A Kv3-like persistent, outwardly rectifying, Cs⁺-permeable, K⁺ current in rat subthalamic nucleus neurones

Abstract

1. A persistent outward K⁺ current (I_PO), activated by depolarization from resting potential, has been identified and characterized in rat subthalamic nucleus (SThN) neurones using whole-cell voltage-clamp recording in brain slices.

2. I_PO both rapidly activated (τ= 8 ms at +5 mV) and deactivated (τ= 2 ms at −68 mV), while showing little inactivation. Tail current reversal potentials varied with extracellular K⁺ concentration in a Nernstian manner.

3. Intracellular Cs⁺ did not alter either I_PO amplitude or the voltage dependence of activation, but blocked transient (A-like) outward currents activated by depolarization. When extracellular K⁺ was replaced with Cs⁺, I_PO tail current reversal potentials were dependent upon the extracellular Cs⁺ concentration, indicating an ability to conduct Cs⁺, as well as K⁺.

4. I_PO was blocked by Ba²⁺ (1 mm), 4-aminopyridine (1 mm) and tetraethylammonium (TEA; 20 mm), with an IC₅₀ for TEA of 0.39 mm.

5. The I_PO conductance appeared maximal (38 nS) at around +27 mV, half-maximal at −13 mV, with the threshold for activation at around −38 mV.

6. TEA (1 mm) blocked the action potential after-hyperpolarization and permitted accommodation of action potential firing at frequencies greater than around 200 Hz.

7. We conclude that I_PO, which shares many characteristics of currents attributable to Kv3.1 K⁺ channels, enables high-frequency spike trains in SThN neurones.

The subthalamic nucleus (SThN) is a small, densely packed nucleus situated in the ventral midbrain. It contains neurones making glutamatergic projections primarily to the external segment of the globus pallidus and substantia nigra pars reticulata (Albin et al. 1989; Parent & Hazrati, 1995). The importance of these projections to other basal ganglia nuclei has become increasingly appreciated in recent years. Thus, not only can lesions of the SThN produce the hyperkinetic syndrome of ballism (Guridi & Obeso, 1997), but also the akinesia in animal models of Parkinson's disease can be ameliorated by SThN lesions (Bergman et al. 1990; Wichmann et al. 1994b). Moreover, surgical SThN lesions have proved beneficial in Parkinson's disease (Krack et al. 1999) and similar benefits may be obtained by bilateral high-frequency stimulation of the SThN (Benazzouz et al. 1993; Limousin et al. 1995; Krack et al. 1999), which has been proposed to inactivate SThN neurones by depolarization block (Benazzouz et al. 1993; Limousin et al. 1995). In animal models of akinesia or dyskinesia of basal ganglia origin, cells in several of the component nuclei, including the SThN, display an increased tendency to fire synchronously which, in parkinsonian models, is associated with an increased tendency to fire in bursts (Bergman et al. 1994; Hassani et al. 1996; Wichmann & Delong, 1996). This change in firing pattern probably results from increased input convergence, reflecting breakdown of separation of ‘channels’ of sensorimotor information, processed in parallel within the cortico-basal ganglia-thalamo-cortical circuit (Alexander et al. 1990; Wichmann & Delong, 1996; Brown & Marsden, 1998). The pivotal role of the SThN in this circuit, where its activation seems able to curtail execution of ongoing motor tasks (Delong, 1990), provides the context in which an understanding of the mechanisms that pattern SThN neurone firing might be applied to better understand – and manipulate – extrapyramidal motor function.

In our examination of the cellular mechanisms that govern firing patterns of SThN neurones we, and others (Beurrier et al. 1999; Bevan & Wilson, 1999), have employed whole-cell patch recording from brain slice preparations. In the course of our studies we have observed a large persistent outward current (I_PO) activated by depolarization from resting potential, whose properties we describe here. SThN neurones are capable of firing at frequencies in excess of 300 Hz in vitro (Nakanishi et al. 1987), and can respond with transient, high-frequency bursts in association with limb or eye movements in vivo (Matsumara et al. 1992; Wichmann et al. 1994a). Other fast-firing neuronal types, including those of the medial nucleus of the trapezoid body (Brew & Forsythe, 1995; Wang et al. 1998b), and cortical (Erisir et al. 1999) and hippocampal (Du et al. 1996; Martina et al. 1998) interneurones, also have persistent outwardly rectifying K⁺ currents, in many cases attributed to expression of the K⁺ channel Kv3.1, which serve to facilitate rapid action potential firing. We present evidence for I_PO being an outwardly rectifying K⁺ current that operates to restrict action potential duration and accommodation, enabling sustained firing at high frequencies. Our biophysical and pharmacological description of its characteristics invites comparison with the properties of expressed K⁺ channels of known molecular sequence. Some of these results have been reported previously in abstract form (Wigmore & Lacey, 1999).

METHODS

Parasagittal slices of rat ventral midbrain were prepared from Wistar rats (70–130 g) as described previously (Miyazaki & Lacey, 1998). Briefly, animals were anaesthetized using 3.5 % halothane (‘Fluothane’) in oxygen, followed by cervical dislocation, in accordance with the Animals (Scientific Procedures) Act, 1986, UK. Brains were quickly removed (within 3 min) into ice-cold artificial cerebrospinal fluid (aCSF) comprising (mm): NaCl, 126; KCl, 3; CaCl₂, 1; MgCl₂, 5; NaH₂PO₄, 1.2; D-glucose, 20; NaHCO₃, 26; and lactate, 4. Slices of 300 μm thickness were cut on a DTK-1000 tissue slicer (Dosaka, Kyoto, Japan) and transferred to a storage chamber or placed directly under a microscope (see below), where they were submerged in aCSF comprising (mm): NaCl, 126; KCl, 2.5; CaCl₂, 2.4; MgCl₂, 1.3; NaH₂PO₄, 1.2; D-glucose, 10; NaHCO₃, 26; continuously bubbled with 95 % O₂-5 % CO₂ and warmed to 30°C. Slices were stored thus for at least 30 min prior to commencing recording.

Recording pipettes (resistance, 3–5 MΩ) were pulled from 1.2 mm external diameter borosilicate glass (Clark ElectroMedical, Pangbourne, UK) on a horizontal puller (P-97, Sutter Instruments). Pipettes were filled with a solution containing (mm): either potassium or caesium gluconate, 125; NaCl, 10; CaCl₂, 1; MgCl₂, 2; BAPTA, 1; Hepes, 10; buffered to pH 7.3–7.4 with either KOH or CsOH. ATP (2 mm) and GTP (0.3 mm) were added on the day of recording from frozen aliquots dissolved in water. For recordings, slices were placed on the base of a bath (volume, < 0.5 ml) and continuously superperfused (> 2 ml min⁻¹) with aCSF of the same composition as that used for the storage of slices. Experiments were performed at 32°C, unless stated otherwise. The slice was stabilized by placing a titanium grid (Agar Scientific, UK) across the SThN and holding it down with flattened platinum wire (Goodfellow, UK). Drugs were applied by a gravity-fed system in which one solution was exchanged for another differing only in the one constituent; the dead time before the bath was reached was 20–30 s. CsCl, tetraethylammonium chloride (TEA), 4-aminopyridine (4-AP), BaCl₂ and 4-ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidium chloride (ZD7288) were purchased from Sigma Chemicals (Poole, UK), and tetrodotoxin (TTX) was obtained variously from Sigma, Tocris Cookson Ltd (Bristol, UK) or Alamone Labs Ltd (Jerusalem, Israel). The ‘low-Ca²⁺ aCSF’ contained 0.24 mm CaCl₂, with [MgCl₂] raised to 10 mm.

Single cells were visualized for recording using infrared differential interference contrast (‘Nomarski’) optics with an Olympus BX50WI microscope with ×40 objective, connected to a CCD camera (KP-M1, Hitachi, Tokyo, Japan) and Sony monitor. The image intensity was boosted using a digital constrast enhancer (CE-4, Brian Reece Scientific, Newbury, UK). The SThN could be observed using a low-power objective as a small, dark region on the ventral surface of the brain, bounded by the internal capsule, thalamus and substantia nigra. Whole-cell recordings were made by advancing the recording pipettes slowly into the slice under visual control using a piezoelectric manipulator (PCS-5000, Burleigh Instruments, Fishers, USA), mounted on a Gibraltar stage (Burleigh Instruments). Following formation by suction of a gigaohm seal, the whole-cell recording configuration was attained by applying additional negative pressure to the pipette. All voltage-clamp recordings were made using an Axopatch 1D amplifier, while current-clamp measurements were made with an Axopatch 200A amplifier (both Axon Instruments, CA, USA). Signals were filtered at 1–5 kHz, and data were acquired using pCLAMP software (version 6; Axon Instruments) running on a PC. Series resistances were in the range 4–20 MΩ and were compensated by 60–80 %. Whole-cell capacitance was in the range 10–20 pF when K⁺ was the principal intra-pipette cation, and 15–30 pF when Cs⁺ was used. Series resistance was monitored at least once every 5 min during recordings and cells were discarded if the uncompensated series resistance rose above 25 MΩ. At the end of the experiment the pipette was removed clear of the cell surface whilst still in the superfusate and any DC offset potential noted; membrane potentials were corrected for this offset retrospectively. Membrane potentials are generally reported without compensation for liquid junction potentials (LJPs), except for those given in the Summary, and where stated specifically elsewhere, when a correction of +8 mV was made. Membrane currents were subjected to leak subtractions to remove the passive component of the membrane conductance; this was performed on-line using a P/4 or P/5 protocol using pCLAMP, and currents are shown leak subtracted unless otherwise stated in the text.

Experimental data were extracted and analysed using Clampfit (pCLAMP suite). Further data analysis was performed with MS-Excel 98 and plotted using Origin 5.0 (MicroCal). The depression of I_PO by TEA (expressed as a percentage) was quantified as [1 – (I_TEA/I_control)]× 100, where I_TEA and I_control are the mean I_PO amplitudes in TEA and under control conditions, respectively, calculated as the mean from three separate trials. To construct the concentration-effect plot, the following equation was used for fitting (by Origin), and iterated until the χ² value reached a minimum:

where A₁ and A₂ are the minimum and maximum y values, respectively, p is the slope and x_o the IC₅₀. Curve fitting was performed off-line using Clampfit, except for current activation curves, for which Origin was employed. The equation used for the activation plot was a Boltzmann equation of the form:

where A₁ and A₂ are the minimum and maximum membrane conductances, respectively, x_o is the midpoint voltage and dx is the slope. All pooled data are expressed as means ± standard error of the mean (s.e.m.).

RESULTS

Data presented here were obtained from recordings from 56 cells, using either caesium gluconate- (n = 33) or potassium gluconate (n = 23)-based pipette solutions.

Characteristics of SThN neurones using potassium gluconate-based pipette solution

SThN neurones commonly showed spontaneous action potential firing after whole-cell access was gained with potassium gluconate-based pipette solution (Fig. 1A), with a mean firing frequency in the first 2 min of recording of 19.5 ± 6.4 Hz (n = 7). However, spontaneous firing was commonly lost after several minutes of whole-cell dialysis. Under these conditions the input resistance was 520.8 ± 78.1 MΩ (n = 10), estimated from a voltage step from −60 to −70 mV under voltage-clamp conditions. Cells showed a small time-dependent inward current (attributable to the hyperpolarization-activated inward current, I_h) when stepped to membrane potentials negative to −75 mV under voltage clamp (Fig. 1B). They also often demonstrated a region of negative slope conductance positive to −45 mV (Fig. 1B and C). These properties correspond to other descriptions of SThN neurones (Nakanishi et al. 1987; Beurrier et al. 1999; Bevan & Wilson, 1999). In the presence of TTX (0.5–1 μm) and low-Ca²⁺ aCSF, spontaneous firing and inward currents activated by depolarization were abolished. Under these conditions, outward currents activated by depolarization were examined using voltage-clamp recording.

Figure 1. Basic properties of SThN neurones recorded with potassium gluconate-based pipette solution.

A, voltage recording showing spontaneous action potential firing (at 20 Hz) at rest. Full spike amplitude is inaccurately displayed due to sampling frequency limitations. B, membrane currents (upper records) evoked in another SThN neurone by voltage steps from −60 mV to potentials in the range −140 to −40 mV (lower records). C, current-voltage relationships displaying both the instantaneous current (▪) and the steady-state current (•) derived from the voltage steps in B– the difference may be attributed to I_h. Currents were not leak subtracted.

Depolarization activates a persistent outward K⁺ current

Depolarizing steps from holding potentials in the region of −50 to −60 mV resulted in the activation of a large and sustained outward current that activated at potentials positive to around −30 mV (Fig. 2). This persistent outward current (I_PO) showed little inactivation over several hundred milliseconds, even when the membrane potential was stepped to 0 mV. However, it is likely that with potassium gluconate-based pipette solution this current was contaminated with other outward currents. Indeed, a transient A-like component was observed, particularly when the membrane was depolarized from a more negative potential (−80 mV; Fig. 2).

Figure 2. Depolarizing voltage steps elicited a persistent outward current, I_PO.

A, steps from a holding potential of −80 mV (lower records) resulted in an outward current with both a transient and steady-state component (upper records). B, changing the holding potential to −60 mV (lower records) markedly reduced the transient component of the outward current (upper records). Currents were recorded from the same cell using a potassium gluconate-based pipette solution in low-Ca²⁺ aCSF containing TTX (0.5 μm).

In order to study I_PO further, tail currents observed on current deactivation were examined. Families of tail currents were elicited by stepping the membrane potential from −50 mV to between −10 and −16 mV for 500 ms (to allow transient currents to inactivate), before hyperpolarizing to potentials in the range −20 to −80 mV. Tail current amplitudes were estimated as the difference between the instantaneous current and steady-state current, as illustrated in Fig. 3A. To facilitate these measurements, ZD7288 (50–100 μm), a blocker of I_h, was added to the superfusate, and experiments were performed at room temperature (22–25°C). Using extracellular K⁺ concentrations ([K⁺]_o) of 5, 10, 15 and 30 mm, mean reversal potentials (V_rev) of −62.0 ± 1.8, −53.7 ± 0.9, −43.8 ± 1.5 and −33.0 ± 2.8 mV (n = 3–6 cells), respectively, were obtained (Fig. 3B and C). Indeed, after LJP correction, these mean values were within +5 mV of those predicted by the Nernst equation for a pure K⁺ conductance, except that for the lowest [K⁺]_o used (5 mm; +11 mV difference), where V_rev was least easy to measure (see Fig. 3A). The relationship between V_rev and [K⁺]_o, particularly at the higher [K⁺]_o used, also closely approximated that predicted by the Nernst equation if I_PO was a pure K⁺ conductance (Fig. 3C).

Figure 3. Tail current reversal potentials indicate that I_PO is mediated by K⁺ ions.

A, tail currents recorded in 4 different [K⁺]_o as indicated. I_PO was initially activated with a 500 ms step to −14 mV, which also allowed any transient currents to inactivate (not shown), and subsequently deactivated by the voltage protocol shown in the lower panel. Tail current amplitude is the difference between the instantaneous (Inst) and steady-state (SS) current, shown for 30 mm[K⁺]_o, recorded at the start and end of the hyperpolarizing step, respectively. Currents were leak subtracted and recorded at room temperature in low-Ca²⁺ aCSF additionally containing TTX (0.5 μm) and ZD7288 (50 μm). Records are all from the same cell. B, plot of tail current amplitude versus membrane potential (V_m) derived from records in A. C, pooled data (means ±s.e.m.) from several such experiments (numbers in parentheses) showing estimated V_rev plotted against [K⁺]_o on a semilogarithmic scale. The slope predicted by the Nernst equation for a pure K⁺ conductance (59 mV per 10-fold change in [K⁺]_o) is shown by the dashed line, and closely approximates that of the data points.

I_PO can be carried by Cs⁺

In an attempt to block I_PO, recordings were made with a caesium gluconate-based pipette solution, as many K⁺ channels are blocked by internal Cs⁺ ions. Under these conditions, the majority (> 90 %) of cells showed spontaneous firing immediately after whole-cell access was gained, but within 1 min depolarized to and subsequently maintained a potential of between −20 and −30 mV, presumably a result of blockade of K⁺ channels contributing to the leak conductance. Again, in the presence of TTX (0.5 μm) and low-Ca²⁺ aCSF, to block inward currents, I_PO was still evident (Fig. 4). Comparison of pooled data from cells recorded under these conditions (n = 9) and data from cells recorded with potassium gluconate-based pipette solutions (n = 7) showed that neither I_PO amplitude nor the voltage dependence of its activation was altered by internal Cs⁺ (Fig. 4B). The persistence of I_PO in the absence of internal K⁺ ions indicated that this current could also be carried by Cs⁺ ions. This was investigated further by examination of tail currents using Cs⁺-based pipette solutions, and with extracellular K⁺ replaced with Cs⁺. Due to the improved voltage control of the cell membrane when caesium gluconate was used in the pipette, these experiments were performed at 32°C. The membrane potential was depolarized to between −10 and −26 mV to activate I_PO and then stepped to potentials in the range −100 to −106 mV in the presence of [Cs⁺]_o (10, 20 or 30 mm) in K⁺-free aCSF (Fig. 5A). The use of extracellular Cs⁺ blocked I_h, obviating the need for ZD7288. Replacement of K⁺ with Cs⁺ ions as charge carrier gave a slope value of +45 mV per 10-fold change in [Cs⁺]_o (estimated by linear regression from data pooled from 7 experiments; Fig. 5B). The Nernst equation predicts this slope value to be +60 mV for a completely Cs⁺-selective conductance, suggesting that under these conditions Cs⁺ ions are the principal, but perhaps not the only, charge carrier. The values for the tail current reversal potentials for each [Cs⁺]_o examined (Fig. 5B), once corrected for LJP, deviated increasingly negatively from the predicted equilibrium potential for Cs⁺ (E_Cs) with higher [Cs⁺]_o (by up to −8 mV with 30 mm). Thus in addition to using K⁺ as the major permeant ion, it appears that I_PO can equally well carry Cs⁺, depending on cation availability.

Figure 4. I_PO persists in cells recorded using caesium gluconate-filled pipettes.

A, upper records: currents resulting from the voltage steps shown in the lower panel in a cell recorded with a caesium gluconate-filled pipette in low-Ca²⁺ aCSF containing TTX (0.5 μm). B, data pooled from several cells obtained in experiments such as that in A comparing the peak I_PO recorded using potassium gluconate- (▴; n = 7) or caesium gluconate (▪; n = 9)-filled pipettes over a range of membrane potentials. There was no difference in activation threshold or current amplitude. Data were obtained under identical conditions except for the intracellular cation; slices were bathed in low-Ca²⁺ aCSF containing TTX (0.5 μm).

Figure 5. I_PO tail current reversal potentials indicate that Cs⁺ can also act as the principal permeant cation.

A, tail currents recorded from the same cell using a caesium gluconate-filled pipette, in 10, 20 and 30 mm[Cs⁺]_o (upper three panels), produced by the voltage steps indicated in the bottom panel. The tail currents reversed polarity at more depolarized potentials as [Cs⁺]_o was increased. Currents were leak subtracted and recorded at 32 °C, in the presence of low-Ca²⁺ aCSF and TTX (0.5 μm), in the absence of extracellular K⁺. B, pooled data (means ±s.e.m.) from 7 such experiments. Tail current reversal potentials were dependent upon [Cs⁺]_o, showing a 45 mV shift per 10-fold change in [Cs⁺]_o.

Pharmacology of I_PO

To characterize I_PO further, its sensitivity to various K⁺ channel blockers was tested. Using low-Ca²⁺ aCSF and caesium gluconate-based pipette solutions, the membrane potential was held at around −50 mV and stepped by positive increments to around +10 mV to activate I_PO. The effect of three K⁺ channel blockers – TEA (20 mm), BaCl₂ (1 mm) and 4-AP (1 mm) – were evaluated by measuring their effect on the amplitude of I_PO activated at +10 mV. TEA (20 mm) reduced I_PO by 98.1 ± 1.4 % (n = 3; Fig. 6A), while BaCl₂ (1 mm) and 4-AP (1 mm) depressed I_PO by 76.4 ± 8.0 and 90.0 ± 2.2 %, respectively (n = 3 in both cases; data not shown). Some washout of the effect of TEA was observed, but this was not complete (Fig. 6A). Blockade with TEA appeared to be voltage independent (Fig. 6B). In an additional series of experiments the potency of TEA was determined. Prior to delivering a test voltage step from around −50 mV to around +10 mV once every 15 s, from which I_PO was measured, the membrane potential was held at −80 mV for 1 s, followed by 1 s at −50 mV. Cells were tested with cumulative applications of 0.1, 1 and 10 mm TEA (Fig. 6C) or 0.3, 3 and 20 mm TEA (n = 3 cells in each group). The degree of depression of (leak-subtracted) current amplitude relative to control was determined and pooled to derive a mean value for each TEA concentration tested. These could be fitted to a logistic concentration-effect curve with a slope value of 0.97, suggesting that TEA is binding to a single site, with an IC₅₀ of 0.39 mm (Fig. 6D).

Figure 6. I_PO is sensitive to block by TEA.

A, outward currents resulting from depolarizing steps from −50 to +15 mV in 5 mV increments under control conditions (a), in the presence of TEA (20 mm; b) and after washout (10 min; c). Recordings were made with caesium gluconate-filled pipette solution in low-Ca²⁺ aCSF containing TTX (0.5 μm). B, resulting current-voltage relationships from data in A for I_PO in the absence (▪) and presence (•) of TEA (20 mm). Block by TEA appeared to be voltage independent. C, plot of I_PO amplitude during cumulative application of TEA (0.1, 1 and 10 mm) during the periods denoted by the bars. Inset shows single leak-subtracted current records (upper panel), corresponding to points a-d, which were evoked every 15 s by 200 ms voltage steps from −52 to +8 mV (lower panel). Recordings were made using caesium gluconate-based pipette solutions, in low-Ca²⁺ aCSF containing TTX (0.5 μm). Not shown is a 1 s conditioning step to −82 mV that preceded a 1 s step to −52 mV, from which the test depolarization was then made. D, concentration-effect curve for the depression of I_PO by TEA pooled from 6 experiments such as that in A. Each point represents the mean data from 3 cells, recorded from different slices. The logarithmic curve fitted to the data had a slope value of 0.97, and gave an IC₅₀ for TEA of 0.39 mm.

I_PO activation

Using caesium gluconate-based pipette solution in the presence of [Cs⁺]_o (10 mm) and no extracellular K⁺, I_PO was elicited by depolarizations in the approximate range −50 to +60 mV in six cells. The conductance (G_PO) for each voltage step was calculated using the peak (non-inactivated) leak-subtracted current, and normalized by dividing by V–E_Cs to account for the driving force on the ion; E_Cs was assumed to be −66 mV, as predicted by the Nernst equation. This permitted a Boltzmann plot of peak G_PO with respect to membrane voltage, which describes the voltage dependence of I_PO activation (see Methods; Fig. 7). The threshold for current activation occurred at around −30 mV. However, the plot obtained did not reach an asymptote at positive potentials, but rather the conductance continued to increase with depolarization (Fig. 7B). This may reflect an inability to reliably control membrane potential in the dendritic tree, such that at somatic potentials sufficient to activate all proximal channels, other more distal channels might still not be activated. As these reach activation threshold they would thus contribute to a steadily increasing conductance value. Indeed, even approaching +100 mV an asymptote was not obtained (data not shown). As the activation curve in Fig. 7B became linear at potentials positive to a mean value of +35 mV, we took this as the potential at which G_PO (at least within the soma) was maximal, at which it was 37.8 ± 1.9 nS. The Boltzmann fit gave half-maximal activation at −5 mV.

Figure 7. Voltage dependence of I_PO activation.

A, a family of leak-subtracted I_PO (upper panel) evoked using the voltage protocol (5 mV increments) shown in the lower panel. The cell was recorded with caesium gluconate-based pipette solution in low-Ca²⁺ aCSF containing TTX (0.5 μm) and Cs⁺ (10 mm), with zero K⁺. B, plot of data pooled from 6 such experiments, with points representing means ±s.e.m. of both I_PO conductance and membrane potential. Leak-subtracted current data from experiments such as those in A were normalized for driving force by dividing by V–E_Cs, with E_Cs derived from that predicted by the Nernst equation. Data points were fitted with a Boltzmann function, assuming maximal somatic I_PO activation at +35 mV (see text). Half-maximal activation was at −5 mV, with the threshold at around −30 mV.

Kinetics of I_PO

In order to facilitate the study of the activation kinetics of I_PO, caesium gluconate-based pipette solutions were used, with an extracellular medium containing zero K⁺ and 5 mm Cs⁺ (n = 3). The time constant (τ) was calculated from a single exponential fit to the peak of the current data, using non-leak-subtracted current to avoid the possibility of leak subtraction distorting the kinetics. The estimated τ for activation decreased from approximately 28 to 8 ms in the range −22 to +13 mV (n = 3; Fig. 8A), predicting significant I_PO activation during the depolarizing phase of an action potential.

Figure 8. Activation and deactivation kinetics of I_PO.

A, activation. Aa, example of a family of currents (upper panel) generated by depolarizing steps (lower panel), from which activation kinetics were derived. Single exponential fits (thick overlays) were made to the peak of the rising phase of the currents (not leak-subtracted) to extract the time constant (τ). Cell recorded with a caesium gluconate-filled pipette in low-Ca²⁺ aCSF containing TTX (0.5 μm) and Cs⁺ (5 mm). Ab, data pooled from 3 such cells showing the relationship between the activation time constant τ and membrane potential. The greater the depolarization, the more rapidly I_PO activated. B, deactivation. Ba, leak-subtracted I_PO tails from a single cell (upper panel) resulting from a series of hyperpolarizing steps from −12 mV (lower panel), preceded by holding at −12 mV for 500 ms (to inactivate any transient currents; not shown). Cell recorded with a potassium gluconate-filled pipette at room temperature in low-Ca²⁺ aCSF containing ZD7288 (50 μm), TTX (0.5 μm) and K⁺ (5 mm). Tail currents were well fitted by a single exponential function (continuous thick lines). Bb, mean values of I_PO deactivation time constant τ derived from 4 such experiments. The greater the membrane repolarization, the faster I_PO deactivated.

The voltage dependence of deactivation was quantified by examination of tail currents recorded using intracellular potassium gluconate and extracellular K⁺ (5 mm) at room temperature, in aCSF additionally containing ZD7288 (50 μm). Cells were depolarized from around −50 mV to between −10 and −16 mV for 450 ms, and then hyperpolarized in 5 mV incremental steps (Fig. 8B). The currents were leak subtracted and the tails could be well fitted using a single exponential function. Data pooled from four cells indicated that τ decreased linearly with hyperpolarization, being approximately 2 ms at around −60 mV (Fig. 8B), showing that I_PO deactivates rapidly at around resting membrane potential.

To study inactivation the membrane potential was held at between −60 and −50 mV and stepped (for up to 4 s) in depolarizing increments up to around +50 mV (using caesium gluconate-based pipette solutions and either extracellular Cs⁺ (5 mm; n = 3) or K⁺ (2.5 mm; n = 3)). Inactivation of I_PO was very slow and could not be well fitted by a single exponential function (Fig. 9A). In an attempt to quantify the inactivation, the peak current amplitude was divided by the current amplitude 3 s after the peak. This ‘inactivation ratio’ (which was 1.66 ± 0.14 at +22.3 ± 0.3 mV; n = 4) appeared to be unrelated to membrane potential (Fig. 9Ab). The slow inactivation of I_PO that developed during a 1 s depolarizing step to around +10 mV could be relieved following interruption by brief (20–200 ms duration) hyperpolarizing steps to potentials in the range 0 to −100 mV (Fig. 9B). The greater the amplitude of the hyperpolarizing steps, the greater the deinactivation. Thus, for 20 ms steps to −90 and −60 mV, the ratios of peak pre-step (non-inactivated) current to peak post-step (deinactivated) current were 1.11 ± 0.03 and 1.20 ± 0.03, respectively (n = 5 in each case; P < 0.01 by Student's paired t test). In addition, longer hyperpolarizing steps gave greater relief from inactivation (Fig. 9B), such that steps to −90 mV for 20 or 200 ms resulted in peak control/peak test current ratios of 1.11 ± 0.03 and 0.99 + 0.02, respectively (n = 5; P < 0.01 by Student's paired t test). This relief of inactivation by brief hyperpolarization suggests that inactivation is a property of the channels carrying I_PO, rather than due to gradual extracellular ion accumulation reducing the driving force on the permeant ion.

Figure 9. Inactivation and deinactivation kinetics of I_PO.

Aa, non-exponential inactivation of I_PO could be observed (upper panel; not leak subtracted) during sustained depolarizing voltage steps (lower panel). Recording made with a caesium gluconate-filled pipette in low-Ca²⁺ aCSF containing TTX (0.5 μm) and Cs⁺ (5 mm); only the first 1 s of the step is shown. Ab, the ratio of peak current to current at 3 s after the peak for the cell in Aa showed no clear relationship to membrane potential. B, the inactivation of I_PO can be relieved by hyperpolarization. Ba, upper panel: overlaid currents from 12 trials, each consisting of an inactivating control (pre-step) current, interrupted by a transient hyperpolarizing step (200 ms) of varying magnitude. Currents truncated negative to +0.5 nA. Lower panel: voltage protocol, where dashed lines represent steps producing the currents that are truncated in the upper panel. Bb, plot of the ratio of peak pre-step current to peak post-step current vs. the step potential for steps of 20 (▪) and 200 ms (•) duration from the cell in Ba. Deinactivation is more apparent the greater the amplitude and duration of the hyperpolarizing step. Relief from inactivation (ratio = 1) occurred at around −70 mV for a 200 ms step. Currents were not leak subtracted.

Influence of I_PO on action potential generation

We sought to identify the role played by I_PO in influencing the behaviour of SThN neurones in current-clamp mode by using TEA (1 mm), which would be expected to block I_PO by around 80 %. TEA (1 mm) caused a pronounced block of the initial phase of the action potential after-hyperpolarization, and also increased the action potential duration by around 50 % (Fig. 10A). This was in contrast to low-Ca²⁺ aCSF, which was without effect on either of these variables (not shown), indicating that this effect was not due to an action of TEA upon I_C, the TEA-sensitive Ca²⁺-dependent K⁺ current mediated by high-conductance Ca²⁺-sensitive K⁺ (BK) channels (Pennefather et al. 1985; Lo et al. 1998). When sustained high-frequency action potential firing was induced by a series of depolarizing current injection steps (300 ms), under control conditions spike frequency adaptation was minimal (see also Bevan & Wilson, 1999), and accommodation was only observed with sustained firing at frequencies higher than around 350 Hz (Fig. 10B). However, in TEA (1 mm), firing rates in excess of around 200 Hz could not be sustained, due to accommodation, for more than a few tens of milliseconds (Fig. 10B). These results suggest that I_PO contributes to the brisk repolarization of the action potential, and that this in turn aids sodium channel deinactivation which is necessary to permit sustained firing at frequencies above around 200 Hz without accommodation.

Figure 10. TEA, at a concentration (1 mm) expected to block I_PO by around 80 %, affects single spike characteristics and increases accommodation at high firing frequencies.

A, two action potentials from the same cell, showing an increase in duration, and reduction in after-hyperpolarization amplitude, in TEA (1 mm). B, TEA (1 mm) impairs high-frequency action potential firing by enhancing accommodation. Two sets of records of membrane potential from the same cell in control conditions (upper records) and in TEA (1 mm; lower records). Records show, for both conditions, the result of injecting depolarizing current (100, 300 and 500 pA) for 300 ms while the cell was at rest. Higher frequencies of action potential firing thus elicited could be better sustained under control conditions (200 Hz, middle panel; initially around 300 Hz in right-hand panel), whereas TEA promoted accommodation, supporting a role for I_PO in action potential repolarization and consequent sodium channel deinactivation.

DISCUSSION

Distinctive characteristics of I_PO

We have demonstrated that rat SThN neurones possess a rapidly activating, very slowly inactivating, outwardly rectifying ‘persistent outward’ K⁺ current that activates positive to around −30 mV, which we have termed I_PO. Other membrane currents active in this range of membrane potentials were abolished by blockers of voltage-gated Ca²⁺ and Na⁺ channels or, in the case of transient A-like K⁺ currents, either by exploiting their rapid inactivation at potentials positive to around −60 mV, or by intracellular dialysis with Cs⁺. Indeed, after 500 ms at membrane potentials positive to around −16 mV, there appears to be only a single component of voltage-dependent membrane current active, as the tail currents seen on repolarization could be fitted by a single exponential function, even with potassium gluconate-containing recording pipettes. The dependence of these tail current reversal potentials on [K⁺]_o was consistent with I_PO being a K⁺ current, although there was some deviation from predicted Nernstian behaviour, particularly at lower [K⁺]_o. This might be because (1) any extracellular accumulation of extruded K⁺ ions through the channel will tend to shift the reversal potential in a positive direction; this would be (and was) particularly evident with low [K⁺]_o. (2) In low [K⁺]_o the currents reverse at potentials at which I_PO deactivation is most rapid, compromising accurate measurement of current amplitude. In addition, the reversal potentials could be influenced by weak permeability of the channel to other ions, and to an influence of incomplete block of other currents.

On discovering that intracellular dialysis with Cs⁺ did not affect I_PO, this condition was used to study several of the other properties of I_PO of interest because of block of A-type currents, and the possibility of better voltage control. Indeed, under conditions in which both intracellular and extracellular K⁺ were nominally replaced by Cs⁺, I_PO current tails displayed a near-Nernstian dependence upon [Cs⁺]_o, confirming the channel's ability to conduct Cs⁺. While Cs⁺ permeability of K⁺ channels is not commonly observed, there are precedents for this in slowly inactivating outwardly rectifying currents in other cell types, such as the avian nucleus magnocellularis (Rathouz & Trussell, 1998), bullfrog sympathetic neurones (Block & Jones, 1997), chick dorsal root ganglion neurones (Trequattrini et al. 1996) and in cells expressing human Kv1.5 channels (Fedida et al. 1999). Many other outwardly rectifying K⁺ currents described elsewhere display block by TEA, Ba²⁺ and 4-AP, as shown for I_PO. However, the high sensitivity of I_PO to block by extracellular TEA, with an IC₅₀ of around 0.4 mm, is particularly noteworthy, resembling instances of slowly inactivating outwardly rectifying currents with IC₅₀ values for TEA block in the range 0.1–1 mm reported in the neurones of medial nucleus of the trapezoid body (Brew & Forsythe, 1995; Wang et al. 1998b), thalamic relay neurones (Huguenard & Prince, 1991), globus pallidus (Baranauskas et al. 1999) and in cortical (Erisir et al. 1999) and hippocampal (Martina et al. 1998) interneurones.

The kinetic analysis of I_PO is likely to be compromised by neuronal geometry, and the potential difficulty in voltage clamping distal parts with sufficient rapidity. However, under the conditions used, it proved possible to fit the kinetics of both activation and deactivation by a single exponential function, minimizing such concerns. The voltage dependence of the activation time constant decreased as the membrane was stepped to more depolarized potentials, and at the peak of an action potential, where the activation time constant τ will be less than 10 ms (at 32°C), I_PO would be expected to activate significantly. Deactivation was also both fast and voltage dependent, with a time constant of approximately 2 ms at −60 mV (at room temperature). Thus, while it would be predicted that I_PO would contribute to action potential repolarization, its influence would be short-lived on the return of the membrane potential to rest, and thus would not impair generation of subsequent action potentials, as long as the interspike interval was in excess of 2–3 ms. I_PO inactivation was very slow and non-exponential and, given that in a spontaneously active neurone I_PO will serve to repolarize the cell and thus not permit inactivation to occur, this characteristic is unlikely to be of great physiological relevance. However, as it could be reversed by a transient deinactivating hyperpolarization, inactivation does appear to be a real property of the current. Again, the voltage dependence and kinetic features of I_PO are comparable to other relatively non-inactivating outward rectifiers described in the neurones of medial nucleus of the trapezoid body (Brew & Forsythe, 1995; Wang et al. 1998b), avian nucleus magnocellularis (Rathouz & Trussell, 1998), thalamic relay neurones (Huguenard & Prince, 1991), islands of Calleja (Halliwell & Horne, 1995), globus pallidus (Baranauskas et al. 1999), and interneurones in cortex (Erisir et al. 1999) and hippocampus (Martina et al. 1998).

I_PO is likely to result from expression of Kv3.1

Outwardly rectifying K⁺ channels can be formed from a number of gene families, namely Kv1, Kv2, Kv3, eag (ether-à-go-go) and KCNQ (Pongs, 1992; Shi et al. 1997, 1998; Robertson, 1997; Wang et al. 1998a), although their properties can be complicated by the coexpression of heteromers or auxiliary subunits (Robertson, 1997; Mathie et al. 1998; Blaine & Ribera, 1998). Nonetheless, in seeking a molecular correlate for the channels carrying I_PO, Kv2, eag and KCNQ are unlikely candidates due to differences in their kinetics and/or TEA sensitivity (Pongs, 1992; Shi et al. 1997, 1998; Robertson, 1997; Wang et al. 1998a). Message encoding members of both the Kv1 and Kv3 families has been detected in the SThN (Perney et al. 1992; Weiser et al. 1994; Verma-Kurvari et al. 1997). As both Kv1 and Kv3 channels show relatively fast deactivation kinetics and TEA sensitivity it is harder to differentiate conclusively between them. However, it has been suggested, for Kv1.1, 1.2, 1.3 and 1.5 at least, that their deactivation time constants at −60 mV are in the order of 15–40 ms (Grissmer et al. 1994), an order of magnitude greater than that of I_PO. In contrast, the deactivation time constant of Kv3.1 is 1–2 ms at −60 mV (Grissmer et al. 1994; Hernández-Pineda et al. 1999), very similar to that of I_PO. Our estimate of 0.39 mm for the IC₅₀ of TEA on I_PO can be compared favourably with that reported in expressed Kv3.1 channels (∼0.3 mm, Grissmer et al. 1994; 0.38 mm, Hernández-Pineda et al. 1999), as well as their relatively low inactivation. However, it remains possible that I_PO is mediated by other members of the Kv3 family, possibly heteromerically expressed with Kv3.1 (Weiser et al. 1994).

Possible role of I_PO in SThN neurones

Kv3.1 in particular has been associated with ‘fast spiking neurones’ able to sustain high rates of firing activity (Martina et al. 1998; Wang et al. 1998b; Baranauskas et al. 1999; Hernández-Pineda et al. 1999) by virtue of its activation at depolarized levels and fast deactivation. Indeed, SThN neurones are also capable of firing at high rates and we have shown that TEA (1 mm), a concentration expected to block I_PO, and Kv3.1, by about 80 %, has a profound effect on the firing activity, by blocking the fast after-hyperpolarization and permitting accommodation at higher firing rates. These effects of TEA confirm predictions of the effect of block of I_PO based upon its kinetics. Such a role in accelerating action potential repolarization, and reducing its duration, has previously been attributed to Kv3.1 in other cell types (Du et al. 1996; Whim & Kaczmarek, 1998; Wang et al. 1998b; Erisir et al. 1999). In demonstrating sensitivity of both action potential repolarization and spike frequency adaptation to 1 mm TEA, it is tempting to conclude that this reflects the result of a selective block of I_PO by TEA at this particular concentration. However, while targeting of BK channels by TEA may be discounted, the possibility of channels of the Kv1 family (Verma-Kurvari et al. 1997) and/or other as yet unidentified TEA-sensitive channels being expressed in SThN neurones renders it prudent to be circumspect about attributing all the effects of TEA shown here to block of I_PO/Kv3.1. Nonetheless, the finding that the block of conductance(s) sensitive to 1 mm TEA prevents sustained firing at frequencies greater than around 200 Hz is very much in line with observations made in neurones of the medial nucleus of the trapezoid body, which were attributed solely to an effect on Kv3.1 (Wang et al. 1998b), and qualitatively similar to those made in cortical interneurones, also attributed to a block of Kv3.1/Kv3.2 channels by TEA (Erisir et al. 1999).

The SThN and the globus pallidus appear to be functionally linked, either by reciprocal interconnectivity or co-activation by common inputs (Plenz & Kitai, 1999; Magill et al. 2000). SThN and globus pallidus neurones can both sustain high rates of firing, and exhibit high-frequency burstiform activity, both in association with normal information processing in the basal ganglia and particularly during the aberrant activity associated with parkinsonian akinesia (Wichmann & Delong, 1996). Demonstration of similar currents to I_PO, subserved by Kv3 channels, in neurones of the globus pallidus (Baranauskas et al. 1999; Hernández-Pineda et al. 1999) indicates that expression of Kv3 K⁺ channels enables this high-frequency activity in both the SThN and globus pallidus. This property is critical to the function of neurones in both these regions and thence in shaping the output of the basal ganglia as a whole.