Neuropharmacology Vol. 3 I, No. 9, pp. 899-907,
1992
0028-3908/92$5.00+ 0.00 Pergamon Press Ltd
Printed in Great Britain
GLIOTOXIC ACTIONS OF EXCITATORY
AMINO ACIDS
R. J. BRIDGES,’C. G. HATALSKI,’ S. N. SHIM,’ B. J. CUMMINGS,~ V. VIJAYAN,~ A. KUNDI~ and C. W. COTMAN~ Departments of ‘Neurology and 2Psychobiology, Irvine Research Unit on Brain Aging, University of California, Irvine, CA 92717, U.S.A. and ‘Department of Human Anatomy, University of California, Davis, CA 95616, U.S.A. (Accepted 5 May 1992)
Summary-Cultures of neonatal Type I astrocytes of the rat were exposed to a series of excitatory amino acid analogs to identify those compounds that were gliotoxic. In addition to L-a-aminoadipate, a previously identified gliotoxin, L-homocysteate, L-serine-O-sulfate, L-a-amino&phosphonobutyrate and ~-or-amino-3-phosphono-propionate were also found to induce a sequence of degenerative events that led to the lysis of the astrocytes. Cellular injury was assessed by quantifying the activity of lactate dehydrogenase present in the surviving astrocytes. Prior to lysis, the cells went through a succession of distinctive morphological changes, the most prominent of which involved nuclear alterations. The nuclei appeared swollen, contained “pale” or “watery” nucleoplasm and exhibited a very prominent nuclear membrane and obvious nucleoli. These astrocytes appeared quite similar in appearance to the Alzheimer’s Type II astrocytes, principally associated with the pathology of hepatic encephalopathy. The nuclear anomalies, which are thought to be indicative of cellular damage and compromised function, were also produced by the endogenous transmitters L-glutamate and L-aspartate, although with time, the affected astrocytes appeared to recover and return to normal morphology, without lyzing. These findings suggest that excessive levels of excitatory amino acids may induce cellular damage to astrocytes, as well as neurons. Once damaged, the resulting reductions in astrocyte function may further contribute to CNS losses and the overall pathology attributed to the excitatory amino acids. Key words-gliotoxic,
astrocytes, excitatory amino acids, excitotoxicity, Alzheimer Type II astrocytes.
The cytotoxic properties of the excitatory amino acid (EAA) neurotransmitters (e.g. L-glutamate and L-aspartate) appear to be an underlying mechanism in the etiology of several neurodegenerative diseases, including: ischemia, hypoglycemia, epilepsy, Huntington’s disease, lathyrisms and Alzheimer’s disease (for review see Choi, 1988; Monaghan, Bridges and Cotman, 1989). While the majority of these studies have been concerned with the specific neurotoxic actions of L-glutamate or related EAA receptor agonists, little is known about their potential effects on glial cells. A notable exception to this is the previously identified gliotoxic action of L-a-aminoadipic acid (L-~-AA) (Olney, de Gubareff and Collin, 1980; Lund Karlsen, Pedersen, Schousboe and Langeland, 1982; Huck, Grass and Hatten, 1984a; Huck, Grass and Hortnagel, 1984b). This dicarboxylic amino acid, which is one methylene group longer than L-glutamate, has been shown to be a potent and stereo-selective toxin when added to cultured astrocytes. Owing to the obvious chemical similarities between L-a-AA and a number of EAA excitotoxins, the pharmacology of this gliotoxic action was examined. Considering the close functional associations between neurons and glia, it is important to assess both the direct impact of neurotoxic conditions on glial cells and the potential neurotoxic consequences of pathologically impaired glia.
Glial cells, and in particular astrocytes, play an essential role in regulating the chemical environment that surrounds neurons and synapses (for review see Kimelberg, 1983; Hertz and Schousboe, 1986; Hosli, Schousboe and Hosli, 1986; Vernadakis, 1988). In EAA transmission, astrocytes are believed to participate in terminating the excitatory signal of L-glutamate by rapidly removing the transmitter from the synaptic cleft through a high affinity transport system (Balcar and Johnston, 1972; Schousboe and Hertz, 1981). In this capacity, transport may play a key role in maintaining a proper balance between that amount of glutamate required for excitatory transmission and the accumulation of excessive amounts of the transmitter that might induce excitotoxic injury. Astrocytes are also thought to participate in the recycling of the pool of L-glutamate transmitter through the “glutamine cycle” (Hamberger, Chiang, Nylen, Scheff and Cotman, 1979; Shank and Aprison, 1981). In this pathway, L-glutamate, taken up into astrocytes, is converted to L-glutamine by the action of glutamine synthetase. The L-glutamine, which appears to be electrophysiologically inactive at EAA receptor gated ion channels, is then transferred back to the presynaptic terminal, where it is reconverted to L-glutamate by the neuronal enzyme glutaminase. Thus, the inability to properly regulate the levels of extracellular L-glutamate could compromise 899
R. J. Btunoss et al.
900
excitatory transmission, disrupt the normal recycling of the transmitter, as well as lead to the accumulation of excitotoxic levels of L-glutamate. If the accumulation of these excitotoxic compounds also has a negative impact on the surrounding astrocytes, a cumulative cycle of damage could arise which increases neuronal vulnerability and the extent of pathological injury. In the present report, a group of EAA analogs, including L-a-AA, is identified that produce a series of stereotypic morphological alterations, particularly nuclear abnormalities, in cultures of neonatal Type I astrocytes that ultimately leads to the lysis of the cells. The endogenous transmitters, L-glutamate and L-aspartate, cause similar morphological changes to those observed during the early phases of this gliotoxic response but without the subsequent glial death. These findings suggest that excessive levels of neurotoxins may have a deleterious effect on astrocytes, that could potentially contribute to the overall pathology of an excitotoxic insult. METHODS
Astrocyte
culture
Primary cultures of protoplasmic astrocytes were prepared from newborn (4 day) cortex of rat and purified, as previously described (Bridges, Hatalski, Shim and Nunn, 1991a), using the procedure of McCarthy and de Vellis (1980). The cells were grown in a 1: 1 mixture of Dulbecco’s modification of Eagle’s medium and Ham’s F12 medium, buffered with a final concentration of 25mM HEPES (N2-hydroxyethylpiperazine-N’-2-ethane-sulfonic acid)/ NaOH and 14.3 mM sodium bicarbonate (pH 7.4). The medium was supplemented with 15% fetal bovine serum (FBS, GIBCO, Long Island, New York), during the initial plating (2 days) and then maintained with a supplement of 10% FBS for the duration of the studies. The cells were cultured in 75 cm* flasks and kept at 37°C in a humid atmosphere of 5% CO,. Media was replenished every 34 days. When 60-70% confluent, the cells were shaken (280 rpm, 24 hr) to remove oligodendrocytes, Type II astrocytes and microglia. Two or three days later the astrocytes were harvested by trypsinization and plated into 12-well plates. These secondary cultures were maintained in a similar environment and allowed to reach confluency (14-21 days), prior to being used in the toxicology studies. Astrocyte
toxicity
Confluent secondary astrocyte cultures of similar age (14-21 days after secondary plating) were given fresh media 2 hr prior to the addition of the EAA analogs. All of the compounds were prepared as concentrated solutions in media (pH 7.4) and added to the cultures in a single 100-200 ~1 aliquot at time zero. In this respect, the concentrations reported represent the initial concentrations to which the
astrocytes were exposed. The astrocytes were examined at regular intervals (2,4, 6, 12, 24, 36 and 48 hr) and photographed with an Olympus IMT-2 inverted phase-contrast microscope. The extent of survival of cells was quantified by determining the activity of lactate dehydrogenase (LDH), present in the surviving cells. At the times indicated, the culture media was removed and the astrocytes were lyzed in media (without FBS), containing 0.1% Triton X-100. The activity of LDH was assayed, as described by Koh and Choi (1987) and compared to that of control cultures. In all experiments a minimum of two preparations were observed and assayed. Materials
N-Methyl-D-aspartate (NMDA), a-amino-3hydroxy-5-methyl-isoxazole4propionate (AMPA), L-2-amino-3-phosphonoproprionate (L-AP3), L-2amino-4-phosphonobutyrate (L-AP4) and quisqualate (QA) were obtained from Tocris (Bristol, England). Kainic acid (KA) and L-a-aminoadipic acid, as well as other general reagents used, were obtained from Sigma (St Louis, Missouri). Tissue culture media was obtained from Gibco (Long Island, New York). RESULTS
Consistent with previous experiments, L-a-AA proved to be a potent gliotoxin. Using a strategy similar to that developed to assess neuronal injury, the extent of astrocyte lysis could be quantified enzymatically by determining the activity of LDH, retained in the surviving cells (Koh and Choi, 1987; Bridges et al., 1991a). When compared to microscopic evaluation, this enzymatic assay was found to be a more sensitive and accurate indicator of cellular lysis than visual observation. Total activity of LDH (that which is released into extracellular media, plus that which was retained in the surviving cells) was determined in a number of experiments (data not shown) and did not change during the course of the exposure to the toxins. As shown in Fig. 1, the gliotoxic activity of L-a-AA exhibited a concentration-dependence, with an LD,, of about 0.6mM 48 hr following a single addition of the amino acid to the culture media (Fig. 1). The curve of the dose-response relationship of this gliotoxic activity was quite steep, as the extent of glial injury produced by 0.4 and 0.75 mM changed from 74+ 10% to 24 f 6% survival, respectively (mean + SEM). Although this narrow range of effective concentrations may suggest a threshold effect, it may also indicate that the lysis of the first cells contributed to the injury of the remaining cells. This could, for example, be attributable to the release or loss of regulation of degradative enzymes that, when emptied into the extracellular space, initiate a pathological chain reaction that rapidly leads to the lysis of the remaining cells.
Gliotoxic actions of exitatory amino acids
120 r
0.1 L-a-Aminoadipate
1
(mM)
Fig. 1. Gliotoxic action of L-a-aminoadipate, as a function of concentration. Astrocyte cultures were exposed to the indicated amount of the compounds for 48 hr. The extent of cell lysis was quantified by determining the amount of activity of LDH remaining in the surviving cells and is reported as mean percentage of control + SEM. The figure includes the cumulative data from 4 different dose-response curves, each using a different astrocyte preparation. The number of determinations (i.e. individual culture wells) at each concentration is shown (n).
Pharmacology
of gliotoxic amino acids
Like the majority of glutamate analogs that exhibit activity at EAA receptors, L-a-AA is an a-amino-
diacid in which the acidic groups are separated by a relative distance of 2-4 carbon units. These chemical similarities raised questions regarding the structure-function relationship between other EAA analogs and their potential gliotoxic activity. Table I summarizes the effect that a variety of these analogs had on confluent cultures of astrocytes of the rat. The toxic action of the compounds was assessed 48 hr after a single exposure to an initial concentration of 1 mM. In addition to the expected activity of L-a-AA, the sulfur-containing amino diacids L-homocysteate, L-serine-O-sulfate and ~-cysteine-sulfinate were found to be exceptionally gliotoxic. Structurally, these amino acids are one methylene group shorter than L-a-AA and possess a sulfonic acid group in place of its 6carboxyl group. L-Cysteate, which is one methylene group shorter than r.-homocysteate, did not exhibit any gliotoxic activity. At a similar concentration, the substitution of a phosphate group for the terminal carboxyl group resulted in a loss of toxic activity, in the instance of L-serine-O-phosphate. In contrast, the phosphonic amino acids, D,L-AP~ and L-AP4 induced a substantial lysis of the astrocytes. Loss of L-a-AA’s amino group, as tested with a-ketoadipate, resulted in the loss of gliotoxic activity. A similar lack of activity was also observed with a-keto analogs of glutamate and aspartate, suggesting that metabolism of these compounds by transamination does not lead to the production of a gliotoxic derivative.
901
In contrast to L-a-AA, D-Cf-aminoadipate was much less effective as a gliotoxin, confirming the stereospecificity first reported by Olney et al. (1980). A similar requirement was observed in the present study, where L-homocysteate and L-AW were considerably more toxic than their respective D-enantiomers. As previous studies indicated that L-a-AA must first be transported into the cells, prior to the cytotoxic response (Huck et al., 1984b), this stereospecificity could reflect the enantioselectivity of either the mechanism directly involved in the toxicity or the transport system responsible for its intracellular accumulation. The well known EAA receptor specific agonists NMDA, KA and AMPA were also examined as potential gliotoxins. In contrast to their potent neurotoxic activity, no cellular damage was detected after 48 hr of exposure to concentrations of up to 2 mM (Table 1). Similarly, QA did not appear to induce any glial loss when it was screened at initial concentrations of 1 mM. Larger concentrations (2-3 mM, data not shown) of this EAA did, on occasion, produce a substantial lysis of the astrocytes, although the activity was considerably more variable than that observed with the other compounds. This action is consistent with the recent study of Haas and Erdo (1991) who described a transient vulnerability of Table I. Effects of excitatory amino acid analogs on cultures of neonatal rat astrocytes’ Astrocyte survival (% of LDH retained)
Compound L-a-Aminoadipate I mM (14) D-or-Aminoadipate I mM (8) L-Glutamate I mM (9) D-Glutamate I mM (26) L-Aspartate 1mM (7) D-Aspartate I mM (17) Quisqualate 1mM (26) Glycine I mM (6) Surfore containing amino acid3 L-Serine-O-sulfate I mM (7) L-Cysteine-sultinate I mM (8) L-Homocysteate I mM (9) D-Homocysteate I mM (7) L-Cysteate I mM (IS) Taurine I mM (7) Phosphate
containing
I mM (8)
EAA
agonists
receptor-spectfk
a-Keto
7*3 13*9 25+11 93 + 6 105 * 9 107 + 8
amino acids
L-Serine-U-phosphate L-AW I mM (6) D-AP~ I mM (6) D,L-AP3 I mM (12) N-Methyl-D-aspartate Kainate I mM (7) Kainate 2 mM (6) AMPA 2 mM (6)
16rt8 80 f 8 97+ II 75* 15 93 + 6 92+_ 14 125 + I6 101 * IO
I mM (7)
100~10 16220 101 + 12 45+15 92* IO 95 f 9 107+2 102*5
aci&
a-Ketoadipate 1 mM (1 I) a-Ketoglutarate I mM (IO) Oxaloacetate I mM (7)
101 -+ 11 92 * 7 97* I2
Cultures of neonatal rat astrocytes were exposed to a single addition of the EAA analogs (initial concentration indicated) as described in Methods. After 48 hr of continuous exposure, the extent of cellular lysis was assessed by quantitating the LDH activity retained in the surviving cells. The values, which are. reported as mean percent of control +_SD, were based on 6-26 (n) determinations (i.e. individual culture wells) from 2-4 different astrocyte preparations.
cultured stration, indiiated at some receptor.
astrocytes to QA. The present demonhowever, that AMPA was not glioioxic, that the action of QA was most probably point other &an the A~PA~QA EAA
DISCUSSION
fn the present study, the authors have begun to i~v~t~gate the ghotoxic property of L-~-AA by first c~~~ac~e~~~~ the pb~a~lo~~l safe of gliotoxic amino acids. Toward this goal, a wide range Morphdogkd changes during t?xposwe of EAA analogs have been screened in cultures of astrocytes of the neonatal rat, using the loss of the Prior to their eventual lysis, the astrocytes exposed cytosohc enzyme LDH to quantify the extent of the to W-AA, ~*hom~y~teate or L-setine-U-sulfa& @ai damage. As expected, t-cr -AA was four& to be underwent a succession of very ~stin~~ve morphoIogicat change.s A time-course of these changes is a potent gliotoxin. Its eytotoxic action was concennation-decedent and exhibited an X.X& of about illust~~cd se~u~~~ally in Figs 2 and 3. The earliest and most prominent alterations, observed in the 0.6 mM, as followed with the LDH protocol. Of the EAA analogs tested, r_-homocysteate, L-cysteine-sulfiastrocytes, involved abnormalities in nuclear marnate, tserine-Q-sulfate and L-AP4 were also found pholugy. The nuclei appeared swollen, contained to exhibit a cytotoxic effect on the euitured astro*‘pale” or “watery” ~ncleopl~m and exhibited a very cytes. The action of ~-hom~yste~c acid is of particuprominent nuclear membrane and obvious nucleoli. far interest because recent evidence suggests that it These chartges, which were observable as early as 6 hr after the addition of the ghotoxic compounds (Fig. 2), may be an endogenous transmitter and excj~o~ox~~ in the neocartex of the rat (Do, Herrling, Streit, were induced most rapidly by those acidic amino Turski and Cuenod, 1986; Knopfel, Zeise, Cuenod acids that proved to be the most toxic. When exposed and ~ieglgans~r~r~ 1987; Olney, Price, Salles, to smaller ~~~~trat~o~s of the g~~oto~~~, &m&r ~ab~~ere, Ryerson, Mahan, Frierdich and Samson, m&i were also observed, even though the ~s~r~yt~~ 1987; Kim, Koh and Choi, 1987). Consistent with did not lyse within the 48 hr experimental interval, these results, s~~taneo~ injections of o+homoThe nuclear mo~hology~ induced by the gli~toxic cysteate have been shown to produce a loss of amino acids, appeared to he very similar to a patho* non-neuronal cells in the retinas and brains of infant logical form of glia that is characteristically associmice, (Olney, Ho and Rhea, 1971). Similarly, D-t-hoated with hepatic e~~h~opa~~ (AIzheimer’s Type rn~~s~e~~acid and D,L-AP~ have been &own to be Xf (AZ IX) astr~~tcs~ Norenberg, $987) In the pretoxic to the glioma C5 ceil line of the rat, although sent e~~~rnen~~ continued exposure to the gliotoxic amino acid Ied to an increase in the prominence of the these cells appeared to be less sensitive to the toxin than the cultures used in the present investigation nuclear abnormalities. These changes were followed by progressive swelling of the cell body and mem* (Kato, Higashida, Higuchi, Hatakenaka and Negishi, 1984). Qf the compounds tested in this study, all of brane blebbing that eventually terminated with the those id~~~ad as toxic to the astrocytes were LWXtysis of the c&t, During the latter stages of the toxic amino diacids, whife the ~-e~a~tjorne~ of homoresponse, the nu&i changed from their swollen AZ cysteate and AP4 had uo visible e&t on the c&s B-type appearance, taking on a shrunken, pyknotic when tested at similar concentrations. This enantioseappearance (Figs 3D and F). lectivity is consistent with earlier studies identifying A different course of events was observed when the astracytes were exposed to high levels of the D-~-AA as ~lioto~cally inactive (also confirmed in the present study) and suggests the preboils deendogenous exettatory t~~srnitt~~s, L-glutamate and t-aspartate. In these onsets, the astrocytes apscribed in r&o actions of bomocysteat~ and AP4, cart peared to undergo the initial sequence of morphobe ascribed to the ~~n~tiorn~~s~ logical changes induced by the gliotoxins, particularly Although a number of the compounds, identified the nuclear anomalies (Fig. 2B) but then returned to as gliotoxins, are well known FAA agonists (e.g, what seemed a normal form, without further swelling L-homocysteic acid, t.-serine-O-sulfate), the lack of or fysis (Fig. 3B). Thus, fX4kr after exposure to an effect by NMDA, KA and AMPA suggests that 1 mM ~~glutamate~ the rmclei took on the appearance the process does not involve the EAA reeptors that of the AZ II astrocytes, ch~cte~stic of those cehs figure prorn~~en~y in exc~totoxic~m~~at~ neuronal that would have eventually been Iyzed by L-~-AA or death {for review see Choi, 1988). Of the three t-homocysteate (Fig. 3). Instead of dying, however, excitotoxins, KA is of particuiar note, as studies have the cells returned to an apparently normal mordemonstrated the presence of a glutamate/kainatephology by 36-48 hr. These findings suggest that the activated Na+ channel on astrocytes (Bowman and ~~~~~ of AZ $1 ~r~~o~~y is no% 8n inewrsK~mal~rg, 1984; Kett~rn~~ and Sehachner, 1985; ible step that signals the eventuai death of the astroSonthetmer, K~t~nrna~n~ Rackus, and Schachner* cyte. ~u~he~~~, as the changes were brought 1988; Rackus, Ketfa~ann and Schacbner, 1989), about by excessive levels of the endogenoua transmitFurther evidence that this receptor does not mediate ter L-glutamate, the pathological appearance of these the gliotoxic response is provided by the action of cells in vitromay he indicative of areas undergoing an L-homocysteic acid, which is only a weak agonist at exeitotox~c insult. this glial receptor_ yet prov& to be one of the most
Fig. 2. Photomicrographs, depicting the L-cl-aminoadipate (D) on confluent cultures (1 mM initial concentration) of the EAA become observable. Control
early effect of L-glutamate (B), r_-homocysteate (C) and of astrocytes of neonatal rat. Six hours after a single addition analog, the stereotyped alterations in nuclear morphology cells are shown in (A). (Bar inset = 100pm.)
903
Fig. 3. Photomicrographs, depicting the long term effect of the L-glutamate (B), L-a-aminoadipate (C) and (D) and L-homocysteate (E) and (F) on confluent cultures of astrocytes of neonatal rat. The cells shown, which are from the same cultures presented in Fig. 2, were exposed to an initial concentration of 1 mM EAA analog. The astrocytes appeared to return to a normal morphology 36 hr after the addition of L-glutamate (B), control cells are shown in (A). In contrast to the action of L-glutamate, the cellular swelling and nuclear abnormalities, induced by L-a-aminoadipate and L-homocysteate continued (C and E, respectively at 24 hr) and eventually led to the lysis of the astrocytes (D and F, respectively at 48 hr). (Bar inset = 100 pm.)
904
Gliotoxic actions of exitatory amino acids
potent gliotoxins. The lack of an involvement of an extracellular receptor also supports the gliotoxic model of Huck et al., who proposed that this L-~-AAinduced cell death is dependent on the intracellular accumulation of the amino acid (Huck et al., 1984b). This conclusion was based on immunocytochemical and autoradiographic studies that demonstrated the sodium-dependency of both ~-x-AA-mediated toxicity and transport and the inability of TTX to block the glial lysis. This would suggest that, in addition to those properties directly responsible for the pathological damage, the gliotoxic L-x-amino diacids must also be substrates of a transport system and accumulate intracellularly. The fact that each of these steps could have its own pharmacological requirement may explain why two structurally very similar acidic amino acids, such as L-serine-O-sulfate and L-serineO-phosphate, exhibited very different physiological effects. Studies by Murphy et al. (Murphy, Miyamoto, Sastre, Schaar and Coyle, 1989; Murphy, Schnaar and Coyle, 1990) have identified an alternative mechanism by which EAA analogs induce a cytotoxic effect that is related to the ability of these compounds to inhibit transport of cystine. As a result of attenuated uptake of cystine, the intracellular levels of glutathione are reduced and the cells succumb to oxidative stress. Although this mechanism is thought to contribute to the EAA-mediated cell loss observed in neuroblastoma cells (Nl8-RE- 105) and immature cortical neurons (E17: 24-72 hr in culture), glial cells in the primary cultures were found to be insensitive (Murphy et al., 1990). Further evidence that this mechanism is not responsible for the gliotoxic activity reported in the present study, comes from the demonstration that little or no cellular lysis is observed in mixed cortical cultures, that have been depleted of glutathione with buthionine sulfoximine (Bridges, Koh, Hatalski and Cotman, 199lb). Thus, even though excessive levels of EAA analogs might act to reduce transport of cystine in these astrocytes, it appears very unlikely that the primary mechanisms underlying EAA-mediated gliotoxicity are related to this pathological scheme. Prior to their eventual lysis, the astrocytes underwent a series of distinct morphological changes, the most prominent of which was marked by the appearance of nuclei that were swollen, contained “pale” or “watery” nucleoplasm and exhibited a very prominent nuclear membrane, with obvious nucleoli (Figs 1B and D). These same nuclear changes were also observed after the administration of the endogenous transmitters L-glutamate and L-aspartate. In contrast to the gliotoxins, however, the astrocytes exposed to these amino acids appeared to return to a normal morphology, without lyzing. Thus, L-glutamate may initially mimic the potentially pathological action of L-LX-AA,although the process is terminated before enough damage has occurred to result in the lysis of the cells. As it appears that the death of
905
the astrocyte is dependent upon the accumulation of the toxic acidic amino acids, the termination of the toxic response may be attributable simply to the metabolism of the excess L-glutamate and L-aspartate. Consistent with this, studies with the C6 glioma cell line indicate that the sensitivity to glutamate- but not L-u-AA-mediated toxicity, could be reduced by chemically inducing activity of glutamine synthetase (Kato et al., 1984). Thus, the gliotoxic response could be avoided by rapidly metabolizing the L-a-amino diacids, thereby preventing their accumulation. The morphological changes observed in response to the gliotoxic a-amino-diacids, as well as to Lglutamate, appear to be similar to AZ II astrocytes that are recognized as a pathological hallmark of hepatic encephalopathy (for review see Norenberg, 1981). Originally named by Waggoner and Malamud (1942), AZ II astrocytes were identified by von Hosslin and Alzheimer (1912) and later associated with hepatic encephalopathy by Adams and Foley (1953). Although their appearance has generally been attributed to high levels of ammonia, the present results suggest that excessive levels of ~-x-amino diacids, including L-glutamate, may also represent a contributing factor. The amount of L-glutamate used in these experiments was less than the levels of ammonia previously used to induce these alterations in vitro (4 days of exposure to 10 mM ammonia, compared to 24-48 hr to 1 mM L-glutamate; Norenberg; 1987). Thus the appearance of the abnormal nuclei may not necessarily be a specific marker of ammonia-mediated damage but, instead, represent a more general marker of metabolic imbalance or cellular pathology. In previous studies, AZ II astrocytes produced by long term exposure of cultures to ammonia (14 days) have been shown to exhibit: (i) a reduction in their ability to transport a-aminobutyric acid (GABA) and L-glutamate, (ii) an impaired capacity to clear extracellular potassium and (iii) a decrease in the intracellular levels of adenosine triphosphate (ATP) (Norenberg, Moses, Papendick and Norenberg, 1985; Norenberg, 1987). If the observed morphological changes are indicative of this damage, the present results would suggest that astrocyte function may also be comprised by excessive levels of L-glutamate. Such an effect may be of particular significance in excitotoxic injury, where the transport and metabolism of L-glutamate by astrocytes are thought to be important in regulating its extracellular concentration and preventing glutamate-mediated neuronal loss (see, for example, Sugiyama, Brunori, and Mayer, 1989). Although initially protective, the continued accumulation of L-glutamate by astrocytes may also compromise their function and reduce the ability of these cells to transport or even maintain their intracellular pools of L-glutamate. A cumulative cycle of damage may ensue, with the end result being an increase in the levels of extracellular L-glutamate and subsequent neuronal loss. This cascade of
906
R. J. BRIIffiEs
pathological events highlights the intricate balance that must exist in the EAA transmission system and illustrates how each component, if not functioning properly, has the potential to cont~bu~ to excitotoxic pathology and neuronai loss. Acknowledgements-This
work was supported in part by
NIH grant NS27056 and
NIA L.E.A.D. award AG07918.
The authors also wish to thank J. S. Kahle for her assistance in the preparation of this manuscript. REFERENCES
Adams R. D. and Foley J. M. (1953) The neurological disorder associated with liver disease. Ass. Res. Nerve Ment. Dis. Proc. 32: 198-237. Backus K. H., Kettenmann H. and Schachner M. (1989) Pha~acolo~cal characterization of the glutamate receptor in cultured astrocytes. J. Neurosei. Res. 22s 274282. Balcar V. S. and Johnston G. A. R. (1972) Glutamate uptake by brain slices and its relation to the depolarization of neurons by acidic amino acids. J. Neurobiof. 3: 295-30 1. Bowman C. L. and Kimelberg H. K. (1984) Excitatory amino acids directly depolarize rat brain astrocytes in primary culture. Nature 311: 656-657. Bridges R. J., Hatalski C. G., Shim S. N. and Nunn P. B. (1991a) Gliotoxic properties of the Lathyrus excitotoxic B-N-oxalyl-L-a$-diaminopropionic acid (fi-L-ODAP). Brain Res. 561: 262-268. Bridges R. J., Koh J., Hatalski C. G. and Cotman C. W. (1991b) Increased excitotoxic vulnerability of cortical cultures with reduced levels of ~u~thione. Eur. J. Pharm. 19% 199-200. Choi D. W. (1988) Glutamate neurotoxicity and diseases of the nervous system. Neuron 1: 623-634. Do K. Q,, Herrling P. L., Streit P., Turski W. A. and Cuenod M. (1986) In vitro release and electrophysiological effects in situ of homocysteic acid, an endogenous N-methyl-D-aspartic acid agonist, in the mammalian striatum. J. Neurosci. 6: 222&2234. Haas J. and Erdo L. E. (1991) Quisqualate-induced excitotoxic death of glial cells: transient vulnerability of cultured astrocytes. Gliu 4: 11l-l 14. Hamberger A. C., Chiang G. H., Nylen E. S., Scheff S. W. and Cotman C. W. (1979) Glutamate as a CNS transmitter. I. Evaluation of glucose and glutamine as precursors for the synthesis of p~ferentially released glutamate. Brain Res. 7: 513-530. Hertz L. and Schousboe A. (1986) Role of astrocytes in compartmentation of amino acid and energy metabolism. In: Astrocytes: Biochemistry, Physiology, and Pharmacology of Astrocytes (Federoff S. and Vernadakis A., Eds), pp. 179-208. Academic Press, New York. Hosli E., Schousboe A. and Hosli L. (1986) Amino acid uptake. In: Asirocytes: Biochemistry, Physiofogy, and Pharmacofogy of Astrocyfes (Federoff S. and Vernadakis A., Eds), pp. 133-153. Academic Press, New York. von Hosslin C. and Alzheimer A. (1912) Ein beitrag zur klinic und pathologischen anotamie der WestphalStrumpellschen pseudosklerose. 2. Neural. Psych&. 8: 183-209. Huck S., Grass F. and Hatten M. E, (1984a) Gliotoxic effects of a-aminoadipic acid on monolayer cultures of dissociated postnatal mouse cerebellum. Neuroscience 12: 783-791. Huck S., Grass F. and Hortnagl H. (1984b) The glutamate analogue a-aminoadipic acid is taken up by astrocytes
et al.
before exertine its stliotoxic effect in vitro. J. Neurosci. 4: 2650-2657. Kato S., Higashida H., Higuchi Y., Hatakenaka S. and Net&hi K. (1984) Sensitive and insensitive states of &ured glioma cells to glutamate damage. Brain Res. 303: 365-373. Kettenmann H. and Schachner M. (1985) Pharmacological properties of GABA-, glutamate-, and aspartate-induced depolarizations in cultured astrocytes. J. Neurosci. 5: 32953301. Kim J. P., Koh J. Y. and Choi D. W. (1987) L-Homocysteate is a potent neurotoxin on cultured cortical neurons. Bruin Res. 4%~ 103-110. Kimelberg H. (1983) Primary astrocyte culture-a key to astrocvte function. Cell. molec. Neurobiol. 3: 1-16. Knopfel-T., Zeise M. L., Cuenod M. and Zieglgansberger W. (1987) L-Homocysteic acid but not L-glutamate is an endogenous N-methyl-o-aspartic acid receptor preferring agonist in rat neocortical neurons in vitro. Neurosci. I&t. 81: 188-192. Koh J. and Choi D. W. (1987) Quantative dete~ination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J. Neuroxi. Meth. 20: 83-90. Lund Karlsen R., Pedersen 0. O., Schousboe A. and Langeland A. (1982) Toxic effects of D,L-a-aminoadipic acid on Muller cells from rats in vivo and cultured cerebral astrocytes. Expl Eye Res. 35: 305-3 11. McCarthv K. D. and de Vellis J. (1980) Preparation of separate astroglial and oligodendroghal cell cultures from rat cerebral tissue. J. Cell Biol. 85: 890-902. Monaghan D. T., Bridges R. J. and Cotman C. W. (1989) The excitatory amino acid receptors: their classes, pharmacology, and distinct properties in the function of the central nervous system. A. Rev. Pharmac. Toxic. 29~ 365-402. Murphy T. H., Schnaar R. L. and Coyle J. T. (1990) Immature cortical neurons are uniquely- sensitive to glutamate toxicity by inhibition of cystine uptake. FASEB J. 4: 1624-1633. Murphy T. H., Miyamoto M., Sastte A., Schnaar R. L. and Covle J. T. (1989) Glutamate toxicity in a neuronal &I line involves inhibition of cystine transport leading to oxidative stress. Neuron 2: 1547-1558. Norenberg M. D. (1981) The astrocyte in liver disease. Adv. cell. Neurobiol. 2: 303-352. Norenbern M. D. (1987) The role of astrocytes in hepatic encephsopathy. ‘Neurachem.Path. 6: 13-33. _ Norenberrz M. D.. Mozes L. W.. Pawndick R. E. and Norenb&g L. 0: B. (1985) Elf&t of’ammonia on glutamate, GABA and rubidium uptake by astrocytes. Ann. Neural. 18: 149. Olney J. W., de Gubareff T. and Collins J. F. (1980) Stereospecificity of the gliotoxic and anti-neurotoxic action of alpha-aminoadipate. Neurosci. Left. 19r 277-282. Olney J. W., Ho 0. L. and Rhee V. (1971) Cytotoxic effects of acidic and sulphur containing amino acids on the infant mouse central nervous system. Expr Brain Res. 14: 61-76. Olney J. W., Price M. T., Salles K. S., Labruyere J., Ryerson R., Mahan K., Frierdich G. and Samson L. (1987) L-Homocysteic acid: An endogenous excitotoxic ligand of the NMDA receptor. Brain Res. B&f. 19: 597-602.
Schousboe A. and Hertz L. (1981) Role of astroglial cells in glutamate homeostasis. Adv. Biochem. Psychophormat. 27: 103-113. Shank R. P. and Aprison M. H. (1981) Present status and significance of the glutamine cycle in neural tissue. I.19 Sci. 28: 837-842.
Gliotoxic actions of exitatory amino acids Sontheimer H., Kettenmann H., Backus K. H. and Schachner M. (1988) Glutamate opens Na + /K + channels in cultured astrocytes. Gliu 1: 328-336. Sugiyama K., Brunori A. and Mayer M. L. (1989) Glial uptake of excitatory amino acids influence neuronal surviva1 in cultures of mouse hippocampus. Neuroscience 32: 779-79
1.
907
Vemadakis A. (1988) Neuron-glia interrelations. Int. Rev. NeurobioI. 30: 149-224.
Waggonner R. W. and Malamud N. (1942) Wilson’s disease in the light of cerebral changes following ordinary acquired liver disorders. J. Nerve. Ment. Dis. 96: 410-423.
1992
0028-3908/92$5.00+ 0.00 Pergamon Press Ltd
Printed in Great Britain
GLIOTOXIC ACTIONS OF EXCITATORY
AMINO ACIDS
R. J. BRIDGES,’C. G. HATALSKI,’ S. N. SHIM,’ B. J. CUMMINGS,~ V. VIJAYAN,~ A. KUNDI~ and C. W. COTMAN~ Departments of ‘Neurology and 2Psychobiology, Irvine Research Unit on Brain Aging, University of California, Irvine, CA 92717, U.S.A. and ‘Department of Human Anatomy, University of California, Davis, CA 95616, U.S.A. (Accepted 5 May 1992)
Summary-Cultures of neonatal Type I astrocytes of the rat were exposed to a series of excitatory amino acid analogs to identify those compounds that were gliotoxic. In addition to L-a-aminoadipate, a previously identified gliotoxin, L-homocysteate, L-serine-O-sulfate, L-a-amino&phosphonobutyrate and ~-or-amino-3-phosphono-propionate were also found to induce a sequence of degenerative events that led to the lysis of the astrocytes. Cellular injury was assessed by quantifying the activity of lactate dehydrogenase present in the surviving astrocytes. Prior to lysis, the cells went through a succession of distinctive morphological changes, the most prominent of which involved nuclear alterations. The nuclei appeared swollen, contained “pale” or “watery” nucleoplasm and exhibited a very prominent nuclear membrane and obvious nucleoli. These astrocytes appeared quite similar in appearance to the Alzheimer’s Type II astrocytes, principally associated with the pathology of hepatic encephalopathy. The nuclear anomalies, which are thought to be indicative of cellular damage and compromised function, were also produced by the endogenous transmitters L-glutamate and L-aspartate, although with time, the affected astrocytes appeared to recover and return to normal morphology, without lyzing. These findings suggest that excessive levels of excitatory amino acids may induce cellular damage to astrocytes, as well as neurons. Once damaged, the resulting reductions in astrocyte function may further contribute to CNS losses and the overall pathology attributed to the excitatory amino acids. Key words-gliotoxic,
astrocytes, excitatory amino acids, excitotoxicity, Alzheimer Type II astrocytes.
The cytotoxic properties of the excitatory amino acid (EAA) neurotransmitters (e.g. L-glutamate and L-aspartate) appear to be an underlying mechanism in the etiology of several neurodegenerative diseases, including: ischemia, hypoglycemia, epilepsy, Huntington’s disease, lathyrisms and Alzheimer’s disease (for review see Choi, 1988; Monaghan, Bridges and Cotman, 1989). While the majority of these studies have been concerned with the specific neurotoxic actions of L-glutamate or related EAA receptor agonists, little is known about their potential effects on glial cells. A notable exception to this is the previously identified gliotoxic action of L-a-aminoadipic acid (L-~-AA) (Olney, de Gubareff and Collin, 1980; Lund Karlsen, Pedersen, Schousboe and Langeland, 1982; Huck, Grass and Hatten, 1984a; Huck, Grass and Hortnagel, 1984b). This dicarboxylic amino acid, which is one methylene group longer than L-glutamate, has been shown to be a potent and stereo-selective toxin when added to cultured astrocytes. Owing to the obvious chemical similarities between L-a-AA and a number of EAA excitotoxins, the pharmacology of this gliotoxic action was examined. Considering the close functional associations between neurons and glia, it is important to assess both the direct impact of neurotoxic conditions on glial cells and the potential neurotoxic consequences of pathologically impaired glia.
Glial cells, and in particular astrocytes, play an essential role in regulating the chemical environment that surrounds neurons and synapses (for review see Kimelberg, 1983; Hertz and Schousboe, 1986; Hosli, Schousboe and Hosli, 1986; Vernadakis, 1988). In EAA transmission, astrocytes are believed to participate in terminating the excitatory signal of L-glutamate by rapidly removing the transmitter from the synaptic cleft through a high affinity transport system (Balcar and Johnston, 1972; Schousboe and Hertz, 1981). In this capacity, transport may play a key role in maintaining a proper balance between that amount of glutamate required for excitatory transmission and the accumulation of excessive amounts of the transmitter that might induce excitotoxic injury. Astrocytes are also thought to participate in the recycling of the pool of L-glutamate transmitter through the “glutamine cycle” (Hamberger, Chiang, Nylen, Scheff and Cotman, 1979; Shank and Aprison, 1981). In this pathway, L-glutamate, taken up into astrocytes, is converted to L-glutamine by the action of glutamine synthetase. The L-glutamine, which appears to be electrophysiologically inactive at EAA receptor gated ion channels, is then transferred back to the presynaptic terminal, where it is reconverted to L-glutamate by the neuronal enzyme glutaminase. Thus, the inability to properly regulate the levels of extracellular L-glutamate could compromise 899
R. J. Btunoss et al.
900
excitatory transmission, disrupt the normal recycling of the transmitter, as well as lead to the accumulation of excitotoxic levels of L-glutamate. If the accumulation of these excitotoxic compounds also has a negative impact on the surrounding astrocytes, a cumulative cycle of damage could arise which increases neuronal vulnerability and the extent of pathological injury. In the present report, a group of EAA analogs, including L-a-AA, is identified that produce a series of stereotypic morphological alterations, particularly nuclear abnormalities, in cultures of neonatal Type I astrocytes that ultimately leads to the lysis of the cells. The endogenous transmitters, L-glutamate and L-aspartate, cause similar morphological changes to those observed during the early phases of this gliotoxic response but without the subsequent glial death. These findings suggest that excessive levels of neurotoxins may have a deleterious effect on astrocytes, that could potentially contribute to the overall pathology of an excitotoxic insult. METHODS
Astrocyte
culture
Primary cultures of protoplasmic astrocytes were prepared from newborn (4 day) cortex of rat and purified, as previously described (Bridges, Hatalski, Shim and Nunn, 1991a), using the procedure of McCarthy and de Vellis (1980). The cells were grown in a 1: 1 mixture of Dulbecco’s modification of Eagle’s medium and Ham’s F12 medium, buffered with a final concentration of 25mM HEPES (N2-hydroxyethylpiperazine-N’-2-ethane-sulfonic acid)/ NaOH and 14.3 mM sodium bicarbonate (pH 7.4). The medium was supplemented with 15% fetal bovine serum (FBS, GIBCO, Long Island, New York), during the initial plating (2 days) and then maintained with a supplement of 10% FBS for the duration of the studies. The cells were cultured in 75 cm* flasks and kept at 37°C in a humid atmosphere of 5% CO,. Media was replenished every 34 days. When 60-70% confluent, the cells were shaken (280 rpm, 24 hr) to remove oligodendrocytes, Type II astrocytes and microglia. Two or three days later the astrocytes were harvested by trypsinization and plated into 12-well plates. These secondary cultures were maintained in a similar environment and allowed to reach confluency (14-21 days), prior to being used in the toxicology studies. Astrocyte
toxicity
Confluent secondary astrocyte cultures of similar age (14-21 days after secondary plating) were given fresh media 2 hr prior to the addition of the EAA analogs. All of the compounds were prepared as concentrated solutions in media (pH 7.4) and added to the cultures in a single 100-200 ~1 aliquot at time zero. In this respect, the concentrations reported represent the initial concentrations to which the
astrocytes were exposed. The astrocytes were examined at regular intervals (2,4, 6, 12, 24, 36 and 48 hr) and photographed with an Olympus IMT-2 inverted phase-contrast microscope. The extent of survival of cells was quantified by determining the activity of lactate dehydrogenase (LDH), present in the surviving cells. At the times indicated, the culture media was removed and the astrocytes were lyzed in media (without FBS), containing 0.1% Triton X-100. The activity of LDH was assayed, as described by Koh and Choi (1987) and compared to that of control cultures. In all experiments a minimum of two preparations were observed and assayed. Materials
N-Methyl-D-aspartate (NMDA), a-amino-3hydroxy-5-methyl-isoxazole4propionate (AMPA), L-2-amino-3-phosphonoproprionate (L-AP3), L-2amino-4-phosphonobutyrate (L-AP4) and quisqualate (QA) were obtained from Tocris (Bristol, England). Kainic acid (KA) and L-a-aminoadipic acid, as well as other general reagents used, were obtained from Sigma (St Louis, Missouri). Tissue culture media was obtained from Gibco (Long Island, New York). RESULTS
Consistent with previous experiments, L-a-AA proved to be a potent gliotoxin. Using a strategy similar to that developed to assess neuronal injury, the extent of astrocyte lysis could be quantified enzymatically by determining the activity of LDH, retained in the surviving cells (Koh and Choi, 1987; Bridges et al., 1991a). When compared to microscopic evaluation, this enzymatic assay was found to be a more sensitive and accurate indicator of cellular lysis than visual observation. Total activity of LDH (that which is released into extracellular media, plus that which was retained in the surviving cells) was determined in a number of experiments (data not shown) and did not change during the course of the exposure to the toxins. As shown in Fig. 1, the gliotoxic activity of L-a-AA exhibited a concentration-dependence, with an LD,, of about 0.6mM 48 hr following a single addition of the amino acid to the culture media (Fig. 1). The curve of the dose-response relationship of this gliotoxic activity was quite steep, as the extent of glial injury produced by 0.4 and 0.75 mM changed from 74+ 10% to 24 f 6% survival, respectively (mean + SEM). Although this narrow range of effective concentrations may suggest a threshold effect, it may also indicate that the lysis of the first cells contributed to the injury of the remaining cells. This could, for example, be attributable to the release or loss of regulation of degradative enzymes that, when emptied into the extracellular space, initiate a pathological chain reaction that rapidly leads to the lysis of the remaining cells.
Gliotoxic actions of exitatory amino acids
120 r
0.1 L-a-Aminoadipate
1
(mM)
Fig. 1. Gliotoxic action of L-a-aminoadipate, as a function of concentration. Astrocyte cultures were exposed to the indicated amount of the compounds for 48 hr. The extent of cell lysis was quantified by determining the amount of activity of LDH remaining in the surviving cells and is reported as mean percentage of control + SEM. The figure includes the cumulative data from 4 different dose-response curves, each using a different astrocyte preparation. The number of determinations (i.e. individual culture wells) at each concentration is shown (n).
Pharmacology
of gliotoxic amino acids
Like the majority of glutamate analogs that exhibit activity at EAA receptors, L-a-AA is an a-amino-
diacid in which the acidic groups are separated by a relative distance of 2-4 carbon units. These chemical similarities raised questions regarding the structure-function relationship between other EAA analogs and their potential gliotoxic activity. Table I summarizes the effect that a variety of these analogs had on confluent cultures of astrocytes of the rat. The toxic action of the compounds was assessed 48 hr after a single exposure to an initial concentration of 1 mM. In addition to the expected activity of L-a-AA, the sulfur-containing amino diacids L-homocysteate, L-serine-O-sulfate and ~-cysteine-sulfinate were found to be exceptionally gliotoxic. Structurally, these amino acids are one methylene group shorter than L-a-AA and possess a sulfonic acid group in place of its 6carboxyl group. L-Cysteate, which is one methylene group shorter than r.-homocysteate, did not exhibit any gliotoxic activity. At a similar concentration, the substitution of a phosphate group for the terminal carboxyl group resulted in a loss of toxic activity, in the instance of L-serine-O-phosphate. In contrast, the phosphonic amino acids, D,L-AP~ and L-AP4 induced a substantial lysis of the astrocytes. Loss of L-a-AA’s amino group, as tested with a-ketoadipate, resulted in the loss of gliotoxic activity. A similar lack of activity was also observed with a-keto analogs of glutamate and aspartate, suggesting that metabolism of these compounds by transamination does not lead to the production of a gliotoxic derivative.
901
In contrast to L-a-AA, D-Cf-aminoadipate was much less effective as a gliotoxin, confirming the stereospecificity first reported by Olney et al. (1980). A similar requirement was observed in the present study, where L-homocysteate and L-AW were considerably more toxic than their respective D-enantiomers. As previous studies indicated that L-a-AA must first be transported into the cells, prior to the cytotoxic response (Huck et al., 1984b), this stereospecificity could reflect the enantioselectivity of either the mechanism directly involved in the toxicity or the transport system responsible for its intracellular accumulation. The well known EAA receptor specific agonists NMDA, KA and AMPA were also examined as potential gliotoxins. In contrast to their potent neurotoxic activity, no cellular damage was detected after 48 hr of exposure to concentrations of up to 2 mM (Table 1). Similarly, QA did not appear to induce any glial loss when it was screened at initial concentrations of 1 mM. Larger concentrations (2-3 mM, data not shown) of this EAA did, on occasion, produce a substantial lysis of the astrocytes, although the activity was considerably more variable than that observed with the other compounds. This action is consistent with the recent study of Haas and Erdo (1991) who described a transient vulnerability of Table I. Effects of excitatory amino acid analogs on cultures of neonatal rat astrocytes’ Astrocyte survival (% of LDH retained)
Compound L-a-Aminoadipate I mM (14) D-or-Aminoadipate I mM (8) L-Glutamate I mM (9) D-Glutamate I mM (26) L-Aspartate 1mM (7) D-Aspartate I mM (17) Quisqualate 1mM (26) Glycine I mM (6) Surfore containing amino acid3 L-Serine-O-sulfate I mM (7) L-Cysteine-sultinate I mM (8) L-Homocysteate I mM (9) D-Homocysteate I mM (7) L-Cysteate I mM (IS) Taurine I mM (7) Phosphate
containing
I mM (8)
EAA
agonists
receptor-spectfk
a-Keto
7*3 13*9 25+11 93 + 6 105 * 9 107 + 8
amino acids
L-Serine-U-phosphate L-AW I mM (6) D-AP~ I mM (6) D,L-AP3 I mM (12) N-Methyl-D-aspartate Kainate I mM (7) Kainate 2 mM (6) AMPA 2 mM (6)
16rt8 80 f 8 97+ II 75* 15 93 + 6 92+_ 14 125 + I6 101 * IO
I mM (7)
100~10 16220 101 + 12 45+15 92* IO 95 f 9 107+2 102*5
aci&
a-Ketoadipate 1 mM (1 I) a-Ketoglutarate I mM (IO) Oxaloacetate I mM (7)
101 -+ 11 92 * 7 97* I2
Cultures of neonatal rat astrocytes were exposed to a single addition of the EAA analogs (initial concentration indicated) as described in Methods. After 48 hr of continuous exposure, the extent of cellular lysis was assessed by quantitating the LDH activity retained in the surviving cells. The values, which are. reported as mean percent of control +_SD, were based on 6-26 (n) determinations (i.e. individual culture wells) from 2-4 different astrocyte preparations.
cultured stration, indiiated at some receptor.
astrocytes to QA. The present demonhowever, that AMPA was not glioioxic, that the action of QA was most probably point other &an the A~PA~QA EAA
DISCUSSION
fn the present study, the authors have begun to i~v~t~gate the ghotoxic property of L-~-AA by first c~~~ac~e~~~~ the pb~a~lo~~l safe of gliotoxic amino acids. Toward this goal, a wide range Morphdogkd changes during t?xposwe of EAA analogs have been screened in cultures of astrocytes of the neonatal rat, using the loss of the Prior to their eventual lysis, the astrocytes exposed cytosohc enzyme LDH to quantify the extent of the to W-AA, ~*hom~y~teate or L-setine-U-sulfa& @ai damage. As expected, t-cr -AA was four& to be underwent a succession of very ~stin~~ve morphoIogicat change.s A time-course of these changes is a potent gliotoxin. Its eytotoxic action was concennation-decedent and exhibited an X.X& of about illust~~cd se~u~~~ally in Figs 2 and 3. The earliest and most prominent alterations, observed in the 0.6 mM, as followed with the LDH protocol. Of the EAA analogs tested, r_-homocysteate, L-cysteine-sulfiastrocytes, involved abnormalities in nuclear marnate, tserine-Q-sulfate and L-AP4 were also found pholugy. The nuclei appeared swollen, contained to exhibit a cytotoxic effect on the euitured astro*‘pale” or “watery” ~ncleopl~m and exhibited a very cytes. The action of ~-hom~yste~c acid is of particuprominent nuclear membrane and obvious nucleoli. far interest because recent evidence suggests that it These chartges, which were observable as early as 6 hr after the addition of the ghotoxic compounds (Fig. 2), may be an endogenous transmitter and excj~o~ox~~ in the neocartex of the rat (Do, Herrling, Streit, were induced most rapidly by those acidic amino Turski and Cuenod, 1986; Knopfel, Zeise, Cuenod acids that proved to be the most toxic. When exposed and ~ieglgans~r~r~ 1987; Olney, Price, Salles, to smaller ~~~~trat~o~s of the g~~oto~~~, &m&r ~ab~~ere, Ryerson, Mahan, Frierdich and Samson, m&i were also observed, even though the ~s~r~yt~~ 1987; Kim, Koh and Choi, 1987). Consistent with did not lyse within the 48 hr experimental interval, these results, s~~taneo~ injections of o+homoThe nuclear mo~hology~ induced by the gli~toxic cysteate have been shown to produce a loss of amino acids, appeared to he very similar to a patho* non-neuronal cells in the retinas and brains of infant logical form of glia that is characteristically associmice, (Olney, Ho and Rhea, 1971). Similarly, D-t-hoated with hepatic e~~h~opa~~ (AIzheimer’s Type rn~~s~e~~acid and D,L-AP~ have been &own to be Xf (AZ IX) astr~~tcs~ Norenberg, $987) In the pretoxic to the glioma C5 ceil line of the rat, although sent e~~~rnen~~ continued exposure to the gliotoxic amino acid Ied to an increase in the prominence of the these cells appeared to be less sensitive to the toxin than the cultures used in the present investigation nuclear abnormalities. These changes were followed by progressive swelling of the cell body and mem* (Kato, Higashida, Higuchi, Hatakenaka and Negishi, 1984). Qf the compounds tested in this study, all of brane blebbing that eventually terminated with the those id~~~ad as toxic to the astrocytes were LWXtysis of the c&t, During the latter stages of the toxic amino diacids, whife the ~-e~a~tjorne~ of homoresponse, the nu&i changed from their swollen AZ cysteate and AP4 had uo visible e&t on the c&s B-type appearance, taking on a shrunken, pyknotic when tested at similar concentrations. This enantioseappearance (Figs 3D and F). lectivity is consistent with earlier studies identifying A different course of events was observed when the astracytes were exposed to high levels of the D-~-AA as ~lioto~cally inactive (also confirmed in the present study) and suggests the preboils deendogenous exettatory t~~srnitt~~s, L-glutamate and t-aspartate. In these onsets, the astrocytes apscribed in r&o actions of bomocysteat~ and AP4, cart peared to undergo the initial sequence of morphobe ascribed to the ~~n~tiorn~~s~ logical changes induced by the gliotoxins, particularly Although a number of the compounds, identified the nuclear anomalies (Fig. 2B) but then returned to as gliotoxins, are well known FAA agonists (e.g, what seemed a normal form, without further swelling L-homocysteic acid, t.-serine-O-sulfate), the lack of or fysis (Fig. 3B). Thus, fX4kr after exposure to an effect by NMDA, KA and AMPA suggests that 1 mM ~~glutamate~ the rmclei took on the appearance the process does not involve the EAA reeptors that of the AZ II astrocytes, ch~cte~stic of those cehs figure prorn~~en~y in exc~totoxic~m~~at~ neuronal that would have eventually been Iyzed by L-~-AA or death {for review see Choi, 1988). Of the three t-homocysteate (Fig. 3). Instead of dying, however, excitotoxins, KA is of particuiar note, as studies have the cells returned to an apparently normal mordemonstrated the presence of a glutamate/kainatephology by 36-48 hr. These findings suggest that the activated Na+ channel on astrocytes (Bowman and ~~~~~ of AZ $1 ~r~~o~~y is no% 8n inewrsK~mal~rg, 1984; Kett~rn~~ and Sehachner, 1985; ible step that signals the eventuai death of the astroSonthetmer, K~t~nrna~n~ Rackus, and Schachner* cyte. ~u~he~~~, as the changes were brought 1988; Rackus, Ketfa~ann and Schacbner, 1989), about by excessive levels of the endogenoua transmitFurther evidence that this receptor does not mediate ter L-glutamate, the pathological appearance of these the gliotoxic response is provided by the action of cells in vitromay he indicative of areas undergoing an L-homocysteic acid, which is only a weak agonist at exeitotox~c insult. this glial receptor_ yet prov& to be one of the most
Fig. 2. Photomicrographs, depicting the L-cl-aminoadipate (D) on confluent cultures (1 mM initial concentration) of the EAA become observable. Control
early effect of L-glutamate (B), r_-homocysteate (C) and of astrocytes of neonatal rat. Six hours after a single addition analog, the stereotyped alterations in nuclear morphology cells are shown in (A). (Bar inset = 100pm.)
903
Fig. 3. Photomicrographs, depicting the long term effect of the L-glutamate (B), L-a-aminoadipate (C) and (D) and L-homocysteate (E) and (F) on confluent cultures of astrocytes of neonatal rat. The cells shown, which are from the same cultures presented in Fig. 2, were exposed to an initial concentration of 1 mM EAA analog. The astrocytes appeared to return to a normal morphology 36 hr after the addition of L-glutamate (B), control cells are shown in (A). In contrast to the action of L-glutamate, the cellular swelling and nuclear abnormalities, induced by L-a-aminoadipate and L-homocysteate continued (C and E, respectively at 24 hr) and eventually led to the lysis of the astrocytes (D and F, respectively at 48 hr). (Bar inset = 100 pm.)
904
Gliotoxic actions of exitatory amino acids
potent gliotoxins. The lack of an involvement of an extracellular receptor also supports the gliotoxic model of Huck et al., who proposed that this L-~-AAinduced cell death is dependent on the intracellular accumulation of the amino acid (Huck et al., 1984b). This conclusion was based on immunocytochemical and autoradiographic studies that demonstrated the sodium-dependency of both ~-x-AA-mediated toxicity and transport and the inability of TTX to block the glial lysis. This would suggest that, in addition to those properties directly responsible for the pathological damage, the gliotoxic L-x-amino diacids must also be substrates of a transport system and accumulate intracellularly. The fact that each of these steps could have its own pharmacological requirement may explain why two structurally very similar acidic amino acids, such as L-serine-O-sulfate and L-serineO-phosphate, exhibited very different physiological effects. Studies by Murphy et al. (Murphy, Miyamoto, Sastre, Schaar and Coyle, 1989; Murphy, Schnaar and Coyle, 1990) have identified an alternative mechanism by which EAA analogs induce a cytotoxic effect that is related to the ability of these compounds to inhibit transport of cystine. As a result of attenuated uptake of cystine, the intracellular levels of glutathione are reduced and the cells succumb to oxidative stress. Although this mechanism is thought to contribute to the EAA-mediated cell loss observed in neuroblastoma cells (Nl8-RE- 105) and immature cortical neurons (E17: 24-72 hr in culture), glial cells in the primary cultures were found to be insensitive (Murphy et al., 1990). Further evidence that this mechanism is not responsible for the gliotoxic activity reported in the present study, comes from the demonstration that little or no cellular lysis is observed in mixed cortical cultures, that have been depleted of glutathione with buthionine sulfoximine (Bridges, Koh, Hatalski and Cotman, 199lb). Thus, even though excessive levels of EAA analogs might act to reduce transport of cystine in these astrocytes, it appears very unlikely that the primary mechanisms underlying EAA-mediated gliotoxicity are related to this pathological scheme. Prior to their eventual lysis, the astrocytes underwent a series of distinct morphological changes, the most prominent of which was marked by the appearance of nuclei that were swollen, contained “pale” or “watery” nucleoplasm and exhibited a very prominent nuclear membrane, with obvious nucleoli (Figs 1B and D). These same nuclear changes were also observed after the administration of the endogenous transmitters L-glutamate and L-aspartate. In contrast to the gliotoxins, however, the astrocytes exposed to these amino acids appeared to return to a normal morphology, without lyzing. Thus, L-glutamate may initially mimic the potentially pathological action of L-LX-AA,although the process is terminated before enough damage has occurred to result in the lysis of the cells. As it appears that the death of
905
the astrocyte is dependent upon the accumulation of the toxic acidic amino acids, the termination of the toxic response may be attributable simply to the metabolism of the excess L-glutamate and L-aspartate. Consistent with this, studies with the C6 glioma cell line indicate that the sensitivity to glutamate- but not L-u-AA-mediated toxicity, could be reduced by chemically inducing activity of glutamine synthetase (Kato et al., 1984). Thus, the gliotoxic response could be avoided by rapidly metabolizing the L-a-amino diacids, thereby preventing their accumulation. The morphological changes observed in response to the gliotoxic a-amino-diacids, as well as to Lglutamate, appear to be similar to AZ II astrocytes that are recognized as a pathological hallmark of hepatic encephalopathy (for review see Norenberg, 1981). Originally named by Waggoner and Malamud (1942), AZ II astrocytes were identified by von Hosslin and Alzheimer (1912) and later associated with hepatic encephalopathy by Adams and Foley (1953). Although their appearance has generally been attributed to high levels of ammonia, the present results suggest that excessive levels of ~-x-amino diacids, including L-glutamate, may also represent a contributing factor. The amount of L-glutamate used in these experiments was less than the levels of ammonia previously used to induce these alterations in vitro (4 days of exposure to 10 mM ammonia, compared to 24-48 hr to 1 mM L-glutamate; Norenberg; 1987). Thus the appearance of the abnormal nuclei may not necessarily be a specific marker of ammonia-mediated damage but, instead, represent a more general marker of metabolic imbalance or cellular pathology. In previous studies, AZ II astrocytes produced by long term exposure of cultures to ammonia (14 days) have been shown to exhibit: (i) a reduction in their ability to transport a-aminobutyric acid (GABA) and L-glutamate, (ii) an impaired capacity to clear extracellular potassium and (iii) a decrease in the intracellular levels of adenosine triphosphate (ATP) (Norenberg, Moses, Papendick and Norenberg, 1985; Norenberg, 1987). If the observed morphological changes are indicative of this damage, the present results would suggest that astrocyte function may also be comprised by excessive levels of L-glutamate. Such an effect may be of particular significance in excitotoxic injury, where the transport and metabolism of L-glutamate by astrocytes are thought to be important in regulating its extracellular concentration and preventing glutamate-mediated neuronal loss (see, for example, Sugiyama, Brunori, and Mayer, 1989). Although initially protective, the continued accumulation of L-glutamate by astrocytes may also compromise their function and reduce the ability of these cells to transport or even maintain their intracellular pools of L-glutamate. A cumulative cycle of damage may ensue, with the end result being an increase in the levels of extracellular L-glutamate and subsequent neuronal loss. This cascade of
906
R. J. BRIIffiEs
pathological events highlights the intricate balance that must exist in the EAA transmission system and illustrates how each component, if not functioning properly, has the potential to cont~bu~ to excitotoxic pathology and neuronai loss. Acknowledgements-This
work was supported in part by
NIH grant NS27056 and
NIA L.E.A.D. award AG07918.
The authors also wish to thank J. S. Kahle for her assistance in the preparation of this manuscript. REFERENCES
Adams R. D. and Foley J. M. (1953) The neurological disorder associated with liver disease. Ass. Res. Nerve Ment. Dis. Proc. 32: 198-237. Backus K. H., Kettenmann H. and Schachner M. (1989) Pha~acolo~cal characterization of the glutamate receptor in cultured astrocytes. J. Neurosei. Res. 22s 274282. Balcar V. S. and Johnston G. A. R. (1972) Glutamate uptake by brain slices and its relation to the depolarization of neurons by acidic amino acids. J. Neurobiof. 3: 295-30 1. Bowman C. L. and Kimelberg H. K. (1984) Excitatory amino acids directly depolarize rat brain astrocytes in primary culture. Nature 311: 656-657. Bridges R. J., Hatalski C. G., Shim S. N. and Nunn P. B. (1991a) Gliotoxic properties of the Lathyrus excitotoxic B-N-oxalyl-L-a$-diaminopropionic acid (fi-L-ODAP). Brain Res. 561: 262-268. Bridges R. J., Koh J., Hatalski C. G. and Cotman C. W. (1991b) Increased excitotoxic vulnerability of cortical cultures with reduced levels of ~u~thione. Eur. J. Pharm. 19% 199-200. Choi D. W. (1988) Glutamate neurotoxicity and diseases of the nervous system. Neuron 1: 623-634. Do K. Q,, Herrling P. L., Streit P., Turski W. A. and Cuenod M. (1986) In vitro release and electrophysiological effects in situ of homocysteic acid, an endogenous N-methyl-D-aspartic acid agonist, in the mammalian striatum. J. Neurosci. 6: 222&2234. Haas J. and Erdo L. E. (1991) Quisqualate-induced excitotoxic death of glial cells: transient vulnerability of cultured astrocytes. Gliu 4: 11l-l 14. Hamberger A. C., Chiang G. H., Nylen E. S., Scheff S. W. and Cotman C. W. (1979) Glutamate as a CNS transmitter. I. Evaluation of glucose and glutamine as precursors for the synthesis of p~ferentially released glutamate. Brain Res. 7: 513-530. Hertz L. and Schousboe A. (1986) Role of astrocytes in compartmentation of amino acid and energy metabolism. In: Astrocytes: Biochemistry, Physiology, and Pharmacology of Astrocytes (Federoff S. and Vernadakis A., Eds), pp. 179-208. Academic Press, New York. Hosli E., Schousboe A. and Hosli L. (1986) Amino acid uptake. In: Asirocytes: Biochemistry, Physiofogy, and Pharmacofogy of Astrocyfes (Federoff S. and Vernadakis A., Eds), pp. 133-153. Academic Press, New York. von Hosslin C. and Alzheimer A. (1912) Ein beitrag zur klinic und pathologischen anotamie der WestphalStrumpellschen pseudosklerose. 2. Neural. Psych&. 8: 183-209. Huck S., Grass F. and Hatten M. E, (1984a) Gliotoxic effects of a-aminoadipic acid on monolayer cultures of dissociated postnatal mouse cerebellum. Neuroscience 12: 783-791. Huck S., Grass F. and Hortnagl H. (1984b) The glutamate analogue a-aminoadipic acid is taken up by astrocytes
et al.
before exertine its stliotoxic effect in vitro. J. Neurosci. 4: 2650-2657. Kato S., Higashida H., Higuchi Y., Hatakenaka S. and Net&hi K. (1984) Sensitive and insensitive states of &ured glioma cells to glutamate damage. Brain Res. 303: 365-373. Kettenmann H. and Schachner M. (1985) Pharmacological properties of GABA-, glutamate-, and aspartate-induced depolarizations in cultured astrocytes. J. Neurosci. 5: 32953301. Kim J. P., Koh J. Y. and Choi D. W. (1987) L-Homocysteate is a potent neurotoxin on cultured cortical neurons. Bruin Res. 4%~ 103-110. Kimelberg H. (1983) Primary astrocyte culture-a key to astrocvte function. Cell. molec. Neurobiol. 3: 1-16. Knopfel-T., Zeise M. L., Cuenod M. and Zieglgansberger W. (1987) L-Homocysteic acid but not L-glutamate is an endogenous N-methyl-o-aspartic acid receptor preferring agonist in rat neocortical neurons in vitro. Neurosci. I&t. 81: 188-192. Koh J. and Choi D. W. (1987) Quantative dete~ination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J. Neuroxi. Meth. 20: 83-90. Lund Karlsen R., Pedersen 0. O., Schousboe A. and Langeland A. (1982) Toxic effects of D,L-a-aminoadipic acid on Muller cells from rats in vivo and cultured cerebral astrocytes. Expl Eye Res. 35: 305-3 11. McCarthv K. D. and de Vellis J. (1980) Preparation of separate astroglial and oligodendroghal cell cultures from rat cerebral tissue. J. Cell Biol. 85: 890-902. Monaghan D. T., Bridges R. J. and Cotman C. W. (1989) The excitatory amino acid receptors: their classes, pharmacology, and distinct properties in the function of the central nervous system. A. Rev. Pharmac. Toxic. 29~ 365-402. Murphy T. H., Schnaar R. L. and Coyle J. T. (1990) Immature cortical neurons are uniquely- sensitive to glutamate toxicity by inhibition of cystine uptake. FASEB J. 4: 1624-1633. Murphy T. H., Miyamoto M., Sastte A., Schnaar R. L. and Covle J. T. (1989) Glutamate toxicity in a neuronal &I line involves inhibition of cystine transport leading to oxidative stress. Neuron 2: 1547-1558. Norenberg M. D. (1981) The astrocyte in liver disease. Adv. cell. Neurobiol. 2: 303-352. Norenbern M. D. (1987) The role of astrocytes in hepatic encephsopathy. ‘Neurachem.Path. 6: 13-33. _ Norenberrz M. D.. Mozes L. W.. Pawndick R. E. and Norenb&g L. 0: B. (1985) Elf&t of’ammonia on glutamate, GABA and rubidium uptake by astrocytes. Ann. Neural. 18: 149. Olney J. W., de Gubareff T. and Collins J. F. (1980) Stereospecificity of the gliotoxic and anti-neurotoxic action of alpha-aminoadipate. Neurosci. Left. 19r 277-282. Olney J. W., Ho 0. L. and Rhee V. (1971) Cytotoxic effects of acidic and sulphur containing amino acids on the infant mouse central nervous system. Expr Brain Res. 14: 61-76. Olney J. W., Price M. T., Salles K. S., Labruyere J., Ryerson R., Mahan K., Frierdich G. and Samson L. (1987) L-Homocysteic acid: An endogenous excitotoxic ligand of the NMDA receptor. Brain Res. B&f. 19: 597-602.
Schousboe A. and Hertz L. (1981) Role of astroglial cells in glutamate homeostasis. Adv. Biochem. Psychophormat. 27: 103-113. Shank R. P. and Aprison M. H. (1981) Present status and significance of the glutamine cycle in neural tissue. I.19 Sci. 28: 837-842.
Gliotoxic actions of exitatory amino acids Sontheimer H., Kettenmann H., Backus K. H. and Schachner M. (1988) Glutamate opens Na + /K + channels in cultured astrocytes. Gliu 1: 328-336. Sugiyama K., Brunori A. and Mayer M. L. (1989) Glial uptake of excitatory amino acids influence neuronal surviva1 in cultures of mouse hippocampus. Neuroscience 32: 779-79
1.
907
Vemadakis A. (1988) Neuron-glia interrelations. Int. Rev. NeurobioI. 30: 149-224.
Waggonner R. W. and Malamud N. (1942) Wilson’s disease in the light of cerebral changes following ordinary acquired liver disorders. J. Nerve. Ment. Dis. 96: 410-423.
