0270.6474/84/0407-1884$02.00/0 The Journal of Neuroscience Copyright 0 Society for Neuroscience Vol. 4, No. 7, pp. 1884-1891 Printed in U.S.A. July 1984 SYNAPTIC RELEASE OF EXCITATORY AMINO ACID NEUROTRANSMITTER MEDIATES ANOXIC NEURONAL DEATH’ STEVEN ROTHMAN Departments of Pediatrics, Neurology, and Anatomy and Neurobiology, Washington University School of Medicine, St. Louis. Missouri 63110 Received December 27, 1983; Revised February 16, 1984; Accepted February 27, 1984 Abstract The pathophysiology of hypoxic neuronal death, which is difficult to study in vivo, was further defined in vitro by placing dispersed cultures of rat hippocampal neurons into an anoxic atmosphere. Previous experiments had demonstrated that the addition of high concentrations of magnesium, which blocks transmitter release, protected anoxic neurons. These more recent experiments have shown that y-D-glutamylglycine (DGG), a postsynaptic blocker of excitatory amino acids, was highly effective in preventing anoxic neuronal death. DGG also completely protected the cultured neurons from the toxicity of exogenous glutamate (GLU) and aspartate (ASP). In parallel physiology experiments, DGG blocked the depolarization produced by GLU and ASP, and dramatically reduced EPSPs in synaptically coupled pairs of neurons. These results provide convincing evidence that the synaptic release of excitatory transmitter, most likely GLU or ASP, mediates the death of anoxic neurons. This result has far-reaching implications regarding the interpretation of the existing literature on cerebral hypoxia. Furthermore, it suggests new strategies that may be effective in preventing the devastating insults produced by cerebral hypoxia and ischemia in man. Cerebral hypoxia, either as an isolated event, or as a duced by hypoxia, unaccompanied by other complicating concomitant of occlusive cerebrovascular disease, peri- variables. Experiments with fetal rat hippocampal neu- natal asphyxia, or cardiorespiratory failure, is a frequent rons maintained in dispersed tissue culture showed that cause of human neurological injury. On occasion, the synaptic activity mediated neuronal death when the cul- deficits produced by hypoxia are reversible, but irre- tures were exposed to an anoxic environment (Rothman, versible brain damage is a far more common event. 1983). Freshly plated neurons, which had not yet formed Despite the importance of the problem, the pathophys- synapses, were very resistant to prolonged anoxia. Ma- iology of hypoxic brain injury has not yet been deter- ture cultures markedly deteriorated in the absence of mined, although it appears that altered calcium homeo- oxygen, and within a few hours all of the neurons had stasis, elevated free fatty acid concentrations, accumu- been replaced by debris. However, when synaptic activity lation of eicosanoids, and increased extracellular lactic was blocked by high concentrations of magnesium, the acid all contribute to neuronal death (Siesjo, 1981; cultures tolerated prolonged anoxia and showed virtually Raichle, 1983). However, resolving all of the factors no morphological changes. responsible for pure hypoxic brain injury has remained While these experiments clearly demonstrated that difficult because hypoxia in vivo is invariably associated synaptic activity was responsible for killing hypoxic neu- with hypotension, hypercarbia, and acidosis, which are rons, they failed to indicate the mechanism of activity- all potentially damaging. induced neuronal death. The experiments described in Recent in vitro observations have provided new in- the present paper provide convincing evidence that syn- sights into the pathophysiology of neuronal injury pro- aptically released neurotransmitter, most likely gluta- mate (GLU) or aspartate (ASP), initiates the cascade of 1 This work was supported by National Institute of Neurological and reactions that culminates in neuronal death in hypoxia. Communicative Disorders and Stroke Grants NS00568 and NS19988 and by the Epilepsy Foundation of America. I would like to thank Drs. Materials and Methods Gerald Fischbach, Richard Hume, Marcus Raichle, and Joseph Volpe, who critiqued earlier drafts of this manuscript; Irene Karl, Ph.D., who Culture. Hippocampi were dissected from the brains of provided the glutamate assay; Miriam Samaie, who provided excellent l&day-old rat fetuses under sterile conditions and were technical help; Robert Freund, who did the photography; and Kim placed in 0.1% trypsin (Gibco) in calcium-magnesium- Kendall, who typed the paper. free Hanks’ Balanced Salt Solution (HBSS) buffered to 1884 The Journal of Neuroscience Excitatory Amino Acids Mediate Effects of Anoxia 1885 pH 7.3 with 10 mM HEPES. After a 15min incubation HEPES (pH 7.3). Tetrodotoxin (TTX), 1 pg/ml, was at 37”C, the hippocampi were washed three times with added to block action potentials in some experiments. HBSS, resuspended in approximately 1 ml of HBSS, and The culture was transferred to the stage of an inverted, dissociated by passage through a series of Pasteur pi- phase contrast microscope, and the temperature was pettes with flame-narrowed tips. The cells were then maintained at 35°C by a heater insert connected to a pipetted into 35mm tissue culture dishes (Falcon) at a feedback system. Evaporation was prevented by covering density of approximately 1.5 to 2.0 X lo4 cells/cm2. Each the surface of the dish with a layer of surgical mineral dish contained 1.5 ml of medium. oil, which allows for adequate gas exchange. Neurons The dishes were precoated with poly-L-lysine hydro- were impaled under direct vision at x 500 magnification. bromide, molecular weight 3.0 to 7.0 X lo4 (Sigma), as Microelectrodes were filled with 4 M potassium acetate previously described (Banker and Cowan, 1977) except and had impedances between 40 and 80 megohms. A that a solution of only 0.01% poly-L-lysine in borate standard bridge circuit was used for recording and cur- buffer was used. rent passage. The culture medium contained equal volumes of Ham’s GLU, ASP, and DGG were delivered by microperfusion F-12 and Dulbecco’s Modified Eagle’s Medium supple- (Choi and Fischbach, 1981). They were dissolved in mented with an additional 120 mg/lOO ml of glucose, 5 medium identical to the bath and applied by pressure pg/ml of bovine insulin, 100 pg/ml of human transferrin, from blunt-tipped pipettes (-5 pm) located approxi- 20 nM progesterone, 100 pM putrescine, and 30 nM sele- mately 30 pm from impaled neurons. A pulse generator nium dioxide (all from Sigma). Penicillin and strepto- activating a solenoid valve (General Valve Corp., Fair- mycin (both from Gibco) were added to final concentra- field, NJ) controlled ejection duration. tions of 20 units/ml and 20 pg/ml, respectively. The final medium also contained 10% heat-inactivated human Results serum. All recent experiments have employed my own In five control cultures exposed to 95% N2/5% CG2, serum, to eliminate the variability seen with different the neurons had markedly degenerated by 6 to 8 hr (Fig. lots of commercial animal sera. However, fetal calf and 1, Al and A2). After 1 day it was impossible to delineate horse sera have supported adequate neuronal growth and outlines of neurons except for their processes, which survival in the past. This medium is almost identical to appeared swollen and vacuolated (Fig. 1, A3). Although the previously described N2 (Bottenstein and Sato, the figures show only a single microscope field followed 1979), except that it omits HEPES and contains serum. over time, they are representative of the appearance of Cultures were maintained in a humidified incubator all five control cultures; no intact neurons were seen with 5% CO2 at 36°C. After 4 to 6 days, division of non- after 6 to 8 hr. There was a delay of a few hours before neuronal cells was halted by the addition of 15 pg/ml of the effects of anoxia became obvious. This is most likely fluorodeoxyuridine and 35 pg/ml of uridine (Ransom et because some oxygen remained dissolved in the medium. al., 1977). the growth medium was never changed, and When aerobic metabolism was blocked by cyanide rather cultures were not refed. Cultures maintained in this than anoxia, the effects were manifest within 1 hr (Roth- manner usually survived 4 to 5 weeks, and this was not man, 1983). demonstrably affected by refeeding. In contrast, glia, which are also present in these hip- Ilzductiorz of anoxia. In early experiments, cultures pocampal cultures (Rothman and Cowan, 1981; Seifert were made anoxic by placing them in a home-made et al., 1981), appear relatively resistant to anoxia, at least chamber which was flushed with 95% N2/5% C02, sealed, over the period of observation. They are not easy to and returned to the incubator. More recently, a commer- distinguish in the figures because they tend to grow as a cially available incubator chamber specifically designed monolayer beneath the neurons. for this purpose has been employed (Billups-Rothenberg, As previous experiments had indicated that blockade Del Mar, CA). Dishes filled with water were placed into of transmitter release by magnesium prevented neuronal both chambers to maintain humidity as close to 100% as death in these cultures, it seemed possible that released possible. transmitter might be responsible for this phenomenon Individual microscope fields were photographed prior (Rothman, 1983). The most likely excitatory transmit- to anoxia and then relocated for photography after vary- ters used by hippocampal neurons in these cultures are ing periods of anoxia. Just prior to anoxic exposure, GLU and ASP (Rothman, 1982), which are well known medium was replaced with Earle’s Minimum Essential neurotoxins (Olney, 1978). Medium (MEM) or MEM supplemented with lop2 M y- To test the hypothesis that released GLU or ASP D-glutamylglycine (DGG; Cambridge Research Biochem- produced the neuronal loss, five cultures were treated icals LTD, Atlantic Beach, NY) as the sodium salt. with lop2 M DGG, a postsynaptic blocker of excitatory Excitatory amino acid toxicity. In these experiments, amino acids (Francis et al., 1980), and were then exposed individual microscope fields were photographed.