Mechanisms of anabolic androgenic steroid modulation of α1β3γ2L GABAA receptors

Neuropharmacology 43 (2002) 619–633 www.elsevier.com/locate/neuropharm

Mechanisms of anabolic androgenic steroid modulation of α1β3γ2L GABAA receptors Paul Yang a, Brian L. Jones a, Leslie P. Henderson a,b,∗ a b

Dartmouth Medical School, Department of Physiology, 03755 Hanover, NH, USA Dartmouth Medical School, Department of Biochemistry, 03755 Hanover, NH, USA

Received 20 May 2002; received in revised form 26 June 2002; accepted 16 July 2002

Abstract Modulation of GABAA receptors induced by both anabolic androgenic steroids (AAS) and the benzodiazepine (BZ) site agonist, zolpidem, show equivalent dependence upon γ subunit composition suggesting that both compounds may be acting at a shared allosteric site. Here we have characterized modulation induced by the AAS, 17α-methyltestosterone (17α-MeT), for responses elicited from α1β3γ2L GABAA receptors and compared it to modulation induced by the BZ site agonists, zolpidem and diazepam. For responses elicited by brief pulses of 20 µM GABA, both the AAS and the BZ site compounds significantly increased the peak current amplitudes and total charge transfer, although 17α-MeT was an appreciably weaker agonist than either diazepam or zolpidem at α1β3γ2L receptors. Neither class of modulator enhanced peak current amplitudes for responses elicited by mM concentrations of GABA. BZ site compounds altered time constants of deactivation, desensitization, and recovery from desensitization, however 17αMeT had no overall effect on these parameters. Experiments in which 17α-MeT and BZ site ligands were applied concomitantly indicated that potentiation elicited by 17α-MeT and zolpidem were additive and that potentiation by 17α-MeT could be elicited in the presence of concentrations of flumazenil that blocked BZ potentiation. Finally, kinetic modeling suggests that while effects of 17α-MeT can be simulated by altering receptor affinity, the data for these α1β3γ2L receptors were best fitted by simulations in which 17α-MeT increases transitions into the singly liganded open state. Taken together, our results suggest that 17α-MeT does not act at the high-affinity BZ site, but may elicit some of its effects at the low affinity BZ site or at a novel site.  2002 Elsevier Science Ltd. All rights reserved. Keywords: Anabolic androgenic steroids; GABAA receptors; HEK293 cells; Fast perfusion; Benzodiazepines; Modeling

1. Introduction Anabolic androgenic steroids (AAS) are synthetic derivatives of testosterone originally developed for clinical purposes, but now predominantly taken as drugs of abuse (for review, Lukas, 1996). AAS abuse is associated with adverse effects on reproductive, cardiac, and liver function, as well as psychological symptoms, including changes in anxiety and aggression (for review, Pope and Katz, 1988; Strauss and Yesalis, 1991; Lukas, 1996). While AAS effects on behavior clearly implicate AAS in having actions in the central nervous system (CNS), little is known about how these steroids act in the brain. A wide range of psychoactive agents is known

∗

Tel.: +1-603-650-1312; fax: +1-603-650-1128. E-mail address: [email protected] (L.P. Henderson).

to elicit effects by inducing allosteric modulation of the γ-aminobutyric acid type A (GABAA) receptor (for review, Sieghart, 1995; Mehta and Ticku, 1999; Lambert et al., 2001). Biochemical studies have demonstrated that two 17α-alkylated AAS, 17α-MeT and stanozolol, when applied at micromolar concentrations, significantly inhibit binding of the BZ site ligand, flunitrazepam, in rat brain cortical membranes (Masonis and McCarthy, 1995) and that stanozolol decreases both the Emax and the EC50 values for GABA-stimulated 36Cl– influx in cortical synaptoneurosomes (Masonis and McCarthy, 1996). More recently, it has been shown that these two 17αalkylated AAS, as well as the 19-nortestosterone AAS, nandrolone, can induce rapid, reversible, and subunitspecific modulation of GABAA receptors. Specifically, all three AAS were found to potentiate GABAA receptormediated currents of recombinant α2β3γ2L receptors expressed in the human embryonic kidney (HEK293)

0028-3908/02/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 8 - 3 9 0 8 ( 0 2 ) 0 0 1 5 5 - 7

620

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

cell line and currents in neurons of the ventromedial nucleus (VMN) of the hypothalamus, which express predominantly α2β3γ2 receptors. All three AAS, however, inhibited currents elicited from recombinant α2β3γ1 receptors and from neurons of the medial preoptic area (mPOA) that express predominantly α2β3γ1 receptors (Jorge-Rivera et al., 2000). While AAS induced negative modulation of both recombinant and native γ1-containing receptors, endogenous neurosteroids induced positive modulation. The opposing modulatory effects of the AAS and the neurosteroids on γ1-containing receptors (Jorge-Rivera et al., 2000), as well as the fact that key structural elements shown to be common to all active neurosteroids (for review, Lambert et al., 1995) are absent from the AAS, do not support the hypothesis that the AAS are acting at the neurosteroid modulatory site. In contrast, the observed reversal in AAS modulation from potentiation to inhibition when a γ1 subunit is substituted for a γ2 subunit resembles the same inversion of modulation that occurs for the benzodiazepine (BZ) type I agonist, zolpidem (Puia et al., 1991; Nett et al., 1999), suggesting that both classes of compounds may be acting at a common modulatory site. Positive BZ site modulators are believed to act by increasing receptor affinity for GABA and thereby increasing the frequency of channel openings (Rogers et al., 1994; Lavoie and Twyman, 1996; Mellor and Randall, 1997; for review, Sieghart, 1995). These BZ modulators will potentiate currents at receptors containing either α1, α2, or α3 subunit isoforms (Puia et al., 1991). While zolpidem shows a higher affinity for α1-containing receptors (for review, Sieghart, 1995), the efficacy of this BZ site modulator is comparable at α1-, α2-, and γ3-containing receptors (Puia et al., 1991; Smith et al., 2001). To determine if the mechanism of AAS modulation of GABAA receptor-mediated currents recapitulates that induced by the BZs, we have characterized AAS potentiation of α1-containing (α1β3γ2L) receptors and compared it to the modulation produced by the BZ site ligands, zolpidem and diazepam.

35 mm tissue culture dishes (BD Falcon, Franklin Lakes, NJ). Rat GABAA receptor cDNAs encoding the α1, β3, and γ2L subunits (all in pCDM8; kindly provided by Dr. Stefano Vicini, Georgetown University Medical Center, Washington, DC) were transiently transfected at a 1:1:1 ratio of 0.8 µg each using Lipofectamine PLUS (Gibco BRL). The pGreenLantern plasmid (Gibco BRL) was cotransfected to permit selection of transfected cells expressing the green fluorescent protein (GFP) under fluorescence optics. Previous studies have shown that greater than 90% of cells expressing GFP also expressed GABAA receptors (Zhu et al., 1996; Jorge-Rivera et al., 2000). Following transfection, cells were incubated for 24–48 hrs prior to electrophysiology. 2.1.1. Drugs and reagents Kynurenic acid was purchased from Tocris (Ballwin, MO), nandrolone from Steraloids (Newport, RI), and all other drugs from Sigma-Aldrich. Stock solutions of zolpidem (N,N-6-trimethyl-2-(4-methylphenyl)-imidazo [1,2-a]-pyridine-3 acetamide), diazepam (7-chloro-1,3dihydro-1-methyl-5-phenyl-2H-1,4-benzodiazepin-2-one), flumazenil (Ro15-1788; ethyl-8-fluoro-5,6-dihydro-5methyl-6-oxo-4H-imidazo[1,5-a]-[1,4]benzodiazepine3-carboxylate), and 17α-MeT (17α-methyl-4-androsten17β-ol-3-one) were made with dimethylsulfoxide (DMSO) as the solvent and were diluted accordingly to achieve a final bath concentration ⱕ 0.01% DMSO. 2.1.2. Electrophysiology 2.1.2.1. Recording conditions Patch pipettes were fabricated from borosilicate glass (Sutter Instrument Co., Novato, CA) with tip diameters of ~2 µm and resistances of 3–5 M⍀. Electronic compensation of ⬎60% was used to reduce the effective series resistance, typically 10–20 M⍀, and the time constant of membrane charging. All recordings were made at room temperature (~24 °C). Unless otherwise stated, all drugs were applied at 1 µM as used previously (Nett et al., 1999; Jorge-Rivera et al., 2000). For all experiments, the effects of each drug were determined for the same cell by comparison of average responses in the presence and absence of drugs.

2. Methods 2.1. Cell culture and transfections HEK293 cells (kindly provided by Dr. Lee Witters, Dartmouth Medical School, Hanover, NH) were incubated (37 °C; 5% CO2/95% air) in the following medium: Dulbecco’s Modified Eagle’s Medium, 10% fetal bovine serum, 20 mM L-glutamine, 50 IU/mL penicillin, and 50 µg/mL streptomycin. Culture reagents were purchased from Gibco BRL (Gaithersburg, MD), Sigma-Aldrich (St. Louis, MO), Mediatech (Herndon, VA), and Fisher (Fair Lawn, NJ). Prior to transfection, cells were re-plated to a minimum 50% confluency in

2.1.2.2. Ultrafast perfusion Ultrafast perfusion was performed on nucleated outside-out patches from transfected HEK293 cells voltage-clamped at –60 mV. GABA was applied to the patch using the LSS-3100 High Speed Positioning System (Burleigh Instruments Inc., Fishers, NY) and a double-barreled theta glass (Sutter Instruments Co.; tip diameter 100–120 µm) as described previously (Nett et al., 1999; Jorge-Rivera et al., 2000). Open tip currents (Lester and Jahr, 1992) showed that the 10–90% of the peak on and off response was achieved in less than 1 ms. The bath was perfused at 2–3 mL/min and theta tube solution flow was stopped between patches. All drugs were pre-equilibrated in both

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

streams from the theta tube. Recordings were made in external solution composed of the following (in mM): 142 NaCl, 8 KCl, 6 MgCl2, 1 CaCl2, 10 HEPES, and 10 dextrose (pH 7.3–7.4). Internal pipette solution was composed of the following (in mM): 153 CsCl, 1 MgCl2, 2 Mg–ATP, 10 HEPES, and 5 EGTA (pH 7.3). The chloride reversal potential was ~0 mV. AAS and BZ modulation of peak current (Ipeak) and current deactivation was assessed for responses elicited by brief (4 ms and 18 ms) pulses of either a low (20 µM) or a high (1 mM) concentration of GABA. Desensitization was assessed in responses elicited by 200–3000 ms steps of 1 mM GABA with up to 300 s intervals between each pulse. Five to ten currents for 4 ms pulses and ⬎3 currents for steps were averaged from each cell. A paired pulse paradigm (Jones and Westbrook, 1995) was used to assess recovery from desensitization. Pairs of 4 ms pulses were applied to the patch with progressive interpulse intervals (IPI) of 25, 50, 100, 200, 400, 800, 1600, 3200, and 6400 ms. Trains of paired pulses were performed twice per drug condition. 2.1.2.3. Bath application protocols For experiments examining the interacting effects of AAS and BZs on currents induced by tonic exposure to GABA, HEK293 cells were transfected as described above. After 48 hours, the transfected cells were replated at low density and 5–8 hours later isolated fluorescent cells were patched. During recordings, the culture dish was perfused with a bath solution composed of the following (in mM): 130 NaCl, 6 KCl, 2 MgCl2, 1 CaCl2, 10 HEPES, 10 dextrose, and 10 sucrose (pH 7.4). Recordings were made in the whole-cell configuration with an internal solution containing the following (in mM): 140 CsCl, 1 MgCl2, 10 HEPES, 5 EGTA, and 2–4 Mg– ATP (pH 7.2). This combination of external and internal solution provided a chloride reversal potential of ~0 mV. Drugs were applied to the bath by gravity flow from a series of manually controlled reservoirs connected to a low-volume perfusion manifold (Automate Scientific, San Francisco, CA). Dish fluid volume was kept low (400–500 µL) to minimize solution exchange time (~20 s). 2.1.2.4. Cerebellar slices Female Balb/C mice were purchased from Charles River Laboratories (Wilmington, MA). All animals were housed in a temperature- and light-controlled (12 hours dark; 12 hours light) facility with lights on at 0700 hours. All animal care procedures were approved by the Institutional Animal Care and Use Committee at Dartmouth. These protocols adhere to both the National Institutes of Health and the American Veterinary Medical Association guidelines to minimize pain and discomfort and to minimize the numbers of animals used. Following CO2 asphyxiation, mice were decapitated and whole brains

621

were quickly removed. Dissections of the cerebellum were made from postnatal day 7–9 animals and were performed in ice-cold solution composed of the following (in mM): 120 sucrose, 3.1 KCl, 26 NaHCO3, 1.25 K2HPO4, 2 CaCl2, 1 MgCl2, and 5 dextrose (pH 7.3– 7.4; saturated with 95% O2/5% CO2). Using a Campden Vibroslice microtome (Stoelting, Wood Dale, IL), 250 µm medial sagittal slices of the cerebellum were prepared and perfused with artificial cerebral spinal fluid (aCSF) at room temperature. The aCSF solution used for recording was identical to the dissecting solution except sucrose was replaced with 120 mM NaCl. Brain slices were viewed using an Olympus BX50 microscope equipped with infrared differential interference contrast optics and a Dage-MTI VE1000 CCD camera system (Optical Analysis Corp., Nashua, NH). Purkinje cells were identified based on cell size, shape, and position. Whole-cell configuration patch clamp recordings were made from Purkinje cells voltage-clamped at –80 mV. Cerebellar slices were perfused at 2–5 mL/min in the recording chamber with room temperature aCSF. The electrode pipette solution contained the following (in mM): 145 CsCl, 5 EGTA, 10 HEPES, and 5 Mg–ATP (pH 7.2). This combination of solutions provided a chloride reversal potential of ~0 mV. GABAA receptormediated miniature inhibitory postsynaptic currents (mIPSCs) were isolated from amino acid-mediated excitatory synapses by addition of 2 mM kynurenic acid and 1 µM tetrodotoxin (TTX) to the bath (Mozrzymas et al., 1999). To confirm that mIPSCs were mediated by GABAA receptors, 10 µM bicuculline was used to reversibly block the events. Recordings consisted of the following: minimum of 3 min of pre-drug control, 3 min of drug application, and 20 minutes of wash. Greater than 50 mIPSCs with rise times of less than 2 ms were collected from 3 min of each condition and averaged. 2.1.2.5. Data acquisition and analysis Data were acquired and analyzed using an EPC-7 or an EPC-9 patch clamp amplifier and HEKA Pulse software (Instrutech Corp., Great Neck, NY). Data were digitized at 50 kHz using Acquire (Bruxton Software, Seattle, WA) and HEKA Pulse. All data were low-pass filtered at 10 kHz for storage. Bath application currents were low-pass filtered at 300 Hz, digitized at 1 kHz for online storage, and averaging of adjacent points was used to improve the signal-to-noise ratio of the recordings. Ultrafast perfusion currents were digitally low-pass filtered at 3 kHz prior to analysis. HEKA Pulsefit and Mini Analysis (Jaejin Software, Leonia, NJ) were used for current averaging and curve fitting. The decays of averaged currents were fitted with exponential components (three for transfected HEK293 cells, one for Purkinje cells) using a simplex algorithm minimizing the root mean square (RMS) error with no restraints imposed on individual components. Previous studies have found that

622

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

current decays for GABAA receptor deactivation are well described by two (e.g., Jones and Westbrook, 1995), three (e.g., Banks and Pearce, 2000), or even as many as four (e.g., Maconochie et al., 1994) components. Although there is no general agreement on how to categorically determine the number of exponentials to apply (Colquhoun and Sigworth, 1995), three exponential fits were deemed most appropriate for currents elicited from HEK293 cells in the present experiments for the following reasons: that based on the algorithm indicated above and assessment by eye, three components unambiguously provided a better fit to the data than two components, and that for Markov processes, data is best described by the maximum number of components that can be justified by kinetic modeling and still be reproduced across multiple experiments (Colquhoun and Sigworth, 1995). Weighted time constants (τw) were calculated by summing the exponential components each weighted to its respective percent contribution to Ipeak. The total charge transfer or total area described by the curve (Atot) was estimated by integrating the current with respect to time. Paired pulse curves were described by two exponential time constants. Percent recovery for intervals of 25 and 50 ms were not included in the fit due to their variation as points of inflection. 2.1.2.6. Kinetic modeling Simulation of channel activity was used to determine kinetic transitions altered by AAS. Models were solved using Q-matrix techniques applied to non-stationary conditions (Colquhoun and Hawkes, 1977, 1981, 1995) using programs locally written (Brian L. Jones) in MatLab 6.1 (The Math Works, Natick, MA). Fitting was implemented by mapping experimental data to an open probability interval using a peak open probability (POMax) determined by nonstationary variance analysis (Sigworth, 1980; Silberberg and Magleby, 1993; Jones and Westbrook, 1995). Model parameters were optimized by minimizing the sum of squares of the residuals using a weighting scheme to nullify the systematic current–variance dependency of the residuals. Several currents elicited by different stimulation protocols (e.g., 20 µM GABA applied for 4 ms, 200 ms, and paired 4 ms pulses) were fitted simultaneously to maximize the dependency of goodness of fit on each of the model’s parameters. This method was found to be adept at recovering parameters from simulated data. Parameter optimization was achieved with a sequential quadratic programming method permitting the boundary value and equality constraints required to maintain thermodynamically feasible solutions. To rule out convergence on local minima, optimization was initiated from a number of starting points. This approach almost always yielded model parameterizations that were identical within termination tolerances. Calculations were performed using a 1.7 GHz Pentium IV-based system (Dell Computer Corp., Round Rock, TX).

The initial parameters used for modeling were based on previously published values (Jones and Westbrook, 1995) and estimates derived from single channel data. Closing rates were fixed to 2387 s–1 (a1) and 523 s–1 (a2) based on the principle open dwell time constants measured from single channel recording made under stationary low agonist concentrations (1 µM GABA) from α1β3γ2L receptors expressed in HEK293 cells (data not shown). Control and drug data were fitted assuming equivalent, independent GABA binding sites unless otherwise specified. The simultaneous fitting of multiple currents evoked by different stimulation protocols insured the parameters were optimized across different patterns of stimulations. This method improved the overall conditioning of the fitting problem leading to faster convergence on solutions and more robust parameter estimates. This was especially important for parameters such as off-rate (koff) that shaped not only the current deactivation elicited by short pulses of GABA, but also played an important role in shaping recovery in pairedpulse experiments. 2.1.2.7. Non-stationary fluctuation analysis Nonstationary fluctuation analysis was applied to estimate single-channel conductance as in Heinemann and Conti (1992). Briefly, 5–10 sweeps from currents elicited by 18 ms application of GABA were averaged and subtracted from each individual trace. The residual currents were divided into 15 to 50 bins sized so each bin contained equal decrements in current. The variance within each bin was calculated and pooled across the equivalent bins in all of the sweeps analyzed. Variance was plotted against mean current and this distribution fitted with the variance equation: I2 s2 ⫽ iI⫺ ⫹ s2base N

(1)

where i is the unitary current, I is the macroscopic current level, N is the number of channels, and σ2 base is the baseline variance. A weighted least-squares technique was used in data fitting to minimize violations of the least-squares assumptions by systematic current– variance relationships. Maximum open probability (POMax) was calculated by: POMax ⫽

IPeak iN

(2)

where Ipeak is the peak macroscopic current. 2.1.2.8. Statistical analysis Values are given as means ± SEM. For each cell, current trace parameters from drug conditions were normalized to control conditions and calculated as percent of control. The agreement of our data with the Student’s t test assumptions was verified by using normal quantile plots, and suspected outliers

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

623

were identified if they fell more than 1.5 interquartile ranges above the 3rd and below the 1st quartile. Results were qualitatively the same whether or not outliers were included in the final analysis. Two-tailed matched pairs t test was performed to determine significance with an α level of 0.05 and a null hypothesis of mean percent control = 100%. Two-tailed independent t test was performed with the same α level to determine significance between the responses of different combinations of drugs unless otherwise specified.

17α-MeT, both diazepam and zolpidem significantly increased Ipeak and %τ3 (Fig. 1D–I). However, in contrast to 17α-MeT, both BZ site compounds significantly increased the values of τ1, τ2, and τ3 (Fig. 1D–I). The overall effect on decay kinetics, as assessed by τw, was to significantly prolong GABA-induced responses for both zolpidem (τw = 187.9 ± 15.5% of control) and diazepam (τw = 176.6 ± 14.7% of control).

3. Results

Under synaptic conditions where receptor occupancy is believed to be maximal, modulators that alter channel affinity, such as the BZs, are predicted to induce no potentiation of Ipeak (Perrais and Ropert, 1999; Ha´ jos et al., 2000), but may still prolong current decays by slowing deactivation (Mellor and Randall, 1997). Likewise, modulators that facilitate entry into open states from those that are sparsely populated under high concentrations of agonist (e.g., singly liganded states) are expected to exhibit similar concentration dependence (Walters et al., 2000; Brian L. Jones, unpublished results). To test if AAS modulation is consistent with either of these mechanisms of action, responses were elicited from α1β3γ2L recombinant receptors by brief (4 ms) pulses of a high concentration (1 mM) of GABA (Ipeak = 301.1 ± 64.3 pA) believed to reflect that present in the synaptic cleft (Maconochie et al., 1994; Puia et al., 1994; Jones and Westbrook, 1996). The decay of these currents were best fitted by three exponential components with time constants of τ1 = 1.3 ± 0.2 ms (58.6 ± 3.6%), τ2 = 11.5 ± 1.3 ms (24.0 ± 1.9%), and τ3 = 241.6 ± 7.6 ms (17.4 ± 2.6%) with τw = 44.4 ± 5.7 ms. This concentration of GABA is thought to be saturating in cultured Purkinje cells (Maconochie et al., 1994) and near saturating (EC90) for recombinant α1β3γ2L receptors in HEK293 cells (Data not shown). The effects of 17α-MeT on these responses were compared to the effects of zolpidem. Neither class of modulator significantly increased average Ipeak values (AAS: 98.5 ± 4.8% of control) for responses elicited by 1 mM GABA (Fig. 2). In fact, in the presence of zolpidem, a small, but significant, decrease in Ipeak (86.0 ± 2.7% of control) was observed (Fig. 2D), which may reflect enhanced receptor desensitization (Jones and Westbrook, 1995; Mellor and Randall, 1997; Zhu and Vicini, 1997). In addition, zolpidem and 17α-MeT had different effects on current decay kinetics (deactivation). Specifically, zolpidem increased the values of all three time constants as reflected in the increase in the weighted time constant (τw = 169.1 ± 11.4% of control; Fig. 2C and D), while 17α-MeT had no effect on either the time constants of deactivation or the percentage of current decay attributed to each component (τw = 109.0 ± 5.1% of control; Fig. 2A and B). The prolongation of current decay induced

Unless otherwise specified, all data were collected from HEK293 cells transiently transfected with rat α1, β3, and γ2L subunit cDNAs. 3.1. AAS modulation of responses elicited by brief pulses of low GABA concentration Currents were elicited by brief (4 ms) pulses of 20 µM GABA (Ipeak = 104.2 ± 23.4 pA), a concentration that corresponds to EC20 in these experiments (data not shown). Least squares analysis indicated that current decays were best fitted by three exponential components with time constants of τ1 = 2.8 ± 0.2 ms (51.1 ± 1.8%), τ2 = 17.4 ± 1.2 ms (37.0 ± 1.9%), and τ3 = 240.8 ± 16.2 ms (11.5 ± 1.0%) with τw = 34.8 ± 2.8 ms. Exposure to 1 µM 17α-MeT significantly and reversibly increased total charge transfer (Atot) to 133.8 ± 7.6% of control, which could be attributed to a significant increase in Ipeak (128.4 ± 6.9% of control). With respect to current decay kinetics, exposure to 1 µM 17α-MeT significantly decreased the mean value of τ3, but also significantly increased the percentage of the current attributable to this slowest component of decay (%τ3) (Fig. 1A–C) such that the there was no overall effect of 17α-MeT on the kinetics of current decay, as indicated by a combined and weighted single time constant, τw (104.2 ± 6.7% of control). 3.2. BZ modulation of responses elicited by brief pulses of low GABA concentration The AAS have been shown to elicit qualitatively similar profiles of allosteric modulation as compounds acting at the BZ site for both native and recombinant α2β3γ1 or 2 receptors (Nett et al., 1999; Jorge-Rivera et al., 2000). Responses elicited from recombinant α1β3γ2L receptors by 4 ms pulses of 20 µM GABA were reversibly potentiated by both diazepam and the BZ type I modulator, zolpidem (Fig. 1D-I), although the enhancement induced by zolpidem (Atot = 634.7 ± 149.6%; Fig. 1F) was significantly greater than that elicited by diazepam (Atot = 239.3 ± 18.4% of control; Fig. 1I). Like

3.3. AAS and BZ modulation of responses elicited by brief pulses of high GABA concentration

624

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

Fig. 1. Effects of AAS and BZ site compounds on currents elicited by brief pulses of 20 µM GABA from recombinant α1β3γ2L GABAA receptors. (A) Representative currents demonstrating reversible potentiation of Ipeak by 1 µM 17α-MeT for currents elicited by 4 ms pulses of 20 µM GABA from a nucleated outside-out patch. (B) Responses shown in (A) normalized to equivalent Ipeak demonstrating a notable, but not significant, prolongation of current decay in the presence of 17α-MeT. (C) Bar graphs for analyzed data indicating relative changes induced by 17α-MeT in the individual time constants of deactivation (τ1–τ3), τw, percent of the Ipeak attributed to each kinetic component of decay (%τ1–%τ3), Ipeak, and Atot. Averaged data are expressed as a percent of control values with control set to 100% (dotted line). Asterisks indicate values that were significantly (p⬍0.05) different from control. All data in the presence of drug were compared to responses in the same cell in the absence of drug (n = 18 cells). (D) Representative currents demonstrating reversible potentiation of Ipeak by 1 µM zolpidem for currents elicited by 4 ms pulses of 20 µM GABA on a nucleated outside-out patch. (E) Responses shown in (D) normalized to equivalent Ipeak demonstrating a significant prolongation of current decay in the presence of zolpidem. (F) Bar graphs for analyzed data indicating relative changes in current amplitude and current decay parameters (same as in C) induced by 1 µM zolpidem (n = 9 cells). (G) Representative currents demonstrating reversible potentiation of peak current amplitude by 1 µM diazepam for currents elicited by 4 ms pulses of 20 µM GABA from a nucleated outside-out patch. (H) Responses shown in (G) normalized to equivalent Ipeak demonstrating a significant prolongation of current decay in the presence of diazepam. (I) Bar graphs for analyzed data indicating relative changes in current amplitude and current decay parameters (same as in C) induced by 1 µM diazepam (n = 7 cells).

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

625

Fig. 3. AAS do not modulate mIPSCs recorded from cerebellar Purkinje neurons. Representative currents demonstrating a small, but not significant, increase in Ipeak for averaged mIPSCs recorded from a Purkinje neuron of a postnatal day nine mouse by nandrolone, an AAS shown to induce modulation of GABAA receptor-mediated currents comparable to that induced by 17α-MeT (Jorge-Rivera et al., 2000). (B) Bar graph for analyzed data indicating absence of changes induced by AAS for mIPSCs. mIPSC decays were fitted by a single time constant (τ) which was 8.48 ± 0.59 ms and 8.57 ± 0.52 ms in the absence and presence of AAS, respectively (data plotted as described for Fig. 1C; n = 10 cells).

Fig. 2. Effects of 17α-MeT and zolpidem on currents elicited by brief pulses of 1 mM GABA from recombinant α1β3γ2L GABAA receptors. (A) Representative currents demonstrating lack of effect on either Ipeak or current decay by 1 µM 17α-MeT for currents elicited by 4 ms pulses of 1 mM GABA from a nucleated outside-out patch. (B) Bar graphs for analyzed data indicating absence of changes induced by 17α-MeT for responses elicited by 1 mM GABA (data plotted as described for Fig. 1C). All data in the presence of drug were compared to responses in the same cell in the absence of drug (n = 9 cells). (C) Representative currents demonstrating prolongation of current decay by 1 µM zolpidem for currents elicited by 4 ms pulses of 1 mM GABA from a nucleated outside-out patch. (D) Bar graphs for analyzed data indicating 1 µM zolpidem increased τ1–τ3, τw, Ipeak, and Atot for responses elicited by 1 mM GABA (data plotted as in Fig. 1C; n = 7 cells).

by zolpidem was sufficient to produce an overall significant potentiation (Atot = 155.7 ± 9.4% of control) of responses elicited by 1 mM GABA in spite of the diminution of Ipeak (Fig. 2D). In contrast, because 17α-MeT had no significant effect on any of the response parameters, there was no overall effect (Atot = 98.4 ± 6.6% of control). Consistent with the lack of modulation induced by AAS for responses elicited by 1 mM GABA applied directly to these α1-containing recombinant receptors in HEK293 cells, no significant effect of AAS on either Ipeak or current decay was observed for mIPSCs recorded from cerebellar Purkinje neurons (Fig. 3) that are believed to express exclusively α1β2/3γ2 receptors (Wisden et al., 1992). 3.4. AAS modulation of desensitization and recovery from desensitization Benzodiazepine site agonists have been shown to enhance desensitization of receptors exposed to long

steps of high concentrations of GABA, and because the desensitized state has a higher affinity for GABA that prolongs channel unbinding (Jones and Westbrook, 1995; Bianchi et al., 2001), BZ site ligands that enhance desensitization also prolong deactivation following the end of the step (Bianchi and Macdonald, 2001). To assess if AAS have effects similar to BZ site agonists on receptor desensitization, long steps of 1 mM GABA were used to determine if 17α-MeT altered desensitization during the step and deactivation following the step, and paired pulses were used to determine if recovery from desensitization was altered. Responses elicited by step applications of 1 mM GABA, even in the absence of modulators, were susceptible to rundown, as has been reported previously (McClellan and Twyman, 1999), and currents in most cells (46 of 50), irrespective of pulse duration (between 200 ms and 3000 ms), pulse interval (30–300 s), and multiple variations of the internal solution, showed a continuous acceleration of desensitization over time even in the absence of modulator. Because of this instability in the response to GABA alone in most cells, population averages of current amplitudes and kinetic parameters were not compiled and compared for data acquired in the presence and absence of modulators. However, in those patches where currents were stable, 17α-MeT did not appear to alter either the fast or slow components of desensitization (Fig. 4A and B) or deactivation following the step (Fig. 4A and C). In contrast to the responses elicited by long step applications of 1 mM GABA, run down was not evident during the paired pulse protocol used to assess recovery from desensitization. In this paradigm, comparisons of Ipeak were made between the first and the

626

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

and τr1 = 51.1 ms (32.4%) and τr2 = 1059.0 ms (67.6%) in the presence of 1 µM zolpidem. Together, these data suggest 17α-MeT had no effect on either desensitization or recovery from desensitization for α1β3γ2L receptors. 3.5. AAS and BZ interactions in the modulation of responses to tonic applications of low GABA concentrations

Fig. 4. AAS do not modulate desensitization or recovery from desensitization for recombinant α1β3γ2L GABAA receptors. (A) Currents elicited by a step application (500 ms) of 1 mM GABA in the presence or absence of 1 µM 17α-MeT. The AAS had no appreciable effect on Ipeak, desensitization during the step, or deactivation following the step. Currents in (B) and (C) are from the current traces in (A) and have been peak scaled and expanded to show that there is no difference in (B) the time course of desensitization or (C) in deactivation following the end of the GABA step in the absence or presence of 17α-MeT. (D) Comparison of the recovery time course of the second response in the paired pulse protocol for exposure to 1 µM 17α-MeT (n = 10). Brief (4 ms) pulses of 1 mM GABA were applied with variable interpulse intervals (see Methods). Percent recovery from desensitization (% Recovery), as assessed by (Ipeak2⫺onset2) / (Ipeak1–onset1) x 100, was plotted as a function of interpulse interval and described by two exponential components with time constants, τr1 and τr2. Inset shows currents elicited by paired pulse paradigm from a representative experiment. (E) Percent recovery from desensitization as shown in (D) but for exposure to 1 µM zolpidem (n = 4). Exposure to the BZ site agonist slowed recovery from desensitization.

second response in the pair of brief (4 ms) pulses of 1 mM GABA presented at incremental intervals of 25 to 6400 ms, and the percent recovery from desensitization was determined according to Jones and Westbrook (1995). The time course of recovery from desensitization was described by two components with time constants, τr1 = 83.6 ms (57.4%) and τr2 = 1054.4 ms (42.6%). Coexposure to 1 µM 17α-MeT was found to have no effect on recovery from desensitization. Time constants in the presence of the AAS were τr1 = 94.4 ms (59.5%) and τr2 = 1119.2 ms (40.5%) with no significant change in the contribution of each component to the recovery process (Fig. 4D). The lack of effect of 1 µM 17α-MeT on recovery from desensitization was in contrast to the effects of 1 µM zolpidem, which induced a small slowing of this process (Fig. 4E). Time constants for the recovery from desensitization were τr1 = 30.8 ms (25.5%) and τr2 = 769.9 ms (74.5%) in control conditions

To assess pharmacologically if BZs and AAS act at a common site, 17α-MeT, zolpidem, and the BZ site antagonist, flumazenil (Hunkeler et al., 1981; Sigel and Baur, 1988), were applied in different combinations to examine modulation of GABA-activated currents elicited by stationary (bath) application of a low (1 µM) concentration of GABA. Bath application of 1 µM GABA elicited small sustained currents (average Isus = 114.6 ± 16.6 pA) (Fig. 5A) that were potentiated by subsequent coapplication of 1 µM zolpidem (Fig. 5B and 6A). In the continued presence of 1 µM GABA and 1 µM zolpidem, subsequent application of 1 µM 17α-MeT induced further potentiation, suggesting the effects of zolpidem and 17α-MeT were additive (Fig. 5B and Fig. 6A). Although the responses of the two classes of modulators were additive, analysis of averaged cumulative responses indicated that the potentiation elicited by 1 µM 17α-MeT in the presence of 1 µM zolpidem was less than that induced by 1 µM 17α-MeT alone (Fig. 6A). When 1 µM 17α-MeT was co-applied with 10 µM zolpidem, potentiation by the steroid, while still significant (Fig. 6A), was severely curtailed with some cells exhibiting no additional potentiation (Fig. 5C). The lack of subsequent potentiation induced by 17αMeT in the presence of high concentrations of zolpidem could arise if both agonists were competing for the same site or, even if acting at separate sites, were acting to accelerate transitions along the activation pathway (see Discussion). Application of 10 µM 17α-MeT following application of 1 µM zolpidem actually diminished the potentiation induced by zolpidem (data not shown). However, this antagonistic effect was nonspecific as this high concentration of 17α-MeT also decreased potentiation induced by the neurosteroid, allopregnanolone (1 µM; data not shown). This nonspecific effect was consistent with previous results demonstrating that while 1 µM 17α-MeT induced maximal potentiation of GABAA receptor-mediated currents from hypothalamic neurons, 10 µM was without effect (Jorge-Rivera et al., 2000). The BZ site antagonist, flumazenil (1 µM), was found to significantly antagonize potentiation elicited by 1 µM zolpidem (Fig. 6B). However, when 1 µM 17α-MeT was applied in the presence of 1 µM flumazenil, potentiation was comparable to that induced by 17α-MeT alone (Fig. 6B). Similarly, flumazenil did not block potentiation elicited by 1 µM 17α-MeT for responses elicited by brief pulses of 20 µM GABA (data not shown).

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

Fig. 5. Effects of AAS and BZ site compounds on currents elicited by bath application of 1 µM GABA to recombinant α1β3γ2L GABAA receptors. Representative whole-cell current demonstrating that the response to a prolonged application of 1 µM GABA was potentiated by concomitant application of 1 µM 17α-MeT. (B) Representative whole-cell current demonstrating that the response to a prolonged application of 1 µM GABA was potentiated by concomitant application of 1 µM zolpidem that both produced a transient increase in the sustained current (Isus) and induced a variable acceleration of desensitization (extent was modest in this cell). Subsequent additional exposure to 1 µM 17α-MeT induced further augmentation of Isus. (C) Representative whole-cell current demonstrating that the response to a prolonged application of 1 µM GABA was potentiated by concomitant application of 10 µM zolpidem (hatched bar) augmented Ihold (note differences in scale bars in A–C), but that subsequent addition of 1 µM 17α-MeT in the continued presence of 10 µM zolpidem induced no further potentiation of Isus or increase in desensitization.

3.6. Kinetic modeling and non-stationary fluctuation analysis of AAS modulation Computer simulations were based on the seven-state model proposed by Jones and Westbrook (1995, 1996) (Fig. 7) and were made for currents elicited by several different pulse durations and GABA concentrations in

627

Fig. 6. Effects of zolpidem or flumazenil on potentiation induced by 1 µM 17α-MeT for currents elicited by steady state application of 1 µM GABA from recombinant α1β3γ2L GABAA receptors. Bar graph for analyzed data indicating the percent potentiation in the amplitude of Isus elicited by stationary exposure to 1 µM GABA and 1 µM 17αMeT (AAS; n = 6 cells) alone or in conjunction with either 1 µM (AAS+1 µM zolp.; n = 5 cells) or 10 µM zolpidem (AAS+10 µM zolp.; n = 7 cells). (B) Bar graph for analyzed data indicating the effects of flumazenil on the potentiation of currents induced by stationary exposure to 1 µM GABA and concomitant exposure to 1 µM zolpidem (zolp.; n = 9 cells) or 1 µM 17α-MeT (AAS; n = 6 cells). In the presence of 1 µM flumazenil, potentiation induced by 1 µM zolpidem was significantly reduced (zolp.+flum.; n = 3 cells). However, exposure to this BZ site antagonist had no effect on the potentiation induced by 1 µM 17α-MeT (AAS+flum.; n = 3 cells). Data are shown normalized to the peak response to agonist alone (either zolpidem or 17α-MeT; set to 100% as indicated by dotted line). Asterisks indicate values that were significantly (p⬍0.05) different from control. One-tailed independent t test was used to determine significance of the inhibition of 17α-MeT-evoked potentiation by 10 µM zolpidem.

628

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

Fig. 7. Quantitative model of GABAA receptor gating. The model is a minor modification of the seven-state Markov model of Jones and Westbrook (1995) and assumes two equivalent agonist binding sites for the receptor. This yields one unbound state (UB), and two bound states (B1 and B2). From the bound states there are also transitions to singly and doubly liganded open (O1 and O2, respectively) and desensitized (D1 and D2, respectively) states. This cartoon illustrates agonist binding and hypothetical conformational changes associated with each kinetic state. The rates for the model were (in s–1): kon = 8.18 x 106 M–1, koff = 50, a1 = 2387, a2 = 523, b1 = 874, b2 = 2559, d1 = 222, d2 = 239, r1 = 36, and r2 = 10. This combination of parameters gives a POMax of 0.7.

the presence and absence of 1 µM 17α-MeT which were fitted simultaneously in order to maximize the quality of the parameter estimates. Although the best quantitative fits to control data were provided by assuming cooperativity between the two GABA binding sites, the assumption of independent binding provided a reduction in the number of free parameters in the model leading to more robust solutions while not qualitatively affecting our conclusions. Consequently, unless otherwise specified, data presented here were modeled with two independent GABA binding sites. The Jones–Westbrook model (Fig. 7) provides a surplus of adjustable rates for modeling modulation by 17αMeT, however, a large number of these rate constants are not viable candidates for simulating 17α-MeT effects in light of the known actions of the AAS determined by ultrafast perfusion experiments. Specifically, 1 µM 17αMeT did not affect receptor desensitization or recovery from desensitization, nor were deactivation kinetics dramatically altered. Increases in the desensitization rates (d1 and d2) and recovery rates (r1 and r2) cannot produce the appropriate changes in Ipeak without producing large changes in desensitization and recovery. Likewise, large decreases in channel closing rates (a1 and a2) that

allowed adequate fits to Ipeak also significantly slowed current deactivation. Changes in the rate opening into the doubly liganded open state (b2) are effective at modulating peak current, but also significantly prolonged deactivation. A notable feature of the modulation of GABA-evoked currents by 17α-MeT is that the level of modulation depends on the concentration of GABA itself. This immediately implicates rate constants (kon and koff) involved in determining the affinity of the receptor for GABA. Changes in kon and koff that increase receptor affinity are most effective at potentiating current at low concentrations of GABA. Modulation of the rate constant (b1) governing transition to the singly liganded open state (O1) can also produce potentiation that accurately predicts our data. The effects of changes in b1 will be most prominent at low concentrations of GABA when receptors in singly liganded states are most abundant. Consistent with effects reported for 4 ms pulses of 20 µM GABA, 17α-MeT was found to reversibly potentiate Ipeak of responses elicited by longer (18 ms) applications of 20 µM GABA (Fig. 8 A and B). Non-stationary fluctuation analysis demonstrated that exposure to 17α-MeT did not alter either the estimated single-channel conductance or POMax, but did lead to an apparent increase in the number of active channels (Fig. 8C and D). Computer simulations indicated that the effects of 17α-MeT were optimally reproduced by altering the kinetics governing channel opening (b1) to the singly liganded open state (O1) or concomitantly increasing kon and decreasing koff . Changes in b1 provided the best fit to the data under a range of experimental conditions. Additional simulations evoking positive cooperativity between the two GABA binding sites were not able to accurately reproduce the lack of effect on the fast components of deactivation and the differences in deactivation observed between high and low concentrations of GABA while simulating the potentiation of Ipeak (Fig. 9). 4. Discussion We have previously demonstrated that both 17αalkylated AAS (17α-MeT and stanozolol) and 19-nortestosterone AAS (nandrolone) can induce rapid and reversible allosteric modulation of native GABAA receptors in rodent forebrain neurons and of recombinant α2β3γ1 or 2 recombinant receptors expressed in HEK293 cells (Jorge-Rivera et al., 2000). Surprisingly, whereas all three AAS potentiated responses from α2β3γ2L receptors, these steroids antagonized responses from α2β3γ1 receptors. Previous studies have demonstrated that both recombinant and native α2βxγ1 receptors show a similar paradoxical pattern of modulation by zolpidem that has been attributed to the presence of a γ1 versus a γ2 subunit (Bormann and Kettenmann, 1988; Puia et al., 1991; Nett

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

Fig. 8. Non-stationary fluctuation analysis of AAS modulation of currents elicited from recombinant α1β3γ2L GABAA receptors. Representative individual (gray trace) and averaged (black trace; n=10) currents elicited by an 18 ms pulse of 20 µM GABA in the absence (A) control or (B) presence of 1 µM 17α-MeT demonstrating an increase in Ipeak elicited by exposure to this AAS. (C) Variance vs. mean current plots (σ2–I) determined from non-stationary fluctuation analysis of simulated GABA-elicited responses for currents elicited in (A) and (B). Individual points represent variance associated with each amplitude bin in control conditions and in the presence of 1 µM 17α-MeT, demonstrating that exposure to this modulator increases N, the number of activated channels, without changing single-channel conductance (γ). (D) Bar graph of variance parameters derived from non-stationary fluctuation analysis demonstrating that exposure to 1 µM 17α-MeT caused a significant increase in Ipeak and the number of channels activated by GABA (N), but had no effect on either single-channel conductance (γ) or POMax.

et al., 1999). Specifically, whereas diazepam potentiates both α2βxγ1 and α2βxγ2 receptors, zolpidem acts as an inverse agonist (or is without effect) at α2βxγ1 receptors. That both zolpidem and the AAS show this same profile of modulation and that the direction of modulation depends upon γ1 versus γ2 subunit composition led us to speculate that the AAS may be acting at the BZ modulatory site (Jorge-Rivera et al., 2000). The experiments presented here were performed both to characterize the biophysical mechanism by which AAS induce allosteric modulation of GABAA receptors and to garner information that would support or refute the hypothesis that these steroids are eliciting their effects by acting at the BZ site. Because the profile of AAS modulation mirrors that produced by zolpidem (Nett et al., 1999; Jorge-Rivera et al., 2000) and zolpidem is known to be selective for α1βxγ2 receptors (for review, Sieghart, 1995), we chose to perform these studies for α1β3γ2L receptors.

629

Fig. 9. Kinetic modeling of AAS modulation of currents elicited from recombinant α1β3γ2L GABAA receptors. Experimental data (i) and computer simulations (ii) are shown for representative responses elicited by either a brief (4 or 18 ms) pulse of 20 µM GABA (A and B) or a 4 ms pulse of 1 mM GABA (C). In each section, immediately below each trace are peak scaled versions. Computer simulations in which kinetic parameters were altered indicated that the best fits to the data were obtained by both increasing kon and decreasing koff or by increasing b1, the rate constant governing transition into O1.

4.1. AAS and BZ demonstrate different profiles of modulation at the GABAA receptor The AAS, 17α-MeT, and the two BZ site agonists, zolpidem and diazepam, were all shown to potentiate currents elicited by brief pulses of micromolar concentrations of GABA from recombinant α1β3γ2L receptors by significantly increasing Ipeak and Atot. Both BZ site compounds significantly prolonged GABA-induced currents, as well as enhanced Ipeak, whereas 17α-MeT had no effect on overall current decay with respect to τw. Similarly, zolpidem was found to significantly prolong the decay of currents elicited by brief application of 1 mM GABA while AAS had no effect on Ipeak or current decay for responses elicited by brief application of 1 mM GABA or on synaptic currents mediated by α1β2/3γ2 receptors expressed in cerebellar Purkinje neurons. The prolongation of current decay by zolpidem for responses

630

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

elicited by millimolar concentrations of GABA has been reported previously for α1β2γ2 recombinant receptors (Krampfl et al., 1998) and for synaptic currents from Purkinje cells (Griebel et al., 1999). Thus, while both the AAS and the BZ site agonists were found to potentiate GABAA receptor-mediated responses in the current study, the effects of the two classes of modulators were divergent with respect to deactivation kinetics. The potentiation induced by 1 µM 17α-MeT was significantly less than that produced by either 1 µM zolpidem or 1 µM diazepam at α1β3γ2L receptors. We observed an Atot relative to control of 134% for 17αMeT, in contrast to 239% and 635% for diazepam and zolpidem, respectively for 20 mM GABA. It had been shown previously that the efficacy of zolpidem was comparable at α1β2γ2 and α1β2γ2 recombinant receptors (Puia et al., 1991; Smith et al., 2001), and we predicted that if 17α-MeT is acting at the BZ site, it should also induce comparable potentiation at α1β3γ2L and α2β3γ2L recombinant receptors. However, 17α-MeT was actually an appreciably weaker agonist at α1β3γ2L receptors than at α2β3γ2L receptors (for 1 mM GABA; Jorge-Rivera et al., 2000). These results suggest that while the AAS and the BZs share a common dependence on γ subunit composition, they have a different profile of effects with respect to α subunit composition than do the BZs. 4.2. AAS potentiation of GABAA receptor-mediated currents is simulated by changes in affinity or b1 The ability of BZ site agonists to increase Ipeak only under subsaturating concentrations of GABA is consistent with a mechanism in which these compounds enhance the affinity of the GABAA receptor by altering either kon (Rogers et al., 1994) or koff (Mellor and Randall, 1997) or both. Experiments performed here indicate that 17α-MeT did not induce changes in desensitization or recovery from desensitization, nor did simulations in which rates governing transitions into (d1 or d2) and out of (r1 or r2) desensitized states were altered provide appropriate fits to the data. Similarly, while AAS effects on Ipeak could be adequately modeled by changes in opening rate (b2), such simulated changes provided poor fits of current decay. Non-stationary fluctuation analysis indicated that 17α-MeT had no effect on single channel conductance but did increase the apparent number of bound receptors, consistent with AAS-induced increases in receptor affinity. However, non-stationary fluctuation analysis may not yield accurate measures of occupancy when applied to channels that exhibit singly liganded openings and long-lived desensitized states (Silver et al., 1996). The actions of 17α-MeT could be simulated by enhanced receptor affinity, specifically with concomitant increases in kon and decreases in koff. However, the best simulations were provided by increases in the opening rate, b1, that governs transitions into the singly liganded

open state. A mechanism by which 17α-MeT acts either to increase receptor affinity or to increase b1 is consistent with observations that this AAS induced no potentiation of responses from α1β3γ2L receptors exposed to millimolar concentrations of GABA where it is expected that activation is near maximal and most receptors reside in the doubly liganded states. Although kinetic simulations of AAS effects were consistent with changes in receptor affinity, the mechanism by which BZs act, our data were more accurately described by changes in b1. Moreover, pharmacological experiments in which BZ site agonists and 17α-MeT were co-applied do not categorically support the hypothesis that the AAS are acting at the high affinity BZ site. Specifically, we found that potentiation induced by 1 µM 17α-MeT was additive with that induced by 1 and 10 µm zolpidem, but that the potentiation decreased with increasing concentrations of zolpidem. This observation is consistent with both allosteric and competitive models of interaction between these sites, as has been suggested previously by Masonis and McCarthy (1995), but it is also consistent with the expected interaction between the kinetic effects elicited during AAS and BZ modulation. In particular, the increases in affinity associated with BZ site ligands would reduce the population of singly liganded receptors and consequently attenuate the potentiation produced by AAS if these steroids act either to enhance b1 or to increase affinity. The most compelling evidence that the AAS are not directly binding at the high affinity BZ site is that 1 µM flumazenil, a concentration of this compound that diminished the potentiation elicited by zolpidem by 81%, had no significant effect on blocking potentiation elicited by 17α-MeT. 4.3. AAS do not act like neurosteroids If AAS are not acting at the BZ site, do they interact with the neurosteroid binding site? This and prior studies indicate that this is not likely. First, the 3α-hydroxyl moiety common to all active neurosteroids (for review, Lambert et al., 1995, 2001) is not present in any of the AAS shown to elicit allosteric modulation of GABAA receptors. Second, the ability of the neurosteroids to elicit potentiation or antagonism does not depend upon γ subunit composition, as is the case for AAS (for review, Lambert et al., 2001; Henderson and Jorge, in press). Third, the kinetic effects of AAS and the prototypical neurosteroid tetrahydrodeoxycorticosterone (THDOC) are in stark contrast. Specifically, if one compares the effects of THDOC and 17α-MeT for α1β3γ2L recombinant receptors, at micromolar concentrations of GABA, THDOC enhances desensitization and prolongs deactivation (Wohlfarth et al., 2002), whereas 17α-MeT did not. Moreover, at millimolar concentrations of GABA, THDOC was shown to inhibit peak current and prolong deactivation (Wohlfarth et al., 2002), whereas AAS were

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

without effect. Finally, unlike THDOC (Wohlfarth et al., 2002), AAS concentrations as high as 10 µM do not directly activate chloride currents (Jorge-Rivera et al., 2000; present study: data not shown). In summary, AAS do not mimic the effects of neurosteroids, are unlikely to interact with the same binding sites, and should be considered as a distinct class of allosteric modulators. 4.4. AAS may act at the low affinity BZ site or a novel modulatory site Whereas these results question the hypothesis that the AAS are acting at the high affinity BZ site, a second possibility is that the AAS are acting at a second low affinity BZ site recently described by Walters et al. (2000), which is not antagonized by flumazenil. Walters et al. (2000) showed that potentiation produced by activation of the low-affinity BZ site was evident at very low concentrations of GABA (EC3–EC8), but was abolished at higher concentrations of GABA (⬎EC16; ~8 µM). The possibility that the AAS are acting at this low affinity BZ site at first seems to be at odds with our finding that AAS induce potentiation at EC20 (20 µM) concentrations of GABA. However, Walters et al. (2000) observed potentiation through the low affinity site during long (~10 s) applications of GABA and diazepam to oocytes. Under stationary conditions, singly liganded openings make an appreciable contribution to the total current only when GABA concentration is very low (⬍EC10). However, during rapid application of higher concentrations of GABA, these singly liganded openings will make a significant contribution to receptor activation and would be reflected in a reduction in first latency and thus an increase in Ipeak, as we have observed for AAS on currents elicited by 4 ms pulses of 20 µM GABA. Such an effect would be obscured by slow solution exchange of higher concentrations of GABA as was performed by Walters et al. (2000). Potentiation by the low affinity BZ site does not depend upon γ subunit composition, and diazepam can elicit modest potentiation of α1β2 receptors (Walters et al., 2000). Consistent with this observation, we find that both 17α-MeT and zolpidem can elicit similar potentiation at α1β3 receptors (data not shown). Interestingly, residues linked to low affinity BZ site activity have been implicated in the activity of a number of structurally diverse compounds including anesthetics (Belelli et al., 1997), barbiturates (Belelli et al., 1999), and loreclezole (Wingrove et al., 1994). Although these data suggest that AAS may elicit some of their effects by acting at the low affinity BZ site, they do not readily explain the following: the reversal of potentiation observed with the γ1 to γ2 substitution, that the efficacy of 17α-MeT is significantly greater for α2β3γ2L than α1β3γ2L receptors, or that 1 µM 17α-MeT induces significant potentiation of

631

currents elicited from α2β3γ2L receptors exposed to 1 mM concentration of GABA (Jorge-Rivera et al., 2000). At the present time, our results do not support the hypothesis that the AAS are acting at either the high affinity BZ or neurosteroid sites. However, the structural complexity of these modulators is greater than what is suggested simply by the presence of low and high affinity BZ-binding sites. For example, there are data indicating that separate structural elements are required for transducing the effects of positive and negative BZ modulators. Specifically, it is known that residues required for the potentiating effects of BZ agonists are not necessary for the action of β-carbolines, compounds that act as BZ inverse agonists (Boileau and Czajkowski, 1999). Interactions of AAS with these negative BZ modulators have not been explored. Thus, although our data may reflect AAS actions at the low affinity BZ site, it is also possible that AAS act at a unique site, but one that shares common elements known to be required either for the binding of BZs to the high affinity site or transduction of BZ effects. Studies employing mutant and chimeric receptors have been valuable for identifying structural elements that are necessary for BZ potentiation (Buhr et al., 1997; Renard et al., 1999; Rudolph et al., 1999; Walters et al., 2000; Teisse´ re and Czajkowski, 2001), and these mutant receptors may prove useful in identifying residues required for either AAS binding or coupling of AAS binding to the observed potentiation of GABA-elicited responses. 4.5. Implications of GABA concentration-dependent modulation by AAS Our finding that 17α-MeT is active at α1β3γ2L receptors at low, but not high, concentrations of GABA suggests mechanisms by which AAS may exert their effects on the CNS. Specifically, in regions where this receptor isoform predominates and synaptic concentrations of GABA are believed to be saturating (e.g., in the cerebellum, Puia et al., 1994) it is more likely that AAS would potentiate tonic currents evoked by low background concentrations of GABA than synaptic currents. This type of tonic inhibition has been shown to alter information processing in the cerebellum (Rossi and Hamann, 1998; Brickley et al., 2001; Hamann et al., 2002). Both the δ and α6 subunits have been implicated as playing an important role in mediating tonic responses to low concentrations of GABA (Rossi and Hamann, 1998; Brickley et al., 2001). Since these subunits also confer insensitivity to BZs (for review, Mehta and Ticku, 1999), it will be of interest to determine if AAS modulate these receptors. It should also be noted that while modulation of tonic inhibition is a likely avenue for AAS activity at α1β3γ2L-containing receptors, it has been previously shown that AAS can significantly modulate IPSCs in the forebrain where α2-containing receptors predominate

632

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

and the synaptic concentration of GABA is hypothesized to be subsaturating (Jorge-Rivera et al., 2000). Therefore, AAS may induce different functional effects in different brain regions depending on both the subunit composition of receptors expressed and the concentration of GABA available at both synaptic and extrasynaptic receptors.

Acknowledgements This work was supported by the NIH (DA/NS14137 and DA14216 to LPH and DA06079 to PY). We thank Dr. Stefano Vicini for providing us with GABAA receptor subunit cDNAs and Dr. Mark Fry for critical review of the manuscript.

References Banks, M.I., Pearce, R.A., 2000. Kinetic differences between synaptic and extrasynaptic GABAA receptors in CA1 pyramidal cells. Journal of Neuroscience 20 (3), 937–948. Belelli, D., Lambert, J.J., Peters, J.A., Wafford, K., Whiting, P.J., 1997. The interaction of the general anesthetic etomidate with the γ-aminobutyric acid type A receptor is influenced by a single amino acid. Proceedings of the National Academy of Science USA 94 (20), 11031–11036. Belelli, D., Pau, D., Cabras, G., Peters, J.A., Lambert, J.J., 1999. A single amino acid confers barbiturate sensitivity upon the GABA ρ1 receptor. British Journal of Pharmacology 127 (3), 601–604. Bianchi, M.T., Haas, K.F., Macdonald, R.L., 2001. Structural determinants of fast desensitization and desensitization–deactivation coupling in GABAA receptors. Journal of Neuroscience 21 (4), 1127– 1136. Bianchi, M.T., Macdonald, R.L., 2001. Agonist trapping by GABAA receptor channels. Journal of Neuroscience 21 (23), 9083–9091. Boileau, A.J., Czajkowski, C., 1999. Identification of transduction elements for benzodiazepine modulation of the GABAA receptor: three residues are required for allosteric coupling. Journal of Neuroscience 19 (23), 10213–10220. Bormann, J., Kettenmann, H., 1988. Patch-clamp study of γ-aminobutyric acid receptor Cl– channels in cultured astrocytes. Proceedings of the National Academy of Science USA 85, 9336–9340. Brickley, S.G., Revilla, V., Cull-Candy, S.G., Wisden, W., Farrant, M., 2001. Adaptive regulation of neuronal excitability by a voltageindependent potassium conductance. Nature 409 (6816), 88–92. Buhr, A., Schaerer, M.T., Sigel, E., 1997. Residues at positions 206 and 209 of the α1 subunit of the γ-aminobutyric acidA receptors influence affinities for benzodiazepine binding site ligands. Molecular Pharmacology 52, 676–682. Colquhoun, D., Hawkes, A.G., 1977. Relaxation and fluctuations of membrane currents that flow through drug-operated channels. Proceedings of the Royal Society of London B Biological Sciences 199 (1135), 231–262. Colquhoun, D., Hawkes, A.G., 1981. On the stochastic properties of single ion channels. Proceedings of the Royal Society of London B Biological Sciences 211 (1183), 205–235. Colquhoun, D., Hawkes, A.G., 1995. Q-Matrix cookbook. In: Sakmann, B., Neher, E. (Eds.), Single-Channel Recording, second edition. Plenum Press, New York. Colquhoun, D., Sigworth, F.J., 1995. Fitting and statisical analysis of

single-channel records. In: Sakmann, B., Neher, E. (Eds.), SingleChannel Recording, second edition. Plenum Press, New York. Griebel, G., Perrault, G., Letang, V., Granger, P., Avenet, P., Schoemaker, H., Sanger, D.J., 1999. New evidence that the pharmacological effects of benzodiazepine receptor ligands can be associated with activities at different BZ (ω) receptor subtypes. Psychopharmacology 146, 205–213. Ha´ jos, N., Nusser, Z., Rancz, E.A., Freund, T.F., Mody, I., 2000. Cell type- and synapse-specific variability in synaptic GABAA receptor occupancy. European Journal of Neuroscience 12, 810–818. Hamann, M., Rossi, D.J., Attwell, D., 2002. Tonic and spillover inhibition of granule cells control information flow through cerebellar cortex. Neuron 33 (4), 625–633. Heinemann, S.H., Conti, F.F., 1992. Nonstationary noise analysis and application to patch clamp recordings. Methods in Enzymology 207, 131–148. Henderson, L.P., Jorge, J.-C., in press. Steroid modulation of GABAA receptors: CNS roles in reproduction, dysfunction and drug abuse. In: Maue, R.A., (Ed), Molecular Insights into Ion Channel Biology in Health and Disease. Elsevier, Amsterdam. Hunkeler, W., Mohler, H., Pieri, L., Polc, P., Bonetti, E.P., Cumin, R., Schaffner, R., Haefely, W., 1981. Selective antagonists of benzodiazepines. Nature 290, 514–516. Jones, M.V., Westbrook, G.L., 1995. Desensitized states prolong GABAA channel responses to brief agonist pulses. Neuron 15, 181–191. Jones, M.V., Westbrook, G.L., 1996. The impact of receptor desensitization on fast synaptic transmission. Trends in Neuroscience 19 (3), 96–101. Jorge-Rivera, J.-C., McIntyre, K.L., Henderson, L.P., 2000. Anabolic steroids induce region- and subunit-specific rapid modulation of GABAA receptor-mediated currents in the rat forebrain. Journal of Neurophysiology 83, 3299–3309. Krampfl, K., Lepier, A., Jahn, K., Franke, C., Bufler, J., 1998. Molecular modulation of recombinant rat α1β2γ2 GABA(A) receptor channels by diazepam. Neuroscience Letters 256 (3), 143–146. Lambert, J.J., Belelli, D., Hill-Venning, C., Peters, J.A., 1995. Neurosteroids and GABAA receptor function. Trends in Pharmacolological Science 16 (9), 295–303. Lambert, J.J., Harney, S.C., Belelli, D., Peters, J.A., 2001. Neurosteroid modulation of recombinant and synaptic GABAA receptors. International Review of Neurobiology 46, 177–205. Lavoie, A.M., Twyman, R.E., 1996. Direct evidence for diazepam modulation of GABAA receptor microscopic affinity. Neuropharmacology 35 (9/10), 1383–1392. Lester, R.A., Jahr, C.E., 1992. NMDA channel behavior depends on agonist affinity. Journal of Neuroscience 12 (2), 635–643. Lukas, S.E., 1996. CNS effects and abuse liability of anabolic–androgenic steroids. Annual Review of Pharmacology & Toxicology 36, 333–357. Maconochie, D.J., Zempel, J.M., Steinbach, J.H., 1994. How quickly can GABAA receptors open? Neuron 12, 61–71. Masonis, A.E.T., McCarthy, M.P., 1995. Direct effects of the anabolic/androgenic steroids, stanozolol and 17α-methyltestosterone, on benzodiazepine binding to the GABAA receptor. Neuroscience Letters 189 (1), 35–38. Masonis, A.E.T., McCarthy, M.P., 1996. Effects of androgenic/anabolic steroid stanozolol on GABAA receptor function: GABA stimulation of 36Cl– influx and 35S TBPS binding. Journal of Pharmacology and Experimental Therapeutics 279 (1), 186–193. McClellan, A.M.L., Twyman, R.E., 1999. Receptor response kinetics reveal functional subtypes of native murine and recombinant human GABAA receptors. Journal of Physiology (London) 515.3, 711–727. Mehta, A.K., Ticku, M.K., 1999. An update on GABAA receptors. Brain Research Reviews 29 (2-3), 196–217.

P. Yang et al. / Neuropharmacology 43 (2002) 619–633

Mellor, J.R., Randall, A.D., 1997. Frequency-dependent actions of benzodiazepines on GABAA receptors in cultured murine cerebellar granule cells. Journal of Physiology (London) 503.2, 353–369. Mozrzymas, J.W., Barberis, A., Michalak, K., Cherubini, E., 1999. Chlorpromazine inhibits miniature GABAergic currents by reducing the binding and by increasing the unbinding rate of GABAA receptors. Journal of Neuroscience 19 (7), 2474–2488. Nett, S.T., Jorge-Rivera, J.-C., Myers, M., Clark, A.S., Henderson, L.P., 1999. Properties and sex-specific differences in GABAA receptors in neurons expressing γ1 subunit mRNA in the preoptic area of the rat. Journal of Neurophysiology 81, 192–203. Perrais, D., Ropert, N., 1999. Effect of zolpidem on miniature IPSCs and occupancy of postsynaptic GABAA receptors in central synapses. Journal of Neuroscience 19 (2), 578–588. Pope, H.G. Jr., Katz, D.L., 1988. Affective and psychiatric symptoms associated with anabolic steroid use. American Journal of Psychiatry 145 (4), 487–490. Puia, G., Costa, E., Vicini, S., 1994. Functional diversity of GABAactivated Cl– currents in Purkinje versus granule neurons in rat cerebellar slices. Neuron 12 (1), 117–126. Puia, G., Vicini, S., Seeburg, P.H., Costa, E., 1991. Influence of recombinant γ-aminobutyric acid-A receptor subunit composition on the action of allosteric modulators of γ -aminobutyric acid-gated Cl– currents. Molecular Pharmacology 39 (6), 691–696. Renard, S., Olivier, A., Granger, P., Avent, P., Grahm, D., Sevrin, M., George, P., Besnard, F., 1999. Structural elements of the γaminobutyric acid type A receptor conferring subunit selectivity for benzodiazepine site ligands. Journal of Biological Chemistry 274 (19), 13370–13374. Rogers, C.J., Twyman, R.E., Macdonald, R.L., 1994. Benzodiazepine and β-carboline regulation of single GABAA receptor channels of mouse spinal cord neurones in culture. Journal of Physiology (London) 475.1, 69–82. Rossi, D.J., Hamann, M., 1998. Spillover-mediated transmission at inhibitory synapses promoted by high affinity α6 subunit GABAA receptors and glomerular geometry. Neuron 20 (4), 783–795. Rudolph, U., Crestani, F., Benke, D., Bru¨ nig, I., Benson, J.A., Fritschy, J.-M., Martin, J.R., Bluethmann, H., Mo¨ hler, H., 1999. Benzodiazepine actions mediated by specific γ-aminobutyric acidA receptor subtypes. Nature 401, 796–800. Sieghart, W., 1995. Structure and pharmacology of γ-aminobutyric acidA receptor subtypes. Pharmacological Reviews 47 (2), 181– 234. Sigel, E., Baur, R., 1988. Allosteric modulation by benzodiazepine

633

receptor ligands of the GABAA receptor channel expressed in Xenopus oocytes. Journal of Neuroscience 8 (1), 289–295. Sigworth, F.J., 1980. The variance of sodium current fluctuations at the node of Ranvier. Journal of Physiology (London) 307, 97–129. Silberberg, S.D., Magleby, K.L., 1993. Preventing errors when estimating single channel properties from the analysis of current fluctuations. Biophysical Journal 65 (4), 1570–1584. Silver, R.A., Cull-Candy, S.G., Takahashi, T., 1996. Non-NMDA glutamate receptor occupancy and open probability at a rat cerebellar synapse with single and multiple release sites. Journal of Physiology 494.1, 231–250. Smith, A.J., Alder, L., Silk, J., Adkins, C., Fletcher, A.E., Scales, T., Kerby, J., Marshall, G., Wafford, K.A., McKernan, R.M., Atack, J.R., 2001. Effect of α subunit on allosteric modulation of ion channel function in stably expressed human recombinant γ-aminobutyric acidA receptors determined using 36Cl ion flux. Molecular Pharmacology 59 (5), 1108–1118. Strauss, R.H., Yesalis, C.E., 1991. Anabolic steroids in the athlete. Annual Review of Medicine 42, 449–457. Teisse´ re, J.A., Czajkowski, C., 2001. A β-strand in the γ2 subunit lines the benzodiazepine binding site of the GABAA receptor: structural rearrangements detected during channel gating. Journal of Neuroscience 21 (14), 4977–4986. Walters, R.J., Hadley, S.H., Morris, K.D.W., Amin, J., 2000. Benzodiazepines act on GABAA receptors via two distinct and separable mechanisms. Nature Neuroscience 3 (12), 1274–1281. Wingrove, P.B., Wafford, K.A., Bain, C., Whiting, P.J., 1994. The modulatory action of loreclezole at the γ-aminobutyric acid type A receptor is determined by a single amino acid in the β2 and β3 subunit. Proceedings of the National Academy of Science USA 91 (10), 4569–4573. Wisden, W., Laurie, D.J., Monyer, H., Seeburg, P.H., 1992. The distribution of 13 GABAA receptor subunit mRNAs in the rat brain. I. Telencephalon, diencephalon, mesencephalon. Journal of Neuroscience 12 (3), 1040–1062. Wohlfarth, K.M., Bianchi, M.T., Macdonald, R.L., 2002. Enhanced neurosteroid potentiation of ternary GABAA receptors containing the δ subunit. Journal of Neuroscience 22 (5), 1541–1549. Zhu, W.J., Vicini, S., 1997. Neurosteroid prolongs GABAA channel deactivation by altering kinetics of desensitized states. Journal of Neuroscience 17 (11), 4022–4031. Zhu, W.J., Wang, J.F., Krueger, K.E., Vicini, S., 1996. δ subunit inhibits neurosteroid modulation of GABAA receptors. Journal of Neuroscience 16 (21), 6648–6656.