Benzodiazepine-Sensitive GABAA Receptors Limit the Activity of the NMDA/NO/Cyclic GMP Pathway

A Microdialysis Study in the Cerebellum of Freely Moving Rats


  • Abbreviations used : cGMP, cyclic GMP ; NO, nitric oxide ; NOS, nitric oxide synthase.

Address correspondence and reprint requests to Dr. M. Raiteri at Sezione di Farmacologia e Tossicologia, Dipartimento di Medicina Sperimentale, Università di Genova, Viale Cembrano 4, 16148 Genova, Italy. E-mail :


In the cerebellum, infusion of NMDA (200 μM) for 20 min evoked a marked (200%) increase of extracellular cyclic GMP (cGMP) levels. The selective GABAA receptor agonist muscimol (0.01-100 μM) was able to counteract the NMDA effect with an EC50 of 0.65 μM ; the inhibitory effect of muscimol (10 μM) was prevented by bicuculline (50 μM). Diazepam (10 μM) significantly potentiated the muscimol (1 μM) inhibition ; furthermore, when coinfused with 0.1 μM muscimol (a concentration not affecting, on its own, the cGMP response to NMDA), diazepam (10 μM) reduced the NMDA effect. Similar results were obtained with zolpidem (0.1-1 μM). Finally, local infusion of the benzodiazepine site antagonist flumazenil (10 μM), together with muscimol and diazepam, almost completely restored the effect of NMDA on extracellular cGMP levels. It is concluded that GABAA receptors potently control the NMDA/nitric oxide/cGMP pathway in the cerebellum in vivo. In terms of the α subunit composition, we can deduce that the cerebellar GABAA receptor does not contain α6 or β4 subunits because it is diazepam-sensitive. Moreover, the observation that zolpidem is active at a rather low concentration, in combination with localization studies present in the literature, tend to exclude the presence of α5 subunits in the receptor composition and suggest the involvement of an α1 subunit.

Abundant studies have indicated that abnormal excitatory neurotransmission, especially through NMDA receptors, underlies neurodegenerative processes that characterize several acute and chronic pathologies. In particular, great interest has emerged with regard to nitric oxide (NO) as it has been shown that, among other harmful chemical species, this gaseous molecule is responsible for NMDA receptor-mediated cell injury (Dawson et al., 1993, 1996 ; Strijbos et al., 1996 ; Gonzalez-Zulueta et al., 1998 ; Urushitani et al., 1998). Following stimulation of NMDA receptors, NO is formed by the activation of NO synthase (NOS), which converts l-arginine into l-citrulline. High concentrations of NO can damage cells by different, but probably concurrent, mechanisms, including reaction with superoxide radicals to form the cytotoxic peroxynitrite ion ; apart from other deleterious effects, recent results have pointed out that NO or peroxynitrite ion can damage DNA strands, thus activating poly(ADP-ribose) synthase, which fatally depletes the ATP energy stores of the cell (for review, see Dawson and Dawson, 1996).

GABA is the major inhibitory neurotransmitter in various regions of the brain where GABAergic circuits play a key role in balancing excitatory neurotransmission, principally mediated by glutamatergic neuronal networks. Under some pathological conditions, i.e., cerebral ischemia, which set in motion glutamate-dependent neurodegenerative processes, increased neuronal excitation seems to be accompanied by reduction of synaptic inhibition (Kawai et al., 1992) and by a decline of GABAA subunit mRNA and GABAA receptor-mediated responses (Li et al., 1993 ; Verheul et al., 1993). In fact, several investigations have found that increased GABAergic synaptic inhibition might result in neuroprotective effects : Administration of drugs acting at GABAA receptors has been shown to prevent neuronal loss in different brain regions of rodents following experimental cerebral ischemia, even when given up to 1 h after the ischemic insult (Sternau et al., 1989 ; Johansen and Diemer, 1991 ; Shuaib et al., 1993 ; Schwartz et al., 1994, 1995).

During the last decade, we have developed an experimental model to study in vivo, by means of the intracerebral microdialysis technique, the production of NO and its modulation. In fact, we have extensively shown that cyclic GMP (cGMP), a close although indirect marker of NO production, can be quantified by microdialysis in the brain of freely moving rats and that its extracellular levels respond, in a NOS-dependent manner, to pharmacological manipulations known to influence glutamatergic neurotransmission (for review, see Fedele and Raiteri, 1999). Subsequently, our model has been validated by other research groups (Laitinen et al., 1994 ; Luo and Vincent, 1994 ; Globus et al., 1995 ; Kendrick et al., 1997 ; Consolo et al., 1998).

In the present microdialysis study, we have examined whether, and to what extent, stimulation of GABAA receptors could limit the NMDA-evoked production of NO/cGMP in the attempt to gain new information that might provide the neurochemical basis for alternative neuroprotective strategies with GABAergic drugs already present in clinical therapeutic regiments, i.e., benzodiazepines.



Drugs were purchased from the companies indicated : NMDA (Tocris Cookson, Bristol, U.K.) and muscimol, bicuculline hydrobromide, diazepam, zolpidem, and flumazenil (Sigma Chemical Co., St. Louis, MO, U.S.A.).

Diazepam, zolpidem, and flumazenil (10 mM) were dissolved in dimethyl sulfoxide and diluted down to the desired concentration in modified Ringer's medium ; at the highest concentration used (0.2%) dimethyl sulfoxide did not modify cGMP basal levels (data not shown). The other drugs were dissolved in modified Ringer's medium. All the final solutions were buffered at pH 7.4.

The experimental procedures were approved by the Ethical Committee of the Pharmacology and Toxicology Section, Department of Experimental Medicine, in accordance with the European legislation (European Communities Council Directive of 24 November 1986, 86/609/EEC).

Neurosurgery and dialysis procedure

Male Sprague-Dawley rats (weighing 250-300 g ; CD-COBS ; Charles River, Calco, Italy) were anesthetized with Equithesin (3 ml/kg), placed in a stereotaxic frame (David Kopf Instruments, West Hempstead, NY, U.S.A.), and implanted with a microdialysis probe that was transversely positioned into the cerebellum according to the following coordinates : AP = -2.3, H = +6.0 from the interaural line (Paxinos and Watson, 1986). A piece of dialysis fiber made of a copolymer of acrylonitrile sodium methallyl sulfonate (AN69HF ; Hospal S.p.A., Bologna, Italy ; 0.3 mm outer diameter ; cutoff, 40 kDa) was covered with epoxy glue to confine dialysis to the area of interest (8-mm glue-free zone). The skull was exposed, and two holes were drilled on the lateral surface at the level of the cerebellum. The dialysis probe, held straight by a tungsten wire inside, was inserted transversely into the brain so that the glue-free zone was exactly located into the target area. The tungsten wire was withdrawn, and stainless steel cannulae (22-gauge diameter, 15-mm long) were glued to the ends of the fiber, bent up, and fixed vertically to the skull with dental cement and modified Eppendorf tips. After a 24-h recovery period, rats were placed into observation cages, and the probes were perfused at a flow rate of 5 μl/min (CMA/100 microinjection pump ; CMA Microdialysis, Stockholm, Sweden) with modified Ringer's medium containing 145 mM NaCl, 3 mM KCl, and 1.26 mM CaCl2, buffered at pH 7.4 with 2 mM phosphate buffer. Consecutive samples were collected every 20 min following a washout period of 1 h. At the end of the experiment, rats were killed, and the correct position of the probe was verified by histological examination of the fiber tract.

The cGMP content in the dialysates was assayed by a commercially available radioimmunoassay kit (Amersham Dual Range ; Amersham Radiochemical Centre, Little Chalfont, Bucks, U.K.) using the acetylation protocol. Aliquots of the samples (10 μl) were assayed in triplicate. Under these experimental conditions, the sensitivity of the assay is 2 fmol/100 μl (standard curve range, 2-128 fmol/100 μl). The percent in vitro recovery for cGMP was 33 ± 5.1% (n = 3).

Statistics and expression of results

Data are expressed as mean ± SEM percentages of the mean basal value, which was determined by averaging the content of the first two or three samples collected before drug treatments. Differences were analyzed by two-way ANOVA with repeated measures over time followed by Mann-Whitney U test. Differences were considered significant at the level of p≤0.05. In the lower panels of Figs. 1 and 3, data represent the net peak effects observed in the different experimental groups. The EC50 for muscimol has been obtained by fitting the data of the concentration-response curve with the following four-parameter logistic equation for nonlinear regression (SigmaPlot version 3.0 software) : Y = (a - d)/[1 + (x/c)b] + d, where a represents the asymptotic maximum, b is the curve slope, c is the value at the inflexion point, and d is the asymptotic minimum.

Figure 1.

GABAA receptor activation inhibits NMDA-induced production of extracellular cGMP in cerebellum of freely moving rats. Rats were transversely implanted in the cerebellum with a dialysis probe that was infused with modified Ringer's medium at a flow rate of 5 μl/min. Following a stabilization period of 1 h, consecutive samples were collected every 20 min until the end of the experiment and assayed for their cGMP content. Upper panel : Time course of the experiments. NMDA was infused for 20 min after the third fraction had been collected (horizontal solid bar) ; when present, muscimol was infused one fraction before NMDA administration and for the time indicated by the horizontal open bar : 200 μM NMDA (•) ; 200 μM NMDA + 0.1 μM muscimol (○) ; 200 μM NMDA + 1 μM muscimol (▴) ; 200 μM NMDA + 3 μM muscimol (♦) ; 200 μM NMDA + 10 μM muscimol (▪) ; and 200 μM NMDA + 100 μM muscimol (□). Lower panel : Muscimol concentration-response curve obtained by plotting the net peak effects of the different experimental groups. Data are mean ± SEM (bars) values of five to nine different experiments. *p≤ 0.05 versus corresponding cGMP basal levels ; #p≤ 0.05 versus 200 μM NMDA. For further technical details, see Materials and Methods.

Figure 3.

Benzodiazepine site agonists enhance the inhibitory effects of muscimol on NMDA-induced production of extracellular cGMP in cerebellum of freely moving rats and reversal by flumazenil. Upper panel : Time course of effects of diazepam or zolpidem on muscimol-induced inhibition of cGMP production evoked by NMDA and the antagonism by flumazenil. NMDA was infused for 20 min (horizontal solid bar), whereas all the other drugs were infused for the time indicated by the horizontal open bar : 200 μM NMDA (○) ; 200 μM NMDA + 0.1 μM muscimol (•) ; 200 μM NMDA + 1 μM muscimol (□) ; 200 μM NMDA + 0.1 μM muscimol + 10 μM diazepam (▴) ; 200 μM NMDA + 1 μM muscimol + 10 μM diazepam (▪) ; 200 μM NMDA + 0.1 μM muscimol + 10 μM diazepam + 10 μM flumazenil (▾) ; 200 μM NMDA + 0.1 μM muscimol + 0.1 μM zolpidem (⋄) ; 200 μM NMDA + 0.1 μM muscimol + 0.3 μM zolpidem (▵) ; and 200 μM NMDA + 0.1 μM muscimol + 10 μM zolpidem (♦). For the sake of clarity, statistical symbols have been omitted. Lower panel : Summary of the data shown in the upper panel. Columns represent the net peak effects observed in the different experimental groups. *p≤ 0.05 versus corresponding cGMP basal levels ; #p≤ 0.05 versus 200 μM NMDA ; §p≤ 0.05 versus 200 μM NMDA + 1 μM muscimol ; °p≤ 0.05 versus 200 μM NMDA + 0.1 μM muscimol + 10 μM diazepam.

FIG. 1.

FIG. 3.


Muscimol limits and prevents the ability of NMDA receptors to enhance extracellular cGMP production in cerebellum of freely moving rats

In keeping with previous results (Vallebuona and Raiteri, 1993 ; Fedele et al., 1997), extracellular basal levels of cGMP in the cerebellum of awake rats amounted to 320.17 ± 11.13 fmol/100 μ1 (mean ± SEM, n = 60).

As shown in Fig. 1 (upper panel) and as already reported (see above), administration of NMDA (200 μM) into the cerebellum by retrodialysis for 20 min enhanced the extracellular levels of the cyclic nucleotide, which peaked in the fraction of the drug infusion and returned to basal values within 40 min following removal of the glutamate receptor agonist from the infusion fluid. Coinfusion of muscimol, a selective GABAA receptor agonist, concentration-dependently (1-100 μM) decreased the cGMP response to NMDA, which was completely abolished at 10 μM. Lower muscimol concentrations (0.01 or 0.1 μM) did not change significantly the NMDA-induced effect. The EC50 for muscimol, calculated by fitting the experimental data with a nonlinear regression four-parameter logistic equation, was 0.65 μM (Fig. 1, lower panel). The selective GABAA receptor antagonist bicuculline (50 μM) was able to prevent the effect of muscimol (10 μM) on the NMDA-evoked increase of extracellular cGMP (Fig. 2). Infusion of bicuculline alone did not affect cGMP basal levels (data not shown).

Figure 2.

Bicuculline prevents the muscimol-induced decrease of extracellular cGMP production caused by NMDA local infusion into cerebellum of freely moving rats. NMDA was infused into the cerebellum for 20 min (horizontal solid bar), whereas muscimol, alone or in the presence of bicuculline, was added to the infusion stream for the time indicated by the horizontal open bar : 200 μM NMDA (•) ; 200 μM NMDA + 10 μM muscimol (▪) ; 200 μM NMDA + 10 μM muscimol + 50 μM bicuculline (○). Data are mean ± SEM (bars) values of four to nine different experiments. *p≤ 0.05 versus corresponding cGMP basal levels ; #p≤ 0.05 versus NMDA.

FIG. 2.

The cerebellar GABAA receptor inhibiting the NMDA-induced cGMP production is benzodiazepine-sensitive

As shown in Fig. 3, infusion of benzodiazepines potentiated the effects of muscimol on the increase of cGMP production evoked by a 20-min NMDA (200 μM) pulse. It can be seen that when muscimol was coinfused at 0.1 μM (a concentration ineffective alone on the NMDA effect) with NMDA (200 μM) in the presence of diazepam (10 μM), the cGMP response to NMDA was markedly reduced (by ~70%). Similar results have been obtained with zolpidem (0.1-1 μM). Moreover, diazepam (10 μM) potentiated the effects of 1 μM muscimol on the NMDA-mediated increase of extracellular cGMP (from 65 to 98% inhibition of the NMDA-evoked response).

When the effect of diazepam (10 μM coinfused with 0.1 μM muscimol) was challenged with flumazenil (10 μM), a selective antagonist at the GABAA benzodiazepine site, the cGMP response to NMDA (200 μM) was completely restored (Fig. 3). Diazepam, zolpidem, or flumazenil alone did not affect either the basal or the NMDA-increased production of cGMP (data not shown).


In the present microdialysis study, we have found that the increase of cGMP extracellular levels by local infusion of NMDA into the cerebellum of freely moving rats can be potently inhibited by activation of GABAA receptors. In fact, muscimol was able to diminish and abolish completely, in a bicuculline-sensitive manner, the NMDA-evoked cGMP response with an EC50 of ~0.7 μM ; furthermore, agonists at the benzodiazepine site, like diazepam or zolpidem, were able to potentiate the activity of this GABAA receptor even when muscimol concentrations in the infusion stream were so low (0.1 μM) that no significant variations of the NMDA effect could be observed. Flumazenil, a selective antagonist at the benzodiazepine allosteric site, reversed the effect of diazepam. Finally, it would seem that GABAA receptors are not tonically stimulated by endogenous GABA released during the spontaneous or stimulated synaptic activity because bicuculline did not change, on its own, either basal or NMDA-increased cGMP levels. The observation that diazepam or zolpidem alone did not affect cGMP dialysate values (in the absence or presence of NMDA) supports this reasoning.

The sensitivity to benzodiazepines of the GABAA receptor here investigated can also lead us to some considerations regarding its molecular subunit arrangement. It is now well established that GABAA receptors are pentameric structures formed by a combination of various subunits (α, β, γ, and δ) and subunit isoforms differently localized across the regions of the brain (Barnard et al., 1998). Expression studies have found that receptors containing only α4 or α6 subunits (in combination with β and γ) are unable to bind classical benzodiazepines like diazepam (Wieland et al., 1992 ; Mohler et al., 1995). Thus, in our case, the GABAA receptor should contain α1, α2, α3, or α5 subunits (alone or in combination) ; however, the α5 subunit seems absent in the cerebellum and possesses a very low affinity for zolpidem (Wieland et al., 1992 ; Wisden et al., 1992 ; Lüddens et al., 1995), and the α2 and α3 subunits are very weakly expressed in this brain region (Wisden, 1995). Although full pharmacological characterization of receptor subunits is extremely difficult in in vivo microdialysis studies, the above-mentioned reasoning together with the data obtained with rather low concentrations of zolpidem (0.3-1 μM in the infusion stream) might suggest that the benzodiazepine-sensitive GABAA receptor inhibiting the NMDA-induced production of cGMP contains the α1 subunit.

At this point, we will try to organize our neurochemical data into a feasible picture for a better understanding of the anatomical/functional relationships between the cerebellar NMDA/NO/cGMP pathway and the GABAA-driven inhibitory system. As mentioned in the introductory section, we have already shown that cGMP elevations in response to NMDA, assessed by intracerebellar microdialysis, are completely dependent on NOS activation and NO production (Fedele and Raiteri, 1999). Purkinje neurons possess α1-containing GABAA receptors ; however, in the adult rat cerebellum, they do not express functional NMDA receptors (Cull-Candy et al., 1998), and, more importantly, they do not express NOS. These neurons, however, seem to be the principal target of NO as they contain soluble guanylyl cyclase, cGMP-dependent protein kinase, and the G-substrate (Schlichter et al., 1980 ; Ariano et al., 1982 ; De Camilli et al., 1984 ; Matsuoka et al., 1992). In contrast, granule cells possess NMDA receptors (Garthwaite and Brodbelt, 1989 ; Seeburg, 1993), contain high levels of NOS (Bredt et al., 1990 ; Vincent and Kimura, 1992), and present GABAA receptors containing the α1 subunit (Wisden et al., 1996). Therefore, it is possible that NMDA evokes NO production in granule cells ; the gas then diffuses into Purkinje neurons, where it increases cGMP production. Benzodiazepine-sensitive GABAA receptors would modulate the excitability of granule cells, thereby decreasing the effects of NMDA receptor stimulation on NOS activity. Alternatively, NO might be formed in basket/stellate cells, which possess NMDA receptors, NOS, and α1-containing GABAA receptors.

Several groups have found that administration of diazepam (or drugs enhancing GABA neurotransmission) before, during, or after experimental ischemia in rodents significantly protects neurons in different brain regions (Sternau et al., 1989 ; Johansen and Diemer, 1991 ; Shuaib et al., 1993 ; Schwartz et al., 1994, 1995). It might be argued that neuroprotection achieved by diazepam is not due to a specific mechanism but is the consequence of a great degree of hypothermia induced by systemic administration of the drug. However, a recent study by Schwartz et al. (1995) has demonstrated that diazepam shows a high degree of neuroprotection in experimental ischemia also when injected directly into the brain tissue, a procedure that did not cause hypothermia. Therefore, because pharmacological manipulations aimed at reducing NO production in cerebral ischemia have been proved to achieve neuroprotection, at a speculative level our data might provide one of the possible mechanisms by which benzodiazepines can prevent cell death in such an acute CNS pathology.