Norepinephrine–gamma-aminobutyric acid (GABA) interaction in limbic stress circuits: effects of reboxetine on GABAergic neurons

Norepinephrine–Gamma-Aminobutyric Acid (GABA) Interaction in Limbic Stress Circuits: Effects of Reboxetine on GABAergic Neurons James P. Herman, Andrew Renda, and Bryan Bodie Background: Reboxetine is a selective norepinephrine (NE) reuptake inhibitor that exerts significant antidepressant action. The current study assessed norepinephrine– ␥-aminobutyric acid (GABA)-ergic mechanisms in reboxetine action, examining glutamic acid decarboxylase (GAD) mRNA expression in limbic neurocircuits following reboxetine within the context of chronic stress. Methods: Male rats received 25 mg/kg reboxetine/day, p.o. Reboxetine and vehicle animals were exposed to 1 week of variable stress exposure or handling. Behavioral responses to stress (open field) were tested on day 7, and animals were killed on day 8 to assess neuroendocrine stress responses and limbic GAD65/67 mRNA regulation (in situ hybridization). Results: Reboxetine significantly decreased behavioral reactivity in the open field. Reboxetine administration did not affect expression of GAD65/67 mRNA in handled rats; however, administration to stressed animals reduced GAD67 (but not GAD65) mRNA in the medial amygdaloid nucleus, posteromedial bed nucleus of the stria terminalis, and dentate gyrus. In contrast, GAD65 mRNA expression was increased by reboxetine in the lateral septum of stressed animals. Conclusions: Norepinephrine pathways appear to modulate synthesis of GABA in central limbic stress circuits. Decreases in GABA synthetic capacity suggest reduced activation of stress-excitatory pathways and enhanced activation of stress-inhibitory circuits, and is consistent with a role for GABA in the antidepressant efficacy of NE reuptake inhibitors. Biol Psychiatry 2003;53: 166 –174 © 2003 Society of Biological Psychiatry Key Words: Depression, antidepressants, norepinephrine reuptake inhibitors, hippocampus, amygdala, septum

From the Department of Psychiatry, University of Cincinnati Medical Center, Cincinnati, Ohio (JPH, BB); and Department of Anatomy and Neurobiology, University of Kentucky School of Medicine, Lexington, Kentucky (JPH, AR). Address reprint requests to James P. Herman, Ph.D., University of Cincinnati Medical Center, Department of Psychiatry, 231 Albert Sabin Way, Cincinnati OH 45267-0559. Received November 2, 2001; revised April 23, 2002; accepted May 6, 2002.

© 2003 Society of Biological Psychiatry

Introduction

N

orepinephrine (NE) systems are known to play a major role in affective disease states. Treatments that enhance central NE neurotransmission have potent antidepressant efficacy, whereas NE depletion can produce depression-like symptoms (c.f., Anand and Charney 2000; Delgado and Moreno 2000; Tanaka et al 2000). In animal models, central NE systems are activated following acute and chronic stress exposure (Nestler et al 1999; Nisenbaum and Abercrombie 1993; Valentino et al 1998; Watanabe et al 1995). Taken together, the data suggest that central NE plays a role in stress adaptation, and this process may be impaired in disease states such as depression. Limbic stress-regulatory regions are prominent targets for agents affecting noradrenergic neurotransmission. Ascending NE axons originating in the locus coeruleus (LC) richly innervate numerous regions implicated in stress integration, including the extended amygdala, hippocampus, prefrontal cortex, and hypothalamus (Moore and Card 1984). Possible targets for NE action are ␥-aminobutyric acid (GABA)-ergic cell populations indigenous to these regions. GABAergic agonists have antidepressant and mood-altering effects in humans (Petty 1995; Shiah and Yatham 1998). Furthermore, previous studies from our group indicate that GABAergic systems are activated by chronic stress exposure in forebrain targets of NE axons, including the bed nucleus of the stria terminalis, preoptic area, dorsomedial hypothalamic nucleus, and hippocampus (Bowers et al 1998). Thus, NE-synthesizing neurons are in prime position to affect stress integration through interactions with GABAergic neurons in key limbic stressintegrative cell populations. The potential for NE–GABA interaction in limbic regions predicts a possible substrate for actions of NEenhancing antidepressant drugs. To test this hypothesis, the current study examined the effect of the selective NE reuptake inhibitor reboxetine on GABAergic transmission and stress responsivity in the rat. Reboxetine is a highly selective NE reuptake inhibitor that serves to effectively increase availability of NE at the synapse, thus enhancing 0006-3223/03/$30.00 doi:10.1016/S0006-3223(02)01449-X

NE–GABA Interactions in Stress Circuitry

the efficacy of NE neurotransmission (Montgomery and Schatzberg 1998; Wong et al 2000). This drug has powerful antidepressant action in humans (Massana 1998; Schatzberg 2000; Versiani et al 1999), indicative of interplay with central stress pathways. The influence of reboxetine on stress responsivity was assessed using a chronic intermittent stress paradigm, which shows reliable stimulation of central and peripheral limbs of the hypothalamic–pituitary–adrenal axis (Herman et al 1995; Paskitti et al 2000) that can be attenuated by antidepressant medications (Lopez et al 1997).

Methods and Materials Subjects Subjects were male Sprague-Dawley rats, weighing 250 –300 g at the time of testing. Animals were housed three per cage in temperature- and humidity-controlled quarters (vivarium) on a 12:12 hours light:dark cycle. Food and water were available ad libitum throughout the procedure.

Stress Procedure Animals were given 2 weeks to acclimate to the vivarium quarters before the initiation of testing. During the second week, water bottles were weighed to determine average daily water consumption per cage. On the ensuing week, reboxetine was dissolved in the drinking water to deliver an average dose of 25 mg/kg/day. The dose of reboxetine was chosen on the basis of a previous report from Wong et al (2000). Treated animals (n ⫽ 12) were maintained on reboxetine-supplemented tap water throughout the procedure, with control rats (n ⫽ 12) administered tap water only. Water consumption was monitored throughout the experiment. Beginning on day 8, six reboxetine-treated and six water-treated animals were submitted to a 7-day chronic variable stress regimen previously characterized by our laboratory (Herman et al 1995; Paskitti et al 2000). Briefly, the stress procedure consisted of twice daily exposure to the following sequence of stressors: Day 1, AM: Rotation (six per cage on orbital shaker platform, 2 hours) Day 1, PM: Cold (placed in 4°C cold room for 2 hours) Day 2, AM: Cold swim (18°C for 2 min) Day 2, PM: Restraint (Plexiglas restrainers, 2 hours) ⫹ overnight crowding (six per cage) Day 3, AM: Cold Day 3, PM: Warm swim (31°C for 35 min) Day 4, AM: Novel environment w/unknown rat (15 min) Day 4, PM: Cold Day 5, AM: Restraint (2 hours) Day 5, PM: Rotation (2 hours) ⫹ overnight isolation (one per cage) Day 6, AM: Restraint (2 hours) Day 6, PM: Rotation Day 7, AM: Open field test (5 min) Day 7, PM: Warm swim (35 min)

BIOL PSYCHIATRY 2003;53:166 –174

167

Control rats were handled for 1 min twice daily. All animals were sacrificed between 9 –10 AM on day 8, 16 hours after the last stress exposure. Core blood samples were collected in ethylenediaminetetraacetic acid (EDTA)-coated Vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ) and spun at 1500 g to separate serum for hormone assays. Brains were removed and frozen in isopentane cooled to ⫺40°C on dry ice. Adrenal and thymus glands were removed, cleaned, and weighed. It should be noted that oral administration was selected to eliminate stress associated with surgery and implantation of miniosmotic pumps. A preliminary study in our laboratory revealed that large-sized osmopumps are required to deliver an equivalent dose (25 mg/kg) of reboxetine. We were concerned that osmopumps of this size would constitute a physical stressor and confound the interpretation of our behavioral and neurochemical assessments. The 7-day pretreatment protocol was based on preliminary studies in the laboratory, showing efficacy of reboxetine on elevated plus maze behavior as early as 7 days following initiation of treatment.

Open Field Behavior The open field sessions on day 7 were videotaped for behavioral analysis. The open field apparatus was a 36 ⫻ 36 –inch white Plexiglas enclosure, divided into 36 squares of equal size. Animals were placed into the apparatus and allowed to explore the environment for 5 min. Videotapes were scored for total maze ambulation, ambulation in the central squares (i.e., squares not adjacent to the outer walls), rearing, and grooming over the 5-min period.

In Situ Hybridization Brains were sectioned at 15 ␮m on a Hacker-Bright (Fairfield, NJ) cryostat. Sections were fixed by immersion in 4% phosphatebuffered paraformaldehyde for 10 min. Slides were then rinsed in 5 mmol/L potassium phosphate-buffered saline (KPBS) for 10 min, followed by 5 mmol/L KPBS with 2% glycine for 10 min, and back into 5 mmol/L KPBS for 10 min. Slides were then acetylated in 0.1 mol/L tetraethylammonium (pH 8.0) containing 0.25% acetic anhydride. Following pretreatment, slides were rinsed twice for 5 min each time in 0.2 ⫻ SSC, dehydrated through graded ethanols and delipidated in chloroform. Hybridizations used probes complementary to glutamic acid decarboxylase (GAD) 65 mRNA (approximately 850 base pair [bp] sequence encoding 3⬘ untranslated and 3⬘ coding region) and GAD67 mRNA (approx. 900 bp sequence encoding 3⬘ untranslated and 3⬘ coding region) mRNAs (courtesy Alan Tobin, University of California, Los Angeles). These probes have been previously characterized (Esclapez et al 1994; Herman et al 1989). Probes were synthesized by in vitro transcription using T3 or T7 RNA polymerase (Roche, Indianapolis, IN) and 35S-UTP (Amersham). Labeling reactions included 60 ␮Ci 35S-UTP, 1 ⫻ transcription buffer (Boehringer Mannheim), 15 mmol/L DTT, 200 ␮mol/L GTP, CTP and ATP, 10 ␮mol/L UTP, 40 units placental RNase inhibitor (40 units/␮l) (Roche), 1 ␮g linearized plasmid DNA, and 20 units of appropriate RNA polymerase (T3 or T7, Roche). Labeling reactions were incubated for 90 min at 37°C, the template DNA digested with 12 units of RNase-free

168

J.P. Herman et al

BIOL PSYCHIATRY 2003;53:166 –174

Table 1. Physiologic Effects of Oral Reboxetine Water

Initial Body Weight (g) Final Body Weight (g) Adrenal Weight (mg/100 g Initial BW) Thymus Weight (mg/100 g Initial BW) Plasma Corticosterone (ng/mL)

Reboxetine

Handled

Stressed

Handled

Stressed

290 ⫾ 5 348 ⫾ 6 13.0 ⫾ 1.1 129.9 ⫾ 9.4 41.9 ⫾ 18.2

281 ⫾ 4 314 ⫾ 5a,b 13.7 ⫾ .9 116.5 ⫾ 14.8 93.9 ⫾ 23.2a

280 ⫾ 1 318 ⫾ 2 14.0 ⫾ .7 135.7 ⫾ 12.1 46.2 ⫾ 9.0

285 ⫾ 2 299 ⫾ 4a,b 16.8 ⫾ 1.2b 127.0 ⫾ 4.9 92.1 ⫾ 33.9a

BW, body weight. a Significantly different from corresponding water group, p ⬍ .05. b Significantly different from corresponding handled group, p ⬍ .05.

DNase I (Roche), and the labeled probe separated from free nucleotide by ammonium acetate precipitation. For hybridization, labeled probes were diluted in a hybridization buffer containing 55% deionized formamide, 10% dextran sulfate, 10 mmol/L Tris-HCl buffer, 0.5 mmol/L EDTA, 0.5 ⫻ Denhardt’s, 0.17 mol/L NaCl, 0.3 mg/mL herring sperm DNA, 150 ␮g/mL yeast tRNA, and 20 mmol/L DTT to yield 1,000,000 cpm/50 ␮L. Fifty ␮L of the hybridization mixture was pipetted onto each slide and the slides then coverslipped. Slides were incubated overnight at 55°C in sealed polystyrene boxes lined with filter paper moistened with 50% formamide. Coverslips were removed, the slides rinsed briefly in 2 ⫻ SSC, and then immersed in 2 ⫻ SSC for 20 min. Slides were then immersed in RNase A (100 ␮g/mL) at 37°C for 30 min. Sections were then washed three times in 0.2 ⫻ SSC for 10 min each time, followed by a 60-min wash at 65°C. Finally, tissue was dehydrated through graded ethanols before being exposed to Kodak BioMAX x-ray film (Rochester, NY) (bed nucleus of the stria terminalis [BST]/forebrain sections) or Packard (Indianapolis, IN) phosphorimager screens (hippocampus/amygdala sections).

Corticosterone Assay Serum corticosterone assays were formed using double antibody radioimmunoassay kits from ICN (Costa Mesa, CA), using 125I corticosterone as trace.

Data Analysis In situ hybridization data were analyzed using National Insititutes of Health Image 1.62 software for Macintosh (x-ray film) or Packard OptiQuant (Phosphorimager) software. GAD65/67 determinations were made at three levels of the forebrain. In the anterior forebrain, GAD mRNA expression was assessed in the anteromedial, ventromedial, and posteromedial subnuclei of the BST and the lateral septum. In the midforebrain, GAD65/67 mRNA hybridization signal was assessed in the medial and central amygdaloid nuclei, dorsomedial hypothalamic nucleus, and reticular thalamic nucleus. Finally, subfields CA1, CA3, and dentate gyrus measurements were taken in the dorsal hippocampal region of the caudal forebrain. All regions were defined according to the rat brain atlas of Paxinos and Watson (1986). Background signal was sampled over nonhybridized regions of tissue and subtracted from all regions to obtain corrected digital luminescence units/mm2 (phosphorimager) or gray level units/ mm2 (film densitometry). Film densitometry was performed on

anterior sections to afford superior resolution of individual subregions of the BST. Because of the large number of areas that expressed GAD65/67 mRNA, separate analyses were performed for anterior forebrain, midforebrain, and hippocampal regions. Data were analyzed by three-way analysis of variance (ANOVA), using region, stress, and drug as independent variables. Following significant drug, stress, or interaction effects, planned comparisons (Fisher’s Protected Least Significant Difference [PLSD] tests) were used to assess effects of drug and stress within each region analyzed. Note that due to sectioning, individual regions may not have been available for each subject. For a region to be analyzed, we required ns ⱖ 4/group. Behavioral data were analyzed using the Mann–Whitney U test.

Results Oral Reboxetine Dosage Average reboxetine dose per animal was 25.0 ⫾ 0.5 mg/kg/day, calculated by dividing total water consumed per cage by the number of animals per cage. As animals were housed three per cage, individual water consumption data could not be assessed.

Response to Chronic Variable Stress Exposure A chronic variable stress regimen was used to assess the effects of reboxetine on measures of organismic stress activation. In accordance with previous reports, chronic stress decreased body weight gain [F(1,20) ⫽ 126.9, P ⬍ .05] and increased plasma corticosterone levels [F(1,20) ⫽ 4.49, P ⬍ .05] in both water- and reboxetine-treated groups. Whereas initial body weight did not differ among the groups, reboxetine significantly decreased body weight in both stressed and unstressed groups to an equivalent extent (Table 1). No effects of stress on adrenal or thymus weight were seen, perhaps owing to the relatively short duration of stress exposure (1 week).

Open Field Behavior As the distribution of ambulation and grooming scores were not normally distributed, analysis of open field behavior was performed by the Mann–Whitney U test. Reboxetine significantly increased the frequency of am-

NE–GABA Interactions in Stress Circuitry

BIOL PSYCHIATRY 2003;53:166 –174

169

Figure 1. Effects of reboxetine (Rebox) on open field behavior in chronically stressed rats. Delivery of reboxetine via the drinking water did not affect total ambulation in an open field, but did increase ambulation in the central squares of the apparatus (p ⬍ .05, Mann– Whitney U test), indicative of reduced behavioral reactivity to novelty. Incidence of grooming was not affected by reboxetine. * indicates significant results.

bulation in the central region of the open field (P ⬍ .05). Stressed reboxetine animals did not differ from stressed water-treated rats in total ambulation or number of grooms or rears (Figure 1). Note that unstressed animals were not behaviorally tested, because novelty would constitute an acute stressor that would have interfered with interpretation of the chronic stress data.

mRNA expression in stressed reboxetine-treated animals relative to stressed control animals. No changes were observed in the central amygdaloid nucleus or reticular thalamic nucleus, indicating that the significant changes were region specific. Finally, there was a significant effect of drug on GAD67 mRNA levels in the hippocampus as well [F(1,51) ⫽ 8.27, P ⬍ .05]. Significant reduction of GAD67 mRNA expression was observed in the dentate

Forebrain GAD mRNA Expression Expression of GAD65 and GAD67 mRNAs were assessed in forebrain regions implicated in hypothalamo–pituitary– adrenocortical (HPA) integration. The distribution of GAD65 and 67 mRNA in stress-relevant forebrain regions are illustrated in Figure 2. Both isoforms of GAD are richly expressed in projection neurons of the lateral septum, all subdivisions of the BST, and the medial and central amygdaloid nuclei. Both GAD65 and 67 are localized in CA1, CA3, and dentate gyrus of the hippocampus and in the cerebral cortex; GAD is believed to be localized primarily to interneurons in these regions (Mugnaini and Oertel 1985; Woodson et al 1989). Semi-quantitative analyses of forebrain GAD expression are illustrated in Figures 3–7. The data suggest site-specific modulation of these species in regions known to integrate behavioral and neuroendocrine responses to stress. Because of the large number of regions under examination, anterior forebrain, mid-forebrain, and hippocampal regions analyses were performed in three separate runs. Densitometric analysis of GAD67 mRNA over the anterior forebrain (lateral septum, anteromedial BST, ventromedial BST, and posteromedial BST) revealed significant effects of reboxetine administration [F(1,72) ⫽ 5.34, P ⬍ .05] and stress [F(1,72) ⫽ 5.04, P ⬍ .05] on GAD67 mRNA expression. No effect of region was observed. Planned analyses within regions revealed significant decreases in GAD67 mRNA in stressed reboxetine animals relative to stressed control animals in the posteromedial BST (Ps ⬍ 0.05, Fisher’s PLSD test). A significant overall effect of drug was observed in the diencephalic and amygdalar analyses [F(1,66) ⫽ 12.1, P ⬍ .05]. Subsequent comparisons revealed significant decreases in medial amygdalar and dorsomedial hypothalamic GAD67

Figure 2. Localization of glutamic acid decarboxylase (GAD) 65 and GAD67 mRNAs in forebain stress circuitry. Images obtained from 35S-sensitive phosphorimager screens. GAD67 (left) and GAD65 (right) mRNAs are highly expressed in numerous stress-relevant CNS regions, including the lateral septum (LS), bed nucleus of the stria terminalis (BST), CA1, dentate gyrus (DG), central amygdaloid nucleus (CEA), and medial amygdaloid nucleus (MeA). GAD is highly expressed in non–stressrelated regions (e.g., reticular thalamic nucleus [RTN]) as well.

170

BIOL PSYCHIATRY 2003;53:166 –174

Figure 3. Regulation of glutamic acid decarboxylase (GAD) 67 mRNA in medial amygdaloid nucleus (MeA), central amygdaloid nucleus (CeA), dorsomedial hypothalamus (DMH), and reticular thalamic nucleus (RTN). Semi-quantitative analysis revealed significant effects of drug on GAD67 mRNA expression in the MeA and DMH; post hoc analysis revealed differences between water and reboxetine (Rebox) groups following stress exposure. No drug or stress effects were seen in the CeA or RTN. * indicates significant results.

gyrus of stressed, reboxetine-treated rats relative to stressed control rats. In contrast with GAD67, GAD65 mRNA expression was not affected by reboxetine or stress in the midforebrain or hippocampal datasets. There was a significant region ⫻ stress ⫻ drug interactions effect in the anterior forebrain [F(3,54) ⫽ 5.48, P ⬍ .05]. Subsequent analysis (Fisher’s PLSD test) revealed that GAD65 mRNA was increased in stressed reboxetine-treated animals relative to both handled and reboxetine-treated animals and stressed control animals; in addition, reboxetine decreased GAD65 mRNA expression in handled reboxetine-treated animals relative to handled control animals.

Discussion Selective inhibition of NE reuptake by reboxetine attenuates behavioral sequelae of chronic stress exposure. Behavioral reactivity to an open field is reduced in chronically stressed animals treated with reboxetine relative to control animals. Reduced behavioral reactivity is accompanied by changes in central GABAergic neurotransmission in the medial amygdala, hippocampus, and lateral

J.P. Herman et al

Figure 4. Regulation of glutamic acid decarboxylase (GAD) 65 mRNA in the medial amygdaloid nucleus (MeA), central amygdaloid nucleus (CeA), dorsomedial hypothalamus (DMH), and reticular thalamic nucleus (RTN). There was no effect of drug on GAD65 mRNA in any region examined.

septum, suggesting that central GABAergic pathways may mediate the stress-reducing effects of NE reuptake inhibition. Efficacy of reboxetine on behavioral performance is in line with previous reports. In humans, reboxetine has antidepressant efficacy, alone or in combination with fluoxetine, and appears to have anxiolytic actions as well (Massana 1998; Schatzberg 2000). In rodents, reboxetine attenuates immobility and defecation in the forced swim test, with effects similar to those seen following administration of selective serotonin reuptake inhibitors (Connor et al 1999; Wong et al 2000). The selective effects of reboxetine on behavior but not HPA axis measures may be related to temporal aspects of the stimulus. Attenuation of behavioral reactivity indicates a decrease in acute stress responsivity, whereas elevated morning glucocorticoid levels represent a cumulative response to repeated stress, which may be more refractory to the effects of NE blockade. Analysis of corticosterone release following open field exposure would be a more sensitive measure of reboxetine effects on HPA axis activity. Reboxetine effects on GABAergic neurotransmission are consistent with the observed effects on behavior. Animals treated with reboxetine show preferential downregulation of GAD67 mRNA expression in the hippocampal dentate gyrus, medial amygdala, and posteromedial

NE–GABA Interactions in Stress Circuitry

Figure 5. Regulation of glutamic acid decarboxylase (GAD) 67 mRNA in the lateral septum (LS) and posteromedial (pm), anteromedial (am), or ventromedial (vm) subdivisions of the bed nucleus of the stria terminalis (BST). Semi-quantitative analysis revealed significant effects of drug on GAD67 mRNA expression in the BSTpm; posthoc analysis revealed differences between water and reboxetine (Rebox) groups following stress exposure. No drug or stress effects were seen in the LS, BSTam or BSTvm. * indicates significant results.

BST following stress, and show enhanced GAD65 synthesis in the lateral septum following stress. All of these regions are implicated in central regulation of stress responses and anxiety (see Figure 8). The anatomical pattern of reboxetine effects on GAD mRNA expression is consistent with a putative antistress activity. For example, reboxetine-induced decreases in GAD67 expression in the medial amygdala and the posteromedial BST of chronically stressed rats suggests reduced outflow from these regions of the extended amygdala. Notably, these regions contain substantial populations of GABAergic projection neurons; indeed, the abundance of GABA neurons in these regions (Swanson and Petrovich 1998) (⬎90% throughout the BST [Cullinan et al 1993]) indicates that a large component of amygdalar function is accomplished by disinhibition. (Evidence for this mode of communication has been well characterized in the extrapyramidal system, where striato–pallido–thalamic GABA–GABA disinhibitory circuits form an essential component of striatal outflow [Graybiel 2000].) As such, loss of GABA synthesis predicts reduced impact of these regions on downstream systems. Both the medial

BIOL PSYCHIATRY 2003;53:166 –174

171

Figure 6. Regulation of glutamic acid decarboxylase (GAD) 65 mRNA in the lateral septum (LS) and posteromedial (pm), anteromedial (am) or ventromedial (vm) subdivisions of the bed nucleus of the stria terminalis (BST). Semi-quantitative analysis revealed significant drug by stress interaction in the LS; post hoc analysis revealed a significant increase in GAD65 mRNA expression in the reboxetine (Rebox) group following stress exposure. No drug or stress effects were seen in the BST. ‡ indicates significant results.

amygdala and its posteromedial BST extension have been implicated in stress– excitation; indeed, lesions of the medial amygdala and BST reduce behavioral reactivity and attenuate neuroendocrine stress responses (Feldman et al 1990, 1994; Gray et al 1993; Herman et al 1994), whereas stimulation produces stress-like behavioral, autonomic, and endocrine responses (Casada and Dafny 1991; Dunn 1987; Dunn and Whitener 1986; Dunn and Williams 1995). Indeed, previous studies indicate that NE inhibits 70% of BST neurons (Casada and Dafny 1993), consistent with NE inhibition of a predominantly GABAergic projection system. Thus, reduced capacity for GABA synthesis may reflect a loss of GABAergic outflow, reducing the stress– excitatory impact of the amygdalar system. In the hippocampus, the vast majority of GABAergic cells are interneurons (Feldblum et al 1993; Mugnaini and Oertel 1985; Woodson et al 1989), involved in inhibition of glutamatergic hippocampal pyramidal cells. These GABAergic cells are known to be regulated by stressful stimuli, showing enhanced GAD65 and/or GAD67 mRNA levels following acute or chronic stress (Bowers et al 1998). As the hippocampus inhibits responses to stress (Jacobson and Sapolsky 1991), enhanced inhibition may

172

BIOL PSYCHIATRY 2003;53:166 –174

Figure 7. Regulation of glutamic acid decarboxylase (GAD) 67 and GAD65 mRNA in hippocampal subfield CA1, dentate gyrus, and frontoparietal cortex. Semi-quantitative analysis revealed significant effects of drug on GAD67 mRNA expression in the dentate gyrus; this effect was marked by a significant reduction in GAD67 levels following stress in reboxetine (Rebox)-treated animals. There was no significant effect of drug or stress in CA1 or CA3. GAD65 mRNA levels were not affected by drug or stress. * indicates significant results.

reduce hippocampal outflow and contribute to behavioral and endocrine hyperreactivity seen with chronic stress. Importantly, reboxetine-induced reduction in GAD67 synthesis in the dentate gyrus may have the opposite effect; loss of GABA would be predicted to enhance inhibitory hippocampal output to stress effector systems. The lateral septum is the only region to show differential regulation of GAD65 mRNA following reboxetine treatment in the context of chronic stress, and is the only region to show reboxetine-driven up-regulation of either GAD isoform. Involvement of this region in stress integration is highlighted by observed neuronal c-fos activation by a variety of psychogenic (but not interoceptive) stressors (c.f., Cullinan et al 1995; Emmert and Herman 1999; Thrivikraman et al 1997). Importantly, stress-induced c-fos induction is reduced in rats showing learned helplessness, an animal model of depressive illness (Steciuk et al 1999). Recent studies indicate that lateral septal serotonin release is negatively correlated with learned immobility in the forced swim test (Kirby and Lucki 1998), suggesting a positive connection between serotonin release and coping behavior in rats. An overall

J.P. Herman et al

inhibitory role for the lateral septum in anxiety and mood has been postulated for some time (Chozick 1985). Thus, selective increase in GAD65 may reflect enhanced lateral septal GABA output during stress, consistent with inhibitory actions on central stress integration. Reboxetine administration does not decrease GAD mRNA levels in all stress effector regions. Most notably, GAD67 levels in the central amygdaloid nucleus and related BST subdivisions are not differentially reduced by reboxetine following stress exposure. These regions are implicated in stress regulation, fear, and anxiety (Davis 1998), and receive rich noradrenergic innervation from the locus coeruleus (c.f., Moore and Bloom 1979). Thus, central amygdala GABA neurons appear to be insensitive to NE reuptake inhibition and may thus be functionally intact during chronic stress exposure. Similarly, there is no significant effect of reboxetine on GABAergic interneurons in pyramidal cell layers of the hippocampus following stress. As such, reboxetine may preferentially affect subdomains of amygdalar and hippocampal function, perhaps in relation to the density of the noradrenergic innervation on GAD neurons contained therein. The preferential effect of reboxetine on GAD67 mRNA has functional implications. For example, GAD65 and 67 are differentially stored and activated; GAD65 is predominantly present in terminals as inactive apoenzyme, whereas a substantially greater proportion of GAD67 is present as active holoenzyme, enriched in cell bodies (Feldblum et al 1993; ). Thus, the selective changes in limbic GAD67 mRNA seen following reboxetine treatment suggest a decrease in the resting level of activated, rather than stored, GAD activity. In contrast, elevated GAD65 mRNA in lateral septum suggests increased terminal storage of GAD, and hence a capacity for increased release in target structures. It is important to note that the current studies examine GABA synthesis at the level of steady-state GAD mRNA levels. Although GAD levels are generally predictive of GABA function (Segovia et al 1990), they represent only one aspect of the molecular cascade culminating in enzyme activity. Both GAD enzymes are activated by phosphorylation, and thus the net levels of RNA predict capacity to make GAD but not direct GAD activity. The latter needs to be assessed by biochemical assessment of GAD activity and/or measurement of GABA levels by microdialysis. The current anatomical studies thus provide a guide for further assessment of the role of NE–GABA interactions in central stress integration. The capacity for reboxetine to alter poststress GABA synthesis is in line with previous studies documenting effects on serotonin and dopamine systems. Previous studies indicate that reboxetine does not appear to affect resting serotonin or dopamine release in cortex, hippocam-

NE–GABA Interactions in Stress Circuitry

BIOL PSYCHIATRY 2003;53:166 –174

173

Figure 8. Schematic diagram of possible norepinephrine (NE)–␥ aminobutyric acid (GABA) interactions in limbic stress circuitry. Ascending noradrenergic neurons project to hippocampus, bed nucleus of the stria terminalis (BST), amygdala and, to a lesser extent, the lateral septum (LS). Our results indicate that increased NE availability decreases glutamic acid decarboxylase (GAD) 67 mRNA in dentate gyrus, posteromedial BST, and medial amygdaloid nucleus (MeA), and increases GAD65 mRNA in the lateral septum. The results of decreased GABA in hippocampus suggests enhanced hippocampal activation; because hippocampus generally inhibits stress effector systems, the net result is reduced stress activation. Decreased GABA synthesis in MeA and BST would impact primarily projection neurons, resulting in less GABAergic outflow from these regions. The net results is a loss of stress excitation (or stress disinhibition). Increased GAD65 synthesis in the LS predicts enhanced activity of GABAergic neurons; these neurons are believed to inhibit stress effectors. Thus, NE reuptake blockade stands to modulate GABAergic tone by both enhanced inhibition and reduced excitation of limbic system pathways controlling neuroendocrine, autonomic and behavioral stress effectors. HPC, hippocampus; LC, locus coeruleus; HPA axis, hypothalamic–pituitary–adrenocortical axis.

pus, or striatum; however, reboxetine attenuates acute stress-induced enhancement of amygdalar and cortical serotonin turnover and prefrontal cortex dopamine release (Connor et al 1999; Sacchetti et al 1999). In combination, the data suggest that enhanced noradrenergic tone preferentially modulates the responsivity of central stress circuits rather than providing a nonspecific enhancement of neurotransmission. As such, it is likely that the effects of enhanced NE availability are most pronounced during periods of stress-circuit activation. The authors acknowledge the expert technical assistance of Mark Dolgas, Megan Paskitti, and Marshall Ney. This work was supported by a grant from Pharmacia-Upjohn.

References Anand A, Charney DS (2000): Norepinephrine dysfunction in depression. J Clin Psychiatry 61:16 –24. Bowers G, Cullinan WE, Herman JP (1998): Region-specific regulation of glutamic acid decarboxylase (GAD) mRNA expression in central stress circuits. J Neurosci 18:5938 – 5947. Casada JH, Dafny N (1991): Restraint and stimulation of the bed nucleus of the stria terminalis produce similar stress-like behaviors. Brain Res Bull 27:207–212. Casada JH, Dafny N (1993): Responses of neurons in bed nucleus of the stria terminalis to microiontophoretically applied morphine, norepinephrine and acetylcholine. Neuropharmacology 32:279 –284.

Chozick BS (1985): The behavioral effects of lesions of the septum: A review. Int J Neurosci 26:197–217. Connor TJ, Kelliher P, Harkin A, Kelly JP, Leonard BE (1999): Reboxetine attenuates forced swim test-induced behavioural and neurochemical alterations in the rat. Eur J Pharmacol 379:125–133. Cullinan WE, Herman JP, Battaglia DF, Akil H, Watson SJ (1995): Pattern and time course of immediate early gene expression in rat brain following acute stress. Neuroscience 64:477–505. Cullinan WE, Herman JP, Watson SJ (1993): Ventral subicular interaction with the hypothalamic paraventricular nucleus: Evidence for a relay in the bed nucleus of the stria terminalis. J Comp Neurol 332:1–20. Davis M (1998): Are different parts of the extended amygdala involved in fear versus anxiety? Biol Psychiatry 44:1239 –1247. Delgado PL, Moreno FA (2000): Role of norepinephrine in depression. J Clin Psychiatry 61:5–12. Dunn JD (1987): Plasma corticosterone responses to electrical stimulation of the bed nucleus of the stria terminalis. Brain Res 407:327–331. Dunn JD, Whitener J (1986): Plasma corticosterone responses to electrical stimulation of the amygdaloid complex: Cytoarchitectonic specificity. Neuroendocrinology 42:211–217. Dunn JD, Williams TJ (1995): Cardiovascular responses to electrical stimulation of the bed nucleus of the stria terminalis. J Comp Neurol 352:227–234. Emmert MH, Herman JP (1999): Differential forebrain c-fos mRNA induction by ether inhalation and novelty: Evidence for distinctive stress pathways. Brain Res 845:60 –67. Esclapez M, Tillakaratne NJK, Kaufman DL, Tobin AJ, Houser CR (1994): Comparative localization of two forms of glu-

174

BIOL PSYCHIATRY 2003;53:166 –174

tamic acid decarboxylase and their mRNAs in rat brain supports the concept of functional differences between the forms. J Neurosci 14:1835–1855. Feldblum S, Erlander MG, Tobin AJ (1993): Different distributions of GAD65 and GAD67 mRNAs suggest that the two glutamate decarboxylases play distinctive functional roles. J Neurosci Res 34:689 –706. Feldman S, Conforti N, Itzik A, Weidenfeld J (1994): Differential effect of amygdaloid lesions of CRF-41, ACTH and corticosterone responses following neural stimuli. Brain Res 658:21–26. Feldman S, Conforti N, Saphier D (1990): The preoptic area and bed nucleus of the stria terminalis are involved in the effects of the amygdala on adrenocortical secretion. Neuroscience 37:775–779. Gray TS, Piechowski RA, Yracheta JM, Rittenhouse PA, Bethea CL, Van de Kar LD (1993): Ibotenic acid lesions in the bed nucleus of the stria terminalis attenuate conditioned stressinduced increases in prolactin, ACTH and corticosterone. Neuroendocrinology 57:517–524. Graybiel AM (2000): The basal ganglia. Curr Biol 10:R509 – R511. Herman JP, Adams D, Prewitt CM (1995): Regulatory changes in neuroendocrine stress-integrative circuitry produced by a variable stress paradigm. Neuroendocrinology 61:180 –190. Herman JP, Cullinan WE, Watson SJ (1994): Involvement of the bed nucleus of the stria terminalis in tonic regulation of paraventricular hypothalamic CRH and AVP mRNA expression. J Neuroendocrinol 6:433–442. Herman JP, Scha¨ fer MK-H, Young EA (1989): Evidence for hippocampal regulation of neuroendocrine neurons of the hypothalamo-pituitary-adrenocortical axis. J Neurosci 9:3072–3082. Jacobson L, Sapolsky RM (1991): The role of the hippocampus in feedback regulation of the hypothalamo-pituitary-adrenocortical axis. Endocrine Rev 12:118 –134. Kirby LG, Lucki I (1998): The effect of repeated exposure to forced swimming on extracellular levels of 5-hydroxytryptamine in the rat. Stress 2:251–263. Lopez JF, Vazquez DM, Chalmers DT, Watson SJ (1997): Regulation of 5-HT receptors and the hypothalamic-pituitaryadrenal axis: Implications for the neurobiology of suicide. Ann N Y Acad Sci 836:106 –134. Massana J (1998): Reboxetine versus fluoxetine: An overview of efficacy and tolerability. J Clin Psychiatry 59:8 –10. Montgomery SA, Schatzberg AF (1998): Introduction. Reboxetine: A new selective antidepressant for the treatment of depression. J Clin Psychiatry 59:3. Moore RY, Bloom FE (1979): Central catecholamine neuron systems: Anatomy and physiology of the norepinephrine and epinephrine systems. Annu Rev Neurosci 2:113–168. Moore RY, Card JP (1984): Noradrenaline-containing neuron systems. In: Bjo¨ rklund A, Ho¨ kfelt T, editors. Handbook of Chemical Neuroanatomy, Vol. 2: Classical Transmitters in the CNS, Part II. Amsterdam: Elsevier, pp 1–124. Mugnaini E, Oertel WH (1985): An atlas of the distribution of GABAergic neurons and terminals in the rat CNS as revealed by GAD immunohistochemistry. In: Bjo¨ rklund A, Ho¨ kfelt T, editors. GABA and Neuropeptides in the CNS, Vol. 1. Amsterdam: Elsevier, pp 436 –622.

J.P. Herman et al

Nestler EJ, Alreja M, Aghajanian GK (1999): Molecular control of locus coeruleus neurotransmission. Biol Psychiatry 46:1131–1139. Nisenbaum LK, Abercrombie ED (1993): Presynaptic alterations associated with enhancement of evoked release and synthesis of norepinephrine in hippocampus of chronically coldstressed rats. Brain Res 608:280 –287. Paskitti ME, McCreary BJ, Herman JP (2000): Stress regulation of adrenocorticosteroid receptor gene transcription and mRNA expression in rat hippocampus: Time-course analysis. Brain Res Mol Brain Res 80:142–152. Paxinos G, Watson C (1986): The Rat Brain in Stereotaxic Coordinates, 2nd ed. New York: Academic Press. Petty F (1995): GABA and mood disorders: A brief review and hypothesis. J Affect Disord 34:275–281. Sacchetti G, Bernini M, Bianchetti A, Parini S, Invernizzi RW, Samanin R (1999): Studies on the acute and chronic effects of reboxetine on extracellular noradrenaline and other monoamines in the rat brain. Br J Pharmacol 128:1332–1338. Schatzberg AF (2000): Clinical efficacy of reboxetine in major depression. J Clin Psychiatry 61:31–38. Segovia J, Tillakaratne NJ, Whelan K, Tobin AJ, Gale K (1990): Parallel increases in striatal glutamic acid decarboxylase activity and mRNA levels in rats with lesions of the nigrostriatal pathway. Brain Res 529:345–348. Shiah IS, Yatham LN (1998): GABA function in mood disorders: An update and critical review. Life Sci 63:1289 –1303. Steciuk M, Kram M, Kramer GL, Petty F (1999): Decrease in stress-induced c-Fos-like immunoreactivity in the lateral septal nucleus of learned helpless rats. Brain Res 822:256 – 259. Swanson LW, Petrovich GD (1998): What is the amygdala? Trends Neurosci 21:323–331. Tanaka M, Yoshida M, Emoto H, Ishii H (2000): Noradrenaline systems in the hypothalamus, amygdala and locus coeruleus are involved in the provocation of anxiety: Basic studies. Eur J Pharmacol 405:397–406. Thrivikraman KV, Su Y, Plotsky PM (1997): Patterns of fosimmunoreactivity in the CNS induced by repeated hemorrhage in conscious rats: Correlations with pituitary-adrenal axis activity. Stress 2:145–158. Valentino RJ, Curtis AL, Page ME, Pavcovich LA, FlorinLechner SM (1998): Activation of the locus ceruleus brain noradrenergic system during stress: Circuitry, consequences, and regulation. Adv Pharmacol 42:781–478. Versiani M, Mehilane L, Gaszner P, Arnaud-Castiglioni R (1999): Reboxetine, a unique selective NRI, prevents relapse and recurrence in long-term treatment of major depressive disorder. J Clin Psychiatry 60:400 –406. Watanabe Y, McKittrick CR, Blanchard DC, Blanchard RJ, McEwen BS, Sakai RR (1995): Effects of chronic social stress on tyrosine hydroxylase mRNA and protein levels. Mol Brain Res 32:176 –180. Wong EH, Sonders MS, Amara SG, Timholt PM, Piercey MF, Hoffman WP, et al (2000): Reboxetine: A pharmacologically potent, selective, and specific norepinephrine reuptake inhibitor. Biol Psychiatry 47:818 –829. Woodson W, Nitecka L, Ben-Ari Y (1989): Organization of the GABAergic system in the rat hippocampal formation: A quantitative immunocytochemical study. J Comp Neurol 280:254 –271.