Permissive role of brain allopregnanolone content in the regulation of pentobarbital-induced righting reflex loss

Neuropharmacology 38 (1999) 955 – 963

Permissive role of brain allopregnanolone content in the regulation of pentobarbital-induced righting reflex loss Kinzo Matsumoto a,b, Veska Uzunova a, Graziano Pinna a,c, Kazuhiro Taki b, Doncho P. Uzunov a, Hroshi Watanabe b, Jean-Marc Mienville a, Alessandro Guidotti a, Erminio Costa a,* a

Department of Psychiatry, The Psychiatric Institute, College of Medicine, Uni6ersity of Illinois at Chicago, 1601 West Taylor Street, Chicago, IL 60612, USA b Department of Pharmacology, Research Institute for Wakan-Yaku, Toyama Medical and Pharmaceutical Uni6ersity, 2630 Sugitani, Toyama 930 -0194, Japan c Department of Radiology and Nuclear Medicine, Klinikum, Benjamin Franklin Free Uni6ersity of Berlin, 12200 Berlin, Germany Accepted 18 January 1999

Abstract Allopregnanolone [3a-hydroxy-5a-pregnan-20-one] (ALLO), a potent neurosteroid that positively modulates g-aminobutyric acid (GABA) action at various GABAA receptor subtypes is synthesized in nanomolar concentrations and stored non uniformly in various brain structures of mammals. We have measured brain ALLO content and its precursors by negative ion chemical ionization-mass-spectrometry after purification and separation of the different steroids with HPLC and gas chromatography. Our procedure measures steroids in the femtomolar range with structural information and unsurpassed selectivity. We were able to establish an association between the decrease in content of ALLO in mouse brain cortex elicited by either long-lasting social isolation or by the administration of 17b-17 [bis (1-methylethyl) amino carbonyl] androstane-3,5-dilene-3-carboxylic acid (SKF 105 111), an inhibitor of Types I and II 5a reductases, and the shortening of the righting reflex loss elicited by pentobarbital (PBT). SKF 105 111 added to cortical brain slices in concentrations up to 10 − 5 M failed per se to alter GABAergic currents or their potentiation by PTB recorded from pyramidal neurons. Fluoxetine (1.45 or 2.9 mmol/kg i.p.) doses that fail to change the PTB-induced loss of righting reflex and the level of brain ALLO in group-housed mice normalized both parameters in socially-isolated mice. In addition, we could detect both fluoxetine actions in socially isolated mice pretreated with doses of p-chlorophenylalanine (1.2 mmol/kg i.p. at 72, 48, and 24 h) that substantially inhibit brain serotonin 5HT synthesis as shown by an 80% drop of brain 5HT content. These studies for the first time have provided evidence suggesting that the endogenous cortical stores of ALLO physiologically upregulate GABAergic tone and by such a mechanism play a permissive or facilitatory role on the PTB-induced loss of the righting reflex. In the absence of such a permissive physiological influence by endogenous ALLO, the righting reflex inhibition by PTB is down regulated. © 1999 Published by Elsevier Science Ltd. All rights reserved. Keywords: Allopregnanolone; HPLC; Gas chromatography-mass spectroscopy; Pentobarbital

1. Introduction Submicromolar concentrations of brain neurosteroids following systemic or intracerebroventricular administration of pregnenolone, pregnenolone sulfate, progesterone, 5a-pregnan-3, 20-dione (5a DHP), 3a-hydroxy-5a-pregnan-20 one [allopregnanolone (ALLO)], or * Corresponding author. Tel.: +1-312-4338361; fax: + 1-3124338358.

dehydroepiandrosterone are sufficient to modulate brain neuronal function by regulating GABAA and NMDA ionotropic receptors and the expression of specific genes (Baulieu and Robel, 1990; McEwen, 1991; Paul and Purdy, 1992; Gee et al., 1995; Lambert et al., 1996; Baulieu and Robel, 1998; Guidotti and Costa, 1998). In spite of the high level of interest triggered by these pharmacological discoveries (Britton and Koob, 1998), it is still unknown whether neurosteroids produced in brain, physiologically regulate specific neurotransmitter functions.

0028-3908/99/$ - see front matter © 1999 Published by Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 8 - 3 9 0 8 ( 9 9 ) 0 0 0 1 8 - 0

956

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

The duration of the righting reflex loss elicited by barbiturates acting via the GABAA receptor is shortened in long-term socially-isolated mice. This response can be normalized by intracerebroventricular or systemic injections of neurosteroids that amplify GABAgated Cl − currents at GABAA receptors (Matsumoto et al., 1996b). The neurosteroids potentiate the pentobarbital (PTB)-induced loss of righting reflex in socially-isolated mice, even when given in doses that per se neither induce the loss of the righting reflex nor potentiate the duration of the PTB-induced loss of the righting reflex in group-housed mice used as controls (Matsumoto et al., 1996b). The most potent neurosteroids that allosterically regulate the actions of GABA and/or PTB at GABAA receptors are ALLO and tetrahydrodeoxycorticosterone (THDOC) (Puia et al., 1990; Paul and Purdy, 1992; Puia et al., 1993; Gee et al., 1995; Olsen and Sapp, 1995; Lambert et al., 1996). Brain cells synthesize ALLO from progesterone (Fig. 1) by converting this steroid first into 5a-DHP, through the action of Types I or II brain 5a-reductases, and, in turn 5a-DHP is converted into ALLO by 3a-hydroxysteroid oxidoreductase (3a-HSOR), also expressed in brain cells (Baulieu and Robel, 1990; Karavolas and Hodges, 1991; Cheney et al., 1995). Presumably, ALLO is the prominent endogenous brain neurosteroid that positively modulates GABAA receptor function (Cheney et al., 1995). In fact, brain tissue fails to synthesize THDOC, and adrenalectomy, which virtually depletes brain THDOC, but not ALLO content (Purdy et al., 1991; Cheney et al., 1995), also fails to reduce the duration of the PTB-induced loss of the righting reflex in group-housed mice or to further shorten the PTB-induced loss of the righting reflex in socially isolated mice (Matsumoto et al., 1996b). ALLO brain content increases during exposure to various stress conditions (Purdy et al., 1991; Barbaccia et al., 1997), but no information is available as to whether brain ALLO biosynthesis is changed in mice exposed to protracted social isolation. Here we tested whether long-term social isolation reduces the strength

of GABAA receptor-mediated synaptic transmission because it decreases brain ALLO content and thereby lowers the efficiency of signal transduction at GABAA receptors elicited by PTB. In mice GABAA receptor efficacy was tested by measuring the duration of the righting reflex loss induced by PTB (Matsumoto et al., 1996b), which directly gates GABAA receptor Cl − channels (Olsen, 1982), and by monitoring possible changes of pregnenolone, progesterone, or ALLO brain expression using a high-performance liquid chromatography –gas chromatographic-mass spectrometric (HPLC-GC-MS) method coupled with negative ion detection by chemical ionization (NICI). This method allows the analysis of femtomolar amounts of neurosteroids while providing specific structural information on the steroid being analyzed (Uzunov et al., 1996; Uzunova et al., 1998). The accessibility of the structural information included in the ion currents selected for quantitative analysis of neurosteroid and of the related internal standards enables the detection of various ALLO stereoisomers (Uzunova et al., 1998) and also obviates the pitfalls that plague radioimmunoassays (Baulieu and Robel, 1998). Moreover, to clarify whether a decrease in brain ALLO content plays a role in shortening the righting reflex loss elicited by PTB, we tested: (a) whether fluoxetine, a potent activator of 3a-HSOR (Uzunov et al., 1996; Griffin et al., 1998; Guidotti & Costa, 1998) (Fig. 1) that increases ALLO content in normal rat brain and normalizes cerebrospinal fluid ALLO content in depressed patients (Uzunov et al., 1996; Uzunova et al., 1998), also normalizes the shortening of the PTB-induced loss of the righting reflex in socially isolated mice; and (b) whether the reduction of brain ALLO content elicited by administration of (17b)-17[bis(1methylethyl) amino carbonyl] androstane -3,5-dilene-3carboxylic acid, (SKF 105 111), an inhibitor of both Types I and II 5a-reductases (Cheney et al., 1995) (Fig. 1), also shortens the PTB-elicited loss of the righting reflex in group-housed mice.

Fig. 1. Conversion of progesterone into allopregnanolone (ALLO) in brain: indicated by the symbol X or ¡ are the possible sites of action of SKF 105 111 and fluoxetine, respectively. 3a-HSOR, 3a-hydroxysteroid-oxidoreductase. SKF 105 111, (17b)-17[bis(1-methylethyl)amino carbonyl]androstane-3,5-dilene-3-carboxylic acid. From Karavolas and Hodges (1991), modified.

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

2. Materials and methods

957

The intra- and the inter-assay coefficients of variation did not exceed 5%.

2.1. Animals 2.4. Neurosteroid determination 2.1.1. (a) Social isolation studies Male ddY mice (18 – 20 g; Japan SLC, Shizuoka, Japan) were obtained at the age of 25 – 30 days. Animals were housed either in groups of 5 – 6/cage (24× 17× 12 cm) or individually (socially isolated) in a cage of the same size for 2 – 10 weeks before the start of our pharmacological and neurochemical experiments. Food and water were given ad libitum. The housing temperature was regulated to be around 24°C, the humidity around 65%, and a 12-h light – dark cycle (lights on: 07:00–19:00) was maintained. 2.1.2. (b) SKF 105 111 studies Male swiss albino mice (20 – 25g, Harlam, Indiana) were group-housed for at least 1 week, under the same conditions described for the dd Y mice before conducting the experiments with SKF 105 111. 2.2. Measurement of PTB-induced loss of the righting reflex The duration of the PTB-induced loss of the righting reflex in group-housed and in socially isolated mice was measured as previously reported (Matsumoto et al., 1996a,b) following intraperitoneal (i.p.) injections of PTB-Na (50 mg/kg).

2.3. PTB extraction and quantification The extraction and the HPLC quantification of PTB were performed using a modification of the method of Shiu and Nemoto (1982). Cerebral cortex samples (approximately 200 mg) were homogenized in 10 volumes of 0.45 M perchloric acid in 150 mM NaCl containing 20 nmol of secobarbital sodium: (5-(methylbutyl)-5-(2propenyl)-2,4,6 (1H, 3H, 5H) pyrimidinetrionemonosodium salt). The homogenate was extracted with an equal volume of chloroform. The organic layer was then extracted with 1 volume of 1 N sodium hydroxide in saline and subsequently with 1.5 volumes of 1 N HCl in diethyl ether. The organic layer was collected, dried under a stream of nitrogen, taken up in distilled water, and injected into a reversed phase Biosil-ODS-5S (250× 4 mm, Bio-Rad) HPLC column equilibrated with 1% trifluoroacetic acid (TFAA) in HPLC-grade distilled water and developed with a gradient of 0.1% TFAA in acetonitrile. PTB and the internal standard were eluted from the column with 35 and 33% acetonitrile, respectively, and were detected by UV absorbance at 220nm. Brain samples were always run along with a concentration-dependent standard curve for PTB. The standard curve for PTB was linear from 5 to 200 nmol.

2.4.1. (a) Extraction and HPLC purification This was performed as previously described (Uzunov et al., 1996; Uzunova et al., 1998). Frontal cortices dissected from mouse brain were homogenized in 10 volumes of distilled water containing 2–5 fmol/ml of a mixture of the 3H-neurosteroids of interest to monitor the recovery (Uzunov et al., 1996). The homogenates were extracted three times with 3 volumes of ethylacetate. The lyophilized extracts were resuspended in 1 ml hexane, filtered through a 13-mm Acrodisc (LC 13 PVDF, 0.2-mm; Gelman), and injected onto an HPLC. HPLC separation of the steroids was performed using a 5-mm Lichrosorb-Diol (4.6×250 mm) column (EM Science, Gibbston, NJ) equilibrated with hexane and developed with two-step tetrahydrofuran gradient: from 0 to 5% in 15 min and from 5 to 9% in additional 40 min, for a total of 55min. The retention times of progesterone, ALLO, and pregnenolone were 32–33, 44–45, and 51–52 min, respectively. The recovery of tritiated steroids through the extraction procedure and the HPLC fractionation ranged between 83–94%. 2.4.2. (b) Quantitati6e analysis of neurosteroids HPLC-purified extracts were derivatized with heptafluorobutyric acid anhydride (HFBA) (50 ml HFBA in 500 ml of ethyl acetate) and injected into the a 58900 GC equipped with a HP/5MS capillary column (length 30 m., i.d. 0.25 mm, film thickness 0.25 mm) coupled to HP 5988 B MS, as previously described (Uzunov et al., 1996). Mass spectrometry was performed operating the MS in the negative ion chemical ionization (NICI) mode, using methane as the reaction gas for pregnenolone and ALLO, and operating in the electron impact (EI) mode for progesterone. The derivatized steroids of interest when subjected to NICI and EI analysis yielded ions in the mass range between m/z 100 and m/z 550. In addition to the HPLC separation and CG retention time characteristic for each steroid (Uzunov et al., 1996; Uzunova et al., 1998), the structural identification of each steroid assayed was provided by its unique mass-fragmentation pattern by using the MS in the selected ion monitoring (SIM) mode focused on the most abundant ion fragments of each steroid derivative, which were m/z 474 and 494 for HFBA-ALLO, 472 and 492 for HFBA-pregnenolone in NICI, and m/z 510 and 495 for HFBA-progesterone using EI mass spectrometry. The internal standard, HFBA-alfaxalone, was monitored by focusing on the specific base ion of the fragmentation spectra, which appeared at m/z 488 for NICI and at m/z 528 for EI. The detection limit for HFBA-ALLO and HFBA preg-

958

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

Fig. 2. Mass-fragmentographic spectrum of HFBA-progesterone in the electron impact (El) mode. Approximately 30 pmol of HFBAprogesterone derivative was used in the experiment.

nenolone in NlCl mode was in the low (1.5 – 3) fmol range; the standard curve was linear between 1 and 1000 fmol. The inter- and the intrasample variability were between 3 and 5%, respectively and failed to reach statistical significance (P B 0.05) using a one way ANOVA (for more details see Cheney et al., 1995; Uzunov et al., 1996). In a previous study (Cheney et al., 1995) using a HP 5971 MSD coupled to an HP 5890A GC we detected only one major ion (m/z ratio 510) with a detection limit in the high nmol range for HFBA-progesterone. However, in the present study, using a more efficient MS, we observed that HFBA-progesterone yielded not only the molecular ion of m/z ratio 510 but also a less abundant ion fragment of m/z ratio 495 (Fig. 2). Using these fragmentation ions, HFBA-progesterone can be detected in EI with a sensitivity of approximately 10 fmol. Since the concentration of progesterone in the mouse brain is in the nmolar range it is possible to determine progesterone content unambiguously in extracts of 100mg of brain tissue applying the hierarchical approach of coupling HPLC with GC and MS in El-mode after HFBA derivatization.

2.5. Electrophysiology The in situ patch-clamp method (Edwards et al., 1989) was used as previously described (Mienville and Barker 1997) to record cortical layer II pyramidal neurons from rat brain slices (postnatal day 12). The drug delivery system consisted of an eight-line bundle converging onto a 100 mm-diameter (i.d.) micromanifold with a dead volume of B1 ml. The tip of the manifold was placed 200 mm from the cell and the solution was ejected at a velocity of 170 mm/ms. This setup allowed rapid (t40 ms) application (for 3 s) and removal of the agonist, as well as preincubation of modulators (for 2 s), before coapplication with the

agonist. GABA was used at a concentration of 50 mM, which is near the EC50 observed with other cortical neurons recorded under the same conditions as those reported in Fig. 6. Low resistance (1.5–2 MV) patch electrodes were used to minimize series resistance changes. They were filled with a solution containing (in mM) 147 CsCl, 10 HEPES and 3.35 CsOH (pH 7.2). Several (two to five) control responses to the agonist were first obtained to check for run-down, and test trials were started after stabilization. Likewise, several (two to three) test responses were collected intermittently and averaged during analysis. Whole-cell currents were voltage-clamped (− 60 mV) and amplified with a List EPC7 unit. They were sampled online via a Digidata 1200B interface, and later automatically measured with the pClamp6 suite of programs (Axon Instruments).

2.6. Drugs and reagents ALLO was purchased from Steraloids (Wilton, NH). Alphaxalone and fluoxetine-HCl were kindly donated by Dr Purdy (Research Service, San Diego VAMC, CA) and Eli Lilly (Indianapolis, IN), respectively. pChlorophenylalanine methylester HCl (p-CPA) was purchased from Sigma (St. Louis, MO). SKF 105 111 was donated by Dr Levy (SmithKline Beecham, King of Prussia, PA). 3H-ALLO, 3H-progesterone, and 3Hpregnenolone were purchased from New England Nuclear (Boston, MA). Fluoxetine HCl was dissolved in saline containing 0.1% Tween 80 just before starting the experiments. p-CPA was dissolved in saline, and SKF 105 111was dissolved in 0.1% DMSO in saline. Test drugs were administered i.p. in a constant volume of 10 ml/kg body weight.

2.7. Drug treatment Fluoxetine HCl (1.45 or 2.9 mmol/kg) or vehicle was administered intraperitoneally (i.p.) 30 min before either PTB injection or decapitation. To reduce brain serotonin (5HT) content, mice were injected with 1.2 mmol/kg of p-CPA i.p. at 72, 48, and 24 h before starting the experiments. We have reported that this treatment depleted, by approximately 80%, the brain content of 5HT (Matsumoto et al., 1996a). To reduce brain ALLO content, mice received SKF 105 111 (Cheney et al., 1995) 30 min before administration of PTB.

2.8. Statistics Data were analyzed with one-way or two-way analyses of variance (ANOVA) followed by the Duncan multiple range test. Differences were considered significant at PB 0.05.

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

959

3. Results

3.1. Fluoxetine re6erses the shortening of the PTBinduced loss of the righting reflex and the reduction of frontal cortex ALLO content in socially isolated mice Consistent with a previous report (Matsumoto et al., 1996a,b), we observed that the PTB-induced loss of the righting reflex was shortened by about 50% in mice socially isolated for 10 weeks (Fig. 3). In initial experiments we observed that administration of fluoxetine in doses of 1.45 or 2.9 mmol/kg (Fig. 3) but not in doses of 0.36 or 0.72 mmol/kg i.p. (not shown here), normalized the duration of the PTB-induced loss of the righting reflex that had been shortened by social isolation without altering the PTB-induced loss of the righting reflex in group-housed animals. A higher dose of fluoxetine (i.e. 58 mmol/kg) also prolongs the PTB-induced loss of the righting reflex in group-housed mice. The fluoxetine-induced normalization of the PTB-induced loss of the righting reflex failed to change the PTB degradation rate; in fact, 20 min after PTB administration, when socially isolated mice treated with vehicle were re-acquiring the righting reflex (see Fig. 3), their brain content of PTB (200918 nmol/g, n =5) was virtually identical to that of the socially isolated mice that received 2.9 mmol/kg of fluoxetine (215 920 nmol/ g, n=5). Since fluoxetine is a potent activator of 3a-HSOR (Guidotti and Costa, 1998; Griffin et al., 1998), the

Fig. 3. Fluoxetine normalizes the social isolation-induced decrease in the pentobarbital (PTB)-induced loss of the righting reflex. DdY mice were either group-housed (open columns) or individually isolated (closed columns) for 6 weeks before beginning the experiment. Fluoxetine (1.45 or 2.9 mmol/kg) was injected i.p. 30 min before PTB-Na injection (50 mg/kg i.p.). Each column represents the mean 9 S.E.M. of eight mice. Fhousing condition × drug (2, 42)= 3.45, PB 0.05. c P B 0.05 compared with vehicle-treated socially-isolated mice. * PB 0.05 compared with vehicle-treated group-housed mice.

Fig. 4. Social isolation-induced decrease in brain frontal cortex ALLO content in mice. DdY mice were either group-housed () or individually isolated (“) for 2 – 10 weeks before the start of the experiment. After HPLC purification of frontal cortex extracts, the content of ALLO was quantitatively analyzed by GC/NICI-MS (see Section 2). Each data point represents the mean9 S.E.M. of four to six mice. * P B0.05 when compared with the respective group-housed mice. Data were analyzed with two-way analysis of variance (ANOVA) followed by the Duncan multiple range test. Differences were considered significant at PB0.05.

enzyme that synthesizes ALLO from 5a-DHP (Fig. 1), we also measured ALLO and related steroids in grouphoused and socially isolated mice receiving the same doses of fluoxetine used in the behavioral experiments of Fig. 3. The content of ALLO in the frontal cortex of grouphoused mice (ddY strain) was about 1.3 pmol/g of tissue, and this level remained constant during a period of 2–10 weeks (Fig. 4). In contrast, in mice socially isolated for 6–10 weeks, the ALLO content of the frontal cortex was approximately 40% lower than that of group-housed mice (Fig. 4). Fluoxetine (1.45 or 2.9 mmoles/kg i.p.) failed to change the frontal cortex ALLO content of group-housed mice, but normalized the frontal cortex ALLO content of mice socially isolated for 10 weeks (Fig. 5). The pregnenolone or progesterone expression in the frontal cortex in grouphoused and socially isolated mice was identical (Table 1). Only traces (below 1.5 fmol) of ALLO stereoisomers 3a5b, 3b5a or 3b5b were found in the cerebrocortical extracts of both groups of mice when measured by NlCl-MS after HPLC purification and GC separation (for details see Uzunova et al., 1998). Fluoxetine is known to be a selective serotonin reuptake inhibitor (SSRI) (Fuller and Wong, 1990), thus it is necessary to exclude the possibility that in socially isolated mice the fluoxetine elicited normalization of the time course of the righting reflex loss and of the decrease of ALLO cortical content is related to an indirect stimulation of serotonin (5HT) receptors. We

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

960

3.2. The inhibition of the 5a-reductases by SKF 105 111 reduces frontal cortex ALLO content and shortens the PTB-induced loss of the righting reflex in group-housed mice

Fig. 5. Fluoxetine doses that fail to change brain frontal cortex ALLO content in group-housed mice normalize ALLO content in socially isolated mice. DdY mice were group-housed (open columns) or individually isolated (closed columns) for 10 weeks before the start of the experiment. ALLO content was determined 30 min after the i.p. injection of fluoxetine (1.45 or 2.9 mmol/kg). Each data point represents the mean9 S.E.M. of four to five mice. * PB 0.01 when vehicle treated isolated mice were compared with vehicle-treated group-housed mice. c PB 0.01 compared with vehicle-treated isolated mice. Fluoxetine was dissolved in saline containing 0.1%, Tween 80, and was administered in a constant volume of 10 ml/kg bodyweight.

investigated whether the fluoxetine-induced normalization of both brain ALLO content and of the PTB-induced loss of the righting reflex would be modified in socially isolated mice by pretreatment with three successive doses of p-CPA (see Table 2), which inhibits 5HT synthesis and reduces 5HT brain content by about 80% (Matsumoto et al., 1996a). This p-CPA pretreatment failed to change the fluoxetine-induced normalization of the PTB-induced loss of the righting reflex and the decrease in ALLO frontal cortex content elicited by social isolation (Table 2). Neither fluoxetine administration nor p-CPA pretreatment changed the brain content of pregnenolone or progesterone in group-housed or socially isolated mice (Table 1).

To further elucidate the permissive role of endogenous ALLO brain content on the duration of the PTB-induced loss of righting reflex, the brain content of ALLO in group-housed mice (Swiss strain) was lowered by a single injection (47 mmol/kg i.p.) of SKF 105 111, (Fig. 1). As shown in Table 3, when SKF 105 111 had decreased the frontal cortex ALLO content by approximately 90%, the duration of the PTB-induced loss of the righting reflex was shortened by approximately 50%. This decrease in the duration of PTB action was independent of changes in the PTB metabolic rate (Table 3). Moreover, SKF 105 111 (up to 50 mM) failed to alter the PTB-induced enhancement of the GABAmediated Cl − current recorded in vitro in neocortical neurons by in situ patch clamp technology (Fig. 6.) This indicates that SKF 105 111 does not interfere directly with PTB action at GABAA receptors.

4. Discussion Among neurosteroids (i.e. pregnenolone, progesterone, 5a-DHP, dehydroepiandrosterone, ALLO and its stereoisomers), only ALLO and its 3a5b stereoisomer pregnanolone positively and allosterically modulate GABA or PTB actions at GABAA receptors at low nanomolar concentrations (Puia et al., 1990; Paul and Purdy, 1992; Puia et al., 1993; Gee et al., 1995; Lambert et al., 1996). Since ALLO is present in the brain and its concentration persists in rat brain after the adrenals and the gonads are removed, it is presumed that it is derived from pregnenolone and progesterone synthesized in the brain (Purdy et al., 1991; Cheney et al., 1995). However, there is virtually no evidence to suggest that brain-produced endogenous ALLO functions as a modulator of the physiological actions of GABA

Table 1 Frontal cortex content of pregnenolone and progesterone following treatment with Fluoxetinea Steroid (pmol/g)

Fluoxetine (2.9 mmol/kg)

Group-housed

Socially isolated mice

Control

Control

p-CPA-treated

Pregnenolone

− +

65 99.1 70 97.2

90 9 11 86 9 10

70 9 6.3 65 9 4.7

Progesterone

− +

0.70 90.22 0.81 90.14

0.51 90.12 0.69 90.22

0.66 9 0.11 0.50 9 0.15

a DdY mice were either group-housed or socially isolated for 10 weeks. Either vehicle or p-CPA (1.2 mmol/kg i.p.) were injected 72, 48 and 24 h before starting the experiment. We have reported that this treatment depletes, by approximately 80%, the brain content of 5HT (Matsumoto et al., 1996a). Fluoxetine was injected i.p. 30 min before the neurosteroid measurements. Each value represents the mean 9S.E.M. obtained from four to six animals.

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

961

Table 2 Effect of pretreatment with p-CPA on the fluoxetine-induced normalization of frontal cortex ALLO content and of the PTB-induced loss of the righting reflex in socially isolated micea Fluoxetine (2.9 mmol/kg)

p-CPA (1.2 mmol/kg)

Loss of righting reflex elicited by PTB (min)

ALLO frontal cortex (pmol/g)

− + − +

− − + +

379 4.0 609 8.5b 3292.0 4892.5b

0.52 90.12 2.09 0.52b 0.60 9 0.25 1.6 9 0.30b

a DdY mice individually isolated for 6 weeks were pretreated three times with p-CPA i.p. before the start of the experiment (see Section 2). A total of 30 min before the experiment, fluoxetine was injected i.p. in socially isolated mice pretreated with either vehicle or p-CPA. Each value represents the mean 9S.E.M. of 8–12 mice. b PB0.05 when fluoxetine-treated mice were compared with the respective vehicle-treated mice. Note that the difference in ALLO levels in mice that received fluoxetine in the absence versus the presence of p-CPA treatment is not statistically significant.

at GABAA receptors. In the present study we have demonstrated that protracted social isolation of mice markedly reduces both ALLO brain content and the duration of the PTB-induced loss of the righting reflex, which depends on PTB action at GABAA receptors. Also, the inhibition of brain Type I and Type II 5a-reductases by SKF 105 111 that produces a marked decrease of brain ALLO content in group-housed mice, shortens the PTB-induced loss of the righting reflex. Neither social isolation nor treatment with SKF 105 111 increase the in vivo degradation rate of PTB. These data support the concept that endogenously produced brain ALLO may have a physiological, permissive role on the PTB action at GABAA receptors. In support of this concept we also demonstrate here that when fluoxetine normalizes cortical ALLO content that is reduced in socially isolated mice, it also normalizes the concurrent shortening of the PTB-induced loss of the righting reflex. Importantly, both effects of fluoxetine are dose dependent and using this property we have shown that fluoxetine can selectively be active in socially isolated mice but not in group-housed mice (Figs. 3 and 5). Virtually all clinical and pharmacological actions of fluoxetine have been attributed to its ability to selectively inhibit 5HT reuptake into serotonergic nerve terminals (Stark et al., 1985; Gorman et al., 1987; Fuller and Wong 1990). This inhibition could also mediate the normalization of the social isolation induced shortening of PTB-induced loss of the righting reflex and of the decrease of brain ALLO content. When socially isolated mice were pretreated with doses of p-CPA that inhibit 5HT synthesis and reduce brain 5HT content to about 20% of normal, however, the fluoxetine-induced normalization of social isolation induced shortening the PTB action and decrease of brain ALLO content persisted. These results are consistent with the view that the fluoxetine actions in socially isolated mice are independent from its action on brain 5HT uptake. Moreover, preliminary studies indicate that, in contrast to fluoxetine, citalopram — a potent

and selective 5HT uptake inhibitor (Hyttel 1978)— which does not increase cortical ALLO content in socially isolated mice-failed to inhibit the decrease in PTB-elicited loss of the righting reflex elicited by social isolation even when given in doses that prolonged PTB effects on the righting reflex in group-housed animals (Matsumoto et al., unpublished data). Taken together, these data virtually exclude a connection between the fluoxetine-elicited inhibition of 5HT uptake and the normalization of social-isolation induced PTB-shortening of righting reflex loss and decrease of ALLO brain content. The long-term decrease in ALLO brain content observed in socially isolated mice could be due at least in part to a decrease in brain uptake of pregnenolone and/or progesterone secondary to their decreased blood levels during a protracted isolation dependent stress. However, in our mice, 10 weeks of social isolation fail to change pregnenolone and/or progesterone brain content (Table 1), indicating that a decrease of blood borne pregnenolone and progesterone do not contribute to the decrease of ALLO brain content. Since the shortening of the PTB-induced loss of the righting reflex in socially isolated mice is independent of the presence of the adrenal glands (Matsumoto et al., 1996b), one might infer that the shortening of PTB action and the decrease of brain ALLO cortical content in socially isolated mice are independent from an adrenal functional deficit and very likely are mediated by a reduced biosynthesis of ALLO in brain neocortex. In fact brain ALLO synthesis is independent of pituitary regulation (Cheney et al., 1995) and is mediated by the brain conversion of progesterone into ALLO catalyzed by 5a-reductases and 3a-HSORs (Fig. 1). Very likely the brain expression of 5a-reductases and 3a-HSORs, is regulated, by steroid hormones secreted by the adrenal cortex and/or ovaries (Karavolas and Hodges, 1991). This steroid hormone secretion or the expression of other putative transcription factors may be decreased during isolation stress, and thereby the expression of the function of 5a-reductases and/or 3a-

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

962

Table 3 Correlation between the PTB-induced loss of the righting reflex, and allopregnanolone or pentobarbital content of frontal cortex of group-housed mice pretreated with SKF 105 111 prior to PTB administrationa Pretreatment

Loss of righting reflex elicited by PTB (min)

ALLO frontal cortex (pmol/g)

PTB frontal cortex (nmol/g)

Vehicle SKF 105 111

82 9 11 41 9 2.8b

3.9 9 0.45 0.26 9 0.010b

190 925 210 915

a Group-housed swiss mice were treated with vehicle or SKF 105 111 (47 mmol/kg i.p.) 30 min before administering PTB (50 mg/kg i.p.). The levels of ALLO in frontal cortex were determined at the end of the PTB-induced loss of the righting reflex. In the olfactory bulb of the same animals ALLO was 14.5 9 2.0 pmol/g in vehicle-and 3.2 9 0.28 pmol/g in SKF 105 111-treated mice, and the cerebellar content was 1.8 90.2 in vehicle-and 0.25 90.01 in SKF 105 111-treated animals. Pregnenolone and progesterone content was not changed. PTB was measured, as described in Section 2, 20 min after the administration of PTB, both in vehicle- and in SKF 105 111-treated animals. Each value is the mean9 S.E.M. obtained from eight to ten mice. b PB0.01 when vehicle-pretreated mice were compared with SKF 105 111-pretreated mice.

HSORs might be reduced. Such a possibility is presently under scrutiny in our laboratory. It has been suggested (Uzunov et al., 1996; Uzunova et al., 1998; Griffin et al., 1998; Guidotti and Costa, 1998; Guidotti et al., 1998), that fluoxetine-elicits an increase in brain ALLO content via an action on the activity of 3a-HSOR, resulting in an acceleration of the reduction rates of 5a-DHP to ALLO or a slowdown of the oxidation of ALLO to 5a-DHP (Fig. 1). Because fluoxetine in the doses used has failed to change the level of ALLO in group-housed mice but normalizes the social isolation-induced decrease in neocortical expression of ALLO independently from its inhibition of 5HT uptake, the normalization of brain ALLO content in socially-isolated mice may be due, at least in part, to a fluoxetine-induced upregulation of 3a-HSOR activity which has a higher sensitivity to fluoxetine in socially-isolated than that in group-housed mice. In cells expressing recombinant GABAA receptors, low nanomolar concentrations of ALLO positively modulate GABA-gated Cl − currents (Lambert et al., 1996; Puia et al., 1990, 1993). As mentioned earlier, 5a-reductases and 3a-HSORs, the enzymes that convert progesterone into ALLO, are unevenly expressed in different brain structures, and are detected in greater amounts in the olfactory bulb. Thus, ALLO concentrations in rat (Cheney et al., 1995) or mouse brain (Pinna et al., unpublished data) range from 1–3 nM in the frontal cortex to 15 – 30 nM in the olfactory bulb. These concentration ranges are compatible with the hypothesis that in vivo they may positively modulate GABA action on GABAA receptors. From our experiments it can be inferred that the decrease of brain ALLO content occurring either during social isolation, or following the 5a-reductase inhibition by SKF 105 111 in group-housed mice, can be correlated with the shortening of the PTB-induced loss of the righting reflex. Of course, more stringent demonstrations that brain ALLO is a physiological modulator with a permissive action at central GABAergic neu-

ronal transmission are needed. For instance, we plan to study whether brain slices of mice pretreated with SKF 105 111 have a reduced responsiveness of GABAA receptors to GABA and whether this reduction can be reversed by preincubation of the slices with ALLO concentrations that per se fail to modify GABA action in slices of naı¨ve mice.

Fig. 6. Lack of effect of SKF 105 111 on GABAergic currents or on their potentiation by pentobarbital (PTB). The upper panel shows typical traces obtained in voltage-clamped cortical pyramidal neurons under the following conditions: (a) GABA (50 mM); (b) GABA + PTB (100 mM); (c) GABA +SKF (50 mM); (d) GABA +SKF+ PTB; (e) GABA (recovery). The lower-panel diagram summarizes data obtained in 12 neurons, after normalizing to current amplitudes obtained with 50 mM GABA alone. GABA-evoked inward currents had large amplitudes, averaging 3968 9 549 pA (range 1224–8010 pA). These currents were unaffected by pre- or co-application of 50 mM SKF 105 111 and were similarly potentiated by PTB (31 9 7%) or PTB+ SKF 105 111 (31 9 9%) in a reversible fashion (recovery= 92 96% of control). PTB-potentiated currents were significantly larger than currents recorded in the presence of SKF or GABA alone after drug washout (PB 0.01), and were not different from PTB-potentiated currents in the presence of SKF 105 111 (one-way ANOVA followed by post-hoc analysis). For experimental conditions see Mienville and Barker (1997).

K. Matsumoto et al. / Neuropharmacology 38 (1999) 955–963

Acknowledgements This work was supported in part by the following grants: MH 56890 (NIMH), A.G.; MH 49486 (NIMH), A.G.; MH 56500-01 (NIMH), E.C. Ministry of Education, Science Sports Culture, Japan ( c 10672148), K.M.; Human Frontiers Science Program Organization (HFSPO) Fellowship, G.P.

References Barbaccia, M.L., Roscetti, G., Trabucchi, M., et al., 1997. The effects of inhibitors of GABAergic transmission and stress on brain and plasma allopregnanolone concentrations. Br. J. Pharmacol. 120, 1582 – 1588. Baulieu, E.E., Robel, P., 1998. Dehydroepiandrosterone (DHEA) and dehydroepiandrosterone sulfate (DHEAS) as neuroactive neurosteroids. Proc. Natl. Acad. Sci. USA 95, 4089–4091. Baulieu, E.E., Robel, P., 1990. Neurosteroids: a new brain function? J. Steroid. Biochem. Mol. Biol. 37, 395–403. Britton, K.T., Koob, G.F., 1998. Premenstrual steroids? Nature 392, 869 – 870. Cheney, D.L., Uzunov, D.P., Costa, E., et al., 1995. Gas chromatographic-mass fragmentographic quantitation of 3a-hydroxy-5apregnan-20-one (allopregnanolone) and its precursors in blood and brain of adrenalectomized and castrated rats. J. Neurosci. 15, 4641 – 4650. Edwards, F.A., Konnerth, A., Sakmann, B., et al., 1989. A thin slice preparation for patch clamp recordings from neurons of the mammalian central nervous system. Pflugers Arch. 414, 600 – 612. Fuller, R.W., Wong, D.T., 1990. Serotonin uptake and serotonin uptake inhibition. Ann. N.Y. Acad. Sci. 600, 68–78. Gee, K.W., McCauley, D.L., Lan, N.C., 1995. A putative receptor for neurosteroids on the GABAA receptor complex. The pharmacological properties and therapeutic potential of epalons. Crit. Rev. Neurobiol. 9, 207–227. Gorman, J.M., et al., 1987. An open trial of fluoxetine in the treatment of panic attacks. J. Clin. Psychopharmacol. 7, 329 – 332. Griffin, L.D., Compagnone, N., Mellon, S., 1998. Selective serotonin reuptake inhibitors-induced alteration in 3a-hydroxysteroid dehydrogenase activity in brain. Neurosci. Abstr. 24, 1012. Guidotti, A., Costa, E., 1998. Can the antidysphoric and anxiolytic profiles of SSRIs be related to their ability to increase brain 3a,5a-tetrahydroprogesterone (ALLO) availability? Biol. Psychiatry 44, 865 – 873. Guidotti, A., Uzunov, D.P., Auta, J., et al., 1996. Application of gas chromatography mass fragmentography with negative ion chemical ionization technology to measure neurosteroids and their biosynthesis rate in the rat brain. In: Genazzani, A.R., Petraglia, F., Purdy, R.H. (Eds.), The Brain: Source and Target for six Steroid Hormones. The Parthenon Publishing Group, New York, pp. 25 – 41.

.

963

Hyttel, J., 1978. Effect of a specific 5-HT uptake inhibitor, citalopram (Lu 10 – 171), on 3H-5-HT uptake in rat brain synaptosomes in vitro. Psychopharmacology 60, 13 – 18. Karavolas, H.J., Hodges, D.R., 1991. Metabolism of progesterone and related steroids by neural and neuroendocrine structures. In: Costa, E., Paul, S.M. (Eds.), Neurosteroids and Brain Function, vol. 8. Thieme, New York, pp. 135 – 145. Lambert, J.J., Belelli, D., Hill-Venning, C., et al., 1996. Neurosteroids and GABAA receptor function. Trends Pharmacol. Sci. 16 (9), 295 – 303. Matsumoto, K., Mizowaki, M., Suchitra, T., et al., 1996a. Central antinociceptive effects of mitragynine in mice: contribution of descending noradrenergic and serotonergic systems. Eur. J. Pharmacol. 317, 75 – 81. Matsumoto, K., Ojima, K., Watanabe, H., 1996b. Neurosteroidal modulation of social isolation-induced decrease in pentobarbital sleep in mice. Brain Res. 708, 1 – 6. McEwen, B.S., 1991. Non-genomic and genomic effects of steroids on neural activity. Trends Pharmacol. Sci. 12, 141 – 147. Mienville, J.M., Barker, J.L., 1997. Potassium current expression during prenatal corticogenesis in the rat. Neuroscience 81, 163– 172. Olsen, R.W., 1982. Drug interactions at the GABA receptorionophore complex. Ann. Rev. Pharmacol. Toxicol. 22, 245– 277. Olsen, R.W., Sapp, D.W., 1995. Neuroactive steroid modulation of GABAA receptors. In: Biggio, G., Sanna, E., Serra, M., Costa, E. (Eds.), GABAA Receptors and Anxiety from Neurobiology to Treatment. In: Advances in Biochemical Psychopharmacology, vol. 48. Ravens, New York, pp. 57 – 74. Paul, S.M., Purdy, R.H., 1992. Neuroactive steroids. FASEB J. 6, 2311 – 2322. Puia, G., Ducic, I., Vicini, S., et al., 1993. Does neurosteroid modulatory efficacy depend on GABAA receptor subunit composition? Recept. Channel. 1, 135 – 142. Puia, G., Santi, M.R., Vicini, S., et al., 1990. Neurosteroids act on recombinant human GABAA receptors. Neuron 4, 759 –765. Purdy, R.H., Morrow, A.L., Moore, P.H. Jr., et al., 1991. Stress-induced elevation of g-aminobutyric acid type A receptor-active steroids in the rat brain. Proc. Natl. Acad. Sci. USA 88, 4553– 4557. Shiu, G.K., Nemoto, E.M., 1982. Simple, rapid and sensitive reversed-phase high-performance liquid chromatographic method for thiopental and pentobarbital determination in plasma and brain tissue. J. Chromatogr. 227 (1), 207 – 212. Stark, P., Fuller, R.W., Wong, D.T, 1985. The pharmacologic profile of fluoxetine. J. Clin. Psychiatry 46, 7 – 13. Uzunov, D.P., Copper, T.B, Costa, E., et al., 1996. Fluoxetine-elicited changes in brain neurosteroid content measured by negative ion mass fragmentography. Proc. Natl. Acad. Sci. USA 93, 12599 – 12604. Uzunova, V., Sheline, Y., Davis, J.M., et al., 1998. Increase in the cerebrospinal fluid content of neurosteroids in patients with unipolar major depression who are receiving fluoxetine or fluvoxamine. Proc. Natl. Acad. Sci. USA 95, 3239 – 3244.