Lactone Modulation of the γ-Aminobutyric Acid A Receptor: Evidence for a Positive Modulatory Site
- Korwyn L. Williams1,
- Joseph B. Tucker1,
- Geoffrey White5,
- David S. Weiss4,
- James A. Ferrendelli,
- Douglas F. Covey3,
- James E. Krause1 and
- Steven M. Rothman1,2
- 1Department of Anatomy and Neurobiology (K.L.W., J.B.T., J.E.K., S.M.R.), Departments of 2Neurology and Pediatrics (S.M.R.), 3Department of Molecular Biology and Pharmacology (D.F.C.), Washington University School of Medicine, Saint Louis, Missouri 63110, 4Department of Neuroscience, University of Alabama at Birmingham, Birmingham, Alabama 35294 (D.S.W.), Department of Neurology, University of Texas, Houston, Texas 77225, and 5Neurogen Corporation, Branford, Connecticut 06425 (G.W.)
Abstract
The γ-aminobutyric acid-A (GABAA) receptor complex is allosterically modulated by a variety of substances, some of clinical importance. Barbiturates and neurosteroids augment GABA-currents and also directly gate the channel. A variety of γ-butyrolactone analogues also modulate GABA-induced currents, with some potentiating and others inhibiting. Because several γ-thiobutyrolactone analogues have biphasic effects on GABA currents, experiments with wild-type and picrotoxinin-insensitive GABAA receptors were performed to analyze whether some γ-thiobutyrolactones interact with two distinguishable sites on the GABAA receptor. β-Ethyl-β-methyl-γ-thiobutyrolactone inhibited GABA-induced currents at low concentrations (0.001–1 mm), but potentiated GABA-induced currents at higher concentrations (3–10 mm) in wild-type α1β2γ2-subunit containing ionophores. The related α-ethyl-α-methyl-γ-thiobutyrolactone potentiated submaximal GABA currents in wild-type receptors at both low and high concentrations (0.1–10 mm). Mutations in the second transmembrane domain of α1, β2, or γ2 conferred picrotoxinin-insensitivity onto GABAA receptor complexes. When these mutated α1, β2, or γ2 subunits were incorporated into the receptor complex, β-ethyl-β-methyl-γ-thiobutyrolactone potentiated GABA currents over the entire concentration range (0.1–10 mm). Neither the potentiating activity nor the EC50 of α-ethyl-α-methyl-γ-thiobutyrolactone changed in the mutant receptors. Further studies demonstrated that the mutations did not affect the EC50 of chlordiazepoxide or phenobarbital. These and our earlier results identify a modulatory site on the GABAA receptor distinct from that interacting with barbiturates, benzodiazepines, and steroids. Additionally, they show that the γ-butyrolactones probably interact at two different sites on the ionophore to produce opposite effects on GABA-mediated current.
The GABAA receptor is responsible for most fast inhibitory neurotransmission in the central nervous system. Consequently, this receptor has been targeted for the pharmacological control of anxiety, sleep, and epilepsy. Numerous natural and synthetic compounds interact with the GABAAreceptor at distinct, yet incompletely defined, sites. These compounds include barbiturates, benzodiazepines, neurosteroids, and picrotoxin (1, 2).
γ-Butyrolactones and the related TBLs are a group of convulsant and anticonvulsant agents that interact with the GABAAreceptor, but not at the benzodiazepine or barbiturate sites (3, 4). Behavioral studies have demonstrated the protective effects of these compounds against various seizure models (5). Displacement studies with [35S]TBPS, a ligand binding at the picrotoxinin site (6), suggested an interaction between the TBLs and the picrotoxinin site on the GABAA receptor. The hypothesis that TBLs interacted with the GABAA receptor at the picrotoxinin site as either agonists or inverse agonists (i.e., blockers or potentiators of GABA-induced currents, respectively) explained most of the findings (7).
The agonist/inverse agonist concept was not novel, because benzodiazepine receptor ligands can negatively (e.g., DMCM) and positively (e.g., diazepam) modulate channel behavior (8). The benzodiazepine receptor ligands are thought to act at just one site (9,10).
Several new observations led to a re-evaluation of the TBL agonist/inverse agonist model. Holland and colleagues (11, 12) showed biphasic effects of several TBLs on GABA currents. They found that the same compound could both block and potentiate GABA currents, albeit with different time and concentration dependencies. They also observed noncompetitive interactions between [35S]TBPS and some TBLs, suggesting two sites of interaction: a negative modulatory site (the picrotoxinin site) and a yet-to-be-defined, positive modulatory site (the lactone site). In the present studies picrotoxinin-insensitive GABAA receptor mutants were used to test this hypothesis. If the two-site model for lactone interaction with the receptor were correct, the biphasic nature of the TBLs on GABA currents should disappear, when one site was ablated.
Materials and Methods
GABAA subunits.
The cDNAs encoding GABAA subunits α1 were provided by A. Tobin (University of California, Los Angeles), β2 by P. Malherbe (Hoffman-La Roche, Switzerland), and γ2L by C. Fraser (National Institute on Alcohol Abuse and Alcoholism).
Mutagenesis.
Mutations were created as previously described (13).
Compounds.
GABA, chlordiazepoxide hydrochloride, and phenobarbital were obtained from Sigma (St. Louis, MO), and DMCM was obtained from Research Biochemicals International (Natick, MA). α-EMTBL and β-EMTBL were synthesized by previously described methods and had the appropriate analytical and spectroscopic properties (5).
cRNA preparation.
The cRNA transcripts were prepared in the following manner. The 5′- and 3′-untranslated region of cDNA templates were removed to improve channel expression, and an optimal Kozak (14) sequence (CCACC) was added upstream of the initiator methionine to enhance mRNA binding to the 40 S ribosomal subunit. cDNA sequences used in the study were verified by automated nucleotide sequence analysis using the Applied Biosystems Incorporated (ABI; Foster City, CA) Prism Dye Terminator cycle sequencing system and an ABI 373A DNA sequencer. The modified cDNA templates were linearized with the appropriate restriction enzyme, and transcripts were prepared with Ambion’s mMessage mMachine kit (Austin, TX). As a final check, each cRNA preparation was subjected to either polyacrylamide or agarose gel analysis to verify size and quantity.
Oocyte collection and injection.
Oocyte preparation was modified from the methods of Gurley et al. (13).Xenopus laevis oocytes were collected from mature, pigmented females obtained from Xenopus One (Northland, MI). After removal, the oocytes were placed in an OR-2 solution containing 100 units of penicillin and 100 μg/ml streptomycin obtained from Sigma and 2 mg/ml collagenase I from Boehringer-Mannheim (Indianapolis, IN). The oocytes were agitated vigorously until the follicular layer was removed. Usually, the oocytes were injected the same day with a 20–60-ng mixture of α, β, and γ subunits, using the Drummond Nanojector (Broomall, PA). The oocytes were incubated for 1–3 days in L-15 medium (GIBCO, Grand Island, NY) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml gentamicin sulfate (Sigma).
Electrophysiology and data analysis.
The electrophysiology was carried out using Warner Instrument’s 725C (Hamden, CT) two-electrode voltage clamp with a virtual ground bath clamp. Electrodes filled with 3 m KCl (resistance < 1 MΩ) were used to impale the oocytes. The oocytes were clamped at a potential of −70 mV. Peak currents were recorded, low pass filtered at 20–30 Hz by an 8-pole Bessel filter (Frequency Devices, Haverhill, MA), digitized by a TL-1 DMA interface (Axon Instruments, Foster City, CA), and stored using Axon Instruments’ AxoTape. The oocytes were perfused with a solution containing the following: 90 mmNaCl, 1 mm kcl, 2 mmMgCl2, 5 mm HEPES and 1 mmCaCl2 (pH 7.3). Solutions were bath applied by a gravity-driven perfusion system until the response reached plateau or until a rapidly desensitizing response peaked, which took anywhere from 3 to 45 sec approximately. No correction in peak amplitude was made for quickly desensitizing responses that were common at high GABA concentrations, high drug concentrations, and also in mutated receptors. In the determination of the EC50 values of α-EMTBL, β-EMTBL, and chlordiazepoxide, the GABA EC10was used as a control. Data were fit using a logistic equation (SigmaPlot; Jandel Scientific, Sausalito, CA), y = [(a − d)/(1 + (x/ EC50)nH)] + d , where y is the response, a is the asymptotic maximum, d is the asymptotic minimum,x is the ligand concentration, and nHis the Hill coefficient. In situations in which the 10 mm response was smaller than the 3 mm response, the 10 mm value was not used in the fitting procedure. This value was ignored, as it probably reflected some degree of both channel block and desensitization in addition to the potentiation; fitting such complex behavior is beyond the limits of the logistic equation. Even after elimination of the 10 mm response, concentration responses for α-EMTBL and β-EMTBL were fit well using the logistic equation and allowing the maximum to float.
Results
GABA sensitivity.
To confirm that the double point mutations in the TM2 did not drastically alter the behavior of the channel, the GABA responses of the mutant ionophores were examined. Wild-type α1β2γ2 GABAA receptors had an EC50 value of 42 μm; the GABA EC50 values of the channels containing mutated subunits changed by less than 5-fold (Fig.1A). The apparent affinity of α1Mβ2γ2 and α1β2Mγ2, but not α1β2γ2M, for GABA was significantly different from that of wild-type channels (ANOVA with Dunnett’s test; p < 0.05). Although these shifts could reflect changes in GABA affinity for the receptor, they could also have resulted from altered GABA gating (15, 16).
A, GABA concentration-response curves in α1β2γ2, α1Mβ2γ2, α1β2Mγ2, and α1β2γ2M-subunit containing ionophores. The EC50 values and Hill coefficients were, respectively, 40 μm and 1.6 (n = 9); 172 μm and 1.3 (n = 5); 114 μm and 1.0 (n = 3); 78 μm and 1.4 (n = 3). The α1Mβ2γ2 and α1β2Mγ2 EC50 values were significantly different (p < 0.05) from α1β2γ2 by a one-way ANOVA with Dunnett’s test; the EC50 for α1β2γ2M was not significantly different (p > 0.05). The GABA EC10concentrations for each receptor combination are: 10 μm(α1β2γ2), 27 μm (α1Mβ2γ2), 18 μm (α1β2Mγ2), and 16 μm(α1β2γ2M). B, Picrotoxinin (PTXNIN) concentration-response in wild-type α1β2γ2 (○) and mutant α1Mβ2γ2 (▴), α1β2Mγ2 (▪), and α1β2γ2M (•).
Picrotoxin and picrotoxinin sensitivity of wild-type α1β2γ2.
Two point mutations in the TM2 of the α1, β2, or γ2 GABA subunit or a single point mutation in the β2 subunit produce picrotoxin-insensitivity (13). The pharmacology of the GABAA receptor was studied using the approximate GABA EC10 as a control. Oocytes expressing the wild-type GABAA receptor were inhibited by both picrotoxin and picrotoxinin (the active component of picrotoxin), with IC50 values of 1.2 and 0.8 μm, respectively (data not shown).
Oocytes injected with α1β2Mγ2 showed a similar insensitivity to picrotoxin and picrotoxinin (Fig. 1B) (13). The other two mutant complexes (α1Mβ2γ2 and α1β2γ2M) also were picrotoxinin-insensitive (Fig.1B). Because of the small magnitude of currents observed when two or more mutated subunits were co-injected, such receptor complexes were not studied.
β-EMTBL modulation of wild-type and mutant receptors.
β-EMTBL both inhibits and potentiates GABA currents in primary cultures of hippocampal neurons (12). At concentrations less than 1 mm, β-EMTBL inhibited GABA currents in wild-type receptors; at 1 mm, β-EMTBL partially relieved its own block; finally, at 3 and 10 mm β-EMTBL, the macroscopic current was potentiated relative to control (Fig. 2A). This biphasic modulation was even more clearly seen in the concentration response curve (Fig. 2C).
β-EMTBL concentration response in wild-type and picrotoxinin-insensitive α1β2γ2-subunit containing ionophores.Arrows, small off-currents seen at 10 mmdrug. A, GABA currents modulated by β-EMTBL in α1β2γ2, demonstrating its ability to inhibit and potentiate a GABA-gated current. Calibration bars: 25 nA, 1 min. B, GABA currents modulated by β-EMTBL in α1Mβ2γ2, demonstrating its ability to only potentiate a GABA-gated current. Calibration bars: 100 nA, 1 min. C, The β-EMTBL concentration-response in wild-type and α1Mβ2γ2, α1β2Mγ2, and α1β2γ2M. Note that the biphasic modulation is lost in the picrotoxinin-insensitive ionophores.
In contrast to its activity in wild-type receptors, β-EMTBL did not inhibit GABA currents in the picrotoxinin-insensitive channels (Fig.2B). Below 300 μm β-EMTBL, GABA currents were no different from control or only slightly potentiated. Above this concentration range, potentiation was robust and occasionally greater than that seen in the wild-type (Fig. 2C).
α-EMTBL modulation of wild-type and mutant receptors.
α-EMTBL is one of the most potent and efficacious anticonvulsant TBL analogues. It does not produce biphasic modulation like β-EMTBL, suggesting it does not interact strongly with the picrotoxinin receptor. With oocytes expressing wild-type α1β2γ2 GABAA receptors, α-EMTBL potentiated submaximal GABA currents. At 10 mm, α-EMTBL exhibited less potentiation than at 3 mm and also exhibited off-currents (Fig.3A).
α-EMTBL concentration response in wild-type and picrotoxinin-insensitive α1β2γ2-subunit containing ionophores.Arrows, off-currents seen at 10 mm drug. A, GABA currents modulated by α-EMTBL in α1β2γ2, demonstrating its ability to potentiate a GABA-gated current. Calibration bars, 100 nA, 1 min. B. GABA currents modulated by α-EMTBL in α1Mβ2γ2, demonstrating its ability to only potentiate a GABA-gated current. Calibration bars, 100 nA, 1 min. C, The α-EMTBL concentration response in α1β2γ2, α1Mβ2γ2, α1β2Mγ2, and α1β2γ2M. The EC50 values and Hill coefficients are: 597 μm and 2.0, 600 μmand 1.6, 501 μm and 1.9, and 537 μm and 1.7, respectively. The differences in EC50 values of the mutants were not significantly different from wild-type values, as assessed by a one-way ANOVA (p = 0.20).
α-EMTBL’s activity was unchanged in the mutated receptor complexes (Fig. 3B). Potentiation was observed over the entire concentration range, even though the 10 mm response once again fell off. Quantitatively, the EC50 values were relatively unaffected: α-EMTBL’s EC50 in the picrotoxinin-insensitive channels differed by less than 20% from wild-type. Yet, the efficacy for one of the complexes (α1Mβ2γ2) was 50% greater than wild-type (Fig. 3C).
Chlordiazepoxide and phenobarbital modulation.
Double point mutations in the TM2 sufficed to eliminate picrotoxinin modulation of the channel; moreover, GABA concentration responses from two types of picrotoxinin-insensitive channels were significantly different from wild-type channels. To address whether these mutations affected more than picrotoxinin and β-EMTBL modulation, we examined three other allosteric modulators of the channel: a benzodiazepine agonist, benzodiazepine inverse agonist, and a barbiturate. Each compounds’ activity was qualitatively unaffected by the double point mutations in the TM2. Chlordiazepoxide’s EC50 was not changed by more than 4-fold by the mutations (Table 1); only the EC50 for α1Mβ2γ2 was statistically different from the wild-type EC50. The efficacy, however, was quite variable, with the α1 mutant giving the most robust response (Table 2).
Determination of EC50 values of chlordiazepoxide and phenobarbital in wild-type and picrotoxinin-insensitive mutants
Efficacy as percentage of GABA EC10 response
DMCM negatively modulates the GABA receptor through the benzodiazepine site, which is demonstrated by flumazenil antagonism (17). In the γ2 mutant DMCM’s efficacy was unchanged, confirming that the point mutations abolished only picrotoxinin’s negative modulation (Table 2).
Phenobarbital could not be modeled using the logistic equation, because its concentration response reflects not only potentiation of GABA-gated currents, but also direct gating by phenobarbital. Hence, only its efficacy is reported, but there were no significant differences (one-way ANOVA: p = 0.42) in phenobarbital efficacy by mutations in the TM2 (Table 2).
Flumazenil inhibition.
Last, flumazenil was used to test whether the lactone interacted with the ionophore at the benzodiazepine receptor. Flumazenil was unable to block α-EMTBL potentiation in a wild-type α1β2γ2-containing ionophore, providing additional evidence that the lactone site is distinct from the benzodiazepine site (Fig. 4).
Comparison of the ability of 1 μmflumazenil to inhibit the approximate EC50 of both chlordiazepoxide and α-EMTBL. Flumazenil did inhibit chlordiazepoxide (CDPX) (two-sided t test:p < 0.0001), but not α-TBL (two-sidedt test: p = 0.2).
Discussion
The results in this study alter our early view that lactones modulate GABAA currents through action at a single site on GABAA channels, the picrotoxinin receptor. Because picrotoxinin contains a γ-butyrolactone ring as part of its structure, it seemed reasonable that the neuroactive γ-butyrolactone analogues could interact at the GABAA receptor through the picrotoxinin binding site (18, 19). Supporting evidence for this hypothesis came from [35S]TBPS displacement studies that revealed a high affinity, competitive interaction between β-substituted TBLs and the picrotoxinin receptor. Moreover, the anticonvulsant α-substituted TBLs also displayed an ability to displace TBPS, albeit with less potency (20). Furthermore, modeling of picrotoxinin and different lactones demonstrated potential regions for interaction with the channel interior (19). Therefore, the agonist/inverse agonist model of lactone activity seemed firmly grounded both experimentally and theoretically.
Recent findings prompted us to consider the hypothesis that lactones may interact at two sites (11, 12). To demonstrate clearly that some TBLs interact with two sites, one of the sites had to be functionally, even if not physically, absent. With the picrotoxinin-like effect absent, β-EMTBL only potentiated GABA currents. The affinity and potentiation activity of α-EMTBL was unchanged, suggesting that the α-substituted lactones have, at best, weak interactions with the picrotoxinin site.
Two issues were considered in interpreting the data. Inasmuch as the mutations were introduced into the TM2, a region of the GABAA subunits critical for proper channel function, it is possible that these mutations could have conformational consequences that led to the loss of the biphasic activity of β-EMTBL. Although the observed picrotoxinin insensitivity could have resulted from disruption of the picrotoxinin binding site, it could also be explained by restricted access to a binding site, or altered channel conformation after picrotoxinin binding. If conformational changes were disturbed by the mutations, then the loss of β-EMTBL’s biphasic activity would be difficult to ascribe to two binding sites. If this were the case, the activity of other modulators might be expected to change as well. Given that the mutants’ responses to GABA, α-EMTBL, chlordiazepoxide, DMCM, and phenobarbital are qualitatively similar to responses in the wild-type channel, these picrotoxinin-insensitive channels do not appear to alter the conformational equilibria. Therefore, these mutations appear to target specifically the picrotoxinin site and do not alter macroscopic responses to other modulators of GABAA currents.
A second concern in interpreting the data was the decrease in potentiation seen at 10 mm lactone (Figs. 2B and 3, A and B). The fall-off in potentiation can be attributed to desensitization or to block at a second site at high drug concentrations (12). The contribution of desensitization can be inferred from similar fall-offs in current seen with high concentration GABA responses (10 mm), chlordiazepoxide (100 μm; data not shown), and phenobarbital (3 mm; data not shown). The other consideration, a second block site, is reflected by the off-currents observed with the washout of high lactone concentrations (arrows in Figs. 2, A and B, and 3, A and B). Similar block has also been seen with washout of 3 mm phenobarbital, so this is not necessarily unique to the lactones.
We believe that our data can be most parsimoniously explained by a lactone site that is distinct from the barbiturate, benzodiazepine, and steroid sites (4, 7, 21-23). Our next goal is to localize the lactone site and determine the physical explanation for its influence on GABAA channel conductance. We believe that this site represents a logical target for drugs that will be effective therapy for epilepsy, spasticity, and sleep disorders.
Footnotes
-
Send reprint requests to: Steven M. Rothman, M.D., Department of Neurology, St. Louis Children’s Hospital, One Children’s Place, Saint Louis, MO 63110. E-mail:rothman{at}kids.wustl.edu
-
This work was supported in part by National Institutes of Health Grants NS14834, NS07027, and AA09212, the Monsanto Fund, and the Seay Fellowship.
- Abbreviations:
- GABAA
- γ-aminobutyric acid-A
- TBL
- γ-thiobutyrolactone
- DMCM
- methyl-6,7-dimethoxy-4-ethyl-β-carboline-3-carboxylate
- TBPS
- tert-butyl-bicyclophosphorothionate
- α-EMTBL
- α-ethyl-α-methyl-γ-thiobutyrolactone
- β-EMTBL
- β-ethyl-β-methyl-γ-thiobutyrolactone
- TM2
- second transmembrane domain
- ANOVA
- analysis of variance
- HEPES
- 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
-
- Received December 19, 1996.
- Accepted March 31, 1997.
- The American Society for Pharmacology and Experimental Therapeutics
