Journal of Neurochemi W y Raven Press, Ltd., New York 8 1994 International Society for Neurochemlstry

Increased GABA Release in the Human Brain Cortex as a Potential Pathogenetic Basis of Hyperosmolar Diabetic Coma

K. Fink, *J. Zentner, and M. Gothert

Institute of Pharmacology and Toxicology and *Clinic of Neurosurgery, University of Bonn, Bonn, F.R.G.

Abstract: Human cerebral cortical slices preincubated with [3H]GABA, [3H]noradrenaline, or 5-[3H]hydroxy- tryptamine and superfused with Krebs solution or Mg2+- free Krebs solution were used to investigate the influence of increased D-glucose concentrations on the release of these [3H]-neurotransmitters evoked by high K+ content or NMDA receptor activation, respectively. An increase in level of D-glucose (normal content, 11.1 mM) by 32, 60, and/or 100 mM (a range characteristic for hyperosmolar diabetic coma) increased the [3H]GABA release and inhibited the [3H]noradrenaline release evoked by both methods of stimulation. The K+-induced 5-[3H]hydroxy- tryptamine release was also inhibited by high D-glucose content. Blockade of GABA, receptors by p-(3-aminopro- pyl)-p-diethoxymethylphosphinic acid (CGP 35348) atten- uated the inhibitory effect of high D-glucose content on the K+-evoked release of [3H]noradrenaline and 5-[3H]hydroxy- tryptamine, suggesting that the effect on monoamine re- lease is, at least to a major part, the result of the increased GABA release and, as a consequence, of an increased GABA concentration at inhibitory GABA, receptors. The membrane-impermeable sorbitol mimicked the increasing effect of D-glucose on [3H]GABA release and its inhibitory effect on 5-[3H]hydroxytryptamine release. However, di- methyl sulfoxide, which is known to permeate rapidly through biological membranes, had no effect at concen- trations equiosmolar to D-glucose. It is concluded that a reduction in brain cell volume caused by increased extra- cellular, compared with cytoplasmic, osmolarity is crucial for the changes in neuronal function observed at high D- glucose and sorbitol content. In view of the fact that GABA is the main inhibitory neurotransmitter in the brain, the in- creased GABA release may be assumed to contribute to the pathogenesis of hyperosmolar diabetic coma. Key Words: GABA release-Noradrenaline release--5-Hy- droxytryptamine release-Hyperosmolar diabetic coma- GABA, receptors-D-Glucose. J. Neurochem. 62,1476-1 481 (1 994).

old of -20-25 mM, D-glucose is osmotically effective in the extracellular space (Rosen and Andrew, 1991). Because a hyperosmolarity-induced modification of neurotransmitter release may be responsible for the altered consciousness at high D-glucose levels in dia- betic patients, the influence of high concentrations of D-glucose, D-fructose, and NaCl on the K+-evoked re- lease of several 3H-neurotransmitters has recently been investigated in rat brain cortical slices and syn- aptosomes (Fink and Gothert, 1993). It was found that high D-glucose content differentially modifies 'H-neurotransmitter release; the most important ob- servation appeared to be the increase in [3H]GABA release.

In consideration of the potentially high significance of this finding for the pathogenesis of diabetic coma and to exclude that the changes observed in the rat brain cortex might be a peculiarity of that species, human cerebral cortical slices were used in the present study to examine the influence of high D-glucose con- tent on the K+-evoked release of ['HIGABA, [3H]- noradrenaline ( ['HINA), and 5-['H]hydroxytrypta- mine ([3H]5-HT). In a further extension of our pre- vious investigations in rat cortical preparations (see above), we also studied the influence of D-glucose on the release of ['HIGABA and [3H]NA evoked by an- other method of stimulation, namely, by activation of NMDA receptors. It will be shown that, similar to the findings in rat cortical tissue, high D-glucose content increases the release of [3H]GABA and inhibits the release of ['HINA and [3H]5-HT in a manner sensi- tive to blockade of GABA, receptors. On the basis of these findings we also investigated whether the changes observed are due to a reduction in brain cell

The neuronal mechanisms underlying hyperosmo- lar coma, a major complication of diabetes mellitus, are not yet known. D-Glucose permeates neuronal cell membranes by facilitated diffusion, a saturable carrier-mediated process, which is assumed to be near maximal already at physiological concentrations. Therefore, at increased concentrations above a thresh-

Received June 1, 1993; revised manuscript received August 16, 1993; accepted August 16, 1993.

Address correspondence and reprint requests to Dr. M. Gothert at Institute of Pharmacology and Toxicology, University of Bonn, Reuterstrasse 2b, D-53 1 13 Bonn, F.R.G.

Abbreviations used: CGP 35348, p-(3-aminopropyl)-p-diethoxy- methylphosphinic acid; CGP 37849, ~~-(E)-2-amino-4-methyl-5- phosphono-3-pentanoic acid; DMSO, dimethyl sulfoxide; 5-HT, 5-hydroxytryptamine; NA, noradrenaline.

1476

1477 GABA RELEASE AND HYPEROSMOLAR DIABETIC COMA

volume caused by the increase in osmotic gradient extra- versus intracellular.

In contrast to the rat cortical slices (see Fink and Gothert, 19931, those from the human cortical tissue could, for technical reasons, not be prepared immedi- ately after removal from the patients' brain; the speci- mens were stored for up to 30 rnin at 4°C in Krebs solution equilibrated with a mixture of 95% O2 and 5% C 0 2 (for details, see Materials and Methods). It is conceivable that, as a consequence of storage of the specimens before preparation of slices, the 'H-neuro- transmitter release in such slices and its modification by high D-glucose content are changed. This possibil- ity was examined systematically (and excluded) by ex- periments in cortical slices prepared from rat brains either immediately after removal or after storage for up to 30 rnin of the brains in cold Krebs solution.

MATERIALS AND METHODS y-[2,3-3H(N)]Aminobutyric acid ( [,H]GABA; specific ac-

tivity, 75.9 Ci/mmol), (-)-[ring-2,5,6-3H]norepinephrine ([,H]NA; specific activity, 56.9 Ci/mmol), and 5-[ 1,2- 'H(N)]hydroxytryptamine creatinine sulfate ( [3H]5-HT; specific activity, ;!7.6 Ci/mmol) were purchased from NEN (Dreieich, F.R.G.); NMDA, aminooxyacetic acid, P-ala- nine, and (*)-nipecotic acid from Sigma (St. Louis, MO, U.S.A.); Dglucose and dimethyl sulfoxide (DMSO) from Merck (Darmstadt, F.R.G.); and dizocilpine hydrogen ma- leate (formerly designated MK-801) from Research Bio- chemicals, Inc. (Natick, MA, U.S.A.). p(3-Aminopropy1)- p-diethoxymethylphosphinic acid (CGP 35348) and DL- (E)- 2-amino-4- methyl- 5 -phosphono- 3 -pentanoic acid (CGP 37849) were donations from Ciba-Geigy (Basel, Swit- zerland).

Slices (0.3 mm thick, 2 mm in diameter) were prepared from cerebral cortical tissue (temporobasolateral region), which was stored in Krebs solution (4°C; equilibrated with 95% O2 and 5% COz) of the following composition: 118 mM NaCI, 4.8 mM KCI, 25 mM NaHCO,, 1.2 mM KHzPO,, 1.3 mMCaCI,, 1.2 mMMgSO,, 0.06 mMascor- bic acid, 0.03 mM disodium EDTA, and I 1.1 mM ~ - g l u - cose. Each of the human cortical specimens used for prepa- ration (within 30 rnin after removal) was part of the tissue that had to be re:moved for therapeutic reasons from 20 patients (either sex, 25.7 t 3.3 years old) who underwent neurosurgery for untractable epilepsy. The patients were premedicated with flunitrazepam and atropine and anesthe- tized with thiopentone, fentanyl, and enflurane or isoflur- ane. Pancuronium was applied for muscular relaxation. The study was approved by the local ethics committee.

Supplementary experiments were camed out in slices from rat cerebral cortex. Either the slices were prepared im- mediately after removal of the brains subsequent to decapi- tation, or whole brains (comparable in size to the largest tissue specimens provided by the neurosurgeons) were first stored for 15 or 3'0 rnin in Krebs solution under identical conditions as the human tissue and then used for prepara- tion of the slices. In any case only the superficial layer of the cortex was used for preparation of slices.

The slices were incubated in Krebs solution of 37°C con- taining 50 nM ,H-neurotransrnitter. The incubation lasted in the case of [,€I]GABA for 15 rnin and in the case of

['HINA or [3H]5-HT for 30 min. In the GABA experiments, incubation with the labeled transmitter was camed out in the presence of 50 MM aminooxyacetic acid for blockade of GABA transaminase and 100 pMB-alanine for blockade of GABA uptake into glial cells; the slices were superfused in the presence of these drugs plus 1 mM nipecotic acid, an inhibitor of neuronal GABA uptake.

The slices were superfused for 62 rnin at a flow rate of 0.6 ml/min with Krebs solution. From 40 to 42 rnin of superfu- sion, tritium efflux was stimulated with 15-25 mMK+; Na+ content was reduced accordingly to maintain isoosmolarity. When stimulation was carried out with NMDA, Mgz+ was omitted from the superfusion solution to avoid blockade of the NMDA-gated cation channel by Mg2+ (Nowak et al., 1984; Collingridge and Lester, 1989). When relevant, the D-glucose level in the superfusion fluid was increased from the 20th rnin of superfusion onward until the end of the experiment. The radioactivity ofthe 5-min superfusate sam- ples and of the slices solubilized with 0.5 ml of Soluene was determined by liquid scintillation counting.

Basal tritium efflux was calculated as a fraction of tissue tritium at the beginning of the respective collection period. Stimulation-evoked tritium overflow was expressed as a per- centage of tissue tritium at the onset of stimulation. It re- flects release of the respective tritiated and endogenous neu- rotransmitter from the corresponding neuron under the present conditions (Starke et al., 1989).

Results are given as mean -t SEM values of n expen- ments. Student's t test for unpaired data was used to com- pare mean values. Bonferroni's procedure was applied if two or more experimental series were compared with one control group (Wallenstein et al., 1980).

RESULTS

Experiments in human cerebral cortical slices D-Glucose, CGP 35348 [a selective GABA, recep-

tor antagonist (Olpe et al., 1990; Waldmeier and Bau- mann, 1990)], sorbitol, DMSO, or D-glucose plus CGP 35348 at the concentrations investigated did not affect the basal efflux of ['HIGABA, [3H]NA, or ['H]5-HT (determined in the superfusate sample col- lected immediately before stimulation; data not shown).

An increase in D-glucose concentration by 32 and 60 mMin addition to the normal D-glucose content of 1 1.1 mM induced a concentration-dependent in- crease of the K+-evoked ['HIGABA release, but fur- ther elevation of D-glucose content did not further enhance the K+-evoked ['HIGABA release (Fig. la). NMDA induced a weaker stimulation of ['HIGABA release than high K+ content (Table 1; compare with legend to Fig. 1). The NMDA-evoked [3H]GABA re- lease was increased by 82% when the D-glucose con- centration was elevated by 60 mM (Table 1).

In contrast to the enhancement of ['HIGABA re- lease, the K+- and NMDA-evoked release of [3H]NA was inhibited by an increase in D-glucose concentra- tion (by 60 and/or 100 mM; Fig. l b and Table 1). An elevation of D-glucose content by 60 mM also inhib- ited the K+-evoked release of ['HIS-HT (Fig. 2b). Blockade ofGABA, receptors by 100 pMCGP 35348

J. Neurochem., Vol. 62, No. 4, 1994

1478 K. FINK ET AL.

32 60 100 mM D-Glucose

FIG. 1. Effect of an increase in D-glucose concentration on the K+-evoked tritium overflow above basal efflux from human cere- bral cortical slices preincubated with a 3H-neurotransmitter and superfused with Krebs solution. For stimulation K+ content was elevated for 2 rnin after 40 rnin of superfusion. The abscissa indi- cates the amount by which the concentration of o-glucose was increased from the 20th min of superfusion onward until the end of the experiments. Data are mean f SEM (bars) values (n = 8- 16). a: Preincubation with r3H]GABA, stimulation with 25 mM Ki. "p < 0.05, **p < 0.005, compared with the corresponding control experiments without extra glucose. Under control conditions (n = 13), the K+-evoked tritium overflow amounted to 1.30 f 0.20 nCi, corresponding to 2.13 0.13% of tissue tritium. b: Preincu- bation with [3H]NA. stimulation with 20 mM K+. o-Glucose was increased either in the absence (open columns) or in the presence (hatched columns; from the 20th rnin of superfusion onward until the end of the experiments) of 100 +M CGP 35348. *p < 0.05, **p < 0.005, compared with the corresponding control experiments without extra glucose and without CGP 35348; ' p < 0.05, com- parison of experiments in the presence of CGP 35348 with those at the same D-glucose concentration but without CGP 35348. Under control conditions, the K+-evoked tritium overflow in the absence (n = 9) and presence (n = 12) of CGP 35348 amounted to 0.60 -t 0.09 and 0.41 f 0.07 nCi, respectively, corresponding to 2.96 k 0.43 and 2.90 f 0.68% of tissue tritium, respectively.

substantially attenuated the inhibitory effect of high D-glucose content on the K+-evoked release of ['HINA and ['H]5-HT (Figs. lb and 2b).

The membrane-impermeable sorbitol(60 mM) in- creased the K+-evoked [3H]GABA release (Fig. 2a) and inhibited the K+-evoked ['H]5-HT release (Fig. 2b) to a similar extent as D-glucose at equiosmolar concentration, whereas 60 mM DMSO, which is eas- ily diffusible through the cell membrane, was ineffec- tive (Fig. 2).

Experiments in rat cerebral cortical slices In cortical slices prepared from rat brains 15 or 30

rnin after storage in cold Krebs solution, no changes

in K+-evoked release of ['HIGABA (Fig. 3a) or [3H]NA (Fig. 3b) were observed when compared with the release in the slices prepared immediately after removal of the brains. The NMDA-evoked ['HI- GABA (Fig. 3c) or ['HINA (Fig. 3d) release was also not significantly modified by storage of the brains in cold Krebs solution for 30 rnin before preparation of the slices, nor did changes in the extent of the ~-g lu- cose-induced increase in NMDA-evoked [ 'HIGABA release (Fig. 3c) and of the inhibition of NMDA- evoked ['HINA release (Fig. 3d) occur.

In experiments on slices prepared immediately after removal of the brains and superfused with Krebs solution, in which the D-glucose concentration was increased by 100 mM, 100 nMdizocilpine (n = 8) and 30 pM CGP 37849 (n = 8) inhibited the NMDA- evoked ['HIGABA release by 83.7 _t 7.9 ( p < 0.005) and 98.6 k 11.4% ( p < 0.005), respectively (data not shown).

DISCUSSION

The present investigation in human cerebral corti- cal slices revealed that an increase in D-glucose con- centration by 32- 100 mM differentially affected the stimulation-evoked release of GABA compared with NA and 5-HT. Whereas release of GABA, the main inhibitory neurotransmitter in the brain, was in- creased, that of the monoamines was inhibited. Simi- lar findings were obtained in slices from normal rat brain cortex (Fink and Gothert, 1993; present study), excluding the possibility that the changes in neuro- transmitter release are due to an impairment of the tissue caused by surgical manipulations or the pa-

TABLE l. Effect of an increase in D-glucose concentration on the NMDA-evoked tritium ovrrflow above basal eflux from human cerebral cortical slices preincubated with ['HJGABA or [3H]NA and superfused with Mg2'-free

Krebs solution

NMDA-evoked tritium overflow

n (9% of tissue tritium)"

['HIGABA Controls 12 0.50 f 0.14 D-Glucose (increase by 60 mM) 8 0.91 f 0.16'

Controls 14 0.97 f 0.10 D-Glucose (increase by 100 mM) 14 0.72 f O.lOb

['HINA

NMDA ( 1 mM) was added to the superfusion buffer for 2 rnin after 40 rnin of superfusion. The concentration of D-glucose was increased from the 20th rnin of superfusion onward until the end of the experiments. In the controls, the absolute amounts of NMDA- evoked tritium overflow were 0.44 f 0. I7 nCi in slices preincubated with ['HIGABA and 0.16 f 0.02 nCi in slices preincubated with ['HINA.

Percentage of tissue tritium at the onset of stimulation. Data are mean f SEM values.

p < 0.05 (compared with the corresponding controls).

J. Neurochem., Vol. 62. No. 4. 1994

GABA RELEASE AND HYPEROSMOLAR DIABETIC COMA 1479

c 0

E l 0

L3 HIGABA

*

- DGluc Sab DMSO

Control MmM MmM MmM

1

b L3H15-HT

1

D G k Sab DMSO Controls MmM G(knM MmM

FIG. 2. Effects of sorbitol (Sorb), DMSO, and an increase in D-

glucose (D-G~uc) concentration on the K+-evoked-tritium overflow above basal efflux from human cerebral cortical slices preincu- bated with (a) [3H]GABA or (b) f'HI5-HT and superfused with Krebs solution. K+ content was elevated (25 mM) for 2 min after 40 rnin of superfusion. The abscissa indicates the amount by which the concentration of o-glucose was increased or, in the case of sorbitol and DMSO, the final concentration of these sub- stances from the 20th rnin of superfusion onward until the end of the experiments either in the absence (open columns) or pres- ence (hatched columns; from the 20th min of superfusion until the end of the experiments) of 100 pM CGP 35348. *p < 0.05, ' *p i 0.005, compared with the corresponding control experiments without sorbitol, DMSO, and extra o-glucose (comparison of val- ues represented by empty columns). Data are means f SEM (bars) values (n = 11-28), ' p i 0.05, compared with the corre- sponding value at the same o-glucose concentration, indicating significant antagonism by CGP 35348 of the D-glucose-induced inhibition (comparison of the value represented by the hatched column with that represented by the corresponding open col- umn). Under control conditions, the K+-evoked tritium overflow in the f'H]GABA (n = 28) as well as f'H15-HT experiments in the absence (n = 15) arid presence (n = 13) of CGP 35348 amounted to 2.59 ? 0.50, 0.30 f 0.05, and 0.20 f 0.04 nCi, corresponding to 2.32 f 0.13, 2.18 f 0.29, and 2.03 f 0.33% of tissue tritium, respectively.

tient's disease. 'The present experiments in rat brain cortical slices argue against the possibility that storage of brain tissue specimens for up to 30 rnin in cold oxygenated Krebs solution before preparation of slices caused impairment of the mechanisms involved in stimulation-evoked 3H-neurotransmitter release and in D-glucose-induced modification of this release. Irrespective of the metabolic state of the brain tissue under the present storage conditions and of the method of stimulation applied, neither the evoked [3H]GABA and [3H]NA release nor the glucose-in- duced facilitation of [3H]GABA release and inhibi- tion of [3H]NA release in rat cortical slices was found to be significantly changed by storage for up to 30 rnin of whole brains before preparation of the slices.

In the present study the effects of high D-glucose content were observed not only when the neuronal cell membrane was depolarized with high K' levels, but also when stimulation was performed with NMDA. This drug activates one of the main classes of

excitatory amino acid receptors, the NMDA recep- tors, via which glutamate exerts its effects (Stone and Burton, 1988; Collingridge and Lester, 1989). The stimulant effect of NMDA receptor activation on

Stimulus: high K'

[3H]GABA [3H]NA

0 15 30 mm 0 15 30

Stimulus: NMDA

[3H]GABA [3H]NA I *

30 0 30 min

FIG. 3. Tritium overflow evoked by high Kf concentration or NMDA from rat brain cortical slices prepared after different times of storage of brains in oxygenated Krebs solution of 4'C and subsequently incubated with a 3H-neurotransmitter. The slices were superfused with Krebs solution or, when indicated, Mg2+- free Krebs solution and stimulated for 2 min after 40 min of super- fusion. The abscissa indicates the time of storage of brains in the cold Krebs solution after decapitation of the rats and removal of the brains. The ordinate indicates the K+- or NMDA-evoked tri- tium overflow expressed as a percentage of the overflow in slices prepared immediately after decapitation of the rats and super- fused without additional o-glucose ("controls at 0 min"). Data are mean f SEM (bars) values from eight experiments in each series. a: Preincubation with [3H]GABA, stimulation with 25 mM K'. The K+-evoked tritium overflow at 0 min amounted to 0.58 ? 0.08 nCi, corresponding to 1.73 f 0.14% of tissue tritium. b: Preincubation with [3H]NA, stimulation with 15 mM K'. The K+-evoked tritium overflow at 0 rnin amounted to 0.79 f 0.1 1 nCi, corresponding to 3.93 k 0.46% of tissue tritium. c: Preincubation with [3H]GABA, stimulation with 1 mM NMDA. Open columns represent slices superfused with Mg2+-free Krebs solution; stippled columns repre- sent slices superfused with Mg2+-free solution in which the D-glu- cose concentration was increased by 100 mM from the 20th min of superfusion onward until the end of the experiments. *p < 0.05, significant increase in the NMDA-evoked tritium overflow by the increased D-glucose concentration. The NMDA-evoked tri- tium overflow in the 0-min controls amounted to 0.20 ? 0.06 nCi, corresponding to 0.33 k 0.07% of tissue tritium. d: Preincubation with [3H]NA, stimulation with 300 p M NMDA. Open columns repre- sent slices superfused with Mg2+-free Krebs solution; stippled columns represent slices superfused with Mg2+-free solution in which the o-glucose concentration was increased by 100 mM from the 20th rnin of superfusion onward until the end of the experiments. *p < 0.05, significant inhibition of the NMDA- evoked tritium overflow by the increased o-glucose concentra- tion. The NMDA-evoked tritium overflow in the 0-min controls amounted to 3.30 * 0.37 nCi, corresponding to 10.77 & 1.26% of tissue tritium.

J . Neurochem., Val. 62, No. 4, 1994

1480 K. FINK ET AL.

GABA release has been proved in primary cultures of GABAergic neurons from rat cerebral cortex (Reyn- olds et al., 1989) but not yet in slices of the cerebral cortex. In the present study, evidence is presented that NMDA receptors are, in fact, involved in the release of [3H]GABA from cortical GABAergic interneurons. This conclusion can be drawn from the ability of the competitive NMDA receptor antagonist CGP 37849 (Fagg et al., 1989; Fink et al., 1992) and ofdizocilpine [which blocks the NMDA-gated ion channel by inter- acting with the phencyclidine site within this channel (Lalies et al., 1988; Lodge and Johnson, 1990)] to counteract the NMDA-evoked [3H]GABA release. A stimulant effect of NMDA receptor activation on NA release in the cerebral cortex has already been estab- lished under the present conditions in rat (Fink et al., 1989) and human (Fink et al., 1992) cortical slices.

The inhibitory effect of high D-glucose content on the release of the 3H-monoamines was attenuated by CGP 35348, a competitive antagonist at GABAB re- ceptors. This finding suggests that the inhibition of [3H]NA and [3H]5-HT release is causally related to the increased GABA release, resulting in an increased stimulation of inhibitory GABA, heteroreceptors. CGP 35348 has recently been reported to block GABAB heteroreceptors on glutamatergic neurons, leaving GABAB autoreceptors on cortical GABAergic interneurons unaffected (Bonanno and Raiteri, 1992); these differential effects of CGP 35348 on GABAB auto- and heteroreceptors, suggesting the involvement of different subtypes of GABAB recep- tors (Bonanno and Raiteri, 1992), are compatible with our interpretation that an increased GABAergic neurotransmission is induced by high D-glucose con- tent. Either presynaptic GABAB heteroreceptors on the noradrenergic and serotoninergic axon terminals may be involved (Bowery et al., 1980; Schlicker et al., 1984; Gray and Green, 1987), or the GABAB hetero- receptors may be located on an unknown excitatory neuron in the human cerebral cortex that innervates the noradrenergic and serotoninergic nerve terminals.

The effects of high D-glucose content on 3H-neuro- transmitter release were shared by equiosmolar con- centrations of the membrane-impermeable sorbitol but not by the easily diffusible DMSO. Because, in contrast to D-glucose and sorbitol, DMSO is assumed to be present at the same concentration on both sides of the cell membrane after the exposure time of 20 min, the reduction in brain cell volume caused by increased extracellular osmolarity, compared with that in cytoplasma, is crucial for the changes in neuro- nal function observed at high D-glucose or sorbitol content. The subcellular mechanism by which extra- cellular hyperosmolarity leads to the increase in GABA release and, as a consequence, to the inhibi- tion of NA and 5-HT release is unknown.

The D-glucose concentrations at which the release of 3H-neurotransmitters was changed in human cere- bral cortical slices are within the range measured in hyperosmotic diabetic coma (Arieff and Carroll,

1974; Gordon and Kabadi, 1976). Even patients with an initial serum glucose level of 264 (Knowles, 1966) or 3 1 1 mM (Soni et al., 1990) have been successfully treated. Accordingly, the present data are compatible with the suggestion that such high D-glucose concen- trations primarily increase the release of GABA, which is the main inhibitory neurotransmitter in the brain (Olsen, 1982). Due to the enhanced concentra- tion of GABA at pre- and postsynaptic inhibitory GABA receptors on neighboring neurons, neuronal excitability and release of other neurotransmitters may be assumed to be decreased, thus probably con- tributing to the pathogenesis of hyperosmolar dia- betic coma.

Acknowledgment: We thank Mrs. H. Burisch and Mrs. I. Konrad for technical assistance. This study was supported by the Deutsche Forschungsgemeinschaft and by a joint grant of the Medical Centre of the University of Bonn and the state of Nordrhein-Westfalen.

REFERENCES Arieff A. 1. and Carroll H. J. (1974) Cerebral edema and depression

of sensorium in nonketotic hyperosmolar coma. Diabetes 23, 525-53 1.

Bonanno G. and Raiteri M. (1992) Functional evidence for multi- ple y-aminobutyric acid, receptor subtypes in the rat cerebral cortex. J. Pharmacol. Exp. Ther. 262, 1 14-1 18.

Bowery N. G., Hill D. R., and Hudson A. L. (1980) (-)Badofen decreases neurotransmitter release in the mammalian CNS by an action at a novel GABA receptor. Nature 283, 92-94.

Collingridge G. L. and Lester R. A. J. (1989) Excitatory amino acid receptors in the vertebrate central nervous system. Pharmacol. Rev. 40, 143-210.

Fagg G. E., Pozza M. F., Olpe H. R., Brugger F., Baumann P., Bittiger H., Schmutz M., Angst C., Brundish D., Allgeier H., Heckendorn R., and Dingwall J. G. ( 1 989) CGP 37849: charac- terization of a novel and potent competitive N-methyl-D- aspartate (NMDA) receptor antagonist. (Abstr.) Br. J . Phar- macol. 97, 582P.

Fink K. and Gothert M. (1993) High D-glucose concentrations in- crease GABA release but inhibit release of norepinephrine and 5-hydroxytryptamine in rat cerebral cortex. Brain Res. 618,

Fink K., Gothert M., Molderings M., and Schlicker E. (1989) N- Methyl-D-aspartate (NMDA) receptors mediating stimulation of noradrenaline release, but not release of other neurotrans- mitters, in the rat brain cortex: receptor location, characteriza- tion and desensitization. Naunyn Schmiedebergs Arch. Phar- macol. 339,s 14-52 1.

Fink K., Schultheilj R., and Gothert M. (1992) Stimulation of nor- adrenaline release in human cerebral cortex mediated by N- methyl-Daspartate (NMDA) and non-NMDA receptors. Br. J. Pharmacol. 106,67-72.

Gordon E. E. and Kabadi U. M. (1976) The hyperglycemic hyper- osmolar syndrome. Am. J. Med. Sci. 271,253-268.

Gray J. A. and Green A. R. ( I 987) GABA,-receptor mediated inhi- bition of potassium-evoked release of endogenous 5-hydroxy- tryptamine from mouse frontal cortex. Br. J. Pharmacol. 91, 5 17-522.

Knowles H. C. (1966) Sirupy blood (editorial). Diabetes 15, 760- 761.

Lalies M., Middlemiss D. N., and Ransom R. (1988) Stereoselec- tive antagonism of NMDA-stimulated noradrenaline release from rat hippocampal slices by MK-801. Neurosci. Lett. 91,

Lodge D. and Johnson K. M. (1990) Noncompetitive excitatory

220-226.

339-342.

J. Neurochem., Vol. 62, No. 4. 1994

GABA RELEASE AND HYPEROSMOLAR DIABETIC COMA

amino acid receptor antagonists. Trends Pharmacol. Sci. 1 I ,

Nowak L., Bregestowsky P., Ascher P., Herbet A., and Prochiantz A. (1984) Magnesium gates glutamate-activated channels in mouse central neurones. Nature 307,462-465.

Olpe H . R., Karlsson G., Pozza M. F., Brugger F., Steinmann M., van Riezen H ., Fagg G., Hall R. G., Froestl W., and Bittiger H. ( 1 990) CGP 35348: a centrally active blocker of GABAB recep- tors. Eur. J. Pharmacol. 187,27-38.

Olsen R. W. (198;!) Drug interactions at the GABA receptor-iono- phore complex. Annu. Rev. Pharmacol. Toxicol. 22, 245-211.

Reynolds I. J., Hams K. M., and Miller R. J. (1989) NMDA recep- tor antagonists that bind to the strychnine-insensitive glycine site and inhibit NMDA-induced Ca2+ fluxes and [3H]GABA release. Eur. .I. Pharmacol. Mol. Pharmacol. Sect. 172,9- 17.

Rosen A. S. and Andrew R. D. (1991) Glucose concentration in-

8 1-86.

1481

versely alters neocortical slice excitability through an osmotic effect. Brain Res. 555, 58-64.

Schlicker E., Classen K., and Gothert M. (1984) GABA, receptor- mediated inhibition of serotonin release in the rat brain. Naunyn Schmiedebergs Arch. Pharmacol. 326,99- 105.

Soni A., Rao S. V., Bajaja R., and Treser G. (1990) Extreme hyper- glycemia and hyperosmolarity. Diabetes Care 13, 18 1-182.

Starke K., Gothert M., and Kilbinger H. (1989) Modulation of neu- rotransmitter release by presynaptic autoreceptors. Physiol. Rev. 69, 864-989.

Stone T. W. and Burton N. R. (1988) NMDA receptors and ligands in the vertebrate CNS. Prog. Neurobiol. 10,333-368.

Waldmeier P. C. and Baumann P. A. (1990) Presynaptic GABA receptors. Ann. NY Acad. Sci. 604, 136- 15 1.

Wallenstein S., Zucker L. L., and Fleiss J. L. (1980) Some statistical methods in circulation research. Circ. Res. 47, 1-9.

J. Neuroehem.. Vol. 62, No. 4, 1994