stimulation of benzodiazepine receptors in the dorsal hippocampus and median raphé reveals...

European Journal of Neuroscience, Vol. 10, pp. 3673–3680, 1998 © European Neuroscience Association

Stimulation of benzodiazepine receptors in the dorsalhippocampus and median raphe reveals differentialGABAergic control in two animal tests of anxiety

Luis E. Gonzalez, Abdel-Mouttalib Ouagazzal and Sandra E. FilePsychopharmacology Research Unit, UMDS, Guy’s Hospital, London SE1 9RT, UK

Keywords: GABA, benzodiazepine receptors, anxiety, midazolam, flumazenil, rat

Abstract

The effects of pharmacological challenges to the benzodiazepine receptors in the dorsal hippocampus andmedian raphe nucleus were investigated in the social interaction and the elevated plus-maze tests of anxiety inrats. In the social interaction test, bilateral administration of midazolam (1 and 2 µg), into the dorsalhippocampus had anxiolytic effects; flumazenil (500 ng) was silent, but was able to antagonize the anxiolyticeffects of midazolam (2 µg). In the social interaction test, midazolam was also anxiolytic when infused into themedian raphe nucleus; flumazenil (100 and 500 ng) increased locomotor activity, but did not change anxietymeasures. As an anatomical control, midazolam (1 and 2 µg) was infused into the adjacent pontine reticularnucleus, and was without effect. In contrast to the social interaction test, local infusion of midazolam (1 and2 µg) and flumazenil (100 and 500 ng) into either the dorsal hippocampus or the median raphe nucleus failed tochange anxiety measures in the elevated plus-maze (trials 1 and 2). These results show that stimulation of thebenzodiazepine receptors in the hippocampus or the median raphe nucleus leads to anxiolytic effects in thesocial interaction test, but not in the elevated plus-maze. It would therefore appear that the two tests detectdifferent types of anxiety that are differentially modulated by GABAA-benzodiazepine receptors in the dorsalhippocampus and the median raphe nucleus.

Introduction

Benzodiazepines are believed to induce their anxiolytic effects byacting on the GABAA receptor complex and enhancing the inhibitoryactions of GABA in the brain (Sieghart, 1992). It has been suggestedthat these effects are due, at least in part, to inhibitory actions on theserotonergic system in the raphe´ nuclei and its projection areas (Steinet al., 1973; Gallager, 1978; Thiebotet al., 1980; Hodgeset al., 1987;Higgins et al., 1988; Peiet al., 1989; Graeffet al., 1996 ). However,most of the behavioural studies using direct administration of benzo-diazepines in the brain have mainly targeted the dorsal raphe´ nucleusand one of its main projection areas, the amygdala. Such experimentshave shown that stimulation of GABAA-benzodiazepine receptors inboth the dorsal raphe´ (Thiebot et al., 1980; Higginset al., 1988;Gonzalez & File, 1997) and the amygdala (Hodgeset al., 1987;Graeff et al., 1993; Green & Vale, 1995; Pesold & Treit, 1995;Sanders & Shekhar, 1995 Gonzalezet al., 1996) can induce anxiolyticeffects in several models of anxiety. Benzodiazepine receptors arealso present in several other brain structures that play a central rolein the modulation of anxiety, such as the median raphe´ nucleus andits major projection area, the dorsal hippocampus (Plazniket al.,1994; File et al., 1996; Andrewset al., 1994, 1997). Yet it issurprising that studies to assess effects of direct administration ofbenzodiazepines to the dorsal hippocampus and the median raphe´nucleus have not been conducted in ethologically based tests, suchas the plus-maze and social interaction tests.

Correspondence: Professor S. E. File, as above. E-mail: [email protected]

Received 1 April 1998, revised 3 July 1998, accepted 10 July 1998

The present study was designed to investigate the roles of thedorsal hippocampus and the median raphe´ nucleus in mediating theanxiolytic effects of benzodiazepines and to establish whether theirrelative importance may vary during distinct behavioural situations.In the social interaction test, animals were tested independently underseveral conditions to evaluate possible changes in the effects producedby benzodiazepine receptor ligands. Because anxiolytic effects ofsystemic administration of benzodiazepines can be detected in bothhigh-light, unfamiliar (HU) and high-light, familiar (HF) test condi-tions (File, 1980) we used both of these to evaluate the potentialanxiolytic effects of midazolam after central administration. The low-light, familiar (LF) condition in which systemic benzodiazepines arewithout effect (File & Hyde, 1978) was also used as a control testcondition.

In the elevated plus-maze, systemic administration of benzo-diazepines induce anxiolytic effects on the first 5 min test (trial 1),but not when the animals are tested a second time (trial 2) (File,1990). Increasing evidence suggest that two different types (or states)of anxiety are generated by the plus-maze on trials 1 and 2 (File,1990; Rodgerset al., 1992; File & Zangrossi, 1993; Fileet al., 1993b).In addition, in previous studies we have found that as little as a 5-min experience in the elevated plus-maze changes the benzodiazepinereceptors in the dorsal raphe´ nucleus to an inverse agonist state

3674 L. E. Gonzalezet al.

FIG. 1. (Top) Diagrammatic representation of a coronal section (3.3 mmposterior to bregma) of the rat brain showing the target area of the dorsalhippocampus as shaded. (Bottom) Mean (6 SEM) time spent in socialinteraction by rats tested in the high-light, unfamiliar (HU); high-light, familiar(HF) and low-light, familiar (LF) arenas after bilateral injection into the dorsalhippocampus of saline (V), midazolam (1 and 2µg), flumazenil (500 ng) orthe combination of FMZ (500 ng)1 MDZ (2 µg for HU and HF, 1µg forLF). *P , 0.05 compared with control and MDZ1 FMZ groups.

(Gonzalez & File, 1997). These findings suggest that the lack ofeffects of benzodiazepines on trial 2 may be due in part to changesin the benzodiazepine receptor function that might occur in severalbrain structures. In the current study we investigated the effects ofcentral injections of midazolam in both trials 1 and 2. In addition weexplored whether a change in benzodiazepine receptor function couldtake place in the median raphe´ nucleus or the dorsal hippocampusby assessing the behavioural changes triggered by local applicationof the benzodiazepine receptor antagonist, flumazenil.

Materials and methods

Animals

Male hooded Lister rats weighing 200–300 g (Harlan, Bicester, Oxon,UK) were housed singly following surgery in cages 28.53 44 cm,with an overall height of 20 cm, and allowed to recover for 7 daysprior to behavioural testing. Food and water were freely availableand the room in which they were housed was lit with dim light andmaintained at 22 °C. Lights were on from 07.00 to 19.00 h.

© 1998 European Neuroscience Association,European Journal of Neuroscience, 10, 3673–3680

Apparatus

The social interaction test arena was a wooden box 603 60 cm, with35-cm-high walls. A camera was mounted vertically above the arenaand the rats were observed on a monitor in an adjacent room by anobserver who was blind to the drug treatment. The time spent insocial interaction (sniffing, following and grooming the partner,boxing and wrestling) provided the measure of anxiety and wasscored only when initiated by the operated animal. Infrared photocellswere mounted in the walls, 4.5 and 12 cm from the floor, and theinterruption of these beams provided automated measures of locomotoractivity and rearing, respectively (for details see File, 1980).

The plus-maze was made of wood and consisted of two oppositeopen arms 503 10 cm, and two opposite arms enclosed by 40-cm-high walls. The arms were connected by a central 103 10-cm square,and thus the maze formed a ‘plus’ shape. The maze was elevated50 cm from the floor and lit by dim light. A closed circuit TV camerawas mounted vertically over the maze and the behaviour was scoredfrom a monitor in an adjacent room. All scores were entered directlyinto an IBM computer. The percentage of time spent and thepercentage of entries onto the open arms of the maze provides themeasure of anxiety and the number of closed arm entries providesthe best measure of locomotor activity in this test (Pellowet al.,1985; File, 1992).

Drugs and chemicals

Midazolam maleate and flumazenil were both kindly donated byRoche Products Ltd (Welwyn Garden City, Herts, UK) and dissolvedin 0.9% sodium chloride solution. The stainless steel injectioncannulae (Cooper’s Needle Works Ltd, Birmingham, UK) were 30gauge 9 mm long for the dorsal hippocampus and 33 gauge 16 mmlong for the median raphe´ nucleus. Infusions were made using amicrodialysis pump (CMA/102 Biotech Instruments, Stockholm,Sweden) set at a rate of 0.5µL/30 s, and the needles were left inposition for an additional 30 s to allow drug diffusion. When bothmidazolam and flumazenil were administered they were dissolvedtogether and given as a single injection. The doses were selected onthe basis of a pilot study and a previous study using administrationinto the basolateral nucleus of the amygdala (Gonzalezet al., 1996).

Surgery

Rats were anaesthetized by inhalation of 3% halothane (May andBaker, Dagenham, Essex, UK) in oxygen and positioned in astereotaxic frame (Kopf Instruments, CA, USA). The skull wasexposed and the incisor bar adjusted such that bregma and lambdawere at the same height. Three indentations were made in the skullto accommodate screws which, together with the application of dentalcement, held the cannulae in place. For bilateral cannulation of thedorsal hippocampus, 7-mm-long stainless steel guide cannulae (23gauge, Cooper’s Needle Works Ltd) were positioned at 3.3 mmposterior to bregma, lateral6 2.4 and vertical –1.2 mm, thus sitingthem 2 mm above the target area according to the atlas of Paxinos& Watson (1986). For cannulation of the median raphe´ nucleus a 14-mm-long stainless steel guide cannula (23 gauge, Cooper’s NeedleWork) was positioned at 7.6 mm posterior to bregma,13.0 mmlateral and –6.5 mm vertical at an angle of 21°, siting it 2 mm abovethe median raphe´ nucleus. Because of concerns that an injectionvolume of 0.5µL to the median raphe´ nucleus may have led to spreadof drug to adjacent areas, we studied the anatomical specificity ofthe anxiolytic effects of midazolam by infusing it into an adjacentarea. The median raphe´ lateral coordinate was moved to13.5 mmand the vertical to 5.5 mm. The injections were thus positioned into

GABAergic modulation of anxiety 3675

TABLE 1. Mean (6 SEM) locomotor activity (beam breaks) and number of rears made by rats in the social interaction test following injections into the dorsalhippocampus and median raphe´ nucleus of 0.5µL saline (V), midazolam (MDZ: 1 and 2µg), flumazenil (FMZ: 100 or 500 ng) or a combination ofmidazolam1 flumazenil 500 ng (MDZ1 FMZ); the dose of midazolam in combination with flumazenil was 2µg for the high light, unfamiliar (HU) and highlight, familiar (HF) conditions and 1µg in the low light, familiar (LF) condition.

Brain area Conditions Drug Locomotor Rears n

Dorsal hippocampus HU V 355.96 23.8 24.96 0.8 6MDZ (1 µg) 344.66 37.4 20.16 2.3 7MDZ (2 µg) 389.16 23.7 21.06 1.2 6FMZ (500 ng) 332.16 14.2 22.26 1.4 6MDZ1FMZ 314.86 22.7 23.96 0.9 6

HF V 374.86 21.8 19.66 1.0 7MDZ (1 µg) 377.56 16.1 20.56 1.2 13MDZ (2 µg) 389.86 16.3 18.76 0.8 12FMZ (500 ng) 374.46 29.6 20.36 1.1 10MDZ1FMZ 386.26 19.3 18.26 1.4 11

LF V 456.36 19.0 17.16 1.6 11MDZ (1 µg) 416.96 29.8 13.66 1.3 10FMZ (500 ng) 473.86 33.2 17.76 2.2 10MDZ1FMZ 423.56 29.1 17.56 1.5 11

Median raphe´ HU V 417.76 44.1 23.86 3.0 6MDZ (1 µg) 379.36 36.3 21.86 3.0 7MDZ (2 µg) 416.56 48.1 19.36 2.2 6

V 399.26 33.8 23.76 1.7 6HF MDZ (1 µg) 501.06 40.3 24.26 2.3 6

MDZ (2 µg) 512.06 40.9* 26.36 0.9 7

V 403.16 19.3 22.46 1.0 13HF FMZ (100 ng) 506.66 21.7* 21.26 1.6 12

FMZ (500 ng) 473.46 26.0* 20.06 1.6 10

*P , 0.05 compared with control, Duncan’s test after analysis of variance.

the right areas of the pontine nuclei and the decussation superior ofthe cerebellar peduncle, next to the median raphe´ nucleus. Cannulaein the dorsal hippocampus and median raphe´ region were kept patentusing 7 and 14-mm-long stainless steel stylets, respectively (30 gauge,Cooper’s Needle Works Ltd). In order to accustom the animals tohandling and to keep the cannulae patent, each day following surgerythe rats were gently wrapped in a cloth and the stylets were replaced.

Behavioural testing

Independent experiments (with different groups of animals) werecarried out for each test condition (i.e. social interaction HU, HF andLF conditions, plus-maze 1 and plus-maze 2) and brain area (i.e.median raphe´ and dorsal hippocampus).

Social interaction test

For the HU and HF conditions, the illuminance was 300 lx and therats were naive to, or had received 10 min familiarization with, thearena 24 h prior to testing. For assessment of possible anxiogeniceffects of flumazenil rats were also tested in the LF condition inwhich control levels are highest and therefore reductions due toanxiogenic effects are easiest to detect. In this case, the illuminancewas 30 lx and the rats had received a 10-min familiarization trialin the test arena. In all cases they were tested with a novel,unoperated partner.

Three minutes after central injection, the rat was placed togetherwith its unoperated partner in the social interaction test arena and thetime spent by the operated rat in active social behaviour scored for4.5 min by an observer blind to the drug treatment. Rats were testedbetween 08.00 and 12.00 h in an order randomized for drug treatment,and the arena was thoroughly wiped after each trial and any faecesremoved.


Elevated plus-maze trial 1

Three minutes after injection each rat was placed in the centre of theplus-maze, facing an open arm and the times spent in, and numberof entries into, open and closed arms were scored for 5 min by anobserver blind to drug treatment. On each test day the animals weretested between 08.00 and 12.00 h in an order randomized for drugtreatment and the maze thoroughly wiped after each trial.

Elevated plus-maze trial 2

All rats received one undrugged 5 min trial in the elevated plus-maze3 days before their central injection. Procedures for central injectionsand testing were the same as in trial 1.

Histology

At the end of testing all animals were killed, the brains removed andthe injection site verified histologically (according to the atlas ofPaxinos & Watson, 1986) by a person blind to drug treatment. Figuresdepicting coronal slices through the dorsal hippocampus and medianrapheshow the target sites as shaded. For the dorsal hippocampus,injection sites between 2.8 and 4.2 mm posterior to bregma and6 1.7and 3.2 mm lateral within the hippocampal borders, between 2 and3 mm below the dura, were accepted. In seven cases the cannula onone side was above the target area and in one it was below, in twocases both cannulae were above the target area. The data from allthese animals were excluded from statistical analysis. Because theseanimals came from different drug treatments and test conditions itwas not possible to analyse their scores in any way. For the medianraphe, in five cases the needle tips failed to reach the target area,which was clearly defined in the sagittal zero plane between 7.3 and8.5 posterior to bregma and 7–8 mm below the dura. The data fromthese animals were excluded from statistical analysis.


FIG. 2. (Top) Diagrammatic representation of a coronal section (7.6 mmposterior to Bregma) of the rat brain showing the target area of the medianrapheas shaded. (Bottom) Mean (6 SEM) time spent in social interaction byrats tested in the high-light, unfamiliar (HU) and high-light, familiar (HF)arena after median raphe´ injections of saline (V) or midazolam (1 and 2µg).*P , 0.05 compared with control group.

Statistics

The scores were analysed by one-way analyses of variance andcomparisons between individual groups were then made with Duncan’stests; it is the significances of these that are shown in the figuresand tables.

Results

Social interaction

Injections into the dorsal hippocampus

Midazolam significantly increased the time spent in social interactionin the HU condition (F4,525 2.7,P , 0.05) andpost-hoctests showedthat the group treated with the dose of 2µg differed significantlyfrom vehicle-treated rats (Fig. 1). This increase in social interactionwas not accompanied by any change in locomotor activity or rears(F4,525 1.1 and 1.4, respectively; Table 1). In the HF condition,midazolam (1 and 2µg) also had a significant anxiolytic effect(F4,555 5.7, P , 0.001, see Fig. 1), without changing locomotoractivity or rears (Table 1). Flumazenil (500 ng) did not produce


FIG. 3. (Top) Diagrammatic representation of a coronal section (7.6 mmposterior to Bregma) of the rat brain showing a target area adjacent to themedian raphe´ as shaded. (Bottom) Mean (6 SEM) time spent in socialinteraction by rats tested in the high-light, unfamiliar (HU) arena afterinjections outside the median raphe´ of saline (V) or midazolam (1 and 2µg).

changes in anxiety, locomotor activity or rears (Fig. 1 and Table 1).Because flumazenil was behaviourally silent, its antagonism ofmidazolam effects was also studied. Flumazenil (500 ng) significantlyreversed the anxiolytic effects of midazolam (2µg) in HU and HFconditions (Fig. 1). Neither midazolam (1µg), flumazenil (500 ng)nor their combination had any effect in the LF condition (Fig. 1).

Injections in the median raphe´.

Midazolam (1 and 2µg) significantly increased the time spent insocial interaction (F2,175 5.0,P , 0.02) in the HU condition, withoutchanging locomotor activity (F2,175 1.9). Post-hocanalysis showedthat both doses had significant anxiolytic effects (Fig. 2).Anxiolyticeffects were also observed under the HF condition (F2,185 8.4,P , 0.005), but only the dose of 1µg had an anxiolytic effect, withoutchanging locomotor activity (see Fig. 2 and Table 1, respectively).Midazolam 2µg increased locomotor activity but not levels of socialinteraction (see Table 1 and Fig. 2, respectively). Administration offlumazenil (100 and 500 ng) to the median raphe´ in the HF conditionhad no effect on the time spent in social interaction (vehicle5


TABLE 2. Mean (6 SEM) perentage number of entries onto open arms, perentage time spent on open arms, number of closed arm entries and time (s) spent inthe central square in the plus-maze by naive (trial 1) and re-exposed (trial 2) rats after injections into the dorsal hippocampus of 0.5µL of saline (V), midazolam(MDZ: 1 and 2µg) or flumazenil (FMZ 100 and 500 ng)

Experiment Drug % time % number Number closed Time centre n

Trial 1Expt. 1 V 33.56 5.1 42.06 3.3 9.36 2.4 77.56 6.6 12

MDZ (1 µg) 35.66 4.7 47.96 3.9 8.86 0.9 79.76 6.6 12MDZ (2 µg) 29.06 4.4 44.06 2.3 8.86 0.8 67.86 5.3 13

Expt. 2 V 44.46 6.1 53.16 5.1 7.16 0.8 61.46 4.3 13MDZ (1 µg) 42.76 5.9 52.66 5.0 9.36 1.4 68.26 4.9 12MDZ (2 µg) 44.46 6.2 48.56 4.7 8.56 1.0 71.26 7.2 12FMZ (100 ng) 44.36 6.8 51.26 4.3 7.26 0.8 57.86 7.8 11

Expt. 3 V 46.46 7.5 56.06 4.4 6.76 0.8 70.56 7.6 12FMZ (500 ng) 40.86 5.5 47.26 3.1 8.26 0.7 67.86 5.4 12

Trial 2Expt. 1 V 13.66 3.8 16.76 4.0 13.56 1.7 104.46 9.7 13

MDZ (1 µg) 13.56 4.8 16.46 4.6 15.36 1.9 102.96 8.5 12MDZ (2 µg) 10.26 3.2 12.46 3.9 17.66 1.4 118.86 9.2 12

Expt. 2 V 13.96 4.6 18.26 5.1 14.36 1.4 100.36 9.5 12MDZ (1 µg) 21.06 5.0 25.76 5.2 11.86 1.1 104.16 10.8 11FMZ (100 ng) 12.56 5.1 18.36 5.5 11.56 1.3 91.16 12.4 11

Expt. 3 V 17.36 4.7 20.66 4.4 12.86 1.0 106.56 10.0 11FMZ (500 ng) 12.76 6.2 16.06 6.4 13.06 1.2 100.76 8.4 11

TABLE 3. Mean (6 SEM) perentage number of entries onto open arms, perentage time spent on open arms, number of closed arm entries and time (s) spent inthe central square in the plus-maze by naive (trial 1) and re-exposed (trial 2) rats after injections into the median raphe´ of 0.5µL of saline (V), midazolam(MDZ: 1 and 2µg) or flumazenil (FMZ 100 and 500 ng)

Experiment Drug % time % number Number closed Time centre n

Trial 1Expt. 1 V 33.06 8.1 39.86 5.6 6.56 1.5 62.86 7.69 10

MDZ (1 µg) 46.76 8.0 55.46 7.2 5.76 1.1 52.36 6.57 7MDZ (2 µg) 33.86 7.3 44.36 4.3 6.36 1.0 58.16 8.69 9

Expt. 2 V 35.26 5.7 43.86 4.2 7.36 1.0 69.56 5.4 16MDZ (1 µg) 47.86 5.0 52.16 4.3 7.56 1.0 64.66 6.9 13MDZ (2 µg) 44.66 6.2 50.06 3.6 7.46 0.7 68.56 6.9 14

Expt. 3 V 54.36 8.3 49.26 6.1 6.96 0.8 72.56 9.1 10FMZ (100 ng) 54.96 8.8 52.06 6.9 6.66 1.2 84.46 19.6 13FMZ (500 ng) 50.16 0.6 42.86 3.3 9.26 0.9 73.86 11.44 14

Trial 2Expt. 1 V 19.86 3.7 25.36 5.3 16.46 2.8 81.66 10.5 10

MDZ (1 µg) 20.46 5.5 20.96 4.7 13.76 2.1 81.86 11.5 11MDZ (2 µg) 25.06 6.9 30.36 5.6 13.06 1.9 81.16 10.6 10

Expt. 2 V 20.06 4.1 26.36 5.9 16.56 5.4 80.56 11.7 9MDZ (1 µg) 25.36 5.6 24.86 4.8 15.96 1.7 94.26 8.8 8MDZ (2 µg) 31.16 8.9 34.06 7.1 12.76 2.0 83.06 13.9 7

Expt. 3 V 30.56 7.9 27.96 6.2 13.76 1.8 63.76 11.8 11FMZ (100 ng) 29.26 6.0 28.76 3.9 14.16 2.3 62.96 17.8 14FMZ (500 ng) 32.76 6.2 39.26 5.5 9.26 1.8 75.56 16.8 15

99.06 9.0; flumazenil 100 ng5 90.1; flumazenil 500 ng596.56 9.2; F2,325 0.3); but clearly increased locomotor activity[F2,325 4.3,P , 0.02; see Table 1]. Because administration of flum-azenil altered general activity, the combination with midazolam wasnot attempted.

Because of concerns that an injection volume of 0.5µL to themedian raphe´ nucleus may have led to spread of drug to adjacentareas, we studied the anatomical specificity of the anxiolytic effectsof midazolam (1 and 2µg, n 5 6–7 per group) by infusing it into the


adjacent area of the pontine nuclei and the decussation superior ofthe cerebellar peduncle. Injections of midazolam in this area did notproduce any anxiolytic effects in the HU condition of the socialinteraction test (Fig. 3). No changes in locomotor activity wereobserved [vehicle5 4026 27.0; midazolam (MDZ) 1µg 5400.06 26.3; MDZ 2µg 5 394.76 19.2].

Elevated plus-maze (trials 1 and 2)

It can be seen from Tables 2 and 3 that none of the drug treatments


given to the dorsal hippocampus or median raphe´ nucleus significantlychanged any of the measures on trial 1 or trial 2 in the plus-maze.

Discussion

In the social interaction test, the benzodiazepine receptor agonist,midazolam, had anxiolytic effects after application to the dorsalhippocampus. These effects were pharmacologically specific sincethey were reversed by a silent dose of the benzodiazepine receptorantagonist flumazenil. When injected into the median raphe´, midazo-lam elicited anxiolytic effects and flumazenil was without effect onanxiety, but increased locomotor activity. The lack of effect of the2 µg dose of midazolam in the HF condition may be related to thelocomotor stimulant effects of this dose in this test condition.Furthermore, the anxiolytic effect of midazolam was shown to beanatomically specific because injections into an area adjacent to themedian raphe´ were without effect. In contrast to the clear-cut effectsin the social interaction test, stimulation of the benzodiazepinereceptors in the dorsal hippocampus or the median raphe´ had nosignificant effects on measures in the elevated plus-maze. It is possiblethat higher doses would have resulted in anxiolytic effects, but centralinjections of high doses of benzodiazepines are complicated becausebenzodiazepines precipitate at high concentrations and the use ofhigher injection volumes compromises anatomical specificity. Theseresults show that the benzodiazepine receptors in both the dorsalhippocampus and the median raphe´ nucleus differentially modulateanxiety in the social interaction and the elevated plus-maze tests. Afurther test of whether benzodiazepine receptors in these areas playany role in behaviour in the plus-maze would be to determine whetheradministration of flumazenil to these sites would antagonize theanxiolytic effects of systemically administered benzodiazepines.

It is well established that benzodiazepines induce their effects byfacilitating GABAergic neurotransmission. The potency of benzo-diazepine receptor agonists is thus limited by the synaptic availabilityof GABA. In the present experiments, bilateral infusion of midazolaminto the dorsal hippocampus induced anxiolytic effects in the socialinteraction test, indicating that in this test situation there is sufficientendogenous GABAergic tone in the dorsal hippocampus. Interestingly,midazolam showed a different degree of efficacy across the threeconditions of the social interaction test. In the most anxiogenic testcondition, HU, an anxiolytic effect was reached only with the higherdose (2µg) of midazolam, suggesting a low GABAergic tone in thedorsal hippocampus; whereas in the HF condition, a moderatelyanxiogenic condition, anxiolytic effects were seen with both doses(1 and 2µg) of midazolam, suggesting an increase in GABAergictone relative to HU. Finally, no anxiolytic effect was seen in the LFtest condition, the least anxiogenic condition, possibly because theGABAergic transmission in these circumstances is high and couldnot be further enhanced by midazolam. Therefore, it seems that theGABAergic tone in the dorsal hippocampus keeps a linear relationshipwith the baseline of social interaction (level of anxiety), which wouldsuggest that the dorsal hippocampal benzodiazepine-GABAergicsystem plays a key role in the modulation of the intensity of anxiety,as measured in the social interaction test.

In contrast to the clear effects observed in the social interactiontest, no effect was induced by the dorsal hippocampal injections ofmidazolam on trial 1 in the elevated plus-maze. The lack of effectsof midazolam can be attributed to a low GABAergic tone in thedorsal hippocampus on trial 1. This is further supported by the lackof effects of dorsal hippocampal injections of flumazenil, which rulesout a possible switch of GABAA receptors to the inverse agoniststate in the dorsal hippocampus. It is interesting to note that administra-


tion of a 5-HT1A receptor agonist or cholinergic antagonists into thedorsal hippocampus, which induce clear-cut anxiogenic effects in thesocial interaction test, have no effects on trial 1 of the plus-maze(File et al., 1996, 1998). It would therefore appear that trial 1 of theplus-maze is less sensitive to pharmacological manipulations in thedorsal hippocampus. Consistent with this view Treit & Menard (1997)found that lesions of the dorsal hippocampus had no effect on anyof the measures on trial 1 in the elevated plus-maze. Altogether theseresults suggest that the dorsal hippocampus may not play an importantrole in the anxiety evoked in the plus-maze (trial 1).

Numerous studies have suggested that on trials 1 and 2 the natureof anxiety and neurotransmitter pathways involved are different(Lister, 1987; Rodgerset al., 1992; File & Zangrossi, 1993; File,1993). Moreover, in previous studies we have found that trials 1 and2 also differ in their dorsal hippocampal modulation of anxiety. Forexample, the administration of a 5-HT1A receptor agonist or muscarinicor nicotinic cholinergic antagonists induced anxiogenic effects ontrial 2 of the elevated plus-maze, effects not seen on trial 1 (Fileet al., 1996, 1998). In the present experiments, both midazolam andflumazenil were without effects on trial 2 suggesting that dorsalhippocampal GABAergic tone is low on trial 2. In line with thissuggestion, previous studies from our laboratory have shown thatexposure to the social interaction test, but not to trial 1 or trial 2 ofthe elevated plus-maze, resulted in increased K1 evoked [14 C]-GABArelease in the hippocampus (Fileet al., 1992, 1993a). Therefore, itseems that a low dorsal hippocampal GABAergic transmission is acommon feature for trials 1 and 2 of the plus-maze.

The median raphe´ nucleus is the main structure in the midbrainwhere the serotonergic afferents to the dorsal hippocampus originate(Azmitia & Segal, 1978). Considerable pharmacological evidencesupports the functional involvement of the median raphe´ in themodulation of anxiety. For instance, stimulation of 5-HT1A receptors inthe median raphe´ produced anxiolytic effects in the three behaviouralsituations studied here (Fileet al., 1996), as well as in a number ofother animal models of anxiety such as the Vogel test (Higginset al.,1992), ultrasonic vocalizations (Schreiber & De Vry, 1993), the black-white crossing test and novelty-induced suppression of drinking (Carli& Samanin, 1988; Carliet al., 1989). In the current study, we havefound that local infusion of midazolam into the median raphe´ leadsto anxiolytic effects in the social interaction test, which is most likelythrough facilitation of the inhibitory action of GABA on neuronalfunction in this brain area. In this respect, it has been shown thatGABA or benzodiazepines also inhibit neuronal firing rate in theraphenuclei (Gallager & Aghajanian, 1976; Hery & Ternaux, 1981;Tao et al., 1996) and 5-HT release in several serotonergic terminalareas, including the hippocampus (Balfour, 1980; Peiet al., 1989).However, we failed to get a clear-cut anxiolytic effect on trial 1 inthe elevated plus-maze after administration of midazolam to themedian raphe´, only a non-significant increase being found. The patternof this finding was replicated in a second experiment using biggergroups of animals. Overall, these results suggest that on trial 1 thereis some GABAergic tone in the median raphe´, but it is not sufficientto allow midazolam to produce a clear anxiolytic effect, such as thatseen in the social interaction test. One alternative explanation is thatbenzodiazepine receptors in the median raphe´ could have a very highagonist or inverse agonist endogenous activity that could prevent theagonist actions of midazolam to emerge. However, this is unlikelybecause the administration of flumazenil did not reveal any endogen-ous activity in this area on trial 1 of the elevated plus-maze.

It is worth noting that local administration of midazolam into thedorsal raphe´, another serotonergic nucleus in the brainstem, had clear-cut anxiolytic effects on trial 1 in the elevated plus-maze. This effect


was observed with a similar baseline (30%) and same doses used asin the present experiments (Gonzalez & File, 1997). Therefore,the dorsal raphe´ seems more sensitive than the median raphe´ tobenzodiazepine receptor stimulation. Such a difference in sensitivitycould be associated with differences in GABAergic neurotransmissionin these areas. Indeed, neurochemical studies have revealed importantdifferences between the median and the dorsal raphe´ nuclei regardingtheir local GABAergic neurotransmission. Taoet al. (1996) havereported that endogenous GABA has a tonic GABAA receptor-mediated inhibitory effect on 5-HT release in the dorsal, but not inthe median, raphe´ nucleus. Furthermore, autoradiographic studieshave shown that GABAA receptors in the median raphe´ are of aparticular kind (type I), whose affinity for benzodiazepines is low,but greatly enhanced by application of GABA (Mennini & Gobbi,1990). It is therefore possible that a combination of these two factors,i.e. the type of GABAA receptor and dynamic changes in GABAergictone in the median raphe´-dorsal hippocampal pathway, might deter-mine the relative potency of benzodiazepines in the elevated plus-maze.

Local infusion of midazolam into the median raphe´ was alsowithout effects on trial 2 of the plus maze. We have previouslyreported that in trial 2 there is an increase in the activity of anendogenous inverse agonist in the dorsal raphe´, since local applicationof flumazenil had anxiolytic effects (Gonzalez & File, 1997). However,in the present experiments, the administration of flumazenil to themedian raphe´ was without effect on trial 2, suggesting there was nochange in the receptor state in this nucleus. These data provide furtherevidence that the benzodiazepine-GABAergic systems differ in thedorsal and median raphe´ nuclei. In the present experiments, therewas a marked increase in the undrugged levels of anxiety from trial1 to trial 2, as has been found in other experiments (Rodgers &Shepherd, 1993; Treitet al., 1993; Gonzalez & File, 1997), withoutany accompanying decrease in locomotor activity, as measured bythe number of closed arm entries. Indeed, this measure showed someincrease from trial 1 to trial 2. This confirms the results of a previousstudy in which three different measures of locomotor activity wereused and which failed to find any evidence of locomotor habituationfrom trial 1 to trial 2 (Gonzalez & File, 1997). It is therefore clearthat locomotor habituation is an insufficient explanation for thechange in anxiety that results from plus-maze experience.

In conclusion, although stimulation of benzodiazepine receptors inthe median raphe´-dorsal hippocampal pathway reduces anxiety in thesocial interaction test, these effects were weak or not seen in theelevated plus-maze, which indicates that these two brain areas areunder high and low GABAergic control in the social interaction andthe plus-maze tests, respectively. This parallel regulation of themedian raphe´ and the dorsal hippocampus by the benzodiazepine-GABA system revealed in our study gives further support to the viewthat both structures are part of one functional pathway regulatingspecific types of anxiety (Graeffet al., 1996).

AcknowledgementsThese experiments were supported by a grant from the Special Trustees ofGuy’s Hospital. We are grateful to Peter Mabbutt for expert technical help.

AbbreviationsHF high light, familiarHU high light, unfamiliarLF low light, familiar

ReferencesAndrews, N., File, S.E., Fernandes, C., Gonzalez, L.E. & Barnes, N.M. (1997)

Evidence that the median raphe´ nucleus – dorsal hippocampal pathway


mediates diazepam withdrawal-induced anxiety.Psychopharmacology, 130,228–234.

Andrews, N., Hogg, S., Gonzalez, L.E. & File, S.E. (1994) 5-HT1A receptorsin the median raphe´ nucleus and dorsal hippocampus may mediate anxiolyticand anxiogenic behaviours respectively.Eur. J. Pharmacol., 264, 259–264.

Azmitia, E.C. & Segal, M. (1978) An autoradiographic analysis of thedifferential ascending projections of the dorsal and median raphe nuclei inthe rat.J. Comp. Neurol., 179, 641– 668.

Balfour, D.J. (1980) Effects of GABA and diazepam on 3H-serotonin releasefrom hippocampal synaptosomes.Eur. J. Pharmacol., 68, 11–16.

Carli, M., Prontera, C. & Samanin, R. (1989) Evidence that central 5-hydroxytryptaminergic neurones are involved in the anxiolytic activity ofbuspirone.Br. J. Pharmacol., 96, 829–836.

Carli, M. & Samanin, R. (1988) Potential anxiolytic properties of 8-hydroxy-2-(Di-N-propylamino) tetralin, a selective serotonin1A receptor agonist.Psychopharmacology, 94, 84–91.

File, S.E. (1980) The use of social interaction as a method for detectinganxiolytic activity of chlordiazepoxide-like drugs.J. Neurosci. Meth., 2,219–238.

File, S.E. (1990) One-trial tolerance to axiolytic effects of chordiazepoxidein the plus-maze.Psychopharmacology, 100, 281–282.

File, S.E. (1992) Behavioural detection of anxiolytic action. In: Elliott, J.M.,Heal D.J. & Marsden, C.A. (eds),Experimental Approaches to Anxiety andDepression. John Wiley, London, pp. 25–44.

File, S.E. (1993) The interplay of learning and anxiety in the elevated plus-maze.Behav. Brain. Res., 58, 199–202.

File, S.E. & Hyde, J.R.G. (1978) Can social interaction be used to measureanxiety?Br. J. Pharmacol., 62, 19–24.

File, S.E., Andrews, N., Zharkovsky, A. & Zangrossi, H. (1992) Modificationof chlordiazepoxide’s behavioural and neurochemical effects by handlingand plus-maze experience.Euro. J. Pharmacol., 218, 9–14.

File, S.E., Gonzalez, L.E. & Andrews, N. (1996) Comparative study of pre-and post-synaptic 5-HT1A receptor modulation of anxiety in two ethologicalanimal tests.J. Neurosci., 16, 4810–4815.

File, S.E., Gonzalez, L.E. & Andrews, N. (1998) Endogenous acetylcholinein the dorsal hippocampus reduces anxiety through actions on nicotinic andmuscarinic1 receptors.Behav Neuroscience, 112, 352–359.

File, S.E. & Zangrossi, H. (1993) One-trial tolerance to the anxiolytic actionsof benzodiazepines in the elevated plus-maze, or the development of aphobic state?Psychopharmacology, 110, 240–244.

File, S.E., Zangrossi, H. & Andrews, N. (1993a) Social interaction andelevated plus-maze tests: changes in release and uptake of 5-HT and GABA.Neuropharmacology, 32, 217–221.

File, S.E., Zangrossi, H., Viana, M. & Graeff, F.G. (1993b) Trial 2 in theelevated plus-maze: A different form of fear?Psychopharmacology, 111,491–494.

Gallager, D.W. (1978) Benzodiazepines: potentiation of a GABA inhibitoryresponse in the dorsal raphe nucleus.Eur. J. Pharmacol., 49, 133–143.

Gallager, D.W. & Aghajanian, G.K. (1976) Effects of antipsychotic drugs onthe firing of dorsal raphe cells. II Reversal by picrotoxin.Eur. J. Pharmacol.,39, 357–364.

Gonzalez, L.E., Andrews, N. & File, S.E. (1996) 5-HT1A and benzodiazepinereceptors in the basolateral amygdala modulate anxiety in the socialinteraction test, but not in the elevated plus-maze.Brain Res., 732, 145–153.

Gonzalez, L.E. & File, S.E. (1997) A 5-min experience in the elevated plus-maze alters the state of the benzodiazepine receptor in the dorsal raphe´nucleus.J. Neurosci., 17, 1505–1511.

Graeff, F.G. (1993) Role of 5-HT in defensive behavior and anxiety.Rev.Neurosci., 4, 181–211.

Graeff, F.G., Guimaraˆes, F.S., De Andrade, T.G.S. & Deakin, J.F.W. (1996)Role of 5-HT in stress, anxiety and depression.Pharmacol. Biochem.Behav., 54, 129–141.

Green, S. & Vale, A.L. (1995) Role of amygdaloid nuclei in the anxiolyticeffects of benzodiazepines in rats.Behav. Pharmacol., 3, 35–41.

Hery, F. & Ternaux, J.P. (1981) Regulation of release processes in centralserotoninergic neurons.J. Physiol. (Paris), 77, 287–301.

Higgins, G.A., Bradbury, A.J., Jones, B.J. & Oakley, N.R. (1988) Behavioraland biochemical consequences following activation of 5-HT1-like andGABA receptors in the dorsal raphe nucleus of the rat.Neuropharmacology,27, 993–1001.

Higgins, G.A., Jones, B.J. & Oakley, N.R. (1992) Effect of 5-HT1A receptoragonists in two models of anxiety after dorsal raphe injection.Psychopharmacology, 106, 261–267.

Hodges, H., Green, S. & Glenn, B. (1987) Evidence that the amygdala isinvolved in benzodiazepine and serotonergic effects on punished respondingbut not on discrimination.Psychopharmacology, 92, 491–504.


Lister, R.G. (1987) The use of a plus-maze to measure anxiety in the mouse.Psychopharmacology, 92, 180–185.

Mennini, T. & Gobbi, M. (1990) Regional distribution of low-affinity GABAreceptors coupled to benzodiazepine receptor subtypes in rat brain: anautoradiographic evaluation.Eur. J. Pharmacol., 189, 143–148.

Paxinos, G. & Watson, C.H. (1986)The Rat Brain in Stereotaxic Coordinates.Academic Press, London.

Pei, Q., Zetterstrom, T. & Fillenz, M. (1989) Both systemic and localadministration of benzodiazepine agonists inhibit the in vivo release of 5-HT from ventral hippocampus.Neuropharmacology, 28, 1061–1066.

Pellow, S., Chopin, P.H., File, S.E. & Briley, M. (1985) Validation ofopen:closed arm entries in an elevated plus-maze as a measure of anxietyin the rat.J. Neurosci. Meth., 14, 149–167.

Pesold, C. & Treit, D. (1995) The central and basolateral amygdaladifferentially mediate the anxiolytic effects of benzodiazepines.Brain Res.,671, 213–221.

Plaznik, A., Jessa, M., Bidzinski, A. & Nazar, M. (1994) The effect ofserotonin depletion and intra-hippocampal midazolam on rat behavior inthe Vogel conflict test.Eur. J. Pharmacol., 257, 293–296.

Rodgers, R.J., Lee, C. & Shepherd, J.K. (1992) Effects of diazepam onbehavioural and antinociceptive responses to the elevated plus-maze inmale mice depend upon treatment regiment and prior maze experience.Psychopharmacology, 106, 102–110.

Rodgers, R.J. & Shepherd, J.K. (1993) Influence of prior maze experience on


behaviour and response to diazepam in the elevated plus-maze and light/dark tests of anxiety in mice.Psychopharmacology, 113, 237–242.

Sanders, S.K. & Shekhar, A. (1995) Regulation of anxiety by GABA receptorsin the rat amygdala.Pharmacol. Biochem.. Behav., 52, 701–706.

Schreiber, R. & De Vry, J. (1993) Neuronal circuits involved in the anxiolyticeffects of the 5-HT1A receptor agonists 8-OH-DPAT, ipsapirone andbuspirone in the rat.Eur. J. Pharmacol., 249, 341–351.

Sieghart, W. (1992) GABAA receptors: ligand-gated Cl- ion channelsmodulated by multiple drug-binding sites.TiPS, 13, 446–450.

Stein, L., Wise, C.D. & Berger, B.D. (1973) Antianxiety action ofbenzodiazepines: decrease in activity of serotonin neurons in the punishmentsystems. In Garattini, S., Mussini, E. & Randall, L.O. (ed.),TheBenzodiazepines. New York, Raven Press, pp. 299–326.

Tao, R., Zhiyuan, M. & Auerbach, S.B. (1996) Differential regulation of 5-hydroxytryptamine release by GABAA and GABAB receptors in midbrainraphe nuclei and forebrain of rats.Br. J. Pharmacol., 119, 1375–1384.

Thiebot, M.-H., Jobert, A. & Soubrie´, P. (1980) Conditioned suppression ofbehavior: its reversal by intra raphe microinjection of chlordiazepoxide andGABA. Neurosci. Lett., 16, 213–217.

Treit, D. & Menard, J. (1997) Dissociations among the anxiolytic effectsof septal, hippocampal, and amygdaloid lesions.Behav. Neurosci., 111,653–658.

Treit, D.T., Pesold, C. & Rotzinger, S. (1993) Dissociating the anti-fear effectsof septal and amygdaloid lesions using two pharmacologically validatedmodels of rat anxiety.Behav. Neurosci., 107, 770–785.

stimulation of benzodiazepine receptors in the dorsal hippocampus and median raphé reveals...

Documents