Introduction

The involvement of the GABAA receptor in anxiety has been extensively studied and confirmed (Nemeroff 2003). The pentameric GABAA receptor consists of five subunits (α1–6, β1–3, γ1–3, δ, ɛ, θ, and π), and the assembly of different combinations of subunits allows the construction of different types of GABAA receptors, each having specific functional and pharmacological properties (Korpi et al. 2002). The majority of GABAA receptors are composed of two α subunits, two β subunits, and one γ subunit (Tretter et al. 1997). Classical benzodiazepines bind to GABAA receptors containing α1, α2, α3, and/or α5 subunits, while binding affinity to α4-containing and α6-containing subunits is extremely weak (Rudolph and Mohler 2004). Besides the preferred anxiolytic action, the use of benzodiazepines is associated with dependence, anticonvulsant activity, sedation, amnesia, and daytime drowsiness (Stewart and Westra 2002). These different benzodiazepine effects are thought to be mediated through different GABAA receptor subtypes. Therefore, the search for new anxiolytics focuses on subunit-selective GABAA receptor agonists. Both genetic and pharmacological studies suggest a major role of the α2 and the α3 GABAA receptor subunit in mediating anxiolysis (Atack et al. 2005; Dias et al. 2005; Low et al. 2000; Rudolph and Mohler 2004). Consistent with this role, the α2 and the α3 GABAA receptor subunits are expressed in anxiety-involved areas like the amygdala and bed nucleus of the stria terminalis (Pirker et al. 2000). The GABAA receptor α1 subunit is associated with sedative and amnesic effects, while it is not involved in anxiolysis (McKernan et al. 2000; Rowlett et al. 2005; Rudolph et al. 1999). Compounds lacking activity at the α1-containing GABAA receptor while still modulating the α2 and/or α3 GABAA receptor subunit appear to be prime candidates for nonsedating anxiolytic drugs (de Haas et al. 2007; Huang et al. 2000). The research for novel anxiolytics has focused on compounds with selective efficacy at different subunits (while binding to all subunits) rather than compounds with different affinities for the subunits (Atack 2005). We therefore aimed to characterize the effects of different GABAA ligands on temperature and locomotor responses to novel cage stress in rats, using home cage telemetry, in order to deduce the relative contributions of α subunits of the GABAA receptor. Somatic stress symptoms are mediated by the autonomic nervous system and constitute a functional response in both humans and animals. The stress-induced hyperthermia (SIH) paradigm uses the physiological transient rise in body temperature in response to stress as a new and translational alternative in anxiety research (Bouwknecht et al. 2007; Vinkers et al. 2008). Using this paradigm, anxiolytic drugs including most benzodiazepines have been shown to dose-dependently attenuate the SIH response (Bouwknecht et al. 2007; Olivier et al. 2002; Van Bogaert et al. 2006). Using telemetry, stress-induced home cage temperature and locomotor activity responses can be simultaneously recorded, thus facilitating the comparison of effects on body temperature and locomotor activity caused by various GABAergic drugs. We hypothesized that anxiolytic effects would cause the SIH response to decrease without influencing basal body temperature and stress-induced locomotor activity responses. GABAergic sedative effects on the other hand would be characterized by a decrease in locomotor activity as well as general hypothermic state.

In the present study, we investigated the nonsubunit-selective GABAA receptor agonist diazepam (Pritchett et al. 1989), the intermediate selective α1 subunit GABAA receptor agonist zolpidem, as well as the selective α3 subunit GABAA receptor agonist TP003 (Dias et al. 2005). Zolpidem is approximately fivefold to tenfold more selective for α1 subunit-containing GABAA receptors than α2 and α3 subunit-containing receptors (Ebert et al. 2006; Petroski et al. 2006). However, zolpidem may demonstrate less selectivity in vivo compared to studies using recombinant receptors (Atack et al. 1999). TP003 has lower efficacies at the different α subtypes with less than 15% potentiation at the α2 and α5 subunit compared to diazepam (Dias et al. 2005). Exposure to higher drug doses may lead to loss of selectivity. We combined these compounds with the selective α1 subunit GABAA receptor antagonist beta-carboline-carboxy-tert-butyl ester (βCCt; Huang et al. 1999).

Less abundant populations of a δ subunit-containing GABAA receptors are often located extrasynaptically and perisynaptically and are thought to be involved in a continuous active inhibitory tone instead of the phasic inhibitory tone caused by intrasynaptic agonists (Jia et al. 2005; Nusser et al. 1998). Alcohol is anxiolytic at low doses and has been shown to bind to extrasynaptic GABAA receptors containing α4 or α6 and δ subunits (Wallner et al. 2003). However, at higher doses, ethanol can modulate excitatory N-methyl-d-aspartic acid (NMDA) and non-NMDA glutamate receptors, serotonin and glycine receptors, as well as potassium and calcium channels (Crews et al. 1996; Davies 2003; Harris 1999). Also, the fact that δ-deficient mice demonstrated a normal anxiolytic and hypothermic response to ethanol and that the alcohol sensitivity of α4β3δ GABAA receptors could not be replicated (Borghese et al. 2006) indicates that the discussion on the mechanism by which ethanol activates the GABAA receptor is still ongoing (Mihalek et al. 2001). Generally, the sedative and anxiolytic effects of alcohol were not altered after deletion of the α1 subunit, suggesting that other yet unexplained factors may play a role (Kralic et al. 2003). Moreover, we studied the nonbenzodiazepine hypnotic drug THIP (gaboxadol) that also binds to extrasynaptic GABAA receptor δ subunits with putative anxiolytic effects (Elfline et al. 2004; Wafford and Ebert 2006).

Materials and methods

Animals

Male Wistar rats (Harlan Zeist, The Netherlands) were used in the current study. Rats were housed socially in a controlled environment with a nonreversed 12 h light/dark cycle (white lights on from 7 am to 7 pm). Animals had unlimited access to food (standard laboratory chow) and water. One week after arrival, telemetry transmitters were implanted. After recovery from surgery, rats were housed in groups of four in type IV Macrolon® cages with a plastic tube as cage enrichment. Food (standard laboratory chow) and tap water were available ad libitum. Once a week, an experimental procedure was carried out. All experiments were carried out with the approval of the ethical committee on animal experiments of the Faculties of Sciences, Utrecht University, The Netherlands, and in accordance with the Declaration of Helsinki.

Surgery

A telemetric device (type ETA-F20, Data Sciences International, St. Paul, MN, USA) was implanted in the abdominal cavity as described earlier (Pattij et al. 2001). Prior to surgery, rats received a subcutaneous (s.c.) injection (2 mL/kg) of the antibiotic Baytrill® (2.5% enrofloxacin). Rats were anesthetized using O2/NO2/isoflurane gas anesthesia. Carprofen (5 mg/kg, s.c.) was given as an analgesic immediately after surgery and twice daily for 2 days after surgery. After surgery, animals were housed individually for 1 week and recovery from surgery was monitored (weight, heart rate, temperature). Also, all rats had access to wet food and solid drinks (gel formula as a water replacement) for 2 days after surgery. Wound recovery was regularly checked. One animal repeatedly opened the abdomen wound and, therefore, had to be sacrificed and replaced. During the experiments, one rat was removed from the experiments due to an inflammation surrounding the telemetry device.

Experimental procedure

On the afternoon before an experimental day, rats were weighed and housed individually in a type III Macrolon® cage, located on a telemetric receiver. The telemetric transmitters were activated using a magnet. Data collection was subsequently started. The day after, the SIH procedure was initiated, consisting of an injection (intraperitoneal [i.p.] or oral [p.o.]) with vehicle or a certain drug dose. Immediately after injection, rats were placed back into their individual cage. Rats were placed in a novel cage (clean cage with fresh bedding and a paper tissue) 60 min later and left undisturbed for approximately 2 h afterwards. At the end of the experimental day, rats were group-housed again and transmitters were turned off. To prevent habituation to the novel cage procedure, the interval between two experiments was set to be at least 1 week. A within-subject design was used, and all animals received all (combined) doses of the drugs.

Drugs

Diazepam, zolpidem, alcohol, and THIP HCl (gaboxadol) were obtained from Sigma Aldrich. βCCt was synthesized by the laboratory of Dr. J. M. Cook, University of Wisconsin-Milwaukee. TP003 was synthesized according to published methods (Dias et al. 2005; Goodacre et al. 2003; Humphries et al. 2006). An injection volume of 2 mL/kg was used for intraperitoneal injections of all drugs, except THIP HCl (5 mL/kg, i.p.) and alcohol (5 mL/kg p.o.). Diazepam, zolpidem, βCCt, TP003, and gaboxadol were suspended in gelatin–mannitol 0.5%/5%. When βCCt was combined with diazepam or zolpidem, βCCt at a dose of 10 mg/kg was injected 10 min prior to diazepam/zolpidem injection. Fresh solutions and suspensions were prepared each testing day.

Data analysis

All data were collected in 5-min blocks and are displayed ±SEM. All experiments were carried out with a within-subject design. Body temperature and locomotor activity were analyzed during the first hour after novel cage using a univariate repeated-measures analysis of variance (ANOVA) with manipulations time and treatment as within-subject factors. Simple contrast tests were used to compare drug with vehicle conditions whenever a significant main effect for drug (indicating an effect on the basal body temperature) or a significant drug × time interaction effect (indicating an effect on the SIH response) was observed. Also, the SIH response was calculated from the telemetry data for each individual rat by subtracting the body temperature at t = 0 from the maximum temperature reached within the first 30 min after the novel cage procedure and compared using a repeated-measures ANOVA with drug as within-subject factor and simple contrasts to compare drug with vehicle conditions. In addition, cumulative locomotor activity after the first 60 min after injection and cumulative locomotor activity after the first 60 min after the novel cage procedure were calculated and compared using repeated-measures ANOVA. A probability level of p < 0.05 was set as statistically significant; probability levels between p = 0.05 and p = 0.1 were regarded as trends.

Results

Diazepam

Summary

Diazepam dose-dependently attenuated the SIH response to novel cage stress without affecting basal body temperature and reduced locomotor activity levels before and after the novel cage procedure only at higher doses (Fig. 1a, c; n = 11).

Fig. 1
figure 1

The effects of diazepam with and without βCCt on the novel cage-induced temperature and locomotor responses (t = −60 injection, t = 0 novel cage stress). *p < 0.05; **p < 0.01; ***p < 0.001. Inset calculated SIH response from the telemetry data. a Diazepam (0–4 mg/kg) dose-dependently reduced the SIH response. Inset calculated SIH response from the telemetry data. b Diazepam at a dose of 4 mg/kg reduced core body temperature and the SIH response. Prior injection with βCCt (10 mg/kg) prevented basal core body temperature reduction without affecting the diazepam-induced reduction of the SIH response *p < 0.05, diazepam effect and diazepam × βCCt interaction. Inset calculated SIH response from the telemetry data. c Diazepam (4 mg/kg) reduced stress-induced locomotor activity responses [Inset white bar cumulative locomotor activity t = −60 to t = 0 (after injection), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage)]. d Diazepam (4 mg/kg) reduced stress-induced locomotor activity responses with no effect of βCCt [Inset white bar cumulative locomotor activity t = −60 to t = 0 (after injection), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage)]. e βCCt (0–20 mg/kg) did not affect SIH responses Inset calculated SIH response from the telemetry data

Full size image

Body temperature

The novel cage SIH response (F(12,120) = 22.0, p < 0.001) was reduced by diazepam (diazepam × time interaction: F(36,360) = 4.31, p < 0.001). Diazepam did not influence basal body temperature (diazepam effect: F(3,30) = 1.04, p = 0.39, NS) (Fig. 1a). The calculated SIH response revealed a diazepam effect on the SIH response (F(3,30) = 12.74, p < 0.001). Simple contrasts revealed SIH attenuation at higher doses (1 mg/kg–vehicle: F(1,10) = 3.85, p = 0.08, trend; 2 mg/kg–vehicle: F(1,10) = 6.03, p < 0.05; 4 mg/kg–vehicle: F(1,10) = 33.51, p < 0.001) (Fig. 1a, inset).

Locomotor activity

The novel cage stress-induced locomotor activity response (F(12,120) = 20.55, p < 0.001) was overall diminished by diazepam (diazepam effect: F(3,30) = 4.98, p < 0.01), although not dependent upon time (diazepam × time interaction: F(3,30) = 1.33, p = 0.11, NS). Only the higher doses of diazepam influenced locomotor activity [planned comparisons: vehicle–1 mg/kg (F(1,10) = 1.68, p = 0.22, NS), vehicle–2 mg/kg (F(1,10) = 8.18, p < 0.05), and vehicle–4 mg/kg (F(1,10) = 7.03, p < 0.05)]. When cumulating locomotor activity levels after injection and after stress (Fig. 1c, inset), diazepam reduced locomotor activity levels at higher doses (main diazepam effect: F(3,30) = 3.03, p < 0.05; simple contrasts: 2 mg/kg vs vehicle F(1,10) = 3.85, p = 0.08, NS; 4 mg/kg vs vehicle F(1,10) = 5.19, p < 0.05). Activity levels were larger after the novel cage procedure than after injection stress (stress effect: F(1,10) = 8.55, p < 0.05).

βCCt

Body temperature

βCCt alone did not affect the SIH response (time effect: F(12,120) = 53.00, p < 0.001; βCCt effect: F(3,30) = 0.70, p = 0.56, NS; βCCt × time interaction: F(36,360) = 1.17, p = 0.24, NS) (Fig. 1e; n = 11).

The calculated SIH response revealed no βCCt effect on the SIH response (F(3,30) = 0.33, p = 0.80, NS). Simple contrasts revealed that there was no attenuation of the SIH response at any dose (3 mg/kg–vehicle: F(1,10) = 0.06, p = 0.83, NS; 10 mg/kg–vehicle: F(1,10) = 0.29, p = 0.61, NS; 20 mg/kg–vehicle: F(1,10) = 0.13, p = 0.72, NS) (Fig. 1e, inset).

Locomotor activity

βCCt did not influence the stress-induced locomotor activity responses (time effect: F(12,120) = 13.72, p < 0.001; βCCt effect: F(3,30) = 0.23, p = 0.88, NS; βCCt × time interaction: F(36,360) = 1.02, p = 0.45, NS) (data not shown).

Diazepam and βCCt

Summary

βCCt was able to partially reverse the diazepam-induced hypothermia without affecting diazepam’s ability to reduce the SIH response. βCCt was not able to reverse the diazepam-induced locomotor reduction (Fig. 1b, d; n = 8).

Body temperature

When combined with βCCt, the SIH response (time effect: F(12,84) = 9.85, p < 0.001) was overall reduced by diazepam (diazepam × time interaction: F(12,84) = 5.17, p < 0.001). βCCt did not influence the SIH response (βCCt × time interaction: F(12,84) = 1.38, p = 0.19, NS). Diazepam reduced basal body temperature (diazepam effect: F(1,7) = 6.96, p < 0.05) and βCCt influenced the diazepam-induced hypothermia (diazepam × βCCt interaction: F(1,7) = 6.18, p < 0.05) without altering the body temperature itself (βCCt effect: F(1,7) = 1.24, p = 0.30, NS). The calculated SIH response revealed that diazepam reduced the SIH response (diazepam effect: F(1,7) = 16.94, p < 0.01). βCCt itself did not affect the SIH response (βCCt effect: F(1,7) = 1.86, p = 0.22, NS) nor did βCCt affect the attenuation of the SIH response by diazepam (diazepam × βCCt interaction: F(1,7) = 0.68, p = 0.44, NS) (Fig. 1b, inset).

Locomotor activity

When diazepam was injected after βCCt, the stress-induced locomotor activity response (time effect: F(12,84) = 7.65, p < 0.001) was generally and time-dependently reduced by diazepam (diazepam effect: F(1,7) = 9.48, p < 0.05; diazepam × time interaction: F(12,84) = 3.45, p < 0.001) with no effect of βCCt (βCCt effect: F(1,7) = 0.20, p = 0.68, NS; βCCt × time interaction: F(12,84) = 0.32, p = 98, NS; βCCt × diazepam interaction: F(1,7) = 0.02, p = 0.88, NS). When cumulating locomotor activity levels after injection and after stress (Fig. 1d, inset), diazepam generally reduced locomotor activity levels (main diazepam effect: F(1,7) = 10.86, p = 0.01; stress effect: F(1,7) = 2.47, p = 0.16, NS) without effect of βCCt (βCCt effect: F(1,7) = 0.05, p = 0.82, NS; βCCt × diazepam interaction: F(1,7) = 0.21, p = 0.66, NS).

Zolpidem

Summary

Zolpidem dose-dependently reduced basal body temperature and the SIH response and attenuated stress-induced and basal locomotor activity levels (Fig. 2a, c; n = 12).

Fig. 2
figure 2

The effects of zolpidem with and without βCCt on the novel cage-induced temperature and locomotor responses (t = −60 injection, t = 0 novel cage stress). *p < 0.05; **p < 0.01; ***p < 0.001. a Zolpidem (0–30 mg/kg) dose-dependently reduced the SIH response and basal body temperature. Inset calculated SIH response from the telemetry data. b Zolpidem at a dose of 10 mg/kg reduced core body temperature and the SIH response. Prior injection with βCCt (10 mg/kg) reversed basal core body temperature reduction without affecting the zolpidem-induced reduction of the SIH response. **p < 0.01, zolpidem effect; *p < 0.05, zolpidem × βCCt interaction. Inset calculated SIH response from the telemetry data. c Zolpidem (0–30 mg/kg) dose-dependently reduced stress-induced locomotor activity responses. Inset: white bar cumulative locomotor activity t = −60 to t = 0 (after injection), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage). d Zolpidem (10 mg/kg) reduced stress-induced locomotor activity responses. βCCt partially reversed zolpidem-induced locomotor sedation. *p < 0.05, zolpidem × βCCt interaction. Inset: white bar cumulative locomotor activity t = −60 to t = 0 (after injection), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage)

Full size image

Body temperature

Zolpidem reduced basal body temperature (main zolpidem effect: F(3,33) = 9.85, p < 0.001). Basal body temperature was found to be reduced in all three dosages [planned comparisons: vehicle–3 mg/kg (F(1,11) = 8.89, p < 0.05), vehicle–10 mg/kg (F(1,11) = 27.98, p < 0.001), and vehicle–30 mg/kg (F(1,11) = 26.73, p < 0.001)]. The SIH response (time effect: F(12,132) = 15.70, p < 0.001) was reduced by zolpidem (zolpidem × time interaction: F(36,396) = 10.12, p < 0.001). The calculated SIH response revealed that zolpidem reduced the SIH response (F(3,33) = 12.71, p < 0.001). Simple contrasts revealed SIH attenuation at all doses (3 mg/kg–vehicle: F(1,11) = 18.17, p < 0.001; 10 mg/kg–vehicle: F(1,11) = 100.61, p < 0.001; 30 mg/kg–vehicle: F(1,11) = 24.10, p < 0.001) (Fig. 2a, inset).

Locomotor activity

The stress-induced locomotor response (main time effect: F(12,132) = 12.48, p < 0.001) was reduced by zolpidem (main zolpidem effect: F(3,33) = 7.41, p < 0.001; zolpidem × time interaction: F(36,396) = 1.98, p < 0.001). Locomotor activity was found to be reduced in all three dosages [planned comparisons: vehicle–3 mg/kg (F(1,11) = 8.27, p < 0.05), vehicle–10 mg/kg (F(1,11) = 12.25, p < 0.01), and vehicle–30 mg/kg (F(1,11) = 13.34, p < 0.01)]. When cumulating locomotor activity levels after injection and after novel cage stress (Fig. 2c, inset), zolpidem was found to reduce overall locomotor activity (main zolpidem effect: F(3,33) = 12.21, p < 0.001; zolpidem × stress interaction: F(3,33) = 0.31, p = 0.69, NS). Simple contrasts showed that all doses of zolpidem reduced cumulative locomotor activity (vehicle–3 mg/kg: F(1,11) = 12.28, p < 0.01; vehicle–10 mg/kg: F(1,11) = 12.37, p < 0.01; vehicle–30 mg/kg: F(1,11) = 25.11, p < 0.001). Novel cage-induced locomotor levels were larger than injection-induced locomotor levels (stress effect: F(3,33) = 17.18, p < 0.01).

Zolpidem and βCCt

Summary

βCCt was able to partially reverse the overall zolpidem-induced hypothermia as well as time-dependently partially reverse the zolpidem-induced locomotor sedation (Fig. 2b, d; n = 8).

Body temperature

When combined with βCCt, zolpidem did not significantly reduce the SIH response (main time effect: F(12,84) = 9.75, p < 0.001; zolpidem × time interaction: F(12,84) = 0.93, p = 0.54, NS). Also, βCCt did not influence the SIH response (βCCt × time interaction: F(12,84) = 0.92, p = 0.53, NS). Zolpidem reduced basal body temperature (main zolpidem effect: F(1,7) = 11.12, p < 0.01), and βCCt influenced the zolpidem-induced hypothermia (zolpidem × βCCt interaction: F(1,7) = 6.31, p < 0.05) without altering the body temperature itself (βCCt effect: F(1,7) = 2.54, p = 0.16, NS). The calculated SIH response revealed that zolpidem reduced the SIH response (diazepam effect: F(1,7) = 11.31, p = 0.01). βCCt itself did not affect the SIH response (βCCt effect: F(1,7) = 0.01, p = 0.97, NS) nor did βCCt affect the attenuation of the SIH response by zolpidem (zolpidem × βCCt interaction: F(1,7) = 0.25, p = 0.63, NS) (Fig. 2b, inset).

Locomotor activity

When combined with βCCt, zolpidem overall reduced locomotor responses (main zolpidem effect: F(1,7) = 7.80, p < 0.05; zolpidem × time interaction: F(1,7) = 1.99, p < 0.05). βCCt had no overall effect on locomotor responses (βCCt effect: F(1,7) = 1.94, p = 0.21, NS). However, βCCt reversed locomotor activity in the zolpidem group dependent upon time (zolpidem × βCCt × time interaction: F(12,84) = 1.91, p < 0.05; zolpidem × βCCt interaction: F(1,7) = 0.18, p = 0.69, NS). When cumulating locomotor activity levels after injection and after novel cage stress (Fig. 2d, inset), zolpidem reduced basal and stress-induced locomotor activity (zolpidem effect: F(1,7) = 16.73, p < 0.01; stress effect: F(1,7) = 1.61, p = 0.25, NS) without overall effect of βCCt (βCCt effect: F(1,7) = 2.09, p = 0.19, NS; βCCt × zolpidem interaction: F(1,7) = 1.47, p = 0.26, NS).

TP003

Summary

TP003 reduced the SIH response at higher doses as well as reduced basal body temperature and attenuated novel cage-induced activity more than injection-induced activity (Fig. 3a, b; n = 10).

Fig. 3
figure 3

The effects of TP003 (0–3 mg/kg, a and b), alcohol (0–3 g/kg, c and d), and THIP (0–10 mg/kg, e and f) on the novel cage-induced temperature and locomotor responses (t = −60 injection, t = 0 novel cage stress). *p < 0.05; **p < 0.01; ***p < 0.001. a TP003 reduced the SIH response at higher doses. Inset: calculated SIH response from the telemetry data. b TP003 dose-dependently reduced stress-induced locomotor activity responses (***). Inset graphs: white bar cumulative locomotor activity t = −60 to t = 0 (after injection stress), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage stress). [TP003 reduced overall locomotor activity (TP003 effect: *p < 0.001), but more so after the novel cage procedure (TP003 × stress effect, *p < 0.001)]. c Alcohol at the highest dose reduced the SIH response (***) and basal body temperature (**). Inset calculated SIH response from the telemetry data. d Alcohol did not affect stress-induced locomotor activity responses. Inset graphs: white bar cumulative locomotor activity t = −60 to t = 0 (after injection stress), gray bar cumulative locomotor activity t = 0 to t = 60 (after novel cage stress). e THIP at the highest dose reduced the SIH response (***) and basal body temperature (**). Inset: calculated SIH response the telemetry data. f THIP at the highest dose reduced stress-induced locomotor activity responses. Inset THIP did not reduce overall locomotor activity, but did reduce locomotor activity after the novel cage procedure (gaboxadol × stress effect, ***p < 0.05)

Full size image

Body temperature

The SIH response (main time effect: F(12,108) = 19.27, p < 0.001) was attenuated by TP003 (TP003 × time interaction: F(36,324) = 1.93, p < 0.01). TP003 did influence basal core body temperature (F(3,27) = 2.96, p = 0.050). Planned comparisons revealed a significant difference between the vehicle and 1 mg/kg condition (F(1,9) = 6.26, p < 0.05), a trend for a difference between vehicle and 3 mg/kg condition (F(1,9) = 3.65, p = 0.09, NS), and no difference between vehicle and 0.3 mg/kg condition (F(1,9) = 0.27, p = 0.62,NS). The calculated SIH response revealed TP003 reduced the SIH response (F(3,27) = 12.57, p < 0.001). Simple contrasts revealed SIH attenuation at all doses (0.3 mg/kg–vehicle: F(1,9) = 22.25, p < 0.001; 1 mg/kg–vehicle: F(1,9) = 25.50, p < 0.001; 3 mg/kg–vehicle: F(1,9) = 27.79, p < 0.001) (Fig. 3a, inset).

Locomotor activity

Stress-induced locomotor responses (main time effect: F(12,108) = 22.43, p < 0.001) were reduced by TP003 (TP003 effect: F(3,27) = 14.43, p < 0.001; TP003 × time interaction: F(36,324) = 1.42, p = 0.06). All three TP003 doses resulted in significant activity reduction after novel cage stress compared to the vehicle group (F(1,9) = 23.94, p < 0.001 for vehicle–0.3 mg/kg, F(1,9) = 25.56, p < 0.001 for vehicle–1 mg/kg, and F(1,9) = 14.09, p < 0.01 for vehicle–3 mg/kg, simple contrasts). When cumulating locomotor activity levels after injection and after novel cage stress (Fig. 3b, inset), TP003 reduced locomotor activity levels after novel cage stress more than after injection (TP003 effect: F(3,27) = 9.34, p < 0.001; TP003 × pre–post interaction: F(1,9) = 9.81, p < 0.001), Although activity levels were comparable after both injection and novel cage (stress effect: F(1,9) = 1.63, p = 0.23, NS).

Alcohol

Summary

Alcohol reduced the SIH response and basal body temperature only at higher doses. Alcohol did not affect locomotor activity levels after injection and after novel cage stress (Fig. 3c, d; n = 11).

Body temperature

Novel cage stress led to a significant increase in temperature (time effect: F(12,120) = 15.68, p < 0.001) and alcohol reduced the SIH response (time × alcohol interaction: F(36,360) = 2.56, p < 0.001). Alcohol decreased basal body temperature (main alcohol effect: F(3,30) = 4.82, p < 0.01). Simple contrasts revealed differences only between vehicle and 3 g/kg regarding basal body temperature (dose contrasts: vehicle–0.3 g/kg: F(1,10) = 0.79, NS; vehicle–1 g/kg: F(1,10) = 0.01, p = 0.98, NS; vehicle–3 g/kg: F(1,10) = 5.71, p < 0.05). The calculated SIH response showed a trend for alcohol to reduce the SIH response (F(3,30) = 2.30, p = 0.09, NS). Simple contrasts revealed SIH attenuation at higher doses (1 g/kg–vehicle: F(1,10) = 1.32, p = 0.28, NS; 2 g/kg–vehicle: F(1,10) = 5.19, p < 0.05; 4 g/kg–vehicle: F(1,10) = 5.12, p < 0.05) (Fig. 3c, inset).

Locomotor activity

The locomotor reaction in response to novel cage stress (time effect: F(12,120) = 16.87, p < 0.001) was not affected by alcohol (main alcohol effect: F(3,30) = 0.62, p = 0.61, NS; alcohol × time interaction: F(36,360) = 1.23, p = 0.18, NS). When cumulating locomotor activity levels after injection and after novel cage stress (Fig. 3d, inset), alcohol did not affect locomotor activity levels after injection and after novel cage stress (alcohol effect: F(3,30) = 1.67, p = 0.20, NS; alcohol × stress: F(3,30) = 0.90, p = 0.45, NS).

THIP

Summary

THIP reduced the SIH response and basal body temperature at its highest dose. THIP reduced locomotor activity after novel cage stress (Fig. 3e, f; n = 10).

Body temperature

Basal body temperature was overall reduced by THIP (THIP effect: F(3,27) = 5.44, p < 0.01). Also, the SIH response (time effect: F(12,108) = 12.59, p < 0.001) was reduced by THIP (THIP × time interaction: F(36,324) = 4.25, p < 0.001). Simple contrasts revealed a significant difference in basal body temperature between the vehicle and 10 mg/kg condition (F(1,9) = 15.40, p < 0.01), whereas the other doses did not affect basal body temperature (vehicle–0.3 mg/kg condition F(1,9) = 1.11, p = 0.32, NS; vehicle–3 mg/kg condition F(1,9) = 0.11, p = 0.75, NS). The calculated SIH response revealed that THIP reduced the SIH response (F(3,27) = 4.64, p = 0.01). Simple contrasts revealed that only the highest dose reduced the SIH response (0.3 mg/kg–vehicle: F(1,9) = 0.31, p = 0.59, NS; 3 mg/kg–vehicle: F(1,9) = 0.25, p = 0.63; 10 mg/kg–vehicle: F(1,10) = 6.15, p < 0.05) (Fig. 3e, inset).

Locomotor activity

The stress-induced locomotor response (time effect: F(12,108) = 12.35, p < 0.001) was reduced by THIP (THIP × time interaction: F(36,324) = 1.96, p < 0.001; THIP effect: F(3,27) = 2.43, p = 0.09, trend). When cumulating locomotor activity levels after injection and after novel cage stress (Fig. 3f, inset), THIP reduced locomotor activity levels only after the novel cage procedure (THIP × stress interaction: F(1,9) = 3.37, p < 0.05), although overall locomotor activity was not reduced (THIP effect: F(3,27) = 1.18, p = 0.34, NS) and overall locomotor activity levels after injection and novel cage stress were not different (stress effect: F(1,9) = 0.18, p = 0.68, NS).

Discussion

In the present study, we examined the effects of various GABAAergic compounds on temperature and locomotor responses to acute stress. The SIH model uses the rise in body temperature in response to stress to assess anxiolytic drug effects and provides a translational approach to anxiety research (Vinkers et al. 2008). We found that the administration of the nonselective GABAA receptor agonist diazepam resulted in a dose-dependent attenuation of the SIH and basal and stress-induced locomotor activity responses, indicating that diazepam induces both anxiolytic and sedative effects. These findings support and extend previous studies on diazepam in the SIH paradigm in mice (Olivier et al. 2002). Zolpidem, an intermediate selective GABAA receptor α1 subunit agonist, decreased basal core body temperature and attenuated basal and stress-induced locomotor and temperature responses in a dose-dependent fashion. The sedative effect of zolpidem in vivo is mediated by the α1 subunit (Crestani et al. 2000), and zolpidem does not possess any anxiolytic properties (Kralic et al. 2002; Mathiasen et al. 2007). However, we cannot exclude that the results of the higher doses of zolpidem may be the result of nonspecific GABAA receptor activation. Also, the reduction of the SIH response by zolpidem is most likely the result of strong hypothermic effects on basal body temperature, disturbing physiological homeostatic mechanisms (Olivier et al. 2003).

βCCt shows a high affinity for the GABAA receptor α1 subunit with considerably lower affinity for GABAA receptor α2, α3, and α4 subunits and has comparable low efficacy at all α subunits. (Basile et al. 2006; Huang et al. 2000). Administration of βCCt alone had no effect on either basal body temperature or novel cage-induced temperature and locomotor activity responses. However, prior injection with βCCt antagonized hypothermic effects of both diazepam and zolpidem and reversed zolpidem-induced locomotor sedation (Figs. 1 and 2). In contrast, βCCt did not antagonize the diazepam-induced locomotor sedation, which may be attributed to the fact that diazepam exerts a broader pharmacological GABAA agonistic profile while zolpidem action is restricted to the GABAA receptor α1 subunit. Basal body temperature reduction after diazepam administration was only observed in the combination βCCt/diazepam experiment and not when solely diazepam was injected. We do not have an explanation for this puzzling observation; the only experimental difference was the presence of a double injection within 10 min. Mice do consistently show a reduction in basal body temperature after administration of diazepam (Olivier et al. 2002; Van Bogaert et al. 2006), and better regulated homeostasis in the rat may account for an absent hypothermia when diazepam was administered. Our results suggest a role for the GABAA receptor α1 subunit in hypothermic and locomotor sedative actions of GABAAergic drugs. Some studies have suggested that anxiolytic effects of benzodiazepines can be reversed with βCCt (Belzung et al. 2000; Griebel et al. 1999). However, this may be the result of decreased sedation rather than a reversal of anxiolytic effects, since sedation caused by diazepam and zolpidem has been shown to be reversible with high doses of βCCt (Basile et al. 2006; Griebel et al. 1999). The hypothermic effects after activation of the GABAA receptor α1 subunit has been extensively studied in mice (Van Bogaert et al. 2006). The α1 subunit is abundantly expressed throughout the brain, and a higher expression of the α1 subunit is present in the hypothalamic preoptic area and dorsomedial hypothalamus compared to the α2 and α3 subunits (Pirker et al. 2000). These areas are thought to play a major role in thermoregulation (Boulant 2000; Dimicco and Zaretsky 2007; Nagashima et al. 2000) and may account for the α1 involvement in the regulation of basal body temperature.

A putative role for the GABAA receptor α3 subunit in anxiety was confirmed with GABAA receptor α3 subunit agonist TP003 that attenuated the SIH response without affecting basal body temperature levels (Fig. 3a). Also, when combined with GABAA receptor α1 subunit antagonist βCCt, diazepam still reduced the SIH response, putatively through activation of the α2/3 subunit. Although transgenic mice lacking benzodiazepine sensitivity in the α3 subunit did not show altered anxiolytic actions of diazepam (Low et al. 2000; Rudolph and Mohler 2004), pharmacological studies have pointed to a role for this subunit in anxiolysis (Atack et al. 2005; Atack et al. 2006; Dias et al. 2005) as might be expected from high α3 subunit expression in brain areas involved in acute stress responses (Pirker et al. 2000). Anxiolytic effects of TP003 were found in the elevated plus maze (rats) and in a conditioned emotional response test (squirrel monkeys) (Dias et al. 2005). Our findings support and extend previous experiments suggesting GABAA receptor α2 and α3 subunits as the main regulatory subunits mediating anxiolytic effects (Atack et al. 2005; Dias et al. 2005). Stress-induced locomotor responses after novel cage stress were reduced at all TP003 doses, but only after novel cage stress and not immediately after injection like in the case of diazepam and zolpidem (Fig. 3b). This contrasts with another study in mice that did not show any sedative locomotor effects of TP003 (Dias et al. 2005). Although being α3 subunit-selective, TP003 also has low modulation via α1-, α2-, and α5-containing subtypes (Dias et al. 2005). Also, differences in metabolizing enzymes exist between animal species, resulting in different clearance rates which are frequently thought to be responsible for differences in behavioral responses. For GABAA receptor agonists, one study of nitrazepam found much higher plasma levels in rats than in mice after a dose of nitrazepam (Takeno et al. 1993), whereas another study found that the oral bioavailability of L-838417, a α1 subunit antagonist and α2/α3 partial subunit agonist, in mice was very poor compared to bioavailability in rats (Scott-Stevens et al. 2005). Therefore, a lack of sedative action of TP003 in mice compared to rats can possibly be ascribed to lower plasma levels of TP003 caused by a more rapid metabolism of TP003 in mice.

Locomotor activity responses to stress are used as an output parameter in various anxiety paradigms such as the elevated plus maze, the open field test, and the light/dark test. Open-arm entries, a lit box, or center of a field all putatively lead to an anxiety state, but also cage exchange as used in the current experiments leads to similar increases in distances traveled and velocities (de Visser et al. 2006). In general, stress-induced behavior in rodents consists of exploration on one hand and anxiety-driven avoidance behavior on the other hand, and there is no easy way of establishing the relationship between exploration and anxiety. Anxiolytic drugs increase explorative behavior and locomotor activity (Belzung and Berton 1997), but in higher doses cause general locomotor sedation, interfering with a good test interpretation (Dawson et al. 1995). Therefore, sedative effects of both diazepam and zolpidem cause a decrease in locomotor activity (Davies et al. 1994; Elliot and White 2001). However, the sedative effects of benzodiazepines in the elevated plus maze are no longer present after a point mutation of the α1 subunit (McKernan et al. 2000; Rudolph et al. 1999), indicating that the α1 subunit is closely involved in benzodiazepine-induced locomotor activity reduction. McKernan et al. showed that diazepam (3 mg/kg) even increased locomotor activity in α1 subunit point-mutated mice compared to wild-type controls. In contrast, myorelaxant effects of diazepam in the rotarod assay remain present in the α1 subunit KO mice, suggesting that the locomotor activity attenuation is not the mere result of muscle relaxation. Other studies showed that βCCt antagonized the locomotor depressant effects of zolpidem and diazepam on open field locomotor activity in mice (Griebel et al. 1999) as well as the elevated plus maze (Savic et al. 2004). All in all, there is ample evidence that locomotor depressant actions of zolpidem and the benzodiazepines are mediated via the α1 subunit of the GABAA receptor. Anxiolytic drugs completely devoid of sedative side effects would, therefore, either increase or not affect locomotor activity parameters after novelty-induced stress.

Alcohol reduced basal body temperature at higher doses without affecting stress-induced locomotor responses (Fig. 3c). Only the highest dose reduced the SIH response, an effect that was already earlier observed in mice (Olivier et al. 2003). Although acute administration of alcohol is known to possess an anxiolytic profile, the effects are known to be different from benzodiazepines (Langen et al. 2002) as alcohol binds to extrasynaptic GABAA receptors containing α4 or α6 and δ subunits (Wallner et al. 2003). However, we used higher doses that could have lost extrasynaptic binding selectivity. Also, alcohol at higher doses may act on NMDA, serotonin, and glycine receptors (Crews et al. 1996; Davies 2003; Harris 1999). THIP reduced basal body temperature and SIH and locomotor activity responses only at the highest dose tested (10 mg/kg), whereas lower doses did not have any effect (Fig. 3e). THIP has been shown to enhance sleep episodes (Lancel and Langebartels 2000) with little affinity for benzodiazepine receptors. Rather, THIP binds to extrasynaptic GABAA receptors containing a δ subunit (Wafford and Ebert 2006), and a role for the GABAA receptor δ subunit in neurosteroid-mediated anxiolytic effects has been proposed (Mihalek et al. 1999). However, it seems more likely that the strong hypothermic effects of the highest dose of THIP are due to interference with physiological thermoregulation (Olivier et al. 2003). Interaction between effects on sleep and thermoregulation are possible because of common neural pathways within the preoptic area and anterior hypothalamus (Frosini et al. 2004). Indeed, THIP synchronized hypothermic and EEG effects in rabbits (Frosini et al. 2004). Interestingly, alcohol did not affect locomotor activity at all doses and THIP affected locomotor activity only at high doses (Fig. 3d, f), whereas the other synaptic compounds all reduced locomotor activity to some extent. Other studies have found that alcohol impaired rotarod performance at lower doses (Zaleski et al. 2001). Although our high doses may have lost extrasynaptic selectivity, this indicates that locomotor activity may be differentially controlled by extrasynaptic and synaptic receptor populations. This is supported by a lack of cross-tolerance in the rotarod test between zolpidem and THIP (Voss et al. 2003).

The SIH amplitude decreased over the course of the experiments from 0.7°C at the start of the experiments to 0.2–0.3°C in the final experiments, as did locomotor activity levels after novel cage stress. Habituation to the experimental procedure may account for a decreased SIH response, although previous methodological testing has not revealed any habituation using a 1-week interval, even when testing occurred for over a year (Bouwknecht et al. 2007; Olivier et al. 2003; Van der Heyden et al. 1997). Also, the manually calculated SIH response from the time graphs is generally in complete agreement with the time graphs. Only when drugs are tested at doses that markedly decrease body temperature, there appears to be a small difference between the calculated SIH response and the time graphs. This difference is attributable to the fact that the calculated SIH response is based on the maximum temperature during the first 30 min after stress. In those cases in which body temperature is decreasing after stress induction, the maximum is likely to be close to the start of that 30-min period. In this way, the calculated SIH response in these cases is likely to yield a result close to 0°C, whereas a decreasing basal body temperature seems to indicate a negative SIH response. The differences, however, are small and do not change the interpretation of our data.

The most important finding in the present study is that the GABAA receptor modulates temperature and locomotor stress responses as well as basal body temperature processes through different GABAA receptor subunits. More specifically, the GABAA α1 receptor subunit was found to be essential for basal body temperature regulation and for inducing locomotor sedation, whereas the GABAA receptor α2 and α3 subunit exerted anxiolytic effects by attenuating the SIH response. Nonbenzodiazepine GABAA activity is less involved in thermoregulation and locomotor sedation, as suggested by the effects of alcohol and THIP. In conclusion, we show that the use of home cage temperature and locomotor stress responses provides a successful approach to anxiety research and possesses an enormous potential to pharmacologically study the effects of GABAAergic drugs. The SIH model uses a simultaneously collected independent parameter and may possess additional value over locomotor activity parameters only.