Glutamate-induced astrocytic metabolism influences the cytosolic redox

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Cellular metabolism is intimately associated with intracellular oxidation-reduction (redox) balance. Here, we describe cytoplasmatic redox changes in cultured hippocampal neurons exposed to glutamate. Neurons were transfected with HyPer, a genetically encoded redox biosensor for hydrogen peroxide which allows real-time imaging of the redox state. HyPer increases its signal with hydrogen peroxide in a dose-dependent manner meanwhile the decay of the fluorescence informs about the reducing capability of the cytoplasm. The rate of decay was found to be augmented by glutamate (10 M) as well as by pharmacological stimulation of NMDA glutamate receptors. Acute chelation of extracellular Ca2+ abolished the glutamate-induced effect observed on HyPer fluorescence. Further experiments indicated that mitochondrial function and hence energetic substrate availability commands the redox state of neurons and is required for the glutamate effect observed on the biosensor signal. Finally, our work pointed out that astrocytic metabolism is involved in the changes of neuronal redox state observed with glutamate.


The brain is one of the most metabolically expensive tissues of the body. It possess a high rate of glucose and oxygen consumption, although corresponding only to 2% of the whole body weight [1,2]. The metabolic demand of the brain shows regional and spatial patterns determined by neuronal transmission, a feature that has been exploited in clinical diagnosis and basic research.

At resting conditions, brain cells supply their energy demand from glucose as the unique metabolite available for oxidation [3]. A different scenario is established when glutamate, the main excitatory neurotransmitter is released to the synaptic cleft. Initially, a high energy demand occurs in brain cells due to the recovery of ionic gradients dissipated during the synaptic transmission. For instance, astrocytes respond to glutamate by stimulating their glucose transport and glycolytic flux to ensure continuous ATP production, a phenomenon coupled to Na+ gradient-driven glutamate reuptake and the plasma membrane Na+/K+ ATPase pump activity [4-8]. On the other hand, neurons apparently do not to activate their glycolytic flux under neurotransmission. Recent evidence provided by Chuquet et al. indicate that neurons from the barrel cortex do not increase their glucose metabolism as astrocytes do upon whisker stimulation in rats [8], confirming differential patterns of glucose metabolism observed before in primary cultures and brain slices [7,9]. These findings indicate that neurons, despite their higher energetic demand [10], do not obtain the energy from metabolizing glucose, they rather prefer an alternative carbon source to fuel the costs of synaptic transmission, likely lactate. Indeed, many reports in vitro have shown that lactate and glucose are equally effective to sustain synaptic function [11-13]. All these findings fit well with the idea proposed by Magistretti et al. where glutamate increases astrocytic glycolysis yielding a net release of lactate to the extracellular space, which is taken up by neurons to be oxidized by the mitochondria. This idea is better known as the astrocyte to neuron shuttle hypothesis, ANLSH [4].

In the brain, blood-borne glucose metabolism produces the primary energy source, adenosine triphosphate (ATP), and reduced nicotinamide adenine dinucleotide (NADH). Oxidative phosphorylation of glucose-derived metabolites, either lactate or piruvate, also render NADH and ATP in the Krebs cycle (referencia libro). Alternatively, glucose-6-phosphate (glc-6-P) can also be metabolized by pentose phosphate pathway to produce biosynthetic molecules and reduced nicotinamide dinucleotide phosphate (NADPH), the cofactor necessary for the regeneration of reduced glutathione (GSH) by glutathione reductase [14-16]. Despite of NADH and NADPH share similar redox characteristics their functions are divergent. Whereas NADH is advocated to energetic functions, NADPH seems to be destined to cellular redox functions. NADPH is not only essential for the regeneration of all antioxidant defense systems, such as GSH, thioredoxins and peroxiredoxins, but also participates in detoxification with cytocrome p450 and in the "oxidative burst" mediated by NADPH oxidase in immune cells [17-19]. These compartmentalized roles find convergence with the NAD+ kinase (NADK), which phosphorylates NAD+ and has shown to exert control over the cellular content of NADPH in mammalian cells [17,20].

The intracellular redox state is basically determined by the ratio of redox-active pairs NADH/NAD+, NADPH/NADP+, GSH/glutathione disulfide (GSSG) and thioredoxin/oxidized thioredoxin. Among those pairs, GSH with an abundance of 2 to 3 mM and a relative ratio of GSH/GSSG around 100:1 is the major anti-oxidant in the cells brain [21]. Healthy cells maintain the cytoplasmatic environment at reducing potential of -250 mV, this balance is held despite of the endogenous generation of reactive species of oxygen (ROS) generated by mitochondria as by products from reduction of O2 and H2O during oxidative phosphorylation. Moderate shifts toward more oxidative potentials are thought to promote cell differentiation. For instance, Sundaresan and colleagues demonstrated that increases in hydrogen peroxide were essential for signaling induced by platelet derived growth factor on vascular smooth cells [22], many other reports have established a physiological role for ROS in cellular events as well [23], for review refer to [24]. More severe increases in ROS formation become harmful to cellular components. DNA damage, lipid peroxidation and unspecific thiol oxidation of proteins are known factors which ultimately affect structural and enzymatic functions leading to cell collapse and death (ref). Together, the evidence indicates that redox state is a dynamic variable in the cell and can be affected by metabolic preferences and signaling triggered by external factors such as growth factors.

Remarkable improvements in measuring fluctuations in the redox state in real time mode have been possible thanks to fluorescent protein-based redox probes. Basically, green and yellow fluorescent proteins (FPs) have been chosen as template to place cysteine pairs able to form disulfide bonds under oxidative environments, which in turn produce changes in the fluorescent spectrum of the protein, giving the property to follow changes in the redox state with changes in fluorescent intensity of FPs, revised by Meyer and Dick[25]. Moreover, genetically encode biosensors have been destined to organelles in order to obtain real time measurements of redox state of endoplasmic reticulum lumen [26] and mitochondria [27,28]. Herein, we took advantage of the fluorescent protein-based redox biosensor, HyPer, which was designed to monitor specifically intracellular hydrogen peroxide and allows ratiometric measurements [27]. However, once HyPer molecule is oxidized, its disulfide bond is susceptible to be reduced back to free thiol groups by the action of reducing mechanisms present in the cell rather than a mere hydrogen peroxide clearance. Thus, HyPer does not only informs about intracellular hydrogen peroxide formation, but gives valuable information about the reducing capacity of the environment as well. Based on those properties, we studied the changes in the reducing cytosolic capacity of neurons submitted to energy demand conditions induced by glutamate, this excitatory neurotransmitter alters brain cell metabolism and in consequence, redox state changes are expected to occur according to metabolic substrate availability.

Materials and methods

MReagents and plasmids. DL-Lactate, sodium piruvate, gramicidin, sodium monensin, hydrogen peroxide, reduced glutathione and standard chemicals were purchased from Sigma (St. Louis, MO). Fetal bovine serum, Minimal Essential Medium, Trypsin, penicillin-estreptomycin, Glutamax, Neurobasal and B27 supplement and fluorescent probes like Fluo-3 AM were purchased from Invitrogen (Carlsbad, CA, USA). pHyPer-cito plasmid was purchased from Evrogen JSC (Moscow, Russia).

Brain cell culture. Sprague-Dawley rats were obtained from the Universidad de Chile. Mixed cultures of neuronal and glia cells were prepared from brains 1-3 days old neonates following the same indication as in Loaiza et al. 1 with minor modifications. In brief, hippocampi were dissected in Hank´s buffer, incubated in 1.25% trypsin for 10 min at 37°C, and mechanically dissociated in MEM-10% fetal bovine serum: The cells were plated on poly-L-lysine coated (0.1 mg/ml) coverslips and maintained at 37°C in a humidified atmosphere with 5%CO2/95% air. After one hour, the media was replaced by Neurobasal/B27 and the cell preparation was maintained by renewal of media every 3 days. Experiments were performed after 7 to 14 days in KRH buffer (in mM, 140 NaCl, 4.7 KCl, 20 HEPES, 1.25 MgSO4 and 1.25 CaCl2.).

PC12 Cell culture. Rat pheochromocytoma PC12 cells (from American Type Culture Collection) were cultured in DMEM supplemented with heat-inactivated 10% horse serum, 5% FBS, 50 units/mL penicillin and 100 mg/mL streptomycin (Invitrogen, Carlsbad, CA). PC12 differentiated neurons were obtained by plating cells onto poly-L-Lysine coated coverslips in the presence of 50 ng/mL NGF (Sigma-Aldrich, St Louis, MO) for a period of 5 days in DMEM medium supplemented with 1% heat-inactivated horse serum and 1% FBS.

HyPer imaging. Cultured cells were transfected with pHyPer-cyto, a plasmid encoding a specific hydrogen peroxide sensor with cytosolic expression for mammalian cells. The transfection method used was either lipofectamine 2000 or calcium phosphate, both rendering similar percent of neuronal cells expressing HyPer. The ratiometric measurement of hydrogen peroxide was achieved in a Olympus microscope equipped with a spinning disk (DSU Model). The preparation was excited at 403/12 nm and 480/20nm (MT20E emission wheel filter) and emission was detected at 555/28 nm with a CCD camera.

Calcium Imaging


Glutamate induces a rapid decrease in HyPer fluorescent signal in hippocampal neurons.

The cytoplasmatic redox state was monitored by means of HyPer, a genetically encoded fluorescent indicator which has been proven to be specific to hydrogen peroxide and no other radical oxygen species [27]. HyPer ratio increased in a dose-dependent manner showing saturation at H2O2 doses over 50 M. In addition, sequential pulses of hydrogen peroxide (100 M) induced comparable increments in the ratio fluorescence although with some loss in the signal amplitude during the time course (Supplemental figure 1). Also it can be noted that the biosensor turns spontaneously back to pre-pulse state. Complementarily, the addition of exogenous glutathione (1 and 5 mM) decreased the HyPer signal at basal level as well as from a pre-oxidized state, (Supplemental figure 1).

In order to investigate the impact of glutamate on the reducing capability of neurons, a pre-pulse of H2O2 100 M was applied to oxidize the biosensor, after this procedure the fluorescence started to diminish with a constant rate (VALOR) named here as (slope a, figure 1A). The addition of glutamate (10 M) accelerated the decay of the signal (slope b, figure 1A). This effect was observed in ~80% of the neurons studied and in average, the neurotransmitter increased the slope in almost seven times.

At this point, is important to clarify that the reduction in the fluorescence from the oxidized HyPer molecule does not obey to a cytoplasmatic clearance of hydrogen peroxide, but rather to reduction of disulfide bonds, as suggested by Meyer and Dick and also, confirmed by the pre-treatment with the catalase inhibitor, 3-amino-triazol, which did not interfere with the glutamate effect on the HyPer signal (data not shown).

Next, we tested if the pharmacological stimulation of the NMDA-sensitive glutamate receptors could mimic the glutamate effect observed above. As shown in the figure 1B and C, NMDA was equally effective as glutamate to decrease the Hyper signal in a neuron previously treated with hydrogen peroxide. Moreover, the pretreatment of ionotropic glutamate receptors blockers, amino-5-phosphonovaleric acid and 6,7-dinitroquinoxaline-2,3-dione, avoided the glutamate-dependent HyPer signal diminishment (figure 1C). Together, the results suggest the glutamate effect on the redox biosensor is mediated by the activation of NMDA-sensitive glutamate receptors.

Due to ionotropic glutamate receptors activation, and in special NMDA-sensitive receptors, would lead to a sodium and calcium influx, we investigated if inducing a glutamate-independent sodium entry could trigger a reduction in the biosensor fluorescence. First, neurons were exposed to monensin, a Na+/H+ antiporter, such treatment provoked a rapid diminishment in the HyPer signal (figure 2A). In excitable cells, like neurons, membrane potential can be perturbed by discrete cation entry, which in turn will open voltage dependent calcium channels, among others voltage dependent channels. Considering this point, cytosolic calcium was imaged in neurons exposed to monensin to confirm cytosolic calcium perturbations by monensin. The figure 2C shows that a massive increase in the cytoplasmatic calcium occurred upon monensin exposure, leaving the calcium signaling as a valid alternative to explain the effect of this sodium ionophore. Of the note, the efficiency of monensin to permeate Na+ and then, to trigger a cytoplasmatic calcium increase in neurons, should be accompanied by a cytoplasmatic proton extrusion. The predicted alkalinization of cytoplasm was confirmed by loading neurons with the cell permeant pH-sensitive dye BCECF, monensin induced a rapid and transient alkalinization as it can be observed in the supplementary figure 2. Alternatively, neurons were exposed to veratridine, an alkaloid that interacts with voltage-dependent Na+ channels in their open state, which induces a gradual sodium influx with the concomitant cytosolic calcium increase [9], but without the pH disturbances generated by monensin. This compound was as effective as monensin to accelerate the reduction in the HyPer signal (see the figure 2B and C).

To test directly the role of calcium influx in the glutamate effect, we performed experiments with an extracellular buffer without calcium and supplemented with the calcium sequestering agent, ethylene glycol tetraacetic acid (EGTA, 5 mM). Under nominal zero calcium condition, glutamate was not able to evoke a diminishment in the HyPer fluorescency. Moreover, the neurotransmitter recovered its capability to affect HyPer signal when calcium was placed back to the extracellular media (Figure 3). Together, these results indicate that a calcium influx is a key cellular messenger in the glutamate-evoked reduction of the biosensor HyPer at the cytoplasmatic neuronal environment.

Astrocytic metabolism is connected with the neuronal redox status

Intriguingly, the decrease in the biosensor fluorescency either induced by a calcium influx with ionophores or by exposure with the neurotransmitter glutamate, could be observed in the absence of any metabolic fuel (e.g. glucose 5 mM). Considering that neurons are not well equipped with energetic reservoirs, it is reasonable to assume that after ten minutes without glucose the glycolitic flux has diminished to negligible level. However, mitochondria still can be producing reducing power by feeding Krebs cycle with intermediaries. To asses that the mitochondrial machinery was involved in the biosensor signal diminishment, the preparation was pre-incubated with the mitochondrial poison, Antimycin A. The figure 4 shows that under this condition, glutamate was not able to affect the fluorescence of the biosensor. This finding unveils that mitocondrial function is necessary to observe the glutamate effect on the HyPer, but which metabolite or from where it comes, are not informed.

The co-culture modality used here allowed us to investigate if astrocytic metabolism is connected in some way with neuronal redox status. In order to impair only the glial metabolism, cells were treated with fluoroacetate (5 mM) for at least 2 hours before the imaging procedure. This compound is only metabolized by astrocytes, where it is converted in fluorocitrate, which finally exerts the toxic action specifically on the aconitase enzyme halting the Krebs cycle [29,30]. In order to ensure the glycogen consumption by astrocytes, thirty minutes before the experiment started, glucose was washed out and the experiment was carried out. The figure 5A shows a time course of HyPer signal recorded from cells which were energetically deprived with fluoracetate, such condition was effective in avoid the glutamate effect on HyPer described before, no significant changes in the basal rate of HyPer recovery before and after glutamate exposure were detected (figure 5C), suggesting that the astrocytic metabolic machinery is pivotal in the HyPer ratio decrease. These results suggest an intercellular transfer from astrocytes to neurons, where the Hyper molecule is located.

In order to eliminate such transferring, the total eradication of glial population from our cultures had been ideal to discard any astrocytic contribution to the glutamate effect on the biosensor, but using cytosine arabinofuranoside, an agent widely used to enriched neuronal cultures is not the best option because it does not guarantee the total astrocytic elimination, but also provokes neuronal toxicity as an undesired secondary effect [31]. Alternatively, we took advantage of PC12 cells, a cell line which can be differentiated to a neuronal phenotype with NGF [32], which lack of any other cellular type that could interfere through intercellular metabolic trafficking. In the figure 5B, dual monitoring of cytosolic calcium and HyPer fluorescency was achieved by loading the preparation with the calcium dye FuraRed, which respond reciprocally to calcium increases. As a general observation, the typical dose of glutamate (10 M) used during this study was ineffective to rise the calcium in differentiated PC12, thus a higher dose (100 M) was applied to evoke a calcium increase, see figure 5. Even when a calcium increase was evoked, likely by NMDA activation [32], no deflection in the HyPer signal was recorded, indicating that not only the glutamate-sensing machinery on the PC12 cells is required to modify the redox environment at the cytoplasm, but an external factor is necessary to evoke a reduction in the biosensor state.


Many cellular changes take place in neurons upon glutamate arrival. From those, changes in the metabolic preference of neurons have been a controversial point since the ANLS hypothesis proposed in 1994. Since then, several reports have given support to the idea that brain cells adapt their metabolism under glutamatergic neurotransmission (montón de referencias). Here, we have shown that neuronal redox state is another cellular aspect which is modulated by glutamate and also that, changes in the reducing capability of neuronal cytoplasm depend on the astrocytic metabolism.

To our knowledge, this is the first time that the YPF-based redox sensor, HyPer, has been used to monitor the redox state in neurons. The experimental strategy consisted in pre-oxidize the biosensor with a hydrogen peroxide pulse in order to monitor the rate of fluorescent ratio decay as a reflex of the reducing capability of the cytoplasm, where the sensor was located. The intracellular reduction of HyPer molecule, and specifically its disulfide bond, is not likely related to hydrogen peroxide clearance nor catalase activity, this process is rather governed by glutathione-mediated reduction mechanisms [25]. A recent work with roGFP1 in neurons, a redox biosensor that like HyPer, also has a cysteine pair to gain sensitivity to oxidation/reduction reactions [33,34] showed no dramatic changes in its baseline fluorescence nor in its response to hydrogen peroxide pulse (only 20% increase in oxidation degree) by chronic treatment with 3-amino-1,2,4-triazole, reinforcing the idea that the intracellular clearance of the oxidant agent has little impact in the reduction of biosensors based in cysteine pairs (funke 2011).

There is an overall consensus about that excitatory neurotransmission is highly expensive in terms of energy. Most of the ATP is consumed by neurons to power ionic pumps in order to restore the dissipated ionic gradients [10,35]. Glucose and lactate could provide the sufficient energy to allow release and reuptake of neurotransmitter in isolated nerves [36] and also, allow synaptic plasticity in hippocampal brain slices [12,37]. However, recent in vivo evidence obtained from rat sensorial cortex indicates that neurons do not modify their glucose uptake during whisker stimulation as astrocytes do [8]. This finding, together with previous work done in cultured brain cells and brain slices [5,7], supports the necessity of an alternative fuel for neurons during activity, most likely, lactate (first proposed by Magistretti and al. [4]). In consequence, neurons should metabolize lactate to piruvate, a process catalyzed by the enzyme lactate dehydrogenase, which renders a NADH molecule. Then, piruvate would be further oxidized by mitochondria in the Krebs cycle, where electrons will finally reduce O2 to H2O producing more NADH and ATP in the process, making neurons more aerobic than astrocytes during neurotransmission. Accordingly with this vision, our experiments indicate that the acute changes in redox state of neuronal cytoplasm toward more reducing potential have a metabolic origin, they were dependent of the mitochondrial function and also, were sensitive to the specific impairment of the astrocytic metabolism. Firstly, we postulate that mitochondrial function, stimulated by exogenous piruvate or lactate, increase reducing potential not only into mitochondrion, but also in the cytoplasmatic compartment. The redox communication between these two compartments has been found in HeLa cells, where mitochondrial depolarization was accompanied temporally with the cytosolic GSH oxidation, both cellular parameters measured by means of TMRM and glutaredoxin-roGFP biosensor fluorescency, respectively [28]. Another experimental evidence of the mitocondrial function impact over cytoplasm redox state was found in cultured cortical neurons. Here, the authors demonstrated that extracellular glucose is necessary for cytoplasmatic oxidation of dihydroethidium by NMDA. Moreover, the inhibition of mitochondrial function increases the dye oxidation and oppositely, piruvate 1mM avoided the effect of NMDA when was applied in the absence of extracellular glucose [38]. The mechanisms implicated in the redox transfer between mitochondrial to cytosolic compartments are varied [39]. Moreover, the presence of enzymatic mechanism that interchanges NADH for NADPH in eukaryotic cells [20] plus other sources of NADPH generation besides the classical pentose phosphate pathway make the redox cellular scene notably complex [17].

Glutamate and NMDA excitotoxic vision

low dose of glutamate

short times of exposure,

rapid effect

Our experimental model to work

Coexistence of astrocytes and neurons, advantages of this model : astrocytes are necessary for neuron maturation (cholesterol, synapsis maturation).

Use f fluroacetate .Discussion :The energetic depletion protocol has been used before to prove the astrocytic lactate transferring a slice model, where the diffusion is a factor that could impair the effectiveness of this maneuver. Matthew Parsons 2010, j neurosci vol 30.

pH issue

relatively minimal acidification with glutamate 10 uM, only 0.1 pH units

even when is impossible to exclude the indirect protons transferring in the "lactate delivery", reducing shift of HyPer were recorded in pH clamp experiments with NMDA.

It is worth to mention that the glucose removal induces a rapid drop in the intracellular pH, the magnitude reaching roughly 0.4 point of pH, probably mediated by Na/K pump, due to the inhibition of that entity generates an acidification in buffer HEPES on cortical neurons, experimental conditions comparables to ours. Toxicon

Volume 50, Issue 4, 15 September 2007, Pages 541-552. However, this experimental maneuver did not affect notoriously the HyPer signal in.

another argument is respect to the lack of effect in pc12 cells at glutamate 100 uM, even when these cells count with the glutamate transport capability and in consequence the chance to acidify their cytoplasm with the neurotransmitter outside (j neurosci 2005-25(26) 6221 and Science 2000 289 (957-960)

Sodium piruvate is transported by the facilitative monocarboxylate carriers in its protonated form with the concomitant acidification of cytoplasm, this scenery that could interfere with HyPer sensing capability, an issue that was also exposed before [25]. To rule out the intracellular acidification as the cause of an impaired H2O2 sensing of HyPer molecule, intracellular pH was clamped to 7.1, such value is close to the acidification observed by adding extracellular lactate 10 mM at two minutes (not shown). Under controlled pH, hydrogen peroxide pulse of 100 M increased the Hyper ratio in about 3 units, an increment comparable to the measurements obtained in standard conditions, see supplemental figure 1. This observation allows us to establish that the sensing properties of the HyPer are not affected by such moderate acidification.