|Home | About | Journals | Submit | Contact Us | Français|
Episodic ataxia type-2 (EA2) is a hereditary cerebellar ataxia associated with mutations in the P/Q-type voltage-gated calcium channels. Therapeutic approaches for treatment of EA2 are very limited. Presently, the potassium channel blocker 4-aminopyridine (4-AP) constitutes the most promising treatment although its mechanism of action is not understood. Here we show that, in contrast to that commonly believed, therapeutic concentrations of 4-AP do not increase the inhibitory drive of cerebellar Purkinje cells. Instead, 4-AP restores the severely diminished precision of pacemaking in Purkinje cells of EA2 mutant mice by prolonging the action potential, and increasing the action potential afterhyperpolarization. Consistent with this mode of action, the therapeutic efficacy of 4-AP was comparable, and not additive, to chlorzoxazone, an activator of calcium-dependent potassium channels that also restores the precision of Purkinje cell pacemaking. The likely target of 4-AP at the concentrations used are the Kv1 family of potassium channels, possibly the Kv1.5 subtype. Because at higher concentrations 4-AP blocks a large array of potassium channels and is a pro-convulsant, use of selective Kv1 channel blockers is likely to be a safer substitute for treatment of cerebellar ataxia.
Episodic ataxia type-2 (EA2) is a form of hereditary neurological disorder caused by cerebellar malfunction and is characterized by interictal ataxia and frequent attacks of dyskinesia, vertigo and imbalance that last for hours to days (Jen et al., 2007). EA2 is associated with mutations in the Cav2.1α1 pore-forming subunit of the P/Q-type voltage gated Ca2+ channel which reduce its current density (Dove et al., 1998; Fletcher et al., 1996; Ophoff et al., 1996). These channels are widely expressed in the CNS and are particularly enriched at synaptic nerve terminals (Stea et al., 1994) and in cerebellar Purkinje cells (Mori et al., 1991; Usowicz et al., 1992)
Current treatments for EA2 are limited (Jen et al., 2007; Strupp et al., 2007). Recently, the K+ channel blocker 4-aminopyridine (4-AP) has been used for treatment of this disorder (Strupp et al., 2008; Strupp and Brandt, 2006) in humans, and also in animal models of EA2 such as the tottering mouse (Weisz et al., 2005). The mode of action of 4-AP is not understood. 4-AP has been prescribed based on the assumption that because of the reduced P/Q-type calcium current density the Purkinje cell inhibitory drive to neurons of deep cerebellar nuclei (DCN) is reduced (Glasauer et al., 2005; Kalla et al., 2007; Strupp et al., 2008). Because in Purkinje cells 4-AP decreases the latency of Ca2+ spikes evoked with large depolarizing pulses (Etzion and Grossman, 2001), it was thought that 4-AP will restore their inhibitory drive onto DCN neurons (Jen et al., 2007; Strupp et al., 2007; Strupp et al., 2008; Strupp and Brandt, 2006). However several recent findings are inconsistent with these assumptions. Unlike that postulated, the firing rates of mutant Purkinje cells in animal models of EA2 are not lower (Hoebeek et al., 2005; Hoebeek et al., 2008; Ovsepian and Friel, 2008). Moreover, despite the presence of morphological synaptic abnormalities, stimulation of Purkinje cells inhibits DCN neurons of tottering mice with the same efficacy (mean latency and pause duration) as it does in the wild type (Hoebeek et al., 2008).
Because as a potassium channel blocker 4-AP is a relatively potent pro-convulsant (Bever, Jr. et al., 1994; Judge et al., 2006), understanding its mode of action is necessary if we are to supersede it by offering safer alternatives. We thus sought to delineate the therapeutic mode of action of 4-AP. We found that at therapeutic concentrations 4-AP neither increased the firing rate nor excitability of Purkinje cells or synaptic drive to DCN neurons. Instead, 4-AP effectively restored the precision of pacemaking in the mutant tottering Purkinje cells by increasing the duration of action potentials and the amplitude of AHPs. 4-AP’s efficacy in restoring the precision of Purkinje cell pacemaking was mimicked by selective blockade of Kv1.5 channels. Consistent with a mode of action on Purkinje cell pacemaking, the therapeutic benefits of 4-AP in the tottering mice were not additive to those of chlorzoxazone (CHZ) which also restores the precision of Purkinje cell pacemaking.
All procedures employed were in accordance with the policies established by the Animal Institute Committee of the Albert Einstein College of Medicine.
P12–19 Wistar rats, 2–3 months old tottering (tg/tg) or age-matched C56Bl/6 mice were anesthetized with halothane and decapitated. The brain was quickly removed and placed in cold extracellular solution containing: (in mM) NaCl 125, KCl 2.5, NaHCO3 26, NaH2PO4 1.25, MgCl2 1, CaCl2 2, glucose 10, pH 7.4 when gassed with 5% CO2:95% O2. The cerebellum was dissected and mounted on a modified Oxford vibratome and 300 µm thick sagittal slices were made. The slices were kept in oxygenated extracellular solution at 34 °C for one hour, and then at room temperature until use (1–5 hours).
Slices were placed in a recording chamber on the stage of a Zeiss Axioskop microscope. Purkinje cells were visually identified using a 40x water-immersion objective with infrared optics. The slices were superfused with the extracellular recording solution at a rate of 1.5–2 ml/min and the temperature adjusted to 35±1°C. Extracellular recordings were obtained from single neurons using a home-made differential amplifier and glass pipette electrodes filled with extracellular solution (tip size 0.3–1 µm). The pipette tip was positioned just above, or lightly touching, the cell body near the axon hillock where the largest potential changes were usually recorded. To isolate the Purkinje cell intrinsic activity, synaptic transmission was blocked using 5 mM kynurenic acid (Spectrum Chemical MFG Corp., Gardena, CA), a broad-spectrum ionotropic glutamate receptor antagonist (Stone, 1993), 100 µM picrotoxin (Sigma, St. Louis, MO), a GABAA receptor antagonist (Yoon et al., 1993), and 1 µM CGP55845 [(2S)-3-[[(1S)-1-(3,4-dichlorophenyl)ethyl]amino-2-hydroxypropyl] (phenylmethyl) phosphinic acid] (Tocris, Ellisville, MO) a GABAB receptors antagonist (Davies et al., 1993). These antagonists were present in all experiments except in those which examined synaptic transmission. 4-aminopyridine (4-AP) was obtained from RBI-Sigma.
Data was sampled at 10 kHz using an analog-to-digital converter (PCI-MIO-16XE-10; National Instruments, Austin, TX), and acquired and analyzed using custom software written in LabView (National Instruments).
In each experiment interspike interval histograms were constructed using a long stretch of spontaneous activity (>5 minutes). These histograms were used to calculate the predominant (instantaneous) firing rate, defined as the reciprocal of the interspike interval most frequently observed (the peak of the histogram, i.e. its mode), and the maximum (instantaneous) firing rate, defined as the reciprocal of the shortest interspike interval which accounted for at least 5% of interspike intervals (Womack and Khodakhah, 2003).
Whole-cell voltage-clamp recordings were performed using an Optopatch amplifier (Cairn Research, Kent, UK) with electrodes pulled from borosilicate glass (1–3 MΩ resistance when filled with intracellular solution). The intracellular solution used was composed of (in mM): 70 Cs-gluconate, 10 CsF, 20 CsCl, 10 EGTA, 10 HEPES, and 3 Na2ATP, pH=7.4 (CsOH). Recordings obtained from deep cerebellar neurons were performed using a high chloride intracellular solution containing (in mM): 138 CsCl, 10 CsF, 10 HEPES, 5 NaCl, 3 MgATP, pH=7.2 (CsOH). For whole-cell current clamp recordings the intracellular solution contained (in mM): 140 K-methyl sulfate, 10 KCl, 5 NaCl, 4 MgATP, 0.01 EGTA, 10 HEPES, pH=7.33 (KOH).
Data was sampled at 20 kHz using an analog-to-digital converter (PCI–MIO–16XE–10; National Instruments, Austin, TX), and acquired and analyzed using a custom software written in LabView (National Instruments).
Parallel fibers were electrically stimulated by a glass pipette (filled with extracellular solution) placed in the molecular layer, above the Purkinje cell under study. In every trial, a single 100 µs current pulse (2–20 µA intensity) was delivered using a constant current isolated stimulator (DS3, Digitimer LTD). To measure the paired pulse ratio, two identical electric pulses were delivered separated by a 50 ms long interval. To stimulate mossy fibers, the stimulation pipette was placed in the white matter tract underneath the granule cell layer that targeted the Purkinje cell under study.
To stimulate Purkinje cells inputs to DCN neurons the stimulation pipette was positioned between the cerebellar cortex and the nuclei in order to prevent direct effects on nuclear cells.
We used the accelerating rotarod test as a paradigm to examine motor performance (Crawley, 2008; Walter et al., 2006). The apparatus consisted of a 3 cm diameter rotating rod (Rotamex-5, Columbus Instruments) elevated 55 cm above a covered platform. Each trial started from stationary position accelerating at a rate of 0.1 cm/s every second. Speed and latency to fall of the animals were automatically recorded by an interfaced computer. Every day, tg/tg mice and +/+ littermates (4–5 months old) were tested in 10 consecutive trials.
To quantify the frequency, severity and duration of episodes of dyskinesia observed in tottering mice, the overall motor behavior was scored every 10 min before and after each rotarod session for up to 2 hours. The scoring followed a previously published scale (Weisz et al., 2005) as follows: 0=normal motor behavior; 1=slightly slowed or abnormal movements; 2=mild impairments, limited ambulation unless disturbed; 3=moderate impairment, limited ambulation even when disturbed, frequent abnormal postures; 4=severe impairment, almost no ambulation, sustained abnormal postures; 5=prolonged immobility in abnormal postures.
The stress and exertion associated with the rotarod trials resulted in long lasting episodes of paroxysmal dyskinesia. The severe attacks (scores of 4–5) typically last for 30–60 minutes and are very stereotyped. These attacks start with the involuntary extension of hind limbs followed by lowering of the hips and extending the knees, ankles and paws. Throughout these movements the back is abnormally arched. This posturing then spreads to the rest of the body, with particularly severe contractions of the neck and face muscles (Campbell et al., 1999). Most of these characteristics are notable every time a tottering mouse has a severe attack making the severe episodes (scores of 4–5) unambiguously distinguishable from milder motor impairments (scores of 2–3).
Scoring of the severity of the symptoms was carried out by one of the authors with prior knowledge of the treatment implemented. To ascertain that this scoring was not biased, a few colleagues who were blind to the treatment were also asked to provide dyskinesia scores by reviewing a number of video-taped episodes. There were no significant differences between the scores assigned by five such observers blind to the treatment of the mice and the scores obtained by the authors.
Chlorzoxazone (CHZ) and 4-aminopyridine (4-AP) were orally administered to tottering and wild type mice by adding it to their drinking water. The CHZ solution was prepared fresh every day by adding CHZ to a 0.1% solution of hydroxypropyl-β-cyclodextrin (Tocris Bioscience, Ellisville, MO, USA), and then adding a few drops of 1 N NaOH until CHZ was fully dissolved. The solution was supplemented with 10% sucrose to improve its taste and thus to ensure its consumption. The 4-AP solution was also prepared daily from a 10X frozen stock. To improve the accuracy of the water consumption measurements, the large water bottle in the cages were replaced with graduated 15 ml plastic tubes. The weight of the animals and the extent of their water intake were monitored daily throughout the experiment.
All data are reported as mean ± S.E.M. To determine statistical significance One-way ANOVA was followed by Bonferroni post test for multiple comparisons or a paired t-test (figure 2H only). Differences were considered to be statistically significant only at p<0.05.
While the output of Purkinje cells onto DCN neurons is dynamically regulated by their synaptic inputs, Purkinje cells are intrinsically active and fire regularly in the absence of any synaptic input (Hausser and Clark, 1997; Raman and Bean, 1999; Womack and Khodakhah, 2002). Therapeutic concentrations of 4-AP may thus increase the inhibitory synaptic drive of Purkinje cells onto DCN neurons by elevating the rate of their intrinsic pacemaking. In humans, 4-AP has a relatively short half life of 3–4 hours (Hayes et al., 2003) and its beneficial effects in alleviating motor symptoms in EA2 and cerebellar and ocular motor disorders have been obtained within an hour of administration of a single 10 mg dose, or with 5 mg doses given thrice daily (Kalla et al., 2007; Strupp et al., 2004). Given its pharmacokinetics in humans, the maximum plasma concentration of 4-AP following a single 10 mg dose corresponds approximately to a concentration just shy of 1 µM. While throughout this study we explored the effects of 1, 5 and 10 µM 4-AP to more clearly delineate trends in potential physiologic effects, it should be noted that therapeutically only the 1 µM concentration is of relevance particularly since higher doses are pro-convulsant (Bever, Jr. et al., 1994).
To explore whether therapeutic concentrations of 4-AP affect the rate of spontaneous activity of Purkinje cells we noninvasively monitored their activity with extracellular recordings in acutely prepared cerebellar slices. After recording the activity of individual Purkinje cells for several minutes to obtain their baseline firing rate we applied 4-AP. Bath perfusion of neither 1 nor 5 µM 4-AP affected the rate of spontaneous activity of Purkinje cells (control firing rate = 42.6±2.4 spikes per second, in 1 µM 4-AP = 43.8.±4.5 spikes per second, p>0.1, and 42.6±5.8 spikes per second in 5 µM 4-AP, p>0.1; Figures 1A–C, n=13 cells). 10 µM 4-AP marginally reduced the firing rate (33.9±3.2 spikes per second, p<0.05), presumably by increasing GABAergic drive onto Purkinje cells because 4-AP did not affect the firing rate of Purkinje cells when synaptic transmission was pharmacologically blocked (Figures 1 D and F). Comparable results were obtained when GABA-ergic synaptic transmission was selectively blocked (data not shown). Thus, the data demonstrate that therapeutic concentrations of 4-AP do not affect intrinsic pacemaking of Purkinje cells. By inference, given that at these concentrations 4-AP did not reduce the firing rate of Purkinje cells in experiments where inhibitory synaptic transmission was intact it is likely that 4-AP also does not appreciably change the firing rate of the molecular layer interneurons.
We next examined whether therapeutic concentrations of 4-AP potentiate the response of Purkinje cells to excitatory synaptic inputs. Purkinje cells receive excitatory inputs from more than 150,000 parallel fiber (PF) synapses. In principle, 4-AP can potentiate PF-evoked responses in Purkinje cells by increasing either the pre- or post-synaptic excitability (or both). At some synapses even low micromolar concentrations of 4-AP can increase neurotransmitter release by broadening the action potential thus leading to greater calcium influx (Sacco and Tempia, 2002). 4-AP can also increase the response of a Purkinje cell to the same synaptic input if it blocks dendritic or perhaps even somatic potassium channels. We explored these possibilities by first quantifying the response of spontaneously firing Purkinje cells to extracellular electrical stimulation of PFs. The intensity of the electrical stimulation was adjusted such that the maximum firing rate of the target Purkinje cells increased to about 100 spikes per second (mean = 113.4±9.1 spikes per second, n=8 cells). 4-AP, in a concentration dependent manner, potentiated these responses (Figures 2A and B) such that PF-evoked responses marginally increased to 129.1±16.4 spikes per second in the presence of 1 µM 4-AP (p>0.1), to 174.6±28.8 spikes per second with 5 µM 4-AP (p<0.05), and to 228.6±37.7 spikes per second with 10 µM 4-AP (p<0.001).
To delineate the mechanism by which 4-AP potentiates PF-evoked responses in Purkinje cells we examined its effect on PF-evoked excitatory post synaptic currents (EPSCs) in voltage-clamped Purkinje cells. Use of a cesium-based internal solution to block potassium channels ensured a better space-clamp and eliminated the possibility of a post-synaptic action of 4-AP. Similar to that seen with PF-evoked increases in the firing rate of Purkinje cells, 4-AP potentiated PF-evoked EPSCs in a concentration dependent manner (Figures 2C and D, n=12 cells). The average peak amplitude of EPSCs was 318±24 pA under control conditions. With 1 µM 4-AP it increased to 399±46 pA although this value was not statistically different from control (p>0.1). The average EPSC amplitude increased further to 473±59 pA with 5 µM (p<0.05), and to 614±59 pA with 10 µM 4-AP (p<0.001). Moreover, the extent to which 4-AP augmented PF-evoked EPSCs was the same as its potentiating effects on PF-evoked increases in the firing rate (Figure 2E). Because in Purkinje cells the increase in the firing rate is a linear function of the charge injected by the PF-evoked EPSCs (Walter and Khodakhah, 2006), our data demonstrate that therapeutic concentrations of 4-AP do not affect integration of synaptic inputs by Purkinje cells by altering their dendritic or somatic excitability. Instead, the effects of low concentrations of 4-AP are solely restricted to increasing the amplitude of EPSCs.
In the experiments reported above, a number of PFs were activated by a brief current applied to an extracellular stimulation electrode placed in the molecular layer. With this stimulation paradigm the increase in the evoked-EPSC amplitudes in the presence of 4-AP could be either due to augmented neurotransmitter release from the same number of PFs, an increase in the number of PFs activated by each stimulation, or both. The former can occur if 4-AP blocks a sufficient number of potassium channels in PF terminals to broaden the pre-synaptic action potential. This will yield a larger calcium influx thus increasing the release probability at each synapse. Alternatively, if blockade of axonal potassium channels with 4-AP were to increase their input resistance more PFs might be brought to threshold with each stimulation. To delineate the potential contribution of each of these mechanisms to the increased EPSC amplitudes in the presence of 4-AP we examined the paired pulse ratio of two successive EPSCs before and after application of 4-AP. At the PF-to-Purkinje cell synapse changes in the paired pulse ratio are faithfully correlated with changes in release probability (Foster et al., 2002). Under control conditions, when PFs were electrically activated with a pair of stimulation pulses 50 ms apart, the ratio of the peak amplitude of the second EPSC to the first was on average 1.43±0.03 (Figure 2F). This paired pulse ratio was not significantly different in the presence of 1 µM 4-AP (1.43±0.14, p>0.3), 5 µM 4-AP (1.36±0.04, p>0.3), or even 10 µM 4-AP (1.48±0.03, p>0.3). As a control we increased the concentration of 4-AP even further. As previously reported (Buckle and Haas, 1982), with 100 µM 4-AP the paired pulse ratio reduced to 1.14±0.12, and with 1 mM 4-AP the paired pulse facilitation changed to significant paired pulse depression (data not shown). These results indicate that at therapeutically relevant concentrations 4-AP does not change the probability of neurotransmitter release at PF synapses. Instead, under our experimental conditions 4-AP simply increased the number of PFs that are brought to threshold with each stimulation (although at the therapeutically relevant concentration of 1 µM the effects were small and not statistically significant). These effects of 4-AP are unlikely to be of physiological significance and possibly reflect the necessarily artifactual manner in which PFs were activated in these experiments. This is because in vivo granule cells, the neurons whose axons form PFs, are brought to threshold for firing of an action potential even by single quanta released from their mossy fiber synaptic inputs (Chadderton et al., 2004) and thus with therapeutic concentrations of 4-AP one would not anticipate a significant increase in the number of PFs activated under physiological conditions. In agreement with this supposition, as shown in Figures 2G and H, 5 µM 4-AP failed to potentiate PF-evoked EPSCs in voltage-clamped Purkinje cells when the stimulation electrode was positioned in the white matter to activate the mossy fiber inputs to the granule cells (control EPSC amplitude = 159±26 pA vs. in 5 µM 4-AP = 162±30 pA, p>0.85, Paired t-test, n=3 cells). Similarly, the paired pulse ratio of EPSCs evoked by mossy fiber stimulation were not affected after addition of 5 µM 4-AP (control = 1.18±0.05 vs. 1.21±0.4 in 4-AP; p>0.1).
Comparable experiments demonstrated that therapeutic concentrations of 4-AP do not alter release probability at the Purkinje cell to DCN synapses (see Supplementary Material 1).
The experiments described above show that therapeutic concentrations of 4-AP do not increase the excitability or rate of activity of Purkinje cells. Moreover, 4-AP does not increase the release probability at the PF to Purkinje cell, or the Purkinje cell to DCN synapses. Combined, in contrast with what is widely believed (Jen et al., 2007; Strupp et al., 2007; Strupp et al., 2008; Strupp and Brandt, 2006), our data suggest that it is unlikely that the therapeutic mode of action of 4-AP in EA2 is by increasing the inhibitory drive of Purkinje cells onto DCN neurons. An alternative hypothesis regarding the therapeutic mode of action of 4-AP proposed by Otis and Jen (Otis and Jen, 2006) is that 4-AP restores the precision of Purkinje cell pacemaking by broadening the duration of action potentials. This mode of action is based on the findings that loss in the precision of Purkinje cell pacemaking has been suggested to contribute to the motor symptoms of EA2 (Walter et al., 2006). Indeed in vivo cerebellar perfusion of KCa channel activator EBIO restores the precision of Purkinje cell pacemaking and significantly improves motor function in mouse models of EA2 (Walter et al., 2006). This potential therapeutic mode of action of 4-AP, while not experimentally examined, has been largely dismissed.
To test this hypothesis we examined the activity of Purkinje cells in acutely prepared cerebellar slices from the tottering mice. These mice suffer from a spontaneous mutation in the pore-forming subunit of the P/Q-type calcium channel which reduces their current density (Fletcher et al., 1996) and are one of the most studied and best established models of EA2 (Jinnah et al., 2005; Pietrobon, 2005). Similar to EA2 patients, the tottering mice show baseline ataxia and episodes of severe dyskinesia which are triggered by stress, caffeine and ethanol (Fureman et al., 2002; Shirley et al., 2008). While the mutation in tottering mice results in a very large decrease in the P/Q calcium current density and more severe episodes of dyskinesia than typically seen in EA2 patients, these mice have served as a very useful and reliable model of EA2 both for examining the etiology of the disorder and also for surveying therapeutic options. In fact, acute administration of 4-AP is as effective in alleviating the motor symptoms of the tottering mice as it is in humans (Weisz et al., 2005).
Consistent with observations in other mutant animal models of EA2 such as the ducky and leaner mice (Ovsepian and Friel, 2008; Walter et al., 2006), we found that the intrinsic pacemaking of tottering Purkinje cells was also highly erratic (Figures 3A–C). Thus, the coefficient of variation of the interspike intervals in tottering Purkinje cells was 0.18±0.03 (Figure 3C, n=15 cells), a value significantly higher than 0.07±0.01 seen in age-matched wild type Purkinje cells (p<0.001, n=22 cells). Application of 4-AP, in a concentration dependent manner, made the firing of tottering Purkinje cells more regular (Figure 3B and C). Even at a concentration of 1 µM, 4-AP reduced the coefficient of variation of the interspike intervals to 0.10±0.01 (p<0.001, n=15 cells). 5 µM 4-AP reduced the coefficient of variation of the interspike intervals to 0.07±0.01 (p<0.001, n=20 cells), and 10 µM to 0.05±0.01 (p<0.001, n=12 cells). The latter values were not statistically different from that seen in age-matched wild type Purkinje cells (p>0.1). While 10 µM 4-AP reduced the firing rate of tottering Purkinje cells from an average of 69.2±4.5 spikes per second to 46.1±2.7 spikes per second (p<0.01), the effects of 1 and 5 µM 4-AP on the firing rate were minor and not statistically significant (1 µM 4-AP = 64.6±5.1 spikes per second, 5 µM 4-AP = 62.4±4.8 spikes per second; Figure 3D, p>0.1). Thus, 4-AP at the therapeutically-relevant concentration of 1 µM significantly improves the precision of pacemaking of tottering Purkinje cells. In a separate set of studies we used low concentrations of cadmium to block calcium channels to decrease the precision of pacemaking in rat Purkinje cells (Walter et al., 2006). We found that the precision of pacemaking in these cells was restored with the same concentrations of 4-AP that were effective in tottering mice (see Supplementary Material 2).
Activators of calcium-dependant potassium channels restore the precision of Purkinje cell pacemaking in EA2 animal models by increasing the amplitude of action potential after-hyperpolarization (AHP) (Walter et al., 2006). We tested the possibility that by blocking a subset of potassium channels 4-AP restores the precision of Purkinje cell pacemaking by broadening their action potentials, thus increasing the amplitude of their AHPs.
We first examined the extracellularly recorded spikes. These voltage deflections correspond to the derivative of the actual membrane potential of the target Purkinje cell. Thus, the start of the negative voltage deflection corresponds to the start of the rising phase of the action potential and, the positive peak of the signal relates to the time at which the rate of membrane potential repolarization is the greatest, that is the time at which the net outward ionic current is at its maximum. The positive voltage thereafter corresponds with the continuation of the down-stroke of the action potential and its AHP. Compared with the direct examination of the action potential width with whole-cell recordings, using the extracellular signals has the advantage that it avoids dialysis of the cell and the associated changes in calcium buffering and homeostasis. Quantification of the data demonstrated that 4-AP broadens the action potential in a concentration-dependent manner (Figures 4A and B). The time to the peak positive voltage deflection (measured as the time taken from the beginning of the negative deflection to the peak of the positive deflection) increased from an average of 0.75±0.04 ms under control conditions to 0.94±0.05 ms in 1 µM 4-AP (p<0.05), and to 1.06±0.03 ms and 1.12±0.06 ms with 5 and 10 µM 4-AP respectively (p<0.001 for both cases). Consistent with the hypothesis that broadening of the action potential was the cause of improved pacemaking, there was a clear linear relationship between the average duration and the coefficient of variation (Figure 4C). In agreement with the notion that 4-AP prolongs the AHP, in the presence of 4-AP the positive deflections of the extracellular signals were longer (see inset in Figure 4A).
As a complementary approach we also examined the effect of 4-AP on the action potential waveform using whole-cell current clamp recordings, taking into consideration the caveat that whole-cell recordings do dialyze the cells and affect calcium buffering (and may also alter other intracellular biochemical signaling pathways). As can be noted in the sample traces shown, 4-AP dose-dependently increased the width of the action potential and also the absolute AHP potential (Figure 4D). The average maximum AHP potential in control conditions was −59.8±1.6 mV (n=6 cells) which after application of 1 µM 4-AP increased to −61.4±0.6 mV (Figure 4E). 5 and 10 µM 4-AP increased the maximum AHP potential further to −63.4±1.1 mV (p<0.05) and −65.8±1.3 mV (p<0.001). In addition to the absolute AHP potential, the AHP amplitude (measured from threshold) increased significantly with 4-AP (Figure 4F, control = 10.9±0.5 mV, 1 µM 4-AP = 11.3±0.2 mV, 5 µM 4-AP = 13.7±0.6 mV, p<0.05, 10 µM 4-AP = 14.8± 0.9 mV, p<0.01). Finally, the width of the action potential (measured as the time taken from threshold to 10% peak action potential amplitude during the down stroke of action potential) also increased when 4-AP was added (Figure 4G; control = 0.50±0.01 ms, 1 µM 4-AP = 0.51±0.017 ms, 5 µM 4-AP = 0.55±0.03 ms, p<0.01, and 10 µM 4-AP = 0.56±0.015 ms, p<0.001).
Collectively the data presented demonstrate that at concentrations effective in the treatment of cerebellar-related motor disorders, 4-AP increases the regularity of a Purkinje cell’s pacemaking by prolonging the duration of its action potentials. The broader action potentials are accompanied with larger AHPs, presumably because of the greater calcium influx associated with the broader action potentials and the additional activation of calcium-dependent potassium conductances. This larger calcium influx is likely to compensate, in part, for the reduced P/Q-type calcium current density associated with EA2 mutations (Barclay et al., 2001; Dove et al., 1998; Fletcher et al., 1996).
Based on the affinity of 4-AP for block of various potassium channel subtypes (Coetzee et al., 1999) and the expression pattern of channels in the cerebellum and specifically in Purkinje cells (Chung et al., 2001; Chung et al., 2005; Hurlock et al., 2008; Madeja et al., 1997; Martina et al., 2003), the most likely target of therapeutic concentrations of 4-AP are the Kv1 family of potassium channels, specifically the Kv1.5 channels which have the highest affinity for 4-AP. To a much lesser extent Kv3.1 or Kv3.3 channels could also be the targets of 4-AP. Therefore we tested whether the Kv1.5 channel blocker 2-isopropyl-5-methylcyclohexyl (DPO) (Lagrutta et al., 2006) could substitute for 4-AP in restoring the precision of pacemaking. We monitored the activity of normal Purkinje cells in slices by extracellular recordings, and as before applied cadmium to partially block calcium channels and make the firing of Purkinje cells irregular. This was reflected as an increase the coefficient of variation of the interspike intervals (Figure 5A).
Bath application of 5 µM DPO decreased the coefficient of variation Purkinje cell firing to levels comparable to control conditions (Figure 5A and B), without altering the firing rate (Figure 5C). The coefficient of variation of interspike intervals in control conditions was 0.052±0.005 which after application of cadmium increased 0.114±0.007 (Figure 5B, n=6–12 cells, p<0.001). Application of DPO reduced it to 0.071±0.003 (p<0.001 vs. cadmium). Furthermore, application of 10 µM 4-AP after DPO did not change further reduce the coefficient of variation of interspike intervals (CV = 0.07±0.003 after application of 4-AP, p<0.001 vs. cadmium, p>0.1 vs. DPO). This occlusion of effects suggests (but of course does not prove) that the effects of 4-AP were likely mediated by the blockade of Kv1 channels.
The data presented above suggest that by restoring the precision of Purkinje cell pacemaking the therapeutic mode of action of 4-AP in EA2 may be the same as that of EBIO, an activator of calcium-dependent potassium (KCa) channels. EBIO improved the precision of Purkinje cell pacemaking in slices of EA2 mutant mice and lessened their ataxia when chronically perfused into their cerebellum in vivo (Walter et al., 2006). While in vivo perfusion of EBIO could in principle affect other cerebellar neurons such as DCN cells, at the concentrations used their most likely target were cerebellar Purkinje cells (Alvina and Khodakhah, 2008). If the primary mode of action of 4-AP is indeed restoring the precision of Purkinje cell pacemaking, then one would anticipate that the beneficial effects of 4-AP in reducing motor symptoms should not be additive to those of KCa channel activators; i.e. co-administration of 4-AP with a KCa channel activator should only be as effective as the sole administration of either drug with the highest efficacy. We tested this hypothesis by examining whether treatment of tottering mice concurrently with both a KCa channel activator and 4-AP was more effective in reducing their motor symptoms than with either one alone.
We first calculated the concentration of 4-AP that needed to be added to the drinking water of tottering mice to achieve the therapeutic concentration of 4-AP found in the plasma of patients (~50 ng/ml). Given that in rodents the half-life of 4-AP is ~2 hr (Capacio et al., 1996), and considering that the tottering mice weigh an average of 20 g, we estimated that a daily 4-AP consumption of ≈0.1 mg will result in an equivalent, but temporally uniform, 4-AP plasma concentration in mice. Since on average the tottering mice consume ~2.5 ml of water daily, we supplied them with a 425 µM solution of 4-AP which they readily drank.
Patients affected by EA2 not only have episodic attacks of dyskinesia, but also show mild baseline ataxia that progress in severity with time (Jen et al., 2004). Thus we examined the efficacy of oral administration of 4-AP in alleviating baseline ataxia in the tg/tg mice using an accelerating rotarod paradigm (Crawley, 2008; Walter et al., 2006). The performance of tottering (n=13), but not the wild type mice (n=14), increased on the rotarod when 4-AP was added to their drinking water (Figure 6A). During the pre-treatment period the performance of the tottering on the rotarod reached on average 10.9±0.8 RPM (Figure 6B), whereas during treatment with 4-AP it increased to 13.8±0.9 RPM (p<0.05). 4-AP had no effect on the performance of the wild type mice (Pre-treatment = 34.5±0.9 RPM, 4-AP = 34.9±1.4 RPM, p>0.1).
We then set out to examine whether the beneficial effects of 4-AP were additive to those of the KCa channel activator chlorzoxazone (CHZ) (Syme et al., 2000). CHZ is an analogue of EBIO (Cao et al., 2001) and crosses the blood brain barrier (Chou et al., 2004) thus allowing it to be supplied to mice by adding it to their drinking. In experiments performed in acutely prepared cerebellar slices we have recently determined the concentrations of CHZ that restores the precision of pacemaking in tottering Purkinje cells (Alviña and Khodakhah, under review). Moreover, using this information, and taking into account the pharmacokinetics of CHZ (Wan et al., 2006), we have found that supplementing the drinking water of the tottering mice with 15 mM CHZ markedly improves their performance on the rotarod (Alviña and Khodakhah, under review).
We thus examined whether in the same group of mice treated with 4-AP the beneficial effects of CHZ were additive. While the performance of the tottering mice on the rotarod improved further when, in addition to 4-AP, CHZ was also added to their drinking water, this increase was small and just shy of being statistically different (from 13.8±0.9 RPM in 4-AP alone to 16.2±0.9 RPM with the addition of CHZ, p=0.06). The performance of tottering mice continued at its high level when 4-AP was removed and the mice only received CHZ (16.2±0.8 RPM, p>0.1 vs. 4AP+CHZ combined), suggesting that there was little benefit in combining the two drugs. Throughout the various treatments, the performance of wild type mice did not change (Pre-treatment = 34.5±0.9 RPM, 4-AP alone = 34.9±1.4 RPM, 4-AP and CHZ = 34.9±1.1 RPM, CHZ alone = 35.0±1.1 RPM and Post-treatment = 35.5±1.1 RPM, p>0.5 for all treatments vs. Pre-treatment). Thus, at least with respect to improving baseline motor function, the data are consistent with the hypothesis that both 4-AP and CHZ may have a common mode of action.
It is plausible that while the mode of action of 4-AP in alleviating baseline ataxia may be the same as that of KCa activators, 4-AP may have a different mechanism of action in preventing the episodes of dyskinesia associated with EA2. We thus similarly compared the efficacy of 4-AP alone with that of its co-administration with CHZ in reducing the frequency, severity, and duration of stress-induced episodes of dyskinesia triggered by the rotarod session in tottering mice (Figure 7A). Both 4-AP and CHZ significantly reduced the average overall severity of symptoms after stress (Figure 7B), although CHZ produced a slightly larger reduction and there was little benefit in the co-administration of the two drugs.
To scrutinize the impact of these treatments on frequency, duration and severity of episodes we first analyzed all attacks of dyskinesia irrespective of their individual severity. 4-AP treatment reduced the frequency of attacks to 81.7±5.2% of the baseline pre-treatment values (Figure 7C, p<0.01). Addition of CHZ to 4-AP further reduced the frequency of attacks to 59.6±7.8% (p<0.001 vs. Pre). When CHZ was administered without 4-AP, the frequency of all attacks were 50.0±5.4% of baseline (p<0.001 vs. Pre), a value statistically different when compared with the 4-AP treatment alone (p<0.05), but not statistically different from CHZ and 4-AP combined (p>0.1). After returning mice to normal drinking water, the frequency of attacks in the tottering mice increased to its pre-treatment values (98.5±1.5%, p>0.1 vs. Pre). In addition, treatment with both 4-AP and CHZ comparably reduced the severity (Figure 7D) and the duration of all attacks of dyskinesia (Figure 7E).
The occurrence of the most severe attacks (those with a maximum score of ≥3.5) was also reduced by all 3 treatments (Figure 7F). Even though both 4-AP and CHZ were effective in reducing the frequency of these severe attacks we found that CHZ was slightly more effective than 4-AP alone (22.1±3.1% of baseline vs. 33.7±3.2%, p<0.05), and there was no advantage in co-application of 4-AP with CHZ. In all cases, when severe attacks occurred they were as severe as those prior to treatment (Figure 7G) although their duration was significantly and comparably briefer with all three treatments (Figure 7H).
The pie charts in Figure 7I directly compare the efficacy of 4-AP, CHZ, and 4-AP with CHZ in reducing the frequency of stress-evoked attacks of dyskinesia in tottering mice. While both 4-AP and CHZ potently reduce the frequency of attacks, CHZ is much more effective in increasing the percentage of attack-free sessions in which stress did not cause any motor impairment whatsoever (50% in CHZ alone vs. 18.3% during 4-AP). Moreover, there was no benefit in co-application of the two compounds.
In summary, our findings are consistent with the hypothesis that both CHZ and 4-AP have a common mode of action. Moreover, it is plausible and perhaps quite likely that their therapeutic efficacies are the consequence of their ability to improve the precision of pacemaking in Purkinje cells.
Episodic ataxia type-2 is the most prevalent form of episodic ataxia (Jen et al., 2007). The symptoms of EA2 patients are mostly cerebellar in origin and in animal models of EA2 removal of the cerebellum, or genetic elimination of Purkinje cells eliminates the episodes of dyskinesia (Campbell et al., 1999; Grusser-Cornehls and Baurle, 2001). The calcium channels which are affected in EA2 are highly expressed throughout the CNS (Evans and Zamponi, 2006) and are particularly enriched in axon terminals and in cerebellar Purkinje cells (Mori et al., 1991; Stea et al., 1994; Usowicz et al., 1992). Calcium influx primarily through these channels mediates synaptic transmission at CNS nerve endings (Evans and Zamponi, 2006) and initial hypotheses regarding the etiology of EA2 focused on the dysfunction of cerebellar synapses. However, while alterations in synaptic transmission at the parallel and climbing fiber synapses are notable in some animal models of EA2, they are relatively minor and not as profound as that anticipated, mainly as a consequence of functional compensation by other voltage gated calcium channels (Matsushita et al., 2002; Ovsepian and Friel, 2008).
Within the cerebellum P/Q-type calcium channels are particularly enriched in Purkinje cells, where the dendritic ones generate calcium action potentials (Llinas and Sugimori, 1980), and the somatic ones provide the sole source of calcium for activation of KCa channels (Womack et al., 2004). Indeed, in Purkinje cells the net calcium-dependent current associated with each action potential is outward (Raman and Bean, 1999) and the P/Q-channel mediated activation of KCa channels is required to maintain the precision of their pacemaking (Walter et al., 2006; Womack and Khodakhah, 2004). In animal models of EA2 the precision of Purkinje pacemaking is significantly deteriorated and there is little evidence for the presence of compensatory mechanisms (Hoebeek et al., 2005; Walter et al., 2006). It has been therefore suggested that irregular firing of cerebellar Purkinje cells contributes to motor symptoms associated with EA2 (Walter et al., 2006).
There are presently few therapeutic options available for EA2 patients with acetazolamide (ACTZ) and 4-AP constituting the most commonly used drugs (Jen et al., 2007; Strupp et al., 2007). Many patients respond well to ACTZ which both improves some of the baseline symptoms and reduces the frequency of episodes of dyskinesia (Friedman and Hollmann, 1987; Harno et al., 2004; Zasorin et al., 1983). However, with time, many patients become non-responsive to ACTZ treatment (Jen et al., 2007; Strupp et al., 2007). The mechanism of action of ACTZ in the treatment of EA2 is not understood. Because ACTZ is a carbonic anhydrase inhibitor (Maren, 1967), it is suggested that ACTZ prevents elevations in the intracellular pH (Strupp et al., 2007) although it is not clear why this should be helpful.
More recently 4-AP has been used to reduce the symptoms of EA2 (Lohle et al., 2008; Strupp et al., 2004) and it is found to be effective in both improving baseline motor coordination and in reducing the frequency and severity of the episodic attacks (Glasauer et al., 2005; Lohle et al., 2008; Strupp et al., 2007). However, 4-AP has to be used with caution since as a potassium channel blocker it can be epileptogenic (Bever, Jr. et al., 1994; Judge and Bever, Jr., 2006).
4-AP was first prescribed based on the notion that, because of the reduced calcium current density associated with EA2 mutations, Purkinje cells must be less active (Glasauer et al., 2005; Strupp et al., 2004; Strupp et al., 2005). The remarkable efficacy of 4-AP in alleviating cerebellar motor symptoms in several patients has indirectly substantiated this presumed mode of action (Glasauer et al., 2005; Kalla et al., 2007; Lohle et al., 2008; Strupp et al., 2004). The data presented here demonstrate that therapeutic concentrations of 4-AP do not alter the rate of activity of Purkinje cells. Nor does 4-AP, at the relevant concentrations, alter synaptic transmission at the parallel fiber-to-Purkinje cell, or the Purkinje cell-to-DCN synapses. The only functional consequence of application of 4-AP that we could discern was that of restoring the precision of pacemaking in the EA2 mutant Purkinje cells by blocking potassium channels and prolonging the AP and increasing its AHP.
The therapeutic mode of action of 4-AP described here is consistent with the finding that in mouse models of EA2 direct perfusion of the KCa channel activator EBIO into the cerebellum to improve the precision of Purkinje cell pacemaking lessens their motor symptoms (Walter et al., 2006). The fact that the beneficial effects of CHZ and 4-AP were not additive also corroborates the hypothesis that the mode of action of 4-AP might be the same as EBIO (although this does not by any means constitute proof). The greater efficacy of CHZ compared with 4-AP in alleviating motor symptoms of the tottering mice (even though in slices their efficacies in restoring the precision of Purkinje cell pacemaking is comparable) may be attributable to differences in their pharmacokinetics, in their ability to cross the blood-brain barrier, or differences in their potential side effects in vivo.
Based on its affinity for potassium channels the most likely target of therapeutic concentrations of 4-AP are the Kv1 family of potassium channels (Coetzee et al., 1999), and possibly the Kv1.5 potassium channels which are thought to be expressed in Purkinje cells (Chung et al., 2001; Chung et al., 2005; Madeja et al., 1997). The efficacy of the Kv1.5 channel blocker DPO (Lagrutta et al., 2006), in restoring the precision of pacemaking in Purkinje cells substantiates the hypothesis that therapeutic concentrations of 4-AP target a member of the Kv1 family of potassium channels. A higher concentrations 4-AP blocks a wide range of potassium channels and is a potent pro-convulsant. Thus selective blockers of Kv1 channels (and pending further confirmation with more specific channel blockers particularly the Kv1.5 subfamily) may constitute safer and more effective candidates for treatment of EA2 and perhaps other cerebellar ataxia.
Studies on the effects of 4-AP on Purkinje cells were initiated by Dr. Simin Khavandgar during her tenure in KK’s lab. We are particularly grateful to her, and also thank members of KK lab for help in scoring the motor behavior of the mice and for discussions. This work was funded by the NIH.