Hyperalgesic effect induced by barbiturates , midazolam and ethanol : pharmacological evidence for GABAA receptor involvement

The involvement of GABA-A receptors in the control of nociception was studied using the tail-flick test in rats. Non-hypnotic doses of the barbiturates phenobarbital (5-50 mg/kg), pentobarbital (17-33 mg/ kg), and thiopental (7.5-30 mg/kg), of the benzodiazepine midazolam (10 mg/kg) or of ethanol (0.4-1.6 g/kg) administered by the systemic route reduced the latency for the tail-flick response, thus inducing a ‘hyperalgesic’ state in the animals. In contrast, non-convulsant doses of the GABA-A antagonist picrotoxin (0.12-1.0 mg/kg) administered systemically induced an increase in the latency for the tail-flick response, therefore characterizing an ‘antinociceptive’ state. Previous picrotoxin (0.12 mg/kg) treatment abolished the hyperalgesic state induced by effective doses of the barbiturates, midazolam or ethanol. Since phenobarbital, midazolam and ethanol reproduced the described hyperalgesic effect of GABA-A-specific agonists (muscimol, THIP), which is specifically antagonized by the GABA-A antagonist picrotoxin, our results suggest that GABA-A receptors are tonically involved in the modulation of nociception in the rat central nervous system. Correspondence


Introduction
Barbiturates, benzodiazepines and ethanol, although belonging to chemically distinct classes of drugs, share some pharmacological properties.The principal medical indication of the barbiturate phenobarbital is the prevention of grand mal epilepsy (1), whereas the benzodiazepine midazolam is used as a pre-anesthetic medication (2).Etha-nol is more a sociological than a medical problem due to its risk of abuse (3).However, all three drugs induce a general depressant effect in the central nervous system (CNS).
Barbiturates, midazolam and ethanol have been shown to increase the inhibitory γ-aminobutyric acid (GABA) transmission which leads to the CNS depression mentioned above (4)(5)(6).At the molecular level, barbiturates and benzodiazepines potentiate GABA actions by a) prolonging the open state of the chloride-associated channels and by b) increasing the frequency of opening of the chloride channels, respectively.These effects can be competitively antagonized by bicuculline or non-competitively by picrotoxin (4,5).
Many reports in the literature have shown the involvement of GABA receptors in the modulation of nociception in the central nervous system (4).Administration of GABA-A agonists seems to interfere with nociception, whereas the GABA-B agonist baclofen can induce analgesia at spinal or supraspinal levels (4).Recent reports in the literature have suggested the involvement of the GABA-A receptor in the modulation of central nociception through the descending inhibitory system (7).
The aim of the present study was to determine the effect of systemic administration of the barbiturates phenobarbital, pentobarbital and thiopental, the benzodiazepine midazolam, and ethanol using the tail-flick model in rats.In addition, picrotoxin alone or in association with these agonists was also used to assess the potential role of the GABA-A receptor in the tail-flick response.

Animals
Male Holtzman rats weighing 180-250 g were used.The animals were housed at a temperature of 24 ± 2 o C, with food and water ad libitum and a 12-h light-dark cycle for 24 h before the experiments.

Measurement of pain threshold
A slight modification of the method described by D'Amour and Smith (20) was used to detect the tail-flick response.Briefly, a voltimeter and an amperimeter were connected to the system to permit variation of the resistance heating.Stimuli of 0.31 cal/ sec caused a 3.5-4.5-seclatency for the tailflick baseline in the rats.Test tail-flick responses in duplicate at 10, 20, 30, 40 and 60 min following drug administration are reported as tail-flick index (TFI).The TFI was the percent of the maximum response obtained before and after drug administration to the animals, as expressed in the following formula: TFI = Post-drug latency (sec) -baseline latency (sec) x 100 7 sec -baseline latency Under our conditions, the cut-off time was defined as 7 sec (in the formula).The mean area under the various TFI curves (AUC) ± SEM from control and drug-treated animals was also obtained.Positive and negative values indicated analgesia and hyperalgesia, respectively, for both TFI and AUC.

Drug treatments
The barbiturates and midazolam, diluted in propylene glycol:saline (50:50, v/v) and in manufacturer-supplied vehicle, respectively, were administered by the intraperitoneal (ip) route in a volume of 0.1 ml/kg.Ethanol was diluted in physiological saline and administered orally (0.2 ml/kg) rather than ip to prevent peritoneal irritation.Picrotoxin diluted in physiological saline was administered ip alone or 10 min before the injection of agonists.In these experiments, control animals received vehicle or saline (0.1 ml/100 g) either ip or orally.

Statistical analysis
The results are reported as mean ± SEM.Data for the time-effect curve were subjected to multivariate analysis of variance (MANOVA).In case of interaction between treatment and time, MANOVA was followed by one-way analysis of variance (ANOVA) for each time.For the analysis of the treatment effect only, MANOVA was followed by ANOVA for the area under the curve.For multiple comparisons, ANOVA was followed by the Duncan test.Statistical significance was accepted at P<0.05.

Results
Intraperitoneal administration of phenobarbital (5-50 mg/kg), pentobarbital (17-33 mg/kg) and thiopental (7.5-30 mg/kg) reduced the latency for the tail-flick response in rats in a dose-dependent manner, therefore inducing hyperalgesia, as shown in Fig- ure 1A.Doses higher than 50, 33 and 30 mg/ kg phenobarbital, pentobarbital and thiopental, respectively, induced hypnotic effects in the rats, thus precluding the use of these doses in our experiments (data not shown).Significant hyperalgesia was observed 10 min after barbiturate administration, with a maximal effect being detected at 30 min, and remained significantly elevated for the subsequent 30 min of observation, as illustrated for 20 mg/kg phenobarbital in Figure 2A.This effect was observed with the three bar-  Intraperitoneal picrotoxin injection (0.12-1.0 mg/kg) induced a dose-dependent increase in the latency for the tail-flick response in rats, which characterized an antinociceptive effect (Figure 3).However, at the lower dose of 0.12 mg/kg, picrotoxin was not able to induce an antinociceptive effect per se, although it could inhibit the hyperalgesic effect induced by phenobarbital, midazolam and ethanol (Figure 2A, B and C, respectively).

Discussion
The precise effect of the administration of barbiturates or benzodiazepines on the mechanisms of pain modulation in humans or rats is still controversial.Analgesia, hyperalgesia or no effect has been attributed to these drugs (13)(14)(15)(16)(17)(18)(19).
In the present study, we have shown that systemic administration of non-hypnotic doses of the barbiturates phenobarbital, pentobarbital and thiopental induced a dosedependent hyperalgesia, as measured by the reduction of the latency for the tail-flick response in rats.Interestingly, non-hypnotic doses of the benzodiazepine midazolam and of ethanol systemically administered also induced a dose-dependent hyperalgesia using the same method.
Barbiturates, benzodiazepines and ethanol are known to be agonists of the GABA-A receptor complex which are ubiquitously distributed throughout the CNS, including the descendent inhibitory system (7,21).A biturates in the dose range used (data not shown).However, barbiturate-induced hyperalgesia was significantly reduced 2 h after drug administration at the higher dose used (data not shown).In addition, midazolam (1.0-10 mg/kg) and ethanol (0.4-1.6 g/ kg) also induced a dose-dependent hyperalgesia (Figure 1B and C, respectively), which showed a similar profile to that observed for facilitating action of the inhibitory neurotransmitter γ-aminobutyric acid (GABA) on the GABA-A receptor by these drugs could ultimately favor the expression of an inhibitory effect on the descending inhibitory system, with a consequent hyperalgesic behavior.This hypothesis was further reinforced by the observed blockade of the hyperalgesia induced by the association of barbiturates, midazolam or ethanol with a low dose of the GABA-A antagonist picrotoxin and a clear-cut antinociceptive effect with higher doses of picrotoxin.If this is so, these results also indicate that there is a GABAergic tonus in the descendent inhibitory system.The use of bicuculline, the specific antagonist of GABA-A receptor, also blocked the hyperalgesia induced by these agonists using the tail-flick test (Yokoro CM and Tatsuo MAKF, unpublished observations).A general depressant effect induced by these drugs at higher doses might obscure this hyperalgesic effect, thus explaining the analgesia or absence of effect formerly related to these drugs in the reported studies (18).
In the present study, however, association of various doses of picrotoxin with midazolam only blocked the hyperalgesic effect induced by the latter, without displaying an analgesic response as observed with barbiturates and ethanol (data not shown).It has been described that the binding site of midazolam is distant from that at which barbiturates or ethanol interact with the GABA-A receptor (21,22).It is possible that the binding of barbiturates or ethanol could facilitate the close association of picrotoxin at the GABA-A receptor, which ultimately leads to the inhibition of the neurotransmitter action.In our experiments, this inhibitory effect was translated as an antinociceptive effect.Midazolam, in turn, interacting with another site, such as the α-subunit of the GABA-A receptor (21), did not display this effect.
Our data demonstrate the involvement of the GABA-A receptor in the modulation of pain pathways.It is suggested that the allosteric change of this receptor induced by barbiturates and ethanol is different from that induced by midazolam, since the association of the latter with picrotoxin did not display the antinociceptive effect observed in the association of the former drugs with picrotoxin.We suggest that this hyperalgesic effect should be taken into account in cases of barbiturate use or ethanol intoxication, since non-hypnotic doses of these drugs are used under such conditions.
kg phenobarbital (Figure2A).Different from phenobarbital, however, midazolam induced a hyperalgesia of shorter duration (50 min), whereas ethanol-induced hyperalgesia was detected at 10 min of drug administration and had a longer duration (2 h; data not shown).Administration of the various vehicles used did not induce a significant variation in the latency of the tail-flick response presented by the animals (control in Figures1, 2 and 3).

Figure 2 -
Figure 2 -Picrotoxin blocked the hyperalgesia induced by systemic phenobarbital (A), midazolam (B) or ethanol (C) administration to rats.Picrotoxin was administered ip 20 min before the tail-flick test.The agonists phenobarbital (20 mg/kg, ip), midazolam (10 mg/kg, ip) or ethanol (1.6 g/kg, oral) were administered 10 min before the test.Control animals were injected ip or orally treated with the respective vehicles.Each point indicates the mean ± SEM of the data from 8 rats.Hyperalgesia is indicated by the negative values on the ordinate; positive values indicate antinociception.*P<0.05 compared to control (Duncan test).TFI, Tail-flick index.

Figure 3 -
Figure 3 -Picrotoxin induced a dose-dependent antinociception in rats, as measured by the tailflick method.Picrotoxin (0.12-1.0 mg/kg) was administered by the intraperitoneal route 20 min before the test.*P<0.05 compared to control (one-way analysis of variance).AUC, Area under the tail-flick index curve.* *