Animal care and tissue dissection.

Sprague-Dawley rat pups were obtained from Charles River and maintained according to the guidelines set by the Canadian Council for Animal Care. Cerebellar slices were prepared from rats at postnatal day 15–24 (P15–24) as described previously9. All chemicals were obtained from Sigma unless otherwise noted. Briefly, rats were killed by intraperitoneal injection of sodium pentobarbital (0.05 mg per kg of body weight, MTC Pharmaceuticals) and the cerebellum was removed in aCSF composed of 125 mM NaCl, 3.25 mM KCl, 1.5 mM CaCl 2 , 1.5 mM MgCl 2 , 25 mM NaHCO 3 and 25 mM d-glucose, preoxygenated by carbogen (95% O 2 , 5% CO 2 ) gas. Parasagittal tissue slices of cerebellum (300 μm) were cut by Vibratome and transiently elevated to 34 °C (60 min) before being stored in carbogen-gassed aCSF at 33–35 °C. Slices were subsequently transferred to a recording chamber on a Zeiss Axioskop FS-2 microscope and maintained at 33–35 °C as a submerged preparation. Stellate cells were directly visualized using differential interference contrast optics according to position in the upper one-third of the molecular layer.

Stellate cell electrophysiology.

Whole-cell recordings were performed at 33–35 °C using a Multiclamp 700B amplifier and pClamp 9.2 software (Molecular Devices). Borosilicate pipettes (1.5 mm outside diameter) had a resistance of 6–8 MΩ with access resistance of 8–15 MΩ (80–90% compensation in voltage clamp). Cells were rejected if access resistance drifted by more than 20%. The internal solution for current-clamp recordings consisted of 130 mM potassium gluconate, 0.1 mM EGTA, 10 mM HEPES, 7 mM NaCl and 0.3 mM MgCl 2 (adjusted to pH 7.3 with KOH), and 5 mM di-Tris-creatine phosphate, 2 mM Tris-ATP and 0.5 mM Na-GTP were added from fresh frozen stock each day. For neuronal voltage-clamp recordings, the internal solution consisted of 140 mM KCl, 0.1 mM EGTA, 10 mM HEPES and 2.5 mM MgCl 2 (pH adjusted to 7.3 with KOH). The bath medium always contained 0.0001 mM tetrodotoxin to isolate I A , 5 mM TEA to block a small amplitude delayed rectifier also expressed by these cells9, and 2 mM CsCl to bock I h . Perforated patch recordings followed previous protocols39 using the KCl-based electrolyte described above and 100 μg ml−1 gramicidin. All recordings were performed in synaptic blockers that were bath applied after obtaining the initial seal: picrotoxin (50 μM), d,l-2-amino-5-phosphonopentanoic acid (AP5, 25 μM), 6,7-dinitroquinoxolinedione (DNQX, 10 μM, Tocris Cookson) and (2S)-3-([(15)-1-(3,4-dichlorophenyl)ethyl]amino-2-hydroxypropyl)(phenylmethyl)phosphinic acid (CGP55845, 1 μM, Tocris). When 0.1 mM [Ca2+] o was bath applied, MgCl 2 was elevated to 2.9 mM. Low-calcium aCSF for local pressure ejections consisted of 150 mM NaCl, 3.25 mM KCl, 0.1 mM CaCl 2 , 2.9 mM MgCl 2 , 20 mM d-glucose, 10 mM HEPES, 0.05 mM picrotoxin, 0.025 mM AP5, 0.01 mM DNQX, 0.001 mM CGP55845, 0.0001 mM TTX, 5 mM TEA and 2 mM CsCl (pH adjusted to 7.3 with NaOH).

The above conditions allowed recordings of 0.5–5.0 nA I A in stellate cells9. T-type current (I T ) was also detectable, but only at the comparatively low level of 20–80 pA, generating a typical ratio of ∼60:1 I A :I T that allowed us to measure I A in the presence or absence of I T . Mibefradil has been shown to be relatively selective for T-type calcium channels at low concentrations40 and was bath applied at 0.5 μM. Direct tests on Kv4 current expressed in isolation in tsA-201 cells confirmed that mibefradil had no effect on Kv4 complex properties (Fig. 3b and Supplementary Fig. 5). The observation that BAPTA was able to occlude the effect of mibefradil also rules out the possibility that mibefradil directly affects Kv4 channel inactivation.

Interrupting KChIP function.

Monoclonal KChIP antibodies (from NeuroMab and Millipore) directed at all isoforms (pan-KChIP; Neuromab) or specific isoforms (KChIP1, 2 or 3, NeuroMab; KChIP3, Millipore) were included in the internal electrolyte at a dilution of 1:100. I A was recorded at 10, 20 and 30 min after entering into the whole-cell configuration. To denature the pan-KChIP antibody, we placed pan-KChIP antibody–containing internal electrolyte in a water bath and maintained it at 95 °C for a minimum of 1 h.

Heterologous expression in tsA-201 cells.

The human Kv4.2 cDNA in pCMV6-XL5 expression vector was obtained from Origene. Human KChIP 1, 2 and 4 cDNA in pCMV-Sport6 expression vector was obtained from ATCC. Human KChIP3 cDNA in pBluescriptR was obtained from ATCC. The coding region of human KChIP3 was excised using KpnI/NotI and the fragment subcloned into the expression vector pCDNA3.1− (Invitrogen). Human DPP10c cDNA was kindly provided by P. Pfaffinger (Baylor College of Medicine). DPP10c cDNA was further subcloned into pCDNA3.1− (Invitrogen) using XhoI/AflIII. The cloning of Cav3.3 has been described previously41. Cav2.2 and Cav3.2 cDNA were donated by T. Snutch (University of British Columbia). To generate the E to K mutation in the P-loop of Domain I, we subcloned a 1.3-kb fragment of the full-length human Cav3.3 calcium channel α 1 subunit into pGEM-T Easy (Promega). Site-directed mutagenesis of this construct was carried out using the QuickChangeII Site-Directed Mutagenesis kit (Stratagene) according to manufacturer's instructions. After confirmation of the appropriate mutations by direct DNA sequencing, the fragment was transferred back into the full-length Cav3.3 calcium channel α 1 subunit in pMT2 using Not I and Mlu I, and the sequence was re-confirmed. PCR was used to amplify the N terminus, C terminus and linker regions of human Cav3.2 and Cav3.3 from their appropriate cDNA templates. The fragments were ligated into pGEM-T Easy vector system (Promega). Following successful verification of PCR product sequence, the fragments were excised using appropriate restriction enzymes and subcloned into pGEX-5X-1 GST Fusion System (Pharmacia). Culturing and transfection of tsA-201 cells has been previously described41.

Electrophysiology in tsA-201 cells.

For consistency, we expressed cDNAs from a representative set of Kv4 complex and Cav3 channels found in stellate cells. Most experiments were conducted using Cav3.3:Kv4.2:KChIP3:DPP10c in a 0.5:1:1:1 ratio, which produced a stable set of kinetic and voltage-dependent properties in tsA-201 cells. Kv4.2 was chosen as the Kv4 complex subunit, as we were able to use available antibodies effectively for both immunolocalization and protein biochemistry. Cells were transfected with combinations of cDNAs: Kv4.2 (5 μg μl−1), Cav3.2, Cav3.3 or mutant Cav3.3 (2.5 μg μl−1), Cav2.2 with β1-b and α2-δ subunits (2.5 μg μl−1), KChIP1, 2, 3 or 4 (5 μg μl−1), DPP10c (5 μg μl−1), and GFP (2 μg μl−1). Cells were washed with fresh media 14 h later and moved to 29 °C for 2–3 d. tsA-201 cells were placed in a recording chamber (21 °C) and perfused with 120 mM NaCl, 3 mM NaHCO 3 , 4.2 mM KCl, 1.2 mM KH 2 PO 4 , 1.5 mM MgCl 2 , 10 mM d-Glucose, 10 mM HEPES and 1.5 mM CaCl 2 (pH adjusted to 7.3 with NaOH). Electrodes were filled with 110 mM potassium gluconate, 30 mM KCl, 1 mM EGTA, 5 mM HEPES and 0.5 mM MgCl 2 (pH adjusted to 7.3 with KOH). As found in previous studies, initial coexpression of KChIP3 with Kv4.2 increased the rate of recovery from inactivation from 58.9 ± 5.2 ms (n = 10) to 28.5 ± 4.6 ms (n = 8), verifying that the basic interaction between these complex members was present.

Immunocytochemistry.

Rats were deeply anesthetized with sodium pentobarbital and killed by intracardial perfusion as described previously10,15. Free-floating 30–40-μm sections were cut on a vibratome in cold phosphate buffer and reactions were carried out in working solution containing 3% normal donkey serum (Jackson), 0.1% TWEEN (vol/vol) and 1% DMSO (vol/vol) in phosphate buffer, with gentle agitation throughout all reactions. Mouse monoclonal antibodies to KChIP1, 2 or 3 (1:250) and DPP6 (1: 250) (R&D Systems) were used as primary antibodies. We used polyclonal rabbit Cav2.2 (1:2,000) and DPP10 (1:250, Santa Cruz Biotechnology) antibodies. We used both monoclonal MAP2 (1:1,000) and polyclonal MAP2 (1:250, AbCam) antibodies. Secondary antibodies were applied (1:1,000) as Alexa Fluor 488– or 594–conjugated donkey or goat antibodies to rabbit IgGs or donkey or goat antibodies to mouse IgGs (Molecular Probes). Controls consisted of omitting the primary antibodies. Immunoreactivity was assessed using a Zeiss AxioImager microscope or an Olympus FV300 BX50 confocal microscope. Images were transferred to Adobe Photoshop and Illustrator and image adjustments were confined to intensity levels.

Co-immunoprecipitation assay.

Adult male rats (200–300 g) were anesthetized using halothane and the brain dissected out and homogenized in ten volumes of ice-cold lysis buffer (150 mM NaCl, 25 mM Tris-HCl (pH 7.4) and protease inhibitors). The homogenate was centrifuged at 700g for 5 min and crude homogenate was again centrifuged at ∼16,000g for 10 min. The supernatants (200–300-μl aliquots) were incubated overnight (4 °C) with antibody to Kv4.2 (C20, Santa Cruz Biotechnology) or antibody to Cav3 (Alomone Labs) at a final concentration of 20 μg ml−1. The mixtures were then incubated with protein A/G beads (GE Healthcare) for 2 h to bring down the antibody-antigen complexes. The beads were washed three times with phosphate-buffered solution by centrifugation (700g) and resuspension. The immunocomplexes were eluted using 40 μl of 8% SDS with 100 mM DTT, mixed with 20 ml of sample buffer (50 mM Tris (pH 6.8), 2.5% SDS, 15% glycerol and 100 mM DTT), and separated by SDS-PAGE. For western blotting, we used antibodies to Kv4.2 (0.3 μg ml−1, Chemicon) and Cav3 (1 μg ml−1, Alomone) as primary antibodies. The bands were detected by horseradish peroxidase–linked secondary antibodies and ECL+ (GE Healthcare).

tsA-201 cells were grown and transfected using the calcium phosphate method41. Cells were washed twice with ice-cold phosphate buffered saline 48 h after transfection and lysed with 0.4 ml of immunoprecipitation buffer (150 mM NaCl, 25 mM Tris-HCl (pH 7.4) and protease inhibitors) supplemented with 1% NP-40 (vol/vol) for 1 h at 4 °C. The cell lysates were then centrifuged to remove cell debris and co-immunoprecipitations performed as stated above. Cav3 or Kv4 antibodies were added to the lysate and incubated at 4 °C overnight with gentle shaking. We then added 50 μl of protein A/G (resuspended in 50% in lysis buffer) to collect immunoprecipitate. The tubes were incubated at 4 °C for 2–3 h and centrifuged at 700g for 1 min and the supernatant decanted. Precipitate was washed three times with 1 ml of phosphate-buffered solution. The immunocomplexes were eluted using 40 ml of 10% SDS with 100 mM DTT. The eluted material was mixed with 20 μl of sample buffer (50 mM Tris (pH 6.8), 2.5% SDS, 15% glycerol and 100 mM DTT) and analyzed by western blot.

GST pulldown assay.

To prepare GST fusion proteins, we transformed Cav3 linkers and C terminus and N-terminus GST fusion protein plasmids into BL21 and induced the expression of recombinant proteins by 0.5 mM isopropyl-d-thiogalactopyranoside. GST fusion proteins were purified from bacteria using glutathione-Sepharose 4B beads (GE Healthcare) according to the protocol recommended by the manufacturer. Fusion proteins were incubated with rat brain homogenate overnight. The beads were washed and bound proteins were eluted in sample buffer, fractionated by SDS-PAGE and analyzed by western blot. The blots were probed with rabbit antibody to Kv4 and antibody-stained bands were visualized by horseradish peroxidase–conjugated antibody to rabbit IgG and ECL+.

Data and statistical analyses.

Voltage-clamp analysis using Origin 8.0 (OriginLab) consisted of fitting the time course of current inactivation with a single exponential decay function, the time course of recovery with a single exponential growth function, and steady-state activation and inactivation curves with Boltzmann functions of the form

Current-clamp analysis was performed using custom software written in MatLab R2007b (MathWorks). A calculated junction potential of −10.7 mV was subtracted from all current-clamp recordings. For neuronal voltage-clamp recordings, the negligible calculated junction potential of −2 mV was not subtracted.

Modeling.

The model was modified from a previous study9 using MatLab R2007b with a fourth-order Runge-Kutta algorithm using a time step (dt) of 0.005 ms with the following equations

for h A∞ : V h = −68 or −78.

The time constants for inactivation of sodium current in the model were voltage dependent and described by a Lorentzian function.

for h: τ(V): V c = −74, w = 28, A = 232 and y 0 = −0.15.

Voltage was integrated according to

The constants that we used in the model were C = 1.5 μF cm−2, E Na + =55 mV, E K + =−80 mV, E leak =−60 mV, E Ca 2 + =22 mV, g Na =35 μS cm−2, g K Na=7 μS cm−2, g leak =0.1 μS cm−2, g A =16 μS cm−2, g T =0.45 μS cm−2.

Average values are indicated as mean ± s.e.m. Unless otherwise noted, all statistical tests were paired t tests or one-way ANOVA with post hoc Tukey's HSD.

Note: Supplementary information is available on the Nature Neuroscience website.