The selective alpha(2) agonist dexmedetomidine may improve neurologic outcome after incomplete ischemia in animals when it is administered either before or after the start of the ischemic insult. To clarify further the mechanisms by which alpha(2) agonists may provide neuroprotective effects, the authors tested the hypotheses that dexmedetomidine decreases synaptic and extrasynaptic glutamate release stimulated by potassium chloride or hypoxia, and decreases postsynaptic glutamate receptor activity during aerobic or hypoxic conditions.
Glutamate released from brain slices (300-microns thick) from rat hippocampus was measured in a cuvette during two experimental stresses: (1) potassium chloride-evoked depolarization (30 mM) with and without 10 nM, 100 nM, or 1,000 nM dexmedetomidine; and (2) hypoxia (95% N2 - 5% CO2) with and without 100 nM dexmedetomidine. Glutamate release was quantified by fluorescence assay using 1 mM nicotinamide adenine dinucleotide, and 5 international units per ml glutamate dehydrogenase. The formation of nicotinamide dinucleotide reduced from nicotinamide adenine dinucleotide by glutamate dehydrogenase was measured fluorometrically (excitation light 340 nm, emission intensity 460 nm) in the solution above the slice. Glutamate receptor activity was determined by the change in cytosolic calcium concentration in CA1 neurons in the presence and absence of 100 nM dexmedetomidine during administration of N-methyl-D-aspartate (100 microM) and during simulated ischemic penumbra conditions (PO2 = 20 mmHg, glutamate 3 mM). Calcium concentration was measured using a microscope fluorometer in fura 2-loaded rat hippocampal brain slices.
Dexmedetomidine attenuated potassium chloride-evoked glutamate release by 37%, 51%, (P = 0.03) and 27%, respectively, for the 10 nM, 100 nM, and 1,000 nM concentrations, and decreased (at 100 nM) the increase in glutamate release in response to hypoxia by 61% (P < 0.0001). Dexmedetomidine (100 nM) had no effect on N-methyl-D-aspartate or hypoxia plus 3 mM L-glutamate-mediated calcium changes.
The selective alpha(2)-adrenergic agonist dexmedetomidine decreases evoked glutamate release from hippocampal rat brain slices during depolarization or hypoxic stress, but does not alter calcium changes mediated by the stimulation of glutamate receptors during aerobic or hypoxic conditions.
Key words: Brain: hippocampus. Ions: cytosolic free calcium. Receptors: glutamate; N-methyl-D-aspartate. Sympathetic nervous system, alpha2-adrenergic agonists: dexmedetomidine.
NEURONAL injury produced by hypoxic or ischemic brain insult is characterized by a series of events that includes release of glutamate and other excitatory neurotransmitters and a subsequent uncontrolled increase of intracellular calcium. Among many factors potentially influencing these events is the balance between alpha1- and alpha2-adrenergic receptor activity: alpha2-adrenergic receptors hyperpolarize neurons [2-4]and decrease depolarization-mediated excitatory neurotransmitter release, whereas alpha1receptors increase the excitability of neurons. Although alpha1stimulation may be deleterious in animal models of cerebral ischemia, alpha2agonists are neuroprotective. Dexmedetomidine, a highly selective alpha2agonist, may improve neurologic outcome after incomplete ischemia in animal models of cerebral ischemia when administered either before or after the start of the ischemic insult.[7-9]Clonidine, a relatively nonspecific alpha sub 2 agonist, has shown similar benefits. Little else is known about the effect of alpha2agonists on pathologic events in in vitro models of cerebral hypoxia/ischemia.
To clarify further the mechanisms by which alpha2agonists may provide neuroprotective effects, we studied the effects of dexmedetomidine (Kifor the alpha2receptor of 1 nM) on key pathologic events in ischemia-vulnerable CA1 neurons in a well-characterized rat hippocampal brain slice preparation. The highly selective nature of dexmedetomidine enabled the study of concentrations (10-100 nM) devoid of alpha1effects, thereby avoiding a factor complicating the mechanistic study of other alpha2agonists. To examine events relevant to the pathology of cerebral ischemia, we tested the hypotheses that: (1) dexmedetomidine decreases synaptic and extrasynaptic glutamate release stimulated by potassium chloride or hypoxia, and (2) dexmedetomidine decreases postsynaptic glutamate receptor activity during aerobic or hypoxic conditions.
Preparation of Brain Slices
With approval from the UCSF Committee on Animal Research, 13-25-day-old Sprague-Dawley rats were killed by decapitation during halothane/oxygen anesthesia. Cerebral cortices were removed rapidly and immersed in ice-cold artificial cerebrospinal fluid (aCSF: Earle's Balanced salt solution; composition in mM: NaCl 116, NaHCO325, KCl 5.4, CaCl21.8, MgCl20.9, NaH2PO40.9, glucose 15, pH 7.40 bubbled with 5% CO2/95% Oxygen2). Sagittal brain slices (300-350 micro meter) were prepared immediately using a vibrating tissue slicer. To permit recovery from slicing trauma, the slices were maintained for 2 h at room temperature (approximately 25 degrees Celsius) in oxygenated aCSF before study. For glutamate release experiments, the majority of cortical tissue was removed from the brain slice, so that the slices contain mainly hippocampal tissue.
Glutamate Release Assay
Glutamate released from brain slices was quantified by fluorescence assay in a Hitachi (Tokyo, Japan) F-2000 fluorometer. Each slice was gently fixed to a mesh holder and placed in a fluorometer cuvette containing 1.6 ml 37 degrees Celsius aCSF, 1 mM nicotinamide adenine dinucleotide, and 5 international units/ml glutamate dehydrogenase. The formation of nicotinamide dinucleotide reduced from nicotinamide adenine dinucleotide by glutamate dehydrogenase was measured fluorometrically (excitation light 340 nm, emission intensity 460 nm) in the solution above the slice. A stir bar ensured rapid detection of released neurotransmitter. The temperature of the cuvette fluid was maintained at 37 degrees Celsius throughout the study. The assay was calibrated by injecting known quantities of L-glutamate into the cuvette, and permitted detection of about 1 nM (approximately 0.01 nM/s) glutamate. To ensure that nicotinamide dinucleotide reduced (or other fluorochromes) and glutamate dehydrogenase were not released from slices during hypoxia, we excluded the fluorochrome and determined whether potassium chloride-evoked glutamate release produced any change in fluorescence signal.
The glutamate release assay measures the net translocation or efflux of glutamate from the slice into the medium, and therefore reflects the balance between cellular release of glutamate and cellular uptake among the cells in the slice.
Measurements of Glutamate Receptor-mediated Calcium Changes
An index of the activity of postsynaptic glutamate receptors was obtained by measuring N-methyl-D-aspartate (NMDA)-evoked changes in cytosolic free calcium concentration. Changes in free calcium were measured in hippocampal slice CA1 neurons using a dual-excitation fluorescence spectrometer (Photon Technology International, South Brunswick, NJ) and a Nikon (Tokyo, Japan) Diaphot inverted microscope. To measure cytosolic free calcium concentration, 1 micro Meter fura-2 acetoxymethyl ester (Molecular Probes, Eugene, OR) was added to the aCSF during the 2-h slice recovery period. Fura-loaded brain slices were placed in a CSF on a nylon mesh support in a temperature-controlled (37 degrees Celsius) recording chamber (Warner Instruments, Hamden, CT) sealed at the top and bottom with glass coverslips and placed on the microscope stage. Slices were perfused intermittently with 37 degrees Celsius oxygenated aCSF. The CA1 pyramidal cell region of the dorsal hippocampus was viewed using a 20X fluorescence objective. An optical aperture in the photometer was adjusted to allow only light from the CA1 neurons to reach the photomultiplier tube. Before reaching the photomultiplier, light passed through a 510-nm narrow-pass filter. Before data collection, a reticle in the viewing lens was used to adjust the photometer slit aperture to a reference size. The total photon counts were recorded at 345 nm excitation at this reference slit size, then the slit aperture was increased or decreased to give 1.0 x 106photons/s. The factor by which the reference aperture counts differed from 1.0 x 106was used to correct the background signal for excitation area in each slice. Background signalswere obtained at both excitation wavelengths in undyed slices undergoing an identical series of exposures to agonists, hypoxia and other study conditions. The background correction eliminated fluorescence changes from hypoxia-generated nicotinamide dinucleotide reduced.
Slices were alternately excited with 345 and 380 nm light, and emitted light intensity at 510 nm was recorded. The ratio of these two background-corrected signals is a well-accepted index of relative changes in free calcium concentration. Reporting calcium changes based on differences in the fura-2 emission intensity ratio eliminates errors due to problems accurately determining fura fluorescence at saturating and zero calcium concentrations. The method is independent of fura-2 loading, dye loss, and photobleaching, and is particularly accurate for detecting changes in [Calcium2+]cnear the KD(dissociation constant) of fura-2 (225 nM).
Glutamate efflux from brain slices was measured during two experimental stresses: (1) potassium chloride-evoked depolarization, in which synaptic glutamate release was induced with potassium chloride (final cuvette concentration 30 mM); and (2) hypoxia, for which slices were placed in aCSF equilibrated with 95% Nitrogen2/5% CO2(P sub O2approximately equal 20 mmHg, PCO2= 40 mmHg, pH 7.35-7.45), i.e., 95% Nitrogen2-5% CO2was used to fill the head space of the cuvette and the fluorometer analysis chamber. For all stresses, the appearance of glutamate in the cuvette was followed for a 5-10-min period, until the rate of formation was stabilized or a plateau was reached. Before study of effects of dexmedetomidine on glutamate release, slices were incubated for 15-25 min in a beaker of oxygenated aCSF containing 10 nM, 100 nM, or 1,000 nM dexmedetomidine for study of potassium chloride-evoked glutamate release and 100 nM for study of hypoxia-evoked glutamate release. To minimize loss, dexmedetomidine was added to the study cuvette immediately before the slice.
For NMDA-stimulated cytosolic calcium concentration measurements, each slice was preincubated for 15-25 min in an oxygenated, fura-2 free, aCSF solution containing 1 micro Meter tetrodotoxin, 1 micro Meter omega conotoxin with or without (control) 100 nM dexmedetomidine. After transferring fura-loaded slices to a 37 degrees Celsius recording chamber, baseline fluorescence recordings were obtained for 50 s. To assess the activity of the NMDA receptor, perfusate containing 100 micro Meter NMDA with or without 100 nM dexmedetomidine was introduced in the recording chamber at a flow rate of 4-5 ml/min. Calcium changes were measured continuously for 15 min. Calcium changes under the conditions of the study primarily reflect events caused by NMDA receptor ion channel activation, because conotoxin and tetrodotoxin were used to prevent calcium-coupled synaptic glutamate release and resulting nonspecific stimulation of other glutamate receptor subtypes.
For hypoxia- and glutamate-stimulated changes in cytosolic calcium concentration, recordings of cytosolic calcium were made in hippocampal CA1 neurons under conditions simulating those thought to occur in the peripheral areas of cerebral infarcts. The test perfusate contained 3 mM L-glutamate (pH 7.35-7.45) with PO2[nearly equal] 20 mmHg, PCO2= 40 mmHg. The slices were pretreated as previously described and the test perfusate contained either 100 nM dexmedetomidine or plain aCSF. Measurements were obtained for 50 s before the change in the perfusion media, and continued for 15 min.
Data are reported as the mean plus/minus SD. P < 0.05 identified statistical significance. Data were analyzed using paired and unpaired t tests for comparisons between specific treatment conditions. Statistical calculations were performed using StatView 4.02 (Abacus Concepts, Berkeley, CA) software.
Effects of Dexmedetomidine on Evoked Glutamate Release
Ten (n = 8), 100 (n = 8), and 1,000 nM (n = 8) dexmedetomidine reduced potassium chloride-evoked glutamate release by 37%, 51% (P = 0.03), and 27%, respectively, compared with control (n = 8; Figure 1, top). Dexmedetomidine (100 nM) also decreased by 61% (P <0.0001) the increase in glutamate release in response to hypoxia (n = 8; Figure 1).
Effects of Dexmedetomidine on N-methyl-D-aspartate Receptor-mediated Calcium Changes
Relative concentration of intracellular free calcium in oxygenated slices (PO2> 400 mmHg, PCO240 mmHg) was not altered by incubation in dexmedetomidine. When unstimulated, cytosolic calcium in CA1 remained stable for > 30 min.
In oxygenated medium, 100 micro Meter NMDA increased relative cytosolic calcium concentration in CA1 by approximately 31%, with a plateau reached in 8-10 min (Figure 2) An early increase in calcium was followed by a gradual increase during the next 8-10 min. Dexmedetomidine (100 nM, a concentration at which alpha2agonist effects predominate) had no effect on NMDA-mediated calcium changes.
Effects of Dexmedetomidine on Calcium Changes during Hypoxia/Glutamate
With hypoxia (PO2approximately 20 mmHg) plus 3 mM L-glutamate, cytosolic calcium concentration in CA1 cell bodies increased gradually (Figure 3) to approximately twice basal concentration over the next 8-10 min. Dexmedetomidine (100 nM) did not change the pattern or extent of calcium changes observed.
Using a well-characterized rat hippocampal brain slice preparation, we have found that the highly selective alpha2-adrenergic agonist dexmedetomidine decreases evoked glutamate release during depolarization or hypoxic stress, but does not alter calcium changes mediated by the stimulation of glutamate receptors during aerobic or hypoxic conditions.
Accumulation of high concentrations of glutamate in the extracellular space is a key pathologic event in cerebral hypoxia/ischemia. By reducing the amount of synaptically or extrasynaptically released glutamate, alpha2agonists might reduce the cascade of harmful events resulting from glutamate receptor hyperactivation. Our results demonstrate that dexmedetomidine reduces depolarization (potassium chloride)-induced glutamate release. This effect may be explained, in part, by the G-protein coupled inhibitory relationship between alpha2-adrenergic receptors and N-type voltage-gated calcium channels ; the latter confer protection from delayed neuronal necrosis. Additionally, hyperpolarization in CA1 neurons or neurons in the locus coeruleus [17,18]caused by G-protein-regulated calcium-sensitive Potassium sup + channels may slow synaptic glutamate release and decrease spontaneous or induced action potential frequency. Potentially protective effects of presynaptic hyperpolarization include inhibition of glutamate-mediated excitatory neurotransmission in CA3 of hippocampus. 
Dexmedetomidine also reduced extrasynaptic glutamate release during anoxia, an effect not related to voltage-gated calcium channels. Extrasynaptic glutamate release during hypoxia is caused by reversal of glutamate transporters caused by ion gradient collapse and is a mechanism distinct from synaptic release processes. Blockade of synaptic voltage-gated calcium channels prevents depolarization-evoked, but not hypoxia-evoked, release in this preparation. Therefore, to reduce hypoxia-evoked release, dexmedetomidine may act on processes that stabilize the ion gradients necessary for normal function of the cell membrane glutamate transporters.
Although dexmedetomidine attenuated potassium chloride-evoked glutamate release, our studies did not examine the effect of dexmedetomidine on cellular survival. Because of the multitude of mechanisms leading to neuronal injury during cerebral ischemia, it is possible that decreasing, or even abolishing, glutamate release may not be sufficient to prevent cellular damage. Further work is needed to evaluate the clinical significance of our findings.
Intracellular Calcium Changes
A major finding of this study is that dexmedetomidine does not decrease glutamate receptor activity or calcium changes during anoxia. A decrease of NMDA-mediated calcium changes was expected because hyperpolarization of CA1 neurons decreases NMDA activity by a voltage-dependent process (the Magnesium2+ block of the NMDA receptor is voltage-sensitive ). In our brain slice model, agents that hyperpolarize CA1 neurons have been found to decrease NMDA receptor-mediated calcium fluxes. 
Calcium changes evoked by anoxia, or even by glutamate, are derived from many different processes in our brain slice model, including calcium entry via glutamate receptors, calcium channels, membrane damage, and release from intracellular calcium stores. Calcium changes owing to glutamate receptor activation during hypoxia in our model are relatively small and derived mainly from hypoxia-induced adenosine triphosphate loss and membrane damage. The effects of alpha2agonists on the various factors that control calcium levels in neurons during hypoxia have not been investigated in detail.
The lack of effect of dexmedetomidine on hypoxia or glutamate-mediated calcium changes need not imply lack of protective effects during or after ischemia in vivo. The injury cascade that occurs during cerebral ischemia is complex, and agents that prevent or reduce distantly related parts of the cascade may be protective. A further complication is that some agents appear to be protective by interrupting processes occurring after the ischemic insult and during the evolution of injury (e.g., N-type voltage-gated calcium channel toxins). Thus, it is possible that dexmedetomidine may be protective in in vivo animal models of cerebral ischemia because of effects that occur many hours after the acute insult.
Alpha-Adrenergic Receptors in Cerebral Ischemia
Alpha2agonists may improve neurologic outcome after incomplete ischemia in animal models when administered either before or after the start of the ischemic insult. [7-9]This is a dose-dependent effect, reversible by alpha2antagonists, and correlated with the level of circulating norepinephrine. [7,25]Additionally, some alpha2receptor-mediated effects are reversible by increased alpha1receptor activity. Thus, the balance of alpha sub 2 /alpha1receptor activity and the role of norepinephrine in neuronal injury warrant attention.
During cerebral ischemia, norepinephrine is released in large quantities to the extracellular fluid in the hippocampus, and may increase neuronal injury by enhancing neuronal excitability and energy consumption. [26,27]Norepinephrine also facilitates neuronal responses to subthreshold concentrations of excitatory transmitters via alpha1receptor-mediated effects, and enhances glutamate response by beta1receptor-mediated effects. alpha2Agonists attenuate the ischemia-induced accumulation of norepinephrine in the hippocampus, an action that may directly and indirectly contribute to the neuroprotective effect of alpha2agonists. Reversal of the protective effect of low extracellular norepinephrine concentrations by systemic administration of norepinephrine in in vivo studies supports the role of norepinephrine in neuronal injury. 
The balance between alpha2and alpha1receptor activity, whether owing to exogenous compounds acting on adrenergic receptors or endogenous norepinephrine, may be significant in mediating the balance between protection or destruction of neurons during ischemia. alpha2Agonists or low concentration of norepinephrine may attenuate the release of glutamate and the deleterious activity of glutamate during ischemia, whereas alpha1receptor activity may overcome these effects and facilitate glutamate-mediated effects. alpha sub 2 Agonists may be protective of neurons via both presynaptic and postsynaptic actions, and, in this way, differ from glutamate receptor antagonists. Therefore, combinations of drugs that stimulate alpha2receptors and decrease norepinephrine and glutamate release should be studied as potential therapeutic agents for cerebral ischemia.
Critique of Methods
Hippocampal CA1 and CA3 neurons are particularly vulnerable to ischemic damage and therefore serve as a valuable model for testing neuroprotective efficacy. The CA1 neurons studied contain abundant alpha sub 2 and alpha1receptors, but the balance of adrenergic innervation varies with brain area. Thus, generalization of the current results to other brain areas is difficult.
As mentioned earlier, the current studies cannot be used to define precisely which mechanisms of cytosolic calcium homeostasis are influenced by alpha2agonists during the stress of hypoxia and high glutamate exposure.
The use of brain slice preparations as a model for understanding the pathology of cerebral ischemia has limitations. These include a variable injury layer in each slice and uncertainty concerning the degree to which biochemical alterations in the slice during hypoxia/ischemia mimic those in vivo. Thus, the extrapolation of in vitro results to intact animals must be made cautiously if at all.
Although the glutamate release assay is commonly used, the continuous conversion of glutamate to alpha-ketoglutarate may prevent accumulation of glutamate within the slice. This experimental condition is different from the in vivo situation.
Because of the high affinity and selectivity of dexmedetomidine to the alpha2-adrenergic receptor, we assume that our findings are mediated via the alpha2-adrenergic receptor. However, we cannot exclude a contribution from other receptors to which dexmedetomidine has weak binding affinity, such as alpha2-adrenergic or imidazole receptors.
We have demonstrated that the alpha2agonist dexmedetomidine decreases glutamate release from hippocampal rat brain slices, and has no effect at the postsynaptic NMDA receptor-mediated changes in intracellular calcium concentrations. Further work is needed in this area to determine the relevance of our findings to the in vivo effects observed with alpha2-agonists during cerebral ischemia.
Abbott Laboratories supplied the dexmedetomidine, and the authors thank B. M. Hansen for her technical assistance and Winifred von Ehrenburg, Ph.D., for her editorial assistance.