INTRODUCTION

Considerable effort has been devoted to discovery of glutamate antagonists (1, 2) in recent years, due to increasing evidence linking glutamate excitotoxicity to various neurological disorders (3). Unfortunately, while known antagonists can provide neuroprotection, excessive action of these classical blocking agents can obtain undesirable side effects (1, 2). To minimize these undesirable side effects, modification of the redox modulatory sulfhydryl groups of the glutamate receptor has been suggested as a possibly superior therapeutic strategy (4). Unlike classical antagonists, that can give complete inhibition by interaction at the glutamate receptor (e.g. CGS 19755) or directly at receptor-linked, calcium ion channels (e.g. phencyclidine or MK-801) (2), inhibition via the redox modulatory sites are expected to give only partial inhibition of function and thereby limit unwanted side effects associated with excessive antagonism (4). At present S-nitrosylation of glutamate receptors by an NO⁺ donor (e.g. nitroglycerin) is the only mechanism for partially blocking receptor response in vivo that would achieve this effect by interaction with the redox modulatory sites (5, 6).

Disulfiram has been used in the treatment of alcoholism for almost 50 years (7, 8). It has recently been demonstrated that disulfiram exerts its anti-alcohol effect in vivo only after bioactivation to the active metabolite S-methyl-N,N-diethylthiolcarbamate sulfoxide (DETC-MeSO)¹ (9) that is a potent and selective carbamoylating agent for sulfhydryl groups (10). We now report that DETC-MeSO also partially blocks glutamate binding to synaptic membrane preparations isolated from the brains of mice, and in addition, DETC-MeSO prevents seizures induced in mice by glutamate analogs or by exposure to hyperbaric oxygen.

MATERIALS AND METHODS

Male Swiss Webster mice (20-30 g) or male Sprague-Dawley rats (250-300 g) were used in the study. All experiments that employed animals were conducted in strict compliance with the National Institutes of Health guidelines on animal use and institutional regulations concerning animal experimentation. Animals were exposed to hyperbaric oxygen in a specially designed pressure chamber as described previously (11, 12). The time to first clonic-tonic seizure after bringing animals to a final pressure of 5 atmospheres of 100% oxygen or after intraperitoneal injection of convulsants was noted by criteria outlined previously (11, 12). Unless otherwise specified, the ability of DETC-MeSO to prevent seizures was tested by intraperitoneal injection of 5.2 mg/kg DETC-MeSO 1-2 h prior to bringing the animal to a final pressure of 5 atmospheres of 100% oxygen or intraperitoneal injection of N-methyl-D-aspartate (NMDA) (125 mg/kg) or L-methionine sulfoximine (MetSOX) (250 mg/kg). Evaluation of the statistics for whole animal experiments or for changes in brain glutamate binding after administration of DETC-MeSO was conducted by use of the program GraphPAD InStat from GraphPAD Software (San Diego, CA).

Synaptic membranes (100 µg of protein) were isolated (13) from whole brain homogenate of male Swiss Webster mice and were incubated in 0.1 ml of 10 mM potassium phosphate, pH 7.4, and DETC-MeSO (1 µM to 1 mM) or 0.5 mM L-glutamate for 30 min at 25 °C. After addition of 50 nM [³H]glutamate, incubation was continued for an additional 45 min. Reactions were terminated by centrifugation at 4 °C to separate membrane-bound from free radioactivity. Nonspecific binding (radioactivity bound in the presence of 0.5 mM unlabeled glutamate) averaged 20-30% of total radioactivity bound. The rate (k) and maximum percent blockage of glutamate binding (M) by DETC-MeSO (I) during the incubation time (t) was determined by fitting the equation: % inhibition = M(1  e^kIt) to these data by use of the program Grafit (Erithicus Software Ltd.). This equation was derived for first-order inactivation by a group-specific reagent that gives a partial effect. If "A" is the percent of binding activity that remains after extensive exposure (t = ) of the receptor to excess DETC-MeSO, then M = 100  A. The percent activity remaining at any time (a) would be equal to: a = (100  A)e^kIt + A; substituting (100  M) for A and [100  % inhibition] for a in this equation gives the one employed for purposes of data analyses. The maximum percent blockage of glutamate binding (M) observed did not depend on the concentration of glutamate employed in binding experiments, such that the modified receptors appeared to be "noncompetitively" inhibited (independent of whether DETC-MeSO was still present or absent at the time that glutamate binding was assayed). A second equation was employed in the analysis of binding data that defines partial, irreversible inactivation of two distinct groups of receptors: % inhibition = M₁(1  e^k₁It) + M₂(1  e^k₂It), where M₁ and M₂ are the maximum amount of inhibition obtained for complete modification of each receptor population, and k₁ and k₂ are the rate constants for these modifications, respectively.

DETC-MeSO was synthesized by the method of Hart and Faiman (14). Carbamoylated glutathione, S-(N,N-diethylcarbamoyl)glutathione (DETC-GS), was prepared essentially as described by Jin et al. (10). The structures and purity of DETC-MeSO and DETC-GS were confirmed by 300 MHz NMR and fast atom bombardment-tandem mass spectrometry. A QE300 NMR spectrometer (General Electric, Fremont, CA) and Autospec-Q tandem hybrid mass spectrometer (Fiscons/VG Analytical Limited, Manchester, United Kingdom) were employed for these analyses. Liver aldehyde dehydrogenase was extracted and assayed as described previously (14). MetSOX, monosodium L-glutamate, glycine, glutathione, and NMDA were purchased from Sigma. L[G-³H]Glutamate (46 Ci/mmol) was obtained from Amersham Life Science (Buckinghamshire, United Kingdom).

RESULTS

Treatment of synaptic membrane preparations (13) from the brains of mice with DETC-MeSO resulted in a time-dependent (k = 25 ± 10 M¹ s¹), partial (M = 58 ± 7%), and irreversible loss of their ability to bind glutamate (Fig. 1). Inhibition of glutamate binding appeared to depend on modification of more than one population of glutamate receptor, each with distinct kinetics. Although a better fit of these data could be obtained by use of a double exponential equation (see "Materials and Methods"), a unique fit of this equation to these data could not be obtained, due to the larger number of independent parameters involved. However, a nominal fit of these data indicated that about one-fourth of the total inhibition (M₁ = 15%) occurs at a much faster rate (k₁ = 900 M¹ s¹), than the rate (k₂ = 8 M¹ s¹) associated with inhibition of the remainder (M₂ = 43%) (Fig. 1). Since the effect of DETC-MeSO on either receptor population is irreversible, glutamate receptor blockage in vivo can be easily determined.

1

1

1

1

1

1

Synaptic membranes were prepared from the brains of mice injected with DETC-MeSO (5.2 mg/kg, intraperitoneal). The ability of synaptic membrane preparations isolated from DETC-MeSO-treated mice to bind glutamate was compared with similar preparations isolated from control animals. Fig. 2A illustrates the results obtained from synaptic membranes prepared from brains of mice killed 2 h after a single dose of DETC-MeSO or 2 h after the last injection of multiple consecutive doses (5.2 mg/kg, intraperitoneal, daily for 7 days). Either single or multiple dosing with DETC-MeSO reduced the capacity of synaptic membranes to bind glutamate by approximately 50%, the maximum effect obtained in vitro (Fig. 2A). For brain synaptic membrane preparations isolated from mice killed 24 h after a single dose of DETC-MeSO, less inhibition of glutamate binding was found (Fig. 2B). By contrast, for brain synaptic membrane preparations isolated from mice killed 24 h after the last dose of seven daily consecutive doses, similar inhibition of glutamate binding was observed to that seen 2-h postinjection of DETC-MeSO (Fig. 2B). The results obtained for the group that received a single dose of DETC-MeSO and were sacrificed after 2 h were significantly different from the group that received a single dose and were sacrificed after 24 h (Bonferroni's p < 0.01 determined by analysis of variance). Comparison of the singly dosed group that was sacrificed after 2 h with the multiply dosed groups that were sacrificed after 2 h or 24 h did not show statistical significance from each other by the same criteria.

It is hard to reconcile the ability of DETC-MeSO to carbamoylate brain glutamate receptors with its extreme lability in vivo. DETC-MeSO rapidly and selectively carbamoylates the sulfhydryl of glutathione (GSH) in vitro (10 M¹ s¹ at 37 °C, pH 7), and since the concentration of GSH in vivo is high (1-6 mM) (15), DETC-MeSO will rapidly be converted (>95% within 5 min) to DETC-GS, Fig. 3. DETC-GS has been detected in the bile of mice treated with disulfiram (10). Unlike DETC-MeSO, DETC-GS is not reactive and does not carbamoylate sulfhydryl groups (10). Although DETC-GS reversibly blocked glutamate binding to synaptic membrane preparations from mouse brain (data not shown), a time-dependent, irreversible blockage of the glutamate receptor, like that observed with DETC-MeSO, was not obtained.

If the effect of DETC-MeSO on glutamate receptors in vivo is mediated by glutathione, then the carbamoylated glutathione requires activation. GSH, oxidized glutathione (GSSG), and DETC-GS reversibly blocked glutamate binding to mouse synaptic membrane preparations (data not shown). However, when the membranes were washed after exposure to GSH, GSSG, or DETC-GS, inhibition was reversed, unlike the effect by DETC-MeSO. Oxidation of the sulfur of DETC-GS to a sulfoxide would make it reactive toward sulfhydryl groups (9), similar in chemistry to DETC-MeSO, and potentially capable of irreversible inhibition of glutamate receptors in vivo.

Carbamoylated glutathione and DETC-MeSO had the same effect in vivo, despite the fact that they had different effects on glutamate receptors in vitro. Intravenous administration of an equimolar concentration of DETC-GS or DETC-MeSO (30 µmol/kg) to mice resulted in a comparable degree of irreversible brain glutamate receptor blockage (26.9 ± 4.3 and 38.2 ± 1.6%, respectively) or liver aldehyde dehydrogenase inhibition (30.1 ± 1.0 or 44.9 ± 1.0%, respectively). Since DETC-GS reversibly blocked glutamate binding to synaptic membranes in vitro (i.e. the inhibition can be reversed by washing the membranes to remove DETC-GS), but in vivo both DETC-MeSO and DETC-GS irreversibly blocked glutamate binding, and DETC-GS has no effect on aldehyde dehydrogenase in vitro (reversible or irreversible), it is suggested that DETC-GS is activated by oxidation in vivo (DETC-GSO, Fig. 3).

To test whether carbamoylation of glutamate receptors might prevent seizures caused by glutamate agonists, the effect of DETC-MeSO on seizures induced by glutamate analogs was examined. Treatment of mice with DETC-MeSO prior to administration of the glutamate analog NMDA (125 mg/kg, intraperitoneal) prevented seizures that result from NMDA administration alone (Table I). Similarly, DETC-MeSO administered to mice or rats prior to injection of the glutamate analog MetSOX more than doubled the time that the animals remain free of seizures (Table I).

Table I. Effect of NMDA, MetSOX, and oxygen on mice and rats Treatment Mean time to seizure Control animals DETC-MeSO-treatedb min NMDA 16  ± 3 >120a (125 mg/kg, mice) (5/8)a (0/6)a MetSOX 127  ± 24 307  ± 60 (250 mg/kg, mice) (8/8) (6/8)a MetSOX 136  ± 6 318  ± 60 (250 mg/kg, rats) (4/4) (4/4) 5 ATA O2c 24  ± 3 >60a (mice) (7/7) (0/5)a a The mean values for the time to the first clonic-tonic seizure are reported ± S.E. Immediately below this value in parentheses is the number of animals that exhibited the effect divided by the total number of animals in that group (number of animals to which value applies/total number of animals in the group). Comparison of the means for control (untreated) and DETC-MeSO-treated animals by a two-tailed t test gave p < 0.001. In the case of NMDA treatment, three of the eight animals in the control group failed to exhibit any seizures, even after a second injection of NMDA. Resistance to NMDA-induced seizures appears to be due to P450-mediated metabolism of this glutamate analog, based on the ability of N-benzylimidazole, a P450 inhibitor, to produce sensitivity in NMDA-resistant animals. None of the animals in the DETC-MeSO-treated group (six out of six animals) exhibited any NMDA-induced seizures for the duration of the observation period (2 h). In the case of MetSOX treatment, two out of eight animals in the DETC-MeSO-treated group remained free of seizures for the period of observation (6 h). In the case of hyperbaric oxygen exposure (5 atmospheres absolute O2), all of the animals in the DETC-MeSO-treated group (five out of five animals) remained free of seizures for the period of observation (1 h). b DETC-MeSO was administered at a dose of 5.2 mg/kg by intraperitoneal injection 1-2 h prior to exposure to hyperbaric oxygen, NMDA, or MetSOX. c Mice were exposed to 5 atmospheres absolute (ATA) of 100% oxygen in a pressure chamber for 60 min, then depressurized.

It has been shown that glutamate is released by rat hippocampal (brain) slices subjected to oxidative stress (16). Therefore, we examined the affect of DETC-MeSO on oxygen-induced seizures. Administration of DETC-MeSO (5.2 mg/kg, intraperitoneal) to mice two hours before exposure to 5 atmospheres of 100% oxygen, prevented the seizures that occurred after 24 min in control animals (Table I).

NMDA is a selective agonist for a major subtype of ionotropic (calcium ion channel-linked) glutamate receptor (17). As determined in the results illustrated in Figs. 1 and 2, the effect of DETC-MeSO on glutamate binding to synaptic membrane preparations is not a measure of DETC-MeSO's modification of this receptor subtype. Although up to 34% of the total glutamate binding capacity of synaptic membranes is attributable to NMDA receptors (18), the effect of glutamate on these receptors is dependent on glycine (19), that was not included in the studies presented in Figs. 1 and 2. Under the conditions of these experiments, NMDA does not block glutamate binding to synaptic membrane preparations. When these binding experiments were repeated in the presence of glycine (0.1 mM), reversible blockage of glutamate binding to mouse brain synaptic membrane preparations by NMDA (0.5 mM, 32 ± 2% blockage) and irreversible blockage by DETC-MeSO (0.1 mM, 48 ± 3% blockage) was observed. The similar degree of inhibition observed by DETC-MeSO in the presence and absence of glycine (48 and 58% (Fig. 1), respectively) indicates that both NMDA and non-NMDA glutamate receptor subtypes are affected to a similar extent by carbamoylation.

DISCUSSION

DETC-MeSO is effective in partially preventing glutamate binding to brain synaptic membrane preparations in vitro and in vivo. Inhibition of the glutamate receptor by DETC-MeSO in vivo is suggested to be mediated by GSH. DETC-MeSO carbamoylates GSH to form DETC-GS. DETC-GS crosses the blood-brain barrier and is then oxidized to DETC-GSO at the site of action. DETC-GSO would be the ultimate carbamoylating agent in vivo. The interconversion of DETC-MeSO, DETC-GS, and DETC-GSO is illustrated in Fig. 3. This novel and efficient method for delivering the carbamoyl moiety across the blood-brain barrier converts a small fraction of the circulating glutathione into a carbamoyl delivery system. Reaction of the oxidized, reactive (sulfoxide) form of carbamoylated glutathione (DETC-GSO) with another molecule of glutathione simply converts the carbamoyl moiety from a reactive form (DETC-GSO) to a latent one (DETC-GS). Oxidation of the latent form of carbamoylated glutathione (DETC-GS) would convert it once again to the reactive form (DETC-GSO). Thus, the carbamoyl moiety could cycle many times between latent and reactive forms prior to delivery at its site of action in vivo.

Evidence that brain glutamate receptors are carbamoylated in vivo comes from: 1) the chemistry of carbamoyl sulfoxides, that are selective carbamoylating agents for sulfhydryl groups (9, 10); 2) the effect in vivo is chemically (but not biologically) irreversible, as would be expected for carbamoylated sulfhydryl residues (9, 10); 3) the rate at which the majority of glutamate binding is lost in vitro as a consequence of receptor exposure to DETC-MeSO (8-25 M¹ s¹) is comparable with the rate at which DETC-MeSO reacts with the sulfhydryl of glutathione (10 M¹ s¹); 4) the size of the maximum effect of DETC-MeSO in vitro (58% loss of glutamate binding capacity by synaptic membrane preparations, Fig. 1) is approximately equal to the maximum effect observed in vivo (53% loss of glutamate binding capacity after multiple serial doses of DETC-MeSO, Fig. 2A); 5) the partial effect of other sulfhydryl modifying reagents on the glutamate receptor (NMDA subtype) in vitro, indicating that DETC-MeSO modifies the known "redox modulatory site" of the receptor (4); 6) the protection by DETC-MeSO against seizures induced by NMDA, MetSOX, or oxygen at high pressure. While it is possible that DETC-MeSO elicits the effect at the glutamate receptor by a mechanism other than carbamoylation and/or prevents seizures by effects in vivo that are unrelated to modification of the glutamate receptor, these possibilities seem rather remote.

NMDA is recognized to be a selective agonist of a specific subtype of glutamate receptor that is associated with calcium ion channels (17). As such, seizures induced as a consequence of NMDA administration are commonly thought to be due to interaction of NMDA with glutamate receptors. By contrast, MetSOX is best known as an extremely potent inhibitor of brain glutamine synthetase (20). However, inhibition of glutamine synthetase persists long after MetSOX-induced seizures have subsided (21), indicating that the cause of convulsions induced by this glutamate analog is not related to glutamine synthetase inhibition. It is not presently known which glutamate receptor subtypes may be affected by MetSOX or how potent these interactions may be. The discovery that glutamine synthetase inhibition does not correlate with the seizures induced by MetSOX (21) predates by more than a decade the discovery of glutamate receptors and their essential role in the central nervous system (3).

The observation that DETC-MeSO acts as a glutamate antagonist offers an explanation for several neurological effects of the drug disulfiram. First, it is consistent with the previous observation that disulfiram also prevents oxygen-induced seizures (11, 12) and requires bioactivation to DETC-MeSO (9). Also, occasionally patients treated with disulfiram have been reported to exhibit various neurological disorders, such as encephalopathy (including schizophrenic-like symptoms), parkinsonism, ataxia, choreoathetosis, seizures, optic neuritis, and peripheral neuropathy (22-24). Patients with a clinical diagnosis of schizophrenia are thought to be especially prone to disorientation, impaired memory, and hallucinations upon treatment with disulfiram (24). Some of these rare side effects associated with disulfiram are consistent with those expected due to excessive blockage of glutamate function in the central nervous system (1-3). However, in a controlled clinical study of the incidence of neurological side effects associated with disulfiram use (250 mg/day), there was no statistically higher incidence of neurological problems than were seen in a control population (25). If, in humans, the normal extent of glutamate antagonism by disulfiram were partial (e.g. 60% as observed for mice), then the effects of excessive blockage would only be manifest in individuals with unusual response (e.g. >60%) to carbamoylation of their glutamate receptors. As such, the average incidence of undesirable neurological problems would be quite low. In any case, since there is evidence that the density of glutamate receptors is increased as a consequence of chronic alcohol consumption (26, 27), the effect on the glutamate receptors by DETC-MeSO may actually be of positive benefit during treatment of the alcoholic with disulfiram. Furthermore, the physiology of the disulfiram-ethanol reaction cannot be completely explained by accumulation of acetaldehyde due to inactivation of aldehyde dehydrogenase (25). It is possible that part of the success of disulfiram in treating alcoholism has, in fact, relied upon its previously unrecognized effect on glutamate receptors. Much of the variability in disulfiram's effectiveness in the treatment of alcoholism can be attributed to variable extents of bioactivation to DETC-MeSO (9).

Not all of the adverse neurological effects of disulfiram can be attributed to modification of glutamate receptors. Disulfiram is metabolized to carbon disulfide, a known neurotoxin, and potently inhibits copper enzymes, such as superoxide dismutase and dopamine -hydroxylase, through the action of another disulfiram metabolite, diethyldithiocarbamate (28, 29). DETC-MeSO does not share any of these latter effects with disulfiram (carbon disulfide formation or copper enzyme inhibition),² so it would be expected to more selectively affect glutamate receptors in vivo. In particular, DETC-MeSO is not likely to cause the seizures, optic neuritis, and peripheral neuropathy linked to higher doses (>500 mg/day) of disulfiram, that are most probably a consequence of CS₂ formation (22-24).

A strong correlation exists between glutamate excitotoxicity and damage due to free radicals (3). Brain hippocampal slices exposed to superoxide selectively release glutamate into the media, without releasing other intracellular constituents (16). The mechanism by which superoxide triggers the release of glutamate is not known. However, it is tempting to speculate that oxidation of sulfhydryl groups on presynaptic neurons, perhaps even sulfhydryl groups associated with presynaptic glutamate receptors (30), are responsible for triggering glutamate release. Carbamoylation of presynaptic sulfhydryls should render them less susceptible to oxidation by reactive oxygen species. Furthermore, carbamoylation of postsynaptic glutamate receptors (i.e. NMDA receptors) should ameliorate the consequences of reactive oxygen-induced glutamate release. Thus, DETC-MeSO may play a dual role in preventing O₂-induced seizures, in that it could both prevent the triggering of glutamate release presynaptically and prevent the consequences of glutamate release postsynaptically. There is not a clear consensus on the origin of seizures induced by oxygen at high pressure. The ability of DETC-MeSO to prevent O₂-induced seizures, while consistent with these seizures having their origin in glutamate excitotoxicity, does not necessarily resolve the complete etiology of these seizures. The molecular mechanism by which the glutamate system becomes perturbed and its overall role in the physiology of these seizures remains to be fully defined. Clearly, further investigation is required to elucidate the details of the mechanism(s) by which DETC-MeSO prevents hyperbaric oxygen-induced seizures.