Intermediate conductance calcium-activated potassium channels modulate summation of parallel fiber input in cerebellar Purkinje cells

Abstract

Encoding sensory input requires the expression of postsynaptic ion channels to transform key features of afferent input to an appropriate pattern of spike output. Although Ca²⁺-activated K⁺ channels are known to control spike frequency in central neurons, Ca²⁺-activated K⁺ channels of intermediate conductance (KCa3.1) are believed to be restricted to peripheral neurons. We now report that cerebellar Purkinje cells express KCa3.1 channels, as evidenced through single-cell RT-PCR, immunocytochemistry, pharmacology, and single-channel recordings. Furthermore, KCa3.1 channels coimmunoprecipitate and interact with low voltage-activated Cav3.2 Ca²⁺ channels at the nanodomain level to support a previously undescribed transient voltage- and Ca²⁺-dependent current. As a result, subthreshold parallel fiber excitatory postsynaptic potentials (EPSPs) activate Cav3 Ca²⁺ influx to trigger a KCa3.1-mediated regulation of the EPSP and subsequent after-hyperpolarization. The Cav3-KCa3.1 complex provides powerful control over temporal summation of EPSPs, effectively suppressing low frequencies of parallel fiber input. KCa3.1 channels thus contribute to a high-pass filter that allows Purkinje cells to respond preferentially to high-frequency parallel fiber bursts characteristic of sensory input.

Central neurons receive an enormous number of spontaneously active synaptic inputs, but exhibit the capacity to differentiate features of sensory input from background noise. Cerebellar Purkinje cells are contacted by up to ∼150,000 parallel fibers from granule cells, of which only a subset will convey sensory information at any given time. The activation of a peripheral receptive field is transmitted to the cerebellar cortex by mossy fibers in the form of high-frequency spike bursts (1). The resulting temporal summation of excitatory postsynaptic potentials (EPSPs) generates a similar high-frequency burst in granule cells (2). Purkinje cells should then also possess the means to respond effectively to bursts of parallel fiber input that convey sensory information compared with background activity.

Postsynaptic membrane excitability can be controlled by activation of K⁺ channels. There are two established types of Ca²⁺-activated K⁺ (KCa) channels in CNS neurons: small conductance (SK, KCa2.x) and big conductance (BK, KCa1.1) (3, 4). A third class of intermediate conductance (KCa3.1, SK4, IK1) KCa channel is thought to be expressed only in microglia and endothelial cells in the CNS (3, 5, 6). KCa3.1 channels are gated by calmodulin in a similar manner to KCa2.x channels but are insensitive to block by apamin and tetraethylammonium (TEA) (6–8). Instead, the KCa3.1 α-subunit, encoded by the gene KCNN4, has specific residues that bind charybdotoxin and 1-[(2-chlorophenyl)diphenylmethyl]-1H-pyrazole (TRAM-34) (5–7, 9, 10).

In cerebellar Purkinje cells, KCa1.1 and KCa2.2 channels are activated during a spike by high voltage-activated (HVA) P-type Ca²⁺ channels (11). In contrast, low voltage-activated (LVA) Cav3 (T-type) Ca²⁺ channels that are active during the subthreshold interspike interval (12–14) have not been associated with any specific KCa channel. We now report that KCa3.1 channels are expressed in cerebellar Purkinje cells and act to shape the parallel fiber EPSP waveform, providing unique evidence for a functional role for KCa3.1 channels in a CNS neuron. Moreover, KCa3.1 channels colocalize with Cav3.2 Ca²⁺ channels to allow even subthreshold parallel fiber EPSPs to activate KCa3.1 and suppress temporal summation, contributing to a high-pass filter that allows Purkinje cells to respond to parallel fiber input indicative of sensory input.

Results

Parallel Fiber EPSPs Activate a Ca²⁺-Dependent After-Hyperpolarization.

To examine the potential for parallel fiber EPSPs to activate postsynaptic currents, we stimulated parallel fiber inputs in the presence of picrotoxin and CGP55845 to block GABAergic transmission. The subthreshold parallel fiber-evoked EPSP was followed by a graded after-hyperpolarization (AHP) of up to ∼250-ms duration (Fig. S1A). To avoid any effects on presynaptic transmisson during pharmacological tests, we simulated EPSPs by injecting excitatory postsynaptic current (EPSC) waveforms at the soma (simEPSCs) in the presence of inhibitory and excitatory synaptic blockers, which produced very similar voltage responses (Fig. S1A). Some portion of this AHP can be attributed to I_H (15). However, given that blocking I_H substantially changes the membrane time constant, all experiments were conducted in the presence of I_H from a membrane potential of −75 mV.

We first tested the ability for blockers against HVA or LVA Ca²⁺ channels to affect the simEPSP-AHP. We found no significant effect on the simEPSP-AHP by the Cav2.1 (P-type) channel blocker ω-agatoxin IVA (AgTx; n = 4, P = 0.93) (Fig. 1A). Similarly, no effects were obtained upon pressure ejection of a mixture of other HVA Ca²⁺ channel blockers, including ω-conotoxin GVIA, nifedipine, and SNX-482 (n = 6, P = 0.084), or the general HVA blocker Cd²⁺ (n = 5, P = 0.097) (Fig. 1A and Fig. S2). However, both the simEPSP rate of decay and AHP were significantly reduced by low concentrations of the putative T-type channel blockers Ni²⁺ (n = 6, P = 0.00050) and mibefradil (n = 8, P = 0.00018) (Fig. 1A) (16, 17). Indeed, Ni²⁺ could affect several aspects of the simEPSP-AHP (Fig. S1B), but the most reliable effect was a reduction in the simEPSP rate of decay (Fig. 1A).

Fig. 1.

Subthreshold parallel fiber EPSPs generate an AHP consistent with activation of KCa3.1. All records were evoked using simEPSCs to test postsynaptic channel contributions, with drug effects on the rate of EPSP decay normalized to the control 5 mV simEPSP. (A and B) Representative recordings (Left) and bar plots (Right) showing the effects of Ca²⁺ and K⁺ channel blockers. (A) The simEPSP rate of decay is not significantly affected by AgTx (200 nM) or other HVA Ca²⁺ channel blockers (ω-conotoxin GVIA, 1 μM; nifedipine, 1 μM; SNX-482, 200 nM) or Cd²⁺ (30 μM), but is reduced by putative T-type Ca²⁺ channel blockers Ni²⁺ (100 μM) and mibefradil (Mib,1 μM). (B) The simEPSP rate of decay is unaffected by the KCa2.x blocker apamin (100 nM), or KCa1.1 blockers IbTx (200 nM), TEA (5 mM), or paxilline (100 nM), but is significantly reduced by ChTx (100 nM) and TRAM-34 (100 nM). (C) Internal dialysis of Camstatin (5 μM) occludes the effect of TRAM-34 on the simEPSP rate of decay. (D) Pretreatment with Ni²⁺ occludes the action of TRAM-34 on the simEPSP rate of decay. Sample numbers are shown in brackets at the base of bar graphs. Average values are mean ± SEM; **P < 0.01, ***P < 0.001.

The relative lack of specificity of T-type channel blockers leaves open the possibility that these effects were not entirely mediated through T-type channels. First, Ni²⁺ and mibefradil also block R-type Ca²⁺ channels (18, 19). However, the lack of effect by either SNX-482 or Cd²⁺ on the simEPSP rules out R-type channel involvement (Fig. 1A and Fig. S2). The effects of Ni²⁺ on the simEPSP-AHP were also separate from those of blocking I_H (Fig. S1C). The effects of Ni²⁺ were further maintained in the presence of internal heparin (4 mg/mL) and bath-applied cyclopiazonic acid (6 μM) (n = 4, P = 0.0032), indicating no involvement of IP₃-mediated Ca²⁺ release or Ca²⁺ -ATPases (Fig. S1D). Ryanodine also had no significant effect on the simEPSP rate of decay (n = 4, P = 0.12), indicating no role for ryanodine receptor-mediated Ca²⁺-induced Ca²⁺ release (Fig. S1E). Taken together, these data suggest that the simEPSP activates Ca²⁺ influx through Cav3 Ca²⁺ channels to control an outward current with an onset early enough to affect both the EPSP rate of decay and AHP.

EPSP Rate of Decay Is Shaped by a K⁺ Current Consistent with KCa3.1 K⁺ Channels.

We next examined the identity of K⁺ channels activated by the simEPSP. Surprisingly, the simEPSP rate of decay was entirely unaffected by bath application of the KCa2.2 channel blocker apamin (n = 6, P = 0.62) or by KCa1.1 channel blockers iberiotoxin (IbTx; n = 5, P = 0.57), TEA (n = 5, P = 0.29), or paxilline (n = 3, P = 0.97) (Fig. 1B) (7), despite confirmed effects of each drug on spontaneous firing. In contrast, the simEPSP rate of decay was reduced 41.7 ± 6.4% by charybdotoxin (ChTx; n = 6, P = 0.00459) (Fig. 1B). ChTx is known to block Ca²⁺-activated KCa1.1 and KCa3.1 channels and specific Kv1.x channels (7, 20). KCa3.1 channels are apamin-insensitive but have specific binding sites for ChTx and the clotrimazole-related compound TRAM-34 (9, 21). TRAM-34 has been established as a selective KCa3.1 blocker with no effects on KCa1.1, KCa2.2, or a wide array of Kv channels (9, 10) or Cav3 channels (Fig. S3). A role for KCa3.1 channels in Purkinje cells was thus suggested when the simEPSP rate of decay was reduced 24.7 ± 3.9% by TRAM-34 (n = 9, P = 0.000218) (Fig. 1B). The simEPSP also proved to be insensitive to TEA at concentrations < 25 mM (n = 13), a finding consistent with KCa3.1 channels but eliminating several other channels (KCa1.1, Kv1.1, Kv1.6, Kv2.x, Kv3.x channels) (20). Finally, we found that including the calmodulin blocker Camstatin (22) in the electrode occluded the effects of TRAM-34 (Fig. 1C) (n = 7, P = 0.31).

These results strongly implicate KCa3.1 channels given that KCa1.1 channels are highly sensitive to TEA (20) and neither KCa1.1 nor Kv1.x channels are calmodulin-dependent, as shown for KCa3.1 (3). There was also no significant difference between the effects of Ni²⁺, mibefradil, ChTx, or TRAM-34 on the simEPSP rate of decay (F = 2.37, one-way ANOVA), suggesting actions on a common pathway. In fact, the effects of TRAM-34 on the simEPSP-AHP were occluded by prior Ni²⁺ treatment (Fig. 1D) (n = 4, P = 0.67). Taken together, these data reveal a postsynaptic current with a pharmacological profile that is unique to KCa3.1 channels (10, 20) and activated by Cav3-mediated Ca²⁺ influx.

Purkinje Cells Express KCa3.1 Channels.

Identifying an outward current with KCa3.1-like properties was unexpected. Although KCCN4 mRNA has been detected in the cerebellar Purkinje cell layer (23), KCa3.1 expression has not been reported in CNS neurons (3). Of the three Cav3 channel isoforms expressed in Purkinje cells (13, 14, 24), Cav3.2 is preferentially blocked by 100 μM Ni²⁺ (16), as found for the evoked AHP (Fig. 1A). We thus used dual-label immunocytochemistry to detect KCa3.1 in relation to Cav3.2 protein or calbindin, a cytosolic protein distributed throughout the soma and dendrites (Fig. 2 and Fig. S4). KCa3.1 immunolabel was detected at the soma and over restricted segments of primary and secondary dendritic branches of Purkinje cells, regions that were closely matched by that of Cav3.2 immunolabel (13), revealing colocalization (Fig. 2 A–F and Fig. S4). To further test for KCa3.1 expression, we performed RT-PCR using primers directed against a 602-bp region of rat KCCN4 mRNA that includes the pore region of the KCa3.1 channel (25). We also tested for the expression of microglial response factor-1 (MRF-1) as a specific marker for microglia, which also express KCa3.1 channels (26). These tests revealed the presence of KCa3.1 cDNA at the predicted product weight in homogenates of rat cerebellum and KCa1.1, KCa2.2, and KCa3.1 in cytosolic extracts from single Purkinje cells (Fig. 2G). As predicted by a reported decrease in KCa2.2 mRNA through development (27), the KCa2.2 band was detectable but relatively faint compared with KCa1.1 or KCa3.1 (Fig. 2G). The band for KCa3.1 in Purkinje cell samples also matched that of KCa3.1 in lysates of cultured endothelial cells known to express KCa3.1 (Fig. 2G) (28). In contrast, MRF-1 cDNA was present only in cerebellar homogenate (Fig. 2G), indicating that KCa3.1 mRNA in Purkinje cell samples did not reflect microglial contamination. Furthermore, sequence analysis of the purified KCa3.1 band in Purkinje cells confirmed that the product represented the region spanning the KCa3.1 channel pore, and included the known binding sites for TRAM-34 and ChTx (25), but not an apamin binding site (29).

Fig. 2.

Purkinje cells express KCa3.1 channels that colocalize with Cav3.2 protein. (A–C) Dual-label immunocytochemistry for Cav3.2 (A) and KCa3.1 (B) in a coronal section reveals protein colocalized (arrows) at the soma (asterisks) and restricted segments of dendritic branches (C). (D–F) High-power view of Purkinje cell dendrites in a sagittal section dual-labeled for Cav3.2 (D) and KCa3.1 protein (E), with overlay (F) illustrating colocalization over specific segments of dendritic branches. (G) RT-PCR reveals KCa3.1 and MRF-1 mRNA in whole cerebellum (Left), and KCa1.1, KCa2.2, and KCa3.1 but not MRF-1 in single Purkinje cell cytoplasmic extracts (Center). The KCa3.1 product in Purkinje cells matches that found in endothelial cells (Right). (Scale bars, 20 μm.)

Direct tests for the expression of K⁺ channels that were both Ca²⁺-sensitive and of intermediate conductance were carried out using on-cell recordings from Purkinje cell somata. On-cell patch electrodes were filled with Hepes-buffered artificial cerebrospinal fluid (aCSF) to provide a physiological level of Ca²⁺ outside, along with tetrodotoxin (TTX), Cd²⁺, apamin, TEA, 4-AP, and Cs⁺ in both the electrode and external medium to block sodium, HVA Ca²⁺, KCa2.2, KCa1.1, Kv, and HCN channels (Materials and Methods and Fig. S2). T-type channel blockers were specifically excluded and the availability of Cav3 channels promoted by applying a +60 mV potential to the pipette at rest to hyperpolarize the patch transmembrane potential. Channel conductance was determined by applying steady-state commands (5 min) up to +30 mV from the resting condition to depolarize the patch to within the subthreshold voltage range. Spontaneous channel openings were readily detected in all patches, with either single or multiple channel openings per patch that changed linearly in amplitude with applied voltage (Fig. 3 A and B). Mean conductance, as determined from unit amplitude over a range of voltage steps, was 36.3 ± 0.1 pS (n = 5) (Fig. 3B), a value within the range previously reported for KCa3.1 channels (6). Importantly, bath application of the Ca²⁺ ionophore A23187 (n = 6) dramatically increased single channel activity (Fig. 3C), indicating an increase in probability for activation with Ca²⁺ influx. Moreover, channel openings were blocked by perfusion of TRAM-34 (n = 6), a lipophillic drug that acts at the cytoplasmic face (Fig. 3C) (9). These results, combined with the mRNA, immunolabel, and pharmacological profiles established in Figs. 1 and 2 provide strong evidence that KCa3.1 channels are expressed in Purkinje cells.

Fig. 3.

Purkinje cells express Ca²⁺-activated intermediate conductance K⁺ channels. (A) Spontaneous on-cell single-channel recordings from three different cells during a steady-state (5 min) pipette holding potential of 30 mV. Open (o) and closed (c) states are indicated. (B) Plot of mean single-channel amplitudes in on-cell recordings at steady-state potentials up to +30 mV reveals a mean conductance of 36.3 pS (n = 5). (C) (Left) Long-duration on-cell channel recordings (+30 mV pipette potential) before and after perfusing the Ca²⁺ ionophore A23187 (2 μM) and block by TRAM-34 (100 nM) (n = 6). (Right) Bar plots show channel open probability under the conditions shown in the channel recordings (n = 6). Average values are mean ± SEM; *P < 0.05; **P < 0.01.

Cav3 and KCa3.1 Channels Exhibit Nanodomain Coupling to Evoke a Transient Voltage-Dependent K⁺ Current.

The selective reduction of the simEPSP rate of decay by T-type Ca²⁺ channel blockers (Fig. 1) raised the possibility that Cav3 and KCa3.1 channels are part of a physical signaling complex. We tested for an association between Cav3.2 and KCa3.1 channels, finding that Cav3.2 channels coimmunoprecipitated with KCa3.1 channels from homogenates of rat cerebellum (Fig. 4A). In comparison, we found no coimmunoprecipitation between KCa3.1 and Cav2.1 channels that are expressed by Purkinje cells (Fig. S5A).

Fig. 4.

KCa3.1 channel activation is coupled to Cav3-mediated Ca²⁺ influx. (A) Western blot showing coimmunoprecipitation of Cav3.2 and KCa3.1 channels from cerebellar lysate. (B) Outside-out recordings from separate Purkinje cell somata in response to steps from −110 mV to 0 mV. Shown are currents calculated as the difference from those evoked at a −40 mV holding potential, or blocked by TRAM-34 (100 nM) or Ni²⁺ (100 μM). (C) Mean I-V plots for currents isolated as in B indicate a common activation in the low voltage range. (D) Mean I-V plots of TRAM-34-sensitive currents in outside-out recordings with either high EGTA or BAPTA in the electrode. All recordings were obtained from the somata of P 18–25 Purkinje cells in 1.5 mM external Ca²⁺. Average values are mean ± SEM.

KCa3.1 channels are Ca²⁺-activated but show no intrinsic voltage dependence (7). Therefore, we used outside-out recordings from Purkinje cell somata to test the hypothesis that a closely associated Cav3 channel could confer a voltage dependence on KCa3.1 current. Cav3 and KCa3.1 channels were isolated by perfusion of TTX, Cd²⁺, apamin, TEA, 4-AP, Cs⁺, and synaptic blockers (Materials and Methods), and the current-voltage relationship of Cav3 or KCa3.1 currents determined for steps from −110 mV to 0 mV. These recordings revealed a voltage-dependent outward current evoked from a holding potential of −110 mV that was absent at a holding potential of −40 mV (Fig. 4B and Fig. S5B), a characteristic property of channels that undergo voltage-dependent inactivation. Subtraction of records obtained at −40 mV vs. −110 mV isolated a transient outward current of up to 50 pA (24.8 ± 6.2 pA, n = 6) that was fast inactivating (t_1/2 = 22.6 ± 1.4 ms) (Fig. 4B and Table S1). Importantly, this outward current cannot reflect activation of KCa1.1, KCa2.2, Kv1, Kv3, or Kv4 channels, as these are substantially blocked by apamin or 5 mM TEA and 4-AP (20). HVA Ca²⁺ currents (including R-type) are also highly effectively blocked by the 30 μM Cd²⁺ included here (Fig. S2), leaving Cav3 channels as the predicted primary source for Ca²⁺ influx in these recordings. We further isolated the Ni²⁺ and TRAM-sensitive currents evoked by steps from −110 mV using the same conditions of HVA Ca²⁺ and K⁺ channel block. These tests showed that TRAM-34 (n = 4) or Ni²⁺ (n = 4) blocked a fast inactivating current of up to 30 pA, with mean values of 11.4 ± 2.3 pA and t_1/2 inactivation of 23.6 ± 6.0 ms at −30 mV (n = 8) (Fig. 4B and Table S1). TRAM-34 and Ni²⁺-sensitive currents also activated in the low voltage range in a manner similar to the inactivating current isolated by membrane voltage (Fig. 4C). In support of this, current-clamp recordings indicated a clear threshold for activation of a Ni²⁺-sensitive AHP for membrane potentials positive to −80 mV, while voltage-ramp commands revealed Cav3 and KCa3.1 activation from membrane voltages between −80 and −90 mV (Fig. S6). Finally, to test the proximity of Cav3 to KCa3.1 channels, we used outside-out recordings to measure TRAM-34–sensitive current in the presence of either EGTA or BAPTA (10 mM) in the internal electrolyte. TRAM-34–sensitive current was only recorded in the presence of internal EGTA (n = 5) but not BAPTA (n = 6) (Fig. 4D), indicating an interaction at the nanodomain level (30).

Given that KCa3.1 channel open probability is only a function of internal Ca²⁺ concentration (31), the voltage dependence of KCa3.1 activation must indirectly reflect the voltage dependence of Ca²⁺ influx through Cav3 channels. Taken together, these data indicate that an association between Cav3 and KCa3.1 channels at the nanodomain level allows T-type Ca²⁺ influx to activate KCa3.1 as a Ca²⁺-dependent and transient outward current in the low voltage range.

Cav3–KCa3.1 Interaction Controls Temporal Summation of Parallel Fiber EPSP Trains.

To test the hypothesis that the Cav3–KCa3.1 complex could affect summation of EPSPs, we stimulated parallel fiber inputs at 25 Hz for 2 s. Initial EPSP amplitude was 2 mV, with presynaptic facilitation causing the EPSP amplitude to reach ∼5 mV. Under control conditions, an initial summation of the EPSP was reduced within 250 ms to a lower, stable amplitude that remained below spike threshold for the duration of the stimulus train (Fig. 5A). However, we found a rapid temporal summation of EPSPs within the first five stimuli after applying either Ni²⁺ (n = 8, P = 0.00012) or TRAM-34 (n = 4, P = 0.019) to block the Cav3–KCa3.1 interaction (Fig. 5 A and B). Measuring the baseline membrane voltage immediately preceding each stimulus indicated a rapid increase over the first ∼20 stimuli that reached a sustained level 8–10 mV beyond initial baseline (Fig. 5 A and B). As a result, parallel fiber-evoked EPSPs surpassed spike threshold within 5–10 stimuli to reliably generate spike output. We found no role for mGluR1 receptor activation (14) or presynaptic endocannabinoid receptors (32), as neither JNJ16259685 nor the CB1 receptor blocker AM-251 occluded the effect of Ni²⁺ (n = 4) during these stimuli. In addition, the parallel fiber-evoked EPSP paired-pulse ratio was unaffected by either Ni²⁺ (n = 7, P = 0.95) or TRAM-34 (n = 7, P = 0.582) (Fig. S7 A and B), indicating that the increase in temporal summation did not involve presynaptic effects.

Fig. 5.

The Cav3–KCa3.1 complex regulates temporal summation of parallel fiber EPSPs. (A and B) Representative recordings and plots of the baseline membrane voltage during 25-Hz trains of parallel fiber-evoked EPSPs before and after applying Ni²⁺ (A, 100 μM, green) or TRAM-34 (B, 100 nM, red). Stimulus intensity was adjusted to evoke an initial EPSP of 2 mV. (C) Recordings and plots of baseline voltage during 25 Hz parallel fiber stimulus trains in a coronal slice in the absence of picrotoxin to preserve feed-forward inhibition. TRAM-34 (red) substantially increases temporal summation, with an additional increase upon addition of picrotoxin (50 μM, blue) to block GABAergic inhibition. Statistical significance tested for last 10 pulses of stimulus trains in A–C (indicated by bars). Spikes were truncated in A–C and average values are mean ± SEM; *P < 0.05; **P < 0.01, ***P < 0.001.

Temporal summation of parallel fiber EPSPs has also been shown to be controlled through feed-forward inhibition (33). However, when we repeated these tests in coronal slices in the absence of picrotoxin to preserve feed-forward inhibition, TRAM-34 rapidly increased temporal summation of parallel fiber-evoked EPSPs (Fig. 5C) (n = 5, P = 0.04, one-way ANOVA with post hoc Tukey test). Subsequent addition of picrotoxin to block GABAergic inputs led to a further increase in temporal summation during the stimulus train (Fig. 5C) (n = 5, P = 0.08, one-way ANOVA with post hoc Tukey test). These results indicate that postsynaptic control of temporal summation by the Cav3–KCa3.1 complex is effective in the presence of feed-forward inhibition.

Cav3–KCa3.1 Complex Is Active During Membrane Depolarizations and Tonic Firing.

Given that Cav3 channels inactivate upon depolarization, it is important to determine the availability of Cav3 current at physiological potentials. We thus tested if the Cav3–KCa3.1 interaction is active at more depolarized levels of membrane potential that promote spontaneous spike firing. To measure Cav3 activation and inactivation over a full voltage range, we used whole-cell recordings from P10–P12 Purkinje cells when the dendritic tree has not fully developed to avoid space clamp errors (n = 7) (Fig. 6 A and B). The Boltzmann fits for activation and inactivation data (Fig. 6B) indicated that a Cav3 window current exists for membrane voltages up to at least ∼−20 mV, and thus well into the suprathreshold range (12, 34) (Fig. 6 A and B).

Fig. 6.

Cav3 window current straddles spike threshold in Purkinje cells and the Cav3–KCa3.1 complex affects EPSP summation under physiological conditions. (A) Mean conductance and inactivation plots calculated for whole-cell Cav3 current recorded from P10–12 Purkinje cells (Inset). (B) Expanded view of the fits for activation and inactivation curves shown in A reveal that Cav3 window current (gray) in relation to spike threshold (dashed line). (C–E) The effects of blocking the Cav3–KCa3.1 complex during a five-pulse train of parallel fiber stimulation (100 Hz, arrows). Resting potential was depolarized to a level sufficient to evoke ∼50-Hz tonic firing, with spike threshold indicated at Left. Blocking the Cav3–KCa3.1 complex with either TRAM-34 (C, 100 nM, red) or Ni²⁺ (D, 100 μM, green) reveals substantial control over EPSP summation and spike frequency (E). Spikes were truncated for display in C and D, and picrotoxin (50 μM) was present for all tests in C–E. All average values are mean ± SEM; *P < 0.05.

We further tested the degree to which parallel fiber EPSP summation could be modified by the Cav3–KCa3.1 interaction during tonic firing. Bias current was applied to sustain a tonic firing frequency of ∼50 Hz (49.7 ± 1.8 Hz, n = 9) before and after drug application. Parallel fiber inputs were stimulated at 100 Hz (five pulses) to mimic frequencies of input relevant to sensory stimulation (1, 35). After application of TRAM-34, there was an increase in both peak baseline voltage (59 ± 25.6%, n = 5, P = 0.025) and peak frequency during stimulation (53 ± 16.2%, n = 5, P = 0.021) (Fig. 6 C and E). Similarly, Ni²⁺ increased peak baseline voltage (96 ± 36%, n = 4, P = 0.024) and spike frequency (114 ± 16.5%, n = 4, P = 0.012) during the stimulation (Fig. 6 D and E). Neither result could be attributed to a change in the amplitude of the evoked EPSP (Fig. S7 C and D). These results demonstrate that both Cav3 and KCa3.1 channels are available at depolarized potentials supporting tonic spike firing, and modulate temporal summation of parallel fiber inputs at physiologically relevant input frequencies.

Discussion

The present study provides unique evidence for the activation of KCa3.1 K⁺ channels in a CNS neuron and its role in creating a postsynaptic frequency filter for synaptic input. Although KCa3.1 channel activation has been documented in cells of the enteric and myenteric nervous systems (36, 37), several lines of evidence now support the expression of KCa3.1 channels in Purkinje cells, including single cell RT-PCR, KCa3.1 immunolabel, single channels with intermediate conductance, and macropatch recordings of KCa currents with a pharmacological profile that is unique to KCa3.1 channels (3, 7). Moreover, Cav3.2 Ca²⁺ channels colocalize and associate with KCa3.1 channels to provide Ca²⁺-dependent regulation at the nanodomain level to control temporal summation of parallel fiber EPSPs. This unique Cav3–KCa3.1 complex differs significantly from other ion channel complexes involving T-type channels reported in previous studies. The ability to evoke the Cav3–KCa3.1 interaction using postsynaptic simEPSCs indicates that the complex does not require Ca²⁺ influx through ligand-gated channels, a known interaction for KCa2.x channels (38). A functional coupling between T-type Ca²⁺ and KCa2.x channels in select neuronal subtypes (39–41) is also reported to operate at the microdomain level compared with the nanodomain demonstrated here. A Cav3–Kv4 K⁺ channel complex employs K⁺ channel interacting protein 3 to mediate Ca²⁺ sensing for voltage-gated Kv4 channels (42, 43). The Cav3-KCa3.1 complex instead depends on calmodulin that has been shown to directly gate the KCa3.1 channel (3).

All members of the KCa2.x and KCa3 families are known to be Ca²⁺- but not voltage-dependent (3). However, the close association between Cav3 and KCa3.1 channels allows KCa3.1 to acquire the properties of a low voltage-activated current, as well as the fast inactivating kinetics of Cav3 current. KCa3.1 channels also exhibit a two-to-three times greater sensitivity to internal Ca²⁺ than KCa2.x channels (36). These properties are consistent with the ability for a transient T-type Ca²⁺ influx to activate KCa3.1 with little delay at the onset of an EPSP and yet generate an AHP of up to 250-ms duration. KCa3.1 channels can thus be activated by single, low-amplitude EPSPs but also exhibit a cumulative activation during repetitive activity, an advantage over other K⁺ channels for modulating temporal summation of synaptic depolarizations.

Control over temporal summation of parallel fiber EPSPs in Purkinje cells has also been reported through other mechanisms. One is a role for I_H to reduce EPSP width and temporal summation (15). Because all tests conducted here were performed with I_H intact, KCa3.1 clearly has a role distinct from I_H in producing the AHP and modifying temporal summation. Molecular layer interneurons provide a feed-forward inhibitory influence that reduces parallel fiber EPSP summation (33). This function is again different, in that the Cav3–KCa3.1 complex exerts a significant effect on temporal summation even in the presence of feed-forward inhibition. The influence of the Cav3–KCa3.1 interaction is also fully functional during tonic firing, a result attributable to the wide extent of Cav3 window current in the suprathreshold voltage range. The present results then suggest that the Cav3–KCa3.1 complex functions synergistically with I_H and feed-forward inhibition to reduce Purkinje cell responsiveness to background granule cell activity yet permit activation by high-frequency trains of parallel fiber input. Taken together, these factors underlie a high-pass filter function that allows spike bursts produced by granule cells to preferentially evoke Purkinje cell spike output (2).

Materials and Methods

Molecular Biology and Cytochemical Methods.

Materials and methods related to coimmunoprecipitation, RT-PCR, immunocytochemistry, and use of tSA-201 cells are found in SI Materials and Methods.

Animals.

Timed-pregnant Sprague-Dawley rats (Charles River) were maintained according to the Canadian Council on Animal Care; male pups ≥ postnatal day (P) 18 were used, unless otherwise indicated. Rats were anesthetized by inhalation of isoflurane until unresponsive to tail pinch. Tissue dissection and preparation of parasagittal (300 μm) slices from the cerebellar vermis were previously described (24).

Solutions.

Chemicals were obtained from Sigma unless otherwise indicated. aCSF was composed of 125 mM NaCl, 3.25 mM KCl, 1.5 mM CaCl₂, 1.5 mM MgCl₂, 25 mM NaHCO₃, and 25 mM d-glucose. Picrotoxin (50 μM), 10 μM 6,7-dinitroquinoxolinedione (DNQX; Tocris), 25 μM DL-AP5 (Ascent Scientific), and 1 μM CGP55485 (Tocris) were added to the aCSF for all recordings that did not require stimulation. DNQX, as well as nifedipine, were dissolved in dimethyl sulfoxide before use (final DMSO concentration less than 0.1%). Apamin was first dissolved in 50 mM acetic acid before preparing stock solution.

Bath perfused channel blockers and ionophores were applied at concentrations of 100 nM TTX, 100 μM NiCl₂, 1 μM mibefradil, 30 μM CdCl₂, 2 μM A23187, 100 nM apamin, 5 mM TEA, 5 mM 4-AP, 2 mM CsCl₂, 100 nM TRAM-34 (Tocris) (10), 100 nM paxilline, 6 μM cyclopiazonic acid (CPA; Ascent Scientific), 10 μM ryanodine, and 1.5 μM JNJ16259685. Heparin (4 mg/mL) or camstatin (5 μM) was included in the pipette electrolyte. Toxins applied locally were ejected from a pressure electrode containing Hepes (10 mM)-buffered aCSF with 0.1% BSA as carrier medium at concentrations of 100 nM ChTx, 200 nM IbTx, 200 nM AgTx, 1 μM ω-conotoxin GVIA, 200 nM SNX-482, 1 μM nifedipine, with appropriate synaptic blockers. When applying HVA Ca²⁺ channel blockers, TTX was included in the bath and pressure electrode.

TTX, NiCl₂, CdCl₂, CsCl, TEA, 4-AP, and picrotoxin were prepared daily from stock solutions and all other drugs daily from frozen aliquots.

For current clamp recordings the pipette electrolyte consisted of 130 mM K-gluconate, 0.1 mM EGTA, 10 mM Hepes, 7 mM NaCl, 0.3 mM MgCl₂, pH 7.3 with KOH, providing an E_Cl of −76 mV and E_K of −97 mV. Di-Tris-creatine phosphate (5 mM), 2 mM Tris-ATP and 0.5 mM Na-GTP were added daily from frozen stock solutions. A junction potential of −10.7 mV was subtracted from current clamp recordings. Outside-out voltage-clamp recordings of KCa3.1 were obtained using an electrolyte of 140 mM KCl, 1 mM MgCl₂, 5 mM EGTA, and 10 mM Hepes, pH 7.3 with KOH, with bath perfusion of TTX, apamin, TEA, 4-AP, CdCl₂, and CsCl. Whole-cell voltage-clamp recordings of Purkinje cell Cav3 current from P10–P12 animals used an electrolyte consisting of 140 mM CsCl, 1 mM MgCl₂, 5 mM EGTA, and 10 mM Hepes, pH 7.3 with CsOH, and bath perfusion of TTX, apamin, CdCl₂, TEA, 4-AP, and CsCl. On-cell recordings of KCa3.1 single channels used an electrolyte of Hepes-buffered aCSF containing 150 mM NaCl, 3.25 mM KCl, 1.5 mM CaCl₂, 1.5 mM MgCl₂, 10 mM Hepes, and 20 mM d-glucose, pH 7.3 with NaOH, and TTX, apamin, TEA, 4-AP, CdCl₂, CsCl and synaptic blockers in both the electrode and external medium.

Electrophysiology.

Whole-cell current-clamp somatic recordings were made using Axoclamp amplifiers and Digidata 1322 with a DC-10 kHz band-pass filter and pClamp software. Negative bias current of less than 1.0 nA was applied during current-clamp recordings to maintain Purkinje cell resting potential at ∼−75 mV and below the level of tonic firing. See SI Materials and Methods for further details.

Data Analysis and Statistics.

Statistical analysis was carried out in OriginPro 8. Unless otherwise indicated, paired-sample Student t tests were used to determine significance. The Tukey HSD post hoc comparison was used to test significance between means following one-way or repeated-measures ANOVA. Average values are expressed as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001.