Neuronal excitability in the periaqueductal grey matter during the estrous cycle in female Wistar rats

Neuroscience 144 (2007) 325–335

NEURONAL EXCITABILITY IN THE PERIAQUEDUCTAL GREY MATTER DURING THE ESTROUS CYCLE IN FEMALE WISTAR RATS K. E. BRACK AND T. A. LOVICK*

Key words: estrous cycle, GABA, panic, periaqueductal grey, CCKB receptors, rat.

Department of Physiology, University of Birmingham, Birmingham B15 2TT, UK

In women, the premenstrual (late luteal) phase of the menstrual cycle is commonly associated with psychological changes that include irritability, anxiety and mood swings (Halbreich, 2003). Women afflicted by other disease states, e.g. catamenial epilepsy, irritable bowel syndrome and panic disorder, also experience exacerbation of symptoms during the premenstrual period (Ensom, 2000). Cyclical changes in female behavior are not confined to humans. During diestrus, female Wistar rats also display increases in aggressive behaviors and heightened responsiveness to panic-inducing drugs (Olsson et al., 2003). The premenstrual phase of the menstrual cycle in women and late diestrus (LD) in rats are both associated with a rapid fall in plasma progesterone (McLaughlan et al., 1987; Watanabe et al., 1990). Several studies have shown that a fall in progesterone precipitated by withdrawal from prolonged dosing with the steroid, leads to plasticity of GABAA receptor subunit expression in several brain regions. Upregulation of ␣4 and ␦ GABAA receptor subunit mRNA and protein levels has been reported in hippocampus, amygdala and periaqueductal gray matter (PAG) (Smith et al., 1998a,b; Gulinello et al., 2003; Griffiths and Lovick, 2005a). In the PAG a parallel increase in ␣4, ␦ and ␤1 GABAA receptor subunit expression was also seen in rats during the LD phase of the estrous cycle (Lovick et al., 2005). Thus the natural fall in endogenous progesterone levels that occurs during LD appears to be a sufficient stimulus to trigger plasticity of GABAA receptor subunit expression in the PAG (Lovick, 2006). The dorsal half of the periaqueductal gray matter (dPAG) contains neural circuitry that can initiate panic behavior. Stimulation in the dPAG in humans and in experimental animals evokes autonomic and behavioral signs of panic-like anxiety (Lovick, 2000). In humans panic can also be triggered by systemic injection of agents that are agonists at cholecystokinin (CCK)B receptors (e.g. van Megen et al., 1994). CCKB receptor agonists are also panicogenic in rats and a major site of action appears to be the dPAG (Bertoglio and Zangrossi, 2005; Zanovelli et al., 2004). A recent study in anesthetized female rats has shown that the panic-like pattern of cardio-respiratory response evoked by i.v. injection of the CCKB agonist pentagastrin (PG) was enhanced during LD (Brack et al., 2006). This suggests that the responsiveness of the panic circuitry to PG may change according to the stage of the estrous cycle. The central actions of CCK are thought to be linked to activity in GABAergic systems (Acosta, 2001;

Abstract—Extracellular recordings were made from output neurons in the dorsal half of the periaqueductal gray matter (dPAG) in urethane-anesthetized female Wistar rats. All the neurons were quiescent. A basal level of firing was therefore induced by continuous iontophoretic application of D,L-homocysteic acid (DLH). In the presence of the GABAA receptor antagonist bicuculline methiodide (BIC 0 –30 nA) the DLHinduced firing increased further, revealing the presence of ongoing GABAergic inhibitory tone on the recorded neurons. The BIC-induced increase in firing rate was significantly greater in neurons recorded during estrus (Est) and late diestrus (LD) compared with proestrus (Pro) and early diestrus (ED) suggesting that GABAergic tone was lower in Est and LD. I.v. injection of the panicogenic cholecystokinin (CCK)B receptor agonist pentagastrin (PG, 40 ␮g kgⴚ1) produced an increase in firing rate in 12/17 (70%) of neurons tested in the dPAG. Iontophoretic application of PG (10 –30 nA) also produced a current-related increase in firing rate in 73.6% of the neurons tested. The excitatory response was reduced during application of the selective CCKB receptor antagonist ␤-[2-([2(8-azaspiro[4.5]dec-8-ylcarbonyl)-4,6-dimethylphenyl]amino)-2oxoethyl]-(R)-napthalenepropanoic acid (CR2945) (60 nA, nⴝ6). The PG-evoked increase in firing rate was significantly greater in neurons recorded during Est and LD compared with during Pro and ED. Juxtacellular labeling with neurobiotin in eight neurons revealed multipolar cells 12– 44 ␮m diameter with up to six primary dendrites. In three of eight neurons, a filled axon was present and coursed without branching toward the perimeter of the periaqueductal gray matter (PAG). The estrous cycle-related change in responsiveness to BIC and PG suggests that the panic circuitry in the PAG may become more responsive to panicogenic agents during estrus and late diestrus as a consequence of a decrease in the intrinsic level of inhibitory GABAergic tone. The findings may have implications for understanding the neural processes that underlie the development of premenstrual dysphorias in women. © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. *Corresponding author. Tel: ⫹44 (0) 121 414 6929; fax: ⫹44 (0) 121 414 6919. E-mail address: [email protected] (T. A. Lovick). Abbreviations: BIC, bicuculline methiodide; CCK, cholecystokinin; CR2945, ␤ -[2-([2-(8-azaspiro[4.5]dec-8-ylcarbonyl)-4,6dimethylphenyl]amino)-2-oxoethyl]-(R)-napthalenepropanoic acid; DLH, D,L-homocysteic acid; dPAG, dorsal half of the midbrain periaqueductal gray matter; ED, early diestrus; Est, estrus; HR, heart rate; LD, late diestrus; MAP, mean arterial pressure; PAG, periaqueductal gray matter; PB, phosphate buffer; PG, pentagastrin; Pro, proestrus; PSB, Pontamine Sky Blue; RR, rate of respiration.

0306-4522/07$30.00⫹0.00 © 2006 IBRO. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.neuroscience.2006.08.058

325

326

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

Ferraro et al., 2000; Ranovska, 1995; Perez de la Mora et al., 1993; Siniscalchi et al., 2003). Within the PAG, estrous cycle-related upregulation of ␣4, ␤1 and ␦ GABAA receptor subunit expression was confined to GABAergic interneurons (Griffiths and Lovick, 2005b). Thus it is possible that changes in the functional properties of the GABAergic cell population in the PAG during the estrous cycle are associated with changes in responsiveness of the panic circuitry to CCKB agonists. In order to test this hypothesis, we investigated the functional excitability of output neurons in the PAG at different stages of the estrous cycle in anesthetized female rats. Some of the results have been published in abstract form (Brack et al., 2004; Brack and Lovick, 2004; Jeffery et al., 2005).

EXPERIMENTAL PROCEDURES All experiments were carried out on female Wistar rats (244.6⫾3.1 g body weight, n⫽50), approximate age 10 –12 weeks. Animals were housed in pairs at 21⫾1 °C under a 12-h light/dark cycle (lights on at 7 a.m.) and were given free access to food and water throughout their maintenance. All procedures were undertaken in accordance with the University of Birmingham local guidelines on the ethical use of animals, the UK Animals (Scientific Procedures) Act 1986 and conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1985). All attempts were made to minimize the number and suffering of animals used.

Determination of the stage of the estrous cycle Prior to starting any surgical procedures, a vaginal smear was taken from the unanesthetized animal and stained using a QuickDiff FIX staining set (Reagena, Takojantie, Toivala, Finland) to determine the stage of the estrous cycle (Brack et al., 2006). Additional smears were taken from the anesthetized animal throughout the day to ensure that each animal remained in the same stage of the estrous cycle for the duration of the experiment.

Surgical preparation Rats were anesthetized with urethane (0.5 ml 100 g, 20% solution i.p., Sigma, Poole, Dorset, UK) and instrumented to record femoral arterial pressure. Heart rate (HR) was derived from the ECG signal recorded from a chest electrode positioned at approximately V4. Respiratory rate was recorded using a spirometer (AD Instruments, Chalgrove, Oxfordshire, UK) attached to a tracheal cannula. Rectal temperature was maintained at 37 °C throughout the experiment via a heating blanket. A femoral vein was cannulated for administration of drugs and supplementary fluids: saline or the plasma substitute gelofusine (B Braun Medical, Sheffield, South Yorkshire, UK). Rats were held in a stereotaxic frame in the attitude described by Paxinos and Watson (1986). A craniotomy was performed and the dura was cut and reflected to expose the cortical surface overlying the PAG. Once the surgical preparation was finished a stabilization period of at least 1 h was allowed before experimentation began. The depth of anesthesia was monitored throughout the experiment by testing corneal and pedal reflexes. Supplementary doses of sodium pentobarbitone (Sagatal, 1–3 mg i.v., Rhone Merieux, Harlow, Essex, UK) were given as needed to maintain a stable and consistent level of anesthesia.

Neuronal recording Extracellular recordings of single unit activity in the PAG were made using five-barreled glass microelectrodes, overall tip diameter 5 ␮m. The barrels were filled respectively with 4 M NaCl for recording, 1% Pontamine Sky Blue (PSB) in 0.5 M sodium acetate (pH 7.7) for current balancing and marking recording sites, D,Lhomocysteic acid (DLH, 0.1 M, pH 8.4) for activating neurons and two of the following: GABA (0.25 M, pH 4), the GABAA receptor antagonist bicuculline methiodide (BIC, 0.2 M, pH 4), the CCKB receptor agonist PG (13 mM, pH 9) or the CCKB receptor antagonist ␤-[2-([2-(8-azaspiro[4.5]dec-8-ylcarbonyl)-4,6-dimethylphenyl]amino)-2-oxoethyl]-(R)-napthalenepropanoic acid (CR2945) (Revel et al., 1998) (10 mM, pH 9). For barrels filled with GABA and BIC, a retaining current of ⫺15 nA was applied in between ejection periods to prevent diffusion of drugs from the pipette tip. For barrels filled with DLH, PG and CR2945 the retaining current was ⫹15 nA. The electrode tip was positioned within the dorsal half of the PAG at sites between 5.6 mm and 8.7 mm caudal to bregma and between 3.5 mm to 5.5 mm below the cortical surface (Paxinos and Watson, 1986). Spike activity was amplified using a Neurolog amplifier (NL104, Digitimer, Welwyn Garden City, Hertfordshire, UK) and fed through a window discriminator and a histogram of firing rate was computed on-line. A four-channel iontophoretic device (Grayden Electronics, Birmingham, West Midlands, UK) was used to apply drugs directly in the vicinity of each recorded neuron. Recording sites were marked by iontophoretic deposition of PSB (5 ␮A for 10 min negative current). At the end of each experiment the brain was removed and fixed in 10% formol saline. Frozen sections 60 ␮m thick were stained with Neutral Red. The anatomical location of blue spots in the tissue was made with reference to the atlas of Paxinos and Watson (1986). In conscious rats the excitability of the circuitry in the dorsal half of the PAG is normally subject to an ongoing inhibitory GABAergic influence (Schenberg et al., 1983; Brandao et al., 2005). Previous studies in anesthetized animals have reported that the majority of the neurons that could be recorded using multibarrel micropipettes, were quiescent (Lovick, 2001). In the present study searching for neurons was therefore carried out during continuous ejection of DLH (⬍5 nA) in order to induce neuronal firing. Once a single unit recording had been established the level of ejecting current for DLH was adjusted to produce an average firing rate of around 5 Hz. No neurons that were spontaneously active, i.e. firing in the absence of DLH, were investigated.

Juxtacellular labeling In a separate group of animals the neurons from which recordings had been made were identified by juxtacellular labeling with neurobiotin (Pinault 1996). Initial attempts to entrain neuronal activity using traditional five-barreled micropipettes proved unsuccessful. We therefore constructed ‘piggyback’ electrodes as described by Jones et al. (2002) by gluing a single barrel pipette filled with 1.5% neurobiotin in 0.5 M sodium acetate (pH unadjusted, resistance 7–35 m⍀, mean 22.6⫾4.0 m⍀) to a five-barrel micropipette so that the single barrel protruded approximately 10 ␮m beyond the tip of the multibarrel micropipette. The barrels were filled respectively with DLH, PSB and a selection of the solutions described above. Searching for neuronal activity was carried out during continuous ejection of DLH (see above). Recordings were made using an intracellular bridge mode amplifier (model BA-1S, npi Electronic GmbH, Hauptstrasse, Tamm, Germany). Once a spike had been isolated, a qualitative assessment of the responsiveness to pharmacological agents was made. Neurons were then entrained by applying 200 ms on 200 ms off positive current pulses through the recording pipette (1–10 nA for periods between 40 and 360 s), gated by an isolated square pulse voltage stimulator (model SD9, Astro-Med, West Warwick, RI, USA). After

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335 entrainment the electrode was left in place for 1 min and then withdrawn 300 ␮m dorsally where a deposit of PSB was made as described above. This blue spot helped to locate the position of the entrained cell in histological material. Following entrainment, animals remained anesthetized for a further 4 –11.5 h to allow for diffusion of the tracer throughout the neuron.

Fixation and histochemical processing to reveal neurobiotin-labeled cells At the end of the experiment, the preparation was perfused retrogradely through the descending aorta with 100 ml heparinized saline (10 units ml⫺1) followed by 200 ml 4% paraformaldehyde in 0.1 M phosphate buffer (PB) containing 0.5% glutaraldehyde (pH 7.4). After removal from the skull, the brain was post fixed for 2 h and then transferred to 30% sucrose in 0.1 M PB at 4 °C overnight to afford cryoprotection. Sections 40 ␮m thick were cut using a freezing microtome. After washing in PB (3⫻10 min) endogenous peroxidase activity was blocked using hydrogen peroxide (0.15% in 0.1 M PB with 0.1% Triton-X 100 for 60 min). Sections were then rinsed three times with 0.1 M PB and incubated in avidin– biotin complex (1:100 in 0.1 M PB and 0.1% Triton-X 100, ABC Elite, Vector Laboratories, Peterborough, Cambs, UK) for 72 h at 4 °C with gentle agitation. After further washes (2⫻10 min in PB followed by 3⫻10 min 0.1 M PB saline) the reaction product was visualized using a nickel– diaminobenzidine substrate kit (Ni-DAB, SK-4100, Vector Laboratories). Following further washes sections were mounted onto gelatinized slides, dried in air and then dehydrated through ascending series of alcohols, cleared in Histoclear (National Diagnostics, Atlanta, GA, USA) and mounted using Histomount (National Diagnostics). Reconstructions of labeled cells were made using an Olympus BH2 microscope (Olympus, London, UK) with a drawing tube attachment. Selected material was photographed using an Olympus C5060Z camera and the images were imported into Photoshop software (v7.0, Adobe, San Jose, CA, USA) for cropping.

Drugs used All drugs unless otherwise stated were purchased from Sigma (Dorset, UK). DLH and GABA were dissolved in distilled water while BIC was dissolved in saline. PG (for i.v. administration) was initially dissolved by sonication in saline pH 9 and subsequently adjusted to pH 7.4. For iontophoretic application of PG the final pH was kept at pH 9.

Data analysis All cardiovascular and neuronal parameters were captured and stored using a PowerLab 8/s data acquisition system operated in Chart mode (AD Instruments). Neuronal activity was analyzed after height and width discrimination using the Spike Histogram module (v1.2, AD Instruments). Data were analyzed using oneway ANOVA with Fisher’s post hoc test, as appropriate using Statview software (v.4.1, JMP Europe, Marlow, Bucks, UK). Differences were considered significant at the P⬍0.05 level.

RESULTS GABAergic influence on neurons in the dPAG In total, recordings were made from 119 neurons in the dorsal half of the PAG and immediately overlying region (Fig. 1). The response properties of the cells indicated that they were a homogeneous population and no differences in responsiveness with respect to dorso-ventral location were detected. Almost all cells recorded in the PAG were

327

quiescent. DLH was therefore ejected continuously to induce activity. The mean ejection current to produce a mean firing rate of 5.1⫾0.2 Hz was 4.9⫾0.6 nA. Fifty-eight neurons were tested for responsiveness to GABA, which in every case produced a decrease in ongoing activity. The cells in the dPAG were extremely sensitive to GABA. Switching off the retaining current was usually sufficient to elicit an abrupt cessation in firing that recovered equally briskly to the pre-drug rate when the retaining current was switched back on. The presence of intrinsic GABA tone on neurons in the dPAG was demonstrated by an increase in firing rate during ejection of the GABAA receptor antagonist BIC (1–30 nA). Forty-four of 58 neurons increased their firing rate in the presence of BIC while six were inhibited and eight neurons were unaffected. Effect of stage of the estrous cycle When data from rats at different stages of the estrous cycle were considered separately, significant differences emerged with respect to neuronal excitability. Firstly, the amount of DLH required to induce firing at comparable rates was higher during proestrus (Pro) and early diestrus (ED) compared with estrus (Est) and LD (Table 1). There was also a significant effect of cycle stage with respect to the effectiveness of BIC in increasing the firing rate. Responsiveness to BIC was assessed by increasing the ejecting current in steps between 0 (no ejecting current, retaining current off) and 30 nA for a maximum period of 2 min at each current. At currents in excess of 30 nA firing often became very erratic and/or the spike height became depressed. In LD and Est the threshold ejecting current at which BIC produced a significant increase in firing rate was lower (⫹3 nA and ⫹5 nA for LD and Est respectively) compared with Pro and ED (⫹10 nA in both stages) (Fig. 2). The firing rate in response to ⫹30 nA ejecting current (the maximum used) was also significantly higher during Est and LD (36.5⫾5.5 Hz and 33.0⫾3.7 Hz respectively) compared with Pro and ED (18.8⫾2.3 Hz and 18.0⫾1.8 Hz respectively) (Fig. 2; P⬍0.05, Pro and ED vs. Est and LD). Iontophoretically applied BIC did not produce changes in any of the cardiorespiratory variables measured. Responsiveness to a panicogenic agent I.v. administration. Responsiveness to i.v. injection of PG (40 ␮g kg⫺1) was tested on 17 cells in 10 rats. PG evoked changes in firing rate of 16/17 neurons. In the majority of responsive cells (n⫽12, 75%) firing rate increased (Fig. 3) while the remaining four (25%) showed a decrease in firing rate. Injection of 2⫻100 ␮l saline to replicate the volume of fluid injected when PG was administered (n⫽5), did not affect neuronal firing rate. PG also evoked changes in cardiorespiratory variables after i.v. administration. Resting mean arterial pressure (MAP), HR and rate of respiration (RR) were 96.3⫾3.7 mm Hg, 436.1⫾12.1 bpm and 90.2⫾3.9 breaths per minute respectively. Following i.v. injection of PG MAP increased to 111.1⫾3.4 mm Hg, HR to 448.6⫾10.6 bpm and RR to 101.8⫾5.4 breaths per minute (P⬍0.001, n⫽15 (due to technical difficulties complete data sets were unavailable

328

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

Fig. 1. Location of recording sites (filled circles) of neurons tested for responsiveness to iontophoretic application of BIC and GABA (upper panel), PG (middle panel) i.v. injection of 40 ␮g kg⫺1 PG (lower panel). Open circles in the lower panel represent recording sites of neurons that were tested for responsiveness to control injections of saline or gelofusin. Outlines taken from the atlas of Paxinos and Watson (1986). P numbers indicate distances in mm from bregma.

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335 Table 1. Ejecting current (nA) for DLH required to evoke a baseline neuronal discharge close to 5 Hz for neurones in recorded in the dPAG at different phases of the estrous cycle using five-barreled micropipettes Phase

Ejecting current (nA)

Pro Est ED LD

⫺5.7⫾1.8 (27) ⫺3.8⫾0.9 (34)* ⫺7.1⫾0.9 (29) ⫺3.2⫾0.8 (29)*

Data are mean⫾S.E.M. Numbers of cells recorded at each phase are shown in parentheses. * P⬍0.05 vs. ED, ANOVA with Fisher’s post hoc test.

in two cases in which the corresponding electrophysiological recordings were made from one responsive and one unresponsive cell)). The onset and peak increases in neuronal firing rate occurred later than the onset and peak of the increase in MAP, HR and RR. The duration of the increase in neuronal firing rate was also longer than the change in MAP, HR and RR (Table 2). To control for possible effects of the changes in blood pressure induced by PG, 0.5 ml of the plasma expander gelofusin was given as a rapid i.v. bolus injection to induce a transient increase in blood pressure of similar magnitude to that produced by i.v. administration of the peptide. Gelofusin had no effect on firing rate (n⫽4). Iontophoretic application. PG was applied iontophoretically to 53 neurons recorded within the dorsal half of the PAG. The majority (73.6%) showed an increase in firing rate while 20.8% showed a decrease and 5.6% were unaffected. Firing rate started to increase within 1– 8 s of drug application but returned to baseline more slowly over 5–20 s once the ejecting current was switched back to the retain current. Current–response relationships were examined in six cells that were excited by PG. Ejecting currents up to 30 nA evoked a linear increase in firing rate (Fig. 4). At currents in excess of 30 nA, the response became less predictable with four of the cells showing a further increase in firing rate but the remaining two showing a decrease. Maximum ejection currents were therefore limited to 30 nA for the remainder of the study. No change was seen in any of the cardiorespiratory variables measured after iontophoretic application of PG. In six neurons the effect of PG was tested in the presence of iontophoretically applied CR2945, a selective CCKB receptor antagonist (Revel et al., 1998). In the presence of CR2945 (60 nA) the increase in firing rate achieved during application of PG was reduced by 64⫾ 4.9%. (P⬍0.01, Fig. 5).

329

[LD]). However, the peak firing rate evoked by PG, ejected using a standard current of 30 nA, was significantly higher during estrus and late diestrus (12.9⫾1.7 Hz and 14.0⫾1.5 Hz respectively) compared with Pro and ED (7.3⫾0.9 Hz and 8.9⫾0.7 Hz respectively; P⬍0.05, Pro and ED vs. Est and LD) (Fig. 6). Although we did not select rats at a particular stage of the estrous cycle for experiments in which effects of i.v. administered PG were tested, examination of the vaginal smears that were taken routinely following completion of a recording period revealed that the group was composed of rats in either LD or ED. Interestingly, 90% (eight of nine cells) recorded in rats in LD were excited by i.v. injection of PG while only 57% (four of seven cells) were excited in animals in ED. These findings are in line with the cyclerelated differences in responsiveness to iontophoretically applied PG described above. Responsiveness to PG and level of GABAergic tone Responsiveness to PG and BIC was compared in 10 neurons that were excited by PG. There was a direct relationship between the size of the excitatory response to PG and the increase in firing rate induced in the presence of BIC (r⫽0.79) suggesting that the level of ongoing GABAergic tone was inversely correlated with the responsiveness to the peptide. Juxtacellular labeling A total of 20 cells recorded using ‘piggyback’ electrodes showed strong entrainment in response to depolarizing current pulses delivered using a 200 ms on 200 ms off duty cycle (Fig. 7). The electrophysiological characteristics of these neurons were qualitatively similar to those recorded using conventional five-barreled microelectrodes. All the cells were quiescent so that firing had to be evoked by ejection of DLH. The ejecting current for DLH needed to induce a mean firing rate of 5.1⫾0.5 Hz using piggyback

Responsiveness to PG at different stages of the estrous cycle Responsiveness to PG was compared in rats at different stages of the estrous cycle. Baseline firing rate evoked by continuous ejection of DLH did not differ significantly between different phases of the estrous cycle (3.9⫾0.2 Hz [Pro], 5.3⫾0.5 Hz [Est], 5.7⫾0.7 Hz [ED], 4.3⫾4.3 Hz

Fig. 2. Current–response relationship for BIC applied iontophoretically to neurons in the dPAG during Pro, Est, ED and LD. Data show means⫾S.E.M. # P⬍0.05: Pro vs. LD, * P⬍0.05: Pro and ED vs. Est and LD, ANOVA with Fisher’s post hoc test.

330

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

Fig. 3. (A) Example of the cardiorespiratory response and (B) change in neuronal firing rate evoked by i.v. PG (40 ␮g kg⫺1, time of injection indicated by the broken line). Sample spike records taken at times indicated by arrows.

311.4⫾82.5 ␮m). In three neurons a fine caliber axon was seen to arise from the soma and to course, without branching, toward the perimeter of the PAG (Fig. 7). On nine further occasions when a well-entrained cell had been recorded, the blue marking spot in the tissue dorsal to the recording sites revealed the recording site had been in the superior colliculus overlying the PAG or in the PAG ventral to the level of the aqueduct. The proportion of these cells that were recovered (33%) in histological material was similar to that of cells recorded within the dPAG.

electrodes (10.0⫾1.5 nA) was higher than that required to evoke similar levels of activity (5.1⫾0.2 Hz) using standard five-barreled microelectrodes (DLH 4.9⫾0.6 nA). However, this was not unexpected since higher concentrations of drug at the tip of the drug barrel of the piggyback electrodes would be needed for the tissue concentration at the tip of the protruding recording pipette to be equivalent to that achieved using traditional electrodes, in which the recording and drug ejection barrels are not displaced. Using piggyback electrodes, the proportions of cells that showed an increase in firing rate in response to PG (73%) was similar to the value using traditional five-barreled pipettes (74%). However, BIC produced an increase in firing rate in a lower proportion of cells (61% versus 82% for traditional multibarrel pipettes). Out of 20 entrained neurons in the dPAG, only eight filled cells (40%) were recovered successfully in histological material. The filled neurons were entrained for periods of 40 –360 s (mean 144.3⫾42.0 s). However, other cells that underwent similar or even longer periods of robust entrainment were not recovered. Neurobiotin-filled cells had multipolar or fusiform-shaped cell bodies between 12 ␮m and 40 ␮m in diameter (mean 21.0⫾3.1 ␮m). Between two and six primary dendrites, which were either smooth or varicose emanated from the cell body. Secondary dendrites were given off within 5–100 ␮m of the soma (mean 29.0⫾8.1 ␮m). The spread of the dendritic tree along its longest axis ranged from 59 to 675 ␮m (mean

DISCUSSION GABAergic tone in the PAG In the present study the population of neurons from which recordings were made in the dPAG in anesthetized female rats was quiescent. We therefore induced ongoing activity by continuous iontophoretic application of DLH. This strategy allowed us to use iontophoretic administration of BIC to investigate the level of intrinsic GABAergic inhibitory tone on the cells although it is inevitable that in the presence of DLH the new baseline membrane potential of the cells was depolarized beyond the normal resting level. Many anesthetics including urethane, act to potentiate the effects of GABA within the CNS (Harra and Harris, 2002). Thus spontaneous neuronal activity may have been depressed as a result of the anesthesia. However, in slice

Table 2. Time course of the cardiorespiratory response and change in firing rate for 11 of the 12 neurones in the dPAG that were excited after administration of i.v. PG Time point

Blood pressure

HR

RR

Neuronal firing rate

Latency to onset (s) Peak response (s) End of response (s)

8.0⫾0.7 18.7⫾1.5 56.4⫾5.5

8.6⫾0.7 17.8⫾1.5 47.2⫾4.1

10.3⫾1.5 23.7⫾2.2 64.5⫾5.1

43.3⫾20.5 63.3⫾23.9 87.8⫾24.9

Data for one cell omitted due to incomplete cardiorespiratory data set. Data are mean⫾S.E.M.

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

331

labeled cells was similar to those described previously for output neurons in the PAG, i.e. neurons whose axons projected without branching, toward the boundaries of the PAG (Lovick and Stezhka, 1999). The electrophysiological properties and morphological characteristics of the neurons recorded in the present study are therefore more akin to those expected of output neurons rather than an active inhibitory interneuron population. However, positive identification by other methods such as antidromic activation from medullary sites will be necessary to confirm this. Changes in neuronal excitability during the estrous cycle

Fig. 4. Current–response relationship for iontophoretically-applied PG. Data show means⫾S.E.M., n⫽6.

preparations of midbrain where anesthesia is not a confounding factor, we found that most neurons recorded in the dPAG using multibarreled glass microelectrodes similar to those used in the present study, were also quiescent (Stezhka and Lovick, 1994). Other studies using brain slices have also reported excitatory effects of BIC on neurons in the PAG (Ogawa et al., 1994), suggesting that ongoing GABA tone is present in the absence of anesthesia. In conscious rats, local microinjection of GABA antagonists into the PAG has been shown to result in arousal and escape behavior that mimicked the responses evoked by direct activation of neurons (Schenberg et al., 1983). Thus a high level of ongoing GABAergic activity appears to be a characteristic feature of the neural circuitry in the dPAG. A population of GABAergic interneurons in the PAG (Reichling, 1991) may be the source of the inhibitory tone although it is possible that inhibitory inputs arise from extrinsic sources as well.

In the present study, continuous iontophoretic application of DLH was used to offset the tonic GABAergic influence on the neurons being recorded. The dose of DLH was adjusted so that the baseline level of firing was similar for

Neuronal phenotype In the present study, neurons were recorded using either conventional multibarrel micropipettes or piggyback electrodes in which a multibarrel pipette used for iontophoretic application of drugs was set back from the tip of a single barrel recording pipette. The physiological and pharmacological characteristics of the cells recorded using either electrode type were qualitatively similar suggesting that they recorded from the same neuronal population within the dPAG. Several factors suggest that the recorded cells were projection neurons. The PAG is rich in GABA-containing interneurons (Reichling, 1991), especially in the dorsal half where they represent up to 50% of the total neuronal population (Mugnaini and Oertel, 1985). Measurements of the diameters of the subpopulation of GABAergic neurons that express GABAA receptor subunits (Griffiths and Lovick, 2005b) have also shown that this interneuron population is significantly smaller in diameter than the neurobiotin-filled cells investigated in the present study. The cells recorded in the present study were quiescent due to the presence of a tonic GABAergic inhibitory influence. The morphology of the neurobiotin-

Fig. 5. Effect of CR2945 on response iontophoretically applied PG. (A) Rate meter record from one representative neuron shows that the reproducible excitation evoked by PG (30 nA, left panel) was decreased in the presence of CR2945 (60 nA, middle panel) and recovered (right panel) within 5 min of ceasing to eject the antagonist. (B) Mean data from six neurons (⫾S.E.M.) illustrating the effect of CR2945 on PG-evoked excitation. Key: BL, baseline-firing rate evoked by continuous application of DLH. * P⬍0.05, ANOVA with Fisher’s post hoc test.

332

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

Fig. 6. Mean increase in firing rate evoked by iontophoretic application of 30 nA PG during Pro, Est, ED, LD. Data show means⫾S.E.M. * P⬍0.05, ANOVA with Fisher’s post hoc test.

each cell. Interestingly, the ejecting currents required were lower in rats in estrus and LD compared with Pro and ED suggesting that in estrus and LD there was a lower level of inhibitory tone to counteract. The effect of BIC was also estrous cycle dependent. A standard submaximal dose of BIC evoked a significantly greater increase in firing in rats in estrus and LD compared with rats in Pro and ED. The threshold current needed to elicit a significant increase in firing rate was also lower in estrus and LD. These factors suggest that BIC was able to more completely block the level of ongoing inhibition present during estrus and LD compared with Pro and ED. The majority of neurons recorded in the dPAG were excited by local or systemic administration of the panicogenic CCKB receptor agonist PG. Blockade of the response in the presence of CR2945 indicated the effect was CCKB receptor-mediated. These findings are in line with previous reports of CCK-evoked excitatory effects on neuronal activity in the PAG (Liu et al., 1994). The mechanism of action of PG in the dPAG is not known. In other brain regions CCKB receptor-mediated effects have been shown to be due to modulation of activity of GABAergic systems. In the hippocampus, CCK produced a CCKB receptormediated decrease in the firing frequency of GABAergic neurons (Deng and Lei, 2006). In nucleus accumbens excitatory effects of CCK produced by interactions with GABAergic systems have also been described (Kombian et al., 2004, 2005). Although CCKB receptors have been reported in the PAG (Mercer and Beart, 1997; Mercer et al., 2000) their relationship to the GABAergic neuronal population is not known. In the present study the neuronal responses to i.v. administration of PG were accompanied by cardiovascular changes. We have shown previously that the cardiovascular effects were CCKB-mediated (Brack et al., 2006). Moreover, they could be replicated by microinjection of PG directly into the dorsal PAG (Brack et al., 2006). Thus we suggest that at least part of the effects of systemic admin-

istration of PG is mediated centrally, by an action in the PAG. Like the response to BIC, the response to PG was estrous cycle dependent. Excitatory responses to iontophoretic application of the peptide were significantly greater in animals in LD and estrus compared with other stages of the cycle. I.v. injection of PG also evoked increases in firing rate in a higher proportion of neurons in rats in LD compared with ED (responses to i.v. PG in rats in estrus and Pro were not tested). Thus increased responsiveness to PG occurred when the level of ongoing GABAergic activity within the PAG was decreased. Recently, we found that cardiovascular responses to i.v. injection of PG were significantly greater in LD and estrus compared with other stages of the cycle (Brack et al., 2006). Thus the estrous cycle-related changes in neuronal responsiveness to PG in the dPAG correspond to a parallel increase in the functional responsiveness of the whole animal. Moreover, increased responsiveness to PG occurred at a time when there was a reduction in GABAergic inhibitory tone in the dPAG. Underlying mechanisms Recent studies have shown that GABAA receptors in the PAG and other brain regions can undergo considerable plasticity during the estrous cycle (Lovick et al., 2005; Maguire et al., 2005). These effects are likely to occur as a consequence of fluctuating plasma and brain levels of progesterone since the changes in subunit expression that occur at the end estrous cycle, when progesterone levels are falling, are mimicked during withdrawal from an exogenous progesterone dosing regime (Smith et al., 1998; Griffiths and Lovick, 2005a; Gulinello et al., 2002, 2003; Lovick et al., 2005, but see Maguire et al., 2005). In the PAG, expression of ␣4, ␤1 and ␦ subunits increased significantly during LD (Lovick et al., 2005). Moreover, the increased expression was confined principally to the GABAergic neuronal subpopulation (Griffiths and Lovick, 2005b). Recombinant ␣4␤1␦ GABAA receptors show a low EC50 for GABA that is consistent with activation by GABA at the concentration present in the extracellular fluid (Lerma et al., 1986; Lovick et al., 2005). Receptors that contain ␦ subunits are likely to be located extrasynaptically and to generate tonic currents (Mody, 2001, 2005; Farrant and Nusser, 2005). An increase in the tonic current carried by GABAergic neurons would be expected to decrease their excitability. Thus when expression of ␣4, ␤1 and ␦ GABAA receptor subunits is increased, the level of GABAergic tone on output neurons should decrease, rendering them more excitable. Since the effects of PG are likely to be mediated via an effect on GABAergic systems (see above), we can speculate that there might also be a causal relationship between falling progesterone levels and increased responsiveness to PG. Although increased expression of ␣4, ␤1 and ␦ GABAA receptor subunits may underlie the excitability changes that occur in the dPAG during LD, the increased responsiveness that occurs during estrus is more difficult to explain since GABAA receptor subunit expression is stable at

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

333

Fig. 7. Juxtacellular labeling of neurons in the dPAG. (A) Reconstruction of a labeled neuron entrained for 3 min and brain fixed 4.5 h after entrainment. (B) Photomicrograph of a section containing the soma and some proximal dendrites of the neurobiotin-filled neuron and the dorsallylocated PSB orientation mark. (C) Entrainment of the cell shown in A and B, using a 200 ms on 200 ms off positive current of 9 nA. Large deflections at the onset and offset of each current pulse are stimulus artifacts. (D) Location of recovered neurons in the PAG plotted onto an outline of a representative section through the PAG 7.3 mm caudal to bregma. Abbreviations: DR, dorsal raphe nucleus; mlf, medial longitudinal fasciculus.

this time. Although a short lasting (⬍8 h) surge in progesterone levels occurs in Pro (Butcher et al., 1974), this may not be long enough to induce changes in subunit expression, which have been demonstrated only after longer periods of exposure to or withdrawal from the hormone (Gulinello et al., 2001; Smith et al., 1998b). Another possibility is that hormones other than progesterone are responsible for the increase in neuronal excitability in the dPAG during estrus. The PAG contains high concentrations of estrogen ␣ receptors (ER␣) (Vanderhorst et al., 2005). At estrus plasma estrogen has fallen from the high level achieved during Pro (Watanabe et al., 1990). In ovariectomized rats, a single dose of 17-␤ estradiol produced a transient reduction in inhibitory GABAergic tone on hippocampal CA1 neurons that was maximal 24 h after injection of the steroid (Rudick and Woolley, 2001). This effect was associated with a decrease in expression of the GABA synthesizing hormone GAD65 (Rudick and Woolley, 2001). By analogy, similar events might be expected to occur in the PAG during estrus, around 24 h after the peak in

estrogen levels peak in Pro. In support of this idea both the binding of 3[H]muscimol to GABAA receptors in the PAG and the level of GABAergic tone are known to be decreased after estrogen treatment (O’Connor et al., 1988; Schumacher et al., 1989). However since the kinetics of these effects are not known, further experimentation is required.

CONCLUSION The present study in female rats has shown that neuronal excitability in the dPAG varies according to the stage of the estrous cycle. However, the changes in excitability do not parallel exactly the fluctuations in ovarian hormone levels. The increased neuronal responsiveness of putative output neurons to local application of a panicogenic agent seen in estrus and LD correlated with a decrease in the level of ongoing GABAergic inhibition. We suggest that in LD, increased expression of extrasynaptic ␣4␤1␦ GABAA receptors on inhibitory GABAergic interneurons as proges-

334

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335

terone levels are falling, may contribute to the increased excitability of the output neurons which renders the cells more susceptible to activation by panicogenic stimuli. The reason behind the increased neuronal excitability during estrous is puzzling as both progesterone and estrogen levels are low and relatively stable at this time. However, it may related to a delayed response linked to the hormonal status in Pro. Clearly, further experimentation is required. Nevertheless, some of the abovementioned processes may underlie the development of premenstrual symptoms in women, which include increased susceptibility to panic. Acknowledgments—This work was supported by the Wellcome Trust.

REFERENCES Acosta GB (2001) A possible interaction between CCKergic and GABAergic systems in the rat brain. Comp Biochem Physiol C Toxicol Pharmacol 128:11–17. Bertoglio LJ, Zangrossi H (2005) Involvement of dorsolateral periaqueductal gray cholecystokinin-2 receptors in regulation of a panicrelated behavior in rats. Brain Res 1059:46 –51. Brack JE, Griffiths J, Lovick TA (2004) Decreased levels of GABAergic tone in the periaqueductal grey matter during late dioestrus correlate with increased expression of ␣4, ␤1 and ␦ GABAA receptor subunits on GABAergic neurones. J Physiol 360P:C28. Brack KE, Jeffery SMT, Lovick TA (2006) Cardiorespiratory responses to a panicogenic agent in female Wistar rats at different stages of the oestrous cycle. Eur J Neurosci 23:3309 –3318. Brack KE, Lovick TA (2004) Changes in GABAergic inhibitory tone within the periaqueductal grey matter during the oestrous cycle in rats. Soc Neurosci Abstr 34:624.4. Brandao ML, Borelli KG, Nobre MJ, Santos JM, Albrecht-Souza L, Oliveira AR, Martinez RC (2005) GABAergic regulation of the neural organization of fear in the midbrain tectum. Neurosci Biobehav Rev 29:1299 –1311. Butcher RL, Collins WE, Fugo NW (1974) Plasma concentration of LH, FSH, prolactin, progesterone and estradiol-17beta throughout the 4-day estrous cycle of the rat. Endocrinology 94:1704 –1709. Deng P-Y, Lei S (2006) Bi-directional modulation of GABAergic transmission by cholecytsokinin in hippocampal dentate gyrus granule cells of juvenile rats. J Physiol 572:425– 442. Ensom MH (2000) Gender-based differences and menstrual cyclerelated changes in specific diseases: implications for pharmacotherapy. Pharmacotherapy 20:523–539. Farrant M, Nusser Z (2005) Variations on an inhibitory theme: phasic and tonic activation of GABA(A) receptors. Nat Rev Neurosci 6:215–229. Ferraro L, O’Connor WT, Glennon J, Tomasini MC, Tanganellim S, Antonelli T (2000) Evidence for a nucleus accumbens CCK2 receptor regulation of rat ventral pallidal GABA levels: a dual microdialysis study. Life Sci 68:483– 496. Griffiths JL, Lovick TA (2005a) Withdrawal from progesterone increases expression of ␣4, ␤1 and ␦ GABAA receptor subunits in neurons in the periaqueductal gray matter in female Wistar rats. J Comp Neurol 486:89 –97. Griffiths JL, Lovick TA (2005b) GABAergic neurones in the rat periaqueductal grey matter express ␣4, ␤1 and ␦ GABAA receptor subunits: plasticity during the oestrous cycle. Neuroscience 136: 457– 466. Gulinello M, Gong QH, Li X, Smith SS (2001) Short-term exposure to a neuroactive steroid increases ␣4 GABAA receptor subunit levels in association with increased anxiety in the female rat. Brain Res 910:55– 66. Gulinello M, Gong QH, Smith SS (2002) Progesterone withdrawal increases the alpha4 subunit of the GABA(A) receptor in male rats

in association with anxiety and altered pharmacology—a comparison with female rats. Neuropharmacology 34:701–714. Gulinello M, Orman R, Smith SS (2003) Sex differences in anxiety, sensorimotor gating and expression of the alpha4 subunit of the GABAA receptor in the amygdala after progesterone withdrawal. Eur J Neurosci 17:641– 648. Halbreich U (2003) The etiology, biology, and evolving pathology of premenstrual syndromes. Psychoneuroendocrinology 28 (Suppl 3): 55–99. Harra K, Harris RA (2002) The anaesthetic mechanism of urethane: effects on neurotransmitter-gated ion channels. Anesth Analg 94: 313–318. Jones GA, Llewellyn-Smith IJ, Jordan D (2002) Physiological, pharmacological, and immunohistochemical characterisation of juxtacellularly labelled neurones in rat nucleus tractus solitarius. Auton Neurosci 98:12–26. Jeffery S, Brack KE, Lovick TA (2005) Functional consequences of increased expression of ␣4, ␤1 and ␦ GABAA receptor subunits in the periaqueductal grey matter: heightened responsiveness to a panicogenic challenge in female rats. J Physiol 567P:C78. Kombian SB, Ananthalakshmi KVV, Parvathy SP, Motowe WC (2004) Cholecystokinin activates CCKB receptors to excite cells and depress EPSCs in the rat rostral nucleus accumbens in vitro. J Physiol 555:71– 84. Kombian SB, Ananthalakshmi KV, Parvathy SS, Matowe WC (2005) Cholecystokinin inhibits evoked inhibitory postsynaptic currents in the rat nucleus accumbens indirectly through gamma-aminobutyric acid and gamma-aminobutyric acid type B receptors. J Neurosci Res 79:412– 420. Lerma J, Herranz AS, Herreras O, Abraira V, DelRio RM (1986) In vivo determination of extracellular concentration of amino-acids in the rat hippocampus: a method based on brain dialysis and computerised analysis. Brain Res 384:145–155. Liu H, Chandler S, Beitz AJ, Shipley MT, Behbehani MM (1994) Characterisation of the effect of cholecystokinin (CCK) on neurones in the periaqueductal gray of the rat: immunocytochemical and in vivo and in vitro electrophysiological studies. Brain Res 642:83–94. Lovick TA (2000) Panic disorder: A malfunction of multiple transmitter control systems with in the periaqueductal gray matter? Neuroscientist 6:48 –59. Lovick TA (2001) Involvement of GABA in medullary raphe-evoked inhibition of neuronal activity within the periaqueductal grey matter. Exp Brain Res 137:214 –218. Lovick TA (2006) Plasticity of GABAA receptor subunit expression during the oestrous cycle of the rat: implications for premenstrual syndrome in women. Exp Physiol 91:655– 860. Lovick TA, Griffiths JL, Dunn SMJ, Martin IL (2005) Changes in GABAA receptor subunit expression during the oestrous cycle in Wistar rats. Neuroscience 131:397– 405. Lovick TA, Stezhka VV (1999) Neurones in the dorsolateral periaqueductal grey matter in coronal slices of midbrain: electrophysiological and morphological characteristics. Exp Brain Res 124:53–58. Maguire JL, Stell BM, Rafizadeh M, Mody I (2005) Ovarian cyclelinked changes in GABAA receptors mediating tonic inhibition alter seizure susceptibility and anxiety. Nat Neurosci Rev 8:797– 804. McLaughlan RI, Robertson DM, Healy DL, Burger HG, de Kretser DM (1987) Circulating immunoreactive inhibin levels during the normal human menstrual cycle. J Clin Endocrinol Metab 65:954 –961. Mercer LD, Beart PM (1997) Histochemistry in rat brain and spinal cord with an antibody directed at the cholecystokininA receptor. Neurosci Lett 225:97–100. Mercer LD, Le VQ, Nuna J, Jones NM, Beart P (2000) Direct visualization of cholecystokinin subtype2 receptors in rat central nervous system using anti-peptide antibodies. Neurosci Lett 293:167–170. Mody I (2001) Distinguishing between tonic and phasic conductances. Neurochem Res 26:907–913.

K. E. Brack and T. A. Lovick / Neuroscience 144 (2007) 325–335 Mody I (2005) Aspects of the homeostatic plasticity of GABAA receptor-mediated inhibition. J Physiol 56:237–246. 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: Handbook of chemical neuroanatomy, Vol. 4: GABA and neuropeptides in the CNS, part 1 (Björkland A, Hökfelt T, eds), pp 436 – 608. Amsterdam: Elsevier. O’Connor LH, Nock B, McEwen BS (1988) Regional specificity of gamma-aminobutyric acid receptor regulation by estradiol. Neuroendocrinology 47:473– 481. Ogawa S, Kow L-M, Pfaff DW (1994) In vitro electrophysiological characterization of midbrain periaqueductal gray neurons in female rats: responses to GABA- and met-enkephalin-related agents. Brain Res 666:239 –249. Olsson M, Ho HP, Annerbrink K, Melchior LK, Hedner J, Eriksson E (2003) Association between estrus cycle-related changes in respiration and estrus cycle-related aggression in outbred female Wistar rats. Neuropsychopharmacology 28:704 –710. Paxinos G, Watson S (1986) A stereotaxic atlas of the rat brain. 2nd ed. New York: Academic Press. Perez de la Mora M, Hernandez-Gomez AM, Mendez-Franco J, Fuxe K (1993) Cholecystokinin-8 increases K(⫹)-evoked [3H] gammaaminobutyric acid release in slices from various brain areas. Eur J Pharmacol 250:423– 430. Pinault D (1996) A novel single-cell staining procedure performed in vivo under electrophysiological control: morpho-functional features of juxtacellularly labelled thalamic cells and other central neurons with biocytin or neurobiotin. J Neurosci Methods 65:113–136. Ranovska A (1995) Cholecystokinin-GABA interactions in rat striatum. Neuropeptides 29:257–262. Reichling DB (1991) GABAergic neuronal circuitry in the periaqueductal gray matter. In: The midbrain periaqueductal gray matter. Functional, anatomical and neurochemical organization (Depaulis A, Bandler R, eds), pp 329 –344. New York: Plenum Press. Revel L, Mennuni L, Garafalo P, Makovec F (1998) CR2954: a novel CCKB receptor antagonist with anxiolytic-like activity. Behav Pharmacol 9:183–194. Rudick CN, Woolley CS (2001) Estrogen regulates functional inhibition of hippocampal CA1 pyramidal cells in the adult female rat. J Neurosci 21:6532– 6543.

335

Schenberg LC, DeAguiar C, Graeff FG (1983) GABA modulation of the defense reaction induced by brain electrical stimulation. Physiol Behav 31:429 – 443. Schumacher M, Coirini H, McEwen BS (1989) Regulation of high affinity GABAa receptors in specific brain regions by ovarian hormones. Neuroendocrinology 50:315–310. Siniscalchi A, Rodi D, Cavallini S, Marino S, Fewrraro L, Beani L, Bianchi C (2003) Effects of cholecystokinin tetrapeptide (CCK(4)) and of anxiolytic drugs on GABA outflow from the cerebral cortex of freely moving rats. Neurochem Int 42:87–92. Smith SS, Gong QH, Hsu F-C, Marcowitz RS, Ffrench-Mullen JM, Li X (1998a) GABAA receptor ␣4 subunit expression prevents withdrawal properties of an endogenous steroid. Nature 382:926 –930. Smith SS, Gong QH, Li X, Moran MH, Bitran D, Frye CA, Hsu FC (1998b) Withdrawal from 3alpha-OH-5alpha-pregnan-20-one using a pseudopregnancy model alters the kinetics of hippocampal GABAA-gated current and increases the kinetics of hippocampal GABAA-gated current and increases GABAA receptor alpha 4 subunit in association with increased anxiety. J Neurosci 18:5275– 5284. Stezhka VV, Lovick TA (1994) Inhibitory and excitatory projections from the dorsal raphe nucleus to neurones in the dorsolateral PAG in slices of midbrain maintained in vitro. Neuroscience 62:177–187. Vanderhorst VGJM, Gustafsson J-A, Ulfhake B (2005) Estrogen receptor-␣ and -␤ immunoreactive neurons in the brainstem and spinal cord of male and female mice: relationships to monoaminergic, cholinergic, and spinal projection systems. J Comp Neurol 488:152–179. van Megen HJ, Westenberg HG, den Boer JA, Haigh JR, Traub M (1994) Pentagastrin induced panic attacks: enhanced sensitivity in panic disorder patients Psychopharmacology 114:449 – 455. Watanabe G, Taya K, Sasamoto S (1990) Dynamics of ovarian inhibin secretion during the oestrous cycle of the rat. J Endocrinol 126: 151–157. Zanovelli JM, Netto FS, Guimaraes H, Zangrossi (2004) Systemic and intra-dorsal periaqueductal gray injections of cholecystokinin sulphated octapeptide (CCK-8s) induce a panic-like response in rats submitted to the elevated T-maze. Peptides 25:1935–1941.

(Accepted 29 August 2006) (Available online 11 October 2006)