Abstract
Benzodiazepines (BDZs) depress neuronal excitability via positive allosteric modulation of inhibitory GABAA receptors (GABAAR). BDZs and other positive GABAAR modulators, including barbiturates, ethanol, and neurosteroids, can also inhibit L-type voltage-gated calcium channels (L-VGCCs), which could contribute to reduced neuronal excitability. Because neuronal L-VGCC function is up-regulated after long-term GABAAR modulator exposure, an interaction with L-VGCCs may also play a role in physical dependence. The current studies assessed the effects of BDZs (diazepam, flurazepam, and desalkylflurazepam), allopregnanolone, pentobarbital, and ethanol on whole-cell Ba2+ currents through recombinant neuronal Cav1.2 and Cav1.3 L-VGCCs expressed with β3 and α2δ-1 in HEK293T cells. Allopregnanolone was the most potent inhibitor (IC50, ∼10 μM), followed by BDZs (IC50, ∼50 μM), pentobarbital (IC50, 0.3–1 mM), and ethanol (IC50, ∼300 mM). Cav1.3 channels were less sensitive to pentobarbital inhibition than Cav1.2 channels, similar to dihydropyridine (DHP) L-VGCC antagonists. All GABAAR modulators induced a negative shift in the steady-state inactivation curve of Cav1.3 channels, but only BDZs and pentobarbital induced a negative shift in Cav1.2 channel inactivation. Mutation of the high-affinity DHP binding site (T1039Y and Q1043M) in Cav1.2 channels reduced pentobarbital potency. Despite the structural similarity between benzothiazepines and BDZs, mutation of an amino acid important for diltiazem potency (I1150A) did not affect diazepam potency. Although L-VGCC inhibition by BDZs occurred at concentrations that are possibly too high to be clinically relevant and is not likely to play a role in the up-regulation of L-VGCCs during long-term treatment, pentobarbital and ethanol inhibited L-VGCCs at clinically relevant concentrations.
Introduction
Benzodiazepines (BDZs) are clinically useful anxiolytics, sedatives, and anticonvulsants. The clinical effects of BDZs are attributed to their ability to allosterically enhance the inhibitory action of GABAA receptors (GABAAR). However, there is also evidence that BDZs directly inhibit L-type voltage-gated calcium channels (L-VGCCs) in muscle and nerve cells at low micromolar concentrations (Yamakage et al., 1999; Xiang et al., 2008). Other GABAAR modulators, such as the neurosteroid allopregnanolone (3α-hydroxy-5α-pregnan-20-one), the barbiturate pentobarbital, and ethanol, can also inhibit neuronal L-VGCCs (Messing et al., 1986; ffrench-Mullen et al., 1993; Hu et al., 2007).
The interaction of GABAAR modulators with L-VGCCs could play a role in their short-term effects to reduce neuronal excitability and may also contribute to the neuronal hyperexcitability seen after withdrawal from long-term drug exposure. It is noteworthy that long-term treatment with BDZs, barbiturates, and ethanol increases neuronal Ca2+ influx in vitro and in vivo, possibly through L-VGCCs (Messing et al., 1986; Rabbani and Little, 1999; Katsura et al., 2006, 2007; Xiang et al., 2008). An up-regulation of L-VGCC function during long-term treatment with the various GABAAR modulators is implicated in contributing to withdrawal hyperexcitability (Dolin et al., 1987; Rabbani and Little, 1999; Xiang and Tietz, 2007; Xiang et al., 2008).
Although it is known that GABAAR modulators inhibit L-VGCCs, there have been no studies to assess the selective effects of these drugs on the neuronal L-VGCC α1 subtypes Cav1.2 and Cav1.3. By use of the whole-cell voltage-clamp technique, the current studies investigated the effects of BDZs (diazepam, flurazepam, and desalkylflurazepam), allopregnanolone, pentobarbital, and ethanol on recombinant L-VGCCs containing the neuronal L-type α1 subunit, Cav1.2 or Cav1.3, along with β3 and α2δ-1 subunits. We found that positive allosteric GABAAR modulators reversibly inhibited both Cav1.2- and Cav1.3-containing L-VGCCs in a state-dependent manner. The order of potency was allopregnanolone > BDZs > pentobarbital > ethanol, with BDZs displaying 2- to 6-fold less potency than the benzothiazepine (BTZ) diltiazem. Previous studies reported that Cav1.3 channels were less sensitive to inhibition by dihydropyridines (DHPs), phenylalkylamines (PAAs), and BTZs than Cav1.2 channels (Koschak et al., 2001; Xu and Lipscombe, 2001; Tarabova et al., 2007). The structural similarities among BDZs and BTZs (see Fig. 1) suggested the possibility that L-VGCCs might also display differential sensitivity to GABAAR modulators. Compared with Cav1.2, recombinant Cav1.3 channels were 1.5-fold less sensitive to inhibition by flurazepam, 3-fold less sensitive to inhibition by pentobarbital, and 3-fold more sensitive to inhibition by diltiazem. Studies using Cav1.2 mutants revealed that pentobarbital potency was reduced 3.5-fold by mutation of the high-affinity DHP binding site (T1039Y and Q1043M). Despite the aforementioned structural similarities, diltiazem did not compete with diazepam inhibition, and mutation of an amino acid within the BTZ binding site (I1150A) had no effect on diazepam potency. Based on their concentration-dependent effects, inhibition of L-VGCCs by pentobarbital and ethanol may contribute to their acute clinical actions. A direct interaction of pentobarbital and ethanol with L-VGCCs may also contribute to the adaptive up-regulation of L-VGCC function after long-term use of these nonselective central nervous system depressants.
Materials and Methods
Cell Culture and Transient Transfection.
SV40 large T-antigen stably transfected HEK293 (HEK293T) cells were used for transient expression of L-VGCCs. Cells were maintained at 37°C and 5% CO2 in Dulbecco's modified Eagle's medium with GlutaMAX-I (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO), 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were grown to approximately 80% confluence and transfected with α1 (Cav1.2 or Cav1.3), β3, α2δ-1, and monomeric red fluorescent protein plasmids (1:1:1:0.5 μg, respectively) using Lipofectamine 2000 reagent (Invitrogen). One day after transfection, cells were trypsinized and replated on 35-mm culture dishes before electrophysiological recording 2 to 3 days after transfection. With the exception of experiments with mutated Cav1.2, all experiments used mouse neuronal Cav1.2 (Helton et al., 2005), rat neuronal Cav1.3 (+exon11, Δexon32, +exon42a; Xu and Lipscombe, 2001), rat neuronal β3, and rat neuronal α2δ-1 (Lin et al., 2004) subunit plasmids, which were generously provided by Dr. Diane Lipscombe (Brown University, Providence, RI). The monomeric red fluorescent protein plasmid was generously provided by Dr. Zi-Jian Xie (University of Toledo College of Medicine, Toledo, OH). A DHP binding site (T1039Y and Q1043M) mutant Cav1.2 (Hockerman et al., 2000) and its wild-type form from rat brain (Snutch et al., 1991) were generously provided by Dr. Gregory Hockerman (Purdue University, West Lafayette, IN).
Site-Directed Mutagenesis.
A mutation previously shown to be important for diltiazem potency, I1150A (Hockerman et al., 2000), was inserted into rat brain wild-type Cav1.2 using the QuikChange II XL site-directed mutagenesis kit (Stratagene Cloning Systems, La Jolla, CA) and the following primers: 5′-GTGGAGATCTCCATCTTCTTCGCGATCTACATCATCATCATTGCC-3′ and 5′-GGCAATGATGATGATGTAGATCGCGAAGAAGATGGAGATCTC-3′ (I1150A mutation in bold). The manufacturer's protocol was followed with a few modifications. In brief, 50 ng of Cav1.2 plasmid template was used in polymerase chain reaction with the following parameters: one cycle: 95°C, 30 s; 18 cycles: 95°C, 50 s, 60°C, 50 s, 68°C, 18 min (1.5 min/kb); one cycle: 68°C, 7 min. After DpnI treatment, the entire polymerase chain reaction product was ethanol-precipitated, resuspended in 5 μl of ddH2O, and transformed into XL10-Gold ultracompetent cells. Transformed cells were plated on LB agar dishes containing 100 μg/ml ampicillin for >16 h at 37°C. Single colonies were grown for >16 h in LB broth containing 100 μg/ml ampicillin; plasmid DNA then was extracted using phenol/chloroform/isoamyl alcohol (25:24:1). Mutation was confirmed by sequencing (Eurofins MWG Operon, Huntsville, AL). A clone positive for the mutation was grown in LB broth containing 100 μg/ml ampicillin, and the plasmid DNA was purified using QIAGEN Plasmid Maxi Kit (QIAGEN Inc., Valencia, CA).
Recording Solutions.
The external solution (300 mOsM) contained 135 mM choline Cl, 1 mM MgCl2, 5 mM BaCl2, and 10 mM HEPES, adjusted to pH 7.4 with TEAOH. As indicated in the results, a number of experiments used 2 mM Ca2+ as the charge carrier instead of 5 mM Ba2+. Glass-recording pipettes were pulled from micro-hematocrit capillary tubes (Thermo Fisher Scientific, Pittsburgh, PA) to a resistance of 2 to 5 MΩ. To facilitate seal formation, recording pipette tips were filled with a lower osmolarity internal solution (270 mOsM) lacking ATP. Pipettes were then back-filled with internal solution (285 mOsM) containing 135 mM CsCl, 4 mM MgCl2, 4 mM ATP, 10 mM HEPES, 10 mM EGTA, and 1 mM EDTA, adjusted to pH 7.2 with TEAOH.
Whole-Cell Voltage-Clamp Electrophysiology.
Red-fluorescing cells were visualized on an Olympus 1×51 fluorescent microscope (Olympus, Center Valley, PA). Whole-cell currents were recorded with an Axopatch 200A amplifier (Molecular Devices, Sunnyvale, CA). Data were sampled at 2 to 10 kHz, low pass-filtered at 1 kHz, and analyzed using pClamp v9.2 (Molecular Devices). After pipette and membrane capacitance compensation, the series resistance was corrected by 80 to 90% with a 20- to 100-μs lag time. Cells with a series resistance greater than 20 MΩ were not used for analysis. After whole-cell break-in, Ba2+ currents were elicited in voltage-clamp mode by stepping from a holding potential (Vh) of −80 mV to test potentials (Vt) of 0 (Cav1.2) or −25 mV (Cav1.3) for 200 ms. Current rundown, plotted as total charge transfer elicited at 10-s intervals (or higher frequencies, as indicated), was fitted with a one- or two-phase exponential decay curve. Recordings using this protocol were not leak-subtracted.
Drugs were applied by gravity flow and washed out rapidly by external solution applied with a separate pipette. Percentage inhibition was calculated with the following equation: % inhibition = (1 − [Iactual/Iexpected]) × 100, where Iactual is the actual current remaining during steady-state inhibition by the drug, and Iexpected is the expected current remaining at the same time point based on the exponential decay fit of the current rundown. Concentration-response curves were fitted with a four-parameter logistic equation: % inhibition = % min + (% max − % min)/{1 + 10(logic50 − log[drug]) × nH}, where % min represents percentage inhibition by vehicle, % max represents maximal percentage inhibition by the drug, IC50 was the drug concentration that yielded half-maximal inhibition, and nH was the Hill slope. % min was constrained to zero because average inhibition by vehicle was not significantly different from zero. Current decay was measured as the ratio of current remaining after a 200-ms depolarization relative to peak current, which was defined as the r200 value, similar to previous studies (Cai et al., 1997).
For voltage-dependent activation and inactivation protocols, data obtained from each voltage step was leak-subtracted online by removing the sum of the current obtained from four consecutive waveforms of 1/4 amplitude and opposite polarity to the test potentials (a P/4 protocol). Voltage dependence of activation was determined using either 5 mM Ba2+ or 2 mM Ca2+ as charge carriers. Channels were activated with 200-ms voltage steps from Vh = −80 mV to various test potentials, Vt = −75 to +40 mV (Cav1.3) or Vt = −50 to +65 mV (Cav1.2) in 5-mV steps. Peak current plotted against the test potential was fitted with the following equation: I = Gmax(Vt − Vrev)/{1 + exp[(V50 − Vt)/k]}, where Gmax is the maximal conductance of the cell, Vrev is the reversal potential, V50 is the potential for half-maximal activation, and k is the slope factor of activation. As the test potential approached Vrev, currents deviated from the linear relationship predicted by this equation. Therefore, currents obtained at potentials that were more positive than 25 mV were excluded from fitting analysis.
The voltage dependence of L-VGCC steady-state inactivation was determined by stepping to various conditioning pulse potentials for 1.5 s, preceded by a control pulse to −25 (Cav1.3) or 0 mV (Cav1.2) and followed by a test pulse to −25 (Cav1.3) or 0 mV (Cav1.2). The current elicited by the control pulse was used to control for current rundown. Inactivation curves were plotted as normalized peak currents (test peak/control peak) versus the conditioning pulse potential and fit with the Boltzmann equation: In = In(min) + (In(max) − In(min))/{1 + exp[(V50 − Vt)/k]}, where In is the normalized peak current, In(min) is minimal normalized current, In(max) is maximal normalized current, V50 is the potential for half-maximal inactivation, and k is the slope factor of inactivation. Drug effects on steady inactivation were tested by directly incubating cells in external solution containing the drug for at least 5 min before running the inactivation protocol.
Drugs.
The three BDZs tested were the prototype diazepam, the relatively water-soluble flurazepam, and its major bioactive metabolite in rat and man, desalkylflurazepam (Lau et al., 1987). Flumazenil, a competitive BDZ antagonist, was also tested. Stocks of nimodipine, diazepam, desalkylflurazepam, flumazenil, and allopregnanolone were made in dimethyl sulfoxide (DMSO). Stocks of (+)-cis-diltiazem, flurazepam, pH 5.8, and racemic pentobarbital were made in dH2O. Drug stocks were dissolved into external solution by vortexing. Drugs dissolved in DMSO were diluted at least 1:1000, with the exception that 300 mM diazepam and desalkylflurazepam stocks were diluted 1:300 to achieve the final concentration of 1 mM. DMSO diluted 1:1000 or 1:300 in external solution was used as a vehicle control for some experiments and resulted in negligible inhibition of currents (mean inhibition typically <5% in each experiment). Absolute ethanol (empirical density of 0.75 g/ml) was dissolved directly into external solution. Flurazepam was generously provided by the Drug Supply Program of the National Institutes of Health National Institute on Drug Abuse. Flumazenil and desalkylflurazepam were obtained from Roche (Nutley, NJ). Allopregnanolone was obtained from Tocris Bioscience (Ellisville, MO). Ethanol was obtained from Pharmco-AAPER (Shelbyville, KY). All other drugs were obtained from Sigma-Aldrich (St. Louis, MO).
Statistical Analyses.
All curve fits were made using Prism version 5 (GraphPad Software, Inc., La Jolla, CA). With the exception of curve-fit parameters, all data are reported as mean ± S.E.M. Curve-fit parameters (i.e., IC50, Hill slope, V50, and slope factor k) are reported as the best-fit value to the experimental data set with the associated 95% confidence interval from a Levenberg-Marquardt least-squares fitting algorithm. Nonoverlapping confidence intervals between values were accepted as significant differences (p < 0.05). A two-tailed Student's t test was used to compare percentage inhibition by select drug concentrations between two experimental conditions. Threshold drug concentrations were defined as the lowest concentration to yield significant inhibition measured by repeated measures one-way ANOVA with post hoc comparison by Dunnett's test. Mean r200 values were analyzed by repeated measures one-way ANOVA with post hoc comparison by Dunnett's test or by a paired t test. A p value of 0.05 or less was considered significant.
Results
Functional Characterization of Recombinant L-VGCCs.
As a confirmation that the transient transfection of HEK293T cells resulted in functional L-VGCCs, the voltage dependence of activation was evaluated for both Cav1.2- and Cav1.3-containing L-VGCCs and was similar to previous reports for these channels (Koschak et al., 2001; Xu and Lipscombe, 2001). Figure 2A shows Cav1.3 and Cav1.2 L-VGCC current-voltage responses, with peak 5 mM Ba2+ or 2 mM Ca2+ currents (picoamperes) normalized to cell capacitance (picofarads). It is noteworthy that activation of Cav1.3 channels occurred at more negative potentials than Cav1.2 channels. Half-maximal inactivation also occurred at more negative potentials in Cav1.3 compared with Cav1.2 channels (see Fig. 6 and Table 3). As an additional test of channel integrity and appropriate pharmacologic response, channels were inhibited by the DHP, nimodipine, in a concentration-dependent manner. As shown in Fig. 2B, Cav1.3 channels were less sensitive to inhibition by nimodipine than Cav1.2 channels, consistent with previous reports of recombinant L-VGCC inhibition by DHPs (Koschak et al., 2001; Xu and Lipscombe, 2001).
GABAAR Modulators Reversibly Inhibit Recombinant L-VGCCs.
In control cells transfected with Cav1.3 L-VGCCs (n = 4) in which no drug was applied, the average percentage current remaining 2, 5, and 10 min from the start of recording was 71 ± 11, 57 ± 13, and 42 ± 9%, respectively. This extent of rundown was similar to that of Cav1.2 L-VGCCs in this study and as reported previously (Kepplinger et al., 2000). There was no apparent correlation between the extent of rundown in different cells and drug concentration-response, suggesting that the mechanisms causing the rundown phenomenon were not a significant factor in modifying L-VGCC inhibition by GABAAR modulators. As seen in Fig. 3A, 5 mM Ba2+ currents through Cav1.3 L-VGCCs were inhibited in a concentration-dependent manner by diazepam. Inhibition by all drugs tested showed at least partial current recovery upon washout. Figure 3, B and C, illustrates that inhibition of Cav1.2 and Cav1.3 L-VGCCs, respectively, was rapid for most GABAAR modulators tested. For BDZs, pentobarbital, and ethanol, a maximal response was typically obtained by the third activating pulse after drug application. Inhibition was slower for allopregnanolone and diltiazem, generally requiring at least four activation pulses to reach maximal inhibition. For allopregnanolone, inhibition was slower in Cav1.2 channels, and for diltiazem, inhibition was slower in Cav1.3 channels. Inhibition of Cav1.3 channels by desalkylflurazepam was similar when 2 mM Ca2+ was used as the charge carrier (data not shown).
Figure 4, B and C, shows the concentration-response relationship of diltiazem and all GABAAR modulators tested in Cav1.2 and Cav1.3 L-VGCCs, respectively. Threshold concentrations, IC50 values, Hill slopes, and the numbers of cells tested are reported in Table 1. BDZs had relatively similar potencies, with the exception that Cav1.3 channels were 1.5-fold less sensitive to flurazepam than Cav1.2 channels (p < 0.05). The BDZs were approximately 2- to 6-fold less potent than diltiazem, with Cav1.3 channels displaying 3-fold greater sensitivity to diltiazem than Cav1.2 channels (p < 0.05). The BDZ receptor-competitive antagonist flumazenil had no effect to inhibit Cav1.2 or Cav1.3 channels at concentrations up to 300 μM (data not shown). Flumazenil (100–300 μM) was unable to antagonize diazepam inhibition of Cav1.2 channels (Table 1). It is interesting that coapplication with 100 μM flumazenil resulted in a 1.8-fold reduction in desalkylflurazepam potency in Cav1.3 channels (Table 1). Although allopregnanolone was the most potent GABAAR modulator tested (approximately 6-fold more potent than BDZs), it incompletely inhibited L-VGCCs and was less efficacious at inhibiting Cav1.3 than Cav1.2 channels (29 versus 63% maximal response, respectively). Pentobarbital was approximately 7- to 21-fold less potent than BDZs. Cav1.3 channels were 3-fold less sensitive to pentobarbital than Cav1.2 channels (p < 0.05). Ethanol was the least potent drug tested, approximately 4500-fold less potent than BDZs.
GABAAR Receptor Modulators Enhance L-VGCC Current Decay.
Inhibition by various L-VGCC antagonists is known to depend on the state of the channel (Lee and Tsien, 1983; Bean, 1984; Uehara and Hume, 1985; Cai et al., 1997; Hockerman et al., 1997). In particular, antagonists can stabilize the inactive state or display enhanced binding during channel opening (Hockerman et al., 1997), resulting in increased current decay during prolonged depolarization. Figure 5A illustrates that the decay of Cav1.3 Ba2+ current is enhanced by diazepam. Figure 5B shows that the ratio of residual current at 200 ms to peak current (r200) was reduced in a concentration-dependent manner by diazepam. Table 2 summarizes L-VGCC r200 values in the presence and absence of diltiazem and GABAAR modulators at approximately equipotent concentrations. A significant decrease in the r200 value of both L-VGCCs was observed for all GABAAR modulators, with the exception of flurazepam at Cav1.3 channels and ethanol at Cav1.2 channels because of variability and the low numbers of cells tested. By comparison, diltiazem at its IC50 concentration resulted in substantially less enhancement of L-VGCC current decay.
GABAAR Modulators Induce a Negative Shift in L-VGCC Steady-State Inactivation.
The voltage dependence of L-VGCC steady-state inactivation was measured as described in the methods in the presence and absence of drug. As illustrated in Fig. 6A, there was a significant shift in the voltage dependence of Cav1.2 L-VGCC steady-state inactivation toward more negative potentials in the presence of approximately equipotent concentrations of diazepam (60 μM) and pentobarbital (300 μM) but not allopregnanolone (30 μM) or ethanol (300 mM). The potentials of half-maximal inactivation (V50) in the presence and absence of diltiazem and GABAAR modulators are reported in Table 3. It is noteworthy that there was a concentration-dependent effect of diazepam to shift Cav1.2 inactivation toward more negative potentials, which was significant at 60 and 100 μM. Desalkylflurazepam (100 μM) caused a significant shift in the Cav1.2 inactivation curve but was smaller in magnitude than the shift by 100 μM diazepam. A similar result was obtained with 30 μM diltiazem. Consistent with data obtained from 0.1-Hz activation, flumazenil (300 μM) did not shift Cav1.2 inactivation and was unable to antagonize the shift by 100 μM diazepam.
Figure 6B shows a drug-induced shift of Cav1.3 steady-state inactivation toward more negative potentials in the presence of approximately equipotent concentrations of diazepam (60 μM), allopregnanolone (30 μM), pentobarbital (1 mM), and ethanol (300 mM), although ethanol was weakly effective relative to the other drugs. As shown in Table 3, 100 μM desalkylflurazepam caused a significant shift in the Cav1.3 inactivation curve but was of smaller magnitude than the shift by 100 μM diazepam. Diltiazem at its IC50 concentration of 10 μM was completely ineffective to shift the Cav1.3 inactivation curve.
Inhibition of Cav1.2 Channels by Desalkylflurazepam Is State- and Frequency-Dependent.
To test the hypothesis that BDZ potency is enhanced by channel inactivation, the inhibitory action of desalkylflurazepam was measured on channels activated from a more depolarized holding potential resulting in a greater extent of channels in the inactive state. As illustrated in Fig. 7A, desalkylflurazepam inhibited Cav1.2 channels 2-fold more potently when channels were activated from a more depolarized holding potential, with a significantly greater degree of inhibition occurring at 30 and 100 μM desalkylflurazepam. For Cav1.3 channels, which inactivate at more negative membrane potentials, a holding potential of −60 mV was used. As illustrated in Fig. 7B, desalkylflurazepam was similarly potent when Cav1.3 channels were activated from a more depolarized holding potential, and the degree of inhibition was not significantly different at any concentration at or above threshold.
To test whether BDZ potency might be similarly enhanced by the frequency of channel activation, desalkylflurazepam potency was tested on L-VGCCs activated at 1 Hz. As illustrated in Fig. 7C, desalkylflurazepam inhibited Cav1.2 L-VGCCs 2-fold more potently when channels were activated at 1 Hz compared with activation at 0.1 Hz, with a significantly greater degree of inhibition occurring at 10, 30, and 100 μM desalkylflurazepam. As illustrated in Fig. 7D, inhibition of Cav1.3 L-VGCCs by desalkylflurazepam was not significantly different when channels were activated at 1 Hz compared with activation at 0.1 Hz and showed no significant difference in inhibition at any concentration at or above threshold.
Two Amino Acids Important for DHP Potency Reduce Pentobarbital Potency.
Although the various classes of L-VGCC antagonists have unique binding sites, they also have overlapping amino acids that contribute to binding. Thus, it was of interest to see whether single and double amino acid mutations, already characterized to affect binding of specific L-VGCC antagonists, might also affect binding of GABAAR modulators. Previous studies have found two amino acids in domain IIIS5 of Cav1.2, Thr1039 and Gln1043, which are critical for DHP affinity (Mitterdorfer et al., 1996; Hockerman et al., 2000). Mutation of these two amino acids thus creates a DHP-insensitive Cav1.2 (DHPI). As can be seen in Fig. 8A, half-maximal inhibition of wild-type (WT) Cav1.2 L-VGCCs occurred at approximately 0.1 μM nimodipine and maximal block at 1 μM (data represent inhibition at 0-mV test potential shown in Fig. 2B). In two cells transfected with DHPI L-VGCCs, application of 1 μM nimodipine resulted in negligible inhibition (5 ± 14%). Application of 10 μM nimodipine resulted in substantial inhibition (58 ± 3%) but was approximately 100-fold less potent than in WT, confirming that DHPI channels are resistant to DHPs.
As shown in Fig. 8B, DHPI L-VGCCs were significantly less sensitive to inhibition by pentobarbital compared with WT. Although this suggests that Thr1039 and Gln1043 may play a role in pentobarbital binding to Cav1.2, the 3.5-fold reduction in potency is quite small compared with the estimated 100-fold reduction in nimodipine potency. As shown in Fig. 8, C and E, inhibition of Cav1.2 channels by diazepam and ethanol, respectively, was unaffected by mutation of the DHP binding site. Figure 8D illustrates that DHPI channels were slightly more sensitive to allopregnanolone than WT Cav1.2 channels, but only 10 μM allopregnanolone resulted in significantly more inhibition.
It should be noted that the WT Cav1.2 used for comparison of nimodipine, allopregnanolone, and ethanol inhibition was a mouse brain clone, whereas the DHPI mutant was created from a rat brain Cav1.2 clone. Thus, the small difference in inhibition by allopregnanolone might be due to species differences between the two channels. Additional controls in rat WT Cav1.2 were obtained for diazepam, pentobarbital, and diltiazem (see below).
A Single Amino Mutation That Affects Diltiazem Potency Does Not Affect Diazepam Potency.
Based on the structural similarity between BDZs and BTZs, it was hypothesized that they might share a similar binding site at L-VGCCs. A single amino acid mutation, I1150A, was previously shown to affect diltiazem potency but not other classes of L-VGCC antagonists (Hockerman et al., 2000). As shown in Fig. 9A, mutation of Ile1150 in rat Cav1.2 resulted in a significant 2-fold decrease in diltiazem potency, with a significantly lower degree of inhibition by 10, 30, and 300 μM diltiazem. However, as shown in Fig. 9B, diazepam potency was not significantly affected by the mutation, with no difference in the degree of inhibition by any concentration tested. To confirm that diazepam probably does not share a binding site with diltiazem, inhibition of WT Cav1.2 channels by diazepam was tested in the presence of an IC50 concentration of diltiazem (30 μM). As seen in Fig. 9C, the presence of diltiazem did not significantly affect inhibition at any diazepam concentration tested.
Discussion
GABAAR Modulator Pharmacology at L-VGCCs: Comparison with GABAARs.
Prior studies investigated inhibition of native L-VGCCs by GABAAR modulators either in non-neuronal or neuronal cells that also contain GABAARs and multiple types of calcium channels, making it difficult to ascertain the selective effects of these drugs on the neuronal L-VGCC subtypes Cav1.2 and Cav1.3. The current studies analyzed the inhibitory action of GABAAR modulators on recombinant neuronal Cav1.2 and Cav1.3 L-VGCCs assembled with β3 and α2δ subunits. Inhibition by modulators was not due to an indirect action at the GABAAR, because HEK293T cells do not express functional GABAARs (Davies et al., 2000).
The BDZs were approximately 3 orders of magnitude less potent (Thomas et al., 1997), allopregnanolone was approximately 1.5 orders of magnitude less potent (Maitra and Reynolds, 1998), pentobarbital was approximately 1 order of magnitude less potent (Thomas et al., 1997), and ethanol was slightly less than 1 order of magnitude less potent (Wallner et al., 2003) at L-VGCCs than at GABAARs. Although diazepam was severalfold more potent at GABAARs than flurazepam (Thomas et al., 1997), they displayed similar potency at L-VGCCs. The competitive BDZ antagonist, flumazenil, did not inhibit L-VGCCs, yet it antagonized BDZ inhibition of Cav1.3 but not Cav1.2 channels. This suggests that the Cav1.3 α1 subunit may have a flumazenil binding site analogous to GABAARs.
GABAAR Modulator Manner of L-VGCC Inhibition: Comparison with L-VGCC Antagonists.
Maximal inhibition by L-VGCC antagonists is dependent upon channel use, with PAAs displaying high levels of use dependence, followed by BTZs and then DHPs, which display little use dependence (Lee and Tsien, 1983; Uehara and Hume, 1985). Although the current studies were not designed to assess use dependence, maximal inhibition by BDZs, pentobarbital, and ethanol consistently required fewer activating pulses compared with inhibition by allopregnanolone and diltiazem (Fig. 3, B and C). Although these time course differences may be due to differential kinetics of drug binding and/or drug action, it could also suggest that the former compounds may not be as dependent on channel opening to bind and can better access their L-VGCC binding sites when channels are in a resting confirmation. Alternatively, antagonists that slow recovery from inactivation may result in an accumulation of channels in the inactive state during repeated activation and have greater effect at higher frequencies of activation (Uehara and Hume, 1985). Although inhibition of Cav1.2 channels by desalkylflurazepam was slightly enhanced in a frequency-dependent manner (Fig. 7), it is possible that these modulators have a limited ability to slow recovery from inactivation relative to BTZs and PAAs (Uehara and Hume, 1985).
GABAAR modulators enhance L-VGCC current decay, similar to other L-VGCC antagonists (Lee and Tsien, 1983; Cai et al., 1997). This could represent open channel block and/or stabilization of the inactive state. In support of the latter possibility, desalkylflurazepam potency was increased at Cav1.2 channels by recording paradigms that enhance channel inactivation (Fig. 7). In addition, GABAAR modulators induced a significant negative shift in the Cav1.3 steady-state inactivation curve, and BDZs and pentobarbital also induced a significant negative shift in the Cav1.2 steady-state inactivation curves (Fig. 6 and Table 3). Altogether, the data provide evidence for state-dependent block by GABAAR modulators, preferentially interacting with inactive and possibly open states.
Cav1.2 and Cav1.3 L-VGCC Differential Sensitivity to GABAAR Modulators.
Recombinant Cav1.3 channels have been shown to be less sensitive to inhibition by DHPs than Cav1.2 channels (Koschak et al., 2001; Xu and Lipscombe, 2001). Native Cav1.3 channels in auditory hair cells were also found to be less sensitive to inhibition by BTZs and PAAs, although this effect may be cell type-specific (Tarabova et al., 2007). Indeed, recombinant Cav1.3 channels were 3-fold more sensitive to diltiazem inhibition than Cav1.2 channels. For the GABAAR modulators, Cav1.3 channels were 1.5-fold less sensitive to flurazepam and 3-fold less sensitive to pentobarbital than Cav1.2 channels. Diazepam, desalkylflurazepam, allopregnanolone, and ethanol displayed no significant differential potency at L-VGCCs. However, because desalkylflurazepam inhibition of Cav1.2, but not Cav1.3 channels, was increased by protocols that enhance inactivation (Fig. 7), it is possible that Cav1.2 L-VGCCs may display greater sensitivity to the latter compounds in a state-dependent manner as proposed previously for DHPs (Koschak et al., 2001). Whether this differential state-dependent effect may be related to the greater extent of inactivation observed for Cav1.2 than Cav1.3 channels (Table 2) or a more specific BDZ interaction with the channels is currently unknown. The differential potency of these drugs at Cav1.2 and Cav1.3 channels may allow them to selectively modify distinct L-VGCC subunit-mediated neuronal functions. In particular, reduced Cav1.2 channel function impairs hippocampal spatial memory, whereas reduced Cav1.3 channel function affects neuronal firing patterns and may be protective in reducing age-related neuronal degeneration, as occurs in Parkinson's disease (Striessnig and Koschak, 2008).
GABAAR Modulator Site of Action.
Organic L-VGCC antagonists have unique binding sites on the Cav1.2 subunit that share a number of overlapping amino acids (Hockerman et al., 1997). Thus, it was of interest to see whether amino acid mutations that are known to affect binding of specific L-VGCC antagonists could similarly reduce inhibition by GABAAR modulators. Two Cav1.2 mutants were used in the current studies: a double amino acid mutation (T1039Y and Q1043M) that reduces DHP potency (Mitterdorfer et al., 1996; Hockerman et al., 2000), and a single mutation, I1150A, that reduces diltiazem potency (Hockerman et al., 2000). Neither mutation affected diazepam potency, suggesting that BDZs may have an L-VGCC binding site distinct from DHPs and BTZs. Furthermore, diltiazem did not competitively interfere with diazepam inhibition of Cav1.2 channels, and these compounds were somewhat dissimilar in their manner of block in terms of state and use dependence. Thus, despite the structural similarity between BDZs and BTZs and their relatively similar potencies, the evidence suggests that these drugs act via distinct sites. It is noteworthy that pentobarbital potency was significantly reduced 3.5-fold by the DHP mutant, suggesting that the pentobarbital binding site on Cav1.2 might overlap that of DHPs.
Clinical Relevance of L-VGCC Inhibition by GABAAR Modulators and Implications for Physical Dependence.
Calcium influx through postsynaptic L-VGCCs is linked to neuronal activity and gene transcription, and L-VGCC inhibition has anxiolytic, antidepressant, and anticonvulsant actions (Striessnig et al., 2006). Thus, inhibition of neuronal L-VGCCs could play a role in the acute clinical actions of GABAAR modulators. Furthermore, long-term inhibition of L-VGCCs may lead to an adaptive up-regulation of L-VGCC function during prolonged GABAAR modulator treatment and contribute to physical dependence (Dolin et al., 1987; Rabbani and Little, 1999; Xiang and Tietz, 2007; Xiang et al., 2008).
Free plasma and cerebrospinal fluid concentrations of BDZs, barbiturates, and ethanol exist in an approximate 1:1 ratio (Richards, 1972; Hallstrom and Lader, 1980; Harris et al., 2008). Taking into account extensive binding (>97%) of diazepam and its active metabolite, desmethyldiazepam, to plasma proteins (Hallstrom and Lader, 1980; Divoll and Greenblatt, 1981), free plasma concentrations of approximately 20 to 60 nM are necessary for acute relief of convulsant activity (Rey et al., 1999), and approximately 50 to 100 nM is needed to produce sedation (Lundgren, 1987). Because of rapid tolerance to the sedative effects of diazepam, higher concentrations up to 200 nM are achieved during long-term treatment for psychiatric therapy (Rutherford et al., 1978; Hallstrom and Lader, 1980; Greenblatt et al., 1981). Because the threshold concentration of BDZs to significantly inhibit recombinant L-VGCCs was approximately 10 μM, inhibition of neuronal L-VGCCs is unlikely to contribute to BDZ clinical actions. However, studies in hippocampal cultures revealed that 1 μM flurazepam significantly inhibited VGCC current in a use-dependent manner (Xiang et al., 2008), suggesting that greater potency can be observed in neuronal systems. The threshold concentration of allopregnanolone to inhibit L-VGCCs was 3 μM, which is much greater than the 5 nM concentration estimated to be produced from progesterone in the brain (Uzunova et al., 1998). The threshold concentration of pentobarbital to inhibit L-VGCCs was 100 μM, within the 200 μM range required to induce anesthesia (Richards, 1972). Unlike pentobarbital, phenobarbital has anticonvulsant actions with minimal sedative effects at 20 to 90 μM (Schulz and Macdonald, 1981). It would be of interest to see whether phenobarbital inhibits L-VGCCs at concentrations required for anticonvulsant activity or whether L-VGCC inhibition is more likely to pertain to the anesthetic effects of barbiturates (ffrench-Mullen et al., 1993). The threshold concentration of ethanol to inhibit L-VGCCs was 30 mM, within range of intoxicating ethanol concentrations (10–100 mM) (Harris et al., 2008). Thus, inhibition of L-VGCCs may attribute to the intoxicating effects of ethanol, including mood alteration and memory impairment.
The enhancement of neuronal L-VGCC-mediated Ca2+ influx observed after long-term BDZ exposure is unlikely to be the result of a direct effect at L-VGCCs because this up-regulation occurs with BDZ concentrations as low as 0.3 μM in vitro (Katsura et al., 2007) and ∼1 μM measured in rat brain homogenates during long-term BDZ treatment in vivo (Xiang et al., 2008). Thus, persistent allosteric enhancement of GABAARs seems the most plausible mechanism associated with L-VGCC regulation during long-term BDZ administration because L-VGCC up-regulation in cortical cultures can be prevented by nanomolar concentrations of the BDZ-competitive antagonist, flumazenil (Katsura et al., 2007). In addition, a bicarbonate-dependent GABAAR-mediated depolarization that arises during BDZ withdrawal (Zeng and Tietz, 2000) may contribute to the hyperexcitable state (Van Sickle et al., 2004), which could exacerbate the downstream Ca2+-mediated effects of L-VGCC up-regulation. As with other GABAAR modulators, withdrawal from long-term exposure to allopregnanolone is associated with a withdrawal syndrome characterized by anxiety and increased seizure susceptibility (Smith, 2002). However, a role for L-VGCC up-regulation in this phenomenon has not been established. Inhibition of L-VGCCs may contribute to the up-regulation of L-VGCC function after long-term exposure to barbiturates (Rabbani and Little, 1999), as well as ethanol (Messing et al., 1986; Katsura et al., 2006). Collectively, these findings may provide a molecular basis for the adaptive changes that contribute to drug withdrawal symptoms and physical dependence on GABAAR modulators.
Authorship Contributions
Participated in research design: Earl and Tietz.
Conducted experiments: Earl.
Performed data analysis: Earl.
Wrote or contributed to the writing of the manuscript: Earl and Tietz.
Acknowledgments
We thank Drs. L. John Greenfield, Jr., David Giovannucci, Paromita Das, and Nathalie Boulineau for technical assistance and advice with plasmid preparation, transient transfection procedures, and experimental design. We thank Dr. Scott Molitor for critical review of the data analyses and manuscript and Drs. Diane Lipscombe and Gregory Hockermann for the generous contribution of L-VGCC subunit cDNA plasmids.
Footnotes
This work was supported by the National Institutes of Health National Institute on Drug Abuse [Grants R01-DA184342, F30-DA026675 (to E.I.T. and D.E.E., respectively)].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
doi:10.1124/jpet.110.178244.
-
ABBREVIATIONS:
- BDZ
- benzodiazepine
- ANOVA
- analysis of variance
- BTZ
- benzothiazepine
- DMSO
- dimethyl sulfoxide
- DHP
- dihydropyridine
- L-VGCC
- L-type voltage-gated calcium channel
- PAA
- phenylalkylamine
- TEA
- tetraethylammonium
- WT
- wild type
- LB
- Luria-Bertani
- HEK
- human embryonic kidney.
- Received December 14, 2010.
- Accepted January 21, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics