Glutamatergic signaling has been linked to the recently discovered neurotransmitter/neuromodulator nitric oxide (NO), and several classes of anesthetics block some step in glutamatergic signaling. This study was designed to determine whether or not ketamine would prevent NO-dependent cGMP production stimulated by glutamate (GLU) and the GLU analogs NMDA, quisqualate (QUIS), and kainate (KAIN).
Primary cultures of cortical neurons and glia (prepared from 16-day gestational rat fetuses) were used after 12-16 days in culture. Reactions were carried out in magnesium-free buffer containing 100 microM 3-isobutyl-1-methylxanthine, and cGMP content of cultures was used as a bioassay of NO production.
Cyclic GMP production stimulated by sodium nitroprusside (100 microM) occurred predominately in neurons and not in glia. Neurons were spontaneously active in these cultures; basal cGMP production was decreased by 50% in the presence of 1 microM tetrodotoxin (TTX). Glutamate (100 microM), NMDA (100 microM), QUIS (300 microM), and KAIN (100 microM) each increased cGMP content of neuronal cultures. L-NMMA (100 microM), a NO synthase inhibitor, prevented the stimulation of cGMP production by GLU or its analogs. Pretreatment with MK-801 (1 microM) or ketamine (10-100 microM) inhibited GLU-, NMDA-, and QUIS-stimulated cGMP production. Quisqualate-stimulated responses were the most sensitive to inhibition by ketamine and NMDA-stimulated responses were the least sensitive to inhibition. MK-801 and ketamine did not significantly inhibit KAIN-stimulated cGMP production. CNQX (10 microns) blocked KAIN-stimulated cGMP production only.
The authors' data demonstrate that ketamine inhibited NO synthesis stimulated by NMDA- and non-NMDA-receptor specific analogs. Our findings indicate that blockade of QUIS- as well as NMDA-subtypes of GLU- receptor may be important in the development of ketamine-induced anesthesia.
Key words: Agonist: NMDA. Amino acid, excitatory: kainate; quisqualate. Anesthetics, intravenous: ketamine. Antagonists: MK 801. Cultured neurons. Neurotransmitters: glutamate.
GLUTAMATE (GLU) is the major excitatory neurotransmitter in the vertebrate central nervous system. L-glutamate activates several subtypes of receptors leading to an increase in intracellular calcium ion concentration ([Ca2+]i). This increase in [Ca2+]ioccurs by at least three mechanisms. First, Calcium2+ enters the cells from the extracellular space through ion channels that are integral to GLU receptors that are activated selectively by the GLU analog N-methyl-D-aspartate (NMDA). [1–4] Second, Calcium2+ enters the cells through voltage-dependent calcium channels that are opened secondary to an influx of sodium ions through ligand gated channels that are activated selectively by the GLU analogs kainate (KAIN), quisqualate (QUIS), alpha -amino-3-hydroxy-5-methyl-4-isoxazoleproprionic acid (AMPA), or NMDA. [1,5,4] Third, metabotropic GLU receptors that are selectively activated by QUIS or (plus/minus)-1-aminocylopentane-trans-1,3-dicarboxilic acid (ACPD) stimulate phospholipase C through a G-protein to produce inositol 1,4,5-trisphosphate (IP3) and release of Calcium2+ from intracellular stores. [5,6] Activation of GLU receptors and subsequent cellular responses are associated with the production of excitatory postsynaptic potentials (EPSPs) and long-term modulation of neuronal behavior, including changes in neuronal excitability, nerve cell architecture, and gene expression. [7–13].
Although many responses to GLU in the central nervous system can be directly attributed to these changes in membrane polarization and [Ca2+]i, it has been demonstrated recently that activation of GLU receptors in the central nervous system results in the synthesis of nitric oxide (NO). This response was first observed by Garthwaite et al. * RF 14,15* Nitric oxide produced after GLU stimulation acts as a paracrine/autocrine substance where it stimulates cyclic GMP (cGMP) accumulation,  and may have other actions, as well. [17,18].
There is increasing evidence that blocking the function of the GLU-Calcium2+-NO-cGMP pathways may be associated with the development of anesthesia. Riluzole (53274 RP), an inhibitor of GLU neurotransmission with both pre- and postsynaptic effects, has anesthetic action.  Noncompetitive antagonists at the NMDA receptor, including (+)-5-methyl- 10, 11-dihydro-5H-di-beno(a,d) cyclohepten-5, 10-amine maleate (MK-801), (+)-N-allyl-normetazocine [(+)-SKF 10,047], phencyclidine, dextrorphan, and ketamine, increase the potency of a variety of general anesthetics and have analgesic and anesthetic properties. [20,21] The competitive antagonist at the NMDA receptor, CGS-19755, also has analgesic and anesthetic effects. [22,23] A competitive antagonist at AMPA receptors, 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo(f)quinoxaline (NBQX), reduces the MAC for halothane in rats.  Because GLU has been associated with both direct and indirect postsynaptic responses (e.g., ion flux, EPSP, and NO release), the possibility has been raised that the anesthetic properties of at least some drugs may be attributable to inhibition of NO synthesis/release. In support of this possibility, it has been shown that antagonists of NMDA receptors inhibit GLU-stimulated increases in NO synthesis, [25,26] and blocking the synthesis of NO increases the potency of the volatile anesthetic halothane.  Also, NO synthase inhibition has been shown to be antinociceptive, [28–33] although other work indicates that NO may mediate some forms of analgesia. [31,34–37] In addition, the administration of GLU or its analogs can reverse ketamine anesthesia. [38,39].
In vivo and in vitro studies from several laboratories indicate that various general anesthetics block the GLU-Calcium2+-NO-cGMP pathway at several sites. Martin et al.  demonstrated that enflurane inhibited GLU-stimulated MK-801 binding to NMDA receptors in membranes prepared from rat cerebral cortex; because MK-801 binds to the open channel, this indicated that enflurane decreased opening of the anion channel. Halothane and isoflurane reduce increases in [Ca2+] sub i in response to GLU and NMDA in neurons cultured from rat hippocampus.  In cortical wedges from mice, electrophysiologic responses of neurons to AMPA and NMDA were reduced by chloroform, halothane, thiopental, diethyl ether, and isoflurane.  Enflurane also decreased cGMP content of the cerebellum of mice. .
Ketamine is the single anesthetic drug used in clinical practice that is believed to act primarily by blockade of GLU receptors. [44–47] Whether ketamine works exclusively by blockade of NMDA receptors, or whether it blocks the function of other subtypes of GLU receptor, is not known with certainty. In studies of cortical wedges from mice, ketamine blocked responses to NMDA but not to AMPA, but agonists at other subtypes of GLU receptors were not tested.  Studies of isolated spinal cord from one of the most primitive vertebrates, the lamprey, showed that ketamine specifically blocked NMDA responses and not responses to KAIN or QUIS.  Studies in another model system, locust muscle, indicate that ketamine can block a QUIS receptor.  The current study was designed to examine how ketamine affects the function of the various subtypes of GLU receptors. Our main experimental goal was to determine whether or not ketamine would prevent NO-dependent cGMP production stimulated by the GLU analogs QUIS and KAIN, as well as the responses to GLU and NMDA in cortical neurons cultured from the rat.
Materials and Methods
Chemicals used in the current study were obtained from the following sources: GLU, KAIN, NMDA, tetrodotoxin (TTX), cytosine- beta-D-arabinofuranoside (Ara-C), 3-isobutyl-l-methylxanthine (IBMX), Hank's balanced salt solution (HBSS), Dulbecco's modified Eagle's medium with 25 mM HEPES (DMEM), and sodium nitroprusside (SNP) from Sigma (St. Louis, MO); Ham's F-12 and penicillin/streptomycin from JRH Biosciences (Lenexa, KS);(+)-quisqualate and 6-cyano-7-nitro-quinoxaline-2,3-dione (CNQX) from Research Biochemicals, Inc. (Natick, MA); ketamine HCl from Parke-Davis (Morris Plains, NJ); and NG-monomethyl-L-arginine (L-NMMA) from Calbiochem (San Diego, CA). Fetal calf serum was obtained from HyClone (Logan, UT).
Cultures of cerebral cortical neurons were prepared essentially as described by Dichter  and Atkinson and Minneman.  Whole cerebral hemispheres were collected in HBSS from fetuses of 16–17-day gestational Sprague-Dawley rats. They were washed twice in phosphate buffered saline, digested using trypsin (5 mg *symbol* ml sup -1) in DMEM, washed three times in HBSS, and triturated with a “fire polished” pasteur pipette into DMEM containing 25 mM HEPES supplemented with 10% Ham's F12 media, 10% heat inactivated fetal calf serum, 100 micro gram *symbol* ml1streptomycin, and 100 U.ml penicillin. Cells were plated at a density of 4 x 105cells per 1.2-cm-diameter tissue culture well that had been pretreated with 10 micro gram *symbol* ml sup -1 poly-D-lysine. The cultures were grown in a 37 degrees Celsius humidified incubator containing 57% CO2in air, and were fed approximately every 2–3 days with fresh medium. On day 5, cells were treated with 10 micro gram *symbol* ml sup -1 Ara-C for 24 h to stop nonneuronal cell proliferation. Subsequent feeding was with the above-described media without Ham's F-12, to eliminate GLU from the feeding. Experiments were carried out after 12–16 days in culture. Preliminary experiments indicated that the capacity of the cultures to respond to GLU with an increase in cGMP was well developed by this time.
Glial cell cultures were prepared as described by Atkinson and Minneman  by trypsinizing cells at day 5 in culture and replating at a density of 8 x 104cells per 1.2-cm-diameter well. The cultures were also fed every 2–3 days, but were not treated with Ara-C. This procedure resulted in preparations that were essentially devoid of neuronal cells. Similar to the neurons, experiments were performed after 12–16 total days in culture.
Measurement of cGMP Production
Experiments were performed at 37 degrees Celsius. Cells were washed three times in buffer (in mM: NaCl, 137; KCl, 0.4; CaCl2, 1.3; KH2PO4, 0.44; Na2HPO4, 0.7; glucose, 6; HEPES, 20; pH 7.4) with a final addition of buffer as the reaction medium. Tetrodotoxin (1 micro Meter) was added at this time in most experiments to block spontaneous action potentials and transmitter release. IBMX (100 micro Meter), a phosphodiesterase inhibitor, various GLU antagonists, or vehicle were added at the beginning of the timed 10-min incubation period. After the incubation period, agonist or vehicle was added for 90 s, except as noted for the time course experiments. At this time, the supernatant was removed and 0.1 M HCl was added to stop the reaction and to extract cyclic nucleotides. Cyclic GMP was measured by radioimmunoassay.  Protein content was determined using the coomassie blue method.  Results were expressed as pmol cGMP *symbol* mg sup -1 protein.
Data were expressed as mean plus/minus SEM and differences between treatment groups were determined using Student's t test or ANOVA and post hoc analysis with the method of Bonferroni, as indicated in the figure legends. Differences were considered significant when P < 0.05.
The primary cultures of mixed cerebral cortical cells used in these experiments consisted of a bed of glial cells on which the neurons grew. The glial cultures were similar, but without the overlying neuronal cells. To determine whether pure cultures of glia, as well as the mixed cultures containing mostly neurons, could respond to NO with an increase in cGMP, we determined the time course of cGMP accumulation on exposure to the NO donor sodium nitroprusside (SNP), as shown in Figure 1. Although both types of cultures did respond to SNP with an increase in cGMP, the response in the neuronal cultures was of much greater magnitude, indicating that the preponderance of the cGMP response occurred in neurons. Figure 1demonstrates that the increase in cGMP in response to 100 micro Meter SNP was maximal by 90 s. Therefore, all subsequent experiments were terminated 90 s after the addition of agonist.
Neuronal cells in culture form multiple complex intercellular contacts and are spontaneously active.  To determine whether spontaneous neuronal activity, including action potential-stimulated neurotransmitter release, was responsible for a portion of the basal cGMP content observed, the cells were treated with 1 micro Meter TTX, a sodium channel blocker. As shown in Figure 2, TTX-treated neurons had significantly reduced basal cGMP content. To prevent spontaneous generation of action potentials and, thus, transmitter release, TTX was included in all subsequent experiments.
Treatment of neuronal cultures with agonists at GLU receptors, including GLU (100 micro Meter), NMDA (100 micro Meter), QUIS (300 micro Meter), or KAIN (100 micro Meter), for 90 s each produced significant increases in cGMP content, as shown in Figure 3. Pretreatment of cultures with the NO synthase inhibitorL-NMMA (100 micro Meter) decreased cGMP production to below basal levels and prevented any increase in the presence of GLU or its analogs. These data demonstrated that GLU and its analogs stimulated cGMP production via the NO pathway and that any increase in cGMP observed in response to these agonists was subsequent to NO synthesis and release.
Pretreatment of cultures with ketamine (100 micro Meter) or MK-801 (1 micro Meter) blocked GLU-, NMDA-, and QUIS-stimulated cGMP production, as shown in Figure 4(A and B). The non-NMDA GLU-channel antagonist, CNQX (10 micro Meter), blocked cGMP responses to KAIN, but did not inhibit responses to GLU, NMDA, or QUIS, as shown in Figure 4(C).
Additional experiments were carried out with cultures pretreated with lower concentrations of ketamine to determine the dose-response relationship, if any. As shown in Figure 5, ketamine at concentrations as low as 10 micro Meter blocked the response to QUIS. The responses to NMDA, however, were not significantly affected by ketamine at concentrations less than 100 micro Meter. The response to GLU was inhibited by ketamine with a sensitivity between that of QUIS and NMDA.
Initial studies indicating that activation of GLU receptors in the central nervous system resulted in NO synthesis and increases in cGMP have implicated the NMDA receptor as the responsible subtype. [14,15] Subsequently, other GLU analogs have been shown to stimulate NO synthesis through activation of specific receptor subtypes. [25,26,54,55] Our results demonstrate that ketamine, and its sister compound MK-801, both drugs primarily believed to block the NMDA receptor channel, can also inhibit QUIS receptor mediated NO synthesis. However, ketamine and MK-801 had no significant effect on KAIN-stimulated NO synthesis.
We monitored changes in NO synthesis by measuring changes in cGMP content in cultures of cells from fetal rat brain after agonist stimulation. Although both neurons and glia can respond to NO with an increase in cGMP, we found that preparations of mixed cells from brain that contain neurons respond to SNP with a much greater increase in cGMP than cultures of nonneuronal cells (Figure 1). These findings indicated that the primary site of response to NO was in neurons and, more importantly, that the changes in cGMP production that we observed were from changes occurring in neurons, in which anesthetic drugs are believed to work.
The experiments with and without TTX (Figure 2) showed that neurons in these cultures were spontaneously active. These data indicated that there was, most likely, a basal release of endogenous neurotransmitter in these cultures that was caused by this spontaneous depolarization that stimulated NO synthesis. This phenomenon has been described previously in electrophysiologic terms in cultured neurons,  but we know of no other report indicating that a substantial portion of this basal generation of NO is caused by this mechanism. To minimize spontaneous activity and release of endogenous transmitter, subsequent experiments were all carried out in the presence of TTX. Thus, the cGMP measured is almost exclusively caused by neuronal stimulation by exogenous agonist and is not caused by intrinsic or spontaneous activity.
sub L -NMMA pretreatment clearly demonstrated that the observed stimulation of cGMP production was indeed through the NO pathway (Figure 3). The concentration ofL-NMMA used, 100 micro Meter, is maximally effective in inhibiting NO synthase. [14,16] Other investigators [26,54,55] have also shown that GLU-stimulated cGMP was caused by activation of NO synthesis in cultured cortical, striatal, and cerebellar neurons from rat and mouse, demonstrating that this pathway is conserved among brain regions and among species. Using supramaximal doses of agonists, our studies further showed that the GLU-agonists NMDA, QUIS, and KAIN were as effective as GLU in leading to the stimulation of NO synthase, indicating that full activation of all of the GLU pathways simultaneously give a response less than the sum of the responses of each pathway activated independently.
As expected, 100 micro Meter ketamine blocked cGMP production stimulated by the GLU analog NMDA. Blockade of NMDA receptor activation by ketamine has been shown previously using other indicators of NMDA-receptor stimulation [44–47] and is the basis for the view that NMDA-receptor activation is specifically coupled to the stimulation of NO synthase. In addition to acting at the NMDA-receptor channel, our data showed that 100 micro Meter ketamine blocked GLU- and QUIS-stimulated NO production. Quisqualate-stimulated NO production was more sensitive to ketamine blockade than NMDA-stimulated responses, as shown in Figure 5. Ketamine at concentrations of 10, 30, and 100 micro Meter prevented QUIS-stimulated cGMP production. This effect on QUIS-stimulated responses has not been reported previously.
The concentrations of ketamine used in this study are comparable with those measured in the plasma of rats and humans anesthetized by ketamine. The potency of ketamine is greater in humans compared with rats. In rats, plasma ketamine concentrations greater than 50 micro Meter are required to produce anesthesia and the righting reflex is lost only at concentrations greater than 20 micro Meter.  In humans anesthetized with ketamine, the plasma concentration has been measured to be approximately 5–10 micro Meter. [57,58] Thus, the range of concentrations used in this study spanned the range present clinically in humans and, more importantly, the concentrations required for anesthesia in rats, the species from which cells were obtained.
The ketamine analog MK-801 (1 micro Meter) blocked GLU-, NMDA-, and QUIS-stimulated cGMP production but had no significant effect on KAIN-stimulated responses. The effect of supposedly specific NMDA-antagonists such as MK-801 on responses to QUIS is controversial. Marin et al.  also demonstrated effective inhibition of QUIS-stimulated cGMP production by MK-801 in striatal neurons cultured from fetal mice. Wood et al.,  however, did not find that MK-801 inhibited QUIS-stimulaced cGMP production in the hippocampus in vivo. There are several possible mechanisms by which MK-801 and ketamine may block QUIS-stimulated cGMP production. Ketamine may block direct QUIS activation of the NMDA channel, but this is not likely. Quisqualate can act as an agonist at the NMDA receptor, [7,60] but the affinity of QUIS for the NMDA receptor is reported to be 2- to 23-fold less than that of NMDA and the total contribution of QUIS-activated current through NMDA channels is very small compared with that through the QUIS/AMPA channel in whole cell preparations. Another possible mechanism by which MK-801 and ketamine may block QUIS-stimulated cGMP production is by directly blocking activation of the AMPA/QUIS receptor or the metabotropic GLU receptor. A prominent effect on the metabotropic receptor (a nonionotropic receptor) would not have been observed by others, because the electrophysiologic techniques they used [44–47] would not have detected this biochemical response.
The effects of ketamine and MK-801 were specific for NMDA and QUIS receptors, because neither drug inhibited KAIN-stimulated cGMP production in our study. CNQX specifically blocked KAIN-stimulated cGMP production without effect on the activation of NMDA or QUIS receptors. This apparent lack of effect of CNQX on QUIS responses in our study may be explained by either rapid desensitization of the AMPA/QUIS receptor, therefore preventing any agonist-mediated response through this receptor,  or by the fact that QUIS activation of the metabotropic receptor gives a maximal cellular response. The lack of blockade of the GLU-stimulated response by CNQX supports the suggestion by others that GLU does not activate KAIN receptors. [7,61].
These findings may also help to explain the role of ketamine in the protection of neurons from excitotoxicity. Several investigators have reported that ketamine can protect the brain against ischemic injury. [62,63] Ischemia is believed to lead to the release of GLU and the subsequent activation of GLU receptors activates the cellular processes that can result in neuronal death. [64,65] Activation of NMDA and QUIS receptors are involved in neuronal death, but the production of nitric oxide and cGMP production may or may not be involved in neuronal death. [64,65] Protection of neurons from excitotoxicity by ketamine may, therefore, result from blocking the activation of both NMDA and QUIS subtypes of GLU receptors.
In conclusion, our study demonstrates that GLU and the GLU analogs NMDA, KAIN, and QUIS stimulate NO production in cultures of cerebral neurons from the rat. Ketamine, an anesthetic drug that is pharmacologically similar to MK-801 and blocks the NMDA-subtype of GLU receptor, also blocks the effect of activation of GLU receptors activated by QUIS. These results are consistent with the hypothesis that blockade of NO production, and, thus, inhibition of the subsequent increase in neuronal cGMP content, results in anesthesia; these results allow one to speculate that interruption of other than the NMDA subtype of GLU receptor may also contribute to the state of anesthesia.
The authors thank Chris Cuspud and Iris Mendez-Bobe for their technical assistance.