Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Generalized, or petit mal, absence epilepsies are characterized by periods of synchronized neuronal activity, generating a 3-Hz spike-wave pattern on the electroencephalogram. Considerable evidence suggests that spike-wave discharges are produced by synchronized oscillations of cortical, reticular thalamic, and corticothalamic neurons (Gloor and Fariello, 1988; Steriade et al., 1993). Oscillatory firing of corticothalamic neurons requires the activity of low threshold Ca²⁺ spikes (LTS; Steriade and Llinas, 1988; McCormick and Bal, 1997),which are produced by T-type Ca²⁺ channels (Crunelli et al., 1989; Suzuki and Rogawski, 1989).

Ethosuximide is considered the prototypical absence seizure drug that works by inhibition of T-type channels. This hypothesis was based on the finding that ethosuximide could partially block T-type currents in thalamic neurons at therapeutically relevant concentrations (Coulter et al., 1989a). Support for this hypothesis came from studies with related analogs: 1) T-type current block was also observed at therapeutically relevant concentrations of methyl-phenylsuccinimide (MPS), the active metabolite of methsuximide; and 2) no block was observed using either the inactive analog succinimide or the convulsant analog tetra-methylsuccinimide (Coulter et al., 1990a). The sensitivity of T-type channels to MPS has been confirmed using dorsal root ganglion neurons (Todorovic and Lingle, 1998). With one exception (Kostyuk et al., 1992), other studies have failed to confirm ethosuximide block of T-type currents at therapeutically relevant concentrations (Todorovic and Lingle, 1998), leading to the suggestion that block of other ionic channels may be more relevant (Leresche et al., 1998).

Many factors interfere with the pharmacological characterization of native T-type channels, such as interference from other ionic conductances, the lack of highly selective blockers, and adequate voltage control. Most studies have been performed in isolated neurons that have lost their dendritic trees and, hence, most of their T-type channels (Destexhe et al., 1998). We have recently cloned the cDNAs of three T-type Ca²⁺ channels (1G, Ca_v3.1; 1H, Ca_v3.2; and 1I, Ca_v3.3), and we now report the cloning of human 1I (Perez-Reyes, 1999; Cribbs et al., 2000). Expression of the cloned channels in HEK-293 cells provides an excellent assay system to test the pharmacology of T-type currents, since they contain very little background currents under appropriate assay conditions. We report for the first time the block of human T-type channels by two succinimide antiepileptic drugs, ethosuximide and MPS, the active metabolite of methsuximide. Our results show that block is state-dependent, with less block observed at highly hyperpolarized holding potentials. This finding may partially explain the controversy of whether ethosuximide block of T-type currents is therapeutically relevant.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. The following stably transfected HEK-293 cell lines were used in this study: h1G-Q39, containing the human 1G channel, Ca_v3.1a (GenBank accession number AF190860; Cribbs et al., 2000); h1H-Q31, containing the Hh7 plasmid construct of human 1H, Ca_v3.2 (GenBank accession number AF073931; Cribbs et al., 1998); and h1I-14, containing the LT4 plasmid construct of 1I, Ca_v3.3 (described below). All chemicals were from either Sigma (St. Louis, MO) or Aldrich (Milwaukee, WI).

Cloning of a Human 1I-cDNA. Human brain cDNA libraries derived from either fetal brain or cerebellum (CLONTECH, Palo Alto, CA) were screened using the rat 1I as probe. Screening was done by filter hybridization according to the manufacturer's protocol. The cDNA probes were synthesized using [-³²P]dCTP and the Ready-To-Go labeling kit (Amersham Pharmacia Biotech Piscataway, NJ). Positive clones were plaque-purified and then subcloned into pUC18 for sequencing. BLAST searches of the GenBank with the rat 1I identified the genomic DNA encoding the human CACNA1I gene (AL022312), allowing us to design PCR primers to clone the 5' end. Overlapping clones were selected and ligated in the vector pcDNA3 (Invitrogen, Carlsbad, CA), generating the clone LT4. The nucleotide sequence of human 1I has also been reported by two groups (Mittman et al., 1999; Monteil et al., 2000). The sequence of LT4 is similar to that reported in AF211189, in that it lacks putative exon 9, but is longer at the 3' end, extending to nucleotide 6214 of AF393329.

Generation of a Stably Transfected Cell Line. Stable cell lines expressing human 1I were constructed by transfecting HEK-293 cells with linearized LT4 plasmid. HEK cells were maintained in DMEM, 10% fetal bovine serum, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells were plated at a density of 1 × 10⁶ per 100-mm plate. One day later, fresh media was added, and the cells were transfected with 10 µg of plasmid DNA by the calcium phosphate method (CalPhos Maximizer Transfection Kit, CLONTECH) according to the manufacturer's protocol. Following transfection, the cells were selected in 1.0 mg/ml G-418 (Invitrogen) for 2 to 3 weeks. Single colonies were isolated, expanded, and then screened electrophysiologically for the expression of T-type Ca²⁺ current. One clone, h1I-14, was selected for further study.

Electrophysiological Analysis of HEK-293 Transfected Cells. HEK-293 cells were dissociated by digestion with 0.25% trypsin plus 1 mM EDTA for 2 min and then diluted 20-fold with DMEM. The cells were triturated, diluted 2-fold with DMEM, and then plated on cover slips. The cells were incubated at least 4 h and up to 2 days before electrophysiological studies. The standard internal pipette solution contained 135 mM CsCl, 10 mM EGTA, 4 mM Mg-ATP, 0.3 mM Na₃GTP, and 10 mM HEPES, pH adjusted to 7.3 with CsOH. Most experiments were performed using the following solution: 5 mM CaCl₂, 155 mM tetraethylammonium (TEA) chloride, and 10 mM HEPES, pH adjusted to 7.4 with TEA-OH. As noted in the figure legends, some experiments were performed using Ba²⁺ as charge carrier in 10 mM BaCl₂, 160 mM TEA chloride, 6 mM CsCl, and 10 mM HEPES, pH adjusted to 7.4 with TEA-OH. Addition of 1 mM Mg²⁺ to the external solution had no effect on the block of either 1G or 1I produced by 3 mM MPS.

Dose-Response Analysis. Succinimide analogs were freshly dissolved in external solution at a concentration of either 30 or 100 mM and then diluted in 10-fold increments. The recording chamber was a RC-25 (Warner Instrument Corp., Hampden, CT), which has a volume of 0.15 ml. Each test solution was perfused at 1 to 3 ml/min.

Data Analysis. Peak currents and exponential fits to currents were determined using Clampfit software (Axon Instruments). Dose-response analysis and graphing of the data were with Prism (GraphPad, San Diego, CA). The following equation was used to fit dose-response data
where X is the logarithm of concentration and Y is the response. The voltage dependence of activation was calculated using the solver function of Excel (Microsoft, Redmond, WA) to minimize the residual sum of squares from the simultaneous fit of the data with the Goldman-Hodgkin-Katz equation (Hille, 1992)

and a Boltzmann equation
where P_Ca is the Ca²⁺ permeability; z, the valence of the Ca²⁺ ion; V, the test potential; V_1/2, the mid-point of activation; F, Faraday's constant; R, the gas constant; T, absolute temperature; [Ca²⁺]_i, intracellular Ca²⁺ concentration (25 nM); [Ca²⁺]_o, extracellular Ca²⁺ (5 mM); and k is the slope factor. Values are reported as the means ± S.E.M.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Cloning of a Human 1I-cDNA. The human gene encoding Ca_v3.3 (1I), CACNA1I, was identified using homology searches of the GenBank HTGS database (Lee et al., 1999a). This sequence information was originally used to clone the rat 1I-cDNA. We then used the rat cDNA to screen human brain cDNA libraries. Overlapping cDNAs were assembled to produce a 6,268-base pair cDNA (LT4-pcDNA3) that encodes a protein of 220,706 Da. The cloning of human 1I was previously reported by Monteil et al. (2000). As noted under Experimental Procedures, our clone contains additional coding sequence at the 3' end. Transient transfection studies indicated that LT4 encoded a functional low voltage-activated Ca²⁺ channel with similar electrophysiological characteristics as observed with the rat (Lee et al., 1999a). The LT4 plasmid was then used to generate stably transfected cell lines, and h1I-14 was chosen for further study.

Ethosuximide Block of Cloned T-Type Ca²⁺ Channels. The effect of ethosuximide was tested on human T-type currents using whole cell recording from HEK-293 cell lines that were stably transfected with Ca_v3.1 (1G), Ca_v3.2 (1H), or Ca_v3.3 (1I). Currents were measured using either 5 mM Ca²⁺ or 10 mM Ba²⁺ as charge carrier and Cs-based intracellular solutions that contain ATP. Under these recording conditions the current density is approximately 40 pA/pF, such that the typically sized cell has 1000 pA of current during a test pulse to 30 mV. The size of the currents is quite stable for most recordings, although there is some run-up observed immediately after establishment of the whole cell clamp and some run-down after ~15 min of recording. Therefore, the stability of the currents was monitored in all experiments, and results are only presented for cells whose run-down rate was less than 2%/min. The effects of compounds were tested by dissolving each compound directly in external solution and then perfusing the compound over the cell. Various concentrations were tested on each cell without (Fig. 1, A and B) or with (Fig. 2) washout between concentrations. Little difference was observed among the results obtained with these two protocols, so the results have been pooled. The ethosuximide dose-response curve for block of human 1G currents is shown in Fig. 1C. The data were fit with the dose-response equation, yielding an apparent IC₅₀ of 10 ± 2 mM (n = 31 cells). Similar results were obtained using 5 mM Ca²⁺ as the charge carrier (IC₅₀ = 12 ± 2 mM, n = 16 cells). The response to concentrations below 1 mM was greater than expected from simple binding of a drug molecule to a single site on the channel. Since the deviation occurred in the therapeutic range of plasma ethosuximide concentrations (40-100 µg/ml or 0.3-0.7 mM; Browne et al., 1975), these measurements were repeated extensively. Clear inhibitory effects of ethosuximide could be detected with concentrations as low as 0.1 mM (Fig. 1D). The response to 100 mM also deviated from the fit. Since such a high concentration alters the osmolarity of the external solution and might cause a nonspecific effect, we adjusted for this by reducing the TEA concentration. In either case the block was 100%. Although the data are consistent with two binding sites, an alternative hypothesis is that ethosuximide block is state-dependent (Hille, 1992).

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Inhibition of human 1G T-type Ca2+ channels by ethosuximide. A, representative whole cell recordings showing the effect of various concentrations of ethosuximide on T-type currents. Whole cell Ba2+ (10 mM) currents recorded from an HEK-293 cell stably transfected with 1G subunit in response to voltage steps to 30 mV from a holding potential of 90 mV applied every 10 s. B, time course of peak currents normalized to control for the same cell shown in A. Ordinate axis, peak current during steady-state exposure to ethosuximide, normalized by the peak current before drug exposure, defined as the fraction of IBa remaining. C, dose-response relationship for ethosuximide block. Fraction of unblocked peak current is plotted against drug concentration. Number of cells investigated for each concentration is indicated near each data point. Data were fitted (smooth line) using a Hill equation (see Experimental Procedures). Note that data points representing drug concentrations lower that 1 mM deviated significantly from the fitted curve. D, inhibitory effect of 0.1 mM ethosuximide on peak currents averaged from five cells.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Inhibition of human T-type Ca2+ channels by MPS. A, representative whole cell recordings showing the effect of various concentrations of MPS on T-type currents. Whole cell Ca2+ (5 mM) currents recorded from an HEK-293 cell stably transfected with 1I subunit in response to voltage steps to 30 mV from a holding potential of 90 mV applied every 10 s. B, time course of peak current for the same cell shown in A. C, dose-response relationship for MPS block on Ca2+ currents expressed by 1G, 1H, and 1I channels. Fraction of unblocked peak current is plotted against drug concentration. IC50 values from fitted data (smooth lines) using a Hill equation were: 1G, 1.95 ± 0.2 mM; 1H, 3.03 ± 0.26 mM; and 1I, 1.82 ± 0.16 mM.

Methyl-Phenylsuccinimide Block of Cloned T-Type Channels. The antiepileptic drug methsuximide is rapidly demethylated to N-desmethylmethsuximide, which is also called MPS. Therapeutic plasma levels of MPS in well controlled patients range between 10 and 40 µg/ml or 50 and 200 µM (Strong et al., 1974; Porter et al., 1979; Wilder and Buchanan, 1981). The sensitivity of human T-type channels to block by MPS was measured as above. Figure 2 shows the time course of a typical experiment using 1I. MPS produced a rapid block of current, which was rapidly and almost completely reversed upon washout. The dose-response relationship of MPS block of the three cloned human T-type channels was measured using the peak Ca²⁺ current elicited during a pulse to 30 mV (Fig. 2C) from a holding potential of 90 mV. The 1G and 1I channels were equally sensitive, displaying apparent IC₅₀ values of 1.95 ± 0.19 (n = 7 cells) and 1.82 ± 0.16 mM (n = 14 cells), respectively, whereas 1H channels were slightly less sensitive (IC₅₀ = 3.0 ± 0.3 mM, n = 6 cells). The Hill coefficients were significantly greater than one: ~1.3 for 1G and 1I and 1.7 for 1H. As observed with ethosuximide, these results suggested that block was more complex than simple 1:1 binding of drug to channel/receptor; therefore we investigated the mechanisms by which the succinimide antiepileptics blocked the channels.

Block Is Voltage-Dependent. Current-voltage relationships (I-V curves) for 1I were measured using 500-ms step depolarizations to varying potentials from a holding potential of 90 mV. Typical current traces are shown in Fig. 3A, measured before (C) and during (M) exposure to 3 mM MPS. Average I-V curves are shown in Fig. 3B (n = 4 cells). The percent block of the peak current was calculated, and then plotted as a function of the test potential (Fig. 3C). MPS block was greatest during test potentials to 50 mV, and then gradually decreased at more positive potentials. To calculate the V_1/2 of activation, the permeability of the channels was estimated using constant field equations (Hille, 1992) and fit with a Boltzmann equation. MPS produced an apparent +7 mV shift, which was reversed upon washout (V_1/2 in mV): control, 41 ± 1; MPS, 34 ± 1; and washout, 41 ± 2. Somewhat surprisingly, the outward currents, which are due to Cs⁺ efflux through the T-type channel (Lee et al., 1999a), were hardly blocked at all. These results indicate that MPS affects the voltage dependence of gating, and that it interacts with the permeant ion.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Voltage-dependent block of MPS. A, sample records obtained at the indicated values of Vm from a holding potential of 100 mV. Current traces were obtained from an HEK-293 cell stably expressing 1I channels in the absence (C) and in the presence of 3 mM MPS (M). Charge carrier was 5 mM Ca2+. B, current-voltage relationships of 1I channels obtained before (), in the presence of 3 mM MPS (), and after washout the drug (). Data points are averages of five cells. The smooth lines are spline curves fit to the calculated values of current using the product of the Goldman-Hodgkin-Katz and Boltzmann equations (see Experimental Procedures). C, percent block of peak current as a function of test potential; from same cells shown in B. Data are not shown for voltages near the reversal potential.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   MPS accelerates decay of human T-type currents. A, superimposed current traces at 30 mV from the same cell illustrated in Fig. 3. MPS (3 mM) and wash traces were normalized to the amplitude of the control trace. Current signals plotted with dots and solid lines represent fits with two exponentials, one for the activation and the other for the inactivation of the current. B, voltage dependence of the inactivation  in the three conditions. Current traces obtained in response to an I-V protocol like in Fig. 3A were fitted as mentioned above, and the respective inactivation time constant was plotted as a function of voltage for the indicated experimental conditions. C, concentration-dependent effect of MPS on inactivation . Inactivation time constants were obtained as described in B for 1G, 1H, and 1I currents at 30 mV and were plotted versus the concentration of MPS.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Ethosuximide accelerates decay of human T-type currents. A, effect of increasing concentrations of ethosuximide on Ba2+ currents recorded from a HEK-293 cell stably transfected with 1I subunit in response to voltage steps to 20 mV. Dotted line corresponds to the washout of the drug. B, same traces as in A normalized to the control amplitude. C, dose-response relationship for ethosuximide block of Ba2+ currents measured at peak () and after 150 ms of depolarization (). Smooth lines represent the fitted curve using the Hill equation. The calculated IC50 values were 8 ± 2 and 0.6 ± 0.2 mM for the peak and the sustained current, respectively.

Block Is State-Dependent. The modulated receptor hypothesis was based on observations that drugs had varying affinities for closed/rested and inactivated states. Binding to inactivated states is an important property of drugs because it can provide selectivity to their action, as clearly exemplified by dihydropyridines that preferentially block smooth muscle L-type channels with little effect on cardiac channels (Triggle, 1999). Voltage-gated Ca²⁺ channels do not need to open before inactivating, and transitions from closed to inactivated states can be measured using long prepulses at subthreshold potentials, commonly referred to as the steady-state inactivation curve, or h. Drugs that bind and stabilize the inactivated state shift the equilibrium from closed/rested to inactivated states, and hence shift this steady-state inactivation curve to more negative voltages (Bean et al., 1983). The effect of MPS and ethosuximide on the inactivation curves of the three T-type channels was estimated using prepulses of the following duration: 0.3 s for 1G (Fig. 6A); 5 s for 1H prepulses (Fig. 6, C and D); and 30 s for 1I (Fig. 6B) and 1G (Table 1). In each case, MPS was capable of shifting the curve 5 to 10 mV to more negative potentials. The amount of shift was dose-dependent; 1 mM MPS shifted the 1G curve 5.5 mV (data not shown), whereas 3 mM shifted it 9.5 mV (Table 1). Similar results were obtained using 10 mM ethosuximide, which shifted the 1H inactivation curve 5 mV toward more negative potentials (Fig. 6D). Reversal of this effect by washing out the drug was in many cases incomplete. This is due in part to time-dependent shifts in the inactivation curve, which can be largely prevented by inclusion of ATP in the intracellular pipette solution (Zhang et al., 2000). The lack of full reversal may also be due to residual drug in the bath.

View larger version (34K): [in this window] [in a new window]   Fig. 6.   MPS and ethosuximide shift the voltage dependence of steady-state inactivation of human T-type Ca2+ channels. Steps lasting 0.3 s (1G), 5 s (1H), and 30 s (1I) to the indicated voltages were followed by 150-ms test pulses to 30 mV in A and B or 20 mV in C and D. Protocols were run before (), during, and after () exposure to 3 mM MPS () or 10 mM ethosuximide (). Charge carrier was 5 mM Ca2+ in A and B, and 10 mM Ba2+ in C and D. Currents were normalized to the value at 100 mV at each experimental condition. The smooth curves are fits using the Boltzmann equation. The corresponding parameters are shown in Table 1.

View this table: [in this window] [in a new window]   TABLE 1 Effect of MPS on inactivation The effects of MPS on the inactivation properties were determined using a double pulse protocol (see Fig. 6). The prepulse duration is shown in parenthesis. Data from each cell were collected before, in the presence of 3 mM MPS, or after washout. The results from each cell were fit with a Boltzmann equation and then averaged. All V1/2 values are in millivolts. Steady-state inactivation curves using 30-s prepulses were also performed on cells that were preincubated (>4 min) in 3 mM MPS before patching the cell and recorded in the continued presence of drug. The double pulse protocol was run 2 min after establishment of the whole cell configuration. As control for these experiments, the protocol was run on other cells a similar time after patching. The V1/2 and k values for the curves of Fig. 6D were: control, 71.7, 3.8; 10 mM ethosuximide, 76.5, 3.9; wash, 74.0, 3.8; respectively. Charge carrier was 5 mM Ca2+ in 1G and 1I experiments and 10 mM Ba2+ in the case of 1H.

View larger version (37K): [in this window] [in a new window]   Fig. 7.   Block by succinimide antiepileptic drugs is dependent on the holding potential. Dose-response relationship for MPS block of peak Ca2+ currents expressed by 1G (A), 1H (B), and 1I (C) channels. Data points were obtained by performing experiments like those described in Fig. 2 (5 mM Ca2+ as charge carrier), but using holding potentials of either 110 mV (open symbols) or 75 mV (filled symbols). Smooth lines represent the fit to the data using the Hill equation. The third set of data () and the corresponding fit in C indicate the block of the current measured after 150 ms of depolarization using a holding potential of 75 mV. The calculated IC50 for each human T-type channel is plotted against holding potential in D. The corresponding relationship for ethosuximide block of 1G is shown in E; in this case, the holding potentials were 100 mV () and 75 mV (), and the charge carrier was 10 mM Ba2+. F, estimated ethosuximide IC50 values from the data shown in E and Fig. 1.

View this table: [in this window] [in a new window]   TABLE 2 Apparent affinities of MPS for resting and inactivated channels Apparent affinities are reported in units of millimolar. KR was measured as the block of the peak current elicited during pulses to 30 mV from a holding potential of 110 mV (Fig. 7). The shift in the midpoint of the steady-state inactivation curve (h) was measured using 30-s prepulses. KI was calculated using eq. 1. Kapp was measured as the block of the peak current elicited during pulses to 30 mV from a holding potential of 75 mV (Fig. 7). The value of h was determined for each cell by dividing the baseline current obtained with a holding potential of 75 mV with that obtained previously at 110 mV. KI* was calculated using eq. 2. Data were obtained using 5 mM Ca2+ as charge carrier in the external solution except for h of 1H, which was measured in 10 mM Ba2+.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Window currents are reduced by MPS and ethosuximide. A, overlapping area of voltage-dependent activation (lines originating from 80 mV) and steady-state inactivation curves (lines ending at 40 mV) for 1I channels in the absence (solid lines) and presence (dotted lines) of 3 mM MPS. All lines represent Boltzmann function fits to experimental data, which have been omitted for clarity purposes. MPS shifted the window current region ~5 mV to more negative potentials and reduced the size of the region ~50%. B, effect of 0.7 mM ethosuximide on Ca2+ current carried through human 1G channels obtained in response to a mock subthreshold LTS.

View larger version (27K): [in this window] [in a new window]   Fig. 9.   Block of human T-type Ca2+ channels by succinimide nonantiepileptic drugs is much less potent. A, time course of peak current during exposure to trimethylsuccinimide and succinimide recorded from a cell expressing human 1H channels. The insets show representative Ca2+ current traces at 30 mV obtained under the indicated experimental conditions. The holding potential was 90 mV, and the charge carrier was 5 mM Ca2+. B, fraction of Ca2+ current blocked at 30 mV by succinimide (10 mM), TMS (6 mM), and MPS (6 mM) for the three human T-type Ca2+ channels.

Analysis of Drug Block. Both ethosuximide and MPS produced measurable shifts in the h curve of all three cloned T-type channels. The shift in the mid-point of inactivation (h) can be used to calculate the apparent affinity of the drug for the inactivated state, K_I (Bean et al., 1983). Rearrangement of this equation yields
where D represents the drug concentration, and k represents the slope of the h curve estimated from Boltzmann fits to the control data (Fig. 6). As shown in Table 2, the apparent affinity of MPS for inactivated channels is 5- to 6-fold greater than closed/rested channels. Similar results were obtained using the h induced by either 1 mM ethosuximide or MPS, as predicted by the model (Bean et al., 1983).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

This study describes the block of the cloned human T-type channels 1G, 1H, and 1I by succinimide antiepileptic drugs. Previous studies have used currents from rat or cat neurons (Coulter et al., 1989a, 1990a; Leresche et al., 1998; Todorovic and Lingle, 1998). The amino acid sequences of T-type channels are not identical between rats and humans. For example, the human 1G sequence is only 93% identical to the rat (Cribbs et al., 2000). More importantly, the pharmacology of human 1G also differs from the rat, being 2-fold more sensitive to block by nickel (Lee et al., 1999b; Cribbs et al., 2000). Considering the recent challenge to the hypothesis that the mechanism of action of ethosuximide involves blockade of T-type channels (Leresche et al., 1998), we decided to clone the human T-type channels and to test their sensitivity to these drugs. The main conclusions of this study are that block is observed at clinically relevant concentrations and that block is state-dependent.

The present results indicate that the apparent potency of ethosuximide and MPS is highly dependent on the recording conditions. Block is greater at threshold potentials, if measured at the end of a depolarizing pulse, and if the holding potential is more positive than 90 mV. T-type channels do not need to open to be blocked since there is block of peak currents in the first pulse after drug treatment in the absence of depolarizing pulses (data not shown). This result indicates that there is substantial block of channels in deep rested states. The affinities of the drugs for the closed/rested state were estimated by measuring the dose-response relationship at a holding potential of 110 mV (K_R, Fig. 7).

Open Channel Block. T-type channels, especially those formed by 1I (Fig. 4), do not inactivate completely during a long depolarizing pulse, leading to measurable persistent current, or plateau. This property is difficult to measure with native channels since plateau currents are often contaminated with currents through high voltage-activated Ca²⁺ channels (Sayer et al., 1993). Block of the T-type channel increased during the pulse, such that the plateau current was eliminated at doses that had little effect on the peak current. This had a dramatic effect on the apparent potency of ethosuximide, decreasing the apparent IC₅₀ 13-fold to 0.6 mM (Fig. 5). However, at the pulse frequency used (0.1 Hz) there was little evidence of use dependence, indicating that channels recovered from this additional block when the holding potential was 90 mV or more negative. Two possible explanations for the effect on kinetics are 1) antiepileptic drugs accelerate the transition from open to inactivated states and/or 2) that there is additional block of open states.

Preferential Binding to Inactivated States. Two experiments indicated that succinimide antiepileptic drugs preferentially blocked inactivated states of the T-type channel: 1) the drugs shifted the steady-state inactivation curve, and 2) potency was enhanced by depolarizing shifts in the holding potential. Analysis of these experiments allowed independent calculations of the drug affinity for inactivated states (Table 2). For most channel subtypes these two methods produced similar estimates for K_I. The selectivity for inactivated states over rested states of the channel was modest, ranging from 3- to 9-fold. Changing the holding potential also increases the potency of ethosuximide to block a mouse 1G (Lacinova et al., 2000).

Ethosuximide Sensitivity of T-Type Channels. Studies on the dose response of human 1G indicated that therapeutically relevant concentrations of ethosuximide were capable of producing modest block (~15%). Since a small degree of run-down could complicate the measurement of such a small effect, the experiment was repeated using a variety of set-ups, external solutions (10 mM Ba²⁺ or 5 mM Ca²⁺), and flow rates. In each case, there was evidence for a plateau in the response to concentrations ranging from 0.1 to 1 mM. The original observation that ethosuximide could inhibit T-type channels in isolated rat thalamic neurons also had similar results; block was partial (32%) and appeared to plateau at 2 mM (Coulter et al., 1989a). Higher concentrations were not reported in that study. One study using isolated rat dorsal root ganglion (DRG) neurons also found evidence for ethosuximide block at low concentrations; block was complete and had an apparent IC₅₀ of 0.007 mM (Kostyuk et al., 1992). Subsequent studies on rat DRG neurons failed to confirm block at concentrations below 1 mM, but did find block at higher concentrations (Todorovic and Lingle, 1998). The apparent IC₅₀ was 23 mM. Recent studies on isolated and slice preparations of rat thalamic neurons found no effect of 0.75 mM ethosuximide on the T-type current (Leresche et al., 1998). Although there are many differences in the recording conditions used in these studies, one notable difference is the holding potential. Since studies that reported no effect used very negative holding potentials (110 mV), we suggest these experiments tested block of rested channels, which are less sensitive to block. Preferential block of inactivated states can produce biphasic dose-response curves (Cai et al., 1997). Therefore, the partial block originally observed in thalamic neurons (Coulter et al., 1989a) may have corresponded to block of inactivated channels.

Sensitivity of T-Type Channels to MPS. In contrast to the discrepant findings with ethosuximide, there is general agreement that MPS is a potent blocker of rat T-type currents (Coulter et al., 1990a; Huguenard and Prince, 1992; Todorovic and Lingle, 1998). The present results indicate that MPS is also a potent blocker of all three cloned human T-type channels. Block was complete and had apparent IC₅₀ values between 2 and 3 mM. Similar results were obtained previously with the rat 1G and human 1H (Todorovic et al., 2000). Block of thalamic T-type currents was also complete and displayed a similar IC₅₀ of 1.1 mM (Coulter et al., 1990a). Three millimolar MPS blocked 45% of the T-type currents recorded from thalamic neurons isolated from either the ventrobasal or the reticular nuclei (Huguenard and Prince, 1992). In situ hybridization studies indicated that 1G is the predominant isoform expressed in the rat ventrobasal nucleus, whereas 1H and 1I were expressed in the reticular nucleus (Talley et al., 1999). Therefore, MPS blocks both native and cloned channels with similar potency. In contrast, MPS block of DRG T-type currents was only partial (~30%), but occurred at lower doses (IC₅₀ ~ 0.18 mM; Todorovic and Lingle, 1998; Todorovic et al., 2000). It is unclear why MPS produces only partial block of DRG currents, but complete block of currents from either thalamic neurons or cloned channels.

Structure-Activity Relationships of Succinimides. Tetramethylsuccinimide has been reported to induce seizures in mice (Klunk et al., 1982). It was found to have no effect on either thalamic T-type currents (1 mM) or their block by MPS (Coulter et al., 1990a). This result rules out a single binding site where compounds have agonist or antagonist properties (Klunk et al., 1982). Evidence indicates that this second site is on -aminobutyric acid receptors (Coulter et al., 1990b). We were unable to test its effect because it is no longer commercially available. In contrast to ethosuximide, the trimethylsuccinimide analog (,-dimethyl--methylsuccinimide) has been shown to be ineffective at reducing spontaneous epileptiform discharges in rat thalamocortical slices (1 mM; Zhang et al., 1996). The present study shows that trimethylsuccinimide is similarly ineffective at blocking T-type currents. The base molecule, succinimide, has been reported to be devoid of anticonvulsant activity (Ferrendelli and Kupferberg, 1980). It is also ineffective at blocking rat thalamic T-type currents (0.5 mM; Coulter et al., 1989b), and at blocking epileptiform discharges in thalamocortical slices (1 mM; Zhang et al., 1996). The present results indicate it was also ineffective in blocking the cloned human T-type channels. These results confirm the correlation that only succinimides that are capable of blocking T-type channels are effective anticonvulsants.

Therapeutic Implications of State-Dependent Block. T-type Ca²⁺ channels mediate low threshold Ca²⁺ spikes in a number of neurons (Crunelli et al., 1989; Suzuki and Rogawski, 1989). Such spikes play an important pacemaker role in the initiation of burst firing of thalamic neurons, and such firing underlies the spike-wave activity observed in the EEG during absence seizures (McCormick and Bal, 1997; Huguenard, 1999). Similar synchronized oscillations occur during slow wave sleep. However, they are not observed during rapid eye movement sleep or in wakeful states. In these states, the thalamus fires single Na⁺-dependent spikes (tonic mode), because the membrane potential of the neuron is depolarized to such an extent that T-type channels are inactivated (Steriade and Llinas, 1988). Therefore, clinically relevant targets for antiepileptic drugs include T-type channels in inactivated states, and blocking these states may prevent the transition from tonic to burst firing. The present results show that both ethosuximide and MPS blocks are enhanced by holding the membrane at a potential that is similar to the resting potential of many neurons. We also found that block was greatest at threshold potentials. These two effects combine such that therapeutically relevant concentrations can cause significant block of current elicited with mock LTS waveforms (Fig. 8).