Selective antagonism of 5α-reduced neurosteroid effects at GABA A receptors

Article abstract of DOI:10.1124/mol.65.5.1191

Although neurosteroids have rapid effects on GABA_A receptors, study of steroid actions at GABA receptors has been hampered by a lack of pharmacological antagonists. In this study, we report the synthesis and characterization of a steroid analog, (3α,5α)-17-phenylandrost-16- en-3-ol (17PA), that selectively antagonized neurosteroid potentiation of GABA responses. We examined 17PA using the α1β2γ2 subunit combination expressed in Xenopus laevis oocytes. 17PA had little or no effect on baseline GABA responses but antagonized both the response augmentation and the direct gating of GABA receptors by 5α-reduced potentiating steroids. The effect was selective for 5α-reduced potentiating steroids; 5β-reduced potentiators were only weakly affected. Likewise, 17PA did not affect barbiturate and benzodiazepine potentiation. 17PA acted primarily by shifting the concentration response for steroid potentiation to the right, suggesting the possibility of a competitive component to the antagonism. 17PA also antagonized 5α-reduced steroid potentiation and gating in hippocampal neurons and inhibited anesthetic actions in X. laevis tadpoles. Analogous to benzodiazepine site antagonists, the development of neurosteroid antagonists may help clarify the role of GABA-potentiating neurosteroids in health and disease.

Full text of DOI:10.1124/mol.65.5.1191

0026-895X/04/6505-1191–1197$20.00
MOLECULAR PHARMACOLOGY
Copyright © 2004 The American Society for Pharmacology and Experimental Therapeutics
Mol Pharmacol 65:1191–1197, 2004
Vol. 65, No. 5
3079/1143632
Printed in U.S.A.
Selective Antagonism of 5␣-Reduced Neurosteroid Effects at
GABAA Receptors
Steven Mennerick, Yejun He,1 Xin Jiang, Brad D. Manion, Mingde Wang,2 Amanda Shute,
Ann Benz, Alex S. Evers, Douglas F. Covey, and Charles F. Zorumski
Departments of Psychiatry (S.M., Y.H., M.W., A.S., A.B., C.F.Z.), Molecular Biology and Pharmacology (X.J., A.S.E., D.F.C.),
Anatomy and Neurobiology (S.M., C.F.Z.), and Anesthesiology (B.D.M., A.S.E.), Washington University School of Medicine, St.
Louis, Missouri
Received November 5, 2003; accepted January 23, 2004
This article is available online at http://molpharm.aspetjournals.org
ABSTRACT
Although neurosteroids have rapid effects on GABAA receptors, for 5␣-reduced potentiating steroids; 5␤-reduced potentiators
study of steroid actions at GABA receptors has been hampered were only weakly affected. Likewise, 17PA did not affect bar-
by a lack of pharmacological antagonists. In this study, we biturate and benzodiazepine potentiation. 17PA acted primarily
report the synthesis and characterization of a steroid analog, by shifting the concentration response for steroid potentiation
(3␣,5␣)-17-phenylandrost-16-en-3-ol (17PA), that selectively to the right, suggesting the possibility of a competitive compo-
antagonized neurosteroid potentiation of GABA responses. We nent to the antagonism. 17PA also antagonized 5␣-reduced
examined 17PA using the ␣1␤2␥2 subunit combination ex- steroid potentiation and gating in hippocampal neurons and
pressed in Xenopus laevis oocytes. 17PA had little or no effect inhibited anesthetic actions in X. laevis tadpoles. Analogous to
on baseline GABA responses but antagonized both the re- benzodiazepine site antagonists, the development of neuroste-
sponse augmentation and the direct gating of GABA receptors roid antagonists may help clarify the role of GABA-potentiating
by 5␣-reduced potentiating steroids. The effect was selective neurosteroids in health and disease.
GABAA receptors are important participants in central forts, a steroid binding site on the GABAA receptor has not
nervous system inhibition and are the target of many clini-
cally important compounds, including benzodiazepines, bar-
biturates, neuroactive steroids, and other general anesthet-
ics (Macdonald and Olsen, 1994; Belelli et al., 1999). By
allosterically potentiating GABAA receptor-mediated re-
sponses, these compounds generally increase inhibition and
decrease overall activity of neuronal circuits (Macdonald and
Olsen, 1994). Neuroactive steroids are particularly interest-
ing because they may be present endogenously at concentra-
tions that modulate GABAA receptor function (Robel and
Baulieu, 1994).
been identified. GABA-potentiating steroids have two dis-
tinct structural classes: 5␣-reduced and 5␤-reduced (Covey et
al., 2001). Both classes potentiate GABAA responses with
similar potency and efficacy; whether the two classes work at
the same site or by the same mechanisms is unclear. In
addition, the role of endogenous neurosteroids in modulating
GABAA receptors is obscure. New pharmacological tools for
exploring neuroactive steroids, including neurosteroid antag-
onists, would advance efforts to understand the interaction of
steroids with GABAA receptors.
Several earlier studies explored potential steroid antagonists.
[3H]Flunitrazepam binding assays suggest that 3␤-hydroxys-
teroids competitively antagonize the actions of the potentiating
steroids (3␣,5␣)-3-hydroxypregnan-20-one (3␣5␣P) and (3␣,5␤)-
3-hydroxypregnan-20-one (3␣5␤P), two standard representa-
tives of the 5␣-reduced and 5␤-reduced classes of potentiating
steroids (Prince and Simmonds, 1992, 1993). In electrophysio-
logical studies, however, 3␤-hydroxypregnane steroids noncom-
petitively antagonize potentiating steroids through activation-
dependent block of GABAA receptors (Wang et al., 2002). The
block by 3␤-hydroxysteroids is not specific; barbiturate poten-
tiation is reduced, as are responses to GABA itself at high
Many important questions remain unanswered regarding
steroid actions at GABAA receptors. Despite concerted ef-
This study was supported by National Institutes of Health grant MH45493,
by a gift from the Bantly Foundation (to C.F.Z.), by National Institutes of
Health grants GM47969 (to D.F.C., C.F.Z., and A.S.E.), NS40488 and AA12952
(to S.M.), and a Klingenstein Foundation grant (to S.M.). M.W. was supported
by a postdoctoral fellowship from Umeå University-Washington University
Network and EU regional fund.
1 Present address: Systems and Cognitive Neuroscience, National Institute
of Neurological Disorders and Stroke, Neuroscience Center, Suite 2115, 6001
Executive Boulevard, Bethesda, MD 20892-9521.
2 Present address: Department of Clinical Science, Section of Obstetrics and
Gynecology, Umeå, University, S-90187 Umeå, Sweden.
ABBREVIATIONS: 3␣5␣P, (3␣,5␣)-3-hydroxypregnan-20-one; 3␣5␤P, (3␣,5␤)-3-hydroxypregnan-20-one; 17PA, (3␣,5␣)-17-phenylandrost-16-
en-3-ol; DMSO, dimethyl sulfoxide; LRR, loss of righting reflex; LSR, loss of swimming reflex.
1191

1192
Mennerick et al.
GABA concentrations (Wang et al., 2002). Therefore, 3␤-hy-
droxysteroids, although interesting functional antagonists of
steroid potentiation, lack the specificity desired of a tool for
exploring steroid potentiation of GABAA receptors.
In the present study, we examined the actions of a syn-
thetic neuroactive steroid analog that was inactive as a po-
tentiator at GABAA receptors in our functional screens,
(3␣,5␣)-17-phenylandrost-16-en-3-ol (17PA). 17PA repre-
sents the first described selective antagonist of neurosteroid
potentiation at GABAA receptors. It is noteworthy that 17PA
antagonized the physiological and anesthetic actions of 5␣-
reduced steroids but had little effect on 5␤-reduced steroids.
Furthermore, 17PA inhibited direct gating of GABAA recep-
tor channels by 5␣-reduced but not 5␤-reduced steroids. In-
hibition of 5␣-reduced steroid potentiation decreased with
increasing potentiating steroid concentration, suggesting the
possibility of a competitive interaction between the antago-
nist and potentiating steroids. 17PA did not antagonize bar-
biturate or benzodiazepine potentiation. Our results suggest
that 5␣-reduced and 5␤-reduced steroids are likely to possess
different sites or mechanisms of potentiation and support the
hypothesis that GABAA receptors are critical participants in
the behavioral anesthesia induced by neuroactive steroids.
Materials and Methods
17PA Synthesis. Fig. 1A shows the reaction scheme for the
synthesis of 17PA.
(3␣,5␣,17␤)-17-Phenylandrostane-3,17-diol (Compound 2).
Phenyllithium (1.8 M in cyclohexane-ether, 2.77 ml) was added to
compound 1 (290 mg, 1.0 mmol) in anhydrous tetrahydrofuran (25
ml) at 0°C. The resultant solution was stirred at ambient tempera-
ture for 5 h. Saturated NH4Cl (5 ml) was added, and the product was
extracted with EtOAc. The combined extracts were washed with
NH4Cl and brine and dried over Na2SO4. The solvent was removed
under reduced pressure, and the residue was purified by column
chromatography (silica gel; hexanes/EtOAc, 8:1) to give compound 2
(162 mg, 44%).
Compound 2 was obtained as white crystals: mp. 193°C to 196°C
(EtOAc-hexanes). 1H NMR ␦ 0.75 (s, 3H), 1.04 (s, 3H), 2.08 (m, 1H),
2.37 (m, 1H), 3.97 (m, 1H), 7.23 to 7.41 (m, 5H); 13C NMR ␦ 11.1, 14.9,
20.3, 24.3, 28.4, 28.9, 31.6, 32.0, 33.6, 35.8, 36.0, 36.2, 38.6, 39.1,
46.7, 49.1, 53.7, 66.4, 86.0, 126.7, 127.2 (2 ϫ C), 127.3 (2 ϫ C), 146.1;
IR ␯max 3403, 2926, 1446, 1002, 908 cmϪ1. Anal. Calcd for C25H36O2:
C, 81.47; H, 9.85; found: C, 81.60; H, 9.89.
(3␣,5␣,17␤)-17-Phenylandrostane-3,17-diol, 3-acetate (Com-
pound 3). 4-(N,N-Dimethylamino)pyridine (16 mg) was added to the
solution of compound 2 (160 mg, 0.43 mmol) in Ac2O (0.5 ml) and
pyridine (1.5 ml). The mixture was stirred at room temperature for 5
min and poured into 10% NaHCO3 (20 ml). After the mixture was
stirred for 10 min, the product was extracted with EtOAc. The
combined extracts were washed with 5% HCl and water and dried
over Na2SO4. The solvent was removed under reduced pressure, and
the residue was purified by column chromatography (silica gel; hex-
anes/EtOAc, 10:1) to give compound 3 (169 mg, 95%).
Compound 3 was obtained as white crystals: mp. 196°C to 198°C.
1H NMR ␦ 0.77 (s, 3H), 1.05 (s, 3H), 1.98 (s, 3H), 2.09 (m, 1H), 2.38
(m, 1H), 4.96 (m, 1H), 7.23 to 7.42 (m, 5H); 13C NMR ␦ 11.3, 14.9,
20.3, 21.5, 24.3, 26.0, 28.3, 31.6, 32.7, 32.8, 33.6, 35.7, 36.2, 38.6,
40.0, 46.7, 49.1, 53.6, 70.1, 86.0, 126.7, 127.2 (2 ϫ C), 127.4 (2 ϫ C),
146.1, 170.6; IR ␯max 3511, 2927, 1719, 1445, 1269, 1024 cmϪ1. Anal.
Calcd for C27H38O3: C, 78.98; H, 9.33; found: C, 79.04; H, 9.22.
(3␣,5␣)-17-Phenylandrost-16-en-3-ol Acetate (Compound 4).
Et3N (0.60 ml, 4.31 mmol) and MsCl (0.17 ml, 2.20 mmol) were added
to compound 3 (160 mg, 0.39 mmol) in CH2Cl2 (15 ml) at 0°C and
Fig. 1. 17PA effects on 5␣-reduced neurosteroid potentiation of GABAA
receptors. A, synthetic scheme for 17PA. Reagents and conditions: a,
PhLi, tetrahydrofuran, 0°C to 25°C, 5 h, 44%; b, Ac2O, pyridine, 4-(N,N-
dimethylamino)pyridine, 25°C, 5 min, 95%; c, MsCl, Et3N, CH2Cl2, 0°C,
40 min, 99%; and d, NaOH, H2O, MeOH, 65°C, 1 h, 95%. B to D, 17PA
effect on GABA responses. B, 17PA (10 ␮M; thick trace) had no effect on
responses to 2 ␮M GABA in an X. laevis oocyte expressing the ␣1␤2␥2
GABAA receptor subunit combination. C, oocyte responses to high GABA
concentrations were unaffected by 10 ␮M 17PA. D, in contrast, the GABA
response from the same oocyte depicted in C was markedly depressed by
3␤5␤P (10 ␮M). For B to D, thin traces represent responses to GABA
alone; thick traces represent steroid co-application with GABA. E, effects
of 10 ␮M 17PA on increasing potentiating steroid concentrations. Note
that the 17PA effect became weaker as potentiating steroid concentration
was increased. F, summary of experiments like that shown in E (n ϭ 11
oocytes). Solid symbols represent 3␣5␣P effects in the absence of 17PA;
open symbols represent 3␣5␣P effects in the presence of 10 ␮M 17PA. The
solid lines are fits of the Hill equation with the minimum level of poten-
tiation constrained to zero but with maximum potentiation uncon-
strained. Parameters for the fits in the absence of 17PA were 0.7 ␮M for
the EC50, 8.4 for maximum potentiation, and 1.4 for the Hill coefficient.
Parameters for fits in the presence of 17PA were 1.3 ␮M for the EC50, 6.7
for maximum potentiation, and 1.5 for the Hill coefficient. G, summary of
the effects of varied concentrations of 17PA against a fixed 3␣5␣P con-
centration (500 nM). The IC50 for 17PA estimated from a fit to the Hill
equation was 2.2 ␮M. Data were averaged from three oocytes.

A GABAA Neurosteroid Antagonist
1193
stirred for 40 min. After CH2Cl2 was removed under reduced pres- cording solution consisted of: 138 mM NaCl, 4 mM KCl, 2 mM CaCl2,
sure, the residue was purified by column chromatography (silica gel;
hexanes/EtOAc, 15:1) to give compound 4 (151 mg, 99%).
1 mM MgCl2, 10 mM glucose, 10 mM HEPES, 0.025 mM D-2-amin-
ophosphonovalerate, 0.001 mM 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-
Compound 4 was obtained as white crystals: mp. 152°C to 153°C benzo[f]quinoxaline-7-sulfonamide, and 0.0002 mM tetrodotoxin, pH
(hexanes). 1H NMR ␦ 0.85 (s, 3H), 1.02 (s, 3H), 2.05 (s, 3H), 2.10 (m, 7.25. Solutions were exchanged via a local multibarrel perfusion
1H), 5.02 (m, 1H), 5.90 (m, 1H), 7.19 to 7.39 (m, 5H); 13C NMR ␦ 11.3, pipette with a common delivery port placed 0.5 mm from the cell
16.7, 20.8, 21.5, 26.1, 28.3, 31.5, 31.8, 32.7, 32.9, 34.0, 35.5, 36.0,
40.3, 47.4, 54.5, 57.6, 70.1, 126.7 (3 ϫ C), 127.2, 128.0 (2 ϫ C), 137.4,
154.9, 170.7; IR ␯max 2948, 1724, 1494, 1446, 1374, 1246 cmϪ1. Anal.
Calcd for C27H36O2: C, 82.61; H, 9.24; found: C, 82.48; H, 9.02.
17PA. 5N NaOH (1 ml) was added to compound 4 (145 mg, 0.37
mmol) in MeOH (15 ml). The mixture was refluxed for 1 h and cooled
to room temperature. Water was added, and the product was ex-
tracted with CH2Cl2. The combined extracts were washed with water
to neutral and dried over Na2SO4. The solvent was removed under
reduced pressure, and the residue was purified by column chroma-
tography (silica gel; hexanes/EtOAc, 10:1) to give compound 5 (123
mg, 95%).
Compound 17PA was obtained as white crystals: mp. 174.5°C to
175.5°C (EtOAc-hexanes). 1H NMR ␦ 0.83 (s, 3H), 1.02 (s, 3H), 2.02
(m, 2H), 2.20 (m, 1H), 4.04 (m, 1H), 5.89 (m, 1H), 7.18 to 7.42 (m, 5H);
13C NMR ␦ 11.2, 16.7, 20.8, 28.5, 29.0, 31.5, 31.8, 32.0, 34.0, 35.5,
35.9, 36.3, 39.3, 47.4, 54.6, 57.6, 66.5, 126.6, 126.7 (2 ϫ C), 127.2,
under study. The pipette solution contained: 140 mM CsCl, 4 mM
NaCl, 0.5 mM CaCl2, 5 mM EGTA, and 10 mM HEPES, pH 7.25. In
some experiments (Fig. 4, C and D), CsCl was replaced with Cs
methanesulfonate, and neurons were clamped at Ϫ20 mV.
Tadpole Anesthesia. Groups of 10 early prelimb bud X. laevis
tadpoles (Nasco, Fort Atkinson, WI) were placed in 100 ml of oxy-
genated Ringer’s solution containing various concentrations of com-
pound (Wittmer et al., 1996). Compounds were added from a 10 mM
DMSO stock (final DMSO concentration Յ 0.2%). After equilibrating
at room temperature for 3 h, tadpoles were evaluated for loss of
righting reflex (LRR) or loss of swimming reflex (LSR) behavioral
endpoints. LRR was defined as failure of the tadpole to right itself
within 5 s after being flipped by a smooth glass rod. LSR was defined
by a failure of purposeful tail movement within 10 s of gently sliding
the tadpole around the beaker with a glass rod for 5 s. Control
beakers containing up to 0.6% DMSO produced no LRR in tadpoles.
Data Analysis. Electrophysiology data acquisition and analysis
were performed with pCLAMP (Axon Instruments Inc.). Data plot-
ting and curve fitting were done with Sigma Plot (SPSS Inc., Chi-
cago, IL). Data are presented as mean Ϯ S.E. Statistical differences
were determined using a two-tailed Student’s t test. Fitting of con-
centration-response data for electrophysiology and behavioral assays
was performed using the Hill equation: r ϭ Rmax/[1 ϩ (EC50/[con-
128.0 (2 ϫ C), 137.4, 154.9; IR ␯
3305, 2928, 1494, 1444, 1002
max
cmϪ1. Anal. Calcd for C25H34O: C, 85.66; H, 9.78; found: C, 85.74; H,
9.95.
(3␤,5␣)-3-hydroxypregnan-20-one sulfate was obtained from Ster-
aloids (Newport, RI), and all other drugs were obtained from Sigma/
RBI (Natick, MA). Steroids were dissolved in DMSO, with final
working concentrations of DMSO typically Յ0.1%. DMSO concentra-
tion was matched across all working solutions to account for any
actions of solvent alone. Pentobarbital was dissolved in 0.1 N NaOH.
GABA Receptor Expression in Xenopus laevis Oocytes.
Stage V to VI oocytes were harvested from mature female X. laevis
(Xenopus I, Ann Arbor, MI) and were defolliculated and injected with
cRNA encoding rat GABAA receptor ␣1, ␤2, and ␥2L subunits.
Capped cRNA was transcribed using the mMESSAGE mMachine kit
(Ambion, Austin, TX). Oocytes were incubated up to 5 days at 18°C
in ND96 medium containing: 96 mM NaCl, 1 mM KCl, 1 mM MgCl2,
2 mM CaCl2, and 5 mM HEPES at pH 7.4, supplemented with
pyruvate (5 mM), penicillin (100 U/ml), streptomycin (100 ␮g/ml),
and gentamycin (50 ␮g/ml). Two-electrode voltage-clamp experi-
ments were performed 2 to 5 days after RNA injection. Extracellular
recording solution was ND96 medium, and 1-M⍀ intracellular re-
cording pipettes were filled with 3 M KCl. Cells were clamped at Ϫ70
mV for all experiments, and current at the end of 20- to 60-s drug
applications was measured for quantification of current amplitudes.
Steroids and other modulators were co-applied with GABA by grav-
ity perfusion.
c])n ], where Rmax is the maximum effect, [conc] is the steroid con-
H
centration, EC50 is the half-maximal effective concentration, and nH
is the Hill coefficient.
Results
Synthesis of the Neuroactive Steroid Antagonist
17PA. As noted above, because modifications of 3␤-hydrox-
ysteroids result in noncompetitive, activation-dependent
GABA receptor antagonists rather than specific antagonists
of neuroactive steroids (Wang et al., 2002), we have recently
shifted our focus toward altering substituents at or near C17,
another region of neuroactive steroids known to be critical for
interactions at GABAA receptors (Phillipps, 1975; Covey et
al., 2001). The steroid 17PA was prepared as part of this
effort (Fig. 1A).
Effects of 17PA on Potentiating Steroids. Fig. 1B
shows the lack of effect of 17PA on GABA responses in X.
laevis oocytes expressing the ␣1␤2␥2 subunit combination.
GABA was used at 2 ␮M in this experiment, representing a
response amplitude approximately 5-fold below the EC50
(Wang et al., 2002). At 20 ␮M GABA, responses were still
unaffected (Fig. 1C). Overall, 17PA altered responses to 2 ␮M
GABA by Ϫ5 Ϯ 4% (n ϭ 8) and responses to 20 ␮M GABA by
ϩ2 Ϯ 1% (n ϭ 3). Previous results have suggested that
Hippocampal Culture Recordings. Primary cultures of hip-
pocampal cells were prepared from 1- to 3-day old postnatal Sprague-
Dawley rats and plated on microdroplets of collagen using estab-
lished methods (Mennerick et al., 1995). In brief, minced
hippocampal slices were dissociated with 1 mg/ml papain in oxygen-
ated Leibovitz L-15 medium and triturated in modified Eagle’s me-
dium containing 5% horse serum, 5% fetal calf serum, 17 mM D-
glucose, 400 ␮M glutamine, 50 U/ml penicillin, and 50 ␮g/ml 3␤-hydroxysteroids might be selective antagonists of neuro-
streptomycin. Dissociated cells were plated at 75 cells/mm2 onto
35-mm plastic culture dishes previously sprayed with collagen mi-
crodroplets (Mennerick et al., 1995). Cultures were treated with
cytosine arabinoside (5–10 ␮M) after 3 days in vitro to halt glial
active steroids (Prince and Simmonds, 1992, 1993). However,
our previous results suggested that 3␤-hydroxysteroids may
be activation-dependent noncompetitive GABA receptor an-
tagonists (Wang et al., 2002). In agreement with these re-
sults, the putative steroid antagonist 3␤5␤P directly inhib-
ited responses to 20 ␮M GABA (Fig. 1D). Thus, 17PA, unlike
3␤-hydroxysteroids, is not a direct antagonist of GABA re-
sponses.
Although it did not affect GABA responses, 10 ␮M 17PA
proliferation.
Electrophysiological recordings were done 8 to 21 days after plat-
ing. Whole-cell recordings at room temperature were performed us-
ing an Axopatch 1D amplifier (Axon Instruments Inc., Union City,
CA) interfaced to a Pentium III-based computer via a Digidata 1200
acquisition board (Axon Instruments Inc.). Access resistance was
typically Ͻ12 M⍀ and was not compensated. The extracellular re- suppressed the potentiation of GABA responses evoked by

1194
Mennerick et al.
the endogenous potentiating neurosteroid 3␣5␣P (Fig. 1, GABA receptor function (Majewska and Schwartz, 1987). We
E–G). Increasing the concentration of potentiator largely assessed whether 17PA interacted with effects of sulfated
overcame the 17PA effect, suggesting that a competitive steroids. Using a sulfated pregnane steroid with 5␣-reduced
mechanism might contribute to the interaction (Fig. 1, E and stereochemistry, (3␣,5␣)-3-hydroxypregnan-20-one sulfate,
F). However, analysis of concentration-response curves pre- we found no change in the degree of inhibition. GABA re-
dicted a maximum response that was smaller than the max- sponses were depressed by 36 Ϯ 4% by 0.5 ␮M (3␣,5␣)-3-
imum in the absence of potentiator (Fig. 1F). In addition, hydroxypregnan-20-one sulfate (data not shown). In the pres-
when directly tested against 20 ␮M 3␣5␣P, 17PA still signif- ence of 10 ␮M 17PA, responses were depressed similarly
icantly reduced the potentiated response (data not shown). (38 Ϯ 3%, n ϭ 6). In sum, 17PA effects are selective to
Steroid solubility and receptor desensitization precluded potentiating neurosteroids.
study of potentiator concentrations above 20 ␮M. These re-
Effects of 17PA on Direct Gating of the GABAA Re-
sults suggest that the mechanism of 17PA block is not simple ceptor by Steroids. Potentiating neuroactive steroids also
competition, although potential explanations for the de- are able to gate the GABAA receptor directly in the absence of
creased maximum response are discussed below. When GABA (Barker et al., 1987). 17PA effectively blocked cur-
3␣5␣P concentration was fixed at 500 nM, an IC50 of approx- rents directly gated by 3␣5␣P (Fig. 3D), but currents gated by
imately 2 ␮M was observed (Fig. 1G).
5 ␮M 3␣5␤P were unaffected (Fig. 3D), consistent with di-
Specificity of 17PA. We also assessed the ability of 17PA astereoselective effects of 17PA on GABA potentiation. Cur-
to antagonize the potentiation of GABA responses by two rents gated by pentobarbital were not affected by 17PA (data
other classes of commonly used GABAA receptor modulators, not shown). On average, 17PA blocked currents gated by 5
benzodiazepines and barbiturates. 17PA had no effect on ␮M 3␣5␣P by 74 Ϯ 5% and currents gated by 5 ␮M 3␣5␤P by
potentiation evoked by either pentobarbital or lorazepam 7 Ϯ 6% (n ϭ 4). Notably, at a similar concentration of 3␣5␣P,
(Fig. 2), suggesting specificity of the antagonist for potenti- 17PA was more effective at reversing that steroid’s direct
ating steroids.
gating effects than it was at reversing its potentiation effects
Specificity was further probed by examining the effect of (Fig. 1, E and F).
17PA on potentiation by 5␤-reduced steroids. 3␣5␤P poten-
tiated GABA responses nearly as well as 3␣5␣P, but 17PA
had essentially no effect on responses enhanced by 3␣5␤P
(Fig. 3, A and B). Even at the lowest concentrations of 3␣5␤P
at which potentiation was observed, there was no detectable
effect of 17PA (Fig. 3C). Thus, 17PA ineffectiveness did not
result from saturation of potentiation by 3␣5␤P. We also
examined the ability of 17PA to antagonize potentiation by
another pair of 5␣-reduced and 5␤-reduced steroids, (3␣,5␣)-
3,21-dihydroxypregnan-20-one
and
(3␣,5␤)-3,21-dihy-
droxypregnan-20-one (500 nM). Again, although the 5␣-re-
duced steroid response was significantly reduced (38 Ϯ 4%,
n ϭ 5), the response to co-application of GABA and the
5␤-reduced steroid was nearly unaffected (13 Ϯ 4%).
Pregnane steroids sulfated at carbon 3 represent another
class of GABA-active steroids, but these steroids inhibit
Fig. 3. 17PA only weakly affects 5␤-reduced steroids. A, GABA responses
under the indicated conditions. GABA was used at 2 ␮M, 3␣5␣P and
3␣5␤P at 0.5 ␮M, and 17PA at 10 ␮M. B, summary of effects like that
shown in A from four oocytes. C, summary of potentiation by 3␣5␤P at
very low concentrations and the effect of 10 ␮M 17PA (n ϭ 3, 4, and 7,
respectively, at the low, medium, and high concentrations of potentiator).
D, 17 PA inhibited currents directly gated by a 5␣-reduced steroid. Top,
3␣5␣P (5 ␮M) was applied and gated a slowly developing and decaying
current in an oocyte expressing the ␣1␤2␥2 subunit combination (left and
right traces). The current was attenuated when 10 ␮M 17PA was co-
applied (middle trace). Bottom, the same protocol was used, but 3␣5␤P
was substituted for 3␣5␣P.
Fig. 2. Potentiation by barbiturates and benzodiazepines was unaffected
by 17PA. A, potentiation by 20 ␮M pentobarbital (Ptb) was unaffected by
10 ␮M 17PA. B, lorazepam (Lzpm; 0.1 ␮M) potentiation was also unaf-
fected. GABA was used at 2 ␮M in both experiments. C, summary of the
pentobarbital experiments (n ϭ 12). D, summary of the lorazepam exper-
iments (n ϭ 8).

A GABAA Neurosteroid Antagonist
1195
Effects of 17PA in Hippocampal Neurons. We also alone. The current relaxed slowly back toward levels of
examined the effects of 17PA in cultured hippocampal neu- GABA alone. Note that even after washing away potentiator
rons and found that 10 ␮M failed to detectably alter the for Ͼ20 s, the current still had not returned to the level of
holding current of neurons when administered by itself and GABA alone, reflecting the extremely slow rate of potentiator
had no significant effect on responses to 0.5 ␮M GABA when dissociation. Washout of GABA at the end of the protocol was
co-applied with GABA (12 Ϯ 8% potentiation, n ϭ 6). As in rapid on the time scale of the sweep, so it is unlikely that
oocytes, 17PA significantly inhibited the GABA potentiating drug delivery was rate limiting in this experiment.
actions of 3␣5␣P (Fig. 4, A and B). Thus, in native cells,
Figure 4C, lower left, shows the effect of again co-applying
where subunits, posttranslational processing, and other fac- GABA plus 3␣5␣P for the first half of the drug application
tors are likely to differ from the oocyte heterologous expres- protocol. Midway through the protocol, the solution was
sion system, 17PA also antagonized potentiating steroid ef- switched to one containing a mixture of GABA, potentiator,
fects.
and 10 ␮M 17PA. The rate of block in this phase of the
We used the more rapid solution exchanges attainable with experiment was similar to the rate of potentiator dissocia-
hippocampal neurons, along with the characteristic slow ki- tion, suggesting that potentiator dissociation is probably rate
netics of potentiating steroid dissociation, to probe the idea limiting in the development of 17PA block. Comparable re-
that 17PA acts competitively with respect to potentiating sults were obtained in nine neurons, where the time constant
steroid. We reasoned that if 17PA competitively antagonizes of potentiator washout (determined from single exponential
potentiation, block of potentiation should not develop faster fits to the decay of the potentiated current) was indistin-
than the rate of dissociation of potentiator from receptors. If guishable from the time constant of block development (17 Ϯ
block develops faster than potentiator dissociation, then it is 3 s versus 13 Ϯ 2 s; p Ͼ 0.05). As a positive control, we
unlikely that 17PA acts in a competitive manner. Figure 4C examined the kinetics of bicuculline block, which acts non-
shows an experiment designed to test the rate of 17PA block competitively with respect to 3␣5␣P (Ueno et al., 1996). The
of prepotentiated receptors. Figure 4C, upper left, shows a kinetics of inhibition were clearly faster than those of 17PA
hippocampal neuron’s response to 0.5 ␮M GABA. At upper (0.5 Ϯ 0.08 s, n ϭ 4).
right is shown the same cell’s response to GABA, potentiated
The slow kinetics of 17PA block in this experiment alone do
by 100 nM 3␣5␣P. Midway through the co-application of not exclude the possibility that the antagonist acts noncom-
GABA and 3␣5␣P, the solution was switched back to GABA petitively with an extremely slow association or inhibition
rate, although the results of co-application experiments like
those in Fig. 4A suggest that 17PA can equilibrate with
receptors rapidly when simultaneously applied with potenti-
ator. These results suggest that competition between 17PA
and 5␣-reduced neurosteroids is the mechanism by which
antagonism occurs. However, the data do not preclude mod-
els in which potentiator binding allosterically depresses
binding of potentiating steroid at a different site or models in
which bound potentiator allosterically alters the ability of
bound antagonist to inhibit the potentiated response.
We also examined directly gated currents in hippocampal
neurons. 3␣5␣P (100–200 nM) gated extremely slowly rising
and decaying currents in hippocampal neurons, and these
currents were more effectively blocked by 17PA than poten-
tiated currents in the presence of GABA (Fig. 4D; compare
Fig. 4A). Overall, steroid-gated currents were inhibited by
84 Ϯ 6% by 10 ␮M 17PA (n ϭ 5). It is interesting that when
the steroid-gated current was challenged with bicuculline,
bicuculline completely blocked the steroid-gated current
along with a small tonic current, which 17PA did not affect
(Fig. 4D, top). Bicuculline removal caused a tail current that
mimicked the washout of 3␣5␣P alone (Fig. 4D, bottom). This
tail current is consistent with a noncompetitive interaction
between bicuculline and neurosteroid agonist (Ueno et al.,
1996) and suggests that 3␣5␣P still interacts with receptors
in the presence of bicuculline. This tail current was not
observed when bicuculline was administered in the absence
of 3␣5␣P (data not shown) and was never observed in the
presence of 17PA (Fig. 4D, top).
Effects of 17PA on Anesthesia. Neuroactive steroids are
potent anesthetics. Although it is believed that the relevant
site of action for anesthetic steroids is the GABAA receptor,
evidence for this assumption is largely correlative (Zorumski
et al., 2000; Lambert et al., 2001). If steroids induce anesthe-
sia through actions at GABAA receptors, we would expect
Fig. 4. 17PA effects in cultured hippocampal neurons. A, 17PA effect on
GABA currents potentiated by 3␣5␣P in hippocampal neurons. Drug
concentrations were 0.5 ␮M GABA, 0.1 ␮M 3␣5␣P, and 10 ␮M 17PA. B,
summary of results from the experiment depicted in C over five neurons.
Amplitudes of potentiated currents were depressed 40 Ϯ 3% by 17PA (p Ͻ
0.05). C, block kinetics tracked with potentiator dissociation. The indi-
cated protocols were performed on a hippocampal neuron using 0.5 ␮M
GABA, 0.1 ␮M 3␣5␣P, and 10 ␮M 17PA. See text for details. Arrows
denote the decay of currents under various conditions. D, 17PA and
bicuculline effects on direct gating in hippocampal neurons. Top, 3␣5␣P
(200 nM) gated a slowly activating and deactivating current in a hip-
pocampal neuron that was nearly abolished by 10 ␮M 17PA. Bottom, in
the same cell, the current gated by 3␣5␣P is superimposed on a trace in
which 25 ␮M bicuculline was co-applied with 3␣5␣P. Note the tail current
upon bicuculline washout.

1196
Mennerick et al.
17PA to antagonize the anesthetic effect of steroids. We eval- tive antagonist of GABA-active neurosteroids. Our results
uated behavioral effects of 17PA in tadpoles, a well-validated suggest several novel conclusions regarding steroid interac-
anesthesia model (Wittmer et al., 1996). When administered tion with GABA receptors. First, 17PA distinguishes steroid
alone, 17PA had no effect on LSR at any concentration tested modulation from modulation by other potentiators, notably
(up to 10 ␮M). At 10 ␮M, LRR was observed in only 2 of 10 barbiturates and benzodiazepines. Second, 17PA selectively
animals (data not shown). As hypothesized, when combined reduces the effects of 5␣-reduced steroids compared with
with steroid anesthetics, 17PA antagonized steroid anesthe- 5␤-reduced steroids, suggesting differences in the mecha-
sia induced by 3␣5␣P but not anesthesia produced by 3␣5␤P nism and/or site of action for these classes. Finally, 17PA
(Fig. 5). At sufficiently high anesthetic concentrations, an- reduces both potentiation and direct gating by 5␣-reduced
tagonism was not apparent (Fig. 5).
steroids, suggesting a relationship between these two mech-
The experiments in Fig. 5 examined co-administration of anisms not previously recognized. The development of selec-
anesthetic and 17PA and demonstrate that 17PA prevents tive steroid antagonists should be important for studies
anesthesia in a manner that can be overcome by sufficiently aimed at understanding the endogenous role of neuros-
high anesthetic concentration. We also assessed whether teroids, which have been implicated in seizure disorders
17PA reverses anesthesia after induction. We evaluated con- (Reddy et al., 2001), depression, sleep disorders, and ethanol
centrations of 3␣5␣P (2 ␮M) and 3␣5␤P (1 ␮M) that yielded effects (for review, see Gasior et al., 1999; Zorumski et al.,
equivalent and near-maximum LRR anesthesia (Fig. 5). An 2000). In addition, neuroactive steroids may participate in
observer naive to the experimental conditions ranked tadpole modulation of GABAergic activity in the central nervous
behavioral responses and found that after a 2-h exposure, 18 system.
of 20 animals exposed to 3␣5␣P, and 20 of 20 animals ex-
The results with 17PA are consistent with at least a par-
posed to 3␣5␤P showed LRR. 17PA (10 ␮M) was then added tially competitive mechanism. Although the pharmacological
to the anesthetic bath solution for half of the animals. After profile of 17PA suggests a competitive component with re-
an additional 3 h., 8 of 10 animals exposed continuously (5 h) spect to potentiating steroids (Fig. 1, E and F), 17PA consis-
to 3␣5␣P alone and 10 of 10 animals exposed to 3␣5␤P alone tently depressed the maximum potentiated response. One
showed LRR. In contrast, 0 of 10 animals exposed to the possibility to explain this result and retain the hypothesis
combination of 3␣5␣P plus 17PA showed LRR. 17PA failed to that 17PA acts through competition is that direct steroid
reverse anesthesia induced by 3␣5␤P; 10 of 10 animals ex- gating and steroid potentiation both contribute to total re-
posed to 3␣5␤P plus 17PA exhibited LRR. Overall, these sponse amplitude at high concentrations of potentiating ste-
results are consistent with the cellular experiments and with roid. Because 17PA is a stronger blocker of direct steroid
the hypothesis that steroid actions at GABAA receptors par- gating (e.g., Fig. 4), apparent maximum potentiation values
ticipate critically in behavioral anesthesia.
are depressed in the concentration-response relationship
shown in Fig. 1F.
Another caveat of the concentration-response relationships
is that the ability to overcome antagonism with high agonist
concentrations does not necessarily imply competition for a
binding site. For instance, inverse benzodiazepine agonists
shift the concentration-response curve for GABA to the right
without changing maximum responses to GABA (Jensen and
Lambert, 1986; Sigel and Baur, 1988), but benzodiazepines
apparently do not compete with GABA for binding. Similarly,
at ␦-containing GABA receptor subunit combinations, Zn2ϩ
shifts the GABA concentration-response curve to the right
without affecting maximum responses (Krishek et al., 1998).
Again, it is unlikely that Zn2ϩ and GABA compete for bind-
ing. Block kinetics for prepotentiated receptors largely fol-
lowed the time course of agonist dissociation. In addition,
17PA inhibition was not associated with large tail currents
observed with bicuculline washout (Fig. 4D), consistent with
the idea that 17PA, but not bicuculline, prevents potentiator
association with receptors. Therefore, if 17PA and 3␣5␣P do
not directly compete for a site, binding of one occludes the
other’s actions.
Discussion
17PA represents a lead compound in the development of
GABA-active neurosteroid antagonists and is the first selec-
17PA was particularly effective at inhibiting directly gated
5␣-reduced steroid responses. For instance, 10 ␮M 17PA was
relatively weak against 100 nM 3␣5␣P as a potentiator in
hippocampal neurons but nearly completely blocked currents
directly gated by 100 to 200 nM 3␣5␣P. Oocytes yielded
similar results (compare Figs. 1, E and F, and 3D). Therefore,
gating is clearly more strongly blocked by 17PA than poten-
tiation. The relationship between gating and potentiation
has been unclear in the literature (Wittmer et al., 1996). One
possible explanation for our results is that gating and poten-
Fig. 5. 17PA effects on steroid anesthesia in tadpoles. A, concentration-
response curve for LRR in tadpoles using 3␣5␣P as anesthetic in one
group of animals (X) and combined with 10 ␮M 17PA (open circles) in
another group of animals. B, concentration-response curve for LSR using
3␣5␣P. Each point represents the fraction of 10 animals that met the
criteria for LRR and LSR. The fits estimated a shift from an EC50 of 1.2
to 3.2 ␮M (LRR) and 2.9 to 10.2 ␮M. C and D, similar plots using 3␣5␤P
as anesthetic. Note that points in the presence and absence of antagonist
superimpose.

A GABAA Neurosteroid Antagonist
1197
Krishek BJ, Moss SJ, and Smart TG (1998) Interaction of Hϩ and Zn2ϩ on recom-
binant and native rat neuronal GABAA receptors. J Physiol 507:639–652.
Lambert JJ, Belelli D, Harney SC, Peters JA, and Frenguelli BG (2001) Modulation
of native and recombinant GABAA receptors by endogenous and synthetic neuro-
active steroids. Brain Res Brain Res Rev 37:68–80.
Macdonald RL and Olsen RW (1994) GABAA receptor channels. Annu Rev Neurosci
17:569–602.
Majewska MD and Schwartz RD (1987) Pregnenolone-sulfate: an endogenous antag-
onist of the gamma-aminobutyric acid receptor complex in brain? Brain Res
404:355–360.
Mennerick S, Que J, Benz A, and Zorumski CF (1995) Passive and synaptic proper-
ties of neurons grown in microcultures and in mass cultures. J Neurophysiol
73:320–332.
Phillipps GH (1975) Structure-activity relationships in steroidal anesthetics. J Ste-
roid Biochem 6:607–613.
Prince RJ and Simmonds MA (1992) 5beta-pregnan-3beta-ol-20-one, a specific an-
tagonist at the neurosteroid site of the GABAA receptor-complex. Neurosci Lett
135:273–275.
Prince RJ and Simmonds MA (1993) Differential antagonism by epipregnanolone of
alphaxalone and pregnanolone potentiation of [3H]flunitrazepam binding suggests
more than one class of binding site for steroids at GABAA receptors. Neurophar-
macology 32:59–63.
tiation may result from 3␣5␣P acting at a single binding site,
whose accessibility to 3␣5␣P is altered by GABA presence on
the receptor. Future work using 17PA and similar com-
pounds can be directed at explicit tests of this model and
others.
In summary, this is the first report, to our knowledge, of a
selective antagonist of neurosteroid actions at GABAA recep-
tors. 17PA exhibits selectivity for 5␣-reduced steroid poten-
tiation of GABA responses and 5␣-reduced steroid direct
receptor gating. Several aspects of the antagonist are consis-
tent with a competitive mechanism. 17PA and future deriv-
atives may be useful for probing the mechanism of neuroac-
tive steroid actions at GABAA receptors and for probing the
endogenous role of neurosteroids in nervous system excitabil-
ity.
Reddy DS, Kim HY, and Rogawski MA (2001) Neurosteroid withdrawal model of
perimenstrual catamenial epilepsy. Epilepsia 42:328–336.
Robel P and Baulieu E-E (1994) Neurosteroids: biosynthesis and function, in Neu-
robiology of Steroids: Methods in the Neurosciences 22 (de Kloet R and Sutanto W
eds) pp 36–50, Academic Press, San Diego, CA.
Sigel E and Baur R (1988) Allosteric modulation by benzodiazepine receptor ligands
of the GABAA receptor channel expressed in Xenopus oocytes. J Neurosci 8:289–
295.
Acknowledgments
We thank Christopher Fields for help in collecting initial data, and
we thank laboratory members Gustav Akk and Joe Henry Steinbach
for discussion.
Ueno S, Zorumski C, Bracamontes J, and Steinbach JH (1996) Endogenous subunits
can cause ambiguities in the pharmacology of exogenous ␥-aminobutyric acidA
receptors expressed in human embryonic kidney 293 cells. Mol Pharmacol 50:931–
938.
Wang M, He Y, Eisenman LN, Fields C, Zeng CM, Mathews J, Benz A, Fu T,
Zorumski E, Steinbach JH, et al. (2002) 3beta-hydroxypregnane steroids are
pregnenolone sulfate-like GABAA receptor antagonists. J Neurosci 22:3366–3375.
Wittmer LL, Hu Y, Kalkbrenner M, Evers AS, Zorumski CF, and Covey DF (1996)
Enantioselectivity of steroid-induced ␥-aminobutyric acidA receptor modulation
and anesthesia. Mol Pharmacol 50:1581–1586.
References
Barker JL, Harrison NL, Lange GD, and Owen DG (1987) Potentiation of gamma-
aminobutyric-acid-activated chloride conductance by a steroid anesthetic in cul-
tured rat spinal neurones. J Physiol 386:485–501.
Belelli D, Pistis M, Peters JA, and Lambert JJ (1999) The interaction of general
anesthetics and neurosteroids with GABAA and glycine receptors. Neurochem Int
34:447–452.
Covey DF, Evers AS, Mennerick S, Zorumski CF, and Purdy RH (2001) Recent
developments in structure-activity relationships for steroid modulators of GABAA
receptors. Brain Res Brain Res Rev 37:91–97.
Gasior M, Carter RB, and Witkin JM (1999) Neuroactive steroids: potential thera-
peutic use in neurological and psychiatric disorders. Trends Pharmacol Sci 20:
107–112.
Zorumski CF, Mennerick S, Isenberg KE, and Covey DF (2000) Potential clinical
uses of neuroactive steroids. Curr Opin Investig Drugs 1:360–369.
Address correspondence to: Dr. Steven Mennerick, Department of Psychi-
atry, Washington University School of Medicine, 660 S. Euclid Avenue, Cam-
pus Box 8134, St. Louis, MO 63110. E-mail: menneris@psychiatry.wustl.edu
Jensen MS and Lambert JD (1986) Electrophysiological studies in cultured mouse
CNS neurones of the actions of an agonist and an inverse agonist at the benzodi-
azepine receptor. Br J Pharmacol 88:717–731.