Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Nearly 2 decades ago, T-type calcium channels were discovered and described in neurons (Carbone and Lux, 1984) and heart cells (Bean, 1985; Nilius et al., 1985). These low-voltage-activated calcium channels open upon small depolarizations of the membrane potential, inactivate within milliseconds, and have a low single-channel conductance for Ba²⁺, compared with the high-voltage-activated calcium channels (Bean, 1985; Nilius et al., 1985; Droogmans and Nilius, 1989). T-type calcium channels are found in a variety of tissues in which their role is only partially understood (Vassort and Alvarez, 1994). T-type channels may differ considerably in their pharmacological properties depending on tissue or cell type (Huguenard, 1996).

The molecular composition of the T-type calcium channel was elusive for many years, whereas in the case of high-voltage-activated channels, seven different genes encoding the pore-forming 1 subunit of the channel protein have been defined (1S and 1A-F). Within the last 3 years, three new 1 subunits were cloned and named the 1G, 1H, and 1I subunits. This corresponds to Ca_v3.1, Ca_v3.2, and Ca_v3.3, respectively, according to the recent nomenclature (Ertel et al., 2000). Extensive whole-cell studies of these recombinant subunits expressed in Xenopus laevis oocytes or human embryonic kidney cells reveal typical biophysical characteristics of native T-type calcium channels (Cribbs et al., 1998; Perez-Reyes et al., 1998; Lee et al., 1999), but at the single-channel level, only their conductance has been investigated. Another issue is whether any of the known auxiliary subunits associate with and modulate Ca_v3.X channels. Previous studies did not resolve this question (Dolphin et al., 1999; Lacinova et al., 1999; Klugbauer et al., 2000).

In this study, we explored for the first time single-channel pharmacology of human native T-type and recombinant Ca_v3.2 channels. Our rationale was that single-molecule biophysics and their modulation by drugs may give deeper insight into channel behavior and possibly its modulation by as-yet-unknown auxiliary subunits. For the native channels, we used a human cell line [human medullary thyroid carcinoma (hMTC)] because it expresses pure T-type calcium currents at the whole-cell level (Biagi et al., 1992), and because a Ca_v3.2 (1H) channel was cloned from this source (Williams et al., 1999). At least some auxiliary subunits (Hobom et al., 2000) are also expressed in hMTC cells. As a recombinant system, human Ca_v3.2 stably overexpressed (Cribbs et al., 1998) in human embryonic kidney (HEK) 293 cells was used in comparison. The main pharmacological tool employed was mibefradil, a known selective nondihydropyridine blocker of native (Mishra and Hermsmeyer, 1994; see also Bezprozvanny and Tsien, 1995) and recombinant (Cribbs et al., 1998) T-type channels. Mibefradil was also of interest because it differentiates between coexpressed auxiliary -subunits when interacting with another voltage-dependent calcium channel, Ca_v1.2 (Welling et al., 1995).

To our initial surprise, a high-voltage-activated calcium channel is also observed in hMTC cells at the single-channel level. We provide evidence that those two channels can easily be discriminated from each other and represent T- and L-type channels, respectively. The single-channel pharmacology of the T-type is extensively analyzed and compared with its recombinant counterpart. We demonstrate that mibefradil blocks both native T-type and recombinant Ca_v3.2 channels by two mechanisms (i.e., a reduction of the fraction of active sweeps and a slight reduction of open times).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture. hMTCcells were grown in RPMI 1640 medium (Biochrom KG, Berlin, Germany) supplemented with 10% fetal bovine serum (Sigma, Deisenhofen, Germany), penicillin (50 units/ml), and streptomycin (50 µg/ml). Cells were plated onto polystyrene dishes. HEK 293 cells have been stable transfected with the human Ca_v3.2 without auxiliary subunits as described previously (Cribbs et al., 1998; Lee et al., 1999). Cells were grown in Dulbecco's modified Eagle's medium (Biochrom KG) supplemented with fetal bovine serum (Sigma) and G418 (1 mg/ml; Invitrogen, Karlsruhe, Germany). Cells were grown onto polystyrene dishes coated with 0.1% gelatin, which also served as recording chambers. For electrophysiological experiments, cells were used on the second day (hMTC) or on the first to third day (HEK 293) after plating.

Electrophysiology. Single calcium channels were recorded in the cell-attached configuration of the patch-clamp technique. Experiments were performed in an external solution containing 120 mM K-glutamate, 25 mM KCl, 2 mM MgCl₂, 10 mM HEPES, 2 mM EGTA, 1 mM CaCl₂, 1 mM Na-ATP, and 10 mM dextrose, pH 7.4. Pipettes (borosilicate glass, 6-7 M) were filled with 110 mM BaCl₂ and 10 mM HEPES, pH 7.4. Ba²⁺ currents were elicited by depolarizing test pulses delivered at 0.5 Hz, recorded at 10 kHz and filtered at 2 kHz (3 dB, four-pole Bessel) using an Axopatch 1D or Axopatch 200 A (Axon Instruments, Union City, CA). The pClamp software (versions 5.5 and 6.0, Axon Instruments) was used for data acquisition and analysis. Drugs were added to the bath as a 20-µl bolus of an appropriate stock solution, assuming a bath volume of 2 ml. At the end of the experiments, the actual bath volume was measured, and the exact drug concentration was calculated. All experiments were carried out at room temperature (21-23°C).

Drugs. Mibefradil was a gift of Hoffmann-La Roche (Basel, Switzerland). It was prepared as a 10 mM stock solution in H₂O, which was freshly diluted with bath solution to 0.1 mM on each experimental day. (S)-Bay K 8644 (Sigma/RBI, Natick, MA) and nitrendipine (Sigma) were dissolved and stored light-protected in absolute ethanol as 10 mM stock solutions and diluted to 0.1 mM for daily use. The final ethanol concentration in the bath was  0.11%.

Data Analysis. Linear leak and capacity currents were digitally subtracted using the average currents of nonactive sweeps. Openings and closures were identified using the half-height criterion (pClamp 6.0). For analysis of data obtained with Ca_v3.2 we also used custom-made software (Patch 0.2) that employs the same idealization algorithm but allows for improved leak subtraction and data-editing features. Data were further analyzed by histogram analysis (see below). Closed time analysis was restricted to patches where only one channel was present. NP_o (the product of the number of channels in the patch times their individual open probability) was calculated as the ratio between the total open time and the total recording time at the test potential. P_o (the single channel open probability) was calculated as the ratio of the total open time and the total time recorded at the test potential, divided by the number of channels in the patch (N) where necessary and possible. f_active (the fraction of sweeps containing at least one opening) was corrected by the square-root method: 1  f_{active, corrected} = (1  f_{active, uncorrected})^1/N for multichannel patches in hMTC cells. The number of channels in the patch was estimated here by dividing the maximum current observed with stacked openings through the unitary amplitude. P_{o, active} (the open probability within active sweeps) was computed as the ratio P_o/f_active. The voltage-dependence of f_active and P_o was analyzed using the Boltzmann function: Y = Y_max/{1 + e ^{(V_0.5  V/k)}}, where V_0.5 is the voltage of half-maximal (in)activation and k is the slope factor. Single-channel amplitudes (i) were determined by direct measurements of fully resolved openings or as the maximum of Gaussian fits to all-point amplitude histograms of "semi"-idealized openings [semi-idealized traces: raw data during openings but idealized (zeroed) baseline]. I_peak was determined as the maximum of ensemble average currents. The time course of inactivation (i) was analyzed by fitting nonlinear approximation the ensemble average current I to the equation I_(t) = I_max. × (1  e^t/a)² × e^t/i (Droogmans and Nilius, 1989; Huguenard, 1996). A two-tailed t test was used for statistical examinations, using unpaired format for comparison of different channels and paired format for testing drug effects. A value of p < 0.05 was considered significant. Data are given as means ± S.E.M. Open-time constants (_open) were obtained from single experiments using simple exponential fits of open-time histograms by maximum likelihood method. Single  values were averaged as pooled mean according to the equation _{open, pooled} = [((_e)·(1/SD_e²))/((1/SD_e²))], where SD_e is the individual standard deviation. SD_e was estimated by SD_e = _e/n^1/2, where _e is the estimated open-time constant and n means the number of events used for estimation of _e. Standard deviation of the pooled mean (_{open, pooled}) was then calculated from SD_pooled² = 1/((1/SD_e²)). Average means of open-time constants were compared using the unpaired two-tailed t test.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

In cell-attached recordings, we obtained two visually distinct types of single-channel activity, which we initially termed "type I" and "type II" behavior. Representative examples from two experiments are compiled in Fig. 1. At first glance, types I and II differ in at least four different aspects. For type I, it activates at lower voltages, it has a lower single-channel amplitude at any given test potential, openings occur in well-separated bursts at any given test potential, and the ensemble average current inactivates rapidly. For type II, openings are more evenly distributed, and average currents do not visibly undergo time-dependent inactivation. We will systematically compare these properties one by the other in the following.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Original traces (A) and ensemble average currents (B) from two experiments, illustrating the differences between type I and type II channel behavior in hMTC cells. Patches were depolarized from 90 mV to various test potentials (TP), as indicated in the figure. In A, two traces are shown for each indicated TP. Scale bars indicate 15 ms and 0.5 pA, respectively. In B, average currents from 60-sweep ensembles are displayed for various TP. Scale bars indicate 15 ms and 50 fA in this case.

The voltage-dependences of activation and inactivation were analyzed by plotting f_active (measured from at least 60 test pulses) against the voltage of the test potential or holding potential, respectively. The pooled data are shown in Fig. 2, A and B. Individual type I channels activate half-maximally at 17.7 ± 2.9 mV, with a slope factor of 4.3 ± 1.0 mV (n = 4). Half-maximal inactivation occurs at 51.1 ± 3.5 mV with a slope of 7.6 ± 1.4 mV (n = 7). A small window current is suggested by the overlap of these curves between 40 mV and 20 mV (Cribbs et al., 1998). For type II, inactivation lies in the same order of magnitude (V_0.5, 48.8 ± 2.3 mV; slope, 7.9 ± 0.7 mV; n = 3), but activation occurs at much more positive potentials (V_0.5, 8.0 ± 0.9 mV; slope, 7.5 ± 2.0 mV; n = 4, p < 0.05). Maximum f_active was always  50%. When analyzing P_o in a similar manner (Fig. 2, C and D), the maximum values of activation curves were more variable, and saturation was not clearly evident at the most positive test potentials where openings still could be resolved (Cachelin et al., 1983). Yet, potentials of half-maximum activation and inactivation gave results consistent with f_active for both type I (activation: V_0.5, 13.9 ± 4.7 mV; slope, 4.9 ± 0.6 mV; n = 6; inactivation: V_0.5, 66.3 ± 4.6 mV; slope, 12.5 ± 3.1 mV; n = 5) and type II (activation: V_0.5, 17.9 ± 5.2 mV; slope, 7.2 ± 0.7 mV; n = 5; inactivation: V_0.5, 45.8 ± 4.9 mV; slope, 6.9 ± 2.5 mV; n = 4) channels, respectively.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Voltage-dependence of single-channel activity of type I (A and C) and type II (B and D) channels in hMTC cells. Patches were depolarized every 2 s from 90 mV to various test potentials (activation), or from various holding potentials (inactivation) to 10 mV (A and C) or +10 mV (B and D). Data were averaged from n = 10 (A, activation), 8 (A, inactivation), 10 (B, activation), 7 (B, inactivation), 10 (C, activation), 9 (C, inactivation), 9 (D, activation), and 7 (D, inactivation) experiments. The curves shown here were fitted using the Boltzmann equation and the pooled data as displayed. Mean values as indicated in the text were derived from Boltzmann fits of individual experiments.

The difference between type I and type II regarding single-channel conductance is illustrated in Fig. 3. Type I channels (closed symbols) have a lower single-channel current amplitude at any given test potential, and the slope conductance, calculated individually by linear regression, is significantly lower, too (type I, 7.2 ± 0.6 pS, n = 10; type II, 14 ± 1 pS, n = 10).

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Single-channel amplitude i of type I (closed symbols, n = 10) and type II (open symbols, n = 10) channels as a function of test potential. Holding potential was 90 mV throughout. Values of i were measured using fully resolved openings. Slope conductances as depicted here were calculated based on pooled data from type I and type II channels, respectively. See text for average values determined by linear regression in each individual experiment.

The most prominent difference regarding rapid gating was the bursting pattern of openings, which was evident over the whole voltage range in the case of type I channels. Therefore, we analyzed the closed-time distribution of type I and II channels at a test potential of 0 mV, where single-channel events could still be resolved for type I channels, and channel activity was already substantial for type II (see Fig. 1). As seen in Fig. 4, both channel types clearly reveal two components in their closed-time histograms. The slow component had a comparable time constant, ₂, that amounted to 22.1 ± 4.2 ms (n = 4) for type I and 22.0 ± 2.8 ms (n = 5) for type II channels. The ₁ extrapolated from maximum likelihood fits, however, seemed shorter in type I (0.20 ± 0.01 ms) than in type II (0.60 ± 0.11 ms). Furthermore, the computed size of the fast component seemed to be more prominent in the case of type I channels. Taken together, these two features illustrate the clear-cut bursting pattern of type I activity. They translate into a markedly, significantly shorter overall mean closed time (type I, 0.79 ± 0.20 ms, n = 6; type II, 9.22 ± 1.72 ms, n = 7).

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Closed-time distribution of type I (A) and type II (B) channels, obtained at a test potential of 0 mV (holding potential 90 mV). Data (plotted according to Sigworth and Sine, 1987) were pooled from one-channel experiments (n = 4 and 5, respectively), where the amount of data allowed for individual maximum likelihood estimation of the 1 and 2 values (see text). Dotted lines represent maximum likelihood fits of the pooled data as displayed.

A rapid inactivation during the test pulse was regularly observed for type I channels but not type II channels. For a number of experiments like those shown in Fig. 1, we were able to analyze the voltage-dependence of the inactivation time course (i) by nonlinear approximation (see Materials and Methods). In the case of type I, inactivation was fast and accelerated by more positive test voltages: the i values were 62 ± 23 ms, 31 ± 11 ms, and 31 ± 2 ms (n = 3-5) at voltages of 10 mV, 0 mV, and +10 mV, respectively. In contrast, i values increased with voltage in the case of type II: 62 ± 3 ms at 0 mV, 77 ± 13 ms at +10 mV, 349 ± 126 ms at +20 mV, and 458 ± 120 ms at +30 mV (n = 3-4). Further examples illustrating the different inactivation behavior can be seen in Figs. 6 and 7.

To exclude the possibility that our analysis of type I versus type II channels is biased by a selection artifact, we checked our identification by plotting two of the proposed distinctive biophysical features against each other (Fig. 5), single-channel conductance and mean closed time (at 0 mV, from one-channel patches only). It can be seen that there is no single case of overlap between those two parameters, which are not inherently dependent on each other.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Comparison between channels classified visually as type I (, n = 6) and type II (, n = 7) using two independent biophysical criteria: conductance (abscissa, see Fig. 3) and mean closed time (ordinate, see Table 1). Only one-channel patches were included. There is no overlap between the values covered by each criterion, and there is no case of ambiguity, as indicated by the two empty quadrants.

In a second series of experiments, we tested our hypothesis that type I and II channels represent different entities by using a pharmacological approach. Channels were classified as type I or II during the early course of the experiments according to the criteria mentioned above. Type I channels were then step-depolarized to 10 mV (Fig. 6) and type II channels to +10 mV (Fig. 7), offering optimal activity and recording conditions for each case. After a control period, channels were then exposed to either solvent, the L-type calcium-channel agonist (S)-Bay K 8644, or the T-type channel blocker mibefradil. Under control conditions, type II but not type I channels underwent some decline in activity ("run down", see Figs. 8 and 9). The time course of P_o observed after drug application is depicted in Figs. 8 and 9. The most prominent drug effects were profound inhibition of type I by mibefradil (10.0 ± 0.1 µM; Fig. 8, middle), and stimulation of type II by (S)-Bay K 8644 (1.01 ± 0.01 µM; Fig. 9, bottom) because of an increase in open times and therefore in open probability within active sweeps (P_{o, active}, Table 1). (S)-Bay K 8644 (1.02 ± 0.01 µM) had no effect on type I channels. Mibefradil (9.7 ± 0.3 µM) inhibited I_peak of type II channels by 50% of predrug values. This effect was caused by a corresponding change in f_active, whereas open times remained unaffected.

View larger version (71K): [in this window] [in a new window]   Fig. 6.   Inhibition of a type I channel (control, left) by mibefradil (right). The top illustrates the pulse protocol (holding potential 90 mV, test potential 10 mV). The bottom trace depicts ensemble averages from 235 traces before drug and 174 traces after drug. The middle shows consecutive sweeps. Scale bars indicate 15 ms and 1 pA (individual traces) or 20 fA (ensemble averages), respectively.

View larger version (57K): [in this window] [in a new window]   Fig. 7.   Stimulation of a type II channel (control, left) by (S)-Bay K 8644 (right). The top illustrates the pulse protocol (holding potential 90 mV, test potential +10 mV). The bottom trace depicts ensemble averages from 117 traces before drug and 179 traces after drug. The middle shows consecutive sweeps. Scale bars indicate 15 ms and 1 pA (individual traces) or 75 fA (ensemble averages), respectively. Besides a prominent increase in open times and open probability (see Table 1), the drug enlarged single-channel amplitude (from 0.89 ± 0.04 pA to 1.13 ± 0.09 pA). Conductance was raised to 23.1 ± 1.0 pS (n = 9).

View larger version (13K): [in this window] [in a new window]   Fig. 8.   Open probability Po (averaged over 2-min recording periods) of type I channels versus time. Channels were depolarized to 10 mV and exposed after 6 min to solvent (top, n = 8), mibefradil (middle, 10.0 ± 0.1 µM, n = 13), or (S)-Bay K 8644 (bottom, 1.02 ± 0.01 µM, n = 10).

View larger version (13K): [in this window] [in a new window]   Fig. 9.   Open probability Po (averaged over 2-min recording periods) of type II channels versus time. Channels were depolarized to +10 mV and exposed after 6 min to solvent (top, n = 10), mibefradil (middle, 9.7 ± 0.3 µM, n = 7), or (S)-Bay K 8644 (bottom, 1.01 ± 0.01 µM, n = 9).

In contrast, mibefradil effects on type I channel ensemble average current are composed of two components. First, the fraction of active sweeps again is strongly depressed; in addition, open probability within active sweeps is also reduced to a significant extent, because of a corresponding change in mean open times (Table 1). To address these two components of inhibition in a more quantitative manner, further analysis was performed. The slow gating kinetics of channel "availability" (i.e., transitions between active and nonactive, or blank sweeps), can be analyzed by sweep histogram analysis (Herzig et al., 1993). Blank-sweep histograms were constructed by plotting the cumulative probability density function of series of continuously nonactive sweeps. In the seven experiments with one type I channel in the patch, blank-sweep histograms were well characterized by a single exponential component before mibefradil was applied. The time constant of the underlying nonactive state amounted to  = 2.68 ± 0.26 s. When data from the seven experiments were pooled (1611 sweeps in total), a similar  value of 2.66 s was obtained (Fig. 10 A). After application of 10 µM mibefradil, a second, slow component became obvious, ₂ = 22.1 ± 4.8 s (n = 7) together with the faster component, the ₁ (2.84 ± 0.36 s, n = 7), which matched the  value obtained under control conditions (see above). Again, results from the pooled analysis (Fig. 10B) were similar (₁ = 2.82 s; ₂ = 16.7 s). This indicates that fraction of active sweeps is reduced by mibefradil because the channel occasionally adopts a long-lived nonactive, blank state. The lifetime of this state can be taken as a measure of the rate of dissociation of the drug (1/  0.05/s).

View larger version (11K): [in this window] [in a new window]   Fig. 10.   Blank-sweep histogram from seven one-channel experiments before (A) and after (B) exposure to mibefradil. The drug induces long series of continuously blank sweeps, which constitute a second, long-lived component of the histogram in B (see text).

Open-time histograms confirm the significant effect of mibefradil seen on mean open time (Table 1). As exemplified in Fig. 11, the dwell time of the open state is adequately described by a single exponential. _open was reduced from 0.48 ms to 0.37 ms in the case depicted and from 0.50 ± 0.01 ms to 0.43 ± 0.03 ms (p < 0.05, n = 13; _{open, pooled} ± S.D._pooled) on average. A clear decrease of _open was observed in nine experiments (by 24.8 ± 3.5%). The results obtained from open-times analysis indicate that mibefradil exerts an additional effect on the short (millisecond) time scale of single-channel gating. The fact that open probability within active sweeps (P_{o, active}) and open times are reduced to a similar extent (about 20%, Table 1), together with a lack of effect on closed times, is also compatible with the idea that the drug leads to a rapid, short-lived block of channels on the order of < 1 ms.

View larger version (21K): [in this window] [in a new window]   Fig. 11.   Open-time histogram from a type I channel before (A) and after (B) exposure to mibefradil. Test potential was 10 mV. The maximum likelihood fit reveals a single exponential open-state component, the lifetime open of which is shortened (from 0.48 ms to 0.37 ms in this experiment) by the drug.

HEK 293 cells expressing Ca_v3.2 channels were first examined using a pulse protocol in which channels were fully available (holding at 90 mV, test pulses to 20 mV). This gave rise to large, rapidly activating and inactivating currents (Fig. 12 A), but clear-cut unitary events were discernible only during the second half of the test pulse. Every single tracewith some variance regarding peak amplituderesembled the current wave form obtained by averaging the whole ensemble (Fig. 12A, bottom trace). Mibefradil (10 µM; 8.9 ± 0.6 µM, n = 5) considerably reduced channel activity regarding peak currents. Open probability was markedly reduced, such that nonstacked unitary events were now also observed early in the pulse (Fig. 12 B). The ensemble averages still revealed the typical features of a rapidly activating and inactivating channel. Mibefradil lowered the maximum ensemble average current (I_peak) significantly from 567 ± 122 to 189 ± 40 fA. Hence, overexpression had led to a high channel density in every patch examined, and we made no further attempts to characterize single-channel gating (or mibefradil effects) under the conditions of this voltage protocol.

View larger version (23K): [in this window] [in a new window]   Fig. 12.   Single-channel records from an experiment with Cav3.2 expressing HEK 293 cells. The top illustrates the pulse protocol (holding potential 90 mV, test potential 20 mV). Four sequential traces are shown in the center, whereas the bottom displays the ensemble average current. Scale bars indicate 20 ms and 1 pA for single traces and 200 fA for ensemble averages, respectively. A, control before mibefradil. B, traces obtained in the presence of mibefradil (10 µM).

Instead, depolarizing steps from 50 to 20 mV were chosen to induce voltage-dependent steady-state inactivation, which dramatically minimized the number of simultaneously available channels, compared with a holding potential of 90 mV. This made it possible to detect and analyze separate unitary events in both the absence and the presence of mibefradil (Fig. 13). Stacked openings of twice the unitary amplitude were rarely observed. Channel activity remained stable over time in cells not exposed to drug (Fig. 14, top). In the presence of 10 µM mibefradil (9.3 ± 0.5 µM, n = 8), the peak inward current obtained from ensemble averages was found to decrease significantly from 46 ± 5 fA to 24 ± 4 fA. In view of the (sub)micromolar IC₅₀ values reported for mibefradil in whole-cell studies on Ca_v3.2 (Cribbs et al., 1998; Williams et al., 1999) we tested a lower concentration of 3.3 ± 0.1 µM (Table 2). This concentration caused qualitatively similar but smaller effects (I_peak, from 93 ± 24 fA to 69 ± 24 fA; NP_o, from 11 ± 3.2% to 7.0 ± 2.9%, p < 0.05, n = 6). Selectivity of the action of 10 µM mibefradil on these channels was further examined by applying the same concentration of nitrendipine (10.3 ± 0.2 µM, n = 6), a selective blocker of single L-type calcium channels (Kawashima and Ochi, 1988). As expected, Ca_v3.2 channels were not blocked significantly by this compound (Fig. 14; Table 2).

View larger version (67K): [in this window] [in a new window]   Fig. 13.   Blockade of Cav3.2 channels in HEK 293 cells. As in Fig. 12, the top indicates the pulse protocols (50 mV holding potential, 20 mV test potential). The bottom trace indicates ensemble average currents before (n = 240 sweeps) and after (n = 418 sweeps) mibefradil application (10 µM). Left, control before mibefradil; right, traces obtained in the presence of mibefradil (10 µM). Scale bars indicate 15 ms and 1 pA (individual traces) or 20 fA (ensemble average currents).

View larger version (15K): [in this window] [in a new window]   Fig. 14.   Open probability NPo (averaged over 2-min recording periods) of recombinant Cav3.2 channels versus time. Channels were held at 50 mV and depolarized to 20 mV. They were exposed either to no drug (top, n = 6), or to mibefradil after 6 min (middle, 9.3 ± 0.5 µM, n = 8), or to nitrendipine after 4 min (bottom, 10.3 ± 0.2 µM, n = 5).

The mechanism of inhibition by 10 µM mibefradil was now studied in more detail at the single-channel level. As illustrated in Fig. 13, the number of sweeps containing channel activity f_active was significantly lowered by 10 µM mibefradil from 67 ± 5 to 40 ± 5%. It should be noted that these numbers only give a lower-limit estimate for the effect on true f_active of each single channel involved, because we are unable to determine the total number of channels participating in channel activity. For similar reasons, open probability is indicated here as the conventional raw NP_o [i.e., the product of the (unknown) number of channels times their individual open probability (not corrected for true f_active, unlike the data obtained from hMTC cells)]. This parameterwhich combines drug effects on f_active and those on rapid gating during the available statewas decreased by 10 µM mibefradil from 2.5 ± 0.2 to 1.1 ± 0.2% (p < 0.05), hinting at an additional effect on rapid gating. Indeed, mean open times tended to decrease after mibefradil (from 0.73 ± 0.09 ms to 0.63 ± 0.08 ms, see Table 2). Analysis of open-time histograms (Fig. 15) confirmed this tendency: 10 µM mibefradil decreased open-time constants (_open) from 0.59 ms to 0.45 ms in the case depicted and from 0.45 ± 0.03 ms to 0.39 ± 0.02 ms (p < 0.05, n = 8, _{open, pooled} ± S.D._pooled) on average. Note that in five experiments, _open dropped clearly (by 24.2 ± 7.5%). We were concerned that the drug effect on open times could be artificial, because of fusion of sequential openings of distinct channels (by bandwidth-induced absences of very short closures or very short stacked events). This problem, however, should be less severe during the second half of the test pulses, where openings are very rare and fusion rather unlikely. Indeed, when studying the second half of test pulses, the mibefradil effect on mean open times was well preserved (from 0.76 ± 0.10 ms to 0.64 ± 0.09 ms, p = 0.06, n = 8).

View larger version (21K): [in this window] [in a new window]   Fig. 15.   Open-time histograms from a representative experiment obtained from a Cav3.2 expressing HEK 293 cell before (A) and after (B) application of 10 µM mibefradil (50 mV holding potential). open obtained from this experiment by maximum likelihood fits decreased from 0.58 ms before to 0.45 ms after drug.

To examine the possibility of an additional, very rapid block in the range of microseconds, we constructed all-point histograms (Fig. 16) from raw data after semi-idealization of the data sets (see Materials and Methods). Single-channel amplitude, as determined by Gaussian fits to the nontruncated portion of histograms, remained unaffected by 10 µM mibefradil both in native (Fig. 16A, control: 0.40 ± 0.02 pA; B, after mibefradil: 0.41 ± 0.03 pA, n = 13, n.s.) and in recombinant channels (Fig. 16C, control: 0.41 ± 0.01 pA; D, after mibefradil: 0.42 ± 0.02 pA, n = 8, n.s.).

View larger version (23K): [in this window] [in a new window]   Fig. 16.   All-point histograms from two representative experiments obtained from an hMTC cell (A and B) and a Cav3.2 expressing HEK 293 cell (C and D) before (A and C) and after (B and D) application of 10 µM mibefradil. Histograms were constructed from raw data after removing data below the idealization threshold (semi-idealization, see Materials and Methods). This inevitably leads to truncation of events below or near the half-height criterion. Mean values obtained by Gaussian fits amounted to 0.43 pA and 0.41 pA (A and B, hMTC) and to 0.41 pA and 0.45 pA (C and D, HEK 293) before and after mibefradil, respectively.

In summary, mibefradil exerted a dual mechanism of block, affecting f_active and open times, in Ca_v3.2 channels. The "pharmacological phenotype" of the recombinant channelsas far as it can be resolved by analysis of multichannel patchesis therefore indistinguishable from its native counterpart expressed in hMTC cells.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The main results of this study are that two types of single calcium channels are present in a human C-cell tumor line (hMTC), that these channels can easily be distinguished by their biophysical properties, that these channels fulfill the necessary criteria to be classified as T-type or L-type channels respectively, that mibefradil blocks T-type channels by reducing fraction of active sweeps and open times, but not single-channel amplitude, and that mibefradil effects are identical when studied in recombinant Ca_v3.2 channels.

The large-conductance channel (type II) inactivates slowly (Fig. 1B) and activates at high voltages (Fig. 2B). The voltage-dependence of activation and inactivation, measured under continuous pulsing rather than with conventional two-pulse protocols, is reminiscent of native cardiac human L-type channels (Handrock et al., 1998; Kreuzberg et al., 2000). The single-channel conductance under basal conditions (14 pS) does not discriminate between various members of the high-voltage-activated channel family, but conductance can be more accurately determined using the well-resolved openings in the presence of the calcium-channel agonist (S)-Bay K 8644 (Fig. 7). The value obtained here (23 pS) is in line with reports on L-type channels (McDonald et al., 1994) under similar conditions, and such an increase found with calcium agonists has been noted previously (Lacerda and Brown, 1989; Handrock et al., 1998; Kreuzberg et al., 2000). Importantly, a stimulatory response to calcium agonists per se (Figs. 7 and 9) is a hallmark feature of L-type (Hess et al., 1984) but not other types of voltage-activated calcium channels. In characteristic distinction (Nilius et al., 1985; McDonald et al., 1994) to the low-voltage-activated channel we observed, type II channels exhibited time-dependent loss of activity [i.e., "run down" (Figures 8 and 9)]. On the basis of our results, the molecular entity (1C, 1D, 1F, or 1S, corresponding to Ca_v1.2, Ca_v1.3, Ca_v1.4, and Ca_v1.1, respectively) that represents the pore-forming unit cannot be determined, although the source of cells (endocrine) favors the first two possibilities (Ertel et al., 2000). Furthermore, it remains unclear why previous investigators did not detect an L-type current at the whole-cell level (Biagi et al., 1992; but see Mehrke et al., 1994), despite the observation that hormone secretion in primary human MTC-cells is stimulated and inhibited by dihydropyridine agonist and antagonist, respectively (Raue et al., 1989). We observed marked periodic changes in the occurrence of type II, but not type I, channels over the course of the project, which were independent of passage number or batch of cells (not shown). The inhibition of the L-type channel by mibefradil, known from other systems (e.g., Bezprozvanny and Tsien, 1995; Welling et al., 1995) but seen here for the first time with single channels, merits further analysis once the molecular channel entity is identified (see above).

The type I behavior reported here fulfills all tested criteria of a T-type, Ca_v3.X channel, namely activation in the low-voltage range (Figs. 1-3) and low conductance of 7 to 8 pS (Fig. 3) with isotonic Ba²⁺ (Nilius et al., 1985; note that the lower value of 5.3 pA reported by Cribbs et al., 1998, was obtained at much more negative potentials), fast inactivation (Fig. 1B), and sensitivity toward mibefradil (Cribbs et al., 1998). Deactivation was not studied here, as the long pulses used (150 ms) gave rise to only a few tail openings, which were not further analyzed. The clear-cut two components of the closed-time distribution, with an ultrashort first component, have also been described previously as characteristic for native T-type channels (Droogmans and Nilius, 1989; Chen and Hess, 1990). Based on cloning work (Williams et al., 1999), we suggest that type I channels described here represents the human 1H (or Ca_v3.2) isoform in its native environment. Notably, the effect of mibefradil on this channel comprises two components: a major reduction of fraction of active sweeps due to a long-lived interaction and a minor component reflected by a shortening of open times.

This study is the first to examine the single-channel mechanism of mibefradil to block a recombinant human low-voltage-activated calcium channel, Ca_v3.2. Unlike the situation with native channels in hMTC cells, the high level of overexpression of channels in HEK 293 cells made it inevitable to perform multichannel recordings. Yet, partial depolarization of the holding potential enabled us to perform a focused analysis of single-channel gating before and after drug application.

Interestingly, mibefradil-induced inhibition again seemed to consist of two components. Blockade that leads to a reduced fraction of active sweeps is indicative for a long-lived drug-channel interaction outlasting the duration of the test pulse (here, 150 ms). This type of effect is known for classical calcium-channel blockade in Ca_v1.X, as with gallopamil (Pelzer et al., 1984) or nitrendipine (Hess et al., 1984), and can be analyzed kinetically (Kawashima and Ochi, 1988). In our case, only single-channel patches from hMTC cells allowed such an analysis, revealing a dissociation rate on the order of 0.05/s in the presence of mibefradil (see above). In addition, and in clear kinetic distinction, open times were shortened, which points to another type of interaction on the order of milliseconds and implies open-channel blockade. Of note, Gomora et al. (2000) measured a mibefradil effect of similar size on whole-cell deactivation kinetics, although these authors did not highlight this result. Open-channel block is known for other calcium-channel blockers, at least at high concentrations (Pelzer et al., 1984; Kawashima and Ochi, 1988). Very rapid block in the range of microsecondswhich would manifest as a reduced apparent single-channel current amplitude at our recording bandwidth (Hille, 1992)was absent both in native and recombinant channels (Fig. 16). Such type of block is more common for small pore-blocking particles but has been observed with larger organic molecules (Gingrich et al., 1993).

Largely different time scales of channel blockas seen hereare commonly assigned to different types of channel blockers, each with particular dwell times of binding (Hille, 1992). Because we observed different kinetics within one drug, we may speculate about factors affecting the mechanism of block. The chemical structure of mibefradil contains a benzimidazolyl group and a tertiary amino group. Thus, resulting charge heterogeneity may cause differences in channel interaction (Kass and Arena, 1989; Abernethy, 1997).

We were quite concerned about the high concentrations of mibefradil necessary for an approximately half-maximum block. Given the (sub)micromolar inhibition constant reported for Ca_v3.2 under whole-cell conditions (Cribbs et al., 1998; Martin et al., 2000) we provide several lines of evidence that the  10-fold lower affinity of block is still indicative of a specific drug-channel interaction: a dihydropyridine agonist [(S)-Bay K 8644] and an antagonist (nitrendipine 10 µM) have no effect in our systems, and mibefradil at a 3-fold lower concentration exerted only minor effects, indicating that we are working in the appropriate range of concentrations. These findings indicate that the low affinity of mibefradil is caused by the particular experimental conditions necessary for single-channel recording. We cannot decide at present whether accessibility limitations imposed by the cell-attached configuration, the high concentration of divalent cations, or both are responsible. The latter mechanism is supported by whole-cell studies (Martin et al., 2000). The former idea is less likely, given the ease of cellular accumulation of mibefradil (Wu et al., 2000), but can be rigorously tested using the inside-out or outside-out excised patch conformation.

HEK 293 cells used in this work overexpress pore-forming Ca_v3.2 channel subunits (Cribbs et al., 1998). The existence of endogenous auxiliary subunits cannot be firmly excluded, because native HEK 293 can express endogenous small low-voltage-activated currents under some conditions (Berjukow et al., 1996). However, we consider it unlikely that those subunits are quantitatively sufficient to modulate transfected Ca_v3.2. Therefore, we conclude that the natural set of accessory subunits possibly coexpressed (in stoichiometric amounts) with native T-type calcium channels in hMTC cells plays no detectable role with regard to single-channel activity under baseline conditions or to mibefradil block. However, a thorough biophysical analysis of recombinant channels under baseline conditions was hampered by technical limitations of the expression system: holding potentials negative to 50 mV inevitably led to multiple overlapping channel openings (Fig. 12), and test potentials positive to 20 mV could not be analyzed because of inadequate signal-to-noise ratios and insufficient seal stability.

This article presents the first description of mibefradil blockade of native and recombinant Ca_v3.X channels at the single-channel level. We found that mibefradil caused a profound blockade of T-type- and Ca_v3.2-mediated calcium channel currents, which consisted of two kinetically distinct components. Our results show that recombinant (Ca_v3.2) and native T-type calcium channels share a common pharmacological phenotype as probed by mibefradil. Hence, we conclude that hMTC cells express both T-type and L-type calcium channels, hMTC cells and HEK 293 cells overexpressing the Ca_v3.2 subunit are useful tools for single-channel studies on low-voltage-activated calcium channels, offering distinct advantages and limitations, and mibefradil block in both systems reveals two kinetically distinct components at the single-channel level, suggesting two distinct mechanisms of block.