Blood-mediated scavenging of cerebrospinal fluid glutamate

Authors


Address correspondence and reprint requests to V.I. Teichberg, Department of Neurobiology, The Weizmann Institute of Science, Rehovot 76100, Israel. E-mail: Vivian.teichberg@weizmann.ac.il

Abstract

The maintenance of brain extracellular glutamate (Glu) at levels below its excitotoxic threshold is performed by Glu transporters present on glia and neurons as well as on brain capillary endothelial cells which remove brain Glu into blood. The feasibility of accelerating the naturally occurring brain-to-blood Glu efflux was studied using paradigms based on the fate of Glu present in the cerebrospinal fluid or infused into the brain ventricles and monitored before, during, and after decreasing blood Glu levels with pyruvate and oxaloacetate, the respective Glu co-substrates of the blood resident enzymes glutamate–pyruvate transaminase and glutamate–oxaloacetate transaminase. Results from cerebroventricular perfusions with [3H]Glu, intracerebroventricular injections of [3H]Glu, and measurements of the basal CSF Glu levels point out to the same conclusion that the intravenous administration of pyruvate and oxaloacetate which decreases blood Glu levels accelerates the brain-to-blood Glu efflux. We conclude that the brain extracellular Glu levels can be controlled in part by the blood Glu levels. The results may provide not only a rational explanation for the inhibition of Glu release and neuroprotective effects of parentally administered pyruvate in hemorrhagic shock and forebrain ischemia but could also outline a potential strategy for the removal of excess Glu in various neurodegenerative disorders.

Abbreviations used
CSF

cerebrospinal fluid

Glu

glutamate

GOT

glutamate–oxaloacetate transaminase

GPT

glutamate–pyruvate transaminase

ISF

interstitial fluid

PBS

phosphate-buffered saline

PCA

perchloric acid

In recent years, great progress have been made in the understanding of the ways brain protects itself from the potential neurotoxic effects of the major excitatory neurotransmitter glutamate (Glu). Particular emphasis was put on members of a large family of Glu transporters present both on nerve terminals and perisynaptic astrocytes which bind and take up Glu and guarantee that the very high concentrations of Glu transiently present in the synaptic cleft after release (1 mm) are soon decreased in the synaptic and neighboring extracellular/interstitial fluid to concentrations (around 1 µm) at which Glu exerts neither overt excitatory nor excitotoxic activities (Danbolt 2001).

So far, little attention has been given to the Glu transporters present on brain blood vessels (O'Kane et al. 1999) and on their role in the control of extracellular Glu. Two interconnected compartments ought here to be distinguished: the vascular endothelium forming the blood–brain barrier within the brain interstitial compartment, and the choroid epithelium forming the blood–cerebrospinal fluid (CSF) barrier. In the first compartment, the average intercapillary distance (24 microns; Pawlik et al. 1981) and intersynaptic distance (0.5 microns; Kullmann et al. 1999) are compatible with the possibility of some Glu ‘spillover’ from synapses to blood capillaries. In the other, the choroid plexus, has been implicated in the active clearance of Glu from CSF to blood (Segal et al. 1990) and probably mediates the rapid brain-to-blood efflux of Glu observed by Berl et al. (1961) following intracerebroventricular Glu administration. In both compartments, the capillary Glu transporters could be involved in brain Glu detoxification in all the pathological situations in which abnormally high Glu levels are found in the brain interstitial fluid (ISF) and CSF.

O'Kane et al. (1999) have proposed a mechanism that accounts for the elimination of Glu from brain into blood in the face of the highly unfavorable concentration gradient between ISF/CSF Glu (1–10 µm) and blood plasma (40–60 µm): extracellular Glu is transported via Na+-dependent transporters located on the antiluminal membrane and accumulates into endothelial cells. When its concentration exceeds that in plasma, Glu is facilitatively transported across the luminal membrane into blood.

On that basis, we tested here the hypothesis that a larger Glu concentration gradient between ISF/CSF and blood plasma could provide an increased driving force for the brain-to-blood Glu efflux. To achieve a decrease of blood Glu levels, we made use of the Glu scavenging properties of the blood resident enzymes glutamate–pyruvate transaminase (GPT) and glutamate–oxaloacetate transaminase (GOT), which transform Glu into 2-ketoglutarate in the presence of the respective Glu co-substrates, pyruvate and oxaloacetate.

We provide now the evidence that intravenous pyruvate and oxaloacetate cause an increased brain-to-blood Glu efflux and conclude that the CSF Glu levels can be controlled in part by the blood Glu levels. The results are discussed both in light of the neuroprotective effects of parentally administered pyruvate in hemorrhagic shock (Mongan et al. 1999, 2001) and forebrain ischemia (Lee et al. 2001), as well as a prospective strategy for the removal of excess Glu in various neurodegenerative disorders.

Materials and methods

Materials

Glu, sodium pyruvate, sodium oxaloacetate, NADH, lactate dehydrogenase, and malate dehydrogenase were from Sigma-Aldrich (Rehovot, Israel). Glu dehydrogenase was from Boehringer Mannheim Biochemicals (Indianapolis, IN, USA).

L-[U-3H]Glutamic acid (42 Ci/mmole) and L-[U-14C]Glutamic acid (249 mCi/mmole) were purchased from Amersham Biosciences (Amersham, UK).

Intracerebroventricular injections

All experimental procedures were approved by the Animal Committee of the Weizmann Institute of Science. Female Sprague–Dawley rats (250–300 g) were anesthesized with an intraperitoneal injection of urethane 0.125 g/0.2 mL for 100 g body weight. Catheterization of the tail vein (for drug injections) and of the femoral vein (for blood aliquots withdrawals) were performed using PE10 polyethylene tubings linked to PE50 polyethylene tubing. All catheters were secured with 5–0 silk thread and flushed with heparin (3–5 µL of 182 U/mL). A steel cannula made out from a 27G needle was implanted in the right lateral ventricle using the following stereotactic co-ordinates: 0.8 mm posterior to bregma; 1.4 mm lateral from the midline; 4 mm below the skull surface or 3.5 mm from the dura. [3H]Glu solutions in phosphate-buffered saline (PBS) were injected into the lateral ventricle through the implanted cannula using a Hamilton syringe (25 µL) connected to PE20 tube filled with solution. A total volume of 11 µL was injected freehand in approximately 2 min. For radioactivity determination, 50-µL blood samples collected from the femoral vein were diluted in 500 µL H2O and added to 16 mL of scintillator. Measured cpm were corrected for quenching as determined by comparing the measured cpm of a set volume of [3H]Glu added to water or to diluted blood. Body temperature was maintained with a lamp and rectal temperature was monitored. Rat pulse rate was monitored using a Periflux system 500 and a laser Doppler probe placed onto the skull. Intravenous injections of pyruvate and oxaloacetate diluted in phosphate-buffered saline (PBS) were carried out at a rate of 0.05 mL/min for 30 min with a Pharmacia pump P-1 (Uppsala, Sweden). During injections and at several time points after the injections (in general, every 15 min), aliquots of 150 µL blood were removed from the femoral vein.

Ventriculo-cisternal perfusion

Ventriculo-cisternal perfusion was carried out according to the procedure described by Davson et al. (1982). Cannulas (27G) were placed in the two lateral ventricles and another into the cisterna magna. The two cannulas implanted in the lateral ventricles were connected to PE10 polyethylene tubings attached to two 5-mL syringes. The latter syringes were driven by a Harvard apparatus infusion pump to release 26 µL/min of [3H] Glu in artificial CSF (122 mm NaCl; 25 mm NaHCO3; 3 mm KCl; 1.4 mm CaCl2; 1.2 mm MgCl2; 0.4 mm K2HPO4; 10 mm HEPES; 10 mm glucose; pH 7.42). Each syringe contained 4 mL of artificial CSF, 0.2 µm[3H]Glu, and, when needed, various amounts of non-radioactive Glu. The cannula implanted in the cisterna magna was connected to a PE10 tubing with its outlet kept 17.5 cm belong the aural line. Upon infusion into the lateral ventricles, the fluid emerging from the cisterna was collected as a function of time.

Glu determination

Whole-blood and plasma samples were deproteinized by adding an equal volume of ice-cold 1 m perchloric acid (PCA) and then centrifuged at 16 000 g for 10 min at 4°C. The pellet was discarded and supernatant collected, adjusted to pH 7.2 with 2 m K2CO3, and, if needed, stored at −20°C for later analysis.

Glu concentration was measured in the supernatant using the fluorometric method of Graham and Aprison (1966) A 20-µL aliquot from PCA supernatant was added to 480 µL HG buffer containing 15 U of Glu dehydrogenase in 0.2 mm NAD, 0.3 m glycine, 0.25 m hydrazine hydrate adjusted to pH 8.6 with 1 N H2SO4. After incubation for 30–45 min at room temperature, the fluorescence was measured at 460 nm with excitation at 350 nm. A Glu standard curve was established with concentrations ranging from 0 to 6 µm. All determinations were done at least in duplicates. The results are expressed as mean ± SD. When CSF Glu levels were measured, the enzymatic cycling method of Kato et al. (1973) was applied.

Attribution of blood radioactivity to authentic [14C]Glu

The identification of the chemical identity of the radioactive molecule(s) present in blood following intracerebroventricular injection of [14C] Glu was performed as follows: collected blood (1 mL) was hemolyzed with water and deproteinized by ultrafiltration (5000 g, 2 h). The ultrafiltrate, concentrated by speedvac, was applied onto a silica gel-coated thin-layer chromatography (TLC) containing a fluorescent indicator (Silica gel 60F254; Merck, Darmstadt, Germany) along with markers of cold Glu and 2-ketoglutarate. TLC was performed using a 2 : 2 : 1 mixture of methanol : chloroform : 95% NH4OH. The plates were exposed to a storage Phosphor screen for up to 3 days and examined with a 445 SI phosphoimager (Molecular Dynamics, Sunnyvale, CA, USA). The pixel intensity values were measured for the 50 contiguous segments representing the chromatogram.

Results

Scavenging blood Glu in vitro

We first defined the in vitro conditions under which Glu levels could be reduced in plasma by activation of the blood resident transaminases GPT and GOT with their respective Glu co-substrates pyruvate and oxaloacetate. Figure 1(a) illustrates the changes of Glu blood levels upon addition, at t = 0, 15, 30 min, of 1 mm pyruvate, 1 mm oxaloacetate or their mixture. The results show that the basal levels of the blood resident transaminases GPT and GOT are large enough in order to scavenge, with pyruvate and oxaloacetate, respectively, up to 50% of Glu. Pyruvate and oxaloacetate are utilized as substrates as repetitive additions to blood are needed to produce a decline of Glu levels. They act in synergy to decrease Glu levels but their effects are not truly additive because the activity of the transaminases is limited by the build-up of 2-ketoglutarate that facilitates the back reaction of Glu formation. Addition of 15 µm pyridoxal phosphate did not contribute to a further decrease of Glu levels.

Figure 1.

In vitro scavenging of blood Glu. (a) Evolution of Glu blood levels upon supplementation at t = 0, 15, 30 min of 1 mm pyruvate (▪), 1 mm oxaloacetate (▴) or a mixture of 1 mm pyruvate and 1 mm oxaloacetate (ellipses). Glu concentration at t = 0 was 166 ± 13 µmol/L. Each point represents the average ± SD of four experiments. (b) Evolution of the Glu levels in blood cellular pool (rectangles) and in plasma (◆◆) following the repeated additions (at t = 0, 15, 30, 45 min) to blood of pyruvate and oxaloacetate (both at a final 1 mm concentration). Each point represents the average of eight experiments ± standard error of the mean.

As blood contains two major compartments, a cell compartment consisting of various cells mainly erythrocytes, leukocytes, and platelets, and a plasma compartment, we also determined the fate of Glu in these two compartments following the activation of endogenous GPT and GOT. In both cases, a similar decrease of Glu levels was observed but was faster in the cell compartment than in plasma (Fig. 1b), probably because the Glu concentration in the cell compartment (120 ± 4.9 µm) is larger than that of plasma (40.8 ± 5.8 µm). These results suggest that pyruvate and oxaloacetate are able to scavenge Glu both in plasma as well as in the blood cell compartment.

Scavenging blood Glu in vivo

Duplication in vivo of the in vitro conditions leading to a decrease in blood Glu levels, i.e activation of GPT/GOT by repeated injections of 1 mm pyruvate/oxaloacetate, failed to achieve any significant decrease of blood Glu levels. This is somehow expected as in vitro conditions are obviously not affected by pharmacokinetic parameters such as elimination rate constants, volume of distribution or renal clearance from plasma. To overcome the latter processes, higher concentrations of pyruvate/oxaloacetate had to be injected in blood in order to build up effective concentrations allowing the activation of GPT and GOT. Figure 2(a) illustrates the effects of pyruvate and oxaloacetate administered continuously through an intravenous cathether at a rate of 50 µmole/min for a duration of 30 min. The blood levels of Glu, pyruvate, and oxaloacetate were monitored in parallel.

Figure 2.

In vivo scavenging of blood Glu. (a) Evolution of blood Glu levels in vivo (◆) upon intravenous administration for a duration of 30 min (black bar) of pyruvate and oxaloacetate (50 µmole each/min). The blood levels of pyruvate (▪) and oxaloacetate (▴) were monitored in parallel. One representative experiment is presented. Symbols show averages of two independent determinations. (b) Evolution of blood Glu levels in vivo upon intravenous administration for a duration of 30 min (black bar) of pyruvate and oxaloacetate (50 µmoles each/min). Symbols show averages of two Glu determinations from seven experiments.

It can be seen that a significant build-up of both pyruvate and oxaloacetate, up to 0.45 mm, takes place after 15 min and is accompanied, in a mirror image, by a marked decrease of blood Glu. However, as soon as the administration of the GOT/GPT co-substrates is stopped, their blood concentration decreases and that of Glu increases concomitantly.

Figure 2(b) shows that, when the levels of Glu are monitored for about 200 min after the completion of the infusion of pyruvate and oxaloacetate, a clear overshoot of Glu above blood basal levels is observed, suggesting that the increase of Glu levels is not only due to the eventual reversal of the GOT/GPT reactions (as a result of the build-up of the enzymatic products, 2-ketoglutarate, alanine, and aspartate) but possibly to additional compensatory processes causing various organs to release their Glu content into plasma to adjust to the new situation. It should be mentioned that the administration of pyruvate and oxaloacetate was not accompanied by changes in the rat pulse rate, blood pressure or rectal temperature but produced a diuretic effect likely to be due to the increase of blood Na+ ion concentration caused by the injection of pyruvate and oxaloacetate as sodium salts.

Brain-to-blood Glu efflux

To follow the fate of excess Glu in the CSF, a bolus injection of 10 µCi of [3H]Glu (230 pmole Glu/10 µL) was made into rat lateral ventricles and the appearance of radioactivity in blood was followed up with time. Figure 3(a) shows that radioactivity appears in blood as soon as [3H]Glu is injected in the lateral ventricle, increases with time and reaches a plateau where it remains constant for at least 40 min. During the first phase, the initial rate of appearance of radioactivity in blood is 0.8% ± 0.1 (n = 8) of the radioactivity input in brain/per min.

Figure 3.

Brain-to-blood [3H]Glu efflux. (a) Evolution of blood radioactivity following the intracerebroventricular infusion of 11 µCi in 11 µL of [3H]Glu. Each point represents the average of four independent experiments ± SD. (b) Evolution of blood radioactivity following a bolus intravenous injection of 1 µCi [3H]Glu. A total of 0.15-mL blood aliquots were removed with time from two injected rats.

The intuitive explanation for the plateau is that it corresponds to a steady state between the brain-to-blood [3H]Glu efflux rate and the rate of disappearance of [3H]Glu from blood.

To investigate whether the radioactivity in blood can be attributed to authentic Glu, we repeated the above experiments using [14C]Glu. Blood was hemolyzed, deproteinized by ultrafiltration, and the concentrated ultrafiltrate submitted to thin-layer ascending chromatography. Figure 4 presents an image analysis of a typical thin-layer chromatogram showing the presence of a major peak at a position similar to that of authentic Glu.

Figure 4.

Thin-layer chromatography analysis of blood radioactivity. The arrows indicate the respective migration positions of non-radioactive Glu and 2-ketoglutarate.

These results suggest that authentic radioactive Glu is transported from brain to blood and is not significantly metabolized. They also confirm the results of Hosoya et al. (1999) who observed that l-Glu in the brain ISF is transported across the blood–brain barrier in an intact form.

To estimate the rate of disappearance of [3H]Glu from blood, i.e. the Glu life time in blood, we carried out bolus intravenous injections of 1 µCi radiolabeled Glu and monitored the radioactivity in blood with time. Figure 3(b) shows that, as expected, the evolution of [3H]Glu radioactivity in blood takes place in two phases: a fast phase terminating within less than 60 s which corresponds most likely to the distribution phase and a slow one that corresponds to the elimination phase. From the latter phase, one can determine the elimination half-life of [3H]Glu in blood as being 19.8 ± 0.4 min from which one can calculate that the rate of disappearance of [3H]Glu from blood. Accordingly, the equivalent brain-to-blood [3H]Glu efflux rate should be equal to ln 2/19.8 = 0.035/min and thus, is very similar to the value of 0.0346/min measured, by the brain efflux index method, upon rat intracortical injection of [3H]Glu (Hosoya et al. 1999).

Impact of blood Glu scavenging on the brain-to-blood Glu efflux

Figure 5 shows the changes of radioactivity in blood upon intracerebroventricular injection of 10 µCi of [3H]Glu followed by blood Glu scavenging. In these experiments, once a steady-state level of radioactivity was reached in blood, the Glu blood levels were transiently decreased by the intravenous administration of pyruvate and oxaloacetate.

Figure 5.

Impact of blood Glu scavenging on the brain-to-blood [3H]Glu efflux. Evolution of blood radioactivity (◆) and blood Glu levels (▪) following the intracerebroventricular infusion of 11 µCi of [3H]Glu and its modulation by the intravenous administration of pyruvate and oxaloacetate. The injection of the latter compounds was started at 55 min at a rate of 50 µmoles/min of each compound for a duration of 30 min (black bar). This regimen causes the build up of a blood concentration of 0.45 mm for both pyruvate and oxaloacetate (see Fig. 2). Each point represents the average of three independent experiments ± SD. An unpaired two-tailed Student's t-test performed for the steady-state [3H]Glu levels reached before and during the infusion of pyruvate and oxaloacetate shows a p-value of 0.002. The broken lines show the steady-state levels reached before the intravenous administration of pyruvate and oxaloacetate.

It can be seen that the changes of blood radioactivity originating from brain displays an almost mirror image to that of blood Glu. While the latter decreases by about 50% during the administration of pyruvate and oxaloacetate and then increases, the blood radioactivity increases by about 40% and then decreases.

Impact of blood Glu scavenging on the brain absorption of excess Glu

In order to provide further evidence that the intravenous administration of pyruvate and oxaloacetate that reduces blood Glu levels can influence the extracellular levels of brain Glu, we carried out experiments of ventriculo-cisternal perfusion of [3H]Glu and measured the extent of its elimination from the perfused fluid. In this paradigm, a [3H]Glu-containing solution is continuously perfused through cannulas implanted in the lateral ventricles and is collected as it emerges from a cannula implanted in the cisterna magna. The ratio R of the radioactivity input per unit volume to that of the output, provides an index of the percentage of [3H]Glu absorbed from the perfused fluid. The absorption is due to the diffusion of [3H]Glu into the brain CSF and ISF, and to its uptake into cellular compartments via the Glu transporters present on the choroid plexus epithelial cells and those associated with the antiluminal membranes of brain capillary endothelial cells. Figure 6 shows that the ventriculo-cisternal perfusion of a 30-µm Glu solution containing 0.2 µCi [3H]Glu/mL leads to a steady-state absorption of about 36 ± 5% which increases to 43.7 ± 3% (p = 0.035) upon the intravenous administration of pyruvate and oxaloacetate and decreases back towards the basal steady-state absorption level upon completion of the infusion of pyruvate and oxaloacetate. Thus, the absorption of radioactive Glu from the perfusion fluid increases while the blood Glu levels decrease.

Figure 6.

Impact of blood Glu scavenging on the brain [3H]Glu absorption. Ventriculo-cisternal perfusion of [3H]Glu and evolution of the percentage of [3H]Glu absorbed. The latter was calculated as (1-R) × 100 where R is the ratio of the radioactivity input per unit volume to that of the output collected at the cisterna magna. (a) Perfusion of a 30-µm Glu solution in artificial CSF containing 0.2 µCi [3H]Glu/mL (▪). After 60 min, an intravenous infusion of pyruvate and oxaloacetate was started at a rate of 50 µmoles/min for a duration of 50 min. A Student's t-test performed for the steady-state values reached before and during the infusion of pyruvate and oxaloacetate shows a p-value of 0.008. (b) Perfusion of a 250-µm Glu solution in artificial CSF containing 0.2 µCi [3H]Glu/mL (▴). The perfusion was carried out at a flow rate of 26 µL/min. After 55 min, an intravenous infusion of pyruvate and oxaloacetate was started at a rate of 50 µmoles/min for a duration of 50 min. Under these conditions an average of 50 ± 10% reduction of blood Glu is achieved (Fig. 3). Each point represents the average of three independent experiments ± SD. A Student's t-test performed for the steady-state values reached before and during the infusion of pyruvate and oxaloacetate shows a p-value of 0.58. The broken lines show the steady-state levels reached before the intravenous administration of pyruvate and oxaloacetate.

When the perfusion is performed with a solution containing 250 µm Glu solution and 0.2 µCi [3H]Glu/mL, the absorption of [3H]Glu decreases to a steady state of 30.8 ± 0.5% because of the competition for the brain absorption sites by unlabeled Glu. Upon intravenous injection of pyruvate and oxaloacetate, the absorption of radioactive Glu from the perfusion fluid slightly increases to an apparent steady state of 33.8 ± 0.9% as some of the competing unlabeled Glu is pumped into blood decreasing thereby the competition with [3H]Glu for uptake by the various Glu transporters.

Impact of blood Glu scavenging on the basal CSF Glu levels

Figure 7 illustrates the finding that, as expected, the intravenous administration of the GOT/GPT co-substrates causes a reduction of CSF Glu. In these experiments, CSF aliquots were collected from a cannula implanted in the cisterna magna, before, during, and after the intravenous administration of pyruvate and oxaloacetate.

Figure 7.

Impact of blood Glu scavenging on CSF Glu levels. Evolution of CSFGlu levels upon intravenous administration for a duration of 30 min (black bar) of pyruvate and oxaloacetate (50 µmoles each/min). Under these conditions an average of 50 ± 10% reduction of blood Glu is achieved (Fig. 3). One representative experiment out of five performed. Each point represents the average of two Glu determinations ± SD.

The CSF Glu levels were determined here by the cycling method of Kato et al. (1973). It can be seen that, under conditions that cause a 50% decrease of blood Glu levels, a parallel decrease is observed in Glu CSF.

Discussion

In this paper, we examined the prediction that a decrease of blood Glu levels should increase the ability of brain to remove Glu from the CSF into blood. Though known for more than 40 years (Berl et al. 1961), the latter process of brain-to-blood Glu efflux has attracted much less attention than the process of the restricted Glu entry into the brain (Yudilevich et al. 1972; Banos et al. 1975; al-Sarraf et al. 1995, 1997). However, the very extensive vascularization of the brain parenchyma with an average intercapillary distance of only 24 µm (Pawlik et al. 1981) and the presence of Glu transporters on the antiluminal (brain) side of the blood capillaries (O'Kane et al. 1999) strongly suggest that the homeostasis of Glu in the brain extracellular spaces is maintained not only by the glial and neuronal Glu transporters but also by those present on the brain capillary endothelial cells. Indeed, it has been observed that 40% of [3H]Glu microinjected into the rat parietal cortex was eliminated from brain-to-blood in 20 min (Hosoya et al. 1999). CSF excess Glu, however, is likely to be eliminated to blood via the choroid plexus epithelial cells (Preston and Segal 1990), though some incorporation into the periventricular parenchyma also takes place (Levin et al. 1966).

To account for the observation that the intracerebroventricular injection of [3H]Glu, seen from the blood perspective, is characterized by an initial fast increase followed by a steady state during which the blood [3H]Glu levels remain constant, we suggest that the CSF acts as a temporary reservoir of diluted [3H]Glu and transports Glu into blood at a rate equal to that of Glu elimination from blood. As we estimated the latter to have a rate constant of 0.035/min, it follows that the brain-to-blood [3H]Glu efflux rate constant is identical to the value of 0.0346/min reported using the very different brain efflux index method. The latter measures the brain-to-blood efflux rate on the basis of the percentage of radioactivity remaining in brain as a function of time (Hosoya et al. 1999).

We have attempted to demonstrate here the feasibility of increasing the brain-to-blood Glu efflux by using two basic paradigms based on the fate of radiolabeled Glu infused into brain. In the first, we followed its appearance in blood before, during, and after decreasing blood Glu levels. In the other, we followed its disappearance from brain before, during, and after decreasing blood Glu levels. In both cases, the results obtained point out to the same conclusion that the intravenous administration of pyruvate and oxaloacetate that reduces blood Glu levels also increases the brain-to-blood Glu efflux. The fact that the reduction of blood Glu levels is not closely matched by an equivalent increased efflux of [3H]Glu is due to the fact that the latter is affected by the concentration of non-labeled Glu present in the CSF which competes with [3H]Glu for the efflux to blood. However, when the CSF levels of non-labeled Glu are monitored, there is a good match between the decrease of blood Glu levels and those of the basal CSF Glu levels. That a decrease of blood Glu causes an increased Glu efflux from brain confirms the early observation of Drewes et al. (1977) that perfusion of the dog brain with plasma containing lower Glu levels than normal caused a larger efflux of brain Glu. It is also in line with the mechanism proposed by O'Kane et al. (1999) to account for the elimination of Glu from brain into blood in spite of the highly unfavorable Glu concentration gradient between ISF/CSF Glu and blood plasma. Lowering plasma Glu facilitates thus the Glu transport from brain ISF/CSF across the capillary luminal membrane, and decreases the sodium-dependent transport of Glu from the endothelial cell into the brain CSF/ISF, causing thereby an enhanced vectorial transport of Glu from brain-to-blood. An alternative possibility that we cannot rule out at this stage is that the scavenging of blood Glu would increase the clearance rate of bulk CSF to blood (al-Sarraf et al. 2000). We are, however, unable to suggest any specific mechanism by which such process would take place.

In any event, the results presented here reveal that the blood Glu levels control the movements of Glu from brain-to-blood. Though upon discontinuation of the pyruvate and oxaloacetate administration, the Glu levels tend to recover their original values both in blood and in the CSF/ISF, the two recovery processes are unlikely to be linked because of the relative impermeability of the blood–brain barrier to a Glu influx from blood into brain. While the recovery in blood can possibly be attributed to a reparative increased Glu efflux from Glu-containing organs such as liver and muscle (Hediger and Welbourne 1999), the recovery of the CSF/ISF Glu levels is probably due to an enhanced Glu leakage from neurons and glia. The kinetics of recovery of ISF/CSF Glu levels is in fact a critical factor in the proposed strategy of decreasing blood Glu levels for therapeutic purposes. In the various cases in which the abnormally high CSF/ISF Glu levels are suggested to have an important pathological role, the kinetics of recovery of Glu following its blood-mediated depletion will dictate whether the treatment with pyruvate/oxaloacetate will be effective or not.

The present study emphasizes the ability of pyruvate/oxaloacetate to decrease both blood and CSF/ISF Glu levels, and raises the possibility of using these natural GPT and GOT co-substrates for therapeutic purposes in the context of the large number of neurodegenerative conditions in which the excess ISF/CSF Glu is thought to exert a crucial excitotoxicity. Though the neuroprotective properties of oxaloacetate have not been studied so far, those of pyruvate are well established (Izumi et al. 1994; Matsumoto et al. 1994; Ruiz et al. 1998; Maus et al. 1999; Mongan et al. 1999, 2001; Lee et al. 2001) but have not been used yet in the clinic. The observed neuroprotective effects of pyruvate in hemorrhagic shock (Mongan et al. 2001) and forebrain ischemia (Lee et al. 2001) have been accounted by the ability of pyruvate to cross the blood–brain barrier and to stabilize the cerebral neocortical energetics including an inhibition of Glu release. However, the direct topical application of pyruvate (20 mm) onto the cortex of rats submitted to brain ischemia did not decrease but rather increased Glu release (Phillis et al. 2001). We therefore suggest that an additional explanation for the neuroprotective properties of pyruvate could be a blood pyruvate-mediated enhanced brain-to-blood Glu efflux.

In summary, we have brought here the experimental evidence establishing that the intravenous administration of the GPT and GPT Glu co-substrates pyruvate and oxaloacetate causes a decrease of CSF Glu levels by virtue of their ability to decrease blood plasma Glu. This enhanced brain-to-blood Glu efflux is suggested to account for the neuroprotective effects of pyruvate in forebrain ischemia and hemorrhagic shock.

Acknowledgements

MG was supported in part by a VEGA 7231 grant from the Slovak Republic. This work was supported by grants from the Dominic Institute, the Nella Benoziyo Center and from the Julius and Ray Charlestein Foundation. VIT holds the Louis and Florence Katz-Cohen professorial chair of neuropharmacology.

Ancillary