels or their calcium sources appear to be involved in both of these pathways. However, one may postulate that BK channels (or their calcium sources) are regulated at two different phosphorylation sites, one of which results in increased channel activity, leading to gain decreases, while the other results in decreased channel activity, causing increases in gain. These two sites could be phosphorylated by two different kinases. One of these kinases appears to be CaMK II, but we have not identified the other(s). Although inhibitors of protein kinase A and C (PKA and PKC) have not been effective in reducing gain, this may be attributed to differences in kinase isoforms in the vestibular nuclei or to the presence of additional kinases which mediate phosphorylation. Receptor tyrosine kinases, or cGMP-dependent protein kinases, for example, may be responsible for the gain increases seen with phosphatase inhibitors in this study.

Intrinsic ion channel modulation as a possible mediator of behavioral gain changes

The gain of the VOR can be dynamically regulated both by unilateral loss of peripheral vestibular function and by the persistent conjunction of image motion and head movement. The precise cellular conditions that induce these behavioral changes are not known but are thought to involve neuronal activity in both the cerebellar flocculus and brain stem vestibular nuclei (du Lac et al. 1995; Raymond et al. 1996). Neurons in the vestibular nuclei show changes in the gain of their firing responses to head movement that parallel adaptive changes in VOR gain (Lisberger and Pavelko 1988; Lisberger et al. 1994; Newlands and Perachio 1990a; Partsalis et al. 1995).

Although it is typically assumed that changes in synaptic strength underlie these firing response changes and the consequent modulation of VOR gain, regulation of intrinsic firing response gains could also play a role. In fact, recent studies in the vestibular system of rodents indicate that VOR plasticity following loss of peripheral function is mediated in part by changes in intrinsic firing properties of vestibular nucleus neurons (Cameron and Dutia 1997; Him and Dutia 2001; Ris et al. 1995, 2001). Our results provide a candidate substrate for the intrinsic neuronal component of vestibular plasticity. Given that vestibular nucleus neurons transform head movement signals into the appropriate oculomotor commands, modulations of SK or BK channels or of their calcium sources would result in a change in the gain of the VOR. Regulation of these channels or of the kinases and phosphatases that modulate them could be initiated via a number of potential mechanisms. For example, the dramatic drop in spontaneous firing of vestibular nucleus neurons following loss of the vestibular function (McCabe and Ryu 1969; Newlands and Perachio 1990a; Precht et al. 1966; Ris et al. 1995; Smith and Curthoys 1988) would lead to decreases in calcium influx that could reduce SK and BK currents directly or indirectly via CaMK II activity. Due to their capacity to regulate channel function on many time scales, kinases and phosphatases and their actions on intrinsic membrane currents provide intriguing potential substrates for the cellular changes underlying adaptive gain control in vestibular reflexes.